From nicholas at mxc.ca  Mon May  2 00:05:29 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Mon, 02 May 2011 05:05:29 -0000
Subject: [llvm-commits] [llvm] r130684 -
	/llvm/trunk/lib/Object/COFFObjectFile.cpp
Message-ID: <20110502050529.F046D2A6C138@llvm.org>

Author: nicholas
Date: Mon May  2 00:05:29 2011
New Revision: 130684

URL: http://llvm.org/viewvc/llvm-project?rev=130684&view=rev
Log:
Remove dead variable pointed out by GCC 4.6 warnings, and reflow this a little
to scope a variable more tightly per llvm coding style. No functional change.

Modified:
    llvm/trunk/lib/Object/COFFObjectFile.cpp

Modified: llvm/trunk/lib/Object/COFFObjectFile.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Object/COFFObjectFile.cpp?rev=130684&r1=130683&r2=130684&view=diff
==============================================================================
--- llvm/trunk/lib/Object/COFFObjectFile.cpp (original)
+++ llvm/trunk/lib/Object/COFFObjectFile.cpp Mon May  2 00:05:29 2011
@@ -186,11 +186,8 @@
     return ret;
 
   uint32_t Characteristics = 0;
-  uint32_t PointerToRawData = 0;
-  const coff_section *Section = getSection(symb->SectionNumber);
-  if (Section) {
+  if (const coff_section *Section = getSection(symb->SectionNumber)) {
     Characteristics = Section->Characteristics;
-    PointerToRawData = Section->PointerToRawData;
   }
 
   switch (symb->SectionNumber) {



From nicholas at mxc.ca  Mon May  2 00:24:48 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Mon, 02 May 2011 05:24:48 -0000
Subject: [llvm-commits] [llvm] r130685 -
	/llvm/trunk/lib/Support/CommandLine.cpp
Message-ID: <20110502052448.116592A6C138@llvm.org>

Author: nicholas
Date: Mon May  2 00:24:47 2011
New Revision: 130685

URL: http://llvm.org/viewvc/llvm-project?rev=130685&view=rev
Log:
In option typo correction, consider -foo=VALUE flags as two distinct parts. The
comments claimed it did this, but the LHS value was actually an unused variable.

The new system considers only the '-foo' part when comparing it for typos
against flags that have values, but still look at the whole string for flags
that don't. That way, we'll still correct '-inst=combine' to '-instcombine'.

Modified:
    llvm/trunk/lib/Support/CommandLine.cpp

Modified: llvm/trunk/lib/Support/CommandLine.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/CommandLine.cpp?rev=130685&r1=130684&r2=130685&view=diff
==============================================================================
--- llvm/trunk/lib/Support/CommandLine.cpp (original)
+++ llvm/trunk/lib/Support/CommandLine.cpp Mon May  2 00:24:47 2011
@@ -186,12 +186,14 @@
 /// have already been stripped.
 static Option *LookupNearestOption(StringRef Arg,
                                    const StringMap<Option*> &OptionsMap,
-                                   const char *&NearestString) {
+                                   std::string &NearestString) {
   // Reject all dashes.
   if (Arg.empty()) return 0;
 
   // Split on any equal sign.
-  StringRef LHS = Arg.split('=').first;
+  std::pair<StringRef, StringRef> SplitArg = Arg.split('=');
+  StringRef &LHS = SplitArg.first;  // LHS == Arg when no '=' is present.
+  StringRef &RHS = SplitArg.second;
 
   // Find the closest match.
   Option *Best = 0;
@@ -204,14 +206,19 @@
     if (O->ArgStr[0])
       OptionNames.push_back(O->ArgStr);
 
+    bool PermitValue = O->getValueExpectedFlag() != cl::ValueDisallowed;
+    StringRef Flag = PermitValue ? LHS : Arg;
     for (size_t i = 0, e = OptionNames.size(); i != e; ++i) {
       StringRef Name = OptionNames[i];
       unsigned Distance = StringRef(Name).edit_distance(
-        Arg, /*AllowReplacements=*/true, /*MaxEditDistance=*/BestDistance);
+        Flag, /*AllowReplacements=*/true, /*MaxEditDistance=*/BestDistance);
       if (!Best || Distance < BestDistance) {
         Best = O;
-        NearestString = OptionNames[i];
         BestDistance = Distance;
+	if (RHS.empty() || !PermitValue)
+	  NearestString = OptionNames[i];
+	else
+	  NearestString = std::string(OptionNames[i]) + "=" + RHS.str();
       }
     }
   }
@@ -611,7 +618,7 @@
   for (int i = 1; i < argc; ++i) {
     Option *Handler = 0;
     Option *NearestHandler = 0;
-    const char *NearestHandlerString = 0;
+    std::string NearestHandlerString;
     StringRef Value;
     StringRef ArgName = "";
 



From stoklund at 2pi.dk  Mon May  2 00:29:56 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Mon, 02 May 2011 05:29:56 -0000
Subject: [llvm-commits] [llvm] r130686 -
	/llvm/trunk/include/llvm/CodeGen/SlotIndexes.h
Message-ID: <20110502052956.57A992A6C138@llvm.org>

Author: stoklund
Date: Mon May  2 00:29:56 2011
New Revision: 130686

URL: http://llvm.org/viewvc/llvm-project?rev=130686&view=rev
Log:
Add a SlotIndexes::insertMachineInstrInMaps to insert the instruction after any null indexes.

This makes a difference if a live interval is referring to a deleted
instruction. It can be important to insert an instruction before or after a
deleted instruction to avoid interference.

Modified:
    llvm/trunk/include/llvm/CodeGen/SlotIndexes.h

Modified: llvm/trunk/include/llvm/CodeGen/SlotIndexes.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/SlotIndexes.h?rev=130686&r1=130685&r2=130686&view=diff
==============================================================================
--- llvm/trunk/include/llvm/CodeGen/SlotIndexes.h (original)
+++ llvm/trunk/include/llvm/CodeGen/SlotIndexes.h Mon May  2 00:29:56 2011
@@ -511,6 +511,40 @@
       return nextNonNull;
     }
 
+    /// getIndexBefore - Returns the index of the last indexed instruction
+    /// before MI, or the the start index of its basic block.
+    /// MI is not required to have an index.
+    SlotIndex getIndexBefore(const MachineInstr *MI) const {
+      const MachineBasicBlock *MBB = MI->getParent();
+      assert(MBB && "MI must be inserted inna basic block");
+      MachineBasicBlock::const_iterator I = MI, B = MBB->begin();
+      for (;;) {
+        if (I == B)
+          return getMBBStartIdx(MBB);
+        --I;
+        Mi2IndexMap::const_iterator MapItr = mi2iMap.find(I);
+        if (MapItr != mi2iMap.end())
+          return MapItr->second;
+      }
+    }
+
+    /// getIndexAfter - Returns the index of the first indexed instruction
+    /// after MI, or the end index of its basic block.
+    /// MI is not required to have an index.
+    SlotIndex getIndexAfter(const MachineInstr *MI) const {
+      const MachineBasicBlock *MBB = MI->getParent();
+      assert(MBB && "MI must be inserted inna basic block");
+      MachineBasicBlock::const_iterator I = MI, E = MBB->end();
+      for (;;) {
+        ++I;
+        if (I == E)
+          return getMBBEndIdx(MBB);
+        Mi2IndexMap::const_iterator MapItr = mi2iMap.find(I);
+        if (MapItr != mi2iMap.end())
+          return MapItr->second;
+      }
+    }
+
     /// Return the (start,end) range of the given basic block number.
     const std::pair<SlotIndex, SlotIndex> &
     getMBBRange(unsigned Num) const {
@@ -604,7 +638,10 @@
 
     /// Insert the given machine instruction into the mapping. Returns the
     /// assigned index.
-    SlotIndex insertMachineInstrInMaps(MachineInstr *mi) {
+    /// If Late is set and there are null indexes between mi's neighboring
+    /// instructions, create the new index after the null indexes instead of
+    /// before them.
+    SlotIndex insertMachineInstrInMaps(MachineInstr *mi, bool Late = false) {
       assert(mi2iMap.find(mi) == mi2iMap.end() && "Instr already indexed.");
       // Numbering DBG_VALUE instructions could cause code generation to be
       // affected by debug information.
@@ -614,34 +651,25 @@
 
       assert(mbb != 0 && "Instr must be added to function.");
 
-      MachineBasicBlock::iterator miItr(mi);
-      IndexListEntry *newEntry;
-      // Get previous index, considering that not all instructions are indexed.
-      IndexListEntry *prevEntry;
-      for (;;) {
-        // If mi is at the mbb beginning, get the prev index from the mbb.
-        if (miItr == mbb->begin()) {
-          prevEntry = &getMBBStartIdx(mbb).entry();
-          break;
-        }
-        // Otherwise rewind until we find a mapped instruction.
-        Mi2IndexMap::const_iterator itr = mi2iMap.find(--miItr);
-        if (itr != mi2iMap.end()) {
-          prevEntry = &itr->second.entry();
-          break;
-        }
+      // Get the entries where mi should be inserted.
+      IndexListEntry *prevEntry, *nextEntry;
+      if (Late) {
+        // Insert mi's index immediately before the following instruction.
+        nextEntry = &getIndexAfter(mi).entry();
+        prevEntry = nextEntry->getPrev();
+      } else {
+        // Insert mi's index immediately after the preceeding instruction.
+        prevEntry = &getIndexBefore(mi).entry();
+        nextEntry = prevEntry->getNext();
       }
 
-      // Get next entry from previous entry.
-      IndexListEntry *nextEntry = prevEntry->getNext();
-
       // Get a number for the new instr, or 0 if there's no room currently.
       // In the latter case we'll force a renumber later.
       unsigned dist = ((nextEntry->getIndex() - prevEntry->getIndex())/2) & ~3u;
       unsigned newNumber = prevEntry->getIndex() + dist;
 
       // Insert a new list entry for mi.
-      newEntry = createEntry(mi, newNumber);
+      IndexListEntry *newEntry = createEntry(mi, newNumber);
       insert(nextEntry, newEntry);
 
       // Renumber locally if we need to.



From stoklund at 2pi.dk  Mon May  2 00:29:59 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Mon, 02 May 2011 05:29:59 -0000
Subject: [llvm-commits] [llvm] r130687 - in /llvm/trunk/lib/CodeGen:
 LiveRangeEdit.cpp LiveRangeEdit.h SplitKit.cpp
Message-ID: <20110502052959.22D8D2A6C139@llvm.org>

Author: stoklund
Date: Mon May  2 00:29:58 2011
New Revision: 130687

URL: http://llvm.org/viewvc/llvm-project?rev=130687&view=rev
Log:
Minimize the slot indexes spanned by register ranges created when splitting.

When an interfering live range ends at a dead slot index between two
instructions, make sure that the inserted copy instruction gets a slot index
after the dead ones. This makes it possible to avoid the interference.

Ideally, there shouldn't be interference ending at a deleted instruction, but
physical register coalescing can sometimes do that to sub-registers.

This fixes PR9823.

Modified:
    llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp
    llvm/trunk/lib/CodeGen/LiveRangeEdit.h
    llvm/trunk/lib/CodeGen/SplitKit.cpp

Modified: llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp?rev=130687&r1=130686&r2=130687&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp (original)
+++ llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp Mon May  2 00:29:58 2011
@@ -140,11 +140,13 @@
                                          const Remat &RM,
                                          LiveIntervals &lis,
                                          const TargetInstrInfo &tii,
-                                         const TargetRegisterInfo &tri) {
+                                         const TargetRegisterInfo &tri,
+                                         bool Late) {
   assert(RM.OrigMI && "Invalid remat");
   tii.reMaterialize(MBB, MI, DestReg, 0, RM.OrigMI, tri);
   rematted_.insert(RM.ParentVNI);
-  return lis.InsertMachineInstrInMaps(--MI).getDefIndex();
+  return lis.getSlotIndexes()->insertMachineInstrInMaps(--MI, Late)
+           .getDefIndex();
 }
 
 void LiveRangeEdit::eraseVirtReg(unsigned Reg, LiveIntervals &LIS) {

Modified: llvm/trunk/lib/CodeGen/LiveRangeEdit.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LiveRangeEdit.h?rev=130687&r1=130686&r2=130687&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/LiveRangeEdit.h (original)
+++ llvm/trunk/lib/CodeGen/LiveRangeEdit.h Mon May  2 00:29:58 2011
@@ -165,7 +165,8 @@
                             const Remat &RM,
                             LiveIntervals&,
                             const TargetInstrInfo&,
-                            const TargetRegisterInfo&);
+                            const TargetRegisterInfo&,
+                            bool Late = false);
 
   /// markRematerialized - explicitly mark a value as rematerialized after doing
   /// it manually.

Modified: llvm/trunk/lib/CodeGen/SplitKit.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SplitKit.cpp?rev=130687&r1=130686&r2=130687&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SplitKit.cpp (original)
+++ llvm/trunk/lib/CodeGen/SplitKit.cpp Mon May  2 00:29:58 2011
@@ -579,15 +579,20 @@
   SlotIndex Def;
   LiveInterval *LI = Edit->get(RegIdx);
 
+  // We may be trying to avoid interference that ends at a deleted instruction,
+  // so always begin RegIdx 0 early and all others late.
+  bool Late = RegIdx != 0;
+
   // Attempt cheap-as-a-copy rematerialization.
   LiveRangeEdit::Remat RM(ParentVNI);
   if (Edit->canRematerializeAt(RM, UseIdx, true, LIS)) {
-    Def = Edit->rematerializeAt(MBB, I, LI->reg, RM, LIS, TII, TRI);
+    Def = Edit->rematerializeAt(MBB, I, LI->reg, RM, LIS, TII, TRI, Late);
   } else {
     // Can't remat, just insert a copy from parent.
     CopyMI = BuildMI(MBB, I, DebugLoc(), TII.get(TargetOpcode::COPY), LI->reg)
                .addReg(Edit->getReg());
-    Def = LIS.InsertMachineInstrInMaps(CopyMI).getDefIndex();
+    Def = LIS.getSlotIndexes()->insertMachineInstrInMaps(CopyMI, Late)
+            .getDefIndex();
   }
 
   // Define the value in Reg.



From chandlerc at gmail.com  Mon May  2 00:49:01 2011
From: chandlerc at gmail.com (Chandler Carruth)
Date: Mon, 02 May 2011 05:49:01 -0000
Subject: [llvm-commits] [llvm] r130688 -
	/llvm/trunk/include/llvm/CodeGen/SlotIndexes.h
Message-ID: <20110502054901.DC6E42A6C139@llvm.org>

Author: chandlerc
Date: Mon May  2 00:49:01 2011
New Revision: 130688

URL: http://llvm.org/viewvc/llvm-project?rev=130688&view=rev
Log:
Remove an unused variable in NDEBUG (found with -Wunused-variable).

Modified:
    llvm/trunk/include/llvm/CodeGen/SlotIndexes.h

Modified: llvm/trunk/include/llvm/CodeGen/SlotIndexes.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/SlotIndexes.h?rev=130688&r1=130687&r2=130688&view=diff
==============================================================================
--- llvm/trunk/include/llvm/CodeGen/SlotIndexes.h (original)
+++ llvm/trunk/include/llvm/CodeGen/SlotIndexes.h Mon May  2 00:49:01 2011
@@ -647,9 +647,7 @@
       // affected by debug information.
       assert(!mi->isDebugValue() && "Cannot number DBG_VALUE instructions.");
 
-      MachineBasicBlock *mbb = mi->getParent();
-
-      assert(mbb != 0 && "Instr must be added to function.");
+      assert(mi->getParent() != 0 && "Instr must be added to function.");
 
       // Get the entries where mi should be inserted.
       IndexListEntry *prevEntry, *nextEntry;



From grosser at fim.uni-passau.de  Mon May  2 02:48:30 2011
From: grosser at fim.uni-passau.de (Tobias Grosser)
Date: Mon, 02 May 2011 07:48:30 -0000
Subject: [llvm-commits] [polly] r130689 - in /polly/trunk/www: ./
 experiments/ experiments/matmul/ images/ images/performance/ publications/
Message-ID: <20110502074833.BA7AB2A6C139@llvm.org>

Author: grosser
Date: Mon May  2 02:48:29 2011
New Revision: 130689

URL: http://llvm.org/viewvc/llvm-project?rev=130689&view=rev
Log:
Add new website for Polly

Use the content of the Polly wiki page[1] to create a new website. I do not yet
plan to officially promote this website, but it is already a solid base that we
can improve and peer review.

[1] http://wiki.llvm.org/Polly

Added:
    polly/trunk/www/
    polly/trunk/www/content.css
    polly/trunk/www/contributors.html
    polly/trunk/www/examples.html
    polly/trunk/www/experiments/
    polly/trunk/www/experiments/matmul/
    polly/trunk/www/experiments/matmul/init_array___%1---%19.jscop
    polly/trunk/www/experiments/matmul/main___%1---%17.jscop
    polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged
    polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged+tiled
    polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged+tiled+vector
    polly/trunk/www/experiments/matmul/matmul.c
    polly/trunk/www/experiments/matmul/matmul.normalopt.exe   (with props)
    polly/trunk/www/experiments/matmul/matmul.normalopt.ll
    polly/trunk/www/experiments/matmul/matmul.normalopt.s
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.exe   (with props)
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.ll
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.s
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.exe   (with props)
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.ll
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.s
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.exe   (with props)
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.ll
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.s
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged.exe   (with props)
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged.ll
    polly/trunk/www/experiments/matmul/matmul.polly.interchanged.s
    polly/trunk/www/experiments/matmul/matmul.preopt.ll
    polly/trunk/www/experiments/matmul/matmul.s
    polly/trunk/www/experiments/matmul/runall.sh   (with props)
    polly/trunk/www/experiments/matmul/scops.init_array.dot
    polly/trunk/www/experiments/matmul/scops.init_array.dot.png
    polly/trunk/www/experiments/matmul/scops.main.dot
    polly/trunk/www/experiments/matmul/scops.main.dot.png
    polly/trunk/www/experiments/matmul/scops.print_array.dot
    polly/trunk/www/experiments/matmul/scops.print_array.dot.png
    polly/trunk/www/experiments/matmul/scopsonly.init_array.dot
    polly/trunk/www/experiments/matmul/scopsonly.init_array.dot.png
    polly/trunk/www/experiments/matmul/scopsonly.main.dot
    polly/trunk/www/experiments/matmul/scopsonly.main.dot.png
    polly/trunk/www/experiments/matmul/scopsonly.print_array.dot
    polly/trunk/www/experiments/matmul/scopsonly.print_array.dot.png
    polly/trunk/www/get_started.html
    polly/trunk/www/images/
    polly/trunk/www/images/architecture.png
    polly/trunk/www/images/iit-madras.png
    polly/trunk/www/images/osu.png
    polly/trunk/www/images/performance/
    polly/trunk/www/images/performance/parallel-large.png
    polly/trunk/www/images/performance/parallel-small.png
    polly/trunk/www/images/performance/sequential-large.png
    polly/trunk/www/images/performance/sequential-small.png
    polly/trunk/www/images/sys-uni.png
    polly/trunk/www/images/uni-passau.png
    polly/trunk/www/index.html
    polly/trunk/www/menu.css
    polly/trunk/www/menu.html.incl
    polly/trunk/www/passes.html
    polly/trunk/www/performance.html
    polly/trunk/www/publications/
    polly/trunk/www/publications.html
    polly/trunk/www/publications/grosser-impact-2011-slides.pdf
    polly/trunk/www/publications/grosser-impact-2011.pdf
    polly/trunk/www/todo.html

Added: polly/trunk/www/content.css
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/content.css?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/content.css (added)
+++ polly/trunk/www/content.css Mon May  2 02:48:29 2011
@@ -0,0 +1,33 @@
+html { margin: 0px; } body { margin: 8px; }
+
+html, body {
+  padding:0px;
+  font-size:small; font-family:"Lucida Grande", "Lucida Sans Unicode", Arial, Verdana, Helvetica, sans-serif; background-color: #fff; color: #222;
+  line-height:1.5;
+}
+
+h1, h2, h3, tt { color: #000 }
+
+h1 { padding-top:0px; margin-top:0px;}
+h2 { color:#333333; padding-top:0.5em; }
+h3 { padding-top: 0.5em; margin-bottom: -0.25em; color:#2d58b7}
+li { padding-bottom: 0.5em; }
+ul { padding-left:1.5em; }
+
+PRE.code {padding-left: 0.5em; background-color: #eeeeee}
+PRE {padding-left: 0.5em}
+
+/* Slides */
+IMG.img_slide {
+    display: block;
+    margin-left: auto;
+    margin-right: auto
+}
+
+.itemTitle { color:#2d58b7 }
+
+span.error { color:red }
+span.caret { color:green; font-weight:bold }
+
+/* Tables */
+tr { vertical-align:top }

Added: polly/trunk/www/contributors.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/contributors.html?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/contributors.html (added)
+++ polly/trunk/www/contributors.html Mon May  2 02:48:29 2011
@@ -0,0 +1,55 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
+          "http://www.w3.org/TR/html4/strict.dtd">
+<html>
+<head>
+  <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1" />
+  <title>Polly - Contributors</title>
+  <link type="text/css" rel="stylesheet" href="menu.css" />
+  <link type="text/css" rel="stylesheet" href="content.css" />
+</head>
+<body>
+
+<!--#include virtual="menu.html.incl"-->
+
+<div id="content">
+
+<h1>Polly: Contributors</h1>
+
+Polly is developed by a team of students supported by different universities.
+
+<h2>People</h2>
+<h3>Raghesh Aloor</h3>
+<p>Raghesh works on OpenMP code generation. He is funded as Google Summer of Code
+Student 2011.</p>
+
+<h3>Tobias Grosser</h3>
+<p>Tobias is one of the two Co-founders of Polly. He designed the overall
+architecture and contributed to almost every part of Polly. He did his work
+during his diploma studies at University of Passau. Furthermore, he spent 6
+months at Ohio State University where he was founded by the U.S. National
+Science Foundation through awards 0811781 and 0926688.</p>
+
+<p>Website: <a href="http://www.grosser.es">www.grosser.es</a></p>
+
+
+<h3>Andreas Simbuerger</h3>
+<p>
+Andreas works on the profiling infrastructure during his PhD at University of
+Passau.
+</p>
+<p>Website: <a href="http://www.infosun.fim.uni-passau.de/cl/staff/simbuerger/">
+http://www.infosun.fim.uni-passau.de/cl/staff/simbuerger/</a></p>
+<h3>Hongbin Zheng</h3>
+<p>Hongbin Zheng is one of the two Co-founders of Polly. He was funded as a
+Google Summer of Code Student 2010 and implemented parts of the Polly frontends
+as well as the automake/cmake infrastructure.</p>
+
+<h2> Universities</h2>
+
+<p>Polly is supported by the following Universities.</p>
+<img src="images/iit-madras.png" style="padding:1em" />
+<img src="images/uni-passau.png" style="padding: 1em; padding-bottom:2em;"/>
+<img src="images/osu.png" style="padding:1em"/>
+<img src="images/sys-uni.png" style="padding:1em"/>
+</body>
+</html>

Added: polly/trunk/www/examples.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/examples.html?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/examples.html (added)
+++ polly/trunk/www/examples.html Mon May  2 02:48:29 2011
@@ -0,0 +1,317 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" 
+          "http://www.w3.org/TR/html4/strict.dtd">
+<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
+<html>
+<head>
+  <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
+  <title>Polly - Examples</title>
+  <link type="text/css" rel="stylesheet" href="menu.css">
+  <link type="text/css" rel="stylesheet" href="content.css">
+</head>
+<body>
+<!--#include virtual="menu.html.incl"-->
+<div id="content">
+  <!--*********************************************************************-->
+  <h1>Polly: Examples</h1>
+  <!--*********************************************************************-->
+<!--=====================================================================-->
+<h2>Optimize Matrix Multiplication Manually</h2>
+<!--=====================================================================-->
+
+<p>Polly does not yet focus on end user, but on research and the development of
+new optimizations. Hence for the users of Polly it is often necessary to
+understand how Polly works internally. To get an overview of the different steps
+taken during polyhedral compilation, we give a step by step example on how to
+use the different Polly passes. For this we optimize a simple matrix
+multiplication kernel. In case you look for a more automated way of executing
+Polly, check out the pollycc tool in utils/pollycc.</p>
+
+The files used and created in this example are available <a
+href="experiments/matmul">here</a>.
+
+<ol>
+<li><h4>Create LLVM-IR from the C code</h4>
+
+Polly works on LLVM-IR. Hence it is necessary to translate the source files into
+LLVM-IR. If more than on file should be optimized the files can be combined into
+a single file with llvm-link.
+
+<pre class="code">clang -S -emit-llvm matmul.c -o matmul.s</pre>
+</li>
+
+
+<li><h4>Load Polly automatically when calling the 'opt' tool</h4>
+
+Polly is not built into opt or bugpoint, but it is a shared library that needs
+to be loaded into these tools explicitally. The Polly library is called
+LVMPolly.so. For a cmake build it is available in the build/lib/ directory,
+autoconf creates the same file in
+build/tools/polly/{Release+Asserts|Asserts|Debug}/lib. For convenience we create
+an alias that automatically loads Polly if 'opt' is called.
+<pre class="code">
+export PATH_TO_POLLY_LIB="~/polly/build/lib/"
+alias opt="opt -load ${PATH_TO_POLLY_LIB}/LLVMPolly.so"</pre>
+</li>
+
+<li><h4>Prepare the LLVM-IR for Polly</h4>
+
+Polly is only able to work with code that matches a canonical form. To translate
+the LLVM-IR into this form we use a set of canonicalication passes. For this
+example only three passes are necessary. To get good coverage on a larger set
+of input files a larger set is needed. pollycc contains a set of passes that has
+shown to be beneficial.
+<pre class="code">opt -S -mem2reg -loop-simplify -indvars matmul.s &gt; matmul.preopt.ll</pre></li>
+
+<li><h4>Show the SCoPs detected by Polly (optional)</h4>
+
+To understand if Polly was able to detect some SCoPs, we print the
+structure of the detected SCoPs. In our example two SCoPs were detected. One in
+'init_array' the other in 'main'.
+
+<pre class="code">opt -basicaa -polly-cloog -analyze -q matmul.preopt.ll</pre>
+
+<pre>
+init_array():
+for (c2=0;c2&lt;=1023;c2++) {
+  for (c4=0;c4&lt;=1023;c4++) {
+    Stmt_5(c2,c4);
+  }
+}
+
+main():
+for (c2=0;c2&lt;=1023;c2++) {
+  for (c4=0;c4&lt;=1023;c4++) {
+    Stmt_4(c2,c4);
+    for (c6=0;c6&lt;=1023;c6++) {
+      Stmt_6(c2,c4,c6);
+    }
+  }
+}
+</pre>
+</li>
+<li><h4>Highlight the detected SCoPs in the CFGs of the program (requires graphviz/dotty)</h4>
+
+Polly can use graphviz to graphically show a CFG in which the detected SCoPs are
+highlighted. It can also create '.dot' files that can be translated by
+the 'dot' utility into various graphic formats.
+
+<pre class="code">opt -basicaa -view-scops -disable-output matmul.preopt.ll
+opt -basicaa -view-scops-only -disable-output matmul.preopt.ll</pre>
+The output for the different functions<br />
+view-scops:
+<a href="experiments/matmul/scops.main.dot.png">main</a>,
+<a href="experiments/matmul/scops.init_array.dot.png">init_array</a>,
+<a href="experiments/matmul/scops.print_array.dot.png">print_array</a><br />
+view-scops-only:
+<a href="experiments/matmul/scopsonly.main.dot.png">main</a>,
+<a href="experiments/matmul/scopsonly.init_array.dot.png">init_array</a>,
+<a href="experiments/matmul/scopsonly.print_array.dot.png">print_array</a>
+</li>
+
+<li><h4>View the polyhedral representation of the SCoPs</h4>
+<pre class="code">opt -basicaa -polly-scops -analyze matmul.preopt.ll</pre>
+<pre>
+[...]
+Printing analysis 'Polly - Create polyhedral description of Scops' for region: '%1 =&gt;&nbsp;%17' in function 'init_array':
+   Context:
+   { [] }
+   Statements {
+   	Stmt_5
+           Domain&nbsp;:=
+               { Stmt_5[i0, i1]&nbsp;: i0 &gt;= 0 and i0 &lt;= 1023 and i1 &gt;= 0 and i1 &lt;= 1023 };
+           Scattering&nbsp;:=
+               { Stmt_5[i0, i1] -&gt; scattering[0, i0, 0, i1, 0] };
+           WriteAccess&nbsp;:= 
+               { Stmt_5[i0, i1] -&gt; MemRef_A[1037i0 + i1] };
+           WriteAccess&nbsp;:= 
+               { Stmt_5[i0, i1] -&gt; MemRef_B[1047i0 + i1] };
+   	FinalRead
+           Domain&nbsp;:=
+               { FinalRead[0] };
+           Scattering&nbsp;:=
+               { FinalRead[i0] -&gt; scattering[200000000, o1, o2, o3, o4] };
+           ReadAccess&nbsp;:= 
+               { FinalRead[i0] -&gt; MemRef_A[o0] };
+           ReadAccess&nbsp;:= 
+               { FinalRead[i0] -&gt; MemRef_B[o0] };
+   }
+Printing analysis 'Polly - Create polyhedral description of Scops' for region: '%0 =&gt; &lt;Function  Return&gt;' in function 'init_array':
+[...]
+Printing analysis 'Polly - Create polyhedral description of Scops' for region: '%1 =&gt;&nbsp;%17' in function 'main':
+   Context:
+   { [] }
+   Statements {
+   	Stmt_4
+           Domain&nbsp;:=
+               { Stmt_4[i0, i1]&nbsp;: i0 &gt;= 0 and i0 &lt;= 1023 and i1 &gt;= 0 and i1 &lt;= 1023 };
+           Scattering&nbsp;:=
+               { Stmt_4[i0, i1] -&gt; scattering[0, i0, 0, i1, 0, 0, 0] };
+           WriteAccess&nbsp;:= 
+               { Stmt_4[i0, i1] -&gt; MemRef_C[1067i0 + i1] };
+   	Stmt_6
+           Domain&nbsp;:=
+               { Stmt_6[i0, i1, i2]&nbsp;: i0 &gt;= 0 and i0 &lt;= 1023 and i1 &gt;= 0 and i1 &lt;= 1023 and i2 &gt;= 0 and i2 &lt;= 1023 };
+           Scattering&nbsp;:=
+               { Stmt_6[i0, i1, i2] -&gt; scattering[0, i0, 0, i1, 1, i2, 0] };
+           ReadAccess&nbsp;:= 
+               { Stmt_6[i0, i1, i2] -&gt; MemRef_C[1067i0 + i1] };
+           ReadAccess&nbsp;:= 
+               { Stmt_6[i0, i1, i2] -&gt; MemRef_A[1037i0 + i2] };
+           ReadAccess&nbsp;:= 
+               { Stmt_6[i0, i1, i2] -&gt; MemRef_B[i1 + 1047i2] };
+           WriteAccess&nbsp;:= 
+               { Stmt_6[i0, i1, i2] -&gt; MemRef_C[1067i0 + i1] };
+   	FinalRead
+           Domain&nbsp;:=
+               { FinalRead[0] };
+           Scattering&nbsp;:=
+               { FinalRead[i0] -&gt; scattering[200000000, o1, o2, o3, o4, o5, o6] };
+           ReadAccess&nbsp;:= 
+               { FinalRead[i0] -&gt; MemRef_C[o0] };
+           ReadAccess&nbsp;:= 
+               { FinalRead[i0] -&gt; MemRef_A[o0] };
+           ReadAccess&nbsp;:= 
+               { FinalRead[i0] -&gt; MemRef_B[o0] };
+   }
+Printing analysis 'Polly - Create polyhedral description of Scops' for region: '%0 =&gt; &lt;Function  Return&gt;' in function 'main':
+Invalid Scop!
+</pre>
+</li>
+
+<li><h4>Show the dependences for the SCoPs</h4>
+<pre class="code">opt -basicaa -polly-dependences -analyze matmul.preopt.ll</pre>
+<pre>Printing analysis 'Polly - Calculate dependences for SCoP' for region: 'for.cond =&gt; for.end28' in function 'init_array':
+   Must dependences:
+       {  }
+   May dependences:
+       {  }
+   Must no source:
+       {  }
+   May no source:
+       {  }
+Printing analysis 'Polly - Calculate dependences for SCoP' for region: 'for.cond =&gt; for.end48' in function 'main':
+   Must dependences:
+       {  Stmt_4[i0, i1] -&gt; Stmt_6[i0, i1, 0]&nbsp;:
+              i0 &gt;= 0 and i0 &lt;= 1023 and i1 &gt;= 0 and i1 &lt;= 1023;
+          Stmt_6[i0, i1, i2] -&gt; Stmt_6[i0, i1, 1 + i2]&nbsp;:
+              i0 &gt;= 0 and i0 &lt;= 1023 and i1 &gt;= 0 and i1 &lt;= 1023 and i2 &gt;= 0 and i2 &lt;= 1022; 
+          Stmt_6[i0, i1, 1023] -&gt; FinalRead[0]&nbsp;:
+              i1 &lt;= 1091540 - 1067i0 and i1 &gt;= -1067i0 and i1 &gt;= 0 and i1 &lt;= 1023;
+          Stmt_6[1023, i1, 1023] -&gt; FinalRead[0]&nbsp;:
+              i1 &gt;= 0 and i1 &lt;= 1023 
+       }
+   May dependences:
+       {  }
+   Must no source:
+       {  Stmt_6[i0, i1, i2] -&gt; MemRef_A[1037i0 + i2]&nbsp;:
+              i0 &gt;= 0 and i0 &lt;= 1023 and i1 &gt;= 0 and i1 &lt;= 1023 and i2 &gt;= 0 and i2 &lt;= 1023; 
+          Stmt_6[i0, i1, i2] -&gt; MemRef_B[i1 + 1047i2]&nbsp;:
+              i0 &gt;= 0 and i0 &lt;= 1023 and i1 &gt;= 0 and i1 &lt;= 1023 and i2 &gt;= 0 and i2 &lt;= 1023; 
+          FinalRead[0] -&gt; MemRef_A[o0];
+          FinalRead[0] -&gt; MemRef_B[o0]
+          FinalRead[0] -&gt; MemRef_C[o0]&nbsp;:
+              o0 &gt;= 1092565 or (exists (e0 = [(o0)/1067]: o0 &lt;= 1091540 and o0 &gt;= 0
+              and 1067e0 &lt;= -1024 + o0 and 1067e0 &gt;= -1066 + o0)) or o0 &lt;= -1;
+       }
+   May no source:
+       {  }
+</pre></li>
+
+<li><h4>Export jscop files</h4>
+
+Polly can export the polyhedral representation in so called jscop files. Jscop
+files contain the polyhedral representation stored in a JSON file.
+<pre class="code">opt -basicaa -polly-export-jscop matmul.preopt.ll</pre>
+<pre>Writing SCoP 'for.cond =&gt; for.end28' in function 'init_array' to './init_array___%for.cond---%for.end28.jscop'.
+Writing SCoP 'for.cond =&gt; for.end48' in function 'main' to './main___%for.cond---%for.end48.jscop'.
+</pre></li>
+
+<li><h4>Import the changed jscop files and print the updated SCoP structure
+(optional)</h4>
+<p>Polly can import jscop files, where the schedules of the statements were
+changed. With the help of these updated files we can import transformations into
+Polly. It is possible to import different jscop files by providing the postfix
+of the jscop file that is imported.</p>
+<p> The optimized jscop files for this example are hand written. The schedule
+used was inspired by looking at the optimizations PoCC performs. If PoCC is
+installed Polly can often calculate such schedules fully automatically.</p>
+
+
+<pre class="code">opt -basicaa -polly-import-jscop -polly-print -disable-output matmul.preopt.ll -polly-import-jscop-postfix=.opt</pre>
+<pre>Cannot open file: ./init_array___%for.cond---%for.end28.jscop.opt
+Skipping import.
+In function: 'init_array' SCoP: for.cond =&gt; for.end28:
+for (c2=0;c2&lt;=1023;c2++) {
+  for (c4=0;c4&lt;=1023;c4++) {
+   &nbsp;%for.body4(c2,c4);
+  }
+}
+Reading SCoP 'for.cond =&gt; for.end48' in function 'main' from './main___%for.cond---%for.end48.scop.opt.opt'.
+In function: 'main' SCoP: for.cond =&gt; for.end48:
+for (c2=0;c2&lt;=1023;c2++) {
+  for (c4=0;c4&lt;=1023;c4++) {
+   &nbsp;%for.body4(c2,c4);
+  }
+}
+for (c2=0;c2&lt;=1023;c2++) {
+  for (c3=0;c3&lt;=1023;c3++) {
+    for (c4=0;c4&lt;=1023;c4++) {
+     &nbsp;%for.body12(c2,c4,c3);
+    }
+  }
+}
+</pre></li>
+
+<li><h4>Codegenerate the SCoPs</h4>
+This generates new code for the SCoPs detected by polly.
+If -polly-import is present, transformations specified in the imported openscop
+files will be applied.
+<pre class="code">opt -basicaa -polly-import -polly-import-postfix=.opt -polly-codegen matmul.preopt.ll | opt -O3 &gt; matmul.pollyopt.ll</pre>
+<pre>
+Cannot open file: ./init_array___%for.cond---%for.end28.scop.opt
+Skipping import.
+Reading SCoP 'for.cond =&gt; for.end48' in function 'main' from './main___%for.cond---%for.end48.scop.opt'.</pre>
+
+<pre class="code">opt matmul.preopt.ll | opt -O3 &gt; matmul.normalopt.ll</pre></li>
+
+<li><h4>Create the executables</h4>
+
+Create one executable optimized with plain -O3 as well as a set of executables
+optimized in different ways with Polly. One changes only the loop structure, the
+other adds tiling, the next adds vectorization and finally we use OpenMP
+parallelism.
+<pre class="code">
+llc matmul.normalopt.ll -o matmul.normalopt.s &amp;&amp; \
+    gcc matmul.normalopt.s -o matmul.normalopt.exe
+llc matmul.polly.interchanged.ll -o matmul.polly.interchanged.s &amp;&amp; \
+    gcc matmul.polly.interchanged.s -o matmul.polly.interchanged.exe
+llc matmul.polly.interchanged+tiled.ll -o matmul.polly.interchanged+tiled.s &amp;&amp; \
+    gcc matmul.polly.interchanged+tiled.s -o matmul.polly.interchanged+tiled.exe
+llc matmul.polly.interchanged+tiled+vector.ll -o matmul.polly.interchanged+tiled+vector.s &amp;&amp; \
+    gcc matmul.polly.interchanged+tiled+vector.s -o matmul.polly.interchanged+tiled+vector.exe
+llc matmul.polly.interchanged+tiled+vector+openmp.ll -o matmul.polly.interchanged+tiled+vector+openmp.s &amp;&amp; \
+    gcc -lgomp matmul.polly.interchanged+tiled+vector+openmp.s -o matmul.polly.interchanged+tiled+vector+openmp.exe
+    </pre>
+
+<li><h4>Compare the runtime of the executables</h4>
+
+By comparing the runtimes of the different code snippets we see that a simple
+loop interchange gives here the largest performance boost. However by adding
+vectorization and by using OpenMP we can further improve the performance
+significantly.
+<pre class="code">time ./matmul.normalopt.exe</pre>
+<pre>42.68 real, 42.55 user, 0.00 sys</pre>
+<pre class="code">time ./matmul.polly.interchanged.exe</pre>
+<pre>04.33 real, 4.30 user, 0.01 sys</pre>
+<pre class="code">time ./matmul.polly.interchanged+tiled.exe</pre>
+<pre>04.11 real, 4.10 user, 0.00 sys</pre>
+<pre class="code">time ./matmul.polly.interchanged+tiled+vector.exe</pre>
+<pre>01.39 real, 1.36 user, 0.01 sys</pre>
+<pre class="code">time ./matmul.polly.interchanged+tiled+vector+openmp.exe</pre>
+<pre>00.66 real, 2.58 user, 0.02 sys</pre>
+</li>
+</ol>
+
+</div>
+</body>
+</html>

Added: polly/trunk/www/experiments/matmul/init_array___%1---%19.jscop
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/init_array___%251---%2519.jscop?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/init_array___%1---%19.jscop (added)
+++ polly/trunk/www/experiments/matmul/init_array___%1---%19.jscop Mon May  2 02:48:29 2011
@@ -0,0 +1,21 @@
+{
+   "context" : "{ [] }",
+   "name" : "%1 => %19",
+   "statements" : [
+      {
+         "accesses" : [
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_5[i0, i1] -> MemRef_A[1536i0 + i1] }"
+            },
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_5[i0, i1] -> MemRef_B[1536i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_5[i0, i1] : i0 >= 0 and i0 <= 1535 and i1 >= 0 and i1 <= 1535 }",
+         "name" : "Stmt_5",
+         "schedule" : "{ Stmt_5[i0, i1] -> scattering[0, i0, 0, i1, 0] }"
+      }
+   ]
+}

Added: polly/trunk/www/experiments/matmul/main___%1---%17.jscop
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/main___%251---%2517.jscop?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/main___%1---%17.jscop (added)
+++ polly/trunk/www/experiments/matmul/main___%1---%17.jscop Mon May  2 02:48:29 2011
@@ -0,0 +1,40 @@
+{
+   "context" : "{ [] }",
+   "name" : "%1 => %17",
+   "statements" : [
+      {
+         "accesses" : [
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_4[i0, i1] -> MemRef_C[1536i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_4[i0, i1] : i0 >= 0 and i0 <= 1535 and i1 >= 0 and i1 <= 1535 }",
+         "name" : "Stmt_4",
+         "schedule" : "{ Stmt_4[i0, i1] -> scattering[0, i0, 0, i1, 0, 0, 0] }"
+      },
+      {
+         "accesses" : [
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_C[1536i0 + i1] }"
+            },
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_A[1536i0 + i2] }"
+            },
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_B[i1 + 1536i2] }"
+            },
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_C[1536i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_6[i0, i1, i2] : i0 >= 0 and i0 <= 1535 and i1 >= 0 and i1 <= 1535 and i2 >= 0 and i2 <= 1535 }",
+         "name" : "Stmt_6",
+         "schedule" : "{ Stmt_6[i0, i1, i2] -> scattering[0, i0, 0, i1, 1, i2, 0] }"
+      }
+   ]
+}

Added: polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/main___%251---%2517.jscop.interchanged?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged (added)
+++ polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged Mon May  2 02:48:29 2011
@@ -0,0 +1,40 @@
+{
+   "context" : "{ [] }",
+   "name" : "%1 => %17",
+   "statements" : [
+      {
+         "accesses" : [
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_4[i0, i1] -> MemRef_C[1067i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_4[i0, i1] : i0 >= 0 and i0 <= 1023 and i1 >= 0 and i1 <= 1023 }",
+         "name" : "Stmt_4",
+         "schedule" : "{ Stmt_4[i0, i1] -> scattering[0, i0, 0, i1, 0, 0, 0] }"
+      },
+      {
+         "accesses" : [
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_C[1067i0 + i1] }"
+            },
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_A[1037i0 + i2] }"
+            },
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_B[i1 + 1047i2] }"
+            },
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_C[1067i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_6[i0, i1, i2] : i0 >= 0 and i0 <= 1023 and i1 >= 0 and i1 <= 1023 and i2 >= 0 and i2 <= 1023 }",
+         "name" : "Stmt_6",
+         "schedule" : "{ Stmt_6[i0, i1, i2] -> scattering[1, i0, 0, i2, 0, i1, 0] }"
+      }
+   ]
+}

Added: polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged+tiled
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/main___%251---%2517.jscop.interchanged%2Btiled?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged+tiled (added)
+++ polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged+tiled Mon May  2 02:48:29 2011
@@ -0,0 +1,40 @@
+{
+   "context" : "{ [] }",
+   "name" : "%1 => %17",
+   "statements" : [
+      {
+         "accesses" : [
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_4[i0, i1] -> MemRef_C[1067i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_4[i0, i1] : i0 >= 0 and i0 <= 1023 and i1 >= 0 and i1 <= 1023 }",
+         "name" : "Stmt_4",
+         "schedule" : "{ Stmt_4[i0, i1] -> scattering[0, i0, 0, i1, 0, 0, 0] }"
+      },
+      {
+         "accesses" : [
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_C[1067i0 + i1] }"
+            },
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_A[1037i0 + i2] }"
+            },
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_B[i1 + 1047i2] }"
+            },
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_C[1067i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_6[i0, i1, i2] : i0 >= 0 and i0 <= 1023 and i1 >= 0 and i1 <= 1023 and i2 >= 0 and i2 <= 1023 }",
+         "name" : "Stmt_6",
+         "schedule" : "{ Stmt_6[i0, i1, i2] -> scattering[1, o0, o1, o2, i0, i2, i1]: o0 <= i0 < o0 + 64 and o1 <= i1 < o1 + 64 and o2 <= i2 < o2 + 64 and o0 % 64 = 0 and o1 % 64 = 0 and o2 % 64 = 0 }"
+      }
+   ]
+}

Added: polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged+tiled+vector
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/main___%251---%2517.jscop.interchanged%2Btiled%2Bvector?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged+tiled+vector (added)
+++ polly/trunk/www/experiments/matmul/main___%1---%17.jscop.interchanged+tiled+vector Mon May  2 02:48:29 2011
@@ -0,0 +1,40 @@
+{
+   "context" : "{ [] }",
+   "name" : "%1 => %17",
+   "statements" : [
+      {
+         "accesses" : [
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_4[i0, i1] -> MemRef_C[1067i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_4[i0, i1] : i0 >= 0 and i0 <= 1023 and i1 >= 0 and i1 <= 1023 }",
+         "name" : "Stmt_4",
+         "schedule" : "{ Stmt_4[i0, i1] -> scattering[0, i0, 0, i1, 0, 0, 0, 0] }"
+      },
+      {
+         "accesses" : [
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_C[1067i0 + i1] }"
+            },
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_A[1037i0 + i2] }"
+            },
+            {
+               "kind" : "read",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_B[i1 + 1047i2] }"
+            },
+            {
+               "kind" : "write",
+               "relation" : "{ Stmt_6[i0, i1, i2] -> MemRef_C[1067i0 + i1] }"
+            }
+         ],
+         "domain" : "{ Stmt_6[i0, i1, i2] : i0 >= 0 and i0 <= 1023 and i1 >= 0 and i1 <= 1023 and i2 >= 0 and i2 <= 1023 }",
+         "name" : "Stmt_6",
+         "schedule" : "{ Stmt_6[i0, i1, i2] -> scattering[1, o0, o1, o2, i0, i2, ii1, i1]: o0 <= i0 < o0 + 64 and o1 <= i1 < o1 + 64 and o2 <= i2 < o2 + 64 and o0 % 64 = 0 and o1 % 64 = 0 and o2 % 64 = 0 and ii1 % 4 = 0 and ii1 <= i1 < ii1 + 4}"
+      }
+   ]
+}

Added: polly/trunk/www/experiments/matmul/matmul.c
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.c?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/matmul.c (added)
+++ polly/trunk/www/experiments/matmul/matmul.c Mon May  2 02:48:29 2011
@@ -0,0 +1,52 @@
+#include <stdio.h>
+
+#define N 1536
+float A[N][N];
+float B[N][N];
+float C[N][N];
+
+void init_array()
+{
+    int i, j;
+
+    for (i=0; i<N; i++) {
+        for (j=0; j<N; j++) {
+            A[i][j] = (1+(i*j)%1024)/2.0;
+            B[i][j] = (1+(i*j)%1024)/2.0;
+        }
+    }
+}
+
+void print_array()
+{
+    int i, j;
+
+    for (i=0; i<N; i++) {
+        for (j=0; j<N; j++) {
+            fprintf(stdout, "%lf ", C[i][j]);
+            if (j%80 == 79) fprintf(stdout, "\n");
+        }
+        fprintf(stdout, "\n");
+    }
+}
+
+int main()
+{
+    int i, j, k;
+    double t_start, t_end;
+
+    init_array();
+
+    for(i=0; i<N; i++)  {
+        for(j=0; j<N; j++)  {
+            C[i][j] = 0;
+            for(k=0; k<N; k++)
+                C[i][j] = C[i][j] + A[i][k] * B[k][j];
+        }
+    }
+
+#ifdef TEST
+    print_array();
+#endif
+    return 0;
+}

Added: polly/trunk/www/experiments/matmul/matmul.normalopt.exe
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.normalopt.exe?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.normalopt.exe (added) and polly/trunk/www/experiments/matmul/matmul.normalopt.exe Mon May  2 02:48:29 2011 differ

Propchange: polly/trunk/www/experiments/matmul/matmul.normalopt.exe
------------------------------------------------------------------------------
    svn:executable = *

Added: polly/trunk/www/experiments/matmul/matmul.normalopt.ll
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.normalopt.ll?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.normalopt.ll (added) and polly/trunk/www/experiments/matmul/matmul.normalopt.ll Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/matmul.normalopt.s
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.normalopt.s?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/matmul.normalopt.s (added)
+++ polly/trunk/www/experiments/matmul/matmul.normalopt.s Mon May  2 02:48:29 2011
@@ -0,0 +1,203 @@
+	.file	"matmul.normalopt.ll"
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI0_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.globl	init_array
+	.align	16, 0x90
+	.type	init_array, at function
+init_array:                             # @init_array
+# BB#0:
+	xorl	%eax, %eax
+	movsd	.LCPI0_0(%rip), %xmm0
+	movq	%rax, %rcx
+	.align	16, 0x90
+.LBB0_1:                                # %.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB0_2 Depth 2
+	movq	$-1536, %rdx            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB0_2:                                #   Parent Loop BB0_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movl	%esi, %edi
+	sarl	$31, %edi
+	shrl	$22, %edi
+	addl	%esi, %edi
+	andl	$-1024, %edi            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%edi
+	leal	1(%rsi,%rdi), %edi
+	cvtsi2sd	%edi, %xmm1
+	mulsd	%xmm0, %xmm1
+	cvtsd2ss	%xmm1, %xmm1
+	movss	%xmm1, A+6144(%rax,%rdx,4)
+	movss	%xmm1, B+6144(%rax,%rdx,4)
+	addl	%ecx, %esi
+	incq	%rdx
+	jne	.LBB0_2
+# BB#3:                                 #   in Loop: Header=BB0_1 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	incq	%rcx
+	cmpq	$1536, %rcx             # imm = 0x600
+	jne	.LBB0_1
+# BB#4:
+	ret
+.Ltmp0:
+	.size	init_array, .Ltmp0-init_array
+
+	.globl	print_array
+	.align	16, 0x90
+	.type	print_array, at function
+print_array:                            # @print_array
+# BB#0:
+	pushq	%r14
+	pushq	%rbx
+	pushq	%rax
+	movq	$-9437184, %rbx         # imm = 0xFFFFFFFFFF700000
+	.align	16, 0x90
+.LBB1_1:                                # %.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB1_2 Depth 2
+	xorl	%r14d, %r14d
+	movq	stdout(%rip), %rdi
+	.align	16, 0x90
+.LBB1_2:                                #   Parent Loop BB1_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movss	C+9437184(%rbx,%r14,4), %xmm0
+	cvtss2sd	%xmm0, %xmm0
+	movl	$.L.str, %esi
+	movb	$1, %al
+	callq	fprintf
+	movslq	%r14d, %rax
+	imulq	$1717986919, %rax, %rcx # imm = 0x66666667
+	movq	%rcx, %rdx
+	shrq	$63, %rdx
+	sarq	$37, %rcx
+	addl	%edx, %ecx
+	imull	$80, %ecx, %ecx
+	subl	%ecx, %eax
+	cmpl	$79, %eax
+	jne	.LBB1_4
+# BB#3:                                 #   in Loop: Header=BB1_2 Depth=2
+	movq	stdout(%rip), %rsi
+	movl	$10, %edi
+	callq	fputc
+.LBB1_4:                                #   in Loop: Header=BB1_2 Depth=2
+	incq	%r14
+	movq	stdout(%rip), %rsi
+	cmpq	$1536, %r14             # imm = 0x600
+	movq	%rsi, %rdi
+	jne	.LBB1_2
+# BB#5:                                 #   in Loop: Header=BB1_1 Depth=1
+	movl	$10, %edi
+	callq	fputc
+	addq	$6144, %rbx             # imm = 0x1800
+	jne	.LBB1_1
+# BB#6:
+	addq	$8, %rsp
+	popq	%rbx
+	popq	%r14
+	ret
+.Ltmp1:
+	.size	print_array, .Ltmp1-print_array
+
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI2_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.globl	main
+	.align	16, 0x90
+	.type	main, at function
+main:                                   # @main
+# BB#0:
+	xorl	%eax, %eax
+	movsd	.LCPI2_0(%rip), %xmm0
+	movq	%rax, %rcx
+	.align	16, 0x90
+.LBB2_1:                                # %.preheader.i
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_2 Depth 2
+	movq	$-1536, %rdx            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB2_2:                                #   Parent Loop BB2_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movl	%esi, %edi
+	sarl	$31, %edi
+	shrl	$22, %edi
+	addl	%esi, %edi
+	andl	$-1024, %edi            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%edi
+	leal	1(%rsi,%rdi), %edi
+	cvtsi2sd	%edi, %xmm1
+	mulsd	%xmm0, %xmm1
+	cvtsd2ss	%xmm1, %xmm1
+	movss	%xmm1, A+6144(%rax,%rdx,4)
+	movss	%xmm1, B+6144(%rax,%rdx,4)
+	addl	%ecx, %esi
+	incq	%rdx
+	jne	.LBB2_2
+# BB#3:                                 #   in Loop: Header=BB2_1 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	incq	%rcx
+	xorl	%edx, %edx
+	cmpq	$1536, %rcx             # imm = 0x600
+	jne	.LBB2_1
+	.align	16, 0x90
+.LBB2_4:                                # %.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_5 Depth 2
+                                        #       Child Loop BB2_6 Depth 3
+	xorl	%eax, %eax
+	xorl	%ecx, %ecx
+	.align	16, 0x90
+.LBB2_5:                                #   Parent Loop BB2_4 Depth=1
+                                        # =>  This Loop Header: Depth=2
+                                        #       Child Loop BB2_6 Depth 3
+	movl	$0, C(%rcx,%rdx)
+	leaq	B(%rcx), %rsi
+	pxor	%xmm0, %xmm0
+	movq	%rax, %rdi
+	.align	16, 0x90
+.LBB2_6:                                #   Parent Loop BB2_4 Depth=1
+                                        #     Parent Loop BB2_5 Depth=2
+                                        # =>    This Inner Loop Header: Depth=3
+	movss	A(%rdx,%rdi,4), %xmm1
+	mulss	(%rsi), %xmm1
+	addss	%xmm1, %xmm0
+	addq	$6144, %rsi             # imm = 0x1800
+	incq	%rdi
+	cmpq	$1536, %rdi             # imm = 0x600
+	jne	.LBB2_6
+# BB#7:                                 #   in Loop: Header=BB2_5 Depth=2
+	movss	%xmm0, C(%rcx,%rdx)
+	addq	$4, %rcx
+	cmpq	$6144, %rcx             # imm = 0x1800
+	jne	.LBB2_5
+# BB#8:                                 # %init_array.exit
+                                        #   in Loop: Header=BB2_4 Depth=1
+	addq	$6144, %rdx             # imm = 0x1800
+	cmpq	$9437184, %rdx          # imm = 0x900000
+	jne	.LBB2_4
+# BB#9:
+	xorl	%eax, %eax
+	ret
+.Ltmp2:
+	.size	main, .Ltmp2-main
+
+	.type	A, at object               # @A
+	.comm	A,9437184,16
+	.type	B, at object               # @B
+	.comm	B,9437184,16
+	.type	.L.str, at object          # @.str
+	.section	.rodata.str1.1,"aMS", at progbits,1
+.L.str:
+	.asciz	 "%lf "
+	.size	.L.str, 5
+
+	.type	C, at object               # @C
+	.comm	C,9437184,16
+
+	.section	".note.GNU-stack","", at progbits

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.exe
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled%2Bvector%2Bopenmp.exe?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.exe (added) and polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.exe Mon May  2 02:48:29 2011 differ

Propchange: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.exe
------------------------------------------------------------------------------
    svn:executable = *

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.ll
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled%2Bvector%2Bopenmp.ll?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.ll (added) and polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.ll Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.s
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled%2Bvector%2Bopenmp.s?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.s (added)
+++ polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector+openmp.s Mon May  2 02:48:29 2011
@@ -0,0 +1,628 @@
+	.file	"matmul.polly.interchanged+tiled+vector+openmp.ll"
+	.text
+	.globl	init_array
+	.align	16, 0x90
+	.type	init_array, at function
+init_array:                             # @init_array
+# BB#0:                                 # %pollyBB
+	pushq	%rbx
+	subq	$16, %rsp
+	movq	$A, (%rsp)
+	movq	$B, 8(%rsp)
+	movl	$init_array.omp_subfn, %edi
+	leaq	(%rsp), %rbx
+	xorl	%edx, %edx
+	xorl	%ecx, %ecx
+	movl	$1536, %r8d             # imm = 0x600
+	movl	$1, %r9d
+	movq	%rbx, %rsi
+	callq	GOMP_parallel_loop_runtime_start
+	movq	%rbx, %rdi
+	callq	init_array.omp_subfn
+	callq	GOMP_parallel_end
+	addq	$16, %rsp
+	popq	%rbx
+	ret
+.Ltmp0:
+	.size	init_array, .Ltmp0-init_array
+
+	.globl	print_array
+	.align	16, 0x90
+	.type	print_array, at function
+print_array:                            # @print_array
+# BB#0:
+	pushq	%r14
+	pushq	%rbx
+	pushq	%rax
+	movq	$-9437184, %rbx         # imm = 0xFFFFFFFFFF700000
+	.align	16, 0x90
+.LBB1_1:                                # %.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB1_2 Depth 2
+	xorl	%r14d, %r14d
+	movq	stdout(%rip), %rdi
+	.align	16, 0x90
+.LBB1_2:                                #   Parent Loop BB1_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movss	C+9437184(%rbx,%r14,4), %xmm0
+	cvtss2sd	%xmm0, %xmm0
+	movl	$.L.str, %esi
+	movb	$1, %al
+	callq	fprintf
+	movslq	%r14d, %rax
+	imulq	$1717986919, %rax, %rcx # imm = 0x66666667
+	movq	%rcx, %rdx
+	shrq	$63, %rdx
+	sarq	$37, %rcx
+	addl	%edx, %ecx
+	imull	$80, %ecx, %ecx
+	subl	%ecx, %eax
+	cmpl	$79, %eax
+	jne	.LBB1_4
+# BB#3:                                 #   in Loop: Header=BB1_2 Depth=2
+	movq	stdout(%rip), %rsi
+	movl	$10, %edi
+	callq	fputc
+.LBB1_4:                                #   in Loop: Header=BB1_2 Depth=2
+	incq	%r14
+	movq	stdout(%rip), %rsi
+	cmpq	$1536, %r14             # imm = 0x600
+	movq	%rsi, %rdi
+	jne	.LBB1_2
+# BB#5:                                 #   in Loop: Header=BB1_1 Depth=1
+	movl	$10, %edi
+	callq	fputc
+	addq	$6144, %rbx             # imm = 0x1800
+	jne	.LBB1_1
+# BB#6:
+	addq	$8, %rsp
+	popq	%rbx
+	popq	%r14
+	ret
+.Ltmp1:
+	.size	print_array, .Ltmp1-print_array
+
+	.globl	main
+	.align	16, 0x90
+	.type	main, at function
+main:                                   # @main
+# BB#0:                                 # %pollyBB
+	pushq	%rbp
+	movq	%rsp, %rbp
+	pushq	%r15
+	pushq	%r14
+	pushq	%r13
+	pushq	%r12
+	pushq	%rbx
+	subq	$56, %rsp
+	movq	$A, -72(%rbp)
+	movq	$B, -64(%rbp)
+	movl	$init_array.omp_subfn, %edi
+	leaq	-72(%rbp), %rbx
+	movq	%rbx, %rsi
+	xorl	%edx, %edx
+	xorl	%ecx, %ecx
+	movl	$1536, %r8d             # imm = 0x600
+	movl	$1, %r9d
+	callq	GOMP_parallel_loop_runtime_start
+	movq	%rbx, %rdi
+	callq	init_array.omp_subfn
+	callq	GOMP_parallel_end
+	movl	$main.omp_subfn, %edi
+	leaq	-96(%rbp), %rsi
+	movq	$C, -96(%rbp)
+	movq	$A, -88(%rbp)
+	movq	$B, -80(%rbp)
+	xorl	%edx, %edx
+	xorl	%ecx, %ecx
+	movl	$1536, %r8d             # imm = 0x600
+	movl	$1, %r9d
+	callq	GOMP_parallel_loop_runtime_start
+	leaq	-48(%rbp), %rdi
+	leaq	-56(%rbp), %rsi
+	callq	GOMP_loop_runtime_next
+	testb	$1, %al
+	je	.LBB2_6
+# BB#1:
+	leaq	-48(%rbp), %rbx
+	leaq	-56(%rbp), %r14
+	.align	16, 0x90
+.LBB2_3:                                # %omp.loadIVBounds.i
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_5 Depth 2
+	movq	-56(%rbp), %r15
+	decq	%r15
+	movq	-48(%rbp), %r12
+	cmpq	%r15, %r12
+	jg	.LBB2_2
+# BB#4:                                 # %polly.loop_header2.preheader.lr.ph.i
+                                        #   in Loop: Header=BB2_3 Depth=1
+	leaq	(%r12,%r12,2), %rax
+	shlq	$11, %rax
+	leaq	C(%rax), %r13
+	.align	16, 0x90
+.LBB2_5:                                # %polly.loop_header2.preheader.i
+                                        #   Parent Loop BB2_3 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movq	%r13, %rdi
+	xorl	%esi, %esi
+	movl	$6144, %edx             # imm = 0x1800
+	callq	memset
+	addq	$6144, %r13             # imm = 0x1800
+	incq	%r12
+	cmpq	%r15, %r12
+	jle	.LBB2_5
+.LBB2_2:                                # %omp.checkNext.loopexit.i
+                                        #   in Loop: Header=BB2_3 Depth=1
+	movq	%rbx, %rdi
+	movq	%r14, %rsi
+	callq	GOMP_loop_runtime_next
+	testb	$1, %al
+	jne	.LBB2_3
+.LBB2_6:                                # %main.omp_subfn.exit
+	callq	GOMP_loop_end_nowait
+	callq	GOMP_parallel_end
+	movq	%rsp, %rax
+	leaq	-32(%rax), %rbx
+	movl	$main.omp_subfn1, %edi
+	xorl	%ecx, %ecx
+	movl	$1536, %r8d             # imm = 0x600
+	movl	$64, %r9d
+	movq	%rbx, %rsp
+	movq	$C, -32(%rax)
+	movq	$A, -24(%rax)
+	movq	$B, -16(%rax)
+	movq	%rbx, %rsi
+	xorl	%edx, %edx
+	callq	GOMP_parallel_loop_runtime_start
+	movq	%rbx, %rdi
+	callq	main.omp_subfn1
+	callq	GOMP_parallel_end
+	xorl	%eax, %eax
+	leaq	-40(%rbp), %rsp
+	popq	%rbx
+	popq	%r12
+	popq	%r13
+	popq	%r14
+	popq	%r15
+	popq	%rbp
+	ret
+.Ltmp2:
+	.size	main, .Ltmp2-main
+
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI3_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.align	16, 0x90
+	.type	init_array.omp_subfn, at function
+init_array.omp_subfn:                   # @init_array.omp_subfn
+.Leh_func_begin3:
+.Ltmp6:
+	.cfi_startproc
+# BB#0:                                 # %omp.setup
+	pushq	%r14
+.Ltmp7:
+	.cfi_def_cfa_offset 16
+	pushq	%rbx
+.Ltmp8:
+	.cfi_def_cfa_offset 24
+	subq	$24, %rsp
+.Ltmp9:
+	.cfi_def_cfa_offset 48
+.Ltmp10:
+	.cfi_offset 3, -24
+.Ltmp11:
+	.cfi_offset 14, -16
+	leaq	16(%rsp), %rdi
+	leaq	8(%rsp), %rsi
+	callq	GOMP_loop_runtime_next
+	testb	$1, %al
+	je	.LBB3_2
+# BB#1:
+	leaq	16(%rsp), %rbx
+	leaq	8(%rsp), %r14
+	jmp	.LBB3_4
+.LBB3_2:                                # %omp.exit
+	callq	GOMP_loop_end_nowait
+	addq	$24, %rsp
+	popq	%rbx
+	popq	%r14
+	ret
+	.align	16, 0x90
+.LBB3_3:                                # %omp.checkNext.loopexit
+                                        #   in Loop: Header=BB3_4 Depth=1
+	movq	%rbx, %rdi
+	movq	%r14, %rsi
+	callq	GOMP_loop_runtime_next
+	testb	$1, %al
+	je	.LBB3_2
+.LBB3_4:                                # %omp.loadIVBounds
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB3_7 Depth 2
+                                        #       Child Loop BB3_8 Depth 3
+	movq	8(%rsp), %rax
+	decq	%rax
+	movq	16(%rsp), %rcx
+	cmpq	%rax, %rcx
+	jg	.LBB3_3
+# BB#5:                                 # %polly.loop_header2.preheader.lr.ph
+                                        #   in Loop: Header=BB3_4 Depth=1
+	movq	%rcx, %rdx
+	shlq	$11, %rdx
+	leaq	(%rdx,%rdx,2), %rdx
+	jmp	.LBB3_7
+	.align	16, 0x90
+.LBB3_6:                                # %polly.loop_header.loopexit
+                                        #   in Loop: Header=BB3_7 Depth=2
+	addq	$6144, %rdx             # imm = 0x1800
+	incq	%rcx
+	cmpq	%rax, %rcx
+	jg	.LBB3_3
+.LBB3_7:                                # %polly.loop_header2.preheader
+                                        #   Parent Loop BB3_4 Depth=1
+                                        # =>  This Loop Header: Depth=2
+                                        #       Child Loop BB3_8 Depth 3
+	movq	$-1536, %rsi            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%edi, %edi
+	.align	16, 0x90
+.LBB3_8:                                # %polly.loop_body3
+                                        #   Parent Loop BB3_4 Depth=1
+                                        #     Parent Loop BB3_7 Depth=2
+                                        # =>    This Inner Loop Header: Depth=3
+	movl	%edi, %r8d
+	sarl	$31, %r8d
+	shrl	$22, %r8d
+	addl	%edi, %r8d
+	andl	$-1024, %r8d            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%r8d
+	leal	1(%rdi,%r8), %r8d
+	cvtsi2sd	%r8d, %xmm0
+	mulsd	.LCPI3_0(%rip), %xmm0
+	cvtsd2ss	%xmm0, %xmm0
+	movss	%xmm0, A+6144(%rdx,%rsi,4)
+	movss	%xmm0, B+6144(%rdx,%rsi,4)
+	addl	%ecx, %edi
+	incq	%rsi
+	jne	.LBB3_8
+	jmp	.LBB3_6
+.Ltmp12:
+	.size	init_array.omp_subfn, .Ltmp12-init_array.omp_subfn
+.Ltmp13:
+	.cfi_endproc
+.Leh_func_end3:
+
+	.align	16, 0x90
+	.type	main.omp_subfn, at function
+main.omp_subfn:                         # @main.omp_subfn
+.Leh_func_begin4:
+.Ltmp20:
+	.cfi_startproc
+# BB#0:                                 # %omp.setup
+	pushq	%r15
+.Ltmp21:
+	.cfi_def_cfa_offset 16
+	pushq	%r14
+.Ltmp22:
+	.cfi_def_cfa_offset 24
+	pushq	%r13
+.Ltmp23:
+	.cfi_def_cfa_offset 32
+	pushq	%r12
+.Ltmp24:
+	.cfi_def_cfa_offset 40
+	pushq	%rbx
+.Ltmp25:
+	.cfi_def_cfa_offset 48
+	subq	$16, %rsp
+.Ltmp26:
+	.cfi_def_cfa_offset 64
+.Ltmp27:
+	.cfi_offset 3, -48
+.Ltmp28:
+	.cfi_offset 12, -40
+.Ltmp29:
+	.cfi_offset 13, -32
+.Ltmp30:
+	.cfi_offset 14, -24
+.Ltmp31:
+	.cfi_offset 15, -16
+	leaq	8(%rsp), %rdi
+	leaq	(%rsp), %rsi
+	callq	GOMP_loop_runtime_next
+	testb	$1, %al
+	je	.LBB4_2
+# BB#1:
+	leaq	8(%rsp), %rbx
+	leaq	(%rsp), %r14
+	jmp	.LBB4_4
+.LBB4_2:                                # %omp.exit
+	callq	GOMP_loop_end_nowait
+	addq	$16, %rsp
+	popq	%rbx
+	popq	%r12
+	popq	%r13
+	popq	%r14
+	popq	%r15
+	ret
+	.align	16, 0x90
+.LBB4_3:                                # %omp.checkNext.loopexit
+                                        #   in Loop: Header=BB4_4 Depth=1
+	movq	%rbx, %rdi
+	movq	%r14, %rsi
+	callq	GOMP_loop_runtime_next
+	testb	$1, %al
+	je	.LBB4_2
+.LBB4_4:                                # %omp.loadIVBounds
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB4_6 Depth 2
+	movq	(%rsp), %r15
+	decq	%r15
+	movq	8(%rsp), %r12
+	cmpq	%r15, %r12
+	jg	.LBB4_3
+# BB#5:                                 # %polly.loop_header2.preheader.lr.ph
+                                        #   in Loop: Header=BB4_4 Depth=1
+	leaq	(%r12,%r12,2), %rax
+	shlq	$11, %rax
+	leaq	C(%rax), %r13
+	.align	16, 0x90
+.LBB4_6:                                # %polly.loop_header2.preheader
+                                        #   Parent Loop BB4_4 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movq	%r13, %rdi
+	xorl	%esi, %esi
+	movl	$6144, %edx             # imm = 0x1800
+	callq	memset
+	addq	$6144, %r13             # imm = 0x1800
+	incq	%r12
+	cmpq	%r15, %r12
+	jle	.LBB4_6
+	jmp	.LBB4_3
+.Ltmp32:
+	.size	main.omp_subfn, .Ltmp32-main.omp_subfn
+.Ltmp33:
+	.cfi_endproc
+.Leh_func_end4:
+
+	.align	16, 0x90
+	.type	main.omp_subfn1, at function
+main.omp_subfn1:                        # @main.omp_subfn1
+.Leh_func_begin5:
+.Ltmp41:
+	.cfi_startproc
+# BB#0:                                 # %omp.setup
+	pushq	%rbp
+.Ltmp42:
+	.cfi_def_cfa_offset 16
+	pushq	%r15
+.Ltmp43:
+	.cfi_def_cfa_offset 24
+	pushq	%r14
+.Ltmp44:
+	.cfi_def_cfa_offset 32
+	pushq	%r13
+.Ltmp45:
+	.cfi_def_cfa_offset 40
+	pushq	%r12
+.Ltmp46:
+	.cfi_def_cfa_offset 48
+	pushq	%rbx
+.Ltmp47:
+	.cfi_def_cfa_offset 56
+	subq	$40, %rsp
+.Ltmp48:
+	.cfi_def_cfa_offset 96
+.Ltmp49:
+	.cfi_offset 3, -56
+.Ltmp50:
+	.cfi_offset 12, -48
+.Ltmp51:
+	.cfi_offset 13, -40
+.Ltmp52:
+	.cfi_offset 14, -32
+.Ltmp53:
+	.cfi_offset 15, -24
+.Ltmp54:
+	.cfi_offset 6, -16
+	leaq	32(%rsp), %rdi
+	leaq	24(%rsp), %rsi
+	jmp	.LBB5_1
+	.align	16, 0x90
+.LBB5_4:                                # %omp.loadIVBounds
+                                        #   in Loop: Header=BB5_1 Depth=1
+	movq	24(%rsp), %rax
+	decq	%rax
+	movq	%rax, (%rsp)            # 8-byte Spill
+	movq	32(%rsp), %rcx
+	cmpq	%rax, %rcx
+	jg	.LBB5_3
+# BB#5:                                 # %polly.loop_header2.preheader.lr.ph
+                                        #   in Loop: Header=BB5_1 Depth=1
+	leaq	(%rcx,%rcx,2), %rax
+	movq	%rcx, %rdx
+	shlq	$9, %rdx
+	leaq	(%rdx,%rdx,2), %rdx
+	movq	%rdx, 16(%rsp)          # 8-byte Spill
+	shlq	$11, %rax
+	leaq	A(%rax), %rax
+	movq	%rax, 8(%rsp)           # 8-byte Spill
+	jmp	.LBB5_7
+	.align	16, 0x90
+.LBB5_6:                                # %polly.loop_header.loopexit
+                                        #   in Loop: Header=BB5_7 Depth=2
+	addq	$98304, 16(%rsp)        # 8-byte Folded Spill
+                                        # imm = 0x18000
+	addq	$393216, 8(%rsp)        # 8-byte Folded Spill
+                                        # imm = 0x60000
+	addq	$64, %rcx
+	cmpq	(%rsp), %rcx            # 8-byte Folded Reload
+	jg	.LBB5_3
+.LBB5_7:                                # %polly.loop_header2.preheader
+                                        #   Parent Loop BB5_1 Depth=1
+                                        # =>  This Loop Header: Depth=2
+                                        #       Child Loop BB5_9 Depth 3
+                                        #         Child Loop BB5_11 Depth 4
+                                        #           Child Loop BB5_14 Depth 5
+                                        #             Child Loop BB5_18 Depth 6
+                                        #               Child Loop BB5_19 Depth 7
+	leaq	63(%rcx), %rax
+	xorl	%edx, %edx
+	jmp	.LBB5_9
+	.align	16, 0x90
+.LBB5_8:                                # %polly.loop_header2.loopexit
+                                        #   in Loop: Header=BB5_9 Depth=3
+	addq	$64, %rdx
+	cmpq	$1536, %rdx             # imm = 0x600
+	je	.LBB5_6
+.LBB5_9:                                # %polly.loop_header7.preheader
+                                        #   Parent Loop BB5_1 Depth=1
+                                        #     Parent Loop BB5_7 Depth=2
+                                        # =>    This Loop Header: Depth=3
+                                        #         Child Loop BB5_11 Depth 4
+                                        #           Child Loop BB5_14 Depth 5
+                                        #             Child Loop BB5_18 Depth 6
+                                        #               Child Loop BB5_19 Depth 7
+	movq	16(%rsp), %rsi          # 8-byte Reload
+	leaq	(%rsi,%rdx), %rsi
+	leaq	63(%rdx), %rdi
+	xorl	%r8d, %r8d
+	movq	8(%rsp), %r9            # 8-byte Reload
+	movq	%rdx, %r10
+	jmp	.LBB5_11
+	.align	16, 0x90
+.LBB5_10:                               # %polly.loop_header7.loopexit
+                                        #   in Loop: Header=BB5_11 Depth=4
+	addq	$256, %r9               # imm = 0x100
+	addq	$98304, %r10            # imm = 0x18000
+	addq	$64, %r8
+	cmpq	$1536, %r8              # imm = 0x600
+	je	.LBB5_8
+.LBB5_11:                               # %polly.loop_body8
+                                        #   Parent Loop BB5_1 Depth=1
+                                        #     Parent Loop BB5_7 Depth=2
+                                        #       Parent Loop BB5_9 Depth=3
+                                        # =>      This Loop Header: Depth=4
+                                        #           Child Loop BB5_14 Depth 5
+                                        #             Child Loop BB5_18 Depth 6
+                                        #               Child Loop BB5_19 Depth 7
+	movabsq	$9223372036854775744, %r11 # imm = 0x7FFFFFFFFFFFFFC0
+	cmpq	%r11, %rcx
+	jg	.LBB5_10
+# BB#12:                                # %polly.loop_body13.lr.ph
+                                        #   in Loop: Header=BB5_11 Depth=4
+	leaq	63(%r8), %r11
+	movq	%rcx, %rbx
+	movq	%rsi, %r14
+	movq	%r9, %r15
+	jmp	.LBB5_14
+	.align	16, 0x90
+.LBB5_13:                               # %polly.loop_header12.loopexit
+                                        #   in Loop: Header=BB5_14 Depth=5
+	addq	$1536, %r14             # imm = 0x600
+	addq	$6144, %r15             # imm = 0x1800
+	incq	%rbx
+	cmpq	%rax, %rbx
+	jg	.LBB5_10
+.LBB5_14:                               # %polly.loop_body13
+                                        #   Parent Loop BB5_1 Depth=1
+                                        #     Parent Loop BB5_7 Depth=2
+                                        #       Parent Loop BB5_9 Depth=3
+                                        #         Parent Loop BB5_11 Depth=4
+                                        # =>        This Loop Header: Depth=5
+                                        #             Child Loop BB5_18 Depth 6
+                                        #               Child Loop BB5_19 Depth 7
+	cmpq	%r11, %r8
+	jg	.LBB5_13
+# BB#15:                                # %polly.loop_body13
+                                        #   in Loop: Header=BB5_14 Depth=5
+	cmpq	%rdi, %rdx
+	jg	.LBB5_13
+# BB#16:                                # %polly.loop_body23.lr.ph.preheader
+                                        #   in Loop: Header=BB5_14 Depth=5
+	xorl	%r12d, %r12d
+	movq	%r10, %r13
+	jmp	.LBB5_18
+	.align	16, 0x90
+.LBB5_17:                               # %polly.loop_header17.loopexit
+                                        #   in Loop: Header=BB5_18 Depth=6
+	addq	$1536, %r13             # imm = 0x600
+	incq	%r12
+	cmpq	$64, %r12
+	je	.LBB5_13
+.LBB5_18:                               # %polly.loop_body23.lr.ph
+                                        #   Parent Loop BB5_1 Depth=1
+                                        #     Parent Loop BB5_7 Depth=2
+                                        #       Parent Loop BB5_9 Depth=3
+                                        #         Parent Loop BB5_11 Depth=4
+                                        #           Parent Loop BB5_14 Depth=5
+                                        # =>          This Loop Header: Depth=6
+                                        #               Child Loop BB5_19 Depth 7
+	movss	(%r15,%r12,4), %xmm0
+	pshufd	$0, %xmm0, %xmm0        # xmm0 = xmm0[0,0,0,0]
+	xorl	%ebp, %ebp
+	.align	16, 0x90
+.LBB5_19:                               # %polly.loop_body23
+                                        #   Parent Loop BB5_1 Depth=1
+                                        #     Parent Loop BB5_7 Depth=2
+                                        #       Parent Loop BB5_9 Depth=3
+                                        #         Parent Loop BB5_11 Depth=4
+                                        #           Parent Loop BB5_14 Depth=5
+                                        #             Parent Loop BB5_18 Depth=6
+                                        # =>            This Inner Loop Header: Depth=7
+	movaps	B(%rbp,%r13,4), %xmm1
+	mulps	%xmm0, %xmm1
+	addps	C(%rbp,%r14,4), %xmm1
+	movaps	%xmm1, C(%rbp,%r14,4)
+	addq	$16, %rbp
+	cmpq	$256, %rbp              # imm = 0x100
+	jne	.LBB5_19
+	jmp	.LBB5_17
+.LBB5_3:                                # %omp.checkNext.loopexit
+                                        #   in Loop: Header=BB5_1 Depth=1
+	leaq	32(%rsp), %rax
+	movq	%rax, %rdi
+	leaq	24(%rsp), %rax
+	movq	%rax, %rsi
+.LBB5_1:                                # %omp.setup
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB5_7 Depth 2
+                                        #       Child Loop BB5_9 Depth 3
+                                        #         Child Loop BB5_11 Depth 4
+                                        #           Child Loop BB5_14 Depth 5
+                                        #             Child Loop BB5_18 Depth 6
+                                        #               Child Loop BB5_19 Depth 7
+	callq	GOMP_loop_runtime_next
+	testb	$1, %al
+	jne	.LBB5_4
+# BB#2:                                 # %omp.exit
+	callq	GOMP_loop_end_nowait
+	addq	$40, %rsp
+	popq	%rbx
+	popq	%r12
+	popq	%r13
+	popq	%r14
+	popq	%r15
+	popq	%rbp
+	ret
+.Ltmp55:
+	.size	main.omp_subfn1, .Ltmp55-main.omp_subfn1
+.Ltmp56:
+	.cfi_endproc
+.Leh_func_end5:
+
+	.type	A, at object               # @A
+	.comm	A,9437184,16
+	.type	B, at object               # @B
+	.comm	B,9437184,16
+	.type	.L.str, at object          # @.str
+	.section	.rodata.str1.1,"aMS", at progbits,1
+.L.str:
+	.asciz	 "%lf "
+	.size	.L.str, 5
+
+	.type	C, at object               # @C
+	.comm	C,9437184,16
+
+	.section	".note.GNU-stack","", at progbits

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.exe
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled%2Bvector.exe?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.exe (added) and polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.exe Mon May  2 02:48:29 2011 differ

Propchange: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.exe
------------------------------------------------------------------------------
    svn:executable = *

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.ll
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled%2Bvector.ll?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.ll (added) and polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.ll Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.s
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled%2Bvector.s?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.s (added)
+++ polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled+vector.s Mon May  2 02:48:29 2011
@@ -0,0 +1,318 @@
+	.file	"matmul.polly.interchanged+tiled+vector.ll"
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI0_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.globl	init_array
+	.align	16, 0x90
+	.type	init_array, at function
+init_array:                             # @init_array
+# BB#0:                                 # %pollyBB
+	xorl	%eax, %eax
+	movsd	.LCPI0_0(%rip), %xmm0
+	movq	%rax, %rcx
+	.align	16, 0x90
+.LBB0_2:                                # %polly.loop_header1.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB0_3 Depth 2
+	movq	$-1536, %rdx            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB0_3:                                # %polly.loop_body2
+                                        #   Parent Loop BB0_2 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movl	%esi, %edi
+	sarl	$31, %edi
+	shrl	$22, %edi
+	addl	%esi, %edi
+	andl	$-1024, %edi            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%edi
+	leal	1(%rsi,%rdi), %edi
+	cvtsi2sd	%edi, %xmm1
+	mulsd	%xmm0, %xmm1
+	cvtsd2ss	%xmm1, %xmm1
+	movss	%xmm1, A+6144(%rax,%rdx,4)
+	movss	%xmm1, B+6144(%rax,%rdx,4)
+	addl	%ecx, %esi
+	incq	%rdx
+	jne	.LBB0_3
+# BB#1:                                 # %polly.loop_header.loopexit
+                                        #   in Loop: Header=BB0_2 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	incq	%rcx
+	cmpq	$1536, %rcx             # imm = 0x600
+	jne	.LBB0_2
+# BB#4:                                 # %polly.after_loop
+	ret
+.Ltmp0:
+	.size	init_array, .Ltmp0-init_array
+
+	.globl	print_array
+	.align	16, 0x90
+	.type	print_array, at function
+print_array:                            # @print_array
+# BB#0:
+	pushq	%r14
+	pushq	%rbx
+	pushq	%rax
+	movq	$-9437184, %rbx         # imm = 0xFFFFFFFFFF700000
+	.align	16, 0x90
+.LBB1_1:                                # %.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB1_2 Depth 2
+	xorl	%r14d, %r14d
+	movq	stdout(%rip), %rdi
+	.align	16, 0x90
+.LBB1_2:                                #   Parent Loop BB1_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movss	C+9437184(%rbx,%r14,4), %xmm0
+	cvtss2sd	%xmm0, %xmm0
+	movl	$.L.str, %esi
+	movb	$1, %al
+	callq	fprintf
+	movslq	%r14d, %rax
+	imulq	$1717986919, %rax, %rcx # imm = 0x66666667
+	movq	%rcx, %rdx
+	shrq	$63, %rdx
+	sarq	$37, %rcx
+	addl	%edx, %ecx
+	imull	$80, %ecx, %ecx
+	subl	%ecx, %eax
+	cmpl	$79, %eax
+	jne	.LBB1_4
+# BB#3:                                 #   in Loop: Header=BB1_2 Depth=2
+	movq	stdout(%rip), %rsi
+	movl	$10, %edi
+	callq	fputc
+.LBB1_4:                                #   in Loop: Header=BB1_2 Depth=2
+	incq	%r14
+	movq	stdout(%rip), %rsi
+	cmpq	$1536, %r14             # imm = 0x600
+	movq	%rsi, %rdi
+	jne	.LBB1_2
+# BB#5:                                 #   in Loop: Header=BB1_1 Depth=1
+	movl	$10, %edi
+	callq	fputc
+	addq	$6144, %rbx             # imm = 0x1800
+	jne	.LBB1_1
+# BB#6:
+	addq	$8, %rsp
+	popq	%rbx
+	popq	%r14
+	ret
+.Ltmp1:
+	.size	print_array, .Ltmp1-print_array
+
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI2_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.globl	main
+	.align	16, 0x90
+	.type	main, at function
+main:                                   # @main
+# BB#0:                                 # %pollyBB
+	pushq	%rbp
+	pushq	%r15
+	pushq	%r14
+	pushq	%r13
+	pushq	%r12
+	pushq	%rbx
+	subq	$24, %rsp
+	xorl	%eax, %eax
+	movsd	.LCPI2_0(%rip), %xmm0
+	movq	%rax, %rcx
+	.align	16, 0x90
+.LBB2_1:                                # %polly.loop_header1.preheader.i
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_2 Depth 2
+	movq	$-1536, %rdx            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB2_2:                                # %polly.loop_body2.i
+                                        #   Parent Loop BB2_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movl	%esi, %edi
+	sarl	$31, %edi
+	shrl	$22, %edi
+	addl	%esi, %edi
+	andl	$-1024, %edi            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%edi
+	leal	1(%rsi,%rdi), %edi
+	cvtsi2sd	%edi, %xmm1
+	mulsd	%xmm0, %xmm1
+	cvtsd2ss	%xmm1, %xmm1
+	movss	%xmm1, A+6144(%rax,%rdx,4)
+	movss	%xmm1, B+6144(%rax,%rdx,4)
+	addl	%ecx, %esi
+	incq	%rdx
+	jne	.LBB2_2
+# BB#3:                                 # %polly.loop_header.loopexit.i
+                                        #   in Loop: Header=BB2_1 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	incq	%rcx
+	cmpq	$1536, %rcx             # imm = 0x600
+	jne	.LBB2_1
+# BB#4:                                 # %polly.loop_header.preheader
+	movl	$C, %edi
+	xorl	%esi, %esi
+	movl	$9437184, %edx          # imm = 0x900000
+	callq	memset
+	xorl	%eax, %eax
+	movq	%rax, 16(%rsp)          # 8-byte Spill
+	movq	%rax, (%rsp)            # 8-byte Spill
+	jmp	.LBB2_6
+	.align	16, 0x90
+.LBB2_5:                                # %polly.loop_header7.loopexit
+                                        #   in Loop: Header=BB2_6 Depth=1
+	addq	$393216, (%rsp)         # 8-byte Folded Spill
+                                        # imm = 0x60000
+	movq	16(%rsp), %rax          # 8-byte Reload
+	addq	$64, %rax
+	movq	%rax, 16(%rsp)          # 8-byte Spill
+	cmpq	$1536, %rax             # imm = 0x600
+	je	.LBB2_7
+.LBB2_6:                                # %polly.loop_header12.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_9 Depth 2
+                                        #       Child Loop BB2_11 Depth 3
+                                        #         Child Loop BB2_14 Depth 4
+                                        #           Child Loop BB2_18 Depth 5
+                                        #             Child Loop BB2_19 Depth 6
+	movq	16(%rsp), %rax          # 8-byte Reload
+	leaq	63(%rax), %rax
+	movq	(%rsp), %rcx            # 8-byte Reload
+	leaq	A(%rcx), %rdx
+	movq	%rdx, 8(%rsp)           # 8-byte Spill
+	xorl	%edx, %edx
+	jmp	.LBB2_9
+	.align	16, 0x90
+.LBB2_8:                                # %polly.loop_header12.loopexit
+                                        #   in Loop: Header=BB2_9 Depth=2
+	addq	$256, %rcx              # imm = 0x100
+	addq	$64, %rdx
+	cmpq	$1536, %rdx             # imm = 0x600
+	je	.LBB2_5
+.LBB2_9:                                # %polly.loop_header17.preheader
+                                        #   Parent Loop BB2_6 Depth=1
+                                        # =>  This Loop Header: Depth=2
+                                        #       Child Loop BB2_11 Depth 3
+                                        #         Child Loop BB2_14 Depth 4
+                                        #           Child Loop BB2_18 Depth 5
+                                        #             Child Loop BB2_19 Depth 6
+	leaq	63(%rdx), %rsi
+	xorl	%edi, %edi
+	movq	8(%rsp), %r8            # 8-byte Reload
+	movq	%rdx, %r9
+	jmp	.LBB2_11
+	.align	16, 0x90
+.LBB2_10:                               # %polly.loop_header17.loopexit
+                                        #   in Loop: Header=BB2_11 Depth=3
+	addq	$256, %r8               # imm = 0x100
+	addq	$98304, %r9             # imm = 0x18000
+	addq	$64, %rdi
+	cmpq	$1536, %rdi             # imm = 0x600
+	je	.LBB2_8
+.LBB2_11:                               # %polly.loop_body18
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        # =>    This Loop Header: Depth=3
+                                        #         Child Loop BB2_14 Depth 4
+                                        #           Child Loop BB2_18 Depth 5
+                                        #             Child Loop BB2_19 Depth 6
+	cmpq	%rax, 16(%rsp)          # 8-byte Folded Reload
+	jg	.LBB2_10
+# BB#12:                                # %polly.loop_body23.lr.ph
+                                        #   in Loop: Header=BB2_11 Depth=3
+	leaq	63(%rdi), %r10
+	xorl	%r11d, %r11d
+	jmp	.LBB2_14
+	.align	16, 0x90
+.LBB2_13:                               # %polly.loop_header22.loopexit
+                                        #   in Loop: Header=BB2_14 Depth=4
+	addq	$6144, %r11             # imm = 0x1800
+	cmpq	$393216, %r11           # imm = 0x60000
+	je	.LBB2_10
+.LBB2_14:                               # %polly.loop_body23
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        #       Parent Loop BB2_11 Depth=3
+                                        # =>      This Loop Header: Depth=4
+                                        #           Child Loop BB2_18 Depth 5
+                                        #             Child Loop BB2_19 Depth 6
+	cmpq	%r10, %rdi
+	jg	.LBB2_13
+# BB#15:                                # %polly.loop_body23
+                                        #   in Loop: Header=BB2_14 Depth=4
+	cmpq	%rsi, %rdx
+	jg	.LBB2_13
+# BB#16:                                # %polly.loop_body33.lr.ph.preheader
+                                        #   in Loop: Header=BB2_14 Depth=4
+	leaq	(%r8,%r11), %rbx
+	xorl	%r14d, %r14d
+	movq	%r9, %r15
+	movq	%r14, %r12
+	jmp	.LBB2_18
+	.align	16, 0x90
+.LBB2_17:                               # %polly.loop_header27.loopexit
+                                        #   in Loop: Header=BB2_18 Depth=5
+	addq	$1536, %r15             # imm = 0x600
+	incq	%r12
+	cmpq	$64, %r12
+	je	.LBB2_13
+.LBB2_18:                               # %polly.loop_body33.lr.ph
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        #       Parent Loop BB2_11 Depth=3
+                                        #         Parent Loop BB2_14 Depth=4
+                                        # =>        This Loop Header: Depth=5
+                                        #             Child Loop BB2_19 Depth 6
+	movss	(%rbx,%r12,4), %xmm0
+	pshufd	$0, %xmm0, %xmm0        # xmm0 = xmm0[0,0,0,0]
+	movq	%r14, %r13
+	.align	16, 0x90
+.LBB2_19:                               # %polly.loop_body33
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        #       Parent Loop BB2_11 Depth=3
+                                        #         Parent Loop BB2_14 Depth=4
+                                        #           Parent Loop BB2_18 Depth=5
+                                        # =>          This Inner Loop Header: Depth=6
+	movaps	B(%r13,%r15,4), %xmm1
+	mulps	%xmm0, %xmm1
+	leaq	(%r11,%r13), %rbp
+	addps	C(%rcx,%rbp), %xmm1
+	movaps	%xmm1, C(%rcx,%rbp)
+	addq	$16, %r13
+	cmpq	$256, %r13              # imm = 0x100
+	jne	.LBB2_19
+	jmp	.LBB2_17
+.LBB2_7:                                # %polly.after_loop9
+	xorl	%eax, %eax
+	addq	$24, %rsp
+	popq	%rbx
+	popq	%r12
+	popq	%r13
+	popq	%r14
+	popq	%r15
+	popq	%rbp
+	ret
+.Ltmp2:
+	.size	main, .Ltmp2-main
+
+	.type	A, at object               # @A
+	.comm	A,9437184,16
+	.type	B, at object               # @B
+	.comm	B,9437184,16
+	.type	.L.str, at object          # @.str
+	.section	.rodata.str1.1,"aMS", at progbits,1
+.L.str:
+	.asciz	 "%lf "
+	.size	.L.str, 5
+
+	.type	C, at object               # @C
+	.comm	C,9437184,16
+
+	.section	".note.GNU-stack","", at progbits

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.exe
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled.exe?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.exe (added) and polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.exe Mon May  2 02:48:29 2011 differ

Propchange: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.exe
------------------------------------------------------------------------------
    svn:executable = *

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.ll
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled.ll?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.ll (added) and polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.ll Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.s
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged%2Btiled.s?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.s (added)
+++ polly/trunk/www/experiments/matmul/matmul.polly.interchanged+tiled.s Mon May  2 02:48:29 2011
@@ -0,0 +1,323 @@
+	.file	"matmul.polly.interchanged+tiled.ll"
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI0_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.globl	init_array
+	.align	16, 0x90
+	.type	init_array, at function
+init_array:                             # @init_array
+# BB#0:                                 # %pollyBB
+	xorl	%eax, %eax
+	movsd	.LCPI0_0(%rip), %xmm0
+	movq	%rax, %rcx
+	.align	16, 0x90
+.LBB0_2:                                # %polly.loop_header1.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB0_3 Depth 2
+	movq	$-1536, %rdx            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB0_3:                                # %polly.loop_body2
+                                        #   Parent Loop BB0_2 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movl	%esi, %edi
+	sarl	$31, %edi
+	shrl	$22, %edi
+	addl	%esi, %edi
+	andl	$-1024, %edi            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%edi
+	leal	1(%rsi,%rdi), %edi
+	cvtsi2sd	%edi, %xmm1
+	mulsd	%xmm0, %xmm1
+	cvtsd2ss	%xmm1, %xmm1
+	movss	%xmm1, A+6144(%rax,%rdx,4)
+	movss	%xmm1, B+6144(%rax,%rdx,4)
+	addl	%ecx, %esi
+	incq	%rdx
+	jne	.LBB0_3
+# BB#1:                                 # %polly.loop_header.loopexit
+                                        #   in Loop: Header=BB0_2 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	incq	%rcx
+	cmpq	$1536, %rcx             # imm = 0x600
+	jne	.LBB0_2
+# BB#4:                                 # %polly.after_loop
+	ret
+.Ltmp0:
+	.size	init_array, .Ltmp0-init_array
+
+	.globl	print_array
+	.align	16, 0x90
+	.type	print_array, at function
+print_array:                            # @print_array
+# BB#0:
+	pushq	%r14
+	pushq	%rbx
+	pushq	%rax
+	movq	$-9437184, %rbx         # imm = 0xFFFFFFFFFF700000
+	.align	16, 0x90
+.LBB1_1:                                # %.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB1_2 Depth 2
+	xorl	%r14d, %r14d
+	movq	stdout(%rip), %rdi
+	.align	16, 0x90
+.LBB1_2:                                #   Parent Loop BB1_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movss	C+9437184(%rbx,%r14,4), %xmm0
+	cvtss2sd	%xmm0, %xmm0
+	movl	$.L.str, %esi
+	movb	$1, %al
+	callq	fprintf
+	movslq	%r14d, %rax
+	imulq	$1717986919, %rax, %rcx # imm = 0x66666667
+	movq	%rcx, %rdx
+	shrq	$63, %rdx
+	sarq	$37, %rcx
+	addl	%edx, %ecx
+	imull	$80, %ecx, %ecx
+	subl	%ecx, %eax
+	cmpl	$79, %eax
+	jne	.LBB1_4
+# BB#3:                                 #   in Loop: Header=BB1_2 Depth=2
+	movq	stdout(%rip), %rsi
+	movl	$10, %edi
+	callq	fputc
+.LBB1_4:                                #   in Loop: Header=BB1_2 Depth=2
+	incq	%r14
+	movq	stdout(%rip), %rsi
+	cmpq	$1536, %r14             # imm = 0x600
+	movq	%rsi, %rdi
+	jne	.LBB1_2
+# BB#5:                                 #   in Loop: Header=BB1_1 Depth=1
+	movl	$10, %edi
+	callq	fputc
+	addq	$6144, %rbx             # imm = 0x1800
+	jne	.LBB1_1
+# BB#6:
+	addq	$8, %rsp
+	popq	%rbx
+	popq	%r14
+	ret
+.Ltmp1:
+	.size	print_array, .Ltmp1-print_array
+
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI2_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.globl	main
+	.align	16, 0x90
+	.type	main, at function
+main:                                   # @main
+# BB#0:                                 # %pollyBB
+	pushq	%rbp
+	pushq	%r15
+	pushq	%r14
+	pushq	%r13
+	pushq	%r12
+	pushq	%rbx
+	subq	$40, %rsp
+	xorl	%eax, %eax
+	movsd	.LCPI2_0(%rip), %xmm0
+	movq	%rax, %rcx
+	.align	16, 0x90
+.LBB2_1:                                # %polly.loop_header1.preheader.i
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_2 Depth 2
+	movq	$-1536, %rdx            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB2_2:                                # %polly.loop_body2.i
+                                        #   Parent Loop BB2_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movl	%esi, %edi
+	sarl	$31, %edi
+	shrl	$22, %edi
+	addl	%esi, %edi
+	andl	$-1024, %edi            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%edi
+	leal	1(%rsi,%rdi), %edi
+	cvtsi2sd	%edi, %xmm1
+	mulsd	%xmm0, %xmm1
+	cvtsd2ss	%xmm1, %xmm1
+	movss	%xmm1, A+6144(%rax,%rdx,4)
+	movss	%xmm1, B+6144(%rax,%rdx,4)
+	addl	%ecx, %esi
+	incq	%rdx
+	jne	.LBB2_2
+# BB#3:                                 # %polly.loop_header.loopexit.i
+                                        #   in Loop: Header=BB2_1 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	incq	%rcx
+	cmpq	$1536, %rcx             # imm = 0x600
+	jne	.LBB2_1
+# BB#4:                                 # %polly.loop_header.preheader
+	movl	$C, %eax
+	movq	%rax, 8(%rsp)           # 8-byte Spill
+	xorl	%esi, %esi
+	movl	$9437184, %edx          # imm = 0x900000
+	movl	$C, %edi
+	callq	memset
+	movl	$A, %eax
+	movq	%rax, 16(%rsp)          # 8-byte Spill
+	movq	$0, 32(%rsp)            # 8-byte Folded Spill
+	jmp	.LBB2_6
+	.align	16, 0x90
+.LBB2_5:                                # %polly.loop_header7.loopexit
+                                        #   in Loop: Header=BB2_6 Depth=1
+	addq	$393216, 16(%rsp)       # 8-byte Folded Spill
+                                        # imm = 0x60000
+	addq	$393216, 8(%rsp)        # 8-byte Folded Spill
+                                        # imm = 0x60000
+	movq	32(%rsp), %rax          # 8-byte Reload
+	addq	$64, %rax
+	movq	%rax, 32(%rsp)          # 8-byte Spill
+	cmpq	$1536, %rax             # imm = 0x600
+	je	.LBB2_7
+.LBB2_6:                                # %polly.loop_header12.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_9 Depth 2
+                                        #       Child Loop BB2_11 Depth 3
+                                        #         Child Loop BB2_14 Depth 4
+                                        #           Child Loop BB2_18 Depth 5
+                                        #             Child Loop BB2_19 Depth 6
+	movq	32(%rsp), %rax          # 8-byte Reload
+	leaq	63(%rax), %rax
+	movl	$B, %ecx
+	movq	%rcx, 24(%rsp)          # 8-byte Spill
+	xorl	%ecx, %ecx
+	movq	8(%rsp), %rdx           # 8-byte Reload
+	jmp	.LBB2_9
+	.align	16, 0x90
+.LBB2_8:                                # %polly.loop_header12.loopexit
+                                        #   in Loop: Header=BB2_9 Depth=2
+	addq	$256, %rdx              # imm = 0x100
+	addq	$256, 24(%rsp)          # 8-byte Folded Spill
+                                        # imm = 0x100
+	addq	$64, %rcx
+	cmpq	$1536, %rcx             # imm = 0x600
+	je	.LBB2_5
+.LBB2_9:                                # %polly.loop_header17.preheader
+                                        #   Parent Loop BB2_6 Depth=1
+                                        # =>  This Loop Header: Depth=2
+                                        #       Child Loop BB2_11 Depth 3
+                                        #         Child Loop BB2_14 Depth 4
+                                        #           Child Loop BB2_18 Depth 5
+                                        #             Child Loop BB2_19 Depth 6
+	leaq	63(%rcx), %rsi
+	xorl	%edi, %edi
+	movq	16(%rsp), %r8           # 8-byte Reload
+	movq	24(%rsp), %r9           # 8-byte Reload
+	jmp	.LBB2_11
+	.align	16, 0x90
+.LBB2_10:                               # %polly.loop_header17.loopexit
+                                        #   in Loop: Header=BB2_11 Depth=3
+	addq	$256, %r8               # imm = 0x100
+	addq	$393216, %r9            # imm = 0x60000
+	addq	$64, %rdi
+	cmpq	$1536, %rdi             # imm = 0x600
+	je	.LBB2_8
+.LBB2_11:                               # %polly.loop_body18
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        # =>    This Loop Header: Depth=3
+                                        #         Child Loop BB2_14 Depth 4
+                                        #           Child Loop BB2_18 Depth 5
+                                        #             Child Loop BB2_19 Depth 6
+	cmpq	%rax, 32(%rsp)          # 8-byte Folded Reload
+	jg	.LBB2_10
+# BB#12:                                # %polly.loop_body23.lr.ph
+                                        #   in Loop: Header=BB2_11 Depth=3
+	leaq	63(%rdi), %r10
+	xorl	%r11d, %r11d
+	jmp	.LBB2_14
+	.align	16, 0x90
+.LBB2_13:                               # %polly.loop_header22.loopexit
+                                        #   in Loop: Header=BB2_14 Depth=4
+	addq	$6144, %r11             # imm = 0x1800
+	cmpq	$393216, %r11           # imm = 0x60000
+	je	.LBB2_10
+.LBB2_14:                               # %polly.loop_body23
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        #       Parent Loop BB2_11 Depth=3
+                                        # =>      This Loop Header: Depth=4
+                                        #           Child Loop BB2_18 Depth 5
+                                        #             Child Loop BB2_19 Depth 6
+	cmpq	%r10, %rdi
+	jg	.LBB2_13
+# BB#15:                                # %polly.loop_body23
+                                        #   in Loop: Header=BB2_14 Depth=4
+	cmpq	%rsi, %rcx
+	jg	.LBB2_13
+# BB#16:                                # %polly.loop_body33.lr.ph.preheader
+                                        #   in Loop: Header=BB2_14 Depth=4
+	leaq	(%rdx,%r11), %rbx
+	leaq	(%r8,%r11), %r14
+	xorl	%r15d, %r15d
+	movq	%r9, %r12
+	movq	%r15, %r13
+	jmp	.LBB2_18
+	.align	16, 0x90
+.LBB2_17:                               # %polly.loop_header27.loopexit
+                                        #   in Loop: Header=BB2_18 Depth=5
+	addq	$6144, %r12             # imm = 0x1800
+	incq	%r13
+	cmpq	$64, %r13
+	je	.LBB2_13
+.LBB2_18:                               # %polly.loop_body33.lr.ph
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        #       Parent Loop BB2_11 Depth=3
+                                        #         Parent Loop BB2_14 Depth=4
+                                        # =>        This Loop Header: Depth=5
+                                        #             Child Loop BB2_19 Depth 6
+	movss	(%r14,%r13,4), %xmm0
+	movq	%r15, %rbp
+	.align	16, 0x90
+.LBB2_19:                               # %polly.loop_body33
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        #       Parent Loop BB2_11 Depth=3
+                                        #         Parent Loop BB2_14 Depth=4
+                                        #           Parent Loop BB2_18 Depth=5
+                                        # =>          This Inner Loop Header: Depth=6
+	movss	(%r12,%rbp,4), %xmm1
+	mulss	%xmm0, %xmm1
+	addss	(%rbx,%rbp,4), %xmm1
+	movss	%xmm1, (%rbx,%rbp,4)
+	incq	%rbp
+	cmpq	$64, %rbp
+	jne	.LBB2_19
+	jmp	.LBB2_17
+.LBB2_7:                                # %polly.after_loop9
+	xorl	%eax, %eax
+	addq	$40, %rsp
+	popq	%rbx
+	popq	%r12
+	popq	%r13
+	popq	%r14
+	popq	%r15
+	popq	%rbp
+	ret
+.Ltmp2:
+	.size	main, .Ltmp2-main
+
+	.type	A, at object               # @A
+	.comm	A,9437184,16
+	.type	B, at object               # @B
+	.comm	B,9437184,16
+	.type	.L.str, at object          # @.str
+	.section	.rodata.str1.1,"aMS", at progbits,1
+.L.str:
+	.asciz	 "%lf "
+	.size	.L.str, 5
+
+	.type	C, at object               # @C
+	.comm	C,9437184,16
+
+	.section	".note.GNU-stack","", at progbits

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged.exe
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged.exe?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.polly.interchanged.exe (added) and polly/trunk/www/experiments/matmul/matmul.polly.interchanged.exe Mon May  2 02:48:29 2011 differ

Propchange: polly/trunk/www/experiments/matmul/matmul.polly.interchanged.exe
------------------------------------------------------------------------------
    svn:executable = *

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged.ll
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged.ll?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/matmul.polly.interchanged.ll (added) and polly/trunk/www/experiments/matmul/matmul.polly.interchanged.ll Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/matmul.polly.interchanged.s
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.polly.interchanged.s?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/matmul.polly.interchanged.s (added)
+++ polly/trunk/www/experiments/matmul/matmul.polly.interchanged.s Mon May  2 02:48:29 2011
@@ -0,0 +1,217 @@
+	.file	"matmul.polly.interchanged.ll"
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI0_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.globl	init_array
+	.align	16, 0x90
+	.type	init_array, at function
+init_array:                             # @init_array
+# BB#0:                                 # %pollyBB
+	xorl	%eax, %eax
+	movsd	.LCPI0_0(%rip), %xmm0
+	movq	%rax, %rcx
+	.align	16, 0x90
+.LBB0_2:                                # %polly.loop_header1.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB0_3 Depth 2
+	movq	$-1536, %rdx            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB0_3:                                # %polly.loop_body2
+                                        #   Parent Loop BB0_2 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movl	%esi, %edi
+	sarl	$31, %edi
+	shrl	$22, %edi
+	addl	%esi, %edi
+	andl	$-1024, %edi            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%edi
+	leal	1(%rsi,%rdi), %edi
+	cvtsi2sd	%edi, %xmm1
+	mulsd	%xmm0, %xmm1
+	cvtsd2ss	%xmm1, %xmm1
+	movss	%xmm1, A+6144(%rax,%rdx,4)
+	movss	%xmm1, B+6144(%rax,%rdx,4)
+	addl	%ecx, %esi
+	incq	%rdx
+	jne	.LBB0_3
+# BB#1:                                 # %polly.loop_header.loopexit
+                                        #   in Loop: Header=BB0_2 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	incq	%rcx
+	cmpq	$1536, %rcx             # imm = 0x600
+	jne	.LBB0_2
+# BB#4:                                 # %polly.after_loop
+	ret
+.Ltmp0:
+	.size	init_array, .Ltmp0-init_array
+
+	.globl	print_array
+	.align	16, 0x90
+	.type	print_array, at function
+print_array:                            # @print_array
+# BB#0:
+	pushq	%r14
+	pushq	%rbx
+	pushq	%rax
+	movq	$-9437184, %rbx         # imm = 0xFFFFFFFFFF700000
+	.align	16, 0x90
+.LBB1_1:                                # %.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB1_2 Depth 2
+	xorl	%r14d, %r14d
+	movq	stdout(%rip), %rdi
+	.align	16, 0x90
+.LBB1_2:                                #   Parent Loop BB1_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movss	C+9437184(%rbx,%r14,4), %xmm0
+	cvtss2sd	%xmm0, %xmm0
+	movl	$.L.str, %esi
+	movb	$1, %al
+	callq	fprintf
+	movslq	%r14d, %rax
+	imulq	$1717986919, %rax, %rcx # imm = 0x66666667
+	movq	%rcx, %rdx
+	shrq	$63, %rdx
+	sarq	$37, %rcx
+	addl	%edx, %ecx
+	imull	$80, %ecx, %ecx
+	subl	%ecx, %eax
+	cmpl	$79, %eax
+	jne	.LBB1_4
+# BB#3:                                 #   in Loop: Header=BB1_2 Depth=2
+	movq	stdout(%rip), %rsi
+	movl	$10, %edi
+	callq	fputc
+.LBB1_4:                                #   in Loop: Header=BB1_2 Depth=2
+	incq	%r14
+	movq	stdout(%rip), %rsi
+	cmpq	$1536, %r14             # imm = 0x600
+	movq	%rsi, %rdi
+	jne	.LBB1_2
+# BB#5:                                 #   in Loop: Header=BB1_1 Depth=1
+	movl	$10, %edi
+	callq	fputc
+	addq	$6144, %rbx             # imm = 0x1800
+	jne	.LBB1_1
+# BB#6:
+	addq	$8, %rsp
+	popq	%rbx
+	popq	%r14
+	ret
+.Ltmp1:
+	.size	print_array, .Ltmp1-print_array
+
+	.section	.rodata.cst8,"aM", at progbits,8
+	.align	8
+.LCPI2_0:
+	.quad	4602678819172646912     # double 5.000000e-01
+	.text
+	.globl	main
+	.align	16, 0x90
+	.type	main, at function
+main:                                   # @main
+# BB#0:                                 # %pollyBB
+	pushq	%rax
+	xorl	%eax, %eax
+	movsd	.LCPI2_0(%rip), %xmm0
+	movq	%rax, %rcx
+	.align	16, 0x90
+.LBB2_1:                                # %polly.loop_header1.preheader.i
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_2 Depth 2
+	movq	$-1536, %rdx            # imm = 0xFFFFFFFFFFFFFA00
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB2_2:                                # %polly.loop_body2.i
+                                        #   Parent Loop BB2_1 Depth=1
+                                        # =>  This Inner Loop Header: Depth=2
+	movl	%esi, %edi
+	sarl	$31, %edi
+	shrl	$22, %edi
+	addl	%esi, %edi
+	andl	$-1024, %edi            # imm = 0xFFFFFFFFFFFFFC00
+	negl	%edi
+	leal	1(%rsi,%rdi), %edi
+	cvtsi2sd	%edi, %xmm1
+	mulsd	%xmm0, %xmm1
+	cvtsd2ss	%xmm1, %xmm1
+	movss	%xmm1, A+6144(%rax,%rdx,4)
+	movss	%xmm1, B+6144(%rax,%rdx,4)
+	addl	%ecx, %esi
+	incq	%rdx
+	jne	.LBB2_2
+# BB#3:                                 # %polly.loop_header.loopexit.i
+                                        #   in Loop: Header=BB2_1 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	incq	%rcx
+	cmpq	$1536, %rcx             # imm = 0x600
+	jne	.LBB2_1
+# BB#4:                                 # %polly.loop_header.preheader
+	movl	$C, %edi
+	xorl	%esi, %esi
+	movl	$9437184, %edx          # imm = 0x900000
+	callq	memset
+	xorl	%eax, %eax
+	jmp	.LBB2_6
+	.align	16, 0x90
+.LBB2_5:                                # %polly.loop_header7.loopexit
+                                        #   in Loop: Header=BB2_6 Depth=1
+	addq	$6144, %rax             # imm = 0x1800
+	cmpq	$9437184, %rax          # imm = 0x900000
+	je	.LBB2_7
+.LBB2_6:                                # %polly.loop_header12.preheader
+                                        # =>This Loop Header: Depth=1
+                                        #     Child Loop BB2_9 Depth 2
+                                        #       Child Loop BB2_10 Depth 3
+	leaq	A(%rax), %rcx
+	movq	$-9437184, %rdx         # imm = 0xFFFFFFFFFF700000
+	jmp	.LBB2_9
+	.align	16, 0x90
+.LBB2_8:                                # %polly.loop_header12.loopexit
+                                        #   in Loop: Header=BB2_9 Depth=2
+	addq	$4, %rcx
+	addq	$6144, %rdx             # imm = 0x1800
+	je	.LBB2_5
+.LBB2_9:                                # %polly.loop_header17.preheader
+                                        #   Parent Loop BB2_6 Depth=1
+                                        # =>  This Loop Header: Depth=2
+                                        #       Child Loop BB2_10 Depth 3
+	movss	(%rcx), %xmm0
+	xorl	%esi, %esi
+	.align	16, 0x90
+.LBB2_10:                               # %polly.loop_body18
+                                        #   Parent Loop BB2_6 Depth=1
+                                        #     Parent Loop BB2_9 Depth=2
+                                        # =>    This Inner Loop Header: Depth=3
+	movss	B+9437184(%rdx,%rsi,4), %xmm1
+	mulss	%xmm0, %xmm1
+	addss	C(%rax,%rsi,4), %xmm1
+	movss	%xmm1, C(%rax,%rsi,4)
+	incq	%rsi
+	cmpq	$1536, %rsi             # imm = 0x600
+	jne	.LBB2_10
+	jmp	.LBB2_8
+.LBB2_7:                                # %polly.after_loop9
+	xorl	%eax, %eax
+	popq	%rdx
+	ret
+.Ltmp2:
+	.size	main, .Ltmp2-main
+
+	.type	A, at object               # @A
+	.comm	A,9437184,16
+	.type	B, at object               # @B
+	.comm	B,9437184,16
+	.type	.L.str, at object          # @.str
+	.section	.rodata.str1.1,"aMS", at progbits,1
+.L.str:
+	.asciz	 "%lf "
+	.size	.L.str, 5
+
+	.type	C, at object               # @C
+	.comm	C,9437184,16
+
+	.section	".note.GNU-stack","", at progbits

Added: polly/trunk/www/experiments/matmul/matmul.preopt.ll
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.preopt.ll?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/matmul.preopt.ll (added)
+++ polly/trunk/www/experiments/matmul/matmul.preopt.ll Mon May  2 02:48:29 2011
@@ -0,0 +1,180 @@
+; ModuleID = 'matmul.s'
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-unknown-linux-gnu"
+
+%struct._IO_FILE = type { i32, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, %struct._IO_marker*, %struct._IO_FILE*, i32, i32, i64, i16, i8, [1 x i8], i8*, i64, i8*, i8*, i8*, i8*, i64, i32, [20 x i8] }
+%struct._IO_marker = type { %struct._IO_marker*, %struct._IO_FILE*, i32 }
+
+ at A = common global [1536 x [1536 x float]] zeroinitializer, align 16
+ at B = common global [1536 x [1536 x float]] zeroinitializer, align 16
+ at stdout = external global %struct._IO_FILE*
+ at .str = private unnamed_addr constant [5 x i8] c"%lf \00"
+ at C = common global [1536 x [1536 x float]] zeroinitializer, align 16
+ at .str1 = private unnamed_addr constant [2 x i8] c"\0A\00"
+
+define void @init_array() nounwind {
+; <label>:0
+  br label %1
+
+; <label>:1                                       ; preds = %18, %0
+  %2 = phi i64 [ %indvar.next2, %18 ], [ 0, %0 ]
+  %exitcond5 = icmp ne i64 %2, 1536
+  br i1 %exitcond5, label %3, label %19
+
+; <label>:3                                       ; preds = %1
+  br label %4
+
+; <label>:4                                       ; preds = %16, %3
+  %indvar = phi i64 [ %indvar.next, %16 ], [ 0, %3 ]
+  %scevgep4 = getelementptr [1536 x [1536 x float]]* @A, i64 0, i64 %2, i64 %indvar
+  %scevgep = getelementptr [1536 x [1536 x float]]* @B, i64 0, i64 %2, i64 %indvar
+  %tmp = mul i64 %2, %indvar
+  %tmp3 = trunc i64 %tmp to i32
+  %exitcond = icmp ne i64 %indvar, 1536
+  br i1 %exitcond, label %5, label %17
+
+; <label>:5                                       ; preds = %4
+  %6 = srem i32 %tmp3, 1024
+  %7 = add nsw i32 1, %6
+  %8 = sitofp i32 %7 to double
+  %9 = fdiv double %8, 2.000000e+00
+  %10 = fptrunc double %9 to float
+  store float %10, float* %scevgep4
+  %11 = srem i32 %tmp3, 1024
+  %12 = add nsw i32 1, %11
+  %13 = sitofp i32 %12 to double
+  %14 = fdiv double %13, 2.000000e+00
+  %15 = fptrunc double %14 to float
+  store float %15, float* %scevgep
+  br label %16
+
+; <label>:16                                      ; preds = %5
+  %indvar.next = add i64 %indvar, 1
+  br label %4
+
+; <label>:17                                      ; preds = %4
+  br label %18
+
+; <label>:18                                      ; preds = %17
+  %indvar.next2 = add i64 %2, 1
+  br label %1
+
+; <label>:19                                      ; preds = %1
+  ret void
+}
+
+define void @print_array() nounwind {
+; <label>:0
+  br label %1
+
+; <label>:1                                       ; preds = %19, %0
+  %indvar1 = phi i64 [ %indvar.next2, %19 ], [ 0, %0 ]
+  %exitcond3 = icmp ne i64 %indvar1, 1536
+  br i1 %exitcond3, label %2, label %20
+
+; <label>:2                                       ; preds = %1
+  br label %3
+
+; <label>:3                                       ; preds = %15, %2
+  %indvar = phi i64 [ %indvar.next, %15 ], [ 0, %2 ]
+  %scevgep = getelementptr [1536 x [1536 x float]]* @C, i64 0, i64 %indvar1, i64 %indvar
+  %j.0 = trunc i64 %indvar to i32
+  %exitcond = icmp ne i64 %indvar, 1536
+  br i1 %exitcond, label %4, label %16
+
+; <label>:4                                       ; preds = %3
+  %5 = load %struct._IO_FILE** @stdout, align 8
+  %6 = load float* %scevgep
+  %7 = fpext float %6 to double
+  %8 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %5, i8* getelementptr inbounds ([5 x i8]* @.str, i32 0, i32 0), double %7)
+  %9 = srem i32 %j.0, 80
+  %10 = icmp eq i32 %9, 79
+  br i1 %10, label %11, label %14
+
+; <label>:11                                      ; preds = %4
+  %12 = load %struct._IO_FILE** @stdout, align 8
+  %13 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %12, i8* getelementptr inbounds ([2 x i8]* @.str1, i32 0, i32 0))
+  br label %14
+
+; <label>:14                                      ; preds = %11, %4
+  br label %15
+
+; <label>:15                                      ; preds = %14
+  %indvar.next = add i64 %indvar, 1
+  br label %3
+
+; <label>:16                                      ; preds = %3
+  %17 = load %struct._IO_FILE** @stdout, align 8
+  %18 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %17, i8* getelementptr inbounds ([2 x i8]* @.str1, i32 0, i32 0))
+  br label %19
+
+; <label>:19                                      ; preds = %16
+  %indvar.next2 = add i64 %indvar1, 1
+  br label %1
+
+; <label>:20                                      ; preds = %1
+  ret void
+}
+
+declare i32 @fprintf(%struct._IO_FILE*, i8*, ...)
+
+define i32 @main() nounwind {
+; <label>:0
+  call void @init_array()
+  br label %1
+
+; <label>:1                                       ; preds = %16, %0
+  %indvar3 = phi i64 [ %indvar.next4, %16 ], [ 0, %0 ]
+  %exitcond9 = icmp ne i64 %indvar3, 1536
+  br i1 %exitcond9, label %2, label %17
+
+; <label>:2                                       ; preds = %1
+  br label %3
+
+; <label>:3                                       ; preds = %14, %2
+  %indvar1 = phi i64 [ %indvar.next2, %14 ], [ 0, %2 ]
+  %scevgep8 = getelementptr [1536 x [1536 x float]]* @C, i64 0, i64 %indvar3, i64 %indvar1
+  %exitcond6 = icmp ne i64 %indvar1, 1536
+  br i1 %exitcond6, label %4, label %15
+
+; <label>:4                                       ; preds = %3
+  store float 0.000000e+00, float* %scevgep8
+  br label %5
+
+; <label>:5                                       ; preds = %12, %4
+  %indvar = phi i64 [ %indvar.next, %12 ], [ 0, %4 ]
+  %scevgep5 = getelementptr [1536 x [1536 x float]]* @A, i64 0, i64 %indvar3, i64 %indvar
+  %scevgep = getelementptr [1536 x [1536 x float]]* @B, i64 0, i64 %indvar, i64 %indvar1
+  %exitcond = icmp ne i64 %indvar, 1536
+  br i1 %exitcond, label %6, label %13
+
+; <label>:6                                       ; preds = %5
+  %7 = load float* %scevgep8
+  %8 = load float* %scevgep5
+  %9 = load float* %scevgep
+  %10 = fmul float %8, %9
+  %11 = fadd float %7, %10
+  store float %11, float* %scevgep8
+  br label %12
+
+; <label>:12                                      ; preds = %6
+  %indvar.next = add i64 %indvar, 1
+  br label %5
+
+; <label>:13                                      ; preds = %5
+  br label %14
+
+; <label>:14                                      ; preds = %13
+  %indvar.next2 = add i64 %indvar1, 1
+  br label %3
+
+; <label>:15                                      ; preds = %3
+  br label %16
+
+; <label>:16                                      ; preds = %15
+  %indvar.next4 = add i64 %indvar3, 1
+  br label %1
+
+; <label>:17                                      ; preds = %1
+  ret i32 0
+}

Added: polly/trunk/www/experiments/matmul/matmul.s
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/matmul.s?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/matmul.s (added)
+++ polly/trunk/www/experiments/matmul/matmul.s Mon May  2 02:48:29 2011
@@ -0,0 +1,255 @@
+; ModuleID = 'matmul.c'
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-unknown-linux-gnu"
+
+%struct._IO_FILE = type { i32, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, i8*, %struct._IO_marker*, %struct._IO_FILE*, i32, i32, i64, i16, i8, [1 x i8], i8*, i64, i8*, i8*, i8*, i8*, i64, i32, [20 x i8] }
+%struct._IO_marker = type { %struct._IO_marker*, %struct._IO_FILE*, i32 }
+
+ at A = common global [1536 x [1536 x float]] zeroinitializer, align 16
+ at B = common global [1536 x [1536 x float]] zeroinitializer, align 16
+ at stdout = external global %struct._IO_FILE*
+ at .str = private unnamed_addr constant [5 x i8] c"%lf \00"
+ at C = common global [1536 x [1536 x float]] zeroinitializer, align 16
+ at .str1 = private unnamed_addr constant [2 x i8] c"\0A\00"
+
+define void @init_array() nounwind {
+  %i = alloca i32, align 4
+  %j = alloca i32, align 4
+  store i32 0, i32* %i, align 4
+  br label %1
+
+; <label>:1                                       ; preds = %41, %0
+  %2 = load i32* %i, align 4
+  %3 = icmp slt i32 %2, 1536
+  br i1 %3, label %4, label %44
+
+; <label>:4                                       ; preds = %1
+  store i32 0, i32* %j, align 4
+  br label %5
+
+; <label>:5                                       ; preds = %37, %4
+  %6 = load i32* %j, align 4
+  %7 = icmp slt i32 %6, 1536
+  br i1 %7, label %8, label %40
+
+; <label>:8                                       ; preds = %5
+  %9 = load i32* %i, align 4
+  %10 = load i32* %j, align 4
+  %11 = mul nsw i32 %9, %10
+  %12 = srem i32 %11, 1024
+  %13 = add nsw i32 1, %12
+  %14 = sitofp i32 %13 to double
+  %15 = fdiv double %14, 2.000000e+00
+  %16 = fptrunc double %15 to float
+  %17 = load i32* %j, align 4
+  %18 = sext i32 %17 to i64
+  %19 = load i32* %i, align 4
+  %20 = sext i32 %19 to i64
+  %21 = getelementptr inbounds [1536 x [1536 x float]]* @A, i32 0, i64 %20
+  %22 = getelementptr inbounds [1536 x float]* %21, i32 0, i64 %18
+  store float %16, float* %22
+  %23 = load i32* %i, align 4
+  %24 = load i32* %j, align 4
+  %25 = mul nsw i32 %23, %24
+  %26 = srem i32 %25, 1024
+  %27 = add nsw i32 1, %26
+  %28 = sitofp i32 %27 to double
+  %29 = fdiv double %28, 2.000000e+00
+  %30 = fptrunc double %29 to float
+  %31 = load i32* %j, align 4
+  %32 = sext i32 %31 to i64
+  %33 = load i32* %i, align 4
+  %34 = sext i32 %33 to i64
+  %35 = getelementptr inbounds [1536 x [1536 x float]]* @B, i32 0, i64 %34
+  %36 = getelementptr inbounds [1536 x float]* %35, i32 0, i64 %32
+  store float %30, float* %36
+  br label %37
+
+; <label>:37                                      ; preds = %8
+  %38 = load i32* %j, align 4
+  %39 = add nsw i32 %38, 1
+  store i32 %39, i32* %j, align 4
+  br label %5
+
+; <label>:40                                      ; preds = %5
+  br label %41
+
+; <label>:41                                      ; preds = %40
+  %42 = load i32* %i, align 4
+  %43 = add nsw i32 %42, 1
+  store i32 %43, i32* %i, align 4
+  br label %1
+
+; <label>:44                                      ; preds = %1
+  ret void
+}
+
+define void @print_array() nounwind {
+  %i = alloca i32, align 4
+  %j = alloca i32, align 4
+  store i32 0, i32* %i, align 4
+  br label %1
+
+; <label>:1                                       ; preds = %32, %0
+  %2 = load i32* %i, align 4
+  %3 = icmp slt i32 %2, 1536
+  br i1 %3, label %4, label %35
+
+; <label>:4                                       ; preds = %1
+  store i32 0, i32* %j, align 4
+  br label %5
+
+; <label>:5                                       ; preds = %26, %4
+  %6 = load i32* %j, align 4
+  %7 = icmp slt i32 %6, 1536
+  br i1 %7, label %8, label %29
+
+; <label>:8                                       ; preds = %5
+  %9 = load %struct._IO_FILE** @stdout, align 8
+  %10 = load i32* %j, align 4
+  %11 = sext i32 %10 to i64
+  %12 = load i32* %i, align 4
+  %13 = sext i32 %12 to i64
+  %14 = getelementptr inbounds [1536 x [1536 x float]]* @C, i32 0, i64 %13
+  %15 = getelementptr inbounds [1536 x float]* %14, i32 0, i64 %11
+  %16 = load float* %15
+  %17 = fpext float %16 to double
+  %18 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %9, i8* getelementptr inbounds ([5 x i8]* @.str, i32 0, i32 0), double %17)
+  %19 = load i32* %j, align 4
+  %20 = srem i32 %19, 80
+  %21 = icmp eq i32 %20, 79
+  br i1 %21, label %22, label %25
+
+; <label>:22                                      ; preds = %8
+  %23 = load %struct._IO_FILE** @stdout, align 8
+  %24 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %23, i8* getelementptr inbounds ([2 x i8]* @.str1, i32 0, i32 0))
+  br label %25
+
+; <label>:25                                      ; preds = %22, %8
+  br label %26
+
+; <label>:26                                      ; preds = %25
+  %27 = load i32* %j, align 4
+  %28 = add nsw i32 %27, 1
+  store i32 %28, i32* %j, align 4
+  br label %5
+
+; <label>:29                                      ; preds = %5
+  %30 = load %struct._IO_FILE** @stdout, align 8
+  %31 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %30, i8* getelementptr inbounds ([2 x i8]* @.str1, i32 0, i32 0))
+  br label %32
+
+; <label>:32                                      ; preds = %29
+  %33 = load i32* %i, align 4
+  %34 = add nsw i32 %33, 1
+  store i32 %34, i32* %i, align 4
+  br label %1
+
+; <label>:35                                      ; preds = %1
+  ret void
+}
+
+declare i32 @fprintf(%struct._IO_FILE*, i8*, ...)
+
+define i32 @main() nounwind {
+  %1 = alloca i32, align 4
+  %i = alloca i32, align 4
+  %j = alloca i32, align 4
+  %k = alloca i32, align 4
+  %t_start = alloca double, align 8
+  %t_end = alloca double, align 8
+  store i32 0, i32* %1
+  call void @init_array()
+  store i32 0, i32* %i, align 4
+  br label %2
+
+; <label>:2                                       ; preds = %57, %0
+  %3 = load i32* %i, align 4
+  %4 = icmp slt i32 %3, 1536
+  br i1 %4, label %5, label %60
+
+; <label>:5                                       ; preds = %2
+  store i32 0, i32* %j, align 4
+  br label %6
+
+; <label>:6                                       ; preds = %53, %5
+  %7 = load i32* %j, align 4
+  %8 = icmp slt i32 %7, 1536
+  br i1 %8, label %9, label %56
+
+; <label>:9                                       ; preds = %6
+  %10 = load i32* %j, align 4
+  %11 = sext i32 %10 to i64
+  %12 = load i32* %i, align 4
+  %13 = sext i32 %12 to i64
+  %14 = getelementptr inbounds [1536 x [1536 x float]]* @C, i32 0, i64 %13
+  %15 = getelementptr inbounds [1536 x float]* %14, i32 0, i64 %11
+  store float 0.000000e+00, float* %15
+  store i32 0, i32* %k, align 4
+  br label %16
+
+; <label>:16                                      ; preds = %49, %9
+  %17 = load i32* %k, align 4
+  %18 = icmp slt i32 %17, 1536
+  br i1 %18, label %19, label %52
+
+; <label>:19                                      ; preds = %16
+  %20 = load i32* %j, align 4
+  %21 = sext i32 %20 to i64
+  %22 = load i32* %i, align 4
+  %23 = sext i32 %22 to i64
+  %24 = getelementptr inbounds [1536 x [1536 x float]]* @C, i32 0, i64 %23
+  %25 = getelementptr inbounds [1536 x float]* %24, i32 0, i64 %21
+  %26 = load float* %25
+  %27 = load i32* %k, align 4
+  %28 = sext i32 %27 to i64
+  %29 = load i32* %i, align 4
+  %30 = sext i32 %29 to i64
+  %31 = getelementptr inbounds [1536 x [1536 x float]]* @A, i32 0, i64 %30
+  %32 = getelementptr inbounds [1536 x float]* %31, i32 0, i64 %28
+  %33 = load float* %32
+  %34 = load i32* %j, align 4
+  %35 = sext i32 %34 to i64
+  %36 = load i32* %k, align 4
+  %37 = sext i32 %36 to i64
+  %38 = getelementptr inbounds [1536 x [1536 x float]]* @B, i32 0, i64 %37
+  %39 = getelementptr inbounds [1536 x float]* %38, i32 0, i64 %35
+  %40 = load float* %39
+  %41 = fmul float %33, %40
+  %42 = fadd float %26, %41
+  %43 = load i32* %j, align 4
+  %44 = sext i32 %43 to i64
+  %45 = load i32* %i, align 4
+  %46 = sext i32 %45 to i64
+  %47 = getelementptr inbounds [1536 x [1536 x float]]* @C, i32 0, i64 %46
+  %48 = getelementptr inbounds [1536 x float]* %47, i32 0, i64 %44
+  store float %42, float* %48
+  br label %49
+
+; <label>:49                                      ; preds = %19
+  %50 = load i32* %k, align 4
+  %51 = add nsw i32 %50, 1
+  store i32 %51, i32* %k, align 4
+  br label %16
+
+; <label>:52                                      ; preds = %16
+  br label %53
+
+; <label>:53                                      ; preds = %52
+  %54 = load i32* %j, align 4
+  %55 = add nsw i32 %54, 1
+  store i32 %55, i32* %j, align 4
+  br label %6
+
+; <label>:56                                      ; preds = %6
+  br label %57
+
+; <label>:57                                      ; preds = %56
+  %58 = load i32* %i, align 4
+  %59 = add nsw i32 %58, 1
+  store i32 %59, i32* %i, align 4
+  br label %2
+
+; <label>:60                                      ; preds = %2
+  ret i32 0
+}

Added: polly/trunk/www/experiments/matmul/runall.sh
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/runall.sh?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/runall.sh (added)
+++ polly/trunk/www/experiments/matmul/runall.sh Mon May  2 02:48:29 2011
@@ -0,0 +1,92 @@
+#!/bin/sh -a
+
+
+echo "--> 1. Create LLVM-IR from C"
+clang -S -emit-llvm matmul.c -o matmul.s
+
+echo "--> 2. Load Polly automatically when calling the 'opt' tool"
+export PATH_TO_POLLY_LIB="~/Projekte/polly/build_clang/lib/"
+alias opt="opt -load ${PATH_TO_POLLY_LIB}/LLVMPolly.so"
+
+echo "--> 3. Prepare the LLVM-IR for Polly"
+opt -S -mem2reg -loop-simplify -indvars matmul.s > matmul.preopt.ll
+
+echo "--> 4. Show the SCoPs detected by Polly"
+opt -basicaa -polly-cloog -analyze -q matmul.preopt.ll
+
+echo "--> 5.1 Highlight the detected SCoPs in the CFGs of the program"
+# We only create .dot files, as directly -view-scops directly calls graphviz
+# which would require user interaction to continue the script.
+# opt -basicaa -view-scops -disable-output matmul.preopt.ll
+opt -basicaa -dot-scops -disable-output matmul.preopt.ll
+
+echo "--> 5.2 Highlight the detected SCoPs in the CFGs of the program (print \
+no instructions)"
+# We only create .dot files, as directly -view-scops-only directly calls
+# graphviz which would require user interaction to continue the script.
+# opt -basicaa -view-scops-only -disable-output matmul.preopt.ll
+opt -basicaa -dot-scops-only -disable-output matmul.preopt.ll
+
+echo "--> 5.3 Create .png files from the .dot files"
+for i in `ls *.dot`; do dot -Tpng $i > $i.png; done
+
+echo "--> 6. View the polyhedral representation of the SCoPs"
+opt -basicaa -polly-scops -analyze matmul.preopt.ll
+
+echo "--> 7. Show the dependences for the SCoPs"
+opt -basicaa -polly-dependences -analyze matmul.preopt.ll
+
+echo "--> 8. Export jscop files"
+opt -basicaa -polly-export-jscop matmul.preopt.ll
+
+echo "--> 9. Import the updated jscop files and print the new SCoPs. (optional)"
+opt -basicaa -polly-import-jscop -polly-cloog -analyze matmul.preopt.ll \
+    -polly-import-jscop-postfix=interchanged
+opt -basicaa -polly-import-jscop -polly-cloog -analyze matmul.preopt.ll \
+    -polly-import-jscop-postfix=interchanged+tiled
+
+echo "--> 10. Codegenerate the SCoPs"
+opt -basicaa -polly-import-jscop -polly-import-jscop-postfix=interchanged \
+    -polly-codegen \
+    matmul.preopt.ll | opt -O3 > matmul.polly.interchanged.ll
+opt -basicaa -polly-import-jscop \
+    -polly-import-jscop-postfix=interchanged+tiled -polly-codegen \
+    matmul.preopt.ll | opt -O3 > matmul.polly.interchanged+tiled.ll
+opt -basicaa -polly-import-jscop \
+    -polly-import-jscop-postfix=interchanged+tiled+vector -polly-codegen \
+    matmul.preopt.ll -enable-polly-vector\
+    | opt -O3 > matmul.polly.interchanged+tiled+vector.ll
+opt -basicaa -polly-import-jscop \
+    -polly-import-jscop-postfix=interchanged+tiled+vector -polly-codegen \
+    matmul.preopt.ll -enable-polly-vector -enable-polly-openmp\
+    | opt -O3 > matmul.polly.interchanged+tiled+vector+openmp.ll
+opt matmul.preopt.ll | opt -O3 > matmul.normalopt.ll
+
+echo "--> 11. Create the executables"
+llc matmul.polly.interchanged.ll -o matmul.polly.interchanged.s && gcc matmul.polly.interchanged.s \
+    -o matmul.polly.interchanged.exe
+llc matmul.polly.interchanged+tiled.ll -o matmul.polly.interchanged+tiled.s && gcc matmul.polly.interchanged+tiled.s \
+    -o matmul.polly.interchanged+tiled.exe
+llc matmul.polly.interchanged+tiled+vector.ll \
+    -o matmul.polly.interchanged+tiled+vector.s \
+    && gcc matmul.polly.interchanged+tiled+vector.s \
+    -o matmul.polly.interchanged+tiled+vector.exe
+llc matmul.polly.interchanged+tiled+vector+openmp.ll \
+    -o matmul.polly.interchanged+tiled+vector+openmp.s \
+    && gcc -lgomp matmul.polly.interchanged+tiled+vector+openmp.s \
+    -o matmul.polly.interchanged+tiled+vector+openmp.exe
+llc matmul.normalopt.ll -o matmul.normalopt.s && gcc matmul.normalopt.s \
+    -o matmul.normalopt.exe
+
+echo "--> 12. Compare the runtime of the executables"
+
+echo "time ./matmul.normalopt.exe"
+time -f "%E real, %U user, %S sys" ./matmul.normalopt.exe
+echo "time ./matmul.polly.interchanged.exe"
+time -f "%E real, %U user, %S sys" ./matmul.polly.interchanged.exe
+echo "time ./matmul.polly.interchanged+tiled.exe"
+time -f "%E real, %U user, %S sys" ./matmul.polly.interchanged+tiled.exe
+echo "time ./matmul.polly.interchanged+tiled+vector.exe"
+time -f "%E real, %U user, %S sys" ./matmul.polly.interchanged+tiled+vector.exe
+echo "time ./matmul.polly.interchanged+tiled+vector+openmp.exe"
+time -f "%E real, %U user, %S sys" ./matmul.polly.interchanged+tiled+vector+openmp.exe

