linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v2] virtio-scsi: Fix the race condition in virtscsi_handle_event
@ 2015-01-06  7:33 Fam Zheng
  2015-01-06  9:01 ` Venkatesh Srinivas
  2015-01-06  9:35 ` Michael S. Tsirkin
  0 siblings, 2 replies; 4+ messages in thread
From: Fam Zheng @ 2015-01-06  7:33 UTC (permalink / raw)
  To: linux-kernel
  Cc: James E.J. Bottomley, linux-scsi, Christoph Hellwig,
	Michael S. Tsirkin, Venkatesh Srinivas

There is a race condition in virtscsi_handle_event, when many device
hotplug/unplug events flush in quickly.

The scsi_remove_device in virtscsi_handle_transport_reset may trigger
the BUG_ON in scsi_target_reap, because the state is altered behind it,
probably by scsi_scan_host of another event. I'm able to reproduce it by
repeatedly plugging and unplugging a scsi disk with the same lun number.

To make is safe, a single thread workqueue local to the module is added
which runs the scan work. With this change, the panic goes away.

Signed-off-by: Fam Zheng <famz@redhat.com>

---

v2: Use a single threaded workqueue instead of mutex to serialize work
(Venkatesh)
---
 drivers/scsi/virtio_scsi.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/drivers/scsi/virtio_scsi.c b/drivers/scsi/virtio_scsi.c
index c52bb5d..71b0091 100644
--- a/drivers/scsi/virtio_scsi.c
+++ b/drivers/scsi/virtio_scsi.c
@@ -120,6 +120,8 @@ struct virtio_scsi {
 
 static struct kmem_cache *virtscsi_cmd_cache;
 static mempool_t *virtscsi_cmd_pool;
+static struct workqueue_struct *virtscsi_scan_wq;
+
 
 static inline struct Scsi_Host *virtio_scsi_host(struct virtio_device *vdev)
 {
@@ -404,7 +406,7 @@ static void virtscsi_complete_event(struct virtio_scsi *vscsi, void *buf)
 	struct virtio_scsi_event_node *event_node = buf;
 
 	if (!vscsi->stop_events)
-		queue_work(system_freezable_wq, &event_node->work);
+		queue_work(virtscsi_scan_wq, &event_node->work);
 }
 
 static void virtscsi_event_done(struct virtqueue *vq)
@@ -1119,6 +1121,12 @@ static int __init init(void)
 		pr_err("mempool_create() for virtscsi_cmd_pool failed\n");
 		goto error;
 	}
+	virtscsi_scan_wq = create_singlethread_workqueue("virtscsi-scan");
+	if (!virtscsi_scan_wq) {
+		pr_err("create_singlethread_workqueue() for virtscsi_scan_wq failed\n");
+		goto error;
+	}
+
 	ret = register_virtio_driver(&virtio_scsi_driver);
 	if (ret < 0)
 		goto error;
@@ -1126,6 +1134,9 @@ static int __init init(void)
 	return 0;
 
 error:
+	if (virtscsi_scan_wq) {
+		destroy_workqueue(virtscsi_scan_wq);
+	}
 	if (virtscsi_cmd_pool) {
 		mempool_destroy(virtscsi_cmd_pool);
 		virtscsi_cmd_pool = NULL;
@@ -1142,6 +1153,7 @@ static void __exit fini(void)
 	unregister_virtio_driver(&virtio_scsi_driver);
 	mempool_destroy(virtscsi_cmd_pool);
 	kmem_cache_destroy(virtscsi_cmd_cache);
+	destroy_workqueue(virtscsi_scan_wq);
 }
 module_init(init);
 module_exit(fini);
-- 
1.9.3


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH v2] virtio-scsi: Fix the race condition in virtscsi_handle_event
  2015-01-06  7:33 [PATCH v2] virtio-scsi: Fix the race condition in virtscsi_handle_event Fam Zheng
@ 2015-01-06  9:01 ` Venkatesh Srinivas
  2015-01-06  9:35 ` Michael S. Tsirkin
  1 sibling, 0 replies; 4+ messages in thread
From: Venkatesh Srinivas @ 2015-01-06  9:01 UTC (permalink / raw)
  To: Fam Zheng
  Cc: Linux Kernel Developers List, James E.J. Bottomley, linux-scsi,
	Christoph Hellwig, Michael S. Tsirkin

