From patchwork Tue Nov 22 18:57:34 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Andi Shyti X-Patchwork-Id: 13052661 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 C4E32C433FE for ; Tue, 22 Nov 2022 18:58:01 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id F13D910E440; Tue, 22 Nov 2022 18:58:00 +0000 (UTC) Received: from mga02.intel.com (mga02.intel.com [134.134.136.20]) by gabe.freedesktop.org (Postfix) with ESMTPS id 0015010E440; Tue, 22 Nov 2022 18:57:56 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1669143477; x=1700679477; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=fODuMoinzOdDwjAfZvH/qIcFnv+R7LeV1DfmW4KLWfE=; b=eMNJYVmL/XhZywKlgMrNodsn6fu16WcFFIOo+gl4aEVSmEBfvkYUnfJS WGhRPYjYZd5tITdMllGgUr1MxznRZxvWcLQf/a2ZvACl2BOpsqvkAsiQ5 GW4y+REGtONzYjZWO3ZFIGSw4bCCdKDH5dKfSmCE+0AMK2VQv5D+dpxTj dFWTeRobmj3rHr0ac7giSJH26SJm3vcypvEgkFcSuUtpHHctAmSIMFtxc wCV8mdin5z0YQQzc01cR8AINlwjLMh9HXtcUA8J05uEEVTt/aKQxsLWZs Jek16HcRF8DBVjtLJ56sZs6lagRUl05BmxpztQiX+91pu/ieVWtDbP+wM A==; X-IronPort-AV: E=McAfee;i="6500,9779,10539"; a="301444292" X-IronPort-AV: E=Sophos;i="5.96,184,1665471600"; d="scan'208";a="301444292" Received: from orsmga001.jf.intel.com ([10.7.209.18]) by orsmga101.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 22 Nov 2022 10:57:54 -0800 X-IronPort-AV: E=McAfee;i="6500,9779,10539"; a="674448237" X-IronPort-AV: E=Sophos;i="5.96,184,1665471600"; d="scan'208";a="674448237" Received: from fuhrberg-mobl1.ger.corp.intel.com (HELO intel.com) ([10.251.209.136]) by orsmga001-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 22 Nov 2022 10:57:51 -0800 From: Andi Shyti To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org Date: Tue, 22 Nov 2022 19:57:34 +0100 Message-Id: <20221122185737.96459-2-andi.shyti@linux.intel.com> X-Mailer: git-send-email 2.38.1 In-Reply-To: <20221122185737.96459-1-andi.shyti@linux.intel.com> References: <20221122185737.96459-1-andi.shyti@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH v2 1/4] drm/i915: Wrap all access to i915_vma.node.start|size X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: =?utf-8?q?Thomas_Hellstr=C3=B6m?= , Chris Wilson , Matthew Auld Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" From: Chris Wilson We already wrap i915_vma.node.start for use with the GGTT, as there we can perform additional sanity checks that the node belongs to the GGTT and fits within the 32b registers. In the next couple of patches, we will introduce guard pages around the objects _inside_ the drm_mm_node allocation. That is we will offset the vma->pages so that the first page is at drm_mm_node.start + vma->guard (not 0 as is currently the case). All users must then not use i915_vma.node.start directly, but compute the guard offset, thus all users are converted to use a i915_vma_offset() wrapper. The notable exceptions are the selftests that are testing exact behaviour of i915_vma_pin/i915_vma_insert. Signed-off-by: Chris Wilson Signed-off-by: Tejas Upadhyay Co-developed-by: Thomas Hellström Signed-off-by: Thomas Hellström Signed-off-by: Tvrtko Ursulin Signed-off-by: Andi Shyti Reviewed-by: Tvrtko Ursulin --- drivers/gpu/drm/i915/display/intel_fbdev.c | 2 +- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 33 ++++++------ drivers/gpu/drm/i915/gem/i915_gem_mman.c | 2 +- drivers/gpu/drm/i915/gem/i915_gem_shrinker.c | 2 +- drivers/gpu/drm/i915/gem/i915_gem_tiling.c | 4 +- .../gpu/drm/i915/gem/selftests/huge_pages.c | 2 +- .../i915/gem/selftests/i915_gem_client_blt.c | 23 +++++---- .../drm/i915/gem/selftests/i915_gem_context.c | 15 +++--- .../drm/i915/gem/selftests/i915_gem_mman.c | 2 +- .../drm/i915/gem/selftests/igt_gem_utils.c | 7 +-- drivers/gpu/drm/i915/gt/gen7_renderclear.c | 2 +- drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c | 3 +- drivers/gpu/drm/i915/gt/intel_renderstate.c | 2 +- .../gpu/drm/i915/gt/intel_ring_submission.c | 2 +- drivers/gpu/drm/i915/gt/selftest_engine_cs.c | 8 +-- drivers/gpu/drm/i915/gt/selftest_execlists.c | 18 +++---- drivers/gpu/drm/i915/gt/selftest_hangcheck.c | 15 +++--- drivers/gpu/drm/i915/gt/selftest_lrc.c | 16 +++--- .../drm/i915/gt/selftest_ring_submission.c | 2 +- drivers/gpu/drm/i915/gt/selftest_rps.c | 12 ++--- .../gpu/drm/i915/gt/selftest_workarounds.c | 8 +-- drivers/gpu/drm/i915/i915_cmd_parser.c | 4 +- drivers/gpu/drm/i915/i915_debugfs.c | 2 +- drivers/gpu/drm/i915/i915_perf.c | 2 +- drivers/gpu/drm/i915/i915_vma.c | 25 ++++----- drivers/gpu/drm/i915/i915_vma.h | 51 +++++++++++++++++-- drivers/gpu/drm/i915/i915_vma_resource.h | 10 ++-- drivers/gpu/drm/i915/selftests/i915_request.c | 20 ++++---- drivers/gpu/drm/i915/selftests/igt_spinner.c | 8 +-- 29 files changed, 180 insertions(+), 122 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_fbdev.c b/drivers/gpu/drm/i915/display/intel_fbdev.c index 5575d7abdc092..03ed4607a46d2 100644 --- a/drivers/gpu/drm/i915/display/intel_fbdev.c +++ b/drivers/gpu/drm/i915/display/intel_fbdev.c @@ -286,7 +286,7 @@ static int intelfb_create(struct drm_fb_helper *helper, /* Our framebuffer is the entirety of fbdev's system memory */ info->fix.smem_start = - (unsigned long)(ggtt->gmadr.start + vma->node.start); + (unsigned long)(ggtt->gmadr.start + i915_ggtt_offset(vma)); info->fix.smem_len = vma->size; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 29e9e8d5b6fec..86956b902c978 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -379,22 +379,25 @@ eb_vma_misplaced(const struct drm_i915_gem_exec_object2 *entry, const struct i915_vma *vma, unsigned int flags) { - if (vma->node.size < entry->pad_to_size) + const u64 start = i915_vma_offset(vma); + const u64 size = i915_vma_size(vma); + + if (size < entry->pad_to_size) return true; - if (entry->alignment && !IS_ALIGNED(vma->node.start, entry->alignment)) + if (entry->alignment && !IS_ALIGNED(start, entry->alignment)) return true; if (flags & EXEC_OBJECT_PINNED && - vma->node.start != entry->offset) + start != entry->offset) return true; if (flags & __EXEC_OBJECT_NEEDS_BIAS && - vma->node.start < BATCH_OFFSET_BIAS) + start < BATCH_OFFSET_BIAS) return true; if (!(flags & EXEC_OBJECT_SUPPORTS_48B_ADDRESS) && - (vma->node.start + vma->node.size + 4095) >> 32) + (start + size + 4095) >> 32) return true; if (flags & __EXEC_OBJECT_NEEDS_MAP && @@ -440,7 +443,7 @@ eb_pin_vma(struct i915_execbuffer *eb, int err; if (vma->node.size) - pin_flags = vma->node.start; + pin_flags = __i915_vma_offset(vma); else pin_flags = entry->offset & PIN_OFFSET_MASK; @@ -663,8 +666,8 @@ static int eb_reserve_vma(struct i915_execbuffer *eb, if (err) return err; - if (entry->offset != vma->node.start) { - entry->offset = vma->node.start | UPDATE; + if (entry->offset != i915_vma_offset(vma)) { + entry->offset = i915_vma_offset(vma) | UPDATE; eb->args->flags |= __EXEC_HAS_RELOC; } @@ -984,8 +987,8 @@ static int eb_validate_vmas(struct i915_execbuffer *eb) return err; if (!err) { - if (entry->offset != vma->node.start) { - entry->offset = vma->node.start | UPDATE; + if (entry->offset != i915_vma_offset(vma)) { + entry->offset = i915_vma_offset(vma) | UPDATE; eb->args->flags |= __EXEC_HAS_RELOC; } } else { @@ -1066,7 +1069,7 @@ static inline u64 relocation_target(const struct drm_i915_gem_relocation_entry *reloc, const struct i915_vma *target) { - return gen8_canonical_addr((int)reloc->delta + target->node.start); + return gen8_canonical_addr((int)reloc->delta + i915_vma_offset(target)); } static void reloc_cache_init(struct reloc_cache *cache, @@ -1275,7 +1278,7 @@ static void *reloc_iomap(struct i915_vma *batch, if (err) /* no inactive aperture space, use cpu reloc */ return NULL; } else { - cache->node.start = vma->node.start; + cache->node.start = i915_ggtt_offset(vma); cache->node.mm = (void *)vma; } } @@ -1438,7 +1441,7 @@ eb_relocate_entry(struct i915_execbuffer *eb, * more work needs to be done. */ if (!DBG_FORCE_RELOC && - gen8_canonical_addr(target->vma->node.start) == reloc->presumed_offset) + gen8_canonical_addr(i915_vma_offset(target->vma)) == reloc->presumed_offset) return 0; /* Check that the relocation address is valid... */ @@ -2368,7 +2371,7 @@ static int eb_request_submit(struct i915_execbuffer *eb, } err = rq->context->engine->emit_bb_start(rq, - batch->node.start + + i915_vma_offset(batch) + eb->batch_start_offset, batch_len, eb->batch_flags); @@ -2379,7 +2382,7 @@ static int eb_request_submit(struct i915_execbuffer *eb, GEM_BUG_ON(intel_context_is_parallel(rq->context)); GEM_BUG_ON(eb->batch_start_offset); err = rq->context->engine->emit_bb_start(rq, - eb->trampoline->node.start + + i915_vma_offset(eb->trampoline) + batch_len, 0, 0); if (err) return err; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_mman.c b/drivers/gpu/drm/i915/gem/i915_gem_mman.c index c29efdef8313a..d73ba0f5c4c52 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_mman.