platform/upstream/pulseaudio.git
11 years agovirtual-sink: Removed the option to specify sample format.
Ștefan Săftescu [Fri, 6 Apr 2012 21:57:43 +0000 (00:57 +0300)]
virtual-sink: Removed the option to specify sample format.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=46529
11 years agocpulimit: Explicitly ignore pa_read() return value.
Tanu Kaskinen [Tue, 3 Apr 2012 07:18:55 +0000 (10:18 +0300)]
cpulimit: Explicitly ignore pa_read() return value.

It doesn't matter if the function fails (I'm not sure if
it's even possible), because the read data isn't used for
anything and the daemon will terminate in any case. The
void cast should get rid of a Coverity warning.

Removing the whole pa_read() call should be ok too, but I
guess it's nice to clean up the pipe before terminating...

11 years agooss: Check pa_read() return value.
Tanu Kaskinen [Mon, 2 Apr 2012 13:13:54 +0000 (16:13 +0300)]
oss: Check pa_read() return value.

11 years agomainloop-signal: Explicitly ignore pa_write() return value.
Tanu Kaskinen [Fri, 30 Mar 2012 08:04:33 +0000 (11:04 +0300)]
mainloop-signal: Explicitly ignore pa_write() return value.

Coverity warned about an ignored return value. I'm not sure
if there's something that should be done if writing fails;
at least I couldn't think of anything. Would logging an
error be acceptable here?

11 years agosmoother-test: Fix array overflow.
Tanu Kaskinen [Fri, 30 Mar 2012 07:31:48 +0000 (10:31 +0300)]
smoother-test: Fix array overflow.

11 years agoFix pa_parse_boolean() return value checking.
Tanu Kaskinen [Thu, 29 Mar 2012 12:24:02 +0000 (15:24 +0300)]
Fix pa_parse_boolean() return value checking.

pa_parse_boolean() return value shouldn't be stored in
pa_bool_t, because 1 and -1 need to be distinguished.

11 years agopasuspender: Check pa_context_connect() return value.
Tanu Kaskinen [Thu, 29 Mar 2012 09:18:32 +0000 (12:18 +0300)]
pasuspender: Check pa_context_connect() return value.

11 years agobluetooth: Don't free read_smoother in pa__done().
Tanu Kaskinen [Thu, 22 Mar 2012 10:23:13 +0000 (12:23 +0200)]
bluetooth: Don't free read_smoother in pa__done().

pa__done() calls stop_thread(), and stop_thread() already
frees the smoother. The duplicate freeing is not strictly
a bug, but static analyzers (in this case Coverity) may
complain about double-freeing, because when pa__done()
"frees" the smoother (which doesn't actually ever happen),
the pointer is not nulled. pa__done() then calls
bt_transport_release(), which will also free the smoother
if it's not NULL.

The analyzer complaint could be silenced also by nulling
the pointer in pa__done(), but since this is clearly
redundant code, I chose to remove it.

11 years agoqpaeq: Try to load equalizer module before failing, better error messages
Matěj Laitl [Mon, 19 Mar 2012 21:44:30 +0000 (22:44 +0100)]
qpaeq: Try to load equalizer module before failing, better error messages

This fixes bug 38728 [1]. When equalizer features are unavailable in running
pulseaudio daemon, try to load relevant module. If this fails, following error
is printed on stderr instead of a confusing traceback:

It seems that running pulseaudio does not support equalizer features and
loading module-equalizer-sink module failed. Exiting...

[1] https://bugs.freedesktop.org/show_bug.cgi?id=38728

Signed-off-by: Matěj Laitl <matej@laitl.cz>
11 years agobluetooth: Make pa_bt_audio_state_from_string() private.
Tanu Kaskinen [Sun, 16 Dec 2012 05:03:00 +0000 (07:03 +0200)]
bluetooth: Make pa_bt_audio_state_from_string() private.

11 years agobluetooth: Improve transport state change log message.
Tanu Kaskinen [Sun, 16 Dec 2012 04:57:22 +0000 (06:57 +0200)]
bluetooth: Improve transport state change log message.

11 years agobluetooth: Replace acquire param 'start' with 'optional'
Mikel Astiz [Fri, 14 Dec 2012 14:14:40 +0000 (15:14 +0100)]
bluetooth: Replace acquire param 'start' with 'optional'

Make the internal function bt_transport_acquire() consistent with the
API in bluetooth-util by replacing the old 'start' parameter with
exactly the opposite: 'optional'.

Therefore, all calls to the function need to negate the second
parameter.

Note also that the name is more accurate now that setup_stream() is not
called inside bt_transport_acquire().

11 years agobluetooth: Call setup_stream() manually
Mikel Astiz [Fri, 14 Dec 2012 14:14:39 +0000 (15:14 +0100)]
bluetooth: Call setup_stream() manually

Do not call setup_stream() automatically inside bt_transport_acquire().
Instead, the caller is responsible to use both functions as necessary.

