[PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window

[PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window

[Philipp Zabel]

There is a cooldown period after watchdog timer setup, and also after each
watchdog ping. If a ping is issued too early, the watchdog triggers the
watchdog error condition and causes a system reset.

The 256 ms period length is a best guess based on the same value being used
for the DA9053 predecessor PMIC and experiments that have shown 200 ms are
not enough to avoid the error.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
---
 drivers/watchdog/da9063_wdt.c | 88 ++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 79 insertions(+), 9 deletions(-)

diff --git a/drivers/watchdog/da9063_wdt.c b/drivers/watchdog/da9063_wdt.c
index e2fe2eb..b2e9201 100644
--- a/drivers/watchdog/da9063_wdt.c
+++ b/drivers/watchdog/da9063_wdt.c
@@ -35,11 +35,15 @@ static const unsigned int wdt_timeout[] = { 0, 2, 4, 8, 16, 32, 65, 131 };
 #define DA9063_WDT_MIN_TIMEOUT		wdt_timeout[DA9063_TWDSCALE_MIN]
 #define DA9063_WDT_MAX_TIMEOUT		wdt_timeout[DA9063_TWDSCALE_MAX]
 #define DA9063_WDG_TIMEOUT		wdt_timeout[3]
+#define DA9063_TWDMIN			256
 
 struct da9063_watchdog {
 	struct da9063 *da9063;
 	struct watchdog_device wdtdev;
 	struct notifier_block restart_handler;
+	struct delayed_work ping_work;
+	struct mutex mutex;
+	bool defer_ping;
 };
 
 static unsigned int da9063_wdt_timeout_to_sel(unsigned int secs)
@@ -54,10 +58,23 @@ static unsigned int da9063_wdt_timeout_to_sel(unsigned int secs)
 	return DA9063_TWDSCALE_MAX;
 }
 
-static int _da9063_wdt_set_timeout(struct da9063 *da9063, unsigned int regval)
+static int _da9063_wdt_set_timeout(struct da9063_watchdog *wdt,
+				   unsigned int regval)
 {
-	return regmap_update_bits(da9063->regmap, DA9063_REG_CONTROL_D,
-				  DA9063_TWDSCALE_MASK, regval);
+	struct da9063 *da9063 = wdt->da9063;
+	int ret;
+
+	mutex_lock(&wdt->mutex);
+
+	ret = regmap_update_bits(da9063->regmap, DA9063_REG_CONTROL_D,
+				 DA9063_TWDSCALE_MASK, regval);
+
+	wdt->defer_ping = false;
+	schedule_delayed_work(&wdt->ping_work, msecs_to_jiffies(DA9063_TWDMIN));
+
+	mutex_unlock(&wdt->mutex);
+
+	return ret;
 }
 
 static int da9063_wdt_start(struct watchdog_device *wdd)
@@ -67,12 +84,16 @@ static int da9063_wdt_start(struct watchdog_device *wdd)
 	int ret;
 
 	selector = da9063_wdt_timeout_to_sel(wdt->wdtdev.timeout);
-	ret = _da9063_wdt_set_timeout(wdt->da9063, selector);
-	if (ret)
+	ret = _da9063_wdt_set_timeout(wdt, selector);
+	if (ret) {
 		dev_err(wdt->da9063->dev, "Watchdog failed to start (err = %d)\n",
 			ret);
+		return ret;
+	}
 
-	return ret;
+	wdd->timeout = wdt_timeout[selector];
+
+	return 0;
 }
 
 static int da9063_wdt_stop(struct watchdog_device *wdd)
@@ -80,6 +101,8 @@ static int da9063_wdt_stop(struct watchdog_device *wdd)
 	struct da9063_watchdog *wdt = watchdog_get_drvdata(wdd);
 	int ret;
 
+	cancel_delayed_work_sync(&wdt->ping_work);
+
 	ret = regmap_update_bits(wdt->da9063->regmap, DA9063_REG_CONTROL_D,
 				 DA9063_TWDSCALE_MASK, DA9063_TWDSCALE_DISABLE);
 	if (ret)
@@ -89,9 +112,8 @@ static int da9063_wdt_stop(struct watchdog_device *wdd)
 	return ret;
 }
 
