Users reported oopses on list corruptions when using i915 perf with a
number of concurrently running graphics applications. That indicates we
are currently missing some important tests for such scenarios. Cover
that gap.
v3: locate the new subtest in a new tests, intedned for exercising
intentionally racy barrier tasks list processing,
- fixes and optimizations (see the patch changelog for details).
v2: drop open-race subtest for now, not capable of triggering the user
reported bug, but triggering other bugs which I can't see any fixes
for queued yet,
- move the other new subtest out of tests/i915/perf.c (Ashutosh).
Janusz Krzysztofik (1):
tests: Exercise remote request vs barrier handling race
tests/i915/gem_barrier_race.c | 159 ++++++++++++++++++++++++++++++++++
tests/meson.build | 8 ++
2 files changed, 167 insertions(+)
create mode 100644 tests/i915/gem_barrier_race.c
--
2.25.1
