Debian Bug report logs -
#760765
intel-microcode: breaks initrd for newer kernels
Reported by: A Mennucc <mennucc1@debian.org>
Date: Sun, 7 Sep 2014 17:18:01 UTC
Severity: grave
Found in version intel-microcode/2.20140624.1
Done: A Mennucc <mennucc1@debian.org>
Bug is archived. No further changes may be made.
Toggle useless messages
Report forwarded
to debian-bugs-dist@lists.debian.org, Henrique de Moraes Holschuh <hmh@debian.org>:
Bug#760765; Package intel-microcode.
(Sun, 07 Sep 2014 17:18:06 GMT) (full text, mbox, link).
Acknowledgement sent
to A Mennucc <mennucc1@debian.org>:
New Bug report received and forwarded. Copy sent to Henrique de Moraes Holschuh <hmh@debian.org>.
(Sun, 07 Sep 2014 17:18:07 GMT) (full text, mbox, link).
Message #5 received at submit@bugs.debian.org (full text, mbox, reply):
[Message part 1 (text/plain, inline)]
Package: intel-microcode
Version: 2.20140624.1
Severity: critical
Justification: breaks the whole system
hi
I have installed a small wheezy system, and then promptly
upgraded it to jessie. This system has rootfs crypted
with cryptsetup (altogh I am not sure this is relevant).
I have two kernel currently in /boot, 3.14-2-amd64 and 3.2.0-4-amd64,
I can boot the system with the latter, but not with the former.
When I try to boot using 3.14-2-amd64 , the scripts in initrd
complain that it cannot find the rootfs.
I tried to investigate the problem, and found this startling
fact. Usually initrd.img is a CPIO archive, compressed
with GZIP. Instead, initrd.img-3.14-2-amd64 is a CPIO file
but not compressed, and it seems to contain only this file
$ cpio -t < /boot/initrd.img-3.14-2-amd64
kernel
kernel/x86
kernel/x86/microcode
kernel/x86/microcode/GenuineIntel.bin
(even if initrd.img-3.14-2-amd64 is itself quite large, roughly 14MB).
After some fiddling, I noted that if I delete the file
/usr/share/initramfs-tools/hooks/intel_microcode
and regenerate initrd.img-3.14-2-amd64
then it is fine. Unfortunately I cannot understand
why it is broken.
You may find the broken initrd in
http://mennucc1.debian.net/tmp/initrd.img-3.14-2-amd64
I attach 3 files, that are the output of
$ update-initramfs -v -u -k 3.14-2-amd64
3.14~no~microcode : the output when /usr/share/initramfs-tools/hooks/intel_microcode is deleted
3.14~with~microcode : the output when /usr/share/initramfs-tools/hooks/intel_microcode is present
3.14~with~microcode~xz : the output when I add 'set -zx' to /usr/share/initramfs-tools/hooks/intel_microcode
a.
-- System Information:
Debian Release: jessie/sid
APT prefers testing
APT policy: (500, 'testing')
Architecture: amd64 (x86_64)
Kernel: Linux 3.2.0-4-amd64 (SMP w/2 CPU cores)
Locale: LANG=it_IT.UTF-8, LC_CTYPE=it_IT.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
Versions of packages intel-microcode depends on:
ii iucode-tool 1.0.3-1
Versions of packages intel-microcode recommends:
ii initramfs-tools 0.116
intel-microcode suggests no packages.
-- no debconf information
--
Andrea Mennucc
"E' un mondo difficile. Che vita intensa!" (Tonino Carotone)
[3.14~no~microcode.gz (application/gzip, attachment)]
[3.14~with~microcode.gz (application/gzip, attachment)]
[3.14~with~microcode~xz.gz (application/gzip, attachment)]
[signature.asc (application/pgp-signature, inline)]
Information forwarded
to debian-bugs-dist@lists.debian.org:
Bug#760765; Package intel-microcode.
(Sun, 07 Sep 2014 20:33:21 GMT) (full text, mbox, link).
Acknowledgement sent
to Henrique de Moraes Holschuh <hmh@debian.org>:
Extra info received and forwarded to list.
(Sun, 07 Sep 2014 20:33:21 GMT) (full text, mbox, link).
Message #10 received at 760765@bugs.debian.org (full text, mbox, reply):
severity 760765 grave
thanks
On Sun, 07 Sep 2014, A Mennucc wrote:
> Package: intel-microcode
> Version: 2.20140624.1
> Severity: critical
> Justification: breaks the whole system
Urk.
> I have installed a small wheezy system, and then promptly
> upgraded it to jessie. This system has rootfs crypted
> with cryptsetup (altogh I am not sure this is relevant).
This is likely an important point, as the early initramfs is known to work
in the general case (I use it everywhere, for example).
And I know for a fact a lot of users are using it safely with 3.14 (or at
least were using it safely until a week ago), so I will lower the severity
to grave for the moment.
