Bug#593245: linux-image-2.6.32-5-xen-amd64_2.6.32-20_amd64.deb can't start VM's due to udev rule mismatch

["Joseph M. Deming" <joseph.deming@matrix.msu.edu>]

Ack, response yesterday didn't get through apparently.

I am the OP and as far as I know this issue was addressed within 2-3 
weeks of my original post.  I can't isolate exactly which patch fixed 
things up, all I know is that there were several changes to 
linux-image-2.6.32-5 around that time and I think the fix was in there, 
but I can't say surely that it wasn't a fix to the xen codebase that 
also helped.  What I can say is that, currently running 2.6.32-26 this 
issue no longer presents itself and I have been able to start machines 
normally, without any held-back patches since around late August, early 
September.

On 2.6.32-26 this is my output: find /dev|grep evt
/dev/.udev/db/misc:xen/evtchn
/dev/xen/evtchn

>find /sys | grep evtchn
/sys/devices/virtual/misc/xen!evtchn
/sys/devices/virtual/misc/xen!evtchn/uevent
/sys/devices/virtual/misc/xen!evtchn/dev
/sys/devices/virtual/misc/xen!evtchn/subsystem
/sys/devices/virtual/misc/xen!evtchn/power
/sys/devices/virtual/misc/xen!evtchn/power/wakeup
/sys/class/misc/xen!evtchn
/sys/module/xen_evtchn
/sys/module/xen_evtchn/holders
/sys/module/xen_evtchn/initstate
/sys/module/xen_evtchn/refcnt
/sys/module/xen_evtchn/sections
/sys/module/xen_evtchn/sections/.note.gnu.build-id
/sys/module/xen_evtchn/sections/.text
/sys/module/xen_evtchn/sections/.exit.text
/sys/module/xen_evtchn/sections/.init.text
/sys/module/xen_evtchn/sections/.rodata
/sys/module/xen_evtchn/sections/.parainstructions
/sys/module/xen_evtchn/sections/.rodata.str1.1
/sys/module/xen_evtchn/sections/__bug_table
/sys/module/xen_evtchn/sections/.data
/sys/module/xen_evtchn/sections/.gnu.linkonce.this_module
/sys/module/xen_evtchn/sections/.bss
/sys/module/xen_evtchn/sections/.symtab
/sys/module/xen_evtchn/sections/.strtab
/sys/module/xen_evtchn/notes
/sys/module/xen_evtchn/notes/.note.gnu.build-id

As I said all has seemed long-since fixed up.  For the record, I run 
Sid, so I can't verify Squeeze for sure, but it would likely be fixed up 
also.  We will be applying our monthly updates on Wednesday (tomorrow) 
and that will take us up to 2.6.32-27, if you wish for any more feedback 
let me know.

- jmd


On 11/23/2010 03:52 AM, Timo Juhani Lindfors wrote:
> Ian Campbell<ijc@hellion.org.uk>  writes:
>    
> > The version of xen-utils-common may also have some bearing on this since
> > it is the package which provides /lib/udev/rules.d/xend.rules
> >      
> Here in working case I have
>
> $ dpkg-query -W xen*
> xen-hypervisor-amd64
> xen-linux-system-2.6.32-5+bug580889.1-xen-amd64
> xen-tools
> xen-utils-common        4.0.0-1
> xenstore-utils  4.0.1-1
>
>    
> > It's also worth checking for stale entries in /etc/udev/rules.d (or
> > whatever the previous path was) which might be interfering.
> >      
> $ md5sum /lib/udev/rules.d/*xen*
> cea42daba332764f255dadec112f64c5  /lib/udev/rules.d/xen-backend.rules
> 5a306f5dba657c3c962c158782e848fb  /lib/udev/rules.d/xend.rules
>
>    


--
Joseph M. Deming
System Administrator
MATRIX/History
415 Nat Sci Bldg
East Lansing, MI 48824
(517) 884-2472
joseph.deming@matrix.msu.edu