From bugzilla-daemon at cs.uiuc.edu  Tue Jul  1 08:31:38 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 1 Jul 2008 08:31:38 -0500
Subject: [LLVMbugs] [Bug 2510] New: Non-power-of-2 _Bool vector asserts
Message-ID: <bug-2510-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2510

           Summary: Non-power-of-2 _Bool vector asserts
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


The following trivial code, declaring a vector of 3 _Bool's lets clang crash
with an assert:
   _Bool __attribute__((vector_size(3))) X;

$ clang -emit-llvm  < tmp.c 
clang: ASTContext.cpp:317: std::pair<long long unsigned int, unsigned int>
clang::ASTContext::getTypeInfo(clang::QualType): Assertion `Align && (Align &
(Align-1)) == 0 && "Alignment must be power of 2"' failed.
clang[0x8637ffa]
clang[0x863812c]
[0xffffe420]
/lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7d32201]
/lib/tls/i686/cmov/libc.so.6(__assert_fail+0xee)[0xb7d29b6e]
clang(_ZN5clang10ASTContext11getTypeInfoENS_8QualTypeE+0x70f)[0x84ba1d3]
clang(_ZN5clang10ASTContext12getTypeAlignENS_8QualTypeE+0x1f)[0x83bb46f]
clang(_ZN5clang7CodeGen13CodeGenModule17EmitGlobalVarInitEPKNS_7VarDeclE+0x55c)[0x83b0cb2]
clang(_ZN5clang7CodeGen13CodeGenModule13EmitGlobalVarEPKNS_7VarDeclE+0x7e)[0x83b0f9a]
clang(_ZN5clang7CodeGen13CodeGenModule23EmitGlobalVarDeclaratorEPKNS_7VarDeclE+0x29)[0x83b0fc5]
clang(_ZN46_GLOBAL__N_ModuleBuilder.cpp_00000000_B4E8DBEF13CodeGenerator18HandleTopLevelDeclEPN5clang4DeclE+0x181)[0x83aee8d]
clang(_ZN5clang8ParseASTERNS_12PreprocessorEPNS_11ASTConsumerEb+0x1e1)[0x846ea0d]
clang[0x83a2562]
clang(main+0x66b)[0x83a2f6b]
/lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe0)[0xb7d1c050]
clang[0x8373321]
Aborted (core dumped)

When replacing the _Bool with int, clang properly generates the following
error:

   <stdin>:1:20: error: vector size not an integral multiple of component size
   int __attribute__((vector_size(3))) X;
                      ^           ~

I guess this should also happen with _Bool, don't know why those are handled
differently.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  1 15:45:57 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 1 Jul 2008 15:45:57 -0500
Subject: [LLVMbugs] [Bug 2511] New: Assertion
 `!OpInfo.AssignedRegs.Regs.empty() && "Couldn' t allocate input reg!"'
 failed.
Message-ID: <bug-2511-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2511

           Summary: Assertion `!OpInfo.AssignedRegs.Regs.empty() &&
                    "Couldn't allocate input reg!"' failed.
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Register Allocator
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: alenhar2 at uiuc.edu
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1791)
 --> (http://llvm.org/bugs/attachment.cgi?id=1791)
foo.bc

llc foo.bc results in an assertion in the register allocator (x86 backend).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  1 16:40:22 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 1 Jul 2008 16:40:22 -0500
Subject: [LLVMbugs] [Bug 2512] New: DomTree verifier checks when it shouldn't
Message-ID: <bug-2512-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2512

           Summary: DomTree verifier checks when it shouldn't
           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


test/CodeGen/X86/nancvt.ll is failing because dominator information is not
preserved by loop unrolling, but loop unrolling does not claim to preserve
dominator information, so this shouldn't cause the verifier to check 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  Tue Jul  1 18:41:43 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 1 Jul 2008 18:41:43 -0500
Subject: [LLVMbugs] [Bug 2481] . s file contains __eh_frame even if building
 with -fno-exceptions
In-Reply-To: <bug-2481-206@http.llvm.org/bugs/>
Message-ID: <200807012341.m61Nfhsh007034@zion.cs.uiuc.edu>

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


Bill Wendling <isanbard at gmail.com> changed:

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




--- Comment #21 from Bill Wendling <isanbard at gmail.com>  2008-07-01 18:41:42 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080630/064401.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 Jul  2 11:13:55 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 2 Jul 2008 11:13:55 -0500
Subject: [LLVMbugs] [Bug 2513] New: invalid bitcode bug
Message-ID: <bug-2513-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2513

           Summary: invalid bitcode 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 using svn 53007 on Ubuntu Feisty on ia32.

[regehr at babel tmp36]$ /home/regehr/z/llvm/Release/bin/clang small.c --emit-llvm
-o - | llvm-as | opt -std-compile-opts | llc
opt[0x8409bf4]
llc: bitcode didn't read correctly.
Reason: Invalid bitcode signature

[regehr at babel tmp36]$ cat small.c
volatile short g_2;
unsigned char g_14;
int g_107;
volatile short g_121;
int g_131;
short g_154;

int func_10 (short p_11)
{
  return g_14;
}

int func_32 (short p_33, short p_34, unsigned char p_36, unsigned p_38,
             short p_39)
{
  for (0; 1; g_14--)
    if (1 != (func_63 (1, g_107, 1, 1)))
      for (1; 1; g_154++);
}

int func_63 (short p_64, short p_65, short p_66, int p_67)
{
  for (1; g_2; --g_2)
    {
      (func_72 (p_67));
      if (g_131)
        return 1;
    }
  return p_65;
}

int func_72 (short p_73)
{
  if (func_10 (1))
    {
      if (p_73)
        return g_2;
      for (1; 1; p_73--);
    }
  for (0; g_121; --g_121);
  for (g_107 = 1; 0; g_107--);
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  2 13:06:06 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 2 Jul 2008 13:06:06 -0500
Subject: [LLVMbugs] [Bug 2509] Jump threading produces funky output
In-Reply-To: <bug-2509-206@http.llvm.org/bugs/>
Message-ID: <200807021806.m62I66KB020341@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2509


Owen Anderson <resistor at mac.com> changed:

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




--- Comment #8 from Owen Anderson <resistor at mac.com>  2008-07-02 13:06:05 ---
Part 1 is done as of r53038.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  2 22:31:23 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 2 Jul 2008 22:31:23 -0500
Subject: [LLVMbugs] [Bug 2020] clang crash redeclaring a union as an enum
In-Reply-To: <bug-2020-206@http.llvm.org/bugs/>
Message-ID: <200807030331.m633VNnA006436@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2020


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-07-02 22:31:23 ---
Fixed, thanks!
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080630/006355.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 Jul  2 22:48:07 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 2 Jul 2008 22:48:07 -0500
Subject: [LLVMbugs] [Bug 2252] Unexpected diagnostics in PP expression
In-Reply-To: <bug-2252-206@http.llvm.org/bugs/>
Message-ID: <200807030348.m633m7xi006938@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2252


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-07-02 22:48:06 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080630/006356.html


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

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  3 02:05:42 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 3 Jul 2008 02:05:42 -0500
Subject: [LLVMbugs] [Bug 2512] DomTree verifier checks when it shouldn't
In-Reply-To: <bug-2512-206@http.llvm.org/bugs/>
Message-ID: <200807030705.m6375gFj014016@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2512


Devang Patel <dpatel at apple.com> changed:

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




--- Comment #1 from Devang Patel <dpatel at apple.com>  2008-07-03 02:05:41 ---
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  Thu Jul  3 04:35:34 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 3 Jul 2008 04:35:34 -0500
Subject: [LLVMbugs] [Bug 2514] New: Relocation when using llvm in shared
	library
Message-ID: <bug-2514-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2514

           Summary: Relocation when using llvm in shared library
           Product: libraries
           Version: 2.3
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: cyb at lepi.org
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1794)
 --> (http://llvm.org/bugs/attachment.cgi?id=1794)
Workaround patch for llvm 2.3

When I try to link a shared library with llvm's x86 code generator, on a 64bits
machine, I get an error about a R_X86_64_PC32 relocation error. The error is
caused by the assembly function called in 'lib/Target/X86/X86JITInfo.cpp'.

I have attached a patch that work arround the issue (for llvm 2.3), in waiting
for a better solution.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  3 04:55:11 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 3 Jul 2008 04:55:11 -0500
Subject: [LLVMbugs] [Bug 2515] New: execution of scanf functions broken in
	lli
Message-ID: <bug-2515-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2515

           Summary: execution of scanf functions broken in lli
           Product: tools
           Version: 2.3
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: critical
          Priority: P2
         Component: lli
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: hvdieren at elis.ugent.be
                CC: llvmbugs at cs.uiuc.edu


Hi,

When running SPEC benchmarks in lli, I found that the execution of scanf()
functions is broken. Here is a description of what goes wrong:

1. mcf benchmark calls sscanf()
2. lli ends up in ByteswapSCANFResults()
(lib/ExecutionEngine/Interpreter/ExternalFunctions.cpp)
   which executes this code for every argument to sscanf():
        if (Size) {
          GenericValue GV;
          void *Arg = Args[ArgNo++];
          memcpy(&GV, Arg, Size);
          TheInterpreter->StoreValueToMemory(GV, (GenericValue*)Arg, Ty);
        }
  The problem is in the definition of GenericValue and the use of memcpy():
include/llvm/ExecutionEngine/GenericValue.h:
struct GenericValue {
  union {
    double          DoubleVal;
    float           FloatVal;
    PointerTy       PointerVal;
    struct { unsigned int first; unsigned int second; } UIntPairVal;
    unsigned char   Untyped[8];
  };
  APInt IntVal;   // also used for long doubles

  GenericValue() : DoubleVal(0.0), IntVal(1,0) {}
  explicit GenericValue(void *V) : PointerVal(V), IntVal(1,0) { }
};

After scanning the value 5985, memcpy() leaves this value into GV:
(gdb)
#5  0x00000000008d666e in ByteswapSCANFResults (Fmt=0x11aaa23 " %ld",
Arg0=0x11ab410, Arg1=0x11aba60, Arg2=0x7fff4a881a20, Arg3=0x10f5690,
Arg4=0x7fff4a8819f0,
    Arg5=0x8fa236, Arg6=0x11aad70, Arg7=0x7fff4a881a38, Arg8=0x0) at
/.../llvm-2.3/lib/ExecutionEngine/Interpreter/ExternalFunctions.cpp:466
466               TheInterpreter->StoreValueToMemory(GV, (GenericValue*)Arg,
Ty);
(gdb) p GV
$33 = {{DoubleVal = 2.9569828903598606e-320, FloatVal = 8.38677131e-42,
PointerVal = 0x1761, UIntPairVal = {first = 5985, second = 0},
    Untyped = "a\027\000\000\000\000\000"}, IntVal = {BitWidth = 1, {VAL = 0,
pVal = 0x0}}}

The memcpy call does not overwrite the IntVal field.

The problem appears to be that, for the memcpy() trick to work, the IntVal must
be part of the union in GenericValue. It appears, however, that the structure
of GenericValue has been unchanged since 15 months. (So how about tests for
this code?)

The problem is also signalled by an assert(), because the bitwidth of the
IntVal is initialized to 1, but this is not conforming the size of the type Ty.
Assertion `(IntVal.getBitWidth()+7)/8 >= StoreBytes && "Integer too small!"'

All of this is on x86-64 running redhat linux.

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  Thu Jul  3 16:21:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 3 Jul 2008 16:21:30 -0500
Subject: [LLVMbugs] [Bug 2482] llvm-gcc emits twice as much debug info when
 compiling instcombine at -O0 -g than GCC
In-Reply-To: <bug-2482-206@http.llvm.org/bugs/>
Message-ID: <200807032121.m63LLUMo024469@zion.cs.uiuc.edu>

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


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

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




--- Comment #7 from Evan Cheng <evan.cheng at apple.com>  2008-07-03 16:21:29 ---
I'll close this now. There are additional refinements to be made but they are
no as critical as this one.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  3 22:15:02 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 3 Jul 2008 22:15:02 -0500
Subject: [LLVMbugs] [Bug 2516] New: llvm-gcc miscompiles crtbegin.o on
	i386-pc-solaris*
Message-ID: <bug-2516-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2516

           Summary: llvm-gcc miscompiles crtbegin.o on i386-pc-solaris*
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Solaris
            Status: NEW
          Severity: minor
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: Nathan.Keynes at sun.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1796)
 --> (http://llvm.org/bugs/attachment.cgi?id=1796)
Disassembly of crtbegin.o compiled with llvm-gcc

Currently (since at least LLVM 2.2) under i386-pc-solaris, llvm-gcc produces a
crtbegin.o that crashes immediately on startup. As a side effect, this also
means that the gcc build fails as it cannot create working executables.

Attached disassembly of bad (llvm) and good (gcc) crtbegin.o, as well as the
preprocessed source. The actual crash is in frame_dummy at 7c: cmpl $0x0, 0x0.
It looks to me like llvm-gcc is ignoring the -fPIC, which may or may not be the
cause of the problem


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

From bugzilla-daemon at cs.uiuc.edu  Thu Jul  3 23:38:37 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 3 Jul 2008 23:38:37 -0500
Subject: [LLVMbugs] =?utf-8?q?=5BBug_2517=5D_New=3A__scan-build_reports_?=
 =?utf-8?q?=E2=80=9Cmissing_-dealloc=E2=80=9D_where_-dealloc_is_unnecessar?=
 =?utf-8?q?y?=
Message-ID: <bug-2517-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2517

           Summary: scan-build reports ???missing -dealloc??? where -dealloc
                    is unnecessary
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: minor
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rix.rob at gmail.com
                CC: llvmbugs at cs.uiuc.edu


One such instance of this is when you have a class which has no object ivars,
but has integers instead. A -dealloc implementation is unnecessary, since all
you???d have it do is call [super dealloc];, but scan-build/ccc-analyzer
complains anyway.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  3 23:42:23 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 3 Jul 2008 23:42:23 -0500
Subject: [LLVMbugs] [Bug 2518] New: scan-build treats -[NSAutoreleasePool
 drain] as -[ NSAutoreleasePool release]
Message-ID: <bug-2518-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2518

           Summary: scan-build treats -[NSAutoreleasePool drain] as -
                    [NSAutoreleasePool release]
           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: rix.rob at gmail.com
                CC: llvmbugs at cs.uiuc.edu


This was found with the common idiom of having an autorelease pool around a
loop:

NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];

while(???) {
  ???do some stuff???
  [pool drain];
}

[pool release];

In this example, scan-build will report Use-After-Release bugs on [pool drain]
(for the second pass through the loop) and on [pool release] (after a single
pass through the loop).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  4 00:12:08 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 4 Jul 2008 00:12:08 -0500
Subject: [LLVMbugs] [Bug 2428] Windows Vista don't support format %n
In-Reply-To: <bug-2428-206@http.llvm.org/bugs/>
Message-ID: <200807040512.m645C87n006871@zion.cs.uiuc.edu>

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


????????? <yonggangluo at hotmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|FIXED                       |INVALID




--- Comment #12 from ????????? <yonggangluo at hotmail.com>  2008-07-04 00:12:08 ---
Now I know why the diff file that I generate is so hudge, Because the original
genmodes.c is encoding with Unix stype Line Encoding, and the file that I
generated is using Windows stype Line Encoding, so the file is very hudge,
Maybe it's would be good that if diff support for Windows stype Line Encoding!.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  4 01:44:50 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 4 Jul 2008 01:44:50 -0500
Subject: [LLVMbugs] [Bug 2519] New: scan-build ignores CFRelease
Message-ID: <bug-2519-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2519

           Summary: scan-build ignores CFRelease
           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: boredzo at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1801)
 --> (http://llvm.org/bugs/attachment.cgi?id=1801)
Test program. Use `make static-analysis` to run the analyzer.

The static analyzer flags false memory leaks on some objects that the
program??under analysis actually does release, if it does so using CFRelease.

The problem seems to be related to putting the object into a CF collection
before releasing it. The program also releases the collection, also using
CFRelease. Simply creating a CF object and releasing it is not enough to
reproduce the 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  Fri Jul  4 06:59:10 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 4 Jul 2008 06:59:10 -0500
Subject: [LLVMbugs] [Bug 2489] clang -g segfaults on functions with arguments
In-Reply-To: <bug-2489-206@http.llvm.org/bugs/>
Message-ID: <200807041159.m64BxAEh032191@zion.cs.uiuc.edu>

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


Matthijs Kooijman <matthijs at stdin.nl> changed:

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




--- Comment #1 from Matthijs Kooijman <matthijs at stdin.nl>  2008-07-04 06:59:09 ---
Sanjiv fixed this in
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080630/006390.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 Jul  5 18:08:22 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 5 Jul 2008 18:08:22 -0500
Subject: [LLVMbugs] [Bug 2146] Linker does not merge aliases and prototypes
	correctly
In-Reply-To: <bug-2146-206@http.llvm.org/bugs/>
Message-ID: <200807052308.m65N8M1S008430@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2146


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>  2008-07-05 18:08:21 ---
Fixed in:

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080630/064569.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080630/064570.html

surprisingly, it took much less time, than I expected. It seems, that Chris'
recent revamp of LinkFunctionProtos() simplified stuff alot.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  5 18:43:11 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 5 Jul 2008 18:43:11 -0500
Subject: [LLVMbugs] [Bug 2463] linking weak and alias fails
In-Reply-To: <bug-2463-206@http.llvm.org/bugs/>
Message-ID: <200807052343.m65NhB6u009438@zion.cs.uiuc.edu>

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


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

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




--- Comment #1 from Anton Korobeynikov <asl at math.spbu.ru>  2008-07-05 18:43:11 ---
Fixed in:

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080630/064572.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080630/064573.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 Jul  6 05:15:22 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 6 Jul 2008 05:15:22 -0500
Subject: [LLVMbugs] [Bug 2428] Windows Vista don't support format %n
In-Reply-To: <bug-2428-206@http.llvm.org/bugs/>
Message-ID: <200807061015.m66AFM3g030461@zion.cs.uiuc.edu>

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


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

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|INVALID                     |FIXED




--- Comment #16 from Anton Korobeynikov <asl at math.spbu.ru>  2008-07-06 05:15:21 ---
(In reply to comment #12)
> Maybe it's would be good that if diff support for Windows stype Line Encoding!.
It does support. However, it cannot silently ignore difference in line endings,
because it's real difference you can ask for. It will be much better if your
favorite editor just preserve original file line endings. It's easy!

Anyway, snipped applied here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080630/064575.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 Jul  6 19:07:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 6 Jul 2008 19:07:09 -0500
Subject: [LLVMbugs] [Bug 2520] New: fabs not marked readnone
Message-ID: <bug-2520-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2520

           Summary: fabs not marked readnone
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


According to fabs(3), none of fabs fabsf or fabsl can have any errors. As such,
they should be marked readnone since all they ever do is return their own
input.

$ cat fabs.c
#include <math.h>

double f(double *x, double *y) { return fabs(*x + *y); }

$ llvm-gcc -O2 fabs.c -S -o - -emit-llvm
; ModuleID = 'fabs.c'
target datalayout =
"e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32"
target triple = "i386-pc-linux-gnu"

define double @f(double* %x, double* %y) nounwind  {
entry:
        %tmp2 = load double* %x, align 8                ; <double> [#uses=1]
        %tmp4 = load double* %y, align 8                ; <double> [#uses=1]
        %tmp5 = add double %tmp2, %tmp4         ; <double> [#uses=1]
        %tmp6 = tail call double @fabs( double %tmp5 )          ; <double>
[#uses=1]
        ret double %tmp6
}

declare double @fabs(double)

This stops some optimizations from working (GVN in particular) because it
doesn't know that fabs() won't be writing to one of the double* arguments.


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

From bugzilla-daemon at cs.uiuc.edu  Sun Jul  6 21:50:43 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 6 Jul 2008 21:50:43 -0500
Subject: [LLVMbugs] [Bug 2521] New: opt prints gibberish (outputs binary) to
 the console with MSYS on win32
Message-ID: <bug-2521-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2521

           Summary: opt prints gibberish (outputs binary) to the console
                    with MSYS on win32
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: opt
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: dominik at vga.hr
                CC: llvmbugs at cs.uiuc.edu


running a line 'opt -std-compile-opts test.bc' - notice that it is without -o
test2.bc option, it prints out gibberish to the MSYS shell as one would get
with -f flag. This is only symptomatic to MSYS, standard cmd shell in w32 does
not show this behaviour.

System: Windows XP Service Pack 3
MSYS: 1.0.10
opt: llvm version 2.3


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  7 05:45:55 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 05:45:55 -0500
Subject: [LLVMbugs] [Bug 2522] New: cannot register variable
Message-ID: <bug-2522-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2522

           Summary: cannot register variable
           Product: tools
           Version: 2.3
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ddk128 at gmail.com
                CC: llvmbugs at cs.uiuc.edu


While attempting to compile a C file generated by GHC. llvm-gcc chocked with
the error message "sorry, unimplemented: LLVM cannot register variable 'R1'."
and requested I report a bug.

The GHC version was ghc-6.8.3.

>From the root of the GHC installation the path to the header llvm-gcc was
parsing is "include/Regs.h" and the error occured at line 220.

