This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: sh casesi patch
- To: rth at cygnus dot com (Richard Henderson)
- Subject: Re: sh casesi patch
- From: Joern Rennecke <amylaar at cygnus dot co dot uk>
- Date: Tue, 9 Mar 1999 23:19:39 +0000 (GMT)
- Cc: MoritaT at sega dot com, egcs-patches at egcs dot cygnus dot com
> The braf code label is not emitted until it is needed, so there's
> no chance it will get optimized away. This obviates the need for
> the dummy_jump hack, which is what was causing flow indigestion.
>
> It's somewhat less than clear to me exactly what fixup_addr_diff_vecs
> is doing. It's quite possible the two fixup functions could be merged.
>
>
> r~
>
>
> * sh.c (fixup_braf_labels): New function.
> (machine_dependent_reorg): Call it.
This is likely to confuse reorg, since you don't update the basic block
structure.
AFAICR I actually had fixup_addr_diff_vecs emitting the labels initially,
but abandoned it because of this problem.