Ramblings and progress with the RK3399

[pbies]

Hello All,

 

I recently bought Orange Pi RK3399 and had some problems with it. At this moment I've solved most of them, but still no luck with such an old OS like Ubuntu Server 16.04 or Debian 9 which were provided by the manufacturer (images to flash).

 

What I am thinking now is to move this board to Armbian. Will it work?

 

How should I in this case organize the partitions on eMMC and where to load kernel and rootfs?

 

I don't see this board in supported ones.

[hjc]

2 hours ago, pbies said:

What I am thinking now is to move this board to Armbian. Will it work?

Of course it could work, but as I said before, you'd need at least mainline or a recent (2017.x) u-boot working. This means if your board vendor does not provide mainline u-boot support (like most RK3399 devices except Firefly), you need to somehow write a u-boot device tree for your board, just like what I did for NanoPC T4. Unfortunately, you need to do this by yourself.

 

2 hours ago, pbies said:

How should I in this case organize the partitions on eMMC and where to load kernel and rootfs?

Armbian use a single ext4 partition, so you don't need to mess up with those default Rockchip partition layout (7 partitions is really a mess for an SBC)

 

BTW, I don't know exactly how Orange Pi RK3399 works, but normally, RK3399 will boot from SD card if eMMC/SPI flash is not available or there's no u-boot SPL on eMMC/SPI flash. You may erase the eMMC (blkdiscard) and use an SD card for debugging, which is more convenient.

[pbies]

1 hour ago, hjc said:

normally, RK3399 will boot from SD card if eMMC/SPI flash is not available

 

This would be not a bad idea, if the SD card wouldn't be few times slower than the eMMC.

 

As I have good SD card, exactly microSD, it does not give reasonable speeds on this board. 20MB/s for linear read is not enough.  eMMC is giving me 200MB/s in average. I suspect that OS is too old to provide faster speed for this SD card. If it is possible to run it faster.

 

I am not enough advanced about these boards to mess with u-boot.

[chwe]

5 hours ago, hjc said:

Armbian use a single ext4 partition, so you don't need to mess up with those default Rockchip partition layout (7 partitions is really a mess for an SBC)

funnily mediatek had 5 or so...  I think this comes from this 'fallback' methods for updates.. An SBC is allowed to fail.. some random TV box/router shouldn't end in a boot loop when doing updates.. But as soon as  u-boot works and is able to read ext4 this doesn't matter anymore..  

 

3 hours ago, pbies said:

As I have good SD card, exactly microSD, it does not give reasonable speeds on this board. 20MB/s for linear read is not enough.  eMMC is giving me 200MB/s in average. I suspect that OS is too old to provide faster speed for this SD card. If it is possible to run it faster.

seq. read write performance doesn't matter.. read through @tkaiser comments on this topic...

 

for sure, good eMMC will perform better. But thinking about eMMC before the board boots from SD-Card is not worth to think about.. step by step...

 

4 hours ago, pbies said:

I am not enough advanced about these boards to mess with u-boot.

read through this one:

and this one:

 

I think @hjc and me were beginners when we started our board bring up... you can learn it by reading and testing.. There's a android source code folder in orangepi's downloadsite.. Look inside, look what's there and what is missing. RK provides a bunch of stuff on their github repo. And there are several RK3399 boards in WIP to get some background how stuff is done. But cause nobody else with experience in board bring up has this board at hand, you've to do the work on your own. Bring up a device which you don't have at hand to test doesn't work. 

[tkaiser]

27 minutes ago, chwe said:

seq. read write performance doesn't matter

 

For some use cases they do matter but unfortunately some users prefer questionable 'benchmarking' methods and ignore the importance the block size makes when accessing storage.

 

That's why common tools like hdparm, dd and pv all simply suck when it's about to determine maximum storage performance or insights in general:

 

SBC users seem to not want to accept most simple facts:

• testing with hdparm is always wrong since it uses just 128k block sizes when reading from disk. 128k were a sufficient number last century when the defaults were defined since back then the fastest HDD storage was slow like hell. Today it's just a joke to test with 128k.
• testing with pv is always wrong when not using -B switch (which no one does) since buffer size will then be calculated dynamically [1] and you get random numbers
• testing with dd can result in 'good' numbers if dd flags and block size has been set accordingly. Problem is: Usually these informations are not shared when dd 'benchmark' numbers are published so you simply not know whether testing was correct or just the usual 'benchmarking gone wrong'. That's why looking at dd numbers is almost always useless

[1] quoting pv manual page:

-B BYTES, --buffer-size BYTES
    Use a transfer buffer size of BYTES bytes. A suffix of "K", "M", "G", or "T" can be added to 
    denote kibibytes (*1024), mebibytes, and so on. The default buffer size is the block size of 
    the input file's filesystem multiplied by 32 (512KiB max), or 400KiB if the block size cannot 
    be determined.

