• 1
  • 78
  • 79
  • 80(current)
  • 81
  • 82
  • 89
OpenELEC Testbuilds for RaspberryPi (Kodi 15.0) Part 2
New OpenELEC Isengard build #0622: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.1.0 #1 Mon Jun 22 21:15:13 BST 2015 armv6l GNU/Linux

# vcgencmd version
Jun 22 2015 20:06:24
Copyright (c) 2012 Broadcom
version f6fc02bbbe78132c0328c1f74b66876d8988e1a3 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20150622211425-#0622-gabcff54 [Build #0622]

# vcdbg log msg 2>&1 | grep DTOK
001570.252: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (abcff54b, changelog) and tip of XBMC master (95a4fdc3, changelog) with the following modifications: Build Highlights:
  1. New firmware and newclock4 updates with 3D detection fixes:
    popcornmix Wrote:There are a few issues with 3D video playback. I'm working on some tweaks to avoid them.
    One issue is that EDID reporting of 3D support is not always reliable, so I'm planning to allow 3D modes to be enabled that aren't reported
    (a bit like hdmi_force_edid_3d=1). I'm also trying to simplify the logic for switching 3d modes.

    There will probably be a (minor) firmware update later, as well as some kodi tweaks. Would be useful if you can include them in tonight's build.
  2. Kernel 4.1
Build Details:
  1. Firmware (Jun 22):
    • firmware: hdmi: Don't reject 3D modes based on edid Allows 3D to be switched to without explicit support in edid
    • Update to latest overlays README
  2. OpenELEC:
    • Syntax error in shell conditional (PR:4207, 1 commit, 1 file changed)
    • linux: update to linux-4.1 (bfd3b42f)
    • vdr-plugin-vnsiserver: update to 1c63a8a (53b012df)
  3. XBMC:
    • [pvr.wmc] Fix android lib name (fixes #16044) (PR:7324, 1 commit, 1 file changed)
    • video library: show "Set content" even during library scan (fixes #16061) (PR:7318, 2 commits, 1 file changed)
    • dvdplayer fix sorting of subtitles - fixes trac 16038 (PR:7284, 1 commit, 1 file changed)
    • [confluence] increase button group height in the PVR group manager dialog (PR:7327, 1 commit, 1 file changed)
    • music library: don't force a rescan after update past version 35 if not needed (PR:7321, 1 commit, 1 file changed)
  4. pvr.wmc:
    • Correct android lib name (PR:11, 1 commit, 2 files changed)
  5. newclock4:
    • New commits in this build:
      • Revert "[3D] don't back out from 3D handling on resolution changes if we're switching to a different 3D mode" (b3415f42)
      • [WinSystemEGL] Remove unused Support3D function (2f334182)
      • [3d] Avoid switching stereo mode based on current display mode (982f4943)
      • [rbp] Drop reporting 3D modes and just use current rendering mode to request 3D signalling (3d84d2e9)
      • squash: fix dependency of framepacking on mvc (f3067868)
      • [EGL] Don't skip out when HDMI mode is considered the same (bc3d54b7)
    • Commits no longer in build:
      • [graphics] Allow switching to a more suitable 3D resolution (0be23102)
      • [3D] Support switching to 3D resolutions (40e60b8a)
      • [graphics] Make pixel ratio for 3d modes consistent (8ebacac4)
  6. kernel 4.1.y:
    • Commits no longer in build:
      • squash: Fix inconsistency in pinctrl Makefile patch (fd08db11)
      • platform: squash: Remove IRQF_DISABLED (05376fce)
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 remove PR:7020 from builds as it does not seem like it will be included in isengard
OK... why do PVR changes have to be so contentious? Sad

Although as a side note I include PRs in these builds irrespective of whether they're for Isengard or a future (15.1/J*) release - if the PR is good, has a benefit and is without obvious blockers then I'll include it - if it's well tested it improves the chances of it merging sooner than later.

Only when it becomes apparent that a PR is not likely to be merged at all (ie. too many issues or disagreement) - or no longer applies cleanly on top of master - will I stop including it and PR7020 now falls into this category (too much disagreement).
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.
(2015-06-21, 12:15)popcornmix Wrote: Can you post a debug log (wiki) when showing a 3D video being played.

I would, but I'm on #0622 now, and what do you know? It sort of works, puts the projector into 3D mode, delivers video at 1920x1080p60 with TAB, and the DLPLink glasses work. It had a few hickups, like the GUI switching back to 2D, but the projector still sticking to 3D, thus giving a messed up GUI, but overall it works. And the first few tries, the projector showed some very interesting things. And yeah, I still have my "Video -> Acceleration -> Use Full HD HDMI modes for 3D" disabled, so I'm not sure if this is the intended behavior.

Hifiberry-Digi+ also still working, so no breaking changes on that front.

I'm unsure if it wouldn't be still wise to have an option for the automatic 3D-FP detection/switching to be disabled and be allowed to manually set how the movie should be displayed. Helix did a good job with many legacy devices like mine, and now in some configurations I even lose quality when I have to stream an SBS video as TAB. This reduces the resolution vertically and horizontally, whereas with Helix, I could match the video format and the way the projector interpreted the video.

AirPlay is still broken, plays one second and then halts. UPnP over the same source and RPI does work, though.

The log is 15 MB, but I'll need to clean out passwords before I can post them, because there are a lot of them in there.
New OpenELEC Isengard build #0623: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.1.0 #1 Tue Jun 23 21:04:42 BST 2015 armv6l GNU/Linux

# vcgencmd version
Jun 23 2015 18:20:38
Copyright (c) 2012 Broadcom
version 144b1da894429d00352f650408d9b1a436302f7d (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20150623210354-#0623-gabcff54 [Build #0623]

# vcdbg log msg 2>&1 | grep DTOK
001566.359: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (abcff54b, changelog) and tip of XBMC master (e916ffa8, changelog) with the following modifications: Build Highlights:
  1. New firmware
  2. Flip default of prefer pts timestamps
Build Details:
  1. Firmware (Jun 23):
    • firmware: arm_display: Fix fb_base alias returned from mailbox property interface. See: link
    • firmware: mem_unlock - prevent decrementing lock count below 0
    • firmware: camera: Support multi-channel raw image capture. firmware: camera: write_raw copies input frame if HDR is enabled. firmware: camera: Write correct camera mode from write_raw_md_stage. firmware: AGC tuner: Correction to setting default digital gain to x1.0
  2. XBMC:
    • [pvr] bump pvr.vbox to 1.3.3 (PR:7332, 1 commit, 1 file changed)
    • [dvdplayer][fix] Initialize SelectionStream members. (PR:7333, 1 commit, 1 file changed)
  3. pvr.vbox:
    • compare channel names case-insensitively in Guide::GetChannelId() (fixes #73) (9fed22af)
    • ensure startup state is not changed if a retrieval request fails (fixes #68) (b2c8fb44)
    • add identifiers to request and log that when throwing RequestFailedException (bf26dca4)
    • use the connection timeout setting when opening timeshift buffer inputs (186ec037)
    • properly close the timeshift buffer if it failed to open, otherwise we may (b1e4d1bc)
    • store channels as shared pointers, this way we don't need a copy of the (cc818c7e)
    • let Kodi play radio channels directly without using the streaming interface, (9ec3e61a)
    • improved failure handling in RetrieveExternalGuide() (cd204988)
    • bump version to 1.3.3 (0f86e76b)
  4. newclock4:
    • New commits in this build:
      • [rbp] Use default resampling setting on Pi2 (6bbaf8e5)
      • [rbp] Flip default of prefer pts timestamps (30bd8d2b)
      • force empty repository refresh on browse (9fdc1796)
      • fixed: Don't do GetLength() off-thread is it may be (b)locked by a pending read() (fixes #16046) (5ccadbd1)
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.
(2015-06-23, 20:44)Milhouse Wrote: OK... why do PVR changes have to be so contentious? Sad

Although as a side note I include PRs in these builds irrespective of whether they're for Isengard or a future (15.1/J*) release - if the PR is good, has a benefit and is without obvious blockers then I'll include it - if it's well tested it improves the chances of it merging sooner than later.

Only when it becomes apparent that a PR is not likely to be merged at all (ie. too many issues or disagreement) - or no longer applies cleanly on top of master - will I stop including it and PR7020 now falls into this category (too much disagreement).

You are right 7020 did solve my issues with multiple backends but since it seems it will not be included there is no point in having you compile it in your builds at the risk of impacting something else.
(2015-06-24, 01:25)gendo Wrote: You are right 7020 did solve my issues with multiple backends but since it seems it will not be included there is no point in having you compile it in your builds at the risk of impacting something else.

