drm_hwcomposer: Make flattening thread-based instead of vsync-based
Using vsync means consume some CPU time every frame for
IRQ -> Kernel_Thread -> UserSpace_listener thread transitions.
Framework tries to reduce vsync usage to the minimum, by using timelines
instead.
New flattening controller thread wakes-up only once per second.
This commit also removes flattening info from dumpsys. Practice shows
that it is almost useless. Instead debugging can be done using ALOGV
dumps.
Signed-off-by: Roman Stratiienko <r.stratiienko@gmail.com>
diff --git a/compositor/FlatteningController.cpp b/compositor/FlatteningController.cpp
new file mode 100644
index 0000000..257f8a0
--- /dev/null
+++ b/compositor/FlatteningController.cpp
@@ -0,0 +1,98 @@
+/*
+ * Copyright (C) 2023 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.
+ */
+
+/*
+ * Usually, display controllers do not use intermediate buffer for composition
+ * results. Instead, they scan-out directly from the input buffers, composing
+ * the planes on the fly every VSYNC.
+ *
+ * Flattening is a technique that reduces memory bandwidth and power consumption
+ * by converting non-updating multi-plane composition into a single-plane.
+ * Additionally, flattening also makes more shared planes available for use by
+ * other CRTCs.
+ *
+ * If the client is not updating layers for 1 second, FlatCon triggers a
+ * callback to refresh the screen. The compositor should mark all layers to be
+ * composed by the client into a single framebuffer using GPU.
+ */
+
+#define LOG_TAG "hwc-flatcon"
+
+#include "FlatteningController.h"
+
+#include "utils/log.h"
+
+namespace android {
+
+auto FlatteningController::CreateInstance(FlatConCallbacks &cbks)
+ -> std::shared_ptr<FlatteningController> {
+ auto fc = std::shared_ptr<FlatteningController>(new FlatteningController());
+
+ fc->cbks_ = cbks;
+
+ std::thread(&FlatteningController::ThreadFn, fc).detach();
+
+ return fc;
+}
+
+/* Compositor should call this every frame */
+bool FlatteningController::NewFrame() {
+ bool wake_it = false;
+ auto lock = std::lock_guard<std::mutex>(mutex_);
+
+ if (flatten_next_frame_) {
+ flatten_next_frame_ = false;
+ return true;
+ }
+
+ sleep_until_ = std::chrono::system_clock::now() + kTimeout;
+ if (disabled_) {
+ wake_it = true;
+ disabled_ = false;
+ }
+
+ if (wake_it)
+ cv_.notify_all();
+
+ return false;
+}
+
+void FlatteningController::ThreadFn(
+ const std::shared_ptr<FlatteningController> &fc) {
+ for (;;) {
+ std::unique_lock<std::mutex> lock(fc->mutex_);
+ if (fc.use_count() == 1 || !fc->cbks_.trigger)
+ break;
+
+ if (fc->sleep_until_ <= std::chrono::system_clock::now() &&
+ !fc->disabled_) {
+ fc->disabled_ = true;
+ fc->flatten_next_frame_ = true;
+ ALOGV("Timeout. Sending an event to compositor");
+ fc->cbks_.trigger();
+ }
+
+ if (fc->disabled_) {
+ ALOGV("Wait");
+ fc->cv_.wait(lock);
+ } else {
+ ALOGV("Wait_until");
+ fc->cv_.wait_until(lock, fc->sleep_until_);
+ }
+ }
+}
+
+} // namespace android