On Mon 2025-03-17 21:31:13, Mikael Morin wrote:
> Le 28/02/2025 à 17:01, Filip Kastl a écrit :
> > diff --git a/gcc/gimple-ssa-sccopy.cc b/gcc/gimple-ssa-sccopy.cc
> > index 9f25fbaff36..7ffb5718ab6 100644
> > --- a/gcc/gimple-ssa-sccopy.cc
> > +++ b/gcc/gimple-ssa-sccopy.cc
> > @@ -568,6 +568,19 @@ scc_copy_prop::propagate ()
> > {
> > vec<gimple *> scc = worklist.pop ();
> > + /* When we do 'replace_scc_by_value' it may happen that some EH edges
> > + get removed. That means parts of CFG get removed. Those may
> > + contain copy statements. For that reason we prune SCCs here. */
> > + unsigned i;
> > + for (i = 0; i < scc.length (); i++)
> > + if (gimple_bb (scc[i]) == NULL)
> > + scc.unordered_remove (i);
> Hello,
> this may not be important, but don't you need to skip the increment of i if
> the item is removed?
> for (i = 0; i < scc.length ();)
> if (gimple_bb (scc[i]) == NULL)
> scc.unordered_remove (i);
> else
> i++;
>
> Mikael
Hi Mikael,
Yes! You're right and this is something I overlooked. Thanks for the
correction. I'll create a patch and mention you as Reported-By:
Filip
