linux-kernel - Re: 29-rc-mmotm - HID/USB wedge w/ WARNING: at kernel/workqueue.c:371

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <200903180848.51755.oliver@neukum.org>
Date:	Wed, 18 Mar 2009 08:48:50 +0100
From:	Oliver Neukum <oliver@...kum.org>
To:	Lai Jiangshan <laijs@...fujitsu.com>
Cc:	Andrew Morton <akpm@...ux-foundation.org>, Valdis.Kletnieks@...edu,
	jkosina@...e.cz, gregkh@...e.de, linux-kernel@...r.kernel.org,
	Oleg Nesterov <oleg@...sign.ru>
Subject: Re: 29-rc-mmotm - HID/USB wedge w/ WARNING: at kernel/workqueue.c:371

Am Mittwoch 18 März 2009 04:26:42 schrieb Lai Jiangshan:
> Andrew Morton wrote:
> > On Wed, 18 Mar 2009 10:29:40 +0800 Lai Jiangshan <laijs@...fujitsu.com>
> > wrote:
> >
> >
> > static void run_workqueue(struct cpu_workqueue_struct *cwq)
> > {
> > 	spin_lock_irq(&cwq->lock);
> > 	cwq->run_depth++;
> > 	if (cwq->run_depth > 3) {
> > 		/* morton gets to eat his hat */
> > 		printk("%s: recursion depth exceeded: %d\n",
> > 			__func__, cwq->run_depth);
> > 		dump_stack();
> > 	}
> >
> > That was added five or six years ago, and I never ever got to eat my hat.
>
> If Valdis.Kletnieks let lockdep work, he will see the complain
> from lockdep at first.
>
> >> If it's hard for you avoid flush_scheduled_work() in
> >> keventd's work fuction by other fix, you can create
> >> another workqueue to handle your works, IMO.
> >
> > Why do we need to change anything here?  No known bugs were fixed, and
> > some new ones were added.
>
> I prefer the code which has no known bugs and no known potential bugs.
>
> I defend myself:
> I did not add a new bug as you said, it's the new code hid_cease_io()
> adds a new unsafe usage of workqueue.

Very well, here's a patch.
Please test.

	Regards
		Oliver
diff --git a/drivers/hid/usbhid/hid-core.c b/drivers/hid/usbhid/hid-core.c
index bc017cd..4306cb1 100644
--- a/drivers/hid/usbhid/hid-core.c
+++ b/drivers/hid/usbhid/hid-core.c
@@ -1210,7 +1210,6 @@ static void hid_cease_io(struct usbhid_device *usbhid)
 	usb_kill_urb(usbhid->urbin);
 	usb_kill_urb(usbhid->urbctrl);
 	usb_kill_urb(usbhid->urbout);
-	flush_scheduled_work();
 }
 
 /* Treat USB reset pretty much the same as suspend/resume */
@@ -1222,6 +1221,7 @@ static int hid_pre_reset(struct usb_interface *intf)
 	spin_lock_irq(&usbhid->lock);
 	set_bit(HID_RESET_PENDING, &usbhid->iofl);
 	spin_unlock_irq(&usbhid->lock);
+	cancel_work_sync(&usbhid->restart_work);
 	hid_cease_io(usbhid);
 
 	return 0;
.