As a first trivial step, setup_stream() is now called manually after
all calls to bt_transport_acquire(u, TRUE), with the exception of
setup_transport() where the thread is still about to start and thus
setup_stream() will be called later on from thread_func().

11 years agobluetooth: Remove D-Bus dependency in module-bluetooth-device
Mikel Astiz [Fri, 14 Dec 2012 14:14:38 +0000 (15:14 +0100)]
bluetooth: Remove D-Bus dependency in module-bluetooth-device

All D-Bus infrastructure is now unused after bluetooth-util has covered
the pieces that were pending. Therefore, all D-Bus related code in
module-bluetooth-device can be safely removed.

11 years agobluetooth: Use transport state instead of profile states
Mikel Astiz [Fri, 14 Dec 2012 14:14:37 +0000 (15:14 +0100)]
bluetooth: Use transport state instead of profile states

The transport state also reflects the state of the audio interface. The
state redundancy can thus be minimized by always using the first one,
and avoiding the use of profile-specific states with the exception of
finding out the initial state of a transport.

11 years agobluetooth: Refactor dependency to org.bluez.Audio
Mikel Astiz [Fri, 14 Dec 2012 14:14:36 +0000 (15:14 +0100)]
bluetooth: Refactor dependency to org.bluez.Audio

The state of this interface is needed for one single reason: we need to
wait until all profiles have been connected (or more precisely, until
are connection attempts are finished). This can be made more explicit in
the code by just checking the CONNECTING state (and not loading
module-bluetooth-device during that state), but otherwise treating all
transport types equally.

Ideally, audio_state should be completely removed but it's left there to
avoid an issue with module-card-restore, as documented in the source
code's comments.

11 years agobluetooth: Trivially remove bt_transport_is_acquired()
Mikel Astiz [Fri, 14 Dec 2012 14:14:35 +0000 (15:14 +0100)]
bluetooth: Trivially remove bt_transport_is_acquired()

The function body is now trivial and can thus be inlined.

11 years agobluetooth: Abstract transport access types inside bluetooth-util
Mikel Astiz [Fri, 14 Dec 2012 14:14:34 +0000 (15:14 +0100)]
bluetooth: Abstract transport access types inside bluetooth-util

Transports can be acquired with different access rights, but in practice
"rw" was always used inside module-bluetooth-device. In addition, this
feature is removed in BlueZ 5.0 and therefore it is convenient to
abstract all this inside bluetooth-util.

11 years agobluetooth: Use transport state to update port availability
Mikel Astiz [Fri, 14 Dec 2012 14:14:33 +0000 (15:14 +0100)]
bluetooth: Use transport state to update port availability

Use transport state to calculate the corresponding port availability,
and while doing so use bluetooth-util to receive profile state updates
instead of directly parsing D-Bus PropertyChanged signals.

11 years agobluetooth: Refactor parsing of profile state changes
Mikel Astiz [Fri, 14 Dec 2012 14:14:32 +0000 (15:14 +0100)]
bluetooth: Refactor parsing of profile state changes

Split filter_cb() to separate the parsing of the D-Bus message from
the internal logic handling the new state.

11 years agobluetooth: Move profile_to_string() to bluetooth-util
Mikel Astiz [Fri, 14 Dec 2012 14:14:31 +0000 (15:14 +0100)]
bluetooth: Move profile_to_string() to bluetooth-util

Move the function to the utility library where the enum is defined. At
same time avoid using the default clause in order to make sure the
compiler will complain if the enum type gets extended.

11 years agobluetooth: Use bluetooth-util for mic/speaker gain control
Mikel Astiz [Fri, 14 Dec 2012 14:14:30 +0000 (15:14 +0100)]
bluetooth: Use bluetooth-util for mic/speaker gain control

Use the new abstraction in bluetooth-util to both receive volume updates
and request them.

11 years agobluetooth: Abstract speaker gain in transport
Mikel Astiz [Fri, 14 Dec 2012 14:14:29 +0000 (15:14 +0100)]
bluetooth: Abstract speaker gain in transport

Similarly to the microphone gain, the speaker gain can be abstracted
inside the transport object, even though the actual D-Bus interface in
BlueZ differs.

11 years agobluetooth: Abstract microphone gain in transport
Mikel Astiz [Fri, 14 Dec 2012 14:14:28 +0000 (15:14 +0100)]
bluetooth: Abstract microphone gain in transport

The microphone gain represents the volume of the incoming audio stream
from the headset. This can be nicely abstracted inside the transport
object in bluetooth-util, so the modules don't have to take care about
the D-Bus details.

11 years agoecho-cancel: Fix null implementation to setup one channel
Stefan Huber [Tue, 4 Dec 2012 13:55:01 +0000 (14:55 +0100)]
echo-cancel: Fix null implementation to setup one channel

memcpy() of the null implementation's run() copied data for only one
channel. Set the number of channels to 1 in init() in order to guarantee
this.

