[PATCH] drm/i915/gvt: Add mdev device attribute group

[Tian, Kevin]

> From: Zhenyu Wang [mailto:zhenyuw at linux.intel.com]
> Sent: Monday, March 13, 2017 2:40 PM
> 
> On 2017.03.13 06:34:35 +0000, Tian, Kevin wrote:
> > > From: Zhenyu Wang [mailto:zhenyuw at linux.intel.com]
> > > Sent: Monday, March 13, 2017 2:30 PM
> > >
> > > On 2017.03.13 06:26:49 +0000, Tian, Kevin wrote:
> > > > > From: Zhenyu Wang
> > > > > Sent: Wednesday, March 08, 2017 4:42 PM
> > > > >
> > > > > On 2017.03.08 16:19:48 +0800, Zhenyu Wang wrote:
> > > > > > This adds initial attribute group for mdev to hold vGPU related
> > > > > > for each mdev device, currently just vGPU id is shown.
> > > > > >
> > > > > > Signed-off-by: Zhenyu Wang <zhenyuw at linux.intel.com>
> > > > > > ---
> > > > >
> > > > > As this will try to create gvt specific attrs directory for each mdev,
> > > > > named for "vgpu" now, if anyone prefers other name like "intel_vgpu"?
> > > >
> > > > vgpu is OK, but curious what's the point of reporting such internal ID?
> > > >
> > >
> > > One reason is that for driver message which only contains vGPU id, we need
> > > a way to link that for mdev device which is created by UUID, otherwise can't
> > > tell which VM the driver message is for.
> > >
> >
> > what about printing vGPU id when the vGPU instance is created?
> >
> 
> I think an interface to read that is more clean than parsing log.
> And as this will create attrs group for each mdev, there will be
> more attrs to add in future for vgpu state or even control, etc.
> 

It depends on how you want to use that vGPU id. If the only purpose
is to use it when reading driver message, I don't think it a big burden
for user to further understand one more message before reading following.

On the other hand, if the vGPU id is required as an input parameter to
other interfaces, then it's necessary. But in this manner I suppose the
UUID already serves the purpose.

Better finding another example? :-)

Thanks
Kevin