main: add configurable fault injection interval

[nhitar]

Adds a parameter to configure the fault injection interval, which is locked at default 100.

Can be used to make the fault injection probability checks more frequent. This would lead to a more thorough test of system stability.

[nhitar]

@acerv hello, can you review these changes please?

[acerv]

Hi, what's the point of having a probability bigger than 100 if the kernel threats any value above 100 as 100 ?

[nhitar]

Can be used to make the fault injection probability checks more frequent.

Hi, what's the point of having a probability bigger than 100 if the kernel threats any value above 100 as 100 ?

The idea is to modify the interval between probability checks, rather than the probability itself. I haven't really changed the probability in code.

Right now 100% probability of failure grants that every 100th call will fail (interval=100). So, with different intervals, it is possible, for example, to have fails (or probability checks if below 100%) every 10th call (interval=10). Or, with an interval set to 1 and a probability to 10%, each call may fail with that probability.

docs:

/sys/kernel/debug/fail*/interval:

specifies the interval between failures, for calls to should_fail() that pass all the other tests.

Note that if you enable this, by setting interval>1, you will probably want to set probability=100.

[acerv]

Of course, I was switching between reviews and messed up this one. Now it's clear thanks

[acerv]

This is a bit counterintuitive. If the value is below 1, we should default it to 1.

[nhitar]

Changed it to return max(1, ret)

[acerv]

We should add a test for real values. Maybe we can create a parametric test passing them as following:

    @pytest.mark.parametrize("val", ("0", "20", "100", "2000"))
    def test_finterval_config(self, val):
        assert libkirk.main._finterval_config(val) == int(val)

[acerv]

Default interval value is 1, according to documentation. This has to be changed.

[nhitar]

I agree, but thought of keeping the original value. Changed it now.