Propchange: polly/trunk/www/experiments/matmul/runall.sh
------------------------------------------------------------------------------
    svn:executable = *

Added: polly/trunk/www/experiments/matmul/scops.init_array.dot
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scops.init_array.dot?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/scops.init_array.dot (added)
+++ polly/trunk/www/experiments/matmul/scops.init_array.dot Mon May  2 02:48:29 2011
@@ -0,0 +1,47 @@
+digraph "Scop Graph for 'init_array' function" {
+	label="Scop Graph for 'init_array' function";
+
+	Node0x26ade30 [shape=record,label="{%0:\l\l  br label %1\l}"];
+	Node0x26ade30 -> Node0x26acdd0;
+	Node0x26acdd0 [shape=record,label="{%1:\l\l  %2 = phi i64 [ %indvar.next2, %18 ], [ 0, %0 ]\l  %exitcond5 = icmp ne i64 %2, 1536\l  br i1 %exitcond5, label %3, label %19\l}"];
+	Node0x26acdd0 -> Node0x26acdf0;
+	Node0x26acdd0 -> Node0x26adce0;
+	Node0x26acdf0 [shape=record,label="{%3:\l\l  br label %4\l}"];
+	Node0x26acdf0 -> Node0x26addc0;
+	Node0x26addc0 [shape=record,label="{%4:\l\l  %indvar = phi i64 [ %indvar.next, %16 ], [ 0, %3 ]\l  %scevgep4 = getelementptr [1536 x [1536 x float]]* @A, i64 0, i64 %2, i64 %indvar\l  %scevgep = getelementptr [1536 x [1536 x float]]* @B, i64 0, i64 %2, i64 %indvar\l  %tmp = mul i64 %2, %indvar\l  %tmp3 = trunc i64 %tmp to i32\l  %exitcond = icmp ne i64 %indvar, 1536\l  br i1 %exitcond, label %5, label %17\l}"];
+	Node0x26addc0 -> Node0x26ace70;
+	Node0x26addc0 -> Node0x26ad010;
+	Node0x26ace70 [shape=record,label="{%5:\l\l  %6 = srem i32 %tmp3, 1024\l  %7 = add nsw i32 1, %6\l  %8 = sitofp i32 %7 to double\l  %9 = fdiv double %8, 2.000000e+00\l  %10 = fptrunc double %9 to float\l  store float %10, float* %scevgep4\l  %11 = srem i32 %tmp3, 1024\l  %12 = add nsw i32 1, %11\l  %13 = sitofp i32 %12 to double\l  %14 = fdiv double %13, 2.000000e+00\l  %15 = fptrunc double %14 to float\l  store float %15, float* %scevgep\l  br label %16\l}"];
+	Node0x26ace70 -> Node0x26ace90;
+	Node0x26ace90 [shape=record,label="{%16:\l\l  %indvar.next = add i64 %indvar, 1\l  br label %4\l}"];
+	Node0x26ace90 -> Node0x26addc0[constraint=false];
+	Node0x26ad010 [shape=record,label="{%17:\l\l  br label %18\l}"];
+	Node0x26ad010 -> Node0x26ad6c0;
+	Node0x26ad6c0 [shape=record,label="{%18:\l\l  %indvar.next2 = add i64 %2, 1\l  br label %1\l}"];
+	Node0x26ad6c0 -> Node0x26acdd0[constraint=false];
+	Node0x26adce0 [shape=record,label="{%19:\l\l  ret void\l}"];
+	colorscheme = "paired12"
+        subgraph cluster_0x26a94c0 {
+          label = "";
+          style = solid;
+          color = 1
+          subgraph cluster_0x26aa4e0 {
+            label = "";
+            style = filled;
+            color = 3            subgraph cluster_0x26a9780 {
+              label = "";
+              style = solid;
+              color = 5
+              Node0x26addc0;
+              Node0x26ace70;
+              Node0x26ace90;
+            }
+            Node0x26acdd0;
+            Node0x26acdf0;
+            Node0x26ad010;
+            Node0x26ad6c0;
+          }
+          Node0x26ade30;
+          Node0x26adce0;
+        }
+}

Added: polly/trunk/www/experiments/matmul/scops.init_array.dot.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scops.init_array.dot.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/scops.init_array.dot.png (added) and polly/trunk/www/experiments/matmul/scops.init_array.dot.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/scops.main.dot
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scops.main.dot?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/scops.main.dot (added)
+++ polly/trunk/www/experiments/matmul/scops.main.dot Mon May  2 02:48:29 2011
@@ -0,0 +1,65 @@
+digraph "Scop Graph for 'main' function" {
+	label="Scop Graph for 'main' function";
+
+	Node0x26ace10 [shape=record,label="{%0:\l\l  call void @init_array()\l  br label %1\l}"];
+	Node0x26ace10 -> Node0x26acd60;
+	Node0x26acd60 [shape=record,label="{%1:\l\l  %indvar3 = phi i64 [ %indvar.next4, %16 ], [ 0, %0 ]\l  %exitcond9 = icmp ne i64 %indvar3, 1536\l  br i1 %exitcond9, label %2, label %17\l}"];
+	Node0x26acd60 -> Node0x26acd80;
+	Node0x26acd60 -> Node0x26af2e0;
+	Node0x26acd80 [shape=record,label="{%2:\l\l  br label %3\l}"];
+	Node0x26acd80 -> Node0x26aee80;
+	Node0x26aee80 [shape=record,label="{%3:\l\l  %indvar1 = phi i64 [ %indvar.next2, %14 ], [ 0, %2 ]\l  %scevgep8 = getelementptr [1536 x [1536 x float]]* @C, i64 0, i64 %indvar3, i64 %indvar1\l  %exitcond6 = icmp ne i64 %indvar1, 1536\l  br i1 %exitcond6, label %4, label %15\l}"];
+	Node0x26aee80 -> Node0x26aeea0;
+	Node0x26aee80 -> Node0x26aeec0;
+	Node0x26aeea0 [shape=record,label="{%4:\l\l  store float 0.000000e+00, float* %scevgep8\l  br label %5\l}"];
+	Node0x26aeea0 -> Node0x26aced0;
+	Node0x26aced0 [shape=record,label="{%5:\l\l  %indvar = phi i64 [ %indvar.next, %12 ], [ 0, %4 ]\l  %scevgep5 = getelementptr [1536 x [1536 x float]]* @A, i64 0, i64 %indvar3, i64 %indvar\l  %scevgep = getelementptr [1536 x [1536 x float]]* @B, i64 0, i64 %indvar, i64 %indvar1\l  %exitcond = icmp ne i64 %indvar, 1536\l  br i1 %exitcond, label %6, label %13\l}"];
+	Node0x26aced0 -> Node0x26ace60;
+	Node0x26aced0 -> Node0x26af5e0;
+	Node0x26ace60 [shape=record,label="{%6:\l\l  %7 = load float* %scevgep8\l  %8 = load float* %scevgep5\l  %9 = load float* %scevgep\l  %10 = fmul float %8, %9\l  %11 = fadd float %7, %10\l  store float %11, float* %scevgep8\l  br label %12\l}"];
+	Node0x26ace60 -> Node0x26af640;
+	Node0x26af640 [shape=record,label="{%12:\l\l  %indvar.next = add i64 %indvar, 1\l  br label %5\l}"];
+	Node0x26af640 -> Node0x26aced0[constraint=false];
+	Node0x26af5e0 [shape=record,label="{%13:\l\l  br label %14\l}"];
+	Node0x26af5e0 -> Node0x26af6e0;
+	Node0x26af6e0 [shape=record,label="{%14:\l\l  %indvar.next2 = add i64 %indvar1, 1\l  br label %3\l}"];
+	Node0x26af6e0 -> Node0x26aee80[constraint=false];
+	Node0x26aeec0 [shape=record,label="{%15:\l\l  br label %16\l}"];
+	Node0x26aeec0 -> Node0x26af740;
+	Node0x26af740 [shape=record,label="{%16:\l\l  %indvar.next4 = add i64 %indvar3, 1\l  br label %1\l}"];
+	Node0x26af740 -> Node0x26acd60[constraint=false];
+	Node0x26af2e0 [shape=record,label="{%17:\l\l  ret i32 0\l}"];
+	colorscheme = "paired12"
+        subgraph cluster_0x26a8b20 {
+          label = "";
+          style = solid;
+          color = 1
+          subgraph cluster_0x26a9220 {
+            label = "";
+            style = filled;
+            color = 3            subgraph cluster_0x26ad500 {
+              label = "";
+              style = solid;
+              color = 5
+              subgraph cluster_0x26ad480 {
+                label = "";
+                style = solid;
+                color = 7
+                Node0x26aced0;
+                Node0x26ace60;
+                Node0x26af640;
+              }
+              Node0x26aee80;
+              Node0x26aeea0;
+              Node0x26af5e0;
+              Node0x26af6e0;
+            }
+            Node0x26acd60;
+            Node0x26acd80;
+            Node0x26aeec0;
+            Node0x26af740;
+          }
+          Node0x26ace10;
+          Node0x26af2e0;
+        }
+}

Added: polly/trunk/www/experiments/matmul/scops.main.dot.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scops.main.dot.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/scops.main.dot.png (added) and polly/trunk/www/experiments/matmul/scops.main.dot.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/scops.print_array.dot
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scops.print_array.dot?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/scops.print_array.dot (added)
+++ polly/trunk/www/experiments/matmul/scops.print_array.dot Mon May  2 02:48:29 2011
@@ -0,0 +1,60 @@
+digraph "Scop Graph for 'print_array' function" {
+	label="Scop Graph for 'print_array' function";
+
+	Node0x26ac9a0 [shape=record,label="{%0:\l\l  br label %1\l}"];
+	Node0x26ac9a0 -> Node0x26acd00;
+	Node0x26acd00 [shape=record,label="{%1:\l\l  %indvar1 = phi i64 [ %indvar.next2, %19 ], [ 0, %0 ]\l  %exitcond3 = icmp ne i64 %indvar1, 1536\l  br i1 %exitcond3, label %2, label %20\l}"];
+	Node0x26acd00 -> Node0x26a8ac0;
+	Node0x26acd00 -> Node0x26ac9c0;
+	Node0x26a8ac0 [shape=record,label="{%2:\l\l  br label %3\l}"];
+	Node0x26a8ac0 -> Node0x26ad940;
+	Node0x26ad940 [shape=record,label="{%3:\l\l  %indvar = phi i64 [ %indvar.next, %15 ], [ 0, %2 ]\l  %scevgep = getelementptr [1536 x [1536 x float]]* @C, i64 0, i64 %indvar1, i64 %indvar\l  %j.0 = trunc i64 %indvar to i32\l  %exitcond = icmp ne i64 %indvar, 1536\l  br i1 %exitcond, label %4, label %16\l}"];
+	Node0x26ad940 -> Node0x26acde0;
+	Node0x26ad940 -> Node0x26ad9e0;
+	Node0x26acde0 [shape=record,label="{%4:\l\l  %5 = load %struct._IO_FILE** @stdout, align 8\l  %6 = load float* %scevgep\l  %7 = fpext float %6 to double\l  %8 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %5, i8* getelementptr inbounds ([5 x i8]* @.str, i32 0, i32 0), double %7)\l  %9 = srem i32 %j.0, 80\l  %10 = icmp eq i32 %9, 79\l  br i1 %10, label %11, label %14\l}"];
+	Node0x26acde0 -> Node0x26ada40;
+	Node0x26acde0 -> Node0x26acfa0;
+	Node0x26ada40 [shape=record,label="{%11:\l\l  %12 = load %struct._IO_FILE** @stdout, align 8\l  %13 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %12, i8* getelementptr inbounds ([2 x i8]* @.str1, i32 0, i32 0))\l  br label %14\l}"];
+	Node0x26ada40 -> Node0x26acfa0;
+	Node0x26acfa0 [shape=record,label="{%14:\l\l  br label %15\l}"];
+	Node0x26acfa0 -> Node0x26ad6c0;
+	Node0x26ad6c0 [shape=record,label="{%15:\l\l  %indvar.next = add i64 %indvar, 1\l  br label %3\l}"];
+	Node0x26ad6c0 -> Node0x26ad940[constraint=false];
+	Node0x26ad9e0 [shape=record,label="{%16:\l\l  %17 = load %struct._IO_FILE** @stdout, align 8\l  %18 = call i32 (%struct._IO_FILE*, i8*, ...)* @fprintf(%struct._IO_FILE* %17, i8* getelementptr inbounds ([2 x i8]* @.str1, i32 0, i32 0))\l  br label %19\l}"];
+	Node0x26ad9e0 -> Node0x26ace00;
+	Node0x26ace00 [shape=record,label="{%19:\l\l  %indvar.next2 = add i64 %indvar1, 1\l  br label %1\l}"];
+	Node0x26ace00 -> Node0x26acd00[constraint=false];
+	Node0x26ac9c0 [shape=record,label="{%20:\l\l  ret void\l}"];
+	colorscheme = "paired12"
+        subgraph cluster_0x26adae0 {
+          label = "";
+          style = solid;
+          color = 1
+          subgraph cluster_0x26aa030 {
+            label = "";
+            style = solid;
+            color = 6
+            subgraph cluster_0x26a9fb0 {
+              label = "";
+              style = solid;
+              color = 5
+              subgraph cluster_0x26adb60 {
+                label = "";
+                style = solid;
+                color = 7
+                Node0x26acde0;
+                Node0x26ada40;
+              }
+              Node0x26ad940;
+              Node0x26acfa0;
+              Node0x26ad6c0;
+            }
+            Node0x26acd00;
+            Node0x26a8ac0;
+            Node0x26ad9e0;
+            Node0x26ace00;
+          }
+          Node0x26ac9a0;
+          Node0x26ac9c0;
+        }
+}

Added: polly/trunk/www/experiments/matmul/scops.print_array.dot.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scops.print_array.dot.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/scops.print_array.dot.png (added) and polly/trunk/www/experiments/matmul/scops.print_array.dot.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/scopsonly.init_array.dot
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scopsonly.init_array.dot?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/scopsonly.init_array.dot (added)
+++ polly/trunk/www/experiments/matmul/scopsonly.init_array.dot Mon May  2 02:48:29 2011
@@ -0,0 +1,47 @@
+digraph "Scop Graph for 'init_array' function" {
+	label="Scop Graph for 'init_array' function";
+
+	Node0x24dfca0 [shape=record,label="{%0}"];
+	Node0x24dfca0 -> Node0x24dfdf0;
+	Node0x24dfdf0 [shape=record,label="{%1}"];
+	Node0x24dfdf0 -> Node0x24dee50;
+	Node0x24dfdf0 -> Node0x24def50;
+	Node0x24dee50 [shape=record,label="{%3}"];
+	Node0x24dee50 -> Node0x24deec0;
+	Node0x24deec0 [shape=record,label="{%4}"];
+	Node0x24deec0 -> Node0x24dfdc0;
+	Node0x24deec0 -> Node0x24df0c0;
+	Node0x24dfdc0 [shape=record,label="{%5}"];
+	Node0x24dfdc0 -> Node0x24defb0;
+	Node0x24defb0 [shape=record,label="{%16}"];
+	Node0x24defb0 -> Node0x24deec0[constraint=false];
+	Node0x24df0c0 [shape=record,label="{%17}"];
+	Node0x24df0c0 -> Node0x24deee0;
+	Node0x24deee0 [shape=record,label="{%18}"];
+	Node0x24deee0 -> Node0x24dfdf0[constraint=false];
+	Node0x24def50 [shape=record,label="{%19}"];
+	colorscheme = "paired12"
+        subgraph cluster_0x24db4c0 {
+          label = "";
+          style = solid;
+          color = 1
+          subgraph cluster_0x24dc4e0 {
+            label = "";
+            style = filled;
+            color = 3            subgraph cluster_0x24db780 {
+              label = "";
+              style = solid;
+              color = 5
+              Node0x24deec0;
+              Node0x24dfdc0;
+              Node0x24defb0;
+            }
+            Node0x24dfdf0;
+            Node0x24dee50;
+            Node0x24df0c0;
+            Node0x24deee0;
+          }
+          Node0x24dfca0;
+          Node0x24def50;
+        }
+}

Added: polly/trunk/www/experiments/matmul/scopsonly.init_array.dot.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scopsonly.init_array.dot.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/scopsonly.init_array.dot.png (added) and polly/trunk/www/experiments/matmul/scopsonly.init_array.dot.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/scopsonly.main.dot
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scopsonly.main.dot?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/scopsonly.main.dot (added)
+++ polly/trunk/www/experiments/matmul/scopsonly.main.dot Mon May  2 02:48:29 2011
@@ -0,0 +1,65 @@
+digraph "Scop Graph for 'main' function" {
+	label="Scop Graph for 'main' function";
+
+	Node0x24deb60 [shape=record,label="{%0}"];
+	Node0x24deb60 -> Node0x24deaa0;
+	Node0x24deaa0 [shape=record,label="{%1}"];
+	Node0x24deaa0 -> Node0x24e12a0;
+	Node0x24deaa0 -> Node0x24e0e30;
+	Node0x24e12a0 [shape=record,label="{%2}"];
+	Node0x24e12a0 -> Node0x24e0e00;
+	Node0x24e0e00 [shape=record,label="{%3}"];
+	Node0x24e0e00 -> Node0x24e1410;
+	Node0x24e0e00 -> Node0x24e1470;
+	Node0x24e1410 [shape=record,label="{%4}"];
+	Node0x24e1410 -> Node0x24e1380;
+	Node0x24e1380 [shape=record,label="{%5}"];
+	Node0x24e1380 -> Node0x24deaf0;
+	Node0x24e1380 -> Node0x24e1620;
+	Node0x24deaf0 [shape=record,label="{%6}"];
+	Node0x24deaf0 -> Node0x24e1680;
+	Node0x24e1680 [shape=record,label="{%12}"];
+	Node0x24e1680 -> Node0x24e1380[constraint=false];
+	Node0x24e1620 [shape=record,label="{%13}"];
+	Node0x24e1620 -> Node0x24e16e0;
+	Node0x24e16e0 [shape=record,label="{%14}"];
+	Node0x24e16e0 -> Node0x24e0e00[constraint=false];
+	Node0x24e1470 [shape=record,label="{%15}"];
+	Node0x24e1470 -> Node0x24e01a0;
+	Node0x24e01a0 [shape=record,label="{%16}"];
+	Node0x24e01a0 -> Node0x24deaa0[constraint=false];
+	Node0x24e0e30 [shape=record,label="{%17}"];
+	colorscheme = "paired12"
+        subgraph cluster_0x24dfc10 {
+          label = "";
+          style = solid;
+          color = 1
+          subgraph cluster_0x24de570 {
+            label = "";
+            style = filled;
+            color = 3            subgraph cluster_0x24de7a0 {
+              label = "";
+              style = solid;
+              color = 5
+              subgraph cluster_0x24de720 {
+                label = "";
+                style = solid;
+                color = 7
+                Node0x24e1380;
+                Node0x24deaf0;
+                Node0x24e1680;
+              }
+              Node0x24e0e00;
+              Node0x24e1410;
+              Node0x24e1620;
+              Node0x24e16e0;
+            }
+            Node0x24deaa0;
+            Node0x24e12a0;
+            Node0x24e1470;
+            Node0x24e01a0;
+          }
+          Node0x24deb60;
+          Node0x24e0e30;
+        }
+}

Added: polly/trunk/www/experiments/matmul/scopsonly.main.dot.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scopsonly.main.dot.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/scopsonly.main.dot.png (added) and polly/trunk/www/experiments/matmul/scopsonly.main.dot.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/experiments/matmul/scopsonly.print_array.dot
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scopsonly.print_array.dot?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/experiments/matmul/scopsonly.print_array.dot (added)
+++ polly/trunk/www/experiments/matmul/scopsonly.print_array.dot Mon May  2 02:48:29 2011
@@ -0,0 +1,60 @@
+digraph "Scop Graph for 'print_array' function" {
+	label="Scop Graph for 'print_array' function";
+
+	Node0x24df2c0 [shape=record,label="{%0}"];
+	Node0x24df2c0 -> Node0x24df2a0;
+	Node0x24df2a0 [shape=record,label="{%1}"];
+	Node0x24df2a0 -> Node0x24dee90;
+	Node0x24df2a0 -> Node0x24dee20;
+	Node0x24dee90 [shape=record,label="{%2}"];
+	Node0x24dee90 -> Node0x24debd0;
+	Node0x24debd0 [shape=record,label="{%3}"];
+	Node0x24debd0 -> Node0x24df150;
+	Node0x24debd0 -> Node0x24de990;
+	Node0x24df150 [shape=record,label="{%4}"];
+	Node0x24df150 -> Node0x24df3a0;
+	Node0x24df150 -> Node0x24defb0;
+	Node0x24df3a0 [shape=record,label="{%11}"];
+	Node0x24df3a0 -> Node0x24defb0;
+	Node0x24defb0 [shape=record,label="{%14}"];
+	Node0x24defb0 -> Node0x24df530;
+	Node0x24df530 [shape=record,label="{%15}"];
+	Node0x24df530 -> Node0x24debd0[constraint=false];
+	Node0x24de990 [shape=record,label="{%16}"];
+	Node0x24de990 -> Node0x24df9a0;
+	Node0x24df9a0 [shape=record,label="{%19}"];
+	Node0x24df9a0 -> Node0x24df2a0[constraint=false];
+	Node0x24dee20 [shape=record,label="{%20}"];
+	colorscheme = "paired12"
+        subgraph cluster_0x24dbe40 {
+          label = "";
+          style = solid;
+          color = 1
+          subgraph cluster_0x24db6e0 {
+            label = "";
+            style = solid;
+            color = 6
+            subgraph cluster_0x24db660 {
+              label = "";
+              style = solid;
+              color = 5
+              subgraph cluster_0x24db5e0 {
+                label = "";
+                style = solid;
+                color = 7
+                Node0x24df150;
+                Node0x24df3a0;
+              }
+              Node0x24debd0;
+              Node0x24defb0;
+              Node0x24df530;
+            }
+            Node0x24df2a0;
+            Node0x24dee90;
+            Node0x24de990;
+            Node0x24df9a0;
+          }
+          Node0x24df2c0;
+          Node0x24dee20;
+        }
+}

Added: polly/trunk/www/experiments/matmul/scopsonly.print_array.dot.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/scopsonly.print_array.dot.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/experiments/matmul/scopsonly.print_array.dot.png (added) and polly/trunk/www/experiments/matmul/scopsonly.print_array.dot.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/get_started.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/get_started.html?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/get_started.html (added)
+++ polly/trunk/www/get_started.html Mon May  2 02:48:29 2011
@@ -0,0 +1,136 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
+          "http://www.w3.org/TR/html4/strict.dtd">
+<html>
+<head>
+  <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1" />
+  <title>Polly - Getting Started</title>
+  <link type="text/css" rel="stylesheet" href="menu.css" />
+  <link type="text/css" rel="stylesheet" href="content.css" />
+</head>
+<body>
+
+<!--#include virtual="menu.html.incl"-->
+
+<div id="content">
+
+<h1>Getting Started: Building and Installing Polly</h1>
+
+
+<h2 id="prerequisites"> Prerequisites </h2>
+
+The following prerequisites can be installed through the package management
+system of your operating system.
+
+<ul>
+<li>libgmp (library + developer package)</li>
+</ul>
+
+<h3> Install ISL / CLooG libraries </h3>
+
+Polly requires the latest versions of <a href="http://www.cloog.org">CLooG</a>
+and <a href="http://repo.or.cz/w/isl.git">isl</a> to be installed. The CLooG git
+repository contains both the latest version of CLooG and isl.
+
+<pre>
+git clone git://repo.or.cz/cloog.git
+cd cloog
+./get_submodules.sh
+./autogen.sh
+./configure --with-gmp-prefix=/path/to/gmp/installation --prefix=/path/to/cloog/installation
+make
+make install
+</pre>
+
+<h3> Install Pocc (Optional) </h3>
+
+Polly can use <a href="http://www.cse.ohio-state.edu/~pouchet/software/pocc">
+PoCC</a> as an external optimizer. PoCC provides an
+integrated version of <a href="http://pluto.sf.net">Pluto</a>, an advanced
+data-locality and tileability optimizer.  To enable this feature install PoCC
+1.0-rc3.1 (the one with Polly support) and add it to your PATH.
+
+<pre>
+wget <a
+href="http://www.cse.ohio-state.edu/~pouchet/software/pocc/download/pocc-1.0-rc3.1-full.tar.gz">http://www.cse.ohio-state.edu/~pouchet/software/pocc/download/pocc-1.0-rc3.1-full.tar.gz</a>
+tar xzf pocc-1.0-rc3.1-full.tar.gz
+cd pocc-1.0-rc3.1
+./install.sh
+export PATH=`pwd`/bin
+</pre>
+
+Furthermore, scoplib-0.2.0 has to be installed such that polly can link to
+it.
+
+<pre>
+wget <a
+href="http://www.cse.ohio-state.edu/~pouchet/software/pocc/download/modules/scoplib-0.2.0.tar.gz"
+>http://www.cse.ohio-state.edu/~pouchet/software/pocc/download/modules/scoplib-0.2.0.tar.gz</a>
+tar xzf  scoplib-0.2.0.tar.gz
+cd scoplib-0.2.0
+./configure --enable-mp-version --prefix=/path/to/scoplib/installation
+make && make install
+</pre>
+
+<h2 id="source"> Get the code </h2>
+
+<p>
+The Polly source code is available in the LLVM SVN repository. For convenience
+we also provide a git mirror. To build Polly we extract its source code into the
+<em>tools</em> directory of the llvm sources.</p>
+<b>A recent LLVM checkout is needed. Older versions may not work!</b>
+
+<h3>SVN</h3>
+<pre>
+export LLVM_SRC=`pwd`/llvm
+svn checkout http://llvm.org/svn/llvm-project/llvm/trunk ${LLVM_SRC}
+cd ${LLVM_SRC}/tools
+svn checkout http://llvm.org/svn/llvm-project/polly/trunk polly
+</pre>
+<h3>GIT</h3>
+<pre>
+export LLVM_SRC=`pwd`/llvm
+git clone http://llvm.org/git/llvm.git ${LLVM_SRC}
+cd ${LLVM_SRC}/tools
+git clone git://repo.or.cz/polly.git
+</pre>
+
+
+
+<h2 id="build">Build Polly</h2>
+
+To build Polly you can either use the autoconf or the cmake build system. At the
+moment only the autoconf build system allows to run the llvm test-suite and only
+the cmake build system allows to run 'make polly-test'.
+
+<h3>CMake</h3>
+
+<pre>
+mkdir build
+cd build
+cmake ${LLVM_SRC}
+
+# If CMAKE cannot find CLooG and ISL
+cmake -DCMAKE_PREFIX_PATH=/cloog/installation .
+
+# To point CMAKE to the scoplib source
+cmake -DCMAKE_PREFIX_PATH=/scoplib/installation .
+
+make
+</pre>
+
+<h3> Autoconf </h2>
+
+<pre>
+mkdir build
+cd build
+${LLVM_SRC}/configure --with-cloog=/cloog/installation --with-isl=/cloog/installation --with-scoplib=/scoplib/installation
+make
+</pre>
+
+<h2> Test Polly</h2>
+
+To check if Polly works correctly you can run <em>make polly-test</em>. This
+currently works only with a cmake build.
+</div>
+</body>
+</html>

Added: polly/trunk/www/images/architecture.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/architecture.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/architecture.png (added) and polly/trunk/www/images/architecture.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/images/iit-madras.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/iit-madras.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/iit-madras.png (added) and polly/trunk/www/images/iit-madras.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/images/osu.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/osu.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/osu.png (added) and polly/trunk/www/images/osu.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/images/performance/parallel-large.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/performance/parallel-large.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/performance/parallel-large.png (added) and polly/trunk/www/images/performance/parallel-large.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/images/performance/parallel-small.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/performance/parallel-small.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/performance/parallel-small.png (added) and polly/trunk/www/images/performance/parallel-small.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/images/performance/sequential-large.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/performance/sequential-large.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/performance/sequential-large.png (added) and polly/trunk/www/images/performance/sequential-large.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/images/performance/sequential-small.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/performance/sequential-small.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/performance/sequential-small.png (added) and polly/trunk/www/images/performance/sequential-small.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/images/sys-uni.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/sys-uni.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/sys-uni.png (added) and polly/trunk/www/images/sys-uni.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/images/uni-passau.png
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/images/uni-passau.png?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/images/uni-passau.png (added) and polly/trunk/www/images/uni-passau.png Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/index.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/index.html?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/index.html (added)
+++ polly/trunk/www/index.html Mon May  2 02:48:29 2011
@@ -0,0 +1,73 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" 
+          "http://www.w3.org/TR/html4/strict.dtd">
+<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
+<html>
+<head>
+  <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
+  <title>Polly - Polyhedral optimizations for LLVM</title>
+  <link type="text/css" rel="stylesheet" href="menu.css">
+  <link type="text/css" rel="stylesheet" href="content.css">
+</head>
+<body>
+<!--#include virtual="menu.html.incl"-->
+<div id="content">
+  <!--*********************************************************************-->
+  <h1>Polly: Polyhedral optimizations for LLVM</h1>
+  <!--*********************************************************************-->
+
+  <p>Polly is a project that works on advanced optimizations for data-locality
+  and parallelism. It uses the polyhedral model, a high-level mathematical
+  abstraction, to analyse and optimize the memory access pattern of a program.
+  Due to the use of a polyhedral representation Polly can easily calculate
+  detailed data dependency information which it uses to derive an optimized loop
+  structure. Polly can speed up sequential code by improving memory locality and
+  consequently the cache use.  Furthermore, Polly is able to expose different
+  kinds of parallelism which it exploits by introducing (basic) OpenMP and SIMD
+  code.  The automatic use of vector accelerators is planned and will take
+  avantage of the ongoing work on the LLVM PTX backend.
+  </p>
+
+  <em> Polly is still a research project, that is not production quality. We are
+  working on a robust implementation of Polly's core. You are invited to join us
+  by directly contributing to Polly or by using it for your own research.</em>
+
+  <!--=====================================================================-->
+  <h2>Major changes in Polly</h2>
+  <!--=====================================================================-->
+
+  <ul>
+  <li>April 2011 - Polly moves to the LLVM infrastructure </li>
+  <li>March 2011 - Polly is presented at <a
+  href="http://impact2011.inrialpes.fr/">CGO/IMPACT 2011</a>, Polly can compile
+  polybench 2.0 with vectorization and OpenMP code generation.  </li>
+  <li> Februar 2011 - pollycc - a script to automatically compile with
+  polyhedral optimizations </li>
+  <li> Januar 2011 - Basic OpenMP support, Alias analysis integration,
+  Pluto/POCC support </li>
+  <li> Dezember 2010 - Basic vectorization support </li>
+  <li> November 2010 - Talk about Polly at the <a
+  href="http://llvm.org/devmtg/2010-11/">LLVM Developer Meeting</a> </li>
+  <li> October 2010 - Added dependency analysis </li>
+  <li> October 2010 - Finished Phase 1 - Get something working </li>
+  <li> October 2010 - Support for scalar dependences and sequential SCoPs </li>
+  <li> August 2010 - RegionInfo pass committed to llvm </li>
+  <li> August 2010 - llvm-test suite compiles </li>
+  <li> July 2010 - Code generation works for normal SCoPs.  </li>
+  <li> June 2010 - OpenSCoP import/export works (as far as openscop is finished)
+  </li>
+  <li> May 2010 - The CLooG AST can be parsed </li>
+  <li> April 2010 - SCoPs can automatically be detected (WIP) </li>
+  <li> March 2010 - The RegionInfo framework is almost completed.  </li>
+  <li> February 2010 - Translating a simple loop to Polly-IR and passing it to
+  CLooG-isl to regenerate a loop structure works.  </li>
+  <li> February 2010 - ISL and CLooG are integrated.  </li>
+  <li> January 2010 - The RegionInfo pass is finished.  </li>
+  <li> End of 2009 - Work on the infrastructure started.  </li>
+  </ul>
+  <!--=====================================================================-->
+  <h2> The architecture of Polly</h2>
+  <!--=====================================================================-->
+  <img src='images/architecture.png' />
+</div>
+</body>
+</html>

Added: polly/trunk/www/menu.css
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/menu.css?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/menu.css (added)
+++ polly/trunk/www/menu.css Mon May  2 02:48:29 2011
@@ -0,0 +1,39 @@
+/***************/
+/* page layout */
+/***************/
+
+[id=menu] {
+	width:25ex;
+        float: left;
+}
+[id=content] {
+	/* *****  EDIT THIS VALUE IF CONTENT OVERLAPS MENU ***** */
+	position:absolute;
+  left:29ex;
+	padding-right:4ex;
+}
+
+/**************/
+/* menu style */
+/**************/
+
+#menu .submenu {
+	padding-top:1em;
+	display:block;
+}
+
+#menu label {
+	display:block;
+	font-weight: bold;
+	text-align: center;
+	background-color: rgb(192,192,192);
+}
+#menu a {
+	padding:0 .2em;
+	display:block;
+	text-align: center;
+	background-color: rgb(235,235,235);
+}
+#menu a:visited {
+	color:rgb(100,50,100);
+}

Added: polly/trunk/www/menu.html.incl
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/menu.html.incl?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/menu.html.incl (added)
+++ polly/trunk/www/menu.html.incl Mon May  2 02:48:29 2011
@@ -0,0 +1,36 @@
+<div id="menu">
+  <div>
+    <a href="http://llvm.org/">LLVM Home</a>
+  </div>
+
+  <div class="submenu">
+    <label>Polly Info</label>
+    <a href="index.html">About</a>
+    <a href="todo.html">Todo</a>
+    <a href="passes.html">LLVM Passes</a>
+<!--    <a href="examples.html">Examples</a> -->
+    <a href="performance.html">Performance</a>
+    <a href="publications.html">Publications</a>
+    <a href="contributors.html">Contributors</a>
+  </div>
+
+  <div class="submenu">
+    <label>Communication</label>
+    <a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits">
+      llvm-commits List
+    </a>
+    <a href="http://groups.google.com/group/polly-dev">polly-dev List</a>
+    <a href="http://llvm.org/bugs/">Bug Reports</a>
+  </div>
+
+  <div class="submenu">
+    <label>The Code</label>
+    <a href="get_started.html#prerequisites">Prerequisites</a>
+    <a href="get_started.html#source">Download</a>
+    <a href="get_started.html#build">Build</a>
+    <a href="http://llvm.org/viewvc/llvm-project/polly/trunk/">
+      Browse (ViewVC)
+    </a>
+    <a href="http://repo.or.cz/w/polly-mirror.git">Browse (GitWeb)</a>
+  </div>
+</div>

Added: polly/trunk/www/passes.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/passes.html?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/passes.html (added)
+++ polly/trunk/www/passes.html Mon May  2 02:48:29 2011
@@ -0,0 +1,68 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" 
+          "http://www.w3.org/TR/html4/strict.dtd">
+<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
+<html>
+<head>
+  <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
+  <title>Polly - The available LLVM passes</title>
+  <link type="text/css" rel="stylesheet" href="menu.css">
+  <link type="text/css" rel="stylesheet" href="content.css">
+</head>
+<body>
+<!--#include virtual="menu.html.incl"-->
+<div id="content">
+  <!--*********************************************************************-->
+  <h1>Polly: The available LLVM passes</h1>
+  <!--*********************************************************************-->
+
+  <p>Polly consists of a set of LLVM passes.  </p>
+
+<h2>Front End</h2>
+<ul>
+<li><em>polly-prepare</em> Prepare code for Polly</li>
+<li><em>polly-region-simplify</em> Transform refined regions into simple regions</li>
+<li><em>polly-detect</em> Detect SCoPs in functions</li>
+<li><em>polly-analyze-ir</em> Analyse the LLVM-IR in the detected SCoPs</li>
+<li><em>polly-independent</em> Create independent blocks</li>
+<li><em>polly-scops</em> Create polyhedral description of SCoPs</li>
+</ul>
+<h2>Middle End</h2>
+<ul>
+<li><em>polly-dependences</em> Calculate the dependences in a SCoPs</li>
+<li><em>polly-interchange</em> Perform loop interchange (work in progress)</li>
+<li><em>polly-optimize</em> Optimize the SCoP using PoCC</li>
+<li>Import/Export
+<ul>
+<li><em>polly-export-cloog</em> Export the CLooG input file
+(Writes a .cloog file for each SCoP)</li>
+<li><em>polly-export</em> Export SCoPs with OpenScop library
+(Writes a .scop file for each SCoP)</li>
+<li><em>polly-import</em> Import SCoPs with OpenScop library
+(Reads a .scop file for each SCoP)</li>
+<li><em>polly-export-scoplib</em> Export SCoPs with ScopLib library
+(Writes a .scoplib file for each SCoP)</li>
+<li><em>polly-import-scoplib</em> Import SCoPs with ScopLib library
+(Reads a .scoplib file for each SCoP)</li>
+<li><em>polly-export-jscop</em> Export SCoPs as JSON
+(Writes a .jscop file for each SCoP)</li>
+<li><em>polly-import-jscop</em> Import SCoPs from JSON
+(Reads a .jscop file for each SCoP)</li>
+</ul>
+</li>
+<li>Graphviz
+<ul>
+<li><em>dot-scops</em> Print SCoPs of function</li>
+<li><em>dot-scops-only</em> Print SCoPs of function (without function bodies)</li>
+<li><em>view-scops</em> View SCoPs of function</li>
+<li><em>view-scops-only</em> View SCoPs of function (without function bodies)</li>
+</ul></li>
+</ul>
+<h2>Back End</h2>
+<ul>
+<li><em>polly-cloog</em> Execute CLooG code generation</li>
+<li><em>polly-codegen</em> Create LLVM-IR from the polyhedral information</li>
+</ul>
+
+</div>
+</body>
+</html>

Added: polly/trunk/www/performance.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/performance.html?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/performance.html (added)
+++ polly/trunk/www/performance.html Mon May  2 02:48:29 2011
@@ -0,0 +1,66 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" 
+          "http://www.w3.org/TR/html4/strict.dtd">
+<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
+<html>
+<head> <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
+  <title>Polly - Performance</title>
+  <link type="text/css" rel="stylesheet" href="menu.css">
+  <link type="text/css" rel="stylesheet" href="content.css">
+</head>
+<body>
+<!--#include virtual="menu.html.incl"-->
+<div id="content">
+<h1>Polly: Performance</h1>
+
+<p>To evaluate the performance benefits Polly currently provides we compiled the
+<a href="http://www.cse.ohio-state.edu/~pouchet/software/polybench/">Polybench
+2.0</a> benchmark suite.  Each benchmark was run with double precision floating
+point values on an Intel Core Xeon X5670 CPU @ 2.93GHz (12 cores, 24 thread)
+system. We used <a href="http://pocc.sf.net">PoCC</a> and the included <a
+href="http://pluto-compiler.sf.net">Pluto</a> transformations to optimize the
+code. The source code of Polly and LLVM/clang was checked out on 
+25/03/2011.</p>
+
+<p>The results shown were created fully automatically without manual
+interaction. We did not yet spend any time to tune the results. Hence
+further improvments may be achieved by tuning the code generated by Polly, the
+heuristics used by Pluto or by investigating if more code could be optimized.
+As Pluto was never used at such a low level, its heuristics are probably
+far from perfect. Another area where we expect larger performance improvements
+is the SIMD vector code generation. At the moment, it rarely yields to
+performance improvements, as we did not yet include vectorization in our
+heuristics. By changing this we should be able to significantly increase the
+number of test cases that show improvements.</p>
+
+<p>The polybench test suite contains computation kernels from linear algebra
+routines, stencil computations, image processing and data mining. Polly
+recognices the majority of them and is able to show good speedup. However,
+to show similar speedup on larger examples like the SPEC CPU benchmarks Polly
+still misses support for integer casts, variable-sized multi-dimensional arrays
+and probably several other construts. This support is necessary as such
+constructs appear in larger programs, but not in our limited test suite.
+
+<h2> Sequential runs</h2>
+
+For the sequential runs we used Polly to create a program structure that is
+optimized for data-locality. One of the major optimizations performed is tiling.
+The speedups shown are without the use of any multi-core parallelism. No
+additional hardware is used, but the single available core is used more
+efficiently.
+<h3> Small data size</h3>
+<img src="images/performance/sequential-small.png" /><br />
+<h3> Large data size</h3>
+<img src="images/performance/sequential-large.png" />
+<h2> Parallel runs</h2>
+For the parallel runs we used Polly to expose parallelism and to add calls to an
+OpenMP runtime library. With OpenMP we can use all 12 hardware cores
+instead of the single core that was used before. We can see that in several
+cases we obtain more than linear speedup. This additional speedup is due to
+improved data-locality.
+<h3> Small data size</h3>
+<img src="images/performance/parallel-small.png" /><br />
+<h3> Large data size</h3>
+<img src="images/performance/parallel-large.png" />
+</div>
+</body>
+</html>

Added: polly/trunk/www/publications.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/publications.html?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/publications.html (added)
+++ polly/trunk/www/publications.html Mon May  2 02:48:29 2011
@@ -0,0 +1,43 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" 
+          "http://www.w3.org/TR/html4/strict.dtd">
+<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
+<html>
+<head>
+  <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
+  <title>Polly - Publications</title>
+  <link type="text/css" rel="stylesheet" href="menu.css">
+  <link type="text/css" rel="stylesheet" href="content.css">
+</head>
+<body>
+<!--#include virtual="menu.html.incl"-->
+<div id="content">
+  <!--*********************************************************************-->
+  <h1>Polly: Publications</h1>
+  <!--*********************************************************************-->
+
+  <h2> 2011 </h2>
+  <ul>
+  <li>Polly - Polyhedral Optimization in LLVM<br />
+  Tobias Grosser, Hongbin Zheng, Ragesh Aloor, Andreas Simb&uuml;rger, Armin
+  Gr&ouml;&szlig;linger, Louis-No&euml;l Pouchet<br />
+  IMPACT at CGO 2011 <br />
+  <a
+  href="publications/grosser-impact-2011.pdf">Paper</a>, <a
+  href="publications/grosser-impact-2011-slides.pdf">Slides
+  </a>
+  </li>
+  </ul>
+  <h2> 2010 </h2>
+  <ul>
+  <li>Polly - Polyhedral Transformations for LLVM<br />
+  Tobias Grosser, Hongbin Zheng<br />
+  LLVM Developer Metting <br /><a
+  href="http://llvm.org/devmtg/2010-11/Grosser-Polly.pdf">Slides</a>, <a
+  href="http://llvm.org/devmtg/2010-11/videos/Grosser_Polly-desktop.mp4">Video
+  (Computer)</a>, <a
+  href="http://llvm.org/devmtg/2010-11/videos/Grosser_Polly-mobile.mp4">Video
+  (Mobile)</a></li>
+  </ul>
+</div>
+</body>
+</html>

Added: polly/trunk/www/publications/grosser-impact-2011-slides.pdf
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/publications/grosser-impact-2011-slides.pdf?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/publications/grosser-impact-2011-slides.pdf (added) and polly/trunk/www/publications/grosser-impact-2011-slides.pdf Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/publications/grosser-impact-2011.pdf
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/publications/grosser-impact-2011.pdf?rev=130689&view=auto
==============================================================================
Binary files polly/trunk/www/publications/grosser-impact-2011.pdf (added) and polly/trunk/www/publications/grosser-impact-2011.pdf Mon May  2 02:48:29 2011 differ

Added: polly/trunk/www/todo.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/todo.html?rev=130689&view=auto
==============================================================================
--- polly/trunk/www/todo.html (added)
+++ polly/trunk/www/todo.html Mon May  2 02:48:29 2011
@@ -0,0 +1,363 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" 
+          "http://www.w3.org/TR/html4/strict.dtd">
+<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
+<html>
+<head> <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
+  <title>Polly - Todo</title>
+  <link type="text/css" rel="stylesheet" href="menu.css">
+  <link type="text/css" rel="stylesheet" href="content.css">
+</head>
+<body>
+<!--#include virtual="menu.html.incl"-->
+<div id="content">
+<h3> Setup infrastructure at LLVM </h3>
+
+<p>We are currently moving to the LLVM infrastructure
+</p>
+<table class="wikitable" cellpadding="2">
+
+<tbody><tr>
+<th width="400px"> Task
+</th><th width="150px"> Status
+</th><th>Owner
+</th></tr>
+<tr>
+<th align="left"> Move to LLVM svn
+</th><td align="center"> done
+</td><td> Tobias
+
+</td></tr>
+<tr>
+<th align="left"> Get git mirror at llvm.org/git/polly.git
+</th><td align="center"> done
+</td><td> Tobias
+</td></tr>
+<tr>
+<th align="left"> Setup on commit mails
+</th><td align="center">
+</td><td> Tobias
+</td></tr>
+<tr>
+
+<th align="left"> Add LLVM Bugzilla category
+</th></tr>
+<tr>
+<th align="left"> Create polly.llvm.org website
+</th><td>
+</td></tr>
+<tr>
+<th align="left"> Add Polly buildbot that runs 'polly-test'
+</th><td>
+</td></tr>
+<tr>
+<th align="left"> Run nightly performance/coverage tests with the llvm
+test-suite
+</th><td>
+
+</td></tr>
+</tbody></table>
+<h3> Phase 2 </h3>
+<p>The second phase of Polly can build on a robust, but very limited framework.
+In this phase work on removing limitations and extending the framework is
+planned. Also we plan the first very simple transformations. Furthermore the
+build system will be improved to simplify deployment.
+</p>
+<table class="wikitable" cellpadding="2">
+
+<tbody><tr>
+
+<th colspan="3" style="background: none repeat scroll 0% 0% rgb(239, 239,
+239);"> Frontend
+</th></tr>
+<tr>
+<th width="400px"> Task
+</th><th width="150px"> Status
+</th><th>Owner
+</th></tr>
+<tr>
+<th align="left"> Support for casts in expressions
+</th><td>
+</td><td>
+</td><td>
+</td></tr>
+
+<tr>
+<th align="left"> Support multi dimensional arrays.
+</th><td align="center"> planning
+</td><td> Tobias
+</td></tr>
+<tr>
+<th align="left"> Alias sets
+</th></tr>
+<tr>
+<th colspan="4" style="background: none repeat scroll 0% 0% rgb(239, 239,
+239);"> Middle end
+</th></tr>
+<tr>
+
+<th width="400px"> Task
+</th><th width="80px"> Status
+</th><th>Owner
+</th></tr>
+<tr>
+<th align="left"> Implement ISL dependency analysis pass
+</th><td align="center"> working
+</td><td> Tobias
+</td></tr>
+<tr>
+<th align="left"> Connect pocc/pluto
+
+</th><td align="center"> working
+</td><td> Tobias
+</td></tr>
+<tr>
+<th align="left"> Finish OpenSCoP support
+</th><td>
+</td><td>
+</td><td>
+</td></tr>
+
+<tr>
+<th align="left"> Add SCoPLib 0.2 support to connect pocc
+</th><td align="center">done
+
+</td><td> Tobias
+</td></tr>
+<tr>
+<th align="left"> Write simple loop blocking
+</th><td>
+</td><td>
+</td><td>
+</td></tr>
+<tr>
+<th colspan="4" style="background: none repeat scroll 0% 0% rgb(239, 239,
+239);"> Backend
+</th></tr>
+<tr>
+<th width="400px"> Task
+
+</th><th width="80px"> Status
+</th><th>Owner
+</th></tr>
+<tr>
+<th align="left"> Code generation for non 64bit targets
+</th><td>
+</td><td>
+</td><td>
+</td></tr>
+<tr>
+<th align="left"> Add write support for data access functions
+</th><td>
+</td><td>
+</td><td>
+
+</td></tr>
+<tr>
+<th align="left"> Create vector loops
+</th><td align="center">70% done
+</td><td>Tobias
+</td></tr>
+<tr>
+<th align="left">Create OpenMP
+loops
+</th><td align="center">90% done
+</td><td> Raghesh &amp; Tobias
+
+</td></tr>
+<tr>
+<th colspan="4" style="background: none repeat scroll 0% 0% rgb(239, 239,
+239);"> General tasks
+</th></tr>
+<tr>
+<th width="300px"> Task
+</th><th width="80px"> Status
+</th><th>Owner
+</th></tr>
+
+<tr>
+<th align="left"> Commit RegionPass patch upstream
+</th><td align="center"> done
+
+</td><td> Tobias
+</td></tr>
+<tr>
+<th align="left"> Build against an installed LLVM
+</th><td> works for cmake
+</td><td> ether
+</td></tr>
+<tr>
+<th align="left"> Setup buildbot regression testers using LNT
+</th><td>
+</td><td>
+</td><td>
+
+</td></tr>
+</tbody></table>
+<h3>Phase 1 - Get something
+working (Finished October 2010)</span></h3>
+<p>The first iteration of this project aims to create a minimal working version
+of this framework, that is capable to transform an LLVM-IR program to the
+polyhedral model
+and back to LLVM-IR without applying any transformations.
+</p>
+<table class="wikitable" cellpadding="2">
+
+<tbody><tr>
+<th colspan="4" style="background: none repeat scroll 0% 0% rgb(239, 239,
+239);"> Frontend
+</th></tr>
+
+<tr>
+<th width="300px"> Task
+</th><th width="150px"> Status
+</th><th>Owner
+</th></tr>
+<tr>
+<th align="left"> Region detection
+</td><td> works + Committed upstream
+</td><td>Ether
+</td></tr>
+<tr>
+<th align="left"> Access Functions
+</td><td> Working
+</td><td> John + Ether
+</td></tr>
+<tr>
+<th align="left"> Alias sets
+</td><td> Still open
+</td></tr>
+<tr>
+<th align="left"> Scalar evolution to affine expression
+</td><td> Done
+
+</td><td>
+Ether
+</td></tr>
+<tr>
+<th align="left"> SCoP extraction
+</td><td> Working
+</td><td>Tobias + Ether
+
+</td></tr>
+<tr>
+<th align="left"> SCoPs to polyhedral model
+</td><td> Working
+</td><td>Tobias + Ether
+</td></tr>
+<tr>
+<th colspan="4" style="background: none repeat scroll 0% 0% rgb(239, 239,
+239);"> Middle end
+
+</th></tr>
+<tr>
+<th width="300px"> Task
+</th><th width="80px"> Status
+</th><th>Owner
+</th></tr>
+<tr>
+<th align="left"> Define polyhedral description
+</td><td> Working
+</td><td>Tobias
+
+</td></tr>
+<tr>
+<th align="left"> Import/Export using openscop
+</td><td> working
+</td><td>Tobias
+</td></tr>
+<tr>
+<th colspan="4" style="background: none repeat scroll 0% 0% rgb(239, 239,
+239);"> Backend
+</th></tr>
+<tr>
+
+<th width="300px"> Task
+</th><th width="80px"> Status
+</th><th>Owner
+</th></tr>
+<tr>
+<th align="left"> Create LLVM-IR using CLooG
+</td><td> Working
+</td><td> Tobias
+
+</td></tr>
+<tr>
+<th colspan="4" style="background: none repeat scroll 0% 0% rgb(239, 239,
+239);"> General tasks
+</th></tr>
+<tr>
+<th width="300px"> Task
+</th><th width="80px"> Status
+</th><th>Owner
+</th></tr>
+<tr>
+<th align="left"> Setup git repositories
+
+</td><td> Done
+</td><td> Tobias
+</td></tr>
+<tr>
+<th align="left"> Add CLooG/isl to build system
+</td><td> Works on Unix
+</td><td> Tobias
+
+</td></tr></tbody></table>
+<h3>Further projects </h3>
+<p>There are several great projects related to polly that can already be started
+or are possible in the near future.  </p>
+<h4>Extend the post dominance analysis for infinite loops (small)</h4>
+
+<p>At the moment the post dominance analysis cannot handle infinite loops. All
+basic blocks in the CFG that do not return are - at the moment - not part of the
+post dominance tree.
+However by adding some virtual edges, they could be added to the post dominator
+tree. Where to add the edges needs some research.
+</p><p>This is a small project, that is is well defined. As it is directly in
+LLVM it can be easily committed upstream. It is useful for polly, as the
+RegionInfo pass will be able to detect regions in parts of the CFG that never
+return.
+</p><p><i>A good starter to get into LLVM</i>
+</p>
+<h4>Vectorization </h4>
+<p>It is planned to use Polly to support vectorization in LLVM.
+</p><p>The basic idea is to use Polly and the polyhedral tools to transform code
+such that the innermost loops can be executed in parallel. Afterwards during
+code generation in LLVM the loops will be created using vector instructions.
+</p><p>To start we plan to use <a href="http://pluto-compiler.sf.net"
+class="external text" title="http://pluto-compiler.sf.net"
+rel="nofollow">Pluto</a> to transform the loop nests. Pluto can generate vector
+parallel code and annotate the vector parallel loops. Some impressive results
+were shown on code that was afterwards vectorized by the icc enforcing
+vectorization. We believe LLVM can do even better, as it can interact directly
+with the polyhedral information.
+
+</p><p>As an example simple matrix multiplication:
+</p>
+<pre>
+for(i=0; i&lt;M; i++)
+  for(j=0; j&lt;N; j++)
+    for(k=0; k&lt;K; k++)
+      C[i][j] = beta*C[i][j] + alpha*A[i][k] * B[k][j];
+</pre>
+<p>After plutos transformations with added tiling and
+vectorization hints:
+</p>
+<pre>
+if ((K &gt;= 1) &amp;&amp; (M &gt;= 1) &amp;&amp; (N &gt;= 1))
+  for (t1=0;t1&lt;=floord(M-1,32);t1++)
+    for (t2=0;t2&lt;=floord(N-1,32);t2++)
+      for (t3=0;t3&lt;=floord(K-1,32);t3++)
+        for (t4=32*t1;t4&lt;=min(M-1,32*t1+31);t4++)
+          for (t5=32*t3;t5&lt;=min(K-1,32*t3+31);t5++) {
+            lbv=32*t2;
+            ubv=min(N-1,32*t2+31);
+            #pragma ivdep
+            #pragma vector always
+            for (t6=lbv; t6&lt;=ubv; t6++)
+              C[t4][t6]=beta*C[t4][t6]+alpha*A[t4][t5]*B[t5][t6];;
+         }
+</pre>
+<p>In this example the innermost loop is parallel without any dependencies. </p>
+</div>
+</body>
+</html>



From rafael.espindola at gmail.com  Mon May  2 09:23:41 2011
From: rafael.espindola at gmail.com (Rafael Avila de Espindola)
Date: Mon, 02 May 2011 10:23:41 -0400
Subject: [llvm-commits] [llvm] r130611 -
	/llvm/trunk/lib/MC/MCAsmStreamer.cpp
In-Reply-To: <281BED74-657D-4615-9F0F-C3654FA64166@apple.com>
References: <20110430163457.CB27A2A6C138@llvm.org>
	<281BED74-657D-4615-9F0F-C3654FA64166@apple.com>
Message-ID: <4DBEBE6D.2050507@gmail.com>

> Hey Rafael,
>
> I'm loving your work on the CFI directives.  Can you add a
> verbose-asm comment to directives that makes sense?  It would be
> great to print out the register name in ".cfi_offset" etc.

I think we can do a bit better than that. Roman has added support for 
using symbolic register names in the directives themselves (gas also 
supports this). I created PR9826 to track this.

> Thanks again for working on this!

My pleasure. Looking forward to enabling relaxations to get smaller eh 
frames!

> -Chris

Cheers,
Rafael


From rafael.espindola at gmail.com  Mon May  2 10:00:52 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Mon, 02 May 2011 15:00:52 -0000
Subject: [llvm-commits] [llvm] r130690 - in /llvm/trunk:
 lib/Target/PowerPC/PPCMCAsmInfo.cpp test/CodeGen/PowerPC/2008-12-12-EH.ll
 test/CodeGen/PowerPC/ppc-prologue.ll
Message-ID: <20110502150052.A9FD22A6C130@llvm.org>

Author: rafael
Date: Mon May  2 10:00:52 2011
New Revision: 130690

URL: http://llvm.org/viewvc/llvm-project?rev=130690&view=rev
Log:
Move ppc OS X to cfi too. I am building it on an old ppc mini, but it will take some time.

Modified:
    llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp
    llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll
    llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll

Modified: llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp?rev=130690&r1=130689&r2=130690&view=diff
==============================================================================
--- llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp (original)
+++ llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp Mon May  2 10:00:52 2011
@@ -17,7 +17,7 @@
 PPCMCAsmInfoDarwin::PPCMCAsmInfoDarwin(bool is64Bit) {
   PCSymbol = ".";
   CommentString = ";";
-  ExceptionsType = ExceptionHandling::DwarfTable;
+  ExceptionsType = ExceptionHandling::DwarfCFI;
 
   if (!is64Bit)
     Data64bitsDirective = 0;      // We can't emit a 64-bit unit in PPC32 mode.

Modified: llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll?rev=130690&r1=130689&r2=130690&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll Mon May  2 10:00:52 2011
@@ -1,4 +1,4 @@
-; RUN: llc < %s  -march=ppc32 -mtriple=powerpc-apple-darwin9 | grep ^__Z1fv.eh
+; RUN: llc < %s -disable-cfi -march=ppc32 -mtriple=powerpc-apple-darwin9 | grep ^__Z1fv.eh
 
 define void @_Z1fv() {
 entry:

Modified: llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll?rev=130690&r1=130689&r2=130690&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll Mon May  2 10:00:52 2011
@@ -5,9 +5,7 @@
 ; CHECK-NEXT:  stw r31, -4(r1)
 ; CHECK-NEXT:  stw r0, 8(r1)
 ; CHECK-NEXT:  stwu r1, -80(r1)
-; CHECK-NEXT: Ltmp0:
-; CHECK-NEXT:  mr r31, r1
-; CHECK-NEXT: Ltmp1:
+; CHECK:  mr r31, r1
 entry:
   %a_addr = alloca i32                            ; <i32*> [#uses=2]
   %retval = alloca i32                            ; <i32*> [#uses=2]



From rafael.espindola at gmail.com  Mon May  2 10:42:31 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Mon, 02 May 2011 15:42:31 -0000
Subject: [llvm-commits] [llvm] r130691 - in /llvm/trunk:
 lib/Target/PowerPC/PPCMCAsmInfo.cpp test/CodeGen/PowerPC/2008-12-12-EH.ll
 test/CodeGen/PowerPC/ppc-prologue.ll
Message-ID: <20110502154231.EC7962A6C12C@llvm.org>

Author: rafael
Date: Mon May  2 10:42:31 2011
New Revision: 130691

URL: http://llvm.org/viewvc/llvm-project?rev=130691&view=rev
Log:
Revert while I debug the tests that use march but not mtriple.

Modified:
    llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp
    llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll
    llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll

Modified: llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp?rev=130691&r1=130690&r2=130691&view=diff
==============================================================================
--- llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp (original)
+++ llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp Mon May  2 10:42:31 2011
@@ -17,7 +17,7 @@
 PPCMCAsmInfoDarwin::PPCMCAsmInfoDarwin(bool is64Bit) {
   PCSymbol = ".";
   CommentString = ";";
-  ExceptionsType = ExceptionHandling::DwarfCFI;
+  ExceptionsType = ExceptionHandling::DwarfTable;
 
   if (!is64Bit)
     Data64bitsDirective = 0;      // We can't emit a 64-bit unit in PPC32 mode.

Modified: llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll?rev=130691&r1=130690&r2=130691&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll Mon May  2 10:42:31 2011
@@ -1,4 +1,4 @@
-; RUN: llc < %s -disable-cfi -march=ppc32 -mtriple=powerpc-apple-darwin9 | grep ^__Z1fv.eh
+; RUN: llc < %s  -march=ppc32 -mtriple=powerpc-apple-darwin9 | grep ^__Z1fv.eh
 
 define void @_Z1fv() {
 entry:

Modified: llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll?rev=130691&r1=130690&r2=130691&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll Mon May  2 10:42:31 2011
@@ -5,7 +5,9 @@
 ; CHECK-NEXT:  stw r31, -4(r1)
 ; CHECK-NEXT:  stw r0, 8(r1)
 ; CHECK-NEXT:  stwu r1, -80(r1)
-; CHECK:  mr r31, r1
+; CHECK-NEXT: Ltmp0:
+; CHECK-NEXT:  mr r31, r1
+; CHECK-NEXT: Ltmp1:
 entry:
   %a_addr = alloca i32                            ; <i32*> [#uses=2]
   %retval = alloca i32                            ; <i32*> [#uses=2]



From rafael.espindola at gmail.com  Mon May  2 10:49:52 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Mon, 02 May 2011 15:49:52 -0000
Subject: [llvm-commits] [llvm] r130692 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCFIException.cpp
Message-ID: <20110502154952.98DE52A6C12C@llvm.org>

Author: rafael
Date: Mon May  2 10:49:52 2011
New Revision: 130692

URL: http://llvm.org/viewvc/llvm-project?rev=130692&view=rev
Log:
Only produce the eh_frame section if we have at least one personality function.

Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCFIException.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCFIException.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCFIException.cpp?rev=130692&r1=130691&r2=130692&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCFIException.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCFIException.cpp Mon May  2 10:49:52 2011
@@ -54,25 +54,27 @@
 
   const TargetLoweringObjectFile &TLOF = Asm->getObjFileLowering();
 
-  if (!TLOF.isFunctionEHFrameSymbolPrivate()) {
-    // This is a temporary hack to keep sections in the same order they
-    // were before. This lets us produce bit identical outputs while
-    // transitioning to CFI.
-    Asm->OutStreamer.SwitchSection(TLOF.getEHFrameSection());
-  }
-
   unsigned PerEncoding = TLOF.getPersonalityEncoding();
 
   if ((PerEncoding & 0x70) != dwarf::DW_EH_PE_pcrel)
     return;
 
   // Emit references to all used personality functions
+  bool AtLeastOne = false;
   const std::vector<const Function*> &Personalities = MMI->getPersonalities();
   for (size_t i = 0, e = Personalities.size(); i != e; ++i) {
     if (!Personalities[i])
       continue;
     MCSymbol *Sym = Asm->Mang->getSymbol(Personalities[i]);
     TLOF.emitPersonalityValue(Asm->OutStreamer, Asm->TM, Sym);
+    AtLeastOne = true;
+  }
+
+  if (AtLeastOne && !TLOF.isFunctionEHFrameSymbolPrivate()) {
+    // This is a temporary hack to keep sections in the same order they
+    // were before. This lets us produce bit identical outputs while
+    // transitioning to CFI.
+    Asm->OutStreamer.SwitchSection(TLOF.getEHFrameSection());
   }
 }
 



From rafael.espindola at gmail.com  Mon May  2 10:58:16 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Mon, 02 May 2011 15:58:16 -0000
Subject: [llvm-commits] [llvm] r130693 - in /llvm/trunk:
 lib/Target/PowerPC/PPCMCAsmInfo.cpp test/CodeGen/PowerPC/2008-12-12-EH.ll
 test/CodeGen/PowerPC/ppc-prologue.ll
Message-ID: <20110502155816.BB60B2A6C12C@llvm.org>

Author: rafael
Date: Mon May  2 10:58:16 2011
New Revision: 130693

URL: http://llvm.org/viewvc/llvm-project?rev=130693&view=rev
Log:
Add 130690 back.

Modified:
    llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp
    llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll
    llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll

Modified: llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp?rev=130693&r1=130692&r2=130693&view=diff
==============================================================================
--- llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp (original)
+++ llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp Mon May  2 10:58:16 2011
@@ -17,7 +17,7 @@
 PPCMCAsmInfoDarwin::PPCMCAsmInfoDarwin(bool is64Bit) {
   PCSymbol = ".";
   CommentString = ";";
-  ExceptionsType = ExceptionHandling::DwarfTable;
+  ExceptionsType = ExceptionHandling::DwarfCFI;
 
   if (!is64Bit)
     Data64bitsDirective = 0;      // We can't emit a 64-bit unit in PPC32 mode.

Modified: llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll?rev=130693&r1=130692&r2=130693&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/2008-12-12-EH.ll Mon May  2 10:58:16 2011
@@ -1,4 +1,4 @@
-; RUN: llc < %s  -march=ppc32 -mtriple=powerpc-apple-darwin9 | grep ^__Z1fv.eh
+; RUN: llc < %s -disable-cfi -march=ppc32 -mtriple=powerpc-apple-darwin9 | grep ^__Z1fv.eh
 
 define void @_Z1fv() {
 entry:

Modified: llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll?rev=130693&r1=130692&r2=130693&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/ppc-prologue.ll Mon May  2 10:58:16 2011
@@ -5,9 +5,7 @@
 ; CHECK-NEXT:  stw r31, -4(r1)
 ; CHECK-NEXT:  stw r0, 8(r1)
 ; CHECK-NEXT:  stwu r1, -80(r1)
-; CHECK-NEXT: Ltmp0:
-; CHECK-NEXT:  mr r31, r1
-; CHECK-NEXT: Ltmp1:
+; CHECK:  mr r31, r1
 entry:
   %a_addr = alloca i32                            ; <i32*> [#uses=2]
   %retval = alloca i32                            ; <i32*> [#uses=2]



From baldrick at free.fr  Mon May  2 11:27:02 2011
From: baldrick at free.fr (Duncan Sands)
Date: Mon, 02 May 2011 16:27:02 -0000
Subject: [llvm-commits] [llvm] r130694 - in /llvm/trunk:
 lib/Analysis/InstructionSimplify.cpp
 lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
 test/Transforms/InstSimplify/rem.ll
Message-ID: <20110502162702.CC0B72A6C12D@llvm.org>

Author: baldrick
Date: Mon May  2 11:27:02 2011
New Revision: 130694

URL: http://llvm.org/viewvc/llvm-project?rev=130694&view=rev
Log:
Move some rem transforms out of instcombine and into instsimplify.
This automagically provides a transform noticed by my super-optimizer
as occurring quite often: "rem x, (select cond, x, 1)" -> 0.

Added:
    llvm/trunk/test/Transforms/InstSimplify/rem.ll
Modified:
    llvm/trunk/lib/Analysis/InstructionSimplify.cpp
    llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp

Modified: llvm/trunk/lib/Analysis/InstructionSimplify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/InstructionSimplify.cpp?rev=130694&r1=130693&r2=130694&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/InstructionSimplify.cpp (original)
+++ llvm/trunk/lib/Analysis/InstructionSimplify.cpp Mon May  2 11:27:02 2011
@@ -901,6 +901,111 @@
   return ::SimplifyFDivInst(Op0, Op1, TD, DT, RecursionLimit);
 }
 
+/// SimplifyRem - Given operands for an SRem or URem, see if we can
+/// fold the result.  If not, this returns null.
+static Value *SimplifyRem(Instruction::BinaryOps Opcode, Value *Op0, Value *Op1,
+                          const TargetData *TD, const DominatorTree *DT,
+                          unsigned MaxRecurse) {
+  if (Constant *C0 = dyn_cast<Constant>(Op0)) {
+    if (Constant *C1 = dyn_cast<Constant>(Op1)) {
+      Constant *Ops[] = { C0, C1 };
+      return ConstantFoldInstOperands(Opcode, C0->getType(), Ops, 2, TD);
+    }
+  }
+
+  bool isSigned = Opcode == Instruction::SRem;
+
+  // X % undef -> undef
+  if (match(Op1, m_Undef()))
+    return Op1;
+
+  // undef % X -> 0
+  if (match(Op0, m_Undef()))
+    return Constant::getNullValue(Op0->getType());
+
+  // 0 % X -> 0, we don't need to preserve faults!
+  if (match(Op0, m_Zero()))
+    return Op0;
+
+  // X % 0 -> undef, we don't need to preserve faults!
+  if (match(Op1, m_Zero()))
+    return UndefValue::get(Op0->getType());
+
+  // X % 1 -> 0
+  if (match(Op1, m_One()))
+    return Constant::getNullValue(Op0->getType());
+
+  if (Op0->getType()->isIntegerTy(1))
+    // It can't be remainder by zero, hence it must be remainder by one.
+    return Constant::getNullValue(Op0->getType());
+
+  // X % X -> 0
+  if (Op0 == Op1)
+    return Constant::getNullValue(Op0->getType());
+
+  // If the operation is with the result of a select instruction, check whether
+  // operating on either branch of the select always yields the same value.
+  if (isa<SelectInst>(Op0) || isa<SelectInst>(Op1))
+    if (Value *V = ThreadBinOpOverSelect(Opcode, Op0, Op1, TD, DT, MaxRecurse))
+      return V;
+
+  // If the operation is with the result of a phi instruction, check whether
+  // operating on all incoming values of the phi always yields the same value.
+  if (isa<PHINode>(Op0) || isa<PHINode>(Op1))
+    if (Value *V = ThreadBinOpOverPHI(Opcode, Op0, Op1, TD, DT, MaxRecurse))
+      return V;
+
+  return 0;
+}
+
+/// SimplifySRemInst - Given operands for an SRem, see if we can
+/// fold the result.  If not, this returns null.
+static Value *SimplifySRemInst(Value *Op0, Value *Op1, const TargetData *TD,
+                               const DominatorTree *DT, unsigned MaxRecurse) {
+  if (Value *V = SimplifyRem(Instruction::SRem, Op0, Op1, TD, DT, MaxRecurse))
+    return V;
+
+  return 0;
+}
+
+Value *llvm::SimplifySRemInst(Value *Op0, Value *Op1, const TargetData *TD,
+                              const DominatorTree *DT) {
+  return ::SimplifySRemInst(Op0, Op1, TD, DT, RecursionLimit);
+}
+
+/// SimplifyURemInst - Given operands for a URem, see if we can
+/// fold the result.  If not, this returns null.
+static Value *SimplifyURemInst(Value *Op0, Value *Op1, const TargetData *TD,
+                               const DominatorTree *DT, unsigned MaxRecurse) {
+  if (Value *V = SimplifyRem(Instruction::URem, Op0, Op1, TD, DT, MaxRecurse))
+    return V;
+
+  return 0;
+}
+
+Value *llvm::SimplifyURemInst(Value *Op0, Value *Op1, const TargetData *TD,
+                              const DominatorTree *DT) {
+  return ::SimplifyURemInst(Op0, Op1, TD, DT, RecursionLimit);
+}
+
+static Value *SimplifyFRemInst(Value *Op0, Value *Op1, const TargetData *,
+                               const DominatorTree *, unsigned) {
+  // undef % X -> undef    (the undef could be a snan).
+  if (match(Op0, m_Undef()))
+    return Op0;
+
+  // X % undef -> undef
+  if (match(Op1, m_Undef()))
+    return Op1;
+
+  return 0;
+}
+
+Value *llvm::SimplifyFRemInst(Value *Op0, Value *Op1, const TargetData *TD,
+                              const DominatorTree *DT) {
+  return ::SimplifyFRemInst(Op0, Op1, TD, DT, RecursionLimit);
+}
+
 /// SimplifyShift - Given operands for an Shl, LShr or AShr, see if we can
 /// fold the result.  If not, this returns null.
 static Value *SimplifyShift(unsigned Opcode, Value *Op0, Value *Op1,
@@ -1994,6 +2099,9 @@
   case Instruction::SDiv: return SimplifySDivInst(LHS, RHS, TD, DT, MaxRecurse);
   case Instruction::UDiv: return SimplifyUDivInst(LHS, RHS, TD, DT, MaxRecurse);
   case Instruction::FDiv: return SimplifyFDivInst(LHS, RHS, TD, DT, MaxRecurse);
+  case Instruction::SRem: return SimplifySRemInst(LHS, RHS, TD, DT, MaxRecurse);
+  case Instruction::URem: return SimplifyURemInst(LHS, RHS, TD, DT, MaxRecurse);
+  case Instruction::FRem: return SimplifyFRemInst(LHS, RHS, TD, DT, MaxRecurse);
   case Instruction::Shl:
     return SimplifyShlInst(LHS, RHS, /*isNSW*/false, /*isNUW*/false,
                            TD, DT, MaxRecurse);
@@ -2088,6 +2196,15 @@
   case Instruction::FDiv:
     Result = SimplifyFDivInst(I->getOperand(0), I->getOperand(1), TD, DT);
     break;
+  case Instruction::SRem:
+    Result = SimplifySRemInst(I->getOperand(0), I->getOperand(1), TD, DT);
+    break;
+  case Instruction::URem:
+    Result = SimplifyURemInst(I->getOperand(0), I->getOperand(1), TD, DT);
+    break;
+  case Instruction::FRem:
+    Result = SimplifyFRemInst(I->getOperand(0), I->getOperand(1), TD, DT);
+    break;
   case Instruction::Shl:
     Result = SimplifyShlInst(I->getOperand(0), I->getOperand(1),
                              cast<BinaryOperator>(I)->hasNoSignedWrap(),

Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp?rev=130694&r1=130693&r2=130694&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp (original)
+++ llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp Mon May  2 11:27:02 2011
@@ -492,28 +492,6 @@
   return 0;
 }
 
-/// This function implements the transforms on rem instructions that work
-/// regardless of the kind of rem instruction it is (urem, srem, or frem). It 
-/// is used by the visitors to those instructions.
-/// @brief Transforms common to all three rem instructions
-Instruction *InstCombiner::commonRemTransforms(BinaryOperator &I) {
-  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
-
-  if (isa<UndefValue>(Op0)) {             // undef % X -> 0
-    if (I.getType()->isFPOrFPVectorTy())
-      return ReplaceInstUsesWith(I, Op0);  // X % undef -> undef (could be SNaN)
-    return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
-  }
-  if (isa<UndefValue>(Op1))
-    return ReplaceInstUsesWith(I, Op1);  // X % undef -> undef
-
-  // Handle cases involving: rem X, (select Cond, Y, Z)
-  if (isa<SelectInst>(Op1) && SimplifyDivRemOfSelect(I))
-    return &I;
-
-  return 0;
-}
-
 /// This function implements the transforms common to both integer remainder
 /// instructions (urem and srem). It is called by the visitors to those integer
 /// remainder instructions.
@@ -521,26 +499,11 @@
 Instruction *InstCombiner::commonIRemTransforms(BinaryOperator &I) {
   Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
 
-  if (Instruction *common = commonRemTransforms(I))
-    return common;
-
-  // X % X == 0
-  if (Op0 == Op1)
-    return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
-
-  // 0 % X == 0 for integer, we don't need to preserve faults!
-  if (Constant *LHS = dyn_cast<Constant>(Op0))
-    if (LHS->isNullValue())
-      return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
+  // Handle cases involving: rem X, (select Cond, Y, Z)
+  if (isa<SelectInst>(Op1) && SimplifyDivRemOfSelect(I))
+    return &I;
 
   if (ConstantInt *RHS = dyn_cast<ConstantInt>(Op1)) {
-    // X % 0 == undef, we don't need to preserve faults!
-    if (RHS->equalsInt(0))
-      return ReplaceInstUsesWith(I, UndefValue::get(I.getType()));
-    
-    if (RHS->equalsInt(1))  // X % 1 == 0
-      return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
-
     if (Instruction *Op0I = dyn_cast<Instruction>(Op0)) {
       if (SelectInst *SI = dyn_cast<SelectInst>(Op0I)) {
         if (Instruction *R = FoldOpIntoSelect(I, SI))
@@ -562,6 +525,9 @@
 Instruction *InstCombiner::visitURem(BinaryOperator &I) {
   Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
 
+  if (Value *V = SimplifyURemInst(Op0, Op1, TD))
+    return ReplaceInstUsesWith(I, V);
+
   if (Instruction *common = commonIRemTransforms(I))
     return common;
   
@@ -602,6 +568,9 @@
 Instruction *InstCombiner::visitSRem(BinaryOperator &I) {
   Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
 
+  if (Value *V = SimplifySRemInst(Op0, Op1, TD))
+    return ReplaceInstUsesWith(I, V);
+
   // Handle the integer rem common cases
   if (Instruction *Common = commonIRemTransforms(I))
     return Common;
@@ -660,6 +629,14 @@
 }
 
 Instruction *InstCombiner::visitFRem(BinaryOperator &I) {
-  return commonRemTransforms(I);
-}
+  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
 
+  if (Value *V = SimplifyFRemInst(Op0, Op1, TD))
+    return ReplaceInstUsesWith(I, V);
+
+  // Handle cases involving: rem X, (select Cond, Y, Z)
+  if (isa<SelectInst>(Op1) && SimplifyDivRemOfSelect(I))
+    return &I;
+
+  return 0;
+}

Added: llvm/trunk/test/Transforms/InstSimplify/rem.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstSimplify/rem.ll?rev=130694&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/InstSimplify/rem.ll (added)
+++ llvm/trunk/test/Transforms/InstSimplify/rem.ll Mon May  2 11:27:02 2011
@@ -0,0 +1,17 @@
+; RUN: opt < %s -instsimplify -S | FileCheck %s
+
+define i32 @select1(i32 %x, i1 %b) {
+; CHECK: @select1
+  %rhs = select i1 %b, i32 %x, i32 1
+  %rem = srem i32 %x, %rhs
+  ret i32 %rem
+; CHECK: ret i32 0
+}
+
+define i32 @select2(i32 %x, i1 %b) {
+; CHECK: @select2
+  %rhs = select i1 %b, i32 %x, i32 1
+  %rem = urem i32 %x, %rhs
+  ret i32 %rem
+; CHECK: ret i32 0
+}



From baldrick at free.fr  Mon May  2 11:34:19 2011
From: baldrick at free.fr (Duncan Sands)
Date: Mon, 02 May 2011 16:34:19 -0000
Subject: [llvm-commits] [llvm] r130695 -
	/llvm/trunk/include/llvm/Analysis/InstructionSimplify.h
Message-ID: <20110502163419.59CBC2A6C12C@llvm.org>

Author: baldrick
Date: Mon May  2 11:34:19 2011
New Revision: 130695

URL: http://llvm.org/viewvc/llvm-project?rev=130695&view=rev
Log:
Forgot to commit the changes to this file.

Modified:
    llvm/trunk/include/llvm/Analysis/InstructionSimplify.h

Modified: llvm/trunk/include/llvm/Analysis/InstructionSimplify.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Analysis/InstructionSimplify.h?rev=130695&r1=130694&r2=130695&view=diff
==============================================================================
--- llvm/trunk/include/llvm/Analysis/InstructionSimplify.h (original)
+++ llvm/trunk/include/llvm/Analysis/InstructionSimplify.h Mon May  2 11:34:19 2011
@@ -55,6 +55,21 @@
   Value *SimplifyFDivInst(Value *LHS, Value *RHS, const TargetData *TD = 0,
                           const DominatorTree *DT = 0);
 
+  /// SimplifySRemInst - Given operands for an SRem, see if we can
+  /// fold the result.  If not, this returns null.
+  Value *SimplifySRemInst(Value *LHS, Value *RHS, const TargetData *TD = 0,
+                          const DominatorTree *DT = 0);
+
+  /// SimplifyURemInst - Given operands for a URem, see if we can
+  /// fold the result.  If not, this returns null.
+  Value *SimplifyURemInst(Value *LHS, Value *RHS, const TargetData *TD = 0,
+                          const DominatorTree *DT = 0);
+
+  /// SimplifyFRemInst - Given operands for an FRem, see if we can
+  /// fold the result.  If not, this returns null.
+  Value *SimplifyFRemInst(Value *LHS, Value *RHS, const TargetData *TD = 0,
+                          const DominatorTree *DT = 0);
+
   /// SimplifyShlInst - Given operands for a Shl, see if we can
   /// fold the result.  If not, this returns null.
   Value *SimplifyShlInst(Value *Op0, Value *Op1, bool isNSW, bool isNUW,



From echristo at apple.com  Mon May  2 12:16:43 2011
From: echristo at apple.com (Eric Christopher)
Date: Mon, 02 May 2011 10:16:43 -0700
Subject: [llvm-commits] [llvm] r130694 - in /llvm/trunk:
 lib/Analysis/InstructionSimplify.cpp
 lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
 test/Transforms/InstSimplify/rem.ll
In-Reply-To: <20110502162702.CC0B72A6C12D@llvm.org>
References: <20110502162702.CC0B72A6C12D@llvm.org>
Message-ID: <EF1D4124-3623-4618-AF6A-CED99C2F633D@apple.com>

Hi Duncan,

This put a few warnings into the build:

/Volumes/Data/sources/llvm-clean/lib/Analysis/InstructionSimplify.cpp:916:8: warning: 
      unused variable 'isSigned' [-Wunused-variable]
  bool isSigned = Opcode == Instruction::SRem;


/Volumes/Data/sources/llvm-clean/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp:506:20: warning: 
      unused variable 'RHS' [-Wunused-variable]
  if (ConstantInt *RHS = dyn_cast<ConstantInt>(Op1)) {


Mind cleaning these up?

Thanks!

-eric

On May 2, 2011, at 9:27 AM, Duncan Sands wrote:

> Author: baldrick
> Date: Mon May  2 11:27:02 2011
> New Revision: 130694
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=130694&view=rev
> Log:
> Move some rem transforms out of instcombine and into instsimplify.
> This automagically provides a transform noticed by my super-optimizer
> as occurring quite often: "rem x, (select cond, x, 1)" -> 0.
> 
> Added:
>    llvm/trunk/test/Transforms/InstSimplify/rem.ll
> Modified:
>    llvm/trunk/lib/Analysis/InstructionSimplify.cpp
>    llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
> 
> Modified: llvm/trunk/lib/Analysis/InstructionSimplify.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/InstructionSimplify.cpp?rev=130694&r1=130693&r2=130694&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Analysis/InstructionSimplify.cpp (original)
> +++ llvm/trunk/lib/Analysis/InstructionSimplify.cpp Mon May  2 11:27:02 2011
> @@ -901,6 +901,111 @@
>   return ::SimplifyFDivInst(Op0, Op1, TD, DT, RecursionLimit);
> }
> 
> +/// SimplifyRem - Given operands for an SRem or URem, see if we can
> +/// fold the result.  If not, this returns null.
> +static Value *SimplifyRem(Instruction::BinaryOps Opcode, Value *Op0, Value *Op1,
> +                          const TargetData *TD, const DominatorTree *DT,
> +                          unsigned MaxRecurse) {
> +  if (Constant *C0 = dyn_cast<Constant>(Op0)) {
> +    if (Constant *C1 = dyn_cast<Constant>(Op1)) {
> +      Constant *Ops[] = { C0, C1 };
> +      return ConstantFoldInstOperands(Opcode, C0->getType(), Ops, 2, TD);
> +    }
> +  }
> +
> +  bool isSigned = Opcode == Instruction::SRem;
> +
> +  // X % undef -> undef
> +  if (match(Op1, m_Undef()))
> +    return Op1;
> +
> +  // undef % X -> 0
> +  if (match(Op0, m_Undef()))
> +    return Constant::getNullValue(Op0->getType());
> +
> +  // 0 % X -> 0, we don't need to preserve faults!
> +  if (match(Op0, m_Zero()))
> +    return Op0;
> +
> +  // X % 0 -> undef, we don't need to preserve faults!
> +  if (match(Op1, m_Zero()))
> +    return UndefValue::get(Op0->getType());
> +
> +  // X % 1 -> 0
> +  if (match(Op1, m_One()))
> +    return Constant::getNullValue(Op0->getType());
> +
> +  if (Op0->getType()->isIntegerTy(1))
> +    // It can't be remainder by zero, hence it must be remainder by one.
> +    return Constant::getNullValue(Op0->getType());
> +
> +  // X % X -> 0
> +  if (Op0 == Op1)
> +    return Constant::getNullValue(Op0->getType());
> +
> +  // If the operation is with the result of a select instruction, check whether
> +  // operating on either branch of the select always yields the same value.
> +  if (isa<SelectInst>(Op0) || isa<SelectInst>(Op1))
> +    if (Value *V = ThreadBinOpOverSelect(Opcode, Op0, Op1, TD, DT, MaxRecurse))
> +      return V;
> +
> +  // If the operation is with the result of a phi instruction, check whether
> +  // operating on all incoming values of the phi always yields the same value.
> +  if (isa<PHINode>(Op0) || isa<PHINode>(Op1))
> +    if (Value *V = ThreadBinOpOverPHI(Opcode, Op0, Op1, TD, DT, MaxRecurse))
> +      return V;
> +
> +  return 0;
> +}
> +
> +/// SimplifySRemInst - Given operands for an SRem, see if we can
> +/// fold the result.  If not, this returns null.
> +static Value *SimplifySRemInst(Value *Op0, Value *Op1, const TargetData *TD,
> +                               const DominatorTree *DT, unsigned MaxRecurse) {
> +  if (Value *V = SimplifyRem(Instruction::SRem, Op0, Op1, TD, DT, MaxRecurse))
> +    return V;
> +
> +  return 0;
> +}
> +
> +Value *llvm::SimplifySRemInst(Value *Op0, Value *Op1, const TargetData *TD,
> +                              const DominatorTree *DT) {
> +  return ::SimplifySRemInst(Op0, Op1, TD, DT, RecursionLimit);
> +}
> +
> +/// SimplifyURemInst - Given operands for a URem, see if we can
> +/// fold the result.  If not, this returns null.
> +static Value *SimplifyURemInst(Value *Op0, Value *Op1, const TargetData *TD,
> +                               const DominatorTree *DT, unsigned MaxRecurse) {
> +  if (Value *V = SimplifyRem(Instruction::URem, Op0, Op1, TD, DT, MaxRecurse))
> +    return V;
> +
> +  return 0;
> +}
> +
> +Value *llvm::SimplifyURemInst(Value *Op0, Value *Op1, const TargetData *TD,
> +                              const DominatorTree *DT) {
> +  return ::SimplifyURemInst(Op0, Op1, TD, DT, RecursionLimit);
> +}
> +
> +static Value *SimplifyFRemInst(Value *Op0, Value *Op1, const TargetData *,
> +                               const DominatorTree *, unsigned) {
> +  // undef % X -> undef    (the undef could be a snan).
> +  if (match(Op0, m_Undef()))
> +    return Op0;
> +
> +  // X % undef -> undef
> +  if (match(Op1, m_Undef()))
> +    return Op1;
> +
> +  return 0;
> +}
> +
> +Value *llvm::SimplifyFRemInst(Value *Op0, Value *Op1, const TargetData *TD,
> +                              const DominatorTree *DT) {
> +  return ::SimplifyFRemInst(Op0, Op1, TD, DT, RecursionLimit);
> +}
> +
> /// SimplifyShift - Given operands for an Shl, LShr or AShr, see if we can
> /// fold the result.  If not, this returns null.
> static Value *SimplifyShift(unsigned Opcode, Value *Op0, Value *Op1,
> @@ -1994,6 +2099,9 @@
>   case Instruction::SDiv: return SimplifySDivInst(LHS, RHS, TD, DT, MaxRecurse);
>   case Instruction::UDiv: return SimplifyUDivInst(LHS, RHS, TD, DT, MaxRecurse);
>   case Instruction::FDiv: return SimplifyFDivInst(LHS, RHS, TD, DT, MaxRecurse);
> +  case Instruction::SRem: return SimplifySRemInst(LHS, RHS, TD, DT, MaxRecurse);
> +  case Instruction::URem: return SimplifyURemInst(LHS, RHS, TD, DT, MaxRecurse);
> +  case Instruction::FRem: return SimplifyFRemInst(LHS, RHS, TD, DT, MaxRecurse);
>   case Instruction::Shl:
>     return SimplifyShlInst(LHS, RHS, /*isNSW*/false, /*isNUW*/false,
>                            TD, DT, MaxRecurse);
> @@ -2088,6 +2196,15 @@
>   case Instruction::FDiv:
>     Result = SimplifyFDivInst(I->getOperand(0), I->getOperand(1), TD, DT);
>     break;
> +  case Instruction::SRem:
> +    Result = SimplifySRemInst(I->getOperand(0), I->getOperand(1), TD, DT);
> +    break;
> +  case Instruction::URem:
> +    Result = SimplifyURemInst(I->getOperand(0), I->getOperand(1), TD, DT);
> +    break;
> +  case Instruction::FRem:
> +    Result = SimplifyFRemInst(I->getOperand(0), I->getOperand(1), TD, DT);
> +    break;
>   case Instruction::Shl:
>     Result = SimplifyShlInst(I->getOperand(0), I->getOperand(1),
>                              cast<BinaryOperator>(I)->hasNoSignedWrap(),
> 
> Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp?rev=130694&r1=130693&r2=130694&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp (original)
> +++ llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp Mon May  2 11:27:02 2011
> @@ -492,28 +492,6 @@
>   return 0;
> }
> 
> -/// This function implements the transforms on rem instructions that work
> -/// regardless of the kind of rem instruction it is (urem, srem, or frem). It 
> -/// is used by the visitors to those instructions.
> -/// @brief Transforms common to all three rem instructions
> -Instruction *InstCombiner::commonRemTransforms(BinaryOperator &I) {
> -  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
> -
> -  if (isa<UndefValue>(Op0)) {             // undef % X -> 0
> -    if (I.getType()->isFPOrFPVectorTy())
> -      return ReplaceInstUsesWith(I, Op0);  // X % undef -> undef (could be SNaN)
> -    return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
> -  }
> -  if (isa<UndefValue>(Op1))
> -    return ReplaceInstUsesWith(I, Op1);  // X % undef -> undef
> -
> -  // Handle cases involving: rem X, (select Cond, Y, Z)
> -  if (isa<SelectInst>(Op1) && SimplifyDivRemOfSelect(I))
> -    return &I;
> -
> -  return 0;
> -}
> -
> /// This function implements the transforms common to both integer remainder
> /// instructions (urem and srem). It is called by the visitors to those integer
> /// remainder instructions.
> @@ -521,26 +499,11 @@
> Instruction *InstCombiner::commonIRemTransforms(BinaryOperator &I) {
>   Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
> 
> -  if (Instruction *common = commonRemTransforms(I))
> -    return common;
> -
> -  // X % X == 0
> -  if (Op0 == Op1)
> -    return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
> -
> -  // 0 % X == 0 for integer, we don't need to preserve faults!
> -  if (Constant *LHS = dyn_cast<Constant>(Op0))
> -    if (LHS->isNullValue())
> -      return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
> +  // Handle cases involving: rem X, (select Cond, Y, Z)
> +  if (isa<SelectInst>(Op1) && SimplifyDivRemOfSelect(I))
> +    return &I;
> 
>   if (ConstantInt *RHS = dyn_cast<ConstantInt>(Op1)) {
> -    // X % 0 == undef, we don't need to preserve faults!
> -    if (RHS->equalsInt(0))
> -      return ReplaceInstUsesWith(I, UndefValue::get(I.getType()));
> -    
> -    if (RHS->equalsInt(1))  // X % 1 == 0
> -      return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
> -
>     if (Instruction *Op0I = dyn_cast<Instruction>(Op0)) {
>       if (SelectInst *SI = dyn_cast<SelectInst>(Op0I)) {
>         if (Instruction *R = FoldOpIntoSelect(I, SI))
> @@ -562,6 +525,9 @@
> Instruction *InstCombiner::visitURem(BinaryOperator &I) {
>   Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
> 
> +  if (Value *V = SimplifyURemInst(Op0, Op1, TD))
> +    return ReplaceInstUsesWith(I, V);
> +
>   if (Instruction *common = commonIRemTransforms(I))
>     return common;
> 
> @@ -602,6 +568,9 @@
> Instruction *InstCombiner::visitSRem(BinaryOperator &I) {
>   Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
> 
> +  if (Value *V = SimplifySRemInst(Op0, Op1, TD))
> +    return ReplaceInstUsesWith(I, V);
> +
>   // Handle the integer rem common cases
>   if (Instruction *Common = commonIRemTransforms(I))
>     return Common;
> @@ -660,6 +629,14 @@
> }
> 
> Instruction *InstCombiner::visitFRem(BinaryOperator &I) {
> -  return commonRemTransforms(I);
> -}
> +  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
> 
> +  if (Value *V = SimplifyFRemInst(Op0, Op1, TD))
> +    return ReplaceInstUsesWith(I, V);
> +
> +  // Handle cases involving: rem X, (select Cond, Y, Z)
> +  if (isa<SelectInst>(Op1) && SimplifyDivRemOfSelect(I))
> +    return &I;
> +
> +  return 0;
> +}
> 
> Added: llvm/trunk/test/Transforms/InstSimplify/rem.ll
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstSimplify/rem.ll?rev=130694&view=auto
> ==============================================================================
> --- llvm/trunk/test/Transforms/InstSimplify/rem.ll (added)
> +++ llvm/trunk/test/Transforms/InstSimplify/rem.ll Mon May  2 11:27:02 2011
> @@ -0,0 +1,17 @@
> +; RUN: opt < %s -instsimplify -S | FileCheck %s
> +
> +define i32 @select1(i32 %x, i1 %b) {
> +; CHECK: @select1
> +  %rhs = select i1 %b, i32 %x, i32 1
> +  %rem = srem i32 %x, %rhs
> +  ret i32 %rem
> +; CHECK: ret i32 0
> +}
> +
> +define i32 @select2(i32 %x, i1 %b) {
> +; CHECK: @select2
> +  %rhs = select i1 %b, i32 %x, i32 1
> +  %rem = urem i32 %x, %rhs
> +  ret i32 %rem
> +; CHECK: ret i32 0
> +}
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits


From dpatel at apple.com  Mon May  2 13:19:17 2011
From: dpatel at apple.com (Devang Patel)
Date: Mon, 02 May 2011 18:19:17 -0000
Subject: [llvm-commits] [llvm] r130702 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
Message-ID: <20110502181917.8DB8C2A6C12C@llvm.org>

Author: dpatel
Date: Mon May  2 13:19:17 2011
New Revision: 130702

URL: http://llvm.org/viewvc/llvm-project?rev=130702&view=rev
Log:
Emit debug info for global variables first.
This works around a limitation in gdb which is reported by following inherit.exp test failures from gdb testsuite.

gdb.cp/inherit.exp: print g_vB.vB::vb
gdb.cp/inherit.exp: print g_vB.vB::vx
gdb.cp/inherit.exp: print g_vC.vC::vc
gdb.cp/inherit.exp: print g_vC.vC::vx
gdb.cp/inherit.exp: print g_vD.vB::vb
...


Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130702&r1=130701&r2=130702&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Mon May  2 13:19:17 2011
@@ -1141,16 +1141,16 @@
          E = DbgFinder.compile_unit_end(); I != E; ++I)
     constructCompileUnit(*I);
 
-  // Create DIEs for each subprogram.
-  for (DebugInfoFinder::iterator I = DbgFinder.subprogram_begin(),
-         E = DbgFinder.subprogram_end(); I != E; ++I)
-    constructSubprogramDIE(*I);
-
   // Create DIEs for each global variable.
   for (DebugInfoFinder::iterator I = DbgFinder.global_variable_begin(),
          E = DbgFinder.global_variable_end(); I != E; ++I)
     constructGlobalVariableDIE(*I);
 
+  // Create DIEs for each subprogram.
+  for (DebugInfoFinder::iterator I = DbgFinder.subprogram_begin(),
+         E = DbgFinder.subprogram_end(); I != E; ++I)
+    constructSubprogramDIE(*I);
+
   //getOrCreateTypeDIE
   if (NamedMDNode *NMD = M->getNamedMetadata("llvm.dbg.enum"))
     for (unsigned i = 0, e = NMD->getNumOperands(); i != e; ++i) {



From baldrick at free.fr  Mon May  2 13:41:29 2011
From: baldrick at free.fr (Duncan Sands)
Date: Mon, 02 May 2011 18:41:29 -0000
Subject: [llvm-commits] [llvm] r130705 -
 /llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
Message-ID: <20110502184130.079FA2A6C12C@llvm.org>

Author: baldrick
Date: Mon May  2 13:41:29 2011
New Revision: 130705

URL: http://llvm.org/viewvc/llvm-project?rev=130705&view=rev
Log:
Remove unused variable.

Modified:
    llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp

Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp?rev=130705&r1=130704&r2=130705&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp (original)
+++ llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp Mon May  2 13:41:29 2011
@@ -503,7 +503,7 @@
   if (isa<SelectInst>(Op1) && SimplifyDivRemOfSelect(I))
     return &I;
 
-  if (ConstantInt *RHS = dyn_cast<ConstantInt>(Op1)) {
+  if (isa<ConstantInt>(Op1)) {
     if (Instruction *Op0I = dyn_cast<Instruction>(Op0)) {
       if (SelectInst *SI = dyn_cast<SelectInst>(Op0I)) {
         if (Instruction *R = FoldOpIntoSelect(I, SI))



From baldrick at free.fr  Mon May  2 13:51:41 2011
From: baldrick at free.fr (Duncan Sands)
Date: Mon, 02 May 2011 18:51:41 -0000
Subject: [llvm-commits] [llvm] r130706 - in /llvm/trunk:
 lib/Analysis/InstructionSimplify.cpp
 test/Transforms/InstCombine/2011-05-02-VectorBoolean.ll
Message-ID: <20110502185141.4E4172A6C12C@llvm.org>

Author: baldrick
Date: Mon May  2 13:51:41 2011
New Revision: 130706

URL: http://llvm.org/viewvc/llvm-project?rev=130706&view=rev
Log:
Fix PR9579: when simplifying a compare to "true" or "false", and it was
a vector compare, generate a vector result rather than i1 (and crashing).

Added:
    llvm/trunk/test/Transforms/InstCombine/2011-05-02-VectorBoolean.ll
Modified:
    llvm/trunk/lib/Analysis/InstructionSimplify.cpp

Modified: llvm/trunk/lib/Analysis/InstructionSimplify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/InstructionSimplify.cpp?rev=130706&r1=130705&r2=130706&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/InstructionSimplify.cpp (original)
+++ llvm/trunk/lib/Analysis/InstructionSimplify.cpp Mon May  2 13:51:41 2011
@@ -913,8 +913,6 @@
     }
   }
 
-  bool isSigned = Opcode == Instruction::SRem;
-
   // X % undef -> undef
   if (match(Op1, m_Undef()))
     return Op1;
@@ -1460,46 +1458,48 @@
     default:
       assert(false && "Unknown ICmp predicate!");
     case ICmpInst::ICMP_ULT:
-      return ConstantInt::getFalse(LHS->getContext());
+      // getNullValue also works for vectors, unlike getFalse.
+      return Constant::getNullValue(ITy);
     case ICmpInst::ICMP_UGE:
-      return ConstantInt::getTrue(LHS->getContext());
+      // getAllOnesValue also works for vectors, unlike getTrue.
+      return ConstantInt::getAllOnesValue(ITy);
     case ICmpInst::ICMP_EQ:
     case ICmpInst::ICMP_ULE:
       if (isKnownNonZero(LHS, TD))
-        return ConstantInt::getFalse(LHS->getContext());
+        return Constant::getNullValue(ITy);
       break;
     case ICmpInst::ICMP_NE:
     case ICmpInst::ICMP_UGT:
       if (isKnownNonZero(LHS, TD))
-        return ConstantInt::getTrue(LHS->getContext());
+        return ConstantInt::getAllOnesValue(ITy);
       break;
     case ICmpInst::ICMP_SLT:
       ComputeSignBit(LHS, LHSKnownNonNegative, LHSKnownNegative, TD);
       if (LHSKnownNegative)
-        return ConstantInt::getTrue(LHS->getContext());
+        return ConstantInt::getAllOnesValue(ITy);
       if (LHSKnownNonNegative)
-        return ConstantInt::getFalse(LHS->getContext());
+        return Constant::getNullValue(ITy);
       break;
     case ICmpInst::ICMP_SLE:
       ComputeSignBit(LHS, LHSKnownNonNegative, LHSKnownNegative, TD);
       if (LHSKnownNegative)
-        return ConstantInt::getTrue(LHS->getContext());
+        return ConstantInt::getAllOnesValue(ITy);
       if (LHSKnownNonNegative && isKnownNonZero(LHS, TD))
-        return ConstantInt::getFalse(LHS->getContext());
+        return Constant::getNullValue(ITy);
       break;
     case ICmpInst::ICMP_SGE:
       ComputeSignBit(LHS, LHSKnownNonNegative, LHSKnownNegative, TD);
       if (LHSKnownNegative)
-        return ConstantInt::getFalse(LHS->getContext());
+        return Constant::getNullValue(ITy);
       if (LHSKnownNonNegative)
-        return ConstantInt::getTrue(LHS->getContext());
+        return ConstantInt::getAllOnesValue(ITy);
       break;
     case ICmpInst::ICMP_SGT:
       ComputeSignBit(LHS, LHSKnownNonNegative, LHSKnownNegative, TD);
       if (LHSKnownNegative)
-        return ConstantInt::getFalse(LHS->getContext());
+        return Constant::getNullValue(ITy);
       if (LHSKnownNonNegative && isKnownNonZero(LHS, TD))
-        return ConstantInt::getTrue(LHS->getContext());
+        return ConstantInt::getAllOnesValue(ITy);
       break;
     }
   }
@@ -1791,7 +1791,8 @@
     case ICmpInst::ICMP_EQ:
     case ICmpInst::ICMP_UGT:
     case ICmpInst::ICMP_UGE:
-      return ConstantInt::getFalse(RHS->getContext());
+      // getNullValue also works for vectors, unlike getFalse.
+      return Constant::getNullValue(ITy);
     case ICmpInst::ICMP_SLT:
     case ICmpInst::ICMP_SLE:
       ComputeSignBit(LHS, KnownNonNegative, KnownNegative, TD);
@@ -1801,7 +1802,8 @@
     case ICmpInst::ICMP_NE:
     case ICmpInst::ICMP_ULT:
     case ICmpInst::ICMP_ULE:
-      return ConstantInt::getTrue(RHS->getContext());
+      // getAllOnesValue also works for vectors, unlike getTrue.
+      return Constant::getAllOnesValue(ITy);
     }
   }
   if (RBO && match(RBO, m_URem(m_Value(), m_Specific(LHS)))) {
@@ -1818,7 +1820,8 @@
     case ICmpInst::ICMP_NE:
     case ICmpInst::ICMP_UGT:
     case ICmpInst::ICMP_UGE:
-      return ConstantInt::getTrue(RHS->getContext());
+      // getAllOnesValue also works for vectors, unlike getTrue.
+      return Constant::getAllOnesValue(ITy);
     case ICmpInst::ICMP_SLT:
     case ICmpInst::ICMP_SLE:
       ComputeSignBit(RHS, KnownNonNegative, KnownNegative, TD);
@@ -1828,7 +1831,8 @@
     case ICmpInst::ICMP_EQ:
     case ICmpInst::ICMP_ULT:
     case ICmpInst::ICMP_ULE:
-      return ConstantInt::getFalse(RHS->getContext());
+      // getNullValue also works for vectors, unlike getFalse.
+      return Constant::getNullValue(ITy);
     }
   }
 

Added: llvm/trunk/test/Transforms/InstCombine/2011-05-02-VectorBoolean.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstCombine/2011-05-02-VectorBoolean.ll?rev=130706&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/InstCombine/2011-05-02-VectorBoolean.ll (added)
+++ llvm/trunk/test/Transforms/InstCombine/2011-05-02-VectorBoolean.ll Mon May  2 13:51:41 2011
@@ -0,0 +1,15 @@
+; RUN: opt < %s -instcombine
+; PR9579
+
+define <2 x i16> @entry(<2 x i16> %a) nounwind {
+entry:
+  %a.addr = alloca <2 x i16>, align 4
+  %.compoundliteral = alloca <2 x i16>, align 4
+  store <2 x i16> %a, <2 x i16>* %a.addr, align 4
+  %tmp = load <2 x i16>* %a.addr, align 4
+  store <2 x i16> zeroinitializer, <2 x i16>* %.compoundliteral
+  %tmp1 = load <2 x i16>* %.compoundliteral
+  %cmp = icmp uge <2 x i16> %tmp, %tmp1
+  %sext = sext <2 x i1> %cmp to <2 x i16>
+  ret <2 x i16> %sext
+}



From baldrick at free.fr  Mon May  2 13:58:26 2011
From: baldrick at free.fr (Duncan Sands)
Date: Mon, 02 May 2011 20:58:26 +0200
Subject: [llvm-commits] [llvm] r130694 - in /llvm/trunk:
 lib/Analysis/InstructionSimplify.cpp
 lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
 test/Transforms/InstSimplify/rem.ll
In-Reply-To: <EF1D4124-3623-4618-AF6A-CED99C2F633D@apple.com>
References: <20110502162702.CC0B72A6C12D@llvm.org>
	<EF1D4124-3623-4618-AF6A-CED99C2F633D@apple.com>
Message-ID: <4DBEFED2.1020101@free.fr>

Hi Eric,

> This put a few warnings into the build:
>
> /Volumes/Data/sources/llvm-clean/lib/Analysis/InstructionSimplify.cpp:916:8: warning:
>        unused variable 'isSigned' [-Wunused-variable]
>    bool isSigned = Opcode == Instruction::SRem;
>
>
> /Volumes/Data/sources/llvm-clean/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp:506:20: warning:
>        unused variable 'RHS' [-Wunused-variable]
>    if (ConstantInt *RHS = dyn_cast<ConstantInt>(Op1)) {

sorry about that - looks like I need to increase my -O level :)

> Mind cleaning these up?

Done!

Ciao, Duncan.

>
> Thanks!
>
> -eric
>
> On May 2, 2011, at 9:27 AM, Duncan Sands wrote:
>
>> Author: baldrick
>> Date: Mon May  2 11:27:02 2011
>> New Revision: 130694
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=130694&view=rev
>> Log:
>> Move some rem transforms out of instcombine and into instsimplify.
>> This automagically provides a transform noticed by my super-optimizer
>> as occurring quite often: "rem x, (select cond, x, 1)" ->  0.
>>
>> Added:
>>     llvm/trunk/test/Transforms/InstSimplify/rem.ll
>> Modified:
>>     llvm/trunk/lib/Analysis/InstructionSimplify.cpp
>>     llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
>>
>> Modified: llvm/trunk/lib/Analysis/InstructionSimplify.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/InstructionSimplify.cpp?rev=130694&r1=130693&r2=130694&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/Analysis/InstructionSimplify.cpp (original)
>> +++ llvm/trunk/lib/Analysis/InstructionSimplify.cpp Mon May  2 11:27:02 2011
>> @@ -901,6 +901,111 @@
>>    return ::SimplifyFDivInst(Op0, Op1, TD, DT, RecursionLimit);
>> }
>>
>> +/// SimplifyRem - Given operands for an SRem or URem, see if we can
>> +/// fold the result.  If not, this returns null.
>> +static Value *SimplifyRem(Instruction::BinaryOps Opcode, Value *Op0, Value *Op1,
>> +                          const TargetData *TD, const DominatorTree *DT,
>> +                          unsigned MaxRecurse) {
>> +  if (Constant *C0 = dyn_cast<Constant>(Op0)) {
>> +    if (Constant *C1 = dyn_cast<Constant>(Op1)) {
>> +      Constant *Ops[] = { C0, C1 };
>> +      return ConstantFoldInstOperands(Opcode, C0->getType(), Ops, 2, TD);
>> +    }
>> +  }
>> +
>> +  bool isSigned = Opcode == Instruction::SRem;
>> +
>> +  // X % undef ->  undef
>> +  if (match(Op1, m_Undef()))
>> +    return Op1;
>> +
>> +  // undef % X ->  0
>> +  if (match(Op0, m_Undef()))
>> +    return Constant::getNullValue(Op0->getType());
>> +
>> +  // 0 % X ->  0, we don't need to preserve faults!
>> +  if (match(Op0, m_Zero()))
>> +    return Op0;
>> +
>> +  // X % 0 ->  undef, we don't need to preserve faults!
>> +  if (match(Op1, m_Zero()))
>> +    return UndefValue::get(Op0->getType());
>> +
>> +  // X % 1 ->  0
>> +  if (match(Op1, m_One()))
>> +    return Constant::getNullValue(Op0->getType());
>> +
>> +  if (Op0->getType()->isIntegerTy(1))
>> +    // It can't be remainder by zero, hence it must be remainder by one.
>> +    return Constant::getNullValue(Op0->getType());
>> +
>> +  // X % X ->  0
>> +  if (Op0 == Op1)
>> +    return Constant::getNullValue(Op0->getType());
>> +
>> +  // If the operation is with the result of a select instruction, check whether
>> +  // operating on either branch of the select always yields the same value.
>> +  if (isa<SelectInst>(Op0) || isa<SelectInst>(Op1))
>> +    if (Value *V = ThreadBinOpOverSelect(Opcode, Op0, Op1, TD, DT, MaxRecurse))
>> +      return V;
>> +
>> +  // If the operation is with the result of a phi instruction, check whether
>> +  // operating on all incoming values of the phi always yields the same value.
>> +  if (isa<PHINode>(Op0) || isa<PHINode>(Op1))
>> +    if (Value *V = ThreadBinOpOverPHI(Opcode, Op0, Op1, TD, DT, MaxRecurse))
>> +      return V;
>> +
>> +  return 0;
>> +}
>> +
>> +/// SimplifySRemInst - Given operands for an SRem, see if we can
>> +/// fold the result.  If not, this returns null.
>> +static Value *SimplifySRemInst(Value *Op0, Value *Op1, const TargetData *TD,
>> +                               const DominatorTree *DT, unsigned MaxRecurse) {
>> +  if (Value *V = SimplifyRem(Instruction::SRem, Op0, Op1, TD, DT, MaxRecurse))
>> +    return V;
>> +
>> +  return 0;
>> +}
>> +
>> +Value *llvm::SimplifySRemInst(Value *Op0, Value *Op1, const TargetData *TD,
>> +                              const DominatorTree *DT) {
>> +  return ::SimplifySRemInst(Op0, Op1, TD, DT, RecursionLimit);
>> +}
>> +
>> +/// SimplifyURemInst - Given operands for a URem, see if we can
>> +/// fold the result.  If not, this returns null.
>> +static Value *SimplifyURemInst(Value *Op0, Value *Op1, const TargetData *TD,
>> +                               const DominatorTree *DT, unsigned MaxRecurse) {
>> +  if (Value *V = SimplifyRem(Instruction::URem, Op0, Op1, TD, DT, MaxRecurse))
>> +    return V;
>> +
>> +  return 0;
>> +}
>> +
>> +Value *llvm::SimplifyURemInst(Value *Op0, Value *Op1, const TargetData *TD,
>> +                              const DominatorTree *DT) {
>> +  return ::SimplifyURemInst(Op0, Op1, TD, DT, RecursionLimit);
>> +}
>> +
>> +static Value *SimplifyFRemInst(Value *Op0, Value *Op1, const TargetData *,
>> +                               const DominatorTree *, unsigned) {
>> +  // undef % X ->  undef    (the undef could be a snan).
>> +  if (match(Op0, m_Undef()))
>> +    return Op0;
>> +
>> +  // X % undef ->  undef
>> +  if (match(Op1, m_Undef()))
>> +    return Op1;
>> +
>> +  return 0;
>> +}
>> +
>> +Value *llvm::SimplifyFRemInst(Value *Op0, Value *Op1, const TargetData *TD,
>> +                              const DominatorTree *DT) {
>> +  return ::SimplifyFRemInst(Op0, Op1, TD, DT, RecursionLimit);
>> +}
>> +
>> /// SimplifyShift - Given operands for an Shl, LShr or AShr, see if we can
>> /// fold the result.  If not, this returns null.
>> static Value *SimplifyShift(unsigned Opcode, Value *Op0, Value *Op1,
>> @@ -1994,6 +2099,9 @@
>>    case Instruction::SDiv: return SimplifySDivInst(LHS, RHS, TD, DT, MaxRecurse);
>>    case Instruction::UDiv: return SimplifyUDivInst(LHS, RHS, TD, DT, MaxRecurse);
>>    case Instruction::FDiv: return SimplifyFDivInst(LHS, RHS, TD, DT, MaxRecurse);
>> +  case Instruction::SRem: return SimplifySRemInst(LHS, RHS, TD, DT, MaxRecurse);
>> +  case Instruction::URem: return SimplifyURemInst(LHS, RHS, TD, DT, MaxRecurse);
>> +  case Instruction::FRem: return SimplifyFRemInst(LHS, RHS, TD, DT, MaxRecurse);
>>    case Instruction::Shl:
>>      return SimplifyShlInst(LHS, RHS, /*isNSW*/false, /*isNUW*/false,
>>                             TD, DT, MaxRecurse);
>> @@ -2088,6 +2196,15 @@
>>    case Instruction::FDiv:
>>      Result = SimplifyFDivInst(I->getOperand(0), I->getOperand(1), TD, DT);
>>      break;
>> +  case Instruction::SRem:
>> +    Result = SimplifySRemInst(I->getOperand(0), I->getOperand(1), TD, DT);
>> +    break;
>> +  case Instruction::URem:
>> +    Result = SimplifyURemInst(I->getOperand(0), I->getOperand(1), TD, DT);
>> +    break;
>> +  case Instruction::FRem:
>> +    Result = SimplifyFRemInst(I->getOperand(0), I->getOperand(1), TD, DT);
>> +    break;
>>    case Instruction::Shl:
>>      Result = SimplifyShlInst(I->getOperand(0), I->getOperand(1),
>>                               cast<BinaryOperator>(I)->hasNoSignedWrap(),
>>
>> Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp?rev=130694&r1=130693&r2=130694&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp (original)
>> +++ llvm/trunk/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp Mon May  2 11:27:02 2011
>> @@ -492,28 +492,6 @@
>>    return 0;
>> }
>>
>> -/// This function implements the transforms on rem instructions that work
>> -/// regardless of the kind of rem instruction it is (urem, srem, or frem). It
>> -/// is used by the visitors to those instructions.
>> -/// @brief Transforms common to all three rem instructions
>> -Instruction *InstCombiner::commonRemTransforms(BinaryOperator&I) {
>> -  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
>> -
>> -  if (isa<UndefValue>(Op0)) {             // undef % X ->  0
>> -    if (I.getType()->isFPOrFPVectorTy())
>> -      return ReplaceInstUsesWith(I, Op0);  // X % undef ->  undef (could be SNaN)
>> -    return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
>> -  }
>> -  if (isa<UndefValue>(Op1))
>> -    return ReplaceInstUsesWith(I, Op1);  // X % undef ->  undef
>> -
>> -  // Handle cases involving: rem X, (select Cond, Y, Z)
>> -  if (isa<SelectInst>(Op1)&&  SimplifyDivRemOfSelect(I))
>> -    return&I;
>> -
>> -  return 0;
>> -}
>> -
>> /// This function implements the transforms common to both integer remainder
>> /// instructions (urem and srem). It is called by the visitors to those integer
>> /// remainder instructions.
>> @@ -521,26 +499,11 @@
>> Instruction *InstCombiner::commonIRemTransforms(BinaryOperator&I) {
>>    Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
>>
>> -  if (Instruction *common = commonRemTransforms(I))
>> -    return common;
>> -
>> -  // X % X == 0
>> -  if (Op0 == Op1)
>> -    return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
>> -
>> -  // 0 % X == 0 for integer, we don't need to preserve faults!
>> -  if (Constant *LHS = dyn_cast<Constant>(Op0))
>> -    if (LHS->isNullValue())
>> -      return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
>> +  // Handle cases involving: rem X, (select Cond, Y, Z)
>> +  if (isa<SelectInst>(Op1)&&  SimplifyDivRemOfSelect(I))
>> +    return&I;
>>
>>    if (ConstantInt *RHS = dyn_cast<ConstantInt>(Op1)) {
>> -    // X % 0 == undef, we don't need to preserve faults!
>> -    if (RHS->equalsInt(0))
>> -      return ReplaceInstUsesWith(I, UndefValue::get(I.getType()));
>> -
>> -    if (RHS->equalsInt(1))  // X % 1 == 0
>> -      return ReplaceInstUsesWith(I, Constant::getNullValue(I.getType()));
>> -
>>      if (Instruction *Op0I = dyn_cast<Instruction>(Op0)) {
>>        if (SelectInst *SI = dyn_cast<SelectInst>(Op0I)) {
>>          if (Instruction *R = FoldOpIntoSelect(I, SI))
>> @@ -562,6 +525,9 @@
>> Instruction *InstCombiner::visitURem(BinaryOperator&I) {
>>    Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
>>
>> +  if (Value *V = SimplifyURemInst(Op0, Op1, TD))
>> +    return ReplaceInstUsesWith(I, V);
>> +
>>    if (Instruction *common = commonIRemTransforms(I))
>>      return common;
>>
>> @@ -602,6 +568,9 @@
>> Instruction *InstCombiner::visitSRem(BinaryOperator&I) {
>>    Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
>>
>> +  if (Value *V = SimplifySRemInst(Op0, Op1, TD))
>> +    return ReplaceInstUsesWith(I, V);
>> +
>>    // Handle the integer rem common cases
>>    if (Instruction *Common = commonIRemTransforms(I))
>>      return Common;
>> @@ -660,6 +629,14 @@
>> }
>>
>> Instruction *InstCombiner::visitFRem(BinaryOperator&I) {
>> -  return commonRemTransforms(I);
>> -}
>> +  Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
>>
>> +  if (Value *V = SimplifyFRemInst(Op0, Op1, TD))
>> +    return ReplaceInstUsesWith(I, V);
>> +
>> +  // Handle cases involving: rem X, (select Cond, Y, Z)
>> +  if (isa<SelectInst>(Op1)&&  SimplifyDivRemOfSelect(I))
>> +    return&I;
>> +
>> +  return 0;
>> +}
>>
>> Added: llvm/trunk/test/Transforms/InstSimplify/rem.ll
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstSimplify/rem.ll?rev=130694&view=auto
>> ==============================================================================
>> --- llvm/trunk/test/Transforms/InstSimplify/rem.ll (added)
>> +++ llvm/trunk/test/Transforms/InstSimplify/rem.ll Mon May  2 11:27:02 2011
>> @@ -0,0 +1,17 @@
>> +; RUN: opt<  %s -instsimplify -S | FileCheck %s
>> +
>> +define i32 @select1(i32 %x, i1 %b) {
>> +; CHECK: @select1
>> +  %rhs = select i1 %b, i32 %x, i32 1
>> +  %rem = srem i32 %x, %rhs
>> +  ret i32 %rem
>> +; CHECK: ret i32 0
>> +}
>> +
>> +define i32 @select2(i32 %x, i1 %b) {
>> +; CHECK: @select2
>> +  %rhs = select i1 %b, i32 %x, i32 1
>> +  %rem = urem i32 %x, %rhs
>> +  ret i32 %rem
>> +; CHECK: ret i32 0
>> +}
>>
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>


From echristo at apple.com  Mon May  2 13:59:53 2011
From: echristo at apple.com (Eric Christopher)
Date: Mon, 02 May 2011 11:59:53 -0700
Subject: [llvm-commits] [llvm] r130694 - in /llvm/trunk:
 lib/Analysis/InstructionSimplify.cpp
 lib/Transforms/InstCombine/InstCombineMulDivRem.cpp
 test/Transforms/InstSimplify/rem.ll
In-Reply-To: <4DBEFED2.1020101@free.fr>
References: <20110502162702.CC0B72A6C12D@llvm.org>
	<EF1D4124-3623-4618-AF6A-CED99C2F633D@apple.com>
	<4DBEFED2.1020101@free.fr>
Message-ID: <BD689E0F-5D0E-4139-8BBE-CFF7B930EDB8@apple.com>


On May 2, 2011, at 11:58 AM, Duncan Sands wrote:

> Hi Eric,
> 
>> This put a few warnings into the build:
>> 
>> /Volumes/Data/sources/llvm-clean/lib/Analysis/InstructionSimplify.cpp:916:8: warning:
>>       unused variable 'isSigned' [-Wunused-variable]
>>   bool isSigned = Opcode == Instruction::SRem;
>> 
>> 
>> /Volumes/Data/sources/llvm-clean/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp:506:20: warning:
>>       unused variable 'RHS' [-Wunused-variable]
>>   if (ConstantInt *RHS = dyn_cast<ConstantInt>(Op1)) {
> 
> sorry about that - looks like I need to increase my -O level :)

Heh.

Could use clang! :)

Thanks!

-eric

From rafael.espindola at gmail.com  Mon May  2 14:45:24 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Mon, 02 May 2011 19:45:24 -0000
Subject: [llvm-commits] [llvm] r130712 -
	/llvm/trunk/tools/llvmc/src/Base.td.in
Message-ID: <20110502194524.66F442A6C12C@llvm.org>

Author: rafael
Date: Mon May  2 14:45:24 2011
New Revision: 130712

URL: http://llvm.org/viewvc/llvm-project?rev=130712&view=rev
Log:
llvmc: Pass -disable-cfi to llc.

Modified:
    llvm/trunk/tools/llvmc/src/Base.td.in

Modified: llvm/trunk/tools/llvmc/src/Base.td.in
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/tools/llvmc/src/Base.td.in?rev=130712&r1=130711&r2=130712&view=diff
==============================================================================
--- llvm/trunk/tools/llvmc/src/Base.td.in (original)
+++ llvm/trunk/tools/llvmc/src/Base.td.in Mon May  2 14:45:24 2011
@@ -301,7 +301,7 @@
 [(in_language "llvm-bitcode", "llvm-assembler"),
  (out_language "assembler"),
  (output_suffix "s"),
- (command "llc"),
+ (command "llc -disable-cfi"),
  (actions (case
           (switch_on "S"), (stop_compilation),
           (switch_on "O0"), (forward "O0"),



From rafael.espindola at gmail.com  Mon May  2 14:51:34 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Mon, 02 May 2011 19:51:34 -0000
Subject: [llvm-commits] [llvm] r130713 -
	/llvm/trunk/test/FrontendC/2008-07-29-EHLabel.ll
Message-ID: <20110502195134.9048B2A6C12C@llvm.org>

Author: rafael
Date: Mon May  2 14:51:34 2011
New Revision: 130713

URL: http://llvm.org/viewvc/llvm-project?rev=130713&view=rev
Log:
Pass -disable-cfi to llc.

Modified:
    llvm/trunk/test/FrontendC/2008-07-29-EHLabel.ll

Modified: llvm/trunk/test/FrontendC/2008-07-29-EHLabel.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC/2008-07-29-EHLabel.ll?rev=130713&r1=130712&r2=130713&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC/2008-07-29-EHLabel.ll (original)
+++ llvm/trunk/test/FrontendC/2008-07-29-EHLabel.ll Mon May  2 14:51:34 2011
@@ -1,4 +1,4 @@
-; RUN: llc %s -o - | %llvmgcc -xassembler -c -o /dev/null -
+; RUN: llc -disable-cfi %s -o - | %llvmgcc -xassembler -c -o /dev/null -
 ; PR2609
 	%struct..0._11 = type { i32 }
 	%struct..1__pthread_mutex_s = type { i32, i32, i32, i32, i32, %struct..0._11 }



From stoklund at 2pi.dk  Mon May  2 15:06:28 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Mon, 02 May 2011 20:06:28 -0000
Subject: [llvm-commits] [llvm] r130714 -
	/llvm/trunk/lib/CodeGen/RegisterScavenging.cpp
Message-ID: <20110502200628.797BB2A6C12C@llvm.org>

Author: stoklund
Date: Mon May  2 15:06:28 2011
New Revision: 130714

URL: http://llvm.org/viewvc/llvm-project?rev=130714&view=rev
Log:
Only ignore <undef> use operands, keep the <def,undef> ops.

Def operands may also have an <undef> flag, but that just means that a
sub-register redef doesn't actually read the super-register. For physical
registers, it has no meaning.

Modified:
    llvm/trunk/lib/CodeGen/RegisterScavenging.cpp

Modified: llvm/trunk/lib/CodeGen/RegisterScavenging.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/RegisterScavenging.cpp?rev=130714&r1=130713&r2=130714&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/RegisterScavenging.cpp (original)
+++ llvm/trunk/lib/CodeGen/RegisterScavenging.cpp Mon May  2 15:06:28 2011
@@ -154,13 +154,16 @@
   BitVector DeadRegs(NumPhysRegs);
   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI->getOperand(i);
-    if (!MO.isReg() || MO.isUndef())
+    if (!MO.isReg())
       continue;
     unsigned Reg = MO.getReg();
     if (!Reg || isReserved(Reg))
       continue;
 
     if (MO.isUse()) {
+      // Ignore undef uses.
+      if (MO.isUndef())
+        continue;
       // Two-address operands implicitly kill.
       if (!isPred && (MO.isKill() || MI->isRegTiedToDefOperand(i)))
         addRegWithSubRegs(KillRegs, Reg);



From stoklund at 2pi.dk  Mon May  2 15:06:30 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Mon, 02 May 2011 20:06:30 -0000
Subject: [llvm-commits] [llvm] r130715 -
	/llvm/trunk/lib/CodeGen/MachineFunction.cpp
Message-ID: <20110502200630.D98062A6C12D@llvm.org>

Author: stoklund
Date: Mon May  2 15:06:30 2011
New Revision: 130715

URL: http://llvm.org/viewvc/llvm-project?rev=130715&view=rev
Log:
Use the PrintReg adaptor to correctly print live-in registers in debug output.

Modified:
    llvm/trunk/lib/CodeGen/MachineFunction.cpp

Modified: llvm/trunk/lib/CodeGen/MachineFunction.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineFunction.cpp?rev=130715&r1=130714&r2=130715&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/MachineFunction.cpp (original)
+++ llvm/trunk/lib/CodeGen/MachineFunction.cpp Mon May  2 15:06:30 2011
@@ -300,31 +300,19 @@
     OS << "Function Live Ins: ";
     for (MachineRegisterInfo::livein_iterator
          I = RegInfo->livein_begin(), E = RegInfo->livein_end(); I != E; ++I) {
-      if (TRI)
-        OS << "%" << TRI->getName(I->first);
-      else
-        OS << " %physreg" << I->first;
-      
+      OS << PrintReg(I->first, TRI);
       if (I->second)
-        OS << " in reg%" << I->second;
-
+        OS << " in " << PrintReg(I->second, TRI);
       if (llvm::next(I) != E)
         OS << ", ";
     }
     OS << '\n';
   }
   if (RegInfo && !RegInfo->liveout_empty()) {
-    OS << "Function Live Outs: ";
+    OS << "Function Live Outs:";
     for (MachineRegisterInfo::liveout_iterator
-         I = RegInfo->liveout_begin(), E = RegInfo->liveout_end(); I != E; ++I){
-      if (TRI)
-        OS << '%' << TRI->getName(*I);
-      else
-        OS << "%physreg" << *I;
-
-      if (llvm::next(I) != E)
-        OS << " ";
-    }
+         I = RegInfo->liveout_begin(), E = RegInfo->liveout_end(); I != E; ++I)
+      OS << ' ' << PrintReg(*I, TRI);
     OS << '\n';
   }
   



From echristo at apple.com  Mon May  2 15:16:33 2011
From: echristo at apple.com (Eric Christopher)
Date: Mon, 02 May 2011 20:16:33 -0000
Subject: [llvm-commits] [llvm] r130716 -
	/llvm/trunk/lib/Target/ARM/ARMFastISel.cpp
Message-ID: <20110502201633.6417A2A6C12C@llvm.org>

Author: echristo
Date: Mon May  2 15:16:33 2011
New Revision: 130716

URL: http://llvm.org/viewvc/llvm-project?rev=130716&view=rev
Log:
Apparently the check for direct calls is unnecessary.

Modified:
    llvm/trunk/lib/Target/ARM/ARMFastISel.cpp

Modified: llvm/trunk/lib/Target/ARM/ARMFastISel.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMFastISel.cpp?rev=130716&r1=130715&r2=130716&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMFastISel.cpp (original)
+++ llvm/trunk/lib/Target/ARM/ARMFastISel.cpp Mon May  2 15:16:33 2011
@@ -1839,9 +1839,9 @@
   // Can't handle inline asm or worry about intrinsics yet.
   if (isa<InlineAsm>(Callee) || isa<IntrinsicInst>(CI)) return false;
 
-  // Only handle global variable Callees that are direct calls.
+  // Only handle global variable Callees.
   const GlobalValue *GV = dyn_cast<GlobalValue>(Callee);
-  if (!GV || Subtarget->GVIsIndirectSymbol(GV, TM.getRelocationModel()))
+  if (!GV)
     return false;
 
   // Check the calling convention.



From stoklund at 2pi.dk  Mon May  2 15:36:53 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Mon, 02 May 2011 20:36:53 -0000
Subject: [llvm-commits] [llvm] r130718 -
	/llvm/trunk/lib/CodeGen/RegisterScavenging.cpp
Message-ID: <20110502203653.E90E72A6C12C@llvm.org>

Author: stoklund
Date: Mon May  2 15:36:53 2011
New Revision: 130718

URL: http://llvm.org/viewvc/llvm-project?rev=130718&view=rev
Log:
Handle <def,undef> in the second loop as well.

Modified:
    llvm/trunk/lib/CodeGen/RegisterScavenging.cpp

Modified: llvm/trunk/lib/CodeGen/RegisterScavenging.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/RegisterScavenging.cpp?rev=130718&r1=130717&r2=130718&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/RegisterScavenging.cpp (original)
+++ llvm/trunk/lib/CodeGen/RegisterScavenging.cpp Mon May  2 15:36:53 2011
@@ -181,12 +181,14 @@
   // Verify uses and defs.
   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI->getOperand(i);
-    if (!MO.isReg() || MO.isUndef())
+    if (!MO.isReg())
       continue;
     unsigned Reg = MO.getReg();
     if (!Reg || isReserved(Reg))
       continue;
     if (MO.isUse()) {
+      if (MO.isUndef())
+        continue;
       if (!isUsed(Reg)) {
         // Check if it's partial live: e.g.
         // D0 = insert_subreg D0<undef>, S0



From evan.cheng at apple.com  Mon May  2 16:00:18 2011
From: evan.cheng at apple.com (Evan Cheng)
Date: Mon, 02 May 2011 14:00:18 -0700
Subject: [llvm-commits] [llvm] r130485 - in /llvm/trunk:
 lib/Transforms/InstCombine/InstCombineShifts.cpp
 test/Transforms/InstCombine/shift.ll
In-Reply-To: <C9E399F9-DA42-4972-9BD4-130A27B3D1D5@apple.com>
References: <20110429081541.B7F422A6C138@llvm.org>
	<C8DE37BF-B56C-4F53-A49F-EA158B233245@apple.com>
	<CB365302-F1CC-4578-BA54-7D954227E750@googlemail.com>
	<1B93035B-15A4-47B0-8E78-2D0F8A6B1701@apple.com>
	<5F0816C4-4FC4-4F02-8B7E-51858107AC6D@googlemail.com>
	<C9E399F9-DA42-4972-9BD4-130A27B3D1D5@apple.com>
Message-ID: <E3078F91-A806-4AEE-8F70-448C10F1B4C6@apple.com>


On May 1, 2011, at 12:40 PM, Chris Lattner wrote:

> 
> On Apr 30, 2011, at 3:22 AM, Benjamin Kramer wrote:
> 
>> 
>> On 30.04.2011, at 00:00, Devang Patel wrote:
>> 
>>> 
>>> On Apr 29, 2011, at 2:50 PM, Benjamin Kramer wrote:
>>>> 
>>>> Are you sure this specific transformation destroys the DebugLoc? When InstCombine simply replaces one
>>>> instruction with another it preserves debug information (InstructionCombining.cpp:1592).
>>> 
>>> I added line 1592 however I am not sure that this particular transformation is covered by 1592 in all cases. I have seen "shl" without DebugLoc generated by instcombine before your patch landed in svn. If you're interested I can send you more info.
>> 
>> There is a way this transform can lose a DebugLoc
>> 
>> x = C1 <<  A # with DebugLoc
>> y = x  << C2 # without DebugLoc
>> 
>> InstCombine will then replace 'y' with a new instruction (but there's no DebugLoc to attach to it) and 'x' is DCE'd.
>> I think the only feasible way to improve here is to find out why 'y' has no debug info attached to it.
>> 
>> If you have a test case where InstCombine introduces SHLs without DebugLocs, I'd be interested in seeing it.
> 
> More generally, we don't want to sprinkle setDebugLoc calls into *every* instcombine xform.  We need a better solution.

I believe there is a long term plan being discussed. Devang, can you share what's being discussed so far?

Evan

> 
> -Chris
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits


From rengolin at systemcall.org  Mon May  2 16:03:35 2011
From: rengolin at systemcall.org (Renato Golin)
Date: Mon, 2 May 2011 22:03:35 +0100
Subject: [llvm-commits] [cfe-commits] [PATCH] arm coprocessor builtins
	and intrinsics!
In-Reply-To: <BANLkTikkHzT9r=hQeqAgnND+Sea_TbCT_g@mail.gmail.com>
References: <BANLkTikkHzT9r=hQeqAgnND+Sea_TbCT_g@mail.gmail.com>
Message-ID: <BANLkTiniFrC6OjZywcoH84ktYbRevh9w8w@mail.gmail.com>

On 30 April 2011 01:26, Bruno Cardoso Lopes <bruno.cardoso at gmail.com> wrote:
> I know in some cases we prefer avoiding the usage of intrinsics as
> much as possible and I'm not aware what it's the policy regarding the
> addition of new ARM builtins and intrinsics. ?However, this patch
> implements builtins and intrinsics for some coprocessor instructions,
> they have a clean interface and directly match the real instructions.

Hi Bruno,

I'm not sure about the intrinsic policy in this case, too, but the
change is most welcomed.

I think the patch looks nice, just would be better to have some tests
on the arguments, too. I know most LLVM tests don't do it, but in this
case, we need to make sure the options are going to the right place.

cheers,
--renato


From evan.cheng at apple.com  Mon May  2 16:12:09 2011
From: evan.cheng at apple.com (Evan Cheng)
Date: Mon, 02 May 2011 14:12:09 -0700
Subject: [llvm-commits] [PATCH] arm coprocessor builtins and intrinsics!
In-Reply-To: <BANLkTikkHzT9r=hQeqAgnND+Sea_TbCT_g@mail.gmail.com>
References: <BANLkTikkHzT9r=hQeqAgnND+Sea_TbCT_g@mail.gmail.com>
Message-ID: <60016C47-371D-48B6-A7DE-413D41592962@apple.com>

Hi Bruno,

These look fine to me. However, is it necessary to use def : Pat patterns? Can't you fold the patterns into the instruction defs?

Evan

On Apr 29, 2011, at 5:26 PM, Bruno Cardoso Lopes wrote:

> Hi all,
> 
> I know in some cases we prefer avoiding the usage of intrinsics as
> much as possible and I'm not aware what it's the policy regarding the
> addition of new ARM builtins and intrinsics.  However, this patch
> implements builtins and intrinsics for some coprocessor instructions,
> they have a clean interface and directly match the real instructions.
> 
> Is it OK to commit?
> 
> -- 
> Bruno Cardoso Lopes
> http://www.brunocardoso.cc
> <clang-arm-intrinsics.patch><llvm-arm-intrinsics.patch>_______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits


From dpatel at apple.com  Mon May  2 16:15:17 2011
From: dpatel at apple.com (Devang Patel)
Date: Mon, 02 May 2011 14:15:17 -0700
Subject: [llvm-commits] [llvm] r130485 - in /llvm/trunk:
 lib/Transforms/InstCombine/InstCombineShifts.cpp
 test/Transforms/InstCombine/shift.ll
In-Reply-To: <5F0816C4-4FC4-4F02-8B7E-51858107AC6D@googlemail.com>
References: <20110429081541.B7F422A6C138@llvm.org>
	<C8DE37BF-B56C-4F53-A49F-EA158B233245@apple.com>
	<CB365302-F1CC-4578-BA54-7D954227E750@googlemail.com>
	<1B93035B-15A4-47B0-8E78-2D0F8A6B1701@apple.com>
	<5F0816C4-4FC4-4F02-8B7E-51858107AC6D@googlemail.com>
Message-ID: <57230E3C-4F66-40A9-A57D-C919726D1E5D@apple.com>


On Apr 30, 2011, at 3:22 AM, Benjamin Kramer wrote:

> 
> On 30.04.2011, at 00:00, Devang Patel wrote:
> 
>> 
>> On Apr 29, 2011, at 2:50 PM, Benjamin Kramer wrote:
>>> 
>>> Are you sure this specific transformation destroys the DebugLoc? When InstCombine simply replaces one
>>> instruction with another it preserves debug information (InstructionCombining.cpp:1592).
>> 
>> I added line 1592 however I am not sure that this particular transformation is covered by 1592 in all cases. I have seen "shl" without DebugLoc generated by instcombine before your patch landed in svn. If you're interested I can send you more info.
> 
> There is a way this transform can lose a DebugLoc
> 
> x = C1 <<  A # with DebugLoc
> y = x  << C2 # without DebugLoc
> 
> InstCombine will then replace 'y' with a new instruction (but there's no DebugLoc to attach to it) and 'x' is DCE'd.
> I think the only feasible way to improve here is to find out why 'y' has no debug info attached to it.
> 
> If you have a test case where InstCombine introduces SHLs without DebugLocs, I'd be interested in seeing it.

Sure, I'll send you the test case offline.
-
Devang

From mttjwl at gmail.com  Mon May  2 16:12:10 2011
From: mttjwl at gmail.com (Matthew Wala)
Date: Mon, 02 May 2011 21:12:10 -0000
Subject: [llvm-commits] [poolalloc] r130724 - in /poolalloc/trunk:
 include/dsa/CStdLib.h lib/DSA/StdLibPass.cpp
Message-ID: <20110502211210.C54B82A6C12C@llvm.org>

Author: wala1
Date: Mon May  2 16:12:10 2011
New Revision: 130724

URL: http://llvm.org/viewvc/llvm-project?rev=130724&view=rev
Log:
Fixed entries for pool_bcmp() and pool_bcopy().

