• 1
  • 107
  • 108
  • 109(current)
  • 110
  • 111
  • 168
OpenELEC Testbuilds for RaspberryPi (Kodi 16.0)
(2015-10-13, 12:12)parcel Wrote: I found solution for mystery bookmark crash problem.

There is a lot of "10025-xxxx.fi" files and then I delete them, no more crash.
I don't know what is this file, but first number of filename is window ID.

Delete all ".fi" files in folder '.kodi/temp', it solves bookmark crash problem.

Looks like they are used by FileItem.cpp. Probably from here.
So something cached from window ID 10025.

I'm wondering if the format of what is cached changed between nightly builds, and using a cached item from an older build can cause the crash.
Might explain why some (like me) never saw the crash.

Should always be safe to delete files from ~/.kodi/temp so it's probably a good suggestion for anyone with random crashes.
New OpenELEC Jarvis build #1013: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.1.10 #1 Tue Oct 13 21:03:43 BST 2015 armv6l GNU/Linux

# vcgencmd version
Oct 11 2015 16:21:59
Copyright (c) 2012 Broadcom
version cdea693c2a4280e4855643dc7840392d915f19b8 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20151013210211-#1013-ge5e6606 [Build #1013]

# vcdbg log msg 2>&1 | grep DTOK
001717.731: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (e5e6606a, changelog) and tip of XBMC master (103d2d5e, changelog) with the following modifications: Build Highlights:
  1. Bump curl, glib, freetype, busybox (clean build)
  2. Disable libaacs for all platforms
Build Details:
  1. OpenELEC:
    • libbluray: depends on fontconfig (4c6a7064)
    • libbluray: explicitly enable optional enabled features (651dcf3a)
    • ccache: update to ccache-3.2.3 (7a5c505b)
    • wireless-regdb: update to wireless-regdb-2015.09.25 (961e7a71)
    • curl: update to curl-7.45.0 (615a7df9)
    • glib: update to glib-2.46.0 (b0059be8)
    • freetype: update to freetype-2.6.1 (467ac9f5)
    • busybox: update to busybox-1.24.0 (d5e9be86)
    • kodi: move non-X11 mousewheel support patch (651fa5f3)
    • kodi: add multitouch support patch, borrowed by popcornmix (3bc0e8d2)
  2. XBMC:
    • Clear the cached source listing when removing a video/music source (PR:8202, 2 commits, 2 files changed)
  3. pvr.filmon:
    • [cmake] Added missing CPack include. Closes #32 (PR:33, 1 commit, 1 file changed)
  4. newclock5:
    • New commits in this build:
    • Commits no longer in build:
      • ADD: [demux] support ssif (c4f73377)
      • [mvc] Enable 3d attribute of mvc video automatically (8cc44f83)
      • Revert "ADD: [demux] support ssif" (df89360b)
      • Revert "[mvc] Enable 3d attribute of mvc video automatically" (620898d3)
      • fixup: free unused packets (27e3a3e3)
      • squash: Remove extra demuxer packet free (ed83e7bf)
  5. Additional commits/pull requests/changes not yet merged upstream:
Texture Cache Maintenance Utility: Preload your texture cache for optimal UI performance. Remotely manage media libraries. Purge unused artwork to free up space. Find missing media. Configurable QA check to highlight metadata issues. Aid in diagnosis of library and cache related problems.
popcornmix:

segfault on #1013 on a specific video, others seem to play.
http://sprunge.us/fabe

Im on freenode if a sample would help.
@miigotu: Can you enable debug log (wiki) and repeat the crash, uploading a new crash log. Also, is it possible to upload a sample of this video (or the entire video)? PM a link to myself or popcornmix if necessary.
Texture Cache Maintenance Utility: Preload your texture cache for optimal UI performance. Remotely manage media libraries. Purge unused artwork to free up space. Find missing media. Configurable QA check to highlight metadata issues. Aid in diagnosis of library and cache related problems.
(2015-10-13, 20:21)popcornmix Wrote: I've been playing with setting GUI resolution to [email protected] and playing MVC files. My display seems to switch back correctly to non-3D mode when stopping.
I've stepped through the firmware code, and while it can skip the mode change if it believes nothing has changed, it does check the 3D mode, so should spot switches into and out of 3D mode.

