Hi, the DCMIPP Camera interface (platform/st/stm32/stm32-dcmipp) is able to adjust the capture rate. As an example, even if a sensor is streaming at 30fps, it is possible to either capture: - all frames (aka 30fps) - 1 frame out of 2 (aka 15 fps) - 1 frame out of 4 ... - 1 frame out of 8 ... In such case the dropped frames won't go to the memory and will be dropped internally by the HW. This allows for reduction of the amount of captured data, reduce the bus usage, and can also be useful when the same sensor data must be captured in parallel by several pipelines outputting in different formats/framerates. (the current stm32mp13 has a single pipeline but the stm32mp25 has several). Having a look at existing subdevs within the media tree, I couldn't find similar subdev having an impact on the framerate (apart from the sensors). I first implemented this behavior using frame_interval control on both SINK and SOURCE pads of the subdev however linux-media doc clearly mentions that frame_interval should only be done on one pad. Original idea was to compute the difference of frame_interval between the SOURCE and SINK pad and thus apply the settings. As an example, having a SINK pad frame_interval corresponding to 30fps (coming from the sensor), and setting a frame_interval of 15fps on the SOURCE pad would lead to configuring the DCMIPP to perform capture of 1 frame out of 2. Another approach I could think of would be to have a dedicated ctrl on the subdev to control the frame rate adjustment factor. Do you have guidelines regarding how such functionality should be exposed ? Regards, Alain
