So, there is nothing more we can do here I think. Wait for rockchip to correct upstream.

maybe this bug at Rockchip/mpp is relative : https://github.com/rockchip-linux/mpp/issues/855

Hi The rights seems to be ok : gart@orangepi5:~$ ls -la /dev/dma_heap total 0 drwxr-xr-x 2 root root 100 Jan 1 1970 . drwxr-xr-x 20 root root 3940 Jul 4 19:49 .. crw-rw---- 1 root video 251, 2 Jul 4 19:48 cma crw-rw---- 1 root video 251, 0 Jul 4 19:48 system crw-rw-rw- 1 root video 251, 1 Jul 4 19:48 system-uncached my user is in the video group gart@orangepi5:~$ sudo cat /etc/group | grep video video:x:44:gart There should not be any issue If I launch ffmpeg encoding with sudo, it does'nt work neither I still have this kind of messages in syslog : 2025-07-06T15:32:19.769702+02:00 orangepi5 kernel: rk_vcodec: mpp_collect_msgs:1622: session 1 process cmd 401 ret -22 2025-07-06T15:32:19.769706+02:00 orangepi5 kernel: rk_vcodec: mpp_dev_ioctl:1732: collect msgs failed -22 2025-07-06T15:32:19.769708+02:00 orangepi5 kernel: rk_vcodec: mpp_task_attach_fd:1841: can't import dma-buf 275 2025-07-06T15:32:19.769711+02:00 orangepi5 kernel: rk_vcodec: mpp_translate_reg_address:1897: reg[ 0]: 0x00000113 fd 275 failed 2025-07-06T15:32:19.769713+02:00 orangepi5 kernel: rk_vcodec: mpp_task_dump_mem_region:2084: --- dump task 0 mem region --- 2025-07-06T15:32:19.769716+02:00 orangepi5 kernel: mpp_rkvenc2 fdbd0000.rkvenc-core: no memory region mapped 2025-07-06T15:32:19.769718+02:00 orangepi5 kernel: rk_vcodec: mpp_process_task_default:589: alloc_task failed. 2025-07-06T15:32:19.769721+02:00 orangepi5 kernel: rkvenc2_wait_result:2340: session 00000000b0422e29 pending list is empty! 2025-07-06T15:32:19.769724+02:00 orangepi5 kernel: rk_vcodec: mpp_msgs_wait:1693: session 1 wait result ret -5 2025-07-06T15:32:19.768265+02:00 orangepi5 mpp[3723]: message repeated 2 times: [ mpp_serivce: mpp_service_ioc_attach_fd failed ret -1 errno 22 Invalid argument] 2025-07-06T15:32:19.785062+02:00 orangepi5 mpp[3723]: mpp_serivce: mpp_service_ioc_attach_fd failed ret -1 errno 22 Invalid argument 2025-07-06T15:32:19.786302+02:00 orangepi5 kernel: rk_vcodec: mpp_process_request:1434: can not import fd 143 2025-07-06T15:32:19.786319+02:00 orangepi5 kernel: rk_vcodec: mpp_collect_msgs:1622: session 1 process cmd 401 ret -22 2025-07-06T15:32:19.786321+02:00 orangepi5 kernel: rk_vcodec: mpp_dev_ioctl:1732: collect msgs failed -22 2025-07-06T15:32:19.786322+02:00 orangepi5 kernel: rk_vcodec: mpp_task_attach_fd:1841: can't import dma-buf 274 2025-07-06T15:32:19.786323+02:00 orangepi5 kernel: rk_vcodec: mpp_translate_reg_address:1897: reg[ 0]: 0x00000112 fd 274 failed 2025-07-06T15:32:19.786324+02:00 orangepi5 kernel: rk_vcodec: mpp_task_dump_mem_region:2084: --- dump task 0 mem region --- 2025-07-06T15:32:19.786324+02:00 orangepi5 kernel: mpp_rkvenc2 fdbd0000.rkvenc-core: no memory region mapped 2025-07-06T15:32:19.786325+02:00 orangepi5 kernel: rk_vcodec: mpp_process_task_default:589: alloc_task failed. 2025-07-06T15:32:19.786327+02:00 orangepi5 kernel: rkvenc2_wait_result:2340: session 00000000b0422e29 pending list is empty! I lauch a simple file encoding from an AV1 file to x265 : sudo ffmpeg -i file1.AV1.mkv -c:v hevc_rkmpp -c:a copy ~/Downloads/file1.x265.mkv The result is a file that has a correct size, but is unreadable, black screen.

Since U-Boot v2025.04 works very well on Orange Pi Zero v3 and has a shorter boot process time, if there was a possibility to switch to U-Boot v2504 for Orange Pi Zero V1, I would be grateful, as would probably other users of this SBC Regards

The issue appears to be two things. These two patches and the current defconfig(s) being used. drv-rtc-sun6i-support-RTCs-without-external-LOSCs.patch drv-rtc-sun6i-Add-Allwinner-H616-support.patch When I disable those patches and use my own defconfig everything works as it should. https://paste.armbian.com/tixixocubu The hard part here would be figuring out exactly what in the defconfig(s) are either breaking things or missing. Also if I disable those patches, what does it break on other units? I'm under the impression these patches were rejects from the original h616 bring up in mainline. But I could be wrong?