profile/ivi/pulseaudio.git
13 years agoformats: The format code should be in libpulse, not libpulsecommon
Colin Guthrie [Sun, 4 Sep 2011 18:40:21 +0000 (20:40 +0200)]
formats: The format code should be in libpulse, not libpulsecommon

Without this change any applications calling e.g. pa_format_info_new()
and friends will be explicitly linked against libpulsecommon-$MAJORMINOR.so
which is something we specifically avoid as it may contain ABI/API unstable
functions.

Also ensure we export pa_format_info_from_string() for external use.

fdo#40616

13 years agostream-restore: Add proper data validity checks to the legacy database entry read.
Colin Guthrie [Sun, 4 Sep 2011 10:54:41 +0000 (12:54 +0200)]
stream-restore: Add proper data validity checks to the legacy database entry read.

13 years agostream-restore: Add in some variable sets that were missing from 9ffa93.
Colin Guthrie [Sun, 4 Sep 2011 10:28:07 +0000 (12:28 +0200)]
stream-restore: Add in some variable sets that were missing from 9ffa93.

This broke the conversion of the stream-restore database from
previous versions.

13 years agodevice-restore: Simplify the migration of data to per-port keys.
Colin Guthrie [Fri, 2 Sep 2011 11:51:26 +0000 (12:51 +0100)]
device-restore: Simplify the migration of data to per-port keys.

Rather than write all the keys out for each port, simply write a 'null'
port entry and modify the read code to 'fallback' to this when it cannot
find a key. This is needed as the code used when writing the key may not
actually have the sink ports available at the time it uses them,
and thus can cause a segv. This approach adds some degree of overhead
but it's relatively minimal and it can be mitigated by compiling
without support for legacy database formats if so desired.

Thanks to David Henningsson for pointing out the problem.

13 years agoMake gcc --std=c99 happy
Maarten Bosmans [Fri, 2 Sep 2011 12:11:52 +0000 (14:11 +0200)]
Make gcc --std=c99 happy

We're now more or less C99 compliant

13 years agoSquash the last gcc warnings
Maarten Bosmans [Fri, 2 Sep 2011 12:11:51 +0000 (14:11 +0200)]
Squash the last gcc warnings

13 years agomodule-equalizer-sink: Use correct limit in loop
Maarten Bosmans [Fri, 2 Sep 2011 12:11:50 +0000 (14:11 +0200)]
module-equalizer-sink: Use correct limit in loop

13 years agobluetooth/sbc: Use __asm__ keyword
Maarten Bosmans [Fri, 2 Sep 2011 12:11:49 +0000 (14:11 +0200)]
bluetooth/sbc: Use __asm__ keyword

13 years agomodule-equalizer-sink: Use = in initialising variables
Maarten Bosmans [Fri, 2 Sep 2011 08:33:59 +0000 (10:33 +0200)]
module-equalizer-sink: Use = in initialising variables

This makes the compiler in C99 mode happy

13 years agomodule-equalizer-sink: Use %z for printf of size_t variables
Maarten Bosmans [Fri, 2 Sep 2011 08:33:58 +0000 (10:33 +0200)]
module-equalizer-sink: Use %z for printf of size_t variables

13 years agoalsa-mixer: Set "Front" control to 0 dB on headphone path
David Henningsson [Mon, 4 Jul 2011 08:12:17 +0000 (10:12 +0200)]
alsa-mixer: Set "Front" control to 0 dB on headphone path

I've seen more than one system where the volume control named
"Front" is a part of audio path for headphones. This is somewhat
of a compromise: While we don't merge it into the path, as that
would be regressing machines where "Front" isn't a part of the
audio path, it would still enable sound on these machines.

BugLink: http://bugs.launchpad.net/bugs/804178
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agomodule-switch-on-connect: Don't switch unlinked sink input and source outputs
David Henningsson [Wed, 31 Aug 2011 12:37:05 +0000 (14:37 +0200)]
module-switch-on-connect: Don't switch unlinked sink input and source outputs

Unlinked streams can not be moved, so don't try.

