kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Paolo Bonzini <pbonzini@redhat.com>
To: Xiao Guangrong <guangrong.xiao@linux.intel.com>,
	Jike Song <jike.song@intel.com>
Cc: Alex Williamson <alex.williamson@redhat.com>,
	kwankhede@nvidia.com, cjia@nvidia.com, kevin.tian@intel.com,
	kvm@vger.kernel.org
Subject: Re: [v3 4/5] vfio: implement APIs to set/put kvm to/from vfio group
Date: Wed, 9 Nov 2016 14:31:01 +0100	[thread overview]
Message-ID: <02c27131-80fd-b882-1838-fe56f3f871d3@redhat.com> (raw)
In-Reply-To: <3e5299bc-e08e-f072-1001-7e0432cb1ca8@linux.intel.com>



On 09/11/2016 14:06, Xiao Guangrong wrote:
> 
> 
> On 11/09/2016 08:49 PM, Jike Song wrote:
> 
>> +void vfio_group_attach_kvm(struct vfio_group *group, struct kvm *kvm,
>> +        void (*fn)(struct kvm *))
>> +{
>> +    mutex_lock(&group->udata.lock);
> 
> This lock is needed, please see below.

*not* needed I guess.

>> +
>> +    fn(kvm);
>> +    blocking_notifier_call_chain(&group->udata.notifier,
>> +                VFIO_GROUP_NOTIFY_ATTACH_KVM, kvm);
> 
> As this is a callback before KVM releases its last refcount, i do not
> think vendor driver need to get additional KVM refcount.

The *group* driver doesn't need it indeed.  The mdev vendor driver
however does, so it will use kvm_get_kvm under its own mutex.  That is:

- attach kvm

	mutex_lock(mdev_driver->lock);
	mdev_driver->kvm = opaque;
	kvm_get_kvm(mdev_driver->kvm);
	mutex_unlock(mdev_driver->lock);

- detach kvm

	mutex_lock(mdev_driver->lock);
	kvm_put_kvm(mdev_driver->kvm);
	WARN_ON(mdev_driver->kvm != opaque);
	mdev_driver->kvm = NULL;
	mutex_unlock(mdev_driver->lock);

- use kvm

	mutex_lock(mdev_driver->lock);
	kvm = mdev_driver->kvm;
	...
	mutex_unlock(mdev_driver->lock);

or if safe:

	mutex_lock(mdev_driver->lock);
	kvm = mdev_driver->kvm;
	kvm_get_kvm(kvm);
	mutex_unlock(mdev_driver->lock);
	...
	kvm_put_kvm(kvm);

Thanks,

Paolo

  reply	other threads:[~2016-11-09 13:31 UTC|newest]

Thread overview: 34+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-10-31  6:35 [v3 0/5] plumb kvm/vfio to allow getting kvm from vfio_group Jike Song
2016-10-31  6:35 ` [v3 1/5] vfio: Rearrange functions to get vfio_group from dev Jike Song
2016-10-31  6:35 ` [v3 2/5] vfio: export functions to get vfio_group from device and put it Jike Song
2016-10-31  6:35 ` [v3 3/5] KVM: move kvm_get_kvm to kvm_host.h Jike Song
2016-10-31  8:33   ` Paolo Bonzini
2016-10-31  6:35 ` [v3 4/5] vfio: implement APIs to set/put kvm to/from vfio group Jike Song
2016-11-07 18:04   ` Alex Williamson
2016-11-07 18:10     ` Paolo Bonzini
2016-11-07 18:28       ` Alex Williamson
2016-11-07 20:45         ` Paolo Bonzini
2016-11-09 12:49           ` Jike Song
2016-11-09 13:06             ` Xiao Guangrong
2016-11-09 13:31               ` Paolo Bonzini [this message]
2016-11-09 14:00                 ` Xiao Guangrong
2016-11-09 14:28                   ` Paolo Bonzini
2016-11-10  4:13                   ` Jike Song
2016-11-09 17:53             ` Alex Williamson
2016-11-10  4:10               ` Jike Song
2016-11-10  6:04                 ` Jike Song
2016-11-10 15:37                   ` Alex Williamson
2016-11-11  7:29                     ` Jike Song
2016-11-14 10:19               ` Jike Song
2016-11-14 15:52                 ` Alex Williamson
2016-11-09  2:28         ` Jike Song
2016-11-09  2:52           ` Xiao Guangrong
2016-11-09  3:07             ` Jike Song
2016-10-31  6:35 ` [v3 5/5] KVM: set/clear kvm to/from vfio group during add/delete Jike Song
2016-10-31  8:33   ` Paolo Bonzini
2016-10-31  7:06 ` [v3 0/5] plumb kvm/vfio to allow getting kvm from vfio_group Xiao Guangrong
2016-10-31  7:24   ` Jike Song
2016-10-31  7:24     ` Xiao Guangrong
2016-10-31  7:30       ` Jike Song
2016-10-31  7:35         ` Xiao Guangrong
2016-11-02  1:06 ` Jike Song

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=02c27131-80fd-b882-1838-fe56f3f871d3@redhat.com \
    --to=pbonzini@redhat.com \
    --cc=alex.williamson@redhat.com \
    --cc=cjia@nvidia.com \
    --cc=guangrong.xiao@linux.intel.com \
    --cc=jike.song@intel.com \
    --cc=kevin.tian@intel.com \
    --cc=kvm@vger.kernel.org \
    --cc=kwankhede@nvidia.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).