Logo

Home About Community Download Documentation Planet


PulseAudio 5.0 Release Notes

Welcome to another episode of PulseAudio release notes! The 5.0 release contains the work of last year's GSoC students Alexander Couzens and Damir Jelić, BlueZ 5 support, and much more. It's also noteworthy that Peter Meerwald from bct electronic joined the PulseAudio maintainer team during this release cycle.

Changes at a Glance

  • BlueZ 5 support (A2DP only)
  • Reimplementation of the tunnel modules
  • Native log target support for systemd-journal
  • Small changes here and there
  • Many bug fixes

Notes for Application Developers

The pa_card_profile_info structure is deprecated, consider migrating to pa_card_profile_info2. The old struct will keep working, though, so applications should not break even if they are not updated.

Notes for Packagers

  • PulseAudio now supports BlueZ 5, but only the A2DP profile, which means that in case of headsets, only playback is available, the headset microphone can't be used. BlueZ 4 is still the only way to make the headset and hands-free profiles work. It's possible to build PulseAudio with support for both BlueZ versions, and that's the default behaviour. If you want to disable one version or the other, pass --disable-bluez4 or --disable-bluez5 to the configure script.

  • PulseAudio now supports systemd-journal as a log target. This adds an optional build dependency to libsystemd-journal.

  • The build dependency to libbluetooth has been dropped.

  • We now use upstream gettext instead of the GLib version. The minimum required version is 0.18.1.

  • To avoid accidental weakening of security, the default behaviour on Linux is now to fail during the configure script execution if libcap is not found. To build PulseAudio without libcap support, the --without-caps option can be used.

  • The minimum supported alsa-lib version is now 1.0.19 (down from 1.0.24).

Changes in More Detail

BlueZ 5 support (A2DP only)

First some background: The BlueZ project decided to redesign their client interface. They also decided to drop support for the old client interface. BlueZ 4 has the old interface and BlueZ 5 has the new interface. The decision to drop support for the old interface in BlueZ 5 meant that all applications, including PulseAudio, that used to work with BlueZ 4 didn't work with BlueZ 5.

The support for BlueZ 5 has been gradually added in various applications, and this is the first PulseAudio version that supports BlueZ 5. This means that everything is great, right? Not so. The BlueZ project also decided to drop support for the HSP and HFP profiles, which were the profiles responsible for handling telephony audio. If you have a headset, its microphone won't work with BlueZ 5, because the microphone is only supported by the HSP and HFP profiles.

There are distributions that have migrated to BlueZ 5 without providing users the alternative of staying with BlueZ 4. Some of those distributions probably made the transition without understanding the consequences - they now have a serious regression in functionality, because their users' Bluetooth headsets have stopped working (well, the headsets can still be used for listening to music, but they're useless for VoIP applications).

GNOME made also a decision, possibly misinformed one, to drop support for BlueZ 4 in their last release, which means that upgrading to the current GNOME version (3.10) has one of two problems depending on the BlueZ version in the system: with BlueZ 4, the GNOME UI for managing Bluetooth won't work, and with BlueZ 5 the headset audio functionality will be crippled.

So what's the way forward with getting the HSP and HFP profiles work with BlueZ 5? There is partial support for those in oFono (a telephony daemon), which will hopefully be completed soon, and the next PulseAudio release will then hopefully support HSP/HFP through oFono. This pulls in a telephony stack as a dependency for using your Bluetooth headset, which might be considered overkill. This issue is yet to be resolved with the BlueZ developers.

Now, the technical details of the BlueZ 5 support in PulseAudio: there are now separate modules for BlueZ 4 and BlueZ 5: module-bluez4-discover and module-bluez5-discover. The old module-bluetooth-discover still exists and is used in the default configuration. The role of module-bluetooth-discover is now to act as an abstraction layer: it will load module-bluez4-discover and/or module-bluez5-discover based on what's present in the system. This means that the migration to BlueZ 5 can be done without changing the PulseAudio configuration.

Reimplementation of the tunnel modules

The old module-tunnel-sink and module-tunnel-source have an independent implementation of the PulseAudio client protocol. There's another implementation in libpulse, which is what applications use to talk to the PulseAudio server. This is clear duplication of effort, and makes the maintenance of the tunnel modules much harder than what it should be.

