Conversation
…a 24" This reverts commit 7821b6a.
This reverts commit 67b85dd.
This reverts commit 18d2ec3.
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Upstream will take the multi-planar SAND format, but add back in the downstream single planar variant for backwards compatibility Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
To avoid user complaints that /dev/video0 isn't their USB webcam, add downstream patch that allows setting the preferred video device number. Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Decoder does not need OUTPUT or CAPTURE buffers mapped into memory. Set the queue dma_attrs to indicate this. Signed-off-by: John Cox <jc@kynesim.co.uk>
I tested a checkout that was just this one commit reverted. The result was an images that were solid blocks of green at the end of a gstreamer pipeline. And I tested this full pull request and at 1920/15 it pulled 175 frames then crashed. It was also able pull a couple of correct 4K frames. At one stage I saw an error about calling a certain function, with I think a null, was an indication of a possible bug. but it certainly translated several frames correctly through to the end of a gstreamer pipeline. So it might just be a some resources handling and edge cases. |
3 participants
This brings rpi-6.12.y into line with the next version of the patchset for upstream, and allows for testing the GStreamer MR at https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9247
We may be able to get away with just "Revert "media: hevc_dec: Drop the new image formats until we have Mes…" to get GStreamer working, and I might be tempted to do that rather than this slightly more invasive change. I'll talk to John next week.