Posts by blackride

Quote from chewitt

NB: Until the audio channel mapping issues are figured out users can select the "HDMI" device as a pass-through device but should still select the "Analogue" output device as the main Kodi output. The Analogue PCM device alsa conf has some channel remapping to workaround the problem - the HDMI device does not support this.

I checked your latest image for Odroid C2. And it loads with the default file dtb.

There was also a sound when playing a video.

Found problem - HEVC files have no picture.

Thanks for your work chewitt.

Quote from sailorpito

I using Samsung EVO sdcard

It doesn't matter to me

Quote from chewitt

Minor progress on a couple of things: One of the issues with the audio driver used with GXBB (S905) boards was figured out so there is no longer any need to force output rates to 41KHz to output 48KHz media. See: https://patchwork.kernel.org/p…[email protected]/. This also leads to a major simplification for alsa configuration which we will send upstream. Sadly we still didn't figure out *that* static noise problem for GXBB/GXL/GXM devices yet, and this is what I regard as the minimum fix before some nightly images can resume. In the background I also have the full vendor sources and schematics for Minix Neo U1 (S905) and U9-H (S912) devices. I plan to extract and post the boot fips so that mainline u-boot support is possible. I can also validate the device-trees which have previously been created with user support (U9-H is upstream, U1 still wip).

Hi, chewitt.

Can we hope to improve the LE 10 for the odroid C2?

Quote from chewitt

See odroid-c2 diffs · GitHub for a comparative diff of the working (left) and current (right) dtb files. This flags only two changes in the dtb, neither of which has any connection to emmc.

Change #1: arm64: dts: amlogic: misc DT schema fixups · torvalds/[email protected] · GitHub

Change #2: arm64: dts: meson: set 128bytes FIFO size on uart A · torvalds/[email protected] · GitHub

The only other thing which touched mmc (on GX devices) semi-recently is arm64: dts: amlogic: Assign a fixed index to mmc devices · torvalds/[email protected] · GitHub but we mount the first partition using disk labels not /dev/mmcblkX devices and it's present in both dtb's so it's not relevant.

So

1. USB is an on/off problem with C2 in the upstream kernel so it's possible OTG isn't working this kernel cycle. I'd expect to find issues with devices being hotplugged.

2. I don't see the USB ID's in the kernel driver, which is a simple patch you can test next time I build/push an image update.

3. No HDMI device is deliberate since we don't support IEC958 modes in dw-hdmi yet (no pass-through) but the "Analogue" output works with HDMI fine (output is multi-channel PCM). I'm still tracking down an issue where 48KHz media doesn't output unless Kodi audio configuration is forced to 44.1KHz (used fixed mode). I've seen this on WP2 (also a GXBB device) but not GXL/GXM devices. I've no idea what the issue is.

Display More

You don’t want to say that I am to blame in this case.

The fact remains that your dtb file does not work on my copy of the odroid c2 (probably the hardkernel is blame).

Improvements, as I understand it, will not be in LE10 for the odroid C2. The fate of the odroid С2 is the use of volumio.

It is sad.

Quote from chewitt

...

Chewitt, I have loaded your LE10 build via the microsd card.

As I said, the dtb file is to blame. On my 4GB microsd card, I found a recorded official nightly build (March 2021). From there I took the dtb file and replaced your dtb file. Attached the file dtb - meson-gxbb-odroidc2.zip

Now about the bad news:

1. usb-otg does not work (it is possible that turned off in compilling the build).

2. The adapter TP-Link TL-WN821N (HW V5 -> 2357:0107 (rtl8192eu)) does not work.

3. Sound does not work when playing a video (TV or video file). Only interface sounds work. The choice of a sound device is only analog and bluetooth. There is no HDMI.

Please tell me, what is needed to solve these 3 problems?

Quote from dtech

I'm sorry I couldn't really help you.

The truth is, I’m also trying to provide support for devices I’ve never had at all. And since I can't test these myself that way, and I can't debug any errors myself on them, so it's really just a remote series of experiments. And if something failed, it’s actually just a waste of time for me too. Unfortunately, I never know if actually I made a mistake or the tester messed up something. That’s when I usually come with this sentence: "Are you sure you did exactly that...?" Although I may have actually screwed up something.

Unfortunately, I start to get a little tired of these.

I apologize. Please tell me, do you have any ideas for rtl8723?

