Merge tag 'drm-intel-next-2017-10-12' of git://anongit.freedesktop.org/drm/drm-intel into drm-next

Last batch of drm/i915 features for v4.15:

- transparent huge pages support (Matthew)
- uapi: I915_PARAM_HAS_SCHEDULER into a capability bitmask (Chris)
- execlists: preemption (Chris)
- scheduler: user defined priorities (Chris)
- execlists optimization (Michał)
- plenty of display fixes (Imre)
- has_ipc fix (Rodrigo)
- platform features definition refactoring (Rodrigo)
- legacy cursor update fix (Maarten)
- fix vblank waits for cursor updates (Maarten)
- reprogram dmc firmware on resume, dmc state fix (Imre)
- remove use_mmio_flip module parameter (Maarten)
- wa fixes (Oscar)
- huc/guc firmware refacoring (Sagar, Michal)
- push encoder specific code to encoder hooks (Jani)
- DP MST fixes (Dhinakaran)
- eDP power sequencing fixes (Manasi)
- selftest updates (Chris, Matthew)
- mmu notifier cpu hotplug deadlock fix (Daniel)
- more VBT parser refactoring (Jani)
- max pipe refactoring (Mika Kahola)
- rc6/rps refactoring and separation (Sagar)
- userptr lockdep fix (Chris)
- tracepoint fixes and defunct tracepoint removal (Chris)
- use rcu instead of abusing stop_machine (Daniel)
- plenty of other fixes all around (Everyone)

* tag 'drm-intel-next-2017-10-12' of git://anongit.freedesktop.org/drm/drm-intel: (145 commits)
  drm/i915: Update DRIVER_DATE to 20171012
  drm/i915: Simplify intel_sanitize_enable_ppgtt
  drm/i915/userptr: Drop struct_mutex before cleanup
  drm/i915/dp: limit sink rates based on rate
  drm/i915/dp: centralize max source rate conditions more
  drm/i915: Allow PCH platforms fall back to BIOS LVDS mode
  drm/i915: Reuse normal state readout for LVDS/DVO fixed mode
  drm/i915: Use rcu instead of stop_machine in set_wedged
  drm/i915: Introduce separate status variable for RC6 and LLC ring frequency setup
  drm/i915: Create generic functions to control RC6, RPS
  drm/i915: Create generic function to setup LLC ring frequency table
  drm/i915: Rename intel_enable_rc6 to intel_rc6_enabled
  drm/i915: Name structure in dev_priv that contains RPS/RC6 state as "gt_pm"
  drm/i915: Move rps.hw_lock to dev_priv and s/hw_lock/pcu_lock
  drm/i915: Name i915_runtime_pm structure in dev_priv as "runtime_pm"
  drm/i915: Separate RPS and RC6 handling for CHV
  drm/i915: Separate RPS and RC6 handling for VLV
  drm/i915: Separate RPS and RC6 handling for BDW
  drm/i915: Remove superfluous IS_BDW checks and non-BDW changes from gen8_enable_rps
  drm/i915: Separate RPS and RC6 handling for gen6+
  ...

85 files changed, 5520 insertions, 2212 deletions

diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile
index 5182e3d5557d..66d23b619db1 100644
--- a/drivers/gpu/drm/i915/Makefile
+++ b/drivers/gpu/drm/i915/Makefile
@@ -47,6 +47,7 @@ i915-y += i915_cmd_parser.o \
 	  i915_gem_tiling.o \
 	  i915_gem_timeline.o \
 	  i915_gem_userptr.o \
+	  i915_gemfs.o \
 	  i915_trace_points.o \
 	  i915_vma.o \
 	  intel_breadcrumbs.o \
@@ -59,6 +60,8 @@ i915-y += i915_cmd_parser.o \
 
 # general-purpose microcontroller (GuC) support
 i915-y += intel_uc.o \
+	  intel_uc_fw.o \
+	  intel_guc.o \
 	  intel_guc_ct.o \
 	  intel_guc_log.o \
 	  intel_guc_loader.o \
diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c
index d5892d24f0b6..f6ded475bb2c 100644
--- a/drivers/gpu/drm/i915/gvt/scheduler.c
+++ b/drivers/gpu/drm/i915/gvt/scheduler.c
@@ -174,6 +174,7 @@ static int shadow_context_status_change(struct notifier_block *nb,
 		atomic_set(&workload->shadow_ctx_active, 1);
 		break;
 	case INTEL_CONTEXT_SCHEDULE_OUT:
+	case INTEL_CONTEXT_SCHEDULE_PREEMPTED:
 		atomic_set(&workload->shadow_ctx_active, 0);
 		break;
 	default:
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index b4a6ac60e7c6..0bb6e01121fc 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -30,6 +30,7 @@
 #include <linux/sort.h>
 #include <linux/sched/mm.h>
 #include "intel_drv.h"
+#include "i915_guc_submission.h"
 
 static inline struct drm_i915_private *node_to_i915(struct drm_info_node *node)
 {
@@ -97,7 +98,7 @@ static char get_tiling_flag(struct drm_i915_gem_object *obj)
 
 static char get_global_flag(struct drm_i915_gem_object *obj)
 {
-	return !list_empty(&obj->userfault_link) ? 'g' : ' ';
+	return obj->userfault_count ? 'g' : ' ';
 }
 
 static char get_pin_mapped_flag(struct drm_i915_gem_object *obj)
@@ -118,6 +119,36 @@ static u64 i915_gem_obj_total_ggtt_size(struct drm_i915_gem_object *obj)
 	return size;
 }
 
+static const char *
+stringify_page_sizes(unsigned int page_sizes, char *buf, size_t len)
+{
+	size_t x = 0;
+
+	switch (page_sizes) {
+	case 0:
+		return "";
+	case I915_GTT_PAGE_SIZE_4K:
+		return "4K";
+	case I915_GTT_PAGE_SIZE_64K:
+		return "64K";
+	case I915_GTT_PAGE_SIZE_2M:
+		return "2M";
+	default:
+		if (!buf)
+			return "M";
+
+		if (page_sizes & I915_GTT_PAGE_SIZE_2M)
+			x += snprintf(buf + x, len - x, "2M, ");
+		if (page_sizes & I915_GTT_PAGE_SIZE_64K)
+			x += snprintf(buf + x, len - x, "64K, ");
+		if (page_sizes & I915_GTT_PAGE_SIZE_4K)
+			x += snprintf(buf + x, len - x, "4K, ");
+		buf[x-2] = '\0';
+
+		return buf;
+	}
+}
+
 static void
 describe_obj(struct seq_file *m, struct drm_i915_gem_object *obj)
 {
@@ -155,9 +186,10 @@ describe_obj(struct seq_file *m, struct drm_i915_gem_object *obj)
 		if (!drm_mm_node_allocated(&vma->node))
 			continue;
 
-		seq_printf(m, " (%sgtt offset: %08llx, size: %08llx",
+		seq_printf(m, " (%sgtt offset: %08llx, size: %08llx, pages: %s",
 			   i915_vma_is_ggtt(vma) ? "g" : "pp",
-			   vma->node.start, vma->node.size);
+			   vma->node.start, vma->node.size,
+			   stringify_page_sizes(vma->page_sizes.gtt, NULL, 0));
 		if (i915_vma_is_ggtt(vma)) {
 			switch (vma->ggtt_view.type) {
 			case I915_GGTT_VIEW_NORMAL:
@@ -402,10 +434,12 @@ static int i915_gem_object_info(struct seq_file *m, void *data)
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
 	struct drm_device *dev = &dev_priv->drm;
 	struct i915_ggtt *ggtt = &dev_priv->ggtt;
-	u32 count, mapped_count, purgeable_count, dpy_count;
-	u64 size, mapped_size, purgeable_size, dpy_size;
+	u32 count, mapped_count, purgeable_count, dpy_count, huge_count;
+	u64 size, mapped_size, purgeable_size, dpy_size, huge_size;
 	struct drm_i915_gem_object *obj;
+	unsigned int page_sizes = 0;
 	struct drm_file *file;
+	char buf[80];
 	int ret;
 
 	ret = mutex_lock_interruptible(&dev->struct_mutex);
@@ -419,6 +453,7 @@ static int i915_gem_object_info(struct seq_file *m, void *data)
 	size = count = 0;
 	mapped_size = mapped_count = 0;
 	purgeable_size = purgeable_count = 0;
+	huge_size = huge_count = 0;
 	list_for_each_entry(obj, &dev_priv->mm.unbound_list, global_link) {
 		size += obj->base.size;
 		++count;
@@ -432,6 +467,12 @@ static int i915_gem_object_info(struct seq_file *m, void *data)
 			mapped_count++;
 			mapped_size += obj->base.size;
 		}
+
+		if (obj->mm.page_sizes.sg > I915_GTT_PAGE_SIZE) {
+			huge_count++;
+			huge_size += obj->base.size;
+			page_sizes |= obj->mm.page_sizes.sg;
+		}
 	}
 	seq_printf(m, "%u unbound objects, %llu bytes\n", count, size);
 
@@ -454,6 +495,12 @@ static int i915_gem_object_info(struct seq_file *m, void *data)
 			mapped_count++;
 			mapped_size += obj->base.size;
 		}
+
+		if (obj->mm.page_sizes.sg > I915_GTT_PAGE_SIZE) {
+			huge_count++;
+			huge_size += obj->base.size;
+			page_sizes |= obj->mm.page_sizes.sg;
+		}
 	}
 	seq_printf(m, "%u bound objects, %llu bytes\n",
 		   count, size);
@@ -461,11 +508,18 @@ static int i915_gem_object_info(struct seq_file *m, void *data)
 		   purgeable_count, purgeable_size);
 	seq_printf(m, "%u mapped objects, %llu bytes\n",
 		   mapped_count, mapped_size);
+	seq_printf(m, "%u huge-paged objects (%s) %llu bytes\n",
+		   huge_count,
+		   stringify_page_sizes(page_sizes, buf, sizeof(buf)),
+		   huge_size);
 	seq_printf(m, "%u display objects (pinned), %llu bytes\n",
 		   dpy_count, dpy_size);
 
 	seq_printf(m, "%llu [%llu] gtt total\n",
 		   ggtt->base.total, ggtt->mappable_end);
+	seq_printf(m, "Supported page sizes: %s\n",
+		   stringify_page_sizes(INTEL_INFO(dev_priv)->page_sizes,
+					buf, sizeof(buf)));
 
 	seq_putc(m, '\n');
 	print_batch_pool_stats(m, dev_priv);
@@ -1026,6 +1080,7 @@ DEFINE_SIMPLE_ATTRIBUTE(i915_next_seqno_fops,
 static int i915_frequency_info(struct seq_file *m, void *unused)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	int ret = 0;
 
 	intel_runtime_pm_get(dev_priv);
@@ -1041,9 +1096,19 @@ static int i915_frequency_info(struct seq_file *m, void *unused)
 		seq_printf(m, "Current P-state: %d\n",
 			   (rgvstat & MEMSTAT_PSTATE_MASK) >> MEMSTAT_PSTATE_SHIFT);
 	} else if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) {
-		u32 freq_sts;
+		u32 rpmodectl, freq_sts;
+
+		mutex_lock(&dev_priv->pcu_lock);
+
+		rpmodectl = I915_READ(GEN6_RP_CONTROL);
+		seq_printf(m, "Video Turbo Mode: %s\n",
+			   yesno(rpmodectl & GEN6_RP_MEDIA_TURBO));
+		seq_printf(m, "HW control enabled: %s\n",
+			   yesno(rpmodectl & GEN6_RP_ENABLE));
+		seq_printf(m, "SW control enabled: %s\n",
+			   yesno((rpmodectl & GEN6_RP_MEDIA_MODE_MASK) ==
+				  GEN6_RP_MEDIA_SW_MODE));
 
-		mutex_lock(&dev_priv->rps.hw_lock);
 		freq_sts = vlv_punit_read(dev_priv, PUNIT_REG_GPU_FREQ_STS);
 		seq_printf(m, "PUNIT_REG_GPU_FREQ_STS: 0x%08x\n", freq_sts);
 		seq_printf(m, "DDR freq: %d MHz\n", dev_priv->mem_freq);
@@ -1052,21 +1117,21 @@ static int i915_frequency_info(struct seq_file *m, void *unused)
 			   intel_gpu_freq(dev_priv, (freq_sts >> 8) & 0xff));
 
 		seq_printf(m, "current GPU freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.cur_freq));
+			   intel_gpu_freq(dev_priv, rps->cur_freq));
 
 		seq_printf(m, "max GPU freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.max_freq));
+			   intel_gpu_freq(dev_priv, rps->max_freq));
 
 		seq_printf(m, "min GPU freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.min_freq));
+			   intel_gpu_freq(dev_priv, rps->min_freq));
 
 		seq_printf(m, "idle GPU freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.idle_freq));
+			   intel_gpu_freq(dev_priv, rps->idle_freq));
 
 		seq_printf(m,
 			   "efficient (RPe) frequency: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.efficient_freq));
-		mutex_unlock(&dev_priv->rps.hw_lock);
+			   intel_gpu_freq(dev_priv, rps->efficient_freq));
+		mutex_unlock(&dev_priv->pcu_lock);
 	} else if (INTEL_GEN(dev_priv) >= 6) {
 		u32 rp_state_limits;
 		u32 gt_perf_status;
@@ -1136,10 +1201,17 @@ static int i915_frequency_info(struct seq_file *m, void *unused)
 			pm_iir = I915_READ(GEN8_GT_IIR(2));
 			pm_mask = I915_READ(GEN6_PMINTRMSK);
 		}
+		seq_printf(m, "Video Turbo Mode: %s\n",
+			   yesno(rpmodectl & GEN6_RP_MEDIA_TURBO));
+		seq_printf(m, "HW control enabled: %s\n",
+			   yesno(rpmodectl & GEN6_RP_ENABLE));
+		seq_printf(m, "SW control enabled: %s\n",
+			   yesno((rpmodectl & GEN6_RP_MEDIA_MODE_MASK) ==
+				  GEN6_RP_MEDIA_SW_MODE));
 		seq_printf(m, "PM IER=0x%08x IMR=0x%08x ISR=0x%08x IIR=0x%08x, MASK=0x%08x\n",
 			   pm_ier, pm_imr, pm_isr, pm_iir, pm_mask);
 		seq_printf(m, "pm_intrmsk_mbz: 0x%08x\n",
-			   dev_priv->rps.pm_intrmsk_mbz);
+			   rps->pm_intrmsk_mbz);
 		seq_printf(m, "GT_PERF_STATUS: 0x%08x\n", gt_perf_status);
 		seq_printf(m, "Render p-state ratio: %d\n",
 			   (gt_perf_status & (INTEL_GEN(dev_priv) >= 9 ? 0x1ff00 : 0xff00)) >> 8);
@@ -1159,8 +1231,7 @@ static int i915_frequency_info(struct seq_file *m, void *unused)
 			   rpcurup, GT_PM_INTERVAL_TO_US(dev_priv, rpcurup));
 		seq_printf(m, "RP PREV UP: %d (%dus)\n",
 			   rpprevup, GT_PM_INTERVAL_TO_US(dev_priv, rpprevup));
-		seq_printf(m, "Up threshold: %d%%\n",
-			   dev_priv->rps.up_threshold);
+		seq_printf(m, "Up threshold: %d%%\n", rps->up_threshold);
 
 		seq_printf(m, "RP CUR DOWN EI: %d (%dus)\n",
 			   rpdownei, GT_PM_INTERVAL_TO_US(dev_priv, rpdownei));
@@ -1168,8 +1239,7 @@ static int i915_frequency_info(struct seq_file *m, void *unused)
 			   rpcurdown, GT_PM_INTERVAL_TO_US(dev_priv, rpcurdown));
 		seq_printf(m, "RP PREV DOWN: %d (%dus)\n",
 			   rpprevdown, GT_PM_INTERVAL_TO_US(dev_priv, rpprevdown));
-		seq_printf(m, "Down threshold: %d%%\n",
-			   dev_priv->rps.down_threshold);
+		seq_printf(m, "Down threshold: %d%%\n", rps->down_threshold);
 
 		max_freq = (IS_GEN9_LP(dev_priv) ? rp_state_cap >> 0 :
 			    rp_state_cap >> 16) & 0xff;
@@ -1191,22 +1261,22 @@ static int i915_frequency_info(struct seq_file *m, void *unused)
 		seq_printf(m, "Max non-overclocked (RP0) frequency: %dMHz\n",
 			   intel_gpu_freq(dev_priv, max_freq));
 		seq_printf(m, "Max overclocked frequency: %dMHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.max_freq));
+			   intel_gpu_freq(dev_priv, rps->max_freq));
 
 		seq_printf(m, "Current freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.cur_freq));
+			   intel_gpu_freq(dev_priv, rps->cur_freq));
 		seq_printf(m, "Actual freq: %d MHz\n", cagf);
 		seq_printf(m, "Idle freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.idle_freq));
+			   intel_gpu_freq(dev_priv, rps->idle_freq));
 		seq_printf(m, "Min freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.min_freq));
+			   intel_gpu_freq(dev_priv, rps->min_freq));
 		seq_printf(m, "Boost freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.boost_freq));
+			   intel_gpu_freq(dev_priv, rps->boost_freq));
 		seq_printf(m, "Max freq: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.max_freq));
+			   intel_gpu_freq(dev_priv, rps->max_freq));
 		seq_printf(m,
 			   "efficient (RPe) frequency: %d MHz\n",
-			   intel_gpu_freq(dev_priv, dev_priv->rps.efficient_freq));
+			   intel_gpu_freq(dev_priv, rps->efficient_freq));
 	} else {
 		seq_puts(m, "no P-state info available\n");
 	}
@@ -1447,21 +1517,11 @@ static void print_rc6_res(struct seq_file *m,
 static int vlv_drpc_info(struct seq_file *m)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
-	u32 rpmodectl1, rcctl1, pw_status;
+	u32 rcctl1, pw_status;
 
 	pw_status = I915_READ(VLV_GTLC_PW_STATUS);
-	rpmodectl1 = I915_READ(GEN6_RP_CONTROL);
 	rcctl1 = I915_READ(GEN6_RC_CONTROL);
 
-	seq_printf(m, "Video Turbo Mode: %s\n",
-		   yesno(rpmodectl1 & GEN6_RP_MEDIA_TURBO));
-	seq_printf(m, "Turbo enabled: %s\n",
-		   yesno(rpmodectl1 & GEN6_RP_ENABLE));
-	seq_printf(m, "HW control enabled: %s\n",
-		   yesno(rpmodectl1 & GEN6_RP_ENABLE));
-	seq_printf(m, "SW control enabled: %s\n",
-		   yesno((rpmodectl1 & GEN6_RP_MEDIA_MODE_MASK) ==
-			  GEN6_RP_MEDIA_SW_MODE));
 	seq_printf(m, "RC6 Enabled: %s\n",
 		   yesno(rcctl1 & (GEN7_RC_CTL_TO_MODE |
 					GEN6_RC_CTL_EI_MODE(1))));
@@ -1479,7 +1539,7 @@ static int vlv_drpc_info(struct seq_file *m)
 static int gen6_drpc_info(struct seq_file *m)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
-	u32 rpmodectl1, gt_core_status, rcctl1, rc6vids = 0;
+	u32 gt_core_status, rcctl1, rc6vids = 0;
 	u32 gen9_powergate_enable = 0, gen9_powergate_status = 0;
 	unsigned forcewake_count;
 	int count = 0;
@@ -1498,24 +1558,16 @@ static int gen6_drpc_info(struct seq_file *m)
 	gt_core_status = I915_READ_FW(GEN6_GT_CORE_STATUS);
 	trace_i915_reg_rw(false, GEN6_GT_CORE_STATUS, gt_core_status, 4, true);
 
-	rpmodectl1 = I915_READ(GEN6_RP_CONTROL);
 	rcctl1 = I915_READ(GEN6_RC_CONTROL);
 	if (INTEL_GEN(dev_priv) >= 9) {
 		gen9_powergate_enable = I915_READ(GEN9_PG_ENABLE);
 		gen9_powergate_status = I915_READ(GEN9_PWRGT_DOMAIN_STATUS);
 	}
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	sandybridge_pcode_read(dev_priv, GEN6_PCODE_READ_RC6VIDS, &rc6vids);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
-	seq_printf(m, "Video Turbo Mode: %s\n",
-		   yesno(rpmodectl1 & GEN6_RP_MEDIA_TURBO));
-	seq_printf(m, "HW control enabled: %s\n",
-		   yesno(rpmodectl1 & GEN6_RP_ENABLE));
-	seq_printf(m, "SW control enabled: %s\n",
-		   yesno((rpmodectl1 & GEN6_RP_MEDIA_MODE_MASK) ==
-			  GEN6_RP_MEDIA_SW_MODE));
 	seq_printf(m, "RC1e Enabled: %s\n",
 		   yesno(rcctl1 & GEN6_RC_CTL_RC1e_ENABLE));
 	seq_printf(m, "RC6 Enabled: %s\n",
@@ -1778,6 +1830,7 @@ static int i915_emon_status(struct seq_file *m, void *unused)
 static int i915_ring_freq_table(struct seq_file *m, void *unused)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	int ret = 0;
 	int gpu_freq, ia_freq;
 	unsigned int max_gpu_freq, min_gpu_freq;
@@ -1789,19 +1842,17 @@ static int i915_ring_freq_table(struct seq_file *m, void *unused)
 
 	intel_runtime_pm_get(dev_priv);
 
-	ret = mutex_lock_interruptible(&dev_priv->rps.hw_lock);
+	ret = mutex_lock_interruptible(&dev_priv->pcu_lock);
 	if (ret)
 		goto out;
 
 	if (IS_GEN9_BC(dev_priv) || IS_CANNONLAKE(dev_priv)) {
 		/* Convert GT frequency to 50 HZ units */
-		min_gpu_freq =
-			dev_priv->rps.min_freq_softlimit / GEN9_FREQ_SCALER;
-		max_gpu_freq =
-			dev_priv->rps.max_freq_softlimit / GEN9_FREQ_SCALER;
+		min_gpu_freq = rps->min_freq_softlimit / GEN9_FREQ_SCALER;
+		max_gpu_freq = rps->max_freq_softlimit / GEN9_FREQ_SCALER;
 	} else {
-		min_gpu_freq = dev_priv->rps.min_freq_softlimit;
-		max_gpu_freq = dev_priv->rps.max_freq_softlimit;
+		min_gpu_freq = rps->min_freq_softlimit;
+		max_gpu_freq = rps->max_freq_softlimit;
 	}
 
 	seq_puts(m, "GPU freq (MHz)\tEffective CPU freq (MHz)\tEffective Ring freq (MHz)\n");
@@ -1820,7 +1871,7 @@ static int i915_ring_freq_table(struct seq_file *m, void *unused)
 			   ((ia_freq >> 8) & 0xff) * 100);
 	}
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 out:
 	intel_runtime_pm_put(dev_priv);
@@ -2254,25 +2305,26 @@ static int i915_rps_boost_info(struct seq_file *m, void *data)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
 	struct drm_device *dev = &dev_priv->drm;
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	struct drm_file *file;
 
-	seq_printf(m, "RPS enabled? %d\n", dev_priv->rps.enabled);
+	seq_printf(m, "RPS enabled? %d\n", rps->enabled);
 	seq_printf(m, "GPU busy? %s [%d requests]\n",
 		   yesno(dev_priv->gt.awake), dev_priv->gt.active_requests);
 	seq_printf(m, "CPU waiting? %d\n", count_irq_waiters(dev_priv));
 	seq_printf(m, "Boosts outstanding? %d\n",
-		   atomic_read(&dev_priv->rps.num_waiters));
+		   atomic_read(&rps->num_waiters));
 	seq_printf(m, "Frequency requested %d\n",
-		   intel_gpu_freq(dev_priv, dev_priv->rps.cur_freq));
+		   intel_gpu_freq(dev_priv, rps->cur_freq));
 	seq_printf(m, "  min hard:%d, soft:%d; max soft:%d, hard:%d\n",
-		   intel_gpu_freq(dev_priv, dev_priv->rps.min_freq),
-		   intel_gpu_freq(dev_priv, dev_priv->rps.min_freq_softlimit),
-		   intel_gpu_freq(dev_priv, dev_priv->rps.max_freq_softlimit),
-		   intel_gpu_freq(dev_priv, dev_priv->rps.max_freq));
+		   intel_gpu_freq(dev_priv, rps->min_freq),
+		   intel_gpu_freq(dev_priv, rps->min_freq_softlimit),
+		   intel_gpu_freq(dev_priv, rps->max_freq_softlimit),
+		   intel_gpu_freq(dev_priv, rps->max_freq));
 	seq_printf(m, "  idle:%d, efficient:%d, boost:%d\n",
-		   intel_gpu_freq(dev_priv, dev_priv->rps.idle_freq),
-		   intel_gpu_freq(dev_priv, dev_priv->rps.efficient_freq),
-		   intel_gpu_freq(dev_priv, dev_priv->rps.boost_freq));
+		   intel_gpu_freq(dev_priv, rps->idle_freq),
+		   intel_gpu_freq(dev_priv, rps->efficient_freq),
+		   intel_gpu_freq(dev_priv, rps->boost_freq));
 
 	mutex_lock(&dev->filelist_mutex);
 	list_for_each_entry_reverse(file, &dev->filelist, lhead) {
@@ -2284,15 +2336,15 @@ static int i915_rps_boost_info(struct seq_file *m, void *data)
 		seq_printf(m, "%s [%d]: %d boosts\n",
 			   task ? task->comm : "<unknown>",
 			   task ? task->pid : -1,
-			   atomic_read(&file_priv->rps.boosts));
+			   atomic_read(&file_priv->rps_client.boosts));
 		rcu_read_unlock();
 	}
 	seq_printf(m, "Kernel (anonymous) boosts: %d\n",
-		   atomic_read(&dev_priv->rps.boosts));
+		   atomic_read(&rps->boosts));
 	mutex_unlock(&dev->filelist_mutex);
 
 	if (INTEL_GEN(dev_priv) >= 6 &&
-	    dev_priv->rps.enabled &&
+	    rps->enabled &&
 	    dev_priv->gt.active_requests) {
 		u32 rpup, rpupei;
 		u32 rpdown, rpdownei;
@@ -2305,13 +2357,13 @@ static int i915_rps_boost_info(struct seq_file *m, void *data)
 		intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
 
 		seq_printf(m, "\nRPS Autotuning (current \"%s\" window):\n",
-			   rps_power_to_str(dev_priv->rps.power));
+			   rps_power_to_str(rps->power));
 		seq_printf(m, "  Avg. up: %d%% [above threshold? %d%%]\n",
 			   rpup && rpupei ? 100 * rpup / rpupei : 0,
-			   dev_priv->rps.up_threshold);
+			   rps->up_threshold);
 		seq_printf(m, "  Avg. down: %d%% [below threshold? %d%%]\n",
 			   rpdown && rpdownei ? 100 * rpdown / rpdownei : 0,
-			   dev_priv->rps.down_threshold);
+			   rps->down_threshold);
 	} else {
 		seq_puts(m, "\nRPS Autotuning inactive\n");
 	}
@@ -3238,9 +3290,9 @@ static int i915_display_info(struct seq_file *m, void *unused)
 static int i915_engine_info(struct seq_file *m, void *unused)
 {
 	struct drm_i915_private *dev_priv = node_to_i915(m->private);
-	struct i915_gpu_error *error = &dev_priv->gpu_error;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
+	struct drm_printer p;
 
 	intel_runtime_pm_get(dev_priv);
 
@@ -3249,149 +3301,9 @@ static int i915_engine_info(struct seq_file *m, void *unused)
 	seq_printf(m, "Global active requests: %d\n",
 		   dev_priv->gt.active_requests);
 
-	for_each_engine(engine, dev_priv, id) {
-		struct intel_breadcrumbs *b = &engine->breadcrumbs;
-		struct drm_i915_gem_request *rq;
-		struct rb_node *rb;
-		u64 addr;
-
-		seq_printf(m, "%s\n", engine->name);
-		seq_printf(m, "\tcurrent seqno %x, last %x, hangcheck %x [%d ms], inflight %d\n",
-			   intel_engine_get_seqno(engine),
-			   intel_engine_last_submit(engine),
-			   engine->hangcheck.seqno,
-			   jiffies_to_msecs(jiffies - engine->hangcheck.action_timestamp),
-			   engine->timeline->inflight_seqnos);
-		seq_printf(m, "\tReset count: %d\n",
-			   i915_reset_engine_count(error, engine));
-
-		rcu_read_lock();
-
-		seq_printf(m, "\tRequests:\n");
-
-		rq = list_first_entry(&engine->timeline->requests,
-				      struct drm_i915_gem_request, link);
-		if (&rq->link != &engine->timeline->requests)
-			print_request(m, rq, "\t\tfirst  ");
-
-		rq = list_last_entry(&engine->timeline->requests,
-				     struct drm_i915_gem_request, link);
-		if (&rq->link != &engine->timeline->requests)
-			print_request(m, rq, "\t\tlast   ");
-
-		rq = i915_gem_find_active_request(engine);
-		if (rq) {
-			print_request(m, rq, "\t\tactive ");
-			seq_printf(m,
-				   "\t\t[head %04x, postfix %04x, tail %04x, batch 0x%08x_%08x]\n",
-				   rq->head, rq->postfix, rq->tail,
-				   rq->batch ? upper_32_bits(rq->batch->node.start) : ~0u,
-				   rq->batch ? lower_32_bits(rq->batch->node.start) : ~0u);
-		}
-
-		seq_printf(m, "\tRING_START: 0x%08x [0x%08x]\n",
-			   I915_READ(RING_START(engine->mmio_base)),
-			   rq ? i915_ggtt_offset(rq->ring->vma) : 0);
-		seq_printf(m, "\tRING_HEAD:  0x%08x [0x%08x]\n",
-			   I915_READ(RING_HEAD(engine->mmio_base)) & HEAD_ADDR,
-			   rq ? rq->ring->head : 0);
-		seq_printf(m, "\tRING_TAIL:  0x%08x [0x%08x]\n",
-			   I915_READ(RING_TAIL(engine->mmio_base)) & TAIL_ADDR,
-			   rq ? rq->ring->tail : 0);
-		seq_printf(m, "\tRING_CTL:   0x%08x [%s]\n",
-			   I915_READ(RING_CTL(engine->mmio_base)),
-			   I915_READ(RING_CTL(engine->mmio_base)) & (RING_WAIT | RING_WAIT_SEMAPHORE) ? "waiting" : "");
-
-		rcu_read_unlock();
-
-		addr = intel_engine_get_active_head(engine);
-		seq_printf(m, "\tACTHD:  0x%08x_%08x\n",
-			   upper_32_bits(addr), lower_32_bits(addr));
-		addr = intel_engine_get_last_batch_head(engine);
-		seq_printf(m, "\tBBADDR: 0x%08x_%08x\n",
-			   upper_32_bits(addr), lower_32_bits(addr));
-
-		if (i915_modparams.enable_execlists) {
-			const u32 *hws = &engine->status_page.page_addr[I915_HWS_CSB_BUF0_INDEX];
-			struct intel_engine_execlists * const execlists = &engine->execlists;
-			u32 ptr, read, write;
-			unsigned int idx;
-
-			seq_printf(m, "\tExeclist status: 0x%08x %08x\n",
-				   I915_READ(RING_EXECLIST_STATUS_LO(engine)),
-				   I915_READ(RING_EXECLIST_STATUS_HI(engine)));
-
-			ptr = I915_READ(RING_CONTEXT_STATUS_PTR(engine));
-			read = GEN8_CSB_READ_PTR(ptr);
-			write = GEN8_CSB_WRITE_PTR(ptr);
-			seq_printf(m, "\tExeclist CSB read %d [%d cached], write %d [%d from hws], interrupt posted? %s\n",
-				   read, execlists->csb_head,
-				   write,
-				   intel_read_status_page(engine, intel_hws_csb_write_index(engine->i915)),
-				   yesno(test_bit(ENGINE_IRQ_EXECLIST,
-						  &engine->irq_posted)));
-			if (read >= GEN8_CSB_ENTRIES)
-				read = 0;
-			if (write >= GEN8_CSB_ENTRIES)
-				write = 0;
-			if (read > write)
-				write += GEN8_CSB_ENTRIES;
-			while (read < write) {
-				idx = ++read % GEN8_CSB_ENTRIES;
-				seq_printf(m, "\tExeclist CSB[%d]: 0x%08x [0x%08x in hwsp], context: %d [%d in hwsp]\n",
-					   idx,
-					   I915_READ(RING_CONTEXT_STATUS_BUF_LO(engine, idx)),
-					   hws[idx * 2],
-					   I915_READ(RING_CONTEXT_STATUS_BUF_HI(engine, idx)),
-					   hws[idx * 2 + 1]);
-			}
-
-			rcu_read_lock();
-			for (idx = 0; idx < execlists_num_ports(execlists); idx++) {
-				unsigned int count;
-
-				rq = port_unpack(&execlists->port[idx], &count);
-				if (rq) {
-					seq_printf(m, "\t\tELSP[%d] count=%d, ",
-						   idx, count);
-					print_request(m, rq, "rq: ");
-				} else {
-					seq_printf(m, "\t\tELSP[%d] idle\n",
-						   idx);
-				}
-			}
-			rcu_read_unlock();
-
-			spin_lock_irq(&engine->timeline->lock);
-			for (rb = execlists->first; rb; rb = rb_next(rb)) {
-				struct i915_priolist *p =
-					rb_entry(rb, typeof(*p), node);
-
-				list_for_each_entry(rq, &p->requests,
-						    priotree.link)
-					print_request(m, rq, "\t\tQ ");
-			}
-			spin_unlock_irq(&engine->timeline->lock);
-		} else if (INTEL_GEN(dev_priv) > 6) {
-			seq_printf(m, "\tPP_DIR_BASE: 0x%08x\n",
-				   I915_READ(RING_PP_DIR_BASE(engine)));
-			seq_printf(m, "\tPP_DIR_BASE_READ: 0x%08x\n",
-				   I915_READ(RING_PP_DIR_BASE_READ(engine)));
-			seq_printf(m, "\tPP_DIR_DCLV: 0x%08x\n",
-				   I915_READ(RING_PP_DIR_DCLV(engine)));
-		}
-
-		spin_lock_irq(&b->rb_lock);
-		for (rb = rb_first(&b->waiters); rb; rb = rb_next(rb)) {
-			struct intel_wait *w = rb_entry(rb, typeof(*w), node);
-
-			seq_printf(m, "\t%s [%d] waiting for %x\n",
-				   w->tsk->comm, w->tsk->pid, w->seqno);
-		}
-		spin_unlock_irq(&b->rb_lock);
-
-		seq_puts(m, "\n");
-	}
+	p = drm_seq_file_printer(m);
+	for_each_engine(engine, dev_priv, id)
+		intel_engine_dump(engine, &p);
 
 	intel_runtime_pm_put(dev_priv);
 
@@ -4258,8 +4170,7 @@ fault_irq_set(struct drm_i915_private *i915,
 	mutex_unlock(&i915->drm.struct_mutex);
 
 	/* Flush idle worker to disarm irq */
-	while (flush_delayed_work(&i915->gt.idle_work))
-		;
+	drain_delayed_work(&i915->gt.idle_work);
 
 	return 0;
 
@@ -4392,7 +4303,7 @@ i915_max_freq_get(void *data, u64 *val)
 	if (INTEL_GEN(dev_priv) < 6)
 		return -ENODEV;
 
-	*val = intel_gpu_freq(dev_priv, dev_priv->rps.max_freq_softlimit);
+	*val = intel_gpu_freq(dev_priv, dev_priv->gt_pm.rps.max_freq_softlimit);
 	return 0;
 }
 
@@ -4400,6 +4311,7 @@ static int
 i915_max_freq_set(void *data, u64 val)
 {
 	struct drm_i915_private *dev_priv = data;
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 hw_max, hw_min;
 	int ret;
 
@@ -4408,7 +4320,7 @@ i915_max_freq_set(void *data, u64 val)
 
 	DRM_DEBUG_DRIVER("Manually setting max freq to %llu\n", val);
 
-	ret = mutex_lock_interruptible(&dev_priv->rps.hw_lock);
+	ret = mutex_lock_interruptible(&dev_priv->pcu_lock);
 	if (ret)
 		return ret;
 
@@ -4417,20 +4329,20 @@ i915_max_freq_set(void *data, u64 val)
 	 */
 	val = intel_freq_opcode(dev_priv, val);
 
-	hw_max = dev_priv->rps.max_freq;
-	hw_min = dev_priv->rps.min_freq;
+	hw_max = rps->max_freq;
+	hw_min = rps->min_freq;
 
-	if (val < hw_min || val > hw_max || val < dev_priv->rps.min_freq_softlimit) {
-		mutex_unlock(&dev_priv->rps.hw_lock);
+	if (val < hw_min || val > hw_max || val < rps->min_freq_softlimit) {
+		mutex_unlock(&dev_priv->pcu_lock);
 		return -EINVAL;
 	}
 
-	dev_priv->rps.max_freq_softlimit = val;
+	rps->max_freq_softlimit = val;
 
 	if (intel_set_rps(dev_priv, val))
 		DRM_DEBUG_DRIVER("failed to update RPS to new softlimit\n");
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	return 0;
 }
@@ -4447,7 +4359,7 @@ i915_min_freq_get(void *data, u64 *val)
 	if (INTEL_GEN(dev_priv) < 6)
 		return -ENODEV;
 
-	*val = intel_gpu_freq(dev_priv, dev_priv->rps.min_freq_softlimit);
+	*val = intel_gpu_freq(dev_priv, dev_priv->gt_pm.rps.min_freq_softlimit);
 	return 0;
 }
 
@@ -4455,6 +4367,7 @@ static int
 i915_min_freq_set(void *data, u64 val)
 {
 	struct drm_i915_private *dev_priv = data;
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 hw_max, hw_min;
 	int ret;
 
@@ -4463,7 +4376,7 @@ i915_min_freq_set(void *data, u64 val)
 
 	DRM_DEBUG_DRIVER("Manually setting min freq to %llu\n", val);
 
-	ret = mutex_lock_interruptible(&dev_priv->rps.hw_lock);
+	ret = mutex_lock_interruptible(&dev_priv->pcu_lock);
 	if (ret)
 		return ret;
 
@@ -4472,21 +4385,21 @@ i915_min_freq_set(void *data, u64 val)
 	 */
 	val = intel_freq_opcode(dev_priv, val);
 
-	hw_max = dev_priv->rps.max_freq;
-	hw_min = dev_priv->rps.min_freq;
+	hw_max = rps->max_freq;
+	hw_min = rps->min_freq;
 
 	if (val < hw_min ||
-	    val > hw_max || val > dev_priv->rps.max_freq_softlimit) {
-		mutex_unlock(&dev_priv->rps.hw_lock);
+	    val > hw_max || val > rps->max_freq_softlimit) {
+		mutex_unlock(&dev_priv->pcu_lock);
 		return -EINVAL;
 	}
 
-	dev_priv->rps.min_freq_softlimit = val;
+	rps->min_freq_softlimit = val;
 
 	if (intel_set_rps(dev_priv, val))
 		DRM_DEBUG_DRIVER("failed to update RPS to new softlimit\n");
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 59ac9199b35d..3db5851756f0 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -367,9 +367,18 @@ static int i915_getparam(struct drm_device *dev, void *data,
 		value = i915_gem_mmap_gtt_version();
 		break;
 	case I915_PARAM_HAS_SCHEDULER:
-		value = dev_priv->engine[RCS] &&
-			dev_priv->engine[RCS]->schedule;
+		value = 0;
+		if (dev_priv->engine[RCS] && dev_priv->engine[RCS]->schedule) {
+			value |= I915_SCHEDULER_CAP_ENABLED;
+			value |= I915_SCHEDULER_CAP_PRIORITY;
+
+			if (INTEL_INFO(dev_priv)->has_logical_ring_preemption &&
+			    i915_modparams.enable_execlists &&
+			    !i915_modparams.enable_guc_submission)
+				value |= I915_SCHEDULER_CAP_PREEMPTION;
+		}
 		break;
+
 	case I915_PARAM_MMAP_VERSION:
 		/* Remember to bump this if the version changes! */
 	case I915_PARAM_HAS_GEM:
@@ -606,9 +615,10 @@ static void i915_gem_fini(struct drm_i915_private *dev_priv)
 	intel_uc_fini_hw(dev_priv);
 	i915_gem_cleanup_engines(dev_priv);
 	i915_gem_contexts_fini(dev_priv);
-	i915_gem_cleanup_userptr(dev_priv);
 	mutex_unlock(&dev_priv->drm.struct_mutex);
 
+	i915_gem_cleanup_userptr(dev_priv);
+
 	i915_gem_drain_freed_objects(dev_priv);
 
 	WARN_ON(!list_empty(&dev_priv->contexts.list));
@@ -1007,6 +1017,8 @@ static int i915_driver_init_mmio(struct drm_i915_private *dev_priv)
 
 	intel_uncore_init(dev_priv);
 
+	intel_uc_init_mmio(dev_priv);
+
 	ret = intel_engines_init_mmio(dev_priv);
 	if (ret)
 		goto err_uncore;
@@ -1580,7 +1592,7 @@ static int i915_drm_suspend_late(struct drm_device *dev, bool hibernation)
 
 	intel_display_set_init_power(dev_priv, false);
 
-	fw_csr = !IS_GEN9_LP(dev_priv) &&
+	fw_csr = !IS_GEN9_LP(dev_priv) && !hibernation &&
 		suspend_to_idle(dev_priv) && dev_priv->csr.dmc_payload;
 	/*
 	 * In case of firmware assisted context save/restore don't manually
@@ -2070,11 +2082,14 @@ static int i915_pm_resume(struct device *kdev)
 /* freeze: before creating the hibernation_image */
 static int i915_pm_freeze(struct device *kdev)
 {
+	struct drm_device *dev = &kdev_to_i915(kdev)->drm;
 	int ret;
 
-	ret = i915_pm_suspend(kdev);
-	if (ret)
-		return ret;
+	if (dev->switch_power_state != DRM_SWITCH_POWER_OFF) {
+		ret = i915_drm_suspend(dev);
+		if (ret)
+			return ret;
+	}
 
 	ret = i915_gem_freeze(kdev_to_i915(kdev));
 	if (ret)
@@ -2085,11 +2100,14 @@ static int i915_pm_freeze(struct device *kdev)
 
 static int i915_pm_freeze_late(struct device *kdev)
 {
+	struct drm_device *dev = &kdev_to_i915(kdev)->drm;
 	int ret;
 
-	ret = i915_pm_suspend_late(kdev);
-	if (ret)
-		return ret;
+	if (dev->switch_power_state != DRM_SWITCH_POWER_OFF) {
+		ret = i915_drm_suspend_late(dev, true);
+		if (ret)
+			return ret;
+	}
 
 	ret = i915_gem_freeze_late(kdev_to_i915(kdev));
 	if (ret)
@@ -2485,7 +2503,7 @@ static int intel_runtime_suspend(struct device *kdev)
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	int ret;
 
-	if (WARN_ON_ONCE(!(dev_priv->rps.enabled && intel_enable_rc6())))
+	if (WARN_ON_ONCE(!(dev_priv->gt_pm.rc6.enabled && intel_rc6_enabled())))
 		return -ENODEV;
 
 	if (WARN_ON_ONCE(!HAS_RUNTIME_PM(dev_priv)))
@@ -2527,12 +2545,12 @@ static int intel_runtime_suspend(struct device *kdev)
 	intel_uncore_suspend(dev_priv);
 
 	enable_rpm_wakeref_asserts(dev_priv);
-	WARN_ON_ONCE(atomic_read(&dev_priv->pm.wakeref_count));
+	WARN_ON_ONCE(atomic_read(&dev_priv->runtime_pm.wakeref_count));
 
 	if (intel_uncore_arm_unclaimed_mmio_detection(dev_priv))
 		DRM_ERROR("Unclaimed access detected prior to suspending\n");
 
-	dev_priv->pm.suspended = true;
+	dev_priv->runtime_pm.suspended = true;
 
 	/*
 	 * FIXME: We really should find a document that references the arguments
@@ -2578,11 +2596,11 @@ static int intel_runtime_resume(struct device *kdev)
 
 	DRM_DEBUG_KMS("Resuming device\n");
 
-	WARN_ON_ONCE(atomic_read(&dev_priv->pm.wakeref_count));
+	WARN_ON_ONCE(atomic_read(&dev_priv->runtime_pm.wakeref_count));
 	disable_rpm_wakeref_asserts(dev_priv);
 
 	intel_opregion_notify_adapter(dev_priv, PCI_D0);
-	dev_priv->pm.suspended = false;
+	dev_priv->runtime_pm.suspended = false;
 	if (intel_uncore_unclaimed_mmio(dev_priv))
 		DRM_DEBUG_DRIVER("Unclaimed access during suspend, bios?\n");
 
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 7ca11318ac69..c7b2ca6aff05 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -80,8 +80,8 @@
 
 #define DRIVER_NAME		"i915"
 #define DRIVER_DESC		"Intel Graphics"
-#define DRIVER_DATE		"20170929"
-#define DRIVER_TIMESTAMP	1506682238
+#define DRIVER_DATE		"20171012"
+#define DRIVER_TIMESTAMP	1507831511
 
 /* Use I915_STATE_WARN(x) and I915_STATE_WARN_ON() (rather than WARN() and
  * WARN_ON()) for hw state sanity checks to check for unexpected conditions
@@ -609,7 +609,7 @@ struct drm_i915_file_private {
 
 	struct intel_rps_client {
 		atomic_t boosts;
-	} rps;
+	} rps_client;
 
 	unsigned int bsd_engine;
 
@@ -783,6 +783,7 @@ struct intel_csr {
 	func(has_l3_dpf); \
 	func(has_llc); \
 	func(has_logical_ring_contexts); \
+	func(has_logical_ring_preemption); \
 	func(has_overlay); \
 	func(has_pipe_cxsr); \
 	func(has_pooled_eu); \
@@ -868,6 +869,8 @@ struct intel_device_info {
 	u8 num_sprites[I915_MAX_PIPES];
 	u8 num_scalers[I915_MAX_PIPES];
 
+	unsigned int page_sizes; /* page sizes supported by the HW */
+
 #define DEFINE_FLAG(name) u8 name:1
 	DEV_INFO_FOR_EACH_FLAG(DEFINE_FLAG);
 #undef DEFINE_FLAG
@@ -981,6 +984,7 @@ struct i915_gpu_state {
 			pid_t pid;
 			u32 handle;
 			u32 hw_id;
+			int priority;
 			int ban_score;
 			int active;
 			int guilty;
@@ -1003,6 +1007,7 @@ struct i915_gpu_state {
 			long jiffies;
 			pid_t pid;
 			u32 context;
+			int priority;
 			int ban_score;
 			u32 seqno;
 			u32 head;
@@ -1312,7 +1317,7 @@ struct intel_rps_ei {
 	u32 media_c0;
 };
 
-struct intel_gen6_power_mgmt {
+struct intel_rps {
 	/*
 	 * work, interrupts_enabled and pm_iir are protected by
 	 * dev_priv->irq_lock
@@ -1353,20 +1358,26 @@ struct intel_gen6_power_mgmt {
 	enum { LOW_POWER, BETWEEN, HIGH_POWER } power;
 
 	bool enabled;
-	struct delayed_work autoenable_work;
 	atomic_t num_waiters;
 	atomic_t boosts;
 
 	/* manual wa residency calculations */
 	struct intel_rps_ei ei;
+};
 
-	/*
-	 * Protects RPS/RC6 register access and PCU communication.
-	 * Must be taken after struct_mutex if nested. Note that
-	 * this lock may be held for long periods of time when
-	 * talking to hw - so only take it when talking to hw!
-	 */
-	struct mutex hw_lock;
+struct intel_rc6 {
+	bool enabled;
+};
+
+struct intel_llc_pstate {
+	bool enabled;
+};
+
+struct intel_gen6_power_mgmt {
+	struct intel_rps rps;
+	struct intel_rc6 rc6;
+	struct intel_llc_pstate llc_pstate;
+	struct delayed_work autoenable_work;
 };
 
 /* defined intel_pm.c */
@@ -1508,6 +1519,11 @@ struct i915_gem_mm {
 	/** Usable portion of the GTT for GEM */
 	dma_addr_t stolen_base; /* limited to low memory (32-bit) */
 
+	/**
+	 * tmpfs instance used for shmem backed objects
+	 */
+	struct vfsmount *gemfs;
+
 	/** PPGTT used for aliasing the PPGTT with the GTT */
 	struct i915_hw_ppgtt *aliasing_ppgtt;
 
@@ -2251,8 +2267,11 @@ struct drm_i915_private {
 	wait_queue_head_t gmbus_wait_queue;
 
 	struct pci_dev *bridge_dev;
-	struct i915_gem_context *kernel_context;
 	struct intel_engine_cs *engine[I915_NUM_ENGINES];
+	/* Context used internally to idle the GPU and setup initial state */
+	struct i915_gem_context *kernel_context;
+	/* Context only to be used for injecting preemption commands */
+	struct i915_gem_context *preempt_context;
 	struct i915_vma *semaphore;
 
 	struct drm_dma_handle *status_page_dmah;
@@ -2408,8 +2427,16 @@ struct drm_i915_private {
 	/* Cannot be determined by PCIID. You must always read a register. */
 	u32 edram_cap;
 
-	/* gen6+ rps state */
-	struct intel_gen6_power_mgmt rps;
+	/*
+	 * Protects RPS/RC6 register access and PCU communication.
+	 * Must be taken after struct_mutex if nested. Note that
+	 * this lock may be held for long periods of time when
+	 * talking to hw - so only take it when talking to hw!
+	 */
+	struct mutex pcu_lock;
+
+	/* gen6+ GT PM state */
+	struct intel_gen6_power_mgmt gt_pm;
 
 	/* ilk-only ips/rps state. Everything in here is protected by the global
 	 * mchdev_lock in intel_pm.c */
@@ -2520,7 +2547,7 @@ struct drm_i915_private {
 		bool distrust_bios_wm;
 	} wm;
 
-	struct i915_runtime_pm pm;
+	struct i915_runtime_pm runtime_pm;
 
 	struct {
 		bool initialized;
@@ -2859,6 +2886,21 @@ static inline struct scatterlist *__sg_next(struct scatterlist *sg)
 	     (((__iter).curr += PAGE_SIZE) >= (__iter).max) ?		\
 	     (__iter) = __sgt_iter(__sg_next((__iter).sgp), false), 0 : 0)
 
+static inline unsigned int i915_sg_page_sizes(struct scatterlist *sg)
+{
+	unsigned int page_sizes;
+
+	page_sizes = 0;
+	while (sg) {
+		GEM_BUG_ON(sg->offset);
+		GEM_BUG_ON(!IS_ALIGNED(sg->length, PAGE_SIZE));
+		page_sizes |= sg->length;
+		sg = __sg_next(sg);
+	}
+
+	return page_sizes;
+}
+
 static inline unsigned int i915_sg_segment_size(void)
 {
 	unsigned int size = swiotlb_max_segment();
@@ -3088,6 +3130,10 @@ intel_info(const struct drm_i915_private *dev_priv)
 #define USES_PPGTT(dev_priv)		(i915_modparams.enable_ppgtt)
 #define USES_FULL_PPGTT(dev_priv)	(i915_modparams.enable_ppgtt >= 2)
 #define USES_FULL_48BIT_PPGTT(dev_priv)	(i915_modparams.enable_ppgtt == 3)
+#define HAS_PAGE_SIZES(dev_priv, sizes) ({ \
+	GEM_BUG_ON((sizes) == 0); \
+	((sizes) & ~(dev_priv)->info.page_sizes) == 0; \
+})
 
 #define HAS_OVERLAY(dev_priv)		 ((dev_priv)->info.has_overlay)
 #define OVERLAY_NEEDS_PHYSICAL(dev_priv) \
@@ -3504,7 +3550,8 @@ i915_gem_object_get_dma_address(struct drm_i915_gem_object *obj,
 				unsigned long n);
 
 void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj,
-				 struct sg_table *pages);
+				 struct sg_table *pages,
+				 unsigned int sg_page_sizes);
 int __i915_gem_object_get_pages(struct drm_i915_gem_object *obj);
 
 static inline int __must_check
@@ -3726,8 +3773,6 @@ i915_vm_to_ppgtt(struct i915_address_space *vm)
 }
 
 /* i915_gem_fence_reg.c */
-int __must_check i915_vma_get_fence(struct i915_vma *vma);
-int __must_check i915_vma_put_fence(struct i915_vma *vma);
 struct drm_i915_fence_reg *
 i915_reserve_fence(struct drm_i915_private *dev_priv);
 void i915_unreserve_fence(struct drm_i915_fence_reg *fence);
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 73eeb6b1f1cd..20fcac37c85a 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -35,6 +35,7 @@
 #include "intel_drv.h"
 #include "intel_frontbuffer.h"
 #include "intel_mocs.h"
+#include "i915_gemfs.h"
 #include <linux/dma-fence-array.h>
 #include <linux/kthread.h>
 #include <linux/reservation.h>
@@ -161,8 +162,7 @@ i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
 	return 0;
 }
 
-static struct sg_table *
-i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
+static int i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
 {
 	struct address_space *mapping = obj->base.filp->f_mapping;
 	drm_dma_handle_t *phys;
@@ -170,9 +170,10 @@ i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
 	struct scatterlist *sg;
 	char *vaddr;
 	int i;
+	int err;
 
 	if (WARN_ON(i915_gem_object_needs_bit17_swizzle(obj)))
-		return ERR_PTR(-EINVAL);
+		return -EINVAL;
 
 	/* Always aligning to the object size, allows a single allocation
 	 * to handle all possible callers, and given typical object sizes,
@@ -182,7 +183,7 @@ i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
 			     roundup_pow_of_two(obj->base.size),
 			     roundup_pow_of_two(obj->base.size));
 	if (!phys)
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 
 	vaddr = phys->vaddr;
 	for (i = 0; i < obj->base.size / PAGE_SIZE; i++) {
@@ -191,7 +192,7 @@ i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
 
 		page = shmem_read_mapping_page(mapping, i);
 		if (IS_ERR(page)) {
-			st = ERR_CAST(page);
+			err = PTR_ERR(page);
 			goto err_phys;
 		}
 
@@ -208,13 +209,13 @@ i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
 
 	st = kmalloc(sizeof(*st), GFP_KERNEL);
 	if (!st) {
-		st = ERR_PTR(-ENOMEM);
+		err = -ENOMEM;
 		goto err_phys;
 	}
 
 	if (sg_alloc_table(st, 1, GFP_KERNEL)) {
 		kfree(st);
-		st = ERR_PTR(-ENOMEM);
+		err = -ENOMEM;
 		goto err_phys;
 	}
 
@@ -226,11 +227,15 @@ i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
 	sg_dma_len(sg) = obj->base.size;
 
 	obj->phys_handle = phys;
-	return st;
+
+	__i915_gem_object_set_pages(obj, st, sg->length);
+
+	return 0;
 
 err_phys:
 	drm_pci_free(obj->base.dev, phys);
-	return st;
+
+	return err;
 }
 
 static void __start_cpu_write(struct drm_i915_gem_object *obj)
@@ -353,7 +358,7 @@ static long
 i915_gem_object_wait_fence(struct dma_fence *fence,
 			   unsigned int flags,
 			   long timeout,
-			   struct intel_rps_client *rps)
+			   struct intel_rps_client *rps_client)
 {
 	struct drm_i915_gem_request *rq;
 
@@ -386,11 +391,11 @@ i915_gem_object_wait_fence(struct dma_fence *fence,
 	 * forcing the clocks too high for the whole system, we only allow
 	 * each client to waitboost once in a busy period.
 	 */
-	if (rps) {
+	if (rps_client) {
 		if (INTEL_GEN(rq->i915) >= 6)
-			gen6_rps_boost(rq, rps);
+			gen6_rps_boost(rq, rps_client);
 		else
-			rps = NULL;
+			rps_client = NULL;
 	}
 
 	timeout = i915_wait_request(rq, flags, timeout);
@@ -406,7 +411,7 @@ static long
 i915_gem_object_wait_reservation(struct reservation_object *resv,
 				 unsigned int flags,
 				 long timeout,
-				 struct intel_rps_client *rps)
+				 struct intel_rps_client *rps_client)
 {
 	unsigned int seq = __read_seqcount_begin(&resv->seq);
 	struct dma_fence *excl;
@@ -425,7 +430,7 @@ i915_gem_object_wait_reservation(struct reservation_object *resv,
 		for (i = 0; i < count; i++) {
 			timeout = i915_gem_object_wait_fence(shared[i],
 							     flags, timeout,
-							     rps);
+							     rps_client);
 			if (timeout < 0)
 				break;
 
