linux-kernel - Re: [PATCH 1/2] epoll: use the waitqueue lock to protect ep->wq

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 14 Dec 2017 14:05:28 +0100
From:   Christoph Hellwig <hch@....de>
To:     Jason Baron <jbaron@...mai.com>
Cc:     Christoph Hellwig <hch@....de>, Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Al Viro <viro@...iv.linux.org.uk>,
        linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/2] epoll: use the waitqueue lock to protect ep->wq

On Thu, Dec 07, 2017 at 11:09:11AM -0500, Jason Baron wrote:
> On 12/06/2017 06:52 PM, Christoph Hellwig wrote:
> > The eoll code currently always uses the unlocked waitqueue helpers for
> > ep->wq, but instead of holding the lock inside the waitqueue around these
> > calls, as expected by the epoll code uses its own lock.  Given that the
> > waitqueue is not exposed to the rest of the kernel this actually works
> > ok at the moment, but prevents the epoll locking rules from being
> > enforced using lockdep.  Remove ep->lock and use the waitqueue lock
> > to not only reduce the size of struct eventpoll but also make sure we
> > can assert locking invariations in the waitqueue code.
> > 
> > Signed-off-by: Christoph Hellwig <hch@....de>
> 
> Probably should also fix the locking comments at the top of
> fs/eventpoll.c that refer to ep->lock...

Done.  Note that while doing this I noticed that the epoll code
seems to have sketchy workarounds for the fact that it abused ep->poll
as the waitqueue lock that might be able to be removed now.
But I don't really dare to touch the guts of this code.