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_mman.c @@ -395,7 +395,7 @@ static vm_fault_t vm_fault_gtt(struct vm_fault *vmf) /* Finally, remap it using the new GTT offset */ ret = remap_io_mapping(area, area->vm_start + (vma->gtt_view.partial.offset << PAGE_SHIFT), - (ggtt->gmadr.start + vma->node.start) >> PAGE_SHIFT, + (ggtt->gmadr.start + i915_ggtt_offset(vma)) >> PAGE_SHIFT, min_t(u64, vma->size, area->vm_end - area->vm_start), &ggtt->iomap); if (ret) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_shrinker.c b/drivers/gpu/drm/i915/gem/i915_gem_shrinker.c index 8dc5c8874d8a2..b1672e054b21e 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_shrinker.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_shrinker.c @@ -400,7 +400,7 @@ i915_gem_shrinker_vmap(struct notifier_block *nb, unsigned long event, void *ptr mutex_lock(&to_gt(i915)->ggtt->vm.mutex); list_for_each_entry_safe(vma, next, &to_gt(i915)->ggtt->vm.bound_list, vm_link) { - unsigned long count = vma->node.size >> PAGE_SHIFT; + unsigned long count = i915_vma_size(vma) >> PAGE_SHIFT; struct drm_i915_gem_object *obj = vma->obj; if (!vma->iomap || i915_vma_is_active(vma)) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_tiling.c b/drivers/gpu/drm/i915/gem/i915_gem_tiling.c index fd42b89b7162b..04bb909acdec8 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_tiling.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_tiling.c @@ -168,11 +168,11 @@ static bool i915_vma_fence_prepare(struct i915_vma *vma, return true; size = i915_gem_fence_size(i915, vma->size, tiling_mode, stride); - if (vma->node.size < size) + if (i915_vma_size(vma) < size) return false; alignment = i915_gem_fence_alignment(i915, vma->size, tiling_mode, stride); - if (!IS_ALIGNED(vma->node.start, alignment)) + if (!IS_ALIGNED(i915_ggtt_offset(vma), alignment)) return false; return true; diff --git a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c index beaf27e09e8a9..e0c2ac9c8053a 100644 --- a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c +++ b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c @@ -400,7 +400,7 @@ static int igt_check_page_sizes(struct i915_vma *vma) * Maintaining alignment is required to utilise huge pages in the ppGGT. */ if (i915_gem_object_is_lmem(obj) && - IS_ALIGNED(vma->node.start, SZ_2M) && + IS_ALIGNED(i915_vma_offset(vma), SZ_2M) && vma->page_sizes.sg & SZ_2M && vma->resource->page_sizes_gtt < SZ_2M) { pr_err("gtt pages mismatch for LMEM, expected 2M GTT pages, sg(%u), gtt(%u)\n", diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c index 692a16914ca0f..3bb1f7f0110e6 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c @@ -194,12 +194,12 @@ static int prepare_blit(const struct tiled_blits *t, *cs++ = src_4t | dst_4t | BLT_DEPTH_32 | dst_pitch; *cs++ = 0; *cs++ = t->height << 16 | t->width; - *cs++ = lower_32_bits(dst->vma->node.start); - *cs++ = upper_32_bits(dst->vma->node.start); + *cs++ = lower_32_bits(i915_vma_offset(dst->vma)); + *cs++ = upper_32_bits(i915_vma_offset(dst->vma)); *cs++ = 0; *cs++ = src_pitch; - *cs++ = lower_32_bits(src->vma->node.start); - *cs++ = upper_32_bits(src->vma->node.start); + *cs++ = lower_32_bits(i915_vma_offset(src->vma)); + *cs++ = upper_32_bits(i915_vma_offset(src->vma)); } else { if (ver >= 6) { *cs++ = MI_LOAD_REGISTER_IMM(1); @@ -240,14 +240,14 @@ static int prepare_blit(const struct tiled_blits *t, *cs++ = BLT_DEPTH_32 | BLT_ROP_SRC_COPY | dst_pitch; *cs++ = 0; *cs++ = t->height << 16 | t->width; - *cs++ = lower_32_bits(dst->vma->node.start); + *cs++ = lower_32_bits(i915_vma_offset(dst->vma)); if (use_64b_reloc) - *cs++ = upper_32_bits(dst->vma->node.start); + *cs++ = upper_32_bits(i915_vma_offset(dst->vma)); *cs++ = 0; *cs++ = src_pitch; - *cs++ = lower_32_bits(src->vma->node.start); + *cs++ = lower_32_bits(i915_vma_offset(src->vma)); if (use_64b_reloc) - *cs++ = upper_32_bits(src->vma->node.start); + *cs++ = upper_32_bits(i915_vma_offset(src->vma)); } *cs++ = MI_BATCH_BUFFER_END; @@ -462,7 +462,7 @@ static int pin_buffer(struct i915_vma *vma, u64 addr) { int err; - if (drm_mm_node_allocated(&vma->node) && vma->node.start != addr) { + if (drm_mm_node_allocated(&vma->node) && i915_vma_offset(vma) != addr) { err = i915_vma_unbind_unlocked(vma); if (err) return err; @@ -472,6 +472,7 @@ static int pin_buffer(struct i915_vma *vma, u64 addr) if (err) return err; + GEM_BUG_ON(i915_vma_offset(vma) != addr); return 0; } @@ -518,8 +519,8 @@ tiled_blit(struct tiled_blits *t, err = igt_vma_move_to_active_unlocked(dst->vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, - t->batch->node.start, - t->batch->node.size, + i915_vma_offset(t->batch), + i915_vma_size(t->batch), 0); i915_request_get(rq); i915_request_add(rq); diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c index a0ff51d71d07a..ac02fb0365920 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c @@ -914,8 +914,8 @@ static int rpcs_query_batch(struct drm_i915_gem_object *rpcs, *cmd++ = MI_STORE_REGISTER_MEM_GEN8; *cmd++ = i915_mmio_reg_offset(GEN8_R_PWR_CLK_STATE(engine->mmio_base)); - *cmd++ = lower_32_bits(vma->node.start); - *cmd++ = upper_32_bits(vma->node.start); + *cmd++ = lower_32_bits(i915_vma_offset(vma)); + *cmd++ = upper_32_bits(i915_vma_offset(vma)); *cmd = MI_BATCH_BUFFER_END; __i915_gem_object_flush_map(rpcs, 0, 64); @@ -999,7 +999,8 @@ emit_rpcs_query(struct drm_i915_gem_object *obj, } err = rq->engine->emit_bb_start(rq, - batch->node.start, batch->node.size, + i915_vma_offset(batch), + i915_vma_size(batch), 0); if (err) goto skip_request; @@ -1560,7 +1561,8 @@ static int write_to_scratch(struct i915_gem_context *ctx, goto skip_request; } - err = engine->emit_bb_start(rq, vma->node.start, vma->node.size, 0); + err = engine->emit_bb_start(rq, i915_vma_offset(vma), + i915_vma_size(vma), 0); if (err) goto skip_request; @@ -1665,7 +1667,7 @@ static int read_from_scratch(struct i915_gem_context *ctx, *cmd++ = offset; *cmd++ = MI_STORE_REGISTER_MEM | MI_USE_GGTT; *cmd++ = reg; - *cmd++ = vma->node.start + result; + *cmd++ = i915_vma_offset(vma) + result; *cmd = MI_BATCH_BUFFER_END; i915_gem_object_flush_map(obj); @@ -1694,7 +1696,8 @@ static int read_from_scratch(struct i915_gem_context *ctx, goto skip_request; } - err = engine->emit_bb_start(rq, vma->node.start, vma->node.size, flags); + err = engine->emit_bb_start(rq, i915_vma_offset(vma), + i915_vma_size(vma), flags); if (err) goto skip_request; diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c index 3f658d5717d83..7f6353827735a 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c @@ -1609,7 +1609,7 @@ static int __igt_mmap_gpu(struct drm_i915_private *i915, err = i915_vma_move_to_active(vma, rq, 0); - err = engine->emit_bb_start(rq, vma->node.start, 0, 0); + err = engine->emit_bb_start(rq, i915_vma_offset(vma), 0, 0); i915_request_get(rq); i915_request_add(rq); diff --git a/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c b/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c index 374b10ac430e8..c147038230c46 100644 --- a/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c +++ b/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c @@ -62,8 +62,8 @@ igt_emit_store_dw(struct i915_vma *vma, goto err; } - GEM_BUG_ON(offset + (count - 1) * PAGE_SIZE > vma->node.size); - offset += vma->node.start; + GEM_BUG_ON(offset + (count - 1) * PAGE_SIZE > i915_vma_size(vma)); + offset += i915_vma_offset(vma); for (n = 0; n < count; n++) { if (ver >= 8) { @@ -147,7 +147,8 @@ int igt_gpu_fill_dw(struct intel_context *ce, flags |= I915_DISPATCH_SECURE; err = rq->engine->emit_bb_start(rq, - batch->node.start, batch->node.size, + i915_vma_offset(batch), + i915_vma_size(batch), flags); skip_request: diff --git a/drivers/gpu/drm/i915/gt/gen7_renderclear.c b/drivers/gpu/drm/i915/gt/gen7_renderclear.c index 317efb145787e..d38b914d12069 100644 --- a/drivers/gpu/drm/i915/gt/gen7_renderclear.c +++ b/drivers/gpu/drm/i915/gt/gen7_renderclear.c @@ -106,7 +106,7 @@ static u32 batch_offset(const struct batch_chunk *bc, u32 *cs) static u32 batch_addr(const struct batch_chunk *bc) { - return bc->vma->node.start; + return i915_vma_offset(bc->vma); } static void batch_add(struct batch_chunk *bc, const u32 d) diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c b/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c index 995082d45cb21..7ac8ed13e1fe3 100644 --- a/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c +++ b/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c @@ -220,7 +220,8 @@ static int fence_update(struct i915_fence_reg *fence, return ret; } - fence->start = vma->node.start; + GEM_BUG_ON(vma->fence_size > i915_vma_size(vma)); + fence->start = i915_ggtt_offset(vma); fence->size = vma->fence_size; fence->stride = i915_gem_object_get_stride(vma->obj); fence->tiling = i915_gem_object_get_tiling(vma->obj); diff --git