Active threads

thanks @Arthur Gu, I will test and include into armbian

This is a work-in-progress resource for bare-metal bringup for Rockchip devices. Source code: https://github.com/petabyt/rk https://danielc.dev/rk/

The issue has been resolved. It turns out that the official release has already configured the necessary documents or files before distribution. When I used im‑config, it gave me no response at all. Following the XDG Autostart manual, I copied /usr/share/applications/org.fcitx.Fcitx5.desktop to ~/.config/autostart; after that, fcitx5 starts automatically and works normally when I reboot or log out and back in. Thanks

@sven-ola Sorry, this was indeed a mystery problem with my SSD. I replaced it, and now everything works like a charm (I just need the CLI server; Wi-Fi or GPU isn’t necessary). By the way, when is the stable release planned?

--- Hey everyone! Long-time lurker, first post here. I picked up one of those cheap RK3229 TV boxes and decided to see how far I could push it on a mainline kernel — no Android, no BSP, no blobs. What started as curiosity about whether rkvdec actually worked on kernel 6.6 turned into a full rabbit hole of GStreamer pipelines, PHY register dumps, and DRM memory bandwidth math. The whole debugging and documentation process was done collaboratively with Claude (Anthropic's AI assistant), which made it a lot easier to dig into kernel driver internals and iterate on the pipeline without spending weeks at it. All the results were validated on real hardware. Sharing here in case it saves someone else the same rabbit hole. --- [GUIDE] Hardware H.264 decode at 720p — mainline kernel 6.6, GStreamer, no blobs After spending some time on my RK3229 TV box running Armbian 24.2.5 Bookworm (kernel 6.6.22-current-rockchip), I got hardware H.264 decode working at 720p real-time using only mainline kernel drivers and open-source GStreamer plugins. Posting here in case it helps others. --- What works - H.264 720p@30fps fully hardware decoded via rkvdec (the mainline V4L2 stateless decoder) - Direct HDMI output via kmssink (DRM/KMS — no X11, no Wayland needed) - Audio output via ALSA (HDMI or analog) - YouTube streaming with audio+video in sync using a small proxy server What doesn't work and why - 1080p: The decoder itself handles it, but writing ~90 MB/s of NV12 frames to uncached DRM memory saturates the Cortex-A7 memory bus. Not a software problem — there's no fix without DMA-BUF zero-copy between rkvdec and the DRM subsystem. - YouTube in a browser: Browsers do their own software decode — no VA-API bridge exists for rkvdec on mainline. Even with a desktop environment installed, frame rate will be unusable. - HEVC / AV1: Not supported by rkvdec on RK3228/RK3229. --- The GStreamer pipeline # Local H.264 file — video only gst-launch-1.0 filesrc location=video.mp4 ! qtdemux ! h264parse \ ! v4l2slh264dec ! videoconvert ! kmssink driver-name=rockchip sync=true # YouTube streaming with audio (requires proxy — see repo) gst-launch-1.0 -e \ souphttpsrc location="http://PROXY_IP:8091/play?v=VIDEO_ID&q=720&fmt=ts" automatic-redirect=true ! \ tsdemux name=demux \ demux. ! queue ! h264parse ! v4l2slh264dec ! videoconvert ! kmssink driver-name=rockchip sync=true \ demux. ! queue ! aacparse ! avdec_aac ! audioconvert ! audioresample \ ! "audio/x-raw,rate=44100,channels=2" ! alsasink device=hw:2 Key points: - v4l2slh264dec is the stateless GStreamer element — do not use h264_v4l2m2m (that's for stateful decoders like RPi) - kmssink driver-name=rockchip uses /dev/dri/card0 (Rockchip DRM display), not the Lima GPU - For YouTube, MPEG-TS (fmt=ts) is required — fragmented MP4 with empty_moov breaks GStreamer cap negotiation --- Why this is different from Jock's media framework Jock's framework uses kernel 4.4 + RKMPP proprietary blobs. This uses kernel 6.6 mainline + rkvdec upstream driver + open-source GStreamer. No blobs, works with current Armbian, survives kernel upgrades. --- Requirements - Armbian 24.x with kernel 6.6-current-rockchip (kernel 5.15 / Bullseye not tested — V4L2 stateless API wasn't stable until 5.18) - GStreamer 1.22 from Debian Bookworm sudo apt install -y gstreamer1.0-tools gstreamer1.0-plugins-base \ gstreamer1.0-plugins-good gstreamer1.0-plugins-bad \ gstreamer1.0-libav gstreamer1.0-alsa --- Repo with scripts and full setup guide https://github.com/Reinoldo-Ozy/rk322x-mediaplayer Includes the yt-play playback script and the proxy server (yt_proxy.py) with systemd unit. The README covers the full setup, performance numbers, and a detailed limitations section. Tested on a generic MXQ Pro-style box with RK3229, 2 GB RAM, Armbian 24.2.5 Bookworm, kernel 6.6.22-current-rockchip, DTB rk322x-box.dtb.

I am running mine on Trixie, installed about a year ago. Then it worked ... Core is running inside Docker container so functionality should be O.K., but we might currently have some problem ... I will look into this when I am about to reinstall my setup, but I plan to add Docker install only, which should work well, but it has certain limitations.

@Sergioclr remove those two Failing patches you don’t need them anyways.

@SecLyzer U-Boot can boot from a FAT32 or Linux file system. You can change this in your board configuration file.

