This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] cfg_remove_useless_stmts
- From: law at redhat dot com
- To: Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>
- Cc: Jan Hubicka <hubicka at ucw dot cz>, gcc-patches at gcc dot gnu dot org
- Date: Thu, 13 Nov 2003 14:56:29 -0700
- Subject: Re: [tree-ssa] cfg_remove_useless_stmts
- Reply-to: law at redhat dot com
In message <20031113211818.GA13739@atrey.karlin.mff.cuni.cz>, Zdenek Dvorak wri
tes:
>Hello,
>
>> >> I see... no direct ideas how to deal it more generally then.
>> >> > > THis can be executed only for COND_EXPRs where both arms get nullif
>ied,
>> >> > > right? This should be cared of by cfg_cleanup unless something is w
>rong
>> >> > > with it.
>> >> >
>> >> > Yes it could; no it is not, since cfg_cleanup is not run after un-ssa
>.
>> >>
>> >> It would be nice to unify these two somehow. I will try to look into i
>t
>> >> later. Some of the issues should be dealt with if we get rid of GOTOs
>> >> completely that should not take that long anyway.
>> >
>> >indeed; with the patch to do it I am currently testing all the goto
>> >removal stuff goes away.
>> I'd *strongly* recommend you examine the .optimized and .rtl dumps if
>> you go down this path. The goto removal stuff is still eliminating
>> a lot of crap. If you pull it out, I strongly suspect you'll find that
>> you're suddenly generating a lot more jumps when you drop down from
>> tree to rtl form.
>
>why? With my patch, I do not generate any unnecesary gotos at all in
>the first place.
Have you looked at the RTL dumps before and after your new patch? Again,
I'm going to require you to give us some info based on analyzing the
RTL dumps before I will serious look at a patch which removes the
GOTO_EXPR handling from cfg_remove_useless_stmts.
jeff