On Mon, Apr 23, 2012 at 11:42 PM, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote:
> On Sun, 22 Apr 2012, Ming Lei wrote:
>> If the lock which is to be acquired in the URB complete handler is dropped
>> before calling usb_unlink_urb, one new submitted URB in complete handler
>> may be unlinked, as mentioned by Oliver already.
>
> We are now talking about two locks. One of them is held during the
> call to usb_unlink_urb; the completion handler does not acquire that
> lock if the URB's status is -ECONNRESET. The other lock is dropped
> before usb_unlink_urb is called, so the completion handler can safely
> grab it.
Yes, adding a new unlink lock plus -ECONNRESET status can
fix the problem, and it is a very simple solution, just like below:
diff --git a/drivers/hid/usbhid/hid-core.c b/drivers/hid/usbhid/hid-core.c
index aa1c503..a1adb81 100644
--- a/drivers/hid/usbhid/hid-core.c
+++ b/drivers/hid/usbhid/hid-core.c
@@ -430,6 +430,8 @@ static void hid_irq_out(struct urb *urb)
}
spin_lock_irqsave(&usbhid->lock, flags);
+ if (status != -ECONNRESET)
+ spin_lock(&usbhid->unlink_lock);
if (unplug)
usbhid->outtail = usbhid->outhead;
@@ -438,11 +440,15 @@ static void hid_irq_out(struct urb *urb)
if (usbhid->outhead != usbhid->outtail && !hid_submit_out(hid)) {
/* Successfully submitted next urb in queue */
+ if (status != -ECONNRESET)
+ spin_unlock(&usbhid->unlink_lock);
spin_unlock_irqrestore(&usbhid->lock, flags);
return;
}
clear_bit(HID_OUT_RUNNING, &usbhid->iofl);
+ if (status != -ECONNRESET)
+ spin_unlock(&usbhid->unlink_lock);
spin_unlock_irqrestore(&usbhid->lock, flags);
usb_autopm_put_interface_async(usbhid->intf);
wake_up(&usbhid->wait);
@@ -459,6 +465,8 @@ static void hid_ctrl(struct urb *urb)
int unplug = 0, status = urb->status;
spin_lock(&usbhid->lock);
+ if (status != -ECONNRESET)
+ spin_lock(&usbhid->unlink_lock);
switch (status) {
case 0: /* success */
@@ -486,11 +494,15 @@ static void hid_ctrl(struct urb *urb)
if (usbhid->ctrlhead != usbhid->ctrltail && !hid_submit_ctrl(hid)) {
/* Successfully submitted next urb in queue */
+ if (status != -ECONNRESET)
+ spin_unlock(&usbhid->unlink_lock);
spin_unlock(&usbhid->lock);
return;
}
clear_bit(HID_CTRL_RUNNING, &usbhid->iofl);
+ if (status != -ECONNRESET)
+ spin_unlock(&usbhid->unlink_lock);
spin_unlock(&usbhid->lock);
usb_autopm_put_interface_async(usbhid->intf);
wake_up(&usbhid->wait);
@@ -546,8 +558,13 @@ static void __usbhid_submit_report(struct
hid_device *hid, struct hid_report *re
* no race because this is called under
* spinlock
*/
- if (time_after(jiffies, usbhid->last_out + HZ * 5))
+ spin_lock(&usbhid->unlink_lock);
+ if (time_after(jiffies, usbhid->last_out + HZ * 5)) {
+ spin_unlock(&usbhid->lock);
usb_unlink_urb(usbhid->urbout);
+ spin_lock(&usbhid->lock);
+ }
+ spin_unlock(&usbhid->unlink_lock);
}
return;
}
@@ -594,8 +611,13 @@ static void __usbhid_submit_report(struct
hid_device *hid, struct hid_report *re
* no race because this is called under
* spinlock
*/
- if (time_after(jiffies, usbhid->last_ctrl + HZ * 5))
+ spin_lock(&usbhid->unlink_lock);
+ if (time_after(jiffies, usbhid->last_ctrl + HZ * 5)) {
+ spin_unlock(&usbhid->lock);
usb_unlink_urb(usbhid->urbctrl);
+ spin_lock(&usbhid->lock);
+ }
+ spin_unlock(&usbhid->unlink_lock);
}
}
@@ -1299,6 +1321,7 @@ static int usbhid_probe(struct usb_interface
*intf, const struct usb_device_id *
