Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 23 Oct 2008 15:10:33 -0400
From:      John Baldwin <jhb@freebsd.org>
To:        Ulrich Spoerlein <uspoerlein@gmail.com>
Cc:        freebsd-current@freebsd.org, David Adam <zanchey@ucc.gu.uwa.edu.au>
Subject:   Re: CURRENT kernel crashes on boot with BTX error
Message-ID:  <200810231510.33474.jhb@freebsd.org>
In-Reply-To: <20081023185544.GB1595@roadrunner.spoerlein.net>
References:  <alpine.DEB.1.10.0810221249310.3065@martello.ucc.gu.uwa.edu.au> <alpine.DEB.1.10.0810230107260.3065@martello.ucc.gu.uwa.edu.au> <20081023185544.GB1595@roadrunner.spoerlein.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thursday 23 October 2008 02:55:44 pm Ulrich Spoerlein wrote:
> On Thu, 23.10.2008 at 14:16:44 +0800, David Adam wrote:
> > I did try a loader from the CURRENT sources, which produced the same 
> > output.
> > 
> > However, the plot thickens. When I don't specify the kernel name with 
> > nextboot(8), but instead interrupt the loader and use 'unload; load 
> > /boot/kernel/gd-8; load /boot/kernel/gd-8/geom_mirror.ko' and so on for my 
> > other modules, the kernel boots quite happily - both using the 7.0 loader 
> > and the new one from the CURRENT sources.
> > 
> > Are there likely to be big differences in the boot process between using 
> > nextboot(8) and the unload/load method? I've used nextboot(8) to 
> > successfully load other kernels on this machine before.
> > 
> > In any case, now that I have a successful workaround I can try the patches 
> > I was after in the first place. If there's any more information I can 
> > provide, or you'd like access to this system (it has serial console and 
> > remote power control), let me know.
> 
> Wild guess: I think acpi is no longer built as module on recent -CURRENT
> but included in the kernel. Perhaps the unload dance you did above
> "fixed" the double loading of the module? Could you check if your
> nextboot kernel is containing/missing ACPI directly?

The loader doesn't always do a good job with detecting double loads of modules 
(the module defined in MODULE_VERSION has to match the filename for the 
loader to DTRT I think), but acpi.ko is one of the few cases when it does get 
it right.

-- 
John Baldwin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200810231510.33474.jhb>