This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Refine cross-jumping
- From: Jan Hubicka <jh at suse dot cz>
- To: Richard Kenner <kenner at vlsi1 dot ultra dot nyu dot edu>
- Cc: jh at suse dot cz, gcc-patches at gcc dot gnu dot org
- Date: Sat, 19 Apr 2003 19:50:27 +0200
- Subject: Re: Refine cross-jumping
- References: <10304191446.AA22380@vlsi1.ultra.nyu.edu>
> How much of opurtunities we miss this way? Perhaps we can simply merge
> the blocks conservativly (ie remove the clobbers and keep uses)
>
> I didn't measure, but I doubt it misses very many. After all, each
> such "miss" would have caused this bug and it didn't occur in very many places.
> I don't think it's worth the complexity.
It used to make important difference (about 20% of crossjumping). I
know that because my original crossjumping rewrite didn't contained that
and I had to put the hack in from old code in order to make results
comparable.
We reduced the amount of clobbers/uses in the code since that but still
it may be important. It usually gets things right (modulo confusing
liveness analysis). Only case I can think of where it produces invalid
code is when use gets lost.
Honza