syscalls: Add timer measurement library

This commit adds a timer measurement library, mostly based on changes done to the pselect01.c test and changes all tests that measure timer precision to use it.

The timer testcases that measure timeouts now just define sampling function and
optional setup and cleanup. The rest of the functionality is implemented in the
lib/tst_timer_test.c library. This change not only removes fair amount of duplicated code but also allows us to tune thresholds and define testcases in a
single place for all testcases.

The timer measurement library also supports for passing sleep time and number of iterations as a command-line parameters, can print nifty frequency plot into the terminal, as well as save test measurements into a text file.

Signed-off-by Cyril Hrubis <[email protected]>

---
 include/tst_timer.h                                |  26 ++
 include/tst_timer_test.h                           |  74 ++++
 lib/tst_timer_test.c                               | 434
 +++++++++++++++++++++
 runtest/syscalls                                   |   1 +
 testcases/kernel/syscalls/.gitignore               |   1 +
 .../syscalls/clock_nanosleep/clock_nanosleep01.c   |  67 +---
 .../syscalls/clock_nanosleep/clock_nanosleep02.c   |  49 +++
 .../kernel/syscalls/epoll_wait/epoll_wait02.c      | 125 ++----
 testcases/kernel/syscalls/futex/futex_wait05.c     |  61 +--
 testcases/kernel/syscalls/nanosleep/nanosleep01.c  |  56 +--
 testcases/kernel/syscalls/poll/poll02.c            | 104 ++---
 testcases/kernel/syscalls/pselect/pselect01.c      | 161 ++------
 testcases/kernel/syscalls/select/select04.c        | 103 ++---
 13 files changed, 746 insertions(+), 516 deletions(-)
 create mode 100644 include/tst_timer_test.h
 create mode 100644 lib/tst_timer_test.c
 create mode 100644

testcases/kernel/syscalls/clock_nanosleep/clock_nanosleep02.c

include/tst_timer.h +0 -0

index f0a10bd45..0448f4428 100644

include/tst_timer.h +9 -0

+{
+	struct timespec ret;
+
+	ret.tv_sec = us / 1000;
+	ret.tv_nsec = (us % 1000) * 1000000;
+
+	return ret;
+}
+

include/tst_timer.h +7 -0

+void do_timer_test(long long usec, unsigned int nsamples)
+{
+	long long trunc_mean, median;
+	unsigned int discard = compute_discard(nsamples);
+	unsigned int keep_samples = nsamples - discard;
+	long long threshold = compute_threshold(usec, keep_samples);
+	unsigned int i;

include/tst_timer_test.h line 60

Quoted 7 lines

include/tst_timer_test.h new file +7 -0

+
+struct tst_timer_test {
+	const char *scall;
+	int (*sample)(int clk_id, long long usec);
+	void (*setup)(void);
+	void (*cleanup)(void);
+};

I'd rather keep tst_test struct and expose some new function, that would do all of this.

void test_all()

{
    tst_timer_test("select()", test_sample_function);
}

void tst_timer_test(fn_name, test_sample_function) {
timer_parse_options();
timer_setup();

  for n ... {
     do_timer_test(timer_tcases[n].usec, timer_tcases[n].samples, test_sample_function);

}
timer_cleanup();
}

What I'm afraid of is that we end up mirror-ing lot of functionality in tst_test struct: needsroot, tmpdir, kernelversion, extra parameter

Quoted 12 lines

lib/tst_timer_test.c new file +12 -0

+void do_timer_test(long long usec, unsigned int nsamples)
+{
+	long long trunc_mean, median;
+	unsigned int discard = compute_discard(nsamples);
+	unsigned int keep_samples = nsamples - discard;
+	long long threshold = compute_threshold(usec, keep_samples);
+	unsigned int i;
+	int failed = 0;
+
+	tst_res(TINFO,
+		"%s sleeping for %llius %u iterations, threshold %.2fus",
+		timer_test->scall, usec, nsamples,

lib/tst_timer_test.c line 266

Quoted 5 lines

lib/tst_timer_test.c new file +5 -0

+		1.00 * threshold / (keep_samples));
+
+	cur_sample = 0;
+	for (i = 0; i < nsamples; i++) {
+		if (timer_test->sample(CLOCK_REALTIME, usec)) {

Since we use resolution of monotonic clock, should we also measure with it here?