David


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  7 09:11:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 09:11:39 -0500
Subject: [LLVMbugs] [Bug 2523] New: Asserts with releational operator
 applied to ext_vector_type(X) as argument
Message-ID: <bug-2523-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2523

           Summary: Asserts with releational operator applied to
                    ext_vector_type(X) as argument
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: benedict.gaster at amd.com
                CC: llvmbugs at cs.uiuc.edu


The code:

typedef __attribute__(( ext_vector_type(2) )) float float2;

float2 vec2, vec2_2;

void test2() 
{
  if (vec2 < vec2_2)
    ;
}

causes clang to fail with the following error (command line "clang
-emit-llvm"):

Assertion failed: ((getOperand(0)->getType()->isInteger() ||
isa<PointerType>(getOperand(0)->getType())) && "Invalid operand types for ICmp
instruction"), function ICmpInst, file
/Users/bgaster/llvm-latest/llvm/include/llvm/Instructions.h, line 629.
0   clang                               0x002e7bcc
_ZN4llvm3sys18RemoveFileOnSignalERKNS0_4PathEPSs + 844
1   libSystem.B.dylib                   0x9523309b _sigtramp + 43
2   ???                                 0xffffffff 0x0 + 4294967295
3   libSystem.B.dylib                   0x952abec2 raise + 26
4   libSystem.B.dylib                   0x952bb47f abort + 73
5   libSystem.B.dylib                   0x952ad063 __assert_rtn + 101
6   clang                               0x0006e37c
_ZN4llvm9IRBuilder10CreateICmpENS_7CmpInst9PredicateEPNS_5ValueES4_PKc + 300
7   clang                               0x0006bd54
_ZN5clang7CodeGen15CodeGenFunction14EmitScalarExprEPKNS_4ExprE + 7476
8   clang                               0x0006f7b1
_ZN5clang11StmtVisitorIN45_GLOBAL__N_CGExprScalar.cpp_00000000_1CBA9DDB17ScalarExprEmitterEPN4llvm5ValueEE5VisitEPNS_4StmtE
+ 1521
9   clang                               0x000707da
_ZN5clang11StmtVisitorIN45_GLOBAL__N_CGExprScalar.cpp_00000000_1CBA9DDB17ScalarExprEmitterEPN4llvm5ValueEE5VisitEPNS_4StmtE
+ 5658
10  clang                               0x0006a09b
_ZN5clang7CodeGen15CodeGenFunction14EmitScalarExprEPKNS_4ExprE + 123
11  clang                               0x0004d955
_ZN5clang7CodeGen15CodeGenFunction18EvaluateExprAsBoolEPKNS_4ExprE + 69
12  clang                               0x000807d7
_ZN5clang7CodeGen15CodeGenFunction10EmitIfStmtERKNS_6IfStmtE + 39
13  clang                               0x00080d24
_ZN5clang7CodeGen15CodeGenFunction16EmitCompoundStmtERKNS_12CompoundStmtEbPN4llvm5ValueEb
+ 180
14  clang                               0x0007ef71
_ZN5clang7CodeGen15CodeGenFunction8EmitStmtEPKNS_4StmtE + 465
15  clang                               0x00081991
_ZN5clang7CodeGen15CodeGenFunction16GenerateFunctionEPKNS_4StmtE + 33
16  clang                               0x00082061
_ZN5clang7CodeGen15CodeGenFunction12GenerateCodeEPKNS_12FunctionDeclE + 673
17  clang                               0x00087f69
_ZN5clang7CodeGen13CodeGenModule12EmitFunctionEPKNS_12FunctionDeclE + 73
18  clang                               0x00091fd2
_ZN46_GLOBAL__N_ModuleBuilder.cpp_00000000_146D646313CodeGenerator18HandleTopLevelDeclEPN5clang4DeclE
+ 898
19  clang                               0x0010aebe
_ZN5clang8ParseASTERNS_12PreprocessorEPNS_11ASTConsumerEb + 366
20  clang                               0x0002a21e
_Z16InitializeGCModeRN5clang11LangOptionsE + 4062
21  clang                               0x0002efcb main + 1803
22  clang                               0x000022f6 start + 54
Abort trap

Note I also tried the following code:

typedef __attribute__(( ext_vector_type(2) )) float float2;
typedef __attribute__(( ext_vector_type(2) )) int int2;

float2 vec2, vec2_2;

void test2() 
{
  int2 ivec = vec2 < vec_2;
}

and got the same error, which seems to imply that it is the application of the
relational operator that is causing the assertion. 

Looking at the error it seems that the problem is with the code generation
phase and so I've assigned it here but I do not have experience with the source
to sure on this point.


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

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  7 09:42:59 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 09:42:59 -0500
Subject: [LLVMbugs] [Bug 2524] New: bogus error when taking the address of a
 noproto function decl
Message-ID: <bug-2524-206@http.llvm.org/bugs/>

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

           Summary: bogus error when taking the address of a noproto
                    function decl
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Windows NT
            Status: NEW
          Keywords: compile-fail
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


clang produces a bogus error with the following code, and gcc does not:

extern void vgPlain_x86_linux_SUBST_FOR_sigreturn;

void build_sigframe() {
        int x = (int)&vgPlain_x86_linux_SUBST_FOR_sigreturn;
}


b.c:4:15: error: address expression must be an lvalue or a function designator
        int x = (int)&vgPlain_x86_linux_SUBST_FOR_sigreturn;
                     ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 diagnostic generated.


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

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  7 10:58:24 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 10:58:24 -0500
Subject: [LLVMbugs] [Bug 2525] New: isConstantExpr() doesn't support C99
	shortcuts
Message-ID: <bug-2525-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2525

           Summary: isConstantExpr() doesn't support C99 shortcuts
           Product: clang
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: AST
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


isConstantExpr() doesn't support C99 shortcuts. test:


extern unsigned int __vki_invalid_size_argument_for_IOC;

void vgSysWrap_generic_sys_ioctl_before ()
{
        switch (1) {
                case (1 ? 1 : __vki_invalid_size_argument_for_IOC): break;
        }

        switch (1) {
                case (1 ?: __vki_invalid_size_argument_for_IOC): break;
        }

        switch (1) {
                case (0 ? __vki_invalid_size_argument_for_IOC : 1): break;
        }

        switch (1) {
                case (0 && __vki_invalid_size_argument_for_IOC): break;
        }

        switch (1) {
                case (1 || __vki_invalid_size_argument_for_IOC): break;
        }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  7 14:08:36 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 14:08:36 -0500
Subject: [LLVMbugs] [Bug 2526] New: Static Analyzer: false-positive memory
	leak
Message-ID: <bug-2526-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2526

           Summary: Static Analyzer: false-positive memory leak
           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: thomascl at free.fr
                CC: llvmbugs at cs.uiuc.edu


I noticed a false-positive memory leak when using the NSAlert class.
The reason is that the NSAlert object can be released later when the
'didEndSelector' is called (the call to beginSheet... is asynchronous).

Please see tha attachment.

Using checker-57.

Thanks


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

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  7 14:18:31 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 14:18:31 -0500
Subject: [LLVMbugs] [Bug 2527] New: Static Analyzer: wrong logic provides
 false-positives
Message-ID: <bug-2527-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2527

           Summary: Static Analyzer: wrong logic provides false-positives
           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: thomascl at free.fr
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1806)
 --> (http://llvm.org/bugs/attachment.cgi?id=1806)
test case

I have seen several false-positives due to code structured that way.
The analyzer evaluates the variable to begin TRUE and FALSE while it is not
modified.

See the attachment.

Using checker-57.

Thanks


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

From bugzilla-daemon at cs.uiuc.edu  Mon Jul  7 15:31:53 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 15:31:53 -0500
Subject: [LLVMbugs] [Bug 2528] New: False positive "dead store" for aliased
	pointer
Message-ID: <bug-2528-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2528

           Summary: False positive "dead store" for aliased pointer
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicolasweber at gmx.de
                CC: llvmbugs at cs.uiuc.edu


The attached file contains the lines

/*10*/    const void *bytes = data;
/*11*/    unsigned bg = *((unsigned*)bytes); bytes += sizeof(unsigned);
/*12*/    printf("%d", bg);

Running `scan_build xcodebuild` with a project containing just the attached
file, clang reports "Value stored to 'bytes' is never read" for line 11.
However, the value is read and copied into `bg`, which is then printed.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  7 15:40:57 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 15:40:57 -0500
Subject: [LLVMbugs] [Bug 2528] False positive "dead store" for aliased
	pointer
In-Reply-To: <bug-2528-206@http.llvm.org/bugs/>
Message-ID: <200807072040.m67Kevvo010252@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2528


Ted Kremenek <kremenek at apple.com> changed:

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




--- Comment #2 from Ted Kremenek <kremenek at apple.com>  2008-07-07 15:40:51 ---
I think you're misreading the bug report.  The warning I see is:

<stdin>:11:40: warning: Value stored to 'bytes' is never read
    unsigned bg = *((unsigned*)bytes); bytes += sizeof(unsigned);
                                       ^        ~~~~~~~~~~~~~~~~
1 diagnostic generated.

The warning is not about the assignment to bg, its about the "+=" performed to
"bytes".  Both the HTML output and console output indicate that the bug has to
do with the expression:

   bytes += sizeof(unsigned);

What happens is that the value of bytes is incremented (which is essentially a
load, addition, store), but then that new value stored to bytes is never read. 
So this isn't a false positive.  There is a dead store here (and a dead
operation).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  7 16:51:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 16:51:09 -0500
Subject: [LLVMbugs] [Bug 2529] New: constant folding fails on vicmp
 instructions with undef operands
Message-ID: <bug-2529-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2529

           Summary: constant folding fails on vicmp instructions with undef
                    operands
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sdt at rapidmind.com
                CC: llvmbugs at cs.uiuc.edu


For example, try passing the following through opt -instcombine:

define <4 x i32> @main(i32 %argc, i8** %argv) {
entry:  
        %foo = vicmp slt <4 x i32> <i32 undef, i32 undef, i32 undef, i32
undef>, <i32 undef, i32 undef, i32 undef, i32 undef>
        ret <4 x i32> %foo
}

This generates the following assertion:
opt: Value.cpp:319: void llvm::Value::replaceAllUsesWith(llvm::Value*):
Assertion `New->getType() == getType() && "replaceAllUses of value with new
value of different type!"' failed.

With the following backtrace:
#4  0x08680fa3 in llvm::Value::replaceAllUsesWith (this=0x87d57c8,
New=0x87d4a00) at Value.cpp:318
#5  0x0843b35e in AddReachableCodeToWorklist (BB=0x87d6a78,
Visited=@0xbfaabfd8, IC=@0x87d5180, TD=0x87d50b0) at
InstructionCombining.cpp:11245

Which is the last line of:
      if (Constant *C = ConstantFoldInstruction(Inst, TD)) {
        DOUT << "IC: ConstFold to: " << *C << " from: " << *Inst;
        Inst->replaceAllUsesWith(C);

It appears the vicmp got folded into an i1 undef, instead of a <4 x i32> undef.

The same problem exists for vfcmp as well.

This was tested on r53196.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  7 19:30:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 7 Jul 2008 19:30:28 -0500
Subject: [LLVMbugs] [Bug 2530] New: -fno-builtin-abs does not
Message-ID: <bug-2530-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2530

           Summary: -fno-builtin-abs does not
           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


GCC supports -fno-builtin-abs and other to selectively disable builtins. This
does not work sometimes because SimplifyLibCalls is not informed about this.

One solution is to use an attribute on call instruction.


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

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  8 00:47:06 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 00:47:06 -0500
Subject: [LLVMbugs] [Bug 2529] constant folding fails on vicmp/ vfcmp
 instructions with undef operands
In-Reply-To: <bug-2529-206@http.llvm.org/bugs/>
Message-ID: <200807080547.m685l5sk028955@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2529


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-07-08 00:47:04 ---
Fixed, thanks!
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064662.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 Jul  8 05:13:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 05:13:09 -0500
Subject: [LLVMbugs] [Bug 2520] fabs not marked readnone
In-Reply-To: <bug-2520-206@http.llvm.org/bugs/>
Message-ID: <200807081013.m68AD9bY016482@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2520


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #3 from Duncan Sands <baldrick at free.fr>  2008-07-08 05:13:06 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064680.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 Jul  8 09:25:50 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 09:25:50 -0500
Subject: [LLVMbugs] [Bug 2531] New: static analyzer doesn't understand
	va_start
Message-ID: <bug-2531-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2531

           Summary: static analyzer doesn't understand va_start
           Product: clang
           Version: unspecified
          Platform: Other
        OS/Version: FreeBSD
            Status: NEW
          Severity: major
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: cperciva at freebsd.org
                CC: llvmbugs at cs.uiuc.edu


Running scan-build on code which uses va_start results in a false positive bug
report of "Pass-by-value argument in function is undefined."

Presumably the relevant code just needs to be taught that va_start and va_end
are "special".


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  8 09:57:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 09:57:09 -0500
Subject: [LLVMbugs] [Bug 2532] New: llvm-gcc compilation for arm failed in
	llc
Message-ID: <bug-2532-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2532

           Summary: llvm-gcc compilation for arm failed in llc
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: patrice.gerin at imag.fr
                CC: llvmbugs at cs.uiuc.edu


llvm-gcc cross compilation for ARM target failed in libgcc2.c to generate
_muldc3.o

cc1: ARMISelLowering.cpp:657: llvm::SDOperand LowerRET(llvm::SDOperand,
llvm::SelectionDAG&): Assertion `0 && "Do not know how to return this many
arguments!"' failed.
../../../gcc/libgcc2.c:1914: internal compiler error: Aborted

I emit the LLVM _muldc3.bc (in attachment) file check using directly llc

llc -march=arm _muldc3.bc 
llc: ARMISelLowering.cpp:657: llvm::SDOperand LowerRET(llvm::SDOperand,
llvm::SelectionDAG&): Assertion `0 && "Do not know how to return this many
arguments!"' failed.
/home/gerin/opt/llvm/bin/llc[0x8905b84]
/lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7d3f641]
/lib/tls/i686/cmov/libc.so.6(__assert_fail+0xfb)[0xb7d3743b]
/home/gerin/opt/llvm/bin/llc[0x826191f]


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  8 12:20:52 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 12:20:52 -0500
Subject: [LLVMbugs] [Bug 2496] volatile optimization bug
In-Reply-To: <bug-2496-206@http.llvm.org/bugs/>
Message-ID: <200807081720.m68HKqGm032127@zion.cs.uiuc.edu>

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


Chris Lattner <sabre at nondot.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED
            Summary|likely volatile bug         |volatile optimization bug




--- Comment #4 from Chris Lattner <sabre at nondot.org>  2008-07-08 12:20:41 ---
Fixed: again, very nice and subtle catch :)
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064686.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 Jul  8 13:09:33 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 13:09:33 -0500
Subject: [LLVMbugs] [Bug 2529] constant folding fails on vicmp/ vfcmp
 instructions with undef operands
In-Reply-To: <bug-2529-206@http.llvm.org/bugs/>
Message-ID: <200807081809.m68I9X8E001580@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2529


Stefanus Du Toit <sdt at rapidmind.com> changed:

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




--- Comment #2 from Stefanus Du Toit <sdt at rapidmind.com>  2008-07-08 13:08:06 ---
Thanks for the speedy fix. This fixed vicmp, but vfcmp still fails for me
(r53241):

define <4 x i32> @test(i32 %argc, i8** %argv) {
entry:  
        %foo = vfcmp ueq <4 x float> <float undef, float undef, float undef,
float undef>, <float undef, float undef, float undef, float undef>
        ret <4 x i32> %foo
}

opt -instcombine crashes in the same spot.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  8 16:20:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 16:20:39 -0500
Subject: [LLVMbugs] [Bug 2533] New: Code gen for x86_64 broken,
	example included
Message-ID: <bug-2533-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2533

           Summary: Code gen for x86_64 broken, example included
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: ABI, miscompilation
          Severity: major
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: thomas.knauth at gmx.de
                CC: llvmbugs at cs.uiuc.edu


#include <stdio.h>
#include <inttypes.h>

void dummy(const char* a) {}

typedef struct {
  const char *name;
  int flags;
  union {
    void (*func_arg)(const char *); //FIXME passing error code as int return
would be nicer then exit() in the func
  } u;
} OptionDef;

const OptionDef options[] = {
  /* main options */
  { "a", 0, {dummy} },
  { "b", 0, {dummy} },
  { 0, },
};

int main(int argc, char **argv)
{
  const OptionDef* po;
  for (po=options; po->name!=0; po++) {
    printf("%s\n", po->name);
  }
  return 0;
}

This code should iterate over all elements in 'options'. The code generated
with llvm-gcc (2.3 stable) however, does not do this. When compiled with gcc,
everything works as expected. On x86_32, llvm-gcc generates correct code.


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

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  8 18:09:12 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 18:09:12 -0500
Subject: [LLVMbugs] [Bug 2513] invalid bitcode bug - loop unswitch dom info
In-Reply-To: <bug-2513-206@http.llvm.org/bugs/>
Message-ID: <200807082309.m68N9CDE012429@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2513


Devang Patel <dpatel at apple.com> changed:

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




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

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  8 18:09:27 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 18:09:27 -0500
Subject: [LLVMbugs] [Bug 2455] loop unswitch dom bug
In-Reply-To: <bug-2455-206@http.llvm.org/bugs/>
Message-ID: <200807082309.m68N9RaC012456@zion.cs.uiuc.edu>

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


Devang Patel <dpatel at apple.com> changed:

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




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

From bugzilla-daemon at cs.uiuc.edu  Tue Jul  8 19:12:49 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 19:12:49 -0500
Subject: [LLVMbugs] [Bug 2487] loop index split codegen bug
In-Reply-To: <bug-2487-206@http.llvm.org/bugs/>
Message-ID: <200807090012.m690CnAI014933@zion.cs.uiuc.edu>

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


Devang Patel <dpatel at apple.com> changed:

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




--- Comment #2 from Devang Patel <dpatel at apple.com>  2008-07-08 19:12:48 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064715.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 Jul  8 23:57:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 8 Jul 2008 23:57:28 -0500
Subject: [LLVMbugs] [Bug 2534] New: Unnamed bitfield causes warning
Message-ID: <bug-2534-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2534

           Summary: Unnamed bitfield causes warning
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: andersca at mac.com
                CC: llvmbugs at cs.uiuc.edu


typedef struct {
        int a; int : 24; char b;
} S;

S a = { 1, 2 };

results in 

test.c:5:12: warning: excess elements in array initializer
S a = { 1, 2 };
           ^
1 diagnostic generated.

which is wrong.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 00:31:27 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 00:31:27 -0500
Subject: [LLVMbugs] [Bug 2535] New: Off-by-one bug with codegen of induction
	variable
Message-ID: <bug-2535-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2535

           Summary: Off-by-one bug with codegen of induction variable
           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:

; 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"

@.str = internal constant [4 x i8] c"%d\0A\00"

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

define i32 @main() nounwind {
entry:
        br label %forbody

forbody:
        %i.0 = phi i32 [ 0, %entry ], [ %inc, %forbody ]                ; <i32>
[#uses=3]
        %sub14 = sub i32 1027, %i.0             ; <i32> [#uses=1]
        %mul15 = mul i32 %sub14, 10             ; <i32> [#uses=1]
        %add166 = or i32 %mul15, 1              ; <i32> [#uses=1] *
        call i32 (i8*, ...)* @printf( i8* noalias  getelementptr ([4 x i8]*
@.str, i32 0, i32 0), i32 %add166 ) nounwind
        %inc = add i32 %i.0, 1          ; <i32> [#uses=3]
        %cmp = icmp ult i32 %inc, 1027          ; <i1> [#uses=1]
        br i1 %cmp, label %forbody, label %afterfor

afterfor:               ; preds = %forcond
        ret i32 0
}

This should produce 1027 lines of output, and does using the C backend. 
However, using the X86 backend (i.e. just plain "llc"), the produced output is
1028 lines.

This was a serious pain to reduce, so whoever fixes this should get it right!
:)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 01:01:03 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 01:01:03 -0500
Subject: [LLVMbugs] [Bug 2536] New: possible integer codegen error
Message-ID: <bug-2536-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2536

           Summary: possible integer codegen error
           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 on svn 53265 on Ubuntu Feisty on ia32.

Apologies for the large test input, automatic and manual techniques were not
successful in generating a smaller one.

[regehr at babel tmp1]$ llvm-gcc -O1 -fwrapv small.c -o small
[regehr at babel tmp1]$ ./small
4
[regehr at babel tmp1]$ llvm-gcc -Os -fwrapv small.c -o small
[regehr at babel tmp1]$ ./small
0

[regehr at babel tmp1]$ cat small.c

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

typedef int int32_t;
typedef unsigned short int uint16_t;
typedef unsigned int uint32_t;

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

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

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

uint16_t g_5 = 4L;
int32_t g_104;
uint16_t g_107;
uint16_t g_227;
uint32_t g_229;
int32_t g_241;
uint16_t g_307;

int32_t func_65 (void);
int32_t func_65 (void)
{
  if (rshift_u_s (g_5, 1))
    g_107 = g_5;
  for (1; g_104; --g_104);
  return g_229;
}

int32_t func_54 (uint32_t p_55, uint16_t p_56);
int32_t func_54 (uint32_t p_55, uint16_t p_56)
{
  for (p_55 = 17; p_55; --p_55)
    if ((func_65 ()) * ((1 % mod_rhs (p_56)) || g_227))
      p_56 = 1;
  return 1;
}

int32_t func_28 (uint16_t p_29);
int32_t func_28 (uint16_t p_29)
{
  g_307 || ((0 >= p_29) / div_rhs (func_54 (g_241, 1) <= g_5))
    && (func_65 ());
  return 1;
}

int main (void)
{
  func_28 (1);
  printf ("%d\n", g_107);
  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  Wed Jul  9 09:10:52 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 09:10:52 -0500
Subject: [LLVMbugs] [Bug 2024] Problem with ELF section attribute
In-Reply-To: <bug-2024-206@http.llvm.org/bugs/>
Message-ID: <200807091410.m69EAq4T021603@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2024


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

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




--- Comment #5 from Anton Korobeynikov <asl at math.spbu.ru>  2008-07-09 09:10:51 ---
Fixed with recent series of section names handling.

