qemu-devel
[Top][All Lists]
Advanced

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

Re: [PATCH Kernel v22 0/8] Add UAPIs to support migration for VFIO devic


From: Yan Zhao
Subject: Re: [PATCH Kernel v22 0/8] Add UAPIs to support migration for VFIO devices
Date: Fri, 29 May 2020 00:15:17 -0400
User-agent: Mutt/1.9.4 (2018-02-28)

On Thu, May 28, 2020 at 04:59:06PM -0600, Alex Williamson wrote:
> On Wed, 27 May 2020 09:48:22 +0100
> "Dr. David Alan Gilbert" <dgilbert@redhat.com> wrote:
> > * Yan Zhao (yan.y.zhao@intel.com) wrote:
> > > BTW, for viommu, the downtime data is as below. under the same network
> > > condition and guest memory size, and no running dirty data/memory produced
> > > by device.
> > > (1) viommu off
> > > single-round dirty query: downtime ~100ms   
> > 
> > Fine.
> > 
> > > (2) viommu on
> > > single-round dirty query: downtime 58s   
> > 
> > Youch.
> 
> Double Youch!  But we believe this is because we're getting the dirty
> bitmap one IOMMU leaf page at a time, right?  We've enable the kernel
> to get a dirty bitmap across multiple mappings, but QEMU isn't yet
> taking advantage of it.  Do I have this correct?  Thanks,
>
Yes, I think so, but I haven't looked into it yet.

Thanks
Yan



reply via email to

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