-static int da9063_wdt_ping(struct watchdog_device *wdd)
+static int da9063_wdt_issue_ping(struct da9063_watchdog *wdt)
 {
-	struct da9063_watchdog *wdt = watchdog_get_drvdata(wdd);
 	int ret;
 
 	ret = regmap_write(wdt->da9063->regmap, DA9063_REG_CONTROL_F,
@@ -100,9 +122,45 @@ static int da9063_wdt_ping(struct watchdog_device *wdd)
 		dev_alert(wdt->da9063->dev, "Failed to ping the watchdog (err = %d)\n",
 			  ret);
 
+	wdt->defer_ping = false;
+	schedule_delayed_work(&wdt->ping_work, msecs_to_jiffies(DA9063_TWDMIN));
+
 	return ret;
 }
 
+static int da9063_wdt_ping(struct watchdog_device *wdd)
+{
+	struct da9063_watchdog *wdt = watchdog_get_drvdata(wdd);
+	int ret = 0;
+
+	mutex_lock(&wdt->mutex);
+
+	if (delayed_work_pending(&wdt->ping_work)) {
+		wdt->defer_ping = true;
+
+		goto out;
+	}
+
+	da9063_wdt_issue_ping(wdt);
+out:
+	mutex_unlock(&wdt->mutex);
+
+	return ret;
+}
+
+static void da9063_wdt_ping_work(struct work_struct *work)
+{
+	struct da9063_watchdog *wdt = container_of(to_delayed_work(work),
+					struct da9063_watchdog, ping_work);
+
+	mutex_lock(&wdt->mutex);
+
+	if (wdt->defer_ping)
+		da9063_wdt_issue_ping(wdt);
+
+	mutex_unlock(&wdt->mutex);
+}
+
 static int da9063_wdt_set_timeout(struct watchdog_device *wdd,
 				  unsigned int timeout)
 {
@@ -111,7 +169,7 @@ static int da9063_wdt_set_timeout(struct watchdog_device *wdd,
 	int ret;
 
 	selector = da9063_wdt_timeout_to_sel(timeout);
-	ret = _da9063_wdt_set_timeout(wdt->da9063, selector);
+	ret = _da9063_wdt_set_timeout(wdt, selector);
 	if (ret)
 		dev_err(wdt->da9063->dev, "Failed to set watchdog timeout (err = %d)\n",
 			ret);
@@ -178,6 +236,9 @@ static int da9063_wdt_probe(struct platform_device *pdev)
 
 	wdt->wdtdev.status = WATCHDOG_NOWAYOUT_INIT_STATUS;
 
+	INIT_DELAYED_WORK(&wdt->ping_work, da9063_wdt_ping_work);
+	mutex_init(&wdt->mutex);
+
 	watchdog_set_drvdata(&wdt->wdtdev, wdt);
 	dev_set_drvdata(&pdev->dev, wdt);
 
@@ -202,13 +263,22 @@ static int da9063_wdt_remove(struct platform_device *pdev)
 	unregister_restart_handler(&wdt->restart_handler);
 
 	watchdog_unregister_device(&wdt->wdtdev);
+	cancel_delayed_work_sync(&wdt->ping_work);
 
 	return 0;
 }
 
+static void da9063_wdt_shutdown(struct platform_device *pdev)
+{
+	struct da9063_watchdog *wdt = dev_get_drvdata(&pdev->dev);
+
+	cancel_delayed_work_sync(&wdt->ping_work);
+}
+
 static struct platform_driver da9063_wdt_driver = {
 	.probe = da9063_wdt_probe,
 	.remove = da9063_wdt_remove,
+	.shutdown = da9063_wdt_shutdown,
 	.driver = {
 		.name = DA9063_DRVNAME_WATCHDOG,
 	},
-- 
2.4.6

[PATCH 2/2] watchdog: da9063: Disable and wait before changing timeout

[Philipp Zabel]

The DA9063 watchdog occasionally enters error condition and resets the
system if the timeout is changed quickly after the timer was enabled.

