From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id A620BC47422 for ; Mon, 29 Jan 2024 10:23:03 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 55E3D10E9C6; Mon, 29 Jan 2024 10:23:03 +0000 (UTC) Received: from mgamail.intel.com (mgamail.intel.com [198.175.65.15]) by gabe.freedesktop.org (Postfix) with ESMTPS id 9D1E510E9C6 for ; Mon, 29 Jan 2024 10:23:01 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1706523782; x=1738059782; h=message-id:date:mime-version:subject:to:references:from: in-reply-to:content-transfer-encoding; bh=ljWlOOL8/tASwpNheqn6KU8aT57p45LK3rwghqIx5Bs=; b=EWEbaya4PnxsE2+1K7RRd1mVchc3HHN/QiVmp0B+mlGyHGE+mK2IGGQD rn/2+YeYrhwkJFddXlowC1wtFSTHGSU4kQxU6adqTZqzKXph+L9RB5APG xH4fxpbIHWb9TjZixBlscDklSnFt+UqhH5qUeULnKWrxv0bF4MHMzMxUV sxttzxcrsTrNMtZO7nXUJQ20ddS+wBe0nTY7MOXyOftz6fJGKYozgT1t5 Wl0N2HB6IyStu4Ku9Ce0OPsupd2SUjhDQnNBmrKnB5rpBBP8+tVgzRXW3 XOCLbXjTyH6kz64Hd0PVOFZSZolo5EYJ6kpHw/U8V7JmMvbR7JZWg9c+x w==; X-IronPort-AV: E=McAfee;i="6600,9927,10967"; a="2774881" X-IronPort-AV: E=Sophos;i="6.05,227,1701158400"; d="scan'208";a="2774881" Received: from orviesa003.jf.intel.com ([10.64.159.143]) by orvoesa107.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 29 Jan 2024 02:23:02 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.05,227,1701158400"; d="scan'208";a="3322894" Received: from cbowler-mobl.ger.corp.intel.com (HELO [10.252.19.51]) ([10.252.19.51]) by ORVIESA003-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 29 Jan 2024 02:23:00 -0800 Message-ID: <16cd4709-5a90-489b-8fa0-5abd75b63f5f@intel.com> Date: Mon, 29 Jan 2024 10:22:57 +0000 MIME-Version: 1.0 User-Agent: Mozilla Thunderbird Subject: Re: [PATCH i-g-t v4] tests/kms_prime: Add XE support To: Nidhi Gupta , igt-dev@lists.freedesktop.org References: <20240129050726.19143-1-nidhi1.gupta@intel.com> Content-Language: en-GB From: Matthew Auld In-Reply-To: <20240129050726.19143-1-nidhi1.gupta@intel.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit X-BeenThere: igt-dev@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Development mailing list for IGT GPU Tools List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: igt-dev-bounces@lists.freedesktop.org Sender: "igt-dev" On 29/01/2024 05:07, Nidhi Gupta wrote: > From: Bhanuprakash Modem > > Add XE driver support for kms tests. > > V2: - Use rendercopy method for both i915 & xe > - Minor cleanup > V3: - New patch for cleanup & rendercopy > V4: - Fallback to blitter > V5: - Rebase > v7: - Rebase and patch cleanup > v8: - D3hot subtest is not required for xe > > Signed-off-by: Bhanuprakash Modem > Signed-off-by: Nidhi Gupta > --- > tests/kms_prime.c | 187 +++++++++++++++++++++++++++++++++++++--------- > 1 file changed, 153 insertions(+), 34 deletions(-) > > diff --git a/tests/kms_prime.c b/tests/kms_prime.c > index 135c75168..687633048 100644 > --- a/tests/kms_prime.c > +++ b/tests/kms_prime.c > @@ -36,10 +36,16 @@ > #include "igt_debugfs.h" > #include "igt_sysfs.h" > #include > +#include > > #include > #include > #include > +#include "lib/intel_blt.h" > +#include "lib/intel_mocs.h" > +#include "xe/xe_ioctl.h" > +#include "xe/xe_query.h" > + > > /** > * SUBTEST: D3hot > @@ -120,10 +126,14 @@ static igt_output_t *setup_display(int importer_fd, igt_display_t *display, > igt_display_reset(display); > > igt_output_set_pipe(output, *pipe); > - if (intel_pipe_output_combo_valid(display)) { > - found = true; > - break; > + if ((is_i915_device(importer_fd) && gem_has_lmem(importer_fd)) || > + (is_xe_device(importer_fd) && xe_has_vram(importer_fd))) { > + if (!intel_pipe_output_combo_valid(display)) > + continue; > } > + > + found = true; > + break; > } > > igt_require_f(found, "No valid connector/pipe found\n"); > @@ -131,6 +141,25 @@ static igt_output_t *setup_display(int importer_fd, igt_display_t *display, > return output; > } > > +static igt_output_t *setup_hybrid_display(int importer_fd, igt_display_t *display, > + enum pipe *pipe) > +{ > + igt_output_t *output; > + bool found = false; > + > + for_each_pipe_with_valid_output(display, *pipe, output) { > + igt_display_reset(display); > + > + igt_output_set_pipe(output, *pipe); > + > + found = true; > + break; /*Validation on single pipe is enough*/ > + } > + > + igt_require_f(found, "No valid connector/pipe found\n"); > + > + return output; > +} > static void prepare_scratch(int exporter_fd, struct dumb_bo *scratch, > drmModeModeInfo *mode, uint32_t color) > { > @@ -140,7 +169,29 @@ static void prepare_scratch(int exporter_fd, struct dumb_bo *scratch, > scratch->height = mode->vdisplay; > scratch->bpp = 32; > > - if (!is_i915_device(exporter_fd)) { > + if (is_intel_device(exporter_fd)) { > + igt_calc_fb_size(exporter_fd, mode->hdisplay, mode->vdisplay, DRM_FORMAT_XRGB8888, > + DRM_FORMAT_MOD_LINEAR, &scratch->size, &scratch->pitch); > + > + if (is_i915_device(exporter_fd)) { > + if (gem_has_lmem(exporter_fd)) > + scratch->handle = gem_create_in_memory_regions(exporter_fd, scratch->size, > + REGION_LMEM(0), REGION_SMEM); > + else > + scratch->handle = gem_create_in_memory_regions(exporter_fd, scratch->size, > + REGION_SMEM); > + > + ptr = gem_mmap__device_coherent(exporter_fd, scratch->handle, 0, > + scratch->size, PROT_WRITE | PROT_READ); > + } else { > + scratch->handle = xe_bo_create(exporter_fd, 0, > + ALIGN(scratch->size, xe_get_default_alignment(exporter_fd)), > + vram_if_possible(exporter_fd, 0), 0); Please also set DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM here. And anywhere else where you create VRAM and need CPU access. That way this also works on small-bar systems. > + > + ptr = xe_bo_mmap_ext(exporter_fd, scratch->handle, > + scratch->size, PROT_READ | PROT_WRITE); > + } > + } else { > scratch->handle = kmstest_dumb_create(exporter_fd, > ALIGN(scratch->width, 256), > scratch->height, scratch->bpp, > @@ -148,18 +199,6 @@ static void prepare_scratch(int exporter_fd, struct dumb_bo *scratch, > > ptr = kmstest_dumb_map_buffer(exporter_fd, scratch->handle, > scratch->size, PROT_WRITE); > - } else { > - igt_calc_fb_size(exporter_fd, mode->hdisplay, mode->vdisplay, DRM_FORMAT_XRGB8888, > - DRM_FORMAT_MOD_LINEAR, &scratch->size, &scratch->pitch); > - if (gem_has_lmem(exporter_fd)) > - scratch->handle = gem_create_in_memory_regions(exporter_fd, scratch->size, > - REGION_LMEM(0), REGION_SMEM); > - else > - scratch->handle = gem_create_in_memory_regions(exporter_fd, scratch->size, > - REGION_SMEM); > - > - ptr = gem_mmap__device_coherent(exporter_fd, scratch->handle, 0, scratch->size, > - PROT_WRITE | PROT_READ); > } > > for (size_t idx = 0; idx < scratch->size / sizeof(*ptr); ++idx) > @@ -178,23 +217,52 @@ static void prepare_fb(int importer_fd, struct dumb_bo *scratch, struct igt_fb * > color_encoding, color_range); > } > > +static struct blt_copy_object *blt_fb_init(const struct igt_fb *fb, > + uint32_t memregion, uint32_t pitch) > +{ > + uint32_t name, handle; > + struct blt_copy_object *blt; > + > + blt = malloc(sizeof(*blt)); > + igt_assert(blt); > + > + name = gem_flink(fb->fd, fb->gem_handle); > + handle = gem_open(fb->fd, name); > + > + blt_set_object(blt, handle, fb->size, memregion, > + intel_get_uc_mocs_index(fb->fd), > + 0, 0, 0, 0); > + > + blt_set_geom(blt, pitch, 0, 0, fb->width, fb->height, 0, 0); > + > + blt->plane_offset = 0; > + > + blt->ptr = xe_bo_mmap_ext(fb->fd, handle, fb->size, > + PROT_READ | PROT_WRITE); > + return blt; > +} > + > static void import_fb(int importer_fd, struct igt_fb *fb, > int dmabuf_fd, uint32_t pitch) > { > uint32_t offsets[4] = {}, pitches[4] = {}, handles[4] = {}, temp_buf_handle; > int ret; > + struct igt_fb dst_fb; > > - if (is_i915_device(importer_fd)) { > - if (gem_has_lmem(importer_fd)) { > - uint64_t ahnd = get_reloc_ahnd(importer_fd, 0); > - uint64_t fb_size = 0; > + if ((is_i915_device(importer_fd) && gem_has_lmem(importer_fd)) || > + (is_xe_device(importer_fd) && xe_has_vram(importer_fd))) { > + uint64_t fb_size = 0; > + uint64_t ahnd = 0; > > - igt_info("Importer is dGPU\n"); > - temp_buf_handle = prime_fd_to_handle(importer_fd, dmabuf_fd); > - igt_assert(temp_buf_handle > 0); > - fb->gem_handle = igt_create_bo_with_dimensions(importer_fd, fb->width, fb->height, > - fb->drm_format, fb->modifier, pitch, &fb_size, NULL, NULL); > - igt_assert(fb->gem_handle > 0); > + igt_info("Importer is dGPU\n"); > + temp_buf_handle = prime_fd_to_handle(importer_fd, dmabuf_fd); > + igt_assert(temp_buf_handle > 0); > + fb->gem_handle = igt_create_bo_with_dimensions(importer_fd, fb->width, fb->height, > + fb->drm_format, fb->modifier, pitch, &fb_size, NULL, NULL); > + igt_assert(fb->gem_handle > 0); > + > + if (is_i915_device(importer_fd)) { > + ahnd = get_reloc_ahnd(importer_fd, 0); > > igt_blitter_src_copy(importer_fd, ahnd, 0, NULL, temp_buf_handle, > 0, pitch, fb->modifier, 0, 0, fb_size, fb->width, > @@ -205,7 +273,62 @@ static void import_fb(int importer_fd, struct igt_fb *fb, > gem_close(importer_fd, temp_buf_handle); > put_ahnd(ahnd); > } else { > - fb->gem_handle = prime_fd_to_handle(importer_fd, dmabuf_fd); > + uint32_t xe_bb; > + uint64_t bb_size = 4096; > + struct blt_copy_data blt = {}; > + struct blt_copy_object *src, *dst; > + struct blt_block_copy_data_ext ext = {}; > + uint32_t mem_region; > + intel_ctx_t *xe_ctx; > + uint32_t vm, xe_exec; > + > + struct drm_xe_engine_class_instance inst = { > + .engine_class = DRM_XE_ENGINE_CLASS_COPY, > + }; > + vm = xe_vm_create(importer_fd, DRM_XE_VM_CREATE_FLAG_LR_MODE, 0); > + xe_exec = xe_exec_queue_create(importer_fd, vm, &inst, 0); > + xe_ctx = intel_ctx_xe(importer_fd, vm, xe_exec, 0, 0, 0); > + mem_region = vram_if_possible(importer_fd, 0); > + > + ahnd = intel_allocator_open_full(importer_fd, xe_ctx->vm, 0, 0, > + INTEL_ALLOCATOR_SIMPLE, > + ALLOC_STRATEGY_LOW_TO_HIGH, 0); > + > + bb_size = ALIGN(bb_size + xe_cs_prefetch_size(importer_fd), > + xe_get_default_alignment(importer_fd)); You can now use xe_bb_size(importer_fd, bb_size) here. And anywhere else where you need a correctly sized batch buffer. > + xe_bb = xe_bo_create(importer_fd, 0, bb_size, mem_region, 0); > + > + > + > + igt_init_fb(&dst_fb, importer_fd, fb->width, fb->height, > + DRM_FORMAT_XRGB8888, DRM_FORMAT_MOD_LINEAR, > + IGT_COLOR_YCBCR_BT709, IGT_COLOR_YCBCR_LIMITED_RANGE); > + dst_fb.gem_handle = temp_buf_handle; > + > + src = blt_fb_init(fb, mem_region, pitch); > + dst = blt_fb_init(&dst_fb, mem_region, pitch); > + > + blt_copy_init(importer_fd, &blt); > + blt.color_depth = 32; > + blt_set_copy_object(&blt.src, src); > + blt_set_copy_object(&blt.dst, dst); > + > + blt_set_object_ext(&ext.src, 0, fb->width, fb->height, > + SURFACE_TYPE_2D); > + blt_set_object_ext(&ext.dst, 0, fb->width, fb->height, > + SURFACE_TYPE_2D); > + > + blt_set_batch(&blt.bb, xe_bb, bb_size, mem_region); > + > + blt_block_copy(importer_fd, xe_ctx, NULL, ahnd, &blt, &ext); > + > + blt_destroy_object(importer_fd, dst); > + > + put_ahnd(ahnd); > + gem_close(importer_fd, xe_bb); > + xe_exec_queue_destroy(importer_fd, xe_exec); > + xe_vm_destroy(importer_fd, vm); > + free(xe_ctx); > } > } else { > fb->gem_handle = prime_fd_to_handle(importer_fd, dmabuf_fd); > @@ -332,7 +455,7 @@ static void test_basic_modeset(int drm_fd) > igt_device_set_master(drm_fd); > igt_display_require(&display, drm_fd); > > - output = setup_display(drm_fd, &display, &pipe); > + output = setup_hybrid_display(drm_fd, &display, &pipe); > mode = igt_output_get_mode(output); > igt_assert(mode); > > @@ -470,6 +593,8 @@ igt_main > igt_require(second_fd_vgem >= 0); > if (is_i915_device(first_fd)) > igt_require(!gem_has_lmem(first_fd)); > + if (is_xe_device(first_fd)) > + igt_require(!xe_has_vram(first_fd)); > } > > igt_describe("Make a dumb color buffer, export to another device and" > @@ -480,11 +605,5 @@ igt_main > igt_dynamic("second-to-first") > test_crc(second_fd_vgem, first_fd); > } > - > - igt_fixture > - drm_close_driver(second_fd_vgem); > } > - > - igt_fixture > - drm_close_driver(first_fd); > }