1511 lines
44 KiB
C++
Raw Normal View History

/* //device/extlibs/pv/android/AudioTrack.cpp
**
** Copyright 2007, The Android Open Source Project
**
** Licensed under the Apache License, Version 2.0 (the "License");
** you may not use this file except in compliance with the License.
** You may obtain a copy of the License at
**
** http://www.apache.org/licenses/LICENSE-2.0
**
** Unless required by applicable law or agreed to in writing, software
** distributed under the License is distributed on an "AS IS" BASIS,
** WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
** See the License for the specific language governing permissions and
** limitations under the License.
*/
//#define LOG_NDEBUG 0
#define LOG_TAG "AudioTrack"
#include <stdint.h>
#include <sys/types.h>
#include <limits.h>
#include <sched.h>
#include <sys/resource.h>
#include <private/media/AudioTrackShared.h>
#include <media/AudioSystem.h>
#include <media/AudioTrack.h>
#include <utils/Log.h>
#include <binder/Parcel.h>
#include <binder/IPCThreadState.h>
#include <utils/Timers.h>
#include <utils/Atomic.h>
#include <cutils/bitops.h>
#include <system/audio.h>
#include <system/audio_policy.h>
#define LIKELY( exp ) (__builtin_expect( (exp) != 0, true ))
#define UNLIKELY( exp ) (__builtin_expect( (exp) != 0, false ))
namespace android {
// ---------------------------------------------------------------------------
// static
status_t AudioTrack::getMinFrameCount(
int* frameCount,
int streamType,
uint32_t sampleRate)
{
int afSampleRate;
if (AudioSystem::getOutputSamplingRate(&afSampleRate, streamType) != NO_ERROR) {
return NO_INIT;
}
int afFrameCount;
if (AudioSystem::getOutputFrameCount(&afFrameCount, streamType) != NO_ERROR) {
return NO_INIT;
}
uint32_t afLatency;
if (AudioSystem::getOutputLatency(&afLatency, streamType) != NO_ERROR) {
return NO_INIT;
}
// Ensure that buffer depth covers at least audio hardware latency
uint32_t minBufCount = afLatency / ((1000 * afFrameCount) / afSampleRate);
if (minBufCount < 2) minBufCount = 2;
*frameCount = (sampleRate == 0) ? afFrameCount * minBufCount :
afFrameCount * minBufCount * sampleRate / afSampleRate;
return NO_ERROR;
}
// ---------------------------------------------------------------------------
AudioTrack::AudioTrack()
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
: mStatus(NO_INIT),
mIsTimed(false)
{
}
AudioTrack::AudioTrack(
int streamType,
uint32_t sampleRate,
int format,
int channelMask,
int frameCount,
uint32_t flags,
callback_t cbf,
void* user,
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
int notificationFrames,
int sessionId)
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
: mStatus(NO_INIT),
mIsTimed(false)
{
mStatus = set(streamType, sampleRate, format, channelMask,
frameCount, flags, cbf, user, notificationFrames,
0, false, sessionId);
}
AudioTrack::AudioTrack(
int streamType,
uint32_t sampleRate,
int format,
int channelMask,
const sp<IMemory>& sharedBuffer,
uint32_t flags,
callback_t cbf,
void* user,
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
int notificationFrames,
int sessionId)
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
: mStatus(NO_INIT),
mIsTimed(false)
{
mStatus = set(streamType, sampleRate, format, channelMask,
0, flags, cbf, user, notificationFrames,
sharedBuffer, false, sessionId);
}
AudioTrack::~AudioTrack()
{
LOGV_IF(mSharedBuffer != 0, "Destructor sharedBuffer: %p", mSharedBuffer->pointer());
if (mStatus == NO_ERROR) {
// Make sure that callback function exits in the case where
// it is looping on buffer full condition in obtainBuffer().
// Otherwise the callback thread will never exit.
stop();
if (mAudioTrackThread != 0) {
mAudioTrackThread->requestExitAndWait();
mAudioTrackThread.clear();
}
mAudioTrack.clear();
IPCThreadState::self()->flushCommands();
AudioSystem::releaseAudioSessionId(mSessionId);
}
}
status_t AudioTrack::set(
int streamType,
uint32_t sampleRate,
int format,
int channelMask,
int frameCount,
uint32_t flags,
callback_t cbf,
void* user,
int notificationFrames,
const sp<IMemory>& sharedBuffer,
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
bool threadCanCallJava,
int sessionId)
{
LOGV_IF(sharedBuffer != 0, "sharedBuffer: %p, size: %d", sharedBuffer->pointer(), sharedBuffer->size());
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
if (mAudioTrack != 0) {
LOGE("Track already in use");
return INVALID_OPERATION;
}
int afSampleRate;
if (AudioSystem::getOutputSamplingRate(&afSampleRate, streamType) != NO_ERROR) {
return NO_INIT;
}
uint32_t afLatency;
if (AudioSystem::getOutputLatency(&afLatency, streamType) != NO_ERROR) {
return NO_INIT;
}
// handle default values first.
if (streamType == AUDIO_STREAM_DEFAULT) {
streamType = AUDIO_STREAM_MUSIC;
}
if (sampleRate == 0) {
sampleRate = afSampleRate;
}
// these below should probably come from the audioFlinger too...
if (format == 0) {
format = AUDIO_FORMAT_PCM_16_BIT;
}
if (channelMask == 0) {
channelMask = AUDIO_CHANNEL_OUT_STEREO;
}
// validate parameters
if (!audio_is_valid_format(format)) {
LOGE("Invalid format");
return BAD_VALUE;
}
// force direct flag if format is not linear PCM
if (!audio_is_linear_pcm(format)) {
flags |= AUDIO_POLICY_OUTPUT_FLAG_DIRECT;
}
if (!audio_is_output_channel(channelMask)) {
LOGE("Invalid channel mask");
return BAD_VALUE;
}
uint32_t channelCount = popcount(channelMask);
audio_io_handle_t output = AudioSystem::getOutput(
(audio_stream_type_t)streamType,
sampleRate,format, channelMask,
(audio_policy_output_flags_t)flags);
if (output == 0) {
LOGE("Could not get audio output for stream type %d", streamType);
return BAD_VALUE;
}
mVolume[LEFT] = 1.0f;
mVolume[RIGHT] = 1.0f;
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
mSendLevel = 0;
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
mFrameCount = frameCount;
mNotificationFramesReq = notificationFrames;
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
mSessionId = sessionId;
mAuxEffectId = 0;
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
// create the IAudioTrack
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
status_t status = createTrack_l(streamType,
sampleRate,
(uint32_t)format,
(uint32_t)channelMask,
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
frameCount,
flags,
sharedBuffer,
output,
true);
if (status != NO_ERROR) {
return status;
}
if (cbf != 0) {
mAudioTrackThread = new AudioTrackThread(*this, threadCanCallJava);
if (mAudioTrackThread == 0) {
LOGE("Could not create callback thread");
return NO_INIT;
}
}
mStatus = NO_ERROR;
mStreamType = streamType;
mFormat = (uint32_t)format;
mChannelMask = (uint32_t)channelMask;
mChannelCount = channelCount;
mSharedBuffer = sharedBuffer;
mMuted = false;
mActive = 0;
mCbf = cbf;
mUserData = user;
mLoopCount = 0;
mMarkerPosition = 0;
mMarkerReached = false;
mNewPosition = 0;
mUpdatePeriod = 0;
mFlushed = false;
mFlags = flags;
AudioSystem::acquireAudioSessionId(mSessionId);
mRestoreStatus = NO_ERROR;
return NO_ERROR;
}
status_t AudioTrack::initCheck() const
{
return mStatus;
}
// -------------------------------------------------------------------------
uint32_t AudioTrack::latency() const
{
return mLatency;
}
int AudioTrack::streamType() const
{
return mStreamType;
}
int AudioTrack::format() const
{
return mFormat;
}
int AudioTrack::channelCount() const
{
return mChannelCount;
}
uint32_t AudioTrack::frameCount() const
{
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
return mCblk->frameCount;
}
int AudioTrack::frameSize() const
{
if (audio_is_linear_pcm(mFormat)) {
return channelCount()*audio_bytes_per_sample(mFormat);
} else {
return sizeof(uint8_t);
}
}
sp<IMemory>& AudioTrack::sharedBuffer()
{
return mSharedBuffer;
}
// -------------------------------------------------------------------------
void AudioTrack::start()
{
sp<AudioTrackThread> t = mAudioTrackThread;
status_t status = NO_ERROR;
LOGV("start %p", this);
if (t != 0) {
if (t->exitPending()) {
if (t->requestExitAndWait() == WOULD_BLOCK) {
LOGE("AudioTrack::start called from thread");
return;
}
}
t->mLock.lock();
}
AutoMutex lock(mLock);
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
// acquire a strong reference on the IMemory and IAudioTrack so that they cannot be destroyed
// while we are accessing the cblk
sp <IAudioTrack> audioTrack = mAudioTrack;
sp <IMemory> iMem = mCblkMemory;
audio_track_cblk_t* cblk = mCblk;
if (mActive == 0) {
mFlushed = false;
mActive = 1;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
mNewPosition = cblk->server + mUpdatePeriod;
cblk->lock.lock();
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
cblk->bufferTimeoutMs = MAX_STARTUP_TIMEOUT_MS;
cblk->waitTimeMs = 0;
android_atomic_and(~CBLK_DISABLED_ON, &cblk->flags);
if (t != 0) {
t->run("AudioTrackThread", ANDROID_PRIORITY_AUDIO);
} else {
setpriority(PRIO_PROCESS, 0, ANDROID_PRIORITY_AUDIO);
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
LOGV("start %p before lock cblk %p", this, mCblk);
if (!(cblk->flags & CBLK_INVALID_MSK)) {
cblk->lock.unlock();
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
status = mAudioTrack->start();
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
cblk->lock.lock();
if (status == DEAD_OBJECT) {
android_atomic_or(CBLK_INVALID_ON, &cblk->flags);
}
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
if (cblk->flags & CBLK_INVALID_MSK) {
status = restoreTrack_l(cblk, true);
}
cblk->lock.unlock();
if (status != NO_ERROR) {
LOGV("start() failed");
mActive = 0;
if (t != 0) {
t->requestExit();
} else {
setpriority(PRIO_PROCESS, 0, ANDROID_PRIORITY_NORMAL);
}
}
}
if (t != 0) {
t->mLock.unlock();
}
}
void AudioTrack::stop()
{
sp<AudioTrackThread> t = mAudioTrackThread;
LOGV("stop %p", this);
if (t != 0) {
t->mLock.lock();
}
AutoMutex lock(mLock);
if (mActive == 1) {
mActive = 0;
mCblk->cv.signal();
mAudioTrack->stop();
// Cancel loops (If we are in the middle of a loop, playback
// would not stop until loopCount reaches 0).
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
setLoop_l(0, 0, 0);
// the playback head position will reset to 0, so if a marker is set, we need
// to activate it again
mMarkerReached = false;
// Force flush if a shared buffer is used otherwise audioflinger
// will not stop before end of buffer is reached.
if (mSharedBuffer != 0) {
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
flush_l();
}
if (t != 0) {
t->requestExit();
} else {
setpriority(PRIO_PROCESS, 0, ANDROID_PRIORITY_NORMAL);
}
}
if (t != 0) {
t->mLock.unlock();
}
}
bool AudioTrack::stopped() const
{
return !mActive;
}
void AudioTrack::flush()
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
{
AutoMutex lock(mLock);
flush_l();
}
// must be called with mLock held
void AudioTrack::flush_l()
{
LOGV("flush");
// clear playback marker and periodic update counter
mMarkerPosition = 0;
mMarkerReached = false;
mUpdatePeriod = 0;
if (!mActive) {
mFlushed = true;
mAudioTrack->flush();
// Release AudioTrack callback thread in case it was waiting for new buffers
// in AudioTrack::obtainBuffer()
mCblk->cv.signal();
}
}
void AudioTrack::pause()
{
LOGV("pause");
AutoMutex lock(mLock);
if (mActive == 1) {
mActive = 0;
mAudioTrack->pause();
}
}
void AudioTrack::mute(bool e)
{
mAudioTrack->mute(e);
mMuted = e;
}
bool AudioTrack::muted() const
{
return mMuted;
}
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
status_t AudioTrack::setVolume(float left, float right)
{
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
if (left > 1.0f || right > 1.0f) {
return BAD_VALUE;
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
mVolume[LEFT] = left;
mVolume[RIGHT] = right;
// write must be atomic
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
mCblk->volumeLR = (uint32_t(uint16_t(right * 0x1000)) << 16) | uint16_t(left * 0x1000);
return NO_ERROR;
}
void AudioTrack::getVolume(float* left, float* right)
{
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
if (left != NULL) {
*left = mVolume[LEFT];
}
if (right != NULL) {
*right = mVolume[RIGHT];
}
}
status_t AudioTrack::setAuxEffectSendLevel(float level)
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
{
LOGV("setAuxEffectSendLevel(%f)", level);
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
if (level > 1.0f) {
return BAD_VALUE;
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
mSendLevel = level;
mCblk->sendLevel = uint16_t(level * 0x1000);
return NO_ERROR;
}
void AudioTrack::getAuxEffectSendLevel(float* level)
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
{
if (level != NULL) {
*level = mSendLevel;
}
}
status_t AudioTrack::setSampleRate(int rate)
{
int afSamplingRate;
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
if (mIsTimed) {
return INVALID_OPERATION;
}
if (AudioSystem::getOutputSamplingRate(&afSamplingRate, mStreamType) != NO_ERROR) {
return NO_INIT;
}
// Resampler implementation limits input sampling rate to 2 x output sampling rate.