Testcase is here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064786.html

support for other (non-x86) targets will follow.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 11:08:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 11:08:09 -0500
Subject: [LLVMbugs] [Bug 2525] isConstantExpr() doesn't support C99 shortcuts
In-Reply-To: <bug-2525-206@http.llvm.org/bugs/>
Message-ID: <200807091608.m69G89A6025914@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2525


Eli Friedman <sharparrow1 at yahoo.com> changed:

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




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2008-07-09 11:08:03 ---
>From C99 6.5p6:
"An integer constant expression shall have integer type and shall only have
operands that are integer constants, enumeration constants, character
constants, sizeof expressions whose results are integer constants, and floating
constants that are the immediate operands of casts."

A global variable is none of the above.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 12:08:01 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 12:08:01 -0500
Subject: [LLVMbugs] [Bug 2525] isConstantExpr() doesn't support C99 shortcuts
In-Reply-To: <bug-2525-206@http.llvm.org/bugs/>
Message-ID: <200807091708.m69H81ee028365@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2525


Nuno Lopes <nunoplopes at sapo.pt> changed:

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




--- Comment #2 from Nuno Lopes <nunoplopes at sapo.pt>  2008-07-09 12:07:57 ---
GCC accepts all those constructs (without warnings) and valgrind relies on that
behaviour. Try to compile valgrind yourself with clang to see the tons of
errors.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 13:26:07 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 13:26:07 -0500
Subject: [LLVMbugs] [Bug 2533] Code gen for x86_64 broken, example included
In-Reply-To: <bug-2533-206@http.llvm.org/bugs/>
Message-ID: <200807091826.m69IQ79s031906@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2533


Dale Johannesen <dalej at apple.com> changed:

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




--- Comment #3 from Dale Johannesen <dalej at apple.com>  2008-07-09 13:26:05 ---
Devang says it was a testsuite issue, and if that still works we don't need the
patch any more.  Gone.
http://llvm.org/viewvc/llvm-project?view=rev&revision=53338


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 13:26:49 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 13:26:49 -0500
Subject: [LLVMbugs] [Bug 2519] scan-build mishandles CFDictionaryCreate
In-Reply-To: <bug-2519-206@http.llvm.org/bugs/>
Message-ID: <200807091826.m69IQneJ031966@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2519


Ted Kremenek <kremenek at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
          Component|Semantic Analyzer           |Static Analyzer
         Resolution|                            |FIXED
            Summary|scan-build ignores CFRelease|scan-build mishandles
                   |                            |CFDictionaryCreate




--- Comment #1 from Ted Kremenek <kremenek at apple.com>  2008-07-09 13:26:48 ---
The bug was that the analyzer wasn't properly handling the keys and values
being passed to CFDictionaryCreate.  _value was being passed by reference, and
its value was invalidated by the object referred to was considered leaked.

This is now fixed:

http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080707/006438.html

The change is in checker-59 (now available).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 13:29:20 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 13:29:20 -0500
Subject: [LLVMbugs] [Bug 2531] static analyzer doesn't understand va_start
In-Reply-To: <bug-2531-206@http.llvm.org/bugs/>
Message-ID: <200807091829.m69ITKjs032061@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2531


Colin Percival <cperciva at freebsd.org> changed:

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




--- Comment #13 from Colin Percival <cperciva at freebsd.org>  2008-07-09 13:29:19 ---
Thanks!  I can confirm that ccc now compiles this and scan-build doesn't
complain about va_start on FreeBSD.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 13:40:17 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 13:40:17 -0500
Subject: [LLVMbugs] [Bug 2518] scan-build treats -[NSAutoreleasePool drain]
 as -[ NSAutoreleasePool release]
In-Reply-To: <bug-2518-206@http.llvm.org/bugs/>
Message-ID: <200807091840.m69IeHP4032464@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2518


Ted Kremenek <kremenek at apple.com> changed:

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




--- Comment #4 from Ted Kremenek <kremenek at apple.com>  2008-07-09 13:40:16 ---
> I.e. -drain does everything -release does except for actually releasing the
> autorelease pool. 

I talked to the main GC implementor.  He told me that -drain should be treated
exactly as a -release when using autorelease pools in non-GC mode.

If your code works, it is an artifact of the current implementation (which
Apple can change at any time in a future OS release).

The GC guy also said that your code example illustrates a common bug when
people use -drain, and that the static analysis tool actually enforces the
correct specification.

> In short, -drain does not -release the pool. 

I can see why using -drain in this matter would be useful, but it is the case
that -drain releases the pool.

As far as retain counts, auto release pools shouldn't really be viewed as
normal objects.  That's why the -retain method is invalid for
NSAutoreleasePool.  The fact that the retain count appears to stay 1 is also is
an artifact of the implementation.  It should probably be the case that
-retainCount, like -retain, also raises a runtime error for NSAutoreleasePool. 
I'll file this as a feature request to the Apple runtime team.

> If it did so, it???d be worthless.

I think the reason that -drain works this way is to both match scope with
allocation/deallocation and to maintain parity with -release.  It seems to me
that you could easily rewrite your example as:

while(???) {
   NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init];
  ???do some stuff???
  [pool drain];
}

Thanks for reporting this bug.  It's good to get important API rules like this
clarified.  I was fully prepared to accept that the tool reported a false
positive; the fact that it didn't means that people commonly have a
misconception about how -drain should be used.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul  9 19:46:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 19:46:48 -0500
Subject: [LLVMbugs] [Bug 2529] constant folding fails on vicmp/ vfcmp
 instructions with undef operands
In-Reply-To: <bug-2529-206@http.llvm.org/bugs/>
Message-ID: <200807100046.m6A0km1q012713@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2529


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #4 from Chris Lattner <sabre at nondot.org>  2008-07-09 19:46:48 ---
I've fixed the issues you identified.  If you hit more cases, please open a new
bug and include testcases, rather than reopening this one.  Thanks,

-Chris

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064837.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064840.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064841.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 Jul  9 20:08:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 20:08:48 -0500
Subject: [LLVMbugs] [Bug 2472] Missing VZEXT_MOVL pattern for <4 x i32>
In-Reply-To: <bug-2472-206@http.llvm.org/bugs/>
Message-ID: <200807100108.m6A18m4p013561@zion.cs.uiuc.edu>

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


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

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




--- Comment #2 from Evan Cheng <evan.cheng at apple.com>  2008-07-09 20:08:48 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064843.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 Jul  9 23:50:20 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 9 Jul 2008 23:50:20 -0500
Subject: [LLVMbugs] [Bug 2406] redefine type: segfault
In-Reply-To: <bug-2406-206@http.llvm.org/bugs/>
Message-ID: <200807100450.m6A4oKI0020875@zion.cs.uiuc.edu>

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


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #5 from Chris Lattner <sabre at nondot.org>  2008-07-09 23:50:20 ---
closing as 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  Thu Jul 10 02:03:18 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 10 Jul 2008 02:03:18 -0500
Subject: [LLVMbugs] [Bug 2497] llvm-gcc segfault
In-Reply-To: <bug-2497-206@http.llvm.org/bugs/>
Message-ID: <200807100703.m6A73I0w025573@zion.cs.uiuc.edu>

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


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #2 from Chris Lattner <sabre at nondot.org>  2008-07-10 02:03:18 ---
Fixed by devang's dom hacking.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 10 09:58:36 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 10 Jul 2008 09:58:36 -0500
Subject: [LLVMbugs] [Bug 2537] New: -loop-reduce crashes with extremely
	large stride
Message-ID: <bug-2537-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2537

           Summary: -loop-reduce crashes with extremely large stride
           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"

define void @a() {
entry:
        br label %dobody

dobody:         ; preds = %dobody, %entry
        %y.0 = phi i128 [ 0, %entry ], [ %add, %dobody ]
        %x.0 = phi i128 [ 0, %entry ], [ %add2, %dobody ]
        %add = add i128 %y.0, shl (i128 1, i128 64)
        %add2 = add i128 %x.0, shl (i128 1, i128 48)
        call void @b( i128 %add )
        %cmp = icmp ult i128 %add2, shl (i128 1, i128 64)
        br i1 %cmp, label %dobody, label %afterdo

afterdo:                ; preds = %dobody
        ret void
}

declare void @b(i128)

This blows up in -loop-reduce because it blindly uses getSExtValue() without
checking the size of the value.


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

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 10 10:08:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 10 Jul 2008 10:08:30 -0500
Subject: [LLVMbugs] [Bug 2538] New: llvm-as gives an error on extremely
	large constants
Message-ID: <bug-2538-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2538

           Summary: llvm-as gives an error on extremely large constants
           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"

define void @a() {
entry:
        br label %dobody

dobody:         ; preds = %dobody, %entry
        %y.0 = phi i128 [ 0, %entry ], [ %add, %dobody ]
        %x.0 = phi i128 [ 0, %entry ], [ %add2, %dobody ]
        %add = add i128 %y.0, shl (i128 1, i128 64)
        %add2 = add i128 %x.0, shl (i128 1, i128 48)
        call void @b( i128 %add )
        %cmp = icmp ult i128 %add2, shl (i128 1, i128 64)
        br i1 %cmp, label %dobody, label %afterdo

afterdo:                ; preds = %dobody
        ret void
}

To reproduce, run llvm-as | llvm-dis | llvm-as, and note that llvm-as rejects
the output of llvm-dis with the following error:
/home/eli/llvmbin/Debug/bin/llvm-as: <stdin>:12,0: <stdin>:12: error: syntax
error, unexpected EUAPINTVAL while reading token: '18446744073709551616'


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 10 11:32:41 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 10 Jul 2008 11:32:41 -0500
Subject: [LLVMbugs] [Bug 2539] New: possible integer codegen error
Message-ID: <bug-2539-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2539

           Summary: possible integer codegen error
           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


>From svn 53390 on Ubuntu Feisty on ia32.

[regehr at babel tmp2]$ llvm-gcc -O1 -fwrapv small.c -o small
[regehr at babel tmp2]$ ./small
1
[regehr at babel tmp2]$ llvm-gcc -O2 -fwrapv small.c -o small
[regehr at babel tmp2]$ ./small
0

[regehr at babel tmp2]$ cat small.c
#include <stdint.h>
#include <stdio.h>

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

int16_t g_6;
uint8_t g_155;
int8_t g_156;

int32_t func_87 (void);
int32_t func_87 (void)
{
  return 1;
}

int32_t func_146 (void);
int32_t func_146 (void)
{
  int8_t p_147;
  for (p_147 = 0; 1; p_147 -= 1)

    if (p_147 >= ((func_87 ()) < (p_147 % mod_rhs (g_155))))
      {
        g_156 = 1;
        for (p_147 = 1; p_147; p_147 += 1)
          return 1;
      }
    else
      return 1;  
}

int main (void)
{
  func_146 ();
  printf ("%d\n", g_156);
  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 Jul 10 13:25:08 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 10 Jul 2008 13:25:08 -0500
Subject: [LLVMbugs] [Bug 2540] New: possible integer codegen error
Message-ID: <bug-2540-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2540

           Summary: possible integer codegen error
           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 53390 again, on Ubuntu Feisty on ia32.  Could easily be the
same bug as 2536 and/or 2539 but I figured I'd be safe and submit it.

[regehr at babel tmp3]$ llvm-gcc -O2 -fwrapv small.c -o small
[regehr at babel tmp3]$ ./small
0
[regehr at babel tmp3]$ llvm-gcc -O3 -fwrapv small.c -o small
[regehr at babel tmp3]$ ./small
1

[regehr at babel tmp3]$ cat small.c
#include <stdint.h>
#include <limits.h>
#include <stdio.h>

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


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


uint32_t g_37;
uint32_t g_73;
int16_t g_98 = 1;
uint32_t g_228;
int32_t g_246;
int8_t g_308;
int32_t func_1 (void);
int32_t func_116 (uint32_t p_117, uint32_t p_119, uint32_t p_120,
                  uint32_t p_121);
int32_t func_132 (int16_t p_133, uint32_t p_134, uint32_t p_135);
int32_t
func_1 (void)
{
  uint32_t l_311 = 0xA9D412B1L;
  int16_t l_316 = 1;
  for (g_73 = 1; g_73; --g_73)
    {
      int32_t l_310 = 0;
      uint32_t l_313 = 1;
      uint32_t l_318 = 1;
      for (g_73 = 0; g_73; g_73 += 1)
        {
        }
      if (g_98 +
          ((rshift_s_s (l_310, 1)) >=
           (((1 !=
              (1 & l_310)) % mod_rhs (l_311)) % mod_rhs (func_132 ((1 +
                                                                    g_246), 1,
                                                                   (1 <
                                                                    (1 |
                                                                     l_311) +
                                                                    (g_228
                                                                     &&
                                                                    
l_313)))))))
        {
          uint32_t l_314 = 1;
          func_116 ((1 <=
                     ((7L * l_314) %
                      mod_rhs (1 !=
                               (lshift_s_u (1, g_37)) * (l_316
                                                         || g_73)))), l_318,
                    g_308, 1);
        }
      else
        return 1;
    }
  return 1;
}

int32_t
func_116 (uint32_t p_117, uint32_t p_119, uint32_t p_120, uint32_t p_121)
{
  if (p_117)
    {
    }
  else
    for (g_98 = 0; g_98; ++g_98)
      {
      }
  return 1;
}

int32_t
func_132 (int16_t p_133, uint32_t p_134, uint32_t p_135)
{
  uint32_t l_138 = 0x1206B3A0L;
  return l_138;
}

int
main (void)
{
  func_1 ();
  printf ("%d\n", g_98);
  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 Jul 10 14:39:17 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 10 Jul 2008 14:39:17 -0500
Subject: [LLVMbugs] [Bug 2541] New: Olden benchmark perimeter fails with
	llvm-2.3
Message-ID: <bug-2541-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2541

           Summary: Olden benchmark perimeter fails with llvm-2.3
           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: santosh.nagarakatte at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1821)
 --> (http://llvm.org/bugs/attachment.cgi?id=1821)
gzipped source of the benchmark

A version of Olden benchmark perimeter when compiled with llvm-gcc-4.2 with
llvm-2.3 aborts with the following error

llvm/llvm-2.3/lib/CodeGen/LiveVariables.cpp:143: void
llvm::LiveVariables::HandleVirtRegUse(unsigned int, llvm::MachineBasicBlock*,
llvm::MachineInstr*): Assertion `MBB != MRI->getVRegDef(reg)->getParent() &&
"Should have kill for defblock!"' failed.

Command-Line used to compile the benchmark

llvm-gcc -m64 -O3  -lm -DPLAIN -DSS_PLAIN -DI_TIME -DI_SYS_TIME -DULTRIX -o
test *.c

Same benchmark was compiling successfully with llvm-2.2.

I am using gcc-4.2.0 on X86-64 machine with Suse linux.

I have attached the c source for perusal


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 10 19:31:00 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 10 Jul 2008 19:31:00 -0500
Subject: [LLVMbugs] [Bug 2538] llvm-as gives an error on extremely large
	constants
In-Reply-To: <bug-2538-206@http.llvm.org/bugs/>
Message-ID: <200807110031.m6B0V072007871@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2538


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-07-10 19:31:00 ---
done:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064902.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 Jul 11 00:41:41 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 11 Jul 2008 00:41:41 -0500
Subject: [LLVMbugs] [Bug 2539] possible integer codegen error
In-Reply-To: <bug-2539-206@http.llvm.org/bugs/>
Message-ID: <200807110541.m6B5ff5M017065@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2539


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #11 from Chris Lattner <sabre at nondot.org>  2008-07-11 00:41:41 ---
This is now fixed, I also refactored some pieces of instcombine to simplify and
clean it up.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 11 02:57:52 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 11 Jul 2008 02:57:52 -0500
Subject: [LLVMbugs] [Bug 2543] New: llc failed with MIPS architecture
Message-ID: <bug-2543-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2543

           Summary: llc failed with MIPS architecture
           Product: tools
           Version: 2.3
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: patrice.gerin at imag.fr
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1827)
 --> (http://llvm.org/bugs/attachment.cgi?id=1827)
A simple  .c file

When I compile the simple attached C file, llc seems to enter an infinite loop
with -march=mips.

Here after are the command I write:

llvm-gcc -emit-llvm -c main.c -o main.bc
llc -march=mips -debug main.bc -f -o main.mips.s

I identify that the infinite loop is the while() in the
LatencyPriorityQueue::CalculatePriorities() method (file
CodeGen/SelectionDAG/ScheduleDAGList.cpp). Some SUnit are added continuously to
the WorkList ...

I attach the main.c and main.bc files.

void LatencyPriorityQueue::CalculatePriorities() {
  Latencies.assign(SUnits->size(), -1);
  NumNodesSolelyBlocking.assign(SUnits->size(), 0);

  // For each node, calculate the maximal path from the node to the exit.
  std::vector<std::pair<const SUnit*, unsigned> > WorkList;
  for (unsigned i = 0, e = SUnits->size(); i != e; ++i) {
    const SUnit *SU = &(*SUnits)[i];
    if (SU->Succs.empty())
      WorkList.push_back(std::make_pair(SU, 0U));
  }

  while (!WorkList.empty()) {
    const SUnit *SU = WorkList.back().first;
    unsigned SuccLat = WorkList.back().second;
    WorkList.pop_back();
    int &Latency = Latencies[SU->NodeNum];
    if (Latency == -1 || (SU->Latency + SuccLat) > (unsigned)Latency) {
      Latency = SU->Latency + SuccLat;
      for (SUnit::const_pred_iterator I = SU->Preds.begin(),E =
SU->Preds.end();
           I != E; ++I)
        WorkList.push_back(std::make_pair(I->Dep, Latency));
    }
  }
}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 11 07:58:37 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 11 Jul 2008 07:58:37 -0500
Subject: [LLVMbugs] [Bug 2544] New: vector uitofp does not select on x86
	with SSE2
Message-ID: <bug-2544-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2544

           Summary: vector uitofp does not select on x86 with SSE2
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sdt at rapidmind.com
                CC: llvmbugs at cs.uiuc.edu


The following IR will not compile with -march=x86 -mattr=+sse2:

@foo = global <4 x i32> undef

define <4 x float> @test() {
entry:
  %baz = load <4 x i32>* @foo
  %bar = uitofp <4 x i32> %baz to <4 x float>
  ret <4 x float> %bar
}

The error is:

Cannot yet select: 0x8dfe4d4: v4f32 = uint_to_fp 0x8dfe714

This looks quite simply like ui to fp conversion is not implemented on vectors.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 11 12:38:40 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 11 Jul 2008 12:38:40 -0500
Subject: [LLVMbugs] [Bug 2543] llc failed with MIPS architecture
In-Reply-To: <bug-2543-206@http.llvm.org/bugs/>
Message-ID: <200807111738.m6BHceG3017506@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2543


Bruno Cardoso Lopes <bruno.cardoso at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bruno.cardoso at gmail.com
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #2 from Bruno Cardoso Lopes <bruno.cardoso at gmail.com>  2008-07-11 12:38:39 ---
Hi GERIN,

This bug was solved by Jun 3 and didn't go into 2.3 branch, but is solved in
trunk. Be welcome to report mips bugs! :)

If you're curious, the bug is:
http://llvm.org/viewvc/llvm-project?view=rev&revision=51899

Since then, there were lots of changes into mips backend, I would suggest using
the trunk.


