Date:	Thu, 2 Apr 2015 19:16:06 -0700
From:	Andrew Bresticker <abrestic@...omium.org>
To:	Guenter Roeck <linux@...ck-us.net>
Cc:	James Hogan <james.hogan@...tec.com>,
	Wim Van Sebroeck <wim@...ana.be>,
	"linux-watchdog@...r.kernel.org" <linux-watchdog@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	Ezequiel Garcia <ezequiel.garcia@...tec.com>
Subject: Re: [PATCH V2 1/3] watchdog: imgpdc: Allow timeout to be set in device-tree

Hi Guenter,

On Thu, Apr 2, 2015 at 6:52 PM, Guenter Roeck <linux@...ck-us.net> wrote:
> On 04/02/2015 09:46 AM, Andrew Bresticker wrote:
>>
>> On Wed, Apr 1, 2015 at 6:08 PM, Guenter Roeck <linux@...ck-us.net> wrote:
>>>
>>> On 04/01/2015 03:22 PM, James Hogan wrote:
>>>>
>>>>
>>>> Hi Andrew,
>>>>
>>>> On Wed, Apr 01, 2015 at 10:43:14AM -0700, Andrew Bresticker wrote:
>>>>>
>>>>>
>>>>> Since the heartbeat is statically initialized to its default value,
>>>>> watchdog_init_timeout() will never look in the device-tree for a
>>>>> timeout-sec value.  Instead of statically initializing heartbeat,
>>>>> fall back to the default timeout value if watchdog_init_timeout()
>>>>> fails.
>>>>
>>>>
>>>>
>>>> Whoops. Sorry about that. I wasn't aware that a timeout-sec value was
>>>> expected. It isn't mentioned in the DT binding documentation for this
>>>> device :-(.
>>>>
>>>>>
>>>>> Signed-off-by: Andrew Bresticker <abrestic@...omium.org>
>>>>> Cc: Ezequiel Garcia <ezequiel.garcia@...tec.com>
>>>>> Cc: James Hogan <james.hogan@...tec.com>
>>>>> ---
>>>>> New for v2.
>>>>> ---
>>>>>    drivers/watchdog/imgpdc_wdt.c | 6 +++---
>>>>>    1 file changed, 3 insertions(+), 3 deletions(-)
>>>>>
>>>>> diff --git a/drivers/watchdog/imgpdc_wdt.c
>>>>> b/drivers/watchdog/imgpdc_wdt.c
>>>>> index 0deaa4f..89b2abc 100644
>>>>> --- a/drivers/watchdog/imgpdc_wdt.c
>>>>> +++ b/drivers/watchdog/imgpdc_wdt.c
>>>>> @@ -42,7 +42,7 @@
>>>>>    #define PDC_WDT_MIN_TIMEOUT           1
>>>>>    #define PDC_WDT_DEF_TIMEOUT           64
>>>>>
>>>>> -static int heartbeat = PDC_WDT_DEF_TIMEOUT;
>>>>> +static int heartbeat;
>>>>>    module_param(heartbeat, int, 0);
>>>>>    MODULE_PARM_DESC(heartbeat, "Watchdog heartbeats in seconds "
>>>>>          "(default=" __MODULE_STRING(PDC_WDT_DEF_TIMEOUT) ")");
>>>>> @@ -195,9 +195,9 @@ static int pdc_wdt_probe(struct platform_device
>>>>> *pdev)
>>>>>
>>>>>          ret = watchdog_init_timeout(&pdc_wdt->wdt_dev, heartbeat,
>>>>> &pdev->dev);
>>>>>          if (ret < 0) {
>>>>> -               pdc_wdt->wdt_dev.timeout =
>>>>> pdc_wdt->wdt_dev.max_timeout;
>>>>> +               pdc_wdt->wdt_dev.timeout = PDC_WDT_DEF_TIMEOUT;
>>>>
>>>>
>>>>
>>>> The watchdog_init_timeout kerneldoc comment suggests that the old value
>>>> should be the default timeout, i.e. that timeout should be set to
>>>> PDC_WDT_DEF_TIMEOUT before calling watchdog_init_timeout, rather than
>>>> whenever ret < 0.
>>>>
>>>> Indeed, if heartbeat is set to an invalid non-zero value,
>>>> watchdog_init_timeout will still try and set timeout from DT, but also
>>>> still returns -EINVAL regardless of whether that succeeds, and this
>>>> would incorrectly override the timeout from DT with the hardcoded
>>>> default.
>>>>
>>>>>                  dev_warn(&pdev->dev,
>>>>> -                        "Initial timeout out of range! setting max
>>>>> timeout\n");
>>>>> +                        "Initial timeout out of range! setting default
>>>>> timeout\n");
>>>>
>>>>
>>>>
>>>> It feels wrong for a presumably safe & normal situation (i.e. no default
>>>> in DT, which arguably shouldn't contain policy anyway) to show a
>>>> warning, but it can also show due to an invalid module parameter (or
>>>> invalid DT property) which is most definitely justified.
>>>>
>>>
>>> Agreed. I would suggest to leave that part alone and set the default
>>> prior
>>> to calling watchdog_init_timeout().
>>
>>
>> Yes, but I think James' concern here was that we'd now get a
>> dev_warn() in the normal case where no timeout is specified via module
>> parameter or DT.
>>
> My understanding is that watchdog_init_timeout only returns an error if
> the second parameter is not 0 and invalid, or if the timeout-sec property
> has been provided and is invalid. I am not entirely sure I understand
> why you think this is a problem. Can you please explain ?

Unless I've gone completely insane, I'm pretty sure this will return
-EINVAL if timeout_parm is 0 and timeout-sec is not present:

int watchdog_init_timeout(struct watchdog_device *wdd,
                                unsigned int timeout_parm, struct device *dev)
{
        unsigned int t = 0;
        int ret = 0;

        watchdog_check_min_max_timeout(wdd);

        /* try to get the timeout module parameter first */
        if (!watchdog_timeout_invalid(wdd, timeout_parm) && timeout_parm) {
                wdd->timeout = timeout_parm;
                return ret;
        }
        if (timeout_parm)
                ret = -EINVAL;

        /* try to get the timeout_sec property */
        if (dev == NULL || dev->of_node == NULL)
                return ret;
        of_property_read_u32(dev->of_node, "timeout-sec", &t);
        if (!watchdog_timeout_invalid(wdd, t) && t)
                wdd->timeout = t;
        else
                ret = -EINVAL;

        return ret;
}

That said, the behavior you describe makes more sense, so perhaps
watchdog_init_timeout() should be updated to match.

Thanks,
Andrew
--
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/

Powered by blists - more mailing lists