Skip to content

RFC Async Flip in Atomic ioctl corrections

Arun R Murthy dri-devel intel-gfx intel-xe 3 versions – In review

Get series with b4
b4 am [email protected] change-id 20251010-async-feb09912440b

Versions newest first

  1. v3 RFC Latest0 of 7 patches 11 repliesIn review
  2. v2 RFC2 patches 2 repliesSuperseded inferred link
  3. v1 RFC1 patch 0 repliesSuperseded

Patches in v3 No applies recorded yet

None of the 7 patches the cover letter announces are in this archive.

Reviews trailers per patch

No Reviewed-by, Acked-by or Tested-by trailers for v3 yet.

Read the v3 thread

Cover letter [PATCH RFC v3 0/7] Async Flip in Atomic ioctl corrections

struct drm_crtc_state {
/**
* @async_flip:
*
* This is set when DRM_MODE_PAGE_FLIP_ASYNC is set in the legacy
* PAGE_FLIP IOCTL. It's not wired up for the atomic IOCTL
itself yet.

          */
         bool async_flip;

In the existing code the flag async_flip was intended for the legacy PAGE_FLIP IOCTL. But the same is being used for atomic IOCTL. As per the hardware feature is concerned, async flip is a plane feature and is to be treated per plane basis and not per pipe basis. For a given hardware pipe, among the multiple hardware planes, one can go with sync flip and other 2/3 can go with async flip. Tearing affect will be noticed with this and if any policy should be taken care by the user space. KMD not to include any policy as to allow async on only one plane for a given pipe as all policy done in user and KMD exposes what is supported by the hardware.

There would be a bunch of changes to correct this in the atomic path. Add a new async_flip plane property to allow user enable async flip on the required plane.
Any restriction checks for async_flip will be taken in atomic_check() and not in the atomic_ioctl().
Let the preset code reside as is even in the atomic patch until all the existing drivers and user space implementations move to plane property for async flips.
Changes include removal of the checks we have in atomic path so as to reject any changes(different plane, pipe, connector) along with async flip. This would be replaced with checks so as to reject any change in that particular plane where async is enabled(reject any change in pipe/connector as that would have impact on this plane)

With the above changes, the challenge that we have presently so as to enable async flip on overlays which is handled seperately with if condition in drm_atomic_uapi.c can be moved to driver specific atomic_check code.

This series depicts the changes in the drm core and upon getting feedback on this, driver specific changes for theis will be done in the next revision on the same series.

Please let us know your opinion on this.

Note: The series is partially tested with the IGT so as to ensure the existing implementation is not broken. Full feature testing is pending, just opening the series as RFC so as to get comments so that I can incorporate them.

Thanks and Regards,
Arun R Murthy
-------------

---
Arun R Murthy (7):
drm/atomic/plane: Add plane property for async flip drm/atomic/plane: create async flip property for plane drm/atomic: Re-route the async flip based on the flag drm/atomic: Move the plane property check for async flip to atomic_check drm/atomic: Allow planes with NULL fb along with async flip drm/atomic: flip_done signal for planes
drm/i915/irq: Enable flip_done for each plane on async flip

 drivers/gpu/drm/drm_atomic.c                       | 122 ++++++++++++++++++++-
 drivers/gpu/drm/drm_atomic_helper.c                |   2 +
 drivers/gpu/drm/drm_atomic_uapi.c                  |  74 +++----------
 drivers/gpu/drm/drm_plane.c                        |  33 ++++++
 drivers/gpu/drm/i915/display/intel_display_irq.c   |  98 ++++++++++++++---
 drivers/gpu/drm/i915/display/skl_universal_plane.c |   2 +-
 include/drm/drm_plane.h                            |  19 ++++
 7 files changed, 277 insertions(+), 73 deletions(-)

---
base-commit: b3845fe8af5ec5c1d4b26e40ac6b4c7370e5fa35
change-id: 20251010-async-feb09912440b

Best regards,

Signature · 1 line
Arun R Murthy <[email protected]>