[PATCH] x86: Optimize atomic_compare_and_exchange_[val|bool]_acq [BZ #28537]
Florian Weimer
fweimer@redhat.com
Thu Nov 4 10:15:39 GMT 2021
* H. J. Lu:
>> Shouldn't GCC be fixed to generate the appropriate instruction sequence
>> for this architecture? Should we perhaps switch to compiler atomics for
>> x86 instead of defining our own?
>
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103065
>
> In the meantime, we should help older compilers.
But wouldn't that extra read persist with fixed compilers, due to the
memory barrier? So that there are two reads?
Maybe we need a hint flag for __atomic_compare_exchange_n, encoded in
the memory order?
Thanks,
Florian
More information about the Libc-alpha
mailing list