Larry Rosenman <ler@lerctr.org> writes:
>> We've called that routine s_lock for a *long* time, so it seems
>> like there must be some factor involved that I don't see just yet...
> Didn't your commit message say something about the TAS and NON-TAS
> paths being the same now?
Yeah, but don't tell me you were running on a non-TAS platform...
that stuff didn't work...
regards, tom lane