RE: [PATCH V4 00/43] Live update: vfio and iommufd

Duan, Zhenzhong posted 43 patches 5 months, 2 weeks ago
Only 0 patches received!
RE: [PATCH V4 00/43] Live update: vfio and iommufd
Posted by Duan, Zhenzhong 5 months, 2 weeks ago

>-----Original Message-----
>From: Steve Sistare <steven.sistare@oracle.com>
>Subject: [PATCH V4 00/43] Live update: vfio and iommufd
>
>Support vfio and iommufd devices with the cpr-transfer live migration mode.
>Devices that do not support live migration can still support cpr-transfer,
>allowing live update to a new version of QEMU on the same host, with no loss
>of guest connectivity.

Just curious. My understanding is: for device not supporting live migration, device
will not be stopped during cpr-transfer and there is no device state saving/restore.
But for device supporting live migration, it will be stopped and device state is saved
in source and restored in destination, it that right?

Thanks
Zhenzhong
Re: [PATCH V4 00/43] Live update: vfio and iommufd
Posted by Steven Sistare 5 months, 2 weeks ago
On 6/3/2025 8:09 AM, Duan, Zhenzhong wrote:
>> -----Original Message-----
>> From: Steve Sistare <steven.sistare@oracle.com>
>> Subject: [PATCH V4 00/43] Live update: vfio and iommufd
>>
>> Support vfio and iommufd devices with the cpr-transfer live migration mode.
>> Devices that do not support live migration can still support cpr-transfer,
>> allowing live update to a new version of QEMU on the same host, with no loss
>> of guest connectivity.
> 
> Just curious. My understanding is: for device not supporting live migration, device
> will not be stopped during cpr-transfer and there is no device state saving/restore.

Yes.

> But for device supporting live migration, it will be stopped and device state is saved
> in source and restored in destination, it that right?

As currently written, yes.
However, it may be faster and more reliable if I disable savevm_vfio_handlers and the
associated notifiers during CPR, and only use the CPR mechanism for preserving the device.

- Steve