linux-kernel - Re: [PATCHv3 0/3] watchdog: add f71862fg support

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]

Date:	Sat, 2 Oct 2010 14:33:09 +0200
From:	Wim Van Sebroeck <wim@...ana.be>
To:	Giel van Schijndel <me@...tis.eu>
Cc:	Lutz Ballaschke <vegan.grindcore@...glemail.com>,
	Wim van Sebroeck <wim@...ana.be>,
	linux-watchdog@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCHv3 0/3] watchdog: add f71862fg support

Hi all,

> On Sun, Sep 26, 2010 at 04:06:29PM +0200, Lutz Ballaschke wrote:
> > Furthermore i was thinking about modifying the WDIOC_GETTIMELEFT
> > helper function since it simply returns the content of the WDT timer 
> > register without regarding whether the driver runs in minute-mode or not
> > (thanks to Giel for pointing that out). This means if the driver is loaded 
> > with a timeout of >255 seconds (and switches to minute-mode) WDIOC_GETTIMELEFT
> > ioctl returns minutes instead of seconds and thus doesn't comply with 
> > watchdog-api (e.g. ioctl returns 6(min) not 360(sec)). It might be 
> > confusing if one loads the driver with a timeout of let's say 360 seconds 
> > and gets 6 minutes returned by ioctl. But imho it's not very accurate to
> > return a value of seconds if the wdt doesn't count in seconds. That might
> > be even more confusing. This inconsistency on the part of the hardware 
> > maybe can't get satisfyingly compensated by the driver. So i'd prefer the 
> > most simple solution.
> 
> IMO the worst option would be to just return minutes.  Returning
> 'minutes * 60 = seconds' would be significantly less bad.  So if these
> are your options I'd suggest going with the lesser evil.
> 
> One other possible option could be returning '(minutes - 1) * 60 + 1'.
> That way bad timing decisions based on the (wrong) assumption, that
> there are still 60 seconds left until reset, can be averted.
> 
> I would however like Wim's opinion on this, as this affects the
> Watchdog API's usage.

The return value should be the seconds before the watchdog will reboot.
So value should be seconds, not minutes and is best the safest value
(so (minutes - 1) * 60 + 1 in this case ).
Secondly: the WDIOC_GETTIMELEFT should only be a DEBUGging ioctl.
The reason I created it, is to check if certain drivers are working
or not. We should not use it in normal operations imho.

Kind regards,
Wim.

--
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/