As a GSoC project, Alexander Couzens reimplemented the tunnel modules so that they use libpulse instead of implementing the protocol details themselves. The reimplemented modules are now available as module-tunnel-sink-new and module-tunnel-source-new. The implementation isn't fully complete yet, which is why the old modules still exist, but the new versions can already be tested (if you're having trouble with audio streaming quality of the old tunnel modules, it's worth checking if the new modules work better). Note that if you use module-tunnel-sink-new or module-tunnel-source-new in a configuration file, that configuration file will break when those new modules are finished, because the "-new" suffix will be removed from the module names.

Systemd-journal support

A new log target has been added: "journal". It is used by default, if PulseAudio is built with systemd-journal support. The new log target is also supported in daemon.conf (the log-target option), on the command line (the --log-target option) and in pacmd (the set-log-target command).

Utility changes

parecord and parec now support the --monitor-stream option, which allows recording from a single playback stream.

pacmd doesn't print the welcome message and the command prompt in non-interactive mode any more.

pacmd now supports "newfile:somefile.log" as a target for the set-log-target command. "newfile:somefile.log" means that somefile.log will not be overwritten if it already exists. Instead, a new file with name somefile.log.1 will be created. The "newfile:" target was already supported in the daemon configuration file, this change just brings pacmd to the same level.

Module changes

module-rtp-recv now supports the "latency_msec" argument. It can be used to control the local buffer size. The default latency is 500 ms.

module-rtp-send now supports the "inhibit_auto_suspend" argument, which controls the policy for interacting with the automatic source suspension logic. module-suspend-on-idle suspends devices when no streams are connected to them, or if all connected streams are marked to be ignored by module-suspend-on-idle. This argument controls whether and when the stream that module-rtp-send creates is marked to be ignored by module-suspend-on-idle. There are three options: "always", "never" and "only_with_non_monitor_sources". The "always" option prevents module-suspend-on-idle from ever suspending the source, the "never" option allows module-suspend-on-idle to always suspend the source and the "only_with_non_monitor_sources" option allows module-suspend-on-idle to suspend monitor sources, but not "regular" sources. The important case where this argument really matters is when module-rtp-send is connected to a monitor source. You should choose the "always" option if you want to send silence to the network when nothing is playing to the monitored sink. You should choose the "only_with_non_monitor_sources" option if you want to pause the RTP stream when nothing is playing to the monitored sink. The "never" option is not really useful, because it doesn't make sense to suspend non-monitor sources like microphones. The "never" option only exists as a logical complement the "always" option - if you find the "never" option useful anyway, we'd be interested to hear about your use case.

module-tunnel-sink and module-tunnel-source (the old versions) now support the auto argument. If it's set to true, then the default connection parameters are gathered from the environment in the same way as applications do. This is usually a bad idea, because often that will cause the server to connect to itself, but some people found this feature useful for their purposes.

When module-suspend-on-idle reads the timeout property from sinks and sources, it now treats a negative timeout as "never suspend" instead of ignoring the property.

module-alsa-card now prefers to adjust the "Capture" volume in the ALSA mixer instead of boosts, e.g. "Mic Boost". This should result in less noise in the captured audio on some hardware.

module-alsa-card now supports the "PlaybackRate" and "CaptureRate" UCM device options.

When using ALSA UCM, module-alsa-card now assigns lower priority to ports that combine multiple devices than ports that use only one device.

The maximum configurable latency of module-loopback has been increased from 2 seconds to 30 seconds.

module-bluetooth-proximity has been removed. It was probably not used by anyone, since it required an ancient version of BlueZ to work.

Configuration syntax changes:

In path configuration files (/usr/share/pulseaudio/alsa-mixer/paths/*) the "name" option in the "General" section is not supported any more. The path name is now determined solely based on the file name. The "name" option was also used as a key to a table containing path descriptions, and that functionality is now provided by a new option called "description-key".

A single path can't be used for both input and output any more.

Hardware support changes

Jack detection support for line out jacks has been added. This means that user interfaces can now hide or mark unavailable the line out when nothing is plugged in to the jack.

Proper hardware volume support has been added for some laptops that have internal surround speakers.

A configuration file for Native Instruments Traktor Audio 2 was missing from the 4.0 tarball. It's now shipped with 5.0, so users of that hardware actually get the benefit from the configuration file (which allows PulseAudio to use both outputs of the sound card).

Performance improvements

Improved float->s16 and s16->float sample conversion for ARM NEON.

libpulse changes

  • An available flag has been added to the card profile info. If a profile is not available, activating it doesn't make sense. GUIs that allow the user to control the card profile can use the new flag to hide or grey out unavailable profiles. Currently the availability is only set for Bluetooth cards, and profiles on other cards will always appear as available.