The method of disabling and waiting for > 150 µs before setting the
new timeout is taken from the DA9052 driver.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
---
 drivers/watchdog/da9063_wdt.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/watchdog/da9063_wdt.c b/drivers/watchdog/da9063_wdt.c
index b2e9201..87e14d8 100644
--- a/drivers/watchdog/da9063_wdt.c
+++ b/drivers/watchdog/da9063_wdt.c
@@ -67,6 +67,11 @@ static int _da9063_wdt_set_timeout(struct da9063_watchdog *wdt,
 	mutex_lock(&wdt->mutex);
 
 	ret = regmap_update_bits(da9063->regmap, DA9063_REG_CONTROL_D,
+				 DA9063_TWDSCALE_MASK, DA9063_TWDSCALE_DISABLE);
+
+	usleep_range(150, 300);
+
+	ret = regmap_update_bits(da9063->regmap, DA9063_REG_CONTROL_D,
 				 DA9063_TWDSCALE_MASK, regval);
 
 	wdt->defer_ping = false;
-- 
2.4.6

Re: [PATCH 2/2] watchdog: da9063: Disable and wait before changing timeout

[Guenter Roeck]

On 08/06/2015 09:36 AM, Philipp Zabel wrote:
> The DA9063 watchdog occasionally enters error condition and resets the
> system if the timeout is changed quickly after the timer was enabled.
>
> The method of disabling and waiting for > 150 µs before setting the
> new timeout is taken from the DA9052 driver.
>
Using sleep concerns me a bit, since during that time the watchdog
is disabled. Bad enough that we have to do this to start with,
but using usleep adds more risk to an already bad situation.

How about udelay() instead ?

Thanks,
Guenter

> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
> ---
>   drivers/watchdog/da9063_wdt.c | 5 +++++
>   1 file changed, 5 insertions(+)
>
> diff --git a/drivers/watchdog/da9063_wdt.c b/drivers/watchdog/da9063_wdt.c
> index b2e9201..87e14d8 100644
> --- a/drivers/watchdog/da9063_wdt.c
> +++ b/drivers/watchdog/da9063_wdt.c
> @@ -67,6 +67,11 @@ static int _da9063_wdt_set_timeout(struct da9063_watchdog *wdt,
>   	mutex_lock(&wdt->mutex);
>
>   	ret = regmap_update_bits(da9063->regmap, DA9063_REG_CONTROL_D,
> +				 DA9063_TWDSCALE_MASK, DA9063_TWDSCALE_DISABLE);
> +
> +	usleep_range(150, 300);
> +
> +	ret = regmap_update_bits(da9063->regmap, DA9063_REG_CONTROL_D,
>   				 DA9063_TWDSCALE_MASK, regval);
>
>   	wdt->defer_ping = false;
>

Re: [PATCH 2/2] watchdog: da9063: Disable and wait before changing timeout

[Philipp Zabel]

Am Freitag, den 07.08.2015, 06:40 -0700 schrieb Guenter Roeck:
> On 08/06/2015 09:36 AM, Philipp Zabel wrote:
> > The DA9063 watchdog occasionally enters error condition and resets the
> > system if the timeout is changed quickly after the timer was enabled.
> >
> > The method of disabling and waiting for > 150 µs before setting the
> > new timeout is taken from the DA9052 driver.
> >
> Using sleep concerns me a bit, since during that time the watchdog
> is disabled. Bad enough that we have to do this to start with,
> but using usleep adds more risk to an already bad situation.
> 
> How about udelay() instead ?

Ok, I'll change it.

regards
Philipp

RE: [PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window

[Opensource [Adam Thomson]]

On August 06, 2015 17:37, Philipp Zabel wrote:

> There is a cooldown period after watchdog timer setup, and also after each
> watchdog ping. If a ping is issued too early, the watchdog triggers the
> watchdog error condition and causes a system reset.
> 
> The 256 ms period length is a best guess based on the same value being used
> for the DA9053 predecessor PMIC and experiments that have shown 200 ms are
> not enough to avoid the error.
> 
> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>

My colleague Stephen is out of office for the next two weeks, but really I
think he should review these two patches given his knowledge of the device. If
you can hold off until then, he should be able to provide feedback.

Re: [PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window

[Philipp Zabel]

Hi Adam,

Am Freitag, den 07.08.2015, 10:13 +0000 schrieb Opensource [Adam
Thomson]:
> On August 06, 2015 17:37, Philipp Zabel wrote:
> 
> > There is a cooldown period after watchdog timer setup, and also after each
> > watchdog ping. If a ping is issued too early, the watchdog triggers the
> > watchdog error condition and causes a system reset.
> > 
> > The 256 ms period length is a best guess based on the same value being used
> > for the DA9053 predecessor PMIC and experiments that have shown 200 ms are
> > not enough to avoid the error.
> > 
> > Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
> 
> My colleague Stephen is out of office for the next two weeks, but really I
> think he should review these two patches given his knowledge of the device. If
> you can hold off until then, he should be able to provide feedback.

Absolutely, thank you for the heads-up.

regards
Philipp

RE: [PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window

[Opensource [Steve Twiss]]

On 07 August 2015 11:18, Philipp Zabel wrote: 
> Subject: Re: [PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window
> 
> Hi Adam,
> 
> Am Freitag, den 07.08.2015, 10:13 +0000 schrieb Opensource [Adam Thomson]:
> > On August 06, 2015 17:37, Philipp Zabel wrote:
> >
> > > There is a cooldown period after watchdog timer setup, and also after each
> > > watchdog ping. If a ping is issued too early, the watchdog triggers the
> > > watchdog error condition and causes a system reset.
> > >
> > > The 256 ms period length is a best guess based on the same value being used
> > > for the DA9053 predecessor PMIC and experiments that have shown 200 ms are
> > > not enough to avoid the error.
> > >
> > > Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
> >
> > My colleague Stephen is out of office for the next two weeks, but really I
> > think he should review these two patches given his knowledge of the device. If
> > you can hold off until then, he should be able to provide feedback.
> 
> Absolutely, thank you for the heads-up.

Hi Philipp,

Thanks for waiting until my return on this topic. I will check with the hardware engineers to
see about this DA9063 minimum time.

Regards,
Steve

Re: [PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window

[Guenter Roeck]

On 08/06/2015 09:36 AM, Philipp Zabel wrote:
> There is a cooldown period after watchdog timer setup, and also after each
> watchdog ping. If a ping is issued too early, the watchdog triggers the
> watchdog error condition and causes a system reset.
>
> The 256 ms period length is a best guess based on the same value being used
> for the DA9053 predecessor PMIC and experiments that have shown 200 ms are
> not enough to avoid the error.
>
> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>

Hi Philipp,

instead of using this quite complex approach, would it be possible to use the
same approach as the da9062 driver ?

I understand the downside, which is that the caller may end up sleeping,
but this should not happen too often. Also, I plan to move the enforcement
of a minimal time between heartbeats into the infrastructure, so this would
not least forever.

