Re: [PATCH bpf-next 2/6] selftests/bpf: add batched, mostly in-kernel BPF triggering benchmarks

[Alexei Starovoitov <alexei.starovoitov@xxxxxxxxx>]

On Thu, Mar 21, 2024 at 5:00 PM Andrii Nakryiko <andrii@xxxxxxxxxx> wrote:
> +
> +static error_t parse_arg(int key, char *arg, struct argp_state *state)
> +{
> +       long ret;
> +
> +       switch (key) {
> +       case ARG_TRIG_BATCH_ITERS:
> +               ret = strtol(arg, NULL, 10);
> +               if (ret < 1 || ret > UINT_MAX) {
> +                       fprintf(stderr, "invalid --trig-batch-iters value");
> +                       argp_usage(state);
> +               }

...

> +const volatile int batch_iters = 0;
> +
> +SEC("raw_tp")
> +int trigger_count(void *ctx)
> +{
> +       int i;
> +
> +       for (i = 0; i < batch_iters; i++)
> +               inc_counter();

so it's passing the verifier due to bounded loop logic, right?

Then let's limit cmd line arg to something small instead of UINT_MAX.
Otherwise users will be surprised by the errors
coming from --trig-batch-iters 500k