virtio-net: correctly handle cpu hotplug notifier during resuming
authorJason Wang <jasowang@redhat.com>
Tue, 29 Oct 2013 07:11:07 +0000 (15:11 +0800)
committerDavid S. Miller <davem@davemloft.net>
Wed, 30 Oct 2013 02:43:41 +0000 (22:43 -0400)
commit 3ab098df35f8b98b6553edc2e40234af512ba877 (virtio-net: don't respond to
cpu hotplug notifier if we're not ready) tries to bypass the cpu hotplug
notifier by checking the config_enable and does nothing is it was false. So it
need to try to hold the config_lock mutex which may happen in atomic
environment which leads the following warnings:

[  622.944441] CPU0 attaching NULL sched-domain.
[  622.944446] CPU1 attaching NULL sched-domain.
[  622.944485] CPU0 attaching NULL sched-domain.
[  622.950795] BUG: sleeping function called from invalid context at kernel/mutex.c:616
[  622.950796] in_atomic(): 1, irqs_disabled(): 1, pid: 10, name: migration/1
[  622.950796] no locks held by migration/1/10.
[  622.950798] CPU: 1 PID: 10 Comm: migration/1 Not tainted 3.12.0-rc5-wl-01249-gb91e82d #317
[  622.950799] Hardware name: Bochs Bochs, BIOS Bochs 01/01/2011
[  622.950802]  0000000000000000 ffff88001d42dba0 ffffffff81a32f22 ffff88001bfb9c70
[  622.950803]  ffff88001d42dbb0 ffffffff810edb02 ffff88001d42dc38 ffffffff81a396ed
[  622.950805]  0000000000000046 ffff88001d42dbe8 ffffffff810e861d 0000000000000000
[  622.950805] Call Trace:
[  622.950810]  [<ffffffff81a32f22>] dump_stack+0x54/0x74
[  622.950815]  [<ffffffff810edb02>] __might_sleep+0x112/0x114
[  622.950817]  [<ffffffff81a396ed>] mutex_lock_nested+0x3c/0x3c6
[  622.950818]  [<ffffffff810e861d>] ? up+0x39/0x3e
[  622.950821]  [<ffffffff8153ea7c>] ? acpi_os_signal_semaphore+0x21/0x2d
[  622.950824]  [<ffffffff81565ed1>] ? acpi_ut_release_mutex+0x5e/0x62
[  622.950828]  [<ffffffff816d04ec>] virtnet_cpu_callback+0x33/0x87
[  622.950830]  [<ffffffff81a42576>] notifier_call_chain+0x3c/0x5e
[  622.950832]  [<ffffffff810e86a8>] __raw_notifier_call_chain+0xe/0x10
[  622.950835]  [<ffffffff810c5556>] __cpu_notify+0x20/0x37
[  622.950836]  [<ffffffff810c5580>] cpu_notify+0x13/0x15
[  622.950838]  [<ffffffff81a237cd>] take_cpu_down+0x27/0x3a
[  622.950841]  [<ffffffff81136289>] stop_machine_cpu_stop+0x93/0xf1
[  622.950842]  [<ffffffff81136167>] cpu_stopper_thread+0xa0/0x12f
[  622.950844]  [<ffffffff811361f6>] ? cpu_stopper_thread+0x12f/0x12f
[  622.950847]  [<ffffffff81119710>] ? lock_release_holdtime.part.7+0xa3/0xa8
[  622.950848]  [<ffffffff81135e4b>] ? cpu_stop_should_run+0x3f/0x47
[  622.950850]  [<ffffffff810ea9b0>] smpboot_thread_fn+0x1c5/0x1e3
[  622.950852]  [<ffffffff810ea7eb>] ? lg_global_unlock+0x67/0x67
[  622.950854]  [<ffffffff810e36b7>] kthread+0xd8/0xe0
[  622.950857]  [<ffffffff81a3bfad>] ? wait_for_common+0x12f/0x164
[  622.950859]  [<ffffffff810e35df>] ? kthread_create_on_node+0x124/0x124
[  622.950861]  [<ffffffff81a45ffc>] ret_from_fork+0x7c/0xb0
[  622.950862]  [<ffffffff810e35df>] ? kthread_create_on_node+0x124/0x124
[  622.950876] smpboot: CPU 1 is now offline
[  623.194556] SMP alternatives: lockdep: fixing up alternatives
[  623.194559] smpboot: Booting Node 0 Processor 1 APIC 0x1
...

A correct fix is to unregister the hotcpu notifier during restore and register a
new one in resume.

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Tested-by: Fengguang Wu <fengguang.wu@intel.com>
Cc: Wanlong Gao <gaowanlong@cn.fujitsu.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Wanlong Gao <gaowanlong@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/virtio_net.c

index 9fbdfcd1e1a0693e96edd908928953097b99742d..bbc9cb84ec1fe4f857bc5df63a0ecb3f80659b1b 100644 (file)
@@ -1118,11 +1118,6 @@ static int virtnet_cpu_callback(struct notifier_block *nfb,
 {
        struct virtnet_info *vi = container_of(nfb, struct virtnet_info, nb);
 
-       mutex_lock(&vi->config_lock);
-
-       if (!vi->config_enable)
-               goto done;
-
        switch(action & ~CPU_TASKS_FROZEN) {
        case CPU_ONLINE:
        case CPU_DOWN_FAILED:
@@ -1136,8 +1131,6 @@ static int virtnet_cpu_callback(struct notifier_block *nfb,
                break;
        }
 
-done:
-       mutex_unlock(&vi->config_lock);
        return NOTIFY_OK;
 }
 
@@ -1699,6 +1692,8 @@ static int virtnet_freeze(struct virtio_device *vdev)
        struct virtnet_info *vi = vdev->priv;
        int i;
 
+       unregister_hotcpu_notifier(&vi->nb);
+
        /* Prevent config work handler from accessing the device */
        mutex_lock(&vi->config_lock);
        vi->config_enable = false;
@@ -1747,6 +1742,10 @@ static int virtnet_restore(struct virtio_device *vdev)
        virtnet_set_queues(vi, vi->curr_queue_pairs);
        rtnl_unlock();
 
+       err = register_hotcpu_notifier(&vi->nb);
+       if (err)
+               return err;
+
        return 0;
 }
 #endif