From patchwork Mon Aug 10 10:30:40 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707275 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 1D0EE1392 for ; Mon, 10 Aug 2020 10:33:45 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 0504E2070B for ; Mon, 10 Aug 2020 10:33:45 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 0504E2070B Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 247BA6E3F2; Mon, 10 Aug 2020 10:33:31 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 899FB89C3B for ; Mon, 10 Aug 2020 10:33:26 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:40 +0200 Message-Id: <20200810103103.303818-2-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 01/24] Revert "drm/i915/gem: Async GPU relocations only" 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" This reverts commit 9e0f9464e2ab36b864359a59b0e9058fdef0ce47, and related commit 7ac2d2536dfa7 ("drm/i915/gem: Delete unused code"). Async GPU relocations are not the path forward, we want to remove GPU accelerated relocation support eventually when userspace is fixed to use VM_BIND, and this is the first step towards that. We will keep async gpu relocations around for now, until userspace is fixed. Relocation support will be disabled completely on platforms where there was never any userspace that depends on it, as the hardware doesn't require it from at least gen9+ onward. For older platforms, the plan is to use cpu relocations only. Signed-off-by: Maarten Lankhorst Reviewed-by: Daniel Vetter --- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 319 ++++++++++++++++-- .../i915/gem/selftests/i915_gem_execbuffer.c | 21 +- 2 files changed, 313 insertions(+), 27 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 24a1486d2dc5..c6a613d92a13 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -46,6 +46,13 @@ struct eb_vma_array { struct eb_vma vma[]; }; +enum { + FORCE_CPU_RELOC = 1, + FORCE_GTT_RELOC, + FORCE_GPU_RELOC, +#define DBG_FORCE_RELOC 0 /* choose one of the above! */ +}; + #define __EXEC_OBJECT_HAS_PIN BIT(31) #define __EXEC_OBJECT_HAS_FENCE BIT(30) #define __EXEC_OBJECT_NEEDS_MAP BIT(29) @@ -261,6 +268,8 @@ struct i915_execbuffer { */ struct reloc_cache { struct drm_mm_node node; /** temporary GTT binding */ + unsigned long vaddr; /** Current kmap address */ + unsigned long page; /** Currently mapped page index */ unsigned int gen; /** Cached value of INTEL_GEN */ bool use_64bit_reloc : 1; bool has_llc : 1; @@ -607,6 +616,23 @@ eb_add_vma(struct i915_execbuffer *eb, } } +static inline int use_cpu_reloc(const struct reloc_cache *cache, + const struct drm_i915_gem_object *obj) +{ + if (!i915_gem_object_has_struct_page(obj)) + return false; + + if (DBG_FORCE_RELOC == FORCE_CPU_RELOC) + return true; + + if (DBG_FORCE_RELOC == FORCE_GTT_RELOC) + return false; + + return (cache->has_llc || + obj->cache_dirty || + obj->cache_level != I915_CACHE_NONE); +} + static int eb_reserve_vma(const struct i915_execbuffer *eb, struct eb_vma *ev, u64 pin_flags) @@ -937,6 +963,8 @@ relocation_target(const struct drm_i915_gem_relocation_entry *reloc, static void reloc_cache_init(struct reloc_cache *cache, struct drm_i915_private *i915) { + cache->page = -1; + cache->vaddr = 0; /* Must be a variable in the struct to allow GCC to unroll. */ cache->gen = INTEL_GEN(i915); cache->has_llc = HAS_LLC(i915); @@ -948,6 +976,25 @@ static void reloc_cache_init(struct reloc_cache *cache, cache->target = NULL; } +static inline void *unmask_page(unsigned long p) +{ + return (void *)(uintptr_t)(p & PAGE_MASK); +} + +static inline unsigned int unmask_flags(unsigned long p) +{ + return p & ~PAGE_MASK; +} + +#define KMAP 0x4 /* after CLFLUSH_FLAGS */ + +static inline struct i915_ggtt *cache_to_ggtt(struct reloc_cache *cache) +{ + struct drm_i915_private *i915 = + container_of(cache, struct i915_execbuffer, reloc_cache)->i915; + return &i915->ggtt; +} + #define RELOC_TAIL 4 static int reloc_gpu_chain(struct reloc_cache *cache) @@ -1060,6 +1107,186 @@ static int reloc_gpu_flush(struct reloc_cache *cache) return err; } +static void reloc_cache_reset(struct reloc_cache *cache) +{ + void *vaddr; + + if (!cache->vaddr) + return; + + vaddr = unmask_page(cache->vaddr); + if (cache->vaddr & KMAP) { + if (cache->vaddr & CLFLUSH_AFTER) + mb(); + + kunmap_atomic(vaddr); + i915_gem_object_finish_access((struct drm_i915_gem_object *)cache->node.mm); + } else { + struct i915_ggtt *ggtt = cache_to_ggtt(cache); + + intel_gt_flush_ggtt_writes(ggtt->vm.gt); + io_mapping_unmap_atomic((void __iomem *)vaddr); + + if (drm_mm_node_allocated(&cache->node)) { + ggtt->vm.clear_range(&ggtt->vm, + cache->node.start, + cache->node.size); + mutex_lock(&ggtt->vm.mutex); + drm_mm_remove_node(&cache->node); + mutex_unlock(&ggtt->vm.mutex); + } else { + i915_vma_unpin((struct i915_vma *)cache->node.mm); + } + } + + cache->vaddr = 0; + cache->page = -1; +} + +static void *reloc_kmap(struct drm_i915_gem_object *obj, + struct reloc_cache *cache, + unsigned long pageno) +{ + void *vaddr; + struct page *page; + + if (cache->vaddr) { + kunmap_atomic(unmask_page(cache->vaddr)); + } else { + unsigned int flushes; + int err; + + err = i915_gem_object_prepare_write(obj, &flushes); + if (err) + return ERR_PTR(err); + + BUILD_BUG_ON(KMAP & CLFLUSH_FLAGS); + BUILD_BUG_ON((KMAP | CLFLUSH_FLAGS) & PAGE_MASK); + + cache->vaddr = flushes | KMAP; + cache->node.mm = (void *)obj; + if (flushes) + mb(); + } + + page = i915_gem_object_get_page(obj, pageno); + if (!obj->mm.dirty) + set_page_dirty(page); + + vaddr = kmap_atomic(page); + cache->vaddr = unmask_flags(cache->vaddr) | (unsigned long)vaddr; + cache->page = pageno; + + return vaddr; +} + +static void *reloc_iomap(struct drm_i915_gem_object *obj, + struct reloc_cache *cache, + unsigned long page) +{ + struct i915_ggtt *ggtt = cache_to_ggtt(cache); + unsigned long offset; + void *vaddr; + + if (cache->vaddr) { + intel_gt_flush_ggtt_writes(ggtt->vm.gt); + io_mapping_unmap_atomic((void __force __iomem *) unmask_page(cache->vaddr)); + } else { + struct i915_vma *vma; + int err; + + if (i915_gem_object_is_tiled(obj)) + return ERR_PTR(-EINVAL); + + if (use_cpu_reloc(cache, obj)) + return NULL; + + i915_gem_object_lock(obj); + err = i915_gem_object_set_to_gtt_domain(obj, true); + i915_gem_object_unlock(obj); + if (err) + return ERR_PTR(err); + + vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, + PIN_MAPPABLE | + PIN_NONBLOCK /* NOWARN */ | + PIN_NOEVICT); + if (IS_ERR(vma)) { + memset(&cache->node, 0, sizeof(cache->node)); + mutex_lock(&ggtt->vm.mutex); + err = drm_mm_insert_node_in_range + (&ggtt->vm.mm, &cache->node, + PAGE_SIZE, 0, I915_COLOR_UNEVICTABLE, + 0, ggtt->mappable_end, + DRM_MM_INSERT_LOW); + mutex_unlock(&ggtt->vm.mutex); + if (err) /* no inactive aperture space, use cpu reloc */ + return NULL; + } else { + cache->node.start = vma->node.start; + cache->node.mm = (void *)vma; + } + } + + offset = cache->node.start; + if (drm_mm_node_allocated(&cache->node)) { + ggtt->vm.insert_page(&ggtt->vm, + i915_gem_object_get_dma_address(obj, page), + offset, I915_CACHE_NONE, 0); + } else { + offset += page << PAGE_SHIFT; + } + + vaddr = (void __force *)io_mapping_map_atomic_wc(&ggtt->iomap, + offset); + cache->page = page; + cache->vaddr = (unsigned long)vaddr; + + return vaddr; +} + +static void *reloc_vaddr(struct drm_i915_gem_object *obj, + struct reloc_cache *cache, + unsigned long page) +{ + void *vaddr; + + if (cache->page == page) { + vaddr = unmask_page(cache->vaddr); + } else { + vaddr = NULL; + if ((cache->vaddr & KMAP) == 0) + vaddr = reloc_iomap(obj, cache, page); + if (!vaddr) + vaddr = reloc_kmap(obj, cache, page); + } + + return vaddr; +} + +static void clflush_write32(u32 *addr, u32 value, unsigned int flushes) +{ + if (unlikely(flushes & (CLFLUSH_BEFORE | CLFLUSH_AFTER))) { + if (flushes & CLFLUSH_BEFORE) { + clflushopt(addr); + mb(); + } + + *addr = value; + + /* + * Writes to the same cacheline are serialised by the CPU + * (including clflush). On the write path, we only require + * that it hits memory in an orderly fashion and place + * mb barriers at the start and end of the relocation phase + * to ensure ordering of clflush wrt to the system. + */ + if (flushes & CLFLUSH_AFTER) + clflushopt(addr); + } else + *addr = value; +} + static int reloc_move_to_gpu(struct i915_request *rq, struct i915_vma *vma) { struct drm_i915_gem_object *obj = vma->obj; @@ -1225,6 +1452,17 @@ static u32 *reloc_gpu(struct i915_execbuffer *eb, return cmd; } +static inline bool use_reloc_gpu(struct i915_vma *vma) +{ + if (DBG_FORCE_RELOC == FORCE_GPU_RELOC) + return true; + + if (DBG_FORCE_RELOC) + return false; + + return !dma_resv_test_signaled_rcu(vma->resv, true); +} + static unsigned long vma_phys_addr(struct i915_vma *vma, u32 offset) { struct page *page; @@ -1239,10 +1477,10 @@ static unsigned long vma_phys_addr(struct i915_vma *vma, u32 offset) return addr + offset_in_page(offset); } -static int __reloc_entry_gpu(struct i915_execbuffer *eb, - struct i915_vma *vma, - u64 offset, - u64 target_addr) +static bool __reloc_entry_gpu(struct i915_execbuffer *eb, + struct i915_vma *vma, + u64 offset, + u64 target_addr) { const unsigned int gen = eb->reloc_cache.gen; unsigned int len; @@ -1258,7 +1496,7 @@ static int __reloc_entry_gpu(struct i915_execbuffer *eb, batch = reloc_gpu(eb, vma, len); if (IS_ERR(batch)) - return PTR_ERR(batch); + return false; addr = gen8_canonical_addr(vma->node.start + offset); if (gen >= 8) { @@ -1307,21 +1545,55 @@ static int __reloc_entry_gpu(struct i915_execbuffer *eb, *batch++ = target_addr; } - return 0; + return true; +} + +static bool reloc_entry_gpu(struct i915_execbuffer *eb, + struct i915_vma *vma, + u64 offset, + u64 target_addr) +{ + if (eb->reloc_cache.vaddr) + return false; + + if (!use_reloc_gpu(vma)) + return false; + + return __reloc_entry_gpu(eb, vma, offset, target_addr); } static u64 -relocate_entry(struct i915_execbuffer *eb, - struct i915_vma *vma, +relocate_entry(struct i915_vma *vma, const struct drm_i915_gem_relocation_entry *reloc, + struct i915_execbuffer *eb, const struct i915_vma *target) { u64 target_addr = relocation_target(reloc, target); - int err; - - err = __reloc_entry_gpu(eb, vma, reloc->offset, target_addr); - if (err) - return err; + u64 offset = reloc->offset; + + if (!reloc_entry_gpu(eb, vma, offset, target_addr)) { + bool wide = eb->reloc_cache.use_64bit_reloc; + void *vaddr; + +repeat: + vaddr = reloc_vaddr(vma->obj, + &eb->reloc_cache, + offset >> PAGE_SHIFT); + if (IS_ERR(vaddr)) + return PTR_ERR(vaddr); + + GEM_BUG_ON(!IS_ALIGNED(offset, sizeof(u32))); + clflush_write32(vaddr + offset_in_page(offset), + lower_32_bits(target_addr), + eb->reloc_cache.vaddr); + + if (wide) { + offset += sizeof(u32); + target_addr >>= 32; + wide = false; + goto repeat; + } + } return target->node.start | UPDATE; } @@ -1386,7 +1658,8 @@ eb_relocate_entry(struct i915_execbuffer *eb, * If the relocation already has the right value in it, no * more work needs to be done. */ - if (gen8_canonical_addr(target->vma->node.start) == reloc->presumed_offset) + if (!DBG_FORCE_RELOC && + gen8_canonical_addr(target->vma->node.start) == reloc->presumed_offset) return 0; /* Check that the relocation address is valid... */ @@ -1418,7 +1691,7 @@ eb_relocate_entry(struct i915_execbuffer *eb, ev->flags &= ~EXEC_OBJECT_ASYNC; /* and update the user's relocation entry */ - return relocate_entry(eb, ev->vma, reloc, target->vma); + return relocate_entry(ev->vma, reloc, eb, target->vma); } static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) @@ -1456,8 +1729,10 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) * this is bad and so lockdep complains vehemently. */ copied = __copy_from_user(r, urelocs, count * sizeof(r[0])); - if (unlikely(copied)) - return -EFAULT; + if (unlikely(copied)) { + remain = -EFAULT; + goto out; + } remain -= count; do { @@ -1465,7 +1740,8 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) if (likely(offset == 0)) { } else if ((s64)offset < 0) { - return (int)offset; + remain = (int)offset; + goto out; } else { /* * Note that reporting an error now @@ -1495,8 +1771,9 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) } while (r++, --count); urelocs += ARRAY_SIZE(stack); } while (remain); - - return 0; +out: + reloc_cache_reset(&eb->reloc_cache); + return remain; } static int eb_relocate(struct i915_execbuffer *eb) @@ -2573,7 +2850,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, eb.i915 = i915; eb.file = file; eb.args = args; - if (!(args->flags & I915_EXEC_NO_RELOC)) + if (DBG_FORCE_RELOC || !(args->flags & I915_EXEC_NO_RELOC)) args->flags |= __EXEC_HAS_RELOC; eb.exec = exec; diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c index 57c14d3340cd..a49016f8ee0d 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c @@ -37,14 +37,20 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, return err; /* 8-Byte aligned */ - err = __reloc_entry_gpu(eb, vma, offsets[0] * sizeof(u32), 0); - if (err) + if (!__reloc_entry_gpu(eb, vma, + offsets[0] * sizeof(u32), + 0)) { + err = -EIO; goto unpin_vma; + } /* !8-Byte aligned */ - err = __reloc_entry_gpu(eb, vma, offsets[1] * sizeof(u32), 1); - if (err) + if (!__reloc_entry_gpu(eb, vma, + offsets[1] * sizeof(u32), + 1)) { + err = -EIO; goto unpin_vma; + } /* Skip to the end of the cmd page */ i = PAGE_SIZE / sizeof(u32) - RELOC_TAIL - 1; @@ -54,9 +60,12 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, eb->reloc_cache.rq_size += i; /* Force batch chaining */ - err = __reloc_entry_gpu(eb, vma, offsets[2] * sizeof(u32), 2); - if (err) + if (!__reloc_entry_gpu(eb, vma, + offsets[2] * sizeof(u32), + 2)) { + err = -EIO; goto unpin_vma; + } GEM_BUG_ON(!eb->reloc_cache.rq); rq = i915_request_get(eb->reloc_cache.rq); From patchwork Mon Aug 10 10:30:41 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707257 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 96CE41392 for ; Mon, 10 Aug 2020 10:33:35 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 7F0EA206E9 for ; Mon, 10 Aug 2020 10:33:35 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 7F0EA206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 6BABB6E3EB; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 8288D897E0 for ; Mon, 10 Aug 2020 10:33:26 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:41 +0200 Message-Id: <20200810103103.303818-3-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 02/24] drm/i915: Revert relocation chaining commits. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" This reverts commit 964a9b0f611ee ("drm/i915/gem: Use chained reloc batches") and commit 0e97fbb080553 ("drm/i915/gem: Use a single chained reloc batches for a single execbuf"). When adding ww locking to execbuf, it's hard enough to deal with a single BO that is part of relocation execution. Chaining is hard to get right, and with GPU relocation deprecated, it's best to drop this altogether, instead of trying to fix something we will remove. Signed-off-by: Maarten Lankhorst Reviewed-by: Daniel Vetter --- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 171 ++++-------------- .../i915/gem/selftests/i915_gem_execbuffer.c | 8 +- 2 files changed, 35 insertions(+), 144 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index c6a613d92a13..6acbd08f82f0 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -276,9 +276,7 @@ struct i915_execbuffer { bool has_fence : 1; bool needs_unfenced : 1; - struct i915_vma *target; struct i915_request *rq; - struct i915_vma *rq_vma; u32 *rq_cmd; unsigned int rq_size; } reloc_cache; @@ -973,7 +971,7 @@ static void reloc_cache_init(struct reloc_cache *cache, cache->needs_unfenced = INTEL_INFO(i915)->unfenced_needs_alignment; cache->node.flags = 0; cache->rq = NULL; - cache->target = NULL; + cache->rq_size = 0; } static inline void *unmask_page(unsigned long p) @@ -995,122 +993,29 @@ static inline struct i915_ggtt *cache_to_ggtt(struct reloc_cache *cache) return &i915->ggtt; } -#define RELOC_TAIL 4 - -static int reloc_gpu_chain(struct reloc_cache *cache) +static void reloc_gpu_flush(struct reloc_cache *cache) { - struct intel_gt_buffer_pool_node *pool; - struct i915_request *rq = cache->rq; - struct i915_vma *batch; - u32 *cmd; - int err; - - pool = intel_gt_get_buffer_pool(rq->engine->gt, PAGE_SIZE); - if (IS_ERR(pool)) - return PTR_ERR(pool); - - batch = i915_vma_instance(pool->obj, rq->context->vm, NULL); - if (IS_ERR(batch)) { - err = PTR_ERR(batch); - goto out_pool; - } - - err = i915_vma_pin(batch, 0, 0, PIN_USER | PIN_NONBLOCK); - if (err) - goto out_pool; - - GEM_BUG_ON(cache->rq_size + RELOC_TAIL > PAGE_SIZE / sizeof(u32)); - cmd = cache->rq_cmd + cache->rq_size; - *cmd++ = MI_ARB_CHECK; - if (cache->gen >= 8) - *cmd++ = MI_BATCH_BUFFER_START_GEN8; - else if (cache->gen >= 6) - *cmd++ = MI_BATCH_BUFFER_START; - else - *cmd++ = MI_BATCH_BUFFER_START | MI_BATCH_GTT; - *cmd++ = lower_32_bits(batch->node.start); - *cmd++ = upper_32_bits(batch->node.start); /* Always 0 for gen<8 */ - i915_gem_object_flush_map(cache->rq_vma->obj); - i915_gem_object_unpin_map(cache->rq_vma->obj); - cache->rq_vma = NULL; - - err = intel_gt_buffer_pool_mark_active(pool, rq); - if (err == 0) { - i915_vma_lock(batch); - err = i915_request_await_object(rq, batch->obj, false); - if (err == 0) - err = i915_vma_move_to_active(batch, rq, 0); - i915_vma_unlock(batch); - } - i915_vma_unpin(batch); - if (err) - goto out_pool; + struct drm_i915_gem_object *obj = cache->rq->batch->obj; - cmd = i915_gem_object_pin_map(batch->obj, - cache->has_llc ? - I915_MAP_FORCE_WB : - I915_MAP_FORCE_WC); - if (IS_ERR(cmd)) { - err = PTR_ERR(cmd); - goto out_pool; - } + GEM_BUG_ON(cache->rq_size >= obj->base.size / sizeof(u32)); + cache->rq_cmd[cache->rq_size] = MI_BATCH_BUFFER_END; - /* Return with batch mapping (cmd) still pinned */ - cache->rq_cmd = cmd; - cache->rq_size = 0; - cache->rq_vma = batch; + __i915_gem_object_flush_map(obj, 0, sizeof(u32) * (cache->rq_size + 1)); + i915_gem_object_unpin_map(obj); -out_pool: - intel_gt_buffer_pool_put(pool); - return err; -} + intel_gt_chipset_flush(cache->rq->engine->gt); -static unsigned int reloc_bb_flags(const struct reloc_cache *cache) -{ - return cache->gen > 5 ? 0 : I915_DISPATCH_SECURE; -} - -static int reloc_gpu_flush(struct reloc_cache *cache) -{ - struct i915_request *rq; - int err; - - rq = fetch_and_zero(&cache->rq); - if (!rq) - return 0; - - if (cache->rq_vma) { - struct drm_i915_gem_object *obj = cache->rq_vma->obj; - - GEM_BUG_ON(cache->rq_size >= obj->base.size / sizeof(u32)); - cache->rq_cmd[cache->rq_size++] = MI_BATCH_BUFFER_END; - - __i915_gem_object_flush_map(obj, - 0, sizeof(u32) * cache->rq_size); - i915_gem_object_unpin_map(obj); - } - - err = 0; - if (rq->engine->emit_init_breadcrumb) - err = rq->engine->emit_init_breadcrumb(rq); - if (!err) - err = rq->engine->emit_bb_start(rq, - rq->batch->node.start, - PAGE_SIZE, - reloc_bb_flags(cache)); - if (err) - i915_request_set_error_once(rq, err); - - intel_gt_chipset_flush(rq->engine->gt); - i915_request_add(rq); - - return err; + i915_request_add(cache->rq); + cache->rq = NULL; } static void reloc_cache_reset(struct reloc_cache *cache) { void *vaddr; + if (cache->rq) + reloc_gpu_flush(cache); + if (!cache->vaddr) return; @@ -1309,6 +1214,7 @@ static int reloc_move_to_gpu(struct i915_request *rq, struct i915_vma *vma) static int __reloc_gpu_alloc(struct i915_execbuffer *eb, struct intel_engine_cs *engine, + struct i915_vma *vma, unsigned int len) { struct reloc_cache *cache = &eb->reloc_cache; @@ -1331,7 +1237,7 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, goto out_pool; } - batch = i915_vma_instance(pool->obj, eb->context->vm, NULL); + batch = i915_vma_instance(pool->obj, vma->vm, NULL); if (IS_ERR(batch)) { err = PTR_ERR(batch); goto err_unmap; @@ -1367,6 +1273,16 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, if (err) goto err_request; + err = reloc_move_to_gpu(rq, vma); + if (err) + goto err_request; + + err = eb->engine->emit_bb_start(rq, + batch->node.start, PAGE_SIZE, + cache->gen > 5 ? 0 : I915_DISPATCH_SECURE); + if (err) + goto skip_request; + i915_vma_lock(batch); err = i915_request_await_object(rq, batch->obj, false); if (err == 0) @@ -1381,7 +1297,6 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, cache->rq = rq; cache->rq_cmd = cmd; cache->rq_size = 0; - cache->rq_vma = batch; /* Return with batch mapping (cmd) still pinned */ goto out_pool; @@ -1410,9 +1325,12 @@ static u32 *reloc_gpu(struct i915_execbuffer *eb, { struct reloc_cache *cache = &eb->reloc_cache; u32 *cmd; - int err; + + if (cache->rq_size > PAGE_SIZE/sizeof(u32) - (len + 1)) + reloc_gpu_flush(cache); if (unlikely(!cache->rq)) { + int err; struct intel_engine_cs *engine = eb->engine; if (!reloc_can_use_engine(engine)) { @@ -1421,31 +1339,11 @@ static u32 *reloc_gpu(struct i915_execbuffer *eb, return ERR_PTR(-ENODEV); } - err = __reloc_gpu_alloc(eb, engine, len); + err = __reloc_gpu_alloc(eb, engine, vma, len); if (unlikely(err)) return ERR_PTR(err); } - if (vma != cache->target) { - err = reloc_move_to_gpu(cache->rq, vma); - if (unlikely(err)) { - i915_request_set_error_once(cache->rq, err); - return ERR_PTR(err); - } - - cache->target = vma; - } - - if (unlikely(cache->rq_size + len > - PAGE_SIZE / sizeof(u32) - RELOC_TAIL)) { - err = reloc_gpu_chain(cache); - if (unlikely(err)) { - i915_request_set_error_once(cache->rq, err); - return ERR_PTR(err); - } - } - - GEM_BUG_ON(cache->rq_size + len >= PAGE_SIZE / sizeof(u32)); cmd = cache->rq_cmd + cache->rq_size; cache->rq_size += len; @@ -1793,20 +1691,15 @@ static int eb_relocate(struct i915_execbuffer *eb) /* The objects are in their final locations, apply the relocations. */ if (eb->args->flags & __EXEC_HAS_RELOC) { struct eb_vma *ev; - int flush; list_for_each_entry(ev, &eb->relocs, reloc_link) { err = eb_relocate_vma(eb, ev); if (err) - break; + return err; } - - flush = reloc_gpu_flush(&eb->reloc_cache); - if (!err) - err = flush; } - return err; + return 0; } static int eb_move_to_gpu(struct i915_execbuffer *eb) diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c index a49016f8ee0d..580884cffec3 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c @@ -53,13 +53,13 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, } /* Skip to the end of the cmd page */ - i = PAGE_SIZE / sizeof(u32) - RELOC_TAIL - 1; + i = PAGE_SIZE / sizeof(u32) - 1; i -= eb->reloc_cache.rq_size; memset32(eb->reloc_cache.rq_cmd + eb->reloc_cache.rq_size, MI_NOOP, i); eb->reloc_cache.rq_size += i; - /* Force batch chaining */ + /* Force next batch */ if (!__reloc_entry_gpu(eb, vma, offsets[2] * sizeof(u32), 2)) { @@ -69,9 +69,7 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, GEM_BUG_ON(!eb->reloc_cache.rq); rq = i915_request_get(eb->reloc_cache.rq); - err = reloc_gpu_flush(&eb->reloc_cache); - if (err) - goto put_rq; + reloc_gpu_flush(&eb->reloc_cache); GEM_BUG_ON(eb->reloc_cache.rq); err = i915_gem_object_wait(obj, I915_WAIT_INTERRUPTIBLE, HZ / 2); From patchwork Mon Aug 10 10:30:42 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707253 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 86D8E1392 for ; Mon, 10 Aug 2020 10:33:30 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 6E93D206E9 for ; Mon, 10 Aug 2020 10:33:30 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 6E93D206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 079C389D9B; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id D5BFF89D9B for ; Mon, 10 Aug 2020 10:33:26 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:42 +0200 Message-Id: <20200810103103.303818-4-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 03/24] Revert "drm/i915/gem: Drop relocation slowpath". 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: Matthew Auld , Chris Wilson Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" This reverts commit 7dc8f1143778 ("drm/i915/gem: Drop relocation slowpath"). We need the slowpath relocation for taking ww-mutex inside the page fault handler, and we will take this mutex when pinning all objects. With this, we have a proper working slowpath again, which will allow us to do fault handling with WW locks held. [mlankhorst: Adjusted for reloc_gpu_flush() changes] Cc: Chris Wilson Cc: Matthew Auld Signed-off-by: Maarten Lankhorst Reviewed-by: Daniel Vetter --- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 263 +++++++++++++++++- 1 file changed, 253 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 6acbd08f82f0..12a130f92e72 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -707,7 +707,7 @@ static int eb_reserve(struct i915_execbuffer *eb) if (err) break; } - if (!