[igt-dev] [PATCH i-g-t v2 6/6] tests/gem_exercise_blt: Add fast-copy-emit test
Zbigniew Kempczyński
zbigniew.kempczynski at intel.com
Mon Jan 9 15:38:23 UTC 2023
On Fri, Dec 23, 2022 at 12:13:51PM +0100, Karolina Stolarek wrote:
> Add a subtest where two fast copy commands are executed within the
> single batch buffer.
>
> Signed-off-by: Karolina Stolarek <karolina.stolarek at intel.com>
> ---
> tests/i915/gem_exercise_blt.c | 190 ++++++++++++++++++++++++++++++++--
> 1 file changed, 184 insertions(+), 6 deletions(-)
>
> diff --git a/tests/i915/gem_exercise_blt.c b/tests/i915/gem_exercise_blt.c
> index 67a8b660..b8bfc436 100644
> --- a/tests/i915/gem_exercise_blt.c
> +++ b/tests/i915/gem_exercise_blt.c
> @@ -37,6 +37,146 @@ static struct param {
> if (param.write_png) \
> blt_surface_to_png((fd), (id), (name), (obj), (w), (h)); } while (0)
>
> +struct blt_fast_copy_data {
> + int i915;
> + struct blt_copy_object src;
> + struct blt_copy_object mid;
> + struct blt_copy_object dst;
> +
> + struct blt_copy_batch bb;
> + enum blt_color_depth color_depth;
> +
> + /* debug stuff */
> + bool print_bb;
> +};
> +
> +static int fast_copy_one_bb(int i915,
> + const intel_ctx_t *ctx,
> + const struct intel_execution_engine2 *e,
> + uint64_t ahnd,
> + const struct blt_fast_copy_data *blt)
> +{
> + struct drm_i915_gem_execbuffer2 execbuf = {};
> + struct drm_i915_gem_exec_object2 obj[4] = {};
> + struct blt_copy_data blt_tmp;
> + uint64_t src_offset, mid_offset, dst_offset, bb_offset, alignment;
> + uint64_t bb_pos = 0;
> + uint32_t flags;
> + int ret;
> +
> + alignment = gem_detect_safe_alignment(i915);
> +
> + src_offset = get_offset(ahnd, blt->src.handle, blt->src.size, alignment);
> + mid_offset = get_offset(ahnd, blt->mid.handle, blt->mid.size, alignment);
> + dst_offset = get_offset(ahnd, blt->dst.handle, blt->dst.size, alignment);
> + bb_offset = get_offset(ahnd, blt->bb.handle, blt->bb.size, alignment);
> +
> + /* First blit */
> + memset(&blt_tmp, 0, sizeof(blt_tmp));
> + blt_tmp.src = blt->src;
> + blt_tmp.dst = blt->mid;
> + blt_tmp.bb = blt->bb;
> + blt_tmp.color_depth = blt->color_depth;
> + blt_tmp.print_bb = blt->print_bb;
> + bb_pos = emit_blt_fast_copy(i915, ahnd, &blt_tmp, bb_pos, false);
> +
> + /* Second blit */
> + memset(&blt_tmp, 0, sizeof(blt_tmp));
> + blt_tmp.src = blt->mid;
> + blt_tmp.dst = blt->dst;
> + blt_tmp.bb = blt->bb;
> + blt_tmp.color_depth = blt->color_depth;
> + blt_tmp.print_bb = blt->print_bb;
> + bb_pos = emit_blt_fast_copy(i915, ahnd, &blt_tmp, bb_pos, true);
Pipelined blits? Nice!
> +
> + flags = EXEC_OBJECT_PINNED | EXEC_OBJECT_SUPPORTS_48B_ADDRESS;
> +
> + obj[0].handle = blt->src.handle;
> + obj[0].offset = CANONICAL(src_offset);
> + obj[0].flags = flags;
> +
> + obj[1].handle = blt->mid.handle;
> + obj[1].offset = CANONICAL(mid_offset);
> + obj[1].flags = flags | EXEC_OBJECT_WRITE;
> +
> + obj[2].handle = blt->dst.handle;
> + obj[2].offset = CANONICAL(dst_offset);
> + obj[2].flags = flags;
> +
> + obj[3].handle = blt->bb.handle;
> + obj[3].offset = CANONICAL(bb_offset);
> + obj[3].flags = flags;
> +
> + execbuf.buffer_count = 4;
> + execbuf.buffers_ptr = to_user_pointer(obj);
> + execbuf.rsvd1 = ctx ? ctx->id : 0;
> + execbuf.flags = e ? e->flags : I915_EXEC_BLT;
> + ret = __gem_execbuf(i915, &execbuf);
> +
> + gem_sync(i915, blt->bb.handle);
> +
> + return ret;
> +}
> +
> +static void fast_copy_emit(int i915, const intel_ctx_t *ctx,
> + const struct intel_execution_engine2 *e,
> + uint32_t region1, uint32_t region2,
> + enum blt_tiling_type mid_tiling)
> +{
> + struct blt_fast_copy_data blt = {};
> + struct blt_copy_object *src, *mid, *dst;
> + const uint32_t bpp = 32;
> + uint64_t bb_size = 4096;
> + uint64_t ahnd = intel_allocator_open_full(i915, ctx->id, 0, 0,
> + INTEL_ALLOCATOR_SIMPLE,
> + ALLOC_STRATEGY_LOW_TO_HIGH, 0);
> + uint32_t bb, width = param.width, height = param.height;
> + int result;
> +
> + igt_assert(__gem_create_in_memory_regions(i915, &bb, &bb_size, region1) == 0);
> +
> + src = blt_create_object(i915, region1, width, height, bpp, 0,
