Are you getting occasional .5 sec black screen pauses - that’s my main issue (mostly with mpeg2 or h264)? Or something else. Did kernel 5.10.21 improve things? I’m running bleeding edge on Mesa/kernel/media driver with the tiger lake. The Skylake is running fine (same bleeding edge)
Posts by heitbaum
-
-
Display More
Thx for the reply.
Regarding Tiger Lake, then how is the Intel NUC supported ny Libreelec -->
Intel NUC11TNKi3 Tiger Canyon - Core i3-1115G4 - Barebone | Billig
Intel NUC11TNKi5 Tiger Canyon - Core i5-1135G7 - Barebone | Billig
And if they both are fully supported and good for LibreElec, then what is best for 4K. The I3 or I5 ??
I am not a big fan on the PI hardware for mediacenter....
I have run a NUC6 i5 for many years. And now transitioning to a NuC11 i7 - it is quick, but there are some graphics teething issues with the new device, expect they will be remediate in the near future with intel investing updates in linux and Iris Xe. There is a slight difference on the Iris Xe chips across the i3/5/7 - the difference in the chips is the number of EUs (Execution Units) 48,80,96 respectively. I Note that the i3 has the UHD chip.
Some reading :
Intel announces its new 11th Gen Tiger Lake CPUs, available on laptops this fall - The Verge
-
Do you mean there is issues with normal video playback on 11 Pro NUCs?
I’m getting some black screens during play back, after idle (come back in the morning video off, needing reboot.) I have only been testing since Wednesday. So haven’t got a full detail yet. Some kernel errors. e.g.
[ 1.441261] ------------[ cut here ]------------
[ 1.441261] i915 0000:00:02.0: drm_WARN_ON(len <= 0)
[ 1.441266] WARNING: CPU: 0 PID: 14 at drivers/gpu/drm/i915/display/intel_dp.c:1918 intel_dp_max_link_rate+0x79/0x90
[ 1.441266] Modules linked in:
[ 1.441267] CPU: 0 PID: 14 Comm: kworker/0:1 Tainted: G U W 5.10.17 #1
But all in all - working fine. Just need to address these minors.
-
should it also work on nuc 11 Pro?
Hardware name: Intel(R) Client Systems NUC11PAHi7/NUC11PABi7, BIOS PATGL357.0037.2021.0106.1527 01/06/2021 is running LibreELEC LE10,
some video improvements will be required... but playback is mostly fine
-
In other news......
Ethernet and WiFi also not working in Libreelec on NUC11PAH. I am using a USB Wifi dongle which needs to be unplugged and plugged back in after a reboot.
update firmware for 5.10.16 by heitbaum · Pull Request #28 · LibreELEC/iwlwifi-firmware · GitHub
These will fix your Ethernet and wifi
-
The issue is more the logs and the why. We have identified why now - deinterlacing settings in the vaapi. Next steps will be to perform some tests and get to an end state.
See the bug an pull request.
-
Hi chewitt - dd’ed the vim3 2020-Dec-19 13:16 image to the SD card and booted it.
Not exactly sure on what is not working, after the boot kernel message, we get some orange on the screen then loose hdmi signal, and led light goes off.
From the rescue SD, was able to install a coreelec.kresq image
Also a little unsure on the required update to the u-boot to SPI/emmc from your repo. I’m running 2020.4 (November) on the unit at the moment. With “kbi bootmode w spi”.
A pointer would be great. -
First impressions:
LibreELEC build fine with only one packages removed on Generic
Specifically: nvidia-legacy (licence taint issue)
It runs Matrix 19b2
2 pulls required to get going.
Update RTL wifi drivers to 5.10 by CGarces · Pull Request #4691 · LibreELEC/LibreELEC.tv · GitHub
-
Checking with ffmpeg it definitely decodes, but I can’t tell if it is using the hwaccel? And why/what the profile it has mapped the .ISO (mpeg2video) to.
Codeffmpeg -vaapi_device /dev/dri/renderD128 -hwaccel vaapi -i a.iso -pix_fmt yuv420p a.yuv#6668 (MPEG2 decoding failed with FFMPEG-VAAPI) – FFmpeg
Based on the above post - moeg2video HW decoding was suppose to be fixed in Jan 2020 and in the following ffmpeg branches - master n4.4-dev n4.3.1 n4.3
-
I have been chasing the errors in my kodi.log for a while now. As you would know I think I have made all Intel Generic current or head in the build tree.
Code2020-12-06 21:48:28.754 T:5314 ERROR <general>: VAAPI output: Postproc preinit failed 2020-12-06 21:48:28.787 T:5314 ERROR <general>: VAAPI/vpp vaCreateConfig error: the requested VAProfile is not supported (12)This occurs when playing ISOs (DVD) or mpeg2video as the video stream in a MKV.
I have believe that the that Intel chip is suppose to suppose this decoding. As per the details at VAProfile is not supported (12) With MPEG2 playback · Issue #477 · intel/libva · GitHub VAProfileMPEG2Main is supported.
The media details itself are:
CodeStream #0:1[0x1e0]: Video: mpeg2video (Main), yuv420p(tv, bt470bg, progressive), 720x576 [SAR 16:15 DAR 4:3], 25 fps, 25 tbr, 90k tbn, 50 tbcThe full ffprobe output of the ISO is at - http://ix.io/2GQe
I added vadumpcaps to my build tree - vadumpcaps: add vadumpcaps to show all capabilities of a VAAPI device… · heitbaum/LibreELEC.tv@821ea50 · GitHub and from that it seems like Kodi/.... is incorrectly mapping the media type to 12 (which is JPEG.) - full output at http://ix.io/2GQl
So by my reading of vadumpcaps - VAProfile should be 1 for this media? Is this right / or am I chasing the wrong thing?
Code{ "profile": 1, "name": "MPEG2Main", "description": "MPEG-2 Main Profile", .... { "profile": 12, "name": "JPEGBaseline", "description": "JPEG", -
Took delivery of my N2+ Yesterday (Sunday) - DHL working all days at the moment....
after a bit of stuffing around and a pointer from tobetter. My notes to install Ubuntu 20.10 with 5.10rc and panfrost on the device - http://ix.io/2G19
In reading your chewitt logs in the mainline, it seems this N2+ is really only just now being supported in mainline without the ARM mali binaries.
Looking forward to running up LE on this.need to get back to coding.
-
I would like to suggest increasing the default log_buf_len to 128K or 256K (not the 16M that I have on my dev unit)
The later 5.x kernels boot logs are no longer complete using the default ring buffer size.
# grep log /flash/syslinux.cfg
APPEND boot=LABEL=System disk=LABEL=Storage log_buf_len=16M quiet
# dmesg | wc
961 8316 65437
# uname -a
Linux LibreELEC 5.9.11 #1 SMP Thu Nov 26 12:54:34 UTC 2020 x86_64 GNU/Linux
Happy to raise a PR either as syslinux or CONFIG_LOG_BUF_SHIFT if this makes sense.
-
-
thanks chewitt - Ill create a PR on le10 master - and have a look at paragon / 5.x too. Still busy getting my head around dbus.
-
Hi Team,
Suggesting we move to Version 2017.3.23AR.x as has been done in some mainline distributions.
E.g. Linked packages : ntfs-3g
Details are available at: NTFS-3G: Changelog for the Advanced Versions
Also noting some changes to behaviours of the caching in the driver (read the change log - NTFS-3G Safe Read/Write NTFS Driver Activity )
Specific patches are:
NTFS-3G Safe Read/Write NTFS Driver / ntfs-3g / Commit [39384c]
NTFS-3G Safe Read/Write NTFS Driver / ntfs-3g / Commit [5d6fd5]
If this is an suitable direction to move in, I’ll create a branch and put together a draft pull request.
-
Looks like a false alert. A USB device I had connected was causing the issues and random kernel crashes. Will continue to do further tests.
-
Different Kernel error back on - Linux LibreELEC 5.9.6 #1 SMP Wed Nov 11 06:51:19 UTC 2020 x86_64 GNU/Linux
ext4
-
Hi All,
As FYI only.(I'm rebuilding back to 5.9.6) against GitHub - heitbaum/LibreELEC.tv: Just enough OS for KODI and will advise. How it goes.
I just rebase lined GitHub - heitbaum/LibreELEC.tv at build_rudi + Kernel 5.9.7 against GitHub - heitbaum/LibreELEC.tv: Just enough OS for KODI
The compiled 5.9.7 with the rebase lined LibreELEC/Master "82a92d8d21762f2dc442b7794a6e02e3c5b74316" on the NUC6i5SYB is having kernel issues with: Workqueue: events azx_probe_work [snd_hda_intel]
Logs are at: http://ix.io/2DJH
Linux edbf6b369537 5.9.7 #1 SMP Tue Nov 10 13:49:34 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