(err == -ENOSPC || err == -EAGAIN)) + if (err != -ENOSPC) break; /* Resort *all* the objects into priority order */ @@ -738,13 +738,6 @@ static int eb_reserve(struct i915_execbuffer *eb) } list_splice_tail(&last, &eb->unbound); - if (err == -EAGAIN) { - mutex_unlock(&eb->i915->drm.struct_mutex); - flush_workqueue(eb->i915->mm.userptr_wq); - mutex_lock(&eb->i915->drm.struct_mutex); - continue; - } - switch (pass++) { case 0: break; @@ -1626,7 +1619,9 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) * we would try to acquire the struct mutex again. Obviously * this is bad and so lockdep complains vehemently. */ - copied = __copy_from_user(r, urelocs, count * sizeof(r[0])); + pagefault_disable(); + copied = __copy_from_user_inatomic(r, urelocs, count * sizeof(r[0])); + pagefault_enable(); if (unlikely(copied)) { remain = -EFAULT; goto out; @@ -1674,6 +1669,251 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) return remain; } +static int +eb_relocate_vma_slow(struct i915_execbuffer *eb, struct eb_vma *ev) +{ + const struct drm_i915_gem_exec_object2 *entry = ev->exec; + struct drm_i915_gem_relocation_entry *relocs = + u64_to_ptr(typeof(*relocs), entry->relocs_ptr); + unsigned int i; + int err; + + for (i = 0; i < entry->relocation_count; i++) { + u64 offset = eb_relocate_entry(eb, ev, &relocs[i]); + + if ((s64)offset < 0) { + err = (int)offset; + goto err; + } + } + err = 0; +err: + reloc_cache_reset(&eb->reloc_cache); + return err; +} + +static int check_relocations(const struct drm_i915_gem_exec_object2 *entry) +{ + const char __user *addr, *end; + unsigned long size; + char __maybe_unused c; + + size = entry->relocation_count; + if (size == 0) + return 0; + + if (size > N_RELOC(ULONG_MAX)) + return -EINVAL; + + addr = u64_to_user_ptr(entry->relocs_ptr); + size *= sizeof(struct drm_i915_gem_relocation_entry); + if (!access_ok(addr, size)) + return -EFAULT; + + end = addr + size; + for (; addr < end; addr += PAGE_SIZE) { + int err = __get_user(c, addr); + if (err) + return err; + } + return __get_user(c, end - 1); +} + +static int eb_copy_relocations(const struct i915_execbuffer *eb) +{ + struct drm_i915_gem_relocation_entry *relocs; + const unsigned int count = eb->buffer_count; + unsigned int i; + int err; + + for (i = 0; i < count; i++) { + const unsigned int nreloc = eb->exec[i].relocation_count; + struct drm_i915_gem_relocation_entry __user *urelocs; + unsigned long size; + unsigned long copied; + + if (nreloc == 0) + continue; + + err = check_relocations(&eb->exec[i]); + if (err) + goto err; + + urelocs = u64_to_user_ptr(eb->exec[i].relocs_ptr); + size = nreloc * sizeof(*relocs); + + relocs = kvmalloc_array(size, 1, GFP_KERNEL); + if (!relocs) { + err = -ENOMEM; + goto err; + } + + /* copy_from_user is limited to < 4GiB */ + copied = 0; + do { + unsigned int len = + min_t(u64, BIT_ULL(31), size - copied); + + if (__copy_from_user((char *)relocs + copied, + (char __user *)urelocs + copied, + len)) + goto end; + + copied += len; + } while (copied < size); + + /* + * As we do not update the known relocation offsets after + * relocating (due to the complexities in lock handling), + * we need to mark them as invalid now so that we force the + * relocation processing next time. Just in case the target + * object is evicted and then rebound into its old + * presumed_offset before the next execbuffer - if that + * happened we would make the mistake of assuming that the + * relocations were valid. + */ + if (!user_access_begin(urelocs, size)) + goto end; + + for (copied = 0; copied < nreloc; copied++) + unsafe_put_user(-1, + &urelocs[copied].presumed_offset, + end_user); + user_access_end(); + + eb->exec[i].relocs_ptr = (uintptr_t)relocs; + } + + return 0; + +end_user: + user_access_end(); +end: + kvfree(relocs); + err = -EFAULT; +err: + while (i--) { + relocs = u64_to_ptr(typeof(*relocs), eb->exec[i].relocs_ptr); + if (eb->exec[i].relocation_count) + kvfree(relocs); + } + return err; +} + +static int eb_prefault_relocations(const struct i915_execbuffer *eb) +{ + const unsigned int count = eb->buffer_count; + unsigned int i; + + for (i = 0; i < count; i++) { + int err; + + err = check_relocations(&eb->exec[i]); + if (err) + return err; + } + + return 0; +} + +static noinline int eb_relocate_slow(struct i915_execbuffer *eb) +{ + bool have_copy = false; + struct eb_vma *ev; + int err = 0; + +repeat: + if (signal_pending(current)) { + err = -ERESTARTSYS; + goto out; + } + + /* + * We take 3 passes through the slowpatch. + * + * 1 - we try to just prefault all the user relocation entries and + * then attempt to reuse the atomic pagefault disabled fast path again. + * + * 2 - we copy the user entries to a local buffer here outside of the + * local and allow ourselves to wait upon any rendering before + * relocations + * + * 3 - we already have a local copy of the relocation entries, but + * were interrupted (EAGAIN) whilst waiting for the objects, try again. + */ + if (!err) { + err = eb_prefault_relocations(eb); + } else if (!have_copy) { + err = eb_copy_relocations(eb); + have_copy = err == 0; + } else { + cond_resched(); + err = 0; + } + + flush_workqueue(eb->i915->mm.userptr_wq); + + if (err) + goto out; + + err = mutex_lock_interruptible(&eb->i915->drm.struct_mutex); + if (err) + goto out; + + list_for_each_entry(ev, &eb->relocs, reloc_link) { + if (!have_copy) { + pagefault_disable(); + err = eb_relocate_vma(eb, ev); + pagefault_enable(); + if (err) + break; + } else { + err = eb_relocate_vma_slow(eb, ev); + if (err) + break; + } + } + + reloc_gpu_flush(&eb->reloc_cache); + mutex_unlock(&eb->i915->drm.struct_mutex); + if (err && !have_copy) + goto repeat; + + if (err) + goto err; + + /* + * Leave the user relocations as are, this is the painfully slow path, + * and we want to avoid the complication of dropping the lock whilst + * having buffers reserved in the aperture and so causing spurious + * ENOSPC for random operations. + */ + +err: + if (err == -EAGAIN) + goto repeat; + +out: + if (have_copy) { + const unsigned int count = eb->buffer_count; + unsigned int i; + + for (i = 0; i < count; i++) { + const struct drm_i915_gem_exec_object2 *entry = + &eb->exec[i]; + struct drm_i915_gem_relocation_entry *relocs; + + if (!entry->relocation_count) + continue; + + relocs = u64_to_ptr(typeof(*relocs), entry->relocs_ptr); + kvfree(relocs); + } + } + + return err; +} + static int eb_relocate(struct i915_execbuffer *eb) { int err; @@ -1695,8 +1935,11 @@ static int eb_relocate(struct i915_execbuffer *eb) list_for_each_entry(ev, &eb->relocs, reloc_link) { err = eb_relocate_vma(eb, ev); if (err) - return err; + break; } + + if (err) + return eb_relocate_slow(eb); } return 0; From patchwork Mon Aug 10 10:30:43 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707255 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 0220E13A4 for ; Mon, 10 Aug 2020 10:33:34 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id DE7BD206E9 for ; Mon, 10 Aug 2020 10:33:33 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org DE7BD206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 6BAD96E3EC; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id E9889897E0 for ; Mon, 10 Aug 2020 10:33:26 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:43 +0200 Message-Id: <20200810103103.303818-5-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 04/24] Revert "drm/i915/gem: Split eb_vma into its own allocation" 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" This reverts commit 0f1dd02295f35dcdcbaafcbcbbec0753884ab974. With the WW locking, we will drop all references only at the end, so refcounting can be removed. Signed-off-by: Maarten Lankhorst Reviewed-by: Daniel Vetter Reviewed-by: Thomas Hellström --- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 124 +++++++----------- 1 file changed, 51 insertions(+), 73 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 12a130f92e72..7051af5ad8d8 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -41,11 +41,6 @@ struct eb_vma { u32 handle; }; -struct eb_vma_array { - struct kref kref; - struct eb_vma vma[]; -}; - enum { FORCE_CPU_RELOC = 1, FORCE_GTT_RELOC, @@ -58,6 +53,7 @@ enum { #define __EXEC_OBJECT_NEEDS_MAP BIT(29) #define __EXEC_OBJECT_NEEDS_BIAS BIT(28) #define __EXEC_OBJECT_INTERNAL_FLAGS (~0u << 28) /* all of the above */ +#define __EXEC_OBJECT_RESERVED (__EXEC_OBJECT_HAS_PIN | __EXEC_OBJECT_HAS_FENCE) #define __EXEC_HAS_RELOC BIT(31) #define __EXEC_INTERNAL_FLAGS (~0u << 31) @@ -295,7 +291,6 @@ struct i915_execbuffer { */ int lut_size; struct hlist_head *buckets; /** ht for relocation handles */ - struct eb_vma_array *array; struct eb_fence *fences; unsigned long num_fences; @@ -308,62 +303,8 @@ static inline bool eb_use_cmdparser(const struct i915_execbuffer *eb) eb->args->batch_len); } -static struct eb_vma_array *eb_vma_array_create(unsigned int count) -{ - struct eb_vma_array *arr; - - arr = kvmalloc(struct_size(arr, vma, count), GFP_KERNEL | __GFP_NOWARN); - if (!arr) - return NULL; - - kref_init(&arr->kref); - arr->vma[0].vma = NULL; - - return arr; -} - -static inline void eb_unreserve_vma(struct eb_vma *ev) -{ - struct i915_vma *vma = ev->vma; - - if (unlikely(ev->flags & __EXEC_OBJECT_HAS_FENCE)) - __i915_vma_unpin_fence(vma); - - if (ev->flags & __EXEC_OBJECT_HAS_PIN) - __i915_vma_unpin(vma); - - ev->flags &= ~(__EXEC_OBJECT_HAS_PIN | - __EXEC_OBJECT_HAS_FENCE); -} - -static void eb_vma_array_destroy(struct kref *kref) -{ - struct eb_vma_array *arr = container_of(kref, typeof(*arr), kref); - struct eb_vma *ev = arr->vma; - - while (ev->vma) { - eb_unreserve_vma(ev); - i915_vma_put(ev->vma); - ev++; - } - - kvfree(arr); -} - -static void eb_vma_array_put(struct eb_vma_array *arr) -{ - kref_put(&arr->kref, eb_vma_array_destroy); -} - static int eb_create(struct i915_execbuffer *eb) { - /* Allocate an extra slot for use by the command parser + sentinel */ - eb->array = eb_vma_array_create(eb->buffer_count + 2); - if (!eb->array) - return -ENOMEM; - - eb->vma = eb->array->vma; - if (!(eb->args->flags & I915_EXEC_HANDLE_LUT)) { unsigned int size = 1 + ilog2(eb->buffer_count); @@ -397,10 +338,8 @@ static int eb_create(struct i915_execbuffer *eb) break; } while (--size); - if (unlikely(!size)) { - eb_vma_array_put(eb->array); + if (unlikely(!size)) return -ENOMEM; - } eb->lut_size = size; } else { @@ -511,6 +450,26 @@ eb_pin_vma(struct i915_execbuffer *eb, return !eb_vma_misplaced(entry, vma, ev->flags); } +static inline void __eb_unreserve_vma(struct i915_vma *vma, unsigned int flags) +{ + GEM_BUG_ON(!(flags & __EXEC_OBJECT_HAS_PIN)); + + if (unlikely(flags & __EXEC_OBJECT_HAS_FENCE)) + __i915_vma_unpin_fence(vma); + + __i915_vma_unpin(vma); +} + +static inline void +eb_unreserve_vma(struct eb_vma *ev) +{ + if (!(ev->flags & __EXEC_OBJECT_HAS_PIN)) + return; + + __eb_unreserve_vma(ev->vma, ev->flags); + ev->flags &= ~__EXEC_OBJECT_RESERVED; +} + static int eb_validate_vma(struct i915_execbuffer *eb, struct drm_i915_gem_exec_object2 *entry, @@ -933,13 +892,31 @@ eb_get_vma(const struct i915_execbuffer *eb, unsigned long handle) } } +static void eb_release_vmas(const struct i915_execbuffer *eb) +{ + const unsigned int count = eb->buffer_count; + unsigned int i; + + for (i = 0; i < count; i++) { + struct eb_vma *ev = &eb->vma[i]; + struct i915_vma *vma = ev->vma; + + if (!vma) + break; + + eb->vma[i].vma = NULL; + + if (ev->flags & __EXEC_OBJECT_HAS_PIN) + __eb_unreserve_vma(vma, ev->flags); + + i915_vma_put(vma); + } +} + static void eb_destroy(const struct i915_execbuffer *eb) { GEM_BUG_ON(eb->reloc_cache.rq); - if (eb->array) - eb_vma_array_put(eb->array); - if (eb->lut_size > 0) kfree(eb->buckets); } @@ -2022,12 +1999,9 @@ static int eb_move_to_gpu(struct i915_execbuffer *eb) err = i915_vma_move_to_active(vma, eb->request, flags); i915_vma_unlock(vma); - eb_unreserve_vma(ev); } ww_acquire_fini(&acquire); - eb_vma_array_put(fetch_and_zero(&eb->array)); - if (unlikely(err)) goto err_skip; @@ -2324,7 +2298,6 @@ static int eb_parse(struct i915_execbuffer *eb) eb->vma[eb->buffer_count].vma = i915_vma_get(shadow); eb->vma[eb->buffer_count].flags = __EXEC_OBJECT_HAS_PIN; eb->batch = &eb->vma[eb->buffer_count++]; - eb->vma[eb->buffer_count].vma = NULL; eb->trampoline = trampoline; eb->batch_start_offset = 0; @@ -2990,6 +2963,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, args->flags |= __EXEC_HAS_RELOC; eb.exec = exec; + eb.vma = (struct eb_vma *)(exec + args->buffer_count + 1); + eb.vma[0].vma = NULL; eb.invalid_flags = __EXEC_OBJECT_UNKNOWN_FLAGS; reloc_cache_init(&eb.reloc_cache, eb.i915); @@ -3199,6 +3174,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, if (batch->private) intel_gt_buffer_pool_put(batch->private); err_vma: + if (eb.exec) + eb_release_vmas(&eb); if (eb.trampoline) i915_vma_unpin(eb.trampoline); eb_unpin_engine(&eb); @@ -3218,7 +3195,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, static size_t eb_element_size(void) { - return sizeof(struct drm_i915_gem_exec_object2); + return sizeof(struct drm_i915_gem_exec_object2) + sizeof(struct eb_vma); } static bool check_buffer_count(size_t count) @@ -3274,7 +3251,7 @@ i915_gem_execbuffer_ioctl(struct drm_device *dev, void *data, /* Copy in the exec list from userland */ exec_list = kvmalloc_array(count, sizeof(*exec_list), __GFP_NOWARN | GFP_KERNEL); - exec2_list = kvmalloc_array(count, eb_element_size(), + exec2_list = kvmalloc_array(count + 1, eb_element_size(), __GFP_NOWARN | GFP_KERNEL); if (exec_list == NULL || exec2_list == NULL) { drm_dbg(&i915->drm, @@ -3351,7 +3328,8 @@ i915_gem_execbuffer2_ioctl(struct drm_device *dev, void *data, if (err) return err; - exec2_list = kvmalloc_array(count, eb_element_size(), + /* Allocate an extra slot for use by the command parser */ + exec2_list = kvmalloc_array(count + 1, eb_element_size(), __GFP_NOWARN | GFP_KERNEL); if (exec2_list == NULL) { drm_dbg(&i915->drm, "Failed to allocate exec list for %zd buffers\n", From patchwork Mon Aug 10 10:30:44 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707293 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 71C0413B6 for ; Mon, 10 Aug 2020 10:33:51 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 5A178206E9 for ; Mon, 10 Aug 2020 10:33:51 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 5A178206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id C2CEB6E408; Mon, 10 Aug 2020 10:33:37 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id F1C7189CDE for ; Mon, 10 Aug 2020 10:33:27 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:44 +0200 Message-Id: <20200810103103.303818-6-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 05/24] drm/i915: Add an implementation for i915_gem_ww_ctx locking, v2. 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?= Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" i915_gem_ww_ctx is used to lock all gem bo's for pinning and memory eviction. We don't use it yet, but lets start adding the definition first. To use it, we have to pass a non-NULL ww to gem_object_lock, and don't unlock directly. It is done in i915_gem_ww_ctx_fini. Changes since v1: - Change ww_ctx and obj order in locking functions (Jonas Lahtinen) Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- drivers/gpu/drm/i915/display/intel_display.c | 4 +- .../gpu/drm/i915/gem/i915_gem_client_blt.c | 2 +- drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c | 4 +- drivers/gpu/drm/i915/gem/i915_gem_domain.c | 10 ++-- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 2 +- drivers/gpu/drm/i915/gem/i915_gem_object.h | 38 +++++++++++--- .../gpu/drm/i915/gem/i915_gem_object_types.h | 9 ++++ drivers/gpu/drm/i915/gem/i915_gem_pm.c | 2 +- drivers/gpu/drm/i915/gem/i915_gem_tiling.c | 2 +- .../gpu/drm/i915/gem/selftests/huge_pages.c | 2 +- .../i915/gem/selftests/i915_gem_client_blt.c | 2 +- .../i915/gem/selftests/i915_gem_coherency.c | 10 ++-- .../drm/i915/gem/selftests/i915_gem_context.c | 4 +- .../drm/i915/gem/selftests/i915_gem_mman.c | 4 +- .../drm/i915/gem/selftests/i915_gem_phys.c | 2 +- .../gpu/drm/i915/gt/selftest_workarounds.c | 2 +- drivers/gpu/drm/i915/gvt/cmd_parser.c | 2 +- drivers/gpu/drm/i915/i915_gem.c | 52 +++++++++++++++++-- drivers/gpu/drm/i915/i915_gem.h | 11 ++++ drivers/gpu/drm/i915/selftests/i915_gem.c | 41 +++++++++++++++ drivers/gpu/drm/i915/selftests/i915_vma.c | 2 +- .../drm/i915/selftests/intel_memory_region.c | 2 +- 22 files changed, 170 insertions(+), 39 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index 522c772a2111..5b4434289117 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -2311,7 +2311,7 @@ intel_pin_and_fence_fb_obj(struct drm_framebuffer *fb, void intel_unpin_fb_vma(struct i915_vma *vma, unsigned long flags) { - i915_gem_object_lock(vma->obj); + i915_gem_object_lock(vma->obj, NULL); if (flags & PLANE_HAS_FENCE) i915_vma_unpin_fence(vma); i915_gem_object_unpin_from_display_plane(vma); @@ -17149,7 +17149,7 @@ static int intel_framebuffer_init(struct intel_framebuffer *intel_fb, if (!intel_fb->frontbuffer) return -ENOMEM; - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); tiling = i915_gem_object_get_tiling(obj); stride = i915_gem_object_get_stride(obj); i915_gem_object_unlock(obj); diff --git a/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c b/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c index 947c8aa8e13e..a85907f29c53 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c @@ -288,7 +288,7 @@ int i915_gem_schedule_fill_pages_blt(struct drm_i915_gem_object *obj, dma_fence_init(&work->dma, &clear_pages_work_ops, &fence_lock, 0, 0); i915_sw_fence_init(&work->wait, clear_pages_work_notify); - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_sw_fence_await_reservation(&work->wait, obj->base.resv, NULL, true, 0, I915_FENCE_GFP); diff --git a/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c b/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c index 2679380159fc..27fddc22a7c6 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c @@ -128,7 +128,7 @@ static int i915_gem_begin_cpu_access(struct dma_buf *dma_buf, enum dma_data_dire if (err) return err; - err = i915_gem_object_lock_interruptible(obj); + err = i915_gem_object_lock_interruptible(obj, NULL); if (err) goto out; @@ -149,7 +149,7 @@ static int i915_gem_end_cpu_access(struct dma_buf *dma_buf, enum dma_data_direct if (err) return err; - err = i915_gem_object_lock_interruptible(obj); + err = i915_gem_object_lock_interruptible(obj, NULL); if (err) goto out; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_domain.c b/drivers/gpu/drm/i915/gem/i915_gem_domain.c index 7f76fc68f498..c0acfc97fae3 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_domain.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_domain.c @@ -32,7 +32,7 @@ void i915_gem_object_flush_if_display(struct drm_i915_gem_object *obj) if (!i915_gem_object_is_framebuffer(obj)) return; - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); __i915_gem_object_flush_for_display(obj); i915_gem_object_unlock(obj); } @@ -197,7 +197,7 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj, if (ret) return ret; - ret = i915_gem_object_lock_interruptible(obj); + ret = i915_gem_object_lock_interruptible(obj, NULL); if (ret) return ret; @@ -536,7 +536,7 @@ i915_gem_set_domain_ioctl(struct drm_device *dev, void *data, if (err) goto out; - err = i915_gem_object_lock_interruptible(obj); + err = i915_gem_object_lock_interruptible(obj, NULL); if (err) goto out_unpin; @@ -576,7 +576,7 @@ int i915_gem_object_prepare_read(struct drm_i915_gem_object *obj, if (!i915_gem_object_has_struct_page(obj)) return -ENODEV; - ret = i915_gem_object_lock_interruptible(obj); + ret = i915_gem_object_lock_interruptible(obj, NULL); if (ret) return ret; @@ -630,7 +630,7 @@ int i915_gem_object_prepare_write(struct drm_i915_gem_object *obj, if (!i915_gem_object_has_struct_page(obj)) return -ENODEV; - ret = i915_gem_object_lock_interruptible(obj); + ret = i915_gem_object_lock_interruptible(obj, NULL); if (ret) return ret; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 7051af5ad8d8..2bb414d55f10 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -1076,7 +1076,7 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj, if (use_cpu_reloc(cache, obj)) return NULL; - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_gtt_domain(obj, true); i915_gem_object_unlock(obj); if (err) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object.h b/drivers/gpu/drm/i915/gem/i915_gem_object.h index e5b9276d254c..488459143366 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_object.h @@ -110,20 +110,44 @@ i915_gem_object_put(struct drm_i915_gem_object *obj) #define assert_object_held(obj) dma_resv_assert_held((obj)->base.resv) -static inline void i915_gem_object_lock(struct drm_i915_gem_object *obj) +static inline int __i915_gem_object_lock(struct drm_i915_gem_object *obj, + struct i915_gem_ww_ctx *ww, + bool intr) { - dma_resv_lock(obj->base.resv, NULL); + int ret; + + if (intr) + ret = dma_resv_lock_interruptible(obj->base.resv, ww ? &ww->ctx : NULL); + else + ret = dma_resv_lock(obj->base.resv, ww ? &ww->ctx : NULL); + + if (!ret && ww) + list_add_tail(&obj->obj_link, &ww->obj_list); + if (ret == -EALREADY) + ret = 0; + + if (ret == -EDEADLK) + ww->contended = obj; + + return ret; } -static inline bool i915_gem_object_trylock(struct drm_i915_gem_object *obj) +static inline int i915_gem_object_lock(struct drm_i915_gem_object *obj, + struct i915_gem_ww_ctx *ww) { - return dma_resv_trylock(obj->base.resv); + return __i915_gem_object_lock(obj, ww, ww && ww->intr); } -static inline int -i915_gem_object_lock_interruptible(struct drm_i915_gem_object *obj) +static inline int i915_gem_object_lock_interruptible(struct drm_i915_gem_object *obj, + struct i915_gem_ww_ctx *ww) { - return dma_resv_lock_interruptible(obj->base.resv, NULL); + WARN_ON(ww && !ww->intr); + return __i915_gem_object_lock(obj, ww, true); +} + +static inline bool i915_gem_object_trylock(struct drm_i915_gem_object *obj) +{ + return dma_resv_trylock(obj->base.resv); } static inline void i915_gem_object_unlock(struct drm_i915_gem_object *obj) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_types.h b/drivers/gpu/drm/i915/gem/i915_gem_object_types.h index d0847d7896f9..b5c15557cc87 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object_types.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_object_types.h @@ -123,6 +123,15 @@ struct drm_i915_gem_object { struct list_head lut_list; spinlock_t lut_lock; /* guards lut_list */ + /** + * @obj_link: Link into @i915_gem_ww_ctx.obj_list + * + * When we lock this object through i915_gem_object_lock() with a + * context, we add it to the list to ensure we can unlock everything + * when i915_gem_ww_ctx_backoff() or i915_gem_ww_ctx_fini() are called. + */ + struct list_head obj_link; + /** Stolen memory for this object, instead of being backed by shmem. */ struct drm_mm_node *stolen; union { diff --git a/drivers/gpu/drm/i915/gem/i915_gem_pm.c b/drivers/gpu/drm/i915/gem/i915_gem_pm.c index 3d215164dd5a..40d3e40500fa 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_pm.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_pm.c @@ -84,7 +84,7 @@ void i915_gem_suspend_late(struct drm_i915_private *i915) spin_unlock_irqrestore(&i915->mm.obj_lock, flags); - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); drm_WARN_ON(&i915->drm, i915_gem_object_set_to_gtt_domain(obj, false)); i915_gem_object_unlock(obj); diff --git a/drivers/gpu/drm/i915/gem/i915_gem_tiling.c b/drivers/gpu/drm/i915/gem/i915_gem_tiling.c index ff72ee2fd9cd..ffcaee74a249 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_tiling.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_tiling.c @@ -249,7 +249,7 @@ i915_gem_object_set_tiling(struct drm_i915_gem_object *obj, * whilst executing a fenced command for an untiled object. */ - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); if (i915_gem_object_is_framebuffer(obj)) { i915_gem_object_unlock(obj); return -EBUSY; diff --git a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c index e2f3d014acb2..68c3631f3746 100644 --- a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c +++ b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c @@ -947,7 +947,7 @@ static int gpu_write(struct intel_context *ce, { int err; - i915_gem_object_lock(vma->obj); + i915_gem_object_lock(vma->obj, NULL); err = i915_gem_object_set_to_gtt_domain(vma->obj, true); i915_gem_object_unlock(vma->obj); if (err) 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 299c29e9ad86..4e36d4897ea6 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 @@ -75,7 +75,7 @@ static int __igt_client_fill(struct intel_engine_cs *engine) if (err) goto err_unpin; - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_cpu_domain(obj, false); i915_gem_object_unlock(obj); if (err) diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c index 87d7d8aa080f..1de2959b153c 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c @@ -82,7 +82,7 @@ static int gtt_set(struct context *ctx, unsigned long offset, u32 v) u32 __iomem *map; int err = 0; - i915_gem_object_lock(ctx->obj); + i915_gem_object_lock(ctx->obj, NULL); err = i915_gem_object_set_to_gtt_domain(ctx->obj, true); i915_gem_object_unlock(ctx->obj); if (err) @@ -115,7 +115,7 @@ static int gtt_get(struct context *ctx, unsigned long offset, u32 *v) u32 __iomem *map; int err = 0; - i915_gem_object_lock(ctx->obj); + i915_gem_object_lock(ctx->obj, NULL); err = i915_gem_object_set_to_gtt_domain(ctx->obj, false); i915_gem_object_unlock(ctx->obj); if (err) @@ -147,7 +147,7 @@ static int wc_set(struct context *ctx, unsigned long offset, u32 v) u32 *map; int err; - i915_gem_object_lock(ctx->obj); + i915_gem_object_lock(ctx->obj, NULL); err = i915_gem_object_set_to_wc_domain(ctx->obj, true); i915_gem_object_unlock(ctx->obj); if (err) @@ -170,7 +170,7 @@ static int wc_get(struct context *ctx, unsigned long offset, u32 *v) u32 *map; int err; - i915_gem_object_lock(ctx->obj); + i915_gem_object_lock(ctx->obj, NULL); err = i915_gem_object_set_to_wc_domain(ctx->obj, false); i915_gem_object_unlock(ctx->obj); if (err) @@ -193,7 +193,7 @@ static int gpu_set(struct context *ctx, unsigned long offset, u32 v) u32 *cs; int err; - i915_gem_object_lock(ctx->obj); + i915_gem_object_lock(ctx->obj, NULL); err = i915_gem_object_set_to_gtt_domain(ctx->obj, true); i915_gem_object_unlock(ctx->obj); if (err) 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 d176b015353f..52af9a52b743 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c @@ -950,7 +950,7 @@ emit_rpcs_query(struct drm_i915_gem_object *obj, if (IS_ERR(vma)) return PTR_ERR(vma); - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_gtt_domain(obj, false); i915_gem_object_unlock(obj); if (err) @@ -1709,7 +1709,7 @@ static int read_from_scratch(struct i915_gem_context *ctx, i915_request_add(rq); - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_cpu_domain(obj, false); i915_gem_object_unlock(obj); if (err) 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 9c7402ce5bf9..9fb95a45bcad 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c @@ -103,7 +103,7 @@ static int check_partial_mapping(struct drm_i915_gem_object *obj, GEM_BUG_ON(i915_gem_object_get_tiling(obj) != tile->tiling); GEM_BUG_ON(i915_gem_object_get_stride(obj) != tile->stride); - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_gtt_domain(obj, true); i915_gem_object_unlock(obj); if (err) { @@ -188,7 +188,7 @@ static int check_partial_mappings(struct drm_i915_gem_object *obj, GEM_BUG_ON(i915_gem_object_get_tiling(obj) != tile->tiling); GEM_BUG_ON(i915_gem_object_get_stride(obj) != tile->stride); - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_gtt_domain(obj, true); i915_gem_object_unlock(obj); if (err) { diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_phys.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_phys.c index 34932871b3a5..a94243dc4c5c 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_phys.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_phys.c @@ -44,7 +44,7 @@ static int mock_phys_object(void *arg) } /* Make the object dirty so that put_pages must do copy back the data */ - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_gtt_domain(obj, true); i915_gem_object_unlock(obj); if (err) { diff --git a/drivers/gpu/drm/i915/gt/selftest_workarounds.c b/drivers/gpu/drm/i915/gt/selftest_workarounds.c index febc9e6692ba..61a0532d0f3d 100644 --- a/drivers/gpu/drm/i915/gt/selftest_workarounds.c +++ b/drivers/gpu/drm/i915/gt/selftest_workarounds.c @@ -214,7 +214,7 @@ static int check_whitelist(struct i915_gem_context *ctx, return PTR_ERR(results); err = 0; - i915_gem_object_lock(results); + i915_gem_object_lock(results, NULL); intel_wedge_on_timeout(&wedge, engine->gt, HZ / 5) /* safety net! */ err = i915_gem_object_set_to_cpu_domain(results, false); i915_gem_object_unlock(results); diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c index f1940939260a..943c8d232703 100644 --- a/drivers/gpu/drm/i915/gvt/cmd_parser.c +++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c @@ -2982,7 +2982,7 @@ static int shadow_indirect_ctx(struct intel_shadow_wa_ctx *wa_ctx) goto put_obj; } - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); ret = i915_gem_object_set_to_cpu_domain(obj, false); i915_gem_object_unlock(obj); if (ret) { diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index e1de50780ed5..4bc8fdb94019 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -420,7 +420,7 @@ i915_gem_gtt_pread(struct drm_i915_gem_object *obj, GEM_BUG_ON(!drm_mm_node_allocated(&node)); } - ret = i915_gem_object_lock_interruptible(obj); + ret = i915_gem_object_lock_interruptible(obj, NULL); if (ret) goto out_unpin; @@ -619,7 +619,7 @@ i915_gem_gtt_pwrite_fast(struct drm_i915_gem_object *obj, GEM_BUG_ON(!drm_mm_node_allocated(&node)); } - ret = i915_gem_object_lock_interruptible(obj); + ret = i915_gem_object_lock_interruptible(obj, NULL); if (ret) goto out_unpin; @@ -1290,7 +1290,7 @@ int i915_gem_freeze_late(struct drm_i915_private *i915) i915_gem_drain_freed_objects(i915); list_for_each_entry(obj, &i915->mm.shrink_list, mm.link) { - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); drm_WARN_ON(&i915->drm, i915_gem_object_set_to_cpu_domain(obj, true)); i915_gem_object_unlock(obj); @@ -1326,6 +1326,52 @@ int i915_gem_open(struct drm_i915_private *i915, struct drm_file *file) return ret; } +void i915_gem_ww_ctx_init(struct i915_gem_ww_ctx *ww, bool intr) +{ + ww_acquire_init(&ww->ctx, &reservation_ww_class); + INIT_LIST_HEAD(&ww->obj_list); + ww->intr = intr; + ww->contended = NULL; +} + +static void i915_gem_ww_ctx_unlock_all(struct i915_gem_ww_ctx *ww) +{ + struct drm_i915_gem_object *obj; + + while ((obj = list_first_entry_or_null(&ww->obj_list, struct drm_i915_gem_object, obj_link))) { + list_del(&obj->obj_link); + i915_gem_object_unlock(obj); + } +} + +void i915_gem_ww_ctx_fini(struct i915_gem_ww_ctx *ww) +{ + i915_gem_ww_ctx_unlock_all(ww); + WARN_ON(ww->contended); + ww_acquire_fini(&ww->ctx); +} + +int __must_check i915_gem_ww_ctx_backoff(struct i915_gem_ww_ctx *ww) +{ + int ret = 0; + + if (WARN_ON(!ww->contended)) + return -EINVAL; + + i915_gem_ww_ctx_unlock_all(ww); + if (ww->intr) + ret = dma_resv_lock_slow_interruptible(ww->contended->base.resv, &ww->ctx); + else + dma_resv_lock_slow(ww->contended->base.resv, &ww->ctx); + + if (!ret) + list_add_tail(&ww->contended->obj_link, &ww->obj_list); + + ww->contended = NULL; + + return ret; +} + #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST) #include "selftests/mock_gem_device.c" #include "selftests/i915_gem.c" diff --git a/drivers/gpu/drm/i915/i915_gem.h b/drivers/gpu/drm/i915/i915_gem.h index f333e88a2b6e..427ebd02fceb 100644 --- a/drivers/gpu/drm/i915/i915_gem.h +++ b/drivers/gpu/drm/i915/i915_gem.h @@ -116,4 +116,15 @@ static inline bool __tasklet_is_scheduled(struct tasklet_struct *t) return test_bit(TASKLET_STATE_SCHED, &t->state); } +struct i915_gem_ww_ctx { + struct ww_acquire_ctx ctx; + struct list_head obj_list; + bool intr; + struct drm_i915_gem_object *contended; +}; + +void i915_gem_ww_ctx_init(struct i915_gem_ww_ctx *ctx, bool intr); +void i915_gem_ww_ctx_fini(struct i915_gem_ww_ctx *ctx); +int __must_check i915_gem_ww_ctx_backoff(struct i915_gem_ww_ctx *ctx); + #endif /* __I915_GEM_H__ */ diff --git a/drivers/gpu/drm/i915/selftests/i915_gem.c b/drivers/gpu/drm/i915/selftests/i915_gem.c index 88d400b9df88..23a6132c5f4e 100644 --- a/drivers/gpu/drm/i915/selftests/i915_gem.c +++ b/drivers/gpu/drm/i915/selftests/i915_gem.c @@ -199,11 +199,52 @@ static int igt_gem_hibernate(void *arg) return err; } +static int igt_gem_ww_ctx(void *arg) +{ + struct drm_i915_private *i915 = arg; + struct drm_i915_gem_object *obj, *obj2; + struct i915_gem_ww_ctx ww; + int err = 0; + + obj = i915_gem_object_create_internal(i915, PAGE_SIZE); + if (IS_ERR(obj)) + return PTR_ERR(obj); + + obj2 = i915_gem_object_create_internal(i915, PAGE_SIZE); + if (IS_ERR(obj)) { + err = PTR_ERR(obj); + goto put1; + } + + i915_gem_ww_ctx_init(&ww, true); +retry: + /* Lock the objects, twice for good measure (-EALREADY handling) */ + err = i915_gem_object_lock(obj, &ww); + if (!err) + err = i915_gem_object_lock_interruptible(obj, &ww); + if (!err) + err = i915_gem_object_lock_interruptible(obj2, &ww); + if (!err) + err = i915_gem_object_lock(obj2, &ww); + + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + i915_gem_object_put(obj2); +put1: + i915_gem_object_put(obj); + return err; +} + int i915_gem_live_selftests(struct drm_i915_private *i915) { static const struct i915_subtest tests[] = { SUBTEST(igt_gem_suspend), SUBTEST(igt_gem_hibernate), + SUBTEST(igt_gem_ww_ctx), }; if (intel_gt_is_wedged(&i915->gt)) diff --git a/drivers/gpu/drm/i915/selftests/i915_vma.c b/drivers/gpu/drm/i915/selftests/i915_vma.c index af89c7fc8f59..88c5e9acb84c 100644 --- a/drivers/gpu/drm/i915/selftests/i915_vma.c +++ b/drivers/gpu/drm/i915/selftests/i915_vma.c @@ -892,7 +892,7 @@ static int igt_vma_remapped_gtt(void *arg) unsigned int x, y; int err; - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_gtt_domain(obj, true); i915_gem_object_unlock(obj); if (err) diff --git a/drivers/gpu/drm/i915/selftests/intel_memory_region.c b/drivers/gpu/drm/i915/selftests/intel_memory_region.c index 741270fcb6b9..93a38a323584 100644 --- a/drivers/gpu/drm/i915/selftests/intel_memory_region.c +++ b/drivers/gpu/drm/i915/selftests/intel_memory_region.c @@ -509,7 +509,7 @@ static int igt_lmem_write_cpu(void *arg) if (err) goto out_unpin; - i915_gem_object_lock(obj); + i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_wc_domain(obj, true); i915_gem_object_unlock(obj); if (err) From patchwork Mon Aug 10 10:30:45 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707299 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 1E5BE13B6 for ; Mon, 10 Aug 2020 10:33:53 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 068FD206E9 for ; Mon, 10 Aug 2020 10:33:53 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 068FD206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 7B9AD6E0E3; Mon, 10 Aug 2020 10:33:44 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id EE22689C3B for ; Mon, 10 Aug 2020 10:33:27 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:45 +0200 Message-Id: <20200810103103.303818-7-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 06/24] drm/i915: Remove locking from i915_gem_object_prepare_read/write 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Execbuffer submission will perform its own WW locking, and we cannot rely on the implicit lock there. This also makes it clear that the GVT code will get a lockdep splat when multiple batchbuffer shadows need to be performed in the same instance, fix that up. Signed-off-by: Maarten Lankhorst Reviewed-by: Tvrtko Ursulin --- drivers/gpu/drm/i915/gem/i915_gem_domain.c | 20 ++++++------------- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 13 ++++++++++-- drivers/gpu/drm/i915/gem/i915_gem_object.h | 1 - .../gpu/drm/i915/gem/selftests/huge_pages.c | 5 ++++- .../i915/gem/selftests/i915_gem_coherency.c | 14 +++++++++---- .../drm/i915/gem/selftests/i915_gem_context.c | 12 ++++++++--- drivers/gpu/drm/i915/gvt/cmd_parser.c | 1 + drivers/gpu/drm/i915/i915_gem.c | 20 +++++++++++++++++-- 8 files changed, 59 insertions(+), 27 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_domain.c b/drivers/gpu/drm/i915/gem/i915_gem_domain.c index c0acfc97fae3..8ebceebd11b0 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_domain.