[regression] Bluetooth audio no longer supports A2DP (stuck in HSP/HFP mode)

[Dustin Kirkland ]

[Impact]
Bluetooth audio seemingly refuses to support A2DP (stuck in HSP/HFP mode). A2DP is the high quality profile used for Bluetooth speakers and music in general.

[Test Case]
1. Pair a Bluetooth audio device with Ubuntu.
2. Go to Sound settings and check the device's current "Mode".
3. If the current Mode is not yet A2DP then try to set it to A2DP
Expected: The mode is either A2DP by default or can be set to it.

[Regression Potential]
Low. The SRU focuses on Bluetooth logic in pulseaudio, so that is the main area affected. A combination of related A2DP bugs in pulseaudio in xenial means it is difficult to get working at all without this patch. So highly unlikely Bluetooth audio support could get worse. Since pulseaudio itself is being modified there is always a regression potential in support for other audio devices, but several audio devices have been tested and all continue to work well with the patch.

[Other notes]
Related to bug 1283003 and bug 1582213. Possibly others too. The debdiff which fixes this is attached to bug 1582213.

Please note that Bluetooth and Bluetooth audio support in xenial is still not perfect. This SRU only aims to address a few of the most troublesome issues. Please consider the fact that this is an incremental improvement and some people are still likely to experience some bugs related to pulseaudio and Bluetooth, even after this SRU.

The patch was authored by Luke Yelavich with help from Konrad Zapałowicz. And it has received further testing over the past two months by Konrad, Jim Hodapp and Daniel van Vugt. The debdiff proposed here only changes the changelog wording from Luke's original PPA.

[Original Description]
Just installed 15.04 fresh from the latest ISO (beta2).

I'm bummed to see my bluetooth headset (Bose Soundlink overear) seems to have regressed in functionality.

In 14.10, I was able to set the output profile either to a2dp or hsp/hfp (telephony duplex).

In 15.04, it only works in telephony duplex mode. I can't get high fidelity sound playback to work at all.

This thread seems to be related, though the workaround within did not solve the problem for me:
https://bbs.archlinux.org/viewtopic.php?id=194006

The bug is still present in 16.04 LTS and 16.10.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Gareth Trainor (gtredx69)]

I am trying to connect my Beats pill which I can, but the sound is set to telephony. I have spent 4 hours trying to enable a2dp with no luck. In pulse audio, the device appears can select the high quality setting but nothing changes.

[James C. Davis (jamescdavis)]

I was able to get a2dp working in 15.04 by editing /etc/bluetooth/audio.conf and uncommenting AutoConnect=true under [General] then restarting the bluetooth service with: systemctl restart bluetooth.service
After doing this, my bluetooth headset connected using the a2dp profile automatically.

[RobertFM (robert123m)]

I had the same problem and for several months I have tried all possible solutions I could find to no avail,... until today!

I found the solution in post #17 in the following https://bbs.archlinux.org/viewtopic.php?pid=1526534#p1526534 (tnx SimFox3)

What I did:

# First copy the files I edited

    sudo cp /etc/bluetooth/audio.conf /etc/bluetooth/audio.conf_tmp
    sudo cp /etc/pulse/default.pa /etc/pulse/default.pa_tmp
    sudo cp /usr/bin/start-pulseaudio-x11 /usr/bin/start-pulseaudio-x11_tmp

#Edit /etc/bluetooth/audio.conf

    sudo nano /etc/bluetooth/audio.conf

This is in my audio.conf

    # Configuration file for the audio service
    # This section contains options which are not specific to any
    # particular interface
    [General]
    Enable=Gateway,Source

    # Switch to master role for incoming connections (defaults to true)
    Master=true

    # If we want to disable support for specific services
    # Defaults to supporting all implemented services
    #Disable=Gateway,Source,Socket
    Disable=Socket

    # SCO routing. Either PCM or HCI (in which case audio is routed to/from ALSA)
    # Defaults to HCI
    #SCORouting=HCI

    # Automatically connect both A2DP and HFP/HSP profiles for incoming
    # connections. Some headsets that support both profiles will only connect the
    # other one automatically so the default setting of true is usually a good
    # idea.
    AutoConnect=true

    # Headset interface specific options (i.e. options which affect how the audio
    # service interacts with remote headset devices)
    [Headset]

    # Set to true to support HFP, false means only HSP is supported
    # Defaults to true
    HFP=true

    # Maximum number of connected HSP/HFP devices per adapter. Defaults to 1
    MaxConnected=2

    # Set to true to enable use of fast connectable mode (faster page scanning)
    # for HFP when incoming call starts. Default settings are restored after
    # call is answered or rejected. Page scan interval is much shorter and page
    # scan type changed to interlaced. Such allows faster connection initiated
    # by a headset.
    FastConnectable=true

    # Just an example of potential config options for the other interfaces
    #[A2DP]
    #SBCSources=1
    #MPEG12Sources=0

#Edit /etc/pulse/default.pa

    sudo nano /etc/pulse/default.pa

comment out (with an # at the beginning of the line) the following line

    #load-module module-bluetooth-discover

#now edit /usr/bin/start-pulseaudio-x11

    sudo nano /usr/bin/start-pulseaudio-x11

and after the lines

    if [ x”$SESSION_MANAGER” != x ] ; then
            /usr/bin/pactl load-module module-x11-xsmp “display=$DISPLAY session_manager=$SESSION_MANAGER” > /dev/null
    fi

add the following

    /usr/bin/pactl load-module module-bluetooth-discover

#Restarted and I was finally able tot switch to A2DP

Hope it works for you too.

[Yura (ykuchinskiy)]

confirm fix from post #17, ubuntu 15.04

[Yura (ykuchinskiy)]

sorry for the typo, confirm fix from post #5, ubuntu 15.04

[felipunk (fenieves)]

I will also like to confirm the solution from RobertFM on Post #5, Worked flawlessly.

[Luigi Maselli (grigio)]

I see the profile AD2P sink but the audio is still mono, the instructions in 5 are outdated, some filenams are changed

[Horacio Arroyo (deblike)]

Confirm fix from post #5 is still working.

[Horacio Arroyo (deblike)]

Just applied to a fresh 16.04 beta 2 install, despite having to create /etc/bluetooth/audio.conf manually, it worked again for me.

[Richard Sellam (richard-sellam)]

I have the exact same bug on up to date 16.04 beta (cannot connect to a2dp sink but hsp working), but i didn't have the bug before with 15.10...

If it can help, here is the message showing in syslog when autoconnecting to the bluetooth speaker:
Apr 2 10:44:39 vaio pulseaudio[1890]: [pulseaudio] bluez5-util.c: Transport Acquire() failed for transport /org/bluez/hci0/dev_A0_E9_DB_00_E2_4D/fd1 (Operation Not Authorized)

I can confirm that is it, at least for me, related to https://bugs.freedesktop.org/show_bug.cgi?id=92102 , and that manually connecting to the speaker with "connect A0:E9:DB:00:E2:4D" with bluetoothctl make the a2dp sink available, while just connecting the speaker with the bluetooth applet didn't. No file modification required.

Note that sometimes it fails to connect even with bluetoothctl, but trying once again the connect command connects successfully.

Regards,

[Dariusz Płoszaj (zakapior1983)]

I have the same problem in Ubuntu 16.04 beta 2...I have JBL bluetooth speaker and can't change quality of sound from HSP to A2DP....

[kylea (kylea)]

confirm fix from post #5, ubuntu 16.04 beta 2 Ubuntu Gnome.

Thanks

[Ryan Frank (ryandlf)]

I'd also like to confirm fix from post #5. Ubuntu 14.04

[Karl Grzeszczak (karlgrz)]

I can confirm that ubuntu 16.04 LTS is exhibiting this behavior with three separate bluetooth headsets (JABRA REVO 2.0, Sony MDRXB950BT, and Sennheiser Urbanite XL WIreless).

14.04, 14.10, 15.04, and 15.10 all worked quite well with no additional configuration necessary. By "worked quite well" I mean I could pair all three headsets, they would remain paired through reboots, and I could get A2DP output from each.

On 16.04, in both gnome and unity, it is a hit or miss affair of pairing any of these headsets. I have a Dell XPS 15 and had zero problems prior to installing 16.04.

I used the procedure in post #5, but that does not persist after reboots. I have to unpair the headset completely and then repair in order to get sound. Even then it's hit or miss, sometimes it doesn't work.

I'd be happy to give more details to help resolve this.

[Karl Grzeszczak (karlgrz)]

Update to this, which might help in diagnosing, might not.

I can ensure that the bluetooth A2DP connection will successfully deliver audio to my headset consistently if I switch back to "Headset Head Unit (HSP/HFP)" mode in Sound settings prior to disconnecting the headset.

Then, next time I connect the headset, it will be in "Headset Head Unit (HSP/HFP)" mode, but when I switch to "High Fidelity Playback (A2DP Sink)" every single time it successfully plays back audio.

[Linux Lurker (mark-edcc)]

@karlgrz

I saw the issues you were having and had recently just figured out the same work-around. Just as I was going to post it for you, I see that you too have narrowed it down to this repeatable behavior.

For me however, I do not switch back to HSP, rather I turn the profile OFF in pavucontrol "configuration" tab. Then I disconnect the bluetooth using the its graphic interface.
The next time I connect the bluetooth device, it connects just fine. Then I simply go to the config tab in pavucontrol to switch the device to A2DP sink. It works every time this way.

In the past, switching the profile to OFF in pavucontrol prior to disconnecting the bluetooth device was unnecessary.
I am currently running Kubuntu 16.04 LTS, but have had 16.04 installed since beta. This regression was not present for me in any previous versions.

[Karl Grzeszczak (karlgrz)]

Another update. I had accidentally installed gnome-tweak-tool on my unity install instead of unity-tweak-tool, which pulled in a bunch of gnome-core stuff. After uninstalling the gnome stuff and reverting my configs to default (i.e. removing everything I did in post #5) my bluetooth headphones seem to be connecting properly again.

[Karl Grzeszczak (karlgrz)]

@mark-edcc thanks for the update. I'll try that in the future if necessary, but I really feel like the gnome nonsense caused all this. When I tried the 16.04 beta, it was the gnome flavor, and I experienced the same nonsense on arch linux with gnome. Never had this problem with unity. Weird.

Thanks again.

[Karl Grzeszczak (karlgrz)]

Hah, removing the gnome packages and reverting the configs did not resolve the issue. Still have to remove the bluetooth device and repair upon any reboot or resuming from sleep / hibernate. What a pain in the ass.

[Max Bowsher (maxb)]

I'm not sure if this helps, but I see pulseaudio logging an "Operation Not Authorized" message when things do not work:

Apr 26 01:46:52 zenbook pulseaudio[3346]: [pulseaudio] module-bluez5-device.c: Acquiring transport /org/bluez/hci0/dev_20_74_CF_02_DB_F2/fd0
Apr 26 01:46:52 zenbook pulseaudio[3346]: [pulseaudio] bluez5-util.c: Transport TryAcquire() failed for transport /org/bluez/hci0/dev_20_74_CF_02_DB_F2/fd0 (Operation Not Authorized)

Whereas after doing the switch-to-HSP, disconnect, reconnect, switch-to-A2DP dance to make it work, the equivalent logging is:

Apr 26 01:52:40 zenbook pulseaudio[3346]: [pulseaudio] module-bluez5-device.c: Acquiring transport /org/bluez/hci0/dev_20_74_CF_02_DB_F2/fd2
Apr 26 01:52:40 zenbook bluetoothd[808]: /org/bluez/hci0/dev_20_74_CF_02_DB_F2/fd2: fd(23) ready
Apr 26 01:52:40 zenbook pulseaudio[3346]: [pulseaudio] module-bluez5-device.c: Transport /org/bluez/hci0/dev_20_74_CF_02_DB_F2/fd2 acquired: fd 46

[Karl Grzeszczak (karlgrz)]

@marc-edcc: your method works the most consistently. I've rebooted multiple times, connecting and disconnecting the headset with my phone and other devices, and turning it on and off, and also connecting and disconnecting from the 16.04 laptop, and I haven't had a single problem getting audio to work as expected.

I have to do this dance on each reboot, sleep, hibernate, or even any time I want to just disconnect my headset. For me, this isn't a huge deal, but I would classify this as a blocker bug regarding bluetooth audio.

@maxb: the next time I get into a bad state I will take a look at the logs.

