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  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]
Date:   Thu, 9 Nov 2017 14:39:55 +0100
From:   Rasmus Villemoes <rasmus.villemoes@...vas.dk>
To:     Guenter Roeck <linux@...ck-us.net>
CC:     <linux-watchdog@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        Rasmus Villemoes <rasmus.villemoes@...vas.dk>
Subject: [PATCH] watchdog: gpio_wdt: set WDOG_HW_RUNNING in gpio_wdt_stop

The first patch above (https://patchwork.kernel.org/patch/9970181/)
makes the oops go away, but it just papers over the problem. The real
problem is that the watchdog core clears WDOG_HW_RUNNING in
watchdog_stop, and the gpio driver fails to set it in its stop
function when it doesn't actually stop it. This means that the core
doesn't know that it now has responsibility for petting the device, in
turn causing the device to reset the system (I hadn't noticed this
because the board I'm working on has that reset logic disabled).

How about this (other drivers may of course have the same problem, I
haven't checked). One might say that ->stop should return an error
when the device can't be stopped, but OTOH this brings parity between
a device without a ->stop method and a GPIO wd that has always-running
set. IOW, I think ->stop should only return an error when an actual
attempt to stop the hardware failed.

From: Rasmus Villemoes <rasmus.villemoes@...vas.dk>

The watchdog framework clears WDOG_HW_RUNNING before calling
->stop. If the driver is unable to stop the device, it is supposed to
set that bit again so that the watchdog core takes care of sending
heart-beats while the device is not open from user-space. Update the
gpio_wdt driver to honour that contract (and get rid of the redundant
clearing of WDOG_HW_RUNNING).

Fixes: 3c10bbde10 ("watchdog: core: Clear WDOG_HW_RUNNING before calling the stop function")
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@...vas.dk>
---
 drivers/watchdog/gpio_wdt.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/watchdog/gpio_wdt.c b/drivers/watchdog/gpio_wdt.c
index cb66c2f..7a6279d 100644
--- a/drivers/watchdog/gpio_wdt.c
+++ b/drivers/watchdog/gpio_wdt.c
@@ -80,7 +80,8 @@ static int gpio_wdt_stop(struct watchdog_device *wdd)
 
 	if (!priv->always_running) {
 		gpio_wdt_disable(priv);
-		clear_bit(WDOG_HW_RUNNING, &wdd->status);
+	} else {
+		set_bit(WDOG_HW_RUNNING, &wdd->status);
 	}
 
 	return 0;
-- 
2.7.4

Powered by blists - more mailing lists