https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125597
Tamar Christina <tnfchris at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Summary|ICE: SIGSEGV in |[16/17 Regression] ICE:
|analyze_scalar_evolution_1 |SIGSEGV in
|(tree-scalar-evolution.cc:2 |analyze_scalar_evolution_1
|026) with -O3 |(tree-scalar-evolution.cc:2
|-mcpu=c1-ultra |026) with -O3
|-mautovec-preference=sve-on |-mcpu=c1-ultra
|ly |-mautovec-preference=sve-on
| |ly
Component|target |tree-optimization
Assignee|unassigned at gcc dot gnu.org |tnfchris at gcc dot
gnu.org
--- Comment #4 from Tamar Christina <tnfchris at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #3)
>
> > > 3955 LOOP_VINFO_NITERSM1 (epilogue_vinfo)
> > > 3956 = fold_build2 (MINUS_EXPR, TREE_TYPE (epilogue_niters),
> > > 3957 epilogue_niters,
> > > 3958 build_one_cst (TREE_TYPE
> > > (epilogue_niters)));
> > >
> > > I suspect we need to resort to a more "temporarily VARYING" way here :/
> > > ISTR suggesting some internal function, like insert
> > >
> > > _38 = .VARYING ();
> > >
> > > it would be a NOVOPS (but not CONST/PURE) one maybe.
> >
> > Ok, so the suggestion is to insert a dummy value until we replace it later
> > on just to make df analysis ignore it.
> > Would using a ssa_default_def work here too? or do you prefer the IFN?
>
> A default def would be semantically uninitialized, so also wrong (it would
> solve this ICE of course). There's no "has unknown value" but using
> a volatile load, but I think a new IFN would be better here.
Ack. Mine then.
