From bugzilla-daemon at cs.uiuc.edu  Sun Jun  1 01:43:42 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 01:43:42 -0500
Subject: [LLVMbugs] [Bug 2398] New: callgraph wrong in presence of weak
	functions
Message-ID: <bug-2398-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2398

           Summary: callgraph wrong in presence of weak functions
           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: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


The definition of a weak function can be replaced with
a different definition (this is what "weak linkage" means).
So any deductions made by examining the instructions making
up a weak function definition may be wrong.  So such functions
must not be inlined etc.  This also impacts calculation of the
callgraph: a new definition may call different functions to the
current definition.

The obvious thing to do is to have functions that call
the weak function get an external node in their callgraph
as well as the weak function.  I think the weak function is
needed in the callgraph as well as the external node
because the weak function may call functions with internal
linkage while an external definition cannot, so if there
was only an external node then optimizers might wrongly
think that a call to the weak function cannot access internal
functions.

There is also the question of what the callgraph for the
weak function itself should look like.  I think this should
be the callgraph of the current definition plus the external
node for the same reason.

Note that any new definition of the weak function will not be
able to access internal globals.


-- 
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 Jun  1 11:57:59 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 11:57:59 -0500
Subject: [LLVMbugs] [Bug 2399] New: cbe generates illegal code for sret of
	array
Message-ID: <bug-2399-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2399

           Summary: cbe generates illegal code for sret of 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: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Testcase:

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 = "i686-pc-linux-gnu"
        %union.a = type [4 x i8]
define void @a(%union.a* sret  %agg.result, ...) {
        ret void
}

gcc output from generated code:
<stdin>:109: error: ???a??? declared as function returning an array
<stdin>:110: warning: conflicting types for built-in function ???malloc???
<stdin>:131: error: ???a??? declared as function returning an array
<stdin>: In function ???a???:
<stdin>:134: warning: return makes integer from pointer without a cast
<stdin>:134: warning: function returns address of local variable

llc doesn't seem to have any issues with this construct in general.

(This is generated by clang; I could fix clang not to do this, but I don't see
anything particularly wrong with 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  Sun Jun  1 12:33:04 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 12:33:04 -0500
Subject: [LLVMbugs] [Bug 2400] New: crash with sizeof(incomplete_type)
Message-ID: <bug-2400-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2400

           Summary: crash with sizeof(incomplete_type)
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1693)
 --> (http://llvm.org/bugs/attachment.cgi?id=1693)
test case (delta reduced)

The attached code triggers an assertion in clang (in
ASTContexr::getTypeInfo()), because of a sizeof(incomplete_type).


-- 
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 Jun  1 13:04:18 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 13:04:18 -0500
Subject: [LLVMbugs] [Bug 2401] New: memcpyopt incorrectly removes memcpy
Message-ID: <bug-2401-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2401

           Summary: memcpyopt incorrectly removes memcpy
           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=1694)
 --> (http://llvm.org/bugs/attachment.cgi?id=1694)
Testcase

Testcase attached; exits quietly when run with "lli memtest.bc", but calls
abort() when run with "opt memtest.bc -memcpyopt -o - | lli".

Sorry the testcase is a little big, but I didn't want to break it reducing it,
and bugpoint didn't seem too helpful.  It should be pretty easy to see what's
going on, though, because all the memcpyopt pass does here is remove one
memcpy.


-- 
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 Jun  1 13:32:55 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 13:32:55 -0500
Subject: [LLVMbugs] [Bug 2357] declaration error, from self included file
In-Reply-To: <bug-2357-206@http.llvm.org/bugs/>
Message-ID: <200806011832.m51IWtrs028032@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2357


Nuno Lopes <nunoplopes at sapo.pt> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nunoplopes at sapo.pt
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #3 from Nuno Lopes <nunoplopes at sapo.pt>  2008-06-01 13:32:55 ---
fixed in r51843. The problem was with the multiple #include optimization that
clang performs when files are protected with standard #ifndef/#define/#endif.
In this case, clang was considering the file to have just one big #ifndef
macro, which is not the case.


-- 
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 Jun  1 14:55:04 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 14:55:04 -0500
Subject: [LLVMbugs] [Bug 2402] New: CBE generates wrong code for packed
 struct with long double
Message-ID: <bug-2402-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2402

           Summary: CBE generates wrong code for packed struct with long
                    double
           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


Testcase:

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 = "i686-pc-linux-gnu"
        %struct.anon = type <{ x86_fp80, i8, i8 }>
@e = common global [100 x %struct.anon] zeroinitializer, align 4               
; <[100 x %struct.anon]*> [#uses=3]

define float @a(...) {
entry:
        %tmp = load x86_fp80* getelementptr ([100 x %struct.anon]* @e, i32 0,
i32 50, i32 0)            ; <x86_fp80> [#uses=1]
        %conv = fptrunc x86_fp80 %tmp to float          ; <float> [#uses=1]
        ret float %conv
}

Generated C code with "llc -march=c" (relevant bit):
/* Structure contents */
struct l_struct_2E_anon {
  long double field0;
  unsigned char field1;
  unsigned char field2;
} __attribute__ ((packed));

The LLVM struct is 12 bytes, while the converted one is 14 bytes on x86-Linux,
and 18 bytes on x86-64.

No idea how to fix 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  Sun Jun  1 16:04:23 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 16:04:23 -0500
Subject: [LLVMbugs] [Bug 2397] gvn reads freed memory
In-Reply-To: <bug-2397-206@http.llvm.org/bugs/>
Message-ID: <200806012104.m51L4NEv032242@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2397


Owen Anderson <resistor at mac.com> changed:

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




--- Comment #2 from Owen Anderson <resistor at mac.com>  2008-06-01 16:04:22 ---
Fixed in r51846.  Turned out to be a problem in memdep.


-- 
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 Jun  1 16:52:35 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 16:52:35 -0500
Subject: [LLVMbugs] [Bug 2401] memcpyopt incorrectly removes memcpy
In-Reply-To: <bug-2401-206@http.llvm.org/bugs/>
Message-ID: <200806012152.m51LqZUG001061@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2401


Owen Anderson <resistor at mac.com> changed:

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




--- Comment #1 from Owen Anderson <resistor at mac.com>  2008-06-01 16:52:35 ---
Fixed in r51848.


-- 
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 Jun  1 18:04:01 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 1 Jun 2008 18:04:01 -0500
Subject: [LLVMbugs] [Bug 2403] New: crash on invalid typedef declarations
Message-ID: <bug-2403-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2403

           Summary: crash on invalid typedef declarations
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Keywords: crash-on-invalid
          Severity: enhancement
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


The following code crashes clang (in MergeTypeDefDecl(); although it doesn't
make sense merging the decls..):


typedef struct {}
typedef __builtin_va_list __gnuc_va_list;


-- 
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 Jun  2 04:44:05 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 2 Jun 2008 04:44:05 -0500
Subject: [LLVMbugs] [Bug 2404] New: llvm-gcc-4.2 bootstrap failure:
 triggered assertion in eh_terminate.cc
Message-ID: <bug-2404-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2404

           Summary: llvm-gcc-4.2 bootstrap failure: triggered assertion in
                    eh_terminate.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: kenneth.hoste at elis.ugent.be
                CC: llvmbugs at cs.uiuc.edu


I'm seeing the following ICE while trying to bootstrap llvm-gcc on x86
(32-bit):

/home/kehoste/LLVM/llvm-gcc-4.2_obj/./gcc/xgcc -shared-libgcc
-B/home/kehoste/LLVM/llvm-gcc-4.2_obj/./gcc -nostdinc++
-L/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/src
-L/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/src/.libs
-B/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/bin/
-B/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/lib/ -isystem
/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/include -isystem
/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/sys-include
-I/home/kehoste/LLVM/llvm-gcc-4.2_src/libstdc++-v3/../gcc
-I/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/include/i686-pc-linux-gnu
-I/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/include
-I/home/kehoste/LLVM/llvm-gcc-4.2_src/libstdc++-v3/libsupc++
-fno-implicit-templates -Wall -Wextra -Wwrite-strings -Wcast-qual
-fdiagnostics-show-location=once -ffunction-sections -fdata-sections -g -O2
-D_GNU_SOURCE -c
../../../../llvm-gcc-4.2_src/libstdc++-v3/libsupc++/eh_term_handler.cc  -fPIC
-DPIC -o eh_term_handler.o
/bin/sh ../libtool --tag CXX --tag disable-shared --mode=compile
/home/kehoste/LLVM/llvm-gcc-4.2_obj/./gcc/xgcc -shared-libgcc
-B/home/kehoste/LLVM/llvm-gcc-4.2_obj/./gcc -nostdinc++
-L/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/src
-L/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/src/.libs
-B/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/bin/
-B/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/lib/ -isystem
/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/include -isystem
/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/sys-include
-I/home/kehoste/LLVM/llvm-gcc-4.2_src/libstdc++-v3/../gcc
-I/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/include/i686-pc-linux-gnu
-I/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/include
-I/home/kehoste/LLVM/llvm-gcc-4.2_src/libstdc++-v3/libsupc++ 
-fno-implicit-templates -prefer-pic -Wall -Wextra -Wwrite-strings -Wcast-qual 
-fdiagnostics-show-location=once  -ffunction-sections -fdata-sections  -g -O2 
-D_GNU_SOURCE  -c -o eh_terminate.lo
../../../../llvm-gcc-4.2_src/libstdc++-v3/libsupc++/eh_terminate.cc
/home/kehoste/LLVM/llvm-gcc-4.2_obj/./gcc/xgcc -shared-libgcc
-B/home/kehoste/LLVM/llvm-gcc-4.2_obj/./gcc -nostdinc++
-L/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/src
-L/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/src/.libs
-B/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/bin/
-B/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/lib/ -isystem
/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/include -isystem
/user/kehoste/LLVM/llvm-gcc-4.2/i686-pc-linux-gnu/sys-include
-I/home/kehoste/LLVM/llvm-gcc-4.2_src/libstdc++-v3/../gcc
-I/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/include/i686-pc-linux-gnu
-I/home/kehoste/LLVM/llvm-gcc-4.2_obj/i686-pc-linux-gnu/libstdc++-v3/include
-I/home/kehoste/LLVM/llvm-gcc-4.2_src/libstdc++-v3/libsupc++
-fno-implicit-templates -Wall -Wextra -Wwrite-strings -Wcast-qual
-fdiagnostics-show-location=once -ffunction-sections -fdata-sections -g -O2
-D_GNU_SOURCE -c
../../../../llvm-gcc-4.2_src/libstdc++-v3/libsupc++/eh_terminate.cc  -fPIC
-DPIC -o eh_terminate.o
cc1plus: /home/kehoste/LLVM/llvm/lib/CodeGen/SelectionDAG/ScheduleDAG.cpp:644:
void llvm::ScheduleDAG::AddOperand(llvm::MachineInstr*, llvm::SDOperand,
unsigned int, const llvm::TargetInstrDesc*, llvm::DenseMap<llvm::SDOperand,
unsigned int, llvm::DenseMapInfo<llvm::SDOperand>, llvm::DenseMapInfo<unsigned
int> >&): Assertion `Op.getValueType() != MVT::Other && Op.getValueType() !=
MVT::Flag && "Chain and flag operands should occur at end of operand list!"'
failed.
../../../../llvm-gcc-4.2_src/libstdc++-v3/libsupc++/eh_terminate.cc:83:
internal compiler error: Aborted

I configured/build LLVM and llvm-gcc as follows:

=> llvm:

../llvm/configure --prefix=/home/kehoste/LLVM/llvm-bin

make ENABLE_OPTIMIZED=1

=> llvm-gcc:

../llvm-gcc-4.2_src/configure --prefix=/user/kehoste/LLVM/llvm-gcc-4.2
--program-prefix=llvm- --enable-llvm=/user/kehoste/LLVM/llvm-obj
--enable-languages=c,c++,fortran

make

This is using revision 51865.


-- 
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 Jun  2 12:27:27 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 2 Jun 2008 12:27:27 -0500
Subject: [LLVMbugs] [Bug 2395] crash in BasicAA with gep+bitcasts
In-Reply-To: <bug-2395-206@http.llvm.org/bugs/>
Message-ID: <200806021727.m52HRRMr011226@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2395


Wojciech Matyjewicz <wmatyjewicz at fastmail.fm> changed:

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




--- Comment #4 from Wojciech Matyjewicz <wmatyjewicz at fastmail.fm>  2008-06-02 12:27:26 ---
Applied here:
http://llvm.org/viewvc/llvm-project?view=rev&revision=51875


-- 
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 Jun  2 16:53:32 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 2 Jun 2008 16:53:32 -0500
Subject: [LLVMbugs] [Bug 2405] New: Test results should print
 TARGET_LLCFLAGS value, and others
Message-ID: <bug-2405-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2405

           Summary: Test results should print TARGET_LLCFLAGS value, and
                    others
           Product: Test Suite
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Nightly Tester
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: gohman at apple.com
                CC: llvmbugs at cs.uiuc.edu


We currently have several nightly testers with "-PIC" in their names, and
looking at the current reports it's not clear exactly what specific flags
they're using to do that. Also, many of the llvm-test tests do not work
on x86-64 without the addition of several flags which the nightly
testers currently use.

It would be useful for the nightly tester reports to include the values
of the environment variables TARGET_LLCFLAGS, EXTRA_LLI_OPTS,
DISABLE_LTO, and possibly others. This would make it easier for
people to reproduce their results.


-- 
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 Jun  2 17:04:49 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 2 Jun 2008 17:04:49 -0500
Subject: [LLVMbugs] [Bug 2406] New: redefine type: segfault
Message-ID: <bug-2406-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2406

           Summary: redefine type: segfault
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: csaba.hruska at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1700)
 --> (http://llvm.org/bugs/attachment.cgi?id=1700)
original buggy source file

system: ubuntu linux  32 bit X86 hardy
clang: revision 51883

source code (bug1.c):
#include <stdlib.h>
typedef long long int64_t ;

problem:
ccc -c bug1.c
bug1.c
clang -emit-llvm-bc -x c -o bug1.o bug1.c
clang[0x837d0b8]

but:
ccc -E bug1.c > b.c
ccc -c b.c
and this throws the correct error message.


-- 
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 Jun  2 17:53:50 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 2 Jun 2008 17:53:50 -0500
Subject: [LLVMbugs] [Bug 2372] loop-unswitch crash
In-Reply-To: <bug-2372-206@http.llvm.org/bugs/>
Message-ID: <200806022253.m52MroAl003314@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2372


Devang Patel <dpatel at apple.com> changed:

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




--- Comment #2 from Devang Patel <dpatel at apple.com>  2008-06-02 17:53:48 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080602/063250.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 Jun  3 01:57:40 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 3 Jun 2008 01:57:40 -0500
Subject: [LLVMbugs] [Bug 2407] New: C backend emits invalid inline asm
Message-ID: <bug-2407-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2407

           Summary: C backend emits invalid inline asm
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: C
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


In llvm-test/MultiSource/Applications/Minisat/Output

$ llc -march=c minisat.llvm.bc -o bugpoint-test-program.bc.cbe.c
$ /usr/bin/gcc -x c -fno-strict-aliasing bugpoint-test-program.bc.cbe.c -x none
-o bugpoint-test-program.bc.cbe.c.gcc.exe -lstdc++ -lm -lm -O2 -Wl,-R.
bugpoint-test-program.bc.cbe.c:393: warning: conflicting types for built-in
function ???malloc???
bugpoint-test-program.bc.cbe.c: In function ???main???:
bugpoint-test-program.bc.cbe.c:843: warning: return type of ???main??? is not
???int???
bugpoint-test-program.bc.cbe.c:2617: error: memory input 0 is not directly
addressable

The code in question is this:

  __asm__ volatile ("fldcw %0"
        :
        :"m"((&llvm_cbe_newcw)));

The original llvm asm instruction is:

        call void asm sideeffect "fldcw $0", "*m,~{dirflag},~{fpsr},~{flags}"(
i16* %newcw ) nounwind 

Any ideas?


-- 
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 Jun  3 04:12:12 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 3 Jun 2008 04:12:12 -0500
Subject: [LLVMbugs] [Bug 2404] llvm-gcc-4.2 bootstrap failure: triggered
 assertion in eh_terminate.cc
In-Reply-To: <bug-2404-206@http.llvm.org/bugs/>
Message-ID: <200806030912.m539CCTQ003953@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2404


Kenneth Hoste <kenneth.hoste at elis.ugent.be> changed:

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




--- Comment #7 from Kenneth Hoste <kenneth.hoste at elis.ugent.be>  2008-06-03 04:12:11 ---
(In reply to comment #6)
> Revision 51877 fixes a bug involving the same assertion
> failure as the one reported here. Can you check to see if this
> commit fixes the problem? 

Problem seems to be fixed with revision 51898 on my end, so I'm closing this
bug... 


-- 
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 Jun  3 07:53:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 3 Jun 2008 07:53:19 -0500
Subject: [LLVMbugs] [Bug 2408] New: CBE generates code that violates C
 signed integer overflow rules
Message-ID: <bug-2408-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2408

           Summary: CBE generates code that violates C signed integer
                    overflow rules
           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


Testcase:
define i32 @a(i32 %a) {
entry:
        %shr = ashr i32 %a, 0           ; <i32> [#uses=1]
        %shr2 = ashr i32 2, 0           ; <i32> [#uses=1]
        %mul = mul i32 %shr, %shr2              ; <i32> [#uses=1]
        %shr4 = ashr i32 2, 0           ; <i32> [#uses=1]
        %div = sdiv i32 %mul, %shr4             ; <i32> [#uses=1]
        ret i32 %div
}

Result of "llc -march=c | ~/bin/gcc-4.3 -x c - -o - -S -O3 -fno-strict-aliasing
-fomit-frame-pointer":
a:
        movl    4(%esp), %eax
        ret

The semantics of this are not the same as those of the original code; for
example, take a(INT_MIN).

A bit contrived, but I'm reasonably sure this could affect real code.

To avoid this, the C backend should make sure to always do
addition/subtraction/multiplication in unsigned types.

Alternatively, we could suggest that -fwrapv should always be used with
CBE-generated code, but that seems unnecessary.  (If we go with this option,
anything that invokes gcc with CBE-generated code needs to be 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  Tue Jun  3 16:38:53 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 3 Jun 2008 16:38:53 -0500
Subject: [LLVMbugs] [Bug 2409] New: Fail to compile on SunOs5.8 with
	gcc-4.1.2
Message-ID: <bug-2409-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2409

           Summary: Fail to compile on SunOs5.8 with gcc-4.1.2
           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: pathscale at 126.com
                CC: llvmbugs at cs.uiuc.edu


/project/agassiz05/cslp/llvm/llvm_2.2/src_2.2/lib/Support/APInt.cpp:1324:
error: '::round' has not been declared

It looks there is no function called round() on Solaris.


-- 
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 Jun  4 01:43:17 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 01:43:17 -0500
Subject: [LLVMbugs] [Bug 2410] New: incorrect weak linkage
Message-ID: <bug-2410-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2410

           Summary: incorrect weak linkage
           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: yingyi.liang at gmail.com
                CC: llvmbugs at cs.uiuc.edu


File1.bc:
define weak i32 @foo() nounwind  {
entry:
        ret i32 1
}

File2.bc:
define i32 @foo() nounwind  {
entry:
        ret i32 2
}

The final version of foo depends on the linkage order of File1.bc and File2.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 Jun  4 02:29:33 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 02:29:33 -0500
Subject: [LLVMbugs] [Bug 2411] New: Interprocedural constant propagation
 ignores weak attribute
Message-ID: <bug-2411-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2411

           Summary: Interprocedural constant propagation ignores weak
                    attribute
           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: yingyi.liang at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1705)
 --> (http://llvm.org/bugs/attachment.cgi?id=1705)
Test case

Interprocedural constant propagation (-ipconstprop) propagates the result of
foo to main, regardless the weak attribute of foo:

define weak i32 @foo() nounwind  {
entry:
        ret i32 1
}

define i32 @main() nounwind  {
entry:
        %tmp1 = call i32 @foo( ) nounwind
        ret i32 %tmp1
}

test.ll as attached is the test case and test-ipconstprop.ll is the result
after applying "opt -ipconstprop".


-- 
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 Jun  4 03:27:49 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 03:27:49 -0500
Subject: [LLVMbugs] [Bug 2402] CBE generates wrong code for packed struct
 with long double
In-Reply-To: <bug-2402-206@http.llvm.org/bugs/>
Message-ID: <200806040827.m548RnXh023956@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2402


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #6 from Duncan Sands <baldrick at free.fr>  2008-06-04 03:27:49 ---
Fixed: LLVM packed structs now use the ABI size for the field size,
like gcc:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080602/063306.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080602/063307.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 Jun  4 05:09:46 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 05:09:46 -0500
Subject: [LLVMbugs] [Bug 2412] New: GCC warning is a clang error
Message-ID: <bug-2412-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2412

           Summary: GCC warning is a clang error
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: isanbard at gmail.com
                CC: llvmbugs at cs.uiuc.edu


For this code:

$ cat t.c
void bar() {
  char ng;

  welcome(ng);
}

void welcome(new_game)
     char new_game;
{
  foo(new_game);
}

clang produces this:

$ clang -emit-llvm-bc -x c -o t.o t.c
t.c:7:6: error: conflicting types for 'welcome'
void welcome(new_game)
     ^
t.c:4:3: error: previous implicit declaration is here
  welcome(ng);
  ^
2 diagnostics generated.