[Karl Grzeszczak (karlgrz)]

@marc-edcc: ok, that apparently does not work when sleeping the laptop and resuming ;-(

@maxb: here's how my logs looked when it was failing to connect:

Apr 26 09:37:07 deathstar bluetoothd[1043]: Unable to get io data for Headset Voice gateway: getpeername: Transport endpoint is not connected (107)
Apr 26 09:37:07 deathstar bluetoothd[1043]: connect error: Connection timed out (110)
Apr 26 09:37:07 deathstar bluetoothd[1043]: 00:16:94:14:C4:15: error updating services: Input/output error (5)
Apr 26 09:37:57 deathstar bluetoothd[1043]: connect error: Connection timed out (110)
Apr 26 09:37:57 deathstar bluetoothd[1043]: 00:16:94:14:C4:15: error updating services: Input/output error (5)
Apr 26 09:37:57 deathstar bluetoothd[1043]: Unable to get io data for Headset Voice gateway: getpeername: Transport endpoint is not connected (107)

[Karl Grzeszczak (karlgrz)]

@maxb: also see this a bit later:

Apr 26 09:40:37 deathstar bluetoothd[1043]: Unable to get Headset Voice gateway SDP record: Host is down

[Karl Grzeszczak (karlgrz)]

If I run:

sudo systemctl restart bluetooth

the Bluetooth icon in the system taskbar goes blank, and if I try to re-enable it nothing ever happens. Here's the logs from that state:

Apr 26 09:41:38 deathstar systemd[1]: Stopping Bluetooth service...
Apr 26 09:41:38 deathstar bluetoothd[1043]: Terminating
Apr 26 09:41:38 deathstar bluetoothd[1043]: Endpoint unregistered: sender=:1.81 path=/MediaEndpoint/A2DPSource
Apr 26 09:41:38 deathstar bluetoothd[1043]: Endpoint unregistered: sender=:1.81 path=/MediaEndpoint/A2DPSink
Apr 26 09:41:38 deathstar bluetoothd[1043]: Stopping SDP server
Apr 26 09:41:38 deathstar bluetoothd[1043]: Exit
Apr 26 09:41:38 deathstar systemd[1]: Stopped Bluetooth service.
Apr 26 09:41:38 deathstar systemd[1]: Starting Bluetooth service...
Apr 26 09:41:38 deathstar bluetoothd[14513]: Bluetooth daemon 5.37
Apr 26 09:41:38 deathstar bluetoothd[14513]: Starting SDP server
Apr 26 09:41:38 deathstar systemd[1]: Started Bluetooth service.
Apr 26 09:41:38 deathstar dbus[1055]: [system] Activating via systemd: service name='org.freedesktop.hostname1' unit='dbus-org.freedesktop.hostname1.service'
Apr 26 09:41:38 deathstar bluetoothd[14513]: Bluetooth management interface 1.10 initialized
Apr 26 09:41:38 deathstar bluetoothd[14513]: Failed to obtain handles for "Service Changed" characteristic
Apr 26 09:41:38 deathstar bluetoothd[14513]: Not enough free handles to register service
Apr 26 09:41:38 deathstar bluetoothd[14513]: Error adding Link Loss service
Apr 26 09:41:38 deathstar bluetoothd[14513]: Not enough free handles to register service
Apr 26 09:41:38 deathstar bluetoothd[14513]: message repeated 2 times: [ Not enough free handles to register service]
Apr 26 09:41:38 deathstar bluetoothd[14513]: Current Time Service could not be registered
Apr 26 09:41:38 deathstar bluetoothd[14513]: gatt-time-server: Input/output error (5)
Apr 26 09:41:38 deathstar bluetoothd[14513]: Not enough free handles to register service
Apr 26 09:41:38 deathstar bluetoothd[14513]: Not enough free handles to register service
Apr 26 09:41:38 deathstar bluetoothd[14513]: Sap driver initialization failed.
Apr 26 09:41:38 deathstar bluetoothd[14513]: sap-server: Operation not permitted (1)
Apr 26 09:41:38 deathstar systemd[1]: Starting Hostname Service...
Apr 26 09:41:38 deathstar pulseaudio[2246]: [pulseaudio] backend-ofono.c: Failed to register as a handsfree audio agent with ofono: org.freedesktop.DBus.Error.ServiceUnknown: The name org.ofono was not provided by any .service files
Apr 26 09:41:38 deathstar bluetoothd[14513]: Endpoint registered: sender=:1.81 path=/MediaEndpoint/A2DPSource
Apr 26 09:41:38 deathstar bluetoothd[14513]: Endpoint registered: sender=:1.81 path=/MediaEndpoint/A2DPSink
Apr 26 09:41:38 deathstar dbus[1055]: [system] Successfully activated service 'org.freedesktop.hostname1'
Apr 26 09:41:38 deathstar systemd[1]: Started Hostname Service.

It's good to note that I reverted all of the configuration changes I made as suggested by @robert123m in post #5.

[Karl Grzeszczak (karlgrz)]

Tried again just now with the Sennheiser Urbanite XL Wireless and see this in the syslog:

May 10 08:00:02 deathstar wpa_supplicant[1472]: wlp6s0: Failed to initiate sched scan
May 10 08:00:09 deathstar rtkit-daemon[2045]: Supervising 4 threads of 1 processes of 1 users.
May 10 08:00:09 deathstar rtkit-daemon[2045]: Successfully made thread 19868 of process 2461 (n/a) owned by '1000' RT at priority 5.
May 10 08:00:09 deathstar rtkit-daemon[2045]: Supervising 5 threads of 1 processes of 1 users.
May 10 08:00:09 deathstar kernel: [ 505.747613] Bluetooth: hci0 SCO packet for unknown connection handle 0
May 10 08:00:09 deathstar kernel: [ 505.747616] Bluetooth: hci0 SCO packet for unknown connection handle 0
May 10 08:00:09 deathstar kernel: [ 505.747617] Bluetooth: hci0 SCO packet for unknown connection handle 0
May 10 08:00:14 deathstar pulseaudio[2461]: [pulseaudio] module-bluez5-device.c: Refused to switch profile to a2dp_sink: Not connected
May 10 08:00:14 deathstar bluetoothd[1176]: Unable to get io data for Headset Voice gateway: getpeername: Transport endpo int is not connected (107)
May 10 08:00:14 deathstar kernel: [ 510.979305] Bluetooth: hci0 SCO packet for unknown connection handle 258
May 10 08:00:21 deathstar bluetoothd[1176]: connect error: Host is down (112)
May 10 08:00:21 deathstar bluetoothd[1176]: GLib: Source ID 197 was not found when attempting to remove it

