kms_cursor_crc: Force the test to run in full RGB range
PATCH 1/2 kms_cursor_crc: Force the test to run in full RGB range
In at least SKL and GLK (possibly other devices too), using a cursor plane to scan out an fb might result in a different pipe crc than when using a regular plane at the same position with the same fb. The differences could be caused by the cursor plane being limited to 8 bpc while the regular planes support higher bit depths, since the failures happens with specific color values, but that's speculation.
--- tests/kms_cursor_crc.c | 8 ++++++++ 1 file changed, 8 insertions(+)
tests/kms_cursor_crc.c +8 -0
@@ -372,6 +372,14 @@ static void run_test(data_t *data, void (*testfunc)(data_t *), int cursor_w, intkmstest_pipe_name (data->pipe), igt_output_name(output)); +/* +* Force test to use full range RGB. Limited range causes CRC +* mismatches in SKL and GLK. +*/ +kmstest_set_connector_broadcast_rgb (data->drm_fd, + data->output->config.connector, +BROADCAST_RGB_FULL ); +testfunc (data); igt_info("\n%s on pipe %s, connector %s: PASSED\n\n",
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/2 kms_cursor_limited: Test crc matches between cursor and overlay planes
This test actually *fails* in GLK and possibly SKL and others too. When using limited range, the pipe crc when using the cursor plane is different than the one when using an overlay plane, although the fb and plane position is the same.
Since this test fails it should probably not be merged, but it is here to prove the point of the previous patch.
--- tests/Makefile.sources | 1 + tests/kms_cursor_limited.c | 175 +++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 176 insertions(+) create mode 100644 tests/kms_cursor_limited.c
tests/Makefile.sources +1 -0
@@ -99,6 +99,7 @@ TESTS_progs_M = \ kms_chv_cursor_fail \ kms_cursor_crc \ kms_cursor_legacy \ + kms_cursor_limited \ kms_draw_crc \ kms_fbc_crc \ kms_fbcon_fbt \
tests/kms_cursor_limited.c new file +175 -0
175 lines not on this page · raw message
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 1/1 kms_cursor_crc: Force the test to run in full RGB range
In reply to Ander Conselvan de OliveiraIn at least SKL and GLK (possibly other devices too), using a cursor plane to scan out an fb might result in a different pipe crc than when using a regular plane at the same position with the same fb while using the CSC logic to limit the color range. The differences could be caused by the cursor plane being limited to 8 bpc while the regular planes support higher bit depths, leading to slightly different values to be used internally. This is evidenced by the failures happening with specific color values, 0.5 for example, but that's mostly speculation.
v2: Add more details to the commit message.
--- tests/kms_cursor_crc.c | 8 ++++++++ 1 file changed, 8 insertions(+)
tests/kms_cursor_crc.c +8 -0
@@ -372,6 +372,14 @@ static void run_test(data_t *data, void (*testfunc)(data_t *), int cursor_w, intkmstest_pipe_name (data->pipe), igt_output_name(output)); +/* +* Force test to use full range RGB. Limited range causes CRC +* mismatches in SKL and GLK. +*/ +kmstest_set_connector_broadcast_rgb (data->drm_fd, + data->output->config.connector, +BROADCAST_RGB_FULL ); +testfunc (data); igt_info("\n%s on pipe %s, connector %s: PASSED\n\n",
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 1/1 lib/igt_kms: Force outputs to use full range RGB
In at least SKL and GLK (possibly other devices too), using a cursor plane to scan out an fb might result in a different pipe crc than when using a regular plane at the same position with the same fb while using the CSC logic to limit the color range. The differences could be caused by the cursor plane being limited to 8 bpc while the regular planes support higher bit depths, leading to slightly different values to be used internally. This is evidenced by the failures happening with specific color values, 0.5 for example, but that's mostly speculation.
To avoid misterious failures caused by limited range rgb, force all tests to use full range. It is still possible for tests to override this if necessary.
v2: Add more details to the commit message.
v3: Force all tests to use full range.
