Jump to content

CSC Armbian for RK322x TV box boards

jock

By jock
in Rockchip CPU Boxes

 Share

Recommended Posts

nafanovich

nafanovich

Posted
Posted

hi, i have several boxes that on 5 mainline kernel reboots (on legacy works good)

  Reveal hidden contents

 

and after "finished load/save random seed" box rebooting

 

any ideas?

  Reveal hidden contents
0
paradigman

paradigman

Posted
Posted

I have a box like this and I got stuck with it, please help! For CPU RK3228A, I attached images and UART log. I try multitool, I see the best result with 'rk3228b-box-mkv809vii.dtb', only with this will the LED light turn blue in the end. There is no picture on HDMI and I don't get to the point in the UART log.

 

The outside of the box
 

  Reveal hidden contents

 

The board:
 

  Reveal hidden contents

 

The UART log:
 

  Reveal hidden contents

 

Thanks

0
nafanovich

nafanovich

Posted
Posted
  On 10/13/2021 at 8:11 PM, jock said:

@nafanovich Hello!

You may try to add cpu-stability overlay in /boot/armbianEnv.txt. Just edit the file and append it to other overlays in overlays= line

Expand  

box cycle reboots right after install image from sd. i need to repack image to add this? or exist other way?

0
nafanovich

nafanovich

Posted
Posted
  On 1/14/2021 at 11:00 PM, jock said:

you should be able to burn the armbian image on an USB stick...

Once done, put the USB stick in the OTG port of the board and power on. U-boot should then boot from USB instead of eMMC...

Expand  

by this way (from usb-running image) i can access to emmc?

0
jock

jock

Posted
  • Author
Posted

@nafanovich Well it's strange your image works on the sdcard and does not work when you install it on eMMC.

You can use armbian  (or the multitool) on SD/USB to mount the internal eMMC and modify armbianEnv.txt from there.

Change verbosity=1 to verbosity=7 too, so you see the kernel log on the serial which is way important for your situation.

 

0
jock

jock

Posted
  • Author
Posted
  On 10/13/2021 at 9:05 PM, paradigman said:

I have a box like this and I got stuck with it, please help! For CPU RK3228A, I attached images and UART log. I try multitool, I see the best result with 'rk3228b-box-mkv809vii.dtb', only with this will the LED light turn blue in the end. There is no picture on HDMI and I don't get to the point in the UART log.

Expand  

Good you posted logs, bad you changed the dtb.

From the log I see, your sdcard/emmc is mounted read-only by kernel. Maybe it is faulty? Can't say for sure, since you changed the dtb and who knows what's happening...

 

Follow the instructions, use another sdcard if you can, then came back with fresh logs if you still have problems.

0
paradigman

paradigman

Posted
Posted (edited)
  On 10/14/2021 at 10:46 AM, jock said:

Good you posted logs, bad you changed the dtb.

From the log I see, your sdcard/emmc is mounted read-only by kernel. Maybe it is faulty? Can't say for sure, since you changed the dtb and who knows what's happening...

 

Follow the instructions, use another sdcard if you can, then came back with fresh logs if you still have problems.

Expand  

 

Thank you for your reply.
I don’t think the problem is unique because I have multiple boxes and each one behaves the same if I want to boot with multitool from an SD card.

I made a fresh log with the original dtb that comes with multitool.

The difference compared to the other dtb is that instead of the blue led, the red one lights up and the yellow one flashes at the end.

 

The fresh uart log:

 

  Reveal hidden contents

 

Thanks

Edited by paradigman
add plus information
0
jock

jock

Posted
  • Author
Posted

@paradigman ah ok, now I get it... you posted the multitool uart log.

Anyway, changing dtb is not useful. multitool dtb is made on purpose to have the main blinking led to show that the board and kernel are functioning.

Also multitool dtb is made to boot with very conservative options to increase compatibility and stability with boards, since its purpose is to do maintenance.

 

The problem that you don't see anything on HDMI may be due to multiple factors: the kernel, the HDMI cable and, most of all, the monitor.

If you can try another HDMI cable and monitor we can start to remove some variables from the equation.

0
fabiobassa

fabiobassa

Posted
Posted

@paradigman
hello , when you boot from multitool the rootfs is squash so it is read only.
If you don't have any hdmi it could be due to old/strange monitor resolution not handled properly from the board.
Example given , I had an old interlaced monitor and NOWAY to obtain hdmi output.