Signed-off-by: Stefan Huber <s.huber@bct-electronic.com>
Acked-by: Peter Meerwald <p.meerwald@bct-electronic.com>
11 years agoecho-cancel: Fix apply_diff_time() to use correct sample spec
Stefan Huber [Tue, 4 Dec 2012 13:54:58 +0000 (14:54 +0100)]
echo-cancel: Fix apply_diff_time() to use correct sample spec

apply_diff_time() fails when dropping bytes from the playback stream
and the sample spec of sink and source differ as source's sample spec is
used. Fix this by using sink's sample spec.

Signed-off-by: Stefan Huber <s.huber@bct-electronic.com>
Acked-by: Peter Meerwald <p.meerwald@bct-electronic.com>
11 years agorole-ducking: Apply a ducking effect based on streams roles
Flavio Ceolin [Thu, 13 Dec 2012 19:00:55 +0000 (17:00 -0200)]
role-ducking: Apply a ducking effect based on streams roles

This module works pretty similar to the module-role-cork.
It should be used as an alternative to that module. Basically
it decreases the volume of the streams specified in ducking_roles
in the presence of at least one stream specified in trigger_roles.
Also, it's possible to choice the volume that will be used in the
ducking streams and if it should operates in all devices or not.

For basic reference: http://en.wikipedia.org/wiki/Ducking

11 years agoman: Document the possiblity of unloading modules by name in pulse-cli-syntax.
Tanu Kaskinen [Fri, 14 Dec 2012 07:11:01 +0000 (09:11 +0200)]
man: Document the possiblity of unloading modules by name in pulse-cli-syntax.

11 years agocore-util: Improve get_path() documentation
Tanu Kaskinen [Thu, 13 Dec 2012 06:26:05 +0000 (08:26 +0200)]
core-util: Improve get_path() documentation

11 years agobluetooth: Connect all hooks during module load/unload
Mikel Astiz [Tue, 11 Dec 2012 13:40:24 +0000 (14:40 +0100)]
bluetooth: Connect all hooks during module load/unload

Move the connection of sink/source-related hooks to module
initialization and shutdown, to group all of them together. There is
no need to connect them every time the card profile is changed.

11 years agobluetooth: Remove prefixing from static functions.
Tanu Kaskinen [Tue, 11 Dec 2012 05:18:29 +0000 (07:18 +0200)]
bluetooth: Remove prefixing from static functions.

11 years agobluetooth: Remove deprecated transport hooks
Mikel Astiz [Mon, 10 Dec 2012 07:30:42 +0000 (08:30 +0100)]
bluetooth: Remove deprecated transport hooks

All code has migrated to the new centralized hooks listed in
pa_bluetooth_hook_t so the old transport-specific hooks can be removed.

11 years agobluetooth: Avoid PA_BLUETOOTH_TRANSPORT_HOOK_NREC_CHANGED
Mikel Astiz [Mon, 10 Dec 2012 07:30:41 +0000 (08:30 +0100)]
bluetooth: Avoid PA_BLUETOOTH_TRANSPORT_HOOK_NREC_CHANGED

The hook has been recently moved to pa_bluetooth_hook_t, so make use
of the new version.

11 years agobluetooth: Avoid PA_BLUETOOTH_TRANSPORT_HOOK_REMOVED
Mikel Astiz [Mon, 10 Dec 2012 07:30:40 +0000 (08:30 +0100)]
bluetooth: Avoid PA_BLUETOOTH_TRANSPORT_HOOK_REMOVED

The hook is now deprecated so avoid using it and instead use the
recently introduced PA_BLUETOOTH_HOOK_TRANSPORT_STATE_CHANGED which also
reports the disconnection event.

11 years agobluetooth: Add transport hooks into pa_bluetooth_hook_t
Mikel Astiz [Mon, 10 Dec 2012 07:30:39 +0000 (08:30 +0100)]
bluetooth: Add transport hooks into pa_bluetooth_hook_t

Add the transport-handling hooks to the centralized list of hooks in
pa_bluetooth_hook_t. These are intended to replace the now deprecated
transport-specific hook list in pa_bluetooth_transport_hook_t.

11 years agobluetooth: Add state to transport objects
Mikel Astiz [Mon, 10 Dec 2012 07:30:38 +0000 (08:30 +0100)]
bluetooth: Add state to transport objects

Transport objects have an associated state even though it's not
explicitly exposed in BlueZ's D-Bus API (prior to 5.0). Instead, the
state is implicitly represented in the profile-specific D-Bus interface
(i.e. org.bluez.Headset, org.bluez.AudioSink, etc.) but it can be
convenient that bluetooth-util would abstract this separation.

11 years agobluetooth: Refactor parse_audio_property() to support more properties
Mikel Astiz [Mon, 10 Dec 2012 07:30:37 +0000 (08:30 +0100)]
bluetooth: Refactor parse_audio_property() to support more properties

The old implementation is limited to parsing the profile state, but
the D-Bus API actually exposes many more properties that are currently
not being considered, specially within org.bluez.Headset.

11 years agobluetooth: Use array to store profile states
Mikel Astiz [Thu, 6 Dec 2012 14:55:29 +0000 (15:55 +0100)]
bluetooth: Use array to store profile states

Refactor the code to use an array of states instead of independent
member fields, avoiding duplicated code and improving readability.