While GCC produces this:

$ gcc -o t t.c -c
t.c:8: warning: conflicting types for ???welcome???
t.c:4: warning: previous implicit declaration of ???welcome??? was here


-- 
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 Jun  4 08:57:52 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 08:57:52 -0500
Subject: [LLVMbugs] [Bug 2413] New: void* address cast error
Message-ID: <bug-2413-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2413

           Summary: void* address cast error
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: csaba.hruska at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1707)
 --> (http://llvm.org/bugs/attachment.cgi?id=1707)
buggy example file

buggy code:

void f()
{
        void *addr;
        addr = (void *)( ((long int)addr + 7L) );
}

error:
ccc -c smemory.c
smemory.c
clang -emit-llvm-bc -x c -o smemory.o smemory.c
clang: /media/disk/download/llvm/svn/llvm/include/llvm/ADT/APSInt.h:159:
llvm::APSInt& llvm::APSInt::operator+=(const llvm::APSInt&): Assertion
`IsUnsigned == RHS.IsUnsigned && "Signedness mismatch!"' failed.
clang[0x837e778]
/lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7d38a01]
/lib/tls/i686/cmov/libc.so.6(__assert_fail+0xee)[0xb7d3010e]
clang(_ZNK5clang4Expr21isIntegerConstantExprERN4llvm6APSIntERNS_10ASTContextEPNS_14SourceLocationEb+0x18d3)[0x822bfa3]

with gcc it compiles fine.

system: 32 bit ubuntu (hardy) linux x86
clang svn: At revision 51936.


-- 
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 neil at daikokuya.co.uk  Wed Jun  4 09:25:09 2008
From: neil at daikokuya.co.uk (Neil Booth)
Date: Wed, 4 Jun 2008 23:25:09 +0900
Subject: [LLVMbugs] [Bug 2412] New: GCC warning is a clang error
In-Reply-To: <bug-2412-206@http.llvm.org/bugs/>
References: <bug-2412-206@http.llvm.org/bugs/>
Message-ID: <20080604142509.GB22630@daikokuya.co.uk>

bugzilla-daemon at cs.uiuc.edu wrote:-

> For this code:
> 
> $ cat t.c
> void bar() {
>   char ng;
> 
>   welcome(ng);
> }
> 
> void welcome(new_game)
>      char new_game;
> {
>   foo(new_game);
> }
> 
> clang produces this:
> 
> $ clang -emit-llvm-bc -x c -o t.o t.c
> t.c:7:6: error: conflicting types for 'welcome'
> void welcome(new_game)
>      ^
> t.c:4:3: error: previous implicit declaration is here
>   welcome(ng);
>   ^
> 2 diagnostics generated.
> 
> While GCC produces this:
> 
> $ gcc -o t t.c -c
> t.c:8: warning: conflicting types for ???welcome???
> t.c:4: warning: previous implicit declaration of ???welcome??? was here

This is undefined behaviour in C90 and a constraint violation in C99.
It is a good thing that compilers other than GCC reject this kind of
shoddy code.  It doesn't take much to fix the code, whatever it was
actually intended to mean.

I believe more modern compilers should strive to far higher goals than
"accept anything invalid just because GCC does", except perhaps in a
well-documented GCC compatibility mode.  Even GCC doesn't strive to
achieve that w.r.t. prior versions; every release is stricter in
various areas than the prior one even for C.

Neil.

From bugzilla-daemon at cs.uiuc.edu  Wed Jun  4 12:23:26 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 12:23:26 -0500
Subject: [LLVMbugs] [Bug 2414] New: typedefed funtion type check assert
Message-ID: <bug-2414-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2414

           Summary: typedefed funtion type check assert
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: csaba.hruska at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1708)
 --> (http://llvm.org/bugs/attachment.cgi?id=1708)
buggy code

buggy code:

typedef void filter_func_t();
filter_func_t mono_filter;

void addfilter2(filter_func_t *func){}

void setup_filters()
{
        addfilter2( mono_filter);
}

error:
ccc -c player.c
player.c
clang -emit-llvm-bc -x c -o player.o player.c
clang: /media/disk/download/llvm/svn/llvm/include/llvm/Support/Casting.h:199:
typename llvm::cast_retty<To, From>::ret_type llvm::cast(const Y&) [with X =
clang::FunctionType, Y = clang::QualType]: Assertion `isa<X>(Val) &&
"cast<Ty>() argument of incompatible type!"' failed.
clang[0x837f828]
/lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7d10a01]
/lib/tls/i686/cmov/libc.so.6(__assert_fail+0xee)[0xb7d0810e]
clang[0x81184b5]

gcc compiles it.

system: 32 bit x86 ubuntu hardy
clang and llvm version: At revision 51948.

Cheers,
Csaba Hruska


-- 
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 Jun  4 12:59:16 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 12:59:16 -0500
Subject: [LLVMbugs] [Bug 2414] typedefed funtion type check assert
In-Reply-To: <bug-2414-206@http.llvm.org/bugs/>
Message-ID: <200806041759.m54HxG9D010612@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2414


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>  2008-06-04 12:59:15 ---
Fixed in r51951.


-- 
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 Jun  4 13:04:40 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 13:04:40 -0500
Subject: [LLVMbugs] [Bug 2407] C backend emits invalid inline asm
In-Reply-To: <bug-2407-206@http.llvm.org/bugs/>
Message-ID: <200806041804.m54I4eKc010875@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2407


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #5 from Chris Lattner <sabre at nondot.org>  2008-06-04 13:04:39 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080602/063330.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 Jun  4 14:29:56 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 4 Jun 2008 14:29:56 -0500
Subject: [LLVMbugs] [Bug 2415] New: passing function pointers as argument
	crashes clang
Message-ID: <bug-2415-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2415

           Summary: passing function pointers as argument crashes clang
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


passing function pointers (with 1 or more arguments) as an argument to a
function crashes clang.


-- 
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 isanbard at gmail.com  Wed Jun  4 15:20:14 2008
From: isanbard at gmail.com (Bill Wendling)
Date: Wed, 4 Jun 2008 13:20:14 -0700
Subject: [LLVMbugs] [Bug 2412] New: GCC warning is a clang error
In-Reply-To: <20080604142509.GB22630@daikokuya.co.uk>
References: <bug-2412-206@http.llvm.org/bugs/>
	<20080604142509.GB22630@daikokuya.co.uk>
Message-ID: <16e5fdf90806041320u608dc355pde2d5d4f4f733315@mail.gmail.com>

On Wed, Jun 4, 2008 at 7:25 AM, Neil Booth <neil at daikokuya.co.uk> wrote:
> bugzilla-daemon at cs.uiuc.edu wrote:-
>
>> For this code:
>>
>> $ cat t.c
>> void bar() {
>>   char ng;
>>
>>   welcome(ng);
>> }
>>
>> void welcome(new_game)
>>      char new_game;
>> {
>>   foo(new_game);
>> }
>>
> This is undefined behaviour in C90 and a constraint violation in C99.
> It is a good thing that compilers other than GCC reject this kind of
> shoddy code.  It doesn't take much to fix the code, whatever it was
> actually intended to mean.
>
> I believe more modern compilers should strive to far higher goals than
> "accept anything invalid just because GCC does", except perhaps in a
> well-documented GCC compatibility mode.  Even GCC doesn't strive to
> achieve that w.r.t. prior versions; every release is stricter in
> various areas than the prior one even for C.
>
ICC also emits a warning instead of an error.

-bw

From bugzilla-daemon at cs.uiuc.edu  Thu Jun  5 01:53:44 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 5 Jun 2008 01:53:44 -0500
Subject: [LLVMbugs] [Bug 686] llvmc enhancements needed
In-Reply-To: <bug-686-206@http.llvm.org/bugs/>
Message-ID: <200806050653.m556riIU000902@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=686


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

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




--- Comment #9 from Anton Korobeynikov <asl at math.spbu.ru>  2008-06-05 01:53:43 ---
The basic ideas listed here were really implemented and overall design looks
pretty promising.


-- 
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 Jun  5 02:22:43 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 5 Jun 2008 02:22:43 -0500
Subject: [LLVMbugs] [Bug 2415] passing function pointers as argument crashes
	clang
In-Reply-To: <bug-2415-206@http.llvm.org/bugs/>
Message-ID: <200806050722.m557Mhwp001899@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2415


Matthijs Kooijman <matthijs at stdin.nl> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |matthijs at stdin.nl
             Status|NEW                         |RESOLVED
         Resolution|                            |DUPLICATE




--- Comment #4 from Matthijs Kooijman <matthijs at stdin.nl>  2008-06-05 02:22:43 ---
I've just tried the attached test case, it crashed using yesterday's version
but works now. From the looks of it, this is indeed the same bug as #2414, the
assertion is the same as well. So, fixed!

*** This bug has been marked as a duplicate of bug 2414 ***


-- 
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 Jun  5 04:56:38 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 5 Jun 2008 04:56:38 -0500
Subject: [LLVMbugs] [Bug 2416] New: acceptance of invalid enum syntax
Message-ID: <bug-2416-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2416

           Summary: acceptance of invalid enum syntax
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: csaba.hruska at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1711)
 --> (http://llvm.org/bugs/attachment.cgi?id=1711)
buggy source file

buggy source code(enum.c):

enum en {};

error:
ccc -c enum.c
enum.c
clang -emit-llvm-bc -x c -o enum.o enum.c

proper:
gcc -c enum.c
enum.c:1: error: expected identifier before ???}??? token


clang svn: At revision 51993.


-- 
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 Jun  5 14:49:05 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 5 Jun 2008 14:49:05 -0500
Subject: [LLVMbugs] [Bug 2417] New: clang emits wrong bitcode with
 initialization of a global variable with a static function ptr
Message-ID: <bug-2417-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2417

           Summary: clang emits wrong bitcode with initialization of a
                    global variable with a static function ptr
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


clang emits wrong bitcode with initialization of a global variable with a
static function pointer.

the error is:
Global is external, but doesn't have external or dllimport or weak linkage!
i32 (...)* @zm_activate_zlib
invalid linkage type for function declaration
i32 (...)* @zm_activate_zlib
Broken module found, compilation aborted!

I don't know much about LLVM's bitcode, but maybe a static function must have
external linkage when it's referenced by a global symbol. This seems to be what
llvm-gcc does..


-- 
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 Jun  6 06:45:50 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 6 Jun 2008 06:45:50 -0500
Subject: [LLVMbugs] [Bug 2418] New: llc produces C file with duplicate
	symbols
Message-ID: <bug-2418-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2418

           Summary: llc produces C file with duplicate symbols
           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: ccris at doc.ic.ac.uk
                CC: llvmbugs at cs.uiuc.edu


This test is against repository version l52043.

To reproduce, do the following using the attached main.bc:
  >llc -march=c main.bc -o main.c
  >gcc -c main.c
main.c:381: error: redefinition of ???L_OBJC_METH_VAR_NAME_0???
main.c:379: error: previous definition of ???L_OBJC_METH_VAR_NAME_0??? was here

Original posting on llvmdev:
http://lists.cs.uiuc.edu/pipermail/llvmdev/2008-June/015118.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 Jun  6 11:10:15 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 6 Jun 2008 11:10:15 -0500
Subject: [LLVMbugs] [Bug 2418] CBE produces C file with duplicate symbols
In-Reply-To: <bug-2418-206@http.llvm.org/bugs/>
Message-ID: <200806061610.m56GAF5D032169@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2418


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

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




--- Comment #3 from Anton Korobeynikov <asl at math.spbu.ru>  2008-06-06 11:10:15 ---
Fixed in:

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080602/063418.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080602/063419.html

Please verify


-- 
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 Jun  6 11:55:53 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 6 Jun 2008 11:55:53 -0500
Subject: [LLVMbugs] [Bug 1956] Sema+Codegen fail to use '((struct
 foo*)0)->m' as initializer
In-Reply-To: <bug-1956-206@http.llvm.org/bugs/>
Message-ID: <200806061655.m56GtrBW001121@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=1956


Nuno Lopes <nunoplopes at sapo.pt> changed:

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




--- Comment #4 from Nuno Lopes <nunoplopes at sapo.pt>  2008-06-06 11:55:53 ---
Fixed in r52049 and several other patches by Eli.


-- 
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 Jun  6 12:14:57 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 6 Jun 2008 12:14:57 -0500
Subject: [LLVMbugs] [Bug 2419] New: empty union initialization crashes clang
Message-ID: <bug-2419-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2419

           Summary: empty union initialization crashes clang
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


The following crashes clang -emit-llvm:

struct Mem {
        union {
        } u;
};

struct Mem *columnMem(){
        static const struct Mem nullMem = { {} };
}


-- 
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 Jun  6 15:13:08 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 6 Jun 2008 15:13:08 -0500
Subject: [LLVMbugs] [Bug 2419] empty union initialization crashes clang
In-Reply-To: <bug-2419-206@http.llvm.org/bugs/>
Message-ID: <200806062013.m56KD8no011384@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2419


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>  2008-06-06 15:13:08 ---
Fixed in PR52060.


-- 
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 Jun  6 18:30:31 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 6 Jun 2008 18:30:31 -0500
Subject: [LLVMbugs] [Bug 2420] New: aggregate store : (Accum->isInteger() &&
 "Unknown FP type! ")
Message-ID: <bug-2420-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2420

           Summary: aggregate store :  (Accum->isInteger() && "Unknown FP
                    type!")
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: dpatel at apple.com
                CC: llvmbugs at cs.uiuc.edu


-- small.bc --
; ModuleID = 'small.bc'
        %union.S1387 = type [8 x i8]

define void @check1387va(i32 %z, ...) nounwind  {
entry:
        %arg2 = alloca [2 x i32], align 4               ; <[2 x i32]*>
[#uses=2]
        %arg2.sub = getelementptr [2 x i32]* %arg2, i32 0, i32 0               
; <i32*> [#uses=0]
        %tmpcast = bitcast [2 x i32]* %arg2 to %union.S1387*            ;
<%union.S1387*> [#uses=1]
        store %union.S1387 zeroinitializer, %union.S1387* %tmpcast, align 4
        ret void
}
---

$ opt -disable-output -scalarrepl small.bc
triggers 
Assertion failed: (Accum->isInteger() && "Unknown FP type!"), function
MergeInType, file ScalarReplAggregates.cpp, line 918.


-- 
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 Jun  8 07:39:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 8 Jun 2008 07:39:39 -0500
Subject: [LLVMbugs] [Bug 2420] aggregate store : (Accum->isInteger() &&
 "Unknown FP type!")
In-Reply-To: <bug-2420-206@http.llvm.org/bugs/>
Message-ID: <200806081239.m58CddmF022132@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2420


Wojciech Matyjewicz <wmatyjewicz at fastmail.fm> changed:

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




--- Comment #3 from Wojciech Matyjewicz <wmatyjewicz at fastmail.fm>  2008-06-08 07:39:38 ---
Right. I've omitted this update. Now it works for me too, so I think we may
close this bug.


-- 
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 Jun  8 15:47:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 8 Jun 2008 15:47:30 -0500
Subject: [LLVMbugs] [Bug 2422] New: segfault ice
Message-ID: <bug-2422-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2422

           Summary: segfault ice
           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


Bug is for svn 52094 on Ubuntu Feisty for x86.

[regehr at babel tmp17]$ llvm-gcc -Os -c small.c
small.c:29: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://developer.apple.com/bugreporter> for instructions.

[regehr at babel tmp17]$ cat small.c
typedef signed char int8_t;
typedef int int32_t;
typedef unsigned int uint32_t;
static unsigned long int
mod_rhs (const long int rhs)
{
  return rhs;
}

uint32_t g_15;
uint32_t g_38;
int32_t
func_18 (int32_t p_19)
{
  return g_15;
}

int32_t
func_29 (uint32_t p_30)
{
  int8_t l_31;
  for (1; 1; ++l_31)
    if (1 % mod_rhs (func_18 (1)))
      for (g_38 = 0; (g_38 <= 0); ++g_38)
        if (func_18 (1))
          g_38 = g_15;
        else
          return 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  Sun Jun  8 18:10:47 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 8 Jun 2008 18:10:47 -0500
Subject: [LLVMbugs] [Bug 406] [ia64] C++ programs compiled by LLVM are not
 ABI compliant with system compiler
In-Reply-To: <bug-406-206@http.llvm.org/bugs/>
Message-ID: <200806082310.m58NAlIH008932@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=406


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-06-08 18:10:47 ---
IA64 has numerous issues.  Until the backend is more complete, we don't need to
track details like 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  Mon Jun  9 02:48:32 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 9 Jun 2008 02:48:32 -0500
Subject: [LLVMbugs] [Bug 2410] incorrect weak linkage
In-Reply-To: <bug-2410-206@http.llvm.org/bugs/>
Message-ID: <200806090748.m597mWVr005333@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2410


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #4 from Chris Lattner <sabre at nondot.org>  2008-06-09 02:48:32 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080609/063500.html

Thanks for the nicely reduced testcase!


-- 
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 Jun  9 02:59:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 9 Jun 2008 02:59:30 -0500
Subject: [LLVMbugs] [Bug 2411] Interprocedural constant propagation ignores
 weak attribute
In-Reply-To: <bug-2411-206@http.llvm.org/bugs/>
Message-ID: <200806090759.m597xUZI005755@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2411


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #3 from Chris Lattner <sabre at nondot.org>  2008-06-09 02:59:30 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080609/063502.html

Nice testcase!


-- 
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 Jun  9 03:00:55 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 9 Jun 2008 03:00:55 -0500
Subject: [LLVMbugs] [Bug 2423] New: Miscompilation with scalarrepl
Message-ID: <bug-2423-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2423

           Summary: Miscompilation with scalarrepl
           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=1715)
 --> (http://llvm.org/bugs/attachment.cgi?id=1715)
Testcase

To reproduce: compare output of "lli bugpoint-passinput.bc" with "opt
-scalarrepl bugpoint-passinput.bc -o - | lli".  (The former silently returns,
the latter aborts.)

The testcase is a bit big, but I like to be conservative with miscompilation
bugs, since it's easy to break the testcase.

This testcase comes from clang compiling struct-layout-1.exp.  It's possible
this is clang's fault, but it seems unlikely.


-- 
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 Jun  9 03:28:44 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 9 Jun 2008 03:28:44 -0500
Subject: [LLVMbugs] [Bug 2337] SimplfyLibCalls should inline calls to abs
In-Reply-To: <bug-2337-206@http.llvm.org/bugs/>
Message-ID: <200806090828.m598Si61007227@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2337


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #4 from Chris Lattner <sabre at nondot.org>  2008-06-09 03:28:44 ---
Implemented, patch here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080609/063514.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 Jun  9 04:18:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 9 Jun 2008 04:18:28 -0500
Subject: [LLVMbugs] [Bug 2424] New: libprofile_rt.so installed correctly,
 but libprofile_rt.  a not
Message-ID: <bug-2424-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2424

           Summary: libprofile_rt.so installed correctly, but
                    libprofile_rt.a not
           Product: Runtime Libraries
           Version: trunk
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: libprofile library
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: hs4233 at mail.mn-solutions.de
                CC: llvmbugs at cs.uiuc.edu


I configure llvm with

mkdir obj.llvm
cd obj.llvm
../svn.llvm/configure \
  --prefix=`pwd`/dist \
  ... (other options)

Later I wanted to install the profile libraries. I did:

cd obj.llvm/runtime
make install

and the following happened:

*** Install runtime
cd /usr/src/llvm/obj.llvm/runtime; make install
make[1]: Entering directory `/usr/src/llvm/obj.llvm/runtime'
make[2]: Entering directory `/usr/src/llvm/obj.llvm/runtime/libprofile'
llvm[2]: Installing Release Shared Library
/usr/src/llvm/dist/lib/libprofile_rt.so
mklib: install: warning: remember to run `mklib --finish
/usr/src/llvm/obj.llvm/Release/lib'
PATH="$PATH:/sbin" ldconfig -n /usr/src/llvm/dist/lib
----------------------------------------------------------------------
Libraries have been installed in:
   /usr/src/llvm/dist/lib

If you ever happen to want to link against installed libraries
in a given directory, LIBDIR, you must either use libtool, and
specify the full pathname of the library, or use the `-LLIBDIR'
flag during linking and do at least one of the following:
   - add LIBDIR to the `LD_LIBRARY_PATH' environment variable
     during execution
   - add LIBDIR to the `LD_RUN_PATH' environment variable
     during linking
   - use the `-Wl,--rpath -Wl,LIBDIR' linker flag
   - have your system administrator add LIBDIR to `/etc/ld.so.conf'

See any operating system documentation about shared libraries for
more information, such as the ld(1) and ld.so(8) manual pages.
----------------------------------------------------------------------
llvm[2]: Installing Release Bytecode Archive /lib/libprofile_rt.a
/usr/bin/install: cannot create regular file `/lib/libprofile_rt.a': Permission
denied
make[2]: *** [/lib/libprofile_rt.a] Error 1
make[2]: Leaving directory `/usr/src/llvm/obj.llvm/runtime/libprofile'
make[1]: *** [libprofile/.makeinstall] Error 2
make[1]: Leaving directory `/usr/src/llvm/obj.llvm/runtime'
make: *** [instrt] Error 2


As you can see, the installation of libprofile_rt.a didn't honor the --prefix
parameter.


-- 
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 Jun  9 17:13:58 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 9 Jun 2008 17:13:58 -0500
Subject: [LLVMbugs] [Bug 2425] New: merge of function declarations doesn't
 ignore __declspec()
Message-ID: <bug-2425-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2425

           Summary: merge of function declarations doesn't ignore
                    __declspec()
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: filcab at gmail.com
                CC: llvmbugs at cs.uiuc.edu


clang doesn't compile the following code:
__declspec(dllexport) void f(void);

void f(void)
{
}



But with __attribute((dllexport)), it compiles.


-- 
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 Jun 10 08:50:31 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 10 Jun 2008 08:50:31 -0500
Subject: [LLVMbugs] [Bug 2426] New: llc -mcpu=help doesn't work
Message-ID: <bug-2426-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2426

           Summary: llc -mcpu=help doesn't work
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: minor
          Priority: P2
         Component: llc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


llc -help says

    -mcpu=<cpu-name>             - Target a specific cpu type (-mcpu=help for
details)

however, llc -mcpu=help or llc -march=x86 -mcpu=help gives no output and llc
simply hangs indefinitely (I guess it simply ignores the help option and just
waits for input to appear on stdin). Looking at the code, there does not seem
to
be any code that handles the -mcpu=help case. The same holds for -mattrs=help.