Modified:
    poolalloc/trunk/include/dsa/CStdLib.h
    poolalloc/trunk/lib/DSA/StdLibPass.cpp

Modified: poolalloc/trunk/include/dsa/CStdLib.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/dsa/CStdLib.h?rev=130724&r1=130723&r2=130724&view=diff
==============================================================================
--- poolalloc/trunk/include/dsa/CStdLib.h (original)
+++ poolalloc/trunk/include/dsa/CStdLib.h Mon May  2 16:12:10 2011
@@ -40,7 +40,7 @@
     { "pool_strcspn",     2 },
     { "pool_memccpy",     2 },
     { "pool_memchr",      1 },
-    { "pool_bcmp",        1 },
+    { "pool_bcmp",        2 },
     { "pool_bcopy",       2 },
     { "pool_index",       1 },
     { "pool_rindex",      1 },

Modified: poolalloc/trunk/lib/DSA/StdLibPass.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/DSA/StdLibPass.cpp?rev=130724&r1=130723&r2=130724&view=diff
==============================================================================
--- poolalloc/trunk/lib/DSA/StdLibPass.cpp (original)
+++ poolalloc/trunk/lib/DSA/StdLibPass.cpp Mon May  2 16:12:10 2011
@@ -287,8 +287,8 @@
   {"pool_memcmp",     {NRET_NNYARGS, NRET_NARGS,    NRET_NARGS, NRET_NARGS,   false}},
   {"pool_strcasecmp", {NRET_NNYARGS, NRET_NARGS,    NRET_NARGS, NRET_NARGS,   false}},
   {"pool_strncasecmp",{NRET_NNYARGS, NRET_NARGS,    NRET_NARGS, NRET_NARGS,   false}},
-  {"pool_bcopy",      {NRET_NYARGS,  NRET_NNNYARGS, NRET_NARGS, NRET_NNYARGS,  true}},
-  {"pool_bcmp",       {NRET_YARGS,   NRET_NARGS,    NRET_NARGS, NRET_NARGS,    true}},
+  {"pool_bcopy",      {NRET_NNYARGS, NRET_NNNYARGS, NRET_NARGS, NRET_NNYARGS,  true}},
+  {"pool_bcmp",       {NRET_NNYARGS, NRET_NARGS,    NRET_NARGS, NRET_NARGS,    true}},
   {"pool_index",      {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, YRET_NYARGS,   true}},
   {"pool_rindex",     {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, YRET_NYARGS,   true}},
 



From dpatel at apple.com  Mon May  2 16:17:20 2011
From: dpatel at apple.com (Devang Patel)
Date: Mon, 02 May 2011 14:17:20 -0700
Subject: [llvm-commits] [llvm] r130485 - in /llvm/trunk:
 lib/Transforms/InstCombine/InstCombineShifts.cpp
 test/Transforms/InstCombine/shift.ll
In-Reply-To: <C9E399F9-DA42-4972-9BD4-130A27B3D1D5@apple.com>
References: <20110429081541.B7F422A6C138@llvm.org>
	<C8DE37BF-B56C-4F53-A49F-EA158B233245@apple.com>
	<CB365302-F1CC-4578-BA54-7D954227E750@googlemail.com>
	<1B93035B-15A4-47B0-8E78-2D0F8A6B1701@apple.com>
	<5F0816C4-4FC4-4F02-8B7E-51858107AC6D@googlemail.com>
	<C9E399F9-DA42-4972-9BD4-130A27B3D1D5@apple.com>
Message-ID: <268B56A5-C381-4C9D-8D87-21784E40E070@apple.com>


On May 1, 2011, at 12:40 PM, Chris Lattner wrote:

> On Apr 30, 2011, at 3:22 AM, Benjamin Kramer wrote:
> 
>> 
>> On 30.04.2011, at 00:00, Devang Patel wrote:
>> 
>>> 
>>> On Apr 29, 2011, at 2:50 PM, Benjamin Kramer wrote:
>>>> 
>>>> Are you sure this specific transformation destroys the DebugLoc? When InstCombine simply replaces one
>>>> instruction with another it preserves debug information (InstructionCombining.cpp:1592).
>>> 
>>> I added line 1592 however I am not sure that this particular transformation is covered by 1592 in all cases. I have seen "shl" without DebugLoc generated by instcombine before your patch landed in svn. If you're interested I can send you more info.
>> 
>> There is a way this transform can lose a DebugLoc
>> 
>> x = C1 <<  A # with DebugLoc
>> y = x  << C2 # without DebugLoc
>> 
>> InstCombine will then replace 'y' with a new instruction (but there's no DebugLoc to attach to it) and 'x' is DCE'd.
>> I think the only feasible way to improve here is to find out why 'y' has no debug info attached to it.
>> 
>> If you have a test case where InstCombine introduces SHLs without DebugLocs, I'd be interested in seeing it.
> 
> More generally, we don't want to sprinkle setDebugLoc calls into *every* instcombine xform.  

One other alternative is to consistently use IRBuilder.

> We need a better solution.

I agree.
-
Devang


From dpatel at apple.com  Mon May  2 16:33:06 2011
From: dpatel at apple.com (Devang Patel)
Date: Mon, 02 May 2011 14:33:06 -0700
Subject: [llvm-commits] [llvm] r130485 - in /llvm/trunk:
 lib/Transforms/InstCombine/InstCombineShifts.cpp
 test/Transforms/InstCombine/shift.ll
In-Reply-To: <E3078F91-A806-4AEE-8F70-448C10F1B4C6@apple.com>
References: <20110429081541.B7F422A6C138@llvm.org>
	<C8DE37BF-B56C-4F53-A49F-EA158B233245@apple.com>
	<CB365302-F1CC-4578-BA54-7D954227E750@googlemail.com>
	<1B93035B-15A4-47B0-8E78-2D0F8A6B1701@apple.com>
	<5F0816C4-4FC4-4F02-8B7E-51858107AC6D@googlemail.com>
	<C9E399F9-DA42-4972-9BD4-130A27B3D1D5@apple.com>
	<E3078F91-A806-4AEE-8F70-448C10F1B4C6@apple.com>
Message-ID: <2E25739D-6707-4365-9DC5-E22C7C8E9644@apple.com>

Evan,

On May 2, 2011, at 2:00 PM, Evan Cheng wrote:

>> More generally, we don't want to sprinkle setDebugLoc calls into *every* instcombine xform.  We need a better solution.
> 
> I believe there is a long term plan being discussed. Devang, can you share what's being discussed so far?

As I understand, the long term plan being discussed will not help here. Instead, the plan assumes that all instructions and machine instruction will have DebugLoc. So in other words, the plan requires that we solve this particular issue first.

Without further delay, the plan being discussed seeks to find easier way to preserve debug values during code gen passes.
	- The idea is to make position of DBG_VALUE machine instructions in instruction stream irrelevant by using a new mechanism (we have not detailed what exactly, but it could be just another intrinsic or metadata pair or something else...) to pair a LLVM definition to respective source level variable information.  
	- At the end of compilation just before emitting out put, a super intelligent pass is employed to 
		a) look at variable info, 
		b) find corresponding value definitions, 
		c) find scopes and ranges of these definitions based on line number info attached with each machine instruction 
		d) construct complete debug info picture based on these information.

I am sure, others will correct me if I am missing something.
-
Devang

From bob.wilson at apple.com  Mon May  2 16:33:45 2011
From: bob.wilson at apple.com (Bob Wilson)
Date: Mon, 02 May 2011 21:33:45 -0000
Subject: [llvm-commits] [llvm-gcc-4.2] r130727 - in
 /llvm-gcc-4.2/trunk/gcc/config/arm: arm-cores.def arm.h darwin.h t-arm-elf
 t-darwin
Message-ID: <20110502213345.38D122A6C12C@llvm.org>

Author: bwilson
Date: Mon May  2 16:33:45 2011
New Revision: 130727

URL: http://llvm.org/viewvc/llvm-project?rev=130727&view=rev
Log:
Add some cortex-a9 and cortex-a9-mp entries in llvm-gcc.

Modified:
    llvm-gcc-4.2/trunk/gcc/config/arm/arm-cores.def
    llvm-gcc-4.2/trunk/gcc/config/arm/arm.h
    llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h
    llvm-gcc-4.2/trunk/gcc/config/arm/t-arm-elf
    llvm-gcc-4.2/trunk/gcc/config/arm/t-darwin

Modified: llvm-gcc-4.2/trunk/gcc/config/arm/arm-cores.def
URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/config/arm/arm-cores.def?rev=130727&r1=130726&r2=130727&view=diff
==============================================================================
--- llvm-gcc-4.2/trunk/gcc/config/arm/arm-cores.def (original)
+++ llvm-gcc-4.2/trunk/gcc/config/arm/arm-cores.def Mon May  2 16:33:45 2011
@@ -123,6 +123,7 @@
 ARM_CORE("cortex-a8",	  cortexa8,	7A,				 FL_LDSCHED, 9e)
 /* LLVM LOCAL begin */
 ARM_CORE("cortex-a9",	  cortexa9,	7A,				 FL_LDSCHED, 9e)
+ARM_CORE("cortex-a9-mp",  cortexa9mp,	7A,				 FL_LDSCHED, 9e)
 ARM_CORE("cortex-r4",	  cortexr4,	7R,				 FL_LDSCHED, 9e)
 ARM_CORE("cortex-m3",	  cortexm3,	7M,				 FL_LDSCHED, 9e)
 ARM_CORE("cortex-m4",	  cortexm4,	7M,				 FL_LDSCHED, 9e)

Modified: llvm-gcc-4.2/trunk/gcc/config/arm/arm.h
URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/config/arm/arm.h?rev=130727&r1=130726&r2=130727&view=diff
==============================================================================
--- llvm-gcc-4.2/trunk/gcc/config/arm/arm.h (original)
+++ llvm-gcc-4.2/trunk/gcc/config/arm/arm.h Mon May  2 16:33:45 2011
@@ -3498,6 +3498,7 @@
     case arm1156t2fs:   F.setCPU("arm1156t2f-s"); break; \
     case cortexa8:      F.setCPU("cortex-a8"); break; \
     case cortexa9:      F.setCPU("cortex-a9"); break; \
+    case cortexa9mp:    F.setCPU("cortex-a9-mp"); break; \
     case cortexr4:      F.setCPU("cortex-r4"); break; \
     case cortexm3:      F.setCPU("cortex-m3"); break; \
     case cortexm4:      F.setCPU("cortex-m4"); break; \

Modified: llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h
URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h?rev=130727&r1=130726&r2=130727&view=diff
==============================================================================
--- llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h (original)
+++ llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h Mon May  2 16:33:45 2011
@@ -236,6 +236,8 @@
    mcpu=cortex-a8:armv7;			\
    mcpu=cortex-r4:armv7;			\
    mcpu=cortex-m3:armv7;			\
+   mcpu=cortex-a9:armv7;			\
+   mcpu=cortex-a9-mp:armv7;			\
    :arm -force_cpusubtype_ALL}"
 
 #define DARWIN_MINVERSION_SPEC "3.0"

Modified: llvm-gcc-4.2/trunk/gcc/config/arm/t-arm-elf
URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/config/arm/t-arm-elf?rev=130727&r1=130726&r2=130727&view=diff
==============================================================================
--- llvm-gcc-4.2/trunk/gcc/config/arm/t-arm-elf (original)
+++ llvm-gcc-4.2/trunk/gcc/config/arm/t-arm-elf Mon May  2 16:33:45 2011
@@ -21,6 +21,7 @@
 #MULTILIB_MATCHES      += march?armv7=mcpu?cortex-a8
 #MULTILIB_MATCHES      += march?armv7=mcpu?cortex-r4
 #MULTILIB_MATCHES      += march?armv7=mcpu?cortex-m3
+#MULTILIB_MATCHES      += march?armv7=mcpu?cortex-a9
 # APPLE LOCAL end v7 support. Merge from mainline
 
 # MULTILIB_OPTIONS    += mcpu=ep9312

Modified: llvm-gcc-4.2/trunk/gcc/config/arm/t-darwin
URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/config/arm/t-darwin?rev=130727&r1=130726&r2=130727&view=diff
==============================================================================
--- llvm-gcc-4.2/trunk/gcc/config/arm/t-darwin (original)
+++ llvm-gcc-4.2/trunk/gcc/config/arm/t-darwin Mon May  2 16:33:45 2011
@@ -67,7 +67,8 @@
 ifneq (,$(findstring armv7,$(ARM_MULTILIB_ARCHS)))
 MULTILIB_MATCHES += \
   march?armv7a=march?armv7-a \
-  march?armv7a=mcpu?cortex-a8
+  march?armv7a=mcpu?cortex-a8 \
+  march?armv7a=mcpu?cortex-a9
 endif
 
 # APPLE LOCAL end 6611402 configurable multilib architectures



From bruno.cardoso at gmail.com  Mon May  2 16:38:56 2011
From: bruno.cardoso at gmail.com (Bruno Cardoso Lopes)
Date: Mon, 2 May 2011 18:38:56 -0300
Subject: [llvm-commits] [PATCH] arm coprocessor builtins and intrinsics!
In-Reply-To: <60016C47-371D-48B6-A7DE-413D41592962@apple.com>
References: <BANLkTikkHzT9r=hQeqAgnND+Sea_TbCT_g@mail.gmail.com>
	<60016C47-371D-48B6-A7DE-413D41592962@apple.com>
Message-ID: <BANLkTike9ws0xjUCFJAG0hg_-eM82ZW6Jw@mail.gmail.com>

On Mon, May 2, 2011 at 6:12 PM, Evan Cheng <evan.cheng at apple.com> wrote:
> Hi Bruno,
>
> These look fine to me. However, is it necessary to use def : Pat patterns? Can't you fold the patterns into the instruction defs?

Nice catch, I forgot about that! :)


-- 
Bruno Cardoso Lopes
http://www.brunocardoso.cc

From evan.cheng at apple.com  Mon May  2 16:44:02 2011
From: evan.cheng at apple.com (Evan Cheng)
Date: Mon, 02 May 2011 14:44:02 -0700
Subject: [llvm-commits] [llvm] r130551 - in /llvm/trunk:
 include/llvm/Transforms/Utils/BasicBlockUtils.h
 lib/Transforms/Utils/BasicBlockUtils.cpp
 lib/Transforms/Utils/SSAUpdater.cpp
In-Reply-To: <4DBBFF5A.5000505@free.fr>
References: <20110429222859.4F3242A6C135@llvm.org> <4DBBFF5A.5000505@free.fr>
Message-ID: <4770EA2C-AFAC-4D7D-BCA4-07BF0DD4EB04@apple.com>

Hi Devang,

I share Duncan's concern. This may impact compile time. Is the scanning the only solution?

Evan

On Apr 30, 2011, at 5:23 AM, Duncan Sands wrote:

> Hi Devang,
> 
>> +/// GetFirstDebugLocInBasicBlock - Return first valid DebugLoc entry in a
>> +/// given basic block.
>> +DebugLoc llvm::GetFirstDebugLocInBasicBlock(const BasicBlock *BB) {
>> +  for (BasicBlock::const_iterator BI = BB->begin(), BE = BB->end();
>> +       BI != BE; ++BI) {
>> +    DebugLoc DL = BI->getDebugLoc();
>> +    if (!DL.isUnknown())
>> +      return DL;
>> +  }
>> +  return DebugLoc();
>> +}
> 
> this potentially scans through the entire basic block which might be expensive
> if it is big.  Also, it's not clear to me that this makes any sense.  Suppose
> the first instructions in the basic block have no line number info.  Why should
> the phi node get the same line number as some random instruction deep down in
> the basic block?  How about just giving the phi node the same line info as the
> first instruction in the basic block?  If it doesn't have any then too bad.
> 
> Ciao, Duncan.
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits


From dpatel at apple.com  Mon May  2 16:54:39 2011
From: dpatel at apple.com (Devang Patel)
Date: Mon, 02 May 2011 14:54:39 -0700
Subject: [llvm-commits] [llvm] r130551 - in /llvm/trunk:
 include/llvm/Transforms/Utils/BasicBlockUtils.h
 lib/Transforms/Utils/BasicBlockUtils.cpp
 lib/Transforms/Utils/SSAUpdater.cpp
In-Reply-To: <4770EA2C-AFAC-4D7D-BCA4-07BF0DD4EB04@apple.com>
References: <20110429222859.4F3242A6C135@llvm.org> <4DBBFF5A.5000505@free.fr>
	<4770EA2C-AFAC-4D7D-BCA4-07BF0DD4EB04@apple.com>
Message-ID: <9CA12E64-E501-4843-9FF4-F9D70047FC51@apple.com>


On May 2, 2011, at 2:44 PM, Evan Cheng wrote:

> Hi Devang,
> 
> I share Duncan's concern. This may impact compile time. Is the scanning the only solution?

I understand the concern. I have not come up with a better alternative. But for now, I'll update it to use first instruction's location.

-
Devang

> 
> Evan
> 
> On Apr 30, 2011, at 5:23 AM, Duncan Sands wrote:
> 
>> Hi Devang,
>> 
>>> +/// GetFirstDebugLocInBasicBlock - Return first valid DebugLoc entry in a
>>> +/// given basic block.
>>> +DebugLoc llvm::GetFirstDebugLocInBasicBlock(const BasicBlock *BB) {
>>> +  for (BasicBlock::const_iterator BI = BB->begin(), BE = BB->end();
>>> +       BI != BE; ++BI) {
>>> +    DebugLoc DL = BI->getDebugLoc();
>>> +    if (!DL.isUnknown())
>>> +      return DL;
>>> +  }
>>> +  return DebugLoc();
>>> +}
>> 
>> this potentially scans through the entire basic block which might be expensive
>> if it is big.  Also, it's not clear to me that this makes any sense.  Suppose
>> the first instructions in the basic block have no line number info.  Why should
>> the phi node get the same line number as some random instruction deep down in
>> the basic block?  How about just giving the phi node the same line info as the
>> first instruction in the basic block?  If it doesn't have any then too bad.
>> 
>> Ciao, Duncan.
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
> 


From dpatel at apple.com  Mon May  2 16:57:00 2011
From: dpatel at apple.com (Devang Patel)
Date: Mon, 02 May 2011 21:57:00 -0000
Subject: [llvm-commits] [llvm] r130729 -
	/llvm/trunk/lib/Transforms/Utils/BasicBlockUtils.cpp
Message-ID: <20110502215700.635DE2A6C12C@llvm.org>

Author: dpatel
Date: Mon May  2 16:57:00 2011
New Revision: 130729

URL: http://llvm.org/viewvc/llvm-project?rev=130729&view=rev
Log:
Scanning entire basic block may be too expensive in terms of compile time. Instead, just use whatever location info first non-phi instruction has.


Modified:
    llvm/trunk/lib/Transforms/Utils/BasicBlockUtils.cpp

Modified: llvm/trunk/lib/Transforms/Utils/BasicBlockUtils.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Utils/BasicBlockUtils.cpp?rev=130729&r1=130728&r2=130729&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Utils/BasicBlockUtils.cpp (original)
+++ llvm/trunk/lib/Transforms/Utils/BasicBlockUtils.cpp Mon May  2 16:57:00 2011
@@ -542,11 +542,9 @@
 /// GetFirstDebugLocInBasicBlock - Return first valid DebugLoc entry in a 
 /// given basic block.
 DebugLoc llvm::GetFirstDebugLocInBasicBlock(const BasicBlock *BB) {
-  for (BasicBlock::const_iterator BI = BB->begin(), BE = BB->end(); 
-       BI != BE; ++BI) {
-    DebugLoc DL = BI->getDebugLoc();
-    if (!DL.isUnknown())
-      return DL;
-  }
+  if (const Instruction *I = BB->getFirstNonPHI())
+    return I->getDebugLoc();
+  // Scanning entire block may be too expensive, if the first instruction
+  // does not have valid location info.
   return DebugLoc();
 }



From resistor at mac.com  Mon May  2 17:25:45 2011
From: resistor at mac.com (Owen Anderson)
Date: Mon, 02 May 2011 22:25:45 -0000
Subject: [llvm-commits] [llvm] r130733 -
	/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
Message-ID: <20110502222545.B18482A6C12C@llvm.org>

Author: resistor
Date: Mon May  2 17:25:45 2011
New Revision: 130733

URL: http://llvm.org/viewvc/llvm-project?rev=130733&view=rev
Log:
Other parts of the SelectionDAG framework assume that targets use their pointer type for vector indices.  Make the vector unrolling code respect that.

Modified:
    llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp?rev=130733&r1=130732&r2=130733&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAG.cpp Mon May  2 17:25:45 2011
@@ -6304,7 +6304,7 @@
         Operands[j] = getNode(ISD::EXTRACT_VECTOR_ELT, dl,
                               OperandEltVT,
                               Operand,
-                              getConstant(i, MVT::i32));
+                              getConstant(i, TLI.getPointerTy()));
       } else {
         // A scalar operand; just use it as is.
         Operands[j] = Operand;



From gohman at apple.com  Mon May  2 19:46:49 2011
From: gohman at apple.com (Dan Gohman)
Date: Tue, 03 May 2011 00:46:49 -0000
Subject: [llvm-commits] [llvm] r130743 - in /llvm/trunk:
 include/llvm/Target/TargetLowering.h lib/Target/ARM/ARMISelLowering.cpp
 lib/Target/ARM/ARMISelLowering.h
 lib/Transforms/Scalar/LoopStrengthReduce.cpp
 test/CodeGen/ARM/lsr-unfolded-offset.ll
Message-ID: <20110503004649.753432A6C12C@llvm.org>

Author: djg
Date: Mon May  2 19:46:49 2011
New Revision: 130743

URL: http://llvm.org/viewvc/llvm-project?rev=130743&view=rev
Log:
Add an unfolded offset field to LSR's Formula record. This is used to
model constants which can be added to base registers via add-immediate
instructions which don't require an additional register to materialize
the immediate.

Added:
    llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll
Modified:
    llvm/trunk/include/llvm/Target/TargetLowering.h
    llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
    llvm/trunk/lib/Target/ARM/ARMISelLowering.h
    llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp

Modified: llvm/trunk/include/llvm/Target/TargetLowering.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Target/TargetLowering.h?rev=130743&r1=130742&r2=130743&view=diff
==============================================================================
--- llvm/trunk/include/llvm/Target/TargetLowering.h (original)
+++ llvm/trunk/include/llvm/Target/TargetLowering.h Mon May  2 19:46:49 2011
@@ -1583,6 +1583,14 @@
     return true;
   }
 
+  /// isLegalAddImmediate - Return true if the specified immediate is legal
+  /// add immediate, that is the target has add instructions which can add
+  /// a register with the immediate without having to materialize the
+  /// immediate into a register.
+  virtual bool isLegalAddImmediate(int64_t Imm) const {
+    return true;
+  }
+
   //===--------------------------------------------------------------------===//
   // Div utility functions
   //

Modified: llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp?rev=130743&r1=130742&r2=130743&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp Mon May  2 19:46:49 2011
@@ -6984,6 +6984,14 @@
   return Imm >= 0 && Imm <= 255;
 }
 
+/// isLegalAddImmediate - Return true if the specified immediate is legal
+/// add immediate, that is the target has add instructions which can add
+/// a register with the immediate without having to materialize the
+/// immediate into a register.
+bool ARMTargetLowering::isLegalAddImmediate(int64_t Imm) const {
+  return ARM_AM::getSOImmVal(Imm) != -1;
+}
+
 static bool getARMIndexedAddressParts(SDNode *Ptr, EVT VT,
                                       bool isSEXTLoad, SDValue &Base,
                                       SDValue &Offset, bool &isInc,

Modified: llvm/trunk/lib/Target/ARM/ARMISelLowering.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMISelLowering.h?rev=130743&r1=130742&r2=130743&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMISelLowering.h (original)
+++ llvm/trunk/lib/Target/ARM/ARMISelLowering.h Mon May  2 19:46:49 2011
@@ -264,6 +264,12 @@
     /// the immediate into a register.
     virtual bool isLegalICmpImmediate(int64_t Imm) const;
 
+    /// isLegalAddImmediate - Return true if the specified immediate is legal
+    /// add immediate, that is the target has add instructions which can
+    /// add a register and the immediate without having to materialize
+    /// the immediate into a register.
+    virtual bool isLegalAddImmediate(int64_t Imm) const;
+
     /// getPreIndexedAddressParts - returns true by value, base pointer and
     /// offset pointer and addressing mode by reference if the node's address
     /// can be legally represented as pre-indexed load / store address.

Modified: llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp?rev=130743&r1=130742&r2=130743&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/LoopStrengthReduce.cpp Mon May  2 19:46:49 2011
@@ -209,7 +209,12 @@
   /// when AM.Scale is not zero.
   const SCEV *ScaledReg;
 
-  Formula() : ScaledReg(0) {}
+  /// UnfoldedOffset - An additional constant offset which added near the
+  /// use. This requires a temporary register, but the offset itself can
+  /// live in an add immediate field rather than a register.
+  int64_t UnfoldedOffset;
+
+  Formula() : ScaledReg(0), UnfoldedOffset(0) {}
 
   void InitialMatch(const SCEV *S, Loop *L, ScalarEvolution &SE);
 
@@ -379,6 +384,10 @@
       OS << "<unknown>";
     OS << ')';
   }
+  if (UnfoldedOffset != 0) {
+    if (!First) OS << " + "; else First = false;
+    OS << "imm(" << UnfoldedOffset << ')';
+  }
 }
 
 void Formula::dump() const {
@@ -771,8 +780,10 @@
     RatePrimaryRegister(BaseReg, Regs, L, SE, DT);
   }
 
-  if (F.BaseRegs.size() > 1)
-    NumBaseAdds += F.BaseRegs.size() - 1;
+  // Determine how many (unfolded) adds we'll need inside the loop.
+  size_t NumBaseParts = F.BaseRegs.size() + (F.UnfoldedOffset != 0);
+  if (NumBaseParts > 1)
+    NumBaseAdds += NumBaseParts - 1;
 
   // Tally up the non-zero immediates.
   for (SmallVectorImpl<int64_t>::const_iterator I = Offsets.begin(),
@@ -1945,7 +1956,8 @@
         if (F.BaseRegs == OrigF.BaseRegs &&
             F.ScaledReg == OrigF.ScaledReg &&
             F.AM.BaseGV == OrigF.AM.BaseGV &&
-            F.AM.Scale == OrigF.AM.Scale) {
+            F.AM.Scale == OrigF.AM.Scale &&
+            F.UnfoldedOffset == OrigF.UnfoldedOffset) {
           if (F.AM.BaseOffs == 0)
             return &LU;
           // This is the formula where all the registers and symbols matched;
@@ -2313,8 +2325,29 @@
       if (InnerSum->isZero())
         continue;
       Formula F = Base;
-      F.BaseRegs[i] = InnerSum;
-      F.BaseRegs.push_back(*J);
+
+      // Add the remaining pieces of the add back into the new formula.
+      const SCEVConstant *InnerSumSC = dyn_cast<SCEVConstant>(InnerSum);
+      if (TLI && InnerSumSC &&
+          SE.getTypeSizeInBits(InnerSumSC->getType()) <= 64 &&
+          TLI->isLegalAddImmediate((uint64_t)F.UnfoldedOffset +
+                                   InnerSumSC->getValue()->getZExtValue())) {
+        F.UnfoldedOffset = (uint64_t)F.UnfoldedOffset +
+                           InnerSumSC->getValue()->getZExtValue();
+        F.BaseRegs.erase(F.BaseRegs.begin() + i);
+      } else
+        F.BaseRegs[i] = InnerSum;
+
+      // Add J as its own register, or an unfolded immediate.
+      const SCEVConstant *SC = dyn_cast<SCEVConstant>(*J);
+      if (TLI && SC && SE.getTypeSizeInBits(SC->getType()) <= 64 &&
+          TLI->isLegalAddImmediate((uint64_t)F.UnfoldedOffset +
+                                   SC->getValue()->getZExtValue()))
+        F.UnfoldedOffset = (uint64_t)F.UnfoldedOffset +
+                           SC->getValue()->getZExtValue();
+      else
+        F.BaseRegs.push_back(*J);
+
       if (InsertFormula(LU, LUIdx, F))
         // If that formula hadn't been seen before, recurse to find more like
         // it.
@@ -2482,6 +2515,13 @@
         continue;
     }
 
+    // Check that multiplying with the unfolded offset doesn't overflow.
+    if (F.UnfoldedOffset != 0) {
+      F.UnfoldedOffset = (uint64_t)F.UnfoldedOffset * Factor;
+      if (F.UnfoldedOffset / Factor != Base.UnfoldedOffset)
+        continue;
+    }
+
     // If we make it here and it's legal, add it.
     (void)InsertFormula(LU, LUIdx, F);
   next:;
@@ -2664,7 +2704,7 @@
       // other orig regs.
       ImmMapTy::const_iterator OtherImms[] = {
         Imms.begin(), prior(Imms.end()),
-        Imms.upper_bound((Imms.begin()->first + prior(Imms.end())->first) / 2)
+        Imms.lower_bound((Imms.begin()->first + prior(Imms.end())->first) / 2)
       };
       for (size_t i = 0, e = array_lengthof(OtherImms); i != e; ++i) {
         ImmMapTy::const_iterator M = OtherImms[i];
@@ -2738,8 +2778,13 @@
           Formula NewF = F;
           NewF.AM.BaseOffs = (uint64_t)NewF.AM.BaseOffs + Imm;
           if (!isLegalUse(NewF.AM, LU.MinOffset, LU.MaxOffset,
-                          LU.Kind, LU.AccessTy, TLI))
-            continue;
+                          LU.Kind, LU.AccessTy, TLI)) {
+            if (!TLI ||
+                !TLI->isLegalAddImmediate((uint64_t)NewF.UnfoldedOffset + Imm))
+              continue;
+            NewF = F;
+            NewF.UnfoldedOffset = (uint64_t)NewF.UnfoldedOffset + Imm;
+          }
           NewF.BaseRegs[N] = SE.getAddExpr(NegImmS, BaseReg);
 
           // If the new formula has a constant in a register, and adding the
@@ -3488,6 +3533,14 @@
     }
   }
 
+  // Expand the unfolded offset portion.
+  int64_t UnfoldedOffset = F.UnfoldedOffset;
+  if (UnfoldedOffset != 0) {
+    // Just add the immediate values.
+    Ops.push_back(SE.getUnknown(ConstantInt::getSigned(IntTy,
+                                                       UnfoldedOffset)));
+  }
+
   // Emit instructions summing all the operands.
   const SCEV *FullS = Ops.empty() ?
                       SE.getConstant(IntTy, 0) :

Added: llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll?rev=130743&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll (added)
+++ llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll Mon May  2 19:46:49 2011
@@ -0,0 +1,80 @@
+; RUN: llc < %s | FileCheck %s
+
+; LSR shouldn't introduce more induction variables than needed, increasing
+; register pressure and therefore spilling. There is more room for improvement
+; here.
+
+; CHECK: sub sp, #32
+
+; CHECK:      ldr r{{.*}}, [sp, #4]
+; CHECK-NEXT: ldr r{{.*}}, [sp, #16]
+; CHECK-NEXT: ldr r{{.*}}, [sp, #12]
+; CHECK-NEXT: adds
+
+target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:32:64-v128:32:128-a0:0:32-n32"
+target triple = "thumbv7-apple-macosx10.7.0"
+
+%struct.partition_entry = type { i32, i32, i64, i64 }
+
+define i32 @partition_overlap_check(%struct.partition_entry* nocapture %part, i32 %num_entries) nounwind readonly optsize ssp {
+entry:
+  %cmp79 = icmp sgt i32 %num_entries, 0
+  br i1 %cmp79, label %outer.loop, label %for.end72
+
+outer.loop:                                 ; preds = %for.inc69, %entry
+  %overlap.081 = phi i32 [ %overlap.4, %for.inc69 ], [ 0, %entry ]
+  %0 = phi i32 [ %inc71, %for.inc69 ], [ 0, %entry ]
+  %offset = getelementptr %struct.partition_entry* %part, i32 %0, i32 2
+  %len = getelementptr %struct.partition_entry* %part, i32 %0, i32 3
+  %tmp5 = load i64* %offset, align 4, !tbaa !0
+  %tmp15 = load i64* %len, align 4, !tbaa !0
+  %add = add nsw i64 %tmp15, %tmp5
+  br label %inner.loop
+
+inner.loop:                                       ; preds = %for.inc, %outer.loop
+  %overlap.178 = phi i32 [ %overlap.081, %outer.loop ], [ %overlap.4, %for.inc ]
+  %1 = phi i32 [ 0, %outer.loop ], [ %inc, %for.inc ]
+  %cmp23 = icmp eq i32 %0, %1
+  br i1 %cmp23, label %for.inc, label %if.end
+
+if.end:                                           ; preds = %inner.loop
+  %len39 = getelementptr %struct.partition_entry* %part, i32 %1, i32 3
+  %offset28 = getelementptr %struct.partition_entry* %part, i32 %1, i32 2
+  %tmp29 = load i64* %offset28, align 4, !tbaa !0
+  %tmp40 = load i64* %len39, align 4, !tbaa !0
+  %add41 = add nsw i64 %tmp40, %tmp29
+  %cmp44 = icmp sge i64 %tmp29, %tmp5
+  %cmp47 = icmp slt i64 %tmp29, %add
+  %or.cond = and i1 %cmp44, %cmp47
+  %overlap.2 = select i1 %or.cond, i32 1, i32 %overlap.178
+  %cmp52 = icmp sle i64 %add41, %add
+  %cmp56 = icmp sgt i64 %add41, %tmp5
+  %or.cond74 = and i1 %cmp52, %cmp56
+  %overlap.3 = select i1 %or.cond74, i32 1, i32 %overlap.2
+  %cmp61 = icmp sgt i64 %tmp29, %tmp5
+  %cmp65 = icmp slt i64 %add41, %add
+  %or.cond75 = or i1 %cmp61, %cmp65
+  br i1 %or.cond75, label %for.inc, label %if.then66
+
+if.then66:                                        ; preds = %if.end
+  br label %for.inc
+
+for.inc:                                          ; preds = %if.end, %if.then66, %inner.loop
+  %overlap.4 = phi i32 [ %overlap.178, %inner.loop ], [ 1, %if.then66 ], [ %overlap.3, %if.end ]
+  %inc = add nsw i32 %1, 1
+  %exitcond = icmp eq i32 %inc, %num_entries
+  br i1 %exitcond, label %for.inc69, label %inner.loop
+
+for.inc69:                                        ; preds = %for.inc
+  %inc71 = add nsw i32 %0, 1
+  %exitcond83 = icmp eq i32 %inc71, %num_entries
+  br i1 %exitcond83, label %for.end72, label %outer.loop
+
+for.end72:                                        ; preds = %for.inc69, %entry
+  %overlap.0.lcssa = phi i32 [ 0, %entry ], [ %overlap.4, %for.inc69 ]
+  ret i32 %overlap.0.lcssa
+}
+
+!0 = metadata !{metadata !"long long", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA", null}



From echristo at apple.com  Mon May  2 20:28:32 2011
From: echristo at apple.com (Eric Christopher)
Date: Tue, 03 May 2011 01:28:32 -0000
Subject: [llvm-commits] [llvm] r130747 - in /llvm/trunk:
 lib/Target/X86/X86InstrSSE.td test/MC/X86/x86-32-coverage.s
Message-ID: <20110503012833.4AD0F2A6C12C@llvm.org>

Author: echristo
Date: Mon May  2 20:28:32 2011
New Revision: 130747

URL: http://llvm.org/viewvc/llvm-project?rev=130747&view=rev
Log:
xmm0 is an implicit parameter in this and so shouldn't be in the
string template.

Fixes rdar://8493866

Modified:
    llvm/trunk/lib/Target/X86/X86InstrSSE.td
    llvm/trunk/test/MC/X86/x86-32-coverage.s

Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrSSE.td?rev=130747&r1=130746&r2=130747&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrSSE.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrSSE.td Mon May  2 20:28:32 2011
@@ -4733,14 +4733,14 @@
     def rr0 : SS48I<opc, MRMSrcReg, (outs VR128:$dst),
                     (ins VR128:$src1, VR128:$src2),
                     !strconcat(OpcodeStr,
-                     "\t{%xmm0, $src2, $dst|$dst, $src2, %xmm0}"),
+                     "\t{$src2, $dst|$dst, $src2}"),
                     [(set VR128:$dst, (IntId VR128:$src1, VR128:$src2, XMM0))]>,
                     OpSize;
 
     def rm0 : SS48I<opc, MRMSrcMem, (outs VR128:$dst),
                     (ins VR128:$src1, i128mem:$src2),
                     !strconcat(OpcodeStr,
-                     "\t{%xmm0, $src2, $dst|$dst, $src2, %xmm0}"),
+                     "\t{$src2, $dst|$dst, $src2}"),
                     [(set VR128:$dst,
                       (IntId VR128:$src1,
                        (bitconvert (memopv16i8 addr:$src2)), XMM0))]>, OpSize;

Modified: llvm/trunk/test/MC/X86/x86-32-coverage.s
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/MC/X86/x86-32-coverage.s?rev=130747&r1=130746&r2=130747&view=diff
==============================================================================
--- llvm/trunk/test/MC/X86/x86-32-coverage.s (original)
+++ llvm/trunk/test/MC/X86/x86-32-coverage.s Mon May  2 20:28:32 2011
@@ -1,3 +1,7 @@
+# Build output for (null)
+# Generated at (null)
+# Using (null) configuration, (null) architecture for (null) target of (null) project
+
 // RUN: llvm-mc -triple i386-unknown-unknown %s --show-encoding  | FileCheck %s
 
 // CHECK: 	movb	$127, 3735928559(%ebx,%ecx,8)
@@ -19570,3 +19574,8 @@
 
 // CHECK: 	aeskeygenassist	$125, (%edx,%eax,4), %xmm2
                 aeskeygenassist $125, (%edx,%eax,4), %xmm2
+
+// CHECK:   blendvps	(%rax), %xmm1   # encoding: [0x66,0x0f,0x38,0x14,0x08]
+            blendvps (%rax), %xmm1
+// CHECK:   blendvps	%xmm2, %xmm1    # encoding: [0x66,0x0f,0x38,0x14,0xca]
+            blendvps %xmm2, %xmm1



From evan.cheng at apple.com  Mon May  2 21:57:10 2011
From: evan.cheng at apple.com (Evan Cheng)
Date: Mon, 02 May 2011 19:57:10 -0700
Subject: [llvm-commits] [llvm] r130551 - in /llvm/trunk:
 include/llvm/Transforms/Utils/BasicBlockUtils.h
 lib/Transforms/Utils/BasicBlockUtils.cpp
 lib/Transforms/Utils/SSAUpdater.cpp
In-Reply-To: <9CA12E64-E501-4843-9FF4-F9D70047FC51@apple.com>
References: <20110429222859.4F3242A6C135@llvm.org> <4DBBFF5A.5000505@free.fr>
	<4770EA2C-AFAC-4D7D-BCA4-07BF0DD4EB04@apple.com>
	<9CA12E64-E501-4843-9FF4-F9D70047FC51@apple.com>
Message-ID: <F1F92552-7AE2-46CB-AD47-46E223209890@apple.com>


On May 2, 2011, at 2:54 PM, Devang Patel wrote:

> 
> On May 2, 2011, at 2:44 PM, Evan Cheng wrote:
> 
>> Hi Devang,
>> 
>> I share Duncan's concern. This may impact compile time. Is the scanning the only solution?
> 
> I understand the concern. I have not come up with a better alternative. But for now, I'll update it to use first instruction's location.

Thanks.

Evan
> 
> -
> Devang
> 
>> 
>> Evan
>> 
>> On Apr 30, 2011, at 5:23 AM, Duncan Sands wrote:
>> 
>>> Hi Devang,
>>> 
>>>> +/// GetFirstDebugLocInBasicBlock - Return first valid DebugLoc entry in a
>>>> +/// given basic block.
>>>> +DebugLoc llvm::GetFirstDebugLocInBasicBlock(const BasicBlock *BB) {
>>>> +  for (BasicBlock::const_iterator BI = BB->begin(), BE = BB->end();
>>>> +       BI != BE; ++BI) {
>>>> +    DebugLoc DL = BI->getDebugLoc();
>>>> +    if (!DL.isUnknown())
>>>> +      return DL;
>>>> +  }
>>>> +  return DebugLoc();
>>>> +}
>>> 
>>> this potentially scans through the entire basic block which might be expensive
>>> if it is big.  Also, it's not clear to me that this makes any sense.  Suppose
>>> the first instructions in the basic block have no line number info.  Why should
>>> the phi node get the same line number as some random instruction deep down in
>>> the basic block?  How about just giving the phi node the same line info as the
>>> first instruction in the basic block?  If it doesn't have any then too bad.
>>> 
>>> Ciao, Duncan.
>>> _______________________________________________
>>> llvm-commits mailing list
>>> llvm-commits at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>> 
> 


From bigcheesegs at gmail.com  Mon May  2 22:42:50 2011
From: bigcheesegs at gmail.com (Michael J. Spencer)
Date: Tue, 03 May 2011 03:42:50 -0000
Subject: [llvm-commits] [llvm] r130749 - /llvm/trunk/lib/Target/X86/X86.td
Message-ID: <20110503034250.4BD842A6C12C@llvm.org>

Author: mspencer
Date: Mon May  2 22:42:50 2011
New Revision: 130749

URL: http://llvm.org/viewvc/llvm-project?rev=130749&view=rev
Log:
Add pentium{3,4}m cpus. Patch by Alexander Best!

Modified:
    llvm/trunk/lib/Target/X86/X86.td

Modified: llvm/trunk/lib/Target/X86/X86.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86.td?rev=130749&r1=130748&r2=130749&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86.td (original)
+++ llvm/trunk/lib/Target/X86/X86.td Mon May  2 22:42:50 2011
@@ -101,8 +101,10 @@
 def : Proc<"pentiumpro",      [FeatureCMOV]>;
 def : Proc<"pentium2",        [FeatureMMX, FeatureCMOV]>;
 def : Proc<"pentium3",        [FeatureSSE1]>;
+def : Proc<"pentium3m",       [FeatureSSE1, FeatureSlowBTMem]>;
 def : Proc<"pentium-m",       [FeatureSSE2, FeatureSlowBTMem]>;
 def : Proc<"pentium4",        [FeatureSSE2]>;
+def : Proc<"pentium4m",       [FeatureSSE2, FeatureSlowBTMem]>;
 def : Proc<"x86-64",          [FeatureSSE2,   Feature64Bit, FeatureSlowBTMem]>;
 def : Proc<"yonah",           [FeatureSSE3, FeatureSlowBTMem]>;
 def : Proc<"prescott",        [FeatureSSE3, FeatureSlowBTMem]>;



From grosser at fim.uni-passau.de  Tue May  3 04:40:41 2011
From: grosser at fim.uni-passau.de (Tobias Grosser)
Date: Tue, 03 May 2011 09:40:41 -0000
Subject: [llvm-commits] [polly] r130751 - in /polly/trunk/www: examples.html
 experiments/matmul/runall.sh menu.css menu.html.incl
Message-ID: <20110503094041.2AD0F2A6C12C@llvm.org>

Author: grosser
Date: Tue May  3 04:40:40 2011
New Revision: 130751

URL: http://llvm.org/viewvc/llvm-project?rev=130751&view=rev
Log:
www: Finish first draft of the matmul example

Modified:
    polly/trunk/www/examples.html
    polly/trunk/www/experiments/matmul/runall.sh
    polly/trunk/www/menu.css
    polly/trunk/www/menu.html.incl

Modified: polly/trunk/www/examples.html
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/examples.html?rev=130751&r1=130750&r2=130751&view=diff
==============================================================================
--- polly/trunk/www/examples.html (original)
+++ polly/trunk/www/examples.html Tue May  3 04:40:40 2011
@@ -20,14 +20,15 @@
 
 <p>Polly does not yet focus on end user, but on research and the development of
 new optimizations. Hence for the users of Polly it is often necessary to
-understand how Polly works internally. To get an overview of the different steps
+understand how Polly works internally. To get an to know the different steps
 taken during polyhedral compilation, we give a step by step example on how to
 use the different Polly passes. For this we optimize a simple matrix
 multiplication kernel. In case you look for a more automated way of executing
 Polly, check out the pollycc tool in utils/pollycc.</p>
 
 The files used and created in this example are available <a
-href="experiments/matmul">here</a>.
+href="experiments/matmul">here</a>. They can be created automatically by running
+the <a href="experiments/matmul/runall.sh">runall.sh</a> script.
 
 <ol>
 <li><h4>Create LLVM-IR from the C code</h4>
@@ -57,14 +58,14 @@
 
 Polly is only able to work with code that matches a canonical form. To translate
 the LLVM-IR into this form we use a set of canonicalication passes. For this
-example only three passes are necessary. To get good coverage on a larger set
-of input files a larger set is needed. pollycc contains a set of passes that has
-shown to be beneficial.
+example only three passes are necessary. To get good coverage on more
+complicated input files often more canonicalization passes are needed. pollycc
+contains a list of passes that have shown to be beneficial.
 <pre class="code">opt -S -mem2reg -loop-simplify -indvars matmul.s &gt; matmul.preopt.ll</pre></li>
 
 <li><h4>Show the SCoPs detected by Polly (optional)</h4>
 
-To understand if Polly was able to detect some SCoPs, we print the
+To understand if Polly was able to detect SCoPs, we print the
 structure of the detected SCoPs. In our example two SCoPs were detected. One in
 'init_array' the other in 'main'.
 
@@ -112,7 +113,8 @@
 <pre class="code">opt -basicaa -polly-scops -analyze matmul.preopt.ll</pre>
 <pre>
 [...]
-Printing analysis 'Polly - Create polyhedral description of Scops' for region: '%1 =&gt;&nbsp;%17' in function 'init_array':
+Printing analysis 'Polly - Create polyhedral description of Scops' for region:
+'%1 =&gt;&nbsp;%17' in function 'init_array':
    Context:
    { [] }
    Statements {
@@ -135,9 +137,9 @@
            ReadAccess&nbsp;:= 
                { FinalRead[i0] -&gt; MemRef_B[o0] };
    }
-Printing analysis 'Polly - Create polyhedral description of Scops' for region: '%0 =&gt; &lt;Function  Return&gt;' in function 'init_array':
 [...]
-Printing analysis 'Polly - Create polyhedral description of Scops' for region: '%1 =&gt;&nbsp;%17' in function 'main':
+Printing analysis 'Polly - Create polyhedral description of Scops' for region:
+'%1 =&gt;&nbsp;%17' in function 'main':
    Context:
    { [] }
    Statements {
@@ -173,14 +175,14 @@
            ReadAccess&nbsp;:= 
                { FinalRead[i0] -&gt; MemRef_B[o0] };
    }
-Printing analysis 'Polly - Create polyhedral description of Scops' for region: '%0 =&gt; &lt;Function  Return&gt;' in function 'main':
-Invalid Scop!
+[...]
 </pre>
 </li>
 
 <li><h4>Show the dependences for the SCoPs</h4>
 <pre class="code">opt -basicaa -polly-dependences -analyze matmul.preopt.ll</pre>
-<pre>Printing analysis 'Polly - Calculate dependences for SCoP' for region: 'for.cond =&gt; for.end28' in function 'init_array':
+<pre>Printing analysis 'Polly - Calculate dependences for SCoP' for region:
+'for.cond =&gt; for.end28' in function 'init_array':
    Must dependences:
        {  }
    May dependences:
@@ -189,7 +191,8 @@
        {  }
    May no source:
        {  }
-Printing analysis 'Polly - Calculate dependences for SCoP' for region: 'for.cond =&gt; for.end48' in function 'main':
+Printing analysis 'Polly - Calculate dependences for SCoP' for region:
+'for.cond =&gt; for.end48' in function 'main':
    Must dependences:
        {  Stmt_4[i0, i1] -&gt; Stmt_6[i0, i1, 0]&nbsp;:
               i0 &gt;= 0 and i0 &lt;= 1023 and i1 &gt;= 0 and i1 &lt;= 1023;
@@ -228,51 +231,191 @@
 
 <li><h4>Import the changed jscop files and print the updated SCoP structure
 (optional)</h4>
-<p>Polly can import jscop files, where the schedules of the statements were
-changed. With the help of these updated files we can import transformations into
-Polly. It is possible to import different jscop files by providing the postfix
+<p>Polly can reimport jscop files, in which the schedules of the statements are
+changed. These changed schedules are used to descripe transformations.
+It is possible to import different jscop files by providing the postfix
 of the jscop file that is imported.</p>
-<p> The optimized jscop files for this example are hand written. The schedule
-used was inspired by looking at the optimizations PoCC performs. If PoCC is
-installed Polly can often calculate such schedules fully automatically.</p>
+<p> We apply three different transformations on the SCoP in the main function.
+The jscop files describing these transformations are hand written. If PoCC is
+installed Polly can sometimes calculate such schedules fully automatically.
+Hwever, this is still an area we are actively working on.</p>
 
+<h5>No Polly</h5>
 
-<pre class="code">opt -basicaa -polly-import-jscop -polly-print -disable-output matmul.preopt.ll -polly-import-jscop-postfix=.opt</pre>
-<pre>Cannot open file: ./init_array___%for.cond---%for.end28.jscop.opt
-Skipping import.
-In function: 'init_array' SCoP: for.cond =&gt; for.end28:
-for (c2=0;c2&lt;=1023;c2++) {
-  for (c4=0;c4&lt;=1023;c4++) {
-   &nbsp;%for.body4(c2,c4);
+<p>As a baseline we do not call any Polly code generation, but only apply the
+normal -O3 optimizations.</p>
+
+<pre class="code">
+opt matmul.preopt.ll -basicaa \
+    -polly-import-jscop \
+    -polly-cloog -analyze
+</pre>
+<pre>
+[...]
+main():
+for (c2=0;c2&ltg;=1535;c2++) {
+  for (c4=0;c4&ltg;=1535;c4++) {
+    Stmt_4(c2,c4);
+    for (c6=0;c6&ltg;=1535;c6++) {
+      Stmt_6(c2,c4,c6);
+    }
   }
 }
-Reading SCoP 'for.cond =&gt; for.end48' in function 'main' from './main___%for.cond---%for.end48.scop.opt.opt'.
-In function: 'main' SCoP: for.cond =&gt; for.end48:
-for (c2=0;c2&lt;=1023;c2++) {
-  for (c4=0;c4&lt;=1023;c4++) {
-   &nbsp;%for.body4(c2,c4);
+[...]
+</pre>
+<h5>Interchange (and Fission to allow the interchange)</h5>
+<p>We split the loops and can now apply an interchange of the loop dimensions that
+enumerate Stmt_6.</p>
+<pre class="code">
+opt matmul.preopt.ll -basicaa \
+    -polly-import-jscop -polly-import-jscop-postfix=interchanged \
+    -polly-cloog -analyze
+</pre>
+<pre>
+[...]
+Reading JScop '%1 =&gt; %17' in function 'main' from './main___%1---%17.jscop.interchanged'.
+[...]
+main():
+for (c2=0;c2&lt;=1535;c2++) {
+  for (c4=0;c4&lt;=1535;c4++) {
+    Stmt_4(c2,c4);
   }
 }
-for (c2=0;c2&lt;=1023;c2++) {
-  for (c3=0;c3&lt;=1023;c3++) {
-    for (c4=0;c4&lt;=1023;c4++) {
-     &nbsp;%for.body12(c2,c4,c3);
+for (c2=0;c2&lt;=1535;c2++) {
+  for (c4=0;c4&lt;=1535;c4++) {
+    for (c6=0;c6&lt;=1535;c6++) {
+      Stmt_6(c2,c6,c4);
     }
   }
 }
-</pre></li>
+[...]
+</pre>
+<h5>Interchange + Tiling</h5>
+<p>In addition to the interchange we tile now the second loop nest.</p>
+
+<pre class="code">
+opt matmul.preopt.ll -basicaa \
+    -polly-import-jscop -polly-import-jscop-postfix=interchanged+tiled \
+    -polly-cloog -analyze
+</pre>
+<pre>
+[...]
+Reading JScop '%1 =&gt; %17' in function 'main' from './main___%1---%17.jscop.interchanged+tiled'.
+[...]
+main():
+for (c2=0;c2&lt;=1535;c2++) {
+  for (c4=0;c4&lt;=1535;c4++) {
+    Stmt_4(c2,c4);
+  }
+}
+for (c2=0;c2&lt;=1535;c2+=64) {
+  for (c3=0;c3&lt;=1535;c3+=64) {
+    for (c4=0;c4&lt;=1535;c4+=64) {
+      for (c5=c2;c5&lt;=c2+63;c5++) {
+        for (c6=c4;c6&lt;=c4+63;c6++) {
+          for (c7=c3;c7&lt;=c3+63;c7++) {
+            Stmt_6(c5,c7,c6);
+          }
+        }
+      }
+    }
+  }
+}
+[...]
+</pre>
+<h5>Interchange + Tiling + Strip-mining to prepare vectorization</h5>
+To later allow vectorization we create a so called trivially parallelizable
+loop. It is innermost, parallel and has only four iterations. It can be
+replaced by 4-element SIMD instructions.
+<pre class="code">
+opt matmul.preopt.ll -basicaa \
+    -polly-import-jscop -polly-import-jscop-postfix=interchanged+tiled+vector \
+    -polly-cloog -analyze </pre>
+
+<pre>
+[...]
+Reading JScop '%1 =&gt; %17' in function 'main' from './main___%1---%17.jscop.interchanged+tiled+vector'.
+[...]
+main():
+for (c2=0;c2&lt;=1535;c2++) {
+  for (c4=0;c4&lt;=1535;c4++) {
+    Stmt_4(c2,c4);
+  }
+}
+for (c2=0;c2&lt;=1535;c2+=64) {
+  for (c3=0;c3&lt;=1535;c3+=64) {
+    for (c4=0;c4&lt;=1535;c4+=64) {
+      for (c5=c2;c5&lt;=c2+63;c5++) {
+        for (c6=c4;c6&lt;=c4+63;c6++) {
+          for (c7=c3;c7&lt;=c3+63;c7+=4) {
+            for (c8=c7;c8&lt;=c7+3;c8++) {
+              Stmt_6(c5,c8,c6);
+            }
+          }
+        }
+      }
+    }
+  }
+}
+[...]
+</pre>
+
+</li>
 
 <li><h4>Codegenerate the SCoPs</h4>
+<p>
 This generates new code for the SCoPs detected by polly.
 If -polly-import is present, transformations specified in the imported openscop
-files will be applied.
-<pre class="code">opt -basicaa -polly-import -polly-import-postfix=.opt -polly-codegen matmul.preopt.ll | opt -O3 &gt; matmul.pollyopt.ll</pre>
+files will be applied.</p>
+<pre class="code">opt matmul.preopt.ll | opt -O3 &gt; matmul.normalopt.ll</pre>
+<pre class="code">
+opt -basicaa \
+    -polly-import-jscop -polly-import-jscop-postfix=interchanged \
+    -polly-codegen matmul.preopt.ll \
+   | opt -O3 &gt; matmul.polly.interchanged.ll</pre>
 <pre>
-Cannot open file: ./init_array___%for.cond---%for.end28.scop.opt
-Skipping import.
-Reading SCoP 'for.cond =&gt; for.end48' in function 'main' from './main___%for.cond---%for.end48.scop.opt'.</pre>
-
-<pre class="code">opt matmul.preopt.ll | opt -O3 &gt; matmul.normalopt.ll</pre></li>
+Reading JScop '%1 =&gt; %19' in function 'init_array' from
+    './init_array___%1---%19.jscop.interchanged'.
+File could not be read: No such file or directory
+Reading JScop '%1 =&gt; %17' in function 'main' from
+    './main___%1---%17.jscop.interchanged'.
+</pre>
+<pre class="code">
+opt -basicaa \
+    -polly-import-jscop -polly-import-jscop-postfix=interchanged+tiled \
+    -polly-codegen matmul.preopt.ll \
+   | opt -O3 &gt; matmul.polly.interchanged+tiled.ll</pre>
+<pre>
+Reading JScop '%1 =&gt; %19' in function 'init_array' from
+    './init_array___%1---%19.jscop.interchanged+tiled'.
+File could not be read: No such file or directory
+Reading JScop '%1 =&gt; %17' in function 'main' from
+    './main___%1---%17.jscop.interchanged+tiled'.
+</pre>
+<pre class="code">
+opt -basicaa \
+    -polly-import-jscop -polly-import-jscop-postfix=interchanged+tiled+vector \
+    -polly-codegen -enable-polly-vector matmul.preopt.ll \
+   | opt -O3 &gt; matmul.polly.interchanged+tiled+vector.ll</pre>
+<pre>
+Reading JScop '%1 =&gt; %19' in function 'init_array' from
+    './init_array___%1---%19.jscop.interchanged+tiled+vector'.
+File could not be read: No such file or directory
+Reading JScop '%1 =&gt; %17' in function 'main' from
+    './main___%1---%17.jscop.interchanged+tiled+vector'.
+</pre>
+<pre class="code">
+opt -basicaa \
+    -polly-import-jscop -polly-import-jscop-postfix=interchanged+tiled+vector \
+    -polly-codegen -enable-polly-vector -enable-polly-openmp matmul.preopt.ll \
+  | opt -O3 &gt; matmul.polly.interchanged+tiled+openmp.ll</pre>
+<pre>
+Reading JScop '%1 =&gt; %19' in function 'init_array' from
+    './init_array___%1---%19.jscop.interchanged+tiled+vector'.
+File could not be read: No such file or directory
+Reading JScop '%1 =&gt; %17' in function 'main' from
+    './main___%1---%17.jscop.interchanged+tiled+vector'.
+</pre>
 
 <li><h4>Create the executables</h4>
 
@@ -290,8 +433,7 @@
 llc matmul.polly.interchanged+tiled+vector.ll -o matmul.polly.interchanged+tiled+vector.s &amp;&amp; \
     gcc matmul.polly.interchanged+tiled+vector.s -o matmul.polly.interchanged+tiled+vector.exe
 llc matmul.polly.interchanged+tiled+vector+openmp.ll -o matmul.polly.interchanged+tiled+vector+openmp.s &amp;&amp; \
-    gcc -lgomp matmul.polly.interchanged+tiled+vector+openmp.s -o matmul.polly.interchanged+tiled+vector+openmp.exe
-    </pre>
+    gcc -lgomp matmul.polly.interchanged+tiled+vector+openmp.s -o matmul.polly.interchanged+tiled+vector+openmp.exe </pre>
 
 <li><h4>Compare the runtime of the executables</h4>
 

Modified: polly/trunk/www/experiments/matmul/runall.sh
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/experiments/matmul/runall.sh?rev=130751&r1=130750&r2=130751&view=diff
==============================================================================
--- polly/trunk/www/experiments/matmul/runall.sh (original)
+++ polly/trunk/www/experiments/matmul/runall.sh Tue May  3 04:40:40 2011
@@ -1,11 +1,10 @@
 #!/bin/sh -a
 
-
 echo "--> 1. Create LLVM-IR from C"
 clang -S -emit-llvm matmul.c -o matmul.s
 
 echo "--> 2. Load Polly automatically when calling the 'opt' tool"
-export PATH_TO_POLLY_LIB="~/Projekte/polly/build_clang/lib/"
+export PATH_TO_POLLY_LIB="~/polly/build/lib/"
 alias opt="opt -load ${PATH_TO_POLLY_LIB}/LLVMPolly.so"
 
 echo "--> 3. Prepare the LLVM-IR for Polly"
@@ -40,10 +39,13 @@
 opt -basicaa -polly-export-jscop matmul.preopt.ll
 
 echo "--> 9. Import the updated jscop files and print the new SCoPs. (optional)"
+opt -basicaa -polly-import-jscop -polly-cloog -analyze matmul.preopt.ll
 opt -basicaa -polly-import-jscop -polly-cloog -analyze matmul.preopt.ll \
     -polly-import-jscop-postfix=interchanged
 opt -basicaa -polly-import-jscop -polly-cloog -analyze matmul.preopt.ll \
     -polly-import-jscop-postfix=interchanged+tiled
+opt -basicaa -polly-import-jscop -polly-cloog -analyze matmul.preopt.ll \
+    -polly-import-jscop-postfix=interchanged+tiled+vector
 
 echo "--> 10. Codegenerate the SCoPs"
 opt -basicaa -polly-import-jscop -polly-import-jscop-postfix=interchanged \

Modified: polly/trunk/www/menu.css
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/menu.css?rev=130751&r1=130750&r2=130751&view=diff
==============================================================================
--- polly/trunk/www/menu.css (original)
+++ polly/trunk/www/menu.css Tue May  3 04:40:40 2011
@@ -11,6 +11,7 @@
 	position:absolute;
   left:29ex;
 	padding-right:4ex;
+        max-width: 50em;
 }
 
 /**************/

Modified: polly/trunk/www/menu.html.incl
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/www/menu.html.incl?rev=130751&r1=130750&r2=130751&view=diff
==============================================================================
--- polly/trunk/www/menu.html.incl (original)
+++ polly/trunk/www/menu.html.incl Tue May  3 04:40:40 2011
@@ -8,7 +8,7 @@
     <a href="index.html">About</a>
     <a href="todo.html">Todo</a>
     <a href="passes.html">LLVM Passes</a>
-<!--    <a href="examples.html">Examples</a> -->
+    <a href="examples.html">Examples</a>
     <a href="performance.html">Performance</a>
     <a href="publications.html">Publications</a>
     <a href="contributors.html">Contributors</a>



From etherzhhb at gmail.com  Tue May  3 08:46:59 2011
From: etherzhhb at gmail.com (Hongbin Zheng)
Date: Tue, 03 May 2011 13:46:59 -0000
Subject: [llvm-commits] [polly] r130752 - in /polly/trunk:
 include/polly/Dependences.h lib/Analysis/Dependences.cpp
 lib/CodeGeneration.cpp
Message-ID: <20110503134659.0FEF62A6C12C@llvm.org>

Author: ether
Date: Tue May  3 08:46:58 2011
New Revision: 130752

URL: http://llvm.org/viewvc/llvm-project?rev=130752&view=rev
Log:
Refactor: Move 'isParallelFor' from codegen backend to Dependences analysis, so other passes can also use it.

Modified:
    polly/trunk/include/polly/Dependences.h
    polly/trunk/lib/Analysis/Dependences.cpp
    polly/trunk/lib/CodeGeneration.cpp

Modified: polly/trunk/include/polly/Dependences.h
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/include/polly/Dependences.h?rev=130752&r1=130751&r2=130752&view=diff
==============================================================================
--- polly/trunk/include/polly/Dependences.h (original)
+++ polly/trunk/include/polly/Dependences.h Tue May  3 08:46:58 2011
@@ -31,6 +31,7 @@
 struct isl_union_set;
 struct isl_map;
 struct isl_set;
+struct clast_for;
 
 using namespace llvm;
 
@@ -69,6 +70,16 @@
     ///              valid for the scattering domain subset given.
     bool isParallelDimension(isl_set *loopDomain, unsigned parallelDimension);
 
+    /// @brief Check if a loop is parallel
+    ///
+    /// Detect if a clast_for loop can be executed in parallel.
+    ///
+    /// @param f The clast for loop to check.
+    ///
+    /// @return bool Returns true if the incoming clast_for statement can
+    ///              execute in parallel.
+    bool isParallelFor(const clast_for *f);
+
     bool runOnScop(Scop &S);
     void printScop(raw_ostream &OS) const;
     virtual void releaseMemory();

Modified: polly/trunk/lib/Analysis/Dependences.cpp
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/lib/Analysis/Dependences.cpp?rev=130752&r1=130751&r2=130752&view=diff
==============================================================================
--- polly/trunk/lib/Analysis/Dependences.cpp (original)
+++ polly/trunk/lib/Analysis/Dependences.cpp Tue May  3 08:46:58 2011
@@ -31,8 +31,9 @@
 #include "llvm/Support/CommandLine.h"
 
 #include <isl/flow.h>
-#include <isl/map.h>
-#include <isl/constraint.h>
+#define CLOOG_INT_GMP 1
+#include <cloog/cloog.h>
+#include <cloog/isl/cloog.h>
 
 using namespace polly;
 using namespace llvm;
@@ -351,6 +352,13 @@
     && isl_union_set_is_empty(nonValid_waw);
 }
 
+bool Dependences::isParallelFor(const clast_for *f) {
+  isl_set *loopDomain = isl_set_from_cloog_domain(f->domain);
+  assert(loopDomain && "Cannot access domain of loop");
+
+  return isParallelDimension(loopDomain, isl_set_n_dim(loopDomain));
+}
+
 void Dependences::printScop(raw_ostream &OS) const {
   OS.indent(4) << "Must dependences:\n";
   OS.indent(8) << stringFromIslObj(must_dep) << "\n";

Modified: polly/trunk/lib/CodeGeneration.cpp
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/lib/CodeGeneration.cpp?rev=130752&r1=130751&r2=130752&view=diff
==============================================================================
--- polly/trunk/lib/CodeGeneration.cpp (original)
+++ polly/trunk/lib/CodeGeneration.cpp Tue May  3 08:46:58 2011
@@ -826,25 +826,6 @@
     Builder.SetInsertPoint(AfterBB);
   }
 
-  /// @brief Check if a loop is parallel
-  ///
-  /// Detect if a clast_for loop can be executed in parallel.
-  ///
-  /// @param f The clast for loop to check.
-  bool isParallelFor(const clast_for *f) {
-    isl_set *loopDomain = isl_set_from_cloog_domain(f->domain);
-    assert(loopDomain && "Cannot access domain of loop");
-
-    bool isParallel = DP->isParallelDimension(loopDomain,
-                                              isl_set_n_dim(loopDomain));
-
-    if (isParallel)
-      DEBUG(dbgs() << "Parallel loop with induction variable '" << f->iterator
-            << "' found\n";);
-
-    return isParallel;
-  }
-
   /// @brief Add a new definition of an openmp subfunction.
   Function* addOpenMPSubfunction(Module *M) {
     Function *F = Builder.GetInsertBlock()->getParent();
@@ -1157,11 +1138,11 @@
   }
 
   void codegen(const clast_for *f) {
-    if (Vector && isInnermostLoop(f) && isParallelFor(f)
+    if (Vector && isInnermostLoop(f) && DP->isParallelFor(f)
         && (-1 != getNumberOfIterations(f))
         && (getNumberOfIterations(f) <= 16)) {
       codegenForVector(f);
-    } else if (OpenMP && !parallelCodeGeneration && isParallelFor(f)) {
+    } else if (OpenMP && !parallelCodeGeneration && DP->isParallelFor(f)) {
       parallelCodeGeneration = true;
       parallelLoops.push_back(f->iterator);
       codegenForOpenMP(f);



From rafael.espindola at gmail.com  Tue May  3 10:26:56 2011
From: rafael.espindola at gmail.com (Rafael Avila de Espindola)
Date: Tue, 03 May 2011 11:26:56 -0400
Subject: [llvm-commits] What requires __debug_frame?
Message-ID: <4DC01EC0.7080101@gmail.com>

I was wondering what software requires __debug_frame. I decided to try 
changing the default and see what would break. Surprisingly, it was just 
a a single test that I needed to update to expect cfi directives.

I used a clang modified with the attached patch to build a debug version 
of clang itself and gdb is happy with it. I can set breakpoints, get 
backtraces, etc.

So, is __debug_frame just an historical thing right now?

Thanks,
Rafael
-------------- next part --------------
A non-text attachment was scrubbed...
Name: debug_frame.patch
Type: text/x-patch
Size: 1427 bytes
Desc: not available
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110503/dd8cf7eb/attachment.bin 

From rafael.espindola at gmail.com  Tue May  3 10:56:16 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Tue, 03 May 2011 15:56:16 -0000
Subject: [llvm-commits] [llvm] r130754 - in /llvm/trunk/test/FrontendC++:
 2006-11-06-StackTrace.cpp 2009-04-21-DtorNames-dbg.cpp
 2010-08-31-ByValArg.cpp
Message-ID: <20110503155616.AFB9A2A6C12C@llvm.org>

Author: rafael
Date: Tue May  3 10:56:16 2011
New Revision: 130754

URL: http://llvm.org/viewvc/llvm-project?rev=130754&view=rev
Log:
Pass --disable-cfi to llc.

Modified:
    llvm/trunk/test/FrontendC++/2006-11-06-StackTrace.cpp
    llvm/trunk/test/FrontendC++/2009-04-21-DtorNames-dbg.cpp
    llvm/trunk/test/FrontendC++/2010-08-31-ByValArg.cpp

Modified: llvm/trunk/test/FrontendC++/2006-11-06-StackTrace.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC%2B%2B/2006-11-06-StackTrace.cpp?rev=130754&r1=130753&r2=130754&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC++/2006-11-06-StackTrace.cpp (original)
+++ llvm/trunk/test/FrontendC++/2006-11-06-StackTrace.cpp Tue May  3 10:56:16 2011
@@ -1,7 +1,7 @@
 // This is a regression test on debug info to make sure that we can get a
 // meaningful stack trace from a C++ program.
 // RUN: %llvmgcc -S -O0 -g %s -o - | \
-// RUN:    llc --disable-fp-elim -o %t.s -O0 -relocation-model=pic
+// RUN:    llc --disable-cfi --disable-fp-elim -o %t.s -O0 -relocation-model=pic
 // RUN: %compile_c %t.s -o %t.o
 // RUN: %link %t.o -o %t.exe
 // RUN: echo {break DeepStack::deepest\nrun 17\nwhere\n} > %t.in 

Modified: llvm/trunk/test/FrontendC++/2009-04-21-DtorNames-dbg.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC%2B%2B/2009-04-21-DtorNames-dbg.cpp?rev=130754&r1=130753&r2=130754&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC++/2009-04-21-DtorNames-dbg.cpp (original)
+++ llvm/trunk/test/FrontendC++/2009-04-21-DtorNames-dbg.cpp Tue May  3 10:56:16 2011
@@ -1,4 +1,4 @@
-// RUN: %llvmgcc -S -g %s -o - | llc -O0 -o %t.s
+// RUN: %llvmgcc -S -g %s -o - | llc --disable-cfi -O0 -o %t.s
 // RUN: %compile_c %t.s -o %t.o
 // PR4025
 

Modified: llvm/trunk/test/FrontendC++/2010-08-31-ByValArg.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC%2B%2B/2010-08-31-ByValArg.cpp?rev=130754&r1=130753&r2=130754&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC++/2010-08-31-ByValArg.cpp (original)
+++ llvm/trunk/test/FrontendC++/2010-08-31-ByValArg.cpp Tue May  3 10:56:16 2011
@@ -1,7 +1,7 @@
 // This regression test checks byval arguments' debug info.
 // Radar 8367011
 // RUN: %llvmgcc -S -O0 -g %s -o - | \
-// RUN:    llc --disable-fp-elim -o %t.s -O0 -relocation-model=pic
+// RUN:    llc --disable-cfi --disable-fp-elim -o %t.s -O0 -relocation-model=pic
 // RUN: %compile_c %t.s -o %t.o
 // RUN: %link %t.o -o %t.exe
 // RUN: echo {break get\nrun\np missing_arg.b} > %t.in 



From benny.kra at googlemail.com  Tue May  3 11:00:27 2011
From: benny.kra at googlemail.com (Benjamin Kramer)
Date: Tue, 03 May 2011 16:00:27 -0000
Subject: [llvm-commits] [llvm] r130755 - in /llvm/trunk/lib:
 AsmParser/LLParser.cpp Target/Sparc/SparcISelLowering.cpp
 Transforms/Instrumentation/PathProfiling.cpp
Message-ID: <20110503160027.89F502A6C12C@llvm.org>

Author: d0k
Date: Tue May  3 11:00:27 2011
New Revision: 130755

URL: http://llvm.org/viewvc/llvm-project?rev=130755&view=rev
Log:
Remove unused variables caught by GCC's -Wunused-but-set-variable.

Modified:
    llvm/trunk/lib/AsmParser/LLParser.cpp
    llvm/trunk/lib/Target/Sparc/SparcISelLowering.cpp
    llvm/trunk/lib/Transforms/Instrumentation/PathProfiling.cpp

Modified: llvm/trunk/lib/AsmParser/LLParser.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/AsmParser/LLParser.cpp?rev=130755&r1=130754&r2=130755&view=diff
==============================================================================
--- llvm/trunk/lib/AsmParser/LLParser.cpp (original)
+++ llvm/trunk/lib/AsmParser/LLParser.cpp Tue May  3 11:00:27 2011
@@ -3003,7 +3003,6 @@
   case lltok::kw_sub:
   case lltok::kw_mul:
   case lltok::kw_shl: {
-    LocTy ModifierLoc = Lex.getLoc();
     bool NUW = EatIfPresent(lltok::kw_nuw);
     bool NSW = EatIfPresent(lltok::kw_nsw);
     if (!NUW) NUW = EatIfPresent(lltok::kw_nuw);

Modified: llvm/trunk/lib/Target/Sparc/SparcISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Sparc/SparcISelLowering.cpp?rev=130755&r1=130754&r2=130755&view=diff
==============================================================================
--- llvm/trunk/lib/Target/Sparc/SparcISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/Sparc/SparcISelLowering.cpp Tue May  3 11:00:27 2011
@@ -182,8 +182,6 @@
     }
 
     if (VA.isRegLoc()) {
-      EVT RegVT = VA.getLocVT();
-
       if (VA.needsCustom()) {
         assert(VA.getLocVT() == MVT::f64);
         unsigned VRegHi = RegInfo.createVirtualRegister(&SP::IntRegsRegClass);

Modified: llvm/trunk/lib/Transforms/Instrumentation/PathProfiling.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Instrumentation/PathProfiling.cpp?rev=130755&r1=130754&r2=130755&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Instrumentation/PathProfiling.cpp (original)
+++ llvm/trunk/lib/Transforms/Instrumentation/PathProfiling.cpp Tue May  3 11:00:27 2011
@@ -1351,8 +1351,6 @@
     return false;
   }
 
-  BasicBlock::iterator insertPoint = Main->getEntryBlock().getFirstNonPHI();
-
   llvmIncrementHashFunction = M.getOrInsertFunction(
     "llvm_increment_path_count",
     Type::getVoidTy(*Context), // return type



From dpatel at apple.com  Tue May  3 11:12:10 2011
From: dpatel at apple.com (Devang Patel)
Date: Tue, 03 May 2011 09:12:10 -0700
Subject: [llvm-commits] What requires __debug_frame?
In-Reply-To: <4DC01EC0.7080101@gmail.com>
References: <4DC01EC0.7080101@gmail.com>
Message-ID: <84C9465D-9B69-429D-8631-FB15463CC930@apple.com>


On May 3, 2011, at 8:26 AM, Rafael Avila de Espindola wrote:

> I was wondering what software requires __debug_frame. I decided to try changing the default and see what would break. Surprisingly, it was just a a single test that I needed to update to expect cfi directives.
> 
> I used a clang modified with the attached patch to build a debug version of clang itself and gdb is happy with it. I can set breakpoints, get backtraces, etc.

It is used by unwinder, so you want to test eh unwinding also.
-
Devang
> 
> So, is __debug_frame just an historical thing right now?
> 
> Thanks,
> Rafael
> <debug_frame.patch>


From dpatel at apple.com  Tue May  3 11:18:28 2011
From: dpatel at apple.com (Devang Patel)
Date: Tue, 03 May 2011 16:18:28 -0000
Subject: [llvm-commits] [llvm] r130756 - in /llvm/trunk:
 docs/SourceLevelDebugging.html lib/Analysis/DIBuilder.cpp
Message-ID: <20110503161828.4AC562A6C12E@llvm.org>

Author: dpatel
Date: Tue May  3 11:18:28 2011
New Revision: 130756

URL: http://llvm.org/viewvc/llvm-project?rev=130756&view=rev
Log:
Use llvm.dbg.cu  named metadata to collect compile units.

Modified:
    llvm/trunk/docs/SourceLevelDebugging.html
    llvm/trunk/lib/Analysis/DIBuilder.cpp

Modified: llvm/trunk/docs/SourceLevelDebugging.html
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/SourceLevelDebugging.html?rev=130756&r1=130755&r2=130756&view=diff
==============================================================================
--- llvm/trunk/docs/SourceLevelDebugging.html (original)
+++ llvm/trunk/docs/SourceLevelDebugging.html Tue May  3 11:18:28 2011
@@ -342,7 +342,9 @@
    that produced it.</p>
 
 <p>Compile unit descriptors provide the root context for objects declared in a
-   specific compilation unit. File descriptors are defined using this context.</p>
+   specific compilation unit. File descriptors are defined using this context.
+   These descriptors are collected by a named metadata 
+   <tt>!llvm.dbg.cu</tt>.
 
 </div>
 

Modified: llvm/trunk/lib/Analysis/DIBuilder.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/DIBuilder.cpp?rev=130756&r1=130755&r2=130756&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/DIBuilder.cpp (original)
+++ llvm/trunk/lib/Analysis/DIBuilder.cpp Tue May  3 11:18:28 2011
@@ -51,6 +51,10 @@
     ConstantInt::get(Type::getInt32Ty(VMContext), RunTimeVer)
   };
   TheCU = DICompileUnit(MDNode::get(VMContext, Elts));
+
+  // Create a named metadata so that it is easier to find cu in a module.
+  NamedMDNode *NMD = M.getOrInsertNamedMetadata("llvm.dbg.cu");
+  NMD->addOperand(TheCU);
 }
 
 /// createFile - Create a file descriptor to hold debugging information



From dpatel at apple.com  Tue May  3 11:45:22 2011
From: dpatel at apple.com (Devang Patel)
Date: Tue, 03 May 2011 16:45:22 -0000
Subject: [llvm-commits] [llvm] r130759 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
Message-ID: <20110503164522.EB4642A6C12C@llvm.org>

Author: dpatel
Date: Tue May  3 11:45:22 2011
New Revision: 130759

URL: http://llvm.org/viewvc/llvm-project?rev=130759&view=rev
Log:
If the front end has emitted llvm.dbg.cu and other debug info anchors (clang does it now) then use them directly. This saves one scan of entire module, to collect debug info, which in turns saves few machine cycles at compile time.

Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130759&r1=130758&r2=130759&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Tue May  3 11:45:22 2011
@@ -1113,44 +1113,66 @@
   if (DisableDebugInfoPrinting)
     return;
 
-  DebugInfoFinder DbgFinder;
-  DbgFinder.processModule(*M);
-
   bool HasDebugInfo = false;
 
-  // Scan all the compile-units to see if there are any marked as the main unit.
-  // if not, we do not generate debug info.
-  for (DebugInfoFinder::iterator I = DbgFinder.compile_unit_begin(),
-       E = DbgFinder.compile_unit_end(); I != E; ++I) {
-    if (DICompileUnit(*I).isMain()) {
+  // If module has named metadata anchors then use them, otherwise scan the module
+  // using debug info finder to collect debug info.
+  NamedMDNode *CU_Nodes = M->getNamedMetadata("llvm.dbg.cu");
+  if (CU_Nodes) {
+    for (unsigned i = 0, e = CU_Nodes->getNumOperands(); i != e; ++i) {
       HasDebugInfo = true;
-      break;
+      constructCompileUnit(CU_Nodes->getOperand(i));
     }
-  }
+    if (!HasDebugInfo)
+      return;
 
-  if (!HasDebugInfo) return;
+    if (NamedMDNode *NMD = M->getNamedMetadata("llvm.dbg.gv"))
+      for (unsigned i = 0, e = NMD->getNumOperands(); i != e; ++i)
+        constructGlobalVariableDIE(NMD->getOperand(i));
+
+    if (NamedMDNode *NMD = M->getNamedMetadata("llvm.dbg.sp"))
+      for (unsigned i = 0, e = NMD->getNumOperands(); i != e; ++i)
+        constructSubprogramDIE(NMD->getOperand(i));
+    
+  } else {
 
+    DebugInfoFinder DbgFinder;
+    DbgFinder.processModule(*M);
+    
+    // Scan all the compile-units to see if there are any marked as the main unit.
+    // if not, we do not generate debug info.
+    for (DebugInfoFinder::iterator I = DbgFinder.compile_unit_begin(),
+           E = DbgFinder.compile_unit_end(); I != E; ++I) {
+      if (DICompileUnit(*I).isMain()) {
+        HasDebugInfo = true;
+        break;
+      }
+    }
+    
+    // Create all the compile unit DIEs.
+    for (DebugInfoFinder::iterator I = DbgFinder.compile_unit_begin(),
+           E = DbgFinder.compile_unit_end(); I != E; ++I)
+      constructCompileUnit(*I);
+    
+    // Create DIEs for each global variable.
+    for (DebugInfoFinder::iterator I = DbgFinder.global_variable_begin(),
+           E = DbgFinder.global_variable_end(); I != E; ++I)
+      constructGlobalVariableDIE(*I);
+    
+    // Create DIEs for each subprogram.
+    for (DebugInfoFinder::iterator I = DbgFinder.subprogram_begin(),
+           E = DbgFinder.subprogram_end(); I != E; ++I)
+      constructSubprogramDIE(*I);
+  }
+  
+  if (!HasDebugInfo) return;
+  
   // Tell MMI that we have debug info.
   MMI->setDebugInfoAvailability(true);
-
+  
   // Emit initial sections.
   EmitSectionLabels();
 
-  // Create all the compile unit DIEs.
-  for (DebugInfoFinder::iterator I = DbgFinder.compile_unit_begin(),
-         E = DbgFinder.compile_unit_end(); I != E; ++I)
-    constructCompileUnit(*I);
-
-  // Create DIEs for each global variable.
-  for (DebugInfoFinder::iterator I = DbgFinder.global_variable_begin(),
-         E = DbgFinder.global_variable_end(); I != E; ++I)
-    constructGlobalVariableDIE(*I);
-
-  // Create DIEs for each subprogram.
-  for (DebugInfoFinder::iterator I = DbgFinder.subprogram_begin(),
-         E = DbgFinder.subprogram_end(); I != E; ++I)
-    constructSubprogramDIE(*I);
-
   //getOrCreateTypeDIE
   if (NamedMDNode *NMD = M->getNamedMetadata("llvm.dbg.enum"))
     for (unsigned i = 0, e = NMD->getNumOperands(); i != e; ++i) {



From bob.wilson at apple.com  Tue May  3 12:22:56 2011
From: bob.wilson at apple.com (Bob Wilson)
Date: Tue, 03 May 2011 17:22:56 -0000
Subject: [llvm-commits] [llvm] r130761 -
	/llvm/trunk/include/llvm/ADT/Triple.h
Message-ID: <20110503172256.638DB2A6C12C@llvm.org>

Author: bwilson
Date: Tue May  3 12:22:56 2011
New Revision: 130761

URL: http://llvm.org/viewvc/llvm-project?rev=130761&view=rev
Log:
Fix a comment.

Modified:
    llvm/trunk/include/llvm/ADT/Triple.h

Modified: llvm/trunk/include/llvm/ADT/Triple.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/ADT/Triple.h?rev=130761&r1=130760&r2=130761&view=diff
==============================================================================
--- llvm/trunk/include/llvm/ADT/Triple.h (original)
+++ llvm/trunk/include/llvm/ADT/Triple.h Tue May  3 12:22:56 2011
@@ -225,7 +225,7 @@
   /// if the environment component is present).
   StringRef getOSAndEnvironmentName() const;
 
-  /// getOSNumber - Parse the version number from the OS name component of the
+  /// getOSVersion - Parse the version number from the OS name component of the
   /// triple, if present.
   ///
   /// For example, "fooos1.2.3" would return (1, 2, 3).



From bruno.cardoso at gmail.com  Tue May  3 12:29:22 2011
From: bruno.cardoso at gmail.com (Bruno Cardoso Lopes)
Date: Tue, 03 May 2011 17:29:22 -0000
Subject: [llvm-commits] [llvm] r130763 - in /llvm/trunk:
 include/llvm/IntrinsicsARM.td lib/Target/ARM/ARMInstrFormats.td
 lib/Target/ARM/ARMInstrInfo.td lib/Target/ARM/ARMInstrThumb.td
 lib/Target/ARM/ARMInstrThumb2.td test/CodeGen/ARM/intrinsics.ll
Message-ID: <20110503172922.A8E992A6C12C@llvm.org>

Author: bruno
Date: Tue May  3 12:29:22 2011
New Revision: 130763

URL: http://llvm.org/viewvc/llvm-project?rev=130763&view=rev
Log:
Add a few ARM coprocessor intrinsics. Testcases included

Added:
    llvm/trunk/test/CodeGen/ARM/intrinsics.ll
Modified:
    llvm/trunk/include/llvm/IntrinsicsARM.td
    llvm/trunk/lib/Target/ARM/ARMInstrFormats.td
    llvm/trunk/lib/Target/ARM/ARMInstrInfo.td
    llvm/trunk/lib/Target/ARM/ARMInstrThumb.td
    llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td

Modified: llvm/trunk/include/llvm/IntrinsicsARM.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/IntrinsicsARM.td?rev=130763&r1=130762&r2=130763&view=diff
==============================================================================
--- llvm/trunk/include/llvm/IntrinsicsARM.td (original)
+++ llvm/trunk/include/llvm/IntrinsicsARM.td Tue May  3 12:29:22 2011
@@ -50,6 +50,43 @@
 }
 
 //===----------------------------------------------------------------------===//
+// Coprocessor
+
+let TargetPrefix = "arm" in {  // All intrinsics start with "llvm.arm.".
+  // Move to coprocessor
+  def int_arm_mcr : GCCBuiltin<"__builtin_arm_mcr">,
+     Intrinsic<[], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
+                    llvm_i32_ty, llvm_i32_ty, llvm_i32_ty], []>;
+  def int_arm_mcr2 : GCCBuiltin<"__builtin_arm_mcr2">,
+     Intrinsic<[], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
+                    llvm_i32_ty, llvm_i32_ty, llvm_i32_ty], []>;
+
+  // Move from coprocessor
+  def int_arm_mrc : GCCBuiltin<"__builtin_arm_mrc">,
+     Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
+                               llvm_i32_ty, llvm_i32_ty], []>;
+  def int_arm_mrc2 : GCCBuiltin<"__builtin_arm_mrc2">,
+     Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
+                               llvm_i32_ty, llvm_i32_ty], []>;
+
+  // Coprocessor data processing
+  def int_arm_cdp : GCCBuiltin<"__builtin_arm_cdp">,
+     Intrinsic<[], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
+                    llvm_i32_ty, llvm_i32_ty, llvm_i32_ty], []>;
+  def int_arm_cdp2 : GCCBuiltin<"__builtin_arm_cdp2">,
+     Intrinsic<[], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
+                    llvm_i32_ty, llvm_i32_ty, llvm_i32_ty], []>;
+
+  // Move from two registers to coprocessor
+  def int_arm_mcrr : GCCBuiltin<"__builtin_arm_mcrr">,
+     Intrinsic<[], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
+                    llvm_i32_ty, llvm_i32_ty], []>;
+  def int_arm_mcrr2 : GCCBuiltin<"__builtin_arm_mcrr2">,
+     Intrinsic<[], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty,
+                    llvm_i32_ty, llvm_i32_ty], []>;
+}
+
+//===----------------------------------------------------------------------===//
 // Advanced SIMD (NEON)
 
 let TargetPrefix = "arm" in {  // All intrinsics start with "llvm.arm.".

Modified: llvm/trunk/lib/Target/ARM/ARMInstrFormats.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrFormats.td?rev=130763&r1=130762&r2=130763&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrFormats.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrFormats.td Tue May  3 12:29:22 2011
@@ -860,6 +860,9 @@
 class ARMPat<dag pattern, dag result> : Pat<pattern, result> {
   list<Predicate> Predicates = [IsARM];
 }
+class ARMV5TPat<dag pattern, dag result> : Pat<pattern, result> {
+  list<Predicate> Predicates = [IsARM, HasV5T];
+}
 class ARMV5TEPat<dag pattern, dag result> : Pat<pattern, result> {
   list<Predicate> Predicates = [IsARM, HasV5TE];
 }
@@ -1208,6 +1211,11 @@
   list<Predicate> Predicates = [IsThumb, IsThumb1Only];
 }
 
+// T2v6Pat - Same as Pat<>, but requires V6T2 Thumb2 mode.
+class T2v6Pat<dag pattern, dag result> : Pat<pattern, result> {
+  list<Predicate> Predicates = [IsThumb2, HasV6T2];
+}
+
 // T2Pat - Same as Pat<>, but requires that the compiler be in Thumb2 mode.
 class T2Pat<dag pattern, dag result> : Pat<pattern, result> {
   list<Predicate> Predicates = [IsThumb2];

Modified: llvm/trunk/lib/Target/ARM/ARMInstrInfo.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrInfo.td?rev=130763&r1=130762&r2=130763&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrInfo.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrInfo.td Tue May  3 12:29:22 2011
@@ -3415,6 +3415,10 @@
   let Inst{23-20} = opc1;
 }
 
+def : ARMPat<(int_arm_cdp imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                          imm:$CRm, imm:$opc2),
+             (CDP imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn, imm:$CRm,imm:$opc2)>;
+
 def CDP2 : ABXI<0b1110, (outs), (ins p_imm:$cop, i32imm:$opc1,
                c_imm:$CRd, c_imm:$CRn, c_imm:$CRm, i32imm:$opc2),
                NoItinerary, "cdp2\t$cop, $opc1, $CRd, $CRn, $CRm, $opc2",
@@ -3436,6 +3440,11 @@
   let Inst{23-20} = opc1;
 }
 
+def : ARMPat<(int_arm_cdp2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                          imm:$CRm, imm:$opc2),
+             (CDP2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn, imm:$CRm,
+                                                            imm:$opc2)>;
+
 class ACI<dag oops, dag iops, string opc, string asm,
           IndexMode im = IndexModeNone>
   : InoP<oops, iops, AddrModeNone, Size4Bytes, im, BrFrm, NoItinerary,
@@ -3572,6 +3581,12 @@
                     (outs GPR:$Rt), (ins p_imm:$cop, i32imm:$opc1,
                                          c_imm:$CRn, c_imm:$CRm, i32imm:$opc2)>;
 
+def : ARMPat<(int_arm_mcr imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                           imm:$CRm, imm:$opc2),
+             (MCR imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn, imm:$CRm, imm:$opc2)>;
+def : ARMPat<(int_arm_mrc imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2),
+             (MRC imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2)>;
+
 class MovRCopro2<string opc, bit direction, dag oops, dag iops>
   : ABXI<0b1110, oops, iops, NoItinerary,
          !strconcat(opc, "\t$cop, $opc1, $Rt, $CRn, $CRm, $opc2"),
@@ -3604,6 +3619,14 @@
                                            c_imm:$CRn, c_imm:$CRm,
                                            i32imm:$opc2)>;
 
+def : ARMV5TPat<(int_arm_mcr2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                              imm:$CRm, imm:$opc2),
+                (MCR2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                      imm:$CRm, imm:$opc2)>;
+def : ARMV5TPat<(int_arm_mrc2 imm:$cop, imm:$opc1, imm:$CRn,
+                              imm:$CRm, imm:$opc2),
+                (MRC2 imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2)>;
+
 class MovRRCopro<string opc, bit direction>
   : ABI<0b1100, (outs), (ins p_imm:$cop, i32imm:$opc1,
         GPR:$Rt, GPR:$Rt2, c_imm:$CRm),
@@ -3628,6 +3651,10 @@
 def MCRR : MovRRCopro<"mcrr", 0 /* from ARM core register to coprocessor */>;
 def MRRC : MovRRCopro<"mrrc", 1 /* from coprocessor to ARM core register */>;
 
+def : ARMV5TEPat<(int_arm_mcrr imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2,
+                               imm:$CRm),
+                 (MCRR imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm)>;
+
 class MovRRCopro2<string opc, bit direction>
   : ABXI<0b1100, (outs), (ins p_imm:$cop, i32imm:$opc1,
          GPR:$Rt, GPR:$Rt2, c_imm:$CRm),
@@ -3653,6 +3680,9 @@
 def MCRR2 : MovRRCopro2<"mcrr2", 0 /* from ARM core register to coprocessor */>;
 def MRRC2 : MovRRCopro2<"mrrc2", 1 /* from coprocessor to ARM core register */>;
 
+def : ARMV6Pat<(int_arm_mcrr2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm),
+               (MCRR2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm)>;
+
 //===----------------------------------------------------------------------===//
 // Move between special register and ARM core register -- for disassembly only
 //

Modified: llvm/trunk/lib/Target/ARM/ARMInstrThumb.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrThumb.td?rev=130763&r1=130762&r2=130763&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrThumb.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrThumb.td Tue May  3 12:29:22 2011
@@ -1381,6 +1381,14 @@
            (outs GPR:$Rt), (ins p_imm:$cop, i32imm:$opc1, c_imm:$CRn,
                                 c_imm:$CRm, i32imm:$opc2)>;
 
+def : Pat<(int_arm_mcr imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                          imm:$CRm, imm:$opc2),
+          (tMCR imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                imm:$CRm, imm:$opc2)>, Requires<[IsThumb, HasV6T2]>;
+def : Pat<(int_arm_mrc imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2),
+          (tMRC imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2)>,
+          Requires<[IsThumb, HasV6T2]>;
+
 class tMovRRCopro<string opc, bit direction>
   : T1Cop<(outs), (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, GPR:$Rt2, c_imm:$CRm),
           !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"),
@@ -1405,6 +1413,10 @@
 def tMCRR : tMovRRCopro<"mcrr", 0 /* from ARM core register to coprocessor */>;
 def tMRRC : tMovRRCopro<"mrrc", 1 /* from coprocessor to ARM core register */>;
 
+def : Pat<(int_arm_mcrr imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm),
+          (tMCRR imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm)>,
+          Requires<[IsThumb, HasV6T2]>;
+
 //===----------------------------------------------------------------------===//
 // Other Coprocessor Instructions.  For disassembly only.
 //
@@ -1430,6 +1442,11 @@
   let Inst{23-20} = opc1;
 }
 
+def : Pat<(int_arm_cdp imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                          imm:$CRm, imm:$opc2),
+          (tCDP imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                imm:$CRm, imm:$opc2)>, Requires<[IsThumb, HasV6T2]>;
+
 //===----------------------------------------------------------------------===//
 // TLS Instructions
 //

Modified: llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td?rev=130763&r1=130762&r2=130763&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td Tue May  3 12:29:22 2011
@@ -3376,6 +3376,14 @@
              (outs GPR:$Rt), (ins p_imm:$cop, i32imm:$opc1, c_imm:$CRn,
                                   c_imm:$CRm, i32imm:$opc2)>;
 
+def : T2v6Pat<(int_arm_mcr2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                            imm:$CRm, imm:$opc2),
+              (t2MCR2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                      imm:$CRm, imm:$opc2)>;
+def : T2v6Pat<(int_arm_mrc2 imm:$cop, imm:$opc1, imm:$CRn,
+                            imm:$CRm, imm:$opc2),
+              (t2MRC2 imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2)>;
+
 class t2MovRRCopro<string opc, bit direction>
   : T2Cop<(outs), (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, GPR:$Rt2, c_imm:$CRm),
           !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"),
@@ -3402,6 +3410,9 @@
 def t2MRRC2 : t2MovRRCopro<"mrrc2",
                            1 /* from coprocessor to ARM core register */>;
 
+def : T2v6Pat<(int_arm_mcrr2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm),
+              (t2MCRR2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm)>;
+
 //===----------------------------------------------------------------------===//
 // Other Coprocessor Instructions.  For disassembly only.
 //
@@ -3427,3 +3438,8 @@
   let Inst{19-16} = CRn;
   let Inst{23-20} = opc1;
 }
+
+def : T2v6Pat<(int_arm_cdp2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                            imm:$CRm, imm:$opc2),
+              (t2CDP2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                      imm:$CRm, imm:$opc2)>;

Added: llvm/trunk/test/CodeGen/ARM/intrinsics.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/intrinsics.ll?rev=130763&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/intrinsics.ll (added)
+++ llvm/trunk/test/CodeGen/ARM/intrinsics.ll Tue May  3 12:29:22 2011
@@ -0,0 +1,39 @@
+; RUN: llc < %s -mtriple=armv7-eabi -mcpu=cortex-a8 | FileCheck %s
+; RUN: llc < %s -march=thumb -mtriple=thumbv7-eabi -mcpu=cortex-a8 | FileCheck %s
+
+define void @coproc() nounwind {
+entry:
+  ; CHECK: mrc
+  %0 = tail call i32 @llvm.arm.mrc(i32 7, i32 1, i32 1, i32 1, i32 4) nounwind
+  ; CHECK: mcr
+  tail call void @llvm.arm.mcr(i32 7, i32 1, i32 %0, i32 1, i32 1, i32 4) nounwind
+  ; CHECK: mrc2
+  %1 = tail call i32 @llvm.arm.mrc2(i32 7, i32 1, i32 1, i32 1, i32 4) nounwind
+  ; CHECK: mcr2
+  tail call void @llvm.arm.mcr2(i32 7, i32 1, i32 %1, i32 1, i32 1, i32 4) nounwind
+  ; CHECK: mcrr
+  tail call void @llvm.arm.mcrr(i32 7, i32 1, i32 %0, i32 %1, i32 1) nounwind
+  ; CHECK: mcrr2
+  tail call void @llvm.arm.mcrr2(i32 7, i32 1, i32 %0, i32 %1, i32 1) nounwind
+  ; CHECK: cdp
+  tail call void @llvm.arm.cdp(i32 7, i32 3, i32 1, i32 1, i32 1, i32 5) nounwind
+  ; CHECK: cdp2
+  tail call void @llvm.arm.cdp2(i32 7, i32 3, i32 1, i32 1, i32 1, i32 5) nounwind
+  ret void
+}
+
+declare void @llvm.arm.cdp2(i32, i32, i32, i32, i32, i32) nounwind
+
+declare void @llvm.arm.cdp(i32, i32, i32, i32, i32, i32) nounwind
+
+declare void @llvm.arm.mcrr2(i32, i32, i32, i32, i32) nounwind
+
+declare void @llvm.arm.mcrr(i32, i32, i32, i32, i32) nounwind
+
+declare void @llvm.arm.mcr2(i32, i32, i32, i32, i32, i32) nounwind
+
+declare i32 @llvm.arm.mrc2(i32, i32, i32, i32, i32) nounwind
+
+declare void @llvm.arm.mcr(i32, i32, i32, i32, i32, i32) nounwind
+
+declare i32 @llvm.arm.mrc(i32, i32, i32, i32, i32) nounwind



From bruno.cardoso at gmail.com  Tue May  3 12:29:29 2011
From: bruno.cardoso at gmail.com (Bruno Cardoso Lopes)
Date: Tue, 03 May 2011 17:29:29 -0000
Subject: [llvm-commits] [llvm] r130764 - in /llvm/trunk/lib/Target/ARM:
 ARMInstrInfo.td ARMInstrThumb.td ARMInstrThumb2.td
Message-ID: <20110503172930.1DF902A6C12C@llvm.org>

Author: bruno
Date: Tue May  3 12:29:29 2011
New Revision: 130764

URL: http://llvm.org/viewvc/llvm-project?rev=130764&view=rev
Log:
Fold ARM coprocessor intrinsics patterns into the instructions defs whenever
it's possible.

Modified:
    llvm/trunk/lib/Target/ARM/ARMInstrInfo.td
    llvm/trunk/lib/Target/ARM/ARMInstrThumb.td
    llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td

Modified: llvm/trunk/lib/Target/ARM/ARMInstrInfo.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrInfo.td?rev=130764&r1=130763&r2=130764&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrInfo.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrInfo.td Tue May  3 12:29:29 2011
@@ -3398,7 +3398,8 @@
 def CDP : ABI<0b1110, (outs), (ins p_imm:$cop, i32imm:$opc1,
             c_imm:$CRd, c_imm:$CRn, c_imm:$CRm, i32imm:$opc2),
             NoItinerary, "cdp", "\t$cop, $opc1, $CRd, $CRn, $CRm, $opc2",
-            [/* For disassembly only; pattern left blank */]> {
+            [(int_arm_cdp imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                          imm:$CRm, imm:$opc2)]> {
   bits<4> opc1;
   bits<4> CRn;
   bits<4> CRd;
@@ -3415,14 +3416,11 @@
   let Inst{23-20} = opc1;
 }
 
-def : ARMPat<(int_arm_cdp imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
-                          imm:$CRm, imm:$opc2),
-             (CDP imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn, imm:$CRm,imm:$opc2)>;
-
 def CDP2 : ABXI<0b1110, (outs), (ins p_imm:$cop, i32imm:$opc1,
                c_imm:$CRd, c_imm:$CRn, c_imm:$CRm, i32imm:$opc2),
                NoItinerary, "cdp2\t$cop, $opc1, $CRd, $CRn, $CRm, $opc2",
-               [/* For disassembly only; pattern left blank */]> {
+               [(int_arm_cdp2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                              imm:$CRm, imm:$opc2)]> {
   let Inst{31-28} = 0b1111;
   bits<4> opc1;
   bits<4> CRn;
@@ -3440,11 +3438,6 @@
   let Inst{23-20} = opc1;
 }
 
-def : ARMPat<(int_arm_cdp2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
-                          imm:$CRm, imm:$opc2),
-             (CDP2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn, imm:$CRm,
-                                                            imm:$opc2)>;
-
 class ACI<dag oops, dag iops, string opc, string asm,
           IndexMode im = IndexModeNone>
   : InoP<oops, iops, AddrModeNone, Size4Bytes, im, BrFrm, NoItinerary,
@@ -3551,10 +3544,10 @@
 // Move between coprocessor and ARM core register -- for disassembly only
 //
 
-class MovRCopro<string opc, bit direction, dag oops, dag iops>
+class MovRCopro<string opc, bit direction, dag oops, dag iops,
+                list<dag> pattern>
   : ABI<0b1110, oops, iops, NoItinerary, opc,
-        "\t$cop, $opc1, $Rt, $CRn, $CRm, $opc2",
-        [/* For disassembly only; pattern left blank */]> {
+        "\t$cop, $opc1, $Rt, $CRn, $CRm, $opc2", pattern> {
   let Inst{20} = direction;
   let Inst{4} = 1;
 
@@ -3574,23 +3567,23 @@
 }
 
 def MCR : MovRCopro<"mcr", 0 /* from ARM core register to coprocessor */,
-                    (outs), (ins p_imm:$cop, i32imm:$opc1,
-                                 GPR:$Rt, c_imm:$CRn, c_imm:$CRm,
-                                 i32imm:$opc2)>;
+                    (outs),
+                    (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, c_imm:$CRn,
+                         c_imm:$CRm, i32imm:$opc2),
+                    [(int_arm_mcr imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                                  imm:$CRm, imm:$opc2)]>;
 def MRC : MovRCopro<"mrc", 1 /* from coprocessor to ARM core register */,
-                    (outs GPR:$Rt), (ins p_imm:$cop, i32imm:$opc1,
-                                         c_imm:$CRn, c_imm:$CRm, i32imm:$opc2)>;
+                    (outs GPR:$Rt),
+                    (ins p_imm:$cop, i32imm:$opc1, c_imm:$CRn, c_imm:$CRm,
+                         i32imm:$opc2), []>;
 
-def : ARMPat<(int_arm_mcr imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                           imm:$CRm, imm:$opc2),
-             (MCR imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn, imm:$CRm, imm:$opc2)>;
 def : ARMPat<(int_arm_mrc imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2),
              (MRC imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2)>;
 
-class MovRCopro2<string opc, bit direction, dag oops, dag iops>
+class MovRCopro2<string opc, bit direction, dag oops, dag iops,
+                 list<dag> pattern>
   : ABXI<0b1110, oops, iops, NoItinerary,
-         !strconcat(opc, "\t$cop, $opc1, $Rt, $CRn, $CRm, $opc2"),
-         [/* For disassembly only; pattern left blank */]> {
+         !strconcat(opc, "\t$cop, $opc1, $Rt, $CRn, $CRm, $opc2"), pattern> {
   let Inst{31-28} = 0b1111;
   let Inst{20} = direction;
   let Inst{4} = 1;
@@ -3611,27 +3604,25 @@
 }
 
 def MCR2 : MovRCopro2<"mcr2", 0 /* from ARM core register to coprocessor */,
-                      (outs), (ins p_imm:$cop, i32imm:$opc1,
-                                   GPR:$Rt, c_imm:$CRn, c_imm:$CRm,
-                                   i32imm:$opc2)>;
+                      (outs),
+                      (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, c_imm:$CRn,
+                           c_imm:$CRm, i32imm:$opc2),
+                      [(int_arm_mcr2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                                     imm:$CRm, imm:$opc2)]>;
 def MRC2 : MovRCopro2<"mrc2", 1 /* from coprocessor to ARM core register */,
-                      (outs GPR:$Rt), (ins p_imm:$cop, i32imm:$opc1,
-                                           c_imm:$CRn, c_imm:$CRm,
-                                           i32imm:$opc2)>;
+                      (outs GPR:$Rt),
+                      (ins p_imm:$cop, i32imm:$opc1, c_imm:$CRn, c_imm:$CRm,
+                           i32imm:$opc2), []>;
 
-def : ARMV5TPat<(int_arm_mcr2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                              imm:$CRm, imm:$opc2),
-                (MCR2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                      imm:$CRm, imm:$opc2)>;
 def : ARMV5TPat<(int_arm_mrc2 imm:$cop, imm:$opc1, imm:$CRn,
                               imm:$CRm, imm:$opc2),
                 (MRC2 imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2)>;
 
-class MovRRCopro<string opc, bit direction>
+class MovRRCopro<string opc, bit direction,
+                 list<dag> pattern = [/* For disassembly only */]>
   : ABI<0b1100, (outs), (ins p_imm:$cop, i32imm:$opc1,
         GPR:$Rt, GPR:$Rt2, c_imm:$CRm),
-        NoItinerary, opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm",
-        [/* For disassembly only; pattern left blank */]> {
+        NoItinerary, opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm", pattern> {
   let Inst{23-21} = 0b010;
   let Inst{20} = direction;
 
@@ -3648,18 +3639,16 @@
   let Inst{3-0}   = CRm;
 }
 
-def MCRR : MovRRCopro<"mcrr", 0 /* from ARM core register to coprocessor */>;
+def MCRR : MovRRCopro<"mcrr", 0 /* from ARM core register to coprocessor */,
+                      [(int_arm_mcrr imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2,
+                                     imm:$CRm)]>;
 def MRRC : MovRRCopro<"mrrc", 1 /* from coprocessor to ARM core register */>;
 
-def : ARMV5TEPat<(int_arm_mcrr imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2,
-                               imm:$CRm),
-                 (MCRR imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm)>;
-
-class MovRRCopro2<string opc, bit direction>
+class MovRRCopro2<string opc, bit direction,
+                  list<dag> pattern = [/* For disassembly only */]>
   : ABXI<0b1100, (outs), (ins p_imm:$cop, i32imm:$opc1,
-         GPR:$Rt, GPR:$Rt2, c_imm:$CRm),
-         NoItinerary, !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"),
-         [/* For disassembly only; pattern left blank */]> {
+         GPR:$Rt, GPR:$Rt2, c_imm:$CRm), NoItinerary,
+         !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"), pattern> {
   let Inst{31-28} = 0b1111;
   let Inst{23-21} = 0b010;
   let Inst{20} = direction;
@@ -3677,12 +3666,11 @@
   let Inst{3-0}   = CRm;
 }
 
-def MCRR2 : MovRRCopro2<"mcrr2", 0 /* from ARM core register to coprocessor */>;
+def MCRR2 : MovRRCopro2<"mcrr2", 0 /* from ARM core register to coprocessor */,
+                        [(int_arm_mcrr2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2,
+                                        imm:$CRm)]>;
 def MRRC2 : MovRRCopro2<"mrrc2", 1 /* from coprocessor to ARM core register */>;
 
-def : ARMV6Pat<(int_arm_mcrr2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm),
-               (MCRR2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm)>;
-
 //===----------------------------------------------------------------------===//
 // Move between special register and ARM core register -- for disassembly only
 //

Modified: llvm/trunk/lib/Target/ARM/ARMInstrThumb.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrThumb.td?rev=130764&r1=130763&r2=130764&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrThumb.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrThumb.td Tue May  3 12:29:29 2011
@@ -1352,9 +1352,10 @@
 // Move between coprocessor and ARM core register -- for disassembly only
 //
 
-class tMovRCopro<string opc, bit direction, dag oops, dag iops>
+class tMovRCopro<string opc, bit direction, dag oops, dag iops,
+                 list<dag> pattern>
   : T1Cop<oops, iops, !strconcat(opc, "\t$cop, $opc1, $Rt, $CRn, $CRm, $opc2"),
-          [/* For disassembly only; pattern left blank */]> {
+          pattern> {
   let Inst{27-24} = 0b1110;
   let Inst{20} = direction;
   let Inst{4} = 1;
@@ -1375,24 +1376,24 @@
 }
 
 def tMCR : tMovRCopro<"mcr", 0 /* from ARM core register to coprocessor */,
-           (outs), (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, c_imm:$CRn,
-                        c_imm:$CRm, i32imm:$opc2)>;
+           (outs),
+           (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, c_imm:$CRn,
+                c_imm:$CRm, i32imm:$opc2),
+           [(int_arm_mcr imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                         imm:$CRm, imm:$opc2)]>;
 def tMRC : tMovRCopro<"mrc", 1 /* from coprocessor to ARM core register */,
-           (outs GPR:$Rt), (ins p_imm:$cop, i32imm:$opc1, c_imm:$CRn,
-                                c_imm:$CRm, i32imm:$opc2)>;
+           (outs GPR:$Rt),
+           (ins p_imm:$cop, i32imm:$opc1, c_imm:$CRn, c_imm:$CRm, i32imm:$opc2),
+           []>;
 
-def : Pat<(int_arm_mcr imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                          imm:$CRm, imm:$opc2),
-          (tMCR imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                imm:$CRm, imm:$opc2)>, Requires<[IsThumb, HasV6T2]>;
 def : Pat<(int_arm_mrc imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2),
           (tMRC imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2)>,
           Requires<[IsThumb, HasV6T2]>;
 
-class tMovRRCopro<string opc, bit direction>
+class tMovRRCopro<string opc, bit direction,
+                  list<dag> pattern = [/* For disassembly only */]>
   : T1Cop<(outs), (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, GPR:$Rt2, c_imm:$CRm),
-          !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"),
-          [/* For disassembly only; pattern left blank */]> {
+          !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"), pattern> {
   let Inst{27-24} = 0b1100;
   let Inst{23-21} = 0b010;
   let Inst{20} = direction;
@@ -1410,20 +1411,19 @@
   let Inst{3-0}   = CRm;
 }
 
-def tMCRR : tMovRRCopro<"mcrr", 0 /* from ARM core register to coprocessor */>;
+def tMCRR : tMovRRCopro<"mcrr", 0 /* from ARM core register to coprocessor */,
+                        [(int_arm_mcrr imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2,
+                                       imm:$CRm)]>;
 def tMRRC : tMovRRCopro<"mrrc", 1 /* from coprocessor to ARM core register */>;
 
-def : Pat<(int_arm_mcrr imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm),
-          (tMCRR imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm)>,
-          Requires<[IsThumb, HasV6T2]>;
-
 //===----------------------------------------------------------------------===//
 // Other Coprocessor Instructions.  For disassembly only.
 //
 def tCDP : T1Cop<(outs), (ins p_imm:$cop, i32imm:$opc1,
                  c_imm:$CRd, c_imm:$CRn, c_imm:$CRm, i32imm:$opc2),
                  "cdp\t$cop, $opc1, $CRd, $CRn, $CRm, $opc2",
-                 [/* For disassembly only; pattern left blank */]> {
+                 [(int_arm_cdp imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                               imm:$CRm, imm:$opc2)]> {
   let Inst{27-24} = 0b1110;
 
   bits<4> opc1;
@@ -1442,11 +1442,6 @@
   let Inst{23-20} = opc1;
 }
 
-def : Pat<(int_arm_cdp imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
-                          imm:$CRm, imm:$opc2),
-          (tCDP imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
-                imm:$CRm, imm:$opc2)>, Requires<[IsThumb, HasV6T2]>;
-
 //===----------------------------------------------------------------------===//
 // TLS Instructions
 //

Modified: llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td?rev=130764&r1=130763&r2=130764&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td Tue May  3 12:29:29 2011
@@ -3347,9 +3347,10 @@
 // Move between coprocessor and ARM core register -- for disassembly only
 //
 
-class t2MovRCopro<string opc, bit direction, dag oops, dag iops>
+class t2MovRCopro<string opc, bit direction, dag oops, dag iops,
+                  list<dag> pattern>
   : T2Cop<oops, iops, !strconcat(opc, "\t$cop, $opc1, $Rt, $CRn, $CRm, $opc2"),
-          [/* For disassembly only; pattern left blank */]> {
+          pattern> {
   let Inst{27-24} = 0b1110;
   let Inst{20} = direction;
   let Inst{4} = 1;
@@ -3371,23 +3372,21 @@
 
 def t2MCR2 : t2MovRCopro<"mcr2", 0 /* from ARM core register to coprocessor */,
              (outs), (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, c_imm:$CRn,
-                          c_imm:$CRm, i32imm:$opc2)>;
+                          c_imm:$CRm, i32imm:$opc2),
+             [(int_arm_mcr2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
+                            imm:$CRm, imm:$opc2)]>;
 def t2MRC2 : t2MovRCopro<"mrc2", 1 /* from coprocessor to ARM core register */,
              (outs GPR:$Rt), (ins p_imm:$cop, i32imm:$opc1, c_imm:$CRn,
-                                  c_imm:$CRm, i32imm:$opc2)>;
+                                  c_imm:$CRm, i32imm:$opc2), []>;
 
-def : T2v6Pat<(int_arm_mcr2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                            imm:$CRm, imm:$opc2),
-              (t2MCR2 imm:$cop, imm:$opc1, GPR:$Rt, imm:$CRn,
-                      imm:$CRm, imm:$opc2)>;
 def : T2v6Pat<(int_arm_mrc2 imm:$cop, imm:$opc1, imm:$CRn,
                             imm:$CRm, imm:$opc2),
               (t2MRC2 imm:$cop, imm:$opc1, imm:$CRn, imm:$CRm, imm:$opc2)>;
 
-class t2MovRRCopro<string opc, bit direction>
+class t2MovRRCopro<string opc, bit direction,
+                   list<dag> pattern = [/* For disassembly only */]>
   : T2Cop<(outs), (ins p_imm:$cop, i32imm:$opc1, GPR:$Rt, GPR:$Rt2, c_imm:$CRm),
-          !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"),
-          [/* For disassembly only; pattern left blank */]> {
+          !strconcat(opc, "\t$cop, $opc1, $Rt, $Rt2, $CRm"), pattern> {
   let Inst{27-24} = 0b1100;
   let Inst{23-21} = 0b010;
   let Inst{20} = direction;
@@ -3406,13 +3405,12 @@
 }
 
 def t2MCRR2 : t2MovRRCopro<"mcrr2",
-                           0 /* from ARM core register to coprocessor */>;
+                           0 /* from ARM core register to coprocessor */,
+                           [(int_arm_mcrr2 imm:$cop, imm:$opc1, GPR:$Rt,
+                                           GPR:$Rt2, imm:$CRm)]>;
 def t2MRRC2 : t2MovRRCopro<"mrrc2",
                            1 /* from coprocessor to ARM core register */>;
 
-def : T2v6Pat<(int_arm_mcrr2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm),
-              (t2MCRR2 imm:$cop, imm:$opc1, GPR:$Rt, GPR:$Rt2, imm:$CRm)>;
-
 //===----------------------------------------------------------------------===//
 // Other Coprocessor Instructions.  For disassembly only.
 //
@@ -3420,7 +3418,8 @@
 def t2CDP2 : T2Cop<(outs), (ins p_imm:$cop, i32imm:$opc1,
                    c_imm:$CRd, c_imm:$CRn, c_imm:$CRm, i32imm:$opc2),
                    "cdp2\t$cop, $opc1, $CRd, $CRn, $CRm, $opc2",
-                   [/* For disassembly only; pattern left blank */]> {
+                   [(int_arm_cdp2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
+                                  imm:$CRm, imm:$opc2)]> {
   let Inst{27-24} = 0b1110;
 
   bits<4> opc1;
@@ -3438,8 +3437,3 @@
   let Inst{19-16} = CRn;
   let Inst{23-20} = opc1;
 }
-
-def : T2v6Pat<(int_arm_cdp2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
-                            imm:$CRm, imm:$opc2),
-              (t2CDP2 imm:$cop, imm:$opc1, imm:$CRd, imm:$CRn,
-                      imm:$CRm, imm:$opc2)>;



From bob.wilson at apple.com  Tue May  3 12:33:22 2011
From: bob.wilson at apple.com (Bob Wilson)
Date: Tue, 03 May 2011 17:33:22 -0000
Subject: [llvm-commits] [llvm] r130766 -
	/llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
Message-ID: <20110503173322.A71232A6C12C@llvm.org>

Author: bwilson
Date: Tue May  3 12:33:22 2011
New Revision: 130766

URL: http://llvm.org/viewvc/llvm-project?rev=130766&view=rev
Log:
Temporarily disable use of divmod compiler-rt functions for iOS.

Modified:
    llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp

Modified: llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp?rev=130766&r1=130765&r2=130766&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/ARM/ARMISelLowering.cpp Tue May  3 12:33:22 2011
@@ -398,12 +398,6 @@
     setLibcallCallingConv(RTLIB::UDIV_I32, CallingConv::ARM_AAPCS);
   }
 
-  // Use divmod iOS compiler-rt calls.
-  if (Subtarget->getTargetTriple().getOS() == Triple::IOS) {
-    setLibcallName(RTLIB::SDIVREM_I32, "__divmodsi4");
-    setLibcallName(RTLIB::UDIVREM_I32, "__udivmodsi4");
-  }
-
   if (Subtarget->isThumb1Only())
     addRegisterClass(MVT::i32, ARM::tGPRRegisterClass);
   else



From bob.wilson at apple.com  Tue May  3 12:54:49 2011
From: bob.wilson at apple.com (Bob Wilson)
Date: Tue, 03 May 2011 17:54:49 -0000
Subject: [llvm-commits] [llvm] r130769 -
	/llvm/trunk/test/CodeGen/ARM/divmod.ll
Message-ID: <20110503175449.7C9142A6C12C@llvm.org>

Author: bwilson
Date: Tue May  3 12:54:49 2011
New Revision: 130769

URL: http://llvm.org/viewvc/llvm-project?rev=130769&view=rev
Log:
Remove test for iOS divmod function, since that is disabled for now.

Removed:
    llvm/trunk/test/CodeGen/ARM/divmod.ll

Removed: llvm/trunk/test/CodeGen/ARM/divmod.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/divmod.ll?rev=130768&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/divmod.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/divmod.ll (removed)
@@ -1,58 +0,0 @@
-; RUN: llc < %s -mtriple=arm-apple-ios | FileCheck %s
-
-define void @foo(i32 %x, i32 %y, i32* nocapture %P) nounwind ssp {
-entry:
-; CHECK: foo:
-; CHECK: bl ___divmodsi4
-; CHECK-NOT: bl ___divmodsi4
-  %div = sdiv i32 %x, %y
-  store i32 %div, i32* %P, align 4
-  %rem = srem i32 %x, %y
-  %arrayidx6 = getelementptr inbounds i32* %P, i32 1
-  store i32 %rem, i32* %arrayidx6, align 4
-  ret void
-}
-
-define void @bar(i32 %x, i32 %y, i32* nocapture %P) nounwind ssp {
-entry:
-; CHECK: bar:
-; CHECK: bl ___udivmodsi4
-; CHECK-NOT: bl ___udivmodsi4
-  %div = udiv i32 %x, %y
-  store i32 %div, i32* %P, align 4
-  %rem = urem i32 %x, %y
-  %arrayidx6 = getelementptr inbounds i32* %P, i32 1
-  store i32 %rem, i32* %arrayidx6, align 4
-  ret void
-}
-
-; rdar://9280991
- at flags = external unnamed_addr global i32
- at tabsize = external unnamed_addr global i32
-
-define void @do_indent(i32 %cols) nounwind {
-entry:
-; CHECK: do_indent:
-  %0 = load i32* @flags, align 4
-  %1 = and i32 %0, 67108864
-  %2 = icmp eq i32 %1, 0
-  br i1 %2, label %bb1, label %bb
-
-bb:
-; CHECK: bl ___divmodsi4
-  %3 = load i32* @tabsize, align 4
-  %4 = srem i32 %cols, %3
-  %5 = sdiv i32 %cols, %3
-  %6 = tail call i32 @llvm.objectsize.i32(i8* null, i1 false)
-  %7 = tail call i8* @__memset_chk(i8* null, i32 9, i32 %5, i32 %6) nounwind
-  br label %bb1
-
-bb1:
-  %line_indent_len.0 = phi i32 [ %4, %bb ], [ 0, %entry ]
-  %8 = getelementptr inbounds i8* null, i32 %line_indent_len.0
-  store i8 0, i8* %8, align 1
-  ret void
-}
-
-declare i32 @llvm.objectsize.i32(i8*, i1) nounwind readnone
-declare i8* @__memset_chk(i8*, i32, i32, i32) nounwind



From ahatanak at gmail.com  Tue May  3 13:41:55 2011
From: ahatanak at gmail.com (Akira Hatanaka)
Date: Tue, 03 May 2011 18:41:55 -0000
Subject: [llvm-commits] [llvm] r130774 -
	/llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp
Message-ID: <20110503184155.2AD7B2A6C12C@llvm.org>

Author: ahatanak
Date: Tue May  3 13:41:54 2011
New Revision: 130774

URL: http://llvm.org/viewvc/llvm-project?rev=130774&view=rev
Log:
Fix function  MipsRegisterInfo::getRegisterNumbering.

Modified:
    llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp

Modified: llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp?rev=130774&r1=130773&r2=130774&view=diff
==============================================================================
--- llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp (original)
+++ llvm/trunk/lib/Target/Mips/MipsRegisterInfo.cpp Tue May  3 13:41:54 2011
@@ -65,16 +65,16 @@
     case Mips::T5   : case Mips::F13: return 13;
     case Mips::T6   : case Mips::F14: case Mips::D7: return 14;
     case Mips::T7   : case Mips::F15: return 15;
-    case Mips::T8   : case Mips::F16: case Mips::D8: return 16;
-    case Mips::T9   : case Mips::F17: return 17;
-    case Mips::S0   : case Mips::F18: case Mips::D9: return 18;
-    case Mips::S1   : case Mips::F19: return 19;
-    case Mips::S2   : case Mips::F20: case Mips::D10: return 20;
-    case Mips::S3   : case Mips::F21: return 21;
-    case Mips::S4   : case Mips::F22: case Mips::D11: return 22;
-    case Mips::S5   : case Mips::F23: return 23;
-    case Mips::S6   : case Mips::F24: case Mips::D12: return 24;
-    case Mips::S7   : case Mips::F25: return 25;
+    case Mips::S0   : case Mips::F16: case Mips::D8: return 16;
+    case Mips::S1   : case Mips::F17: return 17;
+    case Mips::S2   : case Mips::F18: case Mips::D9: return 18;
+    case Mips::S3   : case Mips::F19: return 19;
+    case Mips::S4   : case Mips::F20: case Mips::D10: return 20;
+    case Mips::S5   : case Mips::F21: return 21;
+    case Mips::S6   : case Mips::F22: case Mips::D11: return 22;
+    case Mips::S7   : case Mips::F23: return 23;
+    case Mips::T8   : case Mips::F24: case Mips::D12: return 24;
+    case Mips::T9   : case Mips::F25: return 25;
     case Mips::K0   : case Mips::F26: case Mips::D13: return 26;
     case Mips::K1   : case Mips::F27: return 27;
     case Mips::GP   : case Mips::F28: case Mips::D14: return 28;



From evan.cheng at apple.com  Tue May  3 14:09:32 2011
From: evan.cheng at apple.com (Evan Cheng)
Date: Tue, 03 May 2011 19:09:32 -0000
Subject: [llvm-commits] [llvm] r130778 -
	/llvm/trunk/test/CodeGen/ARM/fnmscs.ll
Message-ID: <20110503190932.E0D132A6C12C@llvm.org>

Author: evancheng
Date: Tue May  3 14:09:32 2011
New Revision: 130778

URL: http://llvm.org/viewvc/llvm-project?rev=130778&view=rev
Log:
Make the test less likely to fail with minor changes.

Modified:
    llvm/trunk/test/CodeGen/ARM/fnmscs.ll

Modified: llvm/trunk/test/CodeGen/ARM/fnmscs.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/fnmscs.ll?rev=130778&r1=130777&r2=130778&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/fnmscs.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/fnmscs.ll Tue May  3 14:09:32 2011
@@ -29,7 +29,7 @@
 ; NEON: vnmla.f32
 
 ; A8: t2:
-; A8: vnmul.f32 s{{[0123]}}, s{{[0123]}}, s{{[0123]}}
+; A8: vnmul.f32 s{{[01234]}}, s{{[01234]}}, s{{[01234]}}
 ; A8: vsub.f32 d{{[0-9]}}, d{{[0-9]}}, d{{[0-9]}}
 	%0 = fmul float %a, %b
 	%1 = fmul float -1.0, %0



From grosbach at apple.com  Tue May  3 14:09:57 2011
From: grosbach at apple.com (Jim Grosbach)
Date: Tue, 03 May 2011 19:09:57 -0000
Subject: [llvm-commits] [llvm] r130779 -
	/llvm/trunk/utils/TableGen/AsmMatcherEmitter.cpp
Message-ID: <20110503190957.0F4782A6C12C@llvm.org>

Author: grosbach
Date: Tue May  3 14:09:56 2011
New Revision: 130779

URL: http://llvm.org/viewvc/llvm-project?rev=130779&view=rev
Log:
Tidy up. Add missing newline to generated file.

Modified:
    llvm/trunk/utils/TableGen/AsmMatcherEmitter.cpp

Modified: llvm/trunk/utils/TableGen/AsmMatcherEmitter.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/utils/TableGen/AsmMatcherEmitter.cpp?rev=130779&r1=130778&r2=130779&view=diff
==============================================================================
--- llvm/trunk/utils/TableGen/AsmMatcherEmitter.cpp (original)
+++ llvm/trunk/utils/TableGen/AsmMatcherEmitter.cpp Tue May  3 14:09:56 2011
@@ -2321,7 +2321,7 @@
   OS << "    for (unsigned i = 0; i != " << MaxNumOperands << "; ++i) {\n";
   OS << "      if (i + 1 >= Operands.size()) {\n";
   OS << "        OperandsValid = (it->Classes[i] == " <<"InvalidMatchClass);\n";
-  OS << "        break;";
+  OS << "        break;\n";
   OS << "      }\n";
   OS << "      if (ValidateOperandClass(Operands[i+1], it->Classes[i]))\n";
   OS << "        continue;\n";



From baldrick at free.fr  Tue May  3 14:53:10 2011
From: baldrick at free.fr (Duncan Sands)
Date: Tue, 03 May 2011 19:53:10 -0000
Subject: [llvm-commits] [llvm] r130780 - in /llvm/trunk:
 include/llvm/Support/PatternMatch.h lib/Analysis/InstructionSimplify.cpp
 test/Transforms/InstSimplify/maxmin.ll
Message-ID: <20110503195310.C41B42A6C12C@llvm.org>

Author: baldrick
Date: Tue May  3 14:53:10 2011
New Revision: 130780

URL: http://llvm.org/viewvc/llvm-project?rev=130780&view=rev
Log:
Implement some basic simplifications involving min/max, for example
max(a,b) >= a -> true.  According to my super-optimizer, these are
by far the most common simplifications (of the -instsimplify kind)
that occur in the testsuite and aren't caught by -std-compile-opts.

Added:
    llvm/trunk/test/Transforms/InstSimplify/maxmin.ll
Modified:
    llvm/trunk/include/llvm/Support/PatternMatch.h
    llvm/trunk/lib/Analysis/InstructionSimplify.cpp

Modified: llvm/trunk/include/llvm/Support/PatternMatch.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Support/PatternMatch.h?rev=130780&r1=130779&r2=130780&view=diff
==============================================================================
--- llvm/trunk/include/llvm/Support/PatternMatch.h (original)
+++ llvm/trunk/include/llvm/Support/PatternMatch.h Tue May  3 14:53:10 2011
@@ -694,6 +694,99 @@
   return brc_match<Cond_t>(C, T, F);
 }
 
+
+//===----------------------------------------------------------------------===//
+// Matchers for max/min idioms, eg: "select (sgt x, y), x, y" -> smax(x,y).
+//
+
+template<typename LHS_t, typename RHS_t, typename Pred_t>
+struct MaxMin_match {
+  LHS_t L;
+  RHS_t R;
+
+  MaxMin_match(const LHS_t &LHS, const RHS_t &RHS)
+    : L(LHS), R(RHS) {}
+
+  template<typename OpTy>
+  bool match(OpTy *V) {
+    // Look for "(x pred y) ? x : y" or "(x pred y) ? y : x".
+    SelectInst *SI = dyn_cast<SelectInst>(V);
+    if (!SI)
+      return false;
+    ICmpInst *Cmp = dyn_cast<ICmpInst>(SI->getCondition());
+    if (!Cmp)
+      return false;
+    // At this point we have a select conditioned on a comparison.  Check that
+    // it is the values returned by the select that are being compared.
+    Value *TrueVal = SI->getTrueValue();
+    Value *FalseVal = SI->getFalseValue();
+    Value *LHS = Cmp->getOperand(0);
+    Value *RHS = Cmp->getOperand(1);
+    if ((TrueVal != LHS || FalseVal != RHS) &&
+        (TrueVal != RHS || FalseVal != LHS))
+      return false;
+    ICmpInst::Predicate Pred = LHS == TrueVal ?
+      Cmp->getPredicate() : Cmp->getSwappedPredicate();
+    // Does "(x pred y) ? x : y" represent the desired max/min operation?
+    if (!Pred_t::match(Pred))
+      return false;
+    // It does!  Bind the operands.
+    return L.match(LHS) && R.match(RHS);
+  }
+};
+
+/// smax_pred_ty - Helper class for identifying signed max predicates.
+struct smax_pred_ty {
+  static bool match(ICmpInst::Predicate Pred) {
+    return Pred == CmpInst::ICMP_SGT || Pred == CmpInst::ICMP_SGE;
+  }
+};
+
+/// smin_pred_ty - Helper class for identifying signed min predicates.
+struct smin_pred_ty {
+  static bool match(ICmpInst::Predicate Pred) {
+    return Pred == CmpInst::ICMP_SLT || Pred == CmpInst::ICMP_SLE;
+  }
+};
+
+/// umax_pred_ty - Helper class for identifying unsigned max predicates.
+struct umax_pred_ty {
+  static bool match(ICmpInst::Predicate Pred) {
+    return Pred == CmpInst::ICMP_UGT || Pred == CmpInst::ICMP_UGE;
+  }
+};
+
+/// umin_pred_ty - Helper class for identifying unsigned min predicates.
+struct umin_pred_ty {
+  static bool match(ICmpInst::Predicate Pred) {
+    return Pred == CmpInst::ICMP_ULT || Pred == CmpInst::ICMP_ULE;
+  }
+};
+
+template<typename LHS, typename RHS>
+inline MaxMin_match<LHS, RHS, smax_pred_ty>
+m_SMax(const LHS &L, const RHS &R) {
+  return MaxMin_match<LHS, RHS, smax_pred_ty>(L, R);
+}
+
+template<typename LHS, typename RHS>
+inline MaxMin_match<LHS, RHS, smin_pred_ty>
+m_SMin(const LHS &L, const RHS &R) {
+  return MaxMin_match<LHS, RHS, smin_pred_ty>(L, R);
+}
+
+template<typename LHS, typename RHS>
+inline MaxMin_match<LHS, RHS, umax_pred_ty>
+m_UMax(const LHS &L, const RHS &R) {
+  return MaxMin_match<LHS, RHS, umax_pred_ty>(L, R);
+}
+
+template<typename LHS, typename RHS>
+inline MaxMin_match<LHS, RHS, umin_pred_ty>
+m_UMin(const LHS &L, const RHS &R) {
+  return MaxMin_match<LHS, RHS, umin_pred_ty>(L, R);
+}
+
 } // end namespace PatternMatch
 } // end namespace llvm
 

Modified: llvm/trunk/lib/Analysis/InstructionSimplify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/InstructionSimplify.cpp?rev=130780&r1=130779&r2=130780&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/InstructionSimplify.cpp (original)
+++ llvm/trunk/lib/Analysis/InstructionSimplify.cpp Tue May  3 14:53:10 2011
@@ -1868,6 +1868,124 @@
     }
   }
 
+  // Simplify comparisons involving max/min.
+  Value *A, *B;
+  CmpInst::Predicate P = CmpInst::BAD_ICMP_PREDICATE;
+  CmpInst::Predicate EqP; // Chosen so that "A == max/min(A,B)" iff "A EqP B".
+
+  // Signed max/min.
+  if (match(LHS, m_SMax(m_Value(A), m_Value(B))) && (A == RHS || B == RHS)) {
+    if (A != RHS) std::swap(A, B); // smax(A, B) pred A.
+    EqP = CmpInst::ICMP_SGE; // "A == smax(A, B)" iff "A sge B".
+    // We analyze this as smax(A, B) pred A.
+    P = Pred;
+  } else if (match(RHS, m_SMax(m_Value(A), m_Value(B))) &&
+             (A == LHS || B == LHS)) {
+    if (A != LHS) std::swap(A, B); // A pred smax(A, B).
+    EqP = CmpInst::ICMP_SGE; // "A == smax(A, B)" iff "A sge B".
+    // We analyze this as smax(A, B) swapped-pred A.
+    P = CmpInst::getSwappedPredicate(Pred);
+  } else if (match(LHS, m_SMin(m_Value(A), m_Value(B))) &&
+             (A == RHS || B == RHS)) {
+    if (A != RHS) std::swap(A, B); // smin(A, B) pred A.
+    EqP = CmpInst::ICMP_SLE; // "A == smin(A, B)" iff "A sle B".
+    // We analyze this as smax(-A, -B) swapped-pred -A.
+    // Note that we do not need to actually form -A or -B thanks to EqP.
+    P = CmpInst::getSwappedPredicate(Pred);
+  } else if (match(RHS, m_SMin(m_Value(A), m_Value(B))) &&
+             (A == LHS || B == LHS)) {
+    if (A != LHS) std::swap(A, B); // A pred smin(A, B).
+    EqP = CmpInst::ICMP_SLE; // "A == smin(A, B)" iff "A sle B".
+    // We analyze this as smax(-A, -B) pred -A.
+    // Note that we do not need to actually form -A or -B thanks to EqP.
+    P = Pred;
+  }
+  if (P != CmpInst::BAD_ICMP_PREDICATE) {
+    // Cases correspond to "max(A, B) p A".
+    switch (P) {
+    default:
+      break;
+    case CmpInst::ICMP_EQ:
+    case CmpInst::ICMP_SLE:
+      // Equivalent to "A EqP B".
+      if (MaxRecurse)
+        if (Value *V = SimplifyICmpInst(EqP, A, B, TD, DT, MaxRecurse-1))
+          return V;
+      break;
+    case CmpInst::ICMP_NE:
+    case CmpInst::ICMP_SGT:
+      // Equivalent to "A inverse-EqP B".
+      if (MaxRecurse)
+        if (Value *V = SimplifyICmpInst(CmpInst::getInversePredicate(EqP), A, B,
+                                        TD, DT, MaxRecurse-1))
+          return V;
+      break;
+    case CmpInst::ICMP_SGE:
+      // Always true.
+      return Constant::getAllOnesValue(ITy);
+    case CmpInst::ICMP_SLT:
+      // Always false.
+      return Constant::getNullValue(ITy);
+    }
+  }
+
+  // Unsigned max/min.
+  P = CmpInst::BAD_ICMP_PREDICATE;
+  if (match(LHS, m_UMax(m_Value(A), m_Value(B))) && (A == RHS || B == RHS)) {
+    if (A != RHS) std::swap(A, B); // umax(A, B) pred A.
+    EqP = CmpInst::ICMP_UGE; // "A == umax(A, B)" iff "A uge B".
+    // We analyze this as umax(A, B) pred A.
+    P = Pred;
+  } else if (match(RHS, m_UMax(m_Value(A), m_Value(B))) &&
+             (A == LHS || B == LHS)) {
+    if (A != LHS) std::swap(A, B); // A pred umax(A, B).
+    EqP = CmpInst::ICMP_UGE; // "A == umax(A, B)" iff "A uge B".
+    // We analyze this as umax(A, B) swapped-pred A.
+    P = CmpInst::getSwappedPredicate(Pred);
+  } else if (match(LHS, m_UMin(m_Value(A), m_Value(B))) &&
+             (A == RHS || B == RHS)) {
+    if (A != RHS) std::swap(A, B); // umin(A, B) pred A.
+    EqP = CmpInst::ICMP_ULE; // "A == umin(A, B)" iff "A ule B".
+    // We analyze this as umax(-A, -B) swapped-pred -A.
+    // Note that we do not need to actually form -A or -B thanks to EqP.
+    P = CmpInst::getSwappedPredicate(Pred);
+  } else if (match(RHS, m_UMin(m_Value(A), m_Value(B))) &&
+             (A == LHS || B == LHS)) {
+    if (A != LHS) std::swap(A, B); // A pred umin(A, B).
+    EqP = CmpInst::ICMP_ULE; // "A == umin(A, B)" iff "A ule B".
+    // We analyze this as umax(-A, -B) pred -A.
+    // Note that we do not need to actually form -A or -B thanks to EqP.
+    P = Pred;
+  }
+  if (P != CmpInst::BAD_ICMP_PREDICATE) {
+    // Cases correspond to "max(A, B) p A".
+    switch (P) {
+    default:
+      break;
+    case CmpInst::ICMP_EQ:
+    case CmpInst::ICMP_ULE:
+      // Equivalent to "A EqP B".
+      if (MaxRecurse)
+        if (Value *V = SimplifyICmpInst(EqP, A, B, TD, DT, MaxRecurse-1))
+          return V;
+      break;
+    case CmpInst::ICMP_NE:
+    case CmpInst::ICMP_UGT:
+      // Equivalent to "A inverse-EqP B".
+      if (MaxRecurse)
+        if (Value *V = SimplifyICmpInst(CmpInst::getInversePredicate(EqP), A, B,
+                                        TD, DT, MaxRecurse-1))
+          return V;
+      break;
+    case CmpInst::ICMP_UGE:
+      // Always true.
+      return Constant::getAllOnesValue(ITy);
+    case CmpInst::ICMP_ULT:
+      // Always false.
+      return Constant::getNullValue(ITy);
+    }
+  }
+
   // If the comparison is with the result of a select instruction, check whether
   // comparing with either branch of the select always yields the same value.
   if (isa<SelectInst>(LHS) || isa<SelectInst>(RHS))

Added: llvm/trunk/test/Transforms/InstSimplify/maxmin.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstSimplify/maxmin.ll?rev=130780&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/InstSimplify/maxmin.ll (added)
+++ llvm/trunk/test/Transforms/InstSimplify/maxmin.ll Tue May  3 14:53:10 2011
@@ -0,0 +1,145 @@
+; RUN: opt < %s -instsimplify -S | FileCheck %s
+
+define i1 @max1(i32 %x, i32 %y) {
+; CHECK: @max1
+  %c = icmp sgt i32 %x, %y
+  %m = select i1 %c, i32 %x, i32 %y
+  %r = icmp slt i32 %m, %x
+  ret i1 %r
+; CHECK: ret i1 false
+}
+
+define i1 @max2(i32 %x, i32 %y) {
+; CHECK: @max2
+  %c = icmp sge i32 %x, %y
+  %m = select i1 %c, i32 %x, i32 %y
+  %r = icmp sge i32 %m, %x
+  ret i1 %r
+; CHECK: ret i1 true
+}
+
+define i1 @max3(i32 %x, i32 %y) {
+; CHECK: @max3
+  %c = icmp ugt i32 %x, %y
+  %m = select i1 %c, i32 %x, i32 %y
+  %r = icmp ult i32 %m, %x
+  ret i1 %r
+; CHECK: ret i1 false
+}
+
+define i1 @max4(i32 %x, i32 %y) {
+; CHECK: @max4
+  %c = icmp uge i32 %x, %y
+  %m = select i1 %c, i32 %x, i32 %y
+  %r = icmp uge i32 %m, %x
+  ret i1 %r
+; CHECK: ret i1 true
+}
+
+define i1 @max5(i32 %x, i32 %y) {
+; CHECK: @max5
+  %c = icmp sgt i32 %x, %y
+  %m = select i1 %c, i32 %x, i32 %y
+  %r = icmp sgt i32 %x, %m
+  ret i1 %r
+; CHECK: ret i1 false
+}
+
+define i1 @max6(i32 %x, i32 %y) {
+; CHECK: @max6
+  %c = icmp sge i32 %x, %y
+  %m = select i1 %c, i32 %x, i32 %y
+  %r = icmp sle i32 %x, %m
+  ret i1 %r
+; CHECK: ret i1 true
+}
+
+define i1 @max7(i32 %x, i32 %y) {
+; CHECK: @max7
+  %c = icmp ugt i32 %x, %y
+  %m = select i1 %c, i32 %x, i32 %y
+  %r = icmp ugt i32 %x, %m
+  ret i1 %r
+; CHECK: ret i1 false
+}
+
+define i1 @max8(i32 %x, i32 %y) {
+; CHECK: @max8
+  %c = icmp uge i32 %x, %y
+  %m = select i1 %c, i32 %x, i32 %y
+  %r = icmp ule i32 %x, %m
+  ret i1 %r
+; CHECK: ret i1 true
+}
+
+define i1 @min1(i32 %x, i32 %y) {
+; CHECK: @min1
+  %c = icmp sgt i32 %x, %y
+  %m = select i1 %c, i32 %y, i32 %x
+  %r = icmp sgt i32 %m, %x
+  ret i1 %r
+; CHECK: ret i1 false
+}
+
+define i1 @min2(i32 %x, i32 %y) {
+; CHECK: @min2
+  %c = icmp sge i32 %x, %y
+  %m = select i1 %c, i32 %y, i32 %x
+  %r = icmp sle i32 %m, %x
+  ret i1 %r
+; CHECK: ret i1 true
+}
+
+define i1 @min3(i32 %x, i32 %y) {
+; CHECK: @min3
+  %c = icmp ugt i32 %x, %y
+  %m = select i1 %c, i32 %y, i32 %x
+  %r = icmp ugt i32 %m, %x
+  ret i1 %r
+; CHECK: ret i1 false
+}
+
+define i1 @min4(i32 %x, i32 %y) {
+; CHECK: @min4
+  %c = icmp uge i32 %x, %y
+  %m = select i1 %c, i32 %y, i32 %x
+  %r = icmp ule i32 %m, %x
+  ret i1 %r
+; CHECK: ret i1 true
+}
+
+define i1 @min5(i32 %x, i32 %y) {
+; CHECK: @min5
+  %c = icmp sgt i32 %x, %y
+  %m = select i1 %c, i32 %y, i32 %x
+  %r = icmp slt i32 %x, %m
+  ret i1 %r
+; CHECK: ret i1 false
+}
+
+define i1 @min6(i32 %x, i32 %y) {
+; CHECK: @min6
+  %c = icmp sge i32 %x, %y
+  %m = select i1 %c, i32 %y, i32 %x
+  %r = icmp sge i32 %x, %m
+  ret i1 %r
+; CHECK: ret i1 true
+}
+
+define i1 @min7(i32 %x, i32 %y) {
+; CHECK: @min7
+  %c = icmp ugt i32 %x, %y
+  %m = select i1 %c, i32 %y, i32 %x
+  %r = icmp ult i32 %x, %m
+  ret i1 %r
+; CHECK: ret i1 false
+}
+
+define i1 @min8(i32 %x, i32 %y) {
+; CHECK: @min8
+  %c = icmp uge i32 %x, %y
+  %m = select i1 %c, i32 %y, i32 %x
+  %r = icmp uge i32 %x, %m
+  ret i1 %r
+; CHECK: ret i1 true
+}



From rafael.espindola at gmail.com  Tue May  3 15:26:43 2011
From: rafael.espindola at gmail.com (Rafael Avila de Espindola)
Date: Tue, 03 May 2011 16:26:43 -0400
Subject: [llvm-commits] What requires __debug_frame?
In-Reply-To: <84C9465D-9B69-429D-8631-FB15463CC930@apple.com>
References: <4DC01EC0.7080101@gmail.com>
	<84C9465D-9B69-429D-8631-FB15463CC930@apple.com>
Message-ID: <4DC06503.3030306@gmail.com>

On 11-05-03 12:12 PM, Devang Patel wrote:
>
> On May 3, 2011, at 8:26 AM, Rafael Avila de Espindola wrote:
>
>> I was wondering what software requires __debug_frame. I decided to
>> try changing the default and see what would break. Surprisingly, it
>> was just a a single test that I needed to update to expect cfi
>> directives.
>>
>> I used a clang modified with the attached patch to build a debug
>> version of clang itself and gdb is happy with it. I can set
>> breakpoints, get backtraces, etc.
>
> It is used by unwinder, so you want to test eh unwinding also.