if (rate <= 0 || rate > afSamplingRate*2 ) return BAD_VALUE;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
mCblk->sampleRate = rate;
return NO_ERROR;
}
uint32_t AudioTrack::getSampleRate()
{
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
if (mIsTimed) {
return INVALID_OPERATION;
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
return mCblk->sampleRate;
}
status_t AudioTrack::setLoop(uint32_t loopStart, uint32_t loopEnd, int loopCount)
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
{
AutoMutex lock(mLock);
return setLoop_l(loopStart, loopEnd, loopCount);
}
// must be called with mLock held
status_t AudioTrack::setLoop_l(uint32_t loopStart, uint32_t loopEnd, int loopCount)
{
audio_track_cblk_t* cblk = mCblk;
Mutex::Autolock _l(cblk->lock);
if (loopCount == 0) {
cblk->loopStart = UINT_MAX;
cblk->loopEnd = UINT_MAX;
cblk->loopCount = 0;
mLoopCount = 0;
return NO_ERROR;
}
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
if (mIsTimed) {
return INVALID_OPERATION;
}
if (loopStart >= loopEnd ||
loopEnd - loopStart > cblk->frameCount ||
cblk->server > loopStart) {
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
LOGE("setLoop invalid value: loopStart %d, loopEnd %d, loopCount %d, framecount %d, user %d", loopStart, loopEnd, loopCount, cblk->frameCount, cblk->user);
return BAD_VALUE;
}
if ((mSharedBuffer != 0) && (loopEnd > cblk->frameCount)) {
LOGE("setLoop invalid value: loop markers beyond data: loopStart %d, loopEnd %d, framecount %d",
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
loopStart, loopEnd, cblk->frameCount);
return BAD_VALUE;
}
cblk->loopStart = loopStart;
cblk->loopEnd = loopEnd;
cblk->loopCount = loopCount;
mLoopCount = loopCount;
return NO_ERROR;
}
status_t AudioTrack::getLoop(uint32_t *loopStart, uint32_t *loopEnd, int *loopCount)
{
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
if (loopStart != 0) {
*loopStart = mCblk->loopStart;
}
if (loopEnd != 0) {
*loopEnd = mCblk->loopEnd;
}
if (loopCount != 0) {
if (mCblk->loopCount < 0) {
*loopCount = -1;
} else {
*loopCount = mCblk->loopCount;
}
}
return NO_ERROR;
}
status_t AudioTrack::setMarkerPosition(uint32_t marker)
{
if (mCbf == 0) return INVALID_OPERATION;
mMarkerPosition = marker;
mMarkerReached = false;
return NO_ERROR;
}
status_t AudioTrack::getMarkerPosition(uint32_t *marker)
{
if (marker == 0) return BAD_VALUE;
*marker = mMarkerPosition;
return NO_ERROR;
}
status_t AudioTrack::setPositionUpdatePeriod(uint32_t updatePeriod)
{
if (mCbf == 0) return INVALID_OPERATION;
uint32_t curPosition;
getPosition(&curPosition);
mNewPosition = curPosition + updatePeriod;
mUpdatePeriod = updatePeriod;
return NO_ERROR;
}
status_t AudioTrack::getPositionUpdatePeriod(uint32_t *updatePeriod)
{
if (updatePeriod == 0) return BAD_VALUE;
*updatePeriod = mUpdatePeriod;
return NO_ERROR;
}
status_t AudioTrack::setPosition(uint32_t position)
{
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
if (mIsTimed) return INVALID_OPERATION;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
Mutex::Autolock _l(mCblk->lock);
if (!stopped()) return INVALID_OPERATION;
if (position > mCblk->user) return BAD_VALUE;
mCblk->server = position;
android_atomic_or(CBLK_FORCEREADY_ON, &mCblk->flags);
return NO_ERROR;
}
status_t AudioTrack::getPosition(uint32_t *position)
{
if (position == 0) return BAD_VALUE;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
*position = mFlushed ? 0 : mCblk->server;
return NO_ERROR;
}
status_t AudioTrack::reload()
{
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
if (!stopped()) return INVALID_OPERATION;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
flush_l();
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
mCblk->stepUser(mCblk->frameCount);
return NO_ERROR;
}
audio_io_handle_t AudioTrack::getOutput()
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
{
AutoMutex lock(mLock);
return getOutput_l();
}
// must be called with mLock held
audio_io_handle_t AudioTrack::getOutput_l()
{
return AudioSystem::getOutput((audio_stream_type_t)mStreamType,
mCblk->sampleRate, mFormat, mChannelMask, (audio_policy_output_flags_t)mFlags);
}
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
int AudioTrack::getSessionId()
{
return mSessionId;
}
status_t AudioTrack::attachAuxEffect(int effectId)
{
LOGV("attachAuxEffect(%d)", effectId);
status_t status = mAudioTrack->attachAuxEffect(effectId);
if (status == NO_ERROR) {
mAuxEffectId = effectId;
}
return status;
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
}
// -------------------------------------------------------------------------
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
// must be called with mLock held
status_t AudioTrack::createTrack_l(
int streamType,
uint32_t sampleRate,
uint32_t format,
uint32_t channelMask,
int frameCount,
uint32_t flags,
const sp<IMemory>& sharedBuffer,
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
audio_io_handle_t output,
bool enforceFrameCount)
{
status_t status;
const sp<IAudioFlinger>& audioFlinger = AudioSystem::get_audio_flinger();
if (audioFlinger == 0) {
LOGE("Could not get audioflinger");
return NO_INIT;
}
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
int afSampleRate;
if (AudioSystem::getOutputSamplingRate(&afSampleRate, streamType) != NO_ERROR) {
return NO_INIT;
}
int afFrameCount;
if (AudioSystem::getOutputFrameCount(&afFrameCount, streamType) != NO_ERROR) {
return NO_INIT;
}
uint32_t afLatency;
if (AudioSystem::getOutputLatency(&afLatency, streamType) != NO_ERROR) {
return NO_INIT;
}
mNotificationFramesAct = mNotificationFramesReq;
if (!audio_is_linear_pcm(format)) {
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
if (sharedBuffer != 0) {
frameCount = sharedBuffer->size();
}
} else {
// Ensure that buffer depth covers at least audio hardware latency
uint32_t minBufCount = afLatency / ((1000 * afFrameCount)/afSampleRate);
if (minBufCount < 2) minBufCount = 2;