11 years agobluetooth: Move device hooks into pa_bluetooth_hook_t
Mikel Astiz [Thu, 6 Dec 2012 14:55:28 +0000 (15:55 +0100)]
bluetooth: Move device hooks into pa_bluetooth_hook_t

Centralize the Bluetooth hooks in one single place, starting with
the device hooks, while removing the duplicated ones (in this case
PA_BLUETOOTH_DEVICE_HOOK_REMOVED).

11 years agobluetooth: Extend discovery to support multiple hooks
Mikel Astiz [Thu, 6 Dec 2012 14:55:27 +0000 (15:55 +0100)]
bluetooth: Extend discovery to support multiple hooks

Add the infrastructure to support several hooks inside
pa_bluetooth_discovery, while using hook names that describe more
accurately their purpose.

11 years agobluetooth: Avoid PA_BLUETOOTH_DEVICE_HOOK_REMOVED
Mikel Astiz [Thu, 6 Dec 2012 14:55:26 +0000 (15:55 +0100)]
bluetooth: Avoid PA_BLUETOOTH_DEVICE_HOOK_REMOVED

The hook PA_BLUETOOTH_HOOK_DEVICE_CONNECTION_CHANGED gets fired also
when a device is being removed, so there is actually no need to have
this duplicated hook.

11 years agobluetooth: Check if BlueZ tries to set the configuration twice for the same transport.
Tanu Kaskinen [Thu, 6 Dec 2012 04:49:22 +0000 (06:49 +0200)]
bluetooth: Check if BlueZ tries to set the configuration twice for the same transport.

11 years agobluetooth: Remove pa_bluetooth_device_get_transport()
Mikel Astiz [Thu, 6 Dec 2012 09:35:20 +0000 (10:35 +0100)]
bluetooth: Remove pa_bluetooth_device_get_transport()

With the use of an array to represent a device's transpors, the function
becomes trivial and thus can be removed.

11 years agobluetooth: Use transport array instead of hashmap for devices
Mikel Astiz [Thu, 6 Dec 2012 09:35:19 +0000 (10:35 +0100)]
bluetooth: Use transport array instead of hashmap for devices

Devices will have zero or one transports per profile, and besides the
typical lookup is also profile-based. Therefore, replace the old hashmap
(which used the transport path as key) with a simple array which holds
a transport pointer per profile.

11 years agobluetooth: Add transport hashmap to discovery
Mikel Astiz [Thu, 6 Dec 2012 09:35:18 +0000 (10:35 +0100)]
bluetooth: Add transport hashmap to discovery

Path-based transport lookups are required in a discovery basis, before
the associated device is known. Therefore, it makes more sense to
maintain a hashmap in the discovery structure itself, instead of
iterating all devices.

11 years agobluetooth: Use round() to convert PA<->BT volumes
Mikel Astiz [Thu, 6 Dec 2012 09:35:17 +0000 (10:35 +0100)]
bluetooth: Use round() to convert PA<->BT volumes

The code can be simplified since it's just trying to round the result of
the division. Note that the resulting behavior is slightly different,
specially when the volume is 0. In this case, it will remain at 0,
instead of being set to 1.

11 years agobluetooth: Add device pointer to transport
Mikel Astiz [Wed, 5 Dec 2012 16:23:15 +0000 (17:23 +0100)]
bluetooth: Add device pointer to transport

Transports always have an associated device, so add the pointer as a
member to the structure, and remove the discovery pointer since it
already exists in the device object.

11 years agobluetooth: Remove unused pa_bluetooth_discovery_get_transport()
Mikel Astiz [Wed, 5 Dec 2012 16:23:14 +0000 (17:23 +0100)]
bluetooth: Remove unused pa_bluetooth_discovery_get_transport()

The function is not used and it also exposes D-Bus-related information
in the internal API, which is in general undesired.

11 years agobluetooth: Make pa_bluetooth_transport_parse_property() private
Mikel Astiz [Wed, 5 Dec 2012 16:23:12 +0000 (17:23 +0100)]
bluetooth: Make pa_bluetooth_transport_parse_property() private

The function is not being used outside bluetooth-util and thus can be
made private.

11 years agobluetooth: Fix condition to load module
Mikel Astiz [Wed, 5 Dec 2012 16:23:11 +0000 (17:23 +0100)]
bluetooth: Fix condition to load module

d->hfgw_state is just another profile that should be considered exactly
as the rest inside device_audio_is_ready(), which is being used to
decide if the discovery hook gets triggered.

Therefore, there seems to be no reason to make an exception for this
profile and skip checking if the condition d->audio_state !=
PA_BT_AUDIO_STATE_INVALID holds true.

This change makes no practical difference but delaying the load of the
module also for HFGW until Audio.State is received. The benefit is
that the behavior and the code are more consistent across profiles.

