OpenSolaris (previously working) no longer boots: kernel panics early

Bug #590201 reported by Ed Hynan
16
This bug affects 2 people
Affects Status Importance Assigned to Milestone
qemu-kvm (Ubuntu)
Triaged
Medium
Unassigned
Lucid
Won't Fix
Medium
Unassigned
Maverick
Won't Fix
Medium
Unassigned
Natty
Won't Fix
Medium
Unassigned

Bug Description

Binary package hint: qemu-kvm

% lsb_release -rd
Description: Ubuntu 10.04 LTS
Release: 10.04

% kvm -h | head -1
QEMU PC emulator version 0.12.3 (qemu-kvm-0.12.3), Copyright (c) 2003-2008 Fabrice Bellard

% cat /proc/cpuinfo
[snip]
AMD Turion(tm) 64 X2 Mobile Technology TL-60
[snip]

% uname -a
Linux alligator 2.6.32-22-generic #36-Ubuntu SMP Thu Jun 3 19:31:57 UTC 2010 x86_64 GNU/Linux

% kvm -enable-kvm -cpu kvm64 -M pc -smp 1 -rtc base=localtime -monitor telnet:127.0.0.1:5448,server,nowait -m 1024M -hda /home/<my acct>/VM/OpenSolaris_vm1 -cdrom /dev/cdrom -net nic,model=rtl8139 -net user,hostfwd=tcp::5555-:22,hostname= -vnc 127.0.0.1:8 -smb /home/<my acct>/VM/mssmb_share -soundhw sb16 /home/<my acct>/tmp/kvm-3414-audio -name OpenSolaris_vm1 /home/<my acct>/VM/OpenSolaris_vm1

Running OpenSolaris had been a qualified success (no usb, unusable w/ cpu > 1; otherwise OK).
Now, after passing OpenSolaris' grub menu the (OSAL) kernel panics; kvm spins.

I can boot (this same machine) into Ubuntu 9.10, which has not had the latest security updates,
and with that the OpenSolaris VM works as expected.

This Ubuntu 10.04 has had all package updates (including recent kernel security issue). Sorry,
but I have not tried to start the OpenSolaris VM for some time, so I cannot associate the
problem with a particular update. It *did* work with the fresh 10.04 install (after changing -soundhw
to work around cpu hogging).

Working with this system: Ubuntu 10.04 Server 32 bit; Ubuntu 9.04 Desktop 64 bit;
MS Vista 32 bit; Plan9

== Regression details ==
Discovered in version: 10.04
Last known good version: 9.10

Revision history for this message
C de-Avillez (hggdh2) wrote :

thank you for opening this bug and helping make Ubuntu better.

Can you please give us the exact package version strings for QEMU, and the linux kernel?

dpkg -l \*kvm\* linux-image\* | egrep ^ii

Changed in qemu-kvm (Ubuntu):
importance: Undecided → Low
status: New → Incomplete
tags: added: regression-potential
Revision history for this message
Ed Hynan (ehynan-suffolk) wrote :

> Can you please give us the exact package version strings [...]

Sorry about that.

% dpkg -l \*kvm\* linux-image\* | egrep '^ii'
ii kvm 1:84+dfsg-0ubuntu16+0.12.3+noroms+0ubuntu9 dummy transitional pacakge from kvm to qemu-kvm
ii kvm-pxe 5.4.4-1ubuntu1 PXE ROM's for KVM
ii linux-image-2.6.32-21-generic 2.6.32-21.32 Linux kernel image for version 2.6.32 on x86/x8
ii linux-image-2.6.32-22-generic 2.6.32-22.36 Linux kernel image for version 2.6.32 on x86/x8
ii linux-image-generic 2.6.32.22.23 Generic Linux kernel image
ii qemu-kvm 0.12.3+noroms-0ubuntu9 Full virtualization on i386 and amd64 hardware
ii qemu-kvm-extras 0.12.3+noroms-0ubuntu9 fast processor emulator binaries for non-x86 ar

Revision history for this message
Ed Hynan (ehynan-suffolk) wrote :

Well, no comment on this for some time. I'll say that I've updated kvm* from the proposed repository, and that did not help.