int minFrameCount = (afFrameCount*sampleRate*minBufCount)/afSampleRate;
if (sharedBuffer == 0) {
if (frameCount == 0) {
frameCount = minFrameCount;
}
if (mNotificationFramesAct == 0) {
mNotificationFramesAct = frameCount/2;
}
// Make sure that application is notified with sufficient margin
// before underrun
if (mNotificationFramesAct > (uint32_t)frameCount/2) {
mNotificationFramesAct = frameCount/2;
}
if (frameCount < minFrameCount) {
LOGW_IF(enforceFrameCount, "Minimum buffer size corrected from %d to %d",
frameCount, minFrameCount);
frameCount = minFrameCount;
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
}
} else {
// Ensure that buffer alignment matches channelcount
int channelCount = popcount(channelMask);
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
if (((uint32_t)sharedBuffer->pointer() & (channelCount | 1)) != 0) {
LOGE("Invalid buffer alignement: address %p, channelCount %d", sharedBuffer->pointer(), channelCount);
return BAD_VALUE;
}
frameCount = sharedBuffer->size()/channelCount/sizeof(int16_t);
}
}
sp<IAudioTrack> track = audioFlinger->createTrack(getpid(),
streamType,
sampleRate,
format,
channelMask,
frameCount,
((uint16_t)flags) << 16,
sharedBuffer,
output,
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
mIsTimed,
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
&mSessionId,
&status);
if (track == 0) {
LOGE("AudioFlinger could not create track, status: %d", status);
return status;
}
sp<IMemory> cblk = track->getCblk();
if (cblk == 0) {
LOGE("Could not get control block");
return NO_INIT;
}
mAudioTrack.clear();
mAudioTrack = track;
mCblkMemory.clear();
mCblkMemory = cblk;
mCblk = static_cast<audio_track_cblk_t*>(cblk->pointer());
android_atomic_or(CBLK_DIRECTION_OUT, &mCblk->flags);
if (sharedBuffer == 0) {
mCblk->buffers = (char*)mCblk + sizeof(audio_track_cblk_t);
} else {
mCblk->buffers = sharedBuffer->pointer();
// Force buffer full condition as data is already present in shared memory
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
mCblk->stepUser(mCblk->frameCount);
}
Issue 2667801: [Audio Effect Framework] AudioFlinger, AudioMixer AudioTrack modifications. First drop of audio framework modifications for audio effects support. - AudioTrack/AudioRecord: Added support for auxiliary effects in AudioTrack Added support for audio sessions Fixed left right channel inversion in setVolume() - IAudioFlinger: Added interface methods for effect enumeraiton and instantiation Added support for audio sessions. - IAudioTrack: Added method to attach auxiliary effect. - AudioFlinger Created new classes to control effect engines in effect library and manage effect connections to tracks or output mix: EffectModule: wrapper object controlling the effect engine implementation in the effect library. There is one EffectModule per instance of an effect in a given audio session EffectChain: group of effects associated to one audio session. There is one EffectChain per audio session. EffectChain for session 0 is for output mix effects, other chains are attached to audio tracks with same session ID. Each chain contains a variable number of EffectModules EffectHandle: implements the IEffect interface. There is one EffectHandle object for each application controlling (or using) an effect module. THe EffectModule maintians a list of EffectHandles. Added support for effect modules and effect chains creation in PlaybackThread. modified mixer thread loop to allow track volume control by effect modules and call effect processing. -AudioMixer Each track now specifies its output buffer used by mixer for accumulation Modified mixer process functions to process tracks by groups of tracks with same buffer Modified track process functions to support accumulation to auxiliary channel Change-Id: I26d5f7c9e070a89bdd383e1a659f8b7ca150379c
2010-06-01 23:49:17 -07:00
mCblk->volumeLR = (uint32_t(uint16_t(mVolume[RIGHT] * 0x1000)) << 16) | uint16_t(mVolume[LEFT] * 0x1000);
mCblk->sendLevel = uint16_t(mSendLevel * 0x1000);
mAudioTrack->attachAuxEffect(mAuxEffectId);
mCblk->bufferTimeoutMs = MAX_STARTUP_TIMEOUT_MS;
mCblk->waitTimeMs = 0;
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
mRemainingFrames = mNotificationFramesAct;
mLatency = afLatency + (1000*mCblk->frameCount) / sampleRate;
return NO_ERROR;
}
status_t AudioTrack::obtainBuffer(Buffer* audioBuffer, int32_t waitCount)
{
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
int active;
status_t result = NO_ERROR;
audio_track_cblk_t* cblk = mCblk;
uint32_t framesReq = audioBuffer->frameCount;
uint32_t waitTimeMs = (waitCount < 0) ? cblk->bufferTimeoutMs : WAIT_PERIOD_MS;
audioBuffer->frameCount = 0;
audioBuffer->size = 0;
uint32_t framesAvail = cblk->framesAvailable();
cblk->lock.lock();
if (cblk->flags & CBLK_INVALID_MSK) {
goto create_new_track;
}
cblk->lock.unlock();
if (framesAvail == 0) {
cblk->lock.lock();
goto start_loop_here;
while (framesAvail == 0) {
active = mActive;
if (UNLIKELY(!active)) {
LOGV("Not active and NO_MORE_BUFFERS");
cblk->lock.unlock();
return NO_MORE_BUFFERS;
}
if (UNLIKELY(!waitCount)) {
cblk->lock.unlock();
return WOULD_BLOCK;
}
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
if (!(cblk->flags & CBLK_INVALID_MSK)) {
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
mLock.unlock();
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
result = cblk->cv.waitRelative(cblk->lock, milliseconds(waitTimeMs));
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
cblk->lock.unlock();
mLock.lock();
if (mActive == 0) {
return status_t(STOPPED);
}
cblk->lock.lock();
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
if (cblk->flags & CBLK_INVALID_MSK) {
goto create_new_track;
}
if (__builtin_expect(result!=NO_ERROR, false)) {
cblk->waitTimeMs += waitTimeMs;
if (cblk->waitTimeMs >= cblk->bufferTimeoutMs) {
// timing out when a loop has been set and we have already written upto loop end
// is a normal condition: no need to wake AudioFlinger up.
