WD_PASSIVE kernel based tickling of the watchdog - request for ideas

From: Nick Hibma <nick_at_van-laarhoven.org>
Date: Fri, 15 Dec 2006 23:16:18 +0100 (CET)
I've got two requests for a clue ... er... ideas today on how to 
implement the WD_PASSIVE flag in the watchdog(9) interface:

- How would I best implement the tickling of the watchdog at specified 
intervals (e.g. 1/2 the watchdog timeout value)?

By using a

 	- timeout
 	- kernel thread sleeping most of the time
 	- hardclock()

considering the trade-off between large variance in frequency vs. making 
sure we notice a (permanent) freeze in the kernel somewhere. For example 
burning a CD sometimes makes my laptop freeze for a short period. Will 
the watchdog fire in that case?

- What do we consider a 'bad' situation in which the watchdog should not 
be tickled? What kind of checks would we need to perform?

I'd be interested to hear your thoughts.

Nick

P.S.: I'm personally not interested in passive tickling of the watchdog, 
so if there is no response, I'll leave the implementation as is (return 
EOPNOTSUPP at the moment).
Received on Fri Dec 15 2006 - 21:16:22 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:04 UTC