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_domain.c @@ -576,19 +576,17 @@ int i915_gem_object_prepare_read(struct drm_i915_gem_object *obj, if (!i915_gem_object_has_struct_page(obj)) return -ENODEV; - ret = i915_gem_object_lock_interruptible(obj, NULL); - if (ret) - return ret; + assert_object_held(obj); ret = i915_gem_object_wait(obj, I915_WAIT_INTERRUPTIBLE, MAX_SCHEDULE_TIMEOUT); if (ret) - goto err_unlock; + return ret; ret = i915_gem_object_pin_pages(obj); if (ret) - goto err_unlock; + return ret; if (obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_READ || !static_cpu_has(X86_FEATURE_CLFLUSH)) { @@ -616,8 +614,6 @@ int i915_gem_object_prepare_read(struct drm_i915_gem_object *obj, err_unpin: i915_gem_object_unpin_pages(obj); -err_unlock: - i915_gem_object_unlock(obj); return ret; } @@ -630,20 +626,18 @@ int i915_gem_object_prepare_write(struct drm_i915_gem_object *obj, if (!i915_gem_object_has_struct_page(obj)) return -ENODEV; - ret = i915_gem_object_lock_interruptible(obj, NULL); - if (ret) - return ret; + assert_object_held(obj); ret = i915_gem_object_wait(obj, I915_WAIT_INTERRUPTIBLE | I915_WAIT_ALL, MAX_SCHEDULE_TIMEOUT); if (ret) - goto err_unlock; + return ret; ret = i915_gem_object_pin_pages(obj); if (ret) - goto err_unlock; + return ret; if (obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_WRITE || !static_cpu_has(X86_FEATURE_CLFLUSH)) { @@ -680,7 +674,5 @@ int i915_gem_object_prepare_write(struct drm_i915_gem_object *obj, err_unpin: i915_gem_object_unpin_pages(obj); -err_unlock: - i915_gem_object_unlock(obj); return ret; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 2bb414d55f10..ebb45fdfad0e 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -991,11 +991,14 @@ static void reloc_cache_reset(struct reloc_cache *cache) vaddr = unmask_page(cache->vaddr); if (cache->vaddr & KMAP) { + struct drm_i915_gem_object *obj = + (struct drm_i915_gem_object *)cache->node.mm; if (cache->vaddr & CLFLUSH_AFTER) mb(); kunmap_atomic(vaddr); - i915_gem_object_finish_access((struct drm_i915_gem_object *)cache->node.mm); + i915_gem_object_finish_access(obj); + i915_gem_object_unlock(obj); } else { struct i915_ggtt *ggtt = cache_to_ggtt(cache); @@ -1031,10 +1034,16 @@ static void *reloc_kmap(struct drm_i915_gem_object *obj, unsigned int flushes; int err; - err = i915_gem_object_prepare_write(obj, &flushes); + err = i915_gem_object_lock_interruptible(obj, NULL); if (err) return ERR_PTR(err); + err = i915_gem_object_prepare_write(obj, &flushes); + if (err) { + i915_gem_object_unlock(obj); + return ERR_PTR(err); + } + BUILD_BUG_ON(KMAP & CLFLUSH_FLAGS); BUILD_BUG_ON((KMAP | CLFLUSH_FLAGS) & PAGE_MASK); diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object.h b/drivers/gpu/drm/i915/gem/i915_gem_object.h index 488459143366..45d79d75e73a 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_object.h @@ -432,7 +432,6 @@ static inline void i915_gem_object_finish_access(struct drm_i915_gem_object *obj) { i915_gem_object_unpin_pages(obj); - i915_gem_object_unlock(obj); } static inline struct intel_engine_cs * diff --git a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c index 68c3631f3746..5daf4a2be422 100644 --- a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c +++ b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c @@ -964,9 +964,10 @@ __cpu_check_shmem(struct drm_i915_gem_object *obj, u32 dword, u32 val) unsigned long n; int err; + i915_gem_object_lock(obj, NULL); err = i915_gem_object_prepare_read(obj, &needs_flush); if (err) - return err; + goto err_unlock; for (n = 0; n < obj->base.size >> PAGE_SHIFT; ++n) { u32 *ptr = kmap_atomic(i915_gem_object_get_page(obj, n)); @@ -986,6 +987,8 @@ __cpu_check_shmem(struct drm_i915_gem_object *obj, u32 dword, u32 val) } i915_gem_object_finish_access(obj); +err_unlock: + i915_gem_object_unlock(obj); return err; } diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c index 1de2959b153c..dcdfc396f2f8 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c @@ -27,9 +27,10 @@ static int cpu_set(struct context *ctx, unsigned long offset, u32 v) u32 *cpu; int err; + i915_gem_object_lock(ctx->obj, NULL); err = i915_gem_object_prepare_write(ctx->obj, &needs_clflush); if (err) - return err; + goto out; page = i915_gem_object_get_page(ctx->obj, offset >> PAGE_SHIFT); map = kmap_atomic(page); @@ -46,7 +47,9 @@ static int cpu_set(struct context *ctx, unsigned long offset, u32 v) kunmap_atomic(map); i915_gem_object_finish_access(ctx->obj); - return 0; +out: + i915_gem_object_unlock(ctx->obj); + return err; } static int cpu_get(struct context *ctx, unsigned long offset, u32 *v) @@ -57,9 +60,10 @@ static int cpu_get(struct context *ctx, unsigned long offset, u32 *v) u32 *cpu; int err; + i915_gem_object_lock(ctx->obj, NULL); err = i915_gem_object_prepare_read(ctx->obj, &needs_clflush); if (err) - return err; + goto out; page = i915_gem_object_get_page(ctx->obj, offset >> PAGE_SHIFT); map = kmap_atomic(page); @@ -73,7 +77,9 @@ static int cpu_get(struct context *ctx, unsigned long offset, u32 *v) kunmap_atomic(map); i915_gem_object_finish_access(ctx->obj); - return 0; +out: + i915_gem_object_unlock(ctx->obj); + return err; } static int gtt_set(struct context *ctx, unsigned long offset, u32 v) 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 52af9a52b743..9160a335c684 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c @@ -461,9 +461,10 @@ static int cpu_fill(struct drm_i915_gem_object *obj, u32 value) unsigned int n, m, need_flush; int err; + i915_gem_object_lock(obj, NULL); err = i915_gem_object_prepare_write(obj, &need_flush); if (err) - return err; + goto out; for (n = 0; n < real_page_count(obj); n++) { u32 *map; @@ -479,7 +480,9 @@ static int cpu_fill(struct drm_i915_gem_object *obj, u32 value) i915_gem_object_finish_access(obj); obj->read_domains = I915_GEM_DOMAIN_GTT | I915_GEM_DOMAIN_CPU; obj->write_domain = 0; - return 0; +out: + i915_gem_object_unlock(obj); + return err; } static noinline int cpu_check(struct drm_i915_gem_object *obj, @@ -488,9 +491,10 @@ static noinline int cpu_check(struct drm_i915_gem_object *obj, unsigned int n, m, needs_flush; int err; + i915_gem_object_lock(obj, NULL); err = i915_gem_object_prepare_read(obj, &needs_flush); if (err) - return err; + goto out_unlock; for (n = 0; n < real_page_count(obj); n++) { u32 *map; @@ -527,6 +531,8 @@ static noinline int cpu_check(struct drm_i915_gem_object *obj, } i915_gem_object_finish_access(obj); +out_unlock: + i915_gem_object_unlock(obj); return err; } diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c index 943c8d232703..d0a599b51bfe 100644 --- a/drivers/gpu/drm/i915/gvt/cmd_parser.c +++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c @@ -1923,6 +1923,7 @@ static int perform_bb_shadow(struct parser_exec_state *s) if (ret) goto err_unmap; + i915_gem_object_unlock(bb->obj); INIT_LIST_HEAD(&bb->list); list_add(&bb->list, &s->workload->shadow_bb); diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 4bc8fdb94019..55158e487a7f 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -335,12 +335,20 @@ i915_gem_shmem_pread(struct drm_i915_gem_object *obj, u64 remain; int ret; - ret = i915_gem_object_prepare_read(obj, &needs_clflush); + ret = i915_gem_object_lock_interruptible(obj, NULL); if (ret) return ret; + ret = i915_gem_object_prepare_read(obj, &needs_clflush); + if (ret) { + i915_gem_object_unlock(obj); + return ret; + } + fence = i915_gem_object_lock_fence(obj); i915_gem_object_finish_access(obj); + i915_gem_object_unlock(obj); + if (!fence) return -ENOMEM; @@ -734,12 +742,20 @@ i915_gem_shmem_pwrite(struct drm_i915_gem_object *obj, u64 remain; int ret; - ret = i915_gem_object_prepare_write(obj, &needs_clflush); + ret = i915_gem_object_lock_interruptible(obj, NULL); if (ret) return ret; + ret = i915_gem_object_prepare_write(obj, &needs_clflush); + if (ret) { + i915_gem_object_unlock(obj); + return ret; + } + fence = i915_gem_object_lock_fence(obj); i915_gem_object_finish_access(obj); + i915_gem_object_unlock(obj); + if (!fence) return -ENOMEM; From patchwork Mon Aug 10 10:30:46 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707279 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 0A7161392 for ; Mon, 10 Aug 2020 10:33:47 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id E68B3206E9 for ; Mon, 10 Aug 2020 10:33:46 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org E68B3206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id AA0F06E3DB; Mon, 10 Aug 2020 10:33:36 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id ECC07897E0 for ; Mon, 10 Aug 2020 10:33:27 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:46 +0200 Message-Id: <20200810103103.303818-8-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 07/24] drm/i915: Parse command buffer earlier in eb_relocate(slow) 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" We want to introduce backoff logic, but we need to lock the pool object as well for command parsing. Because of this, we will need backoff logic for the engine pool obj, move the batch validation up slightly to eb_lookup_vmas, and the actual command parsing in a separate function which can get called from execbuf relocation fast and slowpath. Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 68 ++++++++++--------- 1 file changed, 37 insertions(+), 31 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index ebb45fdfad0e..23811fc4c8bb 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -296,6 +296,8 @@ struct i915_execbuffer { unsigned long num_fences; }; +static int eb_parse(struct i915_execbuffer *eb); + static inline bool eb_use_cmdparser(const struct i915_execbuffer *eb) { return intel_engine_requires_cmd_parser(eb->engine) || @@ -843,6 +845,7 @@ static struct i915_vma *eb_lookup_vma(struct i915_execbuffer *eb, u32 handle) static int eb_lookup_vmas(struct i915_execbuffer *eb) { + struct drm_i915_private *i915 = eb->i915; unsigned int batch = eb_batch_index(eb); unsigned int i; int err = 0; @@ -856,18 +859,37 @@ static int eb_lookup_vmas(struct i915_execbuffer *eb) vma = eb_lookup_vma(eb, eb->exec[i].handle); if (IS_ERR(vma)) { err = PTR_ERR(vma); - break; + goto err; } err = eb_validate_vma(eb, &eb->exec[i], vma); if (unlikely(err)) { i915_vma_put(vma); - break; + goto err; } eb_add_vma(eb, i, batch, vma); } + if (unlikely(eb->batch->flags & EXEC_OBJECT_WRITE)) { + drm_dbg(&i915->drm, + "Attempting to use self-modifying batch buffer\n"); + return -EINVAL; + } + + if (range_overflows_t(u64, + eb->batch_start_offset, eb->batch_len, + eb->batch->vma->size)) { + drm_dbg(&i915->drm, "Attempting to use out-of-bounds batch\n"); + return -EINVAL; + } + + if (eb->batch_len == 0) + eb->batch_len = eb->batch->vma->size - eb->batch_start_offset; + + return 0; + +err: eb->vma[i].vma = NULL; return err; } @@ -1802,7 +1824,7 @@ static int eb_prefault_relocations(const struct i915_execbuffer *eb) return 0; } -static noinline int eb_relocate_slow(struct i915_execbuffer *eb) +static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) { bool have_copy = false; struct eb_vma *ev; @@ -1868,6 +1890,11 @@ static noinline int eb_relocate_slow(struct i915_execbuffer *eb) if (err) goto err; + /* as last step, parse the command buffer */ + err = eb_parse(eb); + if (err) + goto err; + /* * Leave the user relocations as are, this is the painfully slow path, * and we want to avoid the complication of dropping the lock whilst @@ -1900,7 +1927,7 @@ static noinline int eb_relocate_slow(struct i915_execbuffer *eb) return err; } -static int eb_relocate(struct i915_execbuffer *eb) +static int eb_relocate_parse(struct i915_execbuffer *eb) { int err; @@ -1925,10 +1952,10 @@ static int eb_relocate(struct i915_execbuffer *eb) } if (err) - return eb_relocate_slow(eb); + return eb_relocate_parse_slow(eb); } - return 0; + return eb_parse(eb); } static int eb_move_to_gpu(struct i915_execbuffer *eb) @@ -3046,7 +3073,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, if (unlikely(err)) goto err_context; - err = eb_relocate(&eb); + err = eb_relocate_parse(&eb); if (err) { /* * If the user expects the execobject.offset and @@ -3059,33 +3086,10 @@ i915_gem_do_execbuffer(struct drm_device *dev, goto err_vma; } - if (unlikely(eb.batch->flags & EXEC_OBJECT_WRITE)) { - drm_dbg(&i915->drm, - "Attempting to use self-modifying batch buffer\n"); - err = -EINVAL; - goto err_vma; - } - - if (range_overflows_t(u64, - eb.batch_start_offset, eb.batch_len, - eb.batch->vma->size)) { - drm_dbg(&i915->drm, "Attempting to use out-of-bounds batch\n"); - err = -EINVAL; - goto err_vma; - } - - if (eb.batch_len == 0) - eb.batch_len = eb.batch->vma->size - eb.batch_start_offset; - - err = eb_parse(&eb); - if (err) - goto err_vma; - /* * snb/ivb/vlv conflate the "batch in ppgtt" bit with the "non-secure * batch" bit. Hence we need to pin secure batches into the global gtt. * hsw should have this fixed, but bdw mucks it up again. */ - batch = eb.batch->vma; if (eb.batch_flags & I915_DISPATCH_SECURE) { struct i915_vma *vma; @@ -3099,13 +3103,15 @@ i915_gem_do_execbuffer(struct drm_device *dev, * fitting due to fragmentation. * So this is actually safe. */ - vma = i915_gem_object_ggtt_pin(batch->obj, NULL, 0, 0, 0); + vma = i915_gem_object_ggtt_pin(eb.batch->vma->obj, NULL, 0, 0, 0); if (IS_ERR(vma)) { err = PTR_ERR(vma); goto err_parse; } batch = vma; + } else { + batch = eb.batch->vma; } /* All GPU relocation batches must be submitted prior to the user rq */ From patchwork Mon Aug 10 10:30:47 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707263 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id EB98013A4 for ; Mon, 10 Aug 2020 10:33:39 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id D3E76206E9 for ; Mon, 10 Aug 2020 10:33:39 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org D3E76206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 7AB4689FC9; Mon, 10 Aug 2020 10:33:30 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 1016889FC9 for ; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:47 +0200 Message-Id: <20200810103103.303818-9-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 08/24] drm/i915: Use per object locking in execbuf, v12. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Now that we changed execbuf submission slightly to allow us to do all pinning in one place, we can now simply add ww versions on top of struct_mutex. All we have to do is a separate path for -EDEADLK handling, which needs to unpin all gem bo's before dropping the lock, then starting over. This finally allows us to do parallel submission, but because not all of the pinning code uses the ww ctx yet, we cannot completely drop struct_mutex yet. Changes since v1: - Keep struct_mutex for now. :( Changes since v2: - Make sure we always lock the ww context in slowpath. Changes since v3: - Don't call __eb_unreserve_vma in eb_move_to_gpu now; this can be done on normal unlock path. - Unconditionally release vmas and context. Changes since v4: - Rebased on top of struct_mutex reduction. Changes since v5: - Remove training wheels. Changes since v6: - Fix accidentally broken -ENOSPC handling. Changes since v7: - Handle gt buffer pool better. Changes since v8: - Properly clear variables, to make -EDEADLK handling not BUG. Change since v9: - Fix unpinning fence on pnv and below. Changes since v10: - Make relocation gpu chaining working again. Changes since v11: - Remove relocation chaining, pain to make it work. Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 364 +++++++++++------- .../i915/gem/selftests/i915_gem_execbuffer.c | 62 +-- drivers/gpu/drm/i915/i915_gem.c | 6 + drivers/gpu/drm/i915/i915_gem.h | 1 + 4 files changed, 262 insertions(+), 171 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 23811fc4c8bb..5d08ce71f341 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -257,6 +257,8 @@ struct i915_execbuffer { /** list of vma that have execobj.relocation_count */ struct list_head relocs; + struct i915_gem_ww_ctx ww; + /** * Track the most recently used object for relocations, as we * frequently have to perform multiple relocations within the same @@ -275,14 +277,18 @@ struct i915_execbuffer { struct i915_request *rq; u32 *rq_cmd; unsigned int rq_size; + struct intel_gt_buffer_pool_node *pool; } reloc_cache; + struct intel_gt_buffer_pool_node *reloc_pool; /** relocation pool for -EDEADLK handling */ + u64 invalid_flags; /** Set of execobj.flags that are invalid */ u32 context_flags; /** Set of execobj.flags to insert from the ctx */ u32 batch_start_offset; /** Location within object of batch */ u32 batch_len; /** Length of batch within object */ u32 batch_flags; /** Flags composed for emit_bb_start() */ + struct intel_gt_buffer_pool_node *batch_pool; /** pool node for batch buffer */ /** * Indicate either the size of the hastable used to resolve @@ -452,23 +458,16 @@ eb_pin_vma(struct i915_execbuffer *eb, return !eb_vma_misplaced(entry, vma, ev->flags); } -static inline void __eb_unreserve_vma(struct i915_vma *vma, unsigned int flags) -{ - GEM_BUG_ON(!(flags & __EXEC_OBJECT_HAS_PIN)); - - if (unlikely(flags & __EXEC_OBJECT_HAS_FENCE)) - __i915_vma_unpin_fence(vma); - - __i915_vma_unpin(vma); -} - static inline void eb_unreserve_vma(struct eb_vma *ev) { if (!(ev->flags & __EXEC_OBJECT_HAS_PIN)) return; - __eb_unreserve_vma(ev->vma, ev->flags); + if (unlikely(ev->flags & __EXEC_OBJECT_HAS_FENCE)) + __i915_vma_unpin_fence(ev->vma); + + __i915_vma_unpin(ev->vma); ev->flags &= ~__EXEC_OBJECT_RESERVED; } @@ -563,16 +562,6 @@ eb_add_vma(struct i915_execbuffer *eb, eb->batch = ev; } - - if (eb_pin_vma(eb, entry, ev)) { - if (entry->offset != vma->node.start) { - entry->offset = vma->node.start | UPDATE; - eb->args->flags |= __EXEC_HAS_RELOC; - } - } else { - eb_unreserve_vma(ev); - list_add_tail(&ev->bind_link, &eb->unbound); - } } static inline int use_cpu_reloc(const struct reloc_cache *cache, @@ -657,10 +646,6 @@ static int eb_reserve(struct i915_execbuffer *eb) * This avoid unnecessary unbinding of later objects in order to make * room for the earlier objects *unless* we need to defragment. */ - - if (mutex_lock_interruptible(&eb->i915->drm.struct_mutex)) - return -EINTR; - pass = 0; do { list_for_each_entry(ev, &eb->unbound, bind_link) { @@ -669,7 +654,7 @@ static int eb_reserve(struct i915_execbuffer *eb) break; } if (err != -ENOSPC) - break; + return err; /* Resort *all* the objects into priority order */ INIT_LIST_HEAD(&eb->unbound); @@ -709,20 +694,15 @@ static int eb_reserve(struct i915_execbuffer *eb) err = i915_gem_evict_vm(eb->context->vm); mutex_unlock(&eb->context->vm->mutex); if (err) - goto unlock; + return err; break; default: - err = -ENOSPC; - goto unlock; + return -ENOSPC; } pin_flags = PIN_USER; } while (1); - -unlock: - mutex_unlock(&eb->i915->drm.struct_mutex); - return err; } static unsigned int eb_batch_index(const struct i915_execbuffer *eb) @@ -851,7 +831,6 @@ static int eb_lookup_vmas(struct i915_execbuffer *eb) int err = 0; INIT_LIST_HEAD(&eb->relocs); - INIT_LIST_HEAD(&eb->unbound); for (i = 0; i < eb->buffer_count; i++) { struct i915_vma *vma; @@ -894,6 +873,48 @@ static int eb_lookup_vmas(struct i915_execbuffer *eb) return err; } +static int eb_validate_vmas(struct i915_execbuffer *eb) +{ + unsigned int i; + int err; + + INIT_LIST_HEAD(&eb->unbound); + + for (i = 0; i < eb->buffer_count; i++) { + struct drm_i915_gem_exec_object2 *entry = &eb->exec[i]; + struct eb_vma *ev = &eb->vma[i]; + struct i915_vma *vma = ev->vma; + + err = i915_gem_object_lock(vma->obj, &eb->ww); + if (err) + return err; + + if (eb_pin_vma(eb, entry, ev)) { + if (entry->offset != vma->node.start) { + entry->offset = vma->node.start | UPDATE; + eb->args->flags |= __EXEC_HAS_RELOC; + } + } else { + eb_unreserve_vma(ev); + + list_add_tail(&ev->bind_link, &eb->unbound); + if (drm_mm_node_allocated(&vma->node)) { + err = i915_vma_unbind(vma); + if (err) + return err; + } + } + + GEM_BUG_ON(drm_mm_node_allocated(&vma->node) && + eb_vma_misplaced(&eb->exec[i], vma, ev->flags)); + } + + if (!list_empty(&eb->unbound)) + return eb_reserve(eb); + + return 0; +} + static struct eb_vma * eb_get_vma(const struct i915_execbuffer *eb, unsigned long handle) { @@ -914,7 +935,7 @@ eb_get_vma(const struct i915_execbuffer *eb, unsigned long handle) } } -static void eb_release_vmas(const struct i915_execbuffer *eb) +static void eb_release_vmas(const struct i915_execbuffer *eb, bool final) { const unsigned int count = eb->buffer_count; unsigned int i; @@ -926,12 +947,10 @@ static void eb_release_vmas(const struct i915_execbuffer *eb) if (!vma) break; - eb->vma[i].vma = NULL; - - if (ev->flags & __EXEC_OBJECT_HAS_PIN) - __eb_unreserve_vma(vma, ev->flags); + eb_unreserve_vma(ev); - i915_vma_put(vma); + if (final) + i915_vma_put(vma); } } @@ -950,6 +969,14 @@ relocation_target(const struct drm_i915_gem_relocation_entry *reloc, return gen8_canonical_addr((int)reloc->delta + target->node.start); } +static void reloc_cache_clear(struct reloc_cache *cache) +{ + cache->rq = NULL; + cache->rq_cmd = NULL; + cache->pool = NULL; + cache->rq_size = 0; +} + static void reloc_cache_init(struct reloc_cache *cache, struct drm_i915_private *i915) { @@ -962,8 +989,7 @@ static void reloc_cache_init(struct reloc_cache *cache, cache->has_fence = cache->gen < 4; cache->needs_unfenced = INTEL_INFO(i915)->unfenced_needs_alignment; cache->node.flags = 0; - cache->rq = NULL; - cache->rq_size = 0; + reloc_cache_clear(cache); } static inline void *unmask_page(unsigned long p) @@ -985,7 +1011,23 @@ static inline struct i915_ggtt *cache_to_ggtt(struct reloc_cache *cache) return &i915->ggtt; } -static void reloc_gpu_flush(struct reloc_cache *cache) +static void reloc_cache_put_pool(struct i915_execbuffer *eb, struct reloc_cache *cache) +{ + if (!cache->pool) + return; + + /* + * This is a bit nasty, normally we keep objects locked until the end + * of execbuffer, but we already submit this, and have to unlock before + * dropping the reference. Fortunately we can only hold 1 pool node at + * a time, so this should be harmless. + */ + i915_gem_ww_unlock_single(cache->pool->obj); + intel_gt_buffer_pool_put(cache->pool); + cache->pool = NULL; +} + +static void reloc_gpu_flush(struct i915_execbuffer *eb, struct reloc_cache *cache) { struct drm_i915_gem_object *obj = cache->rq->batch->obj; @@ -998,15 +1040,18 @@ static void reloc_gpu_flush(struct reloc_cache *cache) intel_gt_chipset_flush(cache->rq->engine->gt); i915_request_add(cache->rq); - cache->rq = NULL; + reloc_cache_put_pool(eb, cache); + reloc_cache_clear(cache); + + eb->reloc_pool = NULL; } -static void reloc_cache_reset(struct reloc_cache *cache) +static void reloc_cache_reset(struct reloc_cache *cache, struct i915_execbuffer *eb) { void *vaddr; if (cache->rq) - reloc_gpu_flush(cache); + reloc_gpu_flush(eb, cache); if (!cache->vaddr) return; @@ -1020,7 +1065,6 @@ static void reloc_cache_reset(struct reloc_cache *cache) kunmap_atomic(vaddr); i915_gem_object_finish_access(obj); - i915_gem_object_unlock(obj); } else { struct i915_ggtt *ggtt = cache_to_ggtt(cache); @@ -1056,15 +1100,9 @@ static void *reloc_kmap(struct drm_i915_gem_object *obj, unsigned int flushes; int err; - err = i915_gem_object_lock_interruptible(obj, NULL); - if (err) - return ERR_PTR(err); - err = i915_gem_object_prepare_write(obj, &flushes); - if (err) { - i915_gem_object_unlock(obj); + if (err) return ERR_PTR(err); - } BUILD_BUG_ON(KMAP & CLFLUSH_FLAGS); BUILD_BUG_ON((KMAP | CLFLUSH_FLAGS) & PAGE_MASK); @@ -1107,9 +1145,7 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj, if (use_cpu_reloc(cache, obj)) return NULL; - i915_gem_object_lock(obj, NULL); err = i915_gem_object_set_to_gtt_domain(obj, true); - i915_gem_object_unlock(obj); if (err) return ERR_PTR(err); @@ -1198,7 +1234,7 @@ static int reloc_move_to_gpu(struct i915_request *rq, struct i915_vma *vma) struct drm_i915_gem_object *obj = vma->obj; int err; - i915_vma_lock(vma); + assert_vma_held(vma); if (obj->cache_dirty & ~obj->cache_coherent) i915_gem_clflush_object(obj, 0); @@ -1208,8 +1244,6 @@ static int reloc_move_to_gpu(struct i915_request *rq, struct i915_vma *vma) if (err == 0) err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); - i915_vma_unlock(vma); - return err; } @@ -1219,15 +1253,22 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, unsigned int len) { struct reloc_cache *cache = &eb->reloc_cache; - struct intel_gt_buffer_pool_node *pool; + struct intel_gt_buffer_pool_node *pool = eb->reloc_pool; struct i915_request *rq; struct i915_vma *batch; u32 *cmd; int err; - pool = intel_gt_get_buffer_pool(engine->gt, PAGE_SIZE); - if (IS_ERR(pool)) - return PTR_ERR(pool); + if (!pool) { + pool = intel_gt_get_buffer_pool(engine->gt, PAGE_SIZE); + if (IS_ERR(pool)) + return PTR_ERR(pool); + } + eb->reloc_pool = NULL; + + err = i915_gem_object_lock(pool->obj, &eb->ww); + if (err) + goto err_pool; cmd = i915_gem_object_pin_map(pool->obj, cache->has_llc ? @@ -1235,7 +1276,7 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, I915_MAP_FORCE_WC); if (IS_ERR(cmd)) { err = PTR_ERR(cmd); - goto out_pool; + goto err_pool; } batch = i915_vma_instance(pool->obj, vma->vm, NULL); @@ -1284,11 +1325,10 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, if (err) goto skip_request; - i915_vma_lock(batch); + assert_vma_held(batch); err = i915_request_await_object(rq, batch->obj, false); if (err == 0) err = i915_vma_move_to_active(batch, rq, 0); - i915_vma_unlock(batch); if (err) goto skip_request; @@ -1298,9 +1338,10 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, cache->rq = rq; cache->rq_cmd = cmd; cache->rq_size = 0; + cache->pool = pool; /* Return with batch mapping (cmd) still pinned */ - goto out_pool; + return 0; skip_request: i915_request_set_error_once(rq, err); @@ -1310,8 +1351,8 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, i915_vma_unpin(batch); err_unmap: i915_gem_object_unpin_map(pool->obj); -out_pool: - intel_gt_buffer_pool_put(pool); +err_pool: + eb->reloc_pool = pool; return err; } @@ -1328,7 +1369,7 @@ static u32 *reloc_gpu(struct i915_execbuffer *eb, u32 *cmd; if (cache->rq_size > PAGE_SIZE/sizeof(u32) - (len + 1)) - reloc_gpu_flush(cache); + reloc_gpu_flush(eb, cache); if (unlikely(!cache->rq)) { int err; @@ -1376,7 +1417,7 @@ static unsigned long vma_phys_addr(struct i915_vma *vma, u32 offset) return addr + offset_in_page(offset); } -static bool __reloc_entry_gpu(struct i915_execbuffer *eb, +static int __reloc_entry_gpu(struct i915_execbuffer *eb, struct i915_vma *vma, u64 offset, u64 target_addr) @@ -1394,7 +1435,9 @@ static bool __reloc_entry_gpu(struct i915_execbuffer *eb, len = 3; batch = reloc_gpu(eb, vma, len); - if (IS_ERR(batch)) + if (batch == ERR_PTR(-EDEADLK)) + return (s64)-EDEADLK; + else if (IS_ERR(batch)) return false; addr = gen8_canonical_addr(vma->node.start + offset); @@ -1447,7 +1490,7 @@ static bool __reloc_entry_gpu(struct i915_execbuffer *eb, return true; } -static bool reloc_entry_gpu(struct i915_execbuffer *eb, +static int reloc_entry_gpu(struct i915_execbuffer *eb, struct i915_vma *vma, u64 offset, u64 target_addr) @@ -1469,8 +1512,12 @@ relocate_entry(struct i915_vma *vma, { u64 target_addr = relocation_target(reloc, target); u64 offset = reloc->offset; + int reloc_gpu = reloc_entry_gpu(eb, vma, offset, target_addr); - if (!reloc_entry_gpu(eb, vma, offset, target_addr)) { + if (reloc_gpu < 0) + return reloc_gpu; + + if (!reloc_gpu) { bool wide = eb->reloc_cache.use_64bit_reloc; void *vaddr; @@ -1673,7 +1720,7 @@ static int eb_relocate_vma(struct i915_execbuffer *eb, struct eb_vma *ev) urelocs += ARRAY_SIZE(stack); } while (remain); out: - reloc_cache_reset(&eb->reloc_cache); + reloc_cache_reset(&eb->reloc_cache, eb); return remain; } @@ -1696,7 +1743,7 @@ eb_relocate_vma_slow(struct i915_execbuffer *eb, struct eb_vma *ev) } err = 0; err: - reloc_cache_reset(&eb->reloc_cache); + reloc_cache_reset(&eb->reloc_cache, eb); return err; } @@ -1836,6 +1883,10 @@ static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) goto out; } + /* We may process another execbuffer during the unlock... */ + eb_release_vmas(eb, false); + i915_gem_ww_ctx_fini(&eb->ww); + /* * We take 3 passes through the slowpatch. * @@ -1861,12 +1912,17 @@ static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) flush_workqueue(eb->i915->mm.userptr_wq); + i915_gem_ww_ctx_init(&eb->ww, true); if (err) goto out; - err = mutex_lock_interruptible(&eb->i915->drm.struct_mutex); + /* reacquire the objects */ +repeat_validate: + err = eb_validate_vmas(eb); if (err) - goto out; + goto err; + + GEM_BUG_ON(!eb->batch); list_for_each_entry(ev, &eb->relocs, reloc_link) { if (!have_copy) { @@ -1882,8 +1938,9 @@ static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) } } - reloc_gpu_flush(&eb->reloc_cache); - mutex_unlock(&eb->i915->drm.struct_mutex); + if (err == -EDEADLK) + goto err; + if (err && !have_copy) goto repeat; @@ -1903,6 +1960,13 @@ static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) */ err: + if (err == -EDEADLK) { + eb_release_vmas(eb, false); + err = i915_gem_ww_ctx_backoff(&eb->ww); + if (!err) + goto repeat_validate; + } + if (err == -EAGAIN) goto repeat; @@ -1931,15 +1995,12 @@ static int eb_relocate_parse(struct i915_execbuffer *eb) { int err; - err = eb_lookup_vmas(eb); - if (err) - return err; - - if (!list_empty(&eb->unbound)) { - err = eb_reserve(eb); - if (err) - return err; - } +retry: + err = eb_validate_vmas(eb); + if (err == -EAGAIN) + goto slow; + else if (err) + goto err; /* The objects are in their final locations, apply the relocations. */ if (eb->args->flags & __EXEC_HAS_RELOC) { @@ -1951,45 +2012,46 @@ static int eb_relocate_parse(struct i915_execbuffer *eb) break; } - if (err) - return eb_relocate_parse_slow(eb); + if (err == -EDEADLK) + goto err; + else if (err) + goto slow; + } + + if (!err) + err = eb_parse(eb); + +err: + if (err == -EDEADLK) { + eb_release_vmas(eb, false); + err = i915_gem_ww_ctx_backoff(&eb->ww); + if (!err) + goto retry; } - return eb_parse(eb); + return err; + +slow: + err = eb_relocate_parse_slow(eb); + if (err) + /* + * If the user expects the execobject.offset and + * reloc.presumed_offset to be an exact match, + * as for using NO_RELOC, then we cannot update + * the execobject.offset until we have completed + * relocation. + */ + eb->args->flags &= ~__EXEC_HAS_RELOC; + + return err; } static int eb_move_to_gpu(struct i915_execbuffer *eb) { const unsigned int count = eb->buffer_count; - struct ww_acquire_ctx acquire; - unsigned int i; + unsigned int i = count; int err = 0; - ww_acquire_init(&acquire, &reservation_ww_class); - - for (i = 0; i < count; i++) { - struct eb_vma *ev = &eb->vma[i]; - struct i915_vma *vma = ev->vma; - - err = ww_mutex_lock_interruptible(&vma->resv->lock, &acquire); - if (err == -EDEADLK) { - GEM_BUG_ON(i == 0); - do { - int j = i - 1; - - ww_mutex_unlock(&eb->vma[j].vma->resv->lock); - - swap(eb->vma[i], eb->vma[j]); - } while (--i); - - err = ww_mutex_lock_slow_interruptible(&vma->resv->lock, - &acquire); - } - if (err) - break; - } - ww_acquire_done(&acquire); - while (i--) { struct eb_vma *ev = &eb->vma[i]; struct i915_vma *vma = ev->vma; @@ -2033,10 +2095,7 @@ static int eb_move_to_gpu(struct i915_execbuffer *eb) if (err == 0) err = i915_vma_move_to_active(vma, eb->request, flags); - - i915_vma_unlock(vma); } - ww_acquire_fini(&acquire); if (unlikely(err)) goto err_skip; @@ -2228,36 +2287,26 @@ static int eb_parse_pipeline(struct i915_execbuffer *eb, if (err) goto err_commit; - err = dma_resv_lock_interruptible(pw->batch->resv, NULL); - if (err) - goto err_commit; - err = dma_resv_reserve_shared(pw->batch->resv, 1); if (err) - goto err_commit_unlock; + goto err_commit; /* Wait for all writes (and relocs) into the batch to complete */ err = i915_sw_fence_await_reservation(&pw->base.chain, pw->batch->resv, NULL, false, 0, I915_FENCE_GFP); if (err < 0) - goto err_commit_unlock; + goto err_commit; /* Keep the batch alive and unwritten as we parse */ dma_resv_add_shared_fence(pw->batch->resv, &pw->base.dma); - dma_resv_unlock(pw->batch->resv); - /* Force execution to wait for completion of the parser */ - dma_resv_lock(shadow->resv, NULL); dma_resv_add_excl_fence(shadow->resv, &pw->base.dma); - dma_resv_unlock(shadow->resv); dma_fence_work_commit_imm(&pw->base); return 0; -err_commit_unlock: - dma_resv_unlock(pw->batch->resv); err_commit: i915_sw_fence_set_error_once(&pw->base.chain, err); dma_fence_work_commit_imm(&pw->base); @@ -2275,7 +2324,7 @@ static int eb_parse_pipeline(struct i915_execbuffer *eb, static int eb_parse(struct i915_execbuffer *eb) { struct drm_i915_private *i915 = eb->i915; - struct intel_gt_buffer_pool_node *pool; + struct intel_gt_buffer_pool_node *pool = eb->batch_pool; struct i915_vma *shadow, *trampoline; unsigned int len; int err; @@ -2298,9 +2347,16 @@ static int eb_parse(struct i915_execbuffer *eb) len += I915_CMD_PARSER_TRAMPOLINE_SIZE; } - pool = intel_gt_get_buffer_pool(eb->engine->gt, len); - if (IS_ERR(pool)) - return PTR_ERR(pool); + if (!pool) { + pool = intel_gt_get_buffer_pool(eb->engine->gt, len); + if (IS_ERR(pool)) + return PTR_ERR(pool); + eb->batch_pool = pool; + } + + err = i915_gem_object_lock(pool->obj, &eb->ww); + if (err) + goto err; shadow = shadow_batch_pin(pool->obj, eb->context->vm, PIN_USER); if (IS_ERR(shadow)) { @@ -2346,7 +2402,6 @@ static int eb_parse(struct i915_execbuffer *eb) err_shadow: i915_vma_unpin(shadow); err: - intel_gt_buffer_pool_put(pool); return err; } @@ -3001,6 +3056,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, eb.exec = exec; eb.vma = (struct eb_vma *)(exec + args->buffer_count + 1); eb.vma[0].vma = NULL; + eb.reloc_pool = eb.batch_pool = NULL; eb.invalid_flags = __EXEC_OBJECT_UNKNOWN_FLAGS; reloc_cache_init(&eb.reloc_cache, eb.i915); @@ -3073,6 +3129,14 @@ i915_gem_do_execbuffer(struct drm_device *dev, if (unlikely(err)) goto err_context; + err = eb_lookup_vmas(&eb); + if (err) { + eb_release_vmas(&eb, true); + goto err_engine; + } + + i915_gem_ww_ctx_init(&eb.ww, true); + err = eb_relocate_parse(&eb); if (err) { /* @@ -3086,6 +3150,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, goto err_vma; } + ww_acquire_done(&eb.ww.ctx); + /* * snb/ivb/vlv conflate the "batch in ppgtt" bit with the "non-secure * batch" bit. Hence we need to pin secure batches into the global gtt. @@ -3106,7 +3172,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, vma = i915_gem_object_ggtt_pin(eb.batch->vma->obj, NULL, 0, 0, 0); if (IS_ERR(vma)) { err = PTR_ERR(vma); - goto err_parse; + goto err_vma; } batch = vma; @@ -3158,8 +3224,8 @@ i915_gem_do_execbuffer(struct drm_device *dev, * to explicitly hold another reference here. */ eb.request->batch = batch; - if (batch->private) - intel_gt_buffer_pool_mark_active(batch->private, eb.request); + if (eb.batch_pool) + intel_gt_buffer_pool_mark_active(eb.batch_pool, eb.request); trace_i915_request_queue(eb.request, eb.batch_flags); err = eb_submit(&eb, batch); @@ -3185,14 +3251,18 @@ i915_gem_do_execbuffer(struct drm_device *dev, err_batch_unpin: if (eb.batch_flags & I915_DISPATCH_SECURE) i915_vma_unpin(batch); -err_parse: - if (batch->private) - intel_gt_buffer_pool_put(batch->private); err_vma: - if (eb.exec) - eb_release_vmas(&eb); + eb_release_vmas(&eb, true); if (eb.trampoline) i915_vma_unpin(eb.trampoline); + WARN_ON(err == -EDEADLK); + i915_gem_ww_ctx_fini(&eb.ww); + + if (eb.batch_pool) + intel_gt_buffer_pool_put(eb.batch_pool); + if (eb.reloc_pool) + intel_gt_buffer_pool_put(eb.reloc_pool); +err_engine: eb_unpin_engine(&eb); err_context: i915_gem_context_put(eb.gem_context); diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c index 580884cffec3..bc08c02b5767 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c @@ -32,25 +32,23 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, if (IS_ERR(vma)) return PTR_ERR(vma); + err = i915_gem_object_lock(obj, &eb->ww); + if (err) + return err; + err = i915_vma_pin(vma, 0, 0, PIN_USER | PIN_HIGH); if (err) return err; /* 8-Byte aligned */ - if (!__reloc_entry_gpu(eb, vma, - offsets[0] * sizeof(u32), - 0)) { - err = -EIO; - goto unpin_vma; - } + err = __reloc_entry_gpu(eb, vma, offsets[0] * sizeof(u32), 0); + if (err <= 0) + goto reloc_err; /* !8-Byte aligned */ - if (!__reloc_entry_gpu(eb, vma, - offsets[1] * sizeof(u32), - 1)) { - err = -EIO; - goto unpin_vma; - } + err = __reloc_entry_gpu(eb, vma, offsets[1] * sizeof(u32), 1); + if (err <= 0) + goto reloc_err; /* Skip to the end of the cmd page */ i = PAGE_SIZE / sizeof(u32) - 1; @@ -60,16 +58,13 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, eb->reloc_cache.rq_size += i; /* Force next batch */ - if (!__reloc_entry_gpu(eb, vma, - offsets[2] * sizeof(u32), - 2)) { - err = -EIO; - goto unpin_vma; - } + err = __reloc_entry_gpu(eb, vma, offsets[2] * sizeof(u32), 2); + if (err <= 0) + goto reloc_err; GEM_BUG_ON(!eb->reloc_cache.rq); rq = i915_request_get(eb->reloc_cache.rq); - reloc_gpu_flush(&eb->reloc_cache); + reloc_gpu_flush(eb, &eb->reloc_cache); GEM_BUG_ON(eb->reloc_cache.rq); err = i915_gem_object_wait(obj, I915_WAIT_INTERRUPTIBLE, HZ / 2); @@ -101,6 +96,11 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, unpin_vma: i915_vma_unpin(vma); return err; + +reloc_err: + if (!err) + err = -EIO; + goto unpin_vma; } static int igt_gpu_reloc(void *arg) @@ -122,6 +122,8 @@ static int igt_gpu_reloc(void *arg) goto err_scratch; } + intel_gt_pm_get(&eb.i915->gt); + for_each_uabi_engine(eb.engine, eb.i915) { reloc_cache_init(&eb.reloc_cache, eb.i915); memset(map, POISON_INUSE, 4096); @@ -132,15 +134,26 @@ static int igt_gpu_reloc(void *arg) err = PTR_ERR(eb.context); goto err_pm; } + eb.reloc_pool = NULL; + i915_gem_ww_ctx_init(&eb.ww, false); +retry: err = intel_context_pin(eb.context); - if (err) - goto err_put; + if (!err) { + err = __igt_gpu_reloc(&eb, scratch); + + intel_context_unpin(eb.context); + } + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&eb.ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&eb.ww); - err = __igt_gpu_reloc(&eb, scratch); + if (eb.reloc_pool) + intel_gt_buffer_pool_put(eb.reloc_pool); - intel_context_unpin(eb.context); -err_put: intel_context_put(eb.context); err_pm: intel_engine_pm_put(eb.engine); @@ -151,6 +164,7 @@ static int igt_gpu_reloc(void *arg) if (igt_flush_test(eb.i915)) err = -EIO; + intel_gt_pm_put(&eb.i915->gt); err_scratch: i915_gem_object_put(scratch); return err; diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 55158e487a7f..f6615d942b60 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -1360,6 +1360,12 @@ static void i915_gem_ww_ctx_unlock_all(struct i915_gem_ww_ctx *ww) } } +void i915_gem_ww_unlock_single(struct drm_i915_gem_object *obj) +{ + list_del(&obj->obj_link); + i915_gem_object_unlock(obj); +} + void i915_gem_ww_ctx_fini(struct i915_gem_ww_ctx *ww) { i915_gem_ww_ctx_unlock_all(ww); diff --git a/drivers/gpu/drm/i915/i915_gem.h b/drivers/gpu/drm/i915/i915_gem.h index 427ebd02fceb..a4cad3f154ca 100644 --- a/drivers/gpu/drm/i915/i915_gem.h +++ b/drivers/gpu/drm/i915/i915_gem.h @@ -126,5 +126,6 @@ struct i915_gem_ww_ctx { void i915_gem_ww_ctx_init(struct i915_gem_ww_ctx *ctx, bool intr); void i915_gem_ww_ctx_fini(struct i915_gem_ww_ctx *ctx); int __must_check i915_gem_ww_ctx_backoff(struct i915_gem_ww_ctx *ctx); +void i915_gem_ww_unlock_single(struct drm_i915_gem_object *obj); #endif /* __I915_GEM_H__ */ From patchwork Mon Aug 10 10:30:48 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707267 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id E118E1392 for ; Mon, 10 Aug 2020 10:33:41 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id C9719206E9 for ; Mon, 10 Aug 2020 10:33:41 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org C9719206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 952DF6E3D0; Mon, 10 Aug 2020 10:33:30 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 1A1C26E0E3 for ; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:48 +0200 Message-Id: <20200810103103.303818-10-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 09/24] drm/i915: make lockdep slightly happier about execbuf. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" As soon as we install fences, we should stop allocating memory in order to prevent any potential deadlocks. This is required later on, when we start adding support for dma-fence annotations, and also required for userptr. Signed-off-by: Maarten Lankhorst --- drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c | 18 ++++++++++++------ drivers/gpu/drm/i915/i915_vma.c | 8 +++++--- drivers/gpu/drm/i915/i915_vma.h | 3 +++ 3 files changed, 20 insertions(+), 9 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 5d08ce71f341..12397fbc0971 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -48,11 +48,12 @@ enum { #define DBG_FORCE_RELOC 0 /* choose one of the above! */ }; -#define __EXEC_OBJECT_HAS_PIN BIT(31) -#define __EXEC_OBJECT_HAS_FENCE BIT(30) -#define __EXEC_OBJECT_NEEDS_MAP BIT(29) -#define __EXEC_OBJECT_NEEDS_BIAS BIT(28) -#define __EXEC_OBJECT_INTERNAL_FLAGS (~0u << 28) /* all of the above */ +/* __EXEC_OBJECT_NO_RESERVE is BIT(31), defined in i915_vma.h */ +#define __EXEC_OBJECT_HAS_PIN BIT(30) +#define __EXEC_OBJECT_HAS_FENCE BIT(29) +#define __EXEC_OBJECT_NEEDS_MAP BIT(28) +#define __EXEC_OBJECT_NEEDS_BIAS BIT(27) +#define __EXEC_OBJECT_INTERNAL_FLAGS (~0u << 27) /* all of the above + */ #define __EXEC_OBJECT_RESERVED (__EXEC_OBJECT_HAS_PIN | __EXEC_OBJECT_HAS_FENCE) #define __EXEC_HAS_RELOC BIT(31) @@ -2094,7 +2095,8 @@ static int eb_move_to_gpu(struct i915_execbuffer *eb) } if (err == 0) - err = i915_vma_move_to_active(vma, eb->request, flags); + err = i915_vma_move_to_active(vma, eb->request, + flags | __EXEC_OBJECT_NO_RESERVE); } if (unlikely(err)) @@ -2291,6 +2293,10 @@ static int eb_parse_pipeline(struct i915_execbuffer *eb, if (err) goto err_commit; + err = dma_resv_reserve_shared(shadow->resv, 1); + if (err) + goto err_commit; + /* Wait for all writes (and relocs) into the batch to complete */ err = i915_sw_fence_await_reservation(&pw->base.chain, pw->batch->resv, NULL, false, diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c index c6bf04ca2032..8066f167d6b9 100644 --- a/drivers/gpu/drm/i915/i915_vma.c +++ b/drivers/gpu/drm/i915/i915_vma.c @@ -1227,9 +1227,11 @@ int i915_vma_move_to_active(struct i915_vma *vma, obj->write_domain = I915_GEM_DOMAIN_RENDER; obj->read_domains = 0; } else { - err = dma_resv_reserve_shared(vma->resv, 1); - if (unlikely(err)) - return err; + if (!(flags & __EXEC_OBJECT_NO_RESERVE)) { + err = dma_resv_reserve_shared(vma->resv, 1); + if (unlikely(err)) + return err; + } dma_resv_add_shared_fence(vma->resv, &rq->fence); obj->write_domain = 0; diff --git a/drivers/gpu/drm/i915/i915_vma.h b/drivers/gpu/drm/i915/i915_vma.h index d0d01f909548..4b325a670a04 100644 --- a/drivers/gpu/drm/i915/i915_vma.h +++ b/drivers/gpu/drm/i915/i915_vma.h @@ -52,6 +52,9 @@ static inline bool i915_vma_is_active(const struct i915_vma *vma) return !i915_active_is_idle(&vma->active); } +/* do not reserve memory to prevent deadlocks */ +#define __EXEC_OBJECT_NO_RESERVE BIT(31) + int __must_check __i915_vma_move_to_active(struct i915_vma *vma, struct i915_request *rq); int __must_check i915_vma_move_to_active(struct i915_vma *vma, From patchwork Mon Aug 10 10:30:49 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707281 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id E0BBF13A4 for ; Mon, 10 Aug 2020 10:33:47 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id C90EA206E9 for ; Mon, 10 Aug 2020 10:33:47 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org C90EA206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id D44FB6E3F3; Mon, 10 Aug 2020 10:33:36 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 21F396E3D0 for ; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:49 +0200 Message-Id: <20200810103103.303818-11-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 10/24] drm/i915: Use ww locking in intel_renderstate. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" We want to start using ww locking in intel_context_pin, for this we need to lock multiple objects, and the single i915_gem_object_lock is not enough. Convert to using ww-waiting, and make sure we always pin intel_context_state, even if we don't have a renderstate object. Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- drivers/gpu/drm/i915/gt/intel_gt.c | 21 +++--- drivers/gpu/drm/i915/gt/intel_renderstate.c | 73 +++++++++++++++------ drivers/gpu/drm/i915/gt/intel_renderstate.h | 9 ++- 3 files changed, 71 insertions(+), 32 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c index e0755f1a904b..8694ddbdac4c 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt.c +++ b/drivers/gpu/drm/i915/gt/intel_gt.c @@ -406,21 +406,20 @@ static int __engines_record_defaults(struct intel_gt *gt) /* We must be able to switch to something! */ GEM_BUG_ON(!engine->kernel_context); - err = intel_renderstate_init(&so, engine); - if (err) - goto out; - ce = intel_context_create(engine); if (IS_ERR(ce)) { err = PTR_ERR(ce); goto out; } - rq = intel_context_create_request(ce); + err = intel_renderstate_init(&so, ce); + if (err) + goto err; + + rq = i915_request_create(ce); if (IS_ERR(rq)) { err = PTR_ERR(rq); - intel_context_put(ce); - goto out; + goto err_fini; } err = intel_engine_emit_ctx_wa(rq); @@ -434,9 +433,13 @@ static int __engines_record_defaults(struct intel_gt *gt) err_rq: requests[id] = i915_request_get(rq); i915_request_add(rq); - intel_renderstate_fini(&so); - if (err) +err_fini: + intel_renderstate_fini(&so, ce); +err: + if (err) { + intel_context_put(ce); goto out; + } } /* Flush the default context image to memory, and enable powersaving. */ diff --git a/drivers/gpu/drm/i915/gt/intel_renderstate.c b/drivers/gpu/drm/i915/gt/intel_renderstate.c index 1bfad589c63b..5164de04049d 100644 --- a/drivers/gpu/drm/i915/gt/intel_renderstate.c +++ b/drivers/gpu/drm/i915/gt/intel_renderstate.c @@ -27,6 +27,7 @@ #include "i915_drv.h" #include "intel_renderstate.h" +#include "gt/intel_context.h" #include "intel_ring.h" static const struct intel_renderstate_rodata * @@ -157,33 +158,47 @@ static int render_state_setup(struct intel_renderstate *so, #undef OUT_BATCH int intel_renderstate_init(struct intel_renderstate *so, - struct intel_engine_cs *engine) + struct intel_context *ce) { - struct drm_i915_gem_object *obj; + struct intel_engine_cs *engine = ce->engine; + struct drm_i915_gem_object *obj = NULL; int err; memset(so, 0, sizeof(*so)); so->rodata = render_state_get_rodata(engine); - if (!so->rodata) - return 0; + if (so->rodata) { + if (so->rodata->batch_items * 4 > PAGE_SIZE) + return -EINVAL; + + obj = i915_gem_object_create_internal(engine->i915, PAGE_SIZE); + if (IS_ERR(obj)) + return PTR_ERR(obj); + + so->vma = i915_vma_instance(obj, &engine->gt->ggtt->vm, NULL); + if (IS_ERR(so->vma)) { + err = PTR_ERR(so->vma); + goto err_obj; + } + } - if (so->rodata->batch_items * 4 > PAGE_SIZE) - return -EINVAL; + i915_gem_ww_ctx_init(&so->ww, true); +retry: + err = intel_context_pin(ce); + if (err) + goto err_fini; - obj = i915_gem_object_create_internal(engine->i915, PAGE_SIZE); - if (IS_ERR(obj)) - return PTR_ERR(obj); + /* return early if there's nothing to setup */ + if (!err && !so->rodata) + return 0; - so->vma = i915_vma_instance(obj, &engine->gt->ggtt->vm, NULL); - if (IS_ERR(so->vma)) { - err = PTR_ERR(so->vma); - goto err_obj; - } + err = i915_gem_object_lock(so->vma->obj, &so->ww); + if (err) + goto err_context; err = i915_vma_pin(so->vma, 0, 0, PIN_GLOBAL | PIN_HIGH); if (err) - goto err_obj; + goto err_context; err = render_state_setup(so, engine->i915); if (err) @@ -193,8 +208,18 @@ int intel_renderstate_init(struct intel_renderstate *so, err_unpin: i915_vma_unpin(so->vma); +err_context: + intel_context_unpin(ce); +err_fini: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&so->ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&so->ww); err_obj: - i915_gem_object_put(obj); + if (obj) + i915_gem_object_put(obj); so->vma = NULL; return err; } @@ -208,11 +233,9 @@ int intel_renderstate_emit(struct intel_renderstate *so, if (!so->vma) return 0; - i915_vma_lock(so->vma); err = i915_request_await_object(rq, so->vma->obj, false); if (err == 0) err = i915_vma_move_to_active(so->vma, rq, 0); - i915_vma_unlock(so->vma); if (err) return err; @@ -233,7 +256,17 @@ int intel_renderstate_emit(struct intel_renderstate *so, return 0; } -void intel_renderstate_fini(struct intel_renderstate *so) +void intel_renderstate_fini(struct intel_renderstate *so, + struct intel_context *ce) { - i915_vma_unpin_and_release(&so->vma, 0); + if (so->vma) { + i915_vma_unpin(so->vma); + i915_vma_close(so->vma); + } + + intel_context_unpin(ce); + i915_gem_ww_ctx_fini(&so->ww); + + if (so->vma) + i915_gem_object_put(so->vma->obj); } diff --git a/drivers/gpu/drm/i915/gt/intel_renderstate.h b/drivers/gpu/drm/i915/gt/intel_renderstate.h index 5700be69a05a..713aa1e86c80 100644 --- a/drivers/gpu/drm/i915/gt/intel_renderstate.h +++ b/drivers/gpu/drm/i915/gt/intel_renderstate.h @@ -25,9 +25,10 @@ #define _INTEL_RENDERSTATE_H_ #include +#include "i915_gem.h" struct i915_request; -struct intel_engine_cs; +struct intel_context; struct i915_vma; struct intel_renderstate_rodata { @@ -49,6 +50,7 @@ extern const struct intel_renderstate_rodata gen8_null_state; extern const struct intel_renderstate_rodata gen9_null_state; struct intel_renderstate { + struct i915_gem_ww_ctx ww; const struct intel_renderstate_rodata *rodata; struct i915_vma *vma; u32 batch_offset; @@ -58,9 +60,10 @@ struct intel_renderstate { }; int intel_renderstate_init(struct intel_renderstate *so, - struct intel_engine_cs *engine); + struct intel_context *ce); int intel_renderstate_emit(struct intel_renderstate *so, struct i915_request *rq); -void intel_renderstate_fini(struct intel_renderstate *so); +void intel_renderstate_fini(struct intel_renderstate *so, + struct intel_context *ce); #endif /* _INTEL_RENDERSTATE_H_ */ From patchwork Mon Aug 10 10:30:50 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707271 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 958E013A4 for ; Mon, 10 Aug 2020 10:33:43 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 7DDC6206E9 for ; Mon, 10 Aug 2020 10:33:43 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 7DDC6206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id DEA636E3EF; Mon, 10 Aug 2020 10:33:30 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 470156E3DB for ; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:50 +0200 Message-Id: <20200810103103.303818-12-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 11/24] drm/i915: Add ww context handling to context_barrier_task 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" This is required if we want to pass a ww context in intel_context_pin and gen6_ppgtt_pin(). Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- drivers/gpu/drm/i915/gem/i915_gem_context.c | 55 ++++++++++++++----- .../drm/i915/gem/selftests/i915_gem_context.c | 22 +++----- 2 files changed, 48 insertions(+), 29 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_context.c b/drivers/gpu/drm/i915/gem/i915_gem_context.c index db893f6c516b..34c8b0dd85e0 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_context.c @@ -1106,6 +1106,7 @@ I915_SELFTEST_DECLARE(static intel_engine_mask_t context_barrier_inject_fault); static int context_barrier_task(struct i915_gem_context *ctx, intel_engine_mask_t engines, bool (*skip)(struct intel_context *ce, void *data), + int (*pin)(struct intel_context *ce, struct i915_gem_ww_ctx *ww, void *data), int (*emit)(struct i915_request *rq, void *data), void (*task)(void *data), void *data) @@ -1113,6 +1114,7 @@ static int context_barrier_task(struct i915_gem_context *ctx, struct context_barrier_task *cb; struct i915_gem_engines_iter it; struct i915_gem_engines *e; + struct i915_gem_ww_ctx ww; struct intel_context *ce; int err = 0; @@ -1150,10 +1152,21 @@ static int context_barrier_task(struct i915_gem_context *ctx, if (skip && skip(ce, data)) continue; - rq = intel_context_create_request(ce); + i915_gem_ww_ctx_init(&ww, true); +retry: + err = intel_context_pin(ce); + if (err) + goto err; + + if (pin) + err = pin(ce, &ww, data); + if (err) + goto err_unpin; + + rq = i915_request_create(ce); if (IS_ERR(rq)) { err = PTR_ERR(rq); - break; + goto err_unpin; } err = 0; @@ -1163,6 +1176,16 @@ static int context_barrier_task(struct i915_gem_context *ctx, err = i915_active_add_request(&cb->base, rq); i915_request_add(rq); +err_unpin: + intel_context_unpin(ce); +err: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + if (err) break; } @@ -1218,6 +1241,17 @@ static void set_ppgtt_barrier(void *data) i915_vm_close(old); } +static int pin_ppgtt_update(struct intel_context *ce, struct i915_gem_ww_ctx *ww, void *data) +{ + struct i915_address_space *vm = ce->vm; + + if (!HAS_LOGICAL_RING_CONTEXTS(vm->i915)) + /* ppGTT is not part of the legacy context image */ + return gen6_ppgtt_pin(i915_vm_to_ppgtt(vm)); + + return 0; +} + static int emit_ppgtt_update(struct i915_request *rq, void *data) { struct i915_address_space *vm = rq->context->vm; @@ -1274,20 +1308,10 @@ static int emit_ppgtt_update(struct i915_request *rq, void *data) static bool skip_ppgtt_update(struct intel_context *ce, void *data) { - if (!test_bit(CONTEXT_ALLOC_BIT, &ce->flags)) - return true; - if (HAS_LOGICAL_RING_CONTEXTS(ce->engine->i915)) - return false; - - if (!atomic_read(&ce->pin_count)) - return true; - - /* ppGTT is not part of the legacy context image */ - if (gen6_ppgtt_pin(i915_vm_to_ppgtt(ce->vm))) - return true; - - return false; + return !ce->state; + else + return !atomic_read(&ce->pin_count); } static int set_ppgtt(struct drm_i915_file_private *file_priv, @@ -1338,6 +1362,7 @@ static int set_ppgtt(struct drm_i915_file_private *file_priv, */ err = context_barrier_task(ctx, ALL_ENGINES, skip_ppgtt_update, + pin_ppgtt_update, emit_ppgtt_update, set_ppgtt_barrier, old); 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 9160a335c684..fa40006b453a 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c @@ -1920,8 +1920,8 @@ static int mock_context_barrier(void *arg) return -ENOMEM; counter = 0; - err = context_barrier_task(ctx, 0, - NULL, NULL, mock_barrier_task, &counter); + err = context_barrier_task(ctx, 0, NULL, NULL, NULL, + mock_barrier_task, &counter); if (err) { pr_err("Failed at line %d, err=%d\n", __LINE__, err); goto out; @@ -1933,11 +1933,8 @@ static int mock_context_barrier(void *arg) } counter = 0; - err = context_barrier_task(ctx, ALL_ENGINES, - skip_unused_engines, - NULL, - mock_barrier_task, - &counter); + err = context_barrier_task(ctx, ALL_ENGINES, skip_unused_engines, + NULL, NULL, mock_barrier_task, &counter); if (err) { pr_err("Failed at line %d, err=%d\n", __LINE__, err); goto out; @@ -1957,8 +1954,8 @@ static int mock_context_barrier(void *arg) counter = 0; context_barrier_inject_fault = BIT(RCS0); - err = context_barrier_task(ctx, ALL_ENGINES, - NULL, NULL, mock_barrier_task, &counter); + err = context_barrier_task(ctx, ALL_ENGINES, NULL, NULL, NULL, + mock_barrier_task, &counter); context_barrier_inject_fault = 0; if (err == -ENXIO) err = 0; @@ -1972,11 +1969,8 @@ static int mock_context_barrier(void *arg) goto out; counter = 0; - err = context_barrier_task(ctx, ALL_ENGINES, - skip_unused_engines, - NULL, - mock_barrier_task, - &counter); + err = context_barrier_task(ctx, ALL_ENGINES, skip_unused_engines, + NULL, NULL, mock_barrier_task, &counter); if (err) { pr_err("Failed at line %d, err=%d\n", __LINE__, err); goto out; From patchwork Mon Aug 10 10:30:51 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707283 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 7DC0713B6 for ; Mon, 10 Aug 2020 10:33:48 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 65E18206E9 for ; Mon, 10 Aug 2020 10:33:48 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 65E18206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 3A4856E3F7; Mon, 10 Aug 2020 10:33:37 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 53C446E3E5 for ; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:51 +0200 Message-Id: <20200810103103.303818-13-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 12/24] drm/i915: Nuke arguments to eb_pin_engine 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Those arguments are already set as eb.file and eb.args, so kill off the extra arguments. This will allow us to move eb_pin_engine() to after we reserved all BO's. Signed-off-by: Maarten Lankhorst Reviewed-by: Tvrtko Ursulin Reviewed-by: Thomas Hellström --- drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c | 17 +++++++---------- 1 file changed, 7 insertions(+), 10 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 12397fbc0971..c9280ff491dd 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -2616,11 +2616,10 @@ static void eb_unpin_engine(struct i915_execbuffer *eb) } static unsigned int -eb_select_legacy_ring(struct i915_execbuffer *eb, - struct drm_file *file, - struct drm_i915_gem_execbuffer2 *args) +eb_select_legacy_ring(struct i915_execbuffer *eb) { struct drm_i915_private *i915 = eb->i915; + struct drm_i915_gem_execbuffer2 *args = eb->args; unsigned int user_ring_id = args->flags & I915_EXEC_RING_MASK; if (user_ring_id != I915_EXEC_BSD && @@ -2635,7 +2634,7 @@ eb_select_legacy_ring(struct i915_execbuffer *eb, unsigned int bsd_idx = args->flags & I915_EXEC_BSD_MASK; if (bsd_idx == I915_EXEC_BSD_DEFAULT) { - bsd_idx = gen8_dispatch_bsd_engine(i915, file); + bsd_idx = gen8_dispatch_bsd_engine(i915, eb->file); } else if (bsd_idx >= I915_EXEC_BSD_RING1 && bsd_idx <= I915_EXEC_BSD_RING2) { bsd_idx >>= I915_EXEC_BSD_SHIFT; @@ -2660,18 +2659,16 @@ eb_select_legacy_ring(struct i915_execbuffer *eb, } static int -eb_pin_engine(struct i915_execbuffer *eb, - struct drm_file *file, - struct drm_i915_gem_execbuffer2 *args) +eb_pin_engine(struct i915_execbuffer *eb) { struct intel_context *ce; unsigned int idx; int err; if (i915_gem_context_user_engines(eb->gem_context)) - idx = args->flags & I915_EXEC_RING_MASK; + idx = eb->args->flags & I915_EXEC_RING_MASK; else - idx = eb_select_legacy_ring(eb, file, args); + idx = eb_select_legacy_ring(eb); ce = i915_gem_context_get_engine(eb->gem_context, idx); if (IS_ERR(ce)) @@ -3131,7 +3128,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, if (unlikely(err)) goto err_destroy; - err = eb_pin_engine(&eb, file, args); + err = eb_pin_engine(&eb); if (unlikely(err)) goto err_context; From patchwork Mon Aug 10 10:30:52 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707289 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 45D021392 for ; Mon, 10 Aug 2020 10:33:50 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 2E394206E9 for ; Mon, 10 Aug 2020 10:33:50 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 2E394206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 868206E405; Mon, 10 Aug 2020 10:33:37 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 64B18897E0 for ; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:52 +0200 Message-Id: <20200810103103.303818-14-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 13/24] drm/i915: Pin engine before pinning all objects, v5. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" We want to lock all gem objects, including the engine context objects, rework the throttling to ensure that we can do this. Now we only throttle once, but can take eb_pin_engine while acquiring objects. This means we will have to drop the lock to wait. If we don't have to throttle we can still take the fastpath, if not we will take the slowpath and wait for the throttle request while unlocked. The engine has to be pinned as first step, otherwise gpu relocations won't work. Changes since v1: - Only need to get a throttled request in the fastpath, no need for a global flag any more. - Always free the waited request correctly. Changes since v2: - Use intel_engine_pm_get()/put() to keeep engine pool alive during EDEADLK handling. Changes since v3: - Fix small rq leak. Changes since v4: - Use a single reloc_context, for intel_context_pin_ww(). Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 214 ++++++++++++------ .../i915/gem/selftests/i915_gem_execbuffer.c | 3 + 2 files changed, 152 insertions(+), 65 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index c9280ff491dd..604e26adea23 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -57,7 +57,8 @@ enum { #define __EXEC_OBJECT_RESERVED (__EXEC_OBJECT_HAS_PIN | __EXEC_OBJECT_HAS_FENCE) #define __EXEC_HAS_RELOC BIT(31) -#define __EXEC_INTERNAL_FLAGS (~0u << 31) +#define __EXEC_ENGINE_PINNED BIT(30) +#define __EXEC_INTERNAL_FLAGS (~0u << 30) #define UPDATE PIN_OFFSET_FIXED #define BATCH_OFFSET_BIAS (256*1024) @@ -282,6 +283,7 @@ struct i915_execbuffer { } reloc_cache; struct intel_gt_buffer_pool_node *reloc_pool; /** relocation pool for -EDEADLK handling */ + struct intel_context *reloc_context; u64 invalid_flags; /** Set of execobj.flags that are invalid */ u32 context_flags; /** Set of execobj.flags to insert from the ctx */ @@ -304,6 +306,9 @@ struct i915_execbuffer { }; static int eb_parse(struct i915_execbuffer *eb); +static struct i915_request *eb_pin_engine(struct i915_execbuffer *eb, + bool throttle); +static void eb_unpin_engine(struct i915_execbuffer *eb); static inline bool eb_use_cmdparser(const struct i915_execbuffer *eb) { @@ -936,7 +941,7 @@ eb_get_vma(const struct i915_execbuffer *eb, unsigned long handle) } } -static void eb_release_vmas(const struct i915_execbuffer *eb, bool final) +static void eb_release_vmas(struct i915_execbuffer *eb, bool final) { const unsigned int count = eb->buffer_count; unsigned int i; @@ -953,6 +958,8 @@ static void eb_release_vmas(const struct i915_execbuffer *eb, bool final) if (final) i915_vma_put(vma); } + + eb_unpin_engine(eb); } static void eb_destroy(const struct i915_execbuffer *eb) @@ -1293,19 +1300,26 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, if (engine == eb->context->engine) { rq = i915_request_create(eb->context); } else { - struct intel_context *ce; + struct intel_context *ce = eb->reloc_context; - ce = intel_context_create(engine); - if (IS_ERR(ce)) { - err = PTR_ERR(ce); - goto err_unpin; + if (!ce) { + ce = intel_context_create(engine); + if (IS_ERR(ce)) { + err = PTR_ERR(ce); + goto err_unpin; + } + + i915_vm_put(ce->vm); + ce->vm = i915_vm_get(eb->context->vm); + eb->reloc_context = ce; } - i915_vm_put(ce->vm); - ce->vm = i915_vm_get(eb->context->vm); + err = intel_context_pin(ce); + if (err) + goto err_unpin; - rq = intel_context_create_request(ce); - intel_context_put(ce); + rq = i915_request_create(ce); + intel_context_unpin(ce); } if (IS_ERR(rq)) { err = PTR_ERR(rq); @@ -1872,7 +1886,8 @@ static int eb_prefault_relocations(const struct i915_execbuffer *eb) return 0; } -static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) +static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb, + struct i915_request *rq) { bool have_copy = false; struct eb_vma *ev; @@ -1888,6 +1903,21 @@ static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) eb_release_vmas(eb, false); i915_gem_ww_ctx_fini(&eb->ww); + if (rq) { + /* nonblocking is always false */ + if (i915_request_wait(rq, I915_WAIT_INTERRUPTIBLE, + MAX_SCHEDULE_TIMEOUT) < 0) { + i915_request_put(rq); + rq = NULL; + + err = -EINTR; + goto err_relock; + } + + i915_request_put(rq); + rq = NULL; + } + /* * We take 3 passes through the slowpatch. * @@ -1911,14 +1941,25 @@ static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) err = 0; } - flush_workqueue(eb->i915->mm.userptr_wq); + if (!err) + flush_workqueue(eb->i915->mm.userptr_wq); +err_relock: i915_gem_ww_ctx_init(&eb->ww, true); if (err) goto out; /* reacquire the objects */ repeat_validate: + rq = eb_pin_engine(eb, false); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + goto err; + } + + /* We didn't throttle, should be NULL */ + GEM_WARN_ON(rq); + err = eb_validate_vmas(eb); if (err) goto err; @@ -1989,14 +2030,49 @@ static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb) } } + if (rq) + i915_request_put(rq); + return err; } static int eb_relocate_parse(struct i915_execbuffer *eb) { int err; + struct i915_request *rq = NULL; + bool throttle = true; retry: + rq = eb_pin_engine(eb, throttle); + if (IS_ERR(rq)) { + err = PTR_ERR(rq); + rq = NULL; + if (err != -EDEADLK) + return err; + + goto err; + } + + if (rq) { + bool nonblock = eb->file->filp->f_flags & O_NONBLOCK; + + /* Need to drop all locks now for throttling, take slowpath */ + err = i915_request_wait(rq, I915_WAIT_INTERRUPTIBLE, 0); + if (err == -ETIME) { + if (nonblock) { + err = -EWOULDBLOCK; + i915_request_put(rq); + goto err; + } + goto slow; + } + i915_request_put(rq); + rq = NULL; + } + + /* only throttle once, even if we didn't need to throttle */ + throttle = false; + err = eb_validate_vmas(eb); if (err == -EAGAIN) goto slow; @@ -2033,7 +2109,7 @@ static int eb_relocate_parse(struct i915_execbuffer *eb) return err; slow: - err = eb_relocate_parse_slow(eb); + err = eb_relocate_parse_slow(eb, rq); if (err) /* * If the user expects the execobject.offset and @@ -2492,7 +2568,7 @@ static const enum intel_engine_id user_ring_map[] = { [I915_EXEC_VEBOX] = VECS0 }; -static struct i915_request *eb_throttle(struct intel_context *ce) +static struct i915_request *eb_throttle(struct i915_execbuffer *eb, struct intel_context *ce) { struct intel_ring *ring = ce->ring; struct intel_timeline *tl = ce->timeline; @@ -2526,22 +2602,17 @@ static struct i915_request *eb_throttle(struct intel_context *ce) return i915_request_get(rq); } -static int __eb_pin_engine(struct i915_execbuffer *eb, struct intel_context *ce) +static struct i915_request *eb_pin_engine(struct i915_execbuffer *eb, bool throttle) { + struct intel_context *ce = eb->context; struct intel_timeline *tl; - struct i915_request *rq; + struct i915_request *rq = NULL; int err; - /* - * ABI: Before userspace accesses the GPU (e.g. execbuffer), report - * EIO if the GPU is already wedged. - */ - err = intel_gt_terminally_wedged(ce->engine->gt); - if (err) - return err; + GEM_BUG_ON(eb->args->flags & __EXEC_ENGINE_PINNED); if (unlikely(intel_context_is_banned(ce))) - return -EIO; + return ERR_PTR(-EIO); /* * Pinning the contexts may generate requests in order to acquire @@ -2550,7 +2621,7 @@ static int __eb_pin_engine(struct i915_execbuffer *eb, struct intel_context *ce) */ err = intel_context_pin(ce); if (err) - return err; + return ERR_PTR(err); /* * Take a local wakeref for preparing to dispatch the execbuf as @@ -2562,45 +2633,17 @@ static int __eb_pin_engine(struct i915_execbuffer *eb, struct intel_context *ce) */ tl = intel_context_timeline_lock(ce); if (IS_ERR(tl)) { - err = PTR_ERR(tl); - goto err_unpin; + intel_context_unpin(ce); + return ERR_CAST(tl); } intel_context_enter(ce); - rq = eb_throttle(ce); - + if (throttle) + rq = eb_throttle(eb, ce); intel_context_timeline_unlock(tl); - if (rq) { - bool nonblock = eb->file->filp->f_flags & O_NONBLOCK; - long timeout; - - timeout = MAX_SCHEDULE_TIMEOUT; - if (nonblock) - timeout = 0; - - timeout = i915_request_wait(rq, - I915_WAIT_INTERRUPTIBLE, - timeout); - i915_request_put(rq); - - if (timeout < 0) { - err = nonblock ? -EWOULDBLOCK : timeout; - goto err_exit; - } - } - - eb->engine = ce->engine; - eb->context = ce; - return 0; - -err_exit: - mutex_lock(&tl->mutex); - intel_context_exit(ce); - intel_context_timeline_unlock(tl); -err_unpin: - intel_context_unpin(ce); - return err; + eb->args->flags |= __EXEC_ENGINE_PINNED; + return rq; } static void eb_unpin_engine(struct i915_execbuffer *eb) @@ -2608,6 +2651,11 @@ static void eb_unpin_engine(struct i915_execbuffer *eb) struct intel_context *ce = eb->context; struct intel_timeline *tl = ce->timeline; + if (!(eb->args->flags & __EXEC_ENGINE_PINNED)) + return; + + eb->args->flags &= ~__EXEC_ENGINE_PINNED; + mutex_lock(&tl->mutex); intel_context_exit(ce); mutex_unlock(&tl->mutex); @@ -2659,7 +2707,7 @@ eb_select_legacy_ring(struct i915_execbuffer *eb) } static int -eb_pin_engine(struct i915_execbuffer *eb) +eb_select_engine(struct i915_execbuffer *eb) { struct intel_context *ce; unsigned int idx; @@ -2674,12 +2722,45 @@ eb_pin_engine(struct i915_execbuffer *eb) if (IS_ERR(ce)) return PTR_ERR(ce); - err = __eb_pin_engine(eb, ce); - intel_context_put(ce); + intel_gt_pm_get(ce->engine->gt); + if (!test_bit(CONTEXT_ALLOC_BIT, &ce->flags)) { + err = intel_context_alloc_state(ce); + if (err) + goto err; + } + + /* + * ABI: Before userspace accesses the GPU (e.g. execbuffer), report + * EIO if the GPU is already wedged. + */ + err = intel_gt_terminally_wedged(ce->engine->gt); + if (err) + goto err; + + eb->context = ce; + eb->engine = ce->engine; + + /* + * Make sure engine pool stays alive even if we call intel_context_put + * during ww handling. The pool is destroyed when last pm reference + * is dropped, which breaks our -EDEADLK handling. + */ + return err; + +err: + intel_gt_pm_put(ce->engine->gt); + intel_context_put(ce); return err; } +static void +eb_put_engine(struct i915_execbuffer *eb) +{ + intel_gt_pm_put(eb->engine->gt); + intel_context_put(eb->context); +} + static void __free_fence_array(struct eb_fence *fences, unsigned int n) { @@ -3060,6 +3141,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, eb.vma = (struct eb_vma *)(exec + args->buffer_count + 1); eb.vma[0].vma = NULL; eb.reloc_pool = eb.batch_pool = NULL; + eb.reloc_context = NULL; eb.invalid_flags = __EXEC_OBJECT_UNKNOWN_FLAGS; reloc_cache_init(&eb.reloc_cache, eb.i915); @@ -3128,7 +3210,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, if (unlikely(err)) goto err_destroy; - err = eb_pin_engine(&eb); + err = eb_select_engine(&eb); if (unlikely(err)) goto err_context; @@ -3265,8 +3347,10 @@ i915_gem_do_execbuffer(struct drm_device *dev, intel_gt_buffer_pool_put(eb.batch_pool); if (eb.reloc_pool) intel_gt_buffer_pool_put(eb.reloc_pool); + if (eb.reloc_context) + intel_context_put(eb.reloc_context); err_engine: - eb_unpin_engine(&eb); + eb_put_engine(&eb); err_context: i915_gem_context_put(eb.gem_context); err_destroy: diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c index bc08c02b5767..563839cbaf1c 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c @@ -135,6 +135,7 @@ static int igt_gpu_reloc(void *arg) goto err_pm; } eb.reloc_pool = NULL; + eb.reloc_context = NULL; i915_gem_ww_ctx_init(&eb.ww, false); retry: @@ -153,6 +154,8 @@ static int igt_gpu_reloc(void *arg) if (eb.reloc_pool) intel_gt_buffer_pool_put(eb.reloc_pool); + if (eb.reloc_context) + intel_context_put(eb.reloc_context); intel_context_put(eb.context); err_pm: From patchwork Mon Aug 10 10:30:53 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707295 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 02B9C13A4 for ; Mon, 10 Aug 2020 10:33:52 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id DEFD3206E9 for ; Mon, 10 Aug 2020 10:33:51 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org DEFD3206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 165096E417; Mon, 10 Aug 2020 10:33:39 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 6E2596E3EE for ; Mon, 10 Aug 2020 10:33:28 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:53 +0200 Message-Id: <20200810103103.303818-15-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 14/24] drm/i915: Rework intel_context pinning to do everything outside of pin_mutex 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Instead of doing everything inside of pin_mutex, we move all pinning outside. Because i915_active has its own reference counting and pinning is also having the same issues vs mutexes, we make sure everything is pinned first, so the pinning in i915_active only needs to bump refcounts. This allows us to take pin refcounts correctly all the time. Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- drivers/gpu/drm/i915/gt/intel_context.c | 232 +++++++++++------- drivers/gpu/drm/i915/gt/intel_context_types.h | 4 +- drivers/gpu/drm/i915/gt/intel_lrc.c | 34 ++- .../gpu/drm/i915/gt/intel_ring_submission.c | 13 +- drivers/gpu/drm/i915/gt/mock_engine.c | 13 +- 5 files changed, 190 insertions(+), 106 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/intel_context.c b/drivers/gpu/drm/i915/gt/intel_context.c index 52db2bde44a3..efe9a7a89ede 100644 --- a/drivers/gpu/drm/i915/gt/intel_context.c +++ b/drivers/gpu/drm/i915/gt/intel_context.c @@ -93,79 +93,6 @@ static void intel_context_active_release(struct intel_context *ce) i915_active_release(&ce->active); } -int __intel_context_do_pin(struct intel_context *ce) -{ - int err; - - if (unlikely(!test_bit(CONTEXT_ALLOC_BIT, &ce->flags))) { - err = intel_context_alloc_state(ce); - if (err) - return err; - } - - err = i915_active_acquire(&ce->active); - if (err) - return err; - - if (mutex_lock_interruptible(&ce->pin_mutex)) { - err = -EINTR; - goto out_release; - } - - if (unlikely(intel_context_is_closed(ce))) { - err = -ENOENT; - goto out_unlock; - } - - if (likely(!atomic_add_unless(&ce->pin_count, 1, 0))) { - err = intel_context_active_acquire(ce); - if (unlikely(err)) - goto out_unlock; - - err = ce->ops->pin(ce); - if (unlikely(err)) - goto err_active; - - CE_TRACE(ce, "pin ring:{start:%08x, head:%04x, tail:%04x}\n", - i915_ggtt_offset(ce->ring->vma), - ce->ring->head, ce->ring->tail); - - smp_mb__before_atomic(); /* flush pin before it is visible */ - atomic_inc(&ce->pin_count); - } - - GEM_BUG_ON(!intel_context_is_pinned(ce)); /* no overflow! */ - GEM_BUG_ON(i915_active_is_idle(&ce->active)); - goto out_unlock; - -err_active: - intel_context_active_release(ce); -out_unlock: - mutex_unlock(&ce->pin_mutex); -out_release: - i915_active_release(&ce->active); - return err; -} - -void intel_context_unpin(struct intel_context *ce) -{ - if (!atomic_dec_and_test(&ce->pin_count)) - return; - - CE_TRACE(ce, "unpin\n"); - ce->ops->unpin(ce); - - /* - * Once released, we may asynchronously drop the active reference. - * As that may be the only reference keeping the context alive, - * take an extra now so that it is not freed before we finish - * dereferencing it. - */ - intel_context_get(ce); - intel_context_active_release(ce); - intel_context_put(ce); -} - static int __context_pin_state(struct i915_vma *vma) { unsigned int bias = i915_ggtt_pin_bias(vma) | PIN_OFFSET_BIAS; @@ -225,6 +152,138 @@ static void __ring_retire(struct intel_ring *ring) intel_ring_unpin(ring); } +static int intel_context_pre_pin(struct intel_context *ce) +{ + int err; + + CE_TRACE(ce, "active\n"); + + err = __ring_active(ce->ring); + if (err) + return err; + + err = intel_timeline_pin(ce->timeline); + if (err) + goto err_ring; + + if (!ce->state) + return 0; + + err = __context_pin_state(ce->state); + if (err) + goto err_timeline; + + + return 0; + +err_timeline: + intel_timeline_unpin(ce->timeline); +err_ring: + __ring_retire(ce->ring); + return err; +} + +static void intel_context_post_unpin(struct intel_context *ce) +{ + if (ce->state) + __context_unpin_state(ce->state); + + intel_timeline_unpin(ce->timeline); + __ring_retire(ce->ring); +} + +int __intel_context_do_pin(struct intel_context *ce) +{ + bool handoff = false; + void *vaddr; + int err = 0; + + if (unlikely(!test_bit(CONTEXT_ALLOC_BIT, &ce->flags))) { + err = intel_context_alloc_state(ce); + if (err) + return err; + } + + /* + * We always pin the context/ring/timeline here, to ensure a pin + * refcount for __intel_context_active(), which prevent a lock + * inversion of ce->pin_mutex vs dma_resv_lock(). + */ + err = intel_context_pre_pin(ce); + if (err) + return err; + + err = i915_active_acquire(&ce->active); + if (err) + goto err_ctx_unpin; + + err = ce->ops->pre_pin(ce, &vaddr); + if (err) + goto err_release; + + err = mutex_lock_interruptible(&ce->pin_mutex); + if (err) + goto err_post_unpin; + + if (unlikely(intel_context_is_closed(ce))) { + err = -ENOENT; + goto err_unlock; + } + + if (likely(!atomic_add_unless(&ce->pin_count, 1, 0))) { + err = intel_context_active_acquire(ce); + if (unlikely(err)) + goto err_unlock; + + err = ce->ops->pin(ce, vaddr); + if (err) { + intel_context_active_release(ce); + goto err_unlock; + } + + CE_TRACE(ce, "pin ring:{start:%08x, head:%04x, tail:%04x}\n", + i915_ggtt_offset(ce->ring->vma), + ce->ring->head, ce->ring->tail); + + handoff = true; + smp_mb__before_atomic(); /* flush pin before it is visible */ + atomic_inc(&ce->pin_count); + } + + GEM_BUG_ON(!intel_context_is_pinned(ce)); /* no overflow! */ + +err_unlock: + mutex_unlock(&ce->pin_mutex); +err_post_unpin: + if (!handoff) + ce->ops->post_unpin(ce); +err_release: + i915_active_release(&ce->active); +err_ctx_unpin: + intel_context_post_unpin(ce); + return err; +} + +void intel_context_unpin(struct intel_context *ce) +{ + if (!atomic_dec_and_test(&ce->pin_count)) + return; + + CE_TRACE(ce, "unpin\n"); + ce->ops->unpin(ce); + ce->ops->post_unpin(ce); + + /* + * Once released, we may asynchronously drop the active reference. + * As that may be the only reference keeping the context alive, + * take an extra now so that it is not freed before we finish + * dereferencing it. + */ + intel_context_get(ce); + intel_context_active_release(ce); + intel_context_put(ce); +} + __i915_active_call static void __intel_context_retire(struct i915_active *active) { @@ -235,12 +294,7 @@ static void __intel_context_retire(struct i915_active *active) intel_context_get_avg_runtime_ns(ce)); set_bit(CONTEXT_VALID_BIT, &ce->flags); - if (ce->state) - __context_unpin_state(ce->state); - - intel_timeline_unpin(ce->timeline); - __ring_retire(ce->ring); - + intel_context_post_unpin(ce); intel_context_put(ce); } @@ -249,29 +303,25 @@ static int __intel_context_active(struct i915_active *active) struct intel_context *ce = container_of(active, typeof(*ce), active); int err; - CE_TRACE(ce, "active\n"); - intel_context_get(ce); + /* everything should already be activated by intel_context_pre_pin() */ err = __ring_active(ce->ring); - if (err) + if (GEM_WARN_ON(err)) goto err_put; err = intel_timeline_pin(ce->timeline); - if (err) + if (GEM_WARN_ON(err)) goto err_ring; - if (!ce->state) - return 0; - - err = __context_pin_state(ce->state); - if (err) - goto err_timeline; + if (ce->state) { + GEM_WARN_ON(!i915_active_acquire_if_busy(&ce->state->active)); + __i915_vma_pin(ce->state); + i915_vma_make_unshrinkable(ce->state); + } return 0; -err_timeline: - intel_timeline_unpin(ce->timeline); err_ring: __ring_retire(ce->ring); err_put: diff --git a/drivers/gpu/drm/i915/gt/intel_context_types.h b/drivers/gpu/drm/i915/gt/intel_context_types.h index 4954b0df4864..ca8e05b4d3ef 100644 --- a/drivers/gpu/drm/i915/gt/intel_context_types.h +++ b/drivers/gpu/drm/i915/gt/intel_context_types.h @@ -30,8 +30,10 @@ struct intel_ring; struct intel_context_ops { int (*alloc)(struct intel_context *ce); - int (*pin)(struct intel_context *ce); + int (*pre_pin)(struct intel_context *ce, void **vaddr); + int (*pin)(struct intel_context *ce, void *vaddr); void (*unpin)(struct intel_context *ce); + void (*post_unpin)(struct intel_context *ce); void (*enter)(struct intel_context *ce); void (*exit)(struct intel_context *ce); diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c index 417f6b0c6c61..f4390559a3d5 100644 --- a/drivers/gpu/drm/i915/gt/intel_lrc.c +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c @@ -3303,7 +3303,10 @@ static void execlists_context_unpin(struct intel_context *ce) { check_redzone((void *)ce->lrc_reg_state - LRC_STATE_OFFSET, ce->engine); +} +static void execlists_context_post_unpin(struct intel_context *ce) +{ i915_gem_object_unpin_map(ce->state->obj); } @@ -3465,20 +3468,23 @@ __execlists_update_reg_state(const struct intel_context *ce, } static int -__execlists_context_pin(struct intel_context *ce, - struct intel_engine_cs *engine) +execlists_context_pre_pin(struct intel_context *ce, void **vaddr) { - void *vaddr; - GEM_BUG_ON(!ce->state); GEM_BUG_ON(!i915_vma_is_pinned(ce->state)); - vaddr = i915_gem_object_pin_map(ce->state->obj, - i915_coherent_map_type(engine->i915) | + *vaddr = i915_gem_object_pin_map(ce->state->obj, + i915_coherent_map_type(ce->engine->i915) | I915_MAP_OVERRIDE); - if (IS_ERR(vaddr)) - return PTR_ERR(vaddr); + return PTR_ERR_OR_ZERO(*vaddr); +} + +static int +__execlists_context_pin(struct intel_context *ce, + struct intel_engine_cs *engine, + void *vaddr) +{ ce->lrc.lrca = lrc_descriptor(ce, engine) | CTX_DESC_FORCE_RESTORE; ce->lrc_reg_state = vaddr + LRC_STATE_OFFSET; __execlists_update_reg_state(ce, engine, ce->ring->tail); @@ -3486,9 +3492,9 @@ __execlists_context_pin(struct intel_context *ce, return 0; } -static int execlists_context_pin(struct intel_context *ce) +static int execlists_context_pin(struct intel_context *ce, void *vaddr) { - return __execlists_context_pin(ce, ce->engine); + return __execlists_context_pin(ce, ce->engine, vaddr); } static int execlists_context_alloc(struct intel_context *ce) @@ -3514,8 +3520,10 @@ static void execlists_context_reset(struct intel_context *ce) static const struct intel_context_ops execlists_context_ops = { .alloc = execlists_context_alloc, + .pre_pin = execlists_context_pre_pin, .pin = execlists_context_pin, .unpin = execlists_context_unpin, + .post_unpin = execlists_context_post_unpin, .enter = intel_context_enter_engine, .exit = intel_context_exit_engine, @@ -5454,12 +5462,12 @@ static int virtual_context_alloc(struct intel_context *ce) return __execlists_context_alloc(ce, ve->siblings[0]); } -static int virtual_context_pin(struct intel_context *ce) +static int virtual_context_pin(struct intel_context *ce, void *vaddr) { struct virtual_engine *ve = container_of(ce, typeof(*ve), context); /* Note: we must use a real engine class for setting up reg state */ - return __execlists_context_pin(ce, ve->siblings[0]); + return __execlists_context_pin(ce, ve->siblings[0], vaddr); } static void virtual_context_enter(struct intel_context *ce) @@ -5487,8 +5495,10 @@ static void virtual_context_exit(struct intel_context *ce) static const struct intel_context_ops virtual_context_ops = { .alloc = virtual_context_alloc, + .pre_pin = execlists_context_pre_pin, .pin = virtual_context_pin, .unpin = execlists_context_unpin, + .post_unpin = execlists_context_post_unpin, .enter = virtual_context_enter, .exit = virtual_context_exit, diff --git a/drivers/gpu/drm/i915/gt/intel_ring_submission.c b/drivers/gpu/drm/i915/gt/intel_ring_submission.c index a3b10f3c83eb..93cf72cfd318 100644 --- a/drivers/gpu/drm/i915/gt/intel_ring_submission.c +++ b/drivers/gpu/drm/i915/gt/intel_ring_submission.c @@ -499,6 +499,10 @@ static void __context_unpin_ppgtt(struct intel_context *ce) } static void ring_context_unpin(struct intel_context *ce) +{ +} + +static void ring_context_post_unpin(struct intel_context *ce) { __context_unpin_ppgtt(ce); } @@ -587,11 +591,16 @@ static int ring_context_alloc(struct intel_context *ce) return 0; } -static int ring_context_pin(struct intel_context *ce) +static int ring_context_pre_pin(struct intel_context *ce, void **unused) { return __context_pin_ppgtt(ce); } +static int ring_context_pin(struct intel_context *ce, void *unused) +{ + return 0; +} + static void ring_context_reset(struct intel_context *ce) { intel_ring_reset(ce->ring, ce->ring->emit); @@ -600,8 +609,10 @@ static void ring_context_reset(struct intel_context *ce) static const struct intel_context_ops ring_context_ops = { .alloc = ring_context_alloc, + .pre_pin = ring_context_pre_pin, .pin = ring_context_pin, .unpin = ring_context_unpin, + .post_unpin = ring_context_post_unpin, .enter = intel_context_enter_engine, .exit = intel_context_exit_engine, diff --git a/drivers/gpu/drm/i915/gt/mock_engine.c b/drivers/gpu/drm/i915/gt/mock_engine.c index 79764305b8ec..c8e631222f23 100644 --- a/drivers/gpu/drm/i915/gt/mock_engine.c +++ b/drivers/gpu/drm/i915/gt/mock_engine.c @@ -131,6 +131,10 @@ static void mock_context_unpin(struct intel_context *ce) { } +static void mock_context_post_unpin(struct intel_context *ce) +{ +} + static void mock_context_destroy(struct kref *ref) { struct intel_context *ce = container_of(ref, typeof(*ce), ref); @@ -163,7 +167,12 @@ static int mock_context_alloc(struct intel_context *ce) return 0; } -static int mock_context_pin(struct intel_context *ce) +static int mock_context_pre_pin(struct intel_context *ce, void **unused) +{ + return 0; +} + +static int mock_context_pin(struct intel_context *ce, void *unused) { return 0; } @@ -175,8 +184,10 @@ static void mock_context_reset(struct intel_context *ce) static const struct intel_context_ops mock_context_ops = { .alloc = mock_context_alloc, + .pre_pin = mock_context_pre_pin, .pin = mock_context_pin, .unpin = mock_context_unpin, + .post_unpin = mock_context_post_unpin, .enter = intel_context_enter_engine, .exit = intel_context_exit_engine, From patchwork Mon Aug 10 10:30:54 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707277 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 254FC13A4 for ; Mon, 10 Aug 2020 10:33:46 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 0DA192070B for ; Mon, 10 Aug 2020 10:33:46 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 0DA192070B Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id EFDCD6E3FB; Mon, 10 Aug 2020 10:33:31 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 67DA96E3DB for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:54 +0200 Message-Id: <20200810103103.303818-16-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 15/24] drm/i915: Make sure execbuffer always passes ww state to i915_vma_pin. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" As a preparation step for full object locking and wait/wound handling during pin and object mapping, ensure that we always pass the ww context in i915_gem_execbuffer.c to i915_vma_pin, use lockdep to ensure this happens. This also requires changing the order of eb_parse slightly, to ensure we pass ww at a point where we could still handle -EDEADLK safely. Signed-off-by: Maarten Lankhorst --- drivers/gpu/drm/i915/display/intel_display.c | 2 +- drivers/gpu/drm/i915/gem/i915_gem_context.c | 4 +- .../gpu/drm/i915/gem/i915_gem_execbuffer.c | 140 ++++++++++-------- .../i915/gem/selftests/i915_gem_execbuffer.c | 4 +- drivers/gpu/drm/i915/gt/gen6_ppgtt.c | 4 +- drivers/gpu/drm/i915/gt/gen6_ppgtt.h | 4 +- drivers/gpu/drm/i915/gt/intel_context.c | 65 +++++--- drivers/gpu/drm/i915/gt/intel_context.h | 13 ++ drivers/gpu/drm/i915/gt/intel_context_types.h | 3 +- drivers/gpu/drm/i915/gt/intel_engine_cs.c | 2 +- drivers/gpu/drm/i915/gt/intel_gt.c | 2 +- drivers/gpu/drm/i915/gt/intel_lrc.c | 5 +- drivers/gpu/drm/i915/gt/intel_renderstate.c | 2 +- drivers/gpu/drm/i915/gt/intel_ring.c | 10 +- drivers/gpu/drm/i915/gt/intel_ring.h | 3 +- .../gpu/drm/i915/gt/intel_ring_submission.c | 15 +- drivers/gpu/drm/i915/gt/intel_timeline.c | 12 +- drivers/gpu/drm/i915/gt/intel_timeline.h | 3 +- drivers/gpu/drm/i915/gt/mock_engine.c | 3 +- drivers/gpu/drm/i915/gt/selftest_lrc.c | 2 +- drivers/gpu/drm/i915/gt/selftest_timeline.c | 4 +- drivers/gpu/drm/i915/gt/uc/intel_guc.c | 2 +- drivers/gpu/drm/i915/i915_drv.h | 13 +- drivers/gpu/drm/i915/i915_gem.c | 11 +- drivers/gpu/drm/i915/i915_vma.c | 13 +- drivers/gpu/drm/i915/i915_vma.h | 13 +- 26 files changed, 217 insertions(+), 137 deletions(-) diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index 5b4434289117..aa5a88340d10 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -3451,7 +3451,7 @@ initial_plane_vma(struct drm_i915_private *i915, if (IS_ERR(vma)) goto err_obj; - if (i915_ggtt_pin(vma, 0, PIN_MAPPABLE | PIN_OFFSET_FIXED | base)) + if (i915_ggtt_pin(vma, NULL, 0, PIN_MAPPABLE | PIN_OFFSET_FIXED | base)) goto err_obj; if (i915_gem_object_is_tiled(obj) && diff --git a/drivers/gpu/drm/i915/gem/i915_gem_context.c b/drivers/gpu/drm/i915/gem/i915_gem_context.c index 34c8b0dd85e0..cf5ecbde9e06 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_context.c @@ -1154,7 +1154,7 @@ static int context_barrier_task(struct i915_gem_context *ctx, i915_gem_ww_ctx_init(&ww, true); retry: - err = intel_context_pin(ce); + err = intel_context_pin_ww(ce, &ww); if (err) goto err; @@ -1247,7 +1247,7 @@ static int pin_ppgtt_update(struct intel_context *ce, struct i915_gem_ww_ctx *ww if (!HAS_LOGICAL_RING_CONTEXTS(vm->i915)) /* ppGTT is not part of the legacy context image */ - return gen6_ppgtt_pin(i915_vm_to_ppgtt(vm)); + return gen6_ppgtt_pin(i915_vm_to_ppgtt(vm), ww); return 0; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 604e26adea23..94bfdc54f035 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -437,16 +437,17 @@ eb_pin_vma(struct i915_execbuffer *eb, pin_flags |= PIN_GLOBAL; /* Attempt to reuse the current location if available */ - if (unlikely(i915_vma_pin(vma, 0, 0, pin_flags))) { + /* TODO: Add -EDEADLK handling here */ + if (unlikely(i915_vma_pin_ww(vma, &eb->ww, 0, 0, pin_flags))) { if (entry->flags & EXEC_OBJECT_PINNED) return false; /* Failing that pick any _free_ space if suitable */ - if (unlikely(i915_vma_pin(vma, - entry->pad_to_size, - entry->alignment, - eb_pin_flags(entry, ev->flags) | - PIN_USER | PIN_NOEVICT))) + if (unlikely(i915_vma_pin_ww(vma, &eb->ww, + entry->pad_to_size, + entry->alignment, + eb_pin_flags(entry, ev->flags) | + PIN_USER | PIN_NOEVICT))) return false; } @@ -587,7 +588,7 @@ static inline int use_cpu_reloc(const struct reloc_cache *cache, obj->cache_level != I915_CACHE_NONE); } -static int eb_reserve_vma(const struct i915_execbuffer *eb, +static int eb_reserve_vma(struct i915_execbuffer *eb, struct eb_vma *ev, u64 pin_flags) { @@ -602,7 +603,7 @@ static int eb_reserve_vma(const struct i915_execbuffer *eb, return err; } - err = i915_vma_pin(vma, + err = i915_vma_pin_ww(vma, &eb->ww, entry->pad_to_size, entry->alignment, eb_pin_flags(entry, ev->flags) | pin_flags); if (err) @@ -1133,9 +1134,10 @@ static void *reloc_kmap(struct drm_i915_gem_object *obj, } static void *reloc_iomap(struct drm_i915_gem_object *obj, - struct reloc_cache *cache, + struct i915_execbuffer *eb, unsigned long page) { + struct reloc_cache *cache = &eb->reloc_cache; struct i915_ggtt *ggtt = cache_to_ggtt(cache); unsigned long offset; void *vaddr; @@ -1157,10 +1159,13 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj, if (err) return ERR_PTR(err); - vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, - PIN_MAPPABLE | - PIN_NONBLOCK /* NOWARN */ | - PIN_NOEVICT); + vma = i915_gem_object_ggtt_pin_ww(obj, &eb->ww, NULL, 0, 0, + PIN_MAPPABLE | + PIN_NONBLOCK /* NOWARN */ | + PIN_NOEVICT); + if (vma == ERR_PTR(-EDEADLK)) + return vma; + if (IS_ERR(vma)) { memset(&cache->node, 0, sizeof(cache->node)); mutex_lock(&ggtt->vm.mutex); @@ -1196,9 +1201,10 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj, } static void *reloc_vaddr(struct drm_i915_gem_object *obj, - struct reloc_cache *cache, + struct i915_execbuffer *eb, unsigned long page) { + struct reloc_cache *cache = &eb->reloc_cache; void *vaddr; if (cache->page == page) { @@ -1206,7 +1212,7 @@ static void *reloc_vaddr(struct drm_i915_gem_object *obj, } else { vaddr = NULL; if ((cache->vaddr & KMAP) == 0) - vaddr = reloc_iomap(obj, cache, page); + vaddr = reloc_iomap(obj, eb, page); if (!vaddr) vaddr = reloc_kmap(obj, cache, page); } @@ -1293,7 +1299,7 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, goto err_unmap; } - err = i915_vma_pin(batch, 0, 0, PIN_USER | PIN_NONBLOCK); + err = i915_vma_pin_ww(batch, &eb->ww, 0, 0, PIN_USER | PIN_NONBLOCK); if (err) goto err_unmap; @@ -1314,7 +1320,7 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb, eb->reloc_context = ce; } - err = intel_context_pin(ce); + err = intel_context_pin_ww(ce, &eb->ww); if (err) goto err_unpin; @@ -1537,8 +1543,7 @@ relocate_entry(struct i915_vma *vma, void *vaddr; repeat: - vaddr = reloc_vaddr(vma->obj, - &eb->reloc_cache, + vaddr = reloc_vaddr(vma->obj, eb, offset >> PAGE_SHIFT); if (IS_ERR(vaddr)) return PTR_ERR(vaddr); @@ -1954,6 +1959,7 @@ static noinline int eb_relocate_parse_slow(struct i915_execbuffer *eb, rq = eb_pin_engine(eb, false); if (IS_ERR(rq)) { err = PTR_ERR(rq); + rq = NULL; goto err; } @@ -2238,7 +2244,8 @@ static int i915_reset_gen7_sol_offsets(struct i915_request *rq) } static struct i915_vma * -shadow_batch_pin(struct drm_i915_gem_object *obj, +shadow_batch_pin(struct i915_execbuffer *eb, + struct drm_i915_gem_object *obj, struct i915_address_space *vm, unsigned int flags) { @@ -2249,7 +2256,7 @@ shadow_batch_pin(struct drm_i915_gem_object *obj, if (IS_ERR(vma)) return vma; - err = i915_vma_pin(vma, 0, 0, flags); + err = i915_vma_pin_ww(vma, &eb->ww, 0, 0, flags); if (err) return ERR_PTR(err); @@ -2403,16 +2410,33 @@ static int eb_parse_pipeline(struct i915_execbuffer *eb, return err; } +static struct i915_vma *eb_dispatch_secure(struct i915_execbuffer *eb, struct i915_vma *vma) +{ + /* + * snb/ivb/vlv conflate the "batch in ppgtt" bit with the "non-secure + * batch" bit. Hence we need to pin secure batches into the global gtt. + * hsw should have this fixed, but bdw mucks it up again. */ + if (eb->batch_flags & I915_DISPATCH_SECURE) + return i915_gem_object_ggtt_pin_ww(vma->obj, &eb->ww, NULL, 0, 0, 0); + + return NULL; +} + static int eb_parse(struct i915_execbuffer *eb) { struct drm_i915_private *i915 = eb->i915; struct intel_gt_buffer_pool_node *pool = eb->batch_pool; - struct i915_vma *shadow, *trampoline; + struct i915_vma *shadow, *trampoline, *batch; unsigned int len; int err; - if (!eb_use_cmdparser(eb)) - return 0; + if (!eb_use_cmdparser(eb)) { + batch = eb_dispatch_secure(eb, eb->batch->vma); + if (IS_ERR(batch)) + return PTR_ERR(batch); + + goto secure_batch; + } len = eb->batch_len; if (!CMDPARSER_USES_GGTT(eb->i915)) { @@ -2440,7 +2464,7 @@ static int eb_parse(struct i915_execbuffer *eb) if (err) goto err; - shadow = shadow_batch_pin(pool->obj, eb->context->vm, PIN_USER); + shadow = shadow_batch_pin(eb, pool->obj, eb->context->vm, PIN_USER); if (IS_ERR(shadow)) { err = PTR_ERR(shadow); goto err; @@ -2452,7 +2476,7 @@ static int eb_parse(struct i915_execbuffer *eb) if (CMDPARSER_USES_GGTT(eb->i915)) { trampoline = shadow; - shadow = shadow_batch_pin(pool->obj, + shadow = shadow_batch_pin(eb, pool->obj, &eb->engine->gt->ggtt->vm, PIN_GLOBAL); if (IS_ERR(shadow)) { @@ -2465,19 +2489,34 @@ static int eb_parse(struct i915_execbuffer *eb) eb->batch_flags |= I915_DISPATCH_SECURE; } + batch = eb_dispatch_secure(eb, shadow); + if (IS_ERR(batch)) { + err = PTR_ERR(batch); + goto err_trampoline; + } + err = eb_parse_pipeline(eb, shadow, trampoline); if (err) - goto err_trampoline; + goto err_unpin_batch; - eb->vma[eb->buffer_count].vma = i915_vma_get(shadow); - eb->vma[eb->buffer_count].flags = __EXEC_OBJECT_HAS_PIN; eb->batch = &eb->vma[eb->buffer_count++]; + eb->batch->vma = i915_vma_get(shadow); + eb->batch->flags = __EXEC_OBJECT_HAS_PIN; eb->trampoline = trampoline; eb->batch_start_offset = 0; +secure_batch: + if (batch) { + eb->batch = &eb->vma[eb->buffer_count++]; + eb->batch->flags = __EXEC_OBJECT_HAS_PIN; + eb->batch->vma = i915_vma_get(batch); + } return 0; +err_unpin_batch: + if (batch) + i915_vma_unpin(batch); err_trampoline: if (trampoline) i915_vma_unpin(trampoline); @@ -2619,7 +2658,7 @@ static struct i915_request *eb_pin_engine(struct i915_execbuffer *eb, bool throt * GGTT space, so do this first before we reserve a seqno for * ourselves. */ - err = intel_context_pin(ce); + err = intel_context_pin_ww(ce, &eb->ww); if (err) return ERR_PTR(err); @@ -3237,33 +3276,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, ww_acquire_done(&eb.ww.ctx); - /* - * snb/ivb/vlv conflate the "batch in ppgtt" bit with the "non-secure - * batch" bit. Hence we need to pin secure batches into the global gtt. - * hsw should have this fixed, but bdw mucks it up again. */ - if (eb.batch_flags & I915_DISPATCH_SECURE) { - struct i915_vma *vma; - - /* - * So on first glance it looks freaky that we pin the batch here - * outside of the reservation loop. But: - * - The batch is already pinned into the relevant ppgtt, so we - * already have the backing storage fully allocated. - * - No other BO uses the global gtt (well contexts, but meh), - * so we don't really have issues with multiple objects not - * fitting due to fragmentation. - * So this is actually safe. - */ - vma = i915_gem_object_ggtt_pin(eb.batch->vma->obj, NULL, 0, 0, 0); - if (IS_ERR(vma)) { - err = PTR_ERR(vma); - goto err_vma; - } - - batch = vma; - } else { - batch = eb.batch->vma; - } + batch = eb.batch->vma; /* All GPU relocation batches must be submitted prior to the user rq */ GEM_BUG_ON(eb.reloc_cache.rq); @@ -3272,7 +3285,7 @@ i915_gem_do_execbuffer(struct drm_device *dev, eb.request = i915_request_create(eb.context); if (IS_ERR(eb.request)) { err = PTR_ERR(eb.request); - goto err_batch_unpin; + goto err_vma; } if (in_fence) { @@ -3333,9 +3346,6 @@ i915_gem_do_execbuffer(struct drm_device *dev, } i915_request_put(eb.request); -err_batch_unpin: - if (eb.batch_flags & I915_DISPATCH_SECURE) - i915_vma_unpin(batch); err_vma: eb_release_vmas(&eb, true); if (eb.trampoline) @@ -3423,7 +3433,9 @@ i915_gem_execbuffer_ioctl(struct drm_device *dev, void *data, /* Copy in the exec list from userland */ exec_list = kvmalloc_array(count, sizeof(*exec_list), __GFP_NOWARN | GFP_KERNEL); - exec2_list = kvmalloc_array(count + 1, eb_element_size(), + + /* Allocate extra slots for use by the command parser */ + exec2_list = kvmalloc_array(count + 2, eb_element_size(), __GFP_NOWARN | GFP_KERNEL); if (exec_list == NULL || exec2_list == NULL) { drm_dbg(&i915->drm, @@ -3500,8 +3512,8 @@ i915_gem_execbuffer2_ioctl(struct drm_device *dev, void *data, if (err) return err; - /* Allocate an extra slot for use by the command parser */ - exec2_list = kvmalloc_array(count + 1, eb_element_size(), + /* Allocate extra slots for use by the command parser */ + exec2_list = kvmalloc_array(count + 2, eb_element_size(), __GFP_NOWARN | GFP_KERNEL); if (exec2_list == NULL) { drm_dbg(&i915->drm, "Failed to allocate exec list for %zd buffers\n", diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c index 563839cbaf1c..e1d50a5a1477 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_execbuffer.c @@ -36,7 +36,7 @@ static int __igt_gpu_reloc(struct i915_execbuffer *eb, if (err) return err; - err = i915_vma_pin(vma, 0, 0, PIN_USER | PIN_HIGH); + err = i915_vma_pin_ww(vma, &eb->ww, 0, 0, PIN_USER | PIN_HIGH); if (err) return err; @@ -139,7 +139,7 @@ static int igt_gpu_reloc(void *arg) i915_gem_ww_ctx_init(&eb.ww, false); retry: - err = intel_context_pin(eb.context); + err = intel_context_pin_ww(eb.context, &eb.ww); if (!err) { err = __igt_gpu_reloc(&eb, scratch); diff --git a/drivers/gpu/drm/i915/gt/gen6_ppgtt.c b/drivers/gpu/drm/i915/gt/gen6_ppgtt.c index 7e5a86b774a7..fd0d24d28763 100644 --- a/drivers/gpu/drm/i915/gt/gen6_ppgtt.c +++ b/drivers/gpu/drm/i915/gt/gen6_ppgtt.c @@ -368,7 +368,7 @@ static struct i915_vma *pd_vma_create(struct gen6_ppgtt *ppgtt, int size) return vma; } -int gen6_ppgtt_pin(struct i915_ppgtt *base) +int gen6_ppgtt_pin(struct i915_ppgtt *base, struct i915_gem_ww_ctx *ww) { struct gen6_ppgtt *ppgtt = to_gen6_ppgtt(base); int err; @@ -394,7 +394,7 @@ int gen6_ppgtt_pin(struct i915_ppgtt *base) */ err = 0; if (!atomic_read(&ppgtt->pin_count)) - err = i915_ggtt_pin(ppgtt->vma, GEN6_PD_ALIGN, PIN_HIGH); + err = i915_ggtt_pin(ppgtt->vma, ww, GEN6_PD_ALIGN, PIN_HIGH); if (!err) atomic_inc(&ppgtt->pin_count); mutex_unlock(&ppgtt->pin_mutex); diff --git a/drivers/gpu/drm/i915/gt/gen6_ppgtt.h b/drivers/gpu/drm/i915/gt/gen6_ppgtt.h index 7249672e5802..3357228f3304 100644 --- a/drivers/gpu/drm/i915/gt/gen6_ppgtt.h +++ b/drivers/gpu/drm/i915/gt/gen6_ppgtt.h @@ -8,6 +8,8 @@ #include "intel_gtt.h" +struct i915_gem_ww_ctx; + struct gen6_ppgtt { struct i915_ppgtt base; @@ -67,7 +69,7 @@ static inline struct gen6_ppgtt *to_gen6_ppgtt(struct i915_ppgtt *base) (pt = i915_pt_entry(pd, iter), true); \ ++iter) -int gen6_ppgtt_pin(struct i915_ppgtt *base); +int gen6_ppgtt_pin(struct i915_ppgtt *base, struct i915_gem_ww_ctx *ww); void gen6_ppgtt_unpin(struct i915_ppgtt *base); void gen6_ppgtt_unpin_all(struct i915_ppgtt *base); void gen6_ppgtt_enable(struct intel_gt *gt); diff --git a/drivers/gpu/drm/i915/gt/intel_context.c b/drivers/gpu/drm/i915/gt/intel_context.c index efe9a7a89ede..c05ef213bdc2 100644 --- a/drivers/gpu/drm/i915/gt/intel_context.c +++ b/drivers/gpu/drm/i915/gt/intel_context.c @@ -93,12 +93,12 @@ static void intel_context_active_release(struct intel_context *ce) i915_active_release(&ce->active); } -static int __context_pin_state(struct i915_vma *vma) +static int __context_pin_state(struct i915_vma *vma, struct i915_gem_ww_ctx *ww) { unsigned int bias = i915_ggtt_pin_bias(vma) | PIN_OFFSET_BIAS; int err; - err = i915_ggtt_pin(vma, 0, bias | PIN_HIGH); + err = i915_ggtt_pin(vma, ww, 0, bias | PIN_HIGH); if (err) return err; @@ -127,11 +127,12 @@ static void __context_unpin_state(struct i915_vma *vma) __i915_vma_unpin(vma); } -static int __ring_active(struct intel_ring *ring) +static int __ring_active(struct intel_ring *ring, + struct i915_gem_ww_ctx *ww) { int err; - err = intel_ring_pin(ring); + err = intel_ring_pin(ring, ww); if (err) return err; @@ -152,24 +153,25 @@ static void __ring_retire(struct intel_ring *ring) intel_ring_unpin(ring); } -static int intel_context_pre_pin(struct intel_context *ce) +static int intel_context_pre_pin(struct intel_context *ce, + struct i915_gem_ww_ctx *ww) { int err; CE_TRACE(ce, "active\n"); - err = __ring_active(ce->ring); + err = __ring_active(ce->ring, ww); if (err) return err; - err = intel_timeline_pin(ce->timeline); + err = intel_timeline_pin(ce->timeline, ww); if (err) goto err_ring; if (!ce->state) return 0; - err = __context_pin_state(ce->state); + err = __context_pin_state(ce->state, ww); if (err) goto err_timeline; @@ -192,7 +194,8 @@ static void intel_context_post_unpin(struct intel_context *ce) __ring_retire(ce->ring); } -int __intel_context_do_pin(struct intel_context *ce) +int __intel_context_do_pin_ww(struct intel_context *ce, + struct i915_gem_ww_ctx *ww) { bool handoff = false; void *vaddr; @@ -209,7 +212,14 @@ int __intel_context_do_pin(struct intel_context *ce) * refcount for __intel_context_active(), which prevent a lock * inversion of ce->pin_mutex vs dma_resv_lock(). */ - err = intel_context_pre_pin(ce); + + err = i915_gem_object_lock(ce->timeline->hwsp_ggtt->obj, ww); + if (!err && ce->ring->vma->obj) + err = i915_gem_object_lock(ce->ring->vma->obj, ww); + if (!err && ce->state) + err = i915_gem_object_lock(ce->state->obj, ww); + if (!err) + err = intel_context_pre_pin(ce, ww); if (err) return err; @@ -217,7 +227,7 @@ int __intel_context_do_pin(struct intel_context *ce) if (err) goto err_ctx_unpin; - err = ce->ops->pre_pin(ce, &vaddr); + err = ce->ops->pre_pin(ce, ww, &vaddr); if (err) goto err_release; @@ -264,6 +274,23 @@ int __intel_context_do_pin(struct intel_context *ce) return err; } +int __intel_context_do_pin(struct intel_context *ce) +{ + struct i915_gem_ww_ctx ww; + int err; + + i915_gem_ww_ctx_init(&ww, true); +retry: + err = __intel_context_do_pin_ww(ce, &ww); + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + return err; +} + void intel_context_unpin(struct intel_context *ce) { if (!atomic_dec_and_test(&ce->pin_count)) @@ -301,18 +328,14 @@ static void __intel_context_retire(struct i915_active *active) static int __intel_context_active(struct i915_active *active) { struct intel_context *ce = container_of(active, typeof(*ce), active); - int err; intel_context_get(ce); /* everything should already be activated by intel_context_pre_pin() */ - err = __ring_active(ce->ring); - if (GEM_WARN_ON(err)) - goto err_put; + GEM_WARN_ON(!i915_active_acquire_if_busy(&ce->ring->vma->active)); + __intel_ring_pin(ce->ring); - err = intel_timeline_pin(ce->timeline); - if (GEM_WARN_ON(err)) - goto err_ring; + __intel_timeline_pin(ce->timeline); if (ce->state) { GEM_WARN_ON(!i915_active_acquire_if_busy(&ce->state->active)); @@ -321,12 +344,6 @@ static int __intel_context_active(struct i915_active *active) } return 0; - -err_ring: - __ring_retire(ce->ring); -err_put: - intel_context_put(ce); - return err; } void diff --git a/drivers/gpu/drm/i915/gt/intel_context.h b/drivers/gpu/drm/i915/gt/intel_context.h index 07be021882cc..fda2eba81e22 100644 --- a/drivers/gpu/drm/i915/gt/intel_context.h +++ b/drivers/gpu/drm/i915/gt/intel_context.h @@ -25,6 +25,8 @@ ##__VA_ARGS__); \ } while (0) +struct i915_gem_ww_ctx; + void intel_context_init(struct intel_context *ce, struct intel_engine_cs *engine); void intel_context_fini(struct intel_context *ce); @@ -81,6 +83,8 @@ static inline void intel_context_unlock_pinned(struct intel_context *ce) } int __intel_context_do_pin(struct intel_context *ce); +int __intel_context_do_pin_ww(struct intel_context *ce, + struct i915_gem_ww_ctx *ww); static inline bool intel_context_pin_if_active(struct intel_context *ce) { @@ -95,6 +99,15 @@ static inline int intel_context_pin(struct intel_context *ce) return __intel_context_do_pin(ce); } +static inline int intel_context_pin_ww(struct intel_context *ce, + struct i915_gem_ww_ctx *ww) +{ + if (likely(intel_context_pin_if_active(ce))) + return 0; + + return __intel_context_do_pin_ww(ce, ww); +} + static inline void __intel_context_pin(struct intel_context *ce) { GEM_BUG_ON(!intel_context_is_pinned(ce)); diff --git a/drivers/gpu/drm/i915/gt/intel_context_types.h b/drivers/gpu/drm/i915/gt/intel_context_types.h index ca8e05b4d3ef..552cb57a2e8c 100644 --- a/drivers/gpu/drm/i915/gt/intel_context_types.h +++ b/drivers/gpu/drm/i915/gt/intel_context_types.h @@ -23,6 +23,7 @@ DECLARE_EWMA(runtime, 3, 8); struct i915_gem_context; +struct i915_gem_ww_ctx; struct i915_vma; struct intel_context; struct intel_ring; @@ -30,7 +31,7 @@ struct intel_ring; struct intel_context_ops { int (*alloc)(struct intel_context *ce); - int (*pre_pin)(struct intel_context *ce, void **vaddr); + int (*pre_pin)(struct intel_context *ce, struct i915_gem_ww_ctx *ww, void **vaddr); int (*pin)(struct intel_context *ce, void *vaddr); void (*unpin)(struct intel_context *ce); void (*post_unpin)(struct intel_context *ce); diff --git a/drivers/gpu/drm/i915/gt/intel_engine_cs.c b/drivers/gpu/drm/i915/gt/intel_engine_cs.c index ea4ba2afe9f9..f231edd3fa3a 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine_cs.c +++ b/drivers/gpu/drm/i915/gt/intel_engine_cs.c @@ -635,7 +635,7 @@ static int pin_ggtt_status_page(struct intel_engine_cs *engine, else flags = PIN_HIGH; - return i915_ggtt_pin(vma, 0, flags); + return i915_ggtt_pin(vma, NULL, 0, flags); } static int init_status_page(struct intel_engine_cs *engine) diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c index 8694ddbdac4c..39b428c5049c 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt.c +++ b/drivers/gpu/drm/i915/gt/intel_gt.c @@ -356,7 +356,7 @@ static int intel_gt_init_scratch(struct intel_gt *gt, unsigned int size) goto err_unref; } - ret = i915_ggtt_pin(vma, 0, PIN_HIGH); + ret = i915_ggtt_pin(vma, NULL, 0, PIN_HIGH); if (ret) goto err_unref; diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c index f4390559a3d5..df4688d4d264 100644 --- a/drivers/gpu/drm/i915/gt/intel_lrc.c +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c @@ -3468,7 +3468,8 @@ __execlists_update_reg_state(const struct intel_context *ce, } static int -execlists_context_pre_pin(struct intel_context *ce, void **vaddr) +execlists_context_pre_pin(struct intel_context *ce, + struct i915_gem_ww_ctx *ww, void **vaddr) { GEM_BUG_ON(!ce->state); GEM_BUG_ON(!i915_vma_is_pinned(ce->state)); @@ -3887,7 +3888,7 @@ static int lrc_setup_wa_ctx(struct intel_engine_cs *engine) goto err; } - err = i915_ggtt_pin(vma, 0, PIN_HIGH); + err = i915_ggtt_pin(vma, NULL, 0, PIN_HIGH); if (err) goto err; diff --git a/drivers/gpu/drm/i915/gt/intel_renderstate.c b/drivers/gpu/drm/i915/gt/intel_renderstate.c index 5164de04049d..ea2a77c7b469 100644 --- a/drivers/gpu/drm/i915/gt/intel_renderstate.c +++ b/drivers/gpu/drm/i915/gt/intel_renderstate.c @@ -184,7 +184,7 @@ int intel_renderstate_init(struct intel_renderstate *so, i915_gem_ww_ctx_init(&so->ww, true); retry: - err = intel_context_pin(ce); + err = intel_context_pin_ww(ce, &so->ww); if (err) goto err_fini; diff --git a/drivers/gpu/drm/i915/gt/intel_ring.c b/drivers/gpu/drm/i915/gt/intel_ring.c index bdb324167ef3..4034a4bac7f0 100644 --- a/drivers/gpu/drm/i915/gt/intel_ring.c +++ b/drivers/gpu/drm/i915/gt/intel_ring.c @@ -21,7 +21,13 @@ unsigned int intel_ring_update_space(struct intel_ring *ring) return space; } -int intel_ring_pin(struct intel_ring *ring) +void __intel_ring_pin(struct intel_ring *ring) +{ + GEM_BUG_ON(!atomic_read(&ring->pin_count)); + atomic_inc(&ring->pin_count); +} + +int intel_ring_pin(struct intel_ring *ring, struct i915_gem_ww_ctx *ww) { struct i915_vma *vma = ring->vma; unsigned int flags; @@ -39,7 +45,7 @@ int intel_ring_pin(struct intel_ring *ring) else flags |= PIN_HIGH; - ret = i915_ggtt_pin(vma, 0, flags); + ret = i915_ggtt_pin(vma, ww, 0, flags); if (unlikely(ret)) goto err_unpin; diff --git a/drivers/gpu/drm/i915/gt/intel_ring.h b/drivers/gpu/drm/i915/gt/intel_ring.h index cc0ebca65167..1700579bdc93 100644 --- a/drivers/gpu/drm/i915/gt/intel_ring.h +++ b/drivers/gpu/drm/i915/gt/intel_ring.h @@ -21,7 +21,8 @@ int intel_ring_cacheline_align(struct i915_request *rq); unsigned int intel_ring_update_space(struct intel_ring *ring); -int intel_ring_pin(struct intel_ring *ring); +void __intel_ring_pin(struct intel_ring *ring); +int intel_ring_pin(struct intel_ring *ring, struct i915_gem_ww_ctx *ww); void intel_ring_unpin(struct intel_ring *ring); void intel_ring_reset(struct intel_ring *ring, u32 tail); diff --git a/drivers/gpu/drm/i915/gt/intel_ring_submission.c b/drivers/gpu/drm/i915/gt/intel_ring_submission.c index 93cf72cfd318..1ca1bac81cf6 100644 --- a/drivers/gpu/drm/i915/gt/intel_ring_submission.c +++ b/drivers/gpu/drm/i915/gt/intel_ring_submission.c @@ -477,14 +477,16 @@ static void ring_context_destroy(struct kref *ref) intel_context_free(ce); } -static int __context_pin_ppgtt(struct intel_context *ce) +static int ring_context_pre_pin(struct intel_context *ce, + struct i915_gem_ww_ctx *ww, + void **unused) { struct i915_address_space *vm; int err = 0; vm = vm_alias(ce->vm); if (vm) - err = gen6_ppgtt_pin(i915_vm_to_ppgtt((vm))); + err = gen6_ppgtt_pin(i915_vm_to_ppgtt((vm)), ww); return err; } @@ -591,11 +593,6 @@ static int ring_context_alloc(struct intel_context *ce) return 0; } -static int ring_context_pre_pin(struct intel_context *ce, void **unused) -{ - return __context_pin_ppgtt(ce); -} - static int ring_context_pin(struct intel_context *ce, void *unused) { return 0; @@ -1272,7 +1269,7 @@ int