a/drivers/gpu/drm/i915/gt/intel_renderstate.c b/drivers/gpu/drm/i915/gt/intel_renderstate.c index 9c1ae070ee7b9..4b56ec3743cf3 100644 --- a/drivers/gpu/drm/i915/gt/intel_renderstate.c +++ b/drivers/gpu/drm/i915/gt/intel_renderstate.c @@ -63,7 +63,7 @@ static int render_state_setup(struct intel_renderstate *so, u32 s = rodata->batch[i]; if (i * 4 == rodata->reloc[reloc_index]) { - u64 r = s + so->vma->node.start; + u64 r = s + i915_vma_offset(so->vma); s = lower_32_bits(r); if (HAS_64BIT_RELOC(i915)) { diff --git a/drivers/gpu/drm/i915/gt/intel_ring_submission.c b/drivers/gpu/drm/i915/gt/intel_ring_submission.c index 356c787e11d38..827adb0cfaea6 100644 --- a/drivers/gpu/drm/i915/gt/intel_ring_submission.c +++ b/drivers/gpu/drm/i915/gt/intel_ring_submission.c @@ -897,7 +897,7 @@ static int clear_residuals(struct i915_request *rq) } ret = engine->emit_bb_start(rq, - engine->wa_ctx.vma->node.start, 0, + i915_vma_offset(engine->wa_ctx.vma), 0, 0); if (ret) return ret; diff --git a/drivers/gpu/drm/i915/gt/selftest_engine_cs.c b/drivers/gpu/drm/i915/gt/selftest_engine_cs.c index 881b64f3e7b99..542ce6d2de192 100644 --- a/drivers/gpu/drm/i915/gt/selftest_engine_cs.c +++ b/drivers/gpu/drm/i915/gt/selftest_engine_cs.c @@ -178,7 +178,7 @@ static int perf_mi_bb_start(void *arg) goto out; err = rq->engine->emit_bb_start(rq, - batch->node.start, 8, + i915_vma_offset(batch), 8, 0); if (err) goto out; @@ -321,7 +321,7 @@ static int perf_mi_noop(void *arg) goto out; err = rq->engine->emit_bb_start(rq, - base->node.start, 8, + i915_vma_offset(base), 8, 0); if (err) goto out; @@ -331,8 +331,8 @@ static int perf_mi_noop(void *arg) goto out; err = rq->engine->emit_bb_start(rq, - nop->node.start, - nop->node.size, + i915_vma_offset(nop), + i915_vma_size(nop), 0); if (err) goto out; diff --git a/drivers/gpu/drm/i915/gt/selftest_execlists.c b/drivers/gpu/drm/i915/gt/selftest_execlists.c index ab2e9a6a24522..a619057543fd9 100644 --- a/drivers/gpu/drm/i915/gt/selftest_execlists.c +++ b/drivers/gpu/drm/i915/gt/selftest_execlists.c @@ -2737,11 +2737,11 @@ static int create_gang(struct intel_engine_cs *engine, MI_SEMAPHORE_POLL | MI_SEMAPHORE_SAD_EQ_SDD; *cs++ = 0; - *cs++ = lower_32_bits(vma->node.start); - *cs++ = upper_32_bits(vma->node.start); + *cs++ = lower_32_bits(i915_vma_offset(vma)); + *cs++ = upper_32_bits(i915_vma_offset(vma)); if (*prev) { - u64 offset = (*prev)->batch->node.start; + u64 offset = i915_vma_offset((*prev)->batch); /* Terminate the spinner in the next lower priority batch. */ *cs++ = MI_STORE_DWORD_IMM_GEN4; @@ -2767,7 +2767,7 @@ static int create_gang(struct intel_engine_cs *engine, err = i915_vma_move_to_active(vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, - vma->node.start, + i915_vma_offset(vma), PAGE_SIZE, 0); i915_vma_unlock(vma); i915_request_add(rq); @@ -3095,7 +3095,7 @@ create_gpr_user(struct intel_engine_cs *engine, *cs++ = MI_MATH_ADD; *cs++ = MI_MATH_STORE(MI_MATH_REG(i), MI_MATH_REG_ACCU); - addr = result->node.start + offset + i * sizeof(*cs); + addr = i915_vma_offset(result) + offset + i * sizeof(*cs); *cs++ = MI_STORE_REGISTER_MEM_GEN8; *cs++ = CS_GPR(engine, 2 * i); *cs++ = lower_32_bits(addr); @@ -3105,8 +3105,8 @@ create_gpr_user(struct intel_engine_cs *engine, MI_SEMAPHORE_POLL | MI_SEMAPHORE_SAD_GTE_SDD; *cs++ = i; - *cs++ = lower_32_bits(result->node.start); - *cs++ = upper_32_bits(result->node.start); + *cs++ = lower_32_bits(i915_vma_offset(result)); + *cs++ = upper_32_bits(i915_vma_offset(result)); } *cs++ = MI_BATCH_BUFFER_END; @@ -3186,7 +3186,7 @@ create_gpr_client(struct intel_engine_cs *engine, err = i915_vma_move_to_active(batch, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, - batch->node.start, + i915_vma_offset(batch), PAGE_SIZE, 0); i915_vma_unlock(batch); i915_vma_unpin(batch); @@ -3518,7 +3518,7 @@ static int smoke_submit(struct preempt_smoke *smoke, err = i915_vma_move_to_active(vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, - vma->node.start, + i915_vma_offset(vma), PAGE_SIZE, 0); i915_vma_unlock(vma); } diff --git a/drivers/gpu/drm/i915/gt/selftest_hangcheck.c b/drivers/gpu/drm/i915/gt/selftest_hangcheck.c index bc05ef48c194a..8b0d84f2aad22 100644 --- a/drivers/gpu/drm/i915/gt/selftest_hangcheck.c +++ b/drivers/gpu/drm/i915/gt/selftest_hangcheck.c @@ -96,7 +96,8 @@ static int hang_init(struct hang *h, struct intel_gt *gt) static u64 hws_address(const struct i915_vma *hws, const struct i915_request *rq) { - return hws->node.start + offset_in_page(sizeof(u32)*rq->fence.context); + return i915_vma_offset(hws) + + offset_in_page(sizeof(u32) * rq->fence.context); } static struct i915_request * @@ -180,8 +181,8 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) *batch++ = MI_NOOP; *batch++ = MI_BATCH_BUFFER_START | 1 << 8 | 1; - *batch++ = lower_32_bits(vma->node.start); - *batch++ = upper_32_bits(vma->node.start); + *batch++ = lower_32_bits(i915_vma_offset(vma)); + *batch++ = upper_32_bits(i915_vma_offset(vma)); } else if (GRAPHICS_VER(gt->i915) >= 6) { *batch++ = MI_STORE_DWORD_IMM_GEN4; *batch++ = 0; @@ -194,7 +195,7 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) *batch++ = MI_NOOP; *batch++ = MI_BATCH_BUFFER_START | 1 << 8; - *batch++ = lower_32_bits(vma->node.start); + *batch++ = lower_32_bits(i915_vma_offset(vma)); } else if (GRAPHICS_VER(gt->i915) >= 4) { *batch++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT; *batch++ = 0; @@ -207,7 +208,7 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) *batch++ = MI_NOOP; *batch++ = MI_BATCH_BUFFER_START | 2 << 6; - *batch++ = lower_32_bits(vma->node.start); + *batch++ = lower_32_bits(i915_vma_offset(vma)); } else { *batch++ = MI_STORE_DWORD_IMM | MI_MEM_VIRTUAL; *batch++ = lower_32_bits(hws_address(hws, rq)); @@ -219,7 +220,7 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) *batch++ = MI_NOOP; *batch++ = MI_BATCH_BUFFER_START | 2 << 6; - *batch++ = lower_32_bits(vma->node.start); + *batch++ = lower_32_bits(i915_vma_offset(vma)); } *batch++ = MI_BATCH_BUFFER_END; /* not reached */ intel_gt_chipset_flush(engine->gt); @@ -234,7 +235,7 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) if (GRAPHICS_VER(gt->i915) <= 5) flags |= I915_DISPATCH_SECURE; - err = rq->engine->emit_bb_start(rq, vma->node.start, PAGE_SIZE, flags); + err = rq->engine->emit_bb_start(rq, i915_vma_offset(vma), PAGE_SIZE, flags); cancel_rq: if (err) { diff --git a/drivers/gpu/drm/i915/gt/selftest_lrc.c b/drivers/gpu/drm/i915/gt/selftest_lrc.c index 7c56ffd2c6597..a61ae9d7e0a22 100644 --- a/drivers/gpu/drm/i915/gt/selftest_lrc.c +++ b/drivers/gpu/drm/i915/gt/selftest_lrc.c @@ -1030,8 +1030,8 @@ store_context(struct intel_context *ce, struct i915_vma *scratch) while (len--) { *cs++ = MI_STORE_REGISTER_MEM_GEN8; *cs++ = hw[dw]; - *cs++ = lower_32_bits(scratch->node.start + x); - *cs++ = upper_32_bits(scratch->node.start + x); + *cs++ = lower_32_bits(i915_vma_offset(scratch) + x); + *cs++ = upper_32_bits(i915_vma_offset(scratch) + x); dw += 2; x += 4; @@ -1098,8 +1098,8 @@ record_registers(struct intel_context *ce, *cs++ = MI_ARB_ON_OFF | MI_ARB_DISABLE; *cs++ = MI_BATCH_BUFFER_START_GEN8 | BIT(8); - *cs++ = lower_32_bits(b_before->node.start); - *cs++ = upper_32_bits(b_before->node.start); + *cs++ = lower_32_bits(i915_vma_offset(b_before)); + *cs++ = upper_32_bits(i915_vma_offset(b_before)); *cs++ = MI_ARB_ON_OFF | MI_ARB_ENABLE; *cs++ = MI_SEMAPHORE_WAIT | @@ -1114,8 +1114,8 @@ record_registers(struct intel_context *ce, *cs++ = MI_ARB_ON_OFF | MI_ARB_DISABLE; *cs++ = MI_BATCH_BUFFER_START_GEN8 | BIT(8); - *cs++ = lower_32_bits(b_after->node.start); - *cs++ = upper_32_bits(b_after->node.start); + *cs++ = lower_32_bits(i915_vma_offset(b_after)); + *cs++ = upper_32_bits(i915_vma_offset(b_after)); intel_ring_advance(rq, cs); @@ -1236,8 +1236,8 @@ static int poison_registers(struct intel_context *ce, u32 poison, u32 *sema) *cs++ = MI_ARB_ON_OFF | MI_ARB_DISABLE; *cs++ = MI_BATCH_BUFFER_START_GEN8 | BIT(8); - *cs++ = lower_32_bits(batch->node.start); - *cs++ = upper_32_bits(batch->node.start); + *cs++ = lower_32_bits(i915_vma_offset(batch)); + *cs++ = upper_32_bits(i915_vma_offset(batch)); *cs++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT; *cs++ = i915_ggtt_offset(ce->engine->status_page.vma) + diff --git a/drivers/gpu/drm/i915/gt/selftest_ring_submission.c b/drivers/gpu/drm/i915/gt/selftest_ring_submission.c index 70f9ac1ec2c76..87ceb0f374b67 100644 --- a/drivers/gpu/drm/i915/gt/selftest_ring_submission.c +++ b/drivers/gpu/drm/i915/gt/selftest_ring_submission.c @@ -50,7 +50,7 @@ static struct i915_vma *create_wally(struct intel_engine_cs *engine) } else { *cs++ = MI_STORE_DWORD_IMM | MI_MEM_VIRTUAL; } - *cs++ = vma->node.start + 4000; + *cs++ = i915_vma_offset(vma) + 4000; *cs++ = STACK_MAGIC; *cs++ = MI_BATCH_BUFFER_END; diff --git a/drivers/gpu/drm/i915/gt/selftest_rps.c