I'm interested in the results of "disconnecting and reconnecting the HDMI cable". Ideally you should add "hdmi_force_hotplug=1" to config.txt to ensure the Pi doesn't notice the unplug/plug events.

Can you also report output of:
Code:
vcgencmd measure_clock pixel

when stopping (and gui being wrong). It should be about 74MHz when in 2D mode and 148MHz when in 3D FP mode.

Disconnecting and then reconnecting the HDMI cable = functions normally (so GUI is NOT displayed in frame-packing garble).

Output = 148352000 in 3d FP mode. 74176000 in the GUI which is stuck in frame packing
(2015-10-14, 12:12)woronczak Wrote: Disconnecting and then reconnecting the HDMI cable = functions normally (so GUI is NOT displayed in frame-packing garble).

Output = 148352000 in 3d FP mode. 74176000 in the GUI which is stuck in frame packing

Okay, that suggest when you are seeing corrupt GUI we are outputting a valid 2D signal, but the display has not reacted to the change.

What is make and model of display. Is there any way of updating firmware?
If you have a receiver connected, is it any different if that is bypassed?
Do you only get this for frame-packing, or do you also get this with "Use Full HD HDMI modes for 3D" disabled (i.e. switching from 3D SBS to 2D).
Went to use the RPi just now and found the CEC remote not working.

I found there was a crashlog and kodi.log created at 05:21, which would have been shortly after I stopped using it and went to bed, so it appears that it rebooted itself around then and as the TV was not on the CEC did not initialise, so when I turned on the TV later the remote was non-functional until I forced a reboot from my PC.

I've zipped and uploaded kodi.log, kodi.old.log and kodi_crashlog here (only 1.3MB) if you want to take a look to see if there's anything that can be done to fix the cause of the crash and/or the CEC problem: https://drive.google.com/file/d/0B1fDI89...sp=sharing
(2015-10-13, 13:18)hdmkv Wrote: For me, it's not just 3D (frame-packed mode and coming out of it) that had the issue, but 2D as well with UI set to 59.94 or 60.00.

Can you clarify this. Are you saying when you are in gui (in 2D mode), and you play a 2D video file it switches from say, 60Hz to 23.97Hz correctly and you get an image.
But when you stop it doesn't switch back to 60Hz gui correctly? What happens, blank screen?
(2015-10-14, 12:52)doveman2 Wrote: Went to use the RPi just now and found the CEC remote not working.

I found there was a crashlog and kodi.log created at 05:21, which would have been shortly after I stopped using it and went to bed, so it appears that it rebooted itself around then and as the TV was not on the CEC did not initialise, so when I turned on the TV later the remote was non-functional until I forced a reboot from my PC.

I've zipped and uploaded kodi.log, kodi.old.log and kodi_crashlog here (only 1.3MB) if you want to take a look to see if there's anything that can be done to fix the cause of the crash and/or the CEC problem: https://drive.google.com/file/d/0B1fDI89...sp=sharing

It seems to have crashed in libnfs.so. The only thing I can suggest is trying the latest debug-enabled build #1011x and see if you can obtain a more detailed crashlog.

You do seem to have a lot of NFS errors in your log, culminating in:
Code:
05:19:39 87832.781250 T:1515189248   DEBUG: NFS: Refreshing context for 192.168.1.64/Media, old: 87805401, new: 87827960
05:19:39 87832.781250 T:1515189248   DEBUG: NFS: Using cached context.
05:19:39 87832.781250 T:1515189248   ERROR: NFS: Failed to stat(Media/Videos/Series/The Blacklist/S2/../logo.png) nfs_service failed
a couple of minutes before the crash. which may be part of the problem.
Texture Cache Maintenance Utility: Preload your texture cache for optimal UI performance. Remotely manage media libraries. Purge unused artwork to free up space. Find missing media. Configurable QA check to highlight metadata issues. Aid in diagnosis of library and cache related problems.
Milhouse can you please build later tonight tvheadend server addon for rpi2 from latest git because some changes has been made to support m3u for iptv and would like to test it to report any possible bugs. TNX
(2015-10-14, 13:54)Milhouse Wrote: It seems to have crashed in libnfs.so. The only thing I can suggest is trying the latest debug-enabled build #1011x and see if you can obtain a more detailed crashlog.

