This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: atomic ops for powerpc
>>>>> Geoff Keating writes:
>> One could test at configure time if lwsync mnemonic is recognized
>> and emit a "sync" instruction if it is not.
Geoff> Why would that be better than using a .long?
No reason. I was being too conservative.
Geoff> You can't generate and emit labels, because the patterns must be
Geoff> atomic. I will try the $-16 stuff. (I'm not too enthusiastic about
Geoff> it, though; it's one thing when it's $+4, it's another when you're
Geoff> skipping around in 7-instruction patterns.)
I guess I'm missing something. Why couldn't one generate labels
and then use %l to emit them in the output template? I appreciate your
trying $-16 as well.
Thanks, David