  • Adding the available flag to the existing pa_card_profile_info struct turned out to break the ABI due to an earlier design mistake, so instead of extending pa_card_profile_info, we had to introduce pa_card_profile_info2, which contains the same information as the old struct plus the new available flag. The old pa_card_profile_info struct is now deprecated. The profiles field in pa_card_port_info and the profiles and active_profile fields in pa_card_info are deprecated too, superseded by the profiles2 field in pa_card_port_info and profiles2 and active_profile2 fields in pa_card_info.

  • pa_stream_connect_playback() and pa_stream_connect_record() now accept single-channel volume even if the stream has multiple channels. The single volume is applied to all channels.

  • pa_simple_flush() can now be used with record streams too, in addition to playback streams.

  • New functions: pa_sample_format_valid(), pa_sample_rate_valid() and pa_channels_valid(). The functions can be used to check the validity of individual sample spec fields. There was already pa_sample_spec_valid() that checks all fields, but sometimes only a subset of the fields are interesting.

  • New function: pa_context_load_cookie_from_file(). This allows clients to load the authentication cookie from a custom file. It may be that there are no applications that need this - the reason for adding the function was that module-tunnel-sink-new needed it.

  • New function: pa_threaded_mainloop_set_name(). This is used internally in PulseAudio, for improving the server log messages when using pa_threaded_mainloop. The usefulness of this new function is probably very limited for applications.

libpulse Vala bindings changes:

The binding for pa_cvolume_set() was missing, now CVolume.set() has been added.

Miscellaneous

module-dbus-protocol has been removed from the default configuration. The module is in an experimental state, and it tends to crash the server sometimes.

Next Steps

HSP/HFP support with BlueZ 5 will hopefully arrive in the next PulseAudio release. There's also ongoing work with making routing and volume handling more flexible for policy modules and for client applications.

git shortlog

Alexander Couzens (8):
      thread-mq: Support a mainloop as thread/consumer
      client-conf: refactor cookie loaders
      tunnel-sink-new: add a rewrite of module-tunnel using libpulse
      context: add pa_context_load_cookie_from_file()
      tunnel-source-new: counterpart to module-tunnel-sink-new
      tunnel-sink-new: remove switch-default from state change callbacks
      tunnel-sink-new: Fix a possible crash
      tunnel-new: add cookie module argument

Alexander E. Patrakov (1):
      Documentation of known misuse of PulseAudio API

Alexander Potashev (1):
      Update Russian translation

Anton Lundin (2):
      build-sys: Support building pulseaudio as a submodule
      build-sys: Fix building with Android toolchain

Arun Raghavan (35):
      core: Add an "internal" suspend cause
      pulsecore: Fix assert in pa_msgobject creation
      mainloop: Add API to set thread name for threaded mainloop
      zeroconf: Make Avahi usage in m-z-publish async
      tests: Add a latency measurement test
      tests: Make loopback latency test run with rt priority
      tests: Adjust latency test calibration
      tests: Factor out loopback setup code
      tests: Make loopback latency test more accurate
      alsa-mixer: Disable headphones when playing to speakers
      doc: Fix building doxygen docs out of tree
      iochannel: Document modified pa_iochannel_write return value
      sink-input: Log format negotiation better
      mainloop: Document need for mainloop lock around pa_mainloop_api_once
      zeroconf: Fix pa_mainloop_api_once usage
      sample: Apply SPARC-specific condition on SPARC only
      doc: Update LICENSE to reflect optional libpulse dep on libdbus
      alsa: Trivial whitespace deletion
      sink-input: Don't assert when removing non-existent volume factor
      hashmap: Add the ability to free keys
      alsa: Allow sample spec override in mappings
      alsa-ucm: Use playback/capture rate specification from UCM
      alsa-ucm: Make combination port names deterministic
      build-sys: Fix warning in use-case.h configure check
      log: Add missing header
      sink: Increase max sink inputs per sink
      mutex: Fix compiler warning when priority inheritance is not supported
      hashmap: Add a key+value iterator
      alsa-ucm: Don't allow Playback/CaptureRate to be 0
      alsa-ucm: Fix bad rate check
      i18n: Update bluetooth bits in POTFILES
      alsa: Log some output if we disable tsched for BATCH devices
      protocol-native: Don't enumerate unlinked sinks and sources
      build-sys: Bump soname
      Revert "protocol-native: Don't enumerate unlinked sinks and sources"

Ben Brewer (4):
      mainloop, glib-mainloop: time_restart could cause incorrect event ordering
      rtp-send: Remove "don't inhibit auto-suspend" flag
      ladspa-sink: Handle empty chunks in sink_input_pop_cb
      ladspa-sink: Set a silence memchunk for the memblockq

Cheng-Chia Tseng (1):
      i18n: Update zh_TW translation

Colin Guthrie (2):
      card: Ensure we still pick a profile even if it's unavailable.
      sink/source: When picking the initial ports, prefer ones that are not unavailable.

