async_hooks: use typed array stack as fast path

- Communicate the current async stack length through a
  typed array field rather than a native binding method
- Add a new fixed-size `async_ids_fast_stack` typed array
  that contains the async ID stack up to a fixed limit.
  This increases performance noticeably, since most of the time
  the async ID stack will not be more than a handful of
  levels deep.
- Make the JS `pushAsyncIds()` and `popAsyncIds()` functions
  do the same thing as the native ones if the fast path
  is applicable.

Benchmarks:

    $ ./node benchmark/compare.js --new ./node --old ./node-master --runs 10 --filter next-tick process | Rscript benchmark/compare.R
    [00:03:25|% 100| 6/6 files | 20/20 runs | 1/1 configs]: Done
                                                   improvement confidence      p.value
     process/next-tick-breadth-args.js millions=4     19.72 %        *** 3.013913e-06
     process/next-tick-breadth.js millions=4          27.33 %        *** 5.847983e-11
     process/next-tick-depth-args.js millions=12      40.08 %        *** 1.237127e-13
     process/next-tick-depth.js millions=12           77.27 %        *** 1.413290e-11
     process/next-tick-exec-args.js millions=5        13.58 %        *** 1.245180e-07
     process/next-tick-exec.js millions=5             16.80 %        *** 2.961386e-07

PR-URL: https://github.com/nodejs/node/pull/17780
Reviewed-By: James M Snell <jasnell@gmail.com>

2 files changed, 44 insertions, 6 deletions

diff --git a/lib/internal/async_hooks.js b/lib/internal/async_hooks.js
index 035dd0634d..7d99985dd1 100644
--- a/lib/internal/async_hooks.js
+++ b/lib/internal/async_hooks.js
@@ -19,6 +19,12 @@ const async_wrap = process.binding('async_wrap');
  *    retrieving the triggerAsyncId value is passing directly to the
  *    constructor -> value set in kDefaultTriggerAsyncId -> executionAsyncId of
  *    the current resource.
+ *
+ * async_ids_fast_stack is a Float64Array that contains part of the async ID
+ * stack. Each pushAsyncIds() call adds two doubles to it, and each
+ * popAsyncIds() call removes two doubles from it.
+ * It has a fixed size, so if that is exceeded, calls to the native
+ * side are used instead in pushAsyncIds() and popAsyncIds().
  */
 const { async_hook_fields, async_id_fields } = async_wrap;
 // Store the pair executionAsyncId and triggerAsyncId in a std::stack on
@@ -26,7 +32,7 @@ const { async_hook_fields, async_id_fields } = async_wrap;
 // current execution stack. This is unwound as each resource exits. In the case
 // of a fatal exception this stack is emptied after calling each hook's after()
 // callback.
-const { pushAsyncIds, popAsyncIds } = async_wrap;
+const { pushAsyncIds: pushAsyncIds_, popAsyncIds: popAsyncIds_ } = async_wrap;
 // For performance reasons, only track Proimses when a hook is enabled.
 const { enablePromiseHook, disablePromiseHook } = async_wrap;
 // Properties in active_hooks are used to keep track of the set of hooks being
@@ -60,8 +66,8 @@ const active_hooks = {
 // async execution. These are tracked so if the user didn't include callbacks
 // for a given step, that step can bail out early.
 const { kInit, kBefore, kAfter, kDestroy, kPromiseResolve,
-        kCheck, kExecutionAsyncId, kAsyncIdCounter,
-        kDefaultTriggerAsyncId } = async_wrap.constants;
+        kCheck, kExecutionAsyncId, kAsyncIdCounter, kTriggerAsyncId,
+        kDefaultTriggerAsyncId, kStackLength } = async_wrap.constants;
 
 // Used in AsyncHook and AsyncResource.
 const init_symbol = Symbol('init');
@@ -329,6 +335,38 @@ function emitDestroyScript(asyncId) {
 }
 
 
+// This is the equivalent of the native push_async_ids() call.
+function pushAsyncIds(asyncId, triggerAsyncId) {
+  const offset = async_hook_fields[kStackLength];
+  if (offset * 2 >= async_wrap.async_ids_stack.length)
+    return pushAsyncIds_(asyncId, triggerAsyncId);
+  async_wrap.async_ids_stack[offset * 2] = async_id_fields[kExecutionAsyncId];
+  async_wrap.async_ids_stack[offset * 2 + 1] = async_id_fields[kTriggerAsyncId];
+  async_hook_fields[kStackLength]++;
+  async_id_fields[kExecutionAsyncId] = asyncId;
+  async_id_fields[kTriggerAsyncId] = triggerAsyncId;
+}
+
+
+// This is the equivalent of the native pop_async_ids() call.
+function popAsyncIds(asyncId) {
+  if (async_hook_fields[kStackLength] === 0) return false;
+  const stackLength = async_hook_fields[kStackLength];
+
+  if (async_hook_fields[kCheck] > 0 &&
+      async_id_fields[kExecutionAsyncId] !== asyncId) {
+    // Do the same thing as the native code (i.e. crash hard).
+    return popAsyncIds_(asyncId);
+  }
+
+  const offset = stackLength - 1;
+  async_id_fields[kExecutionAsyncId] = async_wrap.async_ids_stack[2 * offset];
+  async_id_fields[kTriggerAsyncId] = async_wrap.async_ids_stack[2 * offset + 1];
+  async_hook_fields[kStackLength] = offset;
+  return offset > 0;
+}
+
+
 module.exports = {
   // Private API
   getHookArrays,
diff --git a/lib/internal/bootstrap_node.js b/lib/internal/bootstrap_node.js
index f8a1fa3961..298fc9df30 100644
--- a/lib/internal/bootstrap_node.js
+++ b/lib/internal/bootstrap_node.js
@@ -367,9 +367,9 @@
     // Arrays containing hook flags and ids for async_hook calls.
     const { async_hook_fields, async_id_fields } = async_wrap;
     // Internal functions needed to manipulate the stack.
-    const { clearAsyncIdStack, asyncIdStackSize } = async_wrap;
+    const { clearAsyncIdStack } = async_wrap;
     const { kAfter, kExecutionAsyncId,
-            kDefaultTriggerAsyncId } = async_wrap.constants;
+            kDefaultTriggerAsyncId, kStackLength } = async_wrap.constants;
 
     process._fatalException = function(er) {
       var caught;
@@ -407,7 +407,7 @@
           do {
             NativeModule.require('internal/async_hooks').emitAfter(
               async_id_fields[kExecutionAsyncId]);
-          } while (asyncIdStackSize() > 0);
+          } while (async_hook_fields[kStackLength] > 0);
         // Or completely empty the id stack.
         } else {
           clearAsyncIdStack();