epmd: Fix regression introduced in e522e738e

[jhogberg]

Fixes #11402

[github-actions]

CT Test Results

  2 files   12 suites   7m 57s ⏱️
129 tests 123 ✅ 6 💤 0 ❌
145 runs  139 ✅ 6 💤 0 ❌

Results for commit e3d12c3.

♻️ This comment has been updated with latest results.

To speed up review, make sure that you have read Contributing to Erlang/OTP and that all checks pass.

See the TESTING and DEVELOPMENT HowTo guides for details about how to run test locally.

Artifacts

• Complete CT logs (Download Logs)
• HTML Documentation (Download HTML Docs)
• Windows Installer

// Erlang/OTP Github Action Bot

[Copilot]

Pull request overview

Fixes an epmd regression where binding an implicitly-added IPv6 (or IPv4) loopback socket could cause epmd to exit on systems where that loopback/protocol stack is unavailable (e.g., IPv6 disabled), even when users explicitly configure an IPv4-only address like ERL_EPMD_ADDRESS=127.0.0.1.

Changes:

• Reworks “nonfatal bind failure” tracking from an index threshold to a per-socket bitmask.
• Marks implicitly-added loopback sockets (v4/v6) as nonfatal so bind failures are logged but do not terminate epmd.
• Updates the bind failure path to consult the nonfatal socket bitmask and continue after closing the failed socket.

Comments suppressed due to low confidence (2)

erts/epmd/src/epmd_srv.c:389

• Use an unsigned shift when building the nonfatal socket bitmask to avoid signed left-shift pitfalls and make the intent explicit.

          nonfatal_sockets |= (1 << num_sockets);

erts/epmd/src/epmd_srv.c:513

• Use an unsigned shift when checking the nonfatal socket bitmask to avoid signed left-shift pitfalls and make the intent explicit.

              if (!(nonfatal_sockets & (1 << i))) {

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.