https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118019
Jeffrey A. Law <law at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |law at gcc dot gnu.org
--- Comment #13 from Jeffrey A. Law <law at gcc dot gnu.org> ---
Just a note.
The cost change does trigger a regression on rv32gcv on the
Wstringop-overflow-47 test. The underlying analysis code for that test is
known to be sensitive to vector ops -- occasionally resulting in false positive
warnings or missed warnings depending on whether or not code got vectorized.
I don't think this means we should revert the change, just to raise awareness
that there may be reports of false positives from the stringop-overflow
warnings as a result of differences in code generation.