-- 
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 Jun 10 10:28:40 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 10 Jun 2008 10:28:40 -0500
Subject: [LLVMbugs] [Bug 2427] New: clang redefines size_t silently but then
	complains
Message-ID: <bug-2427-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2427

           Summary: clang redefines size_t silently but then complains
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: filcab at gmail.com
                CC: llvmbugs at cs.uiuc.edu


clang doesn't compile the following code, although gcc happily compiles:
--------------
#include <stddef.h>
extern void* f(void** p, size_t s);

#include <stdlib.h>


void* f(void** p, size_t s)
{
}
--------------

I think the problem is with the redefinition of size_t in stdlib.h (a system
header) to something different from what is defined in stddef.h.

In Mac OS X, clang complains:
[filcab at farnsworth ~] $ clang b.c
b.c:7:7: error: conflicting types for 'f'
void* f(void** p, size_t s)
      ^
b.c:2:14: error: previous declaration is here
extern void* f(void** p, size_t s);
             ^
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 Jun 10 14:44:53 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 10 Jun 2008 14:44:53 -0500
Subject: [LLVMbugs] [Bug 1979] Potential loop codegen improvement
In-Reply-To: <bug-1979-206@http.llvm.org/bugs/>
Message-ID: <200806101944.m5AJirna025394@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=1979


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

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




--- Comment #4 from Evan Cheng <evan.cheng at apple.com>  2008-06-10 14:44:53 ---
I forgot about this one. But this has been fixed by coalescer improvements.


-- 
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 Jun 10 19:21:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 10 Jun 2008 19:21:30 -0500
Subject: [LLVMbugs] [Bug 2428] New: Windows Vista don't support format %n
Message-ID: <bug-2428-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2428

           Summary: Windows Vista don't support format %n
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: yonggangluo at hotmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1716)
 --> (http://llvm.org/bugs/attachment.cgi?id=1716)
The changed source file for  gccsrc\gcc\genmodes.c

Because Windows Vista don't support format %n, so the file 
gcc4.2\trunk\gcc\genmodes.c can't execute correct! So please use a replacement
of %n!
or It's will generate a very hudge file in windows vista! It's so crazy and I
have to change it to something new!
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 Jun 11 01:43:14 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 11 Jun 2008 01:43:14 -0500
Subject: [LLVMbugs] [Bug 2429] New: unwind crashes lli
Message-ID: <bug-2429-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2429

           Summary: unwind crashes lli
           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


The attached .bc file crashes lli unless -force-interpreter is used. It also
crashes when compiled to native code. (this is with the current svn head.)

Here's the IR:
---------------------------------
; ModuleID = 'unwind'

define i32 @"tart.main()->i32"() {
entry:
        invoke void @"tart.foo()"( )
                        to label %nounwind unwind label %catch

catch:
  ret i32 0

nounwind:
        ret i32 1
}

define void @"tart.foo()"() {
entry:
        unwind
}

define i32 @main(i32, i8**) {
entry:
        invoke i32 @"tart.main()->i32"()
                        to label %success unwind label %failure         ;
<i32>:2 [#uses=1]

success:                ; preds = %entry
        ret i32 %2

failure:                ; preds = %entry
        ret i32 -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  Wed Jun 11 02:08:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 11 Jun 2008 02:08:19 -0500
Subject: [LLVMbugs] [Bug 2429] unwind crashes lli
In-Reply-To: <bug-2429-206@http.llvm.org/bugs/>
Message-ID: <200806110708.m5B78Jbp012477@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2429


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #2 from Duncan Sands <baldrick at free.fr>  2008-06-11 02:08:19 ---
Codegen does not support unwind unfortunately.
For the moment you have to run the LowerInvoke
pass with -enable-correct-eh-support.  I know
this sucks, but doing this right is a hard problem,
though it may not appear that way!

*** This bug has been marked as a duplicate of bug 1578 ***


-- 
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 Jun 12 11:48:11 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 12 Jun 2008 11:48:11 -0500
Subject: [LLVMbugs] [Bug 2432] New: llvm-g++ can't cope with lots of cleanups
Message-ID: <bug-2432-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2432

           Summary: llvm-g++ can't cope with lots of cleanups
           Product: tools
           Version: 2.2
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-g++
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jay.foad at antixlabs.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1721)
 --> (http://llvm.org/bugs/attachment.cgi?id=1721)
test case

I'm using the pre-built llvm-gcc4.2-2.2-x86-linux-RHEL4 on debian. With the
attached test case (taken from the GCC 4.2.1 testsuite), I get:

$ ~/llvm/llvm-gcc4.2-2.2-x86-linux-RHEL4/bin/llvm-gcc -c cleanup1.C
[It runs for a couple of minutes and uses up about 1 GB of virtual memory,
then...]
llvm-g++: Internal error: Killed (program cc1plus)
Please submit a full bug report.
See <URL:http://developer.apple.com/bugreporter> for instructions.

If I try the same thing with a normal release of GCC 4.2.1 it completes in a
couple of seconds and uses less than 100 MB.


-- 
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 Jun 12 12:59:33 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 12 Jun 2008 12:59:33 -0500
Subject: [LLVMbugs] [Bug 2433] New: scalar-evolutions BinomialCoefficient
 function fails zext' ing i64 to i64
Message-ID: <bug-2433-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2433

           Summary: scalar-evolutions BinomialCoefficient function fails
                    zext'ing i64 to i64
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Loop Optimizer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1722)
 --> (http://llvm.org/bugs/attachment.cgi?id=1722)
testcase and fix

$ opt -analyze -scalar-evolution bugpoint-reduced-simplified.bc
Printing analysis 'Scalar Evolution Analysis' for function 'main':
Classifying expressions for: main
        %accum.03 = phi i64 [ %tmp14, %bb10 ], [ 0, %entry ]            ; <i64>
[#uses=1]
opt: Constants.cpp:1880: static llvm::Constant*
llvm::ConstantExpr::getZExt(llvm::Constant*, const llvm::Type*): Assertion
`C->getType()->getPrimitiveSizeInBits() < Ty->getPrimitiveSizeInBits()&& "SrcTy
must be smaller than DestTy for ZExt!"' failed.
  --> { 0,+, 1,+, 1}<bb10>              Exits: opt[0x8697f1e]
opt[0x8698180]
[0xffffe500]
[0x6]
/lib/tls/i686/cmov/libc.so.6(abort+0xe9)[0x49739619]
/lib/tls/i686/cmov/libc.so.6(__assert_fail+0x101)[0x49731031]
opt(_ZN4llvm12ConstantExpr7getZExtEPNS_8ConstantEPKNS_4TypeE+0xe4)[0x85d19fe]
opt(_ZN4llvm15ScalarEvolution17getZeroExtendExprERKNS_10SCEVHandleEPKNS_4TypeE+0x3a)[0x8563de8]
opt[0x856a7ae]
opt(_ZNK4llvm14SCEVAddRecExpr19evaluateAtIterationENS_10SCEVHandleERNS_15ScalarEvolutionE+0x6d)[0x856abc7]
opt[0x856eb0c]
opt(_ZNK4llvm15ScalarEvolution14getSCEVAtScopeEPNS_5ValueEPKNS_8LoopBaseINS_10BasicBlockEEE+0x55)[0x856ebb9]
opt(_ZNK4llvm15ScalarEvolution5printERSoPKNS_6ModuleE+0x27f)[0x856ee51]
opt(_ZNK4llvm4Pass5printEPSoPKNS_6ModuleE+0x2c)[0x8369f92]
opt(_ZN36_GLOBAL__N_opt.cpp_00000000_B17F1A4619FunctionPassPrinter13runOnFunctionERN4llvm8FunctionE+0x106)[0x836b818]
opt(_ZN4llvm13FPPassManager13runOnFunctionERNS_8FunctionE+0x13d)[0x86265c9]
opt(_ZN4llvm13FPPassManager11runOnModuleERNS_6ModuleE+0x6e)[0x8626778]
opt(_ZN4llvm13MPPassManager11runOnModuleERNS_6ModuleE+0x108)[0x8626244]
opt(_ZN4llvm15PassManagerImpl3runERNS_6ModuleE+0x6e)[0x86263fe]
opt(_ZN4llvm11PassManager3runERNS_6ModuleE+0x1b)[0x8626453]
opt(main+0xa97)[0x8369519]
/lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xd2)[0x49723ec2]
opt(__gxx_personality_v0+0x321)[0x835b8f9]
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  Thu Jun 12 13:40:22 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 12 Jun 2008 13:40:22 -0500
Subject: [LLVMbugs] [Bug 2434] New: The -indvars pass generates invalid
	bytecode
Message-ID: <bug-2434-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2434

           Summary: The -indvars pass generates invalid bytecode
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: criswell at uiuc.edu
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1723)
 --> (http://llvm.org/bugs/attachment.cgi?id=1723)
LLVM Assembly test case to reproduce problem

The Canonicalize Induction Variables (-indvars) pass generates bitcode where
instructions are used before they are defined.

A reduced testcase is provided.


-- 
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 Jun 12 14:33:59 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 12 Jun 2008 14:33:59 -0500
Subject: [LLVMbugs] [Bug 2432] llvm-g++ can't cope with lots of cleanups
In-Reply-To: <bug-2432-206@http.llvm.org/bugs/>
Message-ID: <200806121933.m5CJXxFw000449@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2432


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>  2008-06-12 14:33:59 ---


*** This bug has been marked as a duplicate of bug 1622 ***


-- 
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 Jun 12 15:18:57 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 12 Jun 2008 15:18:57 -0500
Subject: [LLVMbugs] [Bug 2435] New: likely bad codegen bug
Message-ID: <bug-2435-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2435

           Summary: likely bad codegen bug
           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


For recent svn targeting ia32.

The generated code does not store to g_139 at all.  It should.

[regehr at babel tmp18]$ cat small.c
typedef int int32_t;
typedef unsigned int uint32_t;
int32_t g_139;

void func_56 (int32_t p_60)
{
  g_139 = 1;
  if (p_60) {
    for (1; 1; 1)
      g_139 = 1;
  }
}

[regehr at babel tmp18]$ llvm-gcc -O small.c -S --emit-llvm -o -
; ModuleID = 'small.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"
@g_139 = common global i32 0            ; <i32*> [#uses=0]

define void @func_56(i32 %p_60) nounwind  {
entry:
        %tmp1 = icmp eq i32 %p_60, 0            ; <i1> [#uses=1]
        br i1 %tmp1, label %return, label %bb

bb:             ; preds = %entry, %bb
        br label %bb

return:         ; preds = %entry
        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 Jun 12 17:04:44 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 12 Jun 2008 17:04:44 -0500
Subject: [LLVMbugs] [Bug 2436] New: doesn't track load and store with one
	unknown
Message-ID: <bug-2436-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2436

           Summary: doesn't track load and store with one unknown
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


In this code, the call to abort() doesn't get optimized away.

  extern char *getPtr(void);

  void foo(int i) {
    char *arr[10];

    arr[i] = getPtr();
    arr[i][10] = 42;

    if (arr[i][10] != 42)
      abort();
  }

By constrast, if you remove "int i" everywhere, you'll find that we optimize it
down just fine.


-- 
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 Jun 12 19:51:11 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 12 Jun 2008 19:51:11 -0500
Subject: [LLVMbugs] [Bug 2432] llvm-g++ can't cope with lots of cleanups
In-Reply-To: <bug-2432-206@http.llvm.org/bugs/>
Message-ID: <200806130051.m5D0pBTv010585@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2432


Eli Friedman <sharparrow1 at yahoo.com> changed:

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




--- Comment #2 from Eli Friedman <sharparrow1 at yahoo.com>  2008-06-12 19:51:11 ---
Are you sure this isn't a front-end regression?  This isn't an optimizer bug;
-O0 -emit-llvm by itself is ridiculously slow.


-- 
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 Jun 12 23:39:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 12 Jun 2008 23:39:28 -0500
Subject: [LLVMbugs] [Bug 2433] scalar-evolutions BinomialCoefficient
 function fails zext' ing i64 to i64
In-Reply-To: <bug-2433-206@http.llvm.org/bugs/>
Message-ID: <200806130439.m5D4dSYj017195@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2433


Nick Lewycky <nicholas at mxc.ca> changed:

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




--- Comment #2 from Nick Lewycky <nicholas at mxc.ca>  2008-06-12 23:39:27 ---
Fixed in r52248:

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080609/063590.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 Jun 13 01:45:00 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 13 Jun 2008 01:45:00 -0500
Subject: [LLVMbugs] [Bug 2436] doesn't track load and store with one unknown
In-Reply-To: <bug-2436-206@http.llvm.org/bugs/>
Message-ID: <200806130645.m5D6j0R6020819@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2436


Nick Lewycky <nicholas at mxc.ca> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nicholas at mxc.ca
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID




--- Comment #1 from Nick Lewycky <nicholas at mxc.ca>  2008-06-13 01:45:00 ---
It works with opt -anders-aa -gvn.


-- 
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 Jun 13 03:00:04 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 13 Jun 2008 03:00:04 -0500
Subject: [LLVMbugs] [Bug 2451] New: Anders pass asserts on
	extractelement/insertelement
Message-ID: <bug-2451-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2451

           Summary: Anders pass asserts on extractelement/insertelement
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Global Analyses
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1740)
 --> (http://llvm.org/bugs/attachment.cgi?id=1740)
Test case

Very easy to reproduce: Feed any program containing a insertelement or extract
element to Anders AA pass, and it asserts. I guess nobody told anders about
these instructions yet...

I've attached a small testcase, which is taken from the instcombine testcases
(test/Transforms/InstCombine/2006-03-30-ExtractElement.ll). Run with
  llvm-as <  ./2006-03-30-ExtractElement.ll | opt -anders-aa
and it gives

Unknown instruction:    %V2 = insertelement <4 x float> %V, float 1.000000e+00,
i32 3           ; <<4 x float>> [#uses=1]

Program received signal SIGABRT, Aborted.
[Switching to Thread -1211025728 (LWP 16041)]
0xffffe410 in __kernel_vsyscall ()
(gdb) bt
#0  0xffffe410 in __kernel_vsyscall ()
#1  0xb7d3f875 in raise () from /lib/tls/i686/cmov/libc.so.6
#2  0xb7d41201 in abort () from /lib/tls/i686/cmov/libc.so.6
#3  0x0853e729 in (anonymous namespace)::Andersens::visitInstruction
(this=0x87c9980, I=@0x87c945c) at Andersens.cpp:1149
#4  0x0855347c in llvm::InstVisitor<(anonymous namespace)::Andersens,
void>::visitInsertElementInst (this=0x87c9980, I=@0x87c945c)
    at /home/kooijman/src/llvm-trunk/include/llvm/Support/InstVisitor.h:197
#5  0x08553496 in llvm::InstVisitor<(anonymous namespace)::Andersens,
void>::visitInsertElement (this=0x87c9980, I=@0x87c945c)
    at /home/kooijman/src/llvm-trunk/include/llvm/Instruction.def:165
#6  0x08553b72 in llvm::InstVisitor<(anonymous namespace)::Andersens,
void>::visit (this=0x87c9980, I=@0x87c945c)
    at /home/kooijman/src/llvm-trunk/include/llvm/Instruction.def:165
#7  0x08553c31 in llvm::InstVisitor<(anonymous namespace)::Andersens,
void>::visit<llvm::ilist_iterator<llvm::Instruction> > (this=0x87c9980, Start=
      {<bidirectional_iterator<llvm::Instruction,int>> = {<> = {<No data
fields>}, <No data fields>}, NodePtr = 0x87ca878}, End=
      {<bidirectional_iterator<llvm::Instruction,int>> = {<> = {<No data
fields>}, <No data fields>}, NodePtr = 0x87c9488})
    at /home/kooijman/src/llvm-trunk/include/llvm/Support/InstVisitor.h:92
#8  0x08553ca8 in llvm::InstVisitor<(anonymous namespace)::Andersens,
void>::visit (this=0x87c9980, BB=@0x87c9358)
    at /home/kooijman/src/llvm-trunk/include/llvm/Support/InstVisitor.h:107
#9  0x08553cf3 in llvm::InstVisitor<(anonymous namespace)::Andersens,
void>::visit<llvm::ilist_iterator<llvm::BasicBlock> > (this=0x87c9980, Start=
      {<bidirectional_iterator<llvm::BasicBlock,int>> = {<> = {<No data
fields>}, <No data fields>}, NodePtr = 0x87c9380}, End=
      {<bidirectional_iterator<llvm::BasicBlock,int>> = {<> = {<No data
fields>}, <No data fields>}, NodePtr = 0x87c9380})
    at /home/kooijman/src/llvm-trunk/include/llvm/Support/InstVisitor.h:92
#10 0x08553d6a in llvm::InstVisitor<(anonymous namespace)::Andersens,
void>::visit (this=0x87c9980, F=@0x87c97b8)
    at /home/kooijman/src/llvm-trunk/include/llvm/Support/InstVisitor.h:103
#11 0x08553d88 in llvm::InstVisitor<(anonymous namespace)::Andersens,
void>::visit (this=0x87c9980, F=0x87c97b8)
    at /home/kooijman/src/llvm-trunk/include/llvm/Support/InstVisitor.h:112
#12 0x08541cc7 in (anonymous namespace)::Andersens::CollectConstraints
(this=0x87c9980, M=@0x87c7e30) at Andersens.cpp:1085
#13 0x08553dca in (anonymous namespace)::Andersens::runOnModule
(this=0x87c9980, M=@0x87c7e30) at Andersens.cpp:439
#14 0x08654772 in llvm::MPPassManager::runOnModule (this=0x87c80c8,
M=@0x87c7e30) at PassManager.cpp:1264
#15 0x0865492a in llvm::PassManagerImpl::run (this=0x87c8918, M=@0x87c7e30) at
PassManager.cpp:1338
#16 0x0865497d in llvm::PassManager::run (this=0xbf8f4f34, M=@0x87c7e30) at
PassManager.cpp:1370
#17 0x083958c2 in main (argc=3, argv=0xbf8f5134) at opt.cpp:431


-- 
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 Jun 13 11:41:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 13 Jun 2008 11:41:19 -0500
Subject: [LLVMbugs] [Bug 2452] New: Some tests generate GCC warnings
Message-ID: <bug-2452-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2452

           Summary: Some tests generate GCC warnings
           Product: Test Suite
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: DejaGNU
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


There are some tests that are generating GCC warnings, which until recently
were ignored. Now, stderr output from a test program is cause for failure.
Preferably these warnings should be fixed, but possibly they also mean that the
test is not working properly. Someone who knows details about these tests
should have a look.

It concerns the following tests:

FAIL: /home/kooijman/src/llvm-trunk/test/FrontendC/2003-08-21-WideString.c
  Failed with unknown error (or has stderr output) at line 1
  while running: /home/kooijman/src/llvm-gcc/install/bin/llvm-gcc  -emit-llvm
-S /home/kooijman/src/llvm-trunk/test/FrontendC/2003-08-21-WideString.c -o - |
llvm-as -f -o /dev/null

  /home/kooijman/src/llvm-trunk/test/FrontendC/2003-08-21-WideString.c:5:
warning: initialization from incompatible pointer type

FAIL: /home/kooijman/src/llvm-trunk/test/FrontendC/2004-02-13-IllegalVararg.c
  Failed with unknown error (or has stderr output) at line 1
  while running: /home/kooijman/src/llvm-gcc/install/bin/llvm-gcc  -emit-llvm
-xc /home/kooijman/src/llvm-trunk/test/FrontendC/2004-02-13-IllegalVararg.c -w
-c -o - | llc

  /home/kooijman/src/llvm-trunk/test/FrontendC/2004-02-13-IllegalVararg.c: In
function ???test???:
  /home/kooijman/src/llvm-trunk/test/FrontendC/2004-02-13-IllegalVararg.c:9:
note: if this code is reached, the program will abort

FAIL:
/home/kooijman/src/llvm-trunk/test/FrontendC/2006-09-18-fwrite-cast-crash.c for
PR910
  Failed with unknown error (or has stderr output) at line 1
  while running: /home/kooijman/src/llvm-gcc/install/bin/llvm-gcc  -emit-llvm
/home/kooijman/src/llvm-trunk/test/FrontendC/2006-09-18-fwrite-cast-crash.c -S
-o /dev/null

  /home/kooijman/src/llvm-trunk/test/FrontendC/2006-09-18-fwrite-cast-crash.c:
In function ???Usage???:
 
/home/kooijman/src/llvm-trunk/test/FrontendC/2006-09-18-fwrite-cast-crash.c:11:
warning: function called through a non-compatible type
 
/home/kooijman/src/llvm-trunk/test/FrontendC/2006-09-18-fwrite-cast-crash.c:11:
note: if this code is reached, the program will abort

FAIL: /home/kooijman/src/llvm-trunk/test/FrontendC/2007-09-05-ConstCtor.c for
PR1641
  Failed with unknown error (or has stderr output) at line 1
  while running: /home/kooijman/src/llvm-gcc/install/bin/llvm-gcc  -emit-llvm
-xc -Os -c /home/kooijman/src/llvm-trunk/test/FrontendC/2007-09-05-ConstCtor.c
-o /dev/null

  /home/kooijman/src/llvm-trunk/test/FrontendC/2007-09-05-ConstCtor.c:11:30:
warning: multi-character character constant

FAIL:
/home/kooijman/src/llvm-trunk/test/FrontendC/2007-09-28-PackedUnionMember.c
  Failed with unknown error (or has stderr output) at line 1
  while running: /home/kooijman/src/llvm-gcc/install/bin/llvm-gcc  -emit-llvm
/home/kooijman/src/llvm-trunk/test/FrontendC/2007-09-28-PackedUnionMember.c -S
-o -

  declare i32 @bork(i32 (...)*, %struct.E*)
 
/home/kooijman/src/llvm-trunk/test/FrontendC/2007-09-28-PackedUnionMember.c:28:23:
warning: multi-character character constant
 
/home/kooijman/src/llvm-trunk/test/FrontendC/2007-09-28-PackedUnionMember.c:33:7:
warning: multi-character character constant

I've XFAIL'd these test for 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  Fri Jun 13 11:45:10 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 13 Jun 2008 11:45:10 -0500
Subject: [LLVMbugs] [Bug 2453] New: Testcase (and/or cbe) generates
	uncompilable C code
Message-ID: <bug-2453-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2453

           Summary: Testcase (and/or cbe) generates uncompilable C code
           Product: Test Suite
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: minor
          Priority: P2
         Component: DejaGNU
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


THe misopt-basictest.ll bugpoint test generates failing C code through CBE.
This
cause compilation to fail and bugpoint never gets around to finding the actual
bug. It seems this is caused or at least related to the lack of a datalayout in
the .ll file. If i add:
  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"          

Things compile as expected on my x86-32 linux system. The error in question
concern the typedef for an 128 bit integer type, which isn't even used in the
code.

