Re: [Qemu-devel] [PATCH v3 0/7] cpu: add i386 cpu hot remove support

[Dou Liyang <douly.fnst@cn.fujitsu.com>]

Hi,

Thanks to test.

在 2016年06月22日 20:12, Eduardo Otubo 写道:
> On Mon, Jul 13, 2015 at 09=45=57AM +0800, Zhu Guihua wrote:
>> On 07/09/2015 10:25 PM, Eduardo Otubo wrote:
>>> On Fri, Jun 26, 2015 at 11=37=43AM +0800, Zhu Guihua wrote:
>>>> Hi,
>>>> On 06/24/2015 09:28 PM, Eduardo Otubo wrote:
>>>>> Hello Zhu,
>>>>>
>>>>> Are you still working on this feature? Could you provide a rebased
>>>>> version of this series?
>>>> Sorry for late reply.
>>>>
>>>> Yes, we are still working on this feature.
>>>>
>>>> I have updated my github, you can get the rebased version from it.
>>>>
>>>> https://github.com/zhuguihua/qemu.git  cpu-hotplug
>>> Hi, thanks a lot for rebasing it. I didn't have time to test it, though.
>>> I appologise.
>>>
>>> Regarding your branch, your plan is to stick with device_del and
>>> object_del interface?
>> Yeah, we will only stick with device_del interface.
>> There is no need to use object_del interface for cpu hot remove.
>>
>> Thanks,
>> Zhu
>>
> This is almost a year old email, but I decided to git it a try anyway.
> Are you still working on this?
Yes, we are still working on this feature.

but, as we discuss almost one year age and the Abstract CPU core type is 
defined recently, this patch's QMP/HMP interface will be replace.

we are  following Igor's patches to see what we can do for this feature.
> I'm trying your patches and it seems I
> had a simple issue: EJ0 is sent, guest ack's and removes the CPU, but it
> get's stuck somewhere between the BIOS and Qemu. CPU is successfully
> removed from the guest, but it still appears on QOM:
>
>      (qemu) info cpus
>      * CPU #0: pc=0xffffffff8104f596 (halted) thread_id=22208
>        CPU #1: pc=0xffffffff8104f596 (halted) thread_id=22209
>        CPU #2: pc=0xffffffff810429c8 (halted) thread_id=22281
>      (qemu) info qom-tree
>      /machine (pc-i440fx-2.1-machine)
>        /fw_cfg (fw_cfg_io)
>          /fwcfg[0] (qemu:memory-region)
>      
>      [...]
>      
>          /cpu3 (SandyBridge-x86_64-cpu)
>      
>      [...]
>
> Do you have any clues?
we guess Qom may not be updated during the remove event. we will concern it.

Thanks,
Dou.
>