Eric Wong [off-list ref] wrote:

Linus Torvalds [off-list ref] wrote:

quoted

Please document the barrier that this mb() pairs with, and then give
an explanation for the fix in the commit message, and I'll happily
take it. Even if it's just duplicating the comments above the
wq_has_sleeper() function, except modified for the ep_modify() case.

Hopefully my explanation is correct and makes sense below,
I think both effects of the barrier are needed

I noticed Linus accepted this already.  This should probably go to
stable, right?

From ancient git history[1], it seems this bug exists for all

2.6 kernels:

	commit 424980a87e226d63af46579b2af16ec1b8d17e52
	Author: Davide Libenzi [off-list ref]
	Date:   Thu Nov 14 16:17:23 2002 -0800

	    [PATCH] epoll bits 0.46 ...

[1] - git://git.kernel.org/pub/scm/linux/kernel/git/tglx/history.git