 

[MickMake]

On 8/29/2018 at 8:03 PM, tkaiser said:

For some use cases they do matter but unfortunately some users prefer questionable 'benchmarking' methods and ignore the importance the block size makes when accessing storage.

  

That's why common tools like hdparm, dd and pv all simply suck when it's about to determine maximum storage performance or insights in general:

I agree with this. When I run storage benchmarks for my review videos, (using Phoronix), I run through the full gamut of options, (block sizes, reads, writes, etc), as well as using several different tools. This usually ends up taking a week to complete just the storage benchmarks, but gives a *MUCH* better picture on overall performance.

 

[Myy]

Best post is last post

 

So I got an Orange Pi RK3399 this Saturday morning, and my current objectives are :
1. Be able to install a bootable Linux on the eMMC.
  By bootable, I mean : You plug it, it boots the Linux system, no questions asked.
2. Be able to generate an image of the Linux installation and be able to flash it using the Rockchip rkdeveloptool

3. Compile and install a Rockchip kernel specialized for Orange Pi RK3399 <- Currently focusing on this
4. Integrate the image generation and installation into Armbian
5. Compile and install a mainline kernel (4.19-4.20), and get at least, the following working :
  - HDMI and X11 display
  - OpenGL rendering
  - Networking
  - USB.

 

Anyway, I'm currently at phase 2, so don't hold your breath.
There's so little documentation on the Orange Pi website, that I have to adapt documentation about others RK3399 boards and make do.

 

# Random procedures I used at the moment

 

Now, if you intend to follow any step in these procedures, which I'll advise *against* since you might just have a "easy to install" Armbian image in a few weeks :

 

1. Learn how to get a serial console that works.
This is *VERY* useful.

 

• Plug in an RS232<->USB connector on the board, by following the pinout provided in the Orange Pi RK3399 User Guide
  
  I don't use the "Debug" one, since I have no documentation on this one and I hate trying different combinations with serial connectors.
  To get an output, you need to be at 1500000 bauds.
  I use picocom as my serial terminal. Minicom just doesn't do the trick to me.

  sudo picocom -b 1500000 /dev/ttyUSB0

 

  # If your system is properly configured, just omit the 'sudo'.
  picocom -b 1500000 /dev/ttyUSB0

 

• Power on the board and see if you get something on the screen.
• Check the pinout again if you don't and try again.

 

2. Download the Orange PI RK3399 boot, trust and loader images.
Again, *life savers* :

• Get RKTool_Linux.tar.gz from Orange Pi "Mega" folder : https://mega.nz/#F!K1BQFQjL!W9rYedaoGpV4y8vHfrGEfQ!ekYA0AwR
• Extract it.
• Extract the Linux_rockdev_2015-06-17_for_RK3399.zip archive contained in it
• Copy the RK3399MiniLoaderAll_V1.05_DDR600MHz.bin, uboot.img and trust.img contained in it to some place

 

3. Get rkdeveloptool if you're using a Linux system. And I have no idea how to perform any of the step forementionned from Windows.
You've been warned !
 

git clone https://github.com/rockchip-linux/rkdeveloptool --depth 1
cd rkdeveloptool
autoreconf -i
./configure
make
cp rkdeveloptool /path/to/your/local/bin

Now, when I got the board, I booted on Android, got into developer mode, got an ADB shell and I started to backup a few things :

• The eMMC image : dd if=/dev/block/mmcblk0 of=/storage/path/to/my/USB/Stick/Backup.img
• Made a copy of the cmdline :

 

earlycon=uart8250,mmio32,0xff1a0000 swiotlb=1 androidboot.baseband=N/A androidboot.selinux=permissive androidboot.hardware=rk30board androidboot.console=ttyFIQ0 init=/init mtdparts=rk29xxnand:0x00002000@0x00002000(uboot),0x00002000@0x00004000(trust),0x00002000@0x00006000(misc),0x00008000@0x00008000(resource),0x0000C000@0x00010000(kernel),0x00010000@0x0001C000(boot),0x00018000@0x0002C000(recovery),0x00038000@0x00044000(backup),0x00040000@0x0007C000(cache),0x00600000@0x000BC000(system),0x00008000@0x006BC000(metadata),0x00002000@0x006C4000(baseparamer),0x00000400@0x006C6000(frp),-@0x006C6400(userdata) storagemedia=emmc loader.timestamp=2017-08-23_15:15:41 hdmi.vic=-1 androidboot.mode=emmc

