[igt-dev] [PATCH i-g-t v10 0/4] Add api_intel_bb test

[Zbigniew Kempczyński]

Slowly as London bus move toward say goodbye to libdrm in the
rendercopy tests.

Seriously - add some new functions in intel_buf/intel_bb and
verify most of them work properly and doesn't affect gem_gpgpu_fill,
gem_media_fill and i915_pm_sseu which were previously migrated.

v2: fix mapping
v3: skip test where no context are available
v4: cover gens <=4, 5, and 6+ separately
v5: differentiate flush (MI_FLUSH <= gen4 and MI_FLUSH_DW >= gen5)
    add intentionally failing test to skip full run
v6: add igt_info to print out ring, remove failing test
v7: fix buffer leak, reset relocation count
v8: ensure bb alignment is modulo 8, fix offsets validation 
    in api_intel_bb test
v9: remove flush code, change bb constraints to fill fully bb,
    adding reference counting (for future tests updates),
    adding full-batch test which verifies filling fully bb
v10: synchronizing on bb (executed as async)

Zbigniew Kempczyński (4):
  lib/intel_batchbuffer: Extend intel_bb
  lib/intel_bufops: Add new functions and intel_buf fields
  tests/api_intel_bb: Add intel_bb API test
  HAX: run api_intel_bb tests and check (gpgpu|media)_fill are ok

 lib/intel_batchbuffer.c               | 370 +++++++++++++++-
 lib/intel_batchbuffer.h               |  79 +++-
 lib/intel_bufops.c                    | 135 +++++-
 lib/intel_bufops.h                    |  24 +-
 tests/i915/api_intel_bb.c             | 610 ++++++++++++++++++++++++++
 tests/intel-ci/fast-feedback.testlist | 164 +------
 tests/meson.build                     |   1 +
 7 files changed, 1188 insertions(+), 195 deletions(-)
 create mode 100644 tests/i915/api_intel_bb.c

-- 
2.26.0