Daniel Albers (1):
      combine-sink: fix segfault

David Henningsson (16):
      alsa-mixer: Add "Line Out" path
      alsa: Fix "Scheduling delay of..." message
      iochannel: Avoid unnecessary wakeup after successful write
      pstream: Fixup hangs caused by recent iochannel patch
      protocol-esound/http/simple: Make sure callers can handle iochannel_write changes
      Revert "alsa-mixer: Add a couple of assertions"
      daemon.conf: Document new speex-float-1 default
      alsa-ucm: Make combination ports have lower priority
      alsa-mixer: Prefer moving "Capture" before moving boosts
      module-tunnel: Fixup broken build
      alsa: Turn one assertion into two
      default/system.pa: Do not load module-dbus-protocol
      alsa-mixer: Drop all unused paths, not only unsupported paths
      alsa-mixer: Add internal surround speaker elements
      sink-input/source-output: Fix LFE remixing suddenly enabled
      suspend-on-idle: Ensure we still time out if a stream remains corked

Dominique Leuenberger (1):
      build-sys: Move ORC_SOURCE definitions outside of "if HAVE_ORC"

Héctor Martín (1):
      zeroconf-discover: Specify the interface when connecting to a link-local IPv6 address

Ismo Puustinen (1):
      module: initialize module index to invalid value.

Jan Alexander Steffens (heftig) (3):
      core-util: Clamp RLIMIT_RTTIME to what RealtimeKit accepts
      alsa: Fix crash when loading bare ALSA sink/source
      thread-mq: Initialize thread_mainloop to NULL if not in use

Javier Jardón (4):
      build-sys: Use upstream gettext instead glib one
      build-sys: Use autoreconf in bootstrap.sh
      build-sys: Use build-aux directory to store autogenerated files
      build-sys: Do not AC_SUBST *_CFLAGS and *_LIBS

