[PATCH v3 0/3] FD close cleanup for XE

[Daniel Charles <daniel.charles@intel.com>]

This series adds the suggested logic to synchronously wait for the queue
kill to complete on FD close. This is needed to avoid a race condition
where the scheduler may still be processing messages for a queue that
has been closed, which can lead to errors.

v3:
    - Patch 3 only, fix GT reset seen in CI (xe_create@create-contexts).
      See the patch 3 changelog for details.

v2:
    - Bound the wait with a timeout, GT reset on timeout (Sashiko, Matt B)
    - Drop the flush_work() loop (Matt B)
    - Use xe_assert() for the killed-queue check (Matt B)
    - Merge the put loop into the kill_sync loop (Matt B)

Daniel Charles (1):
  drm/xe: Synchronously wait for queue kill on FD close

Matthew Brost (2):
  drm/xe: Read scheduler message opcode under lock
  drm/xe: Disable scheduling early on FD close to avoid CAT error
    cascade

 drivers/gpu/drm/xe/xe_device.c               |   5 +
 drivers/gpu/drm/xe/xe_exec_queue.c           |  13 +++
 drivers/gpu/drm/xe/xe_exec_queue.h           |   1 +
 drivers/gpu/drm/xe/xe_exec_queue_types.h     |   2 +
 drivers/gpu/drm/xe/xe_gpu_scheduler.c        |  13 ++-
 drivers/gpu/drm/xe/xe_gpu_scheduler_types.h  |   2 +-
 drivers/gpu/drm/xe/xe_gt_sriov_vf.c          |   1 +
 drivers/gpu/drm/xe/xe_guc_exec_queue_types.h |   2 +-
 drivers/gpu/drm/xe/xe_guc_submit.c           | 101 +++++++++++++++++--
 drivers/gpu/drm/xe/xe_guc_submit.h           |   1 +
 drivers/gpu/drm/xe/xe_guc_types.h            |   6 ++
 11 files changed, 134 insertions(+), 13 deletions(-)

-- 
2.43.0