Quoted 5 lines

lib/tst_timer_test.c new file +5 -0

+			tst_res(TINFO, "sampling function failed, exitting");
+			return;
+		}
+	}
+

lib/tst_timer_test.c line 276

Quoted 5 lines

lib/tst_timer_test.c new file +5 -0

+	qsort(samples, nsamples, sizeof(samples[0]), cmp);
+
+	write_to_file();
+
+	for (i = 0; samples[i] > 10 * usec; i++) {

This could also use a range check for length of samples array, just in case all samples are outliners.

Hi!

Quoted 23 lines

include/tst_timer_test.h line 60

Quoted 7 lines

include/tst_timer_test.h new file +7 -0

+
+struct tst_timer_test {
+	const char *scall;
+	int (*sample)(int clk_id, long long usec);
+	void (*setup)(void);
+	void (*cleanup)(void);
+};

I'd rather keep tst_test struct and expose some new function, that would do all of this.

void test_all()

{
    tst_timer_test("select()", test_sample_function);
}

void tst_timer_test(fn_name, test_sample_function) {
timer_parse_options();
timer_setup();

  for n ... {
     do_timer_test(timer_tcases[n].usec, timer_tcases[n].samples,
     test_sample_function);

Quoted 6 lines

}
timer_cleanup();
}

What I'm afraid of is that we end up mirror-ing lot of functionality in tst_test struct: needsroot, tmpdir, kernelversion, extra parameter

Hmm, we can also fill in the standard tst_test structure in the test then "override" some of the fields in the timer test library.

I.e. test defines tst_test structure and the sampling function, timer library main() stores and replaces setup & cleanup, sets the test function then calls the tst_run_tcases() function. Does that sound better?

Quoted 7 lines

Hmm, we can also fill in the standard tst_test structure in the test then "override" some of the fields in the timer test library.

I.e. test defines tst_test structure and the sampling function, timer library main() stores and replaces setup & cleanup, sets the test function then calls the tst_run_tcases() function. Does that sound better?

Yes. This would mean we extend tst_test struct and add sampling function. As long as rest of fields continue to work (with some exceptions [1]), it sounds good.

I see you still want to keep timer library main(), but I'm not sure how you are going to select it instead of tst_test main.

Quoted 11 lines

Quoted 7 lines

Hmm, we can also fill in the standard tst_test structure in the test then "override" some of the fields in the timer test library.

I.e. test defines tst_test structure and the sampling function, timer library main() stores and replaces setup & cleanup, sets the test function then calls the tst_run_tcases() function. Does that sound better?

Yes. This would mean we extend tst_test struct and add sampling function. As long as rest of fields continue to work (with some exceptions [1]), it sounds good.

We can setup the test/sample functions as an union. As we can then distinguish between them based on .tcnt being non-zero in a case of standard test and expect sampling function to be set in a case of the timer test. However this makes the whole interface less type safe and also much harder to extend. I was thinking of adding support for NULL terminated array of test funcitons as well since we have some testcases that use the test() function as dispatcher with a long switch() statement...

So maybe we should just add a few more function pointers there.

Hi!

Quoted 8 lines

Quoted 7 lines

Hmm, we can also fill in the standard tst_test structure in the test then "override" some of the fields in the timer test library.

I.e. test defines tst_test structure and the sampling function, timer library main() stores and replaces setup & cleanup, sets the test function then calls the tst_run_tcases() function. Does that sound better?

Quoted 3 lines

Yes. This would mean we extend tst_test struct and add sampling function. As long as rest of fields continue to work (with some exceptions [1]), it sounds good.

We can setup the test/sample functions as an union. As we can then distinguish between them based on .tcnt being non-zero in a case of standard test and expect sampling function to be set in a case of the timer test.

However this makes the whole interface less type safe and also much harder to extend.

I was thinking of adding support for NULL
terminated array of test funcitons as well since we have some testcases that use the test() function as dispatcher with a long switch() statement...

So maybe we should just add a few more function pointers there.

Quoted 2 lines

I see you still want to keep timer library main(), but I'm not sure how you are going to select it instead of tst_test main.

Only the tst_timer_test.h should be included in the timer testcases which includes tst_test.h after defining TST_NO_DEFAULT_MAIN, then we can define timer specific main that saves cleanup/setup and replaces these with the timer test cleanup/setup that calls the test specifc cleanup/setup functions. The timer library main also has to set up the test function and test options in the tst_test structure since I want to avoid setting these in each of the testcases to avoid copy & paste errors.

What if we modified lib/tst_test.c to make calls to timerlib where needed? User would just include tst_test.h as before, but instead of test_all he sets sampling_func in tst_test struct.