stroma

Hi team, This is not exactly Armbian related, however I hope to find some help here, also it can be useful for someone else in custom board development (my case). For testing I have Olimex RK3328-SOM-EVB and RK3318 TV box boards. They both behave the same way. As summary I'm trying to boot Linux using the Bootrom and USB OTG only, in other words without any external boot NVM attached. To archive this I build U-Boot SPL (u-boot-spl.bin), which support SDP protocol via USB OTG. For SDP file transfer I'm using imx_usb tool. The binaries used in the process are: rk3328_uboot_v1.19.250.bin - consists of u-boot-tpl.bin and u-boot-spl.bin created with boot_merger tool. u-boot.itb - FIT image (ATF, u-boot.bin and dtb), mainline U-Boot v2023.10 kernel.itb - FIT image (Image.gz and dtb), mainline kernel v6.1 I also prepared a proper sdcard image, from above binaries and wrote it to an SD card. My test setup consists of two booting paths, note the binaries in both are the same: With boot path 1, u-boot-tpl, u-boot-spl loaded via USB OTG: RK3328 in Mask ROM mode, SD card removed rkdeveloptool db rk3328_uboot_v1.19.250.bin insert SD card imx_usb - u-boot.itb boot kernel from U-boot prompt: setenv bootargs root=/dev/mmcblk1p2 console=ttyS2,1500000 earlycon ro rootwait; mmc dev 1; load mmc 1 0x800800 kernel.itb; bootm 0x800800 With boot path 2, u-boot-tpl, u-boot-spl loaded from SD card: insert SD card, power on (or reset) the board imx_usb - u-boot.itb boot kernel from U-boot prompt: setenv bootargs root=/dev/mmcblk1p2 console=ttyS2,1500000 earlycon ro rootwait; mmc dev 1; load mmc 1 0x800800 kernel.itb; bootm 0x800800 The overall description of two boot paths is as follow: boot path 1: rk3328_uboot_v1.19.250.bin (bootrom via OTG)-+ +- -> uboot.itb (imx_usb) -> kernel.itb (from SD card) boot path 2: rk3328_uboot_v1.19.250.bin (bootrom via SD) -+ The issues I have: When the SoC is booted through boot path 1 the kernel hangs soon after starting ... [ 0.016620] smp: Bringing up secondary CPUs ... If boot path 2 is used - The kernel boots and everything works as expected. The Bootrom in RK3318/28 first tries eMMC, SPI NOR, SPI NAND and SD card, if those boot devices failed then USB OTG is used for booting. Since the only difference between two boot paths is the way the TPL/SPL is loaded from the Bootrom - USB OTG or SD card, the only logical explanation I can think of is that there is some "leftover" in SoC state after OTG boot, which causes the kernel to hang. Does anyone have any idea how I can narrow down and eventually find the cause for this issue?

@Aurelio: Yes I used method described in the youtube video above. You will need serial cable connection to the PCB and adb. The 4 pin connector on the PCB is used for the serial connection. Cheap USB<->Serial converter can be used, note the voltage level of the signals is 3V3 (only GND, RX and TX pins are needed), serial speed is 115200 bps. I can't say much about the performance compared to RPi. Since the armbian filesystem is on USB 2.0 flash the file operations are rather slow. I successfully installed OctoPrint with web camera support and it seems to fork fine. I still not try to write the image to internal eMMC, I want to be able to backup the stock android and have a method to restore it if needed. For now I backed up the eMMC content with dd if=/dev/mmcblk1 iflag=fullblock oflag=direct bs=512k | gzip > tvbox-backup.img.gz but I'm not sure if this file is useful. Edit: armbianmonitor result

Yes, @Aurelio, it was proudly advertised as 4GB RAM, 64GB eMMC version. I know it was a scam before I buy it. To be honest I expected to get 1G/8G 🙂 However, good luck with the refund.

I managed to boot Armbian image from a usb flash drive. It seems I have 800MB RAM and 3GB eMMC 🙂 root@armbian:~# free -m total used free shared buff/cache available Mem: 803 163 127 4 512 559 Swap: 401 0 401 [ INFO ] The device name: [ Amlogic Meson GXL (S905X) P212 Development Board ] [ INFO ] The device eMMC name: [ /dev/mmcblk2 ] [ INFO ] The device eMMC size: [ 3GB ]

@schultzz The information you need is in first post on this thread, Unbrick paragraph.

