When using visl in automated tests, we need to have output frames that can be compared to reference frames or hash of those to validate that the whole pipeline is working properly. Make sure that a given input stream always outputs the same frames. This is done by removing TPG information like queues status and pointer values. Introduce the tpg_verbose parameter to allow adding the more verbose information that this patchset removes This also adds some stable variation in the frames so that different input give more different output. Changes since v4: - Remove unnecessary changes in visl_get_ref_frames - Fix missing space after && Changes since v3: - Fix some more permissions in vidtv - Rename stable_output to tpg_verbose - Rebase on media_stage/master to include AV1 support - Remove code churn in visl_get_ref_frames() - Remove code duplication in visl_tpg_fill_sequence() Changes since v2: - Set parameters permissions to 0444 in visl and vidtv - Make stable_output true by default - Keep showing frames timestamps and remove vb2 buffer indexes only in stable output mode - Make codec_variablility non optional (remove parameter, keep implementation) - Add details on used variability fields Changes since v1: - Fix typo in parameter documentation Detlev Casanova (4): media: visl,vidtv: Set parameters permissions to 0444 media: visl: Add a tpg_verbose parameter doc: visl: Document tpg_verbose parameter visl: Add codec specific variability on output frames Documentation/admin-guide/media/visl.rst | 12 +- .../media/test-drivers/vidtv/vidtv_bridge.c | 26 +- drivers/media/test-drivers/visl/visl-core.c | 15 +- drivers/media/test-drivers/visl/visl-dec.c | 301 +++++++++++------- drivers/media/test-drivers/visl/visl.h | 1 + 5 files changed, 217 insertions(+), 138 deletions(-) -- 2.43.0