The unwinder uses __eh_frame, no?

> - Devang

Cheers,
Rafael


From fjahanian at apple.com  Tue May  3 15:22:13 2011
From: fjahanian at apple.com (Fariborz Jahanian)
Date: Tue, 03 May 2011 20:22:13 -0000
Subject: [llvm-commits] [test-suite] r130784 -
 /test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c
Message-ID: <20110503202213.CAFCA2A6C12C@llvm.org>

Author: fjahanian
Date: Tue May  3 15:22:13 2011
New Revision: 130784

URL: http://llvm.org/viewvc/llvm-project?rev=130784&view=rev
Log:
Test related to rules for zero-length bitfields
in ms_struct structs. // rdar://8823265

Added:
    test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c

Added: test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c
URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c?rev=130784&view=auto
==============================================================================
--- test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c (added)
+++ test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c Tue May  3 15:22:13 2011
@@ -0,0 +1,56 @@
+#include <stdlib.h>
+
+#define ATTR __attribute__((__ms_struct__))
+
+struct {
+                   unsigned int bf_1 : 12;
+                   unsigned int : 0;
+                   unsigned int bf_2 : 12;
+} ATTR t1 = {1,2};
+static int a1[(sizeof(t1) == 8) -1];
+
+struct
+{
+                   char foo : 4;
+                   short : 0;
+                   char bar;
+} ATTR t2 = {3,4};
+static int a2[(sizeof(t2) == 4) -1];
+
+#pragma ms_struct on
+struct
+{
+                   char foo : 4;
+                   short : 0;
+                   char bar;
+} t3 = {5,6};
+#pragma ms_struct off
+static int a3[(sizeof(t3) == 4) -1];
+
+struct
+{
+  char foo : 6;
+  int : 0;
+} ATTR t4;
+static int a4[(sizeof(t4) == 4) -1];
+
+struct
+{
+  char foo : 4;
+  short : 0;
+  char bar : 8;
+} ATTR t5 = {7,8};
+static int a5[(sizeof(t5) == 4) -1];
+
+int main() {
+  if (t1.bf_1 != 1 || t1.bf_2 != 2)
+    abort();
+  if (t2.foo != 3 || t2.bar != 4)
+    abort();
+  if (t3.foo != 5 || t3.bar != 6)
+    abort();
+  if (t5.foo != 7 || t5.bar != 8)
+    abort();
+  return 0;
+}
+



From stoklund at 2pi.dk  Tue May  3 15:42:14 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Tue, 03 May 2011 20:42:14 -0000
Subject: [llvm-commits] [llvm] r130787 - in /llvm/trunk/lib/CodeGen:
 RegAllocGreedy.cpp SplitKit.cpp SplitKit.h
Message-ID: <20110503204214.149402A6C12C@llvm.org>

Author: stoklund
Date: Tue May  3 15:42:13 2011
New Revision: 130787

URL: http://llvm.org/viewvc/llvm-project?rev=130787&view=rev
Log:
Gracefully handle invalid live ranges. Fix PR9831.

Register coalescing can sometimes create live ranges that end in the middle of a
basic block without any killing instruction. When SplitKit detects this, it will
repair the live range by shrinking it to its uses.

Live range splitting also needs to know about this. When the range shrinks so
much that it becomes allocatable, live range splitting fails because it can't
find a good split point. It is paranoid about making progress, so an allocatable
range is considered an error.

The coalescer should really not be creating these bad live ranges. They appear
when coalescing dead copies.

Modified:
    llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp
    llvm/trunk/lib/CodeGen/SplitKit.cpp
    llvm/trunk/lib/CodeGen/SplitKit.h

Modified: llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp?rev=130787&r1=130786&r2=130787&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp (original)
+++ llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp Tue May  3 15:42:13 2011
@@ -1315,6 +1315,19 @@
 
   SA->analyze(&VirtReg);
 
+  // FIXME: SplitAnalysis may repair broken live ranges coming from the
+  // coalescer. That may cause the range to become allocatable which means that
+  // tryRegionSplit won't be making progress. This check should be replaced with
+  // an assertion when the coalescer is fixed.
+  if (SA->didRepairRange()) {
+    // VirtReg has changed, so all cached queries are invalid.
+    Order.rewind();
+    while (unsigned PhysReg = Order.next())
+      query(VirtReg, PhysReg).clear();
+    if (unsigned PhysReg = tryAssign(VirtReg, Order, NewVRegs))
+      return PhysReg;
+  }
+
   // First try to split around a region spanning multiple blocks.
   unsigned PhysReg = tryRegionSplit(VirtReg, Order, NewVRegs);
   if (PhysReg || !NewVRegs.empty())

Modified: llvm/trunk/lib/CodeGen/SplitKit.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SplitKit.cpp?rev=130787&r1=130786&r2=130787&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SplitKit.cpp (original)
+++ llvm/trunk/lib/CodeGen/SplitKit.cpp Tue May  3 15:42:13 2011
@@ -51,6 +51,7 @@
   UseBlocks.clear();
   ThroughBlocks.clear();
   CurLI = 0;
+  DidRepairRange = false;
 }
 
 SlotIndex SplitAnalysis::computeLastSplitPoint(unsigned Num) {
@@ -119,6 +120,7 @@
   if (!calcLiveBlockInfo()) {
     // FIXME: calcLiveBlockInfo found inconsistencies in the live range.
     // I am looking at you, SimpleRegisterCoalescing!
+    DidRepairRange = true;
     DEBUG(dbgs() << "*** Fixing inconsistent live interval! ***\n");
     const_cast<LiveIntervals&>(LIS)
       .shrinkToUses(const_cast<LiveInterval*>(CurLI));

Modified: llvm/trunk/lib/CodeGen/SplitKit.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SplitKit.h?rev=130787&r1=130786&r2=130787&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SplitKit.h (original)
+++ llvm/trunk/lib/CodeGen/SplitKit.h Tue May  3 15:42:13 2011
@@ -97,6 +97,9 @@
   /// NumThroughBlocks - Number of live-through blocks.
   unsigned NumThroughBlocks;
 
+  /// DidRepairRange - analyze was forced to shrinkToUses().
+  bool DidRepairRange;
+
   SlotIndex computeLastSplitPoint(unsigned Num);
 
   // Sumarize statistics by counting instructions using CurLI.
@@ -113,6 +116,11 @@
   /// split.
   void analyze(const LiveInterval *li);
 
+  /// didRepairRange() - Returns true if CurLI was invalid and has been repaired
+  /// by analyze(). This really shouldn't happen, but sometimes the coalescer
+  /// can create live ranges that end in mid-air.
+  bool didRepairRange() const { return DidRepairRange; }
+
   /// clear - clear all data structures so SplitAnalysis is ready to analyze a
   /// new interval.
   void clear();



From isanbard at gmail.com  Tue May  3 16:11:17 2011
From: isanbard at gmail.com (Bill Wendling)
Date: Tue, 03 May 2011 21:11:17 -0000
Subject: [llvm-commits] [llvm] r130791 - in /llvm/trunk:
 lib/Target/X86/X86InstrSSE.td lib/VMCore/AutoUpgrade.cpp
 test/Assembler/AutoUpgradeIntrinsics.ll test/CodeGen/X86/nontemporal.ll
Message-ID: <20110503211117.54AA62A6C12C@llvm.org>

Author: void
Date: Tue May  3 16:11:17 2011
New Revision: 130791

URL: http://llvm.org/viewvc/llvm-project?rev=130791&view=rev
Log:
Replace the "movnt" intrinsics with a native store + nontemporal metadata bit.
<rdar://problem/8460511>

Added:
    llvm/trunk/test/CodeGen/X86/nontemporal.ll
Modified:
    llvm/trunk/lib/Target/X86/X86InstrSSE.td
    llvm/trunk/lib/VMCore/AutoUpgrade.cpp
    llvm/trunk/test/Assembler/AutoUpgradeIntrinsics.ll

Modified: llvm/trunk/lib/Target/X86/X86InstrSSE.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrSSE.td?rev=130791&r1=130790&r2=130791&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrSSE.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrSSE.td Tue May  3 16:11:17 2011
@@ -1875,21 +1875,6 @@
 // SSE 1 & 2 - Non-temporal stores
 //===----------------------------------------------------------------------===//
 
-def VMOVNTPSmr_Int : VPSI<0x2B, MRMDestMem, (outs),
-                       (ins i128mem:$dst, VR128:$src),
-                       "movntps\t{$src, $dst|$dst, $src}",
-                       [(int_x86_sse_movnt_ps addr:$dst, VR128:$src)]>, VEX;
-def VMOVNTPDmr_Int : VPDI<0x2B, MRMDestMem, (outs),
-                       (ins i128mem:$dst, VR128:$src),
-                       "movntpd\t{$src, $dst|$dst, $src}",
-                       [(int_x86_sse2_movnt_pd addr:$dst, VR128:$src)]>, VEX;
-
-let ExeDomain = SSEPackedInt in
-  def VMOVNTDQmr_Int : VPDI<0xE7, MRMDestMem, (outs),
-                     (ins f128mem:$dst, VR128:$src),
-                     "movntdq\t{$src, $dst|$dst, $src}",
-                     [(int_x86_sse2_movnt_dq addr:$dst, VR128:$src)]>, VEX;
-
 let AddedComplexity = 400 in { // Prefer non-temporal versions
   def VMOVNTPSmr : VPSI<0x2B, MRMDestMem, (outs),
                        (ins f128mem:$dst, VR128:$src),
@@ -1906,12 +1891,16 @@
                         "movntdq\t{$src, $dst|$dst, $src}",
                         [(alignednontemporalstore (v2f64 VR128:$src),
                                                   addr:$dst)]>, VEX;
+
   let ExeDomain = SSEPackedInt in
-  def VMOVNTDQmr : VPDI<0xE7, MRMDestMem, (outs),
-                      (ins f128mem:$dst, VR128:$src),
-                      "movntdq\t{$src, $dst|$dst, $src}",
-                      [(alignednontemporalstore (v4f32 VR128:$src),
-                                                addr:$dst)]>, VEX;
+  def VMOVNTDQmr    : VPDI<0xE7, MRMDestMem, (outs),
+                           (ins f128mem:$dst, VR128:$src),
+                           "movntdq\t{$src, $dst|$dst, $src}",
+                           [(alignednontemporalstore (v4f32 VR128:$src),
+                                                     addr:$dst)]>, VEX;
+
+  def : Pat<(alignednontemporalstore (v2i64 VR128:$src), addr:$dst),
+            (VMOVNTDQmr addr:$dst, VR128:$src)>;
 
   def VMOVNTPSYmr : VPSI<0x2B, MRMDestMem, (outs),
                        (ins f256mem:$dst, VR256:$src),
@@ -1943,18 +1932,6 @@
 def : Pat<(int_x86_avx_movnt_ps_256 addr:$dst, VR256:$src),
           (VMOVNTPSYmr addr:$dst, VR256:$src)>;
 
-def MOVNTPSmr_Int : PSI<0x2B, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
-                    "movntps\t{$src, $dst|$dst, $src}",
-                    [(int_x86_sse_movnt_ps addr:$dst, VR128:$src)]>;
-def MOVNTPDmr_Int : PDI<0x2B, MRMDestMem, (outs), (ins i128mem:$dst, VR128:$src),
-                        "movntpd\t{$src, $dst|$dst, $src}",
-                        [(int_x86_sse2_movnt_pd addr:$dst, VR128:$src)]>;
-
-let ExeDomain = SSEPackedInt in
-def MOVNTDQmr_Int : PDI<0xE7, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
-                        "movntdq\t{$src, $dst|$dst, $src}",
-                        [(int_x86_sse2_movnt_dq addr:$dst, VR128:$src)]>;
-
 let AddedComplexity = 400 in { // Prefer non-temporal versions
 def MOVNTPSmr : PSI<0x2B, MRMDestMem, (outs), (ins f128mem:$dst, VR128:$src),
                     "movntps\t{$src, $dst|$dst, $src}",
@@ -1972,22 +1949,19 @@
                     "movntdq\t{$src, $dst|$dst, $src}",
                     [(alignednontemporalstore (v4f32 VR128:$src), addr:$dst)]>;
 
+def : Pat<(alignednontemporalstore (v2i64 VR128:$src), addr:$dst),
+          (MOVNTDQmr addr:$dst, VR128:$src)>;
+
 // There is no AVX form for instructions below this point
 def MOVNTImr : I<0xC3, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src),
                  "movnti\t{$src, $dst|$dst, $src}",
                  [(nontemporalstore (i32 GR32:$src), addr:$dst)]>,
                TB, Requires<[HasSSE2]>;
-
 def MOVNTI_64mr : RI<0xC3, MRMDestMem, (outs), (ins i64mem:$dst, GR64:$src),
                      "movnti\t{$src, $dst|$dst, $src}",
                      [(nontemporalstore (i64 GR64:$src), addr:$dst)]>,
                   TB, Requires<[HasSSE2]>;
-
 }
-def MOVNTImr_Int  :   I<0xC3, MRMDestMem, (outs), (ins i32mem:$dst, GR32:$src),
-                    "movnti\t{$src, $dst|$dst, $src}",
-                    [(int_x86_sse2_movnt_i addr:$dst, GR32:$src)]>,
-                  TB, Requires<[HasSSE2]>;
 
 //===----------------------------------------------------------------------===//
 // SSE 1 & 2 - Misc Instructions (No AVX form)

Modified: llvm/trunk/lib/VMCore/AutoUpgrade.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/AutoUpgrade.cpp?rev=130791&r1=130790&r2=130791&view=diff
==============================================================================
--- llvm/trunk/lib/VMCore/AutoUpgrade.cpp (original)
+++ llvm/trunk/lib/VMCore/AutoUpgrade.cpp Tue May  3 16:11:17 2011
@@ -533,6 +533,13 @@
       // Calls to these instructions are transformed into unaligned loads.
       NewFn = 0;
       return true;
+    } else if (Name.compare(5, 16, "x86.sse.movnt.ps", 16) == 0 ||
+               Name.compare(5, 17, "x86.sse2.movnt.dq", 17) == 0 ||
+               Name.compare(5, 17, "x86.sse2.movnt.pd", 17) == 0 ||
+               Name.compare(5, 17, "x86.sse2.movnt.i", 16) == 0) {
+      // Calls to these instructions are transformed into nontemporal stores.
+      NewFn = 0;
+      return true;
     } else if (Name.compare(5, 17, "x86.ssse3.pshuf.w", 17) == 0) {
       // This is an SSE/MMX instruction.
       const Type *X86_MMXTy = VectorType::getX86_MMXTy(FTy->getContext());
@@ -975,6 +982,31 @@
 
       // Remove intrinsic.
       CI->eraseFromParent();
+    } else if (F->getName() == "llvm.x86.sse.movnt.ps" ||
+               F->getName() == "llvm.x86.sse2.movnt.dq" ||
+               F->getName() == "llvm.x86.sse2.movnt.pd" ||
+               F->getName() == "llvm.x86.sse2.movnt.i") {
+      IRBuilder<> Builder(C);
+      Builder.SetInsertPoint(CI->getParent(), CI);
+
+      Module *M = F->getParent();
+      SmallVector<Value *, 1> Elts;
+      Elts.push_back(ConstantInt::get(Type::getInt32Ty(C), 1));
+      MDNode *Node = MDNode::get(C, Elts);
+
+      Value *Arg0 = CI->getArgOperand(0);
+      Value *Arg1 = CI->getArgOperand(1);
+
+      // Convert the type of the pointer to a pointer to the stored type.
+      Value *BC = Builder.CreateBitCast(Arg0,
+                                        PointerType::getUnqual(Arg1->getType()),
+                                        "cast");
+      StoreInst *SI = Builder.CreateStore(Arg1, BC);
+      SI->setMetadata(M->getMDKindID("nontemporal"), Node);
+      SI->setAlignment(16);
+
+      // Remove intrinsic.
+      CI->eraseFromParent();
     } else {
       llvm_unreachable("Unknown function for CallInst upgrade.");
     }

Modified: llvm/trunk/test/Assembler/AutoUpgradeIntrinsics.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Assembler/AutoUpgradeIntrinsics.ll?rev=130791&r1=130790&r2=130791&view=diff
==============================================================================
--- llvm/trunk/test/Assembler/AutoUpgradeIntrinsics.ll (original)
+++ llvm/trunk/test/Assembler/AutoUpgradeIntrinsics.ll Tue May  3 16:11:17 2011
@@ -10,6 +10,7 @@
 ; RUN:   not grep {llvm\\.x86\\.sse2\\.loadu}
 ; RUN: llvm-as < %s | llvm-dis | \
 ; RUN:   grep {llvm\\.x86\\.mmx\\.ps} | grep {x86_mmx} | count 16
+; RUN: llvm-as < %s | llvm-dis | FileCheck %s
 
 declare i32 @llvm.ctpop.i28(i28 %val)
 declare i32 @llvm.cttz.i29(i29 %val)
@@ -91,3 +92,20 @@
   %v2 = call <2 x double> @llvm.x86.sse2.loadu.pd(double* %b)
   ret void
 }
+
+declare void @llvm.x86.sse.movnt.ps(i8*, <4 x float>) nounwind readnone 
+declare void @llvm.x86.sse2.movnt.dq(i8*, <2 x double>) nounwind readnone 
+declare void @llvm.x86.sse2.movnt.pd(i8*, <2 x double>) nounwind readnone 
+declare void @llvm.x86.sse2.movnt.i(i8*, i32) nounwind readnone 
+
+define void @f(<4 x float> %A, i8* %B, <2 x double> %C, i32 %D) {
+; CHECK: store{{.*}}nontemporal
+  call void @llvm.x86.sse.movnt.ps(i8* %B, <4 x float> %A)
+; CHECK: store{{.*}}nontemporal
+  call void @llvm.x86.sse2.movnt.dq(i8* %B, <2 x double> %C)
+; CHECK: store{{.*}}nontemporal
+  call void @llvm.x86.sse2.movnt.pd(i8* %B, <2 x double> %C)
+; CHECK: store{{.*}}nontemporal
+  call void @llvm.x86.sse2.movnt.i(i8* %B, i32 %D)
+  ret void
+}

Added: llvm/trunk/test/CodeGen/X86/nontemporal.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/nontemporal.ll?rev=130791&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/nontemporal.ll (added)
+++ llvm/trunk/test/CodeGen/X86/nontemporal.ll Tue May  3 16:11:17 2011
@@ -0,0 +1,19 @@
+; RUN: llc < %s -march=x86 -mattr=+sse2 | FileCheck %s
+
+define void @f(<4 x float> %A, i8* %B, <2 x double> %C, i32 %D, <2 x i64> %E) {
+; CHECK: movntps
+  %cast = bitcast i8* %B to <4 x float>*
+  store <4 x float> %A, <4 x float>* %cast, align 16, !nontemporal !0
+; CHECK: movntdq
+  %cast1 = bitcast i8* %B to <2 x i64>*
+  store <2 x i64> %E, <2 x i64>* %cast1, align 16, !nontemporal !0
+; CHECK: movntpd
+  %cast2 = bitcast i8* %B to <2 x double>*
+  store <2 x double> %C, <2 x double>* %cast2, align 16, !nontemporal !0
+; CHECK: movnti
+  %cast3 = bitcast i8* %B to i32*
+  store i32 %D, i32* %cast3, align 16, !nontemporal !0
+  ret void
+}
+
+!0 = metadata !{i32 1}



From daniel at zuster.org  Tue May  3 16:33:37 2011
From: daniel at zuster.org (Daniel Dunbar)
Date: Tue, 03 May 2011 21:33:37 -0000
Subject: [llvm-commits] [llvm] r130793 - /llvm/trunk/lib/MC/MCDwarf.cpp
Message-ID: <20110503213337.D7F902A6C12C@llvm.org>

Author: ddunbar
Date: Tue May  3 16:33:37 2011
New Revision: 130793

URL: http://llvm.org/viewvc/llvm-project?rev=130793&view=rev
Log:
MCDwarf: Don't save Twine to local variable, this is almost never safe to do
(and should thus never be done).
 - Should fix a crash on win32.

Modified:
    llvm/trunk/lib/MC/MCDwarf.cpp

Modified: llvm/trunk/lib/MC/MCDwarf.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/MC/MCDwarf.cpp?rev=130793&r1=130792&r2=130793&view=diff
==============================================================================
--- llvm/trunk/lib/MC/MCDwarf.cpp (original)
+++ llvm/trunk/lib/MC/MCDwarf.cpp Tue May  3 16:33:37 2011
@@ -740,8 +740,8 @@
   const TargetAsmInfo &asmInfo = context.getTargetAsmInfo();
 
   if (!asmInfo.isFunctionEHFrameSymbolPrivate()) {
-    Twine EHName = frame.Function->getName() + Twine(".eh");
-    MCSymbol *EHSym = context.GetOrCreateSymbol(EHName);
+    MCSymbol *EHSym = context.GetOrCreateSymbol(
+      frame.Function->getName() + Twine(".eh"));
     streamer.EmitEHSymAttributes(frame.Function, EHSym);
     streamer.EmitLabel(EHSym);
   }



From daniel at zuster.org  Tue May  3 16:41:27 2011
From: daniel at zuster.org (Daniel Dunbar)
Date: Tue, 3 May 2011 14:41:27 -0700
Subject: [llvm-commits] [llvm] r130375 - in /llvm/trunk:
 include/llvm/Target/TargetAsmInfo.h lib/MC/MCDwarf.cpp
In-Reply-To: <20110428024642.C47FF2A6C135@llvm.org>
References: <20110428024642.C47FF2A6C135@llvm.org>
Message-ID: <BANLkTimTEuUEzw0Ws1z6HcBusdCjMOWGfA@mail.gmail.com>

Hi Rafael,

On Wed, Apr 27, 2011 at 7:46 PM, Rafael Espindola
<rafael.espindola at gmail.com> wrote:
> Author: rafael
> Date: Wed Apr 27 21:46:42 2011
> New Revision: 130375
>
> URL: http://llvm.org/viewvc/llvm-project?rev=130375&view=rev
> Log:
> Forward isFunctionEHFrameSymbolPrivate. If it is false, produce the foo.eh
> symbols.
>
> Modified:
> ? ?llvm/trunk/include/llvm/Target/TargetAsmInfo.h
> ? ?llvm/trunk/lib/MC/MCDwarf.cpp
>
> Modified: llvm/trunk/include/llvm/Target/TargetAsmInfo.h
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Target/TargetAsmInfo.h?rev=130375&r1=130374&r2=130375&view=diff
> ==============================================================================
> --- llvm/trunk/include/llvm/Target/TargetAsmInfo.h (original)
> +++ llvm/trunk/include/llvm/Target/TargetAsmInfo.h Wed Apr 27 21:46:42 2011
> @@ -62,6 +62,10 @@
> ? ? return TLOF->getFDEEncoding();
> ? }
>
> + ?bool isFunctionEHFrameSymbolPrivate() const {
> + ? ?return TLOF->isFunctionEHFrameSymbolPrivate();
> + ?}
> +
> ? unsigned getDwarfRARegNum(bool isEH) const {
> ? ? return TRI->getDwarfRegNum(TRI->getRARegister(), isEH);
> ? }
>
> Modified: llvm/trunk/lib/MC/MCDwarf.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/MC/MCDwarf.cpp?rev=130375&r1=130374&r2=130375&view=diff
> ==============================================================================
> --- llvm/trunk/lib/MC/MCDwarf.cpp (original)
> +++ llvm/trunk/lib/MC/MCDwarf.cpp Wed Apr 27 21:46:42 2011
> @@ -17,6 +17,7 @@
> ?#include "llvm/MC/MCContext.h"
> ?#include "llvm/MC/MCObjectWriter.h"
> ?#include "llvm/ADT/SmallString.h"
> +#include "llvm/ADT/Twine.h"
> ?#include "llvm/Support/Debug.h"
> ?#include "llvm/Support/ErrorHandling.h"
> ?#include "llvm/Support/raw_ostream.h"
> @@ -713,6 +714,12 @@
> ? MCSymbol *fdeEnd = context.CreateTempSymbol();
> ? const TargetAsmInfo &asmInfo = context.getTargetAsmInfo();
>
> + ?if (!asmInfo.isFunctionEHFrameSymbolPrivate()) {
> + ? ?Twine EHName = frame.Function->getName() + Twine(".eh");
> + ? ?MCSymbol *EHSym = context.GetOrCreateSymbol(EHName);
> + ? ?streamer.EmitLabel(EHSym);

Note that this use of Twine is not safe (I fixed in r130793). You can
almost never safely store local variables of Twine type.

>From Twine.h:
--
  /// A Twine is not intended for use directly and should not be stored, its
  /// implementation relies on the ability to store pointers to temporary stack
  /// objects which may be deallocated at the end of a statement. Twines should
  /// only be used accepted as const references in arguments, when an API wishes
  /// to accept possibly-concatenated strings.
--

 - Daniel

> + ?}
> +
> ? // Length
> ? const MCExpr *Length = MakeStartMinusEndExpr(streamer, *fdeStart, *fdeEnd, 0);
> ? streamer.EmitAbsValue(Length, 4);
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>


From dpatel at apple.com  Tue May  3 16:50:35 2011
From: dpatel at apple.com (Devang Patel)
Date: Tue, 03 May 2011 21:50:35 -0000
Subject: [llvm-commits] [llvm] r130794 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
Message-ID: <20110503215035.321BF2A6C12C@llvm.org>

Author: dpatel
Date: Tue May  3 16:50:34 2011
New Revision: 130794

URL: http://llvm.org/viewvc/llvm-project?rev=130794&view=rev
Log:
Even if the subprogram is going to use AT_specification, emit DW_AT_MIPS_linkage_name. This helps gdb and fixes var-path-expr.exp regression reported by gdb testsuite.

Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130794&r1=130793&r2=130794&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Tue May  3 16:50:34 2011
@@ -332,6 +332,11 @@
   // Add function template parameters.
   SPCU->addTemplateParams(*SPDie, SP.getTemplateParams());
 
+  StringRef LinkageName = SP.getLinkageName();
+  if (!LinkageName.empty())
+    SPCU->addString(SPDie, dwarf::DW_AT_MIPS_linkage_name, dwarf::DW_FORM_string,
+                    getRealLinkageName(LinkageName));
+
   // If this DIE is going to refer declaration info using AT_specification
   // then there is no need to add other attributes.
   if (SP.getFunctionDeclaration().isSubprogram())
@@ -342,11 +347,6 @@
     SPCU->addString(SPDie, dwarf::DW_AT_name, dwarf::DW_FORM_string, 
                     SP.getName());
 
-  StringRef LinkageName = SP.getLinkageName();
-  if (!LinkageName.empty())
-    SPCU->addString(SPDie, dwarf::DW_AT_MIPS_linkage_name, dwarf::DW_FORM_string,
-                    getRealLinkageName(LinkageName));
-
   SPCU->addSourceLine(SPDie, SP);
 
   if (SP.isPrototyped()) 



From fjahanian at apple.com  Tue May  3 17:17:46 2011
From: fjahanian at apple.com (Fariborz Jahanian)
Date: Tue, 03 May 2011 22:17:46 -0000
Subject: [llvm-commits] [test-suite] r130798 -
 /test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c
Message-ID: <20110503221746.DC9342A6C12C@llvm.org>

Author: fjahanian
Date: Tue May  3 17:17:46 2011
New Revision: 130798

URL: http://llvm.org/viewvc/llvm-project?rev=130798&view=rev
Log:
More unit tests for ms_struct and zero-length
bitfield alignment rules.

Modified:
    test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c

Modified: test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c
URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c?rev=130798&r1=130797&r2=130798&view=diff
==============================================================================
--- test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c (original)
+++ test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield-init-1.c Tue May  3 17:17:46 2011
@@ -42,6 +42,44 @@
 } ATTR t5 = {7,8};
 static int a5[(sizeof(t5) == 4) -1];
 
+struct
+{
+                   char foo : 4;
+                   short : 0;
+                   long  :0;
+                   char bar;
+} ATTR t6 = {5, 10};
+static int a6[(sizeof(t6) == 4) -1];
+
+struct
+{
+                   char foo : 4;
+                   int  :0;
+                   short : 0;
+                   char bar;
+} ATTR t7 = {5, 10};
+static int a7[(sizeof(t7) == 8) -1];
+
+struct
+{
+                   char foo : 4;
+                   short : 0;
+                   int  :0;
+                   char bar:7;
+} ATTR t8 = {5, 10};
+static int a8[(sizeof(t8) == 4) -1];
+
+struct
+{
+                   char foo : 4;
+                   char : 0;
+                   short : 0;
+                   int : 0;
+                   long  :0;
+                   char bar;
+} ATTR t10 = {5, 10};
+static int a10[(sizeof(t10) == 2) -1];
+
 int main() {
   if (t1.bf_1 != 1 || t1.bf_2 != 2)
     abort();
@@ -51,6 +89,14 @@
     abort();
   if (t5.foo != 7 || t5.bar != 8)
     abort();
+  if (t6.foo != 5 || t6.bar != 10)
+    abort();
+  if (t7.foo != 5 || t7.bar != 10)
+    abort();
+  if (t8.foo != 5 || t8.bar != 10)
+    abort();
+  if (t10.foo != 5 || t10.bar != 10)
+    abort();
   return 0;
 }
 



From atrick at apple.com  Tue May  3 17:24:10 2011
From: atrick at apple.com (Andrew Trick)
Date: Tue, 03 May 2011 22:24:10 -0000
Subject: [llvm-commits] [llvm] r130799 -
	/llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp
Message-ID: <20110503222410.395582A6C12C@llvm.org>

Author: atrick
Date: Tue May  3 17:24:10 2011
New Revision: 130799

URL: http://llvm.org/viewvc/llvm-project?rev=130799&view=rev
Log:
indvars: Added canExpandBackEdgeTakenCount.

Only create a canonical IV for backedge taken count if it will
actually be used by LinearFunctionTestReplace. And some related
cleanup, preparing to reduce dependence on canonical IVs.
No significant effect on x86 or arm in the test-suite.

Modified:
    llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp

Modified: llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp?rev=130799&r1=130798&r2=130799&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp Tue May  3 17:24:10 2011
@@ -105,11 +105,12 @@
     void EliminateIVRemainders();
     void RewriteNonIntegerIVs(Loop *L);
 
+    bool canExpandBackedgeTakenCount(Loop *L,
+                                     const SCEV *BackedgeTakenCount);
+
     ICmpInst *LinearFunctionTestReplace(Loop *L, const SCEV *BackedgeTakenCount,
-                                   PHINode *IndVar,
-                                   BasicBlock *ExitingBlock,
-                                   BranchInst *BI,
-                                   SCEVExpander &Rewriter);
+                                        PHINode *IndVar,
+                                        SCEVExpander &Rewriter);
     void RewriteLoopExitValues(Loop *L, SCEVExpander &Rewriter);
 
     void RewriteIVExpressions(Loop *L, SCEVExpander &Rewriter);
@@ -183,17 +184,24 @@
   return true;
 }
 
-/// LinearFunctionTestReplace - This method rewrites the exit condition of the
-/// loop to be a canonical != comparison against the incremented loop induction
-/// variable.  This pass is able to rewrite the exit tests of any loop where the
-/// SCEV analysis can determine a loop-invariant trip count of the loop, which
-/// is actually a much broader range than just linear tests.
-ICmpInst *IndVarSimplify::LinearFunctionTestReplace(Loop *L,
-                                   const SCEV *BackedgeTakenCount,
-                                   PHINode *IndVar,
-                                   BasicBlock *ExitingBlock,
-                                   BranchInst *BI,
-                                   SCEVExpander &Rewriter) {
+/// canExpandBackedgeTakenCount - Return true if this loop's backedge taken
+/// count expression can be safely and cheaply expanded into an instruction
+/// sequence that can be used by LinearFunctionTestReplace.
+bool IndVarSimplify::
+canExpandBackedgeTakenCount(Loop *L,
+                            const SCEV *BackedgeTakenCount) {
+  if (isa<SCEVCouldNotCompute>(BackedgeTakenCount) ||
+      BackedgeTakenCount->isZero())
+    return false;
+
+  if (!L->getExitingBlock())
+    return false;
+
+  // Can't rewrite non-branch yet.
+  BranchInst *BI = dyn_cast<BranchInst>(L->getExitingBlock()->getTerminator());
+  if (!BI)
+    return false;
+
   // Special case: If the backedge-taken count is a UDiv, it's very likely a
   // UDiv that ScalarEvolution produced in order to compute a precise
   // expression, rather than a UDiv from the user's code. If we can't find a
@@ -208,16 +216,31 @@
       const SCEV *L = SE->getSCEV(OrigCond->getOperand(0));
       L = SE->getMinusSCEV(L, SE->getConstant(L->getType(), 1));
       if (L != BackedgeTakenCount)
-        return 0;
+        return false;
     }
   }
+  return true;
+}
+
+/// LinearFunctionTestReplace - This method rewrites the exit condition of the
+/// loop to be a canonical != comparison against the incremented loop induction
+/// variable.  This pass is able to rewrite the exit tests of any loop where the
+/// SCEV analysis can determine a loop-invariant trip count of the loop, which
+/// is actually a much broader range than just linear tests.
+ICmpInst *IndVarSimplify::
+LinearFunctionTestReplace(Loop *L,
+                          const SCEV *BackedgeTakenCount,
+                          PHINode *IndVar,
+                          SCEVExpander &Rewriter) {
+  assert(canExpandBackedgeTakenCount(L, BackedgeTakenCount) && "precondition");
+  BranchInst *BI = cast<BranchInst>(L->getExitingBlock()->getTerminator());
 
   // If the exiting block is not the same as the backedge block, we must compare
   // against the preincremented value, otherwise we prefer to compare against
   // the post-incremented value.
   Value *CmpIndVar;
   const SCEV *RHS = BackedgeTakenCount;
-  if (ExitingBlock == L->getLoopLatch()) {
+  if (L->getExitingBlock() == L->getLoopLatch()) {
     // Add one to the "backedge-taken" count to get the trip count.
     // If this addition may overflow, we have to be more pessimistic and
     // cast the induction variable before doing the add.
@@ -240,7 +263,7 @@
     // The BackedgeTaken expression contains the number of times that the
     // backedge branches to the loop header.  This is one less than the
     // number of times the loop executes, so use the incremented indvar.
-    CmpIndVar = IndVar->getIncomingValueForBlock(ExitingBlock);
+    CmpIndVar = IndVar->getIncomingValueForBlock(L->getExitingBlock());
   } else {
     // We have to use the preincremented value...
     RHS = SE->getTruncateOrZeroExtend(BackedgeTakenCount,
@@ -533,7 +556,6 @@
   // transform them to use integer recurrences.
   RewriteNonIntegerIVs(L);
 
-  BasicBlock *ExitingBlock = L->getExitingBlock(); // may be null
   const SCEV *BackedgeTakenCount = SE->getBackedgeTakenCount(L);
 
   // Create a rewriter object which we'll use to transform the code with.
@@ -558,23 +580,26 @@
   // a canonical induction variable should be inserted.
   const Type *LargestType = 0;
   bool NeedCannIV = false;
-  if (!isa<SCEVCouldNotCompute>(BackedgeTakenCount)) {
-    LargestType = BackedgeTakenCount->getType();
-    LargestType = SE->getEffectiveSCEVType(LargestType);
+  bool ExpandBECount = canExpandBackedgeTakenCount(L, BackedgeTakenCount);
+  if (ExpandBECount) {
     // If we have a known trip count and a single exit block, we'll be
     // rewriting the loop exit test condition below, which requires a
     // canonical induction variable.
-    if (ExitingBlock)
-      NeedCannIV = true;
+    NeedCannIV = true;
+    const Type *Ty = BackedgeTakenCount->getType();
+    if (!LargestType ||
+        SE->getTypeSizeInBits(Ty) >
+        SE->getTypeSizeInBits(LargestType))
+      LargestType = SE->getEffectiveSCEVType(Ty);
   }
   for (IVUsers::const_iterator I = IU->begin(), E = IU->end(); I != E; ++I) {
+    NeedCannIV = true;
     const Type *Ty =
       SE->getEffectiveSCEVType(I->getOperandValToReplace()->getType());
     if (!LargestType ||
         SE->getTypeSizeInBits(Ty) >
           SE->getTypeSizeInBits(LargestType))
       LargestType = Ty;
-    NeedCannIV = true;
   }
 
   // Now that we know the largest of the induction variable expressions
@@ -614,15 +639,13 @@
   // If we have a trip count expression, rewrite the loop's exit condition
   // using it.  We can currently only handle loops with a single exit.
   ICmpInst *NewICmp = 0;
-  if (!isa<SCEVCouldNotCompute>(BackedgeTakenCount) &&
-      !BackedgeTakenCount->isZero() &&
-      ExitingBlock) {
+  if (ExpandBECount) {
+    assert(canExpandBackedgeTakenCount(L, BackedgeTakenCount) &&
+           "canonical IV disrupted BackedgeTaken expansion");
     assert(NeedCannIV &&
            "LinearFunctionTestReplace requires a canonical induction variable");
-    // Can't rewrite non-branch yet.
-    if (BranchInst *BI = dyn_cast<BranchInst>(ExitingBlock->getTerminator()))
-      NewICmp = LinearFunctionTestReplace(L, BackedgeTakenCount, IndVar,
-                                          ExitingBlock, BI, Rewriter);
+    NewICmp = LinearFunctionTestReplace(L, BackedgeTakenCount, IndVar,
+                                        Rewriter);
   }
 
   // Rewrite IV-derived expressions.



From stoklund at 2pi.dk  Tue May  3 17:31:21 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Tue, 03 May 2011 22:31:21 -0000
Subject: [llvm-commits] [llvm] r130800 - in /llvm/trunk/test/CodeGen/ARM:
 2011-04-07-schediv.ll 2011-04-11-MachineLICMBug.ll bfi.ll fcopysign.ll
 ldst-f32-2-i32.ll select-imm.ll stm.ll vldlane.ll
Message-ID: <20110503223122.0AE5B2A6C12C@llvm.org>

Author: stoklund
Date: Tue May  3 17:31:21 2011
New Revision: 130800

URL: http://llvm.org/viewvc/llvm-project?rev=130800&view=rev
Log:
Fix a bunch of ARM tests to be register allocation independent.

Modified:
    llvm/trunk/test/CodeGen/ARM/2011-04-07-schediv.ll
    llvm/trunk/test/CodeGen/ARM/2011-04-11-MachineLICMBug.ll
    llvm/trunk/test/CodeGen/ARM/bfi.ll
    llvm/trunk/test/CodeGen/ARM/fcopysign.ll
    llvm/trunk/test/CodeGen/ARM/ldst-f32-2-i32.ll
    llvm/trunk/test/CodeGen/ARM/select-imm.ll
    llvm/trunk/test/CodeGen/ARM/stm.ll
    llvm/trunk/test/CodeGen/ARM/vldlane.ll

Modified: llvm/trunk/test/CodeGen/ARM/2011-04-07-schediv.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/2011-04-07-schediv.ll?rev=130800&r1=130799&r2=130800&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/2011-04-07-schediv.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/2011-04-07-schediv.ll Tue May  3 17:31:21 2011
@@ -13,6 +13,7 @@
 ; Make sure the scheduler schedules all uses of the preincrement
 ; induction variable before defining the postincrement value.
 ; CHECK: t:
+; CHECK: %bb
 ; CHECK-NOT: mov
 bb:                                               ; preds = %entry, %bb
   %j.05 = phi i32 [ %2, %bb ], [ 0, %entry ]

Modified: llvm/trunk/test/CodeGen/ARM/2011-04-11-MachineLICMBug.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/2011-04-11-MachineLICMBug.ll?rev=130800&r1=130799&r2=130800&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/2011-04-11-MachineLICMBug.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/2011-04-11-MachineLICMBug.ll Tue May  3 17:31:21 2011
@@ -14,15 +14,15 @@
   br i1 %cmp, label %for.body, label %return
 
 for.body:
-; CHECK: %for.body
-; CHECK: movs r{{[0-9]+}}, #1
+; CHECK: %for.
+; CHECK: movs r{{[0-9]+}}, #{{[01]}}
   %arrayidx = getelementptr i32* %A, i32 %0
   %tmp4 = load i32* %arrayidx, align 4
   %cmp6 = icmp eq i32 %tmp4, %value
   br i1 %cmp6, label %return, label %for.inc
 
-; CHECK: %for.cond
-; CHECK: movs r{{[0-9]+}}, #0
+; CHECK: %for.
+; CHECK: movs r{{[0-9]+}}, #{{[01]}}
 
 for.inc:
   %inc = add i32 %0, 1

Modified: llvm/trunk/test/CodeGen/ARM/bfi.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/bfi.ll?rev=130800&r1=130799&r2=130800&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/bfi.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/bfi.ll Tue May  3 17:31:21 2011
@@ -31,8 +31,7 @@
 entry:
 ; CHECK: f3
 ; CHECK: lsr{{.*}} #7
-; CHECK: mov r0, r1
-; CHECK: bfi r0, r2, #7, #16
+; CHECK: bfi {{.*}}, #7, #16
   %and = and i32 %A, 8388480                      ; <i32> [#uses=1]
   %and2 = and i32 %B, -8388481                    ; <i32> [#uses=1]
   %or = or i32 %and2, %and                        ; <i32> [#uses=1]
@@ -42,8 +41,8 @@
 ; rdar://8752056
 define i32 @f4(i32 %a) nounwind {
 ; CHECK: f4
-; CHECK: movw r1, #3137
-; CHECK: bfi r1, r0, #15, #5
+; CHECK: movw [[R1:r[0-9]+]], #3137
+; CHECK: bfi [[R1]], {{r[0-9]+}}, #15, #5
   %1 = shl i32 %a, 15
   %ins7 = and i32 %1, 1015808
   %ins12 = or i32 %ins7, 3137

Modified: llvm/trunk/test/CodeGen/ARM/fcopysign.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/fcopysign.ll?rev=130800&r1=130799&r2=130800&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/fcopysign.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/fcopysign.ll Tue May  3 17:31:21 2011
@@ -10,7 +10,7 @@
 
 ; HARD: test1:
 ; HARD: vmov.i32 [[REG1:(d[0-9]+)]], #0x80000000
-; HARD: vbsl [[REG1]], d2, d0
+; HARD: vbsl [[REG1]], d
   %0 = tail call float @copysignf(float %x, float %y) nounwind
   ret float %0
 }

Modified: llvm/trunk/test/CodeGen/ARM/ldst-f32-2-i32.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/ldst-f32-2-i32.ll?rev=130800&r1=130799&r2=130800&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/ldst-f32-2-i32.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/ldst-f32-2-i32.ll Tue May  3 17:31:21 2011
@@ -10,8 +10,8 @@
   br i1 %0, label %return, label %bb
 
 bb:
-; CHECK: ldr [[REGISTER:(r[0-9]+)]], [r1], r3
-; CHECK: str [[REGISTER]], [r2], #4
+; CHECK: ldr [[REGISTER:(r[0-9]+)]], [{{r[0-9]+}}], {{r[0-9]+}}
+; CHECK: str [[REGISTER]], [{{r[0-9]+}}], #4
   %j.05 = phi i32 [ %2, %bb ], [ 0, %entry ]
   %tmp = mul i32 %j.05, %index
   %uglygep = getelementptr i8* %src6, i32 %tmp

Modified: llvm/trunk/test/CodeGen/ARM/select-imm.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/select-imm.ll?rev=130800&r1=130799&r2=130800&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/select-imm.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/select-imm.ll Tue May  3 17:31:21 2011
@@ -5,8 +5,8 @@
 define i32 @t1(i32 %c) nounwind readnone {
 entry:
 ; ARM: t1:
-; ARM: mov r1, #101
-; ARM: orr r1, r1, #1, #24
+; ARM: mov [[R1:r[0-9]+]], #101
+; ARM: orr [[R1b:r[0-9]+]], [[R1]], #1, #24
 ; ARM: movgt r0, #123
 
 ; ARMT2: t1:
@@ -34,7 +34,7 @@
 ; ARMT2: movwgt r0, #357
 
 ; THUMB2: t2:
-; THUMB2: mov.w r0, #123
+; THUMB2: mov{{(s|\.w)}} r0, #123
 ; THUMB2: movwgt r0, #357
 
   %0 = icmp sgt i32 %c, 1
@@ -53,7 +53,7 @@
 ; ARMT2: moveq r0, #1
 
 ; THUMB2: t3:
-; THUMB2: mov.w r0, #0
+; THUMB2: mov{{(s|\.w)}} r0, #0
 ; THUMB2: moveq r0, #1
   %0 = icmp eq i32 %a, 160
   %1 = zext i1 %0 to i32
@@ -67,11 +67,11 @@
 ; ARM: movlt
 
 ; ARMT2: t4:
-; ARMT2: movwlt r0, #65365
-; ARMT2: movtlt r0, #65365
+; ARMT2: movwlt [[R0:r[0-9]+]], #65365
+; ARMT2: movtlt [[R0]], #65365
 
 ; THUMB2: t4:
-; THUMB2: mvnlt.w r0, #11141290
+; THUMB2: mvnlt.w [[R0:r[0-9]+]], #11141290
   %0 = icmp slt i32 %a, %b
   %1 = select i1 %0, i32 4283826005, i32 %x
   ret i32 %1

Modified: llvm/trunk/test/CodeGen/ARM/stm.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/stm.ll?rev=130800&r1=130799&r2=130800&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/stm.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/stm.ll Tue May  3 17:31:21 2011
@@ -9,7 +9,7 @@
 entry:
 ; CHECK: main
 ; CHECK: push
-; CHECK: stmib
+; CHECK: stm
 	%0 = tail call i32 (i8*, ...)* @printf(i8* getelementptr ([26 x i8]* @"\01LC1", i32 0, i32 0), i32 -2, i32 -3, i32 2, i32 -6) nounwind		; <i32> [#uses=0]
 	%1 = tail call i32 (i8*, ...)* @printf(i8* getelementptr ([32 x i8]* @"\01LC", i32 0, i32 0), i32 0, i32 1, i32 0, i32 1, i32 0, i32 1) nounwind		; <i32> [#uses=0]
 	ret i32 0

Modified: llvm/trunk/test/CodeGen/ARM/vldlane.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/vldlane.ll?rev=130800&r1=130799&r2=130800&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/vldlane.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/vldlane.ll Tue May  3 17:31:21 2011
@@ -125,7 +125,7 @@
 ;Check for a post-increment updating load.
 define <2 x i32> @vld2lanei32_update(i32** %ptr, <2 x i32>* %B) nounwind {
 ;CHECK: vld2lanei32_update:
-;CHECK: vld2.32 {d16[1], d17[1]}, [r1]!
+;CHECK: vld2.32 {d16[1], d17[1]}, [{{r[0-9]+}}]!
 	%A = load i32** %ptr
 	%tmp0 = bitcast i32* %A to i8*
 	%tmp1 = load <2 x i32>* %B
@@ -153,7 +153,7 @@
 define <8 x i16> @vld2laneQi16(i16* %A, <8 x i16>* %B) nounwind {
 ;CHECK: vld2laneQi16:
 ;Check the (default) alignment.
-;CHECK: vld2.16 {d17[1], d19[1]}, [r0]
+;CHECK: vld2.16 {d17[1], d19[1]}, [{{r[0-9]+}}]
 	%tmp0 = bitcast i16* %A to i8*
 	%tmp1 = load <8 x i16>* %B
 	%tmp2 = call %struct.__neon_int16x8x2_t @llvm.arm.neon.vld2lane.v8i16(i8* %tmp0, <8 x i16> %tmp1, <8 x i16> %tmp1, i32 5, i32 1)
@@ -166,7 +166,7 @@
 define <4 x i32> @vld2laneQi32(i32* %A, <4 x i32>* %B) nounwind {
 ;CHECK: vld2laneQi32:
 ;Check the alignment value.  Max for this instruction is 64 bits:
-;CHECK: vld2.32 {d17[0], d19[0]}, [r0, :64]
+;CHECK: vld2.32 {d17[0], d19[0]}, [{{r[0-9]+}}, :64]
 	%tmp0 = bitcast i32* %A to i8*
 	%tmp1 = load <4 x i32>* %B
 	%tmp2 = call %struct.__neon_int32x4x2_t @llvm.arm.neon.vld2lane.v4i32(i8* %tmp0, <4 x i32> %tmp1, <4 x i32> %tmp1, i32 2, i32 16)
@@ -222,7 +222,7 @@
 define <4 x i16> @vld3lanei16(i16* %A, <4 x i16>* %B) nounwind {
 ;CHECK: vld3lanei16:
 ;Check the (default) alignment value.  VLD3 does not support alignment.
-;CHECK: vld3.16 {d16[1], d17[1], d18[1]}, [r0]
+;CHECK: vld3.16 {d16[1], d17[1], d18[1]}, [{{r[0-9]+}}]
 	%tmp0 = bitcast i16* %A to i8*
 	%tmp1 = load <4 x i16>* %B
 	%tmp2 = call %struct.__neon_int16x4x3_t @llvm.arm.neon.vld3lane.v4i16(i8* %tmp0, <4 x i16> %tmp1, <4 x i16> %tmp1, <4 x i16> %tmp1, i32 1, i32 8)
@@ -265,7 +265,7 @@
 define <8 x i16> @vld3laneQi16(i16* %A, <8 x i16>* %B) nounwind {
 ;CHECK: vld3laneQi16:
 ;Check the (default) alignment value.  VLD3 does not support alignment.
-;CHECK: vld3.16 {d16[1], d18[1], d20[1]}, [r0]
+;CHECK: vld3.16 {d16[1], d18[1], d20[1]}, [{{r[0-9]+}}]
 	%tmp0 = bitcast i16* %A to i8*
 	%tmp1 = load <8 x i16>* %B
 	%tmp2 = call %struct.__neon_int16x8x3_t @llvm.arm.neon.vld3lane.v8i16(i8* %tmp0, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1, i32 1, i32 8)
@@ -344,7 +344,7 @@
 define <8 x i8> @vld4lanei8(i8* %A, <8 x i8>* %B) nounwind {
 ;CHECK: vld4lanei8:
 ;Check the alignment value.  Max for this instruction is 32 bits:
-;CHECK: vld4.8 {d16[1], d17[1], d18[1], d19[1]}, [r0, :32]
+;CHECK: vld4.8 {d16[1], d17[1], d18[1], d19[1]}, [{{r[0-9]+}}, :32]
 	%tmp1 = load <8 x i8>* %B
 	%tmp2 = call %struct.__neon_int8x8x4_t @llvm.arm.neon.vld4lane.v8i8(i8* %A, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, i32 1, i32 8)
         %tmp3 = extractvalue %struct.__neon_int8x8x4_t %tmp2, 0
@@ -360,7 +360,7 @@
 ;Check for a post-increment updating load.
 define <8 x i8> @vld4lanei8_update(i8** %ptr, <8 x i8>* %B) nounwind {
 ;CHECK: vld4lanei8_update:
-;CHECK: vld4.8 {d16[1], d17[1], d18[1], d19[1]}, [r1, :32]!
+;CHECK: vld4.8 {d16[1], d17[1], d18[1], d19[1]}, [{{r[0-9]+}}, :32]!
 	%A = load i8** %ptr
 	%tmp1 = load <8 x i8>* %B
 	%tmp2 = call %struct.__neon_int8x8x4_t @llvm.arm.neon.vld4lane.v8i8(i8* %A, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, <8 x i8> %tmp1, i32 1, i32 8)
@@ -380,7 +380,7 @@
 ;CHECK: vld4lanei16:
 ;Check that a power-of-two alignment smaller than the total size of the memory
 ;being loaded is ignored.
-;CHECK: vld4.16 {d16[1], d17[1], d18[1], d19[1]}, [r0]
+;CHECK: vld4.16 {d16[1], d17[1], d18[1], d19[1]}, [{{r[0-9]+}}]
 	%tmp0 = bitcast i16* %A to i8*
 	%tmp1 = load <4 x i16>* %B
 	%tmp2 = call %struct.__neon_int16x4x4_t @llvm.arm.neon.vld4lane.v4i16(i8* %tmp0, <4 x i16> %tmp1, <4 x i16> %tmp1, <4 x i16> %tmp1, <4 x i16> %tmp1, i32 1, i32 4)
@@ -398,7 +398,7 @@
 ;CHECK: vld4lanei32:
 ;Check the alignment value.  An 8-byte alignment is allowed here even though
 ;it is smaller than the total size of the memory being loaded.
-;CHECK: vld4.32 {d16[1], d17[1], d18[1], d19[1]}, [r0, :64]
+;CHECK: vld4.32 {d16[1], d17[1], d18[1], d19[1]}, [{{r[0-9]+}}, :64]
 	%tmp0 = bitcast i32* %A to i8*
 	%tmp1 = load <2 x i32>* %B
 	%tmp2 = call %struct.__neon_int32x2x4_t @llvm.arm.neon.vld4lane.v2i32(i8* %tmp0, <2 x i32> %tmp1, <2 x i32> %tmp1, <2 x i32> %tmp1, <2 x i32> %tmp1, i32 1, i32 8)
@@ -431,7 +431,7 @@
 define <8 x i16> @vld4laneQi16(i16* %A, <8 x i16>* %B) nounwind {
 ;CHECK: vld4laneQi16:
 ;Check the alignment value.  Max for this instruction is 64 bits:
-;CHECK: vld4.16 {d16[1], d18[1], d20[1], d22[1]}, [r0, :64]
+;CHECK: vld4.16 {d16[1], d18[1], d20[1], d22[1]}, [{{r[0-9]+}}, :64]
 	%tmp0 = bitcast i16* %A to i8*
 	%tmp1 = load <8 x i16>* %B
 	%tmp2 = call %struct.__neon_int16x8x4_t @llvm.arm.neon.vld4lane.v8i16(i8* %tmp0, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1, <8 x i16> %tmp1, i32 1, i32 16)
@@ -448,7 +448,7 @@
 define <4 x i32> @vld4laneQi32(i32* %A, <4 x i32>* %B) nounwind {
 ;CHECK: vld4laneQi32:
 ;Check the (default) alignment.
-;CHECK: vld4.32 {d17[0], d19[0], d21[0], d23[0]}, [r0]
+;CHECK: vld4.32 {d17[0], d19[0], d21[0], d23[0]}, [{{r[0-9]+}}]
 	%tmp0 = bitcast i32* %A to i8*
 	%tmp1 = load <4 x i32>* %B
 	%tmp2 = call %struct.__neon_int32x4x4_t @llvm.arm.neon.vld4lane.v4i32(i8* %tmp0, <4 x i32> %tmp1, <4 x i32> %tmp1, <4 x i32> %tmp1, <4 x i32> %tmp1, i32 2, i32 1)



From stoklund at 2pi.dk  Tue May  3 17:31:24 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Tue, 03 May 2011 22:31:24 -0000
Subject: [llvm-commits] [llvm] r130801 - in /llvm/trunk/lib/Target/ARM:
 ARMInstrInfo.td ARMInstrThumb.td ARMInstrThumb2.td
Message-ID: <20110503223124.EF5252A6C12D@llvm.org>

Author: stoklund
Date: Tue May  3 17:31:24 2011
New Revision: 130801

URL: http://llvm.org/viewvc/llvm-project?rev=130801&view=rev
Log:
Mark ultra-super-registers QQQQ as call-clobbered instead of the D sub-registers.

LiveVariables doesn't understand that clobbering D0 and D1 completely overwrites
Q0, so if Q0 is live-in to a function, its live range will extend beyond a
function call that only clobbers D0 and D1. This shows up in the
ARM/2009-11-01-NeonMoves test case.

LiveVariables should probably implement the much stricter rules for physreg
liveness that RAFast imposes - a physreg is killed by the first use of any
alias.

Modified:
    llvm/trunk/lib/Target/ARM/ARMInstrInfo.td
    llvm/trunk/lib/Target/ARM/ARMInstrThumb.td
    llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td

Modified: llvm/trunk/lib/Target/ARM/ARMInstrInfo.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrInfo.td?rev=130801&r1=130800&r2=130801&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrInfo.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrInfo.td Tue May  3 17:31:24 2011
@@ -1321,10 +1321,7 @@
   // FIXME:  Do we really need a non-predicated version? If so, it should
   // at least be a pseudo instruction expanding to the predicated version
   // at MC lowering time.
-  Defs = [R0,  R1,  R2,  R3,  R12, LR,
-          D0,  D1,  D2,  D3,  D4,  D5,  D6,  D7,
-          D16, D17, D18, D19, D20, D21, D22, D23,
-          D24, D25, D26, D27, D28, D29, D30, D31, CPSR, FPSCR],
+  Defs = [R0,  R1,  R2,  R3,  R12, LR, QQQQ0, QQQQ2, QQQQ3, CPSR, FPSCR],
   Uses = [SP] in {
   def BL  : ABXI<0b1011, (outs), (ins bl_target:$func, variable_ops),
                 IIC_Br, "bl\t$func",
@@ -1378,10 +1375,7 @@
   // On Darwin R9 is call-clobbered.
   // R7 is marked as a use to prevent frame-pointer assignments from being
   // moved above / below calls.
-  Defs = [R0,  R1,  R2,  R3,  R9,  R12, LR,
-          D0,  D1,  D2,  D3,  D4,  D5,  D6,  D7,
-          D16, D17, D18, D19, D20, D21, D22, D23,
-          D24, D25, D26, D27, D28, D29, D30, D31, CPSR, FPSCR],
+  Defs = [R0,  R1,  R2,  R3,  R9,  R12, LR, QQQQ0, QQQQ2, QQQQ3, CPSR, FPSCR],
   Uses = [R7, SP] in {
   def BLr9  : ARMPseudoInst<(outs), (ins bltarget:$func, variable_ops),
                 Size4Bytes, IIC_Br,
@@ -1420,10 +1414,7 @@
 // FIXME: The Thumb versions of these should live in ARMInstrThumb.td
 let isCall = 1, isTerminator = 1, isReturn = 1, isBarrier = 1 in {
   // Darwin versions.
-  let Defs = [R0, R1, R2, R3, R9, R12,
-              D0, D1, D2, D3, D4, D5, D6, D7,
-              D16, D17, D18, D19, D20, D21, D22, D23, D24, D25, D26,
-              D27, D28, D29, D30, D31, PC],
+  let Defs = [R0, R1, R2, R3, R9, R12, QQQQ0, QQQQ2, QQQQ3, PC],
       Uses = [SP] in {
     def TCRETURNdi : PseudoInst<(outs), (ins i32imm:$dst, variable_ops),
                        IIC_Br, []>, Requires<[IsDarwin]>;
@@ -1449,10 +1440,7 @@
   }
 
   // Non-Darwin versions (the difference is R9).
-  let Defs = [R0, R1, R2, R3, R12,
-              D0, D1, D2, D3, D4, D5, D6, D7,
-              D16, D17, D18, D19, D20, D21, D22, D23, D24, D25, D26,
-              D27, D28, D29, D30, D31, PC],
+  let Defs = [R0, R1, R2, R3, R12, QQQQ0, QQQQ2, QQQQ3, PC],
       Uses = [SP] in {
     def TCRETURNdiND : PseudoInst<(outs), (ins i32imm:$dst, variable_ops),
                        IIC_Br, []>, Requires<[IsNotDarwin]>;
@@ -3757,10 +3745,8 @@
 // These are pseudo-instructions and are lowered to individual MC-insts, so
 // no encoding information is necessary.
 let Defs =
-  [ R0,  R1,  R2,  R3,  R4,  R5,  R6,  R7,  R8,  R9,  R10, R11, R12, LR,  D0,
-    D1,  D2,  D3,  D4,  D5,  D6,  D7,  D8,  D9,  D10, D11, D12, D13, D14, D15,
-    D16, D17, D18, D19, D20, D21, D22, D23, D24, D25, D26, D27, D28, D29, D30,
-    D31 ], hasSideEffects = 1, isBarrier = 1 in {
+  [ R0,  R1,  R2,  R3,  R4,  R5,  R6,  R7,  R8,  R9,  R10, R11, R12, LR,
+    QQQQ0, QQQQ1, QQQQ2, QQQQ3 ], hasSideEffects = 1, isBarrier = 1 in {
   def Int_eh_sjlj_setjmp : PseudoInst<(outs), (ins GPR:$src, GPR:$val),
                                NoItinerary,
                          [(set R0, (ARMeh_sjlj_setjmp GPR:$src, GPR:$val))]>,

Modified: llvm/trunk/lib/Target/ARM/ARMInstrThumb.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrThumb.td?rev=130801&r1=130800&r2=130801&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrThumb.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrThumb.td Tue May  3 17:31:24 2011
@@ -423,10 +423,7 @@
 // potentially appearing dead.
 let isCall = 1,
   // On non-Darwin platforms R9 is callee-saved.
-  Defs = [R0,  R1,  R2,  R3,  R12, LR,
-          D0,  D1,  D2,  D3,  D4,  D5,  D6,  D7,
-          D16, D17, D18, D19, D20, D21, D22, D23,
-          D24, D25, D26, D27, D28, D29, D30, D31, CPSR, FPSCR],
+  Defs = [R0,  R1,  R2,  R3,  R12, LR, QQQQ0, QQQQ2, QQQQ3, CPSR, FPSCR],
   Uses = [SP] in {
   // Also used for Thumb2
   def tBL  : TIx2<0b11110, 0b11, 1,
@@ -476,10 +473,7 @@
   // On Darwin R9 is call-clobbered.
   // R7 is marked as a use to prevent frame-pointer assignments from being
   // moved above / below calls.
-  Defs = [R0,  R1,  R2,  R3,  R9,  R12, LR,
-          D0,  D1,  D2,  D3,  D4,  D5,  D6,  D7,
-          D16, D17, D18, D19, D20, D21, D22, D23,
-          D24, D25, D26, D27, D28, D29, D30, D31, CPSR, FPSCR],
+  Defs = [R0,  R1,  R2,  R3,  R9,  R12, LR, QQQQ0, QQQQ2, QQQQ3, CPSR, FPSCR],
   Uses = [R7, SP] in {
   // Also used for Thumb2
   def tBLr9 : TIx2<0b11110, 0b11, 1,

Modified: llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td?rev=130801&r1=130800&r2=130801&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td (original)
+++ llvm/trunk/lib/Target/ARM/ARMInstrThumb2.td Tue May  3 17:31:24 2011
@@ -2965,10 +2965,9 @@
 //   all of the callee-saved resgisters, which is exactly what we want.
 //   $val is a scratch register for our use.
 let Defs =
-  [ R0,  R1,  R2,  R3,  R4,  R5,  R6,  R7,  R8,  R9,  R10, R11, R12, LR,  D0,
-    D1,  D2,  D3,  D4,  D5,  D6,  D7,  D8,  D9,  D10, D11, D12, D13, D14, D15,
-    D16, D17, D18, D19, D20, D21, D22, D23, D24, D25, D26, D27, D28, D29, D30,
-    D31 ], hasSideEffects = 1, isBarrier = 1, isCodeGenOnly = 1 in {
+  [ R0,  R1,  R2,  R3,  R4,  R5,  R6,  R7,  R8,  R9,  R10, R11, R12, LR,
+    QQQQ0, QQQQ1, QQQQ2, QQQQ3 ],
+  hasSideEffects = 1, isBarrier = 1, isCodeGenOnly = 1 in {
   def t2Int_eh_sjlj_setjmp : Thumb2XI<(outs), (ins tGPR:$src, tGPR:$val),
                                AddrModeNone, SizeSpecial, NoItinerary, "", "",
                           [(set R0, (ARMeh_sjlj_setjmp tGPR:$src, tGPR:$val))]>,



From pichet2000 at gmail.com  Tue May  3 17:39:46 2011
From: pichet2000 at gmail.com (Francois Pichet)
Date: Tue, 3 May 2011 18:39:46 -0400
Subject: [llvm-commits] [llvm] r130793 - /llvm/trunk/lib/MC/MCDwarf.cpp
In-Reply-To: <20110503213337.D7F902A6C12C@llvm.org>
References: <20110503213337.D7F902A6C12C@llvm.org>
Message-ID: <BANLkTimZk1eSxMYLgnxOpmLQY=+2EVyj-Q@mail.gmail.com>

On Tue, May 3, 2011 at 5:33 PM, Daniel Dunbar <daniel at zuster.org> wrote:
> Author: ddunbar
> Date: Tue May ?3 16:33:37 2011
> New Revision: 130793
>
> URL: http://llvm.org/viewvc/llvm-project?rev=130793&view=rev
> Log:
> MCDwarf: Don't save Twine to local variable, this is almost never safe to do
> (and should thus never be done).
> ?- Should fix a crash on win32.

Yes crash is gone. great


From ahatanak at gmail.com  Tue May  3 18:18:12 2011
From: ahatanak at gmail.com (Akira Hatanaka)
Date: Tue, 3 May 2011 16:18:12 -0700
Subject: [llvm-commits] [PATCH] Mips backend. Emit $gp restore instructions.
Message-ID: <BANLkTinnyohuVJ9a+JwZXkEURxJFNb5e8Q@mail.gmail.com>

This patch prevents instructions using $gp from being placed between a jalr
and the instruction that restores the clobbered $gp (please see the
following example).

The current implementation creates load nodes that restore $gp when call
instructions are lowered in MipsISelLowering::LowerCall. However, it does
not add chains or glues that are needed to make sure load/store nodes that
have $gp as an operand are not scheduled between JALR nodes and nodes that
restore $gp after the call.

This patch runs a pre-regalloc pass that emits $gp restore instructions
instead.

- bitcode
@p = external global i32
...
tail call void (...)* @f1() nounwind
%tmp = load i32* @p, align 4, !tbaa !0

- currently produces following code:
...
lw  $25, %call16(f1)($gp)
jalr  $25                        # calls f1. $gp is clobbered.
nop
lw  $3, 16($sp)              # load $gp from stack.
lw  $2, %got(p)($gp)      # this instr uses a clobbered $gp.
addu  $gp, $zero, $3     # $gp is restored here.
...

- produces following code after patch is applied.
...
lw  $25, %call16(f1)($gp)
lw  $16, %got(p)($gp)  # $gp has correct value here.
jalr  $25                     # calls f1. clobbers $gp.
nop
lw  $gp, 16($sp)          # $gp restored.
...
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110503/a7102878/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gprestore.patch
Type: text/x-patch
Size: 4971 bytes
Desc: not available
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110503/a7102878/attachment.bin 

From clattner at apple.com  Tue May  3 19:46:31 2011
From: clattner at apple.com (Chris Lattner)
Date: Tue, 03 May 2011 17:46:31 -0700
Subject: [llvm-commits] [llvm] r130485 - in /llvm/trunk:
 lib/Transforms/InstCombine/InstCombineShifts.cpp
 test/Transforms/InstCombine/shift.ll
In-Reply-To: <2E25739D-6707-4365-9DC5-E22C7C8E9644@apple.com>
References: <20110429081541.B7F422A6C138@llvm.org>
	<C8DE37BF-B56C-4F53-A49F-EA158B233245@apple.com>
	<CB365302-F1CC-4578-BA54-7D954227E750@googlemail.com>
	<1B93035B-15A4-47B0-8E78-2D0F8A6B1701@apple.com>
	<5F0816C4-4FC4-4F02-8B7E-51858107AC6D@googlemail.com>
	<C9E399F9-DA42-4972-9BD4-130A27B3D1D5@apple.com>
	<E3078F91-A806-4AEE-8F70-448C10F1B4C6@apple.com>
	<2E25739D-6707-4365-9DC5-E22C7C8E9644@apple.com>
Message-ID: <1A474230-04EB-4BA5-9B1C-A14B6117C315@apple.com>


On May 2, 2011, at 2:33 PM, Devang Patel wrote:

> Evan,
> 
> On May 2, 2011, at 2:00 PM, Evan Cheng wrote:
> 
>>> More generally, we don't want to sprinkle setDebugLoc calls into *every* instcombine xform.  We need a better solution.
>> 
>> I believe there is a long term plan being discussed. Devang, can you share what's being discussed so far?
> 
> As I understand, the long term plan being discussed will not help here. Instead, the plan assumes that all instructions and machine instruction will have DebugLoc. So in other words, the plan requires that we solve this particular issue first.

Ok.  Instcombine has an IRBuilder that it uses, changing xforms to use the IRBuilder is vastly preferable to using setDebugLoc everywhere.

> Without further delay, the plan being discussed seeks to find easier way to preserve debug values during code gen passes.
> 	- The idea is to make position of DBG_VALUE machine instructions in instruction stream irrelevant by using a new mechanism (we have not detailed what exactly, but it could be just another intrinsic or metadata pair or something else...) to pair a LLVM definition to respective source level variable information.  
> 	- At the end of compilation just before emitting out put, a super intelligent pass is employed to 
> 		a) look at variable info, 
> 		b) find corresponding value definitions, 
> 		c) find scopes and ranges of these definitions based on line number info attached with each machine instruction 
> 		d) construct complete debug info picture based on these information.
> 
> I am sure, others will correct me if I am missing something.

It's hard to really comment on this without more details, but as long as there aren't tons of new setDebugLoc calls in InstCombine, I'll be happy :)

-Chris

From echristo at apple.com  Tue May  3 19:48:02 2011
From: echristo at apple.com (Eric Christopher)
Date: Wed, 04 May 2011 00:48:02 -0000
Subject: [llvm-commits] [llvm] r130812 -
	/llvm/trunk/test/MC/X86/x86-32-coverage.s
Message-ID: <20110504004803.20F272A6C12C@llvm.org>

Author: echristo
Date: Tue May  3 19:48:02 2011
New Revision: 130812

URL: http://llvm.org/viewvc/llvm-project?rev=130812&view=rev
Log:
Remove some random comments that snuck in from somewhere.

Modified:
    llvm/trunk/test/MC/X86/x86-32-coverage.s

Modified: llvm/trunk/test/MC/X86/x86-32-coverage.s
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/MC/X86/x86-32-coverage.s?rev=130812&r1=130811&r2=130812&view=diff
==============================================================================
--- llvm/trunk/test/MC/X86/x86-32-coverage.s (original)
+++ llvm/trunk/test/MC/X86/x86-32-coverage.s Tue May  3 19:48:02 2011
@@ -1,7 +1,3 @@
-# Build output for (null)
-# Generated at (null)
-# Using (null) configuration, (null) architecture for (null) target of (null) project
-
 // RUN: llvm-mc -triple i386-unknown-unknown %s --show-encoding  | FileCheck %s
 
 // CHECK: 	movb	$127, 3735928559(%ebx,%ecx,8)



From stoklund at 2pi.dk  Tue May  3 20:01:36 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 01:01:36 -0000
Subject: [llvm-commits] [llvm] r130814 -
	/llvm/trunk/lib/Target/MSP430/MSP430RegisterInfo.h
Message-ID: <20110504010136.D21662A6C12C@llvm.org>

Author: stoklund
Date: Tue May  3 20:01:36 2011
New Revision: 130814

URL: http://llvm.org/viewvc/llvm-project?rev=130814&view=rev
Log:
Implement MSP430RegisterInfo::getMatchingSuperRegClass to enable cross-class
coalescing.

Modified:
    llvm/trunk/lib/Target/MSP430/MSP430RegisterInfo.h

Modified: llvm/trunk/lib/Target/MSP430/MSP430RegisterInfo.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/MSP430/MSP430RegisterInfo.h?rev=130814&r1=130813&r2=130814&view=diff
==============================================================================
--- llvm/trunk/lib/Target/MSP430/MSP430RegisterInfo.h (original)
+++ llvm/trunk/lib/Target/MSP430/MSP430RegisterInfo.h Tue May  3 20:01:36 2011
@@ -39,6 +39,13 @@
   BitVector getReservedRegs(const MachineFunction &MF) const;
   const TargetRegisterClass* getPointerRegClass(unsigned Kind = 0) const;
 
+  const TargetRegisterClass *
+  getMatchingSuperRegClass(const TargetRegisterClass *A,
+                           const TargetRegisterClass *B, unsigned Idx) const {
+    // No sub-classes makes this really easy.
+    return A;
+  }
+
   void eliminateCallFramePseudoInstr(MachineFunction &MF,
                                      MachineBasicBlock &MBB,
                                      MachineBasicBlock::iterator I) const;



From stoklund at 2pi.dk  Tue May  3 20:01:39 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 01:01:39 -0000
Subject: [llvm-commits] [llvm] r130815 -
	/llvm/trunk/test/CodeGen/MSP430/Inst8rr.ll
Message-ID: <20110504010139.75E6C2A6C12D@llvm.org>

Author: stoklund
Date: Tue May  3 20:01:39 2011
New Revision: 130815

URL: http://llvm.org/viewvc/llvm-project?rev=130815&view=rev
Log:
Fix register-dependent test in MSP430.

Modified:
    llvm/trunk/test/CodeGen/MSP430/Inst8rr.ll

Modified: llvm/trunk/test/CodeGen/MSP430/Inst8rr.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/MSP430/Inst8rr.ll?rev=130815&r1=130814&r2=130815&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/MSP430/Inst8rr.ll (original)
+++ llvm/trunk/test/CodeGen/MSP430/Inst8rr.ll Tue May  3 20:01:39 2011
@@ -10,7 +10,7 @@
 
 define i8 @add(i8 %a, i8 %b) nounwind {
 ; CHECK: add:
-; CHECK: add.b	r12, r15
+; CHECK: add.b
 	%1 = add i8 %a, %b
 	ret i8 %1
 }



From stoklund at 2pi.dk  Tue May  3 20:01:41 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 01:01:41 -0000
Subject: [llvm-commits] [llvm] r130816 -
	/llvm/trunk/test/CodeGen/XCore/mul64.ll
Message-ID: <20110504010141.D9F6C2A6C12E@llvm.org>

Author: stoklund
Date: Tue May  3 20:01:41 2011
New Revision: 130816

URL: http://llvm.org/viewvc/llvm-project?rev=130816&view=rev
Log:
Fix register-dependent XCore tests

Modified:
    llvm/trunk/test/CodeGen/XCore/mul64.ll

Modified: llvm/trunk/test/CodeGen/XCore/mul64.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/XCore/mul64.ll?rev=130816&r1=130815&r2=130816&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/XCore/mul64.ll (original)
+++ llvm/trunk/test/CodeGen/XCore/mul64.ll Tue May  3 20:01:41 2011
@@ -9,7 +9,7 @@
 }
 ; CHECK: umul_lohi:
 ; CHECK: ldc [[REG:r[0-9]+]], 0
-; CHECK-NEXT: lmul r1, r0, r1, r0, [[REG]], [[REG]]
+; CHECK-NEXT: lmul {{.*}}, [[REG]], [[REG]]
 ; CHECK-NEXT: retsp 0
 
 define i64 @smul_lohi(i32 %a, i32 %b) {
@@ -23,9 +23,7 @@
 ; CHECK: ldc
 ; CHECK-NEXT: mov
 ; CHECK-NEXT: maccs
-; CHECK-NEXT: mov r0,
-; CHECK-NEXT: mov r1,
-; CHECK-NEXT: retsp 0
+; CHECK: retsp 0
 
 define i64 @mul64(i64 %a, i64 %b) {
 entry:
@@ -37,7 +35,6 @@
 ; CHECK-NEXT: lmul
 ; CHECK-NEXT: mul
 ; CHECK-NEXT: lmul
-; CHECK-NEXT: mov r0,
 
 define i64 @mul64_2(i64 %a, i32 %b) {
 entry:
@@ -50,4 +47,4 @@
 ; CHECK-NEXT: lmul
 ; CHECK-NEXT: mul
 ; CHECK-NEXT: add r1,
-; CHECK-NEXT: retsp 0
+; CHECK: retsp 0



From stoklund at 2pi.dk  Tue May  3 20:01:44 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 01:01:44 -0000
Subject: [llvm-commits] [llvm] r130817 -
	/llvm/trunk/test/CodeGen/Mips/o32_cc_vararg.ll
Message-ID: <20110504010144.70AF22A6C12C@llvm.org>

Author: stoklund
Date: Tue May  3 20:01:44 2011
New Revision: 130817

URL: http://llvm.org/viewvc/llvm-project?rev=130817&view=rev
Log:
Don't run this test through -regalloc=basic.

The basic allocator is really bad about hinting, so it doesn't eliminate all
copies when physreg joining is disabled.

Modified:
    llvm/trunk/test/CodeGen/Mips/o32_cc_vararg.ll

Modified: llvm/trunk/test/CodeGen/Mips/o32_cc_vararg.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Mips/o32_cc_vararg.ll?rev=130817&r1=130816&r2=130817&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Mips/o32_cc_vararg.ll (original)
+++ llvm/trunk/test/CodeGen/Mips/o32_cc_vararg.ll Tue May  3 20:01:44 2011
@@ -1,5 +1,4 @@
 ; RUN: llc -march=mipsel -mcpu=mips2 -pre-RA-sched=source < %s | FileCheck %s
-; RUN: llc -march=mipsel -mcpu=mips2 -pre-RA-sched=source < %s -regalloc=basic | FileCheck %s
 
 
 ; All test functions do the same thing - they return the first variable



From stoklund at 2pi.dk  Tue May  3 20:01:47 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 01:01:47 -0000
Subject: [llvm-commits] [llvm] r130818 - in /llvm/trunk/test/CodeGen/X86:
 x86-64-extend-shift.ll xor.ll
Message-ID: <20110504010147.42DA62A6C12D@llvm.org>

Author: stoklund
Date: Tue May  3 20:01:47 2011
New Revision: 130818

URL: http://llvm.org/viewvc/llvm-project?rev=130818&view=rev
Log:
Don't depend on the physreg coalescing order.

Modified:
    llvm/trunk/test/CodeGen/X86/x86-64-extend-shift.ll
    llvm/trunk/test/CodeGen/X86/xor.ll

Modified: llvm/trunk/test/CodeGen/X86/x86-64-extend-shift.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/x86-64-extend-shift.ll?rev=130818&r1=130817&r2=130818&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/x86-64-extend-shift.ll (original)
+++ llvm/trunk/test/CodeGen/X86/x86-64-extend-shift.ll Tue May  3 20:01:47 2011
@@ -2,7 +2,7 @@
 ; Formerly there were two shifts.
 
 define i64 @baz(i32 %A) nounwind {
-; CHECK:  shlq  $49, %rax
+; CHECK:  shlq  $49, %r
         %tmp1 = shl i32 %A, 17
         %tmp2 = zext i32 %tmp1 to i64
         %tmp3 = shl i64 %tmp2, 32

Modified: llvm/trunk/test/CodeGen/X86/xor.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/xor.ll?rev=130818&r1=130817&r2=130818&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/xor.ll (original)
+++ llvm/trunk/test/CodeGen/X86/xor.ll Tue May  3 20:01:47 2011
@@ -29,9 +29,8 @@
         ret i32 %tmp4
         
 ; X64: test3:
-; X64:	notl	[[A1:%esi|%edx]]
-; X64:	andl	[[A0:%edi|%ecx]], [[A1]]
-; X64:	movl	[[A1]], %eax
+; X64:	notl
+; X64:	andl
 ; X64:	shrl	%eax
 ; X64:	ret
 
@@ -139,7 +138,7 @@
   %t2 = add i32 %t1, -1
   ret i32 %t2
 ; X64: test8:
-; X64:   notl %eax
+; X64:   notl {{%eax|%edi|%ecx}}
 ; X32: test8:
 ; X32:   notl %eax
 }



From nicholas at mxc.ca  Tue May  3 20:03:02 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Wed, 04 May 2011 01:03:02 -0000
Subject: [llvm-commits] [llvm] r130819 - in /llvm/trunk/utils/valgrind:
 i386-pc-linux-gnu.supp x86_64-pc-linux-gnu.supp
Message-ID: <20110504010302.A8DDD2A6C12C@llvm.org>

Author: nicholas
Date: Tue May  3 20:03:02 2011
New Revision: 130819

URL: http://llvm.org/viewvc/llvm-project?rev=130819&view=rev
Log:
The system suppression file should catch these, but since they *once again* are
not, I'll just add them here and be done with it.

Modified:
    llvm/trunk/utils/valgrind/i386-pc-linux-gnu.supp
    llvm/trunk/utils/valgrind/x86_64-pc-linux-gnu.supp

Modified: llvm/trunk/utils/valgrind/i386-pc-linux-gnu.supp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/utils/valgrind/i386-pc-linux-gnu.supp?rev=130819&r1=130818&r2=130819&view=diff
==============================================================================
--- llvm/trunk/utils/valgrind/i386-pc-linux-gnu.supp (original)
+++ llvm/trunk/utils/valgrind/i386-pc-linux-gnu.supp Tue May  3 20:03:02 2011
@@ -39,3 +39,10 @@
    fun:malloc
    obj:/usr/bin/python*
 }
+
+{
+   We don't care about anything ld.so does.
+   Memcheck:Cond
+   obj:/lib/ld*.so
+}
+

Modified: llvm/trunk/utils/valgrind/x86_64-pc-linux-gnu.supp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/utils/valgrind/x86_64-pc-linux-gnu.supp?rev=130819&r1=130818&r2=130819&view=diff
==============================================================================
--- llvm/trunk/utils/valgrind/x86_64-pc-linux-gnu.supp (original)
+++ llvm/trunk/utils/valgrind/x86_64-pc-linux-gnu.supp Tue May  3 20:03:02 2011
@@ -44,3 +44,10 @@
    fun:malloc
    obj:/usr/bin/python*
 }
+
+{
+   We don't care about anything ld.so does.
+   Memcheck:Cond
+   obj:/lib/ld*.so
+}
+



From srhines at google.com  Tue May  3 20:24:04 2011
From: srhines at google.com (Stephen Hines)
Date: Tue, 3 May 2011 18:24:04 -0700
Subject: [llvm-commits] Fix an encoding issue with ARM PKH* instructions for
	non-MC path
Message-ID: <BANLkTime=_oB=FscVLUykMYaw5gWMfpVpQ@mail.gmail.com>

Hi,

I am sending this patch because I found that the non-MC ARM path was
incorrectly encoding ARM PKH* instructions. They would essentially be
encoded twice, where the second encoding would unwrap the shift amount in a
way that register fields could be clobbered (by the shift amount/encoding).
I do not have commit privileges, so I would appreciate it if someone could
submit this for me. All current tests pass with this change, since I am
using the non-MC path.

Steve
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110503/83aaf4db/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pkh.patch
Type: text/x-patch
Size: 543 bytes
Desc: not available
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110503/83aaf4db/attachment.bin 

From bruno.cardoso at gmail.com  Tue May  3 20:31:24 2011
From: bruno.cardoso at gmail.com (Bruno Cardoso Lopes)
Date: Tue, 3 May 2011 22:31:24 -0300
Subject: [llvm-commits] [PATCH] Mips backend. Emit $gp restore
	instructions.
In-Reply-To: <BANLkTinnyohuVJ9a+JwZXkEURxJFNb5e8Q@mail.gmail.com>
References: <BANLkTinnyohuVJ9a+JwZXkEURxJFNb5e8Q@mail.gmail.com>
Message-ID: <BANLkTimhGzW3_k=sKAMShzbVGQHep+_ayA@mail.gmail.com>

On Tue, May 3, 2011 at 8:18 PM, Akira Hatanaka <ahatanak at gmail.com> wrote:
> This patch prevents instructions using $gp from being placed between a jalr
> and the instruction that restores the clobbered $gp (please see the
> following example).
>
> The current implementation creates load nodes that restore $gp when call
> instructions are lowered in MipsISelLowering::LowerCall. However, it does
> not add chains or glues that are needed to make sure load/store nodes that
> have $gp as an operand are not scheduled between JALR nodes and nodes that
> restore $gp after the call.
>
> This patch runs a pre-regalloc pass that emits $gp restore instructions
> instead.
>
> - bitcode
> @p = external global i32
> ...
> tail call void (...)* @f1() nounwind
> %tmp = load i32* @p, align 4, !tbaa !0
>
> - currently produces following code:
> ...
> lw? $25, %call16(f1)($gp)
> jalr? $25??????????????????????? # calls f1. $gp is clobbered.
> nop
> lw? $3, 16($sp)????????????? # load $gp from stack.
> lw? $2, %got(p)($gp)????? # this instr uses a clobbered $gp.
> addu? $gp, $zero, $3???? # $gp is restored here.
> ...
>
> - produces following code after patch is applied.
> ...
> lw? $25, %call16(f1)($gp)
> lw? $16, %got(p)($gp)? # $gp has correct value here.
> jalr? $25???????????????????? # calls f1. clobbers $gp.
> nop
> lw? $gp, 16($sp)????????? # $gp restored.
> ...

I've tried other approaches to solve this issue but they we're not
clean enough, I can't figure out now another way besides doing a
separate pass. If someone else knows a better a approach we can go for
it, but in the meantime go for it. Please commit.



-- 
Bruno Cardoso Lopes
http://www.brunocardoso.cc


From bruno.cardoso at gmail.com  Tue May  3 20:34:22 2011
From: bruno.cardoso at gmail.com (Bruno Cardoso Lopes)
Date: Tue, 3 May 2011 22:34:22 -0300
Subject: [llvm-commits] Fix an encoding issue with ARM PKH* instructions
 for non-MC path
In-Reply-To: <BANLkTime=_oB=FscVLUykMYaw5gWMfpVpQ@mail.gmail.com>
References: <BANLkTime=_oB=FscVLUykMYaw5gWMfpVpQ@mail.gmail.com>
Message-ID: <BANLkTimGh7E1CGtoyyPYWsK_27v-2dOAJg@mail.gmail.com>

Hi Stephen,

On Tue, May 3, 2011 at 10:24 PM, Stephen Hines <srhines at google.com> wrote:
> Hi,
> I am sending this patch because I found that the non-MC ARM path was
> incorrectly encoding ARM PKH* instructions. They would essentially be
> encoded twice, where the second encoding would unwrap the shift amount in a
> way that register fields could be clobbered (by the shift amount/encoding).
> I do not have commit privileges, so I would appreciate it if someone could
> submit this for me. All current tests pass with this change, since I am
> using the non-MC path.

Nice! Could you add an encoding test to the patch? See test/MC/ARM for examples.

-- 
Bruno Cardoso Lopes
http://www.brunocardoso.cc

From srhines at google.com  Tue May  3 20:35:21 2011
From: srhines at google.com (Stephen Hines)
Date: Tue, 3 May 2011 18:35:21 -0700
Subject: [llvm-commits] Fix an encoding issue with ARM PKH* instructions
 for non-MC path
In-Reply-To: <BANLkTimGh7E1CGtoyyPYWsK_27v-2dOAJg@mail.gmail.com>
References: <BANLkTime=_oB=FscVLUykMYaw5gWMfpVpQ@mail.gmail.com>
	<BANLkTimGh7E1CGtoyyPYWsK_27v-2dOAJg@mail.gmail.com>
Message-ID: <BANLkTimeMJ=g7cdAsN5y0j3G6GwHidnCZg@mail.gmail.com>

The MC path is actually fine. This is for the non-MC path which seems to not
have any tests.

Steve

On Tue, May 3, 2011 at 6:34 PM, Bruno Cardoso Lopes <bruno.cardoso at gmail.com
> wrote:

> Hi Stephen,
>
> On Tue, May 3, 2011 at 10:24 PM, Stephen Hines <srhines at google.com> wrote:
> > Hi,
> > I am sending this patch because I found that the non-MC ARM path was
> > incorrectly encoding ARM PKH* instructions. They would essentially be
> > encoded twice, where the second encoding would unwrap the shift amount in
> a
> > way that register fields could be clobbered (by the shift
> amount/encoding).
> > I do not have commit privileges, so I would appreciate it if someone
> could
> > submit this for me. All current tests pass with this change, since I am
> > using the non-MC path.
>
> Nice! Could you add an encoding test to the patch? See test/MC/ARM for
> examples.
>
> --
> Bruno Cardoso Lopes
> http://www.brunocardoso.cc
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110503/588c21a3/attachment-0001.html 

From srhines at google.com  Tue May  3 20:50:22 2011
From: srhines at google.com (Stephen Hines)
Date: Tue, 3 May 2011 18:50:22 -0700
Subject: [llvm-commits] Fix an encoding issue with ARM PKH* instructions
 for non-MC path
In-Reply-To: <BANLkTimeMJ=g7cdAsN5y0j3G6GwHidnCZg@mail.gmail.com>
References: <BANLkTime=_oB=FscVLUykMYaw5gWMfpVpQ@mail.gmail.com>
	<BANLkTimGh7E1CGtoyyPYWsK_27v-2dOAJg@mail.gmail.com>
	<BANLkTimeMJ=g7cdAsN5y0j3G6GwHidnCZg@mail.gmail.com>
Message-ID: <BANLkTi=VkS5DMWoXAnxk9wayeyXbtgB_xA@mail.gmail.com>

I should be a bit more clear with my answer. The
file test/CodeGen/ARM/pack.ll does contain tests for pkhbt and pkhtb, but
they do not test the actual encoding of the instruction bits. The readable
assembly code is correct, but it is the encoding where the problem shows up.
If there are actual encoding tests for the classic code generation path on
ARM, I would be more than happy to update them. The MC version does
correctly generate the proper encoding (as seen in
test/MC/ARM/simple-encoding.ll).

Steve

On Tue, May 3, 2011 at 6:35 PM, Stephen Hines <srhines at google.com> wrote:

> The MC path is actually fine. This is for the non-MC path which seems to
> not have any tests.
>
> Steve
>
>
> On Tue, May 3, 2011 at 6:34 PM, Bruno Cardoso Lopes <
> bruno.cardoso at gmail.com> wrote:
>
>> Hi Stephen,
>>
>> On Tue, May 3, 2011 at 10:24 PM, Stephen Hines <srhines at google.com>
>> wrote:
>> > Hi,
>> > I am sending this patch because I found that the non-MC ARM path was
>> > incorrectly encoding ARM PKH* instructions. They would essentially be
>> > encoded twice, where the second encoding would unwrap the shift amount
>> in a
>> > way that register fields could be clobbered (by the shift
>> amount/encoding).
>> > I do not have commit privileges, so I would appreciate it if someone
>> could
>> > submit this for me. All current tests pass with this change, since I am
>> > using the non-MC path.
>>
>> Nice! Could you add an encoding test to the patch? See test/MC/ARM for
>> examples.
>>
>> --
>> Bruno Cardoso Lopes
>> http://www.brunocardoso.cc
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110503/bc5b040b/attachment.html 

From daniel at zuster.org  Tue May  3 20:48:36 2011
From: daniel at zuster.org (Daniel Dunbar)
Date: Wed, 04 May 2011 01:48:36 -0000
Subject: [llvm-commits] [zorg] r130825 -
	/zorg/trunk/lnt/lnt/lnttool/__init__.py
Message-ID: <20110504014836.B3A4B2A6C12C@llvm.org>

Author: ddunbar
Date: Tue May  3 20:48:36 2011
New Revision: 130825

URL: http://llvm.org/viewvc/llvm-project?rev=130825&view=rev
Log:
lnt: Add a --version argument.

Modified:
    zorg/trunk/lnt/lnt/lnttool/__init__.py

Modified: zorg/trunk/lnt/lnt/lnttool/__init__.py
URL: http://llvm.org/viewvc/llvm-project/zorg/trunk/lnt/lnt/lnttool/__init__.py?rev=130825&r1=130824&r2=130825&view=diff
==============================================================================
--- zorg/trunk/lnt/lnt/lnttool/__init__.py (original)
+++ zorg/trunk/lnt/lnt/lnttool/__init__.py Tue May  3 20:48:36 2011
@@ -5,6 +5,7 @@
 from optparse import OptionParser, OptionGroup
 
 import StringIO
+import lnt
 from lnt import testing
 from lnt.db import perfdb
 
@@ -401,28 +402,27 @@
 
 commands = dict((name[7:], f) for name,f in locals().items()
                 if name.startswith('action_'))
-
-def usage():
-    print >>sys.stderr, "Usage: %s command [options]" % (
-        os.path.basename(sys.argv[0]))
-    print >>sys.stderr
-    print >>sys.stderr, "Available commands:"
+def main():
     cmds_width = max(map(len, commands))
-    for name,func in sorted(commands.items()):
-        print >>sys.stderr, "  %-*s - %s" % (cmds_width, name, func.__doc__)
-    sys.exit(1)
+    parser = OptionParser("""\
+%%prog [options] <command> ... arguments ...
 
-def main():
-    import sys
+Available commands:
+%s""" % ("\n".join("  %-*s - %s" % (cmds_width, name, func.__doc__)
+                   for name, func in sorted(commands.items()))),
+                          version = "lnt version %s" % lnt.__version__)
+    parser.disable_interspersed_args()
+    (opts, args) = parser.parse_args()
 
-    if len(sys.argv) < 2 or sys.argv[1] not in commands:
-        if len(sys.argv) >= 2:
-            print >>sys.stderr,"error: invalid command %r\n" % sys.argv[1]
+    if not args:
+        parser.print_usage()
+        return
 
-        usage()
+    cmd = args[0]
+    if cmd not in commands:
+        parser.error("invalid command: %r" % cmd)
 
-    cmd = sys.argv[1]
-    commands[cmd](cmd, sys.argv[2:])
+    commands[cmd](cmd, args[1:])
 
 if __name__ == '__main__':
     main()



From daniel at zuster.org  Tue May  3 20:48:39 2011
From: daniel at zuster.org (Daniel Dunbar)
Date: Wed, 04 May 2011 01:48:39 -0000
Subject: [llvm-commits] [zorg] r130826 -
	/zorg/trunk/lnt/lnt/lnttool/__init__.py
Message-ID: <20110504014839.C1A7C2A6C12D@llvm.org>

Author: ddunbar
Date: Tue May  3 20:48:39 2011
New Revision: 130826

URL: http://llvm.org/viewvc/llvm-project?rev=130826&view=rev
Log:
lnt runserver: Switch --use-flask-app to default.

Modified:
    zorg/trunk/lnt/lnt/lnttool/__init__.py

Modified: zorg/trunk/lnt/lnt/lnttool/__init__.py
URL: http://llvm.org/viewvc/llvm-project/zorg/trunk/lnt/lnt/lnttool/__init__.py?rev=130826&r1=130825&r2=130826&view=diff
==============================================================================
--- zorg/trunk/lnt/lnt/lnttool/__init__.py (original)
+++ zorg/trunk/lnt/lnt/lnttool/__init__.py Tue May  3 20:48:39 2011
@@ -27,9 +27,6 @@
     parser.add_option("", "--processes", dest="processes", type=int,
                       metavar="N", help="number of processes to use [%default]",
                       default=1)
-    parser.add_option("", "--use-flask-app", dest="use_flask_app",
-                      help="use the Flask based UI [%default]",
-                      action="store_true", default=False)
 
     (opts, args) = parser.parse_args(args)
     if len(args) != 1:
@@ -46,23 +43,16 @@
     if not config or not os.path.exists(config):
         raise SystemExit,"error: invalid config: %r" % config
 
-    if opts.use_flask_app:
-        import lnt.server.ui.app
-        instance = lnt.server.ui.app.App.create_standalone(
-            config_path = config)
-        if opts.debugger:
-            instance.debug = True
-        instance.run(opts.hostname, opts.port,
-                     use_reloader = opts.reloader,
-                     use_debugger = opts.debugger,
-                     threaded = opts.threaded,
-                     processes = opts.processes)
-    else:
-        from werkzeug import run_simple
-        from lnt.viewer import app
-        run_simple(opts.hostname, opts.port, app.create_app(config),
-                   opts.reloader, opts.debugger,
-                   False, None, 1, opts.threaded, opts.processes)
+    import lnt.server.ui.app
+    instance = lnt.server.ui.app.App.create_standalone(
+        config_path = config)
+    if opts.debugger:
+        instance.debug = True
+    instance.run(opts.hostname, opts.port,
+                 use_reloader = opts.reloader,
+                 use_debugger = opts.debugger,
+                 threaded = opts.threaded,
+                 processes = opts.processes)
 
 from create import action_create
 from convert import action_convert



From daniel at zuster.org  Tue May  3 20:48:42 2011
From: daniel at zuster.org (Daniel Dunbar)
Date: Wed, 04 May 2011 01:48:42 -0000
Subject: [llvm-commits] [zorg] r130827 - /zorg/trunk/lnt/lnt/db/runinfo.py
Message-ID: <20110504014842.22F312A6C12E@llvm.org>

Author: ddunbar
Date: Tue May  3 20:48:41 2011
New Revision: 130827

URL: http://llvm.org/viewvc/llvm-project?rev=130827&view=rev
Log:
lnt.db.runinfo: Change default value_precision to match current output of LLVM
test suite, and add an argument to allow disabling the logic which ignores any
change under 1%.

Modified:
    zorg/trunk/lnt/lnt/db/runinfo.py

Modified: zorg/trunk/lnt/lnt/db/runinfo.py
URL: http://llvm.org/viewvc/llvm-project/zorg/trunk/lnt/lnt/db/runinfo.py?rev=130827&r1=130826&r2=130827&view=diff
==============================================================================
--- zorg/trunk/lnt/lnt/db/runinfo.py (original)
+++ zorg/trunk/lnt/lnt/db/runinfo.py Tue May  3 20:48:41 2011
@@ -37,7 +37,8 @@
             else:
                 return UNCHANGED_PASS
 
-    def get_value_status(self, confidence_interval=2.576, value_precision=0.01):
+    def get_value_status(self, confidence_interval=2.576,
+                         value_precision=0.0001, ignore_small=True):
         if self.current is None or self.previous is None:
             return None
 
@@ -61,7 +62,7 @@
 
         # Always ignore percentage changes below 1%, for now, we just don't have
         # enough time to investigate that level of stuff.
-        if abs(self.pct_delta) < .01:
+        if ignore_small and abs(self.pct_delta) < .01:
             return UNCHANGED_PASS
 
         # If we have a comparison window, then measure using a symmetic



From eli.friedman at gmail.com  Tue May  3 21:07:09 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Tue, 3 May 2011 19:07:09 -0700
Subject: [llvm-commits] Fix an encoding issue with ARM PKH* instructions
 for non-MC path
In-Reply-To: <BANLkTi=VkS5DMWoXAnxk9wayeyXbtgB_xA@mail.gmail.com>
References: <BANLkTime=_oB=FscVLUykMYaw5gWMfpVpQ@mail.gmail.com>
	<BANLkTimGh7E1CGtoyyPYWsK_27v-2dOAJg@mail.gmail.com>
	<BANLkTimeMJ=g7cdAsN5y0j3G6GwHidnCZg@mail.gmail.com>
	<BANLkTi=VkS5DMWoXAnxk9wayeyXbtgB_xA@mail.gmail.com>
Message-ID: <BANLkTi=vTxbBxKg7mUaaf4xa-W9wixGNwg@mail.gmail.com>

On Tue, May 3, 2011 at 6:50 PM, Stephen Hines <srhines at google.com> wrote:
> I should be a bit more clear with my answer. The
> file?test/CodeGen/ARM/pack.ll does contain tests for pkhbt and pkhtb, but
> they do not test the actual encoding of the instruction bits. The readable
> assembly code is correct, but it is the encoding where the problem shows up.
> If there are actual encoding tests for the classic code generation path on
> ARM, I would be more than happy to update them. The MC version does
> correctly generate the proper encoding (as seen in
> test/MC/ARM/simple-encoding.ll).

We don't really have any in-tree tests for the old JIT codepath.

-Eli


From atrick at apple.com  Tue May  3 21:10:14 2011
From: atrick at apple.com (Andrew Trick)
Date: Wed, 04 May 2011 02:10:14 -0000
Subject: [llvm-commits] [llvm] r130829 - in /llvm/trunk/lib:
 Analysis/IVUsers.cpp Transforms/Scalar/IndVarSimplify.cpp
Message-ID: <20110504021014.1C1302A6C12C@llvm.org>

Author: atrick
Date: Tue May  3 21:10:13 2011
New Revision: 130829

URL: http://llvm.org/viewvc/llvm-project?rev=130829&view=rev
Log:
indvars: Added DisableIVRewrite and WidenIVs.

This adds functionality to remove size/zero extension during indvars
without generating a canonical IV and rewriting all IV users. It's
disabled by default so should have no effect on codegen. Work in progress.

Modified:
    llvm/trunk/lib/Analysis/IVUsers.cpp
    llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp

Modified: llvm/trunk/lib/Analysis/IVUsers.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/IVUsers.cpp?rev=130829&r1=130828&r2=130829&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/IVUsers.cpp (original)
+++ llvm/trunk/lib/Analysis/IVUsers.cpp Tue May  3 21:10:13 2011
@@ -21,6 +21,7 @@
 #include "llvm/Analysis/Dominators.h"
 #include "llvm/Analysis/LoopPass.h"
 #include "llvm/Analysis/ScalarEvolutionExpressions.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Target/TargetData.h"
 #include "llvm/Assembly/Writer.h"
 #include "llvm/ADT/STLExtras.h"
@@ -38,6 +39,15 @@
 INITIALIZE_PASS_END(IVUsers, "iv-users",
                       "Induction Variable Users", false, true)
 
+// IVUsers behavior currently depends on this temporary indvars mode. The
+// option must be defined upstream from its uses.
+namespace llvm {
+  bool DisableIVRewrite = false;
+}
+cl::opt<bool, true> DisableIVRewriteOpt(
+  "disable-iv-rewrite", cl::Hidden, cl::location(llvm::DisableIVRewrite),
+  cl::desc("Disable canonical induction variable rewriting"));
+
 Pass *llvm::createIVUsersPass() {
   return new IVUsers();
 }
@@ -90,6 +100,11 @@
   if (Width > 64 || (TD && !TD->isLegalInteger(Width)))
     return false;
 
+  // We expect Sign/Zero extension to be eliminated from the IR before analyzing
+  // any downstream uses.
+  if (DisableIVRewrite && (isa<SExtInst>(I) || isa<ZExtInst>(I)))
+    return false;
+
   if (!Processed.insert(I))
     return true;    // Instruction already handled.
 

Modified: llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp?rev=130829&r1=130828&r2=130829&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/IndVarSimplify.cpp Tue May  3 21:10:13 2011
@@ -52,33 +52,41 @@
 #include "llvm/Analysis/LoopInfo.h"
 #include "llvm/Analysis/LoopPass.h"
 #include "llvm/Support/CFG.h"
-#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Transforms/Utils/Local.h"
 #include "llvm/Transforms/Utils/BasicBlockUtils.h"
+#include "llvm/Target/TargetData.h"
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/STLExtras.h"
 using namespace llvm;
 
 STATISTIC(NumRemoved , "Number of aux indvars removed");
+STATISTIC(NumWidened , "Number of indvars widened");
 STATISTIC(NumInserted, "Number of canonical indvars added");
 STATISTIC(NumReplaced, "Number of exit values replaced");
 STATISTIC(NumLFTR    , "Number of loop exit tests replaced");
 
+// DisableIVRewrite mode currently affects IVUsers, so is defined in libAnalysis
+// and referenced here.
+namespace llvm {
+  extern bool DisableIVRewrite;
+}
+
 namespace {
   class IndVarSimplify : public LoopPass {
     IVUsers         *IU;
     LoopInfo        *LI;
     ScalarEvolution *SE;
     DominatorTree   *DT;
+    TargetData      *TD;
     SmallVector<WeakVH, 16> DeadInsts;
     bool Changed;
   public:
 
     static char ID; // Pass identification, replacement for typeid
-    IndVarSimplify() : LoopPass(ID) {
+    IndVarSimplify() : LoopPass(ID), IU(0), LI(0), SE(0), DT(0), TD(0) {
       initializeIndVarSimplifyPass(*PassRegistry::getPassRegistry());
     }
 
@@ -104,6 +112,7 @@
     void EliminateIVComparisons();
     void EliminateIVRemainders();
     void RewriteNonIntegerIVs(Loop *L);
+    const Type *WidenIVs(Loop *L, SCEVExpander &Rewriter);
 
     bool canExpandBackedgeTakenCount(Loop *L,
                                      const SCEV *BackedgeTakenCount);
@@ -111,6 +120,7 @@
     ICmpInst *LinearFunctionTestReplace(Loop *L, const SCEV *BackedgeTakenCount,
                                         PHINode *IndVar,
                                         SCEVExpander &Rewriter);
+
     void RewriteLoopExitValues(Loop *L, SCEVExpander &Rewriter);
 
     void RewriteIVExpressions(Loop *L, SCEVExpander &Rewriter);
@@ -123,7 +133,7 @@
 
 char IndVarSimplify::ID = 0;
 INITIALIZE_PASS_BEGIN(IndVarSimplify, "indvars",
-                "Canonicalize Induction Variables", false, false)
+                "Induction Variable Simplification", false, false)
 INITIALIZE_PASS_DEPENDENCY(DominatorTree)
 INITIALIZE_PASS_DEPENDENCY(LoopInfo)
 INITIALIZE_PASS_DEPENDENCY(ScalarEvolution)
@@ -131,7 +141,7 @@
 INITIALIZE_PASS_DEPENDENCY(LCSSA)
 INITIALIZE_PASS_DEPENDENCY(IVUsers)
 INITIALIZE_PASS_END(IndVarSimplify, "indvars",
-                "Canonicalize Induction Variables", false, false)
+                "Induction Variable Simplification", false, false)
 
 Pass *llvm::createIndVarSimplifyPass() {
   return new IndVarSimplify();
@@ -209,7 +219,7 @@
   // rewriting the loop.
   if (isa<SCEVUDivExpr>(BackedgeTakenCount)) {
     ICmpInst *OrigCond = dyn_cast<ICmpInst>(BI->getCondition());
-    if (!OrigCond) return 0;
+    if (!OrigCond) return false;
     const SCEV *R = SE->getSCEV(OrigCond->getOperand(1));
     R = SE->getMinusSCEV(R, SE->getConstant(R->getType(), 1));
     if (R != BackedgeTakenCount) {
@@ -549,6 +559,8 @@
   LI = &getAnalysis<LoopInfo>();
   SE = &getAnalysis<ScalarEvolution>();
   DT = &getAnalysis<DominatorTree>();
+  TD = getAnalysisIfAvailable<TargetData>();
+
   DeadInsts.clear();
   Changed = false;
 
@@ -560,6 +572,13 @@
 
   // Create a rewriter object which we'll use to transform the code with.
   SCEVExpander Rewriter(*SE);
+  if (DisableIVRewrite)
+    Rewriter.disableCanonicalMode();
+
+  const Type *LargestType = 0;
+  if (DisableIVRewrite) {
+    LargestType = WidenIVs(L, Rewriter);
+  }
 
   // Check to see if this loop has a computable loop-invariant execution count.
   // If so, this means that we can compute the final value of any expressions
@@ -578,7 +597,6 @@
 
   // Compute the type of the largest recurrence expression, and decide whether
   // a canonical induction variable should be inserted.
-  const Type *LargestType = 0;
   bool NeedCannIV = false;
   bool ExpandBECount = canExpandBackedgeTakenCount(L, BackedgeTakenCount);
   if (ExpandBECount) {
@@ -598,8 +616,19 @@
       SE->getEffectiveSCEVType(I->getOperandValToReplace()->getType());
     if (!LargestType ||
         SE->getTypeSizeInBits(Ty) >
+        SE->getTypeSizeInBits(LargestType))
+      LargestType = SE->getEffectiveSCEVType(Ty);
+  }
+  if (!DisableIVRewrite) {
+    for (IVUsers::const_iterator I = IU->begin(), E = IU->end(); I != E; ++I) {
+      NeedCannIV = true;
+      const Type *Ty =
+        SE->getEffectiveSCEVType(I->getOperandValToReplace()->getType());
+      if (!LargestType ||
+          SE->getTypeSizeInBits(Ty) >
           SE->getTypeSizeInBits(LargestType))
-      LargestType = Ty;
+        LargestType = Ty;
+    }
   }
 
   // Now that we know the largest of the induction variable expressions
@@ -647,9 +676,9 @@
     NewICmp = LinearFunctionTestReplace(L, BackedgeTakenCount, IndVar,
                                         Rewriter);
   }
-
   // Rewrite IV-derived expressions.
-  RewriteIVExpressions(L, Rewriter);
+  if (!DisableIVRewrite)
+    RewriteIVExpressions(L, Rewriter);
 
   // Clear the rewriter cache, because values that are in the rewriter's cache
   // can be deleted in the loop below, causing the AssertingVH in the cache to
@@ -721,6 +750,83 @@
   return false;
 }
 
+/// Widen the type of any induction variables that are sign/zero extended and
+/// remove the [sz]ext uses.
+///
+/// FIXME: This may currently create extra IVs which could increase regpressure
+/// (without LSR to cleanup).
+///
+/// FIXME: may factor this with RewriteIVExpressions once it stabilizes.
+const Type *IndVarSimplify::WidenIVs(Loop *L, SCEVExpander &Rewriter) {
+  const Type *LargestType = 0;
+  for (IVUsers::iterator UI = IU->begin(), E = IU->end(); UI != E; ++UI) {
+    Instruction *ExtInst = UI->getUser();
+    if (!isa<SExtInst>(ExtInst) && !isa<ZExtInst>(ExtInst))
+      continue;
+    const SCEV *AR = SE->getSCEV(ExtInst);
+    // Only widen this IV is SCEV tells us it's safe.
+    if (!isa<SCEVAddRecExpr>(AR) && !isa<SCEVAddExpr>(AR))
+      continue;
+
+    if (!L->contains(UI->getUser())) {
+      const SCEV *ExitVal = SE->getSCEVAtScope(AR, L->getParentLoop());
+      if (SE->isLoopInvariant(ExitVal, L))
+        AR = ExitVal;
+    }
+
+    // Only expand affine recurences.
+    if (!isSafe(AR, L, SE))
+      continue;
+
+    const Type *Ty =
+      SE->getEffectiveSCEVType(ExtInst->getType());
+
+    // Only remove [sz]ext if the wide IV is still a native type.
+    //
+    // FIXME: We may be able to remove the copy of this logic in
+    // IVUsers::AddUsersIfInteresting.
+    uint64_t Width = SE->getTypeSizeInBits(Ty);
+    if (Width > 64 || (TD && !TD->isLegalInteger(Width)))
+      continue;
+
+    // Now expand it into actual Instructions and patch it into place.
+    //
+    // FIXME: avoid creating a new IV.
+    Value *NewVal = Rewriter.expandCodeFor(AR, Ty, ExtInst);
+
+    DEBUG(dbgs() << "INDVARS: Widened IV '" << *AR << "' " << *ExtInst << '\n'
+                 << "   into = " << *NewVal << "\n");
+
+    if (!isValidRewrite(ExtInst, NewVal)) {
+      DeadInsts.push_back(NewVal);
+      continue;
+    }
+
+    ++NumWidened;
+    Changed = true;
+
+    if (!LargestType ||
+        SE->getTypeSizeInBits(Ty) >
+        SE->getTypeSizeInBits(LargestType))
+      LargestType = Ty;
+
+    SE->forgetValue(ExtInst);
+
+    // Patch the new value into place.
+    if (ExtInst->hasName())
+      NewVal->takeName(ExtInst);
+    ExtInst->replaceAllUsesWith(NewVal);
+
+    // The old value may be dead now.
+    DeadInsts.push_back(ExtInst);
+
+    // UI is a linked list iterator, so AddUsersIfInteresting effectively pushes
+    // nodes on the worklist.
+    IU->AddUsersIfInteresting(ExtInst);
+  }
+  return LargestType;
+}
+
 void IndVarSimplify::RewriteIVExpressions(Loop *L, SCEVExpander &Rewriter) {
   // Rewrite all induction variable expressions in terms of the canonical
   // induction variable.



From bruno.cardoso at gmail.com  Tue May  3 21:28:31 2011
From: bruno.cardoso at gmail.com (Bruno Cardoso Lopes)
Date: Tue, 3 May 2011 23:28:31 -0300
Subject: [llvm-commits] Fix an encoding issue with ARM PKH* instructions
 for non-MC path
In-Reply-To: <BANLkTi=VkS5DMWoXAnxk9wayeyXbtgB_xA@mail.gmail.com>
References: <BANLkTime=_oB=FscVLUykMYaw5gWMfpVpQ@mail.gmail.com>
	<BANLkTimGh7E1CGtoyyPYWsK_27v-2dOAJg@mail.gmail.com>
	<BANLkTimeMJ=g7cdAsN5y0j3G6GwHidnCZg@mail.gmail.com>
	<BANLkTi=VkS5DMWoXAnxk9wayeyXbtgB_xA@mail.gmail.com>
Message-ID: <BANLkTikS6ad_7Te5n6kGyDC6Wkr=vdeffA@mail.gmail.com>

On Tue, May 3, 2011 at 10:50 PM, Stephen Hines <srhines at google.com> wrote:
> I should be a bit more clear with my answer. The
> file?test/CodeGen/ARM/pack.ll does contain tests for pkhbt and pkhtb, but
> they do not test the actual encoding of the instruction bits. The readable
> assembly code is correct, but it is the encoding where the problem shows up.
> If there are actual encoding tests for the classic code generation path on
> ARM, I would be more than happy to update them. The MC version does
> correctly generate the proper encoding (as seen in
> test/MC/ARM/simple-encoding.ll).

I misread the "non-MC path" on the subject :)


-- 
Bruno Cardoso Lopes
http://www.brunocardoso.cc


From isanbard at gmail.com  Tue May  3 21:40:54 2011
From: isanbard at gmail.com (Bill Wendling)
Date: Wed, 04 May 2011 02:40:54 -0000
Subject: [llvm-commits] [llvm] r130831 -
	/llvm/trunk/include/llvm/IntrinsicsX86.td
Message-ID: <20110504024054.312C82A6C12C@llvm.org>

Author: void
Date: Tue May  3 21:40:54 2011
New Revision: 130831

URL: http://llvm.org/viewvc/llvm-project?rev=130831&view=rev
Log:
Remove dead intrinsics.

Modified:
    llvm/trunk/include/llvm/IntrinsicsX86.td

Modified: llvm/trunk/include/llvm/IntrinsicsX86.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/IntrinsicsX86.td?rev=130831&r1=130830&r2=130831&view=diff
==============================================================================
--- llvm/trunk/include/llvm/IntrinsicsX86.td (original)
+++ llvm/trunk/include/llvm/IntrinsicsX86.td Tue May  3 21:40:54 2011
@@ -224,9 +224,6 @@
 
 // Cacheability support ops
 let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
-  def int_x86_sse_movnt_ps : GCCBuiltin<"__builtin_ia32_movntps">,
-              Intrinsic<[], [llvm_ptr_ty,
-                         llvm_v4f32_ty], []>;
   def int_x86_sse_sfence : GCCBuiltin<"__builtin_ia32_sfence">,
               Intrinsic<[], [], []>;
 }
@@ -536,19 +533,6 @@
                          llvm_v4i32_ty], []>;
 }
 
-// Cacheability support ops
-let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
-  def int_x86_sse2_movnt_dq : GCCBuiltin<"__builtin_ia32_movntdq">,
-              Intrinsic<[], [llvm_ptr_ty,
-                         llvm_v2i64_ty], []>;
-  def int_x86_sse2_movnt_pd : GCCBuiltin<"__builtin_ia32_movntpd">,
-              Intrinsic<[], [llvm_ptr_ty,
-                         llvm_v2f64_ty], []>;
-  def int_x86_sse2_movnt_i : GCCBuiltin<"__builtin_ia32_movnti">,
-              Intrinsic<[], [llvm_ptr_ty,
-                         llvm_i32_ty], []>;
-}
-
 // Misc.
 let TargetPrefix = "x86" in {  // All intrinsics start with "llvm.x86.".
   def int_x86_sse2_packsswb_128 : GCCBuiltin<"__builtin_ia32_packsswb128">,



From nicholas at mxc.ca  Tue May  3 22:58:45 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Wed, 04 May 2011 03:58:45 -0000
Subject: [llvm-commits] [llvm] r130834 -
	/llvm/trunk/runtime/libprofile/GCDAProfiling.c
Message-ID: <20110504035845.4191F2A6C12C@llvm.org>

Author: nicholas
Date: Tue May  3 22:58:45 2011
New Revision: 130834

URL: http://llvm.org/viewvc/llvm-project?rev=130834&view=rev
Log:
Fix crash when not setting GCOV_PREFIX.

Modified:
    llvm/trunk/runtime/libprofile/GCDAProfiling.c

Modified: llvm/trunk/runtime/libprofile/GCDAProfiling.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/runtime/libprofile/GCDAProfiling.c?rev=130834&r1=130833&r2=130834&view=diff
==============================================================================
--- llvm/trunk/runtime/libprofile/GCDAProfiling.c (original)
+++ llvm/trunk/runtime/libprofile/GCDAProfiling.c Tue May  3 22:58:45 2011
@@ -54,7 +54,7 @@
   prefix = getenv("GCOV_PREFIX");
 
   if (!prefix)
-    return strdup(filename);
+    return strdup(orig_filename);
 
   filename = malloc(strlen(prefix) + 1 + strlen(orig_filename) + 1);
   strcpy(filename, prefix);



From nicholas at mxc.ca  Tue May  3 23:03:04 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Wed, 04 May 2011 04:03:04 -0000
Subject: [llvm-commits] [llvm] r130835 -
	/llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
Message-ID: <20110504040304.AAAF72A6C12C@llvm.org>

Author: nicholas
Date: Tue May  3 23:03:04 2011
New Revision: 130835

URL: http://llvm.org/viewvc/llvm-project?rev=130835&view=rev
Log:
Emit gcov data files to the directory specified in the metadata produced by the
frontend, if applicable.

Modified:
    llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp

Modified: llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp?rev=130835&r1=130834&r2=130835&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp (original)
+++ llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp Tue May  3 23:03:04 2011
@@ -40,7 +40,6 @@
 
 namespace {
   class GCOVProfiler : public ModulePass {
-    bool runOnModule(Module &M);
   public:
     static char ID;
     GCOVProfiler()
@@ -57,6 +56,8 @@
     }
 
   private:
+    bool runOnModule(Module &M);
+
     // Create the GCNO files for the Module based on DebugInfo.
     void emitGCNO(DebugInfoFinder &DIF);
 
@@ -88,6 +89,8 @@
                                SmallVector<std::pair<GlobalVariable *,
                                                      uint32_t>, 8> &);
 
+    std::string mangleName(DICompileUnit CU, std::string NewStem);
+
     bool EmitNotes;
     bool EmitData;
 
@@ -323,6 +326,22 @@
   return (sys::path::stem(OrigFilename) + "." + NewStem).str();
 }
 
+std::string GCOVProfiler::mangleName(DICompileUnit CU, std::string NewStem) {
+  if (NamedMDNode *GCov = M->getNamedMetadata("llvm.gcov")) {
+    for (int i = 0, e = GCov->getNumOperands(); i != e; ++i) {
+      MDNode *N = GCov->getOperand(i);
+      if (N->getNumOperands() != 2) continue;
+      MDString *Path = dyn_cast<MDString>(N->getOperand(0));
+      MDNode *CompileUnit = dyn_cast<MDNode>(N->getOperand(1));
+      if (!Path || !CompileUnit) continue;
+      if (CompileUnit == CU)
+        return (Path->getString() + "/" +
+                replaceStem(CU.getFilename(), NewStem)).str();
+    }
+  }
+  return replaceStem(CU.getFilename(), NewStem);
+}
+
 bool GCOVProfiler::runOnModule(Module &M) {
   this->M = &M;
   Ctx = &M.getContext();
@@ -346,8 +365,8 @@
     DICompileUnit CU(*I);
     raw_fd_ostream *&out = GcnoFiles[CU];
     std::string ErrorInfo;
-    out = new raw_fd_ostream(replaceStem(CU.getFilename(), "gcno").c_str(),
-                             ErrorInfo, raw_fd_ostream::F_Binary);
+    out = new raw_fd_ostream(mangleName(CU, "gcno").c_str(), ErrorInfo,
+                             raw_fd_ostream::F_Binary);
     out->write("oncg*404MVLL", 12);
   }
 
@@ -615,7 +634,7 @@
   for (DebugInfoFinder::iterator CUI = DIF.compile_unit_begin(),
            CUE = DIF.compile_unit_end(); CUI != CUE; ++CUI) {
     DICompileUnit compile_unit(*CUI);
-    std::string FilenameGcda = replaceStem(compile_unit.getFilename(), "gcda");
+    std::string FilenameGcda = mangleName(compile_unit, "gcda");
     Builder.CreateCall(StartFile,
                        Builder.CreateGlobalStringPtr(FilenameGcda));
     for (SmallVector<std::pair<GlobalVariable *, uint32_t>, 8>::iterator



From baldrick at free.fr  Wed May  4 11:05:05 2011
From: baldrick at free.fr (Duncan Sands)
Date: Wed, 04 May 2011 16:05:05 -0000
Subject: [llvm-commits] [llvm] r130840 - in /llvm/trunk:
 lib/Analysis/InstructionSimplify.cpp test/Transforms/InstSimplify/maxmin.ll
Message-ID: <20110504160505.F0B572A6C12C@llvm.org>

Author: baldrick
Date: Wed May  4 11:05:05 2011
New Revision: 130840

URL: http://llvm.org/viewvc/llvm-project?rev=130840&view=rev
Log:
Add variations on: max(x,y) >= min(x,z) folds to true.  This isn't that common,
but according to my super-optimizer there are only two missed simplifications
of -instsimplify kind when compiling bzip2, and this is one of them.  It amuses
me to have bzip2 be perfectly optimized as far as instsimplify goes!

Modified:
    llvm/trunk/lib/Analysis/InstructionSimplify.cpp
    llvm/trunk/test/Transforms/InstSimplify/maxmin.ll

Modified: llvm/trunk/lib/Analysis/InstructionSimplify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/InstructionSimplify.cpp?rev=130840&r1=130839&r2=130840&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/InstructionSimplify.cpp (original)
+++ llvm/trunk/lib/Analysis/InstructionSimplify.cpp Wed May  4 11:05:05 2011
@@ -1873,7 +1873,7 @@
   CmpInst::Predicate P = CmpInst::BAD_ICMP_PREDICATE;
   CmpInst::Predicate EqP; // Chosen so that "A == max/min(A,B)" iff "A EqP B".
 
-  // Signed max/min.
+  // Signed variants on "max(a,b)>=a -> true".
   if (match(LHS, m_SMax(m_Value(A), m_Value(B))) && (A == RHS || B == RHS)) {
     if (A != RHS) std::swap(A, B); // smax(A, B) pred A.
     EqP = CmpInst::ICMP_SGE; // "A == smax(A, B)" iff "A sge B".
@@ -1929,7 +1929,7 @@
     }
   }
 
-  // Unsigned max/min.
+  // Unsigned variants on "max(a,b)>=a -> true".
   P = CmpInst::BAD_ICMP_PREDICATE;
   if (match(LHS, m_UMax(m_Value(A), m_Value(B))) && (A == RHS || B == RHS)) {
     if (A != RHS) std::swap(A, B); // umax(A, B) pred A.
@@ -1986,6 +1986,50 @@
     }
   }
 
+  // Variants on "max(x,y) >= min(x,z)".
+  Value *C, *D;
+  if (match(LHS, m_SMax(m_Value(A), m_Value(B))) &&
+      match(RHS, m_SMin(m_Value(C), m_Value(D))) &&
+      (A == C || A == D || B == C || B == D)) {
+    // max(x, ?) pred min(x, ?).
+    if (Pred == CmpInst::ICMP_SGE)
+      // Always true.
+      return Constant::getAllOnesValue(ITy);
+    if (Pred == CmpInst::ICMP_SLT)
+      // Always false.
+      return Constant::getNullValue(ITy);
+  } else if (match(LHS, m_SMin(m_Value(A), m_Value(B))) &&
+             match(RHS, m_SMax(m_Value(C), m_Value(D))) &&
+             (A == C || A == D || B == C || B == D)) {
+    // min(x, ?) pred max(x, ?).
+    if (Pred == CmpInst::ICMP_SLE)
+      // Always true.
+      return Constant::getAllOnesValue(ITy);
+    if (Pred == CmpInst::ICMP_SGT)
+      // Always false.
+      return Constant::getNullValue(ITy);
+  } else if (match(LHS, m_UMax(m_Value(A), m_Value(B))) &&
+             match(RHS, m_UMin(m_Value(C), m_Value(D))) &&
+             (A == C || A == D || B == C || B == D)) {
+    // max(x, ?) pred min(x, ?).
+    if (Pred == CmpInst::ICMP_UGE)
+      // Always true.
+      return Constant::getAllOnesValue(ITy);
+    if (Pred == CmpInst::ICMP_ULT)
+      // Always false.
+      return Constant::getNullValue(ITy);
+  } else if (match(LHS, m_UMin(m_Value(A), m_Value(B))) &&
+             match(RHS, m_UMax(m_Value(C), m_Value(D))) &&
+             (A == C || A == D || B == C || B == D)) {
+    // min(x, ?) pred max(x, ?).
+    if (Pred == CmpInst::ICMP_ULE)
+      // Always true.
+      return Constant::getAllOnesValue(ITy);
+    if (Pred == CmpInst::ICMP_UGT)
+      // Always false.
+      return Constant::getNullValue(ITy);
+  }
+
   // If the comparison is with the result of a select instruction, check whether
   // comparing with either branch of the select always yields the same value.
   if (isa<SelectInst>(LHS) || isa<SelectInst>(RHS))

Modified: llvm/trunk/test/Transforms/InstSimplify/maxmin.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstSimplify/maxmin.ll?rev=130840&r1=130839&r2=130840&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/InstSimplify/maxmin.ll (original)
+++ llvm/trunk/test/Transforms/InstSimplify/maxmin.ll Wed May  4 11:05:05 2011
@@ -143,3 +143,91 @@
   ret i1 %r
 ; CHECK: ret i1 true
 }
+
+define i1 @maxmin1(i32 %x, i32 %y, i32 %z) {
+; CHECK: @maxmin1
+  %c1 = icmp sge i32 %x, %y
+  %max = select i1 %c1, i32 %x, i32 %y
+  %c2 = icmp sge i32 %x, %z
+  %min = select i1 %c2, i32 %z, i32 %x
+  %c = icmp sge i32 %max, %min
+  ret i1 %c
+; CHECK: ret i1 true
+}
+
+define i1 @maxmin2(i32 %x, i32 %y, i32 %z) {
+; CHECK: @maxmin2
+  %c1 = icmp sge i32 %x, %y
+  %max = select i1 %c1, i32 %x, i32 %y
+  %c2 = icmp sge i32 %x, %z
+  %min = select i1 %c2, i32 %z, i32 %x
+  %c = icmp sgt i32 %min, %max
+  ret i1 %c
+; CHECK: ret i1 false
+}
+
+define i1 @maxmin3(i32 %x, i32 %y, i32 %z) {
+; CHECK: @maxmin3
+  %c1 = icmp sge i32 %x, %y
+  %max = select i1 %c1, i32 %x, i32 %y
+  %c2 = icmp sge i32 %x, %z
+  %min = select i1 %c2, i32 %z, i32 %x
+  %c = icmp sle i32 %min, %max
+  ret i1 %c
+; CHECK: ret i1 true
+}
+
+define i1 @maxmin4(i32 %x, i32 %y, i32 %z) {
+; CHECK: @maxmin4
+  %c1 = icmp sge i32 %x, %y
+  %max = select i1 %c1, i32 %x, i32 %y
+  %c2 = icmp sge i32 %x, %z
+  %min = select i1 %c2, i32 %z, i32 %x
+  %c = icmp slt i32 %max, %min
+  ret i1 %c
+; CHECK: ret i1 false
+}
+
+define i1 @maxmin5(i32 %x, i32 %y, i32 %z) {
+; CHECK: @maxmin5
+  %c1 = icmp uge i32 %x, %y
+  %max = select i1 %c1, i32 %x, i32 %y
+  %c2 = icmp uge i32 %x, %z
+  %min = select i1 %c2, i32 %z, i32 %x
+  %c = icmp uge i32 %max, %min
+  ret i1 %c
+; CHECK: ret i1 true
+}
+
+define i1 @maxmin6(i32 %x, i32 %y, i32 %z) {
+; CHECK: @maxmin6
+  %c1 = icmp uge i32 %x, %y
+  %max = select i1 %c1, i32 %x, i32 %y
+  %c2 = icmp uge i32 %x, %z
+  %min = select i1 %c2, i32 %z, i32 %x
+  %c = icmp ugt i32 %min, %max
+  ret i1 %c
+; CHECK: ret i1 false
+}
+
+define i1 @maxmin7(i32 %x, i32 %y, i32 %z) {
+; CHECK: @maxmin7
+  %c1 = icmp uge i32 %x, %y
+  %max = select i1 %c1, i32 %x, i32 %y
+  %c2 = icmp uge i32 %x, %z
+  %min = select i1 %c2, i32 %z, i32 %x
+  %c = icmp ule i32 %min, %max
+  ret i1 %c
+; CHECK: ret i1 true
+}
+
+define i1 @maxmin8(i32 %x, i32 %y, i32 %z) {
+; CHECK: @maxmin8
+  %c1 = icmp uge i32 %x, %y
+  %max = select i1 %c1, i32 %x, i32 %y
+  %c2 = icmp uge i32 %x, %z
+  %min = select i1 %c2, i32 %z, i32 %x
+  %c = icmp ult i32 %max, %min
+  ret i1 %c
+; CHECK: ret i1 false
+}



From dpatel at apple.com  Wed May  4 11:34:02 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 16:34:02 -0000
Subject: [llvm-commits] [llvm] r130842 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
Message-ID: <20110504163402.BC0312A6C12C@llvm.org>

Author: dpatel
Date: Wed May  4 11:34:02 2011
New Revision: 130842

URL: http://llvm.org/viewvc/llvm-project?rev=130842&view=rev
Log:
Tighten up check for empty (i.e. no meaningful debug info) module. This fixes dwarf-die2.c test case from gcc test suite.

Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130842&r1=130841&r2=130842&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Wed May  4 11:34:02 2011
@@ -1113,32 +1113,35 @@
   if (DisableDebugInfoPrinting)
     return;
 
-  bool HasDebugInfo = false;
-
   // If module has named metadata anchors then use them, otherwise scan the module
   // using debug info finder to collect debug info.
   NamedMDNode *CU_Nodes = M->getNamedMetadata("llvm.dbg.cu");
   if (CU_Nodes) {
-    for (unsigned i = 0, e = CU_Nodes->getNumOperands(); i != e; ++i) {
-      HasDebugInfo = true;
-      constructCompileUnit(CU_Nodes->getOperand(i));
-    }
-    if (!HasDebugInfo)
+
+    NamedMDNode *GV_Nodes = M->getNamedMetadata("llvm.dbg.gv");
+    NamedMDNode *SP_Nodes = M->getNamedMetadata("llvm.dbg.sp");
+    if (!GV_Nodes && !SP_Nodes)
+      // If there are not any global variables or any functions then
+      // there is not any debug info in this module.
       return;
 
-    if (NamedMDNode *NMD = M->getNamedMetadata("llvm.dbg.gv"))
-      for (unsigned i = 0, e = NMD->getNumOperands(); i != e; ++i)
-        constructGlobalVariableDIE(NMD->getOperand(i));
-
-    if (NamedMDNode *NMD = M->getNamedMetadata("llvm.dbg.sp"))
-      for (unsigned i = 0, e = NMD->getNumOperands(); i != e; ++i)
-        constructSubprogramDIE(NMD->getOperand(i));
+    for (unsigned i = 0, e = CU_Nodes->getNumOperands(); i != e; ++i)
+      constructCompileUnit(CU_Nodes->getOperand(i));
+
+    if (GV_Nodes)
+      for (unsigned i = 0, e = GV_Nodes->getNumOperands(); i != e; ++i)
+        constructGlobalVariableDIE(GV_Nodes->getOperand(i));
+
+    if (SP_Nodes)
+      for (unsigned i = 0, e = SP_Nodes->getNumOperands(); i != e; ++i)
+        constructSubprogramDIE(SP_Nodes->getOperand(i));
     
   } else {
 
     DebugInfoFinder DbgFinder;
     DbgFinder.processModule(*M);
     
+    bool HasDebugInfo = false;
     // Scan all the compile-units to see if there are any marked as the main unit.
     // if not, we do not generate debug info.
     for (DebugInfoFinder::iterator I = DbgFinder.compile_unit_begin(),
@@ -1148,6 +1151,7 @@
         break;
       }
     }
+    if (!HasDebugInfo) return;
     
     // Create all the compile unit DIEs.
     for (DebugInfoFinder::iterator I = DbgFinder.compile_unit_begin(),
@@ -1165,8 +1169,6 @@
       constructSubprogramDIE(*I);
   }
   
-  if (!HasDebugInfo) return;
-  
   // Tell MMI that we have debug info.
   MMI->setDebugInfoAvailability(true);
   



From dpatel at apple.com  Wed May  4 11:39:26 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 09:39:26 -0700
Subject: [llvm-commits] [llvm] r130485 - in /llvm/trunk:
 lib/Transforms/InstCombine/InstCombineShifts.cpp
 test/Transforms/InstCombine/shift.ll
In-Reply-To: <1A474230-04EB-4BA5-9B1C-A14B6117C315@apple.com>
References: <20110429081541.B7F422A6C138@llvm.org>
	<C8DE37BF-B56C-4F53-A49F-EA158B233245@apple.com>
	<CB365302-F1CC-4578-BA54-7D954227E750@googlemail.com>
	<1B93035B-15A4-47B0-8E78-2D0F8A6B1701@apple.com>
	<5F0816C4-4FC4-4F02-8B7E-51858107AC6D@googlemail.com>
	<C9E399F9-DA42-4972-9BD4-130A27B3D1D5@apple.com>
	<E3078F91-A806-4AEE-8F70-448C10F1B4C6@apple.com>
	<2E25739D-6707-4365-9DC5-E22C7C8E9644@apple.com>
	<1A474230-04EB-4BA5-9B1C-A14B6117C315@apple.com>
Message-ID: <FC302293-2480-4D4D-8FF6-0B1F2EE7168F@apple.com>


On May 3, 2011, at 5:46 PM, Chris Lattner wrote:

> Ok.  Instcombine has an IRBuilder that it uses, changing xforms to use the IRBuilder is vastly preferable to using setDebugLoc everywhere.
...
...

> It's hard to really comment on this without more details, but as long as there aren't tons of new setDebugLoc calls in InstCombine, I'll be happy :)

ok :)
-
Devang

From stoklund at 2pi.dk  Wed May  4 11:45:05 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 16:45:05 -0000
Subject: [llvm-commits] [llvm] r130844 -
	/llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp
Message-ID: <20110504164505.C7B9A2A6C12C@llvm.org>

Author: stoklund
Date: Wed May  4 11:45:05 2011
New Revision: 130844

URL: http://llvm.org/viewvc/llvm-project?rev=130844&view=rev
Log:
Rename -disable-physical-join to -join-physregs and invert it.

Physreg joining is still on by default, but I will turn it off shortly.

Modified:
    llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp

Modified: llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp?rev=130844&r1=130843&r2=130844&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp (original)
+++ llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp Wed May  4 11:45:05 2011
@@ -61,9 +61,9 @@
                cl::init(false), cl::Hidden);
 
 static cl::opt<bool>
-DisablePhysicalJoin("disable-physical-join",
-               cl::desc("Avoid coalescing physical register copies"),
-               cl::init(false), cl::Hidden);
+EnablePhysicalJoin("join-physregs",
+                   cl::desc("Join physical register copies"),
+                   cl::init(true), cl::Hidden);
 
 static cl::opt<bool>
 VerifyCoalescing("verify-coalescing",
@@ -927,7 +927,7 @@
   if (!Allocatable && CP.isFlipped() && JoinVInt.containsOneValue())
     return true;
 
-  if (DisablePhysicalJoin) {
+  if (!EnablePhysicalJoin) {
     DEBUG(dbgs() << "\tPhysreg joins disabled.\n");
     return false;
   }



From aggarwa4 at illinois.edu  Wed May  4 11:56:12 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Wed, 04 May 2011 16:56:12 -0000
Subject: [llvm-commits] [poolalloc] r130845 -
	/poolalloc/trunk/lib/DSA/StdLibPass.cpp
Message-ID: <20110504165612.C73F22A6C12C@llvm.org>

Author: aggarwa4
Date: Wed May  4 11:56:12 2011
New Revision: 130845

URL: http://llvm.org/viewvc/llvm-project?rev=130845&view=rev
Log:
Add comment to identify index/rindex functions
as similiar to strchr/strrchr. This might he helpful
if we make changes to either's summary.

Modified:
    poolalloc/trunk/lib/DSA/StdLibPass.cpp

Modified: poolalloc/trunk/lib/DSA/StdLibPass.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/DSA/StdLibPass.cpp?rev=130845&r1=130844&r2=130845&view=diff
==============================================================================
--- poolalloc/trunk/lib/DSA/StdLibPass.cpp (original)
+++ poolalloc/trunk/lib/DSA/StdLibPass.cpp Wed May  4 11:56:12 2011
@@ -270,8 +270,12 @@
   {"pool_strncpy",    {NRET_NNYARGS, YRET_NNYARGS,  NRET_NARGS, YRET_NNYARGS,  true}},
   {"pool_strcpy",     {NRET_NNYARGS, YRET_NNYARGS,  NRET_NARGS, YRET_NNYARGS,  true}},
   {"pool_stpcpy",     {NRET_NNYARGS, YRET_NNYARGS,  NRET_NARGS, YRET_NNYARGS,  true}},
+  // strchr and index have same functionality
   {"pool_strchr",     {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, YRET_NYARGS,   true}},
+  {"pool_index",      {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, YRET_NYARGS,   true}},
+  // strrchr and rindex have same functionality
   {"pool_strrchr",    {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, YRET_NYARGS,   true}},
+  {"pool_rindex",     {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, YRET_NYARGS,   true}},
   {"pool_strcat",     {NRET_NNYARGS, YRET_NNYARGS,  NRET_NARGS, YRET_NNYARGS,  true}},
   {"pool_strncat",    {NRET_NNYARGS, YRET_NNYARGS,  NRET_NARGS, YRET_NNYARGS,  true}},
   {"pool_strstr",     {NRET_NNYARGS, YRET_NARGS,    NRET_NARGS, YRET_NNYNARGS, true}},
@@ -289,8 +293,6 @@
   {"pool_strncasecmp",{NRET_NNYARGS, NRET_NARGS,    NRET_NARGS, NRET_NARGS,   false}},
   {"pool_bcopy",      {NRET_NNYARGS, NRET_NNNYARGS, NRET_NARGS, NRET_NNYARGS,  true}},
   {"pool_bcmp",       {NRET_NNYARGS, NRET_NARGS,    NRET_NARGS, NRET_NARGS,    true}},
-  {"pool_index",      {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, YRET_NYARGS,   true}},
-  {"pool_rindex",     {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, YRET_NYARGS,   true}},
 
 #if 0
   {"wait",       {false, false, false, false,  true, false, false, false, false}},



From rafael.espindola at gmail.com  Wed May  4 12:44:07 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Wed, 04 May 2011 17:44:07 -0000
Subject: [llvm-commits] [llvm] r130846 - in /llvm/trunk:
 include/llvm/MC/MCAsmInfo.h lib/CodeGen/AsmPrinter/DwarfDebug.cpp
 lib/MC/MCAsmInfo.cpp lib/MC/MCAsmInfoDarwin.cpp test/DebugInfo/stmt-list.ll
Message-ID: <20110504174407.2F66D2A6C12C@llvm.org>

Author: rafael
Date: Wed May  4 12:44:06 2011
New Revision: 130846

URL: http://llvm.org/viewvc/llvm-project?rev=130846&view=rev
Log:
Producing a DW_FORM_addr for DW_AT_stmt_list is probably correct, but
it is both inefficient and unexpected by dwarfdump. Change to
a DW_FORM_data4.

While in here, change the predicate name to reflect that the position
is not really absolute (it is an offset), just that the linker needs a
relocation.

Added:
    llvm/trunk/test/DebugInfo/stmt-list.ll
Modified:
    llvm/trunk/include/llvm/MC/MCAsmInfo.h
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
    llvm/trunk/lib/MC/MCAsmInfo.cpp
    llvm/trunk/lib/MC/MCAsmInfoDarwin.cpp

Modified: llvm/trunk/include/llvm/MC/MCAsmInfo.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/MC/MCAsmInfo.h?rev=130846&r1=130845&r2=130846&view=diff
==============================================================================
--- llvm/trunk/include/llvm/MC/MCAsmInfo.h (original)
+++ llvm/trunk/include/llvm/MC/MCAsmInfo.h Wed May  4 12:44:06 2011
@@ -279,9 +279,9 @@
     /// DwarfSectionOffsetDirective - Special section offset directive.
     const char* DwarfSectionOffsetDirective; // Defaults to NULL
 
-    /// DwarfUsesAbsoluteLabelForStmtList - True if DW_AT_stmt_list needs
-    /// absolute label instead of offset.
-    bool DwarfUsesAbsoluteLabelForStmtList;  // Defaults to true;
+    /// DwarfRequiresRelocationForStmtList - True if DW_AT_stmt_list needs
+    /// a relocation to the correct offset.
+    bool DwarfRequiresRelocationForStmtList;  // Defaults to true;
 
     // DwarfUsesLabelOffsetDifference - True if Dwarf2 output can
     // use EmitLabelOffsetDifference.
@@ -476,8 +476,8 @@
     const char *getDwarfSectionOffsetDirective() const {
       return DwarfSectionOffsetDirective;
     }
-    bool doesDwarfUsesAbsoluteLabelForStmtList() const {
-      return DwarfUsesAbsoluteLabelForStmtList;
+    bool doesDwarfRequireRelocationForStmtList() const {
+      return DwarfRequiresRelocationForStmtList;
     }
     bool doesDwarfUsesLabelOffsetForRanges() const {
       return DwarfUsesLabelOffsetForRanges;

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130846&r1=130845&r2=130846&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Wed May  4 12:44:06 2011
@@ -896,8 +896,8 @@
   NewCU->addUInt(Die, dwarf::DW_AT_entry_pc, dwarf::DW_FORM_addr, 0);
   // DW_AT_stmt_list is a offset of line number information for this
   // compile unit in debug_line section.
-  if (Asm->MAI->doesDwarfUsesAbsoluteLabelForStmtList())
-    NewCU->addLabel(Die, dwarf::DW_AT_stmt_list, dwarf::DW_FORM_addr,
+  if(Asm->MAI->doesDwarfRequireRelocationForStmtList())
+    NewCU->addLabel(Die, dwarf::DW_AT_stmt_list, dwarf::DW_FORM_data4,
                     Asm->GetTempSymbol("section_line"));
   else
     NewCU->addUInt(Die, dwarf::DW_AT_stmt_list, dwarf::DW_FORM_data4, 0);

Modified: llvm/trunk/lib/MC/MCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/MC/MCAsmInfo.cpp?rev=130846&r1=130845&r2=130846&view=diff
==============================================================================
--- llvm/trunk/lib/MC/MCAsmInfo.cpp (original)
+++ llvm/trunk/lib/MC/MCAsmInfo.cpp Wed May  4 12:44:06 2011
@@ -76,7 +76,7 @@
   ExceptionsType = ExceptionHandling::None;
   DwarfRequiresFrameSection = true;
   DwarfUsesInlineInfoSection = false;
-  DwarfUsesAbsoluteLabelForStmtList = true;
+  DwarfRequiresRelocationForStmtList = true;
   DwarfSectionOffsetDirective = 0;
   DwarfUsesLabelOffsetForRanges = true;
   HasMicrosoftFastStdCallMangling = false;

Modified: llvm/trunk/lib/MC/MCAsmInfoDarwin.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/MC/MCAsmInfoDarwin.cpp?rev=130846&r1=130845&r2=130846&view=diff
==============================================================================
--- llvm/trunk/lib/MC/MCAsmInfoDarwin.cpp (original)
+++ llvm/trunk/lib/MC/MCAsmInfoDarwin.cpp Wed May  4 12:44:06 2011
@@ -56,6 +56,6 @@
   HasNoDeadStrip = true;
   HasSymbolResolver = true;
 
-  DwarfUsesAbsoluteLabelForStmtList = false;
+  DwarfRequiresRelocationForStmtList = false;
   DwarfUsesLabelOffsetForRanges = false;
 }

Added: llvm/trunk/test/DebugInfo/stmt-list.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/DebugInfo/stmt-list.ll?rev=130846&view=auto
==============================================================================
--- llvm/trunk/test/DebugInfo/stmt-list.ll (added)
+++ llvm/trunk/test/DebugInfo/stmt-list.ll Wed May  4 12:44:06 2011
@@ -0,0 +1,19 @@
+; RUN: llc -mtriple x86_64-pc-linux-gnu < %s | FileCheck %s
+
+; CHECK:      .section        .debug_line,"", at progbits
+; CHECK-NEXT: .Lsection_line:
+
+; CHECK:      .long   .Lsection_line          # DW_AT_stmt_list
+
+define void @f() {
+entry:
+  ret void
+}
+
+!llvm.dbg.sp = !{!0}
+
+!0 = metadata !{i32 589870, i32 0, metadata !1, metadata !"f", metadata !"f", metadata !"", metadata !1, i32 1, metadata !3, i1 false, i1 true, i32 0, i32 0, i32 0, i32 256, i1 true, void ()* @f, null, null} ; [ DW_TAG_subprogram ]
+!1 = metadata !{i32 589865, metadata !"test2.c", metadata !"/home/espindola/llvm", metadata !2} ; [ DW_TAG_file_type ]
+!2 = metadata !{i32 589841, i32 0, i32 12, metadata !"test2.c", metadata !"/home/espindola/llvm", metadata !"clang version 3.0 ()", i1 true, i1 true, metadata !"", i32 0} ; [ DW_TAG_compile_unit ]
+!3 = metadata !{i32 589845, metadata !1, metadata !"", metadata !1, i32 0, i64 0, i64 0, i32 0, i32 0, i32 0, metadata !4, i32 0, i32 0} ; [ DW_TAG_subroutine_type ]
+!4 = metadata !{null}



From ahatanak at gmail.com  Wed May  4 12:54:28 2011
From: ahatanak at gmail.com (Akira Hatanaka)
Date: Wed, 04 May 2011 17:54:28 -0000
Subject: [llvm-commits] [llvm] r130847 - in /llvm/trunk:
 lib/Target/Mips/Mips.h lib/Target/Mips/MipsEmitGPRestore.cpp
 lib/Target/Mips/MipsISelLowering.cpp lib/Target/Mips/MipsTargetMachine.cpp
 lib/Target/Mips/MipsTargetMachine.h test/CodeGen/Mips/gprestore.ll
 test/CodeGen/Mips/internalfunc.ll
Message-ID: <20110504175428.349262A6C12C@llvm.org>

Author: ahatanak
Date: Wed May  4 12:54:27 2011
New Revision: 130847

URL: http://llvm.org/viewvc/llvm-project?rev=130847&view=rev
Log:
Prevent instructions using $gp from being placed between a jalr and the instruction that restores the clobbered $gp.

Added:
    llvm/trunk/lib/Target/Mips/MipsEmitGPRestore.cpp
    llvm/trunk/test/CodeGen/Mips/gprestore.ll
Modified:
    llvm/trunk/lib/Target/Mips/Mips.h
    llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp
    llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp
    llvm/trunk/lib/Target/Mips/MipsTargetMachine.h
    llvm/trunk/test/CodeGen/Mips/internalfunc.ll

Modified: llvm/trunk/lib/Target/Mips/Mips.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/Mips.h?rev=130847&r1=130846&r2=130847&view=diff
==============================================================================
--- llvm/trunk/lib/Target/Mips/Mips.h (original)
+++ llvm/trunk/lib/Target/Mips/Mips.h Wed May  4 12:54:27 2011
@@ -26,6 +26,7 @@
   FunctionPass *createMipsISelDag(MipsTargetMachine &TM);
   FunctionPass *createMipsDelaySlotFillerPass(MipsTargetMachine &TM);
   FunctionPass *createMipsExpandPseudoPass(MipsTargetMachine &TM);
+  FunctionPass *createMipsEmitGPRestorePass(MipsTargetMachine &TM);
 
   extern Target TheMipsTarget;
   extern Target TheMipselTarget;

Added: llvm/trunk/lib/Target/Mips/MipsEmitGPRestore.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsEmitGPRestore.cpp?rev=130847&view=auto
==============================================================================
--- llvm/trunk/lib/Target/Mips/MipsEmitGPRestore.cpp (added)
+++ llvm/trunk/lib/Target/Mips/MipsEmitGPRestore.cpp Wed May  4 12:54:27 2011
@@ -0,0 +1,80 @@
+//===-- MipsEmitGPRestore.cpp - Emit GP restore instruction-----------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===-----------------------------------------------------------------------===//
+//
+// This pass emits instructions that restore $gp right
+// after jalr instructions.
+//
+//===-----------------------------------------------------------------------===//
+
+#define DEBUG_TYPE "emit-gp-restore"
+
+#include "Mips.h"
+#include "MipsTargetMachine.h"
+#include "MipsMachineFunction.h"
+#include "llvm/CodeGen/MachineFunctionPass.h"
+#include "llvm/CodeGen/MachineInstrBuilder.h"
+#include "llvm/Target/TargetInstrInfo.h"
+#include "llvm/ADT/Statistic.h"
+
+using namespace llvm;
+
+namespace {
+  struct Inserter : public MachineFunctionPass {
+
+    TargetMachine &TM;
+    const TargetInstrInfo *TII;
+
+    static char ID;
+    Inserter(TargetMachine &tm)
+      : MachineFunctionPass(ID), TM(tm), TII(tm.getInstrInfo()) { }
+
+    virtual const char *getPassName() const {
+      return "Mips Emit GP Restore";
+    }
+
+    bool runOnMachineBasicBlock(MachineBasicBlock &MBB);
+    bool runOnMachineFunction(MachineFunction &F);
+  };
+  char Inserter::ID = 0;
+} // end of anonymous namespace
+
+bool Inserter::runOnMachineFunction(MachineFunction &F) {
+  if (TM.getRelocationModel() != Reloc::PIC_)
+    return false;
+
+  bool Changed = false;
+  int FI =  F.getInfo<MipsFunctionInfo>()->getGPFI();
+
+  for (MachineFunction::iterator MFI = F.begin(), MFE = F.end();
+       MFI != MFE; ++MFI) {
+    MachineBasicBlock& MBB = *MFI;
+    MachineBasicBlock::iterator I = MFI->begin();
+
+    while (I != MFI->end()) {
+      if (I->getOpcode() != Mips::JALR) {
+        ++I;
+        continue;
+      }
+
+      DebugLoc dl = I->getDebugLoc();
+      // emit lw $gp, ($gp save slot on stack) after jalr
+      BuildMI(MBB, ++I, dl, TII->get(Mips::LW), Mips::GP).addImm(0).addFrameIndex(FI);
+      Changed = true;
+    }
+  } 
+
+  return Changed;
+}
+
+/// createMipsEmitGPRestorePass - Returns a pass that emits instructions that
+/// restores $gp clobbered by jalr instructions.
+FunctionPass *llvm::createMipsEmitGPRestorePass(MipsTargetMachine &tm) {
+  return new Inserter(tm);
+}
+

Modified: llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp?rev=130847&r1=130846&r2=130847&view=diff
==============================================================================
--- llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp (original)
+++ llvm/trunk/lib/Target/Mips/MipsISelLowering.cpp Wed May  4 12:54:27 2011
@@ -1298,17 +1298,6 @@
         }
         MipsFI->setGPStackOffset(LastArgStackLoc);
       }
-
-      // Reload GP value.
-      FI = MipsFI->getGPFI();
-      SDValue FIN = DAG.getFrameIndex(FI, getPointerTy());
-      SDValue GPLoad = DAG.getLoad(MVT::i32, dl, Chain, FIN,
-                                   MachinePointerInfo::getFixedStack(FI),
-                                   false, false, 0);
-      Chain = GPLoad.getValue(1);
-      Chain = DAG.getCopyToReg(Chain, dl, DAG.getRegister(Mips::GP, MVT::i32),
-                               GPLoad, SDValue(0,0));
-      InFlag = Chain.getValue(1);
   }
 
   // Create the CALLSEQ_END node.

Modified: llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp?rev=130847&r1=130846&r2=130847&view=diff
==============================================================================
--- llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp (original)
+++ llvm/trunk/lib/Target/Mips/MipsTargetMachine.cpp Wed May  4 12:54:27 2011
@@ -77,6 +77,12 @@
 }
 
 bool MipsTargetMachine::
+addPreRegAlloc(PassManagerBase &PM, CodeGenOpt::Level OptLevel) {
+  PM.add(createMipsEmitGPRestorePass(*this));
+  return true;
+}
+
+bool MipsTargetMachine::
 addPostRegAlloc(PassManagerBase &PM, CodeGenOpt::Level OptLevel) {
   PM.add(createMipsExpandPseudoPass(*this));
   return true;

Modified: llvm/trunk/lib/Target/Mips/MipsTargetMachine.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/MipsTargetMachine.h?rev=130847&r1=130846&r2=130847&view=diff
==============================================================================
--- llvm/trunk/lib/Target/Mips/MipsTargetMachine.h (original)
+++ llvm/trunk/lib/Target/Mips/MipsTargetMachine.h Wed May  4 12:54:27 2011
@@ -63,6 +63,8 @@
                                  CodeGenOpt::Level OptLevel);
     virtual bool addPreEmitPass(PassManagerBase &PM,
                                 CodeGenOpt::Level OptLevel);
+    virtual bool addPreRegAlloc(PassManagerBase &PM,
+                                CodeGenOpt::Level OptLevel);
     virtual bool addPostRegAlloc(PassManagerBase &, CodeGenOpt::Level);
   };
 

Added: llvm/trunk/test/CodeGen/Mips/gprestore.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Mips/gprestore.ll?rev=130847&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/Mips/gprestore.ll (added)
+++ llvm/trunk/test/CodeGen/Mips/gprestore.ll Wed May  4 12:54:27 2011
@@ -0,0 +1,35 @@
+; RUN: llc -march=mips < %s | FileCheck %s
+
+ at p = external global i32
+ at q = external global i32
+ at r = external global i32
+
+define void @f0() nounwind {
+entry:
+; CHECK: jalr
+; CHECK-NOT: got({{.*}})($gp)
+; CHECK: lw $gp
+; CHECK: jalr
+; CHECK-NOT: got({{.*}})($gp)
+; CHECK: lw $gp
+; CHECK: jalr
+; CHECK-NOT: got({{.*}})($gp)
+; CHECK: lw $gp
+  tail call void (...)* @f1() nounwind
+  %tmp = load i32* @p, align 4, !tbaa !0
+  tail call void @f2(i32 %tmp) nounwind
+  %tmp1 = load i32* @q, align 4, !tbaa !0
+  %tmp2 = load i32* @r, align 4, !tbaa !0
+  tail call void @f3(i32 %tmp1, i32 %tmp2) nounwind
+  ret void
+}
+
+declare void @f1(...)
+
+declare void @f2(i32)
+
+declare void @f3(i32, i32)
+
+!0 = metadata !{metadata !"int", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA", null}

Modified: llvm/trunk/test/CodeGen/Mips/internalfunc.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Mips/internalfunc.ll?rev=130847&r1=130846&r2=130847&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Mips/internalfunc.ll (original)
+++ llvm/trunk/test/CodeGen/Mips/internalfunc.ll Wed May  4 12:54:27 2011
@@ -1,4 +1,4 @@
-; RUN: llc  < %s -march=mips | FileCheck %s
+; RUN: llc  < %s -march=mipsel -mcpu=4ke  | FileCheck %s
 
 @caller.sf1 = internal unnamed_addr global void (...)* null, align 4
 @gf1 = external global void (...)*



From ahatanak at gmail.com  Wed May  4 13:28:36 2011
From: ahatanak at gmail.com (Akira Hatanaka)
Date: Wed, 04 May 2011 18:28:36 -0000
Subject: [llvm-commits] [llvm] r130849 -
	/llvm/trunk/test/CodeGen/Mips/gprestore.ll
Message-ID: <20110504182836.5E6BB2A6C12C@llvm.org>

Author: ahatanak
Date: Wed May  4 13:28:36 2011
New Revision: 130849

URL: http://llvm.org/viewvc/llvm-project?rev=130849&view=rev
Log:
Remove LLVM IR metadata in test case committed in r130847.

Modified:
    llvm/trunk/test/CodeGen/Mips/gprestore.ll

Modified: llvm/trunk/test/CodeGen/Mips/gprestore.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Mips/gprestore.ll?rev=130849&r1=130848&r2=130849&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Mips/gprestore.ll (original)
+++ llvm/trunk/test/CodeGen/Mips/gprestore.ll Wed May  4 13:28:36 2011
@@ -16,10 +16,10 @@
 ; CHECK-NOT: got({{.*}})($gp)
 ; CHECK: lw $gp
   tail call void (...)* @f1() nounwind
-  %tmp = load i32* @p, align 4, !tbaa !0
+  %tmp = load i32* @p, align 4
   tail call void @f2(i32 %tmp) nounwind
-  %tmp1 = load i32* @q, align 4, !tbaa !0
-  %tmp2 = load i32* @r, align 4, !tbaa !0
+  %tmp1 = load i32* @q, align 4
+  %tmp2 = load i32* @r, align 4
   tail call void @f3(i32 %tmp1, i32 %tmp2) nounwind
   ret void
 }
@@ -30,6 +30,3 @@
 
 declare void @f3(i32, i32)
 
-!0 = metadata !{metadata !"int", metadata !1}
-!1 = metadata !{metadata !"omnipotent char", metadata !2}
-!2 = metadata !{metadata !"Simple C/C++ TBAA", null}



From rafael.espindola at gmail.com  Wed May  4 13:46:57 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Wed, 04 May 2011 18:46:57 -0000
Subject: [llvm-commits] [llvm] r130850 -
	/llvm/trunk/lib/Target/Mips/CMakeLists.txt
Message-ID: <20110504184657.0D3C22A6C12C@llvm.org>

Author: rafael
Date: Wed May  4 13:46:56 2011
New Revision: 130850

URL: http://llvm.org/viewvc/llvm-project?rev=130850&view=rev
Log:
Fix cmake build.

Modified:
    llvm/trunk/lib/Target/Mips/CMakeLists.txt

Modified: llvm/trunk/lib/Target/Mips/CMakeLists.txt
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/Mips/CMakeLists.txt?rev=130850&r1=130849&r2=130850&view=diff
==============================================================================
--- llvm/trunk/lib/Target/Mips/CMakeLists.txt (original)
+++ llvm/trunk/lib/Target/Mips/CMakeLists.txt Wed May  4 13:46:56 2011
@@ -13,6 +13,7 @@
 add_llvm_target(MipsCodeGen
   MipsAsmPrinter.cpp
   MipsDelaySlotFiller.cpp
+  MipsEmitGPRestore.cpp
   MipsExpandPseudo.cpp
   MipsInstrInfo.cpp
   MipsISelDAGToDAG.cpp



From fjahanian at apple.com  Wed May  4 13:57:05 2011
From: fjahanian at apple.com (Fariborz Jahanian)
Date: Wed, 04 May 2011 18:57:05 -0000
Subject: [llvm-commits] [test-suite] r130852 -
 /test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield.c
Message-ID: <20110504185705.148DE2A6C12C@llvm.org>

Author: fjahanian
Date: Wed May  4 13:57:04 2011
New Revision: 130852

URL: http://llvm.org/viewvc/llvm-project?rev=130852&view=rev
Log:
Test for packing of ms_struct bitfields.
// rdar://8823265

Added:
    test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield.c

Added: test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield.c
URL: http://llvm.org/viewvc/llvm-project/test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield.c?rev=130852&view=auto
==============================================================================
--- test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield.c (added)
+++ test-suite/trunk/SingleSource/UnitTests/ms_struct-bitfield.c Wed May  4 13:57:04 2011
@@ -0,0 +1,160 @@
+extern void abort ();
+
+#define ATTR __attribute__((__ms_struct__))
+
+#define size_struct_0 1
+#define size_struct_1 4
+#define size_struct_2 24
+#define size_struct_3 8
+#define size_struct_4 32
+#define size_struct_5 12
+#define size_struct_6 40
+#define size_struct_7 8
+#define size_struct_8 20
+#define size_struct_9 32
+
+struct _struct_0
+{
+  char member_0;
+} ATTR;
+typedef struct _struct_0 struct_0;
+
+struct _struct_1
+{
+  char member_0;
+  short member_1:13;
+} ATTR;
+typedef struct _struct_1 struct_1;
+
+struct _struct_2
+{
+  double member_0;
+  unsigned char member_1:8;
+  int member_2:32;
+  unsigned char member_3:5;
+  short member_4:14;
+  short member_5:13;
+  unsigned char:0;
+} ATTR;
+typedef struct _struct_2 struct_2;
+
+struct _struct_3
+{
+  unsigned int member_0:26;
+  unsigned char member_1:2;
+
+} ATTR;
+typedef struct _struct_3 struct_3;
+
+struct _struct_4
+{
+  unsigned char member_0:7;
+  double member_1;
+  double member_2;
+  short member_3:5;
+  char member_4:2;
+
+} ATTR;
+typedef struct _struct_4 struct_4;
+
+struct _struct_5
+{
+  unsigned short member_0:12;
+  int member_1:1;
+  unsigned short member_2:6;
+
+} ATTR;
+typedef struct _struct_5 struct_5;
+
+struct _struct_6
+{
+  unsigned char member_0:7;
+  unsigned int member_1:25;
+  char member_2:1;
+  double member_3;
+  short member_4:9;
+  double member_5;
+
+} ATTR;
+typedef struct _struct_6 struct_6;
+
+struct _struct_7
+{
+  double member_0;
+
+} ATTR;
+typedef struct _struct_7 struct_7;
+
+struct _struct_8
+{
+  unsigned char member_0:7;
+  int member_1:11;
+  int member_2:5;
+  int:0;
+  char member_4:8;
+  unsigned short member_5:4;
+  unsigned char member_6:3;
+  int member_7:23;
+
+} ATTR;
+typedef struct _struct_8 struct_8;
+
+struct _struct_9
+{
+  double member_0;
+  unsigned int member_1:6;
+  int member_2:17;
+  double member_3;
+  unsigned int member_4:22;
+
+} ATTR;
+typedef struct _struct_9 struct_9;
+
+struct_0 test_struct_0 = { 123 };
+struct_1 test_struct_1 = { 82, 1081 };
+struct_2 test_struct_2 = { 20.0, 31, 407760, 1, 14916, 6712 };
+struct_3 test_struct_3 = { 64616999, 1 };
+struct_4 test_struct_4 = { 61, 20.0, 20.0, 12, 0 };
+struct_5 test_struct_5 = { 909, 1, 57 };
+struct_6 test_struct_6 = { 12, 21355796, 0, 20.0, 467, 20.0 };
+struct_7 test_struct_7 = { 20.0 };
+struct_8 test_struct_8 = { 126, 1821, 22, 125, 6, 0, 2432638 };
+struct_9 test_struct_9 = { 20.0, 3, 23957, 20.0, 1001631 };
+
+
+int
+main (void)
+{
+
+  if (size_struct_0 != sizeof (struct_0))
+    abort ();
+
+  if (size_struct_1 != sizeof (struct_1))
+    abort ();
+
+  if (size_struct_2 != sizeof (struct_2))
+    abort ();
+
+  if (size_struct_3 != sizeof (struct_3))
+    abort ();
+
+  if (size_struct_4 != sizeof (struct_4))
+    abort ();
+
+  if (size_struct_5 != sizeof (struct_5))
+    abort ();
+
+  if (size_struct_6 != sizeof (struct_6))
+    abort ();
+
+  if (size_struct_7 != sizeof (struct_7))
+    abort ();
+
+  if (size_struct_8 != sizeof (struct_8))
+    abort ();
+
+  if (size_struct_9 != sizeof (struct_9))
+    abort ();
+
+  return 0;
+}



From dpatel at apple.com  Wed May  4 14:00:57 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 19:00:57 -0000
Subject: [llvm-commits] [llvm] r130854 - in /llvm/trunk:
 lib/Target/ARM/ARMAsmPrinter.cpp test/CodeGen/ARM/debug-info-sreg2.ll
Message-ID: <20110504190057.EB3AE2A6C12C@llvm.org>

Author: dpatel
Date: Wed May  4 14:00:57 2011
New Revision: 130854

URL: http://llvm.org/viewvc/llvm-project?rev=130854&view=rev
Log:
Do not emit location expression size twice.

Added:
    llvm/trunk/test/CodeGen/ARM/debug-info-sreg2.ll
Modified:
    llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp

Modified: llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp?rev=130854&r1=130853&r2=130854&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp (original)
+++ llvm/trunk/lib/Target/ARM/ARMAsmPrinter.cpp Wed May  4 14:00:57 2011
@@ -188,7 +188,6 @@
       
       unsigned SReg = Reg - ARM::S0;
       unsigned Rx = 256 + (SReg >> 1);
-      OutStreamer.AddComment("Loc expr size");
       // DW_OP_regx + ULEB + DW_OP_bit_piece + ULEB + ULEB
       //   1 + ULEB(Rx) + 1 + 1 + 1
       return 4 + MCAsmInfo::getULEB128Size(Rx);
@@ -203,7 +202,6 @@
       unsigned D1 = 256 + 2 * QReg;
       unsigned D2 = D1 + 1;
       
-      OutStreamer.AddComment("Loc expr size");
       // DW_OP_regx + ULEB + DW_OP_piece + ULEB(8) +
       // DW_OP_regx + ULEB + DW_OP_piece + ULEB(8);
       //   6 + ULEB(D1) + ULEB(D2)
@@ -229,10 +227,6 @@
       unsigned SReg = Reg - ARM::S0;
       bool odd = SReg & 0x1;
       unsigned Rx = 256 + (SReg >> 1);
-      OutStreamer.AddComment("Loc expr size");
-      // DW_OP_regx + ULEB + DW_OP_bit_piece + ULEB + ULEB
-      //   1 + ULEB(Rx) + 1 + 1 + 1
-      EmitInt16(4 + MCAsmInfo::getULEB128Size(Rx));
 
       OutStreamer.AddComment("DW_OP_regx for S register");
       EmitInt8(dwarf::DW_OP_regx);
@@ -260,12 +254,6 @@
       unsigned D1 = 256 + 2 * QReg;
       unsigned D2 = D1 + 1;
       
-      OutStreamer.AddComment("Loc expr size");
-      // DW_OP_regx + ULEB + DW_OP_piece + ULEB(8) +
-      // DW_OP_regx + ULEB + DW_OP_piece + ULEB(8);
-      //   6 + ULEB(D1) + ULEB(D2)
-      EmitInt16(6 + MCAsmInfo::getULEB128Size(D1) + MCAsmInfo::getULEB128Size(D2));
-
       OutStreamer.AddComment("DW_OP_regx for Q register: D1");
       EmitInt8(dwarf::DW_OP_regx);
       EmitULEB128(D1);

Added: llvm/trunk/test/CodeGen/ARM/debug-info-sreg2.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/debug-info-sreg2.ll?rev=130854&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/debug-info-sreg2.ll (added)
+++ llvm/trunk/test/CodeGen/ARM/debug-info-sreg2.ll Wed May  4 14:00:57 2011
@@ -0,0 +1,59 @@
+; RUN: llc < %s - | FileCheck %s
+; Radar 9376013
+target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:32:64-v128:32:128-a0:0:32-n32"
+target triple = "thumbv7-apple-macosx10.6.7"
+
+;CHECK: Ldebug_loc0:
+;CHECK-NEXT:        .long   Ltmp0
+;CHECK-NEXT:        .long   Ltmp2
+;CHECK-NEXT:        .short  6                       @ Loc expr size
+;CHECK-NEXT:        .byte   144                     @ DW_OP_regx for S register
+
+define void @_Z3foov() optsize ssp {
+entry:
+  %call = tail call float @_Z3barv() optsize, !dbg !11
+  tail call void @llvm.dbg.value(metadata !{float %call}, i64 0, metadata !5), !dbg !11
+  %call16 = tail call float @_Z2f2v() optsize, !dbg !12
+  %cmp7 = fcmp olt float %call, %call16, !dbg !12
+  br i1 %cmp7, label %for.body, label %for.end, !dbg !12
+
+for.body:                                         ; preds = %entry, %for.body
+  %k.08 = phi float [ %inc, %for.body ], [ %call, %entry ]
+  %call4 = tail call float @_Z2f3f(float %k.08) optsize, !dbg !13
+  %inc = fadd float %k.08, 1.000000e+00, !dbg !14
+  %call1 = tail call float @_Z2f2v() optsize, !dbg !12
+  %cmp = fcmp olt float %inc, %call1, !dbg !12
+  br i1 %cmp, label %for.body, label %for.end, !dbg !12
+
+for.end:                                          ; preds = %for.body, %entry
+  ret void, !dbg !15
+}
+
+declare float @_Z3barv() optsize
+
+declare float @_Z2f2v() optsize
+
+declare float @_Z2f3f(float) optsize
+
+declare void @llvm.dbg.value(metadata, i64, metadata) nounwind readnone
+
+!llvm.dbg.cu = !{!0}
+!llvm.dbg.sp = !{!1}
+!llvm.dbg.lv._Z3foov = !{!5, !8}
+
+!0 = metadata !{i32 589841, i32 0, i32 4, metadata !"k.cc", metadata !"/private/tmp", metadata !"clang version 3.0 (trunk 130845)", i1 true, i1 true, metadata !"", i32 0} ; [ DW_TAG_compile_unit ]
+!1 = metadata !{i32 589870, i32 0, metadata !2, metadata !"foo", metadata !"foo", metadata !"_Z3foov", metadata !2, i32 5, metadata !3, i1 false, i1 true, i32 0, i32 0, i32 0, i32 256, i1 true, void ()* @_Z3foov, null, null} ; [ DW_TAG_subprogram ]
+!2 = metadata !{i32 589865, metadata !"k.cc", metadata !"/private/tmp", metadata !0} ; [ DW_TAG_file_type ]
+!3 = metadata !{i32 589845, metadata !2, metadata !"", metadata !2, i32 0, i64 0, i64 0, i32 0, i32 0, i32 0, metadata !4, i32 0, i32 0} ; [ DW_TAG_subroutine_type ]
+!4 = metadata !{null}
+!5 = metadata !{i32 590080, metadata !6, metadata !"k", metadata !2, i32 6, metadata !7, i32 0} ; [ DW_TAG_auto_variable ]
+!6 = metadata !{i32 589835, metadata !1, i32 5, i32 12, metadata !2, i32 0} ; [ DW_TAG_lexical_block ]
+!7 = metadata !{i32 589860, metadata !0, metadata !"float", null, i32 0, i64 32, i64 32, i64 0, i32 0, i32 4} ; [ DW_TAG_base_type ]
+!8 = metadata !{i32 590080, metadata !9, metadata !"y", metadata !2, i32 8, metadata !7, i32 0} ; [ DW_TAG_auto_variable ]
+!9 = metadata !{i32 589835, metadata !10, i32 7, i32 25, metadata !2, i32 2} ; [ DW_TAG_lexical_block ]
+!10 = metadata !{i32 589835, metadata !6, i32 7, i32 3, metadata !2, i32 1} ; [ DW_TAG_lexical_block ]
+!11 = metadata !{i32 6, i32 18, metadata !6, null}
+!12 = metadata !{i32 7, i32 3, metadata !6, null}
+!13 = metadata !{i32 8, i32 20, metadata !9, null}
+!14 = metadata !{i32 7, i32 20, metadata !10, null}
+!15 = metadata !{i32 10, i32 1, metadata !6, null}



From stoklund at 2pi.dk  Wed May  4 14:01:59 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 19:01:59 -0000
Subject: [llvm-commits] [llvm] r130855 - in /llvm/trunk/test:
 CodeGen/PowerPC/2010-03-09-indirect-call.ll MC/ARM/simple-encoding.ll
Message-ID: <20110504190159.564432A6C12C@llvm.org>

Author: stoklund
Date: Wed May  4 14:01:59 2011
New Revision: 130855

URL: http://llvm.org/viewvc/llvm-project?rev=130855&view=rev
Log:
Explicitly request -join-physregs for some tests that depend on it.

Modified:
    llvm/trunk/test/CodeGen/PowerPC/2010-03-09-indirect-call.ll
    llvm/trunk/test/MC/ARM/simple-encoding.ll

Modified: llvm/trunk/test/CodeGen/PowerPC/2010-03-09-indirect-call.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/2010-03-09-indirect-call.ll?rev=130855&r1=130854&r2=130855&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/2010-03-09-indirect-call.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/2010-03-09-indirect-call.ll Wed May  4 14:01:59 2011
@@ -1,4 +1,4 @@
-; RUN: llc < %s -march=ppc32 -mcpu=g5 -mtriple=powerpc-apple-darwin10.0 | FileCheck %s
+; RUN: llc < %s -march=ppc32 -mcpu=g5 -mtriple=powerpc-apple-darwin10.0 -join-physregs | FileCheck %s
 ; ModuleID = 'nn.c'
 target datalayout = "E-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f128:64:128"
 target triple = "powerpc-apple-darwin11.0"

Modified: llvm/trunk/test/MC/ARM/simple-encoding.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/MC/ARM/simple-encoding.ll?rev=130855&r1=130854&r2=130855&view=diff
==============================================================================
--- llvm/trunk/test/MC/ARM/simple-encoding.ll (original)
+++ llvm/trunk/test/MC/ARM/simple-encoding.ll Wed May  4 14:01:59 2011
@@ -1,4 +1,4 @@
-;RUN: llc -mtriple=armv7-apple-darwin -show-mc-encoding -disable-cgp-branch-opts < %s | FileCheck %s
+;RUN: llc -mtriple=armv7-apple-darwin -show-mc-encoding -disable-cgp-branch-opts -join-physregs < %s | FileCheck %s
 
 
 ;FIXME: Once the ARM integrated assembler is up and going, these sorts of tests



From stoklund at 2pi.dk  Wed May  4 14:02:04 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 19:02:04 -0000
Subject: [llvm-commits] [llvm] r130857 - in /llvm/trunk/lib/Target/SystemZ:
 SystemZRegisterInfo.cpp SystemZRegisterInfo.h
Message-ID: <20110504190204.690B02A6C12E@llvm.org>

Author: stoklund
Date: Wed May  4 14:02:04 2011
New Revision: 130857

URL: http://llvm.org/viewvc/llvm-project?rev=130857&view=rev
Log:
Implement SystemZRegisterInfo::getMatchingSuperRegClass to enable cross-class joins.

Modified:
    llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.cpp
    llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.h

Modified: llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.cpp?rev=130857&r1=130856&r2=130857&view=diff
==============================================================================
--- llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.cpp (original)
+++ llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.cpp Wed May  4 14:02:04 2011
@@ -58,6 +58,20 @@
   return Reserved;
 }
 
+const TargetRegisterClass*
+SystemZRegisterInfo::getMatchingSuperRegClass(const TargetRegisterClass *A,
+                                              const TargetRegisterClass *B,
+                                              unsigned Idx) const {
+  switch(Idx) {
+  // Exact sub-classes don't exist for the other sub-register indexes.
+  default: return 0;
+  case SystemZ::subreg_32bit:
+    if (B == SystemZ::ADDR32RegisterClass)
+      return A->getSize() == 8 ? SystemZ::ADDR64RegisterClass : 0;
+    return A;
+  }
+}
+
 void SystemZRegisterInfo::
 eliminateCallFramePseudoInstr(MachineFunction &MF, MachineBasicBlock &MBB,
                               MachineBasicBlock::iterator I) const {

Modified: llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.h?rev=130857&r1=130856&r2=130857&view=diff
==============================================================================
--- llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.h (original)
+++ llvm/trunk/lib/Target/SystemZ/SystemZRegisterInfo.h Wed May  4 14:02:04 2011
@@ -34,6 +34,10 @@
 
   BitVector getReservedRegs(const MachineFunction &MF) const;
 
+  const TargetRegisterClass*
+  getMatchingSuperRegClass(const TargetRegisterClass *A,
+                           const TargetRegisterClass *B, unsigned Idx) const;
+
   void eliminateCallFramePseudoInstr(MachineFunction &MF,
                                      MachineBasicBlock &MBB,
                                      MachineBasicBlock::iterator I) const;



From stoklund at 2pi.dk  Wed May  4 14:02:01 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 19:02:01 -0000
Subject: [llvm-commits] [llvm] r130856 -
	/llvm/trunk/test/CodeGen/PowerPC/big-endian-formal-args.ll
Message-ID: <20110504190201.CE1232A6C12D@llvm.org>

Author: stoklund
Date: Wed May  4 14:02:01 2011
New Revision: 130856

URL: http://llvm.org/viewvc/llvm-project?rev=130856&view=rev
Log:
FileCheckize and break dependence on coalescing order.

Modified:
    llvm/trunk/test/CodeGen/PowerPC/big-endian-formal-args.ll

Modified: llvm/trunk/test/CodeGen/PowerPC/big-endian-formal-args.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PowerPC/big-endian-formal-args.ll?rev=130856&r1=130855&r2=130856&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PowerPC/big-endian-formal-args.ll (original)
+++ llvm/trunk/test/CodeGen/PowerPC/big-endian-formal-args.ll Wed May  4 14:02:01 2011
@@ -1,14 +1,12 @@
-; RUN: llc < %s -march=ppc32 -mtriple=powerpc-unknown-linux-gnu | \
-; RUN:   grep {li 6, 3}
-; RUN: llc < %s -march=ppc32 -mtriple=powerpc-unknown-linux-gnu | \
-; RUN:   grep {li 4, 2}
-; RUN: llc < %s -march=ppc32 -mtriple=powerpc-unknown-linux-gnu | \
-; RUN:   grep {li 3, 0}
-; RUN: llc < %s -march=ppc32 -mtriple=powerpc-unknown-linux-gnu | \
-; RUN:   grep {mr 5, 3}
+; RUN: llc < %s -march=ppc32 -mtriple=powerpc-unknown-linux-gnu | FileCheck %s
 
 declare void @bar(i64 %x, i64 %y)
 
+; CHECK: li 4, 2
+; CHECK: li {{[53]}}, 0
+; CHECK: li 6, 3
+; CHECK: mr {{[53]}}, {{[53]}}
+
 define void @foo() {
   call void @bar(i64 2, i64 3)
   ret void



From stoklund at 2pi.dk  Wed May  4 14:02:07 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 19:02:07 -0000
Subject: [llvm-commits] [llvm] r130858 - in /llvm/trunk/test/CodeGen/Thumb2:
 thumb2-cmn.ll thumb2-cmp.ll thumb2-cmp2.ll thumb2-teq.ll thumb2-teq2.ll
 thumb2-tst.ll thumb2-tst2.ll
Message-ID: <20110504190207.DA84B2A6C12C@llvm.org>

Author: stoklund
Date: Wed May  4 14:02:07 2011
New Revision: 130858

URL: http://llvm.org/viewvc/llvm-project?rev=130858&view=rev
Log:
Explicitly request physreg coalesing for a bunch of Thumb2 unit tests.

These tests all follow the same pattern:

	mov	r2, r0
	movs	r0, #0
	$CMP	r2, r1
	it	eq
	moveq	r0, #1
	bx	lr

The first 'mov' can be eliminated by rematerializing 'movs r0, #0' below the
test instruction:

	$CMP	r0, r1
	mov.w	r0, #0
	it	eq
	moveq	r0, #1
	bx	lr

So far, only physreg coalescing can do that. The register allocators won't yet
split live ranges just to eliminate copies. They can learn, but this particular
problem is not likely to show up in real code. It only appears because r0 is
used for both the function argument and return value.

Modified:
    llvm/trunk/test/CodeGen/Thumb2/thumb2-cmn.ll
    llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp.ll
    llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp2.ll
    llvm/trunk/test/CodeGen/Thumb2/thumb2-teq.ll
    llvm/trunk/test/CodeGen/Thumb2/thumb2-teq2.ll
    llvm/trunk/test/CodeGen/Thumb2/thumb2-tst.ll
    llvm/trunk/test/CodeGen/Thumb2/thumb2-tst2.ll

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-cmn.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-cmn.ll?rev=130858&r1=130857&r2=130858&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-cmn.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-cmn.ll Wed May  4 14:02:07 2011
@@ -1,4 +1,7 @@
-; RUN: llc < %s -march=thumb -mattr=+thumb2 | FileCheck %s
+; RUN: llc < %s -march=thumb -mattr=+thumb2 -join-physregs | FileCheck %s
+
+; These tests implicitly depend on 'movs r0, #0' being rematerialized below the
+; test as 'mov.w r0, #0'. So far, that requires physreg joining.
 
 define i1 @f1(i32 %a, i32 %b) {
     %nb = sub i32 0, %b

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp.ll?rev=130858&r1=130857&r2=130858&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp.ll Wed May  4 14:02:07 2011
@@ -1,4 +1,7 @@
-; RUN: llc < %s -march=thumb -mattr=+thumb2 | FileCheck %s
+; RUN: llc < %s -march=thumb -mattr=+thumb2 -join-physregs | FileCheck %s
+
+; These tests implicitly depend on 'movs r0, #0' being rematerialized below the
+; test as 'mov.w r0, #0'. So far, that requires physreg joining.
 
 ; 0x000000bb = 187
 define i1 @f1(i32 %a) {

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp2.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp2.ll?rev=130858&r1=130857&r2=130858&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp2.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-cmp2.ll Wed May  4 14:02:07 2011
@@ -1,4 +1,7 @@
-; RUN: llc < %s -march=thumb -mattr=+thumb2 | FileCheck %s
+; RUN: llc < %s -march=thumb -mattr=+thumb2 -join-physregs | FileCheck %s
+
+; These tests implicitly depend on 'movs r0, #0' being rematerialized below the
+; test as 'mov.w r0, #0'. So far, that requires physreg joining.
 
 define i1 @f1(i32 %a, i32 %b) {
 ; CHECK: f1:

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-teq.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-teq.ll?rev=130858&r1=130857&r2=130858&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-teq.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-teq.ll Wed May  4 14:02:07 2011
@@ -1,5 +1,7 @@
-; RUN: llc < %s -march=thumb -mattr=+thumb2 | FileCheck %s
+; RUN: llc < %s -march=thumb -mattr=+thumb2 -join-physregs | FileCheck %s
 
+; These tests implicitly depend on 'movs r0, #0' being rematerialized below the
+; test as 'mov.w r0, #0'. So far, that requires physreg joining.
 
 ; 0x000000bb = 187
 define i1 @f1(i32 %a) {

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-teq2.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-teq2.ll?rev=130858&r1=130857&r2=130858&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-teq2.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-teq2.ll Wed May  4 14:02:07 2011
@@ -1,4 +1,7 @@
-; RUN: llc < %s -march=thumb -mattr=+thumb2 | FileCheck %s
+; RUN: llc < %s -march=thumb -mattr=+thumb2 -join-physregs | FileCheck %s
+
+; These tests implicitly depend on 'movs r0, #0' being rematerialized below the
+; tst as 'mov.w r0, #0'. So far, that requires physreg joining.
 
 define i1 @f1(i32 %a, i32 %b) {
 ; CHECK: f1

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-tst.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-tst.ll?rev=130858&r1=130857&r2=130858&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-tst.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-tst.ll Wed May  4 14:02:07 2011
@@ -1,5 +1,7 @@
-; RUN: llc < %s -march=thumb -mattr=+thumb2 | FileCheck %s
+; RUN: llc < %s -march=thumb -mattr=+thumb2 -join-physregs | FileCheck %s
 
+; These tests implicitly depend on 'movs r0, #0' being rematerialized below the
+; tst as 'mov.w r0, #0'. So far, that requires physreg joining.
 
 ; 0x000000bb = 187
 define i1 @f1(i32 %a) {

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-tst2.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-tst2.ll?rev=130858&r1=130857&r2=130858&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-tst2.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-tst2.ll Wed May  4 14:02:07 2011
@@ -1,4 +1,7 @@
-; RUN: llc < %s -march=thumb -mattr=+thumb2 | FileCheck %s
+; RUN: llc < %s -march=thumb -mattr=+thumb2 -join-physregs | FileCheck %s
+
+; These tests implicitly depend on 'movs r0, #0' being rematerialized below the
+; tst as 'mov.w r0, #0'. So far, that requires physreg joining.
 
 define i1 @f1(i32 %a, i32 %b) {
 ; CHECK: f1:



From stoklund at 2pi.dk  Wed May  4 14:02:11 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 19:02:11 -0000
Subject: [llvm-commits] [llvm] r130859 - in /llvm/trunk/test/CodeGen:
 SPARC/2011-01-22-SRet.ll Thumb/2009-08-20-ISelBug.ll
 Thumb2/2010-08-10-VarSizedAllocaBug.ll Thumb2/bfi.ll
Message-ID: <20110504190211.3F63F2A6C12D@llvm.org>

Author: stoklund
Date: Wed May  4 14:02:11 2011
New Revision: 130859

URL: http://llvm.org/viewvc/llvm-project?rev=130859&view=rev
Log:
Fix more register and coalescing dependencies.

Modified:
    llvm/trunk/test/CodeGen/SPARC/2011-01-22-SRet.ll
    llvm/trunk/test/CodeGen/Thumb/2009-08-20-ISelBug.ll
    llvm/trunk/test/CodeGen/Thumb2/2010-08-10-VarSizedAllocaBug.ll
    llvm/trunk/test/CodeGen/Thumb2/bfi.ll

Modified: llvm/trunk/test/CodeGen/SPARC/2011-01-22-SRet.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/SPARC/2011-01-22-SRet.ll?rev=130859&r1=130858&r2=130859&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/SPARC/2011-01-22-SRet.ll (original)
+++ llvm/trunk/test/CodeGen/SPARC/2011-01-22-SRet.ll Wed May  4 14:02:11 2011
@@ -6,7 +6,6 @@
 entry:
 ;CHECK: make_foo
 ;CHECK: ld [%fp+64], {{.+}}
-;CHECK: or {{.+}}, {{.+}}, %i0
 ;CHECK: jmp %i7+12
   %0 = getelementptr inbounds %struct.foo_t* %agg.result, i32 0, i32 0
   store i32 %a, i32* %0, align 4

Modified: llvm/trunk/test/CodeGen/Thumb/2009-08-20-ISelBug.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb/2009-08-20-ISelBug.ll?rev=130859&r1=130858&r2=130859&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb/2009-08-20-ISelBug.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb/2009-08-20-ISelBug.ll Wed May  4 14:02:11 2011
@@ -11,7 +11,7 @@
 
 define i32 @t(%struct.asl_file_t* %s, i64 %off, i64* %out) nounwind optsize {
 ; CHECK: t:
-; CHECK: adds r0, #8
+; CHECK: adds {{r[0-7]}}, #8
 entry:
   %val = alloca i64, align 4                      ; <i64*> [#uses=3]
   %0 = icmp eq %struct.asl_file_t* %s, null       ; <i1> [#uses=1]

Modified: llvm/trunk/test/CodeGen/Thumb2/2010-08-10-VarSizedAllocaBug.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/2010-08-10-VarSizedAllocaBug.ll?rev=130859&r1=130858&r2=130859&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/2010-08-10-VarSizedAllocaBug.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/2010-08-10-VarSizedAllocaBug.ll Wed May  4 14:02:11 2011
@@ -7,8 +7,8 @@
 ; CHECK: Callee:
 ; CHECK: push
 ; CHECK: mov r4, sp
-; CHECK: sub.w r12, r4, #1000
-; CHECK: mov sp, r12
+; CHECK: sub.w [[R12:r[0-9]+]], r4, #1000
+; CHECK: mov sp, [[R12]]
   %0 = icmp eq i32 %i, 0                          ; <i1> [#uses=1]
   br i1 %0, label %bb2, label %bb
 

Modified: llvm/trunk/test/CodeGen/Thumb2/bfi.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/bfi.ll?rev=130859&r1=130858&r2=130859&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/bfi.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/bfi.ll Wed May  4 14:02:11 2011
@@ -30,9 +30,8 @@
 define i32 @f3(i32 %A, i32 %B) nounwind readnone optsize {
 entry:
 ; CHECK: f3
-; CHECK: lsrs  r2, r0, #7
-; CHECK: mov r0, r1
-; CHECK: bfi r0, r2, #7, #16
+; CHECK: lsrs {{.*}}, #7
+; CHECK: bfi {{.*}}, #7, #16
   %and = and i32 %A, 8388480                      ; <i32> [#uses=1]
   %and2 = and i32 %B, -8388481                    ; <i32> [#uses=1]
   %or = or i32 %and2, %and                        ; <i32> [#uses=1]
@@ -42,8 +41,8 @@
 ; rdar://8752056
 define i32 @f4(i32 %a) nounwind {
 ; CHECK: f4
-; CHECK: movw r1, #3137
-; CHECK: bfi r1, r0, #15, #5
+; CHECK: movw [[R1:r[0-9]+]], #3137
+; CHECK: bfi [[R1]], {{.*}}, #15, #5
   %1 = shl i32 %a, 15
   %ins7 = and i32 %1, 1015808
   %ins12 = or i32 %ins7, 3137



From baldrick at free.fr  Wed May  4 14:13:01 2011
From: baldrick at free.fr (Duncan Sands)
Date: Wed, 04 May 2011 19:13:01 -0000
Subject: [llvm-commits] [llvm] r130860 -
	/llvm/trunk/tools/gold/gold-plugin.cpp
Message-ID: <20110504191301.9102F2A6C12C@llvm.org>

Author: baldrick
Date: Wed May  4 14:13:01 2011
New Revision: 130860

URL: http://llvm.org/viewvc/llvm-project?rev=130860&view=rev
Log:
Remove an unused variable.

Modified:
    llvm/trunk/tools/gold/gold-plugin.cpp

Modified: llvm/trunk/tools/gold/gold-plugin.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/tools/gold/gold-plugin.cpp?rev=130860&r1=130859&r2=130860&view=diff
==============================================================================
--- llvm/trunk/tools/gold/gold-plugin.cpp (original)
+++ llvm/trunk/tools/gold/gold-plugin.cpp Wed May  4 14:13:01 2011
@@ -408,7 +408,6 @@
     if (options::generate_bc_file == options::BC_ONLY)
       exit(0);
   }
-  size_t bufsize = 0;
   const char *objPath;
   if (lto_codegen_compile_to_file(code_gen, &objPath)) {
     (*message)(LDPL_ERROR, "Could not produce a combined object file\n");



From eli.friedman at gmail.com  Wed May  4 14:54:24 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Wed, 04 May 2011 19:54:24 -0000
Subject: [llvm-commits] [llvm] r130862 - in /llvm/trunk:
 lib/CodeGen/MachineCSE.cpp test/CodeGen/Thumb2/thumb2-cbnz.ll
 test/CodeGen/X86/cmp-redundant.ll
Message-ID: <20110504195424.C53702A6C12C@llvm.org>

Author: efriedma
Date: Wed May  4 14:54:24 2011
New Revision: 130862

URL: http://llvm.org/viewvc/llvm-project?rev=130862&view=rev
Log:
Teach MachineCSE how to do simple cross-block CSE involving physregs.  This allows, for example, eliminating duplicate cmpl's on x86. Part of rdar://problem/8259436 .


Added:
    llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
Modified:
    llvm/trunk/lib/CodeGen/MachineCSE.cpp
    llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll

Modified: llvm/trunk/lib/CodeGen/MachineCSE.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineCSE.cpp?rev=130862&r1=130861&r2=130862&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/MachineCSE.cpp (original)
+++ llvm/trunk/lib/CodeGen/MachineCSE.cpp Wed May  4 14:54:24 2011
@@ -82,7 +82,8 @@
                                 MachineBasicBlock::const_iterator E) const ;
     bool hasLivePhysRegDefUses(const MachineInstr *MI,
                                const MachineBasicBlock *MBB,
-                               SmallSet<unsigned,8> &PhysRefs) const;
+                               SmallSet<unsigned,8> &PhysRefs,
+                               SmallVector<unsigned,8> &PhysDefs) const;
     bool PhysRegDefsReach(MachineInstr *CSMI, MachineInstr *MI,
                           SmallSet<unsigned,8> &PhysRefs) const;
     bool isCSECandidate(MachineInstr *MI);
@@ -189,7 +190,8 @@
 /// instruction does not uses a physical register.
 bool MachineCSE::hasLivePhysRegDefUses(const MachineInstr *MI,
                                        const MachineBasicBlock *MBB,
-                                       SmallSet<unsigned,8> &PhysRefs) const {
+                                       SmallSet<unsigned,8> &PhysRefs,
+                                       SmallVector<unsigned,8> &PhysDefs) const{
   MachineBasicBlock::const_iterator I = MI; I = llvm::next(I);
   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI->getOperand(i);
@@ -206,6 +208,7 @@
     if (MO.isDef() &&
         (MO.isDead() || isPhysDefTriviallyDead(Reg, I, MBB->end())))
       continue;
+    PhysDefs.push_back(Reg);
     PhysRefs.insert(Reg);
     for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias)
       PhysRefs.insert(*Alias);
@@ -216,35 +219,40 @@
 
 bool MachineCSE::PhysRegDefsReach(MachineInstr *CSMI, MachineInstr *MI,
                                   SmallSet<unsigned,8> &PhysRefs) const {
-  // For now conservatively returns false if the common subexpression is
-  // not in the same basic block as the given instruction.
-  MachineBasicBlock *MBB = MI->getParent();
-  if (CSMI->getParent() != MBB)
-    return false;
-  MachineBasicBlock::const_iterator I = CSMI; I = llvm::next(I);
-  MachineBasicBlock::const_iterator E = MI;
+  // Look backward from MI to find CSMI.
   unsigned LookAheadLeft = LookAheadLimit;
+  MachineBasicBlock::const_reverse_iterator I(MI);
+  MachineBasicBlock::const_reverse_iterator E(MI->getParent()->rend());
   while (LookAheadLeft) {
-    // Skip over dbg_value's.
-    while (I != E && I->isDebugValue())
-      ++I;
-
-    if (I == E)
-      return true;
+    while (LookAheadLeft && I != E) {
+      // Skip over dbg_value's.
+      while (I != E && I->isDebugValue())
+        ++I;
+
+      if (&*I == CSMI)
+        return true;
+
+      for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
+        const MachineOperand &MO = I->getOperand(i);
+        if (!MO.isReg() || !MO.isDef())
+          continue;
+        unsigned MOReg = MO.getReg();
+        if (TargetRegisterInfo::isVirtualRegister(MOReg))
+          continue;
+        if (PhysRefs.count(MOReg))
+          return false;
+      }
 
-    for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
-      const MachineOperand &MO = I->getOperand(i);
-      if (!MO.isReg() || !MO.isDef())
-        continue;
-      unsigned MOReg = MO.getReg();
-      if (TargetRegisterInfo::isVirtualRegister(MOReg))
-        continue;
-      if (PhysRefs.count(MOReg))
-        return false;
+      --LookAheadLeft;
+      ++I;
     }
-
-    --LookAheadLeft;
-    ++I;
+    // Go back another BB; for now, only go back at most one BB.
+    MachineBasicBlock *CSBB = CSMI->getParent();
+    MachineBasicBlock *BB = MI->getParent();
+    if (!CSBB->isSuccessor(BB) || BB->pred_size() != 1)
+      return false;
+    I = CSBB->rbegin();
+    E = CSBB->rend();
   }
 
   return false;
@@ -395,7 +403,8 @@
     // used, then it's not safe to replace it with a common subexpression.
     // It's also not safe if the instruction uses physical registers.
     SmallSet<unsigned,8> PhysRefs;
-    if (FoundCSE && hasLivePhysRegDefUses(MI, MBB, PhysRefs)) {
+    SmallVector<unsigned,8> DirectPhysRefs;
+    if (FoundCSE && hasLivePhysRegDefUses(MI, MBB, PhysRefs, DirectPhysRefs)) {
       FoundCSE = false;
 
       // ... Unless the CS is local and it also defines the physical register
@@ -448,6 +457,13 @@
         MRI->clearKillFlags(CSEPairs[i].second);
       }
       MI->eraseFromParent();
+      if (!DirectPhysRefs.empty() && CSMI->getParent() != MBB) {
+        assert(CSMI->getParent()->isSuccessor(MBB));
+        SmallVector<unsigned,8>::iterator PI = DirectPhysRefs.begin(),
+                                          PE = DirectPhysRefs.end();
+        for (; PI != PE; ++PI)
+          MBB->addLiveIn(*PI);
+      }
       ++NumCSEs;
       if (!PhysRefs.empty())
         ++NumPhysCSEs;

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll?rev=130862&r1=130861&r2=130862&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll Wed May  4 14:54:24 2011
@@ -21,8 +21,8 @@
 
 bb9:                                              ; preds = %bb7
 ; CHECK:      cmp r0, #0
-; CHECK:      cmp r0, #0
-; CHECK-NEXT: cbnz
+; CHECK-NOT: cmp
+; CHECK: cbnz
   %0 = tail call  double @floor(double %b) nounwind readnone ; <double> [#uses=0]
   br label %bb11
 

Added: llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/cmp-redundant.ll?rev=130862&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/cmp-redundant.ll (added)
+++ llvm/trunk/test/CodeGen/X86/cmp-redundant.ll Wed May  4 14:54:24 2011
@@ -0,0 +1,22 @@
+; RUN: llc < %s -mtriple=x86_64-apple-darwin10 | FileCheck %s
+
+define i32 @cmp(i32* %aa, i32* %bb) nounwind readnone ssp {
+entry:
+  %a = load i32* %aa
+  %b = load i32* %bb
+  %cmp = icmp sgt i32 %a, %b
+  br i1 %cmp, label %return, label %if.end
+; CHECK: cmp:
+; CHECK: cmpl
+; CHECK: jg
+if.end:                                           ; preds = %entry
+; CHECK-NOT: cmpl
+; CHECK: cmov
+  %cmp4 = icmp slt i32 %a, %b
+  %. = select i1 %cmp4, i32 2, i32 111
+  br label %return
+
+return:                                           ; preds = %if.end, %entry
+  %retval.0 = phi i32 [ 1, %entry ], [ %., %if.end ]
+  ret i32 %retval.0
+}



From dmalyshev at accesssoftek.com  Wed May  4 15:05:27 2011
From: dmalyshev at accesssoftek.com (Danil Malyshev)
Date: Wed, 4 May 2011 13:05:27 -0700
Subject: [llvm-commits] Check output stream opened in binary mode instead of
 text mode on MinGW
Message-ID: <6AE1604EE3EC5F4296C096518C6B77EE17E4EAF874@mail.accesssoftek.com>

Hello everyone,

Many LLVM tests relies on standard output stream be in the binary mode.
Which is not always the case (on Windows in particular).

The proposed patch adds a test to verify that the standard output stream is actually in the binary mode.
This will save us a lot of troubleshooting time.

Please review.


Thank you,
Danil
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110504/3415d5de/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test-archive-binary-output-03.diff
Type: application/octet-stream
Size: 467 bytes
Desc: test-archive-binary-output-03.diff
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110504/3415d5de/attachment.obj 

From baldrick at free.fr  Wed May  4 15:16:08 2011
From: baldrick at free.fr (Duncan Sands)
Date: Wed, 04 May 2011 20:16:08 -0000
Subject: [llvm-commits] [dragonegg] r130863 - in /dragonegg/trunk: README
	src/Backend.cpp
Message-ID: <20110504201608.739932A6C12C@llvm.org>

Author: baldrick
Date: Wed May  4 15:16:08 2011
New Revision: 130863

URL: http://llvm.org/viewvc/llvm-project?rev=130863&view=rev
Log:
Remove the -fplugin-arg-dragonegg-disable-llvm-optzns option, and instead
make it possible to specify any desired LLVM optimization level using the
-fplugin-arg-dragonegg-llvm-ir-optimize=N and
-fplugin-arg-dragonegg-llvm-codegen-optimize=N flags.

Modified:
    dragonegg/trunk/README
    dragonegg/trunk/src/Backend.cpp

Modified: dragonegg/trunk/README
URL: http://llvm.org/viewvc/llvm-project/dragonegg/trunk/README?rev=130863&r1=130862&r2=130863&view=diff
==============================================================================
--- dragonegg/trunk/README (original)
+++ dragonegg/trunk/README Wed May  4 15:16:08 2011
@@ -125,8 +125,19 @@
 -fplugin-arg-dragonegg-debug-pass-structure
   Output information about the passes being run.
 
--fplugin-arg-dragonegg-disable-llvm-optzns
-  Do not perform any LLVM IR optimizations even if compiling at -O1, -O2 etc.
+-fplugin-arg-dragonegg-llvm-ir-optimize=N
+  Run the LLVM IR optimizers at optimization level N, overriding the GCC
+  optimization level.  Usually if you pass -O1, -O2 etc to GCC then the
+  LLVM IR level optimizers are also run at -O1, -O2 etc.  Use this option
+  to change this, disassociating the LLVM optimization level from the GCC
+  one.  For example, -fplugin-arg-dragonegg-llvm-ir-optimize=0 disables
+  all LLVM IR optimizations.
+
+-fplugin-arg-dragonegg-llvm-codegen-optimize=N
+  Run the LLVM code generator optimizers at optimization level N, overriding
+  the GCC optimization level.  Usually if you pass -O1, -O2 etc to GCC then
+  the LLVM code generators optimize at a corresponding level.  Use this option
+  to change this, disassociating the LLVM optimization level from the GCC one.
 
 -fplugin-arg-dragonegg-enable-gcc-optzns
   Run the GCC tree optimizers as well as the LLVM IR optimizers.  Only early GCC

Modified: dragonegg/trunk/src/Backend.cpp
URL: http://llvm.org/viewvc/llvm-project/dragonegg/trunk/src/Backend.cpp?rev=130863&r1=130862&r2=130863&view=diff
==============================================================================
--- dragonegg/trunk/src/Backend.cpp (original)
+++ dragonegg/trunk/src/Backend.cpp Wed May  4 15:16:08 2011
@@ -103,10 +103,11 @@
 
 static bool DebugPassArguments;
 static bool DebugPassStructure;
-static bool DisableLLVMOptimizations;
 static bool EnableGCCOptimizations;
 static bool EmitIR;
 static bool SaveGCCOutput;
+static int LLVMCodeGenOptimizeArg = -1;
+static int LLVMIROptimizeArg = -1;
 
 std::vector<std::pair<Constant*, int> > StaticCtors, StaticDtors;
 SmallSetVector<Constant*, 32> AttributeUsedGlobals;
@@ -265,6 +266,24 @@
   }
 }
 
+/// CodeGenOptLevel - The optimization level to be used by the code generators.
+static CodeGenOpt::Level CodeGenOptLevel() {
+  int OptLevel = LLVMCodeGenOptimizeArg >= 0 ?
+    LLVMCodeGenOptimizeArg : optimize;
+  if (OptLevel <= 0)
+    return CodeGenOpt::None;
+  if (OptLevel == 1)
+    return CodeGenOpt::Less;
+  if (OptLevel == 2) // Includes -Os.
+    return CodeGenOpt::Default;
+  return CodeGenOpt::Aggressive;
+}
+
+/// IROptLevel - The optimization level to be used by the IR level optimizers.
+static int IROptLevel() {
+  return LLVMIROptimizeArg >= 0 ? LLVMIROptimizeArg : optimize;
+}
+
 // GuessAtInliningThreshold - Figure out a reasonable threshold to pass llvm's
 // inliner.  gcc has many options that control inlining, but we have decided
 // not to support anything like that for llvm-gcc.
@@ -273,7 +292,7 @@
     // Reduce inline limit.
     return 75;
 
-  if (optimize >= 3)
+  if (IROptLevel() >= 3)
     return 275;
   return 225;
 }
@@ -688,7 +707,7 @@
   HasPerFunctionPasses = true;
 #endif
 
-  if (optimize > 0 && !DisableLLVMOptimizations) {
+  if (IROptLevel() > 0) {
     HasPerFunctionPasses = true;
 
     TargetLibraryInfo *TLI =
@@ -698,7 +717,7 @@
     PerFunctionPasses->add(TLI);
 
     PerFunctionPasses->add(createCFGSimplificationPass());
-    if (optimize == 1)
+    if (IROptLevel() == 1)
       PerFunctionPasses->add(createPromoteMemoryToRegisterPass());
     else
       PerFunctionPasses->add(createScalarReplAggregatesPass());
@@ -715,15 +734,6 @@
     FunctionPassManager *PM = PerFunctionPasses;
     HasPerFunctionPasses = true;
 
-    CodeGenOpt::Level OptLevel = CodeGenOpt::Default;  // -O2, -Os, and -Oz
-    if (optimize == 0)
-      OptLevel = CodeGenOpt::None;
-    else if (optimize == 1)
-      OptLevel = CodeGenOpt::Less;
-    else if (optimize == 3)
-      // -O3 and above.
-      OptLevel = CodeGenOpt::Aggressive;
-
     // Request that addPassesToEmitFile run the Verifier after running
     // passes which modify the IR.
 #ifndef NDEBUG
@@ -737,7 +747,7 @@
     InitializeOutputStreams(false);
     if (TheTarget->addPassesToEmitFile(*PM, FormattedOutStream,
                                        TargetMachine::CGFT_AssemblyFile,
-                                       OptLevel, DisableVerify))
+                                       CodeGenOptLevel(), DisableVerify))
       DieAbjectly("Error interfacing to target machine!");
   }
 
@@ -759,40 +769,38 @@
   PerModulePasses->add(new TargetData(*TheTarget->getTargetData()));
   bool HasPerModulePasses = false;
 
-  if (!DisableLLVMOptimizations) {
-    TargetLibraryInfo *TLI =
-      new TargetLibraryInfo(Triple(TheModule->getTargetTriple()));
-    if (flag_no_simplify_libcalls)
-      TLI->disableAllFunctions();
-    PerModulePasses->add(TLI);
-
-    bool NeedAlwaysInliner = false;
-    llvm::Pass *InliningPass = 0;
-    if (flag_inline_small_functions && !flag_no_inline) {
-      InliningPass = createFunctionInliningPass();    // Inline small functions
-    } else {
-      // If full inliner is not run, check if always-inline is needed to handle
-      // functions that are  marked as always_inline.
-      // TODO: Consider letting the GCC inliner do this.
-      for (Module::iterator I = TheModule->begin(), E = TheModule->end();
-           I != E; ++I)
-        if (I->hasFnAttr(Attribute::AlwaysInline)) {
-          NeedAlwaysInliner = true;
-          break;
-        }
-
-      if (NeedAlwaysInliner)
-        InliningPass = createAlwaysInlinerPass();  // Inline always_inline funcs
-    }
+  TargetLibraryInfo *TLI =
+    new TargetLibraryInfo(Triple(TheModule->getTargetTriple()));
+  if (flag_no_simplify_libcalls)
+    TLI->disableAllFunctions();
+  PerModulePasses->add(TLI);
+
+  bool NeedAlwaysInliner = false;
+  llvm::Pass *InliningPass = 0;
+  if (flag_inline_small_functions && !flag_no_inline) {
+    InliningPass = createFunctionInliningPass();    // Inline small functions
+  } else {
+    // If full inliner is not run, check if always-inline is needed to handle
+    // functions that are  marked as always_inline.
+    // TODO: Consider letting the GCC inliner do this.
+    for (Module::iterator I = TheModule->begin(), E = TheModule->end();
+         I != E; ++I)
+      if (I->hasFnAttr(Attribute::AlwaysInline)) {
+        NeedAlwaysInliner = true;
+        break;
+      }
 
-    HasPerModulePasses = true;
-    createStandardModulePasses(PerModulePasses, optimize,
-                               optimize_size,
-                               flag_unit_at_a_time, flag_unroll_loops,
-                               !flag_no_simplify_libcalls, flag_exceptions,
-                               InliningPass);
+    if (NeedAlwaysInliner)
+      InliningPass = createAlwaysInlinerPass();  // Inline always_inline funcs
   }
 
+  HasPerModulePasses = true;
+  createStandardModulePasses(PerModulePasses, IROptLevel(),
+                             optimize_size,
+                             flag_unit_at_a_time, flag_unroll_loops,
+                             !flag_no_simplify_libcalls, flag_exceptions,
+                             InliningPass);
+
   if (EmitIR && 0) {
     // Emit an LLVM .bc file to the output.  This is used when passed
     // -emit-llvm -c to the GCC driver.
@@ -817,14 +825,6 @@
         new FunctionPassManager(TheModule);
       PM->add(new TargetData(*TheTarget->getTargetData()));
 
-      CodeGenOpt::Level OptLevel = CodeGenOpt::Default;
-
-      switch (optimize) {
-      default: break;
-      case 0: OptLevel = CodeGenOpt::None; break;
-      case 3: OptLevel = CodeGenOpt::Aggressive; break;
-      }
-
       // Request that addPassesToEmitFile run the Verifier after running
       // passes which modify the IR.
 #ifndef NDEBUG
@@ -838,7 +838,7 @@
       InitializeOutputStreams(false);
       if (TheTarget->addPassesToEmitFile(*PM, FormattedOutStream,
                                          TargetMachine::CGFT_AssemblyFile,
-                                         OptLevel, DisableVerify))
+                                         CodeGenOptLevel(), DisableVerify))
         DieAbjectly("Error interfacing to target machine!");
     }
   }
@@ -2400,7 +2400,6 @@
 static FlagDescriptor PluginFlags[] = {
     { "debug-pass-structure", &DebugPassStructure},
     { "debug-pass-arguments", &DebugPassArguments},
-    { "disable-llvm-optzns", &DisableLLVMOptimizations },
     { "enable-gcc-optzns", &EnableGCCOptimizations },
     { "emit-ir", &EmitIR },
     { "save-gcc-output", &SaveGCCOutput },
@@ -2454,27 +2453,47 @@
     int argc = plugin_info->argc;
 
     for (int i = 0; i < argc; ++i) {
-      bool Found = false;
+      if (!strcmp (argv[i].key, "llvm-ir-optimize") ||
+          !strcmp (argv[i].key, "llvm-codegen-optimize")) {
+        if (!argv[i].value) {
+          error(G_("no value supplied for option '-fplugin-arg-%s-%s'"),
+                plugin_name, argv[i].key);
+          return 1;
+        }
+        if (argv[i].value[0] < '0' || argv[i].value[0] > '9' || argv[i].value[1]) {
+          error(G_("invalid option argument '-fplugin-arg-%s-%s=%s'"),
+                plugin_name, argv[i].key, argv[i].value);
+          return 1;
+        }
+        int OptLevel = argv[i].value[0] - '0';
+        if (argv[i].key[5] == 'i')
+          LLVMIROptimizeArg = OptLevel;
+        else
+          LLVMCodeGenOptimizeArg = OptLevel;
+        continue;
+      }
+
+      // All remaining options are flags, so complain if there is an argument.
+      if (argv[i].value) {
+        error(G_("invalid option argument '-fplugin-arg-%s-%s=%s'"),
+              plugin_name, argv[i].key, argv[i].value);
+        return 1;
+      }
 
       // Look for a matching flag.
-      for (FlagDescriptor *F = PluginFlags; F->Key; ++F) {
-        if (strcmp (argv[i].key, F->Key))
-          continue;
-
-        if (argv[i].value)
-          warning (0, G_("option '-fplugin-arg-%s-%s=%s' ignored"
-                         " (superfluous '=%s')"),
-                   plugin_name, argv[i].key, argv[i].value, argv[i].value);
-        else
+      bool Found = false;
+      for (FlagDescriptor *F = PluginFlags; F->Key; ++F)
+        if (!strcmp (argv[i].key, F->Key)) {
+          Found = true;
           *F->Flag = true;
+          break;
+        }
 
-        Found = true;
-        break;
+      if (!Found) {
+        error(G_("invalid option '-fplugin-arg-%s-%s'"),
+              plugin_name, argv[i].key);
+        return 1;
       }
-
-      if (!Found)
-        warning (0, G_("plugin %qs: unrecognized argument %qs ignored"),
-                 plugin_name, argv[i].key);
     }
   }
 



From eli.friedman at gmail.com  Wed May  4 15:48:42 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Wed, 04 May 2011 20:48:42 -0000
Subject: [llvm-commits] [llvm] r130867 - in /llvm/trunk:
 lib/CodeGen/MachineCSE.cpp test/CodeGen/Thumb2/thumb2-cbnz.ll
 test/CodeGen/X86/cmp-redundant.ll
Message-ID: <20110504204842.5DD162A6C12C@llvm.org>

Author: efriedma
Date: Wed May  4 15:48:42 2011
New Revision: 130867

URL: http://llvm.org/viewvc/llvm-project?rev=130867&view=rev
Log:
Back out r130862; it appears to be breaking bootstrap.


Removed:
    llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
Modified:
    llvm/trunk/lib/CodeGen/MachineCSE.cpp
    llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll

Modified: llvm/trunk/lib/CodeGen/MachineCSE.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineCSE.cpp?rev=130867&r1=130866&r2=130867&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/MachineCSE.cpp (original)
+++ llvm/trunk/lib/CodeGen/MachineCSE.cpp Wed May  4 15:48:42 2011
@@ -82,8 +82,7 @@
                                 MachineBasicBlock::const_iterator E) const ;
     bool hasLivePhysRegDefUses(const MachineInstr *MI,
                                const MachineBasicBlock *MBB,
-                               SmallSet<unsigned,8> &PhysRefs,
-                               SmallVector<unsigned,8> &PhysDefs) const;
+                               SmallSet<unsigned,8> &PhysRefs) const;
     bool PhysRegDefsReach(MachineInstr *CSMI, MachineInstr *MI,
                           SmallSet<unsigned,8> &PhysRefs) const;
     bool isCSECandidate(MachineInstr *MI);
@@ -190,8 +189,7 @@
 /// instruction does not uses a physical register.
 bool MachineCSE::hasLivePhysRegDefUses(const MachineInstr *MI,
                                        const MachineBasicBlock *MBB,
-                                       SmallSet<unsigned,8> &PhysRefs,
-                                       SmallVector<unsigned,8> &PhysDefs) const{
+                                       SmallSet<unsigned,8> &PhysRefs) const {
   MachineBasicBlock::const_iterator I = MI; I = llvm::next(I);
   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI->getOperand(i);
@@ -208,7 +206,6 @@
     if (MO.isDef() &&
         (MO.isDead() || isPhysDefTriviallyDead(Reg, I, MBB->end())))
       continue;
-    PhysDefs.push_back(Reg);
     PhysRefs.insert(Reg);
     for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias)
       PhysRefs.insert(*Alias);
@@ -219,40 +216,35 @@
 
 bool MachineCSE::PhysRegDefsReach(MachineInstr *CSMI, MachineInstr *MI,
                                   SmallSet<unsigned,8> &PhysRefs) const {
-  // Look backward from MI to find CSMI.
+  // For now conservatively returns false if the common subexpression is
+  // not in the same basic block as the given instruction.
+  MachineBasicBlock *MBB = MI->getParent();
+  if (CSMI->getParent() != MBB)
+    return false;
+  MachineBasicBlock::const_iterator I = CSMI; I = llvm::next(I);
+  MachineBasicBlock::const_iterator E = MI;
   unsigned LookAheadLeft = LookAheadLimit;
-  MachineBasicBlock::const_reverse_iterator I(MI);
-  MachineBasicBlock::const_reverse_iterator E(MI->getParent()->rend());
   while (LookAheadLeft) {
-    while (LookAheadLeft && I != E) {
-      // Skip over dbg_value's.
-      while (I != E && I->isDebugValue())
-        ++I;
-
-      if (&*I == CSMI)
-        return true;
-
-      for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
-        const MachineOperand &MO = I->getOperand(i);
-        if (!MO.isReg() || !MO.isDef())
-          continue;
-        unsigned MOReg = MO.getReg();
-        if (TargetRegisterInfo::isVirtualRegister(MOReg))
-          continue;
-        if (PhysRefs.count(MOReg))
-          return false;
-      }
-
-      --LookAheadLeft;
+    // Skip over dbg_value's.
+    while (I != E && I->isDebugValue())
       ++I;
+
+    if (I == E)
+      return true;
+
+    for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
+      const MachineOperand &MO = I->getOperand(i);
+      if (!MO.isReg() || !MO.isDef())
+        continue;
+      unsigned MOReg = MO.getReg();
+      if (TargetRegisterInfo::isVirtualRegister(MOReg))
+        continue;
+      if (PhysRefs.count(MOReg))
+        return false;
     }
-    // Go back another BB; for now, only go back at most one BB.
-    MachineBasicBlock *CSBB = CSMI->getParent();
-    MachineBasicBlock *BB = MI->getParent();
-    if (!CSBB->isSuccessor(BB) || BB->pred_size() != 1)
-      return false;
-    I = CSBB->rbegin();
-    E = CSBB->rend();
+
+    --LookAheadLeft;
+    ++I;
   }
 
   return false;
@@ -403,8 +395,7 @@
     // used, then it's not safe to replace it with a common subexpression.
     // It's also not safe if the instruction uses physical registers.
     SmallSet<unsigned,8> PhysRefs;
-    SmallVector<unsigned,8> DirectPhysRefs;
-    if (FoundCSE && hasLivePhysRegDefUses(MI, MBB, PhysRefs, DirectPhysRefs)) {
+    if (FoundCSE && hasLivePhysRegDefUses(MI, MBB, PhysRefs)) {
       FoundCSE = false;
 
       // ... Unless the CS is local and it also defines the physical register
@@ -457,13 +448,6 @@
         MRI->clearKillFlags(CSEPairs[i].second);
       }
       MI->eraseFromParent();
-      if (!DirectPhysRefs.empty() && CSMI->getParent() != MBB) {
-        assert(CSMI->getParent()->isSuccessor(MBB));
-        SmallVector<unsigned,8>::iterator PI = DirectPhysRefs.begin(),
-                                          PE = DirectPhysRefs.end();
-        for (; PI != PE; ++PI)
-          MBB->addLiveIn(*PI);
-      }
       ++NumCSEs;
       if (!PhysRefs.empty())
         ++NumPhysCSEs;

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll?rev=130867&r1=130866&r2=130867&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll Wed May  4 15:48:42 2011
@@ -21,8 +21,8 @@
 
 bb9:                                              ; preds = %bb7
 ; CHECK:      cmp r0, #0
-; CHECK-NOT: cmp
-; CHECK: cbnz
+; CHECK:      cmp r0, #0
+; CHECK-NEXT: cbnz
   %0 = tail call  double @floor(double %b) nounwind readnone ; <double> [#uses=0]
   br label %bb11
 

Removed: llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/cmp-redundant.ll?rev=130866&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/cmp-redundant.ll (original)
+++ llvm/trunk/test/CodeGen/X86/cmp-redundant.ll (removed)
@@ -1,22 +0,0 @@
-; RUN: llc < %s -mtriple=x86_64-apple-darwin10 | FileCheck %s
-
-define i32 @cmp(i32* %aa, i32* %bb) nounwind readnone ssp {
-entry:
-  %a = load i32* %aa
-  %b = load i32* %bb
-  %cmp = icmp sgt i32 %a, %b
-  br i1 %cmp, label %return, label %if.end
-; CHECK: cmp:
-; CHECK: cmpl
-; CHECK: jg
-if.end:                                           ; preds = %entry
-; CHECK-NOT: cmpl
-; CHECK: cmov
-  %cmp4 = icmp slt i32 %a, %b
-  %. = select i1 %cmp4, i32 2, i32 111
-  br label %return
-
-return:                                           ; preds = %if.end, %entry
-  %retval.0 = phi i32 [ 1, %entry ], [ %., %if.end ]
-  ret i32 %retval.0
-}



From aggarwa4 at illinois.edu  Wed May  4 16:20:18 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Wed, 04 May 2011 21:20:18 -0000
Subject: [llvm-commits] [poolalloc] r130868 - in /poolalloc/trunk:
 include/dsa/DSNode.h lib/DSA/Local.cpp
Message-ID: <20110504212018.80FF22A6C12C@llvm.org>

Author: aggarwa4
Date: Wed May  4 16:20:18 2011
New Revision: 130868

URL: http://llvm.org/viewvc/llvm-project?rev=130868&view=rev
Log:
Allow PtrToInt and IntToPtr values to be used
locally for compares, without setting those flags.
This should help improve type safety for some
subset of the program in SAFECode

Modified:
    poolalloc/trunk/include/dsa/DSNode.h
    poolalloc/trunk/lib/DSA/Local.cpp

Modified: poolalloc/trunk/include/dsa/DSNode.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/dsa/DSNode.h?rev=130868&r1=130867&r2=130868&view=diff
==============================================================================
--- poolalloc/trunk/include/dsa/DSNode.h (original)
+++ poolalloc/trunk/include/dsa/DSNode.h Wed May  4 16:20:18 2011
@@ -103,8 +103,11 @@
     ArrayNode       = 1 << 9,   // This node is treated like an array
     CollapsedNode   = 1 << 10,  // This node is collapsed
     ExternalNode    = 1 << 11,  // This node comes from an external source
-    IntToPtrNode    = 1 << 12,   // This node comes from an int cast
-    PtrToIntNode    = 1 << 13,  // This node escapes to an int cast
+    IntToPtrNode    = 1 << 12,  // This node comes from an int cast
+                                // and is used in pointer operations
+                                // like geps, loads, stores
+    PtrToIntNode    = 1 << 13,  // This node escapes to an int cast 
+                                // and DSA does not track it further.
     VAStartNode     = 1 << 14,  // This node is from a vastart call
 
     //#ifndef NDEBUG

Modified: poolalloc/trunk/lib/DSA/Local.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/DSA/Local.cpp?rev=130868&r1=130867&r2=130868&view=diff
==============================================================================
--- poolalloc/trunk/lib/DSA/Local.cpp (original)
+++ poolalloc/trunk/lib/DSA/Local.cpp Wed May  4 16:20:18 2011
@@ -433,12 +433,10 @@
 
 void GraphBuilder::visitIntToPtrInst(IntToPtrInst &I) {
   DSNode *N = createNode();
-  if(TypeInferenceOptimize) {
-    if(I.getNumUses() == 1) {
-      if(isa<ICmpInst>(I.use_begin())) {
-        NumBoringIntToPtr++;
-        return;
-      }
+  if(I.getNumUses() == 1) {
+    if(isa<ICmpInst>(I.use_begin())) {
+      NumBoringIntToPtr++;
+      return;
     }
   } else {
     N->setIntToPtrMarker();
@@ -449,30 +447,26 @@
 
 void GraphBuilder::visitPtrToIntInst(PtrToIntInst& I) {
   DSNode* N = getValueDest(I.getOperand(0)).getNode();
-  if(TypeInferenceOptimize) {
-    if(I.getNumUses() == 1) {
-      if(isa<ICmpInst>(I.use_begin())) {
-        NumBoringIntToPtr++;
-        return;
-      }
+  if(I.getNumUses() == 1) {
+    if(isa<ICmpInst>(I.use_begin())) {
+      NumBoringIntToPtr++;
+      return;
     }
   }
-  if(TypeInferenceOptimize) {
-    if(I.getNumUses() == 1) {
-      Value *V = dyn_cast<Value>(I.use_begin());
-      while(V && V->getNumUses() == 1) {
-        if(isa<LoadInst>(V))
-          break;
-        if(isa<StoreInst>(V))
-          break;
-        if(isa<CallInst>(V))
-          break;
-        V = dyn_cast<Value>(V->use_begin());
-      }
-      if(isa<BranchInst>(V)){
-        NumBoringIntToPtr++;
-        return;
-      }
+  if(I.getNumUses() == 1) {
+    Value *V = dyn_cast<Value>(I.use_begin());
+    while(V && V->getNumUses() == 1) {
+      if(isa<LoadInst>(V))
+        break;
+      if(isa<StoreInst>(V))
+        break;
+      if(isa<CallInst>(V))
+        break;
+      V = dyn_cast<Value>(V->use_begin());
+    }
+    if(isa<BranchInst>(V)){
+      NumBoringIntToPtr++;
+      return;
     }
   }
   if(N)



From vladimir.stefanovic at rt-rk.com  Wed May  4 12:01:43 2011
From: vladimir.stefanovic at rt-rk.com (Vladimir Stefanovic)
Date: Wed, 04 May 2011 19:01:43 +0200
Subject: [llvm-commits] [llvm] [PATCH] Mips JIT implementation
Message-ID: <4DC18677.20706@rt-rk.com>


-------------- next part --------------
A non-text attachment was scrubbed...
Name: Mips-JIT.diff
Type: text/x-patch
Size: 39036 bytes
Desc: not available
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110504/f41d63a9/attachment.bin 

From dpatel at apple.com  Wed May  4 16:37:05 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 21:37:05 -0000
Subject: [llvm-commits] [llvm] r130869 -
	/llvm/trunk/lib/Transforms/Scalar/LoopIdiomRecognize.cpp
Message-ID: <20110504213705.B676A2A6C12C@llvm.org>

Author: dpatel
Date: Wed May  4 16:37:05 2011
New Revision: 130869

URL: http://llvm.org/viewvc/llvm-project?rev=130869&view=rev
Log:
preserve line number info.

Modified:
    llvm/trunk/lib/Transforms/Scalar/LoopIdiomRecognize.cpp

Modified: llvm/trunk/lib/Transforms/Scalar/LoopIdiomRecognize.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/LoopIdiomRecognize.cpp?rev=130869&r1=130868&r2=130869&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/LoopIdiomRecognize.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/LoopIdiomRecognize.cpp Wed May  4 16:37:05 2011
@@ -589,14 +589,15 @@
   Value *NumBytes =
     Expander.expandCodeFor(NumBytesS, IntPtr, Preheader->getTerminator());
 
-  Value *NewCall =
+  CallInst *NewCall =
     Builder.CreateMemCpy(StoreBasePtr, LoadBasePtr, NumBytes,
                          std::min(SI->getAlignment(), LI->getAlignment()));
+  NewCall->setDebugLoc(SI->getDebugLoc());
 
   DEBUG(dbgs() << "  Formed memcpy: " << *NewCall << "\n"
                << "    from load ptr=" << *LoadEv << " at: " << *LI << "\n"
                << "    from store ptr=" << *StoreEv << " at: " << *SI << "\n");
-  (void)NewCall;
+  
 
   // Okay, the memset has been formed.  Zap the original store and anything that
   // feeds into it.



From dpatel at apple.com  Wed May  4 16:57:22 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 21:57:22 -0000
Subject: [llvm-commits] [llvm] r130874 -
	/llvm/trunk/include/llvm/Support/IRBuilder.h
Message-ID: <20110504215722.CDF272A6C12C@llvm.org>

Author: dpatel
Date: Wed May  4 16:57:22 2011
New Revision: 130874

URL: http://llvm.org/viewvc/llvm-project?rev=130874&view=rev
Log:
If builder is initialized using an instruction as insertion point, then use the instruction's debug location as current debug location.

Modified:
    llvm/trunk/include/llvm/Support/IRBuilder.h

Modified: llvm/trunk/include/llvm/Support/IRBuilder.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Support/IRBuilder.h?rev=130874&r1=130873&r2=130874&view=diff
==============================================================================
--- llvm/trunk/include/llvm/Support/IRBuilder.h (original)
+++ llvm/trunk/include/llvm/Support/IRBuilder.h Wed May  4 16:57:22 2011
@@ -324,6 +324,7 @@
   explicit IRBuilder(Instruction *IP)
     : IRBuilderBase(IP->getContext()), Folder() {
     SetInsertPoint(IP);
+    SetCurrentDebugLocation(IP->getDebugLoc());
   }
   
   IRBuilder(BasicBlock *TheBB, BasicBlock::iterator IP, const T& F)



From gkistanova at gmail.com  Wed May  4 16:57:45 2011
From: gkistanova at gmail.com (Galina Kistanova)
Date: Wed, 04 May 2011 21:57:45 -0000
Subject: [llvm-commits] [llvm] r130875 -
	/llvm/trunk/test/CodeGen/ARM/2009-09-09-fpcmp-ole.ll
Message-ID: <20110504215745.2F6B42A6C12C@llvm.org>

Author: gkistanova
Date: Wed May  4 16:57:44 2011
New Revision: 130875

URL: http://llvm.org/viewvc/llvm-project?rev=130875&view=rev
Log:
This test fails on ARM. The test shouldn't explicitly specify alignment (and alignment 4 is wrong) and requires hard-float.

Modified:
    llvm/trunk/test/CodeGen/ARM/2009-09-09-fpcmp-ole.ll

Modified: llvm/trunk/test/CodeGen/ARM/2009-09-09-fpcmp-ole.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/2009-09-09-fpcmp-ole.ll?rev=130875&r1=130874&r2=130875&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/2009-09-09-fpcmp-ole.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/2009-09-09-fpcmp-ole.ll Wed May  4 16:57:44 2011
@@ -1,16 +1,16 @@
-; RUN: llc -O1 -march=arm -mattr=+vfp2 < %s | FileCheck %s
+; RUN: llc -O1 -march=arm -mattr=+vfp2 -float-abi=hard < %s | FileCheck %s
 ; pr4939
 
 define void @test(double* %x, double* %y) nounwind {
-  %1 = load double* %x, align 4
-  %2 = load double* %y, align 4
+  %1 = load double* %x
+  %2 = load double* %y
   %3 = fsub double -0.000000e+00, %1
   %4 = fcmp ugt double %2, %3
   br i1 %4, label %bb1, label %bb2
 
 bb1:
 ;CHECK: vstrhi.64
-  store double %1, double* %y, align 4
+  store double %1, double* %y
   br label %bb2
 
 bb2:



From dpatel at apple.com  Wed May  4 16:58:58 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 21:58:58 -0000
Subject: [llvm-commits] [llvm] r130876 -
	/llvm/trunk/lib/Transforms/Scalar/MemCpyOptimizer.cpp
Message-ID: <20110504215858.5FE812A6C12C@llvm.org>

Author: dpatel
Date: Wed May  4 16:58:58 2011
New Revision: 130876

URL: http://llvm.org/viewvc/llvm-project?rev=130876&view=rev
Log:
Preserve line number info.

Modified:
    llvm/trunk/lib/Transforms/Scalar/MemCpyOptimizer.cpp

Modified: llvm/trunk/lib/Transforms/Scalar/MemCpyOptimizer.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/MemCpyOptimizer.cpp?rev=130876&r1=130875&r2=130876&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/MemCpyOptimizer.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/MemCpyOptimizer.cpp Wed May  4 16:58:58 2011
@@ -459,7 +459,10 @@
           for (unsigned i = 0, e = Range.TheStores.size(); i != e; ++i)
             dbgs() << *Range.TheStores[i] << '\n';
           dbgs() << "With: " << *AMemSet << '\n');
-    
+
+    if (!Range.TheStores.empty())
+      AMemSet->setDebugLoc(Range.TheStores[0]->getDebugLoc());
+
     // Zap all the stores.
     for (SmallVector<Instruction*, 16>::const_iterator
          SI = Range.TheStores.begin(),



From anton at korobeynikov.info  Wed May  4 17:13:24 2011
From: anton at korobeynikov.info (Anton Korobeynikov)
Date: Thu, 5 May 2011 02:13:24 +0400
Subject: [llvm-commits] [llvm] r130875 -
	/llvm/trunk/test/CodeGen/ARM/2009-09-09-fpcmp-ole.ll
In-Reply-To: <20110504215745.2F6B42A6C12C@llvm.org>
References: <20110504215745.2F6B42A6C12C@llvm.org>
Message-ID: <BANLkTincDWQWzHC2DnsX6cyqxGSYwB-whA@mail.gmail.com>

Hi Galina,

> This test fails on ARM. The test shouldn't explicitly specify alignment (and alignment 4 is wrong) and requires hard-float.
In fact it doesn't. The difference is just APCS vs AAPCS, so, explicit
target triplet would be fine :)

-- 
With best regards, Anton Korobeynikov
Faculty of Mathematics and Mechanics, Saint Petersburg State University

From eli.friedman at gmail.com  Wed May  4 17:10:37 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Wed, 04 May 2011 22:10:37 -0000
Subject: [llvm-commits] [llvm] r130877 - in /llvm/trunk:
 lib/CodeGen/MachineCSE.cpp test/CodeGen/Thumb2/thumb2-cbnz.ll
 test/CodeGen/X86/cmp-redundant.ll
Message-ID: <20110504221037.2B8C92A6C12C@llvm.org>

Author: efriedma
Date: Wed May  4 17:10:36 2011
New Revision: 130877

URL: http://llvm.org/viewvc/llvm-project?rev=130877&view=rev
Log:
Re-commit r130862 with a minor change to avoid an iterator running off the edge in some cases.

Original message:

Teach MachineCSE how to do simple cross-block CSE involving physregs.  This allows, for example, eliminating duplicate cmpl's on x86. Part of rdar://problem/8259436 .


Added:
    llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
Modified:
    llvm/trunk/lib/CodeGen/MachineCSE.cpp
    llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll

Modified: llvm/trunk/lib/CodeGen/MachineCSE.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineCSE.cpp?rev=130877&r1=130876&r2=130877&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/MachineCSE.cpp (original)
+++ llvm/trunk/lib/CodeGen/MachineCSE.cpp Wed May  4 17:10:36 2011
@@ -33,6 +33,8 @@
 STATISTIC(NumCSEs,      "Number of common subexpression eliminated");
 STATISTIC(NumPhysCSEs,
           "Number of physreg referencing common subexpr eliminated");
+STATISTIC(NumCrossBlockPhysCSEs,
+          "Number of physreg common subexprs cross-block eliminated");
 STATISTIC(NumCommutes,  "Number of copies coalesced after commuting");
 
 namespace {
@@ -82,7 +84,8 @@
                                 MachineBasicBlock::const_iterator E) const ;
     bool hasLivePhysRegDefUses(const MachineInstr *MI,
                                const MachineBasicBlock *MBB,
-                               SmallSet<unsigned,8> &PhysRefs) const;
+                               SmallSet<unsigned,8> &PhysRefs,
+                               SmallVector<unsigned,8> &PhysDefs) const;
     bool PhysRegDefsReach(MachineInstr *CSMI, MachineInstr *MI,
                           SmallSet<unsigned,8> &PhysRefs) const;
     bool isCSECandidate(MachineInstr *MI);
@@ -189,7 +192,8 @@
 /// instruction does not uses a physical register.
 bool MachineCSE::hasLivePhysRegDefUses(const MachineInstr *MI,
                                        const MachineBasicBlock *MBB,
-                                       SmallSet<unsigned,8> &PhysRefs) const {
+                                       SmallSet<unsigned,8> &PhysRefs,
+                                       SmallVector<unsigned,8> &PhysDefs) const{
   MachineBasicBlock::const_iterator I = MI; I = llvm::next(I);
   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI->getOperand(i);
@@ -206,6 +210,7 @@
     if (MO.isDef() &&
         (MO.isDead() || isPhysDefTriviallyDead(Reg, I, MBB->end())))
       continue;
+    PhysDefs.push_back(Reg);
     PhysRefs.insert(Reg);
     for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias)
       PhysRefs.insert(*Alias);
@@ -216,35 +221,43 @@
 
 bool MachineCSE::PhysRegDefsReach(MachineInstr *CSMI, MachineInstr *MI,
                                   SmallSet<unsigned,8> &PhysRefs) const {
-  // For now conservatively returns false if the common subexpression is
-  // not in the same basic block as the given instruction.
-  MachineBasicBlock *MBB = MI->getParent();
-  if (CSMI->getParent() != MBB)
-    return false;
-  MachineBasicBlock::const_iterator I = CSMI; I = llvm::next(I);
-  MachineBasicBlock::const_iterator E = MI;
+  // Look backward from MI to find CSMI.
   unsigned LookAheadLeft = LookAheadLimit;
+  MachineBasicBlock *CurBB = MI->getParent();
+  MachineBasicBlock::const_reverse_iterator I(MI);
+  MachineBasicBlock::const_reverse_iterator E(CurBB->rend());
   while (LookAheadLeft) {
-    // Skip over dbg_value's.
-    while (I != E && I->isDebugValue())
-      ++I;
-
-    if (I == E)
-      return true;
+    while (LookAheadLeft && I != E) {
+      // Skip over dbg_value's.
+      while (I != E && I->isDebugValue())
+        ++I;
+
+      if (I == E) break;
+
+      if (&*I == CSMI)
+        return true;
+
+      for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
+        const MachineOperand &MO = I->getOperand(i);
+        if (!MO.isReg() || !MO.isDef())
+          continue;
+        unsigned MOReg = MO.getReg();
+        if (TargetRegisterInfo::isVirtualRegister(MOReg))
+          continue;
+        if (PhysRefs.count(MOReg))
+          return false;
+      }
 
-    for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
-      const MachineOperand &MO = I->getOperand(i);
-      if (!MO.isReg() || !MO.isDef())
-        continue;
-      unsigned MOReg = MO.getReg();
-      if (TargetRegisterInfo::isVirtualRegister(MOReg))
-        continue;
-      if (PhysRefs.count(MOReg))
-        return false;
+      --LookAheadLeft;
+      ++I;
     }
-
-    --LookAheadLeft;
-    ++I;
+    // Go back another BB; for now, only go back at most one BB.
+    MachineBasicBlock *CSBB = CSMI->getParent();
+    if (!CSBB->isSuccessor(CurBB) || CurBB->pred_size() != 1)
+      return false;
+    CurBB = CSBB;
+    I = CSBB->rbegin();
+    E = CSBB->rend();
   }
 
   return false;
@@ -395,7 +408,8 @@
     // used, then it's not safe to replace it with a common subexpression.
     // It's also not safe if the instruction uses physical registers.
     SmallSet<unsigned,8> PhysRefs;
-    if (FoundCSE && hasLivePhysRegDefUses(MI, MBB, PhysRefs)) {
+    SmallVector<unsigned,8> DirectPhysRefs;
+    if (FoundCSE && hasLivePhysRegDefUses(MI, MBB, PhysRefs, DirectPhysRefs)) {
       FoundCSE = false;
 
       // ... Unless the CS is local and it also defines the physical register
@@ -448,6 +462,14 @@
         MRI->clearKillFlags(CSEPairs[i].second);
       }
       MI->eraseFromParent();
+      if (!DirectPhysRefs.empty() && CSMI->getParent() != MBB) {
+        assert(CSMI->getParent()->isSuccessor(MBB));
+        ++NumCrossBlockPhysCSEs;
+        SmallVector<unsigned,8>::iterator PI = DirectPhysRefs.begin(),
+                                          PE = DirectPhysRefs.end();
+        for (; PI != PE; ++PI)
+          MBB->addLiveIn(*PI);
+      }
       ++NumCSEs;
       if (!PhysRefs.empty())
         ++NumPhysCSEs;

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll?rev=130877&r1=130876&r2=130877&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll Wed May  4 17:10:36 2011
@@ -21,8 +21,8 @@
 
 bb9:                                              ; preds = %bb7
 ; CHECK:      cmp r0, #0
-; CHECK:      cmp r0, #0
-; CHECK-NEXT: cbnz
+; CHECK-NOT: cmp
+; CHECK: cbnz
   %0 = tail call  double @floor(double %b) nounwind readnone ; <double> [#uses=0]
   br label %bb11
 

Added: llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/cmp-redundant.ll?rev=130877&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/cmp-redundant.ll (added)
+++ llvm/trunk/test/CodeGen/X86/cmp-redundant.ll Wed May  4 17:10:36 2011
@@ -0,0 +1,22 @@
+; RUN: llc < %s -mtriple=x86_64-apple-darwin10 | FileCheck %s
+
+define i32 @cmp(i32* %aa, i32* %bb) nounwind readnone ssp {
+entry:
+  %a = load i32* %aa
+  %b = load i32* %bb
+  %cmp = icmp sgt i32 %a, %b
+  br i1 %cmp, label %return, label %if.end
+; CHECK: cmp:
+; CHECK: cmpl
+; CHECK: jg
+if.end:                                           ; preds = %entry
+; CHECK-NOT: cmpl
+; CHECK: cmov
+  %cmp4 = icmp slt i32 %a, %b
+  %. = select i1 %cmp4, i32 2, i32 111
+  br label %return
+
+return:                                           ; preds = %if.end, %entry
+  %retval.0 = phi i32 [ 1, %entry ], [ %., %if.end ]
+  ret i32 %retval.0
+}



From nicholas at mxc.ca  Wed May  4 17:34:29 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Wed, 04 May 2011 22:34:29 -0000
Subject: [llvm-commits] [llvm] r130879 -
	/llvm/trunk/runtime/libprofile/GCDAProfiling.c
Message-ID: <20110504223430.089502A6C12C@llvm.org>

Author: nicholas
Date: Wed May  4 17:34:29 2011
New Revision: 130879

URL: http://llvm.org/viewvc/llvm-project?rev=130879&view=rev
Log:
Create the parent directories to place the .gcda files in if they don't exist.
That's kinda weird because the .gcno files are supposed to already be there,
but libgcov does this and somehow Google has managed to depend on it.

Modified:
    llvm/trunk/runtime/libprofile/GCDAProfiling.c

Modified: llvm/trunk/runtime/libprofile/GCDAProfiling.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/runtime/libprofile/GCDAProfiling.c?rev=130879&r1=130878&r2=130879&view=diff
==============================================================================
--- llvm/trunk/runtime/libprofile/GCDAProfiling.c (original)
+++ llvm/trunk/runtime/libprofile/GCDAProfiling.c Wed May  4 17:34:29 2011
@@ -24,6 +24,8 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <sys/stat.h>
+#include <sys/types.h>
 
 /* #define DEBUG_GCDAPROFILING */
 
@@ -64,6 +66,21 @@
   return filename;
 }
 
+static void recursive_mkdir(const char *filename) {
+  char *pathname;
+  int i, e;
+
+  for (i = 1, e = strlen(filename); i != e; ++i) {
+    if (filename[i] == '/') {
+      pathname = malloc(i + 1);
+      strncpy(pathname, filename, i);
+      pathname[i] = '\0';
+      mkdir(pathname, 0750);  /* some of these will fail, ignore it. */
+      free(pathname);
+    }
+  }
+}
+
 /*
  * --- LLVM line counter API ---
  */
@@ -75,6 +92,7 @@
 void llvm_gcda_start_file(const char *orig_filename) {
   char *filename;
   filename = mangle_filename(orig_filename);
+  recursive_mkdir(filename);
   output_file = fopen(filename, "wb");
 
   /* gcda file, version 404*, stamp LLVM. */



From dpatel at apple.com  Wed May  4 17:48:19 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 22:48:19 -0000
Subject: [llvm-commits] [llvm] r130880 -
	/llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp
Message-ID: <20110504224819.CC6C32A6C12C@llvm.org>

Author: dpatel
Date: Wed May  4 17:48:19 2011
New Revision: 130880

URL: http://llvm.org/viewvc/llvm-project?rev=130880&view=rev
Log:
Preserve line number information while threading jumps.

Modified:
    llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp

Modified: llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp?rev=130880&r1=130879&r2=130880&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/JumpThreading.cpp Wed May  4 17:48:19 2011
@@ -929,9 +929,10 @@
   if (UnavailablePred) {
     assert(UnavailablePred->getTerminator()->getNumSuccessors() == 1 &&
            "Can't handle critical edge here!");
-    Value *NewVal = new LoadInst(LoadedPtr, LI->getName()+".pr", false,
+    LoadInst *NewVal = new LoadInst(LoadedPtr, LI->getName()+".pr", false,
                                  LI->getAlignment(),
                                  UnavailablePred->getTerminator());
+    NewVal->setDebugLoc(LI->getDebugLoc());
     AvailablePreds.push_back(std::make_pair(UnavailablePred, NewVal));
   }
 
@@ -944,6 +945,7 @@
   PHINode *PN = PHINode::Create(LI->getType(), std::distance(PB, PE), "",
                                 LoadBB->begin());
   PN->takeName(LI);
+  PN->setDebugLoc(LI->getDebugLoc());
 
   // Insert new entries into the PHI for each predecessor.  A single block may
   // have multiple entries here.
@@ -1375,7 +1377,8 @@
 
   // We didn't copy the terminator from BB over to NewBB, because there is now
   // an unconditional jump to SuccBB.  Insert the unconditional jump.
-  BranchInst::Create(SuccBB, NewBB);
+  BranchInst *NewBI =BranchInst::Create(SuccBB, NewBB);
+  NewBI->setDebugLoc(BB->getTerminator()->getDebugLoc());
 
   // Check to see if SuccBB has PHI nodes. If so, we need to add entries to the
   // PHI nodes for NewBB now.



From isanbard at gmail.com  Wed May  4 17:54:05 2011
From: isanbard at gmail.com (Bill Wendling)
Date: Wed, 04 May 2011 22:54:05 -0000
Subject: [llvm-commits] [llvm] r130881 - in /llvm/trunk:
 lib/CodeGen/MachineVerifier.cpp
 test/CodeGen/ARM/2011-05-04-MultipleLandingPadSuccs.ll
Message-ID: <20110504225405.ABD3E2A6C12C@llvm.org>

Author: void
Date: Wed May  4 17:54:05 2011
New Revision: 130881

URL: http://llvm.org/viewvc/llvm-project?rev=130881&view=rev
Log:
SjLj EH could produce a machine basic block that legitimately has more than one
landing pad as its successor.

SjLj exception handling jumps to the correct landing pad via a switch statement
that's generated right before code-gen. Loosen the constraint in the machine
instruction verifier to allow for this. Note, this isn't the most rigorous check
since we cannot determine where that switch statement came from. But it's
marginally better than turning this check off when SjLj exceptions are used.
<rdar://problem/9187612>

Added:
    llvm/trunk/test/CodeGen/ARM/2011-05-04-MultipleLandingPadSuccs.ll
Modified:
    llvm/trunk/lib/CodeGen/MachineVerifier.cpp

Modified: llvm/trunk/lib/CodeGen/MachineVerifier.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineVerifier.cpp?rev=130881&r1=130880&r2=130881&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/MachineVerifier.cpp (original)
+++ llvm/trunk/lib/CodeGen/MachineVerifier.cpp Wed May  4 17:54:05 2011
@@ -23,6 +23,7 @@
 // the verifier errors.
 //===----------------------------------------------------------------------===//
 
+#include "llvm/Instructions.h"
 #include "llvm/Function.h"
 #include "llvm/CodeGen/LiveIntervalAnalysis.h"
 #include "llvm/CodeGen/LiveVariables.h"
@@ -32,6 +33,7 @@
 #include "llvm/CodeGen/MachineMemOperand.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/Passes.h"
+#include "llvm/MC/MCAsmInfo.h"
 #include "llvm/Target/TargetMachine.h"
 #include "llvm/Target/TargetRegisterInfo.h"
 #include "llvm/Target/TargetInstrInfo.h"
@@ -394,7 +396,13 @@
     if ((*I)->isLandingPad())
       LandingPadSuccs.insert(*I);
   }
-  if (LandingPadSuccs.size() > 1)
+
+  const MCAsmInfo *AsmInfo = TM->getMCAsmInfo();
+  const BasicBlock *BB = MBB->getBasicBlock();
+  if (LandingPadSuccs.size() > 1 &&
+      !(AsmInfo &&
+        AsmInfo->getExceptionHandlingType() == ExceptionHandling::SjLj &&
+        BB && isa<SwitchInst>(BB->getTerminator())))
     report("MBB has more than one landing pad successor", MBB);
 
   // Call AnalyzeBranch. If it succeeds, there several more conditions to check.

Added: llvm/trunk/test/CodeGen/ARM/2011-05-04-MultipleLandingPadSuccs.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/2011-05-04-MultipleLandingPadSuccs.ll?rev=130881&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/2011-05-04-MultipleLandingPadSuccs.ll (added)
+++ llvm/trunk/test/CodeGen/ARM/2011-05-04-MultipleLandingPadSuccs.ll Wed May  4 17:54:05 2011
@@ -0,0 +1,93 @@
+; RUN: llc < %s -verify-machineinstrs
+; <rdar://problem/9187612>
+target datalayout = "e-p:32:32:32-i1:8:32-i8:8:32-i16:16:32-i32:32:32-i64:32:32-f32:32:32-f64:32:32-v64:64:64-v128:128:128-a0:0:32-n32"
+target triple = "thumbv7-apple-darwin"
+
+define void @func() unnamed_addr align 2 {
+entry:
+  br label %for.cond
+
+for.cond:
+  %tmp2 = phi i32 [ 0, %entry ], [ %add, %for.cond.backedge ]
+  %cmp = icmp ult i32 %tmp2, 14
+  br i1 %cmp, label %for.body, label %for.end
+
+for.body:
+  %add = add i32 %tmp2, 1
+  switch i32 %tmp2, label %sw.default [
+    i32 0, label %sw.bb
+    i32 1, label %sw.bb
+    i32 2, label %sw.bb
+    i32 4, label %sw.bb
+    i32 5, label %sw.bb
+    i32 10, label %sw.bb
+  ]
+
+sw.bb:
+  invoke void @foo()
+          to label %invoke.cont17 unwind label %lpad
+
+invoke.cont17:
+  invoke void @foo()
+          to label %for.cond.backedge unwind label %lpad26
+
+for.cond.backedge:
+  br label %for.cond
+
+lpad:
+  %exn = tail call i8* @llvm.eh.exception() nounwind
+  %eh.selector = tail call i32 (i8*, i8*, ...)* @llvm.eh.selector(i8* %exn, i8* bitcast (i32 (...)* @__gxx_personality_sj0 to i8*), i8* null) nounwind
+  invoke void @foo()
+          to label %eh.resume unwind label %terminate.lpad
+
+lpad26:
+  %exn27 = tail call i8* @llvm.eh.exception() nounwind
+  %eh.selector28 = tail call i32 (i8*, i8*, ...)* @llvm.eh.selector(i8* %exn27, i8* bitcast (i32 (...)* @__gxx_personality_sj0 to i8*), i8* null) nounwind
+  invoke void @foo()
+          to label %eh.resume unwind label %terminate.lpad
+
+sw.default:
+  br label %for.cond.backedge
+
+for.end:
+  invoke void @foo()
+          to label %call8.i.i.i.noexc unwind label %lpad44
+
+call8.i.i.i.noexc:
+  ret void
+
+lpad44:
+  %exn45 = tail call i8* @llvm.eh.exception() nounwind
+  %eh.selector46 = tail call i32 (i8*, i8*, ...)* @llvm.eh.selector(i8* %exn45, i8* bitcast (i32 (...)* @__gxx_personality_sj0 to i8*), i8* null) nounwind
+  invoke void @foo()
+          to label %eh.resume unwind label %terminate.lpad
+
+eh.resume:
+  %exn.slot.0 = phi i8* [ %exn27, %lpad26 ], [ %exn, %lpad ], [ %exn45, %lpad44 ]
+  tail call void @_Unwind_SjLj_Resume_or_Rethrow(i8* %exn.slot.0) noreturn
+  unreachable
+
+terminate.lpad:
+  %exn51 = tail call i8* @llvm.eh.exception() nounwind
+  %eh.selector52 = tail call i32 (i8*, i8*, ...)* @llvm.eh.selector(i8* %exn51, i8* bitcast (i32 (...)* @__gxx_personality_sj0 to i8*), i8* null) nounwind
+  tail call void @_ZSt9terminatev() noreturn nounwind
+  unreachable
+}
+
+declare void @foo()
+
+declare i8* @llvm.eh.exception() nounwind readonly
+
+declare i32 @__gxx_personality_sj0(...)
+
+declare i32 @llvm.eh.selector(i8*, i8*, ...) nounwind
+
+declare void @_Unwind_SjLj_Resume_or_Rethrow(i8*)
+
+declare void @_ZSt9terminatev()
+
+!0 = metadata !{metadata !"any pointer", metadata !1}
+!1 = metadata !{metadata !"omnipotent char", metadata !2}
+!2 = metadata !{metadata !"Simple C/C++ TBAA", null}
+!3 = metadata !{metadata !"bool", metadata !1}
+!4 = metadata !{metadata !"int", metadata !1}



From gohman at apple.com  Wed May  4 18:14:02 2011
From: gohman at apple.com (Dan Gohman)
Date: Wed, 04 May 2011 23:14:02 -0000
Subject: [llvm-commits] [llvm] r130883 -
	/llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll
Message-ID: <20110504231402.AA0462A6C12C@llvm.org>

Author: djg
Date: Wed May  4 18:14:02 2011
New Revision: 130883

URL: http://llvm.org/viewvc/llvm-project?rev=130883&view=rev
Log:
Give this test an explicit register allocator, so that it can work even if
the default register allocator is changed.

Modified:
    llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll

Modified: llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll?rev=130883&r1=130882&r2=130883&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll (original)
+++ llvm/trunk/test/CodeGen/ARM/lsr-unfolded-offset.ll Wed May  4 18:14:02 2011
@@ -1,4 +1,4 @@
-; RUN: llc < %s | FileCheck %s
+; RUN: llc -regalloc=greedy < %s | FileCheck %s
 
 ; LSR shouldn't introduce more induction variables than needed, increasing
 ; register pressure and therefore spilling. There is more room for improvement



From kremenek at apple.com  Wed May  4 18:26:59 2011
From: kremenek at apple.com (Ted Kremenek)
Date: Wed, 04 May 2011 23:26:59 -0000
Subject: [llvm-commits] [llvm] r130885 -
	/llvm/trunk/include/llvm/Support/CrashRecoveryContext.h
Message-ID: <20110504232659.841E42A6C12C@llvm.org>

Author: kremenek
Date: Wed May  4 18:26:59 2011
New Revision: 130885

URL: http://llvm.org/viewvc/llvm-project?rev=130885&view=rev
Log:
Add explicit 'unregister' method to CrashRecoveryConextCleanupRegistrar.

Modified:
    llvm/trunk/include/llvm/Support/CrashRecoveryContext.h

Modified: llvm/trunk/include/llvm/Support/CrashRecoveryContext.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Support/CrashRecoveryContext.h?rev=130885&r1=130884&r2=130885&view=diff
==============================================================================
--- llvm/trunk/include/llvm/Support/CrashRecoveryContext.h (original)
+++ llvm/trunk/include/llvm/Support/CrashRecoveryContext.h Wed May  4 18:26:59 2011
@@ -186,8 +186,13 @@
   }
 
   ~CrashRecoveryContextCleanupRegistrar() {
+    unregister();
+  }
+  
+  void unregister() {
     if (cleanup && !cleanup->cleanupFired)
-        cleanup->getContext()->unregisterCleanup(cleanup);
+      cleanup->getContext()->unregisterCleanup(cleanup);
+    cleanup = 0;
   }
 };
 }



From isanbard at gmail.com  Wed May  4 18:40:14 2011
From: isanbard at gmail.com (Bill Wendling)
Date: Wed, 04 May 2011 23:40:14 -0000
Subject: [llvm-commits] [llvm] r130889 - /llvm/trunk/docs/CodeGenerator.html
Message-ID: <20110504234014.DE33B2A6C12C@llvm.org>

Author: void
Date: Wed May  4 18:40:14 2011
New Revision: 130889

URL: http://llvm.org/viewvc/llvm-project?rev=130889&view=rev
Log:
Document the automatic alias printing of InstAliases.

Modified:
    llvm/trunk/docs/CodeGenerator.html

Modified: llvm/trunk/docs/CodeGenerator.html
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/docs/CodeGenerator.html?rev=130889&r1=130888&r2=130889&view=diff
==============================================================================
--- llvm/trunk/docs/CodeGenerator.html (original)
+++ llvm/trunk/docs/CodeGenerator.html Wed May  4 18:40:14 2011
@@ -2019,6 +2019,11 @@
 <p>Instruction aliases can also have a Requires clause to make them
 subtarget specific.</p>
 
+<p>If the back-end supports it, the instruction printer can automatically emit
+   the alias rather than what's being aliased. It typically leads to better,
+   more readable code. If it's better to print out what's being aliased, then
+   pass a '0' as the third parameter to the InstAlias definition.</p>
+
 </div>
 
 </div>



From stoklund at 2pi.dk  Wed May  4 18:54:51 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 23:54:51 -0000
Subject: [llvm-commits] [llvm] r130891 - in /llvm/trunk/test/CodeGen/X86:
 sse3.ll sse42.ll tail-opts.ll use-add-flags.ll vec_shuffle-16.ll
 win64_alloca_dynalloca.ll x86-64-and-mask.ll
Message-ID: <20110504235451.C29F32A6C12C@llvm.org>

Author: stoklund
Date: Wed May  4 18:54:51 2011
New Revision: 130891

URL: http://llvm.org/viewvc/llvm-project?rev=130891&view=rev
Log:
Fix a batch of x86 tests to be coalescer independent.

Most of these tests require a single mov instruction that can come either before
or after a 2-addr instruction. -join-physregs changes the behavior, but the
results are equivalent.

Modified:
    llvm/trunk/test/CodeGen/X86/sse3.ll
    llvm/trunk/test/CodeGen/X86/sse42.ll
    llvm/trunk/test/CodeGen/X86/tail-opts.ll
    llvm/trunk/test/CodeGen/X86/use-add-flags.ll
    llvm/trunk/test/CodeGen/X86/vec_shuffle-16.ll
    llvm/trunk/test/CodeGen/X86/win64_alloca_dynalloca.ll
    llvm/trunk/test/CodeGen/X86/x86-64-and-mask.ll

Modified: llvm/trunk/test/CodeGen/X86/sse3.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/sse3.ll?rev=130891&r1=130890&r2=130891&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/sse3.ll (original)
+++ llvm/trunk/test/CodeGen/X86/sse3.ll Wed May  4 18:54:51 2011
@@ -62,11 +62,10 @@
 	%tmp = shufflevector <8 x i16> %A, <8 x i16> %B, <8 x i32> < i32 0, i32 7, i32 2, i32 3, i32 1, i32 5, i32 6, i32 5 >
 	ret <8 x i16> %tmp
 ; X64: t4:
-; X64: 	pextrw	$7, %xmm0, %eax
-; X64: 	pshufhw	$100, %xmm0, %xmm1
-; X64: 	pinsrw	$1, %eax, %xmm1
-; X64: 	pextrw	$1, %xmm0, %eax
-; X64: 	movdqa	%xmm1, %xmm0
+; X64: 	pextrw	$7, [[XMM0:%xmm[0-9]+]], %eax
+; X64: 	pshufhw	$100, [[XMM0]], [[XMM1:%xmm[0-9]+]]
+; X64: 	pinsrw	$1, %eax, [[XMM1]]
+; X64: 	pextrw	$1, [[XMM0]], %eax
 ; X64: 	pinsrw	$4, %eax, %xmm0
 ; X64: 	ret
 }
@@ -251,13 +250,13 @@
         %tmp9 = shufflevector <16 x i8> %tmp8, <16 x i8> %T0,  <16 x i32> < i32 0, i32 1, i32 2, i32 17,  i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef, i32 undef , i32 undef >
         ret <16 x i8> %tmp9
 ; X64: 	t16:
-; X64: 		pinsrw	$0, %eax, %xmm1
-; X64: 		pextrw	$8, %xmm0, %eax
-; X64: 		pinsrw	$1, %eax, %xmm1
-; X64: 		pextrw	$1, %xmm1, %ecx
-; X64: 		movd	%xmm1, %edx
-; X64: 		pinsrw	$0, %edx, %xmm1
-; X64: 		pinsrw	$1, %eax, %xmm0
+; X64: 		pinsrw	$0, %eax, [[X1:%xmm[0-9]+]]
+; X64: 		pextrw	$8, [[X0:%xmm[0-9]+]], %eax
+; X64: 		pinsrw	$1, %eax, [[X1]]
+; X64: 		pextrw	$1, [[X1]], %ecx
+; X64: 		movd	[[X1]], %edx
+; X64: 		pinsrw	$0, %edx, %xmm
+; X64: 		pinsrw	$1, %eax, %xmm
 ; X64: 		ret
 }
 

Modified: llvm/trunk/test/CodeGen/X86/sse42.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/sse42.ll?rev=130891&r1=130890&r2=130891&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/sse42.ll (original)
+++ llvm/trunk/test/CodeGen/X86/sse42.ll Wed May  4 18:54:51 2011
@@ -12,7 +12,7 @@
 ; X32:     crc32b   8(%esp), %eax
 
 ; X64: _crc32_8:
-; X64:     crc32b   %sil, %eax
+; X64:     crc32b   %sil,
 }
 
 
@@ -23,7 +23,7 @@
 ; X32:     crc32w   8(%esp), %eax
 
 ; X64: _crc32_16:
-; X64:     crc32w   %si, %eax
+; X64:     crc32w   %si,
 }
 
 
@@ -34,5 +34,5 @@
 ; X32:     crc32l   8(%esp), %eax
 
 ; X64: _crc32_32:
-; X64:     crc32l   %esi, %eax
+; X64:     crc32l   %esi,
 }

Modified: llvm/trunk/test/CodeGen/X86/tail-opts.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/tail-opts.ll?rev=130891&r1=130890&r2=130891&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/tail-opts.ll (original)
+++ llvm/trunk/test/CodeGen/X86/tail-opts.ll Wed May  4 18:54:51 2011
@@ -412,9 +412,9 @@
 ; can fall-through into the ret and the other side has to branch anyway.
 
 ; CHECK: TESTE:
-; CHECK: imulq
-; CHECK-NEXT: LBB8_2:
-; CHECK-NEXT: ret
+; CHECK: ret
+; CHECK-NOT: ret
+; CHECK: size TESTE
 
 define i64 @TESTE(i64 %parami, i64 %paraml) nounwind readnone {
 entry:

Modified: llvm/trunk/test/CodeGen/X86/use-add-flags.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/use-add-flags.ll?rev=130891&r1=130890&r2=130891&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/use-add-flags.ll (original)
+++ llvm/trunk/test/CodeGen/X86/use-add-flags.ll Wed May  4 18:54:51 2011
@@ -7,10 +7,10 @@
 ; Use the flags on the add.
 
 ; CHECK: test1:
-;      CHECK: addl    (%r[[A0:di|cx]]), {{%esi|%edx}}
-; CHECK-NEXT: movl    {{%edx|%r8d}}, %eax
-; CHECK-NEXT: cmovnsl {{%ecx|%r9d}}, %eax
-; CHECK-NEXT: ret
+;     CHECK: addl
+; CHECK-NOT: test
+;     CHECK: cmovnsl
+;     CHECK: ret
 
 define i32 @test1(i32* %x, i32 %y, i32 %a, i32 %b) nounwind {
 	%tmp2 = load i32* %x, align 4		; <i32> [#uses=1]
@@ -42,7 +42,7 @@
 ; Do use the flags result of the and here, since the and has another use.
 
 ; CHECK: test3:
-;      CHECK: andl    $16, %e[[A0]]
+;      CHECK: andl    $16, %e
 ; CHECK-NEXT: jne
 
 define void @test3(i32 %x) nounwind {

Modified: llvm/trunk/test/CodeGen/X86/vec_shuffle-16.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/vec_shuffle-16.ll?rev=130891&r1=130890&r2=130891&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/vec_shuffle-16.ll (original)
+++ llvm/trunk/test/CodeGen/X86/vec_shuffle-16.ll Wed May  4 18:54:51 2011
@@ -1,8 +1,9 @@
 ; RUN: llc < %s -march=x86 -mattr=+sse,-sse2 -mtriple=i386-apple-darwin | FileCheck %s -check-prefix=sse
 ; RUN: llc < %s -march=x86 -mattr=+sse2 -mtriple=i386-apple-darwin | FileCheck %s -check-prefix=sse2
 
+; sse:  t1:
+; sse2: t1:
 define <4 x float> @t1(<4 x float> %a, <4 x float> %b) nounwind  {
-; sse: movaps
 ; sse: shufps
 ; sse2: pshufd
 ; sse2-NEXT: ret
@@ -10,6 +11,8 @@
         ret <4 x float> %tmp1
 }
 
+; sse:  t2:
+; sse2: t2:
 define <4 x float> @t2(<4 x float> %A, <4 x float> %B) nounwind {
 ; sse: shufps
 ; sse2: pshufd
@@ -18,8 +21,9 @@
 	ret <4 x float> %tmp
 }
 
+; sse:  t3:
+; sse2: t3:
 define <4 x float> @t3(<4 x float> %A, <4 x float> %B) nounwind {
-; sse: movaps
 ; sse: shufps
 ; sse2: pshufd
 ; sse2-NEXT: ret
@@ -27,7 +31,10 @@
 	ret <4 x float> %tmp
 }
 
+; sse:  t4:
+; sse2: t4:
 define <4 x float> @t4(<4 x float> %A, <4 x float> %B) nounwind {
+
 ; sse: shufps
 ; sse2: pshufd
 ; sse2-NEXT: ret

Modified: llvm/trunk/test/CodeGen/X86/win64_alloca_dynalloca.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/win64_alloca_dynalloca.ll?rev=130891&r1=130890&r2=130891&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/win64_alloca_dynalloca.ll (original)
+++ llvm/trunk/test/CodeGen/X86/win64_alloca_dynalloca.ll Wed May  4 18:54:51 2011
@@ -1,9 +1,12 @@
-; RUN: llc < %s -mtriple=x86_64-mingw32     | FileCheck %s -check-prefix=M64
-; RUN: llc < %s -mtriple=x86_64-win32       | FileCheck %s -check-prefix=W64
-; RUN: llc < %s -mtriple=x86_64-win32-macho | FileCheck %s -check-prefix=EFI
+; RUN: llc < %s -join-physregs -mtriple=x86_64-mingw32     | FileCheck %s -check-prefix=M64
+; RUN: llc < %s -join-physregs -mtriple=x86_64-win32       | FileCheck %s -check-prefix=W64
+; RUN: llc < %s -join-physregs -mtriple=x86_64-win32-macho | FileCheck %s -check-prefix=EFI
 ; PR8777
 ; PR8778
 
+; Passing the same value in two registers creates a false interference that
+; only -join-physregs resolves. It could also be handled by a parallel copy.
+
 define i64 @foo(i64 %n, i64 %x) nounwind {
 entry:
 

Modified: llvm/trunk/test/CodeGen/X86/x86-64-and-mask.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/x86-64-and-mask.ll?rev=130891&r1=130890&r2=130891&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/x86-64-and-mask.ll (original)
+++ llvm/trunk/test/CodeGen/X86/x86-64-and-mask.ll Wed May  4 18:54:51 2011
@@ -39,7 +39,7 @@
 
 ; This requires a mov and a 64-bit and.
 ; CHECK: ddd:
-; CHECK: movabsq $4294967296, %rax
+; CHECK: movabsq $4294967296, %r
 ; CHECK: andq %rax, %rdi
 
 define void @ddd(i64 %x) nounwind {



From stoklund at 2pi.dk  Wed May  4 18:54:54 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 23:54:54 -0000
Subject: [llvm-commits] [llvm] r130892 -
	/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp
Message-ID: <20110504235454.85D702A6C12D@llvm.org>

Author: stoklund
Date: Wed May  4 18:54:54 2011
New Revision: 130892

URL: http://llvm.org/viewvc/llvm-project?rev=130892&view=rev
Log:
Fix X86RegisterInfo::getMatchingSuperRegClass for sub_8bit_hi.

It is OK for B to be any GR8_ABCD_H superclass, the returned register class
doesn't have to map surjectively onto B.

Modified:
    llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp

Modified: llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp?rev=130892&r1=130891&r2=130892&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86RegisterInfo.cpp Wed May  4 18:54:54 2011
@@ -229,19 +229,14 @@
     }
     break;
   case X86::sub_8bit_hi:
-    if (B == &X86::GR8_ABCD_HRegClass) {
-      if (A == &X86::GR64RegClass || A == &X86::GR64_ABCDRegClass ||
-          A == &X86::GR64_NOREXRegClass ||
-          A == &X86::GR64_NOSPRegClass ||
-          A == &X86::GR64_NOREX_NOSPRegClass)
-        return &X86::GR64_ABCDRegClass;
-      else if (A == &X86::GR32RegClass || A == &X86::GR32_ABCDRegClass ||
-               A == &X86::GR32_NOREXRegClass || A == &X86::GR32_NOSPRegClass)
-        return &X86::GR32_ABCDRegClass;
-      else if (A == &X86::GR16RegClass || A == &X86::GR16_ABCDRegClass ||
-               A == &X86::GR16_NOREXRegClass)
-        return &X86::GR16_ABCDRegClass;
-    }
+    if (B == &X86::GR8_ABCD_HRegClass ||
+        B->hasSubClass(&X86::GR8_ABCD_HRegClass))
+      switch (A->getSize()) {
+        case 2: return getCommonSubClass(A, &X86::GR16_ABCDRegClass);
+        case 4: return getCommonSubClass(A, &X86::GR32_ABCDRegClass);
+        case 8: return getCommonSubClass(A, &X86::GR64_ABCDRegClass);
+        default: return 0;
+      }
     break;
   case X86::sub_16bit:
     if (B == &X86::GR16RegClass) {



From stoklund at 2pi.dk  Wed May  4 18:54:59 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 23:54:59 -0000
Subject: [llvm-commits] [llvm] r130893 - in /llvm/trunk/test/CodeGen/X86:
 2008-12-19-EarlyClobberBug.ll 2009-03-13-PHIElimBug.ll abi-isel.ll add.ll
 andimm8.ll byval7.ll coalescer-commute2.ll dbg-value-range.ll
 optimize-max-3.ll phys_subreg_coalesce-2.ll phys_subreg_coalesce-3.ll
 pmul.ll pr2659.ll sibcall.ll sse-minmax.ll
Message-ID: <20110504235459.9CCB82A6C12C@llvm.org>

Author: stoklund
Date: Wed May  4 18:54:59 2011
New Revision: 130893

URL: http://llvm.org/viewvc/llvm-project?rev=130893&view=rev
Log:
Prepare remaining tests for -join-physreg going away.

