4 files changed, 133 insertions, 556 deletions

diff --git a/liblog/tests/Android.mk b/liblog/tests/Android.mk
index 39b52ac35..cfa849ba4 100644
--- a/liblog/tests/Android.mk
+++ b/liblog/tests/Android.mk
@@ -24,13 +24,12 @@ test_module_prefix := liblog-
 test_tags := tests
 
 benchmark_c_flags := \
-    -Ibionic/tests \
-    -Wall -Wextra \
+    -Wall \
+    -Wextra \
     -Werror \
     -fno-builtin \
 
 benchmark_src_files := \
-    benchmark_main.cpp \
     liblog_benchmark.cpp
 
 # Build benchmarks for the device. Run with:
@@ -41,7 +40,7 @@ LOCAL_MODULE_TAGS := $(test_tags)
 LOCAL_CFLAGS += $(benchmark_c_flags)
 LOCAL_SHARED_LIBRARIES += liblog libm libbase
 LOCAL_SRC_FILES := $(benchmark_src_files)
-include $(BUILD_NATIVE_TEST)
+include $(BUILD_NATIVE_BENCHMARK)
 
 # -----------------------------------------------------------------------------
 # Unit tests.
diff --git a/liblog/tests/benchmark.h b/liblog/tests/benchmark.h
deleted file mode 100644
index e9280f63e..000000000
--- a/liblog/tests/benchmark.h
+++ /dev/null
@@ -1,147 +0,0 @@
-/*
- * Copyright (C) 2012-2014 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.
- */
-
-#include <stdint.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-
-#include <vector>
-
-#ifndef BIONIC_BENCHMARK_H_
-#define BIONIC_BENCHMARK_H_
-
-namespace testing {
-
-class Benchmark;
-template <typename T> class BenchmarkWantsArg;
-template <typename T> class BenchmarkWithArg;
-
-void BenchmarkRegister(Benchmark* bm);
-int PrettyPrintInt(char* str, int len, unsigned int arg);
-
-class Benchmark {
- public:
-  Benchmark(const char* name, void (*fn)(int)) : name_(strdup(name)), fn_(fn) {
-    BenchmarkRegister(this);
-  }
-  explicit Benchmark(const char* name) : name_(strdup(name)), fn_(NULL) {}
-
-  virtual ~Benchmark() {
-    free(name_);
-  }
-
-  const char* Name() { return name_; }
-  virtual const char* ArgName() { return NULL; }
-  virtual void RunFn(int iterations) { fn_(iterations); }
-
- protected:
-  char* name_;
-
- private:
-  void (*fn_)(int);
-};
-
-template <typename T>
-class BenchmarkWantsArgBase : public Benchmark {
- public:
-  BenchmarkWantsArgBase(const char* name, void (*fn)(int, T)) : Benchmark(name) {
-    fn_arg_ = fn;
-  }
-
-  BenchmarkWantsArgBase<T>* Arg(const char* arg_name, T arg) {
-    BenchmarkRegister(new BenchmarkWithArg<T>(name_, fn_arg_, arg_name, arg));
-    return this;
-  }
-
- protected:
-  virtual void RunFn(int) { printf("can't run arg benchmark %s without arg\n", Name()); }
-  void (*fn_arg_)(int, T);
-};
-
-template <typename T>
-class BenchmarkWithArg : public BenchmarkWantsArg<T> {
- public:
-  BenchmarkWithArg(const char* name, void (*fn)(int, T), const char* arg_name, T arg) :
-      BenchmarkWantsArg<T>(name, fn), arg_(arg) {
-    arg_name_ = strdup(arg_name);
-  }
-
-  virtual ~BenchmarkWithArg() {
-    free(arg_name_);
-  }
-
-  virtual const char* ArgName() { return arg_name_; }
-
- protected:
-  virtual void RunFn(int iterations) { BenchmarkWantsArg<T>::fn_arg_(iterations, arg_); }
-
- private:
-  T arg_;
-  char* arg_name_;
-};
-
-template <typename T>
-class BenchmarkWantsArg : public BenchmarkWantsArgBase<T> {
- public:
-  BenchmarkWantsArg<T>(const char* name, void (*fn)(int, T)) :
-    BenchmarkWantsArgBase<T>(name, fn) { }
-};
-
-template <>
-class BenchmarkWantsArg<int> : public BenchmarkWantsArgBase<int> {
- public:
-  BenchmarkWantsArg<int>(const char* name, void (*fn)(int, int)) :
-    BenchmarkWantsArgBase<int>(name, fn) { }
-
-  BenchmarkWantsArg<int>* Arg(int arg) {
-    char arg_name[100];
-    PrettyPrintInt(arg_name, sizeof(arg_name), arg);
-    BenchmarkRegister(new BenchmarkWithArg<int>(name_, fn_arg_, arg_name, arg));
-    return this;
-  }
-};
-
-static inline Benchmark* BenchmarkFactory(const char* name, void (*fn)(int)) {
-  return new Benchmark(name, fn);
-}
-
-template <typename T>
-static inline BenchmarkWantsArg<T>* BenchmarkFactory(const char* name, void (*fn)(int, T)) {
