race between geom and root mount

John Hay jhay at icomtek.csir.co.za
Tue Sep 21 00:53:28 PDT 2004

On Tue, Sep 21, 2004 at 09:22:21AM +0200, Pawel Jakub Dawidek wrote:
> On Tue, Sep 21, 2004 at 08:11:10AM +0200, John Hay wrote:
> +> > Maybe you're a victim of the problem which was just fixed in -CURRENT.
> +> > Could you try g_mirror.c rev. 1.27 and g_label.c rev. 1.10?
> +> 
> +> I have built a RELENG_5 kernel with g_mirror.c 1.28 and g_label.c 1.11,
> +> but it still does the same. I booted over a serial console and here is
> +> the capture of a boot -v. The order of the last few lines is a little
> +> different when booting using a serial and vga console, but neither
> +> works. :-) With a vga console some of the GEOM messages arrive after
> +> the Mounting Root message. On both some of the last messages are pretty
> +> garbled.
> Not sure how can I help now. This is a new issue, but there is quite simlar
> issue already: when one of the mirror's components is down, root file
> system also can't be mounted, because gmirror waits some time for all
> components to come.
> I've a patch which should solve your problem:
> 	http://people.freebsd.org/~pjd/patches/root_mount.patch
> but it is not going to be committed, as phk asked me to hold on, he will
> provide some more complete solution.

Yes, you are right, this patch does solve my problem. I'll keep it in
my local tree for now.


John Hay -- John.Hay at icomtek.csir.co.za / jhay at FreeBSD.org

More information about the freebsd-current mailing list