On Wed, 2015-01-21 at 10:50 +0100, Peter Krempa wrote:
On Wed, Jan 21, 2015 at 16:20:16 +0800, Zhu Guihua wrote:
> Now qemu has already supported memory hotplug, so this patchset will make
> libvirt support hotplug memory device for qemu driver.
As I'm already working on this I can see a few problems with your
series.
>
> First, add two parameters slots and maxmem for memory hotplug, and the xml
> format can be like this.
> <memory slot='10'>102400</memory>
> <maxMemory>1024000</maxMemory>
>
> Second, memory device's xml format can be like this.
> <dimm driver='pc-dimm' addr='0' node='0'
slot='1'>
Your code states that addr=0 auto-allocates the address, but I don't see
any code that would retrieve it afterwards. Without that the migration
can't work reliably.
> <backend type='ram' size='131072'/>
Your code doesn't support specifying source nodes for the memory device.
This wouldn't work with libvirt with numa pinnnig enabled.
Aditionally the target numa node mode that is configured in the numatune
section is not honored.
The hugepage backend (memory-backend-file) part in your impl requires
the user to specify the path for the memory device. Libvirt has a lot
code that does this detection and it should be reused.
> </dimm>
I'm pretty far in my effort so I think it's not worth for you to
continue working on this feature.
Thank you for pointing out the problems, and we hope the feature could
be supported in libvirt as soon as possible.
So,
1) could you tell us your schedule for this feature?
2) When can you send the patch series to the community?
3) If there are any else work we can do for this feature?
Regards,
Zhu
Peter