[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#766607: marked as done (linux: ("regression") No serial console output when connecting to virtual serial console for the guest during d-i installation initiated via virt-install)

Your message dated Fri, 24 Oct 2014 14:29:00 +0200
with message-id <20141024122900.GA15185@lorien.valinor.li>
and subject line Re: Bug#766607: linux: ("regression") No serial console output when connecting to virtual serial console for the guest during d-i installation initiated via virt-install
has caused the Debian Bug report #766607,
regarding linux: ("regression") No serial console output when connecting to virtual serial console for the guest during d-i installation initiated via virt-install
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact owner@bugs.debian.org
immediately.)


-- 
766607: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=766607
Debian Bug Tracking System
Contact owner@bugs.debian.org with problems
--- Begin Message --- 
Source: linux
Version: 3.16~rc5-1~exp1
Severity: important

Hi Ben

I choosed important as severity, as it causes a regression from
wheezy->jessie for installation of kvm guests with kernel in the d-i,
but actually it might be overrated, please downgrade if you don't
agree.

Background: For tests I often use the following instruction to quickly
create vms (either preseeded or without preseed):

http://honk.sigxcpu.org/con/Preseeding_Debian_virtual_machines_with_virt_install.html

also having the console output to ttyS0, the kernel arguments are 

auto keymap=us hostname=foo netcfg/get_ipaddress=192.168.122.123 suite=unstable -- console=ttyS0,115200n8

in the example case.

When altough using the current d-i with a 3.16.2-2 the kernel I don't
see anymore neither the kernel boot messages nor the started d-i
anymore when virt-install connects to the virtual serial console for
the guest and cannot proceed with installation in d-i.

I was able to track this down to a change between 3.15.5-1~exp1 and
3.16~rc5-1~exp1 but unfortuantely don't have a specific commit.

To reproduce I attach the small 'reproducer.sh' shellscript which
fetches the d-i linux and initrd.gz from
-location=${HOME}/virt/installer-amd64/ (see below) and also sets 

auto keymap=us hostname=foo netcfg/get_ipaddress=192.168.122.123 suite=unstable -- console=ttyS0,115200n8

as kernel arguments. It is

----cut---------cut---------cut---------cut---------cut---------cut-----
#!/bin/bash

virt-install --connect=qemu:///system \
        --graphics none \
	--location=${HOME}/virt/installer-amd64/ \
	--extra-args="auto keymap=us hostname=foo netcfg/get_ipaddress=192.168.122.123 suite=unstable -- console=ttyS0,115200n8" \
	--name foo \
	--ram 1024 \
	--disk=pool=default,size=5,format=qcow2,bus=virtio,cache=writeback
----cut---------cut---------cut---------cut---------cut---------cut-----

${HOME}/virt/installer-amd64 would look as follows:

----cut---------cut---------cut---------cut---------cut---------cut-----
$ ls -1R installer-amd64
installer-amd64:
current

installer-amd64/current:
images

installer-amd64/current/images:
MANIFEST
netboot

installer-amd64/current/images/netboot:
debian-installer

installer-amd64/current/images/netboot/debian-installer:
amd64

installer-amd64/current/images/netboot/debian-installer/amd64:
initrd.gz
linux
----cut---------cut---------cut---------cut---------cut---------cut-----

and the MANIFEST file only need to contain

----cut---------cut---------cut---------cut---------cut---------cut-----
netboot/debian-installer        -- PXE boot directory for tftp server
----cut---------cut---------cut---------cut---------cut---------cut-----

So basically stripped down content as you would find in
http://ftp.ch.debian.org/debian/dists/unstable/main/installer-amd64/

If I use initrd.gz and linux from 20140802 installation works fine if
I use the ones from current (20141002) the above behavious is
observed, no installer ist seen.

I then afterwards replaced the linux image to find closer version were
the problem appear, and looks like 3.15.5-1~exp1 is good,
3.16~rc5-1~exp1 is bad. Quite some changes with merge in commit
49eb7b0750d9483c74e9c14ae6ea1e9d62481c3c

The mentioned setup is running on Debian stable with
libvirt/0.9.12.3-1+deb7u1, qemu-kvm/1.1.2+dfsg-6+deb7u4.

Regards,
Salvatore

--- End Message ---
--- Begin Message --- 
Hi,

On Fri, Oct 24, 2014 at 01:25:06PM +0200, Salvatore Bonaccorso wrote:
> auto keymap=us hostname=foo netcfg/get_ipaddress=192.168.122.123 suite=unstable -- console=ttyS0,115200n8
                                                                                  ^^<--

Thanks to Cyril Brulebois (KiBi), noticed that I used -- to separte th
arguments and pointed me to https://bugs.debian.org/762007 and
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=51e158c12aca3c9ac63988611a97c05109b14dc9
. So even though the change was introduced between v3.15 and 3.16 it
is not something that should be reverted clearly.

Closing the bug, sorry for the noise.

Regards,
Salvatore

--- End Message ---
Reply to: