Comment # 23 on bug 1112292 from Takashi Iwai

(In reply to Daniel Pecka from comment #22)
> (In reply to Takashi Iwai from comment #21)
> > The change is on purpose, and any rest issues (e.g. kmix) must be an
> > application bug.  Please report to each one that appears broken.
> > 
> > The volume map?  That's just a bogus value before the actual deployment of
> > UCM profile.  All these incorrect information have to be wiped out, really...
> 
> I am annoyed enough that I'm writing for you analysis .. what annoys me even
> more is, that you DID'T BOTHER to read logs I've attached :( .. and
> everything is there

Thanks for the information update, now finally I have something that
made me understanding the issue.

A log can't speak by itself, it's merely an evidence; hence it alone can't help
unless you give a proper explanation how to interpret.
So at the next time, PLEASE give the proper explanation at first before giving
any data.  Otherwise it would make little sense.

Now, let's dive into details:

> 1) profiles are gone (see list-cards and profiles: section)

Do you mean the profiles are "changed"?  In the former case, it had
lots of bogus profiles (IEC958 on this device? no way), and they are
corrected.  The "change" of the profile itself shouldn't be a big
matter, but... let's go next.

> 2) you (OMG) hardwired via config file 3 ports and sources that are not
> there (see list-cards and ports section)

Hmm, "ports" subsection of index#1 contains three ports, two [Out] and
one [In].

> 3) you hardwired via your config that two sinks are used altogether - guess
> what it does haha (see section sinks: in list-cards)

These sinks can be chosen individually by PA, too.  It's nothing
special in comparison with other UCM profiles.

> 4) you've broke icon(s)

Do you mean the device.icon_name for each port?
Then I see the point, which should have been handled better in PA
side...

> 5) now in list-sinks are multiple outputs with broken priorities, result ==
> volume doesn't work properly, volume is not properly restored

OK, that can be added in the UCM profile.  This can be easily
corrected.

> 6) in list-sources are broken priorities, incorrect hardware flags (so mute
> button + volume fn+Fx buttons doesn't work properly) ..

Another PA issue.  PA UCM parser doesn't check the hardware volume /
mute control even if UCM profile gives.

Aactually, the upstream fix with a UCM profile in ALSA side is a correct thing,
per se,
but its interpretation in PA side doesn't look ideal, unfortunately.
At best, we may fix PA to deal with UCM better, but a shorter path is
to cherry-pick the recent change in the upstream PA for the Dell dock
profile.

I'll try to bring up the updated PA (that follows PA upstream) so that we can
revert the WD15 Dock UCM profile gracefully.