if (cblk->user < cblk->loopEnd) {
LOGW( "obtainBuffer timed out (is the CPU pegged?) %p "
"user=%08x, server=%08x", this, cblk->user, cblk->server);
//unlock cblk mutex before calling mAudioTrack->start() (see issue #1617140)
cblk->lock.unlock();
result = mAudioTrack->start();
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
cblk->lock.lock();
if (result == DEAD_OBJECT) {
android_atomic_or(CBLK_INVALID_ON, &cblk->flags);
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
create_new_track:
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
result = restoreTrack_l(cblk, false);
}
if (result != NO_ERROR) {
LOGW("obtainBuffer create Track error %d", result);
cblk->lock.unlock();
return result;
}
}
cblk->waitTimeMs = 0;
}
if (--waitCount == 0) {
cblk->lock.unlock();
return TIMED_OUT;
}
}
// read the server count again
start_loop_here:
framesAvail = cblk->framesAvailable_l();
}
cblk->lock.unlock();
}
// restart track if it was disabled by audioflinger due to previous underrun
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
if (mActive && (cblk->flags & CBLK_DISABLED_MSK)) {
android_atomic_and(~CBLK_DISABLED_ON, &cblk->flags);
LOGW("obtainBuffer() track %p disabled, restarting", this);
mAudioTrack->start();
}
cblk->waitTimeMs = 0;
if (framesReq > framesAvail) {
framesReq = framesAvail;
}
uint32_t u = cblk->user;
uint32_t bufferEnd = cblk->userBase + cblk->frameCount;
if (u + framesReq > bufferEnd) {
framesReq = bufferEnd - u;
}
audioBuffer->flags = mMuted ? Buffer::MUTE : 0;
audioBuffer->channelCount = mChannelCount;
audioBuffer->frameCount = framesReq;
audioBuffer->size = framesReq * cblk->frameSize;
if (audio_is_linear_pcm(mFormat)) {
audioBuffer->format = AUDIO_FORMAT_PCM_16_BIT;
} else {
audioBuffer->format = mFormat;
}
audioBuffer->raw = (int8_t *)cblk->buffer(u);
active = mActive;
return active ? status_t(NO_ERROR) : status_t(STOPPED);
}
void AudioTrack::releaseBuffer(Buffer* audioBuffer)
{
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
AutoMutex lock(mLock);
mCblk->stepUser(audioBuffer->frameCount);
}
// -------------------------------------------------------------------------
ssize_t AudioTrack::write(const void* buffer, size_t userSize)
{
if (mSharedBuffer != 0) return INVALID_OPERATION;
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
if (mIsTimed) return INVALID_OPERATION;
if (ssize_t(userSize) < 0) {
// sanity-check. user is most-likely passing an error code.
LOGE("AudioTrack::write(buffer=%p, size=%u (%d)",
buffer, userSize, userSize);
return BAD_VALUE;
}
LOGV("write %p: %d bytes, mActive=%d", this, userSize, mActive);
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
// acquire a strong reference on the IMemory and IAudioTrack so that they cannot be destroyed
// while we are accessing the cblk
mLock.lock();
sp <IAudioTrack> audioTrack = mAudioTrack;
sp <IMemory> iMem = mCblkMemory;
mLock.unlock();
ssize_t written = 0;
const int8_t *src = (const int8_t *)buffer;
Buffer audioBuffer;
size_t frameSz = (size_t)frameSize();
do {
audioBuffer.frameCount = userSize/frameSz;
// Calling obtainBuffer() with a negative wait count causes
// an (almost) infinite wait time.
