platform/upstream/pulseaudio.git
10 years agosink/source: Initialize port before fixate hook (fixes volume/mute not saved)
David Henningsson [Fri, 21 Mar 2014 09:19:19 +0000 (10:19 +0100)]
sink/source: Initialize port before fixate hook (fixes volume/mute not saved)

In case a port has not yet been saved, which is e g often the case
if a sink/source has only one port, reading volume/mute will be done
without port, whereas writing volume/mute will be done with port.

Work around this by setting a default port before the fixate hook,
so module-device-restore can read volume/mute for the correct port.

BugLink: https://bugs.launchpad.net/bugs/1289515
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
10 years agointrospect: Use pa_direction_valid() to validate port direction
Tanu Kaskinen [Wed, 26 Mar 2014 09:52:34 +0000 (11:52 +0200)]
introspect: Use pa_direction_valid() to validate port direction

There's no behavioral change, just nicer code.

10 years agodirection: Add a couple of direction helper functions
Tanu Kaskinen [Wed, 26 Mar 2014 09:52:33 +0000 (11:52 +0200)]
direction: Add a couple of direction helper functions

10 years agodevice-restore: log restored mute state
Wim Taymans [Wed, 26 Mar 2014 14:57:16 +0000 (15:57 +0100)]
device-restore: log restored mute state

Log the mute state as it was restored

10 years agoalsa-util: Reset hwparams_copy before the second try of buffer setup
Peter Ujfalusi [Fri, 21 Mar 2014 07:18:40 +0000 (09:18 +0200)]
alsa-util: Reset hwparams_copy before the second try of buffer setup

hwparams_copy needs to be reset (as it is also reset for the third and
fourth try) before the second try.

If the reset is not done and the first try fails:
D: [lt-pulseaudio] alsa-util.c: Maximum hw buffer size is 743 ms
I: [lt-pulseaudio] alsa-util.c: snd_pcm_hw_params_set_buffer_size_near() failed: Invalid argument
I: [lt-pulseaudio] alsa-util.c: snd_pcm_hw_params_set_period_size_near() failed: Invalid argument
I: [lt-pulseaudio] alsa-util.c: snd_pcm_hw_params_set_buffer_size_near() failed: Invalid argument
D: [lt-pulseaudio] alsa-util.c: Set only period size (to 1102 samples).

We have three failures and finally the fourth (only period size) succeed.

With this patch:
D: [lt-pulseaudio] alsa-util.c: Maximum hw buffer size is 743 ms
I: [lt-pulseaudio] alsa-util.c: snd_pcm_hw_params_set_buffer_size_near() failed: Invalid argument
D: [lt-pulseaudio] alsa-util.c: Set period size first (to 1102 samples), buffer size second (to 4408 samples).

We only fail with the first try, the second (period followed by buffer) is
fine.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
10 years agodoxygen: Update the configuration file
Tanu Kaskinen [Sat, 15 Mar 2014 10:43:12 +0000 (12:43 +0200)]
doxygen: Update the configuration file

I reviewed the changes, and I didn't find anything that would actually
have effects on the generated output.

10 years agozeroconf-publish: Don't react to messages while shutting down
Tanu Kaskinen [Sat, 15 Mar 2014 07:37:06 +0000 (09:37 +0200)]
zeroconf-publish: Don't react to messages while shutting down

This fixes a case where pa__done() is called while
AVAHI_MESSAGE_PUBLISH_ALL is waiting for processing. The
pa_asyncmsgq_wait_for(AVAHI_MESSAGE_SHUTDOWN_COMPLETE) call will
process all pending messages, and processing AVAHI_MESSAGE_PUBLISH_ALL
causes publish_all_services(), and that in turn accesses u->services,
which has been already freed at this point. If we are shutting down,
we shouldn't react to any of the messages that the Avahi thread is
sending to the main thread.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=76184
10 years agozeroconf-publish: Add locking around pa_mainloop_api_once()
Tanu Kaskinen [Sat, 15 Mar 2014 07:37:05 +0000 (09:37 +0200)]
zeroconf-publish: Add locking around pa_mainloop_api_once()

Commit e1e154c7377779377fa1a36655a10effd693f7b5 added locking around
pa_mainloop_api_once() everywhere except pa__done(). This fixes that
omission.

