linux-kernel - Re: [PATCH] watchdog: make DesignWare watchdog allow users to set bigger timeout value

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]

Message-ID: <b0fcaae6-c9da-f551-5af4-ff112b08e851@roeck-us.net>
Date:   Wed, 20 Nov 2019 19:40:58 -0800
From:   Guenter Roeck <linux@...ck-us.net>
To:     "Wang, Peng 1. (NSB - CN/Hangzhou)" <peng.1.wang@...ia-sbell.com>
Cc:     "wim@...ux-watchdog.org" <wim@...ux-watchdog.org>,
        "linux-watchdog@...r.kernel.org" <linux-watchdog@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] watchdog: make DesignWare watchdog allow users to set
 bigger timeout value

On 11/20/19 5:29 PM, Wang, Peng 1. (NSB - CN/Hangzhou) wrote:
> Hi Guenter,
> 
> Thank you for your time.
> - I will remove the unnecessary {}
> - wdd->max_hw_heartbeat_ms is the max timeout value which HW can support, this value is limited according to the input clock, say. It only supports 20 seconds, if users requires to set timeout to be say. 60 seconds, the watchdog device driver 'watchdog_dev.c' checks if wdd->timeout is bigger than wdd->max_hw_heartbeat_ms, if yes, watchdog_dev.c feeds the watchdog by a worker queue itself to help to feed the watchdog before 60 seconds elapse. Here the issue of dw_wdt.c is that, the original codes update wdd->timeout to the value which HW can support, which means if users requires 60 seconds to be the timeout, then dw_wdt.c updates the timeout value to 20 seconds, this makes the "feeding helper" mechanism in watchdog_dev.c not take effect. That's why I add this check.
> 

Yes, I understand you need a check. What I am saying is that the check is wrong.

You need something like

	if (top_s > DW_WDT_MAX_TOP)
		wdt->timeout = top_s;
	else
		wdt->timeout = dw_wdt_top_in_seconds(dw_wdt, top_val);

Guenter

> Thanks,
> Peng Wang
> 
> -----Original Message-----
> From: Guenter Roeck [mailto:groeck7@...il.com] On Behalf Of Guenter Roeck
> Sent: Thursday, November 21, 2019 1:15 AM
> To: Wang, Peng 1. (NSB - CN/Hangzhou) <peng.1.wang@...ia-sbell.com>
> Cc: wim@...ux-watchdog.org; linux-watchdog@...r.kernel.org; linux-kernel@...r.kernel.org
> Subject: Re: [PATCH] watchdog: make DesignWare watchdog allow users to set bigger timeout value
> 
> On Wed, Nov 20, 2019 at 10:07:57AM +0000, Wang, Peng 1. (NSB - CN/Hangzhou) wrote:
>>  From 1d051b7c081083751dc0bab97d3ab9efbba0f4a7 Mon Sep 17 00:00:00 2001
>> From: Peng Wang <peng.1.wang@...ia-sbell.com>
>> Date: Wed, 20 Nov 2019 15:12:59 +0800
>> Subject: [PATCH] watchdog: make DesignWare watchdog allow users to set
>> bigger  timeout value
>>
>> watchdog_dev.c provides means to allow users to set bigger timeout
>> value than HW can support, make DesignWare watchdog align with this.
>>
>> Signed-off-by: Peng Wang <peng.1.wang@...ia-sbell.com>
>> ---
>>   drivers/watchdog/dw_wdt.c | 11 +++++++++--
>>   1 file changed, 9 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/watchdog/dw_wdt.c b/drivers/watchdog/dw_wdt.c
>> index fef7c61..8911e5e 100644
>> --- a/drivers/watchdog/dw_wdt.c
>> +++ b/drivers/watchdog/dw_wdt.c
>> @@ -113,8 +113,15 @@ static int dw_wdt_set_timeout(struct watchdog_device *wdd, unsigned int top_s)
>>   	 */
>>   	writel(top_val | top_val << WDOG_TIMEOUT_RANGE_TOPINIT_SHIFT,
>>   	       dw_wdt->regs + WDOG_TIMEOUT_RANGE_REG_OFFSET);
>> -
>> -	wdd->timeout = dw_wdt_top_in_seconds(dw_wdt, top_val);
>> +
>> +    /*
>> +     * In case users set bigger timeout value than HW can support,
>> +     * kernel(watchdog_dev.c) helps to feed watchdog before
>> +     * wdd->timeout
>> +     */
>> +    if ( wdd->timeout * 1000 <= wdd->max_hw_heartbeat_ms ) {
>> +	    wdd->timeout = dw_wdt_top_in_seconds(dw_wdt, top_val);
>> +    }
> 
> { } is unnecessary here. Also, the above code compares the _old_ timeout againt the maximum supported timeout, which doesn't look correct.
> 
> Thanks,
> Guenter
> 
>>   
>>   	return 0;
>>   }
>> --
>> 1.8.3.1
>>
>