INIT_WORK(&usbhid->reset_work, hid_reset);
setup_timer(&usbhid->io_retry, hid_retry_timeout, (unsigned long) hid);
spin_lock_init(&usbhid->lock);
+ spin_lock_init(&usbhid->unlink_lock);
INIT_WORK(&usbhid->led_work, hid_led);
diff --git a/drivers/hid/usbhid/usbhid.h b/drivers/hid/usbhid/usbhid.h
index 1883d7b..69af387 100644
--- a/drivers/hid/usbhid/usbhid.h
+++ b/drivers/hid/usbhid/usbhid.h
@@ -90,6 +90,7 @@ struct usbhid_device {
unsigned long last_out; /* record of last output for timeouts */
spinlock_t lock; /* fifo spinlock */
+ spinlock_t unlink_lock; /* unlink spinlock */
unsigned long iofl; /*
I/O flags (CTRL_RUNNING, OUT_RUNNING) */
struct timer_list io_retry; /*
Retry timer */
unsigned long stop_retry; /*
Time to give up, in jiffies */
>
> On Mon, 23 Apr 2012, Oliver Neukum wrote:
>
>> > If the URB completes asynchronously after unlinking, its status is still
>> > -ECONNRESET, so extra race may be caused without holding the lock
>> > because complete handler will access some global data.
>>
>> That is the race. And you need not invoke global data. The original
>> race opens again if you are submitting a new URB without the lock
>> held.
>> This is because we cannot be sure that the same URB is unlinked
>> only once. A subsequent timeout may kill the wrong URB if the
>> first is unlinked so that the callback really comes in interrupt.
>>
>> But the basic idea is brilliant. It's just that the one way logical implication:
>> recursive direct call of the callback -> status == -ECONNRESET
>> is not strong enough. But that is very easy to fix. As we know whether
>> the callback is directly called or not, all we need to do is differentiate
>> the cases in urb->status, by introducing a new error code.
>
> I don't like the idea of changing the status codes. It would mean
> changing usb_kill_urb too.
>
> Instead of changing return codes or adding locks, how about
> implementing a small state machine for each URB?
>
> Initially the state is ACTIVE.
>
> When the URB times out, acquire the lock. If the state is not
> equal to ACTIVE, drop the lock and return immediately (the URB
> is being unlinked concurrently). Otherwise set the state to
> UNLINK_STARTED, drop the lock, call usb_unlink_urb, and
> reacquire the lock. If the state hasn't changed, set it back
> to ACTIVE. But if the state has changed to UNLINK_FINISHED,
> set it to ACTIVE and resubmit.
>
> In the completion handler, grab the lock. If the state
> is ACTIVE, resubmit. But if the state is UNLINK_STARTED,
> change it to UNLINK_FINISHED and don't resubmit.
Sounds a smart solution, but extra care should be put on submit urb
in other context, maybe the lock is to be held to check URB state before
submitting it because the lock is released during usb_unlink_urb.
Also basically the URB state is maintained by device driver, so looks
better to add the state into specific device driver instead of struct URB to
save each URB storage, but if so, the solution will become more
complicated than the way of adding lock.
> This is a better approach, in that it doesn't make any assumptions
> regarding synchronous vs. asynchronous unlinks. If you want, you could
> have two different ACTIVE substates, one for URBs which haven't yet
> been unlinked and one for URBs which have been. Then you could avoid
> unlinking the same URB twice.
Thanks,
--
Ming Lei
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
