Bug#987788: debian-installer: qemu-system-s390x installation fails due to segfault in main-menu

[Valentin Vidic]

Just to report I don't get the crash anymore with bullseye rc3
installer, so most probably this was the same problem as:

  #987368 Installer fails at first menu "Choose language"

-- 
Valentin

Bug#987788: debian-installer: qemu-system-s390x installation fails due to segfault in main-menu

[Valentin Vidic]

On Mon, May 03, 2021 at 06:38:05PM +0200, Samuel Thibault wrote:
> How did you start it? I couldn't manage to make it start (with qemu from
> unstable or from stable), I only got the kernel start and a couple of
> lines. I was taking
> 
> http://ftp.de.debian.org/debian/dists/bullseye/main/installer-s390x/current/images/generic/initrd.debian
> http://ftp.de.debian.org/debian/dists/bullseye/main/installer-s390x/current/images/generic/kernel.debian
> 
> and run 
> 
> qemu-system-s390x -kernel /tmp/s390x/kernel.debian -initrd 
> /tmp/s390x/initrd.debian -m 1G -serial stdio
> 
> and got
> 
> [..kernel logs..]
> [4.349417] Run /init as init process
> [5.499051] failover: module verification failed: signature and/or 
> required key missing - tainting kernel
> [6.530501] virtio_net virtio0 enc0: renamed from eth0
> steal-ctty: No such file or directory
> [   83.282353] random: crng init done
> 
> and nothing more.

Hi, I start it with:

virt-install --arch=s390x --name test --memory 1024 --disk size=3 -l 
http://ftp.de.debian.org/debian/dists/bullseye/main/installer-s390x/ 
--extra-args='BOOT_DEBUG=3'

but it should be more or less the same thing as your command. The problem
is you are hitting steal-ctty bug from #961056. I use BOOT_DEBUG=3 to
get a shell before the installer starts and edit /sbin/reopen-console
with this fix:

https://salsa.debian.org/installer-team/rootskel/-/merge_requests/2/diffs

After that installer should start and sometimes (but not always) fail in
the first step (configuring network interfaces). The problem is I don't
have a shell after that segfault to get more info from the logs or dmesg.

-- 
Valentin

Bug#987788: debian-installer: qemu-system-s390x installation fails due to segfault in main-menu

[Samuel Thibault]

Hello,

Valentin Vidic, le lun. 03 mai 2021 18:15:22 +0200, a ecrit:
> On Mon, May 03, 2021 at 09:16:03AM +0200, Cyril Brulebois wrote:
> > Do we have any idea whether that could be something that's only
> > triggered within QEMU (maybe try other QEMU versions?), or something
> > that affects bare metal systems?
> 
> I don't have access to real s390x hardware, but I tried to run the
> installation with QEMU 5.2 from unstable.

How did you start it? I couldn't manage to make it start (with qemu from
unstable or from stable), I only got the kernel start and a couple of
lines. I was taking

http://ftp.de.debian.org/debian/dists/bullseye/main/installer-s390x/current/images/generic/initrd.debian
http://ftp.de.debian.org/debian/dists/bullseye/main/installer-s390x/current/images/generic/kernel.debian

and run 

qemu-system-s390x -kernel /tmp/s390x/kernel.debian -initrd 
/tmp/s390x/initrd.debian -m 1G -serial stdio

and got

[..kernel logs..]
[4.349417] Run /init as init process
[5.499051] failover: module verification failed: signature and/or required 
key missing - tainting kernel
[6.530501] virtio_net virtio0 enc0: renamed from eth0
steal-ctty: No such file or directory
[   83.282353] random: crng init done

and nothing more.

Samuel

Bug#987788: debian-installer: qemu-system-s390x installation fails due to segfault in main-menu

[Valentin Vidic]

On Mon, May 03, 2021 at 09:16:03AM +0200, Cyril Brulebois wrote:
> Do we have any idea whether that could be something that's only
> triggered within QEMU (maybe try other QEMU versions?), or something
> that affects bare metal systems?

I don't have access to real s390x hardware, but I tried to run the
installation with QEMU 5.2 from unstable. Similar to QEMU 3.1 from
stable the installation sometimes starts correctly but most often
fails in the first step (Configure the network device).

-- 
Valentin

Bug#987788: debian-installer: qemu-system-s390x installation fails due to segfault in main-menu

[Cyril Brulebois]

Hi,

Valentin Vidic  (2021-04-29):
> Dear Maintainer,
> 
> Running the installation inside a QEMU instance does not work in bullseye:
> 
> $ virt-install --arch=s390x --name test --memory 1024 --disk none -l 
> http://ftp.de.debian.org/debian/dists/bullseye/main/installer-s390x/
> ...
> 
>  ┌───┤ [!!] Configure the network device ├───┐
>  │   │
>  │ Installation step failed  │
>  │ An installation step failed. You can try to run the failing item  │
>  │ again from the menu, or skip it and choose something else. The│
>  │ failing step is: Configure the network device │
>  │   │
>  │ │
>  │   │
>  └───┘
> 
> [  122.379000] User process fault: interruption code 003b ilc:3 in 
> main-menu[2aa33f0+4000]
> [  122.381112] Failing address:  TEID: 0800
> [  122.381149] Fault in primary space mode while using user ASCE.
> [  122.381239] AS:02ae01c7 R3:0024
> 
> Seems like a segfault in main-menu binary related to network detection?

Do we have any idea whether that could be something that's only
triggered within QEMU (maybe try other QEMU versions?), or something
that affects bare metal systems?


Cheers,
-- 
Cyril Brulebois (k...@debian.org)
D-I release manager -- Release team member -- Freelance Consultant


signature.asc
Description: PGP signature

Bug#987788: debian-installer: qemu-system-s390x installation fails due to segfault in main-menu

[Valentin Vidic]

Package: debian-installer
Severity: important

Dear Maintainer,

Running the installation inside a QEMU instance does not work in bullseye:

$ virt-install --arch=s390x --name test --memory 1024 --disk none -l 
http://ftp.de.debian.org/debian/dists/bullseye/main/installer-s390x/
...

 ┌───┤ [!!] Configure the network device ├───┐
 │   │
 │ Installation step failed  │
 │ An installation step failed. You can try to run the failing item  │
 │ again from the menu, or skip it and choose something else. The│
 │ failing step is: Configure the network device │
 │   │
 │ │
 │   │
 └───┘

[  122.379000] User process fault: interruption code 003b ilc:3 in 
main-menu[2aa33f0+4000]
[  122.381112] Failing address:  TEID: 0800
[  122.381149] Fault in primary space mode while using user ASCE.
[  122.381239] AS:02ae01c7 R3:0024

Seems like a segfault in main-menu binary related to network detection?

-- 
Valentin