vhost_net: clear msg.control for non-zerocopy case during tx
authorJason Wang <jasowang@redhat.com>
Wed, 5 Jun 2013 07:40:46 +0000 (15:40 +0800)
committerDavid S. Miller <davem@davemloft.net>
Mon, 10 Jun 2013 21:31:45 +0000 (14:31 -0700)
When we decide not use zero-copy, msg.control should be set to NULL otherwise
macvtap/tap may set zerocopy callbacks which may decrease the kref of ubufs
wrongly.

Bug were introduced by commit cedb9bdce099206290a2bdd02ce47a7b253b6a84
(vhost-net: skip head management if no outstanding).

This solves the following warnings:

WARNING: at include/linux/kref.h:47 handle_tx+0x477/0x4b0 [vhost_net]()
Modules linked in: vhost_net macvtap macvlan tun nfsd exportfs bridge stp llc openvswitch kvm_amd kvm bnx2 megaraid_sas [last unloaded: tun]
CPU: 5 PID: 8670 Comm: vhost-8668 Not tainted 3.10.0-rc2+ #1566
Hardware name: Dell Inc. PowerEdge R715/00XHKG, BIOS 1.5.2 04/19/2011
ffffffffa0198323 ffff88007c9ebd08 ffffffff81796b73 ffff88007c9ebd48
ffffffff8103d66b 000000007b773e20 ffff8800779f0000 ffff8800779f43f0
ffff8800779f8418 000000000000015c 0000000000000062 ffff88007c9ebd58
Call Trace:
[<ffffffff81796b73>] dump_stack+0x19/0x1e
[<ffffffff8103d66b>] warn_slowpath_common+0x6b/0xa0
[<ffffffff8103d6b5>] warn_slowpath_null+0x15/0x20
[<ffffffffa0197627>] handle_tx+0x477/0x4b0 [vhost_net]
[<ffffffffa0197690>] handle_tx_kick+0x10/0x20 [vhost_net]
[<ffffffffa019541e>] vhost_worker+0xfe/0x1a0 [vhost_net]
[<ffffffffa0195320>] ? vhost_attach_cgroups_work+0x30/0x30 [vhost_net]
[<ffffffffa0195320>] ? vhost_attach_cgroups_work+0x30/0x30 [vhost_net]
[<ffffffff81061f46>] kthread+0xc6/0xd0
[<ffffffff81061e80>] ? kthread_freezable_should_stop+0x70/0x70
[<ffffffff817a1aec>] ret_from_fork+0x7c/0xb0
[<ffffffff81061e80>] ? kthread_freezable_should_stop+0x70/0x70

Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/vhost/net.c

index 2b51e2336aa20006e47cb0f1ac7ea146120413a3..b07d96b8c0d1cc2e3a28ae616837db8f8d8a561b 100644 (file)
@@ -436,7 +436,8 @@ static void handle_tx(struct vhost_net *net)
                                kref_get(&ubufs->kref);
                        }
                        nvq->upend_idx = (nvq->upend_idx + 1) % UIO_MAXIOV;
-               }
+               } else
+                       msg.msg_control = NULL;
                /* TODO: Check specific error and bomb out unless ENOBUFS? */
                err = sock->ops->sendmsg(NULL, sock, &msg, len);
                if (unlikely(err < 0)) {