https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121148
Jonathan Wakely <redi at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |joel at gcc dot gnu.org
--- Comment #6 from Jonathan Wakely <redi at gcc dot gnu.org> ---
If the CC'd target maintainers could confirm my understanding of their area
above, I would be very grateful. I don't think I need any of you to change the
code, just let me know if the assembly is free of undefined behaviour on
integer overflow.
I think the only changes needed are to the implementations that use plain C++
arithmetic, and I reckon I can figure out how to fix those myself!
It might also be worth considering whether we want to keep these config/cpu/*
definitions, or just use the generic version defined in terms of
__atomic_fetch_add. If the compiler already generates good code for
__atomic_fetch_add on signed int, should we remove these cpu-specific files?