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

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 11 15:53:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 11 Jul 2008 15:53:28 -0500
Subject: [LLVMbugs] [Bug 2532] ARM: i128 result (aka double complex) not
	supported
In-Reply-To: <bug-2532-206@http.llvm.org/bugs/>
Message-ID: <200807112053.m6BKrSaf023686@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2532


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #11 from Chris Lattner <sabre at nondot.org>  2008-07-11 15:53:27 ---
Please give this a try:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064941.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 Jul 11 20:57:01 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 11 Jul 2008 20:57:01 -0500
Subject: [LLVMbugs] [Bug 2536] regalloc subreg miscompilation
In-Reply-To: <bug-2536-206@http.llvm.org/bugs/>
Message-ID: <200807120157.m6C1v1U7001792@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2536


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

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




--- Comment #14 from Evan Cheng <evan.cheng at apple.com>  2008-07-11 20:57:00 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064959.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 Jul 12 02:01:06 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 12 Jul 2008 02:01:06 -0500
Subject: [LLVMbugs] [Bug 2171] using iplist<T>::iterator operator[] should
 cause a compile error
In-Reply-To: <bug-2171-206@http.llvm.org/bugs/>
Message-ID: <200807120701.m6C7169t010720@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2171


Nick Lewycky <nicholas at mxc.ca> changed:

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




--- Comment #2 from Nick Lewycky <nicholas at mxc.ca>  2008-07-12 02:01:05 ---
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  Sat Jul 12 02:05:44 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 12 Jul 2008 02:05:44 -0500
Subject: [LLVMbugs] [Bug 2070] Indvars makes loop more complicated
In-Reply-To: <bug-2070-206@http.llvm.org/bugs/>
Message-ID: <200807120705.m6C75isE010895@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2070


Nick Lewycky <nicholas at mxc.ca> changed:

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




--- Comment #7 from Nick Lewycky <nicholas at mxc.ca>  2008-07-12 02:05:44 ---


*** This bug has been marked as a duplicate of bug 2261 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 12 02:42:37 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 12 Jul 2008 02:42:37 -0500
Subject: [LLVMbugs] [Bug 2261] SCEV creates unneeded SMAX node
In-Reply-To: <bug-2261-206@http.llvm.org/bugs/>
Message-ID: <200807120742.m6C7gbfb015602@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2261


Nick Lewycky <nicholas at mxc.ca> changed:

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




--- Comment #2 from Nick Lewycky <nicholas at mxc.ca>  2008-07-12 02:42:36 ---
Fix checked in.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 12 11:06:55 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 12 Jul 2008 11:06:55 -0500
Subject: [LLVMbugs] [Bug 2545] New: THUMB_DIV_MOD_BODY problem with standard
 arm assembler when building llvm-gcc frontend
Message-ID: <bug-2545-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2545

           Summary: THUMB_DIV_MOD_BODY problem with standard arm assembler
                    when building llvm-gcc frontend
           Product: tools
           Version: 2.3
          Platform: Other
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: treat at kde.org
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1830)
 --> (http://llvm.org/bugs/attachment.cgi?id=1830)
Fix for thumb_div_mod_body

Hi,

When building the llvm-gcc frontend for an ARM cross compiler I ran into the
following error:

-c ../../llvm-gcc4.2-2.3.source/gcc/config/arm/lib1funcs.asm -o
libgcc/thumb/_udivsi3.o
../../llvm-gcc4.2-2.3.source/gcc/config/arm/lib1funcs.asm: Assembler messages:
../../llvm-gcc4.2-2.3.source/gcc/config/arm/lib1funcs.asm:847: Error: bad
instruction `thumb_div_mod_body 0'

It seems that apple's arm assembler behaves differently.  Changing
THUMB_DIV_MOD_BODY 0 into THUMB_DIV_MOD_BODY(0) seems to fix the problem when
using the standard arm assembler from binutils.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 13 01:49:24 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 13 Jul 2008 01:49:24 -0500
Subject: [LLVMbugs] [Bug 2546] New: create SCEVInfinite
Message-ID: <bug-2546-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2546

           Summary: create SCEVInfinite
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: Loop Optimizer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


Currently we use SCEVCouldNotCompute for cases where the analysis failed and
when the loop was successfully analyzed as being infinite. We should split
those apart and return a new SCEVInfinite type for loops that are provable
infinite.

Internally, a SCEVInfinite early on should skip the brute force loop count
computation. Once we're no longer wasting time brute forcing infinite loops, we
can try cranking up the number of brute force iterations.

Also, some passes, like loop deletion, might want to see SCEVInfinite.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 13 14:35:42 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 13 Jul 2008 14:35:42 -0500
Subject: [LLVMbugs] [Bug 2231] opt -internalize -std-compile-opts should run
 internalize first
In-Reply-To: <bug-2231-206@http.llvm.org/bugs/>
Message-ID: <200807131935.m6DJZgGb026147@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2231


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-07-13 14:35:42 ---
Implemented:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064970.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 Jul 13 17:30:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 13 Jul 2008 17:30:48 -0500
Subject: [LLVMbugs] [Bug 2540] possible integer codegen error
In-Reply-To: <bug-2540-206@http.llvm.org/bugs/>
Message-ID: <200807132230.m6DMUmRg032147@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2540


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #6 from Chris Lattner <sabre at nondot.org>  2008-07-13 17:30:47 ---
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064980.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 Jul 13 19:16:10 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 13 Jul 2008 19:16:10 -0500
Subject: [LLVMbugs] [Bug 2506] Invalid instcombine with select+div
In-Reply-To: <bug-2506-206@http.llvm.org/bugs/>
Message-ID: <200807140016.m6E0GA8G004183@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2506


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #7 from Chris Lattner <sabre at nondot.org>  2008-07-13 19:16:10 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064982.html

Thanks 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  Sun Jul 13 19:47:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 13 Jul 2008 19:47:19 -0500
Subject: [LLVMbugs] [Bug 2486] llvm incorrectly inlines functions containing
	setjmp
In-Reply-To: <bug-2486-206@http.llvm.org/bugs/>
Message-ID: <200807140047.m6E0lJiq005132@zion.cs.uiuc.edu>

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


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #3 from Chris Lattner <sabre at nondot.org>  2008-07-13 19:47:18 ---
Fixed, thanks:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080707/064988.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 Jul 13 23:15:03 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 13 Jul 2008 23:15:03 -0500
Subject: [LLVMbugs] [Bug 2547] New: Implement support for Type Based Alias
	Analysis
Message-ID: <bug-2547-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2547

           Summary: Implement support for Type Based Alias Analysis
           Product: libraries
           Version: 1.0
          Platform: PC
        OS/Version: All
            Status: NEW
          Keywords: code-quality
          Severity: normal
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sabre at nondot.org
                CC: llvmbugs at cs.uiuc.edu


TBAA is a feature that llvm has never supported, and it is capable of providing
significant speedups.  For example, here is the performance of
SingleSource/Benchmarks/Himeno with various compilers:

gcc 4.0 -O3: 4.69
gcc 4.0 -O3 -fno-strict-aliasing: 5.93
gcc 4.2 -O3: 4.54
gcc 4.2 -O3 -fno-strict-aliasing: 6.14
llvm-gcc -O3: 5.43 [which is no strict aliasing]
llvm-gcc -O4: 3.46 [which is no strict aliasing]

As you can see, at -O3, we lose to GCC by ~20%.  Empirically, looking at the
code, this is because loads aren't getting hoisted out of loops, which makes
strides loop variant, which prevents LSR from eliminating multiplies.  -O4/LTO
makes up for the lack of TBAA in this case with extra inlining etc.

An interesting detail is that TBAA is disabled by default in GCC on the mac,
but enabled by default everywhere else.  This is probably why our performance
is generally seen as worse (compared to GCC) on linux than it is on the mac.

-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  Mon Jul 14 00:18:06 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 14 Jul 2008 00:18:06 -0500
Subject: [LLVMbugs] [Bug 2317] vicmp/vfcmp constant exprs not being
 read/written correctly
In-Reply-To: <bug-2317-206@http.llvm.org/bugs/>
Message-ID: <200807140518.m6E5I6kE012944@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2317


Chris Lattner <sabre at nondot.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED
            Summary|new vicmp/vfcmp constant    |vicmp/vfcmp constant exprs
                   |exprs not being read/written|not being read/written
                   |correctly?                  |correctly




--- Comment #3 from Chris Lattner <sabre at nondot.org>  2008-07-14 00:18:05 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/064992.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 Jul 14 00:22:46 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 14 Jul 2008 00:22:46 -0500
Subject: [LLVMbugs] [Bug 2541] Olden benchmark perimeter fails with llvm-2.3
In-Reply-To: <bug-2541-206@http.llvm.org/bugs/>
Message-ID: <200807140522.m6E5Mk4f013119@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2541


Chris Lattner <sabre at nondot.org> changed:

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




--- Comment #1 from Chris Lattner <sabre at nondot.org>  2008-07-14 00:22:46 ---
Works for me with mainline on darwin/x86-64.  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  Mon Jul 14 06:50:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 14 Jul 2008 06:50:30 -0500
Subject: [LLVMbugs] [Bug 2548] New: wide strings versus endianness
Message-ID: <bug-2548-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2548

           Summary: wide strings versus endianness
           Product: tools
           Version: 2.3
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: jay.foad at antixlabs.com
                CC: llvmbugs at cs.uiuc.edu


I built llvm-gcc as a cross compiler, running on x86 linux but targetting
sparc.

When I compile this source:

$ cat wide.c
int f() { return L'a'; }
int g() { return L"a"[0]; }

I get:

$ cc1 -quiet -emit-llvm wide.c
$ cat wide.s
; ModuleID = 'wide.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-f64:64:64-v64:64:64-v128:128:128-a0:0:64-f128:128:128"
target triple = "sparc-linux"
@.str = internal constant [2 x i32] [ i32 1627389952, i32 0 ]           ; <[2 x
i32]*> [#uses=1]

define i32 @f() nounwind  {
entry:
        %retval = alloca i32            ; <i32*> [#uses=2]
        %tmp = alloca i32               ; <i32*> [#uses=2]
        %"alloca point" = bitcast i32 0 to i32          ; <i32> [#uses=0]
        store i32 97, i32* %tmp, align 4
        %tmp1 = load i32* %tmp, align 4         ; <i32> [#uses=1]
        store i32 %tmp1, i32* %retval, align 4
        br label %return

return:         ; preds = %entry
        %retval2 = load i32* %retval            ; <i32> [#uses=1]
        ret i32 %retval2
}

define i32 @g() nounwind  {
entry:
        %retval = alloca i32            ; <i32*> [#uses=2]
        %tmp = alloca i32               ; <i32*> [#uses=2]
        %"alloca point" = bitcast i32 0 to i32          ; <i32> [#uses=0]
        %tmp1 = load i32* getelementptr ([2 x i32]* @.str, i32 0, i32 0), align
4               ; <i32> [#uses=1]
        store i32 %tmp1, i32* %tmp, align 4
        %tmp2 = load i32* %tmp, align 4         ; <i32> [#uses=1]
        store i32 %tmp2, i32* %retval, align 4
        br label %return

return:         ; preds = %entry
        %retval3 = load i32* %retval            ; <i32> [#uses=1]
        ret i32 %retval3
}


So f() will return 97 (0x61) but g() will return 1627389952 (0x61000000). This
looks like an endianness problem.


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

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 14 13:05:17 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 14 Jul 2008 13:05:17 -0500
Subject: [LLVMbugs] [Bug 2523] Asserts with releational operator applied to
 ext_vector_type(X) as argument
In-Reply-To: <bug-2523-206@http.llvm.org/bugs/>
Message-ID: <200807141805.m6EI5HVU015796@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2523


Nate Begeman <natebegeman at mac.com> changed:

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




--- Comment #3 from Nate Begeman <natebegeman at mac.com>  2008-07-14 13:05:16 ---
Now correctly returns an error that if() expects a scalar.  The vector
comparison produces a vector; it is expected that something a la vec_any(a < b)
would  be required to be used with if().

Patch at:
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080714/006483.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 Jul 14 15:24:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 14 Jul 2008 15:24:39 -0500
Subject: [LLVMbugs] [Bug 2549] New: Vector spills unaligned despite aligned
	stack
Message-ID: <bug-2549-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2549

           Summary: Vector spills unaligned despite aligned stack
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicolas at capens.net
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1831)
 --> (http://llvm.org/bugs/attachment.cgi?id=1831)
Bitcode file generating movups for vector spills

Loads and stores between SSE registers and the stack due to spills currently
use the unaligned move instruction (movups), despite the stack being properly
16-byte aligned. It could use aligned moves (movaps) instead for better
performance.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 15 00:34:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 15 Jul 2008 00:34:39 -0500
Subject: [LLVMbugs] [Bug 2546] create SCEVInfinite
In-Reply-To: <bug-2546-206@http.llvm.org/bugs/>
Message-ID: <200807150534.m6F5YdtE007195@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2546


Nick Lewycky <nicholas at mxc.ca> changed:

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




--- Comment #3 from Nick Lewycky <nicholas at mxc.ca>  2008-07-15 00:34:38 ---
Oh, wow, I didn't realize that. Indeed, I tried it out by modifying SCEV and
there are none!


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 16 02:31:14 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 16 Jul 2008 02:31:14 -0500
Subject: [LLVMbugs] [Bug 2296] Instcombine shouldn' t turn
 x86_sse2_storel_dq into a full-width store
In-Reply-To: <bug-2296-206@http.llvm.org/bugs/>
Message-ID: <200807160731.m6G7VEmS010333@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2296


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

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




--- Comment #8 from Evan Cheng <evan.cheng at apple.com>  2008-07-16 02:31:13 ---
This is fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/065092.html

It now compiles to:
_x:
        movq    %xmm0, _G
        ret


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 16 17:29:35 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 16 Jul 2008 17:29:35 -0500
Subject: [LLVMbugs] [Bug 2552] New: FunctionPassManager crashed with
 function inling pass
Message-ID: <bug-2552-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2552

           Summary: FunctionPassManager crashed with function inling pass
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Interprocedural Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


The following code makes llvm crash:

std::string err;
MemoryBuffer* buf = MemoryBuffer::getFile(filename, &err)
Module* module = ParseBitcodeFile(buf, &err)
ModuleProvider* provider = new ExistingModuleProvider(module);
ExecutionEngine* engine = ExecutionEngine::create(provider);

FunctionPassManager* pass_manager = new FunctionPassManager(provider);
pass_manager->add(new TargetData(*engine->getTargetData()));
pass_manager->add(createFunctionInliningPass());


gdb stack trace:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb7505910 (LWP 6122)]
0xb72153fd in llvm::PMStack::top (this=0x9c600c8) at
/cvs/llvm/include/llvm/PassManagers.h:133
133       inline PMDataManager *top() { return S.back(); }
Current language:  auto; currently c++
(gdb) bt
#0  0xb72153fd in llvm::PMStack::top (this=0x9c600c8) at
/cvs/llvm/include/llvm/PassManagers.h:133
#1  0xb72d696f in llvm::ModulePass::assignPassManager (this=0x9b815c8,
PMS=@0x9c600c8, PreferredType=llvm::PMT_Unknown) at PassManager.cpp:1519
#2  0xb72e0ad8 in llvm::FunctionPassManagerImpl::addTopLevelPass
(this=0x9c60058, P=0x9b815c8) at PassManager.cpp:172
#3  0xb72d5332 in llvm::PMTopLevelManager::schedulePass (this=0x9c600c4,
P=0x9b815c8) at PassManager.cpp:462
#4  0xb72d52da in llvm::PMTopLevelManager::schedulePass (this=0x9c600c4,
P=0x95cc7b8) at PassManager.cpp:455
#5  0xb72e0b17 in llvm::FunctionPassManagerImpl::add (this=0x9c60058,
P=0x95cc7b8) at PassManager.cpp:140
#6  0xb72d54f9 in llvm::FunctionPassManager::add (this=0x89d39e0, P=0x95cc7b8)
at PassManager.cpp:1128
# 7 ...
(gdb) p S.back()
$1 = (class llvm::PMDataManager *&) @0x1fb: <error reading variable>


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 16 20:27:23 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 16 Jul 2008 20:27:23 -0500
Subject: [LLVMbugs] [Bug 2553] New: Invalid float optimization with NaN
Message-ID: <bug-2553-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2553

           Summary: Invalid float optimization with NaN
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: NetBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: krister.walfridsson at gmail.com
                CC: llvmbugs at cs.uiuc.edu


llvm-gcc-4.2 (svn revision 53688) miscompiles some functions in libgcc (for
example __mulxc3) because llvm simplifies the code for its isfinite() a bit too
much.

You can see this for:

  #define isnan(x)      ((x) != (x))
  #define isfinite(x)   (!isnan((x) - (x)))

  void foo(float x)
  {
      if (isfinite(x))
          bar();
  }

Running

  gcc -O0 -S -emit-llvm bug.c

gives me the attached bug.ll, and executing

  llvm-as < bug.ll | opt -std-compile-opts | llvm-dis

gives

  define void @foo(float %x) nounwind  {
  entry:
        tail call i32 (...)* @bar( ) nounwind           ; <i32>:0 [#uses=0]
        ret void
  }

  declare i32 @bar(...)

i.e. it has optimized away the test for isfinite().


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 16 23:39:38 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 16 Jul 2008 23:39:38 -0500
Subject: [LLVMbugs] [Bug 2554] New: assertion failure in
 X86TargetLowering::LowerVECTOR_SHUFFLE
Message-ID: <bug-2554-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2554

           Summary: assertion failure in
                    X86TargetLowering::LowerVECTOR_SHUFFLE
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


This turned up in make check:

llvm-as < test/CodeGen/PowerPC/vec_spat.ll | llc -march=x86 -mattr=+sse2
llc: /home/nicholas/llvm-commit/include/llvm/Support/Casting.h:199: typename
llvm::cast_retty<To, From>::ret_type llvm::cast(const Y&) [with X =
llvm::ConstantSDNode, Y = llvm::SDOperand]: Assertion `isa<X>(Val) &&
"cast<Ty>() argument of incompatible type!"' failed.
llc: /home/nicholas/llvm-commit/include/llvm/Support/Casting.h:199: typename
llvm::cast_retty<To, From>::ret_type llvm::cast(const Y&) [with X =
llvm::ConstantSDNode, Y = llvm::SDOperand]: Assertion `isa<X>(Val) &&
"cast<Ty>() argument of incompatible type!"' failed.
Debug/bin/llc[0x8b1cab2]
Debug/bin/llc[0x8b1cbe4]
[0xffffe400]
Aborted

Backtrace:

[New Thread 0xf7c9a6c0 (LWP 28956)]

Program received signal SIGABRT, Aborted.
[Switching to Thread 0xf7c9a6c0 (LWP 28956)]
0xffffe425 in __kernel_vsyscall ()
(gdb) bt
#0  0xffffe425 in __kernel_vsyscall ()
#1  0xf7cc75e0 in raise () from /lib/i686/cmov/libc.so.6
#2  0xf7cc8fb8 in abort () from /lib/i686/cmov/libc.so.6
#3  0xf7cc056e in __assert_fail () from /lib/i686/cmov/libc.so.6
#4  0x084515bc in llvm::cast<llvm::ConstantSDNode, llvm::SDOperand> (
    Val=@0xffd999ec)
    at /home/nicholas/llvm-commit/include/llvm/Support/Casting.h:199
#5  0x0874d600 in isUndefShuffle (N=0x8d0485c) at X86ISelLowering.cpp:2702
#6  0x0876646c in llvm::X86TargetLowering::LowerVECTOR_SHUFFLE (
    this=0x8cf9ab8, Op={Val = 0x8d0485c, ResNo = 0}, DAG=@0xffd9b6cc)
    at X86ISelLowering.cpp:3654
#7  0x0876d7f3 in llvm::X86TargetLowering::LowerOperation (this=0x8cf9ab8, Op=
      {Val = 0x8d0485c, ResNo = 0}, DAG=@0xffd9b6cc)
    at X86ISelLowering.cpp:5714
#8  0x08869191 in LegalizeOp (this=0xffd9b33c, Op={Val = 0x8d0485c, ResNo = 0})
    at LegalizeDAG.cpp:1522
#9  0x088935d1 in HandleOp (this=0xffd9b33c, Op={Val = 0x8d0485c, ResNo = 0})
    at LegalizeDAG.cpp:455
#10 0x088939e3 in LegalizeDAG (this=0xffd9b33c) at LegalizeDAG.cpp:334
#11 0x08893b23 in llvm::SelectionDAG::Legalize (this=0xffd9b6cc)
    at LegalizeDAG.cpp:7045
#12 0x08818629 in llvm::SelectionDAGISel::CodeGenAndEmitDAG (this=0x8cf9a48, 
    DAG=@0xffd9b6cc) at SelectionDAGISel.cpp:5313
#13 0x08830978 in llvm::SelectionDAGISel::SelectBasicBlock (this=0x8cf9a48, 
    LLVMBB=0x8cef0f0, MF=@0x8d02188, FuncInfo=@0xffd9bb14, 
    AllNodes=@0xffd9b9d4) at SelectionDAGISel.cpp:5409
#14 0x0883215d in llvm::SelectionDAGISel::SelectAllBasicBlocks (
    this=0x8cf9a48, Fn=@0x8cee478, MF=@0x8d02188, FuncInfo=@0xffd9bb14)
    at SelectionDAGISel.cpp:5389
#15 0x08832bfc in llvm::SelectionDAGISel::runOnFunction (this=0x8cf9a48, 
    Fn=@0x8cee478) at SelectionDAGISel.cpp:4883
#16 0x086509a1 in runOnFunction (this=0x8cf9a48, Fn=@0x8cee478)
    at X86ISelDAGToDAG.cpp:123
#17 0x08aad7aa in llvm::FPPassManager::runOnFunction (this=0x8cf21b8, 
    F=@0x8cee478) at PassManager.cpp:1241
#18 0x08aada2a in llvm::FunctionPassManagerImpl::run (this=0x8cef398, 
    F=@0x8cee478) at PassManager.cpp:1196
#19 0x08aadb81 in llvm::FunctionPassManager::run (this=0xffd9bd3c, 
    F=@0x8cee478) at PassManager.cpp:1141
#20 0x083ef7da in main (argc=3, argv=0xffd9be54) at llc.cpp:302
(gdb) up 5
#5  0x0874d600 in isUndefShuffle (N=0x8d0485c) at X86ISelLowering.cpp:2702
2702          unsigned Val = cast<ConstantSDNode>(Arg)->getValue();
(gdb) p Arg
$1 = {Val = 0x8d044a4, ResNo = 0}
(gdb) p Arg->Val
$2 = (class llvm::SDNode *) 0x8d044a4
(gdb) p Arg->Val->dump()
0x8d044a4: v4i32 = BUILD_VECTOR 0x8d04064, 0x8d04064, 0x8d03fdc, 0x8d04064$3 =
void


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

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 17 01:07:59 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 01:07:59 -0500
Subject: [LLVMbugs] [Bug 2553] Invalid float optimization with NaN
In-Reply-To: <bug-2553-206@http.llvm.org/bugs/>
Message-ID: <200807170607.m6H67xdJ032360@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2553


Chris Lattner <clattner at apple.com> changed:

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




--- Comment #3 from Chris Lattner <clattner at apple.com>  2008-07-17 01:07:58 ---
Fixed, thanks!
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/065150.html


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

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 17 06:08:32 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 06:08:32 -0500
Subject: [LLVMbugs] [Bug 2555] New: APFloat does not support IEEEquad
Message-ID: <bug-2555-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2555

           Summary: APFloat does not support IEEEquad
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Support Libraries
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rich at pennware.com
                CC: llvmbugs at cs.uiuc.edu


IEEE 128 bit values are not supported by APFloat. Also, looking at the code, it
is not clear to me how values with multiple significand words work. It seems
that the code here and in APInt is host endian sensitive. For example,
APInt(80, 2, words) copies the words array into the big int. This seems wrong.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 17 07:26:34 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 07:26:34 -0500
Subject: [LLVMbugs] [Bug 2556] New: Multiple return values "Couldn't find
 the register class" assert failure
Message-ID: <bug-2556-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2556

           Summary: Multiple return values "Couldn't find the register
                    class" assert failure
           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


Running llc on the following testcase causes an assertion failure:

lib/Target/TargetRegisterInfo.cpp:72: const llvm::TargetRegisterClass*
llvm::TargetRegisterInfo::getPhysicalRegisterRegClass(unsigned int, llvm::MVT)
const: Assertion `false && "Couldn't find the register class"' failed.

