Bug 449786

Summary: kernel-2.6.26-0.45.rc4.git2.fc10.i686 will not boot with kvm
Product: [Fedora] Fedora Reporter: Clyde E. Kunkel <clydekunkel7734>
Component: kvmAssignee: Glauber Costa <gcosta>
Status: CLOSED DUPLICATE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: low Docs Contact:
Priority: low    
Version: rawhideCC: berrange, clalance, dcantrell, katzj, wwoods
Target Milestone: ---   
Target Release: ---   
Hardware: i386   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2008-07-07 18:20:52 EDT Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---

Description Clyde E. Kunkel 2008-06-03 11:44:00 EDT
Description of problem:
This may not be kvm problem, but I have no experience to aid me.

While in rawhide, when booting kernel-2.6.26-0.45.rc4.git2.fc10.i686, msg
Booting the kernel appears and then nothing. 

Version-Release number of selected component (if applicable):
kvm-69-1.fc10.i386

How reproducible:
every time in rawhide

Steps to Reproduce:
1. boot rawhide
2. start virt-manager from a gnome terminal (won't start from menu)
3. start vm and allow vm to auto boot kernel

  
Actual results:
msg:  Booting the kernel appears and then the vm hangs

Expected results:
normal startup of rawhide in VM

Additional info:
[root@P5K-EWIFI ~]# rpm -q kvm
kvm-69-1.fc10.i386
[root@P5K-EWIFI ~]# virsh
Welcome to virsh, the virtualization interactive terminal.

Type:  'help' for help with commands
       'quit' to quit

virsh # version
Compiled against library: libvir 0.4.2
Using library: libvir 0.4.2
Using API: QEMU 0.4.2
Running hypervisor: QEMU 0.9.1

Also see this in dmesg | tail after several attempts.

virbr0: port 1(vnet0) entering learning state
virbr0: topology change detected, propagating
virbr0: port 1(vnet0) entering forwarding state
/usr/share/virt[3823]: segfault at 1c ip 0100080f sp bff27140 error 4 in
libgtk-vnc-1.0.so.0.0.1[fef000+1c000]
virbr0: port 1(vnet0) entering disabled state
device vnet0 left promiscuous mode
virbr0: port 1(vnet0) entering disabled state
device vnet0 entered promiscuous mode
virbr0: port 1(vnet0) entering listening state
kvm: guest NX capability removed
kvm: guest NX capability removed
vnet0: no IPv6 routers present
virbr0: port 1(vnet0) entering learning state
virbr0: topology change detected, propagating
virbr0: port 1(vnet0) entering forwarding state
/usr/share/virt[3830]: segfault at 1c ip 0100080f sp bf96f380 error 4 in
libgtk-vnc-1.0.so.0.0.1[fef000+1c000]
Comment 1 Glauber Costa 2008-06-04 01:40:44 EDT
From what you describe, it seems a bug in the vnc client, so the machine might
well be going all right. Is that what you mean by a "hang" ?
Comment 2 Clyde E. Kunkel 2008-06-04 11:50:42 EDT
By hang I mean that there are no additional msgs on the virtual console after
the booting the kernel msg.  Clicking pause kills the virt-manager gui. 
Shut-down has no effect.  After restarting the virt-manager, it shows the guest
with cpu % oscillating between 49.95 and 50.00%.

I reverted vnc, vnc-libs, gtk-vnc and gtk-vnc-python to F9 final versions, but
that didn't change anything that I could see.  Pausing the guest still kills
virt-manager and the F10 kernel still won't boot.  Now it appears that the guest
disk images are corrupt since I used destroy to stop then since shutdown wasn't
working.  Now I can't boot the f9 kernel which was working ok.  I may have to
recreate the guest now to continue trouble shooting, but am willing to do that.  

Any trouble shooting steps/advice much appreciated.

Comment 3 Will Woods 2008-06-26 14:09:09 EDT
AFAICT kernel-2.6.26-0.93.rc8.f10.i686 doesn't work either. Same result: virtual
machine starts up, kernel/initrd are loaded into memory, and then.. nothing.

This is with kvm-70-2.fc10. 

I'm told there may be some kvm fixes in the pipeline for the kernel; we'll see.
Comment 4 Jeremy Katz 2008-06-26 14:45:42 EDT
FYI, kvm-70 userspace + f9 kernel is okay... so it's something in the kernel
bits that are in 2.6.26 interacting with kvm-70

Haven't gotten further in looking at it
Comment 5 Chuck Ebbert 2008-06-26 14:47:17 EDT
(In reply to comment #3)
> AFAICT kernel-2.6.26-0.93.rc8.f10.i686 doesn't work either. Same result: virtual
> machine starts up, kernel/initrd are loaded into memory, and then.. nothing.
> 
> This is with kvm-70-2.fc10. 
> 
> I'm told there may be some kvm fixes in the pipeline for the kernel; we'll see.

Those fixes are already in 2.6.26-rc8
Comment 6 Mark McLoughlin 2008-07-07 18:20:52 EDT
Pretty sure this is bug #453981

To confirm, set up a serial console and set earlyprintk=serial and you should
see see the guest hang at e.g. :

     0MB HIGHMEM available.
     511MB LOWMEM available.

*** This bug has been marked as a duplicate of 453981 ***