Re: System boot failure related to commit 'irqdomain: Switch to per-domain locking'

From: Marc Zyngier
Date: Wed Mar 01 2023 - 09:47:07 EST


On Wed, 01 Mar 2023 11:17:21 +0000,
Bingbu Cao <bingbu.cao@xxxxxxxxxxxxxxx> wrote:
>
>
> On 2/28/23 8:45 AM, Marc Zyngier wrote:
> > On 2023-02-27 10:46, Bingbu Cao wrote:
> >> Hi, Johan and Zyngier,
> >>
> >> I am using a Dell XPS laptop(Intel Processor) just update my
> >> Linux kernel to latest tag 6.2.0, and then I see that the kernel
> >> cannot boot successfully, it reported:
> >> --------------------------------------------
> >> Gave up waiting for root file system device. Common problems:
> >> - Boot args (cat /proc/cmdline)
> >> - Check rootdelay= (did the system wait long enough?)
> >> - Missing modules (cat /proc/modules; ls /dev)
> >>
> >> ALERT! UUID=xxxxxxx does not exist. Dropping to shell!
> >> --------------------------------------------
> >>
> >> And then it drop into initramfs shell, I try to use 'blkid' to
> >> get block devices information, but it showed nothing.
> >>
> >> I also tried add 'rootdelay' and 'rootwait' in bootargs, but it did
> >> not work.
> >>
> >> I am sure that my previous kernel 6.2.0-rc4 work normally, so I
> >> did some bisect and found the commit below cause the failure on
> >> my system:
> >>
> >> 9dbb8e3452ab irqdomain: Switch to per-domain locking
> >>
> >> I really have no idea why it cause my problem, but I see just
> >> reverting this commit really help me.
> >>
> >> Do you have any idea?
> >
> > Please provide us with a kernel boot log. It is very hard
> > to figure out what is going on without it. It would also
> > help if you indicated what sort of device is your root
> > filesystem on (NVMe, SATA, USB...), as it would narrow the
> > search for the culprit.
>
> Unfortunately, I have not find a way to capture the console log, no
> serial for me.

You don't need serial access. Since you're able to interact with the
machine, you can save the dmesg log on some other mass storage. Just
make sure that USB, for example is in your initramfs, and dump the log
there.

> I am using a NVMe for my rootfs. By checking the screen log, I see
> that 1 kernel message is missing:
>
> [ 4.193375] EXT4-fs (nvme0n1p3): mounted filesystem a9e1243b-332f-46ce-a5e7-cea86b44f797 with ordered data mode. Quota mode: none.

OK, at least we know that NVMe is in the loop, but we don't know *why*
yet. Please try and get the dmesg for us. I'm sure someone at Intel
can help you with this.

Thanks,

M.

--
Without deviation from the norm, progress is not possible.