This is stopping the llvm-gcc Ada front-end from building on x86-64 linux.

target datalayout =
"e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
target triple = "x86_64-unknown-linux-gnu"

define { <2 x float>, i64 } @ada__complex_text_io__get__3(i64 %from.0, i64
%from.1) {
entry:
        ret <2 x float> zeroinitializer, i64 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 Jul 17 07:38:23 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 07:38:23 -0500
Subject: [LLVMbugs] [Bug 2557] New: Anders pass asserts on
	extractvalue/insertvalue
Message-ID: <bug-2557-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2557

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


Very easy to reproduce: Feed any program containing a insertvalue or
extractvalue to Anders AA pass, and it asserts. These instructions were
recently added (for the "first class aggregate" feature), but no one told
anders about them yet.

Also see bug #2451, which is the same bug, but for the vector insert/extract
instructions.

Using a testcase from InstCombine, we can easily check this:

llvm-as < test/Transforms/InstCombine/extractvalue.ll | opt -anders-aa
-disable-output
Unknown instruction:    %s1.1 = insertvalue { i32, i32 } undef, i32 0, 0       
        ; <{ i32, i32 }> [#uses=1]
opt[0x86db1a6]
opt[0x86db2d8]
[0xffffe420]
/lib/tls/i686/cmov/libc.so.6(abort+0x101)[0xb7d4f201]
opt[0x854b1e3]
opt(_ZN4llvm11InstVisitorIN42_GLOBAL__N_Andersens.cpp_00000000_00A7C8139AndersensEvE20visitInsertValueInstERNS_15InsertValueInstE+0x18)[0x8560582]
opt(_ZN4llvm11InstVisitorIN42_GLOBAL__N_Andersens.cpp_00000000_00A7C8139AndersensEvE16visitInsertValueERNS_15InsertValueInstE+0x18)[0x856059c]
opt(_ZN4llvm11InstVisitorIN42_GLOBAL__N_Andersens.cpp_00000000_00A7C8139AndersensEvE5visitERNS_11InstructionE+0x4fc)[0x8560bf8]
opt(_ZN4llvm11InstVisitorIN42_GLOBAL__N_Andersens.cpp_00000000_00A7C8139AndersensEvE5visitINS_14ilist_iteratorINS_11InstructionEEEEEvT_S8_+0x45)[0x8560c67]
opt(_ZN4llvm11InstVisitorIN42_GLOBAL__N_Andersens.cpp_00000000_00A7C8139AndersensEvE5visitERNS_10BasicBlockE+0x5e)[0x8560cde]
opt(_ZN4llvm11InstVisitorIN42_GLOBAL__N_Andersens.cpp_00000000_00A7C8139AndersensEvE5visitINS_14ilist_iteratorINS_10BasicBlockEEEEEvT_S8_+0x45)[0x8560d29]
opt(_ZN4llvm11InstVisitorIN42_GLOBAL__N_Andersens.cpp_00000000_00A7C8139AndersensEvE5visitERNS_8FunctionE+0x5e)[0x8560da0]
opt(_ZN4llvm11InstVisitorIN42_GLOBAL__N_Andersens.cpp_00000000_00A7C8139AndersensEvE5visitEPNS_8FunctionE+0x18)[0x8560dbe]
opt[0x854e781]
opt[0x8560e00]
opt(_ZN4llvm13MPPassManager11runOnModuleERNS_6ModuleE+0x108)[0x8667420]
opt(_ZN4llvm15PassManagerImpl3runERNS_6ModuleE+0x6e)[0x86675d8]
opt(_ZN4llvm11PassManager3runERNS_6ModuleE+0x1b)[0x866762b]
opt(main+0xc10)[0x839ffda]
/lib/tls/i686/cmov/libc.so.6(__libc_start_main+0xe0)[0xb7d39050]
opt(__gxx_personality_v0+0x335)[0x8391a81]
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  Thu Jul 17 07:49:16 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 07:49:16 -0500
Subject: [LLVMbugs] [Bug 2558] New: Argpromotion does not handle first class
	aggregates
Message-ID: <bug-2558-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2558

           Summary: Argpromotion does not handle first class aggregates
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Interprocedural Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


Argumentpromotion does not currently promote first-class structs and arrays.

A testcase is as follows:

; This testcase makes sure that argpromotion promotes first class aggregates
; properly.
; RUN: llvm-as < %s | opt -argpromotion | llvm-dis > %t
; (These grep lines were not tested)
; RUN: cat %t | grep {define.*@sum(i32 %.*, i32 %.*)}
; RUN: cat %t | grep {define.*@sum2(i32 %.*, i32 %.*)}

define internal i32 @sum({i32, i32} %A) {
  %B = extractvalue {i32, i32} %A, 0
  %C = extractvalue {i32, i32} %A, 0
  %D = add i32 %B, %C
  ret i32 %D
}

define internal i32 @sum2([2 x i32] %A) {
  %B = extractvalue {i32, i32} %A, 0
  %C = extractvalue {i32, i32} %A, 0
  %D = add i32 %B, %C
  ret i32 %D
}

define void @run(i32 %A, i32 %B) {
  %C = insertvalue {i32, i32} undef, i32 %A, 0
  %D = insertvalue {i32, i32} %C, i32 %B, 1
  %E = call i32 @sum({i32, i32} %D);
  %F = insertvalue [ 2 x i32 ] undef, i32 %A, 0
  %G = insertvalue [ 2 x i32 ] %F, i32 %B, 1
  %H = call i32 @sum({i32, i32} %D);
  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 Jul 17 09:15:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 09:15:19 -0500
Subject: [LLVMbugs] [Bug 2559] New: Finish first-class aggregates support
Message-ID: <bug-2559-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2559

           Summary: Finish first-class aggregates support
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu
        Depends on: 2557,2558


In the past months support for first class aggregates has been largely
implemented. The original design was detailed in
http://nondot.org/sabre/LLVMNotes/FirstClassAggregates.txt

By now, the insertvalue and extractvalue instructions are implemented and the
verifier accepts aggregates as first class values (ie, in function arguments,
return values, loads, stores, etc.).

A number of passes have added support for manipulating and simplifying
constructs containing first class structs.

I'm not really sure how the status of codegen for first class structs is, since
I'm not using codegen atm. Perhaps someone else can create PRs for the
unsupported parts in codegen?

There are still a number of other things that need to be done, I will be
creating PRs for those things and make them block this PR.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 17 09:23:46 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 09:23:46 -0500
Subject: [LLVMbugs] [Bug 2560] New: Remove multiple return value support
Message-ID: <bug-2560-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2560

           Summary: Remove multiple return value support
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu
            Blocks: 2559


Now that most of the first class aggregate support is completed, we should be
able te remove the old multiple return value support sometime soon. This mostly
means two things:
 * return instructions should no longer be allowed to take more than one
operand.
 * The getresult instruction should be removed.

Returning multiple values can now be implemented by creating a struct using
insertvalue, returning it, and accessing the results using extractvalue
(instead of getresult).

This can be done as soon as all frontends have been changed to no longer use
multiple return values (not sure which currently do so).


Alternatively, we could add a pass that transforms the old multiple return
value instruction into insertvalues and replaces getresult with extractvalue
(or perhaps change the assembly/bitcode readers to do so).

This will allow support for multiple return values to be removed from all other
passes, while retaining backward compatibility for a while (how long?)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 17 12:02:54 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 12:02:54 -0500
Subject: [LLVMbugs] [Bug 2561] New: flexible array member initializers
	generate warning
Message-ID: <bug-2561-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2561

           Summary: flexible array member initializers generate warning
           Product: clang
           Version: unspecified
          Platform: Macintosh
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


Given:
--
struct one {
  int a;
  int values[];
} x = {5, {1, 2, 3}};
--
clang generates an excess elements in array initializer warning


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 17 12:46:33 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 12:46:33 -0500
Subject: [LLVMbugs] [Bug 2562] New: MMX inserts fail
Message-ID: <bug-2562-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2562

           Summary: MMX inserts fail
           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: nicolas at capens.net
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1838)
 --> (http://llvm.org/bugs/attachment.cgi?id=1838)
fibonacci.cpp replacement generating the IR

Attempting to insert a 16-bit or 32-bit value into a 64-bit vector fails. Both
currently lower to a shufflevector but I get a "Cannot yet select" assert in
X86GenDAGISel.inc.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 17 14:30:06 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 14:30:06 -0500
Subject: [LLVMbugs] [Bug 2554] assertion failure in
 X86TargetLowering::LowerVECTOR_SHUFFLE
In-Reply-To: <bug-2554-206@http.llvm.org/bugs/>
Message-ID: <200807171930.m6HJU64i005196@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2554


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #5 from Duncan Sands <baldrick at free.fr>  2008-07-17 14:30:05 ---
DAG combiner bug.  Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/065164.html


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

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 17 17:06:38 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 17:06:38 -0500
Subject: [LLVMbugs] [Bug 2563] New: SimplifyCFG is not collapsing unneeded
	switches
Message-ID: <bug-2563-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2563

           Summary: SimplifyCFG is not collapsing unneeded switches
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: resistor at mac.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1841)
 --> (http://llvm.org/bugs/attachment.cgi?id=1841)
Example bitcode

In the attached testcase, note the blocks:

bb4.i.i:                ; preds = %grad.exit21.i
        switch i32 %154, label %bb9.i.i [
                 i32 12, label %bb12.i.i
                 i32 14, label %bb12.i.i
        ]

bb9.i.i:                ; preds = %bb4.i.i
        br label %bb12.i.i

bb12.i.i:
 ...

The switch in bb4.i.i is unnecessary, as is bb9.i.i.  SimplifyCFG (or some
other pass) should simplify this away, but doesn't.  This idiom occurs multiple
times in this case, probably as a result of LICM.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 17 22:09:44 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 22:09:44 -0500
Subject: [LLVMbugs] [Bug 2564] New: assertion "BitWidth == RHS.BitWidth && "
 Bit widths must be the same"" failed
Message-ID: <bug-2564-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2564

           Summary: assertion "BitWidth == RHS.BitWidth && "Bit widths must
                    be the same"" failed
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: NetBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Basic
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: krister.walfridsson at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1842)
 --> (http://llvm.org/bugs/attachment.cgi?id=1842)
File triggering the bug

Pointer arithmetic may make clang (svn revision 53738, NetBSD, x86_64)
trigger an assert:

> clang -fsyntax-only bug.c
assertion "BitWidth == RHS.BitWidth && "Bit widths must be the same"" failed:
file "APInt.cpp", line 283, function "llvm::APInt&
llvm::APInt::operator-=(const llvm::APInt&)"
Abort (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  Thu Jul 17 22:56:51 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 17 Jul 2008 22:56:51 -0500
Subject: [LLVMbugs] [Bug 2565] New: assertion
 "Src->getType()->isFloatingPoint() && " Unknown real conversion"" failed
Message-ID: <bug-2565-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2565

           Summary: assertion "Src->getType()->isFloatingPoint() && "Unknown
                    real conversion"" failed
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: NetBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: LLVM Codegen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: krister.walfridsson at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1843)
 --> (http://llvm.org/bugs/attachment.cgi?id=1843)
File triggering the bug

Compiling the following program with clang (svn revision 53738, NetBSD, x86_64)
trigger an assertion:

> clang bug.c -emit-llvm
assertion "Src->getType()->isFloatingPoint() && "Unknown real conversion""
failed: file "CGExprScalar.cpp", line 402, function
"llvm::Value*<unnamed>::ScalarExprEmitter::EmitScalarConversion(llvm::Value*,
clang::QualType, clang::QualType)"
Abort (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 Jul 18 07:45:07 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 18 Jul 2008 07:45:07 -0500
Subject: [LLVMbugs] [Bug 2566] New: Combine of bitcast and shuffle assumes
 vector to vector cast
Message-ID: <bug-2566-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2566

           Summary: Combine of bitcast and shuffle assumes vector to vector
                    cast
           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: nicolas at capens.net
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1844)
 --> (http://llvm.org/bugs/attachment.cgi?id=1844)
Replacement of fibonacci.cpp generating the LLVM IR

When a cast from a plain integer to a vector is followed by a shuffle, I'm
hitting an assert originating in DAGCombiner::visitVECTOR_SHUFFLE:

Assertion failed: isVector() && "Invalid vector type!", file
d:\Projects\LLVM\in
clude\llvm/CodeGen/ValueTypes.h, line 333

It looks like it's incorrectly assuming the cast to be a cast from one vector
to another.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 18 11:40:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 18 Jul 2008 11:40:28 -0500
Subject: [LLVMbugs] [Bug 2280] clang: no output should be produced when
	errors occurs
In-Reply-To: <bug-2280-206@http.llvm.org/bugs/>
Message-ID: <200807181640.m6IGeSXf022485@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2280


Daniel Dunbar <daniel at zuster.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |daniel at zuster.org
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED




--- Comment #1 from Daniel Dunbar <daniel at zuster.org>  2008-07-18 11:40:25 ---
Fixed in: http://llvm.org/viewvc/llvm-project?view=rev&revision=53756


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 18 13:44:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 18 Jul 2008 13:44:30 -0500
Subject: [LLVMbugs] [Bug 2408] CBE generates code that violates C signed
 integer overflow rules
In-Reply-To: <bug-2408-206@http.llvm.org/bugs/>
Message-ID: <200807181844.m6IIiUmY026404@zion.cs.uiuc.edu>

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


Dan Gohman <gohman at apple.com> changed:

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




--- Comment #3 from Dan Gohman <gohman at apple.com>  2008-07-18 13:44:30 ---
I applied the attached patch.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/065190.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 Jul 18 15:15:53 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 18 Jul 2008 15:15:53 -0500
Subject: [LLVMbugs] [Bug 2554] assertion failure in
 X86TargetLowering::LowerVECTOR_SHUFFLE
In-Reply-To: <bug-2554-206@http.llvm.org/bugs/>
Message-ID: <200807182015.m6IKFrRd029257@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2554


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #7 from Duncan Sands <baldrick at free.fr>  2008-07-18 15:15:52 ---
I've reverted the fix because I think it is wrong.
I reckon the real fix is to not have the shufflevec
as an operand at all, but as a list of numbers stored
in the node.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 19 01:33:02 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 19 Jul 2008 01:33:02 -0500
Subject: [LLVMbugs] [Bug 2549] Vector spills unaligned despite aligned stack
In-Reply-To: <bug-2549-206@http.llvm.org/bugs/>
Message-ID: <200807190633.m6J6X2lm015667@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2549


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

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




--- Comment #5 from Anton Korobeynikov <asl at math.spbu.ru>  2008-07-19 01:33:01 ---
Thanks for nice testcase.

Should be fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/065205.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/065206.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 Jul 19 08:44:12 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 19 Jul 2008 08:44:12 -0500
Subject: [LLVMbugs] [Bug 2364] signedness error in scalar-evolution
	HowFarToZero
In-Reply-To: <bug-2364-206@http.llvm.org/bugs/>
Message-ID: <200807191344.m6JDiC0p007036@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2364


Wojciech Matyjewicz <wmatyjewicz at fastmail.fm> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wmatyjewicz at fastmail.fm
         Resolution|FIXED                       |DUPLICATE




--- Comment #3 from Wojciech Matyjewicz <wmatyjewicz at fastmail.fm>  2008-07-19 08:44:10 ---
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/065214.html
revealed the patch (which is correct) didn't fully fix the bug, but made that
currently ComputeIterationCountExhaustively handles the testcase (and gives
correct answer).

This bug will be fully fixed with the patch for PR2088.


*** This bug has been marked as a duplicate of bug 2088 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 19 13:15:40 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 19 Jul 2008 13:15:40 -0500
Subject: [LLVMbugs] [Bug 2568] New: possible integer codegen bug
Message-ID: <bug-2568-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2568

           Summary: possible 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


Created an attachment (id=1851)
 --> (http://llvm.org/bugs/attachment.cgi?id=1851)
reduced code for this bug

This is for svn 53781 on Ubuntu Feisty on ia32.

[regehr at babel tmp6]$ llvm-gcc -O0 -fwrapv small.c -o small
[regehr at babel tmp6]$ ./small
0
[regehr at babel tmp6]$ llvm-gcc -O1 -fwrapv small.c -o small
[regehr at babel tmp6]$ ./small
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 Jul 19 13:33:13 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 19 Jul 2008 13:33:13 -0500
Subject: [LLVMbugs] [Bug 2569] New: JITResolver::getFunctionStub() doesn' t
 update the global mapping
Message-ID: <bug-2569-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2569

           Summary: JITResolver::getFunctionStub() doesn't update the global
                    mapping
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Target-Independent JIT
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nunoplopes at sapo.pt
                CC: llvmbugs at cs.uiuc.edu


JITResolver::getFunctionStub() doesn't update the global mapping when creating
stubs for non-external functions.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 19 16:05:25 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 19 Jul 2008 16:05:25 -0500
Subject: [LLVMbugs] [Bug 2570] New: possible infinite loop in cc1
Message-ID: <bug-2570-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2570

           Summary: possible infinite loop in cc1
           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





-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 20 11:04:18 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 20 Jul 2008 11:04:18 -0500
Subject: [LLVMbugs] [Bug 2088] Scalar evolution fails to compute loop
	iteration count
In-Reply-To: <bug-2088-206@http.llvm.org/bugs/>
Message-ID: <200807201604.m6KG4I8x025369@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2088


Wojciech Matyjewicz <wmatyjewicz at fastmail.fm> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
          Component|new bugs                    |Global Analyses
            Product|new-bugs                    |libraries
         Resolution|                            |FIXED
   Target Milestone|---                         |2.4
            Version|unspecified                 |trunk




--- Comment #13 from Wojciech Matyjewicz <wmatyjewicz at fastmail.fm>  2008-07-20 11:04:17 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080714/065225.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 Jul 20 20:25:43 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 20 Jul 2008 20:25:43 -0500
Subject: [LLVMbugs] [Bug 2571] New: Shootout-C++/ackermann has gcc/llc of
	0.23x on Linux
Message-ID: <bug-2571-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2571

           Summary: Shootout-C++/ackermann has gcc/llc of 0.23x on Linux
           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: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


For whatever reason, ackermann runs very slowly on Linux

  GCC: 0.92s
  CBE: 2.19s
  LLC: 4.01s

making it the slowest test in the Linux results. This bug is to figure out why.
More numbers:

  Using g++ (Debian 4.3.1-6) 4.3.1
  g++ -O2: 5.9s
  g++ -O2 -fwhole-program: 3.1s
  g++ -O3: 0.93s
  g++ -O3 -fno-strict-aliasing: 0.93s

Is it just unrolling loops?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 20 23:31:11 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 20 Jul 2008 23:31:11 -0500
Subject: [LLVMbugs] [Bug 2572] New: kimwitu++ lacks include of cstring for
	strlen
Message-ID: <bug-2572-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2572

           Summary: kimwitu++ lacks include of cstring for strlen
           Product: Test Suite
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Nightly Tester
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicholas at mxc.ca
                CC: llvmbugs at cs.uiuc.edu


kimwitu++ doesn't compile on my Linux GCC 4.3 machine because it lacks an
#include <cstring> when using strlen. Unfortunately, kimwitu++ appears to be
generated code. Does anyone know how this works?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 20 23:59:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 20 Jul 2008 23:59:48 -0500
Subject: [LLVMbugs] [Bug 2572] kimwitu++ lacks include of cstring for strlen
In-Reply-To: <bug-2572-206@http.llvm.org/bugs/>
Message-ID: <200807210459.m6L4xmp3017660@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2572


Nick Lewycky <nicholas at mxc.ca> changed:

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




--- Comment #2 from Nick Lewycky <nicholas at mxc.ca>  2008-07-20 23:59:48 ---
Okay then! Fixed in 53828.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 21 16:18:05 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 21 Jul 2008 16:18:05 -0500
Subject: [LLVMbugs] [Bug 2574] New: scalar_to_vector not selecting on x86
 (-mcpu=core2) for v2f32
Message-ID: <bug-2574-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2574

           Summary: scalar_to_vector not selecting on x86 (-mcpu=core2) for
                    v2f32
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sdt at rapidmind.com
                CC: llvmbugs at cs.uiuc.edu


Consider the following test case:

define void @entry(i32 %m_task_id, i32 %start_x, i32 %end_x) {
; <label>:0
        br i1 false, label %bb.nph, label %._crit_edge

bb.nph:         ; preds = %bb.nph, %0
        %t2206f2.0 = phi <2 x float> [ %2, %bb.nph ], [ undef, %0 ]            
; <<2 x float>> [#uses=1]
        insertelement <2 x float> %t2206f2.0, float 0.000000e+00, i32 0        
; <<2 x float>>:1 [#uses=1]
        insertelement <2 x float> %1, float 0.000000e+00, i32 1         ; <<2 x
float>>:2 [#uses=1]
        br label %bb.nph

