[Intel-gfx] Vblank timestamping improvements/fixes for Linux drm.

Alex Deucher alexdeucher at gmail.com
Mon Oct 28 16:54:09 CET 2013

On Sat, Oct 26, 2013 at 4:27 AM, Mario Kleiner
<mario.kleiner.de at gmail.com> wrote:
> Hi all,
> this patch set for the kernel pushes the latency sensitive bits of
> vblank scanoutpos timestamping from the drm core into the kms drivers.
> A change in the locking of the intel-kms driver for Linux 3.11 made
> the old approach too inaccurate and also incompatible with the
> PREEMPT_RT realtime kernel patch set. These patches fix that problem
> and restore the old level of precision and reliability.
> The patch set changes the prototype of driver->get_scanout_position()
> to require/allow kms drivers to perform the ktime_get() system time
> queries which go along with actual scanout position readout in a way
> that provides maximum precision and to return those timestamps to
> the drm. It also converts the only two kms drivers which use this api
> so far (intel-kms and radeon-kms) to the new api and improves precision
> and reliability of the intel-kms a lot.
> Patches have been tested on Intel and AMD Radeon hardware and the Intel
> bits have received some review and feedback by Ville Syrjälä.
> Please review and apply if possible.

Looks good to me.  For the series:

Reviewed-by: Alex Deucher <alexander.deucher at amd.com>

> Thanks,
> -mario

More information about the Intel-gfx mailing list