Re: wait_on_page_bit_common(TASK_KILLABLE, EXCLUSIVE) can miss wakeup?

[Oleg Nesterov]

On 06/30, Nicholas Piggin wrote:
>
> My patch is what actually introduced this ugly
> bit test, but do we even need it at all? If we do then it's
> under-commented, I can't see it wouldn't be racy though. Can we just
> get rid of it entirely?

But then we will need to move io_schedule() down, after test_and_set_bit().
And we will have the same problem with task->state != RUNNING. Plus more
complications with "behavior == DROP".

Oleg.