From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:54260) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bpL7m-0002C8-5J for qemu-devel@nongnu.org; Wed, 28 Sep 2016 16:06:39 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bpL7i-00086y-TT for qemu-devel@nongnu.org; Wed, 28 Sep 2016 16:06:38 -0400 Received: from hqemgate14.nvidia.com ([216.228.121.143]:11059) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bpL7i-00086m-LI for qemu-devel@nongnu.org; Wed, 28 Sep 2016 16:06:34 -0400 Date: Wed, 28 Sep 2016 13:06:31 -0700 From: Neo Jia Message-ID: <20160928200631.GC26800@nvidia.com> References: <20160920084323.55a032fc@t450s.home> <20160920105025.4ef2cd40@t450s.home> <20160921130353.62bf309c@t450s.home> <11355037-d88e-0f28-bd07-14a7c86f85b0@nvidia.com> <20160922081921.57d31e47@t450s.home> <20160922142638.GR352@redhat.com> <20160928192233.GA25369@nvidia.com> <20160928135547.280daff0@t450s.home> MIME-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Disposition: inline Content-Transfer-Encoding: quoted-printable In-Reply-To: <20160928135547.280daff0@t450s.home> Subject: Re: [Qemu-devel] [libvirt] [RFC v2] libvirt vGPU QEMU integration List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Alex Williamson Cc: "Daniel P. Berrange" , Kirti Wankhede , Andy Currid , "Tian, Kevin" , "libvir-list@redhat.com" , qemu-devel , "Song, Jike" , Gerd Hoffmann , Paolo Bonzini , "bjsdjshi@linux.vnet.ibm.com" On Wed, Sep 28, 2016 at 01:55:47PM -0600, Alex Williamson wrote: > On Wed, 28 Sep 2016 12:22:35 -0700 > Neo Jia wrote: >=20 > > On Thu, Sep 22, 2016 at 03:26:38PM +0100, Daniel P. Berrange wrote: > > > On Thu, Sep 22, 2016 at 08:19:21AM -0600, Alex Williamson wrote: =20 > > > > On Thu, 22 Sep 2016 09:41:20 +0530 > > > > Kirti Wankhede wrote: > > > > =20 > > > > > >>>>> My concern is that a type id seems arbitrary but we're spec= ifying that > > > > > >>>>> it be unique. We already have something unique, the name. = So why try > > > > > >>>>> to make the type id unique as well? A vendor can accidenta= lly create > > > > > >>>>> their vendor driver so that a given name means something ve= ry > > > > > >>>>> specific. On the other hand they need to be extremely deli= berate to > > > > > >>>>> coordinate that a type id means a unique thing across all t= heir product > > > > > >>>>> lines. > > > > > >>>>> =20 > > > > > >>>> > > > > > >>>> Let me clarify, type id should be unique in the list of > > > > > >>>> mdev_supported_types. You can't have 2 directories in with s= ame name. =20 > > > > > >>> > > > > > >>> Of course, but does that mean it's only unique to the machine= I'm > > > > > >>> currently running on? Let's say I have a Tesla P100 on my sy= stem and > > > > > >>> type-id 11 is named "GRID-M60-0B". At some point in the futu= re I > > > > > >>> replace the Tesla P100 with a Q1000 (made up). Is type-id 11= on that > > > > > >>> new card still going to be a "GRID-M60-0B"? If not then we'v= e based > > > > > >>> our XML on the wrong attribute. If the new device does not s= upport > > > > > >>> "GRID-M60-0B" then we should generate an error, not simply in= itialize > > > > > >>> whatever type-id 11 happens to be on this new card. > > > > > >>> =20 > > > > > >> > > > > > >> If there are 2 M60 in the system then you would find '11' type= directory > > > > > >> in mdev_supported_types of both M60. If you have P100, '11' ty= pe would > > > > > >> not be there in its mdev_supported_types, it will have differe= nt types. > > > > > >> > > > > > >> For example, if you replace M60 with P100, but XML is not upda= ted. XML > > > > > >> have type '11'. When libvirt would try to create mdev device, = libvirt > > > > > >> would have to find 'create' file