[PATCH i-g-t v7 5/5] tests/intel/xe_ccs: Add compression support for Lunarlake
Zbigniew Kempczyński
zbigniew.kempczynski at intel.com
Fri Dec 15 08:11:32 UTC 2023
On Fri, Dec 15, 2023 at 11:41:49AM -0800, Akshata Jahagirdar wrote:
> In XE2 IGFX platform only sysmem participates in compression.
> So for XE2 IGFX platform, create all blt objects in sysmem itself,
> For other platform, it will stay the same way as per pervious implementation.
> Also update the pat-index to reflect the compression status.
> Alter the range to default_alignment before vm bind/unbind operation to
> align the buffer object size with page size.
>
> Signed-off-by: Akshata Jahagirdar <akshata.jahagirdar at intel.com>
> ---
> lib/xe/xe_util.c | 2 +-
> tests/intel/xe_ccs.c | 38 ++++++++++++++++++++++++--------------
> 2 files changed, 25 insertions(+), 15 deletions(-)
>
> diff --git a/lib/xe/xe_util.c b/lib/xe/xe_util.c
> index ba8eece71..42cac9dc2 100644
> --- a/lib/xe/xe_util.c
> +++ b/lib/xe/xe_util.c
> @@ -148,7 +148,7 @@ static struct drm_xe_vm_bind_op *xe_alloc_bind_ops(int xe,
> ops->flags = flags;
> ops->obj_offset = 0;
> ops->addr = obj->offset;
> - ops->range = obj->size;
> + ops->range = ALIGN(obj->size, xe_get_default_alignment(xe));
Objects in system memory might be 4K, whereas xe_get_default_alignment()
will return 64K for device memory on discrete (then it will fail). I would
align to 4K as this should be safe both for system and device memory.
Apart of this I would separate this change (single line) to another
patch.
--
Zbigniew
> ops->prefetch_mem_region_instance = 0;
> if (obj->pat_index == DEFAULT_PAT_INDEX)
> ops->pat_index = intel_get_pat_idx_wb(xe);
> diff --git a/tests/intel/xe_ccs.c b/tests/intel/xe_ccs.c
> index ac0805017..1444b7a66 100644
> --- a/tests/intel/xe_ccs.c
> +++ b/tests/intel/xe_ccs.c
> @@ -96,20 +96,27 @@ static void surf_copy(int xe,
> struct blt_ctrl_surf_copy_data surf = {};
> uint32_t bb1, bb2, ccs, ccs2, *ccsmap, *ccsmap2;
> uint64_t bb_size, ccssize = mid->size / CCS_RATIO(xe);
> + uint64_t ccs_bo_size = xe_get_default_alignment(xe);
> uint32_t *ccscopy;
> uint8_t uc_mocs = intel_get_uc_mocs_index(xe);
> uint32_t sysmem = system_memory(xe);
> + uint8_t comp_pat_index = DEFAULT_PAT_INDEX;
> + uint16_t cpu_caching = __xe_default_cpu_caching(xe, sysmem, 0);
> int result;
>
> igt_assert(mid->compression);
> + if (AT_LEAST_GEN(intel_get_drm_devid(xe), 20) && mid->compression) {
> + comp_pat_index = intel_get_pat_idx_uc_comp(xe);
> + cpu_caching = DRM_XE_GEM_CPU_CACHING_WC;
> + }
> ccscopy = (uint32_t *) malloc(ccssize);
> - ccs = xe_bo_create(xe, 0, ccssize, sysmem, 0);
> - ccs2 = xe_bo_create(xe, 0, ccssize, sysmem, 0);
> + ccs = xe_bo_create_caching(xe, 0, ccs_bo_size, sysmem, 0, cpu_caching);
> + ccs2 = xe_bo_create_caching(xe, 0, ccs_bo_size, sysmem, 0, cpu_caching);
>
> blt_ctrl_surf_copy_init(xe, &surf);
> surf.print_bb = param.print_bb;
> blt_set_ctrl_surf_object(&surf.src, mid->handle, mid->region, mid->size,
> - uc_mocs, DEFAULT_PAT_INDEX, BLT_INDIRECT_ACCESS);
> + uc_mocs, comp_pat_index, BLT_INDIRECT_ACCESS);
> blt_set_ctrl_surf_object(&surf.dst, ccs, sysmem, ccssize, uc_mocs,
> DEFAULT_PAT_INDEX, DIRECT_ACCESS);
> bb_size = xe_get_default_alignment(xe);
> @@ -157,7 +164,7 @@ static void surf_copy(int xe,