Modified:
    llvm/trunk/test/CodeGen/X86/2008-12-19-EarlyClobberBug.ll
    llvm/trunk/test/CodeGen/X86/2009-03-13-PHIElimBug.ll
    llvm/trunk/test/CodeGen/X86/abi-isel.ll
    llvm/trunk/test/CodeGen/X86/add.ll
    llvm/trunk/test/CodeGen/X86/andimm8.ll
    llvm/trunk/test/CodeGen/X86/byval7.ll
    llvm/trunk/test/CodeGen/X86/coalescer-commute2.ll
    llvm/trunk/test/CodeGen/X86/dbg-value-range.ll
    llvm/trunk/test/CodeGen/X86/optimize-max-3.ll
    llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-2.ll
    llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-3.ll
    llvm/trunk/test/CodeGen/X86/pmul.ll
    llvm/trunk/test/CodeGen/X86/pr2659.ll
    llvm/trunk/test/CodeGen/X86/sibcall.ll
    llvm/trunk/test/CodeGen/X86/sse-minmax.ll

Modified: llvm/trunk/test/CodeGen/X86/2008-12-19-EarlyClobberBug.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2008-12-19-EarlyClobberBug.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/2008-12-19-EarlyClobberBug.ll (original)
+++ llvm/trunk/test/CodeGen/X86/2008-12-19-EarlyClobberBug.ll Wed May  4 18:54:59 2011
@@ -4,7 +4,7 @@
 
 ; CHECK:         ## InlineAsm End
 ; CHECK-NEXT: BB0_2:
-; CHECK-NEXT:    movl	%esi, %eax
+; CHECK-NEXT:    {{movl	%esi, %eax|addl	%edi, %esi}}
 
 
 @"\01LC" = internal constant [7 x i8] c"n0=%d\0A\00"		; <[7 x i8]*> [#uses=1]

Modified: llvm/trunk/test/CodeGen/X86/2009-03-13-PHIElimBug.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2009-03-13-PHIElimBug.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/2009-03-13-PHIElimBug.ll (original)
+++ llvm/trunk/test/CodeGen/X86/2009-03-13-PHIElimBug.ll Wed May  4 18:54:59 2011
@@ -28,5 +28,5 @@
 }
 
 ; CHECK: call{{.*}}f
-; CHECK-NEXT: Ltmp0:
-; CHECK-NEXT: movl %eax, %esi
+; CHECK: movl %eax, %esi
+; CHECK: call{{.*}}g

Modified: llvm/trunk/test/CodeGen/X86/abi-isel.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/abi-isel.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/abi-isel.ll (original)
+++ llvm/trunk/test/CodeGen/X86/abi-isel.ll Wed May  4 18:54:59 2011
@@ -12,16 +12,16 @@
 ; RUN: llc < %s -asm-verbose=0 -mtriple=x86_64-apple-darwin -march=x86-64 -relocation-model=dynamic-no-pic -code-model=small | FileCheck %s -check-prefix=DARWIN-64-DYNAMIC
 ; RUN: llc < %s -asm-verbose=0 -mtriple=x86_64-apple-darwin -march=x86-64 -relocation-model=pic -code-model=small | FileCheck %s -check-prefix=DARWIN-64-PIC
 
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=i686-unknown-linux-gnu -march=x86 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=LINUX-32-STATIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=i686-unknown-linux-gnu -march=x86 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=LINUX-32-PIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=x86_64-unknown-linux-gnu -march=x86-64 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=LINUX-64-STATIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=x86_64-unknown-linux-gnu -march=x86-64 -relocation-model=pic -code-model=small | FileCheck %s -check-prefix=LINUX-64-PIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=i686-apple-darwin -march=x86 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=DARWIN-32-STATIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=i686-apple-darwin -march=x86 -relocation-model=dynamic-no-pic -code-model=small | FileCheck %s -check-prefix=DARWIN-32-DYNAMIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=i686-apple-darwin -march=x86 -relocation-model=pic -code-model=small | FileCheck %s -check-prefix=DARWIN-32-PIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=x86_64-apple-darwin -march=x86-64 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=DARWIN-64-STATIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=x86_64-apple-darwin -march=x86-64 -relocation-model=dynamic-no-pic -code-model=small | FileCheck %s -check-prefix=DARWIN-64-DYNAMIC
-; RUN: llc < %s -asm-verbose=0 -regalloc=basic -mtriple=x86_64-apple-darwin -march=x86-64 -relocation-model=pic -code-model=small | FileCheck %s -check-prefix=DARWIN-64-PIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=i686-unknown-linux-gnu -march=x86 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=LINUX-32-STATIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=i686-unknown-linux-gnu -march=x86 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=LINUX-32-PIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=x86_64-unknown-linux-gnu -march=x86-64 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=LINUX-64-STATIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=x86_64-unknown-linux-gnu -march=x86-64 -relocation-model=pic -code-model=small | FileCheck %s -check-prefix=LINUX-64-PIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=i686-apple-darwin -march=x86 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=DARWIN-32-STATIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=i686-apple-darwin -march=x86 -relocation-model=dynamic-no-pic -code-model=small | FileCheck %s -check-prefix=DARWIN-32-DYNAMIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=i686-apple-darwin -march=x86 -relocation-model=pic -code-model=small | FileCheck %s -check-prefix=DARWIN-32-PIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=x86_64-apple-darwin -march=x86-64 -relocation-model=static -code-model=small | FileCheck %s -check-prefix=DARWIN-64-STATIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=x86_64-apple-darwin -march=x86-64 -relocation-model=dynamic-no-pic -code-model=small | FileCheck %s -check-prefix=DARWIN-64-DYNAMIC
+; RUN: llc < %s -asm-verbose=0 -regalloc=basic -join-physregs -mtriple=x86_64-apple-darwin -march=x86-64 -relocation-model=pic -code-model=small | FileCheck %s -check-prefix=DARWIN-64-PIC
 
 @src = external global [131072 x i32]
 @dst = external global [131072 x i32]

Modified: llvm/trunk/test/CodeGen/X86/add.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/add.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/add.ll (original)
+++ llvm/trunk/test/CodeGen/X86/add.ll Wed May  4 18:54:59 2011
@@ -1,6 +1,8 @@
 ; RUN: llc < %s -march=x86 | FileCheck %s -check-prefix=X32
-; RUN: llc < %s -mtriple=x86_64-linux | FileCheck %s -check-prefix=X64
-; RUN: llc < %s -mtriple=x86_64-win32 | FileCheck %s -check-prefix=X64
+; RUN: llc < %s -mtriple=x86_64-linux -join-physregs | FileCheck %s -check-prefix=X64
+; RUN: llc < %s -mtriple=x86_64-win32 -join-physregs | FileCheck %s -check-prefix=X64
+
+; Some of these tests depend on -join-physregs to commute instructions.
 
 ; The immediate can be encoded in a smaller way if the
 ; instruction is a sub instead of an add.

Modified: llvm/trunk/test/CodeGen/X86/andimm8.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/andimm8.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/andimm8.ll (original)
+++ llvm/trunk/test/CodeGen/X86/andimm8.ll Wed May  4 18:54:59 2011
@@ -1,4 +1,4 @@
-; RUN: llc < %s -march=x86-64 -mtriple=x86_64-pc-linux-gnu -show-mc-encoding | FileCheck %s
+; RUN: llc < %s -march=x86-64 -mtriple=x86_64-pc-linux-gnu -show-mc-encoding -join-physregs | FileCheck %s
 
 ; PR8365
 ; CHECK: andl	$-64, %edi              # encoding: [0x83,0xe7,0xc0]

Modified: llvm/trunk/test/CodeGen/X86/byval7.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/byval7.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/byval7.ll (original)
+++ llvm/trunk/test/CodeGen/X86/byval7.ll Wed May  4 18:54:59 2011
@@ -9,7 +9,6 @@
 ; CHECK: main:
 ; CHECK: movl $1, (%esp)
 ; CHECK: leal 16(%esp), %edi
-; CHECK: movl $36, %ecx
 ; CHECK: leal 160(%esp), %esi
 ; CHECK: rep;movsl
 	%s = alloca %struct.S		; <%struct.S*> [#uses=2]

Modified: llvm/trunk/test/CodeGen/X86/coalescer-commute2.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/coalescer-commute2.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/coalescer-commute2.ll (original)
+++ llvm/trunk/test/CodeGen/X86/coalescer-commute2.ll Wed May  4 18:54:59 2011
@@ -1,4 +1,4 @@
-; RUN: llc < %s -mtriple=x86_64-linux | FileCheck %s
+; RUN: llc < %s -mtriple=x86_64-linux -join-physregs | FileCheck %s
 ; CHECK-NOT:     mov
 ; CHECK:     paddw
 ; CHECK-NOT:     mov

Modified: llvm/trunk/test/CodeGen/X86/dbg-value-range.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/dbg-value-range.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/dbg-value-range.ll (original)
+++ llvm/trunk/test/CodeGen/X86/dbg-value-range.ll Wed May  4 18:54:59 2011
@@ -1,5 +1,5 @@
 ; RUN: llc -mtriple=x86_64-apple-darwin10 < %s | FileCheck %s
-; RUN: llc -mtriple=x86_64-apple-darwin10 -regalloc=basic < %s | FileCheck %s
+; RUN: llc -mtriple=x86_64-apple-darwin10 -regalloc=basic -join-physregs < %s | FileCheck %s
 
 %struct.a = type { i32 }
 

Modified: llvm/trunk/test/CodeGen/X86/optimize-max-3.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/optimize-max-3.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/optimize-max-3.ll (original)
+++ llvm/trunk/test/CodeGen/X86/optimize-max-3.ll Wed May  4 18:54:59 2011
@@ -41,14 +41,13 @@
 
 ;      CHECK:         jle
 ;  CHECK-NOT:         cmov
-;      CHECK:         xorl    {{%edi, %edi|%ecx, %ecx}}
+;      CHECK:         xorl    {{%edi, %edi|%ecx, %ecx|%eax, %eax}}
 ; CHECK-NEXT:         align
 ; CHECK-NEXT: BB1_2:
-; CHECK-NEXT:         callq
+; CHECK:              callq
 ; CHECK-NEXT:         incl    [[BX:%[a-z0-9]+]]
 ; CHECK-NEXT:         cmpl    [[R14:%[a-z0-9]+]], [[BX]]
-; CHECK-NEXT:         movq    %rax, %r{{di|cx}}
-; CHECK-NEXT:         jl
+; CHECK:              jl
 
 define void @_Z18GenerateStatusPagei(i32 %jobs_to_display) nounwind {
 entry:

Modified: llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-2.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-2.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-2.ll (original)
+++ llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-2.ll Wed May  4 18:54:59 2011
@@ -1,4 +1,4 @@
-; RUN: llc < %s -march=x86 | grep mov | count 4
+; RUN: llc < %s -march=x86 | FileCheck %s
 ; PR2659
 
 define i32 @binomial(i32 %n, i32 %k) nounwind {
@@ -12,7 +12,8 @@
 
 ifthen:		; preds = %entry
 	ret i32 0
-
+; CHECK: forbody
+; CHECK-NOT: mov
 forbody:		; preds = %forbody, %forcond.preheader
 	%indvar = phi i32 [ 0, %forcond.preheader ], [ %divisor.02, %forbody ]		; <i32> [#uses=3]
 	%accumulator.01 = phi i32 [ 1, %forcond.preheader ], [ %div, %forbody ]		; <i32> [#uses=1]

Modified: llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-3.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-3.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-3.ll (original)
+++ llvm/trunk/test/CodeGen/X86/phys_subreg_coalesce-3.ll Wed May  4 18:54:59 2011
@@ -1,6 +1,11 @@
-; RUN: llc < %s -mtriple=i386-apple-darwin | FileCheck %s
+; RUN: llc < %s -mtriple=i386-apple-darwin -join-physregs | FileCheck %s
 ; rdar://5571034
 
+; This requires physreg joining, %vreg13 is live everywhere:
+; 304L		%CL<def> = COPY %vreg13:sub_8bit; GR32_ABCD:%vreg13
+; 320L		%vreg15<def> = COPY %vreg19; GR32:%vreg15 GR32_NOSP:%vreg19
+; 336L		%vreg15<def> = SAR32rCL %vreg15, %EFLAGS<imp-def,dead>, %CL<imp-use,kill>; GR32:%vreg15
+
 define void @foo(i32* nocapture %quadrant, i32* nocapture %ptr, i32 %bbSize, i32 %bbStart, i32 %shifts) nounwind ssp {
 ; CHECK: foo:
 entry:

Modified: llvm/trunk/test/CodeGen/X86/pmul.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/pmul.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/pmul.ll (original)
+++ llvm/trunk/test/CodeGen/X86/pmul.ll Wed May  4 18:54:59 2011
@@ -1,7 +1,9 @@
-; RUN: llc < %s -march=x86 -mattr=sse41 -stack-alignment=16 > %t
+; RUN: llc < %s -march=x86 -mattr=sse41 -stack-alignment=16 -join-physregs > %t
 ; RUN: grep pmul %t | count 12
 ; RUN: grep mov %t | count 11
 
+; The f() arguments in %xmm0 and %xmm1 cause an extra movdqa without -join-physregs.
+
 define <4 x i32> @a(<4 x i32> %i) nounwind  {
         %A = mul <4 x i32> %i, < i32 117, i32 117, i32 117, i32 117 >
         ret <4 x i32> %A

Modified: llvm/trunk/test/CodeGen/X86/pr2659.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/pr2659.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/pr2659.ll (original)
+++ llvm/trunk/test/CodeGen/X86/pr2659.ll Wed May  4 18:54:59 2011
@@ -1,5 +1,4 @@
-; RUN: llc < %s -march=x86 -mtriple=i686-apple-darwin9.4.0 | grep movl | count 4
-; RUN: llc < %s -march=x86 -mtriple=i686-apple-darwin9.4.0 | FileCheck %s
+; RUN: llc < %s -march=x86 -mtriple=i686-apple-darwin9.4.0 -disable-branch-fold | FileCheck %s
 ; PR2659
 
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
@@ -19,7 +18,11 @@
 ; CHECK-NOT: xorl
 ; CHECK-NOT: movl
 ; CHECK-NOT: LBB
-; CHECK: je
+; CHECK: jne
+
+; There should be no moves required in the for loop body.
+; CHECK: %forbody
+; CHECK-NOT: mov
 
 ifthen:         ; preds = %entry
   ret i32 0

Modified: llvm/trunk/test/CodeGen/X86/sibcall.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/sibcall.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/sibcall.ll (original)
+++ llvm/trunk/test/CodeGen/X86/sibcall.ll Wed May  4 18:54:59 2011
@@ -229,7 +229,7 @@
 ; 64: t14:
 ; 64: movq 32(%rdi)
 ; 64-NOT: movq 16(%rdi)
-; 64: jmpq *16(%rdi)
+; 64: jmpq *16({{%rdi|%rax}})
   %0 = getelementptr inbounds %struct.__block_literal_2* %.block_descriptor, i64 0, i32 5 ; <void ()**> [#uses=1]
   %1 = load void ()** %0, align 8                 ; <void ()*> [#uses=2]
   %2 = bitcast void ()* %1 to %struct.__block_literal_1* ; <%struct.__block_literal_1*> [#uses=1]

Modified: llvm/trunk/test/CodeGen/X86/sse-minmax.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/sse-minmax.ll?rev=130893&r1=130892&r2=130893&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/sse-minmax.ll (original)
+++ llvm/trunk/test/CodeGen/X86/sse-minmax.ll Wed May  4 18:54:59 2011
@@ -1,6 +1,6 @@
-; RUN: llc < %s -march=x86-64 -asm-verbose=false | FileCheck %s
-; RUN: llc < %s -march=x86-64 -asm-verbose=false -enable-unsafe-fp-math -enable-no-nans-fp-math | FileCheck -check-prefix=UNSAFE %s
-; RUN: llc < %s -march=x86-64 -asm-verbose=false -enable-no-nans-fp-math | FileCheck -check-prefix=FINITE %s
+; RUN: llc < %s -march=x86-64 -asm-verbose=false -join-physregs | FileCheck %s
+; RUN: llc < %s -march=x86-64 -asm-verbose=false -join-physregs -enable-unsafe-fp-math -enable-no-nans-fp-math | FileCheck -check-prefix=UNSAFE %s
+; RUN: llc < %s -march=x86-64 -asm-verbose=false -join-physregs -enable-no-nans-fp-math | FileCheck -check-prefix=FINITE %s
 
 ; Some of these patterns can be matched as SSE min or max. Some of
 ; then can be matched provided that the operands are swapped.
@@ -12,6 +12,9 @@
 ; y_ : use -0.0 instead of %y
 ; _inverse : swap the arms of the select.
 
+; Some of these tests depend on -join-physregs commuting instructions to
+; eliminate copies.
+
 ; CHECK:      ogt:
 ; CHECK-NEXT: maxsd %xmm1, %xmm0
 ; CHECK-NEXT: ret



From dpatel at apple.com  Wed May  4 18:58:22 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 23:58:22 -0000
Subject: [llvm-commits] [llvm] r130894 -
	/llvm/trunk/lib/Transforms/Utils/LCSSA.cpp
Message-ID: <20110504235822.F2B842A6C12C@llvm.org>

Author: dpatel
Date: Wed May  4 18:58:22 2011
New Revision: 130894

URL: http://llvm.org/viewvc/llvm-project?rev=130894&view=rev
Log:
Set debug location for new PHI nodes created in exit block. 

Modified:
    llvm/trunk/lib/Transforms/Utils/LCSSA.cpp

Modified: llvm/trunk/lib/Transforms/Utils/LCSSA.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Utils/LCSSA.cpp?rev=130894&r1=130893&r2=130894&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Utils/LCSSA.cpp (original)
+++ llvm/trunk/lib/Transforms/Utils/LCSSA.cpp Wed May  4 18:58:22 2011
@@ -36,6 +36,7 @@
 #include "llvm/Analysis/Dominators.h"
 #include "llvm/Analysis/LoopPass.h"
 #include "llvm/Analysis/ScalarEvolution.h"
+#include "llvm/Transforms/Utils/BasicBlockUtils.h"
 #include "llvm/Transforms/Utils/SSAUpdater.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/STLExtras.h"
@@ -226,6 +227,7 @@
                                   PredCache.GetNumPreds(ExitBB),
                                   Inst->getName()+".lcssa",
                                   ExitBB->begin());
+    PN->setDebugLoc(GetFirstDebugLocInBasicBlock(ExitBB));
 
     // Add inputs from inside the loop for this PHI.
     for (BasicBlock **PI = PredCache.GetPreds(ExitBB); *PI; ++PI) {



From dpatel at apple.com  Wed May  4 18:58:50 2011
From: dpatel at apple.com (Devang Patel)
Date: Wed, 04 May 2011 23:58:50 -0000
Subject: [llvm-commits] [llvm] r130895 -
	/llvm/trunk/lib/Transforms/Scalar/GVN.cpp
Message-ID: <20110504235850.785C62A6C12C@llvm.org>

Author: dpatel
Date: Wed May  4 18:58:50 2011
New Revision: 130895

URL: http://llvm.org/viewvc/llvm-project?rev=130895&view=rev
Log:
Set debug loc for new instructions.

Modified:
    llvm/trunk/lib/Transforms/Scalar/GVN.cpp

Modified: llvm/trunk/lib/Transforms/Scalar/GVN.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/GVN.cpp?rev=130895&r1=130894&r2=130895&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/GVN.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/GVN.cpp Wed May  4 18:58:50 2011
@@ -952,12 +952,12 @@
       IntegerType::get(LoadTy->getContext(), NewLoadSize*8);
     DestPTy = PointerType::get(DestPTy, 
                        cast<PointerType>(PtrVal->getType())->getAddressSpace());
-    
+    Builder.SetCurrentDebugLocation(SrcVal->getDebugLoc());
     PtrVal = Builder.CreateBitCast(PtrVal, DestPTy);
     LoadInst *NewLoad = Builder.CreateLoad(PtrVal);
     NewLoad->takeName(SrcVal);
     NewLoad->setAlignment(SrcVal->getAlignment());
-    
+
     DEBUG(dbgs() << "GVN WIDENED LOAD: " << *SrcVal << "\n");
     DEBUG(dbgs() << "TO: " << *NewLoad << "\n");
     
@@ -2118,7 +2118,7 @@
 
       VN.add(Phi, ValNo);
       addToLeaderTable(ValNo, Phi, CurrentBlock);
-
+      Phi->setDebugLoc(CurInst->getDebugLoc());
       CurInst->replaceAllUsesWith(Phi);
       if (Phi->getType()->isPointerTy()) {
         // Because we have added a PHI-use of the pointer value, it has now



From stoklund at 2pi.dk  Wed May  4 18:59:00 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Wed, 04 May 2011 23:59:00 -0000
Subject: [llvm-commits] [llvm] r130896 -
	/llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp
Message-ID: <20110504235900.5441D2A6C12C@llvm.org>

Author: stoklund
Date: Wed May  4 18:59:00 2011
New Revision: 130896

URL: http://llvm.org/viewvc/llvm-project?rev=130896&view=rev
Log:
Disable physical register coalescing by default.

Joining physregs is inherently dangerous because it uses a heuristic to avoid
creating invalid code. Linear scan had an emergency spilling mechanism to deal
with those rare cases. The new greedy allocator does not.

The greedy register allocator is much better at taking hints, so this has almost
no impact on code size and quality. The few cases where it matters show up as
unit tests that now have -join-physregs enabled explicitly.

Modified:
    llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp

Modified: llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp?rev=130896&r1=130895&r2=130896&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp (original)
+++ llvm/trunk/lib/CodeGen/SimpleRegisterCoalescing.cpp Wed May  4 18:59:00 2011
@@ -63,7 +63,7 @@
 static cl::opt<bool>
 EnablePhysicalJoin("join-physregs",
                    cl::desc("Join physical register copies"),
-                   cl::init(true), cl::Hidden);
+                   cl::init(false), cl::Hidden);
 
 static cl::opt<bool>
 VerifyCoalescing("verify-coalescing",



From nicholas at mxc.ca  Wed May  4 19:03:30 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Thu, 05 May 2011 00:03:30 -0000
Subject: [llvm-commits] [llvm] r130897 -
	/llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
Message-ID: <20110505000330.4157E2A6C12C@llvm.org>

Author: nicholas
Date: Wed May  4 19:03:30 2011
New Revision: 130897

URL: http://llvm.org/viewvc/llvm-project?rev=130897&view=rev
Log:
When the path wasn't emitted by the frontend, discard any path on the source
filename.

Modified:
    llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp

Modified: llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp?rev=130897&r1=130896&r2=130897&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp (original)
+++ llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp Wed May  4 19:03:30 2011
@@ -331,15 +331,20 @@
     for (int i = 0, e = GCov->getNumOperands(); i != e; ++i) {
       MDNode *N = GCov->getOperand(i);
       if (N->getNumOperands() != 2) continue;
-      MDString *Path = dyn_cast<MDString>(N->getOperand(0));
+      MDString *GCovFile = dyn_cast<MDString>(N->getOperand(0));
       MDNode *CompileUnit = dyn_cast<MDNode>(N->getOperand(1));
-      if (!Path || !CompileUnit) continue;
-      if (CompileUnit == CU)
-        return (Path->getString() + "/" +
-                replaceStem(CU.getFilename(), NewStem)).str();
+      if (!GCovFile || !CompileUnit) continue;
+      if (CompileUnit == CU) {
+        SmallString<128> Filename = GCovFile->getString();
+        sys::path::replace_extension(Filename, NewStem);
+        return Filename.str();
+      }
     }
   }
-  return replaceStem(CU.getFilename(), NewStem);
+
+  SmallString<128> Filename = CU.getFilename();
+  sys::path::replace_extension(Filename, NewStem);
+  return sys::path::filename(Filename.str());
 }
 
 bool GCOVProfiler::runOnModule(Module &M) {



From aggarwa4 at illinois.edu  Wed May  4 19:09:39 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Thu, 05 May 2011 00:09:39 -0000
Subject: [llvm-commits] [poolalloc] r130900 - in
 /poolalloc/trunk/lib/AssistDS: GEPExprArg.cpp LoadArgs.cpp
Message-ID: <20110505000939.EB2782A6C12D@llvm.org>

Author: aggarwa4
Date: Wed May  4 19:09:39 2011
New Revision: 130900

URL: http://llvm.org/viewvc/llvm-project?rev=130900&view=rev
Log:
Lower the number of clones to control code
size.

Modified:
    poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp
    poolalloc/trunk/lib/AssistDS/LoadArgs.cpp

Modified: poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp?rev=130900&r1=130899&r2=130900&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp Wed May  4 19:09:39 2011
@@ -88,7 +88,7 @@
               const FunctionType *NewFTy = FunctionType::get(CI->getType(), TP, false);
               Function *NewF;
               numSimplified++;
-              if(numSimplified > 1000) 
+              if(numSimplified > 800) 
                 return true;
 
               NewF = Function::Create(NewFTy,

Modified: poolalloc/trunk/lib/AssistDS/LoadArgs.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/LoadArgs.cpp?rev=130900&r1=130899&r2=130900&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/LoadArgs.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/LoadArgs.cpp Wed May  4 19:09:39 2011
@@ -108,7 +108,7 @@
               const FunctionType *NewFTy = FunctionType::get(CI->getType(), TP, false);
               Function *NewF;
               numSimplified++;
-              if(numSimplified > 500) //26
+              if(numSimplified > 400) //26
                 return true;
 
               NewF = Function::Create(NewFTy,



From aggarwa4 at illinois.edu  Wed May  4 19:10:15 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Thu, 05 May 2011 00:10:15 -0000
Subject: [llvm-commits] [poolalloc] r130901 -
	/poolalloc/trunk/lib/AssistDS/LoadArgs.cpp
Message-ID: <20110505001015.161022A6C12D@llvm.org>

Author: aggarwa4
Date: Wed May  4 19:10:14 2011
New Revision: 130901

URL: http://llvm.org/viewvc/llvm-project?rev=130901&view=rev
Log:
Remove unnessary comment

Modified:
    poolalloc/trunk/lib/AssistDS/LoadArgs.cpp

Modified: poolalloc/trunk/lib/AssistDS/LoadArgs.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/LoadArgs.cpp?rev=130901&r1=130900&r2=130901&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/LoadArgs.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/LoadArgs.cpp Wed May  4 19:10:14 2011
@@ -108,7 +108,7 @@
               const FunctionType *NewFTy = FunctionType::get(CI->getType(), TP, false);
               Function *NewF;
               numSimplified++;
-              if(numSimplified > 400) //26
+              if(numSimplified > 400)
                 return true;
 
               NewF = Function::Create(NewFTy,



From nicholas at mxc.ca  Wed May  4 19:17:34 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Thu, 05 May 2011 00:17:34 -0000
Subject: [llvm-commits] [llvm] r130903 -
	/llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
Message-ID: <20110505001734.BB7A82A6C12C@llvm.org>

Author: nicholas
Date: Wed May  4 19:17:34 2011
New Revision: 130903

URL: http://llvm.org/viewvc/llvm-project?rev=130903&view=rev
Log:
Remove dead function.

Modified:
    llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp

Modified: llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp?rev=130903&r1=130902&r2=130903&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp (original)
+++ llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp Wed May  4 19:17:34 2011
@@ -321,11 +321,6 @@
   };
 }
 
-// Replace the stem of a file, or add one if missing.
-static std::string replaceStem(std::string OrigFilename, std::string NewStem) {
-  return (sys::path::stem(OrigFilename) + "." + NewStem).str();
-}
-
 std::string GCOVProfiler::mangleName(DICompileUnit CU, std::string NewStem) {
   if (NamedMDNode *GCov = M->getNamedMetadata("llvm.gcov")) {
     for (int i = 0, e = GCov->getNumOperands(); i != e; ++i) {



From eli.friedman at gmail.com  Wed May  4 19:35:04 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Wed, 4 May 2011 17:35:04 -0700
Subject: [llvm-commits] [llvm] r130877 - in /llvm/trunk:
 lib/CodeGen/MachineCSE.cpp test/CodeGen/Thumb2/thumb2-cbnz.ll
 test/CodeGen/X86/cmp-redundant.ll
In-Reply-To: <20110504221037.2B8C92A6C12C@llvm.org>
References: <20110504221037.2B8C92A6C12C@llvm.org>
Message-ID: <BANLkTi=LtwfhU0VyVCfMfHdbidQY6mZK+A@mail.gmail.com>

On Wed, May 4, 2011 at 3:10 PM, Eli Friedman <eli.friedman at gmail.com> wrote:
> Author: efriedma
> Date: Wed May ?4 17:10:36 2011
> New Revision: 130877
>
> URL: http://llvm.org/viewvc/llvm-project?rev=130877&view=rev
> Log:
> Re-commit r130862 with a minor change to avoid an iterator running off the edge in some cases.
>
> Original message:
>
> Teach MachineCSE how to do simple cross-block CSE involving physregs. ?This allows, for example, eliminating duplicate cmpl's on x86. Part of rdar://problem/8259436 .
>
>

This is breaking a few of the buildbots with the following assertion:

lib/CodeGen/LiveInterval.cpp:267: llvm::LiveRange*
llvm::LiveInterval::addRangeFrom(llvm::LiveRange, llvm::LiveRange*):
Assertion `B->end <= Start && "Cannot overlap two LiveRanges with
differing ValID's" " (did you def the same reg twice in a
MachineInstr?)"' failed.

I'm not sure how to go about reproducing this, so it would be very
much appreciated if someone could send me a bitcode file where this
reproduces.

Thanks,
Eli


From echristo at apple.com  Wed May  4 19:46:19 2011
From: echristo at apple.com (Eric Christopher)
Date: Thu, 05 May 2011 00:46:19 -0000
Subject: [llvm-commits] [llvm-gcc-4.2] r130905 -
	/llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h
Message-ID: <20110505004619.717212A6C12C@llvm.org>

Author: echristo
Date: Wed May  4 19:46:19 2011
New Revision: 130905

URL: http://llvm.org/viewvc/llvm-project?rev=130905&view=rev
Log:
Fix extra long comment.

Modified:
    llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h

Modified: llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h
URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h?rev=130905&r1=130904&r2=130905&view=diff
==============================================================================
--- llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h (original)
+++ llvm-gcc-4.2/trunk/gcc/config/arm/darwin.h Wed May  4 19:46:19 2011
@@ -48,9 +48,10 @@
 
 #define REGISTER_PREFIX 	""
 
-/* The assembler's names for the registers.  Note that the ?xx registers are * there so that VFPv3/NEON registers D16-D31 have the same spacing as D0-D15
- * (each of which is overlaid on two S registers), although there are no
- * actual single-precision registers which correspond to D16-D31.  */
+/* The assembler's names for the registers.  Note that the ?xx registers are 
+   there so that VFPv3/NEON registers D16-D31 have the same spacing as D0-D15
+   (each of which is overlaid on two S registers), although there are no
+   actual single-precision registers which correspond to D16-D31.  */
 #ifndef REGISTER_NAMES
 #define REGISTER_NAMES				   \
 {				                   \



From clattner at apple.com  Wed May  4 19:57:02 2011
From: clattner at apple.com (Chris Lattner)
Date: Wed, 04 May 2011 17:57:02 -0700
Subject: [llvm-commits] [llvm] r130894 -
 /llvm/trunk/lib/Transforms/Utils/LCSSA.cpp
In-Reply-To: <20110504235822.F2B842A6C12C@llvm.org>
References: <20110504235822.F2B842A6C12C@llvm.org>
Message-ID: <5B056F2B-45B0-44A5-B5E0-7F26BBBDC2A6@apple.com>


On May 4, 2011, at 4:58 PM, Devang Patel wrote:

> Author: dpatel
> Date: Wed May  4 18:58:22 2011
> New Revision: 130894
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=130894&view=rev
> Log:
> Set debug location for new PHI nodes created in exit block. 

Devang,

What does a debug loc even mean for a PHI node?  Conceptually a phi node is a copy in each of the predecessors, certainly it can't have a single debug loc?

-Chris

> 
> Modified:
>    llvm/trunk/lib/Transforms/Utils/LCSSA.cpp
> 
> Modified: llvm/trunk/lib/Transforms/Utils/LCSSA.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Utils/LCSSA.cpp?rev=130894&r1=130893&r2=130894&view=diff
> ==============================================================================
> --- llvm/trunk/lib/Transforms/Utils/LCSSA.cpp (original)
> +++ llvm/trunk/lib/Transforms/Utils/LCSSA.cpp Wed May  4 18:58:22 2011
> @@ -36,6 +36,7 @@
> #include "llvm/Analysis/Dominators.h"
> #include "llvm/Analysis/LoopPass.h"
> #include "llvm/Analysis/ScalarEvolution.h"
> +#include "llvm/Transforms/Utils/BasicBlockUtils.h"
> #include "llvm/Transforms/Utils/SSAUpdater.h"
> #include "llvm/ADT/Statistic.h"
> #include "llvm/ADT/STLExtras.h"
> @@ -226,6 +227,7 @@
>                                   PredCache.GetNumPreds(ExitBB),
>                                   Inst->getName()+".lcssa",
>                                   ExitBB->begin());
> +    PN->setDebugLoc(GetFirstDebugLocInBasicBlock(ExitBB));
> 
>     // Add inputs from inside the loop for this PHI.
>     for (BasicBlock **PI = PredCache.GetPreds(ExitBB); *PI; ++PI) {
> 
> 
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits


From clattner at apple.com  Wed May  4 19:58:08 2011
From: clattner at apple.com (Chris Lattner)
Date: Wed, 04 May 2011 17:58:08 -0700
Subject: [llvm-commits] [llvm] r130791 - in /llvm/trunk:
 lib/Target/X86/X86InstrSSE.td lib/VMCore/AutoUpgrade.cpp
 test/Assembler/AutoUpgradeIntrinsics.ll test/CodeGen/X86/nontemporal.ll
In-Reply-To: <20110503211117.54AA62A6C12C@llvm.org>
References: <20110503211117.54AA62A6C12C@llvm.org>
Message-ID: <B28786FC-B253-4809-8EF3-D38B900333EC@apple.com>


On May 3, 2011, at 2:11 PM, Bill Wendling wrote:

> Author: void
> Date: Tue May  3 16:11:17 2011
> New Revision: 130791
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=130791&view=rev
> Log:
> Replace the "movnt" intrinsics with a native store + nontemporal metadata bit.
> <rdar://problem/8460511>

Thanks Bill,

This is probably worth mentioning in the X86 section of CodeGenerator.html if it isn't already,

-Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110504/dab91797/attachment.html 

From aggarwa4 at illinois.edu  Wed May  4 20:47:38 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Thu, 05 May 2011 01:47:38 -0000
Subject: [llvm-commits] [poolalloc] r130908 - in /poolalloc/trunk:
 include/assistDS/TypeChecks.h lib/AssistDS/TypeChecks.cpp
Message-ID: <20110505014738.AB9EC2A6C12C@llvm.org>

Author: aggarwa4
Date: Wed May  4 20:47:38 2011
New Revision: 130908

URL: http://llvm.org/viewvc/llvm-project?rev=130908&view=rev
Log:
Also handle Invoke Insts.


Modified:
    poolalloc/trunk/include/assistDS/TypeChecks.h
    poolalloc/trunk/lib/AssistDS/TypeChecks.cpp

Modified: poolalloc/trunk/include/assistDS/TypeChecks.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/TypeChecks.h?rev=130908&r1=130907&r2=130908&view=diff
==============================================================================
--- poolalloc/trunk/include/assistDS/TypeChecks.h (original)
+++ poolalloc/trunk/include/assistDS/TypeChecks.h Wed May  4 20:47:38 2011
@@ -57,11 +57,12 @@
   bool initShadow(Module &M, Instruction &I);
   bool unmapShadow(Module &M, Instruction &I);
   bool visitCallInst(Module &M, CallInst &CI);
+  bool visitInvokeInst(Module &M, InvokeInst &CI);
   bool visitCallSite(Module &M, CallSite CS);
   bool visitLoadInst(Module &M, LoadInst &LI);
+  bool visitStoreInst(Module &M, StoreInst &SI);
   bool visitGlobal(Module &M, GlobalVariable &GV, 
                    Constant *C, Instruction &I, unsigned offset);
-  bool visitStoreInst(Module &M, StoreInst &SI);
   bool visitCopyingStoreInst(Module &M, StoreInst &SI, Value *SS);
   bool visitInputFunctionValue(Module &M, Value *V, Instruction *CI);
 

Modified: poolalloc/trunk/lib/AssistDS/TypeChecks.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/TypeChecks.cpp?rev=130908&r1=130907&r2=130908&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/TypeChecks.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/TypeChecks.cpp Wed May  4 20:47:38 2011
@@ -133,6 +133,9 @@
         }
       } else if (CallInst *CI = dyn_cast<CallInst>(&I)) {
         modified |= visitCallInst(M, *CI);
+      } else if (InvokeInst *II = dyn_cast<InvokeInst>(&I)) {
+        II->dump();
+        modified |= visitInvokeInst(M, *II);
       }
     }
   }
@@ -276,6 +279,11 @@
   return visitCallSite(M, &CI);
 }
 
+// Insert runtime checks for certain call instructions
+bool TypeChecks::visitInvokeInst(Module &M, InvokeInst &II) {
+  return visitCallSite(M, &II);
+}
+
 bool TypeChecks::visitCallSite(Module &M, CallSite CS) {
   //
   // Get the called value.  Strip off any casts which are lossless.



From nicholas at mxc.ca  Wed May  4 21:46:38 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Thu, 05 May 2011 02:46:38 -0000
Subject: [llvm-commits] [llvm] r130911 - in /llvm/trunk:
 lib/Transforms/Instrumentation/GCOVProfiling.cpp
 runtime/libprofile/GCDAProfiling.c
Message-ID: <20110505024638.CD7872A6C12C@llvm.org>

Author: nicholas
Date: Wed May  4 21:46:38 2011
New Revision: 130911

URL: http://llvm.org/viewvc/llvm-project?rev=130911&view=rev
Log:
Update the gcov version used slightly, to make it stop causing modern gcov's to
crash.

Modified:
    llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
    llvm/trunk/runtime/libprofile/GCDAProfiling.c

Modified: llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp?rev=130911&r1=130910&r2=130911&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp (original)
+++ llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp Wed May  4 21:46:38 2011
@@ -87,7 +87,7 @@
     // list.
     void insertCounterWriteout(DebugInfoFinder &,
                                SmallVector<std::pair<GlobalVariable *,
-                                                     uint32_t>, 8> &);
+                                                     MDNode *>, 8> &);
 
     std::string mangleName(DICompileUnit CU, std::string NewStem);
 
@@ -261,12 +261,13 @@
       ReturnBlock = new GCOVBlock(i++, os);
 
       writeBytes(FunctionTag, 4);
-      uint32_t BlockLen = 1 + 1 + 1 + lengthOfGCOVString(SP.getName()) +
+      uint32_t BlockLen = 1 + 1 + 1 + 1 + lengthOfGCOVString(SP.getName()) +
           1 + lengthOfGCOVString(SP.getFilename()) + 1;
       write(BlockLen);
       uint32_t Ident = reinterpret_cast<intptr_t>((MDNode*)SP);
       write(Ident);
-      write(0);  // checksum
+      write(0);  // checksum #1
+      write(0);  // checksum #2
       writeGCOVString(SP.getName());
       writeGCOVString(SP.getFilename());
       write(SP.getLineNumber());
@@ -418,7 +419,7 @@
   if (DIF.subprogram_begin() == DIF.subprogram_end())
     return false;
 
-  SmallVector<std::pair<GlobalVariable *, uint32_t>, 8> CountersByIdent;
+  SmallVector<std::pair<GlobalVariable *, MDNode *>, 8> CountersBySP;
   for (DebugInfoFinder::iterator SPI = DIF.subprogram_begin(),
            SPE = DIF.subprogram_end(); SPI != SPE; ++SPI) {
     DISubprogram SP(*SPI);
@@ -441,8 +442,7 @@
                            GlobalValue::InternalLinkage,
                            Constant::getNullValue(CounterTy),
                            "__llvm_gcov_ctr", 0, false, 0);
-    CountersByIdent.push_back(
-        std::make_pair(Counters, reinterpret_cast<intptr_t>((MDNode*)SP)));
+    CountersBySP.push_back(std::make_pair(Counters, (MDNode*)SP));
 
     UniqueVector<BasicBlock *> ComplexEdgePreds;
     UniqueVector<BasicBlock *> ComplexEdgeSuccs;
@@ -509,7 +509,7 @@
     }
   }
 
-  insertCounterWriteout(DIF, CountersByIdent);
+  insertCounterWriteout(DIF, CountersBySP);
 
   return true;
 }
@@ -580,7 +580,10 @@
 }
 
 Constant *GCOVProfiler::getEmitFunctionFunc() {
-  const Type *Args[] = { Type::getInt32Ty(*Ctx) };
+  const Type *Args[2] = {
+    Type::getInt32Ty(*Ctx),    // uint32_t ident
+    Type::getInt8PtrTy(*Ctx),  // const char *function_name
+  };
   const FunctionType *FTy = FunctionType::get(Type::getVoidTy(*Ctx),
                                               Args, false);
   return M->getOrInsertFunction("llvm_gcda_emit_function", FTy);
@@ -616,7 +619,7 @@
 
 void GCOVProfiler::insertCounterWriteout(
     DebugInfoFinder &DIF,
-    SmallVector<std::pair<GlobalVariable *, uint32_t>, 8> &CountersByIdent) {
+    SmallVector<std::pair<GlobalVariable *, MDNode *>, 8> &CountersBySP) {
   const FunctionType *WriteoutFTy =
       FunctionType::get(Type::getVoidTy(*Ctx), false);
   Function *WriteoutF = Function::Create(WriteoutFTy,
@@ -637,11 +640,15 @@
     std::string FilenameGcda = mangleName(compile_unit, "gcda");
     Builder.CreateCall(StartFile,
                        Builder.CreateGlobalStringPtr(FilenameGcda));
-    for (SmallVector<std::pair<GlobalVariable *, uint32_t>, 8>::iterator
-             I = CountersByIdent.begin(), E = CountersByIdent.end();
+    for (SmallVector<std::pair<GlobalVariable *, MDNode *>, 8>::iterator
+             I = CountersBySP.begin(), E = CountersBySP.end();
          I != E; ++I) {
-      Builder.CreateCall(EmitFunction, ConstantInt::get(Type::getInt32Ty(*Ctx),
-                                                        I->second));
+      DISubprogram SP(I->second);
+      intptr_t ident = reinterpret_cast<intptr_t>(I->second);
+      Builder.CreateCall2(EmitFunction,
+                          ConstantInt::get(Type::getInt32Ty(*Ctx), ident),
+                          Builder.CreateGlobalStringPtr(SP.getName()));
+                                                        
       GlobalVariable *GV = I->first;
       unsigned Arcs =
           cast<ArrayType>(GV->getType()->getElementType())->getNumElements();

Modified: llvm/trunk/runtime/libprofile/GCDAProfiling.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/runtime/libprofile/GCDAProfiling.c?rev=130911&r1=130910&r2=130911&view=diff
==============================================================================
--- llvm/trunk/runtime/libprofile/GCDAProfiling.c (original)
+++ llvm/trunk/runtime/libprofile/GCDAProfiling.c Wed May  4 21:46:38 2011
@@ -48,6 +48,17 @@
   write_int32(hi);
 }
 
+static uint32_t length_of_string(const char *s) {
+  return (strlen(s) + 5) / 4;
+}
+
+static void write_string(const char *s) {
+  uint32_t len = length_of_string(s);
+  write_int32(len);
+  fwrite(s, strlen(s), 1, output_file);
+  fwrite("\0\0\0\0", 4 - (strlen(s) % 4), 1, output_file);
+}
+
 static char *mangle_filename(const char *orig_filename) {
   /* TODO: handle GCOV_PREFIX_STRIP */
   const char *prefix;
@@ -129,16 +140,18 @@
 #endif
 }
 
-void llvm_gcda_emit_function(uint32_t ident) {
+void llvm_gcda_emit_function(uint32_t ident, const char *function_name) {
 #ifdef DEBUG_GCDAPROFILING
   printf("llvmgcda: function id=%x\n", ident);
 #endif
 
   /* function tag */  
   fwrite("\0\0\0\1", 4, 1, output_file);
-  write_int32(2);
+  write_int32(3 + 1 + length_of_string(function_name));
   write_int32(ident);
   write_int32(0);
+  write_int32(0);
+  write_string(function_name);
 }
 
 void llvm_gcda_emit_arcs(uint32_t num_counters, uint64_t *counters) {



From foldr at codedgers.com  Wed May  4 23:24:58 2011
From: foldr at codedgers.com (Mikhail Glushenkov)
Date: Thu, 05 May 2011 04:24:58 -0000
Subject: [llvm-commits] [llvm] r130914 -
	/llvm/trunk/utils/TableGen/LLVMCConfigurationEmitter.cpp
Message-ID: <20110505042458.7E5A92A6C12C@llvm.org>

Author: foldr
Date: Wed May  4 23:24:58 2011
New Revision: 130914

URL: http://llvm.org/viewvc/llvm-project?rev=130914&view=rev
Log:
llvmc: Make it possible to provide an argument to (join).

Modified:
    llvm/trunk/utils/TableGen/LLVMCConfigurationEmitter.cpp

Modified: llvm/trunk/utils/TableGen/LLVMCConfigurationEmitter.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/utils/TableGen/LLVMCConfigurationEmitter.cpp?rev=130914&r1=130913&r2=130914&view=diff
==============================================================================
--- llvm/trunk/utils/TableGen/LLVMCConfigurationEmitter.cpp (original)
+++ llvm/trunk/utils/TableGen/LLVMCConfigurationEmitter.cpp Wed May  4 23:24:58 2011
@@ -74,6 +74,25 @@
   return val.getValue();
 }
 
+bool InitPtrToBool(const Init* ptr) {
+  bool ret = false;
+  const DefInit& val = dynamic_cast<const DefInit&>(*ptr);
+  const std::string& str = val.getAsString();
+
+  if (str == "true") {
+    ret = true;
+  }
+  else if (str == "false") {
+    ret = false;
+  }
+  else {
+    throw "Incorrect boolean value: '" + str +
+      "': must be either 'true' or 'false'";
+  }
+
+  return ret;
+}
+
 const std::string& InitPtrToString(const Init* ptr) {
   const StringInit& val = dynamic_cast<const StringInit&>(*ptr);
   return val.getValue();
@@ -95,13 +114,7 @@
 
 /// CheckBooleanConstant - Check that the provided value is a boolean constant.
 void CheckBooleanConstant(const Init* I) {
-  const DefInit& val = dynamic_cast<const DefInit&>(*I);
-  const std::string& str = val.getAsString();
-
-  if (str != "true" && str != "false") {
-    throw "Incorrect boolean value: '" + str +
-      "': must be either 'true' or 'false'";
-  }
+  InitPtrToBool(I);
 }
 
 // CheckNumberOfArguments - Ensure that the number of args in d is
@@ -935,8 +948,22 @@
   }
 
   void onJoin (const DagInit& d) {
-    CheckNumberOfArguments(d, 0);
-    toolDesc_.setJoin();
+    bool isReallyJoin = false;
+
+    if (d.getNumArgs() == 0) {
+      isReallyJoin = true;
+    }
+    else {
+      Init* I = d.getArg(0);
+      isReallyJoin = InitPtrToBool(I);
+    }
+
+    // Is this *really* a join tool? We allow (join false) for generating two
+    // tool descriptions from a single generic one.
+    // TOFIX: come up with a cleaner solution.
+    if (isReallyJoin) {
+      toolDesc_.setJoin();
+    }
   }
 
   void onOutLanguage (const DagInit& d) {



From foldr at codedgers.com  Wed May  4 23:25:03 2011
From: foldr at codedgers.com (Mikhail Glushenkov)
Date: Thu, 05 May 2011 04:25:03 -0000
Subject: [llvm-commits] [llvm] r130915 -
	/llvm/trunk/include/llvm/CompilerDriver/Common.td
Message-ID: <20110505042503.AB7812A6C12D@llvm.org>

Author: foldr
Date: Wed May  4 23:25:03 2011
New Revision: 130915

URL: http://llvm.org/viewvc/llvm-project?rev=130915&view=rev
Log:
llvmc: Make 'true' and 'false' instances of a 'Bool' class.

Modified:
    llvm/trunk/include/llvm/CompilerDriver/Common.td

Modified: llvm/trunk/include/llvm/CompilerDriver/Common.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CompilerDriver/Common.td?rev=130915&r1=130914&r2=130915&view=diff
==============================================================================
--- llvm/trunk/include/llvm/CompilerDriver/Common.td (original)
+++ llvm/trunk/include/llvm/CompilerDriver/Common.td Wed May  4 23:25:03 2011
@@ -56,8 +56,11 @@
 def case;
 
 // Boolean constants.
-def true;
-def false;
+class Bool<bit val> {
+      bit Value = val;
+}
+def true : Bool<1>;
+def false : Bool<0>;
 
 // Boolean operators.
 def and;



From aggarwa4 at illinois.edu  Thu May  5 00:11:12 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Thu, 05 May 2011 05:11:12 -0000
Subject: [llvm-commits] [poolalloc] r130917 - in /poolalloc/trunk:
 lib/AssistDS/TypeChecks.cpp runtime/DynamicTypeChecks/TypeRuntime.c
Message-ID: <20110505051113.0684F2A6C12C@llvm.org>

Author: aggarwa4
Date: Thu May  5 00:11:12 2011
New Revision: 130917

URL: http://llvm.org/viewvc/llvm-project?rev=130917&view=rev
Log:
Added some preliminary tracking of byval 
arguments. Assumes they are all TOP. Also 
makes sure we clear the metadata when we return.

Modified:
    poolalloc/trunk/lib/AssistDS/TypeChecks.cpp
    poolalloc/trunk/runtime/DynamicTypeChecks/TypeRuntime.c

Modified: poolalloc/trunk/lib/AssistDS/TypeChecks.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/TypeChecks.cpp?rev=130917&r1=130916&r2=130917&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/TypeChecks.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/TypeChecks.cpp Thu May  5 00:11:12 2011
@@ -21,6 +21,9 @@
 #include "llvm/Support/raw_ostream.h"
 #include "llvm/Intrinsics.h"
 
+#include <set>
+#include <vector>
+
 using namespace llvm;
 
 char TypeChecks::ID = 0;
@@ -109,7 +112,8 @@
     IncorporateType(MI->getType());
     Function &F = *MI;
 
-    // Loop over all of the instructions in the function, adding their return type as well as the types of their operands.
+    // Loop over all of the instructions in the function, 
+    // adding their return type as well as the types of their operands.
     for (inst_iterator II = inst_begin(F), IE = inst_end(F); II != IE; ++II) {
       Instruction &I = *II;
 
@@ -140,6 +144,71 @@
     }
   }
 
+  // Record types for byval arguments.
+
+  // Split fn into 2 clones. One internal, such that
+  // we can find all call sites to it, and also pass in the
+  // original ptr/metadata.
+  // One that sets it just to initialized memory
+  // so that it can be calle from external code.
+  for (Module::iterator MI = M.begin(), ME = M.end(); MI != ME; ++MI) {
+    Function &F = *MI;
+    typedef SmallVector<Value *, 4> RegisteredArgTy;
+    RegisteredArgTy registeredArguments;
+    for (Function::arg_iterator I = F.arg_begin(), E = F.arg_end(); I != E; ++I) {
+      if (I->hasByValAttr()) {
+        assert (isa<PointerType>(I->getType()));
+        const PointerType * PT = cast<PointerType>(I->getType());
+        const Type * ET = PT->getElementType();
+        Value * AllocSize = ConstantInt::get(Int64Ty, TD->getTypeAllocSize(ET));
+        Instruction * InsertBefore = &(F.getEntryBlock().front());
+        CastInst *BCI = BitCastInst::CreatePointerCast(I, VoidPtrTy, "", InsertBefore);
+        std::vector<Value *> Args;
+        Args.push_back(BCI);
+        Args.push_back(AllocSize);
+        Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
+        Constant *F = M.getOrInsertFunction("trackInitInst", VoidTy, VoidPtrTy, Int64Ty, Int32Ty, NULL);
+        CallInst::Create(F, Args.begin(), Args.end(), "", InsertBefore);
+        registeredArguments.push_back(&*I);
+      }
+    }
+    //
+    // Find all basic blocks which terminate the function.
+    //
+    std::set<BasicBlock *> exitBlocks;
+    for (inst_iterator I = inst_begin(&F), E = inst_end(&F); I != E; ++I) {
+      if (isa<ReturnInst>(*I) || isa<UnwindInst>(*I)) {
+        exitBlocks.insert(I->getParent());
+      }
+    }
+
+    //
+    // At each function exit, insert code to deregister all byval arguments.
+    //
+    for (std::set<BasicBlock*>::const_iterator BI = exitBlocks.begin(),
+         BE = exitBlocks.end();
+         BI != BE; ++BI) {
+      for (RegisteredArgTy::const_iterator I = registeredArguments.begin(),
+           E = registeredArguments.end();
+           I != E; ++I) {
+        SmallVector<Value *, 2> args;
+        Instruction * Pt = &((*BI)->back());
+        const PointerType * PT = cast<PointerType>((*I)->getType());
+        const Type * ET = PT->getElementType();
+        Value * AllocSize = ConstantInt::get(Int64Ty, TD->getTypeAllocSize(ET));
+        CastInst *BCI = BitCastInst::CreatePointerCast(*I, VoidPtrTy, "", Pt);
+        std::vector<Value *> Args;
+        Args.push_back(BCI);
+        Args.push_back(AllocSize);
+        Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
+        Constant *F = M.getOrInsertFunction("trackUnInitInst", VoidTy, VoidPtrTy, Int64Ty, Int32Ty, NULL);
+        CallInst::Create(F, Args.begin(), Args.end(), "", Pt);
+      }
+    }
+
+
+  }
+
   return modified;
 }
 
@@ -184,7 +253,7 @@
 
 bool TypeChecks::visitGlobal(Module &M, GlobalVariable &GV, 
                              Constant *C, Instruction &I, unsigned offset) {
-  
+
   if(ConstantArray *CA = dyn_cast<ConstantArray>(C)) {
     const Type * ElementType = CA->getType()->getElementType();
     unsigned int t = TD->getTypeStoreSize(ElementType);

Modified: poolalloc/trunk/runtime/DynamicTypeChecks/TypeRuntime.c
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/runtime/DynamicTypeChecks/TypeRuntime.c?rev=130917&r1=130916&r2=130917&view=diff
==============================================================================
--- poolalloc/trunk/runtime/DynamicTypeChecks/TypeRuntime.c (original)
+++ poolalloc/trunk/runtime/DynamicTypeChecks/TypeRuntime.c Thu May  5 00:11:12 2011
@@ -123,6 +123,11 @@
   p &= 0xFFFFFFFF;
   memset(&shadow_begin[p], 0xFF, size);
 }
+void trackUnInitInst(void *ptr, uint64_t size, uint32_t tag) {
+  uintptr_t p = (uintptr_t)ptr;
+  p &= 0xFFFFFFFF;
+  memset(&shadow_begin[p], 0x00, size);
+}
 
 /**
  * Copy size bits of metadata from src ptr to dest ptr.



From isanbard at gmail.com  Thu May  5 01:49:15 2011
From: isanbard at gmail.com (Bill Wendling)
Date: Thu, 05 May 2011 06:49:15 -0000
Subject: [llvm-commits] [llvm] r130918 - in /llvm/trunk:
 include/llvm/Target/TargetLoweringObjectFile.h
 lib/CodeGen/AsmPrinter/DwarfTableException.cpp
 lib/CodeGen/TargetLoweringObjectFileImpl.cpp
 lib/Target/TargetLoweringObjectFile.cpp
Message-ID: <20110505064915.62BE32A6C12C@llvm.org>

Author: void
Date: Thu May  5 01:49:15 2011
New Revision: 130918

URL: http://llvm.org/viewvc/llvm-project?rev=130918&view=rev
Log:
Remove a flag that would set the ".eh" symbol as .globl. MachO was the only one
who used this flag, and it now emits CFI and doesn't emit this anymore. All
other targets left this flag "false".
<rdar://problem/8486371>

Modified:
    llvm/trunk/include/llvm/Target/TargetLoweringObjectFile.h
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp
    llvm/trunk/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
    llvm/trunk/lib/Target/TargetLoweringObjectFile.cpp

Modified: llvm/trunk/include/llvm/Target/TargetLoweringObjectFile.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Target/TargetLoweringObjectFile.h?rev=130918&r1=130917&r2=130918&view=diff
==============================================================================
--- llvm/trunk/include/llvm/Target/TargetLoweringObjectFile.h (original)
+++ llvm/trunk/include/llvm/Target/TargetLoweringObjectFile.h Thu May  5 01:49:15 2011
@@ -97,10 +97,6 @@
   /// weak_definition of constant 0 for an omitted EH frame.
   bool SupportsWeakOmittedEHFrame;
   
-  /// IsFunctionEHSymbolGlobal - This flag is set to true if the ".eh" symbol
-  /// for a function should be marked .globl.
-  bool IsFunctionEHSymbolGlobal;
-  
   /// IsFunctionEHFrameSymbolPrivate - This flag is set to true if the
   /// "EH_frame" symbol for EH information should be an assembler temporary (aka
   /// private linkage, aka an L or .L label) or false if it should be a normal
@@ -119,9 +115,6 @@
     Ctx = &ctx;
   }
   
-  bool isFunctionEHSymbolGlobal() const {
-    return IsFunctionEHSymbolGlobal;
-  }
   bool isFunctionEHFrameSymbolPrivate() const {
     return IsFunctionEHFrameSymbolPrivate;
   }

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp?rev=130918&r1=130917&r2=130918&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp Thu May  5 01:49:15 2011
@@ -172,11 +172,6 @@
 
   Asm->OutStreamer.SwitchSection(TLOF.getEHFrameSection());
 
-  // Externally visible entry into the functions eh frame info. If the
-  // corresponding function is static, this should not be externally visible.
-  if (!TheFunc->hasLocalLinkage() && TLOF.isFunctionEHSymbolGlobal())
-    Asm->OutStreamer.EmitSymbolAttribute(EHFrameInfo.FunctionEHSym,MCSA_Global);
-
   // If corresponding function is weak definition, this should be too.
   if (TheFunc->isWeakForLinker() && Asm->MAI->getWeakDefDirective())
     Asm->OutStreamer.EmitSymbolAttribute(EHFrameInfo.FunctionEHSym,

Modified: llvm/trunk/lib/CodeGen/TargetLoweringObjectFileImpl.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/TargetLoweringObjectFileImpl.cpp?rev=130918&r1=130917&r2=130918&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/TargetLoweringObjectFileImpl.cpp (original)
+++ llvm/trunk/lib/CodeGen/TargetLoweringObjectFileImpl.cpp Thu May  5 01:49:15 2011
@@ -484,11 +484,6 @@
 
 void TargetLoweringObjectFileMachO::Initialize(MCContext &Ctx,
                                                const TargetMachine &TM) {
-  // _foo.eh symbols are currently always exported so that the linker knows
-  // about them.  This is not necessary on 10.6 and later, but it
-  // doesn't hurt anything.
-  // FIXME: I need to get this from Triple.
-  IsFunctionEHSymbolGlobal = true;
   IsFunctionEHFrameSymbolPrivate = false;
   SupportsWeakOmittedEHFrame = false;
 

Modified: llvm/trunk/lib/Target/TargetLoweringObjectFile.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/TargetLoweringObjectFile.cpp?rev=130918&r1=130917&r2=130918&view=diff
==============================================================================
--- llvm/trunk/lib/Target/TargetLoweringObjectFile.cpp (original)
+++ llvm/trunk/lib/Target/TargetLoweringObjectFile.cpp Thu May  5 01:49:15 2011
@@ -58,7 +58,6 @@
   DwarfRangesSection = 0;
   DwarfMacroInfoSection = 0;
   
-  IsFunctionEHSymbolGlobal = false;
   IsFunctionEHFrameSymbolPrivate = true;
   SupportsWeakOmittedEHFrame = true;
 }



From etherzhhb at gmail.com  Thu May  5 03:03:42 2011
From: etherzhhb at gmail.com (ether zhhb)
Date: Thu, 5 May 2011 16:03:42 +0800
Subject: [llvm-commits] [PATCH 1/2] Minor change: Fix the typo in
	RegionPass.h and RegionPass.cpp.
Message-ID: <BANLkTimLPKZJaoDAT5vnt1-zX-dG+-saxw@mail.gmail.com>

hi,

This patch fix two typo in RegionPass.h and RegionPass.cpp.

best regards
ether

---
 include/llvm/Analysis/RegionPass.h |    2 +-
 lib/Analysis/RegionPass.cpp        |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/include/llvm/Analysis/RegionPass.h
b/include/llvm/Analysis/RegionPass.h
index aedc06a..6c4e226 100644
--- a/include/llvm/Analysis/RegionPass.h
+++ b/include/llvm/Analysis/RegionPass.h
@@ -109,7 +109,7 @@ public:
   /// @brief Print passes managed by this manager.
   void dumpPassStructure(unsigned Offset);

-  /// @brief Print passes contained by this manager.
+  /// @brief Get passes contained by this manager.
   Pass *getContainedPass(unsigned N) {
     assert(N < PassVector.size() && "Pass number out of range!");
     Pass *FP = static_cast<Pass *>(PassVector[N]);
diff --git a/lib/Analysis/RegionPass.cpp b/lib/Analysis/RegionPass.cpp
index 3269dcc..80eda79 100644
--- a/lib/Analysis/RegionPass.cpp
+++ b/lib/Analysis/RegionPass.cpp
@@ -249,7 +249,7 @@ void RegionPass::assignPassManager(PMStack &PMS,
     assert (!PMS.empty() && "Unable to create Region Pass Manager");
     PMDataManager *PMD = PMS.top();

-    // [1] Create new Call Graph Pass Manager
+    // [1] Create new Region Pass Manager
     RGPM = new RGPassManager(PMD->getDepth() + 1);
     RGPM->populateInheritedAnalysis(PMS);

-- 
1.7.4
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Minor-change-Fix-the-typo-in-RegionPass.h-and-Region.patch
Type: application/octet-stream
Size: 1514 bytes
Desc: not available
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110505/2bc8a341/attachment.obj 

From etherzhhb at gmail.com  Thu May  5 03:07:41 2011
From: etherzhhb at gmail.com (ether zhhb)
Date: Thu, 5 May 2011 16:07:41 +0800
Subject: [llvm-commits] [PATCH 2/2] Allow RegionPass to do post processing
 after all passes have run.
Message-ID: <BANLkTi=ZHgtbkft4nL6fBsBQ-FLOvYwLpA@mail.gmail.com>

hi,

This patch add a module level finalization function to regionpass.

best regards
ether

---
 include/llvm/Analysis/RegionPass.h |   16 ++++++++++++++--
 lib/Analysis/RegionPass.cpp        |    9 +++++++++
 2 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/include/llvm/Analysis/RegionPass.h
b/include/llvm/Analysis/RegionPass.h
index 6c4e226..82f4f76 100644
--- a/include/llvm/Analysis/RegionPass.h
+++ b/include/llvm/Analysis/RegionPass.h
@@ -61,6 +61,12 @@ public:

   virtual bool doInitialization(Region *R, RGPassManager &RGM) {
return false; }
   virtual bool doFinalization() { return false; }
+
+  /// @brief Virtual method overridden by subclasses to do any post processing
+  /// needed after all passes have run.
+  ///
+  /// @return True if the pass modifies the module.
+  virtual bool doFinalization(Module &M) { return false; }
   //@}

   //===--------------------------------------------------------------------===//
@@ -95,6 +101,12 @@ public:
   /// @return True if any of the passes modifies the function.
   bool runOnFunction(Function &F);

+  /// @brief Execute all of the passes to do any post processing needed after
+  ///        all passes have run.
+  ///
+  /// @return True if any of the passes modifies the module.
+  bool doFinalization(Module &M);
+
   /// Pass Manager itself does not invalidate any analysis info.
   /// RGPassManager needs RegionInfo.
   void getAnalysisUsage(AnalysisUsage &Info) const;
@@ -110,9 +122,9 @@ public:
   void dumpPassStructure(unsigned Offset);

   /// @brief Get passes contained by this manager.
-  Pass *getContainedPass(unsigned N) {
+  RegionPass *getContainedPass(unsigned N) {
     assert(N < PassVector.size() && "Pass number out of range!");
-    Pass *FP = static_cast<Pass *>(PassVector[N]);
+    RegionPass *FP = static_cast<RegionPass *>(PassVector[N]);
     return FP;
   }

diff --git a/lib/Analysis/RegionPass.cpp b/lib/Analysis/RegionPass.cpp
index 80eda79..9b1c3ab 100644
--- a/lib/Analysis/RegionPass.cpp
+++ b/lib/Analysis/RegionPass.cpp
@@ -165,6 +165,15 @@ bool RGPassManager::runOnFunction(Function &F) {
   return Changed;
 }

+bool RGPassManager::doFinalization(Module &M) {
+  bool Changed = false;
+
+  for (unsigned Index = 0; Index < getNumContainedPasses(); ++Index)
+    Changed |= getContainedPass(Index)->doFinalization(M);
+
+  return Changed;
+}
+
 /// Print passes managed by this manager
 void RGPassManager::dumpPassStructure(unsigned Offset) {
   errs().indent(Offset*2) << "Region Pass Manager\n";
-- 
1.7.4
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0002-Allow-RegionPass-to-do-post-processing-after-all-pas.patch
Type: application/octet-stream
Size: 2691 bytes
Desc: not available
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110505/254921d5/attachment.obj 

From nadav.rotem at intel.com  Thu May  5 06:32:27 2011
From: nadav.rotem at intel.com (Rotem, Nadav)
Date: Thu, 5 May 2011 14:32:27 +0300
Subject: [llvm-commits] [PATCH] A patch to add vector SHL/SRL/SRA support
	for x86
Message-ID: <6594DDFF12B03D4E89690887C2486994027D601E8B@hasmsx504.ger.corp.intel.com>

This patch addresses 9597.
---------------------------------------------------------------------
Intel Israel (74) Limited

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110505/1695e3c2/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sra_llvm_trunk.diff
Type: application/octet-stream
Size: 11610 bytes
Desc: sra_llvm_trunk.diff
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110505/1695e3c2/attachment.obj 

From tobias at grosser.es  Thu May  5 06:55:40 2011
From: tobias at grosser.es (Tobias Grosser)
Date: Thu, 05 May 2011 13:55:40 +0200
Subject: [llvm-commits] [PATCH 1/2] Minor change: Fix the typo in
	RegionPass.h and RegionPass.cpp.
In-Reply-To: <BANLkTimLPKZJaoDAT5vnt1-zX-dG+-saxw@mail.gmail.com>
References: <BANLkTimLPKZJaoDAT5vnt1-zX-dG+-saxw@mail.gmail.com>
Message-ID: <4DC2903C.6080104@grosser.es>

On 05/05/2011 10:03 AM, ether zhhb wrote:
> hi,
>
> This patch fix two typo in RegionPass.h and RegionPass.cpp.

Nice. Please commit.

Tobi

From tobias at grosser.es  Thu May  5 07:04:54 2011
From: tobias at grosser.es (Tobias Grosser)
Date: Thu, 05 May 2011 14:04:54 +0200
Subject: [llvm-commits] [PATCH 2/2] Allow RegionPass to do post
 processing after all passes have run.
In-Reply-To: <BANLkTi=ZHgtbkft4nL6fBsBQ-FLOvYwLpA@mail.gmail.com>
References: <BANLkTi=ZHgtbkft4nL6fBsBQ-FLOvYwLpA@mail.gmail.com>
Message-ID: <4DC29266.3090105@grosser.es>

On 05/05/2011 10:07 AM, ether zhhb wrote:
> hi,
>
> This patch add a module level finalization function to regionpass.

Hi ether,

this looks good.

For what are you planning to use this? Are you planning to clean up the 
OpenMP code generation?

Cheers
Tobi

From etherzhhb at gmail.com  Thu May  5 08:07:26 2011
From: etherzhhb at gmail.com (ether zhhb)
Date: Thu, 5 May 2011 21:07:26 +0800
Subject: [llvm-commits] [PATCH 1/2] Minor change: Fix the typo in
	RegionPass.h and RegionPass.cpp.
In-Reply-To: <4DC2903C.6080104@grosser.es>
References: <BANLkTimLPKZJaoDAT5vnt1-zX-dG+-saxw@mail.gmail.com>
	<4DC2903C.6080104@grosser.es>
Message-ID: <BANLkTinHncqZ=pUCxJjVw-m98-LszKLaGg@mail.gmail.com>

hi tobi,

sorry i do not have the commit access, please commit it for me :)

best regards
ether

On Thu, May 5, 2011 at 7:55 PM, Tobias Grosser <tobias at grosser.es> wrote:
> On 05/05/2011 10:03 AM, ether zhhb wrote:
>>
>> hi,
>>
>> This patch fix two typo in RegionPass.h and RegionPass.cpp.
>
> Nice. Please commit.
>
> Tobi
>

From etherzhhb at gmail.com  Thu May  5 08:16:32 2011
From: etherzhhb at gmail.com (ether zhhb)
Date: Thu, 5 May 2011 21:16:32 +0800
Subject: [llvm-commits] [PATCH 2/2] Allow RegionPass to do post
 processing after all passes have run.
In-Reply-To: <4DC29266.3090105@grosser.es>
References: <BANLkTi=ZHgtbkft4nL6fBsBQ-FLOvYwLpA@mail.gmail.com>
	<4DC29266.3090105@grosser.es>
Message-ID: <BANLkTi=vBji461pQoWTvrJaXLRMVX4aJaw@mail.gmail.com>

hi tobi,

this provide one possible way to clean up the OpenMP codegen, but not
a very straight forward way, i figure out another "just work" approach
to decouple  OpenMP code generation and ScopDetection slightly.  I
will list them in polly-dev list later.

best regards
ether

On Thu, May 5, 2011 at 8:04 PM, Tobias Grosser <tobias at grosser.es> wrote:
> On 05/05/2011 10:07 AM, ether zhhb wrote:
>>
>> hi,
>>
>> This patch add a module level finalization function to regionpass.
>
> Hi ether,
>
> this looks good.
>
> For what are you planning to use this? Are you planning to clean up the
> OpenMP code generation?
>
> Cheers
> Tobi
>

From tobias at grosser.es  Thu May  5 08:23:26 2011
From: tobias at grosser.es (Tobias Grosser)
Date: Thu, 05 May 2011 15:23:26 +0200
Subject: [llvm-commits] [PATCH 1/2] Minor change: Fix the typo in
	RegionPass.h and RegionPass.cpp.
In-Reply-To: <BANLkTinHncqZ=pUCxJjVw-m98-LszKLaGg@mail.gmail.com>
References: <BANLkTimLPKZJaoDAT5vnt1-zX-dG+-saxw@mail.gmail.com>	<4DC2903C.6080104@grosser.es>
	<BANLkTinHncqZ=pUCxJjVw-m98-LszKLaGg@mail.gmail.com>
Message-ID: <4DC2A4CE.9050707@grosser.es>

On 05/05/2011 03:07 PM, ether zhhb wrote:
> hi tobi,
>
> sorry i do not have the commit access, please commit it for me :)

Did you try? I believe your svn account for polly may already work for llvm.

Cheers
Tobi

From nadav.rotem at intel.com  Thu May  5 08:47:30 2011
From: nadav.rotem at intel.com (Rotem, Nadav)
Date: Thu, 5 May 2011 16:47:30 +0300
Subject: [llvm-commits] [PATCH] Initial work on vector-select and modified
	type legalization
Message-ID: <6594DDFF12B03D4E89690887C2486994027D601FBA@hasmsx504.ger.corp.intel.com>

Hi, 

I started working on implementing vector select in LLVM.  Codegen support for vector-select is very important for vectorizing compilers.  I followed Duncan's design proposal and implemented vector-integer promotion in the type-legalizer.  This required changes all over the codegen.  The work is not complete and there are several bugs and missing features, but I am now able to compile the basic vector-select code below.  Almost all of the LIT tests pass and a good number of vectorized programs run successfully. 

This is a big patch and I would like other LLVM developers to go over it and review it.

Thanks,
Nadav

define <4 x float> @vector_code(<4 x i64> %A, <4 x i64> %B, <4 x float> %R0, <4 x float> %R1 )  {
   %C = icmp eq <4 x i64> %A, %B
   %K = xor <4 x i1> <i1 1, i1 1, i1 1, i1 1>, %C
   %D = select <4 x i1> %K, <4 x float> %R1, <4 x float> %R0   
   ret <4 x float> %D
} 


pcmpeqq     XMM1, XMM3
pcmpeqq     XMM0, XMM2
shufps          XMM0, XMM1, -120
xorps            XMM0, XMMWORD PTR [LCPI0_0]
pslld             XMM0, 31
movdqa       XMM1, XMMWORD PTR [ESP + 4]
movdqa       XMM2, XMMWORD PTR [ESP + 20]
blendvps     XMM2, XMM1, %xmm0
movdqa       XMM0, XMM2
ret 

---------------------------------------------------------------------
Intel Israel (74) Limited

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: legalize.diff
Type: application/octet-stream
Size: 92207 bytes
Desc: legalize.diff
Url : http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110505/b4df8e94/attachment-0001.obj 

From etherzhhb at gmail.com  Thu May  5 08:59:38 2011
From: etherzhhb at gmail.com (Hongbin Zheng)
Date: Thu, 05 May 2011 13:59:38 -0000
Subject: [llvm-commits] [llvm] r130920 - in /llvm/trunk:
 include/llvm/Analysis/RegionPass.h lib/Analysis/RegionPass.cpp
Message-ID: <20110505135938.99D742A6C12C@llvm.org>

Author: ether
Date: Thu May  5 08:59:38 2011
New Revision: 130920

URL: http://llvm.org/viewvc/llvm-project?rev=130920&view=rev
Log:
Minor change: Fix the typo in RegionPass.h and RegionPass.cpp.

Modified:
    llvm/trunk/include/llvm/Analysis/RegionPass.h
    llvm/trunk/lib/Analysis/RegionPass.cpp

Modified: llvm/trunk/include/llvm/Analysis/RegionPass.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Analysis/RegionPass.h?rev=130920&r1=130919&r2=130920&view=diff
==============================================================================
--- llvm/trunk/include/llvm/Analysis/RegionPass.h (original)
+++ llvm/trunk/include/llvm/Analysis/RegionPass.h Thu May  5 08:59:38 2011
@@ -109,7 +109,7 @@
   /// @brief Print passes managed by this manager.
   void dumpPassStructure(unsigned Offset);
 
-  /// @brief Print passes contained by this manager.
+  /// @brief Get passes contained by this manager.
   Pass *getContainedPass(unsigned N) {
     assert(N < PassVector.size() && "Pass number out of range!");
     Pass *FP = static_cast<Pass *>(PassVector[N]);

Modified: llvm/trunk/lib/Analysis/RegionPass.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/RegionPass.cpp?rev=130920&r1=130919&r2=130920&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/RegionPass.cpp (original)
+++ llvm/trunk/lib/Analysis/RegionPass.cpp Thu May  5 08:59:38 2011
@@ -249,7 +249,7 @@
     assert (!PMS.empty() && "Unable to create Region Pass Manager");
     PMDataManager *PMD = PMS.top();
 
-    // [1] Create new Call Graph Pass Manager
+    // [1] Create new Region Pass Manager
     RGPM = new RGPassManager(PMD->getDepth() + 1);
     RGPM->populateInheritedAnalysis(PMS);
 



From eli.friedman at gmail.com  Thu May  5 10:24:08 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Thu, 5 May 2011 08:24:08 -0700
Subject: [llvm-commits] [PATCH] A patch to add vector SHL/SRL/SRA
 support for x86
In-Reply-To: <6594DDFF12B03D4E89690887C2486994027D601E8B@hasmsx504.ger.corp.intel.com>
References: <6594DDFF12B03D4E89690887C2486994027D601E8B@hasmsx504.ger.corp.intel.com>
Message-ID: <BANLkTin69Nf1RjtszX4-FKHGQcPkVM=ARA@mail.gmail.com>

On Thu, May 5, 2011 at 4:32 AM, Rotem, Nadav <nadav.rotem at intel.com> wrote:
> This patch addresses 9597.

Your testcase passes without your patch.  See PerformShiftCombine in
X86ISelLowering.cpp.  I woiuldn't be opposed to moving that code to
legalization if that works, though.  (I think it will because of the
separation of vector legalization.)

-Eli

From aggarwa4 at illinois.edu  Thu May  5 10:59:11 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Thu, 05 May 2011 15:59:11 -0000
Subject: [llvm-commits] [poolalloc] r130923 -
	/poolalloc/trunk/lib/AssistDS/TypeChecks.cpp
Message-ID: <20110505155911.6831E2A6C12E@llvm.org>

Author: aggarwa4
Date: Thu May  5 10:59:11 2011
New Revision: 130923

URL: http://llvm.org/viewvc/llvm-project?rev=130923&view=rev
Log:
Formatting changes. No functionality changed.

Modified:
    poolalloc/trunk/lib/AssistDS/TypeChecks.cpp

Modified: poolalloc/trunk/lib/AssistDS/TypeChecks.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/TypeChecks.cpp?rev=130923&r1=130922&r2=130923&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/TypeChecks.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/TypeChecks.cpp Thu May  5 10:59:11 2011
@@ -150,7 +150,7 @@
   // we can find all call sites to it, and also pass in the
   // original ptr/metadata.
   // One that sets it just to initialized memory
-  // so that it can be calle from external code.
+  // so that it can be called from external code.
   for (Module::iterator MI = M.begin(), ME = M.end(); MI != ME; ++MI) {
     Function &F = *MI;
     typedef SmallVector<Value *, 4> RegisteredArgTy;
@@ -205,8 +205,6 @@
         CallInst::Create(F, Args.begin(), Args.end(), "", Pt);
       }
     }
-
-
   }
 
   return modified;



From eli.friedman at gmail.com  Thu May  5 11:18:11 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Thu, 05 May 2011 16:18:11 -0000
Subject: [llvm-commits] [llvm] r130925 -
	/llvm/trunk/lib/CodeGen/MachineCSE.cpp
Message-ID: <20110505161812.9E67F2A6C12C@llvm.org>

Author: efriedma
Date: Thu May  5 11:18:11 2011
New Revision: 130925

URL: http://llvm.org/viewvc/llvm-project?rev=130925&view=rev
Log:
Minor correction to r130877; fixes PR9846 and hopefully the buildbot failures.


Modified:
    llvm/trunk/lib/CodeGen/MachineCSE.cpp

Modified: llvm/trunk/lib/CodeGen/MachineCSE.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineCSE.cpp?rev=130925&r1=130924&r2=130925&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/MachineCSE.cpp (original)
+++ llvm/trunk/lib/CodeGen/MachineCSE.cpp Thu May  5 11:18:11 2011
@@ -468,7 +468,8 @@
         SmallVector<unsigned,8>::iterator PI = DirectPhysRefs.begin(),
                                           PE = DirectPhysRefs.end();
         for (; PI != PE; ++PI)
-          MBB->addLiveIn(*PI);
+          if (!MBB->isLiveIn(*PI))
+            MBB->addLiveIn(*PI);
       }
       ++NumCSEs;
       if (!PhysRefs.empty())



From eli.friedman at gmail.com  Thu May  5 11:25:23 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Thu, 05 May 2011 16:25:23 -0000
Subject: [llvm-commits] [llvm] r130926 -
	/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
Message-ID: <20110505162523.2F84D2A6C12C@llvm.org>

Author: efriedma
Date: Thu May  5 11:25:23 2011
New Revision: 130926

URL: http://llvm.org/viewvc/llvm-project?rev=130926&view=rev
Log:
Small syntax cleanup; we don't need to #define constants in C++.  No functionality change intended.


Modified:
    llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp?rev=130926&r1=130925&r2=130926&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp Thu May  5 11:25:23 2011
@@ -225,9 +225,10 @@
     "vfork",
     "getcontext"
   };
-#define NUM_RETURNS_TWICE_FNS sizeof(ReturnsTwiceFns) / sizeof(const char *)
+  static const size_t NumReturnsTwiceFns = sizeof(ReturnsTwiceFns) /
+                                           sizeof(const char *);
 
-  for (unsigned I = 0; I < NUM_RETURNS_TWICE_FNS; ++I)
+  for (unsigned I = 0; I < NumReturnsTwiceFns; ++I)
     if (const Function *Callee = M->getFunction(ReturnsTwiceFns[I])) {
       if (!Callee->use_empty())
         for (Value::const_use_iterator
@@ -239,7 +240,6 @@
     }
 
   return false;
-#undef NUM_RETURNS_TWICE_FNS
 }
 
 /// SplitCriticalSideEffectEdges - Look for critical edges with a PHI value that



From dpatel at apple.com  Thu May  5 11:31:16 2011
From: dpatel at apple.com (Devang Patel)
Date: Thu, 05 May 2011 09:31:16 -0700
Subject: [llvm-commits] [llvm] r130894 -
 /llvm/trunk/lib/Transforms/Utils/LCSSA.cpp
In-Reply-To: <5B056F2B-45B0-44A5-B5E0-7F26BBBDC2A6@apple.com>
References: <20110504235822.F2B842A6C12C@llvm.org>
	<5B056F2B-45B0-44A5-B5E0-7F26BBBDC2A6@apple.com>
Message-ID: <27357309-4B90-4742-BDB7-01A6F51BD608@apple.com>


On May 4, 2011, at 5:57 PM, Chris Lattner wrote:

> 
> On May 4, 2011, at 4:58 PM, Devang Patel wrote:
> 
>> Author: dpatel
>> Date: Wed May  4 18:58:22 2011
>> New Revision: 130894
>> 
>> URL: http://llvm.org/viewvc/llvm-project?rev=130894&view=rev
>> Log:
>> Set debug location for new PHI nodes created in exit block. 
> 
> Devang,
> 
> What does a debug loc even mean for a PHI node?  Conceptually a phi node is a copy in each of the predecessors, certainly it can't have a single debug loc?
> 

True, but the copy needs a location. Do you have any thoughts on better alternative?
-
Devang


From stoklund at 2pi.dk  Thu May  5 11:48:01 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Thu, 05 May 2011 16:48:01 -0000
Subject: [llvm-commits] [llvm] r130927 -
	/llvm/trunk/test/FrontendC/2010-05-18-asmsched.c
Message-ID: <20110505164801.201992A6C12C@llvm.org>

Author: stoklund
Date: Thu May  5 11:48:00 2011
New Revision: 130927

URL: http://llvm.org/viewvc/llvm-project?rev=130927&view=rev
Log:
Fix test to be less sensitive to coalescing.

This should unbreak llvm-gcc-i386-linux-selfhost.

Modified:
    llvm/trunk/test/FrontendC/2010-05-18-asmsched.c

Modified: llvm/trunk/test/FrontendC/2010-05-18-asmsched.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC/2010-05-18-asmsched.c?rev=130927&r1=130926&r2=130927&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC/2010-05-18-asmsched.c (original)
+++ llvm/trunk/test/FrontendC/2010-05-18-asmsched.c Thu May  5 11:48:00 2011
@@ -3,8 +3,9 @@
 
 void foo(int x, int y) {
 // CHECK: bar
-// CHECK: movq  %r9, %r10
-// CHECK: movq  %rdi, %r9
+// CHECK-NOT: {{, %r9$}}
+// CHECK: movq  %r9,
+// CHECK: movq  {{.*}}, %r9
 // CHECK: bar
   register int lr9 asm("r9") = x;
   register int lr10 asm("r10") = y;



From stoklund at 2pi.dk  Thu May  5 11:55:39 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Thu, 05 May 2011 09:55:39 -0700
Subject: [llvm-commits] [llvm] r130926 -
 /llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
In-Reply-To: <20110505162523.2F84D2A6C12C@llvm.org>
References: <20110505162523.2F84D2A6C12C@llvm.org>
Message-ID: <900F9FD1-82E7-4618-83EC-02186BBE658C@2pi.dk>


On May 5, 2011, at 9:25 AM, Eli Friedman wrote:

> Author: efriedma
> Date: Thu May  5 11:25:23 2011
> New Revision: 130926
> 
> URL: http://llvm.org/viewvc/llvm-project?rev=130926&view=rev
> Log:
> Small syntax cleanup; we don't need to #define constants in C++.  No functionality change intended.
> 
> 
> Modified:
>    llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
> 
> Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp?rev=130926&r1=130925&r2=130926&view=diff
> ==============================================================================
> --- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp (original)
> +++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp Thu May  5 11:25:23 2011
> @@ -225,9 +225,10 @@
>     "vfork",
>     "getcontext"
>   };
> -#define NUM_RETURNS_TWICE_FNS sizeof(ReturnsTwiceFns) / sizeof(const char *)
> +  static const size_t NumReturnsTwiceFns = sizeof(ReturnsTwiceFns) /
> +                                           sizeof(const char *);

Note that we have array_lengthof() in STLExtras.h.



From eli.friedman at gmail.com  Thu May  5 11:53:34 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Thu, 05 May 2011 16:53:34 -0000
Subject: [llvm-commits] [llvm] r130928 - in /llvm/trunk:
 lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp
 lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
 test/CodeGen/X86/2010-06-25-CoalescerSubRegDefDead.ll
 test/CodeGen/X86/cmp-redundant.ll
Message-ID: <20110505165334.D32B72A6C12C@llvm.org>

Author: efriedma
Date: Thu May  5 11:53:34 2011
New Revision: 130928

URL: http://llvm.org/viewvc/llvm-project?rev=130928&view=rev
Log:
Avoid extra vreg copies for arguments passed in registers.  Specifically, this can make MachineCSE more effective in some cases (especially in small functions).  PR8361 / part of rdar://problem/8259436 .


Modified:
    llvm/trunk/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp
    llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
    llvm/trunk/test/CodeGen/X86/2010-06-25-CoalescerSubRegDefDead.ll
    llvm/trunk/test/CodeGen/X86/cmp-redundant.ll

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp?rev=130928&r1=130927&r2=130928&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp Thu May  5 11:53:34 2011
@@ -54,25 +54,6 @@
   return false;
 }
 
-/// isOnlyUsedInEntryBlock - If the specified argument is only used in the
-/// entry block, return true.  This includes arguments used by switches, since
-/// the switch may expand into multiple basic blocks.
-static bool isOnlyUsedInEntryBlock(const Argument *A, bool EnableFastISel) {
-  // With FastISel active, we may be splitting blocks, so force creation
-  // of virtual registers for all non-dead arguments.
-  if (EnableFastISel)
-    return A->use_empty();
-
-  const BasicBlock *Entry = A->getParent()->begin();
-  for (Value::const_use_iterator UI = A->use_begin(), E = A->use_end();
-       UI != E; ++UI) {
-    const User *U = *UI;
-    if (cast<Instruction>(U)->getParent() != Entry || isa<SwitchInst>(U))
-      return false;  // Use not in entry block.
-  }
-  return true;
-}
-
 FunctionLoweringInfo::FunctionLoweringInfo(const TargetLowering &tli)
   : TLI(tli) {
 }
@@ -89,13 +70,6 @@
   CanLowerReturn = TLI.CanLowerReturn(Fn->getCallingConv(), Fn->isVarArg(),
                                       Outs, Fn->getContext());
 
-  // Create a vreg for each argument register that is not dead and is used
-  // outside of the entry block for the function.
-  for (Function::const_arg_iterator AI = Fn->arg_begin(), E = Fn->arg_end();
-       AI != E; ++AI)
-    if (!isOnlyUsedInEntryBlock(AI, EnableFastISel))
-      InitializeRegForValue(AI);
-
   // Initialize the mapping of values to registers.  This is only set up for
   // instruction values that are used outside of the block that defines
   // them.

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp?rev=130928&r1=130927&r2=130928&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/SelectionDAGBuilder.cpp Thu May  5 11:53:34 2011
@@ -6232,6 +6232,25 @@
 
 #include "llvm/CodeGen/SelectionDAGISel.h"
 
+/// isOnlyUsedInEntryBlock - If the specified argument is only used in the
+/// entry block, return true.  This includes arguments used by switches, since
+/// the switch may expand into multiple basic blocks.
+static bool isOnlyUsedInEntryBlock(const Argument *A) {
+  // With FastISel active, we may be splitting blocks, so force creation
+  // of virtual registers for all non-dead arguments.
+  if (EnableFastISel)
+    return A->use_empty();
+
+  const BasicBlock *Entry = A->getParent()->begin();
+  for (Value::const_use_iterator UI = A->use_begin(), E = A->use_end();
+       UI != E; ++UI) {
+    const User *U = *UI;
+    if (cast<Instruction>(U)->getParent() != Entry || isa<SwitchInst>(U))
+      return false;  // Use not in entry block.
+  }
+  return true;
+}
+
 void SelectionDAGISel::LowerArguments(const BasicBlock *LLVMBB) {
   // If this is the entry block, emit arguments.
   const Function &F = *LLVMBB->getParent();
@@ -6375,8 +6394,8 @@
     if (I->use_empty() && NumValues)
       SDB->setUnusedArgValue(I, InVals[i]);
 
-    for (unsigned Value = 0; Value != NumValues; ++Value) {
-      EVT VT = ValueVTs[Value];
+    for (unsigned Val = 0; Val != NumValues; ++Val) {
+      EVT VT = ValueVTs[Val];
       EVT PartVT = TLI.getRegisterType(*CurDAG->getContext(), VT);
       unsigned NumParts = TLI.getNumRegisters(*CurDAG->getContext(), VT);
 
@@ -6395,21 +6414,34 @@
       i += NumParts;
     }
 
+    // We don't need to do anything else for unused arguments.
+    if (ArgValues.empty())
+      continue;
+
     // Note down frame index for byval arguments.
-    if (I->hasByValAttr() && !ArgValues.empty())
+    if (I->hasByValAttr())
       if (FrameIndexSDNode *FI =
           dyn_cast<FrameIndexSDNode>(ArgValues[0].getNode()))
         FuncInfo->setByValArgumentFrameIndex(I, FI->getIndex());
 
-    if (!I->use_empty()) {
-      SDValue Res;
-      if (!ArgValues.empty())
-        Res = DAG.getMergeValues(&ArgValues[0], NumValues,
-                                 SDB->getCurDebugLoc());
-      SDB->setValue(I, Res);
-
-      // If this argument is live outside of the entry block, insert a copy from
-      // wherever we got it to the vreg that other BB's will reference it as.
+    SDValue Res = DAG.getMergeValues(&ArgValues[0], NumValues,
+                                     SDB->getCurDebugLoc());
+    SDB->setValue(I, Res);
+
+    // If this argument is live outside of the entry block, insert a copy from
+    // wherever we got it to the vreg that other BB's will reference it as.
+    if (Res.getOpcode() == ISD::CopyFromReg) {
+      // If we can, though, try to skip creating an unnecessary vreg.
+      // FIXME: This isn't very clean... it would be nice to make this more
+      // general.
+      unsigned Reg = cast<RegisterSDNode>(Res.getOperand(1))->getReg();
+      if (TargetRegisterInfo::isVirtualRegister(Reg)) {
+        FuncInfo->ValueMap[I] = Reg;
+        continue;
+      }
+    }
+    if (!isOnlyUsedInEntryBlock(I)) {
+      FuncInfo->InitializeRegForValue(I);
       SDB->CopyToExportRegsIfNeeded(I);
     }
   }

Modified: llvm/trunk/test/CodeGen/X86/2010-06-25-CoalescerSubRegDefDead.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/2010-06-25-CoalescerSubRegDefDead.ll?rev=130928&r1=130927&r2=130928&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/2010-06-25-CoalescerSubRegDefDead.ll (original)
+++ llvm/trunk/test/CodeGen/X86/2010-06-25-CoalescerSubRegDefDead.ll Thu May  5 11:53:34 2011
@@ -22,7 +22,7 @@
 ; it is.
 ;
 ; CHECK: # %bb
-; CHECK: addq $64036, %rdi
+; CHECK: leaq	64036(%rdx), %rdi
 ; CHECK: rep;stosl
 
   %tmp5 = bitcast i32* %tmp4 to i8*

Modified: llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/cmp-redundant.ll?rev=130928&r1=130927&r2=130928&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/cmp-redundant.ll (original)
+++ llvm/trunk/test/CodeGen/X86/cmp-redundant.ll Thu May  5 11:53:34 2011
@@ -9,6 +9,27 @@
 ; CHECK: cmp:
 ; CHECK: cmpl
 ; CHECK: jg
+
+if.end:                                           ; preds = %entry
+; CHECK-NOT: cmpl
+; CHECK: cmov
+  %cmp4 = icmp slt i32 %a, %b
+  %. = select i1 %cmp4, i32 2, i32 111
+  br label %return
+
+return:                                           ; preds = %if.end, %entry
+  %retval.0 = phi i32 [ 1, %entry ], [ %., %if.end ]
+  ret i32 %retval.0
+}
+
+define i32 @cmp2(i32 %a, i32 %b) nounwind readnone ssp {
+entry:
+  %cmp = icmp sgt i32 %a, %b
+  br i1 %cmp, label %return, label %if.end
+; CHECK: cmp2:
+; CHECK: cmpl
+; CHECK: jg
+
 if.end:                                           ; preds = %entry
 ; CHECK-NOT: cmpl
 ; CHECK: cmov



From stoklund at 2pi.dk  Thu May  5 12:22:53 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Thu, 05 May 2011 17:22:53 -0000
Subject: [llvm-commits] [llvm] r130931 - in /llvm/trunk/lib/CodeGen:
 InlineSpiller.cpp LiveRangeEdit.cpp SplitKit.cpp
Message-ID: <20110505172254.088FD2A6C12C@llvm.org>

Author: stoklund
Date: Thu May  5 12:22:53 2011
New Revision: 130931

URL: http://llvm.org/viewvc/llvm-project?rev=130931&view=rev
Log:
Add some statistics to the splitting and spilling frameworks.

Modified:
    llvm/trunk/lib/CodeGen/InlineSpiller.cpp
    llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp
    llvm/trunk/lib/CodeGen/SplitKit.cpp

Modified: llvm/trunk/lib/CodeGen/InlineSpiller.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/InlineSpiller.cpp?rev=130931&r1=130930&r2=130931&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/InlineSpiller.cpp (original)
+++ llvm/trunk/lib/CodeGen/InlineSpiller.cpp Thu May  5 12:22:53 2011
@@ -16,6 +16,7 @@
 #include "Spiller.h"
 #include "LiveRangeEdit.h"
 #include "VirtRegMap.h"
+#include "llvm/ADT/Statistic.h"
 #include "llvm/Analysis/AliasAnalysis.h"
 #include "llvm/CodeGen/LiveIntervalAnalysis.h"
 #include "llvm/CodeGen/LiveStackAnalysis.h"
@@ -31,6 +32,18 @@
 
 using namespace llvm;
 
+STATISTIC(NumSpilledRanges,   "Number of spilled live ranges");
+STATISTIC(NumSnippets,        "Number of snippets included in spills");
+STATISTIC(NumSpills,          "Number of spills inserted");
+STATISTIC(NumReloads,         "Number of reloads inserted");
+STATISTIC(NumFolded,          "Number of folded stack accesses");
+STATISTIC(NumFoldedLoads,     "Number of folded loads");
+STATISTIC(NumRemats,          "Number of rematerialized defs for spilling");
+STATISTIC(NumOmitReloadSpill, "Number of omitted spills after reloads");
+STATISTIC(NumHoistLocal,      "Number of locally hoisted spills");
+STATISTIC(NumHoistGlobal,     "Number of globally hoisted spills");
+STATISTIC(NumRedundantSpills, "Number of redundant spills identified");
+
 namespace {
 class InlineSpiller : public Spiller {
   MachineFunctionPass &Pass;
@@ -247,10 +260,11 @@
     if (!isSnippet(SnipLI))
       continue;
     SnippetCopies.insert(MI);
-    if (!isRegToSpill(SnipReg))
-      RegsToSpill.push_back(SnipReg);
-
+    if (isRegToSpill(SnipReg))
+      continue;
+    RegsToSpill.push_back(SnipReg);
     DEBUG(dbgs() << "\talso spill snippet " << SnipLI << '\n');
+    ++NumSnippets;
   }
 }
 
@@ -469,9 +483,10 @@
                << *StackInt << '\n');
 
   // Already spilled everywhere.
-  if (SVI.AllDefsAreReloads)
+  if (SVI.AllDefsAreReloads) {
+    ++NumOmitReloadSpill;
     return true;
-
+  }
   // We are going to spill SVI.SpillVNI immediately after its def, so clear out
   // any later spills of the same value.
   eliminateRedundantSpills(SibLI, SVI.SpillVNI);
@@ -493,6 +508,11 @@
   LIS.InsertMachineInstrInMaps(MII);
   VRM.addSpillSlotUse(StackSlot, MII);
   DEBUG(dbgs() << "\thoisted: " << SVI.SpillVNI->def << '\t' << *MII);
+
+  if (MBB == CopyMI->getParent())
+    ++NumHoistLocal;
+  else
+    ++NumHoistGlobal;
   return true;
 }
 
@@ -547,6 +567,7 @@
         // eliminateDeadDefs won't normally remove stores, so switch opcode.
         MI->setDesc(TII.get(TargetOpcode::KILL));
         DeadDefs.push_back(MI);
+        ++NumRedundantSpills;
       }
     }
   } while (!WorkList.empty());
@@ -642,6 +663,7 @@
   if (RM.OrigMI->getDesc().canFoldAsLoad() &&
       foldMemoryOperand(MI, Ops, RM.OrigMI)) {
     Edit->markRematerialized(RM.ParentVNI);
+    ++NumFoldedLoads;
     return true;
   }
 
@@ -668,6 +690,7 @@
   VNInfo *DefVNI = NewLI.getNextValue(DefIdx, 0, LIS.getVNInfoAllocator());
   NewLI.addRange(LiveRange(DefIdx, UseIdx.getDefIndex(), DefVNI));
   DEBUG(dbgs() << "\tinterval: " << NewLI << '\n');
+  ++NumRemats;
   return true;
 }
 
@@ -794,6 +817,7 @@
     VRM.addSpillSlotUse(StackSlot, FoldMI);
   MI->eraseFromParent();
   DEBUG(dbgs() << "\tfolded: " << *FoldMI);
+  ++NumFolded;
   return true;
 }
 
@@ -811,6 +835,7 @@
   VNInfo *LoadVNI = NewLI.getNextValue(LoadIdx, 0,
                                        LIS.getVNInfoAllocator());
   NewLI.addRange(LiveRange(LoadIdx, Idx, LoadVNI));
+  ++NumReloads;
 }
 
 /// insertSpill - Insert a spill of NewLI.reg after MI.
@@ -825,6 +850,7 @@
   DEBUG(dbgs() << "\tspilled: " << StoreIdx << '\t' << *MI);
   VNInfo *StoreVNI = NewLI.getNextValue(Idx, 0, LIS.getVNInfoAllocator());
   NewLI.addRange(LiveRange(Idx, StoreIdx, StoreVNI));
+  ++NumSpills;
 }
 
 /// spillAroundUses - insert spill code around each use of Reg.
@@ -972,6 +998,7 @@
 }
 
 void InlineSpiller::spill(LiveRangeEdit &edit) {
+  ++NumSpilledRanges;
   Edit = &edit;
   assert(!TargetRegisterInfo::isStackSlot(edit.getReg())
          && "Trying to spill a stack slot.");

Modified: llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp?rev=130931&r1=130930&r2=130931&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp (original)
+++ llvm/trunk/lib/CodeGen/LiveRangeEdit.cpp Thu May  5 12:22:53 2011
@@ -15,6 +15,7 @@
 #include "LiveRangeEdit.h"
 #include "VirtRegMap.h"
 #include "llvm/ADT/SetVector.h"
+#include "llvm/ADT/Statistic.h"
 #include "llvm/CodeGen/CalcSpillWeights.h"
 #include "llvm/CodeGen/LiveIntervalAnalysis.h"
 #include "llvm/CodeGen/MachineRegisterInfo.h"
@@ -24,6 +25,10 @@
 
 using namespace llvm;
 
+STATISTIC(NumDCEDeleted,     "Number of instructions deleted by DCE");
+STATISTIC(NumDCEFoldedLoads, "Number of single use loads folded after DCE");
+STATISTIC(NumFracRanges,     "Number of live ranges fractured by DCE");
+
 LiveInterval &LiveRangeEdit::createFrom(unsigned OldReg,
                                         LiveIntervals &LIS,
                                         VirtRegMap &VRM) {
@@ -199,6 +204,7 @@
   UseMI->eraseFromParent();
   DefMI->addRegisterDead(LI->reg, 0);
   Dead.push_back(DefMI);
+  ++NumDCEFoldedLoads;
   return true;
 }
 
@@ -269,6 +275,7 @@
         delegate_->LRE_WillEraseInstruction(MI);
       LIS.RemoveMachineInstrFromMaps(MI);
       MI->eraseFromParent();
+      ++NumDCEDeleted;
     }
 
     if (ToShrink.empty())
@@ -290,6 +297,7 @@
     unsigned NumComp = ConEQ.Classify(LI);
     if (NumComp <= 1)
       continue;
+    ++NumFracRanges;
     DEBUG(dbgs() << NumComp << " components: " << *LI << '\n');
     SmallVector<LiveInterval*, 8> Dups(1, LI);
     for (unsigned i = 1; i != NumComp; ++i) {

Modified: llvm/trunk/lib/CodeGen/SplitKit.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SplitKit.cpp?rev=130931&r1=130930&r2=130931&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SplitKit.cpp (original)
+++ llvm/trunk/lib/CodeGen/SplitKit.cpp Thu May  5 12:22:53 2011
@@ -30,6 +30,8 @@
 
 STATISTIC(NumFinished, "Number of splits finished");
 STATISTIC(NumSimple,   "Number of splits that were simple");
+STATISTIC(NumCopies,   "Number of copies inserted for splitting");
+STATISTIC(NumRemats,   "Number of rematerialized defs for splitting");
 
 //===----------------------------------------------------------------------===//
 //                                 Split Analysis
@@ -589,12 +591,14 @@
   LiveRangeEdit::Remat RM(ParentVNI);
   if (Edit->canRematerializeAt(RM, UseIdx, true, LIS)) {
     Def = Edit->rematerializeAt(MBB, I, LI->reg, RM, LIS, TII, TRI, Late);
+    ++NumRemats;
   } else {
     // Can't remat, just insert a copy from parent.
     CopyMI = BuildMI(MBB, I, DebugLoc(), TII.get(TargetOpcode::COPY), LI->reg)
                .addReg(Edit->getReg());
     Def = LIS.getSlotIndexes()->insertMachineInstrInMaps(CopyMI, Late)
             .getDefIndex();
+    ++NumCopies;
   }
 
   // Define the value in Reg.



From eli.friedman at gmail.com  Thu May  5 12:45:31 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Thu, 05 May 2011 17:45:31 -0000
Subject: [llvm-commits] [llvm] r130932 -
	/llvm/trunk/lib/Target/X86/X86InstrMMX.td
Message-ID: <20110505174531.7FFD02A6C12C@llvm.org>

Author: efriedma
Date: Thu May  5 12:45:31 2011
New Revision: 130932

URL: http://llvm.org/viewvc/llvm-project?rev=130932&view=rev
Log:
PR9848: pandn is not commutative.

No test because I can't think of any way to write one that won't break quickly.


Modified:
    llvm/trunk/lib/Target/X86/X86InstrMMX.td

Modified: llvm/trunk/lib/Target/X86/X86InstrMMX.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86InstrMMX.td?rev=130932&r1=130931&r2=130932&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86InstrMMX.td (original)
+++ llvm/trunk/lib/Target/X86/X86InstrMMX.td Thu May  5 12:45:31 2011
@@ -285,7 +285,7 @@
 defm MMX_PAND : MMXI_binop_rm_int<0xDB, "pand", int_x86_mmx_pand, 1>;
 defm MMX_POR  : MMXI_binop_rm_int<0xEB, "por" , int_x86_mmx_por,  1>;
 defm MMX_PXOR : MMXI_binop_rm_int<0xEF, "pxor", int_x86_mmx_pxor, 1>;
-defm MMX_PANDN : MMXI_binop_rm_int<0xDF, "pandn", int_x86_mmx_pandn, 1>;
+defm MMX_PANDN : MMXI_binop_rm_int<0xDF, "pandn", int_x86_mmx_pandn>;
 
 // Shift Instructions
 defm MMX_PSRLW : MMXI_binop_rmi_int<0xD1, 0x71, MRM2r, "psrlw",



From dpatel at apple.com  Thu May  5 12:54:26 2011
From: dpatel at apple.com (Devang Patel)
Date: Thu, 05 May 2011 17:54:26 -0000
Subject: [llvm-commits] [llvm] r130933 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
Message-ID: <20110505175426.3BC972A6C12C@llvm.org>

Author: dpatel
Date: Thu May  5 12:54:26 2011
New Revision: 130933

URL: http://llvm.org/viewvc/llvm-project?rev=130933&view=rev
Log:
If debug info for inlined function is missing then handle it gracefully.


Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130933&r1=130932&r2=130933&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Thu May  5 12:54:26 2011
@@ -588,12 +588,14 @@
   if (!Scope->getScopeNode())
     return NULL;
   DIScope DS(Scope->getScopeNode());
-  DIE *ScopeDIE = new DIE(dwarf::DW_TAG_inlined_subroutine);
-
   DISubprogram InlinedSP = getDISubprogram(DS);
   CompileUnit *TheCU = getCompileUnit(InlinedSP);
   DIE *OriginDIE = TheCU->getDIE(InlinedSP);
-  assert(OriginDIE && "Unable to find Origin DIE!");
+  if (!OriginDIE) {
+    DEBUG(dbgs() << "Unable to find original DIE for inlined subprogram.");
+    return NULL;
+  }
+  DIE *ScopeDIE = new DIE(dwarf::DW_TAG_inlined_subroutine);
   TheCU->addDIEEntry(ScopeDIE, dwarf::DW_AT_abstract_origin,
                      dwarf::DW_FORM_ref4, OriginDIE);
 



From resistor at mac.com  Thu May  5 12:59:04 2011
From: resistor at mac.com (Owen Anderson)
Date: Thu, 05 May 2011 17:59:04 -0000
Subject: [llvm-commits] [llvm] r130934 - in /llvm/trunk:
 include/llvm/CodeGen/FastISel.h lib/CodeGen/SelectionDAG/FastISel.cpp
Message-ID: <20110505175904.5B4D42A6C12C@llvm.org>

Author: resistor
Date: Thu May  5 12:59:04 2011
New Revision: 130934

URL: http://llvm.org/viewvc/llvm-project?rev=130934&view=rev
Log:
Allow FastISel of three-register-operand instructions.

Modified:
    llvm/trunk/include/llvm/CodeGen/FastISel.h
    llvm/trunk/lib/CodeGen/SelectionDAG/FastISel.cpp

Modified: llvm/trunk/include/llvm/CodeGen/FastISel.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/FastISel.h?rev=130934&r1=130933&r2=130934&view=diff
==============================================================================
--- llvm/trunk/include/llvm/CodeGen/FastISel.h (original)
+++ llvm/trunk/include/llvm/CodeGen/FastISel.h Thu May  5 12:59:04 2011
@@ -241,6 +241,15 @@
                            unsigned Op0, bool Op0IsKill,
                            unsigned Op1, bool Op1IsKill);
 
+  /// FastEmitInst_rrr - Emit a MachineInstr with three register operands
+  /// and a result register in the given register class.
+  ///
+  unsigned FastEmitInst_rrr(unsigned MachineInstOpcode,
+                           const TargetRegisterClass *RC,
+                           unsigned Op0, bool Op0IsKill,
+                           unsigned Op1, bool Op1IsKill,
+                           unsigned Op2, bool Op2IsKill);
+
   /// FastEmitInst_ri - Emit a MachineInstr with a register operand,
   /// an immediate, and a result register in the given register class.
   ///

Modified: llvm/trunk/lib/CodeGen/SelectionDAG/FastISel.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/SelectionDAG/FastISel.cpp?rev=130934&r1=130933&r2=130934&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/SelectionDAG/FastISel.cpp (original)
+++ llvm/trunk/lib/CodeGen/SelectionDAG/FastISel.cpp Thu May  5 12:59:04 2011
@@ -1097,6 +1097,30 @@
   return ResultReg;
 }
 
+unsigned FastISel::FastEmitInst_rrr(unsigned MachineInstOpcode,
+                                   const TargetRegisterClass *RC,
+                                   unsigned Op0, bool Op0IsKill,
+                                   unsigned Op1, bool Op1IsKill,
+                                   unsigned Op2, bool Op2IsKill) {
+  unsigned ResultReg = createResultReg(RC);
+  const TargetInstrDesc &II = TII.get(MachineInstOpcode);
+
+  if (II.getNumDefs() >= 1)
+    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II, ResultReg)
+      .addReg(Op0, Op0IsKill * RegState::Kill)
+      .addReg(Op1, Op1IsKill * RegState::Kill)
+      .addReg(Op2, Op2IsKill * RegState::Kill);
+  else {
+    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
+      .addReg(Op0, Op0IsKill * RegState::Kill)
+      .addReg(Op1, Op1IsKill * RegState::Kill)
+      .addReg(Op2, Op2IsKill * RegState::Kill);
+    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, TII.get(TargetOpcode::COPY),
+            ResultReg).addReg(II.ImplicitDefs[0]);
+  }
+  return ResultReg;
+}
+
 unsigned FastISel::FastEmitInst_ri(unsigned MachineInstOpcode,
                                    const TargetRegisterClass *RC,
                                    unsigned Op0, bool Op0IsKill,



From gkistanova at gmail.com  Thu May  5 13:40:27 2011
From: gkistanova at gmail.com (Galina Kistanova)
Date: Thu, 05 May 2011 18:40:27 -0000
Subject: [llvm-commits] [llvm] r130936 -
	/llvm/trunk/test/Archive/check_binary_output.ll
Message-ID: <20110505184027.A6DFE2A6C12C@llvm.org>

Author: gkistanova
Date: Thu May  5 13:40:27 2011
New Revision: 130936

URL: http://llvm.org/viewvc/llvm-project?rev=130936&view=rev
Log:
Many LLVM tests relies on standard output stream be in the binary mode. Which is not always the case (on Windows in particular). The patch adds a test to verify that the standard output stream is actually in the binary mode.

Added:
    llvm/trunk/test/Archive/check_binary_output.ll

Added: llvm/trunk/test/Archive/check_binary_output.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Archive/check_binary_output.ll?rev=130936&view=auto
==============================================================================
--- llvm/trunk/test/Archive/check_binary_output.ll (added)
+++ llvm/trunk/test/Archive/check_binary_output.ll Thu May  5 13:40:27 2011
@@ -0,0 +1,4 @@
+; This is not an assembly file, this is just to run the test.
+; The test verifies that llvm-ar produces a binary output.
+
+;RUN: llvm-ar p %p/GNU.a very_long_bytecode_file_name.bc | cmp -s %p/very_long_bytecode_file_name.bc -



From rafael.espindola at gmail.com  Thu May  5 13:43:39 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Thu, 05 May 2011 18:43:39 -0000
Subject: [llvm-commits] [llvm] r130937 - in /llvm/trunk:
 lib/MC/MCAsmInfo.cpp test/CodeGen/X86/empty-functions.ll
Message-ID: <20110505184339.EB0382A6C12C@llvm.org>

Author: rafael
Date: Thu May  5 13:43:39 2011
New Revision: 130937

URL: http://llvm.org/viewvc/llvm-project?rev=130937&view=rev
Log:
Don't produce a __debug_frame.
I tested both gdb on a bootstrapped clang and and the gdb testsuite on OS X (snow leopard)
and both are happy using __eh_frame.

Modified:
    llvm/trunk/lib/MC/MCAsmInfo.cpp
    llvm/trunk/test/CodeGen/X86/empty-functions.ll

Modified: llvm/trunk/lib/MC/MCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/MC/MCAsmInfo.cpp?rev=130937&r1=130936&r2=130937&view=diff
==============================================================================
--- llvm/trunk/lib/MC/MCAsmInfo.cpp (original)
+++ llvm/trunk/lib/MC/MCAsmInfo.cpp Thu May  5 13:43:39 2011
@@ -74,7 +74,7 @@
   HasLEB128 = false;
   SupportsDebugInformation = false;
   ExceptionsType = ExceptionHandling::None;
-  DwarfRequiresFrameSection = true;
+  DwarfRequiresFrameSection = false;
   DwarfUsesInlineInfoSection = false;
   DwarfRequiresRelocationForStmtList = true;
   DwarfSectionOffsetDirective = 0;

Modified: llvm/trunk/test/CodeGen/X86/empty-functions.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/empty-functions.ll?rev=130937&r1=130936&r2=130937&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/empty-functions.ll (original)
+++ llvm/trunk/test/CodeGen/X86/empty-functions.ll Thu May  5 13:43:39 2011
@@ -6,10 +6,24 @@
   unreachable
 }
 ; CHECK-NO-FP:     _func:
-; CHECK-NO-FP-NOT: movq %rsp, %rbp
+; CHECK-NO-FP-NEXT: :
+; CHECK-NO-FP-NEXT: .cfi_startproc
 ; CHECK-NO-FP:     nop
+; CHECK-NO-FP-NEXT: :
+; CHECK-NO-FP-NEXT: .cfi_endproc
 
 ; CHECK-FP:      _func:
-; CHECK-FP:      movq %rsp, %rbp
-; CHECK-FP-NEXT: Ltmp1:
-; CHECK-FP:      nop
+; CHECK-FP-NEXT: :
+; CHECK-FP-NEXT: .cfi_startproc
+; CHECK-FP-NEXT: :
+; CHECK-FP-NEXT: pushq %rbp
+; CHECK-FP-NEXT: :
+; CHECK-FP-NEXT: .cfi_def_cfa_offset 16
+; CHECK-FP-NEXT: :
+; CHECK-FP-NEXT: .cfi_offset 6, -16
+; CHECK-FP-NEXT: movq %rsp, %rbp
+; CHECK-FP-NEXT: :
+; CHECK-FP-NEXT: .cfi_def_cfa_register 6
+; CHECK-FP-NEXT: nop
+; CHECK-FP-NEXT: :
+; CHECK-FP-NEXT: .cfi_endproc



From gkistanova at gmail.com  Thu May  5 13:50:20 2011
From: gkistanova at gmail.com (Galina Kistanova)
Date: Thu, 5 May 2011 11:50:20 -0700
Subject: [llvm-commits] Check output stream opened in binary mode
 instead of text mode on MinGW
In-Reply-To: <6AE1604EE3EC5F4296C096518C6B77EE17E4EAF874@mail.accesssoftek.com>
References: <6AE1604EE3EC5F4296C096518C6B77EE17E4EAF874@mail.accesssoftek.com>
Message-ID: <BANLkTik=w4RPdc4+t91hKz63JV7ko3Frrg@mail.gmail.com>

Hi Danil,

Your patch commited. Revision: 130936.

Thanks

Galina



On Wed, May 4, 2011 at 1:05 PM, Danil Malyshev
<dmalyshev at accesssoftek.com> wrote:
> Hello everyone,
>
>
>
> Many LLVM tests relies on standard output stream be in the binary mode.
>
> Which is not always the case (on Windows in particular).
>
>
>
> The proposed patch adds a test to verify that the standard output stream is
> actually in the binary mode.
>
> This will save us a lot of troubleshooting time.
>
>
>
> Please review.
>
>
>
>
>
> Thank you,
>
> Danil
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>

From atrick at apple.com  Thu May  5 14:24:06 2011
From: atrick at apple.com (Andrew Trick)
Date: Thu, 05 May 2011 19:24:06 -0000
Subject: [llvm-commits] [llvm] r130942 -
	/llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp
Message-ID: <20110505192406.A73372A6C12C@llvm.org>

Author: atrick
Date: Thu May  5 14:24:06 2011
New Revision: 130942

URL: http://llvm.org/viewvc/llvm-project?rev=130942&view=rev
Log:
whitespace

Modified:
    llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp

Modified: llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp?rev=130942&r1=130941&r2=130942&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp (original)
+++ llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp Thu May  5 14:24:06 2011
@@ -120,7 +120,7 @@
     // such aliases.
     if (PSV->isAliased(MFI))
       return 0;
-    
+
     MayAlias = PSV->mayAlias(MFI);
     return V;
   }
@@ -174,7 +174,7 @@
     for (MachineBasicBlock::succ_iterator SI = BB->succ_begin(),
            SE = BB->succ_end(); SI != SE; ++SI)
       for (MachineBasicBlock::livein_iterator I = (*SI)->livein_begin(),
-             E = (*SI)->livein_end(); I != E; ++I) {    
+             E = (*SI)->livein_end(); I != E; ++I) {
         unsigned Reg = *I;
         if (Seen.insert(Reg))
           Uses[Reg].push_back(&ExitSU);
@@ -411,11 +411,11 @@
 #define STORE_LOAD_LATENCY 1
     unsigned TrueMemOrderLatency = 0;
     if (TID.isCall() || MI->hasUnmodeledSideEffects() ||
-        (MI->hasVolatileMemoryRef() && 
+        (MI->hasVolatileMemoryRef() &&
          (!TID.mayLoad() || !MI->isInvariantLoad(AA)))) {
       // Be conservative with these and add dependencies on all memory
       // references, even those that are known to not alias.
-      for (std::map<const Value *, SUnit *>::iterator I = 
+      for (std::map<const Value *, SUnit *>::iterator I =
              NonAliasMemDefs.begin(), E = NonAliasMemDefs.end(); I != E; ++I) {
         I->second->addPred(SDep(SU, SDep::Order, /*Latency=*/0));
       }
@@ -458,9 +458,9 @@
         // A store to a specific PseudoSourceValue. Add precise dependencies.
         // Record the def in MemDefs, first adding a dep if there is
         // an existing def.
-        std::map<const Value *, SUnit *>::iterator I = 
+        std::map<const Value *, SUnit *>::iterator I =
           ((MayAlias) ? AliasMemDefs.find(V) : NonAliasMemDefs.find(V));
-        std::map<const Value *, SUnit *>::iterator IE = 
+        std::map<const Value *, SUnit *>::iterator IE =
           ((MayAlias) ? AliasMemDefs.end() : NonAliasMemDefs.end());
         if (I != IE) {
           I->second->addPred(SDep(SU, SDep::Order, /*Latency=*/0, /*Reg=*/0,
@@ -513,37 +513,37 @@
       if (MI->isInvariantLoad(AA)) {
         // Invariant load, no chain dependencies needed!
       } else {
-        if (const Value *V = 
+        if (const Value *V =
             getUnderlyingObjectForInstr(MI, MFI, MayAlias)) {
           // A load from a specific PseudoSourceValue. Add precise dependencies.
-          std::map<const Value *, SUnit *>::iterator I = 
+          std::map<const Value *, SUnit *>::iterator I =
             ((MayAlias) ? AliasMemDefs.find(V) : NonAliasMemDefs.find(V));
-          std::map<const Value *, SUnit *>::iterator IE = 
+          std::map<const Value *, SUnit *>::iterator IE =
             ((MayAlias) ? AliasMemDefs.end() : NonAliasMemDefs.end());
           if (I != IE)
             I->second->addPred(SDep(SU, SDep::Order, /*Latency=*/0, /*Reg=*/0,
                                     /*isNormalMemory=*/true));
           if (MayAlias)
             AliasMemUses[V].push_back(SU);
-          else 
+          else
             NonAliasMemUses[V].push_back(SU);
         } else {
           // A load with no underlying object. Depend on all
           // potentially aliasing stores.
-          for (std::map<const Value *, SUnit *>::iterator I = 
+          for (std::map<const Value *, SUnit *>::iterator I =
                  AliasMemDefs.begin(), E = AliasMemDefs.end(); I != E; ++I)
             I->second->addPred(SDep(SU, SDep::Order, /*Latency=*/0));
-          
+
           PendingLoads.push_back(SU);
           MayAlias = true;
         }
-        
+
         // Add dependencies on alias and barrier chains, if needed.
         if (MayAlias && AliasChain)
           AliasChain->addPred(SDep(SU, SDep::Order, /*Latency=*/0));
         if (BarrierChain)
           BarrierChain->addPred(SDep(SU, SDep::Order, /*Latency=*/0));
-      } 
+      }
     }
   }
 
@@ -572,11 +572,11 @@
   }
 }
 
-void ScheduleDAGInstrs::ComputeOperandLatency(SUnit *Def, SUnit *Use, 
+void ScheduleDAGInstrs::ComputeOperandLatency(SUnit *Def, SUnit *Use,
                                               SDep& dep) const {
   if (!InstrItins || InstrItins->isEmpty())
     return;
-  
+
   // For a data dependency with a known register...
   if ((dep.getKind() != SDep::Data) || (dep.getReg() == 0))
     return;



From atrick at apple.com  Thu May  5 14:32:21 2011
From: atrick at apple.com (Andrew Trick)
Date: Thu, 05 May 2011 19:32:21 -0000
Subject: [llvm-commits] [llvm] r130943 -
	/llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp
Message-ID: <20110505193221.B847D2A6C12C@llvm.org>

Author: atrick
Date: Thu May  5 14:32:21 2011
New Revision: 130943

URL: http://llvm.org/viewvc/llvm-project?rev=130943&view=rev
Log:
ARM post RA scheduler compile time fix.

BuildSchedGraph was quadratic in the number of calls in the basic
block. After this fix, it keeps only a single call at the top of the
DefList so compile time doesn't blow up on large blocks. This reduces
postRA sched time on an external test case from 81s to 0.3s.  Although
r130800 (reduced ARM register alias defs) also partially fixes the
issue by reducing the constant overhead of checking call interference
by an order of magnitude.

Fixes <rdar://problem/7662664> very poor compile time with post RA scheduling.

Modified:
    llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp

Modified: llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp?rev=130943&r1=130942&r2=130943&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp (original)
+++ llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp Thu May  5 14:32:21 2011
@@ -393,6 +393,18 @@
         UseList.clear();
         if (!MO.isDead())
           DefList.clear();
+
+        // Calls will not be reordered because of chain dependencies (see
+        // below). Since call operands are dead, calls may continue to be added
+        // to the DefList making dependence checking quadratic in the size of
+        // the block. Instead, we leave only one call at the back of the
+        // DefList.
+        //
+        // NOTE: This assumes that the DefList is ordered!
+        if (SU->isCall) {
+          while (!DefList.empty() && DefList.back()->isCall)
+            DefList.pop_back();
+        }
         DefList.push_back(SU);
       } else {
         UseList.push_back(SU);



From rafael.espindola at gmail.com  Thu May  5 14:48:34 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Thu, 05 May 2011 19:48:34 -0000
Subject: [llvm-commits] [llvm] r130944 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
Message-ID: <20110505194834.547A92A6C12C@llvm.org>

Author: rafael
Date: Thu May  5 14:48:34 2011
New Revision: 130944

URL: http://llvm.org/viewvc/llvm-project?rev=130944&view=rev
Log:
List all exception types in a switch.

Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp?rev=130944&r1=130943&r2=130944&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp Thu May  5 14:48:34 2011
@@ -189,21 +189,22 @@
   if (MAI->doesSupportDebugInformation())
     DD = new DwarfDebug(this, &M);
 
-  if (MAI->doesSupportExceptionHandling())
-    switch (MAI->getExceptionHandlingType()) {
-    default:
-    case ExceptionHandling::DwarfTable:
-      DE = new DwarfTableException(this);
-      break;
-    case ExceptionHandling::DwarfCFI:
-      DE = new DwarfCFIException(this);
-      break;
-    case ExceptionHandling::ARM:
-      DE = new ARMException(this);
-      break;
-    }
+  switch (MAI->getExceptionHandlingType()) {
+  case ExceptionHandling::None:
+    return false;
+  case ExceptionHandling::SjLj:
+  case ExceptionHandling::DwarfTable:
+    DE = new DwarfTableException(this);
+    return false;
+  case ExceptionHandling::DwarfCFI:
+    DE = new DwarfCFIException(this);
+    return false;
+  case ExceptionHandling::ARM:
+    DE = new ARMException(this);
+    return false;
+  }
 
-  return false;
+  llvm_unreachable("Unknown exception type.");
 }
 
 void AsmPrinter::EmitLinkage(unsigned Linkage, MCSymbol *GVSym) const {



From gkistanova at gmail.com  Thu May  5 14:58:15 2011
From: gkistanova at gmail.com (Galina Kistanova)
Date: Thu, 5 May 2011 12:58:15 -0700
Subject: [llvm-commits] [llvm] r130877 - in /llvm/trunk:
 lib/CodeGen/MachineCSE.cpp test/CodeGen/Thumb2/thumb2-cbnz.ll
 test/CodeGen/X86/cmp-redundant.ll
In-Reply-To: <BANLkTi=LtwfhU0VyVCfMfHdbidQY6mZK+A@mail.gmail.com>
References: <20110504221037.2B8C92A6C12C@llvm.org>
	<BANLkTi=LtwfhU0VyVCfMfHdbidQY6mZK+A@mail.gmail.com>
Message-ID: <BANLkTi=ETu3cwsqaMjY2yEBR-bocpsFf0Q@mail.gmail.com>

I will look at this.

Thanks

Galina



On Wed, May 4, 2011 at 5:35 PM, Eli Friedman <eli.friedman at gmail.com> wrote:
> On Wed, May 4, 2011 at 3:10 PM, Eli Friedman <eli.friedman at gmail.com> wrote:
>> Author: efriedma
>> Date: Wed May ?4 17:10:36 2011
>> New Revision: 130877
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=130877&view=rev
>> Log:
>> Re-commit r130862 with a minor change to avoid an iterator running off the edge in some cases.
>>
>> Original message:
>>
>> Teach MachineCSE how to do simple cross-block CSE involving physregs. ?This allows, for example, eliminating duplicate cmpl's on x86. Part of rdar://problem/8259436 .
>>
>>
>
> This is breaking a few of the buildbots with the following assertion:
>
> lib/CodeGen/LiveInterval.cpp:267: llvm::LiveRange*
> llvm::LiveInterval::addRangeFrom(llvm::LiveRange, llvm::LiveRange*):
> Assertion `B->end <= Start && "Cannot overlap two LiveRanges with
> differing ValID's" " (did you def the same reg twice in a
> MachineInstr?)"' failed.
>
> I'm not sure how to go about reproducing this, so it would be very
> much appreciated if someone could send me a bitcode file where this
> reproduces.
>
> Thanks,
> Eli
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>


From eli.friedman at gmail.com  Thu May  5 15:33:12 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Thu, 5 May 2011 13:33:12 -0700
Subject: [llvm-commits] [llvm] r130877 - in /llvm/trunk:
 lib/CodeGen/MachineCSE.cpp test/CodeGen/Thumb2/thumb2-cbnz.ll
 test/CodeGen/X86/cmp-redundant.ll
In-Reply-To: <BANLkTi=ETu3cwsqaMjY2yEBR-bocpsFf0Q@mail.gmail.com>
References: <20110504221037.2B8C92A6C12C@llvm.org>
	<BANLkTi=LtwfhU0VyVCfMfHdbidQY6mZK+A@mail.gmail.com>
	<BANLkTi=ETu3cwsqaMjY2yEBR-bocpsFf0Q@mail.gmail.com>
Message-ID: <BANLkTikKacqU-NRruoiin7L5+h4z4=vK0A@mail.gmail.com>

On Thu, May 5, 2011 at 12:58 PM, Galina Kistanova <gkistanova at gmail.com> wrote:
> I will look at this.

Fortunately, I managed to get a testcase through other means;
hopefully this is working with r130925.

-Eli

> On Wed, May 4, 2011 at 5:35 PM, Eli Friedman <eli.friedman at gmail.com> wrote:
>> On Wed, May 4, 2011 at 3:10 PM, Eli Friedman <eli.friedman at gmail.com> wrote:
>>> Author: efriedma
>>> Date: Wed May ?4 17:10:36 2011
>>> New Revision: 130877
>>>
>>> URL: http://llvm.org/viewvc/llvm-project?rev=130877&view=rev
>>> Log:
>>> Re-commit r130862 with a minor change to avoid an iterator running off the edge in some cases.
>>>
>>> Original message:
>>>
>>> Teach MachineCSE how to do simple cross-block CSE involving physregs. ?This allows, for example, eliminating duplicate cmpl's on x86. Part of rdar://problem/8259436 .
>>>
>>>
>>
>> This is breaking a few of the buildbots with the following assertion:
>>
>> lib/CodeGen/LiveInterval.cpp:267: llvm::LiveRange*
>> llvm::LiveInterval::addRangeFrom(llvm::LiveRange, llvm::LiveRange*):
>> Assertion `B->end <= Start && "Cannot overlap two LiveRanges with
>> differing ValID's" " (did you def the same reg twice in a
>> MachineInstr?)"' failed.
>>
>> I'm not sure how to go about reproducing this, so it would be very
>> much appreciated if someone could send me a bitcode file where this
>> reproduces.
>>
>> Thanks,
>> Eli
>>
>> _______________________________________________
>> llvm-commits mailing list
>> llvm-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>>
>


From rafael.espindola at gmail.com  Thu May  5 15:48:31 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Thu, 05 May 2011 20:48:31 -0000
Subject: [llvm-commits] [llvm] r130947 - in
 /llvm/trunk/lib/CodeGen/AsmPrinter: AsmPrinter.cpp CMakeLists.txt
 DwarfException.h DwarfSjLjException.cpp
Message-ID: <20110505204831.62E592A6C12C@llvm.org>

Author: rafael
Date: Thu May  5 15:48:31 2011
New Revision: 130947

URL: http://llvm.org/viewvc/llvm-project?rev=130947&view=rev
Log:
Implement a really simple DwarfSjLjException.

Added:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp
Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
    llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp?rev=130947&r1=130946&r2=130947&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp Thu May  5 15:48:31 2011
@@ -193,6 +193,8 @@
   case ExceptionHandling::None:
     return false;
   case ExceptionHandling::SjLj:
+    DE = new DwarfSjLjException(this);
+    return false;
   case ExceptionHandling::DwarfTable:
     DE = new DwarfTableException(this);
     return false;

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt?rev=130947&r1=130946&r2=130947&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt Thu May  5 15:48:31 2011
@@ -8,6 +8,7 @@
   DwarfCompileUnit.cpp
   DwarfDebug.cpp
   DwarfException.cpp
+  DwarfSjLjException.cpp
   DwarfTableException.cpp
   OcamlGCPrinter.cpp
   )

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h?rev=130947&r1=130946&r2=130947&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h Thu May  5 15:48:31 2011
@@ -238,6 +238,25 @@
   virtual void EndFunction();
 };
 
+class DwarfSjLjException : public DwarfException {
+public:
+  //===--------------------------------------------------------------------===//
+  // Main entry points.
+  //
+  DwarfSjLjException(AsmPrinter *A);
+  virtual ~DwarfSjLjException();
+
+  /// EndModule - Emit all exception information that should come after the
+  /// content.
+  virtual void EndModule();
+
+  /// BeginFunction - Gather pre-function exception information.  Assumes being
+  /// emitted immediately after the function entry point.
+  virtual void BeginFunction(const MachineFunction *MF);
+
+  /// EndFunction - Gather and emit post-function exception information.
+  virtual void EndFunction();
+};
 
 class ARMException : public DwarfException {
   /// shouldEmitTable - Per-function flag to indicate if EH tables should

Added: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp?rev=130947&view=auto
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp (added)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp Thu May  5 15:48:31 2011
@@ -0,0 +1,46 @@
+//===-- CodeGen/AsmPrinter/DwarfTableException.cpp - Dwarf Exception Impl --==//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This file is a simple implementation of DwarfException that just produces
+// the exception table for use with SjLj.
+//
+//===----------------------------------------------------------------------===//
+
+#include "DwarfException.h"
+#include "llvm/CodeGen/MachineLocation.h"
+#include "llvm/CodeGen/MachineModuleInfo.h"
+using namespace llvm;
+
+DwarfSjLjException::DwarfSjLjException(AsmPrinter *A) : DwarfException(A) {
+}
+
+DwarfSjLjException::~DwarfSjLjException() {}
+
+/// EndModule - Emit all exception information that should come after the
+/// content.
+void DwarfSjLjException::EndModule() {
+}
+
+/// BeginFunction - Gather pre-function exception information. Assumes it's
+/// being emitted immediately after the function entry point.
+void DwarfSjLjException::BeginFunction(const MachineFunction *MF) {
+}
+
+/// EndFunction - Gather and emit post-function exception information.
+///
+void DwarfSjLjException::EndFunction() {
+  // Record if this personality index uses a landing pad.
+  bool HasLandingPad = !MMI->getLandingPads().empty();
+
+  // Map all labels and get rid of any dead landing pads.
+  MMI->TidyLandingPads();
+
+  if (HasLandingPad)
+    EmitExceptionTable();
+}



From nicholas at mxc.ca  Thu May  5 16:27:15 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Thu, 05 May 2011 21:27:15 -0000
Subject: [llvm-commits] [llvm] r130949 -
	/llvm/trunk/include/llvm/Support/StandardPasses.h
Message-ID: <20110505212715.257E42A6C12C@llvm.org>

Author: nicholas
Date: Thu May  5 16:27:14 2011
New Revision: 130949

URL: http://llvm.org/viewvc/llvm-project?rev=130949&view=rev
Log:
Fix typo. No functional change.

Modified:
    llvm/trunk/include/llvm/Support/StandardPasses.h

Modified: llvm/trunk/include/llvm/Support/StandardPasses.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Support/StandardPasses.h?rev=130949&r1=130948&r2=130949&view=diff
==============================================================================
--- llvm/trunk/include/llvm/Support/StandardPasses.h (original)
+++ llvm/trunk/include/llvm/Support/StandardPasses.h Thu May  5 16:27:14 2011
@@ -61,7 +61,7 @@
   /// \arg SimplifyLibCalls - Allow library calls to be simplified.
   /// \arg HaveExceptions - Whether the module may have code using exceptions.
   /// \arg InliningPass - The inlining pass to use, if any, or null. This will
-  /// always be added, even at -O0.a
+  /// always be added, even at -O0.
   static inline void createStandardModulePasses(PassManagerBase *PM,
                                                 unsigned OptimizationLevel,
                                                 bool OptimizeSize,



From isanbard at gmail.com  Thu May  5 16:30:59 2011
From: isanbard at gmail.com (Bill Wendling)
Date: Thu, 05 May 2011 21:30:59 -0000
Subject: [llvm-commits] [llvm-gcc-4.2] r130950 -
	/llvm-gcc-4.2/trunk/gcc/config/arm/arm.h
Message-ID: <20110505213100.054422A6C12C@llvm.org>

Author: void
Date: Thu May  5 16:30:59 2011
New Revision: 130950

URL: http://llvm.org/viewvc/llvm-project?rev=130950&view=rev
Log:
ARM was generating C-strings that were unnecessarily well-aligned. This wastes
memory because one aligned C-string makes *all* strings in that section
aligned. But C-strings don't need to be so well-aligned.
<rdar://problem/8961909>

Modified:
    llvm-gcc-4.2/trunk/gcc/config/arm/arm.h

Modified: llvm-gcc-4.2/trunk/gcc/config/arm/arm.h
URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/config/arm/arm.h?rev=130950&r1=130949&r2=130950&view=diff
==============================================================================
--- llvm-gcc-4.2/trunk/gcc/config/arm/arm.h (original)
+++ llvm-gcc-4.2/trunk/gcc/config/arm/arm.h Thu May  5 16:30:59 2011
@@ -646,10 +646,15 @@
 /* Make strings word-aligned so strcpy from constants will be faster.  */
 #define CONSTANT_ALIGNMENT_FACTOR (TARGET_THUMB || ! arm_tune_xscale ? 1 : 2)
 
-#define CONSTANT_ALIGNMENT(EXP, ALIGN)				\
-   ((TREE_CODE (EXP) == STRING_CST				\
-     && (ALIGN) < BITS_PER_WORD * CONSTANT_ALIGNMENT_FACTOR)	\
-    ? BITS_PER_WORD * CONSTANT_ALIGNMENT_FACTOR : (ALIGN))
+/* LLVM LOCAL begin - <rdar://problem/8961909> */
+#define CONSTANT_ALIGNMENT(EXP, ALIGN)                                 \
+  (TARGET_MACHO &&  TREE_CODE (EXP) == STRING_CST ?                    \
+   (ALIGN) :                                                           \
+   (TREE_CODE (EXP) == STRING_CST                                      \
+    && (ALIGN) < BITS_PER_WORD * CONSTANT_ALIGNMENT_FACTOR) ?          \
+   BITS_PER_WORD * CONSTANT_ALIGNMENT_FACTOR :                         \
+   (ALIGN))
+/* LLVM LOCAL end - <rdar://problem/8961909> */
 
 /* Setting STRUCTURE_SIZE_BOUNDARY to 32 produces more efficient code, but the
    value set in previous versions of this toolchain was 8, which produces more



From rafael.espindola at gmail.com  Thu May  5 16:34:33 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Thu, 05 May 2011 21:34:33 -0000
Subject: [llvm-commits] [llvm] r130951 -
	/llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp
Message-ID: <20110505213433.542122A6C12C@llvm.org>

Author: rafael
Date: Thu May  5 16:34:33 2011
New Revision: 130951

URL: http://llvm.org/viewvc/llvm-project?rev=130951&view=rev
Log:
Move PPC Linux to CFI.

