https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121382
Tamar Christina <tnfchris at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |rguenth at gcc dot gnu.org
--- Comment #7 from Tamar Christina <tnfchris at gcc dot gnu.org> ---
Yeah I think the simplification shouldn't be allowed if any wraparound occurs
during the constant folding.
So in this case the -9 shouldn't have been folded.
Rejecting the simplification on wraparound should still keep the optimization
but prevent these wrong codegen where the signed IV no longer computes the same
final bit pattern as the unsigned.
Agreed Richi? If so I can modify tree_to_aff_combination with an option to
reject
such folding.
