• 1
  • 138
  • 139
  • 140(current)
  • 141
  • 142
  • 168
OpenELEC Testbuilds for RaspberryPi (Kodi 16.0)
(2015-11-14, 14:34)popcornmix Wrote: This issue was commented on here.
It's a problem with I2S driver and 4.3 kernel. It occurs when driver is closed. Does setting "keep audio alive" to "always" avoid this issue?
I don't think the kernel backtrace is related to the device not showing up in kodi.

Something strange seems to be going on with ALSA device enumeration. The bcm2835 alsa driver (dtparam=audio=on in config.txt) also doesn't show up in kodi in builds 1111b and newer.

In 1110 everything's fine.
aplay -l: http://sprunge.us/ghJT
dmesg: http://sprunge.us/eKcX
kodi.log: http://sprunge.us/SCDA

In 1111b bcm2835 ALSA works fine from the command line (using aplay file.wav) but doesn't show up in kodi.
aplay -l: http://sprunge.us/eLOP
dmesg: http://sprunge.us/HCLJ
kodi.log: http://sprunge.us/CLWe

But an USB audio device shows up in kodi with the 1111b build.
aplay -l: http://sprunge.us/iFCb
dmesg: http://sprunge.us/bWHH
kodi.log: http://sprunge.us/DNMH

so long,

Hias
(2015-11-14, 16:56)HiassofT Wrote: Something strange seems to be going on with ALSA device enumeration. The bcm2835 alsa driver (dtparam=audio=on in config.txt) also doesn't show up in kodi in builds 1111b and newer.

In 1110 everything's fine.
aplay -l: http://sprunge.us/ghJT
dmesg: http://sprunge.us/eKcX
kodi.log: http://sprunge.us/SCDA

In 1111b bcm2835 ALSA works fine from the command line (using aplay file.wav) but doesn't show up in kodi.
aplay -l: http://sprunge.us/eLOP
dmesg: http://sprunge.us/HCLJ
kodi.log: http://sprunge.us/CLWe

dtparam=audio=on is working okay for me in raspbian dev environment (with latest firmware and latest 4.3.y_rebase kernel).
Can you test 1111b image with start.elf/fixup.dat from 1110?
(2015-11-14, 17:26)popcornmix Wrote: dtparam=audio=on is working okay for me in raspbian dev environment (with latest firmware and latest 4.3.y_rebase kernel).
Can you test 1111b image with start.elf/fixup.dat from 1110?
That didn't make a difference, bcm2835 ALSA still not showing up in kodi.

# vcgencmd version
Code:
Nov  6 2015 21:55:08
Copyright (c) 2012 Broadcom
version 40eee6c6d44edb1db4302217927a66a8dffd9374 (clean) (release)
aplay -l: http://sprunge.us/OWNW
dmesg: http://sprunge.us/SLQE
kodi.log: http://sprunge.us/LFXX

so long,

Hias
(2015-11-14, 20:29)HiassofT Wrote:
(2015-11-14, 17:26)popcornmix Wrote: dtparam=audio=on is working okay for me in raspbian dev environment (with latest firmware and latest 4.3.y_rebase kernel).
Can you test 1111b image with start.elf/fixup.dat from 1110?
That didn't make a difference, bcm2835 ALSA still not showing up in kodi.

@Milhouse any thoughts? Seems the only change left is "Dropped ALSA/Pulse bridge and BT commits, meaning there is no PulseAudio in this build".
I did some more testing with earlier builds

In 1102 bcm2835 ALSA showed up in kodi
in 1103 it didn't show up anymore
in 1107 it showed up again

Could pulse audio have masked/fixed a bug introduced in 1103? Maybe something in PR:4411 ?
(2015-11-14, 21:28)HiassofT Wrote: I did some more testing with earlier builds

In 1102 bcm2835 ALSA showed up in kodi
in 1103 it didn't show up anymore
in 1107 it showed up again

Could pulse audio have masked/fixed a bug introduced in 1103? Maybe something in PR:4411 ?

Sounds plausible This line looks suspicious:
https://github.com/lrusak/OpenELEC.tv/co...72e7516R13
(2015-11-14, 20:36)popcornmix Wrote: @Milhouse any thoughts? Seems the only change left is "Dropped ALSA/Pulse bridge and BT commits, meaning there is no PulseAudio in this build".

None other than PulseAudio is masking the i2s problem. I'll re-introduce PulseAudio (as it was in #1110) in tonight's build to see if this stops the i2s backtraces.

We introduced PulseAudio in #1107 before the i2s patches in #1109, so the fact the i2s patches were working without a problem may have been a side effect of the earlier Pulse Audio change. Now we've dropped PA, the i2s driver has fallen apart. Or maybe it's a coincidence, as I'm not seeing anything else that changed in #1111b that comes near to explaining this "new" i2s behaviour.
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.
(2015-11-14, 21:32)popcornmix Wrote: Sounds plausible This line looks suspicious:
https://github.com/lrusak/OpenELEC.tv/co...72e7516R13

I can revert PR:4411 in tonight's build (leaving PA out of the build, as it is in #1111b) and we can see what effect that has?
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 #1114: RPi / RPi2
(Supercedes previous build)

Code:
# uname -a
Linux rpi512 4.3.0 #1 Sat Nov 14 21:02:41 GMT 2015 armv6l GNU/Linux

# vcgencmd version
Nov 11 2015 21:36:20
Copyright (c) 2012 Broadcom
version 54011a8ad59a9ae1c40bd07cddd9bcf90e779b66 (clean) (release)

# lsb_release
OpenELEC (Milhouse) - Version: devel-20151114210148-#1114-g60c4d40 [Build #1114]

# vcdbg log msg 2>&1 | grep DTOK
001693.536: Kernel trailer DTOK property says yes

# Kernel device tree status: Enabled

Based on tip of OpenELEC master (60c4d40d, changelog) and tip of XBMC master (68775348, changelog) with the following modifications: Build Highlights:
  1. Revert PR:4411 to test if PulseAudio/i2s issue is related
Build Details:
  1. XBMC:
    • FIX: [amc] skip frame render when changing surface size (PR:8382, 1 commit, 1 file changed)
    • FIX: [droid] Only sw for SD MPEG (PR:8383, 1 commit, 1 file changed)
    • [amcodec] partialy revert 344af8f (PR:8384, 1 commit, 1 file changed)
  2. pvr.hts:
    • Remove actual file when lifetime expires (PR:125, 2 commits, 8 files changed)
  3. Additional commits/pull requests/changes not yet merged upstream:
    • Reverted: [env] PR:4411: various pulseaudio fixes (try reverting, see if fixes i2s issues)
    • Reverted: [pkg] PR:8382: FIX: [amc] skip frame render when changing surface size (Reason: Requires VideoPlayer rebase)
    • Reverted: [pkg] PR:8383: FIX: [droid] Only sw for SD MPEG (Reason: Requires VideoPlayer rebase)
    • Reverted: [pkg] PR:8384: [amcodec] partialy revert 344af8f (Reason: Requires VideoPlayer rebase)
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.
Those with the i2s/Hifiberry audio problems please test #1114 and report results, post dmesg. Thanks.
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.
(2015-11-15, 00:45)Milhouse Wrote: Those with the i2s/Hifiberry audio problems please test #1114 and report results, post dmesg. Thanks.

i2s/Hifiberry now working OK again on #1114.

dmesg http://sprunge.us/NGRf
(2015-11-15, 01:01)beertank Wrote: i2s/Hifiberry now working OK again on #1114.

dmesg http://sprunge.us/NGRf

Many thanks. Smile

@popcornmix/@HiassofT: So is this a bug in the i2s driver (that is triggered by PR:4411), or is it a bug only in PR:4411? Or is there a problem in both the driver and PR:4411?
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.
Somehow, since the last 2 or 3 updates, the are WAY less skins compatible - any1 else noticed thatHuh
I still have skip counter when osd or video codec info screen is up when using mmal advanced or autoselect deinterlace mode.

i have a debug log if it is necessery
(2015-11-15, 11:47)FattyMcDirty Wrote: Somehow, since the last 2 or 3 updates, the are WAY less skins compatible - any1 else noticed thatHuh

Incompatible in what way, exactly? Skin backward compatibility was removed in build #11101.
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
  • 138
  • 139
  • 140(current)
  • 141
  • 142
  • 168

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