On Mon, Jan 5, 2015 at 11:33 PM, Fam Zheng <famz@redhat.com> wrote:
> There is a race condition in virtscsi_handle_event, when many device
> hotplug/unplug events flush in quickly.
>
> The scsi_remove_device in virtscsi_handle_transport_reset may trigger
> the BUG_ON in scsi_target_reap, because the state is altered behind it,
> probably by scsi_scan_host of another event. I'm able to reproduce it by
> repeatedly plugging and unplugging a scsi disk with the same lun number.
>
> To make is safe, a single thread workqueue local to the module is added
> which runs the scan work. With this change, the panic goes away.
>
> Signed-off-by: Fam Zheng <famz@redhat.com>
>
> ---
>
> v2: Use a single threaded workqueue instead of mutex to serialize work
> (Venkatesh)
> ---
>  drivers/scsi/virtio_scsi.c | 14 +++++++++++++-
>  1 file changed, 13 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/scsi/virtio_scsi.c b/drivers/scsi/virtio_scsi.c
> index c52bb5d..71b0091 100644
> --- a/drivers/scsi/virtio_scsi.c
> +++ b/drivers/scsi/virtio_scsi.c
> @@ -120,6 +120,8 @@ struct virtio_scsi {
>
>  static struct kmem_cache *virtscsi_cmd_cache;
>  static mempool_t *virtscsi_cmd_pool;
> +static struct workqueue_struct *virtscsi_scan_wq;
> +
>
>  static inline struct Scsi_Host *virtio_scsi_host(struct virtio_device *vdev)
>  {
> @@ -404,7 +406,7 @@ static void virtscsi_complete_event(struct virtio_scsi *vscsi, void *buf)
>         struct virtio_scsi_event_node *event_node = buf;
>
>         if (!vscsi->stop_events)
> -               queue_work(system_freezable_wq, &event_node->work);
> +               queue_work(virtscsi_scan_wq, &event_node->work);
>  }
>
>  static void virtscsi_event_done(struct virtqueue *vq)
> @@ -1119,6 +1121,12 @@ static int __init init(void)
>                 pr_err("mempool_create() for virtscsi_cmd_pool failed\n");
>                 goto error;
>         }
> +       virtscsi_scan_wq = create_singlethread_workqueue("virtscsi-scan");
> +       if (!virtscsi_scan_wq) {
> +               pr_err("create_singlethread_workqueue() for virtscsi_scan_wq failed\n");
> +               goto error;
> +       }
> +
>         ret = register_virtio_driver(&virtio_scsi_driver);
>         if (ret < 0)
>                 goto error;
> @@ -1126,6 +1134,9 @@ static int __init init(void)
>         return 0;
>
>  error:
> +       if (virtscsi_scan_wq) {
> +               destroy_workqueue(virtscsi_scan_wq);
> +       }
>         if (virtscsi_cmd_pool) {
>                 mempool_destroy(virtscsi_cmd_pool);
>                 virtscsi_cmd_pool = NULL;
> @@ -1142,6 +1153,7 @@ static void __exit fini(void)
>         unregister_virtio_driver(&virtio_scsi_driver);
>         mempool_destroy(virtscsi_cmd_pool);
>         kmem_cache_destroy(virtscsi_cmd_cache);
> +       destroy_workqueue(virtscsi_scan_wq);
>  }
>  module_init(init);
>  module_exit(fini);
> --
> 1.9.3

Reviewed-by: Venkatesh Srinivas <venkateshs@google.com>

-- vs;

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH v2] virtio-scsi: Fix the race condition in virtscsi_handle_event
  2015-01-06  7:33 [PATCH v2] virtio-scsi: Fix the race condition in virtscsi_handle_event Fam Zheng
  2015-01-06  9:01 ` Venkatesh Srinivas
@ 2015-01-06  9:35 ` Michael S. Tsirkin
  2015-01-06 13:25   ` Fam Zheng
  1 sibling, 1 reply; 4+ messages in thread
From: Michael S. Tsirkin @ 2015-01-06  9:35 UTC (permalink / raw)
  To: Fam Zheng
  Cc: linux-kernel, James E.J. Bottomley, linux-scsi, Christoph Hellwig,
	Venkatesh Srinivas

