[Intel-gfx] [PATCH i-g-t v2 0/7] Scheduler tests

Derek Morton derek.j.morton at intel.com
Wed Mar 2 18:10:38 UTC 2016


This patch set adds scheduler tests.
Patch 1 Makes gem_has_ring() non static as the test will need to call it
Patch 2 Separate ring BSD1 from BSD2 in gem_require_ring() to prevent the
BSD1 specific tests from skipping if there is no BSD2 ring. Based on a patch
originally submitted by Gabriel Feceoru.
Patch 3 adds library code used by the tests. There are other tests under
development which are planned to reuse some of these libraries.
Patch 4 adds some basic tests, read dependency tests and write dependency tests.
Patch 5 Is the patch previously submitted by John Harrison to update
gem_ctx_param_basic with ioctls to set context priorities. It is included as
part of this patch set as Patch 6 & 7 are dependant on it.
Patch 6 adds tests to check sheduler behaviour for batch buffers submitted at
differing priorities.
Patch 7 adds tests to check priority bumping behaviour.

v2: Updates for comments from Daniele Ceraolo Spurio and Daniel Vetter
    Added tests requested by Joonas Lahtinen

Derek Morton (6):
  ioctl_wrappers: make gem_has_ring non static
  lib/ioctl_wrappers: Separate ring BSD1 from BSD2 checks
  lib/intel_batchbuffer: Add functions to be used in the scheduler test
  tests/gem_scheduler: Add gem_scheduler test
  tests/gem_scheduler: Add subtests to test batch priority behaviour
  gem_scheduler: Added subtests to test priority bumping

John Harrison (1):
  igt/gem_ctx_param_basic: Updated to support scheduler priority
    interface

 lib/intel_batchbuffer.c     | 384 ++++++++++++++++++++++++++-
 lib/intel_batchbuffer.h     |  14 +
 lib/ioctl_wrappers.c        |   6 +-
 lib/ioctl_wrappers.h        |   2 +
 tests/Makefile.sources      |   1 +
 tests/gem_ctx_param_basic.c |  34 ++-
 tests/gem_scheduler.c       | 622 ++++++++++++++++++++++++++++++++++++++++++++
 7 files changed, 1055 insertions(+), 8 deletions(-)
 create mode 100644 tests/gem_scheduler.c

-- 
1.9.1



More information about the Intel-gfx mailing list