[00/11] drm/virtio: fixes and cleanups for vbuf queuing

Message

Chia-I Wu Feb. 5, 2020, 6:19 p.m. UTC

This series consists of fixes and cleanups for
virtio_gpu_queue_fenced_ctrl_buffer, except for the last patch.  The fixes are
for corner cases that were overlooked.  The cleanups make the last patch
easier, but they should be good in themselves as well.

The last patch changes the disable_notify mechanism to call
virtqueue_kick_prepare only once in virtio_gpu_enable_notify.  It should be
more efficient than doing that after each command is queued.

There is a follow-up patch that replaces the global disable_notify state by
command-level bools to disable notify for individual commands, so that one
process cannot affect another process.  I can include it in v2 if you want to
review it together with this series.

Comments

Gerd Hoffmann Feb. 6, 2020, 11:17 a.m. UTC | #1

On Wed, Feb 05, 2020 at 10:19:44AM -0800, Chia-I Wu wrote:
> This series consists of fixes and cleanups for
> virtio_gpu_queue_fenced_ctrl_buffer, except for the last patch.  The fixes are
> for corner cases that were overlooked.  The cleanups make the last patch
> easier, but they should be good in themselves as well.

Pushed most patches.

> The last patch changes the disable_notify mechanism to call
> virtqueue_kick_prepare only once in virtio_gpu_enable_notify.  It should be
> more efficient than doing that after each command is queued.
> 
> There is a follow-up patch that replaces the global disable_notify state by
> command-level bools to disable notify for individual commands, so that one
> process cannot affect another process.  I can include it in v2 if you want to
> review it together with this series.

Can you rebase & resend so I can see all notify patches together?

thanks,
  Gerd