sm_coretest_timer_handoff_1

Core test for the hand-off of a timer created on the state machine thread to the executor in the signal detector’s polling loop. It is sm_coretest_forced_finish_1’s cycle run at 100 ms instead of 1 s, which is the shape that reproduced the missed hand-off (finding W): a plain two-state timer ping-pong with no forced finish did not.

What it does

StA  ──EvCountdownEnd──▶  StB  ──EvTimer<CbTimerCountdownLoop>──▶  StA

• StA runs CbSleepFor(30s) in OrSleep and CbTimerCountdownOnce(100ms) in OrTimer, with an SrEventCountdown(1) reactor on the timer event. The timer always wins, so every StA exit forces the sleep to finish: the state machine thread holds the exit for the forced-finish wait, then enters StB, whose CbTimerCountdownLoop(100ms) creates a new ROS timer in its onEntry.
• The polling loop’s executor has to pick that timer up. A hand-off it misses leaves the timer ready and never executed, and the machine sits in StB for good.
• About three cycles a second; a healthy 60 s run logs some 300 Entered lines with no gap longer than a cycle. The gate’s invariant is the gap: no two consecutive state entries more than 5 s apart, and the last one within 5 s of the interrupt.

Running it

ros2 launch sm_coretest_timer_handoff_1 sm_coretest_timer_handoff_1.launch.py

Output is teed to /tmp/sm_coretest_timer_handoff_1_latest.log. The plain build is the one that matters: the sanitizer allocators delay the reuse of freed memory, and the race this test exists for has only ever shown in a plain build.