-  return new BenchmarkWantsArg<T>(name, fn);
-}
-
-}  // namespace testing
-
-template <typename T>
-static inline void BenchmarkAddArg(::testing::Benchmark* b, const char* name, T arg) {
-  ::testing::BenchmarkWantsArg<T>* ba;
-  ba = static_cast< ::testing::BenchmarkWantsArg<T>* >(b);
-  ba->Arg(name, arg);
-}
-
-void SetBenchmarkBytesProcessed(uint64_t);
-void ResetBenchmarkTiming(void);
-void StopBenchmarkTiming(void);
-void StartBenchmarkTiming(void);
-void StartBenchmarkTiming(uint64_t);
-void StopBenchmarkTiming(uint64_t);
-
-#define BENCHMARK(f) \
-    static ::testing::Benchmark* _benchmark_##f __attribute__((unused)) = /* NOLINT */ \
-        (::testing::Benchmark*)::testing::BenchmarkFactory(#f, f) /* NOLINT */
-
-#endif // BIONIC_BENCHMARK_H_
diff --git a/liblog/tests/benchmark_main.cpp b/liblog/tests/benchmark_main.cpp
deleted file mode 100644
index 7367f1b52..000000000
--- a/liblog/tests/benchmark_main.cpp
+++ /dev/null
@@ -1,247 +0,0 @@
-/*
- * Copyright (C) 2012-2014 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.
- */
-
-#include <benchmark.h>
-
-#include <inttypes.h>
-#include <math.h>
-#include <regex.h>
-#include <stdio.h>
-#include <stdlib.h>
-
-#include <map>
-#include <string>
-#include <vector>
-
-static uint64_t gBytesProcessed;
-static uint64_t gBenchmarkTotalTimeNs;
-static uint64_t gBenchmarkTotalTimeNsSquared;
-static uint64_t gBenchmarkNum;
-static uint64_t gBenchmarkStartTimeNs;
-
-typedef std::vector< ::testing::Benchmark*> BenchmarkList;
-static BenchmarkList* gBenchmarks;
-
-static int Round(int n) {
-  int base = 1;
-  while (base * 10 < n) {
-    base *= 10;
-  }
-  if (n < 2 * base) {
-    return 2 * base;
-  }
-  if (n < 5 * base) {
-    return 5 * base;
-  }
-  return 10 * base;
-}
-
-static uint64_t NanoTime() {
-  struct timespec t;
-  t.tv_sec = t.tv_nsec = 0;
-  clock_gettime(CLOCK_MONOTONIC, &t);
-  return static_cast<uint64_t>(t.tv_sec) * 1000000000ULL + t.tv_nsec;
-}
-
-namespace testing {
-
-int PrettyPrintInt(char* str, int len, unsigned int arg) {
-  if (arg >= (1 << 30) && arg % (1 << 30) == 0) {
-    return snprintf(str, len, "%uGi", arg / (1 << 30));
-  } else if (arg >= (1 << 20) && arg % (1 << 20) == 0) {
-    return snprintf(str, len, "%uMi", arg / (1 << 20));
-  } else if (arg >= (1 << 10) && arg % (1 << 10) == 0) {
-    return snprintf(str, len, "%uKi", arg / (1 << 10));
-  } else if (arg >= 1000000000 && arg % 1000000000 == 0) {
-    return snprintf(str, len, "%uG", arg / 1000000000);
-  } else if (arg >= 1000000 && arg % 1000000 == 0) {
-    return snprintf(str, len, "%uM", arg / 1000000);
-  } else if (arg >= 1000 && arg % 1000 == 0) {
-    return snprintf(str, len, "%uK", arg / 1000);
-  } else {
-    return snprintf(str, len, "%u", arg);
-  }
-}
-
-bool ShouldRun(Benchmark* b, int argc, char* argv[]) {
-  if (argc == 1) {
-    return true;  // With no arguments, we run all benchmarks.
-  }
-  // Otherwise, we interpret each argument as a regular expression and
-  // see if any of our benchmarks match.
-  for (int i = 1; i < argc; i++) {
-    regex_t re;
-    if (regcomp(&re, argv[i], 0) != 0) {
-      fprintf(stderr, "couldn't compile \"%s\" as a regular expression!\n",
-              argv[i]);
-      exit(EXIT_FAILURE);
-    }
-    int match = regexec(&re, b->Name(), 0, NULL, 0);
-    regfree(&re);
-    if (match != REG_NOMATCH) {
-      return true;
-    }
-  }
-  return false;
-}
-
-void BenchmarkRegister(Benchmark* b) {
-  if (gBenchmarks == NULL) {
-    gBenchmarks = new BenchmarkList;
-  }
-  gBenchmarks->push_back(b);
-}
-
-void RunRepeatedly(Benchmark* b, int iterations) {
-  gBytesProcessed = 0;
-  ResetBenchmarkTiming();
-  uint64_t StartTimeNs = NanoTime();
-  b->RunFn(iterations);
-  // Catch us if we fail to log anything.
-  if ((gBenchmarkTotalTimeNs == 0) && (StartTimeNs != 0) &&
-      (gBenchmarkStartTimeNs == 0)) {
-    gBenchmarkTotalTimeNs = NanoTime() - StartTimeNs;
-  }
-}
-
-void Run(Benchmark* b) {
