Split-up of "drm/i915: Only bind each object rather than for every execbuffer"

Split-up of "drm/i915: Only bind each object rather than for every execbuffer"

[Daniel Vetter]

Apparently I've lost a game of chicken. For review I'd like to sign up Jani for
the overall series and Ben to upgrade his ack on the actual bugfix to a full
r-b.

Cheers, Daniel

[PATCH 1/9] drm/i915: Consolidate binding parameters into flags

[Daniel Vetter]

Anything more than just one bool parameter is just a pain to read,
symbolic constants are much better.

Split out from Chris' vma-binding rework patch.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_drv.h            | 15 ++++----
 drivers/gpu/drm/i915/i915_gem.c            | 62 ++++++++++++------------------
 drivers/gpu/drm/i915/i915_gem_context.c    |  9 ++---
 drivers/gpu/drm/i915/i915_gem_evict.c      | 10 ++---
 drivers/gpu/drm/i915/i915_gem_execbuffer.c | 16 +++++---
 drivers/gpu/drm/i915/i915_gem_gtt.c        |  2 +-
 drivers/gpu/drm/i915/i915_trace.h          | 20 +++++-----
 drivers/gpu/drm/i915/intel_overlay.c       |  2 +-
 drivers/gpu/drm/i915/intel_pm.c            |  2 +-
 drivers/gpu/drm/i915/intel_ringbuffer.c    | 15 ++++----
 10 files changed, 71 insertions(+), 82 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 3971e7c3943d..dcab2b0a09ef 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -2030,11 +2030,13 @@ void i915_init_vm(struct drm_i915_private *dev_priv,
 void i915_gem_free_object(struct drm_gem_object *obj);
 void i915_gem_vma_destroy(struct i915_vma *vma);
 
+#define PIN_MAPPABLE 0x1
+#define PIN_GLOBAL 0x2
+#define PIN_NONBLOCK 0x4
 int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,
 				     struct i915_address_space *vm,
 				     uint32_t alignment,
-				     bool map_and_fenceable,
-				     bool nonblocking);
+				     unsigned flags);
 void i915_gem_object_ggtt_unpin(struct drm_i915_gem_object *obj);
 int __must_check i915_vma_unbind(struct i915_vma *vma);
 int __must_check i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj);
@@ -2237,11 +2239,9 @@ i915_gem_obj_ggtt_size(struct drm_i915_gem_object *obj)
 static inline int __must_check
 i915_gem_obj_ggtt_pin(struct drm_i915_gem_object *obj,
 		      uint32_t alignment,
-		      bool map_and_fenceable,
-		      bool nonblocking)
+		      unsigned flags)
 {
-	return i915_gem_object_pin(obj, obj_to_ggtt(obj), alignment,
-				   map_and_fenceable, nonblocking);
+	return i915_gem_object_pin(obj, obj_to_ggtt(obj), alignment, flags | PIN_GLOBAL);
 }
 
 /* i915_gem_context.c */
@@ -2280,8 +2280,7 @@ int __must_check i915_gem_evict_something(struct drm_device *dev,
 					  int min_size,
 					  unsigned alignment,
 					  unsigned cache_level,
-					  bool mappable,
-					  bool nonblock);
+					  unsigned flags);
 int i915_gem_evict_vm(struct i915_address_space *vm, bool do_idle);
 int i915_gem_evict_everything(struct drm_device *dev);
 
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 39770f7b333f..9f4dab7df5c0 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -43,12 +43,6 @@ static void i915_gem_object_flush_cpu_write_domain(struct drm_i915_gem_object *o
 static __must_check int
 i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj,
 			       bool readonly);
-static __must_check int
-i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
-			   struct i915_address_space *vm,
-			   unsigned alignment,
-			   bool map_and_fenceable,
-			   bool nonblocking);
 static int i915_gem_phys_pwrite(struct drm_device *dev,
 				struct drm_i915_gem_object *obj,
 				struct drm_i915_gem_pwrite *args,
@@ -605,7 +599,7 @@ i915_gem_gtt_pwrite_fast(struct drm_device *dev,
 	char __user *user_data;
 	int page_offset, page_length, ret;
 
-	ret = i915_gem_obj_ggtt_pin(obj, 0, true, true);
+	ret = i915_gem_obj_ggtt_pin(obj, 0, PIN_MAPPABLE | PIN_NONBLOCK);
 	if (ret)
 		goto out;
 
@@ -1399,7 +1393,7 @@ int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 	}
 
 	/* Now bind it into the GTT if needed */
-	ret = i915_gem_obj_ggtt_pin(obj,  0, true, false);
+	ret = i915_gem_obj_ggtt_pin(obj, 0, PIN_MAPPABLE);
 	if (ret)
 		goto unlock;
 
@@ -2769,7 +2763,6 @@ int i915_vma_unbind(struct i915_vma *vma)
 
 	if (!drm_mm_node_allocated(&vma->node)) {
 		i915_gem_vma_destroy(vma);
-
 		return 0;
 	}
 
@@ -3267,14 +3260,13 @@ static int
 i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 			   struct i915_address_space *vm,
 			   unsigned alignment,
-			   bool map_and_fenceable,
-			   bool nonblocking)
+			   unsigned flags)
 {
 	struct drm_device *dev = obj->base.dev;
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 size, fence_size, fence_alignment, unfenced_alignment;
 	size_t gtt_max =
-		map_and_fenceable ? dev_priv->gtt.mappable_end : vm->total;
+		flags & PIN_MAPPABLE ? dev_priv->gtt.mappable_end : vm->total;
 	struct i915_vma *vma;
 	int ret;
 
@@ -3286,18 +3278,18 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 						     obj->tiling_mode, true);
 	unfenced_alignment =
 		i915_gem_get_gtt_alignment(dev,
-						    obj->base.size,
-						    obj->tiling_mode, false);
+					   obj->base.size,
+					   obj->tiling_mode, false);
 
 	if (alignment == 0)
-		alignment = map_and_fenceable ? fence_alignment :
+		alignment = flags & PIN_MAPPABLE ? fence_alignment :
 						unfenced_alignment;
-	if (map_and_fenceable && alignment & (fence_alignment - 1)) {
+	if (flags & PIN_MAPPABLE && alignment & (fence_alignment - 1)) {
 		DRM_ERROR("Invalid object alignment requested %u\n", alignment);
 		return -EINVAL;
 	}
 
-	size = map_and_fenceable ? fence_size : obj->base.size;
+	size = flags & PIN_MAPPABLE ? fence_size : obj->base.size;
 
 	/* If the object is bigger than the entire aperture, reject it early
 	 * before evicting everything in a vain attempt to find space.
@@ -3305,7 +3297,7 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 	if (obj->base.size > gtt_max) {
 		DRM_ERROR("Attempting to bind an object larger than the aperture: object=%zd > %s aperture=%zu\n",
 			  obj->base.size,
-			  map_and_fenceable ? "mappable" : "total",
+			  flags & PIN_MAPPABLE ? "mappable" : "total",
 			  gtt_max);
 		return -E2BIG;
 	}
@@ -3329,9 +3321,7 @@ search_free:
 						  DRM_MM_SEARCH_DEFAULT);
 	if (ret) {
 		ret = i915_gem_evict_something(dev, vm, size, alignment,
-					       obj->cache_level,
-					       map_and_fenceable,
-					       nonblocking);
+					       obj->cache_level, flags);
 		if (ret == 0)
 			goto search_free;
 
@@ -3362,9 +3352,9 @@ search_free:
 		obj->map_and_fenceable = mappable && fenceable;
 	}
 
-	WARN_ON(map_and_fenceable && !obj->map_and_fenceable);
+	WARN_ON(flags & PIN_MAPPABLE && !obj->map_and_fenceable);
 
-	trace_i915_vma_bind(vma, map_and_fenceable);
+	trace_i915_vma_bind(vma, flags);
 	i915_gem_verify_gtt(dev);
 	return 0;
 
@@ -3735,7 +3725,7 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj,
 	 * (e.g. libkms for the bootup splash), we have to ensure that we
 	 * always use map_and_fenceable for all scanout buffers.
 	 */
-	ret = i915_gem_obj_ggtt_pin(obj, alignment, true, false);
+	ret = i915_gem_obj_ggtt_pin(obj, alignment, PIN_MAPPABLE);
 	if (ret)
 		goto err_unpin_display;
 
@@ -3891,30 +3881,28 @@ int
 i915_gem_object_pin(struct drm_i915_gem_object *obj,
 		    struct i915_address_space *vm,
 		    uint32_t alignment,
-		    bool map_and_fenceable,
-		    bool nonblocking)
+		    unsigned flags)
 {
-	const u32 flags = map_and_fenceable ? GLOBAL_BIND : 0;
 	struct i915_vma *vma;
 	int ret;
 