 

• Made a copy of the parameters.txt used in their U-Boot :  

FIRMWARE_VER: 6.0.1
MACHINE_MODEL: RK3399
MACHINE_ID: 007
MANUFACTURER: RK3399
MAGIC: 0x5041524B
ATAG: 0x00200800
MACHINE: 3399
CHECK_MASK: 0x80
PWR_HLD: 0,0,A,0,1
#KERNEL_IMG: 0x00280000
#FDT_NAME: rk-kernel.dtb
#RECOVER_KEY: 1,1,0,20,0
#in section; per section 512(0x200) bytes
CMDLINE: androidboot.baseband=N/A androidboot.selinux=permissive androidboot.hardware=rk30board androidboot.console=ttyFIQ0 init=/init
mtdparts=rk29xxnand:0x00002000@0x00002000(uboot),0x00002000@0x00004000(trust),0x00002000@0x00006000(moasc),0x00008000@0x00008000(resource),0x0000C000@0x00010000(kernel),0x00010000@0x0001C000(boot),0x00018000@0x0002C000(recovery),0x00038000@0x00044000(backup),0x00040000@0x0007C000(cache),0x00600000@0x000BC000(system),0x00008000@0x006BC000(metadata),0x00002000@0x006C4000(baseparamer),0x00000400@0x006C6000(frp),-@0x006C6400(userdata)

 

• All these information proved to be quite useless, beside the 0x00002000@0x00004000(trust) which... weren't that useful anyway.

 

I then botched the eMMC and... you could do the same thing by just doing dd if=/dev/zero of=/dev/block/mmcblk0 ...
I just used their "update_tool" and fired an "update_tool ef" instead. Same thing.

 

Which led me to a common issue with Rockchip boards :

The dreaded quiet serial console on reboot.
However, the board was pluged into my PC and lsusb still showed this :
Bus 001 Device 012: ID 2207:330c  
The important part is : ID 2207:330c  

 

If you get into this situation, you can get a "noisy UBoot" back again by plugging the board to your PC, with an USBC<->USB cable and do :
 

/path/to/your/rkdeveloptool db /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin
/paht/to/your/rkdeveloptool ul /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin

 

However, before doing that, it might be a good moment to compile yourself a uboot image, since it's the *only moment* where you can flash a U-boot image easily.

 

To compile U-Boot, you'll need the arm-trusted-firmware.

 

To compile the arm-trusted-firmware for RK3399, you'll need the arm-none-eabi- toolset.
You can grab it here : https://developer.arm.com/open-source/gnu-toolchain/gnu-rm/downloads
The x64 version link : https://developer.arm.com/-/media/Files/downloads/gnu-rm/7-2018q2/gcc-arm-none-eabi-7-2018-q2-update-linux.tar.bz2?revision=bc2c96c0-14b5-4bb4-9f18-bceb4050fee7

• Unpack the toolset somewhere, and then export PATH=$PATH:/path/to/where/you/unpacked/the/toolset/bin/
• Check that arm-none-eabi-gcc can run.
• Then, compile the ARM Trusted Firmware :

git clone https://github.com/ARM-software/arm-trusted-firmware.git --depth 1
cd arm-trusted-firmware
make realclean
make CROSS_COMPILE=aarch64-unknown-linux-gnu- PLAT=rk3399 bl31
cp arm-trusted-firmware/build/rk3399/release/bl31/bl31.elf /tmp

 

Then, compile U-Boot :

git clone https://github.com/u-boot/u-boot --depth 1
cd u-boot/
export CROSS_COMPILE=aarch64-unknown-linux-gnu- # Set this to your ARM64 GCC toolset prefix. Set nothing if you're compiling on an ARM64 machine.
export PLAT=rk3399
export ARCH=arm64
make firefly-rk3399_defconfig
make -j4 # N virtual CPU. On high-end AMD Ryzen, for example, put -j16
cp /tmp/bl31.elf .
make u-boot.itb
cp u-boot-dtb.bin /tmp

 

Then get the "Rockchip tools" and make a "Rockchip compatible U-Boot image" :

git clone https://github.com/rockchip-linux/rkbin.git --depth 1
cd rkbin/tools
./loaderimage --pack --uboot /tmp/u-boot-dtb.bin /tmp/uboot.img

 

Then remember where you backed up RK3399MiniLoaderAll_V1.05_DDR600MHz.bin
And do :

/path/to/your/rkdeveloptool db /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin # Write the Bootloader
/paht/to/your/rkdeveloptool ul /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin # Which is also their firmware
/path/to/your/rkdeveloptool wl 0x4000 /path/to/uboot.img # Write the U-Boot image
/path/to/your/rkdeveloptool wl 0x6000 /path/to/trust.img # Write the Trust bits
# Unplug and plug the power cable again to provoke a hard reboot of the board.
# The "reboot" button might not be sufficient to reinitialize the eMMC correctly, leading to U-Boot stalling when loading.
# Just unplug and plug the power cable again.

