On 3/17/21 9:06 AM, Shuah Khan wrote:

On 3/17/21 12:21 AM, Tetsuo Handa wrote:

quoted

Shuah, this driver is getting more and more cryptic and buggy.
Please explain the strategy for serialization before you write patches.

quoted

- Fix attach_store() to check usbip_event_happened() before
   waking up threads.

No, this helps nothing.

quoted

diff --git a/drivers/usb/usbip/vhci_sysfs.c 

b/drivers/usb/usbip/vhci_sysfs.c
index c4b4256e5dad3..f0a770adebd97 100644

--- a/drivers/usb/usbip/vhci_sysfs.c
+++ b/drivers/usb/usbip/vhci_sysfs.c

@@ -418,6 +418,15 @@ static ssize_t attach_store(struct device *dev, 

struct device_attribute *attr,
      spin_unlock_irqrestore(&vhci->lock, flags);
      /* end the lock */
+    if (usbip_event_happened(&vdev->ud)) {
+        /*
+         * something went wrong - event handler shutting
+         * the connection and doing reset - bail out
+         */
+        dev_err(dev, "Event happended - handler is active\n");
+        return -EAGAIN;
+    }
+

detach_store() can queue shutdown event as soon as reaching "/* end 
the lock */" line
but attach_store() might be preempted immediately after verifying that
usbip_event_happened() was false (i.e. at this location) in order to 
wait for
shutdown event posted by detach_store() to be processed.

Please don't review code that isn't sent upstream. This repo you are
looking at is a private branch created just to verify fixes on syzbot.

I understand the race window you are talking about. I have my way of
working to resolve it.

thanks,
-- Shuah