Merge tag 'drm-intel-next-2016-11-08' of git://anongit.freedesktop.org/git/drm-intel into drm-next

- gpu idling rework for s/r (Imre)
- vlv mappable scanout fix
- speed up probing in resume (Lyude)
- dp audio workarounds for gen9 (Dhinakaran)
- more conversion to using dev_priv internally (Ville)
- more gen9+ wm fixes and cleanups (Maarten)
- shrinker cleanup&fixes (Chris)
- reorg plane init code (Ville)
- implement support for multiple timelines (prep work for scheduler)
  from Chris and all
- untangle dev->struct_mutex locking as prep for multiple timelines
  (Chris)
- refactor bxt phy code and collect it all in intel_dpio_phy.c (Ander)
- another gvt with bugfixes all over from Zhenyu
- piles of lspcon fixes from Imre
- 90/270 rotation fixes (Ville)
- guc log buffer support (Akash+Sagar)
- fbc fixes from Paulo
- untangle rpm vs. tiling-fences/mmaps (Chris)
- fix atomic commit to wait on the right fences (Daniel Stone)

* tag 'drm-intel-next-2016-11-08' of git://anongit.freedesktop.org/git/drm-intel: (181 commits)
  drm/i915: Update DRIVER_DATE to 20161108
  drm/i915: Mark CPU cache as dirty when used for rendering
  drm/i915: Add assert for no pending GPU requests during suspend/resume in LR mode
  drm/i915: Make sure engines are idle during GPU idling in LR mode
  drm/i915: Avoid early GPU idling due to race with new request
  drm/i915: Avoid early GPU idling due to already pending idle work
  drm/i915: Limit Valleyview and earlier to only using mappable scanout
  drm/i915: Round tile chunks up for constructing partial VMAs
  drm/i915: Remove the vma from the object list upon close
  drm/i915: Reinit polling before hpd when resuming
  drm/i915: Remove redundant reprobe in i915_drm_resume
  drm/i915/dp: Extend BDW DP audio workaround to GEN9 platforms
  drm/i915/dp: BDW cdclk fix for DP audio
  drm/i915: Fix pages pin counting around swizzle quirk
  drm/i915: Fix test on inputs for vma_compare()
  drm/i915/guc: Cache the client mapping
  drm/i915: Tidy slab cache allocations
  drm/i915: Introduce HAS_64BIT_RELOC
  drm/i915: Show the execlist queue in debugfs/i915_engine_info
  drm/i915: Unify global_list into global_link
  ...

1 files changed, 22 insertions, 55 deletions

diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index e52affdcc125..fb5b44339f71 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -34,7 +34,6 @@
 #include <drm/i915_drm.h>
 
 #include "i915_drv.h"
-#include "i915_gem_dmabuf.h"
 #include "i915_trace.h"
 #include "intel_drv.h"
 #include "intel_frontbuffer.h"
@@ -332,7 +331,8 @@ static void reloc_cache_init(struct reloc_cache *cache,
 	cache->page = -1;
 	cache->vaddr = 0;
 	cache->i915 = i915;
-	cache->use_64bit_reloc = INTEL_GEN(cache->i915) >= 8;
+	/* Must be a variable in the struct to allow GCC to unroll. */
+	cache->use_64bit_reloc = HAS_64BIT_RELOC(i915);
 	cache->node.allocated = false;
 }
 
@@ -418,15 +418,6 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj,
 	unsigned long offset;
 	void *vaddr;
 
-	if (cache->node.allocated) {
-		wmb();
-		ggtt->base.insert_page(&ggtt->base,
-				       i915_gem_object_get_dma_address(obj, page),
-				       cache->node.start, I915_CACHE_NONE, 0);
-		cache->page = page;
-		return unmask_page(cache->vaddr);
-	}
-
 	if (cache->vaddr) {
 		io_mapping_unmap_atomic((void __force __iomem *) unmask_page(cache->vaddr));
 	} else {
@@ -466,6 +457,7 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj,
 
 	offset = cache->node.start;
 	if (cache->node.allocated) {
+		wmb();
 		ggtt->base.insert_page(&ggtt->base,
 				       i915_gem_object_get_dma_address(obj, page),
 				       offset, I915_CACHE_NONE, 0);
@@ -1109,44 +1101,20 @@ err:
 	return ret;
 }
 