 

Now reboot and you should have a noisy Rockchip U-Boot again, that then boots into your freshly compiled UBoot.
If you want to reflash a UBoot image, you generally HAVE TO reflash the Miniloader image and let it boot so that it accepts writing anything with rkdeveloptool.

 

Now, the freshly compiled UBoot can boot an SDCard and, also, have a 5 seconds timer (instead of 0...).

I used that way and burned an Armbian image for NanoPC T4 on an SDCard and booted it with that UBoot.

However, if you insert the SDCard on boot, then Rockchip's Loader crashes... Seriously...

So, I just waited until everything went on, U-Boot tried to find a bootable image and failed... And THEN I put the SDCard and typed :
boot
On the serial console.

 

Now, when booting the Armbian image, the serial console is setup at 115000 bauds so you'll see random characters on the serial console.
Just drop the speed of the serial console. On picocom, do CTRL-A CTRL-D until you drop at 115200. You can up the speed back with CTRL-A CTRL-U.
Then push a few times Enter when the scribblings stopped appearing on the serial console and you should see :
 

nanopct4 login:  

Type "root" and then "1234", and then "1234" again and then type a new password, create a new user and an X11 session should open...

 

Anyway, the X11 session was useless to me since the NanoPC T4 image is not compatible with Orange Pi boards USB-Wise !
So you'll have no USB. Which means no mouse, no keyboard. And also no Wifi/Bluetooth, just in case you're wondering...

 

Still, the serial console worked, so I was able to partition the eMMC, put a GPT partition table and create a partition that starts after the first 16MB (or MibiBytes) just to be sure.

Then I put the Armbian image into this eMMC partition by doing a stupidly complex procedure, which I do not recommend :

• Halt the system
• Mount the SDCard on another Linux system
• Extract the ext4 partition of the Armbian NanoPC T4 image file by first :
  • configure losetup and mount the partition inside the file ( https://askubuntu.com/questions/69363/mount-single-partition-from-image-of-entire-disk-device )
  • sudo dd if=/dev/loop0 of=/my/SDCard/mountpoint/ArmbianPartitionOnly.img
• Boot the board back with the SDCard back
• Log as root and check which /dev/mmcblk is the eMMC, using cfdisk.
  The eMMC is 16GB, while my SDCard is 64 GB
  In my case /dev/mmcblk1 was the eMMC
• Type dd if=/path/to/ArmbianPartitionOnly.img of=/dev/mmcblk1p1

 

I'm sure there's way safer and easier ways to burn an Armbian image into an eMMC partition.

 

I then halted the board, removed the SDCard, booted the board again and saw it boots into Armbian automatically, through the eMMC partition.

 

So, I'm now at phase 2 : Get a snapshot of the partition and be able to flash it again with rkdeveloptool.

 

I'm putting this here, since it might help other people tinkering with the board, trying to get something out of it besides "getting Android working".

Still, if you just want an Armbian image : WAIT.

 

Attachments :

* A picture showing wow to get the Debug USART using the standard GPIO block.

* OrangePI RK3399 Schematics

* What seems to be the main 5V converter chip.

* What seems to be the main 3.3V converter chip

* What seems to be the main charge controller

 

orangepi-rk3399-v1_5_Shcematic.pdf

nb679_r1.03.pdf

nb680_r1.03.pdf

bq25700a.pdf

[Seasalt]

Why did you buy the Orange Pi rk3399?

 

It appeared to me the Armbian team were all rallying around the Friendly Elec Rk3399.

 

How long do you think it will be before you have a stable Orangepi RK3399 image?

 

If my experience with the OrangePi PC2 is anything to go by,  it took 18 months to get it stable.

[Myy]

21 hours ago, Seasalt said:

Why did you buy the Orange Pi rk3399?

I was offered this board to see what I can do with it. Just like the other boards I got (MiQi, Tinkerboard, ...)

 

Concerning "plans", I won't have any until I at least get a way to install a Linux image with a Rockchip kernel that "kind of work" correctly, in an "automated fashion" (scripted and useable with rkdeveloptool and Windows tools, so that you can reflash it at any moment if things go wrong).