Thank you

Quote from chewitt

Code

[    3.682808] mmc0: tuning execution failed: -5
[    3.682820] mmc0: ddr50 tuning failed
[    3.682838] mmc0: new ultra high speed DDR50 SDHC card at address 0001
[    3.684028] mmcblk0: mmc0:0001 SD32G 29.1 GiB
[    6.330769] mmc0: tuning execution failed: -5
[    6.330793] mmc0: ddr50 tuning failed
[    6.332062] blk_update_request: I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
[    6.332124] Buffer I/O error on dev mmcblk0, logical block 0, async page read
[    6.335152] blk_update_request: I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
[    6.335165] Buffer I/O error on dev mmcblk0, logical block 0, async page read
[    6.335197] ldm_validate_partition_table(): Disk read failed.
[    6.337727] blk_update_request: I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
[    6.337738] Buffer I/O error on dev mmcblk0, logical block 0, async page read
[    6.340279] blk_update_request: I/O error, dev mmcblk0, sector 0 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
[    6.340290] Buffer I/O error on dev mmcblk0, logical block 0, async page read
[    6.340321]  mmcblk0: unable to read partition table
[    6.344660] blk_update_request: I/O error, dev mmcblk0, sector 61069184 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 0
[    6.347382] blk_update_request: I/O error, dev mmcblk0, sector 61069184 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
[    6.347397] Buffer I/O error on dev mmcblk0, logical block 7633648, async page read

Display More

^ To me it looks like a bad SD card, or perhaps a power issue making the board unstable, but that's unlikely if the older CE image runs fine.

Chewitt, maybe this dmesg help? I wrote down the official LE 9.0.2 on the same 32GB Kingston Canvas microsd card.

The problem with the power was earlier - at the first cold boot of the LE / CE, always froze the Odroid С2. I had to disconnect and reconnect power. And I could not figure out what was the matter until I soldered the power wires directly to the board (alternative power). The connector dc, the inner diameter was wider than needed and the power was less because of this.

I have no problem with alternative power. However, even with poor nutrition, was a launched the old LE10 builds from the balbes150 (True, now I do not have these builds LE10, the author deleted them on his Yandex disk so that I can take dtb file).

I am using two microsd cards. The very first one that I bought in a local store is the original Sandisk 64GB class 10 XC microsd card. I never had any problems with her (I started with LE7). Another card was also bought from a local store, the original Kingston Canvas 32GB class 10 A1. Also, i bought for a long time the "original" Samsung Evo microsd card for aliexpress and it turned out to be a fake, although it worked in the C2 odroid. Therefore, I no longer buy microsd cards there. Only in local good shops.

So, Sandisk and Kingston show the same picture with the latest LE10 assemblies.

In may, I recorded the night builds of LE10 for Odroid N2 on the sandisk microsd card and everything worked.

So the microsd cards are not to blame - I think so.

Quote from chewitt

C2 has issues with USB generally (even in the old vendor kernels) so YMMV with USB booting.

As promised - your build LE10 was launched on a USB flash (please, see my photo):

I'm sure your LE10 build would most likely be more responsive on a microsd flash drive. But I cannot boot from the microsd stick to check this.

Is it possible to compile the LE10 build yourself with some change in the dtb file, so that my odroid, which differs from yours, can be launched from a microsd flash drive?

Or make some corrections in the form of a patch in the dtb file on your part?

Perhaps you need some information?

I can show the output of commands from your LE10 build running on a USB flash drive.

Just write what commands to enter.

For example, dmesg:

Quote from gst:

blackride

2. I will not do that. I only had such a Error with a corrupt download or a broken micro sd card reader or sd card. My test sd card have 32Gb and emmc 64GB.
You can try Coreelec CoreELEC - Legacy Builds 9.2.7 (Stable) selekt device... and for the libwidevine.so fix update with [ARM] Add suport for libwidevine.so >= 4.10.2252.0 by Portisch · Pull Request #287 · CoreELEC/CoreELEC · GitHub

Good luck

What does the CE update have to do with it?

CE 9.2.7 on Odroid C2 works without problems. And in my odroid there are no hardware problems with the card reader. I tested builds LE10 from balbes150 - they worked.

But for some reason, chewitt builds refuse to start.

Tomorrow I will try to run the LE10 build chewitt with usb flash drive. Then it will be clear that the dtb file is designed only for launching from a USB flash drive.

To summarize, I do all the same actions that you do. Writing the image to a microsd flash and starting the download. Only now LE9 starts up, and LE10 does not see the bootloader.

Quote from chewitt

In the old forum thread you found the user has a generic S912 image where you must configure the correct dtb file for the box to boot. In the LE10 C2 image I shared everything is correctly preconfigured so the dtb does not need "fixing" to work.