Modified:
    llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp

Modified: llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp?rev=130951&r1=130950&r2=130951&view=diff
==============================================================================
--- llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp (original)
+++ llvm/trunk/lib/Target/PowerPC/PPCMCAsmInfo.cpp Thu May  5 16:34:33 2011
@@ -48,7 +48,7 @@
 
   // Exceptions handling
   if (!is64Bit)
-    ExceptionsType = ExceptionHandling::DwarfTable;
+    ExceptionsType = ExceptionHandling::DwarfCFI;
     
   ZeroDirective = "\t.space\t";
   Data64bitsDirective = is64Bit ? "\t.quad\t" : 0;



From aggarwa4 at illinois.edu  Thu May  5 16:48:50 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Thu, 05 May 2011 21:48:50 -0000
Subject: [llvm-commits] [poolalloc] r130952 -
	/poolalloc/trunk/lib/AssistDS/LoadArgs.cpp
Message-ID: <20110505214850.E00F92A6C12C@llvm.org>

Author: aggarwa4
Date: Thu May  5 16:48:50 2011
New Revision: 130952

URL: http://llvm.org/viewvc/llvm-project?rev=130952&view=rev
Log:
Minor cleanup.

Modified:
    poolalloc/trunk/lib/AssistDS/LoadArgs.cpp

Modified: poolalloc/trunk/lib/AssistDS/LoadArgs.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/LoadArgs.cpp?rev=130952&r1=130951&r2=130952&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/LoadArgs.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/LoadArgs.cpp Thu May  5 16:48:50 2011
@@ -106,12 +106,11 @@
 
               //return type is same as that of original instruction
               const FunctionType *NewFTy = FunctionType::get(CI->getType(), TP, false);
-              Function *NewF;
               numSimplified++;
               if(numSimplified > 400)
                 return true;
 
-              NewF = Function::Create(NewFTy,
+              Function *NewF = Function::Create(NewFTy,
                                       GlobalValue::InternalLinkage,
                                       F->getNameStr() + ".TEST",
                                       &M);



From eli.friedman at gmail.com  Thu May  5 16:59:18 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Thu, 05 May 2011 21:59:18 -0000
Subject: [llvm-commits] [llvm] r130954 - in /llvm/trunk:
 lib/Analysis/InstructionSimplify.cpp
 lib/Transforms/InstCombine/InstCombineCompares.cpp
 test/Transforms/InstCombine/icmp.ll
Message-ID: <20110505215918.4F94A2A6C12C@llvm.org>

Author: efriedma
Date: Thu May  5 16:59:18 2011
New Revision: 130954

URL: http://llvm.org/viewvc/llvm-project?rev=130954&view=rev
Log:
PR9838: Fix transform introduced in r127064 to not trigger when only one side of the icmp is an exact shift.


Modified:
    llvm/trunk/lib/Analysis/InstructionSimplify.cpp
    llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp
    llvm/trunk/test/Transforms/InstCombine/icmp.ll

Modified: llvm/trunk/lib/Analysis/InstructionSimplify.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Analysis/InstructionSimplify.cpp?rev=130954&r1=130953&r2=130954&view=diff
==============================================================================
--- llvm/trunk/lib/Analysis/InstructionSimplify.cpp (original)
+++ llvm/trunk/lib/Analysis/InstructionSimplify.cpp Thu May  5 16:59:18 2011
@@ -1847,7 +1847,7 @@
       // fall-through
     case Instruction::SDiv:
     case Instruction::AShr:
-      if (!LBO->isExact() && !RBO->isExact())
+      if (!LBO->isExact() || !RBO->isExact())
         break;
       if (Value *V = SimplifyICmpInst(Pred, LBO->getOperand(0),
                                       RBO->getOperand(0), TD, DT, MaxRecurse-1))

Modified: llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp?rev=130954&r1=130953&r2=130954&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp (original)
+++ llvm/trunk/lib/Transforms/InstCombine/InstCombineCompares.cpp Thu May  5 16:59:18 2011
@@ -2400,7 +2400,7 @@
         // fall-through
       case Instruction::SDiv:
       case Instruction::AShr:
-        if (!BO0->isExact() && !BO1->isExact())
+        if (!BO0->isExact() || !BO1->isExact())
           break;
         return new ICmpInst(I.getPredicate(), BO0->getOperand(0),
                             BO1->getOperand(0));

Modified: llvm/trunk/test/Transforms/InstCombine/icmp.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/InstCombine/icmp.ll?rev=130954&r1=130953&r2=130954&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/InstCombine/icmp.ll (original)
+++ llvm/trunk/test/Transforms/InstCombine/icmp.ll Thu May  5 16:59:18 2011
@@ -510,3 +510,14 @@
   ret i1 %A
 }
 
+; PR9838
+; CHECK: @test53
+; CHECK-NEXT: ashr exact
+; CHECK-NEXT: ashr
+; CHECK-NEXT: icmp
+define i1 @test53(i32 %a, i32 %b) nounwind {
+ %x = ashr exact i32 %a, 30
+ %y = ashr i32 %b, 30
+ %z = icmp eq i32 %x, %y
+ ret i1 %z
+}



From dpatel at apple.com  Thu May  5 17:00:09 2011
From: dpatel at apple.com (Devang Patel)
Date: Thu, 05 May 2011 22:00:09 -0000
Subject: [llvm-commits] [llvm] r130955 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
Message-ID: <20110505220009.256422A6C12C@llvm.org>

Author: dpatel
Date: Thu May  5 17:00:08 2011
New Revision: 130955

URL: http://llvm.org/viewvc/llvm-project?rev=130955&view=rev
Log:
Remove little used statistical counter.

Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130955&r1=130954&r2=130955&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Thu May  5 17:00:08 2011
@@ -56,10 +56,6 @@
      cl::desc("Make an absence of debug location information explicit."),
      cl::init(false));
 
-#ifndef NDEBUG
-STATISTIC(BlocksWithoutLineNo, "Number of blocks without any line number");
-#endif
-
 namespace {
   const char *DWARFGroupName = "DWARF Emission";
   const char *DbgTimerName = "DWARF Debug Writer";
@@ -1861,37 +1857,12 @@
   return DebugLoc();
 }
 
-#ifndef NDEBUG
-/// CheckLineNumbers - Count basicblocks whose instructions do not have any
-/// line number information.
-static void CheckLineNumbers(const MachineFunction *MF) {
-  for (MachineFunction::const_iterator I = MF->begin(), E = MF->end();
-       I != E; ++I) {
-    bool FoundLineNo = false;
-    for (MachineBasicBlock::const_iterator II = I->begin(), IE = I->end();
-         II != IE; ++II) {
-      const MachineInstr *MI = II;
-      if (!MI->getDebugLoc().isUnknown()) {
-        FoundLineNo = true;
-        break;
-      }
-    }
-    if (!FoundLineNo && I->size())
-      ++BlocksWithoutLineNo;      
-  }
-}
-#endif
-
 /// beginFunction - Gather pre-function debug information.  Assumes being
 /// emitted immediately after the function entry point.
 void DwarfDebug::beginFunction(const MachineFunction *MF) {
   if (!MMI->hasDebugInfo()) return;
   if (!extractScopeInformation()) return;
 
-#ifndef NDEBUG
-  CheckLineNumbers(MF);
-#endif
-
   FunctionBeginSym = Asm->GetTempSymbol("func_begin",
                                         Asm->getFunctionNumber());
   // Assumes in correct section after the entry point.



From dpatel at apple.com  Thu May  5 17:05:57 2011
From: dpatel at apple.com (Devang Patel)
Date: Thu, 05 May 2011 22:05:57 -0000
Subject: [llvm-commits] [llvm] r130957 -
	/llvm/trunk/lib/VMCore/DebugInfoProbe.cpp
Message-ID: <20110505220557.DCDD62A6C12C@llvm.org>

Author: dpatel
Date: Thu May  5 17:05:57 2011
New Revision: 130957

URL: http://llvm.org/viewvc/llvm-project?rev=130957&view=rev
Log:
In debug output, clearly list new instructions without DebugLoc.

Modified:
    llvm/trunk/lib/VMCore/DebugInfoProbe.cpp

Modified: llvm/trunk/lib/VMCore/DebugInfoProbe.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/VMCore/DebugInfoProbe.cpp?rev=130957&r1=130956&r2=130957&view=diff
==============================================================================
--- llvm/trunk/lib/VMCore/DebugInfoProbe.cpp (original)
+++ llvm/trunk/lib/VMCore/DebugInfoProbe.cpp Thu May  5 17:05:57 2011
@@ -51,7 +51,6 @@
     unsigned NumDbgLineLost, NumDbgValueLost;
     std::string PassName;
     Function *TheFn;
-    std::set<unsigned> LineNos;
     std::set<MDNode *> DbgVariables;
     std::set<Instruction *> MissingDebugLoc;
   };
@@ -60,32 +59,13 @@
 //===----------------------------------------------------------------------===//
 // DebugInfoProbeImpl
 
-static void collect(Function &F, std::set<unsigned> &Lines) {
-  for (Function::iterator FI = F.begin(), FE = F.end(); FI != FE; ++FI)
-    for (BasicBlock::iterator BI = FI->begin(), BE = FI->end(); 
-         BI != BE; ++BI) {
-      const DebugLoc &DL = BI->getDebugLoc();
-      unsigned LineNo = 0;
-      if (!DL.isUnknown()) {
-        if (MDNode *N = DL.getInlinedAt(F.getContext()))
-          LineNo = DebugLoc::getFromDILocation(N).getLine();
-        else
-          LineNo = DL.getLine();
-
-        Lines.insert(LineNo);
-      }
-    }
-}
-
 /// initialize - Collect information before running an optimization pass.
 void DebugInfoProbeImpl::initialize(StringRef PName, Function &F) {
   if (!EnableDebugInfoProbe) return;
   PassName = PName;
 
-  LineNos.clear();
   DbgVariables.clear();
   TheFn = &F;
-  collect(F, LineNos);
 
   for (Function::iterator FI = F.begin(), FE = F.end(); FI != FE; ++FI)
     for (BasicBlock::iterator BI = FI->begin(), BE = FI->end(); 
@@ -130,30 +110,16 @@
 /// must be used after initialization.
 void DebugInfoProbeImpl::finalize(Function &F) {
   if (!EnableDebugInfoProbe) return;
-  std::set<unsigned> LineNos2;
-  collect(F, LineNos2);
   assert (TheFn == &F && "Invalid function to measure!");
 
-  for (std::set<unsigned>::iterator I = LineNos.begin(),
-         E = LineNos.end(); I != E; ++I) {
-    unsigned LineNo = *I;
-    if (LineNos2.count(LineNo) == 0) {
-      DEBUG(dbgs() 
-            << "DebugInfoProbe("
-            << PassName
-            << "): Losing dbg info for source line " 
-            << LineNo << "\n");
-      ++NumDbgLineLost;
-    }
-  }
-
   std::set<MDNode *>DbgVariables2;
   for (Function::iterator FI = F.begin(), FE = F.end(); FI != FE; ++FI)
     for (BasicBlock::iterator BI = FI->begin(), BE = FI->end(); 
          BI != BE; ++BI) {
       if (BI->getDebugLoc().isUnknown() &&
           MissingDebugLoc.count(BI) == 0) {
-        DEBUG(dbgs() << "DebugInfoProbe(" << PassName << "): --- ");
+        ++NumDbgLineLost;
+        DEBUG(dbgs() << "DebugInfoProbe (" << PassName << "): --- ");
         DEBUG(BI->print(dbgs()));
         DEBUG(dbgs() << "\n");
       }



From rafael.espindola at gmail.com  Thu May  5 17:14:31 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Thu, 05 May 2011 22:14:31 -0000
Subject: [llvm-commits] [llvm] r130959 - in /llvm/trunk:
 include/llvm/MC/MCAsmInfo.h lib/CodeGen/AsmPrinter/AsmPrinter.cpp
 lib/CodeGen/LLVMTargetMachine.cpp
Message-ID: <20110505221431.A5FA82A6C12C@llvm.org>

Author: rafael
Date: Thu May  5 17:14:31 2011
New Revision: 130959

URL: http://llvm.org/viewvc/llvm-project?rev=130959&view=rev
Log:
Remove the DwarfTable enum.

Modified:
    llvm/trunk/include/llvm/MC/MCAsmInfo.h
    llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
    llvm/trunk/lib/CodeGen/LLVMTargetMachine.cpp

Modified: llvm/trunk/include/llvm/MC/MCAsmInfo.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/MC/MCAsmInfo.h?rev=130959&r1=130958&r2=130959&view=diff
==============================================================================
--- llvm/trunk/include/llvm/MC/MCAsmInfo.h (original)
+++ llvm/trunk/include/llvm/MC/MCAsmInfo.h Thu May  5 17:14:31 2011
@@ -29,7 +29,7 @@
   /// MCAsmInfo - This class is intended to be used as a base class for asm
   /// properties and features specific to the target.
   namespace ExceptionHandling {
-    enum ExceptionsType { None, DwarfTable, DwarfCFI, SjLj, ARM };
+    enum ExceptionsType { None, DwarfCFI, SjLj, ARM };
   }
 
   class MCAsmInfo {
@@ -462,8 +462,7 @@
     }
     bool isExceptionHandlingDwarf() const {
       return
-        (ExceptionsType == ExceptionHandling::DwarfTable ||
-         ExceptionsType == ExceptionHandling::DwarfCFI ||
+        (ExceptionsType == ExceptionHandling::DwarfCFI ||
          ExceptionsType == ExceptionHandling::ARM);
     }
 

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp?rev=130959&r1=130958&r2=130959&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp Thu May  5 17:14:31 2011
@@ -195,9 +195,6 @@
   case ExceptionHandling::SjLj:
     DE = new DwarfSjLjException(this);
     return false;
-  case ExceptionHandling::DwarfTable:
-    DE = new DwarfTableException(this);
-    return false;
   case ExceptionHandling::DwarfCFI:
     DE = new DwarfCFIException(this);
     return false;

Modified: llvm/trunk/lib/CodeGen/LLVMTargetMachine.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LLVMTargetMachine.cpp?rev=130959&r1=130958&r2=130959&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/LLVMTargetMachine.cpp (original)
+++ llvm/trunk/lib/CodeGen/LLVMTargetMachine.cpp Thu May  5 17:14:31 2011
@@ -324,7 +324,6 @@
     PM.add(createSjLjEHPass(getTargetLowering()));
     // FALLTHROUGH
   case ExceptionHandling::DwarfCFI:
-  case ExceptionHandling::DwarfTable:
   case ExceptionHandling::ARM:
     PM.add(createDwarfEHPass(this));
     break;



From aggarwa4 at illinois.edu  Thu May  5 17:38:31 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Thu, 05 May 2011 22:38:31 -0000
Subject: [llvm-commits] [poolalloc] r130962 - in /poolalloc/trunk:
 include/assistDS/TypeChecks.h lib/AssistDS/TypeChecks.cpp
Message-ID: <20110505223831.0E8272A6C12C@llvm.org>

Author: aggarwa4
Date: Thu May  5 17:38:30 2011
New Revision: 130962

URL: http://llvm.org/viewvc/llvm-project?rev=130962&view=rev
Log:
Updated handling of byval arguments, to take care
of both internal/external functions, correctly.

Detailed comments inline.

Olden/bh now works.

Modified:
    poolalloc/trunk/include/assistDS/TypeChecks.h
    poolalloc/trunk/lib/AssistDS/TypeChecks.cpp

Modified: poolalloc/trunk/include/assistDS/TypeChecks.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/TypeChecks.h?rev=130962&r1=130961&r2=130962&view=diff
==============================================================================
--- poolalloc/trunk/include/assistDS/TypeChecks.h (original)
+++ poolalloc/trunk/include/assistDS/TypeChecks.h Thu May  5 17:38:30 2011
@@ -59,6 +59,9 @@
   bool visitCallInst(Module &M, CallInst &CI);
   bool visitInvokeInst(Module &M, InvokeInst &CI);
   bool visitCallSite(Module &M, CallSite CS);
+  bool visitInternalFunction(Module &M, Function &F); 
+  bool visitExternalFunction(Module &M, Function &F); 
+  bool visitByValFunction(Module &M, Function &F); 
   bool visitLoadInst(Module &M, LoadInst &LI);
   bool visitStoreInst(Module &M, StoreInst &SI);
   bool visitGlobal(Module &M, GlobalVariable &GV, 

Modified: poolalloc/trunk/lib/AssistDS/TypeChecks.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/TypeChecks.cpp?rev=130962&r1=130961&r2=130962&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/TypeChecks.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/TypeChecks.cpp Thu May  5 17:38:30 2011
@@ -13,6 +13,7 @@
 
 #include "assistDS/TypeChecks.h"
 #include "llvm/Constants.h"
+#include "llvm/Transforms/Utils/Cloning.h"
 #include "llvm/DerivedTypes.h"
 #include "llvm/Module.h"
 #include "llvm/Assembly/Writer.h"
@@ -108,9 +109,11 @@
     modified |= visitGlobal(M, *I, I->getInitializer(), *MainI, 0);
   }
 
+  std::vector<Function *> toProcess;
   for (Module::iterator MI = M.begin(), ME = M.end(); MI != ME; ++MI) {
     IncorporateType(MI->getType());
     Function &F = *MI;
+    toProcess.push_back(&F);
 
     // Loop over all of the instructions in the function, 
     // adding their return type as well as the types of their operands.
@@ -138,7 +141,6 @@
       } else if (CallInst *CI = dyn_cast<CallInst>(&I)) {
         modified |= visitCallInst(M, *CI);
       } else if (InvokeInst *II = dyn_cast<InvokeInst>(&I)) {
-        II->dump();
         modified |= visitInvokeInst(M, *II);
       }
     }
@@ -146,70 +148,259 @@
 
   // Record types for byval arguments.
 
-  // Split fn into 2 clones. One internal, such that
-  // we can find all call sites to it, and also pass in the
-  // original ptr/metadata.
-  // One that sets it just to initialized memory
-  // so that it can be called from external code.
-  for (Module::iterator MI = M.begin(), ME = M.end(); MI != ME; ++MI) {
-    Function &F = *MI;
-    typedef SmallVector<Value *, 4> RegisteredArgTy;
-    RegisteredArgTy registeredArguments;
-    for (Function::arg_iterator I = F.arg_begin(), E = F.arg_end(); I != E; ++I) {
-      if (I->hasByValAttr()) {
-        assert (isa<PointerType>(I->getType()));
-        const PointerType * PT = cast<PointerType>(I->getType());
-        const Type * ET = PT->getElementType();
-        Value * AllocSize = ConstantInt::get(Int64Ty, TD->getTypeAllocSize(ET));
-        Instruction * InsertBefore = &(F.getEntryBlock().front());
-        CastInst *BCI = BitCastInst::CreatePointerCast(I, VoidPtrTy, "", InsertBefore);
-        std::vector<Value *> Args;
-        Args.push_back(BCI);
-        Args.push_back(AllocSize);
-        Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
-        Constant *F = M.getOrInsertFunction("trackInitInst", VoidTy, VoidPtrTy, Int64Ty, Int32Ty, NULL);
-        CallInst::Create(F, Args.begin(), Args.end(), "", InsertBefore);
-        registeredArguments.push_back(&*I);
-      }
+  while(!toProcess.empty()) {
+    Function *F = toProcess.back();
+    toProcess.pop_back();
+    if(F->isDeclaration())
+      continue;
+    modified |= visitByValFunction(M, *F);
+  }
+
+  return modified;
+}
+
+bool
+TypeChecks::visitByValFunction(Module &M, Function &F) {
+
+  bool hasByValArg = false;
+  for (Function::arg_iterator I = F.arg_begin(), E = F.arg_end(); I != E; ++I) {
+    if (I->hasByValAttr()) {
+      hasByValArg = true;
+      break;
     }
-    //
-    // Find all basic blocks which terminate the function.
-    //
-    std::set<BasicBlock *> exitBlocks;
-    for (inst_iterator I = inst_begin(&F), E = inst_end(&F); I != E; ++I) {
-      if (isa<ReturnInst>(*I) || isa<UnwindInst>(*I)) {
-        exitBlocks.insert(I->getParent());
-      }
+  }
+  if(!hasByValArg)
+    return false;
+
+  // For internal functions
+  //   Replace with a cloned function with extra arguments
+  //   That takes as argument the original pointers without a byval parameter too
+  //   Use them to copy the metadata over to the byval arguments
+  //   Also update all the call sites.
+
+  // For external functions
+  //  Create an internal clone (treated same as internal functions)
+  //  Modify the original function
+  //  To assume that the metadata for the byval arguments is TOP 
+  
+  if(F.hasInternalLinkage()) {
+    visitInternalFunction(M, F);
+  } else {
+    // create internal clone
+    Function *F_clone = CloneFunction(&F);
+    F_clone->setName(F.getNameStr() + "internal");
+    F.setLinkage(GlobalValue::InternalLinkage);
+    F.getParent()->getFunctionList().push_back(F_clone);
+    visitInternalFunction(M, *F_clone);
+    visitExternalFunction(M, F);
+  }
+  return true;
+}
+
+bool TypeChecks::visitInternalFunction(Module &M, Function &F) {
+
+  // Create a list of the argument types in the new function.
+  std::vector<const Type*>TP;
+  for (Function::arg_iterator I = F.arg_begin(), E = F.arg_end(); I != E; ++I) {
+    TP.push_back(I->getType());
+    // for every byval argument, add a new argument that indicates the source of
+    // the metadata. It is of the same type as the byval argument.
+    if (I->hasByValAttr())
+      TP.push_back(I->getType());
+  }
+  // Create the new function prototype
+  const FunctionType *NewFTy = FunctionType::get(F.getReturnType(), TP, false);
+  Function *NewF = Function::Create(NewFTy,
+                                    GlobalValue::InternalLinkage,
+                                    F.getNameStr() + ".INT",
+                                    &M);
+
+  Function::arg_iterator NI = NewF->arg_begin();
+  DenseMap<const Value*, Value*> ValueMap;
+  for (Function::arg_iterator II = F.arg_begin(); NI != NewF->arg_end(); ++II, ++NI) {
+    // Each new argument maps to the argument in the old function
+    // For these arguments, also copy over the attributes
+    ValueMap[II] = NI;
+    NI->setName(II->getName());
+    NI->addAttr(F.getAttributes().getParamAttributes(II->getArgNo() + 1));
+    // If we have encountered a byval argument in the old function
+    // We must skip over the next argument in the new function, as that is 
+    // the newly added source argument.
+    if(II->hasByValAttr()) {
+      NI++;
+      // Give this new argument some name, for clarity
+      NI->setName("src");
+    }
+  }
+  // Copy over the attributes for the function.
+  NewF->setAttributes(NewF->getAttributes()
+                      .addAttr(0, F.getAttributes().getRetAttributes()));
+  NewF->setAttributes(NewF->getAttributes().addAttr(~0, F.getAttributes().getFnAttributes()));
+
+  // Perform the cloning.
+  SmallVector<ReturnInst*,100> Returns;
+  CloneFunctionInto(NewF, &F, ValueMap, Returns);
+ 
+  // Add calls to the runtime to copy metadata from source to the byval argument pointer. 
+  typedef SmallVector<Value *, 4> RegisteredArgTy;
+  // Keep track of the byval arguments.
+  RegisteredArgTy registeredArguments;
+  for (Function::arg_iterator I = NewF->arg_begin(), E = NewF->arg_end(); I != E; ++I) {
+    if (I->hasByValAttr()) {
+      registeredArguments.push_back(&*I);
+      assert (isa<PointerType>(I->getType()));
+      const PointerType * PT = cast<PointerType>(I->getType());
+      const Type * ET = PT->getElementType();
+      Value * AllocSize = ConstantInt::get(Int64Ty, TD->getTypeAllocSize(ET));
+      Instruction * InsertBefore = &(NewF->getEntryBlock().front());
+      // If I is the byval argument, the next argument is the source
+      CastInst *BCI_Dest = BitCastInst::CreatePointerCast(I, VoidPtrTy, "", InsertBefore);
+      CastInst *BCI_Src = BitCastInst::CreatePointerCast(++I, VoidPtrTy, "", InsertBefore);
+      std::vector<Value *> Args;
+      Args.push_back(BCI_Dest);
+      Args.push_back(BCI_Src);
+      Args.push_back(AllocSize);
+      Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
+      Constant *F = M.getOrInsertFunction("copyTypeInfo", VoidTy, VoidPtrTy, VoidPtrTy, Int64Ty, Int32Ty, NULL);
+      CallInst::Create(F, Args.begin(), Args.end(), "", InsertBefore);
+    }
+  }
+  
+  // Find all basic blocks which terminate the function.
+  std::set<BasicBlock *> exitBlocks;
+  for (inst_iterator I = inst_begin(NewF), E = inst_end(NewF); I != E; ++I) {
+    if (isa<ReturnInst>(*I) || isa<UnwindInst>(*I)) {
+      exitBlocks.insert(I->getParent());
     }
+  }
 
-    //
-    // At each function exit, insert code to deregister all byval arguments.
-    //
-    for (std::set<BasicBlock*>::const_iterator BI = exitBlocks.begin(),
-         BE = exitBlocks.end();
-         BI != BE; ++BI) {
-      for (RegisteredArgTy::const_iterator I = registeredArguments.begin(),
-           E = registeredArguments.end();
-           I != E; ++I) {
-        SmallVector<Value *, 2> args;
-        Instruction * Pt = &((*BI)->back());
-        const PointerType * PT = cast<PointerType>((*I)->getType());
-        const Type * ET = PT->getElementType();
-        Value * AllocSize = ConstantInt::get(Int64Ty, TD->getTypeAllocSize(ET));
-        CastInst *BCI = BitCastInst::CreatePointerCast(*I, VoidPtrTy, "", Pt);
-        std::vector<Value *> Args;
-        Args.push_back(BCI);
-        Args.push_back(AllocSize);
-        Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
-        Constant *F = M.getOrInsertFunction("trackUnInitInst", VoidTy, VoidPtrTy, Int64Ty, Int32Ty, NULL);
-        CallInst::Create(F, Args.begin(), Args.end(), "", Pt);
-      }
+  // At each function exit, insert code to set the metadata as uninitialized.
+  for (std::set<BasicBlock*>::const_iterator BI = exitBlocks.begin(),
+       BE = exitBlocks.end();
+       BI != BE; ++BI) {
+    for (RegisteredArgTy::const_iterator I = registeredArguments.begin(),
+         E = registeredArguments.end();
+         I != E; ++I) {
+      SmallVector<Value *, 2> args;
+      Instruction * Pt = &((*BI)->back());
+      const PointerType * PT = cast<PointerType>((*I)->getType());
+      const Type * ET = PT->getElementType();
+      Value * AllocSize = ConstantInt::get(Int64Ty, TD->getTypeAllocSize(ET));
+      CastInst *BCI = BitCastInst::CreatePointerCast(*I, VoidPtrTy, "", Pt);
+      std::vector<Value *> Args;
+      Args.push_back(BCI);
+      Args.push_back(AllocSize);
+      Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
+      Constant *F = M.getOrInsertFunction("trackUnInitInst", VoidTy, VoidPtrTy, Int64Ty, Int32Ty, NULL);
+      CallInst::Create(F, Args.begin(), Args.end(), "", Pt);
     }
   }
 
-  return modified;
+  // Update the call sites
+  for(Value::use_iterator ui = F.use_begin(), ue = F.use_end();
+      ui != ue;)  {
+    // Check that F is the called value
+    if(CallInst *CI = dyn_cast<CallInst>(ui++)) {
+      if(CI->getCalledFunction() == &F) {
+        SmallVector<Value*, 8> Args;
+        SmallVector<AttributeWithIndex, 8> AttributesVec;
+
+        // Get the initial attributes of the call
+        AttrListPtr CallPAL = CI->getAttributes();
+        Attributes RAttrs = CallPAL.getRetAttributes();
+        Attributes FnAttrs = CallPAL.getFnAttributes();
+        
+        Function::arg_iterator II = F.arg_begin();
+
+        for(unsigned j =1;j<CI->getNumOperands();j++, II++) {
+          // Add the original argument
+          Args.push_back(CI->getOperand(j));
+          // If there are attributes on this argument, copy them to the correct 
+          // position in the AttributesVec
+          if (Attributes Attrs = CallPAL.getParamAttributes(j))
+            AttributesVec.push_back(AttributeWithIndex::get(Args.size(), Attrs));
+          // If it is a value passed as byval, add it again, as the source
+          if(II->hasByValAttr()) 
+            Args.push_back(CI->getOperand(j));
+        }
+        
+        // Create the new attributes vec.
+        if (FnAttrs != Attribute::None)
+          AttributesVec.push_back(AttributeWithIndex::get(~0, FnAttrs));
+        if (RAttrs)
+          AttributesVec.push_back(AttributeWithIndex::get(0, RAttrs));
+
+        AttrListPtr NewCallPAL = AttrListPtr::get(AttributesVec.begin(),
+                                                  AttributesVec.end());
+
+
+        // Create the substitute call
+        CallInst *CallI = CallInst::Create(NewF,Args.begin(), Args.end(),"", CI);
+        CallI->setCallingConv(CI->getCallingConv());
+        CallI->setAttributes(NewCallPAL);
+        CI->replaceAllUsesWith(CallI);
+        CI->eraseFromParent();
+      }
+    }
+  }
+  return true;
 }
 
+bool TypeChecks::visitExternalFunction(Module &M, Function &F) {
+
+  // A list of the byval arguments that we are setting metadata for
+  typedef SmallVector<Value *, 4> RegisteredArgTy;
+  RegisteredArgTy registeredArguments;
+  for (Function::arg_iterator I = F.arg_begin(), E = F.arg_end(); I != E; ++I) {
+    if (I->hasByValAttr()) {
+      assert (isa<PointerType>(I->getType()));
+      const PointerType * PT = cast<PointerType>(I->getType());
+      const Type * ET = PT->getElementType();
+      Value * AllocSize = ConstantInt::get(Int64Ty, TD->getTypeAllocSize(ET));
+      Instruction * InsertBefore = &(F.getEntryBlock().front());
+      CastInst *BCI = BitCastInst::CreatePointerCast(I, VoidPtrTy, "", InsertBefore);
+      std::vector<Value *> Args;
+      Args.push_back(BCI);
+      Args.push_back(AllocSize);
+      Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
+      // Set the metadata for the byval argument to TOP/Initialized
+      Constant *F = M.getOrInsertFunction("trackInitInst", VoidTy, VoidPtrTy, Int64Ty, Int32Ty, NULL);
+      CallInst::Create(F, Args.begin(), Args.end(), "", InsertBefore);
+      registeredArguments.push_back(&*I);
+    }
+  }
+  
+  // Find all basic blocks which terminate the function.
+  std::set<BasicBlock *> exitBlocks;
+  for (inst_iterator I = inst_begin(&F), E = inst_end(&F); I != E; ++I) {
+    if (isa<ReturnInst>(*I) || isa<UnwindInst>(*I)) {
+      exitBlocks.insert(I->getParent());
+    }
+  }
+
+  // At each function exit, insert code to set the metadata as uninitialized.
+  for (std::set<BasicBlock*>::const_iterator BI = exitBlocks.begin(),
+       BE = exitBlocks.end();
+       BI != BE; ++BI) {
+    for (RegisteredArgTy::const_iterator I = registeredArguments.begin(),
+         E = registeredArguments.end();
+         I != E; ++I) {
+      SmallVector<Value *, 2> args;
+      Instruction * Pt = &((*BI)->back());
+      const PointerType * PT = cast<PointerType>((*I)->getType());
+      const Type * ET = PT->getElementType();
+      Value * AllocSize = ConstantInt::get(Int64Ty, TD->getTypeAllocSize(ET));
+      CastInst *BCI = BitCastInst::CreatePointerCast(*I, VoidPtrTy, "", Pt);
+      std::vector<Value *> Args;
+      Args.push_back(BCI);
+      Args.push_back(AllocSize);
+      Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
+      Constant *F = M.getOrInsertFunction("trackUnInitInst", VoidTy, VoidPtrTy, Int64Ty, Int32Ty, NULL);
+      CallInst::Create(F, Args.begin(), Args.end(), "", Pt);
+    }
+  }
+  return true;
+}
 // Print the types found in the module. If the optional Module parameter is
 // passed in, then the types are printed symbolically if possible, using the
 // symbol table from the module.



From rafael.espindola at gmail.com  Thu May  5 18:20:24 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Thu, 05 May 2011 23:20:24 -0000
Subject: [llvm-commits] [llvm] r130964 - in
 /llvm/trunk/lib/CodeGen/AsmPrinter: CMakeLists.txt DwarfException.h
 DwarfSjLjException.cpp DwarfTableException.cpp
Message-ID: <20110505232111.67B9F2A6C124@llvm.org>

Author: rafael
Date: Thu May  5 18:19:54 2011
New Revision: 130964

URL: http://llvm.org/viewvc/llvm-project?rev=130964&view=rev
Log:
Remove DwarfTableException.

Removed:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp
Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt?rev=130964&r1=130963&r2=130964&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/CMakeLists.txt Thu May  5 18:19:54 2011
@@ -9,7 +9,6 @@
   DwarfDebug.cpp
   DwarfException.cpp
   DwarfSjLjException.cpp
-  DwarfTableException.cpp
   OcamlGCPrinter.cpp
   )
 

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h?rev=130964&r1=130963&r2=130964&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfException.h Thu May  5 18:19:54 2011
@@ -171,73 +171,6 @@
   virtual void EndFunction();
 };
 
-class DwarfTableException : public DwarfException {
-  /// shouldEmitTable - Per-function flag to indicate if EH tables should
-  /// be emitted.
-  bool shouldEmitTable;
-
-  /// shouldEmitMoves - Per-function flag to indicate if frame moves info
-  /// should be emitted.
-  bool shouldEmitMoves;
-
-  /// shouldEmitTableModule - Per-module flag to indicate if EH tables
-  /// should be emitted.
-  bool shouldEmitTableModule;
-
-  /// shouldEmitMovesModule - Per-module flag to indicate if frame moves
-  /// should be emitted.
-  bool shouldEmitMovesModule;
-
-  struct FunctionEHFrameInfo {
-    MCSymbol *FunctionEHSym;  // L_foo.eh
-    unsigned Number;
-    unsigned PersonalityIndex;
-    bool adjustsStack;
-    bool hasLandingPads;
-    std::vector<MachineMove> Moves;
-    const Function *function;
-
-    FunctionEHFrameInfo(MCSymbol *EHSym, unsigned Num, unsigned P,
-                        bool hC, bool hL,
-                        const std::vector<MachineMove> &M,
-                        const Function *f):
-      FunctionEHSym(EHSym), Number(Num), PersonalityIndex(P),
-      adjustsStack(hC), hasLandingPads(hL), Moves(M), function (f) { }
-  };
-
-  std::vector<FunctionEHFrameInfo> EHFrames;
-
-  /// UsesLSDA - Indicates whether an FDE that uses the CIE at the given index
-  /// uses an LSDA. If so, then we need to encode that information in the CIE's
-  /// augmentation.
-  DenseMap<unsigned, bool> UsesLSDA;
-
-  /// EmitCIE - Emit a Common Information Entry (CIE). This holds information
-  /// that is shared among many Frame Description Entries.  There is at least
-  /// one CIE in every non-empty .debug_frame section.
-  void EmitCIE(const Function *Personality, unsigned Index);
-
-  /// EmitFDE - Emit the Frame Description Entry (FDE) for the function.
-  void EmitFDE(const FunctionEHFrameInfo &EHFrameInfo);
-public:
-  //===--------------------------------------------------------------------===//
-  // Main entry points.
-  //
-  DwarfTableException(AsmPrinter *A);
-  virtual ~DwarfTableException();
-
-  /// EndModule - Emit all exception information that should come after the
-  /// content.
-  virtual void EndModule();
-
-  /// BeginFunction - Gather pre-function exception information.  Assumes being
-  /// emitted immediately after the function entry point.
-  virtual void BeginFunction(const MachineFunction *MF);
-
-  /// EndFunction - Gather and emit post-function exception information.
-  virtual void EndFunction();
-};
-
 class DwarfSjLjException : public DwarfException {
 public:
   //===--------------------------------------------------------------------===//

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp?rev=130964&r1=130963&r2=130964&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfSjLjException.cpp Thu May  5 18:19:54 2011
@@ -1,4 +1,4 @@
-//===-- CodeGen/AsmPrinter/DwarfTableException.cpp - Dwarf Exception Impl --==//
+//===-- CodeGen/AsmPrinter/DwarfSjLjException.cpp - Dwarf Exception Impl --==//
 //
 //                     The LLVM Compiler Infrastructure
 //

Removed: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp?rev=130963&view=auto
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfTableException.cpp (removed)
@@ -1,344 +0,0 @@
-//===-- CodeGen/AsmPrinter/DwarfTableException.cpp - Dwarf Exception Impl --==//
-//
-//                     The LLVM Compiler Infrastructure
-//
-// This file is distributed under the University of Illinois Open Source
-// License. See LICENSE.TXT for details.
-//
-//===----------------------------------------------------------------------===//
-//
-// This file contains support for writing DWARF exception info into asm files.
-// The implementation emits all the necessary tables "by hands".
-//
-//===----------------------------------------------------------------------===//
-
-#include "DwarfException.h"
-#include "llvm/Module.h"
-#include "llvm/CodeGen/AsmPrinter.h"
-#include "llvm/CodeGen/MachineModuleInfo.h"
-#include "llvm/CodeGen/MachineFrameInfo.h"
-#include "llvm/CodeGen/MachineFunction.h"
-#include "llvm/CodeGen/MachineLocation.h"
-#include "llvm/MC/MCAsmInfo.h"
-#include "llvm/MC/MCContext.h"
-#include "llvm/MC/MCExpr.h"
-#include "llvm/MC/MCSection.h"
-#include "llvm/MC/MCStreamer.h"
-#include "llvm/MC/MCSymbol.h"
-#include "llvm/Target/Mangler.h"
-#include "llvm/Target/TargetData.h"
-#include "llvm/Target/TargetFrameLowering.h"
-#include "llvm/Target/TargetLoweringObjectFile.h"
-#include "llvm/Target/TargetMachine.h"
-#include "llvm/Target/TargetOptions.h"
-#include "llvm/Target/TargetRegisterInfo.h"
-#include "llvm/Support/Dwarf.h"
-#include "llvm/Support/FormattedStream.h"
-#include "llvm/ADT/SmallString.h"
-#include "llvm/ADT/StringExtras.h"
-#include "llvm/ADT/Twine.h"
-using namespace llvm;
-
-DwarfTableException::DwarfTableException(AsmPrinter *A)
-  :  DwarfException(A),
-     shouldEmitTable(false), shouldEmitMoves(false),
-     shouldEmitTableModule(false), shouldEmitMovesModule(false) {}
-
-DwarfTableException::~DwarfTableException() {}
-
-/// EmitCIE - Emit a Common Information Entry (CIE). This holds information that
-/// is shared among many Frame Description Entries.  There is at least one CIE
-/// in every non-empty .debug_frame section.
-void DwarfTableException::EmitCIE(const Function *PersonalityFn, unsigned Index) {
-  // Size and sign of stack growth.
-  int stackGrowth = Asm->getTargetData().getPointerSize();
-  if (Asm->TM.getFrameLowering()->getStackGrowthDirection() ==
-      TargetFrameLowering::StackGrowsDown)
-    stackGrowth *= -1;
-
-  const TargetLoweringObjectFile &TLOF = Asm->getObjFileLowering();
-
-  // Begin eh frame section.
-  Asm->OutStreamer.SwitchSection(TLOF.getEHFrameSection());
-
-  MCSymbol *EHFrameSym;
-  if (TLOF.isFunctionEHFrameSymbolPrivate())
-    EHFrameSym = Asm->GetTempSymbol("EH_frame", Index);
-  else
-    EHFrameSym = Asm->OutContext.GetOrCreateSymbol(Twine("EH_frame") +
-                                                   Twine(Index));
-  Asm->OutStreamer.EmitLabel(EHFrameSym);
-
-  Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("section_eh_frame", Index));
-
-  // Define base labels.
-  Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("eh_frame_common", Index));
-
-  // Define the eh frame length.
-  Asm->OutStreamer.AddComment("Length of Common Information Entry");
-  Asm->EmitLabelDifference(Asm->GetTempSymbol("eh_frame_common_end", Index),
-                           Asm->GetTempSymbol("eh_frame_common_begin", Index),
-                           4);
-
-  // EH frame header.
-  Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("eh_frame_common_begin",Index));
-  Asm->OutStreamer.AddComment("CIE Identifier Tag");
-  Asm->OutStreamer.EmitIntValue(0, 4/*size*/, 0/*addrspace*/);
-  Asm->OutStreamer.AddComment("DW_CIE_VERSION");
-  Asm->OutStreamer.EmitIntValue(dwarf::DW_CIE_VERSION, 1/*size*/, 0/*addr*/);
-
-  // The personality presence indicates that language specific information will
-  // show up in the eh frame.  Find out how we are supposed to lower the
-  // personality function reference:
-
-  unsigned LSDAEncoding = TLOF.getLSDAEncoding();
-  unsigned FDEEncoding = TLOF.getFDEEncoding(false);
-  unsigned PerEncoding = TLOF.getPersonalityEncoding();
-
-  char Augmentation[6] = { 0 };
-  unsigned AugmentationSize = 0;
-  char *APtr = Augmentation + 1;
-
-  if (PersonalityFn) {
-    // There is a personality function.
-    *APtr++ = 'P';
-    AugmentationSize += 1 + Asm->GetSizeOfEncodedValue(PerEncoding);
-  }
-
-  if (UsesLSDA[Index]) {
-    // An LSDA pointer is in the FDE augmentation.
-    *APtr++ = 'L';
-    ++AugmentationSize;
-  }
-
-  if (FDEEncoding != dwarf::DW_EH_PE_absptr) {
-    // A non-default pointer encoding for the FDE.
-    *APtr++ = 'R';
-    ++AugmentationSize;
-  }
-
-  if (APtr != Augmentation + 1)
-    Augmentation[0] = 'z';
-
-  Asm->OutStreamer.AddComment("CIE Augmentation");
-  Asm->OutStreamer.EmitBytes(StringRef(Augmentation, strlen(Augmentation)+1),0);
-
-  // Round out reader.
-  Asm->EmitULEB128(1, "CIE Code Alignment Factor");
-  Asm->EmitSLEB128(stackGrowth, "CIE Data Alignment Factor");
-  Asm->OutStreamer.AddComment("CIE Return Address Column");
-
-  const TargetRegisterInfo *RI = Asm->TM.getRegisterInfo();
-  const TargetFrameLowering *TFI = Asm->TM.getFrameLowering();
-  Asm->EmitInt8(RI->getDwarfRegNum(RI->getRARegister(), true));
-
-  if (Augmentation[0]) {
-    Asm->EmitULEB128(AugmentationSize, "Augmentation Size");
-
-    // If there is a personality, we need to indicate the function's location.
-    if (PersonalityFn) {
-      Asm->EmitEncodingByte(PerEncoding, "Personality");
-      Asm->OutStreamer.AddComment("Personality");
-      Asm->EmitReference(PersonalityFn, PerEncoding);
-    }
-    if (UsesLSDA[Index])
-      Asm->EmitEncodingByte(LSDAEncoding, "LSDA");
-    if (FDEEncoding != dwarf::DW_EH_PE_absptr)
-      Asm->EmitEncodingByte(FDEEncoding, "FDE");
-  }
-
-  // Indicate locations of general callee saved registers in frame.
-  std::vector<MachineMove> Moves;
-  TFI->getInitialFrameState(Moves);
-  Asm->EmitFrameMoves(Moves, 0, true);
-
-  // On Darwin the linker honors the alignment of eh_frame, which means it must
-  // be 8-byte on 64-bit targets to match what gcc does.  Otherwise you get
-  // holes which confuse readers of eh_frame.
-  Asm->EmitAlignment(Asm->getTargetData().getPointerSize() == 4 ? 2 : 3);
-  Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("eh_frame_common_end", Index));
-}
-
-/// EmitFDE - Emit the Frame Description Entry (FDE) for the function.
-void DwarfTableException::EmitFDE(const FunctionEHFrameInfo &EHFrameInfo) {
-  assert(!EHFrameInfo.function->hasAvailableExternallyLinkage() &&
-         "Should not emit 'available externally' functions at all");
-
-  const Function *TheFunc = EHFrameInfo.function;
-  const TargetLoweringObjectFile &TLOF = Asm->getObjFileLowering();
-
-  unsigned LSDAEncoding = TLOF.getLSDAEncoding();
-  unsigned FDEEncoding = TLOF.getFDEEncoding(false);
-
-  Asm->OutStreamer.SwitchSection(TLOF.getEHFrameSection());
-
-  // If corresponding function is weak definition, this should be too.
-  if (TheFunc->isWeakForLinker() && Asm->MAI->getWeakDefDirective())
-    Asm->OutStreamer.EmitSymbolAttribute(EHFrameInfo.FunctionEHSym,
-                                         MCSA_WeakDefinition);
-
-  // If corresponding function is hidden, this should be too.
-  if (TheFunc->hasHiddenVisibility())
-    if (MCSymbolAttr HiddenAttr = Asm->MAI->getHiddenVisibilityAttr())
-      Asm->OutStreamer.EmitSymbolAttribute(EHFrameInfo.FunctionEHSym,
-                                           HiddenAttr);
-
-  // If there are no calls then you can't unwind.  This may mean we can omit the
-  // EH Frame, but some environments do not handle weak absolute symbols. If
-  // UnwindTablesMandatory is set we cannot do this optimization; the unwind
-  // info is to be available for non-EH uses.
-  if (!EHFrameInfo.adjustsStack && !UnwindTablesMandatory &&
-      (!TheFunc->isWeakForLinker() ||
-       !Asm->MAI->getWeakDefDirective() ||
-       TLOF.getSupportsWeakOmittedEHFrame())) {
-    Asm->OutStreamer.EmitAssignment(EHFrameInfo.FunctionEHSym,
-                                    MCConstantExpr::Create(0, Asm->OutContext));
-    // This name has no connection to the function, so it might get
-    // dead-stripped when the function is not, erroneously.  Prohibit
-    // dead-stripping unconditionally.
-    if (Asm->MAI->hasNoDeadStrip())
-      Asm->OutStreamer.EmitSymbolAttribute(EHFrameInfo.FunctionEHSym,
-                                           MCSA_NoDeadStrip);
-  } else {
-    Asm->OutStreamer.EmitLabel(EHFrameInfo.FunctionEHSym);
-
-    // EH frame header.
-    Asm->OutStreamer.AddComment("Length of Frame Information Entry");
-    Asm->EmitLabelDifference(
-                Asm->GetTempSymbol("eh_frame_end", EHFrameInfo.Number),
-                Asm->GetTempSymbol("eh_frame_begin", EHFrameInfo.Number), 4);
-
-    Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("eh_frame_begin",
-                                                  EHFrameInfo.Number));
-
-    Asm->OutStreamer.AddComment("FDE CIE offset");
-    Asm->EmitLabelDifference(
-                       Asm->GetTempSymbol("eh_frame_begin", EHFrameInfo.Number),
-                       Asm->GetTempSymbol("eh_frame_common",
-                                          EHFrameInfo.PersonalityIndex), 4);
-
-    MCSymbol *EHFuncBeginSym =
-      Asm->GetTempSymbol("eh_func_begin", EHFrameInfo.Number);
-
-    Asm->OutStreamer.AddComment("FDE initial location");
-    Asm->EmitReference(EHFuncBeginSym, FDEEncoding);
-
-    Asm->OutStreamer.AddComment("FDE address range");
-    Asm->EmitLabelDifference(Asm->GetTempSymbol("eh_func_end",
-                                                EHFrameInfo.Number),
-                             EHFuncBeginSym,
-                             Asm->GetSizeOfEncodedValue(FDEEncoding));
-
-    // If there is a personality and landing pads then point to the language
-    // specific data area in the exception table.
-    if (MMI->getPersonalities()[0] != NULL) {
-      unsigned Size = Asm->GetSizeOfEncodedValue(LSDAEncoding);
-
-      Asm->EmitULEB128(Size, "Augmentation size");
-      Asm->OutStreamer.AddComment("Language Specific Data Area");
-      if (EHFrameInfo.hasLandingPads)
-        Asm->EmitReference(Asm->GetTempSymbol("exception", EHFrameInfo.Number),
-                           LSDAEncoding);
-      else
-        Asm->OutStreamer.EmitIntValue(0, Size/*size*/, 0/*addrspace*/);
-
-    } else {
-      Asm->EmitULEB128(0, "Augmentation size");
-    }
-
-    // Indicate locations of function specific callee saved registers in frame.
-    Asm->EmitFrameMoves(EHFrameInfo.Moves, EHFuncBeginSym, true);
-
-    // On Darwin the linker honors the alignment of eh_frame, which means it
-    // must be 8-byte on 64-bit targets to match what gcc does.  Otherwise you
-    // get holes which confuse readers of eh_frame.
-    Asm->EmitAlignment(Asm->getTargetData().getPointerSize() == 4 ? 2 : 3);
-    Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("eh_frame_end",
-                                                  EHFrameInfo.Number));
-
-    // If the function is marked used, this table should be also.  We cannot
-    // make the mark unconditional in this case, since retaining the table also
-    // retains the function in this case, and there is code around that depends
-    // on unused functions (calling undefined externals) being dead-stripped to
-    // link correctly.  Yes, there really is.
-    if (MMI->isUsedFunction(EHFrameInfo.function))
-      if (Asm->MAI->hasNoDeadStrip())
-        Asm->OutStreamer.EmitSymbolAttribute(EHFrameInfo.FunctionEHSym,
-                                             MCSA_NoDeadStrip);
-  }
-  Asm->OutStreamer.AddBlankLine();
-}
-
-/// EndModule - Emit all exception information that should come after the
-/// content.
-void DwarfTableException::EndModule() {
-  if (!Asm->MAI->isExceptionHandlingDwarf())
-    return;
-
-  if (!shouldEmitMovesModule && !shouldEmitTableModule)
-    return;
-
-  const std::vector<const Function*> &Personalities = MMI->getPersonalities();
-
-  for (unsigned I = 0, E = Personalities.size(); I < E; ++I)
-    EmitCIE(Personalities[I], I);
-
-  for (std::vector<FunctionEHFrameInfo>::iterator
-         I = EHFrames.begin(), E = EHFrames.end(); I != E; ++I)
-    EmitFDE(*I);
-}
-
-/// BeginFunction - Gather pre-function exception information. Assumes it's
-/// being emitted immediately after the function entry point.
-void DwarfTableException::BeginFunction(const MachineFunction *MF) {
-  shouldEmitTable = shouldEmitMoves = false;
-
-  // If any landing pads survive, we need an EH table.
-  shouldEmitTable = !MMI->getLandingPads().empty();
-
-  // See if we need frame move info.
-  shouldEmitMoves =
-    !Asm->MF->getFunction()->doesNotThrow() || UnwindTablesMandatory;
-
-  if (shouldEmitMoves || shouldEmitTable)
-    // Assumes in correct section after the entry point.
-    Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("eh_func_begin",
-                                                  Asm->getFunctionNumber()));
-
-  shouldEmitTableModule |= shouldEmitTable;
-  shouldEmitMovesModule |= shouldEmitMoves;
-}
-
-/// EndFunction - Gather and emit post-function exception information.
-///
-void DwarfTableException::EndFunction() {
-  if (!shouldEmitMoves && !shouldEmitTable) return;
-
-  Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("eh_func_end",
-                                                Asm->getFunctionNumber()));
-
-  // Record if this personality index uses a landing pad.
-  bool HasLandingPad = !MMI->getLandingPads().empty();
-  UsesLSDA[MMI->getPersonalityIndex()] |= HasLandingPad;
-
-  // Map all labels and get rid of any dead landing pads.
-  MMI->TidyLandingPads();
-
-  if (HasLandingPad)
-    EmitExceptionTable();
-
-  const TargetLoweringObjectFile &TLOF = Asm->getObjFileLowering();
-  MCSymbol *FunctionEHSym =
-    Asm->GetSymbolWithGlobalValueBase(Asm->MF->getFunction(), ".eh",
-                                      TLOF.isFunctionEHFrameSymbolPrivate());
-
-  // Save EH frame information
-  EHFrames.
-    push_back(FunctionEHFrameInfo(FunctionEHSym,
-                                  Asm->getFunctionNumber(),
-                                  MMI->getPersonalityIndex(),
-                                  Asm->MF->getFrameInfo()->adjustsStack(),
-                                  !MMI->getLandingPads().empty(),
-                                  MMI->getFrameMoves(),
-                                  Asm->MF->getFunction()));
-}



From nicholas at mxc.ca  Thu May  5 18:52:18 2011
From: nicholas at mxc.ca (Nick Lewycky)
Date: Thu, 05 May 2011 23:52:18 -0000
Subject: [llvm-commits] [llvm] r130967 - in /llvm/trunk:
 lib/Transforms/Instrumentation/GCOVProfiling.cpp
 runtime/libprofile/GCDAProfiling.c
Message-ID: <20110505235218.44D1F2A6C12C@llvm.org>

Author: nicholas
Date: Thu May  5 18:52:18 2011
New Revision: 130967

URL: http://llvm.org/viewvc/llvm-project?rev=130967&view=rev
Log:
The computation of string length is not that complicated. Fix it, again. :)

Modified:
    llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
    llvm/trunk/runtime/libprofile/GCDAProfiling.c

Modified: llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp?rev=130967&r1=130966&r2=130967&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp (original)
+++ llvm/trunk/lib/Transforms/Instrumentation/GCOVProfiling.cpp Thu May  5 18:52:18 2011
@@ -140,7 +140,7 @@
       // A GCOV string is a length, followed by a NUL, then between 0 and 3 NULs
       // padding out to the next 4-byte word. The length is measured in 4-byte
       // words including padding, not bytes of actual string.
-      return (s.size() + 5) / 4;
+      return (s.size() / 4) + 1;
     }
 
     void writeGCOVString(StringRef s) {

Modified: llvm/trunk/runtime/libprofile/GCDAProfiling.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/runtime/libprofile/GCDAProfiling.c?rev=130967&r1=130966&r2=130967&view=diff
==============================================================================
--- llvm/trunk/runtime/libprofile/GCDAProfiling.c (original)
+++ llvm/trunk/runtime/libprofile/GCDAProfiling.c Thu May  5 18:52:18 2011
@@ -49,7 +49,7 @@
 }
 
 static uint32_t length_of_string(const char *s) {
-  return (strlen(s) + 5) / 4;
+  return (strlen(s) / 4) + 1;
 }
 
 static void write_string(const char *s) {



From isanbard at gmail.com  Thu May  5 19:14:25 2011
From: isanbard at gmail.com (Bill Wendling)
Date: Fri, 06 May 2011 00:14:25 -0000
Subject: [llvm-commits] [llvm-gcc-4.2] r130969 -
	/llvm-gcc-4.2/trunk/gcc/objc/objc-act.c
Message-ID: <20110506001425.A2EBB2A6C12C@llvm.org>

Author: void
Date: Thu May  5 19:14:25 2011
New Revision: 130969

URL: http://llvm.org/viewvc/llvm-project?rev=130969&view=rev
Log:
This is the llvm-gcc version of the "packing ivar offsets together" fix for
clang (r130870). Basically, we place all ivars into their own section so that if
they slide, they don't dirty as many pages as when they are mixed in with other
objects that don't "slide".
<rdar://problem/9393640>

Modified:
    llvm-gcc-4.2/trunk/gcc/objc/objc-act.c

Modified: llvm-gcc-4.2/trunk/gcc/objc/objc-act.c
URL: http://llvm.org/viewvc/llvm-project/llvm-gcc-4.2/trunk/gcc/objc/objc-act.c?rev=130969&r1=130968&r2=130969&view=diff
==============================================================================
--- llvm-gcc-4.2/trunk/gcc/objc/objc-act.c (original)
+++ llvm-gcc-4.2/trunk/gcc/objc/objc-act.c Thu May  5 19:14:25 2011
@@ -7725,6 +7725,14 @@
      by the front-end. Otherwise, FE asserts when attempting to do futher
      math on the tree whose one operand is one of these offsets. */
   offset = create_extern_decl (TREE_TYPE (size_zero_node), var_offset_name);
+
+  /* APPLE LOCAL begin - radar 9393640 */
+  {
+    const char string[] = "__DATA,__objc_ivar";
+    DECL_SECTION_NAME (offset) = build_string (strlen(string), string);
+  }
+  /* APPLE LOCAL end - radar 9393640 */
+
   return offset;
 }
 
@@ -7802,6 +7810,13 @@
   /* APPLE LOCAL radar 4441049 */
   offset = create_extern_decl (TREE_TYPE (size_zero_node), var_offset_name);
 
+  /* APPLE LOCAL begin - radar 9393640 */
+  {
+    const char string[] = "__DATA,__objc_ivar";
+    DECL_SECTION_NAME (offset) = build_string (strlen(string), string);
+  }
+  /* APPLE LOCAL end - radar 9393640 */
+
   ftype = TREE_TYPE (field);
 
   /* (char*)datum */
@@ -11148,6 +11163,13 @@
   /* APPLE LOCAL begin radar 4441049 */
   *chain = tree_cons (decl, byte_position (field_decl), NULL_TREE);
   /* APPLE LOCAL end radar 4441049 */
+
+  /* APPLE LOCAL begin - radar 9393640 */
+  {
+    const char string[] = "__DATA,__objc_ivar";
+    DECL_SECTION_NAME (decl) = build_string (strlen(string), string);
+  }
+  /* APPLE LOCAL end - radar 9393640 */
   
   return decl;
 }



From mttjwl at gmail.com  Thu May  5 20:18:20 2011
From: mttjwl at gmail.com (Matthew Wala)
Date: Fri, 06 May 2011 01:18:20 -0000
Subject: [llvm-commits] [poolalloc] r130972 - in /poolalloc/trunk:
 include/dsa/CStdLib.h lib/DSA/StdLibPass.cpp
Message-ID: <20110506011820.209272A6C12C@llvm.org>

Author: wala1
Date: Thu May  5 20:18:19 2011
New Revision: 130972

URL: http://llvm.org/viewvc/llvm-project?rev=130972&view=rev
Log:
Added entries for pool_strcasestr().

Modified:
    poolalloc/trunk/include/dsa/CStdLib.h
    poolalloc/trunk/lib/DSA/StdLibPass.cpp

Modified: poolalloc/trunk/include/dsa/CStdLib.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/dsa/CStdLib.h?rev=130972&r1=130971&r2=130972&view=diff
==============================================================================
--- poolalloc/trunk/include/dsa/CStdLib.h (original)
+++ poolalloc/trunk/include/dsa/CStdLib.h Thu May  5 20:18:19 2011
@@ -44,6 +44,7 @@
     { "pool_bcopy",       2 },
     { "pool_index",       1 },
     { "pool_rindex",      1 },
+    { "pool_strcasestr",  2 },
     {  0,                 0 }
   };
 

Modified: poolalloc/trunk/lib/DSA/StdLibPass.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/DSA/StdLibPass.cpp?rev=130972&r1=130971&r2=130972&view=diff
==============================================================================
--- poolalloc/trunk/lib/DSA/StdLibPass.cpp (original)
+++ poolalloc/trunk/lib/DSA/StdLibPass.cpp Thu May  5 20:18:19 2011
@@ -279,6 +279,7 @@
   {"pool_strcat",     {NRET_NNYARGS, YRET_NNYARGS,  NRET_NARGS, YRET_NNYARGS,  true}},
   {"pool_strncat",    {NRET_NNYARGS, YRET_NNYARGS,  NRET_NARGS, YRET_NNYARGS,  true}},
   {"pool_strstr",     {NRET_NNYARGS, YRET_NARGS,    NRET_NARGS, YRET_NNYNARGS, true}},
+  {"pool_strcasestr", {NRET_NNYARGS, YRET_NARGS,    NRET_NARGS, YRET_NNYNARGS, true}},
   {"pool_strpbrk",    {NRET_NNYARGS, YRET_NARGS,    NRET_NARGS, YRET_NNYNARGS, true}},
   {"pool_strspn",     {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, NRET_NARGS,    true}},
   {"pool_strcspn",    {NRET_NYARGS,  YRET_NARGS,    NRET_NARGS, NRET_NARGS,    true}},



From etherzhhb at gmail.com  Thu May  5 21:38:20 2011
From: etherzhhb at gmail.com (Hongbin Zheng)
Date: Fri, 06 May 2011 02:38:20 -0000
Subject: [llvm-commits] [polly] r130975 - in /polly/trunk:
 include/polly/ScopDetection.h lib/Analysis/ScopDetection.cpp
 lib/CodeGeneration.cpp test/CodeGen/OpenMP/invalidate_subfn_scops.ll
Message-ID: <20110506023820.391392A6C12C@llvm.org>

Author: ether
Date: Thu May  5 21:38:20 2011
New Revision: 130975

URL: http://llvm.org/viewvc/llvm-project?rev=130975&view=rev
Log:
ScopDetection: Remember the functions generated by backend in a pointer set, so we
  do not re-generate code for these functions.

Modified:
    polly/trunk/include/polly/ScopDetection.h
    polly/trunk/lib/Analysis/ScopDetection.cpp
    polly/trunk/lib/CodeGeneration.cpp
    polly/trunk/test/CodeGen/OpenMP/invalidate_subfn_scops.ll

Modified: polly/trunk/include/polly/ScopDetection.h
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/include/polly/ScopDetection.h?rev=130975&r1=130974&r2=130975&view=diff
==============================================================================
--- polly/trunk/include/polly/ScopDetection.h (original)
+++ polly/trunk/include/polly/ScopDetection.h Thu May  5 21:38:20 2011
@@ -102,6 +102,10 @@
   typedef std::set<const Region*> RegionSet;
   RegionSet ValidRegions;
 
+  // Remember the invalid functions producted by backends;
+  typedef std::set<const Function*> FunctionSet;
+  FunctionSet InvalidFunctions;
+
   // Try to expand the region R. If R can be expanded return the expanded
   // region, NULL otherwise.
   Region *expandRegion(Region &R);
@@ -248,6 +252,14 @@
     ValidRegions.erase(&R);
   }
 
+  /// @brief Mark the function as invalid so we will not extract any scop from
+  ///        the function.
+  ///
+  /// @param F The function to mark as invalid.
+  void markFunctionAsInvalid(const Function *F) {
+    InvalidFunctions.insert(F);
+  }
+
   /// @brief Verify if all valid Regions in this Function are still valid
   /// after some transformations.
   void verifyAnalysis() const;

Modified: polly/trunk/lib/Analysis/ScopDetection.cpp
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/lib/Analysis/ScopDetection.cpp?rev=130975&r1=130974&r2=130975&view=diff
==============================================================================
--- polly/trunk/lib/Analysis/ScopDetection.cpp (original)
+++ polly/trunk/lib/Analysis/ScopDetection.cpp Thu May  5 21:38:20 2011
@@ -594,12 +594,7 @@
 }
 
 bool ScopDetection::isValidFunction(llvm::Function &F) {
-  const std::string &Name = F.getNameStr();
-  size_t found = Name.find(".omp_subfn");
-  if (found != std::string::npos)
-    return false;
-  else
-    return true;
+  return !InvalidFunctions.count(&F);
 }
 
 bool ScopDetection::runOnFunction(llvm::Function &F) {
@@ -650,6 +645,7 @@
 
 void ScopDetection::releaseMemory() {
   ValidRegions.clear();
+  // Do not clear the invalid function set.
 }
 
 char ScopDetection::ID = 0;

Modified: polly/trunk/lib/CodeGeneration.cpp
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/lib/CodeGeneration.cpp?rev=130975&r1=130974&r2=130975&view=diff
==============================================================================
--- polly/trunk/lib/CodeGeneration.cpp (original)
+++ polly/trunk/lib/CodeGeneration.cpp Thu May  5 21:38:20 2011
@@ -678,8 +678,8 @@
   // The Scop we code generate.
   Scop *S;
   ScalarEvolution &SE;
-
   DominatorTree *DT;
+  ScopDetection *SD;
   Dependences *DP;
   TargetData *TD;
 
@@ -834,6 +834,8 @@
     std::vector<const Type*> Arguments(1, Builder.getInt8PtrTy());
     FunctionType *FT = FunctionType::get(Builder.getVoidTy(), Arguments, false);
     Function *FN = Function::Create(FT, Function::InternalLinkage, Name, M);
+    // Do not run any polly pass on the new function.
+    SD->markFunctionAsInvalid(FN);
 
     Function::arg_iterator AI = FN->arg_begin();
     AI->setName("omp.userContext");
@@ -1251,8 +1253,10 @@
   }
 
   ClastStmtCodeGen(Scop *scop, ScalarEvolution &se, DominatorTree *dt,
-                   Dependences *dp, TargetData *td, IRBuilder<> &B) :
-    S(scop), SE(se), DT(dt), DP(dp), TD(td), Builder(B), ExpGen(Builder, NULL) {}
+                   ScopDetection *sd, Dependences *dp, TargetData *td,
+                   IRBuilder<> &B) :
+    S(scop), SE(se), DT(dt), SD(sd), DP(dp), TD(td), Builder(B),
+    ExpGen(Builder, NULL) {}
 
 };
 }
@@ -1369,7 +1373,7 @@
 
     const clast_root *clast = (const clast_root *) C->getClast();
 
-    ClastStmtCodeGen CodeGen(S, *SE, DT, DP, TD, Builder);
+    ClastStmtCodeGen CodeGen(S, *SE, DT, SD, DP, TD, Builder);
 
     if (OpenMP)
       addOpenMPDefinitions(Builder);

Modified: polly/trunk/test/CodeGen/OpenMP/invalidate_subfn_scops.ll
URL: http://llvm.org/viewvc/llvm-project/polly/trunk/test/CodeGen/OpenMP/invalidate_subfn_scops.ll?rev=130975&r1=130974&r2=130975&view=diff
==============================================================================
--- polly/trunk/test/CodeGen/OpenMP/invalidate_subfn_scops.ll (original)
+++ polly/trunk/test/CodeGen/OpenMP/invalidate_subfn_scops.ll Thu May  5 21:38:20 2011
@@ -1,5 +1,5 @@
 ; ModuleID = 'test.ll'
-; RUN: opt %loadPolly %defaultOpts -basicaa -polly-codegen -enable-polly-openmp  < %s | not FileCheck %s
+; RUN: opt %loadPolly %defaultOpts -polly-codegen -enable-polly-openmp -analyze %s -debug-only=polly-detect 2>&1 | not FileCheck %s
 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32-n8:16:32"
 target triple = "i386-pc-linux-gnu"
 
@@ -35,4 +35,4 @@
 }
 
 
-; CHECK: define internal void @main.omp_subfn.omp_subfn
+; CHECK: Checking region: omp.setup



From aggarwa4 at illinois.edu  Thu May  5 21:51:38 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Fri, 06 May 2011 02:51:38 -0000
Subject: [llvm-commits] [poolalloc] r130976 -
	/poolalloc/trunk/lib/AssistDS/TypeChecks.cpp
Message-ID: <20110506025138.2A9062A6C12C@llvm.org>

Author: aggarwa4
Date: Thu May  5 21:51:37 2011
New Revision: 130976

URL: http://llvm.org/viewvc/llvm-project?rev=130976&view=rev
Log:
Add support for read. 
Needed for gzip.

Modified:
    poolalloc/trunk/lib/AssistDS/TypeChecks.cpp

Modified: poolalloc/trunk/lib/AssistDS/TypeChecks.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/TypeChecks.cpp?rev=130976&r1=130975&r2=130976&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/TypeChecks.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/TypeChecks.cpp Thu May  5 21:51:37 2011
@@ -578,6 +578,16 @@
         CallInst::Create(F, Args.begin(), Args.end(), "", I);
         break;
       }
+    } else if(F->getNameStr() == std::string("read")) {
+      CastInst *BCI = BitCastInst::CreatePointerCast(I->getOperand(2), VoidPtrTy);
+      BCI->insertAfter(I);
+      std::vector<Value *> Args;
+      Args.push_back(BCI);
+      Args.push_back(I);
+      Args.push_back(ConstantInt::get(Int32Ty, tagCounter++));
+      Constant *F = M.getOrInsertFunction("trackInitInst", VoidTy, VoidPtrTy, I->getType(), Int32Ty, NULL);
+      CallInst *CI = CallInst::Create(F, Args.begin(), Args.end());
+      CI->insertAfter(BCI);
     } else if(F->getNameStr() == std::string("calloc")) {
       CastInst *BCI = BitCastInst::CreatePointerCast(I, VoidPtrTy);
       BCI->insertAfter(I);
@@ -596,8 +606,6 @@
       F = M.getOrInsertFunction("trackGlobalArray", VoidTy, VoidPtrTy, I->getOperand(2)->getType(), I->getOperand(1)->getType(), Int32Ty, NULL);
       CallInst *CI_Arr = CallInst::Create(F, Args1.begin(), Args1.end());
       CI_Arr->insertAfter(CI);
-
-
     } else if(F->getNameStr() ==  std::string("realloc")) {
       CastInst *BCI_Src = BitCastInst::CreatePointerCast(I->getOperand(1), VoidPtrTy);
       CastInst *BCI_Dest = BitCastInst::CreatePointerCast(I, VoidPtrTy);



From nobled at dreamwidth.org  Thu May  5 22:41:16 2011
From: nobled at dreamwidth.org (nobled)
Date: Thu, 5 May 2011 23:41:16 -0400
Subject: [llvm-commits] [PATCHES] fix duplication and FIXMEs in JIT code
In-Reply-To: <BANLkTikY6UseB3du3ajhtpGtd4hB2E1zog@mail.gmail.com>
References: <BANLkTi=xMh=W_iY6MqYf2OLmDnDUmDGZCg@mail.gmail.com>
	<BANLkTikY6UseB3du3ajhtpGtd4hB2E1zog@mail.gmail.com>
Message-ID: <BANLkTikX6wxZLR3oGfex=tbx6YAx-07fsw@mail.gmail.com>

ping?

On Sun, Apr 24, 2011 at 6:20 PM, nobled <nobled at dreamwidth.org> wrote:
> Whoops, forgot to join the mailing list, so this got stuck in the
> moderation queue...
>
> ---------- Forwarded message ----------
> From: nobled <nobled at dreamwidth.org>
> Date: Sun, Apr 24, 2011 at 11:05 AM
> Subject: [PATCHES] fix duplication and FIXMEs in JIT code
> To: llvm-commits at cs.uiuc.edu
>
>
> (Just so you know, I do have commit access.)
>
> These patches address this comment from the MCJIT code:
>
> ?// FIXME: This should be lifted out of here, it isn't something which should
> ?// be part of the JIT policy, rather the burden for this selection should be
> ?// pushed to clients.
> ====
> [PATCH 1/3] ExecutionEngine: fix JIT/MCJIT selectTarget() duplication
>
> (Merges the near-identical files lib/EE/JIT/TargetSelect.cpp and
> lib/EE/MCJIT/TargetSelect.cpp into lib/EE/TargetSelect.cpp.)
>
> This prepares for making JITCtor/MCJITCtor take a
> TargetMachine* directly from clients like EngineBuilder.
> ====
> [PATCH 2/3] ExecutionEngine: push TargetMachine creation into clients
>
> In particular, into EngineBuilder. This should only impact
> the private API between the EE and EB classes, not external
> clients, since JITCtor and MCJITCtor are both protected members.
> ====
> [PATCH 3/3] ExecutionEngine: move createJIT definition
>
> As an ExecutionEngine class function, its definition
> really belongs in ExecutionEngine.cpp instead of in JIT.cpp.
> ====
>
> As for this comment:
>
> ? // Try to register the program as a source of symbols to resolve against.
> ? //
> ? // FIXME: Don't do this here.
> ? sys::DynamicLibrary::LoadLibraryPermanently(0, NULL);
>
> where in the stack *should* that go, ideally?
>


From eli.friedman at gmail.com  Fri May  6 00:23:07 2011
From: eli.friedman at gmail.com (Eli Friedman)
Date: Fri, 06 May 2011 05:23:07 -0000
Subject: [llvm-commits] [llvm] r130977 - in /llvm/trunk:
 lib/CodeGen/MachineCSE.cpp test/CodeGen/Thumb2/thumb2-cbnz.ll
 test/CodeGen/X86/cmp-redundant.ll
Message-ID: <20110506052307.7FEE82A6C12C@llvm.org>

Author: efriedma
Date: Fri May  6 00:23:07 2011
New Revision: 130977

URL: http://llvm.org/viewvc/llvm-project?rev=130977&view=rev
Log:
Re-revert r130877; it's apparently causing a regression on 197.parser,
possibly related to cbnz formation.


Removed:
    llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
Modified:
    llvm/trunk/lib/CodeGen/MachineCSE.cpp
    llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll

Modified: llvm/trunk/lib/CodeGen/MachineCSE.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/MachineCSE.cpp?rev=130977&r1=130976&r2=130977&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/MachineCSE.cpp (original)
+++ llvm/trunk/lib/CodeGen/MachineCSE.cpp Fri May  6 00:23:07 2011
@@ -33,8 +33,6 @@
 STATISTIC(NumCSEs,      "Number of common subexpression eliminated");
 STATISTIC(NumPhysCSEs,
           "Number of physreg referencing common subexpr eliminated");
-STATISTIC(NumCrossBlockPhysCSEs,
-          "Number of physreg common subexprs cross-block eliminated");
 STATISTIC(NumCommutes,  "Number of copies coalesced after commuting");
 
 namespace {
@@ -84,8 +82,7 @@
                                 MachineBasicBlock::const_iterator E) const ;
     bool hasLivePhysRegDefUses(const MachineInstr *MI,
                                const MachineBasicBlock *MBB,
-                               SmallSet<unsigned,8> &PhysRefs,
-                               SmallVector<unsigned,8> &PhysDefs) const;
+                               SmallSet<unsigned,8> &PhysRefs) const;
     bool PhysRegDefsReach(MachineInstr *CSMI, MachineInstr *MI,
                           SmallSet<unsigned,8> &PhysRefs) const;
     bool isCSECandidate(MachineInstr *MI);
@@ -192,8 +189,7 @@
 /// instruction does not uses a physical register.
 bool MachineCSE::hasLivePhysRegDefUses(const MachineInstr *MI,
                                        const MachineBasicBlock *MBB,
-                                       SmallSet<unsigned,8> &PhysRefs,
-                                       SmallVector<unsigned,8> &PhysDefs) const{
+                                       SmallSet<unsigned,8> &PhysRefs) const {
   MachineBasicBlock::const_iterator I = MI; I = llvm::next(I);
   for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) {
     const MachineOperand &MO = MI->getOperand(i);
@@ -210,7 +206,6 @@
     if (MO.isDef() &&
         (MO.isDead() || isPhysDefTriviallyDead(Reg, I, MBB->end())))
       continue;
-    PhysDefs.push_back(Reg);
     PhysRefs.insert(Reg);
     for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias)
       PhysRefs.insert(*Alias);
@@ -221,43 +216,35 @@
 
 bool MachineCSE::PhysRegDefsReach(MachineInstr *CSMI, MachineInstr *MI,
                                   SmallSet<unsigned,8> &PhysRefs) const {
-  // Look backward from MI to find CSMI.
+  // For now conservatively returns false if the common subexpression is
+  // not in the same basic block as the given instruction.
+  MachineBasicBlock *MBB = MI->getParent();
+  if (CSMI->getParent() != MBB)
+    return false;
+  MachineBasicBlock::const_iterator I = CSMI; I = llvm::next(I);
+  MachineBasicBlock::const_iterator E = MI;
   unsigned LookAheadLeft = LookAheadLimit;
-  MachineBasicBlock *CurBB = MI->getParent();
-  MachineBasicBlock::const_reverse_iterator I(MI);
-  MachineBasicBlock::const_reverse_iterator E(CurBB->rend());
   while (LookAheadLeft) {
-    while (LookAheadLeft && I != E) {
-      // Skip over dbg_value's.
-      while (I != E && I->isDebugValue())
-        ++I;
-
-      if (I == E) break;
-
-      if (&*I == CSMI)
-        return true;
-
-      for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
-        const MachineOperand &MO = I->getOperand(i);
-        if (!MO.isReg() || !MO.isDef())
-          continue;
-        unsigned MOReg = MO.getReg();
-        if (TargetRegisterInfo::isVirtualRegister(MOReg))
-          continue;
-        if (PhysRefs.count(MOReg))
-          return false;
-      }
-
-      --LookAheadLeft;
+    // Skip over dbg_value's.
+    while (I != E && I->isDebugValue())
       ++I;
+
+    if (I == E)
+      return true;
+
+    for (unsigned i = 0, e = I->getNumOperands(); i != e; ++i) {
+      const MachineOperand &MO = I->getOperand(i);
+      if (!MO.isReg() || !MO.isDef())
+        continue;
+      unsigned MOReg = MO.getReg();
+      if (TargetRegisterInfo::isVirtualRegister(MOReg))
+        continue;
+      if (PhysRefs.count(MOReg))
+        return false;
     }
-    // Go back another BB; for now, only go back at most one BB.
-    MachineBasicBlock *CSBB = CSMI->getParent();
-    if (!CSBB->isSuccessor(CurBB) || CurBB->pred_size() != 1)
-      return false;
-    CurBB = CSBB;
-    I = CSBB->rbegin();
-    E = CSBB->rend();
+
+    --LookAheadLeft;
+    ++I;
   }
 
   return false;
@@ -408,8 +395,7 @@
     // used, then it's not safe to replace it with a common subexpression.
     // It's also not safe if the instruction uses physical registers.
     SmallSet<unsigned,8> PhysRefs;
-    SmallVector<unsigned,8> DirectPhysRefs;
-    if (FoundCSE && hasLivePhysRegDefUses(MI, MBB, PhysRefs, DirectPhysRefs)) {
+    if (FoundCSE && hasLivePhysRegDefUses(MI, MBB, PhysRefs)) {
       FoundCSE = false;
 
       // ... Unless the CS is local and it also defines the physical register
@@ -462,15 +448,6 @@
         MRI->clearKillFlags(CSEPairs[i].second);
       }
       MI->eraseFromParent();
-      if (!DirectPhysRefs.empty() && CSMI->getParent() != MBB) {
-        assert(CSMI->getParent()->isSuccessor(MBB));
-        ++NumCrossBlockPhysCSEs;
-        SmallVector<unsigned,8>::iterator PI = DirectPhysRefs.begin(),
-                                          PE = DirectPhysRefs.end();
-        for (; PI != PE; ++PI)
-          if (!MBB->isLiveIn(*PI))
-            MBB->addLiveIn(*PI);
-      }
       ++NumCSEs;
       if (!PhysRefs.empty())
         ++NumPhysCSEs;

Modified: llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll?rev=130977&r1=130976&r2=130977&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll (original)
+++ llvm/trunk/test/CodeGen/Thumb2/thumb2-cbnz.ll Fri May  6 00:23:07 2011
@@ -21,8 +21,8 @@
 
 bb9:                                              ; preds = %bb7
 ; CHECK:      cmp r0, #0
-; CHECK-NOT: cmp
-; CHECK: cbnz
+; CHECK:      cmp r0, #0
+; CHECK-NEXT: cbnz
   %0 = tail call  double @floor(double %b) nounwind readnone ; <double> [#uses=0]
   br label %bb11
 

Removed: llvm/trunk/test/CodeGen/X86/cmp-redundant.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/cmp-redundant.ll?rev=130976&view=auto
==============================================================================
--- llvm/trunk/test/CodeGen/X86/cmp-redundant.ll (original)
+++ llvm/trunk/test/CodeGen/X86/cmp-redundant.ll (removed)
@@ -1,43 +0,0 @@
-; RUN: llc < %s -mtriple=x86_64-apple-darwin10 | FileCheck %s
-
-define i32 @cmp(i32* %aa, i32* %bb) nounwind readnone ssp {
-entry:
-  %a = load i32* %aa
-  %b = load i32* %bb
-  %cmp = icmp sgt i32 %a, %b
-  br i1 %cmp, label %return, label %if.end
-; CHECK: cmp:
-; CHECK: cmpl
-; CHECK: jg
-
-if.end:                                           ; preds = %entry
-; CHECK-NOT: cmpl
-; CHECK: cmov
-  %cmp4 = icmp slt i32 %a, %b
-  %. = select i1 %cmp4, i32 2, i32 111
-  br label %return
-
-return:                                           ; preds = %if.end, %entry
-  %retval.0 = phi i32 [ 1, %entry ], [ %., %if.end ]
-  ret i32 %retval.0
-}
-
-define i32 @cmp2(i32 %a, i32 %b) nounwind readnone ssp {
-entry:
-  %cmp = icmp sgt i32 %a, %b
-  br i1 %cmp, label %return, label %if.end
-; CHECK: cmp2:
-; CHECK: cmpl
-; CHECK: jg
-
-if.end:                                           ; preds = %entry
-; CHECK-NOT: cmpl
-; CHECK: cmov
-  %cmp4 = icmp slt i32 %a, %b
-  %. = select i1 %cmp4, i32 2, i32 111
-  br label %return
-
-return:                                           ; preds = %if.end, %entry
-  %retval.0 = phi i32 [ 1, %entry ], [ %., %if.end ]
-  ret i32 %retval.0
-}



From baldrick at free.fr  Fri May  6 01:24:10 2011
From: baldrick at free.fr (Duncan Sands)
Date: Fri, 06 May 2011 08:24:10 +0200
Subject: [llvm-commits] [llvm] r130943
	-	/llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp
In-Reply-To: <20110505193221.B847D2A6C12C@llvm.org>
References: <20110505193221.B847D2A6C12C@llvm.org>
Message-ID: <4DC3940A.2070800@free.fr>

Hi Andrew,

> --- llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp (original)
> +++ llvm/trunk/lib/CodeGen/ScheduleDAGInstrs.cpp Thu May  5 14:32:21 2011
> @@ -393,6 +393,18 @@
>           UseList.clear();
>           if (!MO.isDead())
>             DefList.clear();
> +
> +        // Calls will not be reordered because of chain dependencies (see
> +        // below). Since call operands are dead, calls may continue to be added
> +        // to the DefList making dependence checking quadratic in the size of
> +        // the block. Instead, we leave only one call at the back of the
> +        // DefList.
> +        //
> +        // NOTE: This assumes that the DefList is ordered!

that suggests adding a check that DefList is ordered.

Ciao, Duncan.

From baldrick at free.fr  Fri May  6 05:30:37 2011
From: baldrick at free.fr (Duncan Sands)
Date: Fri, 06 May 2011 10:30:37 -0000
Subject: [llvm-commits] [llvm] r130979 - in /llvm/trunk:
 lib/Transforms/Scalar/ScalarReplAggregates.cpp
 test/Transforms/ScalarRepl/2011-05-06-CapturedAlloca.ll
Message-ID: <20110506103037.EF3D02A6C12E@llvm.org>

Author: baldrick
Date: Fri May  6 05:30:37 2011
New Revision: 130979

URL: http://llvm.org/viewvc/llvm-project?rev=130979&view=rev
Log:
Fix PR9820: a read-only call differs from a load in that a load doesn't
return the pointer being dereferenced, it returns the pointee, but a call
might return the pointer itself.

Added:
    llvm/trunk/test/Transforms/ScalarRepl/2011-05-06-CapturedAlloca.ll
Modified:
    llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp

Modified: llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp?rev=130979&r1=130978&r2=130979&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp (original)
+++ llvm/trunk/lib/Transforms/Scalar/ScalarReplAggregates.cpp Fri May  6 05:30:37 2011
@@ -2481,19 +2481,22 @@
     }
 
     if (CallSite CS = U) {
-      // If this is a readonly/readnone call site, then we know it is just a
-      // load and we can ignore it.
-      if (CS.onlyReadsMemory())
-        continue;
-
       // If this is the function being called then we treat it like a load and
       // ignore it.
       if (CS.isCallee(UI))
         continue;
 
+      // If this is a readonly/readnone call site, then we know it is just a
+      // load (but one that potentially returns the value itself), so we can
+      // ignore it if we know that the value isn't captured.
+      unsigned ArgNo = CS.getArgumentNo(UI);
+      if (CS.onlyReadsMemory() &&
+          (CS.getInstruction()->use_empty() ||
+           CS.paramHasAttr(ArgNo+1, Attribute::NoCapture)))
+        continue;
+
       // If this is being passed as a byval argument, the caller is making a
       // copy, so it is only a read of the alloca.
-      unsigned ArgNo = CS.getArgumentNo(UI);
       if (CS.paramHasAttr(ArgNo+1, Attribute::ByVal))
         continue;
     }

Added: llvm/trunk/test/Transforms/ScalarRepl/2011-05-06-CapturedAlloca.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/ScalarRepl/2011-05-06-CapturedAlloca.ll?rev=130979&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/ScalarRepl/2011-05-06-CapturedAlloca.ll (added)
+++ llvm/trunk/test/Transforms/ScalarRepl/2011-05-06-CapturedAlloca.ll Fri May  6 05:30:37 2011
@@ -0,0 +1,26 @@
+; RUN: opt < %s -instcombine -S | FileCheck %s
+; PR9820
+
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-unknown-linux-gnu"
+
+ at func_1.l_10 = internal unnamed_addr constant [4 x i32] [i32 1, i32 0, i32 0, i32 0], align 16
+
+define i32* @noop(i32* %p_29) nounwind readnone {
+entry:
+  ret i32* %p_29
+}
+
+define i32 @main() nounwind {
+entry:
+  %l_10 = alloca [4 x i32], align 16
+  %tmp = bitcast [4 x i32]* %l_10 to i8*
+  call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* bitcast ([4 x i32]* @func_1.l_10 to i8*), i64 16, i32 16, i1 false)
+; CHECK: call void @llvm.memcpy
+  %arrayidx = getelementptr inbounds [4 x i32]* %l_10, i64 0, i64 0
+  %call = call i32* @noop(i32* %arrayidx)
+  store i32 0, i32* %call
+  ret i32 0
+}
+
+declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind



From justin.holewinski at gmail.com  Fri May  6 06:40:36 2011
From: justin.holewinski at gmail.com (Justin Holewinski)
Date: Fri, 06 May 2011 11:40:36 -0000
Subject: [llvm-commits] [llvm] r130980 - in /llvm/trunk:
 lib/Target/PTX/PTX.td lib/Target/PTX/PTXSubtarget.cpp
 lib/Target/PTX/PTXSubtarget.h test/CodeGen/PTX/options.ll
Message-ID: <20110506114036.3F86A2A6C12C@llvm.org>

Author: jholewinski
Date: Fri May  6 06:40:36 2011
New Revision: 130980

URL: http://llvm.org/viewvc/llvm-project?rev=130980&view=rev
Log:
PTX:  add PTX 2.3 language target

Patch by Wei-Ren Chen

Modified:
    llvm/trunk/lib/Target/PTX/PTX.td
    llvm/trunk/lib/Target/PTX/PTXSubtarget.cpp
    llvm/trunk/lib/Target/PTX/PTXSubtarget.h
    llvm/trunk/test/CodeGen/PTX/options.ll

Modified: llvm/trunk/lib/Target/PTX/PTX.td
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PTX/PTX.td?rev=130980&r1=130979&r2=130980&view=diff
==============================================================================
--- llvm/trunk/lib/Target/PTX/PTX.td (original)
+++ llvm/trunk/lib/Target/PTX/PTX.td Fri May  6 06:40:36 2011
@@ -38,6 +38,10 @@
                                     "Use PTX Language Version 2.2",
                                     [FeaturePTX21]>;
 
+def FeaturePTX23 : SubtargetFeature<"ptx23", "PTXVersion", "PTX_VERSION_2_3",
+                                    "Use PTX Language Version 2.3",
+                                    [FeaturePTX22]>;
+
 //===- PTX Shader Model ---------------------------------------------------===//
 
 def FeatureSM10 : SubtargetFeature<"sm10", "PTXShaderModel", "PTX_SM_1_0",

Modified: llvm/trunk/lib/Target/PTX/PTXSubtarget.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PTX/PTXSubtarget.cpp?rev=130980&r1=130979&r2=130980&view=diff
==============================================================================
--- llvm/trunk/lib/Target/PTX/PTXSubtarget.cpp (original)
+++ llvm/trunk/lib/Target/PTX/PTXSubtarget.cpp Fri May  6 06:40:36 2011
@@ -41,6 +41,7 @@
     case PTX_VERSION_2_0: return "2.0";
     case PTX_VERSION_2_1: return "2.1";
     case PTX_VERSION_2_2: return "2.2";
+    case PTX_VERSION_2_3: return "2.3";
   }
 }
 

Modified: llvm/trunk/lib/Target/PTX/PTXSubtarget.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/PTX/PTXSubtarget.h?rev=130980&r1=130979&r2=130980&view=diff
==============================================================================
--- llvm/trunk/lib/Target/PTX/PTXSubtarget.h (original)
+++ llvm/trunk/lib/Target/PTX/PTXSubtarget.h Fri May  6 06:40:36 2011
@@ -37,7 +37,8 @@
       enum PTXVersionEnum {
         PTX_VERSION_2_0,  /*< PTX Version 2.0 */
         PTX_VERSION_2_1,  /*< PTX Version 2.1 */
-        PTX_VERSION_2_2   /*< PTX Version 2.2 */
+        PTX_VERSION_2_2,  /*< PTX Version 2.2 */
+        PTX_VERSION_2_3   /*< PTX Version 2.3 */
       };
 
       /// Shader Model supported on the target GPU.

Modified: llvm/trunk/test/CodeGen/PTX/options.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/PTX/options.ll?rev=130980&r1=130979&r2=130980&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/PTX/options.ll (original)
+++ llvm/trunk/test/CodeGen/PTX/options.ll Fri May  6 06:40:36 2011
@@ -1,6 +1,7 @@
 ; RUN: llc < %s -march=ptx32 -mattr=ptx20 | grep ".version 2.0"
 ; RUN: llc < %s -march=ptx32 -mattr=ptx21 | grep ".version 2.1"
 ; RUN: llc < %s -march=ptx32 -mattr=ptx22 | grep ".version 2.2"
+; RUN: llc < %s -march=ptx32 -mattr=ptx23 | grep ".version 2.3"
 ; RUN: llc < %s -march=ptx32 -mattr=sm10 | grep ".target sm_10"
 ; RUN: llc < %s -march=ptx32 -mattr=sm13 | grep ".target sm_13"
 ; RUN: llc < %s -march=ptx32 -mattr=sm20 | grep ".target sm_20"



From justin.holewinski at gmail.com  Fri May  6 06:46:48 2011
From: justin.holewinski at gmail.com (Justin Holewinski)
Date: Fri, 6 May 2011 07:46:48 -0400
Subject: [llvm-commits] [LLVMdev] [PATCH ]Add Subtarget ptx23
In-Reply-To: <20110506093714.GA41682@cs.nctu.edu.tw>
References: <20110506093714.GA41682@cs.nctu.edu.tw>
Message-ID: <BANLkTikm+X52m8Pvo9R6bfb6Yj5Qx4Bjcg@mail.gmail.com>

On Fri, May 6, 2011 at 5:37 AM, ??? <chenwj at iis.sinica.edu.tw> wrote:

> Hi, Justin
>
>  PTX version in CUDA 4.0 has changed from 2.2 to 2.3.
> I add ptx23 subtarget and update a testcase.
>
>  Is that O.K.?
>

Looks good.  Committed in r130980.  Thanks!

In the future, patches should be sent to llvm-commits instead of llvmdev.


>
> Regards,
> chenwj
>
> --
> Wei-Ren Chen (???)
> Computer Systems Lab, Institute of Information Science,
> Academia Sinica, Taiwan (R.O.C.)
> Tel:886-2-2788-3799 #1667
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
>


-- 

Thanks,

Justin Holewinski
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cs.uiuc.edu/pipermail/llvm-commits/attachments/20110506/2d0a992c/attachment.html 

From rafael.espindola at gmail.com  Fri May  6 09:56:22 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Fri, 06 May 2011 14:56:22 -0000
Subject: [llvm-commits] [llvm] r130984 - in /llvm/trunk:
 include/llvm/MC/MCAsmInfo.h lib/CodeGen/AsmPrinter/AsmPrinter.cpp
 lib/CodeGen/AsmPrinter/DwarfDebug.cpp lib/CodeGen/AsmPrinter/DwarfDebug.h
 lib/MC/MCAsmInfo.cpp lib/Target/ARM/ARMMCAsmInfo.cpp
 lib/Target/X86/X86MCAsmInfo.cpp
Message-ID: <20110506145623.0C0792A6C12C@llvm.org>

Author: rafael
Date: Fri May  6 09:56:22 2011
New Revision: 130984

URL: http://llvm.org/viewvc/llvm-project?rev=130984&view=rev
Log:
Dead code elimination.

Modified:
    llvm/trunk/include/llvm/MC/MCAsmInfo.h
    llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.h
    llvm/trunk/lib/MC/MCAsmInfo.cpp
    llvm/trunk/lib/Target/ARM/ARMMCAsmInfo.cpp
    llvm/trunk/lib/Target/X86/X86MCAsmInfo.cpp

Modified: llvm/trunk/include/llvm/MC/MCAsmInfo.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/MC/MCAsmInfo.h?rev=130984&r1=130983&r2=130984&view=diff
==============================================================================
--- llvm/trunk/include/llvm/MC/MCAsmInfo.h (original)
+++ llvm/trunk/include/llvm/MC/MCAsmInfo.h Fri May  6 09:56:22 2011
@@ -269,9 +269,6 @@
     /// SupportsExceptionHandling - True if target supports exception handling.
     ExceptionHandling::ExceptionsType ExceptionsType; // Defaults to None
 
-    /// RequiresFrameSection - true if the Dwarf2 output needs a frame section
-    bool DwarfRequiresFrameSection;          // Defaults to true.
-
     /// DwarfUsesInlineInfoSection - True if DwarfDebugInlineSection is used to
     /// encode inline subroutine information.
     bool DwarfUsesInlineInfoSection;         // Defaults to false.
@@ -465,10 +462,6 @@
         (ExceptionsType == ExceptionHandling::DwarfCFI ||
          ExceptionsType == ExceptionHandling::ARM);
     }
-
-    bool doesDwarfRequireFrameSection() const {
-      return DwarfRequiresFrameSection;
-    }
     bool doesDwarfUsesInlineInfoSection() const {
       return DwarfUsesInlineInfoSection;
     }

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp?rev=130984&r1=130983&r2=130984&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp Fri May  6 09:56:22 2011
@@ -596,7 +596,7 @@
   if (UnwindTablesMandatory)
     return true;
 
-  if (MMI->hasDebugInfo() && !MAI->doesDwarfRequireFrameSection())
+  if (MMI->hasDebugInfo())
     return true;
 
   if (MF->getFunction()->doesNotThrow())
@@ -608,8 +608,7 @@
 void AsmPrinter::emitPrologLabel(const MachineInstr &MI) {
   MCSymbol *Label = MI.getOperand(0).getMCSymbol();
 
-  if (MAI->doesDwarfRequireFrameSection() ||
-      MAI->getExceptionHandlingType() != ExceptionHandling::DwarfCFI)
+  if (MAI->getExceptionHandlingType() != ExceptionHandling::DwarfCFI)
     OutStreamer.EmitLabel(Label);
 
   if (MAI->getExceptionHandlingType() != ExceptionHandling::DwarfCFI)

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130984&r1=130983&r2=130984&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Fri May  6 09:56:22 2011
@@ -256,7 +256,7 @@
     CurrentFnDbgScope(0), PrevLabel(NULL) {
   NextStringPoolNumber = 0;
 
-  DwarfFrameSectionSym = DwarfInfoSectionSym = DwarfAbbrevSectionSym = 0;
+  DwarfInfoSectionSym = DwarfAbbrevSectionSym = 0;
   DwarfStrSectionSym = TextSectionSym = 0;
   DwarfDebugRangeSectionSym = DwarfDebugLocSectionSym = 0;
   FunctionBeginSym = FunctionEndSym = 0;
@@ -1261,14 +1261,6 @@
     Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("section_end", i));
   }
 
-  // Emit common frame information.
-  emitCommonDebugFrame();
-
-  // Emit function debug frame information
-  for (std::vector<FunctionDebugFrameInfo>::iterator I = DebugFrames.begin(),
-         E = DebugFrames.end(); I != E; ++I)
-    emitFunctionDebugFrame(*I);
-
   // Compute DIE offsets and sizes.
   computeSizeAndOffsets();
 
@@ -2261,11 +2253,6 @@
   const TargetLoweringObjectFile &TLOF = Asm->getObjFileLowering();
 
   // Dwarf sections base addresses.
-  if (Asm->MAI->doesDwarfRequireFrameSection()) {
-    DwarfFrameSectionSym =
-      EmitSectionSym(Asm, TLOF.getDwarfFrameSection(), "section_debug_frame");
-   }
-
   DwarfInfoSectionSym =
     EmitSectionSym(Asm, TLOF.getDwarfInfoSection(), "section_info");
   DwarfAbbrevSectionSym =
@@ -2473,91 +2460,6 @@
   Asm->EmitInt8(1);
 }
 
-/// emitCommonDebugFrame - Emit common frame info into a debug frame section.
-///
-void DwarfDebug::emitCommonDebugFrame() {
-  if (!Asm->MAI->doesDwarfRequireFrameSection())
-    return;
-
-  int stackGrowth = Asm->getTargetData().getPointerSize();
-  if (Asm->TM.getFrameLowering()->getStackGrowthDirection() ==
-      TargetFrameLowering::StackGrowsDown)
-    stackGrowth *= -1;
-
-  // Start the dwarf frame section.
-  Asm->OutStreamer.SwitchSection(
-                              Asm->getObjFileLowering().getDwarfFrameSection());
-
-  Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("debug_frame_common"));
-  Asm->OutStreamer.AddComment("Length of Common Information Entry");
-  Asm->EmitLabelDifference(Asm->GetTempSymbol("debug_frame_common_end"),
-                           Asm->GetTempSymbol("debug_frame_common_begin"), 4);
-
-  Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("debug_frame_common_begin"));
-  Asm->OutStreamer.AddComment("CIE Identifier Tag");
-  Asm->EmitInt32((int)dwarf::DW_CIE_ID);
-  Asm->OutStreamer.AddComment("CIE Version");
-  Asm->EmitInt8(dwarf::DW_CIE_VERSION);
-  Asm->OutStreamer.AddComment("CIE Augmentation");
-  Asm->OutStreamer.EmitIntValue(0, 1, /*addrspace*/0); // nul terminator.
-  Asm->EmitULEB128(1, "CIE Code Alignment Factor");
-  Asm->EmitSLEB128(stackGrowth, "CIE Data Alignment Factor");
-  Asm->OutStreamer.AddComment("CIE RA Column");
-  const TargetRegisterInfo *RI = Asm->TM.getRegisterInfo();
-  const TargetFrameLowering *TFI = Asm->TM.getFrameLowering();
-  Asm->EmitInt8(RI->getDwarfRegNum(RI->getRARegister(), false));
-
-  std::vector<MachineMove> Moves;
-  TFI->getInitialFrameState(Moves);
-
-  Asm->EmitFrameMoves(Moves, 0, false);
-
-  Asm->EmitAlignment(2);
-  Asm->OutStreamer.EmitLabel(Asm->GetTempSymbol("debug_frame_common_end"));
-}
-
-/// emitFunctionDebugFrame - Emit per function frame info into a debug frame
-/// section.
-void DwarfDebug::
-emitFunctionDebugFrame(const FunctionDebugFrameInfo &DebugFrameInfo) {
-  if (!Asm->MAI->doesDwarfRequireFrameSection())
-    return;
-
-  // Start the dwarf frame section.
-  Asm->OutStreamer.SwitchSection(
-                              Asm->getObjFileLowering().getDwarfFrameSection());
-
-  Asm->OutStreamer.AddComment("Length of Frame Information Entry");
-  MCSymbol *DebugFrameBegin =
-    Asm->GetTempSymbol("debug_frame_begin", DebugFrameInfo.Number);
-  MCSymbol *DebugFrameEnd =
-    Asm->GetTempSymbol("debug_frame_end", DebugFrameInfo.Number);
-  Asm->EmitLabelDifference(DebugFrameEnd, DebugFrameBegin, 4);
-
-  Asm->OutStreamer.EmitLabel(DebugFrameBegin);
-
-  Asm->OutStreamer.AddComment("FDE CIE offset");
-  Asm->EmitSectionOffset(Asm->GetTempSymbol("debug_frame_common"),
-                         DwarfFrameSectionSym);
-
-  Asm->OutStreamer.AddComment("FDE initial location");
-  MCSymbol *FuncBeginSym =
-    Asm->GetTempSymbol("func_begin", DebugFrameInfo.Number);
-  Asm->OutStreamer.EmitSymbolValue(FuncBeginSym,
-                                   Asm->getTargetData().getPointerSize(),
-                                   0/*AddrSpace*/);
-
-
-  Asm->OutStreamer.AddComment("FDE address range");
-  Asm->EmitLabelDifference(Asm->GetTempSymbol("func_end",DebugFrameInfo.Number),
-                           FuncBeginSym, Asm->getTargetData().getPointerSize());
-
-  Asm->EmitFrameMoves(DebugFrameInfo.Moves, FuncBeginSym, false);
-
-  Asm->EmitAlignment(2);
-  Asm->OutStreamer.EmitLabel(DebugFrameEnd);
-}
-
 /// emitDebugPubNames - Emit visible names into a debug pubnames section.
 ///
 void DwarfDebug::emitDebugPubNames() {

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.h?rev=130984&r1=130983&r2=130984&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.h (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.h Fri May  6 09:56:22 2011
@@ -269,7 +269,7 @@
   // Section Symbols: these are assembler temporary labels that are emitted at
   // the beginning of each supported dwarf section.  These are used to form
   // section offsets and are created by EmitSectionLabels.
-  MCSymbol *DwarfFrameSectionSym, *DwarfInfoSectionSym, *DwarfAbbrevSectionSym;
+  MCSymbol *DwarfInfoSectionSym, *DwarfAbbrevSectionSym;
   MCSymbol *DwarfStrSectionSym, *TextSectionSym, *DwarfDebugRangeSectionSym;
   MCSymbol *DwarfDebugLocSectionSym;
   MCSymbol *FunctionBeginSym, *FunctionEndSym;
@@ -338,14 +338,6 @@
   ///
   void emitEndOfLineMatrix(unsigned SectionEnd);
 
-  /// emitCommonDebugFrame - Emit common frame info into a debug frame section.
-  ///
-  void emitCommonDebugFrame();
-
-  /// emitFunctionDebugFrame - Emit per function frame info into a debug frame
-  /// section.
-  void emitFunctionDebugFrame(const FunctionDebugFrameInfo &DebugFrameInfo);
-
   /// emitDebugPubNames - Emit visible names into a debug pubnames section.
   ///
   void emitDebugPubNames();

Modified: llvm/trunk/lib/MC/MCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/MC/MCAsmInfo.cpp?rev=130984&r1=130983&r2=130984&view=diff
==============================================================================
--- llvm/trunk/lib/MC/MCAsmInfo.cpp (original)
+++ llvm/trunk/lib/MC/MCAsmInfo.cpp Fri May  6 09:56:22 2011
@@ -74,7 +74,6 @@
   HasLEB128 = false;
   SupportsDebugInformation = false;
   ExceptionsType = ExceptionHandling::None;
-  DwarfRequiresFrameSection = false;
   DwarfUsesInlineInfoSection = false;
   DwarfRequiresRelocationForStmtList = true;
   DwarfSectionOffsetDirective = 0;

Modified: llvm/trunk/lib/Target/ARM/ARMMCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/ARM/ARMMCAsmInfo.cpp?rev=130984&r1=130983&r2=130984&view=diff
==============================================================================
--- llvm/trunk/lib/Target/ARM/ARMMCAsmInfo.cpp (original)
+++ llvm/trunk/lib/Target/ARM/ARMMCAsmInfo.cpp Fri May  6 09:56:22 2011
@@ -70,8 +70,6 @@
   WeakRefDirective = "\t.weak\t";
   HasLCOMMDirective = true;
 
-  DwarfRequiresFrameSection = false;
-
   SupportsDebugInformation = true;
 
   // Exceptions handling

Modified: llvm/trunk/lib/Target/X86/X86MCAsmInfo.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/X86/X86MCAsmInfo.cpp?rev=130984&r1=130983&r2=130984&view=diff
==============================================================================
--- llvm/trunk/lib/Target/X86/X86MCAsmInfo.cpp (original)
+++ llvm/trunk/lib/Target/X86/X86MCAsmInfo.cpp Fri May  6 09:56:22 2011
@@ -108,8 +108,6 @@
   // Exceptions handling
   ExceptionsType = ExceptionHandling::DwarfCFI;
 
-  DwarfRequiresFrameSection = false;
-
   // OpenBSD has buggy support for .quad in 32-bit mode, just split into two
   // .words.
   if (T.getOS() == Triple::OpenBSD && T.getArch() == Triple::x86)



From rafael.espindola at gmail.com  Fri May  6 10:22:26 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Fri, 06 May 2011 15:22:26 -0000
Subject: [llvm-commits] [llvm] r130985 - in /llvm/trunk:
 include/llvm/CodeGen/AsmPrinter.h
 lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp
Message-ID: <20110506152226.BAA132A6C12C@llvm.org>

Author: rafael
Date: Fri May  6 10:22:26 2011
New Revision: 130985

URL: http://llvm.org/viewvc/llvm-project?rev=130985&view=rev
Log:
More dead code elimination.

Modified:
    llvm/trunk/include/llvm/CodeGen/AsmPrinter.h
    llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp

Modified: llvm/trunk/include/llvm/CodeGen/AsmPrinter.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/AsmPrinter.h?rev=130985&r1=130984&r2=130985&view=diff
==============================================================================
--- llvm/trunk/include/llvm/CodeGen/AsmPrinter.h (original)
+++ llvm/trunk/include/llvm/CodeGen/AsmPrinter.h Fri May  6 10:22:26 2011
@@ -398,8 +398,6 @@
 
     /// EmitFrameMoves - Emit frame instructions to describe the layout of the
     /// frame.
-    void EmitFrameMoves(const std::vector<MachineMove> &Moves,
-                        MCSymbol *BaseLabel, bool isEH) const;
     void EmitCFIFrameMove(const MachineMove &Move) const;
     void EmitCFIFrameMoves(const std::vector<MachineMove> &Moves) const;
 

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp?rev=130985&r1=130984&r2=130985&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp Fri May  6 10:22:26 2011
@@ -206,77 +206,6 @@
 // Dwarf Lowering Routines
 //===----------------------------------------------------------------------===//
 
-
-/// EmitFrameMoves - Emit frame instructions to describe the layout of the
-/// frame.
-void AsmPrinter::EmitFrameMoves(const std::vector<MachineMove> &Moves,
-                                MCSymbol *BaseLabel, bool isEH) const {
-  const TargetRegisterInfo *RI = TM.getRegisterInfo();
-  
-  int stackGrowth = TM.getTargetData()->getPointerSize();
-  if (TM.getFrameLowering()->getStackGrowthDirection() !=
-      TargetFrameLowering::StackGrowsUp)
-    stackGrowth *= -1;
-  
-  for (unsigned i = 0, N = Moves.size(); i < N; ++i) {
-    const MachineMove &Move = Moves[i];
-    MCSymbol *Label = Move.getLabel();
-    // Throw out move if the label is invalid.
-    if (Label && !Label->isDefined()) continue; // Not emitted, in dead code.
-    
-    const MachineLocation &Dst = Move.getDestination();
-    const MachineLocation &Src = Move.getSource();
-    
-    // Advance row if new location.
-    if (BaseLabel && Label) {
-      MCSymbol *ThisSym = Label;
-      if (ThisSym != BaseLabel) {
-        EmitCFAByte(dwarf::DW_CFA_advance_loc4);
-        EmitLabelDifference(ThisSym, BaseLabel, 4);
-        BaseLabel = ThisSym;
-      }
-    }
-    
-    // If advancing cfa.
-    if (Dst.isReg() && Dst.getReg() == MachineLocation::VirtualFP) {
-      assert(!Src.isReg() && "Machine move not supported yet.");
-      
-      if (Src.getReg() == MachineLocation::VirtualFP) {
-        EmitCFAByte(dwarf::DW_CFA_def_cfa_offset);
-      } else {
-        EmitCFAByte(dwarf::DW_CFA_def_cfa);
-        EmitULEB128(RI->getDwarfRegNum(Src.getReg(), isEH), "Register");
-      }
-      
-      EmitULEB128(-Src.getOffset(), "Offset");
-      continue;
-    }
-    
-    if (Src.isReg() && Src.getReg() == MachineLocation::VirtualFP) {
-      assert(Dst.isReg() && "Machine move not supported yet.");
-      EmitCFAByte(dwarf::DW_CFA_def_cfa_register);
-      EmitULEB128(RI->getDwarfRegNum(Dst.getReg(), isEH), "Register");
-      continue;
-    }
-    
-    unsigned Reg = RI->getDwarfRegNum(Src.getReg(), isEH);
-    int Offset = Dst.getOffset() / stackGrowth;
-    
-    if (Offset < 0) {
-      EmitCFAByte(dwarf::DW_CFA_offset_extended_sf);
-      EmitULEB128(Reg, "Reg");
-      EmitSLEB128(Offset, "Offset");
-    } else if (Reg < 64) {
-      EmitCFAByte(dwarf::DW_CFA_offset + Reg);
-      EmitULEB128(Offset, "Offset");
-    } else {
-      EmitCFAByte(dwarf::DW_CFA_offset_extended);
-      EmitULEB128(Reg, "Reg");
-      EmitULEB128(Offset, "Offset");
-    }
-  }
-}
-
 /// EmitFrameMoves - Emit a frame instruction.
 void AsmPrinter::EmitCFIFrameMove(const MachineMove &Move) const {
   const TargetRegisterInfo *RI = TM.getRegisterInfo();



From rafael.espindola at gmail.com  Fri May  6 10:28:56 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Fri, 06 May 2011 15:28:56 -0000
Subject: [llvm-commits] [llvm] r130987 - in /llvm/trunk:
 include/llvm/CodeGen/AsmPrinter.h
 lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp
Message-ID: <20110506152856.875362A6C12C@llvm.org>

Author: rafael
Date: Fri May  6 10:28:56 2011
New Revision: 130987

URL: http://llvm.org/viewvc/llvm-project?rev=130987&view=rev
Log:
Update comments.

Modified:
    llvm/trunk/include/llvm/CodeGen/AsmPrinter.h
    llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp

Modified: llvm/trunk/include/llvm/CodeGen/AsmPrinter.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/AsmPrinter.h?rev=130987&r1=130986&r2=130987&view=diff
==============================================================================
--- llvm/trunk/include/llvm/CodeGen/AsmPrinter.h (original)
+++ llvm/trunk/include/llvm/CodeGen/AsmPrinter.h Fri May  6 10:28:56 2011
@@ -396,9 +396,12 @@
     // Dwarf Lowering Routines
     //===------------------------------------------------------------------===//
 
-    /// EmitFrameMoves - Emit frame instructions to describe the layout of the
+    /// EmitCFIFrameMove - Emit frame instruction to describe the layout of the
     /// frame.
     void EmitCFIFrameMove(const MachineMove &Move) const;
+
+    /// EmitCFIFrameMoves - Emit frame instructions to describe the layout of
+    /// the frame.
     void EmitCFIFrameMoves(const std::vector<MachineMove> &Moves) const;
 
     //===------------------------------------------------------------------===//

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp?rev=130987&r1=130986&r2=130987&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp Fri May  6 10:28:56 2011
@@ -206,7 +206,7 @@
 // Dwarf Lowering Routines
 //===----------------------------------------------------------------------===//
 
-/// EmitFrameMoves - Emit a frame instruction.
+/// EmitCFIFrameMove - Emit a frame instruction.
 void AsmPrinter::EmitCFIFrameMove(const MachineMove &Move) const {
   const TargetRegisterInfo *RI = TM.getRegisterInfo();
 
@@ -233,7 +233,7 @@
   }
 }
 
-/// EmitFrameMoves - Emit frame instructions to describe the layout of the
+/// EmitCFIFrameMoves - Emit frame instructions to describe the layout of the
 /// frame.
 void AsmPrinter::EmitCFIFrameMoves(const std::vector<MachineMove> &Moves) const {
   for (unsigned i = 0, N = Moves.size(); i < N; ++i) {



From rafael.espindola at gmail.com  Fri May  6 10:31:55 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Fri, 06 May 2011 15:31:55 -0000
Subject: [llvm-commits] [llvm] r130988 - in /llvm/trunk:
 include/llvm/CodeGen/AsmPrinter.h
 lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp
Message-ID: <20110506153155.847D32A6C12C@llvm.org>

Author: rafael
Date: Fri May  6 10:31:55 2011
New Revision: 130988

URL: http://llvm.org/viewvc/llvm-project?rev=130988&view=rev
Log:
Yet more dead code.

Modified:
    llvm/trunk/include/llvm/CodeGen/AsmPrinter.h
    llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp

Modified: llvm/trunk/include/llvm/CodeGen/AsmPrinter.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/CodeGen/AsmPrinter.h?rev=130988&r1=130987&r2=130988&view=diff
==============================================================================
--- llvm/trunk/include/llvm/CodeGen/AsmPrinter.h (original)
+++ llvm/trunk/include/llvm/CodeGen/AsmPrinter.h Fri May  6 10:31:55 2011
@@ -400,10 +400,6 @@
     /// frame.
     void EmitCFIFrameMove(const MachineMove &Move) const;
 
-    /// EmitCFIFrameMoves - Emit frame instructions to describe the layout of
-    /// the frame.
-    void EmitCFIFrameMoves(const std::vector<MachineMove> &Moves) const;
-
     //===------------------------------------------------------------------===//
     // Inline Asm Support
     //===------------------------------------------------------------------===//

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp?rev=130988&r1=130987&r2=130988&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinterDwarf.cpp Fri May  6 10:31:55 2011
@@ -232,16 +232,3 @@
                               Dst.getOffset());
   }
 }
-
-/// EmitCFIFrameMoves - Emit frame instructions to describe the layout of the
-/// frame.
-void AsmPrinter::EmitCFIFrameMoves(const std::vector<MachineMove> &Moves) const {
-  for (unsigned i = 0, N = Moves.size(); i < N; ++i) {
-    const MachineMove &Move = Moves[i];
-    MCSymbol *Label = Move.getLabel();
-    // Throw out move if the label is invalid.
-    if (Label && !Label->isDefined()) continue; // Not emitted, in dead code.
-
-    EmitCFIFrameMove(Move);
-  }
-}



From rafael.espindola at gmail.com  Fri May  6 10:44:29 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Fri, 06 May 2011 15:44:29 -0000
Subject: [llvm-commits] [llvm] r130989 -
	/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
Message-ID: <20110506154429.422FD2A6C12C@llvm.org>

Author: rafael
Date: Fri May  6 10:44:29 2011
New Revision: 130989

URL: http://llvm.org/viewvc/llvm-project?rev=130989&view=rev
Log:
Nothing else uses this label.

Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp?rev=130989&r1=130988&r2=130989&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/AsmPrinter.cpp Fri May  6 10:44:29 2011
@@ -609,9 +609,6 @@
   MCSymbol *Label = MI.getOperand(0).getMCSymbol();
 
   if (MAI->getExceptionHandlingType() != ExceptionHandling::DwarfCFI)
-    OutStreamer.EmitLabel(Label);
-
-  if (MAI->getExceptionHandlingType() != ExceptionHandling::DwarfCFI)
     return;
 
   if (!needsCFIMoves())



From dpatel at apple.com  Fri May  6 11:57:54 2011
From: dpatel at apple.com (Devang Patel)
Date: Fri, 06 May 2011 16:57:54 -0000
Subject: [llvm-commits] [llvm] r130991 - in
 /llvm/trunk/lib/CodeGen/AsmPrinter: DwarfCompileUnit.cpp DwarfDebug.cpp
Message-ID: <20110506165754.60A322A6C12C@llvm.org>

Author: dpatel
Date: Fri May  6 11:57:54 2011
New Revision: 130991

URL: http://llvm.org/viewvc/llvm-project?rev=130991&view=rev
Log:
Move CompileUnit::getOrCreateNameSpace() and CompileUnit::addPubType() from DwarfDebug.cpp to DwarfCompileUnit.cpp

Modified:
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp
    llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp?rev=130991&r1=130990&r2=130991&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfCompileUnit.cpp Fri May  6 11:57:54 2011
@@ -585,6 +585,27 @@
   Entity->addValue(dwarf::DW_AT_type, dwarf::DW_FORM_ref4, Entry);
 }
 
+/// addPubTypes - Add type for pubtypes section.
+void CompileUnit::addPubTypes(DISubprogram SP) {
+  DICompositeType SPTy = SP.getType();
+  unsigned SPTag = SPTy.getTag();
+  if (SPTag != dwarf::DW_TAG_subroutine_type)
+    return;
+
+  DIArray Args = SPTy.getTypeArray();
+  for (unsigned i = 0, e = Args.getNumElements(); i != e; ++i) {
+    DIType ATy(Args.getElement(i));
+    if (!ATy.Verify())
+      continue;
+    DICompositeType CATy = getDICompositeType(ATy);
+    if (DIDescriptor(CATy).Verify() && !CATy.getName().empty()
+        && !CATy.isForwardDecl()) {
+      if (DIEEntry *Entry = getDIEEntry(CATy))
+        addGlobalType(CATy.getName(), Entry->getEntry());
+    }
+  }
+}
+
 /// constructTypeDIE - Construct basic type die from DIBasicType.
 void CompileUnit::constructTypeDIE(DIE &Buffer, DIBasicType BTy) {
   // Get core information.
@@ -809,6 +830,20 @@
   return ParamDIE;
 }
 
+/// getOrCreateNameSpace - Create a DIE for DINameSpace.
+DIE *CompileUnit::getOrCreateNameSpace(DINameSpace NS) {
+  DIE *NDie = getDIE(NS);
+  if (NDie)
+    return NDie;
+  NDie = new DIE(dwarf::DW_TAG_namespace);
+  insertDIE(NS, NDie);
+  if (!NS.getName().empty())
+    addString(NDie, dwarf::DW_AT_name, dwarf::DW_FORM_string, NS.getName());
+  addSourceLine(NDie, NS);
+  addToContextOwner(NDie, NS.getContext());
+  return NDie;
+}
+
 /// constructSubrangeDIE - Construct subrange DIE from DISubrange.
 void CompileUnit::constructSubrangeDIE(DIE &Buffer, DISubrange SR, DIE *IndexTy){
   DIE *DW_Subrange = new DIE(dwarf::DW_TAG_subrange_type);

Modified: llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp?rev=130991&r1=130990&r2=130991&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp (original)
+++ llvm/trunk/lib/CodeGen/AsmPrinter/DwarfDebug.cpp Fri May  6 11:57:54 2011
@@ -748,26 +748,6 @@
 
 }
 
-void CompileUnit::addPubTypes(DISubprogram SP) {
-  DICompositeType SPTy = SP.getType();
-  unsigned SPTag = SPTy.getTag();
-  if (SPTag != dwarf::DW_TAG_subroutine_type)
-    return;
-
-  DIArray Args = SPTy.getTypeArray();
-  for (unsigned i = 0, e = Args.getNumElements(); i != e; ++i) {
-    DIType ATy(Args.getElement(i));
-    if (!ATy.Verify())
-      continue;
-    DICompositeType CATy = getDICompositeType(ATy);
-    if (DIDescriptor(CATy).Verify() && !CATy.getName().empty()
-        && !CATy.isForwardDecl()) {
-      if (DIEEntry *Entry = getDIEEntry(CATy))
-        addGlobalType(CATy.getName(), Entry->getEntry());
-    }
-  }
-}
-
 /// constructScopeDIE - Construct a DIE for this scope.
 DIE *DwarfDebug::constructScopeDIE(DbgScope *Scope) {
   if (!Scope || !Scope->getScopeNode())
@@ -860,20 +840,6 @@
   return SrcId;
 }
 
-/// getOrCreateNameSpace - Create a DIE for DINameSpace.
-DIE *CompileUnit::getOrCreateNameSpace(DINameSpace NS) {
-  DIE *NDie = getDIE(NS);
-  if (NDie)
-    return NDie;
-  NDie = new DIE(dwarf::DW_TAG_namespace);
-  insertDIE(NS, NDie);
-  if (!NS.getName().empty())
-    addString(NDie, dwarf::DW_AT_name, dwarf::DW_FORM_string, NS.getName());
-  addSourceLine(NDie, NS);
-  addToContextOwner(NDie, NS.getContext());
-  return NDie;
-}
-
 /// constructCompileUnit - Create new CompileUnit for the given
 /// metadata node with tag DW_TAG_compile_unit.
 void DwarfDebug::constructCompileUnit(const MDNode *N) {



From atrick at apple.com  Fri May  6 12:09:08 2011
From: atrick at apple.com (Andrew Trick)
Date: Fri, 06 May 2011 17:09:08 -0000
Subject: [llvm-commits] [llvm] r130992 -
	/llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp
Message-ID: <20110506170908.3920D2A6C12C@llvm.org>

Author: atrick
Date: Fri May  6 12:09:08 2011
New Revision: 130992

URL: http://llvm.org/viewvc/llvm-project?rev=130992&view=rev
Log:
Post-RA scheduler compile time fix. Quadratic computation of DAG node depth.

The post-ra scheduler was explicitly updating the depth of a node's
successors after scheduling it, regardless of whether the successor
was ready. This is quadratic for DAGs with transitively redundant
edges. I simply removed the useless update of depth, which is lazilly
computed later.
Fixes <rdar://problem/9044332> compiler takes way too long to build TextInput.

Modified:
    llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp

Modified: llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp?rev=130992&r1=130991&r2=130992&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp (original)
+++ llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp Fri May  6 12:09:08 2011
@@ -540,10 +540,16 @@
 #endif
   --SuccSU->NumPredsLeft;
 
-  // Compute how many cycles it will be before this actually becomes
-  // available.  This is the max of the start time of all predecessors plus
-  // their latencies.
-  SuccSU->setDepthToAtLeast(SU->getDepth() + SuccEdge->getLatency());
+  // Standard scheduler algorithms will recomute the depth of the successor
+  // here as such:
+  //   SuccSU->setDepthToAtLeast(SU->getDepth() + SuccEdge->getLatency());
+  //
+  // However, we lazily compute node depth instead. Note that
+  // ScheduleNodeTopDown has already updated the depth of this node which causes
+  // all descendents to be marked dirty. Setting the successor depth explicitly
+  // here would cause depth to be recomputed for all its ancestors. If the
+  // successor is not yet ready (because of a transitively redundant edge) then
+  // this causes depth computation to be quadratic in the size of the DAG.
 
   // If all the node's predecessors are scheduled, this node is ready
   // to be scheduled. Ignore the special ExitSU node.



From arplynn at gmail.com  Fri May  6 12:49:04 2011
From: arplynn at gmail.com (Alistair Lynn)
Date: Fri, 6 May 2011 18:49:04 +0100
Subject: [llvm-commits] [llvm] r130992 -
	/llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp
In-Reply-To: <20110506170908.3920D2A6C12C@llvm.org>
References: <20110506170908.3920D2A6C12C@llvm.org>
Message-ID: <B8F2C7EC-A5D6-40B4-8C28-66AE92459E18@gmail.com>

Hi Andrew-

> +  // Standard scheduler algorithms will recomute the depth of the successor
> +  // here as such:

? recompute?

Alistair



From rafael.espindola at gmail.com  Fri May  6 12:44:59 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Fri, 06 May 2011 17:44:59 -0000
Subject: [llvm-commits] [llvm] r130995 -
	/llvm/trunk/test/FrontendC/2010-01-05-LinkageName.c
Message-ID: <20110506174459.1D0F12A6C12C@llvm.org>

Author: rafael
Date: Fri May  6 12:44:58 2011
New Revision: 130995

URL: http://llvm.org/viewvc/llvm-project?rev=130995&view=rev
Log:
Pass -disable-cfi.

Modified:
    llvm/trunk/test/FrontendC/2010-01-05-LinkageName.c

Modified: llvm/trunk/test/FrontendC/2010-01-05-LinkageName.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC/2010-01-05-LinkageName.c?rev=130995&r1=130994&r2=130995&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC/2010-01-05-LinkageName.c (original)
+++ llvm/trunk/test/FrontendC/2010-01-05-LinkageName.c Fri May  6 12:44:58 2011
@@ -1,4 +1,4 @@
-// RUN: %llvmgcc -O2 -S -g %s -o - | llc -o 2010-01-05-LinkageName.s -O0 
+// RUN: %llvmgcc -O2 -S -g %s -o - | llc -disable-cfi -o 2010-01-05-LinkageName.s -O0 
 // RUN: %compile_c 2010-01-05-LinkageName.s -o 2010-01-05-LinkageName.s
 
 struct tm {};



From stoklund at 2pi.dk  Fri May  6 12:59:57 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Fri, 06 May 2011 17:59:57 -0000
Subject: [llvm-commits] [llvm] r130996 -
	/llvm/trunk/lib/CodeGen/VirtRegMap.cpp
Message-ID: <20110506175957.6E5AD2A6C12C@llvm.org>

Author: stoklund
Date: Fri May  6 12:59:57 2011
New Revision: 130996

URL: http://llvm.org/viewvc/llvm-project?rev=130996&view=rev
Log:
Also count identity copies.

Modified:
    llvm/trunk/lib/CodeGen/VirtRegMap.cpp

Modified: llvm/trunk/lib/CodeGen/VirtRegMap.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/VirtRegMap.cpp?rev=130996&r1=130995&r2=130996&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/VirtRegMap.cpp (original)
+++ llvm/trunk/lib/CodeGen/VirtRegMap.cpp Fri May  6 12:59:57 2011
@@ -42,6 +42,7 @@
 using namespace llvm;
 
 STATISTIC(NumSpills  , "Number of register spills");
+STATISTIC(NumIdCopies, "Number of identity moves eliminated after rewriting");
 
 //===----------------------------------------------------------------------===//
 //  VirtRegMap implementation
@@ -318,6 +319,7 @@
 
       // Finally, remove any identity copies.
       if (MI->isIdentityCopy()) {
+        ++NumIdCopies;
         if (MI->getNumOperands() == 2) {
           DEBUG(dbgs() << "Deleting identity copy.\n");
           RemoveMachineInstrFromMaps(MI);



From stoklund at 2pi.dk  Fri May  6 12:59:59 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Fri, 06 May 2011 17:59:59 -0000
Subject: [llvm-commits] [llvm] r130997 -
	/llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp
Message-ID: <20110506180000.0DCD42A6C12E@llvm.org>

Author: stoklund
Date: Fri May  6 12:59:59 2011
New Revision: 130997

URL: http://llvm.org/viewvc/llvm-project?rev=130997&view=rev
Log:
Use TargetMachine hooks to properly print debug variable locations.

Modified:
    llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp

Modified: llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp?rev=130997&r1=130996&r2=130997&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp (original)
+++ llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp Fri May  6 12:59:59 2011
@@ -228,7 +228,7 @@
   /// Only first one needs DebugLoc to identify variable's lexical scope
   /// in source file.
   DebugLoc findDebugLoc();
-  void print(raw_ostream&, const TargetRegisterInfo*);
+  void print(raw_ostream&, const TargetMachine*);
 };
 } // namespace
 
@@ -300,7 +300,7 @@
 };
 } // namespace
 
-void UserValue::print(raw_ostream &OS, const TargetRegisterInfo *TRI) {
+void UserValue::print(raw_ostream &OS, const TargetMachine *TM) {
   if (const MDString *MDS = dyn_cast<MDString>(variable->getOperand(2)))
     OS << "!\"" << MDS->getString() << "\"\t";
   if (offset)
@@ -312,15 +312,17 @@
     else
       OS << I.value();
   }
-  for (unsigned i = 0, e = locations.size(); i != e; ++i)
-    OS << " Loc" << i << '=' << locations[i];
+  for (unsigned i = 0, e = locations.size(); i != e; ++i) {
+    OS << " Loc" << i << '=';
+    locations[i].print(OS, TM);
+  }
   OS << '\n';
 }
 
 void LDVImpl::print(raw_ostream &OS) {
   OS << "********** DEBUG VARIABLES **********\n";
   for (unsigned i = 0, e = userValues.size(); i != e; ++i)
-    userValues[i]->print(OS, TRI);
+    userValues[i]->print(OS, &MF->getTarget());
 }
 
 void UserValue::coalesceLocation(unsigned LocNo) {
@@ -701,7 +703,6 @@
     }
     coalesceLocation(LocNo);
   }
-  DEBUG(print(dbgs(), &TRI));
 }
 
 /// findInsertLocation - Find an iterator for inserting a DBG_VALUE



From stoklund at 2pi.dk  Fri May  6 13:00:02 2011
From: stoklund at 2pi.dk (Jakob Stoklund Olesen)
Date: Fri, 06 May 2011 18:00:02 -0000
Subject: [llvm-commits] [llvm] r130998 - in /llvm/trunk/lib/CodeGen:
 LiveDebugVariables.cpp LiveDebugVariables.h LiveRangeEdit.h
 RegAllocGreedy.cpp
Message-ID: <20110506180003.0C90B2A6C12F@llvm.org>

Author: stoklund
Date: Fri May  6 13:00:02 2011
New Revision: 130998

URL: http://llvm.org/viewvc/llvm-project?rev=130998&view=rev
Log:
Update LiveDebugVariables after live range splitting.

After a virtual register is split, update any debug user variables that resided
in the old register. This ensures that the LiveDebugVariables are still correct
after register allocation.

This may create DBG_VALUE instructions that place a user variable in a register
in parts of the function and in a stack slot in other parts. DwarfDebug
currently doesn't support that.

Modified:
    llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp
    llvm/trunk/lib/CodeGen/LiveDebugVariables.h
    llvm/trunk/lib/CodeGen/LiveRangeEdit.h
    llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp

Modified: llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp?rev=130998&r1=130997&r2=130998&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp (original)
+++ llvm/trunk/lib/CodeGen/LiveDebugVariables.cpp Fri May  6 13:00:02 2011
@@ -101,9 +101,13 @@
   void insertDebugValue(MachineBasicBlock *MBB, SlotIndex Idx, unsigned LocNo,
                         LiveIntervals &LIS, const TargetInstrInfo &TII);
 
+  /// splitLocation - Replace OldLocNo ranges with NewRegs ranges where NewRegs
+  /// is live. Returns true if any changes were made.
+  bool splitLocation(unsigned OldLocNo, ArrayRef<LiveInterval*> NewRegs);
+
 public:
   /// UserValue - Create a new UserValue.
-  UserValue(const MDNode *var, unsigned o, DebugLoc L, 
+  UserValue(const MDNode *var, unsigned o, DebugLoc L,
             LocMap::Allocator &alloc)
     : variable(var), offset(o), dl(L), leader(this), next(0), locInts(alloc)
   {}
@@ -215,6 +219,10 @@
   void renameRegister(unsigned OldReg, unsigned NewReg, unsigned SubIdx,
                       const TargetRegisterInfo *TRI);
 
+  /// splitRegister - Replace OldReg ranges with NewRegs ranges where NewRegs is
+  /// live. Returns true if any changes were made.
+  bool splitRegister(unsigned OldLocNo, ArrayRef<LiveInterval*> NewRegs);
+
   /// rewriteLocations - Rewrite virtual register locations according to the
   /// provided virtual register map.
   void rewriteLocations(VirtRegMap &VRM, const TargetRegisterInfo &TRI);
@@ -293,6 +301,9 @@
   /// renameRegister - Replace all references to OldReg with NewReg:SubIdx.
   void renameRegister(unsigned OldReg, unsigned NewReg, unsigned SubIdx);
 
+  /// splitRegister -  Replace all references to OldReg with NewRegs.
+  void splitRegister(unsigned OldReg, ArrayRef<LiveInterval*> NewRegs);
+
   /// emitDebugVariables - Recreate DBG_VALUE instruction from data structures.
   void emitDebugValues(VirtRegMap *VRM);
 
@@ -679,6 +690,140 @@
     static_cast<LDVImpl*>(pImpl)->renameRegister(OldReg, NewReg, SubIdx);
 }
 
+//===----------------------------------------------------------------------===//
+//                           Live Range Splitting
+//===----------------------------------------------------------------------===//
+
+bool
+UserValue::splitLocation(unsigned OldLocNo, ArrayRef<LiveInterval*> NewRegs) {
+  DEBUG({
+    dbgs() << "Splitting Loc" << OldLocNo << '\t';
+    print(dbgs(), 0);
+  });
+  bool DidChange = false;
+  LocMap::iterator LocMapI;
+  LocMapI.setMap(locInts);
+  for (unsigned i = 0; i != NewRegs.size(); ++i) {
+    LiveInterval *LI = NewRegs[i];
+    if (LI->empty())
+      continue;
+
+    // Don't allocate the new LocNo until it is needed.
+    unsigned NewLocNo = ~0u;
+
+    // Iterate over the overlaps between locInts and LI.
+    LocMapI.find(LI->beginIndex());
+    if (!LocMapI.valid())
+      continue;
+    LiveInterval::iterator LII = LI->advanceTo(LI->begin(), LocMapI.start());
+    LiveInterval::iterator LIE = LI->end();
+    while (LocMapI.valid() && LII != LIE) {
+      // At this point, we know that LocMapI.stop() > LII->start.
+      LII = LI->advanceTo(LII, LocMapI.start());
+      if (LII == LIE)
+        break;
+
+      // Now LII->end > LocMapI.start(). Do we have an overlap?
+      if (LocMapI.value() == OldLocNo && LII->start < LocMapI.stop()) {
+        // Overlapping correct location. Allocate NewLocNo now.
+        if (NewLocNo == ~0u) {
+          MachineOperand MO = MachineOperand::CreateReg(LI->reg, false);
+          MO.setSubReg(locations[OldLocNo].getSubReg());
+          NewLocNo = getLocationNo(MO);
+          DidChange = true;
+        }
+
+        SlotIndex LStart = LocMapI.start();
+        SlotIndex LStop  = LocMapI.stop();
+
+        // Trim LocMapI down to the LII overlap.
+        if (LStart < LII->start)
+          LocMapI.setStartUnchecked(LII->start);
+        if (LStop > LII->end)
+          LocMapI.setStopUnchecked(LII->end);
+
+        // Change the value in the overlap. This may trigger coalescing.
+        LocMapI.setValue(NewLocNo);
+
+        // Re-insert any removed OldLocNo ranges.
+        if (LStart < LocMapI.start()) {
+          LocMapI.insert(LStart, LocMapI.start(), OldLocNo);
+          ++LocMapI;
+          assert(LocMapI.valid() && "Unexpected coalescing");
+        }
+        if (LStop > LocMapI.stop()) {
+          ++LocMapI;
+          LocMapI.insert(LII->end, LStop, OldLocNo);
+          --LocMapI;
+        }
+      }
+
+      // Advance to the next overlap.
+      if (LII->end < LocMapI.stop()) {
+        if (++LII == LIE)
+          break;
+        LocMapI.advanceTo(LII->start);
+      } else {
+        ++LocMapI;
+        if (!LocMapI.valid())
+          break;
+        LII = LI->advanceTo(LII, LocMapI.start());
+      }
+    }
+  }
+
+  // Finally, remove any remaining OldLocNo intervals and OldLocNo itself.
+  locations.erase(locations.begin() + OldLocNo);
+  LocMapI.goToBegin();
+  while (LocMapI.valid()) {
+    unsigned v = LocMapI.value();
+    if (v == OldLocNo) {
+      DEBUG(dbgs() << "Erasing [" << LocMapI.start() << ';'
+                   << LocMapI.stop() << ")\n");
+      LocMapI.erase();
+    } else {
+      if (v > OldLocNo)
+        LocMapI.setValueUnchecked(v-1);
+      ++LocMapI;
+    }
+  }
+
+  DEBUG({dbgs() << "Split result: \t"; print(dbgs(), 0);});
+  return DidChange;
+}
+
+bool
+UserValue::splitRegister(unsigned OldReg, ArrayRef<LiveInterval*> NewRegs) {
+  bool DidChange = false;
+  for (unsigned LocNo = 0, E = locations.size(); LocNo != E; ++LocNo) {
+    const MachineOperand *Loc = &locations[LocNo];
+    if (!Loc->isReg() || Loc->getReg() != OldReg)
+      continue;
+    DidChange |= splitLocation(LocNo, NewRegs);
+  }
+  return DidChange;
+}
+
+void LDVImpl::splitRegister(unsigned OldReg, ArrayRef<LiveInterval*> NewRegs) {
+  bool DidChange = false;
+  for (UserValue *UV = lookupVirtReg(OldReg); UV; UV = UV->getNext())
+    DidChange |= UV->splitRegister(OldReg, NewRegs);
+
+  if (!DidChange)
+    return;
+
+  // Map all of the new virtual registers.
+  UserValue *UV = lookupVirtReg(OldReg);
+  for (unsigned i = 0; i != NewRegs.size(); ++i)
+    mapVirtReg(NewRegs[i]->reg, UV);
+}
+
+void LiveDebugVariables::
+splitRegister(unsigned OldReg, ArrayRef<LiveInterval*> NewRegs) {
+  if (pImpl)
+    static_cast<LDVImpl*>(pImpl)->splitRegister(OldReg, NewRegs);
+}
+
 void
 UserValue::rewriteLocations(VirtRegMap &VRM, const TargetRegisterInfo &TRI) {
   // Iterate over locations in reverse makes it easier to handle coalescing.

Modified: llvm/trunk/lib/CodeGen/LiveDebugVariables.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LiveDebugVariables.h?rev=130998&r1=130997&r2=130998&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/LiveDebugVariables.h (original)
+++ llvm/trunk/lib/CodeGen/LiveDebugVariables.h Fri May  6 13:00:02 2011
@@ -21,10 +21,12 @@
 #ifndef LLVM_CODEGEN_LIVEDEBUGVARIABLES_H
 #define LLVM_CODEGEN_LIVEDEBUGVARIABLES_H
 
+#include "llvm/ADT/ArrayRef.h"
 #include "llvm/CodeGen/MachineFunctionPass.h"
 
 namespace llvm {
 
+class LiveInterval;
 class VirtRegMap;
 
 class LiveDebugVariables : public MachineFunctionPass {
@@ -42,6 +44,11 @@
   ///               register.
   void renameRegister(unsigned OldReg, unsigned NewReg, unsigned SubIdx);
 
+  /// splitRegister - Move any user variables in OldReg to the live ranges in
+  /// NewRegs where they are live. Mark the values as unavailable where no new
+  /// register is live.
+  void splitRegister(unsigned OldReg, ArrayRef<LiveInterval*> NewRegs);
+
   /// emitDebugValues - Emit new DBG_VALUE instructions reflecting the changes
   /// that happened during register allocation.
   /// @param VRM Rename virtual registers according to map.

