On 05.05.2017 16:43, Georg Chini wrote: > On 03.05.2017 12:51, Tanu Kaskinen wrote: >> On Wed, 2017-05-03 at 11:23 +0800, Hui Wang wrote: >>> Today I tested the latest master branch (already including your two >>> patches), seems it did not fix the problem. Below is my testing detail: >>> >>> Install ubuntu 16.04 on my laptop. >>> clone the latest pulseaudio source code, checkout master branch >>> build the source code: ./autogen.sh;make >>> kill the existing pulseaudio and run the latest pulseaudio: pulseaudio >>> --kill && ./src/pulseaudio -vvvnF ./src/default.pa -p $(pwd)/src >>> >>> On this machine, we have sound cards, sinks and ports like below: >>> >>> Card#0(HDA INTEL HDMI)-> Sink#0(hdmi-stereo)->hdmi-output(priority: >>> 5900) >>> Card#1(HDA INTEL PCH)->Sink#1(analog-stereo)->headphones(priority: >>> 9000) >> No speaker port at all on the PCH card? >> >>> with the new pulseaudio, I can only see one sink (analog-stereo) >>> through pactl list sinks at the beginning. >>> >>> After I plug the HDMI cable (headphone is not plugged), I can see two >>> sinks via pactl list sinks, but the default sink is still the analog- >>> stereo, here I expect the default sink is hdmi-stereo since the >>> headphone is not plugged yet. >> That's what I'd expect too. Is the headphone port marked as unavailable >> as it should? >> >>> Then I plug the headphone, and I select the hdmi (hdmi is also >>> plugged in) as the default sink, after a while I unplug the hdmi >>> cable, here I expect the default sink is back to analog-stereo since >>> headphone is still plugged, but the default sink did not switch to >>> analog-stereo. BTW after I unplug the hdmi cable, I can still see two >>> sinks. >> The default sink is set in pa_core_update_default_sink() in >> src/pulsecore/core.c. That function uses the compare_sinks() function, >> and the first thing compare_sinks() does is to check if one of the >> compared sinks has an unavailable port active. Can you figure out what >> goes wrong? Is pa_core_update_default_sink() not called when you plug >> in or unplug things, or does pa_core_update_default_sink() really >> choose a sink whose active port is unavailable? >> > Looking through the patch it seems to me that availability changes of > a port > do not trigger a recalculation of the default sink. Or am I missing > something? Sorry, I was wrong. The actual commit is slightly different from the patch in patchwork and includes the necessary calls.