b/drivers/gpu/drm/i915/gt/selftest_rps.c index 39f1b7564170c..6755bbc4ebdac 100644 --- a/drivers/gpu/drm/i915/gt/selftest_rps.c +++ b/drivers/gpu/drm/i915/gt/selftest_rps.c @@ -122,14 +122,14 @@ create_spin_counter(struct intel_engine_cs *engine, if (srm) { *cs++ = MI_STORE_REGISTER_MEM_GEN8; *cs++ = i915_mmio_reg_offset(CS_GPR(COUNT)); - *cs++ = lower_32_bits(vma->node.start + end * sizeof(*cs)); - *cs++ = upper_32_bits(vma->node.start + end * sizeof(*cs)); + *cs++ = lower_32_bits(i915_vma_offset(vma) + end * sizeof(*cs)); + *cs++ = upper_32_bits(i915_vma_offset(vma) + end * sizeof(*cs)); } } *cs++ = MI_BATCH_BUFFER_START_GEN8; - *cs++ = lower_32_bits(vma->node.start + loop * sizeof(*cs)); - *cs++ = upper_32_bits(vma->node.start + loop * sizeof(*cs)); + *cs++ = lower_32_bits(i915_vma_offset(vma) + loop * sizeof(*cs)); + *cs++ = upper_32_bits(i915_vma_offset(vma) + loop * sizeof(*cs)); GEM_BUG_ON(cs - base > end); i915_gem_object_flush_map(obj); @@ -655,7 +655,7 @@ int live_rps_frequency_cs(void *arg) err = i915_vma_move_to_active(vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, - vma->node.start, + i915_vma_offset(vma), PAGE_SIZE, 0); i915_request_add(rq); if (err) @@ -794,7 +794,7 @@ int live_rps_frequency_srm(void *arg) err = i915_vma_move_to_active(vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, - vma->node.start, + i915_vma_offset(vma), PAGE_SIZE, 0); i915_request_add(rq); if (err) diff --git a/drivers/gpu/drm/i915/gt/selftest_workarounds.c b/drivers/gpu/drm/i915/gt/selftest_workarounds.c index 96e3861706d6f..9c5449709161c 100644 --- a/drivers/gpu/drm/i915/gt/selftest_workarounds.c +++ b/drivers/gpu/drm/i915/gt/selftest_workarounds.c @@ -521,7 +521,7 @@ static int check_dirty_whitelist(struct intel_context *ce) for (i = 0; i < engine->whitelist.count; i++) { u32 reg = i915_mmio_reg_offset(engine->whitelist.list[i].reg); struct i915_gem_ww_ctx ww; - u64 addr = scratch->node.start; + u64 addr = i915_vma_offset(scratch); struct i915_request *rq; u32 srm, lrm, rsvd; u32 expect; @@ -640,7 +640,7 @@ static int check_dirty_whitelist(struct intel_context *ce) goto err_request; err = engine->emit_bb_start(rq, - batch->node.start, PAGE_SIZE, + i915_vma_offset(batch), PAGE_SIZE, 0); if (err) goto err_request; @@ -870,7 +870,7 @@ static int read_whitelisted_registers(struct intel_context *ce, } for (i = 0; i < engine->whitelist.count; i++) { - u64 offset = results->node.start + sizeof(u32) * i; + u64 offset = i915_vma_offset(results) + sizeof(u32) * i; u32 reg = i915_mmio_reg_offset(engine->whitelist.list[i].reg); /* Clear non priv flags */ @@ -942,7 +942,7 @@ static int scrub_whitelisted_registers(struct intel_context *ce) goto err_request; /* Perform the writes from an unprivileged "user" batch */ - err = engine->emit_bb_start(rq, batch->node.start, 0, 0); + err = engine->emit_bb_start(rq, i915_vma_offset(batch), 0, 0); err_request: err = request_add_sync(rq, err); diff --git a/drivers/gpu/drm/i915/i915_cmd_parser.c b/drivers/gpu/drm/i915/i915_cmd_parser.c index f93e6122f2478..ddf49c2dbb917 100644 --- a/drivers/gpu/drm/i915/i915_cmd_parser.c +++ b/drivers/gpu/drm/i915/i915_cmd_parser.c @@ -1471,8 +1471,8 @@ int intel_engine_cmd_parser(struct intel_engine_cs *engine, /* Defer failure until attempted use */ jump_whitelist = alloc_whitelist(batch_length); - shadow_addr = gen8_canonical_addr(shadow->node.start); - batch_addr = gen8_canonical_addr(batch->node.start + batch_offset); + shadow_addr = gen8_canonical_addr(i915_vma_offset(shadow)); + batch_addr = gen8_canonical_addr(i915_vma_offset(batch) + batch_offset); /* * We use the batch length as size because the shadow object is as diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c index 6c7ac73b69a52..a356ca490159e 100644 --- a/drivers/gpu/drm/i915/i915_debugfs.c +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -183,7 +183,7 @@ i915_debugfs_describe_obj(struct seq_file *m, struct drm_i915_gem_object *obj) seq_printf(m, " (%s offset: %08llx, size: %08llx, pages: %s", stringify_vma_type(vma), - vma->node.start, vma->node.size, + i915_vma_offset(vma), i915_vma_size(vma), stringify_page_sizes(vma->resource->page_sizes_gtt, NULL, 0)); if (i915_vma_is_ggtt(vma) || i915_vma_is_dpt(vma)) { diff --git a/drivers/gpu/drm/i915/i915_perf.c b/drivers/gpu/drm/i915/i915_perf.c index 00e09bb18b13b..79a454a17c287 100644 --- a/drivers/gpu/drm/i915/i915_perf.c +++ b/drivers/gpu/drm/i915/i915_perf.c @@ -2259,7 +2259,7 @@ emit_oa_config(struct i915_perf_stream *stream, goto err_add_request; err = rq->engine->emit_bb_start(rq, - vma->node.start, 0, + i915_vma_offset(vma), 0, I915_DISPATCH_SECURE); if (err) goto err_add_request; diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c index 703fee6b5f752..2232118babeb3 100644 --- a/drivers/gpu/drm/i915/i915_vma.c +++ b/drivers/gpu/drm/i915/i915_vma.c @@ -418,8 +418,8 @@ i915_vma_resource_init_from_vma(struct i915_vma_resource *vma_res, i915_vma_resource_init(vma_res, vma->vm, vma->pages, &vma->page_sizes, obj->mm.rsgt, i915_gem_object_is_readonly(obj), i915_gem_object_is_lmem(obj), obj->mm.region, - vma->ops, vma->private, vma->node.start, - vma->node.size, vma->size); + vma->ops, vma->private, __i915_vma_offset(vma), + __i915_vma_size(vma), vma->size); } /** @@ -447,7 +447,7 @@ int i915_vma_bind(struct i915_vma *vma, lockdep_assert_held(&vma->vm->mutex); GEM_BUG_ON(!drm_mm_node_allocated(&vma->node)); - GEM_BUG_ON(vma->size > vma->node.size); + GEM_BUG_ON(vma->size > i915_vma_size(vma)); if (GEM_DEBUG_WARN_ON(range_overflows(vma->node.start, vma->node.size, @@ -569,8 +569,8 @@ void __iomem *i915_vma_pin_iomap(struct i915_vma *vma) vma->obj->base.size); } else if (i915_vma_is_map_and_fenceable(vma)) { ptr = io_mapping_map_wc(&i915_vm_to_ggtt(vma->vm)->iomap, - vma->node.start, - vma->node.size); + i915_vma_offset(vma), + i915_vma_size(vma)); } else { ptr = (void __iomem *) i915_gem_object_pin_map(vma->obj, I915_MAP_WC); @@ -659,22 +659,22 @@ bool i915_vma_misplaced(const struct i915_vma *vma, if (test_bit(I915_VMA_ERROR_BIT, __i915_vma_flags(vma))) return true; - if (vma->node.size < size) + if (i915_vma_size(vma) < size) return true; GEM_BUG_ON(alignment && !is_power_of_2(alignment)); - if (alignment && !IS_ALIGNED(vma->node.start, alignment)) + if (alignment && !IS_ALIGNED(i915_vma_offset(vma), alignment)) return true; if (flags & PIN_MAPPABLE && !i915_vma_is_map_and_fenceable(vma)) return true; if (flags & PIN_OFFSET_BIAS && - vma->node.start < (flags & PIN_OFFSET_MASK)) + i915_vma_offset(vma) < (flags & PIN_OFFSET_MASK)) return true; if (flags & PIN_OFFSET_FIXED && - vma->node.start != (flags & PIN_OFFSET_MASK)) + i915_vma_offset(vma) != (flags & PIN_OFFSET_MASK)) return true; return false; @@ -687,10 +687,11 @@ void __i915_vma_set_map_and_fenceable(struct i915_vma *vma) GEM_BUG_ON(!i915_vma_is_ggtt(vma)); GEM_BUG_ON(!vma->fence_size); - fenceable = (vma->node.size >= vma->fence_size && - IS_ALIGNED(vma->node.start, vma->fence_alignment)); + fenceable = (i915_vma_size(vma) >= vma->fence_size && + IS_ALIGNED(i915_vma_offset(vma), vma->fence_alignment)); - mappable = vma->node.start + vma->fence_size <= i915_vm_to_ggtt(vma->vm)->mappable_end; + mappable = i915_ggtt_offset(vma) + vma->fence_size <= + i915_vm_to_ggtt(vma->vm)->mappable_end; if (mappable && fenceable) set_bit(I915_VMA_CAN_FENCE_BIT, __i915_vma_flags(vma)); diff --git a/drivers/gpu/drm/i915/i915_vma.h b/drivers/gpu/drm/i915/i915_vma.h index 0757977a489bc..3fd4512b1f65f 100644 --- a/drivers/gpu/drm/i915/i915_vma.h +++ b/drivers/gpu/drm/i915/i915_vma.h @@ -125,13 +125,58 @@ static inline bool i915_vma_is_closed(const struct i915_vma *vma) return !list_empty(&vma->closed_link); } +/* Internal use only. */ +static inline u64 __i915_vma_size(const struct i915_vma *vma) +{ + return vma->node.size; +} + +/** + * i915_vma_offset - Obtain the va range size of the vma + * @vma: The vma + * + * GPU virtual address space may be allocated with padding. This + * function returns the effective virtual address range size + * with padding subtracted. + * + * Return: The effective virtual address range size. + */ +static inline u64 i915_vma_size(const struct i915_vma *vma) +{ + GEM_BUG_ON(!drm_mm_node_allocated(&vma->node)); + return __i915_vma_size(vma); +} + +/* Internal use only. */ +static inline u64 __i915_vma_offset(const struct i915_vma *vma) +{ + return vma->node.start; +} + +/** + * i915_vma_offset - Obtain the va offset of the vma + * @vma: The vma + * + * GPU virtual address space may be allocated with padding. This + * function returns the effective virtual address offset the gpu + * should use to access the bound data. + * + * Return: The effective virtual address offset. + */ +static inline u64 i915_vma_offset(const struct i915_vma *vma) +{ + GEM_BUG_ON(!drm_mm_node_allocated(&vma->node)); + return __i915_vma_offset(vma); +} + static inline u32 i915_ggtt_offset(const struct i915_vma *vma) { GEM_BUG_ON(!i915_vma_is_ggtt(vma)); GEM_BUG_ON(!drm_mm_node_allocated(&vma->node)); - GEM_BUG_ON(upper_32_bits(vma->node.start)); - GEM_BUG_ON(upper_32_bits(vma->node.start + vma->node.size - 1)); - return lower_32_bits(vma->node.start); + GEM_BUG_ON(upper_32_bits(i915_vma_offset(vma))); + GEM_BUG_ON(upper_32_bits(i915_vma_offset(vma) + + i915_vma_size(vma) - 1)); + return lower_32_bits(i915_vma_offset(vma)); } static inline u32 i915_ggtt_pin_bias(struct i915_vma *vma) diff --git a/drivers/gpu/drm/i915/i915_vma_resource.h b/drivers/gpu/drm/i915/i915_vma_resource.h index 06923d1816e7e..54edf3739ca0b 100644 --- a/drivers/gpu/drm/i915/i915_vma_resource.h +++ b/drivers/gpu/drm/i915/i915_vma_resource.h @@ -52,8 +52,10 @@ struct i915_page_sizes { * @mr: The memory region of the object pointed to by the vma. * @ops: Pointer to the backend i915_vma_ops. * @private: Bind backend private info. - * @start: Offset into the address space of bind range start. - * @node_size: Size of the allocated range manager node. + * @start: Offset into the address space of bind range start. Note that + * this is after any padding that might have been allocated. + * @node_size: Size of the allocated range manager node with padding + * subtracted. * @vma_size: Bind size. * @page_sizes_gtt: Resulting page sizes from the bind operation. * @bound_flags: Flags indicating binding status. @@ -174,8 +176,8 @@ static inline void i915_vma_resource_put(struct i915_vma_resource *vma_res) * @mr: The memory region of the object the vma points to. * @ops: The backend ops. * @private: Bind backend private info. - * @start: Offset into the address space of bind range start. - * @node_size: Size of the allocated range manager node. + * @start: Offset into the address space of bind range start after padding. + * @node_size: Size of the allocated range manager node minus padding. * @size: Bind size. * * Initializes a vma resource allocated using i915_vma_resource_alloc(). diff --git a/drivers/gpu/drm/i915/selftests/i915_request.c b/drivers/gpu/drm/i915/selftests/i915_request.c index 0daa8669181da..6fe22b096bdd7 100644 --- a/drivers/gpu/drm/i915/selftests/i915_request.c +++ b/drivers/gpu/drm/i915/selftests/i915_request.c @@ -1017,8 +1017,8 @@ empty_request(struct intel_engine_cs *engine, return request; err = engine->emit_bb_start(request, - batch->node.start, - batch->node.size, + i915_vma_offset(batch), + i915_vma_size(batch), I915_DISPATCH_SECURE); if (err) goto out_request; @@ -1138,14 +1138,14 @@ static struct i915_vma *recursive_batch(struct drm_i915_private *i915) if (ver >= 8) { *cmd++ = MI_BATCH_BUFFER_START | 1 << 8 | 1; - *cmd++ = lower_32_bits(vma->node.start); - *cmd++ = upper_32_bits(vma->node.start); + *cmd++ = lower_32_bits(i915_vma_offset(vma)); + *cmd++ = upper_32_bits(i915_vma_offset(vma)); } else if (ver >= 6) { *cmd++ = MI_BATCH_BUFFER_START | 1 << 8; - *cmd++ = lower_32_bits(vma->node.start); + *cmd++ = lower_32_bits(i915_vma_offset(vma)); } else { *cmd++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT; - *cmd++ = lower_32_bits(vma->node.start); + *cmd++ = lower_32_bits(i915_vma_offset(vma)); } *cmd++ = MI_BATCH_BUFFER_END; /* terminate early in case of error */ @@ -1227,8 +1227,8 @@ static int live_all_engines(void *arg) GEM_BUG_ON(err); err = engine->emit_bb_start(request[idx], - batch->node.start, - batch->node.size, + i915_vma_offset(batch), + i915_vma_size(batch), 0); GEM_BUG_ON(err); request[idx]->batch = batch; @@ -1354,8 +1354,8 @@ static int live_sequential_engines(void *arg) GEM_BUG_ON(err); err = engine->emit_bb_start(request[idx], - batch->node.start, - batch->node.size, + i915_vma_offset(batch), + i915_vma_size(batch), 0); GEM_BUG_ON(err); request[idx]->batch = batch; diff --git a/drivers/gpu/drm/i915/selftests/igt_spinner.c b/drivers/gpu/drm/i915/selftests/igt_spinner.c index 16978ac597978..618d9386d5549 100644 --- a/drivers/gpu/drm/i915/selftests/igt_spinner.c +++ b/drivers/gpu/drm/i915/selftests/igt_spinner.c @@ -116,7 +116,7 @@ static unsigned int seqno_offset(u64 fence) static u64 hws_address(const struct i915_vma *hws, const struct i915_request *rq) { - return hws->node.start + seqno_offset(rq->fence.context); + return i915_vma_offset(hws) + seqno_offset(rq->fence.context); } struct i915_request * @@ -187,8 +187,8 @@ igt_spinner_create_request(struct igt_spinner *spin, *batch++ = MI_BATCH_BUFFER_START; else *batch++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT; - *batch++ = lower_32_bits(vma->node.start); - *batch++ = upper_32_bits(vma->node.start); + *batch++ = lower_32_bits(i915_vma_offset(vma)); + *batch++ = upper_32_bits(i915_vma_offset(vma)); *batch++ = MI_BATCH_BUFFER_END; /* not reached */ @@ -203,7 +203,7 @@ igt_spinner_create_request(struct igt_spinner *spin, flags = 0; if (GRAPHICS_VER(rq->engine->i915) <= 5) flags |= I915_DISPATCH_SECURE; - err = engine->emit_bb_start(rq, vma->node.start, PAGE_SIZE, flags); + err = engine->emit_bb_start(rq, i915_vma_offset(vma), PAGE_SIZE, flags); cancel_rq: if (err) { From patchwork Tue Nov 22 18:57:35 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Andi Shyti X-Patchwork-Id: 13052662 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 921DAC433FE for ; Tue, 22 Nov 2022 18:58:07 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id ACA8F10E445; Tue, 22 Nov 2022 18:58:06 +0000 (UTC) Received: from mga09.intel.com (mga09.intel.com [134.134.136.24]) by gabe.freedesktop.org (Postfix) with ESMTPS id 57CE410E443; Tue, 22 Nov 2022 18:58:03 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1669143483; x=1700679483; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=GU43MkICcZ3sQFh036vvHQfySBw2mkGdxXVWLfPo9To=; b=mhFz+z7N1UfAp85rKJKRn+vUt42xLFSXqDRyM0584XtMOFF/KBPJKoOM kBJDa9EiheTxMs8kH/NicKfiBIWQPWAJgN7gNcLODjoH6iixKeTa5nVl2 dPfWCq9Fj5Ng0Mf9ryw2k74dhJX2jvedtL+OlXQsLu4TiFDm2DP97ATRo XYBuNTkcVGfbsz5BCoqZhFKENdyVDxZ4Y3Irt8TUzoNR3pqvToYfmO+cg CySYQfVNWAnxPI+NWhHEgHtddWviFSur0niUe1L3Hzf90MS8kZj+JhoaZ qm7mhaz2PmeaMaNv+ngn0LgaADp9uL1FM+E2q+EGRIMTKWnhl8p+lCNmn w==; X-IronPort-AV: E=McAfee;i="6500,9779,10539"; a="315042014" X-IronPort-AV: E=Sophos;i="5.96,184,1665471600"; d="scan'208";a="315042014" Received: from orsmga001.jf.intel.com ([10.7.209.18]) by orsmga102.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 22 Nov 2022 10:58:02 -0800 X-IronPort-AV: E=McAfee;i="6500,9779,10539"; a="674448249" X-IronPort-AV: E=Sophos;i="5.96,184,1665471600"; d="scan'208";a="674448249" Received: from fuhrberg-mobl1.ger.corp.intel.com (HELO intel.com) ([10.251.209.136]) by orsmga001-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 22 Nov 2022 10:57:57 -0800 From: Andi Shyti To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org Date: Tue, 22 Nov 2022 19:57:35 +0100 Message-Id: <20221122185737.96459-3-andi.shyti@linux.intel.com> X-Mailer: git-send-email 2.38.1 In-Reply-To: <20221122185737.96459-1-andi.shyti@linux.intel.com> References: <20221122185737.96459-1-andi.shyti@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH v2 2/4] drm/i915: Introduce guard pages to i915_vma X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: =?utf-8?q?Thomas_Hellstr=C3=B6m?= , Chris Wilson , Matthew Auld Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" From: Chris Wilson Introduce the concept of padding the i915_vma with guard pages before and after. The major consequence is that all ordinary uses of i915_vma must use i915_vma_offset/i915_vma_size and not i915_vma.node.start/size directly, as the drm_mm_node will include the guard pages that surround our object. The biggest connundrum is how exactly to mix requesting a fixed address with guard pages, particularly through the existing uABI. The user does not know about guard pages, so such must be transparent to the user, and so the execobj.offset must be that of the object itself excluding the guard. So a PIN_OFFSET_FIXED must then be exclusive of the guard pages. The caveat is that some placements will be impossible with guard pages, as wrap arounds need to be avoided, and the vma itself will require a larger node. We must not report EINVAL but ENOSPC as these are unavailable locations within the GTT rather than conflicting user requirements. In the next patch, we start using guard pages for scanout objects. While these are limited to GGTT vma, on a few platforms these vma (or at least an alias of the vma) is shared with userspace, so we may leak the existence of such guards if we are not careful to ensure that the execobj.offset is transparent and excludes the guards. (On such platforms like ivb, without full-ppgtt, userspace has to use relocations so the presence of more untouchable regions within its GTT such be of no further issue.) Signed-off-by: Chris Wilson Signed-off-by: Tejas Upadhyay