> blt_set_ctrl_surf_object(&surf.src, ccs, sysmem, ccssize,
> uc_mocs, DEFAULT_PAT_INDEX, DIRECT_ACCESS);
> blt_set_ctrl_surf_object(&surf.dst, mid->handle, mid->region, mid->size,
> - uc_mocs, DEFAULT_PAT_INDEX, INDIRECT_ACCESS);
> + uc_mocs, comp_pat_index, INDIRECT_ACCESS);
> blt_ctrl_surf_copy(xe, ctx, NULL, ahnd, &surf);
> intel_ctx_xe_sync(ctx, true);
>
> @@ -234,10 +241,10 @@ static int blt_block_copy3(int xe,
> igt_assert_f(blt3, "block-copy3 requires data to do blit\n");
>
> alignment = xe_get_default_alignment(xe);
> - get_offset(ahnd, blt3->src.handle, blt3->src.size, alignment);
> - get_offset(ahnd, blt3->mid.handle, blt3->mid.size, alignment);
> - get_offset(ahnd, blt3->dst.handle, blt3->dst.size, alignment);
> - get_offset(ahnd, blt3->final.handle, blt3->final.size, alignment);
> + get_offset_pat_index(ahnd, blt3->src.handle, blt3->src.size, alignment, blt3->src.pat_index);
> + get_offset_pat_index(ahnd, blt3->mid.handle, blt3->mid.size, alignment, blt3->mid.pat_index);
> + get_offset_pat_index(ahnd, blt3->dst.handle, blt3->dst.size, alignment, blt3->dst.pat_index);
> + get_offset_pat_index(ahnd, blt3->final.handle, blt3->final.size, alignment, blt3->final.pat_index);
> bb_offset = get_offset(ahnd, blt3->bb.handle, blt3->bb.size, alignment);
>
> /* First blit src -> mid */
> @@ -291,8 +298,9 @@ static void block_copy(int xe,
> uint64_t bb_size = xe_get_default_alignment(xe);
> uint64_t ahnd = intel_allocator_open(xe, ctx->vm, INTEL_ALLOCATOR_RELOC);
> uint32_t run_id = mid_tiling;
> - uint32_t mid_region = region2, bb;
> - uint32_t width = param.width, height = param.height;
> + uint32_t mid_region = (AT_LEAST_GEN(intel_get_drm_devid(xe), 20) &
> + !xe_has_vram(xe)) ? region1 : region2;
> + uint32_t width = param.width, height = param.height, bb;
> enum blt_compression mid_compression = config->compression;
> int mid_compression_format = param.compression_format;
> enum blt_compression_type comp_type = COMPRESSION_TYPE_3D;
> @@ -413,8 +421,9 @@ static void block_multicopy(int xe,
> uint64_t bb_size = xe_get_default_alignment(xe);
> uint64_t ahnd = intel_allocator_open(xe, ctx->vm, INTEL_ALLOCATOR_RELOC);
> uint32_t run_id = mid_tiling;
> - uint32_t mid_region = region2, bb;
> - uint32_t width = param.width, height = param.height;
> + uint32_t mid_region = (AT_LEAST_GEN(intel_get_drm_devid(xe), 20) &
> + !xe_has_vram(xe)) ? region1 : region2;
> + uint32_t width = param.width, height = param.height, bb;
> enum blt_compression mid_compression = config->compression;
> int mid_compression_format = param.compression_format;
> enum blt_compression_type comp_type = COMPRESSION_TYPE_3D;
> @@ -539,8 +548,9 @@ static void block_copy_test(int xe,
> region1 = igt_collection_get_value(regions, 0);
> region2 = igt_collection_get_value(regions, 1);
>
> - /* Compressed surface must be in device memory */
> - if (config->compression && !XE_IS_VRAM_MEMORY_REGION(xe, region2))
> + /* if not XE2, then Compressed surface must be in device memory */
> + if (config->compression && !(AT_LEAST_GEN((intel_get_drm_devid(xe)), 20)) &&
> + !XE_IS_VRAM_MEMORY_REGION(xe, region2))
> continue;
>
> regtxt = xe_memregion_dynamic_subtest_name(xe, regions);
> --
> 2.34.1
>
More information about the igt-dev
mailing list