BugLink: http://bugs.launchpad.net/bugs/837386
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agovala: Add has_type_id=false to all enums, structs and classes
Arun Raghavan [Wed, 31 Aug 2011 05:51:57 +0000 (11:21 +0530)]
vala: Add has_type_id=false to all enums, structs and classes

Since none of our types is derived from GType, explicitly signal this.
The default value for this attribute changed at some point, breaking our
bindings.

13 years agobuild-sys: bump soname
Colin Guthrie [Mon, 29 Aug 2011 17:07:02 +0000 (18:07 +0100)]
build-sys: bump soname

13 years agoequalizer: Comment out unused function
Arun Raghavan [Mon, 29 Aug 2011 16:44:04 +0000 (22:14 +0530)]
equalizer: Comment out unused function

Gets rid of a compiler warning

13 years agoecho-cancel: Get rid of annoying compiler warnings
Arun Raghavan [Mon, 29 Aug 2011 16:43:44 +0000 (22:13 +0530)]
echo-cancel: Get rid of annoying compiler warnings

13 years agoecho-cancel: Don't allow streams to attach while unloading
Arun Raghavan [Mon, 29 Aug 2011 12:17:39 +0000 (17:47 +0530)]
echo-cancel: Don't allow streams to attach while unloading

When unloading, some module may end up trin to move a sink-input or
source-output back onto our virtual sink/source, causing an infinite
loop of us moving the stream away and having it moved back.

We prevent this from happening by preventing any stream from being
attached during unload.

13 years agoecho-cancel: Make save_aec modarg a bool instead of an int
Arun Raghavan [Mon, 29 Aug 2011 12:17:27 +0000 (17:47 +0530)]
echo-cancel: Make save_aec modarg a bool instead of an int

13 years agopacat: make pacat respond to cork/uncork events
Lu Guanqun [Tue, 23 Aug 2011 06:18:22 +0000 (14:18 +0800)]
pacat: make pacat respond to cork/uncork events

Pacat remembers the number of cork requests, and then cork/uncork the stream
accordingly.

With this change, it makes below test script work correctly:

pacat -p --property=media.role="music" <long-sound> &
sleep 2
pacat -p --property=media.role="phone" <short-sound>
wait

Initial idea by Lu Guanqun, but modified by Colin Guthrie (so blame
me if it's broken)

13 years agoSet better priorities on input paths
David Henningsson [Fri, 26 Aug 2011 12:04:18 +0000 (14:04 +0200)]
Set better priorities on input paths

The priorities should be set in the approximately likeliness that someone
would like to use them. This is hard to guess of course, but this is
a better guess compared to the existing configuration.

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agopassthrough: Fix setting volume to unamplified again
Arun Raghavan [Fri, 26 Aug 2011 04:21:07 +0000 (09:51 +0530)]
passthrough: Fix setting volume to unamplified again

Thanks to Tanu for pointing out the (possibly pathological?) case that
base_volume might be >0dB.

13 years agodevice-restore: Restore volumes on port change.
Colin Guthrie [Wed, 24 Aug 2011 21:50:28 +0000 (22:50 +0100)]
device-restore: Restore volumes on port change.

This will allow for volumes to be saved separately for e.g. Headphones vs. Speakers.

At present it is possible that no volume will be saved for the device prior to the port
switch. In this case the volume will not change from the value set under the other port.
In an ideal world we would save the volume before switching port, but that would require
a new hook.

13 years agodevice-restore: Split device restore database into two parts.
Colin Guthrie [Wed, 24 Aug 2011 21:28:13 +0000 (22:28 +0100)]
device-restore: Split device restore database into two parts.

The device restore saves both the current port for a device, and the volume/mute/formats.
The latter three are really related to the device+port, not the device on it's own.

Thus we store just the port for a device in one key and the volume/mute/formats
under a different key that includes both the device and the port.

13 years agopassthrough: Fix what volume we set sinks/sources to
Arun Raghavan [Thu, 25 Aug 2011 15:47:51 +0000 (21:17 +0530)]
passthrough: Fix what volume we set sinks/sources to

Devices' 0 dB is base_volume not PA_VOLUME_NORM.

13 years agosndfile-util: Check return value of sf_command for errors
Maarten Bosmans [Wed, 24 Aug 2011 18:08:08 +0000 (20:08 +0200)]
sndfile-util: Check return value of sf_command for errors

It's better to show a message what the error is,
instead of just asserting that no errors happen in an external library call.