[Karl Grzeszczak (karlgrz)]

Also see this in dmesg:

[ 505.747613] Bluetooth: hci0 SCO packet for unknown connection handle 0
[ 505.747616] Bluetooth: hci0 SCO packet for unknown connection handle 0
[ 505.747617] Bluetooth: hci0 SCO packet for unknown connection handle 0
[ 510.979305] Bluetooth: hci0 SCO packet for unknown connection handle 258

[Karl Grzeszczak (karlgrz)]

    4 May 10 08:13:07 deathstar systemd[1]: Stopping User Manager for UID 108...
    5 May 10 08:13:07 deathstar systemd[1880]: Stopped target Default.
    6 May 10 08:13:07 deathstar systemd[1880]: Reached target Shutdown.
    7 May 10 08:13:07 deathstar systemd[1880]: Stopped target Basic System.
    8 May 10 08:13:07 deathstar systemd[1880]: Stopped target Timers.
    9 May 10 08:13:07 deathstar systemd[1880]: Stopped target Sockets.
   10 May 10 08:13:07 deathstar systemd[1880]: Stopped target Paths.
   11 May 10 08:13:07 deathstar systemd[1880]: Starting Exit the Session...
   12 May 10 08:13:07 deathstar systemd[1880]: Received SIGRTMIN+24 from PID 3313 (kill).
   13 May 10 08:13:07 deathstar systemd[1]: Stopped User Manager for UID 108.
   14 May 10 08:13:07 deathstar systemd[1]: Removed slice User Slice of lightdm.
   15 May 10 08:13:42 deathstar rtkit-daemon[1995]: Supervising 4 threads of 1 processes of 1 users.
   16 May 10 08:13:42 deathstar rtkit-daemon[1995]: Successfully made thread 3335 of process 2415 (n/a) owned by '1000' RT at priority 5.
   17 May 10 08:13:42 deathstar rtkit-daemon[1995]: Supervising 5 threads of 1 processes of 1 users.
   18 May 10 08:13:42 deathstar kernel: [ 166.886756] Bluetooth: hci0 SCO packet for unknown connection handle 0
   19 May 10 08:13:42 deathstar kernel: [ 166.886762] Bluetooth: hci0 SCO packet for unknown connection handle 0
   20 May 10 08:13:45 deathstar wpa_supplicant[1568]: wlp6s0: Failed to initiate sched scan
   21 May 10 08:13:51 deathstar bluetoothd[1084]: Start: Connection timed out (110)
   22 May 10 08:13:51 deathstar pulseaudio[2415]: [pulseaudio] bluez5-util.c: Transport Acquire() failed for transport /org/bl uez/hci0/dev_00_16_94_14_C4_15/fd1 (Input/output error)
   23 May 10 08:13:53 deathstar bluetoothd[1084]: Abort: Connection timed out (110)
   24 May 10 08:13:57 deathstar pulseaudio[2415]: [pulseaudio] backend-native.c: connect(): Device or resource busy
   25 May 10 08:14:03 deathstar bluetoothd[1084]: Unable to get io data for Headset Voice gateway: getpeername: Transport endp oint is not connected (107)
   26 May 10 08:14:03 deathstar bluetoothd[1084]: 00:16:94:14:C4:15: error updating services: Input/output error (5)
   27 May 10 08:14:03 deathstar kernel: [ 188.733312] Bluetooth: hci0 SCO packet for unknown connection handle 257
   28 May 10 08:14:03 deathstar kernel: [ 188.733315] Bluetooth: hci0 SCO packet for unknown connection handle 257
   29 May 10 08:14:03 deathstar kernel: [ 188.733316] Bluetooth: hci0 SCO packet for unknown connection handle 257
   30 May 10 08:14:23 deathstar bluetoothd[1084]: /org/bluez/hci0/dev_00_16_94_14_C4_15/fd2: fd(21) ready

[Karl Grzeszczak (karlgrz)]

Happy update. Resolved my issues with at least the Sennheiser Urbanite XL Wireless by disconnecting from my android phone before trying to connect to Ubuntu 16.04. I'll take it.

[jordan rubin (delorean1)]

Confirming this issue also with the Phillips Bluetooth speaker model BT2500B/37. Cannot set to higher quality sound in pulseaudio. Using gnome-ubuntu 16.04

[Romano Giannetti (romano-giannetti)]

Updated to 16.04, have the same problem. The fix in comment#5 fixes it!
Thanks!

[Alexander Menk (alex-menk)]

Same problem here. The workaround partially fixes it, but still I have sometimes problems to connect. Is it dependent on the bluetooth adapter, headset or both? Are there know-to-work adapters?

[Louis Bouchard (louis)]

Testing with a JBL SB400 BT on a Lenovo Thinkpad T450S with the intel ibt driver on F/W version 37081001103110e23 works correctly.

Doing a parallel test on my laptop (HP evo 850) that has :

[12383.666088] Bluetooth: hci0: read Intel version: 370710018002030d00
[12383.666095] Bluetooth: hci0: Intel Bluetooth firmware file: intel/ibt-hw-37.7.10-fw-1.80.2.3.d.bseq

Fails according to the bug report.

So it may be firmware related

[Vudentz (luiz-dentz-gmail)]

This is a headset bug actually, the policy of reconnecting profiles is from the initiator as it may or may not have disabled profiles, like cars normally do have options to have phone calls only or media only.

It is actually pretty strange coming from Sony, this used to work quite well with their devices in the past, maybe they have decided to move to another stack an now there is this sort of problem, please add the HCI trace maybe there is something preventing the L2CAP connection to A2DP.

[anonymous (anonymous-anonymous-1234)]

(In reply to Luiz Augusto von Dentz from comment #7)
> stack an now there is this sort of problem, please add the HCI trace maybe
> there is something preventing the L2CAP connection to A2DP.

Unfortunately, the sony headset was taken by someone else. I don't have it anymore.

[Fabrice-b (fabrice-b)]

(In reply to Luiz Augusto von Dentz from comment #7)
> This is a headset bug actually, the policy of reconnecting profiles is from
> the initiator as it may or may not have disabled profiles, like cars
> normally do have options to have phone calls only or media only.
>
> It is actually pretty strange coming from Sony, this used to work quite well
> with their devices in the past, maybe they have decided to move to another
> stack an now there is this sort of problem, please add the HCI trace maybe
> there is something preventing the L2CAP connection to A2DP.

oh yes, you're right, there's an action to be performed on the device (press on the play/pause jog switch), that makes the a2dp_sink profile available and active in pulseaudio.

[Cyril Humbert (cyril-h5t)]

Same issue with the folowing BT headphones:

 * Plantronic BackBeat FIT.

 * Plantronic BackBeat PRO.

 * B&O Beoplay H8.

The computer has a BT USB dongle:

  Cambridge Silicon Radio, Ltd Bluetooth Dongle (HCI mode)
  CSR8510 A10
  USB ID = 0a12:0001

The following procedure solved the problem for us:

 1. Purge Ubuntu pulseaudio packages and conf files. Backup and purge
    pulseaudio user conf files too.

 2. Install the corresponding packages from Debian testing (June 2016).
    See attached script (get-pulseaudio8-debian-amd64.sh).

 3. Restart.

 4. A2DP profile should now work as expected with the default
    bluez/pulseaudio configuration (Use pavucontrol to select A2DP
    if needed).

Since Debian Testing (June 2016) and Ubuntu 16.04 are based on the
same pulseaudio upstream version (8.0) this suggests that this
regression is maybe due to a difference in Debian and Ubuntu
packaging: either the compilation options, the default conf or the
specific patches (e.g. Ubuntu packages include more specific patches).

[Cyril Humbert (cyril-h5t)]

Updated procedure for comment #35, allowing to install more easily
Debian Testinf pulseaudio package on Ubunutu 16.04.

 1. Stop pulseaudio. Then, backup and purge pulseudio user conf files.

 2. Install pulseaudio Debian Testing (June 2016) packages over Ubuntu
    pulseaudio packages. This will appear as a kind of downgrade. See
    get-pulseaudio8-debian-amd64.sh script. If you use
    pulseaudio-module-x11 too, use the corresponding Debian Testing
    version too to satisfy dependencies.

 3. Restart.

 4. A2DP profile should now work as expected with the default
    BlueZ, pulseaudio configuration. Use pavucontrol to select
    A2DP profile if needed.

[Potato110 (aardappel110)]

Same here with a Teufel Airy bluetooth headset. #18 even doesn't work for me.

[Max Bowsher (maxb)]

Acting on Cyril's suggestion above, I downloaded the Debian 8.0-2 source package, rebuilt it locally, and installed it. It seems to have fixed the problem for me. So perhaps the problem really is in the Ubuntu delta?

[radiantptz (radiantptz)]

Confirmed fixed on @robert123m post #5 above on ubuntu 16.04.1 LTS. Can't thank you enough. Here is my env for reference:

$ dpkg -l | grep 'pulse\|bluetooth'
ii blueman 2.0.4-1ubuntu2
ii gir1.2-gnomebluetooth-1.0:amd64 3.18.2-1ubuntu2
ii gnome-bluetooth 3.18.2-1ubuntu2
ii gstreamer1.0-pulseaudio:amd64 1.8.2-1ubuntu0.1
ii libbluetooth3:amd64 5.37-0ubuntu5
ii libcanberra-pulse:amd64 0.30-2.1ubuntu1
ii libgnome-bluetooth13:amd64 3.18.2-1ubuntu2
ii libpulse-mainloop-glib0:amd64 1:8.0-0ubuntu3
ii libpulse0:amd64 1:8.0-0ubuntu3
ii libpulse0:i386 1:8.0-0ubuntu3
ii libpulsedsp:amd64 1:8.0-0ubuntu3
ii pulseaudio 1:8.0-0ubuntu3
ii pulseaudio-module-bluetooth 1:8.0-0ubuntu3
ii pulseaudio-module-x11 1:8.0-0ubuntu3
ii pulseaudio-utils 1:8.0-0ubuntu3

[Ayberk Özgür (equilibriumtr)]

I also confirm that #5 fixes audio output (was able to connect before but no audio output) with Sony MDRXB950BT on Kubuntu 16.04.

[Elias E. del Real (pixelshader)]

I have a Sony MDR-ZX220BT bluetooth headset that encounters this problem.
I'm using Ubuntu 16.10, bluez "5.41-0ubuntu3", and pulseaudio "1:9.0-2ubuntu2.1".

From
https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/1438510
https://bugs.launchpad.net/ubuntu/+source/indicator-sound/+bug/1577197
it looks like the sequence that works is:

1. Select the "off" profile in pulseaudio.
2.1 Disconnect the bluetooth earphones.
2.2 Reconnect the bluetooth earphones.
3. Select the "a2dp_sink" profile in pulseaudio.

Or for commands:
1. pactl set-card-profile bluez_card.00_18_09_62_4B_98 off
2. bluetoothctl
2.1 disconnect 00:18:09:62:4B:98
2.2 connect 00:18:09:62:4B:98
3. pactl set-card-profile bluez_card.00_18_09_62_4B_98 a2dp_sink

Bluetooth users are resorting to a scripted version of the above as a workaround:
https://gist.github.com/pylover/d68be364adac5f946887b85e6ed6e7ae#file-a2dp-py

What I'm seeing in /var/log/syslog if I select "a2dp_sink" as the profile, then disconnect, and reconnect:

Nov 25 00:47:50 fireflower pulseaudio[2178]: [pulseaudio] module-bluez5-device.c: Default profile not connected, selecting off profile
Nov 25 00:47:50 fireflower bluetoothd[970]: /org/bluez/hci0/dev_00_18_09_62_4B_98/fd17: fd(24) ready
Nov 25 00:47:50 fireflower pulseaudio[2178]: [pulseaudio] bluez5-util.c: Transport TryAcquire() failed for transport /org/bluez/hci0/dev_00_18_09_62_4B_98/fd17 (Operation Not Authorized)
Nov 25 00:47:50 fireflower kernel: [ 7852.918655] input: 00:18:09:62:4B:98 as /devices/virtual/input/input37

The message from bluetoothd "fd(24) ready" showing up then makes me wonder if this is a race between pulseaudio and the bluetooth connection?

If I then run:

pactl set-card-profile bluez_card.00_18_09_62_4B_98 a2dp_sink

I get nothing new in /var/log/syslog .

