Re: Our "fallback" atomics implementation doesn't actually work

Andres Freund <andres(at)anarazel(dot)de> writes:
> On 2016-10-05 14:01:05 -0400, Tom Lane wrote:
>> I think what is happening is that there are circular assumptions that end
>> up trying to implement a spinlock in terms of a spinlock, or otherwise
>> somehow recursively use the process's semaphore. It's a bit hard to tell
>> though because the atomics code is such an underdocumented rat's nest of
>> #ifdefs.

> I don't think that should be the case, but I'll look into it. How long
> did it take for you to reproduce the issue?

It hangs up within 10 or 20 seconds for me. I didn't try hard to get a
census of where, but at least some of the callers are trying to acquire
buffer partition locks.

regards, tom lane