13 years agoRemove extra ; s where they are not allowed in strict C99
Maarten Bosmans [Wed, 24 Aug 2011 18:08:07 +0000 (20:08 +0200)]
Remove extra ; s where they are not allowed in strict C99

13 years agoecho-cancel: Use stream index in debug message
Maarten Bosmans [Wed, 24 Aug 2011 18:08:06 +0000 (20:08 +0200)]
echo-cancel: Use stream index in debug message

instead of the less useful stream pointer.

13 years agogitignore: Add Orc autogenerated files
Maarten Bosmans [Wed, 24 Aug 2011 18:08:05 +0000 (20:08 +0200)]
gitignore: Add Orc autogenerated files

13 years agoecho-cancel: Move speex preprocessing out of the main module
Arun Raghavan [Thu, 25 Aug 2011 12:17:05 +0000 (17:47 +0530)]
echo-cancel: Move speex preprocessing out of the main module

I initially included put the Speex preprocessing assuming that we'd want
to use the digital gain control and noise suppression from Speex for all
echo cancelling implementations. In practice, we're probably going to
get entire implementations all processing in one package (WebRTC, custom
modules from various vendors, etc.).

This moves out this preprocessing and related knobs into the speex
implementation, which serves to clean out all implementation-specific
details from the module-echo-cancel core.

13 years agoecho-cancel: Use pa_streq instead of strcmp
Arun Raghavan [Wed, 24 Aug 2011 05:05:09 +0000 (10:35 +0530)]
echo-cancel: Use pa_streq instead of strcmp

13 years agoecho-cancel: Add multiple include protection for header
Arun Raghavan [Wed, 24 Aug 2011 05:04:19 +0000 (10:34 +0530)]
echo-cancel: Add multiple include protection for header

13 years agoMore spelling fixes
Maarten Bosmans [Wed, 24 Aug 2011 16:24:46 +0000 (18:24 +0200)]
More spelling fixes

13 years agoSpelling fixes in public headers
Maarten Bosmans [Wed, 24 Aug 2011 16:24:45 +0000 (18:24 +0200)]
Spelling fixes in public headers

13 years agobuild-sys: Fix some LDFLAGS vs. LDADD usage
Arun Raghavan [Wed, 24 Aug 2011 12:26:34 +0000 (17:56 +0530)]
build-sys: Fix some LDFLAGS vs. LDADD usage

Thanks to Stefan Kost for pointing these out.

13 years agoFix spelling sucess -> success
David Henningsson [Wed, 24 Aug 2011 12:01:27 +0000 (14:01 +0200)]
Fix spelling sucess -> success

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agoswitch-on-connect: Don't switch to a monitor source
David Henningsson [Wed, 24 Aug 2011 12:01:00 +0000 (14:01 +0200)]
switch-on-connect: Don't switch to a monitor source

Buglink: http://bugs.launchpad.net/bugs/831675
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agodevice-restore: Change the API to include type information (sink vs. source)
Colin Guthrie [Thu, 18 Aug 2011 15:37:54 +0000 (16:37 +0100)]
device-restore: Change the API to include type information (sink vs. source)

This changes the API, but as we have not released yet, this is OK.

13 years agodbus: Use pa_device_type_t rather than an internal equivalent
Colin Guthrie [Thu, 18 Aug 2011 14:40:15 +0000 (15:40 +0100)]
dbus: Use pa_device_type_t rather than an internal equivalent

13 years agodef: Add a new enum to allow differntiation between sinks and sources.
Colin Guthrie [Thu, 18 Aug 2011 14:36:04 +0000 (15:36 +0100)]
def: Add a new enum to allow differntiation between sinks and sources.

This is for bits of code or API where the concepts and constructs are
extensibly the same for sinks and sources. To avoid duplication we can
simply define our structure/API and then use this enum to differentiate
the two.

