On Fri, Mar 15, 2019 at 01:25:57PM +0300, Andrey Belevantsev wrote:
> As explained in the PR trail, we incorrectly update the availability sets
> in the rare case of several successors and one of them having another
> fence. Fixed as follows. Ok for trunk?
>
> Best,
> Andrey
>
> 2019-03-15 Andrey Belevantsev <a...@ispras.ru>
>
> PR middle-end/89676
> * sel-sched.c (compute_av_set_at_bb_end): When we have an ineligible
> successor,
> use NULL as its av set.
Just formatting nits, will defer actual review to some scheduler maintainer.
Too long line in ChangeLog above, "successor," should be on the next line
already together with the rest of the description.
> diff --git a/gcc/sel-sched.c b/gcc/sel-sched.c
> index 315f2c0c0ab..2053694b196 100644
> --- a/gcc/sel-sched.c
> +++ b/gcc/sel-sched.c
> @@ -2820,10 +2820,12 @@ compute_av_set_at_bb_end (insn_t insn, ilist_t p, int
> ws)
> FOR_EACH_VEC_ELT (sinfo->succs_ok, is, succ)
> {
> basic_block succ_bb = BLOCK_FOR_INSN (succ);
> + av_set_t av_succ = (is_ineligible_successor (succ, p)
> + ? NULL
> + : BB_AV_SET (succ_bb));
>
> gcc_assert (BB_LV_SET_VALID_P (succ_bb));
> - mark_unavailable_targets (av1, BB_AV_SET (succ_bb),
> - BB_LV_SET (succ_bb));
> + mark_unavailable_targets (av1, av_succ, BB_LV_SET (succ_bb));
The above line should use tab instead of 8 spaces.
Jakub
