On an ARM SoC device the GPU is used for 2D/3D rendering (GUI) only and hardware acceleration (codecs) are separate things. As Panfrost supports the GPU hardware the GUI works, while decoding requires specific drivers that need adapting to silicon changes with each new chip. The latter are always non-trivial effort and the gene pool of people with the skillsets for the task is tiny; and most of them are commercial developers not the folks working pro-bono that everything normally depends upon. As such the process is slow .. and support probably doesn't exist yet. Over SSH use 'kodi-remote' to press 'o' on the keyboard to bring up the OSD that shows whether you are using (SW) or (HW) decode; but we already know the answer to that.
Posts by chewitt
-
-
It's 4K UHD H265 10-bit media that appears to have some kind of DV profile embedded. I can't see anything wrong with the media, but there are numerous things wrong with the current (unfinished since Q1/2020) H265 hardware decoder. At this stage if it works for you that's great. If it doesn't work, there's really nothing we can do, unless you have $250k spare to finance some commercial development to resume/resurrect codec development.
-
Device is not detected, I think due to Flirc new version.
The owner of flirc says flirc_util should support all their hardware. If your device is not detected you need to get in contact with flirc support and let them investigate.
-
there's a flirc_uril add-on in the LE add-on repo (install then reboot to use it)
The same flirc_util add-on is available for RPi3 and our Amlogic image (AMLGX) running on an N2 board. In both cases you install the add-on and reboot (until reboot ^ the binary is not in $PATH and will not be found) and then run "flirc_util" and the binary will output a bunch of info, for example:
CodeRPi5:~ # flirc_util flirc_util version 11d52588b5a4b3e5c88825888e18ef03ce8658ac [11d5258+] Firmware Detected FW Version: v4.6.3v4.6.3 SKU: Flirc 2.0 [dori] Branch: master Config: release Hash: 0x93A109D3
^ hardware detected.
I understand Fliirc command doesn't exist, only flirc_util, not easy to record IR codes. Am I right?
I'm using a macOS laptop and there's a simple macOS (and Windows) app for updating the firmware and configuring flirc dongles (easily downloaded from the flirc website) and for that reason I've never bothered or needed to use flirc_util and cannot speak for how easy or complicated it is to use.
If you are sat behind a computer with macOS or Windows and a USB port. Are you using the right tool for the job?
-
-
The AMLGX "box" image should work. Most devices with an RTL8189 WiFi chip seem to be S905 (no X) boards and the p200 or p201 device-tree files should work. If the device is S905X then start with p212 Read https://wiki.libreelec.tv/hardware/amlogic first as the instructions for AMLGX are different to other distros and older LE images that use the legacy (vendor) kernel.
The test images here: https://chewitt.libreelec.tv/testing/ have support for the RTL8189 wifi chip. LE release images do not.
-
The "flirc_util" add-on (complete with flirc logo) is in our repo under 'Program Add-ons' and testing it against your flirc hardware will take 2 minutes. If your flirc hardware shows an error or is flagged as unsupported by flirc software you need to contact flirc, because it's their software. All LE did was compile code from https://github.com/flirc/sdk and package it as an installable add-on.
I only have v2 flirc hardware and "it works for me" so
-
Is the drive powered from the USB port? or it has its own independent PSU? - and what is the HTPC device?
Share the system log after a successful mounting, e.g. run "journalctl | paste" and share the URL generated.
-
I assume you mean "flirc_util" and the flirc USB IR receiever device?
If yes, there's a flirc_uril add-on in the LE add-on repo (install then reboot to use it). If for some reason that doesn't work with the flirc USB hardware that you have, contact flirc support and get them to investigate and fix that, and then we can update the add-on.
There are also Windows/macOS/Linux (Debian based distros) installers downloadable from the flirc.tv website.
-
This shows how packaging could be done: https://github.com/Ran-Thegoth/uwe5622
Might be a newer version? https://github.com/SUISHUI/uwe5622_driver
This seems to have lots of patches: https://github.com/misuzu/uwe5622-nixos but also shows the driver has issues on recent kernels (would be normal for a downstream vendor driver).
Some Armbian fixes: https://github.com/armbian/build/pull/7615
However if you have no intention to work on improving the driver or even working your (not our) draft PR into better shape there's no point in leaving it open to rot on GitHub, so I've closed it.
-
Arokhaerr I'm not aware of any noteworthy changes to the SSH server between LE12 and LE13 but perhaps delete any stored profile or connection in the app and recreate in case there are cipher changes. Using a normal computer, check SSH in to prove the SSH service is enabled/running and that password auth is allowed. If yes, perhaps clear any stored connection/profile in the app and recreate. We don't change certs after initial boot but the accepted ciphers can/do change over time (albeit not often).
-
Feel free to test, improve, and merge.
It's the responsibility of the submitter to submit something that is mergeable, not the responsibility of the maintinainer to accept all changes submitted. Your goal is not aligned with our goals; I've commented on the pull-request.
-
Update with https://chewitt.libreelec.tv/testing/LibreE…_64-12.80.2.tar and check/test again. I've enabled some more kernel modules; I'm intentionally doing them in bits to see which ones make a difference (or not). Also share the URL generated by "pastekodi" (run from the SSH console) after updating so I can see the system log.
Information on IR remotes is here: https://wiki.libreelec.tv/configuration/ir-remotes but an RC6/MCE remote should basically work out of box. Worst case you might need to swap some buttons around via XML maps. If you have some other kind of remote you'll need to create a toml keymap (not hard to do).
I'd advise against Amlogic purchases due to no LE support on anything recent. RPi5 is nice. Mine just has some standoffs fitted to the board with plastic bits top/bottom to give some basic protection. It sits in a cupboard out of sight though so appearance and protection aren't important, and I use a BT remote so line-of-sight with an IR remote isn't required (CEC is disabled in the AVR as I have too many CEC capable devices connected and this only causes problems).
-
Merged. Should be in the repo in the next 24h.
-
Download the same file again. It's updated.
-
LE and RPiOS use the same kernels but RPiOS uses wpa_supplicant whereas LE uses iwd and the issue is proven to not occur when using wpa_supplicant. Reverting LE images to wpa_supplicant can be done but aren't going to entertain that idea as wpa_supplicant has its own set of very-legacy-code problems and the Linux community is keen to iwd mature and replace it. Signal strength is the most important factor as poor signal introduces all kinds of errors and iwd is probably not handling all of them as gracefully as it needs to; although a specific issue with WPA3 on Broadcom cards (which RPi boards use) was resolved and things have generally improved to the point where invalid-key is rarely reported now.
Cables are always best. External dongles or anything with a proper antenna beats the RPi on-board one.
-
Hauppague cards are generally supported well in recent-ish kernels. Cross fingers and roll dice
-
It was updating library DB files to the newer version (the image is LE13/K22).
I've enabled the CEC build option and pushed an updated image.