Messages in this thread | | | From | Dexuan Cui <> | Subject | RE: [PATCH net-next 2/3] vsock: fix vsock_dequeue/enqueue_accept race | Date | Fri, 18 Aug 2017 21:13:05 +0000 |
| |
> From: Stefan Hajnoczi [mailto:stefanha@redhat.com] > Sent: Thursday, August 17, 2017 07:06 > > On Tue, Aug 15, 2017 at 10:15:39PM +0000, Dexuan Cui wrote: > > With the current code, when vsock_dequeue_accept() is removing a sock > > from the list, nothing prevents vsock_enqueue_accept() from adding a new > > sock into the list concurrently. We should add a lock to protect the list. > > The listener sock is locked, preventing concurrent modification. I have > checked both the virtio and vmci transports. Can you post an example > where the listener sock isn't locked? > > Stefan Sorry, I was not careful when checking the vmci code. Please ignore the patch.
Now I realized the expectation is that the individual transport drivers should do the locking for vsock_enqueue_accept(), but for vsock_dequeue_accept(), the locking is done by the common vsock driver.
Thanks, -- Dexuan
| |