status_t err = obtainBuffer(&audioBuffer, -1);
if (err < 0) {
// out of buffers, return #bytes written
if (err == status_t(NO_MORE_BUFFERS))
break;
return ssize_t(err);
}
size_t toWrite;
if (mFormat == AUDIO_FORMAT_PCM_8_BIT && !(mFlags & AUDIO_POLICY_OUTPUT_FLAG_DIRECT)) {
// Divide capacity by 2 to take expansion into account
toWrite = audioBuffer.size>>1;
// 8 to 16 bit conversion
int count = toWrite;
int16_t *dst = (int16_t *)(audioBuffer.i8);
while(count--) {
*dst++ = (int16_t)(*src++^0x80) << 8;
}
} else {
toWrite = audioBuffer.size;
memcpy(audioBuffer.i8, src, toWrite);
src += toWrite;
}
userSize -= toWrite;
written += toWrite;
releaseBuffer(&audioBuffer);
} while (userSize >= frameSz);
return written;
}
// -------------------------------------------------------------------------
Media framework changes for Tungsten. 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>
2011-08-15 13:28:26 -07:00
TimedAudioTrack::TimedAudioTrack() {
mIsTimed = true;
}
status_t TimedAudioTrack::allocateTimedBuffer(size_t size, sp<IMemory>* buffer)
{
return mAudioTrack->allocateTimedBuffer(size, buffer);
}
status_t TimedAudioTrack::queueTimedBuffer(const sp<IMemory>& buffer,
int64_t pts)
{
// restart track if it was disabled by audioflinger due to previous underrun
if (mActive && (mCblk->flags & CBLK_DISABLED_MSK)) {
mCblk->flags &= ~CBLK_DISABLED_ON;
LOGW("queueTimedBuffer() track %p disabled, restarting", this);
mAudioTrack->start();
}
return mAudioTrack->queueTimedBuffer(buffer, pts);
}
status_t TimedAudioTrack::setMediaTimeTransform(const LinearTransform& xform,
TargetTimeline target)
{
return mAudioTrack->setMediaTimeTransform(xform, target);
}
// -------------------------------------------------------------------------
bool AudioTrack::processAudioBuffer(const sp<AudioTrackThread>& thread)
{
Buffer audioBuffer;
uint32_t frames;
size_t writtenSize;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
mLock.lock();
// acquire a strong reference on the IMemory and IAudioTrack so that they cannot be destroyed
// while we are accessing the cblk
sp <IAudioTrack> audioTrack = mAudioTrack;
sp <IMemory> iMem = mCblkMemory;
audio_track_cblk_t* cblk = mCblk;
mLock.unlock();
// Manage underrun callback
if (mActive && (cblk->framesAvailable() == cblk->frameCount)) {
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
LOGV("Underrun user: %x, server: %x, flags %04x", cblk->user, cblk->server, cblk->flags);
if (!(android_atomic_or(CBLK_UNDERRUN_ON, &cblk->flags) & CBLK_UNDERRUN_MSK)) {
mCbf(EVENT_UNDERRUN, mUserData, 0);
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
if (cblk->server == cblk->frameCount) {
mCbf(EVENT_BUFFER_END, mUserData, 0);
}
if (mSharedBuffer != 0) return false;
}
}
// Manage loop end callback
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
while (mLoopCount > cblk->loopCount) {
int loopCount = -1;
mLoopCount--;
if (mLoopCount >= 0) loopCount = mLoopCount;
mCbf(EVENT_LOOP_END, mUserData, (void *)&loopCount);
}
// Manage marker callback
if (!mMarkerReached && (mMarkerPosition > 0)) {
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
if (cblk->server >= mMarkerPosition) {
mCbf(EVENT_MARKER, mUserData, (void *)&mMarkerPosition);
mMarkerReached = true;
}
}
// Manage new position callback
if (mUpdatePeriod > 0) {
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
while (cblk->server >= mNewPosition) {
mCbf(EVENT_NEW_POS, mUserData, (void *)&mNewPosition);
mNewPosition += mUpdatePeriod;
}
}
// If Shared buffer is used, no data is requested from client.
if (mSharedBuffer != 0) {
frames = 0;
} else {
frames = mRemainingFrames;
}
int32_t waitCount = -1;
if (mUpdatePeriod || (!mMarkerReached && mMarkerPosition) || mLoopCount) {
waitCount = 1;
}
do {
audioBuffer.frameCount = frames;
// Calling obtainBuffer() with a wait count of 1
// limits wait time to WAIT_PERIOD_MS. This prevents from being
// stuck here not being able to handle timed events (position, markers, loops).
status_t err = obtainBuffer(&audioBuffer, waitCount);
if (err < NO_ERROR) {
if (err != TIMED_OUT) {
LOGE_IF(err != status_t(NO_MORE_BUFFERS), "Error obtaining an audio buffer, giving up.");
return false;
}
break;
}
if (err == status_t(STOPPED)) return false;
// Divide buffer size by 2 to take into account the expansion
// due to 8 to 16 bit conversion: the callback must fill only half
// of the destination buffer
if (mFormat == AUDIO_FORMAT_PCM_8_BIT && !(mFlags & AUDIO_POLICY_OUTPUT_FLAG_DIRECT)) {
audioBuffer.size >>= 1;
}
size_t reqSize = audioBuffer.size;
mCbf(EVENT_MORE_DATA, mUserData, &audioBuffer);
writtenSize = audioBuffer.size;
// Sanity check on returned size
if (ssize_t(writtenSize) <= 0) {
// The callback is done filling buffers
// Keep this thread going to handle timed events and
// still try to get more data in intervals of WAIT_PERIOD_MS
// but don't just loop and block the CPU, so wait
usleep(WAIT_PERIOD_MS*1000);
break;
}
if (writtenSize > reqSize) writtenSize = reqSize;
if (mFormat == AUDIO_FORMAT_PCM_8_BIT && !(mFlags & AUDIO_POLICY_OUTPUT_FLAG_DIRECT)) {
// 8 to 16 bit conversion
const int8_t *src = audioBuffer.i8 + writtenSize-1;
int count = writtenSize;
int16_t *dst = audioBuffer.i16 + writtenSize-1;
while(count--) {
*dst-- = (int16_t)(*src--^0x80) << 8;
}
writtenSize <<= 1;
}
audioBuffer.size = writtenSize;
// NOTE: mCblk->frameSize is not equal to AudioTrack::frameSize() for
// 8 bit PCM data: in this case, mCblk->frameSize is based on a sampel size of
// 16 bit.
audioBuffer.frameCount = writtenSize/mCblk->frameSize;
frames -= audioBuffer.frameCount;
releaseBuffer(&audioBuffer);
}
while (frames);
if (frames == 0) {
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
mRemainingFrames = mNotificationFramesAct;
} else {
mRemainingFrames = frames;
}
return true;
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
// must be called with mLock and cblk.lock held. Callers must also hold strong references on
// the IAudioTrack and IMemory in case they are recreated here.
// If the IAudioTrack is successfully restored, the cblk pointer is updated
status_t AudioTrack::restoreTrack_l(audio_track_cblk_t*& cblk, bool fromStart)
{
status_t result;
if (!(android_atomic_or(CBLK_RESTORING_ON, &cblk->flags) & CBLK_RESTORING_MSK)) {
LOGW("dead IAudioTrack, creating a new one from %s TID %d",
fromStart ? "start()" : "obtainBuffer()", gettid());
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
// signal old cblk condition so that other threads waiting for available buffers stop
// waiting now
cblk->cv.broadcast();
cblk->lock.unlock();
// refresh the audio configuration cache in this process to make sure we get new
// output parameters in getOutput_l() and createTrack_l()
AudioSystem::clearAudioConfigCache();
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
// if the new IAudioTrack is created, createTrack_l() will modify the
// following member variables: mAudioTrack, mCblkMemory and mCblk.
