On Sun, Aug 27, 2017 at 2:40 PM, Linus Torvalds <torva...@linux-foundation.org> wrote: > > End result: page is unlocked, CPU3 is waiting, nothing will wake CPU3 up.
Not CPU3. CPU3 was the waker. It's thread 2 that is waiting and never got woken up, of course. Other than that, the scenario still looks real to me. Ideas? Linus