@@ -442,7 +447,8 @@ i915_gem_object_wait_reservation(struct reservation_object *resv,
 	}
 
 	if (excl && timeout >= 0) {
-		timeout = i915_gem_object_wait_fence(excl, flags, timeout, rps);
+		timeout = i915_gem_object_wait_fence(excl, flags, timeout,
+						     rps_client);
 		prune_fences = timeout >= 0;
 	}
 
@@ -538,7 +544,7 @@ int
 i915_gem_object_wait(struct drm_i915_gem_object *obj,
 		     unsigned int flags,
 		     long timeout,
-		     struct intel_rps_client *rps)
+		     struct intel_rps_client *rps_client)
 {
 	might_sleep();
 #if IS_ENABLED(CONFIG_LOCKDEP)
@@ -550,7 +556,7 @@ i915_gem_object_wait(struct drm_i915_gem_object *obj,
 
 	timeout = i915_gem_object_wait_reservation(obj->resv,
 						   flags, timeout,
-						   rps);
+						   rps_client);
 	return timeout < 0 ? timeout : 0;
 }
 
@@ -558,7 +564,7 @@ static struct intel_rps_client *to_rps_client(struct drm_file *file)
 {
 	struct drm_i915_file_private *fpriv = file->driver_priv;
 
-	return &fpriv->rps;
+	return &fpriv->rps_client;
 }
 
 static int
@@ -1050,7 +1056,9 @@ i915_gem_gtt_pread(struct drm_i915_gem_object *obj,
 
 	intel_runtime_pm_get(i915);
 	vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0,
-				       PIN_MAPPABLE | PIN_NONBLOCK);
+				       PIN_MAPPABLE |
+				       PIN_NONFAULT |
+				       PIN_NONBLOCK);
 	if (!IS_ERR(vma)) {
 		node.start = i915_ggtt_offset(vma);
 		node.allocated = false;
@@ -1234,7 +1242,9 @@ i915_gem_gtt_pwrite_fast(struct drm_i915_gem_object *obj,
 
 	intel_runtime_pm_get(i915);
 	vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0,
-				       PIN_MAPPABLE | PIN_NONBLOCK);
+				       PIN_MAPPABLE |
+				       PIN_NONFAULT |
+				       PIN_NONBLOCK);
 	if (!IS_ERR(vma)) {
 		node.start = i915_ggtt_offset(vma);
 		node.allocated = false;
@@ -1905,22 +1915,27 @@ int i915_gem_fault(struct vm_fault *vmf)
 	if (ret)
 		goto err_unpin;
 
-	ret = i915_vma_get_fence(vma);
+	ret = i915_vma_pin_fence(vma);
 	if (ret)
 		goto err_unpin;
 
-	/* Mark as being mmapped into userspace for later revocation */
-	assert_rpm_wakelock_held(dev_priv);
-	if (list_empty(&obj->userfault_link))
-		list_add(&obj->userfault_link, &dev_priv->mm.userfault_list);
-
 	/* Finally, remap it using the new GTT offset */
 	ret = remap_io_mapping(area,
 			       area->vm_start + (vma->ggtt_view.partial.offset << PAGE_SHIFT),
 			       (ggtt->mappable_base + vma->node.start) >> PAGE_SHIFT,
 			       min_t(u64, vma->size, area->vm_end - area->vm_start),
 			       &ggtt->mappable);
+	if (ret)
+		goto err_fence;
 
+	/* Mark as being mmapped into userspace for later revocation */
+	assert_rpm_wakelock_held(dev_priv);
+	if (!i915_vma_set_userfault(vma) && !obj->userfault_count++)
+		list_add(&obj->userfault_link, &dev_priv->mm.userfault_list);
+	GEM_BUG_ON(!obj->userfault_count);
+
+err_fence:
+	i915_vma_unpin_fence(vma);
 err_unpin:
 	__i915_vma_unpin(vma);
 err_unlock:
@@ -1972,6 +1987,25 @@ err:
 	return ret;
 }
 
+static void __i915_gem_object_release_mmap(struct drm_i915_gem_object *obj)
+{
+	struct i915_vma *vma;
+
+	GEM_BUG_ON(!obj->userfault_count);
+
+	obj->userfault_count = 0;
+	list_del(&obj->userfault_link);
+	drm_vma_node_unmap(&obj->base.vma_node,
+			   obj->base.dev->anon_inode->i_mapping);
+
+	list_for_each_entry(vma, &obj->vma_list, obj_link) {
+		if (!i915_vma_is_ggtt(vma))
+			break;
+
+		i915_vma_unset_userfault(vma);
+	}
+}
+
 /**
  * i915_gem_release_mmap - remove physical page mappings
  * @obj: obj in question
@@ -2002,12 +2036,10 @@ i915_gem_release_mmap(struct drm_i915_gem_object *obj)
 	lockdep_assert_held(&i915->drm.struct_mutex);
 	intel_runtime_pm_get(i915);
 
-	if (list_empty(&obj->userfault_link))
+	if (!obj->userfault_count)
 		goto out;
 
-	list_del_init(&obj->userfault_link);
-	drm_vma_node_unmap(&obj->base.vma_node,
-			   obj->base.dev->anon_inode->i_mapping);
+	__i915_gem_object_release_mmap(obj);
 
 	/* Ensure that the CPU's PTE are revoked and there are not outstanding
 	 * memory transactions from userspace before we return. The TLB
@@ -2035,11 +2067,8 @@ void i915_gem_runtime_suspend(struct drm_i915_private *dev_priv)
 	 */
 
 	list_for_each_entry_safe(obj, on,
-				 &dev_priv->mm.userfault_list, userfault_link) {
-		list_del_init(&obj->userfault_link);
-		drm_vma_node_unmap(&obj->base.vma_node,
-				   obj->base.dev->anon_inode->i_mapping);
-	}
+				 &dev_priv->mm.userfault_list, userfault_link)
+		__i915_gem_object_release_mmap(obj);
 
 	/* The fence will be lost when the device powers down. If any were
 	 * in use by hardware (i.e. they are pinned), we should not be powering
@@ -2062,7 +2091,7 @@ void i915_gem_runtime_suspend(struct drm_i915_private *dev_priv)
 		if (!reg->vma)
 			continue;
 
-		GEM_BUG_ON(!list_empty(&reg->vma->obj->userfault_link));
+		GEM_BUG_ON(i915_vma_has_userfault(reg->vma));
 		reg->dirty = true;
 	}
 }
@@ -2261,6 +2290,8 @@ void __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
 	if (!IS_ERR(pages))
 		obj->ops->put_pages(obj, pages);
 
+	obj->mm.page_sizes.phys = obj->mm.page_sizes.sg = 0;
+
 unlock:
 	mutex_unlock(&obj->mm.lock);
 }
@@ -2291,8 +2322,7 @@ static bool i915_sg_trim(struct sg_table *orig_st)
 	return true;
 }
 
-static struct sg_table *
-i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
+static int i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
 {
 	struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
 	const unsigned long page_count = obj->base.size / PAGE_SIZE;
@@ -2304,6 +2334,7 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
 	struct page *page;
 	unsigned long last_pfn = 0;	/* suppress gcc warning */
 	unsigned int max_segment = i915_sg_segment_size();
+	unsigned int sg_page_sizes;
 	gfp_t noreclaim;
 	int ret;
 
@@ -2316,12 +2347,12 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
 
 	st = kmalloc(sizeof(*st), GFP_KERNEL);
 	if (st == NULL)
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 
 rebuild_st:
 	if (sg_alloc_table(st, page_count, GFP_KERNEL)) {
 		kfree(st);
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 	}
 
 	/* Get the list of pages out of our struct file.  They'll be pinned
@@ -2335,6 +2366,7 @@ rebuild_st:
 
 	sg = st->sgl;
 	st->nents = 0;
+	sg_page_sizes = 0;
 	for (i = 0; i < page_count; i++) {
 		const unsigned int shrink[] = {
 			I915_SHRINK_BOUND | I915_SHRINK_UNBOUND | I915_SHRINK_PURGEABLE,
@@ -2387,8 +2419,10 @@ rebuild_st:
 		if (!i ||
 		    sg->length >= max_segment ||
 		    page_to_pfn(page) != last_pfn + 1) {
-			if (i)
+			if (i) {
+				sg_page_sizes |= sg->length;
 				sg = sg_next(sg);
+			}
 			st->nents++;
 			sg_set_page(sg, page, PAGE_SIZE, 0);
 		} else {
@@ -2399,8 +2433,10 @@ rebuild_st:
 		/* Check that the i965g/gm workaround works. */
 		WARN_ON((gfp & __GFP_DMA32) && (last_pfn >= 0x00100000UL));
 	}
-	if (sg) /* loop terminated early; short sg table */
+	if (sg) { /* loop terminated early; short sg table */
+		sg_page_sizes |= sg->length;
 		sg_mark_end(sg);
+	}
 
 	/* Trim unused sg entries to avoid wasting memory. */
 	i915_sg_trim(st);
@@ -2429,7 +2465,9 @@ rebuild_st:
 	if (i915_gem_object_needs_bit17_swizzle(obj))
 		i915_gem_object_do_bit_17_swizzle(obj, st);
 
-	return st;
+	__i915_gem_object_set_pages(obj, st, sg_page_sizes);
+
+	return 0;
 
 err_sg:
 	sg_mark_end(sg);
@@ -2450,12 +2488,17 @@ err_pages:
 	if (ret == -ENOSPC)
 		ret = -ENOMEM;
 
-	return ERR_PTR(ret);
+	return ret;
 }
 
 void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj,
-				 struct sg_table *pages)
+				 struct sg_table *pages,
+				 unsigned int sg_page_sizes)
 {
+	struct drm_i915_private *i915 = to_i915(obj->base.dev);
+	unsigned long supported = INTEL_INFO(i915)->page_sizes;
+	int i;
+
 	lockdep_assert_held(&obj->mm.lock);
 
 	obj->mm.get_page.sg_pos = pages->sgl;
@@ -2469,23 +2512,40 @@ void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj,
 		__i915_gem_object_pin_pages(obj);
 		obj->mm.quirked = true;
 	}
+
+	GEM_BUG_ON(!sg_page_sizes);
+	obj->mm.page_sizes.phys = sg_page_sizes;
+
+	/*
+	 * Calculate the supported page-sizes which fit into the given
+	 * sg_page_sizes. This will give us the page-sizes which we may be able
+	 * to use opportunistically when later inserting into the GTT. For
+	 * example if phys=2G, then in theory we should be able to use 1G, 2M,
+	 * 64K or 4K pages, although in practice this will depend on a number of
+	 * other factors.
+	 */
+	obj->mm.page_sizes.sg = 0;
+	for_each_set_bit(i, &supported, ilog2(I915_GTT_MAX_PAGE_SIZE) + 1) {
+		if (obj->mm.page_sizes.phys & ~0u << i)
+			obj->mm.page_sizes.sg |= BIT(i);
+	}
+
+	GEM_BUG_ON(!HAS_PAGE_SIZES(i915, obj->mm.page_sizes.sg));
 }
 
 static int ____i915_gem_object_get_pages(struct drm_i915_gem_object *obj)
 {
-	struct sg_table *pages;
+	int err;
 
 	if (unlikely(obj->mm.madv != I915_MADV_WILLNEED)) {
 		DRM_DEBUG("Attempting to obtain a purgeable object\n");
 		return -EFAULT;
 	}
 
-	pages = obj->ops->get_pages(obj);
-	if (unlikely(IS_ERR(pages)))
-		return PTR_ERR(pages);
+	err = obj->ops->get_pages(obj);
+	GEM_BUG_ON(!err && IS_ERR_OR_NULL(obj->mm.pages));
 
-	__i915_gem_object_set_pages(obj, pages);
-	return 0;
+	return err;
 }
 
 /* Ensure that the associated pages are gathered from the backing storage
@@ -2796,7 +2856,17 @@ i915_gem_reset_prepare_engine(struct intel_engine_cs *engine)
 {
 	struct drm_i915_gem_request *request = NULL;
 
-	/* Prevent the signaler thread from updating the request
+	/*
+	 * During the reset sequence, we must prevent the engine from
+	 * entering RC6. As the context state is undefined until we restart
+	 * the engine, if it does enter RC6 during the reset, the state
+	 * written to the powercontext is undefined and so we may lose
+	 * GPU state upon resume, i.e. fail to restart after a reset.
+	 */
+	intel_uncore_forcewake_get(engine->i915, FORCEWAKE_ALL);
+
+	/*
+	 * Prevent the signaler thread from updating the request
 	 * state (by calling dma_fence_signal) as we are processing
 	 * the reset. The write from the GPU of the seqno is
 	 * asynchronous and the signaler thread may see a different
@@ -2807,7 +2877,8 @@ i915_gem_reset_prepare_engine(struct intel_engine_cs *engine)
 	 */
 	kthread_park(engine->breadcrumbs.signaler);
 
-	/* Prevent request submission to the hardware until we have
+	/*
+	 * Prevent request submission to the hardware until we have
 	 * completed the reset in i915_gem_reset_finish(). If a request
 	 * is completed by one engine, it may then queue a request
 	 * to a second via its engine->irq_tasklet *just* as we are
@@ -2997,6 +3068,8 @@ void i915_gem_reset_finish_engine(struct intel_engine_cs *engine)
 {
 	tasklet_enable(&engine->execlists.irq_tasklet);
 	kthread_unpark(engine->breadcrumbs.signaler);
+
+	intel_uncore_forcewake_put(engine->i915, FORCEWAKE_ALL);
 }
 
 void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
@@ -3016,49 +3089,76 @@ static void nop_submit_request(struct drm_i915_gem_request *request)
 {
 	GEM_BUG_ON(!i915_terminally_wedged(&request->i915->gpu_error));
 	dma_fence_set_error(&request->fence, -EIO);
+
 	i915_gem_request_submit(request);
-	intel_engine_init_global_seqno(request->engine, request->global_seqno);
 }
 
-static void engine_set_wedged(struct intel_engine_cs *engine)
+static void nop_complete_submit_request(struct drm_i915_gem_request *request)
 {
-	/* We need to be sure that no thread is running the old callback as
-	 * we install the nop handler (otherwise we would submit a request
-	 * to hardware that will never complete). In order to prevent this
-	 * race, we wait until the machine is idle before making the swap
-	 * (using stop_machine()).
-	 */
-	engine->submit_request = nop_submit_request;
+	unsigned long flags;
 
-	/* Mark all executing requests as skipped */
-	engine->cancel_requests(engine);
+	GEM_BUG_ON(!i915_terminally_wedged(&request->i915->gpu_error));
+	dma_fence_set_error(&request->fence, -EIO);
 
-	/* Mark all pending requests as complete so that any concurrent
-	 * (lockless) lookup doesn't try and wait upon the request as we
-	 * reset it.
-	 */
-	intel_engine_init_global_seqno(engine,
-				       intel_engine_last_submit(engine));
+	spin_lock_irqsave(&request->engine->timeline->lock, flags);
+	__i915_gem_request_submit(request);
+	intel_engine_init_global_seqno(request->engine, request->global_seqno);
+	spin_unlock_irqrestore(&request->engine->timeline->lock, flags);
 }
 
-static int __i915_gem_set_wedged_BKL(void *data)
+void i915_gem_set_wedged(struct drm_i915_private *i915)
 {
-	struct drm_i915_private *i915 = data;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 
+	/*
+	 * First, stop submission to hw, but do not yet complete requests by
+	 * rolling the global seqno forward (since this would complete requests
+	 * for which we haven't set the fence error to EIO yet).
+	 */
 	for_each_engine(engine, i915, id)
-		engine_set_wedged(engine);
+		engine->submit_request = nop_submit_request;
 
-	set_bit(I915_WEDGED, &i915->gpu_error.flags);
-	wake_up_all(&i915->gpu_error.reset_queue);
+	/*
+	 * Make sure no one is running the old callback before we proceed with
+	 * cancelling requests and resetting the completion tracking. Otherwise
+	 * we might submit a request to the hardware which never completes.
+	 */
+	synchronize_rcu();
 
-	return 0;
-}
+	for_each_engine(engine, i915, id) {
+		/* Mark all executing requests as skipped */
+		engine->cancel_requests(engine);
 
-void i915_gem_set_wedged(struct drm_i915_private *dev_priv)
-{
-	stop_machine(__i915_gem_set_wedged_BKL, dev_priv, NULL);
+		/*
+		 * Only once we've force-cancelled all in-flight requests can we
+		 * start to complete all requests.
+		 */
+		engine->submit_request = nop_complete_submit_request;
+	}
+
+	/*
+	 * Make sure no request can slip through without getting completed by
+	 * either this call here to intel_engine_init_global_seqno, or the one
+	 * in nop_complete_submit_request.
+	 */
+	synchronize_rcu();
+
+	for_each_engine(engine, i915, id) {
+		unsigned long flags;
+
+		/* Mark all pending requests as complete so that any concurrent
+		 * (lockless) lookup doesn't try and wait upon the request as we
+		 * reset it.
+		 */
+		spin_lock_irqsave(&engine->timeline->lock, flags);
+		intel_engine_init_global_seqno(engine,
+					       intel_engine_last_submit(engine));
+		spin_unlock_irqrestore(&engine->timeline->lock, flags);
+	}
+
+	set_bit(I915_WEDGED, &i915->gpu_error.flags);
+	wake_up_all(&i915->gpu_error.reset_queue);
 }
 
 bool i915_gem_unset_wedged(struct drm_i915_private *i915)
@@ -3959,42 +4059,47 @@ i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj,
 
 	lockdep_assert_held(&obj->base.dev->struct_mutex);
 
+	if (!view && flags & PIN_MAPPABLE) {
+		/* If the required space is larger than the available
+		 * aperture, we will not able to find a slot for the
+		 * object and unbinding the object now will be in
+		 * vain. Worse, doing so may cause us to ping-pong
+		 * the object in and out of the Global GTT and
+		 * waste a lot of cycles under the mutex.
+		 */
+		if (obj->base.size > dev_priv->ggtt.mappable_end)
+			return ERR_PTR(-E2BIG);
+
+		/* If NONBLOCK is set the caller is optimistically
+		 * trying to cache the full object within the mappable
+		 * aperture, and *must* have a fallback in place for
+		 * situations where we cannot bind the object. We
+		 * can be a little more lax here and use the fallback
+		 * more often to avoid costly migrations of ourselves
+		 * and other objects within the aperture.
+		 *
+		 * Half-the-aperture is used as a simple heuristic.
+		 * More interesting would to do search for a free
+		 * block prior to making the commitment to unbind.
+		 * That caters for the self-harm case, and with a
+		 * little more heuristics (e.g. NOFAULT, NOEVICT)
+		 * we could try to minimise harm to others.
+		 */
+		if (flags & PIN_NONBLOCK &&
+		    obj->base.size > dev_priv->ggtt.mappable_end / 2)
+			return ERR_PTR(-ENOSPC);
+	}
+
 	vma = i915_vma_instance(obj, vm, view);
 	if (unlikely(IS_ERR(vma)))
 		return vma;
 
 	if (i915_vma_misplaced(vma, size, alignment, flags)) {
-		if (flags & PIN_NONBLOCK &&
-		    (i915_vma_is_pinned(vma) || i915_vma_is_active(vma)))
-			return ERR_PTR(-ENOSPC);
+		if (flags & PIN_NONBLOCK) {
+			if (i915_vma_is_pinned(vma) || i915_vma_is_active(vma))
+				return ERR_PTR(-ENOSPC);
 
-		if (flags & PIN_MAPPABLE) {
-			/* If the required space is larger than the available
-			 * aperture, we will not able to find a slot for the
-			 * object and unbinding the object now will be in
-			 * vain. Worse, doing so may cause us to ping-pong
-			 * the object in and out of the Global GTT and
-			 * waste a lot of cycles under the mutex.
-			 */
-			if (vma->fence_size > dev_priv->ggtt.mappable_end)
-				return ERR_PTR(-E2BIG);
-
-			/* If NONBLOCK is set the caller is optimistically
-			 * trying to cache the full object within the mappable
-			 * aperture, and *must* have a fallback in place for
-			 * situations where we cannot bind the object. We
-			 * can be a little more lax here and use the fallback
-			 * more often to avoid costly migrations of ourselves
-			 * and other objects within the aperture.
-			 *
-			 * Half-the-aperture is used as a simple heuristic.
-			 * More interesting would to do search for a free
-			 * block prior to making the commitment to unbind.
-			 * That caters for the self-harm case, and with a
-			 * little more heuristics (e.g. NOFAULT, NOEVICT)
-			 * we could try to minimise harm to others.
-			 */
-			if (flags & PIN_NONBLOCK &&
+			if (flags & PIN_MAPPABLE &&
 			    vma->fence_size > dev_priv->ggtt.mappable_end / 2)
 				return ERR_PTR(-ENOSPC);
 		}
@@ -4221,7 +4326,6 @@ void i915_gem_object_init(struct drm_i915_gem_object *obj,
 	mutex_init(&obj->mm.lock);
 
 	INIT_LIST_HEAD(&obj->global_link);
-	INIT_LIST_HEAD(&obj->userfault_link);
 	INIT_LIST_HEAD(&obj->vma_list);
 	INIT_LIST_HEAD(&obj->lut_list);
 	INIT_LIST_HEAD(&obj->batch_pool_link);
@@ -4251,6 +4355,30 @@ static const struct drm_i915_gem_object_ops i915_gem_object_ops = {
 	.pwrite = i915_gem_object_pwrite_gtt,
 };
 
+static int i915_gem_object_create_shmem(struct drm_device *dev,
+					struct drm_gem_object *obj,
+					size_t size)
+{
+	struct drm_i915_private *i915 = to_i915(dev);
+	unsigned long flags = VM_NORESERVE;
+	struct file *filp;
+
+	drm_gem_private_object_init(dev, obj, size);
+
+	if (i915->mm.gemfs)
+		filp = shmem_file_setup_with_mnt(i915->mm.gemfs, "i915", size,
+						 flags);
+	else
+		filp = shmem_file_setup("i915", size, flags);
+
+	if (IS_ERR(filp))
+		return PTR_ERR(filp);
+
+	obj->filp = filp;
+
+	return 0;
+}
+
 struct drm_i915_gem_object *
 i915_gem_object_create(struct drm_i915_private *dev_priv, u64 size)
 {
@@ -4275,7 +4403,7 @@ i915_gem_object_create(struct drm_i915_private *dev_priv, u64 size)
 	if (obj == NULL)
 		return ERR_PTR(-ENOMEM);
 
-	ret = drm_gem_object_init(&dev_priv->drm, &obj->base, size);
+	ret = i915_gem_object_create_shmem(&dev_priv->drm, &obj->base, size);
 	if (ret)
 		goto fail;
 
@@ -4378,6 +4506,7 @@ static void __i915_gem_free_objects(struct drm_i915_private *i915,
 
 	llist_for_each_entry_safe(obj, on, freed, freed) {
 		GEM_BUG_ON(obj->bind_count);
+		GEM_BUG_ON(obj->userfault_count);
 		GEM_BUG_ON(atomic_read(&obj->frontbuffer_bits));
 		GEM_BUG_ON(!list_empty(&obj->lut_list));
 
@@ -4547,8 +4676,7 @@ int i915_gem_suspend(struct drm_i915_private *dev_priv)
 	/* As the idle_work is rearming if it detects a race, play safe and
 	 * repeat the flush until it is definitely idle.
 	 */
-	while (flush_delayed_work(&dev_priv->gt.idle_work))
-		;
+	drain_delayed_work(&dev_priv->gt.idle_work);
 
 	/* Assert that we sucessfully flushed all the work and
 	 * reset the GPU back to its idle, low power state.
@@ -4595,6 +4723,7 @@ void i915_gem_resume(struct drm_i915_private *dev_priv)
 
 	mutex_lock(&dev->struct_mutex);
 	i915_gem_restore_gtt_mappings(dev_priv);
+	i915_gem_restore_fences(dev_priv);
 
 	/* As we didn't flush the kernel context before suspend, we cannot
 	 * guarantee that the context image is complete. So let's just reset
@@ -4757,6 +4886,15 @@ int i915_gem_init(struct drm_i915_private *dev_priv)
 
 	mutex_lock(&dev_priv->drm.struct_mutex);
 
+	/*
+	 * We need to fallback to 4K pages since gvt gtt handling doesn't
+	 * support huge page entries - we will need to check either hypervisor
+	 * mm can support huge guest page or just do emulation in gvt.
+	 */
+	if (intel_vgpu_active(dev_priv))
+		mkwrite_device_info(dev_priv)->page_sizes =
+			I915_GTT_PAGE_SIZE_4K;
+
 	dev_priv->mm.unordered_timeline = dma_fence_context_alloc(1);
 
 	if (!i915_modparams.enable_execlists) {
@@ -4914,6 +5052,10 @@ i915_gem_load_init(struct drm_i915_private *dev_priv)
 
 	spin_lock_init(&dev_priv->fb_tracking.lock);
 
+	err = i915_gemfs_init(dev_priv);
+	if (err)
+		DRM_NOTE("Unable to create a private tmpfs mount, hugepage support will be disabled(%d).\n", err);
+
 	return 0;
 
 err_priorities:
@@ -4952,6 +5094,8 @@ void i915_gem_load_cleanup(struct drm_i915_private *dev_priv)
 
 	/* And ensure that our DESTROY_BY_RCU slabs are truly destroyed */
 	rcu_barrier();
+
+	i915_gemfs_fini(dev_priv);
 }
 
 int i915_gem_freeze(struct drm_i915_private *dev_priv)
@@ -5341,6 +5485,7 @@ err_unlock:
 #include "selftests/scatterlist.c"
 #include "selftests/mock_gem_device.c"
 #include "selftests/huge_gem_object.c"
+#include "selftests/huge_pages.c"
 #include "selftests/i915_gem_object.c"
 #include "selftests/i915_gem_coherency.c"
 #endif
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c
index 921ee369c74d..5bf96a258509 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -416,14 +416,43 @@ out:
 	return ctx;
 }
 
+static struct i915_gem_context *
+create_kernel_context(struct drm_i915_private *i915, int prio)
+{
+	struct i915_gem_context *ctx;
+
+	ctx = i915_gem_create_context(i915, NULL);
+	if (IS_ERR(ctx))
+		return ctx;
+
+	i915_gem_context_clear_bannable(ctx);
+	ctx->priority = prio;
+	ctx->ring_size = PAGE_SIZE;
+
+	GEM_BUG_ON(!i915_gem_context_is_kernel(ctx));
+
+	return ctx;
+}
+
+static void
+destroy_kernel_context(struct i915_gem_context **ctxp)
+{
+	struct i915_gem_context *ctx;
+
+	/* Keep the context ref so that we can free it immediately ourselves */
+	ctx = i915_gem_context_get(fetch_and_zero(ctxp));
+	GEM_BUG_ON(!i915_gem_context_is_kernel(ctx));
+
+	context_close(ctx);
+	i915_gem_context_free(ctx);
+}
+
 int i915_gem_contexts_init(struct drm_i915_private *dev_priv)
 {
 	struct i915_gem_context *ctx;
+	int err;
 
-	/* Init should only be called once per module load. Eventually the
-	 * restriction on the context_disabled check can be loosened. */
-	if (WARN_ON(dev_priv->kernel_context))
-		return 0;
+	GEM_BUG_ON(dev_priv->kernel_context);
 
 	INIT_LIST_HEAD(&dev_priv->contexts.list);
 	INIT_WORK(&dev_priv->contexts.free_work, contexts_free_worker);
@@ -441,28 +470,38 @@ int i915_gem_contexts_init(struct drm_i915_private *dev_priv)
 	BUILD_BUG_ON(MAX_CONTEXT_HW_ID > INT_MAX);
 	ida_init(&dev_priv->contexts.hw_ida);
 
-	ctx = i915_gem_create_context(dev_priv, NULL);
+	/* lowest priority; idle task */
+	ctx = create_kernel_context(dev_priv, I915_PRIORITY_MIN);
 	if (IS_ERR(ctx)) {
-		DRM_ERROR("Failed to create default global context (error %ld)\n",
-			  PTR_ERR(ctx));
-		return PTR_ERR(ctx);
+		DRM_ERROR("Failed to create default global context\n");
+		err = PTR_ERR(ctx);
+		goto err;
 	}
-
-	/* For easy recognisablity, we want the kernel context to be 0 and then
+	/*
+	 * For easy recognisablity, we want the kernel context to be 0 and then
 	 * all user contexts will have non-zero hw_id.
 	 */
 	GEM_BUG_ON(ctx->hw_id);
-
-	i915_gem_context_clear_bannable(ctx);
-	ctx->priority = I915_PRIORITY_MIN; /* lowest priority; idle task */
 	dev_priv->kernel_context = ctx;
 
-	GEM_BUG_ON(!i915_gem_context_is_kernel(ctx));
+	/* highest priority; preempting task */
+	ctx = create_kernel_context(dev_priv, INT_MAX);
+	if (IS_ERR(ctx)) {
+		DRM_ERROR("Failed to create default preempt context\n");
+		err = PTR_ERR(ctx);
+		goto err_kernel_context;
+	}
+	dev_priv->preempt_context = ctx;
 
 	DRM_DEBUG_DRIVER("%s context support initialized\n",
 			 dev_priv->engine[RCS]->context_size ? "logical" :
 			 "fake");
 	return 0;
+
+err_kernel_context:
+	destroy_kernel_context(&dev_priv->kernel_context);
+err:
+	return err;
 }
 
 void i915_gem_contexts_lost(struct drm_i915_private *dev_priv)
@@ -507,15 +546,10 @@ void i915_gem_contexts_lost(struct drm_i915_private *dev_priv)
 
 void i915_gem_contexts_fini(struct drm_i915_private *i915)
 {
-	struct i915_gem_context *ctx;
-
 	lockdep_assert_held(&i915->drm.struct_mutex);
 
-	/* Keep the context so that we can free it immediately ourselves */
-	ctx = i915_gem_context_get(fetch_and_zero(&i915->kernel_context));
-	GEM_BUG_ON(!i915_gem_context_is_kernel(ctx));
-	context_close(ctx);
-	i915_gem_context_free(ctx);
+	destroy_kernel_context(&i915->preempt_context);
+	destroy_kernel_context(&i915->kernel_context);
 
 	/* Must free all deferred contexts (via flush_workqueue) first */
 	ida_destroy(&i915->contexts.hw_ida);
@@ -1036,6 +1070,9 @@ int i915_gem_context_getparam_ioctl(struct drm_device *dev, void *data,
 	case I915_CONTEXT_PARAM_BANNABLE:
 		args->value = i915_gem_context_is_bannable(ctx);
 		break;
+	case I915_CONTEXT_PARAM_PRIORITY:
+		args->value = ctx->priority;
+		break;
 	default:
 		ret = -EINVAL;
 		break;
@@ -1091,6 +1128,26 @@ int i915_gem_context_setparam_ioctl(struct drm_device *dev, void *data,
 		else
 			i915_gem_context_clear_bannable(ctx);
 		break;
+
+	case I915_CONTEXT_PARAM_PRIORITY:
+		{
+			int priority = args->value;
+
+			if (args->size)
+				ret = -EINVAL;
+			else if (!to_i915(dev)->engine[RCS]->schedule)
+				ret = -ENODEV;
+			else if (priority > I915_CONTEXT_MAX_USER_PRIORITY ||
+				 priority < I915_CONTEXT_MIN_USER_PRIORITY)
+				ret = -EINVAL;
+			else if (priority > I915_CONTEXT_DEFAULT_PRIORITY &&
+				 !capable(CAP_SYS_NICE))
+				ret = -EPERM;
+			else
+				ctx->priority = priority;
+		}
+		break;
+
 	default:
 		ret = -EINVAL;
 		break;
diff --git a/drivers/gpu/drm/i915/i915_gem_dmabuf.c b/drivers/gpu/drm/i915/i915_gem_dmabuf.c
index 6176e589cf09..864439a214c8 100644
--- a/drivers/gpu/drm/i915/i915_gem_dmabuf.c
+++ b/drivers/gpu/drm/i915/i915_gem_dmabuf.c
@@ -256,11 +256,21 @@ struct dma_buf *i915_gem_prime_export(struct drm_device *dev,
 	return drm_gem_dmabuf_export(dev, &exp_info);
 }
 
-static struct sg_table *
-i915_gem_object_get_pages_dmabuf(struct drm_i915_gem_object *obj)
+static int i915_gem_object_get_pages_dmabuf(struct drm_i915_gem_object *obj)
 {
-	return dma_buf_map_attachment(obj->base.import_attach,
-				      DMA_BIDIRECTIONAL);
+	struct sg_table *pages;
+	unsigned int sg_page_sizes;
+
+	pages = dma_buf_map_attachment(obj->base.import_attach,
+				       DMA_BIDIRECTIONAL);
+	if (IS_ERR(pages))
+		return PTR_ERR(pages);
+
+	sg_page_sizes = i915_sg_page_sizes(pages->sgl);
+
+	__i915_gem_object_set_pages(obj, pages, sg_page_sizes);
+
+	return 0;
 }
 
 static void i915_gem_object_put_pages_dmabuf(struct drm_i915_gem_object *obj,
diff --git a/drivers/gpu/drm/i915/i915_gem_evict.c b/drivers/gpu/drm/i915/i915_gem_evict.c
index 4df039ef2ce3..a5a5b7e6daae 100644
--- a/drivers/gpu/drm/i915/i915_gem_evict.c
+++ b/drivers/gpu/drm/i915/i915_gem_evict.c
@@ -82,7 +82,7 @@ mark_free(struct drm_mm_scan *scan,
 	if (i915_vma_is_pinned(vma))
 		return false;
 
-	if (flags & PIN_NONFAULT && !list_empty(&vma->obj->userfault_link))
+	if (flags & PIN_NONFAULT && i915_vma_has_userfault(vma))
 		return false;
 
 	list_add(&vma->evict_link, unwind);
@@ -315,6 +315,11 @@ int i915_gem_evict_for_node(struct i915_address_space *vm,
 			break;
 		}
 
+		if (flags & PIN_NONFAULT && i915_vma_has_userfault(vma)) {
+			ret = -ENOSPC;
+			break;
+		}
+
 		/* Overlap of objects in the same batch? */
 		if (i915_vma_is_pinned(vma)) {
 			ret = -ENOSPC;
diff --git a/drivers/gpu/drm/i915/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
index d733c4d5a500..3d7190764f10 100644
--- a/drivers/gpu/drm/i915/i915_gem_execbuffer.c
+++ b/drivers/gpu/drm/i915/i915_gem_execbuffer.c
@@ -367,12 +367,12 @@ eb_pin_vma(struct i915_execbuffer *eb,
 		return false;
 
 	if (unlikely(exec_flags & EXEC_OBJECT_NEEDS_FENCE)) {
-		if (unlikely(i915_vma_get_fence(vma))) {
+		if (unlikely(i915_vma_pin_fence(vma))) {
 			i915_vma_unpin(vma);
 			return false;
 		}
 
-		if (i915_vma_pin_fence(vma))
+		if (vma->fence)
 			exec_flags |= __EXEC_OBJECT_HAS_FENCE;
 	}
 
@@ -385,7 +385,7 @@ static inline void __eb_unreserve_vma(struct i915_vma *vma, unsigned int flags)
 	GEM_BUG_ON(!(flags & __EXEC_OBJECT_HAS_PIN));
 
 	if (unlikely(flags & __EXEC_OBJECT_HAS_FENCE))
-		i915_vma_unpin_fence(vma);
+		__i915_vma_unpin_fence(vma);
 
 	__i915_vma_unpin(vma);
 }
@@ -563,13 +563,13 @@ static int eb_reserve_vma(const struct i915_execbuffer *eb,
 	}
 
 	if (unlikely(exec_flags & EXEC_OBJECT_NEEDS_FENCE)) {
-		err = i915_vma_get_fence(vma);
+		err = i915_vma_pin_fence(vma);
 		if (unlikely(err)) {
 			i915_vma_unpin(vma);
 			return err;
 		}
 
-		if (i915_vma_pin_fence(vma))
+		if (vma->fence)
 			exec_flags |= __EXEC_OBJECT_HAS_FENCE;
 	}
 
@@ -974,7 +974,9 @@ static void *reloc_iomap(struct drm_i915_gem_object *obj,
 			return ERR_PTR(err);
 
 		vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0,
-					       PIN_MAPPABLE | PIN_NONBLOCK);
+					       PIN_MAPPABLE |
+					       PIN_NONBLOCK |
+					       PIN_NONFAULT);
 		if (IS_ERR(vma)) {
 			memset(&cache->node, 0, sizeof(cache->node));
 			err = drm_mm_insert_node_in_range
diff --git a/drivers/gpu/drm/i915/i915_gem_fence_reg.c b/drivers/gpu/drm/i915/i915_gem_fence_reg.c
index 2783d63bd1ad..012250f25255 100644
--- a/drivers/gpu/drm/i915/i915_gem_fence_reg.c
+++ b/drivers/gpu/drm/i915/i915_gem_fence_reg.c
@@ -240,7 +240,8 @@ static int fence_update(struct drm_i915_fence_reg *fence,
 		/* Ensure that all userspace CPU access is completed before
 		 * stealing the fence.
 		 */
-		i915_gem_release_mmap(fence->vma->obj);
+		GEM_BUG_ON(fence->vma->fence != fence);
+		i915_vma_revoke_mmap(fence->vma);
 
 		fence->vma->fence = NULL;
 		fence->vma = NULL;
@@ -280,8 +281,7 @@ static int fence_update(struct drm_i915_fence_reg *fence,
  *
  * 0 on success, negative error code on failure.
  */
-int
-i915_vma_put_fence(struct i915_vma *vma)
+int i915_vma_put_fence(struct i915_vma *vma)
 {
 	struct drm_i915_fence_reg *fence = vma->fence;
 
@@ -299,6 +299,8 @@ static struct drm_i915_fence_reg *fence_find(struct drm_i915_private *dev_priv)
 	struct drm_i915_fence_reg *fence;
 
 	list_for_each_entry(fence, &dev_priv->mm.fence_list, link) {
+		GEM_BUG_ON(fence->vma && fence->vma->fence != fence);
+
 		if (fence->pin_count)
 			continue;
 
@@ -313,7 +315,7 @@ static struct drm_i915_fence_reg *fence_find(struct drm_i915_private *dev_priv)
 }
 
 /**
- * i915_vma_get_fence - set up fencing for a vma
+ * i915_vma_pin_fence - set up fencing for a vma
  * @vma: vma to map through a fence reg
  *
  * When mapping objects through the GTT, userspace wants to be able to write
@@ -331,10 +333,11 @@ static struct drm_i915_fence_reg *fence_find(struct drm_i915_private *dev_priv)
  * 0 on success, negative error code on failure.
  */
 int
-i915_vma_get_fence(struct i915_vma *vma)
+i915_vma_pin_fence(struct i915_vma *vma)
 {
 	struct drm_i915_fence_reg *fence;
 	struct i915_vma *set = i915_gem_object_is_tiled(vma->obj) ? vma : NULL;
+	int err;
 
 	/* Note that we revoke fences on runtime suspend. Therefore the user
 	 * must keep the device awake whilst using the fence.
@@ -344,6 +347,8 @@ i915_vma_get_fence(struct i915_vma *vma)
 	/* Just update our place in the LRU if our fence is getting reused. */
 	if (vma->fence) {
 		fence = vma->fence;
+		GEM_BUG_ON(fence->vma != vma);
+		fence->pin_count++;
 		if (!fence->dirty) {
 			list_move_tail(&fence->link,
 				       &fence->i915->mm.fence_list);
@@ -353,10 +358,25 @@ i915_vma_get_fence(struct i915_vma *vma)
 		fence = fence_find(vma->vm->i915);
 		if (IS_ERR(fence))
 			return PTR_ERR(fence);
+
+		GEM_BUG_ON(fence->pin_count);
+		fence->pin_count++;
 	} else
 		return 0;
 
-	return fence_update(fence, set);
+	err = fence_update(fence, set);
+	if (err)
+		goto out_unpin;
+
+	GEM_BUG_ON(fence->vma != set);
+	GEM_BUG_ON(vma->fence != (set ? fence : NULL));
+
+	if (set)
+		return 0;
+
+out_unpin:
+	fence->pin_count--;
+	return err;
 }
 
 /**
@@ -429,8 +449,10 @@ void i915_gem_revoke_fences(struct drm_i915_private *dev_priv)
 	for (i = 0; i < dev_priv->num_fence_regs; i++) {
 		struct drm_i915_fence_reg *fence = &dev_priv->fence_regs[i];
 
+		GEM_BUG_ON(fence->vma && fence->vma->fence != fence);
+
 		if (fence->vma)
-			i915_gem_release_mmap(fence->vma->obj);
+			i915_vma_revoke_mmap(fence->vma);
 	}
 }
 
@@ -450,13 +472,15 @@ void i915_gem_restore_fences(struct drm_i915_private *dev_priv)
 		struct drm_i915_fence_reg *reg = &dev_priv->fence_regs[i];
 		struct i915_vma *vma = reg->vma;
 
+		GEM_BUG_ON(vma && vma->fence != reg);
+
 		/*
 		 * Commit delayed tiling changes if we have an object still
 		 * attached to the fence, otherwise just clear the fence.
 		 */
 		if (vma && !i915_gem_object_is_tiled(vma->obj)) {
 			GEM_BUG_ON(!reg->dirty);
-			GEM_BUG_ON(!list_empty(&vma->obj->userfault_link));
+			GEM_BUG_ON(i915_vma_has_userfault(vma));
 
 			list_move(&reg->link, &dev_priv->mm.fence_list);
 			vma->fence = NULL;
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.c b/drivers/gpu/drm/i915/i915_gem_gtt.c
index 4c82ceb8d318..daba55a4fce2 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.c
@@ -135,11 +135,12 @@ static inline void i915_ggtt_invalidate(struct drm_i915_private *i915)
 int intel_sanitize_enable_ppgtt(struct drm_i915_private *dev_priv,
 			       	int enable_ppgtt)
 {
-	bool has_aliasing_ppgtt;
 	bool has_full_ppgtt;
 	bool has_full_48bit_ppgtt;
 
-	has_aliasing_ppgtt = dev_priv->info.has_aliasing_ppgtt;
+	if (!dev_priv->info.has_aliasing_ppgtt)
+		return 0;
+
 	has_full_ppgtt = dev_priv->info.has_full_ppgtt;
 	has_full_48bit_ppgtt = dev_priv->info.has_full_48bit_ppgtt;
 
@@ -149,9 +150,6 @@ int intel_sanitize_enable_ppgtt(struct drm_i915_private *dev_priv,
 		has_full_48bit_ppgtt = intel_vgpu_has_full_48bit_ppgtt(dev_priv);
 	}
 
-	if (!has_aliasing_ppgtt)
-		return 0;
-
 	/*
 	 * We don't allow disabling PPGTT for gen9+ as it's a requirement for
 	 * execlists, the sole mechanism available to submit work.
@@ -188,7 +186,7 @@ int intel_sanitize_enable_ppgtt(struct drm_i915_private *dev_priv,
 			return 2;
 	}
 
-	return has_aliasing_ppgtt ? 1 : 0;
+	return 1;
 }
 
 static int ppgtt_bind_vma(struct i915_vma *vma,
@@ -205,8 +203,6 @@ static int ppgtt_bind_vma(struct i915_vma *vma,
 			return ret;
 	}
 
-	vma->pages = vma->obj->mm.pages;
-
 	/* Currently applicable only to VLV */
 	pte_flags = 0;
 	if (vma->obj->gt_ro)
@@ -222,6 +218,30 @@ static void ppgtt_unbind_vma(struct i915_vma *vma)
 	vma->vm->clear_range(vma->vm, vma->node.start, vma->size);
 }
 
+static int ppgtt_set_pages(struct i915_vma *vma)
+{
+	GEM_BUG_ON(vma->pages);
+
+	vma->pages = vma->obj->mm.pages;
+
+	vma->page_sizes = vma->obj->mm.page_sizes;
+
+	return 0;
+}
+
+static void clear_pages(struct i915_vma *vma)
+{
+	GEM_BUG_ON(!vma->pages);
+
+	if (vma->pages != vma->obj->mm.pages) {
+		sg_free_table(vma->pages);
+		kfree(vma->pages);
+	}
+	vma->pages = NULL;
+
+	memset(&vma->page_sizes, 0, sizeof(vma->page_sizes));
+}
+
 static gen8_pte_t gen8_pte_encode(dma_addr_t addr,
 				  enum i915_cache_level level)
 {
@@ -497,22 +517,63 @@ static void fill_page_dma_32(struct i915_address_space *vm,
 static int
 setup_scratch_page(struct i915_address_space *vm, gfp_t gfp)
 {
-	struct page *page;
+	struct page *page = NULL;
 	dma_addr_t addr;
+	int order;
 
-	page = alloc_page(gfp | __GFP_ZERO);
-	if (unlikely(!page))
-		return -ENOMEM;
+	/*
+	 * In order to utilize 64K pages for an object with a size < 2M, we will
+	 * need to support a 64K scratch page, given that every 16th entry for a
+	 * page-table operating in 64K mode must point to a properly aligned 64K
+	 * region, including any PTEs which happen to point to scratch.
+	 *
+	 * This is only relevant for the 48b PPGTT where we support
+	 * huge-gtt-pages, see also i915_vma_insert().
+	 *
+	 * TODO: we should really consider write-protecting the scratch-page and
+	 * sharing between ppgtt
+	 */
+	if (i915_vm_is_48bit(vm) &&
+	    HAS_PAGE_SIZES(vm->i915, I915_GTT_PAGE_SIZE_64K)) {
+		order = get_order(I915_GTT_PAGE_SIZE_64K);
+		page = alloc_pages(gfp | __GFP_ZERO | __GFP_NOWARN, order);
+		if (page) {
+			addr = dma_map_page(vm->dma, page, 0,
+					    I915_GTT_PAGE_SIZE_64K,
+					    PCI_DMA_BIDIRECTIONAL);
+			if (unlikely(dma_mapping_error(vm->dma, addr))) {
+				__free_pages(page, order);
+				page = NULL;
+			}
 
-	addr = dma_map_page(vm->dma, page, 0, PAGE_SIZE,
-			    PCI_DMA_BIDIRECTIONAL);
-	if (unlikely(dma_mapping_error(vm->dma, addr))) {
-		__free_page(page);
-		return -ENOMEM;
+			if (!IS_ALIGNED(addr, I915_GTT_PAGE_SIZE_64K)) {
+				dma_unmap_page(vm->dma, addr,
+					       I915_GTT_PAGE_SIZE_64K,
+					       PCI_DMA_BIDIRECTIONAL);
+				__free_pages(page, order);
+				page = NULL;
+			}
+		}
+	}
+
+	if (!page) {
+		order = 0;
+		page = alloc_page(gfp | __GFP_ZERO);
+		if (unlikely(!page))
+			return -ENOMEM;
+
+		addr = dma_map_page(vm->dma, page, 0, PAGE_SIZE,
+				    PCI_DMA_BIDIRECTIONAL);
+		if (unlikely(dma_mapping_error(vm->dma, addr))) {
+			__free_page(page);
+			return -ENOMEM;
+		}
 	}
 
 	vm->scratch_page.page = page;
 	vm->scratch_page.daddr = addr;
+	vm->scratch_page.order = order;
+
 	return 0;
 }
 
@@ -520,8 +581,9 @@ static void cleanup_scratch_page(struct i915_address_space *vm)
 {
 	struct i915_page_dma *p = &vm->scratch_page;
 
-	dma_unmap_page(vm->dma, p->daddr, PAGE_SIZE, PCI_DMA_BIDIRECTIONAL);
-	__free_page(p->page);
+	dma_unmap_page(vm->dma, p->daddr, BIT(p->order) << PAGE_SHIFT,
+		       PCI_DMA_BIDIRECTIONAL);
+	__free_pages(p->page, p->order);
 }
 
 static struct i915_page_table *alloc_pt(struct i915_address_space *vm)
@@ -989,6 +1051,105 @@ static void gen8_ppgtt_insert_3lvl(struct i915_address_space *vm,
 
 	gen8_ppgtt_insert_pte_entries(ppgtt, &ppgtt->pdp, &iter, &idx,
 				      cache_level);
+
+	vma->page_sizes.gtt = I915_GTT_PAGE_SIZE;
+}
+
+static void gen8_ppgtt_insert_huge_entries(struct i915_vma *vma,
+					   struct i915_page_directory_pointer **pdps,
+					   struct sgt_dma *iter,
+					   enum i915_cache_level cache_level)
+{
+	const gen8_pte_t pte_encode = gen8_pte_encode(0, cache_level);
+	u64 start = vma->node.start;
+	dma_addr_t rem = iter->sg->length;
+
+	do {
+		struct gen8_insert_pte idx = gen8_insert_pte(start);
+		struct i915_page_directory_pointer *pdp = pdps[idx.pml4e];
+		struct i915_page_directory *pd = pdp->page_directory[idx.pdpe];
+		unsigned int page_size;
+		bool maybe_64K = false;
+		gen8_pte_t encode = pte_encode;
+		gen8_pte_t *vaddr;
+		u16 index, max;
+
+		if (vma->page_sizes.sg & I915_GTT_PAGE_SIZE_2M &&
+		    IS_ALIGNED(iter->dma, I915_GTT_PAGE_SIZE_2M) &&
+		    rem >= I915_GTT_PAGE_SIZE_2M && !idx.pte) {
+			index = idx.pde;
+			max = I915_PDES;
+			page_size = I915_GTT_PAGE_SIZE_2M;
+
+			encode |= GEN8_PDE_PS_2M;
+
+			vaddr = kmap_atomic_px(pd);
+		} else {
+			struct i915_page_table *pt = pd->page_table[idx.pde];
+
+			index = idx.pte;
+			max = GEN8_PTES;
+			page_size = I915_GTT_PAGE_SIZE;
+
+			if (!index &&
+			    vma->page_sizes.sg & I915_GTT_PAGE_SIZE_64K &&
+			    IS_ALIGNED(iter->dma, I915_GTT_PAGE_SIZE_64K) &&
+			    (IS_ALIGNED(rem, I915_GTT_PAGE_SIZE_64K) ||
+			     rem >= (max - index) << PAGE_SHIFT))
+				maybe_64K = true;
+
+			vaddr = kmap_atomic_px(pt);
+		}
+
+		do {
+			GEM_BUG_ON(iter->sg->length < page_size);
+			vaddr[index++] = encode | iter->dma;
+
+			start += page_size;
+			iter->dma += page_size;
+			rem -= page_size;
+			if (iter->dma >= iter->max) {
+				iter->sg = __sg_next(iter->sg);
+				if (!iter->sg)
+					break;
+
+				rem = iter->sg->length;
+				iter->dma = sg_dma_address(iter->sg);
+				iter->max = iter->dma + rem;
+
+				if (maybe_64K && index < max &&
+				    !(IS_ALIGNED(iter->dma, I915_GTT_PAGE_SIZE_64K) &&
+				      (IS_ALIGNED(rem, I915_GTT_PAGE_SIZE_64K) ||
+				       rem >= (max - index) << PAGE_SHIFT)))
+					maybe_64K = false;
+
+				if (unlikely(!IS_ALIGNED(iter->dma, page_size)))
+					break;
+			}
+		} while (rem >= page_size && index < max);
+
+		kunmap_atomic(vaddr);
+
+		/*
+		 * Is it safe to mark the 2M block as 64K? -- Either we have
+		 * filled whole page-table with 64K entries, or filled part of
+		 * it and have reached the end of the sg table and we have
+		 * enough padding.
+		 */
+		if (maybe_64K &&
+		    (index == max ||
+		     (i915_vm_has_scratch_64K(vma->vm) &&
+		      !iter->sg && IS_ALIGNED(vma->node.start +
+					      vma->node.size,
+					      I915_GTT_PAGE_SIZE_2M)))) {
+			vaddr = kmap_atomic_px(pd);
+			vaddr[idx.pde] |= GEN8_PDE_IPS_64K;
+			kunmap_atomic(vaddr);
+			page_size = I915_GTT_PAGE_SIZE_64K;
+		}
+
+		vma->page_sizes.gtt |= page_size;
+	} while (iter->sg);
 }
 
 static void gen8_ppgtt_insert_4lvl(struct i915_address_space *vm,
@@ -1003,11 +1164,18 @@ static void gen8_ppgtt_insert_4lvl(struct i915_address_space *vm,
 		.max = iter.dma + iter.sg->length,
 	};
 	struct i915_page_directory_pointer **pdps = ppgtt->pml4.pdps;
-	struct gen8_insert_pte idx = gen8_insert_pte(vma->node.start);
 
-	while (gen8_ppgtt_insert_pte_entries(ppgtt, pdps[idx.pml4e++], &iter,
-					     &idx, cache_level))
-		GEM_BUG_ON(idx.pml4e >= GEN8_PML4ES_PER_PML4);
+	if (vma->page_sizes.sg > I915_GTT_PAGE_SIZE) {
+		gen8_ppgtt_insert_huge_entries(vma, pdps, &iter, cache_level);
+	} else {
+		struct gen8_insert_pte idx = gen8_insert_pte(vma->node.start);
+
+		while (gen8_ppgtt_insert_pte_entries(ppgtt, pdps[idx.pml4e++],
+						     &iter, &idx, cache_level))
+			GEM_BUG_ON(idx.pml4e >= GEN8_PML4ES_PER_PML4);
+
+		vma->page_sizes.gtt = I915_GTT_PAGE_SIZE;
+	}
 }
 
 static void gen8_free_page_tables(struct i915_address_space *vm,
@@ -1452,6 +1620,8 @@ static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
 	ppgtt->base.cleanup = gen8_ppgtt_cleanup;
 	ppgtt->base.unbind_vma = ppgtt_unbind_vma;
 	ppgtt->base.bind_vma = ppgtt_bind_vma;
+	ppgtt->base.set_pages = ppgtt_set_pages;
+	ppgtt->base.clear_pages = clear_pages;
 	ppgtt->debug_dump = gen8_dump_ppgtt;
 
 	return 0;
@@ -1726,6 +1896,8 @@ static void gen6_ppgtt_insert_entries(struct i915_address_space *vm,
 		}
 	} while (1);
 	kunmap_atomic(vaddr);
+
+	vma->page_sizes.gtt = I915_GTT_PAGE_SIZE;
 }
 
 static int gen6_alloc_va_range(struct i915_address_space *vm,
@@ -1894,6 +2066,8 @@ static int gen6_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
 	ppgtt->base.insert_entries = gen6_ppgtt_insert_entries;
 	ppgtt->base.unbind_vma = ppgtt_unbind_vma;
 	ppgtt->base.bind_vma = ppgtt_bind_vma;
+	ppgtt->base.set_pages = ppgtt_set_pages;
+	ppgtt->base.clear_pages = clear_pages;
 	ppgtt->base.cleanup = gen6_ppgtt_cleanup;
 	ppgtt->debug_dump = gen6_dump_ppgtt;
 
@@ -1961,6 +2135,23 @@ static void gtt_write_workarounds(struct drm_i915_private *dev_priv)
 		I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN9_L3_LRA_1_GPGPU_DEFAULT_VALUE_SKL);
 	else if (IS_GEN9_LP(dev_priv))
 		I915_WRITE(GEN8_L3_LRA_1_GPGPU, GEN9_L3_LRA_1_GPGPU_DEFAULT_VALUE_BXT);
+
+	/*
+	 * To support 64K PTEs we need to first enable the use of the
+	 * Intermediate-Page-Size(IPS) bit of the PDE field via some magical
+	 * mmio, otherwise the page-walker will simply ignore the IPS bit. This
+	 * shouldn't be needed after GEN10.
+	 *
+	 * 64K pages were first introduced from BDW+, although technically they
+	 * only *work* from gen9+. For pre-BDW we instead have the option for
+	 * 32K pages, but we don't currently have any support for it in our
+	 * driver.
+	 */
+	if (HAS_PAGE_SIZES(dev_priv, I915_GTT_PAGE_SIZE_64K) &&
+	    INTEL_GEN(dev_priv) <= 10)
+		I915_WRITE(GEN8_GAMW_ECO_DEV_RW_IA,
+			   I915_READ(GEN8_GAMW_ECO_DEV_RW_IA) |
+			   GAMW_ECO_ENABLE_64K_IPS_FIELD);
 }
 
 int i915_ppgtt_init_hw(struct drm_i915_private *dev_priv)
@@ -2405,12 +2596,6 @@ static int ggtt_bind_vma(struct i915_vma *vma,
 	struct drm_i915_gem_object *obj = vma->obj;
 	u32 pte_flags;
 
-	if (unlikely(!vma->pages)) {
-		int ret = i915_get_ggtt_vma_pages(vma);
-		if (ret)
-			return ret;
-	}
-
 	/* Currently applicable only to VLV */
 	pte_flags = 0;
 	if (obj->gt_ro)
@@ -2420,6 +2605,8 @@ static int ggtt_bind_vma(struct i915_vma *vma,
 	vma->vm->insert_entries(vma->vm, vma, cache_level, pte_flags);
 	intel_runtime_pm_put(i915);
 
+	vma->page_sizes.gtt = I915_GTT_PAGE_SIZE;
+
 	/*
 	 * Without aliasing PPGTT there's no difference between
 	 * GLOBAL/LOCAL_BIND, it's all the same ptes. Hence unconditionally
@@ -2447,12 +2634,6 @@ static int aliasing_gtt_bind_vma(struct i915_vma *vma,
 	u32 pte_flags;
 	int ret;
 
-	if (unlikely(!vma->pages)) {
-		ret = i915_get_ggtt_vma_pages(vma);
-		if (ret)
-			return ret;
-	}
-
 	/* Currently applicable only to VLV */
 	pte_flags = 0;
 	if (vma->obj->gt_ro)
@@ -2467,7 +2648,7 @@ static int aliasing_gtt_bind_vma(struct i915_vma *vma,
 							     vma->node.start,
 							     vma->size);
 			if (ret)
-				goto err_pages;
+				return ret;
 		}
 
 		appgtt->base.insert_entries(&appgtt->base, vma, cache_level,
@@ -2481,17 +2662,6 @@ static int aliasing_gtt_bind_vma(struct i915_vma *vma,
 	}
 
 	return 0;
-
-err_pages:
-	if (!(vma->flags & (I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND))) {
-		if (vma->pages != vma->obj->mm.pages) {
-			GEM_BUG_ON(!vma->pages);
-			sg_free_table(vma->pages);
-			kfree(vma->pages);
-		}
-		vma->pages = NULL;
-	}
-	return ret;
 }
 
 static void aliasing_gtt_unbind_vma(struct i915_vma *vma)
@@ -2529,6 +2699,21 @@ void i915_gem_gtt_finish_pages(struct drm_i915_gem_object *obj,
 	dma_unmap_sg(kdev, pages->sgl, pages->nents, PCI_DMA_BIDIRECTIONAL);
 }
 
+static int ggtt_set_pages(struct i915_vma *vma)
+{
+	int ret;
+
+	GEM_BUG_ON(vma->pages);
+
+	ret = i915_get_ggtt_vma_pages(vma);
+	if (ret)
+		return ret;
+
+	vma->page_sizes = vma->obj->mm.page_sizes;
+
+	return 0;
+}
+
 static void i915_gtt_color_adjust(const struct drm_mm_node *node,
 				  unsigned long color,
 				  u64 *start,
@@ -3151,6 +3336,8 @@ static int gen8_gmch_probe(struct i915_ggtt *ggtt)
 	ggtt->base.cleanup = gen6_gmch_remove;
 	ggtt->base.bind_vma = ggtt_bind_vma;
 	ggtt->base.unbind_vma = ggtt_unbind_vma;
+	ggtt->base.set_pages = ggtt_set_pages;
+	ggtt->base.clear_pages = clear_pages;
 	ggtt->base.insert_page = gen8_ggtt_insert_page;
 	ggtt->base.clear_range = nop_clear_range;
 	if (!USES_FULL_PPGTT(dev_priv) || intel_scanout_needs_vtd_wa(dev_priv))
@@ -3209,6 +3396,8 @@ static int gen6_gmch_probe(struct i915_ggtt *ggtt)
 	ggtt->base.insert_entries = gen6_ggtt_insert_entries;
 	ggtt->base.bind_vma = ggtt_bind_vma;
 	ggtt->base.unbind_vma = ggtt_unbind_vma;
+	ggtt->base.set_pages = ggtt_set_pages;
+	ggtt->base.clear_pages = clear_pages;
 	ggtt->base.cleanup = gen6_gmch_remove;
 
 	ggtt->invalidate = gen6_ggtt_invalidate;
@@ -3254,6 +3443,8 @@ static int i915_gmch_probe(struct i915_ggtt *ggtt)
 	ggtt->base.clear_range = i915_ggtt_clear_range;
 	ggtt->base.bind_vma = ggtt_bind_vma;
 	ggtt->base.unbind_vma = ggtt_unbind_vma;
+	ggtt->base.set_pages = ggtt_set_pages;
+	ggtt->base.clear_pages = clear_pages;
 	ggtt->base.cleanup = i915_gmch_remove;
 
 	ggtt->invalidate = gmch_ggtt_invalidate;
diff --git a/drivers/gpu/drm/i915/i915_gem_gtt.h b/drivers/gpu/drm/i915/i915_gem_gtt.h
index f62fb903dc24..93211a96fdad 100644
--- a/drivers/gpu/drm/i915/i915_gem_gtt.h
+++ b/drivers/gpu/drm/i915/i915_gem_gtt.h
@@ -42,7 +42,13 @@
 #include "i915_gem_request.h"
 #include "i915_selftest.h"
 
-#define I915_GTT_PAGE_SIZE 4096UL
+#define I915_GTT_PAGE_SIZE_4K BIT(12)
+#define I915_GTT_PAGE_SIZE_64K BIT(16)
+#define I915_GTT_PAGE_SIZE_2M BIT(21)
+
+#define I915_GTT_PAGE_SIZE I915_GTT_PAGE_SIZE_4K
+#define I915_GTT_MAX_PAGE_SIZE I915_GTT_PAGE_SIZE_2M
+
 #define I915_GTT_MIN_ALIGNMENT I915_GTT_PAGE_SIZE
 
 #define I915_FENCE_REG_NONE -1
@@ -148,6 +154,9 @@ typedef u64 gen8_ppgtt_pml4e_t;
 #define GEN8_PPAT_GET_AGE(x) ((x) & (3 << 4))
 #define CHV_PPAT_GET_SNOOP(x) ((x) & (1 << 6))
 
+#define GEN8_PDE_IPS_64K BIT(11)
+#define GEN8_PDE_PS_2M   BIT(7)
+
 struct sg_table;
 
 struct intel_rotation_info {
@@ -207,6 +216,7 @@ struct i915_vma;
 
 struct i915_page_dma {
 	struct page *page;
+	int order;
 	union {
 		dma_addr_t daddr;
 
@@ -329,6 +339,8 @@ struct i915_address_space {
 	int (*bind_vma)(struct i915_vma *vma,
 			enum i915_cache_level cache_level,
 			u32 flags);
+	int (*set_pages)(struct i915_vma *vma);
+	void (*clear_pages)(struct i915_vma *vma);
 
 	I915_SELFTEST_DECLARE(struct fault_attr fault_attr);
 };
@@ -341,6 +353,12 @@ i915_vm_is_48bit(const struct i915_address_space *vm)
 	return (vm->total - 1) >> 32;
 }
 
+static inline bool
+i915_vm_has_scratch_64K(struct i915_address_space *vm)
+{
+	return vm->scratch_page.order == get_order(I915_GTT_PAGE_SIZE_64K);
+}
+
 /* The Graphics Translation Table is the way in which GEN hardware translates a
  * Graphics Virtual Address into a Physical Address. In addition to the normal
  * collateral associated with any va->pa translations GEN hardware also has a
diff --git a/drivers/gpu/drm/i915/i915_gem_internal.c b/drivers/gpu/drm/i915/i915_gem_internal.c
index c1f64ddaf8aa..ee83ec838ee7 100644
--- a/drivers/gpu/drm/i915/i915_gem_internal.c
+++ b/drivers/gpu/drm/i915/i915_gem_internal.c
@@ -44,12 +44,12 @@ static void internal_free_pages(struct sg_table *st)
 	kfree(st);
 }
 
-static struct sg_table *
-i915_gem_object_get_pages_internal(struct drm_i915_gem_object *obj)
+static int i915_gem_object_get_pages_internal(struct drm_i915_gem_object *obj)
 {
 	struct drm_i915_private *i915 = to_i915(obj->base.dev);
 	struct sg_table *st;
 	struct scatterlist *sg;
+	unsigned int sg_page_sizes;
 	unsigned int npages;
 	int max_order;
 	gfp_t gfp;
@@ -78,16 +78,17 @@ i915_gem_object_get_pages_internal(struct drm_i915_gem_object *obj)
 create_st:
 	st = kmalloc(sizeof(*st), GFP_KERNEL);
 	if (!st)
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 
 	npages = obj->base.size / PAGE_SIZE;
 	if (sg_alloc_table(st, npages, GFP_KERNEL)) {
 		kfree(st);
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 	}
 
 	sg = st->sgl;
 	st->nents = 0;
+	sg_page_sizes = 0;
 
 	do {
 		int order = min(fls(npages) - 1, max_order);
@@ -105,6 +106,7 @@ create_st:
 		} while (1);
 
 		sg_set_page(sg, page, PAGE_SIZE << order, 0);
+		sg_page_sizes |= PAGE_SIZE << order;
 		st->nents++;
 
 		npages -= 1 << order;
@@ -132,13 +134,17 @@ create_st:
 	 * object are only valid whilst active and pinned.
 	 */
 	obj->mm.madv = I915_MADV_DONTNEED;
-	return st;
+
+	__i915_gem_object_set_pages(obj, st, sg_page_sizes);
+
+	return 0;
 
 err:
 	sg_set_page(sg, NULL, 0, 0);
 	sg_mark_end(sg);
 	internal_free_pages(st);
-	return ERR_PTR(-ENOMEM);
+
+	return -ENOMEM;
 }
 
 static void i915_gem_object_put_pages_internal(struct drm_i915_gem_object *obj,
diff --git a/drivers/gpu/drm/i915/i915_gem_object.h b/drivers/gpu/drm/i915/i915_gem_object.h
index c30d8f808185..d67f1cbe842d 100644
--- a/drivers/gpu/drm/i915/i915_gem_object.h
+++ b/drivers/gpu/drm/i915/i915_gem_object.h
@@ -69,7 +69,7 @@ struct drm_i915_gem_object_ops {
 	 * being released or under memory pressure (where we attempt to
 	 * reap pages for the shrinker).
 	 */
-	struct sg_table *(*get_pages)(struct drm_i915_gem_object *);
+	int (*get_pages)(struct drm_i915_gem_object *);
 	void (*put_pages)(struct drm_i915_gem_object *, struct sg_table *);
 
 	int (*pwrite)(struct drm_i915_gem_object *,
@@ -123,6 +123,7 @@ struct drm_i915_gem_object {
 	/**
 	 * Whether the object is currently in the GGTT mmap.
 	 */
+	unsigned int userfault_count;
 	struct list_head userfault_link;
 
 	struct list_head batch_pool_link;
@@ -169,6 +170,35 @@ struct drm_i915_gem_object {
 		struct sg_table *pages;
 		void *mapping;
 
+		/* TODO: whack some of this into the error state */
+		struct i915_page_sizes {
+			/**
+			 * The sg mask of the pages sg_table. i.e the mask of
+			 * of the lengths for each sg entry.
+			 */
+			unsigned int phys;
+
+			/**
+			 * The gtt page sizes we are allowed to use given the
+			 * sg mask and the supported page sizes. This will
+			 * express the smallest unit we can use for the whole
+			 * object, as well as the larger sizes we may be able
+			 * to use opportunistically.
+			 */
+			unsigned int sg;
+
+			/**
+			 * The actual gtt page size usage. Since we can have
+			 * multiple vma associated with this object we need to
+			 * prevent any trampling of state, hence a copy of this
+			 * struct also lives in each vma, therefore the gtt
+			 * value here should only be read/write through the vma.
+			 */
+			unsigned int gtt;
+		} page_sizes;
+
+		I915_SELFTEST_DECLARE(unsigned int page_mask);
+
 		struct i915_gem_object_page_iter {
 			struct scatterlist *sg_pos;
 			unsigned int sg_idx; /* in pages, but 32bit eek! */
diff --git a/drivers/gpu/drm/i915/i915_gem_request.c b/drivers/gpu/drm/i915/i915_gem_request.c
index 4eb1a76731b2..d140fcf5c6a3 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.c
+++ b/drivers/gpu/drm/i915/i915_gem_request.c
@@ -186,7 +186,7 @@ i915_priotree_init(struct i915_priotree *pt)
 	INIT_LIST_HEAD(&pt->signalers_list);
 	INIT_LIST_HEAD(&pt->waiters_list);
 	INIT_LIST_HEAD(&pt->link);
-	pt->priority = INT_MIN;
+	pt->priority = I915_PRIORITY_INVALID;
 }
 
 static int reset_all_global_seqno(struct drm_i915_private *i915, u32 seqno)
@@ -416,7 +416,7 @@ static void i915_gem_request_retire(struct drm_i915_gem_request *request)
 
 	spin_lock_irq(&request->lock);
 	if (request->waitboost)
-		atomic_dec(&request->i915->rps.num_waiters);
+		atomic_dec(&request->i915->gt_pm.rps.num_waiters);
 	dma_fence_signal_locked(&request->fence);
 	spin_unlock_irq(&request->lock);
 
@@ -556,7 +556,16 @@ submit_notify(struct i915_sw_fence *fence, enum i915_sw_fence_notify state)
 	switch (state) {
 	case FENCE_COMPLETE:
 		trace_i915_gem_request_submit(request);
+		/*
+		 * We need to serialize use of the submit_request() callback with its
+		 * hotplugging performed during an emergency i915_gem_set_wedged().
+		 * We use the RCU mechanism to mark the critical section in order to
+		 * force i915_gem_set_wedged() to wait until the submit_request() is
+		 * completed before proceeding.
+		 */
+		rcu_read_lock();
 		request->engine->submit_request(request);
+		rcu_read_unlock();
 		break;
 
 	case FENCE_FREE:
@@ -587,6 +596,13 @@ i915_gem_request_alloc(struct intel_engine_cs *engine,
 
 	lockdep_assert_held(&dev_priv->drm.struct_mutex);
 
+	/*
+	 * Preempt contexts are reserved for exclusive use to inject a
+	 * preemption context switch. They are never to be used for any trivial
+	 * request!
+	 */
+	GEM_BUG_ON(ctx == dev_priv->preempt_context);
+
 	/* ABI: Before userspace accesses the GPU (e.g. execbuffer), report
 	 * EIO if the GPU is already wedged.
 	 */
diff --git a/drivers/gpu/drm/i915/i915_gem_request.h b/drivers/gpu/drm/i915/i915_gem_request.h
index 96eb52471dad..26249f39de67 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.h
+++ b/drivers/gpu/drm/i915/i915_gem_request.h
@@ -30,6 +30,8 @@
 #include "i915_gem.h"
 #include "i915_sw_fence.h"
 
+#include <uapi/drm/i915_drm.h>
+
 struct drm_file;
 struct drm_i915_gem_object;
 struct drm_i915_gem_request;
@@ -69,9 +71,14 @@ struct i915_priotree {
 	struct list_head waiters_list; /* those after us, they depend upon us */
 	struct list_head link;
 	int priority;
-#define I915_PRIORITY_MAX 1024
-#define I915_PRIORITY_NORMAL 0
-#define I915_PRIORITY_MIN (-I915_PRIORITY_MAX)
+};
+
+enum {
+	I915_PRIORITY_MIN = I915_CONTEXT_MIN_USER_PRIORITY - 1,
+	I915_PRIORITY_NORMAL = I915_CONTEXT_DEFAULT_PRIORITY,
+	I915_PRIORITY_MAX = I915_CONTEXT_MAX_USER_PRIORITY + 1,
+
+	I915_PRIORITY_INVALID = INT_MIN
 };
 
 struct i915_gem_capture_list {
diff --git a/drivers/gpu/drm/i915/i915_gem_stolen.c b/drivers/gpu/drm/i915/i915_gem_stolen.c
index 507c9f0d8df1..54fd4cfa9d07 100644
--- a/drivers/gpu/drm/i915/i915_gem_stolen.c
+++ b/drivers/gpu/drm/i915/i915_gem_stolen.c
@@ -539,12 +539,18 @@ i915_pages_create_for_stolen(struct drm_device *dev,
 	return st;
 }
 
-static struct sg_table *
-i915_gem_object_get_pages_stolen(struct drm_i915_gem_object *obj)
+static int i915_gem_object_get_pages_stolen(struct drm_i915_gem_object *obj)
 {
-	return i915_pages_create_for_stolen(obj->base.dev,
-					    obj->stolen->start,
-					    obj->stolen->size);
+	struct sg_table *pages =
+		i915_pages_create_for_stolen(obj->base.dev,
+					     obj->stolen->start,
+					     obj->stolen->size);
+	if (IS_ERR(pages))
+		return PTR_ERR(pages);
+
+	__i915_gem_object_set_pages(obj, pages, obj->stolen->size);
+
+	return 0;
 }
 
 static void i915_gem_object_put_pages_stolen(struct drm_i915_gem_object *obj,
diff --git a/drivers/gpu/drm/i915/i915_gem_userptr.c b/drivers/gpu/drm/i915/i915_gem_userptr.c
index 2d4996de7331..4d712a4db63b 100644
--- a/drivers/gpu/drm/i915/i915_gem_userptr.c
+++ b/drivers/gpu/drm/i915/i915_gem_userptr.c
@@ -164,7 +164,6 @@ static struct i915_mmu_notifier *
 i915_mmu_notifier_create(struct mm_struct *mm)
 {
 	struct i915_mmu_notifier *mn;
-	int ret;
 
 	mn = kmalloc(sizeof(*mn), GFP_KERNEL);
 	if (mn == NULL)
@@ -179,14 +178,6 @@ i915_mmu_notifier_create(struct mm_struct *mm)
 		return ERR_PTR(-ENOMEM);
 	}
 
-	 /* Protected by mmap_sem (write-lock) */
-	ret = __mmu_notifier_register(&mn->mn, mm);
-	if (ret) {
-		destroy_workqueue(mn->wq);
-		kfree(mn);
-		return ERR_PTR(ret);
-	}
-
 	return mn;
 }
 
@@ -210,23 +201,40 @@ i915_gem_userptr_release__mmu_notifier(struct drm_i915_gem_object *obj)
 static struct i915_mmu_notifier *
 i915_mmu_notifier_find(struct i915_mm_struct *mm)
 {
-	struct i915_mmu_notifier *mn = mm->mn;
+	struct i915_mmu_notifier *mn;
+	int err = 0;
 
 	mn = mm->mn;
 	if (mn)
 		return mn;
 
+	mn = i915_mmu_notifier_create(mm->mm);
+	if (IS_ERR(mn))
+		err = PTR_ERR(mn);
+
 	down_write(&mm->mm->mmap_sem);
 	mutex_lock(&mm->i915->mm_lock);
-	if ((mn = mm->mn) == NULL) {
-		mn = i915_mmu_notifier_create(mm->mm);
-		if (!IS_ERR(mn))
-			mm->mn = mn;
+	if (mm->mn == NULL && !err) {
+		/* Protected by mmap_sem (write-lock) */
+		err = __mmu_notifier_register(&mn->mn, mm->mm);
+		if (!err) {
+			/* Protected by mm_lock */
+			mm->mn = fetch_and_zero(&mn);
+		}
+	} else {
+		/* someone else raced and successfully installed the mmu
+		 * notifier, we can cancel our own errors */
+		err = 0;
 	}
 	mutex_unlock(&mm->i915->mm_lock);
 	up_write(&mm->mm->mmap_sem);
 
-	return mn;
+	if (mn) {
+		destroy_workqueue(mn->wq);
+		kfree(mn);
+	}
+
+	return err ? ERR_PTR(err) : mm->mn;
 }
 
 static int
@@ -405,6 +413,7 @@ __i915_gem_userptr_alloc_pages(struct drm_i915_gem_object *obj,
 {
 	unsigned int max_segment = i915_sg_segment_size();
 	struct sg_table *st;
+	unsigned int sg_page_sizes;
 	int ret;
 
 	st = kmalloc(sizeof(*st), GFP_KERNEL);
@@ -434,6 +443,10 @@ alloc_table:
 		return ERR_PTR(ret);
 	}
 
+	sg_page_sizes = i915_sg_page_sizes(st->sgl);
+
+	__i915_gem_object_set_pages(obj, st, sg_page_sizes);
+
 	return st;
 }
 
@@ -521,7 +534,6 @@ __i915_gem_userptr_get_pages_worker(struct work_struct *_work)
 			pages = __i915_gem_userptr_alloc_pages(obj, pvec,
 							       npages);
 			if (!IS_ERR(pages)) {
-				__i915_gem_object_set_pages(obj, pages);
 				pinned = 0;
 				pages = NULL;
 			}
@@ -582,8 +594,7 @@ __i915_gem_userptr_get_pages_schedule(struct drm_i915_gem_object *obj)
 	return ERR_PTR(-EAGAIN);
 }
 
-static struct sg_table *
-i915_gem_userptr_get_pages(struct drm_i915_gem_object *obj)
+static int i915_gem_userptr_get_pages(struct drm_i915_gem_object *obj)
 {
 	const int num_pages = obj->base.size >> PAGE_SHIFT;
 	struct mm_struct *mm = obj->userptr.mm->mm;
@@ -612,9 +623,9 @@ i915_gem_userptr_get_pages(struct drm_i915_gem_object *obj)
 	if (obj->userptr.work) {
 		/* active flag should still be held for the pending work */
 		if (IS_ERR(obj->userptr.work))
-			return ERR_CAST(obj->userptr.work);
+			return PTR_ERR(obj->userptr.work);
 		else
-			return ERR_PTR(-EAGAIN);
+			return -EAGAIN;
 	}
 
 	pvec = NULL;
@@ -650,7 +661,7 @@ i915_gem_userptr_get_pages(struct drm_i915_gem_object *obj)
 		release_pages(pvec, pinned, 0);
 	kvfree(pvec);
 
-	return pages;
+	return PTR_ERR_OR_ZERO(pages);
 }
 
 static void
diff --git a/drivers/gpu/drm/i915/i915_gemfs.c b/drivers/gpu/drm/i915/i915_gemfs.c
new file mode 100644
index 000000000000..e2993857df37
--- /dev/null
+++ b/drivers/gpu/drm/i915/i915_gemfs.c
@@ -0,0 +1,74 @@
+/*
+ * Copyright © 2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/fs.h>
+#include <linux/mount.h>
+#include <linux/pagemap.h>
+
+#include "i915_drv.h"
+#include "i915_gemfs.h"
+
+int i915_gemfs_init(struct drm_i915_private *i915)
+{
+	struct file_system_type *type;
+	struct vfsmount *gemfs;
+
+	type = get_fs_type("tmpfs");
+	if (!type)
+		return -ENODEV;
+
+	gemfs = kern_mount(type);
+	if (IS_ERR(gemfs))
+		return PTR_ERR(gemfs);
+
+	/*
+	 * Enable huge-pages for objects that are at least HPAGE_PMD_SIZE, most
+	 * likely 2M. Note that within_size may overallocate huge-pages, if say
+	 * we allocate an object of size 2M + 4K, we may get 2M + 2M, but under
+	 * memory pressure shmem should split any huge-pages which can be
+	 * shrunk.
+	 */
+
+	if (has_transparent_hugepage()) {
+		struct super_block *sb = gemfs->mnt_sb;
+		char options[] = "huge=within_size";
+		int flags = 0;
+		int err;
+
+		err = sb->s_op->remount_fs(sb, &flags, options);
+		if (err) {
+			kern_unmount(gemfs);
+			return err;
+		}
+	}
+
+	i915->mm.gemfs = gemfs;
+
+	return 0;
+}
+
+void i915_gemfs_fini(struct drm_i915_private *i915)
+{
+	kern_unmount(i915->mm.gemfs);
+}
diff --git a/drivers/gpu/drm/i915/i915_gemfs.h b/drivers/gpu/drm/i915/i915_gemfs.h
new file mode 100644
index 000000000000..cca8bdc5b93e
--- /dev/null
+++ b/drivers/gpu/drm/i915/i915_gemfs.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright © 2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __I915_GEMFS_H__
+#define __I915_GEMFS_H__
+
+struct drm_i915_private;
+
+int i915_gemfs_init(struct drm_i915_private *i915);
+
+void i915_gemfs_fini(struct drm_i915_private *i915);
+
+#endif
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c
index c14552ab270b..653fb69e7ecb 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.c
+++ b/drivers/gpu/drm/i915/i915_gpu_error.c
@@ -377,9 +377,9 @@ static void error_print_request(struct drm_i915_error_state_buf *m,
 	if (!erq->seqno)
 		return;
 
-	err_printf(m, "%s pid %d, ban score %d, seqno %8x:%08x, emitted %dms ago, head %08x, tail %08x\n",
+	err_printf(m, "%s pid %d, ban score %d, seqno %8x:%08x, prio %d, emitted %dms ago, head %08x, tail %08x\n",
 		   prefix, erq->pid, erq->ban_score,
-		   erq->context, erq->seqno,
+		   erq->context, erq->seqno, erq->priority,
 		   jiffies_to_msecs(jiffies - erq->jiffies),
 		   erq->head, erq->tail);
 }
@@ -388,9 +388,9 @@ static void error_print_context(struct drm_i915_error_state_buf *m,
 				const char *header,
 				const struct drm_i915_error_context *ctx)
 {
-	err_printf(m, "%s%s[%d] user_handle %d hw_id %d, ban score %d guilty %d active %d\n",
+	err_printf(m, "%s%s[%d] user_handle %d hw_id %d, prio %d, ban score %d guilty %d active %d\n",
 		   header, ctx->comm, ctx->pid, ctx->handle, ctx->hw_id,
-		   ctx->ban_score, ctx->guilty, ctx->active);
+		   ctx->priority, ctx->ban_score, ctx->guilty, ctx->active);
 }
 
 static void error_print_engine(struct drm_i915_error_state_buf *m,
@@ -1271,6 +1271,7 @@ static void record_request(struct drm_i915_gem_request *request,
 			   struct drm_i915_error_request *erq)
 {
 	erq->context = request->ctx->hw_id;
+	erq->priority = request->priotree.priority;
 	erq->ban_score = atomic_read(&request->ctx->ban_score);
 	erq->seqno = request->global_seqno;
 	erq->jiffies = request->emitted_jiffies;
@@ -1364,6 +1365,7 @@ static void record_context(struct drm_i915_error_context *e,
 
 	e->handle = ctx->user_handle;
 	e->hw_id = ctx->hw_id;
+	e->priority = ctx->priority;
 	e->ban_score = atomic_read(&ctx->ban_score);
 	e->guilty = atomic_read(&ctx->guilty_count);
 	e->active = atomic_read(&ctx->active_count);
@@ -1672,8 +1674,8 @@ static void i915_capture_gen_state(struct drm_i915_private *dev_priv,
 				   struct i915_gpu_state *error)
 {
 	error->awake = dev_priv->gt.awake;
-	error->wakelock = atomic_read(&dev_priv->pm.wakeref_count);
-	error->suspended = dev_priv->pm.suspended;
+	error->wakelock = atomic_read(&dev_priv->runtime_pm.wakeref_count);
+	error->suspended = dev_priv->runtime_pm.suspended;
 
 	error->iommu = -1;
 #ifdef CONFIG_INTEL_IOMMU
diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c b/drivers/gpu/drm/i915/i915_guc_submission.c
index 04f1281d81a5..a2e8114b739d 100644
--- a/drivers/gpu/drm/i915/i915_guc_submission.c
+++ b/drivers/gpu/drm/i915/i915_guc_submission.c
@@ -21,12 +21,13 @@
  * IN THE SOFTWARE.
  *
  */
-#include <linux/circ_buf.h>
-#include "i915_drv.h"
-#include "intel_uc.h"
 
+#include <linux/circ_buf.h>
 #include <trace/events/dma_fence.h>
 
+#include "i915_guc_submission.h"
+#include "i915_drv.h"
+
 /**
  * DOC: GuC-based command submission
  *
@@ -337,7 +338,7 @@ static void guc_stage_desc_init(struct intel_guc *guc,
 
 	for_each_engine_masked(engine, dev_priv, client->engines, tmp) {
 		struct intel_context *ce = &ctx->engine[engine->id];
-		uint32_t guc_engine_id = engine->guc_id;
+		u32 guc_engine_id = engine->guc_id;
 		struct guc_execlist_context *lrc = &desc->lrc[guc_engine_id];
 
 		/* TODO: We have a design issue to be solved here. Only when we
@@ -387,13 +388,13 @@ static void guc_stage_desc_init(struct intel_guc *guc,
 	gfx_addr = guc_ggtt_offset(client->vma);
 	desc->db_trigger_phy = sg_dma_address(client->vma->pages->sgl) +
 				client->doorbell_offset;
-	desc->db_trigger_cpu = (uintptr_t)__get_doorbell(client);
+	desc->db_trigger_cpu = ptr_to_u64(__get_doorbell(client));
 	desc->db_trigger_uk = gfx_addr + client->doorbell_offset;
 	desc->process_desc = gfx_addr + client->proc_desc_offset;
 	desc->wq_addr = gfx_addr + GUC_DB_SIZE;
 	desc->wq_size = GUC_WQ_SIZE;
 
-	desc->desc_private = (uintptr_t)client;
+	desc->desc_private = ptr_to_u64(client);
 }
 
 static void guc_stage_desc_fini(struct intel_guc *guc,
@@ -499,7 +500,7 @@ static void i915_guc_submit(struct intel_engine_cs *engine)
 	const unsigned int engine_id = engine->id;
 	unsigned int n;
 
-	for (n = 0; n < ARRAY_SIZE(execlists->port); n++) {
+	for (n = 0; n < execlists_num_ports(execlists); n++) {
 		struct drm_i915_gem_request *rq;
 		unsigned int count;
 
@@ -643,48 +644,6 @@ static void i915_guc_irq_handler(unsigned long data)
  * path of i915_guc_submit() above.
  */
 
-/**
- * intel_guc_allocate_vma() - Allocate a GGTT VMA for GuC usage
- * @guc:	the guc
- * @size:	size of area to allocate (both virtual space and memory)
- *
- * This is a wrapper to create an object for use with the GuC. In order to
- * use it inside the GuC, an object needs to be pinned lifetime, so we allocate
- * both some backing storage and a range inside the Global GTT. We must pin
- * it in the GGTT somewhere other than than [0, GUC_WOPCM_TOP) because that
- * range is reserved inside GuC.
- *
- * Return:	A i915_vma if successful, otherwise an ERR_PTR.
- */
-struct i915_vma *intel_guc_allocate_vma(struct intel_guc *guc, u32 size)
-{
-	struct drm_i915_private *dev_priv = guc_to_i915(guc);
-	struct drm_i915_gem_object *obj;
-	struct i915_vma *vma;
-	int ret;
-
-	obj = i915_gem_object_create(dev_priv, size);
-	if (IS_ERR(obj))
-		return ERR_CAST(obj);
-
-	vma = i915_vma_instance(obj, &dev_priv->ggtt.base, NULL);
-	if (IS_ERR(vma))
-		goto err;
-
-	ret = i915_vma_pin(vma, 0, PAGE_SIZE,
-			   PIN_GLOBAL | PIN_OFFSET_BIAS | GUC_WOPCM_TOP);
-	if (ret) {
-		vma = ERR_PTR(ret);
-		goto err;
-	}
-
-	return vma;
-
-err:
-	i915_gem_object_put(obj);
-	return vma;
-}
-
 /* Check that a doorbell register is in the expected state */
 static bool doorbell_ok(struct intel_guc *guc, u16 db_id)
 {
@@ -796,8 +755,8 @@ static int guc_init_doorbell_hw(struct intel_guc *guc)
  */
 static struct i915_guc_client *
 guc_client_alloc(struct drm_i915_private *dev_priv,
-		 uint32_t engines,
-		 uint32_t priority,
+		 u32 engines,
+		 u32 priority,
 		 struct i915_gem_context *ctx)
 {
 	struct i915_guc_client *client;
@@ -1069,6 +1028,7 @@ void i915_guc_submission_fini(struct drm_i915_private *dev_priv)
 
 static void guc_interrupts_capture(struct drm_i915_private *dev_priv)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 	int irqs;
@@ -1105,12 +1065,13 @@ static void guc_interrupts_capture(struct drm_i915_private *dev_priv)
 	 * Here we CLEAR REDIRECT_TO_GUC bit in pm_intrmsk_mbz, which will
 	 * result in the register bit being left SET!
 	 */
-	dev_priv->rps.pm_intrmsk_mbz |= ARAT_EXPIRED_INTRMSK;
-	dev_priv->rps.pm_intrmsk_mbz &= ~GEN8_PMINTR_DISABLE_REDIRECT_TO_GUC;
+	rps->pm_intrmsk_mbz |= ARAT_EXPIRED_INTRMSK;
+	rps->pm_intrmsk_mbz &= ~GEN8_PMINTR_DISABLE_REDIRECT_TO_GUC;
 }
 
 static void guc_interrupts_release(struct drm_i915_private *dev_priv)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
 	int irqs;
@@ -1129,8 +1090,8 @@ static void guc_interrupts_release(struct drm_i915_private *dev_priv)
 	I915_WRITE(GUC_VCS2_VCS1_IER, 0);
 	I915_WRITE(GUC_WD_VECS_IER, 0);
 
-	dev_priv->rps.pm_intrmsk_mbz |= GEN8_PMINTR_DISABLE_REDIRECT_TO_GUC;
-	dev_priv->rps.pm_intrmsk_mbz &= ~ARAT_EXPIRED_INTRMSK;
+	rps->pm_intrmsk_mbz |= GEN8_PMINTR_DISABLE_REDIRECT_TO_GUC;
+	rps->pm_intrmsk_mbz &= ~ARAT_EXPIRED_INTRMSK;
 }
 
 int i915_guc_submission_enable(struct drm_i915_private *dev_priv)
@@ -1212,55 +1173,3 @@ void i915_guc_submission_disable(struct drm_i915_private *dev_priv)
 	guc_client_free(guc->execbuf_client);
 	guc->execbuf_client = NULL;
 }
-
-/**
- * intel_guc_suspend() - notify GuC entering suspend state
- * @dev_priv:	i915 device private
- */
-int intel_guc_suspend(struct drm_i915_private *dev_priv)
-{
-	struct intel_guc *guc = &dev_priv->guc;
-	struct i915_gem_context *ctx;
-	u32 data[3];
-
-	if (guc->fw.load_status != INTEL_UC_FIRMWARE_SUCCESS)
-		return 0;
-
-	gen9_disable_guc_interrupts(dev_priv);
-
-	ctx = dev_priv->kernel_context;
-
-	data[0] = INTEL_GUC_ACTION_ENTER_S_STATE;
-	/* any value greater than GUC_POWER_D0 */
-	data[1] = GUC_POWER_D1;
-	/* first page is shared data with GuC */
-	data[2] = guc_ggtt_offset(ctx->engine[RCS].state) + LRC_GUCSHR_PN * PAGE_SIZE;
-
-	return intel_guc_send(guc, data, ARRAY_SIZE(data));
-}
-
-/**
- * intel_guc_resume() - notify GuC resuming from suspend state
- * @dev_priv:	i915 device private
- */
-int intel_guc_resume(struct drm_i915_private *dev_priv)
-{
-	struct intel_guc *guc = &dev_priv->guc;
-	struct i915_gem_context *ctx;
-	u32 data[3];
-
-	if (guc->fw.load_status != INTEL_UC_FIRMWARE_SUCCESS)
-		return 0;
-
-	if (i915_modparams.guc_log_level >= 0)
-		gen9_enable_guc_interrupts(dev_priv);
-
-	ctx = dev_priv->kernel_context;
-
-	data[0] = INTEL_GUC_ACTION_EXIT_S_STATE;
-	data[1] = GUC_POWER_D0;
-	/* first page is shared data with GuC */
-	data[2] = guc_ggtt_offset(ctx->engine[RCS].state) + LRC_GUCSHR_PN * PAGE_SIZE;
-
-	return intel_guc_send(guc, data, ARRAY_SIZE(data));
-}
diff --git a/drivers/gpu/drm/i915/i915_guc_submission.h b/drivers/gpu/drm/i915/i915_guc_submission.h
new file mode 100644
index 000000000000..cb4353b59059
--- /dev/null
+++ b/drivers/gpu/drm/i915/i915_guc_submission.h
@@ -0,0 +1,80 @@
+/*
+ * Copyright © 2014-2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _I915_GUC_SUBMISSION_H_
+#define _I915_GUC_SUBMISSION_H_
+
+#include <linux/spinlock.h>
+
+#include "i915_gem.h"
+
+struct drm_i915_private;
+
+/*
+ * This structure primarily describes the GEM object shared with the GuC.
+ * The specs sometimes refer to this object as a "GuC context", but we use
+ * the term "client" to avoid confusion with hardware contexts. This
+ * GEM object is held for the entire lifetime of our interaction with
+ * the GuC, being allocated before the GuC is loaded with its firmware.
+ * Because there's no way to update the address used by the GuC after
+ * initialisation, the shared object must stay pinned into the GGTT as
+ * long as the GuC is in use. We also keep the first page (only) mapped
+ * into kernel address space, as it includes shared data that must be
+ * updated on every request submission.
+ *
+ * The single GEM object described here is actually made up of several
+ * separate areas, as far as the GuC is concerned. The first page (kept
+ * kmap'd) includes the "process descriptor" which holds sequence data for
+ * the doorbell, and one cacheline which actually *is* the doorbell; a
+ * write to this will "ring the doorbell" (i.e. send an interrupt to the
+ * GuC). The subsequent  pages of the client object constitute the work
+ * queue (a circular array of work items), again described in the process
+ * descriptor. Work queue pages are mapped momentarily as required.
+ */
+struct i915_guc_client {
+	struct i915_vma *vma;
+	void *vaddr;
+	struct i915_gem_context *owner;
+	struct intel_guc *guc;
+
+	/* bitmap of (host) engine ids */
+	u32 engines;
+	u32 priority;
+	u32 stage_id;
+	u32 proc_desc_offset;
+
+	u16 doorbell_id;
+	unsigned long doorbell_offset;
+
+	spinlock_t wq_lock;
+	/* Per-engine counts of GuC submissions */
+	u64 submissions[I915_NUM_ENGINES];
+};
+
+int i915_guc_submission_init(struct drm_i915_private *dev_priv);
+int i915_guc_submission_enable(struct drm_i915_private *dev_priv);
+void i915_guc_submission_disable(struct drm_i915_private *dev_priv);
+void i915_guc_submission_fini(struct drm_i915_private *dev_priv);
+
+#endif
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index efd7827ff181..b1296a55c1e4 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -404,19 +404,21 @@ void gen6_reset_rps_interrupts(struct drm_i915_private *dev_priv)
 {
 	spin_lock_irq(&dev_priv->irq_lock);
 	gen6_reset_pm_iir(dev_priv, dev_priv->pm_rps_events);
-	dev_priv->rps.pm_iir = 0;
+	dev_priv->gt_pm.rps.pm_iir = 0;
 	spin_unlock_irq(&dev_priv->irq_lock);
 }
 
 void gen6_enable_rps_interrupts(struct drm_i915_private *dev_priv)
 {
-	if (READ_ONCE(dev_priv->rps.interrupts_enabled))
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
+	if (READ_ONCE(rps->interrupts_enabled))
 		return;
 
 	spin_lock_irq(&dev_priv->irq_lock);
-	WARN_ON_ONCE(dev_priv->rps.pm_iir);
+	WARN_ON_ONCE(rps->pm_iir);
 	WARN_ON_ONCE(I915_READ(gen6_pm_iir(dev_priv)) & dev_priv->pm_rps_events);
-	dev_priv->rps.interrupts_enabled = true;
+	rps->interrupts_enabled = true;
 	gen6_enable_pm_irq(dev_priv, dev_priv->pm_rps_events);
 
 	spin_unlock_irq(&dev_priv->irq_lock);
@@ -424,11 +426,13 @@ void gen6_enable_rps_interrupts(struct drm_i915_private *dev_priv)
 
 void gen6_disable_rps_interrupts(struct drm_i915_private *dev_priv)
 {
-	if (!READ_ONCE(dev_priv->rps.interrupts_enabled))
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
+	if (!READ_ONCE(rps->interrupts_enabled))
 		return;
 
 	spin_lock_irq(&dev_priv->irq_lock);
-	dev_priv->rps.interrupts_enabled = false;
+	rps->interrupts_enabled = false;
 
 	I915_WRITE(GEN6_PMINTRMSK, gen6_sanitize_rps_pm_mask(dev_priv, ~0u));
 
@@ -442,7 +446,7 @@ void gen6_disable_rps_interrupts(struct drm_i915_private *dev_priv)
 	 * we will reset the GPU to minimum frequencies, so the current
 	 * state of the worker can be discarded.
 	 */
-	cancel_work_sync(&dev_priv->rps.work);
+	cancel_work_sync(&rps->work);
 	gen6_reset_rps_interrupts(dev_priv);
 }
 
@@ -1119,12 +1123,13 @@ static void vlv_c0_read(struct drm_i915_private *dev_priv,
 
 void gen6_rps_reset_ei(struct drm_i915_private *dev_priv)
 {
-	memset(&dev_priv->rps.ei, 0, sizeof(dev_priv->rps.ei));
+	memset(&dev_priv->gt_pm.rps.ei, 0, sizeof(dev_priv->gt_pm.rps.ei));
 }
 
 static u32 vlv_wa_c0_ei(struct drm_i915_private *dev_priv, u32 pm_iir)
 {
-	const struct intel_rps_ei *prev = &dev_priv->rps.ei;
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+	const struct intel_rps_ei *prev = &rps->ei;
 	struct intel_rps_ei now;
 	u32 events = 0;
 
@@ -1151,28 +1156,29 @@ static u32 vlv_wa_c0_ei(struct drm_i915_private *dev_priv, u32 pm_iir)
 		c0 = max(render, media);
 		c0 *= 1000 * 100 << 8; /* to usecs and scale to threshold% */
 
-		if (c0 > time * dev_priv->rps.up_threshold)
+		if (c0 > time * rps->up_threshold)
 			events = GEN6_PM_RP_UP_THRESHOLD;
-		else if (c0 < time * dev_priv->rps.down_threshold)
+		else if (c0 < time * rps->down_threshold)
 			events = GEN6_PM_RP_DOWN_THRESHOLD;
 	}
 
-	dev_priv->rps.ei = now;
+	rps->ei = now;
 	return events;
 }
 
 static void gen6_pm_rps_work(struct work_struct *work)
 {
 	struct drm_i915_private *dev_priv =
-		container_of(work, struct drm_i915_private, rps.work);
+		container_of(work, struct drm_i915_private, gt_pm.rps.work);
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	bool client_boost = false;
 	int new_delay, adj, min, max;
 	u32 pm_iir = 0;
 
 	spin_lock_irq(&dev_priv->irq_lock);
-	if (dev_priv->rps.interrupts_enabled) {
-		pm_iir = fetch_and_zero(&dev_priv->rps.pm_iir);
-		client_boost = atomic_read(&dev_priv->rps.num_waiters);
+	if (rps->interrupts_enabled) {
+		pm_iir = fetch_and_zero(&rps->pm_iir);
+		client_boost = atomic_read(&rps->num_waiters);
 	}
 	spin_unlock_irq(&dev_priv->irq_lock);
 
@@ -1181,18 +1187,18 @@ static void gen6_pm_rps_work(struct work_struct *work)
 	if ((pm_iir & dev_priv->pm_rps_events) == 0 && !client_boost)
 		goto out;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	pm_iir |= vlv_wa_c0_ei(dev_priv, pm_iir);
 
-	adj = dev_priv->rps.last_adj;
-	new_delay = dev_priv->rps.cur_freq;
-	min = dev_priv->rps.min_freq_softlimit;
-	max = dev_priv->rps.max_freq_softlimit;
+	adj = rps->last_adj;
+	new_delay = rps->cur_freq;
+	min = rps->min_freq_softlimit;
+	max = rps->max_freq_softlimit;
 	if (client_boost)
-		max = dev_priv->rps.max_freq;
-	if (client_boost && new_delay < dev_priv->rps.boost_freq) {
-		new_delay = dev_priv->rps.boost_freq;
+		max = rps->max_freq;
+	if (client_boost && new_delay < rps->boost_freq) {
+		new_delay = rps->boost_freq;
 		adj = 0;
 	} else if (pm_iir & GEN6_PM_RP_UP_THRESHOLD) {
 		if (adj > 0)
@@ -1200,15 +1206,15 @@ static void gen6_pm_rps_work(struct work_struct *work)
 		else /* CHV needs even encode values */
 			adj = IS_CHERRYVIEW(dev_priv) ? 2 : 1;
 
-		if (new_delay >= dev_priv->rps.max_freq_softlimit)
+		if (new_delay >= rps->max_freq_softlimit)
 			adj = 0;
 	} else if (client_boost) {
 		adj = 0;
 	} else if (pm_iir & GEN6_PM_RP_DOWN_TIMEOUT) {
-		if (dev_priv->rps.cur_freq > dev_priv->rps.efficient_freq)
-			new_delay = dev_priv->rps.efficient_freq;
-		else if (dev_priv->rps.cur_freq > dev_priv->rps.min_freq_softlimit)
-			new_delay = dev_priv->rps.min_freq_softlimit;
+		if (rps->cur_freq > rps->efficient_freq)
+			new_delay = rps->efficient_freq;
+		else if (rps->cur_freq > rps->min_freq_softlimit)
+			new_delay = rps->min_freq_softlimit;
 		adj = 0;
 	} else if (pm_iir & GEN6_PM_RP_DOWN_THRESHOLD) {
 		if (adj < 0)
@@ -1216,13 +1222,13 @@ static void gen6_pm_rps_work(struct work_struct *work)
 		else /* CHV needs even encode values */
 			adj = IS_CHERRYVIEW(dev_priv) ? -2 : -1;
 
-		if (new_delay <= dev_priv->rps.min_freq_softlimit)
+		if (new_delay <= rps->min_freq_softlimit)
 			adj = 0;
 	} else { /* unknown event */
 		adj = 0;
 	}
 
-	dev_priv->rps.last_adj = adj;
+	rps->last_adj = adj;
 
 	/* sysfs frequency interfaces may have snuck in while servicing the
 	 * interrupt
@@ -1232,15 +1238,15 @@ static void gen6_pm_rps_work(struct work_struct *work)
 
 	if (intel_set_rps(dev_priv, new_delay)) {
 		DRM_DEBUG_DRIVER("Failed to set new GPU frequency\n");
-		dev_priv->rps.last_adj = 0;
+		rps->last_adj = 0;
 	}
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 out:
 	/* Make sure not to corrupt PMIMR state used by ringbuffer on GEN6 */
 	spin_lock_irq(&dev_priv->irq_lock);
-	if (dev_priv->rps.interrupts_enabled)
+	if (rps->interrupts_enabled)
 		gen6_unmask_pm_irq(dev_priv, dev_priv->pm_rps_events);
 	spin_unlock_irq(&dev_priv->irq_lock);
 }
@@ -1382,10 +1388,8 @@ gen8_cs_irq_handler(struct intel_engine_cs *engine, u32 iir, int test_shift)
 	bool tasklet = false;
 
 	if (iir & (GT_CONTEXT_SWITCH_INTERRUPT << test_shift)) {
-		if (port_count(&execlists->port[0])) {
-			__set_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
-			tasklet = true;
-		}
+		__set_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
+		tasklet = true;
 	}
 
 	if (iir & (GT_RENDER_USER_INTERRUPT << test_shift)) {
@@ -1723,12 +1727,14 @@ static void i9xx_pipe_crc_irq_handler(struct drm_i915_private *dev_priv,
  * the work queue. */
 static void gen6_rps_irq_handler(struct drm_i915_private *dev_priv, u32 pm_iir)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
 	if (pm_iir & dev_priv->pm_rps_events) {
 		spin_lock(&dev_priv->irq_lock);
 		gen6_mask_pm_irq(dev_priv, pm_iir & dev_priv->pm_rps_events);
-		if (dev_priv->rps.interrupts_enabled) {
-			dev_priv->rps.pm_iir |= pm_iir & dev_priv->pm_rps_events;
-			schedule_work(&dev_priv->rps.work);
+		if (rps->interrupts_enabled) {
+			rps->pm_iir |= pm_iir & dev_priv->pm_rps_events;
+			schedule_work(&rps->work);
 		}
 		spin_unlock(&dev_priv->irq_lock);
 	}
@@ -2254,18 +2260,14 @@ static void ivb_err_int_handler(struct drm_i915_private *dev_priv)
 static void cpt_serr_int_handler(struct drm_i915_private *dev_priv)
 {
 	u32 serr_int = I915_READ(SERR_INT);
+	enum pipe pipe;
 
 	if (serr_int & SERR_INT_POISON)
 		DRM_ERROR("PCH poison interrupt\n");
 
-	if (serr_int & SERR_INT_TRANS_A_FIFO_UNDERRUN)
-		intel_pch_fifo_underrun_irq_handler(dev_priv, PIPE_A);
-
-	if (serr_int & SERR_INT_TRANS_B_FIFO_UNDERRUN)
-		intel_pch_fifo_underrun_irq_handler(dev_priv, PIPE_B);
-
-	if (serr_int & SERR_INT_TRANS_C_FIFO_UNDERRUN)
-		intel_pch_fifo_underrun_irq_handler(dev_priv, PIPE_C);
+	for_each_pipe(dev_priv, pipe)
+		if (serr_int & SERR_INT_TRANS_FIFO_UNDERRUN(pipe))
+			intel_pch_fifo_underrun_irq_handler(dev_priv, pipe);
 
 	I915_WRITE(SERR_INT, serr_int);
 }
@@ -3163,10 +3165,17 @@ void gen8_irq_power_well_post_enable(struct drm_i915_private *dev_priv,
 	enum pipe pipe;
 
 	spin_lock_irq(&dev_priv->irq_lock);
+
+	if (!intel_irqs_enabled(dev_priv)) {
+		spin_unlock_irq(&dev_priv->irq_lock);
+		return;
+	}
+
 	for_each_pipe_masked(dev_priv, pipe, pipe_mask)
 		GEN8_IRQ_INIT_NDX(DE_PIPE, pipe,
 				  dev_priv->de_irq_mask[pipe],
 				  ~dev_priv->de_irq_mask[pipe] | extra_ier);
+
 	spin_unlock_irq(&dev_priv->irq_lock);
 }
 
@@ -3176,8 +3185,15 @@ void gen8_irq_power_well_pre_disable(struct drm_i915_private *dev_priv,
 	enum pipe pipe;
 
 	spin_lock_irq(&dev_priv->irq_lock);
+
+	if (!intel_irqs_enabled(dev_priv)) {
+		spin_unlock_irq(&dev_priv->irq_lock);
+		return;
+	}
+
 	for_each_pipe_masked(dev_priv, pipe, pipe_mask)
 		GEN8_IRQ_RESET_NDX(DE_PIPE, pipe);
+
 	spin_unlock_irq(&dev_priv->irq_lock);
 
 	/* make sure we're done processing display irqs */
@@ -3598,16 +3614,15 @@ static void gen8_de_irq_postinstall(struct drm_i915_private *dev_priv)
 	else if (IS_BROADWELL(dev_priv))
 		de_port_enables |= GEN8_PORT_DP_A_HOTPLUG;
 
-	dev_priv->de_irq_mask[PIPE_A] = ~de_pipe_masked;
-	dev_priv->de_irq_mask[PIPE_B] = ~de_pipe_masked;
-	dev_priv->de_irq_mask[PIPE_C] = ~de_pipe_masked;
+	for_each_pipe(dev_priv, pipe) {
+		dev_priv->de_irq_mask[pipe] = ~de_pipe_masked;
 
-	for_each_pipe(dev_priv, pipe)
 		if (intel_display_power_is_enabled(dev_priv,
 				POWER_DOMAIN_PIPE(pipe)))
 			GEN8_IRQ_INIT_NDX(DE_PIPE, pipe,
 					  dev_priv->de_irq_mask[pipe],
 					  de_pipe_enables);
+	}
 
 	GEN3_IRQ_INIT(GEN8_DE_PORT_, ~de_port_masked, de_port_enables);
 	GEN3_IRQ_INIT(GEN8_DE_MISC_, ~de_misc_masked, de_misc_masked);
@@ -4000,11 +4015,12 @@ static irqreturn_t i965_irq_handler(int irq, void *arg)
 void intel_irq_init(struct drm_i915_private *dev_priv)
 {
 	struct drm_device *dev = &dev_priv->drm;
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	int i;
 
 	intel_hpd_init_work(dev_priv);
 
-	INIT_WORK(&dev_priv->rps.work, gen6_pm_rps_work);
+	INIT_WORK(&rps->work, gen6_pm_rps_work);
 
 	INIT_WORK(&dev_priv->l3_parity.error_work, ivybridge_parity_work);
 	for (i = 0; i < MAX_L3_SLICES; ++i)
@@ -4020,7 +4036,7 @@ void intel_irq_init(struct drm_i915_private *dev_priv)
 	else
 		dev_priv->pm_rps_events = GEN6_PM_RPS_EVENTS;
 
-	dev_priv->rps.pm_intrmsk_mbz = 0;
+	rps->pm_intrmsk_mbz = 0;
 
 	/*
 	 * SNB,IVB,HSW can while VLV,CHV may hard hang on looping batchbuffer
@@ -4029,10 +4045,10 @@ void intel_irq_init(struct drm_i915_private *dev_priv)
 	 * TODO: verify if this can be reproduced on VLV,CHV.
 	 */
 	if (INTEL_GEN(dev_priv) <= 7)
-		dev_priv->rps.pm_intrmsk_mbz |= GEN6_PM_RP_UP_EI_EXPIRED;
+		rps->pm_intrmsk_mbz |= GEN6_PM_RP_UP_EI_EXPIRED;
 
 	if (INTEL_GEN(dev_priv) >= 8)
-		dev_priv->rps.pm_intrmsk_mbz |= GEN8_PMINTR_DISABLE_REDIRECT_TO_GUC;
+		rps->pm_intrmsk_mbz |= GEN8_PMINTR_DISABLE_REDIRECT_TO_GUC;
 
 	if (IS_GEN2(dev_priv)) {
 		/* Gen2 doesn't have a hardware frame counter */
@@ -4166,7 +4182,7 @@ int intel_irq_install(struct drm_i915_private *dev_priv)
 	 * interrupts as enabled _before_ actually enabling them to avoid
 	 * special cases in our ordering checks.
 	 */
-	dev_priv->pm.irqs_enabled = true;
+	dev_priv->runtime_pm.irqs_enabled = true;
 
 	return drm_irq_install(&dev_priv->drm, dev_priv->drm.pdev->irq);
 }
@@ -4182,7 +4198,7 @@ void intel_irq_uninstall(struct drm_i915_private *dev_priv)
 {
 	drm_irq_uninstall(&dev_priv->drm);
 	intel_hpd_cancel_work(dev_priv);
-	dev_priv->pm.irqs_enabled = false;
+	dev_priv->runtime_pm.irqs_enabled = false;
 }
 
 /**
@@ -4195,7 +4211,7 @@ void intel_irq_uninstall(struct drm_i915_private *dev_priv)
 void intel_runtime_pm_disable_interrupts(struct drm_i915_private *dev_priv)
 {
 	dev_priv->drm.driver->irq_uninstall(&dev_priv->drm);
-	dev_priv->pm.irqs_enabled = false;
+	dev_priv->runtime_pm.irqs_enabled = false;
 	synchronize_irq(dev_priv->drm.irq);
 }
 
@@ -4208,7 +4224,7 @@ void intel_runtime_pm_disable_interrupts(struct drm_i915_private *dev_priv)
  */
 void intel_runtime_pm_enable_interrupts(struct drm_i915_private *dev_priv)
 {
-	dev_priv->pm.irqs_enabled = true;
+	dev_priv->runtime_pm.irqs_enabled = true;
 	dev_priv->drm.driver->irq_preinstall(&dev_priv->drm);
 	dev_priv->drm.driver->irq_postinstall(&dev_priv->drm);
 }
diff --git a/drivers/gpu/drm/i915/i915_params.c b/drivers/gpu/drm/i915/i915_params.c
index 9dff323a83d3..b4faeb6aa2bd 100644
--- a/drivers/gpu/drm/i915/i915_params.c
+++ b/drivers/gpu/drm/i915/i915_params.c
@@ -146,9 +146,6 @@ i915_param_named(disable_display, bool, 0400,
 i915_param_named_unsafe(enable_cmd_parser, bool, 0400,
 	"Enable command parsing (true=enabled [default], false=disabled)");
 