Then I'll start mimicking their drivers on latest kernels and see if there's any major barrier, though I will have to deal with a few other issues before that.

 

I still see that the VPU is neither documented anywhere on their schematics, so there's at least that. If they could at least document which channel the VPU uses with the DMA subsystem, it might help a bit.

 

For the GPU side, given how fast the Panfrost project is progressing, it might be possible to overcome Mali binary drivers issues with these drivers. The Rockchip team was never able to provide Vulkan-ready binary drivers for their Mali-T7xx and Mali-T8xx chips, and the only one ARM provided with Vulkan support was with fbdev support only. There's of course drivers able to use DRM/KMS but you're "limited" to OpenGL ES 2.x / 3.x and you need an old GPL kernel driver that's starting to pile up patches to get it work with latest mainline kernels.

 

Now, given how the NanoPC T4 4.4 kernel was able to provide : HDMI output, (kind of shaky but working) Ethernet, eMMC and SDCard support, there's chances that "Orange Pi" specific issues might be fewer than expected. What remains, in terms of connections, are : SATA ports, USB, USBC, Audio, Wifi and Bluetooth.

Bluetooth might just be like the Tinkerboard : Wake up the Wifi/Bluetooth chip, flip a few GPIO and then find a way to send the local "binary blob firmware" to the chip.

 

There's also the MIPI connectors, but I got nothing to connect to these ports. Same for the battery.

 

Anyway, I'll be a bit busy this week but I might be able to advance a bit this week-end.

[Myy]

Okay, turns out that you need to write the "trust.img" BEFORE writing a U-Boot image, else the U-Boot image won't be booted with either one of these errors :

 

LoadTrustBL error:-1

hdr 0x0:0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 

 

Even if you install the Trust.img afterwards.

 

I still have to figure out why, but the procedure for recovering from a mute RK3399 with an empty MMC is :

 

/path/to/your/rkdeveloptool db /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin # Write the Bootloader
/paht/to/your/rkdeveloptool ul /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin # Which is also their firmware
/path/to/your/rkdeveloptool rd # Reboot
sleep 5 # If you're just copy-pasting line by line, it should be ok. Else, wait 5 seconds for the system to reboot
/path/to/your/rkdeveloptool wl 0x4000 /path/to/trust.img # Write the Trust bits
/path/to/your/rkdeveloptool rd # Reboot
sleep 5 # If you're just copy-pasting line by line, it should be ok. Else, wait 5 seconds for the system to reboot
/path/to/your/rkdeveloptool wl 0x2000 /path/to/uboot.img # Write the U-Boot image

 

You can use your own compiled U-Boot image. However, I still don't know how to generate the trust.img so I used the one provided by OrangePi in their RKTool_Linux.tar.gz .

[Myy]

It seems that the "wl" instructions write are offseted by 0x2000 . I think I got a glimpse of that information while reading the Firefly RK3399 documentation. However, I can validate that the trust.img is written at 0xC00000 bytes in the eMMC, which translates to 0x6000 with 512 bytes sectors. Since I wrote it as 0x4000, it clearly seems that a +0x2000 offset is added...

 

Also, my U-Boot image is present at 0x800000 which translates to sector 0x4000, which clearly shows the offset again.

 

However, the same U-Boot image is also present at 0x400000, which translates to sector 0x2000. Since I messed up with the "trust.img" thingy, my guess is that during the first "mute phase", every write is done at the correct address. THEN, on reboot, every write is offseted by +0x2000.

So I guess that, if I write a new eMMC image, I'll have to do it during the first phase, since on the second phase, everything will be offseted by +0x2000 sectors (0x400000 bytes) and the partition table won't be written at the right place.

 

I'll try that this Sunday and see how it fares.

[Myy]

Alright, turns out that the

/path/to/your/rkdeveloptool rd # Reboot

step in my procedures wasn't rebooting the board anyway.

So, basically, when in Maskrom mode (the mode you are in when the serial console is quiet, or when you keep the "Recovery" button pushed with the Right U-Boot), all writes are "as-is".

If you reboot (unplug and plug the power cable back again) with a Loader, a Firmware and the Trust image, you'll get into a special "rockusb" mode that offset everything by 0x2000 sectors (0x2000 * 512 bytes).