These builds aren't for Isengard. Once Isengard forks, these builds will stay on master.
So as long as 7020 is expected to be merged after Isengard fork, and may be beneficial to some, then I see no need to drop it.
Obviously if someone reports a regression that seems related to the PR, then we inform PR's author and may drop it if there isn't a quick fix.
Identifying regressions before they get merged is one of the main reasons for these builds.
OK would be great if we could keep it.. fixes issues with multiple pvr backends for me.. just did not want to introduce extra work on you and milhouse.

the Flip default of prefer pts timestamps change in last nights build made a couple of pvr channels (satellite via vuplus addon) stutter.. i switched it of and all is fine now.
(2015-06-23, 22:49)graealex Wrote: I would, but I'm on #0622 now, and what do you know? It sort of works, puts the projector into 3D mode, delivers video at 1920x1080p60 with TAB, and the DLPLink glasses work. It had a few hickups, like the GUI switching back to 2D, but the projector still sticking to 3D, thus giving a messed up GUI, but overall it works. And the first few tries, the projector showed some very interesting things. And yeah, I still have my "Video -> Acceleration -> Use Full HD HDMI modes for 3D" disabled, so I'm not sure if this is the intended behavior.

3D modes supported are half-SBS, half-TAB and FP. The FP mode is enabled when "Use Full HD HDMI modes for 3D" is enabled.

Quote:I'm unsure if it wouldn't be still wise to have an option for the automatic 3D-FP detection/switching to be disabled and be allowed to manually set how the movie should be displayed. Helix did a good job with many legacy devices like mine, and now in some configurations I even lose quality when I have to stream an SBS video as TAB. This reduces the resolution vertically and horizontally, whereas with Helix, I could match the video format and the way the projector interpreted the video.

I'd like to get 3D robust. The update a couple of builds back did improve things for me. I couldn't find any issues when FP was disabled.
If you find any issues with that give a detailed report of what you did and what the outcome was. If projector is still in 3D mode when it shouldn't be,
report output of "tvservice -s" to confirm if we are still signalling 3D. A debug log may be useful too.

With FP enabled it seemed to behave fine for MVC files identified as SBS. I have seen blank screen, or partial screen with FP modes when source was TAB.
I need to investigate that more carefully. It's a bit tricky to debug, as I have no 3D capable TV where I can debug.
For half-SBS/half-TAB that's not too much of a problem as I know what the image should look like.
For FP I don't see anything when debugging so have to do an OE build (or wait for Milhouse's next one) to test on the 3D TV in the lounge which makes progress slower.
Accurate bug reports, with logs and possibly photos of the screen will help me identify problems more quickly.
Quick, potentially stupid question:

If I have full hdmi modes on, will a tab/sbs movie be sent framepacked? Or will it only send MVC (in MKV) FP with this setting on?
(2015-06-24, 13:48)nikc0069 Wrote: If I have full hdmi modes on, will a tab/sbs movie be sent framepacked? Or will it only send MVC (in MKV) FP with this setting on?

Everything. (We don't actually know at the time we change the HDMI mode whether it is MVC or not).
(2015-06-24, 12:48)gendo Wrote: OK would be great if we could keep it.. fixes issues with multiple pvr backends for me.. just did not want to introduce extra work on you and milhouse.

OK I'll reintroduce PR7020 however the level of discussion and lack of agreement in that PR makes me nervous...

It's for this same reason I dropped PR7291 - I'll reintroduce it once it's no longer a moving target.
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 Isengard build #0624: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.1.0 #1 Wed Jun 24 21:02:28 BST 2015 armv6l GNU/Linux

# vcgencmd version
Jun 23 2015 18:20:38
Copyright (c) 2012 Broadcom
version 144b1da894429d00352f650408d9b1a436302f7d (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20150624210140-#0624-gabcff54 [Build #0624]

# vcdbg log msg 2>&1 | grep DTOK
001555.883: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (abcff54b, changelog) and tip of XBMC master (95c0b93c, changelog) with the following modifications: Build Highlights:
  1. Updated libnfs (drop streaming cache patch)
Build Details:
  1. XBMC:
    • [libexif] Fix buffer overflow due to invalid GPS coordinates (PR:7163, 1 commit, 1 file changed)
  2. libnfs:
    • For upstream (PR:118, 9 commits, 7 files changed)
  3. Additional commits/pull requests/changes not yet merged upstream:
    • Added: PR:7020: [pvr] finally make it possible to use multiple PVR clients (r=1045)
    • Added: PR:7331: [addons] force repository update in post-install/uninstall
    • Added: PR:7338: [videodb] use GetNavCommon for tags
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've searched, but I haven't seen any reason why it stopped working, but is there any reason why I now see the Kodi splash screen (specifically the Kodi screen not the Kodi animation) after updating to a newer version (623)?

In my config file. I have the 'disable_splash=1' option, which worked until a more recent build version
You'd normally disable the Kodi splash by adding <splash>false</splash> to advancedsettings.xml - maybe your as.xml has gone AWOL?
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
  • 78
  • 79
  • 80(current)
  • 81
  • 82
  • 89

Logout Mark Read Team Forum Stats Members Help
OpenELEC Testbuilds for RaspberryPi (Kodi 15.0) Part 214