linux-kernel - Re: [PATCH V6 1/2] watchdog: mtk: support pre-timeout when the bark irq is available

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] [day] [month] [year] [list]

Message-ID: <56f16371-5342-5c71-2393-41258cecb516@roeck-us.net>
Date:   Thu, 22 Apr 2021 06:56:27 -0700
From:   Guenter Roeck <linux@...ck-us.net>
To:     王擎 <wangqing@...o.com>
Cc:     Wim Van Sebroeck <wim@...ux-watchdog.org>,
        Rob Herring <robh+dt@...nel.org>,
        Matthias Brugger <matthias.bgg@...il.com>,
        linux-watchdog@...r.kernel.org, devicetree@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org,
        linux-mediatek@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH V6 1/2] watchdog: mtk: support pre-timeout when the bark
 irq is available

On 4/22/21 12:05 AM, 王擎 wrote:
> 
>> On 4/21/21 8:46 PM, 王擎 wrote:
>>>
>>>> On 4/21/21 7:45 PM, Wang Qing wrote:
>>>>> Use the bark interrupt as the pretimeout notifier if available.
>>>>>
>>>>> When the watchdog timer expires in dual mode, an interrupt will be
>>>>> triggered first, then the timing restarts. The reset signal will be
>>>>> initiated when the timer expires again.
>>>>>
>>>>> The pretimeout notification shall occur at timeout-sec/2.
>>>>>
>>>>> V2:
>>>>> - panic() by default if WATCHDOG_PRETIMEOUT_GOV is not enabled.
>>>>>
>>>>> V3:
>>>>> - Modify the pretimeout behavior, manually reset after the pretimeout
>>>>> - is processed and wait until timeout.
>>>>>
>>>>> V4:
>>>>> - Remove pretimeout related processing. 
>>>>> - Add dual mode control separately.
>>>>>
>>>>> V5:
>>>>> - Fix some formatting and printing problems.
>>>>>
>>>>> V6:
>>>>> - Realize pretimeout processing through dualmode.
>>>>>
>>>>> Signed-off-by: Wang Qing <wangqing@...o.com>
>>>>> ---
>>>>>  drivers/watchdog/mtk_wdt.c | 53 +++++++++++++++++++++++++++++++++++++++++-----
>>>>>  1 file changed, 48 insertions(+), 5 deletions(-)
>>>>>
>>>>> diff --git a/drivers/watchdog/mtk_wdt.c b/drivers/watchdog/mtk_wdt.c
>>>>> index 97ca993..ebc648b
>>>>> --- a/drivers/watchdog/mtk_wdt.c
>>>>> +++ b/drivers/watchdog/mtk_wdt.c
>>>>> @@ -25,6 +25,7 @@
>>>>>  #include <linux/reset-controller.h>
>>>>>  #include <linux/types.h>
>>>>>  #include <linux/watchdog.h>
>>>>> +#include <linux/interrupt.h>
>>>>>  
>>>>>  #define WDT_MAX_TIMEOUT		31
>>>>>  #define WDT_MIN_TIMEOUT		1
>>>>> @@ -184,15 +185,22 @@ static int mtk_wdt_set_timeout(struct watchdog_device *wdt_dev,
>>>>>  {
>>>>>  	struct mtk_wdt_dev *mtk_wdt = watchdog_get_drvdata(wdt_dev);
>>>>>  	void __iomem *wdt_base = mtk_wdt->wdt_base;
>>>>> +	unsigned int timeout_interval;
>>>>>  	u32 reg;
>>>>>  
>>>>> -	wdt_dev->timeout = timeout;
>>>>> +	timeout_interval = wdt_dev->timeout = timeout;
>>>>> +	/*
>>>>> +	 * In dual mode, irq will be triggered at timeout/2
>>>>> +	 * the real timeout occurs at timeout
>>>>> +	 */
>>>>> +	if (wdt_dev->pretimeout)
>>>>> +		timeout_interval = wdt_dev->pretimeout = timeout/2;
>>>>
>>>> Please run checkpatch --strict and fix what it reports.
>>>> Also, there should be a set_pretimeout function to set the
>>>> pretimeout. It is ok to update it here, but it should be set
>>>> in its own function to make sure that the actual value
>>>> is reported back to userspace.
>>>>
>>>> Thanks,
>>>> Guenter
>>>
>>> The reason why the set_pretimeout interface is not provided is 
>>> because the pretimeout is fixed after the timeout is set,  we need
>>> to modify timeout after setting pretimeout, which is puzzling.
>>>
>>
>> What you need to do is to set pretimeout = timeout / 2 if a pretimeout
>> is set to a value != 0. Just like we adjust timeout to valid values
>> when set, we adjust pretimeout as well. I don't see a problem with that.
>>
>> Guenter
> 
> Thanks, Guenter. But this will complicate the situation:
> First, set_pretimeout will become an interface for dynamically enable and
> disable the pre-timeout func, instead of adjusting the pretimeout time. 
> 
Effectively yes. That is what it is, based on its limitations. That is
not a problem, and in true for every pretimeout function. Set it to 0,
and it is turned off. Set it to a value other than 0, and it is turned on.

> Secondly, when the irq is not registered, the user cannot be allowed to set
> the pretimeout to non-zero. When irq is registered, it doesn't make any sense
> to turn off pre-timeout func. 
> 

That is your opinion. It is still a user decision to turn it on or off,
just like it is a user decision to set the timeout to a specific value
or to enable the watchdog in the first place. There is no reason to
make it mandatory just because an interrupt has been provided
(or, rather, connected).

Also, if the interrupt is not provided, WDIOF_PRETIMEOUT is not set,
and trying to set the pretimeout would return -EOPNOTSUPP.

Guenter