https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84907
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords| |wrong-code
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
>is actually well defined unaligned load (if aligned to __alignof__(float), but
>not __alignof__(__vector float)).
I don't think it is a well defined unaligned load. I bet if you use
-fsantized=undefined you will get an error about it.
Also on older powerpc the vector load always ignored the lower (unaligned)
bits; I suspect it is still using that instruction here.