._crit_edge:            ; preds = %0
        ret void
}


Running llc -mcpu=core2 on this produces:
Cannot yet select: 0x8b0471c: v2f32 = scalar_to_vector 0x8b049c4

This is a regression since 2.3.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 21 16:20:44 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 21 Jul 2008 16:20:44 -0500
Subject: [LLVMbugs] [Bug 2575] New: v4i32 vsetcc not legalizing correctly
Message-ID: <bug-2575-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2575

           Summary: v4i32 vsetcc not legalizing correctly
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sdt at rapidmind.com
                CC: llvmbugs at cs.uiuc.edu


With llc -mcpu=pentium, the following code (courtesy of bugpoint) crashes llc:

===
define void @entry(i32 %m_task_id, i32 %start_x, i32 %end_x) {
        br i1 false, label %bb.nph, label %._crit_edge

bb.nph:         ; preds = %bb.nph, %0
        vicmp sgt <4 x i32> zeroinitializer, < i32 -128, i32 -128, i32 -128,
i32 -128 >         ; <<4 x i32>>:1 [#uses=1]
        extractelement <4 x i32> %1, i32 3              ; <i32>:2 [#uses=1]
        lshr i32 %2, 31         ; <i32>:3 [#uses=1]
        trunc i32 %3 to i1              ; <i1>:4 [#uses=1]
        select i1 %4, i32 -1, i32 0             ; <i32>:5 [#uses=1]
        insertelement <4 x i32> zeroinitializer, i32 %5, i32 3          ; <<4 x
i32>>:6 [#uses=1]
        and <4 x i32> zeroinitializer, %6               ; <<4 x i32>>:7
[#uses=1]
        bitcast <4 x i32> %7 to <4 x float>             ; <<4 x float>>:8
[#uses=1]
        mul <4 x float> zeroinitializer, %8             ; <<4 x float>>:9
[#uses=1]
        bitcast <4 x float> %9 to <4 x i32>             ; <<4 x i32>>:10
[#uses=1]
        or <4 x i32> %10, zeroinitializer               ; <<4 x i32>>:11
[#uses=1]
        bitcast <4 x i32> %11 to <4 x float>            ; <<4 x float>>:12
[#uses=1]
        mul <4 x float> %12, < float 1.000000e+02, float 1.000000e+02, float
1.000000e+02, float 1.000000e+02 >         ; <<4 x float>>:13 [#uses=1]
        sub <4 x float> %13, < float 1.000000e+02, float 1.000000e+02, float
1.000000e+02, float 1.000000e+02 >         ; <<4 x float>>:14 [#uses=1]
        extractelement <4 x float> %14, i32 3           ; <float>:15 [#uses=1]
        call float @fmaxf( float 0.000000e+00, float %15 )              ;
<float>:16 [#uses=0]
        br label %bb.nph

._crit_edge:            ; preds = %0
        ret void
}

declare float @fmaxf(float, float)
===

llc aborts with:

SplitVectorResult #0: 0x8b04484: v4i32 = vsetcc 0x8b04264, 0x8b04374, 0x8b043fc
llc: LegalizeVectorTypes.cpp:239: void
llvm::DAGTypeLegalizer::SplitVectorResult(llvm::SDNode*, unsigned int):
Assertion `0 && "Do not know how to split the result of this operator!"'
failed.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 21 16:35:31 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 21 Jul 2008 16:35:31 -0500
Subject: [LLVMbugs] [Bug 2576] New: Type check llvm.prefetch intrinsic
Message-ID: <bug-2576-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2576

           Summary: Type check llvm.prefetch intrinsic
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: minor
          Priority: P2
         Component: Core LLVM classes
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


The llvm.prefetch intrinsic should be type checked by the verifier to ensure
that the 2nd and 3rd arguments are constants in the appropriate ranges.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 21 16:42:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 21 Jul 2008 16:42:19 -0500
Subject: [LLVMbugs] [Bug 2577] New: Support __builtin_expect
Message-ID: <bug-2577-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2577

           Summary: Support __builtin_expect
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


We should use the information provided by the gcc __builtin_expect function.

This involves adding some mechanism for encoding the information in the LLVM IR
(most likely an intrinsic) and updating llvm-gcc and clang to emit this
information.


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

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 21 16:46:07 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 21 Jul 2008 16:46:07 -0500
Subject: [LLVMbugs] [Bug 2570] infinite loop in LSR
In-Reply-To: <bug-2570-206@http.llvm.org/bugs/>
Message-ID: <200807212146.m6LLk7P5029806@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2570


Dan Gohman <gohman at apple.com> changed:

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




--- Comment #8 from Dan Gohman <gohman at apple.com>  2008-07-21 16:46:06 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080721/065267.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 Jul 21 16:55:35 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 21 Jul 2008 16:55:35 -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: <200807212155.m6LLtZju030173@zion.cs.uiuc.edu>

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


Dan Gohman <gohman at apple.com> changed:

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




--- Comment #13 from Dan Gohman <gohman at apple.com>  2008-07-21 16:55:35 ---
The fix for this was reverted. I plan to fix it and resubmit it.


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

From bugzilla-daemon at cs.uiuc.edu  Mon Jul 21 22:15:35 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 21 Jul 2008 22:15:35 -0500
Subject: [LLVMbugs] [Bug 2578] New: TrimAllocationToSize will trim blocks to
 smaller than FreeRangeHeader:: getMinBlockSize()
Message-ID: <bug-2578-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2578

           Summary: TrimAllocationToSize will trim blocks to smaller than
                    FreeRangeHeader::getMinBlockSize()
           Product: libraries
           Version: 2.3
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Generic Execution Engine Support
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ddneff at hotmail.com
                CC: llvmbugs at cs.uiuc.edu, ddneff at hotmail.com


There appears to be a bug in MemoryRangeHeader::TrimAllocationToSize in that it
only checks if the split off block will have a size greater than
FreeRangeHeader::getMinBlockSize(), instead of checking if both blocks will
have a size greater than this value.  This will happen when you have a very
small function and endFunctionBody is called.  It will trim the block for the
function to smaller than FreeRangeHeader::getMinBlockSize(), and when this
block gets freed at a later time, the block size written to the end of block
stomps on the previous pointer.  A possible solution is probably to just add:

NewSize = std::min(NewSize, FreeRangeHeader::getMinBlockSize()); 

Steps to reproduce:
1) create a very small function, such as a ret void.
2) JIT the function by calling getPointerToGlobal
3) Release the function by calling freeMachineCodeForFunction

Results: the circular free list is now corrupt.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 22 03:06:03 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 03:06:03 -0500
Subject: [LLVMbugs] [Bug 2579] New: `./configure` check for Carbon.h fails
	with CC=clang
Message-ID: <bug-2579-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2579

           Summary: `./configure` check for Carbon.h fails with CC=clang
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicolasweber at gmx.de
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1859)
 --> (http://llvm.org/bugs/attachment.cgi?id=1859)
Short testcase

When running `./configure` in a project that checks for Carbon.h, configure
tries to compile a program that contains `#include <Carbon/Carbon.h>` and a few
other includes. That program cannot be compiled with clang. A reduced test case
is attached.

This is probably related to #2427.

nico$ cat bla.c
# include <stdlib.h>
#include <Carbon/Carbon.h>
nico$ clang -fsyntax-only bla.c
In file included from bla.c:2:
In file included from
/System/Library/Frameworks/Carbon.framework/Headers/Carbon.h:24:
In file included from
/System/Library/Frameworks/ApplicationServices.framework/Headers/ApplicationServices.h:30:
In file included from
/System/Library/Frameworks/ApplicationServices.framework/Frameworks/QD.framework/Headers/QD.h:25:
In file included from
/System/Library/Frameworks/ApplicationServices.framework/Frameworks/CoreGraphics.framework/Headers/CoreGraphics.h:14:
In file included from
/System/Library/Frameworks/ApplicationServices.framework/Frameworks/CoreGraphics.framework/Headers/CGBitmapContext.h:10:
In file included from
/System/Library/Frameworks/ApplicationServices.framework/Frameworks/CoreGraphics.framework/Headers/CGContext.h:13:
In file included from
/System/Library/Frameworks/ApplicationServices.framework/Frameworks/CoreGraphics.framework/Headers/CGColor.h:12:
In file included from
/System/Library/Frameworks/ApplicationServices.framework/Frameworks/CoreGraphics.framework/Headers/CGColorSpace.h:12:
In file included from
/System/Library/Frameworks/ApplicationServices.framework/Frameworks/CoreGraphics.framework/Headers/CGDataProvider.h:14:
/usr/include/unistd.h:582:7: error: conflicting types for 'valloc'
void    *valloc(size_t);                        
         ^
In file included from bla.c:1:
/usr/include/stdlib.h:312:7: error: previous declaration is here
void    *valloc(size_t);
         ^
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 Jul 22 09:52:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 09:52:28 -0500
Subject: [LLVMbugs] [Bug 2581] New: Useless select not removed
Message-ID: <bug-2581-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2581

           Summary: Useless select not removed
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: matthijs at stdin.nl
                CC: llvmbugs at cs.uiuc.edu


In the following fairly trivial program, the select is obviously completely
useless. However, it is not removed when running opt -std-compile-opts.

define i32 @run(i1 %C) nounwind  {
        br i1 %C, label %exit, label %body

body:           ; preds = %0
        %A = select i1 %C, i32 10, i32 11               ; <i32> [#uses=1]
        ret i32 %A

exit:           ; preds = %0
        ret i32 10
}


If I do a conditional branch instead of a select, simplifycfg seems to catch
that and remove the conditional. It makes sense that simlifycfg doesn't remove
a select, but I'm not sure which pass should catch this. I guess SCCP or
CondProp would be a good condidate?


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 22 14:41:25 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 14:41:25 -0500
Subject: [LLVMbugs] [Bug 2574] scalar_to_vector not selecting on x86
 (-mcpu=core2) for v2f32
In-Reply-To: <bug-2574-206@http.llvm.org/bugs/>
Message-ID: <200807221941.m6MJfPq6023851@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2574


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

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




--- Comment #1 from Evan Cheng <evan.cheng at apple.com>  2008-07-22 14:41:25 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080721/065301.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 Jul 22 15:24:07 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 15:24:07 -0500
Subject: [LLVMbugs] [Bug 2582] New: implicit cast of pointer to integer
 should generate warning
Message-ID: <bug-2582-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2582

           Summary: implicit cast of pointer to integer should generate
                    warning
           Product: clang
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


clang should warn about
--
int foo(float *x) {
  return x;
}
--


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 22 15:43:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 15:43:09 -0500
Subject: [LLVMbugs] [Bug 2566] Combine of bitcast and shuffle assumes vector
 to vector cast
In-Reply-To: <bug-2566-206@http.llvm.org/bugs/>
Message-ID: <200807222043.m6MKh9NX026044@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2566


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

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




--- Comment #8 from Evan Cheng <evan.cheng at apple.com>  2008-07-22 15:43:08 ---
Patch committed. The assumption in getShuffleScalarElt is safe. Also, I'll make
an appointment with an optometrist. :-)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 22 16:31:22 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 16:31:22 -0500
Subject: [LLVMbugs] [Bug 2583] New: Source location not tracked for nested
	macros
Message-ID: <bug-2583-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2583

           Summary: Source location not tracked for nested macros
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: richard at imendio.com
                CC: llvmbugs at cs.uiuc.edu


Constructs such as

  FOO(BAR(m)->abc)

result in error messages like:

FOO(BAR(m)->abc);
^~~
macro.c:18:15: error: no member named 'abc'

while it could ideally be:

  FOO(BAR(m)->abc);
         ~~~~~~  ^

Using macros like this might seem useless but is a very common pattern in
GObject based code (the object system used in GTK+/GNOME). In addition to just
giving clearer error message, it would be useful for a refactoring tool for
GTK+ I'm working on, which would work much better if there was more location
information available for nested macros.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 22 17:19:24 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 17:19:24 -0500
Subject: [LLVMbugs] [Bug 2584] New: Suboptimal codegen for <2 x i32>
	extractelement
Message-ID: <bug-2584-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2584

           Summary: Suboptimal codegen for <2 x i32> extractelement
           Product: new-bugs
           Version: unspecified
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nicolas at capens.net
                CC: llvmbugs at cs.uiuc.edu


The following LLVM IR extracts both the lower and upper dword from a 64-bit
register (MMX):

external constant <2 x i32>             ; <<2 x i32>*>:0 [#uses=1]
external constant i32           ; <i32*>:1 [#uses=1]
external constant i32           ; <i32*>:2 [#uses=1]

define internal void @""() {
        load <2 x i32>* @0, align 8             ; <<2 x i32>>:1 [#uses=2]
        extractelement <2 x i32> %1, i32 0              ; <i32>:2 [#uses=1]
        store i32 %2, i32* @1, align 4
        extractelement <2 x i32> %1, i32 1              ; <i32>:3 [#uses=1]
        store i32 %3, i32* @2, align 4
        ret void
}

It generates the following code:

  sub         esp,14h 
  movq        mm0,mmword ptr ds:[47E8700h] 
  movq        mmword ptr [esp+8],mm0 
  mov         eax,dword ptr [esp+8] 
  mov         dword ptr ds:[47E86FCh],eax 
  movq        mmword ptr [esp],mm0 
  mov         eax,dword ptr [esp+4] 
  mov         dword ptr ds:[47E86F8h],eax 
  add         esp,14h 
  ret   

This could be more optimal, like this:

  movq        mm0,mmword ptr ds:[47E8700h] 
  movd        eax,mm0
  mov         dword ptr ds:[47E86FCh],eax 
  punpckhdq   mm0,mm0 
  movd        eax,mm0
  mov         dword ptr ds:[47E86F8h],eax 
  ret   

That's significantly faster thanks to keeping things in registers (and the
first extract really became just a single movd).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 22 17:54:35 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 17:54:35 -0500
Subject: [LLVMbugs] [Bug 2585] New: Unoptimal vector 'trunc' emulation (and
	crash)
Message-ID: <bug-2585-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2585

           Summary: Unoptimal vector 'trunc' emulation (and crash)
           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: nicolas at capens.net
                CC: llvmbugs at cs.uiuc.edu


In an attempt to do an element-wise trunc of <4 x i32> to <4 x i16> I tried
using the following LLVM IR:

external constant <4 x i32>             ; <<4 x i32>*>:0 [#uses=1]
external constant <4 x i16>             ; <<4 x i16>*>:1 [#uses=1]

define internal void @""() {
        load <4 x i32>* @0, align 16            ; <<4 x i32>>:1 [#uses=1]
        bitcast <4 x i32> %1 to <8 x i16>               ; <<8 x i16>>:2
[#uses=1]
        shufflevector <8 x i16> %2, <8 x i16> undef, <8 x i32> < i32 0, i32 2,
i32 4, i32 6, i32 undef, i32 undef, i32 undef, i32 undef >               ; <<8
x i16>>:3 [#uses=1]
        bitcast <8 x i16> %3 to <2 x i64>               ; <<2 x i64>>:4
[#uses=1]
        extractelement <2 x i64> %4, i32 0              ; <i64>:5 [#uses=1]
        bitcast i64 %5 to <4 x i16>             ; <<4 x i16>>:6 [#uses=1]
        store <4 x i16> %6, <4 x i16>* @1, align 8
        ret void
}

Unfortunately there's a access vilation originating in
LowerVECTOR_SHUFFLEv8i16, due to the undefs in the shuffle mask. Using 02460246
instead gives me the following x86 code:

  push        ebp  
  mov         ebp,esp 
  and         esp,0FFFFFFF0h 
  sub         esp,40h 
  movaps      xmm0,xmmword ptr ds:[48C7700h] 
  pextrw      eax,xmm0,4 
  movaps      xmm1,xmm0 
  punpcklqdq  xmm1,xmm1 
  pshuflw     xmm1,xmm1,88h 
  pshufhw     xmm1,xmm1,88h 
  pinsrw      xmm1,eax,2 
  pextrw      ecx,xmm0,6 
  pinsrw      xmm1,ecx,3 
  pinsrw      xmm1,eax,6 
  pinsrw      xmm1,ecx,7 
  movaps      xmmword ptr [esp],xmm1 
  mov         eax,dword ptr [esp+4] 
  mov         dword ptr [esp+1Ch],eax 
  mov         eax,dword ptr [esp] 
  mov         dword ptr [esp+18h],eax 
  movq        mm0,mmword ptr [esp+18h] 
  movq        mmword ptr ds:[48C76F8h],mm0 
  mov         esp,ebp 
  pop         ebp  
  ret           

This is not very optimal. Instead I expected something like:

  push        ebp  
  mov         ebp,esp 
  and         esp,0FFFFFFF0h 
  sub         esp,40h 
  movaps      xmm0,xmmword ptr ds:[48C7700h] 
  pshuflw     xmm0,xmm0,0x88
  pshufhw     xmm0,xmm0,0x88
  pshufd      xmm0,xmm0,0x88
  movdq2q     mm0,xmm0
  movq        mmword ptr ds:[48C76F8h],mm0 
  mov         esp,ebp 
  pop         ebp  
  ret           

That's essentially 4 instead of 16 instructions!

Since a 'trunc' of <4 x i32> to <4 x i16> can be quite useful I think, it might
be worth it to have specialized codegen for this. Interestingly the
pshuflw+pshufhw already gets generated, but after that it fails to generate
pshufd+movdq2q.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 22 18:11:05 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 18:11:05 -0500
Subject: [LLVMbugs] [Bug 2586] New: clang is less permissive than gcc in
 accepting function redefinitions
Message-ID: <bug-2586-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2586

           Summary: clang is less permissive than gcc in accepting function
                    redefinitions
           Product: clang
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


gcc accepts the following code with a warning, whereas clang rejects
it.
--
void foo(void *);

foo(void *pfile) {
}
--
gcc even accepts
--
void foo(void *, int);

foo(void *pfile) {
}
--


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 22 18:54:57 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 18:54:57 -0500
Subject: [LLVMbugs] [Bug 2575] v4i32 vsetcc not legalizing correctly
In-Reply-To: <bug-2575-206@http.llvm.org/bugs/>
Message-ID: <200807222354.m6MNsvX5001543@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2575


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #3 from Duncan Sands <baldrick at free.fr>  2008-07-22 18:54:57 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080721/065311.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 Jul 22 19:22:37 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 22 Jul 2008 19:22:37 -0500
Subject: [LLVMbugs] [Bug 2485] Do all 4-element SSE shuffles in a maximum of
 two shuffle instructions
In-Reply-To: <bug-2485-206@http.llvm.org/bugs/>
Message-ID: <200807230022.m6N0MbLu002395@zion.cs.uiuc.edu>

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


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

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




--- Comment #10 from Evan Cheng <evan.cheng at apple.com>  2008-07-22 19:22:36 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080721/065312.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 Jul 23 16:06:05 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 23 Jul 2008 16:06:05 -0500
Subject: [LLVMbugs] [Bug 2589] New: The ARM code generator fails for a small
 code snippet
Message-ID: <bug-2589-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2589

           Summary: The ARM code generator fails for a small code snippet
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: ARM
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: rich at pennware.com
                CC: llvmbugs at cs.uiuc.edu


If I use

llvm-as -f t0026.ll
llc -f -march arm t0026.bc

on this snippet, I get an assert. All other targets seem to work.

; ModuleID = 't0026.bc'

define void @main({ i32 }*) {
entry:
        %sret1 = alloca { i32 }         ; <{ i32 }*> [#uses=2]
        load { i32 }* %sret1            ; <{ i32 }>:1 [#uses=1]
        ret void
}


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

From bugzilla-daemon at cs.uiuc.edu  Wed Jul 23 21:04:54 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 23 Jul 2008 21:04:54 -0500
Subject: [LLVMbugs] [Bug 2590] New: TableGen generates different output on
 different platforms
Message-ID: <bug-2590-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2590

           Summary: TableGen generates different output on different
                    platforms
           Product: tools
           Version: trunk
          Platform: PC
        OS/Version: NetBSD
            Status: NEW
          Severity: normal
          Priority: P2
         Component: TableGen
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: krister.walfridsson at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Running the the llvm testsuite on different platforms shows differences for
tests that should give identical result. For example
llvm/test/CodeGen/Alpha/add.ll fails on i386-unknown-netbsdelf4.99.69 but works
on x86_64-unknown-netbsd4.99.69. This is because TableGen generates different
output on different platforms.

The problem is that TableGen keeps some information sorted by pointer values,
and the output is generated by iterating over the list.

I tried to provide a custom ordering operator (that ordered the records by the
Record::Name) to

   std::map<Record*, DAGInstruction> Instructions;

in CodeGenDAGPatterns.h , and that gave me the same testsuite result on the two
machines, but there are still differences between the TableGen-generated files,
so there are probably more maps that should be updated.  But there are several
such candidate variables in CodeGenDAGPatterns.h, e.g.:

   std::map<Record*, SDNodeInfo> SDNodes;
   std::map<Record*, std::pair<Record*, std::string> > SDNodeXForms;
   std::map<Record*, ComplexPattern> ComplexPatterns;
   std::map<Record*, TreePattern*> PatternFragments;
   std::map<Record*, DAGDefaultOperand> DefaultOperands;


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 23 23:32:42 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 23 Jul 2008 23:32:42 -0500
Subject: [LLVMbugs] [Bug 2591] New: Internal compiler error in llvm-g++
Message-ID: <bug-2591-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2591

           Summary: Internal compiler error in llvm-g++
           Product: libraries
           Version: 2.1
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Transformation Utilities
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nadav256 at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Internal compiler error on llvm-g++

The error message:

/home/nadav/file.c: In function ???int main()???:
/home/nadav/file.c:1: 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.


