All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Tian, Kevin" <kevin.tian@intel.com>
To: 'Alex Williamson' <alex.williamson@redhat.com>
Cc: Kirti Wankhede <kwankhede@nvidia.com>,
	"libvir-list@redhat.com" <libvir-list@redhat.com>,
	qemu-devel <qemu-devel@nongnu.org>,
	Paolo Bonzini <pbonzini@redhat.com>, Neo Jia <cjia@nvidia.com>,
	Andy Currid <ACurrid@nvidia.com>,
	"Song, Jike" <jike.song@intel.com>,
	"bjsdjshi@linux.vnet.ibm.com" <bjsdjshi@linux.vnet.ibm.com>,
	Gerd Hoffmann <kraxel@redhat.com>
Subject: Re: [Qemu-devel] [RFC v2] libvirt vGPU QEMU integration
Date: Thu, 22 Sep 2016 02:59:56 +0000	[thread overview]
Message-ID: <AADFC41AFE54684AB9EE6CBC0274A5D18DF8704D@SHSMSX101.ccr.corp.intel.com> (raw)
In-Reply-To: <AADFC41AFE54684AB9EE6CBC0274A5D18DF86F5F@SHSMSX101.ccr.corp.intel.com>

> From: Tian, Kevin
> Sent: Thursday, September 22, 2016 10:33 AM
> 
> >
> > > Another example is class-specific attributes such
> > > as 'resolution', etc. which you listed under 'display class'. All those
> > > attributes should be moved to mdev directory. Only class ID is
> > > required under each type.
> >
> > Can you elaborate on what you're proposing here?  If we don't have
> > attributes like 'resolution' under the type-id then we can't describe
> > to the user the features of the mdev before we create it.  I don't
> > think anybody wants a 'create it and find out' type of interface.
> > Thanks,
> >
> 
> I think such way would be racy. What about two clients creating mdev
> devices simultaneously? How to guarantee their configuration of class
> specific attributes not mutual-impacted since before creation any such
> configuration would be global under that type?
> 
> My feeling is that we'd better keep create simple - just write a UUID
> to "type-id/create". Any class-specific attribute, if we really want to
> support, should be able to be individually configured with required
> resource allocated incrementally on top of basic resources allocated
> at create stage. Then libvirt can set those attributes between create
> and open a mdev device. If this direction is accepted, then naturally
> such attributes should be put under mdev directory. Possibly we
> don't need a class description under type-id. libvirt just checks directly
> whether any known class represented in each mdev directory (vendor
> driver will expose it on demand), and then operate attributes under
> that class.
> 

Have a better understanding of your concern now. User needs to know
and set a list of attributes before requesting to create a new mdev
device. From this angle all attributes should be enumerated per type-id.
But as I explained above, writing multiple sysfs nodes to create a
mdev device is racy. We either need a way to guarantee transactional
operations on those nodes, or having type-id directory to only expose
supported attributes while programming those attributes has to be 
through per-mdev directory after mdev device is created...

Thanks
Kevin

  parent reply	other threads:[~2016-09-22  3:00 UTC|newest]

Thread overview: 62+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-09-19 20:35 [Qemu-devel] [RFC v2] libvirt vGPU QEMU integration Kirti Wankhede
2016-09-19 21:36 ` Alex Williamson
2016-09-19 21:50   ` Paolo Bonzini
2016-09-19 22:25     ` Alex Williamson
2016-09-20 14:35       ` Kirti Wankhede
2016-09-20 14:41         ` Daniel P. Berrange
2016-09-20 14:49           ` Paolo Bonzini
2016-09-20 14:58             ` Daniel P. Berrange
2016-09-20 15:05               ` Paolo Bonzini
2016-09-20 15:14                 ` Daniel P. Berrange
2016-09-20 16:31                   ` Kirti Wankhede
2016-09-20 16:36                     ` Daniel P. Berrange
2016-09-20 16:42                       ` Kirti Wankhede
2016-09-20 16:44                         ` Daniel P. Berrange
2016-09-20 16:46                     ` Daniel P. Berrange
2016-09-20 17:21                   ` Paolo Bonzini
2016-09-21  8:34                     ` Daniel P. Berrange
2016-09-20 14:52         ` Alex Williamson
2016-09-20  1:25   ` Tian, Kevin
2016-09-20 14:21   ` Kirti Wankhede
2016-09-20 14:43     ` Alex Williamson
2016-09-20 16:23       ` Kirti Wankhede
2016-09-20 16:50         ` Alex Williamson
2016-09-21 18:34           ` Kirti Wankhede
2016-09-21 19:03             ` Alex Williamson
2016-09-22  4:11               ` Kirti Wankhede
2016-09-22 14:19                 ` Alex Williamson
2016-09-22 14:26                   ` [Qemu-devel] [libvirt] " Daniel P. Berrange
2016-09-28 19:22                     ` Neo Jia
2016-09-28 19:45                       ` Tian, Kevin
2016-09-28 19:59                         ` Neo Jia
2016-09-28 20:31                           ` Laine Stump
2016-09-28 20:47                             ` Neo Jia
2016-09-28 22:49                           ` Alex Williamson
2016-09-28 19:55                       ` Alex Williamson
2016-09-28 20:06                         ` Neo Jia
2016-09-28 22:39                           ` Alex Williamson
2016-09-29  8:03                       ` Daniel P. Berrange
2016-09-29  8:12                         ` Neo Jia
2016-09-29 14:22               ` [Qemu-devel] " Kirti Wankhede
2016-09-21  4:10         ` Tian, Kevin
2016-09-21  4:43           ` Alex Williamson
2016-09-22  2:43             ` Tian, Kevin
2016-09-22 19:25         ` Tian, Kevin
2016-09-23 18:34           ` Kirti Wankhede
2016-09-21  3:56     ` Tian, Kevin
2016-09-21  4:36       ` Alex Williamson
2016-09-22  2:33         ` Tian, Kevin
2016-09-22  3:01           ` Alex Williamson
2016-09-22  3:42             ` Tian, Kevin
     [not found]         ` <AADFC41AFE54684AB9EE6CBC0274A5D18DF86F5F@SHSMSX101.ccr.corp.intel.com>
2016-09-22  2:59           ` Tian, Kevin [this message]
2016-09-20  1:37 ` Tian, Kevin
2016-09-20  9:47 ` Daniel P. Berrange
2016-09-28 19:48   ` Neo Jia
2016-09-29  8:06     ` Daniel P. Berrange
2016-09-29 14:35       ` Tian, Kevin
2016-09-29 14:38         ` Daniel P. Berrange
2016-09-29 14:42           ` Tian, Kevin
2016-09-30  5:19             ` Kirti Wankhede
2016-10-03  8:20               ` Kirti Wankhede
2016-10-07  5:16                 ` Kirti Wankhede
2016-10-07 19:09                   ` Alex Williamson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=AADFC41AFE54684AB9EE6CBC0274A5D18DF8704D@SHSMSX101.ccr.corp.intel.com \
    --to=kevin.tian@intel.com \
    --cc=ACurrid@nvidia.com \
    --cc=alex.williamson@redhat.com \
    --cc=bjsdjshi@linux.vnet.ibm.com \
    --cc=cjia@nvidia.com \
    --cc=jike.song@intel.com \
    --cc=kraxel@redhat.com \
    --cc=kwankhede@nvidia.com \
    --cc=libvir-list@redhat.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.