See if "boot=/dev/mmcblk0p1 disk=/dev/mmcblk0p2" work (edit the exlinux.conf). If not, I have no idea what the problem is.

I changed file extlinux.conf:

LABEL LibreELEC
  LINUX /KERNEL
  FDT /meson-gxbb-odroidc2.dtb
  APPEND boot=/dev/mmcblk0p1 disk=/dev/mmcblk0p2 quiet systemd.debug_shell=ttyAML0 console=ttyAML0,115200n8 console=tty0

And get the same error where "/dev/mmcblk0p1" is written instead of "LABEL=LIBREELEC".

I can say with certainty that older the official nightly builds of LE10 worked on my C2 (the only thing that worked slowly was the KODI interface). And then in the last nightly official builds of LE10 I got this error. This error is apparently due to the fact that support has been added for booting from a USB flash drive. I have not tried to write to a USB flash drive your builds, but I tried to write a USB flash drive on the latest official night builds of LE10 - if a microsd card is inserted, which also contains the LE10 build and a flash drive, then Odroid C2 is loaded from the USB flash drive. I can try it on your build and I'm sure it will work. But I want to use only a microsd card.

I have another idea. But I think that it will not give anything. Install the official LE9 on the microsd card, run it to create a full size STORAGE. And then replace on it all the contents (files) of the LIBREELEC section from your LE10 build. Don't work, when you turn on the odroid C2, the blue LED is constantly on and nothing else.

Is it possible from your side to write step-by-step commands to start loading LE10 using the connected keyboard to the Odroid C2, when such an error appears?

Quote from gst:

chewitt Thank you for your work. Your images work on my Odroid C2 with SD card and eMMC made with Etcher on Ubuntu 18.04 . The only thing that doesn't work is to wake up via CEC. Maybe you can still incorporate that if it is possible.

en:c2_ir_wake_up [ODROID Wiki]

I tried to write the microsd card via Creator LibreELEC USB-SD, balenaEtcher in windows 10. And even started opensuse in live mode and writing the microsd card via dd (I previously unpacked img from img.gz). And I still get the same error.

Could you do 2 things for me ?:

1. View the revision of the release on the board odroid c2.

2. Via Gparted, reduce section STORAGE on your microsd card (if you have a large microsd card) and make a full dump microsd card via dd and upload img dump to some file hosting (yandex, google disk or another).

Quote from chewitt

blackride The bootloader is unable to find a partition with the label "LIBREELEC" to load the SYSTEM file. It has nothing to do with the device-tree being used. The LIBREELEC label is used with the FAT16/VFAT partition on the .img file we use for imaging SD cards and eMMC modules for the C2 board and the configuation is defined in extlinux.conf where we set boot=LABEL=LIBREELEC and disk=LABEL=STORAGE. If you have swapped between Olegs images (which used non-standard u-boot bootscripts and locations/naming sometimes) and mine or LE10 nightlies the extlinux.conf might be wrong (see if UUID or /dev/device works instead of LABEL). Sometimes sudden power-off without shutdown will result in card corruption; normally fixable with fsck. It's hard to comment without knowing the history of the device, e.g. what was originally installed, from where, and when, and what update was performed.

NB: I've run boot tests with LibreELEC-AMLGX.arm-10.0.0-odroid-c2.img.gz on SD card and an eMMC module and (from a boot/install/run perspectice) the image works fine for me.

No, you did not understand me. I recorded your image on the microsd card, that is, I did not do an update, but a clean installation.

extlinux.conf file on microsd:

LABEL LibreELEC
  LINUX /KERNEL
  FDT /meson-gxbb-odroidc2.dtb
  APPEND boot=LABEL=LIBREELEC disk=LABEL=STORAGE quiet systemd.debug_shell=ttyAML0 console=ttyAML0,115200n8 console=tty0

There is a solution here using the correct dtb file. You also wrote that you can connect a keyboard and via vi fix the bootloader, but you did not write how to do it.

I tried different microsd cards and no result.

Can I fix the dtb file?

Quote from blackride

Hi chewitt.

Thank you for everything you do for us.

Is it possible with your help to solve the following error (photo)?

This is error appeared on builds of LE 10. And also I tried your last build - LibreELEC-AMLGX.arm-10.0.0-odroid-c2.img.gz.

As I understand it, it has to do with the device tree.

Display More

chewitt doesn't answer.

Unfortunately, I have not saved the old builds from the balbes150. So it would be possible to pull out the dtb file from his build.

Who knows how to fix a dtb file?