Modified: llvm/trunk/lib/CodeGen/LiveRangeEdit.h
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/LiveRangeEdit.h?rev=130998&r1=130997&r2=130998&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/LiveRangeEdit.h (original)
+++ llvm/trunk/lib/CodeGen/LiveRangeEdit.h Fri May  6 13:00:02 2011
@@ -18,8 +18,9 @@
 #ifndef LLVM_CODEGEN_LIVERANGEEDIT_H
 #define LLVM_CODEGEN_LIVERANGEEDIT_H
 
-#include "llvm/CodeGen/LiveInterval.h"
+#include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/SmallPtrSet.h"
+#include "llvm/CodeGen/LiveInterval.h"
 
 namespace llvm {
 
@@ -113,6 +114,10 @@
   bool empty() const { return size() == 0; }
   LiveInterval *get(unsigned idx) const { return newRegs_[idx+firstNew_]; }
 
+  ArrayRef<LiveInterval*> regs() const {
+    return ArrayRef<LiveInterval*>(newRegs_).slice(firstNew_);
+  }
+
   /// FIXME: Temporary accessors until we can get rid of
   /// LiveIntervals::AddIntervalsForSpills
   SmallVectorImpl<LiveInterval*> *getNewVRegs() { return &newRegs_; }

Modified: llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp?rev=130998&r1=130997&r2=130998&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp (original)
+++ llvm/trunk/lib/CodeGen/RegAllocGreedy.cpp Fri May  6 13:00:02 2011
@@ -72,6 +72,7 @@
   MachineLoopRanges *LoopRanges;
   EdgeBundles *Bundles;
   SpillPlacement *SpillPlacer;
+  LiveDebugVariables *DebugVars;
 
   // state
   std::auto_ptr<Spiller> SpillerInstance;
@@ -920,6 +921,8 @@
 
   SmallVector<unsigned, 8> IntvMap;
   SE->finish(&IntvMap);
+  DebugVars->splitRegister(VirtReg.reg, LREdit.regs());
+
   LRStage.resize(MRI->getNumVirtRegs());
   unsigned OrigBlocks = SA->getNumThroughBlocks() + SA->getUseBlocks().size();
 
@@ -1284,6 +1287,7 @@
   SlotIndex SegStop  = SE->leaveIntvAfter(Uses[BestAfter]);
   SE->useIntv(SegStart, SegStop);
   SE->finish();
+  DebugVars->splitRegister(VirtReg.reg, LREdit.regs());
   setStage(NewVRegs.begin(), NewVRegs.end(), RS_Local);
   ++NumLocalSplits;
 
@@ -1415,6 +1419,7 @@
   LoopRanges = &getAnalysis<MachineLoopRanges>();
   Bundles = &getAnalysis<EdgeBundles>();
   SpillPlacer = &getAnalysis<SpillPlacement>();
+  DebugVars = &getAnalysis<LiveDebugVariables>();
 
   SA.reset(new SplitAnalysis(*VRM, *LIS, *Loops));
   SE.reset(new SplitEditor(*SA, *LIS, *VRM, *DomTree));
@@ -1433,7 +1438,7 @@
   }
 
   // Write out new DBG_VALUE instructions.
-  getAnalysis<LiveDebugVariables>().emitDebugValues(VRM);
+  DebugVars->emitDebugValues(VRM);
 
   // The pass output is in VirtRegMap. Release all the transient data.
   releaseMemory();



From rafael.espindola at gmail.com  Fri May  6 13:01:58 2011
From: rafael.espindola at gmail.com (Rafael Espindola)
Date: Fri, 06 May 2011 18:01:58 -0000
Subject: [llvm-commits] [llvm] r130999 - in /llvm/trunk/test:
 FrontendC++/2006-11-30-Pubnames.cpp FrontendC/2009-02-17-BitField-dbg.c
 FrontendC/2010-01-14-StaticVariable.c FrontendC/2010-02-16-DbgVarScope.c
 FrontendObjC/2009-08-17-DebugInfo.m
Message-ID: <20110506180158.A61AF2A6C12C@llvm.org>

Author: rafael
Date: Fri May  6 13:01:58 2011
New Revision: 130999

URL: http://llvm.org/viewvc/llvm-project?rev=130999&view=rev
Log:
Pass -disable-cfi to llc.

Modified:
    llvm/trunk/test/FrontendC++/2006-11-30-Pubnames.cpp
    llvm/trunk/test/FrontendC/2009-02-17-BitField-dbg.c
    llvm/trunk/test/FrontendC/2010-01-14-StaticVariable.c
    llvm/trunk/test/FrontendC/2010-02-16-DbgVarScope.c
    llvm/trunk/test/FrontendObjC/2009-08-17-DebugInfo.m

Modified: llvm/trunk/test/FrontendC++/2006-11-30-Pubnames.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC%2B%2B/2006-11-30-Pubnames.cpp?rev=130999&r1=130998&r2=130999&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC++/2006-11-30-Pubnames.cpp (original)
+++ llvm/trunk/test/FrontendC++/2006-11-30-Pubnames.cpp Fri May  6 13:01:58 2011
@@ -1,7 +1,7 @@
 // This is a regression test on debug info to make sure that we can access 
 // qualified global names.
 // RUN: %llvmgcc -S -O0 -g %s -o - | \
-// RUN:   llc --disable-fp-elim -o %t.s -O0
+// RUN:   llc -disable-cfi --disable-fp-elim -o %t.s -O0
 // RUN: %compile_c %t.s -o %t.o
 // RUN: %link %t.o -o %t.exe
 // RUN: %llvmdsymutil %t.exe 

Modified: llvm/trunk/test/FrontendC/2009-02-17-BitField-dbg.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC/2009-02-17-BitField-dbg.c?rev=130999&r1=130998&r2=130999&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC/2009-02-17-BitField-dbg.c (original)
+++ llvm/trunk/test/FrontendC/2009-02-17-BitField-dbg.c Fri May  6 13:01:58 2011
@@ -1,6 +1,6 @@
 // Check bitfields.
 // RUN: %llvmgcc -S -O0 -g %s -o - | \
-// RUN: llc --disable-fp-elim -o 2009-02-17-BitField-dbg.s
+// RUN: llc -disable-cfi --disable-fp-elim -o 2009-02-17-BitField-dbg.s
 // RUN: %compile_c 2009-02-17-BitField-dbg.s -o 2009-02-17-BitField-dbg.o
 // RUN: echo {ptype mystruct} > %t2
 // RUN: gdb -q -batch -n -x %t2 2009-02-17-BitField-dbg.o | \

Modified: llvm/trunk/test/FrontendC/2010-01-14-StaticVariable.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC/2010-01-14-StaticVariable.c?rev=130999&r1=130998&r2=130999&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC/2010-01-14-StaticVariable.c (original)
+++ llvm/trunk/test/FrontendC/2010-01-14-StaticVariable.c Fri May  6 13:01:58 2011
@@ -1,7 +1,7 @@
 // This is a regression test on debug info to make sure that llvm emitted
 // debug info does not crash gdb.
 // RUN: %llvmgcc -S -O0 -g %s -o - | \
-// RUN:    llc --disable-fp-elim -o %t.s -O0 -relocation-model=pic
+// RUN:    llc -disable-cfi --disable-fp-elim -o %t.s -O0 -relocation-model=pic
 // RUN: %compile_c %t.s -o %t.o
 // RUN: echo {quit\n} > %t.in 
 // RUN: gdb -q -batch -n -x %t.in %t.o > /dev/null

Modified: llvm/trunk/test/FrontendC/2010-02-16-DbgVarScope.c
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendC/2010-02-16-DbgVarScope.c?rev=130999&r1=130998&r2=130999&view=diff
==============================================================================
--- llvm/trunk/test/FrontendC/2010-02-16-DbgVarScope.c (original)
+++ llvm/trunk/test/FrontendC/2010-02-16-DbgVarScope.c Fri May  6 13:01:58 2011
@@ -1,5 +1,5 @@
 // RUN: %llvmgcc -S -O0 -g %s -o - | \
-// RUN:    llc --disable-fp-elim -o %t.s -O0 -relocation-model=pic
+// RUN:    llc -disable-cfi --disable-fp-elim -o %t.s -O0 -relocation-model=pic
 // RUN: %compile_c %t.s -o %t.o
 // RUN: %link %t.o -o %t.exe
 // RUN: echo {break 24\nrun\np loc\n} > %t.in 

Modified: llvm/trunk/test/FrontendObjC/2009-08-17-DebugInfo.m
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/FrontendObjC/2009-08-17-DebugInfo.m?rev=130999&r1=130998&r2=130999&view=diff
==============================================================================
--- llvm/trunk/test/FrontendObjC/2009-08-17-DebugInfo.m (original)
+++ llvm/trunk/test/FrontendObjC/2009-08-17-DebugInfo.m Fri May  6 13:01:58 2011
@@ -1,6 +1,6 @@
 // This is a regression test on debug info to make sure that we can set a
 // breakpoint on a objective message.
-// RUN: %llvmgcc -S -O0 -g %s -o - | llc -o %t.s -O0
+// RUN: %llvmgcc -S -O0 -g %s -o - | llc -disable-cfi -o %t.s -O0
 // RUN: %compile_c %t.s -o %t.o
 // RUN: %link %t.o -o %t.exe -framework Foundation
 // RUN: echo {break randomFunc\n} > %t.in 



From atrick at apple.com  Fri May  6 13:14:32 2011
From: atrick at apple.com (Andrew Trick)
Date: Fri, 06 May 2011 18:14:32 -0000
Subject: [llvm-commits] [llvm] r131001 -
	/llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp
Message-ID: <20110506181433.05AA72A6C12C@llvm.org>

Author: atrick
Date: Fri May  6 13:14:32 2011
New Revision: 131001

URL: http://llvm.org/viewvc/llvm-project?rev=131001&view=rev
Log:
Typo: Reviewed by Alistair.

Modified:
    llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp

Modified: llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp?rev=131001&r1=131000&r2=131001&view=diff
==============================================================================
--- llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp (original)
+++ llvm/trunk/lib/CodeGen/PostRASchedulerList.cpp Fri May  6 13:14:32 2011
@@ -540,7 +540,7 @@
 #endif
   --SuccSU->NumPredsLeft;
 
-  // Standard scheduler algorithms will recomute the depth of the successor
+  // Standard scheduler algorithms will recompute the depth of the successor
   // here as such:
   //   SuccSU->setDepthToAtLeast(SU->getDepth() + SuccEdge->getLatency());
   //



From gkistanova at gmail.com  Fri May  6 13:24:46 2011
From: gkistanova at gmail.com (Galina Kistanova)
Date: Fri, 06 May 2011 18:24:46 -0000
Subject: [llvm-commits] [llvm] r131002 - in /llvm/trunk/test: DebugInfo/X86/
 DebugInfo/X86/dg.exp DebugInfo/X86/stmt-list.ll DebugInfo/stmt-list.ll
 Transforms/LoopStrengthReduce/2009-11-10-LSRCrash.ll
 Transforms/LoopStrengthReduce/X86/
 Transforms/LoopStrengthReduce/X86/2009-11-10-LSRCrash.ll
 Transforms/LoopStrengthReduce/X86/dg.exp Transforms/TailDup/X86/
 Transforms/TailDup/X86/dg.exp Transforms/TailDup/X86/if-tail-dup.ll
 Transforms/TailDup/if-tail-dup.ll
Message-ID: <20110506182446.E1BB92A6C12C@llvm.org>

Author: gkistanova
Date: Fri May  6 13:24:46 2011
New Revision: 131002

URL: http://llvm.org/viewvc/llvm-project?rev=131002&view=rev
Log:
Move few target-dependant tests to appropriate directories.

Added:
    llvm/trunk/test/DebugInfo/X86/
    llvm/trunk/test/DebugInfo/X86/dg.exp
    llvm/trunk/test/DebugInfo/X86/stmt-list.ll
    llvm/trunk/test/Transforms/LoopStrengthReduce/X86/
    llvm/trunk/test/Transforms/LoopStrengthReduce/X86/2009-11-10-LSRCrash.ll
    llvm/trunk/test/Transforms/LoopStrengthReduce/X86/dg.exp
    llvm/trunk/test/Transforms/TailDup/X86/
    llvm/trunk/test/Transforms/TailDup/X86/dg.exp
    llvm/trunk/test/Transforms/TailDup/X86/if-tail-dup.ll
Removed:
    llvm/trunk/test/DebugInfo/stmt-list.ll
    llvm/trunk/test/Transforms/LoopStrengthReduce/2009-11-10-LSRCrash.ll
    llvm/trunk/test/Transforms/TailDup/if-tail-dup.ll

Added: llvm/trunk/test/DebugInfo/X86/dg.exp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/DebugInfo/X86/dg.exp?rev=131002&view=auto
==============================================================================
--- llvm/trunk/test/DebugInfo/X86/dg.exp (added)
+++ llvm/trunk/test/DebugInfo/X86/dg.exp Fri May  6 13:24:46 2011
@@ -0,0 +1,5 @@
+load_lib llvm.exp
+
+if { [llvm_supports_target X86] } {
+  RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll}]]
+}

Added: llvm/trunk/test/DebugInfo/X86/stmt-list.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/DebugInfo/X86/stmt-list.ll?rev=131002&view=auto
==============================================================================
--- llvm/trunk/test/DebugInfo/X86/stmt-list.ll (added)
+++ llvm/trunk/test/DebugInfo/X86/stmt-list.ll Fri May  6 13:24:46 2011
@@ -0,0 +1,19 @@
+; RUN: llc -mtriple x86_64-pc-linux-gnu < %s | FileCheck %s
+
+; CHECK:      .section        .debug_line,"", at progbits
+; CHECK-NEXT: .Lsection_line:
+
+; CHECK:      .long   .Lsection_line          # DW_AT_stmt_list
+
+define void @f() {
+entry:
+  ret void
+}
+
+!llvm.dbg.sp = !{!0}
+
+!0 = metadata !{i32 589870, i32 0, metadata !1, metadata !"f", metadata !"f", metadata !"", metadata !1, i32 1, metadata !3, i1 false, i1 true, i32 0, i32 0, i32 0, i32 256, i1 true, void ()* @f, null, null} ; [ DW_TAG_subprogram ]
+!1 = metadata !{i32 589865, metadata !"test2.c", metadata !"/home/espindola/llvm", metadata !2} ; [ DW_TAG_file_type ]
+!2 = metadata !{i32 589841, i32 0, i32 12, metadata !"test2.c", metadata !"/home/espindola/llvm", metadata !"clang version 3.0 ()", i1 true, i1 true, metadata !"", i32 0} ; [ DW_TAG_compile_unit ]
+!3 = metadata !{i32 589845, metadata !1, metadata !"", metadata !1, i32 0, i64 0, i64 0, i32 0, i32 0, i32 0, metadata !4, i32 0, i32 0} ; [ DW_TAG_subroutine_type ]
+!4 = metadata !{null}

Removed: llvm/trunk/test/DebugInfo/stmt-list.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/DebugInfo/stmt-list.ll?rev=131001&view=auto
==============================================================================
--- llvm/trunk/test/DebugInfo/stmt-list.ll (original)
+++ llvm/trunk/test/DebugInfo/stmt-list.ll (removed)
@@ -1,19 +0,0 @@
-; RUN: llc -mtriple x86_64-pc-linux-gnu < %s | FileCheck %s
-
-; CHECK:      .section        .debug_line,"", at progbits
-; CHECK-NEXT: .Lsection_line:
-
-; CHECK:      .long   .Lsection_line          # DW_AT_stmt_list
-
-define void @f() {
-entry:
-  ret void
-}
-
-!llvm.dbg.sp = !{!0}
-
-!0 = metadata !{i32 589870, i32 0, metadata !1, metadata !"f", metadata !"f", metadata !"", metadata !1, i32 1, metadata !3, i1 false, i1 true, i32 0, i32 0, i32 0, i32 256, i1 true, void ()* @f, null, null} ; [ DW_TAG_subprogram ]
-!1 = metadata !{i32 589865, metadata !"test2.c", metadata !"/home/espindola/llvm", metadata !2} ; [ DW_TAG_file_type ]
-!2 = metadata !{i32 589841, i32 0, i32 12, metadata !"test2.c", metadata !"/home/espindola/llvm", metadata !"clang version 3.0 ()", i1 true, i1 true, metadata !"", i32 0} ; [ DW_TAG_compile_unit ]
-!3 = metadata !{i32 589845, metadata !1, metadata !"", metadata !1, i32 0, i64 0, i64 0, i32 0, i32 0, i32 0, metadata !4, i32 0, i32 0} ; [ DW_TAG_subroutine_type ]
-!4 = metadata !{null}

Removed: llvm/trunk/test/Transforms/LoopStrengthReduce/2009-11-10-LSRCrash.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LoopStrengthReduce/2009-11-10-LSRCrash.ll?rev=131001&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/LoopStrengthReduce/2009-11-10-LSRCrash.ll (original)
+++ llvm/trunk/test/Transforms/LoopStrengthReduce/2009-11-10-LSRCrash.ll (removed)
@@ -1,130 +0,0 @@
-; RUN: llc < %s -mtriple=i386-apple-darwin11
-
-define void @_ZN4llvm20SelectionDAGLowering14visitInlineAsmENS_8CallSiteE() nounwind ssp align 2 {
-entry:
-  br i1 undef, label %bb3.i, label %bb4.i
-
-bb3.i:                                            ; preds = %entry
-  unreachable
-
-bb4.i:                                            ; preds = %entry
-  br i1 undef, label %bb.i.i, label %_ZNK4llvm8CallSite14getCalledValueEv.exit
-
-bb.i.i:                                           ; preds = %bb4.i
-  unreachable
-
-_ZNK4llvm8CallSite14getCalledValueEv.exit:        ; preds = %bb4.i
-  br i1 undef, label %_ZN4llvm4castINS_9InlineAsmEPNS_5ValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_.exit, label %bb6.i
-
-bb6.i:                                            ; preds = %_ZNK4llvm8CallSite14getCalledValueEv.exit
-  unreachable
-
-_ZN4llvm4castINS_9InlineAsmEPNS_5ValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_.exit: ; preds = %_ZNK4llvm8CallSite14getCalledValueEv.exit
-  br i1 undef, label %_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit, label %bb.i
-
-bb.i:                                             ; preds = %_ZN4llvm4castINS_9InlineAsmEPNS_5ValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_.exit
-  br label %_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit
-
-_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit: ; preds = %bb.i, %_ZN4llvm4castINS_9InlineAsmEPNS_5ValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_.exit
-  br i1 undef, label %bb50, label %bb27
-
-bb27:                                             ; preds = %_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit
-  br i1 undef, label %bb1.i727, label %bb.i.i726
-
-bb.i.i726:                                        ; preds = %bb27
-  unreachable
-
-bb1.i727:                                         ; preds = %bb27
-  unreachable
-
-bb50:                                             ; preds = %_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit
-  br label %bb107
-
-bb51:                                             ; preds = %bb107
-  br i1 undef, label %bb105, label %bb106
-
-bb105:                                            ; preds = %bb51
-  unreachable
-
-bb106:                                            ; preds = %bb51
-  br label %bb107
-
-bb107:                                            ; preds = %bb106, %bb50
-  br i1 undef, label %bb108, label %bb51
-
-bb108:                                            ; preds = %bb107
-  br i1 undef, label %bb242, label %bb114
-
-bb114:                                            ; preds = %bb108
-  br i1 undef, label %bb141, label %bb116
-
-bb116:                                            ; preds = %bb114
-  br i1 undef, label %bb120, label %bb121
-
-bb120:                                            ; preds = %bb116
-  unreachable
-
-bb121:                                            ; preds = %bb116
-  unreachable
-
-bb141:                                            ; preds = %bb114
-  br i1 undef, label %bb182, label %bb143
-
-bb143:                                            ; preds = %bb141
-  br label %bb157
-
-bb144:                                            ; preds = %bb.i.i.i843
-  switch i32 undef, label %bb155 [
-    i32 2, label %bb153
-    i32 6, label %bb153
-    i32 4, label %bb153
-  ]
-
-bb153:                                            ; preds = %bb144, %bb144, %bb144
-  %indvar.next = add i32 %indvar, 1               ; <i32> [#uses=1]
-  br label %bb157
-
-bb155:                                            ; preds = %bb144
-  unreachable
-
-bb157:                                            ; preds = %bb153, %bb143
-  %indvar = phi i32 [ %indvar.next, %bb153 ], [ 0, %bb143 ] ; <i32> [#uses=2]
-  %0 = icmp eq i32 undef, %indvar                 ; <i1> [#uses=1]
-  switch i16 undef, label %bb6.i841 [
-    i16 9, label %_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit
-    i16 26, label %_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit
-  ]
-
-bb6.i841:                                         ; preds = %bb157
-  unreachable
-
-_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit: ; preds = %bb157, %bb157
-  br i1 undef, label %bb.i.i.i843, label %bb1.i.i.i844
-
-bb.i.i.i843:                                      ; preds = %_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit
-  br i1 %0, label %bb158, label %bb144
-
-bb1.i.i.i844:                                     ; preds = %_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit
-  unreachable
-
-bb158:                                            ; preds = %bb.i.i.i843
-  br i1 undef, label %bb177, label %bb176
-
-bb176:                                            ; preds = %bb158
-  unreachable
-
-bb177:                                            ; preds = %bb158
-  br i1 undef, label %bb179, label %bb178
-
-bb178:                                            ; preds = %bb177
-  unreachable
-
-bb179:                                            ; preds = %bb177
-  unreachable
-
-bb182:                                            ; preds = %bb141
-  unreachable
-
-bb242:                                            ; preds = %bb108
-  unreachable
-}

Added: llvm/trunk/test/Transforms/LoopStrengthReduce/X86/2009-11-10-LSRCrash.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LoopStrengthReduce/X86/2009-11-10-LSRCrash.ll?rev=131002&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/LoopStrengthReduce/X86/2009-11-10-LSRCrash.ll (added)
+++ llvm/trunk/test/Transforms/LoopStrengthReduce/X86/2009-11-10-LSRCrash.ll Fri May  6 13:24:46 2011
@@ -0,0 +1,130 @@
+; RUN: llc < %s -mtriple=i386-apple-darwin11
+
+define void @_ZN4llvm20SelectionDAGLowering14visitInlineAsmENS_8CallSiteE() nounwind ssp align 2 {
+entry:
+  br i1 undef, label %bb3.i, label %bb4.i
+
+bb3.i:                                            ; preds = %entry
+  unreachable
+
+bb4.i:                                            ; preds = %entry
+  br i1 undef, label %bb.i.i, label %_ZNK4llvm8CallSite14getCalledValueEv.exit
+
+bb.i.i:                                           ; preds = %bb4.i
+  unreachable
+
+_ZNK4llvm8CallSite14getCalledValueEv.exit:        ; preds = %bb4.i
+  br i1 undef, label %_ZN4llvm4castINS_9InlineAsmEPNS_5ValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_.exit, label %bb6.i
+
+bb6.i:                                            ; preds = %_ZNK4llvm8CallSite14getCalledValueEv.exit
+  unreachable
+
+_ZN4llvm4castINS_9InlineAsmEPNS_5ValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_.exit: ; preds = %_ZNK4llvm8CallSite14getCalledValueEv.exit
+  br i1 undef, label %_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit, label %bb.i
+
+bb.i:                                             ; preds = %_ZN4llvm4castINS_9InlineAsmEPNS_5ValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_.exit
+  br label %_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit
+
+_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit: ; preds = %bb.i, %_ZN4llvm4castINS_9InlineAsmEPNS_5ValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_.exit
+  br i1 undef, label %bb50, label %bb27
+
+bb27:                                             ; preds = %_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit
+  br i1 undef, label %bb1.i727, label %bb.i.i726
+
+bb.i.i726:                                        ; preds = %bb27
+  unreachable
+
+bb1.i727:                                         ; preds = %bb27
+  unreachable
+
+bb50:                                             ; preds = %_ZL25hasInlineAsmMemConstraintRSt6vectorIN4llvm9InlineAsm14ConstraintInfoESaIS2_EERKNS0_14TargetLoweringE.exit
+  br label %bb107
+
+bb51:                                             ; preds = %bb107
+  br i1 undef, label %bb105, label %bb106
+
+bb105:                                            ; preds = %bb51
+  unreachable
+
+bb106:                                            ; preds = %bb51
+  br label %bb107
+
+bb107:                                            ; preds = %bb106, %bb50
+  br i1 undef, label %bb108, label %bb51
+
+bb108:                                            ; preds = %bb107
+  br i1 undef, label %bb242, label %bb114
+
+bb114:                                            ; preds = %bb108
+  br i1 undef, label %bb141, label %bb116
+
+bb116:                                            ; preds = %bb114
+  br i1 undef, label %bb120, label %bb121
+
+bb120:                                            ; preds = %bb116
+  unreachable
+
+bb121:                                            ; preds = %bb116
+  unreachable
+
+bb141:                                            ; preds = %bb114
+  br i1 undef, label %bb182, label %bb143
+
+bb143:                                            ; preds = %bb141
+  br label %bb157
+
+bb144:                                            ; preds = %bb.i.i.i843
+  switch i32 undef, label %bb155 [
+    i32 2, label %bb153
+    i32 6, label %bb153
+    i32 4, label %bb153
+  ]
+
+bb153:                                            ; preds = %bb144, %bb144, %bb144
+  %indvar.next = add i32 %indvar, 1               ; <i32> [#uses=1]
+  br label %bb157
+
+bb155:                                            ; preds = %bb144
+  unreachable
+
+bb157:                                            ; preds = %bb153, %bb143
+  %indvar = phi i32 [ %indvar.next, %bb153 ], [ 0, %bb143 ] ; <i32> [#uses=2]
+  %0 = icmp eq i32 undef, %indvar                 ; <i1> [#uses=1]
+  switch i16 undef, label %bb6.i841 [
+    i16 9, label %_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit
+    i16 26, label %_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit
+  ]
+
+bb6.i841:                                         ; preds = %bb157
+  unreachable
+
+_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit: ; preds = %bb157, %bb157
+  br i1 undef, label %bb.i.i.i843, label %bb1.i.i.i844
+
+bb.i.i.i843:                                      ; preds = %_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit
+  br i1 %0, label %bb158, label %bb144
+
+bb1.i.i.i844:                                     ; preds = %_ZN4llvm4castINS_14ConstantSDNodeENS_7SDValueEEENS_10cast_rettyIT_T0_E8ret_typeERKS5_.exit
+  unreachable
+
+bb158:                                            ; preds = %bb.i.i.i843
+  br i1 undef, label %bb177, label %bb176
+
+bb176:                                            ; preds = %bb158
+  unreachable
+
+bb177:                                            ; preds = %bb158
+  br i1 undef, label %bb179, label %bb178
+
+bb178:                                            ; preds = %bb177
+  unreachable
+
+bb179:                                            ; preds = %bb177
+  unreachable
+
+bb182:                                            ; preds = %bb141
+  unreachable
+
+bb242:                                            ; preds = %bb108
+  unreachable
+}

Added: llvm/trunk/test/Transforms/LoopStrengthReduce/X86/dg.exp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LoopStrengthReduce/X86/dg.exp?rev=131002&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/LoopStrengthReduce/X86/dg.exp (added)
+++ llvm/trunk/test/Transforms/LoopStrengthReduce/X86/dg.exp Fri May  6 13:24:46 2011
@@ -0,0 +1,5 @@
+load_lib llvm.exp
+
+if { [llvm_supports_target X86] } {
+  RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll}]]
+}

Added: llvm/trunk/test/Transforms/TailDup/X86/dg.exp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/TailDup/X86/dg.exp?rev=131002&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/TailDup/X86/dg.exp (added)
+++ llvm/trunk/test/Transforms/TailDup/X86/dg.exp Fri May  6 13:24:46 2011
@@ -0,0 +1,5 @@
+load_lib llvm.exp
+
+if { [llvm_supports_target X86] } {
+  RunLLVMTests [lsort [glob -nocomplain $srcdir/$subdir/*.{ll}]]
+}

Added: llvm/trunk/test/Transforms/TailDup/X86/if-tail-dup.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/TailDup/X86/if-tail-dup.ll?rev=131002&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/TailDup/X86/if-tail-dup.ll (added)
+++ llvm/trunk/test/Transforms/TailDup/X86/if-tail-dup.ll Fri May  6 13:24:46 2011
@@ -0,0 +1,49 @@
+; RUN: opt < %s -tailduplicate | \
+; RUN:   llc -march=x86 -o %t
+; RUN: grep {\\\<je\\\>} %t
+; RUN: not grep jmp %t
+; END.
+; This should have no unconditional jumps in it.  The C source is:
+
+;void foo(int c, int* P) {
+;  if (c & 1)  P[0] = 1;
+;  if (c & 2)  P[1] = 1;
+;  if (c & 4)  P[2] = 1;
+;  if (c & 8)  P[3] = 1;
+;}
+
+define void @foo(i32 %c, i32* %P) {
+entry:
+	%tmp1 = and i32 %c, 1		; <i32> [#uses=1]
+	%tmp1.upgrd.1 = icmp eq i32 %tmp1, 0		; <i1> [#uses=1]
+	br i1 %tmp1.upgrd.1, label %cond_next, label %cond_true
+cond_true:		; preds = %entry
+	store i32 1, i32* %P
+	br label %cond_next
+cond_next:		; preds = %cond_true, %entry
+	%tmp5 = and i32 %c, 2		; <i32> [#uses=1]
+	%tmp5.upgrd.2 = icmp eq i32 %tmp5, 0		; <i1> [#uses=1]
+	br i1 %tmp5.upgrd.2, label %cond_next10, label %cond_true6
+cond_true6:		; preds = %cond_next
+	%tmp8 = getelementptr i32* %P, i32 1		; <i32*> [#uses=1]
+	store i32 1, i32* %tmp8
+	br label %cond_next10
+cond_next10:		; preds = %cond_true6, %cond_next
+	%tmp13 = and i32 %c, 4		; <i32> [#uses=1]
+	%tmp13.upgrd.3 = icmp eq i32 %tmp13, 0		; <i1> [#uses=1]
+	br i1 %tmp13.upgrd.3, label %cond_next18, label %cond_true14
+cond_true14:		; preds = %cond_next10
+	%tmp16 = getelementptr i32* %P, i32 2		; <i32*> [#uses=1]
+	store i32 1, i32* %tmp16
+	br label %cond_next18
+cond_next18:		; preds = %cond_true14, %cond_next10
+	%tmp21 = and i32 %c, 8		; <i32> [#uses=1]
+	%tmp21.upgrd.4 = icmp eq i32 %tmp21, 0		; <i1> [#uses=1]
+	br i1 %tmp21.upgrd.4, label %return, label %cond_true22
+cond_true22:		; preds = %cond_next18
+	%tmp24 = getelementptr i32* %P, i32 3		; <i32*> [#uses=1]
+	store i32 1, i32* %tmp24
+	ret void
+return:		; preds = %cond_next18
+	ret void
+}

Removed: llvm/trunk/test/Transforms/TailDup/if-tail-dup.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/TailDup/if-tail-dup.ll?rev=131001&view=auto
==============================================================================
--- llvm/trunk/test/Transforms/TailDup/if-tail-dup.ll (original)
+++ llvm/trunk/test/Transforms/TailDup/if-tail-dup.ll (removed)
@@ -1,49 +0,0 @@
-; RUN: opt < %s -tailduplicate | \
-; RUN:   llc -march=x86 -o %t
-; RUN: grep {\\\<je\\\>} %t
-; RUN: not grep jmp %t
-; END.
-; This should have no unconditional jumps in it.  The C source is:
-
-;void foo(int c, int* P) {
-;  if (c & 1)  P[0] = 1;
-;  if (c & 2)  P[1] = 1;
-;  if (c & 4)  P[2] = 1;
-;  if (c & 8)  P[3] = 1;
-;}
-
-define void @foo(i32 %c, i32* %P) {
-entry:
-	%tmp1 = and i32 %c, 1		; <i32> [#uses=1]
-	%tmp1.upgrd.1 = icmp eq i32 %tmp1, 0		; <i1> [#uses=1]
-	br i1 %tmp1.upgrd.1, label %cond_next, label %cond_true
-cond_true:		; preds = %entry
-	store i32 1, i32* %P
-	br label %cond_next
-cond_next:		; preds = %cond_true, %entry
-	%tmp5 = and i32 %c, 2		; <i32> [#uses=1]
-	%tmp5.upgrd.2 = icmp eq i32 %tmp5, 0		; <i1> [#uses=1]
-	br i1 %tmp5.upgrd.2, label %cond_next10, label %cond_true6
-cond_true6:		; preds = %cond_next
-	%tmp8 = getelementptr i32* %P, i32 1		; <i32*> [#uses=1]
-	store i32 1, i32* %tmp8
-	br label %cond_next10
-cond_next10:		; preds = %cond_true6, %cond_next
-	%tmp13 = and i32 %c, 4		; <i32> [#uses=1]
-	%tmp13.upgrd.3 = icmp eq i32 %tmp13, 0		; <i1> [#uses=1]
-	br i1 %tmp13.upgrd.3, label %cond_next18, label %cond_true14
-cond_true14:		; preds = %cond_next10
-	%tmp16 = getelementptr i32* %P, i32 2		; <i32*> [#uses=1]
-	store i32 1, i32* %tmp16
-	br label %cond_next18
-cond_next18:		; preds = %cond_true14, %cond_next10
-	%tmp21 = and i32 %c, 8		; <i32> [#uses=1]
-	%tmp21.upgrd.4 = icmp eq i32 %tmp21, 0		; <i1> [#uses=1]
-	br i1 %tmp21.upgrd.4, label %return, label %cond_true22
-cond_true22:		; preds = %cond_next18
-	%tmp24 = getelementptr i32* %P, i32 3		; <i32*> [#uses=1]
-	store i32 1, i32* %tmp24
-	ret void
-return:		; preds = %cond_next18
-	ret void
-}



From grosbach at apple.com  Fri May  6 13:39:28 2011
From: grosbach at apple.com (Jim Grosbach)
Date: Fri, 06 May 2011 18:39:28 -0000
Subject: [llvm-commits] [llvm] r131004 -
	/llvm/trunk/utils/TableGen/TableGen.cpp
Message-ID: <20110506183928.DB13A2A6C12C@llvm.org>

Author: grosbach
Date: Fri May  6 13:39:28 2011
New Revision: 131004

URL: http://llvm.org/viewvc/llvm-project?rev=131004&view=rev
Log:
ParseFile() may throw, so extend the try/catch to handle that.

Modified:
    llvm/trunk/utils/TableGen/TableGen.cpp

Modified: llvm/trunk/utils/TableGen/TableGen.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/utils/TableGen/TableGen.cpp?rev=131004&r1=131003&r2=131004&view=diff
==============================================================================
--- llvm/trunk/utils/TableGen/TableGen.cpp (original)
+++ llvm/trunk/utils/TableGen/TableGen.cpp Fri May  6 13:39:28 2011
@@ -228,19 +228,19 @@
   cl::ParseCommandLineOptions(argc, argv);
 
 
-  // Parse the input file.
-  if (ParseFile(InputFilename, IncludeDirs, SrcMgr, Records))
-    return 1;
-
-  std::string Error;
-  tool_output_file Out(OutputFilename.c_str(), Error);
-  if (!Error.empty()) {
-    errs() << argv[0] << ": error opening " << OutputFilename
-           << ":" << Error << "\n";
-    return 1;
-  }
-
   try {
+    // Parse the input file.
+    if (ParseFile(InputFilename, IncludeDirs, SrcMgr, Records))
+      return 1;
+
+    std::string Error;
+    tool_output_file Out(OutputFilename.c_str(), Error);
+    if (!Error.empty()) {
+      errs() << argv[0] << ": error opening " << OutputFilename
+        << ":" << Error << "\n";
+      return 1;
+    }
+
     switch (Action) {
     case PrintRecords:
       Out.os() << Records;           // No argument, dump all contents



From grosbach at apple.com  Fri May  6 13:47:45 2011
From: grosbach at apple.com (Jim Grosbach)
Date: Fri, 06 May 2011 18:47:45 -0000
Subject: [llvm-commits] [llvm] r131005 -
	/llvm/trunk/utils/TableGen/Record.cpp
Message-ID: <20110506184745.B92A52A6C12C@llvm.org>

Author: grosbach
Date: Fri May  6 13:47:45 2011
New Revision: 131005

URL: http://llvm.org/viewvc/llvm-project?rev=131005&view=rev
Log:
Improve diagnostics for some parse errors. Not asserting when a user input
error is detected is a good thing.

Modified:
    llvm/trunk/utils/TableGen/Record.cpp

Modified: llvm/trunk/utils/TableGen/Record.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/utils/TableGen/Record.cpp?rev=131005&r1=131004&r2=131005&view=diff
==============================================================================
--- llvm/trunk/utils/TableGen/Record.cpp (original)
+++ llvm/trunk/utils/TableGen/Record.cpp Fri May  6 13:47:45 2011
@@ -583,9 +583,7 @@
         if (Record *D = (CurRec->getRecords()).getDef(Name))
           return new DefInit(D);
 
-        errs() << "Variable not defined: '" + Name + "'\n";
-        assert(0 && "Variable not found");
-        return 0;
+        throw TGError(CurRec->getLoc(), "Undefined reference:'" + Name + "'\n");
       }
     }
     break;
@@ -813,15 +811,13 @@
   OpInit *RHSo = dynamic_cast<OpInit*>(RHS);
 
   if (!RHSo) {
-    errs() << "!foreach requires an operator\n";
-    assert(0 && "No operator for !foreach");
+    throw TGError(CurRec->getLoc(), "!foreach requires an operator\n");
   }
 
   TypedInit *LHSt = dynamic_cast<TypedInit*>(LHS);
 
   if (!LHSt) {
-    errs() << "!foreach requires typed variable\n";
-    assert(0 && "No typed variable for !foreach");
+    throw TGError(CurRec->getLoc(), "!foreach requires typed variable\n");
   }
 
   if ((MHSd && DagType) || (MHSl && ListType)) {



From grosbach at apple.com  Fri May  6 14:00:22 2011
From: grosbach at apple.com (Jim Grosbach)
Date: Fri, 06 May 2011 12:00:22 -0700
Subject: [llvm-commits] Fwd: [PATCHES] fix duplication and FIXMEs in JIT
 code
In-Reply-To: <BANLkTikY6UseB3du3ajhtpGtd4hB2E1zog@mail.gmail.com>
References: <BANLkTi=xMh=W_iY6MqYf2OLmDnDUmDGZCg@mail.gmail.com>
	<BANLkTikY6UseB3du3ajhtpGtd4hB2E1zog@mail.gmail.com>
Message-ID: <435AECFF-2C0B-42BB-865C-05ECEA9080F5@apple.com>

Looks reasonable to me. Thanks!

-Jim

On Apr 24, 2011, at 3:20 PM, nobled wrote:

> Whoops, forgot to join the mailing list, so this got stuck in the
> moderation queue...
> 
> ---------- Forwarded message ----------
> From: nobled <nobled at dreamwidth.org>
> Date: Sun, Apr 24, 2011 at 11:05 AM
> Subject: [PATCHES] fix duplication and FIXMEs in JIT code
> To: llvm-commits at cs.uiuc.edu
> 
> 
> (Just so you know, I do have commit access.)
> 
> These patches address this comment from the MCJIT code:
> 
>  // FIXME: This should be lifted out of here, it isn't something which should
>  // be part of the JIT policy, rather the burden for this selection should be
>  // pushed to clients.
> ====
> [PATCH 1/3] ExecutionEngine: fix JIT/MCJIT selectTarget() duplication
> 
> (Merges the near-identical files lib/EE/JIT/TargetSelect.cpp and
> lib/EE/MCJIT/TargetSelect.cpp into lib/EE/TargetSelect.cpp.)
> 
> This prepares for making JITCtor/MCJITCtor take a
> TargetMachine* directly from clients like EngineBuilder.
> ====
> [PATCH 2/3] ExecutionEngine: push TargetMachine creation into clients
> 
> In particular, into EngineBuilder. This should only impact
> the private API between the EE and EB classes, not external
> clients, since JITCtor and MCJITCtor are both protected members.
> ====
> [PATCH 3/3] ExecutionEngine: move createJIT definition
> 
> As an ExecutionEngine class function, its definition
> really belongs in ExecutionEngine.cpp instead of in JIT.cpp.
> ====
> 
> As for this comment:
> 
>   // Try to register the program as a source of symbols to resolve against.
>   //
>   // FIXME: Don't do this here.
>   sys::DynamicLibrary::LoadLibraryPermanently(0, NULL);
> 
> where in the stack *should* that go, ideally?
> <0001-ExecutionEngine-fix-JIT-MCJIT-selectTarget-duplicati.patch.txt><0002-ExecutionEngine-push-TargetMachine-creation-into-cli.patch.txt><0003-ExecutionEngine-move-createJIT-definition.patch.txt>_______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits


From aggarwa4 at illinois.edu  Fri May  6 14:26:18 2011
From: aggarwa4 at illinois.edu (Arushi Aggarwal)
Date: Fri, 06 May 2011 19:26:18 -0000
Subject: [llvm-commits] [poolalloc] r131006 - in /poolalloc/trunk:
	include/assistDS/ lib/AssistDS/
Message-ID: <20110506192618.A7DAE2A6C12C@llvm.org>

Author: aggarwa4
Date: Fri May  6 14:26:18 2011
New Revision: 131006

URL: http://llvm.org/viewvc/llvm-project?rev=131006&view=rev
Log:
Split pass definitions into separate header files.


Added:
    poolalloc/trunk/include/assistDS/ArgCast.h
    poolalloc/trunk/include/assistDS/FuncSpec.h
    poolalloc/trunk/include/assistDS/GEPExprArgs.h
    poolalloc/trunk/include/assistDS/Int2PtrCmp.h
    poolalloc/trunk/include/assistDS/LoadArgs.h
    poolalloc/trunk/include/assistDS/MergeGEP.h
    poolalloc/trunk/include/assistDS/SimplifyExtractValue.h
    poolalloc/trunk/include/assistDS/SimplifyGEP.h
    poolalloc/trunk/include/assistDS/SimplifyInsertValue.h
    poolalloc/trunk/include/assistDS/StructReturnToPointer.h
    poolalloc/trunk/include/assistDS/VarArgsFunc.h
    poolalloc/trunk/lib/AssistDS/GEPExprArgs.cpp
    poolalloc/trunk/lib/AssistDS/MergeGEP.cpp
Removed:
    poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp
    poolalloc/trunk/lib/AssistDS/MergeArrayIndexGEP.cpp
Modified:
    poolalloc/trunk/lib/AssistDS/ArgCast.cpp
    poolalloc/trunk/lib/AssistDS/FuncSpec.cpp
    poolalloc/trunk/lib/AssistDS/Int2PtrCmp.cpp
    poolalloc/trunk/lib/AssistDS/LoadArgs.cpp
    poolalloc/trunk/lib/AssistDS/SimplifyExtractValue.cpp
    poolalloc/trunk/lib/AssistDS/SimplifyGEP.cpp
    poolalloc/trunk/lib/AssistDS/SimplifyInsertValue.cpp
    poolalloc/trunk/lib/AssistDS/StructReturnToPointer.cpp
    poolalloc/trunk/lib/AssistDS/TypeAnalysis.cpp
    poolalloc/trunk/lib/AssistDS/VarArgsFunc.cpp

Added: poolalloc/trunk/include/assistDS/ArgCast.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/ArgCast.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/ArgCast.h (added)
+++ poolalloc/trunk/include/assistDS/ArgCast.h Fri May  6 14:26:18 2011
@@ -0,0 +1,34 @@
+//===-------- ArgCast.cpp - Cast Arguments to Calls -----------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+// Convert
+// call(bitcast (.., T1 arg, ...)F to(..., T2 arg, ...))(..., T2 val, ...)
+// to
+// val1 = bitcast T2 val to T1
+// call F (..., T1 val1, ...)
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: ArgCast
+  //
+  // Description:
+  //  Implement an LLVM pass that cleans up call sites that take casted args
+  //
+  class ArgCast : public ModulePass {
+  public:
+    static char ID;
+    ArgCast() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/FuncSpec.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/FuncSpec.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/FuncSpec.h (added)
+++ poolalloc/trunk/include/assistDS/FuncSpec.h Fri May  6 14:26:18 2011
@@ -0,0 +1,35 @@
+//===-- FuncSpec.cpp - Clone Functions With Constant Function Ptr Args ----===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// This pass clones functions that take constant function pointers as arguments
+// from some call sites. It changes those call sites to call cloned functions.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: FuncSpec
+  //
+  // Description:
+  //  Implement an LLVM pass that clones functions which are passed
+  //  as an argument
+  //  
+  //
+  class FuncSpec : public ModulePass {
+  public:
+    static char ID;
+    FuncSpec() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/GEPExprArgs.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/GEPExprArgs.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/GEPExprArgs.h (added)
+++ poolalloc/trunk/include/assistDS/GEPExprArgs.h Fri May  6 14:26:18 2011
@@ -0,0 +1,34 @@
+//===-- GEPExprArg.cpp - Promote args if they come from GEPs  -------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Identify GEPs used as arguments to call sites.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: GEPExprArgs
+  //
+  // Description:
+  //  Implement an LLVM pass that clones functions which are passed GEPs
+  //  as an argument
+  //  
+  //
+  class GEPExprArgs : public ModulePass {
+  public:
+    static char ID;
+    GEPExprArgs() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/Int2PtrCmp.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/Int2PtrCmp.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/Int2PtrCmp.h (added)
+++ poolalloc/trunk/include/assistDS/Int2PtrCmp.h Fri May  6 14:26:18 2011
@@ -0,0 +1,40 @@
+//===-- Int2PtrCmp.cpp - Merge inttoptr/ptrtoint --------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Remove unnecessary inttoptr casts
+// Specially ones used in just compares
+// Most cases derived from InstCombine
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+#include "llvm/Target/TargetData.h"
+
+
+namespace llvm {
+  //
+  // Class: Int2PtrCmp
+  //
+  //
+  class Int2PtrCmp : public ModulePass {
+  private:
+    TargetData * TD;
+  public:
+    static char ID;
+    Int2PtrCmp() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+      AU.addRequired<TargetData>();
+    }
+
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/LoadArgs.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/LoadArgs.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/LoadArgs.h (added)
+++ poolalloc/trunk/include/assistDS/LoadArgs.h Fri May  6 14:26:18 2011
@@ -0,0 +1,36 @@
+//===-- LoadArgs.cpp - Promote args if they came from loads ---------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Identify calls, that are passed arguemtns that are LoadInsts.
+// Pass the original pointer instead. Helps improve some
+// context sensitivity.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: LoadArgs
+  //
+  // Description:
+  //  Implement an LLVM pass that clones functions which are passed loads
+  //  as an argument
+  //  
+  //
+  class LoadArgs : public ModulePass {
+  public:
+    static char ID;
+    LoadArgs() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/MergeGEP.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/MergeGEP.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/MergeGEP.h (added)
+++ poolalloc/trunk/include/assistDS/MergeGEP.h Fri May  6 14:26:18 2011
@@ -0,0 +1,30 @@
+//===-- MergeGEP.cpp - Merge GEPs for indexing in arrays ------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Merge chained GEPs; Specially useful for arrays inside structs
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Target/TargetData.h"
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: MergeArrayGEP
+  //
+  class MergeArrayGEP : public ModulePass {
+  public:
+    static char ID;
+    MergeArrayGEP() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/SimplifyExtractValue.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/SimplifyExtractValue.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/SimplifyExtractValue.h (added)
+++ poolalloc/trunk/include/assistDS/SimplifyExtractValue.h Fri May  6 14:26:18 2011
@@ -0,0 +1,31 @@
+//===-- SimplifyExtractValue.cpp - Remove extraneous extractvalue insts----===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Simplify extractvalue
+//
+// Derived from InstCombine
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: SimplifyEV
+  //
+  class SimplifyEV : public ModulePass {
+  public:
+    static char ID;
+    SimplifyEV() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/SimplifyGEP.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/SimplifyGEP.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/SimplifyGEP.h (added)
+++ poolalloc/trunk/include/assistDS/SimplifyGEP.h Fri May  6 14:26:18 2011
@@ -0,0 +1,35 @@
+//===--------------- SimplifyGEP.cpp - Simplify GEPs types  ---------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Simplify GEPs with bitcasts (mostly cloned from InstCombine)
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Target/TargetData.h"
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: SimplifyGEP
+  //
+  class SimplifyGEP : public ModulePass {
+  private:
+    TargetData * TD;
+  public:
+    static char ID;
+    SimplifyGEP() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
+      AU.addRequired<TargetData>();
+    }
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/SimplifyInsertValue.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/SimplifyInsertValue.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/SimplifyInsertValue.h (added)
+++ poolalloc/trunk/include/assistDS/SimplifyInsertValue.h Fri May  6 14:26:18 2011
@@ -0,0 +1,30 @@
+//===-- SimplifyInsertValue.cpp - Remove extraneous insertvalue insts------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Simplify insertvalue
+// Replace insertvalue by storess where possible
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: SimplifyIV
+  //
+  class SimplifyIV : public ModulePass {
+  public:
+    static char ID;
+    SimplifyIV() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/StructReturnToPointer.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/StructReturnToPointer.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/StructReturnToPointer.h (added)
+++ poolalloc/trunk/include/assistDS/StructReturnToPointer.h Fri May  6 14:26:18 2011
@@ -0,0 +1,30 @@
+//===-------- StructReturnToPointer.cpp ------------------------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// For functions that return structures,
+// transform them to return a pointer to the structure instead.
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: StructRet
+  //
+  class StructRet : public ModulePass {
+  public:
+    static char ID;
+    StructRet() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Added: poolalloc/trunk/include/assistDS/VarArgsFunc.h
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/include/assistDS/VarArgsFunc.h?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/include/assistDS/VarArgsFunc.h (added)
+++ poolalloc/trunk/include/assistDS/VarArgsFunc.h Fri May  6 14:26:18 2011
@@ -0,0 +1,32 @@
+//===-- VarArgsFunc.cpp - Simplify calls to bitcasted const funcs --------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+// Convert calls of type
+// call(bitcast F to (...)*) ()
+// to
+// call F()
+// if the number and types of arguments passed matches.
+//===----------------------------------------------------------------------===//
+
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+
+namespace llvm {
+  //
+  // Class: VarArgsFunc
+  //
+  //
+  class VarArgsFunc : public ModulePass {
+  public:
+    static char ID;
+    VarArgsFunc() : ModulePass(&ID) {}
+    virtual bool runOnModule(Module& M);
+  };
+}
+

Modified: poolalloc/trunk/lib/AssistDS/ArgCast.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/ArgCast.cpp?rev=131006&r1=131005&r2=131006&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/ArgCast.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/ArgCast.cpp Fri May  6 14:26:18 2011
@@ -14,10 +14,8 @@
 //===----------------------------------------------------------------------===//
 #define DEBUG_TYPE "argcast"
 
-#include "llvm/Instructions.h"
+#include "assistDS/ArgCast.h"
 #include "llvm/Attributes.h"
-#include "llvm/Module.h"
-#include "llvm/Pass.h"
 #include "llvm/Transforms/Utils/Cloning.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Support/FormattedStream.h"
@@ -32,144 +30,135 @@
 // Pass statistics
 STATISTIC(numChanged,   "Number of Args bitcasted");
 
-namespace {
+//
+// Method: runOnModule()
+//
+// Description:
+//  Entry point for this LLVM pass.
+//  Search for all call sites to casted functions.
+//  Check if they only differ in an argument type
+//  Cast the argument, and call the original function
+//
+// Inputs:
+//  M - A reference to the LLVM module to transform
+//
+// Outputs:
+//  M - The transformed LLVM module.
+//
+// Return value:
+//  true  - The module was modified.
+//  false - The module was not modified.
+//
+bool ArgCast::runOnModule(Module& M) {
+
+  std::vector<CallInst*> worklist;
+  for (Module::iterator I = M.begin(); I != M.end(); ++I)
+    if (!I->isDeclaration() && !I->mayBeOverridden())
+      // Find all uses of this function
+      for(Value::use_iterator ui = I->use_begin(), ue = I->use_end(); ui != ue; ++ui)
+        // check if is ever casted to a different function type
+        if (Constant *C = dyn_cast<Constant>(ui)) 
+          if (ConstantExpr *CE = dyn_cast<ConstantExpr>(C)) 
+            if (CE->getOpcode() == Instruction::BitCast) 
+              if(CE->getOperand(0) == I) 
+                if(const FunctionType *FTy  = dyn_cast<FunctionType>
+                   ((cast<PointerType>(CE->getType()))->getElementType())) {
+                  //casting to a varargs funtion
+                  if(FTy->isVarArg())
+                    for(Value::use_iterator uii = CE->use_begin(),
+                        uee = CE->use_end(); uii != uee; ++uii) {
+                      // Find all uses of the casted value, and check if it is 
+                      // used in a Call Instruction
+                      if (CallInst* CI = dyn_cast<CallInst>(uii)) {
+                        // Check that it is the called value, and not an argument
+                        if(CI->getCalledValue() != CE) 
+                          continue;
+                        // Check that the number of arguments passed, and expected
+                        // by the function are the same.
+                        if(CI->getNumOperands() != I->arg_size() + 1)
+                          continue;
+                        // Check that the return type of the function matches that
+                        // expected by the call inst(ensures that the reason for the
+                        // cast is not the return type).
+                        if(CI->getType() != I->getReturnType())
+                          continue;
 
-  class ArgCast : public ModulePass {
-  public:
-    static char ID;
-    ArgCast() : ModulePass(&ID) {}
-
-    //
-    // Method: runOnModule()
-    //
-    // Description:
-    //  Entry point for this LLVM pass.
-    //  Search for all call sites to casted functions.
-    //  Check if they only differ in an argument type
-    //  Cast the argument, and call the original function
-    //
-    // Inputs:
-    //  M - A reference to the LLVM module to transform
-    //
-    // Outputs:
-    //  M - The transformed LLVM module.
-    //
-    // Return value:
-    //  true  - The module was modified.
-    //  false - The module was not modified.
-    //
-    bool runOnModule(Module& M) {
-
-      std::vector<CallInst*> worklist;
-      for (Module::iterator I = M.begin(); I != M.end(); ++I)
-        if (!I->isDeclaration() && !I->mayBeOverridden())
-          // Find all uses of this function
-          for(Value::use_iterator ui = I->use_begin(), ue = I->use_end(); ui != ue; ++ui)
-            // check if is ever casted to a different function type
-            if (Constant *C = dyn_cast<Constant>(ui)) 
-              if (ConstantExpr *CE = dyn_cast<ConstantExpr>(C)) 
-                if (CE->getOpcode() == Instruction::BitCast) 
-                  if(CE->getOperand(0) == I) 
-                    if(const FunctionType *FTy  = dyn_cast<FunctionType>
-                       ((cast<PointerType>(CE->getType()))->getElementType())) {
-                      //casting to a varargs funtion
-                      if(FTy->isVarArg())
-                        for(Value::use_iterator uii = CE->use_begin(),
-                            uee = CE->use_end(); uii != uee; ++uii) {
-                          // Find all uses of the casted value, and check if it is 
-                          // used in a Call Instruction
-                          if (CallInst* CI = dyn_cast<CallInst>(uii)) {
-                            // Check that it is the called value, and not an argument
-                            if(CI->getCalledValue() != CE) 
-                              continue;
-                            // Check that the number of arguments passed, and expected
-                            // by the function are the same.
-                            if(CI->getNumOperands() != I->arg_size() + 1)
-                              continue;
-                            // Check that the return type of the function matches that
-                            // expected by the call inst(ensures that the reason for the
-                            // cast is not the return type).
-                            if(CI->getType() != I->getReturnType())
-                              continue;
-
-                            // If so, add to worklist
-                            worklist.push_back(CI);
-                          }
-                        }
+                        // If so, add to worklist
+                        worklist.push_back(CI);
                       }
-      
-      // Proces the worklist of potential call sites to transform
-      while(!worklist.empty()) {
-        CallInst *CI = worklist.back();
-        worklist.pop_back();
-        // Get the called Function
-        Function *F = cast<Function>(CI->getCalledValue()->stripPointerCasts());
-        const FunctionType *FTy = F->getFunctionType();
-
-        SmallVector<Value*, 8> Args;
-        unsigned i =0;
-        for(i =0; i< FTy->getNumParams(); ++i) {
-          const Type *ArgType = CI->getOperand(i+1)->getType();
-          const Type *FormalType = FTy->getParamType(i);
-          // If the types for this argument match, just add it to the
-          // parameter list. No cast needs to be inserted.
-          if(ArgType == FormalType) {
-            Args.push_back(CI->getOperand(i+1));
-          }
-          else if(ArgType->isPointerTy() && FormalType->isPointerTy()) {
-            CastInst *CastI = CastInst::CreatePointerCast(CI->getOperand(i+1), 
-                                                          FormalType, "", CI);
+                    }
+                }
+
+  // Proces the worklist of potential call sites to transform
+  while(!worklist.empty()) {
+    CallInst *CI = worklist.back();
+    worklist.pop_back();
+    // Get the called Function
+    Function *F = cast<Function>(CI->getCalledValue()->stripPointerCasts());
+    const FunctionType *FTy = F->getFunctionType();
+
+    SmallVector<Value*, 8> Args;
+    unsigned i =0;
+    for(i =0; i< FTy->getNumParams(); ++i) {
+      const Type *ArgType = CI->getOperand(i+1)->getType();
+      const Type *FormalType = FTy->getParamType(i);
+      // If the types for this argument match, just add it to the
+      // parameter list. No cast needs to be inserted.
+      if(ArgType == FormalType) {
+        Args.push_back(CI->getOperand(i+1));
+      }
+      else if(ArgType->isPointerTy() && FormalType->isPointerTy()) {
+        CastInst *CastI = CastInst::CreatePointerCast(CI->getOperand(i+1), 
+                                                      FormalType, "", CI);
+        Args.push_back(CastI);
+      } else if (ArgType->isIntegerTy() && FormalType->isIntegerTy()) {
+        unsigned SrcBits = ArgType->getScalarSizeInBits();
+        unsigned DstBits = FormalType->getScalarSizeInBits();
+        if(SrcBits > DstBits) {
+          CastInst *CastI = CastInst::CreateIntegerCast(CI->getOperand(i+1), 
+                                                        FormalType, true, "", CI);
+          Args.push_back(CastI);
+        } else {
+          if(F->paramHasAttr(i+1, Attribute::SExt)) {
+            CastInst *CastI = CastInst::CreateIntegerCast(CI->getOperand(i+1), 
+                                                          FormalType, true, "", CI);
+            Args.push_back(CastI);
+          } else if(F->paramHasAttr(i+1, Attribute::ZExt)) {
+            CastInst *CastI = CastInst::CreateIntegerCast(CI->getOperand(i+1), 
+                                                          FormalType, false, "", CI);
             Args.push_back(CastI);
-          } else if (ArgType->isIntegerTy() && FormalType->isIntegerTy()) {
-            unsigned SrcBits = ArgType->getScalarSizeInBits();
-            unsigned DstBits = FormalType->getScalarSizeInBits();
-            if(SrcBits > DstBits) {
-              CastInst *CastI = CastInst::CreateIntegerCast(CI->getOperand(i+1), 
-                                                            FormalType, true, "", CI);
-              Args.push_back(CastI);
-            } else {
-              if(F->paramHasAttr(i+1, Attribute::SExt)) {
-                CastInst *CastI = CastInst::CreateIntegerCast(CI->getOperand(i+1), 
-                                                              FormalType, true, "", CI);
-                Args.push_back(CastI);
-              } else if(F->paramHasAttr(i+1, Attribute::ZExt)) {
-                CastInst *CastI = CastInst::CreateIntegerCast(CI->getOperand(i+1), 
-                                                              FormalType, false, "", CI);
-                Args.push_back(CastI);
-              } else {
-                // Use ZExt in default case.
-                // Derived from InstCombine. Also, the only reason this should happen
-                // is mismatched prototypes.
-                // Seen in case of integer constants which get interpreted as i32, 
-                // even if being used as i64.
-                // TODO: is this correct?
-                CastInst *CastI = CastInst::CreateIntegerCast(CI->getOperand(i+1), 
-                                                              FormalType, false, "", CI);
-                Args.push_back(CastI);
-              } 
-            } 
           } else {
-            DEBUG(ArgType->dump());
-            DEBUG(FormalType->dump());
-            break;
-          }
-        }
-
-        // If we found an argument we could not cast, try the next instruction
-        if(i != FTy->getNumParams())
-          continue;
-
-        // else replace the call instruction
-        CallInst *CINew = CallInst::Create(F, Args.begin(), Args.end(), "", CI);
-        CINew->setCallingConv(CI->getCallingConv());
-        CINew->setAttributes(CI->getAttributes());
-        CI->replaceAllUsesWith(CINew);
-        CI->eraseFromParent();
-        numChanged++;
+            // Use ZExt in default case.
+            // Derived from InstCombine. Also, the only reason this should happen
+            // is mismatched prototypes.
+            // Seen in case of integer constants which get interpreted as i32, 
+            // even if being used as i64.
+            // TODO: is this correct?
+            CastInst *CastI = CastInst::CreateIntegerCast(CI->getOperand(i+1), 
+                                                          FormalType, false, "", CI);
+            Args.push_back(CastI);
+          } 
+        } 
+      } else {
+        DEBUG(ArgType->dump());
+        DEBUG(FormalType->dump());
+        break;
       }
-      return true;
     }
-  };
+
+    // If we found an argument we could not cast, try the next instruction
+    if(i != FTy->getNumParams())
+      continue;
+
+    // else replace the call instruction
+    CallInst *CINew = CallInst::Create(F, Args.begin(), Args.end(), "", CI);
+    CINew->setCallingConv(CI->getCallingConv());
+    CINew->setAttributes(CI->getAttributes());
+    CI->replaceAllUsesWith(CINew);
+    CI->eraseFromParent();
+    numChanged++;
+  }
+  return true;
 }
 
 // Pass ID variable

Modified: poolalloc/trunk/lib/AssistDS/FuncSpec.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/FuncSpec.cpp?rev=131006&r1=131005&r2=131006&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/FuncSpec.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/FuncSpec.cpp Fri May  6 14:26:18 2011
@@ -13,9 +13,8 @@
 //===----------------------------------------------------------------------===//
 #define DEBUG_TYPE "funcspec"
 
-#include "llvm/Instructions.h"
-#include "llvm/Module.h"
-#include "llvm/Pass.h"
+#include "assistDS/FuncSpec.h"
+
 #include "llvm/Transforms/Utils/Cloning.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Support/FormattedStream.h"
@@ -30,95 +29,87 @@
 // Pass statistics
 STATISTIC(numCloned, "Number of Functions Cloned in FuncSpec");
 STATISTIC(numReplaced, "Number of Calls Replaced");
-
-namespace {
-  class FuncSpec : public ModulePass {
-  public:
-    static char ID;
-    FuncSpec() : ModulePass(&ID) {}
-    //
-    // Method: runOnModule()
-    //
-    // Description:
-    //  Entry point for this LLVM pass. Search for call sites, that take functions as arguments
-    //  Clone those functions, and pass the clone.
-    //
-    // Inputs:
-    //  M - A reference to the LLVM module to transform
-    //
-    // Outputs:
-    //  M - The transformed LLVM module.
-    //
-    // Return value:
-    //  true  - The module was modified.
-    //  false - The module was not modified.
-    //
-    bool runOnModule(Module& M) {
-      std::map<CallInst*, std::vector<std::pair<unsigned, Constant*> > > cloneSites;
-      std::map<std::pair<Function*, std::vector<std::pair<unsigned, Constant*> > >, Function* > toClone;
-
-      for (Module::iterator I = M.begin(); I != M.end(); ++I)
-        if (!I->isDeclaration() && !I->mayBeOverridden()) {
-          std::vector<unsigned> FPArgs;
-          for (Function::arg_iterator ii = I->arg_begin(), ee = I->arg_end();
-               ii != ee; ++ii) {
-            // check if this function has a FunctionType(or a pointer to) argument 
-            if (const PointerType* Ty = dyn_cast<PointerType>(ii->getType())) {
-              if (isa<FunctionType>(Ty->getElementType())) {
-                // Store the index of such an argument
-                FPArgs.push_back(ii->getArgNo());
-                DEBUG(errs() << "Eligible: " << I->getNameStr() << "\n");
-              }
-            } else if (isa<FunctionType>(ii->getType())) {
-              // Store the index of such an argument
-              FPArgs.push_back(ii->getArgNo());
-              DEBUG(errs() << "Eligible: " << I->getNameStr() << "\n");
-            } 
+//
+// Method: runOnModule()
+//
+// Description:
+//  Entry point for this LLVM pass. Search for call sites, that take functions as arguments
+//  Clone those functions, and pass the clone.
+//
+// Inputs:
+//  M - A reference to the LLVM module to transform
+//
+// Outputs:
+//  M - The transformed LLVM module.
+//
+// Return value:
+//  true  - The module was modified.
+//  false - The module was not modified.
+//
+bool FuncSpec::runOnModule(Module& M) {
+  std::map<CallInst*, std::vector<std::pair<unsigned, Constant*> > > cloneSites;
+  std::map<std::pair<Function*, std::vector<std::pair<unsigned, Constant*> > >, Function* > toClone;
+
+  for (Module::iterator I = M.begin(); I != M.end(); ++I)
+    if (!I->isDeclaration() && !I->mayBeOverridden()) {
+      std::vector<unsigned> FPArgs;
+      for (Function::arg_iterator ii = I->arg_begin(), ee = I->arg_end();
+           ii != ee; ++ii) {
+        // check if this function has a FunctionType(or a pointer to) argument 
+        if (const PointerType* Ty = dyn_cast<PointerType>(ii->getType())) {
+          if (isa<FunctionType>(Ty->getElementType())) {
+            // Store the index of such an argument
+            FPArgs.push_back(ii->getArgNo());
+            DEBUG(errs() << "Eligible: " << I->getNameStr() << "\n");
           }
-          // Now find all call sites that it is called from
-          for(Value::use_iterator ui = I->use_begin(), ue = I->use_end();
-              ui != ue; ++ui) {
-            if (CallInst* CI = dyn_cast<CallInst>(ui)) {
-              // Check that it is the called value (and not an argument)
-              if(CI->getCalledValue()->stripPointerCasts() == I) {
-                std::vector<std::pair<unsigned, Constant*> > Consts;
-                for (unsigned x = 0; x < FPArgs.size(); ++x)
-                  if (Constant* C = dyn_cast<Constant>(ui->getOperand(FPArgs.at(x) + 1))) {
-                    // If the argument passed, at any of the locations noted earlier
-                    // is a constant function, store the pair
-                    Consts.push_back(std::make_pair(FPArgs.at(x), C));
-                  }
-                if (!Consts.empty()) {
-                  // If at least one of the arguments is a constant function,
-                  // we must clone the function.
-                  cloneSites[CI] = Consts;
-                  toClone[std::make_pair(I, Consts)] = 0;
-                }
+        } else if (isa<FunctionType>(ii->getType())) {
+          // Store the index of such an argument
+          FPArgs.push_back(ii->getArgNo());
+          DEBUG(errs() << "Eligible: " << I->getNameStr() << "\n");
+        } 
+      }
+      // Now find all call sites that it is called from
+      for(Value::use_iterator ui = I->use_begin(), ue = I->use_end();
+          ui != ue; ++ui) {
+        if (CallInst* CI = dyn_cast<CallInst>(ui)) {
+          // Check that it is the called value (and not an argument)
+          if(CI->getCalledValue()->stripPointerCasts() == I) {
+            std::vector<std::pair<unsigned, Constant*> > Consts;
+            for (unsigned x = 0; x < FPArgs.size(); ++x)
+              if (Constant* C = dyn_cast<Constant>(ui->getOperand(FPArgs.at(x) + 1))) {
+                // If the argument passed, at any of the locations noted earlier
+                // is a constant function, store the pair
+                Consts.push_back(std::make_pair(FPArgs.at(x), C));
               }
+            if (!Consts.empty()) {
+              // If at least one of the arguments is a constant function,
+              // we must clone the function.
+              cloneSites[CI] = Consts;
+              toClone[std::make_pair(I, Consts)] = 0;
             }
           }
         }
-
-      numCloned += toClone.size();
-
-      for (std::map<std::pair<Function*, std::vector<std::pair<unsigned, Constant*> > >, Function* >::iterator I = toClone.begin(), E = toClone.end(); I != E; ++I) {
-        // Clone all the functions we need cloned
-        Function* DirectF = CloneFunction(I->first.first);
-        DirectF->setName(I->first.first->getNameStr() + "_SPEC");
-        DirectF->setLinkage(GlobalValue::InternalLinkage);
-        I->first.first->getParent()->getFunctionList().push_back(DirectF);
-        I->second = DirectF;
       }
+    }
 
-      for (std::map<CallInst*, std::vector<std::pair<unsigned, Constant*> > >::iterator ii = cloneSites.begin(), ee = cloneSites.end(); ii != ee; ++ii) {
-        // Transform the call sites, to call the clones
-        ii->first->setOperand(0, toClone[std::make_pair(cast<Function>(ii->first->getOperand(0)), ii->second)]);
-        ++numReplaced;
-      }
+  numCloned += toClone.size();
 
-      return true;
-    }
-  };
+  for (std::map<std::pair<Function*, std::vector<std::pair<unsigned, Constant*> > >, Function* >::iterator I = toClone.begin(), E = toClone.end(); I != E; ++I) {
+    // Clone all the functions we need cloned
+    Function* DirectF = CloneFunction(I->first.first);
+    DirectF->setName(I->first.first->getNameStr() + "_SPEC");
+    DirectF->setLinkage(GlobalValue::InternalLinkage);
+    I->first.first->getParent()->getFunctionList().push_back(DirectF);
+    I->second = DirectF;
+  }
+
+  for (std::map<CallInst*, std::vector<std::pair<unsigned, Constant*> > >::iterator ii = cloneSites.begin(), ee = cloneSites.end(); ii != ee; ++ii) {
+    // Transform the call sites, to call the clones
+    ii->first->setOperand(0, toClone[std::make_pair(cast<Function>(ii->first->getOperand(0)), ii->second)]);
+    ++numReplaced;
+  }
+
+  return true;
 }
 
 // Pass ID variable

Removed: poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp?rev=131005&view=auto
==============================================================================
--- poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/GEPExprArg.cpp (removed)
@@ -1,156 +0,0 @@
-//===-- MergeGEP.cpp - Merge GEPs for indexing in arrays ------------ ----===//
-//
-//                     The LLVM Compiler Infrastructure
-//
-// This file is distributed under the University of Illinois Open Source
-// License. See LICENSE.TXT for details.
-//
-//===----------------------------------------------------------------------===//
-//
-// 
-//===----------------------------------------------------------------------===//
-#define DEBUG_TYPE "gepexprargs"
-
-#include "llvm/Instructions.h"
-#include "llvm/Module.h"
-#include "llvm/Pass.h"
-#include "llvm/Instructions.h"
-#include "llvm/Constants.h"
-#include "llvm/Support/GetElementPtrTypeIterator.h"
-#include "llvm/Transforms/Utils/Cloning.h"
-#include "llvm/ADT/Statistic.h"
-#include "llvm/Support/FormattedStream.h"
-#include "llvm/Support/Debug.h"
-#include "llvm/Use.h"
-#include <vector>
-#include <map>
-
-using namespace llvm;
-STATISTIC(numSimplified, "Number of Calls Modified");
-
-
-namespace {
-  class GEPExprArg : public ModulePass {
-  public:
-    static char ID;
-    GEPExprArg() : ModulePass(&ID) {}
-    bool runOnModule(Module& M) {
-      bool changed;
-      do {
-        changed = false;
-        for (Module::iterator F = M.begin(); F != M.end(); ++F){
-          for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {
-            for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
-              CallInst *CI = dyn_cast<CallInst>(I++);
-              if(!CI)
-                continue;
-
-              if(CI->hasByValArgument())
-                continue;
-              // if the GEP calls a function, that is externally defined,
-              // or might be changed, ignore this call site.
-              Function *F = CI->getCalledFunction();
-
-              if (!F || (F->isDeclaration() || F->mayBeOverridden())) 
-                continue;
-              if(F->hasStructRetAttr())
-                continue;
-              if(F->isVarArg())
-                continue;
-
-              // find the argument we must replace
-              Function::arg_iterator ai = F->arg_begin(), ae = F->arg_end();
-              unsigned argNum = 1;
-              for(; argNum < CI->getNumOperands();argNum++, ++ai) {
-                if(ai->use_empty())
-                  continue;
-                if (isa<GEPOperator>(CI->getOperand(argNum)))
-                  break;
-              }
-
-              // if no argument was a GEP operator to be changed 
-              if(ai == ae)
-                continue;
-
-              GEPOperator *GEP = dyn_cast<GEPOperator>(CI->getOperand(argNum));
-              if(!GEP->hasAllConstantIndices())
-                continue;
-
-              // Construct the new Type
-              // Appends the struct Type at the beginning
-              std::vector<const Type*>TP;
-              TP.push_back(GEP->getPointerOperand()->getType());
-              for(unsigned c = 1; c < CI->getNumOperands();c++) {
-                TP.push_back(CI->getOperand(c)->getType());
-              }
-
-              //return type is same as that of original instruction
-              const FunctionType *NewFTy = FunctionType::get(CI->getType(), TP, false);
-              Function *NewF;
-              numSimplified++;
-              if(numSimplified > 800) 
-                return true;
-
-              NewF = Function::Create(NewFTy,
-                                      GlobalValue::InternalLinkage,
-                                      F->getNameStr() + ".TEST",
-                                      &M);
-
-              Function::arg_iterator NI = NewF->arg_begin();
-              NI->setName("Sarg");
-              ++NI;
-
-              DenseMap<const Value*, Value*> ValueMap;
-
-              for (Function::arg_iterator II = F->arg_begin(); NI != NewF->arg_end(); ++II, ++NI) {
-                ValueMap[II] = NI;
-                NI->setName(II->getName());
-              }
-              // Perform the cloning.
-              SmallVector<ReturnInst*,100> Returns;
-              CloneFunctionInto(NewF, F, ValueMap, Returns);
-              std::vector<Value*> fargs;
-              for(Function::arg_iterator ai = NewF->arg_begin(), 
-                  ae= NewF->arg_end(); ai != ae; ++ai) {
-                fargs.push_back(ai);
-              }
-
-              NewF->setAlignment(F->getAlignment());
-              //Get the point to insert the GEP instr.
-              NI = NewF->arg_begin();
-              SmallVector<Value*, 8> Ops(CI->op_begin()+1, CI->op_end());
-              Instruction *InsertPoint;
-              for (BasicBlock::iterator insrt = NewF->front().begin(); isa<AllocaInst>(InsertPoint = insrt); ++insrt) {;}
-
-              SmallVector<Value*, 8> Indices;
-              Indices.append(GEP->op_begin()+1, GEP->op_end());
-              GetElementPtrInst *GEP_new = GetElementPtrInst::Create(cast<Value>(NI), Indices.begin(), Indices.end(), "", InsertPoint);
-              fargs.at(argNum)->replaceAllUsesWith(GEP_new);
-              unsigned j = argNum + 1;
-              for(; j < CI->getNumOperands();j++) {
-                if(CI->getOperand(j) == GEP)
-                  fargs.at(j)->replaceAllUsesWith(GEP_new);
-              }
-
-              SmallVector<Value*, 8> Args;
-              Args.push_back(GEP->getPointerOperand());
-              for(unsigned j =1;j<CI->getNumOperands();j++) {
-                Args.push_back(CI->getOperand(j));
-              }
-              CallInst *CallI = CallInst::Create(NewF,Args.begin(), Args.end(),"", CI);
-              CallI->setCallingConv(CI->getCallingConv());
-              CI->replaceAllUsesWith(CallI);
-              CI->eraseFromParent();
-              changed = true;
-            }
-          }
-        }
-      } while(changed);
-      return true;
-    }
-  };
-}
-
-char GEPExprArg::ID = 0;
-static RegisterPass<GEPExprArg>
-X("gep-expr-arg", "Find GEP Exprs passed as args");

Added: poolalloc/trunk/lib/AssistDS/GEPExprArgs.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/GEPExprArgs.cpp?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/lib/AssistDS/GEPExprArgs.cpp (added)
+++ poolalloc/trunk/lib/AssistDS/GEPExprArgs.cpp Fri May  6 14:26:18 2011
@@ -0,0 +1,166 @@
+//===-- GEPExprArg.cpp - Promote args if they come from GEPs  -------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+//
+// Identify GEPs used as arguments to call sites.
+// 
+//===----------------------------------------------------------------------===//
+#define DEBUG_TYPE "gepexprargs"
+
+#include "assistDS/GEPExprArgs.h"
+#include "llvm/Constants.h"
+#include "llvm/Support/GetElementPtrTypeIterator.h"
+#include "llvm/Transforms/Utils/Cloning.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/Support/FormattedStream.h"
+#include "llvm/Support/Debug.h"
+#include "llvm/Use.h"
+#include <vector>
+#include <map>
+
+// Pass statistics
+STATISTIC(numSimplified, "Number of Calls Modified");
+
+using namespace llvm;
+
+//
+// Method: runOnModule()
+//
+// Description:
+//  Entry point for this LLVM pass.
+//  Clone functions that take GEPs as arguments
+//
+// Inputs:
+//  M - A reference to the LLVM module to transform
+//
+// Outputs:
+//  M - The transformed LLVM module.
+//
+// Return value:
+//  true  - The module was modified.
+//  false - The module was not modified.
+//
+bool GEPExprArgs::runOnModule(Module& M) {
+  bool changed;
+  do {
+    changed = false;
+    for (Module::iterator F = M.begin(); F != M.end(); ++F){
+      for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {
+        for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
+          CallInst *CI = dyn_cast<CallInst>(I++);
+          if(!CI)
+            continue;
+
+          if(CI->hasByValArgument())
+            continue;
+          // if the GEP calls a function, that is externally defined,
+          // or might be changed, ignore this call site.
+          Function *F = CI->getCalledFunction();
+
+          if (!F || (F->isDeclaration() || F->mayBeOverridden())) 
+            continue;
+          if(F->hasStructRetAttr())
+            continue;
+          if(F->isVarArg())
+            continue;
+
+          // find the argument we must replace
+          Function::arg_iterator ai = F->arg_begin(), ae = F->arg_end();
+          unsigned argNum = 1;
+          for(; argNum < CI->getNumOperands();argNum++, ++ai) {
+            if(ai->use_empty())
+              continue;
+            if (isa<GEPOperator>(CI->getOperand(argNum)))
+              break;
+          }
+
+          // if no argument was a GEP operator to be changed 
+          if(ai == ae)
+            continue;
+
+          GEPOperator *GEP = dyn_cast<GEPOperator>(CI->getOperand(argNum));
+          if(!GEP->hasAllConstantIndices())
+            continue;
+
+          // Construct the new Type
+          // Appends the struct Type at the beginning
+          std::vector<const Type*>TP;
+          TP.push_back(GEP->getPointerOperand()->getType());
+          for(unsigned c = 1; c < CI->getNumOperands();c++) {
+            TP.push_back(CI->getOperand(c)->getType());
+          }
+
+          //return type is same as that of original instruction
+          const FunctionType *NewFTy = FunctionType::get(CI->getType(), TP, false);
+          Function *NewF;
+          numSimplified++;
+          if(numSimplified > 800) 
+            return true;
+
+          NewF = Function::Create(NewFTy,
+                                  GlobalValue::InternalLinkage,
+                                  F->getNameStr() + ".TEST",
+                                  &M);
+
+          Function::arg_iterator NI = NewF->arg_begin();
+          NI->setName("Sarg");
+          ++NI;
+
+          DenseMap<const Value*, Value*> ValueMap;
+
+          for (Function::arg_iterator II = F->arg_begin(); NI != NewF->arg_end(); ++II, ++NI) {
+            ValueMap[II] = NI;
+            NI->setName(II->getName());
+          }
+          // Perform the cloning.
+          SmallVector<ReturnInst*,100> Returns;
+          CloneFunctionInto(NewF, F, ValueMap, Returns);
+          std::vector<Value*> fargs;
+          for(Function::arg_iterator ai = NewF->arg_begin(), 
+              ae= NewF->arg_end(); ai != ae; ++ai) {
+            fargs.push_back(ai);
+          }
+
+          NewF->setAlignment(F->getAlignment());
+          //Get the point to insert the GEP instr.
+          NI = NewF->arg_begin();
+          SmallVector<Value*, 8> Ops(CI->op_begin()+1, CI->op_end());
+          Instruction *InsertPoint;
+          for (BasicBlock::iterator insrt = NewF->front().begin(); isa<AllocaInst>(InsertPoint = insrt); ++insrt) {;}
+
+          SmallVector<Value*, 8> Indices;
+          Indices.append(GEP->op_begin()+1, GEP->op_end());
+          GetElementPtrInst *GEP_new = GetElementPtrInst::Create(cast<Value>(NI), Indices.begin(), Indices.end(), "", InsertPoint);
+          fargs.at(argNum)->replaceAllUsesWith(GEP_new);
+          unsigned j = argNum + 1;
+          for(; j < CI->getNumOperands();j++) {
+            if(CI->getOperand(j) == GEP)
+              fargs.at(j)->replaceAllUsesWith(GEP_new);
+          }
+
+          SmallVector<Value*, 8> Args;
+          Args.push_back(GEP->getPointerOperand());
+          for(unsigned j =1;j<CI->getNumOperands();j++) {
+            Args.push_back(CI->getOperand(j));
+          }
+          CallInst *CallI = CallInst::Create(NewF,Args.begin(), Args.end(),"", CI);
+          CallI->setCallingConv(CI->getCallingConv());
+          CI->replaceAllUsesWith(CallI);
+          CI->eraseFromParent();
+          changed = true;
+        }
+      }
+    }
+  } while(changed);
+  return true;
+}
+
+
+char GEPExprArgs::ID = 0;
+static RegisterPass<GEPExprArgs>
+X("gep-expr-arg", "Find GEP Exprs passed as args");

Modified: poolalloc/trunk/lib/AssistDS/Int2PtrCmp.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/Int2PtrCmp.cpp?rev=131006&r1=131005&r2=131006&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/Int2PtrCmp.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/Int2PtrCmp.cpp Fri May  6 14:26:18 2011
@@ -6,6 +6,7 @@
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
+//
 // Remove unnecessary inttoptr casts
 // Specially ones used in just compares
 // Most cases derived from InstCombine
@@ -13,15 +14,11 @@
 //===----------------------------------------------------------------------===//
 #define DEBUG_TYPE "int2ptr-cmp"
 
-#include "llvm/Instructions.h"
-#include "llvm/Module.h"
-#include "llvm/Pass.h"
-#include "llvm/Transforms/Utils/Cloning.h"
+#include "assistDS/Int2PtrCmp.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Support/FormattedStream.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/PatternMatch.h"
-#include "llvm/Target/TargetData.h"
 
 #include <set>
 #include <map>
@@ -30,71 +27,61 @@
 using namespace llvm;
 using namespace PatternMatch;
 
-namespace {
-  class Int2PtrCmp : public ModulePass {
-  private:
-    TargetData * TD;
-  public:
-    static char ID;
-    Int2PtrCmp() : ModulePass(&ID) {}
-
-    //
-    // Method: runOnModule()
-    //
-    // Description:
-    //  Entry point for this LLVM pass.
-    //  Remove unnecessary inttoptr instructions.
-    //
-    // Inputs:
-    //  M - A reference to the LLVM module to transform
-    //
-    // Outputs:
-    //  M - The transformed LLVM module.
-    //
-    // Return value:
-    //  true  - The module was modified.
-    //  false - The module was not modified.
-    //
-    bool runOnModule(Module& M) {
-      TD = &getAnalysis<TargetData>();
-      for (Module::iterator F = M.begin(); F != M.end(); ++F) {
-        for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {      
-          for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
-            // ptrtoint(inttoptr ty Y) to ty -> Y 
-            if(PtrToIntInst *P2I = dyn_cast<PtrToIntInst>(I++)) {
-              if(IntToPtrInst *I2P = dyn_cast<IntToPtrInst>(P2I->getOperand(0))) {
-                if(I2P->getSrcTy() == P2I->getDestTy()){
-                  P2I->replaceAllUsesWith(I2P->getOperand(0));
-                  P2I->eraseFromParent();
-                  if(I2P->use_empty()) {
-                    // If this is the only use of the cast delete it.
-                    I2P->eraseFromParent();
-                  }
-                }
+//
+// Method: runOnModule()
+//
+// Description:
+//  Entry point for this LLVM pass.
+//  Remove unnecessary inttoptr instructions.
+//
+// Inputs:
+//  M - A reference to the LLVM module to transform
+//
+// Outputs:
+//  M - The transformed LLVM module.
+//
+// Return value:
+//  true  - The module was modified.
+//  false - The module was not modified.
+//
+bool Int2PtrCmp::runOnModule(Module& M) {
+  TD = &getAnalysis<TargetData>();
+  for (Module::iterator F = M.begin(); F != M.end(); ++F) {
+    for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {      
+      for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
+        // ptrtoint(inttoptr ty Y) to ty -> Y 
+        if(PtrToIntInst *P2I = dyn_cast<PtrToIntInst>(I++)) {
+          if(IntToPtrInst *I2P = dyn_cast<IntToPtrInst>(P2I->getOperand(0))) {
+            if(I2P->getSrcTy() == P2I->getDestTy()){
+              P2I->replaceAllUsesWith(I2P->getOperand(0));
+              P2I->eraseFromParent();
+              if(I2P->use_empty()) {
+                // If this is the only use of the cast delete it.
+                I2P->eraseFromParent();
               }
             }
           }
-          for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
-            //icmp pred inttoptr(X), null  -> icmp pred X 0
-            if(ICmpInst *CI = dyn_cast<ICmpInst>(I++)) {
-              Value *Op0 = CI->getOperand(0);
-              Value *Op1 = CI->getOperand(1);
-              if (Constant *RHSC = dyn_cast<Constant>(Op1)) {
-                if (Instruction *LHSI = dyn_cast<Instruction>(Op0)){
-                  if(LHSI->getOpcode() == Instruction::IntToPtr) {
-                    if (RHSC->isNullValue() && TD &&
-                        TD->getIntPtrType(RHSC->getContext()) ==
-                        LHSI->getOperand(0)->getType()){
-                      ICmpInst *CI_new = new ICmpInst(CI, CI->getPredicate(), LHSI->getOperand(0),
-                                                      Constant::getNullValue(LHSI->getOperand(0)->getType()));
-
-                      CI->replaceAllUsesWith(CI_new);
-                      CI->eraseFromParent();
-                      if(LHSI->use_empty()) {
-                        // If this is the only use of the cast delete it.
-                        LHSI->eraseFromParent();
-                      }
-                    }
+        }
+      }
+      for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
+        //icmp pred inttoptr(X), null  -> icmp pred X 0
+        if(ICmpInst *CI = dyn_cast<ICmpInst>(I++)) {
+          Value *Op0 = CI->getOperand(0);
+          Value *Op1 = CI->getOperand(1);
+          if (Constant *RHSC = dyn_cast<Constant>(Op1)) {
+            if (Instruction *LHSI = dyn_cast<Instruction>(Op0)){
+              if(LHSI->getOpcode() == Instruction::IntToPtr) {
+                if (RHSC->isNullValue() && TD &&
+                    TD->getIntPtrType(RHSC->getContext()) ==
+                    LHSI->getOperand(0)->getType()){
+                  ICmpInst *CI_new = new ICmpInst(CI, CI->getPredicate(), LHSI->getOperand(0),
+                                                  Constant::getNullValue(LHSI->getOperand(0)->getType()));
+
+                  CI->replaceAllUsesWith(CI_new);
+                  CI->eraseFromParent();
+                  if(LHSI->use_empty()) {
+                    // If this is the only use of the cast delete it.
+                    LHSI->eraseFromParent();
                   }
                 }
               }
@@ -102,94 +89,91 @@
           }
         }
       }
+    }
+  }
 
-      for (Module::iterator F = M.begin(); F != M.end(); ++F) {
-        for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {      
-          for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
-            if(ICmpInst *ICI = dyn_cast<ICmpInst>(I++)) {
-              Value *Op0 = ICI->getOperand(0);
-              Value *Op1 = ICI->getOperand(1);
-              if (ConstantInt *CI = dyn_cast<ConstantInt>(Op1)) {
-                // Since the RHS is a ConstantInt (CI), if the left hand side is an
-                // instruction, see if that instruction also has constants so that the
-                // instruction can be folded into the icmp
-                if (Instruction *LHSI = dyn_cast<Instruction>(Op0)){
-                  if(LHSI->getOpcode() == Instruction::Or) {
-                    if (!ICI->isEquality() || !CI->isNullValue() || !LHSI->hasOneUse())
-                      break;
-                    Value *P, *Q, *R;
-                    if (match(LHSI, m_Or(m_PtrToInt(m_Value(P)), m_PtrToInt(m_Value(Q))))) {
-                      // Simplify icmp eq (or (ptrtoint P), (ptrtoint Q)), 0
-                      // -> and (icmp eq P, null), (icmp eq Q, null).
-                      Value *ICIP = new ICmpInst(ICI, ICI->getPredicate(), P,
-                                                 Constant::getNullValue(P->getType()));
-                      Value *ICIQ = new ICmpInst(ICI, ICI->getPredicate(), Q,
-                                                 Constant::getNullValue(Q->getType()));
-                      Instruction *Op;
-                      if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
-                        Op = BinaryOperator::CreateAnd(ICIP, ICIQ,"",ICI);
-                      else
-                        Op = BinaryOperator::CreateOr(ICIP, ICIQ, "", ICI);
-                      ICI->replaceAllUsesWith(Op);
-
-                    } else if(match(LHSI, m_Or(m_Or(m_PtrToInt(m_Value(P)), 
-                                                    m_PtrToInt(m_Value(Q))), 
-                                               m_PtrToInt(m_Value(R))))) {
-                      // Simplify icmp eq (or (or (ptrtoint P), (ptrtoint Q)), ptrtoint(R)), 0
-                      // -> and (and (icmp eq P, null), (icmp eq Q, null)), (icmp eq R, null).
-                      Value *ICIP = new ICmpInst(ICI, ICI->getPredicate(), P,
-                                                 Constant::getNullValue(P->getType()));
-                      Value *ICIQ = new ICmpInst(ICI, ICI->getPredicate(), Q,
-                                                 Constant::getNullValue(Q->getType()));
-                      Value *ICIR = new ICmpInst(ICI, ICI->getPredicate(), R,
-                                                 Constant::getNullValue(R->getType()));
-                      Instruction *Op;
-                      if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
-                        Op = BinaryOperator::CreateAnd(ICIP, ICIQ,"",ICI);
-                      else
-                        Op = BinaryOperator::CreateOr(ICIP, ICIQ, "", ICI);
-
-                      if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
-                        Op = BinaryOperator::CreateAnd(Op, ICIR,"",ICI);
-                      else
-                        Op = BinaryOperator::CreateOr(Op, ICIR, "", ICI);
-                      ICI->replaceAllUsesWith(Op);
-
-                    } else if(match(LHSI, m_Or(m_PtrToInt(m_Value(Q)), m_Or(
-                            m_PtrToInt(m_Value(P)), m_PtrToInt(m_Value(R)))))) {
-                      // Simplify icmp eq (or  (ptrtoint P), or((ptrtoint Q), ptrtoint(R))), 0
-                      // -> and (icmp eq P, null), (and (icmp eq Q, null), (icmp eq R, null)).
-                      Value *ICIP = new ICmpInst(ICI, ICI->getPredicate(), P,
-                                                 Constant::getNullValue(P->getType()));
-                      Value *ICIQ = new ICmpInst(ICI, ICI->getPredicate(), Q,
-                                                 Constant::getNullValue(Q->getType()));
-                      Value *ICIR = new ICmpInst(ICI, ICI->getPredicate(), R,
-                                                 Constant::getNullValue(R->getType()));
-                      Instruction *Op;
-                      if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
-                        Op = BinaryOperator::CreateAnd(ICIP, ICIQ,"",ICI);
-                      else
-                        Op = BinaryOperator::CreateOr(ICIP, ICIQ, "", ICI);
-
-                      if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
-                        Op = BinaryOperator::CreateAnd(Op, ICIR,"",ICI);
-                      else
-                        Op = BinaryOperator::CreateOr(Op, ICIR, "", ICI);
-                      ICI->replaceAllUsesWith(Op);
-                    }
-                  }
+  for (Module::iterator F = M.begin(); F != M.end(); ++F) {
+    for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {      
+      for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
+        if(ICmpInst *ICI = dyn_cast<ICmpInst>(I++)) {
+          Value *Op0 = ICI->getOperand(0);
+          Value *Op1 = ICI->getOperand(1);
+          if (ConstantInt *CI = dyn_cast<ConstantInt>(Op1)) {
+            // Since the RHS is a ConstantInt (CI), if the left hand side is an
+            // instruction, see if that instruction also has constants so that the
+            // instruction can be folded into the icmp
+            if (Instruction *LHSI = dyn_cast<Instruction>(Op0)){
+              if(LHSI->getOpcode() == Instruction::Or) {
+                if (!ICI->isEquality() || !CI->isNullValue() || !LHSI->hasOneUse())
+                  break;
+                Value *P, *Q, *R;
+                if (match(LHSI, m_Or(m_PtrToInt(m_Value(P)), m_PtrToInt(m_Value(Q))))) {
+                  // Simplify icmp eq (or (ptrtoint P), (ptrtoint Q)), 0
+                  // -> and (icmp eq P, null), (icmp eq Q, null).
+                  Value *ICIP = new ICmpInst(ICI, ICI->getPredicate(), P,
+                                             Constant::getNullValue(P->getType()));
+                  Value *ICIQ = new ICmpInst(ICI, ICI->getPredicate(), Q,
+                                             Constant::getNullValue(Q->getType()));
+                  Instruction *Op;
+                  if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
+                    Op = BinaryOperator::CreateAnd(ICIP, ICIQ,"",ICI);
+                  else
+                    Op = BinaryOperator::CreateOr(ICIP, ICIQ, "", ICI);
+                  ICI->replaceAllUsesWith(Op);
+
+                } else if(match(LHSI, m_Or(m_Or(m_PtrToInt(m_Value(P)), 
+                                                m_PtrToInt(m_Value(Q))), 
+                                           m_PtrToInt(m_Value(R))))) {
+                  // Simplify icmp eq (or (or (ptrtoint P), (ptrtoint Q)), ptrtoint(R)), 0
+                  // -> and (and (icmp eq P, null), (icmp eq Q, null)), (icmp eq R, null).
+                  Value *ICIP = new ICmpInst(ICI, ICI->getPredicate(), P,
+                                             Constant::getNullValue(P->getType()));
+                  Value *ICIQ = new ICmpInst(ICI, ICI->getPredicate(), Q,
+                                             Constant::getNullValue(Q->getType()));
+                  Value *ICIR = new ICmpInst(ICI, ICI->getPredicate(), R,
+                                             Constant::getNullValue(R->getType()));
+                  Instruction *Op;
+                  if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
+                    Op = BinaryOperator::CreateAnd(ICIP, ICIQ,"",ICI);
+                  else
+                    Op = BinaryOperator::CreateOr(ICIP, ICIQ, "", ICI);
+
+                  if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
+                    Op = BinaryOperator::CreateAnd(Op, ICIR,"",ICI);
+                  else
+                    Op = BinaryOperator::CreateOr(Op, ICIR, "", ICI);
+                  ICI->replaceAllUsesWith(Op);
+
+                } else if(match(LHSI, m_Or(m_PtrToInt(m_Value(Q)), m_Or(
+                        m_PtrToInt(m_Value(P)), m_PtrToInt(m_Value(R)))))) {
+                  // Simplify icmp eq (or  (ptrtoint P), or((ptrtoint Q), ptrtoint(R))), 0
+                  // -> and (icmp eq P, null), (and (icmp eq Q, null), (icmp eq R, null)).
+                  Value *ICIP = new ICmpInst(ICI, ICI->getPredicate(), P,
+                                             Constant::getNullValue(P->getType()));
+                  Value *ICIQ = new ICmpInst(ICI, ICI->getPredicate(), Q,
+                                             Constant::getNullValue(Q->getType()));
+                  Value *ICIR = new ICmpInst(ICI, ICI->getPredicate(), R,
+                                             Constant::getNullValue(R->getType()));
+                  Instruction *Op;
+                  if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
+                    Op = BinaryOperator::CreateAnd(ICIP, ICIQ,"",ICI);
+                  else
+                    Op = BinaryOperator::CreateOr(ICIP, ICIQ, "", ICI);
+
+                  if (ICI->getPredicate() == ICmpInst::ICMP_EQ)
+                    Op = BinaryOperator::CreateAnd(Op, ICIR,"",ICI);
+                  else
+                    Op = BinaryOperator::CreateOr(Op, ICIR, "", ICI);
+                  ICI->replaceAllUsesWith(Op);
                 }
               }
             }
           }
         }
       }
-      return true;
-    }
-    virtual void getAnalysisUsage(AnalysisUsage &AU) const {
-      AU.addRequired<TargetData>();
     }
-  };
+  }
+  return true;
 }
 
 // Pass ID variable

Modified: poolalloc/trunk/lib/AssistDS/LoadArgs.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/LoadArgs.cpp?rev=131006&r1=131005&r2=131006&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/LoadArgs.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/LoadArgs.cpp Fri May  6 14:26:18 2011
@@ -1,4 +1,4 @@
-//===-- MergeGEP.cpp - Merge GEPs for indexing in arrays ------------ ----===//
+//===-- LoadArgs.cpp - Promote args if they came from loads ---------------===//
 //
 //                     The LLVM Compiler Infrastructure
 //
@@ -7,14 +7,14 @@
 //
 //===----------------------------------------------------------------------===//
 //
+// Identify calls, that are passed arguments that are LoadInsts.
+// Pass the original pointer instead. Helps improve some
+// context sensitivity.
 // 
 //===----------------------------------------------------------------------===//
 #define DEBUG_TYPE "ld-args"
 
-#include "llvm/Instructions.h"
-#include "llvm/Module.h"
-#include "llvm/Pass.h"
-#include "llvm/Instructions.h"
+#include "assistDS/LoadArgs.h"
 #include "llvm/Constants.h"
 #include "llvm/Support/GetElementPtrTypeIterator.h"
 #include "llvm/Transforms/Utils/Cloning.h"
@@ -25,142 +25,153 @@
 #include <vector>
 #include <map>
 
-using namespace llvm;
+// Pass statistics
 STATISTIC(numSimplified, "Number of Calls Modified");
 
+using namespace llvm;
+
+//
+// Method: runOnModule()
+//
+// Description:
+//  Entry point for this LLVM pass.
+//  Clone functions that take LoadInsts as arguments
+//
+// Inputs:
+//  M - A reference to the LLVM module to transform
+//
+// Outputs:
+//  M - The transformed LLVM module.
+//
+// Return value:
+//  true  - The module was modified.
+//  false - The module was not modified.
+//
+bool LoadArgs::runOnModule(Module& M) {
+  bool changed;
+  do { 
+    changed = false;
+    for (Module::iterator F = M.begin(); F != M.end(); ++F){
+      for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {
+        for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
+          CallInst *CI = dyn_cast<CallInst>(I++);
+          if(!CI)
+            continue;
+
+          if(CI->hasByValArgument())
+            continue;
+          // if the GEP calls a function, that is externally defined,
+          // or might be changed, ignore this call site.
+          Function *F = CI->getCalledFunction();
+
+          if (!F || (F->isDeclaration() || F->mayBeOverridden())) 
+            continue;
+          if(F->hasStructRetAttr())
+            continue;
+          if(F->isVarArg())
+            continue;
+
+          // find the argument we must replace
+          Function::arg_iterator ai = F->arg_begin(), ae = F->arg_end();
+          unsigned argNum = 1;
+          for(; argNum < CI->getNumOperands();argNum++, ++ai) {
+            if(ai->use_empty())
+              continue;
+            if(F->paramHasAttr(argNum, Attribute::SExt) ||
+               F->paramHasAttr(argNum, Attribute::ZExt)) 
+              continue;
+            if (isa<LoadInst>(CI->getOperand(argNum)))
+              break;
+          }
+
+          // if no argument was a GEP operator to be changed 
+          if(ai == ae)
+            continue;
+
+          LoadInst *LI = dyn_cast<LoadInst>(CI->getOperand(argNum));
+          if(LI->getParent() != CI->getParent())
+            continue;
+          // Also check that there is no store after the load.
+          // TODO: Check if the load/store do not alias.
+          BasicBlock::iterator bii = LI->getParent()->begin();
+          Instruction *BII = bii;
+          while(BII != LI) {
+            ++bii;
+            BII = bii;
+          }
+          while(BII != CI) {
+            if(isa<StoreInst>(BII))
+              break;
+            ++bii;
+            BII = bii;
+          }
+          if(isa<StoreInst>(bii)){
+            continue;
+          }
+
+          // Construct the new Type
+          // Appends the struct Type at the beginning
+          std::vector<const Type*>TP;
+          TP.push_back(LI->getOperand(0)->getType());
+          for(unsigned c = 1; c < CI->getNumOperands();c++) {
+            TP.push_back(CI->getOperand(c)->getType());
+          }
+
+          //return type is same as that of original instruction
+          const FunctionType *NewFTy = FunctionType::get(CI->getType(), TP, false);
+          numSimplified++;
+          if(numSimplified > 400)
+            return true;
+
+          Function *NewF = Function::Create(NewFTy,
+                                            GlobalValue::InternalLinkage,
+                                            F->getNameStr() + ".TEST",
+                                            &M);
+
+          Function::arg_iterator NI = NewF->arg_begin();
+          NI->setName("Sarg");
+          ++NI;
+
+          DenseMap<const Value*, Value*> ValueMap;
+
+          for (Function::arg_iterator II = F->arg_begin(); NI != NewF->arg_end(); ++II, ++NI) {
+            ValueMap[II] = NI;
+            NI->setName(II->getName());
+          }
+          // Perform the cloning.
+          SmallVector<ReturnInst*,100> Returns;
+          CloneFunctionInto(NewF, F, ValueMap, Returns);
+          std::vector<Value*> fargs;
+          for(Function::arg_iterator ai = NewF->arg_begin(), 
+              ae= NewF->arg_end(); ai != ae; ++ai) {
+            fargs.push_back(ai);
+          }
 
-namespace {
-  class LoadArgs : public ModulePass {
-  public:
-    static char ID;
-    LoadArgs() : ModulePass(&ID) {}
-    bool runOnModule(Module& M) {
-      bool changed;
-      do { 
-        changed = false;
-        for (Module::iterator F = M.begin(); F != M.end(); ++F){
-          for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {
-            for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
-              CallInst *CI = dyn_cast<CallInst>(I++);
-              if(!CI)
-                continue;
-
-              if(CI->hasByValArgument())
-                continue;
-              // if the GEP calls a function, that is externally defined,
-              // or might be changed, ignore this call site.
-              Function *F = CI->getCalledFunction();
-
-              if (!F || (F->isDeclaration() || F->mayBeOverridden())) 
-                continue;
-              if(F->hasStructRetAttr())
-                continue;
-              if(F->isVarArg())
-                continue;
-
-              // find the argument we must replace
-              Function::arg_iterator ai = F->arg_begin(), ae = F->arg_end();
-              unsigned argNum = 1;
-              for(; argNum < CI->getNumOperands();argNum++, ++ai) {
-                if(ai->use_empty())
-                  continue;
-                if(F->paramHasAttr(argNum, Attribute::SExt) ||
-                   F->paramHasAttr(argNum, Attribute::ZExt)) 
-                  continue;
-                if (isa<LoadInst>(CI->getOperand(argNum)))
-                  break;
-              }
-
-              // if no argument was a GEP operator to be changed 
-              if(ai == ae)
-                continue;
-
-              LoadInst *LI = dyn_cast<LoadInst>(CI->getOperand(argNum));
-              if(LI->getParent() != CI->getParent())
-                continue;
-              // Also check that there is no store after the load.
-              // TODO: Check if the load/store do not alias.
-              BasicBlock::iterator bii = LI->getParent()->begin();
-              Instruction *BII = bii;
-              while(BII != LI) {
-                ++bii;
-                BII = bii;
-              }
-              while(BII != CI) {
-                if(isa<StoreInst>(BII))
-                  break;
-                ++bii;
-                BII = bii;
-              }
-              if(isa<StoreInst>(bii)){
-                continue;
-              }
-
-              // Construct the new Type
-              // Appends the struct Type at the beginning
-              std::vector<const Type*>TP;
-              TP.push_back(LI->getOperand(0)->getType());
-              for(unsigned c = 1; c < CI->getNumOperands();c++) {
-                TP.push_back(CI->getOperand(c)->getType());
-              }
-
-              //return type is same as that of original instruction
-              const FunctionType *NewFTy = FunctionType::get(CI->getType(), TP, false);
-              numSimplified++;
-              if(numSimplified > 400)
-                return true;
-
-              Function *NewF = Function::Create(NewFTy,
-                                      GlobalValue::InternalLinkage,
-                                      F->getNameStr() + ".TEST",
-                                      &M);
-
-              Function::arg_iterator NI = NewF->arg_begin();
-              NI->setName("Sarg");
-              ++NI;
-
-              DenseMap<const Value*, Value*> ValueMap;
-
-              for (Function::arg_iterator II = F->arg_begin(); NI != NewF->arg_end(); ++II, ++NI) {
-                ValueMap[II] = NI;
-                NI->setName(II->getName());
-              }
-              // Perform the cloning.
-              SmallVector<ReturnInst*,100> Returns;
-              CloneFunctionInto(NewF, F, ValueMap, Returns);
-              std::vector<Value*> fargs;
-              for(Function::arg_iterator ai = NewF->arg_begin(), 
-                  ae= NewF->arg_end(); ai != ae; ++ai) {
-                fargs.push_back(ai);
-              }
-
-              NewF->setAlignment(F->getAlignment());
-              //Get the point to insert the GEP instr.
-              NI = NewF->arg_begin();
-              SmallVector<Value*, 8> Ops(CI->op_begin()+1, CI->op_end());
-              Instruction *InsertPoint;
-              for (BasicBlock::iterator insrt = NewF->front().begin(); isa<AllocaInst>(InsertPoint = insrt); ++insrt) {;}
-
-              LoadInst *LI_new = new LoadInst(cast<Value>(NI), "", InsertPoint);
-              fargs.at(argNum)->replaceAllUsesWith(LI_new);
-
-              SmallVector<Value*, 8> Args;
-              Args.push_back(LI->getOperand(0));
-              for(unsigned j =1;j<CI->getNumOperands();j++) {
-                Args.push_back(CI->getOperand(j));
-              }
-              CallInst *CallI = CallInst::Create(NewF,Args.begin(), Args.end(),"", CI);
-              CallI->setCallingConv(CI->getCallingConv());
-              CI->replaceAllUsesWith(CallI);
-              CI->eraseFromParent();
-              changed = true;
-            }
+          NewF->setAlignment(F->getAlignment());
+          //Get the point to insert the GEP instr.
+          NI = NewF->arg_begin();
+          SmallVector<Value*, 8> Ops(CI->op_begin()+1, CI->op_end());
+          Instruction *InsertPoint;
+          for (BasicBlock::iterator insrt = NewF->front().begin(); isa<AllocaInst>(InsertPoint = insrt); ++insrt) {;}
+
+          LoadInst *LI_new = new LoadInst(cast<Value>(NI), "", InsertPoint);
+          fargs.at(argNum)->replaceAllUsesWith(LI_new);
+
+          SmallVector<Value*, 8> Args;
+          Args.push_back(LI->getOperand(0));
+          for(unsigned j =1;j<CI->getNumOperands();j++) {
+            Args.push_back(CI->getOperand(j));
           }
+          CallInst *CallI = CallInst::Create(NewF,Args.begin(), Args.end(),"", CI);
+          CallI->setCallingConv(CI->getCallingConv());
+          CI->replaceAllUsesWith(CallI);
+          CI->eraseFromParent();
+          changed = true;
         }
-      } while(changed);
-      return true;
+      }
     }
-  };
+  } while(changed);
+  return true;
 }
 
 char LoadArgs::ID = 0;

Removed: poolalloc/trunk/lib/AssistDS/MergeArrayIndexGEP.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/MergeArrayIndexGEP.cpp?rev=131005&view=auto
==============================================================================
--- poolalloc/trunk/lib/AssistDS/MergeArrayIndexGEP.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/MergeArrayIndexGEP.cpp (removed)
@@ -1,164 +0,0 @@
-//===-- MergeArrayIndexGEP.cpp - Merge GEPs for indexing in arrays --------===//
-//
-//                     The LLVM Compiler Infrastructure
-//
-// This file is distributed under the University of Illinois Open Source
-// License. See LICENSE.TXT for details.
-//
-//===----------------------------------------------------------------------===//
-// 
-// Merge chained GEPs; Specially useful for arrays inside structs
-//
-//===----------------------------------------------------------------------===//
-#define DEBUG_TYPE "merge-gep"
-
-#include "llvm/Instructions.h"
-#include "llvm/Module.h"
-#include "llvm/Pass.h"
-#include "llvm/Instructions.h"
-#include "llvm/Constants.h"
-#include "llvm/Support/GetElementPtrTypeIterator.h"
-#include "llvm/Transforms/Utils/Cloning.h"
-#include "llvm/ADT/Statistic.h"
-#include "llvm/Support/FormattedStream.h"
-#include "llvm/Support/Debug.h"
-
-#include <vector>
-// Pass statistics
-STATISTIC(numMerged, "Number of GEPs merged");
-
-using namespace llvm;
-
-namespace {
-  class MergeArrayGEP : public ModulePass {
-  public:
-    static char ID;
-    MergeArrayGEP() : ModulePass(&ID) {}
-    //
-    // Method: runOnModule()
-    //
-    // Description:
-    //  Entry point for this LLVM pass.
-    //  Merge chained GEPs into a single GEP
-    //
-    // Inputs:
-    //  M - A reference to the LLVM module to transform
-    //
-    // Outputs:
-    //  M - The transformed LLVM module.
-    //
-    // Return value:
-    //  true  - The module was modified.
-    //  false - The module was not modified.
-    //
-    bool runOnModule(Module& M) {
-      bool changed;
-      do {
-        changed = false;
-        for (Module::iterator F = M.begin(); F != M.end(); ++F){
-          for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {      
-            for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
-              GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(I++);
-              if(GEP == NULL)
-                continue;
-              simplifyGEP(GEP);
-            }
-          }
-        }
-      } while(changed);
-      return true;
-    }
-    
-    //
-    // Method: simplifyGEP()
-    //
-    // Description:
-    //  Check if this GEP's pointer argument is a GEP(Inst/ConstExpr)
-    //  If so check if we can merge the two GEPs into a single GEP
-    //
-    // Inputs:
-    //  GEP - A pointer to the GEP to simplify
-    //
-    static void simplifyGEP(GetElementPtrInst *GEP) {
-      Value *PtrOp = GEP->getOperand(0);
-      if (GEPOperator *Src = dyn_cast<GEPOperator>(PtrOp)) {
-        // Note that if our source is a gep chain itself that we wait for that
-        // chain to be resolved before we perform this transformation.  This
-        // avoids us creating a TON of code in some cases.
-        //
-        if (GetElementPtrInst *SrcGEP =
-            dyn_cast<GetElementPtrInst>(Src->getOperand(0)))
-          if (SrcGEP->getNumOperands() == 2)
-            return;   // Wait until our source is folded to completion.
-
-        SmallVector<Value*, 8> Indices;
-
-        // Find out whether the last index in the source GEP is a sequential idx.
-        bool EndsWithSequential = false;
-        for (gep_type_iterator I = gep_type_begin(*Src), E = gep_type_end(*Src);
-             I != E; ++I)
-          EndsWithSequential = !(*I)->isStructTy();
-
-        // Can we combine the two pointer arithmetics offsets?
-        if (EndsWithSequential) {
-          // Replace: gep (gep %P, long B), long A, ...
-          // With:    T = long A+B; gep %P, T, ...
-          //
-          Value *Sum;
-          Value *SO1 = Src->getOperand(Src->getNumOperands()-1);
-          Value *GO1 = GEP->getOperand(1);
-          if (SO1 == Constant::getNullValue(SO1->getType())) {
-            Sum = GO1;
-          } else if (GO1 == Constant::getNullValue(GO1->getType())) {
-            Sum = SO1;
-          } else {
-            // If they aren't the same type, then the input hasn't been processed
-            // by the loop above yet (which canonicalizes sequential index types to
-            // intptr_t).  Just avoid transforming this until the input has been
-            // normalized.
-            if (SO1->getType() != GO1->getType())
-              return;
-            Sum = llvm::BinaryOperator::Create(BinaryOperator::Add,
-                                               SO1, GO1, 
-                                               PtrOp->getName()+".sum",GEP);
-          }
-
-          // Update the GEP in place if possible.
-          if (Src->getNumOperands() == 2) {
-            GEP->setOperand(0, Src->getOperand(0));
-            GEP->setOperand(1, Sum);
-            numMerged++;
-            return;
-          }
-          Indices.append(Src->op_begin()+1, Src->op_end()-1);
-          Indices.push_back(Sum);
-          Indices.append(GEP->op_begin()+2, GEP->op_end());
-        } else if (isa<Constant>(GEP->idx_begin()) &&
-                   cast<Constant>(GEP->idx_begin())->isNullValue() &&
-                   Src->getNumOperands() != 1) {
-          // Otherwise we can do the fold if the first index of the GEP is a zero
-          Indices.append(Src->op_begin()+1, Src->op_end());
-          Indices.append(GEP->idx_begin()+1, GEP->idx_end());
-        }
-
-        if (!Indices.empty()){
-          GetElementPtrInst *GEPNew =  (GEP->isInBounds() && Src->isInBounds()) ?
-            GetElementPtrInst::CreateInBounds(Src->getOperand(0), Indices.begin(),
-                                              Indices.end(), GEP->getName(), GEP) :
-            GetElementPtrInst::Create(Src->getOperand(0), Indices.begin(),
-                                      Indices.end(), GEP->getName(), GEP);
-          numMerged++;
-          GEP->replaceAllUsesWith(GEPNew);
-          GEP->eraseFromParent();
-        }
-      }
-    }
-  };
-}
-
-// Pass ID variable
-char MergeArrayGEP::ID = 0;
-
-// Register the pass
-static RegisterPass<MergeArrayGEP>
-X("mergearrgep", "Merge GEPs for arrays indexing");

Added: poolalloc/trunk/lib/AssistDS/MergeGEP.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/MergeGEP.cpp?rev=131006&view=auto
==============================================================================
--- poolalloc/trunk/lib/AssistDS/MergeGEP.cpp (added)
+++ poolalloc/trunk/lib/AssistDS/MergeGEP.cpp Fri May  6 14:26:18 2011
@@ -0,0 +1,159 @@
+//===-- MergeGEP.cpp - Merge GEPs for indexing in arrays ------------------===//
+//
+//                     The LLVM Compiler Infrastructure
+//
+// This file is distributed under the University of Illinois Open Source
+// License. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+// 
+// Merge chained GEPs; Specially useful for arrays inside structs
+//
+//===----------------------------------------------------------------------===//
+#define DEBUG_TYPE "merge-gep"
+
+#include "assistDS/MergeGEP.h"
+#include "llvm/Instructions.h"
+#include "llvm/Module.h"
+#include "llvm/Pass.h"
+#include "llvm/Instructions.h"
+#include "llvm/Constants.h"
+#include "llvm/Support/GetElementPtrTypeIterator.h"
+#include "llvm/Transforms/Utils/Cloning.h"
+#include "llvm/ADT/Statistic.h"
+#include "llvm/Support/FormattedStream.h"
+#include "llvm/Support/Debug.h"
+
+#include <vector>
+// Pass statistics
+STATISTIC(numMerged, "Number of GEPs merged");
+
+using namespace llvm;
+
+static void simplifyGEP(GetElementPtrInst *GEP);
+//
+// Method: runOnModule()
+//
+// Description:
+//  Entry point for this LLVM pass.
+//  Merge chained GEPs into a single GEP
+//
+// Inputs:
+//  M - A reference to the LLVM module to transform
+//
+// Outputs:
+//  M - The transformed LLVM module.
+//
+// Return value:
+//  true  - The module was modified.
+//  false - The module was not modified.
+//
+bool MergeArrayGEP::runOnModule(Module& M) {
+  bool changed;
+  do {
+    changed = false;
+    for (Module::iterator F = M.begin(); F != M.end(); ++F){
+      for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {      
+        for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
+          GetElementPtrInst *GEP = dyn_cast<GetElementPtrInst>(I++);
+          if(GEP == NULL)
+            continue;
+          simplifyGEP(GEP);
+        }
+      }
+    }
+  } while(changed);
+  return true;
+}
+
+//
+// Method: simplifyGEP()
+//
+// Description:
+//  Check if this GEP's pointer argument is a GEP(Inst/ConstExpr)
+//  If so check if we can merge the two GEPs into a single GEP
+//
+// Inputs:
+//  GEP - A pointer to the GEP to simplify
+//
+static void simplifyGEP(GetElementPtrInst *GEP) {
+  Value *PtrOp = GEP->getOperand(0);
+  if (GEPOperator *Src = dyn_cast<GEPOperator>(PtrOp)) {
+    // Note that if our source is a gep chain itself that we wait for that
+    // chain to be resolved before we perform this transformation.  This
+    // avoids us creating a TON of code in some cases.
+    //
+    if (GetElementPtrInst *SrcGEP =
+        dyn_cast<GetElementPtrInst>(Src->getOperand(0)))
+      if (SrcGEP->getNumOperands() == 2)
+        return;   // Wait until our source is folded to completion.
+
+    SmallVector<Value*, 8> Indices;
+
+    // Find out whether the last index in the source GEP is a sequential idx.
+    bool EndsWithSequential = false;
+    for (gep_type_iterator I = gep_type_begin(*Src), E = gep_type_end(*Src);
+         I != E; ++I)
+      EndsWithSequential = !(*I)->isStructTy();
+
+    // Can we combine the two pointer arithmetics offsets?
+    if (EndsWithSequential) {
+      // Replace: gep (gep %P, long B), long A, ...
+      // With:    T = long A+B; gep %P, T, ...
+      //
+      Value *Sum;
+      Value *SO1 = Src->getOperand(Src->getNumOperands()-1);
+      Value *GO1 = GEP->getOperand(1);
+      if (SO1 == Constant::getNullValue(SO1->getType())) {
+        Sum = GO1;
+      } else if (GO1 == Constant::getNullValue(GO1->getType())) {
+        Sum = SO1;
+      } else {
+        // If they aren't the same type, then the input hasn't been processed
+        // by the loop above yet (which canonicalizes sequential index types to
+        // intptr_t).  Just avoid transforming this until the input has been
+        // normalized.
+        if (SO1->getType() != GO1->getType())
+          return;
+        Sum = llvm::BinaryOperator::Create(BinaryOperator::Add,
+                                           SO1, GO1, 
+                                           PtrOp->getName()+".sum",GEP);
+      }
+
+      // Update the GEP in place if possible.
+      if (Src->getNumOperands() == 2) {
+        GEP->setOperand(0, Src->getOperand(0));
+        GEP->setOperand(1, Sum);
+        numMerged++;
+        return;
+      }
+      Indices.append(Src->op_begin()+1, Src->op_end()-1);
+      Indices.push_back(Sum);
+      Indices.append(GEP->op_begin()+2, GEP->op_end());
+    } else if (isa<Constant>(GEP->idx_begin()) &&
+               cast<Constant>(GEP->idx_begin())->isNullValue() &&
+               Src->getNumOperands() != 1) {
+      // Otherwise we can do the fold if the first index of the GEP is a zero
+      Indices.append(Src->op_begin()+1, Src->op_end());
+      Indices.append(GEP->idx_begin()+1, GEP->idx_end());
+    }
+
+    if (!Indices.empty()){
+      GetElementPtrInst *GEPNew =  (GEP->isInBounds() && Src->isInBounds()) ?
+        GetElementPtrInst::CreateInBounds(Src->getOperand(0), Indices.begin(),
+                                          Indices.end(), GEP->getName(), GEP) :
+        GetElementPtrInst::Create(Src->getOperand(0), Indices.begin(),
+                                  Indices.end(), GEP->getName(), GEP);
+      numMerged++;
+      GEP->replaceAllUsesWith(GEPNew);
+      GEP->eraseFromParent();
+    }
+  }
+}
+
+// Pass ID variable
+char MergeArrayGEP::ID = 0;
+
+// Register the pass
+static RegisterPass<MergeArrayGEP>
+X("mergearrgep", "Merge GEPs for arrays indexing");

Modified: poolalloc/trunk/lib/AssistDS/SimplifyExtractValue.cpp
URL: http://llvm.org/viewvc/llvm-project/poolalloc/trunk/lib/AssistDS/SimplifyExtractValue.cpp?rev=131006&r1=131005&r2=131006&view=diff
==============================================================================
--- poolalloc/trunk/lib/AssistDS/SimplifyExtractValue.cpp (original)
+++ poolalloc/trunk/lib/AssistDS/SimplifyExtractValue.cpp Fri May  6 14:26:18 2011
@@ -14,9 +14,7 @@
 //===----------------------------------------------------------------------===//
 #define DEBUG_TYPE "simplifyev"
 
-#include "llvm/Instructions.h"
-#include "llvm/Module.h"
-#include "llvm/Pass.h"
+#include "assistDS/SimplifyExtractValue.h"
 #include "llvm/Transforms/Utils/Cloning.h"
 #include "llvm/ADT/Statistic.h"
 #include "llvm/Support/FormattedStream.h"
@@ -33,42 +31,76 @@
 // Pass statistic
 STATISTIC(numErased, "Number of Instructions Deleted");
 
-namespace {
-  class SimplifyEV : public ModulePass {
-  public:
-    static char ID;
-    SimplifyEV() : ModulePass(&ID) {}
-    //
-    // Method: runOnModule()
-    //
-    // Description:
-    //  Entry point for this LLVM pass. Search for insert/extractvalue instructions
-    //  that can be simplified.
-    //
-    // Inputs:
-    //  M - A reference to the LLVM module to transform.
-    //
-    // Outputs:
-    //  M - The transformed LLVM module.
-    //
-    // Return value:
-    // true  - The module was modified.
-    // false - The module was not modified.
-    //
-    bool runOnModule(Module& M) {
-      // Repeat till no change
-      bool changed;
-      do {
-        changed = false;
-        for (Module::iterator F = M.begin(); F != M.end(); ++F) {
-          for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {      
-            for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
-              ExtractValueInst *EV = dyn_cast<ExtractValueInst>(I++);
-              if(!EV)
-                continue;
-              Value *Agg = EV->getAggregateOperand();
-              if (!EV->hasIndices()) {
-                EV->replaceAllUsesWith(Agg);
+//
+// Method: runOnModule()
+//
+// Description:
+//  Entry point for this LLVM pass. Search for insert/extractvalue instructions
+//  that can be simplified.
+//
+// Inputs:
+//  M - A reference to the LLVM module to transform.
+//
+// Outputs:
+//  M - The transformed LLVM module.
+//
+// Return value:
+// true  - The module was modified.
+// false - The module was not modified.
+//
+bool SimplifyEV::runOnModule(Module& M) {
+  // Repeat till no change
+  bool changed;
+  do {
+    changed = false;
+    for (Module::iterator F = M.begin(); F != M.end(); ++F) {
+      for (Function::iterator B = F->begin(), FE = F->end(); B != FE; ++B) {      
+        for (BasicBlock::iterator I = B->begin(), BE = B->end(); I != BE;) {
+          ExtractValueInst *EV = dyn_cast<ExtractValueInst>(I++);
+          if(!EV)
+            continue;
+          Value *Agg = EV->getAggregateOperand();
+          if (!EV->hasIndices()) {
+            EV->replaceAllUsesWith(Agg);
+            DEBUG(errs() << "EV:");
+            DEBUG(errs() << "ERASE:");
+            DEBUG(EV->dump());
+            EV->eraseFromParent();
+            numErased++;
+            changed = true;
+            continue;
+          }
+          if (Constant *C = dyn_cast<Constant>(Agg)) {
+            if (isa<UndefValue>(C)) {
+              EV->replaceAllUsesWith(UndefValue::get(EV->getType()));
+              DEBUG(errs() << "EV:");
+              DEBUG(errs() << "ERASE:");
+              DEBUG(EV->dump());
+              EV->eraseFromParent();
+              numErased++;
+              changed = true;
+              continue;
+            }
+            if (isa<ConstantAggregateZero>(C)) {
+              EV->replaceAllUsesWith(Constant::getNullValue(EV->getType()));
+              DEBUG(errs() << "EV:");
+              DEBUG(errs() << "ERASE:");
+              DEBUG(EV->dump());
+              EV->eraseFromParent();
+              numErased++;
+              changed = true;
+              continue;
+            }
+            if (isa<ConstantArray>(C) || isa<ConstantStruct>(C)) {
+              // Extract the element indexed by the first index out of the constant
+              Value *V = C->getOperand(*EV->idx_begin());
+              if (EV->getNumIndices() > 1) {
+                // Extract the remaining indices out of the constant indexed by the
+                // first index
+                ExtractValueInst *EV_new = ExtractValueInst::Create(V, 
+                                                                    EV->idx_begin() + 1, 
+                                                                    EV->idx_end(), "", EV);
+                EV->replaceAllUsesWith(EV_new);
                 DEBUG(errs() << "EV:");
                 DEBUG(errs() << "ERASE:");
                 DEBUG(EV->dump());
@@ -76,174 +108,133 @@
                 numErased++;
                 changed = true;
                 continue;
-              }
-              if (Constant *C = dyn_cast<Constant>(Agg)) {
-                if (isa<UndefValue>(C)) {
-                  EV->replaceAllUsesWith(UndefValue::get(EV->getType()));
-                  DEBUG(errs() << "EV:");
-                  DEBUG(errs() << "ERASE:");
-                  DEBUG(EV->dump());
-                  EV->eraseFromParent();
-                  numErased++;
-                  changed = true;
-                  continue;
-                }
-                if (isa<ConstantAggregateZero>(C)) {
-                  EV->replaceAllUsesWith(Constant::getNullValue(EV->getType()));
-                  DEBUG(errs() << "EV:");
-                  DEBUG(errs() << "ERASE:");
-                  DEBUG(EV->dump());
-                  EV->eraseFromParent();
-                  numErased++;
-                  changed = true;
-                  continue;
-                }
-                if (isa<ConstantArray>(C) || isa<ConstantStruct>(C)) {
-                  // Extract the element indexed by the first index out of the constant
-                  Value *V = C->getOperand(*EV->idx_begin());
-                  if (EV->getNumIndices() > 1) {
-                    // Extract the remaining indices out of the constant indexed by the
-                    // first index
-                    ExtractValueInst *EV_new = ExtractValueInst::Create(V, 
-                                                                        EV->idx_begin() + 1, 
-                                                                        EV->idx_end(), "", EV);
-                    EV->replaceAllUsesWith(EV_new);
-                    DEBUG(errs() << "EV:");
-                    DEBUG(errs() << "ERASE:");
-                    DEBUG(EV->dump());
-                    EV->eraseFromParent();
-                    numErased++;
-                    changed = true;
-                    continue;
-                  }  else {
-                    EV->replaceAllUsesWith(V);
-                    DEBUG(errs() << "EV:");
-                    DEBUG(errs() << "ERASE:");
-                    DEBUG(EV->dump());
-                    EV->eraseFromParent();
-                    numErased++;
-                    changed = true;
-                    continue;
-                  }
-                }
-                continue;
-              }
-              if (LoadInst * LI = dyn_cast<LoadInst>(Agg)) {
-                SmallVector<Value*, 8> Indices;
-                const Type *Int32Ty = Type::getInt32Ty(M.getContext());
-                Indices.push_back(Constant::getNullValue(Int32Ty));
-                for (ExtractValueInst::idx_iterator I = EV->idx_begin(), E = EV->idx_end();
-                     I != E; ++I) {
-                  Indices.push_back(ConstantInt::get(Int32Ty, *I));
-                }
-
-                GetElementPtrInst *GEP = GetElementPtrInst::CreateInBounds(LI->getOperand(0), Indices.begin(),
-                                                                           Indices.end(), LI->getName(), LI) ;
-                LoadInst *LINew = new LoadInst(GEP, "", LI);
-                EV->replaceAllUsesWith(LINew);
+              }  else {
+                EV->replaceAllUsesWith(V);
+                DEBUG(errs() << "EV:");
+                DEBUG(errs() << "ERASE:");
+                DEBUG(EV->dump());
                 EV->eraseFromParent();
-                changed = true;
                 numErased++;
+                changed = true;
                 continue;
-
               }
-              if (InsertValueInst *IV = dyn_cast<InsertValueInst>(Agg)) {
-                bool done = false;
-                // We're extracting from an insertvalue instruction, compare the indices
-                const unsigned *exti, *exte, *insi, *inse;
-                for (exti = EV->idx_begin(), insi = IV->idx_begin(),
-                     exte = EV->idx_end(), inse = IV->idx_end();
-                     exti != exte && insi != inse;
-                     ++exti, ++insi) {
-                  if (*insi != *exti) {
-                    // The insert and extract both reference distinctly different elements.
-                    // This means the extract is not influenced by the insert, and we can
-                    // replace the aggregate operand of the extract with the aggregate
-                    // operand of the insert. i.e., replace
-                    // %I = insertvalue { i32, { i32 } } %A, { i32 } { i32 42 }, 1
-                    // %E = extractvalue { i32, { i32 } } %I, 0
-                    // with
-                    // %E = extractvalue { i32, { i32 } } %A, 0
-                    ExtractValueInst *EV_new = ExtractValueInst::Create(IV->getAggregateOperand(),
-                                                                        EV->idx_begin(), EV->idx_end(),"", EV);
-                    EV->replaceAllUsesWith(EV_new);
-                    DEBUG(errs() << "EV:");
-                    DEBUG(errs() << "ERASE:");
-                    DEBUG(EV->dump());
-                    EV->eraseFromParent();
-                    numErased++;
-                    done = true;
-                    changed = true;
-                    break;
-                  }
-                }
-                if(done)
-                  continue;
-                if (exti == exte && insi == inse) {
-                  // Both iterators are at the end: Index lists are identical. Replace
-                  // %B = insertvalue { i32, { i32 } } %A, i32 42, 1, 0
-                  // %C = extractvalue { i32, { i32 } } %B, 1, 0
-                  // with "i32 42"
-                  EV->replaceAllUsesWith(IV->getInsertedValueOperand());
-                  DEBUG(errs() << "EV:");
-                  DEBUG(errs() << "ERASE:");
-                  DEBUG(EV->dump());
-                  EV->eraseFromParent();
-                  numErased++;
-                  changed = true;
-                  continue;
-
-                }
-                if (exti == exte) {
-                  // The extract list is a prefix of the insert list. i.e. replace
-                  // %I = insertvalue { i32, { i32 } } %A, i32 42, 1, 0
-                  // %E = extractvalue { i32, { i32 } } %I, 1
-                  // with
-                  // %X = extractvalue { i32, { i32 } } %A, 1
-                  // %E = insertvalue { i32 } %X, i32 42, 0
-                  // by switching the order of the insert and extract (though the
-                  // insertvalue should be left in, since it may have other uses).
-                  Value *NewEV = ExtractValueInst::Create(IV->getAggregateOperand(),
-                                                          EV->idx_begin(), EV->idx_end(), "", EV);
-                  Value *NewIV = InsertValueInst::Create(NewEV, IV->getInsertedValueOperand(),
-                                                         insi, inse, "", EV);
-                  EV->replaceAllUsesWith(NewIV);
-                  DEBUG(errs() << "EV:");
-                  DEBUG(errs() << "ERASE:");
-                  DEBUG(EV->dump());
-                  EV->eraseFromParent();
-                  numErased++;
-                  changed = true;
-                  continue;
-                }
-                if (insi == inse) {
-                  // The insert list is a prefix of the extract list
-                  // We can simply remove the common indices from the extract and make it
-                  // operate on the inserted value instead of the insertvalue result.
-                  // i.e., replace
-                  // %I = insertvalue { i32, { i32 } } %A, { i32 } { i32 42 }, 1
-                  // %E = extractvalue { i32, { i32 } } %I, 1, 0
-                  // with
-                  // %E extractvalue { i32 } { i32 42 }, 0
-                  ExtractValueInst *EV_new = ExtractValueInst::Create(IV->getInsertedValueOperand(),
-                                                                      exti, exte,"", EV);
-                  EV->replaceAllUsesWith(EV_new);
-                  DEBUG(errs() << "EV:");
-                  DEBUG(errs() << "ERASE:");
-                  DEBUG(EV->dump());
-                  EV->eraseFromParent();
-                  numErased++;
-                  changed = true;
-                  continue;
-                }
+            }
+            continue;
+          }
+          if (LoadInst * LI = dyn_cast<LoadInst>(Agg)) {
+            SmallVector<Value*, 8> Indices;
+            const Type *Int32Ty = Type::getInt32Ty(M.getContext());
+            Indices.push_back(Constant::getNullValue(Int32Ty));
+            for (ExtractValueInst::idx_iterator I = EV->idx_begin(), E = EV->idx_end();
+                 I != E; ++I) {
+              Indices.push_back(ConstantInt::get(Int32Ty, *I));
+            }
+
+            GetElementPtrInst *GEP = GetElementPtrInst::CreateInBounds(LI->getOperand(0), Indices.begin(),
+                                                                       Indices.end(), LI->getName(), LI) ;
+            LoadInst *LINew = new LoadInst(GEP, "", LI);
+            EV->replaceAllUsesWith(LINew);
+            EV->eraseFromParent();
+            changed = true;
+            numErased++;
+            continue;
+
+          }
+          if (InsertValueInst *IV = dyn_cast<InsertValueInst>(Agg)) {
+            bool done = false;
+            // We're extracting from an insertvalue instruction, compare the indices
+            const unsigned *exti, *exte, *insi, *inse;
+            for (exti = EV->idx_begin(), insi = IV->idx_begin(),
+                 exte = EV->idx_end(), inse = IV->idx_end();
+                 exti != exte && insi != inse;
+                 ++exti, ++insi) {
+              if (*insi != *exti) {
+                // The insert and extract both reference distinctly different elements.
+                // This means the extract is not influenced by the insert, and we can
+                // replace the aggregate operand of the extract with the aggregate
+                // operand of the insert. i.e., replace
+                // %I = insertvalue { i32, { i32 } } %A, { i32 } { i32 42 }, 1
+                // %E = extractvalue { i32, { i32 } } %I, 0
+                // with
+                // %E = extractvalue { i32, { i32 } } %A, 0
+                ExtractValueInst *EV_new = ExtractValueInst::Create(IV->getAggregateOperand(),
+                                                                    EV->idx_begin(), EV->idx_end(),"", EV);
+                EV->replaceAllUsesWith(EV_new);
+                DEBUG(errs() << "EV:");
+                DEBUG(errs() << "ERASE:");
+                DEBUG(EV->dump());
+                EV->eraseFromParent();
+                numErased++;
+                done = true;
+                changed = true;
+                break;
               }
             }
+            if(done)
+              continue;
+            if (exti == exte && insi == inse) {
+              // Both iterators are at the end: Index lists are identical. Replace
+              // %B = insertvalue { i32, { i32 } } %A, i32 42, 1, 0
+              // %C = extractvalue { i32, { i32 } } %B, 1, 0
+              // with "i32 42"
+              EV->replaceAllUsesWith(IV->getInsertedValueOperand());
+              DEBUG(errs() << "EV:");
+              DEBUG(errs() << "ERASE:");
+              DEBUG(EV->dump());
+              EV->eraseFromParent();
+              numErased++;
+              changed = true;
+              continue;
+
+            }
+            if (exti == exte) {
+              // The extract list is a prefix of the insert list. i.e. replace
+              // %I = insertvalue { i32, { i32 } } %A, i32 42, 1, 0
+              // %E = extractvalue { i32, { i32 } } %I, 1
+              // with
+              // %X = extractvalue { i32, { i32 } } %A, 1
+              // %E = insertvalue { i32 } %X, i32 42, 0
+              // by switching the order of the insert and extract (though the
+              // insertvalue should be left in, since it may have other uses).
+              Value *NewEV = ExtractValueInst::Create(IV->getAggregateOperand(),
+                                                      EV->idx_begin(), EV->idx_end(), "", EV);
+              Value *NewIV = InsertValueInst::Create(NewEV, IV->getInsertedValueOperand(),
+                                                     insi, inse, "", EV);
+              EV->replaceAllUsesWith(NewIV);
+              DEBUG(errs() << "EV:");
+              DEBUG(errs() << "ERASE:");
+              DEBUG(EV->dump());
+              EV->eraseFromParent();
+              numErased++;
+              changed = true;
+              continue;
+            }
+            if (insi == inse) {
+              // The insert list is a prefix of the extract list
+              // We can simply remove the common indices from the extract and make it
+              // operate on the inserted value instead of the insertvalue result.
+              // i.e., replace
+              // %I = insertvalue { i32, { i32 } } %A, { i32 } { i32 42 }, 1
+              // %E = extractvalue { i32, { i32 } } %I, 1, 0
+              // with
+              // %E extractvalue { i32 } { i32 42 }, 0
+              ExtractValueInst *EV_new = ExtractValueInst::Create(IV->getInsertedValueOperand(),
+                                                                  exti, exte,"", EV);
+              EV->replaceAllUsesWith(EV_new);
+              DEBUG(errs() << "EV:");
+              DEBUG(errs() << "ERASE:");
+              DEBUG(EV->dump());
+              EV->eraseFromParent();
+              numErased++;
+              changed = true;
+              continue;
+            }
           }
         }
-      } while(changed);
-      return (numErased > 0);
+      }
     }
-  };
+  } while(changed);
+  return (numErased > 0);
 }
 
 // Pass ID variable

Modified: poolalloc/trunk/lib/AssistDS/Simpli