-
Notifications
You must be signed in to change notification settings - Fork 116
/
ticker.h
241 lines (207 loc) · 6.11 KB
/
ticker.h
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
#pragma once
#include <cstdint>
#include <atomic>
#include <thread>
#include "core.h"
#include "macros.h"
#include "spinlock.h"
#include "lockguard.h"
class ticker {
public:
#ifdef CHECK_INVARIANTS
static const uint64_t tick_us = 1 * 1000; /* 1 ms */
#else
static const uint64_t tick_us = 40 * 1000; /* 40 ms */
#endif
ticker()
: current_tick_(1), last_tick_inclusive_(0)
{
std::thread thd(&ticker::tickerloop, this);
thd.detach();
}
inline uint64_t
global_current_tick() const
{
return current_tick_.load(std::memory_order_acquire);
}
inline uint64_t
global_last_tick_inclusive() const
{
return last_tick_inclusive_.load(std::memory_order_acquire);
}
inline uint64_t
global_last_tick_exclusive() const
{
return global_last_tick_inclusive() + 1;
}
// should yield a # >= global_last_tick_exclusive()
uint64_t
compute_global_last_tick_exclusive() const
{
uint64_t e = ticks_[0].current_tick_.load(std::memory_order_acquire);
for (size_t i = 1; i < ticks_.size(); i++)
e = std::min(e, ticks_[i].current_tick_.load(std::memory_order_acquire));
return e;
}
// returns true if guard is currently active, along with filling
// cur_epoch out
inline bool
is_locally_guarded(uint64_t &cur_epoch) const
{
const uint64_t core_id = coreid::core_id();
const uint64_t current_tick =
ticks_[core_id].current_tick_.load(std::memory_order_acquire);
const uint64_t current_depth =
ticks_[core_id].depth_.load(std::memory_order_acquire);
if (current_depth)
cur_epoch = current_tick;
return current_depth;
}
inline bool
is_locally_guarded() const
{
uint64_t c;
return is_locally_guarded(c);
}
inline spinlock &
lock_for(uint64_t core_id)
{
INVARIANT(core_id < ticks_.size());
return ticks_[core_id].lock_;
}
// a guard is re-entrant within a single thread
class guard {
public:
guard(ticker &impl)
: impl_(&impl), core_(coreid::core_id()), start_us_(0)
{
tickinfo &ti = impl_->ticks_[core_];
// bump the depth first
const uint64_t prev_depth = util::non_atomic_fetch_add(ti.depth_, 1UL);
// grab the lock
if (!prev_depth) {
ti.lock_.lock();
// read epoch # (try to advance forward)
tick_ = impl_->global_current_tick();
INVARIANT(ti.current_tick_.load(std::memory_order_acquire) <= tick_);
ti.current_tick_.store(tick_, std::memory_order_release);
start_us_ = util::timer::cur_usec();
ti.start_us_.store(start_us_, std::memory_order_release);
} else {
tick_ = ti.current_tick_.load(std::memory_order_acquire);
start_us_ = ti.start_us_.load(std::memory_order_acquire);
}
INVARIANT(ti.lock_.is_locked());
depth_ = prev_depth + 1;
}
guard(guard &&) = default;
guard(const guard &) = delete;
guard &operator=(const guard &) = delete;
~guard()
{
if (!impl_)
return;
INVARIANT(core_ == coreid::core_id());
tickinfo &ti = impl_->ticks_[core_];
INVARIANT(ti.lock_.is_locked());
INVARIANT(tick_ > impl_->global_last_tick_inclusive());
const uint64_t prev_depth = util::non_atomic_fetch_sub(ti.depth_, 1UL);
INVARIANT(prev_depth);
// unlock
if (prev_depth == 1) {
ti.start_us_.store(0, std::memory_order_release);
ti.lock_.unlock();
}
}
inline uint64_t
tick() const
{
INVARIANT(impl_);
return tick_;
}
inline uint64_t
core() const
{
INVARIANT(impl_);
return core_;
}
inline uint64_t
depth() const
{
INVARIANT(impl_);
return depth_;
}
inline const ticker &
impl() const
{
INVARIANT(impl_);
return *impl_;
}
// refers to the start time of the *outermost* scope
inline uint64_t
start_us() const
{
return start_us_;
}
private:
ticker *impl_;
uint64_t core_;
uint64_t tick_;
uint64_t depth_;
uint64_t start_us_;
};
static ticker s_instance CACHE_ALIGNED; // system wide ticker
private:
void
tickerloop()
{
// runs as daemon
util::timer loop_timer;
struct timespec t;
for (;;) {
const uint64_t last_loop_usec = loop_timer.lap();
const uint64_t delay_time_usec = tick_us;
if (last_loop_usec < delay_time_usec) {
const uint64_t sleep_ns = (delay_time_usec - last_loop_usec) * 1000;
t.tv_sec = sleep_ns / ONE_SECOND_NS;
t.tv_nsec = sleep_ns % ONE_SECOND_NS;
nanosleep(&t, nullptr);
loop_timer.lap(); // since we slept away the lag
}
// bump the current tick
// XXX: ignore overflow
const uint64_t last_tick = util::non_atomic_fetch_add(current_tick_, 1UL);
const uint64_t cur_tick = last_tick + 1;
// wait for all threads to finish the last tick
for (size_t i = 0; i < ticks_.size(); i++) {
tickinfo &ti = ticks_[i];
const uint64_t thread_cur_tick =
ti.current_tick_.load(std::memory_order_acquire);
INVARIANT(thread_cur_tick == last_tick ||
thread_cur_tick == cur_tick);
if (thread_cur_tick == cur_tick)
continue;
lock_guard<spinlock> lg(ti.lock_);
ti.current_tick_.store(cur_tick, std::memory_order_release);
}
last_tick_inclusive_.store(last_tick, std::memory_order_release);
}
}
struct tickinfo {
spinlock lock_; // guards current_tick_ and depth_
std::atomic<uint64_t> current_tick_; // last RCU epoch this thread has seen
// (implies completion through current_tick_ - 1)
std::atomic<uint64_t> depth_; // 0 if not in RCU section
std::atomic<uint64_t> start_us_; // 0 if not in RCU section
tickinfo()
: current_tick_(1), depth_(0), start_us_(0)
{
ALWAYS_ASSERT(((uintptr_t)this % CACHELINE_SIZE) == 0);
}
};
percore<tickinfo> ticks_;
std::atomic<uint64_t> current_tick_; // which tick are we currenlty on?
std::atomic<uint64_t> last_tick_inclusive_;
// all threads have *completed* ticks <= last_tick_inclusive_
// (< current_tick_)
};