Thanks,
Guneter

> ---
>   drivers/watchdog/da9063_wdt.c | 88 ++++++++++++++++++++++++++++++++++++++-----
>   1 file changed, 79 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/watchdog/da9063_wdt.c b/drivers/watchdog/da9063_wdt.c
> index e2fe2eb..b2e9201 100644
> --- a/drivers/watchdog/da9063_wdt.c
> +++ b/drivers/watchdog/da9063_wdt.c
> @@ -35,11 +35,15 @@ static const unsigned int wdt_timeout[] = { 0, 2, 4, 8, 16, 32, 65, 131 };
>   #define DA9063_WDT_MIN_TIMEOUT		wdt_timeout[DA9063_TWDSCALE_MIN]
>   #define DA9063_WDT_MAX_TIMEOUT		wdt_timeout[DA9063_TWDSCALE_MAX]
>   #define DA9063_WDG_TIMEOUT		wdt_timeout[3]
> +#define DA9063_TWDMIN			256
>
>   struct da9063_watchdog {
>   	struct da9063 *da9063;
>   	struct watchdog_device wdtdev;
>   	struct notifier_block restart_handler;
> +	struct delayed_work ping_work;
> +	struct mutex mutex;
> +	bool defer_ping;
>   };
>
>   static unsigned int da9063_wdt_timeout_to_sel(unsigned int secs)
> @@ -54,10 +58,23 @@ static unsigned int da9063_wdt_timeout_to_sel(unsigned int secs)
>   	return DA9063_TWDSCALE_MAX;
>   }
>
> -static int _da9063_wdt_set_timeout(struct da9063 *da9063, unsigned int regval)
> +static int _da9063_wdt_set_timeout(struct da9063_watchdog *wdt,
> +				   unsigned int regval)
>   {
> -	return regmap_update_bits(da9063->regmap, DA9063_REG_CONTROL_D,
> -				  DA9063_TWDSCALE_MASK, regval);
> +	struct da9063 *da9063 = wdt->da9063;
> +	int ret;
> +
> +	mutex_lock(&wdt->mutex);
> +
> +	ret = regmap_update_bits(da9063->regmap, DA9063_REG_CONTROL_D,
> +				 DA9063_TWDSCALE_MASK, regval);
> +
> +	wdt->defer_ping = false;
> +	schedule_delayed_work(&wdt->ping_work, msecs_to_jiffies(DA9063_TWDMIN));
> +
> +	mutex_unlock(&wdt->mutex);
> +
> +	return ret;
>   }
>
>   static int da9063_wdt_start(struct watchdog_device *wdd)
> @@ -67,12 +84,16 @@ static int da9063_wdt_start(struct watchdog_device *wdd)
>   	int ret;
>
>   	selector = da9063_wdt_timeout_to_sel(wdt->wdtdev.timeout);
> -	ret = _da9063_wdt_set_timeout(wdt->da9063, selector);
> -	if (ret)
> +	ret = _da9063_wdt_set_timeout(wdt, selector);
> +	if (ret) {
>   		dev_err(wdt->da9063->dev, "Watchdog failed to start (err = %d)\n",
>   			ret);
> +		return ret;
> +	}
>
> -	return ret;
> +	wdd->timeout = wdt_timeout[selector];
> +
> +	return 0;
>   }
>
>   static int da9063_wdt_stop(struct watchdog_device *wdd)
> @@ -80,6 +101,8 @@ static int da9063_wdt_stop(struct watchdog_device *wdd)
>   	struct da9063_watchdog *wdt = watchdog_get_drvdata(wdd);
>   	int ret;
>
> +	cancel_delayed_work_sync(&wdt->ping_work);
> +
>   	ret = regmap_update_bits(wdt->da9063->regmap, DA9063_REG_CONTROL_D,
>   				 DA9063_TWDSCALE_MASK, DA9063_TWDSCALE_DISABLE);
>   	if (ret)
> @@ -89,9 +112,8 @@ static int da9063_wdt_stop(struct watchdog_device *wdd)
>   	return ret;
>   }
>
> -static int da9063_wdt_ping(struct watchdog_device *wdd)
> +static int da9063_wdt_issue_ping(struct da9063_watchdog *wdt)
>   {
> -	struct da9063_watchdog *wdt = watchdog_get_drvdata(wdd);
>   	int ret;
>
>   	ret = regmap_write(wdt->da9063->regmap, DA9063_REG_CONTROL_F,
> @@ -100,9 +122,45 @@ static int da9063_wdt_ping(struct watchdog_device *wdd)
>   		dev_alert(wdt->da9063->dev, "Failed to ping the watchdog (err = %d)\n",
>   			  ret);
>
> +	wdt->defer_ping = false;
> +	schedule_delayed_work(&wdt->ping_work, msecs_to_jiffies(DA9063_TWDMIN));
> +
>   	return ret;
>   }
>
> +static int da9063_wdt_ping(struct watchdog_device *wdd)
> +{
> +	struct da9063_watchdog *wdt = watchdog_get_drvdata(wdd);
> +	int ret = 0;
> +
> +	mutex_lock(&wdt->mutex);
> +
> +	if (delayed_work_pending(&wdt->ping_work)) {
> +		wdt->defer_ping = true;
> +
> +		goto out;
> +	}
> +
> +	da9063_wdt_issue_ping(wdt);
> +out:
> +	mutex_unlock(&wdt->mutex);
> +
> +	return ret;
> +}
> +
> +static void da9063_wdt_ping_work(struct work_struct *work)
> +{
> +	struct da9063_watchdog *wdt = container_of(to_delayed_work(work),
> +					struct da9063_watchdog, ping_work);
> +
> +	mutex_lock(&wdt->mutex);
> +
> +	if (wdt->defer_ping)
> +		da9063_wdt_issue_ping(wdt);
> +
> +	mutex_unlock(&wdt->mutex);
> +}
> +
>   static int da9063_wdt_set_timeout(struct watchdog_device *wdd,
>   				  unsigned int timeout)
>   {
> @@ -111,7 +169,7 @@ static int da9063_wdt_set_timeout(struct watchdog_device *wdd,
>   	int ret;
>
>   	selector = da9063_wdt_timeout_to_sel(timeout);
> -	ret = _da9063_wdt_set_timeout(wdt->da9063, selector);
> +	ret = _da9063_wdt_set_timeout(wdt, selector);
>   	if (ret)
>   		dev_err(wdt->da9063->dev, "Failed to set watchdog timeout (err = %d)\n",
>   			ret);
> @@ -178,6 +236,9 @@ static int da9063_wdt_probe(struct platform_device *pdev)
>
>   	wdt->wdtdev.status = WATCHDOG_NOWAYOUT_INIT_STATUS;
>
> +	INIT_DELAYED_WORK(&wdt->ping_work, da9063_wdt_ping_work);
> +	mutex_init(&wdt->mutex);
> +
>   	watchdog_set_drvdata(&wdt->wdtdev, wdt);
>   	dev_set_drvdata(&pdev->dev, wdt);
>
> @@ -202,13 +263,22 @@ static int da9063_wdt_remove(struct platform_device *pdev)
>   	unregister_restart_handler(&wdt->restart_handler);
>
>   	watchdog_unregister_device(&wdt->wdtdev);
> +	cancel_delayed_work_sync(&wdt->ping_work);
>
>   	return 0;
>   }
>
> +static void da9063_wdt_shutdown(struct platform_device *pdev)
> +{
> +	struct da9063_watchdog *wdt = dev_get_drvdata(&pdev->dev);
> +
> +	cancel_delayed_work_sync(&wdt->ping_work);
> +}
> +
>   static struct platform_driver da9063_wdt_driver = {
>   	.probe = da9063_wdt_probe,
>   	.remove = da9063_wdt_remove,
> +	.shutdown = da9063_wdt_shutdown,
>   	.driver = {
>   		.name = DA9063_DRVNAME_WATCHDOG,
>   	},
>

