This is the mail archive of the
ecos-patches@sourceware.org
mailing list for the eCos project.
Re: AT91 watchdog timer controller driver
- From: David Vrabel <dvrabel at arcom dot com>
- To: Andrew Lunn <andrew at lunn dot ch>
- Cc: eCos Patches <ecos-patches at ecos dot sourceware dot org>
- Date: Mon, 20 Feb 2006 09:46:47 +0000
- Subject: Re: AT91 watchdog timer controller driver
- References: <20060219193110.GE19406@lunn.ch>
Andrew Lunn wrote:
> As i said in my last email, the AT91 system controller has a different
> watchdog timer than the older devices. This patch contains a driver
> for this new hardware.
>
> A couple of points to note:
>
> 1) The watchdog timer is driven by the slow clock of the AT91. This is
> an LC oscillator on the chip. The accuracy of this oscillator is
> dependent on silicon process at manufacturing time and the
> temperature. I found that my development board a watchdog timeout of
> 1000ms actually goes off at around 550ms at room temperature. So be
> generous with your timeouts. This also means that one of the test case
> fails because the timer goes off too early.
Is it possible to measure the rate of this clock at boot time? By
comparing its rate against that of a more accurate clock, for example.
If not, I think you should consider adding a sufficiently a large fudge
factor so that the watchdog never goes off early. Presumably the
datasheet provides minimum and maximum values for the rate of this slow
clock?
David Vrabel
--
David Vrabel, Design Engineer
Arcom, Clifton Road Tel: +44 (0)1223 411200 ext. 3233
Cambridge CB1 7EA, UK Web: http://www.arcom.com/