[Qemu-devel] memory hotplug with 2.1.2

[Qemu-devel] memory hotplug with 2.1.2

[Mikhail Sennikovskii]

Hi all,

Trying to hotplug RAM with qemu 2.1.2 using

object_add memory-backend-ram,id=ram1,size=128M
device_add pc-dimm,id=dimm1,memdev=ram1

gives the following error:
{{{
'DIMM property node has value 0' which exceeds the number of numa nodes: 0
Device 'pc-dimm' could not be initialized
}}}

which goes away if I add a "dummy" numa node via command line at VM 
creation  -numa node,nodeid=0

This behavior was introduced by the commit 
9a72433843d912a45046959b1953861211d1838d, and disabling the check makes 
the hotplug work again for me.

Is this a regression, or just some memory hotplug semantic change? And 
in the latter case, could you specify the proper semantic I should use?

Thanks,
Mikhail

Re: [Qemu-devel] memory hotplug with 2.1.2

[Mikhail Sennikovskii]

Ah, just noticed there is a bug related to this: 
https://bugzilla.redhat.com/show_bug.cgi?id=1150930

Mikhail


On 17.10.2014 17:38, Mikhail Sennikovskii wrote:
> Hi all,
>
> Trying to hotplug RAM with qemu 2.1.2 using
>
> object_add memory-backend-ram,id=ram1,size=128M
> device_add pc-dimm,id=dimm1,memdev=ram1
>
> gives the following error:
> {{{
> 'DIMM property node has value 0' which exceeds the number of numa 
> nodes: 0
> Device 'pc-dimm' could not be initialized
> }}}
>
> which goes away if I add a "dummy" numa node via command line at VM 
> creation  -numa node,nodeid=0
>
> This behavior was introduced by the commit 
> 9a72433843d912a45046959b1953861211d1838d, and disabling the check 
> makes the hotplug work again for me.
>
> Is this a regression, or just some memory hotplug semantic change? And 
> in the latter case, could you specify the proper semantic I should use?
>
> Thanks,
> Mikhail
>

Re: [Qemu-devel] memory hotplug with 2.1.2

[Mikhail Sennikovskii]

[-- Attachment #1: Type: text/plain, Size: 428 bytes --]

Hi Andrey, thank you for your answer.

I know I can work around this by specifying a numa node.
My question is though is the current behaviour considered to be a bug, or
not. And if yes, when it is expected to get fixed.

Thanks,
Mikhail

On Fri, Oct 17, 2014 at 5:54 PM, Andrey Korolyov <andrey@xdel.ru> wrote:

> Please try to populate basic (single-node) NUMA topology to workaround
> this (or specify numa node for dimm).
>

[-- Attachment #2: Type: text/html, Size: 787 bytes --]

Re: [Qemu-devel] memory hotplug with 2.1.2

[zhanghailiang]

Hi,

Have you tried the latest qemu (you can git clone from http://git.qemu.org/qemu.git).

Actually, i have posted a patch fc50ff0666315be5120c70ad00cd0b0097484b84
     pc-dimm: Don't check dimm->node when there is non-NUMA config

     It should not break memory hotplug feature if there is non-NUMA option.

     This patch would also allow to use pc-dimm as replacement for initial memory
     for non-NUMA configs.

     Note: After this patch, the memory hotplug can work normally for Linux guest OS
     when there is non-NUMA option and NUMA option. But not support Windows guest OS
     to hotplug memory with no-NUMA config, actully, it's Windows limitation.

     Reviewed-By: Igor Mammedov <imammedo@redhat.com>
     Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
     Acked-by: Michael S. Tsirkin <mst@redhat.com>
     Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

I think this will fix your problem, it will support hotplug a pc-dimm without configure numa node.

Hope this can help you, Thanks

Best Regards,
zhanghailiang

On 2014/10/18 1:11, Mikhail Sennikovskii wrote:
> Hi Andrey, thank you for your answer.
>
> I know I can work around this by specifying a numa node.
> My question is though is the current behaviour considered to be a bug, or
> not. And if yes, when it is expected to get fixed.
>
> Thanks,
> Mikhail
>
> On Fri, Oct 17, 2014 at 5:54 PM, Andrey Korolyov <andrey@xdel.ru> wrote:
>
>> Please try to populate basic (single-node) NUMA topology to workaround
>> this (or specify numa node for dimm).
>>
>