-	WARN_ON(map_and_fenceable && !i915_is_ggtt(vm));
+	if (WARN_ON(flags & (PIN_GLOBAL | PIN_MAPPABLE) && !i915_is_ggtt(vm)))
+		return -EINVAL;
 
 	vma = i915_gem_obj_to_vma(obj, vm);
-
 	if (vma) {
 		if (WARN_ON(vma->pin_count == DRM_I915_GEM_OBJECT_MAX_PIN_COUNT))
 			return -EBUSY;
 
 		if ((alignment &&
 		     vma->node.start & (alignment - 1)) ||
-		    (map_and_fenceable && !obj->map_and_fenceable)) {
+		    (flags & PIN_MAPPABLE && !obj->map_and_fenceable)) {
 			WARN(vma->pin_count,
 			     "bo is already pinned with incorrect alignment:"
 			     " offset=%lx, req.alignment=%x, req.map_and_fenceable=%d,"
 			     " obj->map_and_fenceable=%d\n",
 			     i915_gem_obj_offset(obj, vm), alignment,
-			     map_and_fenceable,
+			     flags & PIN_MAPPABLE,
 			     obj->map_and_fenceable);
 			ret = i915_vma_unbind(vma);
 			if (ret)
@@ -3923,9 +3911,7 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
 	}
 
 	if (!i915_gem_obj_bound(obj, vm)) {
-		ret = i915_gem_object_bind_to_vm(obj, vm, alignment,
-						 map_and_fenceable,
-						 nonblocking);
+		ret = i915_gem_object_bind_to_vm(obj, vm, alignment, flags);
 		if (ret)
 			return ret;
 
@@ -3933,10 +3919,12 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
 
 	vma = i915_gem_obj_to_vma(obj, vm);
 
-	vma->bind_vma(vma, obj->cache_level, flags);
+	if (flags & PIN_GLOBAL && !obj->has_global_gtt_mapping)
+		vma->bind_vma(vma, obj->cache_level, GLOBAL_BIND);
 
 	i915_gem_obj_to_vma(obj, vm)->pin_count++;
-	obj->pin_mappable |= map_and_fenceable;
+	if (flags & PIN_MAPPABLE)
+		obj->pin_mappable |= true;
 
 	return 0;
 }
@@ -3994,7 +3982,7 @@ i915_gem_pin_ioctl(struct drm_device *dev, void *data,
 	}
 
 	if (obj->user_pin_count == 0) {
-		ret = i915_gem_obj_ggtt_pin(obj, args->alignment, true, false);
+		ret = i915_gem_obj_ggtt_pin(obj, args->alignment, PIN_MAPPABLE);
 		if (ret)
 			goto out;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index fb64ab4a2068..1c9408230492 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -263,8 +263,7 @@ i915_gem_create_context(struct drm_device *dev,
 		 * context.
 		 */
 		ret = i915_gem_obj_ggtt_pin(ctx->obj,
-					    get_context_alignment(dev),
-					    false, false);
+					    get_context_alignment(dev), 0);
 		if (ret) {
 			DRM_DEBUG_DRIVER("Couldn't pin %d\n", ret);
 			goto err_destroy;
@@ -340,8 +339,7 @@ void i915_gem_context_reset(struct drm_device *dev)
 
 		if (i == RCS) {
 			WARN_ON(i915_gem_obj_ggtt_pin(dctx->obj,
-						      get_context_alignment(dev),
-						      false, false));
+						      get_context_alignment(dev), 0));
 			/* Fake a finish/inactive */
 			dctx->obj->base.write_domain = 0;
 			dctx->obj->active = 0;
@@ -613,8 +611,7 @@ static int do_switch(struct intel_ring_buffer *ring,
 	/* Trying to pin first makes error handling easier. */
 	if (ring == &dev_priv->ring[RCS]) {
 		ret = i915_gem_obj_ggtt_pin(to->obj,
-					    get_context_alignment(ring->dev),
-					    false, false);
+					    get_context_alignment(ring->dev), 0);
 		if (ret)
 			return ret;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c
index 4e82ca4a7a52..525b242ac512 100644
--- a/drivers/gpu/drm/i915/i915_gem_evict.c
+++ b/drivers/gpu/drm/i915/i915_gem_evict.c
@@ -50,7 +50,7 @@ mark_free(struct i915_vma *vma, struct list_head *unwind)
 int
 i915_gem_evict_something(struct drm_device *dev, struct i915_address_space *vm,
 			 int min_size, unsigned alignment, unsigned cache_level,
-			 bool mappable, bool nonblocking)
+			 unsigned flags)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct list_head eviction_list, unwind_list;
@@ -58,7 +58,7 @@ i915_gem_evict_something(struct drm_device *dev, struct i915_address_space *vm,
 	int ret = 0;
 	int pass = 0;
 
-	trace_i915_gem_evict(dev, min_size, alignment, mappable);
+	trace_i915_gem_evict(dev, min_size, alignment, flags);
 
 	/*
 	 * The goal is to evict objects and amalgamate space in LRU order.
@@ -84,7 +84,7 @@ i915_gem_evict_something(struct drm_device *dev, struct i915_address_space *vm,
 	 */
 
 	INIT_LIST_HEAD(&unwind_list);
-	if (mappable) {
+	if (flags & PIN_MAPPABLE) {
 		BUG_ON(!i915_is_ggtt(vm));
 		drm_mm_init_scan_with_range(&vm->mm, min_size,
 					    alignment, cache_level, 0,
@@ -99,7 +99,7 @@ search_again:
 			goto found;
 	}
 
-	if (nonblocking)
+	if (flags & PIN_NONBLOCK)
 		goto none;
 
 	/* Now merge in the soon-to-be-expired objects... */
@@ -123,7 +123,7 @@ none:
 	/* Can we unpin some objects such as idle hw contents,
 	 * or pending flips?
 	 */
-	if (nonblocking)
+	if (flags & PIN_NONBLOCK)
 		return -ENOSPC;
 
 	/* Only idle the GPU and repeat the search once */
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 032def901f98..9399a6fa4c2f 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -544,19 +544,23 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,
 	struct drm_i915_gem_object *obj = vma->obj;
 	struct drm_i915_gem_exec_object2 *entry = vma->exec_entry;
 	bool has_fenced_gpu_access = INTEL_INFO(ring->dev)->gen < 4;
-	bool need_fence, need_mappable;
-	u32 flags = (entry->flags & EXEC_OBJECT_NEEDS_GTT) &&
-		!vma->obj->has_global_gtt_mapping ? GLOBAL_BIND : 0;
+	bool need_fence;
+	unsigned flags;
 	int ret;
 
+	flags = 0;
+
 	need_fence =
 		has_fenced_gpu_access &&
 		entry->flags & EXEC_OBJECT_NEEDS_FENCE &&
 		obj->tiling_mode != I915_TILING_NONE;
-	need_mappable = need_fence || need_reloc_mappable(vma);
+	if (need_fence || need_reloc_mappable(vma))
+		flags |= PIN_MAPPABLE;
+
+	if (entry->flags & EXEC_OBJECT_NEEDS_GTT)
+		flags |= PIN_GLOBAL;
 
-	ret = i915_gem_object_pin(obj, vma->vm, entry->alignment, need_mappable,
-				  false);
+	ret = i915_gem_object_pin(obj, vma->vm, entry->alignment, flags);
 	if (ret)
 		return ret;
 
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 6e858e17bb0c..154e1ad8bfab 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -868,7 +868,7 @@ alloc:
 	if (ret == -ENOSPC && !retried) {
 		ret = i915_gem_evict_something(dev, &dev_priv->gtt.base,
 					       GEN6_PD_SIZE, GEN6_PD_ALIGN,
-					       I915_CACHE_NONE, false, true);
+					       I915_CACHE_NONE, PIN_NONBLOCK);
 		if (ret)
 			return ret;
 
diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h
index 6e580c98dede..b95a380958db 100644
--- a/drivers/gpu/drm/i915/i915_trace.h
+++ b/drivers/gpu/drm/i915/i915_trace.h
@@ -34,15 +34,15 @@ TRACE_EVENT(i915_gem_object_create,
 );
 
 TRACE_EVENT(i915_vma_bind,
-	    TP_PROTO(struct i915_vma *vma, bool mappable),
-	    TP_ARGS(vma, mappable),
+	    TP_PROTO(struct i915_vma *vma, unsigned flags),
+	    TP_ARGS(vma, flags),
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
 			     __field(struct i915_address_space *, vm)
 			     __field(u32, offset)
 			     __field(u32, size)
-			     __field(bool, mappable)
+			     __field(unsigned, flags)
 			     ),
 
 	    TP_fast_assign(
@@ -50,12 +50,12 @@ TRACE_EVENT(i915_vma_bind,
 			   __entry->vm = vma->vm;
 			   __entry->offset = vma->node.start;
 			   __entry->size = vma->node.size;
-			   __entry->mappable = mappable;
+			   __entry->flags = flags;
 			   ),
 
 	    TP_printk("obj=%p, offset=%08x size=%x%s vm=%p",
 		      __entry->obj, __entry->offset, __entry->size,
-		      __entry->mappable ? ", mappable" : "",
+		      __entry->flags & PIN_MAPPABLE ? ", mappable" : "",
 		      __entry->vm)
 );
 
@@ -196,26 +196,26 @@ DEFINE_EVENT(i915_gem_object, i915_gem_object_destroy,
 );
 
 TRACE_EVENT(i915_gem_evict,
-	    TP_PROTO(struct drm_device *dev, u32 size, u32 align, bool mappable),
-	    TP_ARGS(dev, size, align, mappable),
+	    TP_PROTO(struct drm_device *dev, u32 size, u32 align, unsigned flags),
+	    TP_ARGS(dev, size, align, flags),
 
