8 Apr
2005
8 Apr
'05
6:42 p.m.
Alexander Terekhov wrote:
Alexander Terekhov wrote:
Peter Dimov wrote: [...]
I don't know, but all of Apple's code specifically clears the reservation. ;-)
It doesn't make (some) Apple's code less broken (msync wise) though. ;-)
Note also that
http://sourceware.org/cgi-bin/cvsweb.cgi/libc/sysdeps/powerpc/bits/atomic.h?...
doesn't seem to exhibit that paranoia (and supposedly runs just fine on the same hardware). I suggest that you uncommit the "fix".
It's good enough for 1.33. A failed lock() is still expensive because I haven't fixed the upper layers to not throw bad_weak_ptr. It might be good enough anyway... it's only two more instructions, after all. stwcx. is uncontended on zero. The paranoia is cheap in this specific case. :-)