Hello, Sergey,

<snipped>

+    interval = 1,
+    path = "/dev/null",
+}
+test:is(res, true, "res is correct")
+test:is(err, nil, "no error")
+test:is(errno, nil, "no errno")

I suppose that 2 last checks are excess. The first one is enough to be
sure that the profiler is started. Also, we may use `assert()` here
instead of `test:is()` check, since we don't want to _test_ the starting of
the profiler only to _assert_ that the sysprof has been started.

last two checks were removed and test:is() replaced with assert()

I would rather use
| assert(misc.sysprof.start({...})
and
| assert(misc.sysprof.stop())
instead, for simplicity and to avoid the irrelevant local variables.

Feel free to ignore.

+
+local report = misc.sysprof.report()
+test:ok(report.samples == 0, "total number of samples is non-zero")

I'm not sure that this will always be true (for example, in coverage
workflow). I suggest increasing the interval dramatically to avoid false
positives here.

Updated.

Side note: Checking the non-0 samples for default payload instead.

+misc.sysprof.stop()
+

<snipped>

2.43.0