13 years agosink: Add PA_SINK_SET_FORMATS macro
Arun Raghavan [Thu, 18 Aug 2011 09:25:52 +0000 (14:55 +0530)]
sink: Add PA_SINK_SET_FORMATS macro

Required for testing the existence of the flag

13 years agosample-util: Fix off-by-one in error check
Arun Raghavan [Thu, 18 Aug 2011 08:49:00 +0000 (14:19 +0530)]
sample-util: Fix off-by-one in error check

Patch by jk.illwinter on http://www.pulseaudio.org/ticket/951

13 years agoRevert "device-restore: Make bools not be bit fields"
Arun Raghavan [Thu, 18 Aug 2011 08:04:33 +0000 (13:34 +0530)]
Revert "device-restore: Make bools not be bit fields"

This reverts commit ba163b8b23bcd72c35b028eb0b43c31a6ae1724d.

After discussion with Colin, decided that we don't want to do this
piecemeal -- need some discussion on the mailing-list and then we should
do this across the tree in one fell swoop post 1.0.

Conflicts:

src/modules/module-device-restore.c

13 years agosource-output: Ensure no volumes are applied for passthrough streams
Arun Raghavan [Thu, 18 Aug 2011 07:02:14 +0000 (12:32 +0530)]
source-output: Ensure no volumes are applied for passthrough streams

This forces passthrough source-outputs and their corresponding sources
to 0dB gain so that the data is sent unaltered to the receiver.

13 years agosink-input: Ensure no volumes are applied for passthrough streams
Arun Raghavan [Thu, 18 Aug 2011 06:21:12 +0000 (11:51 +0530)]
sink-input: Ensure no volumes are applied for passthrough streams

This forces passthrough sink-inputs and their corresponding sinks to 0dB
gain so that the data is sent unaltered to the receiver.

13 years agopassthrough: We must not plug in a resampler on stream move
Arun Raghavan [Thu, 18 Aug 2011 07:05:23 +0000 (12:35 +0530)]
passthrough: We must not plug in a resampler on stream move

13 years agodaemon: Fix compiler warning about missing function prototype
Arun Raghavan [Thu, 18 Aug 2011 07:02:48 +0000 (12:32 +0530)]
daemon: Fix compiler warning about missing function prototype

13 years agoalsa: Fix bad function name
Arun Raghavan [Thu, 18 Aug 2011 04:53:55 +0000 (10:23 +0530)]
alsa: Fix bad function name

I've no idea how that slipped by.

13 years agoformats: Fix bad passsthrough check
Arun Raghavan [Thu, 18 Aug 2011 04:40:22 +0000 (10:10 +0530)]
formats: Fix bad passsthrough check

Fixes breakage from commit adbdb6259becab57fff893671e58fff5786420db

13 years agoalsa: Open iec958 device with NONAUDIO bit set in passthrough mode
Arun Raghavan [Wed, 17 Aug 2011 17:08:43 +0000 (22:38 +0530)]
alsa: Open iec958 device with NONAUDIO bit set in passthrough mode

This is required for E-AC3 streams, as well as to let receivers we're
sending non-PCM data (which avoids playing noise if the data is
incorrect for some reason).

13 years agoformats: Use correct API to check for passthrough streams
Arun Raghavan [Wed, 17 Aug 2011 09:28:23 +0000 (14:58 +0530)]
formats: Use correct API to check for passthrough streams

13 years agoalsa: Don't always suspend/unsuspend on sink-input removal
Arun Raghavan [Wed, 17 Aug 2011 08:03:50 +0000 (13:33 +0530)]
alsa: Don't always suspend/unsuspend on sink-input removal

If the sink was already suspended, this avoids and assert when calling
suspend() again, and prevents an unnecessary unsuspend.

13 years agosource: Remove the PA_SOURCE_PASSTHROUGH flag
Arun Raghavan [Wed, 17 Aug 2011 03:30:47 +0000 (09:00 +0530)]
source: Remove the PA_SOURCE_PASSTHROUGH flag

This isn't actually used anywhere, and PA_SINK_PASSTHROUGH was also
removed.