-i915_param_named_unsafe(use_mmio_flip, int, 0600,
-	"use MMIO flips (-1=never, 0=driver discretion [default], 1=always)");
-
 i915_param_named(mmio_debug, int, 0600,
 	"Enable the MMIO debug code for the first N failures (default: off). "
 	"This may negatively affect performance.");
diff --git a/drivers/gpu/drm/i915/i915_params.h b/drivers/gpu/drm/i915/i915_params.h
index 4f3f8d650194..c7292268ed43 100644
--- a/drivers/gpu/drm/i915/i915_params.h
+++ b/drivers/gpu/drm/i915/i915_params.h
@@ -49,7 +49,6 @@
 	param(int, guc_log_level, -1) \
 	param(char *, guc_firmware_path, NULL) \
 	param(char *, huc_firmware_path, NULL) \
-	param(int, use_mmio_flip, 0) \
 	param(int, mmio_debug, 0) \
 	param(int, edp_vswing, 0) \
 	param(int, reset, 2) \
diff --git a/drivers/gpu/drm/i915/i915_pci.c b/drivers/gpu/drm/i915/i915_pci.c
index da60866b6628..bf467f30c99b 100644
--- a/drivers/gpu/drm/i915/i915_pci.c
+++ b/drivers/gpu/drm/i915/i915_pci.c
@@ -54,8 +54,14 @@
 	.color = { .degamma_lut_size = 512, .gamma_lut_size = 512 }
 #define CHV_COLORS \
 	.color = { .degamma_lut_size = 65, .gamma_lut_size = 257 }
+#define GLK_COLORS \
+	.color = { .degamma_lut_size = 0, .gamma_lut_size = 1024 }
 
 /* Keep in gen based order, and chronological order within a gen */
+
+#define GEN_DEFAULT_PAGE_SIZES \
+	.page_sizes = I915_GTT_PAGE_SIZE_4K
+
 #define GEN2_FEATURES \
 	.gen = 2, .num_pipes = 1, \
 	.has_overlay = 1, .overlay_needs_physical = 1, \
@@ -65,6 +71,7 @@
 	.ring_mask = RENDER_RING, \
 	.has_snoop = true, \
 	GEN_DEFAULT_PIPEOFFSETS, \
+	GEN_DEFAULT_PAGE_SIZES, \
 	CURSOR_OFFSETS
 
 static const struct intel_device_info intel_i830_info __initconst = {
@@ -98,6 +105,7 @@ static const struct intel_device_info intel_i865g_info __initconst = {
 	.ring_mask = RENDER_RING, \
 	.has_snoop = true, \
 	GEN_DEFAULT_PIPEOFFSETS, \
+	GEN_DEFAULT_PAGE_SIZES, \
 	CURSOR_OFFSETS
 
 static const struct intel_device_info intel_i915g_info __initconst = {
@@ -161,6 +169,7 @@ static const struct intel_device_info intel_pineview_info __initconst = {
 	.ring_mask = RENDER_RING, \
 	.has_snoop = true, \
 	GEN_DEFAULT_PIPEOFFSETS, \
+	GEN_DEFAULT_PAGE_SIZES, \
 	CURSOR_OFFSETS
 
 static const struct intel_device_info intel_i965g_info __initconst = {
@@ -203,6 +212,7 @@ static const struct intel_device_info intel_gm45_info __initconst = {
 	.ring_mask = RENDER_RING | BSD_RING, \
 	.has_snoop = true, \
 	GEN_DEFAULT_PIPEOFFSETS, \
+	GEN_DEFAULT_PAGE_SIZES, \
 	CURSOR_OFFSETS
 
 static const struct intel_device_info intel_ironlake_d_info __initconst = {
@@ -226,6 +236,7 @@ static const struct intel_device_info intel_ironlake_m_info __initconst = {
 	.has_rc6p = 1, \
 	.has_aliasing_ppgtt = 1, \
 	GEN_DEFAULT_PIPEOFFSETS, \
+	GEN_DEFAULT_PAGE_SIZES, \
 	CURSOR_OFFSETS
 
 #define SNB_D_PLATFORM \
@@ -269,6 +280,7 @@ static const struct intel_device_info intel_sandybridge_m_gt2_info __initconst =
 	.has_aliasing_ppgtt = 1, \
 	.has_full_ppgtt = 1, \
 	GEN_DEFAULT_PIPEOFFSETS, \
+	GEN_DEFAULT_PAGE_SIZES, \
 	IVB_CURSOR_OFFSETS
 
 #define IVB_D_PLATFORM \
@@ -325,11 +337,12 @@ static const struct intel_device_info intel_valleyview_info __initconst = {
 	.has_snoop = true,
 	.ring_mask = RENDER_RING | BSD_RING | BLT_RING,
 	.display_mmio_offset = VLV_DISPLAY_BASE,
+	GEN_DEFAULT_PAGE_SIZES,
 	GEN_DEFAULT_PIPEOFFSETS,
 	CURSOR_OFFSETS
 };
 
-#define HSW_FEATURES  \
+#define G75_FEATURES  \
 	GEN7_FEATURES, \
 	.ring_mask = RENDER_RING | BSD_RING | BLT_RING | VEBOX_RING, \
 	.has_ddi = 1, \
@@ -341,7 +354,7 @@ static const struct intel_device_info intel_valleyview_info __initconst = {
 	.has_runtime_pm = 1
 
 #define HSW_PLATFORM \
-	HSW_FEATURES, \
+	G75_FEATURES, \
 	.platform = INTEL_HASWELL, \
 	.has_l3_dpf = 1
 
@@ -360,16 +373,18 @@ static const struct intel_device_info intel_haswell_gt3_info __initconst = {
 	.gt = 3,
 };
 
-#define BDW_FEATURES \
-	HSW_FEATURES, \
+#define GEN8_FEATURES \
+	G75_FEATURES, \
 	BDW_COLORS, \
+	.page_sizes = I915_GTT_PAGE_SIZE_4K | \
+		      I915_GTT_PAGE_SIZE_2M, \
 	.has_logical_ring_contexts = 1, \
 	.has_full_48bit_ppgtt = 1, \
 	.has_64bit_reloc = 1, \
 	.has_reset_engine = 1
 
 #define BDW_PLATFORM \
-	BDW_FEATURES, \
+	GEN8_FEATURES, \
 	.gen = 8, \
 	.platform = INTEL_BROADWELL
 
@@ -415,19 +430,31 @@ static const struct intel_device_info intel_cherryview_info __initconst = {
 	.has_reset_engine = 1,
 	.has_snoop = true,
 	.display_mmio_offset = VLV_DISPLAY_BASE,
+	GEN_DEFAULT_PAGE_SIZES,
 	GEN_CHV_PIPEOFFSETS,
 	CURSOR_OFFSETS,
 	CHV_COLORS,
 };
 
-#define SKL_PLATFORM \
-	BDW_FEATURES, \
-	.gen = 9, \
-	.platform = INTEL_SKYLAKE, \
+#define GEN9_DEFAULT_PAGE_SIZES \
+	.page_sizes = I915_GTT_PAGE_SIZE_4K | \
+		      I915_GTT_PAGE_SIZE_64K | \
+		      I915_GTT_PAGE_SIZE_2M
+
+#define GEN9_FEATURES \
+	GEN8_FEATURES, \
+	GEN9_DEFAULT_PAGE_SIZES, \
+	.has_logical_ring_preemption = 1, \
 	.has_csr = 1, \
 	.has_guc = 1, \
+	.has_ipc = 1, \
 	.ddb_size = 896
 
+#define SKL_PLATFORM \
+	GEN9_FEATURES, \
+	.gen = 9, \
+	.platform = INTEL_SKYLAKE
+
 static const struct intel_device_info intel_skylake_gt1_info __initconst = {
 	SKL_PLATFORM,
 	.gt = 1,
@@ -463,6 +490,7 @@ static const struct intel_device_info intel_skylake_gt4_info __initconst = {
 	.has_ddi = 1, \
 	.has_fpga_dbg = 1, \
 	.has_fbc = 1, \
+	.has_psr = 1, \
 	.has_runtime_pm = 1, \
 	.has_pooled_eu = 0, \
 	.has_csr = 1, \
@@ -470,6 +498,7 @@ static const struct intel_device_info intel_skylake_gt4_info __initconst = {
 	.has_rc6 = 1, \
 	.has_dp_mst = 1, \
 	.has_logical_ring_contexts = 1, \
+	.has_logical_ring_preemption = 1, \
 	.has_guc = 1, \
 	.has_aliasing_ppgtt = 1, \
 	.has_full_ppgtt = 1, \
@@ -477,6 +506,7 @@ static const struct intel_device_info intel_skylake_gt4_info __initconst = {
 	.has_reset_engine = 1, \
 	.has_snoop = true, \
 	.has_ipc = 1, \
+	GEN9_DEFAULT_PAGE_SIZES, \
 	GEN_DEFAULT_PIPEOFFSETS, \
 	IVB_CURSOR_OFFSETS, \
 	BDW_COLORS
@@ -491,17 +521,13 @@ static const struct intel_device_info intel_geminilake_info __initconst = {
 	GEN9_LP_FEATURES,
 	.platform = INTEL_GEMINILAKE,
 	.ddb_size = 1024,
-	.color = { .degamma_lut_size = 0, .gamma_lut_size = 1024 }
+	GLK_COLORS,
 };
 
 #define KBL_PLATFORM \
-	BDW_FEATURES, \
+	GEN9_FEATURES, \
 	.gen = 9, \
-	.platform = INTEL_KABYLAKE, \
-	.has_csr = 1, \
-	.has_guc = 1, \
-	.has_ipc = 1, \
-	.ddb_size = 896
+	.platform = INTEL_KABYLAKE
 
 static const struct intel_device_info intel_kabylake_gt1_info __initconst = {
 	KBL_PLATFORM,
@@ -520,13 +546,9 @@ static const struct intel_device_info intel_kabylake_gt3_info __initconst = {
 };
 
 #define CFL_PLATFORM \
-	BDW_FEATURES, \
+	GEN9_FEATURES, \
 	.gen = 9, \
-	.platform = INTEL_COFFEELAKE, \
-	.has_csr = 1, \
-	.has_guc = 1, \
-	.has_ipc = 1, \
-	.ddb_size = 896
+	.platform = INTEL_COFFEELAKE
 
 static const struct intel_device_info intel_coffeelake_gt1_info __initconst = {
 	CFL_PLATFORM,
@@ -544,16 +566,17 @@ static const struct intel_device_info intel_coffeelake_gt3_info __initconst = {
 	.ring_mask = RENDER_RING | BSD_RING | BLT_RING | VEBOX_RING | BSD2_RING,
 };
 
+#define GEN10_FEATURES \
+	GEN9_FEATURES, \
+	.ddb_size = 1024, \
+	GLK_COLORS
+
 static const struct intel_device_info intel_cannonlake_gt2_info __initconst = {
-	BDW_FEATURES,
+	GEN10_FEATURES,
 	.is_alpha_support = 1,
 	.platform = INTEL_CANNONLAKE,
 	.gen = 10,
 	.gt = 2,
-	.ddb_size = 1024,
-	.has_csr = 1,
-	.has_ipc = 1,
-	.color = { .degamma_lut_size = 0, .gamma_lut_size = 1024 }
 };
 
 /*
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index ee0d4f14ac98..d2d0a83c09b6 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -2371,6 +2371,9 @@ enum i915_power_well_id {
 #define GEN9_GAMT_ECO_REG_RW_IA _MMIO(0x4ab0)
 #define   GAMT_ECO_ENABLE_IN_PLACE_DECOMPRESS	(1<<18)
 
+#define GEN8_GAMW_ECO_DEV_RW_IA _MMIO(0x4080)
+#define   GAMW_ECO_ENABLE_64K_IPS_FIELD 0xF
+
 #define GAMT_CHKN_BIT_REG	_MMIO(0x4ab8)
 #define   GAMT_CHKN_DISABLE_DYNAMIC_CREDIT_SHARING	(1<<28)
 #define   GAMT_CHKN_DISABLE_I2M_CYCLE_ON_WR_PORT	(1<<24)
@@ -3819,6 +3822,16 @@ enum {
 #define   PWM2_GATING_DIS		(1 << 14)
 #define   PWM1_GATING_DIS		(1 << 13)
 
+#define _CLKGATE_DIS_PSL_A		0x46520
+#define _CLKGATE_DIS_PSL_B		0x46524
+#define _CLKGATE_DIS_PSL_C		0x46528
+#define   DPF_GATING_DIS		(1 << 10)
+#define   DPF_RAM_GATING_DIS		(1 << 9)
+#define   DPFR_GATING_DIS		(1 << 8)
+
+#define CLKGATE_DIS_PSL(pipe) \
+	_MMIO_PIPE(pipe, _CLKGATE_DIS_PSL_A, _CLKGATE_DIS_PSL_B)
+
 /*
  * GEN10 clock gating regs
  */
@@ -5671,8 +5684,7 @@ enum {
 #define  CBR_PWM_CLOCK_MUX_SELECT	(1<<30)
 
 #define CBR4_VLV			_MMIO(VLV_DISPLAY_BASE + 0x70450)
-#define  CBR_DPLLBMD_PIPE_C		(1<<29)
-#define  CBR_DPLLBMD_PIPE_B		(1<<18)
+#define  CBR_DPLLBMD_PIPE(pipe)		(1<<(7+(pipe)*11)) /* pipes B and C */
 
 /* FIFO watermark sizes etc */
 #define G4X_FIFO_LINE_SIZE	64
@@ -6993,6 +7005,12 @@ enum {
 #define GEN9_CS_DEBUG_MODE1		_MMIO(0x20ec)
 #define GEN9_CTX_PREEMPT_REG		_MMIO(0x2248)
 #define GEN8_CS_CHICKEN1		_MMIO(0x2580)
+#define GEN9_PREEMPT_3D_OBJECT_LEVEL		(1<<0)
+#define GEN9_PREEMPT_GPGPU_LEVEL(hi, lo)	(((hi) << 2) | ((lo) << 1))
+#define GEN9_PREEMPT_GPGPU_MID_THREAD_LEVEL	GEN9_PREEMPT_GPGPU_LEVEL(0, 0)
+#define GEN9_PREEMPT_GPGPU_THREAD_GROUP_LEVEL	GEN9_PREEMPT_GPGPU_LEVEL(0, 1)
+#define GEN9_PREEMPT_GPGPU_COMMAND_LEVEL	GEN9_PREEMPT_GPGPU_LEVEL(1, 0)
+#define GEN9_PREEMPT_GPGPU_LEVEL_MASK		GEN9_PREEMPT_GPGPU_LEVEL(1, 1)
 
 /* GEN7 chicken */
 #define GEN7_COMMON_SLICE_CHICKEN1		_MMIO(0x7010)
@@ -7164,9 +7182,6 @@ enum {
 
 #define SERR_INT			_MMIO(0xc4040)
 #define  SERR_INT_POISON		(1<<31)
-#define  SERR_INT_TRANS_C_FIFO_UNDERRUN	(1<<6)
-#define  SERR_INT_TRANS_B_FIFO_UNDERRUN	(1<<3)
-#define  SERR_INT_TRANS_A_FIFO_UNDERRUN	(1<<0)
 #define  SERR_INT_TRANS_FIFO_UNDERRUN(pipe)	(1<<((pipe)*3))
 
 /* digital port hotplug */
diff --git a/drivers/gpu/drm/i915/i915_suspend.c b/drivers/gpu/drm/i915/i915_suspend.c
index 5c86925a0294..8f3aa4dc0c98 100644
--- a/drivers/gpu/drm/i915/i915_suspend.c
+++ b/drivers/gpu/drm/i915/i915_suspend.c
@@ -108,8 +108,6 @@ int i915_restore_state(struct drm_i915_private *dev_priv)
 
 	mutex_lock(&dev_priv->drm.struct_mutex);
 
-	i915_gem_restore_fences(dev_priv);
-
 	if (IS_GEN4(dev_priv))
 		pci_write_config_word(pdev, GCDGMBUS,
 				      dev_priv->regfile.saveGCDGMBUS);
diff --git a/drivers/gpu/drm/i915/i915_sysfs.c b/drivers/gpu/drm/i915/i915_sysfs.c
index d61c8727f756..791759f632e1 100644
--- a/drivers/gpu/drm/i915/i915_sysfs.c
+++ b/drivers/gpu/drm/i915/i915_sysfs.c
@@ -49,7 +49,7 @@ static u32 calc_residency(struct drm_i915_private *dev_priv,
 static ssize_t
 show_rc6_mask(struct device *kdev, struct device_attribute *attr, char *buf)
 {
-	return snprintf(buf, PAGE_SIZE, "%x\n", intel_enable_rc6());
+	return snprintf(buf, PAGE_SIZE, "%x\n", intel_rc6_enabled());
 }
 
 static ssize_t
@@ -246,7 +246,7 @@ static ssize_t gt_act_freq_mhz_show(struct device *kdev,
 
 	intel_runtime_pm_get(dev_priv);
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) {
 		u32 freq;
 		freq = vlv_punit_read(dev_priv, PUNIT_REG_GPU_FREQ_STS);
@@ -261,7 +261,7 @@ static ssize_t gt_act_freq_mhz_show(struct device *kdev,
 			ret = (rpstat & GEN6_CAGF_MASK) >> GEN6_CAGF_SHIFT;
 		ret = intel_gpu_freq(dev_priv, ret);
 	}
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	intel_runtime_pm_put(dev_priv);
 
@@ -275,7 +275,7 @@ static ssize_t gt_cur_freq_mhz_show(struct device *kdev,
 
 	return snprintf(buf, PAGE_SIZE, "%d\n",
 			intel_gpu_freq(dev_priv,
-				       dev_priv->rps.cur_freq));
+				       dev_priv->gt_pm.rps.cur_freq));
 }
 
 static ssize_t gt_boost_freq_mhz_show(struct device *kdev, struct device_attribute *attr, char *buf)
@@ -284,7 +284,7 @@ static ssize_t gt_boost_freq_mhz_show(struct device *kdev, struct device_attribu
 
 	return snprintf(buf, PAGE_SIZE, "%d\n",
 			intel_gpu_freq(dev_priv,
-				       dev_priv->rps.boost_freq));
+				       dev_priv->gt_pm.rps.boost_freq));
 }
 
 static ssize_t gt_boost_freq_mhz_store(struct device *kdev,
@@ -292,6 +292,7 @@ static ssize_t gt_boost_freq_mhz_store(struct device *kdev,
 				       const char *buf, size_t count)
 {
 	struct drm_i915_private *dev_priv = kdev_minor_to_i915(kdev);
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 val;
 	ssize_t ret;
 
@@ -301,12 +302,12 @@ static ssize_t gt_boost_freq_mhz_store(struct device *kdev,
 
 	/* Validate against (static) hardware limits */
 	val = intel_freq_opcode(dev_priv, val);
-	if (val < dev_priv->rps.min_freq || val > dev_priv->rps.max_freq)
+	if (val < rps->min_freq || val > rps->max_freq)
 		return -EINVAL;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
-	dev_priv->rps.boost_freq = val;
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
+	rps->boost_freq = val;
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	return count;
 }
@@ -318,7 +319,7 @@ static ssize_t vlv_rpe_freq_mhz_show(struct device *kdev,
 
 	return snprintf(buf, PAGE_SIZE, "%d\n",
 			intel_gpu_freq(dev_priv,
-				       dev_priv->rps.efficient_freq));
+				       dev_priv->gt_pm.rps.efficient_freq));
 }
 
 static ssize_t gt_max_freq_mhz_show(struct device *kdev, struct device_attribute *attr, char *buf)
@@ -327,7 +328,7 @@ static ssize_t gt_max_freq_mhz_show(struct device *kdev, struct device_attribute
 
 	return snprintf(buf, PAGE_SIZE, "%d\n",
 			intel_gpu_freq(dev_priv,
-				       dev_priv->rps.max_freq_softlimit));
+				       dev_priv->gt_pm.rps.max_freq_softlimit));
 }
 
 static ssize_t gt_max_freq_mhz_store(struct device *kdev,
@@ -335,6 +336,7 @@ static ssize_t gt_max_freq_mhz_store(struct device *kdev,
 				     const char *buf, size_t count)
 {
 	struct drm_i915_private *dev_priv = kdev_minor_to_i915(kdev);
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 val;
 	ssize_t ret;
 
@@ -344,34 +346,34 @@ static ssize_t gt_max_freq_mhz_store(struct device *kdev,
 
 	intel_runtime_pm_get(dev_priv);
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	val = intel_freq_opcode(dev_priv, val);
 
-	if (val < dev_priv->rps.min_freq ||
-	    val > dev_priv->rps.max_freq ||
-	    val < dev_priv->rps.min_freq_softlimit) {
-		mutex_unlock(&dev_priv->rps.hw_lock);
+	if (val < rps->min_freq ||
+	    val > rps->max_freq ||
+	    val < rps->min_freq_softlimit) {
+		mutex_unlock(&dev_priv->pcu_lock);
 		intel_runtime_pm_put(dev_priv);
 		return -EINVAL;
 	}
 
-	if (val > dev_priv->rps.rp0_freq)
+	if (val > rps->rp0_freq)
 		DRM_DEBUG("User requested overclocking to %d\n",
 			  intel_gpu_freq(dev_priv, val));
 
-	dev_priv->rps.max_freq_softlimit = val;
+	rps->max_freq_softlimit = val;
 
-	val = clamp_t(int, dev_priv->rps.cur_freq,
-		      dev_priv->rps.min_freq_softlimit,
-		      dev_priv->rps.max_freq_softlimit);
+	val = clamp_t(int, rps->cur_freq,
+		      rps->min_freq_softlimit,
+		      rps->max_freq_softlimit);
 
 	/* We still need *_set_rps to process the new max_delay and
 	 * update the interrupt limits and PMINTRMSK even though
 	 * frequency request may be unchanged. */
 	ret = intel_set_rps(dev_priv, val);
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	intel_runtime_pm_put(dev_priv);
 
@@ -384,7 +386,7 @@ static ssize_t gt_min_freq_mhz_show(struct device *kdev, struct device_attribute
 
 	return snprintf(buf, PAGE_SIZE, "%d\n",
 			intel_gpu_freq(dev_priv,
-				       dev_priv->rps.min_freq_softlimit));
+				       dev_priv->gt_pm.rps.min_freq_softlimit));
 }
 
 static ssize_t gt_min_freq_mhz_store(struct device *kdev,
@@ -392,6 +394,7 @@ static ssize_t gt_min_freq_mhz_store(struct device *kdev,
 				     const char *buf, size_t count)
 {
 	struct drm_i915_private *dev_priv = kdev_minor_to_i915(kdev);
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 val;
 	ssize_t ret;
 
@@ -401,30 +404,30 @@ static ssize_t gt_min_freq_mhz_store(struct device *kdev,
 
 	intel_runtime_pm_get(dev_priv);
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	val = intel_freq_opcode(dev_priv, val);
 
-	if (val < dev_priv->rps.min_freq ||
-	    val > dev_priv->rps.max_freq ||
-	    val > dev_priv->rps.max_freq_softlimit) {
-		mutex_unlock(&dev_priv->rps.hw_lock);
+	if (val < rps->min_freq ||
+	    val > rps->max_freq ||
+	    val > rps->max_freq_softlimit) {
+		mutex_unlock(&dev_priv->pcu_lock);
 		intel_runtime_pm_put(dev_priv);
 		return -EINVAL;
 	}
 
-	dev_priv->rps.min_freq_softlimit = val;
+	rps->min_freq_softlimit = val;
 
-	val = clamp_t(int, dev_priv->rps.cur_freq,
-		      dev_priv->rps.min_freq_softlimit,
-		      dev_priv->rps.max_freq_softlimit);
+	val = clamp_t(int, rps->cur_freq,
+		      rps->min_freq_softlimit,
+		      rps->max_freq_softlimit);
 
 	/* We still need *_set_rps to process the new min_delay and
 	 * update the interrupt limits and PMINTRMSK even though
 	 * frequency request may be unchanged. */
 	ret = intel_set_rps(dev_priv, val);
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	intel_runtime_pm_put(dev_priv);
 
@@ -448,14 +451,15 @@ static DEVICE_ATTR(gt_RPn_freq_mhz, S_IRUGO, gt_rp_mhz_show, NULL);
 static ssize_t gt_rp_mhz_show(struct device *kdev, struct device_attribute *attr, char *buf)
 {
 	struct drm_i915_private *dev_priv = kdev_minor_to_i915(kdev);
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 val;
 
 	if (attr == &dev_attr_gt_RP0_freq_mhz)
-		val = intel_gpu_freq(dev_priv, dev_priv->rps.rp0_freq);
+		val = intel_gpu_freq(dev_priv, rps->rp0_freq);
 	else if (attr == &dev_attr_gt_RP1_freq_mhz)
-		val = intel_gpu_freq(dev_priv, dev_priv->rps.rp1_freq);
+		val = intel_gpu_freq(dev_priv, rps->rp1_freq);
 	else if (attr == &dev_attr_gt_RPn_freq_mhz)
-		val = intel_gpu_freq(dev_priv, dev_priv->rps.min_freq);
+		val = intel_gpu_freq(dev_priv, rps->min_freq);
 	else
 		BUG();
 
diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h
index 92f4c5bb7aa7..9cab91ddeb79 100644
--- a/drivers/gpu/drm/i915/i915_trace.h
+++ b/drivers/gpu/drm/i915/i915_trace.h
@@ -345,7 +345,7 @@ TRACE_EVENT(i915_gem_object_create,
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
-			     __field(u32, size)
+			     __field(u64, size)
 			     ),
 
 	    TP_fast_assign(
@@ -353,7 +353,7 @@ TRACE_EVENT(i915_gem_object_create,
 			   __entry->size = obj->base.size;
 			   ),
 
-	    TP_printk("obj=%p, size=%u", __entry->obj, __entry->size)
+	    TP_printk("obj=%p, size=0x%llx", __entry->obj, __entry->size)
 );
 
 TRACE_EVENT(i915_gem_shrink,
@@ -384,7 +384,7 @@ TRACE_EVENT(i915_vma_bind,
 			     __field(struct drm_i915_gem_object *, obj)
 			     __field(struct i915_address_space *, vm)
 			     __field(u64, offset)
-			     __field(u32, size)
+			     __field(u64, size)
 			     __field(unsigned, flags)
 			     ),
 
@@ -396,7 +396,7 @@ TRACE_EVENT(i915_vma_bind,
 			   __entry->flags = flags;
 			   ),
 
-	    TP_printk("obj=%p, offset=%016llx size=%x%s vm=%p",
+	    TP_printk("obj=%p, offset=0x%016llx size=0x%llx%s vm=%p",
 		      __entry->obj, __entry->offset, __entry->size,
 		      __entry->flags & PIN_MAPPABLE ? ", mappable" : "",
 		      __entry->vm)
@@ -410,7 +410,7 @@ TRACE_EVENT(i915_vma_unbind,
 			     __field(struct drm_i915_gem_object *, obj)
 			     __field(struct i915_address_space *, vm)
 			     __field(u64, offset)
-			     __field(u32, size)
+			     __field(u64, size)
 			     ),
 
 	    TP_fast_assign(
@@ -420,18 +420,18 @@ TRACE_EVENT(i915_vma_unbind,
 			   __entry->size = vma->node.size;
 			   ),
 
-	    TP_printk("obj=%p, offset=%016llx size=%x vm=%p",
+	    TP_printk("obj=%p, offset=0x%016llx size=0x%llx vm=%p",
 		      __entry->obj, __entry->offset, __entry->size, __entry->vm)
 );
 
 TRACE_EVENT(i915_gem_object_pwrite,
-	    TP_PROTO(struct drm_i915_gem_object *obj, u32 offset, u32 len),
+	    TP_PROTO(struct drm_i915_gem_object *obj, u64 offset, u64 len),
 	    TP_ARGS(obj, offset, len),
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
-			     __field(u32, offset)
-			     __field(u32, len)
+			     __field(u64, offset)
+			     __field(u64, len)
 			     ),
 
 	    TP_fast_assign(
@@ -440,18 +440,18 @@ TRACE_EVENT(i915_gem_object_pwrite,
 			   __entry->len = len;
 			   ),
 
-	    TP_printk("obj=%p, offset=%u, len=%u",
+	    TP_printk("obj=%p, offset=0x%llx, len=0x%llx",
 		      __entry->obj, __entry->offset, __entry->len)
 );
 
 TRACE_EVENT(i915_gem_object_pread,
-	    TP_PROTO(struct drm_i915_gem_object *obj, u32 offset, u32 len),
+	    TP_PROTO(struct drm_i915_gem_object *obj, u64 offset, u64 len),
 	    TP_ARGS(obj, offset, len),
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
-			     __field(u32, offset)
-			     __field(u32, len)
+			     __field(u64, offset)
+			     __field(u64, len)
 			     ),
 
 	    TP_fast_assign(
@@ -460,17 +460,17 @@ TRACE_EVENT(i915_gem_object_pread,
 			   __entry->len = len;
 			   ),
 
-	    TP_printk("obj=%p, offset=%u, len=%u",
+	    TP_printk("obj=%p, offset=0x%llx, len=0x%llx",
 		      __entry->obj, __entry->offset, __entry->len)
 );
 
 TRACE_EVENT(i915_gem_object_fault,
-	    TP_PROTO(struct drm_i915_gem_object *obj, u32 index, bool gtt, bool write),
+	    TP_PROTO(struct drm_i915_gem_object *obj, u64 index, bool gtt, bool write),
 	    TP_ARGS(obj, index, gtt, write),
 
 	    TP_STRUCT__entry(
 			     __field(struct drm_i915_gem_object *, obj)
-			     __field(u32, index)
+			     __field(u64, index)
 			     __field(bool, gtt)
 			     __field(bool, write)
 			     ),
@@ -482,7 +482,7 @@ TRACE_EVENT(i915_gem_object_fault,
 			   __entry->write = write;
 			   ),
 
-	    TP_printk("obj=%p, %s index=%u %s",
+	    TP_printk("obj=%p, %s index=%llu %s",
 		      __entry->obj,
 		      __entry->gtt ? "GTT" : "CPU",
 		      __entry->index,
@@ -515,14 +515,14 @@ DEFINE_EVENT(i915_gem_object, i915_gem_object_destroy,
 );
 
 TRACE_EVENT(i915_gem_evict,
-	    TP_PROTO(struct i915_address_space *vm, u32 size, u32 align, unsigned int flags),
+	    TP_PROTO(struct i915_address_space *vm, u64 size, u64 align, unsigned int flags),
 	    TP_ARGS(vm, size, align, flags),
 
 	    TP_STRUCT__entry(
 			     __field(u32, dev)
 			     __field(struct i915_address_space *, vm)
-			     __field(u32, size)
-			     __field(u32, align)
+			     __field(u64, size)
+			     __field(u64, align)
 			     __field(unsigned int, flags)
 			    ),
 
@@ -534,43 +534,11 @@ TRACE_EVENT(i915_gem_evict,
 			   __entry->flags = flags;
 			  ),
 
-	    TP_printk("dev=%d, vm=%p, size=%d, align=%d %s",
+	    TP_printk("dev=%d, vm=%p, size=0x%llx, align=0x%llx %s",
 		      __entry->dev, __entry->vm, __entry->size, __entry->align,
 		      __entry->flags & PIN_MAPPABLE ? ", mappable" : "")
 );
 
-TRACE_EVENT(i915_gem_evict_everything,
-	    TP_PROTO(struct drm_device *dev),
-	    TP_ARGS(dev),
-
-	    TP_STRUCT__entry(
-			     __field(u32, dev)
-			    ),
-
-	    TP_fast_assign(
-			   __entry->dev = dev->primary->index;
-			  ),
-
-	    TP_printk("dev=%d", __entry->dev)
-);
-
-TRACE_EVENT(i915_gem_evict_vm,
-	    TP_PROTO(struct i915_address_space *vm),
-	    TP_ARGS(vm),
-
-	    TP_STRUCT__entry(
-			     __field(u32, dev)
-			     __field(struct i915_address_space *, vm)
-			    ),
-
-	    TP_fast_assign(
-			   __entry->dev = vm->i915->drm.primary->index;
-			   __entry->vm = vm;
-			  ),
-
-	    TP_printk("dev=%d, vm=%p", __entry->dev, __entry->vm)
-);
-
 TRACE_EVENT(i915_gem_evict_node,
 	    TP_PROTO(struct i915_address_space *vm, struct drm_mm_node *node, unsigned int flags),
 	    TP_ARGS(vm, node, flags),
@@ -593,12 +561,29 @@ TRACE_EVENT(i915_gem_evict_node,
 			   __entry->flags = flags;
 			  ),
 
-	    TP_printk("dev=%d, vm=%p, start=%llx size=%llx, color=%lx, flags=%x",
+	    TP_printk("dev=%d, vm=%p, start=0x%llx size=0x%llx, color=0x%lx, flags=%x",
 		      __entry->dev, __entry->vm,
 		      __entry->start, __entry->size,
 		      __entry->color, __entry->flags)
 );
 
+TRACE_EVENT(i915_gem_evict_vm,
+	    TP_PROTO(struct i915_address_space *vm),
+	    TP_ARGS(vm),
+
+	    TP_STRUCT__entry(
+			     __field(u32, dev)
+			     __field(struct i915_address_space *, vm)
+			    ),
+
+	    TP_fast_assign(
+			   __entry->dev = vm->i915->drm.primary->index;
+			   __entry->vm = vm;
+			  ),
+
+	    TP_printk("dev=%d, vm=%p", __entry->dev, __entry->vm)
+);
+
 TRACE_EVENT(i915_gem_ring_sync_to,
 	    TP_PROTO(struct drm_i915_gem_request *to,
 		     struct drm_i915_gem_request *from),
@@ -649,29 +634,6 @@ TRACE_EVENT(i915_gem_request_queue,
 		      __entry->flags)
 );
 
-TRACE_EVENT(i915_gem_ring_flush,
-	    TP_PROTO(struct drm_i915_gem_request *req, u32 invalidate, u32 flush),
-	    TP_ARGS(req, invalidate, flush),
-
-	    TP_STRUCT__entry(
-			     __field(u32, dev)
-			     __field(u32, ring)
-			     __field(u32, invalidate)
-			     __field(u32, flush)
-			     ),
-
-	    TP_fast_assign(
-			   __entry->dev = req->i915->drm.primary->index;
-			   __entry->ring = req->engine->id;
-			   __entry->invalidate = invalidate;
-			   __entry->flush = flush;
-			   ),
-
-	    TP_printk("dev=%u, ring=%x, invalidate=%04x, flush=%04x",
-		      __entry->dev, __entry->ring,
-		      __entry->invalidate, __entry->flush)
-);
-
 DECLARE_EVENT_CLASS(i915_gem_request,
 	    TP_PROTO(struct drm_i915_gem_request *req),
 	    TP_ARGS(req),
diff --git a/drivers/gpu/drm/i915/i915_utils.h b/drivers/gpu/drm/i915/i915_utils.h
index 12fc250b47b9..af3d7cc53fa1 100644
--- a/drivers/gpu/drm/i915/i915_utils.h
+++ b/drivers/gpu/drm/i915/i915_utils.h
@@ -99,6 +99,11 @@
 	__T;								\
 })
 
+static inline u64 ptr_to_u64(const void *ptr)
+{
+	return (uintptr_t)ptr;
+}
+
 #define u64_to_ptr(T, x) ({						\
 	typecheck(u64, x);						\
 	(T *)(uintptr_t)(x);						\
@@ -119,4 +124,17 @@ static inline void __list_del_many(struct list_head *head,
 	WRITE_ONCE(head->next, first);
 }
 
+/*
+ * Wait until the work is finally complete, even if it tries to postpone
+ * by requeueing itself. Note, that if the worker never cancels itself,
+ * we will spin forever.
+ */
+static inline void drain_delayed_work(struct delayed_work *dw)
+{
+	do {
+		while (flush_delayed_work(dw))
+			;
+	} while (delayed_work_pending(dw));
+}
+
 #endif /* !__I915_UTILS_H */
diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
index 02d1a5eacb00..4dce2e0197d9 100644
--- a/drivers/gpu/drm/i915/i915_vma.c
+++ b/drivers/gpu/drm/i915/i915_vma.c
@@ -266,6 +266,8 @@ int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,
 	if (bind_flags == 0)
 		return 0;
 
+	GEM_BUG_ON(!vma->pages);
+
 	trace_i915_vma_bind(vma, bind_flags);
 	ret = vma->vm->bind_vma(vma, cache_level, bind_flags);
 	if (ret)
@@ -278,13 +280,16 @@ int i915_vma_bind(struct i915_vma *vma, enum i915_cache_level cache_level,
 void __iomem *i915_vma_pin_iomap(struct i915_vma *vma)
 {
 	void __iomem *ptr;
+	int err;
 
 	/* Access through the GTT requires the device to be awake. */
 	assert_rpm_wakelock_held(vma->vm->i915);
 
 	lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
-	if (WARN_ON(!i915_vma_is_map_and_fenceable(vma)))
-		return IO_ERR_PTR(-ENODEV);
+	if (WARN_ON(!i915_vma_is_map_and_fenceable(vma))) {
+		err = -ENODEV;
+		goto err;
+	}
 
 	GEM_BUG_ON(!i915_vma_is_ggtt(vma));
 	GEM_BUG_ON((vma->flags & I915_VMA_GLOBAL_BIND) == 0);
@@ -294,14 +299,36 @@ void __iomem *i915_vma_pin_iomap(struct i915_vma *vma)
 		ptr = io_mapping_map_wc(&i915_vm_to_ggtt(vma->vm)->mappable,
 					vma->node.start,
 					vma->node.size);
-		if (ptr == NULL)
-			return IO_ERR_PTR(-ENOMEM);
+		if (ptr == NULL) {
+			err = -ENOMEM;
+			goto err;
+		}
 
 		vma->iomap = ptr;
 	}
 
 	__i915_vma_pin(vma);
+
+	err = i915_vma_pin_fence(vma);
+	if (err)
+		goto err_unpin;
+
 	return ptr;
+
+err_unpin:
+	__i915_vma_unpin(vma);
+err:
+	return IO_ERR_PTR(err);
+}
+
+void i915_vma_unpin_iomap(struct i915_vma *vma)
+{
+	lockdep_assert_held(&vma->obj->base.dev->struct_mutex);
+
+	GEM_BUG_ON(vma->iomap == NULL);
+
+	i915_vma_unpin_fence(vma);
+	i915_vma_unpin(vma);
 }
 
 void i915_vma_unpin_and_release(struct i915_vma **p_vma)
@@ -471,25 +498,64 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
 	if (ret)
 		return ret;
 
+	GEM_BUG_ON(vma->pages);
+
+	ret = vma->vm->set_pages(vma);
+	if (ret)
+		goto err_unpin;
+
 	if (flags & PIN_OFFSET_FIXED) {
 		u64 offset = flags & PIN_OFFSET_MASK;
 		if (!IS_ALIGNED(offset, alignment) ||
 		    range_overflows(offset, size, end)) {
 			ret = -EINVAL;
-			goto err_unpin;
+			goto err_clear;
 		}
 
 		ret = i915_gem_gtt_reserve(vma->vm, &vma->node,
 					   size, offset, obj->cache_level,
 					   flags);
 		if (ret)
-			goto err_unpin;
+			goto err_clear;
 	} else {
+		/*
+		 * We only support huge gtt pages through the 48b PPGTT,
+		 * however we also don't want to force any alignment for
+		 * objects which need to be tightly packed into the low 32bits.
+		 *
+		 * Note that we assume that GGTT are limited to 4GiB for the
+		 * forseeable future. See also i915_ggtt_offset().
+		 */
+		if (upper_32_bits(end - 1) &&
+		    vma->page_sizes.sg > I915_GTT_PAGE_SIZE) {
+			/*
+			 * We can't mix 64K and 4K PTEs in the same page-table
+			 * (2M block), and so to avoid the ugliness and
+			 * complexity of coloring we opt for just aligning 64K
+			 * objects to 2M.
+			 */
+			u64 page_alignment =
+				rounddown_pow_of_two(vma->page_sizes.sg |
+						     I915_GTT_PAGE_SIZE_2M);
+
+			/*
+			 * Check we don't expand for the limited Global GTT
+			 * (mappable aperture is even more precious!). This
+			 * also checks that we exclude the aliasing-ppgtt.
+			 */
+			GEM_BUG_ON(i915_vma_is_ggtt(vma));
+
+			alignment = max(alignment, page_alignment);
+
+			if (vma->page_sizes.sg & I915_GTT_PAGE_SIZE_64K)
+				size = round_up(size, I915_GTT_PAGE_SIZE_2M);
+		}
+
 		ret = i915_gem_gtt_insert(vma->vm, &vma->node,
 					  size, alignment, obj->cache_level,
 					  start, end, flags);
 		if (ret)
-			goto err_unpin;
+			goto err_clear;
 
 		GEM_BUG_ON(vma->node.start < start);
 		GEM_BUG_ON(vma->node.start + vma->node.size > end);
@@ -504,6 +570,8 @@ i915_vma_insert(struct i915_vma *vma, u64 size, u64 alignment, u64 flags)
 
 	return 0;
 
+err_clear:
+	vma->vm->clear_pages(vma);
 err_unpin:
 	i915_gem_object_unpin_pages(obj);
 	return ret;
@@ -517,6 +585,8 @@ i915_vma_remove(struct i915_vma *vma)
 	GEM_BUG_ON(!drm_mm_node_allocated(&vma->node));
 	GEM_BUG_ON(vma->flags & (I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND));
 
+	vma->vm->clear_pages(vma);
+
 	drm_mm_remove_node(&vma->node);
 	list_move_tail(&vma->vm_link, &vma->vm->unbound_list);
 
@@ -569,8 +639,8 @@ int __i915_vma_do_pin(struct i915_vma *vma,
 
 err_remove:
 	if ((bound & I915_VMA_BIND_MASK) == 0) {
-		GEM_BUG_ON(vma->pages);
 		i915_vma_remove(vma);
+		GEM_BUG_ON(vma->pages);
 	}
 err_unpin:
 	__i915_vma_unpin(vma);
@@ -620,6 +690,30 @@ static void __i915_vma_iounmap(struct i915_vma *vma)
 	vma->iomap = NULL;
 }
 
+void i915_vma_revoke_mmap(struct i915_vma *vma)
+{
+	struct drm_vma_offset_node *node = &vma->obj->base.vma_node;
+	u64 vma_offset;
+
+	lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);
+
+	if (!i915_vma_has_userfault(vma))
+		return;
+
+	GEM_BUG_ON(!i915_vma_is_map_and_fenceable(vma));
+	GEM_BUG_ON(!vma->obj->userfault_count);
+
+	vma_offset = vma->ggtt_view.partial.offset << PAGE_SHIFT;
+	unmap_mapping_range(vma->vm->i915->drm.anon_inode->i_mapping,
+			    drm_vma_node_offset_addr(node) + vma_offset,
+			    vma->size,
+			    1);
+
+	i915_vma_unset_userfault(vma);
+	if (!--vma->obj->userfault_count)
+		list_del(&vma->obj->userfault_link);
+}
+
 int i915_vma_unbind(struct i915_vma *vma)
 {
 	struct drm_i915_gem_object *obj = vma->obj;
@@ -683,11 +777,13 @@ int i915_vma_unbind(struct i915_vma *vma)
 			return ret;
 
 		/* Force a pagefault for domain tracking on next user access */
-		i915_gem_release_mmap(obj);
+		i915_vma_revoke_mmap(vma);
 
 		__i915_vma_iounmap(vma);
 		vma->flags &= ~I915_VMA_CAN_FENCE;
 	}
+	GEM_BUG_ON(vma->fence);
+	GEM_BUG_ON(i915_vma_has_userfault(vma));
 
 	if (likely(!vma->vm->closed)) {
 		trace_i915_vma_unbind(vma);
@@ -695,13 +791,6 @@ int i915_vma_unbind(struct i915_vma *vma)
 	}
 	vma->flags &= ~(I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND);
 
-	if (vma->pages != obj->mm.pages) {
-		GEM_BUG_ON(!vma->pages);
-		sg_free_table(vma->pages);
-		kfree(vma->pages);
-	}
-	vma->pages = NULL;
-
 	i915_vma_remove(vma);
 
 destroy:
diff --git a/drivers/gpu/drm/i915/i915_vma.h b/drivers/gpu/drm/i915/i915_vma.h
index e811067c7724..1e2bc9b3c3ac 100644
--- a/drivers/gpu/drm/i915/i915_vma.h
+++ b/drivers/gpu/drm/i915/i915_vma.h
@@ -55,6 +55,7 @@ struct i915_vma {
 	void __iomem *iomap;
 	u64 size;
 	u64 display_alignment;
+	struct i915_page_sizes page_sizes;
 
 	u32 fence_size;
 	u32 fence_alignment;
@@ -65,7 +66,7 @@ struct i915_vma {
 	 * that exist in the ctx->handle_vmas LUT for this vma.
 	 */
 	unsigned int open_count;
-	unsigned int flags;
+	unsigned long flags;
 	/**
 	 * How many users have pinned this object in GTT space. The following
 	 * users can each hold at most one reference: pwrite/pread, execbuffer
@@ -87,6 +88,8 @@ struct i915_vma {
 #define I915_VMA_GGTT		BIT(8)
 #define I915_VMA_CAN_FENCE	BIT(9)
 #define I915_VMA_CLOSED		BIT(10)
+#define I915_VMA_USERFAULT_BIT	11
+#define I915_VMA_USERFAULT	BIT(I915_VMA_USERFAULT_BIT)
 
 	unsigned int active;
 	struct i915_gem_active last_read[I915_NUM_ENGINES];
@@ -145,6 +148,22 @@ static inline bool i915_vma_is_closed(const struct i915_vma *vma)
 	return vma->flags & I915_VMA_CLOSED;
 }
 
+static inline bool i915_vma_set_userfault(struct i915_vma *vma)
+{
+	GEM_BUG_ON(!i915_vma_is_map_and_fenceable(vma));
+	return __test_and_set_bit(I915_VMA_USERFAULT_BIT, &vma->flags);
+}
+
+static inline void i915_vma_unset_userfault(struct i915_vma *vma)
+{
+	return __clear_bit(I915_VMA_USERFAULT_BIT, &vma->flags);
+}
+
+static inline bool i915_vma_has_userfault(const struct i915_vma *vma)
+{
+	return test_bit(I915_VMA_USERFAULT_BIT, &vma->flags);
+}
+
 static inline unsigned int i915_vma_get_active(const struct i915_vma *vma)
 {
 	return vma->active;
@@ -243,6 +262,7 @@ bool i915_gem_valid_gtt_space(struct i915_vma *vma, unsigned long cache_level);
 bool i915_vma_misplaced(const struct i915_vma *vma,
 			u64 size, u64 alignment, u64 flags);
 void __i915_vma_set_map_and_fenceable(struct i915_vma *vma);
+void i915_vma_revoke_mmap(struct i915_vma *vma);
 int __must_check i915_vma_unbind(struct i915_vma *vma);
 void i915_vma_unlink_ctx(struct i915_vma *vma);
 void i915_vma_close(struct i915_vma *vma);
@@ -321,12 +341,7 @@ void __iomem *i915_vma_pin_iomap(struct i915_vma *vma);
  * Callers must hold the struct_mutex. This function is only valid to be
  * called on a VMA previously iomapped by the caller with i915_vma_pin_iomap().
  */
-static inline void i915_vma_unpin_iomap(struct i915_vma *vma)
-{
-	lockdep_assert_held(&vma->obj->base.dev->struct_mutex);
-	GEM_BUG_ON(vma->iomap == NULL);
-	i915_vma_unpin(vma);
-}
+void i915_vma_unpin_iomap(struct i915_vma *vma);
 
 static inline struct page *i915_vma_first_page(struct i915_vma *vma)
 {
@@ -349,15 +364,13 @@ static inline struct page *i915_vma_first_page(struct i915_vma *vma)
  *
  * True if the vma has a fence, false otherwise.
  */
-static inline bool
-i915_vma_pin_fence(struct i915_vma *vma)
+int i915_vma_pin_fence(struct i915_vma *vma);
+int __must_check i915_vma_put_fence(struct i915_vma *vma);
+
+static inline void __i915_vma_unpin_fence(struct i915_vma *vma)
 {
-	lockdep_assert_held(&vma->obj->base.dev->struct_mutex);
-	if (vma->fence) {
-		vma->fence->pin_count++;
-		return true;
-	} else
-		return false;
+	GEM_BUG_ON(vma->fence->pin_count <= 0);
+	vma->fence->pin_count--;
 }
 
 /**
@@ -372,10 +385,8 @@ static inline void
 i915_vma_unpin_fence(struct i915_vma *vma)
 {
 	lockdep_assert_held(&vma->obj->base.dev->struct_mutex);
-	if (vma->fence) {
-		GEM_BUG_ON(vma->fence->pin_count <= 0);
-		vma->fence->pin_count--;
-	}
+	if (vma->fence)
+		__i915_vma_unpin_fence(vma);
 }
 
 #endif
diff --git a/drivers/gpu/drm/i915/intel_audio.c b/drivers/gpu/drm/i915/intel_audio.c
index 27743be5b768..0ddba16fde1b 100644
--- a/drivers/gpu/drm/i915/intel_audio.c
+++ b/drivers/gpu/drm/i915/intel_audio.c
@@ -754,7 +754,7 @@ static struct intel_encoder *get_saved_enc(struct drm_i915_private *dev_priv,
 {
 	struct intel_encoder *encoder;
 
-	if (WARN_ON(pipe >= I915_MAX_PIPES))
+	if (WARN_ON(pipe >= INTEL_INFO(dev_priv)->num_pipes))
 		return NULL;
 
 	/* MST */
diff --git a/drivers/gpu/drm/i915/intel_bios.c b/drivers/gpu/drm/i915/intel_bios.c
index b881ce8596ee..9d5b42953436 100644
--- a/drivers/gpu/drm/i915/intel_bios.c
+++ b/drivers/gpu/drm/i915/intel_bios.c
@@ -431,27 +431,6 @@ parse_general_features(struct drm_i915_private *dev_priv,
 		      dev_priv->vbt.fdi_rx_polarity_inverted);
 }
 
-static void
-parse_general_definitions(struct drm_i915_private *dev_priv,
-			  const struct bdb_header *bdb)
-{
-	const struct bdb_general_definitions *general;
-
-	general = find_section(bdb, BDB_GENERAL_DEFINITIONS);
-	if (general) {
-		u16 block_size = get_blocksize(general);
-		if (block_size >= sizeof(*general)) {
-			int bus_pin = general->crt_ddc_gmbus_pin;
-			DRM_DEBUG_KMS("crt_ddc_bus_pin: %d\n", bus_pin);
-			if (intel_gmbus_is_valid_pin(dev_priv, bus_pin))
-				dev_priv->vbt.crt_ddc_pin = bus_pin;
-		} else {
-			DRM_DEBUG_KMS("BDB_GD too small (%d). Invalid.\n",
-				      block_size);
-		}
-	}
-}
-
 static const struct child_device_config *
 child_device_ptr(const struct bdb_general_definitions *defs, int i)
 {
@@ -459,41 +438,24 @@ child_device_ptr(const struct bdb_general_definitions *defs, int i)
 }
 
 static void
-parse_sdvo_device_mapping(struct drm_i915_private *dev_priv,
-			  const struct bdb_header *bdb)
+parse_sdvo_device_mapping(struct drm_i915_private *dev_priv, u8 bdb_version)
 {
 	struct sdvo_device_mapping *mapping;
-	const struct bdb_general_definitions *defs;
 	const struct child_device_config *child;
-	int i, child_device_num, count;
-	u16	block_size;
-
-	defs = find_section(bdb, BDB_GENERAL_DEFINITIONS);
-	if (!defs) {
-		DRM_DEBUG_KMS("No general definition block is found, unable to construct sdvo mapping.\n");
-		return;
-	}
+	int i, count = 0;
 
 	/*
-	 * Only parse SDVO mappings when the general definitions block child
-	 * device size matches that of the *legacy* child device config
-	 * struct. Thus, SDVO mapping will be skipped for newer VBT.
+	 * Only parse SDVO mappings on gens that could have SDVO. This isn't
+	 * accurate and doesn't have to be, as long as it's not too strict.
 	 */
-	if (defs->child_dev_size != LEGACY_CHILD_DEVICE_CONFIG_SIZE) {
-		DRM_DEBUG_KMS("Unsupported child device size for SDVO mapping.\n");
+	if (!IS_GEN(dev_priv, 3, 7)) {
+		DRM_DEBUG_KMS("Skipping SDVO device mapping\n");
 		return;
 	}
-	/* get the block size of general definitions */
-	block_size = get_blocksize(defs);
-	/* get the number of child device */
-	child_device_num = (block_size - sizeof(*defs)) / defs->child_dev_size;
-	count = 0;
-	for (i = 0; i < child_device_num; i++) {
-		child = child_device_ptr(defs, i);
-		if (!child->device_type) {
-			/* skip the device block if device type is invalid */
-			continue;
-		}
+
+	for (i = 0, count = 0; i < dev_priv->vbt.child_dev_num; i++) {
+		child = dev_priv->vbt.child_dev + i;
+
 		if (child->slave_addr != SLAVE_ADDR1 &&
 		    child->slave_addr != SLAVE_ADDR2) {
 			/*
@@ -544,7 +506,6 @@ parse_sdvo_device_mapping(struct drm_i915_private *dev_priv,
 		/* No SDVO device info is found */
 		DRM_DEBUG_KMS("No SDVO device info is found in VBT\n");
 	}
-	return;
 }
 
 static void
@@ -1111,7 +1072,7 @@ static void sanitize_aux_ch(struct drm_i915_private *dev_priv,
 }
 
 static void parse_ddi_port(struct drm_i915_private *dev_priv, enum port port,
-			   const struct bdb_header *bdb)
+			   u8 bdb_version)
 {
 	struct child_device_config *it, *child = NULL;
 	struct ddi_vbt_port_info *info = &dev_priv->vbt.ddi_port_info[port];
@@ -1215,7 +1176,7 @@ static void parse_ddi_port(struct drm_i915_private *dev_priv, enum port port,
 		sanitize_aux_ch(dev_priv, port);
 	}
 
-	if (bdb->version >= 158) {
+	if (bdb_version >= 158) {
 		/* The VBT HDMI level shift values match the table we have. */
 		hdmi_level_shift = child->hdmi_level_shifter_value;
 		DRM_DEBUG_KMS("VBT HDMI level shift for port %c: %d\n",
@@ -1225,7 +1186,7 @@ static void parse_ddi_port(struct drm_i915_private *dev_priv, enum port port,
 	}
 
 	/* Parse the I_boost config for SKL and above */
-	if (bdb->version >= 196 && child->iboost) {
+	if (bdb_version >= 196 && child->iboost) {
 		info->dp_boost_level = translate_iboost(child->dp_iboost_level);
 		DRM_DEBUG_KMS("VBT (e)DP boost level for port %c: %d\n",
 			      port_name(port), info->dp_boost_level);
@@ -1235,40 +1196,52 @@ static void parse_ddi_port(struct drm_i915_private *dev_priv, enum port port,
 	}
 }
 
-static void parse_ddi_ports(struct drm_i915_private *dev_priv,
-			    const struct bdb_header *bdb)
+static void parse_ddi_ports(struct drm_i915_private *dev_priv, u8 bdb_version)
 {
 	enum port port;
 
-	if (!HAS_DDI(dev_priv))
+	if (!HAS_DDI(dev_priv) && !IS_CHERRYVIEW(dev_priv))
 		return;
 
 	if (!dev_priv->vbt.child_dev_num)
 		return;
 
-	if (bdb->version < 155)
+	if (bdb_version < 155)
 		return;
 
 	for (port = PORT_A; port < I915_MAX_PORTS; port++)
-		parse_ddi_port(dev_priv, port, bdb);
+		parse_ddi_port(dev_priv, port, bdb_version);
 }
 
 static void
-parse_device_mapping(struct drm_i915_private *dev_priv,
-		     const struct bdb_header *bdb)
+parse_general_definitions(struct drm_i915_private *dev_priv,
+			  const struct bdb_header *bdb)
 {
 	const struct bdb_general_definitions *defs;
 	const struct child_device_config *child;
-	struct child_device_config *child_dev_ptr;
 	int i, child_device_num, count;
 	u8 expected_size;
 	u16 block_size;
+	int bus_pin;
 
 	defs = find_section(bdb, BDB_GENERAL_DEFINITIONS);
 	if (!defs) {
 		DRM_DEBUG_KMS("No general definition block is found, no devices defined.\n");
 		return;
 	}
+
+	block_size = get_blocksize(defs);
+	if (block_size < sizeof(*defs)) {
+		DRM_DEBUG_KMS("General definitions block too small (%u)\n",
+			      block_size);
+		return;
+	}
+
+	bus_pin = defs->crt_ddc_gmbus_pin;
+	DRM_DEBUG_KMS("crt_ddc_bus_pin: %d\n", bus_pin);
+	if (intel_gmbus_is_valid_pin(dev_priv, bus_pin))
+		dev_priv->vbt.crt_ddc_pin = bus_pin;
+
 	if (bdb->version < 106) {
 		expected_size = 22;
 	} else if (bdb->version < 111) {
@@ -1298,18 +1271,14 @@ parse_device_mapping(struct drm_i915_private *dev_priv,
 		return;
 	}
 
-	/* get the block size of general definitions */
-	block_size = get_blocksize(defs);
 	/* get the number of child device */
 	child_device_num = (block_size - sizeof(*defs)) / defs->child_dev_size;
 	count = 0;
 	/* get the number of child device that is present */
 	for (i = 0; i < child_device_num; i++) {
 		child = child_device_ptr(defs, i);
-		if (!child->device_type) {
-			/* skip the device block if device type is invalid */
+		if (!child->device_type)
 			continue;
-		}
 		count++;
 	}
 	if (!count) {
@@ -1326,36 +1295,18 @@ parse_device_mapping(struct drm_i915_private *dev_priv,
 	count = 0;
 	for (i = 0; i < child_device_num; i++) {
 		child = child_device_ptr(defs, i);
-		if (!child->device_type) {
-			/* skip the device block if device type is invalid */
+		if (!child->device_type)
 			continue;
-		}
-
-		child_dev_ptr = dev_priv->vbt.child_dev + count;
-		count++;
 
 		/*
 		 * Copy as much as we know (sizeof) and is available
 		 * (child_dev_size) of the child device. Accessing the data must
 		 * depend on VBT version.
 		 */
-		memcpy(child_dev_ptr, child,
+		memcpy(dev_priv->vbt.child_dev + count, child,
 		       min_t(size_t, defs->child_dev_size, sizeof(*child)));
-
-		/*
-		 * copied full block, now init values when they are not
-		 * available in current version
-		 */
-		if (bdb->version < 196) {
-			/* Set default values for bits added from v196 */
-			child_dev_ptr->iboost = 0;
-			child_dev_ptr->hpd_invert = 0;
-		}
-
-		if (bdb->version < 192)
-			child_dev_ptr->lspcon = 0;
+		count++;
 	}
-	return;
 }
 
 /* Common defaults which may be overridden by VBT. */
@@ -1536,14 +1487,15 @@ void intel_bios_init(struct drm_i915_private *dev_priv)
 	parse_lfp_panel_data(dev_priv, bdb);
 	parse_lfp_backlight(dev_priv, bdb);
 	parse_sdvo_panel_data(dev_priv, bdb);
-	parse_sdvo_device_mapping(dev_priv, bdb);
-	parse_device_mapping(dev_priv, bdb);
 	parse_driver_features(dev_priv, bdb);
 	parse_edp(dev_priv, bdb);
 	parse_psr(dev_priv, bdb);
 	parse_mipi_config(dev_priv, bdb);
 	parse_mipi_sequence(dev_priv, bdb);
-	parse_ddi_ports(dev_priv, bdb);
+
+	/* Further processing on pre-parsed data */
+	parse_sdvo_device_mapping(dev_priv, bdb->version);
+	parse_ddi_ports(dev_priv, bdb->version);
 
 out:
 	if (!vbt) {
diff --git a/drivers/gpu/drm/i915/intel_cdclk.c b/drivers/gpu/drm/i915/intel_cdclk.c
index 87fc42b19336..b2a6d62b71c0 100644
--- a/drivers/gpu/drm/i915/intel_cdclk.c
+++ b/drivers/gpu/drm/i915/intel_cdclk.c
@@ -503,7 +503,7 @@ static void vlv_set_cdclk(struct drm_i915_private *dev_priv,
 	else
 		cmd = 0;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	val = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ);
 	val &= ~DSPFREQGUAR_MASK;
 	val |= (cmd << DSPFREQGUAR_SHIFT);
@@ -513,7 +513,7 @@ static void vlv_set_cdclk(struct drm_i915_private *dev_priv,
 		     50)) {
 		DRM_ERROR("timed out waiting for CDclk change\n");
 	}
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	mutex_lock(&dev_priv->sb_lock);
 
@@ -590,7 +590,7 @@ static void chv_set_cdclk(struct drm_i915_private *dev_priv,
 	 */
 	cmd = DIV_ROUND_CLOSEST(dev_priv->hpll_freq << 1, cdclk) - 1;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	val = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ);
 	val &= ~DSPFREQGUAR_MASK_CHV;
 	val |= (cmd << DSPFREQGUAR_SHIFT_CHV);
@@ -600,7 +600,7 @@ static void chv_set_cdclk(struct drm_i915_private *dev_priv,
 		     50)) {
 		DRM_ERROR("timed out waiting for CDclk change\n");
 	}
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	intel_update_cdclk(dev_priv);
 
@@ -656,10 +656,10 @@ static void bdw_set_cdclk(struct drm_i915_private *dev_priv,
 		 "trying to change cdclk frequency with cdclk not enabled\n"))
 		return;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	ret = sandybridge_pcode_write(dev_priv,
 				      BDW_PCODE_DISPLAY_FREQ_CHANGE_REQ, 0x0);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 	if (ret) {
 		DRM_ERROR("failed to inform pcode about cdclk change\n");
 		return;
@@ -712,9 +712,9 @@ static void bdw_set_cdclk(struct drm_i915_private *dev_priv,
 			LCPLL_CD_SOURCE_FCLK_DONE) == 0, 1))
 		DRM_ERROR("Switching back to LCPLL failed\n");
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	sandybridge_pcode_write(dev_priv, HSW_PCODE_DE_WRITE_FREQ_REQ, data);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	I915_WRITE(CDCLK_FREQ, DIV_ROUND_CLOSEST(cdclk, 1000) - 1);
 
@@ -928,12 +928,12 @@ static void skl_set_cdclk(struct drm_i915_private *dev_priv,
 
 	WARN_ON((cdclk == 24000) != (vco == 0));
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	ret = skl_pcode_request(dev_priv, SKL_PCODE_CDCLK_CONTROL,
 				SKL_CDCLK_PREPARE_FOR_CHANGE,
 				SKL_CDCLK_READY_FOR_CHANGE,
 				SKL_CDCLK_READY_FOR_CHANGE, 3);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 	if (ret) {
 		DRM_ERROR("Failed to inform PCU about cdclk change (%d)\n",
 			  ret);
@@ -975,9 +975,9 @@ static void skl_set_cdclk(struct drm_i915_private *dev_priv,
 	POSTING_READ(CDCLK_CTL);
 
 	/* inform PCU of the change */
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	sandybridge_pcode_write(dev_priv, SKL_PCODE_CDCLK_CONTROL, pcu_ack);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	intel_update_cdclk(dev_priv);
 }
@@ -1268,10 +1268,10 @@ static void bxt_set_cdclk(struct drm_i915_private *dev_priv,
 	}
 
 	/* Inform power controller of upcoming frequency change */
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	ret = sandybridge_pcode_write(dev_priv, HSW_PCODE_DE_WRITE_FREQ_REQ,
 				      0x80000000);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	if (ret) {
 		DRM_ERROR("PCode CDCLK freq change notify failed (err %d, freq %d)\n",
@@ -1300,10 +1300,10 @@ static void bxt_set_cdclk(struct drm_i915_private *dev_priv,
 		val |= BXT_CDCLK_SSA_PRECHARGE_ENABLE;
 	I915_WRITE(CDCLK_CTL, val);
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	ret = sandybridge_pcode_write(dev_priv, HSW_PCODE_DE_WRITE_FREQ_REQ,
 				      DIV_ROUND_UP(cdclk, 25000));
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	if (ret) {
 		DRM_ERROR("PCode CDCLK freq set failed, (err %d, freq %d)\n",
@@ -1518,12 +1518,12 @@ static void cnl_set_cdclk(struct drm_i915_private *dev_priv,
 	u32 val, divider, pcu_ack;
 	int ret;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	ret = skl_pcode_request(dev_priv, SKL_PCODE_CDCLK_CONTROL,
 				SKL_CDCLK_PREPARE_FOR_CHANGE,
 				SKL_CDCLK_READY_FOR_CHANGE,
 				SKL_CDCLK_READY_FOR_CHANGE, 3);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 	if (ret) {
 		DRM_ERROR("Failed to inform PCU about cdclk change (%d)\n",
 			  ret);
@@ -1575,9 +1575,9 @@ static void cnl_set_cdclk(struct drm_i915_private *dev_priv,
 	I915_WRITE(CDCLK_CTL, val);
 
 	/* inform PCU of the change */
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 	sandybridge_pcode_write(dev_priv, SKL_PCODE_CDCLK_CONTROL, pcu_ack);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	intel_update_cdclk(dev_priv);
 }
diff --git a/drivers/gpu/drm/i915/intel_color.c b/drivers/gpu/drm/i915/intel_color.c
index ff9ecd211abb..b8315bca852b 100644
--- a/drivers/gpu/drm/i915/intel_color.c
+++ b/drivers/gpu/drm/i915/intel_color.c
@@ -74,7 +74,7 @@
 #define I9XX_CSC_COEFF_1_0		\
 	((7 << 12) | I9XX_CSC_COEFF_FP(CTM_COEFF_1_0, 8))
 
-static bool crtc_state_is_legacy(struct drm_crtc_state *state)
+static bool crtc_state_is_legacy_gamma(struct drm_crtc_state *state)
 {
 	return !state->degamma_lut &&
 		!state->ctm &&
@@ -288,7 +288,7 @@ static void cherryview_load_csc_matrix(struct drm_crtc_state *state)
 	}
 
 	mode = (state->ctm ? CGM_PIPE_MODE_CSC : 0);
-	if (!crtc_state_is_legacy(state)) {
+	if (!crtc_state_is_legacy_gamma(state)) {
 		mode |= (state->degamma_lut ? CGM_PIPE_MODE_DEGAMMA : 0) |
 			(state->gamma_lut ? CGM_PIPE_MODE_GAMMA : 0);
 	}
@@ -469,7 +469,7 @@ static void broadwell_load_luts(struct drm_crtc_state *state)
 	struct intel_crtc_state *intel_state = to_intel_crtc_state(state);
 	enum pipe pipe = to_intel_crtc(state->crtc)->pipe;
 
-	if (crtc_state_is_legacy(state)) {
+	if (crtc_state_is_legacy_gamma(state)) {
 		haswell_load_luts(state);
 		return;
 	}
@@ -529,7 +529,7 @@ static void glk_load_luts(struct drm_crtc_state *state)
 
 	glk_load_degamma_lut(state);
 
-	if (crtc_state_is_legacy(state)) {
+	if (crtc_state_is_legacy_gamma(state)) {
 		haswell_load_luts(state);
 		return;
 	}
@@ -551,7 +551,7 @@ static void cherryview_load_luts(struct drm_crtc_state *state)
 	uint32_t i, lut_size;
 	uint32_t word0, word1;
 
-	if (crtc_state_is_legacy(state)) {
+	if (crtc_state_is_legacy_gamma(state)) {
 		/* Turn off degamma/gamma on CGM block. */
 		I915_WRITE(CGM_PIPE_MODE(pipe),
 			   (state->ctm ? CGM_PIPE_MODE_CSC : 0));
@@ -632,12 +632,10 @@ int intel_color_check(struct drm_crtc *crtc,
 		return 0;
 
 	/*
-	 * We also allow no degamma lut and a gamma lut at the legacy
+	 * We also allow no degamma lut/ctm and a gamma lut at the legacy
 	 * size (256 entries).
 	 */
-	if (!crtc_state->degamma_lut &&
-	    crtc_state->gamma_lut &&
-	    crtc_state->gamma_lut->length == LEGACY_LUT_LENGTH)
+	if (crtc_state_is_legacy_gamma(crtc_state))
 		return 0;
 
 	return -EINVAL;
diff --git a/drivers/gpu/drm/i915/intel_crt.c b/drivers/gpu/drm/i915/intel_crt.c
index 954070255b4d..668e8c3e791d 100644
--- a/drivers/gpu/drm/i915/intel_crt.c
+++ b/drivers/gpu/drm/i915/intel_crt.c
@@ -213,6 +213,19 @@ static void pch_post_disable_crt(struct intel_encoder *encoder,
 	intel_disable_crt(encoder, old_crtc_state, old_conn_state);
 }
 
+static void hsw_disable_crt(struct intel_encoder *encoder,
+			    const struct intel_crtc_state *old_crtc_state,
+			    const struct drm_connector_state *old_conn_state)
+{
+	struct drm_crtc *crtc = old_crtc_state->base.crtc;
+	struct drm_i915_private *dev_priv = to_i915(crtc->dev);
+	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+
+	WARN_ON(!intel_crtc->config->has_pch_encoder);
+
+	intel_set_pch_fifo_underrun_reporting(dev_priv, PIPE_A, false);
+}
+
 static void hsw_post_disable_crt(struct intel_encoder *encoder,
 				 const struct intel_crtc_state *old_crtc_state,
 				 const struct drm_connector_state *old_conn_state)
@@ -225,6 +238,58 @@ static void hsw_post_disable_crt(struct intel_encoder *encoder,
 	lpt_disable_iclkip(dev_priv);
 
 	intel_ddi_fdi_post_disable(encoder, old_crtc_state, old_conn_state);
+
+	WARN_ON(!old_crtc_state->has_pch_encoder);
+
+	intel_set_pch_fifo_underrun_reporting(dev_priv, PIPE_A, true);
+}
+
+static void hsw_pre_pll_enable_crt(struct intel_encoder *encoder,
+				   const struct intel_crtc_state *pipe_config,
+				   const struct drm_connector_state *conn_state)
+{
+	struct drm_crtc *crtc = pipe_config->base.crtc;
+	struct drm_i915_private *dev_priv = to_i915(crtc->dev);
+	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+
+	WARN_ON(!intel_crtc->config->has_pch_encoder);
+
+	intel_set_pch_fifo_underrun_reporting(dev_priv, PIPE_A, false);
+}
+
+static void hsw_pre_enable_crt(struct intel_encoder *encoder,
+			       const struct intel_crtc_state *pipe_config,
+			       const struct drm_connector_state *conn_state)
+{
+	struct drm_crtc *crtc = pipe_config->base.crtc;
+	struct drm_i915_private *dev_priv = to_i915(crtc->dev);
+	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+	int pipe = intel_crtc->pipe;
+
+	WARN_ON(!intel_crtc->config->has_pch_encoder);
+
+	intel_set_cpu_fifo_underrun_reporting(dev_priv, pipe, false);
+
+	dev_priv->display.fdi_link_train(intel_crtc, pipe_config);
+}
+
+static void hsw_enable_crt(struct intel_encoder *encoder,
+			   const struct intel_crtc_state *pipe_config,
+			   const struct drm_connector_state *conn_state)
+{
+	struct drm_crtc *crtc = pipe_config->base.crtc;
+	struct drm_i915_private *dev_priv = to_i915(crtc->dev);
+	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+	int pipe = intel_crtc->pipe;
+
+	WARN_ON(!intel_crtc->config->has_pch_encoder);
+
+	intel_crt_set_dpms(encoder, pipe_config, DRM_MODE_DPMS_ON);
+
+	intel_wait_for_vblank(dev_priv, pipe);
+	intel_wait_for_vblank(dev_priv, pipe);
+	intel_set_cpu_fifo_underrun_reporting(dev_priv, pipe, true);
+	intel_set_pch_fifo_underrun_reporting(dev_priv, PIPE_A, true);
 }
 
 static void intel_enable_crt(struct intel_encoder *encoder,
@@ -890,26 +955,31 @@ void intel_crt_init(struct drm_i915_private *dev_priv)
 
 	crt->base.power_domain = POWER_DOMAIN_PORT_CRT;
 
-	crt->base.compute_config = intel_crt_compute_config;
-	if (HAS_PCH_SPLIT(dev_priv)) {
-		crt->base.disable = pch_disable_crt;
-		crt->base.post_disable = pch_post_disable_crt;
-	} else {
-		crt->base.disable = intel_disable_crt;
-	}
-	crt->base.enable = intel_enable_crt;
 	if (I915_HAS_HOTPLUG(dev_priv) &&
 	    !dmi_check_system(intel_spurious_crt_detect))
 		crt->base.hpd_pin = HPD_CRT;
+
+	crt->base.compute_config = intel_crt_compute_config;
 	if (HAS_DDI(dev_priv)) {
 		crt->base.port = PORT_E;
 		crt->base.get_config = hsw_crt_get_config;
 		crt->base.get_hw_state = intel_ddi_get_hw_state;
+		crt->base.pre_pll_enable = hsw_pre_pll_enable_crt;
+		crt->base.pre_enable = hsw_pre_enable_crt;
+		crt->base.enable = hsw_enable_crt;
+		crt->base.disable = hsw_disable_crt;
 		crt->base.post_disable = hsw_post_disable_crt;
 	} else {
+		if (HAS_PCH_SPLIT(dev_priv)) {
+			crt->base.disable = pch_disable_crt;
+			crt->base.post_disable = pch_post_disable_crt;
+		} else {
+			crt->base.disable = intel_disable_crt;
+		}
 		crt->base.port = PORT_NONE;
 		crt->base.get_config = intel_crt_get_config;
 		crt->base.get_hw_state = intel_crt_get_hw_state;
+		crt->base.enable = intel_enable_crt;
 	}
 	intel_connector->get_hw_state = intel_connector_get_hw_state;
 
diff --git a/drivers/gpu/drm/i915/intel_csr.c b/drivers/gpu/drm/i915/intel_csr.c
index cdfb624eb82d..ea5d5c9645a4 100644
--- a/drivers/gpu/drm/i915/intel_csr.c
+++ b/drivers/gpu/drm/i915/intel_csr.c
@@ -216,7 +216,7 @@ static void gen9_set_dc_state_debugmask(struct drm_i915_private *dev_priv)
 
 	mask = DC_STATE_DEBUG_MASK_MEMORY_UP;
 
-	if (IS_BROXTON(dev_priv))
+	if (IS_GEN9_LP(dev_priv))
 		mask |= DC_STATE_DEBUG_MASK_CORES;
 
 	/* The below bit doesn't need to be cleared ever afterwards */
diff --git a/drivers/gpu/drm/i915/intel_ddi.c b/drivers/gpu/drm/i915/intel_ddi.c
index 93cbbcbbc193..b307b6fe1ce3 100644
--- a/drivers/gpu/drm/i915/intel_ddi.c
+++ b/drivers/gpu/drm/i915/intel_ddi.c
@@ -602,8 +602,10 @@ cnl_get_buf_trans_hdmi(struct drm_i915_private *dev_priv, int *n_entries)
 	} else if (voltage == VOLTAGE_INFO_1_05V) {
 		*n_entries = ARRAY_SIZE(cnl_ddi_translations_hdmi_1_05V);
 		return cnl_ddi_translations_hdmi_1_05V;
-	} else
+	} else {
+		*n_entries = 1; /* shut up gcc */
 		MISSING_CASE(voltage);
+	}
 	return NULL;
 }
 
@@ -621,8 +623,10 @@ cnl_get_buf_trans_dp(struct drm_i915_private *dev_priv, int *n_entries)
 	} else if (voltage == VOLTAGE_INFO_1_05V) {
 		*n_entries = ARRAY_SIZE(cnl_ddi_translations_dp_1_05V);
 		return cnl_ddi_translations_dp_1_05V;
-	} else
+	} else {
+		*n_entries = 1; /* shut up gcc */
 		MISSING_CASE(voltage);
+	}
 	return NULL;
 }
 
@@ -641,8 +645,10 @@ cnl_get_buf_trans_edp(struct drm_i915_private *dev_priv, int *n_entries)
 		} else if (voltage == VOLTAGE_INFO_1_05V) {
 			*n_entries = ARRAY_SIZE(cnl_ddi_translations_edp_1_05V);
 			return cnl_ddi_translations_edp_1_05V;
-		} else
+		} else {
+			*n_entries = 1; /* shut up gcc */
 			MISSING_CASE(voltage);
+		}
 		return NULL;
 	} else {
 		return cnl_get_buf_trans_dp(dev_priv, n_entries);
@@ -1214,6 +1220,9 @@ static int cnl_calc_wrpll_link(struct drm_i915_private *dev_priv,
 	dco_freq += (((cfgcr0 & DPLL_CFGCR0_DCO_FRACTION_MASK) >>
 		      DPLL_CFGCR0_DCO_FRACTION_SHIFT) * ref_clock) / 0x8000;
 
+	if (WARN_ON(p0 == 0 || p1 == 0 || p2 == 0))
+		return 0;
+
 	return dco_freq / (p0 * p1 * p2 * 5);
 }
 
@@ -1713,7 +1722,8 @@ bool intel_ddi_get_hw_state(struct intel_encoder *encoder,
 out:
 	if (ret && IS_GEN9_LP(dev_priv)) {
 		tmp = I915_READ(BXT_PHY_CTL(port));
-		if ((tmp & (BXT_PHY_LANE_POWERDOWN_ACK |
+		if ((tmp & (BXT_PHY_CMNLANE_POWERDOWN_ACK |
+			    BXT_PHY_LANE_POWERDOWN_ACK |
 			    BXT_PHY_LANE_ENABLED)) != BXT_PHY_LANE_ENABLED)
 			DRM_ERROR("Port %c enabled but PHY powered down? "
 				  "(PHY_CTL %08x)\n", port_name(port), tmp);
@@ -2161,7 +2171,8 @@ static void intel_ddi_pre_enable_dp(struct intel_encoder *encoder,
 		intel_prepare_dp_ddi_buffers(encoder);
 
 	intel_ddi_init_dp_buf_reg(encoder);
-	intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_ON);
+	if (!link_mst)
+		intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_ON);
 	intel_dp_start_link_train(intel_dp);
 	if (port != PORT_A || INTEL_GEN(dev_priv) >= 9)
 		intel_dp_stop_link_train(intel_dp);
@@ -2205,8 +2216,16 @@ static void intel_ddi_pre_enable(struct intel_encoder *encoder,
 				 const struct intel_crtc_state *pipe_config,
 				 const struct drm_connector_state *conn_state)
 {
+	struct drm_crtc *crtc = pipe_config->base.crtc;
+	struct drm_i915_private *dev_priv = to_i915(crtc->dev);
+	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+	int pipe = intel_crtc->pipe;
 	int type = encoder->type;
 
+	WARN_ON(intel_crtc->config->has_pch_encoder);
+
+	intel_set_cpu_fifo_underrun_reporting(dev_priv, pipe, true);
+
 	if (type == INTEL_OUTPUT_DP || type == INTEL_OUTPUT_EDP) {
 		intel_ddi_pre_enable_dp(encoder,
 					pipe_config->port_clock,
@@ -2235,12 +2254,21 @@ static void intel_ddi_post_disable(struct intel_encoder *intel_encoder,
 	uint32_t val;
 	bool wait = false;
 
-	/* old_crtc_state and old_conn_state are NULL when called from DP_MST */
-
 	if (type == INTEL_OUTPUT_DP || type == INTEL_OUTPUT_EDP) {
+		/*
+		 * old_crtc_state and old_conn_state are NULL when called from
+		 * DP_MST. The main connector associated with this port is never
+		 * bound to a crtc for MST.
+		 */
+		bool is_mst = !old_crtc_state;
 		struct intel_dp *intel_dp = enc_to_intel_dp(encoder);
 
-		intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_OFF);
+		/*
+		 * Power down sink before disabling the port, otherwise we end
+		 * up getting interrupts from the sink on detecting link loss.
+		 */
+		if (!is_mst)
+			intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_OFF);
 	}
 
 	val = I915_READ(DDI_BUF_CTL(port));
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 615c58e48613..7e91dc9a0fcf 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -1539,7 +1539,7 @@ static void chv_enable_pll(struct intel_crtc *crtc,
 		 * DPLLCMD is AWOL. Use chicken bits to propagate
 		 * the value from DPLLBMD to either pipe B or C.
 		 */
-		I915_WRITE(CBR4_VLV, pipe == PIPE_B ? CBR_DPLLBMD_PIPE_B : CBR_DPLLBMD_PIPE_C);
+		I915_WRITE(CBR4_VLV, CBR_DPLLBMD_PIPE(pipe));
 		I915_WRITE(DPLL_MD(PIPE_B), pipe_config->dpll_hw_state.dpll_md);
 		I915_WRITE(CBR4_VLV, 0);
 		dev_priv->chv_dpll_md[pipe] = pipe_config->dpll_hw_state.dpll_md;
@@ -1568,11 +1568,12 @@ static int intel_num_dvo_pipes(struct drm_i915_private *dev_priv)
 	return count;
 }
 
-static void i9xx_enable_pll(struct intel_crtc *crtc)
+static void i9xx_enable_pll(struct intel_crtc *crtc,
+			    const struct intel_crtc_state *crtc_state)
 {
 	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
 	i915_reg_t reg = DPLL(crtc->pipe);
-	u32 dpll = crtc->config->dpll_hw_state.dpll;
+	u32 dpll = crtc_state->dpll_hw_state.dpll;
 	int i;
 
 	assert_pipe_disabled(dev_priv, crtc->pipe);
@@ -1609,7 +1610,7 @@ static void i9xx_enable_pll(struct intel_crtc *crtc)
 
 	if (INTEL_GEN(dev_priv) >= 4) {
 		I915_WRITE(DPLL_MD(crtc->pipe),
-			   crtc->config->dpll_hw_state.dpll_md);
+			   crtc_state->dpll_hw_state.dpll_md);
 	} else {
 		/* The pixel multiplier can only be updated once the
 		 * DPLL is enabled and the clocks are stable.
@@ -1627,15 +1628,6 @@ static void i9xx_enable_pll(struct intel_crtc *crtc)
 	}
 }
 
-/**
- * i9xx_disable_pll - disable a PLL
- * @dev_priv: i915 private structure
- * @pipe: pipe PLL to disable
- *
- * Disable the PLL for @pipe, making sure the pipe is off first.
- *
- * Note!  This is for pre-ILK only.
- */
 static void i9xx_disable_pll(struct intel_crtc *crtc)
 {
 	struct drm_i915_private *dev_priv = to_i915(crtc->base.dev);
@@ -2219,8 +2211,7 @@ intel_pin_and_fence_fb_obj(struct drm_framebuffer *fb, unsigned int rotation)
 		 * something and try to run the system in a "less than optimal"
 		 * mode that matches the user configuration.
 		 */
-		if (i915_vma_get_fence(vma) == 0)
-			i915_vma_pin_fence(vma);
+		i915_vma_pin_fence(vma);
 	}
 
 	i915_vma_get(vma);
@@ -4955,10 +4946,10 @@ void hsw_enable_ips(struct intel_crtc *crtc)
 
 	assert_plane_enabled(dev_priv, crtc->plane);
 	if (IS_BROADWELL(dev_priv)) {
-		mutex_lock(&dev_priv->rps.hw_lock);
+		mutex_lock(&dev_priv->pcu_lock);
 		WARN_ON(sandybridge_pcode_write(dev_priv, DISPLAY_IPS_CONTROL,
 						IPS_ENABLE | IPS_PCODE_CONTROL));
-		mutex_unlock(&dev_priv->rps.hw_lock);
+		mutex_unlock(&dev_priv->pcu_lock);
 		/* Quoting Art Runyan: "its not safe to expect any particular
 		 * value in IPS_CTL bit 31 after enabling IPS through the
 		 * mailbox." Moreover, the mailbox may return a bogus state,
@@ -4988,9 +4979,9 @@ void hsw_disable_ips(struct intel_crtc *crtc)
 
 	assert_plane_enabled(dev_priv, crtc->plane);
 	if (IS_BROADWELL(dev_priv)) {
-		mutex_lock(&dev_priv->rps.hw_lock);
+		mutex_lock(&dev_priv->pcu_lock);
 		WARN_ON(sandybridge_pcode_write(dev_priv, DISPLAY_IPS_CONTROL, 0));
-		mutex_unlock(&dev_priv->rps.hw_lock);
+		mutex_unlock(&dev_priv->pcu_lock);
 		/* wait for pcode to finish disabling IPS, which may take up to 42ms */
 		if (intel_wait_for_register(dev_priv,
 					    IPS_CTL, IPS_ENABLE, 0,
@@ -5459,6 +5450,20 @@ static bool hsw_crtc_supports_ips(struct intel_crtc *crtc)
 	return HAS_IPS(to_i915(crtc->base.dev)) && crtc->pipe == PIPE_A;
 }
 
+static void glk_pipe_scaler_clock_gating_wa(struct drm_i915_private *dev_priv,
+					    enum pipe pipe, bool apply)
+{
+	u32 val = I915_READ(CLKGATE_DIS_PSL(pipe));
+	u32 mask = DPF_GATING_DIS | DPF_RAM_GATING_DIS | DPFR_GATING_DIS;
+
+	if (apply)
+		val |= mask;
+	else
+		val &= ~mask;
+
+	I915_WRITE(CLKGATE_DIS_PSL(pipe), val);
+}
+
 static void haswell_crtc_enable(struct intel_crtc_state *pipe_config,
 				struct drm_atomic_state *old_state)
 {
@@ -5469,13 +5474,11 @@ static void haswell_crtc_enable(struct intel_crtc_state *pipe_config,
 	enum transcoder cpu_transcoder = intel_crtc->config->cpu_transcoder;
 	struct intel_atomic_state *old_intel_state =
 		to_intel_atomic_state(old_state);
+	bool psl_clkgate_wa;
 
 	if (WARN_ON(intel_crtc->active))
 		return;
 
-	if (intel_crtc->config->has_pch_encoder)
-		intel_set_pch_fifo_underrun_reporting(dev_priv, PIPE_A, false);
-
 	intel_encoders_pre_pll_enable(crtc, pipe_config, old_state);
 
 	if (intel_crtc->config->shared_dpll)
@@ -5509,19 +5512,17 @@ static void haswell_crtc_enable(struct intel_crtc_state *pipe_config,
 
 	intel_crtc->active = true;
 
-	if (intel_crtc->config->has_pch_encoder)
-		intel_set_cpu_fifo_underrun_reporting(dev_priv, pipe, false);
-	else
-		intel_set_cpu_fifo_underrun_reporting(dev_priv, pipe, true);
-
 	intel_encoders_pre_enable(crtc, pipe_config, old_state);
 
-	if (intel_crtc->config->has_pch_encoder)
-		dev_priv->display.fdi_link_train(intel_crtc, pipe_config);
-
 	if (!transcoder_is_dsi(cpu_transcoder))
 		intel_ddi_enable_pipe_clock(pipe_config);
 
+	/* Display WA #1180: WaDisableScalarClockGating: glk, cnl */
+	psl_clkgate_wa = (IS_GEMINILAKE(dev_priv) || IS_CANNONLAKE(dev_priv)) &&
+			 intel_crtc->config->pch_pfit.enabled;
+	if (psl_clkgate_wa)
+		glk_pipe_scaler_clock_gating_wa(dev_priv, pipe, true);
+
 	if (INTEL_GEN(dev_priv) >= 9)
 		skylake_pfit_enable(intel_crtc);
 	else
@@ -5555,11 +5556,9 @@ static void haswell_crtc_enable(struct intel_crtc_state *pipe_config,
 
 	intel_encoders_enable(crtc, pipe_config, old_state);
 
-	if (intel_crtc->config->has_pch_encoder) {
+	if (psl_clkgate_wa) {
 		intel_wait_for_vblank(dev_priv, pipe);
-		intel_wait_for_vblank(dev_priv, pipe);
-		intel_set_cpu_fifo_underrun_reporting(dev_priv, pipe, true);
-		intel_set_pch_fifo_underrun_reporting(dev_priv, PIPE_A, true);
+		glk_pipe_scaler_clock_gating_wa(dev_priv, pipe, false);
 	}
 
 	/* If we change the relative order between pipe/planes enabling, we need
@@ -5655,9 +5654,6 @@ static void haswell_crtc_disable(struct intel_crtc_state *old_crtc_state,
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
 	enum transcoder cpu_transcoder = intel_crtc->config->cpu_transcoder;
 
-	if (intel_crtc->config->has_pch_encoder)
-		intel_set_pch_fifo_underrun_reporting(dev_priv, PIPE_A, false);
-
 	intel_encoders_disable(crtc, old_crtc_state, old_state);
 
 	drm_crtc_vblank_off(crtc);
@@ -5682,9 +5678,6 @@ static void haswell_crtc_disable(struct intel_crtc_state *old_crtc_state,
 		intel_ddi_disable_pipe_clock(intel_crtc->config);
 
 	intel_encoders_post_disable(crtc, old_crtc_state, old_state);
-
-	if (old_crtc_state->has_pch_encoder)
-		intel_set_pch_fifo_underrun_reporting(dev_priv, PIPE_A, true);
 }
 
 static void i9xx_pfit_enable(struct intel_crtc *crtc)
@@ -5894,7 +5887,7 @@ static void i9xx_crtc_enable(struct intel_crtc_state *pipe_config,
 
 	intel_encoders_pre_enable(crtc, pipe_config, old_state);
 
-	i9xx_enable_pll(intel_crtc);
+	i9xx_enable_pll(intel_crtc, pipe_config);
 
 	i9xx_pfit_enable(intel_crtc);
 
@@ -8846,11 +8839,11 @@ static uint32_t hsw_read_dcomp(struct drm_i915_private *dev_priv)
 static void hsw_write_dcomp(struct drm_i915_private *dev_priv, uint32_t val)
 {
 	if (IS_HASWELL(dev_priv)) {
-		mutex_lock(&dev_priv->rps.hw_lock);
+		mutex_lock(&dev_priv->pcu_lock);
 		if (sandybridge_pcode_write(dev_priv, GEN6_PCODE_WRITE_D_COMP,
 					    val))
 			DRM_DEBUG_KMS("Failed to write to D_COMP\n");
-		mutex_unlock(&dev_priv->rps.hw_lock);
+		mutex_unlock(&dev_priv->pcu_lock);
 	} else {
 		I915_WRITE(D_COMP_BDW, val);
 		POSTING_READ(D_COMP_BDW);
@@ -10245,58 +10238,44 @@ static void ironlake_pch_clock_get(struct intel_crtc *crtc,
 					 &pipe_config->fdi_m_n);
 }
 
-/** Returns the currently programmed mode of the given pipe. */
-struct drm_display_mode *intel_crtc_mode_get(struct drm_device *dev,
-					     struct drm_crtc *crtc)
+/* Returns the currently programmed mode of the given encoder. */
+struct drm_display_mode *
+intel_encoder_current_mode(struct intel_encoder *encoder)
 {
-	struct drm_i915_private *dev_priv = to_i915(dev);
-	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
-	enum transcoder cpu_transcoder = intel_crtc->config->cpu_transcoder;
+	struct drm_i915_private *dev_priv = to_i915(encoder->base.dev);
+	struct intel_crtc_state *crtc_state;
 	struct drm_display_mode *mode;
-	struct intel_crtc_state *pipe_config;
-	int htot = I915_READ(HTOTAL(cpu_transcoder));
-	int hsync = I915_READ(HSYNC(cpu_transcoder));
-	int vtot = I915_READ(VTOTAL(cpu_transcoder));
-	int vsync = I915_READ(VSYNC(cpu_transcoder));
-	enum pipe pipe = intel_crtc->pipe;
+	struct intel_crtc *crtc;
+	enum pipe pipe;
+
+	if (!encoder->get_hw_state(encoder, &pipe))
+		return NULL;
+
+	crtc = intel_get_crtc_for_pipe(dev_priv, pipe);
 
 	mode = kzalloc(sizeof(*mode), GFP_KERNEL);
 	if (!mode)
 		return NULL;
 
-	pipe_config = kzalloc(sizeof(*pipe_config), GFP_KERNEL);
-	if (!pipe_config) {
+	crtc_state = kzalloc(sizeof(*crtc_state), GFP_KERNEL);
+	if (!crtc_state) {
 		kfree(mode);
 		return NULL;
 	}
 
-	/*
-	 * Construct a pipe_config sufficient for getting the clock info
-	 * back out of crtc_clock_get.
-	 *
-	 * Note, if LVDS ever uses a non-1 pixel multiplier, we'll need
-	 * to use a real value here instead.
-	 */
-	pipe_config->cpu_transcoder = (enum transcoder) pipe;
-	pipe_config->pixel_multiplier = 1;
-	pipe_config->dpll_hw_state.dpll = I915_READ(DPLL(pipe));
-	pipe_config->dpll_hw_state.fp0 = I915_READ(FP0(pipe));
-	pipe_config->dpll_hw_state.fp1 = I915_READ(FP1(pipe));
-	i9xx_crtc_clock_get(intel_crtc, pipe_config);
-
-	mode->clock = pipe_config->port_clock / pipe_config->pixel_multiplier;
-	mode->hdisplay = (htot & 0xffff) + 1;
-	mode->htotal = ((htot & 0xffff0000) >> 16) + 1;
-	mode->hsync_start = (hsync & 0xffff) + 1;
-	mode->hsync_end = ((hsync & 0xffff0000) >> 16) + 1;
-	mode->vdisplay = (vtot & 0xffff) + 1;
-	mode->vtotal = ((vtot & 0xffff0000) >> 16) + 1;
-	mode->vsync_start = (vsync & 0xffff) + 1;
-	mode->vsync_end = ((vsync & 0xffff0000) >> 16) + 1;
+	crtc_state->base.crtc = &crtc->base;
 
-	drm_mode_set_name(mode);
+	if (!dev_priv->display.get_pipe_config(crtc, crtc_state)) {
+		kfree(crtc_state);
+		kfree(mode);
+		return NULL;
+	}
 
-	kfree(pipe_config);
+	encoder->get_config(encoder, crtc_state);
+
+	intel_mode_from_pipe_config(mode, crtc_state);
+
+	kfree(crtc_state);
 
 	return mode;
 }
@@ -11336,6 +11315,18 @@ intel_pipe_config_compare(struct drm_i915_private *dev_priv,
 	PIPE_CONF_CHECK_X(dpll_hw_state.ctrl1);
 	PIPE_CONF_CHECK_X(dpll_hw_state.cfgcr1);
 	PIPE_CONF_CHECK_X(dpll_hw_state.cfgcr2);
+	PIPE_CONF_CHECK_X(dpll_hw_state.cfgcr0);
+	PIPE_CONF_CHECK_X(dpll_hw_state.ebb0);
+	PIPE_CONF_CHECK_X(dpll_hw_state.ebb4);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pll0);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pll1);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pll2);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pll3);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pll6);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pll8);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pll9);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pll10);
+	PIPE_CONF_CHECK_X(dpll_hw_state.pcsdw12);
 
 	PIPE_CONF_CHECK_X(dsi_pll.ctrl);
 	PIPE_CONF_CHECK_X(dsi_pll.div);
@@ -12220,7 +12211,10 @@ static void skl_update_crtcs(struct drm_atomic_state *state)
 			if (updated & cmask || !cstate->base.active)
 				continue;
 
-			if (skl_ddb_allocation_overlaps(entries, &cstate->wm.skl.ddb, i))
+			if (skl_ddb_allocation_overlaps(dev_priv,
+							entries,
+							&cstate->wm.skl.ddb,
+							i))
 				continue;
 
 			updated |= cmask;
@@ -12515,21 +12509,10 @@ static int intel_atomic_commit(struct drm_device *dev,
 	struct drm_i915_private *dev_priv = to_i915(dev);
 	int ret = 0;
 
-	ret = drm_atomic_helper_setup_commit(state, nonblock);
-	if (ret)
-		return ret;
-
 	drm_atomic_state_get(state);
 	i915_sw_fence_init(&intel_state->commit_ready,
 			   intel_atomic_commit_ready);
 
-	ret = intel_atomic_prepare_commit(dev, state);
-	if (ret) {
-		DRM_DEBUG_ATOMIC("Preparing state failed with %i\n", ret);
-		i915_sw_fence_commit(&intel_state->commit_ready);
-		return ret;
-	}
-
 	/*
 	 * The intel_legacy_cursor_update() fast path takes care
 	 * of avoiding the vblank waits for simple cursor
@@ -12538,19 +12521,37 @@ static int intel_atomic_commit(struct drm_device *dev,
 	 * updates happen during the correct frames. Gen9+ have
 	 * double buffered watermarks and so shouldn't need this.
 	 *
-	 * Do this after drm_atomic_helper_setup_commit() and
-	 * intel_atomic_prepare_commit() because we still want
-	 * to skip the flip and fb cleanup waits. Although that
-	 * does risk yanking the mapping from under the display
-	 * engine.
+	 * Unset state->legacy_cursor_update before the call to
+	 * drm_atomic_helper_setup_commit() because otherwise
+	 * drm_atomic_helper_wait_for_flip_done() is a noop and
+	 * we get FIFO underruns because we didn't wait
+	 * for vblank.
 	 *
 	 * FIXME doing watermarks and fb cleanup from a vblank worker
 	 * (assuming we had any) would solve these problems.
 	 */
-	if (INTEL_GEN(dev_priv) < 9)
-		state->legacy_cursor_update = false;
+	if (INTEL_GEN(dev_priv) < 9 && state->legacy_cursor_update) {
+		struct intel_crtc_state *new_crtc_state;
+		struct intel_crtc *crtc;
+		int i;
+
+		for_each_new_intel_crtc_in_state(intel_state, crtc, new_crtc_state, i)
+			if (new_crtc_state->wm.need_postvbl_update ||
+			    new_crtc_state->update_wm_post)
+				state->legacy_cursor_update = false;
+	}
+
+	ret = intel_atomic_prepare_commit(dev, state);
+	if (ret) {
+		DRM_DEBUG_ATOMIC("Preparing state failed with %i\n", ret);
+		i915_sw_fence_commit(&intel_state->commit_ready);
+		return ret;
+	}
+
+	ret = drm_atomic_helper_setup_commit(state, nonblock);
+	if (!ret)
+		ret = drm_atomic_helper_swap_state(state, true);
 
-	ret = drm_atomic_helper_swap_state(state, true);
 	if (ret) {
 		i915_sw_fence_commit(&intel_state->commit_ready);
 
@@ -14748,10 +14749,10 @@ static struct intel_connector *intel_encoder_find_connector(struct intel_encoder
 }
 
 static bool has_pch_trancoder(struct drm_i915_private *dev_priv,
-			      enum transcoder pch_transcoder)
+			      enum pipe pch_transcoder)
 {
 	return HAS_PCH_IBX(dev_priv) || HAS_PCH_CPT(dev_priv) ||
-		(HAS_PCH_LPT_H(dev_priv) && pch_transcoder == TRANSCODER_A);
+		(HAS_PCH_LPT_H(dev_priv) && pch_transcoder == PIPE_A);
 }
 
 static void intel_sanitize_crtc(struct intel_crtc *crtc,
@@ -14834,7 +14835,7 @@ static void intel_sanitize_crtc(struct intel_crtc *crtc,
 		 * PCH transcoders B and C would prevent enabling the south
 		 * error interrupt (see cpt_can_enable_serr_int()).
 		 */
-		if (has_pch_trancoder(dev_priv, (enum transcoder)crtc->pipe))
+		if (has_pch_trancoder(dev_priv, crtc->pipe))
 			crtc->pch_fifo_underrun_disabled = true;
 	}
 }
diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
index 90e756c76f10..b0f446b68f42 100644
--- a/drivers/gpu/drm/i915/intel_dp.c
+++ b/drivers/gpu/drm/i915/intel_dp.c
@@ -137,32 +137,20 @@ static void vlv_steal_power_sequencer(struct drm_device *dev,
 				      enum pipe pipe);
 static void intel_dp_unset_edid(struct intel_dp *intel_dp);
 
-static int intel_dp_num_rates(u8 link_bw_code)
-{
-	switch (link_bw_code) {
-	default:
-		WARN(1, "invalid max DP link bw val %x, using 1.62Gbps\n",
-		     link_bw_code);
-	case DP_LINK_BW_1_62:
-		return 1;
-	case DP_LINK_BW_2_7:
-		return 2;
-	case DP_LINK_BW_5_4:
-		return 3;
-	}
-}
-
 /* update sink rates from dpcd */
 static void intel_dp_set_sink_rates(struct intel_dp *intel_dp)
 {
-	int i, num_rates;
+	int i, max_rate;
 
-	num_rates = intel_dp_num_rates(intel_dp->dpcd[DP_MAX_LINK_RATE]);
+	max_rate = drm_dp_bw_code_to_link_rate(intel_dp->dpcd[DP_MAX_LINK_RATE]);
 
-	for (i = 0; i < num_rates; i++)
+	for (i = 0; i < ARRAY_SIZE(default_rates); i++) {
+		if (default_rates[i] > max_rate)
+			break;
 		intel_dp->sink_rates[i] = default_rates[i];
+	}
 
-	intel_dp->num_sink_rates = num_rates;
+	intel_dp->num_sink_rates = i;
 }
 
 /* Theoretical max between source and sink */
@@ -254,15 +242,15 @@ intel_dp_set_source_rates(struct intel_dp *intel_dp)
 	} else if (IS_GEN9_BC(dev_priv)) {
 		source_rates = skl_rates;
 		size = ARRAY_SIZE(skl_rates);
-	} else {
+	} else if ((IS_HASWELL(dev_priv) && !IS_HSW_ULX(dev_priv)) ||
+		   IS_BROADWELL(dev_priv)) {
 		source_rates = default_rates;
 		size = ARRAY_SIZE(default_rates);
+	} else {
+		source_rates = default_rates;
+		size = ARRAY_SIZE(default_rates) - 1;
 	}
 
-	/* This depends on the fact that 5.4 is last value in the array */
-	if (!intel_dp_source_supports_hbr2(intel_dp))
-		size--;
-
 	intel_dp->source_rates = source_rates;
 	intel_dp->num_source_rates = size;
 }
@@ -1482,14 +1470,9 @@ intel_dp_aux_init(struct intel_dp *intel_dp)
 
 bool intel_dp_source_supports_hbr2(struct intel_dp *intel_dp)
 {
-	struct intel_digital_port *dig_port = dp_to_dig_port(intel_dp);
-	struct drm_i915_private *dev_priv = to_i915(dig_port->base.base.dev);
+	int max_rate = intel_dp->source_rates[intel_dp->num_source_rates - 1];
 
-	if ((IS_HASWELL(dev_priv) && !IS_HSW_ULX(dev_priv)) ||
-	    IS_BROADWELL(dev_priv) || (INTEL_GEN(dev_priv) >= 9))
-		return true;
-	else
-		return false;
+	return max_rate >= 540000;
 }
 
 static void
@@ -2308,8 +2291,8 @@ static void edp_panel_off(struct intel_dp *intel_dp)
 	I915_WRITE(pp_ctrl_reg, pp);
 	POSTING_READ(pp_ctrl_reg);
 
-	intel_dp->panel_power_off_time = ktime_get_boottime();
 	wait_panel_off(intel_dp);
+	intel_dp->panel_power_off_time = ktime_get_boottime();
 
 	/* We got a reference when we enabled the VDD. */
 	intel_display_power_put(dev_priv, intel_dp->aux_power_domain);
@@ -5286,7 +5269,7 @@ intel_dp_init_panel_power_sequencer(struct drm_device *dev,
 	 * seems sufficient to avoid this problem.
 	 */
 	if (dev_priv->quirks & QUIRK_INCREASE_T12_DELAY) {
-		vbt.t11_t12 = max_t(u16, vbt.t11_t12, 900 * 10);
+		vbt.t11_t12 = max_t(u16, vbt.t11_t12, 1300 * 10);
 		DRM_DEBUG_KMS("Increasing T12 panel delay as per the quirk to %d\n",
 			      vbt.t11_t12);
 	}
diff --git a/drivers/gpu/drm/i915/intel_dp_mst.c b/drivers/gpu/drm/i915/intel_dp_mst.c
index 9a396f483f8b..3c131e2544cf 100644
--- a/drivers/gpu/drm/i915/intel_dp_mst.c
+++ b/drivers/gpu/drm/i915/intel_dp_mst.c
@@ -162,14 +162,19 @@ static void intel_mst_post_disable_dp(struct intel_encoder *encoder,
 
 	drm_dp_mst_deallocate_vcpi(&intel_dp->mst_mgr, connector->port);
 
+	/*
+	 * Power down mst path before disabling the port, otherwise we end
+	 * up getting interrupts from the sink upon detecting link loss.
+	 */
+	drm_dp_send_power_updown_phy(&intel_dp->mst_mgr, connector->port,
+				     false);
+
 	intel_dp->active_mst_links--;
 
 	intel_mst->connector = NULL;
 	if (intel_dp->active_mst_links == 0) {
 		intel_dig_port->base.post_disable(&intel_dig_port->base,
 						  NULL, NULL);
-
-		intel_dp_sink_dpms(intel_dp, DRM_MODE_DPMS_OFF);
 	}
 	DRM_DEBUG_KMS("active links %d\n", intel_dp->active_mst_links);
 }
@@ -196,6 +201,7 @@ static void intel_mst_pre_enable_dp(struct intel_encoder *encoder,
 
 	DRM_DEBUG_KMS("active links %d\n", intel_dp->active_mst_links);
 
+	drm_dp_send_power_updown_phy(&intel_dp->mst_mgr, connector->port, true);
 	if (intel_dp->active_mst_links == 0)
 		intel_dig_port->base.pre_enable(&intel_dig_port->base,
 						pipe_config, NULL);
diff --git a/drivers/gpu/drm/i915/intel_dpio_phy.c b/drivers/gpu/drm/i915/intel_dpio_phy.c
index 09b670929786..de38d014ed39 100644
--- a/drivers/gpu/drm/i915/intel_dpio_phy.c
+++ b/drivers/gpu/drm/i915/intel_dpio_phy.c
@@ -208,12 +208,6 @@ static const struct bxt_ddi_phy_info glk_ddi_phy_info[] = {
 	},
 };
 
-static u32 bxt_phy_port_mask(const struct bxt_ddi_phy_info *phy_info)
-{
-	return (phy_info->dual_channel * BIT(phy_info->channel[DPIO_CH1].port)) |
-		BIT(phy_info->channel[DPIO_CH0].port);
-}
-
 static const struct bxt_ddi_phy_info *
 bxt_get_phy_list(struct drm_i915_private *dev_priv, int *count)
 {
@@ -313,7 +307,6 @@ bool bxt_ddi_phy_is_enabled(struct drm_i915_private *dev_priv,
 			    enum dpio_phy phy)
 {
 	const struct bxt_ddi_phy_info *phy_info;
-	enum port port;
 
 	phy_info = bxt_get_phy_info(dev_priv, phy);
 
@@ -335,19 +328,6 @@ bool bxt_ddi_phy_is_enabled(struct drm_i915_private *dev_priv,
 		return false;
 	}
 
-	for_each_port_masked(port, bxt_phy_port_mask(phy_info)) {
-		u32 tmp = I915_READ(BXT_PHY_CTL(port));
-
-		if (tmp & BXT_PHY_CMNLANE_POWERDOWN_ACK) {
-			DRM_DEBUG_DRIVER("DDI PHY %d powered, but common lane "
-					 "for port %c powered down "
-					 "(PHY_CTL %08x)\n",
-					 phy, port_name(port), tmp);
-
-			return false;
-		}
-	}
-
 	return true;
 }
 
diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
index 0cab667fff57..b87946dcc53f 100644
--- a/drivers/gpu/drm/i915/intel_drv.h
+++ b/drivers/gpu/drm/i915/intel_drv.h
@@ -1243,7 +1243,7 @@ void gen6_disable_rps_interrupts(struct drm_i915_private *dev_priv);
 static inline u32 gen6_sanitize_rps_pm_mask(const struct drm_i915_private *i915,
 					    u32 mask)
 {
-	return mask & ~i915->rps.pm_intrmsk_mbz;
+	return mask & ~i915->gt_pm.rps.pm_intrmsk_mbz;
 }
 
 void intel_runtime_pm_disable_interrupts(struct drm_i915_private *dev_priv);
@@ -1254,7 +1254,7 @@ static inline bool intel_irqs_enabled(struct drm_i915_private *dev_priv)
 	 * We only use drm_irq_uninstall() at unload and VT switch, so
 	 * this is the only thing we need to check.
 	 */
-	return dev_priv->pm.irqs_enabled;
+	return dev_priv->runtime_pm.irqs_enabled;
 }
 
 int intel_get_crtc_scanline(struct intel_crtc *crtc);
@@ -1363,8 +1363,9 @@ struct intel_connector *intel_connector_alloc(void);
 bool intel_connector_get_hw_state(struct intel_connector *connector);
 void intel_connector_attach_encoder(struct intel_connector *connector,
 				    struct intel_encoder *encoder);
-struct drm_display_mode *intel_crtc_mode_get(struct drm_device *dev,
-					     struct drm_crtc *crtc);
+struct drm_display_mode *
+intel_encoder_current_mode(struct intel_encoder *encoder);
+
 enum pipe intel_get_pipe_from_connector(struct intel_connector *connector);
 int intel_get_pipe_from_crtc_id(struct drm_device *dev, void *data,
 				struct drm_file *file_priv);
@@ -1790,7 +1791,7 @@ void intel_display_power_put(struct drm_i915_private *dev_priv,
 static inline void
 assert_rpm_device_not_suspended(struct drm_i915_private *dev_priv)
 {
-	WARN_ONCE(dev_priv->pm.suspended,
+	WARN_ONCE(dev_priv->runtime_pm.suspended,
 		  "Device suspended during HW access\n");
 }
 
@@ -1798,7 +1799,7 @@ static inline void
 assert_rpm_wakelock_held(struct drm_i915_private *dev_priv)
 {
 	assert_rpm_device_not_suspended(dev_priv);
-	WARN_ONCE(!atomic_read(&dev_priv->pm.wakeref_count),
+	WARN_ONCE(!atomic_read(&dev_priv->runtime_pm.wakeref_count),
 		  "RPM wakelock ref not held during HW access");
 }
 
@@ -1823,7 +1824,7 @@ assert_rpm_wakelock_held(struct drm_i915_private *dev_priv)
 static inline void
 disable_rpm_wakeref_asserts(struct drm_i915_private *dev_priv)
 {
-	atomic_inc(&dev_priv->pm.wakeref_count);
+	atomic_inc(&dev_priv->runtime_pm.wakeref_count);
 }
 
 /**
@@ -1840,7 +1841,7 @@ disable_rpm_wakeref_asserts(struct drm_i915_private *dev_priv)
 static inline void
 enable_rpm_wakeref_asserts(struct drm_i915_private *dev_priv)
 {
-	atomic_dec(&dev_priv->pm.wakeref_count);
+	atomic_dec(&dev_priv->runtime_pm.wakeref_count);
 }
 
 void intel_runtime_pm_get(struct drm_i915_private *dev_priv);
@@ -1893,7 +1894,8 @@ int intel_enable_sagv(struct drm_i915_private *dev_priv);
 int intel_disable_sagv(struct drm_i915_private *dev_priv);
 bool skl_wm_level_equals(const struct skl_wm_level *l1,
 			 const struct skl_wm_level *l2);
-bool skl_ddb_allocation_overlaps(const struct skl_ddb_entry **entries,
+bool skl_ddb_allocation_overlaps(struct drm_i915_private *dev_priv,
+				 const struct skl_ddb_entry **entries,
 				 const struct skl_ddb_entry *ddb,
 				 int ignore);
 bool ilk_disable_lp_wm(struct drm_device *dev);
@@ -1902,7 +1904,7 @@ int skl_check_pipe_max_pixel_rate(struct intel_crtc *intel_crtc,
 				  struct intel_crtc_state *cstate);
 void intel_init_ipc(struct drm_i915_private *dev_priv);
 void intel_enable_ipc(struct drm_i915_private *dev_priv);
-static inline int intel_enable_rc6(void)
+static inline int intel_rc6_enabled(void)
 {
 	return i915_modparams.enable_rc6;
 }
diff --git a/drivers/gpu/drm/i915/intel_dsi.c b/drivers/gpu/drm/i915/intel_dsi.c
index 20a7b004ffd7..66bbedc5fa01 100644
--- a/drivers/gpu/drm/i915/intel_dsi.c
+++ b/drivers/gpu/drm/i915/intel_dsi.c
@@ -790,14 +790,19 @@ static void intel_dsi_pre_enable(struct intel_encoder *encoder,
 				 const struct intel_crtc_state *pipe_config,
 				 const struct drm_connector_state *conn_state)
 {
-	struct drm_i915_private *dev_priv = to_i915(encoder->base.dev);
 	struct intel_dsi *intel_dsi = enc_to_intel_dsi(&encoder->base);
+	struct drm_crtc *crtc = pipe_config->base.crtc;
+	struct drm_i915_private *dev_priv = to_i915(crtc->dev);
+	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
+	int pipe = intel_crtc->pipe;
 	enum port port;
 	u32 val;
 	bool glk_cold_boot = false;
 
 	DRM_DEBUG_KMS("\n");
 