10 years agopacat: Fix partially translated message
Tanu Kaskinen [Mon, 24 Mar 2014 07:22:20 +0000 (09:22 +0200)]
pacat: Fix partially translated message

The word "not" was not translated.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=76529
10 years agocore-util: Make pa_yes_no() translatable
Tanu Kaskinen [Mon, 24 Mar 2014 07:17:53 +0000 (09:17 +0200)]
core-util: Make pa_yes_no() translatable

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=76529
10 years agodoc: Fix some typos/wording in man/pulse-daemon.conf.5
Peter Meerwald [Mon, 17 Mar 2014 14:36:22 +0000 (15:36 +0100)]
doc: Fix some typos/wording in man/pulse-daemon.conf.5

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agomainloop-api: Improve pa_mainloop_api_once() documentation
Tanu Kaskinen [Sat, 15 Mar 2014 07:52:07 +0000 (09:52 +0200)]
mainloop-api: Improve pa_mainloop_api_once() documentation

10 years agovolume: Increase PA_SW_VOLUME_SNPRINT_DB_MAX
Tanu Kaskinen [Sun, 9 Mar 2014 08:56:07 +0000 (10:56 +0200)]
volume: Increase PA_SW_VOLUME_SNPRINT_DB_MAX

10 bytes isn't enough for "-123.45 dB", including the terminating null
byte.

10 years agoalsa: Use card description in default sink/source prefix when available
Pete Beardmore [Thu, 13 Mar 2014 10:14:40 +0000 (10:14 +0000)]
alsa: Use card description in default sink/source prefix when available

When given an explicit device.description in card_properties, prefer
this information over other default prefixes (e.g. 'Built-in Audio')
when constructing sink/source descriptions.

For example, if I manually configure the card description to be
"FooBar", I then expect that the sinks and created by the card also
have "FooBar" in their description instead of generic "Built-in
Audio".

10 years agotunnel: Fix inverted if condition
Tanu Kaskinen [Thu, 13 Mar 2014 11:04:55 +0000 (13:04 +0200)]
tunnel: Fix inverted if condition

10 years agoalsa-mixer: Fix Analog Input showing up on USB Headset
David Henningsson [Tue, 11 Mar 2014 04:50:10 +0000 (05:50 +0100)]
alsa-mixer: Fix Analog Input showing up on USB Headset

In some cases, "Analog Input" could show up as well as
"Headset Mic" (or "Headphone Mic"), because I forgot to add the
relevant "required-absent" lines when I added the headset mic path.

As a result, both "Analog Input" and "Headset Mic" showed up on the
Logitech USB 530 Headset.

Reported-by: Steve Magoun <steve.magoun@canonical.com>
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
10 years agopactl: Fix a copy-paster error
Tanu Kaskinen [Mon, 10 Mar 2014 10:48:23 +0000 (12:48 +0200)]
pactl: Fix a copy-paster error

10 years agodoc: Use \since doxygen command correctly in mainloop-api.h
Peter Meerwald [Fri, 7 Mar 2014 12:08:35 +0000 (13:08 +0100)]
doc: Use \since doxygen command correctly in mainloop-api.h

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agocore: add missing SET_PORT_LATENCY_OFFSET to dispatcher opcode string table
Peter Meerwald [Fri, 7 Mar 2014 11:33:06 +0000 (12:33 +0100)]
core: add missing SET_PORT_LATENCY_OFFSET to dispatcher opcode string table

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agodoc: Use correct parameter name wait_for_accept in documentation of pa_threaded_mainl...
Peter Meerwald [Fri, 7 Mar 2014 12:34:41 +0000 (13:34 +0100)]
doc: Use correct parameter name wait_for_accept in documentation of pa_threaded_mainloop_signal()

also sneaks in a full stop in documentation elsewhere

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agoFix a few "it's -> its" typos
Alexander E. Patrakov [Fri, 7 Mar 2014 15:26:28 +0000 (21:26 +0600)]
Fix a few "it's -> its" typos

10 years agotunnel-source-new: Fix shadow compiler warning
Peter Meerwald [Thu, 6 Mar 2014 12:15:12 +0000 (13:15 +0100)]
tunnel-source-new: Fix shadow compiler warning

  CC     modules/module_tunnel_sink_la-module-tunnel.lo
modules/module-tunnel-source-new.c: In function 'read_new_samples':
modules/module-tunnel-source-new.c:145:16: warning: declaration of 'read' shadows a global declaration [-Wshadow]

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agodoc: Fix pa_stream_set_latency_update_callback() documentation
Peter Meerwald [Thu, 6 Mar 2014 09:31:32 +0000 (10:31 +0100)]
doc: Fix pa_stream_set_latency_update_callback() documentation

the latency_update callback is supported on playback AND record
streams

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agodoc: Fix a couple of 'to to' typos
Peter Meerwald [Thu, 6 Mar 2014 09:25:35 +0000 (10:25 +0100)]
doc: Fix a couple of 'to to' typos

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agoalsa: Remove a redundant check
Tanu Kaskinen [Wed, 29 Jan 2014 18:59:00 +0000 (20:59 +0200)]
alsa: Remove a redundant check

If mixer_handle is not NULL, then hctl_handle won't be NULL either.
The redundant check was confusing, because it looked like we would
leak the mixer_handle if mixer_handle is non-NULL and hctl_handle is
NULL.