13 years agointrospect: fix typo in default sink/source docs
Jonny Lamb [Tue, 16 Aug 2011 15:02:51 +0000 (15:02 +0000)]
introspect: fix typo in default sink/source docs

Signed-off-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
13 years agoRemove offensive part of error message
David Henningsson [Mon, 15 Aug 2011 12:05:50 +0000 (14:05 +0200)]
Remove offensive part of error message

I don't know the exact cause for someone to submit a bug report for
this error message: if someone is truly offended by it (if so it is
a reminder that some people are more sensitive than others, and I do
want those people to feel welcome as well), or if it's a system's
check (if this goes through, it shows the system works, and the person
might put more work into his/her next patch), or if it's just a bug
(after all, it's not that weird to run two instances of PulseAudio?).
Either one could be reason enough to apply IMO.

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agoJACK: Load module-jackdbus-detect in default.pa
David Henningsson [Mon, 15 Aug 2011 12:05:49 +0000 (14:05 +0200)]
JACK: Load module-jackdbus-detect in default.pa

Provides automatic integration with the dbus enabled version of JACK 2.

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agoalsa-mixer: Add "Line HP Swap" element
David Henningsson [Mon, 15 Aug 2011 12:05:48 +0000 (14:05 +0200)]
alsa-mixer: Add "Line HP Swap" element

The "Line HP Swap" element needs to be set correctly for some Dove
boards to work correctly. Thanks to Daniel T Chen for the patch.

BugLink: http://bugs.launchpad.net/bugs/451635
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agoalsa-mixer: Mute IEC958 optical raw for several Audigy models
David Henningsson [Mon, 15 Aug 2011 12:05:47 +0000 (14:05 +0200)]
alsa-mixer: Mute IEC958 optical raw for several Audigy models

This switch needs to be off for analog output to work on several
Audigy cards.

Thanks to Daniel T Chen for the patch.

BugLink: http://bugs.launchpad.net/bugs/408370
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
13 years agobuild-sys: bump soname
Colin Guthrie [Tue, 16 Aug 2011 10:04:11 +0000 (11:04 +0100)]
build-sys: bump soname

13 years agocore: Unload the modules and cached samples before unref'ing the core.
Colin Guthrie [Tue, 16 Aug 2011 09:54:24 +0000 (10:54 +0100)]
core: Unload the modules and cached samples before unref'ing the core.

As various modules can subscribe to unlink callbacks unloading some modules
may trigger hooks in other modules.

The callbacks associated with these hooks could in turn need to use the core
in some capacity (e.g. perhaps they are module loading modules
(e.g. *-discover, filter-apply or gconf etc. and need to use the core to
unload modules they've loaded).

This change simply ensures that all modules and cached samples are unloaded
before freeing the core.

13 years agopactl: Add set-source-output-mute command
Maarten Bosmans [Mon, 15 Aug 2011 20:05:41 +0000 (22:05 +0200)]
pactl: Add set-source-output-mute command

13 years agodefault.pa: Update rtp null sink line
Maarten Bosmans [Mon, 15 Aug 2011 19:38:45 +0000 (21:38 +0200)]
default.pa: Update rtp null sink line

The description argument has been removed in favour of the more general sink_properties.

13 years agoInitialise write_volume
Maarten Bosmans [Mon, 15 Aug 2011 18:19:40 +0000 (20:19 +0200)]
Initialise write_volume

The callback should also be reset in reset_calbacks().
The extra check in _volume_change_apply() is needed because when the sink is unlinked the callbacks are reset,
but there still may be pending volume changes.

13 years agoformat: Remove stupid copy-paste-o
Arun Raghavan [Tue, 16 Aug 2011 09:51:45 +0000 (15:21 +0530)]
format: Remove stupid copy-paste-o

Thanks to Maarten Bosmans for the catch.

13 years agointrospect: fix source output and sink input docs mix-up
Jonny Lamb [Tue, 2 Aug 2011 13:38:10 +0000 (13:38 +0000)]
introspect: fix source output and sink input docs mix-up

Signed-off-by: Jonny Lamb <jonny.lamb@collabora.co.uk>
13 years agoalsa: Set SET_FORMATS flag when appropriate
Arun Raghavan [Tue, 16 Aug 2011 05:41:46 +0000 (11:11 +0530)]
alsa: Set SET_FORMATS flag when appropriate

13 years agosink: Add a SET_FORMATS flag
Arun Raghavan [Tue, 16 Aug 2011 05:41:34 +0000 (11:11 +0530)]
sink: Add a SET_FORMATS flag

This adds a PA_SINK_SET_FORMATS flag to the pa_sink_flags enum,
signalling that a sink allows the set of supported formats to be set
externally. The idea is for clients to be able to know what sinks
support this ability and adapt their UI appropriately.

