[PATCH i-g-t v3 00/10] tests/intel/xe_drm_fdinfo: Fix noise and improve

[Lucas De Marchi <lucas.demarchi@intel.com>]

2 patch series together:
1) https://patchwork.freedesktop.org/series/137750/
2) https://patchwork.freedesktop.org/series/137742/

The first one was a single patch fixing a bug. It's the same fix here,
but now with a proper justification in the commit message.  I found a
(LNL) machine reproducing the failures every 2 or 3 runs and executed 50
times successfully with patch 1.

Patch 3 was a suggestion from Matt Roper, but for that to work we need
patch 2, as we don't have the relevant assert in igt. Patch 4 comes
as a bonus.

Other patches are the same as the second patch series.

With all patches applied I executed 100 times in that machine and
didn't have any failure.

v2: Remove bogus patch
v3: Fix the stricter == 0 check

Lucas De Marchi (10):
  tests/intel/xe_drm_fdinfo: Extend mercy to the upper end
  tests/intel/xe_drm_fdinfo: Print timestamp for debug
  lib/igt_core: Add igt_assert_lt_double()
  tests/intel/xe_drm_fdinfo: Use igt_assert_lt_double()
  tests/intel/xe_drm_fdinfo: Be strict on == 0 comparison
  tests/intel/xe_drm_fdinfo: Use usec for batch duration
  tests/intel/xe_drm_fdinfo: Half the execution time
  tests/intel/xe_drm_fdinfo: Remove unused flags
  tests/intel/xe_drm_fdinfo: Use enum with expected load
  tests/intel/xe_drm_fdinfo: Rename and reorder tests

 lib/igt_core.h              |  12 ++
 tests/intel/xe_drm_fdinfo.c | 260 +++++++++++++++++++-----------------
 2 files changed, 153 insertions(+), 119 deletions(-)

-- 
2.43.0