 

Since I'm not interested in that special rockusb mode, here's the "right" process for installing a U-Boot and Trust.img when you are in "quiet" mode :

 

/path/to/your/rkdeveloptool db /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin # Write the Bootloader
/path/to/your/rkdeveloptool ul /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin # Which is also their firmware
/path/to/your/rkdeveloptool wl 0x4000 /path/to/uboot.img # Write the U-Boot image
/path/to/your/rkdeveloptool wl 0x6000 /path/to/trust.img # Write the Trust bits
# "Turn it off and on again" by unpluging and plugging the power cable back (Either the USB-C cable or the power plug)
# Rebooting through the little button generate "Tinkerboard"-esque issues with the eMMC, which won't be properly power up anymore.

 

Now, if you already have an eMMC image, you can do this instead :

 

/path/to/your/rkdeveloptool db /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin # Write the Bootloader
/path/to/your/rkdeveloptool ul /path/to/RK3399MiniLoaderAll_V1.05_DDR600MHz.bin # Which is also their firmware
/path/to/your/rkdeveloptool wl 0 /path/to/eMMC.img # This will be STUPIDLY slow. You'll have to wait betweens 10 and 30 minutes.
# Unplug and plug the power cable back again, after the command finishes.

 

I tested successfully these last steps and, therefore, have been able to directly flash my semi-working Armbian installation based on the Firefly RK3399 image.

 

So, the current step now is "Getting a Rockchip kernel working correctly with this board".

[Myy]

I extracted the "rk3399-orangepi.dts" from their Linux archive and... it looks like something done by an employee that didn't have much time to understand how the different drivers worked, so he put #ifdef here and there to disable the devices when they're not needed.

 

While I think it's terribly ugly, at the same time, there isn't many way to disable the initialization of hardware you do not want...

 

Anyway, here's the DTS file. I'll take a look at the different RK3399 kernel trees, see if one got a "slightly better" DTS file :

 

https://gist.github.com/Miouyouyou/00f8b006f96b21b0383196f61045df76

 

 

[Myy]

Okay trying this DTS "as-is" with a Rockchip Linux kernel led me to No HDMI, a barely working serial console and other issues. But, at least, it seems that the USB is alright.

 

I'll try with Orange Pi sources tomorrow and see where it leads me.

[DataMannen]

On 12/3/2018 at 7:38 PM, Myy said:

Okay trying this DTS "as-is" with a Rockchip Linux kernel led me to No HDMI, a barely working serial console and other issues. But, at least, it seems that the USB is alright.

 

I'll try with Orange Pi sources tomorrow and see where it leads me.

 

Hi Myy

 

I just wanna say that I'm following this thread with great interest and a thank you! Much appreciated to finally get something else potentially running on my OPi RK3399 than the images from the OrangePi website.

KEEP UP THE GOOD WORK! :-)

 

Personally, I managed to compile idbloader.img, rk3399_loader_v1.08.106.bin, trust.img, uboot.img, boot.img and then flash the eMMC (together with a rootfs-partition) to and boot up successfully. However, I have used the downloads provided by OrangePi for this and they are not really satisfactory, lacks certain drivers,  have eth0 problems, etc, etc. I have been stuck for quite some time but now you have given me back my hope. :-)

 

Note: To build the images, I have used their provided downloads OrangePi-RK3399_Linux4.4_V1.0_2018_1030.tar.gz (for build-scripts and tools) and OrangePi-rk3399_Debian9_20181016_HDMI IN.tar.gz (for the rootfs partition). Both are available here: https://mega.nz/#F!ggt3QaoJ!ZexfJRVdf7_6C1bs75-fJQ

 

Can I assist you in any way? I'm not really a "pro" at this, but I am learning as things move along...

[Myy]

On 12/15/2018 at 7:48 AM, DataMannen said:

 

Hi Myy

 

I just wanna say that I'm following this thread with great interest and a thank you! Much appreciated to finally get something else potentially running on my OPi RK3399 than the images from the OrangePi website.

KEEP UP THE GOOD WORK! :-)

 

Personally, I managed to compile idbloader.img, rk3399_loader_v1.08.106.bin, trust.img, uboot.img, boot.img and then flash the eMMC (together with a rootfs-partition) to and boot up successfully. However, I have used the downloads provided by OrangePi for this and they are not really satisfactory, lacks certain drivers,  have eth0 problems, etc, etc. I have been stuck for quite some time but now you have given me back my hope. :-)

 

Note: To build the images, I have used their provided downloads OrangePi-RK3399_Linux4.4_V1.0_2018_1030.tar.gz (for build-scripts and tools) and OrangePi-rk3399_Debian9_20181016_HDMI IN.tar.gz (for the rootfs partition). Both are available here: https://mega.nz/#F!ggt3QaoJ!ZexfJRVdf7_6C1bs75-fJQ

 

Can I assist you in any way? I'm not really a "pro" at this, but I am learning as things move along...

