From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ben Widawsky Subject: [PATCH 5/5] [v5] drm/i915: Create VMAs Date: Tue, 16 Jul 2013 16:50:09 -0700 Message-ID: <1374018609-1047-5-git-send-email-ben@bwidawsk.net> References: <1374018609-1047-1-git-send-email-ben@bwidawsk.net> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Received: from shiva.localdomain (unknown [209.20.75.48]) by gabe.freedesktop.org (Postfix) with ESMTP id 44FEAE5DC7 for ; Tue, 16 Jul 2013 16:50:43 -0700 (PDT) In-Reply-To: <1374018609-1047-1-git-send-email-ben@bwidawsk.net> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: intel-gfx-bounces+gcfxdi-intel-gfx=m.gmane.org@lists.freedesktop.org Errors-To: intel-gfx-bounces+gcfxdi-intel-gfx=m.gmane.org@lists.freedesktop.org To: Intel GFX Cc: Ben Widawsky List-Id: intel-gfx@lists.freedesktop.org Formerly: "drm/i915: Create VMAs (part 1)" In a previous patch, the notion of a VM was introduced. A VMA describes an area of part of the VM address space. A VMA is similar to the concept in the linux mm. However, instead of representing regular memory, a VMA is backed by a GEM BO. There may be many VMAs for a given object, one for each VM the object is to be used in. This may occur through flink, dma-buf, or a number of other transient states. Currently the code depends on only 1 VMA per object, for the global GTT (and aliasing PPGTT). The following patches will address this and make the rest of the infrastructure more suited v2: s/i915_obj/i915_gem_obj (Chris) v3: Only move an object to the now global unbound list if there are no more VMAs for the object which are bound into a VM (ie. the list is empty). v4: killed obj->gtt_space some reworks due to rebase v5: Free vma on error path (Imre) Signed-off-by: Ben Widawsky --- drivers/gpu/drm/i915/i915_drv.h | 48 ++++++++++++++++++----- drivers/gpu/drm/i915/i915_gem.c | 71 +++++++++++++++++++++++++++------- drivers/gpu/drm/i915/i915_gem_evict.c | 12 ++++-- drivers/gpu/drm/i915/i915_gem_gtt.c | 5 ++- drivers/gpu/drm/i915/i915_gem_stolen.c | 14 +++++-- 5 files changed, 118 insertions(+), 32 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index b3ba428..1a32412 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -533,6 +533,17 @@ struct i915_hw_ppgtt { int (*enable)(struct drm_device *dev); }; +/* To make things as simple as possible (ie. no refcounting), a VMA's lifetime + * will always be <= an objects lifetime. So object refcounting should cover us. + */ +struct i915_vma { + struct drm_mm_node node; + struct drm_i915_gem_object *obj; + struct i915_address_space *vm; + + struct list_head vma_link; /* Link in the object's VMA list */ +}; + struct i915_ctx_hang_stats { /* This context had batch pending when hang was declared */ unsigned batch_pending; @@ -1229,8 +1240,9 @@ struct drm_i915_gem_object { const struct drm_i915_gem_object_ops *ops; - /** Current space allocated to this object in the GTT, if any. */ - struct drm_mm_node gtt_space; + /** List of VMAs backed by this object */ + struct list_head vma_list; + /** Stolen memory for this object, instead of being backed by shmem. */ struct drm_mm_node *stolen; struct list_head global_list; @@ -1356,18 +1368,32 @@ struct drm_i915_gem_object { #define to_intel_bo(x) container_of(x, struct drm_i915_gem_object, base) -/* Offset of the