Re: [PATCH 3/3] CBUS: Make retu watchdog behave like a standard Linux watchdog

[Tony Lindgren <tony@atomide.com>]

* Igor Stoppa <igor.stoppa@nokia.com> [080512 17:56]:
> Hi Tony,
> On Mon, 2008-05-12 at 17:35 -0700, ext Tony Lindgren wrote:
> > Make retu watchdog behave like a standard Linux watchdog.
> > 
> > Let the kernel do the kicking until the watchdog device is opened.
> 
> This is not always the desidered behavior: the powerdown wd is used to
> ensure that the whole sw stack is healty: doing the kicking in
> kernelspace for free introduces the case where userspace can get stuck
> and the device does not powerdown.

That's why there's CONFIG_WATCHDOG_NOWAYOUT where the ping timer is
not enabled at all, and the watchdog is just set to max until userspace
watchdog software kicking starts.

> Also the unconditional loading of the maximum value during probe is not
> aligned with the original reset logic, which was to have the powerdown
> wd to allow for 2 boot attempts:
> 
> -cold boot -> load max value in retu wd (63s)
>            -> load 30s in omap wd
> -try to kick both wds

Well you can set those values via /dev/watchdog too, right?
And then use CONFIG_WATCHDOG_NOWAYOUT.

And ff CONFIG_WATCHDOG_NOWAYOUT is not set, the kernel ping timer only
happens when /dev/watchdog is not open.

> if fail, then omap reboots, but retu keeps counting down
> 
> -warm boot -> let the retu wd untouched
>            -> load 30s in omap wd
> -try to kick both wds
> 
> if fail, retu powers down
> 
> That was the original idea in 770 times and i still like it.
> 
> To conclude, i'd see inkernel kicking more as a debugging feature while
> one is hacking at the kernel than a desirable quality of a stable
> kernel.

Considering that the /dev/watchdog interface is the standard, I see this
patch as the only way we can get this code ever merged upstream. And it's
easy to patch back the non-standard if you want to.

Tony
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html