Re: [PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window

[Philipp Zabel]

Hi Guenter,

Am Freitag, den 07.08.2015, 06:36 -0700 schrieb Guenter Roeck:
> On 08/06/2015 09:36 AM, Philipp Zabel wrote:
> > There is a cooldown period after watchdog timer setup, and also after each
> > watchdog ping. If a ping is issued too early, the watchdog triggers the
> > watchdog error condition and causes a system reset.
> >
> > The 256 ms period length is a best guess based on the same value being used
> > for the DA9053 predecessor PMIC and experiments that have shown 200 ms are
> > not enough to avoid the error.
> >
> > Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
> 
> Hi Philipp,
> 
> instead of using this quite complex approach, would it be possible to use the
> same approach as the da9062 driver ?
> 
> I understand the downside, which is that the caller may end up sleeping,
> but this should not happen too often.

Depending on the userspace this can happen quite a lot ...

>  Also, I plan to move the enforcement
> of a minimal time between heartbeats into the infrastructure, so this would
> not least forever.

... but certainly taking care of this in the framework is better than
doing it in the drivers.

regards
Philipp

Re: [PATCH 1/2] watchdog: da9063: Ping watchdog only during allowed time window

[Guenter Roeck]

Hi Philipp,

On 08/07/2015 10:08 AM, Philipp Zabel wrote:
> Hi Guenter,
>
> Am Freitag, den 07.08.2015, 06:36 -0700 schrieb Guenter Roeck:
>> On 08/06/2015 09:36 AM, Philipp Zabel wrote:
>>> There is a cooldown period after watchdog timer setup, and also after each
>>> watchdog ping. If a ping is issued too early, the watchdog triggers the
>>> watchdog error condition and causes a system reset.
>>>
>>> The 256 ms period length is a best guess based on the same value being used
>>> for the DA9053 predecessor PMIC and experiments that have shown 200 ms are
>>> not enough to avoid the error.
>>>
>>> Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
>>
>> Hi Philipp,
>>
>> instead of using this quite complex approach, would it be possible to use the
>> same approach as the da9062 driver ?
>>
>> I understand the downside, which is that the caller may end up sleeping,
>> but this should not happen too often.
>
> Depending on the userspace this can happen quite a lot ...
>
Your call, ultimately. I don't want to spend time reviewing the changes
as is, given they are quite substantial and will ultimately go away,
so we'll just have to wait for someone else to spend that time.

Thanks,
Guenter

>>   Also, I plan to move the enforcement
>> of a minimal time between heartbeats into the infrastructure, so this would
>> not least forever.
>
> ... but certainly taking care of this in the framework is better than
> doing it in the drivers.
>
> regards
> Philipp
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-watchdog" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>