From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alex Williamson Subject: Re: [libvirt] [PATCH v7 0/4] Add Mediated device support Date: Fri, 2 Sep 2016 13:15:07 -0600 Message-ID: <20160902131507.0e5aceb8@t450s.home> References: <1472097235-6332-1-git-send-email-kwankhede@nvidia.com> <20160830101638.49df467d@t450s.home> <78fedd65-6d62-e849-ff3b-d5105b2da816@redhat.com> <20160901105948.62f750aa@t450s.home> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Cc: "Song, Jike" , "cjia@nvidia.com" , "kvm@vger.kernel.org" , "libvir-list@redhat.com" , Michal Privoznik , "Tian, Kevin" , "qemu-devel@nongnu.org" , Kirti Wankhede , "kraxel@redhat.com" , Laine Stump , "pbonzini@redhat.com" , "bjsdjshi@linux.vnet.ibm.com" To: Laine Stump Return-path: In-Reply-To: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+gceq-qemu-devel=gmane.org@nongnu.org Sender: "Qemu-devel" List-Id: kvm.vger.kernel.org On Fri, 2 Sep 2016 13:55:19 -0400 Laine Stump wrote: > On 09/01/2016 12:59 PM, Alex Williamson wrote: > > On Thu, 1 Sep 2016 18:47:06 +0200 > > Michal Privoznik wrote: > > =20 > >> On 31.08.2016 08:12, Tian, Kevin wrote: =20 > >>>> From: Alex Williamson [mailto:alex.williamson@redhat.com] > >>>> Sent: Wednesday, August 31, 2016 12:17 AM > >>>> > >>>> Hi folks, > >>>> > >>>> At KVM Forum we had a BoF session primarily around the mediated devi= ce > >>>> sysfs interface. I'd like to share what I think we agreed on and the > >>>> "problem areas" that still need some work so we can get the thoughts > >>>> and ideas from those who weren't able to attend. > >>>> > >>>> DanPB expressed some concern about the mdev_supported_types sysfs > >>>> interface, which exposes a flat csv file with fields like "type", > >>>> "number of instance", "vendor string", and then a bunch of type > >>>> specific fields like "framebuffer size", "resolution", "frame rate > >>>> limit", etc. This is not entirely machine parsing friendly and sort= of > >>>> abuses the sysfs concept of one value per file. Example output taken > >>>> from Neo's libvirt RFC: > >>>> > >>>> cat /sys/bus/pci/devices/0000:86:00.0/mdev_supported_types > >>>> # vgpu_type_id, vgpu_type, max_instance, num_heads, frl_config, fram= ebuffer, > >>>> max_resolution > >>>> 11 ,"GRID M60-0B", 16, 2, 45, 512M, 2560= x1600 > >>>> 12 ,"GRID M60-0Q", 16, 2, 60, 512M, 2560= x1600 > >>>> 13 ,"GRID M60-1B", 8, 2, 45, 1024M, 2560= x1600 > >>>> 14 ,"GRID M60-1Q", 8, 2, 60, 1024M, 2560= x1600 > >>>> 15 ,"GRID M60-2B", 4, 2, 45, 2048M, 2560= x1600 > >>>> 16 ,"GRID M60-2Q", 4, 4, 60, 2048M, 2560= x1600 > >>>> 17 ,"GRID M60-4Q", 2, 4, 60, 4096M, 3840= x2160 > >>>> 18 ,"GRID M60-8Q", 1, 4, 60, 8192M, 3840= x2160 > >>>> > >>>> The create/destroy then looks like this: > >>>> > >>>> echo "$mdev_UUID:vendor_specific_argument_list" > > >>>> /sys/bus/pci/devices/.../mdev_create > >>>> > >>>> echo "$mdev_UUID:vendor_specific_argument_list" > > >>>> /sys/bus/pci/devices/.../mdev_destroy > >>>> > >>>> "vendor_specific_argument_list" is nebulous. > >>>> > >>>> So the idea to fix this is to explode this into a directory structur= e, > >>>> something like: > >>>> > >>>> =E2=94=9C=E2=94=80=E2=94=80 mdev_destroy > >>>> =E2=94=94=E2=94=80=E2=94=80 mdev_supported_types > >>>> =E2=94=9C=E2=94=80=E2=94=80 11 > >>>> =E2=94=82 =E2=94=9C=E2=94=80=E2=94=80 create > >>>> =E2=94=82 =E2=94=9C=E2=94=80=E2=94=80 description > >>>> =E2=94=82 =E2=94=94=E2=94=80=E2=94=80 max_instances > >>>> =E2=94=9C=E2=94=80=E2=94=80 12 > >>>> =E2=94=82 =E2=94=9C=E2=94=80=E2=94=80 create > >>>> =E2=94=82 =E2=94=9C=E2=94=80=E2=94=80 description > >>>> =E2=94=82 =E2=94=94=E2=94=80=E2=94=80 max_instances > >>>> =E2=94=94=E2=94=80=E2=94=80 13 > >>>> =E2=94=9C=E2=94=80=E2=94=80 create > >>>> =E2=94=9C=E2=94=80=E2=94=80 description > >>>> =E2=94=94=E2=94=80=E2=94=80 max_instances > >>>> > >>>> Note that I'm only exposing the minimal attributes here for simplici= ty, > >>>> the other attributes would be included in separate files and we would > >>>> require vendors to create standard attributes for common device clas= ses. =20 > >>> I like this idea. All standard attributes are reflected into this hie= rarchy. > >>> In the meantime, can we still allow optional vendor string in create > >>> interface? libvirt doesn't need to know the meaning, but allows upper > >>> layer to do some vendor specific tweak if necessary. =20 > >> This is not the best idea IMO. Libvirt is there to shadow differences > >> between hypervisors. While doing that, we often hide differences betwe= en > >> various types of HW too. Therefore in order to provide good abstraction > >> we should make vendor specific string as small as possible (ideally an > >> empty string). I mean I see it as bad idea to expose "vgpu_type_id" fr= om > >> example above in domain XML. What I think the better idea is if we let > >> users chose resolution and frame buffer size, e.g.: