/* * Copyright 2013 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. */ #ifndef ANDROID_SF_BUFFERQUEUEINTERPOSER_H #define ANDROID_SF_BUFFERQUEUEINTERPOSER_H #include #include #include // --------------------------------------------------------------------------- namespace android { // --------------------------------------------------------------------------- // BufferQueueInterposers introduce an extra stage between a buffer producer // (the source) and a buffer consumer (the sink), which communicate via the // IGraphicBufferProducer interface. It is designed to be as transparent as // possible to both endpoints, so that they can work the same whether an // interposer is present or not. // // When the interpose is present, the source queues buffers to the // IGraphicBufferProducer implemented by BufferQueueInterposer. A client of // the BufferQueueInterposer can acquire each buffer in turn and read or // modify it, releasing the buffer when finished. When the buffer is released, // the BufferQueueInterposer queues it to the original IGraphicBufferProducer // interface representing the sink. // // A BufferQueueInterposer can be used to do additional rendering to a buffer // before it is consumed -- essentially pipelining two producers. As an // example, SurfaceFlinger uses this to implement mixed GLES and HWC // compositing to the same buffer for virtual displays. If it used two separate // buffer queues, then in GLES-only or mixed GLES+HWC compositing, the HWC // would have to copy the GLES output buffer to the HWC output buffer, using // more bandwidth than having HWC do additional composition "in place" on the // GLES output buffer. // // The goal for this class is to be usable in a variety of situations and be // part of libgui. But both the interface and implementation need some // iteration before then, so for now it should only be used by // VirtualDisplaySurface, which is why it's currently in SurfaceFlinger. // // Some of the problems that still need to be solved are: // // - Refactor the interposer interface along with BufferQueue and ConsumerBase, // so that there is a common interface for the consumer end of a queue. The // existing interfaces have some problems when the implementation isn't the // final consumer. // // - The client of the interposer may need one or more buffers in addition to // those used by the source and sink. IGraphicBufferProducer will probably // need to change to allow the producer to specify how many buffers it needs // to dequeue at a time, and then the interposer can add its requirements to // those of the source. // // - Abandoning, disconnecting, and connecting need to pass through somehow. // There needs to be a way to tell the interposer client to release its // buffer immediately so it can be queued/released, e.g. when the source // calls disconnect(). // // - Right now the source->BQI queue is synchronous even if the BQI->sink queue // is asynchronous. Need to figure out how asynchronous should behave and // implement that. class BufferQueueInterposer : public BnGraphicBufferProducer { public: BufferQueueInterposer(const sp& sink, const String8& name); // // IGraphicBufferProducer interface // virtual status_t requestBuffer(int slot, sp* outBuf); virtual status_t setBufferCount(int bufferCount); virtual status_t dequeueBuffer(int* slot, sp* fence, uint32_t w, uint32_t h, uint32_t format, uint32_t usage); virtual status_t queueBuffer(int slot, const QueueBufferInput& input, QueueBufferOutput* output); virtual void cancelBuffer(int slot, const sp& fence); virtual int query(int what, int* value); virtual status_t setSynchronousMode(bool enabled); virtual status_t connect(int api, QueueBufferOutput* output); virtual status_t disconnect(int api); // // Interposer interface // enum { NO_BUFFER_AVAILABLE = 2, // matches BufferQueue BUFFER_NOT_ACQUIRED, BUFFER_ALREADY_ACQUIRED, }; // Acquire the oldest queued buffer. If no buffers are pending, returns // NO_BUFFER_AVAILABLE. If a buffer is currently acquired, returns // BUFFER_ALREADY_ACQUIRED. status_t acquireBuffer(sp* buf, sp* fence); // Release the currently acquired buffer, queueing it to the sink. If the // current buffer hasn't been acquired, returns BUFFER_NOT_ACQUIRED. status_t releaseBuffer(const sp& fence); // pullEmptyBuffer dequeues a buffer from the sink, then immediately // queues it to the interposer. This makes a buffer available for the // client to acquire even if the source hasn't queued one. status_t pullEmptyBuffer(); private: struct QueuedBuffer { QueuedBuffer(): slot(-1) {} QueuedBuffer(int slot, const QueueBufferInput& qbi): slot(slot) { qbi.deflate(×tamp, &crop, &scalingMode, &transform, &fence); } int slot; int64_t timestamp; Rect crop; int scalingMode; uint32_t transform; sp fence; }; virtual ~BufferQueueInterposer(); status_t flushQueuedBuffersLocked(); const sp mSink; String8 mName; Mutex mMutex; Vector > mBuffers; Vector mQueue; bool mAcquired; QueueBufferOutput mQueueBufferOutput; }; // --------------------------------------------------------------------------- } // namespace android // --------------------------------------------------------------------------- #endif // ANDROID_SF_BUFFERQUEUEINTERPOSER_H