| Age | Commit message (Collapse) | Author |
|---|
|
Use the same input and output formats to disable format conversion as
the purpose of the tests is to test flipping.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The rotation control can be set to a non-default value by a previously
run test, reset it.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Test updating the vertical and horizontal flip controls during video
streaming.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Test all combinations of horizontal flip, vertical flip and rotation.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Test CLU and LUT operation in both RGB and YUV modes with different look
up tables.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
This allows reusing variables internal to vsp-lib and getting rid of the
mdev and format arguments to the vsp-runner invocations.
As part of the transition, handle additional options with an explicit
list of supported options instead of passing them straight to yavta.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
This allows deciding whether to compare output frames using the exact or
fuzzy algorithm internally, removing the need for an explicit argument
to the compare_frames function.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The reference frame generator scaling algorithm doesn't match the VSP
scaler in a pixel-perfect fashion, use fuzzy comparision of the result.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Depending on the test this lowers the conversion errors or ensures a
pixel-perfect match between the reference frame generator and the
hardware. In the latter case we can now use the faster exact output
verification method.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The VSP instance that supports the largest number of RPF might not be
the same as the one that supports the largest number of WPF. To maximize
test coverage, split the test into RPF and WPF tests that can use
different VSP instances.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Instead of getting the number of BRU inputs from an explicit argument to
the compare_frames function, store it internally when configuring the
pipeline. This simplifies the test scripts by removing the reftype
argument to the compare_frames function.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
The information is already known to the pipeline configuration
functions, store it internally and reuse it instead of having to pass it
explicitly.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Remove the need to pipe individual commands to the logger from the test
scripts by moving all logging code internally to vsp-lib and vsp-runner.
As an interesting side effect the removal of shell pipes from the test
scripts avoids running functions in subshells. This makes it possible
for the pipeline configuration functions to set variables that can be
used later.
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
|
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
|
