[PATCH 0/9] Clean up some CI failures for GuC submission

Matthew Brost matthew.brost at intel.com
Wed Aug 11 01:16:13 UTC 2021


Resets are notoriously hard to get fully working and notoriously racey,
especially with selftests / IGTs that do all sorts of wild things that
would be near impossible to hit during normal use cases. Even though
likely impossible to hit, anything selftests / IGTs uncover needs to be
fixed. This series addresses 7 such issues, adds a workaround to a
selftest, and add another selftest to prove scrubbing of G2H during a
reset works. 

v2:
 (Daniel Vetter)
  - Split fixes into individual patches
  - A kernel doc
  - A VLK ref for selftest workaround
 (Checkpatch)
  - Fix warnings

Signed-off-by: Matthew Brost <matthew.brost at intel.com>

Matthew Brost (9):
  drm/i915/guc: Fix blocked context accounting
  drm/i915/guc: outstanding G2H accounting
  drm/i915/guc: Unwind context requests in reverse order
  drm/i915/guc: Don't drop ce->guc_active.lock when unwinding context
  drm/i915/guc: Flush the work queue for GuC generated G2H
  drm/i915/guc: Do not clear enable during reset in an enable is
    inflight
  drm/i915/guc: Don't enable scheduling on a banned context
  drm/i915/selftests: Fix memory corruption in live_lrc_isolation
  drm/i915/selftests: Add initial GuC selftest for scrubbing lost G2H

 drivers/gpu/drm/i915/gt/intel_context_types.h |  18 +++
 drivers/gpu/drm/i915/gt/selftest_lrc.c        |  29 +++-
 .../gpu/drm/i915/gt/uc/intel_guc_submission.c |  69 +++++++---
 drivers/gpu/drm/i915/gt/uc/selftest_guc.c     | 126 ++++++++++++++++++
 .../drm/i915/selftests/i915_live_selftests.h  |   1 +
 .../i915/selftests/intel_scheduler_helpers.c  |  12 ++
 .../i915/selftests/intel_scheduler_helpers.h  |   2 +
 7 files changed, 239 insertions(+), 18 deletions(-)
 create mode 100644 drivers/gpu/drm/i915/gt/uc/selftest_guc.c

-- 
2.32.0



More information about the dri-devel mailing list