Signed-off-by: Tvrtko Ursulin Signed-off-by: Andi Shyti --- drivers/gpu/drm/i915/gt/intel_ggtt.c | 14 ++++++++---- drivers/gpu/drm/i915/i915_gem_gtt.h | 3 ++- drivers/gpu/drm/i915/i915_vma.c | 27 ++++++++++++++++++------ drivers/gpu/drm/i915/i915_vma.h | 5 +++-- drivers/gpu/drm/i915/i915_vma_resource.c | 4 ++-- drivers/gpu/drm/i915/i915_vma_resource.h | 7 +++++- drivers/gpu/drm/i915/i915_vma_types.h | 3 ++- 7 files changed, 46 insertions(+), 17 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt.c b/drivers/gpu/drm/i915/gt/intel_ggtt.c index 8145851ad23d5..133710258eae6 100644 --- a/drivers/gpu/drm/i915/gt/intel_ggtt.c +++ b/drivers/gpu/drm/i915/gt/intel_ggtt.c @@ -287,8 +287,11 @@ static void gen8_ggtt_insert_entries(struct i915_address_space *vm, */ gte = (gen8_pte_t __iomem *)ggtt->gsm; - gte += vma_res->start / I915_GTT_PAGE_SIZE; - end = gte + vma_res->node_size / I915_GTT_PAGE_SIZE; + gte += (vma_res->start - vma_res->guard) / I915_GTT_PAGE_SIZE; + end = gte + vma_res->guard / I915_GTT_PAGE_SIZE; + while (gte < end) + gen8_set_pte(gte++, vm->scratch[0]->encode); + end += (vma_res->node_size + vma_res->guard) / I915_GTT_PAGE_SIZE; for_each_sgt_daddr(addr, iter, vma_res->bi.pages) gen8_set_pte(gte++, pte_encode | addr); @@ -338,9 +341,12 @@ static void gen6_ggtt_insert_entries(struct i915_address_space *vm, dma_addr_t addr; gte = (gen6_pte_t __iomem *)ggtt->gsm; - gte += vma_res->start / I915_GTT_PAGE_SIZE; - end = gte + vma_res->node_size / I915_GTT_PAGE_SIZE; + gte += (vma_res->start - vma_res->guard) / I915_GTT_PAGE_SIZE; + end = gte + vma_res->guard / I915_GTT_PAGE_SIZE; + while (gte < end) + iowrite32(vm->scratch[0]->encode, gte++); + end += (vma_res->node_size + vma_res->guard) / I915_GTT_PAGE_SIZE; for_each_sgt_daddr(addr, iter, vma_res->bi.pages) iowrite32(vm->pte_encode(addr, level, flags), gte++); GEM_BUG_ON(gte > end); diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.h b/drivers/gpu/drm/i915/i915_gem_gtt.h index 8c2f57eb5ddaa..2434197830523 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.h +++ b/drivers/gpu/drm/i915/i915_gem_gtt.h @@ -44,7 +44,8 @@ int i915_gem_gtt_insert(struct i915_address_space *vm, #define PIN_HIGH BIT_ULL(5) #define PIN_OFFSET_BIAS BIT_ULL(6) #define PIN_OFFSET_FIXED BIT_ULL(7) -#define PIN_VALIDATE BIT_ULL(8) /* validate placement only, no need to call unpin() */ +#define PIN_OFFSET_GUARD BIT_ULL(8) +#define PIN_VALIDATE BIT_ULL(9) /* validate placement only, no need to call unpin() */ #define PIN_GLOBAL BIT_ULL(10) /* I915_VMA_GLOBAL_BIND */ #define PIN_USER BIT_ULL(11) /* I915_VMA_LOCAL_BIND */ diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c index 2232118babeb3..457e35e03895f 100644 --- a/drivers/gpu/drm/i915/i915_vma.c +++ b/drivers/gpu/drm/i915/i915_vma.c @@ -419,7 +419,7 @@ i915_vma_resource_init_from_vma(struct i915_vma_resource *vma_res, obj->mm.rsgt, i915_gem_object_is_readonly(obj), i915_gem_object_is_lmem(obj), obj->mm.region, vma->ops, vma->private, __i915_vma_offset(vma), - __i915_vma_size(vma), vma->size); + __i915_vma_size(vma), vma->size, vma->guard); } /** @@ -749,7 +749,7 @@ static int i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, u64 size, u64 alignment, u64 flags) { - unsigned long color; + unsigned long color, guard; u64 start, end; int ret; @@ -757,7 +757,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, GEM_BUG_ON(drm_mm_node_allocated(&vma->node)); size = max(size, vma->size); - alignment = max(alignment, vma->display_alignment); + alignment = max_t(typeof(alignment), alignment, vma->display_alignment); if (flags & PIN_MAPPABLE) { size = max_t(typeof(size), size, vma->fence_size); alignment = max_t(typeof(alignment), @@ -768,6 +768,9 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, GEM_BUG_ON(!IS_ALIGNED(alignment, I915_GTT_MIN_ALIGNMENT)); GEM_BUG_ON(!is_power_of_2(alignment)); + guard = vma->guard; /* retain guard across rebinds */ + guard = ALIGN(guard, alignment); + start = flags & PIN_OFFSET_BIAS ? flags & PIN_OFFSET_MASK : 0; GEM_BUG_ON(!IS_ALIGNED(start, I915_GTT_PAGE_SIZE)); @@ -777,6 +780,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, if (flags & PIN_ZONE_4G) end = min_t(u64, end, (1ULL << 32) - I915_GTT_PAGE_SIZE); GEM_BUG_ON(!IS_ALIGNED(end, I915_GTT_PAGE_SIZE)); + GEM_BUG_ON(2 * guard > end); alignment = max(alignment, i915_vm_obj_min_alignment(vma->vm, vma->obj)); @@ -784,7 +788,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, * aperture has, reject it early before evicting everything in a vain * attempt to find space. */ - if (size > end) { + if (size > end - 2 * guard) { drm_dbg(&to_i915(vma->obj->base.dev)->drm, "Attempting to bind an object larger than the aperture: request=%llu > %s aperture=%llu\n", size, flags & PIN_MAPPABLE ? "mappable" : "total", end); @@ -801,13 +805,23 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, if (!IS_ALIGNED(offset, alignment) || range_overflows(offset, size, end)) return -EINVAL; + /* + * The caller knows not of the guard added by others and + * requests for the offset of the start of its buffer + * to be fixed, which may not be the same as the position + * of the vma->node due to the guard pages. + */ + if (offset < guard || offset + size > end - guard) + return -ENOSPC; ret = i915_gem_gtt_reserve(vma->vm, ww, &vma->node, - size, offset, color, - flags); + size + 2 * guard, + offset - guard, + color, flags); if (ret) return ret; } else { + size += 2 * guard; /* * We only support huge gtt pages through the 48b PPGTT, * however we also don't want to force any alignment for @@ -855,6 +869,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, GEM_BUG_ON(!i915_gem_valid_gtt_space(vma, color)); list_move_tail(&vma->vm_link, &vma->vm->bound_list); + vma->guard = guard; return 0; } diff --git a/drivers/gpu/drm/i915/i915_vma.h b/drivers/gpu/drm/i915/i915_vma.h index 3fd4512b1f65f..ed5c9d682a1b2 100644 --- a/drivers/gpu/drm/i915/i915_vma.h +++ b/drivers/gpu/drm/i915/i915_vma.h @@ -128,7 +128,7 @@ static inline bool i915_vma_is_closed(const struct i915_vma *vma) /* Internal use only. */ static inline u64 __i915_vma_size(const struct i915_vma *vma) { - return vma->node.size; + return vma->node.size - 2 * vma->guard; } /** @@ -150,7 +150,8 @@ static inline u64 i915_vma_size(const struct i915_vma *vma) /* Internal use only. */ static inline u64 __i915_vma_offset(const struct i915_vma *vma) { - return vma->node.start; + /* The actual start of the vma->pages is after the guard pages. */ + return vma->node.start + vma->guard; } /** diff --git a/drivers/gpu/drm/i915/i915_vma_resource.c b/drivers/gpu/drm/i915/i915_vma_resource.c index de1342dbfa128..6ba7a7feceba1 100644 --- a/drivers/gpu/drm/i915/i915_vma_resource.c +++ b/drivers/gpu/drm/i915/i915_vma_resource.c @@ -34,8 +34,8 @@ static struct kmem_cache *slab_vma_resources; * and removal of fences increases as O(ln(pending_unbinds)) instead of * O(1) for a single fence without interval tree. */ -#define VMA_RES_START(_node) ((_node)->start) -#define VMA_RES_LAST(_node) ((_node)->start + (_node)->node_size - 1) +#define VMA_RES_START(_node) ((_node)->start - (_node)->guard) +#define VMA_RES_LAST(_node) ((_node)->start + (_node)->node_size + (_node)->guard - 1) INTERVAL_TREE_DEFINE(struct i915_vma_resource, rb, u64, __subtree_last, VMA_RES_START, VMA_RES_LAST, static, vma_res_itree); diff --git a/drivers/gpu/drm/i915/i915_vma_resource.h b/drivers/gpu/drm/i915/i915_vma_resource.h index 54edf3739ca0b..c1864e3d0b43e 100644 --- a/drivers/gpu/drm/i915/i915_vma_resource.h +++ b/drivers/gpu/drm/i915/i915_vma_resource.h @@ -57,6 +57,7 @@ struct i915_page_sizes { * @node_size: Size of the allocated range manager node with padding * subtracted. * @vma_size: Bind size. + * @guard: The size of guard area preceding and trailing the bind. * @page_sizes_gtt: Resulting page sizes from the bind operation. * @bound_flags: Flags indicating binding status. * @allocated: Backend private data. TODO: Should move into @private. @@ -115,6 +116,7 @@ struct i915_vma_resource { u64 start; u64 node_size; u64 vma_size; + u32 guard; u32 page_sizes_gtt; u32 bound_flags; @@ -179,6 +181,7 @@ static inline void i915_vma_resource_put(struct i915_vma_resource *vma_res) * @start: Offset into the address space of bind range start after padding. * @node_size: Size of the allocated range manager node minus padding. * @size: Bind size. + * @guard: The size of the guard area preceding and trailing the bind. * * Initializes a vma resource allocated using i915_vma_resource_alloc(). * The reason for having separate allocate and initialize function is that @@ -197,7 +200,8 @@ static inline void i915_vma_resource_init(struct i915_vma_resource *vma_res, void *private, u64 start, u64 node_size, - u64 size) + u64 size, + u32 guard) { __i915_vma_resource_init(vma_res); vma_res->vm = vm; @@ -215,6 +219,7 @@ static