About the uart, once fatresize has occurred you should have a uart prompt. Since you were able to obtain tx from board ( rx to ttl converter..) ..... are you sure that even tx from ttl converte ( rx on board) is working properly ?
Because if you just hit " enter" you should have the # prompt

0
paradigman

paradigman

Posted
Posted
  On 10/14/2021 at 7:35 PM, jock said:

@paradigmanah ok, now I get it... you posted the multitool uart log.
...

The problem that you don't see anything on HDMI may be due to multiple factors: the kernel, the HDMI cable and, most of all, the monitor.

If you can try another HDMI cable and monitor we can start to remove some variables from the equation.

Expand  

I understand that we start the investigation at the basics, but it tells a lot about the error that if I boot the box without an SD card, so with the factory android firmware, then with this cable, I can see the image on the monitor flawlessly (which anyway not old: ASUS VX239).

 

@fabiobassa: Yes, if I hit "enter", I get a # prompt. And how from there?

0
fabiobassa

fabiobassa

Posted
Posted
  Quote

Yes, if I hit "enter", I get a # prompt. And how from there?

Expand  

 the hdmi part is more a @jock  's investigation 
The prompt is just telling you are actually logged in the box.  Unfortunately the script to install and manage things on uart is not so good as on hdmi :-(

 

  Quote

I understand that we start the investigation at the basics,

Expand  


Yes unfortunately when we don't have physically the board in our hands not so easy to debug, but please, have you the chanche to try another monitor, since we are quite sure that the cable is ok ?

0
jock

jock

Posted
  • Author
Posted
  On 10/14/2021 at 7:46 PM, paradigman said:

I understand that we start the investigation at the basics, but it tells a lot about the error that if I boot the box without an SD card, so with the factory android firmware, then with this cable, I can see the image on the monitor flawlessly (which anyway not old: ASUS VX239).

Expand  

I understand your doubts, but the fact is that the kernel that runs on android is not the exact copy of the kernel of the multitool, probably neither the same version.

HDMI timings are differently arranged on multitool and armbian kernels, that's something inherited from libreelec patches I merged in, so I can just trust them.

This in turn causes different behaviour than stock android.

 

0
nafanovich

nafanovich

Posted
Posted
  On 10/14/2021 at 10:39 AM, jock said:

@nafanovich Well it's strange your image works on the sdcard and does not work when you install it on eMMC.

You can use armbian  (or the multitool) on SD/USB to mount the internal eMMC and modify armbianEnv.txt from there.

Change verbosity=1 to verbosity=7 too, so you see the kernel log on the serial which is way important for your situation.

 

Expand  

log:

  Reveal hidden contents
0
tommy

tommy

Posted
Posted

Finally, I could compile Lakka (Retroarch fork of LibreElec) for RK322x,  thanks to Lakka  LibreElec and modified knaerzche LibreElec for provide neccessary source code.

Retroarch interface has more functions than LibreElec game addon. Performance is excellent to PSX @60fps level. I could upload binary if someone interested.

 

  On 10/12/2021 at 10:33 AM, jock said:

@tommy

I would say you to give a chance to the mainline kernel too. Lima driver has decent performance and you don't have to lose time around proprietary Mali OpenGLES-only drivers swapping links under the hood. (BTW for legacy kernel there is the media script link on first page that should install the libraries in the right places).

 

For snes9x you should not need particular load on the GPU since it has just to render the framebuffer. Lima performances should suffice.

 

Expand  

 

I just look into mainline kenel. I use Armbian Focal kernel 5.10.y at main download page still study how to use hardware acceleration by Lima driver. Retroarch looks like slow because no active hardware acceleration yet. Hope I will figure out soon.

0
jock

jock

Posted
  • Author
Posted
  On 10/15/2021 at 9:53 AM, tommy said:

I just look into mainline kenel. I use Armbian Focal kernel 5.10.y at main download page still study how to use hardware acceleration by Lima driver. Retroarch looks like slow because no active hardware acceleration yet. Hope I will figure out soon.

Expand  

Can't remember if the upstream image has lima enabled in X11 or not.

Anyway check /etc/X11/xorg.conf.d/40-serverflags.conf and in case Option "AccelMethod" is set to "none" change it to "glamor".

It was set to none because X.org performance is very slow with older lima + mesa and software cursor.

If you wish to try I can provide a cutting edge image with kernel 5.14 and Ubuntu Hirsute (=> much newer Mesa) and hardware cursor, which should radically change the experience.

 

BTW: I'm pretty interested in trying retroarch just for fun; there is a "gaming on arm" club where you may give some hints or write down a tutorial on how to compile. Maybe @NicoD may give you some help.