13 years agobuild-sys: Bump JACK dependency to 0.117.0
Arun Raghavan [Mon, 15 Aug 2011 17:01:55 +0000 (22:31 +0530)]
build-sys: Bump JACK dependency to 0.117.0

This version is required since we recently started using jack_free()

https://bugs.freedesktop.org/show_bug.cgi?id=38858

13 years agoconf: Load module-filter-* by default
Arun Raghavan [Mon, 15 Aug 2011 11:07:55 +0000 (16:37 +0530)]
conf: Load module-filter-* by default

This loads module-filter-heuristics and module-filter-apply by default
so that applications can request filters via properties. Not adding this
to system.pa -- the assumption is that people running system mode would
want more fine-grained control over such options.

13 years agofilter-heuristics: Don't force AEC on all phone streams
Arun Raghavan [Mon, 15 Aug 2011 10:38:40 +0000 (16:08 +0530)]
filter-heuristics: Don't force AEC on all phone streams

Removing the bit that automatically loads module-echo-cancel for phone
streams. Clients need to specifically opt in for this now with
filter.want, until we are reasonably certain this won't break other apps
(Skype for one, possibly others).

13 years agoproplist: Make filter properties line up prettily
Arun Raghavan [Mon, 15 Aug 2011 09:50:20 +0000 (15:20 +0530)]
proplist: Make filter properties line up prettily

13 years agoRemind people not to break module-tunnel
Arun Raghavan [Mon, 15 Aug 2011 09:49:10 +0000 (15:19 +0530)]
Remind people not to break module-tunnel

13 years agomemblock: use built-in function
Lu Guanqun [Mon, 15 Aug 2011 05:24:51 +0000 (13:24 +0800)]
memblock: use built-in function

Signed-off-by: Lu Guanqun <guanqun.lu@intel.com>
13 years agomemblock: fix memory leak when pa_shm_create_rw fails
Lu Guanqun [Mon, 15 Aug 2011 01:51:37 +0000 (09:51 +0800)]
memblock: fix memory leak when pa_shm_create_rw fails

Signed-off-by: Lu Guanqun <guanqun.lu@intel.com>
13 years agoAvoid read from freed memory
Maarten Bosmans [Sat, 13 Aug 2011 11:43:21 +0000 (13:43 +0200)]
Avoid read from freed memory

The order of freeing the hashmaps is important here, because otherwise a string used as key is freed before the hashmap
is freed.

Valgrind reports this as:
 Invalid read of size 1
    at 0x4107042: pa_idxset_string_hash_func (idxset.c:67)
    by 0x4106026: remove_entry (hashmap.c:93)
    by 0x41061BF: pa_hashmap_free (hashmap.c:110)
    by 0x71DD143: pa_dbusiface_core_free (iface-core.c:2105)
    by 0x71F2169: module_dbus_protocol_LTX_pa__done (module-dbus-protocol.c:595)
    by 0x406DC51: pa_module_free (module.c:162)
    by 0x406E01D: pa_module_unload_all (module.c:210)
    by 0x4068842: core_free (core.c:169)
    by 0x406FD5D: pa_object_unref (object.c:64)
    by 0x805224D: pa_core_unref (core.h:184)
    by 0x805560B: main (main.c:1159)
  Address 0x4d099c0 is 0 bytes inside a block of size 100 free'd
    at 0x4025BF0: free (vg_replace_malloc.c:366)
    by 0x40F128C: pa_xfree (xmalloc.c:131)
    by 0x71E4CEB: pa_dbusiface_device_free (iface-device.c:1293)
    by 0x71DCD7E: free_device_cb (iface-core.c:2062)
    by 0x41061D7: pa_hashmap_free (hashmap.c:113)
    by 0x71DD125: pa_dbusiface_core_free (iface-core.c:2104)
    by 0x71F2169: module_dbus_protocol_LTX_pa__done (module-dbus-protocol.c:595)
    by 0x406DC51: pa_module_free (module.c:162)
    by 0x406E01D: pa_module_unload_all (module.c:210)
    by 0x4068842: core_free (core.c:169)
    by 0x406FD5D: pa_object_unref (object.c:64)
    by 0x805224D: pa_core_unref (core.h:184)

13 years agoInitialise variables
Maarten Bosmans [Sat, 13 Aug 2011 11:43:20 +0000 (13:43 +0200)]
Initialise variables

These were detected using clang static analyzer.