-  // run once in case it's expensive
-  unsigned iterations = 1;
-  uint64_t s = NanoTime();
-  RunRepeatedly(b, iterations);
-  s = NanoTime() - s;
-  while (s < 2e9 && gBenchmarkTotalTimeNs < 1e9 && iterations < 1e9) {
-    unsigned last = iterations;
-    if (gBenchmarkTotalTimeNs / iterations == 0) {
-      iterations = 1e9;
-    } else {
-      iterations = 1e9 / (gBenchmarkTotalTimeNs / iterations);
-    }
-    iterations =
-        std::max(last + 1, std::min(iterations + iterations / 2, 100 * last));
-    iterations = Round(iterations);
-    s = NanoTime();
-    RunRepeatedly(b, iterations);
-    s = NanoTime() - s;
-  }
-
-  char throughput[100];
-  throughput[0] = '\0';
-  if (gBenchmarkTotalTimeNs > 0 && gBytesProcessed > 0) {
-    double mib_processed = static_cast<double>(gBytesProcessed) / 1e6;
-    double seconds = static_cast<double>(gBenchmarkTotalTimeNs) / 1e9;
-    snprintf(throughput, sizeof(throughput), " %8.2f MiB/s",
-             mib_processed / seconds);
-  }
-
-  char full_name[100];
-  snprintf(full_name, sizeof(full_name), "%s%s%s", b->Name(),
-           b->ArgName() ? "/" : "", b->ArgName() ? b->ArgName() : "");
-
-  uint64_t mean = gBenchmarkTotalTimeNs / iterations;
-  uint64_t sdev = 0;
-  if (gBenchmarkNum == iterations) {
-    mean = gBenchmarkTotalTimeNs / gBenchmarkNum;
-    uint64_t nXvariance = gBenchmarkTotalTimeNsSquared * gBenchmarkNum -
-                          (gBenchmarkTotalTimeNs * gBenchmarkTotalTimeNs);
-    sdev = (sqrt((double)nXvariance) / gBenchmarkNum / gBenchmarkNum) + 0.5;
-  }
-  if (mean > (10000 * sdev)) {
-    printf("%-25s %10" PRIu64 " %10" PRIu64 "%s\n", full_name,
-           static_cast<uint64_t>(iterations), mean, throughput);
-  } else {
-    printf("%-25s %10" PRIu64 " %10" PRIu64 "(\317\203%" PRIu64 ")%s\n",
-           full_name, static_cast<uint64_t>(iterations), mean, sdev, throughput);
-  }
-  fflush(stdout);
-}
-
-}  // namespace testing
-
-void SetBenchmarkBytesProcessed(uint64_t x) {
-  gBytesProcessed = x;
-}
-
-void ResetBenchmarkTiming() {
-  gBenchmarkStartTimeNs = 0;
-  gBenchmarkTotalTimeNs = 0;
-  gBenchmarkTotalTimeNsSquared = 0;
-  gBenchmarkNum = 0;
-}
-
-void StopBenchmarkTiming(void) {
-  if (gBenchmarkStartTimeNs != 0) {
-    int64_t diff = NanoTime() - gBenchmarkStartTimeNs;
-    gBenchmarkTotalTimeNs += diff;
-    gBenchmarkTotalTimeNsSquared += diff * diff;
-    ++gBenchmarkNum;
-  }
-  gBenchmarkStartTimeNs = 0;
-}
-
-void StartBenchmarkTiming(void) {
-  if (gBenchmarkStartTimeNs == 0) {
-    gBenchmarkStartTimeNs = NanoTime();
-  }
-}
-
-void StopBenchmarkTiming(uint64_t NanoTime) {
-  if (gBenchmarkStartTimeNs != 0) {
-    int64_t diff = NanoTime - gBenchmarkStartTimeNs;
-    gBenchmarkTotalTimeNs += diff;
-    gBenchmarkTotalTimeNsSquared += diff * diff;
-    if (NanoTime != 0) {
-      ++gBenchmarkNum;
-    }
-  }
-  gBenchmarkStartTimeNs = 0;
-}
-
-void StartBenchmarkTiming(uint64_t NanoTime) {
-  if (gBenchmarkStartTimeNs == 0) {
-    gBenchmarkStartTimeNs = NanoTime;
-  }
-}
-
-int main(int argc, char* argv[]) {
-  if (gBenchmarks->empty()) {
-    fprintf(stderr, "No benchmarks registered!\n");
-    exit(EXIT_FAILURE);
-  }
-
-  bool need_header = true;
-  for (auto b : *gBenchmarks) {
-    if (ShouldRun(b, argc, argv)) {
-      if (need_header) {
-        printf("%-25s %10s %10s\n", "", "iterations", "ns/op");
-        fflush(stdout);
-        need_header = false;
-      }
-      Run(b);
-    }
-  }
-
-  if (need_header) {
-    fprintf(stderr, "No matching benchmarks!\n");
-    fprintf(stderr, "Available benchmarks:\n");
-    for (auto b : *gBenchmarks) {
-      fprintf(stderr, "  %s\n", b->Name());
-    }
-    exit(EXIT_FAILURE);
-  }
-
-  return 0;
-}
diff --git a/liblog/tests/liblog_benchmark.cpp b/liblog/tests/liblog_benchmark.cpp
index c4bf95962..706a0f6ae 100644
--- a/liblog/tests/liblog_benchmark.cpp
+++ b/liblog/tests/liblog_benchmark.cpp
@@ -25,12 +25,13 @@
 #include <unordered_set>
 
 #include <android-base/file.h>
+#include <benchmark/benchmark.h>
 #include <cutils/sockets.h>
 #include <log/event_tag_map.h>
 #include <log/log_transport.h>
 #include <private/android_logger.h>
 
-#include "benchmark.h"
+BENCHMARK_MAIN();
 
 // enhanced version of LOG_FAILURE_RETRY to add support for EAGAIN and
 // non-syscall libs. Since we are benchmarking, or using this in the emergency
@@ -51,15 +52,11 @@
  * the log at high pressure. Expect this to be less than double the process
  * wakeup time (2ms?)
  */
-static void BM_log_maximum_retry(int iters) {
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
-    LOG_FAILURE_RETRY(
-        __android_log_print(ANDROID_LOG_INFO, "BM_log_maximum_retry", "%d", i));
+static void BM_log_maximum_retry(benchmark::State& state) {
+  while (state.KeepRunning()) {
+    LOG_FAILURE_RETRY(__android_log_print(
+        ANDROID_LOG_INFO, "BM_log_maximum_retry", "%zu", state.iterations()));
   }
-
-  StopBenchmarkTiming();
 }
 BENCHMARK(BM_log_maximum_retry);
 
@@ -68,14 +65,11 @@ BENCHMARK(BM_log_maximum_retry);
  * at high pressure. Expect this to be less than double the process wakeup
  * time (2ms?)
  */
-static void BM_log_maximum(int iters) {
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
-    __android_log_print(ANDROID_LOG_INFO, "BM_log_maximum", "%d", i);
+static void BM_log_maximum(benchmark::State& state) {
+  while (state.KeepRunning()) {
+    __android_log_print(ANDROID_LOG_INFO, "BM_log_maximum", "%zu",
+                        state.iterations());
   }
-
-  StopBenchmarkTiming();
 }
 BENCHMARK(BM_log_maximum);
 
@@ -87,26 +81,28 @@ static void set_log_default() {
   android_set_log_transport(LOGGER_DEFAULT);
 }
 
-static void BM_log_maximum_null(int iters) {
+static void BM_log_maximum_null(benchmark::State& state) {
   set_log_null();
-  BM_log_maximum(iters);
+  BM_log_maximum(state);
   set_log_default();
 }
 BENCHMARK(BM_log_maximum_null);
 
 /*
  *      Measure the time it takes to collect the time using
- * discrete acquisition (StartBenchmarkTiming() -> StopBenchmarkTiming())
+ * discrete acquisition (state.PauseTiming() to state.ResumeTiming())
  * under light load. Expect this to be a syscall period (2us) or
  * data read time if zero-syscall.
  *
  * vdso support in the kernel and the library can allow
- * clock_gettime to be zero-syscall.
+ * clock_gettime to be zero-syscall, but there there does remain some
+ * benchmarking overhead to pause and resume; assumptions are both are
+ * covered.
  */
-static void BM_clock_overhead(int iters) {
-  for (int i = 0; i < iters; ++i) {
-    StartBenchmarkTiming();
-    StopBenchmarkTiming();
+static void BM_clock_overhead(benchmark::State& state) {
+  while (state.KeepRunning()) {
+    state.PauseTiming();
+    state.ResumeTiming();
   }
 }
 BENCHMARK(BM_clock_overhead);
@@ -114,9 +110,10 @@ BENCHMARK(BM_clock_overhead);
 /*
  * Measure the time it takes to submit the android logging data to pstore
  */
-static void BM_pmsg_short(int iters) {
+static void BM_pmsg_short(benchmark::State& state) {
   int pstore_fd = TEMP_FAILURE_RETRY(open("/dev/pmsg0", O_WRONLY));
   if (pstore_fd < 0) {
+    state.SkipWithError("/dev/pmsg0");
     return;
   }
 
@@ -175,13 +172,12 @@ static void BM_pmsg_short(int iters) {
   newVec[2].iov_base = &buffer;
   newVec[2].iov_len = sizeof(buffer);
 
-  StartBenchmarkTiming();
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     ++snapshot;
     buffer.payload.data = htole32(snapshot);
     writev(pstore_fd, newVec, nr);
   }
-  StopBenchmarkTiming();
+  state.PauseTiming();
   close(pstore_fd);
 }
 BENCHMARK(BM_pmsg_short);
@@ -190,9 +186,10 @@ BENCHMARK(BM_pmsg_short);
  * Measure the time it takes to submit the android logging data to pstore
  * best case aligned single block.
  */
-static void BM_pmsg_short_aligned(int iters) {
+static void BM_pmsg_short_aligned(benchmark::State& state) {
   int pstore_fd = TEMP_FAILURE_RETRY(open("/dev/pmsg0", O_WRONLY));
   if (pstore_fd < 0) {
+    state.SkipWithError("/dev/pmsg0");
     return;
   }
 
@@ -228,7 +225,8 @@ static void BM_pmsg_short_aligned(int iters) {
   memset(buf, 0, sizeof(buf));
   struct packet* buffer = (struct packet*)(((uintptr_t)buf + 7) & ~7);
   if (((uintptr_t)&buffer->pmsg_header) & 7) {
-    fprintf(stderr, "&buffer=0x%p iters=%d\n", &buffer->pmsg_header, iters);
+    fprintf(stderr, "&buffer=0x%p iterations=%zu\n", &buffer->pmsg_header,
+            state.iterations());
   }
 
   buffer->pmsg_header.magic = LOGGER_MAGIC;
@@ -247,15 +245,14 @@ static void BM_pmsg_short_aligned(int iters) {
   uint32_t snapshot = 0;
   buffer->payload.payload.data = htole32(snapshot);
 
-  StartBenchmarkTiming();
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     ++snapshot;
     buffer->payload.payload.data = htole32(snapshot);
     write(pstore_fd, &buffer->pmsg_header,
           sizeof(android_pmsg_log_header_t) + sizeof(android_log_header_t) +
               sizeof(android_log_event_int_t));
   }
-  StopBenchmarkTiming();
+  state.PauseTiming();
   close(pstore_fd);
 }
 BENCHMARK(BM_pmsg_short_aligned);
@@ -264,9 +261,10 @@ BENCHMARK(BM_pmsg_short_aligned);
  * Measure the time it takes to submit the android logging data to pstore
  * best case aligned single block.
  */
-static void BM_pmsg_short_unaligned1(int iters) {
+static void BM_pmsg_short_unaligned1(benchmark::State& state) {
   int pstore_fd = TEMP_FAILURE_RETRY(open("/dev/pmsg0", O_WRONLY));
   if (pstore_fd < 0) {
+    state.SkipWithError("/dev/pmsg0");
     return;
   }
 
@@ -302,7 +300,8 @@ static void BM_pmsg_short_unaligned1(int iters) {
   memset(buf, 0, sizeof(buf));
   struct packet* buffer = (struct packet*)((((uintptr_t)buf + 7) & ~7) + 1);
   if ((((uintptr_t)&buffer->pmsg_header) & 7) != 1) {
-    fprintf(stderr, "&buffer=0x%p iters=%d\n", &buffer->pmsg_header, iters);
+    fprintf(stderr, "&buffer=0x%p iterations=%zu\n", &buffer->pmsg_header,
+            state.iterations());
   }
 
   buffer->pmsg_header.magic = LOGGER_MAGIC;
@@ -321,15 +320,14 @@ static void BM_pmsg_short_unaligned1(int iters) {
   uint32_t snapshot = 0;
   buffer->payload.payload.data = htole32(snapshot);
 
-  StartBenchmarkTiming();
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     ++snapshot;
     buffer->payload.payload.data = htole32(snapshot);
     write(pstore_fd, &buffer->pmsg_header,
           sizeof(android_pmsg_log_header_t) + sizeof(android_log_header_t) +
               sizeof(android_log_event_int_t));
   }
-  StopBenchmarkTiming();
+  state.PauseTiming();
   close(pstore_fd);
 }
 BENCHMARK(BM_pmsg_short_unaligned1);
@@ -338,9 +336,10 @@ BENCHMARK(BM_pmsg_short_unaligned1);
  * Measure the time it takes to submit the android logging data to pstore
  * best case aligned single block.
  */
-static void BM_pmsg_long_aligned(int iters) {
+static void BM_pmsg_long_aligned(benchmark::State& state) {
   int pstore_fd = TEMP_FAILURE_RETRY(open("/dev/pmsg0", O_WRONLY));
   if (pstore_fd < 0) {
+    state.SkipWithError("/dev/pmsg0");
     return;
   }
 
@@ -376,7 +375,8 @@ static void BM_pmsg_long_aligned(int iters) {
   memset(buf, 0, sizeof(buf));
   struct packet* buffer = (struct packet*)(((uintptr_t)buf + 7) & ~7);
   if (((uintptr_t)&buffer->pmsg_header) & 7) {
-    fprintf(stderr, "&buffer=0x%p iters=%d\n", &buffer->pmsg_header, iters);
+    fprintf(stderr, "&buffer=0x%p iterations=%zu\n", &buffer->pmsg_header,
+            state.iterations());
   }
 
   buffer->pmsg_header.magic = LOGGER_MAGIC;
@@ -395,13 +395,12 @@ static void BM_pmsg_long_aligned(int iters) {
   uint32_t snapshot = 0;
   buffer->payload.payload.data = htole32(snapshot);
 
-  StartBenchmarkTiming();
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     ++snapshot;
     buffer->payload.payload.data = htole32(snapshot);
     write(pstore_fd, &buffer->pmsg_header, LOGGER_ENTRY_MAX_PAYLOAD);
   }
-  StopBenchmarkTiming();
+  state.PauseTiming();
   close(pstore_fd);
 }
 BENCHMARK(BM_pmsg_long_aligned);
@@ -410,9 +409,10 @@ BENCHMARK(BM_pmsg_long_aligned);
  * Measure the time it takes to submit the android logging data to pstore
  * best case aligned single block.
  */
-static void BM_pmsg_long_unaligned1(int iters) {
+static void BM_pmsg_long_unaligned1(benchmark::State& state) {
   int pstore_fd = TEMP_FAILURE_RETRY(open("/dev/pmsg0", O_WRONLY));
   if (pstore_fd < 0) {
+    state.SkipWithError("/dev/pmsg0");
     return;
   }
 
@@ -448,7 +448,8 @@ static void BM_pmsg_long_unaligned1(int iters) {
   memset(buf, 0, sizeof(buf));
   struct packet* buffer = (struct packet*)((((uintptr_t)buf + 7) & ~7) + 1);
   if ((((uintptr_t)&buffer->pmsg_header) & 7) != 1) {
-    fprintf(stderr, "&buffer=0x%p iters=%d\n", &buffer->pmsg_header, iters);
+    fprintf(stderr, "&buffer=0x%p iterations=%zu\n", &buffer->pmsg_header,
+            state.iterations());
   }
 
   buffer->pmsg_header.magic = LOGGER_MAGIC;
@@ -467,22 +468,20 @@ static void BM_pmsg_long_unaligned1(int iters) {
   uint32_t snapshot = 0;
   buffer->payload.payload.data = htole32(snapshot);
 
-  StartBenchmarkTiming();
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     ++snapshot;
     buffer->payload.payload.data = htole32(snapshot);
     write(pstore_fd, &buffer->pmsg_header, LOGGER_ENTRY_MAX_PAYLOAD);
   }
-  StopBenchmarkTiming();
+  state.PauseTiming();
   close(pstore_fd);
 }
 BENCHMARK(BM_pmsg_long_unaligned1);
 
 /*
  *      Measure the time it takes to form sprintf plus time using
- * discrete acquisition (StartBenchmarkTiming() -> StopBenchmarkTiming())
- * under light load. Expect this to be a syscall period (2us) or sprintf
- * time if zero-syscall time.
+ * discrete acquisition under light load. Expect this to be a syscall period
+ * (2us) or sprintf time if zero-syscall time.
  */
 /* helper function */
 static void test_print(const char* fmt, ...) {
@@ -498,58 +497,55 @@ static void test_print(const char* fmt, ...) {
 #define logd_sleep() usleep(50)     // really allow logd to catch up
 
 /* performance test */
-static void BM_sprintf_overhead(int iters) {
-  for (int i = 0; i < iters; ++i) {
-    StartBenchmarkTiming();
-    test_print("BM_sprintf_overhead:%d", i);
-    StopBenchmarkTiming();
+static void BM_sprintf_overhead(benchmark::State& state) {
+  while (state.KeepRunning()) {
+    test_print("BM_sprintf_overhead:%zu", state.iterations());
+    state.PauseTiming();
     logd_yield();
+    state.ResumeTiming();
   }
 }
 BENCHMARK(BM_sprintf_overhead);
 
 /*
  *      Measure the time it takes to submit the android printing logging call
- * using discrete acquisition discrete acquisition (StartBenchmarkTiming() ->
- * StopBenchmarkTiming()) under light load. Expect this to be a dozen or so
- * syscall periods (40us) plus time to run *printf
+ * using discrete acquisition discrete acquisition under light load. Expect
+ * this to be a dozen or so syscall periods (40us) plus time to run *printf
  */
-static void BM_log_print_overhead(int iters) {
-  for (int i = 0; i < iters; ++i) {
-    StartBenchmarkTiming();
-    __android_log_print(ANDROID_LOG_INFO, "BM_log_overhead", "%d", i);
-    StopBenchmarkTiming();
+static void BM_log_print_overhead(benchmark::State& state) {
+  while (state.KeepRunning()) {
+    __android_log_print(ANDROID_LOG_INFO, "BM_log_overhead", "%zu",
+                        state.iterations());
+    state.PauseTiming();
     logd_yield();
+    state.ResumeTiming();
   }
 }
 BENCHMARK(BM_log_print_overhead);
 
 /*
  *      Measure the time it takes to submit the android event logging call
- * using discrete acquisition (StartBenchmarkTiming() -> StopBenchmarkTiming())
- * under light load. Expect this to be a long path to logger to convert the
- * unknown tag (0) into a tagname (less than 200us).
+ * using discrete acquisition under light load. Expect this to be a long path
+ * to logger to convert the unknown tag (0) into a tagname (less than 200us).
  */
-static void BM_log_event_overhead(int iters) {
-  for (unsigned long long i = 0; i < (unsigned)iters; ++i) {
-    StartBenchmarkTiming();
+static void BM_log_event_overhead(benchmark::State& state) {
+  for (int64_t i = 0; state.KeepRunning(); ++i) {
     // log tag number 0 is not known, nor shall it ever be known
     __android_log_btwrite(0, EVENT_TYPE_LONG, &i, sizeof(i));
-    StopBenchmarkTiming();
+    state.PauseTiming();
     logd_yield();
+    state.ResumeTiming();
   }
 }
 BENCHMARK(BM_log_event_overhead);
 
 /*
  *      Measure the time it takes to submit the android event logging call
- * using discrete acquisition (StartBenchmarkTiming() -> StopBenchmarkTiming())
- * under light load with a known logtag.  Expect this to be a dozen or so
- * syscall periods (less than 40us)
+ * using discrete acquisition under light load with a known logtag.  Expect
+ * this to be a dozen or so syscall periods (less than 40us)
  */
-static void BM_log_event_overhead_42(int iters) {
-  for (unsigned long long i = 0; i < (unsigned)iters; ++i) {
-    StartBenchmarkTiming();
+static void BM_log_event_overhead_42(benchmark::State& state) {
+  for (int64_t i = 0; state.KeepRunning(); ++i) {
     // In system/core/logcat/event.logtags:
     // # These are used for testing, do not modify without updating
     // # tests/framework-tests/src/android/util/EventLogFunctionalTest.java.
@@ -557,40 +553,42 @@ static void BM_log_event_overhead_42(int iters) {
     // # system/core/liblog/tests/liblog_test.cpp
     // 42    answer (to life the universe etc|3)
     __android_log_btwrite(42, EVENT_TYPE_LONG, &i, sizeof(i));
-    StopBenchmarkTiming();
+    state.PauseTiming();
     logd_yield();
+    state.ResumeTiming();
   }
 }
 BENCHMARK(BM_log_event_overhead_42);
 
-static void BM_log_event_overhead_null(int iters) {
+static void BM_log_event_overhead_null(benchmark::State& state) {
   set_log_null();
-  BM_log_event_overhead(iters);
+  BM_log_event_overhead(state);
   set_log_default();
 }
 BENCHMARK(BM_log_event_overhead_null);
 
 /*
  *      Measure the time it takes to submit the android event logging call
- * using discrete acquisition (StartBenchmarkTiming() -> StopBenchmarkTiming())
- * under very-light load (<1% CPU utilization).
+ * using discrete acquisition under very-light load (<1% CPU utilization).
  */
-static void BM_log_light_overhead(int iters) {
-  for (unsigned long long i = 0; i < (unsigned)iters; ++i) {
-    StartBenchmarkTiming();
+static void BM_log_light_overhead(benchmark::State& state) {
+  for (int64_t i = 0; state.KeepRunning(); ++i) {
     __android_log_btwrite(0, EVENT_TYPE_LONG, &i, sizeof(i));
-    StopBenchmarkTiming();
+    state.PauseTiming();
     usleep(10000);
+    state.ResumeTiming();
   }
 }
 BENCHMARK(BM_log_light_overhead);
 
-static void BM_log_light_overhead_null(int iters) {
+static void BM_log_light_overhead_null(benchmark::State& state) {
   set_log_null();
-  BM_log_light_overhead(iters);
+  BM_log_light_overhead(state);
   set_log_default();
 }
-BENCHMARK(BM_log_light_overhead_null);
+// Default gets out of hand for this test, so we set a reasonable number of
+// iterations for a timely result.
+BENCHMARK(BM_log_light_overhead_null)->Iterations(500);
 
 static void caught_latency(int /*signum*/) {
   unsigned long long v = 0xDEADBEEFA55A5AA5ULL;
@@ -614,10 +612,12 @@ static const int alarm_time = 3;
 
