From bugzilla-daemon at cs.uiuc.edu  Fri May  1 14:18:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 14:18:48 -0500
Subject: [LLVMbugs] [Bug 4116] New: promotion of bit-field differs from gcc
Message-ID: <bug-4116-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4116

           Summary: promotion of bit-field differs from gcc
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


clang and llvm-gcc differ in the promotion rules for bit-fields. I'm not sure
yet who is correct, filing while I try to figure it out.

--
ddunbar at lordcrumb:tmp$ cat t.c
long long f0(void) {
  struct { unsigned f0 : 16; } x = { 18 };
  struct { unsigned f1 : 16; } y = { 22 };
  return (long long) (x.f0 - y.f1);
}

long long f1(void) {
  struct { unsigned f0 : 16; } x = { 18 };
  struct { unsigned f1     ; } y = { 22 };
  return (long long) (x.f0 - y.f1);
}
ddunbar at lordcrumb:tmp$ llvm-gcc -O3 -emit-llvm -S -o - t.c
; ModuleID = 't.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-f80:128:128"
target triple = "i386-apple-darwin10.0"

define i64 @f0() nounwind readnone ssp {
entry:
        ret i64 -4
}

define i64 @f1() nounwind readnone ssp {
entry:
        ret i64 4294967292
}
ddunbar at lordcrumb:tmp$ clang -O3 -emit-llvm -S -o - t.c
; ModuleID = 't.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-f80:128:128"
target triple = "x86_64-apple-darwin10.0"

define i64 @f0() nounwind readnone {
entry:
        ret i64 4294967292
}

define i64 @f1() nounwind readnone {
entry:
        ret i64 4294967292
}
--


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 15:19:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 15:19:54 -0500
Subject: [LLVMbugs] [Bug 4116] promotion of bit-field is incorrect
In-Reply-To: <bug-4116-206@http.llvm.org/bugs/>
Message-ID: <200905012019.n41KJsOO032016@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4116


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-01 15:19:54 ---


*** This bug has been marked as a duplicate of bug 3500 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 15:20:52 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 15:20:52 -0500
Subject: [LLVMbugs] [Bug 4117] New: ./configure fails when there' s
 whitespace in the working directory
Message-ID: <bug-4117-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4117

           Summary: ./configure fails when there's whitespace in the working
                    directory
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Keywords: portability
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: joachimb at gmail.com
                CC: llvmbugs at cs.uiuc.edu


When calling ./configure from this path:
21:56:39 nevyn:llvm$ pwd
/Users/nevyn/Development/Libraries and sources/C/llvm

./configure fails with:
./config.status: line 828: /Users/nevyn/Development/Libraries: No such file or
directory
/bin/sh: /Users/nevyn/Development/Libraries: No such file or directory
config.status: executing lib/Makefile commands
./config.status: line 830: /Users/nevyn/Development/Libraries: No such file or
directory
/bin/sh: /Users/nevyn/Development/Libraries: No such file or directory
config.status: executing lib/sample/Makefile commands
./config.status: line 832: /Users/nevyn/Development/Libraries: No such file or
directory
/bin/sh: /Users/nevyn/Development/Libraries: No such file or directory
config.status: executing tools/Makefile commands
./config.status: line 834: /Users/nevyn/Development/Libraries: No such file or
directory
/bin/sh: /Users/nevyn/Development/Libraries: No such file or directory
config.status: executing tools/sample/Makefile commands
./config.status: line 836: /Users/nevyn/Development/Libraries: No such file or
directory
/bin/sh: /Users/nevyn/Development/Libraries: No such file or directory


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 16:19:23 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 16:19:23 -0500
Subject: [LLVMbugs] [Bug 3500] Wrong type computed for member access to
 unsigned bitfield
In-Reply-To: <bug-3500-206@http.llvm.org/bugs/>
Message-ID: <200905012119.n41LJN78001576@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3500


Douglas Gregor <dgregor at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dgregor at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Douglas Gregor <dgregor at apple.com>  2009-05-01 16:19:19 ---
Fixed here:

 
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090427/016580.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 17:52:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 17:52:22 -0500
Subject: [LLVMbugs] [Bug 4118] New: Register allocator not respecting early
 clobber on inline asm
Message-ID: <bug-4118-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4118

           Summary: Register allocator not respecting early clobber on
                    inline asm
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2916)
 --> (http://llvm.org/bugs/attachment.cgi?id=2916)
Testcase (bitcode)

Testcase attached; to reproduce, use the following command, at least on my
computer:

llc x2.bc -o - | fgrep "REGSA: %eax %eax"

I'm on Ubuntu Linux on x86-32 using a Core Duo (hopefully, that doesn't make a
difference).

The relevant inline asm is as follows:
        %0 = call i32 asm sideeffect "REGSA: $0 $1 $2",
"=&{ax},r,r,~{cx},~{bx},~{dx},~{si},~{memory},~{dirflag},~{fpsr},~{flags}"(i8*
%arrayidx.i.i, %struct.CABACContext* %tmp3.i.i) nounwind         ; <i32>
[#uses=1]

Unfortunately, it's difficult to simplify a testcase like this; the construct
in question isn't generally broken, just not working in this particular case.

Testcase reduced from libavcodec/h264.c in ffmpeg.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 18:08:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 18:08:33 -0500
Subject: [LLVMbugs] [Bug 4119] New: Pointer alignment assumptions incorrect
 for User*, causing bad behaviour in Use
Message-ID: <bug-4119-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4119

           Summary: Pointer alignment assumptions incorrect for User*,
                    causing bad behaviour in Use
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: critical
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: sdt at rapidmind.com
        ReportedBy: sdt at rapidmind.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2917)
 --> (http://llvm.org/bugs/attachment.cgi?id=2917)
Workaround that sets the default available bits to 2 instead of 3

This is a bug that's technically been around for a while, but was uncovered by
r67979. It causes various random crashes under Windows.

>From an email to llvm-dev:

The issue is with PointerLikeTypeTraits<T*>::NumLowBitsAvailable. This is set
to 3, which basically assumes that unless the traits are specialized for a
particular pointer type, objects of that type are allocated with malloc() and
aligned to 8 bytes.

While PointerLikeTypeTraits is overloaded for Use*, it is not overloaded for
User*. lib/VMCore/Use.cpp uses a PointerIntPair<User*, 1, Tag>, and things go
bad. Users are typically allocated in an array after a bunch of Uses, which are
12 bytes in size, so they may actually only be aligned to 4 bytes.

The attached patch (which I don't intend to commit, it's just a workaround)
works around this simply by dropping this down to 2 bits, which gets us past
our problem on Windows.

To actually solve this properly I see two main options:

(1) we could specialize PointerLikeTypeTraits for User*, and leave the default
value of NumLowBitsAvailable at 3.
(2) we could drop the default NumLowBitsAvailable to 2 (or even use _alignof
and similar compiler-specific extensions to determine it), and allow classes
that assert that they are only ever allocated through malloc to relax it back
up to 3.

My preference would be for option (2), due to the difficulty of tracking this
bug down, and the risk of future similar bugs happening. However, I'd
appreciate some feedback from the LLVM developer community on this. Please let
me know what you think, and I'll be happy to prepare a patch.

I'm still wondering why this wasn't an issue on other platforms. One
possibility is that Use is being bumped up to 16 bytes in size, and thus Users
never get placed at less than an 8-byte boundary. However, in that case, the
whole point of the "use diet" optimization is lost! I'll investigate and try to
find out.

I'm also still not sure why the assertion in PointerIntPair::setPointer() did
not get triggered by the User::allocHungOffUses() implementation in Use.cpp.
Perhaps the assertion is wrong (it looks reasonable, though) or perhaps there
is something else going on I haven't seen yet. I'll look into this some more as
well.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 18:49:23 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 18:49:23 -0500
Subject: [LLVMbugs] [Bug 4119] Pointer alignment assumptions incorrect for
 User*, causing bad behaviour in Use
In-Reply-To: <bug-4119-206@http.llvm.org/bugs/>
Message-ID: <200905012349.n41NnNtq006713@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4119


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Chris Lattner <clattner at apple.com>  2009-05-01 18:49:23 ---
I dropped the default to 2 bits:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090427/076954.html

And added a patch to clang to compensate for this:
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090427/016609.html

I'm still curious why no assertion happens, it could be a miscompilation by
your host compiler or something.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 19:33:44 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 19:33:44 -0500
Subject: [LLVMbugs] [Bug 3500] Wrong type computed for member access to
 unsigned bitfield
In-Reply-To: <bug-3500-206@http.llvm.org/bugs/>
Message-ID: <200905020033.n420XiD6008254@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3500


Douglas Gregor <dgregor at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #5 from Douglas Gregor <dgregor at apple.com>  2009-05-01 19:33:43 ---
Thanks, Eli! Re-opening until I get *all* of the cases. 

The (x.x+=1)+1 case is interesting, and it makes me wonder: how hard are we
supposed to try to determine whether an expression is a bit-field? We currently
look through parentheses and implicit casts, but appears that we also have to
look into binary operators that are assignments, too. Now I'm worried about,
e.g., "true? x.x : 0".


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 19:36:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 19:36:33 -0500
Subject: [LLVMbugs] [Bug 4117] ./configure fails when there' s whitespace in
 the working directory
In-Reply-To: <bug-4117-206@http.llvm.org/bugs/>
Message-ID: <200905020036.n420aXCe008393@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4117


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WONTFIX




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-05-01 19:36:33 ---
This is a known problem.  Unfortunately, if you get past the configure script,
the makefiles have a ton of similar problems.  Workaround is to create a
directory without a space and do a symlink to it.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 21:18:49 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 21:18:49 -0500
Subject: [LLVMbugs] [Bug 4121] New: Assertion `it != mi2iMap_.end() &&
 "Invalid instruction!"' failed.
Message-ID: <bug-4121-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4121

           Summary: Assertion `it != mi2iMap_.end() && "Invalid
                    instruction!"' failed.
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu


Seen using r70531 on Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp156$ llvm-gcc -O2 small.c
small.c: In function ???func_30???:
small.c:37: warning: assignment makes integer from pointer without a cast
cc1:
/home/regehr/z/tmp/llvm-r70531/include/llvm/CodeGen/LiveIntervalAnalysis.h:209:
unsigned int llvm::LiveIntervals::getInstructionIndex(llvm::MachineInstr*)
const: Assertion `it != mi2iMap_.end() && "Invalid instruction!"' failed.
small.c: At top level:
small.c:40: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.
regehr at john-home:~/volatile/tmp156$ cat small.c
char safe_add_func_int8_t_s_s (char _si1, char _si2)
{
  return _si1 > 0 && 0 > 127 - _si2 && -128 - _si2 ? : 1 + _si2;
}

short safe_sub_func_int16_t_s_s (short _si1, short _si2)
{
  return _si1 << 1 * 8 - 1 - _si2 < 0 ? _si1 : _si1 - _si2;
}

char safe_mod_func_uint8_t_u_u (char _ui1, char _ui2)
{
  return _ui2 ? : (_ui1 % _ui2);
}

char safe_lshift_func_uint8_t_u_s (char _left, int _right)
{
  return _right >= 1 * 8 || 1 >> _right ? : 1 << _right;
}

int uint8;
char g_51;
char g_109;
int g_159;

char func_30 (int p_31, char p_32, char p_33, char p_34)
{
  char l_52;
  char l;
  if (func_30
      (1,
       safe_add_func_int8_t_s_s (safe_mod_func_uint8_t_u_u
                                 (safe_lshift_func_uint8_t_u_s (l_52, g_109),
                                  safe_sub_func_int16_t_s_s (g_159, 1)),
                                 g_51), 1, 1))
    uint8;
  for (1; 1; l = safe_sub_func_int16_t_s_s)
    {
    }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 21:26:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 21:26:37 -0500
Subject: [LLVMbugs] [Bug 4123] New: clang: Assertion `New != this &&
 "this->replaceAllUsesWith( this) is NOT valid!"' failed.
Message-ID: <bug-4123-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4123

           Summary: clang: Assertion `New != this && "this-
                    >replaceAllUsesWith(this) is NOT valid!"' failed.
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu, jxyang at cs.utah.edu


Seen using r70531 on Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp157$ clang -O2 -w small.c
clang-cc: Value.cpp:328: void llvm::Value::replaceAllUsesWith(llvm::Value*):
Assertion `New != this && "this->replaceAllUsesWith(this) is NOT valid!"'
failed.
0   clang-cc 0x09021428
Stack dump:
0.      Program arguments:
/home/regehr/z/tmp/llvm-gcc-r70531-install/bin/../libexec/clang-cc -triple
i386-pc-linux-gnu -S -disable-free -main-file-name small.c --relocation-model
static --disable-fp-elim --unwind-tables=0 --fmath-errno=1 -O2 -w
-fdiagnostics-show-option -o /tmp/cc-wxGkRp.s -x c small.c 
1.      <eof> parser at end of file
2.      Per-module optimization passes
3.      Running pass 'CallGraph Pass Manager' on module 'small.c'.

regehr at john-home:~/volatile/tmp157$ cat small.c

static char safe_sub_func_uint8_t_u_u (char _ui1, char _ui2)
{
  return _ui1 - _ui2;
}

struct S0
{
  int f0;
};

struct S1
{
  char f0;
  struct S0 f3;
};

struct S2
{
  struct S1 f0;
  int f1;
};

struct S1 func_120 (int p_121, int16p_122, uint32p_123, uint32p_124,
                    struct S2 p_125, struct S1 p_127, uint32p_128);

struct S2 func_96 (int p_98, int16p_99, int16p_101, int16p_102, struct S1
p_103,
                   int16p_104, int16p_105, int16p_106)
{
  for (p_103.f0;; p_103.f0)
    {
    }
}

struct S1 func_113 (char p_114)
{
  struct S2 l_232 = {
    {
     }
    , 5L, 0x65B3L
  };
  struct S2 l_235 = {
    {
     }
    , 0L
  };
  struct S1 l_256 = {
  };
  struct S1 l_263 = {
  };
  struct S1 l_281 = {
  };
  if (func_9func82
      (func_96
       (l_256.f3.f0, 1, 1, 1,
        func_120 (l_235.f1, p_114, p_114, l_263.f0, l_232, l_281, 1), 1, 1,
        1)), 1)
    {
      for (1; 1; p_114 = safe_sub_func_uint8_t_u_u (p_114, 0))
        {
        }
    }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 21:32:06 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 21:32:06 -0500
Subject: [LLVMbugs] [Bug 4124] New: null ptr dereferenced
Message-ID: <bug-4124-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4124

           Summary: null ptr dereferenced
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu, jxyang at cs.utah.edu


Seen using r70531 on Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp158$ valgrind --trace-children=yes llvm-gcc -O3
small.c
==2526== Memcheck, a memory error detector.
==2526== Copyright (C) 2002-2007, and GNU GPL'd, by Julian Seward et al.
==2526== Using LibVEX rev 1804, a library for dynamic binary translation.
==2526== Copyright (C) 2004-2007, and GNU GPL'd, by OpenWorks LLP.
==2526== Using valgrind-3.3.0, a dynamic binary instrumentation framework.
==2526== Copyright (C) 2000-2007, and GNU GPL'd, by Julian Seward et al.
==2526== For more details, rerun with: -v
==2526== 
==2528== Memcheck, a memory error detector.
==2528== Copyright (C) 2002-2007, and GNU GPL'd, by Julian Seward et al.
==2528== Using LibVEX rev 1804, a library for dynamic binary translation.
==2528== Copyright (C) 2004-2007, and GNU GPL'd, by OpenWorks LLP.
==2528== Using valgrind-3.3.0, a dynamic binary instrumentation framework.
==2528== Copyright (C) 2000-2007, and GNU GPL'd, by Julian Seward et al.
==2528== For more details, rerun with: -v
==2528== 
==2528== Invalid read of size 4
==2528==    at 0x88D5655:
llvm::SimpleRegisterCoalescing::JoinIntervals(llvm::LiveInterval&,
llvm::LiveInterval&, bool&) (in
/home/regehr/z/tmp/llvm-gcc-r70531-install/libexec/gcc/i686-pc-linux-gnu/4.2.1/cc1)
==2528==  Address 0x8 is not stack'd, malloc'd or (recently) free'd
small.c:51: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.
==2528== 
==2528== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 23 from 1)
==2528== malloc/free: in use at exit: 542,819 bytes in 1,778 blocks.
==2528== malloc/free: 8,794 allocs, 7,016 frees, 1,627,093 bytes allocated.
==2528== For counts of detected errors, rerun with: -v
==2528== searching for pointers to 1,778 not-freed blocks.
==2528== checked 1,807,200 bytes.
==2528== 
==2528== LEAK SUMMARY:
==2528==    definitely lost: 0 bytes in 0 blocks.
==2528==      possibly lost: 1,028 bytes in 24 blocks.
==2528==    still reachable: 541,791 bytes in 1,754 blocks.
==2528==         suppressed: 0 bytes in 0 blocks.
==2528== Rerun with --leak-check=full to see details of leaked memory.
==2526== 
==2526== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 23 from 1)
==2526== malloc/free: in use at exit: 18,964 bytes in 73 blocks.
==2526== malloc/free: 491 allocs, 418 frees, 42,008 bytes allocated.
==2526== For counts of detected errors, rerun with: -v
==2526== searching for pointers to 73 not-freed blocks.
==2526== checked 153,916 bytes.
==2526== 
==2526== LEAK SUMMARY:
==2526==    definitely lost: 4,314 bytes in 32 blocks.
==2526==      possibly lost: 0 bytes in 0 blocks.
==2526==    still reachable: 14,650 bytes in 41 blocks.
==2526==         suppressed: 0 bytes in 0 blocks.
==2526== Rerun with --leak-check=full to see details of leaked memory.

regehr at john-home:~/volatile/tmp158$ cat small.c

int safe_rshift_func_int64_t_s_u (int _left, int _right)
{
  return 1 * 8 ? _left : 1;
}

short safe_unary_minus_func_uint16_t_u (short _ui)
{
  return -_ui;
}

int g_11;
char g_96;

int func_21 (char p_22, int p_24, char p_25)
{
  if (1 & g_11)
    {
      func_70 (func_59 (1));
    }
}

uint82 (int p_13, char p_14)
{
  for (1; 1; 1)
    {
      func_59 (func_51 (safe_rshift_func_int64_t_s_u (g_11, 0))
               && func_21 (g_96, p_14, 1));
    }
}

int func_51 (int p_52)
{
  if (safe_unary_minus_func_uint16_t_u (p_52) || !p_52)
    for (g_11 = 1; 0; 1)
      {
      }
  else
    {
      return 1;
    }
  return p_52;
}

int func_59 (int p_60, int p_61)
{
  g_11 = 1;
}

int func_70 (int32p_72)
{
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  1 21:47:17 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 1 May 2009 21:47:17 -0500
Subject: [LLVMbugs] [Bug 4125] New: clang wrong code at all optimization
	levels
Message-ID: <bug-4125-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4125

           Summary: clang wrong code at all optimization levels
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu, jxyang at cs.utah.edu


Seen using r70531 on Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp159$ llvm-gcc -O0 small.c -o small -w
regehr at john-home:~/volatile/tmp159$ ./small
checksum = 0
regehr at john-home:~/volatile/tmp159$ clang -O0 small.c -o small -w
regehr at john-home:~/volatile/tmp159$ ./small
checksum = 1
regehr at john-home:~/volatile/tmp159$ cat small.c
#include <stdio.h>
#include <stdint.h>

struct S1
{
  uint8_t f0;
  uint8_t f1;
};

struct S1 g_5 = {
  1L, 0xD8564F5E99A55AEFLL
};

struct S1 g_158 = {
  0xF5395DFB6A07BBA0LL, -9L
};

struct S1 g_297 = {
  0x48FFBA5703788BE3LL, 1L
};

struct S1 g_313 = {
  -8L, 1L
};

struct S1 func_1 (void);

uint32_t func_2 (uint8_t p_4, struct S1 p_6, uint8_t p_7);

int32_t func_23 (int32_t p_24, uint8_t p_25);

uint16_t func_35 (uint8_t p_37, int32_t p_39, int32_t p_40, uint32_t p_42,
                  int64_t p_43);

int32_t func_45 (int32_t p_46, struct S1 p_47);

struct S1 func_72 (uint16_t p_73);

struct S1 func_92 (struct S1 p_93);

struct S1 func_96 (uint32_t p_98, struct S1 p_99);

struct S1 func_1 (void)
{
  uint8_t l_3 = 0;
  struct S1 l_326 = {
    3L, 0xEB43ED56C7AF7ED3LL
  };
  uint8_t l_327 = 1;
  uint8_t l_328 = 1;
  func_2 (1, func_92 (func_72 (1)),
          func_23 (1,
                   func_45 (func_45
                            (1,
                             func_96 (func_35
                                      (g_313.f1, l_327, l_328, l_3, 1),
                                      l_326)), func_96 (l_3,
                                                        func_96 (func_35
                                                                 (g_313.f1,
                                                                  l_326.f1,
                                                                  l_326.f0,
                                                                  g_297.f1,
                                                                  l_326.f1),
                                                                 func_96
                                                                 (g_313.f0,
                                                                  l_326))))));
  return g_158;
}

uint32_t func_2 (uint8_t p_4, struct S1 p_6, uint8_t p_7)
{
  return 1;
}

int32_t func_23 (int32_t p_24, uint8_t p_25)
{
  return g_5.f1;
}

uint16_t func_35 (uint8_t p_37, int32_t p_39, int32_t p_40, uint32_t p_42,
         int64_t p_43)
{
  g_5.f0 = p_42;
  return 1;
}

int32_t func_45 (int32_t p_46, struct S1 p_47)
{
  return 1;
}

struct S1 func_72 (uint16_t p_73)
{
  return g_5;
}

struct S1 func_92 (struct S1 p_93)
{
  return g_158;
}

struct S1 func_96 (uint32_t p_98, struct S1 p_99)
{
  return g_5;
}

int main (void)
{
  func_1 ();
  printf ("checksum = %d\n", g_5.f0);
  return 0;
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May  2 01:13:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 2 May 2009 01:13:22 -0500
Subject: [LLVMbugs] [Bug 4125] clang wrong code at all optimization levels
In-Reply-To: <bug-4125-206@http.llvm.org/bugs/>
Message-ID: <200905020613.n426DMI9019203@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4125


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-02 01:13:21 ---
This testcase has unspecified behavior because the evaluation order of
arguments to a function is unspecified.  clang and llvm-gcc are putting the two
calls to func_35 in different orders.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May  2 12:47:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 2 May 2009 12:47:13 -0500
Subject: [LLVMbugs] [Bug 4126] New: conditional jump to nowhere
Message-ID: <bug-4126-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4126

           Summary: conditional jump to nowhere
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu, jxyang at cs.utah.edu


Seen using r70624 on Ubuntu Hardy on x86.

One of the functions in the emitted code contains this strange-looking body:

uint8func_52:
        pushl   %ebp
        movl    %esp, %ebp
        subl    $8, %esp
        movl    8(%ebp), %eax
        testl   %eax, %eax
        jne     .LBB9_1
        jle     .LBB9_1
        movl    $0, (%esp)
        call    funcmul_
        addl    $8, %esp
        popl    %ebp
        ret

regehr at john-home:~/volatile/tmp158$ llvm-gcc -Os small.c
/tmp/ccgm63CD.o: In function `uint8func_52':
small.c:(.text+0x55): undefined reference to `funcmul_'
small.c:(.text+0x43): undefined reference to `.LBB9_1'
small.c:(.text+0x49): undefined reference to `.LBB9_1'
collect2: ld returned 1 exit status

regehr at john-home:~/volatile/tmp158$ cat small.c

typedef int int16_t;
typedef int int32_t;
__extension__ typedef int int64_t;
typedef char uint8_t;
typedef int uint32_t;
__extension__ typedef int uint64_t;
static uint8_t
safe_mul_func_int16_t_s_s (int16_t _si1, int16_t _si2)
{
  return _si2 > 0 && _si1 > 1 + _si2 && _si2 < -32767 - 1 + _si1 && +1
    || _si2 ? : _si2;
}
static int16_t
safe_div_func_int32_t_s_s (int32_t _si1, uint8_t _si2)
{
  return -2147483647 - 1 == 1 ? : _si1;
}

int crc32 (int x)
{
}

static void
crc32_8bytes (uint64_t val)
{
  crc32 (val >> 0);
  crc32 (val >> 8);
  crc32 (val >> 16);
}
volatile uint8_t g_89;
uint8_t g_93;
uint8_t
func_2 (uint32_t p_4, uint8_t p_5, uint8_t p_7, uint8_t p_8, uint8_t p_9)
{
}

uint82 (uint8_t p_13, uint8_t p_14)
{
}

uint81 (int64p_22)
{
}

uint8func_52 (uint64_t p_53)
{
  uint8_t l_91;
  funcmul_ (func_2
            (1, g_93, safe_div_func_int32_t_s_s (p_53, 1)
             || safe_mul_func_int16_t_s_s (l_91, p_53), 1, 1));
}

uint84 (uint8p_56)
{
}

uint87 (uint8p_58)
{
}

uint89 (int64_t p_61, uint8_t p_62)
{
}
int
main (void)
{
  crc32_8bytes (g_89);
  crc32_8bytes (g_93);
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May  2 23:34:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 2 May 2009 23:34:13 -0500
Subject: [LLVMbugs] [Bug 4084] fix-it diagnostics on multiple lines need
	fixin'
In-Reply-To: <bug-4084-206@http.llvm.org/bugs/>
Message-ID: <200905030434.n434YDSR005295@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4084


Douglas Gregor <dgregor at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dgregor at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Douglas Gregor <dgregor at apple.com>  2009-05-02 23:34:12 ---
Fixed in r70656, by suppressing fix-it output if we can't show *all* of the
fixes needed.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 03:15:32 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 03:15:32 -0500
Subject: [LLVMbugs] [Bug 4129] New: clang parser crash: terminate called
 after throwing an instance of 'std::out_of_range' what():
 basic_string::erase
Message-ID: <bug-4129-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4129

           Summary: clang parser crash: terminate called after throwing an
                    instance of 'std::out_of_range'
                    what():  basic_string::erase
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2920)
 --> (http://llvm.org/bugs/attachment.cgi?id=2920)
testcase-min.i

$ clang testcase-min.i
testcase-min.i:1:62: warning: declaration of 'struct bts_tracer' will not be
visible outside of this function

 struct bts_trace {  struct ds_trace ds;  int (*read)(struct bts_tracer
*tracer, const void *at,       struct bts_struct ...
                                                             ^                  
testcase-min.i:1:111: warning: declaration of 'struct bts_struct' will not be
visible outside of this function              

 struct bts_trace {  struct ds_trace ds;  int (*read)(struct bts_tracer
*tracer, const void *at,       struct bts_struct ...
                                                                               
                              ^             
testcase-min.i:1:150: warning: declaration of 'struct bts_tracer' will not be
visible outside of this function              
terminate called after throwing an instance of 'std::out_of_range'              
  what():  basic_string::erase                                                  
0   clang-cc        0x000000000105dfdf                                          
1   clang-cc        0x000000000105e3d9                                          
2   libpthread.so.0 0x00007f7792b0a7b0                                          
3   libc.so.6       0x00007f7791e2a115 gsignal + 53                             
4   libc.so.6       0x00007f7791e2b633 abort + 387                              
5   libstdc++.so.6  0x00007f77926b0e65 __gnu_cxx::__verbose_terminate_handler()
+ 277                                       
6   libstdc++.so.6  0x00007f77926af256                                          
7   libstdc++.so.6  0x00007f77926af283                                          
8   libstdc++.so.6  0x00007f77926af37e                                          
9   libstdc++.so.6  0x00007f779264ec17 std::__throw_out_of_range(char const*) +
103                                         
10  clang-cc        0x00000000006acfb8
11  clang-cc        0x00000000006ae1ec
12  clang-cc        0x00000000006af60b
13  clang-cc        0x000000000078c285
14  clang-cc        0x00000000005ad925
15  clang-cc        0x00000000005b67a2
16  clang-cc        0x00000000007310d5
17  clang-cc        0x00000000007252a1
18  clang-cc        0x0000000000726077
19  clang-cc        0x0000000000727824
clang::Parser::ParseDirectDeclarator(clang::Declarator&) + 676
20  clang-cc        0x00000000007230aa
21  clang-cc        0x0000000000728963
22  clang-cc        0x000000000072978b
23  clang-cc        0x00000000007316b6
24  clang-cc        0x00000000007252a1
25  clang-cc        0x000000000071d482
26  clang-cc        0x000000000071edf2
27  clang-cc        0x000000000071f206
28  clang-cc        0x00000000005ad5e3
29  clang-cc        0x000000000047f19f
30  clang-cc        0x0000000000480631 main + 1985
31  libc.so.6       0x00007f7791e165a6 __libc_start_main + 230
32  clang-cc        0x000000000042b5a9
Stack dump:
0.      Program arguments: /usr/local/bin/../libexec/clang-cc -triple
x86_64-unknown-linux-gnu -S -disable-free -main-file-name testcase-min.i
--relocation-model static --disable-fp-elim --unwind-tables=1 --fmath-errno=1
-fdiagnostics-show-option -o /tmp/cc-ZzEXpP.s -x cpp-output testcase-min.i
1.      testcase-min.i:1:161: current parser token '*'
2.      testcase-min.i:1:2: parsing struct/union body 'bts_trace'


If I reformat the input using topformflat clang no longer crashes, it only
crashes if everything is on one line.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 03:30:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 03:30:37 -0500
Subject: [LLVMbugs] [Bug 4130] New: clang infloops on invalid C code
Message-ID: <bug-4130-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4130

           Summary: clang infloops on invalid C code
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


$ time gcc testcase-min.i
testcase-min.i:1: error: expected ???=???, ???,???, ???;???, ???asm??? or
???__attribute__??? before ???ptep_get_and_clear???

real    0m0.007s
user    0m0.004s
sys     0m0.008s

However clang infloops:
$ clang testcase-min.i
^C

Here is the testcase-min.i:
static inline __attribute__((always_inline)) pte_t ptep_get_and_clear(struct
mm_struct *mm, unsigned long addr,            pte_t *ptep) {


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 03:39:40 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 03:39:40 -0500
Subject: [LLVMbugs] [Bug 4129] clang parser crash: terminate called after
 throwing an instance of 'std::out_of_range' what(): basic_string::erase
In-Reply-To: <bug-4129-206@http.llvm.org/bugs/>
Message-ID: <200905030839.n438de3i023379@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4129


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-05-03 03:39:39 ---


*** This bug has been marked as a duplicate of bug 4128 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 04:20:50 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 04:20:50 -0500
Subject: [LLVMbugs] [Bug 4131] New: clang: error: indirection requires
 pointer operand ('void' invalid)
Message-ID: <bug-4131-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4131

           Summary: clang: error: indirection requires pointer operand
                    ('void' invalid)
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 4068


clang -MD -MF kernel/.smp.o.d  -nostdinc -isystem include -Iinclude 
-I/home/edwin/builds/linux-2.6/arch/x86/include -include
include/linux/autoconf.h -D__KERNEL__ -Wall -Wundef-Wstrict-prototypes
-Wno-trigraphs -fno-strict-aliasing -fno-common
-Werror-implicit-function-declaration -O2 -m64 -march=core2 -mno-red-zone
-mcmodel=kernel -funit-at-a-time -maccumulate-outgoing-args -DCONFIG_AS_CFI=1
-DCONFIG_AS_CFI_SIGNAL_FRAME=1 -pipe -Wno-sign-compare
-fno-asynchronous-unwind-tables -mno-sse -mno-mmx -mno-sse2 -mno-3dnow
-Iarch/x86/include/asm/mach-default -Wframe-larger-than=2048
-fno-stack-protector -fno-omit-frame-pointer -fno-optimize-sibling-calls
-Wdeclaration-after-statement -Wno-pointer-sign -fwrapv -fno-dwarf2-cfi-asm 
-D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(smp)" 
-D"KBUILD_MODNAME=KBUILD_STR(smp)"  -c -o kernel/.tmp_smp.o kernel/smp.c
warning: unknown warning option: -Wframe-larger-than=2048
kernel/smp.c:14:8: error: variable has incomplete type 'typeof(struct
call_single_queue)' (aka 'struct call_single_queue')

static DEFINE_PER_CPU(struct call_single_queue, call_single_queue);
       ^
<scratch space>:201:1: note: instantiated from:

per_cpu__call_single_queue
^
kernel/smp.c:14:8: note: forward declaration of 'struct call_single_queue'

static DEFINE_PER_CPU(struct call_single_queue, call_single_queue);
       ^
kernel/smp.c:14:30: note: instantiated from:

static DEFINE_PER_CPU(struct call_single_queue, call_single_queue);
                             ^
kernel/smp.c:42:34: error: indirection requires pointer operand ('void'
invalid)
                struct call_single_queue *q = &per_cpu(call_single_queue, i);
                                               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/asm-generic/percpu.h:57:3: note: instantiated from:
        (*SHIFT_PERCPU_PTR(&per_cpu_var(var), per_cpu_offset(cpu)))
         ^
kernel/smp.c:67:35: error: indirection requires pointer operand ('void'
invalid)
        struct call_single_queue *dst = &per_cpu(call_single_queue, cpu);
                                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/asm-generic/percpu.h:57:3: note: instantiated from:
        (*SHIFT_PERCPU_PTR(&per_cpu_var(var), per_cpu_offset(cpu)))
         ^
kernel/smp.c:155:33: error: indirection requires pointer operand ('void'
invalid)
        struct call_single_queue *q = &__get_cpu_var(call_single_queue);
                                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/asm-generic/percpu.h:59:3: note: instantiated from:
        (*SHIFT_PERCPU_PTR(&per_cpu_var(var), my_cpu_offset))
         ^
5 diagnostics generated.
make[1]: *** [kernel/smp.o] Error 1
make: *** [kernel/smp.o] Error 2


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 04:38:18 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 04:38:18 -0500
Subject: [LLVMbugs] [Bug 4131] clang: error: indirection requires pointer
 operand ('void' invalid)
In-Reply-To: <bug-4131-206@http.llvm.org/bugs/>
Message-ID: <200905030938.n439cIeD016426@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4131


T??r??k Edwin <edwintorok at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #2 from T??r??k Edwin <edwintorok at gmail.com>  2009-05-03 04:37:52 ---


*** This bug has been marked as a duplicate of bug 4065 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 04:38:17 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 04:38:17 -0500
Subject: [LLVMbugs] [Bug 4063] clang -MD -MF doesn't create dependency file
 for .S source files
In-Reply-To: <bug-4063-206@http.llvm.org/bugs/>
Message-ID: <200905030938.n439cHpL016400@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4063


Daniel Dunbar <daniel at zuster.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Daniel Dunbar <daniel at zuster.org>  2009-05-03 04:38:01 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090427/016686.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 05:16:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 05:16:54 -0500
Subject: [LLVMbugs] [Bug 4132] New: clang: wrong preprocessing of .S file:
 Error: invalid character '%' in mnemonic
Message-ID: <bug-4132-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4132

           Summary: clang: wrong preprocessing of .S file: Error: invalid
                    character '%' in mnemonic
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 4068


Created an attachment (id=2924)
 --> (http://llvm.org/bugs/attachment.cgi?id=2924)
clang -E

arch/x86/crypto/aes-x86_64-asm_64.S:190: Error: invalid character '%' in
mnemonic

clang -MD -MF arch/x86/crypto/.aes-x86_64-asm_64.o.d  -nostdinc -isystem
include -Iinclude  -I/home/edwin/builds/linux-2.6/arch/x86/include -include
include/linux/autoconf.h -D__KERNEL__ -D__ASSEMBLY__ -m64 -DCONFIG_AS_CFI=1
-DCONFIG_AS_CFI_SIGNAL_FRAME=1 -Iarch/x86/include/asm/mach-default    -DMODULE
-c -o arch/x86/crypto/aes-x86_64-asm_64.o arch/x86/crypto/aes-x86_64-asm_64.S
-E -o x.s

gcc -MD -MF arch/x86/crypto/.aes-x86_64-asm_64.o.d  -nostdinc -isystem
/usr/lib/gcc/x86_64-linux-gnu/4.3.3/include -Iinclude 
-I/home/edwin/builds/linux-2.6/arch/x86/include -include
include/linux/autoconf.h -D__KERNEL__ -D__ASSEMBLY__ -m64 -DCONFIG_AS_CFI=1
-DCONFIG_AS_CFI_SIGNAL_FRAME=1 -Iarch/x86/include/asm/mach-default    -DMODULE 
arch/x86/crypto/aes-x86_64-asm_64.S -E -o y.s

clang's x.s is not compilable by gcc -c, but gcc's y.s is compilable by clang.