// It will also delete the strong references on previous IAudioTrack and IMemory
result = createTrack_l(mStreamType,
cblk->sampleRate,
mFormat,
mChannelMask,
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
mFrameCount,
mFlags,
mSharedBuffer,
getOutput_l(),
false);
if (result == NO_ERROR) {
uint32_t user = cblk->user;
uint32_t server = cblk->server;
// restore write index and set other indexes to reflect empty buffer status
mCblk->user = user;
mCblk->server = user;
mCblk->userBase = user;
mCblk->serverBase = user;
// restore loop: this is not guaranteed to succeed if new frame count is not
// compatible with loop length
setLoop_l(cblk->loopStart, cblk->loopEnd, cblk->loopCount);
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
if (!fromStart) {
mCblk->bufferTimeoutMs = MAX_RUN_TIMEOUT_MS;
// Make sure that a client relying on callback events indicating underrun or
// the actual amount of audio frames played (e.g SoundPool) receives them.
if (mSharedBuffer == 0) {
uint32_t frames = 0;
if (user > server) {
frames = ((user - server) > mCblk->frameCount) ?
mCblk->frameCount : (user - server);
memset(mCblk->buffers, 0, frames * mCblk->frameSize);
}
// restart playback even if buffer is not completely filled.
android_atomic_or(CBLK_FORCEREADY_ON, &mCblk->flags);
// stepUser() clears CBLK_UNDERRUN_ON flag enabling underrun callbacks to
// the client
mCblk->stepUser(frames);
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
}
if (mActive) {
result = mAudioTrack->start();
LOGW_IF(result != NO_ERROR, "restoreTrack_l() start() failed status %d", result);
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
if (fromStart && result == NO_ERROR) {
mNewPosition = mCblk->server + mUpdatePeriod;
}
}
if (result != NO_ERROR) {
android_atomic_and(~CBLK_RESTORING_ON, &cblk->flags);
LOGW_IF(result != NO_ERROR, "restoreTrack_l() failed status %d", result);
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
}
mRestoreStatus = result;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
// signal old cblk condition for other threads waiting for restore completion
android_atomic_or(CBLK_RESTORED_ON, &cblk->flags);
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
cblk->cv.broadcast();
} else {
if (!(cblk->flags & CBLK_RESTORED_MSK)) {
LOGW("dead IAudioTrack, waiting for a new one TID %d", gettid());
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
mLock.unlock();
result = cblk->cv.waitRelative(cblk->lock, milliseconds(RESTORE_TIMEOUT_MS));
if (result == NO_ERROR) {
result = mRestoreStatus;
}
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
cblk->lock.unlock();
mLock.lock();
} else {
LOGW("dead IAudioTrack, already restored TID %d", gettid());
result = mRestoreStatus;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
cblk->lock.unlock();
}
}
LOGV("restoreTrack_l() status %d mActive %d cblk %p, old cblk %p flags %08x old flags %08x",
result, mActive, mCblk, cblk, mCblk->flags, cblk->flags);
if (result == NO_ERROR) {
// from now on we switch to the newly created cblk
cblk = mCblk;
}
cblk->lock.lock();
LOGW_IF(result != NO_ERROR, "restoreTrack_l() error %d TID %d", result, gettid());
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
return result;
}
status_t AudioTrack::dump(int fd, const Vector<String16>& args) const
{
const size_t SIZE = 256;
char buffer[SIZE];
String8 result;
result.append(" AudioTrack::dump\n");
snprintf(buffer, 255, " stream type(%d), left - right volume(%f, %f)\n", mStreamType, mVolume[0], mVolume[1]);
result.append(buffer);
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
snprintf(buffer, 255, " format(%d), channel count(%d), frame count(%d)\n", mFormat, mChannelCount, mCblk->frameCount);
result.append(buffer);
snprintf(buffer, 255, " sample rate(%d), status(%d), muted(%d)\n", (mCblk == 0) ? 0 : mCblk->sampleRate, mStatus, mMuted);
result.append(buffer);
snprintf(buffer, 255, " active(%d), latency (%d)\n", mActive, mLatency);
result.append(buffer);
::write(fd, result.string(), result.size());
return NO_ERROR;
}
// =========================================================================
AudioTrack::AudioTrackThread::AudioTrackThread(AudioTrack& receiver, bool bCanCallJava)
: Thread(bCanCallJava), mReceiver(receiver)
{
}
bool AudioTrack::AudioTrackThread::threadLoop()
{
return mReceiver.processAudioBuffer(this);
}
status_t AudioTrack::AudioTrackThread::readyToRun()
{
return NO_ERROR;
}
void AudioTrack::AudioTrackThread::onFirstRef()
{
}
// =========================================================================
audio_track_cblk_t::audio_track_cblk_t()
: lock(Mutex::SHARED), cv(Condition::SHARED), user(0), server(0),
userBase(0), serverBase(0), buffers(0), frameCount(0),
loopStart(UINT_MAX), loopEnd(UINT_MAX), loopCount(0), volumeLR(0),
sendLevel(0), flags(0)
{
}
uint32_t audio_track_cblk_t::stepUser(uint32_t frameCount)
{
uint32_t u = this->user;
u += frameCount;
// Ensure that user is never ahead of server for AudioRecord
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
if (flags & CBLK_DIRECTION_MSK) {
// If stepServer() has been called once, switch to normal obtainBuffer() timeout period
if (bufferTimeoutMs == MAX_STARTUP_TIMEOUT_MS-1) {
bufferTimeoutMs = MAX_RUN_TIMEOUT_MS;
}
} else if (u > this->server) {
LOGW("stepServer occured after track reset");
u = this->server;
}
if (u >= userBase + this->frameCount) {
userBase += this->frameCount;
}
this->user = u;
// Clear flow control error condition as new data has been written/read to/from buffer.
