Re: Kernel problem?

Thanks to everyone who provided help with this failure of the 4.9.0-5-686-pae kernel. I have tried both "nopti" and "pti=off" kernel parameters to see if it is the fix for "Meltdown" which is causing the problem, but neither parameter changes anything. Since the system doesn't even get to the kernel fsck stage, I don't have any logs to be able to either report a bug or analyse the problem. I also realise now that auto-update is turned on by default in Stretch, so that is how the new kernel became installed. I have temporarily turned auto-update off because I don't want any automatic update to clobber my only working kernel - the 4.9.0-4-686-pae. The updating system only seems to keep 2 generations of kernel?

Any further ideas would be very welcome.

Cheers, Rob Hurle

-----------------------------

Rob Hurle
e-mail: rob1940@gmail.com
Mobile: +61 417 293 603 (Australia)

Telephone: (02) 6236 3895

28 Mirrormere Rd, Burra, NSW 2620, Australia

On 8 January 2018 at 13:11, tv.debian@googlemail.com <tv.debian@googlemail.com> wrote:

On 07/01/2018 21:27, Marc Auslander wrote:

The new kernel implements the "fix" for meltdown. You could try booting
with the fix turned off - I believe the kernel parameter is pti=off
Rob Hurle <rob1940@gmail.com> writes:

Hi All,

I'm running Stretch and yesterday I did my normal:

sudo apt-get update
sudo apt-get upgrade

It seemed to install vmlinuz-4.9.0-5-686-pae (and associated config
and image files, etc) in place of 4.9.0-4-686-pae versions. Now the
system won't boot at all. I have reverted to 4.9.0-4-686-pae and all
is well. My questions are:

1. Does anyone else see this?

2. How can I revert without losing my working 4.9.0-4-686-pae system?
Can I just change the soft links for initrd.img and vmlinuz at / to
point to the 4.9.0-4-686-pae versions instead of the 4.9.0-5-686-pae
ones? Will this break something else for a future upgrade?

Any help much appreciated. Thank you.

Cheers, Rob Hurle

-----------------------------
Rob Hurle
e-mail: rob1940@gmail.com
Mobile: +61 417 293 603 (Australia)
Telephone: (02) 6236 3895
28 Mirrormere Rd, Burra, NSW 2620, Australia

Hi, if the hang is due to memory isolation option, then "nopti" added as a kernel boot parameter will cancel it. If you are on an Intel machine this will leave you exposed to the new class of "Meltdown" attacks.
This kind of boot problem seems to happen to a very small number of systems, patches are already queued in the kernel to (hopefully) correct this, but you will have to wait for them to be merged.

Hope it helps.