yeah. me too !
that's why I'm asking
I've tried also "git ignore" for that commit, but it also fails.
Nightlies are experimental development images so users should expect bugs and delays in fixing things. Our normal preference is to report the fix and then bump to include the resolution patch or another bump rather than revert back. If it becomes clear the issue won't be resolved quickly, then we'll revert the version.
What is the best apoach to compile the current nightly but ignoring the last samba update ?
Will patching packages/network/samba/package.mk (for the old 4.15.6 ) be enough ?
PROJECT=RPi DEVICE=RPi4 ARCH=arm scripts/clean samba
PROJECT=RPi DEVICE=RPi4 ARCH=arm scripts/clean heimdal
PROJECT=RPi DEVICE=RPi4 ARCH=arm make image
as the bug is identified and already known, shouldn't the samba version be reverted back to a working one ?
(in the libreelec nightlies), of course.
From the last nightly versions I've noticed it is not possible to browse the list of shares.
If i try direct access to a share like \\ipaddress_libreelec\storage from a windows 11 machine --> it works ok.
If i just try to list the shares \\ipaddress\ it says "windows cannot access ipaddress" !
I've removed the file /storage/.config/samba.conf and reboot, in order to force a default config, but it has the same behaviour.
In LibreELEC kodi interface sambe is configured with SMB1 as min protocol, and SMB3 as max.
I've just compiled and installed the current master branch from github.
I've been watching TV for the entire afternoon without any visible artefacts or any reported error in tvheadend.log
I'm not proficient enough to understand if there were changes in the code that may have produced this new behaviour (apart some changes in ffmpeg that were merged yesterday in the branch) or if it is just some weird coincidence.
I will keep in tests during the weekend and will mark the thread as solved next monday if that is the case.
Made new tests with 2 Raspberry Pi 4.
Both in the same lan network:
RPi4 #1 -> LibreELEC 9.2.8, dvb-C dvb-S receiver + tvheadend server + tvheadend player addon
RPi4 #2 -> LibreELEC 11 nightly build, tvheadend player addon, connected to RPi4 #1 (tvheadend server)
All goes OK !
The green screens artefacts are only appearing if using both tvheadend server + player in the same Rpi4 when running LE10/LE11.
Hope this helps somehow in solving the problem.
The sample file is corrupt (you can see complaints from VLC here).
Now how decoders cope with corrupt data is variable. In general software decoders have more ability to minimise visible artifacts (i.e. replace missing data with nearby data).
There is nothing that can be done about this, apart from not using the hardware decoder (which may have performance issues).
I don't believe LE 9.2.8 will play that sample file flawlessly.
The issue is that the stream gets corrupt when using a tv server and client on the same Pi + deinterlacing. This is totally RPi-specific.
Thank you both for your comments.
The strange thing here is that using the exact same hw configuration with le 9.2.8, plays flawlessly.
What have changed in le 10 that makes impossible to use tv server + client in the same machine ?
Having some issues while watching tv.
I've created a new thread.
Hope it helps :
Some additional information:
When watching tv with addon pvr.hts running in the same machine as tvheadend service:
- LE 11 nightly build - green artefacts; (*)
- LE 10.0.2 - green artefacts; (*)
- LE 9.2.8 - ALL OK (no issues).
(*) If I turn off "Allow Hardware acceleration with DRM PRIME" the artefacts disappear. But image loosed smoothness ...
When watching tv by remotely connecting from another machine (windows pc with kodi, or vlc, or android tv with kodi)
to libreelec tvheadend service:
- ALL OK !
So, the issue is common to LE10 and LE11 and ONLY when running pvr.hts in the same machine as tvheadend ...
and with "Allow Hardware acceleration with DRM PRIME" activated.
Some more tests and a sample:
- Green Artefacts just appearing if using kodi pvr.hts addon viewer and tvheadend server in the same machine (libreELEC 11 in RPi4).
- Artefacts appearing only after a Continuity Counter Error in tvheadend log;
- Green Artefacts and Continuity Counter Error DO NOT APPEAR if the viewer is running in another machine (example: VLC in WINDOWS or KODI running from Android tv !
The link bellow is for a sample file recorded in LibreELEC -> the green artefacts are displayed if played in LibreELEC 11 RPi4, but almost unnoticeable if playing in VLC player in windows:
Do you see the artifacts in recordings? If so, can you provide a short sample file that shows the issue?
hi, in recordings they do not appear as green artifacts. anyway I will try to get a sample - more thinking where to share it.
I had the same issue on a pi3. The issue was that when running both tvh client and server on the same pi AND using advanced deinterlacing method - green artefacts would appear from time to time, also continuity counter errors in tvh log.
The solution was to use BOB deinterlacing instead of "advanced". However, BOB looks much worse than advanced.
Another solution was to increase the arm_uc value, e.g.:
I don't have pi4 so no idea if this would work.
pi4 using prime for deinterlacing. No options.
changing arm_uc not producing any changes.
But thanks for the suggestions.
I've been using the latest nightly builds on a RPi4.
Since the introduction of deinterlace for RPi4, it was a big evolution for live tv watching (RE: LE10 & deinterlacing in RPi4)
I've installed tvheadend server in LE11 and using the tvheadened pvr add-on to watch live tv on my RPi4.
All ok except ... some green artefacts that appear from time to time in almost every channel.
Watching tv in KODI running on libreelec gives the green artefacts.
As my tv is a SONY Android tv, i've installed KODI 19 in it and connected to the tvheadend server on my RPi with LibreELEC- in this case the image is stable without any artefacts.
The same goes to the tv provider set-top which also plays flawlessly.
So, the difference here is KODI20 in LE11 on RPi4 - any suggestions on how to fiddle with LE11 KODI20 to handle those green artefacts ?
Hi had the same issue.
I've also compiled myself from git and had the same problems.
In my case kodi was complaining about a missing python lib. (I believe pyhton was updated to 3.9 a couple of days ago).
Similar to a problem I had reported in another thread,
I've SOLVED the issue by "make clean" and re-building everything.
Everything is stable now with the last source versions.
As I do not know how the nighlty builds are made,
I humbly suggest to the team to re-build them from scratch.
reverted the patch .. works.
placed the patch back and it also works !!!
if ..... "make clean" and recompile everything
So, after the changes for 5.15.17 / 5.15.18 I should have rebuild everything.
everything is working ok.
smp , thank you for your suggestion it sent me in the correct path.
I assume you are using rpi kernel 5.15.18.
Looking at the git commit history it looks like this commit was merged to kernel 5.15.17, so you may want to revert it for testing.
Correct. 5.15.18 it is.
I will then try swapping the initialization bits for si2157 , recompile and report back.
[FALSE PROBLEM ---- IGNORE THIS POST]
well ... drivers are no working anymore.
I've updated the patch file as the offsets changed (see attach) and it worked for builds compiled with version up to 2nd February 2022.
After that, tvheadend cannot lock signal.
dmesg reports errors:
Any help will be greatly appreciated.