[Intel-gfx] [PATCH 3/3] drm/i915: Add soft-pinning API for execbuffer

[Yang, Rong R]

> -----Original Message-----
> From: Daniel, Thomas
> Sent: Wednesday, October 21, 2015 23:11
> To: Daniel Vetter
> Cc: Chris Wilson; intel-gfx at lists.freedesktop.org; Belgaumkar, Vinay; Yang,
> Rong R
> Subject: RE: [Intel-gfx] [PATCH 3/3] drm/i915: Add soft-pinning API for
> execbuffer
> 
> > -----Original Message-----
> > From: Daniel Vetter [mailto:daniel.vetter at ffwll.ch] On Behalf Of
> > Daniel Vetter
> > Sent: Wednesday, October 21, 2015 4:08 PM
> > To: Daniel, Thomas
> > Cc: Chris Wilson; intel-gfx at lists.freedesktop.org
> > Subject: Re: [Intel-gfx] [PATCH 3/3] drm/i915: Add soft-pinning API
> > for execbuffer
> >
> > On Tue, Oct 06, 2015 at 01:59:12PM +0000, Daniel, Thomas wrote:
> > > > -----Original Message-----
> > > > From: Chris Wilson [mailto:chris at chris-wilson.co.uk]
> > > > Sent: Tuesday, October 6, 2015 11:53 AM
> > > > To: intel-gfx at lists.freedesktop.org
> > > > Cc: Chris Wilson; Daniel, Thomas
> > > > Subject: [PATCH 3/3] drm/i915: Add soft-pinning API for execbuffer
> > > >
> > > > Userspace can pass in an offset that it presumes the object is
> > > > located at. The kernel will then do its utmost to fit the object
> > > > into that location. The assumption is that userspace is handling
> > > > its own object locations (for example along with full-ppgtt) and
> > > > that the kernel will rarely have to make space for the user's requests.
> > > >
> > > > v2: Fix i915_gem_evict_range() (now evict_for_vma) to handle
> > > > ordinary and fixed objects within the same batch
> > > >
> > > > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > > > Cc: "Daniel, Thomas" <thomas.daniel at intel.com>
> > >
> > > This didn't apply cleanly to my tree pulled today (after patches 1
> > > and 2 of this
> > series).
> > > Are you going to post a rebase?
> >
> > It's a really trivial conflict in the uapi flag allocation. Happens
> > all the time with interface extensions.
> >
> > What I'm looking for here is the userspace for this new interface. And
> > the testcases.
> Hm I thought the beignet guys had already posted.
> Vinay has written i-g-t for this

Beignet svm patch haven't post, because the beignet's svm patch only work on i386 linux now, the x86_64 svm depends on 48bits pointer support in Beignet compiler's backend.
If the i386 svm patch is worthy for this patch, I will send it out.
 
> 
> > -Daniel
> > --
> > Daniel Vetter
> > Software Engineer, Intel Corporation
> > http://blog.ffwll.ch