[llvm-commits] [llvm] r54519 - in /llvm/trunk: include/llvm/Target/DarwinTargetAsmInfo.h include/llvm/Target/ELFTargetAsmInfo.h include/llvm/Target/TargetAsmInfo.h lib/Target/ARM/ARMAsmPrinter.cpp lib/Target/DarwinTargetAsmInfo.cpp lib/Target/ELFTargetAsmInfo.cpp lib/Target/Mips/MipsTargetAsmInfo.cpp lib/Target/Mips/MipsTargetAsmInfo.h lib/Target/TargetAsmInfo.cpp test/CodeGen/ARM/2008-08-07-AsmPrintBug.ll

[Evan Cheng]

Yeah, I realized I over-thought the problem after I went to bed:

struct x {
   int a;
   int b;
} X  __attribute__((weak)) = { 1, 2 };

void set(int b) {
   X.b = b;
}

All the extra logic is not needed since the PIC relative expression is  
using GV stub. Feel free to undo my patch (except the ARMAsmPrinter  
part the choose lcomm instead of comm). Or I will get to it later today.

Evan

On Aug 8, 2008, at 3:03 AM, Anton Korobeynikov wrote:

> Hi, Evan
>
>> +    virtual const Section* SelectSectionForGlobal(const  
>> GlobalValue *GV,
>> +                                                 bool NoCoalesce =  
>> false) const;
> This approach seems to be pretty hackish and definitely cannot be
> solution here, since it adds target-specific stuff (coalesced  
> sections)
> into generic interface. Could you please provide a C testcase, I'll  
> try
> to see how mainline gcc (from iPhone SDK) deals with this problem and
> fix the stuff properly.
>
> -- 
> With best regards, Anton Korobeynikov.
>
> Faculty of Mathematics & Mechanics, Saint Petersburg State University.
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits