Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Fri, 28 Jun 2019 09:56:08 -0600
From:   Alex Williamson <alex.williamson@redhat.com>
To:     Cornelia Huck <cohuck@redhat.com>
Cc:     kwankhede@nvidia.com, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] mdev: Send uevents around parent device registration
Message-ID: <20190628095608.7762d6d0@x1.home>
In-Reply-To: <20190627101914.32829440.cohuck@redhat.com>
References: <156155924767.11505.11457229921502145577.stgit@gimli.home>
        <20190627101914.32829440.cohuck@redhat.com>
Organization: Red Hat
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, 27 Jun 2019 10:19:14 +0200
Cornelia Huck <cohuck@redhat.com> wrote:

> On Wed, 26 Jun 2019 08:27:58 -0600
> Alex Williamson <alex.williamson@redhat.com> wrote:
> 
> > This allows udev to trigger rules when a parent device is registered
> > or unregistered from mdev.
> > 
> > Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
> > ---
> >  drivers/vfio/mdev/mdev_core.c |   10 ++++++++--
> >  1 file changed, 8 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/vfio/mdev/mdev_core.c b/drivers/vfio/mdev/mdev_core.c
> > index ae23151442cb..ecec2a3b13cb 100644
> > --- a/drivers/vfio/mdev/mdev_core.c
> > +++ b/drivers/vfio/mdev/mdev_core.c
> > @@ -146,6 +146,8 @@ int mdev_register_device(struct device *dev, const struct mdev_parent_ops *ops)
> >  {
> >  	int ret;
> >  	struct mdev_parent *parent;
> > +	char *env_string = "MDEV_STATE=registered";  
> 
> This string is probably reasonable enough.
> 
> > +	char *envp[] = { env_string, NULL };
> >  
> >  	/* check for mandatory ops */
> >  	if (!ops || !ops->create || !ops->remove || !ops->supported_type_groups)
> > @@ -196,7 +198,8 @@ int mdev_register_device(struct device *dev, const struct mdev_parent_ops *ops)
> >  	list_add(&parent->next, &parent_list);
> >  	mutex_unlock(&parent_list_lock);
> >  
> > -	dev_info(dev, "MDEV: Registered\n");
> > +	kobject_uevent_env(&dev->kobj, KOBJ_CHANGE, envp);  
> 
> I also agree with the positioning here.
> 
> > +
> >  	return 0;
> >  
> >  add_dev_err:
> > @@ -220,6 +223,8 @@ EXPORT_SYMBOL(mdev_register_device);
> >  void mdev_unregister_device(struct device *dev)
> >  {
> >  	struct mdev_parent *parent;
> > +	char *env_string = "MDEV_STATE=unregistered";  
> 
> Ok.
> 
> > +	char *envp[] = { env_string, NULL };
> >  
> >  	mutex_lock(&parent_list_lock);
> >  	parent = __find_parent_device(dev);
> > @@ -228,7 +233,6 @@ void mdev_unregister_device(struct device *dev)
> >  		mutex_unlock(&parent_list_lock);
> >  		return;
> >  	}
> > -	dev_info(dev, "MDEV: Unregistering\n");
> >  
> >  	list_del(&parent->next);
> >  	mutex_unlock(&parent_list_lock);
> > @@ -243,6 +247,8 @@ void mdev_unregister_device(struct device *dev)
> >  	up_write(&parent->unreg_sem);
> >  
> >  	mdev_put_parent(parent);
> > +
> > +	kobject_uevent_env(&dev->kobj, KOBJ_CHANGE, envp);  
> 
> I'm wondering whether we should indicate this uevent earlier: Once we
> have detached from the parent list, we're basically done for all
> practical purposes. So maybe move this right before we grab the
> unreg_sem?

That would make it a "this thing is about to go away" (ie.
"unregistering") rather than "this thing is gone" ("unregistered").  I
was aiming for the latter as the former just seems like it might make
userspace race to remove devices.  Note that I don't actually make use
of this event in mdevctl currently, so we could maybe save it for
later, but the symmetry seemed preferable.  Thanks,

Alex