10 years agobluetooth: Fix timing to count based on decoded data
James Bunton [Sun, 26 Jan 2014 14:14:39 +0000 (01:14 +1100)]
bluetooth: Fix timing to count based on decoded data

Currently the latency information is being updated based on the encoded
SBC data instead of the decoded PCM data. Fixing this required moving
the timing update to be after the packet has been decoded.

10 years agobluetooth: Don't abort on SBC decoding error
James Bunton [Sun, 26 Jan 2014 14:14:38 +0000 (01:14 +1100)]
bluetooth: Don't abort on SBC decoding error

The Nokia E7 running Symbian Belle Refresh seems to generate invalid SBC
packets every few minutes. This causes pulseaudio to disconnect the
stream and log "SBC decoding error (-3)".

If a single packet is bad, pulseaudio should keep playing the stream.

10 years agoRemove all references to the removed KDE files
Tanu Kaskinen [Sun, 26 Jan 2014 15:40:31 +0000 (17:40 +0200)]
Remove all references to the removed KDE files

10 years agodaemon: ensure pulseaudio early startup on kde
Rex Dieter [Wed, 22 Jan 2014 15:43:27 +0000 (09:43 -0600)]
daemon: ensure pulseaudio early startup on kde

10 years agodaemon: consolidate startup scripts
Rex Dieter [Wed, 22 Jan 2014 15:41:35 +0000 (09:41 -0600)]
daemon: consolidate startup scripts

simplify pulseaudio autostart into one, avoid needless extra
work for kde, and avoid possible startup races.

10 years agosuspend-on-idle: Fix monitor source handling upstream/5.0 v5.0
Tanu Kaskinen [Mon, 3 Mar 2014 14:21:21 +0000 (16:21 +0200)]
suspend-on-idle: Fix monitor source handling

Fixes a crash that happened due to calling pa_source_check_suspend()
when d->source was NULL.

10 years agoUpdate NEWS
Tanu Kaskinen [Mon, 3 Mar 2014 13:42:21 +0000 (15:42 +0200)]
Update NEWS

10 years agortp-send: Add "inhibit_auto_suspend" module argument
Tanu Kaskinen [Sun, 23 Feb 2014 17:47:15 +0000 (19:47 +0200)]
rtp-send: Add "inhibit_auto_suspend" module argument

Some people want module-rtp-send to send silence when the sink that is
monitored goes idle, and some people want module-rtp-send to pause the
RTP stream to avoid unnecessary bandwidth consumption.

10 years agosuspend-on-idle: Ensure we still time out if a stream remains corked
David Henningsson [Tue, 25 Feb 2014 15:56:49 +0000 (16:56 +0100)]
suspend-on-idle: Ensure we still time out if a stream remains corked

If a stream is started corked and remains corked, the sink/source
remained idle without being properly suspended. This patch fixes
that issue.

BugLink: https://bugs.launchpad.net/bugs/1284415
Tested-by: Ricardo Salveti <ricardo.salveti@canonical.com>
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
10 years agosink-input/source-output: Fix LFE remixing suddenly enabled
David Henningsson [Fri, 28 Feb 2014 11:16:54 +0000 (12:16 +0100)]
sink-input/source-output: Fix LFE remixing suddenly enabled

Steps to reproduce:
 1) Leave LFE remixing disabled (the default)
 2) Start playback of stereo material on e g 5.1 surround, notice nothing in LFE
 3) Now change profile to e g 4.0 surround and then back to 5.1 surround
 4) Notice that LFE channel is now remixed

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
10 years agoMake tests compile on FreeBSD
Koop Mast [Sat, 22 Feb 2014 23:56:59 +0000 (00:56 +0100)]
Make tests compile on FreeBSD

cpu_set_t is named cpuset_t on FreeBSD.
Add headers to allow for warning free compilation.

10 years agortp-send: Use _cb suffix with callbacks
Tanu Kaskinen [Sun, 23 Feb 2014 17:41:42 +0000 (19:41 +0200)]
rtp-send: Use _cb suffix with callbacks

10 years agodoc: Assorted spelling and grammar fixes for doxygen output
Peter Meerwald [Thu, 20 Feb 2014 20:41:47 +0000 (21:41 +0100)]
doc: Assorted spelling and grammar fixes for doxygen output

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agomodules: Fix uninitialized variable in rtp-send
Peter Meerwald [Fri, 3 Jan 2014 22:52:19 +0000 (23:52 +0100)]
modules: Fix uninitialized variable in rtp-send

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agosconv, svolume: Fix compilation on 32-bit FreeBSD
Koop Mast [Thu, 20 Feb 2014 23:21:46 +0000 (00:21 +0100)]
sconv, svolume: Fix compilation on 32-bit FreeBSD

Don't try to compile this code on 32-bit FreeBSD, it will error out complain
about registers only being available in 64-bit mode.

10 years agoRevert "protocol-native: Don't enumerate unlinked sinks and sources"
Arun Raghavan [Mon, 17 Feb 2014 11:02:40 +0000 (16:32 +0530)]
Revert "protocol-native: Don't enumerate unlinked sinks and sources"

This reverts commit bd0bd9143f8159e5475b1295e30ec026ff0b6fef.