I'm attaching preprocessed files, however clang's preprocessed file is invalid
in this case i think.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 06:01:46 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 06:01:46 -0500
Subject: [LLVMbugs] [Bug 4133] New: clang: symbol declared as always_inline
 and extern considered internal
Message-ID: <bug-4133-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4133

           Summary: clang: symbol declared as always_inline and extern
                    considered internal
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 4068


The kernel fails to link:
net/unix/af_unix.c:(.text+0x6d779): undefined reference to `mutex_lock'
net/built-in.o: In function `unix_dgram_recvmsg':
net/unix/af_unix.c:(.text+0x6e9af): undefined reference to `mutex_lock'
net/built-in.o: In function `unix_autobind':
net/unix/af_unix.c:(.text+0x6f30f): undefined reference to `mutex_lock'
net/built-in.o: In function `unix_stream_recvmsg':
net/unix/af_unix.c:(.text+0x70531): undefined reference to `mutex_lock'
net/unix/af_unix.c:(.text+0x70966): undefined reference to `mutex_lock


Testcase:
$ gcc -c testcase-min.i -o x.o
0000000000000000 T mutex_lock
$ clang testcase-min.i -c -o x.o
$ nm x.o
$ clang testcase-min.i -c -o x.o -std=gnu89
$ nm x.o
$ cat testcase-min.i
void inline __attribute__ ((always_inline))
  __attribute__ ((__section__ (".sched.text"))) mutex_lock (struct mutex
                                                            *lock)
{
}

extern
typeof (mutex_lock)
  mutex_lock;


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 06:11:06 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 06:11:06 -0500
Subject: [LLVMbugs] [Bug 4134] New: clang: __builtin_extract_return_addr not
	implemented
Message-ID: <bug-4134-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4134

           Summary: clang: __builtin_extract_return_addr not implemented
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 4068


Testcase:
unsigned long foo(unsigned long addr)
{
    return (unsigned long) __builtin_extract_return_addr((void*)addr);
}
int main()
{}


$ clang foo.c

tmp/cc-y3kFrH.o: In function `foo':
x.c:(.text+0x1b): undefined reference to `__builtin_extract_return_addr'
collect2: ld returned 1 exit status

$ gcc foo.c
$


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 06:43:02 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 06:43:02 -0500
Subject: [LLVMbugs] [Bug 4135] New: PointerMayBeCaptured() vs.
	readonly/readnone
Message-ID: <bug-4135-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4135

           Summary: PointerMayBeCaptured() vs. readonly/readnone
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fvbommel at wxs.nl
                CC: llvmbugs at cs.uiuc.edu


PointerMayBeCaptured (in Analysis/CaptureTracking) assumes that
readonly/readnone functions returning void can't capture their pointer
arguments. Due to the "visible to caller functions" clauses in the attribute
definitions this isn't technically correct.

(If this is resolved by changing the attribute descriptions to remove those
clauses, please also update PointerMayBeCaptured to handle readonly/readnone
functions returning pointers by tracking the return value)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 10:59:00 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 10:59:00 -0500
Subject: [LLVMbugs] [Bug 4136] New: Infinite recursion in scheduler
Message-ID: <bug-4136-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4136

           Summary: Infinite recursion in scheduler
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: asl at math.spbu.ru
                CC: gohman at apple.com, llvmbugs at cs.uiuc.edu


Created an attachment (id=2926)
 --> (http://llvm.org/bugs/attachment.cgi?id=2926)
Testcase

Consider attached bitecode. It seems that scheduler just infinitely loops on it
and then crashes due to stack exhaustion.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 11:08:40 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 11:08:40 -0500
Subject: [LLVMbugs] [Bug 4137] New: Miscompilation of heavy-phi code
Message-ID: <bug-4137-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4137

           Summary: Miscompilation of heavy-phi code
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: asl at math.spbu.ru
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2927)
 --> (http://llvm.org/bugs/attachment.cgi?id=2927)
Original bytecode

Consider the .bc attached. Currently llc produces invalid code for it (running
the resulting binary would yield "sort test failed"). CBE works fine.

Bugpoint reduced "safe" and "test" parts are attached as well. The bytecode
itself  was derived from
/SingleSource/Benchmarks/Adobe-C++/stepanov_abstraction.cpp after some bc=>bc
transformation.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 12:14:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 12:14:33 -0500
Subject: [LLVMbugs] [Bug 4138] New: GEP is not of right type for indices!
Message-ID: <bug-4138-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4138

           Summary: GEP is not of right type for indices!
           Product: tools
           Version: 2.5
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-ld
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: mattp223 at renzelmann.com
                CC: llvmbugs at cs.uiuc.edu


Hello,

As an experiment, I'm attempting to port user-mode Linux 2.6.29 to LLVM.  In my
day-old attempt at a port, LLVM gets relatively far before encountering a
problem.

Build environment/platform:  LLVM 2.5, using the LLVM-2.5 GCC front end for GCC
4.2.1.  Used GCC 4.2.4 for building LLVM.  Platform is RHEL 5, kernel
2.6.18-128.1.6.el5.

Here's the output of a slightly-reduced test case.  You can download the
object/bitcode files for this test from here:

http://pages.cs.wisc.edu/~mjr/files/test_case.tbz

I can provide the complete source code as well, if desired, since it's straight
from the Linux 2.6.29 kernel.

======================================
> llvm-ld -verify-each -disable-opt -disable-inlining -v   -r -o proc.o base.o proc_tty.o cmdline.o cpuinfo.o

  Linking bitcode file 'base.o'
  Linked in file 'base.o'
  Linking bitcode file 'proc_tty.o'
  Linked in file 'proc_tty.o'
  Linking bitcode file 'cmdline.o'
  Linked in file 'cmdline.o'
  Linking bitcode file 'cpuinfo.o'
  Linked in file 'cpuinfo.o'
GEP is not of right type for indices!
        %29 = getelementptr %struct.inode* %2, i32 0, i32 22            ;
<%struct.inode_operations**> [#uses=1]
 { i32 (%struct.inode*, %struct.dentry*, i32, %struct.nameidata*)*,
%struct.dentry* (%struct.inode*, %struct.dentry*, %struct.nameidata*)*, i32
(%struct.dentry*, %struct.inode*, %struct.dentry*)*, i32 (%struct.inode*,
%struct.dentry*)*, i32 (%struct.inode*, %struct.dentry*, i8*)*, i32
(%struct.inode*, %struct.dentry*, i32)*, i32 (%struct.inode*,
%struct.dentry*)*, i32 (%struct.inode*, %struct.dentry*, i32, i32)*, i32
(%struct.inode*, %struct.dentry*, %struct.inode*, %struct.dentry*)*, i32
(%struct.dentry*, i8*, i32)*, i8* (%struct.dentry*, %struct.nameidata*)*, void
(%struct.dentry*, %struct.nameidata*, i8*)*, void (%struct.inode*)*, i32
(%struct.inode*, i32)*, i32 (%struct.dentry*, %struct.iattr*)*, i32
(%struct.vfsmount*, %struct.dentry*, %struct.kstat*)*, i32 (%struct.dentry*,
i8*, i8*, i32, i32)*, i32 (%struct.dentry*, i8*, i8*, i32)*, i32
(%struct.dentry*, i8*, i32)*, i32 (%struct.dentry*, i8*)*, void
(%struct.inode*, i64, i64)*, i32 (%struct.inode*, i32, i64, i64)*, i32
(%struct.inode*, %struct.fiemap_extent_info*, i64, i64)* }*Instruction does not
dominate all uses!
        %29 = getelementptr %struct.inode* %2, i32 0, i32 22            ;
<%struct.inode_operations**> [#uses=1]
        store %struct.inode_operations* %27, %struct.inode_operations** %29,
align 4
Broken module found, compilation aborted!
0   llvm-ld   0x083571de
1   libc.so.6 0x002ccac1 abort + 257
2   llvm-ld   0x082fa195
./test.sh: line 5: 14575 Aborted                 llvm-ld -verify-each
-disable-opt -disable-inlining -v -r -o proc.o base.o proc_tty.o cmdline.o
cpuinfo.o
======================================

These objects were produced using commands like this:

======================================
llvm-gcc
    -Wp,-MD,fs/proc/.page.o.d 
    -nostdinc
    -isystem
/scratch/sym/llvm-gcc4.2-2.5-x86-linux-RHEL4/bin/../lib/gcc/i686-pc-linux-gnu/4.2.1/include
    -Iinclude 
    -I/scratch/sym/linux-2.6.29-llvm/arch/um/include
    -include include/linux/autoconf.h
    -D__KERNEL__
    -I/scratch/sym/linux-2.6.29-llvm/arch/um/sys-i386
    -I/scratch/sym/linux-2.6.29-llvm/arch/x86/include
    -Wall
    -Wundef
    -Wstrict-prototypes
    -Wno-trigraphs
    -fno-strict-aliasing
    -fno-common
    -Werror-implicit-function-declaration
    -Os
    -emit-llvm
    -D__arch_um__
    -DSUBARCH=\"i386\"
    -I/scratch/sym/linux-2.6.29-llvm/arch/um/include/shared
    -I/scratch/sym/linux-2.6.29-llvm/arch/um/sys-i386/shared
    -I/scratch/sym/linux-2.6.29-llvm/arch/um/include/shared/skas
    -Dvmap=kernel_vmap
    -Din6addr_loopback=kernel_in6addr_loopback
    -Din6addr_any=kernel_in6addr_any
    -funit-at-a-time
    -march=i686
    -Wa,-mtune=generic32
    -mpreferred-stack-boundary=2
    -ffreestanding
    -D_LARGEFILE64_SOURCE
    -Derrno=kernel_errno
    -Dsigprocmask=kernel_sigprocmask
    -Dmktime=kernel_mktime
    -fno-stack-protector
    -fno-omit-frame-pointer
    -fno-optimize-sibling-calls
    -g
    -Wdeclaration-after-statement
    -Wno-pointer-sign
    -fwrapv 
    -D"KBUILD_STR(s)=#s"
    -D"KBUILD_BASENAME=KBUILD_STR(page)" 
    -D"KBUILD_MODNAME=KBUILD_STR(proc)" 
    -c
    -o fs/proc/page.o fs/proc/page.c
======================================

I recognize this is a huge mess compared to some of the test cases you normally
deal with.  If you would like me to try anything else, aside from writing a
patch ;), or you need more information, please let me know.

Thanks and regards,
Matt


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 12:48:57 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 12:48:57 -0500
Subject: [LLVMbugs] [Bug 3373] cannot tie inline asm operands of different
	widths
In-Reply-To: <bug-3373-206@http.llvm.org/bugs/>
Message-ID: <200905031748.n43HmvI1009526@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3373


T??r??k Edwin <edwintorok at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #28 from T??r??k Edwin <edwintorok at gmail.com>  2009-05-03 12:48:56 ---
(In reply to comment #27)
> This appears to be fixed...
> 

Closing then.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 13:00:17 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 13:00:17 -0500
Subject: [LLVMbugs] [Bug 3373] cannot tie inline asm operands of different
	widths
In-Reply-To: <bug-3373-206@http.llvm.org/bugs/>
Message-ID: <200905031800.n43I0HcT009920@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3373


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #29 from Duncan Sands <baldrick at free.fr>  2009-05-03 13:00:16 ---
This was reported as an llvm-gcc problem, but was not fixed in llvm-gcc.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 13:14:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 13:14:34 -0500
Subject: [LLVMbugs] [Bug 4133] clang: symbol declared as always_inline and
 extern considered internal
In-Reply-To: <bug-4133-206@http.llvm.org/bugs/>
Message-ID: <200905031814.n43IEYl2010456@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4133


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-03 13:14:33 ---
Fixed in r70786.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 13:32:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 13:32:48 -0500
Subject: [LLVMbugs] [Bug 4140] New: clang -print-file-name not consistent
	with gcc
Message-ID: <bug-4140-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4140

           Summary: clang -print-file-name not consistent with gcc
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 4068


$ gcc -print-file-name=libncursesw.so
/usr/lib/gcc/x86_64-linux-gnu/4.3.3/../../../../lib/libncursesw.so
$ clang
libncursesw.so

This is causing silent failures during evaluating of certain variables in
Makefile, which in turn leads to link failures because a required -l line is
missing.

$ make HOSTCC=clang CC=clang menuconfig
fails with lots of undefined references to symbols in ncurses

The link line is:
clang  -o scripts/kconfig/mconf scripts/kconfig/mconf.o
scripts/kconfig/zconf.tab.o scripts/kconfig/lxdialog/checklist.o
scripts/kconfig/lxdialog/util.o scripts/kconfig/lxdialog/inputbox.o
scripts/kconfig/lxdialog/textbox.o scripts/kconfig/lxdialog/yesno.o
scripts/kconfig/lxdialog/menubox.o

vs. the link line with gcc:
gcc  -o scripts/kconfig/mconf scripts/kconfig/mconf.o
scripts/kconfig/zconf.tab.o scripts/kconfig/lxdialog/checklist.o
scripts/kconfig/lxdialog/util.o scripts/kconfig/lxdialog/inputbox.o
scripts/kconfig/lxdialog/textbox.o scripts/kconfig/lxdialog/yesno.o
scripts/kconfig/lxdialog/menubox.o -lncursesw

This is because check-lxdialog.sh -ldflags clang outputs nothing:
/bin/sh -x
/home/edwin/builds/linux-2.6/scripts/kconfig/lxdialog/check-lxdialog.sh
-ldflags clang
+ tmp=.lxdialog.tmp                                                             
+ trap 'rm -f .lxdialog.tmp' 0 1 2 3 15                                         
+ '[' 2 -eq 0 ']'                                                               
+ cc=                                                                           
+ case "$1" in                                                                  
+ shift                                                                         
+ cc=clang                                                                      
+ ldflags                                                                       
+ for ext in so a dylib                                                         
+ for lib in ncursesw ncurses curses                                            
+ clang -print-file-name=libncursesw.so                                         
+ grep -q /                                                                     
+ '[' 1 -eq 0 ']'                                                               
+ for lib in ncursesw ncurses curses                                            
+ clang -print-file-name=libncurses.so                                          
+ grep -q /                                                                     
+ '[' 1 -eq 0 ']'                                                               
+ for lib in ncursesw ncurses curses                                            
+ clang -print-file-name=libcurses.so                                           
+ grep -q /                                                                     
+ '[' 1 -eq 0 ']'                                                               
+ for ext in so a dylib                                                         
+ for lib in ncursesw ncurses curses                                            
+ clang -print-file-name=libncursesw.a                                          
+ grep -q /                                                                     
+ '[' 1 -eq 0 ']'
+ for lib in ncursesw ncurses curses
+ clang -print-file-name=libncurses.a
+ grep -q /
+ '[' 1 -eq 0 ']'
+ for lib in ncursesw ncurses curses
+ clang -print-file-name=libcurses.a
+ grep -q /
+ '[' 1 -eq 0 ']'
+ for ext in so a dylib
+ for lib in ncursesw ncurses curses
+ clang -print-file-name=libncursesw.dylib
+ grep -q /
+ '[' 1 -eq 0 ']'
+ for lib in ncursesw ncurses curses
+ clang -print-file-name=libncurses.dylib
+ grep -q /
+ '[' 1 -eq 0 ']'
+ for lib in ncursesw ncurses curses
+ clang -print-file-name=libcurses.dylib
+ grep -q /
+ '[' 1 -eq 0 ']'
+ exit 1
+ rm -f .lxdialog.tmp

This is how gcc behaves:
+ tmp=.lxdialog.tmp
+ trap 'rm -f .lxdialog.tmp' 0 1 2 3 15
+ '[' 2 -eq 0 ']'
+ cc=
+ case "$1" in
+ shift
+ cc=gcc
+ ldflags
+ for ext in so a dylib
+ for lib in ncursesw ncurses curses
+ gcc -print-file-name=libncursesw.so
+ grep -q /
+ '[' 0 -eq 0 ']'
+ echo -lncursesw
-lncursesw
+ exit
+ rm -f .lxdialog.tmp


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 13:51:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 13:51:09 -0500
Subject: [LLVMbugs] [Bug 4135] PointerMayBeCaptured() vs. readonly/readnone
In-Reply-To: <bug-4135-206@http.llvm.org/bugs/>
Message-ID: <200905031851.n43Ip9c8012269@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4135


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #12 from Chris Lattner <clattner at apple.com>  2009-05-03 13:51:09 ---
I updated langref to make it explicit that readnone/readonly cannot throw.  I
don't think there is anything left in this bug to do.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 14:16:17 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 14:16:17 -0500
Subject: [LLVMbugs] [Bug 4132] clang: wrong preprocessing of .S file: Error:
 invalid character '%' in mnemonic
In-Reply-To: <bug-4132-206@http.llvm.org/bugs/>
Message-ID: <200905031916.n43JGHfm013194@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4132


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-03 14:16:17 ---
Fixed in r70793.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 14:24:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 14:24:26 -0500
Subject: [LLVMbugs] [Bug 4134] clang: __builtin_extract_return_addr not
	implemented
In-Reply-To: <bug-4134-206@http.llvm.org/bugs/>
Message-ID: <200905031924.n43JOQ2L013533@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4134


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-03 14:24:26 ---
Fixed in r70794.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 14:33:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 14:33:22 -0500
Subject: [LLVMbugs] [Bug 4141] New: PointerMayBeCaptured() can do better for
 readonly/ readnone function calls.
Message-ID: <bug-4141-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4141

           Summary: PointerMayBeCaptured() can do better for
                    readonly/readnone function calls.
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fvbommel at wxs.nl
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2935)
 --> (http://llvm.org/bugs/attachment.cgi?id=2935)
Proposed patch

PointerMayBeCaptured() currently only treats calls to readonly/readnone
functions differently if they return 'void', but it could also handle them if
they return a pointer (by tracking whether the return value escapes).

>From the testcase in the attached patch:
---
declare i8* @external_identity(i8*) readonly
define void @nc6(i8* %p) {
        call i8* @external_identity(i8* %p)
        ret void
}
---
That %p parameter should be marked 'nocapture' by "opt -functionattrs". There's
no way it could escape.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 17:36:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 17:36:48 -0500
Subject: [LLVMbugs] [Bug 2524] bogus error when taking the address of a
 noproto function decl
In-Reply-To: <bug-2524-206@http.llvm.org/bugs/>
Message-ID: <200905032236.n43Mam5x020102@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2524


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-03 17:36:47 ---
"Fixed" (downgraded to a warning) in r70805.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May  3 17:38:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 3 May 2009 17:38:31 -0500
Subject: [LLVMbugs] [Bug 4036] `opt -verify` crashes on Windows
In-Reply-To: <bug-4036-206@http.llvm.org/bugs/>
Message-ID: <200905032238.n43McVxc020179@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4036


Mikhail Glushenkov <foldr at codedgers.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #9 from Mikhail Glushenkov <foldr at codedgers.com>  2009-05-03 17:38:31 ---
Couldn't reproduce this with recent trunk. This bug really seems to be a
duplicate of bug 4119.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 02:52:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 02:52:38 -0500
Subject: [LLVMbugs] [Bug 4145] New: Bulds fail on Cygwin with error message
 about dwarf version
Message-ID: <bug-4145-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4145

           Summary: Bulds fail on Cygwin with error message about dwarf
                    version
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Windows NT
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: magnus.reftel at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2940)
 --> (http://llvm.org/bugs/attachment.cgi?id=2940)
Output from "make > dwarf_error.txt 2>&1" after a failing build

Since Wednesday or Thursday last week, builds have failed on Cygwin with error
messages like the following:

/usr/lib/gcc/i686-pc-cygwin/4.3.2/../../../../i686-pc-cygwin/bin/ld: Dwarf
Error: found dwarf version '8242', this reader only handles version 2
information.
/c/temp/llvm-svn/Debug/lib/LLVMXCore.o:XCoreTargetMachine.cpp:(.rdata$_ZTVN4llvm18XCoreTargetMachineE[vtable
for llvm::XCoreTargetMachine]+0x4c): undefined reference to
`llvm::LLVMTargetMachine::addPassesToEmitMachineCode(llvm::PassManagerBase&,
llvm::MachineCodeEmitter&, llvm::CodeGenOpt::Level)'

I'm attaching the make output from building again after a failure.

I am currently running revision 70826, but have had autobuild running since
around 15:00 UTC on April 30. No builds have succeeded. The last revision that
I know did not have this problem was 70374. Current svn info output:

$ svn info
Path: .
URL: http://llvm.org/svn/llvm-project/llvm/trunk
Repository Root: http://llvm.org/svn/llvm-project
Repository UUID: 91177308-0d34-0410-b5e6-96231b3b80d8
Revision: 70826
Node Kind: directory
Schedule: normal
Last Changed Author: evancheng
Last Changed Rev: 70821
Last Changed Date: 2009-05-04 05:30:11 +0200 (Mon, 04 May 2009)

And config.status --version:

$ ./config.status --version
llvm config.status 2.6svn
configured by ./configure, generated by GNU Autoconf 2.60,
  with options "'--prefix=/opt/llvm-svn' 'CC=gcc-4'
'CFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include/'
'LDFLAGS=-L/opt/llvm-svn/lib/'
'CPPFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include/' 'CXX=g++-4'
'CXXFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include/'"

Copyright (C) 2006 Free Software Foundation, Inc.
This config.status script is free software; the Free Software Foundation
gives unlimited permission to copy, distribute and modify it.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 03:12:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 03:12:28 -0500
Subject: [LLVMbugs] [Bug 4145] Bulds fail on Cygwin with error message about
 dwarf version
In-Reply-To: <bug-4145-206@http.llvm.org/bugs/>
Message-ID: <200905040812.n448CSFw016764@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4145


Anton Korobeynikov <asl at math.spbu.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |asl at math.spbu.ru
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Anton Korobeynikov <asl at math.spbu.ru>  2009-05-04 03:12:28 ---
(In reply to comment #0)
> /usr/lib/gcc/i686-pc-cygwin/4.3.2/../../../../i686-pc-cygwin/bin/ld: Dwarf
> Error: found dwarf version '8242', this reader only handles version 2
> information.
> /c/temp/llvm-svn/Debug/lib/LLVMXCore.o:XCoreTargetMachine.cpp:(.rdata$_ZTVN4llvm18XCoreTargetMachineE[vtable
> for llvm::XCoreTargetMachine]+0x4c): undefined reference to
> `llvm::LLVMTargetMachine::addPassesToEmitMachineCode(llvm::PassManagerBase&,
> llvm::MachineCodeEmitter&, llvm::CodeGenOpt::Level)'
Your gcc is broken then - it generates invalid debug information. This is not
an LLVM bug. Try release version instead, maybe it will work.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 03:29:07 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 03:29:07 -0500
Subject: [LLVMbugs] [Bug 4146] New: scalarrepl crash on i1 store
Message-ID: <bug-4146-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4146

           Summary: scalarrepl crash on i1 store
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


$ opt -scalarrepl testcase.bc
opt: llvm/lib/VMCore/Instructions.cpp:2356:
llvm::TruncInst::TruncInst(llvm::Value*, const llvm::Type*, const std::string&,
llvm::Instruction*): Assertion `castIsValid(getOpcode(), S, Ty) && "Illegal
Trunc"' failed.

Here's the testcase:

        %wrapper = type { i1 }

define void @f() {
entry:
        %w = alloca %wrapper, align 8           ; <%wrapper*> [#uses=1]
        %0 = getelementptr %wrapper* %w, i64 0, i32 0           ; <i1*>
[#uses=1]
        store i1 true, i1* %0
        ret void
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 03:30:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 03:30:41 -0500
Subject: [LLVMbugs] [Bug 4147] New: llvm-ar rs and llvm-ranlib do not
 produce archive index
Message-ID: <bug-4147-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4147

           Summary: llvm-ar rs and llvm-ranlib do not produce archive index
           Product: tools
           Version: 2.5
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: blocker
          Priority: P2
         Component: llvm-ranlib
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: wagner at it.neclab.eu
                CC: llvmbugs at cs.uiuc.edu


Using LLVM-2.5 I have the problem not being able to link against llvm-ar
created libs, because the linker is missing the index:
> could not read symbols: Archive has no index; run ranlib to add one

However the library was created using 'llvm-ar rs'.

Manually creating the index using 'llvm-ranlib' does not help either.

If the library is created with 'llvm-ar r' only and 'llvm-ranlib' is used to
add the index later, the library file is not altered, size and content do not
change after calling ranlib.

'llvm-ranlib -verbose' does not print a table either.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 04:39:32 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 04:39:32 -0500
Subject: [LLVMbugs] [Bug 4145] Bulds fail on Cygwin with error message about
 dwarf version
In-Reply-To: <bug-4145-206@http.llvm.org/bugs/>
Message-ID: <200905040939.n449dWvR020344@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4145


Magnus Reftel <magnus.reftel at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|INVALID                     |




--- Comment #3 from Magnus Reftel <magnus.reftel at gmail.com>  2009-05-04 04:39:26 ---
Building for debug on a supported platform with the currently shipping version
on the complier should work, or it should be listed as a known issue in the
release notes.

Anyway, the problem may not be the dwarf version, but undefined references. In
the clean build output (see attachment 2941), the dwarf version warnings are
preceeded by the following:

llvm[2]: Linking Debug executable llc
Info: resolving std::cerr  by linking to __imp___ZSt4cerr (auto-import)
Info: resolving std::basic_string<char, std::char_traits<char>,
std::allocator<char> >::_Rep::_S_max_size by linking to
__imp___ZNSs4_Rep11_S_max_sizeE (auto-import)
Info: resolving std::basic_string<char, std::char_traits<char>,
std::allocator<char> >::_Rep::_S_empty_rep_storage by linking to
__imp___ZNSs4_Rep20_S_empty_rep_storageE (auto-import)
Info: resolving std::nothrow  by linking to __imp___ZSt7nothrow (auto-import)
Info: resolving VTT for std::basic_ifstream<char, std::char_traits<char> > by
linking to __imp___ZTTSt14basic_ifstreamIcSt11char_traitsIcEE (auto-import)
Info: resolving vtable for std::basic_ifstream<char, std::char_traits<char> >
by linking to __imp___ZTVSt14basic_ifstreamIcSt11char_traitsIcEE (auto-import)
Info: resolving vtable for std::basic_filebuf<char, std::char_traits<char> > by
linking to __imp___ZTVSt13basic_filebufIcSt11char_traitsIcEE (auto-import)
Info: resolving vtable for std::basic_streambuf<char, std::char_traits<char> >
by linking to __imp___ZTVSt15basic_streambufIcSt11char_traitsIcEE (auto-import)
Info: resolving vtable for std::basic_ios<char, std::char_traits<char> > by
linking to __imp___ZTVSt9basic_iosIcSt11char_traitsIcEE (auto-import)
Info: resolving std::basic_string<char, std::char_traits<char>,
std::allocator<char> >::_Rep::_S_terminal by linking to
__imp___ZNSs4_Rep11_S_terminalE (auto-import)
Info: resolving VTT for std::basic_ofstream<char, std::char_traits<char> > by
linking to __imp___ZTTSt14basic_ofstreamIcSt11char_traitsIcEE (auto-import)
Info: resolving vtable for std::basic_ofstream<char, std::char_traits<char> >
by linking to __imp___ZTVSt14basic_ofstreamIcSt11char_traitsIcEE (auto-import)
Info: resolving VTT for std::basic_ostringstream<char, std::char_traits<char>,
std::allocator<char> > by linking to
__imp___ZTTSt19basic_ostringstreamIcSt11char_traitsIcESaIcEE (auto-import)
Info: resolving vtable for std::basic_ostringstream<char,
std::char_traits<char>, std::allocator<char> > by linking to
__imp___ZTVSt19basic_ostringstreamIcSt11char_traitsIcESaIcEE (auto-import)
Info: resolving vtable for std::basic_stringbuf<char, std::char_traits<char>,
std::allocator<char> > by linking to
__imp___ZTVSt15basic_stringbufIcSt11char_traitsIcESaIcEE (auto-import)
Info: resolving vtable for std::basic_stringstream<char,
std::char_traits<char>, std::allocator<char> > by linking to
__imp___ZTVSt18basic_stringstreamIcSt11char_traitsIcESaIcEE (auto-import)
Info: resolving VTT for std::basic_stringstream<char, std::char_traits<char>,
std::allocator<char> > by linking to
__imp___ZTTSt18basic_stringstreamIcSt11char_traitsIcESaIcEE (auto-import)
Info: resolving std::cout  by linking to __imp___ZSt4cout (auto-import)
Info: resolving std::cin  by linking to __imp___ZSt3cin
(auto-import/usr/lib/gcc/i686-pc-cygwin/4.3.2/../../../../i686-pc-cygwin/bin/ld:
warning: auto-importing has been activated without --enable-auto-import
specified on the command line.
This should work unless it involves constant data structures referencing
symbols from auto-imported DLLs./c/temp/llvm-svn/tools/llc/Debug/llc.o: In
function `ForceCodegenLinking':
/c/temp/llvm-svn/include/llvm/CodeGen/LinkAllCodegenComponents.h:46: undefined
reference to `llvm::createBURRListDAGScheduler(llvm::SelectionDAGISel*,
llvm::CodeGenOpt::Level)'
/c/temp/llvm-svn/include/llvm/CodeGen/LinkAllCodegenComponents.h:47: undefined
reference to `llvm::createTDRRListDAGScheduler(llvm::SelectionDAGISel*,
llvm::CodeGenOpt::Level)'
/c/temp/llvm-svn/include/llvm/CodeGen/LinkAllCodegenComponents.h:48: undefined
reference to `llvm::createTDListDAGScheduler(llvm::SelectionDAGISel*,
llvm::CodeGenOpt::Level)'
/c/temp/llvm-svn/include/llvm/CodeGen/LinkAllCodegenComponents.h:49: undefined
reference to `llvm::createFastDAGScheduler(llvm::SelectionDAGISel*,
llvm::CodeGenOpt::Level)'
/c/temp/llvm-svn/include/llvm/CodeGen/LinkAllCodegenComponents.h:50: undefined
reference to `llvm::createDefaultScheduler(llvm::SelectionDAGISel*,
llvm::CodeGenOpt::Level)'
/c/temp/llvm-svn/tools/llc/Debug/llc.o: In function `main':
/c/temp/llvm-svn/tools/llc/llc.cpp:258: undefined reference to
`llvm::cl::opt<char, false, llvm::cl::parser<char> >::operator char() const'
/c/temp/llvm-svn/tools/llc/Debug/llc.o: In function `basic_parser':
/c/temp/llvm-svn/include/llvm/Support/CommandLine.h:531: undefined reference to
`vtable for llvm::cl::basic_parser<char>'
/c/temp/llvm-svn/tools/llc/Debug/llc.o: In function `parser':
/c/temp/llvm-svn/include/llvm/Support/CommandLine.h:686: undefined reference to
`vtable for llvm::cl::parser<char>'
/c/temp/llvm-svn/tools/llc/Debug/llc.o:/c/temp/llvm-svn/include/llvm/Support/CommandLine.h:281:
undefined reference to `llvm::cl::opt<char, false, llvm::cl::parser<char>
>::setInitialValue(char const&)'
/c/temp/llvm-svn/tools/llc/Debug/llc.o: In function `~basic_parser':
/c/temp/llvm-svn/include/llvm/Support/CommandLine.h:531: undefined reference to
`vtable for llvm::cl::basic_parser<char>'
/c/temp/llvm-svn/tools/llc/Debug/llc.o: In function `~parser':
/c/temp/llvm-svn/include/llvm/Support/CommandLine.h:686: undefined reference to
`vtable for llvm::cl::parser<char>'
/c/temp/llvm-svn/tools/llc/Debug/llc.o: In function `~opt':
/c/temp/llvm-svn/include/llvm/Support/CommandLine.h:833: undefined reference to
`vtable for llvm::cl::opt<char, false, llvm::cl::parser<char> >'
/c/temp/llvm-svn/tools/llc/Debug/llc.o: In function `opt<char [2],
llvm::cl::desc, llvm::cl::FormattingFlags, llvm::cl::NumOccurrences,
llvm::cl::initializer<char> >':
/c/temp/llvm-svn/include/llvm/Support/CommandLine.h:909: undefined reference to
`vtable for llvm::cl::opt<char, false, llvm::cl::parser<char> >'
/c/temp/llvm-svn/include/llvm/Support/CommandLine.h:912: undefined reference to
`llvm::cl::opt<char, false, llvm::cl::parser<char> >::done()'
/usr/lib/gcc/i686-pc-cygwin/4.3.2/../../../../i686-pc-cygwin/bin/ld: Dwarf
Error: Abbrev offset (2007759) greater than or equal to .debug_abbrev size
(24497).
/c/temp/llvm-svn/Debug/lib/LLVMMSP430.o:MSP430AsmPrinter.cpp:(.text+0x87c):
undefined reference to `llvm::AsmPrinter::AsmPrinter(llvm::raw_ostream&,
llvm::TargetMachine&, llvm::TargetAsmInfo const*, llvm::CodeGenOpt::Level,
bool)'
/usr/lib/gcc/i686-pc-cygwin/4.3.2/../../../../i686-pc-cygwin/bin/ld: Dwarf
Error: found dwarf version '30', this reader only handles version 2
information.
/c/temp/llvm-svn/Debug/lib/LLVMMSP430.o:MSP430AsmPrinter.cpp:(.text+0xab5):
undefined reference to `llvm::DwarfWriter::RecordSourceLine(unsigned int,
unsigned int, llvm::DICompileUnit)'

The GCC version I'm using is

$ g++-4 --version
g++-4 (GCC) 4.3.2 20080827 (beta) 2
Copyright (C) 2008 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

and ld is 

$ ld --version
GNU ld (GNU Binutils) 2.18.50.20080625
Copyright 2007 Free Software Foundation, Inc.
This program is free software; you may redistribute it under the terms of
the GNU General Public License version 3 or (at your option) a later version.
This program has absolutely no warranty.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 05:32:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 05:32:09 -0500
Subject: [LLVMbugs] [Bug 4145] Bulds fail on Cygwin with error message about
 dwarf version
In-Reply-To: <bug-4145-206@http.llvm.org/bugs/>
Message-ID: <200905041032.n44AW9sU021999@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4145


Anton Korobeynikov <asl at math.spbu.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Anton Korobeynikov <asl at math.spbu.ru>  2009-05-04 05:31:32 ---
(In reply to comment #3)
> Building for debug on a supported platform with the currently shipping version
> on the complier should work, or it should be listed as a known issue in the
> release notes.
Note that you're using beta snapshot-based compiler which can be pretty buggy
(this already happened with cygwin in the past) especially for gcc secondary
priority platform. I verified that mingw debug build is ok, that's why I assume
that nothing is broken in LLVM codebase itself.

There were already problems with linkers for COFF targets which prevented LLVM
build someday (see note in "Broken compilers and tools" section).

I added an entry here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077100.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 06:03:52 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 06:03:52 -0500
Subject: [LLVMbugs] [Bug 4149] New: JIT fails when requiring exact size
Message-ID: <bug-4149-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4149

           Summary: JIT fails when requiring exact size
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Windows NT
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicolas at capens.net
                CC: llvmbugs at cs.uiuc.edu


When JITMemoryManager.SizeRequired is set to true, recent revisions hit the
assert in X86InstrInfo::sizeOfImm.

