大家好，又见面了，我是你们的朋友全栈君。如果您正在找激活码,请点击查看最新教程,关注关注公众号 “全栈程序员社区” 获取激活教程,可能之前旧版本教程已经失效.最新Idea2022.1教程亲测有效,一键激活。
Jetbrains全系列IDE稳定放心使用
资料：https://blog.csdn.net/luoshengyang/article/details/51348829
总结来说，分块的光栅化过程包含了以下三个主要的步骤：
1. 根据分块的可见性，将它们划分到不同的Bin中。
2. 根据内存限制策略，从优先级较高的Bin中选集出需要光栅化的分块。
3. 为每一个需要光栅化的分块分配光栅化内存，并且分别为它们创建光栅化任务。
光栅化任务的执行过程又主要分为以下两个步骤：
1. 创建画布。画布以前面分配的光栅化内存为后端存储。
2. 调用画布提供的绘制接口执行以前记录的绘制命令。这些绘制命令就作用在前面分配的光栅化内存中。也就是前面分配的光栅化内存保存了分块的光栅化结果。
当所有标记为Acquired For Activation的分块都光栅化完成之后，就会触发一个ACTION_ACTIVATE_PENDING_TREE操作，也就是将CC Pending Layer Tree激活为CC Active Layer Tree。
创建光栅化线程：
void SchedulerWorkerPoolImpl::Start(
const SchedulerWorkerPoolParams& params,
int max_background_tasks,
scoped_refptr<TaskRunner> service_thread_task_runner,
SchedulerWorkerObserver* scheduler_worker_observer,
WorkerEnvironment worker_environment) {
AutoSchedulerLock auto_lock(lock_);
DCHECK(workers_.empty());
...
for (int index = 0; index < num_initial_workers; ++index) {
SchedulerWorker* worker =
CreateRegisterAndStartSchedulerWorkerLockRequired();
// CHECK that the first worker can be started (assume that failure means
// that threads can't be created on this machine).
CHECK(worker || index > 0);
if (worker) {
if (index < num_wake_ups_before_start_) {
worker->WakeUp();
} else {
idle_workers_stack_.Push(worker);
}
}
}
}src/base/task/task_scheduler/scheduler_worker_pool_impl.cc
SchedulerWorker*
SchedulerWorkerPoolImpl::CreateRegisterAndStartSchedulerWorkerLockRequired() {
lock_.AssertAcquired();
DCHECK_LT(workers_.size(), max_tasks_);
DCHECK_LT(workers_.size(), kMaxNumberOfWorkers);
// SchedulerWorker needs |lock_| as a predecessor for its thread lock
// because in WakeUpOneWorker, |lock_| is first acquired and then
// the thread lock is acquired when WakeUp is called on the worker.
scoped_refptr<SchedulerWorker> worker = MakeRefCounted<SchedulerWorker>(
priority_hint_,
std::make_unique<SchedulerWorkerDelegateImpl>(
tracked_ref_factory_.GetTrackedRef()),
task_tracker_, &lock_, backward_compatibility_);
if (!worker->Start(scheduler_worker_observer_))
return nullptr;
workers_.push_back(worker);
DCHECK_LE(workers_.size(), max_tasks_);
if (!cleanup_timestamps_.empty()) {
detach_duration_histogram_->AddTime(TimeTicks::Now() -
cleanup_timestamps_.top());
cleanup_timestamps_.pop();
}
return worker.get();
}src/base/task/task_scheduler/scheduler_worker_pool_impl.cc
运行光栅化线程：
void CategorizedWorkerPool::Run(
const std::vector<cc::TaskCategory>& categories,
base::ConditionVariable* has_ready_to_run_tasks_cv) {
base::AutoLock lock(lock_);
while (true) {
if (!RunTaskWithLockAcquired(categories)) {
// We are no longer running tasks, which may allow another category to
// start running. Signal other worker threads.
SignalHasReadyToRunTasksWithLockAcquired();
// Exit when shutdown is set and no more tasks are pending.
if (shutdown_)
break;
// Wait for more tasks.
has_ready_to_run_tasks_cv->Wait();
continue;
}
}
}src/content/renderer/categorized_worker_pool.cc
发布者：全栈程序员-站长，转载请注明出处：https://javaforall.net/179637.html原文链接：https://javaforall.net