10 years agobuild-sys: Bump soname v4.99.4
Arun Raghavan [Sat, 15 Feb 2014 04:34:39 +0000 (10:04 +0530)]
build-sys: Bump soname

Bumping from 0->3 since we missed bumping correctly for v4.99.{1,2,3}

10 years agoprotocol-native: Don't enumerate unlinked sinks and sources
Arun Raghavan [Sun, 8 Dec 2013 22:18:00 +0000 (14:18 -0800)]
protocol-native: Don't enumerate unlinked sinks and sources

This makes sure that there is no window between pa_sink/source_new()
and _put() where enumerating sinks/sources causes an assert (several
calls in sink/source_get_info need a linked sink or source).

10 years agosink-input, source-output: Don't crash if format negotiation fails
Tanu Kaskinen [Wed, 12 Feb 2014 10:41:34 +0000 (12:41 +0200)]
sink-input, source-output: Don't crash if format negotiation fails

A segfault was reported on this line:

pa_return_val_if_fail(PA_SINK_IS_LINKED(pa_sink_get_state(data->sink)), -PA_ERR_BADSTATE);

After expanding the pa_sink_get_state() macro, the line looks like
this:

pa_return_val_if_fail(PA_SINK_IS_LINKED(data->sink->state), -PA_ERR_BADSTATE);

So data->sink was apparently NULL. That could happen if we try to fall
back to the default sink, but format negotiation fails.

This bug was introduced in commit
71816ecb7f81fe7a124cefa2258cfa64046908a1.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=74646
10 years agoi18n: Update Polish translation
Piotr Drąg [Tue, 11 Feb 2014 14:18:08 +0000 (15:18 +0100)]
i18n: Update Polish translation

10 years agoman: pulse-cli-syntax: Document the "journal" log target
Tanu Kaskinen [Wed, 12 Feb 2014 12:31:26 +0000 (14:31 +0200)]
man: pulse-cli-syntax: Document the "journal" log target

10 years agodbus: Use correct idxset when getting sources
Juho Hämäläinen [Mon, 10 Feb 2014 10:46:28 +0000 (12:46 +0200)]
dbus: Use correct idxset when getting sources

This fixes get_sources() which iterates wrong idxset when
getting source object paths in GetSources or GetAll.

10 years agoman: improve man page and help for cmdline options
Wim Taymans [Thu, 6 Feb 2014 15:08:50 +0000 (16:08 +0100)]
man: improve man page and help for cmdline options

10 years agosource: Increase max outputs per source
Tanu Kaskinen [Wed, 5 Feb 2014 15:55:44 +0000 (17:55 +0200)]
source: Increase max outputs per source

This change is just for being symmetric with sinks, which got this
change in 26bf2df903b1c7439ad3eef0dc3bb1e777835cc3.

10 years agoprotocol-native: Fix a crash
Tanu Kaskinen [Tue, 4 Feb 2014 19:28:44 +0000 (21:28 +0200)]
protocol-native: Fix a crash

Without this check pa_source_output_set_volume() crashes if
so->volume_writable is false. The same check already exists for sink
inputs.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=74529
10 years agothread-win32: Implement pa_thread_free_nojoin()
Tanu Kaskinen [Sun, 26 Jan 2014 17:11:07 +0000 (19:11 +0200)]
thread-win32: Implement pa_thread_free_nojoin()

This fixes a build error with mingw32:

pulsecore/.libs/libpulsecommon_4.99_la-lock-autospawn.o: In function `unref':
/home/abuild/rpmbuild/BUILD/pulseaudio-4.99.2/src/pulsecore/lock-autospawn.c:123: undefined reference to `pa_thread_free_nojoin'
collect2: error: ld returned 1 exit status

pa_thread_free_nojoin() was initially only implemented for the pthread
based pa_thread backend, because it was incorrectly assumed that
autospawning (the only user of pa_thread_free_nojoin()) is not used on
Windows.

Reported-By: Michael DePaulo <mikedep333@gmail.com>
10 years agocard-restore: Fix a memory leak v4.99.3
Tanu Kaskinen [Wed, 29 Jan 2014 18:54:39 +0000 (20:54 +0200)]
card-restore: Fix a memory leak

entry_read() may set entry->profile, so we need to free that string
before we can replace it with a new string.

10 years agoresampler: Resample first followed by remapping if have more out channels than in...
Peter Meerwald [Mon, 2 Dec 2013 16:27:01 +0000 (17:27 +0100)]
resampler: Resample first followed by remapping if have more out channels than in channels

Reintroduces a cleaned-up version of commit 30ce3a14e5ae1cd316a18bec95b831c07ac57a1a which
was reverted by 1ce71cbd8206d1be59ac62274ad83cdbe693a96a; for more information see
http://thread.gmane.org/gmane.comp.audio.pulseaudio.general/17479/focus=17487

The patch intends to reduce computational load when resampling AND remapping. The PA
resampler performs the following steps:

sample format conversion -> remapping -> resampling -> sample format conversion

In case the number of output channels is higher than the number of input channels, the
resampler has to be run more often than necessary. E.g. in case of mono to 4-channel remapping,
the resampler runs on 4 channels separately.

