[LeandroLP23]

What

On Windows, setitimer and alarm expiries land on the host's coarse tick. A 2 ms one-shot ITIMER_REAL timer delivers SIGALRM after 15.9 ms (median of 31). The alarm thread (RunAlarms in core/libs/prx/libkernel/System/src/Signals.cpp) waits on a std::condition_variable, and MinGW's wait_until goes through winpthreads and its GetTickCount64-based timeout. This is another call site of #1471. #11167 left it out of the sigtimedwait change as a separate site, and this PR does it the same way.

The change:

• AlarmState::changed is a TimedWait::Condition, the waiter sigtimedwait, event flags, equeues and semaphores already use. alarm and setitimer wake it with NotifyAll.
• RunAlarms keeps the steady_clock deadline that alarm, setitimer and getitimer share. When the deadline is still ahead, it waits with WaitUntil for the remaining time, converted to a TimedWait::NowNanos() deadline, then loops. It expires the timer only once steady_clock has reached the deadline, so a wake from a re-arm, or the two clocks disagreeing by a few microseconds, never fires it early.
• A periodic timer still re-arms by whole intervals past the deadline it missed. The now < deadline branch of the old period count can no longer be reached, so it is gone.

Tested

Windows 11, Intel Core i5-10600K, WinLibs GCC 15.2.0 r7, Release, full build:

• New guest_itimer_short_expiry (guest_itimer_tests short-expiry): with SIGALRM blocked, it arms a 2 ms one-shot ITIMER_REAL timer 31 times and takes each expiry with sigtimedwait. No expiry may come earlier than 2 ms after arming, and the median must stay under 12 ms, as in fix(libkernel): wait for sigtimedwait timeouts with the precise timed wait on Windows #11167's sigtimedwait test.
• With main's Signals.cpp it fails: median 15858000 ns.
• With the change it passes 10 of 10 (--repeat until-fail:10), with medians between 2008700 and 2017400 ns.
• The 53 tests matching sig|signal|pause|alarm|itimer|pselect|exception|raise pass with --repeat until-fail:3.

Linux and macOS: not run. There TimedWait::Condition uses its condition_variable_any path, which waits the same way the old code did.

Tested on main 5bc84038c4, then rebased onto 801d5b9bc1, which doesn't touch these files. Overlaps are textual only: #9389, #10749 and #11002 edit other parts of Signals.cpp, and none touches the alarm code.

The other guest-visible waits #1471 still lists are AIO (#11416) and libSceUsbd (#9339), both in their own PRs.

Checklist

• Based on current main; no other open PR implements the same functions
• One topic per PR; follow-ups go in a new PR
• No comments in code except technical debt
• Unimplemented paths throw (NotImplemented_nid_no_patch); silent stubs are listed under Silent stubs in their library's technical debt file
• No notes, investigation .md files or images added to the repository (attach them to this PR)
• New third-party code is a submodule built from source
• Generic behaviour, not specific to one title
• Depends on: none
• Fixes: none
• Unblocks: none
• GPUs tested: n/a
• AI-assisted: yes

Refs #1471

[github-actions]

Some comparisons could not be completed because a pull request conflicts with the base branch. Review these shared files manually before concluding that there is no overlap.

Pull request	Shared files not compared
PR 3986	core/libs/prx/libkernel/System/src/Signals.cpp
PR 4563	core/libs/prx/libkernel/CMakeLists.txt
PR 5021	core/libs/prx/libkernel/CMakeLists.txt
PR 6936	core/libs/prx/libkernel/CMakeLists.txt
PR 7192	core/libs/prx/libkernel/CMakeLists.txt
PR 9019	core/libs/prx/libkernel/CMakeLists.txt
PR 9156	core/libs/prx/libkernel/CMakeLists.txt
PR 9293	core/libs/prx/libkernel/CMakeLists.txt
PR 9297	core/libs/prx/libkernel/CMakeLists.txt
PR 9302	core/libs/prx/libkernel/CMakeLists.txt
PR 9383	core/libs/prx/libkernel/CMakeLists.txt
PR 9389	core/libs/prx/libkernel/CMakeLists.txt core/libs/prx/libkernel/System/src/Signals.cpp
PR 9557	core/libs/prx/libkernel/CMakeLists.txt
PR 9566	core/libs/prx/libkernel/CMakeLists.txt
PR 9938	core/libs/prx/libkernel/CMakeLists.txt
PR 11831	core/libs/prx/libkernel/CMakeLists.txt
PR 11896	core/libs/prx/libkernel/CMakeLists.txt
PR 11939	core/libs/prx/libkernel/CMakeLists.txt
PR 12544	core/libs/prx/libkernel/CMakeLists.txt
PR 12970	core/libs/prx/libkernel/CMakeLists.txt
PR 14264	core/libs/prx/libkernel/CMakeLists.txt
PR 14299	core/libs/prx/libkernel/CMakeLists.txt

[B-Howell]

Linux run: Ubuntu 22.04 on WSL2 (kernel 6.18), GCC 11.4, Release, AMD Ryzen 7 7800X3D (16 threads). I built head b81185dd merged onto main f759e165d9; it merged without conflicts.

• With the change: guest_itimer, guest_itimer_pending_default and guest_itimer_short_expiry pass 10 of 10 runs each (--repeat until-fail:10).
• Control: with Signals.cpp from main and this PR's test, all three also pass 10 of 10.

So the slow expiry you measured on Windows (median about 15.9 ms) doesn't happen on this Linux host, and guest_itimer_short_expiry only separates the two versions on Windows. On Linux the change causes no regression in these three tests. I didn't record the per-run medians here.

AI-assisted (Claude Code).