rocksdb/util/timer.h

262 lines
6.9 KiB
C
Raw Normal View History

// Copyright (c) 2011-present, Facebook, Inc. All rights reserved.
// This source code is licensed under both the GPLv2 (found in the
// COPYING file in the root directory) and Apache 2.0 License
// (found in the LICENSE.Apache file in the root directory).
//
#pragma once
#include <functional>
#include <memory>
#include <queue>
#include <unordered_map>
#include <utility>
#include <vector>
#include "monitoring/instrumented_mutex.h"
#include "rocksdb/env.h"
#include "test_util/sync_point.h"
#include "util/mutexlock.h"
namespace ROCKSDB_NAMESPACE {
// A Timer class to handle repeated work.
//
// A single timer instance can handle multiple functions via a single thread.
// It is better to leave long running work to a dedicated thread pool.
//
// Timer can be started by calling `Start()`, and ended by calling `Shutdown()`.
// Work (in terms of a `void function`) can be scheduled by calling `Add` with
// a unique function name and de-scheduled by calling `Cancel`.
// Many functions can be added.
//
// Impl Details:
// A heap is used to keep track of when the next timer goes off.
// A map from a function name to the function keeps track of all the functions.
class Timer {
public:
explicit Timer(Env* env)
: env_(env),
mutex_(env),
cond_var_(&mutex_),
running_(false),
executing_task_(false) {}
~Timer() {}
Fix+clean up handling of mock sleeps (#7101) Summary: We have a number of tests hanging on MacOS and windows due to mishandling of code for mock sleeps. In addition, the code was in terrible shape because the same variable (addon_time_) would sometimes refer to microseconds and sometimes to seconds. One test even assumed it was nanoseconds but was written to pass anyway. This has been cleaned up so that DB tests generally use a SpecialEnv function to mock sleep, for either some number of microseconds or seconds depending on the function called. But to call one of these, the test must first call SetMockSleep (precondition enforced with assertion), which also turns sleeps in RocksDB into mock sleeps. To also removes accounting for actual clock time, call SetTimeElapseOnlySleepOnReopen, which implies SetMockSleep (on DB re-open). This latter setting only works by applying on DB re-open, otherwise havoc can ensue if Env goes back in time with DB open. More specifics: Removed some unused test classes, and updated comments on the general problem. Fixed DBSSTTest.GetTotalSstFilesSize using a sync point callback instead of mock time. For this we have the only modification to production code, inserting a sync point callback in flush_job.cc, which is not a change to production behavior. Removed unnecessary resetting of mock times to 0 in many tests. RocksDB deals in relative time. Any behaviors relying on absolute date/time are likely a bug. (The above test DBSSTTest.GetTotalSstFilesSize was the only one clearly injecting a specific absolute time for actual testing convenience.) Just in case I misunderstood some test, I put this note in each replacement: // NOTE: Presumed unnecessary and removed: resetting mock time in env Strengthened some tests like MergeTestTime, MergeCompactionTimeTest, and FilterCompactionTimeTest in db_test.cc stats_history_test and blob_db_test are each their own beast, rather deeply dependent on MockTimeEnv. Each gets its own variant of a work-around for TimedWait in a mock time environment. (Reduces redundancy and inconsistency in stats_history_test.) Intended follow-up: Remove TimedWait from the public API of InstrumentedCondVar, and only make that accessible through Env by passing in an InstrumentedCondVar and a deadline. Then the Env implementations mocking time can fix this problem without using sync points. (Test infrastructure using sync points interferes with individual tests' control over sync points.) With that change, we can simplify/consolidate the scattered work-arounds. Pull Request resolved: https://github.com/facebook/rocksdb/pull/7101 Test Plan: make check on Linux and MacOS Reviewed By: zhichao-cao Differential Revision: D23032815 Pulled By: pdillinger fbshipit-source-id: 7f33967ada8b83011fb54e8279365c008bd6610b
2020-08-11 21:39:49 +02:00
// repeat_every_us == 0 means do not repeat
void Add(std::function<void()> fn,
const std::string& fn_name,
uint64_t start_after_us,
uint64_t repeat_every_us) {
std::unique_ptr<FunctionInfo> fn_info(new FunctionInfo(
std::move(fn),
fn_name,
env_->NowMicros() + start_after_us,
repeat_every_us));
InstrumentedMutexLock l(&mutex_);
heap_.push(fn_info.get());
map_.emplace(std::make_pair(fn_name, std::move(fn_info)));
cond_var_.Signal();
}
void Cancel(const std::string& fn_name) {
InstrumentedMutexLock l(&mutex_);
// Mark the function with fn_name as invalid so that it will not be
// requeued.