> When I try to boot using 3.14-2-amd64 , the scripts in initrd
> complain that it cannot find the rootfs.
Ok.
> I tried to investigate the problem, and found this startling
> fact. Usually initrd.img is a CPIO archive, compressed
> with GZIP. Instead, initrd.img-3.14-2-amd64 is a CPIO file
> but not compressed, and it seems to contain only this file
This is normal. It is called an hybrid initramfs, with an early initramfs
prepended to the regular initramfs. It is supported by every kernel in
different ways:
The kernel will load cpio *segments* it finds. Each segment can be in any
of the formats it supports. It will just overlay one on the top of the
other.
> $ cpio -t < /boot/initrd.img-3.14-2-amd64
> kernel
> kernel/x86
> kernel/x86/microcode
> kernel/x86/microcode/GenuineIntel.bin
>
> (even if initrd.img-3.14-2-amd64 is itself quite large, roughly 14MB).
There are no userspace tools to process initramfs archives. We have stuff
that handles cpio just fine, but nothing that can really read a
fully-featured initramfs the same way the kernel does.
You can strip off the first initramfs (the one with the microcode), and you
should have the regular compressed initramfs.
> After some fiddling, I noted that if I delete the file
> /usr/share/initramfs-tools/hooks/intel_microcode
> and regenerate initrd.img-3.14-2-amd64
> then it is fine. Unfortunately I cannot understand
> why it is broken.
Just edit /etc/default/intel-microcode to work around the problem.
> You may find the broken initrd in
> http://mennucc1.debian.net/tmp/initrd.img-3.14-2-amd64
Ah, thanks. will inspect.
> Versions of packages intel-microcode depends on:
> ii iucode-tool 1.0.3-1
This could be relevant, as it has a workaround that creates a filename with
NULs at the end (which the kernel, Debian cpio and pax will ignore, but
still maybe some other tool is at play).
--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique Holschuh
Severity set to 'grave' from 'critical'
Request was from Henrique de Moraes Holschuh <hmh@debian.org>
to control@bugs.debian.org.
(Sun, 07 Sep 2014 20:33:28 GMT) (full text, mbox, link).
Information forwarded
to debian-bugs-dist@lists.debian.org:
Bug#760765; Package intel-microcode.
(Sun, 07 Sep 2014 21:27:20 GMT) (full text, mbox, link).
Acknowledgement sent
to Henrique de Moraes Holschuh <hmh@debian.org>:
Extra info received and forwarded to list.
(Sun, 07 Sep 2014 21:27:20 GMT) (full text, mbox, link).
Message #17 received at 760765@bugs.debian.org (full text, mbox, reply):
On Sun, 07 Sep 2014, A Mennucc wrote:
> I have installed a small wheezy system, and then promptly
> upgraded it to jessie. This system has rootfs crypted
> with cryptsetup (altogh I am not sure this is relevant).
Can you send me a recipe to exactly duplicate this install? I'm trying to
duplicate the issue with the wheezy installer's default "all encrypted, all
in one partition, using encrypted lvm" layout, but I am not sure this
duplicates your setup.
--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique Holschuh
Reply sent
to A Mennucc <mennucc1@debian.org>:
You have taken responsibility.
(Mon, 08 Sep 2014 08:00:05 GMT) (full text, mbox, link).
Notification sent
to A Mennucc <mennucc1@debian.org>:
Bug acknowledged by developer.
(Mon, 08 Sep 2014 08:00:05 GMT) (full text, mbox, link).
Message #22 received at 760765-done@bugs.debian.org (full text, mbox, reply):
[Message part 1 (text/plain, inline)]
hi
after fiddling with it for a while, my host would not boot with any
kernel, regardless of the presence of intel-microcode, or not!
So I eventually noted that there was a typo in the UUID for the crypted
root in /etc/crypttab :->
Also I did not know about 'hybrid initramfs' (is there any documentation
around? could not find any); I also found bug
http://bugs.debian.org/717805 , it is really a pity.
So it seems that intel-microcode was not at fault.
Sorry for the noise.
a.
[signature.asc (application/pgp-signature, attachment)]
Bug archived.
Request was from Debbugs Internal Request <owner@bugs.debian.org>
to internal_control@bugs.debian.org.
(Tue, 07 Oct 2014 07:32:13 GMT) (full text, mbox, link).
Send a report that this bug log contains spam.
Debian bug tracking system administrator <owner@bugs.debian.org>.
Last modified:
Thu Jan 4 08:35:13 2018;
Machine Name:
beach
Debian Bug tracking system
Debbugs is free software and licensed under the terms of the GNU
Public License version 2. The current version can be obtained
from https://bugs.debian.org/debbugs-source/.
Copyright © 1999 Darren O. Benham,
1997,2003 nCipher Corporation Ltd,
1994-97 Ian Jackson,
2005-2017 Don Armstrong, and many other contributors.