[PATCH] prime_mmap: Limit aperture testing to machines with an aperture

[Kamil Konieczny]

Hi Pravalika,
On 2025-04-14 at 15:56:40 +0530, Pravalika Gurram wrote:
> From: Chris Wilson <chris.p.wilson at linux.intel.com>
> 
> The test_aperture_limit looks at the handling of internal GGTT mmaps for
> the dmabuf and checking what happens if we try to mmap more than we can
> fit inside the aperture. Thus if we have no aperture, there is no corner
> case regarding the mmapable aperture that requires validation.
> 
> Signed-off-by: Chris Wilson <chris.p.wilson at linux.intel.com>
> Signed-off-by: Pravalika Gurram <pravalika.gurram at intel.com>
> ---
>  tests/intel/prime_mmap.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/tests/intel/prime_mmap.c b/tests/intel/prime_mmap.c
> index 6de54b840..ae0c74dfe 100644
> --- a/tests/intel/prime_mmap.c
> +++ b/tests/intel/prime_mmap.c
> @@ -490,6 +490,10 @@ test_aperture_limit(uint32_t region, uint64_t size)
>  	uint64_t size1 = (gem_mappable_aperture_size(fd) * 7) / 8;
>  	uint64_t size2 = (gem_mappable_aperture_size(fd) * 3) / 8;
>  
> +	igt_require(gem_has_mappable_ggtt(fd));
> +	igt_require(size1 >= size2);
> +	igt_require(size2 >= size);
> +

Looks good.

>  	handle1 = gem_create(fd, size1);
>  	dma_buf_fd1 = prime_handle_to_fd_for_mmap(fd, handle1);
>  	igt_assert(errno == 0);
> @@ -498,7 +502,7 @@ test_aperture_limit(uint32_t region, uint64_t size)
>  	fill_bo_cpu(ptr1, size);
>  	igt_assert(memcmp(ptr1, pattern, sizeof(pattern)) == 0);
>  
> -	handle2 = gem_create(fd, size1);
> +	handle2 = gem_create(fd, size2);

After a second read this looks like a separate fix, especially
that it is not described in commit.

Please split this into a separate patch.

Regards,
Kamil

>  	dma_buf_fd2 = prime_handle_to_fd_for_mmap(fd, handle2);
>  	igt_assert(errno == 0);
>  	ptr2 = mmap(NULL, size2, PROT_READ | PROT_WRITE, MAP_SHARED, dma_buf_fd2, 0);
> -- 
> 2.34.1
>