 	    TP_STRUCT__entry(
 			     __field(u32, dev)
 			     __field(u32, size)
 			     __field(u32, align)
-			     __field(bool, mappable)
+			     __field(unsigned, flags)
 			    ),
 
 	    TP_fast_assign(
 			   __entry->dev = dev->primary->index;
 			   __entry->size = size;
 			   __entry->align = align;
-			   __entry->mappable = mappable;
+			   __entry->flags = flags;
 			  ),
 
 	    TP_printk("dev=%d, size=%d, align=%d %s",
 		      __entry->dev, __entry->size, __entry->align,
-		      __entry->mappable ? ", mappable" : "")
+		      __entry->flags & PIN_MAPPABLE ? ", mappable" : "")
 );
 
 TRACE_EVENT(i915_gem_evict_everything,
diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index 424f0946d8c4..ac519cb46f22 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -1349,7 +1349,7 @@ void intel_setup_overlay(struct drm_device *dev)
 		}
 		overlay->flip_addr = reg_bo->phys_obj->handle->busaddr;
 	} else {
-		ret = i915_gem_obj_ggtt_pin(reg_bo, PAGE_SIZE, true, false);
+		ret = i915_gem_obj_ggtt_pin(reg_bo, PAGE_SIZE, PIN_MAPPABLE);
 		if (ret) {
 			DRM_ERROR("failed to pin overlay register bo\n");
 			goto out_free_bo;
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 4960314a3611..33dc2dd29750 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -2741,7 +2741,7 @@ intel_alloc_context_page(struct drm_device *dev)
 		return NULL;
 	}
 
-	ret = i915_gem_obj_ggtt_pin(ctx, 4096, true, false);
+	ret = i915_gem_obj_ggtt_pin(ctx, 4096, PIN_MAPPABLE);
 	if (ret) {
 		DRM_ERROR("failed to pin power context: %d\n", ret);
 		goto err_unref;
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index d897a19f887f..a0793c929b95 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -531,9 +531,11 @@ init_pipe_control(struct intel_ring_buffer *ring)
 		goto err;
 	}
 
-	i915_gem_object_set_cache_level(ring->scratch.obj, I915_CACHE_LLC);
+	ret = i915_gem_object_set_cache_level(ring->scratch.obj, I915_CACHE_LLC);
+	if (ret)
+		goto err_unref;
 
-	ret = i915_gem_obj_ggtt_pin(ring->scratch.obj, 4096, true, false);
+	ret = i915_gem_obj_ggtt_pin(ring->scratch.obj, 4096, 0);
 	if (ret)
 		goto err_unref;
 
@@ -1273,10 +1275,9 @@ static int init_status_page(struct intel_ring_buffer *ring)
 
 	i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
 
-	ret = i915_gem_obj_ggtt_pin(obj, 4096, true, false);
-	if (ret != 0) {
+	ret = i915_gem_obj_ggtt_pin(obj, 4096, PIN_MAPPABLE);
+	if (ret)
 		goto err_unref;
-	}
 
 	ring->status_page.gfx_addr = i915_gem_obj_ggtt_offset(obj);
 	ring->status_page.page_addr = kmap(sg_page(obj->pages->sgl));
@@ -1356,7 +1357,7 @@ static int intel_init_ring_buffer(struct drm_device *dev,
 
 	ring->obj = obj;
 
-	ret = i915_gem_obj_ggtt_pin(obj, PAGE_SIZE, true, false);
+	ret = i915_gem_obj_ggtt_pin(obj, PAGE_SIZE, PIN_MAPPABLE);
 	if (ret)
 		goto err_unref;
 
@@ -1933,7 +1934,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
 			return -ENOMEM;
 		}
 
-		ret = i915_gem_obj_ggtt_pin(obj, 0, true, false);
+		ret = i915_gem_obj_ggtt_pin(obj, 0, PIN_MAPPABLE);
 		if (ret != 0) {
 			drm_gem_object_unreference(&obj->base);
 			DRM_ERROR("Failed to ping batch bo\n");
-- 
1.8.5.2

[PATCH 2/9] drm/i915: Don't set PIN_MAPPABLE for legacy ringbuffers

[Daniel Vetter]

Tighter code since legacy gem has only mappable anyway.

Split out from Chris vma-bind rework.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/intel_ringbuffer.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index a0793c929b95..c8013571c8b5 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1934,7 +1934,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
 			return -ENOMEM;
 		}
 
-		ret = i915_gem_obj_ggtt_pin(obj, 0, PIN_MAPPABLE);
+		ret = i915_gem_obj_ggtt_pin(obj, 0, 0);
 		if (ret != 0) {
 			drm_gem_object_unreference(&obj->base);
 			DRM_ERROR("Failed to ping batch bo\n");
-- 
1.8.5.2

[PATCH 3/9] drm/i915: Don't pin the status page as mappable

[Daniel Vetter]

We access it through the cpu window. No functional difference expected
atm since we default to a bottom-up allocation scheme. But that might
eventually change so that we prefer the unmappable range for buffers
that don't need cpu gtt access.

Split out from Chris vma-bind rework.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/intel_ringbuffer.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index c8013571c8b5..bc19261eb436 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1275,7 +1275,7 @@ static int init_status_page(struct intel_ring_buffer *ring)
 
 	i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
 
-	ret = i915_gem_obj_ggtt_pin(obj, 4096, PIN_MAPPABLE);
+	ret = i915_gem_obj_ggtt_pin(obj, 4096, 0);
 	if (ret)
 		goto err_unref;
 
-- 
1.8.5.2

[PATCH 4/9] drm/i915: Handle set_cache_level errors in the status page setup

[Daniel Vetter]

Split out from Chris vma-bind rework.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/intel_ringbuffer.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index bc19261eb436..e75d8b325029 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1273,7 +1273,9 @@ static int init_status_page(struct intel_ring_buffer *ring)
 		goto err;
 	}
 
-	i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
+	ret = i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
+	if (ret)
+		goto err_unref;
 
 	ret = i915_gem_obj_ggtt_pin(obj, 4096, 0);
 	if (ret)
-- 
1.8.5.2

[PATCH 5/9] drm/i915: Don't allocate context pages as mappable

[Daniel Vetter]

Only the hardware really access them, so no need to have cpu
gtt access available.

Split out from Chris vma-bind rework.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/intel_pm.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 33dc2dd29750..54e99db6e509 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -2741,7 +2741,7 @@ intel_alloc_context_page(struct drm_device *dev)
 		return NULL;
 	}
 
-	ret = i915_gem_obj_ggtt_pin(ctx, 4096, PIN_MAPPABLE);
+	ret = i915_gem_obj_ggtt_pin(ctx, 4096, 0);
 	if (ret) {
 		DRM_ERROR("failed to pin power context: %d\n", ret);
 		goto err_unref;
-- 
1.8.5.2

[PATCH 6/9] drm/i915: Allow blocking in the PDE alloc when running low on gtt space

[Daniel Vetter]

There's no need not to, really.

Split out from Chris vma-bind rework.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_gem_gtt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 154e1ad8bfab..af3bd9b6fa27 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -868,7 +868,7 @@ alloc:
 	if (ret == -ENOSPC && !retried) {
 		ret = i915_gem_evict_something(dev, &dev_priv->gtt.base,
 					       GEN6_PD_SIZE, GEN6_PD_ALIGN,
-					       I915_CACHE_NONE, PIN_NONBLOCK);
+					       I915_CACHE_NONE, 0);
 		if (ret)
 			return ret;
 
-- 
1.8.5.2

[PATCH 7/9] drm/i915: Simplify i915_gem_object_ggtt_unpin

[Daniel Vetter]

Split out from Chris vma-bind rework.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_drv.h | 10 ++++++++--
 drivers/gpu/drm/i915/i915_gem.c | 20 --------------------
 2 files changed, 8 insertions(+), 22 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index dcab2b0a09ef..13991b8ce602 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -2037,9 +2037,7 @@ int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,
 				     struct i915_address_space *vm,
 				     uint32_t alignment,
 				     unsigned flags);
-void i915_gem_object_ggtt_unpin(struct drm_i915_gem_object *obj);
 int __must_check i915_vma_unbind(struct i915_vma *vma);
-int __must_check i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj);
 int i915_gem_object_put_pages(struct drm_i915_gem_object *obj);
 void i915_gem_release_all_mmaps(struct drm_i915_private *dev_priv);
 void i915_gem_release_mmap(struct drm_i915_gem_object *obj);
@@ -2244,6 +2242,14 @@ i915_gem_obj_ggtt_pin(struct drm_i915_gem_object *obj,
 	return i915_gem_object_pin(obj, obj_to_ggtt(obj), alignment, flags | PIN_GLOBAL);
 }
 
+static inline int
+i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj)
+{
+	return i915_vma_unbind(i915_gem_obj_to_ggtt(obj));
+}
+
+void i915_gem_object_ggtt_unpin(struct drm_i915_gem_object *obj);
+
 /* i915_gem_context.c */
 #define ctx_to_ppgtt(ctx) container_of((ctx)->vm, struct i915_hw_ppgtt, base)
 int __must_check i915_gem_context_init(struct drm_device *dev);
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 9f4dab7df5c0..680b300d7454 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2814,26 +2814,6 @@ int i915_vma_unbind(struct i915_vma *vma)
 	return 0;
 }
 
-/**
- * Unbinds an object from the global GTT aperture.
- */
-int
-i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj)
-{
-	struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
-	struct i915_address_space *ggtt = &dev_priv->gtt.base;
-
-	if (!i915_gem_obj_ggtt_bound(obj))
-		return 0;
-
-	if (i915_gem_obj_to_ggtt(obj)->pin_count)
-		return -EBUSY;
-
-	BUG_ON(obj->pages == NULL);
-
-	return i915_vma_unbind(i915_gem_obj_to_vma(obj, ggtt));
-}
-
 int i915_gpu_idle(struct drm_device *dev)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
-- 
1.8.5.2

[PATCH 8/9] drm/i915: Directly return the vma from bind_to_vm

[Daniel Vetter]

This is prep work for reworking the object_pin logic. Atm
it still does a (now redundant) lookup of the vma. The next
patch will fix this.