The code:
 int main() {
  void* ptr;
  *(long*)&ptr[1] = 0;
 }


My friend Gil Dabah, of distorm fame, told me about 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  Thu Jul 24 05:06:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 24 Jul 2008 05:06:28 -0500
Subject: [LLVMbugs] [Bug 2560] Remove multiple return value support
In-Reply-To: <bug-2560-206@http.llvm.org/bugs/>
Message-ID: <200807241006.m6OA6SKx009348@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2560


Matthijs Kooijman <matthijs at stdin.nl> changed:

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




--- Comment #1 from Matthijs Kooijman <matthijs at stdin.nl>  2008-07-24 05:06:27 ---
In r53941, dan removed all internal support for multiple argument return
instructions and getresult. All code working with the IR does no longer need to
handle those.

There is still support for these instructions in bitcode and assembly files,
for backward compatibility. However, these instructions will get transformed
into insert/extractvalue instructions automatically.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 24 16:52:37 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 24 Jul 2008 16:52:37 -0500
Subject: [LLVMbugs] [Bug 2566] Combine of bitcast and shuffle assumes vector
 to vector cast
In-Reply-To: <bug-2566-206@http.llvm.org/bugs/>
Message-ID: <200807242152.m6OLqblE008889@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2566


Nicolas Capens <nicolas at capens.net> changed:

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




--- Comment #9 from Nicolas Capens <nicolas at capens.net>  2008-07-24 16:52:36 ---
(In reply to comment #8)
> Patch committed.

Thanks!

> The assumption in getShuffleScalarElt is safe.

getShuffleScalarElt itself is indeed correct, however, PerformShuffleCombine in
X86ISelLowering incorrectly assumes it's dealing with a vector to vector
bitcast. So to clarify, I'm getting an assert with the following call stack:

assert
llvm::MVT::getVectorNumElements
llvm::SelectionDAG::getShuffleScalarElt
EltsFromConsecutiveLoads
PerformShuffleCombine
llvm::X86TargetLowering::PerformDAGCombine
...


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 24 17:46:26 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 24 Jul 2008 17:46:26 -0500
Subject: [LLVMbugs] [Bug 2591] Internal compiler error in llvm-g++
In-Reply-To: <bug-2591-206@http.llvm.org/bugs/>
Message-ID: <200807242246.m6OMkQuu010782@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2591


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

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




--- Comment #3 from Evan Cheng <evan.cheng at apple.com>  2008-07-24 17:46:26 ---
Ok, we won't patch 2.1. Please move to 2.3.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 24 19:55:46 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 24 Jul 2008 19:55:46 -0500
Subject: [LLVMbugs] [Bug 2589] The ARM code generator fails for a small code
	snippet
In-Reply-To: <bug-2589-206@http.llvm.org/bugs/>
Message-ID: <200807250055.m6P0tkQ7014870@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2589


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

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




--- Comment #1 from Evan Cheng <evan.cheng at apple.com>  2008-07-24 19:55:45 ---
Fixed.
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080721/065355.html


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

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 24 20:45:46 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 24 Jul 2008 20:45:46 -0500
Subject: [LLVMbugs] [Bug 2594] New: crash in SelectionDAGLowering with
 va_arg applied to struct type
Message-ID: <bug-2594-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2594

           Summary: crash in SelectionDAGLowering with va_arg applied to
                    struct type
           Product: new-bugs
           Version: unspecified
          Platform: Macintosh
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


llc crashes in SelectionDAGLowering on the following code:
--
%S = type <{ i32, i32 }>

define void @test() {
        va_arg i8** null, %S
        unreachable
}
--
ddunbar at il0102b-dhcp47:rt$ llvm-as -o - boom.ll | llc
Assertion failed: (0 && "Unknown type!"), function getMVT, file ValueTypes.cpp,
line 109.
0   llc                                 0x00774783 _ZSt4findIPKccET_S2_S2_RKT0_
+ 97


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 25 10:00:42 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 25 Jul 2008 10:00:42 -0500
Subject: [LLVMbugs] [Bug 2595] New: llc/lli asserts during code generation
Message-ID: <bug-2595-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2595

           Summary: llc/lli asserts during code generation
           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: luked at cs.rochester.edu
                CC: gohman at apple.com, llvmbugs at cs.uiuc.edu


Created an attachment (id=1862)
 --> (http://llvm.org/bugs/attachment.cgi?id=1862)
Bugpoint reduction with partially cleaned names (I got rid of some C++ name
cruft)

Reverting r54000 fixes issue.

Assertion failed: (I->contains(Start) && I->contains(End-1) && "Range is not
entirely in interval!"), function removeRange, file
[llvm-svn]/lib/CodeGen/LiveInterval.cpp, line 257.
0   llc                                 0x0077751b _ZSt4findIPKccET_S2_S2_RKT0_
+ 97
1   llc                                 0x007777e5
_ZN4llvm3sys28PrintStackTraceOnErrorSignalEv + 593
2   libSystem.B.dylib                   0x9369509b _sigtramp + 43
3   ???                                 0xffffffff 0x0 + 4294967295
4   libSystem.B.dylib                   0x9370dec2 raise + 26
5   libSystem.B.dylib                   0x9371d47f abort + 73
6   libSystem.B.dylib                   0x9370f063 __assert_rtn + 101
7   llc                                 0x00537592
_ZN4llvm12LiveInterval11removeRangeEjjb + 270
8   llc                                 0x0053b285
_ZN4llvm13LiveIntervals29spillPhysRegAroundRegDefsUsesERKNS_12LiveIntervalEjRNS_10VirtRegMapE
+ 555
9   llc                                 0x005a4272
_ZN4llvm33createLinearScanRegisterAllocatorEv + 7388
10  llc                                 0x005a51ac
_ZN4llvm33createLinearScanRegisterAllocatorEv + 11286
11  llc                                 0x005a5983
_ZN4llvm33createLinearScanRegisterAllocatorEv + 13293
12  llc                                 0x000547a0
_ZN4llvm19MachineFunctionPass13runOnFunctionERNS_8FunctionE + 40
13  llc                                 0x006fd077
_ZN4llvm13FPPassManager13runOnFunctionERNS_8FunctionE + 331
14  llc                                 0x006fd2ec
_ZN4llvm23FunctionPassManagerImpl3runERNS_8FunctionE + 110
15  llc                                 0x006fd46a
_ZN4llvm19FunctionPassManager3runERNS_8FunctionE + 156
16  llc                                 0x00002e75 main + 2609
17  llc                                 0x00001c36 start + 54
Abort trap


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

From bugzilla-daemon at cs.uiuc.edu  Fri Jul 25 11:59:34 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 25 Jul 2008 11:59:34 -0500
Subject: [LLVMbugs] [Bug 2596] New: valgrind error in VirtRegMap.cpp
Message-ID: <bug-2596-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2596

           Summary: valgrind error in VirtRegMap.cpp
           Product: libraries
           Version: trunk
          Platform: All
        OS/Version: All
            Status: NEW
          Keywords: compile-fail
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: gohman at apple.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1863)
 --> (http://llvm.org/bugs/attachment.cgi?id=1863)
reduced testcase

valgrind turned up an out-of-bounds access in VirtRegMap.cpp.
The following assert demonstrates it, with the attached testcase
Compile with llc -march=x86.

--- lib/CodeGen/VirtRegMap.cpp
+++ lib/CodeGen/VirtRegMap.cpp
@@ -562,6 +562,7 @@ static void InvalidateKills(MachineInstr &MI,
BitVector &RegKills,
    unsigned Reg = MO.getReg();
    if (KillRegs)
      KillRegs->push_back(Reg);
+    assert(Reg < KillOps.size());
    if (KillOps[Reg] == &MO) {
      RegKills.reset(Reg);
      KillOps[Reg] = NULL;


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 25 15:49:35 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 25 Jul 2008 15:49:35 -0500
Subject: [LLVMbugs] [Bug 2598] New: fptosint fails when asked to convert
 v2f32 (0.0, 0, 0)   to v2i32 (0, 0)
Message-ID: <bug-2598-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2598

           Summary: fptosint fails when asked to convert v2f32 (0.0,0,0) to
                    v2i32 (0,0)
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pgurd at rapidmind.com
                CC: llvmbugs at cs.uiuc.edu, sdt at rapidmind.com


Created an attachment (id=1864)
 --> (http://llvm.org/bugs/attachment.cgi?id=1864)
llvm ir file which reproduces the problem

Running the attached file through llvm-as and llc produces the error

Cannot yet select: 0x8e3d1f4: v2i32 = fp_to_sint 0x8e3d524


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 25 18:45:21 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Fri, 25 Jul 2008 18:45:21 -0500
Subject: [LLVMbugs] [Bug 2599] New: Missed DSE case
Message-ID: <bug-2599-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2599

           Summary: Missed DSE case
           Product: libraries
           Version: 1.0
          Platform: PC
        OS/Version: All
            Status: NEW
          Keywords: code-quality
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: clattner at apple.com
                CC: llvmbugs at cs.uiuc.edu


DSE should delete a store to a location when the stored value is the same as
the loaded one.  For example:

void foo(_Complex int *x) {
  *x = ~*x;
}

compiles to:

define void @foo({ i32, i32 }* %x) nounwind  {
entry:
        getelementptr { i32, i32 }* %x, i32 0, i32 0            ; <i32*>:0
[#uses=2]
        load i32* %0, align 4           ; <i32>:1 [#uses=1]
        getelementptr { i32, i32 }* %x, i32 0, i32 1            ; <i32*>:2
[#uses=2]
        load i32* %2, align 4           ; <i32>:3 [#uses=1]
        sub i32 0, %3           ; <i32>:4 [#uses=1]
        store i32 %1, i32* %0, align 4
        store i32 %4, i32* %2, align 4
        ret void
}

The load+store to %0 should be nuked, which would happen if DSE deleted the
store to %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 Jul 26 00:13:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 26 Jul 2008 00:13:19 -0500
Subject: [LLVMbugs] [Bug 2403] crash on invalid typedef declarations
In-Reply-To: <bug-2403-206@http.llvm.org/bugs/>
Message-ID: <200807260513.m6Q5DJXX017491@zion.cs.uiuc.edu>

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


Chris Lattner <clattner at apple.com> changed:

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




--- Comment #1 from Chris Lattner <clattner at apple.com>  2008-07-26 00:13:18 ---
This works now:

t.c:3:1: error: cannot combine with previous 'typedef' declaration specifier
typedef __builtin_va_list __gnuc_va_list;int a() { return ({L:5;}); }
^
t.c:3:9: error: typedef redefinition with different types ('struct <anonymous>'
vs 'char *')
typedef __builtin_va_list __gnuc_va_list;int a() { return ({L:5;}); }
        ^
<predefines>:4:15: error: previous definition is here
typedef char* __builtin_va_list;
              ^
t.c:3:27: error: parse error
typedef __builtin_va_list __gnuc_va_list;int a() { return ({L:5;}); }
                          ^
t.c:3:52: error: incompatible type returning 'void', expected 'int'
typedef __builtin_va_list __gnuc_va_list;int a() { return ({L:5;}); }
                                                   ^      ~~~~~~~~


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 26 14:15:28 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 26 Jul 2008 14:15:28 -0500
Subject: [LLVMbugs] [Bug 2416] acceptance of invalid enum syntax
In-Reply-To: <bug-2416-206@http.llvm.org/bugs/>
Message-ID: <200807261915.m6QJFSgp022577@zion.cs.uiuc.edu>

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


Chris Lattner <clattner at apple.com> changed:

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




--- Comment #1 from Chris Lattner <clattner at apple.com>  2008-07-26 14:15:26 ---
We accept this as an extension and emit a warning when -pedantic is used.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 26 14:51:48 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 26 Jul 2008 14:51:48 -0500
Subject: [LLVMbugs] [Bug 2374] sema calculates incorrect type for statement
 expression with label
In-Reply-To: <bug-2374-206@http.llvm.org/bugs/>
Message-ID: <200807261951.m6QJpmaQ023677@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2374


Chris Lattner <clattner at apple.com> changed:

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




--- Comment #1 from Chris Lattner <clattner at apple.com>  2008-07-26 14:51:48 ---
Fixed:
http://lists.cs.uiuc.edu/pipermail/cfe-commits/Week-of-Mon-20080721/006746.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 Jul 26 23:47:29 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sat, 26 Jul 2008 23:47:29 -0500
Subject: [LLVMbugs] [Bug 2601] New: clang generates wrong AST/ code for
 compound assignment operators
Message-ID: <bug-2601-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2601

           Summary: clang generates wrong AST/code for compound assignment
                    operators
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Semantic Analyzer
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: sharparrow1 at yahoo.com
                CC: llvmbugs at cs.uiuc.edu


Per subject; a couple of testcases:

unsigned char a,b; void c(void) {a <<= b;} void d(void) {a /= b;}

The issue is that the functions for computing the types skip unary promotions
for compound assignments.  I'm not sure what the cleanest way to solve this is,
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  Sun Jul 27 03:23:33 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 27 Jul 2008 03:23:33 -0500
Subject: [LLVMbugs] [Bug 2602] New: OOM in SCEV with polynomial induction
	variable
Message-ID: <bug-2602-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2602

           Summary: OOM in SCEV with polynomial induction variable
           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: nicholas at mxc.ca, llvmbugs at cs.uiuc.edu


Testcase:
int x(int); int a(void) {unsigned q=0,w=0,e=0; do {q+=w;w+=e;e+=1;x(q);} while
(q != 0xFFFFFFFF);}

crashes with OOM in llvm-gcc. The issue is in
SCEVAddRecExpr::getNumIterationsInRange.

I think Nick Lewycky said he would do this on IRC, but I don't want it to get
lost.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 27 16:32:32 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Sun, 27 Jul 2008 16:32:32 -0500
Subject: [LLVMbugs] [Bug 2603] New: assertion emitting constant cast expr in
	AsmPrinter
Message-ID: <bug-2603-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2603

           Summary: assertion emitting constant cast expr in AsmPrinter
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: nlewycky at google.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1865)
 --> (http://llvm.org/bugs/attachment.cgi?id=1865)
testcase

$ x86_64-unknown-linux-gnu/bin/x86_64-unknown-linux-gnu-c++ 000.c
cc1plus:
/usr/local/google/tmp/crosstool.9nJarU/gcc-llvm-glibc-2.3.6-grte/rpmbuild/BUILD/crosstoolv12-gcc-llvm-glibc-2.3.6-grte-0.1/llvm/lib/CodeGen/AsmPrinter.cpp:828:
void llvm::AsmPrinter::EmitConstantValueOnly(const llvm::Constant*): Assertion
`0 && "FIXME: Don't yet support this kind of constant cast expr"' failed.
000.c:19: internal compiler error: Aborted
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://developer.apple.com/bugreporter> for instructions.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 28 11:32:43 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 28 Jul 2008 11:32:43 -0500
Subject: [LLVMbugs] [Bug 2599] Missed DSE case
In-Reply-To: <bug-2599-206@http.llvm.org/bugs/>
Message-ID: <200807281632.m6SGWhLW012422@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2599


Owen Anderson <resistor at mac.com> changed:

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




--- Comment #2 from Owen Anderson <resistor at mac.com>  2008-07-28 11:32:42 ---
Fixed in r54133.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 28 15:13:52 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 28 Jul 2008 15:13:52 -0500
Subject: [LLVMbugs] [Bug 2604] New: mfocrf definition in PPCInstrInfo. td
 contradicts PowerPC spec
Message-ID: <bug-2604-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2604

           Summary: mfocrf definition in PPCInstrInfo.td contradicts PowerPC
                    spec
           Product: libraries
           Version: 2.3
          Platform: PC
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: PowerPC
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: maxburke at gmail.com
                CC: llvmbugs at cs.uiuc.edu


The definition for the "mfocrf" instruction in PPCInstrInfo.td appears to be
incorrect:

def MFOCRF: XFXForm_5a<31, 19, (outs GPRC:$rT), (ins crbitm:$FXM),
                       "mfcr $rT, $FXM", SprMFCR>,

According to the PowerPC programming environment manual, there is only one form
of mfcr which takes only a destination register as a parameter. This should
read as follows:

def MFOCRF: XFXForm_5a<31, 19, (outs GPRC:$rT), (ins crbitm:$FXM),
                       "mfocrf $rT, $FXM", SprMFCR>,


Reference link:
http://publib.boulder.ibm.com/infocenter/systems/topic/com.ibm.aix.aixassem/doc/alangref/mfocrf.htm


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 28 18:09:24 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Mon, 28 Jul 2008 18:09:24 -0500
Subject: [LLVMbugs] [Bug 2605] New: Fail to build AsmParser without bison on
	Win32
Message-ID: <bug-2605-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2605

           Summary: Fail to build AsmParser without bison on Win32
           Product: Build scripts
           Version: trunk
          Platform: PC
        OS/Version: Windows NT
            Status: NEW
          Severity: minor
          Priority: P2
         Component: Makefiles
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: asitgoes at gmail.com
                CC: llvmbugs at cs.uiuc.edu


in 'win32/dobison.cmd' script, the output directory for 'llvmAsmParser.cpp',
and 'llvmAsmParser.h' is different when they are generated using bison from
when they are generated using pre-generated files. Thus, the 'AsmParser'
project fails to build when 'bison' cannot be found in the path. 

This could easily be fixed by make the output directories for both cases the
same and the project has correct path to the generated 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  Tue Jul 29 02:38:45 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 02:38:45 -0500
Subject: [LLVMbugs] [Bug 2578] TrimAllocationToSize will trim blocks to
 smaller than FreeRangeHeader:: getMinBlockSize()
In-Reply-To: <bug-2578-206@http.llvm.org/bugs/>
Message-ID: <200807290738.m6T7cjnC008647@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2578


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>  2008-07-29 02:38:44 ---
Patch committed. Thanks!


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

From bugzilla-daemon at cs.uiuc.edu  Tue Jul 29 05:03:50 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 05:03:50 -0500
Subject: [LLVMbugs] [Bug 2498] Argpromotion promotes possibly invalid loads
In-Reply-To: <bug-2498-206@http.llvm.org/bugs/>
Message-ID: <200807291003.m6TA3oFZ017813@zion.cs.uiuc.edu>

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


Matthijs Kooijman <matthijs at stdin.nl> changed:

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




--- Comment #14 from Matthijs Kooijman <matthijs at stdin.nl>  2008-07-29 05:03:49 ---
It took a while to get this finished, I was a bit busy with other things and
the fix required some changes to GetElementPtrInst as well.

Anyway, I've commited (a modified version of) the fix here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080728/065441.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 Jul 29 10:16:53 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 10:16:53 -0500
Subject: [LLVMbugs] [Bug 2606] New: global variables declared in module A
 but defined in module B are not resolved by the JIT
Message-ID: <bug-2606-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2606

           Summary: global variables declared in module A but defined in
                    module B are not resolved by the JIT
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Generic Execution Engine Support
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: paul at rapidmind.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1868)
 --> (http://llvm.org/bugs/attachment.cgi?id=1868)
modified HowToUseJIT.cpp which reproduces the problem

If you declare and use a global variable in module A and define it in module B
the execution engine will not resolve the symbol in module A.

The attached modified version of HowToUseJIT.cpp shows this problem. You can
manually work around this limitation by explicitly calling addGlobalMapping on
the declaration using the same pointer that's mapped to the definition.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 29 11:25:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 11:25:39 -0500
Subject: [LLVMbugs] [Bug 2607] New: SCEV miscomputes trip count for loop
 with SGT condition
Message-ID: <bug-2607-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2607

           Summary: SCEV miscomputes trip count for loop with SGT condition
           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 <stdio.h>
int a(unsigned j) {unsigned i; for (i = 0; (int)i > (int)j; i--) {} return i;}
int main() {printf("%d", a(0x80000000));}

Compare the output with llvm-gcc with "-O0" vs. "-O3 -fno-inline" to see the
issue.

Relevant chunk from ScalarEvolutionsImpl::ComputeIterationCount:
  case ICmpInst::ICMP_SGT: {
    SCEVHandle TC = HowManyLessThans(SE.getNegativeSCEV(LHS),
                                     SE.getNegativeSCEV(RHS), L, true);
    if (!isa<SCEVCouldNotCompute>(TC)) return TC;
    break;
  }

The calls to getNegativeSCEV should be calls to getNotSCEV; negation doesn't
properly flip the comparison when one side is INT_MIN.  (It looked suspicious
when I saw it, although it took a little while to come up with a testcase.)

I'll throw together a patch later.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 29 13:41:38 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 13:41:38 -0500
Subject: [LLVMbugs] [Bug 2608] New: The function to stup map in
 JITResolverState does not get updated when a function is released
Message-ID: <bug-2608-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2608

           Summary: The function to stup map in JITResolverState does not
                    get updated when a function is released
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: ddneff at hotmail.com
                CC: llvmbugs at cs.uiuc.edu, ddneff at hotmail.com


I am currently running into a crash while using the JIT engine where I try to
call Function B, but Function A ends up getting called instead.  Both Function
A and Function B are extern "C" functions in this case.  The reason for the
crash is that the StubToFunctionMap and the FunctionToStubMap inside of
JITResolverState never get updated when a llvm::Function is deleted, even if
you properly call freeMachineCodeForFunction.  A stub is created for Function A
during the JIT process, but when Function A is later deleted the stub mapping
never goes away.  Eventually another llvm::Function will be created that is
given the same exact address as Function A had (stupid memory manager), and
when you try to call this function it ends up calling the stub for the
previously deleted function because of the stale map entry.   I believe that
someplace in either freeMachineCodeForFunction, llvm::Function::~Function, or
some other relevant location these stub mappings need to be cleared out to
prevent stubs from being used after a function is released.  I have tried
adding such code to JITEmitter::deallocateMemForFunction and it fixes my
problem, but the entire stub process is sufficiently complicated enough that I
am not convinced that my solution will work in all cases (it looks like the
stub is also placed into the GOT, and I'm not sure how to correctly deal with
that when the function goes away).