Let there be no doubt about the the file housing OpenSolaris: I can still boot an older Ubuntu 9.10 (with no upgrades applied since this problem appeared) and use ***the same*** file as OpenSolaris with the old kvm; and it works as expected (or as one would hope).

Is it time to state explicitly for Google and Ubuntu users that kvm under Ubuntu (10.04) does *not* support OpenSolaris?

Revision history for this message
Jean-Baptiste Lallement (jibel) wrote :

Thanks for your report. Can you try with the latest version in Ubuntu 10.10 and tell us if it's still not working ?
Thanks in advance.

Changed in qemu-kvm (Ubuntu):
importance: Low → Medium
status: Incomplete → Triaged
description: updated
Changed in qemu-kvm (Ubuntu Lucid):
status: New → Triaged
importance: Undecided → Medium
tags: added: regression-release
removed: regression-potential
tags: added: lucid
Revision history for this message
Ed Hynan (ehynan-suffolk) wrote :

> Can you try with the latest version in Ubuntu 10.10 [...]

Sorry for the late response; I hadn't updated.

Also sorry to say it's still broken. I tried both the installed OpenSolaris VM
disk image and the installation CD and the result is the same: kernal panic
(message "BAD TRAP: type=d (#gp General Protection) r").

Other systems are working (Linux, MS Vista, Plan 9).

Revision history for this message
Jean-Baptiste Lallement (jibel) wrote :

Thanks for confirming. I'm adding a task for maverick too.

Changed in qemu-kvm (Ubuntu Maverick):
importance: Undecided → Medium
status: New → Triaged
Revision history for this message
Jean-Baptiste Lallement (jibel) wrote :

I'm also adding a task for natty because that's the same version than in maverick.

Revision history for this message
Serge Hallyn (serge-hallyn) wrote :

I just installed and booted opensolaris with no problems under natty.

I'll try under maverick as well.

My cmdline was simply
   qemu-kvm -hda solaris.img -cdrom Downloads/opensolaris.iso -m 1024 -smp 2 -boot d -vnc :1
to install and
   qemu-kvm -hda solaris.img -boot c -m 1024 -smp 2 -vnc :1
to boot.

Changed in qemu-kvm (Ubuntu Natty):
status: Triaged → Fix Released
Revision history for this message
Serge Hallyn (serge-hallyn) wrote :

@Ed,

opensolaris is booting fine for me in an uptodate maverick system on
an intel laptop. Can you please give the precise steps you followed
in your test on 10.10? URL for iso's, and any non-default steps you
took in the installer?

I suspect, however, that this is another bug manifesting itself only
on AMD.

Revision history for this message
Ed Hynan (ehynan-suffolk) wrote :

Serge,

The problem must be related to the hardware. The machine I have the bug
on is a laptop w/ AMD Turion cpu, ATI/AMD chips (RS690, SB600). As you
said it works for you on an Intel laptop, I've managed to try the same VM
image on a HP laptop w/ Intel Core quad, Ubuntu 10.04 fully updated, and
it is working.

Note that the AMD machine still has a Ubuntu 9.04 _not_ updated, and kvm
runs the OpenSolaris VM with that, just not with 10.04 or 10.10; moreover,
VirtualBoxOSE will run the _same_ VM image under 10.04 and 10.10.
So, the problem must be between the hardware and some kernel and/or kvm
that appeared with or shortly after the 10.04 release.

Note also that that since 10.10 there are other problems, such as that display
brightness or pm suspend do not work if booted on main power, but _do_ work
if booted on battery. The kernel has regressed for that hardware.

>> I suspect, however, that this is another bug manifesting itself only
>> on AMD.

I should have tried on Intel sooner. Still, it isn't strictly a hardware problem -
the older software worked.

- Ed

Revision history for this message
Serge Hallyn (serge-hallyn) wrote : Re: [Bug 590201] Re: OpenSolaris (previously working) no longer boots: kernel panics early

Quoting Ed Hynan (<email address hidden>):
> The problem must be related to the hardware. The machine I have the bug
> on is a laptop w/ AMD Turion cpu, ATI/AMD chips (RS690, SB600). As you
> said it works for you on an Intel laptop, I've managed to try the same VM
> image on a HP laptop w/ Intel Core quad, Ubuntu 10.04 fully updated, and
> it is working.