The SizeRequired flag is very useful when using a custom JITMemoryManager
implementation (e.g. to compile functions eagerly instead of lazily).
Unfortunately this feature was broken in revision 68645. For reproducing this
it should suffice to set SizeRequired to true in the constructor of
JITMemoryManager, and then run the Fibonacci sample.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 07:14:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 07:14:31 -0500
Subject: [LLVMbugs] [Bug 4152] New: [Linux kernel] invalid asm constraint 'i'
Message-ID: <bug-4152-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4152

           Summary: [Linux kernel] invalid asm constraint 'i'
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jgarzik at pobox.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 4068


Created an attachment (id=2943)
 --> (http://llvm.org/bugs/attachment.cgi?id=2943)
pre-processed source code: arch/x86/kernel/process_64.c

clang -MD -MF arch/x86/kernel/.process_64.o.d  -nostdinc -isystem include
-Iinclude  -I/spare/repo/linux-2.6/arch/x86/include -include
include/linux/autoconf.h -D__KERNEL__ -Wall -Wundef -Wstrict-prototypes
-Wno-trigraphs -fno-strict-aliasing -fno-common
-Werror-implicit-function-declaration -Os -m64 -mtune=generic -mno-red-zone
-mcmodel=kernel -funit-at-a-time -maccumulate-outgoing-args -DCONFIG_AS_CFI=1
-DCONFIG_AS_CFI_SIGNAL_FRAME=1 -pipe -Wno-sign-compare
-fno-asynchronous-unwind-tables -mno-sse -mno-mmx -mno-sse2 -mno-3dnow
-Wframe-larger-than=2048 -fno-omit-frame-pointer -fno-optimize-sibling-calls -g
-pg -Wdeclaration-after-statement -Wno-pointer-sign -fwrapv -fno-dwarf2-cfi-asm
  -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(process_64)" 
-D"KBUILD_MODNAME=KBUILD_STR(process_64)" -D"DEBUG_HASH=48" -D"DEBUG_HASH2=34"
-c -o arch/x86/kernel/.tmp_process_64.o arch/x86/kernel/process_64.c
clang: warning: the clang compiler does not yet support '-pg'
warning: unknown warning option: -Wframe-larger-than=2048
llvm: error: Invalid operand for inline asm constraint 'i'!
make[1]: *** [arch/x86/kernel/process_64.o] Error 1
make: *** [arch/x86/kernel] Error 2


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 07:49:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 07:49:38 -0500
Subject: [LLVMbugs] [Bug 3967] llvm-gcc handles regparm incorrectly
In-Reply-To: <bug-3967-206@http.llvm.org/bugs/>
Message-ID: <200905041249.n44CncM6026640@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3967


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #6 from Duncan Sands <baldrick at free.fr>  2009-05-04 07:49:35 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077102.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 08:03:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 08:03:34 -0500
Subject: [LLVMbugs] [Bug 4145] Linking lcc fails on Cygwin
In-Reply-To: <bug-4145-206@http.llvm.org/bugs/>
Message-ID: <200905041303.n44D3Y5Y027118@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4145


Magnus Reftel <magnus.reftel at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |
            Summary|Bulds fail on Cygwin with   |Linking lcc fails on Cygwin
                   |error message about dwarf   |
                   |version                     |




--- Comment #5 from Magnus Reftel <magnus.reftel at gmail.com>  2009-05-04 08:03:30 ---
It appears not to be related to debug or to the compiler version. I've build
for retail (using --enable-optimized when configuring) and build with GCC 4 and
GCC 3.4.4, and I get link errors in both. The warnings about the dwarf version
go away though, so it seems that was a red herring.

$ svn info
Path: .
URL: http://llvm.org/svn/llvm-project/llvm/trunk
Repository Root: http://llvm.org/svn/llvm-project
Repository UUID: 91177308-0d34-0410-b5e6-96231b3b80d8
Revision: 70845
Node Kind: directory
Schedule: normal
Last Changed Author: evancheng
Last Changed Rev: 70821
Last Changed Date: 2009-05-04 05:30:11 +0200 (Mon, 04 May 2009)


GCC 4

$ ./config.status --version
llvm config.status 2.6svn
configured by ./configure, generated by GNU Autoconf 2.60,
  with options "'--prefix=/opt/llvm-svn' '--enable-optimized' 'CC=gcc-4'
'CFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include' 'LDFLAGS=-L/opt/llvm-svn/lib'
'CPPFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include' 'CXX=g++-4'
'CXXFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include'"

Copyright (C) 2006 Free Software Foundation, Inc.
This config.status script is free software; the Free Software Foundation
gives unlimited permission to copy, distribute and modify it.

Attaching clean build output as 78045_optimized_gcc4.txt


GCC 3

$ ./config.status --version
llvm config.status 2.6svn
configured by ./configure, generated by GNU Autoconf 2.60,
  with options "'--prefix=/opt/llvm-svn' '--enable-optimized' 'CC=gcc'
'CFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include' 'LDFLAGS=-L/opt/llvm-svn/lib'
'CPPFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include' 'CXX=g++'
'CXXFLAGS=-I/opt/llvm-svn/lib/libffi-3.0.8/include'"

Copyright (C) 2006 Free Software Foundation, Inc.
This config.status script is free software; the Free Software Foundation
gives unlimited permission to copy, distribute and modify it.


Attaching clean build output as 78045_optimized_gcc3.txt


$ gcc --version
gcc (GCC) 3.4.4 (cygming special, gdc 0.12, using dmd 0.125)
Copyright (C) 2004 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 11:51:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 11:51:36 -0500
Subject: [LLVMbugs] [Bug 4141] PointerMayBeCaptured() can do better for
 readonly/ readnone function calls.
In-Reply-To: <bug-4141-206@http.llvm.org/bugs/>
Message-ID: <200905041651.n44Gpac8002212@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4141


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Duncan Sands <baldrick at free.fr>  2009-05-04 11:51:35 ---
I applied your patch with some tweaks:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077110.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 14:26:03 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 14:26:03 -0500
Subject: [LLVMbugs] [Bug 4154] New: -fast and -tailcallopt don't combine
Message-ID: <bug-4154-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4154

           Summary: -fast and -tailcallopt don't combine
           Product: libraries
           Version: 2.5
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: arigo at tunes.org
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2951)
 --> (http://llvm.org/bugs/attachment.cgi?id=2951)
wrong result of "llc -fast -tailcallopt"

"llc -fast -tailcallopt" gives wrong results on the small attached file.

The issue is that there is a "addl 4, %esp" after the "call *%eax", which is
(correctly) not there if compiled with "llc -tailcallopt".

This seems to be tailcallopt not generating the correct calls in "fast" mode. 
It was discovered by playing with the JIT compiler.  (Obviously, the call in
question is not meant to be a tail call; this is part of a more complete set of
functions, and some other calls elsewhere are the ones meant to be tail calls.)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 14:52:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 14:52:36 -0500
Subject: [LLVMbugs] [Bug 4154] -fast and -tailcallopt don't combine
In-Reply-To: <bug-4154-206@http.llvm.org/bugs/>
Message-ID: <200905041952.n44Jqa82010088@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4154


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gohman at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Dan Gohman <gohman at apple.com>  2009-05-04 14:52:35 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077128.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 15:18:27 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 15:18:27 -0500
Subject: [LLVMbugs] [Bug 4147] llvm-ar rs and llvm-ranlib do not produce
	archive index
In-Reply-To: <bug-4147-206@http.llvm.org/bugs/>
Message-ID: <200905042018.n44KIRqf010953@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4147


Nick Lewycky <nlewycky at google.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nlewycky at google.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Nick Lewycky <nlewycky at google.com>  2009-05-04 15:18:26 ---
The llvm-ar format can't be read by your system linker on linux. Use normal
system ar and ranlib.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May  4 23:57:57 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 4 May 2009 23:57:57 -0500
Subject: [LLVMbugs] [Bug 4156] New: ICE: Assertion `isa<X>(Val) &&
 "cast<Ty>() argument of incompatible type!"' failed.
Message-ID: <bug-4156-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4156

           Summary: ICE: Assertion `isa<X>(Val) && "cast<Ty>() argument of
                    incompatible type!"' failed.
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu, jxyang at cs.utah.edu


Seen using r70938 on Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp163$ llvm-gcc -O small.c
small.c: In function ???int322???:
small.c:51: warning: assignment makes integer from pointer without a cast
cc1: /home/regehr/z/tmp/llvm-r70938/include/llvm/Support/Casting.h:199:
typename llvm::cast_retty<To, From>::ret_type llvm::cast(const Y&) [with X =
llvm::CallInst, Y = llvm::Instruction*]: Assertion `isa<X>(Val) && "cast<Ty>()
argument of incompatible type!"' failed.
small.c: At top level:
small.c:58: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.

regehr at john-home:~/volatile/tmp163$ cat small.c

typedef char int8_t;
typedef int int16_t;
typedef char uint8_t;
typedef int uint32_t;
uint8_t
safe_add_func_int8_t_s_s (int8_t _si1, uint8_t _si2)
{
}

uint8_t
safe_add_func_int16_t_s_s (int16_t _si1, uint8_t _si2)
{
  return 32767 - _si2 < -_si2 ? : 1;
}

uint32_t
safe_sub_func_uint32_t_u_u (uint32_t _ui1, uint32_t _ui2)
{
}
uint8_t g_11;
uint8_t g_38;
uint8_t
func_12 (int8_t p_14, uint8_t p_16, uint8_t p_17, uint8_t p_19)
{
  return g_11;
}

uint8_t
func_27 (uint32p_28)
{
  return g_38;
}

uint8_t
func_30 (int64p_31)
{
}
uint8_t
func_33 (uint32p_35)
{
}

int322 (uint32_t p_43)
{
  uint8_t l_44;
  uint8_t l_45;
  for (1; p_43; safe_add_func_int8_t_s_s)
    {
      uint8_t l_48;
      for (1; p_43; 1)
        for (1; l_44; l_44 = safe_sub_func_uint32_t_u_u)
          {
            func_30 (func_27 (l_45));
            func_33 (func_27 (1) || safe_add_func_int16_t_s_s (g_38, 1));
            l_45 = func_12 (p_43, l_48, g_38, 1);
          }
    }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May  5 03:39:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 5 May 2009 03:39:09 -0500
Subject: [LLVMbugs] [Bug 4157] New: Using ENV generates wrong source code
Message-ID: <bug-4157-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4157

           Summary: Using ENV generates wrong source code
           Product: tools
           Version: trunk
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: TableGen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: martin.nowack at inf.tu-dresden.de
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2953)
 --> (http://llvm.org/bugs/attachment.cgi?id=2953)
Patch for ENV bug

Using the ENV macro the generated code contains an additional brace ')' - the
attached patch fixes the behavior.
The problem persists at least since 2.5 and also in trunk.

Thanks.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May  5 04:24:16 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 5 May 2009 04:24:16 -0500
Subject: [LLVMbugs] [Bug 4158] New: support arbitrary-width integers in the
	C backend
Message-ID: <bug-4158-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4158

           Summary: support arbitrary-width integers in the C backend
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: Backend: C
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Currently it is not possible to use the C backend if the bitcode contains i256,
i512, or anything >128, due to:

extern llvmUInt128llc: CBackend.cpp:484:
llvm::raw_ostream&<unnamed>::CWriter::printSimpleType(llvm::raw_ostream&, const
llvm::Type*, bool, const std::string&): Assertion `NumBits <= 128 && "Bit
widths > 128 not implemented yet"' failed.

Would it be possible for the C backend to use the type legalization framework,
and tell it that it supports only i8, i16, i32, i64, (and possibly i128)?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May  5 07:34:47 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 5 May 2009 07:34:47 -0500
Subject: [LLVMbugs] [Bug 4157] Using ENV generates wrong source code
In-Reply-To: <bug-4157-206@http.llvm.org/bugs/>
Message-ID: <200905051234.n45CYl9U022139@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4157


Mikhail Glushenkov <foldr at codedgers.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Mikhail Glushenkov <foldr at codedgers.com>  2009-05-05 07:34:43 ---
Applied, thanks!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May  5 11:04:42 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 5 May 2009 11:04:42 -0500
Subject: [LLVMbugs] [Bug 4159] New: forward_as does not use second argument
Message-ID: <bug-4159-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4159

           Summary: forward_as does not use second argument
           Product: tools
           Version: trunk
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: TableGen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: martin.nowack at inf.tu-dresden.de
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2956)
 --> (http://llvm.org/bugs/attachment.cgi?id=2956)
patch to fix forward

Using forward_as to substitute a parameter with a new one does not use the new
parameter in the generated file. This bug at least exists for 2.5

Patch attached to fix this. 

By the way, please update the online documentation
(http://www.llvm.org/docs/CompilerDriver.html).
The provided example for forward_as omits the comma between the arguments.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May  5 20:41:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 5 May 2009 20:41:28 -0500
Subject: [LLVMbugs] [Bug 4159] forward_as does not use second argument
In-Reply-To: <bug-4159-206@http.llvm.org/bugs/>
Message-ID: <200905060141.n461fSdK020405@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4159


Mikhail Glushenkov <foldr at codedgers.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Mikhail Glushenkov <foldr at codedgers.com>  2009-05-05 20:41:27 ---
Applied, thanks!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May  5 23:34:23 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 5 May 2009 23:34:23 -0500
Subject: [LLVMbugs] [Bug 4161] New: need to pass codegen options through to
	lto
Message-ID: <bug-4161-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4161

           Summary: need to pass codegen options through to lto
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: lto
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: rafael.espindola at gmail.com, llvmbugs at cs.uiuc.edu


Some LTO options such as PIC and FP elimination should be passed from the
frontend through to the backend. For debugging, it'd be nice to pass -fast
through too.

Some of the options (like target triple) are stored in the .bc, but we need to
sort out what options go where and how we make them flow through from llvm-gcc
to the gold plugin, into libLTO and make it all the way to codegen. Every
option that llvm-gcc supports should probably have a matching option unless
it's in the bitcode.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 01:55:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 01:55:14 -0500
Subject: [LLVMbugs] [Bug 3277] llvm-g++ front-end outputing code that is not
 folded by irbuilder?
In-Reply-To: <bug-3277-206@http.llvm.org/bugs/>
Message-ID: <200905060655.n466tENn031784@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3277


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #14 from Duncan Sands <baldrick at free.fr>  2009-05-06 01:55:13 ---
I guess this can be considered fixed now that we have ODR linkage types.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 01:59:16 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 01:59:16 -0500
Subject: [LLVMbugs] [Bug 2163] Codegen illegally reorders division by zero
 before function call
In-Reply-To: <bug-2163-206@http.llvm.org/bugs/>
Message-ID: <200905060659.n466xGEn031945@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2163


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |baldrick at free.fr
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #4 from Duncan Sands <baldrick at free.fr>  2009-05-06 01:59:15 ---
This seems to have been fixed.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 02:08:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 02:08:12 -0500
Subject: [LLVMbugs] [Bug 3403] llc: "Do not know how to split the result of
 this operator!"
In-Reply-To: <bug-3403-206@http.llvm.org/bugs/>
Message-ID: <200905060708.n4678C32032238@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3403


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |baldrick at free.fr
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #1 from Duncan Sands <baldrick at free.fr>  2009-05-06 02:08:11 ---
This seems to have been fixed at some point.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 03:56:59 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 03:56:59 -0500
Subject: [LLVMbugs] [Bug 3754] FunctionAttrs pass marks function with
 MallocInst as readnone
In-Reply-To: <bug-3754-206@http.llvm.org/bugs/>
Message-ID: <200905060856.n468uwOZ014112@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3754


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Duncan Sands <baldrick at free.fr>  2009-05-06 03:55:44 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077169.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 06:25:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 06:25:28 -0500
Subject: [LLVMbugs] [Bug 4165] New: LLVM_ENABLE_THREADS
Message-ID: <bug-4165-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4165

           Summary: LLVM_ENABLE_THREADS
           Product: Build scripts
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Makefiles
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: thebohemian at gmx.net
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2963)
 --> (http://llvm.org/bugs/attachment.cgi?id=2963)
fix to use proper macro in lib/System/Mutex.cpp

In the file lib/System/Mutex.cpp the preprocessor macro ENABLE_THREADS is
evaluated. However there is no such macro when the cmake buildsystem is in use.
Instead there is LLVM_ENABLE_THREADS.

I believe it is the source file which needs to be fixed. However it is possible
that the autotools build system needs to be adjusted accordingly.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 06:32:03 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 06:32:03 -0500
Subject: [LLVMbugs] [Bug 4093] LegalizeDAG does not allow expansion of
 subtraction operation
In-Reply-To: <bug-4093-206@http.llvm.org/bugs/>
Message-ID: <200905061132.n46BW3ST020144@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4093


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Duncan Sands <baldrick at free.fr>  2009-05-06 06:31:56 ---
Applied here, with cleanups (spaces not tabs, use
AllOnesValue to be 128 bit safe, eliminate the
fp case - SUB is not valid for floating point, etc):
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077170.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 07:09:47 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 07:09:47 -0500
Subject: [LLVMbugs] [Bug 4060] building with g++-4.4 causes ocaml test to
	fail
In-Reply-To: <bug-4060-206@http.llvm.org/bugs/>
Message-ID: <200905061209.n46C9lIG021465@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4060


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Duncan Sands <baldrick at free.fr>  2009-05-06 07:09:46 ---
Nick fixed this.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 07:22:39 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 07:22:39 -0500
Subject: [LLVMbugs] [Bug 2752] Ocaml bindings for parameter attributes
In-Reply-To: <bug-2752-206@http.llvm.org/bugs/>
Message-ID: <200905061222.n46CMdcZ021916@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2752


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #13 from Duncan Sands <baldrick at free.fr>  2009-05-06 07:22:23 ---
Applied here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077171.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 07:34:18 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 07:34:18 -0500
Subject: [LLVMbugs] [Bug 4166] New: Assert while trying to compile using ldc
 with arm as target on x86 host
Message-ID: <bug-4166-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4166

           Summary: Assert while trying to compile using ldc with arm as
                    target on x86 host
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fredridv at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2964)
 --> (http://llvm.org/bugs/attachment.cgi?id=2964)
Output from 'ldc -c tango/net/cluster/tina/CacheThread.d -oftest.bc'

A slightly modified ldc (to get it to cross compile for an arm target) running
on linux x86 gives the following error while compiling a file in the tango
library:

ldc: TwoAddressInstructionPass.cpp:800: virtual
bool<unnamed>::TwoAddressInstructionPass::runOnMachineFunction(llvm::MachineFunction&):
Assertion `TargetRegisterInfo::isVirtualRegister(regB) && "cannot update
physical register live information"' failed.
0   ldc 0x08add088
Stack dump:
0.      Running pass 'Two-Address instruction pass' on function
'@_D5tango3net7cluster4tina11CacheThread11CacheThread4loadMFE5tango3net7cluster4tina14ProtocolWriter14ProtocolWriter7CommandAaAaZv'
./build-tango.sh: line 110: 32166 Aborted                 (core dumped) $DC
$ARCH $WARN -c $INLINE $DEBUG $RELEASE $POSIXFLAG -version=Tango -of$OBJNAME
$FILENAME
Compilation of tango/net/cluster/tina/CacheThread.d failed

ldc is from hg, Mon May 04 13:32:16 2009 +0200, rev. 1299:847b767b2d0b,

tango is 0.99.8,

llvm is svn rev. 70630.

I couldn't get bugpoint to reduce the file size, so I attach the bitcode file
produced by ldc.

The error is reproducible by running 'llc -march=arm test.bc'.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 08:47:23 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 08:47:23 -0500
Subject: [LLVMbugs] [Bug 2655] memcpyopt doesn't zap an alloca + memmove
In-Reply-To: <bug-2655-206@http.llvm.org/bugs/>
Message-ID: <200905061347.n46DlNUP024856@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2655


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #8 from Duncan Sands <baldrick at free.fr>  2009-05-06 08:47:16 ---
The memcpy is now no longer present, though perhaps for the wrong reasons.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 09:28:56 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 09:28:56 -0500
Subject: [LLVMbugs] [Bug 4165] LLVM_ENABLE_THREADS
In-Reply-To: <bug-4165-206@http.llvm.org/bugs/>
Message-ID: <200905061428.n46ESuPR026365@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4165


??scar Fuentes <ofv at wanadoo.es> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from ??scar Fuentes <ofv at wanadoo.es>  2009-05-06 09:28:53 ---
Applied. Thanks!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 11:21:09 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 11:21:09 -0500
Subject: [LLVMbugs] [Bug 2655] memcpyopt doesn't zap an alloca + memmove
In-Reply-To: <bug-2655-206@http.llvm.org/bugs/>
Message-ID: <200905061621.n46GL9NU030387@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2655


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |




--- Comment #9 from Chris Lattner <clattner at apple.com>  2009-05-06 11:21:06 ---
The memcpy isn't present, but the memmove is.  Taking the code from the
description, adding nocapture to the stret result and using:
$ llvm-as < t.ll  | opt -memcpyopt | llvm-dis

gives:
define void @foo(%struct.out* nocapture sret %agg.result) {
entry:
        %out = alloca %struct.out, align 4              ; <%struct.out*>
[#uses=3]
        %tmp = getelementptr %struct.out* %out, i32 0, i32 0            ;
<i32*> [#uses=1]
        %call = call i32 @bar()         ; <i32> [#uses=1]
        store i32 %call, i32* %tmp
        %tmp1 = getelementptr %struct.out* %out, i32 0, i32 1           ;
<i32*> [#uses=1]
        %call2 = call i32 @bar()                ; <i32> [#uses=1]
        store i32 %call2, i32* %tmp1
        %tmp3 = bitcast %struct.out* %agg.result to i8*         ; <i8*>
[#uses=1]
        %tmp4 = bitcast %struct.out* %out to i8*                ; <i8*>
[#uses=1]
        call void @llvm.memmove.i32(i8* %tmp3, i8* %tmp4, i32 8, i32 4)
        ret void
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 12:46:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 12:46:08 -0500
Subject: [LLVMbugs] [Bug 2163] Codegen illegally reorders division by zero
 before function call
In-Reply-To: <bug-2163-206@http.llvm.org/bugs/>
Message-ID: <200905061746.n46Hk85g000822@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2163


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gohman at apple.com
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |




--- Comment #5 from Dan Gohman <gohman at apple.com>  2009-05-06 12:45:56 ---
Here's a new testcase which still reproduces the problem:

declare i32 @x()

define i32 @foo(i32 %a, i32 %b) nounwind {
        %t0 = tail call i32 @x() nounwind
        %t1 = sdiv i32 %a, %b
        ret i32 %t1
}

Currently the div is scheduled before the call on x86 and x86-64.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 13:28:53 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 13:28:53 -0500
Subject: [LLVMbugs] [Bug 4166] Assert while trying to compile using ldc with
 arm as target on x86 host
In-Reply-To: <bug-4166-206@http.llvm.org/bugs/>
Message-ID: <200905061828.n46ISru5002346@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4166


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Evan Cheng <evan.cheng at apple.com>  2009-05-06 13:28:49 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077185.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 15:26:32 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 15:26:32 -0500
Subject: [LLVMbugs] [Bug 4169] New: llvm-config --src-root should print
	canonical path
Message-ID: <bug-4169-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4169

           Summary: llvm-config --src-root should print canonical path
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: minor
          Priority: P2
         Component: llvm-config
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fawzi at gmx.ch
                CC: llvmbugs at cs.uiuc.edu


llvm-config --src-root returns

/Users/fawzi/d/ldc/llvm-build/../llvm/

in my configuration, and that becomes invalid if the build directory is
deleted... even if the src directory is still present. 
It is not difficult to keep a directory, but returning a canonicalized path
would be better.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 16:12:19 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 16:12:19 -0500
Subject: [LLVMbugs] [Bug 4126] conditional jump to nowhere
In-Reply-To: <bug-4126-206@http.llvm.org/bugs/>
Message-ID: <200905062112.n46LCJ3S008119@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4126


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Dan Gohman <gohman at apple.com>  2009-05-06 16:12:18 ---
Yes, fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077153.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 20:20:05 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 20:20:05 -0500
Subject: [LLVMbugs] [Bug 4170] New: llc crashes when emitting code for
 variable sized ( zero length) array
Message-ID: <bug-4170-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4170

           Summary: llc crashes when emitting code for variable sized (zero
                    length) array
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jvb at wongr.net
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2967)
 --> (http://llvm.org/bugs/attachment.cgi?id=2967)
test case

Hi

llc crashes on this test case with an assertion failure (.bc file is attached)

define void @test([0 x i32] %x) {
entry:
        %y = extractvalue [0 x i32] %x, 0
        ret void
}

llc:
/home/julian/rcs-data/llvm-svn/lib/CodeGen/SelectionDAG/SelectionDAG.cpp:4011:
llvm::SDVTList llvm::SelectionDAG::getVTList(const llvm::MVT*, unsigned int):
Assertion `0 && "Cannot have nodes without results!"' failed.

I am using a fresh SVN checkout:
$ llc --version
Low Level Virtual Machine (http://llvm.org/):
  llvm version 2.6svn
  Optimized build with assertions.
  Built May  7 2009(01:19:30).


I am not very familiar with the internals of LLVM but the problem might be that
ComputeValueVTs in SelectionDAGBuild.cpp thinks that a zero length array
comprises of no types.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May  6 21:01:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 6 May 2009 21:01:13 -0500
Subject: [LLVMbugs] [Bug 4171] New: darwin-string-literals test failure on
 ppc32 for utf-16 string
Message-ID: <bug-4171-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4171

           Summary: darwin-string-literals test failure on ppc32 for utf-16
                    string
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Keywords: miscompilation, portability
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: cianci66 at mac.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2969)
 --> (http://llvm.org/bugs/attachment.cgi?id=2969)
llvm-ir file produced by test.

The test is being run on powerpc-apple-darwin9.6.0 

The test expects

@__utf16_string_ = internal global [35 x i8] c"h\00e\00l\00l\00o\00 \00\92!
\00\03& \00\90! \00w\00o\00r\00l\00d\00\00", section "__TEXT,__ustring", align
2

to be produced. The test produces

@__utf16_string_ = internal global [35 x i8] c"\00h\00e\00l\00l\00o\00 !\92\00
&\03\00 !\90\00 \00w\00o\00r\00l\00d\00", section "__TEXT,__ustring", align 2

See attached .ll file.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 03:12:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 03:12:37 -0500
Subject: [LLVMbugs] [Bug 4145] Linking lcc fails on Cygwin
In-Reply-To: <bug-4145-206@http.llvm.org/bugs/>
Message-ID: <200905070812.n478Cbq1007706@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4145


Magnus Reftel <magnus.reftel at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |INVALID




--- Comment #11 from Magnus Reftel <magnus.reftel at gmail.com>  2009-05-07 03:12:31 ---
Re-tested with a pristine 70826, and it worked. There must have been some junk
in my working directory that didn't go away after "make clean" but still
managed to ruin my builds.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 09:17:27 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 09:17:27 -0500
Subject: [LLVMbugs] [Bug 4172] New: false positive: Branch condition
 evaluates to an uninitialized value
Message-ID: <bug-4172-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4172

           Summary: false positive: Branch condition evaluates to an
                    uninitialized value
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: dimhen at gmail.com
                CC: llvmbugs at cs.uiuc.edu


scan-build gcc -Wall -O3 tst.c 
produce

ANALYZE: tst.c foo
tst.c:15:5: warning: Branch condition evaluates to an uninitialized value.
    if(k==0)
    ^  ~
1 diagnostic generated.
scan-build: 1 bugs found.

I use rev. #71157

#pr -n tst.c

    1   struct blob {
    2       int n;
    3       void *p;
    4   } B;
    5
    6   extern void bar(struct blob*);
    7
    8   int foo(void)
    9   {
   10       int k;
   11       B.n = sizeof(k);
   12       B.p = (void*) &k;
   13       bar(&B);
   14
   15       if(k==0)
   16           return 0;
   17       else
   18           return 1;
   19   }


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 09:43:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 09:43:36 -0500
Subject: [LLVMbugs] [Bug 4173] New: false positive: Dereference of null
	pointer
Message-ID: <bug-4173-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4173

           Summary: false positive: Dereference of null pointer
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: dimhen at gmail.com
                CC: llvmbugs at cs.uiuc.edu


scan-build gcc -Wall -O3 -c tst1.c

produce

ANALYZE: tst1.c foo
tst1.c:21:10: warning: Dereference of null pointer
            pB->n = 0;
            ~~~~^
1 diagnostic generated.
scan-build: 1 bugs found.

cat tst1.c
#include <stdio.h>

struct blob {
    int n;
    void *p;
};

extern void bar(FILE**);

void foo(int x, int *y)
{
    struct blob *pB = NULL;

    if( !(x & 0x16) )
        bar((FILE**) (void*)&pB);

    if( !(x & 0x16) ) {
        if(*y==0) {
            *y = 5;
        } else {
            pB->n = 0;
        }
    }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 10:42:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 10:42:20 -0500
Subject: [LLVMbugs] [Bug 4174] New: LoopIndexSplit executes loop body too
	many times!
Message-ID: <bug-4174-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4174

           Summary: LoopIndexSplit executes loop body too many times!
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: major
          Priority: P2
         Component: Loop Optimizer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jay.foad at antixlabs.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2970)
 --> (http://llvm.org/bugs/attachment.cgi?id=2970)
test case

With the attached test case, I get:

$ llvm-as -o - s.ll | lli
0
1
2
3

but:

$ llvm-as -o - s.ll | opt -f -loop-index-split | lli
0
1
2
2
3
4

LoopIndexSplit has completely broken how many times the loop iterates! This
test case is reduced from a loop that was mysteriously misbehaving deep inside
a C++ application.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 11:27:55 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 11:27:55 -0500
Subject: [LLVMbugs] [Bug 3841] clang's stdint.h is found before platform
	stdint.h
In-Reply-To: <bug-3841-206@http.llvm.org/bugs/>
Message-ID: <200905071627.n47GRtjf024952@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3841


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-07 11:27:46 ---
This should be fixed with r70806; let me know if there are still issues.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 12:06:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 12:06:25 -0500
Subject: [LLVMbugs] [Bug 4176] New: llvm-gcc regparm uses incorrect calling
 convention with structures
Message-ID: <bug-4176-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4176

           Summary: llvm-gcc regparm uses incorrect calling convention with
                    structures
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


struct a {unsigned char a,b,c;};
__attribute((regparm(1))) int a(struct a x) { return x.a+x.b+x.c; }
struct c {unsigned x,y;};
__attribute((regparm(2))) int d(struct c x) { return x.x+x.y; }
struct h {int a:5,b:5,c:5;};
__attribute((regparm(3))) int h(struct h x) { return x.a+x.b+x.c; }

llvm-gcc passes all of the above in memory instead of registers, which is not
compatible with gcc.  See bug 3782 comment 2 for more related testcases.

These are manufactured testcases, so I'm not sure they matter much, but I
figured I should file it for completeness.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 13:25:06 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 13:25:06 -0500
Subject: [LLVMbugs] [Bug 4173] false positive: Dereference of null pointer
In-Reply-To: <bug-4173-206@http.llvm.org/bugs/>
Message-ID: <200905071825.n47IP6cs028526@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4173


Ted Kremenek <kremenek at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|llvmbugs at cs.uiuc.edu        |xuzhongxing at gmail.com
             Status|NEW                         |ASSIGNED




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 14:06:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 14:06:20 -0500
Subject: [LLVMbugs] [Bug 4177] New: strcpy optimizer gets endian wrong
Message-ID: <bug-4177-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4177

           Summary: strcpy optimizer gets endian wrong
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: md at bu.edu
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2971)
 --> (http://llvm.org/bugs/attachment.cgi?id=2971)
sample program

I have a small program example which uses strcpy to concatenate the strings
"hello\0" and "\n\0".  When I compile through opt -O1, it is okay.  When I
compile through opt -O2 or higher, it makes a mistake.  bugpoint isolates this
to "Found miscompiling passes: -inline -simplify-libcalls -instcombine" but I
can point specifically to the line of asm output which is incorrect.

LLVM opt -O2 outputs:

  movw $2560, 5(%rbx)

and 2560 is 0x0A00 in hex.  The problem is that on a little-endian architecture
this write comes out to:

0x23e2010:      104 'h' 101 'e' 108 'l' 108 'l' 111 'o' 0 '\0'  10 '\n'

the nul and the newline are in the wrong order.  The instruction should have
been

  movw $10, 5(%rbx)

I attached bug.ll which can reproduce the error by:

  cat bug.ll | llvm-as | opt -O2 | llc > bug.s && gcc bug.s


LLVM from SVN (r70985)
Debian Linux 2.6.26 amd64


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 14:08:43 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 14:08:43 -0500
Subject: [LLVMbugs] [Bug 4178] New: missed optimization: global variable not
	inlined
Message-ID: <bug-4178-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4178

           Summary: missed optimization: global variable not inlined
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Windows NT
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Global Analyses
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: micah.villmow at amd.com
                CC: llvmbugs at cs.uiuc.edu


In this example, the global variable can be inlined and thus removing a load
operation.

; ModuleID = 'test-opt.bc'
target datalayout =
"e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16:16-i32:32:32:32-i64:64:64:64-f32:32:32:32-f64:64:64:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
target triple = "i686-pc-win32"
@y = global i32 3, align 4              ; <i32*> [#uses=1]

define void @test_and(i32 %x, i32 addrspace(1)* %res) {
entry:
        %tmp = load i32* @y             ; <i32> [#uses=2]
        %tmp1 = add i32 %tmp, %tmp      ; <i32> [#uses=1]
        %arrayidx = getelementptr i32 addrspace(1)* %res, i32 %tmp1            
; <i32 addrspace(1)*> [#uses=1]
        %arrayidx6 = getelementptr i32 addrspace(1)* %res, i32 %x              
; <i32 addrspace(1)*> [#uses=1]
        %tmp7 = load i32 addrspace(1)* %arrayidx6               ; <i32>
[#uses=1]
        store i32 %tmp7, i32 addrspace(1)* %arrayidx
        ret void
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 14:50:00 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 14:50:00 -0500
Subject: [LLVMbugs] [Bug 4178] missed optimization: global variable not
	inlined
In-Reply-To: <bug-4178-206@http.llvm.org/bugs/>
Message-ID: <200905071950.n47Jo0t3031212@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4178


Anton Korobeynikov <asl at math.spbu.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |asl at math.spbu.ru
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Anton Korobeynikov <asl at math.spbu.ru>  2009-05-07 14:50:00 ---
No, it cannot. global is externally visible and thus can be changed outside.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May  7 19:22:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 7 May 2009 19:22:36 -0500
Subject: [LLVMbugs] [Bug 4123] clang: Assertion `New != this &&
 "this->replaceAllUsesWith(this) is NOT valid!"' failed.
In-Reply-To: <bug-4123-206@http.llvm.org/bugs/>
Message-ID: <200905080022.n480MaYr007455@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4123


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-07 19:22:36 ---
Fixed in r71199.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 01:22:01 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 01:22:01 -0500
Subject: [LLVMbugs] [Bug 4149] JIT fails when requiring exact size
In-Reply-To: <bug-4149-206@http.llvm.org/bugs/>
Message-ID: <200905080622.n486M1FH017501@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4149


Nicolas Capens <nicolas at capens.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Nicolas Capens <nicolas at capens.net>  2009-05-08 01:22:01 ---
Fixed in revision 70929 by Zoltan Varga's patch.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 02:04:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 02:04:48 -0500
Subject: [LLVMbugs] [Bug 4177] strcpy optimizer gets endian wrong
In-Reply-To: <bug-4177-206@http.llvm.org/bugs/>
Message-ID: <200905080704.n4874mMu018732@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4177


Nick Lewycky <nicholas at mxc.ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #7 from Nick Lewycky <nicholas at mxc.ca>  2009-05-08 02:04:46 ---
I've committed the appropriate change to disable this optimization when no
TargetData is present (r71210) but it won't take effect until we actually have
a "no TargetData" state instead of substituting in a default target. That's
being tracked in bug 760.

Closing this bug as invalid.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 07:09:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 07:09:28 -0500
Subject: [LLVMbugs] [Bug 4180] New: MDNodes may be used after being destroyed
Message-ID: <bug-4180-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4180

           Summary: MDNodes may be used after being destroyed
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fvbommel at wxs.nl
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2975)
 --> (http://llvm.org/bugs/attachment.cgi?id=2975)
Valgrind output for example program

I'm seeing crashes due to invalid pointer accesses from
==24533==    at 0x498DDF:
llvm::FoldingSetImpl::FindNodeOrInsertPos(llvm::FoldingSetNodeID const&,
void*&) (in /home/urxae/tmp/llvm-test/mdnode/test)
==24533==    by 0x40C9C1: llvm::MDNode::get(llvm::Constant* const*, unsigned)
(in /home/urxae/tmp/llvm-test/mdnode/test)

These seem to be because MDNodes don't get removed from the MDNodeSet global
upon destruction.