1
tommy

tommy

Posted
Posted

I am interest your new cutting edge armbian image. Could you provide me your links? Thank you in advance.

 

 

 

Lakka image files for rk322x

 

rk3228a-mxq4kpro <- I tested this image

https://mega.nz/file/dJgDXIbB#7yOLyHmuptrP_8aswOCxKdk2aEffOQDUnH-9lIAg8jw

 

rk3229-a95xr1

https://mega.nz/file/hIwFHCrZ#nWeLsKnBCXcCwtsDp8_MklgYsqEF5r0N-a3iVQUeLXI

 

Any images has dtb in first fat partion. You may edit extlinux/extlinux.conf to correct device. It has samba service that can upload game roms but in my case wifi driver not work, only LAN work for my tv box.

 

 

2
jock

jock

Posted
  • Author
Posted

@nafanovich sorry but can't see anything wrong in the log you provided, except for the last message Failed to start Create System Users which is something I really don't have any clue about.

 

You may try to boot from sdcard, mount /dev/mmcblk2p1 on /mnt and add emmc-pins overlay to overlays in /mnt/boot/armbianEnv.txt directory reboot and see if something changes, but otherwise it looks like a pretty clean log to me with no particular hardware errors.

0
tommy

tommy

Posted
Posted

 

  On 10/15/2021 at 5:05 PM, jock said:

@tommy Here it is, the latest 5.14 kernel with Ubuntu Hirsute: https://users.armbian.com/jock/rk322x/Armbian_21.11.0-trunk_Rk322x-box_hirsute_edge_5.14.12_xfce_desktop.img.xz

 

I didn't even test the image, I hope it boots!

Expand  

 

I tested in terminal mode ,'$ sudo systemctl set-default multi-user" which has really good performance. Your cutting edge armbian can achieve PSX full 60fps speed in retroarch (however, in legacy, I use wrong parameter which seriously impact performance) and snes9x framerate improved. I really surprise it is faster than legacy in my case. Really great.

2
nafanovich

nafanovich

Posted
Posted
  On 10/15/2021 at 2:42 PM, jock said:

You may try to boot from sdcard, mount /dev/mmcblk2p1 on /mnt and add emmc-pins overlay to overlays in /mnt/boot/armbianEnv.txt directory reboot and see if something changes, but otherwise it looks like a pretty clean log to me with no particular hardware errors.

Expand  

same things

  Reveal hidden contents

 

where is "first run" scripts located?

0
marshall

marshall

Posted
Posted

Loving the project a lot, got it working on an MXQ PRO 4K rk3229 complete with WiFi support after I figured out the correct led gpio config... But it was running off of the SD card, wasn't able to install to NAND, not sure why but it wasn't a big deal

 

When I tried it on another board, an MXQ PRO 4K 5G it worked smoother with me being able to install to eMMC from multitool, however I'm experiencing gpio issues... The LEDs don't work as well as the other board and this also affects WiFi which is an ESP8089. It seldomly gets detected at random, and most of the time it's not detected at all.

 

Tried all the info I could find about this by changing up the CPU types to lower speeds, editing esp8089.conf setting crystal 26MHz to 1, and trying every LED configurations on rk322x-config... None of those worked, as I'm assuming the gpio pins are still misconfigured

 

If there's a specific procedure to fix the issue, I'd gladly follow to learn more

 

Thanks for this great project, it's been fun

 

Board Images: https://imgur.com/a/2eA3pmE

0
jock

jock

Posted
  • Author
Posted

@marshall

Hello.

The NAND problem is just unsolvable without any clues. The commercial name of the tv box is totally useless: what is outside is not what is inside; the board name may give some more hints about but surely won't solve the mistery.

 

About the latter board, that's a very recent eMCP board, some of which we've never experienced directly but just seen in photos.

Providing a dtb overlay that fixes leds is usually quite simple, less simple is understanding what's wrong with the esp8089.

 

0
marshall

marshall

Posted
Posted

The esp8089 works flawlessly on the stock android firmware, I made a backup of it with the multitool if that would be of any help,

And I mentioned the LEDs because I've never gotten it to work properly on the second board 100%... But I've gotten it to work perfectly on the first board, as a network activity indicator.

On the second board, the system seems to be just using it either as a CPU activity monitor, or just a straight one color  when the system is on... When setting triggers on (iirc from the top of my head) `/sys/class/leds/working` and `/sys/class/leds/auxiliary` it either only gets detected as one LED... Or one of them is always on and only the other is controllable...