intel_ring_submission_setup(struct intel_engine_cs *engine) } GEM_BUG_ON(timeline->has_initial_breadcrumb); - err = intel_timeline_pin(timeline); + err = intel_timeline_pin(timeline, NULL); if (err) goto err_timeline; @@ -1282,7 +1279,7 @@ int intel_ring_submission_setup(struct intel_engine_cs *engine) goto err_timeline_unpin; } - err = intel_ring_pin(ring); + err = intel_ring_pin(ring, NULL); if (err) goto err_ring; diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.c b/drivers/gpu/drm/i915/gt/intel_timeline.c index 2baecf414acb..a2f74cefe4c3 100644 --- a/drivers/gpu/drm/i915/gt/intel_timeline.c +++ b/drivers/gpu/drm/i915/gt/intel_timeline.c @@ -317,14 +317,20 @@ __intel_timeline_create(struct intel_gt *gt, return timeline; } -int intel_timeline_pin(struct intel_timeline *tl) +void __intel_timeline_pin(struct intel_timeline *tl) +{ + GEM_BUG_ON(!atomic_read(&tl->pin_count)); + atomic_inc(&tl->pin_count); +} + +int intel_timeline_pin(struct intel_timeline *tl, struct i915_gem_ww_ctx *ww) { int err; if (atomic_add_unless(&tl->pin_count, 1, 0)) return 0; - err = i915_ggtt_pin(tl->hwsp_ggtt, 0, PIN_HIGH); + err = i915_ggtt_pin(tl->hwsp_ggtt, ww, 0, PIN_HIGH); if (err) return err; @@ -467,7 +473,7 @@ __intel_timeline_get_seqno(struct intel_timeline *tl, goto err_rollback; } - err = i915_ggtt_pin(vma, 0, PIN_HIGH); + err = i915_ggtt_pin(vma, NULL, 0, PIN_HIGH); if (err) { __idle_hwsp_free(vma->private, cacheline); goto err_rollback; diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.h b/drivers/gpu/drm/i915/gt/intel_timeline.h index 06bd06c6595f..9882cd911d8e 100644 --- a/drivers/gpu/drm/i915/gt/intel_timeline.h +++ b/drivers/gpu/drm/i915/gt/intel_timeline.h @@ -88,7 +88,8 @@ static inline bool intel_timeline_sync_is_later(struct intel_timeline *tl, return __intel_timeline_sync_is_later(tl, fence->context, fence->seqno); } -int intel_timeline_pin(struct intel_timeline *tl); +void __intel_timeline_pin(struct intel_timeline *tl); +int intel_timeline_pin(struct intel_timeline *tl, struct i915_gem_ww_ctx *ww); void intel_timeline_enter(struct intel_timeline *tl); int intel_timeline_get_seqno(struct intel_timeline *tl, struct i915_request *rq, diff --git a/drivers/gpu/drm/i915/gt/mock_engine.c b/drivers/gpu/drm/i915/gt/mock_engine.c index c8e631222f23..dfd1cfb8a7ec 100644 --- a/drivers/gpu/drm/i915/gt/mock_engine.c +++ b/drivers/gpu/drm/i915/gt/mock_engine.c @@ -167,7 +167,8 @@ static int mock_context_alloc(struct intel_context *ce) return 0; } -static int mock_context_pre_pin(struct intel_context *ce, void **unused) +static int mock_context_pre_pin(struct intel_context *ce, + struct i915_gem_ww_ctx *ww, void **unused) { return 0; } diff --git a/drivers/gpu/drm/i915/gt/selftest_lrc.c b/drivers/gpu/drm/i915/gt/selftest_lrc.c index f749071f54a7..7faba9f91224 100644 --- a/drivers/gpu/drm/i915/gt/selftest_lrc.c +++ b/drivers/gpu/drm/i915/gt/selftest_lrc.c @@ -3089,7 +3089,7 @@ static struct i915_vma *create_global(struct intel_gt *gt, size_t sz) return vma; } - err = i915_ggtt_pin(vma, 0, 0); + err = i915_ggtt_pin(vma, NULL, 0, 0); if (err) { i915_vma_put(vma); return ERR_PTR(err); diff --git a/drivers/gpu/drm/i915/gt/selftest_timeline.c b/drivers/gpu/drm/i915/gt/selftest_timeline.c index 98ceac2fb77d..96d164a3841d 100644 --- a/drivers/gpu/drm/i915/gt/selftest_timeline.c +++ b/drivers/gpu/drm/i915/gt/selftest_timeline.c @@ -455,7 +455,7 @@ tl_write(struct intel_timeline *tl, struct intel_engine_cs *engine, u32 value) struct i915_request *rq; int err; - err = intel_timeline_pin(tl); + err = intel_timeline_pin(tl, NULL); if (err) { rq = ERR_PTR(err); goto out; @@ -667,7 +667,7 @@ static int live_hwsp_wrap(void *arg) if (!tl->has_initial_breadcrumb || !tl->hwsp_cacheline) goto out_free; - err = intel_timeline_pin(tl); + err = intel_timeline_pin(tl, NULL); if (err) goto out_free; diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.c b/drivers/gpu/drm/i915/gt/uc/intel_guc.c index 861657897c0f..942c7c187adb 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.c @@ -677,7 +677,7 @@ struct i915_vma *intel_guc_allocate_vma(struct intel_guc *guc, u32 size) goto err; flags = PIN_OFFSET_BIAS | i915_ggtt_pin_bias(vma); - ret = i915_ggtt_pin(vma, 0, flags); + ret = i915_ggtt_pin(vma, NULL, 0, flags); if (ret) { vma = ERR_PTR(ret); goto err; diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 05a2624116a1..4fc14923e5a8 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -1788,11 +1788,18 @@ static inline void i915_gem_drain_workqueue(struct drm_i915_private *i915) } struct i915_vma * __must_check +i915_gem_object_ggtt_pin_ww(struct drm_i915_gem_object *obj, + struct i915_gem_ww_ctx *ww, + const struct i915_ggtt_view *view, + u64 size, u64 alignment, u64 flags); + +static inline struct i915_vma * __must_check i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj, const struct i915_ggtt_view *view, - u64 size, - u64 alignment, - u64 flags); + u64 size, u64 alignment, u64 flags) +{ + return i915_gem_object_ggtt_pin_ww(obj, NULL, view, size, alignment, flags); +} int i915_gem_object_unbind(struct drm_i915_gem_object *obj, unsigned long flags); diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index f6615d942b60..bb0c12975f38 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -962,11 +962,10 @@ static void discard_ggtt_vma(struct i915_vma *vma) } struct i915_vma * -i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj, - const struct i915_ggtt_view *view, - u64 size, - u64 alignment, - u64 flags) +i915_gem_object_ggtt_pin_ww(struct drm_i915_gem_object *obj, + struct i915_gem_ww_ctx *ww, + const struct i915_ggtt_view *view, + u64 size, u64 alignment, u64 flags) { struct drm_i915_private *i915 = to_i915(obj->base.dev); struct i915_ggtt *ggtt = &i915->ggtt; @@ -1032,7 +1031,7 @@ i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj, return ERR_PTR(ret); } - ret = i915_vma_pin(vma, size, alignment, flags | PIN_GLOBAL); + ret = i915_vma_pin_ww(vma, ww, size, alignment, flags | PIN_GLOBAL); if (ret) return ERR_PTR(ret); diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c index 8066f167d6b9..52f2af31a7db 100644 --- a/drivers/gpu/drm/i915/i915_vma.c +++ b/drivers/gpu/drm/i915/i915_vma.c @@ -852,13 +852,19 @@ static void vma_unbind_pages(struct i915_vma *vma) __vma_put_pages(vma, count | count << I915_VMA_PAGES_BIAS); } -int i915_vma_pin(struct i915_vma *vma, u64 size, u64 alignment, u64 flags) +int i915_vma_pin_ww(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, + u64 size, u64 alignment, u64 flags) { struct i915_vma_work *work = NULL; intel_wakeref_t wakeref = 0; unsigned int bound; int err; +#ifdef CONFIG_PROVE_LOCKING + if (debug_locks && lockdep_is_held(&vma->vm->i915->drm.struct_mutex)) + WARN_ON(!ww); +#endif + BUILD_BUG_ON(PIN_GLOBAL != I915_VMA_GLOBAL_BIND); BUILD_BUG_ON(PIN_USER != I915_VMA_LOCAL_BIND); @@ -1002,7 +1008,8 @@ static void flush_idle_contexts(struct intel_gt *gt) intel_gt_wait_for_idle(gt, MAX_SCHEDULE_TIMEOUT); } -int i915_ggtt_pin(struct i915_vma *vma, u32 align, unsigned int flags) +int i915_ggtt_pin(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, + u32 align, unsigned int flags) { struct i915_address_space *vm = vma->vm; int err; @@ -1010,7 +1017,7 @@ int i915_ggtt_pin(struct i915_vma *vma, u32 align, unsigned int flags) GEM_BUG_ON(!i915_vma_is_ggtt(vma)); do { - err = i915_vma_pin(vma, 0, align, flags | PIN_GLOBAL); + err = i915_vma_pin_ww(vma, ww, 0, align, flags | PIN_GLOBAL); if (err != -ENOSPC) { if (!err) { err = i915_vma_wait_for_bind(vma); diff --git a/drivers/gpu/drm/i915/i915_vma.h b/drivers/gpu/drm/i915/i915_vma.h index 4b325a670a04..5345ba10b64f 100644 --- a/drivers/gpu/drm/i915/i915_vma.h +++ b/drivers/gpu/drm/i915/i915_vma.h @@ -240,8 +240,17 @@ static inline void i915_vma_unlock(struct i915_vma *vma) } int __must_check -i915_vma_pin(struct i915_vma *vma, u64 size, u64 alignment, u64 flags); -int i915_ggtt_pin(struct i915_vma *vma, u32 align, unsigned int flags); +i915_vma_pin_ww(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, + u64 size, u64 alignment, u64 flags); + +static inline int __must_check +i915_vma_pin(struct i915_vma *vma, u64 size, u64 alignment, u64 flags) +{ + return i915_vma_pin_ww(vma, NULL, size, alignment, flags); +} + +int i915_ggtt_pin(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, + u32 align, unsigned int flags); static inline int i915_vma_pin_count(const struct i915_vma *vma) { From patchwork Mon Aug 10 10:30:55 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707273 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 660B313A4 for ; Mon, 10 Aug 2020 10:33:44 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 4E234206E9 for ; Mon, 10 Aug 2020 10:33:44 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 4E234206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 0933B6E3F0; Mon, 10 Aug 2020 10:33:31 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 63DD189FC9 for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:55 +0200 Message-Id: <20200810103103.303818-17-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 16/24] drm/i915: Convert i915_gem_object/client_blt.c to use ww locking as well, v2. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" This is the last part outside of selftests that still don't use the correct lock ordering of timeline->mutex vs resv_lock. With gem fixed, there are a few places that still get locking wrong: - gvt/scheduler.c - i915_perf.c - Most if not all selftests. Changes since v1: - Add intel_engine_pm_get/put() calls to fix use-after-free when using intel_engine_get_pool(). Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- .../gpu/drm/i915/gem/i915_gem_client_blt.c | 78 +++++++-- .../gpu/drm/i915/gem/i915_gem_object_blt.c | 152 ++++++++++++------ .../gpu/drm/i915/gem/i915_gem_object_blt.h | 3 + 3 files changed, 163 insertions(+), 70 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c b/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c index a85907f29c53..272cf3ea68d5 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_client_blt.c @@ -158,6 +158,7 @@ static void clear_pages_worker(struct work_struct *work) struct clear_pages_work *w = container_of(work, typeof(*w), work); struct drm_i915_gem_object *obj = w->sleeve->vma->obj; struct i915_vma *vma = w->sleeve->vma; + struct i915_gem_ww_ctx ww; struct i915_request *rq; struct i915_vma *batch; int err = w->dma.error; @@ -173,17 +174,20 @@ static void clear_pages_worker(struct work_struct *work) obj->read_domains = I915_GEM_GPU_DOMAINS; obj->write_domain = 0; - err = i915_vma_pin(vma, 0, 0, PIN_USER); - if (unlikely(err)) + i915_gem_ww_ctx_init(&ww, false); + intel_engine_pm_get(w->ce->engine); +retry: + err = intel_context_pin_ww(w->ce, &ww); + if (err) goto out_signal; - batch = intel_emit_vma_fill_blt(w->ce, vma, w->value); + batch = intel_emit_vma_fill_blt(w->ce, vma, &ww, w->value); if (IS_ERR(batch)) { err = PTR_ERR(batch); - goto out_unpin; + goto out_ctx; } - rq = intel_context_create_request(w->ce); + rq = i915_request_create(w->ce); if (IS_ERR(rq)) { err = PTR_ERR(rq); goto out_batch; @@ -225,9 +229,19 @@ static void clear_pages_worker(struct work_struct *work) i915_request_add(rq); out_batch: intel_emit_vma_release(w->ce, batch); -out_unpin: - i915_vma_unpin(vma); +out_ctx: + intel_context_unpin(w->ce); out_signal: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + + i915_vma_unpin(w->sleeve->vma); + intel_engine_pm_put(w->ce->engine); + if (unlikely(err)) { dma_fence_set_error(&w->dma, err); dma_fence_signal(&w->dma); @@ -235,6 +249,44 @@ static void clear_pages_worker(struct work_struct *work) } } +static int pin_wait_clear_pages_work(struct clear_pages_work *w, + struct intel_context *ce) +{ + struct i915_vma *vma = w->sleeve->vma; + struct i915_gem_ww_ctx ww; + int err; + + i915_gem_ww_ctx_init(&ww, false); +retry: + err = i915_gem_object_lock(vma->obj, &ww); + if (err) + goto out; + + err = i915_vma_pin_ww(vma, &ww, 0, 0, PIN_USER); + if (unlikely(err)) + goto out; + + err = i915_sw_fence_await_reservation(&w->wait, + vma->obj->base.resv, NULL, + true, 0, I915_FENCE_GFP); + if (err) + goto err_unpin_vma; + + dma_resv_add_excl_fence(vma->obj->base.resv, &w->dma); + +err_unpin_vma: + if (err) + i915_vma_unpin(vma); +out: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + return err; +} + static int __i915_sw_fence_call clear_pages_work_notify(struct i915_sw_fence *fence, enum i915_sw_fence_notify state) @@ -288,17 +340,9 @@ int i915_gem_schedule_fill_pages_blt(struct drm_i915_gem_object *obj, dma_fence_init(&work->dma, &clear_pages_work_ops, &fence_lock, 0, 0); i915_sw_fence_init(&work->wait, clear_pages_work_notify); - i915_gem_object_lock(obj, NULL); - err = i915_sw_fence_await_reservation(&work->wait, - obj->base.resv, NULL, true, 0, - I915_FENCE_GFP); - if (err < 0) { + err = pin_wait_clear_pages_work(work, ce); + if (err < 0) dma_fence_set_error(&work->dma, err); - } else { - dma_resv_add_excl_fence(obj->base.resv, &work->dma); - err = 0; - } - i915_gem_object_unlock(obj); dma_fence_get(&work->dma); i915_sw_fence_commit(&work->wait); diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c index bfdb32d46877..d93eb36160c9 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.c @@ -14,6 +14,7 @@ struct i915_vma *intel_emit_vma_fill_blt(struct intel_context *ce, struct i915_vma *vma, + struct i915_gem_ww_ctx *ww, u32 value) { struct drm_i915_private *i915 = ce->vm->i915; @@ -39,10 +40,24 @@ struct i915_vma *intel_emit_vma_fill_blt(struct intel_context *ce, goto out_pm; } + err = i915_gem_object_lock(pool->obj, ww); + if (err) + goto out_put; + + batch = i915_vma_instance(pool->obj, ce->vm, NULL); + if (IS_ERR(batch)) { + err = PTR_ERR(batch); + goto out_put; + } + + err = i915_vma_pin_ww(batch, ww, 0, 0, PIN_USER); + if (unlikely(err)) + goto out_put; + cmd = i915_gem_object_pin_map(pool->obj, I915_MAP_WC); if (IS_ERR(cmd)) { err = PTR_ERR(cmd); - goto out_put; + goto out_unpin; } rem = vma->size; @@ -84,19 +99,11 @@ struct i915_vma *intel_emit_vma_fill_blt(struct intel_context *ce, intel_gt_chipset_flush(ce->vm->gt); - batch = i915_vma_instance(pool->obj, ce->vm, NULL); - if (IS_ERR(batch)) { - err = PTR_ERR(batch); - goto out_put; - } - - err = i915_vma_pin(batch, 0, 0, PIN_USER); - if (unlikely(err)) - goto out_put; - batch->private = pool; return batch; +out_unpin: + i915_vma_unpin(batch); out_put: intel_gt_buffer_pool_put(pool); out_pm: @@ -108,11 +115,9 @@ int intel_emit_vma_mark_active(struct i915_vma *vma, struct i915_request *rq) { int err; - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, false); if (err == 0) err = i915_vma_move_to_active(vma, rq, 0); - i915_vma_unlock(vma); if (unlikely(err)) return err; @@ -141,6 +146,7 @@ int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj, struct intel_context *ce, u32 value) { + struct i915_gem_ww_ctx ww; struct i915_request *rq; struct i915_vma *batch; struct i915_vma *vma; @@ -150,17 +156,28 @@ int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj, if (IS_ERR(vma)) return PTR_ERR(vma); - err = i915_vma_pin(vma, 0, 0, PIN_USER); - if (unlikely(err)) - return err; + i915_gem_ww_ctx_init(&ww, true); + intel_engine_pm_get(ce->engine); +retry: + err = i915_gem_object_lock(obj, &ww); + if (err) + goto out; - batch = intel_emit_vma_fill_blt(ce, vma, value); + err = intel_context_pin_ww(ce, &ww); + if (err) + goto out; + + err = i915_vma_pin_ww(vma, &ww, 0, 0, PIN_USER); + if (err) + goto out_ctx; + + batch = intel_emit_vma_fill_blt(ce, vma, &ww, value); if (IS_ERR(batch)) { err = PTR_ERR(batch); - goto out_unpin; + goto out_vma; } - rq = intel_context_create_request(ce); + rq = i915_request_create(ce); if (IS_ERR(rq)) { err = PTR_ERR(rq); goto out_batch; @@ -170,11 +187,9 @@ int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj, if (unlikely(err)) goto out_request; - i915_vma_lock(vma); err = move_obj_to_gpu(vma->obj, rq, true); if (err == 0) err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); - i915_vma_unlock(vma); if (unlikely(err)) goto out_request; @@ -193,8 +208,18 @@ int i915_gem_object_fill_blt(struct drm_i915_gem_object *obj, i915_request_add(rq); out_batch: intel_emit_vma_release(ce, batch); -out_unpin: +out_vma: i915_vma_unpin(vma); +out_ctx: + intel_context_unpin(ce); +out: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + intel_engine_pm_put(ce->engine); return err; } @@ -210,6 +235,7 @@ static bool wa_1209644611_applies(struct drm_i915_private *i915, u32 size) } struct i915_vma *intel_emit_vma_copy_blt(struct intel_context *ce, + struct i915_gem_ww_ctx *ww, struct i915_vma *src, struct i915_vma *dst) { @@ -236,10 +262,24 @@ struct i915_vma *intel_emit_vma_copy_blt(struct intel_context *ce, goto out_pm; } + err = i915_gem_object_lock(pool->obj, ww); + if (err) + goto out_put; + + batch = i915_vma_instance(pool->obj, ce->vm, NULL); + if (IS_ERR(batch)) { + err = PTR_ERR(batch); + goto out_put; + } + + err = i915_vma_pin_ww(batch, ww, 0, 0, PIN_USER); + if (unlikely(err)) + goto out_put; + cmd = i915_gem_object_pin_map(pool->obj, I915_MAP_WC); if (IS_ERR(cmd)) { err = PTR_ERR(cmd); - goto out_put; + goto out_unpin; } rem = src->size; @@ -296,20 +336,11 @@ struct i915_vma *intel_emit_vma_copy_blt(struct intel_context *ce, i915_gem_object_unpin_map(pool->obj); intel_gt_chipset_flush(ce->vm->gt); - - batch = i915_vma_instance(pool->obj, ce->vm, NULL); - if (IS_ERR(batch)) { - err = PTR_ERR(batch); - goto out_put; - } - - err = i915_vma_pin(batch, 0, 0, PIN_USER); - if (unlikely(err)) - goto out_put; - batch->private = pool; return batch; +out_unpin: + i915_vma_unpin(batch); out_put: intel_gt_buffer_pool_put(pool); out_pm: @@ -321,10 +352,9 @@ int i915_gem_object_copy_blt(struct drm_i915_gem_object *src, struct drm_i915_gem_object *dst, struct intel_context *ce) { - struct drm_gem_object *objs[] = { &src->base, &dst->base }; struct i915_address_space *vm = ce->vm; struct i915_vma *vma[2], *batch; - struct ww_acquire_ctx acquire; + struct i915_gem_ww_ctx ww; struct i915_request *rq; int err, i; @@ -332,25 +362,36 @@ int i915_gem_object_copy_blt(struct drm_i915_gem_object *src, if (IS_ERR(vma[0])) return PTR_ERR(vma[0]); - err = i915_vma_pin(vma[0], 0, 0, PIN_USER); - if (unlikely(err)) - return err; - vma[1] = i915_vma_instance(dst, vm, NULL); if (IS_ERR(vma[1])) - goto out_unpin_src; + return PTR_ERR(vma); - err = i915_vma_pin(vma[1], 0, 0, PIN_USER); + i915_gem_ww_ctx_init(&ww, true); + intel_engine_pm_get(ce->engine); +retry: + err = i915_gem_object_lock(src, &ww); + if (!err) + err = i915_gem_object_lock(dst, &ww); + if (!err) + err = intel_context_pin_ww(ce, &ww); + if (err) + goto out; + + err = i915_vma_pin_ww(vma[0], &ww, 0, 0, PIN_USER); + if (err) + goto out_ctx; + + err = i915_vma_pin_ww(vma[1], &ww, 0, 0, PIN_USER); if (unlikely(err)) goto out_unpin_src; - batch = intel_emit_vma_copy_blt(ce, vma[0], vma[1]); + batch = intel_emit_vma_copy_blt(ce, &ww, vma[0], vma[1]); if (IS_ERR(batch)) { err = PTR_ERR(batch); goto out_unpin_dst; } - rq = intel_context_create_request(ce); + rq = i915_request_create(ce); if (IS_ERR(rq)) { err = PTR_ERR(rq); goto out_batch; @@ -360,14 +401,10 @@ int i915_gem_object_copy_blt(struct drm_i915_gem_object *src, if (unlikely(err)) goto out_request; - err = drm_gem_lock_reservations(objs, ARRAY_SIZE(objs), &acquire); - if (unlikely(err)) - goto out_request; - for (i = 0; i < ARRAY_SIZE(vma); i++) { err = move_obj_to_gpu(vma[i]->obj, rq, i); if (unlikely(err)) - goto out_unlock; + goto out_request; } for (i = 0; i < ARRAY_SIZE(vma); i++) { @@ -375,20 +412,19 @@ int i915_gem_object_copy_blt(struct drm_i915_gem_object *src, err = i915_vma_move_to_active(vma[i], rq, flags); if (unlikely(err)) - goto out_unlock; + goto out_request; } if (rq->engine->emit_init_breadcrumb) { err = rq->engine->emit_init_breadcrumb(rq); if (unlikely(err)) - goto out_unlock; + goto out_request; } err = rq->engine->emit_bb_start(rq, batch->node.start, batch->node.size, 0); -out_unlock: - drm_gem_unlock_reservations(objs, ARRAY_SIZE(objs), &acquire); + out_request: if (unlikely(err)) i915_request_set_error_once(rq, err); @@ -400,6 +436,16 @@ int i915_gem_object_copy_blt(struct drm_i915_gem_object *src, i915_vma_unpin(vma[1]); out_unpin_src: i915_vma_unpin(vma[0]); +out_ctx: + intel_context_unpin(ce); +out: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + intel_engine_pm_put(ce->engine); return err; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_blt.h b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.h index 8bcd336a90dc..2409fdcccf0e 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object_blt.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_object_blt.h @@ -13,12 +13,15 @@ #include "i915_vma.h" struct drm_i915_gem_object; +struct i915_gem_ww_ctx; struct i915_vma *intel_emit_vma_fill_blt(struct intel_context *ce, struct i915_vma *vma, + struct i915_gem_ww_ctx *ww, u32 value); struct i915_vma *intel_emit_vma_copy_blt(struct intel_context *ce, + struct i915_gem_ww_ctx *ww, struct i915_vma *src, struct i915_vma *dst); From patchwork Mon Aug 10 10:30:56 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707287 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id AF4AD13B6 for ; Mon, 10 Aug 2020 10:33:49 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 97B1E206E9 for ; Mon, 10 Aug 2020 10:33:49 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 97B1E206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 640226E3FE; Mon, 10 Aug 2020 10:33:37 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 5D7B0897E0 for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:56 +0200 Message-Id: <20200810103103.303818-18-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 17/24] drm/i915: Kill last user of intel_context_create_request outside of selftests 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Instead of using intel_context_create_request(), use intel_context_pin() and i915_create_request directly. Now all those calls are gone outside of selftests. :) Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- drivers/gpu/drm/i915/gt/intel_workarounds.c | 43 ++++++++++++++------- 1 file changed, 29 insertions(+), 14 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds.c b/drivers/gpu/drm/i915/gt/intel_workarounds.c index cef1c122696f..5737d851fb4c 100644 --- a/drivers/gpu/drm/i915/gt/intel_workarounds.c +++ b/drivers/gpu/drm/i915/gt/intel_workarounds.c @@ -2070,6 +2070,7 @@ static int engine_wa_list_verify(struct intel_context *ce, const struct i915_wa *wa; struct i915_request *rq; struct i915_vma *vma; + struct i915_gem_ww_ctx ww; unsigned int i; u32 *results; int err; @@ -2082,29 +2083,34 @@ static int engine_wa_list_verify(struct intel_context *ce, return PTR_ERR(vma); intel_engine_pm_get(ce->engine); - rq = intel_context_create_request(ce); - intel_engine_pm_put(ce->engine); + i915_gem_ww_ctx_init(&ww, false); +retry: + err = i915_gem_object_lock(vma->obj, &ww); + if (err == 0) + err = intel_context_pin_ww(ce, &ww); + if (err) + goto err_pm; + + rq = i915_request_create(ce); if (IS_ERR(rq)) { err = PTR_ERR(rq); - goto err_vma; + goto err_unpin; } - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, true); if (err == 0) err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); - i915_vma_unlock(vma); - if (err) { - i915_request_add(rq); - goto err_vma; - } - - err = wa_list_srm(rq, wal, vma); - if (err) - goto err_vma; + if (err == 0) + err = wa_list_srm(rq, wal, vma); i915_request_get(rq); + if (err) + i915_request_set_error_once(rq, err); i915_request_add(rq); + + if (err) + goto err_rq; + if (i915_request_wait(rq, 0, HZ / 5) < 0) { err = -ETIME; goto err_rq; @@ -2129,7 +2135,16 @@ static int engine_wa_list_verify(struct intel_context *ce, err_rq: i915_request_put(rq); -err_vma: +err_unpin: + intel_context_unpin(ce); +err_pm: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + intel_engine_pm_put(ce->engine); i915_vma_unpin(vma); i915_vma_put(vma); return err; From patchwork Mon Aug 10 10:30:57 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707259 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 06A561392 for ; Mon, 10 Aug 2020 10:33:37 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id E3046206E9 for ; Mon, 10 Aug 2020 10:33:36 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org E3046206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 1C475897E0; Mon, 10 Aug 2020 10:33:30 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 645B36E0E3 for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:57 +0200 Message-Id: <20200810103103.303818-19-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 18/24] drm/i915: Convert i915_perf to ww locking as well 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" We have the ordering of timeline->mutex vs resv_lock wrong, convert the i915_pin_vma and intel_context_pin as well to future-proof this. We may need to do future changes to do this more transaction-like, and only get down to a single i915_gem_ww_ctx, but for now this should work. Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- drivers/gpu/drm/i915/i915_perf.c | 57 +++++++++++++++++++++++--------- 1 file changed, 42 insertions(+), 15 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_perf.c b/drivers/gpu/drm/i915/i915_perf.c index c6f6370283cf..e94976976571 100644 --- a/drivers/gpu/drm/i915/i915_perf.c +++ b/drivers/gpu/drm/i915/i915_perf.c @@ -1195,24 +1195,39 @@ static struct intel_context *oa_pin_context(struct i915_perf_stream *stream) struct i915_gem_engines_iter it; struct i915_gem_context *ctx = stream->ctx; struct intel_context *ce; - int err; + struct i915_gem_ww_ctx ww; + int err = -ENODEV; for_each_gem_engine(ce, i915_gem_context_lock_engines(ctx), it) { if (ce->engine != stream->engine) /* first match! */ continue; - /* - * As the ID is the gtt offset of the context's vma we - * pin the vma to ensure the ID remains fixed. - */ - err = intel_context_pin(ce); - if (err == 0) { - stream->pinned_ctx = ce; - break; - } + err = 0; + break; } i915_gem_context_unlock_engines(ctx); + if (err) + return ERR_PTR(err); + + i915_gem_ww_ctx_init(&ww, true); +retry: + /* + * As the ID is the gtt offset of the context's vma we + * pin the vma to ensure the ID remains fixed. + */ + err = intel_context_pin_ww(ce, &ww); + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + + if (err) + return ERR_PTR(err); + + stream->pinned_ctx = ce; return stream->pinned_ctx; } @@ -1923,15 +1938,22 @@ emit_oa_config(struct i915_perf_stream *stream, { struct i915_request *rq; struct i915_vma *vma; + struct i915_gem_ww_ctx ww; int err; vma = get_oa_vma(stream, oa_config); if (IS_ERR(vma)) return PTR_ERR(vma); - err = i915_vma_pin(vma, 0, 0, PIN_GLOBAL | PIN_HIGH); + i915_gem_ww_ctx_init(&ww, true); +retry: + err = i915_gem_object_lock(vma->obj, &ww); + if (err) + goto err; + + err = i915_vma_pin_ww(vma, &ww, 0, 0, PIN_GLOBAL | PIN_HIGH); if (err) - goto err_vma_put; + goto err; intel_engine_pm_get(ce->engine); rq = i915_request_create(ce); @@ -1953,11 +1975,9 @@ emit_oa_config(struct i915_perf_stream *stream, goto err_add_request; } - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, 0); if (!err) err = i915_vma_move_to_active(vma, rq, 0); - i915_vma_unlock(vma); if (err) goto err_add_request; @@ -1971,7 +1991,14 @@ emit_oa_config(struct i915_perf_stream *stream, i915_request_add(rq); err_vma_unpin: i915_vma_unpin(vma); -err_vma_put: +err: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + + i915_gem_ww_ctx_fini(&ww); i915_vma_put(vma); return err; } From patchwork Mon Aug 10 10:30:58 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707301 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id A28E113A4 for ; Mon, 10 Aug 2020 10:33:53 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 8A957206E9 for ; Mon, 10 Aug 2020 10:33:53 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 8A957206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id E0E2E6E40A; Mon, 10 Aug 2020 10:33:44 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 686F16E3E5 for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:58 +0200 Message-Id: <20200810103103.303818-20-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 19/24] drm/i915: Dirty hack to fix selftests locking inversion 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Some i915 selftests still use i915_vma_lock() as inner lock, and intel_context_create_request() intel_timeline->mutex as outer lock. Fortunately for selftests this is not an issue, they should be fixed but we can move ahead and cleanify lockdep now. Signed-off-by: Maarten Lankhorst Acked-by: Thomas Hellström --- drivers/gpu/drm/i915/gt/intel_context.