[pulseaudio-tickets] [Bug 58746] bluetooth audio out of sync when connection temporarily drops

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=58746

--- Comment #25 from Johannes Larsen <mail at johslarsen.net> ---
(In reply to Georg Chini from comment #24)
> Nobody willing to test?

I have tested the patch with a pair of Mackie CR4BT bluetooth speakers that
suffers from the lag issues after they reconnect. That patch was applied
7f09164e and built as part of the [pulseaudio-git] AUR package.

When the computer initiate connection with the speakers (i.e. `echo connect MAC
| bluetoothctl`) everything works fine, but every few seconds the logs are
flooded with 1000s of these messages: 

  [bluetooth] module-bluez5-device.c: Broken kernel: we got EAGAIN on write()
after POLLOUT!

When the speakers initiate the connection they disconnect after about a second
(even before they start making sound), and when the speakers disconnects
(either a second after they connected or if I turn them off after computer
initiated the connection) pulseaudio aborts with:

  [bluetooth] module-bluez5-device.c: Skipping 2176 us (= 384 bytes) in audio
stream
  [bluetooth] module-bluez5-device.c: Assertion 'u->write_memchunk.length ==
u->write_block_size' failed at modules/bluetooth/module-bluez5-device.c:437,
function a2dp_process_render(). Aborting.


I only observe the out-of-sync audio problems when the connection is initiated
by the audio device (e.g. power on, pushing their connect buttons, after signal
loss). I am happy to test further patches, but I do not see any means to
reproduce the problem as long as pulseaudio aborts whenever any of my devices
tries to reconnect.

I also have a Bose QC35 and a RHA MA 650 Wireless headset to test with, but
they usually take longer to suffer from audio sync problems than the Mackie
CR4BT speakers. I tested the patch with the headsets to, and they observed the
same crashing behavior as when I tested it with the speakers. 


BTW, as a workaround until the problem is fixed I have a script I use whenever
I notice audio sync problems that disconnect then reconnects the device, so
that it is the computer that initiated the connection. This seems to work
consistently and avoids the problem until there are intermittent connection
problem (e.g. I walk too far away from the computer with a headset) or I turn
off and on the audio devices.

[pulseaudio-git]
https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=pulseaudio-git

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/pulseaudio-bugs/attachments/20180303/38d2857b/attachment.html>