Linux Virtualization - Jul 2019 - [PATCH v6 14/18] drm/virtio: rework virtio_gpu_transfer_from_host_ioctl fencing

On Wed, Jul 03, 2019 at 01:05:12PM -0700, Chia-I Wu
wrote:
> On Tue, Jul 2, 2019 at 7:19 AM Gerd Hoffmann <kraxel at redhat.com>
wrote:
> >
> > Switch to the virtio_gpu_array_* helper workflow.
> (just repeating my question on patch 6)
> 
> Does this fix the obj refcount issue?  When was the issue introduced?
obj refcount should be fine in both old and new code.

old code:
  drm_gem_object_lookup
  drm_gem_object_put_unlocked

new code:
  virtio_gpu_array_from_handles
  virtio_gpu_array_put_free (in virtio_gpu_dequeue_ctrl_func).

Or did I miss something?

cheers,
  Gerd

On Thu, Jul 4, 2019 at 4:48 AM Gerd Hoffmann <kraxel at redhat.com>
wrote:
>
> On Wed, Jul 03, 2019 at 01:05:12PM -0700, Chia-I Wu wrote:
> > On Tue, Jul 2, 2019 at 7:19 AM Gerd Hoffmann <kraxel at
redhat.com> wrote:
> > >
> > > Switch to the virtio_gpu_array_* helper workflow.
> > (just repeating my question on patch 6)
> >
> > Does this fix the obj refcount issue?  When was the issue introduced?
>
> obj refcount should be fine in both old and new code.
>
> old code:
>   drm_gem_object_lookup
>   drm_gem_object_put_unlocked
>
> new code:
>   virtio_gpu_array_from_handles
>   virtio_gpu_array_put_free (in virtio_gpu_dequeue_ctrl_func).
>
> Or did I miss something?
In the old code, drm_gem_object_put_unlocked is called before the vbuf
using the object is retired.  Isn't that what object array wants to
fix?

We get away with that because the host only sees  hw_res_handles, and
executes the commands in order.

Maybe it was me who missed something..?
>
> cheers,
>   Gerd
>

On Thu, Jul 04, 2019 at 11:55:59AM -0700, Chia-I Wu
wrote:
> On Thu, Jul 4, 2019 at 4:48 AM Gerd Hoffmann <kraxel at redhat.com>
wrote:
> >
> > On Wed, Jul 03, 2019 at 01:05:12PM -0700, Chia-I Wu wrote:
> > > On Tue, Jul 2, 2019 at 7:19 AM Gerd Hoffmann <kraxel at
redhat.com> wrote:
> > > >
> > > > Switch to the virtio_gpu_array_* helper workflow.
> > > (just repeating my question on patch 6)
> > >
> > > Does this fix the obj refcount issue?  When was the issue
introduced?
> >
> > obj refcount should be fine in both old and new code.
> >
> > old code:
> >   drm_gem_object_lookup
> >   drm_gem_object_put_unlocked
> >
> > new code:
> >   virtio_gpu_array_from_handles
> >   virtio_gpu_array_put_free (in virtio_gpu_dequeue_ctrl_func).
> >
> > Or did I miss something?
> In the old code, drm_gem_object_put_unlocked is called before the vbuf
> using the object is retired.  Isn't that what object array wants to
> fix?
I think the fence keeps the bo alive.

cheers,
  Gerd