-static unsigned int eb_other_engines(struct drm_i915_gem_request *req)
-{
-	unsigned int mask;
-
-	mask = ~intel_engine_flag(req->engine) & I915_BO_ACTIVE_MASK;
-	mask <<= I915_BO_ACTIVE_SHIFT;
-
-	return mask;
-}
-
 static int
 i915_gem_execbuffer_move_to_gpu(struct drm_i915_gem_request *req,
 				struct list_head *vmas)
 {
-	const unsigned int other_rings = eb_other_engines(req);
 	struct i915_vma *vma;
 	int ret;
 
 	list_for_each_entry(vma, vmas, exec_list) {
 		struct drm_i915_gem_object *obj = vma->obj;
-		struct reservation_object *resv;
 
-		if (obj->flags & other_rings) {
-			ret = i915_gem_request_await_object
-				(req, obj, obj->base.pending_write_domain);
-			if (ret)
-				return ret;
-		}
-
-		resv = i915_gem_object_get_dmabuf_resv(obj);
-		if (resv) {
-			ret = i915_sw_fence_await_reservation
-				(&req->submit, resv, &i915_fence_ops,
-				 obj->base.pending_write_domain, 10*HZ,
-				 GFP_KERNEL | __GFP_NOWARN);
-			if (ret < 0)
-				return ret;
-		}
+		ret = i915_gem_request_await_object
+			(req, obj, obj->base.pending_write_domain);
+		if (ret)
+			return ret;
 
 		if (obj->base.write_domain & I915_GEM_DOMAIN_CPU)
 			i915_gem_clflush_object(obj, false);
@@ -1279,6 +1247,12 @@ i915_gem_validate_context(struct drm_device *dev, struct drm_file *file,
 	return ctx;
 }
 
+static bool gpu_write_needs_clflush(struct drm_i915_gem_object *obj)
+{
+	return !(obj->cache_level == I915_CACHE_NONE ||
+		 obj->cache_level == I915_CACHE_WT);
+}
+
 void i915_vma_move_to_active(struct i915_vma *vma,
 			     struct drm_i915_gem_request *req,
 			     unsigned int flags)
@@ -1288,8 +1262,6 @@ void i915_vma_move_to_active(struct i915_vma *vma,
 
 	GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
 
-	obj->dirty = 1; /* be paranoid  */
-
 	/* Add a reference if we're newly entering the active list.
 	 * The order in which we add operations to the retirement queue is
 	 * vital here: mark_active adds to the start of the callback list,
@@ -1297,37 +1269,32 @@ void i915_vma_move_to_active(struct i915_vma *vma,
 	 * add the active reference first and queue for it to be dropped
 	 * *last*.
 	 */
-	if (!i915_gem_object_is_active(obj))
-		i915_gem_object_get(obj);
-	i915_gem_object_set_active(obj, idx);
-	i915_gem_active_set(&obj->last_read[idx], req);
+	if (!i915_vma_is_active(vma))
+		obj->active_count++;
+	i915_vma_set_active(vma, idx);
+	i915_gem_active_set(&vma->last_read[idx], req);
+	list_move_tail(&vma->vm_link, &vma->vm->active_list);
 
 	if (flags & EXEC_OBJECT_WRITE) {
-		i915_gem_active_set(&obj->last_write, req);
+		i915_gem_active_set(&vma->last_write, req);
 
 		intel_fb_obj_invalidate(obj, ORIGIN_CS);
 
 		/* update for the implicit flush after a batch */
 		obj->base.write_domain &= ~I915_GEM_GPU_DOMAINS;
+		if (!obj->cache_dirty && gpu_write_needs_clflush(obj))
+			obj->cache_dirty = true;
 	}
 
 	if (flags & EXEC_OBJECT_NEEDS_FENCE)
 		i915_gem_active_set(&vma->last_fence, req);
-
-	i915_vma_set_active(vma, idx);
-	i915_gem_active_set(&vma->last_read[idx], req);
-	list_move_tail(&vma->vm_link, &vma->vm->active_list);
 }
 
 static void eb_export_fence(struct drm_i915_gem_object *obj,
 			    struct drm_i915_gem_request *req,
 			    unsigned int flags)
 {
-	struct reservation_object *resv;
-
-	resv = i915_gem_object_get_dmabuf_resv(obj);
-	if (!resv)
-		return;
+	struct reservation_object *resv = obj->resv;
 
 	/* Ignore errors from failing to allocate the new fence, we can't
 	 * handle an error right now. Worst case should be missed