Posts by chewitt

    Code
    GXBB:BL1:08dafd:0a8993;FEAT:EDFD718C;POC:3;RCY:0;EMMC:0;READ:0;CHK:F3;SD:800;USB:8;LOOP:1

    ^ This is good, it means eMMC is fully erased (and the board is not bricked, it's simply missing software). You can now boot the board from an SD card .. as long as the SD card has a working u-boot in the correct places. The LE "hub" board image has everything needed although the catch22 will be that you're likely to hit the same problem with serial UART noise. Perhaps try writing the factory image backup written to SD and see if that's found (it might not be since it's an eMMC backup and S905 puts magic boot headers in different offsets for eMMC and SD) but even if it fails at booting into Android it might get you to a recovery console or perhaps the vendor u-boot console.

    If the AMLGX image works for you, that's great. If it doesn't work for you, don't expect anything to change. I've done zero work on Amlogic hardware since April and have no plan to change that until someone shows up with new hardware decode drivers. Comparisons with CE are irrelevant due to 100% of the media driver code being different.

    In my case once the UART pins were resoldered I don't see noise and it boots without getting stuck (the other user forced a pin to 3.3v which had a positive effect). So I didn't need to short eMMC pins on the Hub (but have done that on Play2 and some other boxes before).

    There are two more options I can think of:

    a) https://github.com/superna9999/li…DMI-Boot-Dongle <= but this requires hardware you don't have (and a little $) and will take a while to arrive if in-stock or you have to order parts from China to self-build.

    b) I forget the commands, but it's possible to enter the u-boot console, select the emmc device:partition and then erase a block range on the device to remove u-boot. Once that's done the boot-rom will fail to find u-boot on eMMC and boot the LE board image from an SD card (from where you can dd the factory image).

    RPi0/1/2/3 are 32-bit SoCs so all past/present/future releases are 32-bit kernel and 32-bit userspace. RPi4/400 are 64-bit SoCs and <= LE 11.x are 64-bit kernel with 32-bit userspace. LE 12.x will use 64-bit kernel and 64-bit userspace. The same is true for other ARM SoC vendors where we have historically and/or still do support a mix of 32-bit and 64-bit SoC hardware. Generic LibreELEC releases have always been 64-bit (OpenELEC dropped i386 support @ after v5.0).

    Over time the number of people confused/offended by our lack of publishing specific information on kernel vs. userspace arch is remarkably small, so I wouldn't over think the number of references that need to be made.

    Amlogic hardware is hardcoded (in silicon) to search for bootable firmware on eMMC, so unless eMMC is erased or electrically disabled to prevent u-boot from being found; the boot rom finds it and runs it. The toopick method applies to vendor u-boot, the (Android) recovery mode scripts are not implemented in upstream u-boot.

    I fixed this on my own board by resoldering dry joints on the UART connector. Another user shorted a pin. It's also possible to short eMMC pins to disable it allowing SD boot to the box image and then restore the factory image. I totally understand that not everyone would be comfortable with those options, but from a pure "is it recoverable or not?" perspective, it's recoverable.

    1) PT works here on HDMI but I didn't test S/PDIF in about 4 years (extracting the AVR to cable it is complicated) so that might be different.

    3) Read the big fat warning on AMLGX in public release notes

    4) Read the big fat warning on AMLGX in public release notes

    5) It's unusual but happens occasionally.. easily recovered from

    6) Already explained in the other thread (boot from SD card so emmc is not in-use and run "emmctool w /path/to/dd-backup.tar.gz" to write the dd backup of the factory image back to emmc).

    As written in the release notes, AMLGX will not be great for all users. If it works for you that's great. If it doesn't work for you (and it appears that way) then you should have low expectations of any major development work to improve it. I'd like to do more but my coding skills are limited and most of the long-known problems require the attentions of people who write kernel driver code.