Re: md raid oops on 2.4.25/alpha

From: Marcelo Tosatti
Date: Fri Apr 02 2004 - 07:19:38 EST


On Fri, Apr 02, 2004 at 04:53:01PM +1000, Lewis Shobbrook wrote:
> Hi All,
>
> [Excerpt from linux-raid mailing list]
> > We have some problems with the md code on alpha. We get
> > regular oops when using the md raid1. Here we got another
> > oops when fsck (at boot
> > time) the raid:
> > This was after a fresh reboot. As long as only the raid is
> > *not* mounted of fsck the machine works without any oops.

There is a patch by Ivan against the MD code which avoids
gcc3 < 3.3.3 from generating bad assembly code. This patch
is sitting now in the 2.4 BK tree.

Have you guys tried that one?

> I've found similar results with the unstable Debian 2.4.25.-1-686
> kernel.
>
> > I also can mount the hard disks *without* raid directly as
> > hda1 and hdc1, and do NOT get any errors here, so I suspect
> > that only the md code is the culprit.
>
> Same again here.
>
> I have /dev/hda2 listed in raidtab as a failed disc, lilo points to hda
> as boot. I get kernel panic when attemtpting to boot to it.
>
> I thought this strange as I have another system running the very same
> kernel (even used the same copied kernel deb) with a UU (no failed)raid
> 1 running off a hpt372 onboard controller (as soft md, not hardware raid
> 1). The raid device was present prior to the kernel in this instance.
> I was impressed that the std initrd wokred with both md and hpt372
> without modification, where the process had been more involved in the
> past requiring a custom initrd or "compiled in" kernel.
>
> I thought that the devfs (compiled in with the std debain kernel) may
> have been an issue, but it has never been in the past and the same
> filesystem on the raidtab listed "failed" drive /dev/hda2 is happy.
>
> I can boot the the raid 1 as root filesystem when passing root=/dev/md0
> loading through a system rescue disc (http://www.sysresccd.org/) with a
> 2.4.25 kernel.
>
> Attempting to boot to /dev/md0 with the Debian 2.4.25-1-686 kernel
> panics after attempting to mount the device /dev2/root2 as ext2, minix &
> fat (possibly others that disappear before they can be read) and
> complains the std way when you attempt to mount with the wrong fs.
> ...
> pivot_root: no such file or directory
> /sbin/init: 347: cannot open dev/console : no such file
> Kernel panic: Attempted to kill init !
>
> I've attempted to pre-load the initrd modules, but didn't expect this to
> be a solution and it wasn't.
>
> I'm scratching the head and losing hair...
>
> Never had any trouble of this sort before.

Lewis, dont know what is going wrong here for you. Sorry.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/