 

Hey ! Glad these information were useful to you !

 

I should have a bit more time to get into getting a kernel working correctly with the OrangePi this week, so I'm going to try getting a mainline kernel working with... basic things. Like HDMI, USB and decent network.

 

So, the next step is basically : Rebuilding a DTS file.

 

Like I said in this topic, the NanoPC T4 DTS file is surprisingly a good start.

The DTS file provided by Orange Pi, however, isn't very good. I mean, no HDMI, flacky Ethernet, flaky DMA (which seems to be related).

 

Fortunately, they seem to provide the schematics for these boards.

 

If you'd like to help, I guess that a good start would be to test various RK3399 oriented DTS files and see how they fare (is it better, worse, do you have better USB support, Ethernet, Video, ...).

This would help building a nice OrangePi specific DTS file very quickly.

[pfry]

Interesting. I figured the Rockchip devices would be - by and large - reference designs, given that most features are integrated into the SOC. Excepting things like DRAM type and speed, audio CODEC, Ethernet phy, etc. Certainly the power components seem standard (one board swapped the SY837/838, meaning the big array and GPU would get the other's voltage setting, although I imagine they're similar enough).

You should be able to flash images using Rockchip's AndroidTool under Windows, so long as you can read Chinese (it has English language files, but I didn't see a way to choose the language), although I've read of issues with Win10.

It should also be possible to choose the boot device: (http://opensource.rock-chips.com/wiki_Boot_option)

...but only after U-Boot is loaded from eMMC (or SD, if no eMMC present/enabled). And Armbian seems to do its own thing, partition-wise, and I haven't quite figured out what/how.

Good luck.

[DataMannen]

3 hours ago, pfry said:

You should be able to flash images using Rockchip's AndroidTool under Windows, so long as you can read Chinese (it has English language files, but I didn't see a way to choose the language), although I've read of issues with Win10.

 

I've always flashed my OrangePi RK3399 using Linux (rkdeveloptool) and never actually done it using Windows. However, I did fiddle around with it and noted that in order to change to english, open the file config.ini in the same folder as AndroidTool.exe, and change:

 

[Language]
Kinds=2
Selected=1
 

to

 

[Language]
Kinds=2
Selected=2
 

[DataMannen]

On 12/16/2018 at 9:54 PM, Myy said:

If you'd like to help, I guess that a good start would be to test various RK3399 oriented DTS files and see how they fare (is it better, worse, do you have better USB support, Ethernet, Video, ...).

This would help building a nice OrangePi specific DTS file very quickly.

 

Cool. I'm off for Christmas holidays now but will put my teeth into this once I get back in January.

 

[Myy]

I've been able to boot a mainline kernel with a barely working DTS file that can, still, boot up to X11 and provide a working serial console.

USB is "just broken" though, since the DTS file doesn't follow the schematics at all.

 

https://gitlab.com/snippets/1792126

 

Note that I'm currently using an external MMC file for the moment, since it's the easiest way for me. Replacing broken kernels on the eMMC is a pain, I can't get Gadget mode working with the DesignWare DWC3 connector on U-Boot at the moment (I just tried to enable the option, though, not tinker with the DTB and such...).

So my only way to test and replace kernels quickly is to use a Micro SDCard to boot the system on the OrangePi, and use an USB MicroSD reader on my main PC to replace broken kernels/DTB files.

 

So, tomorrow will be : Schematics to DTS time... Yay...

[Myy]

Ok, I uploaded a few Datasheet concerning the power supply bucks and controller used...

However, these files seem to all have "Unauthorized copy prohibited" written at the bottom of the pages, so I'll replace them by HTTP links tomorrow, in order to avoid any legal issues.

[Myy]

I forgot how fun it is to deal with DTS file. Especially the parts where you lose the SDcard detection when you enable USB3.

 

This... will take some time.

[martinayotte]

I've received my OrangePi-RK3399 recently ...

Just for trials, before doing any real development, I gave a try with a NanoPC-T4 4.19.y image compiled a month ago, and it worked "as is", even eMMC and WiFi is running fine.

That it is a good starting point !

 

Of course, to be able to boot from SDCard, we need to short the eMMC clock using the test point TP50265 :

 

[Myy]

Got the MMC back but it's still terribly slow for some operations.

 

Still I got a nice HDMI output using the CEC line instead of the two I2C lines :

https://gist.github.com/Miouyouyou/c8fbd78abbc39dc38bd050a4e9f6eed7

[Tido]

I post here the request from @balbes150 the title of this topic seems to fit for general testing and progress... hopefully

 

Quote

I would be very grateful if someone has the opportunity to try the latest versions of Armbian for Khadas EDGE on different models of RK3399.

