https://bugs.winehq.org/show_bug.cgi?id=46208
--- Comment #7 from Zebediah Figura z.figura12@gmail.com --- (In reply to Brendan McGrath from comment #6)
That certainly explains what we see and why the problem doesn't occur in the futex implementation.
I guess the current implementation needs a fix regardless (given kernel futex support won't always be available).
Yes, definitely. I'm not entirely sure how, though. I'll have to give it some thought.
Possibly we can implement it on top of RtlWaitOnAddress(), which deserves a futex path anyway.