11 years agomodargs: Adding pa_modargs_get_value_volume()
Flavio Ceolin [Thu, 29 Nov 2012 13:04:12 +0000 (11:04 -0200)]
modargs: Adding pa_modargs_get_value_volume()

This function gets a pa_volume_t from a string.

11 years agosink-input: New volume_factor system
Flavio Ceolin [Thu, 29 Nov 2012 13:04:11 +0000 (11:04 -0200)]
sink-input: New volume_factor system

Implement setting of more than one volume factor.  The
real value of the volume_factor will be the multiplication of these
values.

11 years agocard-restore: Log the restored profile name.
Tanu Kaskinen [Mon, 19 Nov 2012 10:59:31 +0000 (12:59 +0200)]
card-restore: Log the restored profile name.

11 years agoLog the reason for every suspend/resume.
Tanu Kaskinen [Fri, 16 Nov 2012 16:24:34 +0000 (18:24 +0200)]
Log the reason for every suspend/resume.

I was looking at a log that showed that a suspend happened (at
a strange time), but the log didn't tell me why the suspend was done.
This patch tries to make sure that that won't happen again.

11 years agoesound: Suspend/resume also sources on STANDBY/RESUME commands.
Tanu Kaskinen [Fri, 16 Nov 2012 16:03:55 +0000 (18:03 +0200)]
esound: Suspend/resume also sources on STANDBY/RESUME commands.

11 years agomainloop: Check pa_write() return value.
Tanu Kaskinen [Mon, 2 Apr 2012 12:01:06 +0000 (15:01 +0300)]
mainloop: Check pa_write() return value.

11 years agomainloop: Remove redundant wakeup_pipe validity checks.
Tanu Kaskinen [Mon, 2 Apr 2012 12:01:04 +0000 (15:01 +0300)]
mainloop: Remove redundant wakeup_pipe validity checks.

The pipe is opened when creating the mainloop and closed
when freeing the mainloop. The pipe fds can never be less
than zero.

11 years agobluetooth: Add support for transport created by external profile
Luiz Augusto von Dentz [Mon, 12 Nov 2012 10:26:02 +0000 (12:26 +0200)]
bluetooth: Add support for transport created by external profile

With BlueZ 5 it is possible to have profile registered by a third party
process which does not share the same bus id as bluetoothd so it is
necessary to store the sender of the transport to be able to talk to it.

Note that this is backward compatible.

11 years agobluetooth: Handle UUIDs announced later
Mikel Astiz [Fri, 26 Oct 2012 06:23:51 +0000 (08:23 +0200)]
bluetooth: Handle UUIDs announced later

In some cases (typically during pairing) UUIDs might be reported by
BlueZ incrementally, that is, as soon as they have been discovered. At
this point module-bluetooth-device might already be loaded, so the late
UUID announcements need to be handled and additional card profiles
might need to be created accordingly.

11 years agodbus: Add signal org.PulseAudio.Core1.Card.NewProfile
Mikel Astiz [Fri, 26 Oct 2012 06:23:50 +0000 (08:23 +0200)]
dbus: Add signal org.PulseAudio.Core1.Card.NewProfile

Add a new D-Bus signal to report profiles that have been dynamically
created for already existing cards.

11 years agodbus: Support dynamically created card profiles
Mikel Astiz [Fri, 26 Oct 2012 06:23:49 +0000 (08:23 +0200)]
dbus: Support dynamically created card profiles

Use the recently added core hook to detect when new profiles are added
to a card, and update the D-Bus object accordingly.

11 years agoalsa-mixer: Fix the analog-output-speaker-always path
David Henningsson [Wed, 19 Dec 2012 10:24:33 +0000 (11:24 +0100)]
alsa-mixer: Fix the analog-output-speaker-always path

A left over "required-any" made this path useless for most people.
While we're at it, also add "Front Headphone" like for the normal
speaker path.

Tested-by: Colin Guthrie <gmane@colin.guthr.ie>
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
11 years agobuild-sys: Bump soname v3.0
Arun Raghavan [Mon, 17 Dec 2012 09:00:09 +0000 (14:30 +0530)]
build-sys: Bump soname

11 years agoalsa-mixer: Add device.icon-name property for some common ports
David Henningsson [Fri, 7 Dec 2012 15:37:23 +0000 (16:37 +0100)]
alsa-mixer: Add device.icon-name property for some common ports

If we expose this information, UIs can use this to make better
decisions about what icon to display.

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
11 years agobuild-sys: Bump soname v2.99.3
Arun Raghavan [Wed, 5 Dec 2012 03:51:58 +0000 (09:21 +0530)]
build-sys: Bump soname

11 years agoalsa-mixer: Add Dell Inspiron One 2020 to mic whitelist
David Henningsson [Wed, 5 Dec 2012 13:34:54 +0000 (14:34 +0100)]
alsa-mixer: Add Dell Inspiron One 2020 to mic whitelist

BugLink: https://bugs.launchpad.net/bugs/1073420
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
11 years agoalsa udev quirks: Add some more Dell devices to speaker whitelist
David Henningsson [Wed, 5 Dec 2012 13:08:30 +0000 (14:08 +0100)]
alsa udev quirks: Add some more Dell devices to speaker whitelist

Added Dell Inspiron 3420, 3520 and Vostro 2420, 2520.

