diff mbox

xen: avoid deadlock in xenbus

Message ID 20170728145355.17258-1-jgross@suse.com (mailing list archive)
State New, archived
Headers show

Commit Message

Jürgen Groß July 28, 2017, 2:53 p.m. UTC
When starting the xenwatch thread a theoretical deadlock situation is
possible:

xs_init() contains:

    task = kthread_run(xenwatch_thread, NULL, "xenwatch");
    if (IS_ERR(task))
        return PTR_ERR(task);
    xenwatch_pid = task->pid;

And xenwatch_thread() does:

    mutex_lock(&xenwatch_mutex);
    ...
    event->handle->callback();
    ...
    mutex_unlock(&xenwatch_mutex);

The callback could call unregister_xenbus_watch() which does:

    ...
    if (current->pid != xenwatch_pid)
        mutex_lock(&xenwatch_mutex);
    ...

In case a watch is firing before xenwatch_pid could be set and the
callback of that watch unregisters a watch, then a self-deadlock would
occur.

Avoid this by setting xenwatch_pid in xenwatch_thread().

Signed-off-by: Juergen Gross <jgross@suse.com>
---
 drivers/xen/xenbus/xenbus_xs.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Comments

Boris Ostrovsky July 28, 2017, 3:14 p.m. UTC | #1
On 07/28/2017 10:53 AM, Juergen Gross wrote:
> When starting the xenwatch thread a theoretical deadlock situation is
> possible:
>
> xs_init() contains:
>
>     task = kthread_run(xenwatch_thread, NULL, "xenwatch");
>     if (IS_ERR(task))
>         return PTR_ERR(task);
>     xenwatch_pid = task->pid;
>
> And xenwatch_thread() does:
>
>     mutex_lock(&xenwatch_mutex);
>     ...
>     event->handle->callback();
>     ...
>     mutex_unlock(&xenwatch_mutex);
>
> The callback could call unregister_xenbus_watch() which does:
>
>     ...
>     if (current->pid != xenwatch_pid)
>         mutex_lock(&xenwatch_mutex);
>     ...
>
> In case a watch is firing before xenwatch_pid could be set and the
> callback of that watch unregisters a watch, then a self-deadlock would
> occur.
>
> Avoid this by setting xenwatch_pid in xenwatch_thread().
>
> Signed-off-by: Juergen Gross <jgross@suse.com>

Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>

+stable?
Jürgen Groß July 28, 2017, 3:16 p.m. UTC | #2
On 28/07/17 17:14, Boris Ostrovsky wrote:
> On 07/28/2017 10:53 AM, Juergen Gross wrote:
>> When starting the xenwatch thread a theoretical deadlock situation is
>> possible:
>>
>> xs_init() contains:
>>
>>     task = kthread_run(xenwatch_thread, NULL, "xenwatch");
>>     if (IS_ERR(task))
>>         return PTR_ERR(task);
>>     xenwatch_pid = task->pid;
>>
>> And xenwatch_thread() does:
>>
>>     mutex_lock(&xenwatch_mutex);
>>     ...
>>     event->handle->callback();
>>     ...
>>     mutex_unlock(&xenwatch_mutex);
>>
>> The callback could call unregister_xenbus_watch() which does:
>>
>>     ...
>>     if (current->pid != xenwatch_pid)
>>         mutex_lock(&xenwatch_mutex);
>>     ...
>>
>> In case a watch is firing before xenwatch_pid could be set and the
>> callback of that watch unregisters a watch, then a self-deadlock would
>> occur.
>>
>> Avoid this by setting xenwatch_pid in xenwatch_thread().
>>
>> Signed-off-by: Juergen Gross <jgross@suse.com>
> 
> Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
> 
> +stable?

As this problem is purely theoretical, I don't think the patch is
appropriate for stable (at least the stable rules tell me so).


Juergen
diff mbox

Patch

diff --git a/drivers/xen/xenbus/xenbus_xs.c b/drivers/xen/xenbus/xenbus_xs.c
index e46080214955..3e59590c7254 100644
--- a/drivers/xen/xenbus/xenbus_xs.c
+++ b/drivers/xen/xenbus/xenbus_xs.c
@@ -857,6 +857,8 @@  static int xenwatch_thread(void *unused)
 	struct list_head *ent;
 	struct xs_watch_event *event;
 
+	xenwatch_pid = current->pid;
+
 	for (;;) {
 		wait_event_interruptible(watch_events_waitq,
 					 !list_empty(&watch_events));
@@ -925,7 +927,6 @@  int xs_init(void)
 	task = kthread_run(xenwatch_thread, NULL, "xenwatch");
 	if (IS_ERR(task))
 		return PTR_ERR(task);
-	xenwatch_pid = task->pid;
 
 	/* shutdown watches for kexec boot */
 	xs_reset_watches();