Signal workers for priority tasks

[chenBright]

What problem does this PR solve?

Issue Number: resolve N/A

Problem Summary:

Tasks placed into a global priority queue did not signal worker threads after being enqueued. If all eligible workers were sleeping, these tasks could remain pending until another event woke a worker.

What is changed and the side effects?

Changed:

Signal one worker after adding a task to the global priority queue. The signal uses the task's tag so that a worker associated with the correct tag is notified.

Side effects:

• Performance effects:

• Breaking backward compatibility:

---

Check List:

• Please make sure your changes are compilable.
• When providing us with a new feature, it is best to add related tests.
• Please follow Contributor Covenant Code of Conduct.

[Copilot]

Pull request overview

This PR addresses a scheduling gap in bthread’s global priority queue path: after enqueuing a task into the global per-tag priority queue, no worker was explicitly signaled, so eligible workers could remain asleep and leave priority work pending until another wake-up event occurred.

Changes:

• Enqueueing into the global ED priority queue now additionally signals one worker for the task’s tag.
• The priority-queue unit test now keeps track of started urgent children and joins them to ensure completion.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/bthread/task_group.cpp	Signals a worker after pushing a task into the global per-tag priority queue to avoid stalled priority work.
test/bthread_priority_queue_unittest.cpp	Joins urgent child bthreads in a priority-queue test to ensure spawned tasks complete before returning.

---

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

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

Suppressed comments (2)

src/bthread/task_group.cpp:1026

• The new wakeup in priority_to_run() always calls signal_task() even when the BTHREAD_NOSIGNAL flag was set on the newly created bthread (passed through args->nosignal). That changes the documented NOSIGNAL behavior (“worker pthreads are not notified”) and can defeat batching/flush semantics.

    g->control()->push_ed_priority_queue(
        args->tag, args->meta->priority_index, args->meta->tid);

    ++g->_nsignaled;
    g->control()->signal_task(1, args->tag);

src/bthread/task_group.cpp:1025

• There isn’t a regression test that specifically exercises the original failure mode described in the PR (all eligible workers parked, then an ED is enqueued into the global priority queue and must wake a worker). The current unit tests create/execute priority work, but they don’t assert the wakeup behavior from the ED priority queue under idle-worker conditions, so this change could regress without being caught.

    ++g->_nsignaled;
    g->control()->signal_task(1, args->tag);