On 2015-07-30 17:36:52 +0300, Heikki Linnakangas wrote:
> In 9.4, LWLockAcquire holds the spinlock when it marks the lock as held,
> until it has updated the variable. And LWLockWaitForVar() holds the spinlock
> when it checks that the lock is held and that the variable's value matches.
> So it cannot happen on 9.4.
The first paragraph talked about "the same value", but that was just
referring to it not yet having been cleared i think...
> To reiterate, with 9.5, it's possible that a backend is sleeping in
> LWLockWaitForVar(oldvar=123), even though the lock is currently held by
> another backend with value 124. That seems wrong, or surprising at the very
> least.
With my patch that can't really happen that way though? The value is
re-checked after queuing. If it has changed by then we're done. And if
it hasn't yet changed we're guaranteed to be woken up once it's being
changed?
I generaly don't mind adding some sort of flag clearing or such, but I'd
rather not have it in the retry loop in the general LWLockAttemptLock
path - I found that very small amounts of instructions in there have a
measurable impact.
Greetings,
Andres Freund