Note that this is only necessary for kernels 3.3 to 3.5, as 3.6
has phantom jack support.

BugLink: https://bugs.launchpad.net/bugs/1076840
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
11 years agoalsa: Try to support non-standard rates in alsa-sink/source
Arun Raghavan [Mon, 3 Dec 2012 05:57:27 +0000 (11:27 +0530)]
alsa: Try to support non-standard rates in alsa-sink/source

We inadvertantly stopped supporting non-standard rates when the
passthrough work was done. This makes sure that if no standard rates are
supported, we try to fallback to whatever ALSA gives us.

11 years agobluetooth: Fix unacquired transports during sink resume
Mikel Astiz [Wed, 28 Nov 2012 18:20:58 +0000 (19:20 +0100)]
bluetooth: Fix unacquired transports during sink resume

The sink can be resumed while the source is still in PA_SOURCE_INIT.
This is the case if a module such as module-stream-restore routes the
audio to the sink during pa_sink_put(), leading to an inconsistent
state: the sink stays RUNNING but the transport is not actually
acquired.

11 years agobluetooth: Request headset audio during profile switch
Mikel Astiz [Mon, 3 Dec 2012 10:03:59 +0000 (11:03 +0100)]
bluetooth: Request headset audio during profile switch

When a headset is having a profile switch, we can either leave the
SCO state unmodified (as it was before this patch) or we can
alternatively request it (as older versions of PA).

This patch tries to avoid a potential regression in case a module
such as module-suspend-on-idle is not present, due to the provided
resume-on-running policy. Without this patch, and without such a policy,
the sink and sources would stay suspended until the user manually
performed another profile switch (i.e. hsp->off->hsp).

There are several other solutions currently being discussed as a longer
term solution, some of which require extendind the core. This patch is
therefore proposed as a short-term workaround to avoid the regression.

11 years agobluetooth: Do not setup stream before thread starts
Mikel Astiz [Mon, 3 Dec 2012 10:03:58 +0000 (11:03 +0100)]
bluetooth: Do not setup stream before thread starts

bt_transport_acquire() might get called from the main thread, in case
the IO thread hasn't been started yet. In this case, we should not call
setup_stream() since this is going to be called in the beginning of
thread_func().

11 years agobluetooth: Trivially refactor to call setup_stream() directly
Mikel Astiz [Mon, 3 Dec 2012 10:03:57 +0000 (11:03 +0100)]
bluetooth: Trivially refactor to call setup_stream() directly

If the transport is already acquired and the stream needs to be started,
call setup_stream() directly instead of bt_transport_acquire(u, TRUE).
Both calls are identical in these conditions, with the exception of the
log trace which has now been moved to setup_stream().

11 years agoconf: Load bluetooth-policy module by default
Mikel Astiz [Mon, 26 Nov 2012 17:32:09 +0000 (18:32 +0100)]
conf: Load bluetooth-policy module by default

Headset use-cases shouldn't get affected by this module and the support
for A2DP source is interesting, therefore load the module by default.

11 years agobluetooth: Disable profile auto-switch policy for headsets
Mikel Astiz [Mon, 26 Nov 2012 17:32:08 +0000 (18:32 +0100)]
bluetooth: Disable profile auto-switch policy for headsets

Given that headsets have just one single port exposing whether the
audio is streaming (playing) or not, it's not possible that
module-bluetooth-policy would distinguish A2DP/HSP cases, and thus
the automatic selection of the card profile is not deterministic.

For this reason, disable the policy entirely for headsets and focus
only on HFGW and A2DP source profiles.

11 years agobluetooth: Merge headset ports into one
Mikel Astiz [Mon, 26 Nov 2012 17:32:07 +0000 (18:32 +0100)]
bluetooth: Merge headset ports into one

Merge the former "hsp-output" and "a2dp-output" ports into one single
port, in order to fix the regression of having several independent
entries in the UI.

11 years agovirtual-surround: check if resampled memblock is not equal to input
Niels Ole Salscheider [Sat, 24 Nov 2012 11:32:51 +0000 (12:32 +0100)]
virtual-surround: check if resampled memblock is not equal to input

Since commit e32a408b3cdd46857fdf12210c1bf5bdbf3a96f8, we silence the
input memblock in order to give the resampler enough input samples, if
necessary.
But if there is no need to resample the hrir, the resampled memblock is
actually the same as the input memblock. Thus, we have to make sure that
we do not silence it in this case.

11 years agobluetooth: Unload device module when no audio profiles connected
Tanu Kaskinen [Thu, 22 Nov 2012 14:20:28 +0000 (15:20 +0100)]
bluetooth: Unload device module when no audio profiles connected

Without this patch, device modules will be left around after the
device has been disconnected and when they are reconnected, the
discovery module will load duplicate device module instances.

