Some relevant kernel log:

8f2f748b0656257153bcf0941df8d6060acc5ca6 CPU hotplug, cpusets, suspend: Don't 
touch cpusets during suspend/resume
4293f20c19f44ca66e5ac836b411d25e14b9f185 Revert "CPU hotplug, cpusets, suspend: 
Don't touch cpusets during suspend/resume"

$ git describe 8f2f748b0656257153bcf0941df8d6060acc5ca6
v3.3-rc4-36-g8f2f748

git describe 4293f20c19f44ca66e5ac836b411d25e14b9f185
v3.3-rc6-147-g4293f20

So it comes as no surprise that the bug still exists in upstream.  The
commit that was supposed to fix this was reverted due to side effects
(see comment #3).

The cpuset.sh script in this thread works as a workaround (for me at least):
https://www.redhat.com/archives/libvir-list/2012-April/msg00777.html

However, this is just a hack to get around a deficiency in the kernel.
The kernel shouldn't mangle the cpusets like this.

Thomas Gleixner (tglx) is working on refactoring cpu hotplug code, which
is used for suspend/resume support.  It is the hot removal of all cpus
except cpu 0 during suspend that is responsible for the cpusets being
reduced to only cpu 0.

https://lkml.org/lkml/2012/4/20/160

http://git.kernel.org/?p=linux/kernel/git/tip/tip.git;a=shortlog;h=refs/heads/smp/hotplug

I think he's just in the early stages though; nothing to fix this issue
yet.

So question:
Should the above mentioned cpuset.sh be packaged in libvirt-bin and installed 
at /usr/lib/pm-utils/sleep.d/XXcpuset.sh (whatever XX should be) in the 
meantime?  Even when this eventually gets fixed in the kernel, this script 
won't mess things up.  It'll just be redundant.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/993354

Title:
  cpuset for libvirt set to 0 after suspend/resume

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/993354/+subscriptions

-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to