This problem due to the way audio buffers are mixed when
low power mode is active was addressed by commits 19ddf0eb
and 8a04fe03 but only partially. As a matter of fact, when more
than one audio track is playing, the problem is still present.
This is most noticeable when playing music with screen off
and a notification or navigation instruction is played: in this case,
the music or notification is likely to skip.
The fix consists in declaring the mixer ready if all active tracks
are ready. Previous behavior was to declare ready if at least one track was
ready. To avoid that one application failing to fill the track buffer blocks other
tracks indefinitely, this condition is respected only if the mixer was ready
in the previous round.
Issue 5799167.
Change-Id: Iabd4ca08d3d45f563d9824c8a03c2c68a43ae179
The calculation done in prepareTracks_l() for the minimum amount
off frames needed to mix one output buffer had 2 issues:
- the additional sample needed for interpolation was not included
- the fact that the resampler does not acknowledge the frames consumed
immediately after each mixing round but only once all frames requested have been used
was not taken into account.
Thus the number of frames available in track buffer could be considered sufficient although
it was not and the resampler would abort producing a short silence perceived as a click.
Issue 5727099.
Change-Id: I7419847a7474c7d9f9170bedd0a636132262142c
Make the standby time for AudioFlinger configurable using a system
property. Default AudioFlinger behavior is to go into standby
(allowing the audio outputs to underflow) after there has been nothing
to mix and AudioFlinger has just been pumping out silence for the
configured standby time (which defaulted to 3 seconds).
Now, by setting the "ro.audio.flinger_standbytime_ms" property in
their platform init.rc, platforms can override this default and
control the standby time. If the property is missing or malformed,
the old default value of 3 seconds will be used instead.
Change-Id: Ic9fa8b5f5bccee493bc72c65e408d3fd8ddd1059
Signed-off-by: John Grossman <johngro@google.com>
The maximum sleep time allowed in the mixer thread when audio tracks
are enabled but not ready for mixing is derived from the latency
reported by the output stream.
This does not work for A2DP where the latency also reflects encoding, decoding
and transfer time.
Modified activeSleepTimeUs() to take A2DP case into account.
Issue 5682206.
Change-Id: I3784ac01fb6f836b5a6ce6f764fb15347586de35
Progressively reduce the sleep time applied in MixerThread::threadLoop()
in case of consecutive application underruns to avoid starving the audio HAL.
As the default sleep time is longer than the duration of an audio buffer
we ended up writing less data than needed by the audio HAL if
the condition persisted.
Issue 5553055.
Change-Id: I2b23ee79c032efa945025db228beaecd1e07a2e5
When audio effects are enabled, a noise can be heard at the
beginning of the new song when skipping to next song in music app.
This is because some effects (especially virtualizer) have a tail.
This tail was not played when previous song was stopped because effects were
not processed when no tracks were present on a given session. This is to
reduce CPU load when effects are enabled but no audio is playing.
The tail was then rendered when the new song was started.
Added a delay before stopping effect process after all tracks have been removed from a session.
Issue 5584880.
Change-Id: I815e0f7441f9302e8dfe413dc269a94e4cc6fd95
Commit 19ddf0eb introduced a problem with applications (like SoundPool)
relying on an underrun condition to detect end of playback instead of
stopping the track when all data is written.
AudioFlinger would keep waiting for new data in case of partial buffer
filling and never reach the underrun condition.
Added a mechanism to wait no more than once if not enough frames are present
in the track buffer.
Issue 5585490.
Change-Id: I131e605ff6070831a01ddf734e68459e3bf2354b
The addition of low power audio playback mode made that audio buffer consumption
by audio HAL can now happen in bursts. This makes that requesting audio data
from an AudioTrack for mixing can happen at much shorter intervals than before.
This revealed an existing problem where AudioFlinger would consider a track ready
for mixing although not enough frames were ready to completely fill one output buffer,
thus creating short periods of silence.
The fix consists in waiting for enough frames to be ready in AudioTrack buffer before
declaring a track ready for mixing. This minimum is not applied when the track is stopped
to allow the buffer to be emptied completely.
Change-Id: I6d04f9b65db5af85b0b53f0a5674be7ec02f9e9f
The assembly expects arguments to live at fixed offsets from the stack pointer
which are invalid if the code is inlined.
Change-Id: Ie93e93c5c69774079112345754fbc85896fc2f64
The get_next_write_timestamp method introduced to the audio HAL is optional.
HALs which do not implement it leave it set to NULL. Callers (there is
currently only one in the AudioMixer code) need to be certain to check for NULL
before invoking it.
Change-Id: I88ba43bb53bec081c98c9a8842936c4fbfdd44f6
Squashed merge from master-tungsten of the following changes:
commit 73d09e18c4557e583a1684d44d598a1a02fd0cf2
Author: John Grossman <johngro@google.com>
Date: Mon Jun 20 13:57:44 2011 -0700
Remove TungstenMisc and rename LinearTransform
Change-Id: Ie8aa3e24e09fdbf6ef8996c26deb9c5640e20d1b
commit 3114aabe76ad733b59929d87e49c68229f5ae2e8
Author: John Grossman <johngro@google.com>
Date: Fri Jun 3 10:47:16 2011 -0700
Name changes and spelling fixes.
+ Replace the term TungstenTime with the Eugene-approved term CommonTime.
+ Fix a spelling error in a comment I noticed.
Change-Id: I8c10d618206826d16055f78c7724e24443bb03fd
commit cbf2903ab6893b6e662514e2f6d670e268a419df
Author: John Grossman <johngro@google.com>
Date: Fri Apr 15 09:27:54 2011 -0700
Migrate Tungsten code from the HC-Tungsten to the Master-Tungsten branch.
Change-Id: I95372d913a0761d90168edb4016f5ece0ea74502
commit bc7c46aa629f9883e959ef23de8da297f9eb508b
Author: Jason Simmons <jsimmons@google.com>
Date: Mon Jun 20 13:59:17 2011 -0700
Create a separate class for timed AudioTracks
commit 43be3231034ff8537fdd84422a7954780038671f
Author: John Grossman <johngro@google.com>
Date: Mon Jun 27 18:59:12 2011 -0700
Move libaah_rtp over from the vendor directory.
Also move factor PipeEvent out into utils.
Change-Id: Id3877c66efe22d771cf3ef4877107e431b828e37
commit 17526eb3148c9c3d4365b6d5b47e8dc13bca71b6
Author: John Grossman <johngro@google.com>
Date: Mon Jun 27 17:06:49 2011 -0700
Name changes for the TRTP Players s/tungsten/aah/g
Change-Id: I55e9ad13003f6aa6a36955b54426a7efbe31ac51
commit 423fc1bfc0fda799c421a650c83c4b9293b1a08c
Author: Jason Simmons <jsimmons@google.com>
Date: Mon Jun 20 17:56:09 2011 -0700
More timed AudioFlinger changes requested by code review:
* change trimTimedBufferQueue to trimTimedBufferQueue_l
* create one timed audio buffer heap per client process instead of one per track
* grow the silence buffer on demand
* some error handling fixes in timed getNextBuffer
* calculate the next output PTS in all mixer and track hooks
Change-Id: Ifc51a08b55029b7c48902ab2f22933ad7bafe1ad
commit a148e2674b1d3cb73289b82b85c333f0a66824a9
Author: John Grossman <johngro@google.com>
Date: Mon Jun 20 17:02:24 2011 -0700
Move the A@H time service into frameworks/base
Change-Id: I5c570cde70e8931e205516cb33517585804ce841
commit dfa438fa49bdaeeb2ec5fd0d17b30d881608b6b1
Author: John Grossman <johngro@google.com>
Date: Mon Jun 20 11:55:36 2011 -0700
Fix the build after Mike's code moving.
Change-Id: Ia883643ded252168bcc5a70584ab6ce97bb05266
commit 04489474ec8e73efe1bf52918831f41659033162
Author: John Grossman <johngro@google.com>
Date: Fri Jun 17 14:19:50 2011 -0700
Refactor the local/common clock services.
This change is one of a set of 5 changes made to different repositories. Look
for this comment in all of them.
Refactor the local/common clock services in tungsten to match android best
practice. Notable changes include
+ The kernel no longer knows anything about common time. Common time has been
moved completely up into user land. This has an impact on the accuracy of the
timesync debugging code, and the netfilter assisted approach to network based
timesync is going to have to be modified.
+ The timesync driver used by A@H is now just local time driver.
+ The kernel no longer needs access to the linear transform math code, and it
has been removed.
+ A new HAL has been introduced to expose the concept of local time to the
system.
+ A non-slewable stub implementation of the local time HAL based on
CLOCK_MONOTONIC has been added.
+ The TungstenTime library has been eliminated. Its functionality has been
distributed among the common time binder service, the local time hal and the
linear transform utility code.
+ All clients of the old TungstenTime library have been changed to be clients of
the binder service, the hal and the utility code.
+ The reset_tt utilities have been removed, they no longer have a purpose in the
system.
+ more progress has been made in eliminating the word "tungsten" from the code.
Things left to do include
+ Finish getting rid of tungsten from the time service.
+ Move the time service into the framework; AudioFlinger's new timed mode
depends on it and the service cannot continue to live in vendor tungsten.
Change-Id: I999b6cfb4a9d267818a86d747c35eecfc6693101
commit d48194545eed1116a84d81e2fb53315d2b0701a7
Author: Jason Simmons <jsimmons@google.com>
Date: Thu Jun 16 14:22:46 2011 -0700
Change the interface of the AudioMixer and AudioBufferProvider to accept a presentation timestamp
Change-Id: Ice2df5628d45a7f77100e7008103b35b3d3160a4
commit 02561419db82b01ffb28df38000716c612988427
Author: John Grossman <johngro@google.com>
Date: Tue May 10 14:00:21 2011 -0700
Put in a hack for controling master volume in the policy manager.
Fix initial master volume reporting.
Change-Id: Ia6caf2bbc6083c5f99fab852baa40fff10fc5fc7
commit 549cdc3ba115dc654cdade261fb055c72c6cdb79
Author: John Grossman <johngro@google.com>
Date: Wed May 4 11:46:17 2011 -0700
Make certain the logic for computing the output stream mixing point is hardened
against underflow and overflow when input and output sample rates don't match.
Change-Id: I5ebea07c9938107b435bec7413418622767e4e16
commit 8043d8ed63f51e76d452d22be7d453d4a7794530
Author: Jason Simmons <jsimmons@google.com>
Date: Wed Apr 27 18:06:27 2011 -0700
Add the patch for timed audio support to the mono resampler
Change-Id: I526f34ae9d1e8e3b0ed2fb05af3d024d5c5fe711
commit 2be89486ef23f0b0b0cc2dc25a4c0ee691043f00
Author: John Grossman <johngro@google.com>
Date: Wed Apr 27 10:38:57 2011 -0700
Extend the AudioHWInterface to allow it to specify the initial master volume used by AudioFlinger.
Change-Id: I8823330801c927494cf7ca31a6b8f9264fbfbb26
commit ff89a4d5e37e6a05a2b03f79ab4e97833dd66393
Author: John Grossman <johngro@google.com>
Date: Wed Apr 27 09:07:14 2011 -0700
Fix an issue with inconsistent volume reporting.
Changed masterVolume() to return the same value as the last call
to setMasterVolume when the HW layer is implementing master
volume control. The masterVolume/setMasterVolume API seems to be
an idea which was abandonded a long time ago; as of today the
system only ever sets it to 1.0 at startup and then never changes
it. Until we can figure out how the concept of external
amplifier gain control fits into the Android audio framework,
Tungsten is exposing this API via a hack-tastic invoke back door
in the TungstenRXPlayer and needs the getter/setter results to be
consistent.
Change-Id: I2ac730fa8fc9ee28c88f1a8e6f2e493eb5b65544
commit 086511b2d19cceb976747ac23e12b73fc7c28bea
Author: Jason Simmons <jsimmons@google.com>
Date: Mon Apr 25 16:07:19 2011 -0700
Add handling of timed audio tracks in the generic resampling mixer
Change-Id: Ic3be1d21b1117f1b233808be543c28a0dcec4792
Change-Id: I6ec5d2bca9b8ebc0acd395a7dd92e1a48fcdfa9b
Signed-off-by: Mike J. Chen <mjchen@google.com>
Signed-off-by: John Grossman <johngro@google.com>
Signed-off-by: Jason Simmons <jsimmons@google.com>
When AudioEffectTest is executed, an Equalizer is created
and enabled on a MediaPlayer session. Effects on the output
mix are therefore suspended.
Then the MediaPlayer is released with the effect still enabled.
In this case, Audioflinger::purgeStaleEffects_l() fails to restore
the suspended effects when the effect attached to the released audio session
is removed.
When subsequent tests are executed on output mix effects, these effects cannot be
enabled as they are still suspended.
Fixed purgeStaleEffects_l() to restore suspended effects if the effect removed is enabled.
Also fixed EffectHandle::disconnect() to only restore suspended effects if the disconnected
handle actually has control over the effect.
Change-Id: I67232e7c34680b0cc01abfd57d5d510a524e5d4f
AudioFlinger logs a warning when a write to the audio HAL
takes too long to return. The threshold for this warning is
a rule of thumb based on the assumption that the audio HAL will consume
buffers at a regular pace.
The introduction of low power audio mode with larger buffers and writes
occuring in bursts makes that this threshold is often exceeded resulting
in excessive and misleading warnings.
The threshold is raised to remove unwanted warnings but we should reconsider
the usefulness of this warning altogether.
Change-Id: I5ef6898ea28d879cede3e47da542a64092a3cca4
This problem only occurs when audio effects are present and
the music volume is applied by one effect engine.
When connecting or disconnecting A2DP, audio effects are moved from
one mixer thread to another. When removed from the source thread,
the effect is stopped but it is not restarted when added to the
destination thread.
This regression was introduced by commit 21b5c47e.
Change-Id: I4cc578d8d760ec65b185032b6fda98c739d331bc
Fixed several regressions in automated audio effect tests due
to changes in effect framework and visualizer FFT output range.
- Do not suspend Volume effect on session 0 when effects are
enabled on specific sessions.
- Adapt energy detection thresholds to new visualizer FFT range.
- Leave more time for BassBoost and Virtualizer effects to ramp up
before measuring the effect.
- Removed second insert reverb left by mistake on the player session
in preset reverb test.
Change-Id: I7a1ad1372d783fa7900eb9dd1d3b47f54d8d766f
There is a possiblility that the condition on which RecordThread::checkForNewParameters_l()
waits after updating the command completion status is never signalled.
This happens if the thread executing ThreadBase::setParameters() has timed out waiting
for the status (for instance if the audio HAL takes too long to execute the setParameters()
command. Then the RecordThread is stuck forever.
The fix consists in waiting for the condition with a timeout in RecordThread::checkForNewParameters_l().
Change-Id: I7fc671bc2fc43ba4acb65a2beb33ee05742f091e
When the A2DP headset is connected, there is a possible
race condition when the audio tracks are moved from
the mixer thread attached to the speaker output to the thread
attached to A2DP output.
As the request to clear the stream type to output mapping cache in
the client process is asynchronous, it is possible that the flag
indicating to the client audio track to re-create the IAudioTrack
on the new thread is processed before the cache is invalidated.
In this case, the track will be attached to the old thread and
music will continue playing over the device speaker instead of being
redirected to A2DP headset.
Change-Id: Ib2ce1eb5320eaff83287b93779061bf4e7a330df
The interpretation of BT NREC by AudioFlinger to enable
or disable AEC and NS was wrong: NREC to ON (default) means
the phone (Audio Gateway) must enable local AEC and NS.
Change-Id: I88a264e7fc9831c43bbace4f6b585baec73f2006
Do not call audio HAL functions on the primary HW interface
if it could not be initialized properly.
Change-Id: If54059c8fd188d6c1686f9e0439994fe9411478a
AudioManager.isWiredHeadsetOn() should not require permission MODIFY_AUDIO_SETTINGS.
Remove permission checks on all getters in audio policy manager as permission enforcement
is really usefull for setters.
Also deprecate AudioManager.isWiredHeadsetOn() which name and implementation are deceptive.
Change-Id: I38f8df7c26c0d417bf0e2b74e4c11c2d143f2ecd
Before this change, CPU and memory usage for an audio effect were
registered and checked against the limit by audio policy manager
upon effect instantiation. Even if an effect was not enabled
it would prevent another effect to be created if the CPU load budget
was exceeded, which was too restrictive.
This change adds a method to register/unregister CPU load only when
an effect is enabled or disabled.