Split out from Chris vma-bind rework.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_gem.c | 23 +++++++++++------------
 1 file changed, 11 insertions(+), 12 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 680b300d7454..a3534246fbca 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3236,7 +3236,7 @@ static void i915_gem_verify_gtt(struct drm_device *dev)
 /**
  * Finds free space in the GTT aperture and binds the object there.
  */
-static int
+static struct i915_vma *
 i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 			   struct i915_address_space *vm,
 			   unsigned alignment,
@@ -3266,7 +3266,7 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 						unfenced_alignment;
 	if (flags & PIN_MAPPABLE && alignment & (fence_alignment - 1)) {
 		DRM_ERROR("Invalid object alignment requested %u\n", alignment);
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 	}
 
 	size = flags & PIN_MAPPABLE ? fence_size : obj->base.size;
@@ -3279,20 +3279,18 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 			  obj->base.size,
 			  flags & PIN_MAPPABLE ? "mappable" : "total",
 			  gtt_max);
-		return -E2BIG;
+		return ERR_PTR(-E2BIG);
 	}
 
 	ret = i915_gem_object_get_pages(obj);
 	if (ret)
-		return ret;
+		return ERR_PTR(ret);
 
 	i915_gem_object_pin_pages(obj);
 
 	vma = i915_gem_obj_lookup_or_create_vma(obj, vm);
-	if (IS_ERR(vma)) {
-		ret = PTR_ERR(vma);
+	if (IS_ERR(vma))
 		goto err_unpin;
-	}
 
 search_free:
 	ret = drm_mm_insert_node_in_range_generic(&vm->mm, &vma->node,
@@ -3336,15 +3334,16 @@ search_free:
 
 	trace_i915_vma_bind(vma, flags);
 	i915_gem_verify_gtt(dev);
-	return 0;
+	return vma;
 
 err_remove_node:
 	drm_mm_remove_node(&vma->node);
 err_free_vma:
 	i915_gem_vma_destroy(vma);
+	vma = ERR_PTR(ret);
 err_unpin:
 	i915_gem_object_unpin_pages(obj);
-	return ret;
+	return vma;
 }
 
 bool
@@ -3891,10 +3890,10 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
 	}
 
 	if (!i915_gem_obj_bound(obj, vm)) {
-		ret = i915_gem_object_bind_to_vm(obj, vm, alignment, flags);
-		if (ret)
-			return ret;
 
+		vma = i915_gem_object_bind_to_vm(obj, vm, alignment, flags);
+		if (IS_ERR(vma))
+			return PTR_ERR(vma);
 	}
 
 	vma = i915_gem_obj_to_vma(obj, vm);
-- 
1.8.5.2

[PATCH 9/9] drm/i915: Only bind each object rather than for every execbuffer

[Daniel Vetter]

From: Chris Wilson <chris@chris-wilson.co.uk>

One side-effect of the introduction of ppgtt was that we needed to
rebind the object into the appropriate vm (and global gtt in some
peculiar cases). For simplicity this was done twice for every object on
every call to execbuffer. However, that adds a tremendous amount of CPU
overhead (rewriting all the PTE for all objects into WC memory) per
draw. The fix is to push all the decision about which vm to bind into
and when down into the low-level bind routines through hints rather than
inside the execbuffer routine.

v2: Split out prep work and unrelated changes.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=72906
Tested-by: jianx.zhou@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> (v1)
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_gem.c            | 16 ++++++++++------
 drivers/gpu/drm/i915/i915_gem_execbuffer.c |  2 --
 2 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index a3534246fbca..05e79fa02db2 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3333,6 +3333,9 @@ search_free:
 	WARN_ON(flags & PIN_MAPPABLE && !obj->map_and_fenceable);
 
 	trace_i915_vma_bind(vma, flags);
+	vma->bind_vma(vma, obj->cache_level,
+		      flags & (PIN_MAPPABLE | PIN_GLOBAL) ? GLOBAL_BIND : 0);
+
 	i915_gem_verify_gtt(dev);
 	return vma;
 
@@ -3535,7 +3538,9 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj,
 		}
 
 		list_for_each_entry(vma, &obj->vma_list, vma_link)
-			vma->bind_vma(vma, cache_level, 0);
+			if (drm_mm_node_allocated(&vma->node))
+				vma->bind_vma(vma, cache_level,
+					      obj->has_global_gtt_mapping ? GLOBAL_BIND : 0);
 	}
 
 	list_for_each_entry(vma, &obj->vma_list, vma_link)
@@ -3886,22 +3891,21 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
 			ret = i915_vma_unbind(vma);
 			if (ret)
 				return ret;
+
+			vma = NULL;
 		}
 	}
 
-	if (!i915_gem_obj_bound(obj, vm)) {
-
+	if (vma == NULL || !drm_mm_node_allocated(&vma->node)) {
 		vma = i915_gem_object_bind_to_vm(obj, vm, alignment, flags);
 		if (IS_ERR(vma))
 			return PTR_ERR(vma);
 	}
 
-	vma = i915_gem_obj_to_vma(obj, vm);
-
 	if (flags & PIN_GLOBAL && !obj->has_global_gtt_mapping)
 		vma->bind_vma(vma, obj->cache_level, GLOBAL_BIND);
 
-	i915_gem_obj_to_vma(obj, vm)->pin_count++;
+	vma->pin_count++;
 	if (flags & PIN_MAPPABLE)
 		obj->pin_mappable |= true;
 
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 9399a6fa4c2f..d7229ad2bd22 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -589,8 +589,6 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,
 		obj->base.pending_write_domain = I915_GEM_DOMAIN_RENDER;
 	}
 
-	vma->bind_vma(vma, obj->cache_level, flags);
-
 	return 0;
 }
 
-- 
1.8.5.2

Re: [PATCH 1/9] drm/i915: Consolidate binding parameters into flags

[Chris Wilson]

On Mon, Jan 27, 2014 at 07:10:51PM +0100, Daniel Vetter wrote:
> Anything more than just one bool parameter is just a pain to read,
> symbolic constants are much better.

This patch introduces PIN_GLOBAL which is required for ggtt_pin to
remain correct with the flag changes in the later patches, but
PIN_GLOBAL is only implemented in the final patch, patch 9.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

Re: [PATCH 1/9] drm/i915: Consolidate binding parameters into flags

[Daniel Vetter]

On Mon, Jan 27, 2014 at 09:47:12PM +0000, Chris Wilson wrote:
> On Mon, Jan 27, 2014 at 07:10:51PM +0100, Daniel Vetter wrote:
> > Anything more than just one bool parameter is just a pain to read,
> > symbolic constants are much better.
> 
> This patch introduces PIN_GLOBAL which is required for ggtt_pin to
> remain correct with the flag changes in the later patches, but
> PIN_GLOBAL is only implemented in the final patch, patch 9.

Indeed I've fumbled this. Will reshuffle.

The entire global/ppgtt pin stuff is fairly messy in the current code
anyway, dunno whether we can do better. The code as-is just feels like
it's hiding things a bit ...
-Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch

[PATCH] drm/i915: Consolidate binding parameters into flags

[Daniel Vetter]

Anything more than just one bool parameter is just a pain to read,
symbolic constants are much better.