Due to the shortage of memory chips these days, Radxa is using different emmc models in their boards. For example, I recently ordered a big batch of Rock S0 boards, but they had to be delivered with Sandisk iNAND 32GB emmc drives instead of the typical 8GB option. The problem is that Armbian did not reliably boot on these new boards. It randomly gets stuck during initialization with the onboard LED endlessly blinking. Some boards works, some did not, some only works sometimes. Long story short, in case other people also have this problem: I found the solution. I believe it is caused by Armbian's Rock S0 device tree enabling the HS200 high speed mode for the emmc device. (source) I guess not all emmc chips reliably support this. To fix it, I had to remove the line in the device tree linked in the source above. Actually this should probably be done with a proper device tree modification, but in my case, just to test, I modified the boot scripts. In /boot/boot.cmd, add these lines near the bottom, but ABOVE the "booti ..." line: fdt rm /mmc@ff490000 mmc-hs200-1_8v fdt set /mmc@ff490000 max-frequency <0x02faf080> Then recompile in terminal with: mkimage -C none -A arm -T script -d /boot/boot.cmd /boot/boot.scr If you update the kernel/armbian distro these script will probably get overwritten, hence why it should preferably done with a custom device tree instead. But in my case I have frozen updates, and either way this will serve as a starting point for others having the same issue. Actually, I cannot see this troublesome patch on the 6.18 distro source, so maybe this is old news and no longer a problem on modern images anyway. But my image is still on 6.12, so I figured I'd share anyway.

@Sand_DeathDid you extract the image and use BalenaEtcher? It works fine on my Radxa Cubie A7A board.

@Taz Radxa's u-boot won’t load the kernel image. So I used vmlinux.

none which doesn't break other boards.

Armbian has limited resources. And given that there are hundreds of different boards out there, generally that means that new boards only get supported either because the company behind them is willing to contribute to Armbian (and Orange Pi has made if very clear that they don't do anything to support their boards software after the fact as it would reduce sales for new boards they are constantly introducing), or some volunteer from the community steps up to volunteer their time to add support. Most boards under Armbian are community supported. So it relies on volunteers like you to do the work.

Embedded Recipes returns to Nice for its second standalone edition, along with Linux Media Summit, PipeWire, libcamera, GStreamer Spring Hackfest, and new Display Next Hackfest & BlueZ F2F. Visit our table for Tyr, ML video analytics & Flipper One prototype! View the full article

Are you using the rolling release? Otherwise you should not use "full-upgrade" but rather just "upgrade". Not saying that is the reason for missing deps, but it definitely could be.

@Giunti & @samlevy0515 It would be helpful to the community if you responded to the developer trying to fix this issue rather than state a workaround. Thank you.

Did you try Yamagi Quake 2 with OpenGL ES1? A developer optimized it, and I assume it will also be a lot faster on a PBP, but it might be that certain mods don't work with GLES1. You have to build it like this: make -j4 WITH_SDL3=no with_gles1 Select OpenGL ES1 in the Video Settings. If you don't see blood particles (or they are squares), go to the game console and disable a certain module: gl1_pointparameters 0 Restart the game and have fun!

R76S migrated to use only uboot mainline, tested on real hardware, wait the next releases that's at the end of this month.

Collabora is proud to share that we've partnered with Flipper Devices to work together on building an open Linux platform for hardware hackers. The long-awaited Flipper One will be built on the Rockchip RK3576! View the full article

Got the same board running Armbian with ChatGPT. Because we performed a lot of experiments, here are summary of how we actually got it running tl.tr: upgrade_tool is wonderful, but running it on my board destroyed everything and I had to recover from backups. We backuped board completely, then intentionally corrupted android bootloader from adb and managed to get U-Boot running via modification of boot.scr Here is a chat-gpt generated summary:

To help others, here is the solution to my problem: I downloaded some firmware versions available here (https://xdaforums.com/t/help-with-a-s805x-box-official-firmware-update-borked-it.4646963/post-89791881) and used Android's Recovery mode to return the version. I clicked on "Apply update from" and used the .zip file for this. With this, the restart loop problems were resolved and now I can run Armbian directly from the pendrive. I'm trying to install it directly on the eMMC and I hope it works, I'll come back here with news.

Hello everyone , i tried flashing the slimboxtv using phoenix app , but failed in the middle of it and the device wont boot and wont go into bootloader mode anymore . Driver was installed successfully , but what i did that was not mentioned on the tutorial was plugging the usb cable after the power cable whipe holding the reset button as it didnt detect it via usb cable alone . Any way to unbrick ? Anyhelp would be appreciated

Orangepi4pro is pure shit. A lot of marketing and no support at all. No driver to install nvme, I ruined the SPI trying to install Assist, and now I have a unuseful expensive brick.

Yes. I just came to armbian-config and there I select en_US as default, but it does not change Mate to en_US. It change are only on cli.

@iav Thank you for the detailed reponse. I was not blaming the overlay or your build; I know the cpu voltages don't impact on this seperate issue. I have seen this before with the 6.18 kernel on the Helios64 which is why for the moment I am sticking with the 6.12 kernel with the opp microvolt patch which seems to be stable for me at least. I assume something has changed in the underlying PCIe architecture or drivers between 25.11.2 with 6.12.58 and 26.5.0 with 6.18.30 which results in this issue on my hardware. It is interesting that the build works on your hardware though. I will investigate the PCIe issue further when time allows and raise a issue.