qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH v2 1/9] cpus: Reclaim vCPU objects


From: Matthew Rosato
Subject: Re: [Qemu-devel] [PATCH v2 1/9] cpus: Reclaim vCPU objects
Date: Fri, 20 Nov 2015 08:37:40 -0500
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.3.0

On 11/19/2015 09:33 PM, Bharata B Rao wrote:
> On Thu, Nov 19, 2015 at 10:10:06AM -0500, Matthew Rosato wrote:
>> From: Gu Zheng <address@hidden>
>>
>> In order to deal well with the kvm vcpus (which can not be removed without 
>> any
>> protection), we do not close KVM vcpu fd, just record and mark it as stopped
>> into a list, so that we can reuse it for the appending cpu hot-add request if
>> possible. It is also the approach that kvm guys suggested:
>> https://www.mail-archive.com/address@hidden/msg102839.html
>>
>> Signed-off-by: Chen Fan <address@hidden>
>> Signed-off-by: Gu Zheng <address@hidden>
>> Signed-off-by: Zhu Guihua <address@hidden>
>> Signed-off-by: Bharata B Rao <address@hidden>
>>                [Explicit CPU_REMOVE() from qemu_kvm/tcg_destroy_vcpu()
>>                 isn't needed as it is done from cpu_exec_exit()]
> 
> I didn't look very closely but the patch that removes cpu from the list
> from cpu_exec_exit() isn't part of this series. The above change requires
> 
> https://lists.gnu.org/archive/html/qemu-devel/2015-08/msg00656.html
> 
> I have just cleaned that patch a bit and will be posting early next
> week with another patch that does CPU vmstate unregistration too from
> cpu_exec_exit(). I think since we do vmstate registration from cpu_exec_init()
> it makes sense to do unregistration from cpu_exec_exit() instead of
> archs doing it themselves. I had a version of this at
> 
> https://lists.gnu.org/archive/html/qemu-devel/2015-08/msg00649.html
> 
> With the above patch, you woudn't need 7/9 in this series.
> 

Hi Bharata -- Looking at the mailing list discussion from your patch
set, I got the impression that handling this in cpu_exec_exit() might
not be acceptable for all architectures.  So, my patch just tries to
handle the s390 case in patch 7/9, doing list removal and vmstate
unregistration.

FWIW, the 2 patches you referenced would be fine for s390, so if you can
get those approved I'd have no problem dropping 7/9 in favor of your
patches.

Matt




reply via email to

[Prev in Thread] Current Thread [Next in Thread]