Bug#805414: gdm3: disable pulseaudio to prevent capturing A2DP sink on session start

[Felipe Sateler]

Adding the a11y list to CC.

Dear a11y team, a short summary for you: Currently, pulseaudio does not
negotiate bluetooth devices the same way it does with alsa devices, with a
result that is undesirable: if the gdm login screen grabs the BT device, it
becomes unavailable in the logged in session. The change being discussed
here is to disable bluetooth in the gdm login screen.

On Mon, Jul 3, 2017 at 10:20 AM, Antoine Beaupré <anarcat at debian.org> wrote:

> On 2017-07-03 15:46:20, Michael Biebl wrote:
> > Am 03.07.2017 um 00:05 schrieb Antoine Beaupré:
> >> On 2017-07-02 23:43:55, Michael Biebl wrote:
> >>> Am 02.07.2017 um 23:36 schrieb Antoine Beaupré:
> >>>> On 2017-07-02 23:16:19, Michael Biebl wrote:
> >>>
> >>>>> Have you tested the workaround from the arch wiki and can you
> confirm it
> >>>>> works?
> >>>>
> >>>> I cannot, unfortunately, test this anymore, as I have disabled the
> >>>> pulseaudio socket as directed earlier, with:
> >>>>
> >>>> rm /var/lib/gdm3/.config/systemd/user/sockets.target.wants/
> pulseaudio.socket
> >>>>
> >>>> I don't know what that was pointing to, so I can't quite restore that
> >>>> behavior directly.
> >>>
> >>> That's from the postinst:
> >>>
> >>> UNIT=/usr/lib/systemd/user/pulseaudio.socket
> >>> USERUNITDIR=/var/lib/gdm3/.config/systemd/user
> >>> if ! [ -L $USERUNITDIR/sockets.target.wants/pulseaudio.socket ]; then
> >>>   mkdir -p $USERUNITDIR/sockets.target.wants
> >>>   ln -sf $UNIT $USERUNITDIR/sockets.target.wants
> >>> fi
> >>>
> >>> You can run that manually or simply re-install the gdm3 package so that
> >>> code is run again.
> >>
> >> Thanks! I believe I have rolled back the workarounds and implemented the
> >> fix default.pa documented in the Arch wiki. Everything seems to work
> >> normally now, ie. I can correctly connect through A2DP to my bluetooth
> >> speaker.
> >
> > Since concerns were raised that this might break existing a11y setups, I
> > don't plan to make an upload with this change.
>
> I don't understand the impact this could have regarding
> accessibility. There are, as far as I know, currently no usable way to
> have bluetooth audio working at all in the gdm3 login prompt, because
> there's no way to associate a device, so there's no loss of
> functionality here. But there *is* a loss of functionality for
> everyone (*including* screen reader users) *after* the user is logged
> in: they cannot correctly configure bluetooth devices!
>

Bluetooth pairings are global. Once you configured them in your user
session, they will become available to gdm too.


>
> Can we get a better idea of the use case you are worried about here?
>

Here is where the input of the a11y team would be very appreciated. Is
disabling bluetooth in the login screen likely to cause problem?


>
> To be really clear here, the change I am proposing is the one documented
> in the Arch wiki and now also in the Debian wiki:
>
> https://wiki.debian.org/BluetoothUser/a2dp#Workaround_
> 2:_disable_pulseaudio.27s_bluetooth_in_gdm
>
> which currently consists of adding the following
> /var/lib/gdm3/.config/pulse/default.pa file:
>
> #!/usr/bin/pulseaudio -nF
> #
>
> # load system wide configuration
> .include /etc/pulse/default.pa
>
> ### unload driver modules for Bluetooth hardware
> .ifexists module-bluetooth-policy.so
>   unload-module module-bluetooth-policy
> .endif
>
> .ifexists module-bluetooth-discover.so
>   unload-module module-bluetooth-discover
> .endif
>
> How would this break accessibility?
>

My own thought is that if your only speakers are BT, how are you going to
have the screen read if there is no BT?

-- 

Saludos,
Felipe Sateler
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/pkg-gnome-maintainers/attachments/20170703/1fa146c9/attachment-0001.html>