[Fix-Point]

Summary

This PR included:

1. Fix the affinity side-affect — the test pinned threads to a CPU but did not restore the previous affinity, leaking side effects into other tests;
2. Update the comments to describe MAX_THREAD_NUM and the loop count;
3. Add CONFIG_TESTING_OSTEST_SPINLOCK_PRIORITY — spinlock test threads were created at SCHED_PRIORITY_MAX (255), which starves watchdog-feeding tasks on systems with watchdog monitoring (e.g. BMP + WdgM), eventually triggering a WarmReset; the new option (default 255) lets platforms lower the priority, and is used in both spinlock_test() and run_test_thread();
4. Fix the spinlock loop count to match the Kconfig description — TEST_LOOP_SCALE is documented as "N x 10,000" but the code used * 100000; aligned to * 10000;
5. Lower the loop scale under MM_KASAN — avoid excessively slow runs (and watchdog implications) on KASAN builds;
6. Fix a Tasking compiler warning (W597) alignment issue.

Impact

• User impact: test-only; no public API changes.
• Build: no build-system changes; testing/ostest builds as before.
• Hardware: none (validated on qemu-armv7a, see Testing).
• Documentation: none.
• Security: none.
• Compatibility: the spinlock test now restores thread affinity, uses the configurable thread priority, and matches the documented loop counts.

Testing

Host: x86-64 Linux, arm-none-eabi-gcc + QEMU (qemu-system-arm).

Config: qemu-armv7a:smp (-smp 4)

Result: spinlock_test completes on 1/2/3/4 threads with correct final counters (N x 1000000) and 0 errors.

Log:

qemu-system-arm -cpu cortex-a7 -nographic -smp 4 \
-machine virt,virtualization=off,gic-version=2 \
-net none -chardev stdio,id=con,mux=on -serial chardev:con \
-mon chardev=con,mode=readline -kernel ./nuttx
nsh> hello
Hello, World!!
spinlock_test: start......
Thread num: 1, Loop times: 1000000

Test type: spinlock
spinlock: Test Results:
spinlock: Final counter: 1000000
spinlock: Average throughput : 7881812 op/s
spinlock: Total execution time: 127776336 ns

Test type: rspinlock
rspinlock: Test Results:
rspinlock: Final counter: 1000000
rspinlock: Average throughput : 4003758 op/s
spinlock: Total execution time: 250123904 ns

Test type: seqcount
seqcount: Test Results:
seqcount: Final counter: 1000000
seqcount: Average throughput : 7803117 op/s
seqcount: Total execution time: 128437824 ns

Thread num: 2, Loop times: 1000000

Test type: spinlock
spinlock: Test Results:
spinlock: Final counter: 2000000
spinlock: Average throughput : 3639822 op/s
spinlock: Total execution time: 275360688 ns

Test type: rspinlock
rspinlock: Test Results:
rspinlock: Final counter: 2000000
rspinlock: Average throughput : 1967137 op/s
rspinlock: Total execution time: 509021424 ns

Test type: seqcount
seqcount: Test Results:
seqcount: Final counter: 2000000
seqcount: Average throughput : 2913931 op/s
seqcount: Total execution time: 343893376 ns

Thread num: 3, Loop times: 1000000

Test type: spinlock
spinlock: Test Results:
spinlock: Final counter: 3000000
spinlock: Average throughput : 2436294 op/s
spinlock: Total execution time: 426570192 ns

Test type: rspinlock
rspinlock: Test Results:
rspinlock: Final counter: 3000000
rspinlock: Average throughput : 1141461 op/s
rspinlock: Total execution time: 879273104 ns

Test type: seqcount
seqcount: Test Results:
seqcount: Final counter: 3000000
seqcount: Average throughput : 2063639 op/s
seqcount: Total execution time: 528608448 ns

Thread num: 4, Loop times: 1000000

Test type: spinlock
spinlock: Test Results:
spinlock: Final counter: 4000000
spinlock: Average throughput : 1554178 op/s
spinlock: Total execution time: 651331712 ns

Test type: rspinlock
rspinlock: Test Results:
rspinlock: Final counter: 4000000
rspinlock: Average throughput : 679824 op/s
rspinlock: Total execution time: 1506000272 ns

Test type: seqcount
seqcount: Test Results:
seqcount: Final counter: 4000000
seqcount: Average throughput : 1066522 op/s
seqcount: Total execution time: 949742336 ns

spinlock_test: end......
nsh>

[cederom]

Thank you @Fix-Point :-)

[casaroli]

1. testing/ostest/spinlock.c:253: The test needs no priority option: the final counters are right at any priority, and run_test_thread() already reads the priority of the caller with sched_getparam() (line 245) and then overwrites it. ostest runs at SCHED_PRIORITY_DEFAULT, so threads at that priority stay below a watchdog task.

Fix: keep the priority that sched_getparam() returns (here and at line 440), and remove CONFIG_TESTING_OSTEST_SPINLOCK_PRIORITY.

---

2. testing/ostest/spinlock.c:226: getpid() gives the process ID, which is the main ostest thread, so sched_setaffinity() still pins the main thread, also when spinlock_test() runs the test in a new pthread. The affinity side effect stays.

Fix: use sched_setaffinity(0, ...), which changes the calling thread.

---

3. testing/ostest/Kconfig:162: The help says "scale * 100000", but LOOP_TIMES is now CONFIG_TEST_LOOP_SCALE * 10000 (and the prompt says N x 10,000).

Fix: write "scale * 10000".

---

4. testing/ostest/spinlock.c:343: rspinlock_t is the recursive spinlock, not a read-write spinlock. The comment at line 437 says "default - 1", but the code sets the Kconfig value.

Fix: correct both comments.

[casaroli]

Commits 15513a3 ("fix spinlock test loop count to match Kconfig description") and deb5eff ("lower spinlock loop scale under MM_KASAN") are signed off by vela-autotest03 <vela-autotest03@xiaomi.com> and vela-autotest04 <vela-autotest04@xiaomi.com>. These look like automation accounts, and CONTRIBUTING.md 1.5 says: "AI agents MUST NOT add Signed-off-by tags. Only humans can legally certify the commit."

Fix: have the human author of each commit sign it off with their own name and address.