Kodi Community Forum
v17 LibreELEC Testbuilds for x86_64 (Kodi 17.0) - Printable Version

+- Kodi Community Forum (https://forum.kodi.tv)
+-- Forum: Support (https://forum.kodi.tv/forumdisplay.php?fid=33)
+--- Forum: General Support (https://forum.kodi.tv/forumdisplay.php?fid=111)
+---- Forum: Linux (https://forum.kodi.tv/forumdisplay.php?fid=52)
+---- Thread: v17 LibreELEC Testbuilds for x86_64 (Kodi 17.0) (/showthread.php?tid=269815)



RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - Milhouse - 2016-08-05

The nfs:// segfault is also not a new issue - I added details on trac #16425 but it's been open since 15.2. However if you don't access an nfs:// share then you probably won't have a segfault on shutdown, which may go some way to explain the intermittent nature.


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - Milhouse - 2016-08-05

New LibreELEC.tv Krypton build #0805: Generic
(Supercedes previous build)

Code:
# uname -a
Linux NUC 4.7.0 #1 SMP Fri Aug 5 21:07:05 BST 2016 x86_64 GNU/Linux

# lsb_release
LibreELEC (Milhouse) - Version: devel-20160805210449-#0805-g7e85b6d [Build #0805]

Based on tip of LibreELEC.tv master (7e85b6d9, changelog) and tip of XBMC master (efcc4347, changelog) with the following modifications: Build Highlights:
  1. VideoPlayer fixes
Build Details:
  1. LibreELEC.tv:
    • RBP: Add GL symlinks to toolchain (PR:606, 1 commit, 1 file changed)
    • RBP: Use force when creating symbolic links in toolchain (PR:608, 1 commit, 1 file changed)
  2. XBMC:
    • VideoPlayer: fix state of StreamInfo (PR:10229, 3 commits, 7 files changed)
    • inputstream: fix return value of CheckAPIVersion (efcc4347)
  3. xbmc/master (FernetMenta):
    • New commits in this build:
      • VideoPlayer: remove some dead code (54f8c61a)
      • AE: change definition of AEStream::GetCacheTime (4f554e19)
      • VideoPlayer: fix early stall of audio stream, sync is now in AE (7d628566)
  4. Additional commits/pull requests/changes not yet merged upstream:
    • Added: [env] PR:609: init: Report out of space when unable to extract archive/compressed image



RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - C-Quel - 2016-08-05

(2016-08-05, 22:26)Milhouse Wrote: The nfs:// segfault is also not a new issue - I added details on trac #16425 but it's been open since 15.2. However if you don't access an nfs:// share then you probably won't have a segfault on shutdown, which may go some way to explain the intermittent nature.

ok switched tvheadend mounted folder to cifs and changed all shared from nfs:// to smb:// and confirmed no nfs:// shares reside in sources.xml located on my nas. Upon the first shutdown on my client it failed to complete and was forced to pull the power.

would the previous advice still be relevant to you for debugging? i'll pull a debug log anyway with the kernel changes you mentioned before as it seems it is less likely to be nfs:// related


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - dirtydesaster - 2016-08-05

(2016-08-05, 00:45)Milhouse Wrote: [*]VAAPI: Use VAAPI's BOB implementation by default (PR:10170)

Is VAAPI-BOB the best deinterlace method? I have a Skylake-CPU (G4400)


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - fritsch - 2016-08-05

No it's not ... use your eyes and find the best one :-)


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - Milhouse - 2016-08-05

(2016-08-05, 22:47)C-Quel Wrote: ok switched tvheadend mounted folder to cifs and changed all shared from nfs:// to smb:// and confirmed no nfs:// shares reside in sources.xml located on my nas. Upon the first shutdown on my client it failed to complete and was forced to pull the power.

Ok probably not the nfs:// segfault in that case.

(2016-08-05, 22:47)C-Quel Wrote: would the previous advice still be relevant to you for debugging? i'll pull a debug log anyway with the kernel changes you mentioned before as it seems it is less likely to be nfs:// related

Yes. Add the debugging option to your kernel command line, reboot a couple of times, then run "journalctl -a | pastebinit" after you have a hang on shutdown.