I've XFAIL'd the test for now.



FAIL: /home/kooijman/src/llvm-trunk/test/BugPoint/misopt-basictest.ll
Failed with unknown error (or has stderr output) at line 1
while running: bugpoint
/home/kooijman/src/llvm-trunk/test/BugPoint/misopt-basictest.ll -dce
-bugpoint-deletecalls -simplifycfg -silence-passes
Read input file      :
'/home/kooijman/src/llvm-trunk/test/BugPoint/misopt-basictest.ll'
*** All input ok
Running selected passes on program to test for crash: Success!
Initializing execution environment: Found gcc: /usr/bin/gcc
Running the code generator to test for a crash: <cbe>
Generating reference output from raw program: <cbe><cbe><gcc>
Checking to see if we can delete global inits: <cbe>  - Removing all global
inits hides problem!
Checking for crash with only these blocks: : <cbe>
*** Attempting to reduce testcase by deleting instructions: Simplification
Level #1
Checking instruction '': <cbe>
*** Attempting to reduce testcase by deleting instructions: Simplification
Level #0
Checking instruction '': <cbe>
*** Attempting to perform final cleanups: <cbe>Emitted bitcode to
'bugpoint-reduced-simplified.bc'

*** You can reproduce the problem with: opt bugpoint-reduced-simplified.bc -dce
-bugpoint-deletecalls -simplifycfg

Error running tool:
  /usr/bin/gcc -x c -fno-strict-aliasing bugpoint-test-program.bc.cbe.c-BXIQYJ
-x none -o bugpoint-test-program.bc.cbe.c-BXIQYJ.gcc.exe -lm -O2 -Wl,-R.
bugpoint-test-program.bc.cbe.c-BXIQYJ:83: error: unable to emulate ???TI???
bugpoint-test-program.bc.cbe.c-BXIQYJ:84: error: unable to emulate ???TI???
bugpoint-test-program.bc.cbe.c-BXIQYJ:127: warning: conflicting types for
built-in function ???malloc???
bugpoint-test-program.bc.cbe.c-BXIQYJ: In function ???main???:
bugpoint-test-program.bc.cbe.c-BXIQYJ:156: warning: return type of ???main???
is not ???int???
*** Debugging code generator 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 Jun 13 11:47:37 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 13 Jun 2008 11:47:37 -0500
Subject: [LLVMbugs] [Bug 2454] New: Testcase failing to find compile unit
Message-ID: <bug-2454-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2454

           Summary: Testcase failing to find compile unit
           Product: Test Suite
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: DejaGNU
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


The NoCompileUnit testcase is currently failing, but I can't figure out what
it's supposed to do. Anyone got a clue?

FAIL:
/home/kooijman/src/llvm-trunk/test/FrontendC++/2006-11-30-NoCompileUnit.cpp
Failed with unknown error (or has stderr output) at line 5
while running: gdb -q -batch -n -x 2006-11-30-NoCompileUnit.cpp.tmp2
NoCompileUnit.exe |  tee NoCompileUnit.out | not grep {"low == high"}

/home/kooijman/src/llvm-trunk/test/FrontendC++/Output/2006-11-30-NoCompileUnit.cpp.tmp2:3:
Error in sourced command file: No symbol "NoCompileUnit" in current context

/home/kooijman/src/llvm-trunk/test/FrontendC++/Output/NoCompileUnit.cpp.tmp2:3:
Error in sourced command file: No symbol "NoCompileUnit" in current context.


-- 
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 Jun 13 16:19:46 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 13 Jun 2008 16:19:46 -0500
Subject: [LLVMbugs] [Bug 2435] likely bad codegen bug
In-Reply-To: <bug-2435-206@http.llvm.org/bugs/>
Message-ID: <200806132119.m5DLJkNH029836@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2435


Eli Friedman <sharparrow1 at yahoo.com> changed:

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




--- Comment #5 from Eli Friedman <sharparrow1 at yahoo.com>  2008-06-13 16:19:45 ---
Fixed.

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080609/063605.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 Jun 13 18:55:07 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 13 Jun 2008 18:55:07 -0500
Subject: [LLVMbugs] [Bug 2455] New: ice
Message-ID: <bug-2455-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2455

           Summary: ice
           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


[regehr at babel tmp19]$ llvm-gcc -O2 -c small.c
cc1: /uusoc/exports/scratch/regehr/llvm/include/llvm/Instructions.h:1911: void
llvm::PHINode::addIncoming(llvm::Value*, llvm::BasicBlock*): Assertion `V &&
"PHI node got a null value!"' failed.
small.c:28: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://developer.apple.com/bugreporter> for instructions.
[regehr at babel tmp19]$ 



static inline unsigned int lshift_u_s (unsigned int left, int right)
{
  if ((right) || (right >= sizeof (unsigned int))
      || (left > ((1 * 2U + 1U) >> right)))
    return left;
  return left << right;
}
short g_56;
unsigned g_87;
unsigned g_90;
int func_67 (int p_68, char p_69, char p_71)
{
  for (g_87 = 1; 1; g_87 = 1)
    {
      for (p_71 = 1; p_71; --p_71)
        {
          unsigned l_110 = 1;
          if (lshift_u_s (l_110, p_69))
            {
              if (g_90)
                return 1;
              return 0;
            }
          for (g_56 = 0; 0; 1);
        }
      for (0; g_56; 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  Sat Jun 14 12:27:05 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 14 Jun 2008 12:27:05 -0500
Subject: [LLVMbugs] [Bug 2460] New: Module dtor segfaults on empty PHI nodes.
Message-ID: <bug-2460-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2460

           Summary: Module dtor segfaults on empty PHI nodes.
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: quality-of-implementation
          Severity: normal
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: mdevan.foobar at gmail.com
                CC: llvmbugs at cs.uiuc.edu


If a module has an empty PHI node (a PHI node on which addIncomingBlock() has
not been called), then it segfaults while the module object is destroyed. It
can be reproduced by the attached test program.

Probably can be fixed by modifying User::dropHungoffUses().


-- 
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 Jun 14 12:41:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 14 Jun 2008 12:41:09 -0500
Subject: [LLVMbugs] [Bug 2461] New: __attribute__ ((noreturn)) in a function
	parameter
Message-ID: <bug-2461-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2461

           Summary: __attribute__ ((noreturn)) in a function parameter
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: filcab at gmail.com
                CC: llvmbugs at cs.uiuc.edu


clang doesn't allow __attribute__ ((noreturn)) in a function parameter:
[filcab at farnsworth ~] $ cat b.c
extern void LibVEX_Init (
   /* failure exit function */
   __attribute__ ((noreturn))
   void (*failure_exit) ( void )
);
[filcab at farnsworth ~] $ ccc -o b.o -c b.c
clang -emit-llvm-bc -x c -o b.o b.c
b.c:3:20: warning: 'noreturn' attribute only applies to function types
   __attribute__ ((noreturn))
                   ^
1 diagnostic generated.
[filcab at farnsworth ~] $ gcc -o b.o -c b.c
[filcab at farnsworth ~] $


-- 
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 Jun 15 11:06:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 15 Jun 2008 11:06:48 -0500
Subject: [LLVMbugs] [Bug 2462] New: Compilation failure with -O4,
	where -O3 succeeds
Message-ID: <bug-2462-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2462

           Summary: Compilation failure with -O4, where -O3 succeeds
           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: sherlock at genome.stanford.edu
                CC: llvmbugs at cs.uiuc.edu


This is using the binary version of llvm-gcc 2.3, on MacOS 10.5.3, with XCode
3.0 installed:

Steps to reproduce (I made a reduction):

Compile source with:

llvm-gcc -O4 reduction.c -o reduction

result:

ld: warning in /usr/tmp/cc8mQGAH.o, file is not of required architecture
Undefined symbols:
  "_main", referenced from:
      start in crt1.10.5.o
ld: symbol(s) not found
collect2: ld returned 1 exit status

However, changing the -O4 for -O3 makes it compile fine.  Also, using gcc,
instead of llvm-gcc works fine.

The reduction is:

int main(int argc, char *argv[]){

  return 0;

}

Other relevant info:

>llvm-gcc -v
Using built-in specs.
Target: i686-apple-darwin9
Configured with: ../llvm-gcc4.2-2.3.source/configure
--prefix=/Users/tonic/2.3/llvm-gcc/install --program-prefix=llvm-
--enable-llvm=/Users/tonic/2.3/llvm-2.3/ --enable-languages=c,c++,objc,obj-c++
--with-arch=nocona --with-tune=generic
--with-gxx-include-dir=/usr/include/c++/4.0.0 --build=i686-apple-darwin9
--host=i686-apple-darwin9 --target=i686-apple-darwin9
Thread model: posix
gcc version 4.2.1 (Based on Apple Inc. build 5555) (LLVM build 2.3)

>uname -a
Darwin Gavza 9.3.0 Darwin Kernel Version 9.3.0: Fri May 23 00:49:16 PDT 2008;
root:xnu-1228.5.18~1/RELEASE_I386 i386

>gcc -v
Using built-in specs.
Target: i686-apple-darwin9
Configured with: /var/tmp/gcc/gcc-5465~16/src/configure --disable-checking
-enable-werror --prefix=/usr --mandir=/share/man
--enable-languages=c,objc,c++,obj-c++
--program-transform-name=/^[cg][^.-]*$/s/$/-4.0/
--with-gxx-include-dir=/include/c++/4.0.0 --with-slibdir=/usr/lib
--build=i686-apple-darwin9 --with-arch=apple --with-tune=generic
--host=i686-apple-darwin9 --target=i686-apple-darwin9
Thread model: posix
gcc version 4.0.1 (Apple Inc. build 5465)

One final bit of news, using llvm-gcc, instead of gcc to compile my
application, with -O3, made it more than twice as fast (runtime went to 25
minutes from 1 hour 9 minutes - it's a CPU intensive program for analysis of
biological data).  Caveat is that it was compiled with gcc 4.0.1, as Apple
hasn't yet released (at least publicly) XCode 3.1 with gcc 4.2, so I can't say
how much of that improvement was due to gcc version rather than llvm.  Was
interested to try -)4 to see if it further improved.


-- 
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 Jun 15 14:06:59 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 15 Jun 2008 14:06:59 -0500
Subject: [LLVMbugs] [Bug 2462] Compilation failure with -O4,
	where -O3 succeeds
In-Reply-To: <bug-2462-206@http.llvm.org/bugs/>
Message-ID: <200806151906.m5FJ6x06000475@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2462


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #5 from Chris Lattner <sabre at nondot.org>  2008-06-15 14:06:59 ---
Please download the free iphone sdk.  It includes xcode 3.1 which has the
llvm-gcc compatible linker.  If this still doesn't work after doing that,
please reopen.

-Chris


-- 
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 Jun 15 14:09:34 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 15 Jun 2008 14:09:34 -0500
Subject: [LLVMbugs] [Bug 2434] The -indvars pass generates invalid bytecode
In-Reply-To: <bug-2434-206@http.llvm.org/bugs/>
Message-ID: <200806151909.m5FJ9YKT000626@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2434


Wojciech Matyjewicz <wmatyjewicz at fastmail.fm> changed:

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




--- Comment #2 from Wojciech Matyjewicz <wmatyjewicz at fastmail.fm>  2008-06-15 14:09:34 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080609/063665.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 Jun 15 15:52:34 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 15 Jun 2008 15:52:34 -0500
Subject: [LLVMbugs] [Bug 2462] Compilation failure with -O4,
	where -O3 succeeds
In-Reply-To: <bug-2462-206@http.llvm.org/bugs/>
Message-ID: <200806152052.m5FKqYg4004419@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2462


Gavin Sherlock <sherlock at genome.stanford.edu> changed:

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




--- Comment #7 from Gavin Sherlock <sherlock at genome.stanford.edu>  2008-06-15 15:52:34 ---
I installed XCode 3.1.  The error message is different, but the result the
same:

>llvm-gcc -O4 reduction.c -o reduction
ld: in /usr/tmp/cc0E932t.o, could not process object file.  Looks like an llvm
bitcode object file, but libLTO.dylib could not be loaded
collect2: ld returned 1 exit status


-- 
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 Jun 15 20:40:35 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 15 Jun 2008 20:40:35 -0500
Subject: [LLVMbugs] [Bug 2463] New: linking weak and alias fails
Message-ID: <bug-2463-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2463

           Summary: linking weak and alias fails
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Linker
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: alenhar2 at uiuc.edu
                CC: llvmbugs at cs.uiuc.edu


linking an aliased symbol with the same name as a weak symbol fails.

llvm-ld -disable-inlining -r -o c.o  a.bc b.o

a.ll:
; 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-f80:32:32"
target triple = "i386-pc-linux-gnu"

@sched_clock = alias i64 ()* @native_sched_clock                ; <i64 ()*>
[#uses=0]

define i64 @native_sched_clock() nounwind  {
entry:
        ret i64 0
}


b.ll:
; 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-f80:32:32"
target triple = "i386-pc-linux-gnu"


define i64 @sched_clock_cpu(i32 inreg  %cpu) nounwind  {
entry:
        %tmp = call i64 @sched_clock( ) nounwind                ; <i64>
[#uses=2]
        ret i64 %tmp
}

define weak i64 @sched_clock() {
entry:
        ret i64 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  Sun Jun 15 23:03:02 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 15 Jun 2008 23:03:02 -0500
Subject: [LLVMbugs] [Bug 2460] Module dtor segfaults on empty PHI nodes.
In-Reply-To: <bug-2460-206@http.llvm.org/bugs/>
Message-ID: <200806160403.m5G432OK016989@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2460


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #2 from Chris Lattner <sabre at nondot.org>  2008-06-15 23:03:01 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080609/063668.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 Jun 15 23:26:33 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 15 Jun 2008 23:26:33 -0500
Subject: [LLVMbugs] [Bug 2453] Testcase (and/or cbe) generates uncompilable
	C code
In-Reply-To: <bug-2453-206@http.llvm.org/bugs/>
Message-ID: <200806160426.m5G4QXAF017824@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2453


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-06-15 23:26:33 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080609/063672.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 Jun 15 23:30:56 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 15 Jun 2008 23:30:56 -0500
Subject: [LLVMbugs] [Bug 2436] doesn't track load and store with one unknown
In-Reply-To: <bug-2436-206@http.llvm.org/bugs/>
Message-ID: <200806160430.m5G4UuIJ017991@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2436


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #2 from Chris Lattner <sabre at nondot.org>  2008-06-15 23:30:56 ---
basicaa should handle 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  Mon Jun 16 01:20:18 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 16 Jun 2008 01:20:18 -0500
Subject: [LLVMbugs] [Bug 2436] basicaa doesn't take advantage of escape info
 for aliasing
In-Reply-To: <bug-2436-206@http.llvm.org/bugs/>
Message-ID: <200806160620.m5G6KIsB021084@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2436


Chris Lattner <sabre at nondot.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
             Status|REOPENED                    |RESOLVED
         Resolution|                            |FIXED
            Summary|doesn't track load and store|basicaa doesn't take
                   |with one unknown            |advantage of escape info for
                   |                            |aliasing




--- Comment #3 from Chris Lattner <sabre at nondot.org>  2008-06-16 01:20:17 ---
Implemented here, great catch, thanks!
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063676.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 Jun 16 02:17:33 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 16 Jun 2008 02:17:33 -0500
Subject: [LLVMbugs] [Bug 2462] Compilation failure with -O4,
	where -O3 succeeds
In-Reply-To: <bug-2462-206@http.llvm.org/bugs/>
Message-ID: <200806160717.m5G7HXRY022826@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2462


Gavin Sherlock <sherlock at genome.stanford.edu> changed:

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




--- Comment #9 from Gavin Sherlock <sherlock at genome.stanford.edu>  2008-06-16 02:17:32 ---
My bad.  I was still pointing at the llvm installation that I downloaded from
the llvm site.  Once I fixed my path, it works correctly.  Resolved 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  Mon Jun 16 03:35:15 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 16 Jun 2008 03:35:15 -0500
Subject: [LLVMbugs] [Bug 2464] New: Miscompilation: register smashed on ppc
Message-ID: <bug-2464-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2464

           Summary: Miscompilation: register smashed on ppc
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: PowerPC
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: gbenson at redhat.com
                CC: llvmbugs at cs.uiuc.edu


func5 in the attached test.bc gets miscompiled on 32-bit ppc Linux.  The
failing section is instrumented with calls to trace_bytecode and print_value,
which product the following output:

  630: iload
  local_4_113 = 0
  632: iload
  local_5_114 = 57
  local_4_113 = 0
  634: if_icmplt
  132: aload_3
  133: iload
  135: caload
  136: istore
  138: iload
  140: sipush
  143: if_icmpge
  146: iload
  148: iload
  150: if_icmplt
  185: aload
  187: iload
  189: iinc
  192: iload 
  194: i2b
  195: bastore
  196: iinc 
  199: goto 
  630: iload
  local_4_419 = 1 
  632: iload 
  local_5_420 = 261095424
  local_4_419 = 1 

The variable local_5_* is used to see when to exit a loop; it's "sl" in this:

  int sp = 0;
  int sl = whatever;
  while (sp < sl) {
    // do stuff
  }

In the attached test.s (and test.ll):

  lines 2646-2648 (line 3993 in test.ll) print the first "632: iload"
  lines 2649-2652 (line 3994 in test.ll) print "local_5_114 = 57" (correct)

>From line 2651 you can see that the 57 came from r26.

At the bottom: 

  lines 4901-4903 (7791 in test.ll) print the second "632: iload"
  lines 4904-4907 (7792 in test.ll) print "local_5_420 = 261095424" (junk) 

>From line 4906 you can see that the 261095424 also came from r26. Looking at
what happens to r26 in the meantime it seems it's being used to hold temporary
values:

 lines 2684 and 2685 (4048 and 4049 in test.ll calculate an offset into an
array which is then used in line 2687.
 line 2703 (the top half of the 261101980 in lines 4074 and 4075 in test.ll)
stores the high word of a pair of inlined pointers, used in lines 2704 and
2711.

That last one is where the 261095424 comes from.

This is all with svn revision 52213 BTW, but the miscompilation was also
present in 49380.


-- 
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 Jun 16 08:27:22 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 16 Jun 2008 08:27:22 -0500
Subject: [LLVMbugs] [Bug 2465] New: Invalid register allocation
Message-ID: <bug-2465-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2465

           Summary: Invalid register allocation
           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=1750)
 --> (http://llvm.org/bugs/attachment.cgi?id=1750)
Bytecode in questions

Consider the attached .ll file (sorry, I wasn't be able to reduce it).

in bb27.i.i74 we're seeing:

bb27.i.i74:             ; preds = %bb14.i.i
        %tmp30.i.i69 = getelementptr i8* %tmp1097, i32 %x_addr.0.i.i           
; <i8*> [#uses=1]
        %tmp31.i.i70 = load i8* %tmp30.i.i69, align 1           ; <i8>
[#uses=2]
        %tmp34.i.i71 = getelementptr i8* %tmp1094, i32 %y.0.i.i         ; <i8*>
[#uses=1]
        %tmp35.i.i72 = load i8* %tmp34.i.i71, align 1           ; <i8>
[#uses=1]
        %tmp36.i.i73 = icmp eq i8 %tmp31.i.i70, %tmp35.i.i72            ; <i1>
[#uses=1]
        zext i8 %tmp31.i.i70 to i32             ; <i32>:1418 [#uses=2]
        %tmp1.i2918 = and i32 %1418, 1          ; <i32> [#uses=1]
        %tmp2.i2919 = xor i32 %tmp1.i2918, 1            ; <i32> [#uses=1]
        %tmp4.i2920 = sub i32 0, %1418          ; <i32> [#uses=1]
        %tmp.i2921 = icmp ne i32 %tmp2.i2919, %tmp4.i2920               ; <i1>
[#uses=1]
        and i1 %tmp36.i.i73, %tmp.i2921         ; <i1>:1419 [#uses=1]
        br i1 %1419, label %bb9.i.i, label %snake.exit.i

so, two GEPs plus some innocent bit plays.

These are codegen'ed into:

LBB5_811:       # bb27.i.i74
        movl    208(%esp), %ecx
        movb    (%ecx,%ebp), %al
        movzbl  %al, %ecx
        movl    %ecx, %edx
        andl    $1, %edx
        negl    %ecx
        xorl    $1, %edx
        movl    204(%esp), %ebx
        cmpb    (%ebx,%ecx), %al
        jne     LBB5_813        # bb14.i.i.snake.exit.i_crit_edge

Note, that
        %tmp35.i.i72 = load i8* %tmp34.i.i71, align 1           ; <i8>
[#uses=1]
        %tmp36.i.i73 = icmp eq i8 %tmp31.i.i70, %tmp35.i.i72            ; <i1> 

is turned into 

        movl    204(%esp), %ebx
        cmpb    (%ebx,%ecx), %al

which is definitely incorrect, since %ecx was already clobbered by
computations.