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/drivers/gpu/drm/i915/gt/intel_context.c b/drivers/gpu/drm/i915/gt/intel_context.c index c05ef213bdc2..671081b0e4d6 100644 --- a/drivers/gpu/drm/i915/gt/intel_context.c +++ b/drivers/gpu/drm/i915/gt/intel_context.c @@ -459,6 +459,18 @@ struct i915_request *intel_context_create_request(struct intel_context *ce) rq = i915_request_create(ce); intel_context_unpin(ce); + if (IS_ERR(rq)) + return rq; + + /* + * timeline->mutex should be the inner lock, but is used as outer lock. + * Hack around this to shut up lockdep in selftests.. + */ + lockdep_unpin_lock(&ce->timeline->mutex, rq->cookie); + mutex_release(&ce->timeline->mutex.dep_map, _RET_IP_); + mutex_acquire(&ce->timeline->mutex.dep_map, SINGLE_DEPTH_NESTING, 0, _RET_IP_); + rq->cookie = lockdep_pin_lock(&ce->timeline->mutex); + return rq; } From patchwork Mon Aug 10 10:30:59 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707261 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id A5AC51392 for ; Mon, 10 Aug 2020 10:33:38 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 8A489206E9 for ; Mon, 10 Aug 2020 10:33:38 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 8A489206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 246B289C3B; Mon, 10 Aug 2020 10:33:30 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 67A816E3D0 for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:30:59 +0200 Message-Id: <20200810103103.303818-21-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 20/24] drm/i915/selftests: Fix locking inversion in lrc selftest. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" This function does not use intel_context_create_request, so it has to use the same locking order as normal code. This is required to shut up lockdep in selftests. Signed-off-by: Maarten Lankhorst --- drivers/gpu/drm/i915/gt/selftest_lrc.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/selftest_lrc.c b/drivers/gpu/drm/i915/gt/selftest_lrc.c index 7faba9f91224..95d41c01d0e0 100644 --- a/drivers/gpu/drm/i915/gt/selftest_lrc.c +++ b/drivers/gpu/drm/i915/gt/selftest_lrc.c @@ -4996,6 +4996,7 @@ static int __live_lrc_state(struct intel_engine_cs *engine, { struct intel_context *ce; struct i915_request *rq; + struct i915_gem_ww_ctx ww; enum { RING_START_IDX = 0, RING_TAIL_IDX, @@ -5010,7 +5011,11 @@ static int __live_lrc_state(struct intel_engine_cs *engine, if (IS_ERR(ce)) return PTR_ERR(ce); - err = intel_context_pin(ce); + i915_gem_ww_ctx_init(&ww, false); +retry: + err = i915_gem_object_lock(scratch->obj, &ww); + if (!err) + err = intel_context_pin_ww(ce, &ww); if (err) goto err_put; @@ -5039,11 +5044,9 @@ static int __live_lrc_state(struct intel_engine_cs *engine, *cs++ = i915_ggtt_offset(scratch) + RING_TAIL_IDX * sizeof(u32); *cs++ = 0; - i915_vma_lock(scratch); err = i915_request_await_object(rq, scratch->obj, true); if (!err) err = i915_vma_move_to_active(scratch, rq, EXEC_OBJECT_WRITE); - i915_vma_unlock(scratch); i915_request_get(rq); i915_request_add(rq); @@ -5080,6 +5083,12 @@ static int __live_lrc_state(struct intel_engine_cs *engine, err_unpin: intel_context_unpin(ce); err_put: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); intel_context_put(ce); return err; } From patchwork Mon Aug 10 10:31:00 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707291 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id E647B13A4 for ; Mon, 10 Aug 2020 10:33:50 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id CE8A1206E9 for ; Mon, 10 Aug 2020 10:33:50 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org CE8A1206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 6D88B6E402; Mon, 10 Aug 2020 10:33:38 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 8389A89C3B for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:31:00 +0200 Message-Id: <20200810103103.303818-22-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 21/24] drm/i915: Use ww pinning for intel_context_create_request() 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" We want to get rid of intel_context_pin(), convert intel_context_create_request() first. :) Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- drivers/gpu/drm/i915/gt/intel_context.c | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/intel_context.c b/drivers/gpu/drm/i915/gt/intel_context.c index 671081b0e4d6..61b05cd4c47a 100644 --- a/drivers/gpu/drm/i915/gt/intel_context.c +++ b/drivers/gpu/drm/i915/gt/intel_context.c @@ -449,15 +449,25 @@ int intel_context_prepare_remote_request(struct intel_context *ce, struct i915_request *intel_context_create_request(struct intel_context *ce) { + struct i915_gem_ww_ctx ww; struct i915_request *rq; int err; - err = intel_context_pin(ce); - if (unlikely(err)) - return ERR_PTR(err); + i915_gem_ww_ctx_init(&ww, true); +retry: + err = intel_context_pin_ww(ce, &ww); + if (!err) { + rq = i915_request_create(ce); + intel_context_unpin(ce); + } else if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } else { + rq = ERR_PTR(err); + } - rq = i915_request_create(ce); - intel_context_unpin(ce); + i915_gem_ww_ctx_fini(&ww); if (IS_ERR(rq)) return rq; From patchwork Mon Aug 10 10:31:01 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707265 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id EBC9C13A4 for ; Mon, 10 Aug 2020 10:33:40 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id D3BEF206E9 for ; Mon, 10 Aug 2020 10:33:40 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org D3BEF206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 64F0089CDE; Mon, 10 Aug 2020 10:33:30 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 86CF16E3EE for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:31:01 +0200 Message-Id: <20200810103103.303818-23-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 22/24] drm/i915: Move i915_vma_lock in the selftests to avoid lock inversion, v3. 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Make sure vma_lock is not used as inner lock when kernel context is used, and add ww handling where appropriate. Ensure that execbuf selftests keep passing by using ww handling. Changes since v2: - Fix i915_gem_context finally. Signed-off-by: Maarten Lankhorst Reviewed-by: Thomas Hellström --- .../i915/gem/selftests/i915_gem_coherency.c | 26 +++-- .../drm/i915/gem/selftests/i915_gem_context.c | 106 +++++++++--------- .../drm/i915/gem/selftests/i915_gem_mman.c | 41 +++++-- drivers/gpu/drm/i915/gt/selftest_rps.c | 30 +++-- drivers/gpu/drm/i915/selftests/i915_request.c | 18 ++- 5 files changed, 125 insertions(+), 96 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c index dcdfc396f2f8..7049a6bbc03d 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c @@ -201,25 +201,25 @@ static int gpu_set(struct context *ctx, unsigned long offset, u32 v) i915_gem_object_lock(ctx->obj, NULL); err = i915_gem_object_set_to_gtt_domain(ctx->obj, true); - i915_gem_object_unlock(ctx->obj); if (err) - return err; + goto out_unlock; vma = i915_gem_object_ggtt_pin(ctx->obj, NULL, 0, 0, 0); - if (IS_ERR(vma)) - return PTR_ERR(vma); + if (IS_ERR(vma)) { + err = PTR_ERR(vma); + goto out_unlock; + } rq = intel_engine_create_kernel_request(ctx->engine); if (IS_ERR(rq)) { - i915_vma_unpin(vma); - return PTR_ERR(rq); + err = PTR_ERR(rq); + goto out_unpin; } cs = intel_ring_begin(rq, 4); if (IS_ERR(cs)) { - i915_request_add(rq); - i915_vma_unpin(vma); - return PTR_ERR(cs); + err = PTR_ERR(cs); + goto out_rq; } if (INTEL_GEN(ctx->engine->i915) >= 8) { @@ -240,14 +240,16 @@ static int gpu_set(struct context *ctx, unsigned long offset, u32 v) } intel_ring_advance(rq, cs); - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, true); if (err == 0) err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); - i915_vma_unlock(vma); - i915_vma_unpin(vma); +out_rq: i915_request_add(rq); +out_unpin: + i915_vma_unpin(vma); +out_unlock: + i915_gem_object_unlock(ctx->obj); return err; } 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 fa40006b453a..99becb86abd3 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c @@ -893,24 +893,15 @@ static int igt_shared_ctx_exec(void *arg) return err; } -static struct i915_vma *rpcs_query_batch(struct i915_vma *vma) +static int rpcs_query_batch(struct drm_i915_gem_object *rpcs, struct i915_vma *vma) { - struct drm_i915_gem_object *obj; u32 *cmd; - int err; - if (INTEL_GEN(vma->vm->i915) < 8) - return ERR_PTR(-EINVAL); + GEM_BUG_ON(INTEL_GEN(vma->vm->i915) < 8); - obj = i915_gem_object_create_internal(vma->vm->i915, PAGE_SIZE); - if (IS_ERR(obj)) - return ERR_CAST(obj); - - cmd = i915_gem_object_pin_map(obj, I915_MAP_WB); - if (IS_ERR(cmd)) { - err = PTR_ERR(cmd); - goto err; - } + cmd = i915_gem_object_pin_map(rpcs, I915_MAP_WB); + if (IS_ERR(cmd)) + return PTR_ERR(cmd); *cmd++ = MI_STORE_REGISTER_MEM_GEN8; *cmd++ = i915_mmio_reg_offset(GEN8_R_PWR_CLK_STATE); @@ -918,26 +909,12 @@ static struct i915_vma *rpcs_query_batch(struct i915_vma *vma) *cmd++ = upper_32_bits(vma->node.start); *cmd = MI_BATCH_BUFFER_END; - __i915_gem_object_flush_map(obj, 0, 64); - i915_gem_object_unpin_map(obj); + __i915_gem_object_flush_map(rpcs, 0, 64); + i915_gem_object_unpin_map(rpcs); intel_gt_chipset_flush(vma->vm->gt); - vma = i915_vma_instance(obj, vma->vm, NULL); - if (IS_ERR(vma)) { - err = PTR_ERR(vma); - goto err; - } - - err = i915_vma_pin(vma, 0, 0, PIN_USER); - if (err) - goto err; - - return vma; - -err: - i915_gem_object_put(obj); - return ERR_PTR(err); + return 0; } static int @@ -945,52 +922,68 @@ emit_rpcs_query(struct drm_i915_gem_object *obj, struct intel_context *ce, struct i915_request **rq_out) { + struct drm_i915_private *i915 = to_i915(obj->base.dev); struct i915_request *rq; + struct i915_gem_ww_ctx ww; struct i915_vma *batch; struct i915_vma *vma; + struct drm_i915_gem_object *rpcs; int err; GEM_BUG_ON(!intel_engine_can_store_dword(ce->engine)); + if (INTEL_GEN(i915) < 8) + return -EINVAL; + vma = i915_vma_instance(obj, ce->vm, NULL); if (IS_ERR(vma)) return PTR_ERR(vma); - i915_gem_object_lock(obj, NULL); - err = i915_gem_object_set_to_gtt_domain(obj, false); - i915_gem_object_unlock(obj); - if (err) - return err; + rpcs = i915_gem_object_create_internal(i915, PAGE_SIZE); + if (IS_ERR(rpcs)) + return PTR_ERR(rpcs); - err = i915_vma_pin(vma, 0, 0, PIN_USER); - if (err) - return err; - - batch = rpcs_query_batch(vma); + batch = i915_vma_instance(rpcs, ce->vm, NULL); if (IS_ERR(batch)) { err = PTR_ERR(batch); - goto err_vma; + goto err_put; } + i915_gem_ww_ctx_init(&ww, false); +retry: + err = i915_gem_object_lock(obj, &ww); + if (!err) + err = i915_gem_object_lock(rpcs, &ww); + if (!err) + err = i915_gem_object_set_to_gtt_domain(obj, false); + if (!err) + err = i915_vma_pin_ww(vma, &ww, 0, 0, PIN_USER); + if (err) + goto err_put; + + err = i915_vma_pin_ww(batch, &ww, 0, 0, PIN_USER); + if (err) + goto err_vma; + + err = rpcs_query_batch(rpcs, vma); + if (err) + goto err_batch; + rq = i915_request_create(ce); if (IS_ERR(rq)) { err = PTR_ERR(rq); goto err_batch; } - i915_vma_lock(batch); err = i915_request_await_object(rq, batch->obj, false); if (err == 0) err = i915_vma_move_to_active(batch, rq, 0); - i915_vma_unlock(batch); if (err) goto skip_request; - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, true); if (err == 0) err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); - i915_vma_unlock(vma); if (err) goto skip_request; @@ -1006,23 +999,24 @@ emit_rpcs_query(struct drm_i915_gem_object *obj, if (err) goto skip_request; - i915_vma_unpin_and_release(&batch, 0); - i915_vma_unpin(vma); - *rq_out = i915_request_get(rq); - i915_request_add(rq); - - return 0; - skip_request: - i915_request_set_error_once(rq, err); + if (err) + i915_request_set_error_once(rq, err); i915_request_add(rq); err_batch: - i915_vma_unpin_and_release(&batch, 0); + i915_vma_unpin(batch); err_vma: i915_vma_unpin(vma); - +err_put: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + i915_gem_object_put(rpcs); return err; } 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 9fb95a45bcad..d27d87a678c8 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c @@ -528,31 +528,42 @@ static int make_obj_busy(struct drm_i915_gem_object *obj) for_each_uabi_engine(engine, i915) { struct i915_request *rq; struct i915_vma *vma; + struct i915_gem_ww_ctx ww; int err; vma = i915_vma_instance(obj, &engine->gt->ggtt->vm, NULL); if (IS_ERR(vma)) return PTR_ERR(vma); - err = i915_vma_pin(vma, 0, 0, PIN_USER); + i915_gem_ww_ctx_init(&ww, false); +retry: + err = i915_gem_object_lock(obj, &ww); + if (!err) + err = i915_vma_pin_ww(vma, &ww, 0, 0, PIN_USER); if (err) - return err; + goto err; rq = intel_engine_create_kernel_request(engine); if (IS_ERR(rq)) { - i915_vma_unpin(vma); - return PTR_ERR(rq); + err = PTR_ERR(rq); + goto err_unpin; } - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, true); if (err == 0) err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); - i915_vma_unlock(vma); i915_request_add(rq); +err_unpin: i915_vma_unpin(vma); +err: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); if (err) return err; } @@ -1123,6 +1134,7 @@ static int __igt_mmap_gpu(struct drm_i915_private *i915, for_each_uabi_engine(engine, i915) { struct i915_request *rq; struct i915_vma *vma; + struct i915_gem_ww_ctx ww; vma = i915_vma_instance(obj, engine->kernel_context->vm, NULL); if (IS_ERR(vma)) { @@ -1130,9 +1142,13 @@ static int __igt_mmap_gpu(struct drm_i915_private *i915, goto out_unmap; } - err = i915_vma_pin(vma, 0, 0, PIN_USER); + i915_gem_ww_ctx_init(&ww, false); +retry: + err = i915_gem_object_lock(obj, &ww); + if (!err) + err = i915_vma_pin_ww(vma, &ww, 0, 0, PIN_USER); if (err) - goto out_unmap; + goto out_ww; rq = i915_request_create(engine->kernel_context); if (IS_ERR(rq)) { @@ -1140,11 +1156,9 @@ static int __igt_mmap_gpu(struct drm_i915_private *i915, goto out_unpin; } - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, false); if (err == 0) err = i915_vma_move_to_active(vma, rq, 0); - i915_vma_unlock(vma); err = engine->emit_bb_start(rq, vma->node.start, 0, 0); i915_request_get(rq); @@ -1166,6 +1180,13 @@ static int __igt_mmap_gpu(struct drm_i915_private *i915, out_unpin: i915_vma_unpin(vma); +out_ww: + if (err == -EDEADLK) { + err = i915_gem_ww_ctx_backoff(&ww); + if (!err) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); if (err) goto out_unmap; } diff --git a/drivers/gpu/drm/i915/gt/selftest_rps.c b/drivers/gpu/drm/i915/gt/selftest_rps.c index 34b403d47840..3540ba9bd459 100644 --- a/drivers/gpu/drm/i915/gt/selftest_rps.c +++ b/drivers/gpu/drm/i915/gt/selftest_rps.c @@ -77,20 +77,20 @@ create_spin_counter(struct intel_engine_cs *engine, vma = i915_vma_instance(obj, vm, NULL); if (IS_ERR(vma)) { - i915_gem_object_put(obj); - return vma; + err = PTR_ERR(vma); + goto err_put; } err = i915_vma_pin(vma, 0, 0, PIN_USER); - if (err) { - i915_vma_put(vma); - return ERR_PTR(err); - } + if (err) + goto err_unlock; + + i915_vma_lock(vma); base = i915_gem_object_pin_map(obj, I915_MAP_WC); if (IS_ERR(base)) { - i915_gem_object_put(obj); - return ERR_CAST(base); + err = PTR_ERR(base); + goto err_unpin; } cs = base; @@ -134,6 +134,14 @@ create_spin_counter(struct intel_engine_cs *engine, *cancel = base + loop; *counter = srm ? memset32(base + end, 0, 1) : NULL; return vma; + +err_unpin: + i915_vma_unpin(vma); +err_unlock: + i915_vma_unlock(vma); +err_put: + i915_gem_object_put(obj); + return ERR_PTR(err); } static u8 wait_for_freq(struct intel_rps *rps, u8 freq, int timeout_ms) @@ -639,7 +647,6 @@ int live_rps_frequency_cs(void *arg) goto err_vma; } - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, false); if (!err) err = i915_vma_move_to_active(vma, rq, 0); @@ -647,7 +654,6 @@ int live_rps_frequency_cs(void *arg) err = rq->engine->emit_bb_start(rq, vma->node.start, PAGE_SIZE, 0); - i915_vma_unlock(vma); i915_request_add(rq); if (err) goto err_vma; @@ -708,6 +714,7 @@ int live_rps_frequency_cs(void *arg) i915_gem_object_flush_map(vma->obj); i915_gem_object_unpin_map(vma->obj); i915_vma_unpin(vma); + i915_vma_unlock(vma); i915_vma_put(vma); st_engine_heartbeat_enable(engine); @@ -781,7 +788,6 @@ int live_rps_frequency_srm(void *arg) goto err_vma; } - i915_vma_lock(vma); err = i915_request_await_object(rq, vma->obj, false); if (!err) err = i915_vma_move_to_active(vma, rq, 0); @@ -789,7 +795,6 @@ int live_rps_frequency_srm(void *arg) err = rq->engine->emit_bb_start(rq, vma->node.start, PAGE_SIZE, 0); - i915_vma_unlock(vma); i915_request_add(rq); if (err) goto err_vma; @@ -849,6 +854,7 @@ int live_rps_frequency_srm(void *arg) i915_gem_object_flush_map(vma->obj); i915_gem_object_unpin_map(vma->obj); i915_vma_unpin(vma); + i915_vma_unlock(vma); i915_vma_put(vma); st_engine_heartbeat_enable(engine); diff --git a/drivers/gpu/drm/i915/selftests/i915_request.c b/drivers/gpu/drm/i915/selftests/i915_request.c index c1dcd4b91bda..3092ca763789 100644 --- a/drivers/gpu/drm/i915/selftests/i915_request.c +++ b/drivers/gpu/drm/i915/selftests/i915_request.c @@ -862,6 +862,8 @@ static int live_all_engines(void *arg) goto out_free; } + i915_vma_lock(batch); + idx = 0; for_each_uabi_engine(engine, i915) { request[idx] = intel_engine_create_kernel_request(engine); @@ -872,11 +874,9 @@ static int live_all_engines(void *arg) goto out_request; } - i915_vma_lock(batch); err = i915_request_await_object(request[idx], batch->obj, 0); if (err == 0) err = i915_vma_move_to_active(batch, request[idx], 0); - i915_vma_unlock(batch); GEM_BUG_ON(err); err = engine->emit_bb_start(request[idx], @@ -891,6 +891,8 @@ static int live_all_engines(void *arg) idx++; } + i915_vma_unlock(batch); + idx = 0; for_each_uabi_engine(engine, i915) { if (i915_request_completed(request[idx])) { @@ -981,12 +983,13 @@ static int live_sequential_engines(void *arg) goto out_free; } + i915_vma_lock(batch); request[idx] = intel_engine_create_kernel_request(engine); if (IS_ERR(request[idx])) { err = PTR_ERR(request[idx]); pr_err("%s: Request allocation failed for %s with err=%d\n", __func__, engine->name, err); - goto out_request; + goto out_unlock; } if (prev) { @@ -996,16 +999,14 @@ static int live_sequential_engines(void *arg) i915_request_add(request[idx]); pr_err("%s: Request await failed for %s with err=%d\n", __func__, engine->name, err); - goto out_request; + goto out_unlock; } } - i915_vma_lock(batch); err = i915_request_await_object(request[idx], batch->obj, false); if (err == 0) err = i915_vma_move_to_active(batch, request[idx], 0); - i915_vma_unlock(batch); GEM_BUG_ON(err); err = engine->emit_bb_start(request[idx], @@ -1020,6 +1021,11 @@ static int live_sequential_engines(void *arg) prev = request[idx]; idx++; + +out_unlock: + i915_vma_unlock(batch); + if (err) + goto out_request; } idx = 0; From patchwork Mon Aug 10 10:31:02 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707297 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 8E5BE1392 for ; Mon, 10 Aug 2020 10:33:52 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id 7682E206E9 for ; Mon, 10 Aug 2020 10:33:52 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 7682E206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id A58E86E3EE; Mon, 10 Aug 2020 10:33:44 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id 90FD089CDE for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:31:02 +0200 Message-Id: <20200810103103.303818-24-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 23/24] drm/i915: Add ww locking to vm_fault_gtt 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Signed-off-by: Maarten Lankhorst --- drivers/gpu/drm/i915/gem/i915_gem_mman.c | 51 +++++++++++++++--------- 1 file changed, 33 insertions(+), 18 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_mman.c b/drivers/gpu/drm/i915/gem/i915_gem_mman.c index b23368529a40..548ed9fb427d 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_mman.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_mman.c @@ -283,37 +283,46 @@ static vm_fault_t vm_fault_gtt(struct vm_fault *vmf) struct intel_runtime_pm *rpm = &i915->runtime_pm; struct i915_ggtt *ggtt = &i915->ggtt; bool write = area->vm_flags & VM_WRITE; + struct i915_gem_ww_ctx ww; intel_wakeref_t wakeref; struct i915_vma *vma; pgoff_t page_offset; int srcu; int ret; - /* Sanity check that we allow writing into this object */ - if (i915_gem_object_is_readonly(obj) && write) - return VM_FAULT_SIGBUS; - /* We don't use vmf->pgoff since that has the fake offset */ page_offset = (vmf->address - area->vm_start) >> PAGE_SHIFT; trace_i915_gem_object_fault(obj, page_offset, true, write); - ret = i915_gem_object_pin_pages(obj); + wakeref = intel_runtime_pm_get(rpm); + + i915_gem_ww_ctx_init(&ww, true); +retry: + ret = i915_gem_object_lock(obj, &ww); if (ret) - goto err; + goto err_rpm; - wakeref = intel_runtime_pm_get(rpm); + /* Sanity check that we allow writing into this object */ + if (i915_gem_object_is_readonly(obj) && write) { + ret = -EFAULT; + goto err_rpm; + } - ret = intel_gt_reset_trylock(ggtt->vm.gt, &srcu); + ret = i915_gem_object_pin_pages(obj); if (ret) goto err_rpm; + ret = intel_gt_reset_trylock(ggtt->vm.gt, &srcu); + if (ret) + goto err_pages; + /* Now pin it into the GTT as needed */ - vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0, - PIN_MAPPABLE | - PIN_NONBLOCK /* NOWARN */ | - PIN_NOEVICT); - if (IS_ERR(vma)) { + vma = i915_gem_object_ggtt_pin_ww(obj, &ww, NULL, 0, 0, + PIN_MAPPABLE | + PIN_NONBLOCK /* NOWARN */ | + PIN_NOEVICT); + if (IS_ERR(vma) && vma != ERR_PTR(-EDEADLK)) { /* Use a partial view if it is bigger than available space */ struct i915_ggtt_view view = compute_partial_view(obj, page_offset, MIN_CHUNK_PAGES); @@ -328,11 +337,11 @@ static vm_fault_t vm_fault_gtt(struct vm_fault *vmf) * all hope that the hardware is able to track future writes. */ - vma = i915_gem_object_ggtt_pin(obj, &view, 0, 0, flags); - if (IS_ERR(vma)) { + vma = i915_gem_object_ggtt_pin_ww(obj, &ww, &view, 0, 0, flags); + if (IS_ERR(vma) && vma != ERR_PTR(-EDEADLK)) { flags = PIN_MAPPABLE; view.type = I915_GGTT_VIEW_PARTIAL; - vma = i915_gem_object_ggtt_pin(obj, &view, 0, 0, flags); + vma = i915_gem_object_ggtt_pin_ww(obj, &ww, &view, 0, 0, flags); } /* The entire mappable GGTT is pinned? Unexpected! */ @@ -389,10 +398,16 @@ static vm_fault_t vm_fault_gtt(struct vm_fault *vmf) __i915_vma_unpin(vma); err_reset: intel_gt_reset_unlock(ggtt->vm.gt, srcu); +err_pages: + i915_gem_object_unpin_pages(obj); err_rpm: + if (ret == -EDEADLK) { + ret = i915_gem_ww_ctx_backoff(&ww); + if (!ret) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); intel_runtime_pm_put(rpm, wakeref); - i915_gem_object_unpin_pages(obj); -err: return i915_error_to_vmf_fault(ret); } From patchwork Mon Aug 10 10:31:03 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Maarten Lankhorst X-Patchwork-Id: 11707285 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 1A0E21392 for ; Mon, 10 Aug 2020 10:33:49 +0000 (UTC) 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 mail.kernel.org (Postfix) with ESMTPS id F3EDA206E9 for ; Mon, 10 Aug 2020 10:33:48 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org F3EDA206E9 Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=intel-gfx-bounces@lists.freedesktop.org Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 79B886E400; Mon, 10 Aug 2020 10:33:37 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mblankhorst.nl (mblankhorst.nl [141.105.120.124]) by gabe.freedesktop.org (Postfix) with ESMTPS id A45706E3EF for ; Mon, 10 Aug 2020 10:33:29 +0000 (UTC) From: Maarten Lankhorst To: intel-gfx@lists.freedesktop.org Date: Mon, 10 Aug 2020 12:31:03 +0200 Message-Id: <20200810103103.303818-25-maarten.lankhorst@linux.intel.com> X-Mailer: git-send-email 2.28.0 In-Reply-To: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> References: <20200810103103.303818-1-maarten.lankhorst@linux.intel.com> MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH 24/24] drm/i915: Add ww locking to pin_to_display_plane 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: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" Signed-off-by: Maarten Lankhorst --- drivers/gpu/drm/i915/gem/i915_gem_domain.c | 65 ++++++++++++++++------ drivers/gpu/drm/i915/gem/i915_gem_object.h | 1 + 2 files changed, 49 insertions(+), 17 deletions(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_domain.c b/drivers/gpu/drm/i915/gem/i915_gem_domain.c index 8ebceebd11b0..c0d153284984 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_domain.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_domain.c @@ -37,6 +37,12 @@ void i915_gem_object_flush_if_display(struct drm_i915_gem_object *obj) i915_gem_object_unlock(obj); } +void i915_gem_object_flush_if_display_locked(struct drm_i915_gem_object *obj) +{ + if (i915_gem_object_is_framebuffer(obj)) + __i915_gem_object_flush_for_display(obj); +} + /** * Moves a single object to the WC read, and possibly write domain. * @obj: object to act on @@ -197,18 +203,12 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj, if (ret) return ret; - ret = i915_gem_object_lock_interruptible(obj, NULL); - if (ret) - return ret; - /* Always invalidate stale cachelines */ if (obj->cache_level != cache_level) { i915_gem_object_set_cache_coherency(obj, cache_level); obj->cache_dirty = true; } - i915_gem_object_unlock(obj); - /* The cache-level will be applied when each vma is rebound. */ return i915_gem_object_unbind(obj, I915_GEM_OBJECT_UNBIND_ACTIVE | @@ -255,6 +255,7 @@ int i915_gem_set_caching_ioctl(struct drm_device *dev, void *data, struct drm_i915_gem_caching *args = data; struct drm_i915_gem_object *obj; enum i915_cache_level level; + struct i915_gem_ww_ctx ww; int ret = 0; switch (args->caching) { @@ -293,7 +294,18 @@ int i915_gem_set_caching_ioctl(struct drm_device *dev, void *data, goto out; } - ret = i915_gem_object_set_cache_level(obj, level); + i915_gem_ww_ctx_init(&ww, true); +retry: + ret = i915_gem_object_lock(obj, &ww); + if (!ret) + ret = i915_gem_object_set_cache_level(obj, level); + + if (ret == -EDEADLK) { + ret = i915_gem_ww_ctx_backoff(&ww); + if (!ret) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); out: i915_gem_object_put(obj); @@ -313,6 +325,7 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj, unsigned int flags) { struct drm_i915_private *i915 = to_i915(obj->base.dev); + struct i915_gem_ww_ctx ww; struct i915_vma *vma; int ret; @@ -320,6 +333,11 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj, if (HAS_LMEM(i915) && !i915_gem_object_is_lmem(obj)) return ERR_PTR(-EINVAL); + i915_gem_ww_ctx_init(&ww, true); +retry: + ret = i915_gem_object_lock(obj, &ww); + if (ret) + goto err; /* * The display engine is not coherent with the LLC cache on gen6. As * a result, we make sure that the pinning that is about to occur is @@ -334,7 +352,7 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj, HAS_WT(i915) ? I915_CACHE_WT : I915_CACHE_NONE); if (ret) - return ERR_PTR(ret); + goto err; /* * As the user may map the buffer once pinned in the display plane @@ -347,18 +365,31 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj, vma = ERR_PTR(-ENOSPC); if ((flags & PIN_MAPPABLE) == 0 && (!view || view->type == I915_GGTT_VIEW_NORMAL)) - vma = i915_gem_object_ggtt_pin(obj, view, 0, alignment, - flags | - PIN_MAPPABLE | - PIN_NONBLOCK); - if (IS_ERR(vma)) - vma = i915_gem_object_ggtt_pin(obj, view, 0, alignment, flags); - if (IS_ERR(vma)) - return vma; + vma = i915_gem_object_ggtt_pin_ww(obj, &ww, view, 0, alignment, + flags | PIN_MAPPABLE | + PIN_NONBLOCK); + if (IS_ERR(vma) && vma != ERR_PTR(-EDEADLK)) + vma = i915_gem_object_ggtt_pin_ww(obj, &ww, view, 0, + alignment, flags); + if (IS_ERR(vma)) { + ret = PTR_ERR(vma); + goto err; + } vma->display_alignment = max_t(u64, vma->display_alignment, alignment); - i915_gem_object_flush_if_display(obj); + i915_gem_object_flush_if_display_locked(obj); + +err: + if (ret == -EDEADLK) { + ret = i915_gem_ww_ctx_backoff(&ww); + if (!ret) + goto retry; + } + i915_gem_ww_ctx_fini(&ww); + + if (ret) + return ERR_PTR(ret); return vma; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object.h b/drivers/gpu/drm/i915/gem/i915_gem_object.h index 45d79d75e73a..afde1952c119 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_object.h @@ -454,6 +454,7 @@ i915_gem_object_last_write_engine(struct drm_i915_gem_object *obj) void i915_gem_object_set_cache_coherency(struct drm_i915_gem_object *obj, unsigned int cache_level); void i915_gem_object_flush_if_display(struct drm_i915_gem_object *obj); +void i915_gem_object_flush_if_display_locked(struct drm_i915_gem_object *obj); int __must_check i915_gem_object_set_to_wc_domain(struct drm_i915_gem_object *obj, bool write);