Hi Ed,

that may actually be good news. Could you try the kernel from
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/714335/comments/30
on your AMD lucid host and see if that fixes the opensolaris boots?

> Note also that that since 10.10 there are other problems, such as that display
> brightness or pm suspend do not work if booted on main power, but _do_ work
> if booted on battery. The kernel has regressed for that hardware.

Please do open bugs for those against the kernel and display drivers
for those, if you haven't already.

Revision history for this message
Ed Hynan (ehynan-suffolk) wrote :

OK Serge, I've tried that kernel and it didn't help except
maybe with a dmesg line that looks new (I might have
just missed it).

1) With my long kvm invocation, I got the OSol kernel
panic output shown in the .png attachment to an
earlier message; but, also this line dmesg:

  [ 210.924821] kvm: 2594: cpu0 unhandled rdmsr: 0xc0011023

2) With the short commandline you posted a few days ago
the OSol kernel doesn't panic but attempts a BSDish dump, and
no dump device is configured so it just stops. I will attach a
.png of the visible last lines of that. In this case the dmesg
line is not "unhandled rdmsr", but instead:

  [ 580.689021] kvm: emulating exchange as write

Something looks interesting: in 1) kvm says "unhandled rdmsr",
and in 2) the kernel call trace shows a few symbols with
'wrmsr' (see attached .png). Error regarding a model specific
register? I don't have experience at this low level so I am
only hoping there's a clue here.

-Ed

Changed in qemu-kvm (Ubuntu Natty):
status: Fix Released → Triaged
Revision history for this message
Serge Hallyn (serge-hallyn) wrote :

Thanks, Ed. I'd like to see if latest kernel or qemu fixes it. Could you try doing add-apt-repository ppa:kernel-ppa/ppa and then (after apt-get update) install the package linux-image-generic-lts-backport-natty (or linux-image-server-lts-backport-natty)? Then see whether the images boot under 2.6.38-5. If yes, then we'll have to figure out which kernel patches fixed it. If not, then we'll try with qemu 0.14.0.

Revision history for this message
Ed Hynan (ehynan-suffolk) wrote :

>> Thanks, Ed. I'd like to see if latest kernel or qemu fixes it. Could you try doing add-apt-repository ppa:kernel->> >> ppa/ppa and then (after apt-get update) install the package linux-image-generic-lts-backport-natty (or linux->> >> image-server-lts-backport-natty)? Then see whether the images boot under 2.6.38-5.

I added the repository and updated, but that did not make any
kernel packages available. So, I got
   http://ppa.launchpad.net/kernel-ppa/ppa/ubuntu/pool/main/l/linux-lts-backport-natty/linux-image-2.6.38-5-generic_2.6.38-5.32~lucid1_amd64.deb
and installed that.

It did not change the behavior of kvm (it did change the other thing
I mentioned: that kernel would not boot on main power - only on
battery).

Since IIRC the problem began after some update to 10.04, I'll try
a few packages from that in order looking for the change.

Dave Walker (davewalker)
tags: added: server-nrs
Dave Walker (davewalker)
tags: removed: server-nrs
Revision history for this message
Rolf Leggewie (r0lf) wrote :

maverick has seen the end of its life and is no longer receiving any updates. Marking the maverick task for this ticket as "Won't Fix".

Changed in qemu-kvm (Ubuntu Maverick):
status: Triaged → Won't Fix
Revision history for this message
Rolf Leggewie (r0lf) wrote :

natty has seen the end of its life and is no longer receiving any updates. Marking the natty task for this ticket as "Won't Fix".

Changed in qemu-kvm (Ubuntu Natty):
status: Triaged → Won't Fix
Revision history for this message
Rolf Leggewie (r0lf) wrote :

lucid has seen the end of its life and is no longer receiving any updates. Marking the lucid task for this ticket as "Won't Fix".

Changed in qemu-kvm (Ubuntu Lucid):
status: Triaged → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.