[pulseaudio-discuss] Ubuntu or PulseAudio Bug?

[Tanu Kaskinen]

On Mon, 2013-01-21 at 03:34 +0200, Tanu Kaskinen wrote:
> On Sun, 2013-01-20 at 14:13 -0800, Justin Chudgar wrote:
> > Following is one of the last attempts I made. I've experimented with
> > just the onboard IntelHDA and with just the ancient Ensonic PCI. I've
> > verified that the ladspa filter works using the 'applyfilter' and
> > 'analyzefilter' ladspa tools.
> > 
> > I have also let module-udev-detect find the card(s) and used pacmd to
> > add the filter manually with the same symptom of sound no longer coming
> > from the speakers.
> 
> Loading module-ladspa-sink with the parameters that you use resulted in
> the following:
> 
> D: [pulseaudio] module-ladspa-sink.c: Module: lowpass_iir_1891
> D: [pulseaudio] module-ladspa-sink.c: Label: lowpass_iir
> D: [pulseaudio] module-ladspa-sink.c: Unique ID: 1891
> D: [pulseaudio] module-ladspa-sink.c: Name: Glame Lowpass Filter
> D: [pulseaudio] module-ladspa-sink.c: Maker: Alexander Ehlert <mag at glame.de>
> D: [pulseaudio] module-ladspa-sink.c: Copyright: GPL
> D: [pulseaudio] module-ladspa-sink.c: Port 0 is control: Cutoff Frequency
> D: [pulseaudio] module-ladspa-sink.c: Port 1 is control: Stages(2 poles per stage)
> D: [pulseaudio] module-ladspa-sink.c: Port 2 is input: Input
> D: [pulseaudio] module-ladspa-sink.c: Port 3 is output: Output
> D: [pulseaudio] module-ladspa-sink.c: Will run 2 plugin instances
> D: [pulseaudio] module-ladspa-sink.c: Trying to read 2 control values
> D: [pulseaudio] module-ladspa-sink.c: cdata: '110,0.755'
> D: [pulseaudio] module-ladspa-sink.c: Read config value 110.000000 (p=0)
> D: [pulseaudio] module-ladspa-sink.c: Read config value 0.755000 (p=1)
> W: [pulseaudio] module-ladspa-sink.c: Control value 1 below lower bound: 0.755000 (lower bound: 1.000000)
> E: [pulseaudio] module-ladspa-sink.c: Failed to parse, validate or set control parameters
> Segmentation fault
> 
> So, the parameter for "Stages" seems to be invalid, and that then
> triggers some bug in pulseaudio.

The segfault seems to be due to a bug in the lowpass filter: the cleanup
function assumes that the activate function has been called. That
doesn't happen here, because pulseaudio tries to unload the filter
before the activate function has been called.

CC'ing Steve Harris, who I believe is the author of the ladspa plugin in
question.

-- 
Tanu