first PTE pointing to this object */ -static inline unsigned long -i915_gem_obj_ggtt_offset(struct drm_i915_gem_object *o) +/* This is a temporary define to help transition us to real VMAs. If you see + * this, you're either reviewing code, or bisecting it. */ +static inline struct i915_vma * +__i915_gem_obj_to_vma(struct drm_i915_gem_object *obj) { - return o->gtt_space.start; + if (list_empty(&obj->vma_list)) + return NULL; + return list_first_entry(&obj->vma_list, struct i915_vma, vma_link); } /* Whether or not this object is currently mapped by the translation tables */ static inline bool i915_gem_obj_ggtt_bound(struct drm_i915_gem_object *o) { - return drm_mm_node_allocated(&o->gtt_space); + struct i915_vma *vma = __i915_gem_obj_to_vma(o); + if (vma == NULL) + return false; + return drm_mm_node_allocated(&vma->node); +} + +/* Offset of the first PTE pointing to this object */ +static inline unsigned long +i915_gem_obj_ggtt_offset(struct drm_i915_gem_object *o) +{ + BUG_ON(list_empty(&o->vma_list)); + return __i915_gem_obj_to_vma(o)->node.start; } /* The size used in the translation tables may be larger than the actual size of @@ -1377,14 +1403,15 @@ i915_gem_obj_ggtt_bound(struct drm_i915_gem_object *o) static inline unsigned long i915_gem_obj_ggtt_size(struct drm_i915_gem_object *o) { - return o->gtt_space.size; + BUG_ON(list_empty(&o->vma_list)); + return __i915_gem_obj_to_vma(o)->node.size; } static inline void i915_gem_obj_ggtt_set_color(struct drm_i915_gem_object *o, enum i915_cache_level color) { - o->gtt_space.color = color; + __i915_gem_obj_to_vma(o)->node.color = color; } /** @@ -1691,6 +1718,9 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj, struct drm_i915_gem_object *i915_gem_alloc_object(struct drm_device *dev, size_t size); void i915_gem_free_object(struct drm_gem_object *obj); +struct i915_vma *i915_gem_vma_create(struct drm_i915_gem_object *obj, + struct i915_address_space *vm); +void i915_gem_vma_destroy(struct i915_vma *vma); int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj, uint32_t alignment, diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 812275a..3eb12ff 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -2580,6 +2580,7 @@ int i915_gem_object_unbind(struct drm_i915_gem_object *obj) { drm_i915_private_t *dev_priv = obj->base.dev->dev_private; + struct i915_vma *vma; int ret; if (!i915_gem_obj_ggtt_bound(obj)) @@ -2617,11 +2618,20 @@ i915_gem_object_unbind(struct drm_i915_gem_object *obj) i915_gem_object_unpin_pages(obj); list_del(&obj->mm_list); - list_move_tail(&obj->global_list, &dev_priv->mm.unbound_list); /* Avoid an unnecessary call to unbind on rebind. */ obj->map_and_fenceable = true; - drm_mm_remove_node(&obj->gtt_space); + vma = __i915_gem_obj_to_vma(obj); + list_del(&vma->vma_link); + drm_mm_remove_node(&vma->node); + i915_gem_vma_destroy(vma); + + /* Since the unbound list is global, only move to that list if + * no more VMAs exist. + * NB: Until we have real VMAs there will only ever be one */ + WARN_ON(!list_empty(&obj->vma_list)); + if (list_empty(&obj->vma_list)) + list_move_tail(&obj->global_list, &dev_priv->mm.unbound_list); return 0; } @@ -3051,8 +3061,12 @@ i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj, bool mappable, fenceable; size_t gtt_max = map_and_fenceable ? dev_priv->gtt.mappable_end : dev_priv->gtt.base.total; + struct i915_vma *vma; int ret; + if (WARN_ON(!list_empty(&obj->vma_list))) + return -EBUSY; + fence_size = i915_gem_get_gtt_size(dev, obj->base.size, obj->tiling_mode); @@ -3091,9 +3105,15 @@ i915_gem_object_bind_to_gtt(struct drm_i915_gem_object *obj, i915_gem_object_pin_pages(obj); + vma = i915_gem_vma_create(obj, &dev_priv->gtt.base); + if (vma == NULL) { + i915_gem_object_unpin_pages(obj); + return -ENOMEM; + } + search_free: ret = drm_mm_insert_node_in_range_generic(&dev_priv->gtt.base.mm, - &obj->gtt_space, + &vma->node, size, alignment, obj->cache_level, 0, gtt_max); if (ret) { @@ -3107,22 +3127,19 @@ search_free: i915_gem_object_unpin_pages(obj); return ret; } - if (WARN_ON(!i915_gem_valid_gtt_space(dev, &obj->gtt_space, + if (WARN_ON(!i915_gem_valid_gtt_space(dev, &vma->node, obj->cache_level))) { - i915_gem_object_unpin_pages(obj); - drm_mm_remove_node(&obj->gtt_space); - return -EINVAL; + ret = -EINVAL; + goto err_out; } ret = i915_gem_gtt_prepare_object(obj); - if (ret) { - i915_gem_object_unpin_pages(obj); - drm_mm_remove_node(&obj->gtt_space); - return ret; - } + if (ret) + goto err_out; list_move_tail(&obj->global_list, &dev_priv->mm.bound_list); list_add_tail(&obj->mm_list, &vm->inactive_list); + list_add(&vma->vma_link, &obj->vma_list); fenceable = i915_gem_obj_ggtt_size(obj) == fence_size && @@ -3136,6 +3153,12 @@ search_free: trace_i915_gem_object_bind(obj, map_and_fenceable); i915_gem_verify_gtt(dev); return 0; + +err_out: + i915_gem_vma_destroy(vma); + i915_gem_object_unpin_pages(obj); + drm_mm_remove_node(&vma->node); + return ret; } void @@ -3281,6 +3304,7 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj, { struct drm_device *dev = obj->base.dev; drm_i915_private_t *dev_priv = dev->dev_private; + struct i915_vma *vma = __i915_gem_obj_to_vma(obj); int ret; if (obj->cache_level == cache_level) @@ -3291,7 +3315,7 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj, return -EBUSY; } - if (!i915_gem_valid_gtt_space(dev, &obj->gtt_space, cache_level)) { + if (!i915_gem_valid_gtt_space(dev, &vma->node, cache_level)) { ret = i915_gem_object_unbind(obj); if (ret) return ret; @@ -3836,6 +3860,7 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj, INIT_LIST_HEAD(&obj->global_list); INIT_LIST_HEAD(&obj->ring_list); INIT_LIST_HEAD(&obj->exec_list); + INIT_LIST_HEAD(&obj->vma_list); obj->ops = ops; @@ -3956,6 +3981,26 @@ void i915_gem_free_object(struct drm_gem_object *gem_obj) i915_gem_object_free(obj); } +struct i915_vma *i915_gem_vma_create(struct drm_i915_gem_object *obj, + struct i915_address_space *vm) +{ + struct i915_vma *vma = kzalloc(sizeof(*vma), GFP_KERNEL); + if (vma == NULL) + return ERR_PTR(-ENOMEM); + + INIT_LIST_HEAD(&vma->vma_link); + vma->vm = vm; + vma->obj = obj; + + return vma; +} + +void i915_gem_vma_destroy(struct i915_vma *vma) +{ + WARN_ON(vma->node.allocated); + kfree(vma); +} + int i915_gem_idle(struct drm_device *dev) { diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c index 43b8235..df61f33 100644 --- a/drivers/gpu/drm/i915/i915_gem_evict.c +++ b/drivers/gpu/drm/i915/i915_gem_evict.c @@ -34,11 +34,13 @@ static bool mark_free(struct drm_i915_gem_object *obj, struct list_head *unwind) { + struct i915_vma *vma = __i915_gem_obj_to_vma(obj); + if (obj->pin_count) return false; list_add(&obj->exec_list, unwind); - return drm_mm_scan_add_block(&obj->gtt_space); + return drm_mm_scan_add_block(&vma->node); } int @@ -49,6 +51,7 @@ i915_gem_evict_something(struct drm_device *dev, int min_size, drm_i915_private_t *dev_priv = dev->dev_private; struct i915_address_space *vm = &dev_priv->gtt.base; struct list_head eviction_list, unwind_list; + struct i915_vma *vma; struct drm_i915_gem_object *obj; int ret = 0; @@ -106,8 +109,8 @@ none: obj = list_first_entry(&unwind_list, struct drm_i915_gem_object, exec_list); - - ret = drm_mm_scan_remove_block(&obj->gtt_space); + vma = __i915_gem_obj_to_vma(obj); + ret = drm_mm_scan_remove_block(&vma->node); BUG_ON(ret); list_del_init(&obj->exec_list); @@ -127,7 +130,8 @@ found: obj = list_first_entry(&unwind_list, struct drm_i915_gem_object, exec_list); - if (drm_mm_scan_remove_block(&obj->gtt_space)) { + vma = __i915_gem_obj_to_vma(obj); + if (drm_mm_scan_remove_block(&vma->node)) { list_move(&obj->exec_list, &eviction_list); drm_gem_object_reference(&obj->base); continue; diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c index 999ecfe..3b639a9 100644 --- a/drivers/gpu/drm/i915/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/i915_gem_gtt.c @@ -662,16 +662,17 @@ void i915_gem_setup_global_gtt(struct drm_device *dev, /* Mark any preallocated objects as occupied */ list_for_each_entry(obj, &dev_priv->mm.bound_list, global_list) { + struct i915_vma *vma = __i915_gem_obj_to_vma(obj); int ret; DRM_DEBUG_KMS("reserving preallocated space: %lx + %zx\n", i915_gem_obj_ggtt_offset(obj), obj->base.size); WARN_ON(i915_gem_obj_ggtt_bound(obj)); - ret = drm_mm_reserve_node(&dev_priv->gtt.base.mm, - &obj->gtt_space); + ret = drm_mm_reserve_node(&dev_priv->gtt.base.mm, &vma->node); if (ret) DRM_DEBUG_KMS("Reservation failed\n"); obj->has_global_gtt_mapping = 1; + list_add(&vma->vma_link, &obj->vma_list); } dev_priv->gtt.base.start = start; diff --git a/drivers/gpu/drm/i915/i915_gem_stolen.c b/drivers/gpu/drm/i915/i915_gem_stolen.c index 46a9715..a1f2308 100644 --- a/drivers/gpu/drm/i915/i915_gem_stolen.c +++ b/drivers/gpu/drm/i915/i915_gem_stolen.c @@ -354,6 +354,7 @@ i915_gem_object_create_stolen_for_preallocated(struct drm_device *dev, struct i915_address_space *vm = &dev_priv->gtt.base; struct drm_i915_gem_object *obj; struct drm_mm_node *stolen; + struct i915_vma *vma; int ret; if (!drm_mm_initialized(&dev_priv->mm.stolen)) @@ -393,16 +394,21 @@ i915_gem_object_create_stolen_for_preallocated(struct drm_device *dev, if (gtt_offset == I915_GTT_OFFSET_NONE) return obj; + vma = i915_gem_vma_create(obj, &dev_priv->gtt.base); + if (!vma) { + drm_gem_object_unreference(&obj->base); + return NULL; + } + /* To simplify the initialisation sequence between KMS and GTT, * we allow construction of the stolen object prior to * setting up the GTT space. The actual reservation will occur * later. */ - obj->gtt_space.start = gtt_offset; - obj->gtt_space.size = size; + vma->node.start = gtt_offset; + vma->node.size = size; if (drm_mm_initialized(&dev_priv->gtt.base.mm)) { - ret = drm_mm_reserve_node(&dev_priv->gtt.base.mm, - &obj->gtt_space); + ret = drm_mm_reserve_node(&dev_priv->gtt.base.mm, &vma->node); if (ret) { DRM_DEBUG_KMS("failed to allocate stolen GTT space\n"); goto unref_out; -- 1.8.3.3