On Fri, 19 Sep 2025, Mikael Morin wrote:
> Le 18/09/2025 à 09:43, Richard Biener a écrit :
> > diff --git a/gcc/tree-ssa-pre.cc b/gcc/tree-ssa-pre.cc
> > index 99331730bc2..18b36259cb4 100644
> > --- a/gcc/tree-ssa-pre.cc
> > +++ b/gcc/tree-ssa-pre.cc
> > @@ -2118,6 +2121,20 @@ compute_antic_aux (basic_block block, bool
> > block_has_abnormal_pred_edge)
> > first = e;
> > else if (BB_VISITED (e->dest))
> > worklist.quick_push (e);
> > + else if (0 && !(e->flags & EDGE_DFS_BACK))
> This (dead) branch should probably be removed, was it really your intent to
> add it?
Whoops, no. That was an earlier attempt that somehow I failed to
cleanup.
Thanks for noticing.
Richard.
> > + {
> > + /* When our reverse iteration order does not match up with
> > + a forward DFS which can in happen with unfortunate
> > + choices of fake edges to exits from infinite loops, we
> > + have to avoid intermangling two ANTIC iterations by
> > + using ANTIC_IN computed in the previous iteration.
> > + As we cannot easily do this stabilize the iteration by
> > + not allowing a MAX set on such edge initially. */
> > + if (dump_file && (dump_flags & TDF_DETAILS))
> > + fprintf (dump_file, "ANTIC_IN is uncomputed on non-DFS_BACK "
> > + "%d->%d\n", e->src->index, e->dest->index);
> > + worklist.quick_push (e);
> > + }
> > else
> > {
> > /* Unvisited successors get their ANTIC_IN replaced by the
>
>
--
Richard Biener <[email protected]>
SUSE Software Solutions Germany GmbH,
Frankenstrasse 146, 90461 Nuernberg, Germany;
GF: Ivo Totev, Andrew McDonald, Werner Knoblich; (HRB 36809, AG Nuernberg)