+	intel_set_cpu_fifo_underrun_reporting(dev_priv, pipe, true);
+
 	/*
 	 * The BIOS may leave the PLL in a wonky state where it doesn't
 	 * lock. It needs to be fully powered down to fix it.
diff --git a/drivers/gpu/drm/i915/intel_dvo.c b/drivers/gpu/drm/i915/intel_dvo.c
index 5c562e1f56ae..53c9b763f4ce 100644
--- a/drivers/gpu/drm/i915/intel_dvo.c
+++ b/drivers/gpu/drm/i915/intel_dvo.c
@@ -379,32 +379,15 @@ static const struct drm_encoder_funcs intel_dvo_enc_funcs = {
  * chip being on DVOB/C and having multiple pipes.
  */
 static struct drm_display_mode *
-intel_dvo_get_current_mode(struct drm_connector *connector)
+intel_dvo_get_current_mode(struct intel_encoder *encoder)
 {
-	struct drm_device *dev = connector->dev;
-	struct drm_i915_private *dev_priv = to_i915(dev);
-	struct intel_dvo *intel_dvo = intel_attached_dvo(connector);
-	uint32_t dvo_val = I915_READ(intel_dvo->dev.dvo_reg);
-	struct drm_display_mode *mode = NULL;
+	struct drm_display_mode *mode;
 
-	/* If the DVO port is active, that'll be the LVDS, so we can pull out
-	 * its timings to get how the BIOS set up the panel.
-	 */
-	if (dvo_val & DVO_ENABLE) {
-		struct intel_crtc *crtc;
-		int pipe = (dvo_val & DVO_PIPE_B_SELECT) ? 1 : 0;
-
-		crtc = intel_get_crtc_for_pipe(dev_priv, pipe);
-		if (crtc) {
-			mode = intel_crtc_mode_get(dev, &crtc->base);
-			if (mode) {
-				mode->type |= DRM_MODE_TYPE_PREFERRED;
-				if (dvo_val & DVO_HSYNC_ACTIVE_HIGH)
-					mode->flags |= DRM_MODE_FLAG_PHSYNC;
-				if (dvo_val & DVO_VSYNC_ACTIVE_HIGH)
-					mode->flags |= DRM_MODE_FLAG_PVSYNC;
-			}
-		}
+	mode = intel_encoder_current_mode(encoder);
+	if (mode) {
+		DRM_DEBUG_KMS("using current (BIOS) mode: ");
+		drm_mode_debug_printmodeline(mode);
+		mode->type |= DRM_MODE_TYPE_PREFERRED;
 	}
 
 	return mode;
@@ -551,7 +534,7 @@ void intel_dvo_init(struct drm_i915_private *dev_priv)
 			 * mode being output through DVO.
 			 */
 			intel_panel_init(&intel_connector->panel,
-					 intel_dvo_get_current_mode(connector),
+					 intel_dvo_get_current_mode(intel_encoder),
 					 NULL, NULL);
 			intel_dvo->panel_wants_dither = true;
 		}
diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c
index a28e2a864cf1..a59b2a30ff5a 100644
--- a/drivers/gpu/drm/i915/intel_engine_cs.c
+++ b/drivers/gpu/drm/i915/intel_engine_cs.c
@@ -22,6 +22,8 @@
  *
  */
 
+#include <drm/drm_print.h>
+
 #include "i915_drv.h"
 #include "intel_ringbuffer.h"
 #include "intel_lrc.h"
@@ -39,7 +41,7 @@
 
 #define GEN8_LR_CONTEXT_RENDER_SIZE	(20 * PAGE_SIZE)
 #define GEN9_LR_CONTEXT_RENDER_SIZE	(22 * PAGE_SIZE)
-#define GEN10_LR_CONTEXT_RENDER_SIZE	(19 * PAGE_SIZE)
+#define GEN10_LR_CONTEXT_RENDER_SIZE	(18 * PAGE_SIZE)
 
 #define GEN8_LR_CONTEXT_OTHER_SIZE	( 2 * PAGE_SIZE)
 
@@ -613,9 +615,22 @@ int intel_engine_init_common(struct intel_engine_cs *engine)
 	if (IS_ERR(ring))
 		return PTR_ERR(ring);
 
+	/*
+	 * Similarly the preempt context must always be available so that
+	 * we can interrupt the engine at any time.
+	 */
+	if (INTEL_INFO(engine->i915)->has_logical_ring_preemption) {
+		ring = engine->context_pin(engine,
+					   engine->i915->preempt_context);
+		if (IS_ERR(ring)) {
+			ret = PTR_ERR(ring);
+			goto err_unpin_kernel;
+		}
+	}
+
 	ret = intel_engine_init_breadcrumbs(engine);
 	if (ret)
-		goto err_unpin;
+		goto err_unpin_preempt;
 
 	ret = i915_gem_render_state_init(engine);
 	if (ret)
@@ -634,7 +649,10 @@ err_rs_fini:
 	i915_gem_render_state_fini(engine);
 err_breadcrumbs:
 	intel_engine_fini_breadcrumbs(engine);
-err_unpin:
+err_unpin_preempt:
+	if (INTEL_INFO(engine->i915)->has_logical_ring_preemption)
+		engine->context_unpin(engine, engine->i915->preempt_context);
+err_unpin_kernel:
 	engine->context_unpin(engine, engine->i915->kernel_context);
 	return ret;
 }
@@ -660,6 +678,8 @@ void intel_engine_cleanup_common(struct intel_engine_cs *engine)
 	intel_engine_cleanup_cmd_parser(engine);
 	i915_gem_batch_pool_fini(&engine->batch_pool);
 
+	if (INTEL_INFO(engine->i915)->has_logical_ring_preemption)
+		engine->context_unpin(engine, engine->i915->preempt_context);
 	engine->context_unpin(engine, engine->i915->kernel_context);
 }
 
@@ -830,11 +850,6 @@ static int wa_add(struct drm_i915_private *dev_priv,
 #define WA_SET_FIELD_MASKED(addr, mask, value) \
 	WA_REG(addr, mask, _MASKED_FIELD(mask, value))
 
-#define WA_SET_BIT(addr, mask) WA_REG(addr, mask, I915_READ(addr) | (mask))
-#define WA_CLR_BIT(addr, mask) WA_REG(addr, mask, I915_READ(addr) & ~(mask))
-
-#define WA_WRITE(addr, val) WA_REG(addr, 0xffffffff, val)
-
 static int wa_ring_whitelist_reg(struct intel_engine_cs *engine,
 				 i915_reg_t reg)
 {
@@ -845,8 +860,8 @@ static int wa_ring_whitelist_reg(struct intel_engine_cs *engine,
 	if (WARN_ON(index >= RING_MAX_NONPRIV_SLOTS))
 		return -EINVAL;
 
-	WA_WRITE(RING_FORCE_TO_NONPRIV(engine->mmio_base, index),
-		 i915_mmio_reg_offset(reg));
+	I915_WRITE(RING_FORCE_TO_NONPRIV(engine->mmio_base, index),
+		   i915_mmio_reg_offset(reg));
 	wa->hw_whitelist_count[engine->id]++;
 
 	return 0;
@@ -980,7 +995,11 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
 				  GEN9_PBE_COMPRESSED_HASH_SELECTION);
 		WA_SET_BIT_MASKED(GEN9_HALF_SLICE_CHICKEN7,
 				  GEN9_SAMPLER_HASH_COMPRESSED_READ_ADDR);
-		WA_SET_BIT(MMCD_MISC_CTRL, MMCD_PCLA | MMCD_HOTSPOT_EN);
+
+		I915_WRITE(MMCD_MISC_CTRL,
+			   I915_READ(MMCD_MISC_CTRL) |
+			   MMCD_PCLA |
+			   MMCD_HOTSPOT_EN);
 	}
 
 	/* WaClearFlowControlGpgpuContextSave:skl,bxt,kbl,glk,cfl */
@@ -1071,13 +1090,33 @@ static int gen9_init_workarounds(struct intel_engine_cs *engine)
 	I915_WRITE(GEN8_L3SQCREG4, (I915_READ(GEN8_L3SQCREG4) |
 				    GEN8_LQSC_FLUSH_COHERENT_LINES));
 
+	/*
+	 * Supporting preemption with fine-granularity requires changes in the
+	 * batch buffer programming. Since we can't break old userspace, we
+	 * need to set our default preemption level to safe value. Userspace is
+	 * still able to use more fine-grained preemption levels, since in
+	 * WaEnablePreemptionGranularityControlByUMD we're whitelisting the
+	 * per-ctx register. As such, WaDisable{3D,GPGPU}MidCmdPreemption are
+	 * not real HW workarounds, but merely a way to start using preemption
+	 * while maintaining old contract with userspace.
+	 */
+
+	/* WaDisable3DMidCmdPreemption:skl,bxt,glk,cfl,[cnl] */
+	WA_CLR_BIT_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_3D_OBJECT_LEVEL);
+
+	/* WaDisableGPGPUMidCmdPreemption:skl,bxt,blk,cfl,[cnl] */
+	WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_GPGPU_LEVEL_MASK,
+			    GEN9_PREEMPT_GPGPU_COMMAND_LEVEL);
+
 	/* WaVFEStateAfterPipeControlwithMediaStateClear:skl,bxt,glk,cfl */
 	ret = wa_ring_whitelist_reg(engine, GEN9_CTX_PREEMPT_REG);
 	if (ret)
 		return ret;
 
-	/* WaEnablePreemptionGranularityControlByUMD:skl,bxt,kbl,cfl */
-	ret= wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1);
+	/* WaEnablePreemptionGranularityControlByUMD:skl,bxt,kbl,cfl,[cnl] */
+	I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1,
+		   _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL));
+	ret = wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1);
 	if (ret)
 		return ret;
 
@@ -1139,14 +1178,6 @@ static int skl_init_workarounds(struct intel_engine_cs *engine)
 	if (ret)
 		return ret;
 
-	/*
-	 * Actual WA is to disable percontext preemption granularity control
-	 * until D0 which is the default case so this is equivalent to
-	 * !WaDisablePerCtxtPreemptionGranularityControl:skl
-	 */
-	I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1,
-		   _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL));
-
 	/* WaEnableGapsTsvCreditFix:skl */
 	I915_WRITE(GEN8_GARBCNTL, (I915_READ(GEN8_GARBCNTL) |
 				   GEN9_GAPS_TSV_CREDIT_DISABLE));
@@ -1278,7 +1309,16 @@ static int cnl_init_workarounds(struct intel_engine_cs *engine)
 	/* FtrEnableFastAnisoL1BankingFix: cnl */
 	WA_SET_BIT_MASKED(HALF_SLICE_CHICKEN3, CNL_FAST_ANISO_L1_BANKING_FIX);
 
+	/* WaDisable3DMidCmdPreemption:cnl */
+	WA_CLR_BIT_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_3D_OBJECT_LEVEL);
+
+	/* WaDisableGPGPUMidCmdPreemption:cnl */
+	WA_SET_FIELD_MASKED(GEN8_CS_CHICKEN1, GEN9_PREEMPT_GPGPU_LEVEL_MASK,
+			    GEN9_PREEMPT_GPGPU_COMMAND_LEVEL);
+
 	/* WaEnablePreemptionGranularityControlByUMD:cnl */
+	I915_WRITE(GEN7_FF_SLICE_CS_CHICKEN1,
+		   _MASKED_BIT_ENABLE(GEN9_FFSC_PERCTX_PREEMPT_CTRL));
 	ret= wa_ring_whitelist_reg(engine, GEN8_CS_CHICKEN1);
 	if (ret)
 		return ret;
@@ -1578,6 +1618,164 @@ bool intel_engine_can_store_dword(struct intel_engine_cs *engine)
 	}
 }
 
+static void print_request(struct drm_printer *m,
+			  struct drm_i915_gem_request *rq,
+			  const char *prefix)
+{
+	drm_printf(m, "%s%x [%x:%x] prio=%d @ %dms: %s\n", prefix,
+		   rq->global_seqno, rq->ctx->hw_id, rq->fence.seqno,
+		   rq->priotree.priority,
+		   jiffies_to_msecs(jiffies - rq->emitted_jiffies),
+		   rq->timeline->common->name);
+}
+
+void intel_engine_dump(struct intel_engine_cs *engine, struct drm_printer *m)
+{
+	struct intel_breadcrumbs *b = &engine->breadcrumbs;
+	struct i915_gpu_error *error = &engine->i915->gpu_error;
+	struct drm_i915_private *dev_priv = engine->i915;
+	struct drm_i915_gem_request *rq;
+	struct rb_node *rb;
+	u64 addr;
+
+	drm_printf(m, "%s\n", engine->name);
+	drm_printf(m, "\tcurrent seqno %x, last %x, hangcheck %x [%d ms], inflight %d\n",
+		   intel_engine_get_seqno(engine),
+		   intel_engine_last_submit(engine),
+		   engine->hangcheck.seqno,
+		   jiffies_to_msecs(jiffies - engine->hangcheck.action_timestamp),
+		   engine->timeline->inflight_seqnos);
+	drm_printf(m, "\tReset count: %d\n",
+		   i915_reset_engine_count(error, engine));
+
+	rcu_read_lock();
+
+	drm_printf(m, "\tRequests:\n");
+
+	rq = list_first_entry(&engine->timeline->requests,
+			      struct drm_i915_gem_request, link);
+	if (&rq->link != &engine->timeline->requests)
+		print_request(m, rq, "\t\tfirst  ");
+
+	rq = list_last_entry(&engine->timeline->requests,
+			     struct drm_i915_gem_request, link);
+	if (&rq->link != &engine->timeline->requests)
+		print_request(m, rq, "\t\tlast   ");
+
+	rq = i915_gem_find_active_request(engine);
+	if (rq) {
+		print_request(m, rq, "\t\tactive ");
+		drm_printf(m,
+			   "\t\t[head %04x, postfix %04x, tail %04x, batch 0x%08x_%08x]\n",
+			   rq->head, rq->postfix, rq->tail,
+			   rq->batch ? upper_32_bits(rq->batch->node.start) : ~0u,
+			   rq->batch ? lower_32_bits(rq->batch->node.start) : ~0u);
+	}
+
+	drm_printf(m, "\tRING_START: 0x%08x [0x%08x]\n",
+		   I915_READ(RING_START(engine->mmio_base)),
+		   rq ? i915_ggtt_offset(rq->ring->vma) : 0);
+	drm_printf(m, "\tRING_HEAD:  0x%08x [0x%08x]\n",
+		   I915_READ(RING_HEAD(engine->mmio_base)) & HEAD_ADDR,
+		   rq ? rq->ring->head : 0);
+	drm_printf(m, "\tRING_TAIL:  0x%08x [0x%08x]\n",
+		   I915_READ(RING_TAIL(engine->mmio_base)) & TAIL_ADDR,
+		   rq ? rq->ring->tail : 0);
+	drm_printf(m, "\tRING_CTL:   0x%08x [%s]\n",
+		   I915_READ(RING_CTL(engine->mmio_base)),
+		   I915_READ(RING_CTL(engine->mmio_base)) & (RING_WAIT | RING_WAIT_SEMAPHORE) ? "waiting" : "");
+
+	rcu_read_unlock();
+
+	addr = intel_engine_get_active_head(engine);
+	drm_printf(m, "\tACTHD:  0x%08x_%08x\n",
+		   upper_32_bits(addr), lower_32_bits(addr));
+	addr = intel_engine_get_last_batch_head(engine);
+	drm_printf(m, "\tBBADDR: 0x%08x_%08x\n",
+		   upper_32_bits(addr), lower_32_bits(addr));
+
+	if (i915_modparams.enable_execlists) {
+		const u32 *hws = &engine->status_page.page_addr[I915_HWS_CSB_BUF0_INDEX];
+		struct intel_engine_execlists * const execlists = &engine->execlists;
+		u32 ptr, read, write;
+		unsigned int idx;
+
+		drm_printf(m, "\tExeclist status: 0x%08x %08x\n",
+			   I915_READ(RING_EXECLIST_STATUS_LO(engine)),
+			   I915_READ(RING_EXECLIST_STATUS_HI(engine)));
+
+		ptr = I915_READ(RING_CONTEXT_STATUS_PTR(engine));
+		read = GEN8_CSB_READ_PTR(ptr);
+		write = GEN8_CSB_WRITE_PTR(ptr);
+		drm_printf(m, "\tExeclist CSB read %d [%d cached], write %d [%d from hws], interrupt posted? %s\n",
+			   read, execlists->csb_head,
+			   write,
+			   intel_read_status_page(engine, intel_hws_csb_write_index(engine->i915)),
+			   yesno(test_bit(ENGINE_IRQ_EXECLIST,
+					  &engine->irq_posted)));
+		if (read >= GEN8_CSB_ENTRIES)
+			read = 0;
+		if (write >= GEN8_CSB_ENTRIES)
+			write = 0;
+		if (read > write)
+			write += GEN8_CSB_ENTRIES;
+		while (read < write) {
+			idx = ++read % GEN8_CSB_ENTRIES;
+			drm_printf(m, "\tExeclist CSB[%d]: 0x%08x [0x%08x in hwsp], context: %d [%d in hwsp]\n",
+				   idx,
+				   I915_READ(RING_CONTEXT_STATUS_BUF_LO(engine, idx)),
+				   hws[idx * 2],
+				   I915_READ(RING_CONTEXT_STATUS_BUF_HI(engine, idx)),
+				   hws[idx * 2 + 1]);
+		}
+
+		rcu_read_lock();
+		for (idx = 0; idx < execlists_num_ports(execlists); idx++) {
+			unsigned int count;
+
+			rq = port_unpack(&execlists->port[idx], &count);
+			if (rq) {
+				drm_printf(m, "\t\tELSP[%d] count=%d, ",
+					   idx, count);
+				print_request(m, rq, "rq: ");
+			} else {
+				drm_printf(m, "\t\tELSP[%d] idle\n",
+					   idx);
+			}
+		}
+		rcu_read_unlock();
+
+		spin_lock_irq(&engine->timeline->lock);
+		for (rb = execlists->first; rb; rb = rb_next(rb)) {
+			struct i915_priolist *p =
+				rb_entry(rb, typeof(*p), node);
+
+			list_for_each_entry(rq, &p->requests,
+					    priotree.link)
+				print_request(m, rq, "\t\tQ ");
+		}
+		spin_unlock_irq(&engine->timeline->lock);
+	} else if (INTEL_GEN(dev_priv) > 6) {
+		drm_printf(m, "\tPP_DIR_BASE: 0x%08x\n",
+			   I915_READ(RING_PP_DIR_BASE(engine)));
+		drm_printf(m, "\tPP_DIR_BASE_READ: 0x%08x\n",
+			   I915_READ(RING_PP_DIR_BASE_READ(engine)));
+		drm_printf(m, "\tPP_DIR_DCLV: 0x%08x\n",
+			   I915_READ(RING_PP_DIR_DCLV(engine)));
+	}
+
+	spin_lock_irq(&b->rb_lock);
+	for (rb = rb_first(&b->waiters); rb; rb = rb_next(rb)) {
+		struct intel_wait *w = rb_entry(rb, typeof(*w), node);
+
+		drm_printf(m, "\t%s [%d] waiting for %x\n",
+			   w->tsk->comm, w->tsk->pid, w->seqno);
+	}
+	spin_unlock_irq(&b->rb_lock);
+
+	drm_printf(m, "\n");
+}
+
 #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
 #include "selftests/mock_engine.c"
 #endif
diff --git a/drivers/gpu/drm/i915/intel_guc.c b/drivers/gpu/drm/i915/intel_guc.c
new file mode 100644
index 000000000000..9e18c4fb9909
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_guc.c
@@ -0,0 +1,265 @@
+/*
+ * Copyright © 2014-2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#include "intel_guc.h"
+#include "i915_drv.h"
+
+static void gen8_guc_raise_irq(struct intel_guc *guc)
+{
+	struct drm_i915_private *dev_priv = guc_to_i915(guc);
+
+	I915_WRITE(GUC_SEND_INTERRUPT, GUC_SEND_TRIGGER);
+}
+
+static inline i915_reg_t guc_send_reg(struct intel_guc *guc, u32 i)
+{
+	GEM_BUG_ON(!guc->send_regs.base);
+	GEM_BUG_ON(!guc->send_regs.count);
+	GEM_BUG_ON(i >= guc->send_regs.count);
+
+	return _MMIO(guc->send_regs.base + 4 * i);
+}
+
+void intel_guc_init_send_regs(struct intel_guc *guc)
+{
+	struct drm_i915_private *dev_priv = guc_to_i915(guc);
+	enum forcewake_domains fw_domains = 0;
+	unsigned int i;
+
+	guc->send_regs.base = i915_mmio_reg_offset(SOFT_SCRATCH(0));
+	guc->send_regs.count = SOFT_SCRATCH_COUNT - 1;
+
+	for (i = 0; i < guc->send_regs.count; i++) {
+		fw_domains |= intel_uncore_forcewake_for_reg(dev_priv,
+					guc_send_reg(guc, i),
+					FW_REG_READ | FW_REG_WRITE);
+	}
+	guc->send_regs.fw_domains = fw_domains;
+}
+
+void intel_guc_init_early(struct intel_guc *guc)
+{
+	intel_guc_ct_init_early(&guc->ct);
+
+	mutex_init(&guc->send_mutex);
+	guc->send = intel_guc_send_nop;
+	guc->notify = gen8_guc_raise_irq;
+}
+
+int intel_guc_send_nop(struct intel_guc *guc, const u32 *action, u32 len)
+{
+	WARN(1, "Unexpected send: action=%#x\n", *action);
+	return -ENODEV;
+}
+
+/*
+ * This function implements the MMIO based host to GuC interface.
+ */
+int intel_guc_send_mmio(struct intel_guc *guc, const u32 *action, u32 len)
+{
+	struct drm_i915_private *dev_priv = guc_to_i915(guc);
+	u32 status;
+	int i;
+	int ret;
+
+	GEM_BUG_ON(!len);
+	GEM_BUG_ON(len > guc->send_regs.count);
+
+	/* If CT is available, we expect to use MMIO only during init/fini */
+	GEM_BUG_ON(HAS_GUC_CT(dev_priv) &&
+		*action != INTEL_GUC_ACTION_REGISTER_COMMAND_TRANSPORT_BUFFER &&
+		*action != INTEL_GUC_ACTION_DEREGISTER_COMMAND_TRANSPORT_BUFFER);
+
+	mutex_lock(&guc->send_mutex);
+	intel_uncore_forcewake_get(dev_priv, guc->send_regs.fw_domains);
+
+	for (i = 0; i < len; i++)
+		I915_WRITE(guc_send_reg(guc, i), action[i]);
+
+	POSTING_READ(guc_send_reg(guc, i - 1));
+
+	intel_guc_notify(guc);
+
+	/*
+	 * No GuC command should ever take longer than 10ms.
+	 * Fast commands should still complete in 10us.
+	 */
+	ret = __intel_wait_for_register_fw(dev_priv,
+					   guc_send_reg(guc, 0),
+					   INTEL_GUC_RECV_MASK,
+					   INTEL_GUC_RECV_MASK,
+					   10, 10, &status);
+	if (status != INTEL_GUC_STATUS_SUCCESS) {
+		/*
+		 * Either the GuC explicitly returned an error (which
+		 * we convert to -EIO here) or no response at all was
+		 * received within the timeout limit (-ETIMEDOUT)
+		 */
+		if (ret != -ETIMEDOUT)
+			ret = -EIO;
+
+		DRM_WARN("INTEL_GUC_SEND: Action 0x%X failed;"
+			 " ret=%d status=0x%08X response=0x%08X\n",
+			 action[0], ret, status, I915_READ(SOFT_SCRATCH(15)));
+	}
+
+	intel_uncore_forcewake_put(dev_priv, guc->send_regs.fw_domains);
+	mutex_unlock(&guc->send_mutex);
+
+	return ret;
+}
+
+int intel_guc_sample_forcewake(struct intel_guc *guc)
+{
+	struct drm_i915_private *dev_priv = guc_to_i915(guc);
+	u32 action[2];
+
+	action[0] = INTEL_GUC_ACTION_SAMPLE_FORCEWAKE;
+	/* WaRsDisableCoarsePowerGating:skl,bxt */
+	if (!intel_rc6_enabled() ||
+	    NEEDS_WaRsDisableCoarsePowerGating(dev_priv))
+		action[1] = 0;
+	else
+		/* bit 0 and 1 are for Render and Media domain separately */
+		action[1] = GUC_FORCEWAKE_RENDER | GUC_FORCEWAKE_MEDIA;
+
+	return intel_guc_send(guc, action, ARRAY_SIZE(action));
+}
+
+/**
+ * intel_guc_auth_huc() - Send action to GuC to authenticate HuC ucode
+ * @guc: intel_guc structure
+ * @rsa_offset: rsa offset w.r.t ggtt base of huc vma
+ *
+ * Triggers a HuC firmware authentication request to the GuC via intel_guc_send
+ * INTEL_GUC_ACTION_AUTHENTICATE_HUC interface. This function is invoked by
+ * intel_huc_auth().
+ *
+ * Return:	non-zero code on error
+ */
+int intel_guc_auth_huc(struct intel_guc *guc, u32 rsa_offset)
+{
+	u32 action[] = {
+		INTEL_GUC_ACTION_AUTHENTICATE_HUC,
+		rsa_offset
+	};
+
+	return intel_guc_send(guc, action, ARRAY_SIZE(action));
+}
+
+/**
+ * intel_guc_suspend() - notify GuC entering suspend state
+ * @dev_priv:	i915 device private
+ */
+int intel_guc_suspend(struct drm_i915_private *dev_priv)
+{
+	struct intel_guc *guc = &dev_priv->guc;
+	struct i915_gem_context *ctx;
+	u32 data[3];
+
+	if (guc->fw.load_status != INTEL_UC_FIRMWARE_SUCCESS)
+		return 0;
+
+	gen9_disable_guc_interrupts(dev_priv);
+
+	ctx = dev_priv->kernel_context;
+
+	data[0] = INTEL_GUC_ACTION_ENTER_S_STATE;
+	/* any value greater than GUC_POWER_D0 */
+	data[1] = GUC_POWER_D1;
+	/* first page is shared data with GuC */
+	data[2] = guc_ggtt_offset(ctx->engine[RCS].state) +
+		  LRC_GUCSHR_PN * PAGE_SIZE;
+
+	return intel_guc_send(guc, data, ARRAY_SIZE(data));
+}
+
+/**
+ * intel_guc_resume() - notify GuC resuming from suspend state
+ * @dev_priv:	i915 device private
+ */
+int intel_guc_resume(struct drm_i915_private *dev_priv)
+{
+	struct intel_guc *guc = &dev_priv->guc;
+	struct i915_gem_context *ctx;
+	u32 data[3];
+
+	if (guc->fw.load_status != INTEL_UC_FIRMWARE_SUCCESS)
+		return 0;
+
+	if (i915_modparams.guc_log_level >= 0)
+		gen9_enable_guc_interrupts(dev_priv);
+
+	ctx = dev_priv->kernel_context;
+
+	data[0] = INTEL_GUC_ACTION_EXIT_S_STATE;
+	data[1] = GUC_POWER_D0;
+	/* first page is shared data with GuC */
+	data[2] = guc_ggtt_offset(ctx->engine[RCS].state) +
+		  LRC_GUCSHR_PN * PAGE_SIZE;
+
+	return intel_guc_send(guc, data, ARRAY_SIZE(data));
+}
+
+/**
+ * intel_guc_allocate_vma() - Allocate a GGTT VMA for GuC usage
+ * @guc:	the guc
+ * @size:	size of area to allocate (both virtual space and memory)
+ *
+ * This is a wrapper to create an object for use with the GuC. In order to
+ * use it inside the GuC, an object needs to be pinned lifetime, so we allocate
+ * both some backing storage and a range inside the Global GTT. We must pin
+ * it in the GGTT somewhere other than than [0, GUC_WOPCM_TOP) because that
+ * range is reserved inside GuC.
+ *
+ * Return:	A i915_vma if successful, otherwise an ERR_PTR.
+ */
+struct i915_vma *intel_guc_allocate_vma(struct intel_guc *guc, u32 size)
+{
+	struct drm_i915_private *dev_priv = guc_to_i915(guc);
+	struct drm_i915_gem_object *obj;
+	struct i915_vma *vma;
+	int ret;
+
+	obj = i915_gem_object_create(dev_priv, size);
+	if (IS_ERR(obj))
+		return ERR_CAST(obj);
+
+	vma = i915_vma_instance(obj, &dev_priv->ggtt.base, NULL);
+	if (IS_ERR(vma))
+		goto err;
+
+	ret = i915_vma_pin(vma, 0, PAGE_SIZE,
+			   PIN_GLOBAL | PIN_OFFSET_BIAS | GUC_WOPCM_TOP);
+	if (ret) {
+		vma = ERR_PTR(ret);
+		goto err;
+	}
+
+	return vma;
+
+err:
+	i915_gem_object_put(obj);
+	return vma;
+}
diff --git a/drivers/gpu/drm/i915/intel_guc.h b/drivers/gpu/drm/i915/intel_guc.h
new file mode 100644
index 000000000000..aa9a7b55be6e
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_guc.h
@@ -0,0 +1,110 @@
+/*
+ * Copyright © 2014-2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _INTEL_GUC_H_
+#define _INTEL_GUC_H_
+
+#include "intel_uncore.h"
+#include "intel_guc_fwif.h"
+#include "intel_guc_ct.h"
+#include "intel_guc_log.h"
+#include "intel_uc_fw.h"
+#include "i915_guc_reg.h"
+#include "i915_vma.h"
+
+struct intel_guc {
+	struct intel_uc_fw fw;
+	struct intel_guc_log log;
+	struct intel_guc_ct ct;
+
+	/* Log snapshot if GuC errors during load */
+	struct drm_i915_gem_object *load_err_log;
+
+	/* intel_guc_recv interrupt related state */
+	bool interrupts_enabled;
+
+	struct i915_vma *ads_vma;
+	struct i915_vma *stage_desc_pool;
+	void *stage_desc_pool_vaddr;
+	struct ida stage_ids;
+
+	struct i915_guc_client *execbuf_client;
+
+	DECLARE_BITMAP(doorbell_bitmap, GUC_NUM_DOORBELLS);
+	/* Cyclic counter mod pagesize	*/
+	u32 db_cacheline;
+
+	/* GuC's FW specific registers used in MMIO send */
+	struct {
+		u32 base;
+		unsigned int count;
+		enum forcewake_domains fw_domains;
+	} send_regs;
+
+	/* To serialize the intel_guc_send actions */
+	struct mutex send_mutex;
+
+	/* GuC's FW specific send function */
+	int (*send)(struct intel_guc *guc, const u32 *data, u32 len);
+
+	/* GuC's FW specific notify function */
+	void (*notify)(struct intel_guc *guc);
+};
+
+static
+inline int intel_guc_send(struct intel_guc *guc, const u32 *action, u32 len)
+{
+	return guc->send(guc, action, len);
+}
+
+static inline void intel_guc_notify(struct intel_guc *guc)
+{
+	guc->notify(guc);
+}
+
+static inline u32 guc_ggtt_offset(struct i915_vma *vma)
+{
+	u32 offset = i915_ggtt_offset(vma);
+
+	GEM_BUG_ON(offset < GUC_WOPCM_TOP);
+	GEM_BUG_ON(range_overflows_t(u64, offset, vma->size, GUC_GGTT_TOP));
+
+	return offset;
+}
+
+void intel_guc_init_early(struct intel_guc *guc);
+void intel_guc_init_send_regs(struct intel_guc *guc);
+int intel_guc_send_nop(struct intel_guc *guc, const u32 *action, u32 len);
+int intel_guc_send_mmio(struct intel_guc *guc, const u32 *action, u32 len);
+int intel_guc_sample_forcewake(struct intel_guc *guc);
+int intel_guc_auth_huc(struct intel_guc *guc, u32 rsa_offset);
+int intel_guc_suspend(struct drm_i915_private *dev_priv);
+int intel_guc_resume(struct drm_i915_private *dev_priv);
+struct i915_vma *intel_guc_allocate_vma(struct intel_guc *guc, u32 size);
+
+int intel_guc_select_fw(struct intel_guc *guc);
+int intel_guc_init_hw(struct intel_guc *guc);
+u32 intel_guc_wopcm_size(struct drm_i915_private *dev_priv);
+
+#endif
diff --git a/drivers/gpu/drm/i915/intel_guc_fwif.h b/drivers/gpu/drm/i915/intel_guc_fwif.h
index 7eb6b4fa1d6f..1c0a2a3de121 100644
--- a/drivers/gpu/drm/i915/intel_guc_fwif.h
+++ b/drivers/gpu/drm/i915/intel_guc_fwif.h
@@ -178,49 +178,49 @@
  */
 
 struct uc_css_header {
-	uint32_t module_type;
+	u32 module_type;
 	/* header_size includes all non-uCode bits, including css_header, rsa
 	 * key, modulus key and exponent data. */
-	uint32_t header_size_dw;
-	uint32_t header_version;
-	uint32_t module_id;
-	uint32_t module_vendor;
+	u32 header_size_dw;
+	u32 header_version;
+	u32 module_id;
+	u32 module_vendor;
 	union {
 		struct {
-			uint8_t day;
-			uint8_t month;
-			uint16_t year;
+			u8 day;
+			u8 month;
+			u16 year;
 		};
-		uint32_t date;
+		u32 date;
 	};
-	uint32_t size_dw; /* uCode plus header_size_dw */
-	uint32_t key_size_dw;
-	uint32_t modulus_size_dw;
-	uint32_t exponent_size_dw;
+	u32 size_dw; /* uCode plus header_size_dw */
+	u32 key_size_dw;
+	u32 modulus_size_dw;
+	u32 exponent_size_dw;
 	union {
 		struct {
-			uint8_t hour;
-			uint8_t min;
-			uint16_t sec;
+			u8 hour;
+			u8 min;
+			u16 sec;
 		};
-		uint32_t time;
+		u32 time;
 	};
 
 	char username[8];
 	char buildnumber[12];
 	union {
 		struct {
-			uint32_t branch_client_version;
-			uint32_t sw_version;
+			u32 branch_client_version;
+			u32 sw_version;
 	} guc;
 		struct {
-			uint32_t sw_version;
-			uint32_t reserved;
+			u32 sw_version;
+			u32 reserved;
 	} huc;
 	};
-	uint32_t prod_preprod_fw;
-	uint32_t reserved[12];
-	uint32_t header_info;
+	u32 prod_preprod_fw;
+	u32 reserved[12];
+	u32 header_info;
 } __packed;
 
 struct guc_doorbell_info {
diff --git a/drivers/gpu/drm/i915/intel_guc_loader.c b/drivers/gpu/drm/i915/intel_guc_loader.c
index c9e25be4db40..c7a800a3798d 100644
--- a/drivers/gpu/drm/i915/intel_guc_loader.c
+++ b/drivers/gpu/drm/i915/intel_guc_loader.c
@@ -386,10 +386,7 @@ int intel_guc_select_fw(struct intel_guc *guc)
 {
 	struct drm_i915_private *dev_priv = guc_to_i915(guc);
 
-	guc->fw.path = NULL;
-	guc->fw.fetch_status = INTEL_UC_FIRMWARE_NONE;
-	guc->fw.load_status = INTEL_UC_FIRMWARE_NONE;
-	guc->fw.type = INTEL_UC_FW_TYPE_GUC;
+	intel_uc_fw_init(&guc->fw, INTEL_UC_FW_TYPE_GUC);
 
 	if (i915_modparams.guc_firmware_path) {
 		guc->fw.path = i915_modparams.guc_firmware_path;
diff --git a/drivers/gpu/drm/i915/intel_guc_log.c b/drivers/gpu/drm/i915/intel_guc_log.c
index 6571d96704ad..76d3eb1e4614 100644
--- a/drivers/gpu/drm/i915/intel_guc_log.c
+++ b/drivers/gpu/drm/i915/intel_guc_log.c
@@ -21,8 +21,11 @@
  * IN THE SOFTWARE.
  *
  */
+
 #include <linux/debugfs.h>
 #include <linux/relay.h>
+
+#include "intel_guc_log.h"
 #include "i915_drv.h"
 
 static void guc_log_capture_logs(struct intel_guc *guc);
@@ -525,7 +528,8 @@ int intel_guc_log_create(struct intel_guc *guc)
 {
 	struct i915_vma *vma;
 	unsigned long offset;
-	uint32_t size, flags;
+	u32 flags;
+	u32 size;
 	int ret;
 
 	GEM_BUG_ON(guc->log.vma);
diff --git a/drivers/gpu/drm/i915/intel_guc_log.h b/drivers/gpu/drm/i915/intel_guc_log.h
new file mode 100644
index 000000000000..f512cf79339b
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_guc_log.h
@@ -0,0 +1,59 @@
+/*
+ * Copyright © 2014-2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _INTEL_GUC_LOG_H_
+#define _INTEL_GUC_LOG_H_
+
+#include <linux/workqueue.h>
+
+#include "intel_guc_fwif.h"
+
+struct drm_i915_private;
+struct intel_guc;
+
+struct intel_guc_log {
+	u32 flags;
+	struct i915_vma *vma;
+	/* The runtime stuff gets created only when GuC logging gets enabled */
+	struct {
+		void *buf_addr;
+		struct workqueue_struct *flush_wq;
+		struct work_struct flush_work;
+		struct rchan *relay_chan;
+	} runtime;
+	/* logging related stats */
+	u32 capture_miss_count;
+	u32 flush_interrupt_count;
+	u32 prev_overflow_count[GUC_MAX_LOG_BUFFER];
+	u32 total_overflow_count[GUC_MAX_LOG_BUFFER];
+	u32 flush_count[GUC_MAX_LOG_BUFFER];
+};
+
+int intel_guc_log_create(struct intel_guc *guc);
+void intel_guc_log_destroy(struct intel_guc *guc);
+int i915_guc_log_control(struct drm_i915_private *dev_priv, u64 control_val);
+void i915_guc_log_register(struct drm_i915_private *dev_priv);
+void i915_guc_log_unregister(struct drm_i915_private *dev_priv);
+
+#endif
diff --git a/drivers/gpu/drm/i915/intel_huc.c b/drivers/gpu/drm/i915/intel_huc.c
index 8b4b53525422..4b4cf56d29ad 100644
--- a/drivers/gpu/drm/i915/intel_huc.c
+++ b/drivers/gpu/drm/i915/intel_huc.c
@@ -21,9 +21,11 @@
  * IN THE SOFTWARE.
  *
  */
-#include <linux/firmware.h>
+
+#include <linux/types.h>
+
+#include "intel_huc.h"
 #include "i915_drv.h"
-#include "intel_uc.h"
 
 /**
  * DOC: HuC Firmware
@@ -150,10 +152,7 @@ void intel_huc_select_fw(struct intel_huc *huc)
 {
 	struct drm_i915_private *dev_priv = huc_to_i915(huc);
 
-	huc->fw.path = NULL;
-	huc->fw.fetch_status = INTEL_UC_FIRMWARE_NONE;
-	huc->fw.load_status = INTEL_UC_FIRMWARE_NONE;
-	huc->fw.type = INTEL_UC_FW_TYPE_HUC;
+	intel_uc_fw_init(&huc->fw, INTEL_UC_FW_TYPE_HUC);
 
 	if (i915_modparams.huc_firmware_path) {
 		huc->fw.path = i915_modparams.huc_firmware_path;
diff --git a/drivers/gpu/drm/i915/intel_huc.h b/drivers/gpu/drm/i915/intel_huc.h
new file mode 100644
index 000000000000..aaa38b9e5817
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_huc.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright © 2014-2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _INTEL_HUC_H_
+#define _INTEL_HUC_H_
+
+#include "intel_uc_fw.h"
+
+struct intel_huc {
+	/* Generic uC firmware management */
+	struct intel_uc_fw fw;
+
+	/* HuC-specific additions */
+};
+
+void intel_huc_select_fw(struct intel_huc *huc);
+void intel_huc_init_hw(struct intel_huc *huc);
+void intel_huc_auth(struct intel_huc *huc);
+
+#endif
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 61cac26a8b05..fbfcf88d7fe3 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -208,8 +208,9 @@
 
 /* Typical size of the average request (2 pipecontrols and a MI_BB) */
 #define EXECLISTS_REQUEST_SIZE 64 /* bytes */
-
 #define WA_TAIL_DWORDS 2
+#define WA_TAIL_BYTES (sizeof(u32) * WA_TAIL_DWORDS)
+#define PREEMPT_ID 0x1
 
 static int execlists_context_deferred_alloc(struct i915_gem_context *ctx,
 					    struct intel_engine_cs *engine);
@@ -243,8 +244,7 @@ int intel_sanitize_enable_execlists(struct drm_i915_private *dev_priv, int enabl
 		return 0;
 
 	if (HAS_LOGICAL_RING_CONTEXTS(dev_priv) &&
-	    USES_PPGTT(dev_priv) &&
-	    i915_modparams.use_mmio_flip >= 0)
+	    USES_PPGTT(dev_priv))
 		return 1;
 
 	return 0;
@@ -348,6 +348,43 @@ find_priolist:
 	return ptr_pack_bits(p, first, 1);
 }
 
+static void unwind_wa_tail(struct drm_i915_gem_request *rq)
+{
+	rq->tail = intel_ring_wrap(rq->ring, rq->wa_tail - WA_TAIL_BYTES);
+	assert_ring_tail_valid(rq->ring, rq->tail);
+}
+
+static void unwind_incomplete_requests(struct intel_engine_cs *engine)
+{
+	struct drm_i915_gem_request *rq, *rn;
+	struct i915_priolist *uninitialized_var(p);
+	int last_prio = I915_PRIORITY_INVALID;
+
+	lockdep_assert_held(&engine->timeline->lock);
+
+	list_for_each_entry_safe_reverse(rq, rn,
+					 &engine->timeline->requests,
+					 link) {
+		if (i915_gem_request_completed(rq))
+			return;
+
+		__i915_gem_request_unsubmit(rq);
+		unwind_wa_tail(rq);
+
+		GEM_BUG_ON(rq->priotree.priority == I915_PRIORITY_INVALID);
+		if (rq->priotree.priority != last_prio) {
+			p = lookup_priolist(engine,
+					    &rq->priotree,
+					    rq->priotree.priority);
+			p = ptr_mask_bits(p, 1);
+
+			last_prio = rq->priotree.priority;
+		}
+
+		list_add(&rq->priotree.link, &p->requests);
+	}
+}
+
 static inline void
 execlists_context_status_change(struct drm_i915_gem_request *rq,
 				unsigned long status)
@@ -392,6 +429,12 @@ static u64 execlists_update_context(struct drm_i915_gem_request *rq)
 	return ce->lrc_desc;
 }
 
+static inline void elsp_write(u64 desc, u32 __iomem *elsp)
+{
+	writel(upper_32_bits(desc), elsp);
+	writel(lower_32_bits(desc), elsp);
+}
+
 static void execlists_submit_ports(struct intel_engine_cs *engine)
 {
 	struct execlist_port *port = engine->execlists.port;
@@ -417,8 +460,7 @@ static void execlists_submit_ports(struct intel_engine_cs *engine)
 			desc = 0;
 		}
 
-		writel(upper_32_bits(desc), elsp);
-		writel(lower_32_bits(desc), elsp);
+		elsp_write(desc, elsp);
 	}
 }
 
@@ -451,26 +493,43 @@ static void port_assign(struct execlist_port *port,
 	port_set(port, port_pack(i915_gem_request_get(rq), port_count(port)));
 }
 
+static void inject_preempt_context(struct intel_engine_cs *engine)
+{
+	struct intel_context *ce =
+		&engine->i915->preempt_context->engine[engine->id];
+	u32 __iomem *elsp =
+		engine->i915->regs + i915_mmio_reg_offset(RING_ELSP(engine));
+	unsigned int n;
+
+	GEM_BUG_ON(engine->i915->preempt_context->hw_id != PREEMPT_ID);
+	GEM_BUG_ON(!IS_ALIGNED(ce->ring->size, WA_TAIL_BYTES));
+
+	memset(ce->ring->vaddr + ce->ring->tail, 0, WA_TAIL_BYTES);
+	ce->ring->tail += WA_TAIL_BYTES;
+	ce->ring->tail &= (ce->ring->size - 1);
+	ce->lrc_reg_state[CTX_RING_TAIL+1] = ce->ring->tail;
+
+	for (n = execlists_num_ports(&engine->execlists); --n; )
+		elsp_write(0, elsp);
+
+	elsp_write(ce->lrc_desc, elsp);
+}
+
+static bool can_preempt(struct intel_engine_cs *engine)
+{
+	return INTEL_INFO(engine->i915)->has_logical_ring_preemption;
+}
+
 static void execlists_dequeue(struct intel_engine_cs *engine)
 {
-	struct drm_i915_gem_request *last;
 	struct intel_engine_execlists * const execlists = &engine->execlists;
 	struct execlist_port *port = execlists->port;
 	const struct execlist_port * const last_port =
 		&execlists->port[execlists->port_mask];
+	struct drm_i915_gem_request *last = port_request(port);
 	struct rb_node *rb;
 	bool submit = false;
 
-	last = port_request(port);
-	if (last)
-		/* WaIdleLiteRestore:bdw,skl
-		 * Apply the wa NOOPs to prevent ring:HEAD == req:TAIL
-		 * as we resubmit the request. See gen8_emit_breadcrumb()
-		 * for where we prepare the padding after the end of the
-		 * request.
-		 */
-		last->tail = last->wa_tail;
-
 	/* Hardware submission is through 2 ports. Conceptually each port
 	 * has a (RING_START, RING_HEAD, RING_TAIL) tuple. RING_START is
 	 * static for a context, and unique to each, so we only execute
@@ -495,7 +554,65 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
 	spin_lock_irq(&engine->timeline->lock);
 	rb = execlists->first;
 	GEM_BUG_ON(rb_first(&execlists->queue) != rb);
-	while (rb) {
+	if (!rb)
+		goto unlock;
+
+	if (last) {
+		/*
+		 * Don't resubmit or switch until all outstanding
+		 * preemptions (lite-restore) are seen. Then we
+		 * know the next preemption status we see corresponds
+		 * to this ELSP update.
+		 */
+		if (port_count(&port[0]) > 1)
+			goto unlock;
+
+		if (can_preempt(engine) &&
+		    rb_entry(rb, struct i915_priolist, node)->priority >
+		    max(last->priotree.priority, 0)) {
+			/*
+			 * Switch to our empty preempt context so
+			 * the state of the GPU is known (idle).
+			 */
+			inject_preempt_context(engine);
+			execlists->preempt = true;
+			goto unlock;
+		} else {
+			/*
+			 * In theory, we could coalesce more requests onto
+			 * the second port (the first port is active, with
+			 * no preemptions pending). However, that means we
+			 * then have to deal with the possible lite-restore
+			 * of the second port (as we submit the ELSP, there
+			 * may be a context-switch) but also we may complete
+			 * the resubmission before the context-switch. Ergo,
+			 * coalescing onto the second port will cause a
+			 * preemption event, but we cannot predict whether
+			 * that will affect port[0] or port[1].
+			 *
+			 * If the second port is already active, we can wait
+			 * until the next context-switch before contemplating
+			 * new requests. The GPU will be busy and we should be
+			 * able to resubmit the new ELSP before it idles,
+			 * avoiding pipeline bubbles (momentary pauses where
+			 * the driver is unable to keep up the supply of new
+			 * work).
+			 */
+			if (port_count(&port[1]))
+				goto unlock;
+
+			/* WaIdleLiteRestore:bdw,skl
+			 * Apply the wa NOOPs to prevent
+			 * ring:HEAD == req:TAIL as we resubmit the
+			 * request. See gen8_emit_breadcrumb() for
+			 * where we prepare the padding after the
+			 * end of the request.
+			 */
+			last->tail = last->wa_tail;
+		}
+	}
+
+	do {
 		struct i915_priolist *p = rb_entry(rb, typeof(*p), node);
 		struct drm_i915_gem_request *rq, *rn;
 
@@ -547,8 +664,6 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
 			}
 
 			INIT_LIST_HEAD(&rq->priotree.link);
-			rq->priotree.priority = INT_MAX;
-
 			__i915_gem_request_submit(rq);
 			trace_i915_gem_request_in(rq, port_index(port, execlists));
 			last = rq;
@@ -560,11 +675,12 @@ static void execlists_dequeue(struct intel_engine_cs *engine)
 		INIT_LIST_HEAD(&p->requests);
 		if (p->priority != I915_PRIORITY_NORMAL)
 			kmem_cache_free(engine->i915->priorities, p);
-	}
+	} while (rb);
 done:
 	execlists->first = rb;
 	if (submit)
 		port_assign(port, last);
+unlock:
 	spin_unlock_irq(&engine->timeline->lock);
 
 	if (submit)
@@ -575,12 +691,12 @@ static void
 execlist_cancel_port_requests(struct intel_engine_execlists *execlists)
 {
 	struct execlist_port *port = execlists->port;
-	unsigned int num_ports = ARRAY_SIZE(execlists->port);
+	unsigned int num_ports = execlists_num_ports(execlists);
 
 	while (num_ports-- && port_isset(port)) {
 		struct drm_i915_gem_request *rq = port_request(port);
 
-		execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_OUT);
+		execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_PREEMPTED);
 		i915_gem_request_put(rq);
 
 		memset(port, 0, sizeof(*port));
@@ -645,13 +761,6 @@ static void execlists_cancel_requests(struct intel_engine_cs *engine)
 	spin_unlock_irqrestore(&engine->timeline->lock, flags);
 }
 
-static bool execlists_elsp_ready(const struct intel_engine_cs *engine)
-{
-	const struct execlist_port *port = engine->execlists.port;
-
-	return port_count(&port[0]) + port_count(&port[1]) < 2;
-}
-
 /*
  * Check the unread Context Status Buffers and manage the submission of new
  * contexts to the ELSP accordingly.
@@ -660,7 +769,7 @@ static void intel_lrc_irq_handler(unsigned long data)
 {
 	struct intel_engine_cs * const engine = (struct intel_engine_cs *)data;
 	struct intel_engine_execlists * const execlists = &engine->execlists;
-	struct execlist_port *port = execlists->port;
+	struct execlist_port * const port = execlists->port;
 	struct drm_i915_private *dev_priv = engine->i915;
 
 	/* We can skip acquiring intel_runtime_pm_get() here as it was taken
@@ -745,6 +854,23 @@ static void intel_lrc_irq_handler(unsigned long data)
 			if (!(status & GEN8_CTX_STATUS_COMPLETED_MASK))
 				continue;
 
+			if (status & GEN8_CTX_STATUS_ACTIVE_IDLE &&
+			    buf[2*head + 1] == PREEMPT_ID) {
+				execlist_cancel_port_requests(execlists);
+
+				spin_lock_irq(&engine->timeline->lock);
+				unwind_incomplete_requests(engine);
+				spin_unlock_irq(&engine->timeline->lock);
+
+				GEM_BUG_ON(!execlists->preempt);
+				execlists->preempt = false;
+				continue;
+			}
+
+			if (status & GEN8_CTX_STATUS_PREEMPTED &&
+			    execlists->preempt)
+				continue;
+
 			/* Check the context/desc id for this event matches */
 			GEM_DEBUG_BUG_ON(buf[2 * head + 1] != port->context_id);
 
@@ -756,6 +882,7 @@ static void intel_lrc_irq_handler(unsigned long data)
 				execlists_context_status_change(rq, INTEL_CONTEXT_SCHEDULE_OUT);
 
 				trace_i915_gem_request_out(rq);
+				rq->priotree.priority = INT_MAX;
 				i915_gem_request_put(rq);
 
 				execlists_port_complete(execlists, port);
@@ -775,7 +902,7 @@ static void intel_lrc_irq_handler(unsigned long data)
 		}
 	}
 
-	if (execlists_elsp_ready(engine))
+	if (!execlists->preempt)
 		execlists_dequeue(engine);
 
 	intel_uncore_forcewake_put(dev_priv, execlists->fw_domains);
@@ -788,7 +915,7 @@ static void insert_request(struct intel_engine_cs *engine,
 	struct i915_priolist *p = lookup_priolist(engine, pt, prio);
 
 	list_add_tail(&pt->link, &ptr_mask_bits(p, 1)->requests);
-	if (ptr_unmask_bits(p, 1) && execlists_elsp_ready(engine))
+	if (ptr_unmask_bits(p, 1))
 		tasklet_hi_schedule(&engine->execlists.irq_tasklet);
 }
 
@@ -808,11 +935,15 @@ static void execlists_submit_request(struct drm_i915_gem_request *request)
 	spin_unlock_irqrestore(&engine->timeline->lock, flags);
 }
 
+static struct drm_i915_gem_request *pt_to_request(struct i915_priotree *pt)
+{
+	return container_of(pt, struct drm_i915_gem_request, priotree);
+}
+
 static struct intel_engine_cs *
 pt_lock_engine(struct i915_priotree *pt, struct intel_engine_cs *locked)
 {
-	struct intel_engine_cs *engine =
-		container_of(pt, struct drm_i915_gem_request, priotree)->engine;
+	struct intel_engine_cs *engine = pt_to_request(pt)->engine;
 
 	GEM_BUG_ON(!locked);
 
@@ -831,6 +962,8 @@ static void execlists_schedule(struct drm_i915_gem_request *request, int prio)
 	struct i915_dependency stack;
 	LIST_HEAD(dfs);
 
+	GEM_BUG_ON(prio == I915_PRIORITY_INVALID);
+
 	if (prio <= READ_ONCE(request->priotree.priority))
 		return;
 
@@ -866,6 +999,9 @@ static void execlists_schedule(struct drm_i915_gem_request *request, int prio)
 		 * engines.
 		 */
 		list_for_each_entry(p, &pt->signalers_list, signal_link) {
+			if (i915_gem_request_completed(pt_to_request(p->signaler)))
+				continue;
+
 			GEM_BUG_ON(p->signaler->priority < pt->priority);
 			if (prio > READ_ONCE(p->signaler->priority))
 				list_move_tail(&p->dfs_link, &dfs);
@@ -879,7 +1015,7 @@ static void execlists_schedule(struct drm_i915_gem_request *request, int prio)
 	 * execlists_submit_request()), we can set our own priority and skip
 	 * acquiring the engine locks.
 	 */
-	if (request->priotree.priority == INT_MIN) {
+	if (request->priotree.priority == I915_PRIORITY_INVALID) {
 		GEM_BUG_ON(!list_empty(&request->priotree.link));
 		request->priotree.priority = prio;
 		if (stack.dfs_link.next == stack.dfs_link.prev)
@@ -909,8 +1045,6 @@ static void execlists_schedule(struct drm_i915_gem_request *request, int prio)
 	}
 
 	spin_unlock_irq(&engine->timeline->lock);
-
-	/* XXX Do we need to preempt to make room for us and our deps? */
 }
 
 static struct intel_ring *
@@ -1106,6 +1240,8 @@ static u32 *gen8_init_indirectctx_bb(struct intel_engine_cs *engine, u32 *batch)
 				       i915_ggtt_offset(engine->scratch) +
 				       2 * CACHELINE_BYTES);
 
+	*batch++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
+
 	/* Pad to end of cacheline */
 	while ((unsigned long)batch % CACHELINE_BYTES)
 		*batch++ = MI_NOOP;
@@ -1119,26 +1255,10 @@ static u32 *gen8_init_indirectctx_bb(struct intel_engine_cs *engine, u32 *batch)
 	return batch;
 }
 
-/*
- *  This batch is started immediately after indirect_ctx batch. Since we ensure
- *  that indirect_ctx ends on a cacheline this batch is aligned automatically.
- *
- *  The number of DWORDS written are returned using this field.
- *
- *  This batch is terminated with MI_BATCH_BUFFER_END and so we need not add padding
- *  to align it with cacheline as padding after MI_BATCH_BUFFER_END is redundant.
- */
-static u32 *gen8_init_perctx_bb(struct intel_engine_cs *engine, u32 *batch)
-{
-	/* WaDisableCtxRestoreArbitration:bdw,chv */
-	*batch++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
-	*batch++ = MI_BATCH_BUFFER_END;
-
-	return batch;
-}
-
 static u32 *gen9_init_indirectctx_bb(struct intel_engine_cs *engine, u32 *batch)
 {
+	*batch++ = MI_ARB_ON_OFF | MI_ARB_DISABLE;
+
 	/* WaFlushCoherentL3CacheLinesAtContextSwitch:skl,bxt,glk */
 	batch = gen8_emit_flush_coherentl3_wa(engine, batch);
 
@@ -1184,6 +1304,8 @@ static u32 *gen9_init_indirectctx_bb(struct intel_engine_cs *engine, u32 *batch)
 		*batch++ = 0;
 	}
 
+	*batch++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
+
 	/* Pad to end of cacheline */
 	while ((unsigned long)batch % CACHELINE_BYTES)
 		*batch++ = MI_NOOP;
@@ -1251,7 +1373,7 @@ static int intel_init_workaround_bb(struct intel_engine_cs *engine)
 		break;
 	case 8:
 		wa_bb_fn[0] = gen8_init_indirectctx_bb;
-		wa_bb_fn[1] = gen8_init_perctx_bb;
+		wa_bb_fn[1] = NULL;
 		break;
 	default:
 		MISSING_CASE(INTEL_GEN(engine->i915));
@@ -1337,6 +1459,7 @@ static int gen8_init_common_ring(struct intel_engine_cs *engine)
 		   GT_CONTEXT_SWITCH_INTERRUPT << engine->irq_shift);
 	clear_bit(ENGINE_IRQ_EXECLIST, &engine->irq_posted);
 	execlists->csb_head = -1;
+	execlists->preempt = false;
 
