[L-S-D]

no, I'm not overclocked. I have a default config.txt except a mpeg2 licence and enabled an edid file (as either my bed room tv or hdmi cable is broken and I was to lazy to track the culprit yet).
On OE side, everything is pretty standard, default skin, just mysql for video and music library, currently not even a non default addon. Beside that, I have only ser2net running.

[Milhouse]

@L-S-D: a debug log (wiki) would be the first place to look - does it always freeze the same way (same error, same point etc.)? If you're power cycling the Pi, you'll need to look at kodi.old.log to see the output from the previous Kodi (upload kodi.old.log not kodi.log).

If your pvr.hts configuration is easy to install and reproduce then by all means upload it, but please include step-by-step instructions as I know nothing about PVR.

[popcornmix]

Both test builds from today are crashing for me when I start live tv playback. Rolling back to 0727 works fine with the same settings. Nothing of interest in the log file.

Switching deinterlace to "bob" or using force_turbo=1 should work around this for now.

[illiac4]

actually it's a freeze, so in my case no picture any longer (TV tells me nothing is connected) and no network connectivity/ssh/ping. I can only power cycle the Pi2

This was the same behaviour i have seen with latest build and the sample file i have uploaded.

The one explained by Milhouse is the one I have seen in builds before x0729 and the same video file.
The video is a recording from livetv.

[L-S-D]

is zaphod24 experiencing crashes or freezes?

@popcornmix: force_turbo or bob as deinterlacer is supposed to workaround crashes or freezes? Just asking if it makes sense to playaround with the settings in the evening.

[zaphod24]

I'll try removing overclocking tonight and try again. Technically I guess I was getting a hard freeze after starting live TV playback because the playback just stopped after about 1 second. I was already using bcmstat to monitor the speeds and my ssh session stopped as well. There was no crash log and my only recourse was to cut power and then restart the PI2.

I'm using advanced deinterlacing, so I can only guess the issue is with the new auto-turbo in the firmware update? I'll also try switching back to Auto.

[popcornmix]

I'm using advanced deinterlacing, so I can only guess the issue is with the new auto-turbo in the firmware update? I'll also try switching back to Auto.

Yes, it's the auto turbo. Disabling overclock won't help (there is a default 600->900MHz). force_turbo will.
Automatic deinterlace won't help (it still uses advanced for SD). Explicitly setting BOB will work.

I suspect there will be fix for tonight anyway.

[zaphod24]

Awesome as always! It will be nice not to have to force turbo all the time but only have it running that way when playing video (and possibly only when using advanced deinterlacing).

[Milhouse]

That would explain why no crash for me as I have force_turbo=1.

I'm out this evening so assuming the automated build succeeds at 9pm the builds should be available to download just before 10pm, with the release notes following about midnight.

[Mfleigle]

I have a quick question about the build naming; both #727 and #729c have the same ending gc37b5a7; does that mean Kodi (source code) itself wasn't changed, or something else. I've noticed that sometimes a few days in a row have the same number, and thought I'd ask.

[Milhouse]

As you may have guessed, the automated build failed (this new OpenELEC commit isn't compatible with Kodi 16, so I'll just drop it for now). The new build will be uploaded within the hour...

Edit: Turns out that particular commit is quite important for MCE remote control users, so I've had to fix that commit and respin the build...

I have a quick question about the build naming; both #727 and #729c have the same ending gc37b5a7; does that mean Kodi (source code) itself wasn't changed, or something else.

"c37b5a7" is the git hash for the OpenELEC repo, and is added automatically by the OpenELEC build procedure.

In the case of build #0729c, there were no new OpenELEC updates so the hash remained the same as that used by the #0727 build.

I've noticed that sometimes a few days in a row have the same number, and thought I'd ask.

Yes, and it's for this reason that mentioning this hash when discussing/referencing a test build is not a good idea - it's not definitive, and can therefore be confusing.

The only code that should be used when referencing a build is my own build code, ie. #0729c, as this is unambiguous and should never correspond to multiple builds (unless I cock it up when manually publishing experimental/debug builds, which might have happened once or twice!)

FYI, the Kodi git hash is visible in kodi.log - for example, tonight's #0730 build shows the following:

Code:

03:17:23   8.207521 T:1967562752  NOTICE: Starting Kodi (16.0-ALPHA1 Git:4081de3). Platform: Linux ARM 32-bit
...
03:17:23   8.208012 T:1967562752  NOTICE: Running on OpenELEC (Milhouse) - Version: devel-20150731030807-#0730-g0fa7209 [Build #0730] 6.0, kernel: Linux ARM 32-bit version 4.1.3

which maps to the first line in the release note:

Based on tip of OpenELEC master (0fa72090, changelog) and tip of XBMC master (4081de3a, changelog) with the following modifications:

[Milhouse]

New OpenELEC Jarvis build #0730: RPi / RPi2
(Supercedes previous build)

Code:

# uname -a
Linux rpi512 4.1.3 #1 Fri Jul 31 03:08:57 BST 2015 armv6l GNU/Linux

# vcgencmd version
Jul 30 2015 17:56:23
Copyright (c) 2012 Broadcom
version 065d51651eeaced58522f5591150b64cacdc00f8 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20150731030807-#0730-g0fa7209 [Build #0730]

# vcdbg log msg 2>&1 | grep DTOK
001572.095: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (0fa72090, changelog) and tip of XBMC master (4081de3a, changelog) with the following modifications:

• Includes newclock4 patches
  
• Excludes the OpenELEC fernetmenta patches due to conflicts with newclock4
  
• Excludes the OpenELEC linux-01-RPi_support patch in favour of sourcing these and possibly more recent patches directly from kernel branch rpi-4.1.y
  
• Default setting for "Show RSS Feed" changed to disabled (new installs only) [patch details]
  
• Disabled "Total Duration" in Confluence (see build #0221 for details)
  
• Includes latest dcadec master (2a9186e3, ahead +34)
  
• Includes latest kodi-platform master (33b6390b, ahead +1)
  
• Includes latest libcec master (d997b9aa, ahead +28)
  
• Includes latest libnfs master (68941809, ahead +25)
  
• Includes latest platform master (278828aa, ahead +2)
  
• Includes latest addons: adsp.basic (da6ee5dd), adsp.biquad.filters (b6a31a90), adsp.freesurround (4dd3202f), asplib (490de460), pvr.argustv (b5fef151, +22), pvr.demo (bb2c46a6, +13), pvr.dvblink (1855e441, +13), pvr.dvbviewer (65b86471, +12), pvr.filmon (5d1a5c13, +12), pvr.hts (78a6bd68, +34), pvr.iptvsimple (353cb28d, +12), pvr.mediaportal.tvserver (0de32b2b, +26), pvr.mythtv (30b40c93, +51), pvr.nextpvr (a731f3c4, +15), pvr.njoy (43b4698f, +12), pvr.pctv (ff5d5360, +12), pvr.stalker (b8a61cb8, +12), pvr.vbox (48ebfdae, +24), pvr.vdr.vnsi (8dedf209, +11), pvr.vuplus (0702fc50, +14), pvr.wmc (e87f1328, +14)
  
• Exclude kodi-100.10-handle-SIGTERM.patch: Use updated Kodi patch from stefansaraev repo
  
• Include b24d9e06: handle SIGTERM
  
• Include patch: Fix devinput 18b8ba44
  
• Include patch: Add ffmpeg dependency and includes for HEVC optimisations
  
• Include patch: Add experimental splash video
  
• Include patch: Enable pvr addons, disable PVR updates
  
• Include PR:4228: ffmpeg: update to ffmpeg-2.7.1
  
• Include PR:7462: added: hide thumbs for unwatched episodes option to show plot is off
  

Build Highlights:

1. New firmware - don't request turbo mode when doing qpu deinterlace
   
2. eventlircd/devinput - big cleanup, update default mappings, fix KEY_ENTER/KEY_OK confusion. May break some remotes, may fix other remotes - please test!
   

Build Details:

1. Firmware (Jul 30):
   • firmware: di_adv: Revert: Request turbo mode when doing qpu deinterlace
     
2. OpenELEC:
   • xkeyboard-config: update to xkeyboard-config-2.15 (3de54ec1)
     
   • binutils: update to binutils-2.25.1 (5fd79560)
     
   • gettext: update to gettext-0.19.5.1 (49b27144)
     
   • xf86-input-libinput: update to xf86-input-libinput-0.12.0 (b65cffaa)
     
   • libtirpc: update tio libtirpc-0.3.2 (3ee5cae1)
     
   • eventlircd: big cleanup, remove mappings with source=target, use default mapping for ircore and power mappings, use KEY_ENTER instead KEY_OK (f7c59a63)
     
   • mesa: update to mesa-10.6.3 (c28b4061)
     
   • kodi: configure record button for some remotes (441e1a3e)
     
   • kodi: add some more standard mappings for devinput devices (18b8ba44)
     
3. XBMC:
   • [pvr] fix missing return in CGUIWindowPVRBase::OnBack (fixes #16150) (PR:7646, 1 commit, 1 file changed)
     
   • [guilib] fix hitrect for auto-width buttons (PR:7661, 1 commit, 1 file changed)
     
4. Additional commits/pull requests/changes not yet merged upstream:
   • Added: patch: Fix devinput 18b8ba44
     

[illiac4]

Ok this revert seems to fix the problem but....

Lets assume that I have overclock enabled with the following settings

Code:

# mode 'Turbo'                  :   1000   |    500    |    500     |      6
arm_freq=1000
core_freq=500
sdram_freq=500
over_voltage=6
# set to 'force_turbo=1' to disable dynamic overclocking (you can lose your warranty!)
  force_turbo=0

In this case I will still get Gui freezes like Milhouse did with my sample video since dynamic overclocking is enabled and in case it will speedup the CPU to turbo speed it will freeze the gui.

[illiac4]

Ok here is the same freeze as Milhouse has had with the latest build.
http://pastebin.com/TDmSj1CR

[ivota]

@illiac4 try this:

http://forum.kodi.tv/showthread.php?tid=...pid2062764

This config is working 5 days without issue.
I only have two heatsink. TEMP is about max=60C