This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: What is this patch doing?
- To: Richard Henderson <rth at cygnus dot com>
- Subject: Re: What is this patch doing?
- From: Joern Rennecke <amylaar at cygnus dot co dot uk>
- Date: Tue, 1 Aug 2000 23:03:19 +0100 (BST)
- CC: Joern Rennecke <amylaar at pasanda dot cygnus dot co dot uk>, Bernd Schmidt <bernds at pasanda dot cygnus dot co dot uk>, gcc at gcc dot gnu dot org
> But iirc this splitter, and the branchy widgetry it replaced was an
> optimization was it not? There was a simple jump around sequence
> when it couldn't figure out what to do. Can you just output that
> (without splitting) at -O0?
Well, I could, but there are actually six patterns affected overall.
this means that the output pattern would be repeated in a non-splitted
form for each of these patterns.
And the splitters would all have to have a predicate
"reload_completed && optimize_size" .
Maybe we should rather fix the problem by changing split_all_insns so that
it calls rebuild_jump_labels when needed.
Or change try_split to call rebuild_jump_labels on the result - that
would even cover the case where final splits an insn just before outputting.