Bug 121960

Summary:	Core dumps in VM's after update from SuSE 9.3 with XEN 2 to SuSE 10.0 with XEN 3
Product:	[openSUSE] SUSE LINUX 10.0	Reporter:	Ralf Müller <ralf>
Component:	Xen	Assignee:	Charles Coffing <ccoffing>
Status:	RESOLVED WONTFIX	QA Contact:	E-mail List <qa-bugs>
Severity:	Normal
Priority:	P5 - None
Version:	Final
Target Milestone:	---
Hardware:	PC
OS:	All
Whiteboard:
Found By:	Customer	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---

Description Ralf Müller 2005-10-10 13:55:08 UTC

After updating a XEN Server with SuSE 9.3 to SuSE 10.0 different programs   
within the existing SuSE 9.3 VM's generate a core dump. Major players here are  
hwscan and YaST - which crash both. The result is an unusable Virtual Machine.  
It starts with it's inability to configure a network card and ends to be hard  
to configure without YaST.   
   
/sbin/hwscan --silent --boot --fast --isapnp --pci --block --floppy --mouse   
Segmentation fault (core dumped)   
   
==>   
(gdb) bt   
#0  0x4003e94f in read_udevinfo () from /lib/libhd.so.10   
#1  0x4004bf49 in hd_scan_int () from /lib/libhd.so.10   
#2  0x40040023 in hd_scan () from /lib/libhd.so.10   
#3  0x4004167c in hd_list () from /lib/libhd.so.10   
#4  0x08048da4 in ?? ()   
#5  0x0804c050 in ?? ()   
#6  0x00000019 in ?? ()   
#7  0x00000001 in ?? ()   
#8  0x00000000 in ?? ()   
#9  0x00000001 in ?? ()   
#10 0x00001728 in ?? ()   
#11 0x00000001 in ?? ()   
#12 0x00000001 in ?? ()   
#13 0xbfc9f764 in ?? ()   
#14 0x00000000 in ?? ()   
#15 0xbfc9f6c8 in ?? ()   
#16 0x0804962e in ?? ()   
#17 0x0804bb00 in optind ()   
#18 0x08049fa7 in _IO_stdin_used ()   
#19 0x0804b640 in ?? ()   
#20 0x00000000 in ?? ()   
#21 0x00000000 in ?? ()   
#22 0xbfc9f764 in ?? ()   
#23 0x00000000 in ?? ()   
#24 0xbfc9f764 in ?? ()   
#25 0x00000000 in ?? ()   
#26 0x00000000 in ?? ()   
#27 0xbfc9f718 in ?? ()   
#28 0x08049b2e in ?? ()   
#29 0x0804bb00 in optind ()   
#30 0x40016ff4 in ?? () from /lib/ld-linux.so.2   
#31 0x00000001 in ?? ()   
#32 0x40017518 in ?? ()   
#33 0x00000000 in ?? ()   
#34 0x0804b574 in ?? ()   
#35 0xbfc9f6f8 in ?? ()   
#36 0x080489e1 in _init ()   
Previous frame inner to this frame (corrupt stack?)   
  
  
The System has been a SuSE 9.3 32-Bit minimal installation with XEN 2 on a  
DELL PowerEdge SC 1425 and was updated to SuSE 10.0 with XEN 3.

Comment 1 Martin Lasarsch 2005-10-10 14:03:14 UTC

kurt, something for you?

Comment 2 Kurt Garloff 2005-12-05 13:34:31 UTC

Hmmm, does domain0 survive this?
Does a 9.3 domU run stable except for the hwscan crash?

Comment 3 Ralf Müller 2005-12-05 15:47:51 UTC

Both do survive - the domU is just kind of hard to administrate ... no yast, no you ...

Comment 4 Kurt Garloff 2005-12-20 21:43:59 UTC

Hmmm, just tested with the latest 10.0 update packages on my machine:

root@g46: ~ [0]# /sbin/hwscan --silent --boot --fast --isapnp --pci --block --floppy --mouse
root@g46: ~ [0]# cat /etc/SuSE-release 
SuSE Linux 9.3 (i586)
VERSION = 9.3
root@g46: ~ [0]# uname -a
Linux g46 2.6.13-15.7-xen #1 SMP Tue Nov 29 14:32:29 UTC 2005 i686 i686 i386 GNU/Linux
root@g46: ~ [0]# 

So it does work here. Can you confirm it's fixed for you as well with the 10.0 updates?

Comment 5 Ralf Müller 2005-12-21 13:39:54 UTC

(In reply to comment #4)
> Hmmm, just tested with the latest 10.0 update packages on my machine:
> [ worked for you ]

Sorry - but I can't confirm that it works here. I have the same problem as before:

skel:~ # /sbin/hwscan --silent --boot --fast --isapnp --pci --block --floppy --mouse
Segmentation fault
skel:~ # cat /etc/SuSE-release
SuSE Linux 9.3 (i586)
VERSION = 9.3
skel:~ # uname -a
Linux skel 2.6.13-15.7-xen #1 SMP Tue Nov 29 14:32:29 UTC 2005 i686 i686 i386 GNU/Linux

Comment 6 Charles Coffing 2006-02-16 16:22:06 UTC

This doesn't sound like a supported configuration.

SUSE 9.3 VMs just won't work (at least, not easily) on SUSE 10.0 with Xen 3.  If you don't upgrade the kernel (the one within the VM) to a Xen-3 compatible one (e.g., install the SUSE 10.0 kernel within the SUSE 9.3 VM), then it won't work, because the Xen 2->3 transition broke the ABI between the kernel and hypervisor.  But if you do upgrade the kernel, we can't guarantee the 2.6.16-era kernel will work with the 2.6.13-era user-space tools in the VM (/sys has changed; /proc has changed; hotplug/udev have changed, etc.)

The only way to "fix" this is to maintain a Xen-3 compatible 2.6.13 kernel, but we aren't doing that.  We're supporting SLES 9 (2.6.5) and SLES 10 / SL 10.0 (2.6.16) on Xen 3.