I'm attaching full valgrind output for the following program:
=====
#include "llvm/Module.h"
#include "llvm/Type.h"
#include "llvm/DerivedTypes.h"
#include "llvm/Constants.h"
#include "llvm/GlobalVariable.h"

using namespace llvm;

int main() {
    const Type* Ty = Type::Int32Ty;
    char buf[20];
    const unsigned MAX = 100;
    for (unsigned i = 0; i < MAX; i++) {
        snprintf(buf, 20, "var_%u", i);

        Module* M = new Module("test");
        Constant* vals[] = {
            new GlobalVariable(Ty, true, GlobalValue::WeakODRLinkage,
UndefValue::get(Ty), buf, M),
        };
        Constant* node = MDNode::get(vals, sizeof(vals) / sizeof(vals[0]));

        delete M;
    }
    return 0;
}
=====
(If you can't reproduce, try increasing the loop count)

The patch is extremely trivial, once problem is diagnosed:
######
Index: lib/VMCore/Constants.cpp
===================================================================
--- lib/VMCore/Constants.cpp    (revision 71118)
+++ lib/VMCore/Constants.cpp    (working copy)
@@ -1714,6 +1714,7 @@
 }

 void MDNode::destroyConstant() {
+  MDNodeSet->RemoveNode(this);
   destroyConstantImpl();
 }

######


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 07:13:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 07:13:22 -0500
Subject: [LLVMbugs] [Bug 4180] MDNodes may be used after being destroyed
In-Reply-To: <bug-4180-206@http.llvm.org/bugs/>
Message-ID: <200905081213.n48CDMWg006498@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4180


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |baldrick at free.fr
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #1 from Duncan Sands <baldrick at free.fr>  2009-05-08 07:13:20 ---
Looks like another manifestation of PR4046, though I didn't check if this
is really the case.

*** This bug has been marked as a duplicate of bug 4046 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 08:35:59 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 08:35:59 -0500
Subject: [LLVMbugs] [Bug 4181] New: Crash in libgomp: invalid MVT
Message-ID: <bug-4181-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4181

           Summary: Crash in libgomp: invalid MVT
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


The following C code causes Clang to crash. I'm filing this bug here, because
the failed assertion is in libllvmcore:

int
main(int argc, char *argv[])
{
        __sync_val_compare_and_swap((void **)0, (void *)0, (void *)0);
}

Crash:

Assertion failed: (0 && "Invalid MVT!"), function getMVTString, file
/home/ed/projects/freebsd-clang/usr.bin/clang/lib/libllvmcore/../../../../contrib/llvm/lib/VMCore/ValueTypes.cpp,
line 86.
Stack dump:
0.      Program arguments:
/usr/obj/home/ed/projects/freebsd-clang/tmp/usr/bin/../libexec/clang-cc -triple
x86_64-undermydesk-freebsd8.0 -emit-llvm -disable-free -main-file-name
llvm-crash.c --relocation-model static --disable-fp-elim --unwind-tables=1
--mcpu=x86-64 --fmath-errno=1 -fdiagnostics-show-option -o foo.ll -x c
llvm-crash.c 
1.      <eof> parser at end of file
2.      llvm-crash.c:2:1: LLVM IR generation of declaration 'main'
3.      llvm-crash.c:3:1: LLVM IR generation of compound statement ('{}')

Code is based on libgomp sources.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 10:48:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 10:48:20 -0500
Subject: [LLVMbugs] [Bug 4181] Crash in libgomp: invalid MVT
In-Reply-To: <bug-4181-206@http.llvm.org/bugs/>
Message-ID: <200905081548.n48FmKs9014028@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4181


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-08 10:48:18 ---
And this just got fixed in r71218.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 10:55:15 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 10:55:15 -0500
Subject: [LLVMbugs] [Bug 4146] scalarrepl crash on i1 store
In-Reply-To: <bug-4146-206@http.llvm.org/bugs/>
Message-ID: <200905081555.n48FtFi9014308@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4146


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-05-08 10:55:10 ---
Fixed, thanks!
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077248.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 11:01:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 11:01:08 -0500
Subject: [LLVMbugs] [Bug 4182] New: RegionStore.cpp fails to compile under
	MSVC
Message-ID: <bug-4182-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4182

           Summary: RegionStore.cpp fails to compile under MSVC
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Keywords: portability
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: abbeyj at gmail.com
                CC: llvmbugs at cs.uiuc.edu


/lib/Analysis/RegionStore.cpp fails to compile under MSVC.  It gets confused by
having two different variables both named SR but with different types:
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Analysis/RegionStore.cpp?annotate=71213#l1056
and
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Analysis/RegionStore.cpp?annotate=71213#l1066

This can be worked around by renaming the second one to SRs (for example).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 11:10:39 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 11:10:39 -0500
Subject: [LLVMbugs] [Bug 3886] Crash when building Xorg/Cairo's pixman
In-Reply-To: <bug-3886-206@http.llvm.org/bugs/>
Message-ID: <200905081610.n48GAd3M014860@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3886


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Chris Lattner <clattner at apple.com>  2009-05-08 11:10:38 ---
It looks like this works now, presumably due to nate's shuffle fixes?  If this
still fails for you, please reopen with a new testcase.  Thanks!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 13:24:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 13:24:13 -0500
Subject: [LLVMbugs] [Bug 4152] [Linux kernel] invalid asm constraint 'i'
In-Reply-To: <bug-4152-206@http.llvm.org/bugs/>
Message-ID: <200905081824.n48IODP1020719@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4152


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Chris Lattner <clattner at apple.com>  2009-05-08 13:24:13 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077265.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 13:55:46 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 13:55:46 -0500
Subject: [LLVMbugs] [Bug 4130] clang infloops on invalid C code
In-Reply-To: <bug-4130-206@http.llvm.org/bugs/>
Message-ID: <200905081855.n48ItkiK022185@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4130


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-05-08 13:55:44 ---
Works for me, I get:

t.c:1:46: error: unknown type name 'pte_t'
static inline __attribute__((always_inline)) pte_t ptep_get_and_clear(struct
                                             ^
t.c:2:1: warning: declaration of 'struct mm_struct' will not be visible outside
of this function
mm_struct *mm, unsigned long addr,            pte_t *ptep) {
^
t.c:2:47: error: unknown type name 'pte_t'
mm_struct *mm, unsigned long addr,            pte_t *ptep) {
                                              ^
t.c:2:61: error: expected '}'
mm_struct *mm, unsigned long addr,            pte_t *ptep) {
                                                            ^
4 diagnostics generated.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 15:46:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 15:46:14 -0500
Subject: [LLVMbugs] [Bug 4184] New: Type in LangRef, alloca Instruction
Message-ID: <bug-4184-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4184

           Summary: Type in LangRef, alloca Instruction
           Product: Documentation
           Version: trunk
          Platform: PC
               URL: http://llvm.org/docs/LangRef.html#i_alloca
        OS/Version: Linux
            Status: NEW
          Severity: trivial
          Priority: P2
         Component: General docs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: me22.ca at gmail.com
                CC: llvmbugs at cs.uiuc.edu


"The operation is undefiend if"
should be
"The operation is undefined if"


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 15:50:02 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 15:50:02 -0500
Subject: [LLVMbugs] [Bug 4184] Type in LangRef, alloca Instruction
In-Reply-To: <bug-4184-206@http.llvm.org/bugs/>
Message-ID: <200905082050.n48Ko1Fm026514@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4184


Bill Wendling <wendling at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wendling at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Bill Wendling <wendling at apple.com>  2009-05-08 15:49:59 ---
Fixed. Thanks!

Sending        docs/LangRef.html
Transmitting file data .
Committed revision 71260.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 16:33:35 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 16:33:35 -0500
Subject: [LLVMbugs] [Bug 4100] assertion failed,
	"Re-defining a live register!"
In-Reply-To: <bug-4100-206@http.llvm.org/bugs/>
Message-ID: <200905082133.n48LXZmR028283@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4100


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Bob Wilson <bob.wilson at apple.com>  2009-05-08 16:33:34 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077235.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 18:30:04 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 18:30:04 -0500
Subject: [LLVMbugs] [Bug 4182] RegionStore.cpp fails to compile under MSVC
In-Reply-To: <bug-4182-206@http.llvm.org/bugs/>
Message-ID: <200905082330.n48NU4Nc031897@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4182


Zhongxing Xu <xuzhongxing at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Zhongxing Xu <xuzhongxing at gmail.com>  2009-05-08 18:30:03 ---
Does this fix the problem?
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20090504/017026.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May  8 21:54:58 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 8 May 2009 21:54:58 -0500
Subject: [LLVMbugs] [Bug 3886] Crash when building Xorg/Cairo's pixman
In-Reply-To: <bug-3886-206@http.llvm.org/bugs/>
Message-ID: <200905090254.n492sw0f009738@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3886


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #8 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-08 21:54:58 ---
Both testcases are still broken with "clang -m64 -O3".  I'll attach bitcode
testcases in a moment.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May  9 08:02:58 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 9 May 2009 08:02:58 -0500
Subject: [LLVMbugs] [Bug 3171] not optimizing trivial loop (trip count
	computation)
In-Reply-To: <bug-3171-206@http.llvm.org/bugs/>
Message-ID: <200905091302.n49D2wgA007998@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3171


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-09 08:02:51 ---
Fixed in r71354.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May  9 09:34:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 9 May 2009 09:34:33 -0500
Subject: [LLVMbugs] [Bug 4102] Write after free in debug info writer with
	fast isel
In-Reply-To: <bug-4102-206@http.llvm.org/bugs/>
Message-ID: <200905091434.n49EYX61010866@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4102


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #3 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-09 09:34:33 ---
Yeah, I can't reproduce anymore.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May  9 09:39:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 9 May 2009 09:39:31 -0500
Subject: [LLVMbugs] [Bug 1227] [scev] Can't compute trip count of simple loop
In-Reply-To: <bug-1227-206@http.llvm.org/bugs/>
Message-ID: <200905091439.n49EdV0l011071@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=1227


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #9 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-09 09:39:31 ---
This appears to be working now.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 10 09:56:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 10 May 2009 09:56:54 -0500
Subject: [LLVMbugs] [Bug 4136] Infinite recursion in scheduler
In-Reply-To: <bug-4136-206@http.llvm.org/bugs/>
Message-ID: <200905101456.n4AEusid027029@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4136


Anton Korobeynikov <asl at math.spbu.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #6 from Anton Korobeynikov <asl at math.spbu.ru>  2009-05-10 09:56:13 ---
Done:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077272.html

Testcase here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090504/077338.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 10 12:59:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 10 May 2009 12:59:34 -0500
Subject: [LLVMbugs] [Bug 4185] New: x87 inline asm assertion
Message-ID: <bug-4185-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4185

           Summary: x87 inline asm assertion
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: rafael.espindola at gmail.com, llvmbugs at cs.uiuc.edu


Created an attachment (id=2980)
 --> (http://llvm.org/bugs/attachment.cgi?id=2980)
testcase

I thought I'd already filed this once before but couldn't find it while
searching, not even through closed bugs. On x86-32 without SSE, this testcase
crashes:

$ llvm-as < c.ll | llc -mattr=-sse
llc: X86FloatingPoint.cpp:995:
void<unnamed>::FPS::handleSpecialFP(llvm::ilist_iterator<llvm::MachineInstr>&):
Assertion `(StackTop == 1 || StackTop == 2) && "Stack should have one or two
element on it to return!"' failed.
0   llc 0x089a99e2
1   llc 0x089a9f89
2       0xffffe400 __kernel_sigreturn + 0
Stack dump:
0.      Program arguments: llc -mattr=-sse 
1.      Running pass 'X86 FP Stackifier' on function '@test'
Aborted

It'd be awfully nice if that worked.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 10 16:04:58 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 10 May 2009 16:04:58 -0500
Subject: [LLVMbugs] [Bug 4186] New: missed IR opt: eliminate unused bitcast
	in call
Message-ID: <bug-4186-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4186

           Summary: missed IR opt: eliminate unused bitcast in call
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


The bitcast below can be eliminated:
--
ddunbar at frank:test$ cat t.c
void f0(void) {
  f1();
}

void f1(void);
ddunbar at frank:test$ llvm-gcc -O3 -emit-llvm -S -o - t.c
t.c:5: warning: conflicting types for ???f1???
t.c:2: warning: previous implicit declaration of ???f1??? was here
; ModuleID = 't.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-f80:32:32"
target triple = "i386-pc-linux-gnu"

define void @f0() nounwind {
entry:
        %0 = tail call i32 (...)* bitcast (void ()* @f1 to i32 (...)*)()
nounwind               ; <i32> [#uses=0]
        ret void
}

declare void @f1()
--

The example C code is lame, but this situation also occurs during linking.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 10 16:23:35 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 10 May 2009 16:23:35 -0500
Subject: [LLVMbugs] [Bug 4187] New: GVN eliminates load across free?
Message-ID: <bug-4187-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4187

           Summary: GVN eliminates load across free?
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: resistor at mac.com, llvmbugs at cs.uiuc.edu


GVN eliminates the second load in this example, which I believe is invalid.
This didn't happen in 2.4, I believe.

--
; ModuleID = '2007-10-11-illegal-access-after-free-and-branch.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-f80:128:128"
target triple = "i386-apple-darwin10"

declare void @klee_make_symbolic(i8* %addr, i32 %nbytes) nounwind ssp;

define i32 @f0() nounwind ssp {
entry:
        %t0 = malloc [3 x i8]
        %.sub = getelementptr [3 x i8]* %t0, i32 0, i32 0
        call void @klee_make_symbolic(i8* %.sub, i32 3) nounwind
        %t2 = load i8* %.sub, align 4
        free [3 x i8]* %t0
        %t8 = load i8* %.sub, align 4
        %t9 = zext i8 %t8 to i32
        ret i32 %t9
}
--

Here:
--
ddunbar at ubuntu:regression$ llvm-as < load-elim.ll | opt -gvn | llvm-dis
; ModuleID = '<stdin>'
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"
target triple = "i386-apple-darwin10"

declare void @klee_make_symbolic(i8*, i32) nounwind ssp

define i32 @f0() nounwind ssp {
entry:
        %t0 = malloc [3 x i8]           ; <[3 x i8]*> [#uses=2]
        %.sub = getelementptr [3 x i8]* %t0, i32 0, i32 0               ; <i8*>
[#uses=2]
        call void @klee_make_symbolic(i8* %.sub, i32 3) nounwind
        %t2 = load i8* %.sub, align 4           ; <i8> [#uses=1]
        free [3 x i8]* %t0
        %t9 = zext i8 %t2 to i32                ; <i32> [#uses=1]
        ret i32 %t9
}
--


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 10 18:39:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 10 May 2009 18:39:38 -0500
Subject: [LLVMbugs] [Bug 4187] GVN eliminates load across free?
In-Reply-To: <bug-4187-206@http.llvm.org/bugs/>
Message-ID: <200905102339.n4ANdcm6010973@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4187


Owen Anderson <resistor at mac.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #2 from Owen Anderson <resistor at mac.com>  2009-05-10 18:39:37 ---
I talked to Daniel about this, and explained why the transformation is valid.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 10 22:41:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 10 May 2009 22:41:41 -0500
Subject: [LLVMbugs] [Bug 4188] New: ICE: Dereferencing end()!
Message-ID: <bug-4188-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4188

           Summary: ICE: Dereferencing end()!
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: regehr at cs.utah.edu
                CC: llvmbugs at cs.uiuc.edu


Seen using r71415 on Ubuntu Hardy on x86.

regehr at john-home:~/volatile/tmp164$ llvm-gcc -O2 small.c
cc1: /home/regehr/z/tmp/llvm-r71415/include/llvm/ADT/ilist.h:192:
llvm::ilist_iterator<NodeTy>::operator typename bidirectional_iterator<NodeTy,
int>::pointer() const [with NodeTy = llvm::MachineBasicBlock]: Assertion
`Traits::getNext(NodePtr) != 0 && "Dereferencing end()!"' failed.
small.c:19: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.

regehr at john-home:~/volatile/tmp164$ cat small.c

static char foo (int _si1, char _si2)
{
  return 1 < +_si1 && _si2 && _si1 < +_si2 || _si2 && _si1
    && _si2 + _si1 ? : _si1;
}

volatile int g_9;

int func_88 (int p_90, char p_91)
{
}

int86 (int p_87)
{
  for (1;;)
    {
      func_88 (foo (g_9, p_87), 1);
    }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 02:26:55 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 02:26:55 -0500
Subject: [LLVMbugs] [Bug 4186] missed IR opt: eliminate unused bitcast in
	call
In-Reply-To: <bug-4186-206@http.llvm.org/bugs/>
Message-ID: <200905110726.n4B7Qts2023925@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4186


Daniel Dunbar <daniel at zuster.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #5 from Daniel Dunbar <daniel at zuster.org>  2009-05-11 02:26:55 ---
Well, as soon as I click the right button that is.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 05:48:02 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 05:48:02 -0500
Subject: [LLVMbugs] [Bug 4189] New: redundant load from constant not
	eliminated
Message-ID: <bug-4189-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4189

           Summary: redundant load from constant not eliminated
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


void
copy_text (char *p)
{
  const char *text = "hello world/n";
  do
    {
      *p = *text;
    }
  while (*text++);
}

->

@.str = internal constant [14 x i8] c"hello world/n\00"         ; <[14 x i8]*>
[#uses=1]

define void @copy_text(i8* nocapture %p) nounwind {
entry:
        br label %bb

bb:             ; preds = %bb, %entry
        %indvar = phi i64 [ 0, %entry ], [ %indvar.next, %bb ]          ; <i64>
[#uses=2]
        %tmp = add i64 %indvar, ptrtoint ([14 x i8]* @.str to i64)             
; <i64> [#uses=1]
        %text.0 = inttoptr i64 %tmp to i8*              ; <i8*> [#uses=2]
        %0 = load i8* %text.0, align 1          ; <i8> [#uses=1]
        store i8 %0, i8* %p, align 1
        %1 = load i8* %text.0, align 1          ; <i8> [#uses=1]
        %2 = icmp eq i8 %1, 0           ; <i1> [#uses=1]
        %indvar.next = add i64 %indvar, 1               ; <i64> [#uses=1]
        br i1 %2, label %return, label %bb

return:         ; preds = %bb
        ret void
}

The second load is redundant.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 08:29:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 08:29:48 -0500
Subject: [LLVMbugs] [Bug 4046] Use after free in
	test/Feature/embeddedmetadata.ll
In-Reply-To: <bug-4046-206@http.llvm.org/bugs/>
Message-ID: <200905111329.n4BDTmhm016313@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4046


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Duncan Sands <baldrick at free.fr>  2009-05-11 08:29:38 ---
This was fixed recently.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 09:26:58 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 09:26:58 -0500
Subject: [LLVMbugs] [Bug 4190] New: Internal compiler error
Message-ID: <bug-4190-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4190

           Summary: Internal compiler error
           Product: tools
           Version: 2.5
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: blocker
          Priority: P2
         Component: llvm-g++
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: luc_j_bourhis at mac.com
                CC: llvmbugs at cs.uiuc.edu, luc_j_bourhis at mac.com


Short
----
llvm-g++-4.2 -o cctbx/miller/boost_python/phase_transfer.o -c -fPIC
-fno-strict-aliasing -w -ftemplate-depth-130 -g2 -O0 -fno-inline -fopenmp -fPIC
-fno-strict-aliasing -w -g2 -O0 -fno-inline -fopenmp
phase_transfer.preprocessed.cpp

where phase_transfer.preprocessed.cpp is the file attached to this bug report
results in

/Users/luc/Developer/cctbx/cctbx_project/scitbx/array_family/ref.h: In member
function ???const ElementType& scitbx::af::const_ref<ElementType,
AccessorType>::operator[](size_t) const [with ElementType =
cctbx::miller::index<int>, AccessorType = scitbx::af::trivial_accessor]???:
/Users/luc/Developer/cctbx/cctbx_project/scitbx/array_family/ref.h:79: internal
compiler error: Bus error
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://developer.apple.com/bugreporter> for instructions.

Longer
-----
I experimented to compile the CCTBX (http://cctbx.sourceforge.net) with
llvm-g++-4.2 from the 2.5 release. It resulted in an internal compiler error. I
then preprocessed the source file llvm-g++-4.2 choked upon so as to provide a
self-contained source file exposing the problem.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 14:38:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 14:38:31 -0500
Subject: [LLVMbugs] [Bug 4193] New: llvm-gcc segfault on broken code
Message-ID: <bug-4193-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4193

           Summary: llvm-gcc segfault on broken code
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


Got this when delta reducing PR4190.  Unfortunately delta latched onto a
different segmentation fault...

$ llvm-gcc -c barf.ii
barf.ii:4: error: ISO C++ forbids declaration of ???codecvt??? with no type
barf.ii:4: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.
$ cat barf.ii 
struct __sbuf {}
FILE;
extern "C" {
    extern template const codecvt<char, char, mbstate_t>&
use_facet<codecvt<char, char, mbstate_t> >(const locale&);


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 15:59:55 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 15:59:55 -0500
Subject: [LLVMbugs] [Bug 4195] New: assertion failure in CodePlacementOpt
Message-ID: <bug-4195-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4195

           Summary: assertion failure in CodePlacementOpt
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: bob.wilson at apple.com
                CC: llvmbugs at cs.uiuc.edu


The attached testcase causes an assertion failure when compiled with llc.  It
appears to be a recent regression.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 16:56:24 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 16:56:24 -0500
Subject: [LLVMbugs] [Bug 4188] ICE: crash in branchfolding
In-Reply-To: <bug-4188-206@http.llvm.org/bugs/>
Message-ID: <200905112156.n4BLuOhG006060@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4188


Dale Johannesen <dalej at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Dale Johannesen <dalej at apple.com>  2009-05-11 16:56:23 ---
Fixed.
http://llvm.org/viewvc/llvm-project?view=rev&revision=71478


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 18:16:57 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 18:16:57 -0500
Subject: [LLVMbugs] [Bug 4090] Assertion failed: Re-defining a live
 register!", lib/CodeGen/  RegisterScavenging.cpp, line 272
In-Reply-To: <bug-4090-206@http.llvm.org/bugs/>
Message-ID: <200905112316.n4BNGvRG008947@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4090


Lang Hames <lhames at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 21:36:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 21:36:48 -0500
Subject: [LLVMbugs] [Bug 1301] [LSR/IndVars] Ada array traversal with a
 non-word index type optimizations
In-Reply-To: <bug-1301-206@http.llvm.org/bugs/>
Message-ID: <200905120236.n4C2amOg015573@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=1301


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #13 from Dan Gohman <gohman at apple.com>  2009-05-11 21:36:47 ---
After a long series of patches, culminating with this patch:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090511/077422.html
all of these loops are now optimized down to four instructions: store, inc,
cmp, jne. All
the arithmetic is now done at "word" size.

There are a few more minor opportunities remaining; the cmp could be eliminated
by
making the loop count to zero instead of from zero, and the loop could be made
to use
one register instead of two via full strength reduction, however all the issues
specific to
non-"word" index types are now addressed.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 11 22:50:27 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 11 May 2009 22:50:27 -0500
Subject: [LLVMbugs] [Bug 4195] assertion failure in CodePlacementOpt
In-Reply-To: <bug-4195-206@http.llvm.org/bugs/>
Message-ID: <200905120350.n4C3oRv4017921@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4195


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Bob Wilson <bob.wilson at apple.com>  2009-05-11 22:50:26 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090511/077423.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 08:40:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 08:40:14 -0500
Subject: [LLVMbugs] [Bug 4200] New: Pass Manager Hangs
Message-ID: <bug-4200-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4200

           Summary: Pass Manager Hangs
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: npjohnso at cs.princeton.edu
                CC: llvmbugs at cs.uiuc.edu


With this particular pattern of passes requiring one another and LoopSimplify
and CallGraph, pass manager never terminates.

I have confirmed this with gcc 4.1.2 and 4.2.4 against llvm svn revision 68820.

To recreate this bug,

Compile the attached C++ source code to a shared library

Run opt:

opt -load ./libBug.so -after input.bc -o output.bc


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 10:08:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 10:08:36 -0500
Subject: [LLVMbugs] [Bug 4201] New: llvm.smul.with.overflow.i32 not
	implemented on PPC
Message-ID: <bug-4201-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4201

           Summary: llvm.smul.with.overflow.i32 not implemented on PPC
           Product: new-bugs
           Version: unspecified
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ewasylishen at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2988)
 --> (http://llvm.org/bugs/attachment.cgi?id=2988)
A simple test case

Hi, just found that the llvm.*.with.overflow intrinsics are not implemented on
PowerPC, and target independent lowering is also not supported for these, so
you can't use them. I've attached a simple test case.

Thanks.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 11:57:18 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 11:57:18 -0500
Subject: [LLVMbugs] [Bug 4202] New: constant pools out of range
Message-ID: <bug-4202-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4202

           Summary: constant pools out of range
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: bob.wilson at apple.com
                CC: llvmbugs at cs.uiuc.edu


A number of the MultiSource tests are failing on ARM because the constant pools
are too far away from the loads that reference them.  Most of the offsets are
only slightly too large.  For example:

/Volumes/LocalHD/bwilson/nightly/test/llvm/Release/bin/llc -O3
-relocation-model=pic -mcpu=arm1136jf-s -f Output/treecc.llvm.bc -o
Output/treecc.llc.s
gcc Output/treecc.llc.s -o Output/treecc.llc  -lm  -arch armv6 -isysroot
/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS3.0.Internal.sdk
Output/treecc.llc.s:9173:bad immediate value for offset (4112)
Output/treecc.llc.s:9084:bad immediate value for offset (4104)
Output/treecc.llc.s:9056:bad immediate value for offset (4108)
Output/treecc.llc.s:9037:bad immediate value for offset (4108)
Output/treecc.llc.s:8971:bad immediate value for offset (4104)
Output/treecc.llc.s:8948:bad immediate value for offset (4104)
Output/treecc.llc.s:8898:bad immediate value for offset (4100)
Output/treecc.llc.s:8647:bad immediate value for offset (4104)
make[3]: [Output/treecc.llc] Error 1 (ignored)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 13:20:01 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 13:20:01 -0500
Subject: [LLVMbugs] [Bug 4203] New: Linux/PPC ABI: Obvious way of calling a
 C function which takes a struct paramater doesn' t work
Message-ID: <bug-4203-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4203

           Summary: Linux/PPC ABI: Obvious way of calling a C function which
                    takes a struct paramater doesn't work
           Product: new-bugs
           Version: unspecified
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ewasylishen at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2989)
 --> (http://llvm.org/bugs/attachment.cgi?id=2989)
a test case

Hi,

Suppose I compile a C function with gcc which takes a single struct parameter,
struct foo { float a, float b };

I would expect to call it from LLVM in the same way, i.e.:
call void @some_function(%struct.foo %examplestruct)
(where %struct.foo = type { float, float }).


This doesn't work (the C function gets nonsense values.) Instead, what does
work is calling it from LLVM by passing a pointer to the struct I want to pass:
call void @some_function(%struct.foo* %pointer_to_examplestruct)


I have attached a simple test case which demonstrates the problem.
The output is:
./fails
x=262253949499549329478747984903838629888.000000, y=0.000000
./works
x=3.140000, y=15.900000

I think, ideally, the 'fails' executable should produce the correct result,
"x=3.140000, y=15.900000".

Thanks


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 13:46:58 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 13:46:58 -0500
Subject: [LLVMbugs] [Bug 4203] Linux/PPC ABI: Obvious way of calling a C
 function which takes a struct paramater doesn' t work
In-Reply-To: <bug-4203-206@http.llvm.org/bugs/>
Message-ID: <200905121846.n4CIkw3M001356@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4203


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Chris Lattner <clattner at apple.com>  2009-05-12 13:46:58 ---
Unfortunately, this is not how LLVM interacts with native ABIs.  You need to
know about the native ABI to know how structs are passed by value.  This is
annoying, but we don't have a good solution for this.  You'll note that
llvm-gcc (or clang) produce different llvm ir for functions returning structs
when targetting x86-32 and x86-64 for example.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 16:14:15 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 16:14:15 -0500
Subject: [LLVMbugs] [Bug 4204] New: clang rejects valid redeclaration of
 function declared in nested scope
Message-ID: <bug-4204-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4204

           Summary: clang rejects valid redeclaration of function declared
                    in nested scope
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


If I am not mistaken, clang is incorrect to reject the following program:
--
ddunbar at lordcrumb:tmp$ cat t.c
void f0(void) {
  extern void g0(void);
}
g0(void) {}
ddunbar at lordcrumb:tmp$ gcc -std=gnu89 -c t.c
t.c:4: warning: conflicting types for 'g0'
t.c:2: warning: previous declaration of 'g0' was here
ddunbar at lordcrumb:tmp$ clang -std=gnu89 -c t.c
t.c:4:1: error: conflicting types for 'g0'
g0(void) {}
^
t.c:2:15: note: previous declaration is here
  extern void g0(void);
              ^
2 diagnostics generated.
--


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 17:30:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 17:30:38 -0500
Subject: [LLVMbugs] [Bug 4206] New: crash in simplify lib calls
Message-ID: <bug-4206-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4206

           Summary: crash in simplify lib calls
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


simplify lib calls crashes on the attached .bc file, generated by clang:
--
ddunbar at lordcrumb:tmp$ opt -std-compile-opts < t.bc > t.opt.bc
Assertion failed: (S1->getType() == S2->getType() && "Cannot create binary
operator with two operands of differing type!"), function Create, file
Instructions.cpp, line 1606.
0   opt               0x00000001003aba5a PrintStackTrace(void*) + 38
1   opt               0x00000001003abf6a SignalHandler(int) + 288
2   libSystem.B.dylib 0x00007fff87c8890a _sigtramp + 26
3   libSystem.B.dylib 0x00000001012dbc50 _sigtramp + 2036675424
4   libSystem.B.dylib 0x00007fff87ce7072 __assert_rtn + 237
5   opt               0x000000010032384a
llvm::BinaryOperator::Create(llvm::Instruction::BinaryOps, llvm::Value*,
llvm::Value*, std::string const&, llvm::Instruction*) + 106
6   opt               0x000000010018983e
llvm::BinaryOperator::CreateAdd(llvm::Value*, llvm::Value*, std::string const&)
+ 48
7   opt               0x00000001001b2c99 llvm::IRBuilder<true,
llvm::ConstantFolder>::CreateAdd(llvm::Value*, llvm::Value*, char const*) + 137
8   opt               0x00000001001a83b9 (anonymous
namespace)::FFSOpt::CallOptimizer(llvm::Function*, llvm::CallInst*,
llvm::IRBuilder<true, llvm::ConstantFolder>&) + 413
9   opt               0x00000001001a7dd1 (anonymous
namespace)::LibCallOptimization::OptimizeCall(llvm::CallInst*, llvm::TargetData
const&, llvm::IRBuilder<true, llvm::ConstantFolder>&) + 105
10  opt               0x00000001001aecdd (anonymous
namespace)::SimplifyLibCalls::runOnFunction(llvm::Function&) + 495
11  opt               0x000000010033f126
llvm::FPPassManager::runOnFunction(llvm::Function&) + 256
12  opt               0x000000010021f993 (anonymous
namespace)::CGPassManager::runOnModule(llvm::Module&) + 623
13  opt               0x000000010033db93
llvm::MPPassManager::runOnModule(llvm::Module&) + 203
14  opt               0x000000010033f589
llvm::PassManagerImpl::run(llvm::Module&) + 111
15  opt               0x000000010033f5eb llvm::PassManager::run(llvm::Module&)
+ 33
16  opt               0x0000000100056f46 main + 4608
17  opt               0x000000010004aa48 start + 52
18  opt               0x0000000000000002 start + 4294661614
Stack dump:
0.      Running pass 'CallGraph Pass Manager' on module '<stdin>'.
1.      Running pass 'Simplify well-known library calls' on function
'@test_ffsl'
Abort trap
--


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 17:51:17 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 17:51:17 -0500
Subject: [LLVMbugs] [Bug 4207] New: TableGen Cannot Slice List Arguments
Message-ID: <bug-4207-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4207

           Summary: TableGen Cannot Slice List Arguments
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: TableGen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: greened at obbligato.org
                CC: llvmbugs at cs.uiuc.edu


The attached seemingly trivial testcase crashes TableGen with:

tblgen:
/ptmp/dag/llvm-project.modified/llvm/trunk/utils/TableGen/Record.cpp:642:
virtual llvm::Init* llvm::VarInit::resolveListElementReference(llvm::Record&,
const llvm::RecordVal*, unsigned int): Assertion `LI && "Invalid list
element!"' failed.

The problem appears to be that 'vals' is a VarInit in the BB context.

Taking out one layer of classes (e. g. having One and Two derive from BB) AND
commenting out the definition of BBB makes the problem go away.  Strangely,
leaving BBB defined causes the same crash even though it is referenced nowhere
and does not take any slices.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 21:27:49 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 21:27:49 -0500
Subject: [LLVMbugs] [Bug 4204] clang rejects valid redeclaration of function
 declared in nested scope
In-Reply-To: <bug-4204-206@http.llvm.org/bugs/>
Message-ID: <200905130227.n4D2Rn88020316@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4204


Daniel Dunbar <daniel at zuster.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #3 from Daniel Dunbar <daniel at zuster.org>  2009-05-12 21:27:48 ---
Yes.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 12 23:43:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 12 May 2009 23:43:22 -0500
Subject: [LLVMbugs] [Bug 4121] Assertion `it != mi2iMap_.end() && "Invalid
 instruction!"' failed.
In-Reply-To: <bug-4121-206@http.llvm.org/bugs/>
Message-ID: <200905130443.n4D4hMqG025873@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4121


Lang Hames <lhames at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 13 00:15:46 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 13 May 2009 00:15:46 -0500
Subject: [LLVMbugs] [Bug 4208] New: isIntegerType() returns true for vectors
	of integers
Message-ID: <bug-4208-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4208

           Summary: isIntegerType() returns true for vectors of integers
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Keywords: code-cleanup
          Severity: normal
          Priority: P2
         Component: AST
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: clattner at apple.com
                CC: llvmbugs at cs.uiuc.edu


isIntegerType() currently returns true for vectors of integers.  This is really
dangerous because a lot of sema uses this predicate and probably doesn't expect
vectors.  Should should follow the lead of complex and has a isVectorOfInteger
method or something like that instead.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 13 01:26:33 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 13 May 2009 01:26:33 -0500
Subject: [LLVMbugs] [Bug 4206] crash in simplify lib calls
In-Reply-To: <bug-4206-206@http.llvm.org/bugs/>
Message-ID: <200905130626.n4D6QXSd029360@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4206


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Chris Lattner <clattner at apple.com>  2009-05-13 01:26:33 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090511/077483.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 13 05:48:39 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 13 May 2009 05:48:39 -0500
Subject: [LLVMbugs] [Bug 4210] New: optimisation breaks AES code from MySQL
Message-ID: <bug-4210-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4210

           Summary: optimisation breaks AES code from MySQL
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: df at sun.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2993)
 --> (http://llvm.org/bugs/attachment.cgi?id=2993)
standalone test case

When building MySQL 5.0 with clang (at least for a Mac OS X x86_64 target),
turning on optimisation (-O1 or above) breaks AES encryption/decryption
functionality as compared to a build without optimisation (-O0).

The attached standalone test case was extracted from the MySQL 5.0 code base.
I'm sorry that it is still quite large, I've not had much luck with extracting
the individual bit that is optimised badly, but I have reduced it to one code
path in one function that shows the problem.

To compile:

clang -o rijndael_test_O0  rijndael_bug.c -O0
clang -o rijndael_test_O1  rijndael_bug.c -O0
clang -o rijndael_test_O1V rijndael_bug.c -O0 -DVOLATILE=volatile

The output of rijndael_test_O0 is as follows:

39 25 cc 14 d7 5 ef 4e e9 c7 fb cd 40 22 ed 29 

The output of rijndael_test_O1 is as follows:

39 25 cc 14 d7 5 ef 4e e9 c7 fb d1 40 22 ed 29 

Note that the 12th byte is different.

