GVT State Save/Restore

Julian Stecklina julian.stecklina at cyberus-technology.de
Mon Mar 23 14:42:53 UTC 2020


On Sun, 2020-03-22 at 14:45 +0800, Zhenyu Wang wrote:
> On 2020.03.20 11:49:46 +0100, Julian Stecklina wrote:
> > Hi everyone,
> > 
> > I'm researching what needs to be done to enable VMs with a GVT device to be
> > saved to disk and restored. This is similar to live migration in the sense
> > that
> > state of the vGPU needs to be serialized and deserialized, but simpler,
> > because
> > no dirty page tracking is needed.
[...]
> > It doesn't seem like any of the patches actually made it upstream. Are there
> > some newer PoC patches to look at or did I miss some functionality in the
> > mediator?
> > 
> 
> cc Yan, who is working on upstream VFIO interface and GVT change for
> migration.
> 
> We have an old topic branch on 
> https://github.com/intel/gvt-linux/commits/topic/gvt-migration-yan
> which had working GVT migration support based on older interface proposal.
> 

Excellent! We are going to check this out.

Thanks,
Julian



More information about the intel-gvt-dev mailing list