[PATCH i-g-t v5 4/4] tests/intel/kms_dp_linktrain_fallback: commit display reset before fallback

[Sowmiya S <sowmiya.s@intel.com>]

igt_display_reset() only clears IGT's in-memory state without a
commit, so the link from a previous run stays trained as-is. The
next setup_outputs() call re-applies the same mode on the same pipe,
which the driver satisfies with a fastset instead of a real modeset,
so the link never gets retrained from the sink's max. This makes
prev_link_rate/lane_count start from an already-reduced config and
fail with "Fallback unsuccessful".

Fix it by committing the reset, forcing the pipe off and the link
retrained from the max before each run.

Signed-off-by: Sowmiya S <sowmiya.s@intel.com>
---
 tests/intel/kms_dp_linktrain_fallback.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tests/intel/kms_dp_linktrain_fallback.c b/tests/intel/kms_dp_linktrain_fallback.c
index 2e72957e1..4ceaad8b5 100644
--- a/tests/intel/kms_dp_linktrain_fallback.c
+++ b/tests/intel/kms_dp_linktrain_fallback.c
@@ -383,6 +383,7 @@ static bool test_fallback(data_t *data, bool is_mst, bool force_uhbr)
 	retries = SPURIOUS_HPD_RETRY;
 
 	igt_display_reset(&data->display);
+	igt_display_commit2(&data->display, COMMIT_ATOMIC);
 
 	/*
 	 * For the UHBR-to-HBR fallback case, pin the link at the highest
-- 
2.43.0