[pulseaudio-discuss] [alsa-devel] pulseaudio fails to start with kernel 4.11, caused by new snd_hdmi_lpe_audio module)

[Takashi Iwai]

On Wed, 29 Mar 2017 16:40:28 +0200,
Tanu Kaskinen wrote:
> 
> On Wed, 2017-03-29 at 15:26 +0200, Takashi Iwai wrote:
> > On Wed, 29 Mar 2017 15:14:19 +0200,
> > Tanu Kaskinen wrote:
> > > 
> > > On Wed, 2017-03-29 at 15:06 +0200, Takashi Iwai wrote:
> > > > On Wed, 29 Mar 2017 14:59:45 +0200,
> > > > Tanu Kaskinen wrote:
> > > > > 
> > > > > On Wed, 2017-03-29 at 07:21 +0200, Takashi Iwai wrote:
> > > > > > Does PA really need to start streaming at its invocation time?
> > > > > > The crash happens when PA gets started via desktop autostart, and at
> > > > > > this moment, the HDMI graphics state is possible disconnected before
> > > > > > xrandr setup.  The HDMI connection state should have been informed /
> > > > > > notified to PA via the jack interface, so it should be possible to
> > > > > > judge beforehand.
> > > > > > 
> > > > > > Or it might be something wrong in the driver side regarding the jack
> > > > > > state processing?
> > > > > 
> > > > > I had a closer look at the PA log that was posted earlier. It looks
> > > > > like the device numbering is non-standard. Trying to open any of the
> > > > > hdmi:1,x devices fails. hw:1 can be opened, and pulseaudio assumes that
> > > > > it's an analog stereo device. Jack detection isn't going to work in
> > > > > this situation, because pulseaudio doesn't know that it should be
> > > > > looking at the hdmi jacks.
> > > > > 
> > > > > Can the driver be fixed to use the standard HDMI device numbers?
> > > > 
> > > > Hmm, it might be the missing hdmi PCM definition?
> > > > 
> > > > The latest alsa-lib git already contains the card config for HDMI LPE
> > > > audio, and this might help.  (Though, I thought I still saw the same
> > > > PA problem even with the card config.  Unfortunately I can't access
> > > > the box with the DP audio right now, so others may help more quickly
> > > > than me...)
> > > > 
> > > > Can anyone confirm that you have the latest alsa-lib git installed and
> > > > whether the PA issue is fixed or not?
> > > 
> > > If the alsa-lib configuration just maps hdmi:1,0 to hw:1,0, then I
> > > would expect this problem to persist. If hw:1,0 can be opened,
> > > pulseaudio will think that there's an analog stereo output. If hdmi:1,0
> > > works too, then pulseaudio will think that there are two separate
> > > outputs, and if the hdmi jack state says that hdmi is not available,
> > > pulseaudio will use what it thinks is an analog output.
> > 
> > Aha, that explains.
> > 
> > > If it's not possible to change the device numbering in the driver, this
> > > will have to be worked around in pulseaudio somehow (pulseaudio
> > > shouldn't try to use hw:1,0 on this hardware).
> > 
> > Well, it's not impossible to change in the driver side, but I hesitate
> > to do so, since it's not right, IMO.
> > 
> > In general, the PCM device#0 isn't always for an analog output,
> > although it's de facto standard, as most boards have both analog and
> > HDMI/DP.  For HD-audio, we even kept the constant PCM dev# even if no
> > analog output is present; but it's just to make the configuration
> > easier, and it doesn't mean that PCM dev#0 is for the analog I/O.
> 
> I agree, it's not good to assume that hw:x,0 is always analog stereo
> output. Pulseaudio should fall back to hw: only if nothing else works.
> 
> The jack name problem is easily solved in pulseaudio, but there's one
> more problem related to the device numbers: pulseaudio needs to know
> which ELD element corresponds to which PCM device. Currently the
> mapping is hardcoded: hdmi:x,0 is expected to always correspond to the
> ELD element of device 3. Is there any way pulseaudio could query the
> mapping instead of hardcoding it?

Oh that's ugly.  This mess comes from the HD-audio configuration, my
bad.

The ELD index number is aligned with the PCM device number (in a way
of hw:x,y), thus it's 3 for HD-audio.  It's used for HDMI/DP jack,
too.

I guess you can get this from snd_pcm_info() =>
snd_pcm_info_get_device() call for the opened stream?
The usual plugins should return the slsave pcm info, so even for
hdmi:x,y, it should reach to the underlying hardware PCM.


Takashi