[igt-dev] [PATCH i-g-t 2/2] tests/api_intel_allocator: Add default-alignment test

[Dixit, Ashutosh]

On Wed, 23 Feb 2022 23:54:15 -0800, Zbigniew Kempczyński wrote:
>
> diff --git a/tests/i915/api_intel_allocator.c b/tests/i915/api_intel_allocator.c
> index 7ae4a00b7..32d1aad70 100644
> --- a/tests/i915/api_intel_allocator.c
> +++ b/tests/i915/api_intel_allocator.c
> @@ -120,6 +120,34 @@ static void reserve(int fd, uint8_t type)
>	igt_assert_eq(intel_allocator_close(ahnd), true);
>  }
>
> +static void default_alignment(int fd)
> +{
> +	struct test_obj obj[3];
> +	uint64_t ahnd, default_alignment = 0x4000;
> +
> +	ahnd = intel_allocator_open_full(fd, 0, 0, 0, INTEL_ALLOCATOR_SIMPLE,
> +					 ALLOC_STRATEGY_LOW_TO_HIGH,
> +					 default_alignment);
> +
> +	for (int i = 0; i < ARRAY_SIZE(obj); i++) {
> +		obj[i].handle = gem_handle_gen();
> +		obj[i].offset = intel_allocator_alloc(ahnd, obj[i].handle, 4096,
> +				i == 2 ? 4096 : 0);
> +		igt_debug("obj[%d].offset: %llx, handle: %u\n", i,
> +			 (long long) obj[i].offset, obj[i].handle);
> +	}
> +
> +	igt_assert_eq(obj[1].offset - obj[0].offset, default_alignment);

What would happen here if we had not specified default_alignment (i.e. let
the allocator choose safe_alignment silently)? Do we need a
"get_alignment()" API? Or clients know that the default default_alignment
is safe_alignment?

In any case, this patch is fine so:

Reviewed-by: Ashutosh Dixit <ashutosh.dixit at intel.com>