If I run opt -std-compile-opts before, result is pretty sane:

LBB5_744:       # bb27.i.i74
        movl    208(%esp), %esi
        movb    (%esi,%edx), %bl
        movzbl  %bl, %esi
        movl    %esi, %ebp
        negl    %ebp
        andl    $1, %esi
        xorl    $1, %esi
        cmpl    %ebp, %esi
        movl    204(%esp), %esi
        movb    (%esi,%eax), %bh
        je      LBB5_746        # bb14.i.i.snake.exit.i_crit_edge

here %eax and %edx are liveins from prev BBs, where GEP indices were
calculated.


-- 
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 Jun 16 13:33:41 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 16 Jun 2008 13:33:41 -0500
Subject: [LLVMbugs] [Bug 2428] Windows Vista don't support format %n
In-Reply-To: <bug-2428-206@http.llvm.org/bugs/>
Message-ID: <200806161833.m5GIXfrb026923@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2428


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

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




--- Comment #7 from Anton Korobeynikov <asl at math.spbu.ru>  2008-06-16 13:33:41 ---
Fixed in r52342


-- 
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 Jun 16 16:22:21 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 16 Jun 2008 16:22:21 -0500
Subject: [LLVMbugs] [Bug 1860] exponential algorithm when linking types
In-Reply-To: <bug-1860-206@http.llvm.org/bugs/>
Message-ID: <200806162122.m5GLMLEP000980@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=1860


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #8 from Chris Lattner <sabre at nondot.org>  2008-06-16 16:22:21 ---
I committed a slew of fixes the the linker and vmcore for this.  It now runs a
few orders of magnitude faster, please let me know if you hit any other similar
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  Mon Jun 16 23:42:22 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 16 Jun 2008 23:42:22 -0500
Subject: [LLVMbugs] [Bug 2466] New: CodeGenPrepare merges critical edges
 which are immediately split again by OptimizeBlock
Message-ID: <bug-2466-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2466

           Summary: CodeGenPrepare merges critical edges which are
                    immediately split again by OptimizeBlock
           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: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1751)
 --> (http://llvm.org/bugs/attachment.cgi?id=1751)
testcase .bc

On the attached testcase I noticed that EliminateMostlyEmptyBlocks
was eliminating many blocks which later somehow popped up again.
It turns out that it was eliminating all the critical edges, which
were all restored right after by OptimizeBlock.  This seems like a
waste of time and effort: presumably EliminateMostlyEmptyBlocks
should detect this case and leave critical edges alone.


-- 
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 Jun 17 02:13:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 02:13:30 -0500
Subject: [LLVMbugs] [Bug 2467] New: Invalid instcombine of inttoptr -> gep
Message-ID: <bug-2467-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2467

           Summary: Invalid instcombine of inttoptr -> gep
           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


The current inttoptr ->gep optimization in InstCombiner::visitIntToPtr will
always optimize inttoptr(add (ptrtoint x), cst) to gep x, cst.  This is an
invalid transformation in general because unlike integer arithmetic, overflow
for a GEP is undefined.  Therefore, the transformation is only valid if the
resultant pointer is guaranteed to be valid.  (There are a few ways to make
this guarantee: one is if the resulting pointer is immediately loaded from,
another is to ensure x is an object of known size.)

Sample program that breaks when run through clang -emit-llvm-bc | opt
-std-compile-opts (it's a little complicated because it has to trigger both the
bad optimization and an undefined overflow optimization):

int* a(int* a, int i) {return
(int*)((unsigned)((int*)((unsigned)(a)+0x80000000))+0x80000000)+i;}
unsigned b(int* b, int i) {return a(b, i) == b+i;}
int main(int argc, char** argv) {if (!b(&argc, argc)) abort();}

Output of clang -emit-llvm-bc for this program run through mem2reg:
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 = "i686-pc-linux-gnu"

define i32* @a(i32* %a, i32 %i) {
entry:
        %conv = ptrtoint i32* %a to i32
        %add = add i32 %conv, -2147483648
        %conv1 = inttoptr i32 %add to i32*
        %conv2 = ptrtoint i32* %conv1 to i32
        %add3 = add i32 %conv2, -2147483648
        %conv4 = inttoptr i32 %add3 to i32*
        %add.ptr = getelementptr i32* %conv4, i32 %i
        ret i32* %add.ptr
}

define i32 @b(i32* %b, i32 %i) {
entry:
        %call = call i32* @a( i32* %b, i32 %i )
        %add.ptr = getelementptr i32* %b, i32 %i
        %cmp = icmp eq i32* %call, %add.ptr
        %cmp.ext = zext i1 %cmp to i32
        ret i32 %cmp.ext
}

define i32 @main(i32 %argc, i8** %argv) {
entry:
        %argc.addr = alloca i32
        store i32 %argc, i32* %argc.addr
        %tmp = load i32* %argc.addr
        %call = call i32 @b( i32* %argc.addr, i32 %tmp )
        %tobool = icmp ne i32 %call, 0
        %lnot = xor i1 %tobool, true
        br i1 %lnot, label %ifthen, label %ifend

ifthen:         ; preds = %entry
        %call1 = call i32 (...)* @abort( )
        br label %ifend

ifend:          ; preds = %ifthen, %entry
        ret i32 undef
}

Resultant program after running this through opt -std-compile-opts:
; 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-f80:32:32"
target triple = "i686-pc-linux-gnu"

define i32* @a(i32* %a, i32 %i) nounwind  {
entry:
        %conv1.sum = add i32 %i, -1073741824
        %add.ptr = getelementptr i32* %a, i32 %conv1.sum
        ret i32* %add.ptr
}

define i32 @b(i32* %b, i32 %i) nounwind  {
entry:
        %conv1.sum.i = add i32 %i, -1073741824
        %cmp = icmp eq i32 %conv1.sum.i, %i
        %cmp.ext = zext i1 %cmp to i32
        ret i32 %cmp.ext
}

define i32 @main(i32 %argc, i8** %argv) {
entry:
        %conv1.sum.i.i = add i32 %argc, -1073741824
        %cmp.i = icmp eq i32 %conv1.sum.i.i, %argc
        br i1 %cmp.i, label %ifend, label %ifthen

ifthen:         ; preds = %entry
        %call1 = tail call i32 (...)* @abort( )
        ret i32 undef

ifend:          ; preds = %entry
        ret i32 undef
}

declare i32 @abort(...)

There's also another issue with the optimization: it doesn't make sure that the
ptrtoint doesn't lose information by casting to a narrower type.


-- 
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 Jun 17 03:05:26 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 03:05:26 -0500
Subject: [LLVMbugs] [Bug 2468] New: clang mixes up complex numbers
Message-ID: <bug-2468-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2468

           Summary: clang mixes up complex numbers
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: filcab at gmail.com
                CC: llvmbugs at cs.uiuc.edu


clang doesn't give expected results with the following program:

[filcab at farnsworth ~] $ cat b.c
#include <complex.h>
#include <stdio.h>

int main(int argc, char **argv)
{
  complex double long c = 3+I*42;

  printf("%g + I%g\n", creal(c), cimag(c));
  return 0;
}

[filcab at farnsworth ~] $ gcc --std=c99 b.c -o b
[filcab at farnsworth ~] $ ./b
3 + I42
[filcab at farnsworth ~] $ ccc --std=c99 b.c -o b
b.c
clang -emit-llvm-bc -o b.o b.c
llvm-ld -native -disable-internalize -o b b.o
[filcab at farnsworth ~] $ ./b
2.06895e-313 + I-3.81583e-232
[filcab at farnsworth ~] $


-- 
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 Jun 17 04:03:12 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 04:03:12 -0500
Subject: [LLVMbugs] [Bug 2424] libprofile_rt.so installed correctly,
 but libprofile_rt.a not
In-Reply-To: <bug-2424-206@http.llvm.org/bugs/>
Message-ID: <200806170903.m5H93CPF000697@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2424


Holger Schurig <hs4233 at mail.mn-solutions.de> changed:

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




--- Comment #2 from Holger Schurig <hs4233 at mail.mn-solutions.de>  2008-06-17 04:03:10 ---
Your trick would paper over the fault, but I'm not sure if it is the right fix.
Makefile.rules defines in line 308:

CFERuntimeLibDir := $(LLVMGCCDIR)/lib

So either this define is wrong, or my LLVMGCCDIR is wrong. I checked for the
latter and added a

install::
      echo "LLVMGGCCDIR $(LLVMGCCDIR)"

near the top of runtime/Makefile. And indeed, strangely llvmgccdir was empty. I
don't have the slightest idea, because I run configure that way:

/usr/src/llvm/svn.llvm/configure --prefix=/usr/src/llvm/dist
--with-llvmgccdir=/usr/src/llvm/dist --enable-optimized --disable-debug

It's just that the last run of ./configure was ages ago. I re-run, LLVMGCCDIR
got populated and "make -C runtime install" worked. I'm going to close this
bug. However, it might still be buggy somehow, because at the top of
runtime/Makefile there is a check if LLVMGCCDIR is bogus or not. This check
didn't trigger, so maybe the check is buggy.


-- 
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 Jun 17 05:11:29 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 05:11:29 -0500
Subject: [LLVMbugs] [Bug 2469] New: function inlining tries to invoke
	intrinsics
Message-ID: <bug-2469-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2469

           Summary: function inlining tries to invoke intrinsics
           Product: libraries
           Version: 2.2
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Transformation Utilities
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jay.foad at antixlabs.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1753)
 --> (http://llvm.org/bugs/attachment.cgi?id=1753)
bitcode file

I'm using LLVM 2.2 on debian Linux/x86. With the attached bitcode file (derived
from Noiz2sa, http://www.asahi-net.or.jp/~cs8k-cyu/windows/noiz2sa_e.html) I
get:

$ opt -f -o /dev/null reduced.bc -inline
Cannot take the address of an intrinsic!
        %savedstack1 = invoke i8* @llvm.stacksave( )
                        to label %savedstack.noexc unwind label %unwind831     
        ; <i8*> [#uses=1]
Cannot take the address of an intrinsic!
        invoke void @llvm.stackrestore( i8* %savedstack1 )
                        to label %_Z7yyparsev.exit unwind label %unwind831
Broken module found, compilation aborted!
/home/foad/llvm/objdir/Release/bin/opt((anonymous
namespace)::PrintStackTrace()+0x1a)[0x83d981a]
/lib/tls/i686/cmov/libc.so.6(abort+0x109)[0xb7cd0fb9]
/home/foad/llvm/objdir/Release/bin/opt[0x8399c3a]
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 Jun 17 05:33:35 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 05:33:35 -0500
Subject: [LLVMbugs] [Bug 2469] function inlining tries to invoke intrinsics
In-Reply-To: <bug-2469-206@http.llvm.org/bugs/>
Message-ID: <200806171033.m5HAXZim003379@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2469


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>  2008-06-17 05:33:32 ---
I don't see this with svn head, so it seems to have been fixed.
I suggest you try using the 2.3 release instead of 2.2.


-- 
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  Tue Jun 17 12:41:54 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 12:41:54 -0500
Subject: [LLVMbugs] [Bug 2422] segfault ice
In-Reply-To: <bug-2422-206@http.llvm.org/bugs/>
Message-ID: <200806171741.m5HHfsCU016844@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2422


Devang Patel <dpatel at apple.com> changed:

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




--- Comment #5 from Devang Patel <dpatel at apple.com>  2008-06-17 12:41:54 ---
This is fixed by
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063777.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 Jun 17 12:53:00 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 12:53:00 -0500
Subject: [LLVMbugs] [Bug 2455] loop unswitch dom bug
In-Reply-To: <bug-2455-206@http.llvm.org/bugs/>
Message-ID: <200806171753.m5HHr04r017256@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2455


Devang Patel <dpatel at apple.com> changed:

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




--- Comment #5 from Devang Patel <dpatel at apple.com>  2008-06-17 12:53:00 ---
This is also fixed by
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063777.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 Jun 17 15:12:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 15:12:19 -0500
Subject: [LLVMbugs] [Bug 2465] Invalid register allocation
In-Reply-To: <bug-2465-206@http.llvm.org/bugs/>
Message-ID: <200806172012.m5HKCJ4n022003@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2465


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

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




--- Comment #4 from Evan Cheng <evan.cheng at apple.com>  2008-06-17 15:12:19 ---
Fixed. Now the BB looks like this:

LBB5_811:       # bb27.i.i74
        movl    208(%esp), %ecx
        movb    (%ecx,%ebp), %al
        movzbl  %al, %ecx
        movl    %ecx, %edx
        andl    $1, %edx
        negl    %ecx
        xorl    $1, %edx
        movl    204(%esp), %ebx
        cmpb    (%ebx,%ecx), %al
        jne     LBB5_813        # bb14.i.i.snake.exit.i_crit_edge

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063792.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 Jun 17 16:05:26 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 16:05:26 -0500
Subject: [LLVMbugs] [Bug 2470] New: singlely-initialized value not optimized
Message-ID: <bug-2470-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2470

           Summary: singlely-initialized value not optimized
           Product: libraries
           Version: trunk
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


This is a refinement of PR2436:

  extern char *getPtr();

  char f(int i, int j) {
    char *arr[10];

    arr[i] = getPtr();
    arr[i][42] = 123;

    return arr[j][42];
  }

The values of 'j' and 'i' don't actually matter; if they're not equal, the
result is undefined, if they are equal, the function returns 123. The store of
123 can't be eliminated though, just the load of arr[j].


-- 
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 Jun 17 18:03:04 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 17 Jun 2008 18:03:04 -0500
Subject: [LLVMbugs] [Bug 2455] loop unswitch dom bug
In-Reply-To: <bug-2455-206@http.llvm.org/bugs/>
Message-ID: <200806172303.m5HN34Ro027995@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2455


Eli Friedman <sharparrow1 at yahoo.com> changed:

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




--- Comment #6 from Eli Friedman <sharparrow1 at yahoo.com>  2008-06-17 18:03:04 ---
What?  There's no way a patch for IPConstantPropagation could possibly fix this
bug.


-- 
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 Jun 18 04:04:26 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 18 Jun 2008 04:04:26 -0500
Subject: [LLVMbugs] [Bug 2471] New: Illegal instcombine for load of address
	from select
Message-ID: <bug-2471-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2471

           Summary: Illegal instcombine for load of address from select
           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


Consider the following testcase:
declare i32 @x(i32*)
define i32 @b(i32* %a, i32* %b) {
entry:
        %tmp1 = load i32* %a            ; <i32> [#uses=1]
        %tmp3 = load i32* %b            ; <i32> [#uses=1]
        %add = add i32 %tmp1, %tmp3             ; <i32> [#uses=1]
        %call = call i32 @x( i32* %a )          ; <i32> [#uses=0]
        %tobool = icmp ne i32 %add, 0           ; <i1> [#uses=1]
        %cond = select i1 %tobool, i32* %b, i32* %a             ; <i32*>
[#uses=1]
        %tmp8 = load i32* %cond         ; <i32> [#uses=1]
        ret i32 %tmp8
}

Instcombine transforms this to the following:
declare i32 @x(i32*)
define i32 @b(i32* %a, i32* %b) {
entry:
        %tmp1 = load i32* %a            ; <i32> [#uses=1]
        %tmp3 = load i32* %b            ; <i32> [#uses=1]
        %call = call i32 @x( i32* %a )          ; <i32> [#uses=0]
        %add = sub i32 0, %tmp3         ; <i32> [#uses=1]
        %tobool = icmp ne i32 %tmp1, %add               ; <i1> [#uses=1]
        %b.val = load i32* %b           ; <i32> [#uses=1]
        %a.val = load i32* %a           ; <i32> [#uses=1]
        %tmp8 = select i1 %tobool, i32 %b.val, i32 %a.val               ; <i32>
[#uses=1]
        ret i32 %tmp8
}

The extra loads instcombine has inserted are dangerous; for example, suppose %a
was allocated by malloc(), and @x calls free() on %a.

isSafeToLoadUnconditionally is being too aggressive here.


-- 
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 Jun 18 18:43:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 18 Jun 2008 18:43:39 -0500
Subject: [LLVMbugs] [Bug 2472] New: Missing VZEXT_MOVL pattern for <4 x i32>
Message-ID: <bug-2472-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2472

           Summary: Missing VZEXT_MOVL pattern for <4 x i32>
           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


Testcase:
define <4 x i32> @a(<4 x i32> %a) {
entry:
        %vecext = extractelement <4 x i32> %a, i32 0
        insertelement <4 x i32> zeroinitializer, i32 %vecext, i32 0
        %add = add <4 x i32> %a, %0
        ret <4 x i32> %add
}

Run through llvm-as | llc, prints the following and aborts:
Cannot yet select: 0x8a470c8: v4i32 = X86ISD::VZEXT_MOVL 0x8a473d8

It looks like general patterns are also missing for v8i16 and v16i8; I'm not
completely sure if those are supposed to be legal, though.


-- 
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 Jun 18 19:26:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 18 Jun 2008 19:26:19 -0500
Subject: [LLVMbugs] [Bug 2473] New: -ipsccp miscompilation
Message-ID: <bug-2473-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2473

           Summary: -ipsccp miscompilation
           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: torvald at se.inf.tu-dresden.de
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1756)
 --> (http://llvm.org/bugs/attachment.cgi?id=1756)
bugpoint log output, input files, .bc's

ipsccp seems to cause bad code generation. The program (KMeans from
http://stamp.stanford.edu + Tanger transformations (tinystm.org)) works when
opts are disabled for the final linking step, but doesn't if standard opts are
not disabled. Result is a segfault in the final output. Valgrind sees the fault
with opts, but doesn't see any issue without the opts.

Bugpoint log output attached. Cmd line:
bugpoint kmeans-bugpoint.bc -Xlinker='-lpthread' -run-llc --output test
-abs-tolerance=1 -internalize -ipsccp -globalopt -constmerge -deadargelim
-instcombine -basiccg -inline -prune-eh -globalopt -globaldce -basiccg
-argpromotion -instcombine -jump-threading -domtree -domfrontier -scalarrepl
-basiccg -globalsmodref-aa -domtree -loops -loopsimplify -domfrontier
-scalar-evolution -licm -memdep -gvn -memcpyopt -dse -instcombine
-jump-threading -simplifycfg -globaldce -instcombine -simplifycfg -dce
-globaldce -preverify -domtree -verify --args -- -m40 -n40 -t0.00001 -i
inputs/random-n2048-d16-c16.txt -p 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  Thu Jun 19 17:56:36 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 19 Jun 2008 17:56:36 -0500
Subject: [LLVMbugs] [Bug 2474] New: PATCH: ExecutionEngine::
 addGlobalMapping is missing from the C bindings
Message-ID: <bug-2474-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2474

           Summary: PATCH: ExecutionEngine::addGlobalMapping is missing from
                    the C bindings
           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: yann-nicolas.dauphin at polymtl.ca
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1758)
 --> (http://llvm.org/bugs/attachment.cgi?id=1758)
This patch adds ExecutionEngine::addGlobalMapping to the C bindings

ExecutionEngine::addGlobalMapping - without which you can't use LLVM in a VM -
is missing from the C bindings.
I included a patch that adds it.