The output of rijndael_test_O1V is the same as rijndael_test_O0:

39 25 cc 14 d7 5 ef 4e e9 c7 fb cd 40 22 ed 29 

The difference between O1 and O1V is that a "volatile" statement is added to
the variable declaration in the rijndaelEncrypt function. In fact it seems to
be sufficient to declare *one* of these variables volatile to make the problem
go away.

I have inserted one comment into the source:
    /* Value of t1 will be different depending on optimisation */

At this location, a value t1 is computed as follows:
    t1= (Te0[(s1 >> 24)  ] ^
         Te1[(s2 >> 16) & 0xff] ^
         Te2[(s3 >>     8) & 0xff] ^
         Te3[(s0         ) & 0xff] ^
         rk[5]);

The result of this calculation is different when optimisation is turned on. I
wasn't able to reproduce this with the snippet on its own.

The full version of the rijndaelEncrypt function in the MySQL code base has an
alternate code path that can be enabled with -DFULL_UNROLL. I removed this
codepath from the test case because it does not exhibit the problem.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 13 19:05:42 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 13 May 2009 19:05:42 -0500
Subject: [LLVMbugs] [Bug 3343] llc abort with -pre-alloc-split
In-Reply-To: <bug-3343-206@http.llvm.org/bugs/>
Message-ID: <200905140005.n4E05gvl020049@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3343


Owen Anderson <resistor at mac.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Owen Anderson <resistor at mac.com>  2009-05-13 19:05:39 ---
This was fixed a while back.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 13 23:28:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 13 May 2009 23:28:37 -0500
Subject: [LLVMbugs] [Bug 4124] null ptr dereferenced
In-Reply-To: <bug-4124-206@http.llvm.org/bugs/>
Message-ID: <200905140428.n4E4Sava028286@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4124


Lang Hames <lhames at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Lang Hames <lhames at gmail.com>  2009-05-13 23:28:24 ---
This was being caused by TwoAddressInstructionPass::FindLastUseInMBB returning
an instruction that is not, in fact, the nearest previous use. This in turn
lead to incorrect live interval information and finally the crash during
coalescing. The fault occured because FindLastUseInMBB assumed reg_iterator
would iterate over defs/uses in the order they appear in the basic block. I'm
not sure if this assumption _should_ hold, but it certainly didn't in this
case.

Fix committed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090511/077556.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 14 15:40:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 14 May 2009 15:40:26 -0500
Subject: [LLVMbugs] [Bug 4207] TableGen Cannot Slice List Arguments
In-Reply-To: <bug-4207-206@http.llvm.org/bugs/>
Message-ID: <200905142040.n4EKeQbi010762@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4207


David Greene <greened at obbligato.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from David Greene <greened at obbligato.org>  2009-05-14 15:40:20 ---
Fixed in r71787.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 14 19:00:45 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 14 May 2009 19:00:45 -0500
Subject: [LLVMbugs] [Bug 4212] New: bugpoint doesn't reduce global variables
Message-ID: <bug-4212-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4212

           Summary: bugpoint doesn't reduce global variables
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: bugpoint
        AssignedTo: nicholas at mxc.ca
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


The global variable reduction in CrashDebugger.cpp uses the ListReducer, but
the actual implementation of the test is buggy; it always removes all of the
globals instead of leaving some of them intact.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 14 21:25:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 14 May 2009 21:25:12 -0500
Subject: [LLVMbugs] [Bug 4213] New: Cannot create call to void function
	using C++ API
Message-ID: <bug-4213-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4213

           Summary: Cannot create call to void function using C++ API
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthew at loar.name
                CC: llvmbugs at cs.uiuc.edu


When attempting to create a call to a function with a return type of
Type::VoidTy using the C++ API, I get the following assertion failure:

"molc: Value.cpp:182: void llvm::Value::setName(const char*, unsigned int):
Assertion `getType() != Type::VoidTy && "Cannot assign a name to void values!"'
failed."

This appears to be because the CallInst class ultimately is derived from Value.
 The type that is passed to the Value constructor is the return type of the
called function.  The CallInst constructor then calls setName(), triggering the
assert.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 14 23:25:44 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 14 May 2009 23:25:44 -0500
Subject: [LLVMbugs] [Bug 4213] Cannot create call to void function using C++
	API
In-Reply-To: <bug-4213-206@http.llvm.org/bugs/>
Message-ID: <200905150425.n4F4PiFK027395@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4213


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-14 23:25:42 ---
This is definitely invalid; C++ code in LLVM creates calls to void functions
all the time.  Try using an empty name for the instruction.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 15 00:47:01 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 15 May 2009 00:47:01 -0500
Subject: [LLVMbugs] [Bug 4214] New: Internal compile error when function
 declared with void formal parameter
Message-ID: <bug-4214-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4214

           Summary: Internal compile error when function declared with void
                    formal parameter
           Product: new-bugs
           Version: unspecified
          Platform: Sun
        OS/Version: Solaris
            Status: NEW
          Keywords: crash-on-invalid
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: mike.emmerik at sun.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=2997)
 --> (http://llvm.org/bugs/attachment.cgi?id=2997)
C source code that shows the bug

An Internal Compile Error results when llvm-gcc encounters a function
declaration with a void parameter. This is illegal C, but I found it in some
real code (so some compiler, probably an old version of Sun Studio, must allow
this). In any case, llvm should not fall over on this input.

Steps to reproduce:
1) Create test.c with this content:
typedef void vt;
void (*func_ptr)(vt my_vt);
2) llvm-gcc -c -o test.o test.c

Actual results:
test.c:3: warning: parameter 1 ('my_vt') has void type
Assertion failed: ValueType != Type::VoidTy && "Pointer to void is not valid,
use sbyte* instead!", file Type.cpp, line 1311
test.c:3: internal compiler error: Abort
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://developer.apple.com/bugreporter> for instructions.

Expected results:
test.c:3: warning: parameter 1 ('my_vt') has void type
and a valid test.o and test.o.bc.

Build data and platform:
llvm-gcc4.2-2.4 built from source on Solaris 11.

Additional builds and platforms:
Also happens on llvm-gcc4.2-2.5 on Solaris 11.

Additional information:
The following patch fixes the ICE:
*** new.h       Fri May 15 15:07:12 2009
--- ../../llvm-gcc4.2-2.4.source/gcc/llvm-abi.h Fri May 15 15:06:40 2009
***************
*** 421,426 ****
--- 421,428 ----
      // not include variable sized fields here.
      std::vector<const Type*> Elts;
      if (isPassedByInvisibleReference(type)) { // variable size -> by-ref.
+       if (Ty->getTypeID() == Type::VoidTyID)
+         Ty = Type::Int8Ty;  // void* -> sbyte*   Added MVE
        const Type *PtrTy = PointerType::getUnqual(Ty);
        C.HandleByInvisibleReferenceArgument(PtrTy, type);
        ScalarElts.push_back(PtrTy);


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 15 02:02:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 15 May 2009 02:02:26 -0500
Subject: [LLVMbugs] [Bug 4210] optimisation breaks AES code from MySQL
In-Reply-To: <bug-4210-206@http.llvm.org/bugs/>
Message-ID: <200905150702.n4F72QjZ001276@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4210


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #19 from Evan Cheng <evan.cheng at apple.com>  2009-05-15 02:02:25 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090511/077632.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 15 06:50:46 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 15 May 2009 06:50:46 -0500
Subject: [LLVMbugs] [Bug 4216] New: Consecutive small memory accesses not
 merged in a bigger access
Message-ID: <bug-4216-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4216

           Summary: Consecutive small memory accesses not merged in a bigger
                    access
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: code-quality
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: quickslyver at free.fr
                CC: llvmbugs at cs.uiuc.edu


struct control_reg
{
  int a:3;
  int b:2;
  int c:1;
  int d:4;
  int e:3;
  int f:3;
};

struct control_reg test_bitfield(struct control_reg c)
{
  c.a=2;
  c.d=3;
  c.f=4;
  return c;
}

--------------------
With clang -03 this give:
--------------------

define void @test_bitfield(%struct.control_reg* noalias nocapture sret
%agg.result, %struct.control_reg* nocapture byval %c) nounwind {
entry:
        %tmp = bitcast %struct.control_reg* %c to i32*          ; <i32*>
[#uses=2]
        %bf.prev.low = load i32* %tmp           ; <i32> [#uses=1]
        %bf.prev.lo.cleared10 = or i32 %bf.prev.low, 32962              ; <i32>
[#uses=1]
        %bf.new.lo11 = and i32 %bf.prev.lo.cleared10, -25350            ; <i32>
[#uses=1]
        store i32 %bf.new.lo11, i32* %tmp
        %c.0 = getelementptr %struct.control_reg* %c, i32 0, i32 0             
; <i8*> [#uses=1]
        %tmp20 = load i8* %c.0, align 4         ; <i8> [#uses=1]
        %c.1 = getelementptr %struct.control_reg* %c, i32 0, i32 1             
; <i8*> [#uses=1]
        %tmp21 = load i8* %c.1, align 1         ; <i8> [#uses=1]
        %c.2 = getelementptr %struct.control_reg* %c, i32 0, i32 2             
; <i8*> [#uses=1]
        %tmp22 = load i8* %c.2, align 2         ; <i8> [#uses=1]
        %c.3 = getelementptr %struct.control_reg* %c, i32 0, i32 3             
; <i8*> [#uses=1]
        %tmp23 = load i8* %c.3, align 1         ; <i8> [#uses=1]
        %agg.result.0 = getelementptr %struct.control_reg* %agg.result, i32 0,
i32 0            ; <i8*> [#uses=1]
        store i8 %tmp20, i8* %agg.result.0, align 4
        %agg.result.1 = getelementptr %struct.control_reg* %agg.result, i32 0,
i32 1            ; <i8*> [#uses=1]
        store i8 %tmp21, i8* %agg.result.1, align 1
        %agg.result.2 = getelementptr %struct.control_reg* %agg.result, i32 0,
i32 2            ; <i8*> [#uses=1]
        store i8 %tmp22, i8* %agg.result.2, align 2
        %agg.result.3 = getelementptr %struct.control_reg* %agg.result, i32 0,
i32 3            ; <i8*> [#uses=1]
        store i8 %tmp23, i8* %agg.result.3, align 1
        ret void
}

--------------
i8 load and store operations should be merged into i32 operations


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 15 12:19:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 15 May 2009 12:19:54 -0500
Subject: [LLVMbugs] [Bug 4217] New: Eliminate the Annotable and Annotation
 classes by making MachineFunction a pass
Message-ID: <bug-4217-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4217

           Summary: Eliminate the Annotable and Annotation classes by making
                    MachineFunction a pass
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: clattner at apple.com
                CC: llvmbugs at cs.uiuc.edu


The only use of the Annotable and Annotation classes are for putting a
MachineFunction on Function.  This class is bad for a number of reasons and
should be removed.

The fix is to make MachineFunction be an analysis group that is provided by the
various instruction selectors and required/preserved by MachineFunctionPass. 
That way the lifetime of the MachineFunction is managed by PassMgr, not be
explicit code.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 16 06:26:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 16 May 2009 06:26:36 -0500
Subject: [LLVMbugs] [Bug 2412] GCC warning is a clang error
In-Reply-To: <bug-2412-206@http.llvm.org/bugs/>
Message-ID: <200905161126.n4GBQaAb018863@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2412


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #3 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-16 06:26:26 ---
I'm going to tentatively mark this as invalid; this issue doesn't seem common. 
Also, although it's reasonably safe on common platforms to cast a function
pointer like "void (*)(void)" to "int (*)(void)", it's not a property we really
want to guarantee for every platform we might potentially support.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 16 06:46:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 16 May 2009 06:46:36 -0500
Subject: [LLVMbugs] [Bug 3009] Bogus warning on compound literals
In-Reply-To: <bug-3009-206@http.llvm.org/bugs/>
Message-ID: <200905161146.n4GBka1r019539@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3009


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-16 06:46:33 ---
Fixed in r71939.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 16 07:16:23 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 16 May 2009 07:16:23 -0500
Subject: [LLVMbugs] [Bug 2044] clang incorrectly accepts function returning
 pointer to vla
In-Reply-To: <bug-2044-206@http.llvm.org/bugs/>
Message-ID: <200905161216.n4GCGN5g020592@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2044


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-16 07:16:22 ---
Fixed in r71941.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 16 16:31:05 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 16 May 2009 16:31:05 -0500
Subject: [LLVMbugs] [Bug 3942] Clang complains unsigned integer arithmetic
	overflows
In-Reply-To: <bug-3942-206@http.llvm.org/bugs/>
Message-ID: <200905162131.n4GLV5et006340@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3942


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-16 16:31:04 ---
Fixed in r71960.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 16 19:34:01 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 16 May 2009 19:34:01 -0500
Subject: [LLVMbugs] [Bug 4220] New: Use-after-free on invalid code with label
Message-ID: <bug-4220-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4220

           Summary: Use-after-free on invalid code with label
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Testcase:

int a() { goto L; do { L:; } while 0; }

A use-after-free shows up running the testcase through "valgrind clang-cc".


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 16 20:33:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 16 May 2009 20:33:22 -0500
Subject: [LLVMbugs] [Bug 4221] New: -ast-print doesn't print tag definitions
	correctly
Message-ID: <bug-4221-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4221

           Summary: -ast-print doesn't print tag definitions correctly
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: AST
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Testcase:
typedef struct {int x;} t1;
int t2() { return (struct {int x;}){1}.x; }
struct t3 { struct {int x,y;}; int z;};

Actual output:
typedef t1 t1;

int t2() {
  return (struct <anonymous>){ 1 }.x;
}

struct t3 {
  struct  {
    int x;
    int y;
  };
  struct t3::<anonymous>;
  int z;
};

-ast-print should not be printing anything with <anonymous>, and it should
print definitions for all three structs.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 17 15:57:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 17 May 2009 15:57:12 -0500
Subject: [LLVMbugs] [Bug 4222] New: Assertion failed: (BitWidth ==
 RHS.BitWidth && " Bit widths must be the same"), function operator*,
 file APInt.cpp, line 432  .
Message-ID: <bug-4222-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4222

           Summary: Assertion failed: (BitWidth == RHS.BitWidth && "Bit
                    widths must be the same"), function operator*, file
                    APInt.cpp, line 432.
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pawel.worach at gmail.com
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


Created an attachment (id=3002)
 --> (http://llvm.org/bugs/attachment.cgi?id=3002)
Pre-processed source

Assertion when building the FreeBSD kernel with clang r71983 on file 
sys/amd64/amd64/minidump_machdep.c

$ clang -c cc-rr9fHR.i
$ clang -O2 -c cc-rr9fHR.i
Assertion failed: (BitWidth == RHS.BitWidth && "Bit widths must be the same"),
function operator*, file APInt.cpp, line 432.
Stack dump:
0.      Program arguments: /usr/opt/llvm/bin/../libexec/clang-cc -triple
x86_64-unknown-freebsd8.0 -S -disable-free -main-file-name cc-rr9fHR.i
--relocation-model static --disable-fp-elim --unwind-tables=1 --mcpu=x86-64
--fmath-errno=1 -O2 -fdiagnostics-show-option -o /tmp/cc-fXARiP.s -x cpp-output
cc-rr9fHR.i 
1.      <eof> parser at end of file
2.      Code generation
3.      Running pass 'Loop Pass Manager' on function '@minidumpsys'
4.      Running pass 'Loop Strength Reduction' on basic block
'%for.cond.i.preheader'


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 17 21:00:48 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 17 May 2009 21:00:48 -0500
Subject: [LLVMbugs] [Bug 4223] New: Improper handling of "undef" type in
	simplifycfg
Message-ID: <bug-4223-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4223

           Summary: Improper handling of "undef" type in simplifycfg
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: viridia at gmail.com
                CC: llvmbugs at cs.uiuc.edu


It seems that variables whose value is "undef" aren't being handled correctly
in a number of places. When I pass the following *invalid* code to LLVM, I
would expect LLVM to detect the error in the 4th instruction (the store to
undef):

define %2 @"throwSomething()"() {
entry:
        %0 = call %tart.core.Exception* @tart.core.Exception.type.alloc()      
        ; <%tart.core.Exception*> [#uses=2]
        %construct = call %2
@"tart.core.Exception.construct()"(%tart.core.Exception* %0)               ;
<%2> [#uses=0]
        %1 = bitcast %tart.core.Exception* %0 to %3*            ; <%3*>
[#uses=2]
        store %3* %1, %3** undef
        %2 = getelementptr %3* %1, i32 0, i32 1         ; <%4*> [#uses=1]
        %throw = call i32 @_Unwind_RaiseException(%4* %2)               ; <i32>
[#uses=0]
        unreachable
}

However, that is not what is happening.

The simplifycfg pass appears to fail silently on this line - that is, it
doesn't report an error, but neither does it generate any code from the 4th
line on. The resulting function contains only the instructions before the
"store", and the rest is just chopped off.

The function verifier pass did not report any error either when it ran on this
function either.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 17 22:32:01 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 17 May 2009 22:32:01 -0500
Subject: [LLVMbugs] [Bug 4224] New: llvm IR documentation doesn't describe
 string constants
Message-ID: <bug-4224-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4224

           Summary: llvm IR documentation doesn't describe string constants
           Product: Documentation
           Version: trunk
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: General docs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


(Pointed out on IRC). The LLVM IR docs don't ever describe the c"..." syntax.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 00:47:50 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 00:47:50 -0500
Subject: [LLVMbugs] [Bug 4223] Improper handling of "undef" type in
	simplifycfg
In-Reply-To: <bug-4223-206@http.llvm.org/bugs/>
Message-ID: <200905180547.n4I5loLp013952@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4223


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-18 00:47:49 ---
That sounds like the expected behavior: store to undef is undefined, so
SimplifyCFG turns it into "unreachable".


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 04:37:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 04:37:20 -0500
Subject: [LLVMbugs] [Bug 4225] New: Failed assertion: "Wrong MachineOperand
 accessor" in setIsKill()
Message-ID: <bug-4225-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4225

           Summary: Failed assertion: "Wrong MachineOperand accessor" in
                    setIsKill()
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Register Allocator
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


Created an attachment (id=3004)
 --> (http://llvm.org/bugs/attachment.cgi?id=3004)
C testcase

I'm seeing the following failed assertion when building some code from OpenSSL:

Assertion failed: (isReg() && !IsDef && "Wrong MachineOperand accessor"),
function setIsKill, file .../llvm/include/llvm/CodeGen/MachineOperand.h, line
232.
Stack dump:
0.      Program arguments: .../libexec/clang-cc -triple
x86_64-undermydesk-freebsd8.0 -S -disable-free -main-file-name llvm-crash.c
--relocation-model static --disable-fp-elim --unwind-tables=1 --mcpu=x86-64
--fmath-errno=1 -O2 -fdiagnostics-show-option -o /tmp/cc-IyJtsf.s -x c
.../llvm-crash.c 
1.      <eof> parser at end of file
2.      Code generation
3.      Running pass 'Linear Scan Register Allocator' on function
'@sha256_block'

This is with LLVM+Clang, SVN revision 71997.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 08:20:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 08:20:22 -0500
Subject: [LLVMbugs] [Bug 4226] New: ARM does not lower all vshifts
Message-ID: <bug-4226-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4226

           Summary: ARM does not lower all vshifts
           Product: libraries
           Version: 2.5
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: xerxes at zafena.se
                CC: llvmbugs at cs.uiuc.edu


When running the included testcase in llvm test/CodeGen/X86/vshift_split.ll
using the arm backend:

llvm-as < test/CodeGen/X86/vshift_split.ll | llc -march=arm            
llc:
/usr/src/openembedded/overo/tmp/work/armv7a-angstrom-linux-gnueabi/llvm2.6-2.5+svnr20090511-r0/trunk/lib/Target/ARM/ARMISelLowering.cpp:1565:
llvm::SDValue ExpandSRx(llvm::SDNode*, llvm::SelectionDAG&, const
llvm::ARMSubtarget*): Assertion `N->getValueType(0) == MVT::i64 &&
(N->getOpcode() == ISD::SRL || N->getOpcode() == ISD::SRA) && "Unknown shift to
lower!"' failed.
Stack dump:
0.      Program arguments: llc -march=arm 
1.      Running pass 'ARM Instruction Selection' on function '@update'
Aborted
root at overo:~/llvm-test/fail# 

I have tested this on both 2.5 and svn trunk


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 08:33:32 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 08:33:32 -0500
Subject: [LLVMbugs] [Bug 4166] Assert while trying to compile using ldc with
 arm as target on x86 host
In-Reply-To: <bug-4166-206@http.llvm.org/bugs/>
Message-ID: <200905181333.n4IDXWSt013296@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4166


Xerxes R??nby <xerxes at zafena.se> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |xerxes at zafena.se
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |




--- Comment #3 from Xerxes R??nby <xerxes at zafena.se>  2009-05-18 08:33:31 ---
I have tested running the reduced testcase.ll using llc
This bug asserts on ARM using the thumb backend so i have reopened the bug.
The test case work using -march=arm and fails using -march=thumb

*OK*

root at overo:~/llvm-test/fail# llvm-as < pr4166.ll  | llc -march=arm
        .file   "<stdin>"


        .text
        .globl 
_D5tango3net7cluster4tina11CacheThread11CacheThread4loadMFE5tango3net7cluster4tina14ProtocolWriter14ProtocolWriter7CommandAaAaZv
        .align  2
_D5tango3net7cluster4tina11CacheThread11CacheThread4loadMFE5tango3net7cluster4tina14ProtocolWriter14ProtocolWriter7CommandAaAaZv:
        stmfd sp!, {r11, lr}
        sub sp, sp, #8
        @ implicit-def: r3
        str r3, [sp]
        mov r0, #0
        ldr r3, [r0]
        mov r2, #4
        ldr r2, [r2]
        str r2, [sp, #+8]
        str r3, [sp, #+4]
        @ implicit-def: r1
        @ implicit-def: r2
        @ implicit-def: r3
        @ implicit-def: lr
        mov lr, pc
        bx r0
        add sp, sp, #8
        ldmfd sp!, {r11, pc}
        .size
_D5tango3net7cluster4tina11CacheThread11CacheThread4loadMFE5tango3net7cluster4tina14ProtocolWriter14ProtocolWriter7CommandAaAaZv,
.-_D5tango3net7cluster4tina11CacheThread11CacheThread4loadMFE5tango3net7cluster4tina14ProtocolWriter14ProtocolWriter7CommandAaAaZv



*FAIL*

root at overo:~/llvm-test/fail# llvm-as < pr4166.ll  | llc -march=thumb
llc:
/usr/src/openembedded/overo/tmp/work/armv7a-angstrom-linux-gnueabi/llvm2.6-2.5+svnr20090511-r0/trunk/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp:1550:
llvm::SDValue<unnamed>::SelectionDAGLegalize::LegalizeOp(llvm::SDValue):
Assertion `(!IsLegalizingCall || Node->getOpcode() != ISD::CALL ||
Tmp3.getNode()->getOpcode() != ISD::CALLSEQ_END) && "Nested
CALLSEQ_START..CALLSEQ_END not supported."' failed.
Stack dump:
0.      Program arguments: llc -march=thumb 
1.      Running pass 'ARM Instruction Selection' on function
'@_D5tango3net7cluster4tina11CacheThread11CacheThread4loadMFE5tango3net7cluster4tina14ProtocolWriter14ProtocolWriter7CommandAaAaZv'
Aborted
root at overo:~/llvm-test/fail# 


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 08:58:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 08:58:41 -0500
Subject: [LLVMbugs] [Bug 2386] erraneous html with -emit-html
In-Reply-To: <bug-2386-206@http.llvm.org/bugs/>
Message-ID: <200905181358.n4IDwfmQ014354@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2386


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #23 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-18 08:58:35 ---
Fixed in r72000.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 09:24:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 09:24:34 -0500
Subject: [LLVMbugs] [Bug 4227] New: ARM backend fail on
	asm-large-immediate.ll testcase
Message-ID: <bug-4227-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4227

           Summary: ARM backend fail on asm-large-immediate.ll testcase
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: xerxes at zafena.se
                CC: llvmbugs at cs.uiuc.edu


The asm-large-immediate.ll testcase inserts 2 instead of 68719476738 in the
output from llc.

cat test/CodeGen/Generic/asm-large-immediate.ll
; RUN: llvm-as < %s | llc | grep 68719476738

define void @test() {
entry:
        tail call void asm sideeffect "/* result: ${0:c} */",
"i,~{dirflag},~{fpsr},~{flags}"( i64 68719476738 )
        ret void
}

root at overo:~/llvm-test/fail# llvm-as <
test/CodeGen/Generic/asm-large-immediate.ll | llc | grep result
        /* result: 2 */
root at overo:~/llvm-test/fail# llvm-as <
test/CodeGen/Generic/asm-large-immediate.ll | llc | grep 68719476738
root at overo:~/llvm-test/fail# echo $?
1
root at overo:~/llvm-test/fail# 

tested using llvm2.5 and trunk


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 14:04:21 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 14:04:21 -0500
Subject: [LLVMbugs] [Bug 4226] ARM does not lower all vshifts
In-Reply-To: <bug-4226-206@http.llvm.org/bugs/>
Message-ID: <200905181904.n4IJ4LMe025704@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4226


Evan Cheng <evan.cheng at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |evan.cheng at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |LATER




--- Comment #1 from Evan Cheng <evan.cheng at apple.com>  2009-05-18 14:04:16 ---
Currently the ARM backend does not support vector types. What are you hoping to
see?

I'm going to mark this "later" when someone added vector support.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 14:31:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 14:31:31 -0500
Subject: [LLVMbugs] [Bug 4226] ARM does not lower all vshifts
In-Reply-To: <bug-4226-206@http.llvm.org/bugs/>
Message-ID: <200905181931.n4IJVV3w026811@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4226


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|LATER                       |




--- Comment #2 from Chris Lattner <clattner at apple.com>  2009-05-18 14:31:27 ---
I'd expect to see scalar code.  The code generator should not crash on any
valid IR.  Is this a legalize problem?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 15:02:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 15:02:29 -0500
Subject: [LLVMbugs] [Bug 4228] New: Duplicate debug symbol in recursively
	included file
Message-ID: <bug-4228-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4228

           Summary: Duplicate debug symbol in recursively included file
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: jason.haslam at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Including a C file which includes itself confuses the debug generator.  The
attached test case compiles without -g but fails when compiled with -g:

clang -g -c inc_test.c
/var/folders/EQ/EQfWBPm-G2ODRK5-R1QU7E+++TI/-Tmp-/cc-rHhHSz.s:91:FATAL:Symbol
Linfo_begin1 already defined.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 15:57:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 15:57:26 -0500
Subject: [LLVMbugs] [Bug 4202] constant pools out of range
In-Reply-To: <bug-4202-206@http.llvm.org/bugs/>
Message-ID: <200905182057.n4IKvQGC029857@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4202


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Bob Wilson <bob.wilson at apple.com>  2009-05-18 15:57:11 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090518/077720.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 18 16:18:51 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 18 May 2009 16:18:51 -0500
Subject: [LLVMbugs] [Bug 4030] nightly tester build failure emails have
 wrong URL for the build log
In-Reply-To: <bug-4030-206@http.llvm.org/bugs/>
Message-ID: <200905182118.n4ILIph6030597@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4030


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bob.wilson at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Bob Wilson <bob.wilson at apple.com>  2009-05-18 16:18:28 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090518/077725.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 00:54:35 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 00:54:35 -0500
Subject: [LLVMbugs] [Bug 4091] ARM: Inline asm failure!
In-Reply-To: <bug-4091-206@http.llvm.org/bugs/>
Message-ID: <200905190554.n4J5sZmV017052@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4091


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Bob Wilson <bob.wilson at apple.com>  2009-05-19 00:54:34 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090518/077753.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 04:49:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 04:49:38 -0500
Subject: [LLVMbugs] [Bug 4232] New: llvm-ar brings the absolute path into
	archive
Message-ID: <bug-4232-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4232

           Summary: llvm-ar brings the absolute path into archive
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: dodohack at gmail.com
                CC: llvmbugs at cs.uiuc.edu


[hwalin at neptune temp]$ rm temp1.a 
[hwalin at neptune temp]$ rm temp2.a 
[hwalin at neptune temp]$ /events/build-earth/200905/nds32-elf/bin/llvm-ar rc
temp1.a  _lt_df.bc _negvsi2.bc[hwalin at neptune temp]$
/events/build-earth/200905/nds32-elf/bin/llvm-ar rc temp2.a
/nobackup/temp/_lt_df.bc /nobackup/temp/_negvsi2.bc
[hwalin at neptune temp]$ /events/build-earth/200905/nds32-elf/bin/llvm-ar t
temp1.a
_lt_df.bc
_negvsi2.bc
[hwalin at neptune temp]$ /events/build-earth/200905/nds32-elf/bin/llvm-ar t
temp2.a
/nobackup/temp/_lt_df.bc
/nobackup/temp/_negvsi2.bc


when extract the .bc from archive temp2.a, you may find nothing is extracted.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 05:04:11 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 05:04:11 -0500
Subject: [LLVMbugs] [Bug 4058] ARM: i64 arguments may be split between
 registers and stack with AAPCS ABI
In-Reply-To: <bug-4058-206@http.llvm.org/bugs/>
Message-ID: <200905191004.n4JA4Bot005782@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4058


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Bob Wilson <bob.wilson at apple.com>  2009-05-19 05:04:09 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090518/077757.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 05:04:47 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 05:04:47 -0500
Subject: [LLVMbugs] [Bug 4059] ARM: double arguments may be split between
 registers and stack with AAPCS ABI
In-Reply-To: <bug-4059-206@http.llvm.org/bugs/>
Message-ID: <200905191004.n4JA4lQN005812@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4059


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Bob Wilson <bob.wilson at apple.com>  2009-05-19 05:04:47 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090518/077757.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 05:18:17 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 05:18:17 -0500
Subject: [LLVMbugs] [Bug 4233] New: ARM: Unsupported addressing mode
Message-ID: <bug-4233-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4233

           Summary: ARM: Unsupported addressing mode
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: xerxes at zafena.se
                CC: llvmbugs at cs.uiuc.edu


Two of the llvm/testcases asserts with "Unsupported addressing mode" on ARM.

This might be similar to the SPARC bug
http://llvm.org/bugs/show_bug.cgi?id=1557 yet now on ARM.

llvm-as < test/CodeGen/Generic/2007-04-08-MultipleFrameIndices.ll | llc
-march=arm
llc:
/usr/src/openembedded/overo/tmp/work/armv7a-angstrom-linux-gnueabi/llvm2.6-2.5+svnr72105-r1/trunk/lib/Target/ARM/ARMRegisterInfo.cpp:832:
virtual void
llvm::ARMRegisterInfo::eliminateFrameIndex(llvm::ilist_iterator<llvm::MachineInstr>,
int, llvm::RegScavenger*) const: Assertion `0 && "Unsupported addressing
mode!"' failed.
Stack dump:
0.      Program arguments: llc 
1.      Running pass 'Prolog/Epilog Insertion & Frame Finalization' on function
'@stuff'
Aborted


llvm-as < test/CodeGen/Generic/2007-04-27-LargeMemObject.ll | llc -march=arm
llc:
/usr/src/openembedded/overo/tmp/work/armv7a-angstrom-linux-gnueabi/llvm2.6-2.5+svnr72105-r1/trunk/lib/Target/ARM/ARMRegisterInfo.cpp:832:
virtual void
llvm::ARMRegisterInfo::eliminateFrameIndex(llvm::ilist_iterator<llvm::MachineInstr>,
int, llvm::RegScavenger*) const: Assertion `0 && "Unsupported addressing
mode!"' failed.
Stack dump:
0.      Program arguments: llc 
1.      Running pass 'Prolog/Epilog Insertion & Frame Finalization' on function
'@test'
Aborted


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 10:10:15 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 10:10:15 -0500
Subject: [LLVMbugs] [Bug 4234] New: LLVMCodeGen will not build with Visual
	Studio 2005
Message-ID: <bug-4234-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4234

           Summary: LLVMCodeGen will not build with Visual Studio 2005
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Keywords: build-problem
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: alan.s.king at gmail.com
                CC: llvmbugs at cs.uiuc.edu


After updating to the latest version of the trunk repository, LLVM would no
longer build properly in Visual Studio 2005, giving the following compiler
errors:

error C2888: 'LiveIntervals::InstrSlots' : symbol cannot be defined within
namespace 'llvm'     C:\llvm\include\llvm\CodeGen\LiveIntervalAnalysis.h     99

error C2888: 'LiveIntervals::SRInfo' : symbol cannot be defined within
namespace 'llvm' C:\llvm\include\llvm\CodeGen\LiveIntervalAnalysis.h     477

I decided to try and isolate the source of the problem by looking at the recent
changes to files involved with the LLVMCodeGen project and found that two
recently added files (Spiller.cpp/h) use the LiveIntervals class. Digging a
little deeper I found that Spiller.h was using an unusual form of in-line
predeclaration of four classes to function. I have no idea whether this is
technically a legal form of predeclaration, but it is not going well with the
compiler.

I have attached a patch for this file that performs the predeclaration in a
more usual manner found throughout the LLVM project files and fixes the
aforementioned compiler warnings in Visual Studio 2005.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 11:35:15 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 11:35:15 -0500
Subject: [LLVMbugs] [Bug 4235] New: copyFromParts bug with valueVt=f64 ->
	PartVT=i16
Message-ID: <bug-4235-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4235

           Summary: copyFromParts bug with valueVt=f64 -> PartVT=i16
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: compile-fail
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: quickslyver at free.fr
                CC: llvmbugs at cs.uiuc.edu


there is a bug with getCopyFromParts() function from f64  (valueVT)  to i16
(PartVT)
this function try to convert recursively using this path: 
f64 -> f32 -> f16 -> i16 

but f16 doesn't exist and this finish in an assertion failure.

