[llvm-commits] [llvm] r146468 - in /llvm/trunk: lib/MC/ELFObjectWriter.cpp test/CodeGen/Mips/DO-relsym.ll

Bruno Cardoso Lopes bruno.cardoso at gmail.com
Mon Dec 12 19:59:48 PST 2011


Thanks Akira,

On Tue, Dec 13, 2011 at 1:54 AM, Hatanaka, Akira <ahatanaka at mips.com> wrote:
> r146470.
> ________________________________________
> From: Bruno Cardoso Lopes [bruno.cardoso at gmail.com]
> Sent: Monday, December 12, 2011 6:53 PM
> To: Hatanaka, Akira; Carter, Jack
> Cc: llvm-commits at cs.uiuc.edu
> Subject: Re: [llvm-commits] [llvm] r146468 - in /llvm/trunk: lib/MC/ELFObjectWriter.cpp test/CodeGen/Mips/DO-relsym.ll
>
> On Tue, Dec 13, 2011 at 12:27 AM, Akira Hatanaka <ahatanaka at mips.com> wrote:
>> Author: ahatanak
>> Date: Mon Dec 12 20:27:40 2011
>> New Revision: 146468
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=146468&view=rev
>> Log:
>> Relocation against a symbol, instead of against section. We had some extreme
>> test cases where there were a lot of relocations applied relative to a large
>> rodata section. Gas would create a symbol for each of these whereas we would
>> be relative to the beginning of the rodata section. This change mimics what
>> gas does.
>>
>> Patch by Jack Carter.
>>
>> Added:
>>    llvm/trunk/test/CodeGen/Mips/DO-relsym.ll
>> Modified:
>>    llvm/trunk/lib/MC/ELFObjectWriter.cpp
>>
>> Modified: llvm/trunk/lib/MC/ELFObjectWriter.cpp
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/MC/ELFObjectWriter.cpp?rev=146468&r1=146467&r2=146468&view=diff
>> ==============================================================================
>> --- llvm/trunk/lib/MC/ELFObjectWriter.cpp (original)
>> +++ llvm/trunk/lib/MC/ELFObjectWriter.cpp Mon Dec 12 20:27:40 2011
>> @@ -1851,7 +1851,8 @@
>>   assert(Target.getSymA() && "SymA cannot be 0.");
>>   const MCSymbol &Sym = Target.getSymA()->getSymbol();
>>
>> -  if (Sym.getSection().getKind().isMergeable1ByteCString())
>> +  if (Sym.getSection().getKind().isMergeableCString() ||
>> +      Sym.getSection().getKind().isMergeableConst())
>>     return &Sym;
>>
>>   return NULL;
>> @@ -1914,4 +1915,3 @@
>>
>>   return Type;
>>  }
>> -
>>
>> Added: llvm/trunk/test/CodeGen/Mips/DO-relsym.ll
>> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/Mips/DO-relsym.ll?rev=146468&view=auto
>> ==============================================================================
>> --- llvm/trunk/test/CodeGen/Mips/DO-relsym.ll (added)
>> +++ llvm/trunk/test/CodeGen/Mips/DO-relsym.ll Mon Dec 12 20:27:40 2011
>> @@ -0,0 +1,29 @@
>> +; RUN: llc -filetype=obj -mtriple mipsel-unknown-linux %s -o - | elf-dump --dump-section-data  | FileCheck %s
>> +
>> +; Check that the appropriate symbols were created.
>> +
>> +; CHECK: (('st_name', 0x{{[0-9|a-f]+}}) # '$.str'
>> +; CHECK: (('st_name', 0x{{[0-9|a-f]+}}) # '$.str1'
>> +; CHECK: (('st_name', 0x{{[0-9|a-f]+}}) # '$CPI0_0'
>> +; CHECK: (('st_name', 0x{{[0-9|a-f]+}}) # '$CPI0_1'
>> +
>> + at .str = private unnamed_addr constant [6 x i8] c"abcde\00", align 1
>> + at gc1 = external global i8*
>> + at .str1 = private unnamed_addr constant [5 x i8] c"fghi\00", align 1
>> + at gc2 = external global i8*
>> + at gd1 = external global double
>> + at gd2 = external global double
>> +
>> +define void @foo1() nounwind {
>> +entry:
>> +  store i8* getelementptr inbounds ([6 x i8]* @.str, i32 0, i32 0), i8** @gc1, align 4
>> +  store i8* getelementptr inbounds ([5 x i8]* @.str1, i32 0, i32 0), i8** @gc2, align 4
>> +  %0 = load double* @gd1, align 8
>> +  %add = fadd double %0, 2.500000e+00
>> +  store double %add, double* @gd1, align 8
>> +  %1 = load double* @gd2, align 8
>> +  %add1 = fadd double %1, 4.500000e+00
>> +  store double %add1, double* @gd2, align 8
>> +  ret void
>> +}
>> +
>
> Although ARM has some MC tests under CodeGen/ARM, I think it would be
> better to create a new directory: test/MC/Mips and place this test
> there (which is going to be the same dir to have future mips asm
> parsing tests). Also, rename it to "elf-relsym.ll".
>
>
> --
> Bruno Cardoso Lopes
> http://www.brunocardoso.cc



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




More information about the llvm-commits mailing list