Split out from Chris' vma-binding rework patch.

v2: Undo the behaviour change in object_pin that Chris spotted.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_drv.h            | 15 ++++----
 drivers/gpu/drm/i915/i915_gem.c            | 62 ++++++++++++------------------
 drivers/gpu/drm/i915/i915_gem_context.c    |  9 ++---
 drivers/gpu/drm/i915/i915_gem_evict.c      | 10 ++---
 drivers/gpu/drm/i915/i915_gem_execbuffer.c | 16 +++++---
 drivers/gpu/drm/i915/i915_gem_gtt.c        |  2 +-
 drivers/gpu/drm/i915/i915_trace.h          | 20 +++++-----
 drivers/gpu/drm/i915/intel_overlay.c       |  2 +-
 drivers/gpu/drm/i915/intel_pm.c            |  2 +-
 drivers/gpu/drm/i915/intel_ringbuffer.c    | 15 ++++----
 10 files changed, 71 insertions(+), 82 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 3971e7c3943d..dcab2b0a09ef 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -2030,11 +2030,13 @@ void i915_init_vm(struct drm_i915_private *dev_priv,
 void i915_gem_free_object(struct drm_gem_object *obj);
 void i915_gem_vma_destroy(struct i915_vma *vma);
 
+#define PIN_MAPPABLE 0x1
+#define PIN_GLOBAL 0x2
+#define PIN_NONBLOCK 0x4
 int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,
 				     struct i915_address_space *vm,
 				     uint32_t alignment,
-				     bool map_and_fenceable,
-				     bool nonblocking);
+				     unsigned flags);
 void i915_gem_object_ggtt_unpin(struct drm_i915_gem_object *obj);
 int __must_check i915_vma_unbind(struct i915_vma *vma);
 int __must_check i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj);
@@ -2237,11 +2239,9 @@ i915_gem_obj_ggtt_size(struct drm_i915_gem_object *obj)
 static inline int __must_check
 i915_gem_obj_ggtt_pin(struct drm_i915_gem_object *obj,
 		      uint32_t alignment,
-		      bool map_and_fenceable,
-		      bool nonblocking)
+		      unsigned flags)
 {
-	return i915_gem_object_pin(obj, obj_to_ggtt(obj), alignment,
-				   map_and_fenceable, nonblocking);
+	return i915_gem_object_pin(obj, obj_to_ggtt(obj), alignment, flags | PIN_GLOBAL);
 }
 
 /* i915_gem_context.c */
@@ -2280,8 +2280,7 @@ int __must_check i915_gem_evict_something(struct drm_device *dev,
 					  int min_size,
 					  unsigned alignment,
 					  unsigned cache_level,
-					  bool mappable,
-					  bool nonblock);
+					  unsigned flags);
 int i915_gem_evict_vm(struct i915_address_space *vm, bool do_idle);
 int i915_gem_evict_everything(struct drm_device *dev);
 
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 39770f7b333f..50d7be1c4741 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -43,12 +43,6 @@ static void i915_gem_object_flush_cpu_write_domain(struct drm_i915_gem_object *o
 static __must_check int
 i915_gem_object_wait_rendering(struct drm_i915_gem_object *obj,
 			       bool readonly);
-static __must_check int
-i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
-			   struct i915_address_space *vm,
-			   unsigned alignment,
-			   bool map_and_fenceable,
-			   bool nonblocking);
 static int i915_gem_phys_pwrite(struct drm_device *dev,
 				struct drm_i915_gem_object *obj,
 				struct drm_i915_gem_pwrite *args,
@@ -605,7 +599,7 @@ i915_gem_gtt_pwrite_fast(struct drm_device *dev,
 	char __user *user_data;
 	int page_offset, page_length, ret;
 
-	ret = i915_gem_obj_ggtt_pin(obj, 0, true, true);
+	ret = i915_gem_obj_ggtt_pin(obj, 0, PIN_MAPPABLE | PIN_NONBLOCK);
 	if (ret)
 		goto out;
 
@@ -1399,7 +1393,7 @@ int i915_gem_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 	}
 
 	/* Now bind it into the GTT if needed */
-	ret = i915_gem_obj_ggtt_pin(obj,  0, true, false);
+	ret = i915_gem_obj_ggtt_pin(obj, 0, PIN_MAPPABLE);
 	if (ret)
 		goto unlock;
 
@@ -2769,7 +2763,6 @@ int i915_vma_unbind(struct i915_vma *vma)
 
 	if (!drm_mm_node_allocated(&vma->node)) {
 		i915_gem_vma_destroy(vma);
-
 		return 0;
 	}
 
@@ -3267,14 +3260,13 @@ static int
 i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 			   struct i915_address_space *vm,
 			   unsigned alignment,
-			   bool map_and_fenceable,
-			   bool nonblocking)
+			   unsigned flags)
 {
 	struct drm_device *dev = obj->base.dev;
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	u32 size, fence_size, fence_alignment, unfenced_alignment;
 	size_t gtt_max =
-		map_and_fenceable ? dev_priv->gtt.mappable_end : vm->total;
+		flags & PIN_MAPPABLE ? dev_priv->gtt.mappable_end : vm->total;
 	struct i915_vma *vma;
 	int ret;
 
@@ -3286,18 +3278,18 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 						     obj->tiling_mode, true);
 	unfenced_alignment =
 		i915_gem_get_gtt_alignment(dev,
-						    obj->base.size,
-						    obj->tiling_mode, false);
+					   obj->base.size,
+					   obj->tiling_mode, false);
 
 	if (alignment == 0)
-		alignment = map_and_fenceable ? fence_alignment :
+		alignment = flags & PIN_MAPPABLE ? fence_alignment :
 						unfenced_alignment;
-	if (map_and_fenceable && alignment & (fence_alignment - 1)) {
+	if (flags & PIN_MAPPABLE && alignment & (fence_alignment - 1)) {
 		DRM_ERROR("Invalid object alignment requested %u\n", alignment);
 		return -EINVAL;
 	}
 
-	size = map_and_fenceable ? fence_size : obj->base.size;
+	size = flags & PIN_MAPPABLE ? fence_size : obj->base.size;
 
 	/* If the object is bigger than the entire aperture, reject it early
 	 * before evicting everything in a vain attempt to find space.
@@ -3305,7 +3297,7 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
 	if (obj->base.size > gtt_max) {
 		DRM_ERROR("Attempting to bind an object larger than the aperture: object=%zd > %s aperture=%zu\n",
 			  obj->base.size,
-			  map_and_fenceable ? "mappable" : "total",
+			  flags & PIN_MAPPABLE ? "mappable" : "total",
 			  gtt_max);
 		return -E2BIG;
 	}
@@ -3329,9 +3321,7 @@ search_free:
 						  DRM_MM_SEARCH_DEFAULT);
 	if (ret) {
 		ret = i915_gem_evict_something(dev, vm, size, alignment,
-					       obj->cache_level,
-					       map_and_fenceable,
-					       nonblocking);
+					       obj->cache_level, flags);
 		if (ret == 0)
 			goto search_free;
 
@@ -3362,9 +3352,9 @@ search_free:
 		obj->map_and_fenceable = mappable && fenceable;
 	}
 
-	WARN_ON(map_and_fenceable && !obj->map_and_fenceable);
+	WARN_ON(flags & PIN_MAPPABLE && !obj->map_and_fenceable);
 
-	trace_i915_vma_bind(vma, map_and_fenceable);
+	trace_i915_vma_bind(vma, flags);
 	i915_gem_verify_gtt(dev);
 	return 0;
 
@@ -3735,7 +3725,7 @@ i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj,
 	 * (e.g. libkms for the bootup splash), we have to ensure that we
 	 * always use map_and_fenceable for all scanout buffers.
 	 */
-	ret = i915_gem_obj_ggtt_pin(obj, alignment, true, false);
+	ret = i915_gem_obj_ggtt_pin(obj, alignment, PIN_MAPPABLE);
 	if (ret)
 		goto err_unpin_display;
 
@@ -3891,30 +3881,28 @@ int
 i915_gem_object_pin(struct drm_i915_gem_object *obj,
 		    struct i915_address_space *vm,
 		    uint32_t alignment,
-		    bool map_and_fenceable,
-		    bool nonblocking)
+		    unsigned flags)
 {
-	const u32 flags = map_and_fenceable ? GLOBAL_BIND : 0;
 	struct i915_vma *vma;
 	int ret;
 
-	WARN_ON(map_and_fenceable && !i915_is_ggtt(vm));
+	if (WARN_ON(flags & (PIN_GLOBAL | PIN_MAPPABLE) && !i915_is_ggtt(vm)))
+		return -EINVAL;
 
 	vma = i915_gem_obj_to_vma(obj, vm);
-
 	if (vma) {
 		if (WARN_ON(vma->pin_count == DRM_I915_GEM_OBJECT_MAX_PIN_COUNT))
 			return -EBUSY;
 
 		if ((alignment &&
 		     vma->node.start & (alignment - 1)) ||
-		    (map_and_fenceable && !obj->map_and_fenceable)) {
+		    (flags & PIN_MAPPABLE && !obj->map_and_fenceable)) {
 			WARN(vma->pin_count,
 			     "bo is already pinned with incorrect alignment:"
 			     " offset=%lx, req.alignment=%x, req.map_and_fenceable=%d,"
 			     " obj->map_and_fenceable=%d\n",
 			     i915_gem_obj_offset(obj, vm), alignment,
-			     map_and_fenceable,
+			     flags & PIN_MAPPABLE,
 			     obj->map_and_fenceable);
 			ret = i915_vma_unbind(vma);
 			if (ret)
@@ -3923,9 +3911,7 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
 	}
 
 	if (!i915_gem_obj_bound(obj, vm)) {
-		ret = i915_gem_object_bind_to_vm(obj, vm, alignment,
-						 map_and_fenceable,
-						 nonblocking);
+		ret = i915_gem_object_bind_to_vm(obj, vm, alignment, flags);
 		if (ret)
 			return ret;
 
@@ -3933,10 +3919,12 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
 
 	vma = i915_gem_obj_to_vma(obj, vm);
 
