Re: [kvm-unit-tests PATCH] s390x: Add specification exception test

[Cornelia Huck <cohuck@redhat.com>]

On Tue, Jul 06 2021, Janis Schoetterl-Glausch <scgl@linux.ibm.com> wrote:

> Generate specification exceptions and check that they occur.
> Also generate specification exceptions during a transaction,
> which results in another interruption code.
> With the iterations argument one can check if specification
> exception interpretation occurs, e.g. by using a high value and
> checking that the debugfs counters are substantially lower.
> The argument is also useful for estimating the performance benefit
> of interpretation.
>
> Signed-off-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
> ---
>  s390x/Makefile           |   1 +
>  lib/s390x/asm/arch_def.h |   1 +
>  s390x/spec_ex.c          | 344 +++++++++++++++++++++++++++++++++++++++
>  s390x/unittests.cfg      |   3 +
>  4 files changed, 349 insertions(+)
>  create mode 100644 s390x/spec_ex.c

(...)

> +static void lpsw(uint64_t psw)

Maybe call this load_psw(), as you do a bit more than a simple lpsw?

> +{
> +	uint32_t *high, *low;
> +	uint64_t r0 = 0, r1 = 0;
> +
> +	high = (uint32_t *) &fixup_early_pgm_psw.mask;
> +	low = high + 1;
> +
> +	asm volatile (
> +		"	epsw	%0,%1\n"
> +		"	st	%0,%[high]\n"
> +		"	st	%1,%[low]\n"
> +		"	larl	%0,nop%=\n"
> +		"	stg	%0,%[addr]\n"
> +		"	lpsw	%[psw]\n"
> +		"nop%=:	nop\n"
> +		: "+&r"(r0), "+&a"(r1), [high] "=&R"(*high), [low] "=&R"(*low)
> +		, [addr] "=&R"(fixup_early_pgm_psw.addr)
> +		: [psw] "Q"(psw)
> +		: "cc", "memory"
> +	);
> +}

(...)

> +static void test_spec_ex(struct args *args,
> +			 const struct spec_ex_trigger *trigger)
> +{
> +	uint16_t expected_pgm = PGM_INT_CODE_SPECIFICATION;
> +	uint16_t pgm;
> +	unsigned int i;
> +
> +	register_pgm_cleanup_func(trigger->fixup);
> +	for (i = 0; i < args->iterations; i++) {
> +		expect_pgm_int();
> +		trigger->func();
> +		pgm = clear_pgm_int();
> +		if (pgm != expected_pgm) {
> +			report(0,
> +			"Program interrupt: expected(%d) == received(%d)",
> +			expected_pgm,
> +			pgm);

The indentation looks a bit funny here.

> +			return;
> +		}
> +	}
> +	report(1,
> +	"Program interrupt: always expected(%d) == received(%d)",
> +	expected_pgm,
> +	expected_pgm);

Here as well.

> +}

(...)

> +#define report_info_if(cond, fmt, ...)			\
> +	do {						\
> +		if (cond) {				\
> +			report_info(fmt, ##__VA_ARGS__);\
> +		}					\
> +	} while (0)

I'm wondering whether such a wrapper function could be generally useful.