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 30 00:22:55 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 30 Jul 2008 00:22:55 -0500
Subject: [LLVMbugs] [Bug 2611] New: smax scev is redundant with umax
Message-ID: <bug-2611-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2611

           Summary: smax scev is redundant with umax
           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


Per discussion with nicholas on IRC, the SMAX and UMAX scevs are redundant; "a
smax b" can be rewritten as "((a + INT_MIN) umax (b + INT_MIN)) + INT_MIN".

Quick explanation of the reasoning: the range of signed integers is from
INT_MIN to INT_MAX.  The range of unsigned integers is from 0 to UINT_MAX.  By
adding INT_MIN to a number, it is mapped from the signed integer range to the
unsigned integer range in the correct order.  Then, the umax returns the larger
number, and then adding INT_MIN again restores the result into the range of
signed numbers.

For implementation, the idea is mostly straightforward: just replace
getSMaxExpr with a getSMaxSCEV which does this mapping.  The subtle point is
not getting any regressions in code quality.  The ScalarEvolutionExpander
probably should be made a bit smarter, and instcombine probably should be made
smart enough to deal with expressions like this (currently, instcombine can't
fold the naive expansion at all).

2's complement arithmetic and comparisons is so much fun.  Good way to make
your head spin. :)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 30 04:34:30 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 30 Jul 2008 04:34:30 -0500
Subject: [LLVMbugs] [Bug 2612] New: Code snippet fails with assert
Message-ID: <bug-2612-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2612

           Summary: Code snippet fails with assert
           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: rich at pennware.com
                CC: llvmbugs at cs.uiuc.edu


The following code fails with an assert both in llvm-ld and llc:

; ModuleID = 't0034.ubc'
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-s0:0:64-f80:32:32"
target triple = "i686-pc-linux-gnu"
@current_foo = internal global {  } zeroinitializer             ; <{  }*>
[#uses=1]

define i32 @main(...) {
entry:
        %retval = alloca i32            ; <i32*> [#uses=2]
        store i32 0, i32* %retval
        %local_foo = alloca {  }                ; <{  }*> [#uses=1]
        load {  }* @current_foo         ; <{  }>:0 [#uses=1]
        store {  } %0, {  }* %local_foo
        br label %return

return:         ; preds = %entry
        load i32* %retval               ; <i32>:1 [#uses=1]
        ret i32 %1
}

llc:
/home/rich/llvm-trunk-new/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp:1263:
llvm::SDOperand llvm::SelectionDAGLowering::getValue(const llvm::Value*):
Assertion `InReg && "Value not in map!"' failed.
llc[0x8b5c22a]
llc[0x8b5c35c]
[0x110400]
/lib/libc.so.6(abort+0x101)[0xa62f91]
/lib/libc.so.6(__assert_fail+0xee)[0xa5a93e]
llc(_ZN4llvm20SelectionDAGLowering8getValueEPKNS_5ValueE+0x1068)[0x882f89a]
llc(_ZN4llvm20SelectionDAGLowering10visitStoreERNS_9StoreInstE+0x3e)[0x8837a32]
llc(_ZN4llvm20SelectionDAGLowering5visitEjRNS_4UserE+0x28f)[0x885d89b]
llc(_ZN4llvm20SelectionDAGLowering5visitERNS_11InstructionE+0x28)[0x885db38]
llc(_ZN4llvm16SelectionDAGISel17BuildSelectionDAGERNS_12SelectionDAGEPNS_10BasicBlockERSt6vectorISt4pairIPNS_12MachineInstrEjESaIS9_EERNS_20FunctionLoweringInfoE+0x3a4)[0x8840922]
llc(_ZN4llvm16SelectionDAGISel16SelectBasicBlockEPNS_10BasicBlockERNS_15MachineFunctionERNS_20FunctionLoweringInfoERNS_5alistINS_6SDNodeENS_10LoadSDNodeEEE+0xb7)[0x884113b]
llc(_ZN4llvm16SelectionDAGISel20SelectAllBasicBlocksERNS_8FunctionERNS_15MachineFunctionERNS_20FunctionLoweringInfoE+0x8f)[0x8842981]
llc(_ZN4llvm16SelectionDAGISel13runOnFunctionERNS_8FunctionE+0x210)[0x88434e2]
llc[0x8748cbb]
llc(_ZN4llvm13FPPassManager13runOnFunctionERNS_8FunctionE+0x135)[0x8ae9847]
llc(_ZN4llvm23FunctionPassManagerImpl3runERNS_8FunctionE+0x6e)[0x8ae9acc]
llc(_ZN4llvm19FunctionPassManager3runERNS_8FunctionE+0x89)[0x8ae9c1b]
llc(main+0x9a2)[0x83d6244]
/lib/libc.so.6(__libc_start_main+0xe0)[0xa4e390]
llc[0x83d48d1]
Abort

llvm-ld: /home/rich/llvm-trunk-new/include/llvm/Support/Casting.h:199: typename
llvm::cast_retty<To, From>::ret_type llvm::cast(const Y&) [with X =
llvm::StoreInst, Y = llvm::User*]: Assertion `isa<X>(Val) && "cast<Ty>()
argument of incompatible type!"' failed.
llvm-ld[0x84e6ffa]
llvm-ld[0x84e712c]
[0x110400]
/lib/libc.so.6(abort+0x101)[0xa62f91]
/lib/libc.so.6(__assert_fail+0xee)[0xa5a93e]
llvm-ld(_ZN4llvm4castINS_9StoreInstEPNS_4UserEEENS_10cast_rettyIT_T0_E8ret_typeERKS6_+0x4c)[0x831bdfa]
llvm-ld[0x833da78]
llvm-ld(_ZN4llvm13MPPassManager11runOnModuleERNS_6ModuleE+0x100)[0x847534e]
llvm-ld(_ZN4llvm15PassManagerImpl3runERNS_6ModuleE+0x6e)[0x8475506]
llvm-ld(_ZN4llvm11PassManager3runERNS_6ModuleE+0x1b)[0x8475559]
llvm-ld(_ZN4llvm8OptimizeEPNS_6ModuleE+0x415)[0x8264d7b]
llvm-ld(main+0x6b0)[0x826c6a8]
/lib/libc.so.6(__libc_start_main+0xe0)[0xa4e390]
llvm-ld(__gxx_personality_v0+0x309)[0x8264311]
Abort


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 29 19:26:39 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 19:26:39 -0500
Subject: [LLVMbugs] [Bug 2607] SCEV miscomputes trip count for loop with SGT
	condition
In-Reply-To: <bug-2607-206@http.llvm.org/bugs/>
Message-ID: <200807300026.m6U0Qdti015579@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2607


Eli Friedman <sharparrow1 at yahoo.com> changed:

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




--- Comment #1 from Eli Friedman <sharparrow1 at yahoo.com>  2008-07-29 19:26:39 ---
Fixed in r54184.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 29 17:57:50 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 17:57:50 -0500
Subject: [LLVMbugs] [Bug 2610] New: missing function in the ocaml bindings
Message-ID: <bug-2610-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2610

           Summary: missing function in the ocaml bindings
           Product: new-bugs
           Version: unspecified
          Platform: All
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: oandrieu at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1870)
 --> (http://llvm.org/bugs/attachment.cgi?id=1870)
bind the LLVMAddCase function in ocaml

Looks like the ocaml bindings are missing the SwitchInst::addCase method. 

Unless I'm missing something, it makes it a bit hard to generate switch
instructions :)


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 29 16:08:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 16:08:09 -0500
Subject: [LLVMbugs] [Bug 2609] unknown exception handling label
In-Reply-To: <bug-2609-206@http.llvm.org/bugs/>
Message-ID: <200807292108.m6TL89Vv008355@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2609


Duncan Sands <baldrick at free.fr> changed:

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




--- Comment #1 from Duncan Sands <baldrick at free.fr>  2008-07-29 16:08:08 ---
Fixed in commit 54172.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 29 14:42:11 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Tue, 29 Jul 2008 14:42:11 -0500
Subject: [LLVMbugs] [Bug 2609] New: unknown exception handling label
Message-ID: <bug-2609-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2609

           Summary: unknown exception handling label
           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=1869)
 --> (http://llvm.org/bugs/attachment.cgi?id=1869)
testcase .ll

This breaks the llvm-gcc x86-32 linux build.

Compile the attached testcase with llc then try to assemble the result:

$ llc -f istream.bc && as -o istream.o istream.s
istream.s: Assembler messages:
istream.s:222: Error: can't resolve `.Llabel6' {*UND* section} -
`.Leh_func_begin1' {.text section}


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 30 13:34:56 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 30 Jul 2008 13:34:56 -0500
Subject: [LLVMbugs] [Bug 2613] New: Debugging labels are not being
 invalidated after opt pass
Message-ID: <bug-2613-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2613

           Summary: Debugging labels are not being invalidated after opt
                    pass
           Product: libraries
           Version: trunk
          Platform: Other
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Common Code Generator Code
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: maxburke at gmail.com
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1871)
 --> (http://llvm.org/bugs/attachment.cgi?id=1871)
Test case exhibiting a missing label referred to within debug info

Occasionally there seem to be labels popping up in the DWARF line number
information sections that have been removed from the code. The attached .ll
file when built with the following commands refers to a "label23" in the line
numbers but does not exist in code.

llvm-as -f -o=md5.bc md5.ll
opt -o=md5.opt.bc -f md5.bc -std-compile-opts
llc -asm-verbose -march=ppc64 -f -o=md5.s md5.opt.bc

This has been seen in both 2.3 and in the trunk as of revision 54223


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 30 13:37:50 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 30 Jul 2008 13:37:50 -0500
Subject: [LLVMbugs] [Bug 2612] Code snippet fails with assert
In-Reply-To: <bug-2612-206@http.llvm.org/bugs/>
Message-ID: <200807301837.m6UIbo91001737@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2612


Dan Gohman <gohman at apple.com> changed:

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




--- Comment #1 from Dan Gohman <gohman at apple.com>  2008-07-30 13:37:49 ---
Fixed here:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20080728/065508.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 Jul 30 15:49:03 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Wed, 30 Jul 2008 15:49:03 -0500
Subject: [LLVMbugs] [Bug 2612] Code snippet fails with assert
In-Reply-To: <bug-2612-206@http.llvm.org/bugs/>
Message-ID: <200807302049.m6UKn3O4006579@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2612


Richard Pennington <rich at pennware.com> changed:

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




--- Comment #2 from Richard Pennington <rich at pennware.com>  2008-07-30 15:49:03 ---
This works with llc but still fails in llvm-ld. 


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 31 02:52:10 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 31 Jul 2008 02:52:10 -0500
Subject: [LLVMbugs] [Bug 2614] New: Redefined symbols with llvm-gcc with
 -std=c99 and -O1 ( or higher)
Message-ID: <bug-2614-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2614

           Summary: Redefined symbols with llvm-gcc with -std=c99 and -O1
                    (or higher)
           Product: tools
           Version: 2.3
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: llvm-gcc
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: llvm at assumetheposition.nl
                CC: llvmbugs at cs.uiuc.edu


Given two very simple .c files:

main.c
------
#include <stdlib.h>

int
main(int argc, char **argv)
{
}

module.c
--------
#include <stdlib.h>


If I compile each of these with plain gcc using -std=c99 and -O1 followed by
linking them, all is well. But if I replace gcc with llvm-gcc I get multiply
defined symbols:

llvm-gcc -c -O1 -std=c99                 main.c
llvm-gcc -c -O1 -std=c99                 module.c
llvm-gcc -o main main.o module.o
module.o(.text+0x0): In function `strtod':
: multiple definition of `strtod'
main.o(.text+0x0): first defined here
module.o(.text+0x30): In function `strtoull':
: multiple definition of `strtoull'
main.o(.text+0x40): first defined here
module.o(.text+0x60): In function `strtoll':
: multiple definition of `strtoll'
main.o(.text+0x70): first defined here
module.o(.text+0x90): In function `strtold':
: multiple definition of `strtold'
main.o(.text+0xa0): first defined here
module.o(.text+0xc0): In function `strtof':
[...]

Looking at the assembly output it seems the string functions get pulled into
both modules and therefore clash. E.g. llvm-gcc -S -O1 -std=c99 module.c
produces

    .text
    .align  16
    .globl  strtod
    .type   strtod, at function
strtod:
    pushl   %ebp
    movl    %esp, %ebp
    subl    $16, %esp
    movl    12(%ebp), %ecx
    movl    %ecx, 4(%esp)
    movl    8(%ebp), %ecx
    movl    %ecx, (%esp)
    movl    $0, 8(%esp)
    call    __strtod_internal
    addl    $16, %esp
    popl    %ebp
    ret
    .size   strtod, .-strtod


    .align  16
    .globl  strtoull
    .type   strtoull, at function
strtoull:
    pushl   %ebp
    movl    %esp, %ebp
    subl    $16, %esp
    movl    16(%ebp), %ecx
    movl    %ecx, 8(%esp)
    movl    12(%ebp), %ecx
    movl    %ecx, 4(%esp)
    movl    8(%ebp), %ecx
    movl    %ecx, (%esp)
    movl    $0, 12(%esp)
    call    __strtoull_internal
    addl    $16, %esp
    popl    %ebp
    ret
    .size   strtoull, .-strtoull

[...]

while module.s also contains the exact same functions.

Running llvm-gcc -S -O1 -std=c99 module.c (i.e. without the c99 flag) produces
merely

        .section        .note.GNU-stack,"", at progbits
        .ident  "GCC: (GNU) 4.2.1 (Based on Apple Inc. build 5555) (LLVM build
2.3)"

Using -std=c99 without any optimization (-O0) also produces the same output
(jus t the two lines).

This is on a Fedora Core 4 system (whose gcc version is 4.0.2), with LLVM 2.3
and the precompiled RHEL4 llvm-gcc-4.2 binaries.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 31 04:24:01 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 31 Jul 2008 04:24:01 -0500
Subject: [LLVMbugs] [Bug 2614] Redefined symbols with llvm-gcc with -std=c99
 and -O1 (or higher )
In-Reply-To: <bug-2614-206@http.llvm.org/bugs/>
Message-ID: <200807310924.m6V9O1uw014679@zion.cs.uiuc.edu>

http://llvm.org/bugs/show_bug.cgi?id=2614


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

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




--- Comment #9 from Anton Korobeynikov <asl at math.spbu.ru>  2008-07-31 04:24:00 ---
(In reply to comment #6)
> However I get the same if I compile using mainline gcc.
> This is on x86-64 linux with gcc 4.3.  Interestingly
> enough, on x86-32 linux with gcc 4.2 I get the same
> from llvm-gcc, but no errors from gcc-4.2.  In any case
> I see no complaints about strtod and friends.
The 'inline' semantics were changed for 4.2 to be C99-consistent. Bogus
includes (which rely on old semantics) should be fixed by fixincludes during
gcc compilation. Surely, this is not true for this case, since prebuilt
binaries were used. 

See:
http://gcc.gnu.org/ml/gcc-patches/2006-05/msg00328.html

*** This bug has been marked as a duplicate of bug 1699 ***


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 31 04:39:34 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 31 Jul 2008 04:39:34 -0500
Subject: [LLVMbugs] [Bug 2615] New: mips cannot return double.
Message-ID: <bug-2615-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2615

           Summary: mips cannot return 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: rich at pennware.com
                CC: llvmbugs at cs.uiuc.edu


The following fails with llc -march mips:

; ModuleID = 't0037.ubc'

define double @main(...) {
entry:
        %retval = alloca double         ; <double*> [#uses=3]
        store double 0.000000e+00, double* %retval
        %r = alloca double              ; <double*> [#uses=1]
        load double* %r         ; <double>:0 [#uses=1]
        store double %0, double* %retval
        br label %return

return:         ; preds = %entry
        load double* %retval            ; <double>:1 [#uses=1]
        ret double %1
}

[~/ellcc/ellcc] main% llc -march mips t0037.ubc
Return operand #0 has unhandled type f64
llc[0x8b5105e]
llc[0x8b51190]
...


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 31 04:59:56 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 31 Jul 2008 04:59:56 -0500
Subject: [LLVMbugs] [Bug 2616] New: Code generation failure for Cell SPU
Message-ID: <bug-2616-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2616

           Summary: Code generation failure for Cell SPU
           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: rich at pennware.com
                CC: llvmbugs at cs.uiuc.edu


The following fails when generating code for the SPU with llc -march cellspu:

; ModuleID = 't0038.ubc'

; ModuleID = 't0038.ubc'

define i32 @main(...) {
entry:
        %retval = alloca i32            ; <i32*> [#uses=2]
        store i32 0, i32* %retval
        %i = alloca i32         ; <i32*> [#uses=0]
        br label %return

return:         ; preds = %entry
        load i32* %retval               ; <i32>:0 [#uses=1]
        ret i32 %0
}

[~/ellcc/ellcc] main% llc -march cellspu t0038.ubc
Attempt to copy unknown/unsupported register class!
llc[0x8b5105e]
...


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 31 05:25:07 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 31 Jul 2008 05:25:07 -0500
Subject: [LLVMbugs] [Bug 2617] New: A code generation error on ia64
Message-ID: <bug-2617-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2617

           Summary: A code generation error on ia64
           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: rich at pennware.com
                CC: llvmbugs at cs.uiuc.edu


The following fails on the ia64 with llc -march ia64:

; ModuleID = 't0039.ubc'

define i32 @main(...) {
entry:
        %retval = alloca i32            ; <i32*> [#uses=2]
        store i32 0, i32* %retval
        %fn = alloca void ()*           ; <void ()**> [#uses=2]
        %f = alloca [1 x void ()*]              ; <[1 x void ()*]*> [#uses=1]
        getelementptr [1 x void ()*]* %f, i32 0, i32 1          ; <void ()**>:0
[#uses=1]
        bitcast void ()** %0 to void ()*                ; <void ()*>:1
[#uses=1]
        store void ()* %1, void ()** %fn
        load void ()** %fn              ; <void ()*>:2 [#uses=1]
        call void %2( )
        br label %return

return:         ; preds = %entry
        load i32* %retval               ; <i32>:3 [#uses=1]
        ret i32 %3
}

[~/ellcc/ellcc] main% llc -f -march ia64 t0039.ubc
llc: /home/rich/llvm-trunk-new/lib/Target/TargetRegisterInfo.cpp:72: const
llvm::TargetRegisterClass*
llvm::TargetRegisterInfo::getPhysicalRegisterRegClass(unsigned int, llvm::MVT)
const: Assertion `false && "Couldn't find the register class"' failed.
llc[0x8b5105e]
...


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 31 10:25:09 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 31 Jul 2008 10:25:09 -0500
Subject: [LLVMbugs] [Bug 2618] New: Dependency files not supported
Message-ID: <bug-2618-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2618

           Summary: Dependency files not supported
           Product: clang
           Version: unspecified
          Platform: PC
        OS/Version: Windows XP
            Status: NEW
          Severity: enhancement
          Priority: P2
         Component: preprocessor
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: krj at rajaratnam.dk
                CC: llvmbugs at cs.uiuc.edu


clang doesn't support creation of dependency files, which makes it hard to use
it as a GCC replacement.

The proposed fix can be found at:

http://lists.cs.uiuc.edu/pipermail/cfe-dev/2008-July/002398.html


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

From bugzilla-daemon at cs.uiuc.edu  Thu Jul 31 11:35:19 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 31 Jul 2008 11:35:19 -0500
Subject: [LLVMbugs] [Bug 2619] New: poor optimization in presence of
	bitcast'ed pointer
Message-ID: <bug-2619-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2619

           Summary: poor optimization in presence of bitcast'ed pointer
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Scalar Optimizations
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: daniel at zuster.org
                CC: llvmbugs at cs.uiuc.edu


Created an attachment (id=1874)
 --> (http://llvm.org/bugs/attachment.cgi?id=1874)
failing testcase

The attached test case (foo.ll) should optimize down to a constant return. I
believe the presence of the bitcast is inhibiting further optimization though. 

A relatively simple fix (I believe) would be to reduce the stores through a
bitcasted pointer to separate stores into the proper type when obvious. 

Alternatively, this code arises from a lowered memset, which could be changed
to lower in a type specific way. The original .ll is also attached.


-- 
Configure bugmail: http://llvm.org/bugs/userprefs.cgi?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 Jul 31 16:04:00 2008
From: bugzilla-daemon at cs.uiuc.edu (bugzilla-daemon at cs.uiuc.edu)
Date: Thu, 31 Jul 2008 16:04:00 -0500
Subject: [LLVMbugs] [Bug 2620] New: Cannot yet select error from the x86
	backend
Message-ID: <bug-2620-206@http.llvm.org/bugs/>

http://llvm.org/bugs/show_bug.cgi?id=2620

           Summary: Cannot yet select error from the x86 backend
           Product: libraries
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: pgurd at rapidmind.com
                CC: llvmbugs at cs.uiuc.edu, sdt at rapidmind.com


Created an attachment (id=1876)
 --> (http://llvm.org/bugs/attachment.cgi?id=1876)
llvm ir file which reproduces the problem

The attached program produces the message

Cannot yet select: 0x8e1e094: v2i64 = X86ISD::CMPPD 0x8e1df84, 0x8e1df84,
0x8e1e00c

with -mcpu=core2


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