inline void i915_vma_resource_init(struct i915_vma_resource *vma_res, vma_res->start = start; vma_res->node_size = node_size; vma_res->vma_size = size; + vma_res->guard = guard; } static inline void i915_vma_resource_fini(struct i915_vma_resource *vma_res) diff --git a/drivers/gpu/drm/i915/i915_vma_types.h b/drivers/gpu/drm/i915/i915_vma_types.h index ec0f6c9f57d02..77fda2244d161 100644 --- a/drivers/gpu/drm/i915/i915_vma_types.h +++ b/drivers/gpu/drm/i915/i915_vma_types.h @@ -197,14 +197,15 @@ struct i915_vma { struct i915_fence_reg *fence; u64 size; - u64 display_alignment; struct i915_page_sizes page_sizes; /* mmap-offset associated with fencing for this vma */ struct i915_mmap_offset *mmo; + u32 guard; /* padding allocated around vma->pages within the node */ u32 fence_size; u32 fence_alignment; + u32 display_alignment; /** * Count of the number of times this vma has been opened by different From patchwork Tue Nov 22 18:57:36 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Andi Shyti X-Patchwork-Id: 13052663 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 A8A75C4332F for ; Tue, 22 Nov 2022 18:58:30 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 7081810E44B; Tue, 22 Nov 2022 18:58:29 +0000 (UTC) Received: from mga09.intel.com (mga09.intel.com [134.134.136.24]) by gabe.freedesktop.org (Postfix) with ESMTPS id 3E07C10E444; Tue, 22 Nov 2022 18:58:09 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1669143489; x=1700679489; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=xfeq13Qxaa0M+bBo+aW3iHVcha2EItqV1BdFYKsG3d8=; b=atAy+v18vz5TrJdAGUMBDsznvXWSLCDcA85hUYDBJ3fEs/jIYksr/Fmg FtMkznAgH5/Ael19IK4CRYoR4FhVPfP6MB2F+jK2uvRTCaVwdT5Fcw4gl lGMCOSOaDeHIQhIRTwdKwZ7HdUZRE87sImIy25ZhaEICebQOk1ocMPOTQ opc3cSQR0d5ueTE9rpLKp+x4kjDGX+PG7nxFHWyowUYdMe84SLwZgEPnY zQKQNVu7VgvWkQnyJwuDavy53pXZjfF2zVPRgspvnaUapUKOwuASZTvrC I7l2+9Z5wrfSH5R22dQ/B6Cyhb54QIq4lJDdB6oM2MP7nlJ0gAisfGGVv A==; X-IronPort-AV: E=McAfee;i="6500,9779,10539"; a="315042024" X-IronPort-AV: E=Sophos;i="5.96,184,1665471600"; d="scan'208";a="315042024" Received: from orsmga001.jf.intel.com ([10.7.209.18]) by orsmga102.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 22 Nov 2022 10:58:08 -0800 X-IronPort-AV: E=McAfee;i="6500,9779,10539"; a="674448268" X-IronPort-AV: E=Sophos;i="5.96,184,1665471600"; d="scan'208";a="674448268" Received: from fuhrberg-mobl1.ger.corp.intel.com (HELO intel.com) ([10.251.209.136]) by orsmga001-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 22 Nov 2022 10:58:03 -0800 From: Andi Shyti To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org Date: Tue, 22 Nov 2022 19:57:36 +0100 Message-Id: <20221122185737.96459-4-andi.shyti@linux.intel.com> X-Mailer: git-send-email 2.38.1 In-Reply-To: <20221122185737.96459-1-andi.shyti@linux.intel.com> References: <20221122185737.96459-1-andi.shyti@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH v2 3/4] drm/i915: Refine VT-d scanout workaround X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: =?utf-8?q?Thomas_Hellstr=C3=B6m?= , Chris Wilson , Matthew Auld Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" From: Chris Wilson VT-d may cause overfetch of the scanout PTE, both before and after the vma (depending on the scanout orientation). bspec recommends that we provide a tile-row in either directions, and suggests using 168 PTE, warning that the accesses will wrap around the ends of the GGTT. Currently, we fill the entire GGTT with scratch pages when using VT-d to always ensure there are valid entries around every vma, including scanout. However, writing every PTE is slow as on recent devices we perform 8MiB of uncached writes, incurring an extra 100ms during resume. If instead we focus on only putting guard pages around scanout, we can avoid touching the whole GGTT. To avoid having to introduce extra nodes around each scanout vma, we adjust the scanout drm_mm_node to be smaller than the allocated space, and fixup the extra PTE during dma binding. Signed-off-by: Chris Wilson Signed-off-by: Tejas Upadhyay Signed-off-by: Tvrtko Ursulin Signed-off-by: Andi Shyti --- drivers/gpu/drm/i915/gem/i915_gem_domain.c | 13 +++++++++++ drivers/gpu/drm/i915/gt/intel_ggtt.c | 25 +--------------------- drivers/gpu/drm/i915/i915_vma.c | 9 ++++++++ 3 files changed, 23 insertions(+), 24 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_domain.c b/drivers/gpu/drm/i915/gem/i915_gem_domain.c index d44a152ce6800..882b91519f92b 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_domain.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_domain.c @@ -17,6 +17,8 @@ #include "i915_gem_object.h" #include "i915_vma.h" +#define VTD_GUARD (168u * I915_GTT_PAGE_SIZE) /* 168 or tile-row PTE padding */ + static bool gpu_write_needs_clflush(struct drm_i915_gem_object *obj) { struct drm_i915_private *i915 = to_i915(obj->base.dev); @@ -424,6 +426,17 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj, if (ret) return ERR_PTR(ret); + /* VT-d may overfetch before/after the vma, so pad with scratch */ + if (intel_scanout_needs_vtd_wa(i915)) { + unsigned int guard = VTD_GUARD; + + if (i915_gem_object_is_tiled(obj)) + guard = max(guard, + i915_gem_object_get_tile_row_size(obj)); + + flags |= PIN_OFFSET_GUARD | guard; + } + /* * As the user may map the buffer once pinned in the display plane * (e.g. libkms for the bootup splash), we have to ensure that we diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt.c b/drivers/gpu/drm/i915/gt/intel_ggtt.c index 133710258eae6..5ccec5c9206d2 100644 --- a/drivers/gpu/drm/i915/gt/intel_ggtt.c +++ b/drivers/gpu/drm/i915/gt/intel_ggtt.c @@ -367,27 +367,6 @@ static void nop_clear_range(struct i915_address_space *vm, { } -static void gen8_ggtt_clear_range(struct i915_address_space *vm, - u64 start, u64 length) -{ - struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm); - unsigned int first_entry = start / I915_GTT_PAGE_SIZE; - unsigned int num_entries = length / I915_GTT_PAGE_SIZE; - const gen8_pte_t scratch_pte = vm->scratch[0]->encode; - gen8_pte_t __iomem *gtt_base = - (gen8_pte_t __iomem *)ggtt->gsm + first_entry; - const int max_entries = ggtt_total_entries(ggtt) - first_entry; - int i; - - if (WARN(num_entries > max_entries, - "First entry = %d; Num entries = %d (max=%d)\n", - first_entry, num_entries, max_entries)) - num_entries = max_entries; - - for (i = 0; i < num_entries; i++) - gen8_set_pte(>t_base[i], scratch_pte); -} - static void bxt_vtd_ggtt_wa(struct i915_address_space *vm) { /* @@ -959,8 +938,6 @@ static int gen8_gmch_probe(struct i915_ggtt *ggtt) ggtt->vm.cleanup = gen6_gmch_remove; ggtt->vm.insert_page = gen8_ggtt_insert_page; ggtt->vm.clear_range = nop_clear_range; - if (intel_scanout_needs_vtd_wa(i915)) - ggtt->vm.clear_range = gen8_ggtt_clear_range; ggtt->vm.insert_entries = gen8_ggtt_insert_entries; @@ -1121,7 +1098,7 @@ static int gen6_gmch_probe(struct i915_ggtt *ggtt) ggtt->vm.alloc_scratch_dma = alloc_pt_dma; ggtt->vm.clear_range = nop_clear_range; - if (!HAS_FULL_PPGTT(i915) || intel_scanout_needs_vtd_wa(i915)) + if (!HAS_FULL_PPGTT(i915)) ggtt->vm.clear_range = gen6_ggtt_clear_range; ggtt->vm.insert_page = gen6_ggtt_insert_page; ggtt->vm.insert_entries = gen6_ggtt_insert_entries; diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c index 457e35e03895f..840c7daf8bb70 100644 --- a/drivers/gpu/drm/i915/i915_vma.c +++ b/drivers/gpu/drm/i915/i915_vma.c @@ -677,6 +677,10 @@ bool i915_vma_misplaced(const struct i915_vma *vma, i915_vma_offset(vma) != (flags & PIN_OFFSET_MASK)) return true; + if (flags & PIN_OFFSET_GUARD && + vma->guard < (flags & PIN_OFFSET_MASK)) + return true; + return false; } @@ -755,6 +759,7 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, GEM_BUG_ON(i915_vma_is_bound(vma, I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND)); GEM_BUG_ON(drm_mm_node_allocated(&vma->node)); + GEM_BUG_ON(hweight64(flags & (PIN_OFFSET_GUARD | PIN_OFFSET_FIXED | PIN_OFFSET_BIAS)) > 1); size = max(size, vma->size); alignment = max_t(typeof(alignment), alignment, vma->display_alignment); @@ -769,6 +774,10 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, GEM_BUG_ON(!is_power_of_2(alignment)); guard = vma->guard; /* retain guard across rebinds */ + if (flags & PIN_OFFSET_GUARD) { + GEM_BUG_ON(overflows_type(flags & PIN_OFFSET_MASK, u32)); + guard = max_t(u32, guard, flags & PIN_OFFSET_MASK); + } guard = ALIGN(guard, alignment); start = flags & PIN_OFFSET_BIAS ? flags & PIN_OFFSET_MASK : 0; From patchwork Tue Nov 22 18:57:37 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Andi Shyti X-Patchwork-Id: 13052664 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 D50EBC433FE for ; Tue, 22 Nov 2022 18:58:32 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id A926510E44C; Tue, 22 Nov 2022 18:58:31 +0000 (UTC) Received: from mga09.intel.com (mga09.intel.com [134.134.136.24]) by gabe.freedesktop.org (Postfix) with ESMTPS id 6DFA410E44B; Tue, 22 Nov 2022 18:58:13 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1669143493; x=1700679493; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=wAUO3vPuGeEk0LHtAL6dVHzHtqKg3Fou2q2AWzcmr4s=; b=Gxc3FF1He2Jl7ILHjK8QB68rRY3RN4m/n0Lwv2uqtTDBqv6pp5n5b6jK bnK8V13GELtaebFnArmMeEbVJRJpAaUoYqOnlMmR0TVo4px8IBn+5ta08 ch/F4TDdoJqdSKhRLSPpiXs9SptoFn8uuBYRfSBBxvDWgnLflYXIlHZF/ ABkwjH7KWvPhCBjru5Vz/wRVoaPJmBM81z3gRfNgFK20iTsTmneY771R+ qHsTKmBZokuj9nfR+5kzFq9BaSMWFWOQm/+hPRU8y6d/ZuvvtQsqxUJE0 CzNCDkBxzmxLz4iD4ST3AtZGorfKY61F8uyG3X5FOjUHl8brwxoTp6661 w==; X-IronPort-AV: E=McAfee;i="6500,9779,10539"; a="315042054" X-IronPort-AV: E=Sophos;i="5.96,184,1665471600"; d="scan'208";a="315042054" Received: from orsmga001.jf.intel.com ([10.7.209.18]) by orsmga102.