-	vma->bind_vma(vma, obj->cache_level, flags);
+	vma->bind_vma(vma, obj->cache_level,
+		      flags & PIN_GLOBAL ? GLOBAL_BIND : 0);
 
 	i915_gem_obj_to_vma(obj, vm)->pin_count++;
-	obj->pin_mappable |= map_and_fenceable;
+	if (flags & PIN_MAPPABLE)
+		obj->pin_mappable |= true;
 
 	return 0;
 }
@@ -3994,7 +3982,7 @@ i915_gem_pin_ioctl(struct drm_device *dev, void *data,
 	}
 
 	if (obj->user_pin_count == 0) {
-		ret = i915_gem_obj_ggtt_pin(obj, args->alignment, true, false);
+		ret = i915_gem_obj_ggtt_pin(obj, args->alignment, PIN_MAPPABLE);
 		if (ret)
 			goto out;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index fb64ab4a2068..1c9408230492 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -263,8 +263,7 @@ i915_gem_create_context(struct drm_device *dev,
 		 * context.
 		 */
 		ret = i915_gem_obj_ggtt_pin(ctx->obj,
-					    get_context_alignment(dev),
-					    false, false);
+					    get_context_alignment(dev), 0);
 		if (ret) {
 			DRM_DEBUG_DRIVER("Couldn't pin %d\n", ret);
 			goto err_destroy;
@@ -340,8 +339,7 @@ void i915_gem_context_reset(struct drm_device *dev)
 
 		if (i == RCS) {
 			WARN_ON(i915_gem_obj_ggtt_pin(dctx->obj,
-						      get_context_alignment(dev),
-						      false, false));
+						      get_context_alignment(dev), 0));
 			/* Fake a finish/inactive */
 			dctx->obj->base.write_domain = 0;
 			dctx->obj->active = 0;
@@ -613,8 +611,7 @@ static int do_switch(struct intel_ring_buffer *ring,
 	/* Trying to pin first makes error handling easier. */
 	if (ring == &dev_priv->ring[RCS]) {
 		ret = i915_gem_obj_ggtt_pin(to->obj,
-					    get_context_alignment(ring->dev),
-					    false, false);
+					    get_context_alignment(ring->dev), 0);
 		if (ret)
 			return ret;
 	}
diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c
index 4e82ca4a7a52..525b242ac512 100644
--- a/drivers/gpu/drm/i915/i915_gem_evict.c
+++ b/drivers/gpu/drm/i915/i915_gem_evict.c
@@ -50,7 +50,7 @@ mark_free(struct i915_vma *vma, struct list_head *unwind)
 int
 i915_gem_evict_something(struct drm_device *dev, struct i915_address_space *vm,
 			 int min_size, unsigned alignment, unsigned cache_level,
-			 bool mappable, bool nonblocking)
+			 unsigned flags)
 {
 	drm_i915_private_t *dev_priv = dev->dev_private;
 	struct list_head eviction_list, unwind_list;
@@ -58,7 +58,7 @@ i915_gem_evict_something(struct drm_device *dev, struct i915_address_space *vm,
 	int ret = 0;
 	int pass = 0;
 
-	trace_i915_gem_evict(dev, min_size, alignment, mappable);
+	trace_i915_gem_evict(dev, min_size, alignment, flags);
 
 	/*
 	 * The goal is to evict objects and amalgamate space in LRU order.
@@ -84,7 +84,7 @@ i915_gem_evict_something(struct drm_device *dev, struct i915_address_space *vm,
 	 */
 
 	INIT_LIST_HEAD(&unwind_list);
-	if (mappable) {
+	if (flags & PIN_MAPPABLE) {
 		BUG_ON(!i915_is_ggtt(vm));
 		drm_mm_init_scan_with_range(&vm->mm, min_size,
 					    alignment, cache_level, 0,
@@ -99,7 +99,7 @@ search_again:
 			goto found;
 	}
 
-	if (nonblocking)
+	if (flags & PIN_NONBLOCK)
 		goto none;
 
 	/* Now merge in the soon-to-be-expired objects... */
@@ -123,7 +123,7 @@ none:
 	/* Can we unpin some objects such as idle hw contents,
 	 * or pending flips?
 	 */
-	if (nonblocking)
+	if (flags & PIN_NONBLOCK)
 		return -ENOSPC;
 
 	/* Only idle the GPU and repeat the search once */
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 032def901f98..9399a6fa4c2f 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -544,19 +544,23 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,
 	struct drm_i915_gem_object *obj = vma->obj;
 	struct drm_i915_gem_exec_object2 *entry = vma->exec_entry;
 	bool has_fenced_gpu_access = INTEL_INFO(ring->dev)->gen < 4;
-	bool need_fence, need_mappable;
-	u32 flags = (entry->flags & EXEC_OBJECT_NEEDS_GTT) &&
-		!vma->obj->has_global_gtt_mapping ? GLOBAL_BIND : 0;
+	bool need_fence;
+	unsigned flags;
 	int ret;
 
+	flags = 0;
+
 	need_fence =
 		has_fenced_gpu_access &&
 		entry->flags & EXEC_OBJECT_NEEDS_FENCE &&
 		obj->tiling_mode != I915_TILING_NONE;
-	need_mappable = need_fence || need_reloc_mappable(vma);
+	if (need_fence || need_reloc_mappable(vma))
+		flags |= PIN_MAPPABLE;
+
+	if (entry->flags & EXEC_OBJECT_NEEDS_GTT)
+		flags |= PIN_GLOBAL;
 
-	ret = i915_gem_object_pin(obj, vma->vm, entry->alignment, need_mappable,
-				  false);
+	ret = i915_gem_object_pin(obj, vma->vm, entry->alignment, flags);
 	if (ret)
 		return ret;
 
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 6e858e17bb0c..154e1ad8bfab 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -868,7 +868,7 @@ alloc:
 	if (ret == -ENOSPC && !retried) {
 		ret = i915_gem_evict_something(dev, &dev_priv->gtt.base,
 					       GEN6_PD_SIZE, GEN6_PD_ALIGN,
-					       I915_CACHE_NONE, false, true);
+					       I915_CACHE_NONE, PIN_NONBLOCK);
 		if (ret)
 			return ret;
 
diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h
index 6e580c98dede..b95a380958db 100644
--- a/drivers/gpu/drm/i915/i915_trace.h
+++ b/drivers/gpu/drm/i915/i915_trace.h
@@ -34,15 +34,15 @@ TRACE_EVENT(i915_gem_object_create,
 );
 
 TRACE_EVENT(i915_vma_bind,
-	    TP_PROTO(struct i915_vma *vma, bool mappable),
-	    TP_ARGS(vma, mappable),
+	    TP_PROTO(struct i915_vma *vma, unsigned flags),
+	    TP_ARGS(vma, flags),
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
 			     __field(struct i915_address_space *, vm)
 			     __field(u32, offset)
 			     __field(u32, size)
-			     __field(bool, mappable)
+			     __field(unsigned, flags)
 			     ),
 
 	    TP_fast_assign(
@@ -50,12 +50,12 @@ TRACE_EVENT(i915_vma_bind,
 			   __entry->vm = vma->vm;
 			   __entry->offset = vma->node.start;
 			   __entry->size = vma->node.size;
-			   __entry->mappable = mappable;
+			   __entry->flags = flags;
 			   ),
 
 	    TP_printk("obj=%p, offset=%08x size=%x%s vm=%p",
 		      __entry->obj, __entry->offset, __entry->size,
-		      __entry->mappable ? ", mappable" : "",
+		      __entry->flags & PIN_MAPPABLE ? ", mappable" : "",
 		      __entry->vm)
 );
 
@@ -196,26 +196,26 @@ DEFINE_EVENT(i915_gem_object, i915_gem_object_destroy,
 );
 
 TRACE_EVENT(i915_gem_evict,
-	    TP_PROTO(struct drm_device *dev, u32 size, u32 align, bool mappable),
-	    TP_ARGS(dev, size, align, mappable),
+	    TP_PROTO(struct drm_device *dev, u32 size, u32 align, unsigned flags),
+	    TP_ARGS(dev, size, align, flags),
 
 	    TP_STRUCT__entry(
 			     __field(u32, dev)
 			     __field(u32, size)
 			     __field(u32, align)
-			     __field(bool, mappable)
+			     __field(unsigned, flags)
 			    ),
 
 	    TP_fast_assign(
 			   __entry->dev = dev->primary->index;
 			   __entry->size = size;
 			   __entry->align = align;
-			   __entry->mappable = mappable;
+			   __entry->flags = flags;
 			  ),
 
 	    TP_printk("dev=%d, size=%d, align=%d %s",
 		      __entry->dev, __entry->size, __entry->align,
-		      __entry->mappable ? ", mappable" : "")
+		      __entry->flags & PIN_MAPPABLE ? ", mappable" : "")
 );
 
