linux-kernel - Re: [PATCH 2/2] watchdog: gpio_wdt: add "start-at-boot" feature

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 linux-cve-announce 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, 22 Apr 2021 18:28:40 +0200
From:   Francesco Zanella <francesco.zanella@...ar.com>
To:     Guenter Roeck <linux@...ck-us.net>
Cc:     linux-watchdog@...r.kernel.org, wim@...ux-watchdog.org,
        devicetree@...r.kernel.org, robh+dt@...nel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/2] watchdog: gpio_wdt: add "start-at-boot" feature



On 21/04/21 18:42, Guenter Roeck wrote:
> On Wed, Apr 21, 2021 at 06:26:21PM +0200, Francesco Zanella wrote:
>> If "start-at-boot" property is present in the device tree, start pinging
>> hw watchdog at probe, in order to take advantage of kernel configs:
>> - WATCHDOG_HANDLE_BOOT_ENABLED: Avoid possible reboot if hw watchdog was
>>   been enabled before the kernel (by uboot for example) and userspace
>>   doesn't take control of /dev/watchdog in time;
>> - WATCHDOG_OPEN_TIMEOUT: Reboot if userspace doesn't take control of
>>   /dev/watchdog within the timeout.
>>
>> Signed-off-by: Francesco Zanella <francesco.zanella@...ar.com>
>> ---
>>  drivers/watchdog/gpio_wdt.c | 6 +++++-
>>  1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/watchdog/gpio_wdt.c b/drivers/watchdog/gpio_wdt.c
>> index 0923201ce874..1e6f0322ab7a 100644
>> --- a/drivers/watchdog/gpio_wdt.c
>> +++ b/drivers/watchdog/gpio_wdt.c
>> @@ -31,6 +31,7 @@ struct gpio_wdt_priv {
>>  	struct gpio_desc	*gpiod;
>>  	bool			state;
>>  	bool			always_running;
>> +	bool			start_at_boot;
>>  	unsigned int		hw_algo;
>>  	struct watchdog_device	wdd;
>>  };
>> @@ -147,6 +148,9 @@ static int gpio_wdt_probe(struct platform_device *pdev)
>>  	priv->always_running = of_property_read_bool(np,
>>  						     "always-running");
>>  
>> +	priv->start_at_boot = of_property_read_bool(np,
>> +						    "start-at-boot");
>> +
>>  	watchdog_set_drvdata(&priv->wdd, priv);
>>  
>>  	priv->wdd.info		= &gpio_wdt_ident;
>> @@ -161,7 +165,7 @@ static int gpio_wdt_probe(struct platform_device *pdev)
>>  
>>  	watchdog_stop_on_reboot(&priv->wdd);
>>  
>> -	if (priv->always_running)
>> +	if (priv->always_running || priv->start_at_boot)
>>  		gpio_wdt_start(&priv->wdd);
> 
> So the only real difference to always_running is that always_running
> doesn't stop the watchdog on close but keeps it running.
> 
> Does that really warrant another property ? Why not just use
> always-running ?
> 
> The special use case of being able to stop the watchdog doesn't seem
> to be worth the trouble. Please explain your use case.
> 
> Guenter
> 

You got the point.
I would like to be able to stop the watchdog when I want.
>From my point of view always_running is used in the very special
case when the wdt chip can't be stopped.
I want a normal wdt that can be stopped (for example during a system
update), but I want it to start at boot for the 2 reasons that I
explained before:
- I need continuity in hw wdt pinging because I start in uboot,
  so I take advantage of WATCHDOG_HANDLE_BOOT_ENABLED that makes
  the kernel to do that job; but it is triggered only if it is
  started at probe, if I'm not wrong.
- I would like to monitor with the wdt also the kernel at boot,
  and more importantly handle the case when the userspace app
  doesn't take control of /dev/watchdog for whatever reason
  within the timeout set with WATCHDOG_OPEN_TIMEOUT.

Hope that this explain my target.
Thanks for your attention,

Francesco