auto it = map_.find(fn_name);
if (it != map_.end() && it->second) {
it->second->Cancel();
}
// If the currently running function is fn_name, then we need to wait
// until it finishes before returning to caller.
while (!heap_.empty() && executing_task_) {
FunctionInfo* func_info = heap_.top();
assert(func_info);
if (func_info->name == fn_name) {
WaitForTaskCompleteIfNecessary();
} else {
break;
}
}
}
void CancelAll() {
InstrumentedMutexLock l(&mutex_);
CancelAllWithLock();
}
// Start the Timer
bool Start() {
InstrumentedMutexLock l(&mutex_);
if (running_) {
return false;
}
running_ = true;
thread_.reset(new port::Thread(&Timer::Run, this));
return true;
}
// Shutdown the Timer
bool Shutdown() {
{
InstrumentedMutexLock l(&mutex_);
if (!running_) {
return false;
}
running_ = false;
CancelAllWithLock();
cond_var_.SignalAll();
}
if (thread_) {
thread_->join();
}
return true;
}
private:
void Run() {
InstrumentedMutexLock l(&mutex_);
while (running_) {
if (heap_.empty()) {
// wait
TEST_SYNC_POINT("Timer::Run::Waiting");
cond_var_.Wait();
continue;
}
FunctionInfo* current_fn = heap_.top();
assert(current_fn);
if (!current_fn->IsValid()) {
heap_.pop();
map_.erase(current_fn->name);
continue;
}
if (current_fn->next_run_time_us <= env_->NowMicros()) {
executing_task_ = true;
mutex_.Unlock();
// Execute the work
current_fn->fn();
mutex_.Lock();
executing_task_ = false;
cond_var_.SignalAll();
// Remove the work from the heap once it is done executing.
// Note that we are just removing the pointer from the heap. Its
// memory is still managed in the map (as it holds a unique ptr).
// So current_fn is still a valid ptr.
heap_.pop();
// current_fn may be cancelled already.
if (current_fn->IsValid() && current_fn->repeat_every_us > 0) {
assert(running_);
current_fn->next_run_time_us = env_->NowMicros() +
current_fn->repeat_every_us;
// Schedule new work into the heap with new time.
heap_.push(current_fn);
}
} else {
cond_var_.TimedWait(current_fn->next_run_time_us);
}
}
}
void CancelAllWithLock() {
mutex_.AssertHeld();
if (map_.empty() && heap_.empty()) {
return;
}
// With mutex_ held, set all tasks to invalid so that they will not be
// re-queued.
for (auto& elem : map_) {
auto& func_info = elem.second;
assert(func_info);
func_info->Cancel();
}
// WaitForTaskCompleteIfNecessary() may release mutex_
WaitForTaskCompleteIfNecessary();
while (!heap_.empty()) {
heap_.pop();
}
map_.clear();
}
// A wrapper around std::function to keep track when it should run next
// and at what frequency.
struct FunctionInfo {
// the actual work
std::function<void()> fn;
// name of the function
std::string name;
// when the function should run next
uint64_t next_run_time_us;
// repeat interval
uint64_t repeat_every_us;
// controls whether this function is valid.
// A function is valid upon construction and until someone explicitly
// calls `Cancel()`.
bool valid;
FunctionInfo(std::function<void()>&& _fn, const std::string& _name,
const uint64_t _next_run_time_us, uint64_t _repeat_every_us)
: fn(std::move(_fn)),
name(_name),
next_run_time_us(_next_run_time_us),
repeat_every_us(_repeat_every_us),
valid(true) {}
void Cancel() {
valid = false;
}
bool IsValid() const { return valid; }
};
void WaitForTaskCompleteIfNecessary() {
mutex_.AssertHeld();
while (executing_task_) {
TEST_SYNC_POINT("Timer::WaitForTaskCompleteIfNecessary:TaskExecuting");
cond_var_.Wait();
}
}
struct RunTimeOrder {
bool operator()(const FunctionInfo* f1,
const FunctionInfo* f2) {
return f1->next_run_time_us > f2->next_run_time_us;
}
};
Env* const env_;
// This mutex controls both the heap_ and the map_. It needs to be held for
// making any changes in them.
InstrumentedMutex mutex_;
InstrumentedCondVar cond_var_;
std::unique_ptr<port::Thread> thread_;
bool running_;
bool executing_task_;
std::priority_queue<FunctionInfo*,
std::vector<FunctionInfo*>,
RunTimeOrder> heap_;
// In addition to providing a mapping from a function name to a function,
// it is also responsible for memory management.
std::unordered_map<std::string, std::unique_ptr<FunctionInfo>> map_;
};
} // namespace ROCKSDB_NAMESPACE