On Tue, Jan 06, 2015 at 03:33:36PM +0800, Fam Zheng wrote:
> There is a race condition in virtscsi_handle_event, when many device
> hotplug/unplug events flush in quickly.
> 
> The scsi_remove_device in virtscsi_handle_transport_reset may trigger
> the BUG_ON in scsi_target_reap, because the state is altered behind it,
> probably by scsi_scan_host of another event. I'm able to reproduce it by
> repeatedly plugging and unplugging a scsi disk with the same lun number.
> 
> To make is safe, a single thread workqueue local to the module is added
> which runs the scan work. With this change, the panic goes away.
> 
> Signed-off-by: Fam Zheng <famz@redhat.com>
> ---
> 
> v2: Use a single threaded workqueue instead of mutex to serialize work
> (Venkatesh)
> ---
>  drivers/scsi/virtio_scsi.c | 14 +++++++++++++-
>  1 file changed, 13 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/scsi/virtio_scsi.c b/drivers/scsi/virtio_scsi.c
> index c52bb5d..71b0091 100644
> --- a/drivers/scsi/virtio_scsi.c
> +++ b/drivers/scsi/virtio_scsi.c
> @@ -120,6 +120,8 @@ struct virtio_scsi {
>  
>  static struct kmem_cache *virtscsi_cmd_cache;
>  static mempool_t *virtscsi_cmd_pool;
> +static struct workqueue_struct *virtscsi_scan_wq;
> +
>  

Don't add two empty lines please.

>  static inline struct Scsi_Host *virtio_scsi_host(struct virtio_device *vdev)
>  {
> @@ -404,7 +406,7 @@ static void virtscsi_complete_event(struct virtio_scsi *vscsi, void *buf)
>  	struct virtio_scsi_event_node *event_node = buf;
>  
>  	if (!vscsi->stop_events)
> -		queue_work(system_freezable_wq, &event_node->work);
> +		queue_work(virtscsi_scan_wq, &event_node->work);
>  }
>  
>  static void virtscsi_event_done(struct virtqueue *vq)
> @@ -1119,6 +1121,12 @@ static int __init init(void)
>  		pr_err("mempool_create() for virtscsi_cmd_pool failed\n");
>  		goto error;
>  	}
> +	virtscsi_scan_wq = create_singlethread_workqueue("virtscsi-scan");
> +	if (!virtscsi_scan_wq) {
> +		pr_err("create_singlethread_workqueue() for virtscsi_scan_wq failed\n");
> +		goto error;
> +	}
> +
>  	ret = register_virtio_driver(&virtio_scsi_driver);
>  	if (ret < 0)
>  		goto error;
> @@ -1126,6 +1134,9 @@ static int __init init(void)
>  	return 0;
>  
>  error:
> +	if (virtscsi_scan_wq) {
> +		destroy_workqueue(virtscsi_scan_wq);
> +	}
>  	if (virtscsi_cmd_pool) {
>  		mempool_destroy(virtscsi_cmd_pool);
>  		virtscsi_cmd_pool = NULL;
> @@ -1142,6 +1153,7 @@ static void __exit fini(void)
>  	unregister_virtio_driver(&virtio_scsi_driver);
>  	mempool_destroy(virtscsi_cmd_pool);
>  	kmem_cache_destroy(virtscsi_cmd_cache);
> +	destroy_workqueue(virtscsi_scan_wq);


Please destroy in reverse order of initialization:
between unregister_virtio_driver and mempool_destroy.

>  }
>  module_init(init);
>  module_exit(fini);
> -- 
> 1.9.3

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH v2] virtio-scsi: Fix the race condition in virtscsi_handle_event
  2015-01-06  9:35 ` Michael S. Tsirkin
@ 2015-01-06 13:25   ` Fam Zheng
  0 siblings, 0 replies; 4+ messages in thread
From: Fam Zheng @ 2015-01-06 13:25 UTC (permalink / raw)
  To: Michael S. Tsirkin
  Cc: linux-kernel, James E.J. Bottomley, linux-scsi, Christoph Hellwig,
	Venkatesh Srinivas