This bug happen with the following llvm code case on MSP430 target 
( because it's a 16 bit target)
-------------------------------------------

; ModuleID = 'test.c'
target datalayout = "e-p:16:8:8-i8:8:8-i16:8:8-i32:8:8"
target triple = "msp430-unknown-unknown"

define i16 @test(double %d) nounwind {
entry:
        %add = add double %d, 1.000000e+00              ; <double> [#uses=1]
        %call = tail call i16 @funct(double %add) nounwind              ; <i16>
[#uses=1]
        ret i16 %call
}

declare i16 @funct(double)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 12:54:56 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 12:54:56 -0500
Subject: [LLVMbugs] [Bug 4234] LLVMCodeGen will not build with Visual Studio
	2005
In-Reply-To: <bug-4234-206@http.llvm.org/bugs/>
Message-ID: <200905191754.n4JHsufb022306@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4234


Bill Wendling <wendling at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wendling at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Bill Wendling <wendling at apple.com>  2009-05-19 12:54:53 ---
Potentially Fixed:
http://llvm.org/viewvc/llvm-project?rev=72113&view=rev
http://llvm.org/viewvc/llvm-project?rev=72114&view=rev

Please try it out.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 13:33:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 13:33:12 -0500
Subject: [LLVMbugs] [Bug 4236] New: Inline asm in linux-2.4. 37 rejected by
 clang-cc SVN version 72105
Message-ID: <bug-4236-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4236

           Summary: Inline asm in linux-2.4.37 rejected by clang-cc SVN
                    version 72105
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: bagnara at cs.unipr.it
                CC: llvmbugs at cs.uiuc.edu


Both clang-cc SVN version 68830 and gcc 4.3.2 accept the following chunk 
of code (which was extracted from linux-2.4.37):

=========================================================
typedef struct {
        volatile unsigned int lock;
} spinlock_t;

static inline int spin_trylock(spinlock_t *lock)
{
        char oldval;
        __asm__ __volatile__(
                "xchgb %b0,%1"
                :"=q" (oldval), "=m" (lock->lock)
                :"0" (0) : "memory");
        return oldval > 0;
}
=========================================================


In contrast, clang-cc SVN version 72105 rejects it as follows:

bug.c:11:9: error: unsupported inline asm: input with type 'int' 
matching output
       with type 'char'
                 :"0" (0) : "memory");
                       ^
1 diagnostic generated.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 13:37:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 13:37:34 -0500
Subject: [LLVMbugs] [Bug 4233] ARM: Unsupported addressing mode
In-Reply-To: <bug-4233-206@http.llvm.org/bugs/>
Message-ID: <200905191837.n4JIbY8Z023956@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4233


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bob.wilson at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Bob Wilson <bob.wilson at apple.com>  2009-05-19 13:36:27 ---
This is mostly a duplicate of pr4091, but it also required an additional
change:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090518/077768.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 15:38:27 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 15:38:27 -0500
Subject: [LLVMbugs] [Bug 4045] llvm-gcc not using collect2,
	missing paths to .o files
In-Reply-To: <bug-4045-206@http.llvm.org/bugs/>
Message-ID: <200905192038.n4JKcRUI028380@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4045


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bob.wilson at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #3 from Bob Wilson <bob.wilson at apple.com>  2009-05-19 15:38:26 ---
You are running the compiler from the build directory.  If you want that to
work, you'll need to specify one or more -B options to tell it where to find
things.  Notice in the -print-search-dirs output that it is not looking for
programs in the build/prev-gcc directory.  Alternatively, if you install it and
run it out of the install directory (which you can change by specifying
--prefix when you configure), then it should work.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 19 16:28:52 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 19 May 2009 16:28:52 -0500
Subject: [LLVMbugs] [Bug 4227] ARM backend fail on asm-large-immediate.ll
	testcase
In-Reply-To: <bug-4227-206@http.llvm.org/bugs/>
Message-ID: <200905192128.n4JLSqh4030134@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4227


Bob Wilson <bob.wilson at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bob.wilson at apple.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Bob Wilson <bob.wilson at apple.com>  2009-05-19 16:28:51 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090518/077782.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 20 01:04:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 20 May 2009 01:04:25 -0500
Subject: [LLVMbugs] [Bug 4235] copyFromParts bug with valueVt=f64 ->
	PartVT=i16
In-Reply-To: <bug-4235-206@http.llvm.org/bugs/>
Message-ID: <200905200604.n4K64PVR014460@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4235


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-20 01:04:25 ---
Fixed in r72162.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 20 10:25:22 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 20 May 2009 10:25:22 -0500
Subject: [LLVMbugs] [Bug 4238] New: Many passes do not preserve domtree and/
 or domfrontier info properly
Message-ID: <bug-4238-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4238

           Summary: Many passes do not preserve domtree and/or domfrontier
                    info properly
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


Making this change in lib/VMCore/PassManager.cpp

-bool VerifyDomInfo = false;
+bool VerifyDomInfo = true;

so it always verifies dominator info, then running "make check" shows
that many transforms are failing to correctly preserve dom info.  The
loop passes are particularly bad, for example "loop-index-split".


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 21 15:22:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 21 May 2009 15:22:41 -0500
Subject: [LLVMbugs] [Bug 4239] New: clang: incorrect handling of
 __attribute__ ((regparm (3), stdcall))
Message-ID: <bug-4239-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4239

           Summary: clang: incorrect handling of  __attribute__ ((regparm
                    (3), stdcall))
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu, rdivacky at freebsd.org


This bug in clang causes libgnuregex to crash/hang on FreeBSD.
Same file compiled with gcc works.

Testcase:
void abort(void) __attribute__((__noreturn__));
typedef void re_string_t;
typedef void re_dfa_t;
typedef int reg_errcode_t;
static reg_errcode_t re_string_construct (re_string_t *pstr, const char *str,
       int len, char * trans,
       int icase, const re_dfa_t *dfa)
     __attribute__ ((regparm (3), stdcall));
static reg_errcode_t
re_string_construct (pstr, str, len, trans, icase, dfa)
     re_string_t *pstr;
     const char *str;
     int len, icase;
     char * trans;
     const re_dfa_t *dfa;
{
        if (dfa != (void*)0x282020c0)
                abort();
return 0;
}
int main()
{
return re_string_construct(0, 0, 0, 0, 0, (void*)0x282020c0);
}


Works with gcc, llvm-gcc -m32, fails with clang -m32.
Works with clang on x86-64 (64-bit default).

There is something wrong with parameter passing, dfa becomes 0x3.

It seems that the missing x86_stdcallcc in the call causes this.
Shouldn't the verifier reject this assembly then? (incompatible calling
conventions).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 21 15:28:31 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 21 May 2009 15:28:31 -0500
Subject: [LLVMbugs] [Bug 4240] New: clang: broken debug info for restrict
Message-ID: <bug-4240-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4240

           Summary: clang: broken debug info for restrict
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu, rdivacky at freebsd.org


Created an attachment (id=3010)
 --> (http://llvm.org/bugs/attachment.cgi?id=3010)
gnuregex.i

clang compiled gnuregex.i, cannot breakpoint in re_compile_internal, or
stacktrace into that function, because gdb shows a dwarf error:

Die: DW_TAG_restrict_type (abbrev = 33, offset = 27216)
        has children: FALSE
        attributes:
        DW_AT_type (DW_FORM_ref4) constant: 16237
Dwarf Error: Cannot find type of die [in module
/usr/obj/usr/src/gnu/lib/libregex/libgnuregex.so.4]
(gdb) bt
#0  0x280a9f4c in re_string_construct_common (str=0x0, len=3, pstr=0x280b3124,
trans=0xbfbfe860 "", icase=-1077941050, dfa=0x3) at regex_internal.c:192
 #1  0x280aa577 in re_string_construct (pstr=0x280b3124, str=0x0, len=3,
trans=0xbfbfe860 "", icase=-1077941050, dfa=0x3) at regex_internal.c:87
 Die: DW_TAG_restrict_type (abbrev = 33, offset = 48651)
        has children: FALSE
        attributes:
                DW_AT_type (DW_FORM_ref4) constant: 12449
Dwarf Error: Cannot find type of die [in module
/usr/obj/usr/src/gnu/lib/libregex/libgnuregex.so.4]

Preprocessed test file attached.
Compile on FreeBSD 32-bit as:
clang gnuregex.i -o x.so -^Cared -fPIC -DPIC -march=native -std=gnu99 -O0 -g

Then try to breakpoint at re_compile_internal, gdb refuses due to dwarf error.

When I remove all the restrict qualifiers from the gnuregex.i, gdb can
breakpoint and stacktrace that function.

Sorry if this bugreport is a bit sketchy. 
Please ask if you need more information.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 02:05:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 02:05:29 -0500
Subject: [LLVMbugs] [Bug 4241] New: Internal compile error with alias
	attribute
Message-ID: <bug-4241-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4241

           Summary: Internal compile error with alias attribute
           Product: new-bugs
           Version: unspecified
          Platform: Other
        OS/Version: Solaris
            Status: NEW
          Keywords: compile-fail
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: mike.emmerik at sun.com
                CC: llvmbugs at cs.uiuc.edu, mike.emmerik at sun.com


Created an attachment (id=3011)
 --> (http://llvm.org/bugs/attachment.cgi?id=3011)
Patch that WORKS AROUND the problem; may NOT be a complete fix

llvm-backend.cpp assumes that if a function entry is renamed, then an object
with the same name already exists in the symbol table. I speculate that when
the alias attribute is used, this is not valid.

In any case, when the code below is compiled, an ICE results because
getGlobalVariable() returns NULL:

void cfi_cmdset_0001() {}
void  *cfi_cmdset_0003() __attribute__((alias("cfi_cmdset_0001")));
extern typeof(cfi_cmdset_0003) cfi_cmdset_0003;

Steps to reproduce
==================
Compile the above or attached code with
  llvm-gcc -c -o foo test.c

Actual results
==============
test.c:8: internal compiler error: Segmentation Fault
Please submit a full bug report,
with preprocessed source if appropriate.

Expected Results
================
Normal compilation with no error; valid foo.o produced.

Build Date and Platform
=======================
2009/May/20 on Solaris 2.11 (X86) llvm-gcc is 4.2-2.4 (gcc 4.2, llvm 2.4)

Additional platforms
====================
Also happens on llvm 2.5

Workaround
==========

The code can be fixed by the attached patch. It checks for G being null, and if
so, doesn't attempt to replace the presumed old symbol with the cast of the new
one. NOTE: this may not be a proper fix; the test should determine whether
there really has been a rename or an alias, and handle the two cases
appropriately.

      // If FnEntry got renamed, then there is already an object with this name
      // in the symbol table.  If this happens, the old one must be a forward
      // decl, just replace it with a cast of the new one.
      if (FnEntry->getName() != Name) {
        GlobalVariable *G = TheModule->getGlobalVariable(Name, true);
        assert(G && G->isDeclaration() && "A global turned into a function?");
+       if (G) {                    /* G can be NULL! */
        // Replace any uses of "G" with uses of FnEntry.
        Value *GInNewType = TheFolder->CreateBitCast(FnEntry, G->getType());
        G->replaceAllUsesWith(GInNewType);

        // Update the decl that points to G.
        changeLLVMValue(G, GInNewType);

        // Now we can give GV the proper name.
        FnEntry->takeName(G);

        // G is now dead, nuke it.
        G->eraseFromParent();          
+     } /* End if */
      }


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 05:59:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 05:59:12 -0500
Subject: [LLVMbugs] [Bug 4242] New: [x86-64 ABI] struct passed partially in
 registers if there' s an sret parameter
Message-ID: <bug-4242-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4242

           Summary: [x86-64 ABI] struct passed partially in registers if
                    there's an sret parameter
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: ABI
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fvbommel at wxs.nl
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3013)
 --> (http://llvm.org/bugs/attachment.cgi?id=3013)
The code in question

There's an ABI bug in llvm-gcc for x86-64 linux.
The attached code should generate functionally identical asm for both
functions, and does so with stock g++ (version: Ubuntu 4.3.2-1ubuntu12) but not
with llvm-g++; it generates incorrect code for fail().

The issue seems to be that the sret pointer isn't taken into account when the
number of available integer registers for the struct parameter is calculated
causing that parameter to be passed with one half in a register and one in
memory (in violation of the ABI spec, which requires it to be passed entirely
in memory if it does not fit entirely into registers).

g++ output (via objdump)
----
0000000000000000 <fail>:
   0:   48 8b 44 24 08          mov    0x8(%rsp),%rax
   5:   4c 89 4f 10             mov    %r9,0x10(%rdi)
   9:   48 89 07                mov    %rax,(%rdi)
   c:   48 8b 44 24 10          mov    0x10(%rsp),%rax
  11:   48 89 47 08             mov    %rax,0x8(%rdi)
  15:   48 89 f8                mov    %rdi,%rax
  18:   c3                      retq   
----

llvm-g++ output:
----
0000000000000000 <fail>:
   0:   48 89 f8                mov    %rdi,%rax
   3:   4c 89 0f                mov    %r9,(%rdi)
   6:   48 8b 4c 24 08          mov    0x8(%rsp),%rcx
   b:   48 89 4f 08             mov    %rcx,0x8(%rdi)
   f:   48 8b 4c 24 10          mov    0x10(%rsp),%rcx
  14:   48 89 4f 10             mov    %rcx,0x10(%rdi)
  18:   c3                      retq   
----


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 07:46:42 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 07:46:42 -0500
Subject: [LLVMbugs] [Bug 4243] New: Build error during install
Message-ID: <bug-4243-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4243

           Summary: Build error during install
           Product: Build scripts
           Version: 2.5
          Platform: Macintosh
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Makefiles
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: torsten.eichstaedt at web.de
                CC: llvmbugs at cs.uiuc.edu


Ubuntu 8.04 Linux 2.6.24 powerpc 32-bit

"make all" was ok, no errors.  But "make install" (as root) then exited with:
Parsing file
/home/paul/Projects/src/llvm-2.5/lib/Target/CellSPU/SPUGenCodeEmitter.inc...
Reading
/home/paul/Projects/src/llvm-2.5/lib/Target/CellSPU/SPUGenDAGISel.inc...
input buffer overflow, can't enlarge buffer because scanner uses REJECT
Exiting...
make[1]: *** [regendoc] Error 2
make[1]: Leaving directory '/home/paul/Projects/src/llvm-2.5/docs'
make: *** [install] Error 1

Note: My locale was set to de_DE. I translated the message above, so it might
not be 100% acurate.

I tried not to build the cross compilers as a workaround, but the same error
occured. (??? I configure'd with --enable-targets=host-only)

2nd Q: where's llvm-gcc and llvm-g++?

Thanks, Paul


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 08:02:58 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 08:02:58 -0500
Subject: [LLVMbugs] [Bug 4243] Build error during install
In-Reply-To: <bug-4243-206@http.llvm.org/bugs/>
Message-ID: <200905221302.n4MD2wvo027397@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4243


Anton Korobeynikov <asl at math.spbu.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |asl at math.spbu.ru
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Anton Korobeynikov <asl at math.spbu.ru>  2009-05-22 08:02:52 ---
(In reply to comment #0)
> Parsing file
> /home/paul/Projects/src/llvm-2.5/lib/Target/CellSPU/SPUGenCodeEmitter.inc...
> Reading
> /home/paul/Projects/src/llvm-2.5/lib/Target/CellSPU/SPUGenDAGISel.inc...
> input buffer overflow, can't enlarge buffer because scanner uses REJECT
> Exiting...
That's not a LLVM problem. Your doxygen seems to be is broken.

> 2nd Q: where's llvm-gcc and llvm-g++?
In the directory, where you've built it.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 09:24:40 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 09:24:40 -0500
Subject: [LLVMbugs] [Bug 4244] New: IndVarSimplify breaks float loops
Message-ID: <bug-4244-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4244

           Summary: IndVarSimplify breaks float loops
           Product: libraries
           Version: 2.5
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jay.foad at antixlabs.com
                CC: llvmbugs at cs.uiuc.edu


If I compile this code with -O2:

$ cat x.cpp
void f(float *p) {
  int n = 0;
  for (float j = 0.0f; j <= 2.0f; j++)
    for (float i = 0.0f; i <= 13.0f; i++)
      p[n++] += i/10;
}
$ .../llvm-gcc4.2-2.5-x86-linux-RHEL4/bin/llvm-g++ -O2 -S -emit-llvm x.cpp

I get this LLVM output:

define void @_Z1fPf(float* nocapture %p) nounwind {
bb4.thread:
        br label %bb2

bb2:            ; preds = %bb2, %bb4.thread
        %i.0.reg2mem.0.int = phi i32 [ 0, %bb4.thread ], [ %indvar.next, %bb2 ]
        %indvar.conv = sitofp i32 %i.0.reg2mem.0.int to float
        %0 = getelementptr float* %p, i32 %i.0.reg2mem.0.int
        %1 = load float* %0, align 4
        %2 = fdiv float %indvar.conv, 1.000000e+01
        %3 = add float %1, %2
        store float %3, float* %0, align 4
        %indvar.next = add i32 %i.0.reg2mem.0.int, 1
        br label %bb2
}

The original code clearly has a bounded number of iterations, but this LLVM
code clearly loops forever.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 10:02:36 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 10:02:36 -0500
Subject: [LLVMbugs] [Bug 4244] IndVarSimplify breaks float loops
In-Reply-To: <bug-4244-206@http.llvm.org/bugs/>
Message-ID: <200905221502.n4MF2aXb031427@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4244


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |baldrick at free.fr
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #1 from Duncan Sands <baldrick at free.fr>  2009-05-22 10:02:36 ---
This seems to work fine if you use llvm-gcc from svn.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 12:47:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 12:47:14 -0500
Subject: [LLVMbugs] [Bug 4246] New: Share code for generating ABI conformant
 function calls/ returns
Message-ID: <bug-4246-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4246

           Summary: Share code for generating ABI conformant function
                    calls/returns
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


For platforms like x86-64, getting ABI conformance right is hard.
There is one implementation in llvm-gcc, and another in clang.
Currently, anyone writing a compiler and wanting to be able to
interface with external C functions needs to roll their own
implementation.  We should consider developing a generic solution
that can be used by llvm-gcc, clang and other projects.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 13:36:13 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 13:36:13 -0500
Subject: [LLVMbugs] [Bug 4239] clang: incorrect handling of __attribute__
 ((regparm (3), stdcall))
In-Reply-To: <bug-4239-206@http.llvm.org/bugs/>
Message-ID: <200905221836.n4MIaDq5006144@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4239


T??r??k Edwin <edwintorok at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 18:56:27 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 18:56:27 -0500
Subject: [LLVMbugs] [Bug 4247] New: [clang] An array overgflow occurs for
	128 bit floats
Message-ID: <bug-4247-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4247

           Summary: [clang] An array overgflow occurs for 128 bit floats
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rich at pennware.com
                CC: llvmbugs at cs.uiuc.edu


In the function DefineFloatMacros, there is a 60 byte array (MacroBuf) used to
build up macro definitions. When floats are IEEE long double, the string placed
in the array is exactly 60 characters not including the trailing nul:
__LDBL_DENORM_MIN__=4.94065645841246544176568792868221e-324L

This can cause all sorts of strange behavior.

InitPreprocessor.cpp


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 22 21:04:30 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 22 May 2009 21:04:30 -0500
Subject: [LLVMbugs] [Bug 4247] [clang] An array overgflow occurs for 128 bit
	floats
In-Reply-To: <bug-4247-206@http.llvm.org/bugs/>
Message-ID: <200905230204.n4N24Uqj020503@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4247


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-22 21:04:29 ---
Fixed in r72310, but in the future, it's a bit easier if you just mail a patch
to cfe-commits.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 23 06:47:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 23 May 2009 06:47:14 -0500
Subject: [LLVMbugs] [Bug 4251] New: indvars: Instruction does not dominate
	all uses!
Message-ID: <bug-4251-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4251

           Summary: indvars: Instruction does not dominate all uses!
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3016)
 --> (http://llvm.org/bugs/attachment.cgi?id=3016)
reduced .bc

When compiling vim with clang at -O2 using SVN r72324 it crashes.

Bugpoint reduced the testcase to:
$ opt -indvars bugpoint-reduced-simplified.bc
Instruction does not dominate all uses!
        %exitcond = icmp eq i32 %indvar.next, 3         ; <i1> [#uses=2]
        %conv837 = zext i1 %exitcond to i32             ; <i32> [#uses=0]
Broken module found, compilation aborted!                                
0   opt             0x00000000007c844f                                   
1   opt             0x00000000007c8849                                   
2   libpthread.so.0 0x0000003b0a80e7b0                                   
3   libc.so.6       0x0000003b09c32065 gsignal + 53                      
4   libc.so.6       0x0000003b09c35153 abort + 387                       
5   opt             0x000000000078963a
6   opt             0x0000000000762d89
llvm::FPPassManager::runOnFunction(llvm::Function&) + 489
7   opt             0x0000000000762f32
llvm::FPPassManager::runOnModule(llvm::Module&) + 50
8   opt             0x000000000076052d
llvm::MPPassManager::runOnModule(llvm::Module&) + 301
9   opt             0x0000000000760eb0
llvm::PassManagerImpl::run(llvm::Module&) + 160
10  opt             0x00000000004a66e0 main + 1488
11  libc.so.6       0x0000003b09c1e5a6 __libc_start_main + 230
12  opt             0x000000000049b3a9
Stack dump:
0.      Running pass 'Function Pass Manager' on module
'bugpoint-reduced-simplified.bc'.
1.      Running pass 'Module Verifier' on function '@write_vim_spell'
Aborted


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 23 07:42:42 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 23 May 2009 07:42:42 -0500
Subject: [LLVMbugs] [Bug 4252] New: Compiling interp.i takes 54 seconds: 78%
 time spent in Simple Register Coalescing
Message-ID: <bug-4252-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4252

           Summary: Compiling interp.i takes 54 seconds: 78% time spent in
                    Simple Register Coalescing
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3017)
 --> (http://llvm.org/bugs/attachment.cgi?id=3017)
llc interp.bc

Compiling attached file takes more than 54 seconds with clang -O.
Compiling same file with gcc takes 0.002s.

You can reproduce by:
$ llc -time-passes interp.bc

Or by running clang -O:
$ clang interp.i -c -O -ftime-report                                            
===-------------------------------------------------------------------------=== 
                      Instruction Selection and Scheduling                      
===-------------------------------------------------------------------------=== 
  Total Execution Time: 0.2400 seconds (0.2590 wall clock)                      

   ---User Time---   --System Time--   --User+System--   ---Wall Time---  ---
Name ---
   0.0480 ( 21.4%)   0.0040 ( 25.0%)   0.0520 ( 21.6%)   0.0644 ( 24.8%) 
Instruction Scheduling
   0.0400 ( 17.8%)   0.0000 (  0.0%)   0.0400 ( 16.6%)   0.0458 ( 17.7%)  DAG
Legalization      
   0.0480 ( 21.4%)   0.0040 ( 24.9%)   0.0520 ( 21.6%)   0.0443 ( 17.1%)  Type
Legalization     
   0.0200 (  8.9%)   0.0040 ( 25.0%)   0.0240 ( 10.0%)   0.0320 ( 12.3%) 
Instruction Selection 
   0.0360 ( 16.0%)   0.0040 ( 24.9%)   0.0400 ( 16.6%)   0.0298 ( 11.5%) 
Instruction Creation  
   0.0120 (  5.3%)   0.0000 (  0.0%)   0.0120 (  5.0%)   0.0199 (  7.7%)  DAG
Combining 1       
   0.0120 (  5.3%)   0.0000 (  0.0%)   0.0120 (  4.9%)   0.0137 (  5.3%)  DAG
Combining 2       
   0.0080 (  3.5%)   0.0000 (  0.0%)   0.0080 (  3.3%)   0.0079 (  3.0%) 
Instruction Scheduling Cleanup
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0006 (  0.2%)  DAG
Combining after legalize types
   0.2240 (100.0%)   0.0160 (100.0%)   0.2400 (100.0%)   0.2590 (100.0%)  TOTAL 

===-------------------------------------------------------------------------===
                      ... Pass execution timing report ...                     
===-------------------------------------------------------------------------===
  Total Execution Time: 54.6514 seconds (54.6741 wall clock)                   

   ---User Time---   --System Time--   --User+System--   ---Wall Time---  ---
Name ---
  42.7026 ( 78.2%)   0.0240 ( 21.4%)  42.7266 ( 78.1%)  42.7355 ( 78.1%) 
Simple Register Coalescing
   3.7242 (  6.8%)   0.0040 (  3.5%)   3.7282 (  6.8%)   3.7292 (  6.8%) 
Unswitch loops            
   1.4080 (  2.5%)   0.0000 (  0.0%)   1.4080 (  2.5%)   1.4124 (  2.5%) 
Canonicalize natural loops
   1.1840 (  2.1%)   0.0000 (  0.0%)   1.1840 (  2.1%)   1.1913 (  2.1%) 
Canonicalize natural loops
   0.8320 (  1.5%)   0.0000 (  0.0%)   0.8320 (  1.5%)   0.8321 (  1.5%) 
Simplify the CFG          
   0.6840 (  1.2%)   0.0000 (  0.0%)   0.6840 (  1.2%)   0.6861 (  1.2%) 
Optimize for code generation
   0.6080 (  1.1%)   0.0000 (  0.0%)   0.6080 (  1.1%)   0.6070 (  1.1%) 
Simplify the CFG            
   0.5440 (  0.9%)   0.0000 (  0.0%)   0.5440 (  0.9%)   0.5444 (  0.9%) 
Linear Scan Register Allocator
   0.4720 (  0.8%)   0.0320 ( 28.5%)   0.5040 (  0.9%)   0.5015 (  0.9%)  X86
DAG->DAG Instruction Selection
   0.3640 (  0.6%)   0.0000 (  0.0%)   0.3640 (  0.6%)   0.3662 (  0.6%)  Break
critical edges in CFG       
   0.3280 (  0.6%)   0.0000 (  0.0%)   0.3280 (  0.6%)   0.3248 (  0.5%) 
Global Value Numbering            
   0.1920 (  0.3%)   0.0000 (  0.0%)   0.1920 (  0.3%)   0.1925 (  0.3%) 
Control Flow Optimizer            
   0.1120 (  0.2%)   0.0000 (  0.0%)   0.1120 (  0.2%)   0.1104 (  0.2%) 
Eliminate PHI nodes for register allocation
   0.1120 (  0.2%)   0.0040 (  3.5%)   0.1160 (  0.2%)   0.1081 (  0.1%) 
Canonicalize Induction Variables           
   0.1040 (  0.1%)   0.0000 (  0.0%)   0.1040 (  0.1%)   0.0961 (  0.1%)  Loop
Strength Reduction                    
   0.0680 (  0.1%)   0.0000 (  0.0%)   0.0680 (  0.1%)   0.0906 (  0.1%) 
Induction Variable Users                   
   0.0600 (  0.1%)   0.0120 ( 10.7%)   0.0720 (  0.1%)   0.0764 (  0.1%)  Live
Interval Analysis                     
   0.0480 (  0.0%)   0.0200 ( 17.8%)   0.0680 (  0.1%)   0.0714 (  0.1%)  Live
Variable Analysis                     
   0.0720 (  0.1%)   0.0040 (  3.5%)   0.0760 (  0.1%)   0.0709 (  0.1%) 
Loop-Closed SSA Form Pass                  
   0.0680 (  0.1%)   0.0000 (  0.0%)   0.0680 (  0.1%)   0.0685 (  0.1%) 
Simplify the CFG                           
   0.0640 (  0.1%)   0.0000 (  0.0%)   0.0640 (  0.1%)   0.0641 (  0.1%) 
Combine redundant instructions             
   0.0600 (  0.1%)   0.0000 (  0.0%)   0.0600 (  0.1%)   0.0593 (  0.1%) 
Combine redundant instructions             
   0.0520 (  0.0%)   0.0000 (  0.0%)   0.0520 (  0.0%)   0.0528 (  0.0%) 
Promote Memory to Register                 
   0.0520 (  0.0%)   0.0000 (  0.0%)   0.0520 (  0.0%)   0.0511 (  0.0%) 
Loop-Closed SSA Form Pass                  
   0.0360 (  0.0%)   0.0000 (  0.0%)   0.0360 (  0.0%)   0.0475 (  0.0%) 
Induction Variable Users                   
   0.0480 (  0.0%)   0.0000 (  0.0%)   0.0480 (  0.0%)   0.0465 (  0.0%) 
Simplify the CFG                           
   0.0400 (  0.0%)   0.0000 (  0.0%)   0.0400 (  0.0%)   0.0392 (  0.0%) 
Sparse Conditional Constant Propagation    
   0.0400 (  0.0%)   0.0000 (  0.0%)   0.0400 (  0.0%)   0.0392 (  0.0%) 
Dominator Tree Construction                
   0.0320 (  0.0%)   0.0000 (  0.0%)   0.0320 (  0.0%)   0.0317 (  0.0%) 
Two-Address instruction pass               
   0.0280 (  0.0%)   0.0040 (  3.5%)   0.0320 (  0.0%)   0.0290 (  0.0%)  Loop
Invariant Code Motion                 
   0.0240 (  0.0%)   0.0000 (  0.0%)   0.0240 (  0.0%)   0.0231 (  0.0%) 
Simplify the CFG                           
   0.0200 (  0.0%)   0.0000 (  0.0%)   0.0200 (  0.0%)   0.0222 (  0.0%) 
Combine redundant instructions             
   0.0200 (  0.0%)   0.0000 (  0.0%)   0.0200 (  0.0%)   0.0205 (  0.0%) 
Combine redundant instructions             
   0.0240 (  0.0%)   0.0000 (  0.0%)   0.0240 (  0.0%)   0.0204 (  0.0%) 
Dominance Frontier Construction            
   0.0160 (  0.0%)   0.0000 (  0.0%)   0.0160 (  0.0%)   0.0195 (  0.0%) 
Dominance Frontier Construction            
   0.0160 (  0.0%)   0.0000 (  0.0%)   0.0160 (  0.0%)   0.0153 (  0.0%)  Break
critical edges in CFG                
   0.0160 (  0.0%)   0.0000 (  0.0%)   0.0160 (  0.0%)   0.0148 (  0.0%)  Stack
Slot Coloring                        
   0.0120 (  0.0%)   0.0000 (  0.0%)   0.0120 (  0.0%)   0.0141 (  0.0%) 
Dominator Tree Construction                
   0.0120 (  0.0%)   0.0000 (  0.0%)   0.0120 (  0.0%)   0.0129 (  0.0%) 
MachineDominator Tree Construction         
   0.0119 (  0.0%)   0.0000 (  0.0%)   0.0119 (  0.0%)   0.0126 (  0.0%) 
Simplify the CFG                           
   0.0120 (  0.0%)   0.0000 (  0.0%)   0.0120 (  0.0%)   0.0124 (  0.0%) 
Dominator Tree Construction                
   0.0200 (  0.0%)   0.0000 (  0.0%)   0.0200 (  0.0%)   0.0115 (  0.0%) 
Delete dead loops                          
   0.0079 (  0.0%)   0.0000 (  0.0%)   0.0079 (  0.0%)   0.0111 (  0.0%) 
MachineDominator Tree Construction         
   0.0079 (  0.0%)   0.0040 (  3.5%)   0.0120 (  0.0%)   0.0105 (  0.0%) 
Machine Code Deleter                       
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0096 (  0.0%) 
Aggressive Dead Code Elimination           
   0.0079 (  0.0%)   0.0000 (  0.0%)   0.0079 (  0.0%)   0.0095 (  0.0%)  Jump
Threading                             
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0092 (  0.0%) 
Machine code sinking                       
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0078 (  0.0%) 
Dominator Tree Construction                
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0076 (  0.0%) 
Dominator Tree Construction                
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0075 (  0.0%) 
Combine redundant instructions             
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0071 (  0.0%)  Find
Used Types                            
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0069 (  0.0%) 
Dominator Tree Construction                
   0.0079 (  0.0%)   0.0000 (  0.0%)   0.0079 (  0.0%)   0.0068 (  0.0%) 
Natural Loop Information                   
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0068 (  0.0%) 
Dominator Tree Construction                
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0067 (  0.0%)  X86
AT&T-Style Assembly Printer            
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0061 (  0.0%) 
Prolog/Epilog Insertion & Frame Finalization
   0.0000 (  0.0%)   0.0040 (  3.5%)   0.0040 (  0.0%)   0.0058 (  0.0%) 
Dominance Frontier Construction             
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0057 (  0.0%) 
Remove unreachable blocks from the CFG      
   0.0080 (  0.0%)   0.0000 (  0.0%)   0.0080 (  0.0%)   0.0054 (  0.0%) 
Remove unreachable machine basic blocks     
   0.0079 (  0.0%)   0.0000 (  0.0%)   0.0079 (  0.0%)   0.0053 (  0.0%) 
Natural Loop Information                    
   0.0039 (  0.0%)   0.0000 (  0.0%)   0.0039 (  0.0%)   0.0051 (  0.0%) 
Conditional Propagation                     
   0.0039 (  0.0%)   0.0000 (  0.0%)   0.0039 (  0.0%)   0.0046 (  0.0%) 
MachineDominator Tree Construction          
   0.0039 (  0.0%)   0.0000 (  0.0%)   0.0039 (  0.0%)   0.0040 (  0.0%) 
Dominance Frontier Construction             
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0040 (  0.0%) 
Dominance Frontier Construction             
   0.0039 (  0.0%)   0.0000 (  0.0%)   0.0039 (  0.0%)   0.0038 (  0.0%) 
Combine redundant instructions              
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0038 (  0.0%) 
Combine redundant instructions              
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0033 (  0.0%)  Dead
Store Elimination                      
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0032 (  0.0%) 
Virtual Register Map                        
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0029 (  0.0%) 
Machine Natural Loop Construction           
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0027 (  0.0%) 
Machine Natural Loop Construction           
   0.0039 (  0.0%)   0.0000 (  0.0%)   0.0039 (  0.0%)   0.0026 (  0.0%)  X86
FP Stackifier                           
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0023 (  0.0%) 
Loop-Closed SSA Form Pass                   
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0020 (  0.0%) 
Module Verifier                             
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0019 (  0.0%) 
Exception handling preparation              
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0018 (  0.0%) 
Reassociate expressions                     
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0018 (  0.0%) 
MemCpy Optimization                         
   0.0039 (  0.0%)   0.0000 (  0.0%)   0.0039 (  0.0%)   0.0017 (  0.0%) 
Conditional Propagation                     
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0012 (  0.0%)  Dead
Global Elimination                     
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0012 (  0.0%) 
Rotate Loops                                
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0010 (  0.0%) 
Machine Natural Loop Construction           
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0007 (  0.0%)  Tail
Call Elimination                       
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0006 (  0.0%) 
Subregister lowering instruction pass       
   0.0040 (  0.0%)   0.0000 (  0.0%)   0.0040 (  0.0%)   0.0006 (  0.0%) 
Memory Dependence Analysis                  
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0005 (  0.0%) 
Scalar Evolution Analysis                   
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0003 (  0.0%) 
Machine Instruction LICM                    
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0003 (  0.0%) 
Scalar Evolution Analysis                   
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0003 (  0.0%)  Label
Folder                                
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0002 (  0.0%) 
Memory Dependence Analysis                  
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0001 (  0.0%)  Code
Placement Optimizater                  
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0001 (  0.0%)  Basic
CallGraph Construction                
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  Live
Stack Slot Analysis                    
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Remove unused exception handling info       
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Inliner for always_inline functions         
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  X86
Maximal Stack Alignment Calculator      
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Simplify well-known library calls           
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Deduce function attributes                  
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Memory Dependence Analysis                  
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  Dead
Argument Elimination                   
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Scalar Replacement of Aggregates            
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  Dead
Type Elimination                       
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Global Variable Optimizer                   
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Preliminary module verification             
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Interprocedural constant propagation        
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  X86
FP_REG_KILL inserter
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  Basic
Alias Analysis (default AA impl)
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Promote Memory to Register
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Analyze Machine Code For Garbage Collection
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  Strip
Unused Function Prototypes
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Insert stack protectors
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Target Data Layout
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  Raise
allocations from calls to instructions
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)  Lower
Garbage Collection Instructions
   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%)   0.0000 (  0.0%) 
Delete Garbage Collector Information
  54.5394 (100.0%)   0.1120 (100.0%)  54.6514 (100.0%)  54.6741 (100.0%)  TOTAL

clang-cc: /home/edwin/llvm-svn/llvm/include/llvm/Support/Timer.h:158:
llvm::TimerGroup::~TimerGroup(): Assertion `NumTimers == 0 && "TimerGroup
destroyed before all contained timers!"' failed.                                
0   clang-cc        0x00000000010ec94f                                          
1   clang-cc        0x00000000010ecd49                                          
2   libpthread.so.0 0x0000003b0a80e7b0                                          
3   libc.so.6       0x0000003b09c32065 gsignal + 53                             
4   libc.so.6       0x0000003b09c35153 abort + 387                              
5   libc.so.6       0x0000003b09c2b159 __assert_fail + 233                      
6   clang-cc        0x0000000000dbcc5c llvm::TimerGroup::~TimerGroup() + 124    
7   libc.so.6       0x0000003b09c367dd exit + 157                               
8   libc.so.6       0x0000003b09c1e5ad __libc_start_main + 237                  
9   clang-cc        0x000000000042d479


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 23 08:21:11 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 23 May 2009 08:21:11 -0500
Subject: [LLVMbugs] [Bug 4253] New: available_externally causes wrong PIC
 code generation
Message-ID: <bug-4253-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4253

           Summary: available_externally causes wrong PIC code generation
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3019)
 --> (http://llvm.org/bugs/attachment.cgi?id=3019)
tst.i

$ clang tst.i -fPIC -o tst.o -c
$ gcc -shared -o tst.so tst.o
/usr/bin/ld: tst.o: relocation R_X86_64_PC32 against undefined symbol
`atoi@@GLIBC_2.2.5' can not be used when making a shared object; recompile with
-fPIC

The IR looks like this:
define available_externally fastcc i32 @atoi(i8* %__nptr) nounwind readonly {
entry:
        %call = tail call i64 @strtol(i8* nocapture %__nptr, i8** null, i32 10)
nounwind readonly               ; <i64> [#uses=1]
        %conv = trunc i64 %call to i32          ; <i32> [#uses=1]
        ret i32 %conv
}

If I remove the 'available_externally' flag the file is compiled correctly.