Quote from dtech

Okay, maybe we have one more option, so please send the output from these commands under the old 9.2.8.1:

Code

cat /sys/class/net/wlan0/device/vendor
cat /sys/class/net/wlan0/device/device

I need to know the vendor ID and the product ID of the integrated wireless/bluetooth SDIO module.

LibreELEC:~ # cat /sys/class/net/wlan0/device/vendor
0x024c

LibreELEC:~ # cat /sys/class/net/wlan0/device/device
0xb723

Quote from dtech

blackride

Now the 8723bs module will have the problem.

I need some information about the integrated adapter from an older version that was still working.

lshw -C network

I apologize for such a late answer.

LibreELEC (dtech): 9.2.8.1 (MXIII-Plus.arm)
LibreELECa:~ # lshw -C network
  *-network:0
       description: Wireless interface
       physical id: 3
       logical name: wlan0
       serial: 38:1d:d9:f8:26:c5
       capabilities: ethernet physical wireless
       configuration: broadcast=yes driver=rtl8723bs driverversion=3.10.108 firmware=N/A ip=192.168.1.94 link=yes multicast=yes wireless=IEEE 802.11gn
  *-network:1
       description: Ethernet interface
       physical id: 4
       logical name: eth0
       serial: 00:01:23:cd:ee:af
       size: 10Mbit/s
       capacity: 100Mbit/s
       capabilities: ethernet physical tp mii 10bt 10bt-fd 100bt 100bt-fd autonegotiation
       configuration: autonegotiation=on broadcast=yes driver=st_gmac driverversion=March_2013 duplex=half link=no multicast=yes port=MII speed=10Mbit/s

And last build 9.2.8.2 (also bluetooth don't work):

LibreELEC (dtech): 9.2.8.2 (MXIII-Plus.arm)
LibreELECa:~ # lshw -C network
  *-network:0
       description: Wireless interface
       physical id: 3
       bus info: [email protected]:1.3
       logical name: wlan0
       serial: d0:37:45:ca:df:66
       capabilities: ethernet physical wireless
       configuration: broadcast=yes driver=rtl8192eu driverversion=3.10.108 firmware=N/A ip=192.168.1.143 link=yes multicast=yes wireless=IEEE 802.11gn
  *-network:1
       description: Ethernet interface
       physical id: 4
       logical name: eth0
       serial: 00:01:23:cd:ee:af
       size: 10Mbit/s
       capacity: 100Mbit/s
       capabilities: ethernet physical tp mii 10bt 10bt-fd 100bt 100bt-fd autonegotiation
       configuration: autonegotiation=on broadcast=yes driver=st_gmac driverversion=March_2013 duplex=half link=no multicast=yes port=MII speed=10Mbit/s

Quote from dtech

I need a dmesg immediately after the boot.

hastebin

LibreELECa:~ # lsmod
Module                  Size  Used by
8021q                  16085  0
8192eu                996527  0
cfg80211              205322  1 8192eu
wifi_dummy               737  0
wireguard             119358  0
ip_tunnel              11161  1 wireguard
mali                  163808  5
amlvideodri             9161  0
videobuf_res            5077  1 amlvideodri
videobuf_core          15610  2 amlvideodri,videobuf_res
videodev               84856  1 amlvideodri

Quote from chewitt

dtech see if you can backport systemd/busybox: allow configuration of persistent logs and journal by mglae · Pull Request #5537 · LibreELEC/LibreELEC.tv · GitHub and then use /storage/.cache/journald.conf.d/00_settings.conf with something like:

[Journal]

SystemMaxUse=10M

MaxRetentionSec=0

RateLimitInterval=0

RateLimitBurst=0

The settings add-on changes are in Add persistent logs and journal configuration by mglae · Pull Request #233 · LibreELEC/service.libreelec.settings · GitHub but we rewrote/restructured most of the settings add-on since 9.2 so those won't backport.

Display More

Hi chewitt.

Are you can me help please? This is probably related to dtb (post RE: Official LE Test Images for Amlogic (Kodi-19))

Quote from dtech

One step forward, then one step back. This may solve the problem:

Index of /snapshots/20210827/

Bluetooth adapter not found

Quote from dtech

Вышел новый выпуск снимка (20210826):

Индекс / snapshots / 20210826 /

Тот же драйвер, но я отключил встроенные модули в ядре, потому что они были включены.

Пожалуйста, протестируйте его сначала с V5, а затем с V4.

V5 - hastebin

V4 - hastebin

V4 and V5 work.

Version 5 does not light up the green LED.

Version 4 LED is on.

The built-in bluetooth box don't work.

Now the built-in wi-fi boxing module will always be disabled or can you try to solve the modules loading conflict?

Thank you very much