On Wed, Dec 18, 2013 at 11:21:52PM +0100, Steven Bosscher wrote:
> On Wednesday, December 18, 2013, Jakub Jelinek wrote:
> > As discussed in the PR, this testcase ICEs on arm, because ifcvt
> > is relying on active instruction counts from various routines
> > (count_bb_insns, flow_find_cross_jump and flow_find_head_matching_sequence),
> > but each of those routines have different view of what counts as
> > active insns.
> >
> > Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux
> > and tested on the testcase using cross to arm. Ok for trunk?
> >
> > 2013-12-18 Jakub Jelinek <>
> > PR rtl-optimization/58668
> > * cfgcleanup.c (flow_find_cross_jump): Don't count
> > any jumps if dir_p is NULL. Remove p1 variable and make USE/CLOBBER
> > check consistent with other places.
> > (flow_find_head_matching_sequence): Don't count USE or CLOBBER
> > insns.
> > (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
> > counting change.
> > * ifcvt.c (count_bb_insns): Don't count USE or CLOBBER insns.
>
> Why not use active_insn_p instead of hand-checks for USE and CLOBBER insns?
Because it brings in the JUMP_TABLE_DATA mess into the picture? Some of the
places already guard it with INSN_P and similar checks and do more things
than just counting the insns under those conditionals, so in other places
one can't just use say prev_active_insn or next_active_insn anyway, the insn
type has to be checked then and thus active_insn_p would become just a fancy
way of checking for USE/CLOBBER.
Jakub
