Messages in this thread | | | From | Rainer Weikusat <> | Subject | Re: use-after-free in sock_wake_async | Date | Wed, 25 Nov 2015 20:57:59 +0000 |
| |
Eric Dumazet <eric.dumazet@gmail.com> writes: > On Wed, 2015-11-25 at 11:50 -0800, Eric Dumazet wrote: > >> > other->sk_data_ready(other); >> > + unix_state_unlock(other); > > > Also, problem with such construct is that we wakeup a thread that will > block on the lock we hold. > > Beauty of sk_data_ready() is to call it once we hold no lock any more, > to enable another cpu to immediately proceed. > > In this case, 'other' can not disappear, so it should be safe.
I do agree that keeping the ->sk_data_ready outside of the lock will very likely have performance advantages. That's just something I wouldn't have undertaken because I'd be reluctant to make a fairly complicated change to a lot of code in order to improve performance unless performance was actually found to be lacking and because it would step onto to many different people's turf.
| |