This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Fix 34826 (branch probability is not updated with do-loop optimization)
- From: Revital1 Eres <ERES at il dot ibm dot com>
- To: Jan Hubicka <hubicka at ucw dot cz>
- Cc: Ayal Zaks <ZAKS at il dot ibm dot com>, gcc-patches at gcc dot gnu dot org, Mircea Namolaru <NAMOLARU at il dot ibm dot com>, rakdver at kam dot mff dot cuni dot cz
- Date: Thu, 17 Jan 2008 15:42:08 +0200
- Subject: Re: [PATCH] Fix 34826 (branch probability is not updated with do-loop optimization)
Hello,
gcc-patches-owner@gcc.gnu.org wrote on 17/01/2008 14:49:48:
> Hi,
> the patch looks fine, however
> > + /* Update the REG_BR_PROB note. */
> > + if (true_prob_val)
>
> In what cases you can end up with REG_BR_PROB being NULL here?
I did not find such cases so far, running gcc testsuite, but I am not
sure it does not exist.
Thanks,
Revital
>
> Honza
> > + {
> > + /* Seems safer to use the branch probability. */
> > + REG_NOTES (jump_insn) =
> > + gen_rtx_EXPR_LIST (REG_BR_PROB,
> > + GEN_INT (desc->in_edge->probability),
> > + REG_NOTES (jump_insn));
> > + }
> > }
> >
> > /* Process loop described by LOOP validating that the loop is suitable
for
>