> + T_LINEAR, COMPRESSION_DISABLED, 0, true);
> + mid = blt_create_object(i915, region2, width, height, bpp, 0,
> + mid_tiling, COMPRESSION_DISABLED, 0, true);
> + dst = blt_create_object(i915, region1, width, height, bpp, 0,
> + T_LINEAR, COMPRESSION_DISABLED, 0, true);
> + igt_assert(src->size == dst->size);
> +
> + PRINT_SURFACE_INFO("src", src);
> + PRINT_SURFACE_INFO("mid", mid);
> + PRINT_SURFACE_INFO("dst", dst);
> +
> + blt_surface_fill_rect(i915, src, width, height);
> + WRITE_PNG(i915, mid_tiling, "src", src, width, height);
> +
> + memset(&blt, 0, sizeof(blt));
> + blt.color_depth = CD_32bit;
> + blt.print_bb = param.print_bb;
> + blt_set_copy_object(&blt.src, src);
> + blt_set_copy_object(&blt.mid, mid);
> + blt_set_copy_object(&blt.dst, dst);
> + blt_set_batch(&blt.bb, bb, bb_size, region1);
> +
> + fast_copy_one_bb(i915, ctx, e, ahnd, &blt);
> + gem_sync(i915, blt.dst.handle);
> +
> + WRITE_PNG(i915, mid_tiling, "mid", &blt.mid, width, height);
> + WRITE_PNG(i915, mid_tiling, "dst", &blt.dst, width, height);
> +
> + result = memcmp(src->ptr, blt.dst.ptr, src->size);
> +
> + blt_destroy_object(i915, src);
> + blt_destroy_object(i915, mid);
> + blt_destroy_object(i915, dst);
> + gem_close(i915, bb);
> + put_ahnd(ahnd);
> +
> + munmap(&bb, bb_size);
> +
> + igt_assert_f(!result, "source and destination surfaces differs!\n");
> +}
> +
> static void fast_copy(int i915, const intel_ctx_t *ctx,
> const struct intel_execution_engine2 *e,
> uint32_t region1, uint32_t region2,
> @@ -102,12 +242,42 @@ static void fast_copy(int i915, const intel_ctx_t *ctx,
> igt_assert_f(!result, "source and destination surfaces differs!\n");
> }
>
> +enum fast_copy_func {
> + FAST_COPY,
> + FAST_COPY_EMIT
> +};
> +
> +static char
> + *full_subtest_str(char *regtxt, enum blt_tiling_type tiling,
> + enum fast_copy_func func)
> +{
> + const char *multi_str = "emit";
> + char const *tile_str = blt_tiling_name(tiling);
> + char *name;
> + uint32_t len, sub_len;
> +
> + len = strlen(regtxt) + strlen(tile_str) + strlen(multi_str) + 2;
Similar to other patch - looks overengineered. Maybe asprintf() will help
you to avoid calling strlen().
--
Zbigniew
> + name = malloc(len * sizeof(char));
> + igt_assert(name);
> +
> + sub_len = snprintf(name, len, "%s-%s", tile_str, regtxt);
> +
> + if (func == FAST_COPY_EMIT)
> + snprintf(name + sub_len, len, "-%s", multi_str);
> +
> + return name;
> +}
> +
> static void fast_copy_test(int i915,
> const intel_ctx_t *ctx,
> - struct igt_collection *set)
> + struct igt_collection *set,
> + enum fast_copy_func func)
> {
> struct igt_collection *regions;
> const struct intel_execution_engine2 *e;
> + void (*copy_func)(int i915, const intel_ctx_t *ctx,
> + const struct intel_execution_engine2 *e,
> + uint32_t r1, uint32_t r2, enum blt_tiling_type tiling);
> int tiling;
>
> for_each_tiling(tiling) {
> @@ -119,20 +289,23 @@ static void fast_copy_test(int i915,
> continue;
> for_each_variation_r(regions, 2, set) {
> uint32_t region1, region2;
> - char *regtxt;
> + char *regtxt, *test_name;
>
> region1 = igt_collection_get_value(regions, 0);
> region2 = igt_collection_get_value(regions, 1);
> +
> + copy_func = (func == FAST_COPY) ? fast_copy : fast_copy_emit;
> regtxt = memregion_dynamic_subtest_name(regions);
> + test_name = full_subtest_str(regtxt, tiling, func);
>
> - igt_dynamic_f("%s-%s",
> - blt_tiling_name(tiling), regtxt) {
> - fast_copy(i915, ctx, e,
> + igt_dynamic_f("%s", test_name) {
> + copy_func(i915, ctx, e,
> region1, region2,
> tiling);
> }
>
> free(regtxt);
> + free(test_name);
> }
> }
> }
> @@ -206,7 +379,12 @@ igt_main_args("b:pst:W:H:", NULL, help_str, opt_handler, NULL)
>
> igt_describe("Check fast-copy blit");
> igt_subtest_with_dynamic("fast-copy") {
> - fast_copy_test(i915, ctx, set);
> + fast_copy_test(i915, ctx, set, FAST_COPY);
> + }
> +
> + igt_describe("Check multiple fast-copy in one batch");
> + igt_subtest_with_dynamic("fast-copy-emit") {
> + fast_copy_test(i915, ctx, set, FAST_COPY_EMIT);
> }
>
> igt_fixture {
> --
> 2.25.1
>
More information about the igt-dev
mailing list