Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 12 Jul 2007 09:51:35 +0100
From:      David Malone <dwmalone@maths.tcd.ie>
To:        Ivan Voras <ivoras@fer.hr>
Cc:        freebsd-hackers@freebsd.org, freebsd-current@freebsd.org
Subject:   Re: Debugging times
Message-ID:  <20070712085135.GA5332@walton.maths.tcd.ie>
In-Reply-To: <f73koj$a13$1@sea.gmane.org>
References:  <f6u94s$v6o$1@sea.gmane.org> <20070709214216.GA72912@walton.maths.tcd.ie> <f6ucs8$d6t$1@sea.gmane.org> <20070711132202.GA95487@walton.maths.tcd.ie> <f73koj$a13$1@sea.gmane.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Jul 12, 2007 at 12:14:43AM +0200, Ivan Voras wrote:
> I've got interesting results (in the bad sense of the phrase): I do get
> the message "Invalid time in real time clock. Check and reset the time
> immediately" (the i386 message) BUT my time gets reset to 0 (midnight
> 1970.)

Ah - that's interesting. Could you look for the comment in
src/sys/i386/isa/clock.c that says:

	/* Should we set dow = -1 because some clocks don't set it correctly? */

and add a line afterwards to say:

	ct.dow = -1;

and see if that helps?

> I see your patch and it shouldn't do that. Could it be a compiler bug,
> so the effects change after trivial code has been changed?

I think that without the patch, the clock is initialised using a
chunk of uninitialised memory on the stack, which could result in
all sorts of random values being used. With the patch, things may
be a little more deterministic.

	David.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070712085135.GA5332>