From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:49797) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1f6TiA-0008LR-Cr for qemu-devel@nongnu.org; Thu, 12 Apr 2018 00:19:51 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1f6Ti7-0005SX-AA for qemu-devel@nongnu.org; Thu, 12 Apr 2018 00:19:50 -0400 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:47580 helo=mx1.redhat.com) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1f6Ti7-0005SK-4n for qemu-devel@nongnu.org; Thu, 12 Apr 2018 00:19:47 -0400 Date: Thu, 12 Apr 2018 07:19:38 +0300 From: "Michael S. Tsirkin" Message-ID: <20180412071805-mutt-send-email-mst@kernel.org> References: <20180411072027.5656-1-tiwei.bie@intel.com> <20180411161926-mutt-send-email-mst@kernel.org> <20180412011059.yywn73znjdip2cyv@debian> <20180412042724-mutt-send-email-mst@kernel.org> <20180412013942.egucc4isxkokta7z@debian> <20180412044404-mutt-send-email-mst@kernel.org> <20180412023505.xftmhguzu7j55nif@debian> <20180412061108-mutt-send-email-mst@kernel.org> <20180412062216-mutt-send-email-mst@kernel.org> <9314ebd5-8e92-41b1-efd1-4371177075a0@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <9314ebd5-8e92-41b1-efd1-4371177075a0@redhat.com> Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [RFC] vhost-user: introduce F_NEED_ALL_IOTLB protocol feature List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Jason Wang Cc: Tiwei Bie , cunming.liang@intel.com, qemu-devel@nongnu.org, peterx@redhat.com, zhihong.wang@intel.com, dan.daly@intel.com On Thu, Apr 12, 2018 at 11:43:41AM +0800, Jason Wang wrote: >=20 >=20 > On 2018=E5=B9=B404=E6=9C=8812=E6=97=A5 11:35, Michael S. Tsirkin wrote: > > > There are more advantages to using request with PASID: > > >=20 > > > You can use hardware support for nesting, having guest supply 1st l= evel > > > translation and host second level translation. > > >=20 > > > I actually had an idea to do something like this for AMD > > > and ARM which support nesting even for requests with PASID, > > > having intel benefit too would be nice. > > Something else to consider is implementing PRS capability. > >=20 > >=20 > > In theory this could then go like this: > >=20 > > - get page request from device > > - fetch request from VTD page tables > > - use response to issue a page response message > >=20 > >=20 > > This would match the current vhost-user model. >=20 > This requires IOMMU driver can forward this to VFIO and then VFIO can > forward it to userspace. Looks like a lot of changes and it would be ev= en > slower than what is proposed in this patch. >=20 > Thanks It would work better for a static table as only accessed pages would need to be sent. Slower for the dynamic case but dynamic case needs hardware support to work properly in any case. --=20 MST