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  Thu Jun 19 21:20:11 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 19 Jun 2008 21:20:11 -0500
Subject: [LLVMbugs] [Bug 2474] PATCH: ExecutionEngine:: addGlobalMapping is
 missing from the C bindings
In-Reply-To: <bug-2474-206@http.llvm.org/bugs/>
Message-ID: <200806200220.m5K2KBL6018968@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2474


Gordon Henriksen <gordonhenriksen at mac.com> changed:

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




--- Comment #1 from Gordon Henriksen <gordonhenriksen at mac.com>  2008-06-19 21:20:10 ---
Thanks, applied:

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063904.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 Jun 20 00:17:54 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 20 Jun 2008 00:17:54 -0500
Subject: [LLVMbugs] [Bug 2471] Illegal instcombine for load of address from
	select
In-Reply-To: <bug-2471-206@http.llvm.org/bugs/>
Message-ID: <200806200517.m5K5HsTR023867@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2471


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-06-20 00:17:54 ---
Nice catch, fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063907.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 Jun 20 00:33:43 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 20 Jun 2008 00:33:43 -0500
Subject: [LLVMbugs] [Bug 2452] Some tests generate GCC warnings
In-Reply-To: <bug-2452-206@http.llvm.org/bugs/>
Message-ID: <200806200533.m5K5Xh6e024389@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2452


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #4 from Chris Lattner <sabre at nondot.org>  2008-06-20 00:33:42 ---
The remaining 3 xpass's should be fixed 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  Fri Jun 20 06:41:02 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 20 Jun 2008 06:41:02 -0500
Subject: [LLVMbugs] [Bug 2475] New: This is a compiling error
Message-ID: <bug-2475-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2475

           Summary: This is a compiling error
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-g++
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: yonggangluo at hotmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1760)
 --> (http://llvm.org/bugs/attachment.cgi?id=1760)
This a patch for llvm-gcc4.2/gcc/genmodes.c

When I compile the trunk version of llvm-gcc4.2 , My computer get crashed,
My computer is using Windows Vista, and so when my computer to compile the file
in /trunk/gcc/genmodes.c; and run genmodes.exe to generate source files. The
computer were crashed, because genomdes.exe generate a very hudge file. I found
why this happened, and I revised it, and I submit a diff file for this 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  Fri Jun 20 08:15:17 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 20 Jun 2008 08:15:17 -0500
Subject: [LLVMbugs] [Bug 2475] This is a compiling error
In-Reply-To: <bug-2475-206@http.llvm.org/bugs/>
Message-ID: <200806201315.m5KDFHud018812@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2475


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

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




--- Comment #1 from Anton Korobeynikov <asl at math.spbu.ru>  2008-06-20 08:15:16 ---


*** This bug has been marked as a duplicate of bug 2428 ***


-- 
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 Jun 20 11:12:56 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 20 Jun 2008 11:12:56 -0500
Subject: [LLVMbugs] [Bug 2476] New: assertion when custom expanding ISD::ADD
	or ISD::SUB
Message-ID: <bug-2476-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2476

           Summary: assertion when custom expanding ISD::ADD or ISD::SUB
           Product: libraries
           Version: 2.3
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rlsosborne at gmail.com
                CC: llvmbugs at cs.uiuc.edu


In my target I have 32bit registers and must custom expand 64bit adds and subs.
I soon as I do this I hit an assertion in the ExpandOp in LegalizeDAG.cpp:

llc:
/homelocal/richard/llvm/tools_llvm/src/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp:6724:
void<unnamed>::SelectionDAGLegalize::ExpandOp(llvm::SDOperand,
llvm::SDOperand&, llvm::SDOperand&): Assertion `isNew && "Value already
expanded?!?"' failed.
llc[0xe358c6]
llc[0xe359eb]
/lib64/tls/libc.so.6[0x356b22e2f0]
/lib64/tls/libc.so.6(gsignal+0x3d)[0x356b22e25d]
/lib64/tls/libc.so.6(abort+0xfe)[0x356b22fa5e]
/lib64/tls/libc.so.6(__assert_fail+0xf1)[0x356b227ae1]
llc[0xbcdf27]
llc[0xbd560c]
llc[0xbd5990]
llc(_ZN4llvm12SelectionDAG8LegalizeEv+0x4b)[0xbd5adf]
llc(_ZN4llvm16SelectionDAGISel17CodeGenAndEmitDAGERNS_12SelectionDAGE+0xd1)[0xb5962d]
llc(_ZN4llvm16SelectionDAGISel16SelectBasicBlockEPNS_10BasicBlockERNS_15MachineFunctionERNS_20FunctionLoweringInfoE+0xb5)[0xb70b15]
llc(_ZN4llvm16SelectionDAGISel13runOnFunctionERNS_8FunctionE+0x20b)[0xb7295d]
llc(_ZN4llvm13FPPassManager13runOnFunctionERNS_8FunctionE+0x127)[0xdc9305]
llc(_ZN4llvm23FunctionPassManagerImpl3runERNS_8FunctionE+0x74)[0xdc952e]
llc(_ZN4llvm19FunctionPassManager3runERNS_8FunctionE+0x80)[0xdc968c]
llc(main+0x8a8)[0x798d2e]
/lib64/tls/libc.so.6(__libc_start_main+0xdb)[0x356b21c3fb]
llc(atan+0xfa)[0x79780a]
Aborted

The problem seem to be that in the case of ISD::ADD and ISD::SUB Op is set to
the value returned by TLI.LowerOperation which then has ExpandOp called on it.
Instead Op should be left as the original 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 mstrout at CS.ColoState.EDU  Fri Jun 20 14:50:12 2008
From: mstrout at CS.ColoState.EDU (Michelle Strout)
Date: Fri, 20 Jun 2008 15:50:12 -0400
Subject: [LLVMbugs] Problems compiling poolalloc
Message-ID: <03889E2A-18B2-49A5-AB39-3151D99BCE36@CS.ColoState.EDU>

Chris and others,

To get the poolalloc module to compile, I had to copy some llvm  
include files to different names.

         % cd /Users/mstrout/software/include/llvm/ADT/
         % cp hash_map hash_map.h
         % cp hash_set hash_set.h
         % cp iterator iterator.h

Does this mean that poolalloc is not being updated?  What is its status?

Thanks,
Michelle

=======================================
  Michelle Mills Strout
  Assistant Professor

  Colorado State University
  Computer Science Department
  1873 Campus Delivery
  Fort Collins, CO 80523-1873

  (970) 491-7026
  mstrout at cs.colostate.edu
=======================================





From sabre at nondot.org  Fri Jun 20 15:10:19 2008
From: sabre at nondot.org (Chris Lattner)
Date: Fri, 20 Jun 2008 13:10:19 -0700 (PDT)
Subject: [LLVMbugs] Problems compiling poolalloc
In-Reply-To: <03889E2A-18B2-49A5-AB39-3151D99BCE36@CS.ColoState.EDU>
References: <03889E2A-18B2-49A5-AB39-3151D99BCE36@CS.ColoState.EDU>
Message-ID: <Pine.LNX.4.62.0806201309500.18920@nondot.org>

On Fri, 20 Jun 2008, Michelle Strout wrote:
> Chris and others,
>
> To get the poolalloc module to compile, I had to copy some llvm
> include files to different names.

Hi Michelle,

I don't actively hack on poolalloc any longer, but I believe Vikram's 
group is still doing work with it,

-Chris

>         % cd /Users/mstrout/software/include/llvm/ADT/
>         % cp hash_map hash_map.h
>         % cp hash_set hash_set.h
>         % cp iterator iterator.h
>
> Does this mean that poolalloc is not being updated?  What is its status?
>
> Thanks,
> Michelle
>
> =======================================
>  Michelle Mills Strout
>  Assistant Professor
>
>  Colorado State University
>  Computer Science Department
>  1873 Campus Delivery
>  Fort Collins, CO 80523-1873
>
>  (970) 491-7026
>  mstrout at cs.colostate.edu
> =======================================
>
>
>
>
> _______________________________________________
> LLVMbugs mailing list
> LLVMbugs at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmbugs
>

-Chris

-- 
http://nondot.org/sabre/
http://llvm.org/

From criswell at uiuc.edu  Fri Jun 20 15:12:30 2008
From: criswell at uiuc.edu (John Criswell)
Date: Fri, 20 Jun 2008 15:12:30 -0500
Subject: [LLVMbugs] Problems compiling poolalloc
In-Reply-To: <03889E2A-18B2-49A5-AB39-3151D99BCE36@CS.ColoState.EDU>
References: <03889E2A-18B2-49A5-AB39-3151D99BCE36@CS.ColoState.EDU>
Message-ID: <485C0F2E.8060909@uiuc.edu>

Michelle Strout wrote:
> Chris and others,
>
> To get the poolalloc module to compile, I had to copy some llvm
> include files to different names.
>
>          % cd /Users/mstrout/software/include/llvm/ADT/
>          % cp hash_map hash_map.h
>          % cp hash_set hash_set.h
>          % cp iterator iterator.h
>
> Does this mean that poolalloc is not being updated?  What is its status?
>   
We are currently doing active work on llvm-poolalloc; mainline 
llvm-poolalloc should compile with mainline LLVM.

Is your LLVM and llvm-poolalloc trees up to date with everything in the 
Subversion repository?

-- John T.

> Thanks,
> Michelle
>
> =======================================
>   Michelle Mills Strout
>   Assistant Professor
>
>   Colorado State University
>   Computer Science Department
>   1873 Campus Delivery
>   Fort Collins, CO 80523-1873
>
>   (970) 491-7026
>   mstrout at cs.colostate.edu
> =======================================
>
>
>
>
> _______________________________________________
> LLVMbugs mailing list
> LLVMbugs at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmbugs
>   


From bugzilla-daemon at cs.uiuc.edu  Fri Jun 20 15:16:56 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 20 Jun 2008 15:16:56 -0500
Subject: [LLVMbugs] [Bug 2477] New: clang misparses "//*" in C89 mode
Message-ID: <bug-2477-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2477

           Summary: clang misparses "//*" in C89 mode
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: preprocessor
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Testcase:
int a(int x) {return x//**/;
+x;}

In C99/GNU/C++ modes, this should return x+x.  However, in C89 mode, this
should return x.  clang currently gets this wrong in C89 mode.

It would probably be a good idea to warn for this construct.


-- 
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 mstrout at CS.ColoState.EDU  Fri Jun 20 15:29:01 2008
From: mstrout at CS.ColoState.EDU (Michelle Strout)
Date: Fri, 20 Jun 2008 16:29:01 -0400
Subject: [LLVMbugs] Problems compiling poolalloc
In-Reply-To: <485C0F2E.8060909@uiuc.edu>
References: <03889E2A-18B2-49A5-AB39-3151D99BCE36@CS.ColoState.EDU>
	<485C0F2E.8060909@uiuc.edu>
Message-ID: <A81B1159-5B14-4CCC-A9D2-03DE99676FDC@CS.ColoState.EDU>

John,

Thanks for the swift response.

I checked poolalloc out from the repository, and I am compiling it  
against the llvm-2.3 source distribution.  I am doing this all on a  
Mac running 10.4.  I tried a number of different things before I  
stooped to copying header files, but here was my final set of actions.

         % mkdir /tmp/llvm-2.3
         % cd /tmp/llvm-2.3
         % ./configure --prefix=/Users/mstrout/software/  \
         --with-objdir=/tmp/llvm-2.3/ \
         --with-llvmgccdir=/Users/mstrout/package/llvm-gcc4.2-2.3-x86- 
darwin8/bin
         % gmake -k | & tee gnumake.out
	% make install

	% svn co http://llvm.org/svn/llvm-project/poolalloc/trunk poolalloc
	

	% cd /Users/mstrout/software/include/llvm/ADT/
         % cp hash_map hash_map.h
         % cp hash_set hash_set.h
         % cp iterator iterator.h


         % cd ~/SVNWorkDirs/poolalloc
         % ./configure --with-llvmsrc=/Users/mstrout/package/ 
llvm-2.3/ --with-llvmobj=/tmp/llvm-2.3/ --prefix=/Users/mstrout/ 
software/

         % make

I got it to compile, but it took awhile.


Thanks,
Michelle

On Jun 20, 2008, at 4:12 PM, John Criswell wrote:

> Michelle Strout wrote:
>> Chris and others,
>>
>> To get the poolalloc module to compile, I had to copy some llvm
>> include files to different names.
>>
>>          % cd /Users/mstrout/software/include/llvm/ADT/
>>          % cp hash_map hash_map.h
>>          % cp hash_set hash_set.h
>>          % cp iterator iterator.h
>>
>> Does this mean that poolalloc is not being updated?  What is its  
>> status?
>>
> We are currently doing active work on llvm-poolalloc; mainline llvm- 
> poolalloc should compile with mainline LLVM.
>
> Is your LLVM and llvm-poolalloc trees up to date with everything in  
> the Subversion repository?
>
> -- John T.
>
>> Thanks,
>> Michelle
>>
>> =======================================
>>   Michelle Mills Strout
>>   Assistant Professor
>>
>>   Colorado State University
>>   Computer Science Department
>>   1873 Campus Delivery
>>   Fort Collins, CO 80523-1873
>>
>>   (970) 491-7026
>>   mstrout at cs.colostate.edu
>> =======================================
>>
>>
>>
>>
>> _______________________________________________
>> LLVMbugs mailing list
>> LLVMbugs at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmbugs
>>
>

=======================================
  Michelle Mills Strout
  Assistant Professor

  Colorado State University
  Computer Science Department
  1873 Campus Delivery
  Fort Collins, CO 80523-1873

  (970) 491-7026
  mstrout at cs.colostate.edu
=======================================





From bugzilla-daemon at cs.uiuc.edu  Fri Jun 20 15:36:51 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 20 Jun 2008 15:36:51 -0500
Subject: [LLVMbugs] [Bug 2478] New: -O0 compile time: stop llvm-gcc from
 making llvm values named "tmp"
Message-ID: <bug-2478-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2478

           Summary: -O0 compile time: stop llvm-gcc from making llvm values
                    named "tmp"
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Keywords: slow-compile
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sabre at nondot.org
                CC: llvmbugs at cs.uiuc.edu


Duncan found that the symbol table traffic of pushing around local value named
'tmp' are a significant portion of -O0 compile times.  Disabling this sped
things up a lot.  We should consider doing 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  Sat Jun 21 14:16:40 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 14:16:40 -0500
Subject: [LLVMbugs] [Bug 2479] New: likely integer codegen bug
Message-ID: <bug-2479-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2479

           Summary: likely integer codegen bug
           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


This happens with svn 52513 on ubuntu feisty on ia32.

Unfortunately I have not been able to further reduce the test input.

[regehr at babel tmp21]$ llvm-gcc -O2 -fwrapv small.c -o small  
[regehr at babel tmp21]$ ./small
0
[regehr at babel tmp21]$ llvm-gcc -O3 -fwrapv small.c -o small
[regehr at babel tmp21]$ ./small
1



#include <stdint.h>
#include <limits.h>
#include <stdio.h>

static inline unsigned long int
mod_rhs(const long int rhs)
{
    if (rhs == 0) return 1;
    return rhs;
}

static inline int
rshift_s_s(int left, int right)
{
        if ((left < 0)
                || (right < 0)
                || (right >= sizeof(int)*CHAR_BIT)) {
                /* Avoid implementation-defined and undefined behavior. */
                return left;
        }
        return left >> right;
}

static inline int
lshift_s_s(int left, int right)
{
        if ((left < 0)
                || (right < 0)
                || (right >= sizeof(int)*CHAR_BIT)
                || (left > (INT_MAX >> right))) {
                /* Avoid undefined behavior. */
                return left;
        }
        return left << right;
}

uint32_t g_2;
uint32_t g_3 = 0x00252604L;

int32_t func_10 (int8_t p_12);
int32_t func_10 (int8_t p_12)
{
  int32_t l_14 = 1;
  uint16_t l_26;
  uint32_t l_15 = 1;
  if ((l_14 || l_15) % mod_rhs (1 < p_12))
    for (0; l_26; l_26 -= 1)
      {
      }
  return 1;
}

void func_1 (void);
void func_1 (void)
{
  int8_t l_4 = 0x18L;
  int32_t l_5 = 1;
  uint16_t l_6 = 1;
  uint32_t l_7 = 1;
  uint32_t l_8 = 1;
  g_2 =
    ((((g_3 ^ l_4) <= (1 % mod_rhs (1)) && l_4)
      && (g_3 > (l_6 && 1))) != (((rshift_s_s (l_5, 1))
                                  || (l_7 && g_3) <= (l_6 && g_2)) - ((l_8
                                                                       && l_5)
                                                                      %
                                                                      mod_rhs
                                                                     
(lshift_s_s
                                                                       ((1 <
                                                                         l_6),
                                                                        func_10
                                                                       
(g_3))))));
}

int
main (void)
{
  func_1 ();
  printf ("%d\n", g_2);
  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 Jun 21 15:05:01 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 15:05:01 -0500
Subject: [LLVMbugs] [Bug 2478] -O0 compile time: stop llvm-gcc from making
 llvm values named " tmp"
In-Reply-To: <bug-2478-206@http.llvm.org/bugs/>
Message-ID: <200806212005.m5LK51C5003111@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2478


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #4 from Duncan Sands <baldrick at free.fr>  2008-06-21 15:05:00 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063971.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 Jun 21 15:12:00 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 15:12:00 -0500
Subject: [LLVMbugs] [Bug 2480] New: improve assembler syntax for nameless
	values
Message-ID: <bug-2480-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2480

           Summary: improve assembler syntax for nameless values
           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: baldrick at free.fr
                CC: llvmbugs at cs.uiuc.edu


The assembler is hard to read when values have no names.  Here's what it
looks like currently:

        call i8* @f( i32 0 )            ; <i8*>:0 [#uses=1]
        call void @g( i8* %0 )

How about changing it to this:

        %0 = call i8* @f( i32 0 )            ; <i8*> [#uses=1]
        call void @g( i8* %0 )

?

So in the assembler %0 would act like a name for the purpose
of assembling, but internally no name would be assigned to the
value (so the value might be written out to assembler later as
%1 or some other number).


-- 
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 Jun 21 15:20:15 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 15:20:15 -0500
Subject: [LLVMbugs] [Bug 2470] singlely-initialized value not optimized
In-Reply-To: <bug-2470-206@http.llvm.org/bugs/>
Message-ID: <200806212020.m5LKKFsk003621@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2470


Nick Lewycky <nicholas at mxc.ca> changed:

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




--- Comment #6 from Nick Lewycky <nicholas at mxc.ca>  2008-06-21 15:20:14 ---
I contend that we don't need to know that the call to operator new actually
allocates memory, just so long as the pointers it returns don't partially
overlap and we know that we perform the store to them afterwards.

But your second point is a very good one I'd completely forgotten. Each member
function gets the this pointer which could be used to modify the vptr, short of
field-sensitive interprocedural alias analysis proving otherwise.


-- 
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 Jun 21 15:57:20 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 15:57:20 -0500
Subject: [LLVMbugs] [Bug 2481] New: . s file contains __eh_frame even if
 building with -fno-exceptions
Message-ID: <bug-2481-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2481

           Summary: .s file contains __eh_frame even if building with -fno-
                    exceptions
           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: sabre at nondot.org
                CC: llvmbugs at cs.uiuc.edu


When building instcombine with llvm-gcc -O0 -g -fno-exceptions, I get a large
eh_frame section, this is bad.


-- 
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 Jun 21 15:59:05 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 15:59:05 -0500
Subject: [LLVMbugs] [Bug 2482] New: llvm-gcc emits twice as much debug info
 when compiling instcombine at -O0 -g than GCC
Message-ID: <bug-2482-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2482

           Summary: llvm-gcc emits twice as much debug info when compiling
                    instcombine at -O0 -g than GCC
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Keywords: slow-compile
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sabre at nondot.org
                CC: llvmbugs at cs.uiuc.edu


When compiling instcombine at -O0 -g with llvm-g++ we get almost twice as much
debug info as GCC produces:

gcc 4.2:
Section (__DWARF, __debug_info): 671478
Section (__DWARF, __debug_line): 126106

llvm-gcc 4.2:
Section (__DWARF, __debug_info): 1240778
Section (__DWARF, __debug_line): 149797


-- 
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 Jun 21 16:01:36 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 16:01:36 -0500
Subject: [LLVMbugs] [Bug 2483] New: llvm-gcc doesn' t take advantage of
 __const_coal section on darwin
Message-ID: <bug-2483-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2483

           Summary: llvm-gcc doesn't take advantage of __const_coal section
                    on darwin
           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: sabre at nondot.org
                CC: llvmbugs at cs.uiuc.edu


Building instcombine at -O0 -g with llvm-gcc, nothing is emitted into the
__const_coal section.  With GCC, it emits:

        Section (__DATA, __const_coal): 48
        Section (__TEXT, __const_coal): 128

I don't know what this section is, but if useful, we should use 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  Sat Jun 21 19:23:58 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 19:23:58 -0500
Subject: [LLVMbugs] [Bug 2479] likely integer codegen bug
In-Reply-To: <bug-2479-206@http.llvm.org/bugs/>
Message-ID: <200806220023.m5M0NwZd011851@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2479


Eli Friedman <sharparrow1 at yahoo.com> changed:

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




--- Comment #5 from Eli Friedman <sharparrow1 at yahoo.com>  2008-06-21 19:23:58 ---
Fixed.

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063983.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 Jun 21 22:15:51 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 21 Jun 2008 22:15:51 -0500
Subject: [LLVMbugs] [Bug 2418] CBE produces C file with duplicate symbols
In-Reply-To: <bug-2418-206@http.llvm.org/bugs/>
Message-ID: <200806220315.m5M3FpKo017234@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2418


Cristiano Calcagno <ccris at doc.ic.ac.uk> changed:

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




--- Comment #8 from Cristiano Calcagno <ccris at doc.ic.ac.uk>  2008-06-21 22:15:50 ---
The fix was removed in commit 52071, as it was breaking something else.
As far as I can see, the problem is still there, at least using precompiled
llvm-gcc4.2-2.3-x86-darwin9.
Don't know about the svn version of 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 Jun 22 02:51:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 22 Jun 2008 02:51:28 -0500
Subject: [LLVMbugs] [Bug 2484] New: <4 x float> SSE shuffle breaks without
	SSE2
Message-ID: <bug-2484-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2484

           Summary: <4 x float> SSE shuffle breaks without SSE2
           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


Testcase:
define <4 x float> @f4523(<4 x float> %a,<4 x float> %b) nounwind {
entry:
%shuffle = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4,i32
5,i32 2,i32 3>
ret <4 x float> %shuffle
}

When compiled with llvm-as | llc -mattr=-sse2, llc aborts with the following
error:
Cannot yet select: 0x8a47420: v4f32 = vector_shuffle 0x8a472b0, 0x8a47358,
0x8a471a0<4,5,2,3>

Not particularly important, but I ran into it while testing some shuffles.


-- 
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 Jun 22 03:09:04 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 22 Jun 2008 03:09:04 -0500
Subject: [LLVMbugs] [Bug 2485] New: Do all 4-element SSE shuffles in a
 maximum of two shuffle instructions
Message-ID: <bug-2485-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2485

           Summary: Do all 4-element SSE shuffles in a maximum of two
                    shuffle instructions
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Currently, shuffles for <4 x i32> and <4 x float> SSE/SSE2 vectors end up
falling back in the general case to doing three shuffle operations.  This is
unnecessary; all shuffles can be done with at most two shufps:

Suppose the elements all come from one vector. Then the result can be
calculated with a single shufps.  (This logic already exists in the code.)

Suppose there are at most two distinct elements that come from each source
vector.  Then we can use a shufps to make an intermediate vector with all the
elements we need in the final result.  This vector can then be rearranged with
a shufps to the final result.  (This logic already exists, but it doesn't
actually catch all the relevant cases.)

Otherwise, we must have three elements from one vector, call it X, and one
element from the other, call it Y.  First, use a shufps to build an
intermediate vector with the one element from Y and the element from X that
will be in the same half in the final destination (the indexes don't matter).
Then, use a shufps to build the final vector, taking the half containing the
element from Y from the intermediate, and the other half from X.


This might be something to stick into X86/README.txt, but it seems simple
enough to implement that it really should be just done.