 /*
  *      Measure the time it takes for the logd posting call to acquire the
- * timestamp to place into the internal record. Expect this to be less than
- * 4 syscalls (3us).
+ * timestamp to place into the internal record.  Expect this to be less than
+ * 4 syscalls (3us).  This test uses manual injection of timing because it is
+ * comparing the timestamp at send, and then picking up the corresponding log
+ * end-to-end long path from logd to see what actual timestamp was submitted.
  */
-static void BM_log_latency(int iters) {
+static void BM_log_latency(benchmark::State& state) {
   pid_t pid = getpid();
 
   struct logger_list* logger_list =
@@ -631,7 +631,8 @@ static void BM_log_latency(int iters) {
   signal(SIGALRM, caught_latency);
   alarm(alarm_time);
 
-  for (int j = 0, i = 0; i < iters && j < 10 * iters; ++i, ++j) {
+  for (size_t j = 0; state.KeepRunning() && j < 10 * state.iterations(); ++j) {
+  retry:  // We allow transitory errors (logd overloaded) to be retried.
     log_time ts;
     LOG_FAILURE_RETRY((ts = log_time(CLOCK_REALTIME),
                        android_btWriteLog(0, EVENT_TYPE_LONG, &ts, sizeof(ts))));
@@ -642,7 +643,7 @@ static void BM_log_latency(int iters) {
       alarm(alarm_time);
 
       if (ret <= 0) {
-        iters = i;
+        state.SkipWithError("android_logger_list_read");
         break;
       }
       if ((log_msg.entry.len != (4 + 1 + 8)) ||
@@ -658,7 +659,7 @@ static void BM_log_latency(int iters) {
       log_time tx(eventData + 4 + 1);
       if (ts != tx) {
         if (0xDEADBEEFA55A5AA5ULL == caught_convert(eventData + 4 + 1)) {
-          iters = i;
+          state.SkipWithError("signal");
           break;
         }
         continue;
@@ -666,12 +667,8 @@ static void BM_log_latency(int iters) {
 
       uint64_t start = ts.nsec();
       uint64_t end = log_msg.nsec();
-      if (end >= start) {
-        StartBenchmarkTiming(start);
-        StopBenchmarkTiming(end);
-      } else {
-        --i;
-      }
+      if (end < start) goto retry;
+      state.SetIterationTime((end - start) / (double)NS_PER_SEC);
       break;
     }
   }
@@ -681,7 +678,9 @@ static void BM_log_latency(int iters) {
 
   android_logger_list_free(logger_list);
 }
-BENCHMARK(BM_log_latency);
+// Default gets out of hand for this test, so we set a reasonable number of
+// iterations for a timely result.
+BENCHMARK(BM_log_latency)->UseManualTime()->Iterations(200);
 
 static void caught_delay(int /*signum*/) {
   unsigned long long v = 0xDEADBEEFA55A5AA6ULL;
@@ -693,7 +692,7 @@ static void caught_delay(int /*signum*/) {
  *      Measure the time it takes for the logd posting call to make it into
  * the logs. Expect this to be less than double the process wakeup time (2ms).
  */
-static void BM_log_delay(int iters) {
+static void BM_log_delay(benchmark::State& state) {
   pid_t pid = getpid();
 
   struct logger_list* logger_list =
@@ -707,9 +706,7 @@ static void BM_log_delay(int iters) {
   signal(SIGALRM, caught_delay);
   alarm(alarm_time);
 
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     log_time ts(CLOCK_REALTIME);
 
     LOG_FAILURE_RETRY(android_btWriteLog(0, EVENT_TYPE_LONG, &ts, sizeof(ts)));
@@ -720,7 +717,7 @@ static void BM_log_delay(int iters) {
       alarm(alarm_time);
 
       if (ret <= 0) {
-        iters = i;
+        state.SkipWithError("android_logger_list_read");
         break;
       }
       if ((log_msg.entry.len != (4 + 1 + 8)) ||
@@ -736,7 +733,7 @@ static void BM_log_delay(int iters) {
       log_time tx(eventData + 4 + 1);
       if (ts != tx) {
         if (0xDEADBEEFA55A5AA6ULL == caught_convert(eventData + 4 + 1)) {
-          iters = i;
+          state.SkipWithError("signal");
           break;
         }
         continue;
@@ -745,12 +742,11 @@ static void BM_log_delay(int iters) {
       break;
     }
   }
+  state.PauseTiming();
 
   signal(SIGALRM, SIG_DFL);
   alarm(0);
 
-  StopBenchmarkTiming();
-
   android_logger_list_free(logger_list);
 }
 BENCHMARK(BM_log_delay);
@@ -758,45 +754,33 @@ BENCHMARK(BM_log_delay);
 /*
  *      Measure the time it takes for __android_log_is_loggable.
  */
-static void BM_is_loggable(int iters) {
+static void BM_is_loggable(benchmark::State& state) {
   static const char logd[] = "logd";
 
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     __android_log_is_loggable_len(ANDROID_LOG_WARN, logd, strlen(logd),
                                   ANDROID_LOG_VERBOSE);
   }
-
-  StopBenchmarkTiming();
 }
 BENCHMARK(BM_is_loggable);
 
 /*
  *      Measure the time it takes for android_log_clockid.
  */
-static void BM_clockid(int iters) {
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
+static void BM_clockid(benchmark::State& state) {
+  while (state.KeepRunning()) {
     android_log_clockid();
   }
-
-  StopBenchmarkTiming();
 }
 BENCHMARK(BM_clockid);
 
 /*
  *      Measure the time it takes for __android_log_security.
  */
-static void BM_security(int iters) {
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
+static void BM_security(benchmark::State& state) {
+  while (state.KeepRunning()) {
     __android_log_security();
   }
-
-  StopBenchmarkTiming();
 }
 BENCHMARK(BM_security);
 
@@ -824,21 +808,17 @@ static bool prechargeEventMap() {
 /*
  *      Measure the time it takes for android_lookupEventTag_len
  */
-static void BM_lookupEventTag(int iters) {
+static void BM_lookupEventTag(benchmark::State& state) {
   prechargeEventMap();
 
   std::unordered_set<uint32_t>::const_iterator it = set.begin();
 
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     size_t len;
     android_lookupEventTag_len(map, &len, (*it));
     ++it;
     if (it == set.end()) it = set.begin();
   }
-
-  StopBenchmarkTiming();
 }
 BENCHMARK(BM_lookupEventTag);
 
@@ -847,7 +827,7 @@ BENCHMARK(BM_lookupEventTag);
  */
 static uint32_t notTag = 1;
 
-static void BM_lookupEventTag_NOT(int iters) {
+static void BM_lookupEventTag_NOT(benchmark::State& state) {
   prechargeEventMap();
 
   while (set.find(notTag) != set.end()) {
@@ -855,15 +835,11 @@ static void BM_lookupEventTag_NOT(int iters) {
     if (notTag >= USHRT_MAX) notTag = 1;
   }
 
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     size_t len;
     android_lookupEventTag_len(map, &len, notTag);
   }
 
-  StopBenchmarkTiming();
-
   ++notTag;
   if (notTag >= USHRT_MAX) notTag = 1;
 }
@@ -872,42 +848,38 @@ BENCHMARK(BM_lookupEventTag_NOT);
 /*
  *      Measure the time it takes for android_lookupEventFormat_len
  */
-static void BM_lookupEventFormat(int iters) {
+static void BM_lookupEventFormat(benchmark::State& state) {
   prechargeEventMap();
 
   std::unordered_set<uint32_t>::const_iterator it = set.begin();
 
-  StartBenchmarkTiming();
-
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     size_t len;
     android_lookupEventFormat_len(map, &len, (*it));
     ++it;
     if (it == set.end()) it = set.begin();
   }
-
-  StopBenchmarkTiming();
 }
 BENCHMARK(BM_lookupEventFormat);
 
 /*
  *      Measure the time it takes for android_lookupEventTagNum plus above
  */
-static void BM_lookupEventTagNum(int iters) {
+static void BM_lookupEventTagNum(benchmark::State& state) {
   prechargeEventMap();
 
   std::unordered_set<uint32_t>::const_iterator it = set.begin();
 
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     size_t len;
     const char* name = android_lookupEventTag_len(map, &len, (*it));
     std::string Name(name, len);
     const char* format = android_lookupEventFormat_len(map, &len, (*it));
     std::string Format(format, len);
-    StartBenchmarkTiming();
+    state.ResumeTiming();
     android_lookupEventTagNum(map, Name.c_str(), Format.c_str(),
                               ANDROID_LOG_UNKNOWN);
-    StopBenchmarkTiming();
+    state.PauseTiming();
     ++it;
     if (it == set.end()) it = set.begin();
   }
@@ -943,7 +915,7 @@ static void send_to_control(char* buf, size_t len) {
   close(sock);
 }
 
-static void BM_lookupEventTagNum_logd_new(int iters) {
+static void BM_lookupEventTagNum_logd_new(benchmark::State& state) {
   fprintf(stderr,
           "WARNING: "
           "This test can cause logd to grow in size and hit DOS limiter\n");
@@ -965,7 +937,7 @@ static void BM_lookupEventTagNum_logd_new(int iters) {
     data_event_log_tags = empty_event_log_tags;
   }
 
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     char buffer[256];
     memset(buffer, 0, sizeof(buffer));
     log_time now(CLOCK_MONOTONIC);
@@ -973,9 +945,9 @@ static void BM_lookupEventTagNum_logd_new(int iters) {
     snprintf(name, sizeof(name), "a%" PRIu64, now.nsec());
     snprintf(buffer, sizeof(buffer), "getEventTag name=%s format=\"(new|1)\"",
              name);
-    StartBenchmarkTiming();
+    state.ResumeTiming();
     send_to_control(buffer, sizeof(buffer));
-    StopBenchmarkTiming();
+    state.PauseTiming();
   }
 
   // Restore copies (logd still know about them, until crash or reboot)
@@ -1002,12 +974,12 @@ static void BM_lookupEventTagNum_logd_new(int iters) {
 }
 BENCHMARK(BM_lookupEventTagNum_logd_new);
 
-static void BM_lookupEventTagNum_logd_existing(int iters) {
+static void BM_lookupEventTagNum_logd_existing(benchmark::State& state) {
   prechargeEventMap();
 
   std::unordered_set<uint32_t>::const_iterator it = set.begin();
 
-  for (int i = 0; i < iters; ++i) {
+  while (state.KeepRunning()) {
     size_t len;
     const char* name = android_lookupEventTag_len(map, &len, (*it));
     std::string Name(name, len);
@@ -1018,9 +990,9 @@ static void BM_lookupEventTagNum_logd_existing(int iters) {
     snprintf(buffer, sizeof(buffer), "getEventTag name=%s format=\"%s\"",
              Name.c_str(), Format.c_str());
 
-    StartBenchmarkTiming();
+    state.ResumeTiming();
     send_to_control(buffer, sizeof(buffer));
-    StopBenchmarkTiming();
+    state.PauseTiming();
     ++it;
     if (it == set.end()) it = set.begin();
   }