João Paulo Rechi Vita (79):
      bluetooth: Remove unnecessary braces
      bluetooth: Improve code and log readability
      bluetooth: Add 'bluez.alias' property
      bluetooth: Use 'Alias' value as the device description
      bluetooth: Remove the 'bluez.name' property
      module: Fix style
      bluetooth: Remove unused DBusError
      bluetooth: Fix coding style
      build-sys: Fix misuse of "dependant"
      Revert "bluetooth: Support transport auto-release"
      Revert "bluetooth: Update to new BlueZ 5 transport acquire/release API"
      Revert "bluetooth: Support media transport's State property"
      Revert "bluetooth: Parse media transport's properties"
      Revert "bluetooth: Support Properties.PropertiesChanged signal"
      Revert "bluetooth: Support ObjectManager interface add/remove"
      Revert "bluetooth: BlueZ 5 interface rename to org.bluez.MediaTransport1"
      Revert "bluetooth: BlueZ 5 interface rename to org.bluez.Media1"
      Revert "bluetooth: BlueZ 5 interface rename to org.bluez.MediaEndpoint1"
      Revert "bluetooth: Fix error checking style"
      Revert "bluetooth: Parse the tree returned by ObjectManager"
      Revert "bluetooth: Detect BlueZ 5"
      bluetooth: Update modules description and copyright
      bluetooth: Rename bluetooth modules to BlueZ 4
      bluetooth: Rename shared struct to make it BlueZ 4 specific
      bluetooth: Name BlueZ 4 symbols accordingly
      bluetooth: Register BlueZ 4 endpoints in a separate object path
      bluetooth: Prefix BlueZ 4 card profiles enum with PA_BLUEZ4_
      bluetooth: Remove module-bluetooth-proximity
      build: Make the build of bluetooth modules BlueZ 4 specific
      build: Add support for enabling/disabling BlueZ 5 support
      bluetooth: Create stub for module-bluez5-discover
      bluetooth: Create pa_bluetooth_discovery for BlueZ 5
      bluetooth: Track org.bluez for BlueZ 5
      bluetooth: Create infrastrucure for bluetooth hooks
      bluetooth: Create pa_bluetooth_adapter for BlueZ 5 support
      bluetooth: Create pa_bluetooth_device for BlueZ 5 support
      bluetooth: Remove all devices and adapters when org.bluez goes away
      bluetooth: Create pa_bluetooth_transport for BlueZ 5 support
      bluetooth: Register MediaEndpoint1 objects with D-Bus
      bluetooth: Implement org.bluez.MediaEndpoint1.SetConfiguration()
      bluetooth: Implement org.bluez.MediaEndpoint1.SelectConfiguration()
      bluetooth: Implement org.bluez.MediaEndpoint1.ClearConfiguration()
      bluetooth: Add utility function to send D-Bus messages with reply
      bluetooth: Get managed objects
      bluetooth: Create a function to remove only one adapter object
      bluetooth: Handle InterfacesAdded and InterfacesRemoved
      bluetooth: Parse BlueZ 5 D-Bus interfaces
      bluetooth: Parse BlueZ 5 adapter properties
      bluetooth: Register endpoints with BlueZ 5 adapter
      bluetooth: Handle PropertiesChanged for org.bluez.Adapter1
      bluetooth: Parse BlueZ 5 device properties
      bluetooth: Protect from a misbehaving bluetoothd
      bluetooth: Handle PropertiesChanged for org.bluez.Device1
      bluetooth: Handle PropertiesChanged for org.bluez.MediaTransport1
      bluetooth: Track devices in module-bluez5-discover
      bluetooth: Create module-bluez5-device
      bluetooth: Get BlueZ 5 device object
      bluetooth: Unload module-bluez5-device on device's disconnection
      bluetooth: Create BlueZ 5 card
      bluetooth: Create BlueZ 5 card ports
      bluetooth: Create BlueZ 5 card profile for each audio UUID
      bluetooth: Initialize profiles for BlueZ 5 cards
      bluetooth: Create sink for BlueZ 5 cards
      bluetooth: Create source for BlueZ 5 cards
      bluetooth: Start / stop I/O thread for BlueZ 5 cards
      bluetooth: Set card profile for BlueZ 5 cards
      bluetooth: Create I/O thread function for BlueZ 5 cards
      bluetooth: Process sink messages for BlueZ 5 cards
      bluetooth: Process source messages for BlueZ 5 cards
      bluetooth: Handle changes to BlueZ 5 transports state
      bluetooth: Implement get_n_used() for module-bluez5-device
      bluetooth: Fail to load driver if discovery module is not loaded
      module: Create pa_module_exists()
      cli-command: Use pa_module_exists() in .ifexists
      bluetooth: Revive module-bluetooth-discover
      bluetooth: Track discovery modules by index
      console-kit: Remove logging of every message on the bus
      bluetooth: Check adapter address to set device_info_valid
      bluetooth: Fix crash in pa_bluetooth_discovery_get_device_by_address()

Juho Hämäläinen (1):
      dbus: Use correct idxset when getting sources

Kenneth Perry (1):
      bonjour-publish: Return ports in network byteorder

Kerrick Staley (1):
      Fix typos in command_names

Kiran Krishnappa (1):
      sndfile-util: fix format for 24bit depth wav files

Koop Mast (2):
      sconv, svolume: Fix compilation on 32-bit FreeBSD
      Make tests compile on FreeBSD

Lars-Peter Clausen (1):
      alsa: Disable timer-scheduling for PCMs with the BATCH flag

Laurentiu Nicola (2):
      module-rtp-recv: Add an argument for latency
      rtp-recv: Only try adding the group membership for multicast addresses

Luiz Augusto von Dentz (4):
      introspect: Expose card profile available flag
      pactl: Print available flag of card profiles
      introspect: Fix ABI break introduced by b98a2e1
      card: Only set active_profile with available profile

Magnus Ekhall (1):
      pacat: Added support for recording from one specific sink input

Mikel Astiz (12):
      bluetooth: Detect BlueZ 5
      bluetooth: Parse the tree returned by ObjectManager
      bluetooth: BlueZ 5 interface rename to org.bluez.MediaEndpoint1
      bluetooth: BlueZ 5 interface rename to org.bluez.Media1
      bluetooth: BlueZ 5 interface rename to org.bluez.MediaTransport1
      bluetooth: Support ObjectManager interface add/remove
      bluetooth: Support Properties.PropertiesChanged signal
      bluetooth: Parse media transport's properties
      bluetooth: Support media transport's State property
      bluetooth: Update to new BlueZ 5 transport acquire/release API
      bluetooth: Support transport auto-release
      loopback: Fix cork state not updated after move

Mitchell Fang (1):
      thread-mainloop: Fix bug in example code

Nikolay Amiantov (1):
      pactl: Flush stdout buffer when printing subscribe events.

Parin Porecha (1):
      protocol-native: Express XOR asserts more concisely

Peter Meerwald (63):
      build: make ARM NEON check in configure.ac more strict
      fdsem: Remember pa_write() type in pa_fdsem_post()
      .gitignore: Ignore src/*.gcno files created when building with --enable-gcov
      cli: Fix plural in caption of source-outputs list
      pactl: Use colon to separate priority in profiles output
      intended-roles: Fix typo in module description
      rescue-streams: Fix wording of module description
      sconv: avoid multiply in ARM NEON float->s16 conversion
      sconv: avoid multiply in ARM NEON s16->float conversion
      cli: Output prompt and welcome message only when requested by 'hello'
      pacmd: Discriminate between interactive and non-interactive mode
      cli: Use pa_xnew0 to save some 0 inits
      memblock: Fix typo in description of pa_memblock_unref_fixed(), cleanup
      tests: Resampler output memchunk may be reset (zero)
      modargs: Add pa_modargs_get_resample_method()
      modargs: Add pa_modargs_get_sample_rate()
      tests: Extract PA_CPU_TEST* macros to separate header
      resampler: Introduce work frame size (w_fz)
      resampler: Introduce work_channels
      resampler: Change interface, resampler may return the number of leftover frames
      resampler: Handle leftover samples, use new return value
      resampler: Cleanup, rename xxxx_buf_samples to xxxx_buf_size
      resampler: Add fit_buf() helper funtion for buffer allocation
      resampler: Prepare to have leftover data in different output buffers
      resampler: Cleanup, index of memchunks is not used
      resampler: Clean up ffmpeg resampler buffering
      modules: Make use of new function pa_modargs_get_sample_rate()
      module-sine: Add rate argument
      module-remap-sink: Add resample_method argument
      module-remap-source: Add resample_method argument
      core-util: Check that we actually have regexec before we use it
      modargs: Fix pa_modargs_get_sample_rate() if no "rate" is present in the modargs
      modargs: Do not modify alternate_rate in pa_modargs_get_alternate_sample_rate() on failure
      doc: Fix reference to pa_threaded_mainloop
      doc: Refer to seek parameter of pa_stream_write() in description of pa_stream_flush()
      doc: Document that pa_simple_get_latency() can be used on record streams as well
      doc: Document that pa_stream_flush() is applicable to playback and record streams
      simple: Enable pa_simple_flush() on record streams and amend documentation
      modules: Fix resource leak in stream-restore
      modules: Fix resource leak in alsa-card
      modules: Fix resource leak in device-restore
      modules: Fix resource leak in alsa-ucm
      modules: Fix resource leak in oss
      core: Fix resource leak in core-util
      modules: Fix resource leak in tunnel-sink-new
      modules: Fix resource leak in raop_client
      modules: Fix resource leak in card-restore
      core: Fix uninit pointer read in protocol-native
      modules: Fix uninit value in rtp-send
      core: Fix asserting command in protocol-native
      modules: Remove redundant code in virtual-surround-sink
      tests: Fix resource leak in sigbus-test
      core: Use after free in pa_sink_input_new_data_set_formats() and pa_source_output_new_data_set_formats()
      modules: Remove dead code in position-event-sounds
      core: Remove dead code in pa_mainloop_run()
      core: Fix check for io->output_event in enable_events()
      modules: Add assertion in equalizer-sink
      alsa-card: Don't free the modargs in pa__init
      resampler: Extend fit_buf() helper to copy leftover data to new buffer
      resampler: Implement leftover handling in convert_to_work_format()
      resampler: Resample first followed by remapping if have more out channels than in channels
      modules: Fix uninitialized variable in rtp-send
      doc: Assorted spelling and grammar fixes for doxygen output

Pierre Ossman (5):
      alsa: Support ALSA without a use case manager
      module-tunnel: automatically find the PulseAudio server
      daemon conf: "high-priority" setting isn't respected
      core: make sure win32 sockets remain blocking
      Revert "core: Transparently handle non-blocking sockets on Windows"

Piotr Drąg (4):
      i18n: sort POTFILES.in alphabetically
      i18n: add missing files to POTFILES.in
      i18n: remove unneeded files from POTFILES.in
      i18n: Update Polish translation

Rafael Ferreira (1):
      Updated Brazilian Portuguese translation

Ryan Lortie (3):
      bootstrap.sh: use /usr/bin/env to find bash
      core-util: set_scheduler: check for RLIMIT_RTTIME
      Add support for FreeBSD 

Scott Reeves (1):
      protocol-native: Remove written differently but functionally redundant check.

Shuai Fan (1):
      cli, log: Improve the set-log-target functionality

Stefan Sperling (1):
      Fix undefined behaviour in pulseaudio --start.

Tanu Kaskinen (126):
      switch-on-port-available: Prepare for dual-direction ports going away
      switch-on-port-available: Remove a redundant function argument
      switch-on-port-available: Fix error reporting style
      alsa-mixer: Introduce "description-key" option for paths
      alsa-mixer: Remove the "name" option from the "General" section of path configuration files
      device-port: Make it impossible to have dual-direction ports
      .gitignore: Update m4/ ignore list
      core-util: Don't accept random words in pa_parse_boolean()
      alsa: Don't use pa_strna() for port names
      pasuspender: Resume before exiting in case of SIGINT or fork() failure.
      module: Assign the index before calling init()
      bluetooth: Fix error checking style
      bluetooth: Fix input port description
      .gitignore: Add lo-latency-test
      native: Don't enable prebuffering when draining
      combine: Fix crash in output freeing
      bluetooth: Fix too late assertion
      log: Fix error message
      core-util: Fix log message levels
      .gitignore: Add build-aux
      device-port: Assert that ports have a description
      object: Initialize allocated memory to zero
      source: Fix an indentation mistake
      dynarray: Reimplement with nicer semantics
      alsa-mixer: Add a couple of assertions
      waveout: Simplify setting the device description
      device-manager: Add a sanity check for reading entries
      device-manager: Refuse to load empty descriptions
      build-sys: Add native-instruments-traktor-audio2.conf to Makefile.am
      mainloop-test: Use FALSE instead of false for gboolean
      channelmap, volume: Don't refer to bool in the public API documentation
      pulse: Add verbose volume printing functions
      Use pa_(c)volume_snprint_verbose() everywhere
      sink-input: Keep reference ratio up-to-date
      Revert "resampler: Resample first followed by remapping if have more out channels than in channels"
      module: Fix crash in pa_module_unload_all()
      log: Fix thread name printing
      caps: Fix compilation on Windows
      tunnel: Fix reading state from wrong variable
      log: Remove pa_log_set_fd()
      log: Use pa_write() instead of write()
      log: Harmonize log target output
      ucm: Fix an incorrect log message
      sink: Add missing curly braces
      dbus: Fix invalid dbus_bool_t values
      vala: Add CVolume.set()
      loopback: Increase the maximum allowed latency
      resampler: Fix peaks resampler's channel handling
      source-output: Get the correct source for "direct_on_input" streams
      sink, source: Reduce indentation level in *_update_rate()
      sink, source: Don't care about default and alternate rate in passthrough mode
      sink, source: Fix default and alternate rate assertions
      sink, source: Small readability improvement
      sink, source: Return early from pa_*_update_rate(), if there's no need to do anything
      source: Fix monitor source rate changing
      source: When updating a monitor source's rate, update the sink rate too
      sink, source: Fix error reporting style for rate updates
      resampler: Never return zero for max block size
      suspend-on-idle: Allow disabling suspending for specific devices
      .gitignore: Add .dirstamp
      bluetooth: Add a comment about messing with the IDLE suspend cause
      build-sys: Don't define _FORTIFY_SOURCE when building with -O0
      build-sys: Print CPPFLAGS in configure
      bluetooth: Remove adapter_remove_all()
      bluetooth: Fix variable constness
      simple: Improve pa_simple_read() documentation
      sink-input, source-output: Don't assume that proplist has been initialized in free()
      bluetooth: Remove a redundant variable
      bluetooth: Use a helper function for setting device_info_valid
      bluetooth: Don't mark device valid before it has an adapter
      bluetooth: Fire DEVICE_CONNECTION_CHANGED in set_device_info_valid()
      bluetooth: Remove device_remove_all()
      bluetooth: Set device_info_valid to -1 when the device's adapter disappears
      source-output: Log the format negotiation result
      sink, source: Add a TODO comment about removing get_formats()
      idxset: Allow deep copying with pa_idxset_copy()
      ucm: Add a FIXME comment about bad error handling
      Pass the profile object instead of the profile name to pa_card_set_profile()
      card: Move pa_card_add_profile() implementation to a more logical place
      bluetooth: Fix profile priority comparison
      Add pa_sample_format_valid()
      Add pa_sample_rate_valid()
      Add pa_channels_valid()
      sink-input: Fix volume_factor_sink initialization
      caps: Improve log message
      caps: Fix platform checking
      caps: Remove the FreeBSD implementation of pa_drop_caps()
      volume-test: Increase the allowed number of rouding errors
      Move pa_format_info_to_sample_spec_fake() to core-format
      core-format: Add pa_format_info_get_sample_format()
      core-format: Add pa_format_info_get_rate()
      core-format: Add pa_format_info_get_channels()
      core-format: Add pa_format_info_get_channel_map()
      format: Simplify pa_format_info_to_sample_spec()
      core-format: Add pa_format_info_to_sample_spec2()
      core-format: Add pa_format_info_from_sample_spec2()
      sink-input, source-output: Do routing related validity checks immediately after routing
      format, core-format: Constify some function parameters
      stream-util: Add pa_stream_get_volume_channel_map()
      stream: Remove a volume channel validity check
      sink-input, source-output: Interpret missing PCM parameters in format info as a request to decide those parameters at the server end
      def, format: Document how to leave PCM parameters to be decided by the server
      stream: Improve pa_stream_connect_playback() documentation
      stream: Mention pa_stream_new_extended() in the high-level stream creation documentation
      format: Add some error logging
      caps: Use standard macro when checking for Linux
      build-sys: Fix building with --as-needed
      alsa: Mute "Bass Speaker" when using headphones or lineout
      Revert "ladspa-sink: Handle empty chunks in sink_input_pop_cb"
      sink, source: Remove useless attach/detach stuff
      introspect, thread-mainloop: Fix Doxygen "since" tags
      build-sys: Bump sonames
      build-sys: Add tests/runtime-test-util.h to the tarball
      caps: Don't require --without-caps on FreeBSD
      device-manager: Don't try to use unlinked devices
      module: Replace rindex() with strrchr()
      card-restore: Fix a memory leak
      thread-win32: Implement pa_thread_free_nojoin()
      protocol-native: Fix a crash
      source: Increase max outputs per source
      man: pulse-cli-syntax: Document the "journal" log target
      sink-input, source-output: Don't crash if format negotiation fails
      rtp-send: Use _cb suffix with callbacks
      rtp-send: Add "inhibit_auto_suspend" module argument
      Update NEWS
      suspend-on-idle: Fix monitor source handling

Wim Taymans (2):
      daemon: Remove deprecated Encoding from .desktop files
      man: improve man page and help for cmdline options

Yuri Chornoivan (1):
      Update Ukrainian translation

poljar (Damir Jelić) (38):
      device-port: Introduce pa_device_port_new_data
      switch-on-port-available: Silence gcc warning.
      shell-completion: Add list-cards to the pacmd completion.
      Whitespace cleanup: Remove tabs
      Whitespace cleanup: Fix broken indentation by hand
      Style fix: Add a space betwen the if statement and the opening bracket
      Style fix: Add a space between the closing/opening bracket
      Style fix: Remove new lines from opening brackets
      Whitespace cleanup: Remove all multiple newlines
      Remove pa_bool_t and replace it with bool.
      Fix alignment issues introduced by the pa_bool -> bool conversion
      Revert dbus_bool_t variables to use TRUE/FALSE instead of true/false
      Revert windows BOOL variables to use TRUE/FALSE instead of true/false
      macro: Remove unneeded typedef for pa_bool_t.
      bash-completion: Fix device completion for pacat
      resampler: Make resampler struct implementation agnostic
      resampler: Move the fix method logic into a separate function
      resampler: fix_method use switch instead of if/else
      resampler: Move the work format finding logic into a separate function
      resampler: choose_work_format use switch instead of if/else
      resampler: Add a function for comparing sample format precision
      Resampler: Don't use the peaks resampler for upsampling
      resampler: Introduce a implementation specific struct
      resampler: Add assert to set_input/output_rate
      resampler: Get rid of redundant implementation specific structs
      resampler: Add a choose_auto_resampler function
      card-restore: Watch for profiles added after card creation.
      build-sys: Enable subdir-objects option
      build-sys: Silence warning about missing AC_LANG_SOURCE call in configure.ac
      build-sys: Silence warning about missing AC_LANG_SOURCE call in acx_libwrap.m4
      build-sys: Update ax_pthread macro
      resampler: Revert to auto if the rates are equal and copy is chosen
      bash-completion: Add --monitor-stream completion to pacat
      zsh-completion: Fix pacat device completion
      zsh-completion: Add --monitor-stream completion to pacat
      log: Add support for the systemd journal
      bluez5-util: Free the adapter_path in device_free().
      alsa-mixer: Free the description key in alsa_path_free().

victory (1):
      i18n: Japanese translation update