BugLink: https://bugs.freedesktop.org/show_bug.cgi?id=57239
11 years agobluetooth: Run the discovery hook only when necessary
Mikel Astiz [Thu, 22 Nov 2012 14:20:27 +0000 (15:20 +0100)]
bluetooth: Run the discovery hook only when necessary

This is a minor optimization too, but the main benefit is that it's
makes the code easier to understand (I hope), since run_callback()
won't be called at times when it's not needed.

11 years agobluetooth: Add helper pa_bluetooth_device_any_audio_connected()
Tanu Kaskinen [Thu, 22 Nov 2012 14:20:26 +0000 (15:20 +0100)]
bluetooth: Add helper pa_bluetooth_device_any_audio_connected()

The new helper function makes it easier to check whether any audio
profiles are connected. That information is needed by the discovery
module for deciding whether a new device module should be loaded. The
device module should use this information too to unload itself at the
right time, but that's currently not implemented.

11 years agobluetooth: Rename former device_is_audio()
Mikel Astiz [Thu, 22 Nov 2012 14:20:25 +0000 (15:20 +0100)]
bluetooth: Rename former device_is_audio()

Use a more accurate name for the function since it doesn't just check
if it is an audio device (which can be detected quite early), but it
also checks if the most relevant properties (device info, etc.) have
been received.

Besides, add the const qualifier to the pointer since it's not going to
be modified.

11 years agobluetooth: Ignore Device.Connected
Tanu Kaskinen [Thu, 22 Nov 2012 14:20:24 +0000 (15:20 +0100)]
bluetooth: Ignore Device.Connected

The Device.Connected was only used for tracking whether a device module
should be loaded, but that information is already included in the
individual profile state properties. The property can therefore be
completely ignored without any loss in functionality.

11 years agobluetooth: Add a pa_bluetooth_discovery pointer to pa_bluetooth_device.
Tanu Kaskinen [Tue, 20 Nov 2012 12:51:27 +0000 (14:51 +0200)]
bluetooth: Add a pa_bluetooth_discovery pointer to pa_bluetooth_device.

This simplifies some function argument lists.

11 years agostream-restore: Add missing method handler argument.
Juho Hämäläinen [Tue, 20 Nov 2012 06:30:22 +0000 (08:30 +0200)]
stream-restore: Add missing method handler argument.

Stream-restore DBus API method argument list is missing last boolean
argument apply_immediately, causing assert to fail in AddEntry handling.

Signed-off-by: Juho Hämäläinen <jusa@hilvi.org>
11 years agocore: Remove bad free() call
Arun Raghavan [Mon, 19 Nov 2012 15:58:22 +0000 (21:28 +0530)]
core: Remove bad free() call

The string created when trying to use XDG_RUNTIME_DIR is freed before it
is used in a debug message, and is freed again.

https://bugs.freedesktop.org/show_bug.cgi?id=57280

11 years agosink-input, source-output: Check rate update success for passthrough
Arun Raghavan [Mon, 19 Nov 2012 07:22:28 +0000 (12:52 +0530)]
sink-input, source-output: Check rate update success for passthrough

This makes sure we don't try to plug in a passthrough stream if the
final sink/source sample spec doesn't match what we want. In the future,
we might want to change rate updates to try a full sample spec update
for passthrough streams.

https://bugs.freedesktop.org/show_bug.cgi?id=50951

11 years agosource-output: Fix volume fixup for rate update
Arun Raghavan [Mon, 19 Nov 2012 07:03:25 +0000 (12:33 +0530)]
source-output: Fix volume fixup for rate update

The could that should have been after the rate update ended up being
before, which is incorrect.

11 years agobuild-sys: Drop ChangeLog generation v2.99.2
Arun Raghavan [Sat, 17 Nov 2012 05:25:05 +0000 (10:55 +0530)]
build-sys: Drop ChangeLog generation

This drops ChangeLog generation from git. It does not make sense to
distribute 500 kB of ChangeLog given how easy it is these days to browse
history with git.

We might replace this with a release-annoucement-esque high-level
summary of changes.

11 years agobuild-sys: Bump soname
Arun Raghavan [Sat, 17 Nov 2012 05:24:16 +0000 (10:54 +0530)]
build-sys: Bump soname

11 years agobuild-sys: Bump BlueZ dependency to 4.99
Arun Raghavan [Fri, 16 Nov 2012 17:43:26 +0000 (23:13 +0530)]
build-sys: Bump BlueZ dependency to 4.99

Since the old UNIX fd IPC mechanism was removed and we rely on the D-Bus
based API, this is the oldest version that is known to reliably work.

11 years agomodules: Micro-optimisation for rewind_requested paths
Arun Raghavan [Fri, 16 Nov 2012 17:39:15 +0000 (23:09 +0530)]
modules: Micro-optimisation for rewind_requested paths

Since some devices can be chatty with regards to how often they return
from poll(), this adds a PA_UNLIKELY() to all the the rewind_requested
checks in our sink modules to make the general case (no rewind was
requested) the fast path.

