• 1
  • 40
  • 41
  • 42(current)
  • 43
  • 44
  • 168
OpenELEC Testbuilds for RaspberryPi (Kodi 16.0)
(2015-08-11, 22:35)Heiko123 Wrote:
Quote:2. Fix for power button shutdown

What did you mean.
Is it a predefined gpio pinout ?

No, nothing to do with GPIOs, this is the fix for the problem reported by @doveman2 and @MrNice, when Kodi would freeze on shutdown after the remote control "Power off" button is pressed.
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.
New OpenELEC Jarvis build #0811: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.1.4 #1 Tue Aug 11 22:28:29 BST 2015 armv6l GNU/Linux

# vcgencmd version
Aug 10 2015 20:05:05
Copyright (c) 2012 Broadcom
version 010f0680c51cd693b4533061fb19f439451d46e9 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20150811222741-#0811-gcac5a18 [Build #0811]

# vcdbg log msg 2>&1 | grep DTOK
001572.693: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (cac5a18a, changelog) and tip of XBMC master (5e59b3d6, changelog) with the following modifications: Build Highlights:
  1. New Activity logging screen. Accessed from Settings > System > Logging. Click "Show event log" to view events (and if enabled, notifications)
Build Details:
  1. OpenELEC:
    • add docker support to kernel (PR:4256, 2 commits, 2 files changed)
  2. XBMC:
    • [Database/Import] - present the user with removable and network share… (PR:7778, 1 commit, 1 file changed)
    • [binary addons] make sure CORE_SYSTEM_NAME is set (PR:7785, 1 commit, 1 file changed)
    • [win32] Fix trac #16165 (PR:7713, 3 commits, 5 files changed)
    • [coverity] Result is not floating-point. (PR:7723, 1 commit, 5 files changed)
    • [Binary add-ons] Point audio encoders/decoders to master instead of SHA (PR:7725, 2 commits, 12 files changed)
    • [Util.cpp] sap stream was dropped by mistake during HDhomerun cleanup (PR:7787, 1 commit, 1 file changed)
    • Activity log (PR:6746, 20 commits, 65 files changed)
    • settings: use constants for event log settings (a6388d82)
    • events: fix showing event log from settings (5e59b3d6)
  3. pvr.hts:
    • Fix: Use epg data only for creation of epg-based timers. (PR:89, 1 commit, 2 files changed)
  4. pvr.vdr.vnsi:
    • Updates to cmake files: (PR:23, 1 commit, 3 files changed)
  5. Additional commits/pull requests/changes not yet merged upstream:
    • Added: dd9ec795: Add crossguid package
    • Added: patch: Add crossguid dependency
    • Updated: patch: Update with latest version of pull request
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.
(2015-08-11, 18:39)doveman2 Wrote: I did notice a fair bit of stuttering and dropped/skipped frames however. When I first checked, I think it was around 479 skipped but I figured that might have been as a result of toggling full-screen, so I restarted the channel and I had a lot less after that but still noticed some gaps in audio and stutters in the video. I'm using MMAL acceleration only and had deinterlace set to Auto. I tried with deinterlace off and that seemed to reduce them but not entirely and even when starting the channel again with deinterlace disabled, there were quite a few gaps in audio for the first few minutes, although it settled down after that. On the third try, I think there were a total of 73 skipped frames (1 dropped) by the time the programme finished and whilst 49 of those occurred with deinterlace on and after I disabled it, that only increased slightly to 52 over the next few minutes, later on there were a couple of obvious stutters/gaps and on checking it had jumped to 63 and the next time, 72.

Just a bit more info. I left TV running for several hours (not watching it) with deinterlacing disabled and when I checked there was something like 144 drops and 646 skips, so those are obviously not caused by deinterlacing.
(2015-08-11, 19:13)popcornmix Wrote:
(2015-08-11, 16:26)illiac4 Wrote: and the output after Kodi crash and manual(systemctl stop kodi):
http://sprunge.us/BPPd

Okay, there are some clues in there. The video_decode and image_fx pools are still present, which suggests one of those components hung.
The "qpu_malloc" shows that advanced deinterlace was in progress when it hung.

Can you switch from advanced deinterlace to bob and then grab the same log if it hangs. Then disable deinterlace completely and grab the same log if it hangs.

With no deinterlacer:

At the beginning of the stream start:
http://sprunge.us/ViiC
At around 380mb free:
http://sprunge.us/WSMg
At around 300mb free:
http://sprunge.us/KWdP
At around 200mb free:
http://sprunge.us/UHIU
At around 100mb free:
http://sprunge.us/AXTL
At around 50mb free:
http://sprunge.us/XGIH
At around 10mb free:

At freeze it has had 40mb left and could not even ssh into pi as it has fully freezed.
(2015-08-12, 05:25)illiac4 Wrote: At freeze it has had 40mb left and could not even ssh into pi as it has fully freezed.

Yes, it appears it can take a while for the OOM killer to kick in.

The other evening I tested a video with a random external ASS subtitle file (known memory leaker) on a Pi2.

I left the video playing on repeat overnight, and this is the amount of free memory at 10 second intervals: http://sprunge.us/DQSD

Here's a journalctl after Kodi had been killed by the OOM killer: http://sprunge.us/fQFA

Notice how the Pi stopped responding to ssh between 10:52:11 and 11:44:23, until Kodi restarted.
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.
I hope that popcornmix will figure it out. In my video stream there are no subtitles involved.

BTW are for AAS leak already any clues how to fixed it?
(2015-08-12, 06:17)illiac4 Wrote: In my video stream there are no subtitles involved.

Yes, sorry - I was just using that as a demonstration of what to expect when an OOM situation occurs. If you had waited long enough, your kodi.bin process would most likely have restarted, and the contents of journalctl would have confirmed the activity of the OOM killer.

(2015-08-12, 06:17)illiac4 Wrote: BTW are for AAS leak already any clues how to fixed it?

Nope. Sad

The one good thing is that it's not RPi specific!
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.
#0811 broke VDR client specific settings (going into OSD will segfault Kodi), switched to OpenELEC-RPi2.arm-6.0-devel-20150808142827-#0808-g75e1b35.tar.

FTA HD stream froze video after around 3 hrs.
DVDPlayerAudio thread of kodi.bin spinning at 100% CPU.
Logs: http://pastebin.com/LpZA4aRA
No asserts in videocore log.

RPi2 config.txt excerpt:
initial_turbo=30
core_freq=450
gpu_freq=300
gpu_mem=320
max_usb_current=1

Restarting Kodi+VDR brought video back this time. This looks more like a Kodi bug now. Anything I should turn off to test? Thanks.
Katastrophentourist
(2015-08-12, 13:06)MarkT Wrote: #0811 broke VDR client specific settings (going into OSD will segfault Kodi), switched to OpenELEC-RPi2.arm-6.0-devel-20150808142827-#0808-g75e1b35.tar.

No crashlog, no problem... Wink
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.
With #0810 and #0811 my RPi2 becomes completely unresponsive with a black screen when starting an HD live tv channel. I have to pull the plug to reboot it.
There seems to be no problem with the SD channels.
Leopold's Repository: Home of LibreELEC Dev Updater ...
(2015-08-12, 20:28)Leopold Wrote: With #0810 and #0811 my RPi2 becomes completely unresponsive with a black screen when starting an HD live tv channel. I have to pull the plug to reboot it.
There seems to be no problem with the SD channels.

What are your overclock settings? Does it work if you remove them?
(2015-08-12, 20:42)popcornmix Wrote:
(2015-08-12, 20:28)Leopold Wrote: With #0810 and #0811 my RPi2 becomes completely unresponsive with a black screen when starting an HD live tv channel. I have to pull the plug to reboot it.
There seems to be no problem with the SD channels.

What are your overclock settings? Does it work if you remove them?

After tinkering with removing various overclock settings and eventually restoring the original config, I have been unable to reproduce the issue again. So currently I am able to view HD live tv channels with #0811 when overclocked. I think the screen turns black for longer than it used to before the video starts playing, but that could be my imagination.
Leopold's Repository: Home of LibreELEC Dev Updater ...
What format is your HD live tv in? Looks like you're in the UK, I'm in the US and mine is 1080i60 and MPEG2. I'm using the following overclock settings and have enabled the new advanced deinterlacing with great success:

dtoverlay=sdhost,overclock_50=84
gpu_mem_1024=320
arm_freq=1050
core_freq=525
gpu_freq=350
over_voltage=4

Instead of force_turbo=1 I use "echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor" in autostart.sh. I'm running 0810 and don't have any issues other than occasional odd messages in the system log like:

[ 544.967786] mceusb 1-1.5:1.0: Error: urb status = -71
[11904.044682] NYET/NAK/ACK/other in non-error case, 0x00000002

Those do not appear to affect video playback, network or IR receiver response that I can tell.
(2015-08-12, 22:40)Leopold Wrote: After tinkering with removing various overclock settings and eventually restoring the original config, I have been unable to reproduce the issue again. So currently I am able to view HD live tv channels with #0811 when overclocked. I think the screen turns black for longer than it used to before the video starts playing, but that could be my imagination.

