https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94440
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
Assignee|unassigned at gcc dot gnu.org |jakub at gcc dot gnu.org
--- Comment #15 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 49699
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49699&action=edit
gcc11-pr94440.patch
I think best would be to get away with the TREE_OPTIMIZATION_NODE vs.
TARGET_OPTIMIZATION_NODE dichotomy and just use one node for both, but that
won't fly in stage3, so this patch essentially forces new target optimization
nodes to cover even the flag_excess_precision and flag_unsafe_math_optimization
stuff, so that the enabled attr etc. can use those.