11 years agosink: Process rewind requests also when suspended.
Tanu Kaskinen [Thu, 30 Aug 2012 13:50:13 +0000 (16:50 +0300)]
sink: Process rewind requests also when suspended.

When a rewind is requested on a sink input, the request parameters are
stored in the pa_sink_input struct. The parameters are reset during
rewind processing, and if the sink decides to ignore the rewind
request due to being suspended, stale parameters are left in
pa_sink_input. It's particularly problematic if the rewrite_bytes
parameter is left at -1, because that will prevent all future rewind
processing on that sink input. So, in order to avoid stale parameters,
every rewind request needs to be processed, even if the sink is
suspended.

Reported-by: Uoti Urpala
11 years agosink, source: Prevent unnecessary rate update attempts
Arun Raghavan [Thu, 15 Nov 2012 10:21:52 +0000 (15:51 +0530)]
sink, source: Prevent unnecessary rate update attempts

We don't need to try a rate update if the desired sample rate is the
same as the one the sink or source is already using.

11 years agoalsa-mixer: Prefer "Digital Input Source:Digital Mic 1"
David Henningsson [Fri, 16 Nov 2012 12:16:04 +0000 (13:16 +0100)]
alsa-mixer: Prefer "Digital Input Source:Digital Mic 1"

...over "Digital Input Source:Analog Input". It makes life a little
easier for users of Dell xps m1330.

Just an old Ubuntu delta I never upstreamed until now.
The patch was originally written by Daniel T Chen <crimsun@ubuntu.com>.

BugLink: https://bugs.launchpad.net/bugs/453966
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
11 years agomainloop: Don't care about the mainloop state variable when waking up the mainloop.
Tanu Kaskinen [Mon, 2 Apr 2012 12:01:05 +0000 (15:01 +0300)]
mainloop: Don't care about the mainloop state variable when waking up the mainloop.

If the mainloop is just about to enter polling, but m->state
is not POLLING yet when some other thread calls
pa_mainloop_wakeup(), the mainloop will not be woken up.
It's safe to write to the wakeup pipe at any time, so let's
just remove the check.

11 years agomainloop: Change wakeup_requested type from pa_bool_t to pa_atomic_t.
Tanu Kaskinen [Mon, 2 Apr 2012 12:01:02 +0000 (15:01 +0300)]
mainloop: Change wakeup_requested type from pa_bool_t to pa_atomic_t.

The variable is accessed from multiple threads, so it should
be atomic.

11 years agoalsa-mixer: Add "CLFE" and "Bass Speaker" names
David Henningsson [Fri, 9 Nov 2012 06:10:24 +0000 (07:10 +0100)]
alsa-mixer: Add "CLFE" and "Bass Speaker" names

These are sometimes being used in the HDA driver and we should
support them.

Signed-off-by: David Henningsson <david.henningsson@canonical.com>
11 years agoman: Correction for how sample rate switching is disabled
Arun Raghavan [Mon, 12 Nov 2012 16:54:52 +0000 (22:24 +0530)]
man: Correction for how sample rate switching is disabled

Setting to 0 in the parser causes daemon startup to fail.

11 years agobuild-sys: Correct bluez support error if sbc is missing
Sjoerd Simons [Sat, 10 Nov 2012 16:41:27 +0000 (16:41 +0000)]
build-sys: Correct bluez support error if sbc is missing

11 years agoalsa-mixer: Add a few more machines to internal mic whitelist
David Henningsson [Fri, 9 Nov 2012 11:14:18 +0000 (12:14 +0100)]
alsa-mixer: Add a few more machines to internal mic whitelist

Based on feedback in the bug below (comments 128, 129, 131).

BugLink: https://bugs.launchpad.net/bugs/946232
Signed-off-by: David Henningsson <david.henningsson@canonical.com>
11 years agopulse: Fix hole handling in pa_stream_peek().
Tanu Kaskinen [Wed, 7 Nov 2012 14:52:37 +0000 (16:52 +0200)]
pulse: Fix hole handling in pa_stream_peek().

Previously, if there was a hole in a recording stream,
pa_stream_peek() would crash. Holes could be handled silently inside
pa_stream_peek() by generating silence (wouldn't work for compressed
streams, though) or by skipping any holes. However, I think it's
better to let the caller decide how the holes should be handled, so
in case of holes, pa_stream_peek() will return NULL data pointer and
the length of the hole in the nbytes argument.

This change is technically an interface break, because previously the
documentation didn't mention the possibility of holes that need
special handling. However, since holes caused crashing anyway in the
past, it's not a regression if applications keep misbehaving due to
not handing holes properly.

Some words about when holes can appear in recording streams: I think
it would be reasonable behavior if overruns due to the application
reading data too slowly would cause holes. Currently that's not the
case - overruns will just cause audio to be skipped. But the point is
that this might change some day. I'm not sure how holes can occur
with the current code, but as the linked bug shows, they can happen.
It's most likely due to recording from a monitor source where the
thing being monitored has holes in its playback stream.

BugLink: http://bugs.launchpad.net/bugs/1058200