if (flags & CBLK_UNDERRUN_MSK) {
android_atomic_and(~CBLK_UNDERRUN_MSK, &flags);
}
return u;
}
bool audio_track_cblk_t::stepServer(uint32_t frameCount)
{
if (!tryLock()) {
LOGW("stepServer() could not lock cblk");
return false;
}
uint32_t s = this->server;
s += frameCount;
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
if (flags & CBLK_DIRECTION_MSK) {
// Mark that we have read the first buffer so that next time stepUser() is called
// we switch to normal obtainBuffer() timeout period
if (bufferTimeoutMs == MAX_STARTUP_TIMEOUT_MS) {
bufferTimeoutMs = MAX_STARTUP_TIMEOUT_MS - 1;
}
// It is possible that we receive a flush()
// while the mixer is processing a block: in this case,
// stepServer() is called After the flush() has reset u & s and
// we have s > u
if (s > this->user) {
LOGW("stepServer occured after track reset");
s = this->user;
}
}
if (s >= loopEnd) {
LOGW_IF(s > loopEnd, "stepServer: s %u > loopEnd %u", s, loopEnd);
s = loopStart;
if (--loopCount == 0) {
loopEnd = UINT_MAX;
loopStart = UINT_MAX;
}
}
if (s >= serverBase + this->frameCount) {
serverBase += this->frameCount;
}
this->server = s;
Fix issue 3439872: video chat and bluetooth SCO This change fixes the stability problems experienced when using a bluetooth headset supporting both A2DP and SCO. Problems occur when starting the video chat at which time the A2DP output is being stopped to start SCO. At that time, active AudioTracks are invalidated by AudioFlinger so that a new AudioTrack binder interface can be recreated by the client process on the new mixer thread with correct parameters. The problem was that the process to restore the binder interface was not protected against concurrent requests which caused 2 binder interfaces to be created sometimes. This could lead to permanent client deadlock if one of the client threads was waiting for a condition of the first created binder interface while the second one was created (as the AudioFlinger would only signal conditions on the last one created). This concurrent request situation is more likely to happen when a client uses the JAVA AudioTrack as the JNI implementation uses simultaneously the native AudioTrack callback and write push mechanisms. By doing so, the code that checks if the binder interface should be restored (in obtainBuffer()) is much more likely to be called concurrently from two different threads. The fix consists in protecting the critical binder interface restore phase with a flag in the AudioTrack control block. The first thread acting upon the binder interface restore request will raise the flag and the second thread will just wait for a condition to be signaled when the restore process is complete. Also protected all accesses to the AudioTrack control block by a mutex to prevent access while the track is being destroyed and restored. If a mutex cannot be held (e.g because we call a callback function), acquire a strong reference on the IAudioTrack to prevent its destruction while the cblk is being accessed. Modified AudioTrack JNI to use GetByteArrayElements() instead of GetPrimitiveArrayCritical() when writing audio buffers. Entering a critical section would cause the JNI to abort if a mediaserver crash occurs during a write due to the AudioSystem callback being called during the critical section when media server process restarts. Anyway with current JNI implementation, either versions do not copy data most of the times and the criticial version does not guaranty no data copy. The same modifications have been made to AudioRecord. Change-Id: Idc5aa711a04c3eee180cdd03f44fe17f3c4dcb52
2011-03-07 14:52:59 -08:00
if (!(flags & CBLK_INVALID_MSK)) {
cv.signal();
}
lock.unlock();
return true;
}
void* audio_track_cblk_t::buffer(uint32_t offset) const
{
return (int8_t *)this->buffers + (offset - userBase) * this->frameSize;
}
uint32_t audio_track_cblk_t::framesAvailable()
{
Mutex::Autolock _l(lock);
return framesAvailable_l();
}
uint32_t audio_track_cblk_t::framesAvailable_l()
{
uint32_t u = this->user;
uint32_t s = this->server;
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
if (flags & CBLK_DIRECTION_MSK) {
uint32_t limit = (s < loopStart) ? s : loopStart;
return limit + frameCount - u;
} else {
return frameCount + u - s;
}
}
uint32_t audio_track_cblk_t::framesReady()
{
uint32_t u = this->user;
uint32_t s = this->server;
Fix issue 2553359: Pandora does not work well with Passion deskdock / Cardock. The problem is due to a too big difference between the buffer size used at the hardware interface and at the A2DP interface. When no resampling occurs we don't notice problems but the timing is very tight. As soon as resampling is activated, the AudioTrack underruns. This is because the AudioTrack buffers are not resized when moving the AudioTrack from hardware to A2DP output. The AudioTrack buffers are calculated based on a hardware output buffer size of 3072 bytes. Which is much less than the A2DP output buffer size (10240). The solution consists in creating new tracks with new buffers in AudioFlinger when the A2DP output is opened instead of just transfering active tracks from hardware output mixer thread to the new A2DP output mixer thread. To avoid synchronization issues between mixer threads and client processes, this is done by invalidating tracks by setting a flag in their control block and having AudioTrack release the handle on this track (IAudioTrack) and create a new IAudioTrack when this flag is detected next time obtainBuffer() or start() is executed. AudioFlinger modifications: - invalidate the tracks when setStreamOutput() is called - make sure that notifications of output opening/closing and change of stream type to output mapping are sent synchronously to client process. This is necessary so that AudioSystem has the new stream to output mapping when the AudioTrack detects the invalidate flag in the client process. Previously their were sent when the corresponding thread loop was executed. AudioTrack modifications: - move frame count calculation and verification from set() to createTrack() so that is is updated every time a new IAudioTrack is created. - detect track invalidate flag in obtainBuffer() and start() and create a new IAudioTrack. AudioTrackShared modifications - group all flags (out, flowControlFlag, forceReady...) into a single bit filed to save space. Change-Id: I9ac26b6192230627d35084e1449640caaf7d56ee
2010-05-14 03:26:45 -07:00
if (flags & CBLK_DIRECTION_MSK) {
if (u < loopEnd) {
return u - s;
} else {
// do not block on mutex shared with client on AudioFlinger side
if (!tryLock()) {
LOGW("framesReady() could not lock cblk");
return 0;
}
uint32_t frames = UINT_MAX;
if (loopCount >= 0) {
frames = (loopEnd - loopStart)*loopCount + u - s;
}
lock.unlock();
return frames;
}
} else {
return s - u;
}
}
bool audio_track_cblk_t::tryLock()
{
// the code below simulates lock-with-timeout
// we MUST do this to protect the AudioFlinger server
// as this lock is shared with the client.
status_t err;
err = lock.tryLock();
if (err == -EBUSY) { // just wait a bit
usleep(1000);
err = lock.tryLock();
}
if (err != NO_ERROR) {
// probably, the client just died.
return false;
}
return true;
}
// -------------------------------------------------------------------------
}; // namespace android