 

Which is why I'm speculating that it might be a misconfiguration in terms of the pinout, rather than a hardware issue

 

Thanks

0
jock

jock

Posted
  • Author
Posted
  On 10/18/2021 at 7:38 AM, marshall said:

The esp8089 works flawlessly on the stock android firmware, I made a backup of it with the multitool if that would be of any help,

And I mentioned the LEDs because I've never gotten it to work properly on the second board 100%... But I've gotten it to work perfectly on the first board, as a network activity indicator.

On the second board, the system seems to be just using it either as a CPU activity monitor, or just a straight one color  when the system is on... When setting triggers on (iirc from the top of my head) `/sys/class/leds/working` and `/sys/class/leds/auxiliary` it either only gets detected as one LED... Or one of them is always on and only the other is controllable...

 

Which is why I'm speculating that it might be a misconfiguration in terms of the pinout, rather than a hardware issue

 

Thanks

Expand  

Yes thanks, indeed the original firmware will be very useful!

Actually I just need the original device tree overlay, but if you're not able to extract it by yourself, just post the first 32 megabytes of the original firmware and they will suffice.

 

Most surely the led problem on the second board is a misconfiguration: that's something I can adjust taking a look into the dtb. esp8089 driver and firmware are already into the package, but it is very probable that your board has a particular gpio configuration to let it work, and the dtb can reveal that to us. Curious is the fact that it is randomly detected: maybe you can check if the device is correctly detected after warm reboots and not detected after cold reboots, or things like that. Any hint of this kind (plus some dmesg) are helpful with such situations.

 

By the way, I see a big 26MHz crystal near the esp8089 on the board, so surely the .conf file has to be set accordingly.

0
marshall

marshall

Posted
Posted

Here's the backup: https://mega.nz/file/MBV0FbjC#SFhkL9yvnxsE9EKuwhJBSX2J_cGom3HQbZrwg1FjmdM

 

I'm not sure if the literal first 32 bytes of the extracted file is significant, but if it actually is, for convenience, here's the hex dump for the first 32 bytes:  

00 01 83 00 00 02 00 00 00 00 00 00 00 00 00 00 16 00 00 00 00 00 00 00 64 06 00 00 00 00 00 00

 

To my surprise, when I booted the second board up just now, it detects the esp8089 and even lists it at ifconfig as wlan0... And after a few reboots (probably 2 or 3) and switching between crystal_26M_en values, now it doesn't... It randomly works on 26M regardless if set as 1 or 0

 

At that time that it was working though, I still can't connect with it for some reason (using armbian-config), and ifconfig lists the wlan0 interface as an "Ethernet" device, + yeah the LEDs are still acting weirdly

 

This is all using the led-conf5 option btw, which, shows `working` and `auxiliary` for `/sys/class/leds` but uses it as if it's just one LED and switches between red/blue, instead of being 2 individually controllable LEDs, and apparently its default trigger is mmc2 activity

 

I haven't set it up on my network yet to copy/paste dmesg as that would require me to set a static IP for it and place it far away from my workdesk, and near the router as only the ethernet works... but here's a photo of dmesg, grepping through possibly relevant keywords: https://imgur.com/a/VjpJUkr

 

Thanks for your time!

0
paradigman

paradigman

Posted
Posted
  On 10/14/2021 at 8:55 PM, jock said:

I understand your doubts, but the fact is that the kernel that runs on android is not the exact copy of the kernel of the multitool, probably neither the same version.

HDMI timings are differently arranged on multitool and armbian kernels, that's something inherited from libreelec patches I merged in, so I can just trust them.

This in turn causes different behaviour than stock android.

 

Expand  

I tried it with another monitor, but the same result, i.e. no picture on the HDMI output.
If I switch the monitor between the inputs, this extra information will appear on the UART output:
 

  Reveal hidden contents

 

0
jock

jock

Posted
  • Author
Posted

@marshall

Thanks for the 32 bytes of the image, but actually I meant the first 32 megabytes eheheh :D. No problem however, I managed to extract the original dtb from the tvbox image.

 

I took a look into and yes, there is the need to adjust the secondary led and something seems to be a bit different for wifi too.

I will craft a special overlay so you can test the whole thing.

 

 

0
jock

jock

Posted
  • Author
Posted

@paradigman I don't what to say, that's just a problem that has been rarely seen before and I don't have such a problematic board to investigate further.

I see that your board has "R29-MXQ" marking, but we have only seen "R28-MXQ" which is indeed different.

 

I may suggest you, if you didn't already, to try an image with the mainline kernel.

0

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
 Share
Go to topic listing
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines

  I accept