To ímprove this, the PA resampler pipeline is made adaptive:

if out-channels <= in-channels:
  sample format conversion -> remapping -> resampling -> sample format conversion
if out-channels > in-channels:
  sample format conversion -> resampling -> remapping -> sample format conversion

Signed-off-by: Peter Meerwald <p.meerwald@bct-electronic.com>
10 years agoresampler: Implement leftover handling in convert_to_work_format()
Peter Meerwald [Mon, 2 Dec 2013 16:27:00 +0000 (17:27 +0100)]
resampler: Implement leftover handling in convert_to_work_format()

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agoresampler: Extend fit_buf() helper to copy leftover data to new buffer
Peter Meerwald [Mon, 2 Dec 2013 16:26:59 +0000 (17:26 +0100)]
resampler: Extend fit_buf() helper to copy leftover data to new buffer

the patch changes the interface of the (internal) fit_buf() function:
fit_buf() manages the memblock of the buf chunk, it reallocates the memblock
if the requested number of bytes ('len') if larger than the memblock's size ('size')
and optionally preserves 'copy' bytes

the code should be in line with the comment now

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
10 years agomodule: Replace rindex() with strrchr()
Tanu Kaskinen [Mon, 27 Jan 2014 19:00:58 +0000 (21:00 +0200)]
module: Replace rindex() with strrchr()

rindex() appears to be "non-standard" to an extent, and it caused a
build failure on mingw32.

From the man page of rindex(): "POSIX.1-2008 removes the
specifications of index() and rindex(), recommending strchr(3) and
strrchr(3) instead."

10 years agodevice-manager: Don't try to use unlinked devices
Tanu Kaskinen [Sun, 26 Jan 2014 18:13:50 +0000 (20:13 +0200)]
device-manager: Don't try to use unlinked devices

This fixes an assertion crash:

[pulseaudio] source.c: Assertion 'PA_SOURCE_IS_LINKED(s->state)' failed at pulsecore/source.c:734, function pa_source_update_status(). Aborting.

The crash happened when a Bluetooth headset profile was changed from
a2dp to hsp. During the profile change three devices are created:
a sink, a monitor source for the sink, and a regular source. First
pa_sink/source_new() are called for each device, and that puts the
devices to u->core->sinks/sources. Then, pa_sink_put() is called for
the sink, and that in turn calls pa_source_put() for the source. At
that point module-device-manager decides to reroute all source
outputs. The non-monitor source that the Bluetooth card created hasn't
been linked yet at this stage, because it will only be linked after
the sink and the monitor source have been linked. So,
module-device-manager should take into account during the rerouting
that not all sinks and sources are necessarily linked. This patch does
that.

Reported-By: Iskren Hadzhinedev <i.hadzhinedev@gmail.com>
10 years agocaps: Don't require --without-caps on FreeBSD
Tanu Kaskinen [Sun, 26 Jan 2014 14:41:29 +0000 (16:41 +0200)]
caps: Don't require --without-caps on FreeBSD

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=72580
10 years agodaemon: Remove deprecated Encoding from .desktop files
Wim Taymans [Thu, 23 Jan 2014 11:21:53 +0000 (12:21 +0100)]
daemon: Remove deprecated Encoding from .desktop files

Encoding is deprecated according to
http://standards.freedesktop.org/desktop-entry-spec/latest/apc.html

10 years agobuild-sys: Move ORC_SOURCE definitions outside of "if HAVE_ORC"
Dominique Leuenberger [Fri, 24 Jan 2014 20:45:52 +0000 (21:45 +0100)]
build-sys: Move ORC_SOURCE definitions outside of "if HAVE_ORC"

Not having ORC_SOURCE defined results in different tarballs depending on
whether the dev issuing 'make dist' has orc support enabled or disabled.

Specifying ORC_SOURCE unconditionally addresses that, without causing
negative effects on users not having orc in the end.

10 years agobuild-sys: Add tests/runtime-test-util.h to the tarball v4.99.2
Tanu Kaskinen [Thu, 23 Jan 2014 18:34:58 +0000 (20:34 +0200)]
build-sys: Add tests/runtime-test-util.h to the tarball

The file is used by three test binaries, but none of them mentioned
it in Makefile.am, which caused the file to be left out of the
tarball.