To select the desired DTB, you need to edit extlinux/extlinux.conf (specify the available options in the image).

https://yadi.sk/d/ie_dx5x_w2xUIQ

I can give it a try on a Rock Pi 4B

 

[NicoD]

13 hours ago, balbes150 said:

To select the desired DTB , you need to edit extlinux/extlinux.conf (specify the available options in the image).

Привет, Балбес.
Do you have got a dtb file for the NanoPi M4 and/or RockPi4B? They've got working Armbian images. But would be cool to be able to use the same image on both boards. 
I wished all images were like this. Then we would not need new images for different boards with simular SoC's.
Can this then also be done with the LibreElec image?
I tried downloading your LibreElec image for the Edge. Got this message 'Произошла ошибка'. I don't have an Edge, but still wanted to let you know.
https://docviewer.yandex.ru/view/0/?*=hmrjCyQfuYVKHUtYSUu1p%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%3D

 

Спасибо, привет.

[TonyMac32]

5 hours ago, NicoD said:

Then we would not need new images for different boards with simular SoC's.

That would be ideal, but some of these boards have... peculiarities that cause source code issues (Tinker has some gymnastics around reboot that require some ugly hacking of the sd/emmc driver that can break other RK3288 devices) or userspace issues (manually toggling IO's to reset BT adapters).  In general, though, you'll find that most of our images are the same for the same SoC, just with DTB/boot script differences for each particular layout, it's why on any board bringup you see us just manually swapping DTB's to see if it will work out of the box.

[NicoD]

12 hours ago, TonyMac32 said:

That would be ideal, but some of these boards have... peculiarities that cause source code issues (Tinker has some gymnastics around reboot that require some ugly hacking of the sd/emmc driver that can break other RK3288 devices) or userspace issues (manually toggling IO's to reset BT adapters).  In general, though, you'll find that most of our images are the same for the same SoC, just with DTB/boot script differences for each particular layout, it's why on any board bringup you see us just manually swapping DTB's to see if it will work out of the box.

Is it possible to read the board name before it loads the dtb? Then you could protect it from loading the wrong dtb for it, and maybe choose the right one instead. I guess not or else it probably would be done like this.
Compatibility is a big factor that holds down the evolution in the software for these boards. That's why IBM/Intel x86 pc's made it and not the others(except ARM). While there were a lot better choices than that.
It's great ARM is so versitile, that's what kept it alive al those years. But when dealing with SBC's it could be done better.

But if it would be all too easy I wouldn't like it anymore. But then there'll be RISC-V to have the same issue's.

[TonyMac32]

9 minutes ago, NicoD said:

Is it possible to read the board name before it loads the dtb?

I'm afraid not.  A few have some sort of "marker" to show what they are (a resistor somewhere tied to an ADC, and efuse value, etc), but they are as non-standard as the SoC's themselves.

 

11 minutes ago, NicoD said:

But when dealing with SBC's it could be done better.

 

A lot of people feel this way, but it would require an agreement on a standard.  SPI flash or an eeprom drive cost some vendors will never sign up for, and create some problems when needing to upgrade the information stored on them (device tree bugs, bootloader fixes, etc)

13 minutes ago, NicoD said:

But if it would be all too easy I wouldn't like it anymore.

 

And projects like this one wouldn't have much to do either, the "big guys" would take over and you'd have the same basic options as PC, with others like us and Arch being "alternatives".  That said, there is a versatility of application to SBC's that does not exist with PC.  Few strap a PC to a bench with bare I/O monitoring things, that died with the parallel port.  So I think there will always be room.

 

14 minutes ago, NicoD said:

But then there'll be RISC-V to have the same issue's.

An FPGA that runs a RISC-V  powerful enough for linux isn't in my budget just yet, but it does look interesting.

[Myy]

It seems that @martinayotte was able to run the Nano PC T4 image with a 4.19.x kernel just fine, so you might be able to just start off with the Nano PC T4 DTS/DTB used in this image, and advance from there.

 

This week, as all the weeks in this year, has been pretty busy for me since I'm trying to advance some personal projects as best as I can. I'll try to fix the DTS file in the following weeks.

[balbes150]

On 12/30/2018 at 12:56 AM, NicoD said:

I tried downloading your LibreElec image for the Edge. Got this message 'Произошла ошибка'. I don't have an Edge, but still wanted to let you know.

What dtb was used ? Have you tried the "combined" startup option (write the same image to the SD card and USB flash drive, connect the SD and USB media at the same time, and turn on the power) ?