You do seem to have a lot of NFS errors in your log, culminating in:
Code:
05:19:39 87832.781250 T:1515189248   DEBUG: NFS: Refreshing context for 192.168.1.64/Media, old: 87805401, new: 87827960
05:19:39 87832.781250 T:1515189248   DEBUG: NFS: Using cached context.
05:19:39 87832.781250 T:1515189248   ERROR: NFS: Failed to stat(Media/Videos/Series/The Blacklist/S2/../logo.png) nfs_service failed
a couple of minutes before the crash. which may be part of the problem.

Thanks. I'll try the debug build and see if I can trigger it again. I can't see "a lot" of NFS errors in the log and those I do see, like the one you've posted, result from not being able to find the logo.png file, which is normal as it doesn't exist. However, I do recall I put my PC on standby shortly before I went to bed and seconds later I noticed it was still awake and had to do it again. I assumed I'd just done it wrong but I see a WakeOnAccess attempt at 05:19:26, so it may be that OE woke the PC after the first time I put it in standby and when I put it back into standby it upset OE and triggered the crash.
Here's a libnfs crash with debug enabled.
http://sprunge.us/eHYG

Not obviously reproducible unfortunately.
Leopold's Repository: Home of LibreELEC Dev Updater ...
(2015-10-14, 18:55)Leopold Wrote: Here's a libnfs crash with debug enabled.
http://sprunge.us/eHYG

Not obviously reproducible unfortunately.

Thanks. I've added your log and @doveman2's log to the open issue #124. @doveman2 if you can obtain a debug-enabled crash log that would be most helpful.
Texture Cache Maintenance Utility: Preload your texture cache for optimal UI performance. Remotely manage media libraries. Purge unused artwork to free up space. Find missing media. Configurable QA check to highlight metadata issues. Aid in diagnosis of library and cache related problems.
(2015-10-13, 10:08)woronczak Wrote: The W1070 only supports frame packing at "24p" for 1920*1080. It seems like after stopping a 3d MKV / bluray iso, Kodi doesn't send a message to the projector to say that it should turn frame packing off (the projector tells me Frame Packing is still on) so the GUI looks wrong (or in your case, it just goes blank). If I set the GUI at anything other than 23.98 / 24 @ 1920*1080, as the projector does not support frame packing for those rates, it works normally. I don't fully understand what is happening under the hood, but through trial and error, I found that it works that way. No great loss to have the GUI at 50hz (other than a flicker when the projector shifts modes).

@woronczak @hdmkv

Can you try this test firmware:
https://dl.dropboxusercontent.com/u/3669...e_hdmi.zip

Use start_x.elf and fixup_x.dat and replace start.elf and fixup.dat on your sdcard. By default there will be no change, but can you try with config.txt option:
Code:
hdmi_debug=0xffff

That should make hdmi mode changes take about 20 seconds. I'd like to know if cases that weren't working before behave any differently.

Normally when switching hdmi mode, you assert the "AV MUTE" flag a few vsyncs before changing the clocks, and remove it a few vsyncs after.
The new setting makes this programmable. The lower byte is the number of "AV MUTE" vsyncs before the change, and the next byte is the number after the change.

The defaults are 2 before a 3 after (hdmi_debug=0x0302). If 0xffff does help we can work out exactly what number is required.

If that doesn't help, then try:
Code:
hdmi_debug=0x1ffff

This will now always change to VGA60 and then to desired HDMI mode. I'd be interested if that helps.
(2015-10-14, 20:13)Milhouse Wrote: @doveman2 if you can obtain a debug-enabled crash log that would be most helpful.

Sure, I'll do my best Smile
  • 1
  • 107
  • 108
  • 109(current)
  • 110
  • 111
  • 168

Logout Mark Read Team Forum Stats Members Help
OpenELEC Testbuilds for RaspberryPi (Kodi 16.0)10