We need some sort of stress testing that tries to break quark.
The major culprit is us doing something wrong in the bpf probes, or the kernel semantics just changing (like PREEMPT_LAZY). This can output invalid events in the ringbuffer, which makes us take an invalid field offset.
quark-bomb(8) would be a stress testing program that starts something like quark-mon(8) does, but also spawns a bunch of things that will generate events. We can be smart and have tests that stress a single cpu to see if we can preempt probes and whatnot.
We need some sort of stress testing that tries to break quark.
The major culprit is us doing something wrong in the bpf probes, or the kernel semantics just changing (like PREEMPT_LAZY). This can output invalid events in the ringbuffer, which makes us take an invalid field offset.
quark-bomb(8) would be a stress testing program that starts something like quark-mon(8) does, but also spawns a bunch of things that will generate events. We can be smart and have tests that stress a single cpu to see if we can preempt probes and whatnot.