perf_counters: Initialize once only when needed (#1656)

* perf_counters: Initialize once only when needed

This works around some performance problems running Android under QEMU.
Calling `pfm_initialize` was very slow, and was called during dynamic
initialization (before `main` or when loaded as a shared library).
This happened whenever benchmark was linked, even if no benchmarks
were run.

Instead, call `pfm_initialize` at most once, and only when one of:
1. `PerfCounters::Initialize` is called
2. `PerfCounters::Create` is called with a non-empty counter list
3. `PerfCounters::IsCounterSupported` is called

The return value of the first `pfm_initialize()` is saved and
returned from all subsequent `PerfCounters::Initialize` calls.

* perf_counters: Make success var const

* InitLibPfmOnce: Inline function

2 files changed, 14 insertions, 3 deletions

diff --git a/src/perf_counters.cc b/src/perf_counters.cc
index 3980ea0..417acdb 100644
--- a/src/perf_counters.cc
+++ b/src/perf_counters.cc
@@ -57,9 +57,18 @@ size_t PerfCounterValues::Read(const std::vector<int>& leaders) {
 
 const bool PerfCounters::kSupported = true;
 
-bool PerfCounters::Initialize() { return pfm_initialize() == PFM_SUCCESS; }
+// Initializes libpfm only on the first call.  Returns whether that single
+// initialization was successful.
+bool PerfCounters::Initialize() {
+  // Function-scope static gets initialized only once on first call.
+  static const bool success = []() {
+    return pfm_initialize() == PFM_SUCCESS;
+  }();
+  return success;
+}
 
 bool PerfCounters::IsCounterSupported(const std::string& name) {
+  Initialize();
   perf_event_attr_t attr;
   std::memset(&attr, 0, sizeof(attr));
   pfm_perf_encode_arg_t arg;
@@ -73,6 +82,10 @@ bool PerfCounters::IsCounterSupported(const std::string& name) {
 
 PerfCounters PerfCounters::Create(
     const std::vector<std::string>& counter_names) {
+  if (!counter_names.empty()) {
+    Initialize();
+  }
+
   // Valid counters will populate these arrays but we start empty
   std::vector<std::string> valid_names;
   std::vector<int> counter_ids;
diff --git a/src/perf_counters.h b/src/perf_counters.h
index 152a6f2..bf5eb6b 100644
--- a/src/perf_counters.h
+++ b/src/perf_counters.h
@@ -190,8 +190,6 @@ class BENCHMARK_EXPORT PerfCountersMeasurement final {
   PerfCounterValues end_values_;
 };
 
-BENCHMARK_UNUSED static bool perf_init_anchor = PerfCounters::Initialize();
-
 }  // namespace internal
 }  // namespace benchmark