 	/* After a GPU reset, we may have requests to replay */
 	if (!i915_modparams.enable_guc_submission && execlists->first)
@@ -1382,7 +1505,6 @@ static void reset_common_ring(struct intel_engine_cs *engine,
 			      struct drm_i915_gem_request *request)
 {
 	struct intel_engine_execlists * const execlists = &engine->execlists;
-	struct drm_i915_gem_request *rq, *rn;
 	struct intel_context *ce;
 	unsigned long flags;
 
@@ -1400,21 +1522,7 @@ static void reset_common_ring(struct intel_engine_cs *engine,
 	execlist_cancel_port_requests(execlists);
 
 	/* Push back any incomplete requests for replay after the reset. */
-	list_for_each_entry_safe_reverse(rq, rn,
-					 &engine->timeline->requests, link) {
-		struct i915_priolist *p;
-
-		if (i915_gem_request_completed(rq))
-			break;
-
-		__i915_gem_request_unsubmit(rq);
-
-		p = lookup_priolist(engine,
-				    &rq->priotree,
-				    rq->priotree.priority);
-		list_add(&rq->priotree.link,
-			 &ptr_mask_bits(p, 1)->requests);
-	}
+	unwind_incomplete_requests(engine);
 
 	spin_unlock_irqrestore(&engine->timeline->lock, flags);
 
@@ -1451,10 +1559,7 @@ static void reset_common_ring(struct intel_engine_cs *engine,
 	intel_ring_update_space(request->ring);
 
 	/* Reset WaIdleLiteRestore:bdw,skl as well */
-	request->tail =
-		intel_ring_wrap(request->ring,
-				request->wa_tail - WA_TAIL_DWORDS*sizeof(u32));
-	assert_ring_tail_valid(request->ring, request->tail);
+	unwind_wa_tail(request);
 }
 
 static int intel_logical_ring_emit_pdps(struct drm_i915_gem_request *req)
@@ -1513,13 +1618,31 @@ static int gen8_emit_bb_start(struct drm_i915_gem_request *req,
 	if (IS_ERR(cs))
 		return PTR_ERR(cs);
 
+	/*
+	 * WaDisableCtxRestoreArbitration:bdw,chv
+	 *
+	 * We don't need to perform MI_ARB_ENABLE as often as we do (in
+	 * particular all the gen that do not need the w/a at all!), if we
+	 * took care to make sure that on every switch into this context
+	 * (both ordinary and for preemption) that arbitrartion was enabled
+	 * we would be fine. However, there doesn't seem to be a downside to
+	 * being paranoid and making sure it is set before each batch and
+	 * every context-switch.
+	 *
+	 * Note that if we fail to enable arbitration before the request
+	 * is complete, then we do not see the context-switch interrupt and
+	 * the engine hangs (with RING_HEAD == RING_TAIL).
+	 *
+	 * That satisfies both the GPGPU w/a and our heavy-handed paranoia.
+	 */
+	*cs++ = MI_ARB_ON_OFF | MI_ARB_ENABLE;
+
 	/* FIXME(BDW): Address space and security selectors. */
 	*cs++ = MI_BATCH_BUFFER_START_GEN8 |
 		(flags & I915_DISPATCH_SECURE ? 0 : BIT(8)) |
 		(flags & I915_DISPATCH_RS ? MI_BATCH_RESOURCE_STREAMER : 0);
 	*cs++ = lower_32_bits(offset);
 	*cs++ = upper_32_bits(offset);
-	*cs++ = MI_NOOP;
 	intel_ring_advance(req, cs);
 
 	return 0;
@@ -1648,7 +1771,8 @@ static int gen8_emit_flush_render(struct drm_i915_gem_request *request,
  */
 static void gen8_emit_wa_tail(struct drm_i915_gem_request *request, u32 *cs)
 {
-	*cs++ = MI_NOOP;
+	/* Ensure there's always at least one preemption point per-request. */
+	*cs++ = MI_ARB_CHECK;
 	*cs++ = MI_NOOP;
 	request->wa_tail = intel_ring_offset(request, cs);
 }
@@ -1669,7 +1793,6 @@ static void gen8_emit_breadcrumb(struct drm_i915_gem_request *request, u32 *cs)
 
 	gen8_emit_wa_tail(request, cs);
 }
-
 static const int gen8_emit_breadcrumb_sz = 6 + WA_TAIL_DWORDS;
 
 static void gen8_emit_breadcrumb_render(struct drm_i915_gem_request *request,
@@ -1697,7 +1820,6 @@ static void gen8_emit_breadcrumb_render(struct drm_i915_gem_request *request,
 
 	gen8_emit_wa_tail(request, cs);
 }
-
 static const int gen8_emit_breadcrumb_render_sz = 8 + WA_TAIL_DWORDS;
 
 static int gen8_init_rcs_context(struct drm_i915_gem_request *req)
diff --git a/drivers/gpu/drm/i915/intel_lrc.h b/drivers/gpu/drm/i915/intel_lrc.h
index 314adee7127a..689fde1a63a9 100644
--- a/drivers/gpu/drm/i915/intel_lrc.h
+++ b/drivers/gpu/drm/i915/intel_lrc.h
@@ -61,6 +61,7 @@
 enum {
 	INTEL_CONTEXT_SCHEDULE_IN = 0,
 	INTEL_CONTEXT_SCHEDULE_OUT,
+	INTEL_CONTEXT_SCHEDULE_PREEMPTED,
 };
 
 /* Logical Rings */
diff --git a/drivers/gpu/drm/i915/intel_lvds.c b/drivers/gpu/drm/i915/intel_lvds.c
index a55954a89148..38572d65e46e 100644
--- a/drivers/gpu/drm/i915/intel_lvds.c
+++ b/drivers/gpu/drm/i915/intel_lvds.c
@@ -939,10 +939,8 @@ void intel_lvds_init(struct drm_i915_private *dev_priv)
 	struct drm_display_mode *fixed_mode = NULL;
 	struct drm_display_mode *downclock_mode = NULL;
 	struct edid *edid;
-	struct intel_crtc *crtc;
 	i915_reg_t lvds_reg;
 	u32 lvds;
-	int pipe;
 	u8 pin;
 	u32 allowed_scalers;
 
@@ -1113,22 +1111,11 @@ void intel_lvds_init(struct drm_i915_private *dev_priv)
 	 * on.  If so, assume that whatever is currently programmed is the
 	 * correct mode.
 	 */
-
-	/* Ironlake: FIXME if still fail, not try pipe mode now */
-	if (HAS_PCH_SPLIT(dev_priv))
-		goto failed;
-
-	pipe = (lvds & LVDS_PIPEB_SELECT) ? 1 : 0;
-	crtc = intel_get_crtc_for_pipe(dev_priv, pipe);
-
-	if (crtc && (lvds & LVDS_PORT_EN)) {
-		fixed_mode = intel_crtc_mode_get(dev, &crtc->base);
-		if (fixed_mode) {
-			DRM_DEBUG_KMS("using current (BIOS) mode: ");
-			drm_mode_debug_printmodeline(fixed_mode);
-			fixed_mode->type |= DRM_MODE_TYPE_PREFERRED;
-			goto out;
-		}
+	fixed_mode = intel_encoder_current_mode(intel_encoder);
+	if (fixed_mode) {
+		DRM_DEBUG_KMS("using current (BIOS) mode: ");
+		drm_mode_debug_printmodeline(fixed_mode);
+		fixed_mode->type |= DRM_MODE_TYPE_PREFERRED;
 	}
 
 	/* If we still don't have a mode after all that, give up. */
diff --git a/drivers/gpu/drm/i915/intel_pipe_crc.c b/drivers/gpu/drm/i915/intel_pipe_crc.c
index 96043a51c1bf..899839f2f7c6 100644
--- a/drivers/gpu/drm/i915/intel_pipe_crc.c
+++ b/drivers/gpu/drm/i915/intel_pipe_crc.c
@@ -206,11 +206,11 @@ static const char *pipe_crc_source_name(enum intel_pipe_crc_source source)
 static int display_crc_ctl_show(struct seq_file *m, void *data)
 {
 	struct drm_i915_private *dev_priv = m->private;
-	int i;
+	enum pipe pipe;
 
-	for (i = 0; i < I915_MAX_PIPES; i++)
-		seq_printf(m, "%c %s\n", pipe_name(i),
-			   pipe_crc_source_name(dev_priv->pipe_crc[i].source));
+	for_each_pipe(dev_priv, pipe)
+		seq_printf(m, "%c %s\n", pipe_name(pipe),
+			   pipe_crc_source_name(dev_priv->pipe_crc[pipe].source));
 
 	return 0;
 }
@@ -775,11 +775,12 @@ display_crc_ctl_parse_object(const char *buf, enum intel_pipe_crc_object *o)
 	return -EINVAL;
 }
 
-static int display_crc_ctl_parse_pipe(const char *buf, enum pipe *pipe)
+static int display_crc_ctl_parse_pipe(struct drm_i915_private *dev_priv,
+				      const char *buf, enum pipe *pipe)
 {
 	const char name = buf[0];
 
-	if (name < 'A' || name >= pipe_name(I915_MAX_PIPES))
+	if (name < 'A' || name >= pipe_name(INTEL_INFO(dev_priv)->num_pipes))
 		return -EINVAL;
 
 	*pipe = name - 'A';
@@ -828,7 +829,7 @@ static int display_crc_ctl_parse(struct drm_i915_private *dev_priv,
 		return -EINVAL;
 	}
 
-	if (display_crc_ctl_parse_pipe(words[1], &pipe) < 0) {
+	if (display_crc_ctl_parse_pipe(dev_priv, words[1], &pipe) < 0) {
 		DRM_DEBUG_DRIVER("unknown pipe %s\n", words[1]);
 		return -EINVAL;
 	}
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index c66af09e27a7..2fcff9788b6f 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -322,7 +322,7 @@ static void chv_set_memory_dvfs(struct drm_i915_private *dev_priv, bool enable)
 {
 	u32 val;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	val = vlv_punit_read(dev_priv, PUNIT_REG_DDR_SETUP2);
 	if (enable)
@@ -337,14 +337,14 @@ static void chv_set_memory_dvfs(struct drm_i915_private *dev_priv, bool enable)
 		      FORCE_DDR_FREQ_REQ_ACK) == 0, 3))
 		DRM_ERROR("timed out waiting for Punit DDR DVFS request\n");
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 }
 
 static void chv_set_memory_pm5(struct drm_i915_private *dev_priv, bool enable)
 {
 	u32 val;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	val = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ);
 	if (enable)
@@ -353,7 +353,7 @@ static void chv_set_memory_pm5(struct drm_i915_private *dev_priv, bool enable)
 		val &= ~DSP_MAXFIFO_PM5_ENABLE;
 	vlv_punit_write(dev_priv, PUNIT_REG_DSPFREQ, val);
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 }
 
 #define FW_WM(value, plane) \
@@ -2790,11 +2790,11 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv,
 
 		/* read the first set of memory latencies[0:3] */
 		val = 0; /* data0 to be programmed to 0 for first set */
-		mutex_lock(&dev_priv->rps.hw_lock);
+		mutex_lock(&dev_priv->pcu_lock);
 		ret = sandybridge_pcode_read(dev_priv,
 					     GEN9_PCODE_READ_MEM_LATENCY,
 					     &val);
-		mutex_unlock(&dev_priv->rps.hw_lock);
+		mutex_unlock(&dev_priv->pcu_lock);
 
 		if (ret) {
 			DRM_ERROR("SKL Mailbox read error = %d\n", ret);
@@ -2811,11 +2811,11 @@ static void intel_read_wm_latency(struct drm_i915_private *dev_priv,
 
 		/* read the second set of memory latencies[4:7] */
 		val = 1; /* data0 to be programmed to 1 for second set */
-		mutex_lock(&dev_priv->rps.hw_lock);
+		mutex_lock(&dev_priv->pcu_lock);
 		ret = sandybridge_pcode_read(dev_priv,
 					     GEN9_PCODE_READ_MEM_LATENCY,
 					     &val);
-		mutex_unlock(&dev_priv->rps.hw_lock);
+		mutex_unlock(&dev_priv->pcu_lock);
 		if (ret) {
 			DRM_ERROR("SKL Mailbox read error = %d\n", ret);
 			return;
@@ -3608,13 +3608,13 @@ intel_enable_sagv(struct drm_i915_private *dev_priv)
 		return 0;
 
 	DRM_DEBUG_KMS("Enabling the SAGV\n");
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	ret = sandybridge_pcode_write(dev_priv, GEN9_PCODE_SAGV_CONTROL,
 				      GEN9_SAGV_ENABLE);
 
 	/* We don't need to wait for the SAGV when enabling */
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	/*
 	 * Some skl systems, pre-release machines in particular,
@@ -3645,14 +3645,14 @@ intel_disable_sagv(struct drm_i915_private *dev_priv)
 		return 0;
 
 	DRM_DEBUG_KMS("Disabling the SAGV\n");
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	/* bspec says to keep retrying for at least 1 ms */
 	ret = skl_pcode_request(dev_priv, GEN9_PCODE_SAGV_CONTROL,
 				GEN9_SAGV_DISABLE,
 				GEN9_SAGV_IS_DISABLED, GEN9_SAGV_IS_DISABLED,
 				1);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	/*
 	 * Some skl systems, pre-release machines in particular,
@@ -4820,16 +4820,18 @@ static inline bool skl_ddb_entries_overlap(const struct skl_ddb_entry *a,
 	return a->start < b->end && b->start < a->end;
 }
 
-bool skl_ddb_allocation_overlaps(const struct skl_ddb_entry **entries,
+bool skl_ddb_allocation_overlaps(struct drm_i915_private *dev_priv,
+				 const struct skl_ddb_entry **entries,
 				 const struct skl_ddb_entry *ddb,
 				 int ignore)
 {
-	int i;
+	enum pipe pipe;
 
-	for (i = 0; i < I915_MAX_PIPES; i++)
-		if (i != ignore && entries[i] &&
-		    skl_ddb_entries_overlap(ddb, entries[i]))
+	for_each_pipe(dev_priv, pipe) {
+		if (pipe != ignore && entries[pipe] &&
+		    skl_ddb_entries_overlap(ddb, entries[pipe]))
 			return true;
+	}
 
 	return false;
 }
@@ -5619,7 +5621,7 @@ void vlv_wm_get_hw_state(struct drm_device *dev)
 	wm->level = VLV_WM_LEVEL_PM2;
 
 	if (IS_CHERRYVIEW(dev_priv)) {
-		mutex_lock(&dev_priv->rps.hw_lock);
+		mutex_lock(&dev_priv->pcu_lock);
 
 		val = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ);
 		if (val & DSP_MAXFIFO_PM5_ENABLE)
@@ -5649,7 +5651,7 @@ void vlv_wm_get_hw_state(struct drm_device *dev)
 				wm->level = VLV_WM_LEVEL_DDR_DVFS;
 		}
 
-		mutex_unlock(&dev_priv->rps.hw_lock);
+		mutex_unlock(&dev_priv->pcu_lock);
 	}
 
 	for_each_intel_crtc(dev, crtc) {
@@ -5827,6 +5829,12 @@ void intel_enable_ipc(struct drm_i915_private *dev_priv)
 {
 	u32 val;
 
+	/* Display WA #0477 WaDisableIPC: skl */
+	if (IS_SKYLAKE(dev_priv)) {
+		dev_priv->ipc_enabled = false;
+		return;
+	}
+
 	val = I915_READ(DISP_ARB_CTL2);
 
 	if (dev_priv->ipc_enabled)
@@ -5980,6 +5988,7 @@ static void ironlake_disable_drps(struct drm_i915_private *dev_priv)
  */
 static u32 intel_rps_limits(struct drm_i915_private *dev_priv, u8 val)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 limits;
 
 	/* Only set the down limit when we've reached the lowest level to avoid
@@ -5989,13 +5998,13 @@ static u32 intel_rps_limits(struct drm_i915_private *dev_priv, u8 val)
 	 * frequency, if the down threshold expires in that window we will not
 	 * receive a down interrupt. */
 	if (INTEL_GEN(dev_priv) >= 9) {
-		limits = (dev_priv->rps.max_freq_softlimit) << 23;
-		if (val <= dev_priv->rps.min_freq_softlimit)
-			limits |= (dev_priv->rps.min_freq_softlimit) << 14;
+		limits = (rps->max_freq_softlimit) << 23;
+		if (val <= rps->min_freq_softlimit)
+			limits |= (rps->min_freq_softlimit) << 14;
 	} else {
-		limits = dev_priv->rps.max_freq_softlimit << 24;
-		if (val <= dev_priv->rps.min_freq_softlimit)
-			limits |= dev_priv->rps.min_freq_softlimit << 16;
+		limits = rps->max_freq_softlimit << 24;
+		if (val <= rps->min_freq_softlimit)
+			limits |= rps->min_freq_softlimit << 16;
 	}
 
 	return limits;
@@ -6003,39 +6012,40 @@ static u32 intel_rps_limits(struct drm_i915_private *dev_priv, u8 val)
 
 static void gen6_set_rps_thresholds(struct drm_i915_private *dev_priv, u8 val)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	int new_power;
 	u32 threshold_up = 0, threshold_down = 0; /* in % */
 	u32 ei_up = 0, ei_down = 0;
 
-	new_power = dev_priv->rps.power;
-	switch (dev_priv->rps.power) {
+	new_power = rps->power;
+	switch (rps->power) {
 	case LOW_POWER:
-		if (val > dev_priv->rps.efficient_freq + 1 &&
-		    val > dev_priv->rps.cur_freq)
+		if (val > rps->efficient_freq + 1 &&
+		    val > rps->cur_freq)
 			new_power = BETWEEN;
 		break;
 
 	case BETWEEN:
-		if (val <= dev_priv->rps.efficient_freq &&
-		    val < dev_priv->rps.cur_freq)
+		if (val <= rps->efficient_freq &&
+		    val < rps->cur_freq)
 			new_power = LOW_POWER;
-		else if (val >= dev_priv->rps.rp0_freq &&
-			 val > dev_priv->rps.cur_freq)
+		else if (val >= rps->rp0_freq &&
+			 val > rps->cur_freq)
 			new_power = HIGH_POWER;
 		break;
 
 	case HIGH_POWER:
-		if (val < (dev_priv->rps.rp1_freq + dev_priv->rps.rp0_freq) >> 1 &&
-		    val < dev_priv->rps.cur_freq)
+		if (val < (rps->rp1_freq + rps->rp0_freq) >> 1 &&
+		    val < rps->cur_freq)
 			new_power = BETWEEN;
 		break;
 	}
 	/* Max/min bins are special */
-	if (val <= dev_priv->rps.min_freq_softlimit)
+	if (val <= rps->min_freq_softlimit)
 		new_power = LOW_POWER;
-	if (val >= dev_priv->rps.max_freq_softlimit)
+	if (val >= rps->max_freq_softlimit)
 		new_power = HIGH_POWER;
-	if (new_power == dev_priv->rps.power)
+	if (new_power == rps->power)
 		return;
 
 	/* Note the units here are not exactly 1us, but 1280ns. */
@@ -6098,20 +6108,21 @@ static void gen6_set_rps_thresholds(struct drm_i915_private *dev_priv, u8 val)
 		   GEN6_RP_DOWN_IDLE_AVG);
 
 skip_hw_write:
-	dev_priv->rps.power = new_power;
-	dev_priv->rps.up_threshold = threshold_up;
-	dev_priv->rps.down_threshold = threshold_down;
-	dev_priv->rps.last_adj = 0;
+	rps->power = new_power;
+	rps->up_threshold = threshold_up;
+	rps->down_threshold = threshold_down;
+	rps->last_adj = 0;
 }
 
 static u32 gen6_rps_pm_mask(struct drm_i915_private *dev_priv, u8 val)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 mask = 0;
 
 	/* We use UP_EI_EXPIRED interupts for both up/down in manual mode */
-	if (val > dev_priv->rps.min_freq_softlimit)
+	if (val > rps->min_freq_softlimit)
 		mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_DOWN_THRESHOLD | GEN6_PM_RP_DOWN_TIMEOUT;
-	if (val < dev_priv->rps.max_freq_softlimit)
+	if (val < rps->max_freq_softlimit)
 		mask |= GEN6_PM_RP_UP_EI_EXPIRED | GEN6_PM_RP_UP_THRESHOLD;
 
 	mask &= dev_priv->pm_rps_events;
@@ -6124,10 +6135,12 @@ static u32 gen6_rps_pm_mask(struct drm_i915_private *dev_priv, u8 val)
  * update the GEN6_RP_INTERRUPT_LIMITS register accordingly. */
 static int gen6_set_rps(struct drm_i915_private *dev_priv, u8 val)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
 	/* min/max delay may still have been modified so be sure to
 	 * write the limits value.
 	 */
-	if (val != dev_priv->rps.cur_freq) {
+	if (val != rps->cur_freq) {
 		gen6_set_rps_thresholds(dev_priv, val);
 
 		if (INTEL_GEN(dev_priv) >= 9)
@@ -6149,7 +6162,7 @@ static int gen6_set_rps(struct drm_i915_private *dev_priv, u8 val)
 	I915_WRITE(GEN6_RP_INTERRUPT_LIMITS, intel_rps_limits(dev_priv, val));
 	I915_WRITE(GEN6_PMINTRMSK, gen6_rps_pm_mask(dev_priv, val));
 
-	dev_priv->rps.cur_freq = val;
+	rps->cur_freq = val;
 	trace_intel_gpu_freq_change(intel_gpu_freq(dev_priv, val));
 
 	return 0;
@@ -6165,7 +6178,7 @@ static int valleyview_set_rps(struct drm_i915_private *dev_priv, u8 val)
 
 	I915_WRITE(GEN6_PMINTRMSK, gen6_rps_pm_mask(dev_priv, val));
 
-	if (val != dev_priv->rps.cur_freq) {
+	if (val != dev_priv->gt_pm.rps.cur_freq) {
 		err = vlv_punit_write(dev_priv, PUNIT_REG_GPU_FREQ_REQ, val);
 		if (err)
 			return err;
@@ -6173,7 +6186,7 @@ static int valleyview_set_rps(struct drm_i915_private *dev_priv, u8 val)
 		gen6_set_rps_thresholds(dev_priv, val);
 	}
 
-	dev_priv->rps.cur_freq = val;
+	dev_priv->gt_pm.rps.cur_freq = val;
 	trace_intel_gpu_freq_change(intel_gpu_freq(dev_priv, val));
 
 	return 0;
@@ -6188,10 +6201,11 @@ static int valleyview_set_rps(struct drm_i915_private *dev_priv, u8 val)
 */
 static void vlv_set_rps_idle(struct drm_i915_private *dev_priv)
 {
-	u32 val = dev_priv->rps.idle_freq;
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+	u32 val = rps->idle_freq;
 	int err;
 
-	if (dev_priv->rps.cur_freq <= val)
+	if (rps->cur_freq <= val)
 		return;
 
 	/* The punit delays the write of the frequency and voltage until it
@@ -6216,34 +6230,38 @@ static void vlv_set_rps_idle(struct drm_i915_private *dev_priv)
 
 void gen6_rps_busy(struct drm_i915_private *dev_priv)
 {
-	mutex_lock(&dev_priv->rps.hw_lock);
-	if (dev_priv->rps.enabled) {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
+	mutex_lock(&dev_priv->pcu_lock);
+	if (rps->enabled) {
 		u8 freq;
 
 		if (dev_priv->pm_rps_events & GEN6_PM_RP_UP_EI_EXPIRED)
 			gen6_rps_reset_ei(dev_priv);
 		I915_WRITE(GEN6_PMINTRMSK,
-			   gen6_rps_pm_mask(dev_priv, dev_priv->rps.cur_freq));
+			   gen6_rps_pm_mask(dev_priv, rps->cur_freq));
 
 		gen6_enable_rps_interrupts(dev_priv);
 
 		/* Use the user's desired frequency as a guide, but for better
 		 * performance, jump directly to RPe as our starting frequency.
 		 */
-		freq = max(dev_priv->rps.cur_freq,
-			   dev_priv->rps.efficient_freq);
+		freq = max(rps->cur_freq,
+			   rps->efficient_freq);
 
 		if (intel_set_rps(dev_priv,
 				  clamp(freq,
-					dev_priv->rps.min_freq_softlimit,
-					dev_priv->rps.max_freq_softlimit)))
+					rps->min_freq_softlimit,
+					rps->max_freq_softlimit)))
 			DRM_DEBUG_DRIVER("Failed to set idle frequency\n");
 	}
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 }
 
 void gen6_rps_idle(struct drm_i915_private *dev_priv)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
 	/* Flush our bottom-half so that it does not race with us
 	 * setting the idle frequency and so that it is bounded by
 	 * our rpm wakeref. And then disable the interrupts to stop any
@@ -6251,36 +6269,36 @@ void gen6_rps_idle(struct drm_i915_private *dev_priv)
 	 */
 	gen6_disable_rps_interrupts(dev_priv);
 
-	mutex_lock(&dev_priv->rps.hw_lock);
-	if (dev_priv->rps.enabled) {
+	mutex_lock(&dev_priv->pcu_lock);
+	if (rps->enabled) {
 		if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv))
 			vlv_set_rps_idle(dev_priv);
 		else
-			gen6_set_rps(dev_priv, dev_priv->rps.idle_freq);
-		dev_priv->rps.last_adj = 0;
+			gen6_set_rps(dev_priv, rps->idle_freq);
+		rps->last_adj = 0;
 		I915_WRITE(GEN6_PMINTRMSK,
 			   gen6_sanitize_rps_pm_mask(dev_priv, ~0));
 	}
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 }
 
 void gen6_rps_boost(struct drm_i915_gem_request *rq,
-		    struct intel_rps_client *rps)
+		    struct intel_rps_client *rps_client)
 {
-	struct drm_i915_private *i915 = rq->i915;
+	struct intel_rps *rps = &rq->i915->gt_pm.rps;
 	unsigned long flags;
 	bool boost;
 
 	/* This is intentionally racy! We peek at the state here, then
 	 * validate inside the RPS worker.
 	 */
-	if (!i915->rps.enabled)
+	if (!rps->enabled)
 		return;
 
 	boost = false;
 	spin_lock_irqsave(&rq->lock, flags);
 	if (!rq->waitboost && !i915_gem_request_completed(rq)) {
-		atomic_inc(&i915->rps.num_waiters);
+		atomic_inc(&rps->num_waiters);
 		rq->waitboost = true;
 		boost = true;
 	}
@@ -6288,22 +6306,23 @@ void gen6_rps_boost(struct drm_i915_gem_request *rq,
 	if (!boost)
 		return;
 
-	if (READ_ONCE(i915->rps.cur_freq) < i915->rps.boost_freq)
-		schedule_work(&i915->rps.work);
+	if (READ_ONCE(rps->cur_freq) < rps->boost_freq)
+		schedule_work(&rps->work);
 
-	atomic_inc(rps ? &rps->boosts : &i915->rps.boosts);
+	atomic_inc(rps_client ? &rps_client->boosts : &rps->boosts);
 }
 
 int intel_set_rps(struct drm_i915_private *dev_priv, u8 val)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	int err;
 
-	lockdep_assert_held(&dev_priv->rps.hw_lock);
-	GEM_BUG_ON(val > dev_priv->rps.max_freq);
-	GEM_BUG_ON(val < dev_priv->rps.min_freq);
+	lockdep_assert_held(&dev_priv->pcu_lock);
+	GEM_BUG_ON(val > rps->max_freq);
+	GEM_BUG_ON(val < rps->min_freq);
 
-	if (!dev_priv->rps.enabled) {
-		dev_priv->rps.cur_freq = val;
+	if (!rps->enabled) {
+		rps->cur_freq = val;
 		return 0;
 	}
 
@@ -6326,21 +6345,30 @@ static void gen9_disable_rps(struct drm_i915_private *dev_priv)
 	I915_WRITE(GEN6_RP_CONTROL, 0);
 }
 
-static void gen6_disable_rps(struct drm_i915_private *dev_priv)
+static void gen6_disable_rc6(struct drm_i915_private *dev_priv)
 {
 	I915_WRITE(GEN6_RC_CONTROL, 0);
+}
+
+static void gen6_disable_rps(struct drm_i915_private *dev_priv)
+{
 	I915_WRITE(GEN6_RPNSWREQ, 1 << 31);
 	I915_WRITE(GEN6_RP_CONTROL, 0);
 }
 
-static void cherryview_disable_rps(struct drm_i915_private *dev_priv)
+static void cherryview_disable_rc6(struct drm_i915_private *dev_priv)
 {
 	I915_WRITE(GEN6_RC_CONTROL, 0);
 }
 
-static void valleyview_disable_rps(struct drm_i915_private *dev_priv)
+static void cherryview_disable_rps(struct drm_i915_private *dev_priv)
 {
-	/* we're doing forcewake before Disabling RC6,
+	I915_WRITE(GEN6_RP_CONTROL, 0);
+}
+
+static void valleyview_disable_rc6(struct drm_i915_private *dev_priv)
+{
+	/* We're doing forcewake before Disabling RC6,
 	 * This what the BIOS expects when going into suspend */
 	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
 
@@ -6349,6 +6377,11 @@ static void valleyview_disable_rps(struct drm_i915_private *dev_priv)
 	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
 }
 
+static void valleyview_disable_rps(struct drm_i915_private *dev_priv)
+{
+	I915_WRITE(GEN6_RP_CONTROL, 0);
+}
+
 static void intel_print_rc6_info(struct drm_i915_private *dev_priv, u32 mode)
 {
 	if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv)) {
@@ -6471,24 +6504,26 @@ int sanitize_rc6_option(struct drm_i915_private *dev_priv, int enable_rc6)
 
 static void gen6_init_rps_frequencies(struct drm_i915_private *dev_priv)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
 	/* All of these values are in units of 50MHz */
 
 	/* static values from HW: RP0 > RP1 > RPn (min_freq) */
 	if (IS_GEN9_LP(dev_priv)) {
 		u32 rp_state_cap = I915_READ(BXT_RP_STATE_CAP);
-		dev_priv->rps.rp0_freq = (rp_state_cap >> 16) & 0xff;
-		dev_priv->rps.rp1_freq = (rp_state_cap >>  8) & 0xff;
-		dev_priv->rps.min_freq = (rp_state_cap >>  0) & 0xff;
+		rps->rp0_freq = (rp_state_cap >> 16) & 0xff;
+		rps->rp1_freq = (rp_state_cap >>  8) & 0xff;
+		rps->min_freq = (rp_state_cap >>  0) & 0xff;
 	} else {
 		u32 rp_state_cap = I915_READ(GEN6_RP_STATE_CAP);
-		dev_priv->rps.rp0_freq = (rp_state_cap >>  0) & 0xff;
-		dev_priv->rps.rp1_freq = (rp_state_cap >>  8) & 0xff;
-		dev_priv->rps.min_freq = (rp_state_cap >> 16) & 0xff;
+		rps->rp0_freq = (rp_state_cap >>  0) & 0xff;
+		rps->rp1_freq = (rp_state_cap >>  8) & 0xff;
+		rps->min_freq = (rp_state_cap >> 16) & 0xff;
 	}
 	/* hw_max = RP0 until we check for overclocking */
-	dev_priv->rps.max_freq = dev_priv->rps.rp0_freq;
+	rps->max_freq = rps->rp0_freq;
 
-	dev_priv->rps.efficient_freq = dev_priv->rps.rp1_freq;
+	rps->efficient_freq = rps->rp1_freq;
 	if (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv) ||
 	    IS_GEN9_BC(dev_priv) || IS_CANNONLAKE(dev_priv)) {
 		u32 ddcc_status = 0;
@@ -6496,33 +6531,34 @@ static void gen6_init_rps_frequencies(struct drm_i915_private *dev_priv)
 		if (sandybridge_pcode_read(dev_priv,
 					   HSW_PCODE_DYNAMIC_DUTY_CYCLE_CONTROL,
 					   &ddcc_status) == 0)
-			dev_priv->rps.efficient_freq =
+			rps->efficient_freq =
 				clamp_t(u8,
 					((ddcc_status >> 8) & 0xff),
-					dev_priv->rps.min_freq,
-					dev_priv->rps.max_freq);
+					rps->min_freq,
+					rps->max_freq);
 	}
 
 	if (IS_GEN9_BC(dev_priv) || IS_CANNONLAKE(dev_priv)) {
 		/* Store the frequency values in 16.66 MHZ units, which is
 		 * the natural hardware unit for SKL
 		 */
-		dev_priv->rps.rp0_freq *= GEN9_FREQ_SCALER;
-		dev_priv->rps.rp1_freq *= GEN9_FREQ_SCALER;
-		dev_priv->rps.min_freq *= GEN9_FREQ_SCALER;
-		dev_priv->rps.max_freq *= GEN9_FREQ_SCALER;
-		dev_priv->rps.efficient_freq *= GEN9_FREQ_SCALER;
+		rps->rp0_freq *= GEN9_FREQ_SCALER;
+		rps->rp1_freq *= GEN9_FREQ_SCALER;
+		rps->min_freq *= GEN9_FREQ_SCALER;
+		rps->max_freq *= GEN9_FREQ_SCALER;
+		rps->efficient_freq *= GEN9_FREQ_SCALER;
 	}
 }
 
 static void reset_rps(struct drm_i915_private *dev_priv,
 		      int (*set)(struct drm_i915_private *, u8))
 {
-	u8 freq = dev_priv->rps.cur_freq;
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+	u8 freq = rps->cur_freq;
 
 	/* force a reset */
-	dev_priv->rps.power = -1;
-	dev_priv->rps.cur_freq = -1;
+	rps->power = -1;
+	rps->cur_freq = -1;
 
 	if (set(dev_priv, freq))
 		DRM_ERROR("Failed to reset RPS to initial values\n");
@@ -6535,7 +6571,7 @@ static void gen9_enable_rps(struct drm_i915_private *dev_priv)
 
 	/* Program defaults and thresholds for RPS*/
 	I915_WRITE(GEN6_RC_VIDEO_FREQ,
-		GEN9_FREQUENCY(dev_priv->rps.rp1_freq));
+		GEN9_FREQUENCY(dev_priv->gt_pm.rps.rp1_freq));
 
 	/* 1 second timeout*/
 	I915_WRITE(GEN6_RP_DOWN_TIMEOUT,
@@ -6589,7 +6625,7 @@ static void gen9_enable_rc6(struct drm_i915_private *dev_priv)
 	I915_WRITE(GEN9_RENDER_PG_IDLE_HYSTERESIS, 25);
 
 	/* 3a: Enable RC6 */
-	if (intel_enable_rc6() & INTEL_RC6_ENABLE)
+	if (intel_rc6_enabled() & INTEL_RC6_ENABLE)
 		rc6_mask = GEN6_RC_CTL_RC6_ENABLE;
 	DRM_INFO("RC6 %s\n", onoff(rc6_mask & GEN6_RC_CTL_RC6_ENABLE));
 	I915_WRITE(GEN6_RC6_THRESHOLD, 37500); /* 37.5/125ms per EI */
@@ -6609,7 +6645,7 @@ static void gen9_enable_rc6(struct drm_i915_private *dev_priv)
 	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
 }
 
-static void gen8_enable_rps(struct drm_i915_private *dev_priv)
+static void gen8_enable_rc6(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
@@ -6618,7 +6654,7 @@ static void gen8_enable_rps(struct drm_i915_private *dev_priv)
 	/* 1a: Software RC state - RC0 */
 	I915_WRITE(GEN6_RC_STATE, 0);
 
-	/* 1c & 1d: Get forcewake during program sequence. Although the driver
+	/* 1b: Get forcewake during program sequence. Although the driver
 	 * hasn't enabled a state yet where we need forcewake, BIOS may have.*/
 	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
 
@@ -6632,36 +6668,38 @@ static void gen8_enable_rps(struct drm_i915_private *dev_priv)
 	for_each_engine(engine, dev_priv, id)
 		I915_WRITE(RING_MAX_IDLE(engine->mmio_base), 10);
 	I915_WRITE(GEN6_RC_SLEEP, 0);
-	if (IS_BROADWELL(dev_priv))
-		I915_WRITE(GEN6_RC6_THRESHOLD, 625); /* 800us/1.28 for TO */
-	else
-		I915_WRITE(GEN6_RC6_THRESHOLD, 50000); /* 50/125ms per EI */
+	I915_WRITE(GEN6_RC6_THRESHOLD, 625); /* 800us/1.28 for TO */
 
 	/* 3: Enable RC6 */
-	if (intel_enable_rc6() & INTEL_RC6_ENABLE)
+	if (intel_rc6_enabled() & INTEL_RC6_ENABLE)
 		rc6_mask = GEN6_RC_CTL_RC6_ENABLE;
 	intel_print_rc6_info(dev_priv, rc6_mask);
-	if (IS_BROADWELL(dev_priv))
-		I915_WRITE(GEN6_RC_CONTROL, GEN6_RC_CTL_HW_ENABLE |
-				GEN7_RC_CTL_TO_MODE |
-				rc6_mask);
-	else
-		I915_WRITE(GEN6_RC_CONTROL, GEN6_RC_CTL_HW_ENABLE |
-				GEN6_RC_CTL_EI_MODE(1) |
-				rc6_mask);
 
-	/* 4 Program defaults and thresholds for RPS*/
+	I915_WRITE(GEN6_RC_CONTROL, GEN6_RC_CTL_HW_ENABLE |
+			GEN7_RC_CTL_TO_MODE |
+			rc6_mask);
+
+	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
+}
+
+static void gen8_enable_rps(struct drm_i915_private *dev_priv)
+{
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
+	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
+
+	/* 1 Program defaults and thresholds for RPS*/
 	I915_WRITE(GEN6_RPNSWREQ,
-		   HSW_FREQUENCY(dev_priv->rps.rp1_freq));
+		   HSW_FREQUENCY(rps->rp1_freq));
 	I915_WRITE(GEN6_RC_VIDEO_FREQ,
-		   HSW_FREQUENCY(dev_priv->rps.rp1_freq));
+		   HSW_FREQUENCY(rps->rp1_freq));
 	/* NB: Docs say 1s, and 1000000 - which aren't equivalent */
 	I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 100000000 / 128); /* 1 second timeout */
 
 	/* Docs recommend 900MHz, and 300 MHz respectively */
 	I915_WRITE(GEN6_RP_INTERRUPT_LIMITS,
-		   dev_priv->rps.max_freq_softlimit << 24 |
-		   dev_priv->rps.min_freq_softlimit << 16);
+		   rps->max_freq_softlimit << 24 |
+		   rps->min_freq_softlimit << 16);
 
 	I915_WRITE(GEN6_RP_UP_THRESHOLD, 7600000 / 128); /* 76ms busyness per EI, 90% */
 	I915_WRITE(GEN6_RP_DOWN_THRESHOLD, 31300000 / 128); /* 313ms busyness per EI, 70%*/
@@ -6670,7 +6708,7 @@ static void gen8_enable_rps(struct drm_i915_private *dev_priv)
 
 	I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10);
 
-	/* 5: Enable RPS */
+	/* 2: Enable RPS */
 	I915_WRITE(GEN6_RP_CONTROL,
 		   GEN6_RP_MEDIA_TURBO |
 		   GEN6_RP_MEDIA_HW_NORMAL_MODE |
@@ -6679,14 +6717,12 @@ static void gen8_enable_rps(struct drm_i915_private *dev_priv)
 		   GEN6_RP_UP_BUSY_AVG |
 		   GEN6_RP_DOWN_IDLE_AVG);
 
-	/* 6: Ring frequency + overclocking (our driver does this later */
-
 	reset_rps(dev_priv, gen6_set_rps);
 
 	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
 }
 
-static void gen6_enable_rps(struct drm_i915_private *dev_priv)
+static void gen6_enable_rc6(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
@@ -6695,14 +6731,6 @@ static void gen6_enable_rps(struct drm_i915_private *dev_priv)
 	int rc6_mode;
 	int ret;
 
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
-
-	/* Here begins a magic sequence of register writes to enable
-	 * auto-downclocking.
-	 *
-	 * Perhaps there might be some value in exposing these to
-	 * userspace...
-	 */
 	I915_WRITE(GEN6_RC_STATE, 0);
 
 	/* Clear the DBG now so we don't confuse earlier errors */
@@ -6736,7 +6764,7 @@ static void gen6_enable_rps(struct drm_i915_private *dev_priv)
 	I915_WRITE(GEN6_RC6pp_THRESHOLD, 64000); /* unused */
 
 	/* Check if we are enabling RC6 */
-	rc6_mode = intel_enable_rc6();
+	rc6_mode = intel_rc6_enabled();
 	if (rc6_mode & INTEL_RC6_ENABLE)
 		rc6_mask |= GEN6_RC_CTL_RC6_ENABLE;
 
@@ -6756,12 +6784,6 @@ static void gen6_enable_rps(struct drm_i915_private *dev_priv)
 		   GEN6_RC_CTL_EI_MODE(1) |
 		   GEN6_RC_CTL_HW_ENABLE);
 
-	/* Power down if completely idle for over 50ms */
-	I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 50000);
-	I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10);
-
-	reset_rps(dev_priv, gen6_set_rps);
-
 	rc6vids = 0;
 	ret = sandybridge_pcode_read(dev_priv, GEN6_PCODE_READ_RC6VIDS, &rc6vids);
 	if (IS_GEN6(dev_priv) && ret) {
@@ -6779,8 +6801,28 @@ static void gen6_enable_rps(struct drm_i915_private *dev_priv)
 	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
 }
 
+static void gen6_enable_rps(struct drm_i915_private *dev_priv)
+{
+	/* Here begins a magic sequence of register writes to enable
+	 * auto-downclocking.
+	 *
+	 * Perhaps there might be some value in exposing these to
+	 * userspace...
+	 */
+	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
+
+	/* Power down if completely idle for over 50ms */
+	I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 50000);
+	I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10);
+
+	reset_rps(dev_priv, gen6_set_rps);
+
+	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
+}
+
 static void gen6_update_ring_freq(struct drm_i915_private *dev_priv)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	int min_freq = 15;
 	unsigned int gpu_freq;
 	unsigned int max_ia_freq, min_ring_freq;
@@ -6788,7 +6830,7 @@ static void gen6_update_ring_freq(struct drm_i915_private *dev_priv)
 	int scaling_factor = 180;
 	struct cpufreq_policy *policy;
 
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	WARN_ON(!mutex_is_locked(&dev_priv->pcu_lock));
 
 	policy = cpufreq_cpu_get(0);
 	if (policy) {
@@ -6811,11 +6853,11 @@ static void gen6_update_ring_freq(struct drm_i915_private *dev_priv)
 
 	if (IS_GEN9_BC(dev_priv) || IS_CANNONLAKE(dev_priv)) {
 		/* Convert GT frequency to 50 HZ units */
-		min_gpu_freq = dev_priv->rps.min_freq / GEN9_FREQ_SCALER;
-		max_gpu_freq = dev_priv->rps.max_freq / GEN9_FREQ_SCALER;
+		min_gpu_freq = rps->min_freq / GEN9_FREQ_SCALER;
+		max_gpu_freq = rps->max_freq / GEN9_FREQ_SCALER;
 	} else {
-		min_gpu_freq = dev_priv->rps.min_freq;
-		max_gpu_freq = dev_priv->rps.max_freq;
+		min_gpu_freq = rps->min_freq;
+		max_gpu_freq = rps->max_freq;
 	}
 
 	/*
@@ -7066,17 +7108,18 @@ static void valleyview_cleanup_pctx(struct drm_i915_private *dev_priv)
 
 static void vlv_init_gpll_ref_freq(struct drm_i915_private *dev_priv)
 {
-	dev_priv->rps.gpll_ref_freq =
+	dev_priv->gt_pm.rps.gpll_ref_freq =
 		vlv_get_cck_clock(dev_priv, "GPLL ref",
 				  CCK_GPLL_CLOCK_CONTROL,
 				  dev_priv->czclk_freq);
 
 	DRM_DEBUG_DRIVER("GPLL reference freq: %d kHz\n",
-			 dev_priv->rps.gpll_ref_freq);
+			 dev_priv->gt_pm.rps.gpll_ref_freq);
 }
 
 static void valleyview_init_gt_powersave(struct drm_i915_private *dev_priv)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 val;
 
 	valleyview_setup_pctx(dev_priv);
@@ -7098,30 +7141,31 @@ static void valleyview_init_gt_powersave(struct drm_i915_private *dev_priv)
 	}
 	DRM_DEBUG_DRIVER("DDR speed: %d MHz\n", dev_priv->mem_freq);
 
-	dev_priv->rps.max_freq = valleyview_rps_max_freq(dev_priv);
-	dev_priv->rps.rp0_freq = dev_priv->rps.max_freq;
+	rps->max_freq = valleyview_rps_max_freq(dev_priv);
+	rps->rp0_freq = rps->max_freq;
 	DRM_DEBUG_DRIVER("max GPU freq: %d MHz (%u)\n",
-			 intel_gpu_freq(dev_priv, dev_priv->rps.max_freq),
-			 dev_priv->rps.max_freq);
+			 intel_gpu_freq(dev_priv, rps->max_freq),
+			 rps->max_freq);
 
-	dev_priv->rps.efficient_freq = valleyview_rps_rpe_freq(dev_priv);
+	rps->efficient_freq = valleyview_rps_rpe_freq(dev_priv);
 	DRM_DEBUG_DRIVER("RPe GPU freq: %d MHz (%u)\n",
-			 intel_gpu_freq(dev_priv, dev_priv->rps.efficient_freq),
-			 dev_priv->rps.efficient_freq);
+			 intel_gpu_freq(dev_priv, rps->efficient_freq),
+			 rps->efficient_freq);
 
-	dev_priv->rps.rp1_freq = valleyview_rps_guar_freq(dev_priv);
+	rps->rp1_freq = valleyview_rps_guar_freq(dev_priv);
 	DRM_DEBUG_DRIVER("RP1(Guar Freq) GPU freq: %d MHz (%u)\n",
-			 intel_gpu_freq(dev_priv, dev_priv->rps.rp1_freq),
-			 dev_priv->rps.rp1_freq);
+			 intel_gpu_freq(dev_priv, rps->rp1_freq),
+			 rps->rp1_freq);
 
-	dev_priv->rps.min_freq = valleyview_rps_min_freq(dev_priv);
+	rps->min_freq = valleyview_rps_min_freq(dev_priv);
 	DRM_DEBUG_DRIVER("min GPU freq: %d MHz (%u)\n",
-			 intel_gpu_freq(dev_priv, dev_priv->rps.min_freq),
-			 dev_priv->rps.min_freq);
+			 intel_gpu_freq(dev_priv, rps->min_freq),
+			 rps->min_freq);
 }
 
 static void cherryview_init_gt_powersave(struct drm_i915_private *dev_priv)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
 	u32 val;
 
 	cherryview_setup_pctx(dev_priv);
@@ -7142,31 +7186,29 @@ static void cherryview_init_gt_powersave(struct drm_i915_private *dev_priv)
 	}
 	DRM_DEBUG_DRIVER("DDR speed: %d MHz\n", dev_priv->mem_freq);
 
-	dev_priv->rps.max_freq = cherryview_rps_max_freq(dev_priv);
-	dev_priv->rps.rp0_freq = dev_priv->rps.max_freq;
+	rps->max_freq = cherryview_rps_max_freq(dev_priv);
+	rps->rp0_freq = rps->max_freq;
 	DRM_DEBUG_DRIVER("max GPU freq: %d MHz (%u)\n",
-			 intel_gpu_freq(dev_priv, dev_priv->rps.max_freq),
-			 dev_priv->rps.max_freq);
+			 intel_gpu_freq(dev_priv, rps->max_freq),
+			 rps->max_freq);
 
-	dev_priv->rps.efficient_freq = cherryview_rps_rpe_freq(dev_priv);
+	rps->efficient_freq = cherryview_rps_rpe_freq(dev_priv);
 	DRM_DEBUG_DRIVER("RPe GPU freq: %d MHz (%u)\n",
-			 intel_gpu_freq(dev_priv, dev_priv->rps.efficient_freq),
-			 dev_priv->rps.efficient_freq);
+			 intel_gpu_freq(dev_priv, rps->efficient_freq),
+			 rps->efficient_freq);
 
-	dev_priv->rps.rp1_freq = cherryview_rps_guar_freq(dev_priv);
+	rps->rp1_freq = cherryview_rps_guar_freq(dev_priv);
 	DRM_DEBUG_DRIVER("RP1(Guar) GPU freq: %d MHz (%u)\n",
-			 intel_gpu_freq(dev_priv, dev_priv->rps.rp1_freq),
-			 dev_priv->rps.rp1_freq);
+			 intel_gpu_freq(dev_priv, rps->rp1_freq),
+			 rps->rp1_freq);
 
-	dev_priv->rps.min_freq = cherryview_rps_min_freq(dev_priv);
+	rps->min_freq = cherryview_rps_min_freq(dev_priv);
 	DRM_DEBUG_DRIVER("min GPU freq: %d MHz (%u)\n",
-			 intel_gpu_freq(dev_priv, dev_priv->rps.min_freq),
-			 dev_priv->rps.min_freq);
+			 intel_gpu_freq(dev_priv, rps->min_freq),
+			 rps->min_freq);
 
-	WARN_ONCE((dev_priv->rps.max_freq |
-		   dev_priv->rps.efficient_freq |
-		   dev_priv->rps.rp1_freq |
-		   dev_priv->rps.min_freq) & 1,
+	WARN_ONCE((rps->max_freq | rps->efficient_freq | rps->rp1_freq |
+		   rps->min_freq) & 1,
 		  "Odd GPU freq values\n");
 }
 
@@ -7175,13 +7217,11 @@ static void valleyview_cleanup_gt_powersave(struct drm_i915_private *dev_priv)
 	valleyview_cleanup_pctx(dev_priv);
 }
 
-static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
+static void cherryview_enable_rc6(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
-	u32 gtfifodbg, val, rc6_mode = 0, pcbr;
-
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	u32 gtfifodbg, rc6_mode = 0, pcbr;
 
 	gtfifodbg = I915_READ(GTFIFODBG) & ~(GT_FIFO_SBDEDICATE_FREE_ENTRY_CHV |
 					     GT_FIFO_FREE_ENTRIES_CHV);
@@ -7212,7 +7252,7 @@ static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
 	/* TO threshold set to 500 us ( 0x186 * 1.28 us) */
 	I915_WRITE(GEN6_RC6_THRESHOLD, 0x186);
 
-	/* allows RC6 residency counter to work */
+	/* Allows RC6 residency counter to work */
 	I915_WRITE(VLV_COUNTER_CONTROL,
 		   _MASKED_BIT_ENABLE(VLV_COUNT_RANGE_HIGH |
 				      VLV_MEDIA_RC6_COUNT_EN |
@@ -7222,13 +7262,22 @@ static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
 	pcbr = I915_READ(VLV_PCBR);
 
 	/* 3: Enable RC6 */
-	if ((intel_enable_rc6() & INTEL_RC6_ENABLE) &&
+	if ((intel_rc6_enabled() & INTEL_RC6_ENABLE) &&
 	    (pcbr >> VLV_PCBR_ADDR_SHIFT))
 		rc6_mode = GEN7_RC_CTL_TO_MODE;
 
 	I915_WRITE(GEN6_RC_CONTROL, rc6_mode);
 
-	/* 4 Program defaults and thresholds for RPS*/
+	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
+}
+
+static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
+{
+	u32 val;
+
+	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
+
+	/* 1: Program defaults and thresholds for RPS*/
 	I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 1000000);
 	I915_WRITE(GEN6_RP_UP_THRESHOLD, 59400);
 	I915_WRITE(GEN6_RP_DOWN_THRESHOLD, 245000);
@@ -7237,7 +7286,7 @@ static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
 
 	I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10);
 
-	/* 5: Enable RPS */
+	/* 2: Enable RPS */
 	I915_WRITE(GEN6_RP_CONTROL,
 		   GEN6_RP_MEDIA_HW_NORMAL_MODE |
 		   GEN6_RP_MEDIA_IS_GFX |
@@ -7264,13 +7313,11 @@ static void cherryview_enable_rps(struct drm_i915_private *dev_priv)
 	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
 }
 
-static void valleyview_enable_rps(struct drm_i915_private *dev_priv)
+static void valleyview_enable_rc6(struct drm_i915_private *dev_priv)
 {
 	struct intel_engine_cs *engine;
 	enum intel_engine_id id;
-	u32 gtfifodbg, val, rc6_mode = 0;
-
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	u32 gtfifodbg, rc6_mode = 0;
 
 	valleyview_check_pctx(dev_priv);
 
@@ -7281,28 +7328,11 @@ static void valleyview_enable_rps(struct drm_i915_private *dev_priv)
 		I915_WRITE(GTFIFODBG, gtfifodbg);
 	}
 
-	/* If VLV, Forcewake all wells, else re-direct to regular path */
 	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
 
 	/*  Disable RC states. */
 	I915_WRITE(GEN6_RC_CONTROL, 0);
 
-	I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 1000000);
-	I915_WRITE(GEN6_RP_UP_THRESHOLD, 59400);
-	I915_WRITE(GEN6_RP_DOWN_THRESHOLD, 245000);
-	I915_WRITE(GEN6_RP_UP_EI, 66000);
-	I915_WRITE(GEN6_RP_DOWN_EI, 350000);
-
-	I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10);
-
-	I915_WRITE(GEN6_RP_CONTROL,
-		   GEN6_RP_MEDIA_TURBO |
-		   GEN6_RP_MEDIA_HW_NORMAL_MODE |
-		   GEN6_RP_MEDIA_IS_GFX |
-		   GEN6_RP_ENABLE |
-		   GEN6_RP_UP_BUSY_AVG |
-		   GEN6_RP_DOWN_IDLE_CONT);
-
 	I915_WRITE(GEN6_RC6_WAKE_RATE_LIMIT, 0x00280000);
 	I915_WRITE(GEN6_RC_EVALUATION_INTERVAL, 125000);
 	I915_WRITE(GEN6_RC_IDLE_HYSTERSIS, 25);
@@ -7312,7 +7342,7 @@ static void valleyview_enable_rps(struct drm_i915_private *dev_priv)
 
 	I915_WRITE(GEN6_RC6_THRESHOLD, 0x557);
 
-	/* allows RC6 residency counter to work */
+	/* Allows RC6 residency counter to work */
 	I915_WRITE(VLV_COUNTER_CONTROL,
 		   _MASKED_BIT_ENABLE(VLV_COUNT_RANGE_HIGH |
 				      VLV_MEDIA_RC0_COUNT_EN |
@@ -7320,13 +7350,38 @@ static void valleyview_enable_rps(struct drm_i915_private *dev_priv)
 				      VLV_MEDIA_RC6_COUNT_EN |
 				      VLV_RENDER_RC6_COUNT_EN));
 
-	if (intel_enable_rc6() & INTEL_RC6_ENABLE)
+	if (intel_rc6_enabled() & INTEL_RC6_ENABLE)
 		rc6_mode = GEN7_RC_CTL_TO_MODE | VLV_RC_CTL_CTX_RST_PARALLEL;
 
 	intel_print_rc6_info(dev_priv, rc6_mode);
 
 	I915_WRITE(GEN6_RC_CONTROL, rc6_mode);
 
+	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
+}
+
+static void valleyview_enable_rps(struct drm_i915_private *dev_priv)
+{
+	u32 val;
+
+	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);
+
+	I915_WRITE(GEN6_RP_DOWN_TIMEOUT, 1000000);
+	I915_WRITE(GEN6_RP_UP_THRESHOLD, 59400);
+	I915_WRITE(GEN6_RP_DOWN_THRESHOLD, 245000);
+	I915_WRITE(GEN6_RP_UP_EI, 66000);
+	I915_WRITE(GEN6_RP_DOWN_EI, 350000);
+
+	I915_WRITE(GEN6_RP_IDLE_HYSTERSIS, 10);
+
+	I915_WRITE(GEN6_RP_CONTROL,
+		   GEN6_RP_MEDIA_TURBO |
+		   GEN6_RP_MEDIA_HW_NORMAL_MODE |
+		   GEN6_RP_MEDIA_IS_GFX |
+		   GEN6_RP_ENABLE |
+		   GEN6_RP_UP_BUSY_AVG |
+		   GEN6_RP_DOWN_IDLE_CONT);
+
 	/* Setting Fixed Bias */
 	val = VLV_OVERRIDE_EN |
 		  VLV_SOC_TDP_EN |
@@ -7534,7 +7589,7 @@ static unsigned long __i915_gfx_val(struct drm_i915_private *dev_priv)
 
 	lockdep_assert_held(&mchdev_lock);
 