Yes it is pretty crappy box but I got mine for 17$ so I could not resist to buy it. Mine is "4GB/64GB" version it seems they all are 1GB/8GB in reality. Here is couple of pictures of the PCB, the eMMC (BIWIN UM18A8G861S) and DDR (Lanson L5TC2G83EFR) chips are not well known at all.

Thanks @jock I've seen this paragraph but I thought this nvram was already included in never images, according to this post and also the tutorial from @awawa (link on the first page)

Hi I am not able to connect my H96Max 4GB RAM/64GB eMMC with RK3318_V1.4 blue PCB to a wifi network. The wifi module is marked as SP2734C. I'm trying to connect with nmtui, I can see available network SSIDs, but the connection attempt always fails. I tested wifi connection from stock android image and it works good, so the wifi module itself should be OK. pi@rk3318-box:~$ uname -a Linux rk3318-box 5.19.15-rockchip64 #trunk SMP PREEMPT Thu Oct 13 16:33:06 UTC 2022 aarch64 GNU/Linux pi@rk3318-box:~$ dmesg | grep brcm [ 19.505134] brcmfmac: F1 signature read @0x18000000=0x16034334 [ 19.559777] brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac4334-sdio for chip BCM4334/3 [ 19.561271] brcmfmac mmc1:0001:1: Direct firmware load for brcm/brcmfmac4334-sdio.rockchip,rk3318-box.bin failed with error -2 [ 19.585315] usbcore: registered new interface driver brcmfmac [ 19.761972] brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac4334-sdio for chip BCM4334/3 [ 19.762154] brcmfmac: brcmf_c_process_clm_blob: no clm_blob available (err=-2), device may have limited channels available [ 19.762745] brcmfmac: brcmf_c_preinit_dcmds: Firmware: BCM4334/3 wl0: Jul 24 2019 12:58:36 version 6.10.191 (TOB) (r) FWID 01-b049404e [ 20.213106] Bluetooth: hci0: BCM4334B0 'brcm/BCM4334B0.hcd' Patch Also tested with latest mainlined armbian image - same result. Is there something else I can try?

I Just found this youtube video describing hot to install armbian on Q96 Max TV box. It seems the guy forces Q96 to boot an armbian image from USB flash drive. Unfortunately it is in Korean and is not clear to me if the image can be actually written to the eMMC. Perhaps someone can help with the translation of the required steps.

@jock, just tested the experimental image with kernel 5.19.15 and libreelec patches and as you suggested the HDMI output is OK. It's a pity that mainlined armbian has this HDMI issue on my board. It seems the solution is to stay at older kernel revision. Thanks for the help.

Hello, First I want to thank to jock and other guys for the great work big effort to make this project possible. I have H96Max 4GB RAM/64GB eMMC with RK3318_V1.4 blue PCB inside, which seems to be well known here. I successfully booted several Armbian images from the SD card. All seems to work as expected and described in this thread. However when I test newer kernels I have weird problem with the HDMI signal with text being distorted (see attached images). With jocks Armbian_22.05.0-trunk_Rk3318-box_bullseye_current_5.15.35_minimal.img.xz and ...._5.15.16_minimal.img.xz the HDMI is OK and text is not distorted, like on the images below https://ibb.co/cFDyXTC https://ibb.co/vmRTqCX When jock Armbian_22.08.0-trunk_Rk3318-box_bullseye_edge_5.18.10_minimal.img.xz is used the text is weird as if some scan lines are sort of missing or skipped The problem also persists when I tested with nightly stables https://github.com/armbian/community/releases/download/202307/Armbian_23.02.0-trunk_Rk3318-box_sid_edge_6.1.11_minimal.img.xz, as shown on images bwlow https://ibb.co/dLJJL6f https://ibb.co/svB9n4g Please note, I tested HDMI with different monitors and TV, also the video from the stock Android image is OK. I've read the whole thread, and several other threads also but did not find similar issue with newer kernels (>=5.18). Can someone points me the direction where to dig? Thanks.