10 years agobuild-sys: Bump sonames v4.99.1
Tanu Kaskinen [Thu, 23 Jan 2014 10:05:52 +0000 (12:05 +0200)]
build-sys: Bump sonames

The bigger than usual bump in libpulse-simple was warranted by the
change in pa_simple_flush() that allows also record streams to be
flushed. There are no changes to the function signature, but it's in
practice a change in the ABI anyway, because new clients using the new
possibility won't work with older versions of the library.

libpulse-mainloop-glib got a bug fix in commit
68156d3f79b67c77f88050e6c9de180698377c64.

10 years agointrospect, thread-mainloop: Fix Doxygen "since" tags
Tanu Kaskinen [Thu, 23 Jan 2014 09:21:58 +0000 (11:21 +0200)]
introspect, thread-mainloop: Fix Doxygen "since" tags

10 years agocombine-sink: fix segfault
Daniel Albers [Thu, 16 Jan 2014 09:00:42 +0000 (10:00 +0100)]
combine-sink: fix segfault

10 years agoUpdate Russian translation
Alexander Potashev [Sat, 11 Jan 2014 16:27:18 +0000 (20:27 +0400)]
Update Russian translation

Translated in sync with the guidelines at http://l10n.lrn.ru/wiki/Pulseaudio

10 years agoUpdate Ukrainian translation
Yuri Chornoivan [Mon, 13 Jan 2014 16:53:46 +0000 (18:53 +0200)]
Update Ukrainian translation

10 years agosink, source: Remove useless attach/detach stuff
Tanu Kaskinen [Wed, 15 Jan 2014 16:59:18 +0000 (18:59 +0200)]
sink, source: Remove useless attach/detach stuff

The removed stuff wasn't used anywhere.

10 years agoRevert "ladspa-sink: Handle empty chunks in sink_input_pop_cb"
Tanu Kaskinen [Wed, 15 Jan 2014 12:24:25 +0000 (14:24 +0200)]
Revert "ladspa-sink: Handle empty chunks in sink_input_pop_cb"

This reverts commit dbe66b0b5e6b6e24df45bda8bf656836f18e7f4b.

I accidentally pushed this patch that was superseded by
e96785c1c34706e902f28c25844f31d8201db2b2.

10 years agoladspa-sink: Set a silence memchunk for the memblockq
Ben Brewer [Wed, 15 Jan 2014 11:19:46 +0000 (11:19 +0000)]
ladspa-sink: Set a silence memchunk for the memblockq

A crash was observed that was caused by pa_memblockq_peek() returning
a NULL memblock in sink_input_pop_cb(). The scenario where this was
happening was

1. Delete 2 rtp-recv's connected to a ladspa-sink
2. Delete ladspa-sink
3. Delete alsa-sink
4. Create alsa-sink
5. Create ladspa-sink
6. Create 2 rtp-recv's connected to the ladspa-sink

The crash was probably caused by a rewind that made the read index go
negative while the write index was at least zero, causing there to be
a gap in the memblockq. The problematic rewind might have been caused
by adding the rtp-recv stream to the ladspa-sink. That has not been
proven, but this looks very similar to the bug that was fixed in
module-virtual-sink in commit 6bd34156b130c07b130de10111a12ef6dab18b52.

10 years agoladspa-sink: Handle empty chunks in sink_input_pop_cb
Ben Brewer [Wed, 15 Jan 2014 10:30:36 +0000 (10:30 +0000)]
ladspa-sink: Handle empty chunks in sink_input_pop_cb

10 years agozeroconf-discover: Specify the interface when connecting to a link-local IPv6 address
Héctor Martín [Wed, 15 Jan 2014 10:20:27 +0000 (12:20 +0200)]
zeroconf-discover: Specify the interface when connecting to a link-local IPv6 address

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=40785
10 years agoalsa: Mute "Bass Speaker" when using headphones or lineout
Tanu Kaskinen [Wed, 15 Jan 2014 09:37:26 +0000 (11:37 +0200)]
alsa: Mute "Bass Speaker" when using headphones or lineout

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=73604
10 years agortp-send: Remove "don't inhibit auto-suspend" flag
Ben Brewer [Tue, 14 Jan 2014 12:26:47 +0000 (12:26 +0000)]
rtp-send: Remove "don't inhibit auto-suspend" flag

A source is no longer considered idle in pa_source_check_suspend when
module-rtp-send is consuming data from it.

10 years agobuild-sys: Fix building with --as-needed
Tanu Kaskinen [Sat, 11 Jan 2014 14:47:56 +0000 (16:47 +0200)]
build-sys: Fix building with --as-needed

pulsecore/core-format.c was recently added to libpulsecommon, and
core-format.c depends on functions in libpulse, which libpulsecommon
doesn't link to. That broke building with --as-needed. This patch adds
pulse/format.c to libpulsecommon, so that core-format.c doesn't need
to depend on libpulse any more. format.c pulls in also the dependency
to json-c.

Reported-By: Jan Steffens <jan.steffens@gmail.com>
10 years agortp-recv: Only try adding the group membership for multicast addresses
Laurentiu Nicola [Tue, 7 Jan 2014 18:26:50 +0000 (20:26 +0200)]
rtp-recv: Only try adding the group membership for multicast addresses

10 years agocaps: Use standard macro when checking for Linux
Tanu Kaskinen [Fri, 10 Jan 2014 10:41:00 +0000 (12:41 +0200)]
caps: Use standard macro when checking for Linux

linux and __linux are non-standard, according to
http://sourceforge.net/p/predef/wiki/OperatingSystems/

Reported-By: Damir Jelić <poljarinho@gmail.com>
10 years agoalsa-card: Don't free the modargs in pa__init
Peter Meerwald [Fri, 3 Jan 2014 22:21:26 +0000 (23:21 +0100)]
alsa-card: Don't free the modargs in pa__init

The modargs are in both cases (a succesfull as well as a failed module
initialization) freed already in pa__done().

To avoid leaking modargs memory before they are assigned to u->modargs, the
code is reorganized to first allocate userdata, and then allocate the modargs.
Local variable ma is not needed anymore.

discussion here
http://lists.freedesktop.org/archives/pulseaudio-discuss/2013-December/019661.html

Signed-off-by: Peter Meerwald <pmeerw@pmeerw.net>
Reported-by: poljar (Damir Jelić) <poljarinho@gmail.com>
10 years agoformat: Add some error logging
Tanu Kaskinen [Wed, 18 Dec 2013 16:28:55 +0000 (18:28 +0200)]
format: Add some error logging

10 years agostream: Mention pa_stream_new_extended() in the high-level stream creation documentation
Tanu Kaskinen [Wed, 4 Dec 2013 04:18:42 +0000 (06:18 +0200)]
stream: Mention pa_stream_new_extended() in the high-level stream creation documentation

10 years agostream: Improve pa_stream_connect_playback() documentation
Tanu Kaskinen [Tue, 17 Dec 2013 21:42:09 +0000 (23:42 +0200)]
stream: Improve pa_stream_connect_playback() documentation

10 years agodef, format: Document how to leave PCM parameters to be decided by the server
Tanu Kaskinen [Wed, 4 Dec 2013 04:17:11 +0000 (06:17 +0200)]
def, format: Document how to leave PCM parameters to be decided by the server

10 years agosink-input, source-output: Interpret missing PCM parameters in format info as a reque...
Tanu Kaskinen [Wed, 4 Dec 2013 04:09:23 +0000 (06:09 +0200)]
sink-input, source-output: Interpret missing PCM parameters in format info as a request to decide those parameters at the server end

The "fix flags" (PA_SINK_INPUT_FIX_FORMAT etc.) don't work properly
with the pa_stream_new_extended() interface. This patch fixes it so
that the same effect can be achieved by leaving some of the PCM
parameters unspecified in format info objects. Also, when converting
a sample spec to a format info when using the old pa_stream_new()
interface, the "fix flags" are taken into account in that conversion.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=68952
10 years agostream: Remove a volume channel validity check
Tanu Kaskinen [Tue, 17 Dec 2013 19:22:50 +0000 (21:22 +0200)]
stream: Remove a volume channel validity check

The check is done for clients that use pa_stream_new() but not for
clients that use pa_stream_new_extended(). This is inconsistent. We
could check that the volume channels match the channels set in the
format info struct that is passed to pa_stream_new_extended(), but
that doesn't work if the format info doesn't contain the channel
information (that can happen if the client wants the server to choose
the channel count for the stream). And it should also be possible to
pass a mono volume for a multi-channel stream. The check could be
extended to handle all these cases, but I don't see much point in
wasting time on that. The server will anyway validate the stream
parameters, it's not particularly important to fail already when the
stream is being created at the client side.

10 years agostream-util: Add pa_stream_get_volume_channel_map()
Tanu Kaskinen [Tue, 17 Dec 2013 19:17:55 +0000 (21:17 +0200)]
stream-util: Add pa_stream_get_volume_channel_map()

The new function isn't used yet, but it soon will.

10 years agoformat, core-format: Constify some function parameters
Tanu Kaskinen [Tue, 17 Dec 2013 19:14:18 +0000 (21:14 +0200)]
format, core-format: Constify some function parameters

10 years agosink-input, source-output: Do routing related validity checks immediately after routing
Tanu Kaskinen [Tue, 3 Dec 2013 14:54:48 +0000 (16:54 +0200)]
sink-input, source-output: Do routing related validity checks immediately after routing

It's more logical that way.

10 years agocore-format: Add pa_format_info_from_sample_spec2()
Tanu Kaskinen [Tue, 3 Dec 2013 15:49:51 +0000 (17:49 +0200)]
core-format: Add pa_format_info_from_sample_spec2()

The function will be used in pa_sink_input_new() and
pa_source_output_new() to convert the sample spec given by the client
to a format info object. The set_format, set_rate and set_channels
will be set according to the stream flags (PA_SINK_INPUT_FIX_FORMAT
etc.).

