On Wed, Aug 5, 2009 at 7:32 PM, Daniel Dunbar<daniel at zuster.org>
wrote:> On Tue, Jun 30, 2009 at 6:11 AM, Sylvere Teissier<st at invia.fr>
wrote:
>>> 2. The backend (llc) converts that to Dwarf (or something else as
>>> desired by that back-end).
>>>
>> How advanced is LLVM Dwarf generation compared to GCC ?
>
> Devang can probably give a better answer, but my understanding is:
>
> 1. At -O0, the quality is very good, with a few exceptions (like
> inlined functions). With clang I believe there are cases where we are
> more accurate than gcc.
llvm-gcc is not very good at keeping track of lexical scopes in debug
info. (This needs some help from someone who can grok GIMPLE trees
well). However, clang handles it well.
-
Devang
> 2. At -O1+ we only have line number information, which is approximate
> at best. This is substantially worse than gcc (although plenty of
> caveats about debugging optimized code apply).
> - Daniel
>
>>
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>
>
> _______________________________________________
> LLVM Developers mailing list
> LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>
--
-
Devang