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - gjwAudio - 2016-08-06

Hi Milhouse

I'm now on the VPN STOPPED WORKING bandwagon too...

(2016-07-31, 06:28)Milhouse Wrote: "route add" is a Busybox command. The Busybox package is bumped in build #0721. Start testing builds #0720 and #0721, as it's possible the updated Busybox package has introduced a change of "route add" behaviour. However if both of these builds are working OK then you'll need to work through the more recent builds until you identify the first non-working build. Obviously if both builds are not working, you need to work back through the older builds until you find the first working build.

I marched through the builds and - as you predicted - VPN MGR works fine with #0720 and fails with #0721 and above. Just for extra work, I tried with the current #0804 and same deal - not connecting to the remote VPN server.

#0720 debug log | #0720 openvpn log

#0804 debug log | #0804 openvpn log

I hope some way can be found to get this working again Angel

TIA


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - artur9 - 2016-08-06

(2016-08-05, 22:08)C-Quel Wrote: ok ill switch to smb for a few days to rule out nfs:// connections... if successful then i will report back! you may be right as i hold a shared db on my nas which is accessed via nfs:// maybe its not closing this correctly on shutdown

I actually mount my nfs disks directly using the startup scripts. Never had a problem.

SMB restricts too many characters that are useful in naming movies.


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - C-Quel - 2016-08-06

(2016-08-06, 01:39)artur9 Wrote:
(2016-08-05, 22:08)C-Quel Wrote: ok ill switch to smb for a few days to rule out nfs:// connections... if successful then i will report back! you may be right as i hold a shared db on my nas which is accessed via nfs:// maybe its not closing this correctly on shutdown

I actually mount my nfs disks directly using the startup scripts. Never had a problem.

SMB restricts too many characters that are useful in naming movies.

been using nfs with kodi for about... 10 years? never had an issue i still think its this fancy bi323 box and not actually kodi related! will find out tomorrow


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - Milhouse - 2016-08-06

(2016-08-06, 00:15)gjwAudio Wrote: Hi Milhouse

I'm now on the VPN STOPPED WORKING bandwagon too...

Thanks, it really helps to have the first non-working build confirmed.

The problem is this change in Busybox 1.25.0. I've opened a bug for it, #9141.

I'll revert the change from Busybox 1.25.0 in future builds until there's an upstream solution.


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - C-Quel - 2016-08-06

Apologize for the false alarm... simple bios update and clearly stated in the changelog that resuming from S3 or shutdown was a problem from production release! Sad thank you for your time though Milhouse (school boy error) Sad


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - ultraman - 2016-08-06

gjwAudio, can you try adding parameter
Code:
scope 0
at the end of route add command?

Update: Watercoolertje, same for you.


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - Watercoolertje - 2016-08-06

(2016-07-31, 04:32)sciascia88 Wrote: @Milhouse
Here's my Debug Log from the current build. The VPN manager acts as if it connects to the New York server, but actually does not. It works fine on the Alpha build (7.90.003).

Hope that helps.

I have exactly the same problem, but with a Raspberry PI 3. The problem started with the update to busybox 1.25 ( build build #0721 for RPI1/RPI2 ). With this change
the command 'ip route add xxx.xxx.xxx.xxx/yy via xxx.xxx.xxx.xxx' no longer works and leads to several errors in /run/openvpn.log

Going back to a release with busybox 1.24 fixes the problem.


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - Watercoolertje - 2016-08-06

(2016-08-06, 09:20)ultraman Wrote: gjwAudio, can you try adding parameter
Code:
scope 0
at the end of route add command?

Update: Watercoolertje, same for you.

adding 'scope 0' at the end of the route command accepts the command and no error occurs, after this openvpn seems to
work again. Smile


RE: LibreELEC Testbuilds for x86_64 (Kodi 17.0) - Eureka - 2016-08-06

Okay, I'll ask... What differences am I supposed to see on Intel machines on builds after 7/30? I tried one on a 3205u and nothing jumped out at me.