Re: [Qemu-devel] [RFC 0/2] vhost-vfio: introduce mdev based HW vhost backend

["Liang, Cunming" <cunming.liang@intel.com>]

> -----Original Message-----
> From: Jason Wang [mailto:jasowang@redhat.com]
> Sent: Thursday, November 8, 2018 2:16 AM
> To: Liang, Cunming <cunming.liang@intel.com>; Wang, Xiao W
> <xiao.w.wang@intel.com>; mst@redhat.com; alex.williamson@redhat.com
> Cc: qemu-devel@nongnu.org; Bie, Tiwei <tiwei.bie@intel.com>; Ye, Xiaolong
> <xiaolong.ye@intel.com>; Wang, Zhihong <zhihong.wang@intel.com>; Daly, Dan
> <dan.daly@intel.com>
> Subject: Re: [RFC 0/2] vhost-vfio: introduce mdev based HW vhost backend
> 
> 
> On 2018/11/7 下午11:08, Liang, Cunming wrote:
> >>>> believe.
> >>> [LC] Agreed, so it reuses CMD defined by vhost-kernel ioctl. But
> >>> VFIO provides
> >> device specific things (e.g. DMAR, INTR and etc.) which is the extra
> >> APIs being introduced by this transport.
> >>
> >>
> >> I'm not quite sure I understand here. I think having vhost-kernel
> >> compatible ioctl does not conflict of using VFIO ioctl like DMA or INTR?
> >>
> >> Btw, VFIO DMA ioctl is even not a must from my point of view,
> >> vhost-mdev can forward the mem table information to device driver and
> >> let it call DMA API to map/umap pages.
> > [LC] If not regarding vhost-mdev as a device, then forward mem table won't be a
> concern.
> > If introducing a new mdev bus driver (vhost-mdev) which allows mdev instance to
> be a new type of provider for vhost-kernel. It becomes a pretty good alternative to
> fully leverage vhost-kernel ioctl.
> > I'm not sure it's the same view as yours when you says reusing vhost-kernel ioctl.
> >
> 
> Yes it is.
[LC] It sounds a pretty good idea to me. Let us spend some time to figure out the next level detail, and sync-up further plan in community call. :)

> 
> Thanks