This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Fix profiling of fork()
- From: Jan Hubicka <jh at suse dot cz>
- To: Andrew Pinski <pinskia at physics dot uc dot edu>
- Cc: gcc-patches at gcc dot gnu dot org, Jan Hubicka <hubicka at ucw dot cz>
- Date: Thu, 17 Apr 2003 09:14:13 +0200
- Subject: Re: Fix profiling of fork()
- References: <8F2F6748-7085-11D7-9950-000393A6D2F2@physics.uc.edu>
> Since I deleted the original message:
> <http://gcc.gnu.org/ml/gcc-patches/2003-04/msg00978.html>.
>
> I found this patch very useful in fact, I thought it would just work
> before I tried it out a couple of days before with a program i was
It is supposed to just work, but it is not very well tested - we do use
it regulary only for the SPEC benchmarks. I recently did experiments
with some other applications and they are pretty good, but I commonly
hit problems. Lets hope it will be used more commonly in the future. I
did simple experiments and I was able to speedup ./configure script
execution (from the config.cache) by 10-12% via building profiled bash
and sed so it appears to be quite easy.
> creating which used fork and continued to run so I got an error from
> gcc:
> fire.c:314: error: corrupted profile info: prob for 18--2 thought to be
> -10000
> fire.c:314: error: corrupted profile info: prob for 18-19 thought to be
> 20000
>
> Then I saw this patch, I was saved. I would like this patch to go in
> also.
Hope that Richard will eventually catch up :)
Let me know about additional problems (or sucesses :) you have with the
profiling code...
Honza
>
>
> Thanks,
> Andrew Pinski