omx audio_decoder hangs on EmptyThisBuffer() in kernel when 2 omx audio_render used

[stupid-boy]

more info in this thread:
http://www.raspberrypi.org/phpBB3/viewtopic.php?f=31&t=27908

[popcornmix]

Can you provide a simple test app that causes the hang?
It may be easier to debug using something based on hello_audio, or command line omxplayer.

[stupid-boy]

no, just XBMC code, currently flooded with logs.
week ago i have free time and was able to work n this/extract code/produce 'simple' app, now i am back to work.

[stupid-boy]

Bellow are stack traces of 2 of threads when EmptyThisBuffer() hangs. Remaining threads are not related to libopenmaxil.so. Hope this will help any way. If i can collect some more info, please tell me.

Thread 20 (Thread 0xab90c440 (LWP 2669)):
#0 0xb6473700 in sem_wait@@GLIBC_2.4 () from /lib/arm-linux-gnueabihf/libpthread.so.0
#1 0xb6fc1ae0 in ilcs_execute_function_ex () from /opt/vc/lib/libopenmaxil.so
#2 0xb6fc2588 in ilcs_pass_buffer () from /opt/vc/lib/libopenmaxil.so
#3 0x00678478 in COMXCoreComponent::EmptyThisBuffer (this=0xac416170, omx_buffer=0x17811e0) at OMXCore.cpp:448
#4 0x012648d4 in COMXAudio::AddPackets (this=0xac413a70, data=0xab90d010, len=24576, dts=32000, pts=32000) at OMXAudio.cpp:1062
#5 0x0126e574 in OMXPlayerAudio::Decode (this=0xac4136e0, pkt=0x1b0b658, bDropPacket=false) at OMXPlayerAudio.cpp:392
#6 0x0126ec2c in OMXPlayerAudio::Process (this=0xac4136e0) at OMXPlayerAudio.cpp:505
#7 0x0127cd24 in CThread::Action (this=0xac4136e0) at Thread.cpp:218
#8 0x0127c91c in CThread::staticThread (data=0xac4136e0) at Thread.cpp:128
#9 0xb646cbfc in start_thread () from /lib/arm-linux-gnueabihf/libpthread.so.0
#10 0xb4a33758 in ?? () from /lib/arm-linux-gnueabihf/libc.so.6
#11 0xb4a33758 in ?? () from /lib/arm-linux-gnueabihf/libc.so.6
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

Thread 19 (Thread 0xac1c8440 (LWP 2668)):
#0 0xb6473700 in sem_wait@@GLIBC_2.4 () from /lib/arm-linux-gnueabihf/libpthread.so.0
#1 0xb6fc1ae0 in ilcs_execute_function_ex () from /opt/vc/lib/libopenmaxil.so
#2 0xb6fc2320 in ilcs_execute_function () from /opt/vc/lib/libopenmaxil.so
#3 0xb6fc299c in vcil_out_get () from /opt/vc/lib/libopenmaxil.so
#4 0x0f442e80 in ?? ()
#5 0x0f442e80 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)

EDIT:
I pushed my XBMC code. Can be found here: https://github.com/stupid-boy/xbmc/commits/RPI_both_audio

[tthef]

I suspect this might be the same/similar problem as this gst-omx issue, http://www.raspberrypi.org/phpBB3/viewtopic.php?f=38&t=19606 (that discussion has a reference to an earlier firmware that is supposed to work, but unfortunately, the commit hash no longer exists in the repo).

[stupid-boy]

fixed at commit from Feb 07.
thanks a lot!

[popcornmix]

Yes, apologies for this not working. Using hdmi and analogue together is quite rare, so I hadn't realised the GPU was one DMA channel short.
I found it with an aplay to analogue, and an omxplayer to hdmi, but I thought it might fix your issue.

[stupid-boy]

no problem.
everything is fine when it finish fine.