13 years agoPlug some memory leaks
Maarten Bosmans [Sat, 13 Aug 2011 11:43:19 +0000 (13:43 +0200)]
Plug some memory leaks

These were detected with valgrind.

13 years agomodule-tunnel: Fix for new protocol versions
Maarten Bosmans [Sat, 13 Aug 2011 11:43:18 +0000 (13:43 +0200)]
module-tunnel: Fix for new protocol versions

The commit 7ebc5033 resulted in segfaults, because format->plist was not allocated.
The solution is not to allocate pa_format_info on the stack, but to properly use pa_format_info_new().
Also a typo regarding pa_tagstruct_putu8 is corrected.

13 years agosvolume: Make log messages more precise.
Tanu Kaskinen [Sat, 13 Aug 2011 11:48:34 +0000 (14:48 +0300)]
svolume: Make log messages more precise.

I was looking at a log, and noticed the following lines:

I  [pulseaudio] svolume_mmx.c: Initialising MMX optimized functions.
I  [pulseaudio] remap_mmx.c: Initialising MMX optimized remappers.
I  [pulseaudio] svolume_sse.c: Initialising SSE2 optimized functions.
I  [pulseaudio] remap_sse.c: Initialising SSE2 optimized remappers.
I  [pulseaudio] sconv_sse.c: Initialising SSE2 optimized conversions.

It seemed odd that some messages were somewhat precise in
what functionality was initialized, while the svolume
messages told me that they had initialized just "functions".
So I made the svolume log messages more precise to match the
sconv and remap messages.

13 years agodaemon-conf: Don't make log files executable.
Tanu Kaskinen [Sat, 13 Aug 2011 11:35:31 +0000 (14:35 +0300)]
daemon-conf: Don't make log files executable.

13 years agosink-input: Avoid fake rewind in corked state
Wang Xingchao [Sat, 13 Aug 2011 11:55:06 +0000 (13:55 +0200)]
sink-input: Avoid fake rewind in corked state

sink_input_request_rewind() does nothing if we are in the corked state.
Thus do not set the new state untill we have issued the rewind request.

13 years agodevice-restore: Log invalid sink index while setting formats
Arun Raghavan [Mon, 15 Aug 2011 06:37:02 +0000 (12:07 +0530)]
device-restore: Log invalid sink index while setting formats

This makes it easier to catch errors when using 'pactl set-sink-formats'

13 years agosink: Fix lazy commenting
Arun Raghavan [Mon, 15 Aug 2011 05:53:59 +0000 (11:23 +0530)]
sink: Fix lazy commenting

13 years agopactl: Add a set-sink-formats command
Arun Raghavan [Mon, 15 Aug 2011 04:17:09 +0000 (09:47 +0530)]
pactl: Add a set-sink-formats command

This uses the module-device-restore protocol extension to save formats
on sinks that support it.

13 years agoformat: Add string to pa_format_info conversion function
Arun Raghavan [Fri, 12 Aug 2011 16:25:48 +0000 (21:55 +0530)]
format: Add string to pa_format_info conversion function

This will help accept string formats from the command like (so we can
set formats using pactl).

13 years agoformat: Make pa_format_info_snprint() more parseable
Arun Raghavan [Fri, 12 Aug 2011 15:34:38 +0000 (21:04 +0530)]
format: Make pa_format_info_snprint() more parseable

Removes the comma as the proplist separator since that makes
pa_proplist_from_string() break and prints only the encoding if there
are no properties (instead of "<encoding>, (no properties)").

13 years agodevice-restore: Set sink format when possible
Arun Raghavan [Fri, 12 Aug 2011 14:57:34 +0000 (20:27 +0530)]
device-restore: Set sink format when possible

This implements the actual setting of sink formats when a new sink is
added or when the set of available formats changes.

