[PATCH 6/6] riscv: Test for a SBI implementation ID range

[Andrew Jones]

On Thu, Feb 29, 2024 at 01:42:12PM +0100, cem at kernel.org wrote:
> From: Carlos Maiolino <cem at kernel.org>
> 
> This tests the SBI ID against SBI_IMPLID_MAX, so it can be used to test a SBI
> against the available IDs from a specification version.
> 
> Signed-off-by: Carlos Maiolino <cmaiolino at redhat.com>
> ---
>  riscv/sbi.c | 8 ++++++++
>  1 file changed, 8 insertions(+)
> 
> diff --git a/riscv/sbi.c b/riscv/sbi.c
> index 9daab9dc..37f5680c 100644
> --- a/riscv/sbi.c
> +++ b/riscv/sbi.c
> @@ -64,6 +64,14 @@ static void check_base(void)
>  	}
>  	report_prefix_pop();
>  
> +	report_prefix_push("sbi_impl_id_max");
> +	if (env_is_defined("SBI_IMPLID_MAX")) {
> +		expected = strtol(getenv("SBI_IMPLID_MAX"), NULL, 0);
> +
> +		gen_report(&ret, (ret.value <= expected));
> +	}
> +	report_prefix_pop();

tl;dr: let's drop this patch

There's nothing in the spec about a maximum and the previous test already
would have failed if the value wasn't exactly what the tester expected
it to be. I can see a case for having a maximum test instead of the exact
value test, though, if test configurations were meant to work on more than
one SBI implementation, but it's probably best that each configuration be
tailored to specific implementations, so a maximum instead of an explicit
set probably isn't what we want.

Thanks,
drew

> +
>  	report_prefix_push("probe_ext");
>  	if (env_is_defined("PROBE_EXT")) {
>  		expected = strtol(getenv("PROBE_EXT"), NULL, 0);
> -- 
> 2.43.2
> 
> 
> -- 
> kvm-riscv mailing list
> kvm-riscv at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/kvm-riscv