https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119142
--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Sure, but it would be good to track somewhere progress on adjusting the patch
so that the patch can be committed again.
But guess we can do that even on a closed PR.
Note, I wonder why some of the ICEs were in -Os compiled code, shouldn't that
be
caught by
if (optimize_function_for_size_p (cfun))
return 1;
in the patch already? Though, looking at the x86_64 FAILs, I see all those
tests have optimize attribute (though, I don't understand why then the normal
generic tuning or so rules don't kick in and make the memory cost higher).