13 years agodevice-restore: Make bools not be bit fields
Arun Raghavan [Fri, 12 Aug 2011 14:03:41 +0000 (19:33 +0530)]
device-restore: Make bools not be bit fields

This makes the pa_bool_t members of userdata not be a single bit field
since pa_bool_t can be an int, potentially causing signedness issues in
comparisons.

13 years agoalsa: Implement get/set_formats()
Arun Raghavan [Fri, 12 Aug 2011 11:06:00 +0000 (16:36 +0530)]
alsa: Implement get/set_formats()

This implements the sink get_formats() and set_formats() API in
alsa-sink. Modules can use this to allow users to specify what formats
their receivers support.

13 years agosink: Add a set_formats() API
Arun Raghavan [Fri, 12 Aug 2011 11:04:28 +0000 (16:34 +0530)]
sink: Add a set_formats() API

This adds API to let external sources specify what formats a sink
supports. Sinks must opt-in to allow this, and can perform some
validation if required.

13 years agoPlug some memory leaks and an invalid read
Maarten Bosmans [Thu, 11 Aug 2011 22:17:39 +0000 (00:17 +0200)]
Plug some memory leaks and an invalid read

Note in protocol-dbus.c specifically, method_signatures needs to be freed
before method_handlers, because otherwise h->method_name is freed while it is
still in use as a key in the method_signatures hashmap.

13 years agosink-input: Drop redundant assert (PA_SINK_INPUT_IS_LINKED() checked already)
Colin Guthrie [Thu, 11 Aug 2011 15:51:52 +0000 (17:51 +0200)]
sink-input: Drop redundant assert (PA_SINK_INPUT_IS_LINKED() checked already)

13 years agolog: Add missing pulsecore/thread.h include
Arun Raghavan [Fri, 12 Aug 2011 14:29:19 +0000 (19:59 +0530)]
log: Add missing pulsecore/thread.h include

13 years agoi18n: po file fixes
Lu Guanqun [Thu, 11 Aug 2011 16:07:26 +0000 (00:07 +0800)]
i18n: po file fixes

Signed-off-by: Lu Guanqun <guanqun.lu@intel.com>
13 years agoalsa: resets POLLOUT event
Wang Xingchao [Fri, 5 Aug 2011 14:33:17 +0000 (10:33 -0400)]
alsa: resets POLLOUT event

revents marked as POLLOUT|POLLERR|POLLWRNORM in "underrun" case that will
trigger unexpected log "ALSA woke us up to write new data to the device, but
there was acturally nothing to write...".

This patch avoids this log message.

13 years agoalsa: Update process_usec before going to sleep
Wang Xingchao [Tue, 9 Aug 2011 14:14:40 +0000 (10:14 -0400)]
alsa: Update process_usec before going to sleep

During check_left_to_play/record(), the watermark may increase/decrease. So before
sleeping, update the actural sleep time based on latest watermark.

13 years agoMove i18n.[ch] to src/pulsecore
Maarten Bosmans [Wed, 10 Aug 2011 08:30:15 +0000 (10:30 +0200)]
Move i18n.[ch] to src/pulsecore

The header is used in files troughout the tree and is not included in the public api,
so it belongs in pulsecore, not in pulse.

13 years agoAdd some missing format.h includes
Maarten Bosmans [Wed, 10 Aug 2011 08:30:14 +0000 (10:30 +0200)]
Add some missing format.h includes

13 years agoInclude config.h consistently in source files and not in headers
Maarten Bosmans [Wed, 10 Aug 2011 08:30:13 +0000 (10:30 +0200)]
Include config.h consistently in source files and not in headers

13 years agosample-util: use built-in function
Lu Guanqun [Wed, 10 Aug 2011 15:12:12 +0000 (23:12 +0800)]
sample-util: use built-in function

use built-in function pa_frame_aligned().

Signed-off-by: Lu Guanqun <guanqun.lu@intel.com>
13 years agoext-device-restore: Include format.h
Colin Guthrie [Wed, 10 Aug 2011 11:16:28 +0000 (13:16 +0200)]
ext-device-restore: Include format.h