[PATCH i-g-t 0/3] test/kms_async_flip: fixes to get execution

[Melissa Wen]

Hi,

When I was reviewing "tests/kms_async_flip: Speed up the CRC test on
Intel hw" (https://patchwork.freedesktop.org/patch/572473/), I noticied
kms_async_flip aborts in the first test case with non-intel device since
commit 7fa7602ccf ("tests/kms_async_flips: Test all modifiers").
To get the execution back, the first patch changes the name of a
non-intel modifier from "?" to "unknown".

The same commit also claims to allow failure with some modifiers, but
the path for handling fb with modifiers != LINEAR and != intel doesn't
cover other vendors requirements and isn't consistent yet. Ofc,
increasing coverage with other vendor modifiers would be nice, but before
changing from run_test() to run_test_with_modifiers() the test case was
working as expected. Moreover, it seems we need a good rework of the
_with_modifiers path to make it stable again. This regression is the
reason I choose to bring back the original approach for non-intel
devices in the third path of this series. Follow-up patches can be sent
to increase test coverage with modifiers for non-intel device.

Let me know your thoughts.

Thanks,

Melissa

Melissa Wen (3):
  test/kms_async_flips: fix test abortion due to unreadable mod name
  lib/igt_fb: fix documentation of igt_create_color_fb return
  tests/kms_async_flips: test with modifiers only on intel devices

 lib/igt_fb.c            | 5 ++---
 tests/kms_async_flips.c | 5 ++++-
 2 files changed, 6 insertions(+), 4 deletions(-)

-- 
2.43.0