Kodi Community Forum

Full Version: OpenELEC Testbuilds for RaspberryPi (Kodi 15.0) Part 2
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
(2015-05-21, 16:36)Milhouse Wrote: [ -> ]
(2015-05-21, 15:53)Leopold Wrote: [ -> ]I thought the point of the patch was to delay the HDMI mode change until the hello_video.bin process is no longer running? If so then I don't see how that's possible.

Indeed, but you said "Here's one where the hdmi mode changed before the video ended." so somehow the HDMI mode is changing before the video has played out, which might be upsetting the very simple video player used to display the splash video. I'm only guessing though.
Sorry I seem to have confused things with that statement. I should have said completed instead of ended Wink.

I originally thought that the video may still be playing while the screen goes black during the video mode change, prior to the Kodi gui appearing. From the logging and the patched code I can now see that is not the case and that the video must be terminating before the hdmi mode change.
(2015-05-21, 16:22)Milhouse Wrote: [ -> ]@popcornmix: just noticed in the last few logs that have been posted (and also my own) that the default cacheMemBufferSize is being set to 2MB on RPi2 systems with plenty of available ARM mem - shouldn't this be 20MB?

I can confirm: on my Pi2 cache does not exceed 2MB.
(2015-05-21, 17:02)Leopold Wrote: [ -> ]I originally thought that the video may still be playing while the screen goes black during the video mode change, prior to the Kodi gui appearing. From the logging and the patched code I can now see that is not the case and that the video must be terminating before the hdmi mode change.

I'll modify the kodi-splash.sh script in the next build and capture stderr/stdout from the video player to /tmp/splash.log, maybe it's segfaulting.

If it is segfaulting, with the next build a core will be dumped into /storage/.cache/cores - upload the stacktraces as follows:
Code:
gdb /usr/bin/hello_video.bin --batch -ex "thread apply all bt" --core=$(ls -1 /storage/.cache/cores/core.*hello_video.bin*) | pastebinit
Hi,

Everytime I select live TV, kodi freezes and then reboots. Which log do I need to upload to show the crash?

Thanks
(2015-05-21, 16:22)Milhouse Wrote: [ -> ]@popcornmix: just noticed in the last few logs that have been posted (and also my own) that the default cacheMemBufferSize is being set to 2MB on RPi2 systems with plenty of available ARM mem - shouldn't this be 20MB?
Code:
14:02:36   8.754464 T:1967771648  NOTICE: ARM mem: 656MB GPU mem: 352MB MPG2:0 WVC1:0
14:02:36   8.754519 T:1967771648  NOTICE: cacheMemBufferSize: 2MB

Looking at the patch, is the m_arm_mem variable in line 70 the same m_arm_mem variable in line 50?

Okay, I see this. I think I know what it is, but I'm doing a clean build so will take a little while to confirm.
(2015-05-21, 18:45)Marshy33 Wrote: [ -> ]Hi,

Everytime I select live TV, kodi freezes and then reboots. Which log do I need to upload to show the crash?

Thanks

The crashlog - upload the latest file with the following ssh command:
Code:
paste $(ls -1 /storage/.kodi/temp/kodi_crashlog* | tail -1)
(2015-05-21, 18:49)Milhouse Wrote: [ -> ]
(2015-05-21, 18:45)Marshy33 Wrote: [ -> ]Hi,

Everytime I select live TV, kodi freezes and then reboots. Which log do I need to upload to show the crash?

Thanks

The crashlog - upload the latest file with the following ssh command:
Code:
paste $(ls -1 /storage/.kodi/temp/kodi_crashlog* | tail -1)
Thankyou for the quick reply, here is the log...

http://pastebin.com/V65rghmH

I first set up live TV on 512(I believe) and it worked fine, then after updating to a newer version, the crashes started happening.
@Marshy33: Try deleting your EPG database with:
Code:
rm /storage/.kodi/userdata/Database/Epg10.db
and restart Kodi.

If that doesn't work, you'll need to contact the maintainer of your PVR client.
(2015-05-21, 19:03)Milhouse Wrote: [ -> ]@Marshy33: Try deleting your EPG database with:
Code:
rm /storage/.kodi/userdata/Database/Epg10.db
and restart Kodi.

If that doesn't work, you'll need to contact the maintainer of your PVR client.

Thankyou again. I have tried what you suggested but still no luck.

I will contact the maintainer of the PVR client.

Really appreciate the help you give on this forum. Smile
New OpenELEC Isengard build #0521: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.0.4 #1 Thu May 21 21:02:23 BST 2015 armv6l GNU/Linux

# vcgencmd version
May 18 2015 16:26:35
Copyright (c) 2012 Broadcom
version dda584d6907a6c642dcae5a260aa3396f9146cd2 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20150521210136-#0521-gcf51d16 [Build #0521]

# vcdbg log msg 2>&1 | grep DTOK
001539.006: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (cf51d162, changelog) and tip of XBMC master (da65bf6b, changelog) with the following modifications: Build Highlights:
  1. Shadertoy improvements
  2. Temporarily add /tmp/splash.log logging, and core dump capability, to kodi-splash.sh
Build Details:
  1. dcadec:
    • Refactor core extension sync word checks. (5a93570a)
    • Make core extension parsing more robust. (ec3ee289)
    • Fix core profile reporting. (92af012b)
    • Change core info structure. (1d714d86)
  2. newclock4:
    • New commits in this build:
      • [visualisation] Improvements to shadertoy (caee07b7)
  3. kernel 4.0.y:
    • New commits in this build:
      • vchiq: Change logging level for inbound data (03d42eb7)
  4. Additional commits/pull requests/changes not yet merged upstream:
    • Updated: patch: Add experimental splash video
@Leopold/@ivota:

Can you configure your systems as you would normally, so that the splash doesn't fully play out, then upgrade to build #0521.

Once booted into build #0521, and after the splash video has failed to play to completion, can you paste the contents of /tmp/splash.log into this thread.

If it's just
Code:
RES: 0
then there is no error and the video player actually completed successfully, which right now would be a little surprising.

However if the contents of /tmp/splash.log is something like:
Code:
Segmentation fault (core dumped)
RES: 139
then you should have a core file in /storage/.cache/cores - please upload the stacktraces with the following command:
Code:
gdb /usr/bin/hello_video.bin --batch -ex "thread apply all bt" --core=$(ls -1 /storage/.cache/cores/core.*hello_video.bin*) | pastebinit
build #0521
/tmp/splash.log = RES: 0

psmon.log
http://sprunge.us/XGMB
(2015-05-21, 23:52)ivota Wrote: [ -> ]build #0521
/tmp/splash.log = RES: 0

psmon.log
http://sprunge.us/XGMB

Thanks, doesn't look like it's seg faulting, so I'm at a loss.
http://sprunge.us/FPcY

/tmp/splash.log
Code:
RES: 0

No core file

So the process ran for about 3 seconds instead of the full 6 seconds that is expected for normal execution, but still returned an exit status of zero. Strange.
Is this the code for the included hello_video.bin?
In last few builds i see that during watching live tv pi just freezes.
The log was 440 mb big so these are just last lines before it freezes if this is helpful at all.

The last part before freeze:
http://pastebin.com/Z8JNSgXj


This is the first part of sys info:
http://pastebin.com/wDBxy4gb