[PATCH] watchdog: fix watchdog_set_timeout breakage for drivers like imxwd

[PATCH] watchdog: fix watchdog_set_timeout breakage for drivers like imxwd

[Ahmad Fatoum]

Commit 0b944fce55f4 ("watchdog: permit `wd 0` for non-stoppable, but
inactive, watchdogs") silently broke use of non-stoppable, but inactive,
watchdogs altogether. Because those always had watchdog_hw_running
evaluating to false, it early exited without a chance to ever actually
set the timeout (and for watchdog_hw_running to evaluate to != false).

This results in following watchdog drivers being broken in v2020.08.0:

 - imxwd
 - f71808e_wdt
 - at91sam9_wdt

Fixes: 0b944fce55f4 ("watchdog: permit `wd 0` for non-stoppable, but inactive, watchdogs")
Reported-by: Bastian Krause <bst@pengutronix.de>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
---
Please apply to master (and to v2020.08.1?).
---
 drivers/watchdog/wd_core.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/drivers/watchdog/wd_core.c b/drivers/watchdog/wd_core.c
index ab1dcaa40a3a..4247288dbd0a 100644
--- a/drivers/watchdog/wd_core.c
+++ b/drivers/watchdog/wd_core.c
@@ -45,12 +45,11 @@ int watchdog_set_timeout(struct watchdog *wd, unsigned timeout)
 	if (timeout > wd->timeout_max)
 		return -EINVAL;
 
-	if (watchdog_hw_running(wd) == false)
-		return 0;
-
 	pr_debug("setting timeout on %s to %ds\n", watchdog_name(wd), timeout);
 
 	ret = wd->set_timeout(wd, timeout);
+	if (ret == -ENOSYS && watchdog_hw_running(wd) == false)
+		return 0;
 	if (ret)
 		return ret;
 
-- 
2.28.0


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] watchdog: fix watchdog_set_timeout breakage for drivers like imxwd

[Sascha Hauer]

Hi Ahmad,

On Tue, Aug 18, 2020 at 03:18:16PM +0200, Ahmad Fatoum wrote:
> Commit 0b944fce55f4 ("watchdog: permit `wd 0` for non-stoppable, but
> inactive, watchdogs") silently broke use of non-stoppable, but inactive,
> watchdogs altogether.

Didn't this commit break *all* watchdogs? With 0b944fce55f4 an inactive
watchdog will never be activated anymore.

Sascha

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] watchdog: fix watchdog_set_timeout breakage for drivers like imxwd

[Ahmad Fatoum]

On 8/19/20 8:59 AM, Sascha Hauer wrote:
> Hi Ahmad,
> 
> On Tue, Aug 18, 2020 at 03:18:16PM +0200, Ahmad Fatoum wrote:
>> Commit 0b944fce55f4 ("watchdog: permit `wd 0` for non-stoppable, but
>> inactive, watchdogs") silently broke use of non-stoppable, but inactive,
>> watchdogs altogether.
> 
> Didn't this commit break *all* watchdogs? With 0b944fce55f4 an inactive
> watchdog will never be activated anymore.

watchdog_hw_running() returns -ENOSYS for watchdog that don't report whether
they are running, so it broke all those who do report it and are inactive at
driver probe time. 

> 
> Sascha
> 

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] watchdog: fix watchdog_set_timeout breakage for drivers like imxwd

[Sascha Hauer]

On Wed, Aug 19, 2020 at 09:18:44AM +0200, Ahmad Fatoum wrote:
> On 8/19/20 8:59 AM, Sascha Hauer wrote:
> > Hi Ahmad,
> > 
> > On Tue, Aug 18, 2020 at 03:18:16PM +0200, Ahmad Fatoum wrote:
> >> Commit 0b944fce55f4 ("watchdog: permit `wd 0` for non-stoppable, but
> >> inactive, watchdogs") silently broke use of non-stoppable, but inactive,
> >> watchdogs altogether.
> > 
> > Didn't this commit break *all* watchdogs? With 0b944fce55f4 an inactive
> > watchdog will never be activated anymore.
> 
> watchdog_hw_running() returns -ENOSYS for watchdog that don't report whether
> they are running, so it broke all those who do report it and are inactive at
> driver probe time.

Ah, I see. I didn't look at the implementation of watchdog_hw_running().
The code explicitly tested for a boolean result, so it seemed obvious to
me that this function returns either true or false, but not an error.

Sascha

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: [PATCH] watchdog: fix watchdog_set_timeout breakage for drivers like imxwd

[Sascha Hauer]

On Tue, Aug 18, 2020 at 03:18:16PM +0200, Ahmad Fatoum wrote:
> Commit 0b944fce55f4 ("watchdog: permit `wd 0` for non-stoppable, but
> inactive, watchdogs") silently broke use of non-stoppable, but inactive,
> watchdogs altogether. Because those always had watchdog_hw_running
> evaluating to false, it early exited without a chance to ever actually
> set the timeout (and for watchdog_hw_running to evaluate to != false).
> 
> This results in following watchdog drivers being broken in v2020.08.0:
> 
>  - imxwd
>  - f71808e_wdt
>  - at91sam9_wdt
> 
> Fixes: 0b944fce55f4 ("watchdog: permit `wd 0` for non-stoppable, but inactive, watchdogs")
> Reported-by: Bastian Krause <bst@pengutronix.de>
> Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
> ---
> Please apply to master (and to v2020.08.1?).
> ---
>  drivers/watchdog/wd_core.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/watchdog/wd_core.c b/drivers/watchdog/wd_core.c
> index ab1dcaa40a3a..4247288dbd0a 100644
> --- a/drivers/watchdog/wd_core.c
> +++ b/drivers/watchdog/wd_core.c
> @@ -45,12 +45,11 @@ int watchdog_set_timeout(struct watchdog *wd, unsigned timeout)
>  	if (timeout > wd->timeout_max)
>  		return -EINVAL;
>  
> -	if (watchdog_hw_running(wd) == false)
> -		return 0;
> -
>  	pr_debug("setting timeout on %s to %ds\n", watchdog_name(wd), timeout);
>  
>  	ret = wd->set_timeout(wd, timeout);
> +	if (ret == -ENOSYS && watchdog_hw_running(wd) == false)

Instead of testing for a specific error code from set_timeout(), can't
we just do a

	if (!timeout && !watchdog_hw_running(wd))
		return 0;

This would make the intention of the code more clear.

Sascha

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox