Dne 06. 04. 20 v 18:58 Hans de Goede napsal(a):
Hi,
On 4/6/20 5:17 PM, Pierre-Louis Bossart wrote:
So overall it seems rather unlikely that we can support dmic + headset mic in parallel?
Right that is what the ConflictingDevices sections are for.
Removing the JackHWMute does not allow using both add the same time,
it merely allows overriding which of the 2 to use (if the headset mic
is detected by the jack, the JackControl setting is kept).
So lets say the internal mic is crappy, but the user prefers to
have their ears free, then they can use a headset around there neck,
using just the headset-mic and select speakers as output even though
a jack is inserted.
Note pulseaudio will still automatically switch to the headphones /
headsetmic when plugged in, this merely allows overriding that
choice manually.
Jacks which actually physically disable (or re-route) signals are
quite rare now a days so having a JackHWMute in the UCM conf is
often wrong. After this patch only a few cases are left:
[hans@x1 alsa-ucm-conf]$ ack -l JackHWMute ucm2
ucm2/HDA-Intel/HiFi-dual.conf
ucm2/chtrt5650/HiFi.conf
ucm2/hda-dsp/HiFi.conf
ucm2/codecs/es8316/HeadPhones.conf
ucm2/sof-bdw-rt5677/HiFi.conf
ucm2/broxton-rt298/HiFi.conf
ucm2/skylake-rt286/HiFi.conf
The es8316 case is actually correct, the es8316 has only 1
output (which is amplified for e.g. a speaker) and at
least the reference design suggests routing the speaker signal
through the jack and using a jack which will interrupt the
signal when headphones are plugged in.
Note most actual designs implement this with an analog
switch and a GPIO to control that switch, since 3.5 mm
jacks which actually can disconnect a signal are rare
(if they exist at all, but I guess they do). For the
designs using the GPIO controlled analog switch, the
JackHWMute technically also is wrong, but I put it
there (IIRC it was me) because it matches the reference
design.
I'm not familiar enough with any of the others to say anything
about them, but most of them are for HDA setups, where this
is somewhat more normal I guess.
If I follow the logic, I guess we should also remove the JackHWMute for sof-bdw-rt5677, hda-dsp and HDA-Intel. I can't think of a hardware limitation preventing the headset mic from being used while the speakers are used for playback. Likewise the rt5650 is a cheaper version of rt5645 so the same applies.
Right AFAIK:
JackHWMute "foo"
Means that the JackControl listed in the same section does
something at the hardware level which makes it impossible
for the "foo" input / output to function until the jack
is removed. Typically this would be something like actually
disconnecting the speakers at the electrical level.
So almost always using it is wrong.
I agree. The description is really straight (use-case.h):
"Note that JackHWMute should be used only when the hardware enforces the
automatic muting."
Pierre, it seems that hds-dsp uses JackHWMute for Speaker and Headphones, I
think that it should be removed. Otherwise PA thinks that those devices cannot
be used simultaneously when user wants to..
Jaroslav
Regards,
Hans
--
Jaroslav Kysela <perex@xxxxxxxx>
Linux Sound Maintainer; ALSA Project; Red Hat, Inc.