 TRACE_EVENT(i915_gem_evict_everything,
diff --git a/drivers/gpu/drm/i915/intel_overlay.c b/drivers/gpu/drm/i915/intel_overlay.c
index 424f0946d8c4..ac519cb46f22 100644
--- a/drivers/gpu/drm/i915/intel_overlay.c
+++ b/drivers/gpu/drm/i915/intel_overlay.c
@@ -1349,7 +1349,7 @@ void intel_setup_overlay(struct drm_device *dev)
 		}
 		overlay->flip_addr = reg_bo->phys_obj->handle->busaddr;
 	} else {
-		ret = i915_gem_obj_ggtt_pin(reg_bo, PAGE_SIZE, true, false);
+		ret = i915_gem_obj_ggtt_pin(reg_bo, PAGE_SIZE, PIN_MAPPABLE);
 		if (ret) {
 			DRM_ERROR("failed to pin overlay register bo\n");
 			goto out_free_bo;
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 4960314a3611..33dc2dd29750 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -2741,7 +2741,7 @@ intel_alloc_context_page(struct drm_device *dev)
 		return NULL;
 	}
 
-	ret = i915_gem_obj_ggtt_pin(ctx, 4096, true, false);
+	ret = i915_gem_obj_ggtt_pin(ctx, 4096, PIN_MAPPABLE);
 	if (ret) {
 		DRM_ERROR("failed to pin power context: %d\n", ret);
 		goto err_unref;
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index d897a19f887f..a0793c929b95 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -531,9 +531,11 @@ init_pipe_control(struct intel_ring_buffer *ring)
 		goto err;
 	}
 
-	i915_gem_object_set_cache_level(ring->scratch.obj, I915_CACHE_LLC);
+	ret = i915_gem_object_set_cache_level(ring->scratch.obj, I915_CACHE_LLC);
+	if (ret)
+		goto err_unref;
 
-	ret = i915_gem_obj_ggtt_pin(ring->scratch.obj, 4096, true, false);
+	ret = i915_gem_obj_ggtt_pin(ring->scratch.obj, 4096, 0);
 	if (ret)
 		goto err_unref;
 
@@ -1273,10 +1275,9 @@ static int init_status_page(struct intel_ring_buffer *ring)
 
 	i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
 
-	ret = i915_gem_obj_ggtt_pin(obj, 4096, true, false);
-	if (ret != 0) {
+	ret = i915_gem_obj_ggtt_pin(obj, 4096, PIN_MAPPABLE);
+	if (ret)
 		goto err_unref;
-	}
 
 	ring->status_page.gfx_addr = i915_gem_obj_ggtt_offset(obj);
 	ring->status_page.page_addr = kmap(sg_page(obj->pages->sgl));
@@ -1356,7 +1357,7 @@ static int intel_init_ring_buffer(struct drm_device *dev,
 
 	ring->obj = obj;
 
-	ret = i915_gem_obj_ggtt_pin(obj, PAGE_SIZE, true, false);
+	ret = i915_gem_obj_ggtt_pin(obj, PAGE_SIZE, PIN_MAPPABLE);
 	if (ret)
 		goto err_unref;
 
@@ -1933,7 +1934,7 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
 			return -ENOMEM;
 		}
 
-		ret = i915_gem_obj_ggtt_pin(obj, 0, true, false);
+		ret = i915_gem_obj_ggtt_pin(obj, 0, PIN_MAPPABLE);
 		if (ret != 0) {
 			drm_gem_object_unreference(&obj->base);
 			DRM_ERROR("Failed to ping batch bo\n");
-- 
1.8.5.2

[PATCH] drm/i915: Only bind each object rather than for every execbuffer

[Daniel Vetter]

From: Chris Wilson <chris@chris-wilson.co.uk>

One side-effect of the introduction of ppgtt was that we needed to
rebind the object into the appropriate vm (and global gtt in some
peculiar cases). For simplicity this was done twice for every object on
every call to execbuffer. However, that adds a tremendous amount of CPU
overhead (rewriting all the PTE for all objects into WC memory) per
draw. The fix is to push all the decision about which vm to bind into
and when down into the low-level bind routines through hints rather than
inside the execbuffer routine.

v2: Split out prep work and unrelated changes.

v3: Bring back functional change around PIN_GLOBAL that I've
accidentally split out.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=72906
Tested-by: jianx.zhou@intel.com
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> (v1)
Cc: Ben Widawsky <benjamin.widawsky@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
---
 drivers/gpu/drm/i915/i915_gem.c            | 20 ++++++++++++--------
 drivers/gpu/drm/i915/i915_gem_execbuffer.c |  2 --
 2 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index c732f20a86ca..05e79fa02db2 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3333,6 +3333,9 @@ search_free:
 	WARN_ON(flags & PIN_MAPPABLE && !obj->map_and_fenceable);
 
 	trace_i915_vma_bind(vma, flags);
+	vma->bind_vma(vma, obj->cache_level,
+		      flags & (PIN_MAPPABLE | PIN_GLOBAL) ? GLOBAL_BIND : 0);
+
 	i915_gem_verify_gtt(dev);
 	return vma;
 
@@ -3535,7 +3538,9 @@ int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj,
 		}
 
 		list_for_each_entry(vma, &obj->vma_list, vma_link)
-			vma->bind_vma(vma, cache_level, 0);
+			if (drm_mm_node_allocated(&vma->node))
+				vma->bind_vma(vma, cache_level,
+					      obj->has_global_gtt_mapping ? GLOBAL_BIND : 0);
 	}
 
 	list_for_each_entry(vma, &obj->vma_list, vma_link)
@@ -3886,22 +3891,21 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
 			ret = i915_vma_unbind(vma);
 			if (ret)
 				return ret;
+
+			vma = NULL;
 		}
 	}
 
-	if (!i915_gem_obj_bound(obj, vm)) {
-
+	if (vma == NULL || !drm_mm_node_allocated(&vma->node)) {
 		vma = i915_gem_object_bind_to_vm(obj, vm, alignment, flags);
 		if (IS_ERR(vma))
 			return PTR_ERR(vma);
 	}
 
-	vma = i915_gem_obj_to_vma(obj, vm);
-
-	vma->bind_vma(vma, obj->cache_level,
-		      flags & PIN_GLOBAL ? GLOBAL_BIND : 0);
+	if (flags & PIN_GLOBAL && !obj->has_global_gtt_mapping)
+		vma->bind_vma(vma, obj->cache_level, GLOBAL_BIND);
 
-	i915_gem_obj_to_vma(obj, vm)->pin_count++;
+	vma->pin_count++;
 	if (flags & PIN_MAPPABLE)
 		obj->pin_mappable |= true;
 
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index 9399a6fa4c2f..d7229ad2bd22 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -589,8 +589,6 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,
 		obj->base.pending_write_domain = I915_GEM_DOMAIN_RENDER;
 	}
 
-	vma->bind_vma(vma, obj->cache_level, flags);
-
 	return 0;
 }
 
-- 
1.8.5.2

Re: [PATCH] drm/i915: Consolidate binding parameters into flags

[Jani Nikula]

On Tue, 28 Jan 2014, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> Anything more than just one bool parameter is just a pain to read,
> symbolic constants are much better.
>
> Split out from Chris' vma-binding rework patch.
>
> v2: Undo the behaviour change in object_pin that Chris spotted.
>
> Cc: Chris Wilson <chris@chris-wilson.co.uk>
> Cc: Ben Widawsky <benjamin.widawsky@intel.com>
> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

[snip]

> diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> index 032def901f98..9399a6fa4c2f 100644
> --- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> +++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
> @@ -544,19 +544,23 @@ i915_gem_execbuffer_reserve_vma(struct i915_vma *vma,
>  	struct drm_i915_gem_object *obj = vma->obj;
>  	struct drm_i915_gem_exec_object2 *entry = vma->exec_entry;
>  	bool has_fenced_gpu_access = INTEL_INFO(ring->dev)->gen < 4;
> -	bool need_fence, need_mappable;
> -	u32 flags = (entry->flags & EXEC_OBJECT_NEEDS_GTT) &&
> -		!vma->obj->has_global_gtt_mapping ? GLOBAL_BIND : 0;
> +	bool need_fence;
> +	unsigned flags;
>  	int ret;
>  
> +	flags = 0;
> +
>  	need_fence =
>  		has_fenced_gpu_access &&
>  		entry->flags & EXEC_OBJECT_NEEDS_FENCE &&
>  		obj->tiling_mode != I915_TILING_NONE;
> -	need_mappable = need_fence || need_reloc_mappable(vma);
> +	if (need_fence || need_reloc_mappable(vma))
> +		flags |= PIN_MAPPABLE;
> +
> +	if (entry->flags & EXEC_OBJECT_NEEDS_GTT)
> +		flags |= PIN_GLOBAL;

It is not obvious to me that this together with the PIN_GLOBAL handling
in i915_gem_object_pin() do not introduce a functional change. (Stress
on obvious to _me_; it may be obvious to you.)

I would have thought it better to first change the two bool parameters
to two flags, and then add the new flag in a separate patch to not
confuse poor reviewers like myself.

[snip]

> diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
> index d897a19f887f..a0793c929b95 100644
> --- a/drivers/gpu/drm/i915/intel_ringbuffer.c
> +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
> @@ -531,9 +531,11 @@ init_pipe_control(struct intel_ring_buffer *ring)
>  		goto err;
>  	}
>  
> -	i915_gem_object_set_cache_level(ring->scratch.obj, I915_CACHE_LLC);
> +	ret = i915_gem_object_set_cache_level(ring->scratch.obj, I915_CACHE_LLC);
> +	if (ret)
> +		goto err_unref;

This should be split out from the patch just like patch 4/9.