The problem is that llc emits this for available_externally, probably thinking
that atoi is a local function:
call    atoi


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 23 10:19:10 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 23 May 2009 10:19:10 -0500
Subject: [LLVMbugs] [Bug 4254] New: ocaml 3.11.0 miscompiled,
 bugpoint says -instcombine is to blame
Message-ID: <bug-4254-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4254

           Summary: ocaml 3.11.0 miscompiled, bugpoint says -instcombine is
                    to blame
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3021)
 --> (http://llvm.org/bugs/attachment.cgi?id=3021)
bugpoint-tooptimize.bc

Symptom: the ocamlrun build by clang -O rejects pervasives.mli with a syntax
error, the gcc built one doesn't, and neither does a 'clang' built ocamlrun:
File "../stdlib/pervasives.mli", line 29, characters 0-8:
Error: Syntax error

I've run bugpoint like this:
bugpoint -run-jit -safe-run-llc  -output=good -simplifycfg -domtree
-domfrontier -mem2reg -instcombine -simplifycfg -simplify-libcalls -instcombine
-jump-threading -simplifycfg -domtree -domfrontier -scalarrepl -instcombine
-break-crit-edges -condprop -tailcallelim -simplifycfg -reassociate -domtree
-loops -loopsimplify -domfrontier -lcssa -loop-rotate -licm -lcssa
-loop-unswitch -loop-index-split -instcombine -scalar-evolution -lcssa
-iv-users -indvars -loop-deletion -instcombine -memdep -gvn -memdep -memcpyopt
-sccp -instcombine -break-crit-edges -condprop -domtree -memdep -dse -adce
-simplifycfg -preverify -domtree -verify ocamlrun.bc -Xlinker=-lm
-Xlinker=-lpthread -Xlinker=-ldl -Xlinker=-lcurses --args -- ../boot/ocamlc -g
-warn-error A -nostdlib-nopervasives -c ../stdlib/pervasives.mli

And I got 2 bitcodes: bugpoint-tooptimize.bc, and bugpoint-tonotoptimize.bc.
Bugpoint says -instcombine breaks the program.

Here is what instcombine does:
-       %tmp128 = load i8** @intern_src         ; <i8*> [#uses=1]
-       %arrayidx129 = getelementptr i8* %tmp128, i64 -2                ; <i8*>
[#uses=1]
-       %tmp130 = load i8* %arrayidx129         ; <i8> [#uses=1]
-       %conv131 = zext i8 %tmp130 to i32               ; <i32> [#uses=1]
-       %conv132 = sext i32 %conv131 to i64             ; <i64> [#uses=1]
-       %shl133 = shl i64 %conv132, 56          ; <i64> [#uses=1]
-       %shr134 = ashr i64 %shl133, 56          ; <i64> [#uses=1]
-       %shl135 = shl i64 %shr134, 8            ; <i64> [#uses=1]
-       %tmp136 = load i8** @intern_src         ; <i8*> [#uses=1]
-       %arrayidx137 = getelementptr i8* %tmp136, i64 -1                ; <i8*>
[#uses=1]
+       %tmp130 = load i8* %tmp126              ; <i8> [#uses=1]
+       %conv1322 = zext i8 %tmp130 to i64              ; <i64> [#uses=1]
+       %shl133 = shl i64 %conv1322, 56         ; <i64> [#uses=1]
+       %0 = ashr i64 %shl133, 48               ; <i64> [#uses=1]
+       %arrayidx137 = getelementptr i8* %tmp126, i64 1         ; <i8*>
[#uses=1]
        %tmp138 = load i8* %arrayidx137         ; <i8> [#uses=1]
        %conv139 = zext i8 %tmp138 to i64               ; <i64> [#uses=1]
-       %add140 = add i64 %shl135, %conv139             ; <i64> [#uses=1]
+       %add140 = or i64 %0, %conv139           ; <i64> [#uses=1]
        %shl141 = shl i64 %add140, 1            ; <i64> [#uses=1]
-       %add142 = add i64 %shl141, 1            ; <i64> [#uses=1]
-       store i64 %add142, i64* %v
+       %add1421 = or i64 %shl141, 1            ; <i64> [#uses=1]
+       store i64 %add1421, i64* %v


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 23 12:42:24 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 23 May 2009 12:42:24 -0500
Subject: [LLVMbugs] [Bug 4254] ocaml 3.11.0 miscompiled,
	-instcombine or codegen bug
In-Reply-To: <bug-4254-206@http.llvm.org/bugs/>
Message-ID: <200905231742.n4NHgOJ4030112@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4254


T??r??k Edwin <edwintorok at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #10 from T??r??k Edwin <edwintorok at gmail.com>  2009-05-23 12:42:22 ---
ocamlbyterun works correctly now with -O1.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 24 03:55:40 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 24 May 2009 03:55:40 -0500
Subject: [LLVMbugs] [Bug 4255] New: GlobalVariable.h:99: llvm::Constant*
 llvm::GlobalVariable:: getInitializer(): Assertion `hasInitializer() && "GV
 doesn' t have initializer!"' failed.
Message-ID: <bug-4255-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4255

           Summary: GlobalVariable.h:99: llvm::Constant*
                    llvm::GlobalVariable::getInitializer(): Assertion
                    `hasInitializer() && "GV doesn't have initializer!"'
                    failed.
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


$ opt bugpoint-reduced-simplified.bc -inline -scalarrepl
WARNING: You're attempting to print out a bitcode file.
This is inadvisable as it may cause display problems. If
you REALLY want to taste LLVM bitcode first-hand, you
can force output with the `-f' option.

opt: /home/edwin/llvm-svn/llvm/include/llvm/GlobalVariable.h:99:
llvm::Constant* llvm::GlobalVariable::getInitializer(): Assertion
`hasInitializer() && "GV doesn't have initializer!"' failed.
0   opt 0x084ce4e8
Stack dump:
0.      Running pass 'CallGraph Pass Manager' on module
'bugpoint-reduced-simplified.bc'.
Aborted


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 24 04:00:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 24 May 2009 04:00:29 -0500
Subject: [LLVMbugs] [Bug 4256] New: MemoryDependenceAnalysis.cpp:93:
 Assertion `InstIt != ReverseMap.end() && "Reverse map out of sync?"' failed
Message-ID: <bug-4256-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4256

           Summary: MemoryDependenceAnalysis.cpp:93: Assertion `InstIt !=
                    ReverseMap.end() && "Reverse map out of sync?"' failed
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3026)
 --> (http://llvm.org/bugs/attachment.cgi?id=3026)
bugpoint-reduced-simplified.bc

$ opt bugpoint-reduced-simplified.bc -gvn
WARNING: You're attempting to print out a bitcode file.
This is inadvisable as it may cause display problems. If
you REALLY want to taste LLVM bitcode first-hand, you
can force output with the `-f' option.

opt: /home/edwin/llvm-svn/llvm/lib/Analysis/MemoryDependenceAnalysis.cpp:93:
void RemoveFromReverseMap(llvm::DenseMap<llvm::Instruction*,
llvm::SmallPtrSet<KeyTy, 4u>, llvm::DenseMapInfo<llvm::Instruction*>,
llvm::DenseMapInfo<llvm::SmallPtrSet<KeyTy, 4u> > >&, llvm::Instruction*,
KeyTy) [with KeyTy = llvm::PointerIntPair<llvm::Value*, 1u, bool,
llvm::PointerLikeTypeTraits<llvm::Value*> >]: Assertion `InstIt !=
ReverseMap.end() && "Reverse map out of sync?"' failed.
0   opt 0x084ce4e8
Stack dump:
0.      Running pass 'Function Pass Manager' on module
'bugpoint-reduced-simplified.bc'.
1.      Running pass 'Global Value Numbering' on function '@lbackref'
Aborted


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 24 04:12:11 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 24 May 2009 04:12:11 -0500
Subject: [LLVMbugs] [Bug 4257] New: llvm-gcc bootstrap failure in r72358
Message-ID: <bug-4257-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4257

           Summary: llvm-gcc bootstrap failure in r72358
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3027)
 --> (http://llvm.org/bugs/attachment.cgi?id=3027)
x.bc

Reduced to:
$ ~/llvm-bootstrap/llvm-obj/Release/bin/llc -mcpu=i686 x.bc -f                  
llc:
/home/edwin/llvm-bootstrap/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp:996:
llvm::SDValue<unnamed>::SelectionDAGLegalize::LegalizeOp(llvm::SDValue):
Assertion `(isTypeLegal(Node->getOperand(i).getValueType()) ||
Node->getOperand(i).getOpcode() == ISD::TargetConstant) && "Unexpected illegal
type!"' failed.
0   llc             0x0000000000d62b2f
1   llc             0x0000000000d62f29
2   libpthread.so.0 0x0000003b0a80e7b0
3   libc.so.6       0x0000003b09c32065 gsignal + 53
4   libc.so.6       0x0000003b09c35153 abort + 387
5   libc.so.6       0x0000003b09c2b159 __assert_fail + 233
6   llc             0x0000000000a4537c
7   llc             0x0000000000a52c2e
8   llc             0x0000000000a5685d
9   llc             0x0000000000a5b7e8
10  llc             0x0000000000a5bb18
11  llc             0x0000000000a5c79d llvm::SelectionDAG::Legalize(bool,
llvm::CodeGenOpt::Level) + 477
12  llc             0x0000000000a1d64b
llvm::SelectionDAGISel::CodeGenAndEmitDAG() + 2315
13  llc             0x0000000000a20a1a
llvm::SelectionDAGISel::SelectBasicBlock(llvm::BasicBlock*,
llvm::ilist_iterator<llvm::Instruction>,
llvm::ilist_iterator<llvm::Instruction>) + 298
14  llc             0x0000000000a20e12
llvm::SelectionDAGISel::SelectAllBasicBlocks(llvm::Function&,
llvm::MachineFunction&, llvm::MachineModuleInfo*, llvm::DwarfWriter*,
llvm::TargetInstrInfo const&) + 786
15  llc             0x0000000000a22209
llvm::SelectionDAGISel::runOnFunction(llvm::Function&) + 1513
16  llc             0x0000000000cfcb89
llvm::FPPassManager::runOnFunction(llvm::Function&) + 489
17  llc             0x0000000000cfce46
llvm::FunctionPassManagerImpl::run(llvm::Function&) + 166
18  llc             0x0000000000cfd08b
llvm::FunctionPassManager::run(llvm::Function&) + 75
19  llc             0x0000000000516a89 main + 4633
20  libc.so.6       0x0000003b09c1e5a6 __libc_start_main + 230
21  llc             0x0000000000514cb9
Stack dump:
0.      Program arguments: /home/edwin/llvm-bootstrap/llvm-obj/Release/bin/llc
-mcpu=i686 x.bc -f
1.      Running pass 'X86 DAG->DAG Instruction Selection' on function
'@__muldc3'
Aborted


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 24 04:44:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 24 May 2009 04:44:38 -0500
Subject: [LLVMbugs] [Bug 4258] New: -indvars: use after free and crash
Message-ID: <bug-4258-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4258

           Summary: -indvars: use after free and crash
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


$ Release/bin/opt bugpoint-reduced-blocks.bc -inline -simplifycfg -loop-rotate
-licm -loop-index-split -indvars
WARNING: You're attempting to print out a bitcode file.
This is inadvisable as it may cause display problems. If
you REALLY want to taste LLVM bitcode first-hand, you
can force output with the `-f' option.

0   opt             0x00000000007c82df
1   opt             0x00000000007c86d9
2   libpthread.so.0 0x0000003b0a80e7b0
3   opt             0x0000000000603d2a
llvm::RecursivelyDeleteTriviallyDeadInstructions(llvm::Value*) + 314
4   opt             0x0000000000528159
5   opt             0x00000000005299f4
6   opt             0x00000000006848be
llvm::LPPassManager::runOnFunction(llvm::Function&) + 1054
7   opt             0x0000000000762c29
llvm::FPPassManager::runOnFunction(llvm::Function&) + 489
8   opt             0x0000000000648dce
9   opt             0x00000000007603cd
llvm::MPPassManager::runOnModule(llvm::Module&) + 301
10  opt             0x0000000000760d50
llvm::PassManagerImpl::run(llvm::Module&) + 160
11  opt             0x00000000004a66e0 main + 1488
12  libc.so.6       0x0000003b09c1e5a6 __libc_start_main + 230
13  opt             0x000000000049b3a9
Stack dump:
0.      Running pass 'CallGraph Pass Manager' on module
'bugpoint-reduced-blocks.bc'.
1.      Running pass 'Loop Pass Manager' on function 'Segmentation fault


If I create a bitcode with -inline -simplifycfg -loop-rotate -licm
-loop-index-split, and run -indvars on the result it doesn't crash (and its
valgrind clean too).

Valgrind shows use-after-free in indvars:
==13974== Invalid read of size 8
==13974==    at 0x4B21B0: llvm::Value::getType() const (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)
==13974==    by 0x6C1636: llvm::SCEVExpander::visitAddExpr(llvm::SCEVAddExpr
const*) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)                      
==13974==    by 0x6BFA1F: llvm::SCEVExpander::expand(llvm::SCEV const*) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)          
==13974==    by 0x6BFB0D: llvm::SCEVExpander::expandCodeFor(llvm::SCEVHandle,
llvm::Type const*) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)           
==13974==    by 0x52B304: llvm::SCEVExpander::expandCodeFor(llvm::SCEVHandle,
llvm::Type const*, llvm::ilist_iterator<llvm::Instruction>) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)                                  
==13974==    by 0x528096: (anonymous
namespace)::IndVarSimplify::RewriteLoopExitValues(llvm::LoopBase<llvm::BasicBlock>*,
llvm::SCEV const*) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)           
==13974==    by 0x5299F3: (anonymous
namespace)::IndVarSimplify::runOnLoop(llvm::LoopBase<llvm::BasicBlock>*,
llvm::LPPassManager&) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)        
==13974==    by 0x6848BD: llvm::LPPassManager::runOnFunction(llvm::Function&)
(in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)    
==13974==    by 0x762C28: llvm::FPPassManager::runOnFunction(llvm::Function&)
(in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)    
==13974==    by 0x648DCD: (anonymous
namespace)::CGPassManager::runOnModule(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)                                  
==13974==    by 0x7603CC: llvm::MPPassManager::runOnModule(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)        
==13974==    by 0x760D4F: llvm::PassManagerImpl::run(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)
==13974==    by 0x6BFA1F: llvm::SCEVExpander::expand(llvm::SCEV const*) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)          
==13974==    by 0x6BFB0D: llvm::SCEVExpander::expandCodeFor(llvm::SCEVHandle,
llvm::Type const*) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)           
==13974==    by 0x52B304: llvm::SCEVExpander::expandCodeFor(llvm::SCEVHandle,
llvm::Type const*, llvm::ilist_iterator<llvm::Instruction>) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)                                  
==13974==    by 0x528096: (anonymous
namespace)::IndVarSimplify::RewriteLoopExitValues(llvm::LoopBase<llvm::BasicBlock>*,
llvm::SCEV const*) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)           
==13974==    by 0x5299F3: (anonymous
namespace)::IndVarSimplify::runOnLoop(llvm::LoopBase<llvm::BasicBlock>*,
llvm::LPPassManager&) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)        
==13974==    by 0x6848BD: llvm::LPPassManager::runOnFunction(llvm::Function&)
(in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)    
==13974==    by 0x762C28: llvm::FPPassManager::runOnFunction(llvm::Function&)
(in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)    
==13974==    by 0x648DCD: (anonymous
namespace)::CGPassManager::runOnModule(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)                                  
==13974==    by 0x7603CC: llvm::MPPassManager::runOnModule(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)        
==13974==    by 0x760D4F: llvm::PassManagerImpl::run(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)              
==13974==  Address 0x6f5dbf8 is 64 bytes inside a block of size 128 free'd      
==13974==    at 0x4A0711D: operator delete(void*) (vg_replace_malloc.c:342)     
==13974==    by 0x603DB7:
llvm::RecursivelyDeleteTriviallyDeadInstructions(llvm::Value*) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)                                  
==13974==    by 0x528158: (anonymous
namespace)::IndVarSimplify::RewriteLoopExitValues(llvm::LoopBase<llvm::BasicBlock>*,
llvm::SCEV const*) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)           
==13974==    by 0x5299F3: (anonymous
namespace)::IndVarSimplify::runOnLoop(llvm::LoopBase<llvm::BasicBlock>*,
llvm::LPPassManager&) (in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)        
==13974==    by 0x6848BD: llvm::LPPassManager::runOnFunction(llvm::Function&)
(in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)    
==13974==    by 0x762C28: llvm::FPPassManager::runOnFunction(llvm::Function&)
(in /home/edwin/llvm-svn/llvm-obj/Release/bin/opt)    
==13974==    by 0x648DCD: (anonymous
namespace)::CGPassManager::runOnModule(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)                                  
==13974==    by 0x7603CC: llvm::MPPassManager::runOnModule(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)        
==13974==    by 0x760D4F: llvm::PassManagerImpl::run(llvm::Module&) (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)              
==13974==    by 0x4A66DF: main (in
/home/edwin/llvm-svn/llvm-obj/Release/bin/opt)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 24 04:50:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 24 May 2009 04:50:25 -0500
Subject: [LLVMbugs] [Bug 4259] New: bugpoint creates broken module during
 instruction simplification
Message-ID: <bug-4259-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4259

           Summary: bugpoint creates broken module during instruction
                    simplification
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Using svn r72361 that has bug reported in PR4258, if I let bugpoint run further
to simplify instructions in the testcase, it crashes because it creates a
broken module.

Testcase is same as for PR4258, but this time for bugpoint, instead of opt:
http://llvm.org/bugs/attachment.cgi?id=3028&action=edit

$ Release/bin/bugpoint bugpoint-reduced-blocks.bc -inline -simplifycfg
-loop-rotate -licm -loop-index-split -indvars

*** Attempting to reduce testcase by deleting instructions: Simplification
Level 
#1
...
Checking instruction:   call void @llvm.dbg.func.start({ }* bitcast
(%llvm.dbg.subprogram.type* @llvm.dbg.subprogram17754115 to { }*))              
invalid llvm.dbg.declare intrinsic call                                         
        call void @llvm.dbg.declare({ }* null, { }* bitcast
(%llvm.dbg.variable.type* @llvm.dbg.variable22454162 to { }*))          
invalid llvm.dbg.declare intrinsic call                                         
        call void @llvm.dbg.declare({ }* null, { }* bitcast
(%llvm.dbg.variable.type* @llvm.dbg.variable22954167 to { }*))          
Broken module found, compilation aborted!                                       
0   bugpoint        0x000000000081498f  
1   bugpoint        0x0000000000814d89                                          
2   libpthread.so.0 0x0000003b0a80e7b0                                          
3   libc.so.6       0x0000003b09c32065 gsignal + 53                             
4   libc.so.6       0x0000003b09c35153 abort + 387                              
5   bugpoint        0x00000000007d422a                                          
6   bugpoint        0x00000000007adb39
llvm::FPPassManager::runOnFunction(llvm::Function&) + 489                       
7   bugpoint        0x000000000069e51e                                          
8   bugpoint        0x00000000007ab2dd
llvm::MPPassManager::runOnModule(llvm::Module&) + 301                           
9   bugpoint        0x00000000007abc60
llvm::PassManagerImpl::run(llvm::Module&) + 160                                 
10  bugpoint        0x00000000004c1679
llvm::BugDriver::runPassesAsChild(std::vector<llvm::PassInfo const*,
std::allocator<llvm::PassInfo const*> > const&) + 953                           
11  bugpoint        0x00000000004a5af9 llvm::BugDriver::run() + 265             
12  bugpoint        0x00000000004cb678 main + 232                               
13  libc.so.6       0x0000003b09c1e5a6 __libc_start_main + 230                  
14  bugpoint        0x00000000004a4f79                                          
Stack dump:                                                                     
0.      Program arguments: Release/bin/bugpoint -as-child -child-output
bugpoint-output.bc -inline bugpoint-input.bc                
1.      Running pass 'CallGraph Pass Manager' on module 'bugpoint-input.bc'.    
2.      Running pass 'Module Verifier' on function '@BZ2_blockSort'  
Crashed with signal #6
Dumped core
Checking instruction:   %0 = bitcast [256 x i32]* null to { }*          ; <{
}*> [#uses=1]
invalid llvm.dbg.declare intrinsic call
        call void @llvm.dbg.declare({ }* null, { }* bitcast
(%llvm.dbg.variable.type* @llvm.dbg.variable22454162 to { }*))
Broken module found, compilation aborted!
0   bugpoint        0x000000000081498f
1   bugpoint        0x0000000000814d89
2   libpthread.so.0 0x0000003b0a80e7b0
3   libc.so.6       0x0000003b09c32065 gsignal + 53
4   libc.so.6       0x0000003b09c35153 abort + 387
5   bugpoint        0x00000000007d422a
6   bugpoint        0x00000000007adb39
llvm::FPPassManager::runOnFunction(llvm::Function&) + 489
7   bugpoint        0x00000000007adce2
llvm::FPPassManager::runOnModule(llvm::Module&) + 50
8   bugpoint        0x00000000007ab2dd
llvm::MPPassManager::runOnModule(llvm::Module&) + 301
9   bugpoint        0x00000000007abc60
llvm::PassManagerImpl::run(llvm::Module&) + 160
10  bugpoint        0x00000000004b436f
llvm::BugDriver::deleteInstructionFromProgram(llvm::Instruction const*,
unsigned int) const + 831
11  bugpoint        0x00000000004a97d6
12  bugpoint        0x00000000004aa579
llvm::BugDriver::debugOptimizerCrash(std::string const&) + 281
13  bugpoint        0x00000000004a5b42 llvm::BugDriver::run() + 338
14  bugpoint        0x00000000004cb678 main + 232
15  libc.so.6       0x0000003b09c1e5a6 __libc_start_main + 230
16  bugpoint        0x00000000004a4f79
Stack dump:
0.      Program arguments: Release/bin/bugpoint bugpoint-reduced-blocks.bc
-inline -simplifycfg -loop-rotate -licm -loop-index-split -indvars
1.      Running pass 'Function Pass Manager' on module
'bugpoint-reduced-blocks.bc'.
2.      Running pass 'Module Verifier' on function '@fallbackSort'
Aborted


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 24 05:23:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 24 May 2009 05:23:12 -0500
Subject: [LLVMbugs] [Bug 4257] llvm-gcc bootstrap failure in r72358
In-Reply-To: <bug-4257-206@http.llvm.org/bugs/>
Message-ID: <200905241023.n4OANClc006866@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4257


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-24 05:23:05 ---
Fixed in r72363.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 24 09:24:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 24 May 2009 09:24:12 -0500
Subject: [LLVMbugs] [Bug 4258] -indvars: use after free and crash
In-Reply-To: <bug-4258-206@http.llvm.org/bugs/>
Message-ID: <200905241424.n4OEOCXI016176@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4258


T??r??k Edwin <edwintorok at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from T??r??k Edwin <edwintorok at gmail.com>  2009-05-24 09:24:09 ---
Bug was that the SCEV Rewriter was still holding references to a Value* that
was deleted by RecursivelyDeleteTriviallyDeadInstructions.

Fixed in SVN r72364.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 24 23:10:46 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 24 May 2009 23:10:46 -0500
Subject: [LLVMbugs] [Bug 4261] New: doesn't eliminate complex dead load and
	store case
Message-ID: <bug-4261-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4261

           Summary: doesn't eliminate complex dead load and store case
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Interprocedural Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


With 'opt -std-compile-opts' LLVM doesn't simplify this:

@_ZGVZ3foovE1x = internal global i64 0, align 8

define void @_Z3foov() nounwind {
entry:
        %0 = load i8* bitcast (i64* @_ZGVZ3foovE1x to i8*), align 8
        %1 = icmp eq i8 %0, 0
        br i1 %1, label %bb1, label %return

bb1:            ; preds = %entry
        store i8 1, i8* bitcast (i64* @_ZGVZ3foovE1x to i8*), align 8
        ret void

return:         ; preds = %entry
        ret void
}

The thing is that because @_ZGVZ3foovE1x is internal, we know that those are
the only two uses of the global. There are two possible values (without the
bitcasts it converts the global into i1) and neither case much matters. The
whole thing is dead.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 25 01:31:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 25 May 2009 01:31:08 -0500
Subject: [LLVMbugs] [Bug 4212] bugpoint doesn't reduce global variables
In-Reply-To: <bug-4212-206@http.llvm.org/bugs/>
Message-ID: <200905250631.n4P6V8nB020367@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4212


Nick Lewycky <nicholas at mxc.ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Nick Lewycky <nicholas at mxc.ca>  2009-05-25 01:31:07 ---
Fixed in r72378:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090525/077989.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 25 11:01:14 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 25 May 2009 11:01:14 -0500
Subject: [LLVMbugs] [Bug 4253] available_externally causes wrong PIC code
	generation
In-Reply-To: <bug-4253-206@http.llvm.org/bugs/>
Message-ID: <200905251601.n4PG1EMu021180@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4253


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Chris Lattner <clattner at apple.com>  2009-05-25 11:01:13 ---
Looks great to me, thanks!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 25 11:45:54 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 25 May 2009 11:45:54 -0500
Subject: [LLVMbugs] [Bug 4262] New: llvm-gcc should generate
 available_externally function bodies etc
Message-ID: <bug-4262-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4262

           Summary: llvm-gcc should generate available_externally function
                    bodies etc
           Product: tools
           Version: 2.2
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: clattner at apple.com
                CC: llvmbugs at cs.uiuc.edu


llvm-gcc currently doesn't inline this example at -O3 because it doesn't emit
the body of test as an available_externally symbol:

extern __inline__ int test() { return 4; }
int bar() { return test(); }

$ llvm-gcc t.c -std=c89 -O3 -o - -emit-llvm -S
define i32 @bar() nounwind {
entry:
        %0 = tail call i32 (...)* @test() nounwind              ; <i32>
[#uses=1]
        ret i32 %0
}

declare i32 @test(...)


Clang handles this right by emitting @test with a body and available_externally
linkage.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 25 12:47:43 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 25 May 2009 12:47:43 -0500
Subject: [LLVMbugs] [Bug 4263] New: clang: struct layout tests failed
Message-ID: <bug-4263-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4263

           Summary: clang: struct layout tests failed
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: edwintorok at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3032)
 --> (http://llvm.org/bugs/attachment.cgi?id=3032)
makefile hack

I changed the makefiles of clang a bit to make utils/ABITest work on Linux.

utils/ABITest/layout fails though.

../ABITestGen.py --max-args 0 --test-layout --no-bit-field -o test.0.c --min=0 
--count=1000
gcc -std=gnu99 -m32 -o test.0.ref test.0.c
test.0.c: In function ???main???:
test.0.c:8509: warning: implicit declaration of function ???atoi???
./test.0.ref > test.0.ref.out
llvm-gcc -std=gnu99 -m32 -o test.0.x test.0.c
test.0.c: In function ???main???:
test.0.c:8509: warning: implicit declaration of function ???atoi???
./test.0.x > test.0.x.out
diff test.0.ref.out test.0.x.out > test.0.x.diff
make: [test.0.x.diff] Error 1 (ignored)
clang -std=gnu99 -m32 -o test.0.y test.0.c
./test.0.y > test.0.y.out
diff test.0.ref.out test.0.y.out > test.0.y.diff
make: [test.0.y.diff] Error 1 (ignored)
TEST 0: test.0.x.diff failed
TEST 0: test.0.y.diff failed

Attached are the diffs


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 25 16:06:16 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 25 May 2009 16:06:16 -0500
Subject: [LLVMbugs] [Bug 4266] New: llvm-gcc segfaults on void (*f)(enum b);
 if b not declared
Message-ID: <bug-4266-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4266

           Summary: llvm-gcc segfaults on void (*f)(enum b); if b not
                    declared
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sliedes at cc.hut.fi
                CC: llvmbugs at cs.uiuc.edu


1. Create a file test.c with the line

  void (*f)(enum b);

2. Run llvm-gcc -c test.c

3. llvm-gcc crashes:

---------------------------------------------------------
$ llvm-gcc -c test.c
test.c:1: warning: 'enum b' declared inside parameter list
test.c:1: warning: its scope is only this definition or declaration, which is
probably not what you want
test.c:1: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://llvm.org/bugs/> for instructions.
---------------------------------------------------------

I have llvm and llvm-gcc svn head from today. Architecture is i386.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 25 16:30:08 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 25 May 2009 16:30:08 -0500
Subject: [LLVMbugs] [Bug 4189] redundant load from constant not eliminated
In-Reply-To: <bug-4189-206@http.llvm.org/bugs/>
Message-ID: <200905252130.n4PLU8LR002328@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4189


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #13 from Chris Lattner <clattner at apple.com>  2009-05-25 16:30:07 ---
Ah, that is just because it figures out how to completely unroll the loop. 
With this patch, GVN now handles the load/store/load disambiguation case and
-std-compile-opts does remove the redundant load:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090525/078014.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Mon May 25 17:32:25 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 25 May 2009 17:32:25 -0500
Subject: [LLVMbugs] [Bug 4267] New: BasicAliasAnalysis misanalyzes
	over-indexed GEP
Message-ID: <bug-4267-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4267

           Summary: BasicAliasAnalysis misanalyzes over-indexed GEP
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Keywords: miscompilation
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: gohman at apple.com
                CC: llvmbugs at cs.uiuc.edu


Running

  opt -aa-eval -print-all-alias-modref-info

on the attached testcase prints out this:

  NoAlias:      double* %p.0.i.0, double* %p3

This is incorrect because in the second iteration of the loop,
%p.0.i.0 is equal to %p3.

Background: The new indvars GEP reconstruction logic is now emitting
code that does this, and this is resulting in some testcases being
miscompiled. With extra care, indvars could probably avoid
emitting code that triggers this bug in most cases, however it's
a bug regardless.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 26 02:56:49 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 26 May 2009 02:56:49 -0500
Subject: [LLVMbugs] [Bug 4140] clang -print-file-name not consistent with gcc
In-Reply-To: <bug-4140-206@http.llvm.org/bugs/>
Message-ID: <200905260756.n4Q7un3d032668@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4140


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-26 02:56:48 ---
Fix committed in r72412.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 26 11:07:59 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 26 May 2009 11:07:59 -0500
Subject: [LLVMbugs] [Bug 4270] New: getelementptr is underspecified
Message-ID: <bug-4270-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4270

           Summary: getelementptr is underspecified
           Product: Documentation
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: General docs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: gohman at apple.com
                CC: llvmbugs at cs.uiuc.edu


This paragraph from LangRef.html:

Note that it is undefined to access an array out of bounds: array and pointer
indexes must always be within the defined bounds of the array type when
accessed with an instruction that dereferences the pointer (e.g. a load or
store instruction). The one exception for this rule is zero length arrays.
These arrays are defined to be accessible as variable length arrays, which
requires access beyond the zero'th element.

raises several questions. I'm working on adding the concept of undefined
integer arithmetic overflow to LLVM, and also GEP expansion, so I'm filling
this bug in order to work towards clarification of the rules.

The first sentence seems to suggest that it's well defined to compute
arbitrary addresses, as long as they are not dereferenced. Especially since
there is no other mention of C's "one-past-the-end" provision, this
sentance seems to take that role by saying that in LLVM IR, addresses
N-past-the-end, or even N-ahead-of-the-beginning, may be computed, for
any N.

However, the second sentence makes a special provision for
zero-length array types. If N-past-the-end addresses are permitted, this
wouldn't really be an exception, but instead just an example of the
standard rule.

Also, there is also a rumor that GEP overflow is intended to be
undefined behavior. This isn't mentioned in LangRef.html, but it's been
heard spoken in a variety of places, and if it's true, it would seem to
rule out N-past-the-end. However in that case, there's nothing guaranteeing
one-past-the-end, which is needed for C support.

So first, assuming %A points to an array of [10 x double], which of the 
following instructions are intended to be undefined?
  %a = getelementptr double* %A, i64 -1
  %b = getelemnetptr double* %A, i64 9223372036854775807
  %c = getelementptr double* %A, i64 10


Second, is there anything undefined about this code?

  %p3 = getelementptr [3 x [3 x double]]* %p, i64 0, i64 0, i64 3
  store double 0.0, double* %p3

(assume %p3 points to sufficient storage)
The last index 3 is outside the bounds implied by the static type
implied by the base pointer and the gep, however the computed address
is within the bounds of the underlying allocated storage.

The following comment from BasicAliasAnalysis.cpp suggests that this
code is valid and that optimizers should handle it correctly:

  // We have to be careful here about array accesses.  In particular, consider:
  //        A[1][0] vs A[0][i]
  // In this case, we don't *know* that the array will be accessed in bounds:
  // the index could even be negative.  Because of this, we have to
  // conservatively *give up* and return may alias.  We disregard differing
  // array subscripts that are followed by a variable index without going
  // through a struct.


Third, if there are any cases where a getelementptr by itself (with no
load or store) is "undefined", is it Undefined Behavior, as in
"demons may fly out your nose", or is it merely that the getlementptr
may return an unspecified result?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 26 11:53:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 26 May 2009 11:53:34 -0500
Subject: [LLVMbugs] [Bug 4271] New: Assertion failed: (StartV->getType() ==
 PTy && " Pointer type mismatch for GEP!"), function visitAddRecExpr
Message-ID: <bug-4271-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4271

           Summary: Assertion failed: (StartV->getType() == PTy && "Pointer
                    type mismatch for GEP!"), function visitAddRecExpr
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ssen at apple.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=3036)
 --> (http://llvm.org/bugs/attachment.cgi?id=3036)
bugpoint generated testcase

I'm getting a crash on the attached bitcode file for i686-apple-darwin:

[ssen at virgon]$ /Volumes/HD/ltmp/ssen/llvm/build/Debug/bin/opt
bugpoint-reduced-simplified.bc -indvars
WARNING: You're attempting to print out a bitcode file.
This is inadvisable as it may cause display problems. If
you REALLY want to taste LLVM bitcode first-hand, you
can force output with the `-f' option.

