Merge tag 'drm-intel-next-2019-10-07' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

UAPI Changes:
- Never allow userptr into the mappable GGTT (Chris)
  No existing users. Avoid anyone from even trying to
  spare a deadlock scenario.

Cross-subsystem Changes:

Core Changes:

Driver Changes:

- Eliminate struct_mutex use as BKL! (Chris)
  Only used for execbuf serialisation.

- Initialize DDI TC and TBT ports (D-I) on Tigerlake (Lucas)
- Fix DKL link training for 2.7GHz and 1.62GHz (Jose)
- Add Tigerlake DKL PHY programming sequences (Clinton)
- Add Tigerlake Thunderbolt PLL divider values (Imre)

- drm/i915: Use helpers for drm_mm_node booleans (Chris)
- Restrict L3 remapping sysfs interface to dwords (Chris)
- Fix audio power up sequence for gen10+ display (Kai)
- Skip redundant execlist resubmission (Chris)
- Only unwedge if we can reset GPU first (Chris)
- Initialise breadcrumb lists on the virtual engine (Chris)
- Don't rely on kernel context existing during early errors (Matt A)
- Update Icelake+ MG_DP_MODE programming table (Clinton)
- Update DMC firmware for Icelake (Anusha)
- Downgrade DP MST error after unplugging TypeC cable (Srinivasan)
- Limit MST modes based on plane size too (Ville)
- Polish intel_tv_mode_valid() (Ville)
- Fix g4x sprite scaling stride check with GTT remapping (Ville)
- Don't advertize non-exisiting crtcs (Ville)
- Clean up encoder->crtc_mask setup (Ville)
- Use tc_port instead of port parameter to MG registers (Jose)
- Remove static variable for aux last status (Jani)
- Implement a better i945gm vblank irq vs. C-states workaround (Ville)

- Make the object creation interface consistent (CQ)
- Rename intel_vga_msr_write() to intel_vga_reset_io_mem() (Jani, Ville)
- Eliminate previous drm_dbg/drm_err usage (Jani)
- Move gmbus setup down to intel_modeset_init() (Jani)
- Abstract all vgaarb access to intel_vga.[ch] (Jani)
- Split out i915_switcheroo.[ch] from i915_drv.c (Jani)
- Use intel_gt in has_reset* (Chris)
- Eliminate return value for i915_gem_init_early (Matt A)
- Selftest improvements (Chris)
- Update HuC firmware header version number format (Daniele)

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191007134801.GA24313@jlahtine-desk.ger.corp.intel.com

1 files changed, 44 insertions, 39 deletions

diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
index b5f6937369ea..e8ddc2320efa 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c
@@ -252,6 +252,7 @@ struct i915_execbuffer {
 		bool has_fence : 1;
 		bool needs_unfenced : 1;
 
+		struct intel_context *ce;
 		struct i915_request *rq;
 		u32 *rq_cmd;
 		unsigned int rq_size;
@@ -697,7 +698,9 @@ static int eb_reserve(struct i915_execbuffer *eb)
 
 		case 1:
 			/* Too fragmented, unbind everything and retry */
+			mutex_lock(&eb->context->vm->mutex);
 			err = i915_gem_evict_vm(eb->context->vm);
+			mutex_unlock(&eb->context->vm->mutex);
 			if (err)
 				return err;
 			break;
@@ -725,7 +728,7 @@ static int eb_select_context(struct i915_execbuffer *eb)
 		return -ENOENT;
 
 	eb->gem_context = ctx;
-	if (ctx->vm)
+	if (rcu_access_pointer(ctx->vm))
 		eb->invalid_flags |= EXEC_OBJECT_NEEDS_GTT;
 
 	eb->context_flags = 0;
@@ -880,6 +883,9 @@ static void eb_destroy(const struct i915_execbuffer *eb)
 {
 	GEM_BUG_ON(eb->reloc_cache.rq);
 
+	if (eb->reloc_cache.ce)
+		intel_context_put(eb->reloc_cache.ce);
+
 	if (eb->lut_size > 0)
 		kfree(eb->buckets);
 }
@@ -903,6 +909,7 @@ 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.allocated = false;
+	cache->ce = NULL;
 	cache->rq = NULL;
 	cache->rq_size = 0;
 }
@@ -963,11 +970,13 @@ static void reloc_cache_reset(struct reloc_cache *cache)
 		intel_gt_flush_ggtt_writes(ggtt->vm.gt);
 		io_mapping_unmap_atomic((void __iomem *)vaddr);
 
-		if (cache->node.allocated) {
+		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);
 		}
@@ -1042,11 +1051,13 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj,
 					       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 {
@@ -1056,7 +1067,7 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj,
 	}
 
 	offset = cache->node.start;
-	if (cache->node.allocated) {
+	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);
@@ -1145,7 +1156,7 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb,
 	u32 *cmd;
 	int err;
 