10 years agocore-format: Add pa_format_info_to_sample_spec2()
Tanu Kaskinen [Mon, 18 Nov 2013 17:58:37 +0000 (19:58 +0200)]
core-format: Add pa_format_info_to_sample_spec2()

The function will be used in pa_sink_input_new() and
pa_source_output_new(). The fallback parameters are used to merge the
data in the format info with the sink/source sample spec and channel
map, when the format info is lacking some information.

10 years agoformat: Simplify pa_format_info_to_sample_spec()
Tanu Kaskinen [Tue, 3 Dec 2013 14:45:45 +0000 (16:45 +0200)]
format: Simplify pa_format_info_to_sample_spec()

10 years agocore-format: Add pa_format_info_get_channel_map()
Tanu Kaskinen [Mon, 25 Nov 2013 12:59:44 +0000 (14:59 +0200)]
core-format: Add pa_format_info_get_channel_map()

10 years agocore-format: Add pa_format_info_get_channels()
Tanu Kaskinen [Mon, 25 Nov 2013 12:51:37 +0000 (14:51 +0200)]
core-format: Add pa_format_info_get_channels()

This also fixes an issue in pa_format_info_to_sample_spec(): it did
no validation for the channels value. Now the validation is taken care
of in pa_format_info_get_channels().

10 years agocore-format: Add pa_format_info_get_rate()
Tanu Kaskinen [Mon, 25 Nov 2013 12:34:58 +0000 (14:34 +0200)]
core-format: Add pa_format_info_get_rate()

This also fixes an issue in pa_format_info_to_sample_spec(): it did
no validation for the rate value. Now the validation is taken care of
in pa_format_info_get_rate().

10 years agocore-format: Add pa_format_info_get_sample_format()
Tanu Kaskinen [Mon, 25 Nov 2013 14:15:54 +0000 (16:15 +0200)]
core-format: Add pa_format_info_get_sample_format()

10 years agoMove pa_format_info_to_sample_spec_fake() to core-format
Tanu Kaskinen [Mon, 18 Nov 2013 17:43:48 +0000 (19:43 +0200)]
Move pa_format_info_to_sample_spec_fake() to core-format

I will need to use the function from outside libpulse.

I added the channel map argument, because the function will be called
from another function that is expected to initialize the channel map.
I don't know if it's in practice necessary, but it shouldn't do any
harm either.

10 years agovolume-test: Increase the allowed number of rouding errors
Tanu Kaskinen [Sat, 14 Dec 2013 07:21:22 +0000 (09:21 +0200)]
volume-test: Increase the allowed number of rouding errors

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=72374
10 years agomodule-rtp-recv: Add an argument for latency
Laurentiu Nicola [Thu, 26 Dec 2013 13:46:50 +0000 (15:46 +0200)]
module-rtp-recv: Add an argument for latency

10 years agocaps: Remove the FreeBSD implementation of pa_drop_caps()
Tanu Kaskinen [Fri, 3 Jan 2014 11:31:29 +0000 (13:31 +0200)]
caps: Remove the FreeBSD implementation of pa_drop_caps()

Quoting Ryan Lortie from [1]:

I assumed from my reading of the Linux code ("cap_clear()...") that it
was clearing all capabilities of the process when in fact it is only
clearing the "special to root" capabilities.

The FreeBSD version of the code indeed clears _all_ capabilities beyond
ones that the process already has (ie: cannot open any new files, create
sockets, etc.)

This has a pretty obvious adverse effect on pulseaudio's ability to do
what it needs to do -- indeed, it bombs out pretty quickly due to an
inability to read its own config file.

[1] https://bugs.freedesktop.org/show_bug.cgi?id=72580#c11

10 years agoalsa-mixer: Free the description key in alsa_path_free().
poljar (Damir Jelić) [Tue, 24 Dec 2013 21:35:41 +0000 (22:35 +0100)]
alsa-mixer: Free the description key in alsa_path_free().

10 years agobluez5-util: Free the adapter_path in device_free().
poljar (Damir Jelić) [Tue, 24 Dec 2013 21:35:40 +0000 (22:35 +0100)]
bluez5-util: Free the adapter_path in device_free().

10 years agoDocumentation of known misuse of PulseAudio API
Alexander E. Patrakov [Mon, 21 Oct 2013 23:19:27 +0000 (00:19 +0100)]
Documentation of known misuse of PulseAudio API

Hello.

Over time, I became aware of several instances of tempting but
semantically incorrect usage of PulseAudio API (one from my own bad
proposal of "improving" Wine, one from Parole media player and one
from Webkit-GTK). I want to document these gotchas so that other
developers don't fall for that. See the attached patch.

I have checked that the rendered HTML is correct, but need someone to
confirm the factual accuracy of the proposed changes and, possibly, to
improve the wording.

--
Alexander E. Patrakov