• 1
  • 2
  • 3(current)
  • 4
  • 5
  • 146
OpenELEC Testbuilds for RaspberryPi (Kodi 17.0)
#31
(2015-12-07, 22:32)popcornmix Wrote:
(2015-12-07, 21:51)Jdiesel Wrote: I am curious to how colorspace conversion is working on the Videoplayer builds on the RPi. When forced to output “RGB Limited” is the video being scaled 16-235 -> 0-255 -> 16-235?

I have also been testing the Intel EGL builds and have tried forcing RGB full and selecting “limited range 16-235” in Kodi. The issue with that is my display only accepts RGB limited inputs so the method doesn’t work correctly.

After extensive testing I still get the best results with these RPi builds. I was just curious if it was handled any differently than the generic Openelec builds.

It's purely handled by GPU harwdare, so there won't be any common behaviour compared to Intel EGL.
There should only be a single conversion. The YUV data from the video decoder gets converted/composited to RGB on-the-fly and output through HDMI.
There are different matrices used for the different YUV formats (e.g. 601/709 colourspaces) and the limited/full RGB output modes.

That's very good to hear.
#32
(2015-12-08, 12:21)nalor Wrote: Finally it includes only very little changes:

  1. one additional debug line
  2. a correction of the debug output in procedure '_find_config_entry'
  3. and finally the change that corrects the problem: the discid for those 'special' discs is always identical because the file 'UNIT_KEY_RO.INF' contains only NULL-bytes and so I've simply created a constant for this special 'null_byte_discid' and in case the procedure '_calc_title_hash' (that calculates the discid for the current disc) successfully returns there's now a comparison to check if the current disc is a special disc or not - and in case it is simply return 'AACS_SUCCESS' and everything is fine. This is working because the final decryption of the disc is done in procedure 'aacs_decrypt_unit' and the first action inside this procedure is to check if the unit is encrypted at all and in case it isn't it simply returns success (and in our 'special' case nothing is encrypted so everything is working well Smile )

Hopefully this is working for Milhouse too and we can reenable libaacs support Smile

Thanks for investigating. Hopefully we can enable it again without issues.
#33
@Milhouse I installed 1207 that I found on your server. pvr.mythv (3.4.1) seems to load just fine and I'm running a test with it right now. Thanks. Smile
Experience: It's what you get when you were expecting something else.
#34
Glad it's working, bit late with the release post today - Christmas lunch yesterday...
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.
#35
New OpenELEC K* build #1207: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.3.0 #1 Mon Dec 7 21:04:20 GMT 2015 armv6l GNU/Linux

# vcgencmd version
Nov 30 2015 20:58:50
Copyright (c) 2012 Broadcom
version 411fa8fd4bbe6a0a4b0c50f09330ce84c1d70b3d (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20151207210330-#1207-gc5875ae [Build #1207]

# vcdbg log msg 2>&1 | grep DTOK
001710.440: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (c5875ae6, changelog) and tip of XBMC master (c5d208cf, changelog) with the following modifications: Build Highlights:
  1. VideoPlayer now merged
  2. Update PVR clients to 5.10.0 GUI API
Build Details:
  1. XBMC:
    • VideoPlayer (PR:8486, 125 commits, 301 files changed)
    • fix leaking credentials in logs (PR:8500, 2 commits, 3 files changed)
    • [advancedsettings] cleanup pvr related leftovers after videoplayer merge (PR:8509, 1 commit, 2 files changed)
    • AE: fixes for passthrough (PR:8512, 1 commit, 2 files changed)
    • VideoPlayer - PVR: fix crash on channel switch for backends not navin… (PR:8506, 2 commits, 4 files changed)
    • REVERT: [droid] leave MPEG2 h/w decoded; VOB playback is niche (PR:8520, 1 commit, 1 file changed)
    • [win32] Fix post build event copy instruction. (PR:8513, 1 commit, 1 file changed)
    • jsonrpc: fix Video.Details.MusicVideo definition after c5b281b2edf9182c9baebddf2c25dacf2cc74ce1. (319f19db)
  2. dcadec:
    • Use separate variable for number of channels. (10de9721)
    • Improve primary channel set downmixing. (e88b978a)
    • Clarify API comment regarding sample data lifetime. (0fb1fa42)
  3. pvr.mediaportal.tvserver:
    • Updated to GUI API v5.10.0 (PR:27, 1 commit, 2 files changed)
  4. pvr.mythtv:
    • Updated to GUI API v5.10.0 (PR:49, 1 commit, 2 files changed)
  5. pvr.nextpvr:
    • Updated to GUI API v5.10.0 (PR:32, 1 commit, 2 files changed)
  6. pvr.vdr.vnsi:
    • Updated to GUI API v5.10.0 (PR:41, 1 commit, 1 file changed)
  7. pvr.wmc:
    • Updated to GUI API v5.10.0 (PR:28, 1 commit, 2 files changed)
  8. newclock5:
    • New commits in this build:
      • VideoPlayer: realtime streams do require special treatment, audio must be capable of accepting speed adjustments (1a4b4c2f)
      • gfx: fix toggle fullscreen when playing video (a4720ecd)
    • Commits no longer in build:
      • renderer: promote interface up to BaseRenderer (bbf64bf4)
      • renderer: refactor RenderManager and LinuxRendererGL (51a9eb2d)
      • X11: add EGL (0ac305f7)
      • fixed: don't assing std:Confusedtring to nullptr (3b2dfb9d)
      • Fixed compilation of refactored EGL on X11 (f475de72)
      • dvdplayer: fix dropped frame counter - it counted not all drops (458f044c)
      • renderer: fix forced deinterlacing for non-render methods (01152ffb)
      • make video renderer non-global (5653532c)
      • move VideoRenderers (18cefe61)
      • gl: split output stage from yuv2rgb shader (0b280bab)
      • gl: split output stage from convolution shaders (a4eb9924)
      • gl: GLSLOutput helper with dithering (c4a136d2)
      • gl: disable dither texture after rendering only when dithering is used (fb6981aa)
      • settings: add video dither (6eb24719)
      • rename dvdplayer to VideoPlayer (3c9e9607)
      • AE: refactor IAudioCallback (3f39ddaf)
      • move audio sync to AE (337907d4)
      • renderer: fix HasFrame (b3018361)
      • configure: Add raspberry-pi2 platform (253ec417)
      • [gles] Fix build with gles. GL_RED is not defined (a8eeb7df)
      • [GLES] Fix build with VideoPlayer (53784659)
      • [MMAL] Update to support VideoPlayer renderer (8d9069fe)
      • fixed: macro existence should not rely on config.h usage (d0c65013)
      • fixed: unify include paths with other files (b35e7834)
      • VideoPlayer: rework HandlePlayspeed (7f93a6ee)
      • dvdplayer: you are dead, dont you come back anymore' (3407a329)
      • [shader] - fix compilation of glsl shaders for gles (653c5b2b)
      • [shader] - use GLSLOutput object on gles platforms aswell (49d62934)
      • [ios] - add implementation for RendererVTB (a9697855)
      • [VideoRendererGLES] - refactor to support derived hw dec renderer - similar to VideoRendererGL (atm only VTB is implemented. All other hw dec renderers are stil (0403b9af)
      • [RenderManager] - hookup hw dec RendererVTB (87f69fd9)
      • [HwRendererVTB] - add to CMakeLists.txt and Makefile buildsys (d30700a7)
      • [RendererVTB] - some more cleanup (ddb712e0)
      • [HwRendererIMX] - moved IMX video rendering out of LinuxRendererGLES.cpp (8ddfd4f3)
      • [HwRendererOMX] - moved OpenMax (OMX) video rendering out of (94352105)
      • [HwDecRenderers] - get some more platform specific stuff out of (d6da3844)
      • [HwDecRendererStagefrighht] - moved OpenMax (OMX) video rendering out of LinuxRendererGLES.cpp into its own class (2e0474f8)
      • Add X11 EGL depends build support (effd65c9)
      • [VideoPlayer] cleanup DVDVideoCodec.* (70a8a50d)
      • [HwDecRendererMediaCodec] - moved Android MediaCodec (surface) video rendering out of LinuxRendererGLES.cpp into its own class (a641873e)
      • [VideoRenderer/GLES] - drop AddSupportedHwRenderFormats - m_formats will (323c4c66)
      • remove stagefright (bd8f054d)
      • changed: allow multiple callbacks in AE (36f01c0c)
      • RenderManager - cleanup CreateRenderer (5786fd99)
      • VideoPlayer: refactor resolution switching (6dc57572)
      • VideoPlayer: flush renderer when skin is changed (5e99cf1e)
