[components][pthreads] fix cond wait scheduling#11533
Conversation
|
👋 感谢您对 RT-Thread 的贡献！Thank you for your contribution to RT-Thread! 为确保代码符合 RT-Thread 的编码规范，请在你的仓库中执行以下步骤运行代码格式化工作流（如果格式化CI运行失败）。 🛠 操作步骤 | Steps
完成后，提交将自动更新至 如有问题欢迎联系我们，再次感谢您的贡献！💐 |
📌 Code Review Assignment🏷️ Tag: componentsReviewers: @Maihuanyi Changed Files (Click to expand)
🏷️ Tag: components_libcReviewers: @GorrayLi @mysterywolf Changed Files (Click to expand)
📊 Current Review Status (Last Updated: 2026-06-30 09:54 CST)
📝 Review Instructions
|
|
2 participants
What changed
Add the missing scheduler invocation in
_pthread_cond_timedwait()after the current thread has been suspended, the pthread mutex has been released, and the critical section has been exited.Root cause
The wait path put the current thread onto the condition semaphore suspend list, but only called
rt_exit_critical()before readingthread->errorand re-locking the mutex. Withoutrt_schedule(), the suspended thread could continue running instead of yielding to the scheduler, which breaks pthread condition wait users such as pthread read/write locks.Scope
This keeps the fix in the condition variable wait implementation, which is the layer that directly suspends the thread. It does not add duplicate scheduling behavior in pthread rwlock callers.
Closes #11532
Validation
git diff --checkgh search prs 'pthread_cond_timedwait rt_schedule' --repo RT-Thread/rt-thread --state open --limit 20 --json number,title,author,url,isDraftreturned no duplicate open PRs.python -m SCons -non pthread-enabled BSPs (bsp/mipssim,bsp/microchip/samd51-adafruit-metro-m4), but local cross-toolchain paths are not installed, so BSP build validation could not complete locally.