--- lib/igt_kms.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)
lib/igt_kms.c +6 -1
@@ -1455,10 +1455,15 @@ static void igt_output_refresh(igt_output_t *output) -1); } -if (output->config.connector) +if (output->config.connector) {igt_atomic_fill_connector_props (display, output,IGT_NUM_CONNECTOR_PROPS , igt_connector_prop_names); +kmstest_set_connector_broadcast_rgb (display->drm_fd, + output->config.connector, +BROADCAST_RGB_FULL ); + } +if (output->use_override_mode) output->config.default_mode = output->override_mode;
Signature · 6 lines
2.9.3 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH i-g-t v3] lib/igt_kms: Force outputs to use full range RGB
Quoted 13 lines · Ander Conselvan de Oliveira
In at least SKL and GLK (possibly other devices too), using a cursor plane to scan out an fb might result in a different pipe crc than when
using a regular plane at the same position with the same fb while using
the CSC logic to limit the color range. The differences could be caused
by the cursor plane being limited to 8 bpc while the regular planes support higher bit depths, leading to slightly different values to be used internally. This is evidenced by the failures happening with specific color values, 0.5 for example, but that's mostly speculation.
Quoted 4 lines
To avoid misterious failures caused by limited range rgb, force all tests to use full range. It is still possible for tests to override this
if necessary.
By this way, we know for sure what is the color range in use.
Quoted 32 lines
v2: Add more details to the commit message.
v3: Force all tests to use full range.
Cc: Ville Syrjälä <[email protected]>
Signed-off-by: Ander Conselvan de Oliveira
<[email protected]>
---
lib/igt_kms.c | 7 ++++++-1 file changed, 6 insertions(+), 1 deletion(-)lib/igt_kms.c +0
-0@@ -1455,10 +1455,15 @@ static void igt_output_refresh(igt_output_t *output) -1); }lib/igt_kms.c +1
-1-if (output->config.connector) +if (output->config.connector) {
igt_atomic_fill_connector_props(display, output, IGT_NUM_CONNECTOR_PROPS, igt_connector_prop_names);lib/igt_kms.c +2
-0+ kmstest_set_connector_broadcast_rgb(display->drm_fd, + output-Quoted 1 line
config.connector,
lib/igt_kms.c +1
-0+ BROADCAST_RGB_FULL);
lib/igt_kms.c +2
-0+ } +
Quoted 3 lines
if (
output->use_override_mode)output->config.default_mode=output->override_mode;
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH i-g-t v3] lib/igt_kms: Force outputs to use full range RGB
Quoted 14 lines · Mika Kahola
Quoted 13 lines · Ander Conselvan de Oliveira
In at least SKL and GLK (possibly other devices too), using a cursor plane to scan out an fb might result in a different pipe crc than when
using a regular plane at the same position with the same fb while using
the CSC logic to limit the color range. The differences could be caused
by the cursor plane being limited to 8 bpc while the regular planes support higher bit depths, leading to slightly different values to be used internally. This is evidenced by the failures happening with specific color values, 0.5 for example, but that's mostly speculation.
Quoted 8 lines
Quoted 4 lines
To avoid misterious failures caused by limited range rgb, force all tests to use full range. It is still possible for tests to override this
if necessary.By this way, we know for sure what is the color range in use.
Reviewed-by Mika Kahola <[email protected]>
Thanks, patch pushed.
Ander
Quoted 30 lines
Quoted 30 lines
v2: Add more details to the commit message.
v3: Force all tests to use full range.
Cc: Ville Syrjälä <[email protected]>
Signed-off-by: Ander Conselvan de Oliveira
<[email protected]>
---
lib/igt_kms.c | 7 ++++++-1 file changed, 6 insertions(+), 1 deletion(-)lib/igt_kms.c +0
-0@@ -1455,10 +1455,15 @@ static void igt_output_refresh(igt_output_t *output) -1); }lib/igt_kms.c +1
-1-if (output->config.connector) +if (output->config.connector) {
igt_atomic_fill_connector_props(display, output, IGT_NUM_CONNECTOR_PROPS, igt_connector_prop_names);lib/igt_kms.c +2
-0+ kmstest_set_connector_broadcast_rgb(display->drm_fd, + output-Quoted 1 line
config.connector,
Quoted 12 lines
Quoted 7 lines
lib/igt_kms.c +1
-0+ BROADCAST_RGB_FULL);
lib/igt_kms.c +2
-0+ } +if (
output->use_override_mode)output->config.default_mode=output->override_mode;_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
---------------------------------------------------------------------
Intel Finland Oy
Registered Address: PL 281, 00181 Helsinki
Business Identity Code: 0357606 - 4
Domiciled in Helsinki
This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx