linux-kernel - [PATCH 2/2] HID: wacom: fix freeze on open when autosuspend is on

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: <1417452760-2763-2-git-send-email-benjamin.tissoires@redhat.com>
Date:	Mon,  1 Dec 2014 11:52:40 -0500
From:	Benjamin Tissoires <benjamin.tissoires@...hat.com>
To:	Jiri Kosina <jkosina@...e.cz>,
	Dmitry Torokhov <dmitry.torokhov@...il.com>,
	Jason Gerecke <killertofu@...il.com>,
	Ping Cheng <pinglinux@...il.com>,
	Hans Spath <inbox-546@...s-spath.de>
Cc:	linux-input@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: [PATCH 2/2] HID: wacom: fix freeze on open when autosuspend is on

Since the conversion from USB to HID (in v3.17), some people reported a
freeze on boot with the wacom driver. Hans managed to get a stacktrace:

[  240.272331] Call Trace:
[  240.272338]  [<ffffffff813de7b9>] ? usb_hcd_submit_urb+0xa9/0xb10
[  240.272347]  [<ffffffff81555579>] schedule+0x29/0x70
[  240.272355]  [<ffffffff815559e6>] schedule_preempt_disabled+0x16/0x20
[  240.272363]  [<ffffffff81557365>] __mutex_lock_slowpath+0xe5/0x230
[  240.272372]  [<ffffffff815574c7>] mutex_lock+0x17/0x30
[  240.272380]  [<ffffffffa063c1d2>] wacom_resume+0x22/0x50 [wacom]
[  240.272396]  [<ffffffffa01aea8a>] hid_resume_common+0xba/0x110 [usbhid]
[  240.272404]  [<ffffffff813e5890>] ? usb_runtime_suspend+0x80/0x80
[  240.272417]  [<ffffffffa01aeb1d>] hid_resume+0x3d/0x70 [usbhid]
[  240.272425]  [<ffffffff813e44a6>] usb_resume_interface.isra.6+0xb6/0x120
[  240.272432]  [<ffffffff813e4774>] usb_resume_both+0x74/0x140
[  240.272439]  [<ffffffff813e58aa>] usb_runtime_resume+0x1a/0x20
[  240.272446]  [<ffffffff813b1912>] __rpm_callback+0x32/0x70
[  240.272453]  [<ffffffff813b1976>] rpm_callback+0x26/0xa0
[  240.272460]  [<ffffffff813b2d71>] rpm_resume+0x4b1/0x690
[  240.272468]  [<ffffffff812ab992>] ? radix_tree_lookup_slot+0x22/0x50
[  240.272475]  [<ffffffff813b2c1a>] rpm_resume+0x35a/0x690
[  240.272482]  [<ffffffff8116e9c9>] ? zone_statistics+0x89/0xa0
[  240.272489]  [<ffffffff813b2f90>] __pm_runtime_resume+0x40/0x60
[  240.272497]  [<ffffffff813e4272>] usb_autopm_get_interface+0x22/0x60
[  240.272509]  [<ffffffffa01ae8d9>] usbhid_open+0x59/0xe0 [usbhid]
[  240.272517]  [<ffffffffa063ac85>] wacom_open+0x35/0x50 [wacom]
[  240.272525]  [<ffffffff813f37b9>] input_open_device+0x79/0xa0
[  240.272534]  [<ffffffffa048d1c1>] evdev_open+0x1b1/0x200 [evdev]
[  240.272543]  [<ffffffff811c899e>] chrdev_open+0xae/0x1f0
[  240.272549]  [<ffffffff811c88f0>] ? cdev_put+0x30/0x30
[  240.272556]  [<ffffffff811c17e2>] do_dentry_open+0x1d2/0x320
[  240.272562]  [<ffffffff811c1cd1>] finish_open+0x31/0x50
[  240.272571]  [<ffffffff811d2202>] do_last.isra.36+0x652/0xe50
[  240.272579]  [<ffffffff811d2ac7>] path_openat+0xc7/0x6f0
[  240.272586]  [<ffffffff811cf012>] ? final_putname+0x22/0x50
[  240.272594]  [<ffffffff811d42d2>] ? user_path_at_empty+0x72/0xd0
[  240.272602]  [<ffffffff811d43fd>] do_filp_open+0x4d/0xc0
[...]

So here, wacom_open is called, and then wacom_resume is called by the
PM system. However, wacom_open already took the lock when wacom_resume
tries to get it. Freeze.

A little bit of history shows that this already happened in the past
- commit f6cd378372bf ("Input: wacom - fix runtime PM related deadlock"),
and the solution was to call first the PM function before taking the lock.

The lock was introduced in commit commit e722409445fb ("Input: wacom -
implement suspend and autosuspend") when the autosuspend feature has
been added. Given that usbhid already takes care of this very same
locking between suspend/resume, I think we can simply kill the lock
in open/close.

The lock is now used also with LEDs, so we can not remove it completely.

Reported-by: Hans Spath <inbox-546@...s-spath.de>
Tested-by: Hans Spath <inbox-546@...s-spath.de>
CC: stable@...r.kernel.org # v3.17+
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@...hat.com>
---
 drivers/hid/wacom_sys.c | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)

diff --git a/drivers/hid/wacom_sys.c b/drivers/hid/wacom_sys.c
index eb55316..17bf52b 100644
--- a/drivers/hid/wacom_sys.c
+++ b/drivers/hid/wacom_sys.c
@@ -70,22 +70,15 @@ static int wacom_raw_event(struct hid_device *hdev, struct hid_report *report,
 static int wacom_open(struct input_dev *dev)
 {
 	struct wacom *wacom = input_get_drvdata(dev);
-	int retval;
-
-	mutex_lock(&wacom->lock);
-	retval = hid_hw_open(wacom->hdev);
-	mutex_unlock(&wacom->lock);
 
-	return retval;
+	return hid_hw_open(wacom->hdev);
 }
 
 static void wacom_close(struct input_dev *dev)
 {
 	struct wacom *wacom = input_get_drvdata(dev);
 
-	mutex_lock(&wacom->lock);
 	hid_hw_close(wacom->hdev);
-	mutex_unlock(&wacom->lock);
 }
 
 /*
-- 
2.1.0

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/