[Oraclevm-errata] OVMBA-2012-0029 Oracle VM 3.1 xen Bug Fix update

[Errata Announcements for Oracle VM]

Oracle VM Bug Fix Advisory OVMBA-2012-0029

The following updated rpms for Oracle VM 3.1 have been uploaded to the 
Unbreakable Linux Network:

x86_64:
xen-4.1.2-18.el5.6.x86_64.rpm
xen-devel-4.1.2-18.el5.6.x86_64.rpm
xen-tools-4.1.2-18.el5.6.x86_64.rpm


SRPMS:
http://oss.oracle.com/oraclevm/server/3.1/SRPMS-updates/xen-4.1.2-18.el5.6.src.rpm



Description of changes:

[4.1.2-18.el5.6 ]
- fix hotplug script locking using flock
   We already has a fix there, but this one is better:
   1. No timeout: in the old one, if one process holding the lock more 
than 100
   seconds, other processes could steal it.
   2. No leftovers: if a process holding this lock is dead, it will 
close the lock
   file, so it will not affect other processes.
   This patch is from OL5U8 Red Hat xen.
   Signed-off-by: Zhigang Wang <zhigang.x.wang at oracle.com>
   Signed-off-by: Adnan Misherfi <adnan.misherfi at oracle.com>
   Backported-by: Joe Jin <joe.jin at oracle.com> [bug 14216774]

[4.1.2-18.el5.5 ]
- tools/hotplug: fix locking
   The current locking implementation would allow two processes get the lock
   simultaneously:
   ++ echo 16741: /etc/xen/scripts/block
   ++ cut -d : -f 1
   + local pid=16741
   + '[' -n 16741 -a 16741 '!=' unknown -a '!' -f /proc/16741/status ']'
   + '[' 5 -gt 5 ']'
   + sleep 0
   + retries=6
+ '[' 6 -lt 100 ']'
   + mkdir /var/run/xen-hotplug/block
   ++ _lock_owner /var/run/xen-hotplug/block
   ++ cat /var/run/xen-hotplug/block/owner
   + local 'new_owner=16741: /etc/xen/scripts/block'
   + '[' '16741: /etc/xen/scripts/block' '!=' '16741: 
/etc/xen/scripts/block' ']'
   ++ echo 16741: /etc/xen/scripts/block
   ++ cut -d : -f 1
   + local pid=16741
   + '[' -n 16741 -a 16741 '!=' unknown -a '!' -f /proc/16741/status ']'
   + '[' 6 -gt 5 ']'
   + sleep 1
   + do_or_die losetup -r /dev/loop27 
/OVS/Repositories/0004fb00000300000aac184a9bbab7a9/VirtualDisks/0004fb00001200002b0ef651033c8381.img
   + losetup -r /dev/loop27 
/OVS/Repositories/0004fb00000300000aac184a9bbab7a9/VirtualDisks/0004fb00001200002b0ef651033c8381.img
   + do_or_die losetup -r /dev/loop27 
/OVS/Repositories/0004fb00000300000aac184a9bbab7a9/VirtualDisks/0004fb0000120000143173990458f2a7.img
   + losetup -r /dev/loop27 
/OVS/Repositories/0004fb00000300000aac184a9bbab7a9/VirtualDisks/0004fb0000120000143173990458f2a7.img
   + xenstore_write backend/vbd/33/51920/node /dev/loop27
   + _xenstore_write backend/vbd/33/51920/node /dev/loop27
   + log debug 'Writing backend/vbd/33/51920/node' '/dev/loop27 to 
xenstore.'
   + local level=debug
   + shift
   + logger -p daemon.debug -- /etc/xen/scripts/block: 'Writing 
backend/vbd/33/51920/node' '/dev/loop27 to xenstore.'
   ioctl: LOOP_SET_FD: Device or resource busy
   + xenstore-write backend/vbd/33/51920/node /dev/loop27
   + fatal losetup -r /dev/loop27 
'/OVS/Repositories/0004fb00000300000aac184a9bbab7a9/VirtualDisks/0004fb0000120000143173990458f2a7.img 
failed'
   This patch removes the ownner support and fixed this issue per my test.
   Signed-off-by: Zhigang Wang <zhigang.x.wang at oracle.com>
   Signed-off-by: Adnan Misherfi <adnan.misherfi at oracle.com>
   Backported-by: Joe Jin <joe.jin at oracle.com> [bug 14216774]

[4.1.2-18.el5.4 ]
- Add restore (and migrate) functionality back in for PV 
superpage-enabled guests.
   Backported-by: Joe Jin <joe.jin at oracle.com> [bug 14192660]