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 22 Nov 2022 10:58:13 -0800 X-IronPort-AV: E=McAfee;i="6500,9779,10539"; a="674448294" X-IronPort-AV: E=Sophos;i="5.96,184,1665471600"; d="scan'208";a="674448294" Received: from fuhrberg-mobl1.ger.corp.intel.com (HELO intel.com) ([10.251.209.136]) by orsmga001-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 22 Nov 2022 10:58:09 -0800 From: Andi Shyti To: intel-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org Date: Tue, 22 Nov 2022 19:57:37 +0100 Message-Id: <20221122185737.96459-5-andi.shyti@linux.intel.com> X-Mailer: git-send-email 2.38.1 In-Reply-To: <20221122185737.96459-1-andi.shyti@linux.intel.com> References: <20221122185737.96459-1-andi.shyti@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH v2 4/4] Revert "drm/i915: Improve on suspend / resume time with VT-d enabled" X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: =?utf-8?q?Thomas_Hellstr=C3=B6m?= , Chris Wilson , Matthew Auld Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" This reverts commit 2ef6efa79fecd5e3457b324155d35524d95f2b6b. Checking the presence if the IRST (Intel Rapid Start Technology) through the ACPI to decide whether to rebuild or not the GGTT puts us at the mercy of the boot firmware and we need to unnecessarily rely on third parties. Because now we avoid adding scratch pages to the entire GGTT we don't need this hack anymore. Signed-off-by: Andi Shyti Cc: Thomas Hellström Cc: Chris Wilson Cc: Tvrtko Ursulin --- drivers/gpu/drm/i915/gt/intel_ggtt.c | 69 ++++++---------------------- drivers/gpu/drm/i915/gt/intel_gtt.h | 24 ---------- drivers/gpu/drm/i915/i915_driver.c | 16 ------- 3 files changed, 13 insertions(+), 96 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt.c b/drivers/gpu/drm/i915/gt/intel_ggtt.c index 5ccec5c9206d2..9d76a573255f6 100644 --- a/drivers/gpu/drm/i915/gt/intel_ggtt.c +++ b/drivers/gpu/drm/i915/gt/intel_ggtt.c @@ -26,13 +26,6 @@ #include "intel_gtt.h" #include "gen8_ppgtt.h" -static inline bool suspend_retains_ptes(struct i915_address_space *vm) -{ - return GRAPHICS_VER(vm->i915) >= 8 && - !HAS_LMEM(vm->i915) && - vm->is_ggtt; -} - static void i915_ggtt_color_adjust(const struct drm_mm_node *node, unsigned long color, u64 *start, @@ -104,23 +97,6 @@ int i915_ggtt_init_hw(struct drm_i915_private *i915) return 0; } -/* - * Return the value of the last GGTT pte cast to an u64, if - * the system is supposed to retain ptes across resume. 0 otherwise. - */ -static u64 read_last_pte(struct i915_address_space *vm) -{ - struct i915_ggtt *ggtt = i915_vm_to_ggtt(vm); - gen8_pte_t __iomem *ptep; - - if (!suspend_retains_ptes(vm)) - return 0; - - GEM_BUG_ON(GRAPHICS_VER(vm->i915) < 8); - ptep = (typeof(ptep))ggtt->gsm + (ggtt_total_entries(ggtt) - 1); - return readq(ptep); -} - /** * i915_ggtt_suspend_vm - Suspend the memory mappings for a GGTT or DPT VM * @vm: The VM to suspend the mappings for @@ -184,10 +160,7 @@ void i915_ggtt_suspend_vm(struct i915_address_space *vm) i915_gem_object_unlock(obj); } - if (!suspend_retains_ptes(vm)) - vm->clear_range(vm, 0, vm->total); - else - i915_vm_to_ggtt(vm)->probed_pte = read_last_pte(vm); + vm->clear_range(vm, 0, vm->total); vm->skip_pte_rewrite = save_skip_rewrite; @@ -536,8 +509,6 @@ static int init_ggtt(struct i915_ggtt *ggtt) struct drm_mm_node *entry; int ret; - ggtt->pte_lost = true; - /* * GuC requires all resources that we're sharing with it to be placed in * non-WOPCM memory. If GuC is not present or not in use we still need a @@ -1236,20 +1207,11 @@ bool i915_ggtt_resume_vm(struct i915_address_space *vm) { struct i915_vma *vma; bool write_domain_objs = false; - bool retained_ptes; drm_WARN_ON(&vm->i915->drm, !vm->is_ggtt && !vm->is_dpt); - /* - * First fill our portion of the GTT with scratch pages if - * they were not retained across suspend. - */ - retained_ptes = suspend_retains_ptes(vm) && - !i915_vm_to_ggtt(vm)->pte_lost && - !GEM_WARN_ON(i915_vm_to_ggtt(vm)->probed_pte != read_last_pte(vm)); - - if (!retained_ptes) - vm->clear_range(vm, 0, vm->total); + /* First fill our portion of the GTT with scratch pages */ + vm->clear_range(vm, 0, vm->total); /* clflush objects bound into the GGTT and rebind them. */ list_for_each_entry(vma, &vm->bound_list, vm_link) { @@ -1258,16 +1220,16 @@ bool i915_ggtt_resume_vm(struct i915_address_space *vm) atomic_read(&vma->flags) & I915_VMA_BIND_MASK; GEM_BUG_ON(!was_bound); - if (!retained_ptes) { - /* - * Clear the bound flags of the vma resource to allow - * ptes to be repopulated. - */ - vma->resource->bound_flags = 0; - vma->ops->bind_vma(vm, NULL, vma->resource, - obj ? obj->cache_level : 0, - was_bound); - } + + /* + * Clear the bound flags of the vma resource to allow + * ptes to be repopulated. + */ + vma->resource->bound_flags = 0; + vma->ops->bind_vma(vm, NULL, vma->resource, + obj ? obj->cache_level : 0, + was_bound); + if (obj) { /* only used during resume => exclusive access */ write_domain_objs |= fetch_and_zero(&obj->write_domain); obj->read_domains |= I915_GEM_DOMAIN_GTT; @@ -1295,8 +1257,3 @@ void i915_ggtt_resume(struct i915_ggtt *ggtt) intel_ggtt_restore_fences(ggtt); } - -void i915_ggtt_mark_pte_lost(struct drm_i915_private *i915, bool val) -{ - to_gt(i915)->ggtt->pte_lost = val; -} diff --git a/drivers/gpu/drm/i915/gt/intel_gtt.h b/drivers/gpu/drm/i915/gt/intel_gtt.h index 4d75ba4bb41da..eae880dced4d7 100644 --- a/drivers/gpu/drm/i915/gt/intel_gtt.h +++ b/drivers/gpu/drm/i915/gt/intel_gtt.h @@ -355,19 +355,6 @@ struct i915_ggtt { bool do_idle_maps; - /** - * @pte_lost: Are ptes lost on resume? - * - * Whether the system was recently restored from hibernate and - * thus may have lost pte content. - */ - bool pte_lost; - - /** - * @probed_pte: Probed pte value on suspend. Re-checked on resume. - */ - u64 probed_pte; - int mtrr; /** Bit 6 swizzling required for X tiling */ @@ -600,17 +587,6 @@ bool i915_ggtt_resume_vm(struct i915_address_space *vm); void i915_ggtt_suspend(struct i915_ggtt *gtt); void i915_ggtt_resume(struct i915_ggtt *ggtt); -/** - * i915_ggtt_mark_pte_lost - Mark ggtt ptes as lost or clear such a marking - * @i915 The device private. - * @val whether the ptes should be marked as lost. - * - * In some cases pte content is retained across suspend, but typically lost - * across hibernate. Typically they should be marked as lost on - * hibernation restore and such marking cleared on suspend. - */ -void i915_ggtt_mark_pte_lost(struct drm_i915_private *i915, bool val); - void fill_page_dma(struct drm_i915_gem_object *p, const u64 val, unsigned int count); diff --git a/drivers/gpu/drm/i915/i915_driver.c b/drivers/gpu/drm/i915/i915_driver.c index 69103ae377792..94ea2e7f1ad2e 100644 --- a/drivers/gpu/drm/i915/i915_driver.c +++ b/drivers/gpu/drm/i915/i915_driver.c @@ -102,9 +102,6 @@ #include "intel_region_ttm.h" #include "vlv_suspend.h" -/* Intel Rapid Start Technology ACPI device name */ -static const char irst_name[] = "INT3392"; - static const struct drm_driver i915_drm_driver; static void i915_release_bridge_dev(struct drm_device *dev, @@ -1494,8 +1491,6 @@ static int i915_pm_suspend(struct device *kdev) return -ENODEV; } - i915_ggtt_mark_pte_lost(i915, false); - if (i915->drm.switch_power_state == DRM_SWITCH_POWER_OFF) return 0; @@ -1548,14 +1543,6 @@ static int i915_pm_resume(struct device *kdev) if (i915->drm.switch_power_state == DRM_SWITCH_POWER_OFF) return 0; - /* - * If IRST is enabled, or if we can't detect whether it's enabled, - * then we must assume we lost the GGTT page table entries, since - * they are not retained if IRST decided to enter S4. - */ - if (!IS_ENABLED(CONFIG_ACPI) || acpi_dev_present(irst_name, NULL, -1)) - i915_ggtt_mark_pte_lost(i915, true); - return i915_drm_resume(&i915->drm); } @@ -1615,9 +1602,6 @@ static int i915_pm_restore_early(struct device *kdev) static int i915_pm_restore(struct device *kdev) { - struct drm_i915_private *i915 = kdev_to_i915(kdev); - - i915_ggtt_mark_pte_lost(i915, true); return i915_pm_resume(kdev); }