      • VideoPlayer: reqork dithering (a2d5c2df)
      • [win32] Fix compilation errors. (2a6eddbf)
      • VideoPlayer: keep dvdplayer path for libs on Windows, temp (c2831acf)
      • VideoPlayer: limit sleep time for render to 500ms (88b112af)
      • VideoPlayer: win32 fixes for ffmpeg (d9096829)
      • rename OnLostDevice, OnResetDevice (776700a3)
      • fixup: Fix GLES and MMAL builds (26caad66)
      • VideoPlayer: honor flac metatags for channel layouts (9ce560ba)
      • set desired resolution before start of playback (091d54f5)
      • [depends] fix target raspberry-pi (1da2f07b)
      • [EGL] don't always include amlogic (34d7edc5)
      • VideoPlayer: clean audio interface (4ec8088d)
      • disable passthrough when sync playback to display (bc51d231)
      • VideoPlayer: drop useless masterclock (7bbf64a4)
      • X11: delayed OnResetDisplay after change of refresh rate (61a63bca)
      • VideoPlayer: rework player sync to clock (3647ebd9)
      • Advancedsettings: Remove blackbarcolour and set it automatically (1068acb3)
      • Fix iMX6 compilation (0e09c93f)
      • Fix imx buffer lock/release to have proper ref count (41c7a4fe)
      • VideoPlayer: drop annoying asked for new data packet msg (9647b85e)
      • vaapi: do not disassociate egl context (ec310ad5)
      • VideoRenderer: fix state after flush (f9628373)
      • VideoPlayer: reslotion fixes (8a205f19)
      • [omxplayer] Ensure MMAL renderer is used in BYPASS mode (95a72d8a)
      • [omxplayer] Update to new VideoPlayer sync scheme (00d7c8a6)
      • sync XCode project (3ae54b1d)
      • VideoPlayer: allow speed up to 1.5 (5bba2bae)
      • VideoPlayer: add debug log for flushing audio (863a36a0)
      • VideoPlayer: do not flush and drain audio on stall (954f0f2e)
      • AE: fix calculation of bytesToSkip to avoid cracking sounds (128bd897)
      • [AE] Fix large buffertime check (e20827b7)
      • VideoPlayer: fix cachestate after FlushBuffers (0637b49b)
      • [renderer] Remove RenderUpdateCallBack and RenderFeaturesCallBack (c7e722aa)
      • Fixup mediacodec (feb45e4b)
      • Get rid of libstagefright (604bd4d7)
      • baseline amlcodec (14ab8aae)
      • fixup aml (b5b5064a)
      • squash: fix wrong prototype for RenderUpdateVideoHook (b19f4a06)
      • vaapi: cope with comtinuous get_buffer errors (e927a934)
      • VideoPlayer: fix ff after recent changes (7aa4008a)
      • merge error (8f7ea7cd)
      • VideoPlayer: flag and handle realtime streams (c313475f)
      • VideoPlayer: pass CFileItem to input stream (a4fa90c3)
      • vtb: rename after merge (e2c95162)
      • AE: init sample pts (96df0b53)
      • ffmpeg: Update AVPixelFormat and AV_PIX_FMT_* to compile with master (c31b7d37)
      • AE: move iec packing for passthrough into AE, contains squashed code from fritsch (071d9423)
      • AESinkALSA: Explicitely push supported Stream Types and AE_FMT_RAW (47370e0b)
      • AESinkAudiotrack: Add Stream Types (2142282d)
      • omxplayer: fixup: AE: test type of stream for passthough modes (20809aef)
      • AESink: Remove m_frameSamples from sink implementations (82ac71fc)
      • AEUtils: Add Streamtypes to Enumeration output (70619dfe)
      • AEUtils: Fix DataFormatToStr with new AE_FMT_RAW (500e011a)
      • AE: drop old raw data formats (4d17aaa6)
      • DVDAudioCodecPassthrough: Set samplerate when creating StreamInfo (8c8059a1)
      • AE: update OSX sink (9a34c6d0)
      • [win32] Add handling of DispResResource. (e795de6b)
      • [win32] Change way to delayed dll load. (4692d179)
      • AESinkDirectSound: Prepare for new STREAM_TYPES (58d85d97)
      • AESinkWASAPI: Prepare for new STREAM_TYPES (ba483051)
      • AESinkDARWINIOS: Port to new STREAM_TYPE (76a402bc)
      • AE: set channel layout for iec packing (9fb9c974)
      • AE: fixes (truehd) and cleanup (5420c9ef)
      • [OMXAudio] Remove obsolete hw decode support (bcdd886a)
      • [OmxAudio] Make m_format more local (55910364)
      • AE: update omx audio (c5d50afe)
      • AE: update Pi Sink (b744e29a)
      • VideoPlayer: trash method OpenStreamPlayer - enables detection if stream was actually opened or not (e544d90c)
      • [videoplayer] Use a copy of stream info for thread safety (4b118f67)
      • RenderSystem: reduce video latency during video playback - GL part (b2917be3)
      • [omxplayer] Update to support VideoPlayer renderer (37762d0e)
      • [omxplayer] Remove debug code that uses removed GetCacheTotal api (15c42604)
      • VideoPlayer: remove some libstagefright remnants (dc8b218b)
      • addon ae: fixups (1be498ac)
      • fixup VS project (517bdeff)
      • pthreads: use mutex protocol PTHREAD_PRIO_INHERIT (b44348d2)
      • VideoPlayer: fix compile error on Android (7b8c5985)
      • OSX: remove remnants to libGLEW (219566f1)
      • videoplayer: mediacodec surface (e616f790)
      • [videoplayer] Allow passthrough with display clock and pll adjust (72c938a5)
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.
#36
(2015-12-07, 21:31)Pienoet Wrote: It was working with 1205 build but stopped working with 1206.
Sorry it took so look but here a my logs:

log

You'll need to wait until pvr.hts is bumped to the new GUI API.
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.
#37
(2015-12-08, 12:21)nalor Wrote: Hopefully this is working for Milhouse too and we can reenable libaacs support Smile

Great, this is working - I'll include it in the next test build.

Do you have any plans to send this upstream, to libaacs?
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.
#38
(2015-12-08, 16:01)Milhouse Wrote:
(2015-12-08, 12:21)nalor Wrote: Hopefully this is working for Milhouse too and we can reenable libaacs support Smile

Great, this is working - I'll include it in the next test build.

Do you have any plans to send this upstream, to libaacs?

I'll try - but honestly I've never sent something upstream in my life - so I'll first have to figure out how I can contribute something to libaacs Wink
#39
(2015-12-08, 06:57)username145 Wrote: I think I tried, but ran into other problems. Maybe it would try to mount before the network was ready, or I was not too familiar with linux and wasn't able to get it working.

I've opened issue #4483. For now, use autostart.sh to mount filesystems.
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.
#40
I've been running a live TV stream for something like 5 hours and no high CPU usage and only 410 skipped frames using videoplayer on the 1207 build. That's much better reliability than I've had in a very long time. Here's a non-debug log if you guys care to see it.

http://sprunge.us/aXLG

There was one read error on the live stream reported by the addon apparently. I'm guessing that's where the 400 skipped frames came from. At any rate, this is much better. Thanks to popcornmix for whatever he's doing in videoplayer and to Milhouse for making all this possible and last but not least to janbar for providing the myth tv pvr addon. Smile

Is mii-tool and/or iperf available from some repository addon? I'd like to do some testing from my Raspberry Pi's perspective. I'd also like to turn off full duplex for grins. I really wish I had some smart switches, but I'd need at least three and they're kind of expensive.

The following is just more rhetoric about my network issues investigation.

As for my network investigations, I took the gigabit switch out of the livingroom and replaced it with a 100Mb switch. I'm still trying to determine if that made any difference in anything. I had a couple of freezes in the LR playing back recordings and also playing a movie from the Video part of Kodi. It was pulling the video from the mythbackend through a samba share. The movie froze twice in a fairly short time, but proceeded on after a few seconds. After changing out the switch, the movie didn't freeze again, but a recording playback did a few times. It didn't matter whether I used OMX or DVDPlayer. This is all using OE6 released version in the living room. The myth backend is in the same room so the only thing in common was the 100Mb switch. I still get htop freezes in the bedroom and in the livingroom even though the linux box (myth) and my laptop are on the 100Mb switch. The myth box is enabling flow control with the switch. I'm not sure that's a good thing from what I've read. Every bit of this could be from one of my many network devices randomly renegotiating with one of my switches or possibly full-duplex cross-talk on a cable. I may see if I can rent a network tester to certify all my network cabling in the house.
Experience: It's what you get when you were expecting something else.
#41
(2015-12-08, 19:58)afremont Wrote: Is mii-tool and/or iperf available from some repository addon? I'd like to do some testing from my Raspberry Pi's perspective. I'd also like to turn off full duplex for grins. I really wish I had some smart switches, but I'd need at least three and they're kind of expensive.

iperf is in the openelec unofficial add-ons repo.
#42
A technical question about PVRs. When watching TV recordings, does the addon stream the data to the Pi, or does the Pi read the file itself? I'm guessing that the Pi asks for data as it needs it and the addon is the thing actually reading the disk.
Experience: It's what you get when you were expecting something else.
#43
(2015-12-08, 20:19)afremont Wrote: A technical question about PVRs. When watching TV recordings, does the addon stream the data to the Pi, or does the Pi read the file itself? I'm guessing that the Pi asks for data as it needs it and the addon is the thing actually reading the disk.

My understanding is that recordings from the Live TV menu will use the PVR backend to read the file (but I don't know a lot about this)
If the recordings directory is accessed from the Files menu, then the Pi will read the file directly.

The code path for these two cases is quite different, so if one behaves better than the other, then that may be interesting.
#44
(2015-12-08, 15:50)Milhouse Wrote:
(2015-12-07, 21:31)Pienoet Wrote: It was working with 1205 build but stopped working with 1206.
Sorry it took so look but here a my logs:

log

You'll need to wait until pvr.hts is bumped to the new GUI API.

Ok thanks milhouse untill then i guess it isn't possible update to newer builds then 1205 if using pvr.hts client?
#45
Since I can see that there is something odd going on with my network, I'm going to chalk any issues I see up to that for the time being. I played a movie from the videos menu yesterday on the 1206 build and it stuck a few times for 10 seconds or so and then went on. I assume it was using videoplayer since OMXPlayer is disabled. I don't think it really matters though since I've seen OMXPlayer hang up the same way lately. I think the network is stalling so it affects the addon and the video menu file access. I have the same thing happening in my livingroom too and there is only one switch involved and it is not gigabit. I'm beginning to suspect that the mythbackend is causing these stalls somehow even though I can see htop hang when the backend isn't directly involved. But it's the one single thing in common with the video, recording and live TV playback problems.

I'm wondering if the myth backend is jamming the network somehow, either by flooding the switch or sending flow control commands to the switch causing everything to stall. I'm going to see if I can disable flow control on the backend and see what happens. If it is, it still doesn't explain absolutely everything. I'm like Monk and Columbo when it comes to this kind of stuff. Wink I can't let it go until everything adds up properly to one solid answer. dmesg in all devices looks fine, there are no oddball network things showing up there, not even any unexplained link-up/link-down things.
Experience: It's what you get when you were expecting something else.
  • 1
  • 2
  • 3(current)
  • 4
  • 5
  • 146

Logout Mark Read Team Forum Stats Members Help
OpenELEC Testbuilds for RaspberryPi (Kodi 17.0)6