-- 
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 Jun 22 04:43:25 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 22 Jun 2008 04:43:25 -0500
Subject: [LLVMbugs] [Bug 2476] assertion when custom expanding ISD::ADD or
	ISD::SUB
In-Reply-To: <bug-2476-206@http.llvm.org/bugs/>
Message-ID: <200806220943.m5M9hPpH000806@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2476


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #6 from Duncan Sands <baldrick at free.fr>  2008-06-22 04:43:24 ---
This is already handled correctly by the new type legalization
infrastructure.  While waiting for it to be turned on I have
applied your patch here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063986.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 Jun 22 12:46:53 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 22 Jun 2008 12:46:53 -0500
Subject: [LLVMbugs] [Bug 2369] scalarrepl overly aggressive with large arrays
In-Reply-To: <bug-2369-206@http.llvm.org/bugs/>
Message-ID: <200806221746.m5MHkroN019178@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2369


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #2 from Chris Lattner <sabre at nondot.org>  2008-06-22 12:46:52 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080616/063987.html

Great catch!


-- 
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 Jun 22 19:53:47 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 22 Jun 2008 19:53:47 -0500
Subject: [LLVMbugs] [Bug 2486] New: llvm incorrectly inlines functions
	containing setjmp
Message-ID: <bug-2486-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2486

           Summary: llvm incorrectly inlines functions containing setjmp
           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


Testcase:

#include <setjmp.h>      
#include <stdio.h>
jmp_buf g;
static int a() {longjmp(g,1);}
static void b(int* x) {if (setjmp(g))return;*x = 10;a();}
int main() {int a = 1; b(&a); printf("%d\n", a); return 0;}

This program should print out "10".

Output with llvm-gcc -O3 -emit-llvm -S:

target datalayout =
"e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32"
target triple = "i386-pc-linux-gnu"
        %struct.__jmp_buf_tag = type { [6 x i32], i32, %struct.__sigset_t }
        %struct.__sigset_t = type { [32 x i32] }
@g = common global [1 x %struct.__jmp_buf_tag] zeroinitializer, align 32       
; <[1 x %struct.__jmp_buf_tag]*> [#uses=1]
@.str = internal constant [4 x i8] c"%d\0A\00"          ; <[4 x i8]*> [#uses=1]

declare i32 @_setjmp(%struct.__jmp_buf_tag*) nounwind 

declare void @longjmp(%struct.__jmp_buf_tag*, i32) noreturn nounwind 

define i32 @main() nounwind  {
entry:
        tail call i32 @_setjmp( %struct.__jmp_buf_tag* getelementptr ([1 x
%struct.__jmp_buf_tag]* @g, i32 0, i32 0) ) nounwind                 ; <i32>:0
[#uses=1]
        icmp eq i32 %0, 0               ; <i1>:1 [#uses=1]
        br i1 %1, label %bb.i, label %b.exit

bb.i:           ; preds = %entry
        tail call void @longjmp( %struct.__jmp_buf_tag* getelementptr ([1 x
%struct.__jmp_buf_tag]* @g, i32 0, i32 0), i32 1 ) noreturn nounwind 
        unreachable

b.exit:         ; preds = %entry
        tail call i32 (i8*, ...)* @printf( i8* noalias  getelementptr ([4 x
i8]* @.str, i32 0, i32 0), i32 1 ) nounwind                 ; <i32>:2 [#uses=0]
        ret i32 0
}

declare i32 @printf(i8*, ...) nounwind 

As should be obvious, this program prints out "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  Sun Jun 22 22:06:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 22 Jun 2008 22:06:48 -0500
Subject: [LLVMbugs] [Bug 2487] New: loop codegen bug
Message-ID: <bug-2487-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2487

           Summary: loop codegen bug
           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


This is for svn 52617 targeting ia32 hosted on Ubuntu Hardy.

[regehr at babel tmp25]$ llvm-gcc -Wall -fwrapv -O0 small.c -o small
[regehr at babel tmp25]$ ./small
1
[regehr at babel tmp25]$ llvm-gcc -Wall -fwrapv -O1 small.c -o small
[regehr at babel tmp25]$ ./small
0


int g_6;
void func_1 (void);
void func_1 (void)
{
  char l_3;
  for (l_3 = 0; l_3 >= 0; ++l_3) {
    if (!l_3) {
      for (g_6 = 1; 0; --g_6);
    }
  }
}

int main (void)
{
  func_1 ();
  printf ("%d\n", g_6);
  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  Mon Jun 23 16:30:56 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 23 Jun 2008 16:30:56 -0500
Subject: [LLVMbugs] [Bug 2253] loop unroll crashes on example with
	-unroll-count=2
In-Reply-To: <bug-2253-206@http.llvm.org/bugs/>
Message-ID: <200806232130.m5NLUumi024485@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2253


Dan Gohman <gohman at apple.com> changed:

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




--- Comment #12 from Dan Gohman <gohman at apple.com>  2008-06-23 16:30:56 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080623/064041.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 Jun 24 08:38:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 24 Jun 2008 08:38:39 -0500
Subject: [LLVMbugs] [Bug 2488] New: instcombine removes stackrestore after
	alloca
Message-ID: <bug-2488-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2488

           Summary: instcombine removes stackrestore after alloca
           Product: libraries
           Version: 2.3
          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


Created an attachment (id=1769)
 --> (http://llvm.org/bugs/attachment.cgi?id=1769)
test case

The attached test case is based on gcc.c-torture/execute/20040811-1.c in the
GCC 4.2.1 testsuite (I just compiled it with -O1). It has a loop containing a
variable-sized alloca, with a call to llvm.stacksave before the alloca and
llvm.stackrestore after it.

If I run it through "opt -f -o post.o pre.o -instcombine" I get this:

bb44:           ; preds = %bb44, %entry
        %indvar = phi i32 [ 0, %entry ], [ %tmp3857, %bb44 ]            ; <i32>
[#uses=1]
        %tmp3857 = add i32 %indvar, 1           ; <i32> [#uses=3]
        %tmp2 = call i8* @llvm.stacksave( )             ; <i8*> [#uses=0]
        %tmp4 = srem i32 %tmp3857, 1000         ; <i32> [#uses=2]
        %tmp5 = add i32 %tmp4, 1                ; <i32> [#uses=1]
        %tmp27 = alloca i32, i32 %tmp5          ; <i32*> [#uses=3]
        %tmpcast = bitcast i32* %tmp27 to i8*           ; <i8*> [#uses=1]
        store i32 1, i32* %tmp27, align 4
        %tmp34 = getelementptr i32* %tmp27, i32 %tmp4           ; <i32*>
[#uses=1]
        store i32 2, i32* %tmp34, align 4
        volatile store i8* %tmpcast, i8** @p, align 4
        %exitcond = icmp eq i32 %tmp3857, 999999                ; <i1>
[#uses=1]
        br i1 %exitcond, label %bb, label %bb44

It has completely removed the call to llvm.stackrestore!

(See also bug 1851, where instcombine used to remove both the stacksave and the
stackrestore from this test case.)


-- 
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 Jun 25 01:00:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 01:00:09 -0500
Subject: [LLVMbugs] [Bug 2488] instcombine removes stackrestore after alloca
In-Reply-To: <bug-2488-206@http.llvm.org/bugs/>
Message-ID: <200806250600.m5P609CT018003@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2488


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-06-25 01:00:08 ---
Nice catch and nice reduction, fix here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080623/064123.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 Jun 25 03:56:17 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 03:56:17 -0500
Subject: [LLVMbugs] [Bug 2489] New: clang -g segfaults on functions with
	arguments
Message-ID: <bug-2489-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2489

           Summary: clang -g segfaults on functions with arguments
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


It seems that clang debug output (clang -g) breaks on any function that has
arguments.

The problem is easily illustrated using the following trivial foo.c:

  void foo(int blaa) { }

kooijman at eris:~/src/montiumccfe$ gdb --args clang -g -emit-llvm foo.c 
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1211607360 (LWP 19210)]
0x083c4765 in clang::CodeGen::CGDebugInfo::EmitDeclare (this=0x8773d00,
decl=0x8768330, Tag=257, AI=0x877891c, Builder=@0xbfb25ab8) at
CGDebugInfo.cpp:730
730       Variable->setContext(RegionStack.back());
(gdb) bt
#0  0x083c4765 in clang::CodeGen::CGDebugInfo::EmitDeclare (this=0x8773d00,
decl=0x8768330, Tag=257, AI=0x877891c, Builder=@0xbfb25ab8) at
CGDebugInfo.cpp:730
#1  0x083de65a in clang::CodeGen::CodeGenFunction::EmitParmDecl
(this=0xbfb25ab0, D=@0x8768330, Arg=0x87788a8) at CGDecl.cpp:229
#2  0x083dd07a in clang::CodeGen::CodeGenFunction::GenerateCode
(this=0xbfb25ab0, FD=0x8770e88) at CodeGenFunction.cpp:123
#3  0x083b1f07 in clang::CodeGen::CodeGenModule::EmitFunction (this=0x8771938,
FD=0x8770e88) at CodeGenModule.cpp:525
#4  0x083add62 in (anonymous namespace)::CodeGenerator::HandleTopLevelDecl
(this=0x8768b18, D=0x8770e88) at ModuleBuilder.cpp:69
#5  0x0846dc81 in clang::ParseAST (PP=@0x8756f98, Consumer=0x8768b18,
PrintStats=false) at ParseAST.cpp:63
#6  0x083a1572 in ProcessInputFile (PP=@0x8756f98, PPF=@0xbfb261c0,
InFile=@0x8756638) at clang.cpp:1305
#7  0x083a1f7b in main (argc=Cannot access memory at address 0xfffffffc
) at clang.cpp:1527
(gdb) p RegionStack.size()
$1 = 0

It seems that the declarations for the parameters are emitted before the
function start is emitted, since RegionStack is still empty (which causes the
problem). This can probably be fixed by emitting debug info for the parameter
declarations just after emitting the function start, instead of in
EmitParmDecl.


-- 
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 Jun 25 04:15:33 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 04:15:33 -0500
Subject: [LLVMbugs] [Bug 2490] New: clang does not support
 __attribute__((annotate)) on functions
Message-ID: <bug-2490-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2490

           Summary: clang does not support __attribute__((annotate)) on
                    functions
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Basic
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


clang does not support the annotate attribute on functions, only variables.
Though this is consistent with gcc documentation and gcc itself, llvm-gcc does
support annotation on functions. I think it should be useful to make clang do
the same.


-- 
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 Jun 25 04:28:52 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 04:28:52 -0500
Subject: [LLVMbugs] [Bug 2491] New: valgrind errors in FixBaseClassFields()
Message-ID: <bug-2491-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2491

           Summary: valgrind errors in FixBaseClassFields()
           Product: tools
           Version: 2.2
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-g++
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jay.foad at antixlabs.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1773)
 --> (http://llvm.org/bugs/attachment.cgi?id=1773)
valgrind log

On a particular C++ source file, I'm seeing a bunch of valgrind errors like
this (full log attached):

$ valgrind
~/llvm/llvm-gcc4.2-2.2-x86-linux-RHEL4/libexec/gcc/i686-pc-linux-gnu/4.2.1/cc1plus
-quiet -emit-llvm-bc -o foo.o foo.ii
...
==23065== Invalid read of size 4
==23065==    at 0x837D82D: FixBaseClassFields(tree_node*) (DenseMap.h:38)
==23065==    by 0x838205E: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1990)
==23065==    by 0x8380ACF: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:718)
==23065==    by 0x8380AA5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:920)
==23065==    by 0x83815E6: TypeConverter::DecodeStructFields(tree_node*,
StructTypeConversionInfo&) (llvm-types.cpp:1802)
==23065==    by 0x838207A: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1994)
==23065==    by 0x8380ACF: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:718)
==23065==    by 0x8380949: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:828)
==23065==    by 0x8382287: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:2100)
==23065==    by 0x8380ACF: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:718)
==23065==    by 0x8380C4F: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:866)
==23065==    by 0x8383476: TypeConverter::ConvertFunctionType(tree_node*,
tree_node*, tree_node*, unsigned&, llvm::ParamAttrsList const*&)
(Casting.h:115)
==23065==  Address 0x42BF49C is 236 bytes inside a block of size 512 free'd
==23065==    at 0x401C93E: operator delete[](void*) (vg_replace_malloc.c:256)
==23065==    by 0x837D81A: FixBaseClassFields(tree_node*) (DenseMap.h:360)
==23065==    by 0x838205E: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1990)
==23065==    by 0x8380ACF: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:718)
==23065==    by 0x8380AA5: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:920)
==23065==    by 0x83815E6: TypeConverter::DecodeStructFields(tree_node*,
StructTypeConversionInfo&) (llvm-types.cpp:1802)
==23065==    by 0x838207A: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:1994)
==23065==    by 0x8380ACF: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:718)
==23065==    by 0x8380949: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:828)
==23065==    by 0x8382287: TypeConverter::ConvertRECORD(tree_node*, tree_node*)
(llvm-types.cpp:2100)
==23065==    by 0x8380ACF: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:718)
==23065==    by 0x8380C4F: TypeConverter::ConvertType(tree_node*)
(llvm-types.cpp:866)

The problem seems to have gone away in LLVM 2.3. But I thought I'd raise this
bug anyway in case the problem has just moved, not gone away completely. Or can
anyone assure me that the problem has really been fixed?

Unfortunately I can't attach my source file as-is because it contains
proprietary code. I can have a go at reducing it if that would be useful.


-- 
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 Jun 25 09:26:54 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 09:26:54 -0500
Subject: [LLVMbugs] [Bug 2492] New: Unresolved symbol in ocaml bindings
Message-ID: <bug-2492-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2492

           Summary: Unresolved symbol in ocaml bindings
           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: amanous at softlab.ntua.gr
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1774)
 --> (http://llvm.org/bugs/attachment.cgi?id=1774)
Add has_initializer to ocaml binding

While trying to link an ocaml program with llvm I encountered a problem with an
unresolved symbol "llvm_has_initializer". It turns out that indeed this symbol
is not defined anywhere.

This code exists in llvm.mli:

external has_initializer : llvalue -> bool = "llvm_has_initializer"

And the correspinding binding in llvm.ml, but the initializer is nowhere to be
found in C++ code.

I believe the missing glue code is something like the one included in the
attached patch. I am not very familiar with LLVM internals, so this might not
be the correct call.


-- 
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 Jun 25 10:26:59 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 10:26:59 -0500
Subject: [LLVMbugs] [Bug 2493] New: lli / JIT bug in KMeans
Message-ID: <bug-2493-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2493

           Summary: lli / JIT bug in KMeans
           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: torvald at se.inf.tu-dresden.de
                CC: llvmbugs at cs.uiuc.edu


I see sth that looks like an lli/JIT bug. The program is the KMeans benchmark
from the STAMP benchmark suite (http://stamp.stanford.edu). 
Overwrite ./common/Defines.common.mk with the attached files to make CC/LD 
customizable.

in ./kmeans, for
./kmeans -m40 -n40 -t0.00001 -i inputs/random-n65536-d32-c16.txt -p 1 :

make -f Makefile.seq clean

CC='llvm-gcc' LD='llvm-gcc -lm' make -f Makefile.seq
-->works, but is slower than gcc (35s vs 16s on my Intel 32b system)

CC='llvm-gcc -emit-llvm' LD='llvm-ld -lm' make -f Makefile.seq
--> does not work (lot's of NaN's in the output, and much faster (11s)) and
uses lli

CC='llvm-gcc -emit-llvm' LD='llvm-ld -lm -disable-opt' make -f Makefile.seq
--> does not work either, and uses lli

CC='llvm-gcc -emit-llvm' LD='llvm-ld -lm -disable-opt -native' make -f --> 
looks as if it would work (and is faster, 16s)

CC='llvm-gcc -emit-llvm' LD='llvm-ld -lm -native' make -f Makefile.seq
--> looks as if it would work (and is faster, 16s)

CC='llvm-gcc -emit-llvm' LD='llvm-ld -link-as-library' make -f Makefile.seq
llc -f -o kmeans.s kmeans
gcc -o kmeans kmeans.s -lm -lpthread
--> looks as if it would work (and is faster, 16s)


-- 
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 Jun 25 10:33:26 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 10:33:26 -0500
Subject: [LLVMbugs] [Bug 2494] New: bad performance vs. gcc in STAMP Kmeans
Message-ID: <bug-2494-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2494

           Summary: bad performance vs. gcc in STAMP Kmeans
           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: torvald at se.inf.tu-dresden.de
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1777)
 --> (http://llvm.org/bugs/attachment.cgi?id=1777)
Defines.common.mk

The program is the KMeans benchmark from the STAMP benchmark suite
(http://stamp.stanford.edu). Overwrite ./common/Defines.common.mk with the
attached file to make CC/LD customizable.

This happens on an Intel Q6600, Linux 32b.
gcc (GCC) 4.2.3 (Debian 4.2.3-5)
LLVM 2.3 and SVN trunk rev 52676


in ./kmeans, for
./kmeans -m40 -n40 -t0.00001 -i inputs/random-n65536-d32-c16.txt -p 1 :

make -f Makefile.seq clean

/tmp/stamp-0.9.8/kmeans$ LD='gcc -lm' make -f Makefile.seq
cc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c cluster.c -o cluster.o
cc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c common.c -o common.o
cc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c kmeans.c -o kmeans.o
cc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c normal.c -o normal.o
cc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c ../lib/mt19937ar.c -o
../lib/mt19937ar.o
cc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c ../lib/random.c -o
../lib/random.o
cc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c ../lib/thread.c -o
../lib/thread.o
gcc -lm  cluster.o common.o kmeans.o normal.o ../lib/mt19937ar.o
../lib/random.o ../lib/thread.o -lpthread -o kmeans

Runtime: approx. 15s


tmp/stamp-0.9.8/kmeans$ CC='llvm-gcc' LD='llvm-gcc -lm' make -f Makefile.seq
llvm-gcc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c cluster.c -o
cluster.o
llvm-gcc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c common.c -o
common.o
llvm-gcc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c kmeans.c -o
kmeans.o
llvm-gcc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c normal.c -o
normal.o
llvm-gcc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c
../lib/mt19937ar.c -o ../lib/mt19937ar.o
llvm-gcc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c ../lib/random.c
-o ../lib/random.o
llvm-gcc -g -Wall -pthread -O3 -I../lib -DOUTPUT_TO_STDOUT -c ../lib/thread.c
-o ../lib/thread.o
llvm-gcc -lm  cluster.o common.o kmeans.o normal.o ../lib/mt19937ar.o
../lib/random.o ../lib/thread.o -lpthread -o kmeans

Runtime: 35s


-- 
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 Jun 25 11:58:52 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 11:58:52 -0500
Subject: [LLVMbugs] [Bug 2495] New: Dominator queries should be lazy
Message-ID: <bug-2495-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2495

           Summary: Dominator queries should be lazy
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Keywords: slow-compile
          Severity: normal
          Priority: P2
         Component: Global Analyses
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sabre at nondot.org
                CC: llvmbugs at cs.uiuc.edu


llvm-ld runs mem2reg after jump threading because jump threading introduces
allocas in some cases.  However, the number of cases it does this is pretty
low.

Because mem2reg is run after jump threading, the pass manager computes
dominator info and dom frontier info.  However, if there are no promotable
allocas in the function this information is just thrown away again and never
used.

We should make domfrontier and domtree be lazy: compute their info for the
function on the first meaningful query, instead of computing everything in
runOnFunction.

One minor issue with this is that the time to compute this dom info will be
charged (in -time-passes) to the pass that uses it, instead of to the analysis
pass.  You could argue that this is actually a good thing though.

-Chris


-- 
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 Jun 25 13:07:21 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 13:07:21 -0500
Subject: [LLVMbugs] [Bug 2473] -ipsccp miscompilation
In-Reply-To: <bug-2473-206@http.llvm.org/bugs/>
Message-ID: <200806251807.m5PI7LMZ020759@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2473


Torvald <torvald at se.inf.tu-dresden.de> changed:

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




--- Comment #5 from Torvald <torvald at se.inf.tu-dresden.de>  2008-06-25 13:07:18 ---
The fact that there will be no "fixed 2.3" (or will be?) and 2.4 is still some
time away makes this pretty inconvenient.


-- 
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 Jun 25 13:36:59 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 25 Jun 2008 13:36:59 -0500
Subject: [LLVMbugs] [Bug 2496] New: likely volatile bug
Message-ID: <bug-2496-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2496

           Summary: likely volatile bug
           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


This bug is seen using svn 52695 targetting ia32, running on Ubuntu Feisty.

At -O1 the volatile variable g_45 is read 6 times, and at -O3 it is read 4
times.  I compile with -fwrapv to avoid potential undefined integer overflow
behavior.