Assertion failed: (StartV->getType() == PTy && "Pointer type mismatch for
GEP!"), function visitAddRecExpr, file
/Volumes/HD/ltmp/ssen/llvm/lib/Analysis/ScalarEvolutionExpander.cpp, line 452.
0   opt               0x003ad277 PrintStackTrace(void*) + 45
1   opt               0x003ad7e9 SignalHandler(int) + 351
2   libSystem.B.dylib 0x96872f8b _sigtramp + 43
3   libSystem.B.dylib 0xffffffff _sigtramp + 1769525407
4   libSystem.B.dylib 0x968dd6cf raise + 26
5   libSystem.B.dylib 0x968f4595 abort + 93
6   libSystem.B.dylib 0x968e0422 __assert_rtn + 252
7   opt               0x00291941
llvm::SCEVExpander::visitAddRecExpr(llvm::SCEVAddRecExpr const*) + 569
8   opt               0x002944e0 llvm::SCEVVisitor<llvm::SCEVExpander,
llvm::Value*>::visit(llvm::SCEV const*) + 304
9   opt               0x00290da3 llvm::SCEVExpander::expand(llvm::SCEV const*)
+ 143
10  opt               0x00290e1d
llvm::SCEVExpander::expandCodeFor(llvm::SCEVHandle, llvm::Type const*) + 41
11  opt               0x000d771c (anonymous
namespace)::IndVarSimplify::RewriteIVExpressions(llvm::LoopBase<llvm::BasicBlock>*,
llvm::Type const*, llvm::SCEVExpander&) + 2030
12  opt               0x000d8f19 (anonymous
namespace)::IndVarSimplify::runOnLoop(llvm::LoopBase<llvm::BasicBlock>*,
llvm::LPPassManager&) + 1265
13  opt               0x00248f60
llvm::LPPassManager::runOnFunction(llvm::Function&) + 952
14  opt               0x003401fe
llvm::FPPassManager::runOnFunction(llvm::Function&) + 288
15  opt               0x0034039c
llvm::FPPassManager::runOnModule(llvm::Module&) + 84
16  opt               0x0033eaf9
llvm::MPPassManager::runOnModule(llvm::Module&) + 235
17  opt               0x00340620 llvm::PassManagerImpl::run(llvm::Module&) +
112
18  opt               0x00340677 llvm::PassManager::run(llvm::Module&) + 27
19  opt               0x00056c00 main + 5036
20  opt               0x0004a8a9 start + 53
Stack dump:
0.      Running pass 'Function Pass Manager' on module
'bugpoint-reduced-simplified.bc'.
1.      Running pass 'Loop Pass Manager' on function
'@hibernate_kernel_entrypoint'
2.      Running pass 'Canonicalize Induction Variables' on basic block
'%for.body'
Abort trap


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 26 12:07:35 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 26 May 2009 12:07:35 -0500
Subject: [LLVMbugs] [Bug 4251] indvars: Instruction does not dominate all
	uses!
In-Reply-To: <bug-4251-206@http.llvm.org/bugs/>
Message-ID: <200905261707.n4QH7ZsU026142@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4251


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Dan Gohman <gohman at apple.com>  2009-05-26 12:07:34 ---
This was fixed by llvm r72369.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 26 12:46:20 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 26 May 2009 12:46:20 -0500
Subject: [LLVMbugs] [Bug 4271] Assertion failed: (StartV->getType() == PTy
 && " Pointer type mismatch for GEP!"), function visitAddRecExpr
In-Reply-To: <bug-4271-206@http.llvm.org/bugs/>
Message-ID: <200905261746.n4QHkK6W027965@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4271


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Dan Gohman <gohman at apple.com>  2009-05-26 12:46:19 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090525/078046.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 26 18:20:21 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 26 May 2009 18:20:21 -0500
Subject: [LLVMbugs] [Bug 4242] [x86-64 ABI] struct passed partially in
 registers if there' s an sret parameter
In-Reply-To: <bug-4242-206@http.llvm.org/bugs/>
Message-ID: <200905262320.n4QNKLMe009600@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4242


Dale Johannesen <dalej at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Dale Johannesen <dalej at apple.com>  2009-05-26 18:20:20 ---
Fixed here in llvm-gcc.
http://llvm.org/viewvc/llvm-project?view=rev&revision=72442


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 26 20:37:51 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 26 May 2009 20:37:51 -0500
Subject: [LLVMbugs] [Bug 2110] Assertion `0 && "Cannot expand insert element
 operand"' failed.
In-Reply-To: <bug-2110-206@http.llvm.org/bugs/>
Message-ID: <200905270137.n4R1bpEL014399@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2110


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #7 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-26 20:37:49 ---
Yeah, looks like this is working now.  Probably fixed by the introduction of
type legalization.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Tue May 26 21:03:38 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 26 May 2009 21:03:38 -0500
Subject: [LLVMbugs] [Bug 4267] BasicAliasAnalysis misanalyzes over-indexed
	GEP
In-Reply-To: <bug-4267-206@http.llvm.org/bugs/>
Message-ID: <200905270203.n4R23cL3015290@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4267


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Depends on|4270                        |
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Dan Gohman <gohman at apple.com>  2009-05-26 21:03:38 ---
After studying this a bit, I believe it really is a BasicAliasAnalysis bug,
since BasicAliasAnalysis already has special code to check for this
situation, but it only happens to cover non-constant indices.

I've now fixed it to handle constant indices as well, here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090525/078066.html

I've also fixed SCEVExpander to be more careful about constant
indices, however note that this doesn't avoid the problem in all
cases.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 01:32:56 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 01:32:56 -0500
Subject: [LLVMbugs] [Bug 4273] New: Top-down list schedulers do not support
 physreg dependencies
Message-ID: <bug-4273-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4273

           Summary: Top-down list schedulers do not support physreg
                    dependencies
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: asl at math.spbu.ru
                CC: llvmbugs at cs.uiuc.edu


Currently when scheduling for latency (and thus - via topdown list schedulers)
I need to always flag comparison to conditional branch since it's impossible to
reuse the status register due to lack of physreg dependencies.

This is quite unfortunate and yields ugly code :(

I think one can easily reproduce the problem with x86 backend codegen'ing
something like select_cc / br_cc and changing default (pre-RA) scheduler to
list-td


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 01:34:49 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 01:34:49 -0500
Subject: [LLVMbugs] [Bug 4273] Top-down list schedulers do not support
 physreg dependencies
In-Reply-To: <bug-4273-206@http.llvm.org/bugs/>
Message-ID: <200905270634.n4R6Yn4s024684@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4273


Anton Korobeynikov <asl at math.spbu.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #1 from Anton Korobeynikov <asl at math.spbu.ru>  2009-05-27 01:34:49 ---


*** This bug has been marked as a duplicate of bug 2344 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 03:16:10 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 03:16:10 -0500
Subject: [LLVMbugs] [Bug 4275] New: Wrong code generated by JIT with
 --enable-pic on Linux x86
Message-ID: <bug-4275-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4275

           Summary: Wrong code generated by JIT with --enable-pic on Linux
                    x86
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: Dr.Graef at t-online.de
                CC: llvmbugs at cs.uiuc.edu


If LLVM is compiled with --enable-pic (as is the default with current svn
sources), the JIT creates an improper call sequence via PLT. This only affects
Linux x86 systems, x86-64 works fine.

The symptom is a segfault in X86CompilationCallback2 at plt(), see, e.g.,
http://code.google.com/p/pure-lang/issues/detail?id=9 for a detailed bug report
related to this issue.

Remark by Anton Korobeynikov on the llvmdev mailing list:

<quote>
The problem is pretty easy: nobody cared about loading GOT pointer to
ebx before doing a call via PLT. The initial patch for this was x86-64
only (and I explicitly mentioned that it won't work on x86-32 due to
this issue). Unfortunately, it was commited as-is and later when PIC
was enabled by default this subtle bug appeared...

The proper solution will be to conditionalize for x86-32 linux and
assemble a proper call sequence via PLT.
</quote>

Anton suggested that I should file a bug report, so there.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 12:45:03 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 12:45:03 -0500
Subject: [LLVMbugs] [Bug 4276] New: llvm-config --ldflags lists too many libs
Message-ID: <bug-4276-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4276

           Summary: llvm-config --ldflags lists too many libs
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fvbommel at wxs.nl
                CC: llvmbugs at cs.uiuc.edu


llvm-config --ldflags lists all libs used *anywhere* in LLVM, regardless of
whether the user is linking to them.
For instance: this causes a static compiler to link to libffi, which is only
used by the JIT AFAIK[1].

Relevant link:
http://lists.cs.uiuc.edu/pipermail/llvmdev/2009-February/020155.html


[1]: And which is not installed on all systems I might add, so this is a
problem for distributing binaries of said static compiler.

I'm working around this by manually editing llvm-config...


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 13:47:26 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 13:47:26 -0500
Subject: [LLVMbugs] [Bug 3087] "Value does not have a node in the points-to
	graph"
In-Reply-To: <bug-3087-206@http.llvm.org/bugs/>
Message-ID: <200905271847.n4RIlQd9000678@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3087


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Chris Lattner <clattner at apple.com>  2009-05-27 13:47:25 ---
Yes, likely a missing call to deleteValue was added.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 13:48:19 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 13:48:19 -0500
Subject: [LLVMbugs] [Bug 3428] inline asm crash
In-Reply-To: <bug-3428-206@http.llvm.org/bugs/>
Message-ID: <200905271848.n4RImJ7x000720@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3428


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Chris Lattner <clattner at apple.com>  2009-05-27 13:48:18 ---
Verified, thanks!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 15:11:44 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 15:11:44 -0500
Subject: [LLVMbugs] [Bug 4222] LSR assertion failure
In-Reply-To: <bug-4222-206@http.llvm.org/bugs/>
Message-ID: <200905272011.n4RKBijc004826@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4222


Dan Gohman <gohman at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Dan Gohman <gohman at apple.com>  2009-05-27 15:11:44 ---
This is fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090525/078097.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 18:13:10 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 18:13:10 -0500
Subject: [LLVMbugs] [Bug 4277] New: llvm-gcc builds libstdc++ with symbols
	from libm
Message-ID: <bug-4277-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4277

           Summary: llvm-gcc builds libstdc++ with symbols from libm
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: rafael.espindola at gmail.com, llvmbugs at cs.uiuc.edu


On x86-64 linux, the built libstdc++ defines logf and does so badly:

~/llvm-gcc/install/lib64$ objdump -dr libstdc++.so
[...]
00000000000541e0 <logf at plt>:
   541e0:       ff 25 f2 f4 27 00       jmpq   *0x27f4f2(%rip)        # 2d36d8
<_GLOBAL_OFFSET_TABLE_+0x16f0>
   541e6:       68 db 02 00 00          pushq  $0x2db
   541eb:       e9 30 d2 ff ff          jmpq   51420 <_init+0x18>
[...]
000000000009bb40 <logf>:
   9bb40:       48 83 ec 08             sub    $0x8,%rsp
   9bb44:       e8 97 86 fb ff          callq  541e0 <logf at plt>
   9bb49:       48 83 c4 08             add    $0x8,%rsp
   9bb4d:       c3                      retq   
   9bb4e:       66 90                   xchg   %ax,%ax

This causes any program that calls logf() and linked against the libstdc++
built by llvm-gcc to enter an infinite loop of calling itself.

All the other math symbols are there too (hypot, sinf, powl etc), which aren't
in my system libstdc++, and they also seem to have recursive definitions.

Here's the configure flags I passed to llvm-gcc configure to cause this:
--enable-checking --enable-llvm=/home/nlewycky/llvm --program-prefix=llvm-
--prefix=/home/nlewycky/llvm-gcc/install --target=x86_64-unknown-linux-gnu
--disable-nls --enable-threads=posix --enable-symvers=gnu --enable-__cxa_atexit
--enable-c99 --enable-long-long --with-gnu-as --with-gnu-ld --disable-multilib
--enable-shared=libgcc,libmudflap,libssp,libstdc++ --enable-languages=c,c++
--with-native-system-header-dir=/include --with-local-prefix=/
--enable-clocale=gnu


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Wed May 27 23:11:06 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 27 May 2009 23:11:06 -0500
Subject: [LLVMbugs] [Bug 4277] llvm-gcc builds libstdc++ with symbols from
	libm
In-Reply-To: <bug-4277-206@http.llvm.org/bugs/>
Message-ID: <200905280411.n4S4B6BX032300@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4277


Nick Lewycky <nicholas at mxc.ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #4 from Nick Lewycky <nicholas at mxc.ca>  2009-05-27 23:11:00 ---
Fixed in r72512.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 00:11:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 00:11:29 -0500
Subject: [LLVMbugs] [Bug 3886] Crash when building Xorg/Cairo's pixman
In-Reply-To: <bug-3886-206@http.llvm.org/bugs/>
Message-ID: <200905280511.n4S5BToS002266@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3886


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED




--- Comment #15 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-28 00:11:23 ---
Fixed by recent legalization work.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 02:46:47 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 02:46:47 -0500
Subject: [LLVMbugs] [Bug 1853] Codegen doesn't support ExtractElement with
	variable idx
In-Reply-To: <bug-1853-206@http.llvm.org/bugs/>
Message-ID: <200905280746.n4S7kljq014299@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=1853


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-28 02:46:47 ---
I think this has been working for a while now.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 03:19:00 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 03:19:00 -0500
Subject: [LLVMbugs] [Bug 4001] assertion failed in lshr <2 x i64> without
	sse2
In-Reply-To: <bug-4001-206@http.llvm.org/bugs/>
Message-ID: <200905280819.n4S8J0xF020540@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4001


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #7 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-28 03:18:55 ---
This is fixed with my recent legalization work.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 04:32:39 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 04:32:39 -0500
Subject: [LLVMbugs] [Bug 4278] New: X86-64 -tailcallopt calling convention
 specification seems out-of-sync with regular cc
Message-ID: <bug-4278-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4278

           Summary: X86-64 -tailcallopt calling convention specification
                    seems out-of-sync with regular cc
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fvbommel at wxs.nl
                CC: llvmbugs at cs.uiuc.edu


In CC_X86_64_TailCall (X86CallingConv.td), there are several differences from
CC_X86_64_C that are not explained (AFAICT) by the need to support tail calls.
* It doesn't support fp80 arguments (causing a "Formal argument #0 has
unhandled type f80" abort).
* v2f32 doesn't seem to be handled either.
* The code that puts MMX vectors in XMM registers on Darwin doesn't check
CCIfSubtarget<"hasSSE2()"> like it does for regular calls.

Also, the comment for the integer registers is a copy-paste-o; there are only 5
of them for tail calls, not 6.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 05:12:04 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 05:12:04 -0500
Subject: [LLVMbugs] [Bug 4279] New: Tail calls not used when possible
 without -tailcallopt
Message-ID: <bug-4279-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4279

           Summary: Tail calls not used when possible without -tailcallopt
           Product: new-bugs
           Version: unspecified
          Platform: PC
               URL: http://llvm.org/docs/CodeGenerator.html#tailcallopt
        OS/Version: Linux
            Status: NEW
          Keywords: code-quality
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: fvbommel at wxs.nl
                CC: llvmbugs at cs.uiuc.edu


Not all cases where tail calls are possible should require -tailcallopt.
For caller-pops calling conventions (like non-"-tailcallopt" x86-64 fastcc), it
should be possible to generate tailcalls if the arguments to the callee take up
no more stack space than the arguments to the caller.
On targets like x86-64 many functions pass parameters entirely in registers, so
any time one such function calls another in tail call position the optimization
might be applied.


Another related thing I was wondering about: why is CC_X86_64_TailCall using R9
for the function pointer case instead of R11? R11 isn't normally a parameter
register (as opposed to R9) and is caller-saved.
In fact, according to the System V ABI document for x86-64, R11 is available as
scratch register for PLT code, so it can't be used for much else anyway. (This
may only apply to Linux, I'm not sure whether this ABI is used on other
operating systems[1])

As far as I can tell, once bug 4278 is fixed[2] the only difference actually
needed between the regular calling convention and the one for -tailcallopt is
caller-pops/callee-pops.

(At that point, the fastcc requirement also seems rather arbitrary, especially
since ccc uses the exact same calling convention on x86-64 if -tailcallopt
isn't used)


[1]: Not that this should be a problem for fastcc anyway unless R11 is used for
something else I'm unaware of on any of the other operating systems that would
require it to be preserved.
[2]: Minus the comment mismatch if R11 is substituted for R9 and the latter is
re-instated as last integer parameter register, of course.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 11:59:47 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 11:59:47 -0500
Subject: [LLVMbugs] [Bug 4280] New: test/CodeGen/Generic/select.ll failure
 on linux-ppc32
Message-ID: <bug-4280-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4280

           Summary: test/CodeGen/Generic/select.ll failure on linux-ppc32
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: PowerPC
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu, sharparrow1 at yahoo.com


Eli, according to the buildbot your recent DAG changes are responsible for
this. Could you take a look?

$ llvm-as < test/CodeGen/Generic/select.ll | Debug/bin/llc -march=ppc32
[... snip some assembly ...]
        .global boolexpr
        .type   boolexpr, @function
        .align  2
boolexpr:
.Leh_func_begin7:
        srwi 4, 4, 31
        xori 4, 4, 1
        and 3, 3, 4
        blr 
        .size   boolexpr,.-boolexpr
.Leh_func_end7:
llc: LegalizeDAG.cpp:2923:
void<unnamed>::SelectionDAGLegalize::ExpandNode(llvm::SDNode*,
llvm::SmallVectorImpl<llvm::SDValue>&): Assertion `!Tmp2.getNode() && "Can't
legalize BR_CC with legal condition!"' failed.
#0  0x00007ffff6cee095 in raise () from /lib/libc.so.6
#1  0x00007ffff6cefaf0 in abort () from /lib/libc.so.6
#2  0x00007ffff6ce72df in __assert_fail () from /lib/libc.so.6
#3  0x0000000000e663ae in ExpandNode (this=0x7fffffffdc70, Node=0x18a77b8, 
    Results=@0x7fffffffda40) at LegalizeDAG.cpp:2923
#4  0x0000000000e5925d in LegalizeOp (this=0x7fffffffdc70, Op=
      {Node = 0x18a77b8, ResNo = 0}) at LegalizeDAG.cpp:874
#5  0x0000000000e664f2 in LegalizeDAG (this=0x7fffffffdc70)
    at LegalizeDAG.cpp:228
#6  0x0000000000e66602 in llvm::SelectionDAG::Legalize (this=0x18a73e0, 
    TypesNeedLegalizing=false, OptLevel=llvm::CodeGenOpt::Default)
    at LegalizeDAG.cpp:3087
#7  0x0000000000e2e757 in llvm::SelectionDAGISel::CodeGenAndEmitDAG (
    this=0x18a65c0) at SelectionDAGISel.cpp:652
#8  0x0000000000e30c93 in llvm::SelectionDAGISel::SelectBasicBlock (
    this=0x18a65c0, LLVMBB=0x18972b0, Begin=
      {<bidirectional_iterator<llvm::Instruction,long int>> =
{<std::iterator<std::bidirectional_iterator_tag,llvm::Instruction,long
int,llvm::Instruction*,llvm::Instruction&>> = {<No data fields>}, <No data
fields>}, NodePtr = 0x1897440}, End=
      {<bidirectional_iterator<llvm::Instruction,long int>> =
{<std::iterator<std::bidirectional_iterator_tag,llvm::Instruction,long
int,llvm::Instruction*,llvm::Instruction&>> = {<No data fields>}, <No data
fields>}, NodePtr = 0x18972b0})
    at SelectionDAGISel.cpp:496
#9  0x0000000000e316e3 in llvm::SelectionDAGISel::SelectAllBasicBlocks (
    this=0x18a65c0, Fn=@0x1893cd0, MF=@0x18d3cd0, MMI=0x18bea10, DW=0x18abc30, 
    TII=@0x189bc10) at SelectionDAGISel.cpp:886
#10 0x0000000000e32493 in llvm::SelectionDAGISel::runOnFunction (
    this=0x18a65c0, Fn=@0x1893cd0) at SelectionDAGISel.cpp:326
#11 0x0000000000b40b3f in runOnFunction (this=0x18a65c0, Fn=@0x1893cd0)
    at PPCISelDAGToDAG.cpp:60
#12 0x0000000001135cd4 in llvm::FPPassManager::runOnFunction (this=0x189f4b0, 
    F=@0x1893cd0) at PassManager.cpp:1343
#13 0x0000000001136249 in llvm::FunctionPassManagerImpl::run (this=0x188fd60, 
    F=@0x1893cd0) at PassManager.cpp:1296
#14 0x00000000011363f0 in llvm::FunctionPassManager::run (this=0x7fffffffe300, 
    F=@0x1893cd0) at PassManager.cpp:1238
#15 0x000000000084541c in main (argc=4, argv=0x7fffffffe528) at llc.cpp:340
(gdb) up 3
#3  0x0000000000e663ae in ExpandNode (this=0x7fffffffdc70, Node=0x18a77b8, 
    Results=@0x7fffffffda40) at LegalizeDAG.cpp:2923
2923        assert(!Tmp2.getNode() && "Can't legalize BR_CC with legal
condition!");
(gdb) p Tmp2
$1 = {Node = 0x18a8070, ResNo = 0}
(gdb) p Tmp2.getNode()
$2 = (class llvm::SDNode *) 0x18a8070
(gdb) p Tmp2.getNode()->dump()
0x18a8070: i32 = and 0x18a8260, 0x18a7b98$3 = void


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 16:05:10 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 16:05:10 -0500
Subject: [LLVMbugs] [Bug 4280] test/CodeGen/Generic/select.ll failure on
	linux-ppc32
In-Reply-To: <bug-4280-206@http.llvm.org/bugs/>
Message-ID: <200905282105.n4SL5AAU017929@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4280


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-28 16:05:09 ---
Fixed in r72536.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 16:09:11 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 16:09:11 -0500
Subject: [LLVMbugs] [Bug 4281] New: Making typedefs of const stuff makes
	Clang crash
Message-ID: <bug-4281-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4281

           Summary: Making typedefs of const stuff makes Clang crash
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


The following C code makes Clang die:

typedef struct {
        int i;
} something;

typedef const something const_something;

something fail(void);

int
main(int argc, char *argv[])
{
        const_something R = fail();
}

cc -O2 -pipe   llvm-crash.c  -o llvm-crash
Assertion failed: (CGF.getContext().typesAreCompatible(
E->getSubExpr()->getType().getUnqualifiedType(),
E->getType().getUnqualifiedType()) && "Implicit cast types must be
compatible"), function VisitImplicitCastExpr, file
llvm/tools/clang/lib/CodeGen/CGExprAgg.cpp, line 181.
Stack dump:
0.      Program arguments: /usr/bin/../libexec/clang-cc -triple
x86_64-undermydesk-freebsd8.0 -S -disable-free -main-file-name llvm-crash.c
--relocation-model static --disable-fp-elim --unwind-tables=1 --mcpu=x86-64
--fmath-errno=1 -O2 -fdiagnostics-show-option -o /tmp/cc-k2ODg6.s -x c
llvm-crash.c 
1.      <eof> parser at end of file
2.      llvm-crash.c:10:1: LLVM IR generation of declaration 'main'
3.      llvm-crash.c:11:1: LLVM IR generation of compound statement ('{}')


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Thu May 28 18:32:12 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 28 May 2009 18:32:12 -0500
Subject: [LLVMbugs] [Bug 4281] Making typedefs of const stuff makes Clang
	crash
In-Reply-To: <bug-4281-206@http.llvm.org/bugs/>
Message-ID: <200905282332.n4SNWCLQ023193@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4281


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-28 18:32:12 ---
Fixed in r72552.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 29 01:28:57 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 29 May 2009 01:28:57 -0500
Subject: [LLVMbugs] [Bug 4282] New: Crash when returning empty structure and
 using specific compiler flags
Message-ID: <bug-4282-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4282

           Summary: Crash when returning empty structure and using specific
                    compiler flags
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


struct empty_struct {
};

struct empty_struct *
crash(void)
{
}

The above code causes LLVM to crash when being called like this:

clang-cc -triple x86_64-undermydesk-freebsd8.0 -S -disable-free -main-file-name
ppp_l2tp_avp.c --relocation-model static --disable-fp-elim --unwind-tables=1
--mcpu=x86-64 --fmath-errno=1 -g -DTYPED_MEM_UNDEFINE_ORIGINALS
-I/usr/ports/devel/libpdel/work/libpdel-0.5.3 -I/usr/local/include -O2 -Wall
-Wcast-align -Wchar-subscripts -Wcomment -Wformat -Wimplicit
-Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wno-long-long
-Wparentheses -Wpointer-arith -Wreturn-type -Wswitch -Wtrigraphs
-Wuninitialized -Wunused -Wwrite-strings -std=gnu99 -fdiagnostics-show-option
-o /tmp/cc-keDoNh.s -x c llvm-crash.c

Assertion failed: (Begin + idx < End), function operator[], file
llvm/include/llvm/ADT/SmallVector.h, line 125.
Stack dump:
0.      Program arguments: clang-cc -triple x86_64-undermydesk-freebsd8.0 -S
-disable-free -main-file-name ppp_l2tp_avp.c --relocation-model static
--disable-fp-elim --unwind-tables=1 --mcpu=x86-64 --fmath-errno=1 -g
-DTYPED_MEM_UNDEFINE_ORIGINALS -I/usr/ports/devel/libpdel/work/libpdel-0.5.3
-I/usr/local/include -O2 -Wall -Wcast-align -Wchar-subscripts -Wcomment
-Wformat -Wimplicit -Wmissing-declarations -Wmissing-prototypes
-Wnested-externs -Wno-long-long -Wparentheses -Wpointer-arith -Wreturn-type
-Wswitch -Wtrigraphs -Wuninitialized -Wunused -Wwrite-strings -std=gnu99
-fdiagnostics-show-option -o /tmp/cc-keDoNh.s -x c llvm-crash.c 
1.      <eof> parser at end of file
2.      llvm-crash.c:5:1: LLVM IR generation of declaration 'crash'


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 29 06:55:28 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 29 May 2009 06:55:28 -0500
Subject: [LLVMbugs] [Bug 4282] Crash when returning empty structure when
	using -g
In-Reply-To: <bug-4282-206@http.llvm.org/bugs/>
Message-ID: <200905291155.n4TBtSNl027226@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4282


Ed Schouten <ed at 80386.nl> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |WORKSFORME




--- Comment #2 from Ed Schouten <ed at 80386.nl>  2009-05-29 06:55:22 ---
Closing ticket. After updating LLVM to a newer version, I can no longer
reproduce this.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 29 07:52:58 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 29 May 2009 07:52:58 -0500
Subject: [LLVMbugs] [Bug 4283] New: Clang's preprocessor doesn' t support
 multibyte character constants
Message-ID: <bug-4283-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4283

           Summary: Clang's preprocessor doesn't support multibyte character
                    constants
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: preprocessor
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


Believe it or not, GCC accepts this piece of code:

#include <stdio.h>

#if (('1234' >> 24) == '1')
# define ENDIAN "little"
#elif (('4321' >> 24) == '1')
# define ENDIAN "big"
#else
# define ENDIAN "unknown"
#endif

int
main(int argc, char *argv[])
{

        puts("I am " ENDIAN " endian!");
}

When built with GCC, it prints "I am little endian!" on my AMD64 box, while a
Clang built binary prints "I am unknown endian!"

This construct is used by http://www.speakfreely.org/


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Fri May 29 14:45:21 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 29 May 2009 14:45:21 -0500
Subject: [LLVMbugs] [Bug 4275] Wrong code generated by JIT with --enable-pic
 on Linux x86
In-Reply-To: <bug-4275-206@http.llvm.org/bugs/>
Message-ID: <200905291945.n4TJjLxp012577@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4275


Nick Lewycky <nlewycky at google.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nlewycky at google.com
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #2 from Nick Lewycky <nlewycky at google.com>  2009-05-29 14:45:21 ---


*** This bug has been marked as a duplicate of bug 3239 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 30 03:00:07 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 30 May 2009 03:00:07 -0500
Subject: [LLVMbugs] [Bug 2268]
 void<unnamed>::FPS::handleSpecialFP(llvm::ilist_iterator<llvm::
 MachineInstr>&): Assertion `StackTop == 1 && " Stack should have one
 element on it to return!"' failed
In-Reply-To: <bug-2268-206@http.llvm.org/bugs/>
Message-ID: <200905300800.n4U8074C017660@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2268


T??r??k Edwin <edwintorok at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from T??r??k Edwin <edwintorok at gmail.com>  2009-05-30 03:00:06 ---
(In reply to comment #2)
> I can not reproduce this with r72485. Fixed ?
> 

I can't reproduce it either, closing.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 30 08:58:00 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 30 May 2009 08:58:00 -0500
Subject: [LLVMbugs] [Bug 4262] llvm-gcc should generate available_externally
 function bodies etc
In-Reply-To: <bug-4262-206@http.llvm.org/bugs/>
Message-ID: <200905301358.n4UDw022031445@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4262


Duncan Sands <baldrick at free.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #5 from Duncan Sands <baldrick at free.fr>  2009-05-30 08:57:56 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090525/078186.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 30 12:09:37 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 30 May 2009 12:09:37 -0500
Subject: [LLVMbugs] [Bug 4284] New: memcmp() optimizer is not 8-bit clean
Message-ID: <bug-4284-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4284

           Summary: memcmp() optimizer is not 8-bit clean
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Transformation Utilities
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


The following code works when built with Clang without optimization, but fails
when optimization is enabled. The code is obtained from OpenLDAP's configure
script.

#include <stdio.h>
#include <string.h>

int
main(int argc, char *argv[])
{
        char c0 = '\100', c1 = '\200', c2 = '\201';

        if (memcmp(&c0, &c2, 1) >= 0 || memcmp(&c1, &c2, 1) >= 0) {
                puts("Test failed");
                return (1);
        }

        puts("Test passed");
        return (0);
}

; ModuleID = 'memcmp.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-f80:128:128"
target triple = "x86_64-undermydesk-freebsd8.0"
@.str = internal constant [12 x i8] c"Test failed\00"           ; <[12 x i8]*>
[#uses=1]
@.str1 = internal constant [12 x i8] c"Test passed\00"          ; <[12 x i8]*>
[#uses=1]

define i32 @main(i32 %argc, i8** %argv) nounwind {
entry:
        %retval = alloca i32            ; <i32*> [#uses=3]
        %argc.addr = alloca i32         ; <i32*> [#uses=1]
        %argv.addr = alloca i8**                ; <i8***> [#uses=1]
        %c0 = alloca i8, align 1                ; <i8*> [#uses=2]
        %c1 = alloca i8, align 1                ; <i8*> [#uses=2]
        %c2 = alloca i8, align 1                ; <i8*> [#uses=3]
        store i32 %argc, i32* %argc.addr
        store i8** %argv, i8*** %argv.addr
        store i8 64, i8* %c0
        store i8 -128, i8* %c1
        store i8 -127, i8* %c2
        %call = call i32 @memcmp(i8* %c0, i8* %c2, i64 1) readonly             
; <i32> [#uses=1]
        %cmp = icmp sge i32 %call, 0            ; <i1> [#uses=1]
        br i1 %cmp, label %if.then, label %lor.lhs.false

lor.lhs.false:          ; preds = %entry
        %call1 = call i32 @memcmp(i8* %c1, i8* %c2, i64 1) readonly            
; <i32> [#uses=1]
        %cmp2 = icmp sge i32 %call1, 0          ; <i1> [#uses=1]
        br i1 %cmp2, label %if.then, label %if.end

if.then:                ; preds = %lor.lhs.false, %entry
        %call3 = call i32 @puts(i8* getelementptr ([12 x i8]* @.str, i32 0, i32
0))             ; <i32> [#uses=0]
        store i32 1, i32* %retval
        br label %return

if.end:         ; preds = %lor.lhs.false
        %call4 = call i32 @puts(i8* getelementptr ([12 x i8]* @.str1, i32 0,
i32 0))            ; <i32> [#uses=0]
        store i32 0, i32* %retval
        br label %return

return:         ; preds = %if.end, %if.then
        %0 = load i32* %retval          ; <i32> [#uses=1]
        ret i32 %0
}

declare i32 @memcmp(i8*, i8*, i64) readonly

declare i32 @puts(i8*)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sat May 30 13:50:11 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 30 May 2009 13:50:11 -0500
Subject: [LLVMbugs] [Bug 4284] memcmp() optimizer is not 8-bit clean
In-Reply-To: <bug-4284-206@http.llvm.org/bugs/>
Message-ID: <200905301850.n4UIoBrZ009785@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4284


Chris Lattner <clattner at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Chris Lattner <clattner at apple.com>  2009-05-30 13:50:11 ---
Nice catch, the patch looks great applied here with a testcase:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20090525/078192.html


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 31 01:02:21 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 31 May 2009 01:02:21 -0500
Subject: [LLVMbugs] [Bug 4285] New: Wide character initialisation doesn't
 work with const.
Message-ID: <bug-4285-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4285

           Summary: Wide character initialisation doesn't work with const.
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: parser
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


The following code doesn't build with Clang:

#include <stdio.h>
#include <wchar.h>

static const wchar_t foo[] = L"Awesome!";

int
main(int argc, char *argv[])
{

        printf("%ls\n", foo);
        return (0);
}

test.c:4:30: error: initialization with '{...}' expected for array
static const wchar_t foo[] = L"Awesome!";
                             ^~~~~~~~~~~
1 diagnostic generated.

When the const keyword is removed, it builds properly.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 31 05:55:16 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 31 May 2009 05:55:16 -0500
Subject: [LLVMbugs] [Bug 4285] Wide character initialisation doesn't work
	with const.
In-Reply-To: <bug-4285-206@http.llvm.org/bugs/>
Message-ID: <200905311055.n4VAtGTg015247@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=4285


Eli Friedman <sharparrow1 at yahoo.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sharparrow1 at yahoo.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2009-05-31 05:55:15 ---
Fixed in r72663.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 31 09:04:45 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 31 May 2009 09:04:45 -0500
Subject: [LLVMbugs] [Bug 4286] New: Assertion failed: (use_empty() && " Uses
 remain when a value is destroyed!"), function ~Value
Message-ID: <bug-4286-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4286

           Summary: Assertion failed: (use_empty() && "Uses remain when a
                    value is destroyed!"), function ~Value
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


Created an attachment (id=3044)
 --> (http://llvm.org/bugs/attachment.cgi?id=3044)
Reduced testcase

When building the attached C code with Clang, I get the following console
output:

$ cc -O2 -pipe   /home/ed/llvm-crash.c  -o llvm-crash
While deleting: i1* %storetmp.i
Use still stuck around after Def is destroyed:  store i1 true, i1* %storetmp.i

Assertion failed: (use_empty() && "Uses remain when a value is destroyed!"),
function ~Value, file .../llvm/lib/VMCore/Value.cpp, line 70.
Stack dump:
0.      Program arguments: /usr/bin/../libexec/clang-cc -triple
x86_64-undermydesk-freebsd8.0 -S -disable-free -main-file-name llvm-crash.c
--relocation-model static --disable-fp-elim --unwind-tables=1 --mcpu=x86-64
--fmath-errno=1 -O2 -fdiagnostics-show-option -o /tmp/cc-eU01Mt.s -x c
/home/ed/llvm-crash.c 
1.      <eof> parser at end of file
2.      Per-module optimization passes
3.      Running pass 'CallGraph Pass Manager' on module
'/home/ed/llvm-crash.c'.
4.      Running pass 'Scalar Replacement of Aggregates' on function
'@pnm_encode'

The testcase is based on source code from the high profile FreeBSD port
`jasper' (a JPEG decoder): http://www.ece.uvic.ca/~mdadams/jasper/


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 31 09:53:34 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 31 May 2009 09:53:34 -0500
Subject: [LLVMbugs] [Bug 3443] recent regression: undeclared function isn' t
 infered the right type
In-Reply-To: <bug-3443-206@http.llvm.org/bugs/>
Message-ID: <200905311453.n4VErYWa023697@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=3443


Nuno Lopes <nunoplopes at sapo.pt> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|INVALID                     |




--- Comment #3 from Nuno Lopes <nunoplopes at sapo.pt>  2009-05-31 09:53:33 ---
Sorry for this late reopening. I still have this problem.
Currently I cannot compile PHP due to this bug. Can we downgrade the error to a
warning?

gcc says the following:
/cvs/php5/Zend/zend_language_parser.c: In function 'yytnamerr':
/cvs/php5/Zend/zend_language_parser.c:2915: warning: incompatible implicit
declaration of built-in function 'stpcpy'


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 31 16:45:41 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 31 May 2009 16:45:41 -0500
Subject: [LLVMbugs] [Bug 4287] New: Varargs + K&R doesn't work nicely
Message-ID: <bug-4287-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4287

           Summary: Varargs + K&R doesn't work nicely
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: FreeBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: ed at 80386.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 3696


The following code doesn't build with Clang, while it does build when using
GCC. This causes some build failures when building packages for FreeBSD:

#include <stdarg.h>
char *foo = "test";
#if defined(__STDC__) && __STDC__
int test(char*,...);
#endif

int test(fmt)
        char*fmt;
{
        va_list ap;
        char*a;
        int x;

        va_start(ap,fmt);
        a=va_arg(ap,char*);
        x=(a!=foo);
        va_end(ap);
        return x;
}

int main(argc,argv)
        int argc;char**argv;
{

        exit(test("",foo));
}

Error log:

conftest.c:7:5: error: conflicting types for 'test'
int test(fmt)
    ^
conftest.c:4:5: note: previous declaration is here
int test(char*,...);
    ^
conftest.c:14:2: error: 'va_start' used in function with fixed args
        va_start(ap,fmt);
        ^
/usr/include/stdarg.h:45:2: note: instantiated from:
        __builtin_va_start((ap), (last))
        ^
3 diagnostics generated.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

From bugzilla-daemon at cs.uiuc.edu  Sun May 31 16:47:29 2009
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 31 May 2009 16:47:29 -0500
Subject: [LLVMbugs] [Bug 4288] New: Mixed up # line directives with -E -dD
Message-ID: <bug-4288-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=4288

           Summary: Mixed up # line directives with -E -dD
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: preprocessor
        AssignedTo: unassignedclangbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Testcase:
#define A A
/* 1
 * 2
 * 3
 * 4
 * 5
 * 6
 * 7
 * 8
 */
#define B B

Run through "clang -x c - -o - -E -dD", produces something like the following
(ignoring the predefine buffer):

# 1 "<stdin>" 2
#define A A# 11 "<stdin>"
#define B B


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.