On Wed, Nov 15, 2017 at 04:57:26PM +0800, Xin Long wrote:
> Commit dfcb9f4f99f1 ("sctp: deny peeloff operation on asocs with threads
> sleeping on it") fixed the race between peeloff and wait sndbuf by
> checking waitqueue_active(&asoc->wait) in sctp_do_peeloff().
>
> But it actually doesn't work, as even if waitqueue_active returns false
> the waiting sndbuf thread may still not yet hold sk lock. After asoc is
> peeled off, sk is not asoc->base.sk any more, then to hold the old sk
> lock couldn't make assoc safe to access.
>
> This patch is to fix this by changing to hold the new sk lock if sk is
> not asoc->base.sk, meanwhile, also set the sk in sctp_sendmsg with the
> new sk.
>
> With this fix, there is no more race between peeloff and waitbuf, the
> check 'waitqueue_active' in sctp_do_peeloff can be removed.
>
> Thanks Marcelo and Neil for making this clear.
>
> v1->v2:
> fix it by changing to lock the new sock instead of adding a flag in asoc.
>
> Suggested-by: Neil Horman <nhor...@tuxdriver.com>
> Signed-off-by: Xin Long <lucien....@gmail.com>
Acked-by: Neil Horman <nhor...@tuxdriver.com>