If I run:

pactl set-card-profile bluez_card.00_18_09_62_4B_98 off

I get nothing new in /var/log/syslog .

If I again run:

pactl set-card-profile bluez_card.00_18_09_62_4B_98 a2dp_sink

I get this output:

Failure: Input/Output error

and /var/log/syslog logs:

Nov 25 00:52:26 fireflower pulseaudio[2178]: [pulseaudio] bluez5-util.c: Transport Acquire() failed for transport /org/bluez/hci0/dev_00_18_09_62_4B_98/fd17 (Operation Not Authorized)

If I run

pactl list cards | tail -n 26

I see:

Card #18
 Name: bluez_card.00_18_09_62_4B_98
 Driver: module-bluez5-device.c
 Owner Module: 43
 Properties:
  device.description = "MDR-ZX220BT"
  device.string = "00:18:09:62:4B:98"
  device.api = "bluez"
  device.class = "sound"
  device.bus = "bluetooth"
  device.form_factor = "headset"
  bluez.path = "/org/bluez/hci0/dev_00_18_09_62_4B_98"
  bluez.class = "0x240404"
  bluez.alias = "MDR-ZX220BT"
  device.icon_name = "audio-headset-bluetooth"
  device.intended_roles = "phone"
 Profiles:
  headset_head_unit: Headset Head Unit (HSP/HFP) (sinks: 1, sources: 1, priority: 20, available: yes)
  a2dp_sink: High Fidelity Playback (A2DP Sink) (sinks: 1, sources: 0, priority: 10, available: yes)
  off: Off (sinks: 0, sources: 0, priority: 0, available: yes)
 Active Profile: off
 Ports:
  headset-output: Headset (priority: 0, latency offset: 0 usec, available)
   Part of profile(s): headset_h...

[Elias E. del Real (pixelshader)]

Created attachment 128181
btmon log of attempt to connect and automatically use a2dp_sink profile.

Attached a btmon log of attempt to connect and automatically use a2dp_sink profile.

[Vudentz (luiz-dentz-gmail)]

Well the logs looks completely different from the original problem, perhaps the logs of bluetoothd can show if there is perhaps another client acquiring the transport which would explain the error:

Nov 25 00:47:50 fireflower pulseaudio[2178]: [pulseaudio] bluez5-util.c: Transport TryAcquire() failed for transport
/org/bluez/hci0/dev_00_18_09_62_4B_98/fd17 (Operation Not Authorized)

[Tanu Kaskinen (tanuk)]

That another client could be gdm's pulseaudio instance. See comment 56 in bug 73325 for how to disable gdm's use of bluetooth: https://bugs.freedesktop.org/show_bug.cgi?id=73325#c56

[Josh-onn (josh-onn)]

I faced a similar issue with a2dp_sink profile where it is "not available" and what I did prior to connecting the Bluetooth headset in bluetoothctl was to first run pulseaudio --check -v to check if the pulseaudio daemon is running. If it is not running, issue "pulseaudio -D" and then check if it is running. Once the pulseaudio daemon is running, then only perform pairing and connecting in bluetoothctl.

[Armagvvg (armagvvg)]

Hi,

The same bug for me. Ubuntu 16.04. Packets:

  * libbluetooth3:amd64 5.37-0ubuntu5
  * pulseaudio-module-bluetooth 1:8.0-0ubuntu3.1
  * bluez 5.37-0ubuntu5
  * pulseaudio 1:8.0-0ubuntu3.1

Speakers JBL Charge 2 can't return after disconnecting. The python script a2dp.py published above restores connection.

Is there any chance to solve this in 2017? How can I help with this? (I'm a programmer). Is is a bug of pulseaudio or blueZ?

[Tanu Kaskinen (tanuk)]

(In reply to Slava Grig from comment #15)
> Is there any chance to solve this in 2017? How can I help with this? (I'm a
> programmer). Is is a bug of pulseaudio or blueZ?

If you're a programmer, you can dive into pulseaudio and bluez code and figure out why it's not working.

Have you disabled bluetooth for gdm, as hinted to in comment #13? Ruling out the gdm problem would be the first step.

If the a2dp profile in "pactl list cards" shows up as "available: no", then it's likely a bluez bug. An "unavailable" bluetooth profile in pulseaudio means that bluez tells pulseaudio that the profile isn't connected.

[ipatrol (ipatrol6010)]

Raising the priority due to downstream activity (affects a number of popular devices, directly impacts usability, leaves mainstream desktop Linux distros at a feature loss compared to non-free OSes, etc.). Also personally confirming that it affects the Bose Color SoundLink, and the workaround script is no help with it.

This bug should not have gone unresolved for two years.

[Danielius (rarebit)]

A more or less vanilla Ubuntu 16.04 installation, this bug is triggered. Python script provided (a2dp.py) works like a charm and returns sound to device.

[anonymous (anonymous-anonymous-1234)]

(In reply to Daniel from comment #18)
> A more or less vanilla Ubuntu 16.04 installation, this bug is triggered.
> Python script provided (a2dp.py) works like a charm and returns sound to
> device.

Where can I download a2dp.py? I want to make MDR-1RBT work on my linux.

[Danielius (rarebit)]

(In reply to crocket from comment #19)
> (In reply to Daniel from comment #18)
> > A more or less vanilla Ubuntu 16.04 installation, this bug is triggered.
> > Python script provided (a2dp.py) works like a charm and returns sound to
> > device.
>
> Where can I download a2dp.py? I want to make MDR-1RBT work on my linux.

Link provided above in comment #10

https://gist.github.com/pylover/d68be364adac5f946887b85e6ed6e7ae#file-a2dp-py

[anonymous (anonymous-anonymous-1234)]

Unfortunately, with MDR-1RBT, a2dp_sink is not available. Thus, a2dp.py cannot work in my case.

The only real errors I see in journalctl are

E: [pulseaudio] backend-native.c: connect(): Function not implemented
W: [pulseaudio] module-bluez5-device.c: Refused to switch profile to a2dp_sink: Not connected

[anonymous (anonymous-anonymous-1234)]

Wait. The proposed workaround worked for MDR-1RBT.
I followed the steps below.

1. Turn on MDR-1RBT
2. Let it be paired and connected to automatically
3. Execute `connect 00:18:09:1F:C3:29` in bluetoothctl
4. Execute `pactl set-card-profile bluez_card.00_18_09_1F_C3_29 a2dp_sink` on a shell.
5. Select MDR-1RBT's A2DP in sound settings

If there was a workaround, people must have found the exact cause of the problem.
What is the problem? What is the cause? What is the solution?

[Vudentz (luiz-dentz-gmail)]

(In reply to crocket from comment #22)
> Wait. The proposed workaround worked for MDR-1RBT.
> I followed the steps below.
>
> 1. Turn on MDR-1RBT
> 2. Let it be paired and connected to automatically
> 3. Execute `connect 00:18:09:1F:C3:29` in bluetoothctl
> 4. Execute `pactl set-card-profile bluez_card.00_18_09_1F_C3_29 a2dp_sink`
> on a shell.
> 5. Select MDR-1RBT's A2DP in sound settings
>
> If there was a workaround, people must have found the exact cause of the
> problem.
> What is the problem? What is the cause? What is the solution?

If you could please add some logs when connecting, syslog of bluetoothd and pulseaudio, then we can figure out if it needs some special handling.

[anonymous (anonymous-anonymous-1234)]

Created attachment 130342
bluetoothd.log

I attached bluetooth.log as a file because it is somewhat long.
I paste journalctl log below because it is short.

journalctl.log:
----------------------------------------
I don't know what this is.

Mar 21 21:11:42 arch-server pulseaudio[16130]: W: [pulseaudio] module-bluez5-device.c: Profile has no transport

-----------------------------------------
When I tried to select MDR-1RBT in gnome sound settings while MDR-1RBT's active profile was off.

Mar 21 21:12:43 arch-server pulseaudio[16130]: E: [pulseaudio] backend-native.c: connect(): Function not implemented
Mar 21 21:12:47 arch-server pulseaudio[16130]: E: [pulseaudio] backend-native.c: connect(): Function not implemented

[anonymous (anonymous-anonymous-1234)]

Created attachment 130343
bluetooth-headset_head_unit.log

I haven't found a way to get verbose pulseaudio log output on gnome.
According to my test, A2DP works for MDR-1RBT after reconnection.
But, I can't set the active profile to headset_head_unit for MDR-1RBT.
Read the following errors I got when I tried headset_head_unit.

~> pactl set-card-profile bluez_card.00_18_09_1F_C3_29 headset_head_unit
Failure: Input/Output error
journalctl:
E: [pulseaudio] backend-native.c: connect(): Function not implemented

I attached bluetoothd log recorded when I tried headset_head_unit.

[Vudentz (luiz-dentz-gmail)]

bluetoothd[16274]: src/adapter.c:connected_callback() hci0 device 00:18:09:1F:C3:29 connected eir_len 5
bluetoothd[16274]: src/adapter.c:dev_disconnected() Device 00:18:09:1F:C3:29 disconnected, reason 3
bluetoothd[16274]: src/adapter.c:adapter_remove_connection()
bluetoothd[16274]: plugins/policy.c:disconnect_cb() reason 3
bluetoothd[16274]: src/adapter.c:bonding_attempt_complete() hci0 bdaddr 00:18:09:1F:C3:29 type 0 status 0xe
bluetoothd[16274]: src/device.c:device_bonding_complete() bonding (nil) status 0x0e
bluetoothd[16274]: src/device.c:device_bonding_failed() status 14
bluetoothd[16274]: src/adapter.c:resume_discovery()
bluetoothd[16274]: src/adapter.c:connected_callback() hci0 device 00:18:09:1F:C3:29 connected eir_len 5
bluetoothd[16274]: src/adapter.c:dev_disconnected() Device 00:18:09:1F:C3:29 disconnected, reason 3
bluetoothd[16274]: src/adapter.c:adapter_remove_connection()
bluetoothd[16274]: plugins/policy.c:disconnect_cb() reason 3
bluetoothd[16274]: src/adapter.c:bonding_attempt_complete() hci0 bdaddr 00:18:09:1F:C3:29 type 0 status 0xe
bluetoothd[16274]: src/device.c:device_bonding_complete() bonding (nil) status 0x0e
bluetoothd[16274]: src/device.c:device_bonding_failed() status 14
bluetoothd[16274]: src/adapter.c:resume_discovery()
bluetoothd[16274]: src/adapter.c:connected_callback() hci0 device 00:18:09:1F:C3:29 connected eir_len 5
bluetoothd[16274]: src/adapter.c:dev_disconnected() Device 00:18:09:1F:C3:29 disconnected, reason 3
bluetoothd[16274]: src/adapter.c:adapter_remove_connection()
bluetoothd[16274]: plugins/policy.c:disconnect_cb() reason 3
bluetoothd[16274]: src/adapter.c:bonding_attempt_complete() hci0 bdaddr 00:18:09:1F:C3:29 type 0 status 0xe
bluetoothd[16274]: src/device.c:device_bonding_complete() bonding (nil) status 0x0e
bluetoothd[16274]: src/device.c:device_bonding_failed() status 14
bluetoothd[16274]: src/adapter.c:resume_discovery()
bluetoothd[16274]: src/adapter.c:connected_callback() hci0 device 00:18:09:1F:C3:29 connected eir_len 5
bluetoothd[16274]: src/adapter.c:dev_disconnected() Device 00:18:09:1F:C3:29 disconnected, reason 3
bluetoothd[16274]: src/adapter.c:adapter_remove_connection()
bluetoothd[16274]: plugins/policy.c:disconnect_cb() reason 3
bluetoothd[16274]: src/adapter.c:bonding_attempt_complete() hci0 bdaddr 00:18:09:1F:C3:29 type 0 status 0xe
bluetoothd[16274]: src/device.c:device_bonding_complete() bonding (nil) status 0x0e
bluetoothd[16274]: src/device.c:device_bonding_failed() status 14
bluetoothd[16274]: src/adapter.c:resume_discovery()
bluetoothd[16274]: src/adapter.c:connected_callback() hci0 device 00:18:09:1F:C3:29 connected eir_len 5
bluetoothd[16274]: src/profile.c:ext_confirm() incoming connect from 00:18:09:1F:C3:29

This sounds strange, it is as if the connection is interrupted several times before it succeeds, anyway once past this line all looks fine:

bluetoothd[16274]: src/service.c:change_state() 0x250c1a0: device 00:18:09:1F:C3:29 profile Headset Voice gateway state changed: connecting -> connected (0)

bluetoothd[16274]: profiles/audio/sink.c:sink_set_state() State changed /org/bluez/hci0/dev_00_18_09_1F_C3_29: SINK_STATE_CONNECTING -> SINK_STATE_CONNECTED
...

[anonymous (anonymous-anonymous-1234)]

> This sounds strange, it is as if the connection is interrupted several times before it succeeds

When I turn on MDR-1RBT, the connection is severed by an unknown force several times before the connection succeeds.

> Perhaps this is GDM instance taking over? Btw, have we make sure GDM instance disables its endpoints when it is no longer active?

No, GDM is not running pulseaudio. I explicitly disabled pulseaudio in GDM.

~> ps -f (pgrep pulseaudio)
UID PID PPID C STIME TTY STAT TIME CMD
crocket 2038 1936 0 10:52 ? S<sl 0:04 /usr/bin/pulseaudio --daemonize=no

> there is some client acquiring the transport

I played music after A2DP became available to make sure A2DP actually worked.

[Diego Fernandez (aiguofer)]

I'm on Fedora 26, using PulseAudio 10.0 and I'm having the same issue with my Sony MDR100ABN headset. I've tried all sorts of things, and the only thing that consistently works is to:

1. Let it connect automatically when I turn on my headset
2. Disconnect through bluetoothctl
3. Reconnect through bluetoothctl
4. Use pacmd set-card-profile to set a2dp_sink

I've tried disabling the GDM pulseaudio, changing 'load-module module-bluetooth-discover headset=<mode>' in default.pa, adding 'load-module module-switch-on-connect' to default.pa, adding 'Disable=Headset' to '[General]' in bluetooth/main.conf, and pretty much any combination of those changes.

[Johannes Wienke (languitar)]

(In reply to Diego Fernandez from comment #28)
> I'm on Fedora 26, using PulseAudio 10.0 and I'm having the same issue with
> my Sony MDR100ABN headset. I've tried all sorts of things, and the only
> thing that consistently works is to:
>
> 1. Let it connect automatically when I turn on my headset
> 2. Disconnect through bluetoothctl
> 3. Reconnect through bluetoothctl
> 4. Use pacmd set-card-profile to set a2dp_sink

For me this is also the only thing that works, apart from the fact that I can enable A2DP within blueman after the explicit connection.

Archlinux, no GDM running.

[Sergey (hxss)]

My headset is Sony DR-BT50.
4.13.12-1-ARCH.

This is not gdm problem. I use lightdm + xfce. And have this bug. Nevertheless I was try to disable lightdm, reboot, startx - the bug still here.

[Sergey (hxss)]

Created attachment 135608
journalctl log of "automatic" headset connection

[Sergey (hxss)]

Created attachment 135609
pactl-subscribe log of "automatic" headset connection

[Sergey (hxss)]

Created attachment 135610
btmon log of "automatic" headset connection

[Sergey (hxss)]

Created attachment 135611
journalctl log of manual headset connection

[Sergey (hxss)]

Created attachment 135612
pactl-subscribe log of manual headset connection

[Sergey (hxss)]

Created attachment 135613
btmon log of manual headset connection

[Sergey (hxss)]

added some logs. Please check it

My system:
4.13.12-1-ARCH

BT:
local/blueberry 1.1.15-1
    Bluetooth configuration tool
local/bluez 5.47-2
    Daemons for the bluetooth protocol stack
local/bluez-libs 5.47-2
    Deprecated libraries for the bluetooth protocol stack
local/bluez-tools 0.2.0-1
    A set of tools to manage Bluetooth devices for Linux
local/bluez-utils 5.47-2
    Development and debugging utilities for the bluetooth protocol stack
local/gnome-bluetooth 3.26.0+1+g3875844-1
    The GNOME Bluetooth Subsystem
local/pulseaudio-bluetooth 11.1-1
    Bluetooth support for PulseAudio
local/sbc 1.3-1
    Bluetooth Subband Codec (SBC) library

Pulse:
local/libcanberra-pulse 0.30+2+gc0620e4-1
    PulseAudio plugin for libcanberra
local/libpulse 11.1-1
    A featureful, general-purpose sound server (client library)
local/pavucontrol 3.0-4
    PulseAudio Volume Control
local/pulseaudio 11.1-1
    A featureful, general-purpose sound server
local/pulseaudio-alsa 2-3
    ALSA Configuration for PulseAudio
local/pulseaudio-bluetooth 11.1-1
    Bluetooth support for PulseAudio
local/xfce4-pulseaudio-plugin 0.3.2-1 (xfce4-goodies)
    Pulseaudio plugin for Xfce4 panel

[Sergey (hxss)]

wrote this fully automated(you no need to manually run scripts or specify mac addresses) udev rule for temporary solve the problem. It work with already paired and new headsets.
https://gist.github.com/hxss/a3eadb0cc52e58ce7743dff71b92b297

[Knizek-confy (knizek-confy)]

@Sergey: thanks for your udev script.

Would you add a comment to your script that bluez-tools package is a dependency (for bt-device)? (Standard dependency for bluetooth sound seems to be only bluez-utils.)

[Sergey (hxss)]

(In reply to Milan Knížek from comment #39)
> @Sergey: thanks for your udev script.
>
> Would you add a comment to your script that bluez-tools package is a
> dependency (for bt-device)? (Standard dependency for bluetooth sound seems
> to be only bluez-utils.)

Dependencies list added.

[Rathann (dominik-greysector)]

Isn't this a duplicate of bug 73325?

[Sergey (hxss)]

Dominik, no. For me all protocols works fine. But only if I connect to headset manual from bluetoothctl. We need exactly AUTOMATIC connection, when headset trying to connect to PC.

[Felix-steghofer (felix-steghofer)]

FYI:
https://medium.com/@towo/bose-qc35-and-arch-linux-24a260391df3

[Gih (gih)]

Just want to confirm this bug with SONY MDR-XB650BT on Kernel 4.14.52-1-MANJARO.

The script from Comment 38 helped me.

[Freedesktop-h (freedesktop-h)]

Can also confirm the issue with Sony WH-H900N headphones, Arch with kernel 4.17.4, BlueZ 5.50, PulseAudio 12.0.

[Gitlab-migration (gitlab-migration)]

-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/pulseaudio/pulseaudio/issues/525.