On Tue, 01/06 11:35, Michael S. Tsirkin wrote:
> On Tue, Jan 06, 2015 at 03:33:36PM +0800, Fam Zheng wrote:
> > There is a race condition in virtscsi_handle_event, when many device
> > hotplug/unplug events flush in quickly.
> > 
> > The scsi_remove_device in virtscsi_handle_transport_reset may trigger
> > the BUG_ON in scsi_target_reap, because the state is altered behind it,
> > probably by scsi_scan_host of another event. I'm able to reproduce it by
> > repeatedly plugging and unplugging a scsi disk with the same lun number.
> > 
> > To make is safe, a single thread workqueue local to the module is added
> > which runs the scan work. With this change, the panic goes away.
> > 
> > Signed-off-by: Fam Zheng <famz@redhat.com>
> > ---
> > 
> > v2: Use a single threaded workqueue instead of mutex to serialize work
> > (Venkatesh)
> > ---
> >  drivers/scsi/virtio_scsi.c | 14 +++++++++++++-
> >  1 file changed, 13 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/scsi/virtio_scsi.c b/drivers/scsi/virtio_scsi.c
> > index c52bb5d..71b0091 100644
> > --- a/drivers/scsi/virtio_scsi.c
> > +++ b/drivers/scsi/virtio_scsi.c
> > @@ -120,6 +120,8 @@ struct virtio_scsi {
> >  
> >  static struct kmem_cache *virtscsi_cmd_cache;
> >  static mempool_t *virtscsi_cmd_pool;
> > +static struct workqueue_struct *virtscsi_scan_wq;
> > +
> >  
> 
> Don't add two empty lines please.
> 
> >  static inline struct Scsi_Host *virtio_scsi_host(struct virtio_device *vdev)
> >  {
> > @@ -404,7 +406,7 @@ static void virtscsi_complete_event(struct virtio_scsi *vscsi, void *buf)
> >  	struct virtio_scsi_event_node *event_node = buf;
> >  
> >  	if (!vscsi->stop_events)
> > -		queue_work(system_freezable_wq, &event_node->work);
> > +		queue_work(virtscsi_scan_wq, &event_node->work);
> >  }
> >  
> >  static void virtscsi_event_done(struct virtqueue *vq)
> > @@ -1119,6 +1121,12 @@ static int __init init(void)
> >  		pr_err("mempool_create() for virtscsi_cmd_pool failed\n");
> >  		goto error;
> >  	}
> > +	virtscsi_scan_wq = create_singlethread_workqueue("virtscsi-scan");
> > +	if (!virtscsi_scan_wq) {
> > +		pr_err("create_singlethread_workqueue() for virtscsi_scan_wq failed\n");
> > +		goto error;
> > +	}
> > +
> >  	ret = register_virtio_driver(&virtio_scsi_driver);
> >  	if (ret < 0)
> >  		goto error;
> > @@ -1126,6 +1134,9 @@ static int __init init(void)
> >  	return 0;
> >  
> >  error:
> > +	if (virtscsi_scan_wq) {
> > +		destroy_workqueue(virtscsi_scan_wq);
> > +	}
> >  	if (virtscsi_cmd_pool) {
> >  		mempool_destroy(virtscsi_cmd_pool);
> >  		virtscsi_cmd_pool = NULL;
> > @@ -1142,6 +1153,7 @@ static void __exit fini(void)
> >  	unregister_virtio_driver(&virtio_scsi_driver);
> >  	mempool_destroy(virtscsi_cmd_pool);
> >  	kmem_cache_destroy(virtscsi_cmd_cache);
> > +	destroy_workqueue(virtscsi_scan_wq);
> 
> 
> Please destroy in reverse order of initialization:
> between unregister_virtio_driver and mempool_destroy.

OK, will fix this. Thanks.

Fam

> 
> >  }
> >  module_init(init);
> >  module_exit(fini);
> > -- 
> > 1.9.3

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2015-01-06 13:24 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2015-01-06  7:33 [PATCH v2] virtio-scsi: Fix the race condition in virtscsi_handle_event Fam Zheng
2015-01-06  9:01 ` Venkatesh Srinivas
2015-01-06  9:35 ` Michael S. Tsirkin
2015-01-06 13:25   ` Fam Zheng

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).