I think the longer black screen is your imagination.
However the hard lock up I did just catch. Was an issue of changing sdram frequency from the second (vpu) core which I suspect is not safe in the presence of interrupts on first core.
Hopefully there is a fix for that in tonight's (somewhat delayed) build.
New OpenELEC Jarvis build #0812: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.1.4 #1 Thu Aug 13 01:08:47 BST 2015 armv6l GNU/Linux

# vcgencmd version
Aug 13 2015 00:10:56
Copyright (c) 2012 Broadcom
version 2fe75164bde8530ab8e9c384cc0fa4723757ddfa (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20150813010757-#0812-gcac5a18 [Build #0812]

# vcdbg log msg 2>&1 | grep DTOK
001558.313: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (cac5a18a, changelog) and tip of XBMC master (71e6731b, changelog) with the following modifications: Build Highlights:
  1. New firmware with GPU frequency boost to 300Mhz (+50Mhz over stock) when required - disable with "disable_auto_turbo=1" in config.txt
    popcornmix Wrote:This will give a gpu turbo boost to 300MHz when HD advanced deinterlace, HD MVC, or HD high framerate (> 45 fps) content is played.

    This will occur automatically without any config.txt settings needed.

    ...

    The logic is complicated.

    Basically for deinterlace v3d will be bumped to 300MHz, but due to PLL dependency all related clocks will be bumped (core, h264, isp by default, just h264 and isp if using avoid_pwm_pll). For the video decode h264 will be bumped (and dependent clocks will be bumped too).

    Now if you have your own gpu overclock settings things get more complicated. Each clock will be set to the max of user (e.g. h264_freq) and auto (call it h264_boost = 300). So if you have a big overclock on all gpu clocks, then the auto boost will have no effect (apart from enabling turbo at suitable times).

    If you have a partial gpu overclock (e.g. core_freq=500 and h264_freq=250), then it gets more complicated. The core_freq will get 500MHz, which gives a PLL of 2GHz, which means 300MHz is not possible, so h264 will be left at 250MHz. If core_freq was 400MHz, which gives a PLL of 2.4GHz, then 300MHz is possible, so h264 will be boosted to 300MHz.

    So, I think you will never be worse off than before, but some user overclocks may have the effect of disabling the auto boost frequency.
Build Details:
  1. Firmware (Aug 13):
    • firmware: arm_loader: Force turbo changes to first core. Changing sdram isn't safe from second core. See: link
    • firmware: arm_loader: Move the application of user parameters and overlays later. See: link
    • firmware: rtos: Remove breakpoint in zero exception handler
    • firmware: arm_loader: Fix some sanity asserts
    • firmware: di_adv: Update to use chained shaders with larger VPM DMA
    • firmware: arm_loader: Allow temporary gpu frequency boosts to be requested
    • firmware: video_decode: Request boost to 300MHz when playing HD MVC or HD high framerate video
    • firmware: di_adv: Request boost to 300MHz when doing full rate HD advanced deinterlace
    • firmware: arm_loader: Allow auto turbo to be disabled in config with disable_auto_turbo
    • firmware: dma_lib: Allow burst, waits and priority to be specified for 2d dma memcpy
  2. XBMC:
    • [infomanager] add Container(id).CurrentItem / NumItems for grouplists (PR:7768, 1 commit, 3 files changed)
    • [infomanager] add System.HasModalDialog info bool (PR:7776, 1 commit, 2 files changed)
    • [coverity] Add an additional check to prevent an only theoretical out-of… (PR:7724, 1 commit, 1 file changed)
    • cleanup ui sound system (PR:7684, 1 commit, 6 files changed)
    • [guilib] add Container(id).HasNext / HasPrevious for textbox controls (PR:7769, 2 commits, 3 files changed)
    • [gui] new ACTION_MENU key to focus sidebar menu (PR:3885, 3 commits, 21 files changed)
    • [setting] hunt missing setting id constants (PR:7790, 2 commits, 19 files changed)
  3. pvr.hts:
    • Improve connection handling on suspend/wakeup. (PR:90, 1 commit, 4 files changed)
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.
  • 1
  • 40
  • 41
  • 42(current)
  • 43
  • 44
  • 168

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