-	pool = intel_engine_pool_get(&eb->engine->pool, PAGE_SIZE);
+	pool = intel_engine_get_pool(eb->engine, PAGE_SIZE);
 	if (IS_ERR(pool))
 		return PTR_ERR(pool);
 
@@ -1168,7 +1179,7 @@ static int __reloc_gpu_alloc(struct i915_execbuffer *eb,
 	if (err)
 		goto err_unmap;
 
-	rq = i915_request_create(eb->context);
+	rq = intel_context_create_request(cache->ce);
 	if (IS_ERR(rq)) {
 		err = PTR_ERR(rq);
 		goto err_unpin;
@@ -1239,6 +1250,29 @@ static u32 *reloc_gpu(struct i915_execbuffer *eb,
 		if (!intel_engine_can_store_dword(eb->engine))
 			return ERR_PTR(-ENODEV);
 
+		if (!cache->ce) {
+			struct intel_context *ce;
+
+			/*
+			 * The CS pre-parser can pre-fetch commands across
+			 * memory sync points and starting gen12 it is able to
+			 * pre-fetch across BB_START and BB_END boundaries
+			 * (within the same context). We therefore use a
+			 * separate context gen12+ to guarantee that the reloc
+			 * writes land before the parser gets to the target
+			 * memory location.
+			 */
+			if (cache->gen >= 12)
+				ce = intel_context_create(eb->context->gem_context,
+							  eb->engine);
+			else
+				ce = intel_context_get(eb->context);
+			if (IS_ERR(ce))
+				return ERR_CAST(ce);
+
+			cache->ce = ce;
+		}
+
 		err = __reloc_gpu_alloc(eb, vma, len);
 		if (unlikely(err))
 			return ERR_PTR(err);
@@ -1388,7 +1422,7 @@ eb_relocate_entry(struct i915_execbuffer *eb,
 		if (reloc->write_domain == I915_GEM_DOMAIN_INSTRUCTION &&
 		    IS_GEN(eb->i915, 6)) {
 			err = i915_vma_bind(target, target->obj->cache_level,
-					    PIN_GLOBAL);
+					    PIN_GLOBAL, NULL);
 			if (WARN_ONCE(err,
 				      "Unexpected failure to bind target VMA!"))
 				return err;
@@ -1961,7 +1995,7 @@ static struct i915_vma *eb_parse(struct i915_execbuffer *eb, bool is_master)
 	struct i915_vma *vma;
 	int err;
 
-	pool = intel_engine_pool_get(&eb->engine->pool, eb->batch_len);
+	pool = intel_engine_get_pool(eb->engine, eb->batch_len);
 	if (IS_ERR(pool))
 		return ERR_CAST(pool);
 
@@ -2112,35 +2146,6 @@ static struct i915_request *eb_throttle(struct intel_context *ce)
 	return i915_request_get(rq);
 }
 
-static int
-__eb_pin_context(struct i915_execbuffer *eb, struct intel_context *ce)
-{
-	int err;
-
-	if (likely(atomic_inc_not_zero(&ce->pin_count)))
-		return 0;
-
-	err = mutex_lock_interruptible(&eb->i915->drm.struct_mutex);
-	if (err)
-		return err;
-
-	err = __intel_context_do_pin(ce);
-	mutex_unlock(&eb->i915->drm.struct_mutex);
-
-	return err;
-}
-
-static void
-__eb_unpin_context(struct i915_execbuffer *eb, struct intel_context *ce)
-{
-	if (likely(atomic_add_unless(&ce->pin_count, -1, 1)))
-		return;
-
-	mutex_lock(&eb->i915->drm.struct_mutex);
-	intel_context_unpin(ce);
-	mutex_unlock(&eb->i915->drm.struct_mutex);
-}
-
 static int __eb_pin_engine(struct i915_execbuffer *eb, struct intel_context *ce)
 {
 	struct intel_timeline *tl;
@@ -2160,7 +2165,7 @@ static int __eb_pin_engine(struct i915_execbuffer *eb, struct intel_context *ce)
 	 * GGTT space, so do this first before we reserve a seqno for
 	 * ourselves.
 	 */
-	err = __eb_pin_context(eb, ce);
+	err = intel_context_pin(ce);
 	if (err)
 		return err;
 
@@ -2204,7 +2209,7 @@ err_exit:
 	intel_context_exit(ce);
 	intel_context_timeline_unlock(tl);
 err_unpin:
-	__eb_unpin_context(eb, ce);
+	intel_context_unpin(ce);
 	return err;
 }
 
@@ -2217,7 +2222,7 @@ static void eb_unpin_engine(struct i915_execbuffer *eb)
 	intel_context_exit(ce);
 	mutex_unlock(&tl->mutex);
 
-	__eb_unpin_context(eb, ce);
+	intel_context_unpin(ce);
 }
 
 static unsigned int