diff options
author | Jamie Liu <jamieliu@google.com> | 2020-03-25 10:56:36 -0700 |
---|---|---|
committer | gVisor bot <gvisor-bot@google.com> | 2020-03-25 10:59:39 -0700 |
commit | e7fbf6949514f1cf239437dcead4ceed5aac029e (patch) | |
tree | 8419ee413c5e4fc44289675f717957bfd2f5a781 /test/perf/linux | |
parent | d04adebaab86ac30aca463b06528fc22430598ac (diff) |
Fix futex_benchmark.
- Fix definitions of Futex* wrappers.
- Correctly handle glibc syscall() (which returns -1 and sets errno instead of
returning the raw syscall return value).
- De-parameterize FutexWaitBitset, which was apparently intended to test with
deadlines of between 0 and 100000 nanoseconds after the Unix epoch, but was
broken due to the preceding two issues.
- Use wall time to measure the durations of tests that are expected to block
(and thus stop accumulating CPU time).
- Require 5s for all tests to improve robustness in the presence of sentry GC.
- Remove FutexContend and FutexContendDeadline; it's unclear what these are
supposed to measure, given that (1) FutexLock is unrealistically inefficient
and (2) the benchmark rewards slow scheduling (since this reduces
contention).
PiperOrigin-RevId: 302925246
Diffstat (limited to 'test/perf/linux')
-rw-r--r-- | test/perf/linux/futex_benchmark.cc | 144 |
1 files changed, 47 insertions, 97 deletions
diff --git a/test/perf/linux/futex_benchmark.cc b/test/perf/linux/futex_benchmark.cc index b349d50bf..241f39896 100644 --- a/test/perf/linux/futex_benchmark.cc +++ b/test/perf/linux/futex_benchmark.cc @@ -33,24 +33,24 @@ namespace testing { namespace { inline int FutexWait(std::atomic<int32_t>* v, int32_t val) { - return syscall(SYS_futex, v, FUTEX_BITSET_MATCH_ANY, nullptr); + return syscall(SYS_futex, v, FUTEX_WAIT_PRIVATE, val, nullptr); } -inline int FutexWaitRelativeTimeout(std::atomic<int32_t>* v, int32_t val, - const struct timespec* reltime) { - return syscall(SYS_futex, v, FUTEX_WAIT_PRIVATE, reltime); +inline int FutexWaitMonotonicTimeout(std::atomic<int32_t>* v, int32_t val, + const struct timespec* timeout) { + return syscall(SYS_futex, v, FUTEX_WAIT_PRIVATE, val, timeout); } -inline int FutexWaitAbsoluteTimeout(std::atomic<int32_t>* v, int32_t val, - const struct timespec* abstime) { - return syscall(SYS_futex, v, FUTEX_BITSET_MATCH_ANY, abstime); +inline int FutexWaitMonotonicDeadline(std::atomic<int32_t>* v, int32_t val, + const struct timespec* deadline) { + return syscall(SYS_futex, v, FUTEX_WAIT_BITSET_PRIVATE, val, deadline, + nullptr, FUTEX_BITSET_MATCH_ANY); } -inline int FutexWaitBitsetAbsoluteTimeout(std::atomic<int32_t>* v, int32_t val, - int32_t bits, - const struct timespec* abstime) { +inline int FutexWaitRealtimeDeadline(std::atomic<int32_t>* v, int32_t val, + const struct timespec* deadline) { return syscall(SYS_futex, v, FUTEX_WAIT_BITSET_PRIVATE | FUTEX_CLOCK_REALTIME, - val, abstime, nullptr, bits); + val, deadline, nullptr, FUTEX_BITSET_MATCH_ANY); } inline int FutexWake(std::atomic<int32_t>* v, int32_t count) { @@ -62,11 +62,11 @@ void BM_FutexWakeNop(benchmark::State& state) { std::atomic<int32_t> v(0); for (auto _ : state) { - EXPECT_EQ(0, FutexWake(&v, 1)); + TEST_PCHECK(FutexWake(&v, 1) == 0); } } -BENCHMARK(BM_FutexWakeNop); +BENCHMARK(BM_FutexWakeNop)->MinTime(5); // This just uses FUTEX_WAIT on an address whose value has changed, i.e., the // syscall won't wait. @@ -74,43 +74,63 @@ void BM_FutexWaitNop(benchmark::State& state) { std::atomic<int32_t> v(0); for (auto _ : state) { - EXPECT_EQ(-EAGAIN, FutexWait(&v, 1)); + TEST_PCHECK(FutexWait(&v, 1) == -1 && errno == EAGAIN); } } -BENCHMARK(BM_FutexWaitNop); +BENCHMARK(BM_FutexWaitNop)->MinTime(5); // This uses FUTEX_WAIT with a timeout on an address whose value never // changes, such that it always times out. Timeout overhead can be estimated by // timer overruns for short timeouts. -void BM_FutexWaitTimeout(benchmark::State& state) { +void BM_FutexWaitMonotonicTimeout(benchmark::State& state) { const int timeout_ns = state.range(0); std::atomic<int32_t> v(0); auto ts = absl::ToTimespec(absl::Nanoseconds(timeout_ns)); for (auto _ : state) { - EXPECT_EQ(-ETIMEDOUT, FutexWaitRelativeTimeout(&v, 0, &ts)); + TEST_PCHECK(FutexWaitMonotonicTimeout(&v, 0, &ts) == -1 && + errno == ETIMEDOUT); } } -BENCHMARK(BM_FutexWaitTimeout) +BENCHMARK(BM_FutexWaitMonotonicTimeout) + ->MinTime(5) + ->UseRealTime() ->Arg(1) ->Arg(10) ->Arg(100) ->Arg(1000) ->Arg(10000); -// This calls FUTEX_WAIT_BITSET with CLOCK_REALTIME. -void BM_FutexWaitBitset(benchmark::State& state) { +// This uses FUTEX_WAIT_BITSET with a deadline that is in the past. This allows +// estimation of the overhead of setting up a timer for a deadline (as opposed +// to a timeout as specified for FUTEX_WAIT). +void BM_FutexWaitMonotonicDeadline(benchmark::State& state) { std::atomic<int32_t> v(0); - int timeout_ns = state.range(0); - auto ts = absl::ToTimespec(absl::Nanoseconds(timeout_ns)); + struct timespec ts = {}; + for (auto _ : state) { - EXPECT_EQ(-ETIMEDOUT, FutexWaitBitsetAbsoluteTimeout(&v, 0, 1, &ts)); + TEST_PCHECK(FutexWaitMonotonicDeadline(&v, 0, &ts) == -1 && + errno == ETIMEDOUT); } } -BENCHMARK(BM_FutexWaitBitset)->Range(0, 100000); +BENCHMARK(BM_FutexWaitMonotonicDeadline)->MinTime(5); + +// This is equivalent to BM_FutexWaitMonotonicDeadline, but uses CLOCK_REALTIME +// instead of CLOCK_MONOTONIC for the deadline. +void BM_FutexWaitRealtimeDeadline(benchmark::State& state) { + std::atomic<int32_t> v(0); + struct timespec ts = {}; + + for (auto _ : state) { + TEST_PCHECK(FutexWaitRealtimeDeadline(&v, 0, &ts) == -1 && + errno == ETIMEDOUT); + } +} + +BENCHMARK(BM_FutexWaitRealtimeDeadline)->MinTime(5); int64_t GetCurrentMonotonicTimeNanos() { struct timespec ts; @@ -130,11 +150,10 @@ void SpinNanos(int64_t delay_ns) { // Each iteration of FutexRoundtripDelayed involves a thread sending a futex // wakeup to another thread, which spins for delay_us and then sends a futex -// wakeup back. The time per iteration is 2* (delay_us + kBeforeWakeDelayNs + +// wakeup back. The time per iteration is 2 * (delay_us + kBeforeWakeDelayNs + // futex/scheduling overhead). void BM_FutexRoundtripDelayed(benchmark::State& state) { const int delay_us = state.range(0); - const int64_t delay_ns = delay_us * 1000; // Spin for an extra kBeforeWakeDelayNs before invoking FUTEX_WAKE to reduce // the probability that the wakeup comes before the wait, preventing the wait @@ -165,83 +184,14 @@ void BM_FutexRoundtripDelayed(benchmark::State& state) { } BENCHMARK(BM_FutexRoundtripDelayed) + ->MinTime(5) + ->UseRealTime() ->Arg(0) ->Arg(10) ->Arg(20) ->Arg(50) ->Arg(100); -// FutexLock is a simple, dumb futex based lock implementation. -// It will try to acquire the lock by atomically incrementing the -// lock word. If it did not increment the lock from 0 to 1, someone -// else has the lock, so it will FUTEX_WAIT until it is woken in -// the unlock path. -class FutexLock { - public: - FutexLock() : lock_word_(0) {} - - void lock(struct timespec* deadline) { - int32_t val; - while ((val = lock_word_.fetch_add(1, std::memory_order_acquire) + 1) != - 1) { - // If we didn't get the lock by incrementing from 0 to 1, - // do a FUTEX_WAIT with the desired current value set to - // val. If val is no longer what the atomic increment returned, - // someone might have set it to 0 so we can try to acquire - // again. - int ret = FutexWaitAbsoluteTimeout(&lock_word_, val, deadline); - if (ret == 0 || ret == -EWOULDBLOCK || ret == -EINTR) { - continue; - } else { - FAIL() << "unexpected FUTEX_WAIT return: " << ret; - } - } - } - - void unlock() { - // Store 0 into the lock word and wake one waiter. We intentionally - // ignore the return value of the FUTEX_WAKE here, since there may be - // no waiters to wake anyway. - lock_word_.store(0, std::memory_order_release); - (void)FutexWake(&lock_word_, 1); - } - - private: - std::atomic<int32_t> lock_word_; -}; - -FutexLock* test_lock; // Used below. - -void FutexContend(benchmark::State& state, int thread_index, - struct timespec* deadline) { - int counter = 0; - if (thread_index == 0) { - test_lock = new FutexLock(); - } - for (auto _ : state) { - test_lock->lock(deadline); - counter++; - test_lock->unlock(); - } - if (thread_index == 0) { - delete test_lock; - } - state.SetItemsProcessed(state.iterations()); -} - -void BM_FutexContend(benchmark::State& state) { - FutexContend(state, state.thread_index, nullptr); -} - -BENCHMARK(BM_FutexContend)->ThreadRange(1, 1024)->UseRealTime(); - -void BM_FutexDeadlineContend(benchmark::State& state) { - auto deadline = absl::ToTimespec(absl::Now() + absl::Minutes(10)); - FutexContend(state, state.thread_index, &deadline); -} - -BENCHMARK(BM_FutexDeadlineContend)->ThreadRange(1, 1024)->UseRealTime(); - } // namespace } // namespace testing |