audio: Implement compressed offload simulation
Add 'StreamOffloadStub' to 'ModulePrimary' which simulates
offloaded playback ('COMPRESS_OFFLOAD') and can send asynchronous
callbacks. The simulation only keeps track of the playback
time, it does not actually decode compressed audio. This
approach was chosen to avoid linking against any decoder
implementation.
For the same reason, the only supported format is 'APE' ("Monkey
audio") which was chosen due to the simplicity of its header
structure.
As 'StreamOffloadStub' provides a proper implementation of
'EARLY_NOTIFY' drain callbacks, there is no more need to use
'aosp.forceDrainToDraining' property in VTS, and support for
it has been removed.
Bug: 373872271
Bug: 384431822
Test: atest CtsMediaAudioTestCases
Test: atest VtsHalAudioCoreTargetTest
Merged-In: I74ae867227c768d4afe9314ed93168da58362131
Change-Id: I74ae867227c768d4afe9314ed93168da58362131
diff --git a/audio/aidl/default/ModulePrimary.cpp b/audio/aidl/default/ModulePrimary.cpp
index 3da6d48..2a1dba9 100644
--- a/audio/aidl/default/ModulePrimary.cpp
+++ b/audio/aidl/default/ModulePrimary.cpp
@@ -21,12 +21,16 @@
#include <android-base/logging.h>
#include "core-impl/ModulePrimary.h"
+#include "core-impl/StreamOffloadStub.h"
#include "core-impl/StreamPrimary.h"
#include "core-impl/Telephony.h"
+using aidl::android::hardware::audio::common::areAllBitPositionFlagsSet;
using aidl::android::hardware::audio::common::SinkMetadata;
using aidl::android::hardware::audio::common::SourceMetadata;
+using aidl::android::media::audio::common::AudioIoFlags;
using aidl::android::media::audio::common::AudioOffloadInfo;
+using aidl::android::media::audio::common::AudioOutputFlags;
using aidl::android::media::audio::common::AudioPort;
using aidl::android::media::audio::common::AudioPortConfig;
using aidl::android::media::audio::common::MicrophoneInfo;
@@ -43,6 +47,17 @@
return ndk::ScopedAStatus::ok();
}
+ndk::ScopedAStatus ModulePrimary::calculateBufferSizeFrames(
+ const ::aidl::android::media::audio::common::AudioFormatDescription& format,
+ int32_t latencyMs, int32_t sampleRateHz, int32_t* bufferSizeFrames) {
+ if (format.type != ::aidl::android::media::audio::common::AudioFormatType::PCM &&
+ StreamOffloadStub::getSupportedEncodings().count(format.encoding)) {
+ *bufferSizeFrames = sampleRateHz / 2; // 1/2 of a second.
+ return ndk::ScopedAStatus::ok();
+ }
+ return Module::calculateBufferSizeFrames(format, latencyMs, sampleRateHz, bufferSizeFrames);
+}
+
ndk::ScopedAStatus ModulePrimary::createInputStream(StreamContext&& context,
const SinkMetadata& sinkMetadata,
const std::vector<MicrophoneInfo>& microphones,
@@ -54,8 +69,18 @@
ndk::ScopedAStatus ModulePrimary::createOutputStream(
StreamContext&& context, const SourceMetadata& sourceMetadata,
const std::optional<AudioOffloadInfo>& offloadInfo, std::shared_ptr<StreamOut>* result) {
- return createStreamInstance<StreamOutPrimary>(result, std::move(context), sourceMetadata,
- offloadInfo);
+ if (!areAllBitPositionFlagsSet(
+ context.getFlags().get<AudioIoFlags::output>(),
+ {AudioOutputFlags::COMPRESS_OFFLOAD, AudioOutputFlags::NON_BLOCKING})) {
+ return createStreamInstance<StreamOutPrimary>(result, std::move(context), sourceMetadata,
+ offloadInfo);
+ } else {
+ // "Stub" is used because there is no actual decoder. The stream just
+ // extracts the clip duration from the media file header and simulates
+ // playback over time.
+ return createStreamInstance<StreamOutOffloadStub>(result, std::move(context),
+ sourceMetadata, offloadInfo);
+ }
}
int32_t ModulePrimary::getNominalLatencyMs(const AudioPortConfig&) {