in sysfs in following directo= ry format: > > > > > >> > > > > > >> --- mdev_supported_types > > > > > >> |-- 11 > > > > > >> | |-- create > > > > > >> > > > > > >> but now for P100, '11' directory is not there, so libvirt shou= ld throw > > > > > >> error on not able to find '11' directory. =20 > > > > > >=20 > > > > > > This really seems like an accident waiting to happen. What hap= pens > > > > > > when the user replaces their M60 with an Intel XYZ device that = happens > > > > > > to expose a type 11 mdev class gpu device? How is libvirt supp= osed to > > > > > > know that the XML used to refer to a GRID-M60-0B and now it's a= n > > > > > > INTEL-IGD-XYZ? Doesn't basing the XML entry on the name and re= moving > > > > > > yet another arbitrary requirement that we have some sort of glo= bally > > > > > > unique type-id database make a lot of sense? The same issue ap= plies > > > > > > for simple debug-ability, if I'm reviewing the XML for a domain= and the > > > > > > name is the primary index for the mdev device, I know what it i= s. > > > > > > Seeing type-id=3D'11' is meaningless. > > > > > > =20 > > > > >=20 > > > > > Let me clarify again, type '11' is a string that vendor driver wo= uld > > > > > define (see my previous reply below) it could be "11" or "GRID-M6= 0-0B". > > > > > If 2 vendors used same string we can't control that. right? > > > > >=20 > > > > > =20 > > > > > >>>> Lets remove 'id' from type id in XML if that is the concern.= Supported > > > > > >>>> types is going to be defined by vendor driver, so let vendor= driver > > > > > >>>> decide what to use for directory name and same should be use= d in device > > > > > >>>> xml file, it could be '11' or "GRID M60-0B": > > > > > >>>> > > > > > >>>> > > > > > >>>> my-vgpu > > > > > >>>> pci_0000_86_00_0 > > > > > >>>> > > > > > >>>> > > > > > >>>> ... > > > > > >>>> > > > > > >>>> =20 > > > >=20 > > > > Then let's get rid of the 'name' attribute and let the sysfs direct= ory > > > > simply be the name. Then we can get rid of 'type' altogether so we > > > > don't have this '11' vs 'GRID-M60-0B' issue. Thanks, =20 > > >=20 > > > That sounds nice to me - we don't need two unique identifiers if > > > one will do. =20 > >=20 > > Hi Alex and Daniel, > >=20 > > I just had some internal discussions here within NVIDIA and found out t= hat > > actually the name/label potentially might not be unique and the "id" wi= ll be.=20 > > So I think we still would like to keep both so the id is the programmat= ic id > > and the name/label is a human readable string for it, which might get c= hanged to > > be non-unique by outside of engineering. >=20 > I think your discovery only means that for your vendor driver, the name > will be "11" (as a string). Perhaps you'd like some sort of vendor > provided description within each type, but I am not in favor of having > an arbitrary integer value imply something specific within the sysfs > interface. IOW, the NVIDIA vendor driver should be able to create: >=20 > 11 > =E2=94=9C=E2=94=80=E2=94=80 create > =E2=94=9C=E2=94=80=E2=94=80 description > =E2=94=9C=E2=94=80=E2=94=80 etc > =E2=94=94=E2=94=80=E2=94=80 resolution >=20 > While Intel might create: >=20 > Skylake-vGPU > =E2=94=9C=E2=94=80=E2=94=80 create > =E2=94=9C=E2=94=80=E2=94=80 description > =E2=94=9C=E2=94=80=E2=94=80 etc > =E2=94=94=E2=94=80=E2=94=80 resolution >=20 > Maybe "description" is optional for vendors that use useful names? > Thanks, I think we should be able to have a unique vendor type string instead of an arbitrary integer value there as long as we are allowed to have a descripti= on field that can be used to show to the end user as "name / label".=20 Thanks, Neo >=20 > Alex