* [PATCH] watchdog: gpio_wdt: use __udelay() to avoid recursion
@ 2022-09-27 7:45 Rasmus Villemoes
2022-09-27 8:00 ` Stefan Roese
2022-09-27 12:45 ` Stefan Roese
0 siblings, 2 replies; 4+ messages in thread
From: Rasmus Villemoes @ 2022-09-27 7:45 UTC (permalink / raw)
To: u-boot; +Cc: Paul Doelle, Stefan Roese, Rasmus Villemoes
The udelay() function in lib/time.c contains a WATCHDOG_RESET()
call. The only reason this doesn't lead to a catastrophic infinite
recursion is due to the rate-limiting in wdt-uclass.c:
if (time_after_eq(now, priv->next_reset)) {
priv->next_reset = now + priv->reset_period;
wdt_reset(dev);
}
But this would fall apart if ->next_reset was updated after calling the
device's reset method.
This is needlessly fragile, and it's easy enough to avoid that
recursion in the first place by just using __udelay() directly.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
---
drivers/watchdog/gpio_wdt.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/watchdog/gpio_wdt.c b/drivers/watchdog/gpio_wdt.c
index fe06ec8cc9..2920c2c751 100644
--- a/drivers/watchdog/gpio_wdt.c
+++ b/drivers/watchdog/gpio_wdt.c
@@ -31,7 +31,7 @@ static int gpio_wdt_reset(struct udevice *dev)
case HW_ALGO_LEVEL:
/* Pulse */
dm_gpio_set_value(&priv->gpio, 1);
- udelay(1);
+ __udelay(1);
dm_gpio_set_value(&priv->gpio, 0);
break;
}
--
2.37.2
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] watchdog: gpio_wdt: use __udelay() to avoid recursion
2022-09-27 7:45 [PATCH] watchdog: gpio_wdt: use __udelay() to avoid recursion Rasmus Villemoes
@ 2022-09-27 8:00 ` Stefan Roese
2022-09-27 10:21 ` Pali Rohár
2022-09-27 12:45 ` Stefan Roese
1 sibling, 1 reply; 4+ messages in thread
From: Stefan Roese @ 2022-09-27 8:00 UTC (permalink / raw)
To: Rasmus Villemoes, u-boot; +Cc: Paul Doelle, Pali Rohár
Hi Rasmus,
(add Pali to Cc)
On 27.09.22 09:45, Rasmus Villemoes wrote:
> The udelay() function in lib/time.c contains a WATCHDOG_RESET()
> call. The only reason this doesn't lead to a catastrophic infinite
> recursion is due to the rate-limiting in wdt-uclass.c:
>
> if (time_after_eq(now, priv->next_reset)) {
> priv->next_reset = now + priv->reset_period;
> wdt_reset(dev);
> }
>
> But this would fall apart if ->next_reset was updated after calling the
> device's reset method.
>
> This is needlessly fragile, and it's easy enough to avoid that
> recursion in the first place by just using __udelay() directly.
>
> Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
> ---
> drivers/watchdog/gpio_wdt.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/watchdog/gpio_wdt.c b/drivers/watchdog/gpio_wdt.c
> index fe06ec8cc9..2920c2c751 100644
> --- a/drivers/watchdog/gpio_wdt.c
> +++ b/drivers/watchdog/gpio_wdt.c
> @@ -31,7 +31,7 @@ static int gpio_wdt_reset(struct udevice *dev)
> case HW_ALGO_LEVEL:
> /* Pulse */
> dm_gpio_set_value(&priv->gpio, 1);
> - udelay(1);
> + __udelay(1);
> dm_gpio_set_value(&priv->gpio, 0);
> break;
> }
Yes, good catch:
Reviewed-by: Stefan Roese <sr@denx.de>
The max6370_wdt.c driver has the same problem AFAICT. Pali, could you
please take a look a this?
Thanks,
Stefan
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] watchdog: gpio_wdt: use __udelay() to avoid recursion
2022-09-27 8:00 ` Stefan Roese
@ 2022-09-27 10:21 ` Pali Rohár
0 siblings, 0 replies; 4+ messages in thread
From: Pali Rohár @ 2022-09-27 10:21 UTC (permalink / raw)
To: Stefan Roese; +Cc: Rasmus Villemoes, u-boot, Paul Doelle
On Tuesday 27 September 2022 10:00:00 Stefan Roese wrote:
> The max6370_wdt.c driver has the same problem AFAICT. Pali, could you
> please take a look a this?
Done, here is patch:
https://patchwork.ozlabs.org/project/uboot/patch/20220927101919.6999-1-pali@kernel.org/
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] watchdog: gpio_wdt: use __udelay() to avoid recursion
2022-09-27 7:45 [PATCH] watchdog: gpio_wdt: use __udelay() to avoid recursion Rasmus Villemoes
2022-09-27 8:00 ` Stefan Roese
@ 2022-09-27 12:45 ` Stefan Roese
1 sibling, 0 replies; 4+ messages in thread
From: Stefan Roese @ 2022-09-27 12:45 UTC (permalink / raw)
To: Rasmus Villemoes, u-boot; +Cc: Paul Doelle
On 27.09.22 09:45, Rasmus Villemoes wrote:
> The udelay() function in lib/time.c contains a WATCHDOG_RESET()
> call. The only reason this doesn't lead to a catastrophic infinite
> recursion is due to the rate-limiting in wdt-uclass.c:
>
> if (time_after_eq(now, priv->next_reset)) {
> priv->next_reset = now + priv->reset_period;
> wdt_reset(dev);
> }
>
> But this would fall apart if ->next_reset was updated after calling the
> device's reset method.
>
> This is needlessly fragile, and it's easy enough to avoid that
> recursion in the first place by just using __udelay() directly.
>
> Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Applied to u-boot-watchdog/master
Thanks,
Stefan
> ---
> drivers/watchdog/gpio_wdt.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/watchdog/gpio_wdt.c b/drivers/watchdog/gpio_wdt.c
> index fe06ec8cc9..2920c2c751 100644
> --- a/drivers/watchdog/gpio_wdt.c
> +++ b/drivers/watchdog/gpio_wdt.c
> @@ -31,7 +31,7 @@ static int gpio_wdt_reset(struct udevice *dev)
> case HW_ALGO_LEVEL:
> /* Pulse */
> dm_gpio_set_value(&priv->gpio, 1);
> - udelay(1);
> + __udelay(1);
> dm_gpio_set_value(&priv->gpio, 0);
> break;
> }
Viele Grüße,
Stefan Roese
--
DENX Software Engineering GmbH, Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-51 Fax: (+49)-8142-66989-80 Email: sr@denx.de
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2022-09-27 12:46 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-09-27 7:45 [PATCH] watchdog: gpio_wdt: use __udelay() to avoid recursion Rasmus Villemoes
2022-09-27 8:00 ` Stefan Roese
2022-09-27 10:21 ` Pali Rohár
2022-09-27 12:45 ` Stefan Roese
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.