BR,
Jani.


-- 
Jani Nikula, Intel Open Source Technology Center

Re: [PATCH 4/9] drm/i915: Handle set_cache_level errors in the status page setup

[Jani Nikula]

On Mon, 27 Jan 2014, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> Split out from Chris vma-bind rework.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>

> Cc: Chris Wilson <chris@chris-wilson.co.uk>
> Cc: Ben Widawsky <benjamin.widawsky@intel.com>
> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> ---
>  drivers/gpu/drm/i915/intel_ringbuffer.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
> index bc19261eb436..e75d8b325029 100644
> --- a/drivers/gpu/drm/i915/intel_ringbuffer.c
> +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
> @@ -1273,7 +1273,9 @@ static int init_status_page(struct intel_ring_buffer *ring)
>  		goto err;
>  	}
>  
> -	i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
> +	ret = i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
> +	if (ret)
> +		goto err_unref;
>  
>  	ret = i915_gem_obj_ggtt_pin(obj, 4096, 0);
>  	if (ret)
> -- 
> 1.8.5.2
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Jani Nikula, Intel Open Source Technology Center

Re: [PATCH 7/9] drm/i915: Simplify i915_gem_object_ggtt_unpin

[Jani Nikula]

On Mon, 27 Jan 2014, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> Split out from Chris vma-bind rework.

"simplify" does not explain to me *why* you can skip two checks in the
code.

BR,
Jani.

> Cc: Chris Wilson <chris@chris-wilson.co.uk>
> Cc: Ben Widawsky <benjamin.widawsky@intel.com>
> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> ---
>  drivers/gpu/drm/i915/i915_drv.h | 10 ++++++++--
>  drivers/gpu/drm/i915/i915_gem.c | 20 --------------------
>  2 files changed, 8 insertions(+), 22 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> index dcab2b0a09ef..13991b8ce602 100644
> --- a/drivers/gpu/drm/i915/i915_drv.h
> +++ b/drivers/gpu/drm/i915/i915_drv.h
> @@ -2037,9 +2037,7 @@ int __must_check i915_gem_object_pin(struct drm_i915_gem_object *obj,
>  				     struct i915_address_space *vm,
>  				     uint32_t alignment,
>  				     unsigned flags);
> -void i915_gem_object_ggtt_unpin(struct drm_i915_gem_object *obj);
>  int __must_check i915_vma_unbind(struct i915_vma *vma);
> -int __must_check i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj);
>  int i915_gem_object_put_pages(struct drm_i915_gem_object *obj);
>  void i915_gem_release_all_mmaps(struct drm_i915_private *dev_priv);
>  void i915_gem_release_mmap(struct drm_i915_gem_object *obj);
> @@ -2244,6 +2242,14 @@ i915_gem_obj_ggtt_pin(struct drm_i915_gem_object *obj,
>  	return i915_gem_object_pin(obj, obj_to_ggtt(obj), alignment, flags | PIN_GLOBAL);
>  }
>  
> +static inline int
> +i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj)
> +{
> +	return i915_vma_unbind(i915_gem_obj_to_ggtt(obj));
> +}
> +
> +void i915_gem_object_ggtt_unpin(struct drm_i915_gem_object *obj);
> +
>  /* i915_gem_context.c */
>  #define ctx_to_ppgtt(ctx) container_of((ctx)->vm, struct i915_hw_ppgtt, base)
>  int __must_check i915_gem_context_init(struct drm_device *dev);
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index 9f4dab7df5c0..680b300d7454 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -2814,26 +2814,6 @@ int i915_vma_unbind(struct i915_vma *vma)
>  	return 0;
>  }
>  
> -/**
> - * Unbinds an object from the global GTT aperture.
> - */
> -int
> -i915_gem_object_ggtt_unbind(struct drm_i915_gem_object *obj)
> -{
> -	struct drm_i915_private *dev_priv = obj->base.dev->dev_private;
> -	struct i915_address_space *ggtt = &dev_priv->gtt.base;
> -
> -	if (!i915_gem_obj_ggtt_bound(obj))
> -		return 0;
> -
> -	if (i915_gem_obj_to_ggtt(obj)->pin_count)
> -		return -EBUSY;
> -
> -	BUG_ON(obj->pages == NULL);
> -
> -	return i915_vma_unbind(i915_gem_obj_to_vma(obj, ggtt));
> -}
> -
>  int i915_gpu_idle(struct drm_device *dev)
>  {
>  	drm_i915_private_t *dev_priv = dev->dev_private;
> -- 
> 1.8.5.2
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Jani Nikula, Intel Open Source Technology Center

Re: [PATCH 8/9] drm/i915: Directly return the vma from bind_to_vm

[Jani Nikula]

On Mon, 27 Jan 2014, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> This is prep work for reworking the object_pin logic. Atm
> it still does a (now redundant) lookup of the vma. The next
> patch will fix this.
>
> Split out from Chris vma-bind rework.

Reviewed-by: Jani Nikula <jani.nikula@intel.com>

> Cc: Chris Wilson <chris@chris-wilson.co.uk>
> Cc: Ben Widawsky <benjamin.widawsky@intel.com>
> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> ---
>  drivers/gpu/drm/i915/i915_gem.c | 23 +++++++++++------------
>  1 file changed, 11 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index 680b300d7454..a3534246fbca 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -3236,7 +3236,7 @@ static void i915_gem_verify_gtt(struct drm_device *dev)
>  /**
>   * Finds free space in the GTT aperture and binds the object there.
>   */
> -static int
> +static struct i915_vma *
>  i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
>  			   struct i915_address_space *vm,
>  			   unsigned alignment,
> @@ -3266,7 +3266,7 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
>  						unfenced_alignment;
>  	if (flags & PIN_MAPPABLE && alignment & (fence_alignment - 1)) {
>  		DRM_ERROR("Invalid object alignment requested %u\n", alignment);
> -		return -EINVAL;
> +		return ERR_PTR(-EINVAL);
>  	}
>  
>  	size = flags & PIN_MAPPABLE ? fence_size : obj->base.size;
> @@ -3279,20 +3279,18 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
>  			  obj->base.size,
>  			  flags & PIN_MAPPABLE ? "mappable" : "total",
>  			  gtt_max);
> -		return -E2BIG;
> +		return ERR_PTR(-E2BIG);
>  	}
>  
>  	ret = i915_gem_object_get_pages(obj);
>  	if (ret)
> -		return ret;
> +		return ERR_PTR(ret);
>  
>  	i915_gem_object_pin_pages(obj);
>  
>  	vma = i915_gem_obj_lookup_or_create_vma(obj, vm);
> -	if (IS_ERR(vma)) {
> -		ret = PTR_ERR(vma);
> +	if (IS_ERR(vma))
>  		goto err_unpin;
> -	}
>  
>  search_free:
>  	ret = drm_mm_insert_node_in_range_generic(&vm->mm, &vma->node,
> @@ -3336,15 +3334,16 @@ search_free:
>  
>  	trace_i915_vma_bind(vma, flags);
>  	i915_gem_verify_gtt(dev);
> -	return 0;
> +	return vma;
>  
>  err_remove_node:
>  	drm_mm_remove_node(&vma->node);
>  err_free_vma:
>  	i915_gem_vma_destroy(vma);
> +	vma = ERR_PTR(ret);
>  err_unpin:
>  	i915_gem_object_unpin_pages(obj);
> -	return ret;
> +	return vma;
>  }
>  
>  bool
> @@ -3891,10 +3890,10 @@ i915_gem_object_pin(struct drm_i915_gem_object *obj,
>  	}
>  
>  	if (!i915_gem_obj_bound(obj, vm)) {
> -		ret = i915_gem_object_bind_to_vm(obj, vm, alignment, flags);
> -		if (ret)
> -			return ret;
>  
> +		vma = i915_gem_object_bind_to_vm(obj, vm, alignment, flags);
> +		if (IS_ERR(vma))
> +			return PTR_ERR(vma);
>  	}
>  
>  	vma = i915_gem_obj_to_vma(obj, vm);
> -- 
> 1.8.5.2
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Jani Nikula, Intel Open Source Technology Center

Re: Split-up of "drm/i915: Only bind each object rather than for every execbuffer"

[Jani Nikula]

On Mon, 27 Jan 2014, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> Apparently I've lost a game of chicken. For review I'd like to sign up Jani for
> the overall series and Ben to upgrade his ack on the actual bugfix to a full
> r-b.

I'd like someone else to look at patches 2, 3, 5 and 6.

BR,
Jani.


-- 
Jani Nikula, Intel Open Source Technology Center

Re: [PATCH 7/9] drm/i915: Simplify i915_gem_object_ggtt_unpin

[Chris Wilson]

On Tue, Jan 28, 2014 at 03:13:34PM +0200, Jani Nikula wrote:
> On Mon, 27 Jan 2014, Daniel Vetter <daniel.vetter@ffwll.ch> wrote:
> > Split out from Chris vma-bind rework.
> 
> "simplify" does not explain to me *why* you can skip two checks in the
> code.

Because they are all repeated in i915_vma_unbind().
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre