b9732

ggml-org/llama.cppb9732Jun 19, 2026by github-actions[bot]same day as b9731

What changed since b9731

low risk

Server internals refactored: child-to-router communication reworked with wakeup fixes and improved status updates.

Improvements

  • Refactored the child process to router communication path in the server for cleaner internal messaging.
  • Improved the update_status() logic used by the router to track child state.
  • Fixed wakeup handling cases to make inter-process signaling more reliable.
  • Added documentation covering the new child-to-router communication design.

Bug fixes

  • Corrected wakeup edge cases that could leave child processes or the router waiting on signals.

Upgrade notes

This is an internal server refactor with no documented config or API changes; existing flags and endpoints should continue to work. Validate multi-process/router deployments (e.g., parallel slots or child spawning) after upgrading, since the messaging path changed. No migration steps are required.

Analyzed Oct 24, 2026

Upgrading?

Release notes summary

Refactored the internal communication mechanism between child processes and the router in the server architecture.

Key Highlights

  • Refactored communication between child processes and the router
  • Fixed wakeup cases for better reliability
  • Improved update_status() function

Release Notes (excerpt)

<details open> server: refactor child --> router communication (#24821) * server: refactor child --> router communication * fix wakeup case * add docs * improve update_status() * nits </details> **macOS/iOS:** - macOS Apple Silicon (arm64) - macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED - macOS Intel (x64) - iOS XCFramework **Linux:** - Ubuntu x64 (CPU) - Ubuntu arm64 (CPU) - Ubuntu s390x (CPU) - Ubuntu x64 (Vulkan) - Ubuntu arm64 (Vulkan) - Ubuntu x64 (ROCm 7.2)…

Published by ggml-org on GitHub.

Read the full release notes

Related Repositories