[nigrosimone]

AsyncHooks::pop_async_context() calls shrink_to_fit() on native_execution_async_resources_ after every pop. With the MSVC and libc++ standard libraries, shrink_to_fit() on an empty std::deque frees its storage, so every top-level callback scope allocates it again on push and frees it on pop. libstdc++ does nothing here, so on Linux nothing change.

Now the deque keeps its storage, like the async_ids_stack_ next to it, which never shrinks.

Until v24 the shrink ran only when the size was over 16 and under half of the capacity, so only after a deep nesting. Since #56457 it runs on every pop.

The same pattern in a small program, on Window with MSVC: 136-147 ns per callback with shrink_to_fit(), 6 ns without. macOS uses libc++, which frees the empty deque in the same way (not measured).

#include <chrono>
#include <cstdio>
#include <deque>
#include <variant>

std::deque<std::variant<int*, long*>> resources;
int value;

// Same as push_async_context() and pop_async_context() at depth 0.
__declspec(noinline) void Push() { resources.resize(1); resources[0] = &value; }
__declspec(noinline) void Pop() { resources.resize(0); resources.shrink_to_fit(); }

int main() {
  const int n = 5000000;
  auto start = std::chrono::steady_clock::now();
  for (int i = 0; i < n; i++) { Push(); Pop(); }
  auto end = std::chrono::steady_clock::now();
  printf("%.1f ns\n", std::chrono::duration<double, std::nano>(end - start).count() / n);
}

Refs: #66316
Refs: nodejs/performance#24

Disclosure: I used Opus 5.5 (Max) as coding assistant

[github-actions]

Welcome to Node.js, and thank you for your first contribution!

Before review, please take a moment to read:

• the guide for first-time contributors
• the contribution and automation policies
• the pull request guide
• the AI use policy
• the Code of Conduct

Please make sure every commit is signed off. For a first pull request, GitHub Actions require collaborator approval and Jenkins CI must be started by a collaborator or triager, so an initial wait is normal.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.50%. Comparing base (3e5c8af) to head (a979326).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #66344      +/-   ##
==========================================
- Coverage   92.86%   90.50%   -2.36%     
==========================================
  Files         428      797     +369     
  Lines      195100   278361   +83261     
  Branches    30147    53500   +23353     
==========================================
+ Hits       181175   251936   +70761     
- Misses      13597    16809    +3212     
- Partials      328     9616    +9288     

Files with missing lines	Coverage Δ
src/env.cc	84.42% <ø> (ø)

... and 501 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/78166/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/78180/

[nigrosimone]

@anonrig CI is green and this has your approval for more than 7 days, could you add commit-queue?

[gurgunday]

I like this change, however this means on Windows it will never shrink

I am still approving because async_ids_stack_, sitting right next to it, already keeps 16 bytes per level forever, on every platform, so we're already making similar tradeoffs

[nigrosimone]

Rebased on main after #66395 landed, it touched the same lines in pop_async_context(). The change is still only the shrink_to_fit() line.

@anonrig @gurgunday could one of you start a CI and add commit-queue?