selftests/timers: Improve skew_consistency by testing with other clockids

Lei Chen reported a bug with CLOCK_MONOTONIC_COARSE having inconsistencies
when NTP is adjusting the clock frequency.

This has gone seemingly undetected for ~15 years, illustrating a clear gap
in our testing.

The skew_consistency test is intended to catch this sort of problem, but
was focused on only evaluating CLOCK_MONOTONIC, and thus missed the problem
on CLOCK_MONOTONIC_COARSE.

So adjust the test to run with all clockids for 60 seconds each instead of
10 minutes with just CLOCK_MONOTONIC.

Reported-by: Lei Chen <[email protected]>
Signed-off-by: John Stultz <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Link: https://lore.kernel.org/all/[email protected]
Closes: https://lore.kernel.org/lkml/[email protected]/

Author: johnstultz-work

tools/testing/selftests/timers/skew_consistency.c

@@ -47,7 +47,7 @@ int main(int argc, char **argv)
 
 	pid = fork();
 	if (!pid)
-		return system("./inconsistency-check -c 1 -t 600");
+		return system("./inconsistency-check -t 60");
 
 	ppm = 500;
 	ret = 0;