On Sun, 2017-10-08 at 19:48 +0300, Tanu Kaskinen wrote: > The Intel HDMI LPE driver uses hw:0 for HDMI, which is different from > the HDA device numbering scheme. Our jack detection and ELD code assumed > the HDA numbering, these patches remove the assumption. > > If PulseAudio tries to play audio while the monitor is not plugged in, > the HDMI LPE driver will refuse, and the error handling in PulseAudio > will enter an infinite loop, causing the kernel to kill PulseAudio. It > would be good to fix the error handling code, but these patches just > avoid the problem by implementing working jack detection, so PulseAudio > won't try to use HDMI when it's unplugged. I forgot to write what's new in v2. I simplified the first patch so that we always query the hw device, even though it's unclear what index ALSA returns for devices that aren't associated with exactly one hw device. As long as we use the index only with HDMI, this shouldn't be a problem. The simplification allows us to avoid adding the "query-hw- device" option in the mapping configuration. -- Tanu https://www.patreon.com/tanuk