I apologize for the large testcase, I couldn't reduce it further.


unsigned short context;
typedef signed char int8_t;
typedef short int int16_t;
typedef int int32_t;
typedef unsigned short int uint16_t;
typedef unsigned int uint32_t;

static inline unsigned int
lshift_u_s (unsigned int left, int right)
{
  if ((right < 0) || (right >= sizeof (unsigned int) * 8)
      || (left > ((2147483647 * 2U + 1U) >> right)))
    {
      return left;
    }
  return left << right;
}

static inline int
rshift_s_u (int left, unsigned int right)
{
  if ((left < 0) || (right >= sizeof (int) * 8))
    {
      return left;
    }
  return left >> right;
}

static inline unsigned long int
mod_rhs (const long int rhs)
{
  if (rhs == 0)
    return 1;
  return rhs;
}

int32_t volatile g_45 ;
signed char g_47 = (signed char) 1;

int32_t func_38 (int8_t p_39);
int32_t func_38 (int8_t p_39)
{
  return 1;
}

int32_t func_73 (uint32_t p_75);
int32_t func_73 (uint32_t p_75)
{
  return g_47;
}

int32_t func_24 (int16_t p_35);
int32_t func_24 (int16_t p_35)
{
  int32_t tmp;
  if (p_35)
    {
      tmp = func_38 (g_45);
      if (tmp)
        rshift_s_u (g_45, 1);
    }
  else if (g_45)
    {
      g_47 = 0;
      while (g_47)
        g_47 = g_47 - 1;
    }
  return 1;
}

int32_t func_22 (void);
int32_t func_22 (void)
{
  int16_t tmp = func_73 (1);
  int16_t x = tmp;
  func_24 (x);
  return 1;
}

int32_t func_9 (int16_t p_11);
int32_t func_9 (int16_t p_11)
{
  return 1;
}

int32_t func_7 (uint32_t p_8);
int32_t func_7 (uint32_t p_8)
{
  int16_t tmp = func_22 ();
  int32_t tmp___0 = func_22 ();
  func_9 (tmp___0);
  return 1;
}

void func_1 (void);
void func_1 (void)
{
  unsigned long tmp___1;
  unsigned int tmp___2;
  int32_t tmp___3;
  unsigned long tmp___4;
  tmp___2 = lshift_u_s (1, 1);
  tmp___3 = func_7 (1);
  tmp___4 = mod_rhs (1);
  if ((tmp___2 != 1) + tmp___3 % tmp___4)
    {
      func_22 ();
    }
  tmp___1 = mod_rhs (1);
  func_73 (1 % tmp___1);
}

int main (void)
{
  func_1 ();
  printf ("checksum = %04X\n", context);
  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  Thu Jun 26 00:07:12 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 26 Jun 2008 00:07:12 -0500
Subject: [LLVMbugs] [Bug 2497] New: llvm-gcc segfault
Message-ID: <bug-2497-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2497

           Summary: llvm-gcc segfault
           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


This happens with svn 52763 targeting ia32 hosted on Ubuntu Feisty.

[regehr at babel tmp32]$ llvm-gcc -O small.c
small.c:27: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://developer.apple.com/bugreporter> for instructions.

[regehr at babel tmp32]$ cat small.c
typedef signed char int8_t;
typedef short int int16_t;
typedef int int32_t;
typedef unsigned int uint32_t;
uint32_t g_6;
int32_t g_148;
int32_t
func_1 (void)
{
  int16_t l_4;
  uint32_t l_147;
  for (0; 1; l_147 -= 0)
    {
      int8_t l_152;
      if (g_148)
        {
          for (1; 1; --l_152)
            1;
        }
      if (l_4)
        l_4 = g_6;
      else
        {
          g_6 = 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  Thu Jun 26 02:13:42 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 26 Jun 2008 02:13:42 -0500
Subject: [LLVMbugs] [Bug 2498] New: Incorrect optimization
Message-ID: <bug-2498-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2498

           Summary: Incorrect optimization
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: major
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: cristic at stanford.edu
                CC: llvmbugs at cs.uiuc.edu


I found the following bug in the optimizer.  Here's a short contrived example:

====================================
/* llvm-opt-bug.c */
static int foo(char *s) {
  if (s[0] == 0 || s[100000] != 0)
    return 0;

  return 1;
}


int main(int argc, char** argv) {
  return foo(argv[1]);
}
====================================

Note that running this program with an empty string as sole argument should
trigger no errors:

gcc llvm-opt-bug.c
./a.out ""

Compiling w/ LLVM and applying all standard compile-time optimizations (but w/
inlining disabled) produces the following bitcode:

llvm-gcc --emit-llvm -c llvm-opt-bug.c
opt --std-compile-opts --disable-inlining llvm-opt-bug.o -o llvm-opt-bug.opt.o
llvm-dis llvm-opt-bug.opt.o
cat llvm-opt-bug.opt.o.ll

; ModuleID = 'llvm-opt-bug.opt.o'
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 internal fastcc i32 @foo(i8 %s.val, i8 %s.100000.val) nounwind  {
entry:
        %tmp4 = icmp eq i8 %s.val, 0            ; <i1> [#uses=1]
        br i1 %tmp4, label %UnifiedReturnBlock, label %bb

bb:             ; preds = %entry
        %tmp9 = icmp eq i8 %s.100000.val, 0             ; <i1> [#uses=1]
        %retval = zext i1 %tmp9 to i32          ; <i32> [#uses=1]
        ret i32 %retval

UnifiedReturnBlock:             ; preds = %entry
        ret i32 0
}

define i32 @main(i32 %argc, i8** %argv) nounwind  {
entry:
        %tmp2 = getelementptr i8** %argv, i32 1         ; <i8**> [#uses=1]
        %tmp3 = load i8** %tmp2, align 4                ; <i8*> [#uses=2]
        %tmp3.val = load i8* %tmp3              ; <i8> [#uses=1]
        %tmp3.idx = getelementptr i8* %tmp3, i32 100000         ; <i8*>
[#uses=1]
        %tmp3.idx.val = load i8* %tmp3.idx              ; <i8> [#uses=1]
        %tmp4 = tail call fastcc i32 @foo( i8 %tmp3.val, i8 %tmp3.idx.val )
nounwind            ; <i32> [#uses=1]
        ret i32 %tmp4
}

Note that the load s[100000] in foo() was incorrectly lifted before the if
statement in foo().   

I'm using llvm-gcc 4.2 and llvm 2.3:

$ llvm-gcc -v
Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: ../llvm-gcc4.2-2.3.source/configure
--prefix=/mounts/zion/disks/0/localhome/tbrethou/2.3/test/llvm-gcc/install
--program-prefix=llvm-
--enable-llvm=/mounts/zion/disks/0/localhome/tbrethou/2.3/test/llvm-2.3/
--enable-languages=c,c++ --disable-shared --disable-bootstrap
Thread model: posix
gcc version 4.2.1 (Based on Apple Inc. build 5555) (LLVM build 2.3)

$ opt --version
Low Level Virtual Machine (http://llvm.org/):
  llvm version 2.3
  Optimized build with assertions.

Please let me know if you need any additional information.

Thank you,
Cristian

P.S. I apologize if this is a duplicate report; but it doesn't look my
first report was submitted (browser crashed).


-- 
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 Jun 26 03:00:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 26 Jun 2008 03:00:30 -0500
Subject: [LLVMbugs] [Bug 2499] New: Support for DragonFly BSD
Message-ID: <bug-2499-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2499

           Summary: Support for DragonFly BSD
           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: hasso at estpak.ee
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1779)
 --> (http://llvm.org/bugs/attachment.cgi?id=1779)
the patch

Attached trivial patch adds support for DragonFly BSD.


-- 
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 Jun 26 05:45:03 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 26 Jun 2008 05:45:03 -0500
Subject: [LLVMbugs] [Bug 2499] Support for DragonFly BSD
In-Reply-To: <bug-2499-206@http.llvm.org/bugs/>
Message-ID: <200806261045.m5QAj3d8029386@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2499


Matthijs Kooijman <matthijs at stdin.nl> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |matthijs at stdin.nl
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Matthijs Kooijman <matthijs at stdin.nl>  2008-06-26 05:45:02 ---
Commited here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080623/064188.html

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  Thu Jun 26 12:34:50 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 26 Jun 2008 12:34:50 -0500
Subject: [LLVMbugs] [Bug 2500] New: interpreter broken?
Message-ID: <bug-2500-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2500

           Summary: interpreter broken?
           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


Created an attachment (id=1780)
 --> (http://llvm.org/bugs/attachment.cgi?id=1780)
bitcode file

This is for recent svn, running on Ubuntu Feisty on ia32.

[regehr at babel tmp28]$ lli small.s.bc
checksum = 0000

[regehr at babel tmp28]$ lli --force-interpreter small.s.bc
lli: APInt.cpp:518: bool llvm::APInt::operator==(const llvm::APInt&) const:
Assertion `BitWidth == RHS.BitWidth && "Comparison requires equal bit widths"'
failed.
lli[0x86345f4]
/lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7cd2641]
/lib/tls/i686/cmov/libc.so.6(__assert_fail+0xfb)[0xb7cca43b]
lli[0x8607fcd]
[0x0]
Aborted (core dumped)


-- 
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 Jun 27 15:39:32 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 27 Jun 2008 15:39:32 -0500
Subject: [LLVMbugs] [Bug 2491] valgrind errors in FixBaseClassFields()
In-Reply-To: <bug-2491-206@http.llvm.org/bugs/>
Message-ID: <200806272039.m5RKdWP8008404@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2491


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #5 from Chris Lattner <sabre at nondot.org>  2008-06-27 15:39:31 ---
I'm going to close this as fixed, please reopen if it still occurs on mainline.


-- 
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 Jun 27 16:02:15 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 27 Jun 2008 16:02:15 -0500
Subject: [LLVMbugs] [Bug 2501] New: probable clang bug
Message-ID: <bug-2501-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2501

           Summary: probable clang bug
           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


On svn 52832 on ubuntu feisty on ia32:

[regehr at babel tmp33]$ ~/z/llvm/Release/bin/clang small.c --emit-llvm -o - |
llvm-as | opt -std-compile-opts | llc | as -o small.o ; gcc small.o -o small
[regehr at babel tmp33]$ ./small
0

[regehr at babel tmp33]$ llvm-gcc -O small.c -o small
small.c: In function ???main???:
small.c:31: warning: incompatible implicit declaration of built-in function
???printf???
[regehr at babel tmp33]$ ./small
1

[regehr at babel tmp33]$ cat small.c
typedef short int int16_t;
typedef int int32_t;
typedef unsigned char uint8_t;
typedef unsigned int uint32_t;
volatile int16_t g_60;
uint32_t g_120;
uint8_t g_149;
int32_t func_1 (void);
int32_t func_31 (uint8_t p_33, int16_t p_34, int32_t p_35, int32_t p_37);
int32_t
func_1 (void)
{
  uint32_t l_170 = 1;
  g_60 = (func_31 (l_170, g_120, g_149, 1));
  return 1;
}

int32_t
func_31 (uint8_t p_33, int16_t p_34, int32_t p_35, int32_t p_37)
{
  uint32_t l_160 = -9L;
  if (0xA68AL <= l_160)
    return 1;
  return 0;
}

int
main (int argc, char *argv[])
{
  func_1 ();
  printf ("%d\n", g_60);
  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 Jun 28 01:26:06 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 28 Jun 2008 01:26:06 -0500
Subject: [LLVMbugs] [Bug 2501] probable clang bug
In-Reply-To: <bug-2501-206@http.llvm.org/bugs/>
Message-ID: <200806280626.m5S6Q6DQ025907@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2501


Eli Friedman <sharparrow1 at yahoo.com> changed:

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




--- Comment #3 from Eli Friedman <sharparrow1 at yahoo.com>  2008-06-28 01:26:05 ---
Fixed.

http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080623/006282.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 Jun 28 18:31:29 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 28 Jun 2008 18:31:29 -0500
Subject: [LLVMbugs] [Bug 2502] New: Function pointer types are wrongly
	compared
Message-ID: <bug-2502-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2502

           Summary: Function pointer types are wrongly compared
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: filcab at gmail.com
                CC: llvmbugs at cs.uiuc.edu


clang blows up with the following program:
void (*f)(void);
void (*f)() = 0;


But gcc compiles happily. Also, in the draft for the standard (
http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1256.pdf
), it is said that (section 6.9.1 - Function definitions):

typedef int F(void); //type F is ??????function with no parameters
                     // returning int??????
// ...
int g() { /* ... */ } //RIGHT: g has type compatible with F

So I'm assuming it's a bug in clang.


The offending line is (I think) clang/lib/Sema/SemaDecl.cpp:401.

SVN Revision: 52870


-- 
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 Jun 28 21:49:17 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 28 Jun 2008 21:49:17 -0500
Subject: [LLVMbugs] [Bug 2503] New: Infinite recursion crash during link
Message-ID: <bug-2503-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2503

           Summary: Infinite recursion crash during link
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: critical
          Priority: P2
         Component: Linker
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: bryceco at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1782)
 --> (http://llvm.org/bugs/attachment.cgi?id=1782)
llvm code causing crash

The attached .llvm code was emitted by clang. llvm-ld crashes while linking it:

% llvm-as bad1.llvm -o bad1.bc
% llvm-ld bad1.bc
Segmentation fault


-- 
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 Jun 29 14:02:49 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 29 Jun 2008 14:02:49 -0500
Subject: [LLVMbugs] [Bug 2504] New: Codegen for vector select fails on ppc32
	and ppc64
Message-ID: <bug-2504-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2504

           Summary: Codegen for vector select fails on ppc32 and ppc64
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: PowerPC
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: llvm at brendanyounger.com
                CC: llvmbugs at cs.uiuc.edu


Compiling the following code with either
lvm-as < vector_select.ll | llc -march=ppc32 > vector_select.s
or
lvm-as < vector_select.ll | llc -march=ppc64 > vector_select.s
yields a failed assertion
0x1504a30: v2f64 = select_cc 0x15042b0, 0x1504110, 0x1504620, 0x15045c0,
0x1504680Assertion failed: (0 && "Unhandled operation in SplitVectorOp!"),
function SplitVectorOp, file LegalizeDAG.cpp, line 6778.

The offending code in "vector_select.ll" is:
define <2 x double> @vector_select_bug(<2 x double> %x, <2 x double> %y) {
        %x.lo = extractelement <2 x double> %x, i32 0
        %x.lo.ge = fcmp oge double %x.lo, 0.0
        %a.d = select i1 %x.lo.ge, <2 x double> %y, <2 x double> %x
        ret <2 x double> %a.d
}


-- 
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 Jun 29 14:22:07 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 29 Jun 2008 14:22:07 -0500
Subject: [LLVMbugs] [Bug 2505] New: Support for autorelease pools and
 'autorelease' messages in memory leak analysis
Message-ID: <bug-2505-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2505

           Summary: Support for autorelease pools and 'autorelease' messages
                    in memory leak analysis
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Keywords: new-feature
          Severity: enhancement
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nikita at zhuk.fi
                CC: llvmbugs at cs.uiuc.edu


In a non-GC enrivonment, clang static analyzer stops tracking reference counts
as soon as it sees an "autorelease" message sent to an object (tracking is
stopped in CFRefCount.cpp:879). 

This causes clang to miss possible memory leaks and uses after release. Some
concrete examples:

- (void)useAfterRelease
{
  NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
  NSString *str = [[NSString alloc] initWithString:@"some string data"]; //
rc(str)=1

  [str autorelease]; // rc(str)=1, added to local autorelease pool 'pool'
  [pool release]; // rc(str)=0, since local autorelease pool 'pool' is released

  NSLog(@"%@", str); // 'str' is used after release. Not reported by clang
checker.
}

- (void)memoryLeak
{
  NSString *str = [[NSString alloc] initWithString:@"some string data"]; //
rc(str)=1

  [str retain]; // rc(str)=2
  [str autorelease]; // rc(str)=2, and will be 1 after the current event.
   // 'str' is leaked, since it goes out of scope. Not reported by clang
checker.
}

Clang should probably track local autorelease pools (which may be nested) and
handle 'autorelease' messages just as it handles 'release' messages, but taking
into account the knowledge about autorelease pools. 

Clang revision used: 52881


-- 
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 Jun 29 16:42:46 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 29 Jun 2008 16:42:46 -0500
Subject: [LLVMbugs] [Bug 2506] New: Invalid instcombine with select+div
Message-ID: <bug-2506-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2506

           Summary: Invalid instcombine with select+div
           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


Testcase:
define i32 @a(i32 %x, i32 %y) {
entry:
        %tobool = icmp ne i32 %y, 0             ; <i1> [#uses=1]
        %cond = select i1 %tobool, i32 5, i32 0         ; <i32> [#uses=2]
        %call = call i32 @b( i32 %cond )                ; <i32> [#uses=0]
        %div = sdiv i32 %x, %cond               ; <i32> [#uses=1]
        ret i32 %div
}

declare i32 @b(i32)

Run through opt -instcombine produces:

define i32 @a(i32 %x, i32 %y) {
entry:
        %call = call i32 @b( i32 5 )            ; <i32> [#uses=0]
        %div = sdiv i32 %x, 5           ; <i32> [#uses=1]
        ret i32 %div
}

declare i32 @b(i32)

This optimization is wrong because it assumes that b eventually returns, which
is not guaranteed.


-- 
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 Jun 29 21:08:44 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 29 Jun 2008 21:08:44 -0500
Subject: [LLVMbugs] [Bug 1879] Branch folding deficiency
In-Reply-To: <bug-1879-206@http.llvm.org/bugs/>
Message-ID: <200806300208.m5U28i6E020489@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=1879


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #3 from Chris Lattner <sabre at nondot.org>  2008-06-29 21:08:44 ---
This is 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  Sun Jun 29 21:43:29 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 29 Jun 2008 21:43:29 -0500
Subject: [LLVMbugs] [Bug 2504] Codegen for vector select fails on ppc32 and
	ppc64
In-Reply-To: <bug-2504-206@http.llvm.org/bugs/>
Message-ID: <200806300243.m5U2hTnh021526@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2504


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-06-29 21:43:29 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080623/064289.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 Jun 30 04:32:02 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 30 Jun 2008 04:32:02 -0500
Subject: [LLVMbugs] [Bug 2507] New: Anders-AA returns wrong mod/ ref info
 for external function call
Message-ID: <bug-2507-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2507

           Summary: Anders-AA returns wrong mod/ref info for external
                    function call
           Product: libraries
           Version: 2.3
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Interprocedural Analyses
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: hvdieren at elis.ugent.be
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1784)
 --> (http://llvm.org/bugs/attachment.cgi?id=1784)
failing test case

Hi,

When studying some alias analysis results, I found this wrong behavior in
Andersen's Alias Analysis. The C-code for the sample program is listed here,
the bc is attached:
#include <stdio.h>

int main( int argc, char * argv[] ) {
    int n, i, nelem, sum;
    char buf[128];
    int arr[256];

    i = 0;
    while( fgets( buf, sizeof(buf), stdin ) ) {
        n = atoi( buf );
        arr[i++] = n;
    }
    nelem = i;

    sum = 0;
    for( i=0; i < nelem; ++i ) {
        sum += arr[i];
        arr[i] = 0;
    }
    printf( "Sum is: %d\n", sum );
}

Running -anders-aa:
% opt -anders-aa -count-aa -count-aa-print-all-queries -aa-eval
-disable-inlining -disable-opt -f sample2.bc -o sample2.opt.bc  

I get, amongst others:
NoModRef:  Ptr: [128B] [128 x i8]* %buf <->     %tmp = call i32 (...)* @atoi(
[128 x i8]* %buf ) nounwind               ; <i32> [#uses=1]

This should be a Ref result instead of NoModRef.

Hans.


-- 
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 Jun 30 17:31:47 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 30 Jun 2008 17:31:47 -0500
Subject: [LLVMbugs] [Bug 2508] New: -Wno-vla not supported
Message-ID: <bug-2508-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2508

           Summary: -Wno-vla not supported
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


$ llvm-g++ -Wvla x.cpp
cc1plus: error: unrecognized command line option "-Wvla"

It'd be great if llvm-g++ would support the -Wvla (and -Wno-vla) option.


-- 
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 Jun 30 17:50:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 30 Jun 2008 17:50:48 -0500
Subject: [LLVMbugs] [Bug 2508] -Wno-vla not supported
In-Reply-To: <bug-2508-206@http.llvm.org/bugs/>
Message-ID: <200806302250.m5UMom36013666@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2508


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>  2008-06-30 17:50:47 ---
(In reply to comment #0)
> $ llvm-g++ -Wvla x.cpp
> cc1plus: error: unrecognized command line option "-Wvla"
> 
> It'd be great if llvm-g++ would support the -Wvla (and -Wno-vla) option.
This seems to be not supported even by mainline gcc:

$ gcc -v
Using built-in specs.
Target: i686-pc-linux-gnu
Configured with: /var/tmp/portage/sys-devel/gcc-4.2.2/work/gcc-4.2.2/configure
--prefix=/usr --bindir=/usr/i686-pc-linux-gnu/gcc-bin/4.2.2
--includedir=/usr/lib/gcc/i686-pc-linux-gnu/4.2.2/include
--datadir=/usr/share/gcc-data/i686-pc-linux-gnu/4.2.2
--mandir=/usr/share/gcc-data/i686-pc-linux-gnu/4.2.2/man
--infodir=/usr/share/gcc-data/i686-pc-linux-gnu/4.2.2/info
--with-gxx-include-dir=/usr/lib/gcc/i686-pc-linux-gnu/4.2.2/include/g++-v4
--host=i686-pc-linux-gnu --build=i686-pc-linux-gnu --disable-altivec
--enable-nls --without-included-gettext --with-system-zlib --disable-checking
--disable-werror --enable-secureplt --disable-libunwind-exceptions
--disable-multilib --enable-libmudflap --disable-libssp --disable-libgcj
--with-arch=i686 --enable-languages=c,c++,fortran --enable-shared
--enable-threads=posix --enable-__cxa_atexit --enable-clocale=gnu
Thread model: posix
gcc version 4.2.2 (Gentoo 4.2.2 p1.0)

$ touch foo.c && gcc -Wvla foo.c
cc1: error: unrecognized command line option "-Wvla"


-- 
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 Jun 30 19:46:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 30 Jun 2008 19:46:48 -0500
Subject: [LLVMbugs] [Bug 2509] New: Jump threading produces funky output
Message-ID: <bug-2509-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2509

           Summary: Jump threading produces funky output
           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: resistor at mac.com
                CC: llvmbugs at cs.uiuc.edu


In the case attached, jump threading produces output with unreachable blocks
that still branch to reachable blocks.  It would be really nice if it didn't do
this.  This situation causes GVN to miss some optimizations.


-- 
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 Jun 30 19:48:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 30 Jun 2008 19:48:19 -0500
Subject: [LLVMbugs] [Bug 2503] Infinite recursion crash during link
In-Reply-To: <bug-2503-206@http.llvm.org/bugs/>
Message-ID: <200807010048.m610mJ38017454@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2503


Owen Anderson <resistor at mac.com> changed:

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




--- Comment #6 from Owen Anderson <resistor at mac.com>  2008-06-30 19:48:19 ---
Fixed in r52946.  I'm forking the fact that jump threading shouldn't be
producing code like this into PR2509.


-- 
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.