It also adds a mechanism to place all effects on the global output mix
in suspend state (disabled) when an effect is enabled on a specific session.
This will allow applications using session effects to have the priority
over others using global effects.
Also fixes some issues with suspend/restore mechanism:
- avoid taking actions when an effect is disconnected and was not enabled.
- do not remove a session from the suspended sessions list when corresponding
effect chain is destroyed.
Change-Id: I5225278aba1ae13d0d0997bfe26a0c9fb46b17d3
Don't remove effects until the session they are in goes away or all
AudioEffects have been explicitly released. This allows the control
panel process to die without stopping the effects.
Change-Id: I4496e5df080230ca1af149dec95c1309ab8ea888
Some methods would not check that the output orinput stream of a thread
was still valid before calling functions on its interface.
This could cause a crash if those methods where called while the output or
input was being closed by another thread.
Make sure that the output or input stream pointer is cleared before closing the
stream.
Always check that the output or input pointer is not null before calling
functions at the stream interface.
Generalize the use of initCheck() method to verify that the output or input
stream is not null.
Change-Id: I9d9ca6b744d011bcf3a7bbacb4a581ac1477bfa5
Disable AEC and NS when the Bluetooth SCO headset in use indicates it
implements those pre processings.
Change-Id: I93f3d10b0a27243d5dbff7182639576fc0c6d862
Add the possibility for the effect framework to suspend
(temporarily disable process) and restore audio effects.
This feature will be usefull to disable pre processing under certain
conditions and better control coexistence of audio effects
on output mix and specific sources.
Change-Id: I79b195982cc48748d5708308fb1647b9c3c34cc6
If a pre processing effect is detroyed while enabled and capture is active,
there was a possibility that the effect engine is released by the framework
while still processed by the audio HAL.
The fix consists in not releasing the engine in EffectModule::removeHandle()
but just flag the effect as being detroyed to avoid further calls to functions
on the engine effect interface.
The effect interface is then removed from the audio HAL safely in
EffectChain::removeEffect_l() while holding the EffectChain mutex.
Change-Id: I71fab30d9145062af8644f545a1f1d4d3e7e7f02
The problem is that the audio HAL fails to acquire the wake lock when playing the notification.
This is because of a change that removed the mediaserver process form the system group for honeycomb.
The fix consists in requesting the wake lock from PowerManagerService when AudioFlinger mixer
wakes up.
A consequence of this change is that audio HALs or pcm drivers do not have to hold wake locks
anymore as in the past.
Change-Id: I4fb3cc84816c9c408ab7fec75886baf801e1ecb5
Added APIs to control pre processes applied on captured audio.
Those APIs are still hidden until reviewed by API council.
Three types of standard pre processes are supported:
- Automatic Gain Control (AGC) by AutomaticGainControl class
- Acoustic Echo Cancellation (AEC) by AcousticEchoCanceler class
- Noise Suppression (NS) by NoiseSuppressor class
A method is added to AudioEffect class to query audio pre processings
applied by default by the platform on a given AudioRecord session ID.
Change-Id: I0b9fceeb8c704dd06319c3b52b85c96fe871d51d
Dump of media.audio_flinger service was only listing effects on output threads.
Moved the dump of effect chains from PlaybackThread to ThreadBase class so that
pre processings on RecordThread are also listed.
Change-Id: If8bc74023c12b9c2371f1b300743b156ceca7b87
Audio effect framework is extended to suport effects on
output and input audio path.
AudioFlinger: Support for audio effects and effect chains is
moved from PlaybackThread class to ThreadBase class so that
RecordThread can manage effects.
Effects of type pre processing are allowed on record thread
only. When a pre processing is enabled, the effect interface handle is
passed down to the input stream so that the audio HAL can call the
process function. The record thread loop calls the effect chain process
function that will only manage the effect state and commands and skip the
process function.
AudioRecord: The audio session is allocated before calling getInput() into
audio policy serice so that the session is known before the input theead is
created and pre processings can be created on the correct session.
AudioPolicyService: default pre processing for a given input source are
loaded from audio_effects.conf file.
When an input is created, corresponding effects are created and enabled.
Change-Id: Id17119e0979b4dcf189b5c7957fec30dc3478790