-	pxvid = I915_READ(PXVFREQ(dev_priv->rps.cur_freq));
+	pxvid = I915_READ(PXVFREQ(dev_priv->gt_pm.rps.cur_freq));
 	pxvid = (pxvid >> 24) & 0x7f;
 	ext_v = pvid_to_extvid(dev_priv, pxvid);
 
@@ -7821,6 +7876,8 @@ static void intel_init_emon(struct drm_i915_private *dev_priv)
 
 void intel_init_gt_powersave(struct drm_i915_private *dev_priv)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
 	/*
 	 * RPM depends on RC6 to save restore the GT HW context, so make RC6 a
 	 * requirement.
@@ -7831,7 +7888,7 @@ void intel_init_gt_powersave(struct drm_i915_private *dev_priv)
 	}
 
 	mutex_lock(&dev_priv->drm.struct_mutex);
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	/* Initialize RPS limits (for userspace) */
 	if (IS_CHERRYVIEW(dev_priv))
@@ -7842,16 +7899,16 @@ void intel_init_gt_powersave(struct drm_i915_private *dev_priv)
 		gen6_init_rps_frequencies(dev_priv);
 
 	/* Derive initial user preferences/limits from the hardware limits */
-	dev_priv->rps.idle_freq = dev_priv->rps.min_freq;
-	dev_priv->rps.cur_freq = dev_priv->rps.idle_freq;
+	rps->idle_freq = rps->min_freq;
+	rps->cur_freq = rps->idle_freq;
 
-	dev_priv->rps.max_freq_softlimit = dev_priv->rps.max_freq;
-	dev_priv->rps.min_freq_softlimit = dev_priv->rps.min_freq;
+	rps->max_freq_softlimit = rps->max_freq;
+	rps->min_freq_softlimit = rps->min_freq;
 
 	if (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv))
-		dev_priv->rps.min_freq_softlimit =
+		rps->min_freq_softlimit =
 			max_t(int,
-			      dev_priv->rps.efficient_freq,
+			      rps->efficient_freq,
 			      intel_freq_opcode(dev_priv, 450));
 
 	/* After setting max-softlimit, find the overclock max freq */
@@ -7862,16 +7919,16 @@ void intel_init_gt_powersave(struct drm_i915_private *dev_priv)
 		sandybridge_pcode_read(dev_priv, GEN6_READ_OC_PARAMS, &params);
 		if (params & BIT(31)) { /* OC supported */
 			DRM_DEBUG_DRIVER("Overclocking supported, max: %dMHz, overclock: %dMHz\n",
-					 (dev_priv->rps.max_freq & 0xff) * 50,
+					 (rps->max_freq & 0xff) * 50,
 					 (params & 0xff) * 50);
-			dev_priv->rps.max_freq = params & 0xff;
+			rps->max_freq = params & 0xff;
 		}
 	}
 
 	/* Finally allow us to boost to max by default */
-	dev_priv->rps.boost_freq = dev_priv->rps.max_freq;
+	rps->boost_freq = rps->max_freq;
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 	mutex_unlock(&dev_priv->drm.struct_mutex);
 
 	intel_autoenable_gt_powersave(dev_priv);
@@ -7899,7 +7956,7 @@ void intel_suspend_gt_powersave(struct drm_i915_private *dev_priv)
 	if (INTEL_GEN(dev_priv) < 6)
 		return;
 
-	if (cancel_delayed_work_sync(&dev_priv->rps.autoenable_work))
+	if (cancel_delayed_work_sync(&dev_priv->gt_pm.autoenable_work))
 		intel_runtime_pm_put(dev_priv);
 
 	/* gen6_rps_idle() will be called later to disable interrupts */
@@ -7907,90 +7964,168 @@ void intel_suspend_gt_powersave(struct drm_i915_private *dev_priv)
 
 void intel_sanitize_gt_powersave(struct drm_i915_private *dev_priv)
 {
-	dev_priv->rps.enabled = true; /* force disabling */
+	dev_priv->gt_pm.rps.enabled = true; /* force RPS disabling */
+	dev_priv->gt_pm.rc6.enabled = true; /* force RC6 disabling */
 	intel_disable_gt_powersave(dev_priv);
 
 	gen6_reset_rps_interrupts(dev_priv);
 }
 
-void intel_disable_gt_powersave(struct drm_i915_private *dev_priv)
+static inline void intel_disable_llc_pstate(struct drm_i915_private *i915)
 {
-	if (!READ_ONCE(dev_priv->rps.enabled))
+	lockdep_assert_held(&i915->pcu_lock);
+
+	if (!i915->gt_pm.llc_pstate.enabled)
 		return;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	/* Currently there is no HW configuration to be done to disable. */
 
-	if (INTEL_GEN(dev_priv) >= 9) {
+	i915->gt_pm.llc_pstate.enabled = false;
+}
+
+static void intel_disable_rc6(struct drm_i915_private *dev_priv)
+{
+	lockdep_assert_held(&dev_priv->pcu_lock);
+
+	if (!dev_priv->gt_pm.rc6.enabled)
+		return;
+
+	if (INTEL_GEN(dev_priv) >= 9)
 		gen9_disable_rc6(dev_priv);
+	else if (IS_CHERRYVIEW(dev_priv))
+		cherryview_disable_rc6(dev_priv);
+	else if (IS_VALLEYVIEW(dev_priv))
+		valleyview_disable_rc6(dev_priv);
+	else if (INTEL_GEN(dev_priv) >= 6)
+		gen6_disable_rc6(dev_priv);
+
+	dev_priv->gt_pm.rc6.enabled = false;
+}
+
+static void intel_disable_rps(struct drm_i915_private *dev_priv)
+{
+	lockdep_assert_held(&dev_priv->pcu_lock);
+
+	if (!dev_priv->gt_pm.rps.enabled)
+		return;
+
+	if (INTEL_GEN(dev_priv) >= 9)
 		gen9_disable_rps(dev_priv);
-	} else if (IS_CHERRYVIEW(dev_priv)) {
+	else if (IS_CHERRYVIEW(dev_priv))
 		cherryview_disable_rps(dev_priv);
-	} else if (IS_VALLEYVIEW(dev_priv)) {
+	else if (IS_VALLEYVIEW(dev_priv))
 		valleyview_disable_rps(dev_priv);
-	} else if (INTEL_GEN(dev_priv) >= 6) {
+	else if (INTEL_GEN(dev_priv) >= 6)
 		gen6_disable_rps(dev_priv);
-	}  else if (IS_IRONLAKE_M(dev_priv)) {
+	else if (IS_IRONLAKE_M(dev_priv))
 		ironlake_disable_drps(dev_priv);
-	}
 
-	dev_priv->rps.enabled = false;
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	dev_priv->gt_pm.rps.enabled = false;
 }
 
-void intel_enable_gt_powersave(struct drm_i915_private *dev_priv)
+void intel_disable_gt_powersave(struct drm_i915_private *dev_priv)
 {
-	/* We shouldn't be disabling as we submit, so this should be less
-	 * racy than it appears!
-	 */
-	if (READ_ONCE(dev_priv->rps.enabled))
+	mutex_lock(&dev_priv->pcu_lock);
+
+	intel_disable_rc6(dev_priv);
+	intel_disable_rps(dev_priv);
+	if (HAS_LLC(dev_priv))
+		intel_disable_llc_pstate(dev_priv);
+
+	mutex_unlock(&dev_priv->pcu_lock);
+}
+
+static inline void intel_enable_llc_pstate(struct drm_i915_private *i915)
+{
+	lockdep_assert_held(&i915->pcu_lock);
+
+	if (i915->gt_pm.llc_pstate.enabled)
 		return;
 
-	/* Powersaving is controlled by the host when inside a VM */
-	if (intel_vgpu_active(dev_priv))
+	gen6_update_ring_freq(i915);
+
+	i915->gt_pm.llc_pstate.enabled = true;
+}
+
+static void intel_enable_rc6(struct drm_i915_private *dev_priv)
+{
+	lockdep_assert_held(&dev_priv->pcu_lock);
+
+	if (dev_priv->gt_pm.rc6.enabled)
 		return;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	if (IS_CHERRYVIEW(dev_priv))
+		cherryview_enable_rc6(dev_priv);
+	else if (IS_VALLEYVIEW(dev_priv))
+		valleyview_enable_rc6(dev_priv);
+	else if (INTEL_GEN(dev_priv) >= 9)
+		gen9_enable_rc6(dev_priv);
+	else if (IS_BROADWELL(dev_priv))
+		gen8_enable_rc6(dev_priv);
+	else if (INTEL_GEN(dev_priv) >= 6)
+		gen6_enable_rc6(dev_priv);
+
+	dev_priv->gt_pm.rc6.enabled = true;
+}
+
+static void intel_enable_rps(struct drm_i915_private *dev_priv)
+{
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
+	lockdep_assert_held(&dev_priv->pcu_lock);
+
+	if (rps->enabled)
+		return;
 
 	if (IS_CHERRYVIEW(dev_priv)) {
 		cherryview_enable_rps(dev_priv);
 	} else if (IS_VALLEYVIEW(dev_priv)) {
 		valleyview_enable_rps(dev_priv);
 	} else if (INTEL_GEN(dev_priv) >= 9) {
-		gen9_enable_rc6(dev_priv);
 		gen9_enable_rps(dev_priv);
-		if (IS_GEN9_BC(dev_priv) || IS_CANNONLAKE(dev_priv))
-			gen6_update_ring_freq(dev_priv);
 	} else if (IS_BROADWELL(dev_priv)) {
 		gen8_enable_rps(dev_priv);
-		gen6_update_ring_freq(dev_priv);
 	} else if (INTEL_GEN(dev_priv) >= 6) {
 		gen6_enable_rps(dev_priv);
-		gen6_update_ring_freq(dev_priv);
 	} else if (IS_IRONLAKE_M(dev_priv)) {
 		ironlake_enable_drps(dev_priv);
 		intel_init_emon(dev_priv);
 	}
 
-	WARN_ON(dev_priv->rps.max_freq < dev_priv->rps.min_freq);
-	WARN_ON(dev_priv->rps.idle_freq > dev_priv->rps.max_freq);
+	WARN_ON(rps->max_freq < rps->min_freq);
+	WARN_ON(rps->idle_freq > rps->max_freq);
+
+	WARN_ON(rps->efficient_freq < rps->min_freq);
+	WARN_ON(rps->efficient_freq > rps->max_freq);
+
+	rps->enabled = true;
+}
+
+void intel_enable_gt_powersave(struct drm_i915_private *dev_priv)
+{
+	/* Powersaving is controlled by the host when inside a VM */
+	if (intel_vgpu_active(dev_priv))
+		return;
+
+	mutex_lock(&dev_priv->pcu_lock);
 
-	WARN_ON(dev_priv->rps.efficient_freq < dev_priv->rps.min_freq);
-	WARN_ON(dev_priv->rps.efficient_freq > dev_priv->rps.max_freq);
+	intel_enable_rc6(dev_priv);
+	intel_enable_rps(dev_priv);
+	if (HAS_LLC(dev_priv))
+		intel_enable_llc_pstate(dev_priv);
 
-	dev_priv->rps.enabled = true;
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 }
 
 static void __intel_autoenable_gt_powersave(struct work_struct *work)
 {
 	struct drm_i915_private *dev_priv =
-		container_of(work, typeof(*dev_priv), rps.autoenable_work.work);
+		container_of(work,
+			     typeof(*dev_priv),
+			     gt_pm.autoenable_work.work);
 	struct intel_engine_cs *rcs;
 	struct drm_i915_gem_request *req;
 
-	if (READ_ONCE(dev_priv->rps.enabled))
-		goto out;
-
 	rcs = dev_priv->engine[RCS];
 	if (rcs->last_retired_context)
 		goto out;
@@ -8018,9 +8153,6 @@ out:
 
 void intel_autoenable_gt_powersave(struct drm_i915_private *dev_priv)
 {
-	if (READ_ONCE(dev_priv->rps.enabled))
-		return;
-
 	if (IS_IRONLAKE_M(dev_priv)) {
 		ironlake_enable_drps(dev_priv);
 		intel_init_emon(dev_priv);
@@ -8038,7 +8170,7 @@ void intel_autoenable_gt_powersave(struct drm_i915_private *dev_priv)
 		 * runtime resume it's necessary).
 		 */
 		if (queue_delayed_work(dev_priv->wq,
-				       &dev_priv->rps.autoenable_work,
+				       &dev_priv->gt_pm.autoenable_work,
 				       round_jiffies_up_relative(HZ)))
 			intel_runtime_pm_get_noresume(dev_priv);
 	}
@@ -8447,6 +8579,9 @@ static void skl_init_clock_gating(struct drm_i915_private *dev_priv)
 
 static void bdw_init_clock_gating(struct drm_i915_private *dev_priv)
 {
+	/* The GTT cache must be disabled if the system is using 2M pages. */
+	bool can_use_gtt_cache = !HAS_PAGE_SIZES(dev_priv,
+						 I915_GTT_PAGE_SIZE_2M);
 	enum pipe pipe;
 
 	ilk_init_lp_watermarks(dev_priv);
@@ -8481,12 +8616,8 @@ static void bdw_init_clock_gating(struct drm_i915_private *dev_priv)
 	/* WaProgramL3SqcReg1Default:bdw */
 	gen8_set_l3sqc_credits(dev_priv, 30, 2);
 
-	/*
-	 * WaGttCachingOffByDefault:bdw
-	 * GTT cache may not work with big pages, so if those
-	 * are ever enabled GTT cache may need to be disabled.
-	 */
-	I915_WRITE(HSW_GTT_CACHE_EN, GTT_CACHE_EN_ALL);
+	/* WaGttCachingOffByDefault:bdw */
+	I915_WRITE(HSW_GTT_CACHE_EN, can_use_gtt_cache ? GTT_CACHE_EN_ALL : 0);
 
 	/* WaKVMNotificationOnConfigChange:bdw */
 	I915_WRITE(CHICKEN_PAR2_1, I915_READ(CHICKEN_PAR2_1)
@@ -9067,7 +9198,7 @@ int sandybridge_pcode_read(struct drm_i915_private *dev_priv, u32 mbox, u32 *val
 {
 	int status;
 
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	WARN_ON(!mutex_is_locked(&dev_priv->pcu_lock));
 
 	/* GEN6_PCODE_* are outside of the forcewake domain, we can
 	 * use te fw I915_READ variants to reduce the amount of work
@@ -9114,7 +9245,7 @@ int sandybridge_pcode_write(struct drm_i915_private *dev_priv,
 {
 	int status;
 
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	WARN_ON(!mutex_is_locked(&dev_priv->pcu_lock));
 
 	/* GEN6_PCODE_* are outside of the forcewake domain, we can
 	 * use te fw I915_READ variants to reduce the amount of work
@@ -9191,7 +9322,7 @@ int skl_pcode_request(struct drm_i915_private *dev_priv, u32 mbox, u32 request,
 	u32 status;
 	int ret;
 
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	WARN_ON(!mutex_is_locked(&dev_priv->pcu_lock));
 
 #define COND skl_pcode_try_request(dev_priv, mbox, request, reply_mask, reply, \
 				   &status)
@@ -9233,31 +9364,39 @@ out:
 
 static int byt_gpu_freq(struct drm_i915_private *dev_priv, int val)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
 	/*
 	 * N = val - 0xb7
 	 * Slow = Fast = GPLL ref * N
 	 */
-	return DIV_ROUND_CLOSEST(dev_priv->rps.gpll_ref_freq * (val - 0xb7), 1000);
+	return DIV_ROUND_CLOSEST(rps->gpll_ref_freq * (val - 0xb7), 1000);
 }
 
 static int byt_freq_opcode(struct drm_i915_private *dev_priv, int val)
 {
-	return DIV_ROUND_CLOSEST(1000 * val, dev_priv->rps.gpll_ref_freq) + 0xb7;
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
+	return DIV_ROUND_CLOSEST(1000 * val, rps->gpll_ref_freq) + 0xb7;
 }
 
 static int chv_gpu_freq(struct drm_i915_private *dev_priv, int val)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
 	/*
 	 * N = val / 2
 	 * CU (slow) = CU2x (fast) / 2 = GPLL ref * N / 2
 	 */
-	return DIV_ROUND_CLOSEST(dev_priv->rps.gpll_ref_freq * val, 2 * 2 * 1000);
+	return DIV_ROUND_CLOSEST(rps->gpll_ref_freq * val, 2 * 2 * 1000);
 }
 
 static int chv_freq_opcode(struct drm_i915_private *dev_priv, int val)
 {
+	struct intel_rps *rps = &dev_priv->gt_pm.rps;
+
 	/* CHV needs even values */
-	return DIV_ROUND_CLOSEST(2 * 1000 * val, dev_priv->rps.gpll_ref_freq) * 2;
+	return DIV_ROUND_CLOSEST(2 * 1000 * val, rps->gpll_ref_freq) * 2;
 }
 
 int intel_gpu_freq(struct drm_i915_private *dev_priv, int val)
@@ -9288,14 +9427,14 @@ int intel_freq_opcode(struct drm_i915_private *dev_priv, int val)
 
 void intel_pm_setup(struct drm_i915_private *dev_priv)
 {
-	mutex_init(&dev_priv->rps.hw_lock);
+	mutex_init(&dev_priv->pcu_lock);
 
-	INIT_DELAYED_WORK(&dev_priv->rps.autoenable_work,
+	INIT_DELAYED_WORK(&dev_priv->gt_pm.autoenable_work,
 			  __intel_autoenable_gt_powersave);
-	atomic_set(&dev_priv->rps.num_waiters, 0);
+	atomic_set(&dev_priv->gt_pm.rps.num_waiters, 0);
 
-	dev_priv->pm.suspended = false;
-	atomic_set(&dev_priv->pm.wakeref_count, 0);
+	dev_priv->runtime_pm.suspended = false;
+	atomic_set(&dev_priv->runtime_pm.wakeref_count, 0);
 }
 
 static u64 vlv_residency_raw(struct drm_i915_private *dev_priv,
@@ -9348,7 +9487,7 @@ u64 intel_rc6_residency_us(struct drm_i915_private *dev_priv,
 {
 	u64 time_hw, units, div;
 
-	if (!intel_enable_rc6())
+	if (!intel_rc6_enabled())
 		return 0;
 
 	intel_runtime_pm_get(dev_priv);
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 05c08b0bc172..4285f09ff8b8 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -579,7 +579,16 @@ out:
 static void reset_ring_common(struct intel_engine_cs *engine,
 			      struct drm_i915_gem_request *request)
 {
-	/* Try to restore the logical GPU state to match the continuation
+	/*
+	 * RC6 must be prevented until the reset is complete and the engine
+	 * reinitialised. If it occurs in the middle of this sequence, the
+	 * state written to/loaded from the power context is ill-defined (e.g.
+	 * the PP_BASE_DIR may be lost).
+	 */
+	assert_forcewakes_active(engine->i915, FORCEWAKE_ALL);
+
+	/*
+	 * Try to restore the logical GPU state to match the continuation
 	 * of the request queue. If we skip the context/PD restore, then
 	 * the next request may try to execute assuming that its context
 	 * is valid and loaded on the GPU and so may try to access invalid
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 56d7ae9f298b..17186f067408 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -7,6 +7,8 @@
 #include "i915_gem_timeline.h"
 #include "i915_selftest.h"
 
+struct drm_printer;
+
 #define I915_CMD_HASH_ORDER 9
 
 /* Early gen2 devices have a cacheline of just 32 bytes, using 64 is overkill,
@@ -239,6 +241,11 @@ struct intel_engine_execlists {
 	} port[EXECLIST_MAX_PORTS];
 
 	/**
+	 * @preempt: are we currently handling a preempting context switch?
+	 */
+	bool preempt;
+
+	/**
 	 * @port_mask: number of execlist ports - 1
 	 */
 	unsigned int port_mask;
@@ -834,4 +841,6 @@ void intel_engines_reset_default_submission(struct drm_i915_private *i915);
 
 bool intel_engine_can_store_dword(struct intel_engine_cs *engine);
 
+void intel_engine_dump(struct intel_engine_cs *engine, struct drm_printer *p);
+
 #endif /* _INTEL_RINGBUFFER_H_ */
diff --git a/drivers/gpu/drm/i915/intel_runtime_pm.c b/drivers/gpu/drm/i915/intel_runtime_pm.c
index 7933d1bc6a1c..8af286c63d3b 100644
--- a/drivers/gpu/drm/i915/intel_runtime_pm.c
+++ b/drivers/gpu/drm/i915/intel_runtime_pm.c
@@ -187,7 +187,7 @@ bool __intel_display_power_is_enabled(struct drm_i915_private *dev_priv,
 	struct i915_power_well *power_well;
 	bool is_enabled;
 
-	if (dev_priv->pm.suspended)
+	if (dev_priv->runtime_pm.suspended)
 		return false;
 
 	is_enabled = true;
@@ -368,7 +368,7 @@ static void hsw_power_well_enable(struct drm_i915_private *dev_priv,
 {
 	enum i915_power_well_id id = power_well->id;
 	bool wait_fuses = power_well->hsw.has_fuses;
-	enum skl_power_gate pg;
+	enum skl_power_gate uninitialized_var(pg);
 	u32 val;
 
 	if (wait_fuses) {
@@ -785,7 +785,7 @@ static void vlv_set_power_well(struct drm_i915_private *dev_priv,
 	state = enable ? PUNIT_PWRGT_PWR_ON(power_well_id) :
 			 PUNIT_PWRGT_PWR_GATE(power_well_id);
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 #define COND \
 	((vlv_punit_read(dev_priv, PUNIT_REG_PWRGT_STATUS) & mask) == state)
@@ -806,7 +806,7 @@ static void vlv_set_power_well(struct drm_i915_private *dev_priv,
 #undef COND
 
 out:
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 }
 
 static void vlv_power_well_enable(struct drm_i915_private *dev_priv,
@@ -833,7 +833,7 @@ static bool vlv_power_well_enabled(struct drm_i915_private *dev_priv,
 	mask = PUNIT_PWRGT_MASK(power_well_id);
 	ctrl = PUNIT_PWRGT_PWR_ON(power_well_id);
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	state = vlv_punit_read(dev_priv, PUNIT_REG_PWRGT_STATUS) & mask;
 	/*
@@ -852,7 +852,7 @@ static bool vlv_power_well_enabled(struct drm_i915_private *dev_priv,
 	ctrl = vlv_punit_read(dev_priv, PUNIT_REG_PWRGT_CTRL) & mask;
 	WARN_ON(ctrl != state);
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	return enabled;
 }
@@ -1364,7 +1364,7 @@ static bool chv_pipe_power_well_enabled(struct drm_i915_private *dev_priv,
 	bool enabled;
 	u32 state, ctrl;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 	state = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ) & DP_SSS_MASK(pipe);
 	/*
@@ -1381,7 +1381,7 @@ static bool chv_pipe_power_well_enabled(struct drm_i915_private *dev_priv,
 	ctrl = vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ) & DP_SSC_MASK(pipe);
 	WARN_ON(ctrl << 16 != state);
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 
 	return enabled;
 }
@@ -1396,7 +1396,7 @@ static void chv_set_pipe_power_well(struct drm_i915_private *dev_priv,
 
 	state = enable ? DP_SSS_PWR_ON(pipe) : DP_SSS_PWR_GATE(pipe);
 
-	mutex_lock(&dev_priv->rps.hw_lock);
+	mutex_lock(&dev_priv->pcu_lock);
 
 #define COND \
 	((vlv_punit_read(dev_priv, PUNIT_REG_DSPFREQ) & DP_SSS_MASK(pipe)) == state)
@@ -1417,7 +1417,7 @@ static void chv_set_pipe_power_well(struct drm_i915_private *dev_priv,
 #undef COND
 
 out:
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	mutex_unlock(&dev_priv->pcu_lock);
 }
 
 static void chv_pipe_power_well_enable(struct drm_i915_private *dev_priv,
@@ -2809,6 +2809,9 @@ static void cnl_display_core_init(struct drm_i915_private *dev_priv, bool resume
 
 	/* 6. Enable DBUF */
 	gen9_dbuf_enable(dev_priv);
+
+	if (resume && dev_priv->csr.dmc_payload)
+		intel_csr_load_program(dev_priv);
 }
 
 static void cnl_display_core_uninit(struct drm_i915_private *dev_priv)
@@ -3125,7 +3128,7 @@ void intel_runtime_pm_get(struct drm_i915_private *dev_priv)
 	ret = pm_runtime_get_sync(kdev);
 	WARN_ONCE(ret < 0, "pm_runtime_get_sync() failed: %d\n", ret);
 
-	atomic_inc(&dev_priv->pm.wakeref_count);
+	atomic_inc(&dev_priv->runtime_pm.wakeref_count);
 	assert_rpm_wakelock_held(dev_priv);
 }
 
@@ -3159,7 +3162,7 @@ bool intel_runtime_pm_get_if_in_use(struct drm_i915_private *dev_priv)
 			return false;
 	}
 
-	atomic_inc(&dev_priv->pm.wakeref_count);
+	atomic_inc(&dev_priv->runtime_pm.wakeref_count);
 	assert_rpm_wakelock_held(dev_priv);
 
 	return true;
@@ -3190,7 +3193,7 @@ void intel_runtime_pm_get_noresume(struct drm_i915_private *dev_priv)
 	assert_rpm_wakelock_held(dev_priv);
 	pm_runtime_get_noresume(kdev);
 
-	atomic_inc(&dev_priv->pm.wakeref_count);
+	atomic_inc(&dev_priv->runtime_pm.wakeref_count);
 }
 
 /**
@@ -3207,7 +3210,7 @@ void intel_runtime_pm_put(struct drm_i915_private *dev_priv)
 	struct device *kdev = &pdev->dev;
 
 	assert_rpm_wakelock_held(dev_priv);
-	atomic_dec(&dev_priv->pm.wakeref_count);
+	atomic_dec(&dev_priv->runtime_pm.wakeref_count);
 
 	pm_runtime_mark_last_busy(kdev);
 	pm_runtime_put_autosuspend(kdev);
diff --git a/drivers/gpu/drm/i915/intel_sideband.c b/drivers/gpu/drm/i915/intel_sideband.c
index 7d971cb56116..75c872bb8cc9 100644
--- a/drivers/gpu/drm/i915/intel_sideband.c
+++ b/drivers/gpu/drm/i915/intel_sideband.c
@@ -81,7 +81,7 @@ u32 vlv_punit_read(struct drm_i915_private *dev_priv, u32 addr)
 {
 	u32 val = 0;
 
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	WARN_ON(!mutex_is_locked(&dev_priv->pcu_lock));
 
 	mutex_lock(&dev_priv->sb_lock);
 	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_PUNIT,
@@ -95,7 +95,7 @@ int vlv_punit_write(struct drm_i915_private *dev_priv, u32 addr, u32 val)
 {
 	int err;
 
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	WARN_ON(!mutex_is_locked(&dev_priv->pcu_lock));
 
 	mutex_lock(&dev_priv->sb_lock);
 	err = vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_PUNIT,
@@ -125,7 +125,7 @@ u32 vlv_nc_read(struct drm_i915_private *dev_priv, u8 addr)
 {
 	u32 val = 0;
 
-	WARN_ON(!mutex_is_locked(&dev_priv->rps.hw_lock));
+	WARN_ON(!mutex_is_locked(&dev_priv->pcu_lock));
 
 	mutex_lock(&dev_priv->sb_lock);
 	vlv_sideband_rw(dev_priv, PCI_DEVFN(0, 0), IOSF_PORT_NC,
diff --git a/drivers/gpu/drm/i915/intel_sprite.c b/drivers/gpu/drm/i915/intel_sprite.c
index 28a1209d87e2..86fc9b529f2d 100644
--- a/drivers/gpu/drm/i915/intel_sprite.c
+++ b/drivers/gpu/drm/i915/intel_sprite.c
@@ -66,7 +66,13 @@ int intel_usecs_to_scanlines(const struct drm_display_mode *adjusted_mode,
 			    1000 * adjusted_mode->crtc_htotal);
 }
 
+/* FIXME: We should instead only take spinlocks once for the entire update
+ * instead of once per mmio. */
+#if IS_ENABLED(CONFIG_PROVE_LOCKING)
+#define VBLANK_EVASION_TIME_US 250
+#else
 #define VBLANK_EVASION_TIME_US 100
+#endif
 
 /**
  * intel_pipe_update_start() - start update of a set of display registers
diff --git a/drivers/gpu/drm/i915/intel_uc.c b/drivers/gpu/drm/i915/intel_uc.c
index 277477890240..7b938e822fde 100644
--- a/drivers/gpu/drm/i915/intel_uc.c
+++ b/drivers/gpu/drm/i915/intel_uc.c
@@ -22,22 +22,9 @@
  *
  */
 
-#include "i915_drv.h"
 #include "intel_uc.h"
-#include <linux/firmware.h>
-
-/* Cleans up uC firmware by releasing the firmware GEM obj.
- */
-static void __intel_uc_fw_fini(struct intel_uc_fw *uc_fw)
-{
-	struct drm_i915_gem_object *obj;
-
-	obj = fetch_and_zero(&uc_fw->obj);
-	if (obj)
-		i915_gem_object_put(obj);
-
-	uc_fw->fetch_status = INTEL_UC_FIRMWARE_NONE;
-}
+#include "i915_drv.h"
+#include "i915_guc_submission.h"
 
 /* Reset GuC providing us with fresh state for both GuC and HuC.
  */
@@ -94,198 +81,34 @@ void intel_uc_sanitize_options(struct drm_i915_private *dev_priv)
 		i915_modparams.enable_guc_submission = HAS_GUC_SCHED(dev_priv);
 }
 
-static void gen8_guc_raise_irq(struct intel_guc *guc)
-{
-	struct drm_i915_private *dev_priv = guc_to_i915(guc);
-
-	I915_WRITE(GUC_SEND_INTERRUPT, GUC_SEND_TRIGGER);
-}
-
 void intel_uc_init_early(struct drm_i915_private *dev_priv)
 {
-	struct intel_guc *guc = &dev_priv->guc;
-
-	intel_guc_ct_init_early(&guc->ct);
-
-	mutex_init(&guc->send_mutex);
-	guc->send = intel_guc_send_nop;
-	guc->notify = gen8_guc_raise_irq;
-}
-
-static void fetch_uc_fw(struct drm_i915_private *dev_priv,
-			struct intel_uc_fw *uc_fw)
-{
-	struct pci_dev *pdev = dev_priv->drm.pdev;
-	struct drm_i915_gem_object *obj;
-	const struct firmware *fw = NULL;
-	struct uc_css_header *css;
-	size_t size;
-	int err;
-
-	if (!uc_fw->path)
-		return;
-
-	uc_fw->fetch_status = INTEL_UC_FIRMWARE_PENDING;
-
-	DRM_DEBUG_DRIVER("before requesting firmware: uC fw fetch status %s\n",
-			 intel_uc_fw_status_repr(uc_fw->fetch_status));
-
-	err = request_firmware(&fw, uc_fw->path, &pdev->dev);
-	if (err)
-		goto fail;
-	if (!fw)
-		goto fail;
-
-	DRM_DEBUG_DRIVER("fetch uC fw from %s succeeded, fw %p\n",
-			 uc_fw->path, fw);
-
-	/* Check the size of the blob before examining buffer contents */
-	if (fw->size < sizeof(struct uc_css_header)) {
-		DRM_NOTE("Firmware header is missing\n");
-		goto fail;
-	}
-
-	css = (struct uc_css_header *)fw->data;
-
-	/* Firmware bits always start from header */
-	uc_fw->header_offset = 0;
-	uc_fw->header_size = (css->header_size_dw - css->modulus_size_dw -
-			      css->key_size_dw - css->exponent_size_dw) * sizeof(u32);
-
-	if (uc_fw->header_size != sizeof(struct uc_css_header)) {
-		DRM_NOTE("CSS header definition mismatch\n");
-		goto fail;
-	}
-
-	/* then, uCode */
-	uc_fw->ucode_offset = uc_fw->header_offset + uc_fw->header_size;
-	uc_fw->ucode_size = (css->size_dw - css->header_size_dw) * sizeof(u32);
-
-	/* now RSA */
-	if (css->key_size_dw != UOS_RSA_SCRATCH_MAX_COUNT) {
-		DRM_NOTE("RSA key size is bad\n");
-		goto fail;
-	}
-	uc_fw->rsa_offset = uc_fw->ucode_offset + uc_fw->ucode_size;
-	uc_fw->rsa_size = css->key_size_dw * sizeof(u32);
-
-	/* At least, it should have header, uCode and RSA. Size of all three. */
-	size = uc_fw->header_size + uc_fw->ucode_size + uc_fw->rsa_size;
-	if (fw->size < size) {
-		DRM_NOTE("Missing firmware components\n");
-		goto fail;
-	}
-
-	/*
-	 * The GuC firmware image has the version number embedded at a
-	 * well-known offset within the firmware blob; note that major / minor
-	 * version are TWO bytes each (i.e. u16), although all pointers and
-	 * offsets are defined in terms of bytes (u8).
-	 */
-	switch (uc_fw->type) {
-	case INTEL_UC_FW_TYPE_GUC:
-		/* Header and uCode will be loaded to WOPCM. Size of the two. */
-		size = uc_fw->header_size + uc_fw->ucode_size;
-
-		/* Top 32k of WOPCM is reserved (8K stack + 24k RC6 context). */
-		if (size > intel_guc_wopcm_size(dev_priv)) {
-			DRM_ERROR("Firmware is too large to fit in WOPCM\n");
-			goto fail;
-		}
-		uc_fw->major_ver_found = css->guc.sw_version >> 16;
-		uc_fw->minor_ver_found = css->guc.sw_version & 0xFFFF;
-		break;
-
-	case INTEL_UC_FW_TYPE_HUC:
-		uc_fw->major_ver_found = css->huc.sw_version >> 16;
-		uc_fw->minor_ver_found = css->huc.sw_version & 0xFFFF;
-		break;
-
-	default:
-		DRM_ERROR("Unknown firmware type %d\n", uc_fw->type);
-		err = -ENOEXEC;
-		goto fail;
-	}
-
-	if (uc_fw->major_ver_wanted == 0 && uc_fw->minor_ver_wanted == 0) {
-		DRM_NOTE("Skipping %s firmware version check\n",
-			 intel_uc_fw_type_repr(uc_fw->type));
-	} else if (uc_fw->major_ver_found != uc_fw->major_ver_wanted ||
-		   uc_fw->minor_ver_found < uc_fw->minor_ver_wanted) {
-		DRM_NOTE("%s firmware version %d.%d, required %d.%d\n",
-			 intel_uc_fw_type_repr(uc_fw->type),
-			 uc_fw->major_ver_found, uc_fw->minor_ver_found,
-			 uc_fw->major_ver_wanted, uc_fw->minor_ver_wanted);
-		err = -ENOEXEC;
-		goto fail;
-	}
-
-	DRM_DEBUG_DRIVER("firmware version %d.%d OK (minimum %d.%d)\n",
-			 uc_fw->major_ver_found, uc_fw->minor_ver_found,
-			 uc_fw->major_ver_wanted, uc_fw->minor_ver_wanted);
-
-	obj = i915_gem_object_create_from_data(dev_priv, fw->data, fw->size);
-	if (IS_ERR(obj)) {
-		err = PTR_ERR(obj);
-		goto fail;
-	}
-
-	uc_fw->obj = obj;
-	uc_fw->size = fw->size;
-
-	DRM_DEBUG_DRIVER("uC fw fetch status SUCCESS, obj %p\n",
-			 uc_fw->obj);
-
-	release_firmware(fw);
-	uc_fw->fetch_status = INTEL_UC_FIRMWARE_SUCCESS;
-	return;
-
-fail:
-	DRM_WARN("Failed to fetch valid uC firmware from %s (error %d)\n",
-		 uc_fw->path, err);
-	DRM_DEBUG_DRIVER("uC fw fetch status FAIL; err %d, fw %p, obj %p\n",
-			 err, fw, uc_fw->obj);
-
-	release_firmware(fw);		/* OK even if fw is NULL */
-	uc_fw->fetch_status = INTEL_UC_FIRMWARE_FAIL;
+	intel_guc_init_early(&dev_priv->guc);
 }
 
 void intel_uc_init_fw(struct drm_i915_private *dev_priv)
 {
-	fetch_uc_fw(dev_priv, &dev_priv->huc.fw);
-	fetch_uc_fw(dev_priv, &dev_priv->guc.fw);
+	intel_uc_fw_fetch(dev_priv, &dev_priv->huc.fw);
+	intel_uc_fw_fetch(dev_priv, &dev_priv->guc.fw);
 }
 
 void intel_uc_fini_fw(struct drm_i915_private *dev_priv)
 {
-	__intel_uc_fw_fini(&dev_priv->guc.fw);
-	__intel_uc_fw_fini(&dev_priv->huc.fw);
-}
-
-static inline i915_reg_t guc_send_reg(struct intel_guc *guc, u32 i)
-{
-	GEM_BUG_ON(!guc->send_regs.base);
-	GEM_BUG_ON(!guc->send_regs.count);
-	GEM_BUG_ON(i >= guc->send_regs.count);
-
-	return _MMIO(guc->send_regs.base + 4 * i);
+	intel_uc_fw_fini(&dev_priv->guc.fw);
+	intel_uc_fw_fini(&dev_priv->huc.fw);
 }
 
-static void guc_init_send_regs(struct intel_guc *guc)
+/**
+ * intel_uc_init_mmio - setup uC MMIO access
+ *
+ * @dev_priv: device private
+ *
+ * Setup minimal state necessary for MMIO accesses later in the
+ * initialization sequence.
+ */
+void intel_uc_init_mmio(struct drm_i915_private *dev_priv)
 {
-	struct drm_i915_private *dev_priv = guc_to_i915(guc);
-	enum forcewake_domains fw_domains = 0;
-	unsigned int i;
-
-	guc->send_regs.base = i915_mmio_reg_offset(SOFT_SCRATCH(0));
-	guc->send_regs.count = SOFT_SCRATCH_COUNT - 1;
-
-	for (i = 0; i < guc->send_regs.count; i++) {
-		fw_domains |= intel_uncore_forcewake_for_reg(dev_priv,
-					guc_send_reg(guc, i),
-					FW_REG_READ | FW_REG_WRITE);
-	}
-	guc->send_regs.fw_domains = fw_domains;
+	intel_guc_init_send_regs(&dev_priv->guc);
 }
 
 static void guc_capture_load_err_log(struct intel_guc *guc)
@@ -309,8 +132,6 @@ static int guc_enable_communication(struct intel_guc *guc)
 {
 	struct drm_i915_private *dev_priv = guc_to_i915(guc);
 
-	guc_init_send_regs(guc);
-
 	if (HAS_GUC_CT(dev_priv))
 		return intel_guc_enable_ct(guc);
 
@@ -328,27 +149,6 @@ static void guc_disable_communication(struct intel_guc *guc)
 	guc->send = intel_guc_send_nop;
 }
 
-/**
- * intel_guc_auth_huc() - Send action to GuC to authenticate HuC ucode
- * @guc: intel_guc structure
- * @rsa_offset: rsa offset w.r.t ggtt base of huc vma
- *
- * Triggers a HuC firmware authentication request to the GuC via intel_guc_send
- * INTEL_GUC_ACTION_AUTHENTICATE_HUC interface. This function is invoked by
- * intel_huc_auth().
- *
- * Return:	non-zero code on error
- */
-int intel_guc_auth_huc(struct intel_guc *guc, u32 rsa_offset)
-{
-	u32 action[] = {
-		INTEL_GUC_ACTION_AUTHENTICATE_HUC,
-		rsa_offset
-	};
-
-	return intel_guc_send(guc, action, ARRAY_SIZE(action));
-}
-
 int intel_uc_init_hw(struct drm_i915_private *dev_priv)
 {
 	struct intel_guc *guc = &dev_priv->guc;
@@ -480,82 +280,3 @@ void intel_uc_fini_hw(struct drm_i915_private *dev_priv)
 
 	i915_ggtt_disable_guc(dev_priv);
 }
-
-int intel_guc_send_nop(struct intel_guc *guc, const u32 *action, u32 len)
-{
-	WARN(1, "Unexpected send: action=%#x\n", *action);
-	return -ENODEV;
-}
-
-/*
- * This function implements the MMIO based host to GuC interface.
- */
-int intel_guc_send_mmio(struct intel_guc *guc, const u32 *action, u32 len)
-{
-	struct drm_i915_private *dev_priv = guc_to_i915(guc);
-	u32 status;
-	int i;
-	int ret;
-
-	GEM_BUG_ON(!len);
-	GEM_BUG_ON(len > guc->send_regs.count);
-
-	/* If CT is available, we expect to use MMIO only during init/fini */
-	GEM_BUG_ON(HAS_GUC_CT(dev_priv) &&
-		*action != INTEL_GUC_ACTION_REGISTER_COMMAND_TRANSPORT_BUFFER &&
-		*action != INTEL_GUC_ACTION_DEREGISTER_COMMAND_TRANSPORT_BUFFER);
-
-	mutex_lock(&guc->send_mutex);
-	intel_uncore_forcewake_get(dev_priv, guc->send_regs.fw_domains);
-
-	for (i = 0; i < len; i++)
-		I915_WRITE(guc_send_reg(guc, i), action[i]);
-
-	POSTING_READ(guc_send_reg(guc, i - 1));
-
-	intel_guc_notify(guc);
-
-	/*
-	 * No GuC command should ever take longer than 10ms.
-	 * Fast commands should still complete in 10us.
-	 */
-	ret = __intel_wait_for_register_fw(dev_priv,
-					   guc_send_reg(guc, 0),
-					   INTEL_GUC_RECV_MASK,
-					   INTEL_GUC_RECV_MASK,
-					   10, 10, &status);
-	if (status != INTEL_GUC_STATUS_SUCCESS) {
-		/*
-		 * Either the GuC explicitly returned an error (which
-		 * we convert to -EIO here) or no response at all was
-		 * received within the timeout limit (-ETIMEDOUT)
-		 */
-		if (ret != -ETIMEDOUT)
-			ret = -EIO;
-
-		DRM_WARN("INTEL_GUC_SEND: Action 0x%X failed;"
-			 " ret=%d status=0x%08X response=0x%08X\n",
-			 action[0], ret, status, I915_READ(SOFT_SCRATCH(15)));
-	}
-
-	intel_uncore_forcewake_put(dev_priv, guc->send_regs.fw_domains);
-	mutex_unlock(&guc->send_mutex);
-
-	return ret;
-}
-
-int intel_guc_sample_forcewake(struct intel_guc *guc)
-{
-	struct drm_i915_private *dev_priv = guc_to_i915(guc);
-	u32 action[2];
-
-	action[0] = INTEL_GUC_ACTION_SAMPLE_FORCEWAKE;
-	/* WaRsDisableCoarsePowerGating:skl,bxt */
-	if (!intel_enable_rc6() || NEEDS_WaRsDisableCoarsePowerGating(dev_priv))
-		action[1] = 0;
-	else
-		/* bit 0 and 1 are for Render and Media domain separately */
-		action[1] = GUC_FORCEWAKE_RENDER | GUC_FORCEWAKE_MEDIA;
-
-	return intel_guc_send(guc, action, ARRAY_SIZE(action));
-}
diff --git a/drivers/gpu/drm/i915/intel_uc.h b/drivers/gpu/drm/i915/intel_uc.h
index 6966349ed737..e18d3bb02088 100644
--- a/drivers/gpu/drm/i915/intel_uc.h
+++ b/drivers/gpu/drm/i915/intel_uc.h
@@ -24,237 +24,15 @@
 #ifndef _INTEL_UC_H_
 #define _INTEL_UC_H_
 
-#include "intel_guc_fwif.h"
-#include "i915_guc_reg.h"
-#include "intel_ringbuffer.h"
-#include "intel_guc_ct.h"
-#include "i915_vma.h"
+#include "intel_guc.h"
+#include "intel_huc.h"
 
-struct drm_i915_gem_request;
-
-/*
- * This structure primarily describes the GEM object shared with the GuC.
- * The specs sometimes refer to this object as a "GuC context", but we use
- * the term "client" to avoid confusion with hardware contexts. This
- * GEM object is held for the entire lifetime of our interaction with
- * the GuC, being allocated before the GuC is loaded with its firmware.
- * Because there's no way to update the address used by the GuC after
- * initialisation, the shared object must stay pinned into the GGTT as
- * long as the GuC is in use. We also keep the first page (only) mapped
- * into kernel address space, as it includes shared data that must be
- * updated on every request submission.
- *
- * The single GEM object described here is actually made up of several
- * separate areas, as far as the GuC is concerned. The first page (kept
- * kmap'd) includes the "process descriptor" which holds sequence data for
- * the doorbell, and one cacheline which actually *is* the doorbell; a
- * write to this will "ring the doorbell" (i.e. send an interrupt to the
- * GuC). The subsequent  pages of the client object constitute the work
- * queue (a circular array of work items), again described in the process
- * descriptor. Work queue pages are mapped momentarily as required.
- */
-struct i915_guc_client {
-	struct i915_vma *vma;
-	void *vaddr;
-	struct i915_gem_context *owner;
-	struct intel_guc *guc;
-
-	uint32_t engines;		/* bitmap of (host) engine ids	*/
-	uint32_t priority;
-	u32 stage_id;
-	uint32_t proc_desc_offset;
-
-	u16 doorbell_id;
-	unsigned long doorbell_offset;
-
-	spinlock_t wq_lock;
-	/* Per-engine counts of GuC submissions */
-	uint64_t submissions[I915_NUM_ENGINES];
-};
-
-enum intel_uc_fw_status {
-	INTEL_UC_FIRMWARE_FAIL = -1,
-	INTEL_UC_FIRMWARE_NONE = 0,
-	INTEL_UC_FIRMWARE_PENDING,
-	INTEL_UC_FIRMWARE_SUCCESS
-};
-
-/* User-friendly representation of an enum */
-static inline
-const char *intel_uc_fw_status_repr(enum intel_uc_fw_status status)
-{
-	switch (status) {
-	case INTEL_UC_FIRMWARE_FAIL:
-		return "FAIL";
-	case INTEL_UC_FIRMWARE_NONE:
-		return "NONE";
-	case INTEL_UC_FIRMWARE_PENDING:
-		return "PENDING";
-	case INTEL_UC_FIRMWARE_SUCCESS:
-		return "SUCCESS";
-	}
-	return "<invalid>";
-}
-
-enum intel_uc_fw_type {
-	INTEL_UC_FW_TYPE_GUC,
-	INTEL_UC_FW_TYPE_HUC
-};
-
-/* User-friendly representation of an enum */
-static inline const char *intel_uc_fw_type_repr(enum intel_uc_fw_type type)
-{
-	switch (type) {
-	case INTEL_UC_FW_TYPE_GUC:
-		return "GuC";
-	case INTEL_UC_FW_TYPE_HUC:
-		return "HuC";
-	}
-	return "uC";
-}
-
-/*
- * This structure encapsulates all the data needed during the process
- * of fetching, caching, and loading the firmware image into the GuC.
- */
-struct intel_uc_fw {
-	const char *path;
-	size_t size;
-	struct drm_i915_gem_object *obj;
-	enum intel_uc_fw_status fetch_status;
-	enum intel_uc_fw_status load_status;
-
-	uint16_t major_ver_wanted;
-	uint16_t minor_ver_wanted;
-	uint16_t major_ver_found;
-	uint16_t minor_ver_found;
-
-	enum intel_uc_fw_type type;
-	uint32_t header_size;
-	uint32_t header_offset;
-	uint32_t rsa_size;
-	uint32_t rsa_offset;
-	uint32_t ucode_size;
-	uint32_t ucode_offset;
-};
-
-struct intel_guc_log {
-	uint32_t flags;
-	struct i915_vma *vma;
-	/* The runtime stuff gets created only when GuC logging gets enabled */
-	struct {
-		void *buf_addr;
-		struct workqueue_struct *flush_wq;
-		struct work_struct flush_work;
-		struct rchan *relay_chan;
-	} runtime;
-	/* logging related stats */
-	u32 capture_miss_count;
-	u32 flush_interrupt_count;
-	u32 prev_overflow_count[GUC_MAX_LOG_BUFFER];
-	u32 total_overflow_count[GUC_MAX_LOG_BUFFER];
-	u32 flush_count[GUC_MAX_LOG_BUFFER];
-};
-
-struct intel_guc {
-	struct intel_uc_fw fw;
-	struct intel_guc_log log;
-	struct intel_guc_ct ct;
-
-	/* Log snapshot if GuC errors during load */
-	struct drm_i915_gem_object *load_err_log;
-
-	/* intel_guc_recv interrupt related state */
-	bool interrupts_enabled;
-
-	struct i915_vma *ads_vma;
-	struct i915_vma *stage_desc_pool;
-	void *stage_desc_pool_vaddr;
-	struct ida stage_ids;
-
-	struct i915_guc_client *execbuf_client;
-
-	DECLARE_BITMAP(doorbell_bitmap, GUC_NUM_DOORBELLS);
-	uint32_t db_cacheline;		/* Cyclic counter mod pagesize	*/
-
-	/* GuC's FW specific registers used in MMIO send */
-	struct {
-		u32 base;
-		unsigned int count;
-		enum forcewake_domains fw_domains;
-	} send_regs;
-
-	/* To serialize the intel_guc_send actions */
-	struct mutex send_mutex;
-
-	/* GuC's FW specific send function */
-	int (*send)(struct intel_guc *guc, const u32 *data, u32 len);
-
-	/* GuC's FW specific notify function */
-	void (*notify)(struct intel_guc *guc);
-};
-
-struct intel_huc {
-	/* Generic uC firmware management */
-	struct intel_uc_fw fw;
-
-	/* HuC-specific additions */
-};
-
-/* intel_uc.c */
 void intel_uc_sanitize_options(struct drm_i915_private *dev_priv);
 void intel_uc_init_early(struct drm_i915_private *dev_priv);
+void intel_uc_init_mmio(struct drm_i915_private *dev_priv);
 void intel_uc_init_fw(struct drm_i915_private *dev_priv);
 void intel_uc_fini_fw(struct drm_i915_private *dev_priv);
 int intel_uc_init_hw(struct drm_i915_private *dev_priv);
 void intel_uc_fini_hw(struct drm_i915_private *dev_priv);
-int intel_guc_sample_forcewake(struct intel_guc *guc);
-int intel_guc_send_nop(struct intel_guc *guc, const u32 *action, u32 len);
-int intel_guc_send_mmio(struct intel_guc *guc, const u32 *action, u32 len);
-int intel_guc_auth_huc(struct intel_guc *guc, u32 rsa_offset);
-
-static inline int intel_guc_send(struct intel_guc *guc, const u32 *action, u32 len)
-{
-	return guc->send(guc, action, len);
-}
-
-static inline void intel_guc_notify(struct intel_guc *guc)
-{
-	guc->notify(guc);
-}
-
-/* intel_guc_loader.c */
-int intel_guc_select_fw(struct intel_guc *guc);
-int intel_guc_init_hw(struct intel_guc *guc);
-int intel_guc_suspend(struct drm_i915_private *dev_priv);
-int intel_guc_resume(struct drm_i915_private *dev_priv);
-u32 intel_guc_wopcm_size(struct drm_i915_private *dev_priv);
-
-/* i915_guc_submission.c */
-int i915_guc_submission_init(struct drm_i915_private *dev_priv);
-int i915_guc_submission_enable(struct drm_i915_private *dev_priv);
-void i915_guc_submission_disable(struct drm_i915_private *dev_priv);
-void i915_guc_submission_fini(struct drm_i915_private *dev_priv);
-struct i915_vma *intel_guc_allocate_vma(struct intel_guc *guc, u32 size);
-
-/* intel_guc_log.c */
-int intel_guc_log_create(struct intel_guc *guc);
-void intel_guc_log_destroy(struct intel_guc *guc);
-int i915_guc_log_control(struct drm_i915_private *dev_priv, u64 control_val);
-void i915_guc_log_register(struct drm_i915_private *dev_priv);
-void i915_guc_log_unregister(struct drm_i915_private *dev_priv);
-
-static inline u32 guc_ggtt_offset(struct i915_vma *vma)
-{
-	u32 offset = i915_ggtt_offset(vma);
-	GEM_BUG_ON(offset < GUC_WOPCM_TOP);
-	GEM_BUG_ON(range_overflows_t(u64, offset, vma->size, GUC_GGTT_TOP));
-	return offset;
-}
-
-/* intel_huc.c */
-void intel_huc_select_fw(struct intel_huc *huc);
-void intel_huc_init_hw(struct intel_huc *huc);
-void intel_huc_auth(struct intel_huc *huc);
 
 #endif
diff --git a/drivers/gpu/drm/i915/intel_uc_fw.c b/drivers/gpu/drm/i915/intel_uc_fw.c
new file mode 100644
index 000000000000..766b1cbdfbd7
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_uc_fw.c
@@ -0,0 +1,193 @@
+/*
+ * Copyright © 2016-2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#include <linux/firmware.h>
+
+#include "intel_uc_fw.h"
+#include "i915_drv.h"
+
+/**
+ * intel_uc_fw_fetch - fetch uC firmware
+ *
+ * @dev_priv: device private
+ * @uc_fw: uC firmware
+ *
+ * Fetch uC firmware into GEM obj.
+ */
+void intel_uc_fw_fetch(struct drm_i915_private *dev_priv,
+		       struct intel_uc_fw *uc_fw)
+{
+	struct pci_dev *pdev = dev_priv->drm.pdev;
+	struct drm_i915_gem_object *obj;
+	const struct firmware *fw = NULL;
+	struct uc_css_header *css;
+	size_t size;
+	int err;
+
+	if (!uc_fw->path)
+		return;
+
+	uc_fw->fetch_status = INTEL_UC_FIRMWARE_PENDING;
+
+	DRM_DEBUG_DRIVER("before requesting firmware: uC fw fetch status %s\n",
+			 intel_uc_fw_status_repr(uc_fw->fetch_status));
+
+	err = request_firmware(&fw, uc_fw->path, &pdev->dev);
+	if (err)
+		goto fail;
+	if (!fw)
+		goto fail;
+
+	DRM_DEBUG_DRIVER("fetch uC fw from %s succeeded, fw %p\n",
+			 uc_fw->path, fw);
+
+	/* Check the size of the blob before examining buffer contents */
+	if (fw->size < sizeof(struct uc_css_header)) {
+		DRM_NOTE("Firmware header is missing\n");
+		goto fail;
+	}
+
+	css = (struct uc_css_header *)fw->data;
+
+	/* Firmware bits always start from header */
+	uc_fw->header_offset = 0;
+	uc_fw->header_size = (css->header_size_dw - css->modulus_size_dw -
+			      css->key_size_dw - css->exponent_size_dw) *
+			     sizeof(u32);
+
+	if (uc_fw->header_size != sizeof(struct uc_css_header)) {
+		DRM_NOTE("CSS header definition mismatch\n");
+		goto fail;
+	}
+
+	/* then, uCode */
+	uc_fw->ucode_offset = uc_fw->header_offset + uc_fw->header_size;
+	uc_fw->ucode_size = (css->size_dw - css->header_size_dw) * sizeof(u32);
+
+	/* now RSA */
+	if (css->key_size_dw != UOS_RSA_SCRATCH_MAX_COUNT) {
+		DRM_NOTE("RSA key size is bad\n");
+		goto fail;
+	}
+	uc_fw->rsa_offset = uc_fw->ucode_offset + uc_fw->ucode_size;
+	uc_fw->rsa_size = css->key_size_dw * sizeof(u32);
+
+	/* At least, it should have header, uCode and RSA. Size of all three. */
+	size = uc_fw->header_size + uc_fw->ucode_size + uc_fw->rsa_size;
+	if (fw->size < size) {
+		DRM_NOTE("Missing firmware components\n");
+		goto fail;
+	}
+
+	/*
+	 * The GuC firmware image has the version number embedded at a
+	 * well-known offset within the firmware blob; note that major / minor
+	 * version are TWO bytes each (i.e. u16), although all pointers and
+	 * offsets are defined in terms of bytes (u8).
+	 */
+	switch (uc_fw->type) {
+	case INTEL_UC_FW_TYPE_GUC:
+		/* Header and uCode will be loaded to WOPCM. Size of the two. */
+		size = uc_fw->header_size + uc_fw->ucode_size;
+
+		/* Top 32k of WOPCM is reserved (8K stack + 24k RC6 context). */
+		if (size > intel_guc_wopcm_size(dev_priv)) {
+			DRM_ERROR("Firmware is too large to fit in WOPCM\n");
+			goto fail;
+		}
+		uc_fw->major_ver_found = css->guc.sw_version >> 16;
+		uc_fw->minor_ver_found = css->guc.sw_version & 0xFFFF;
+		break;
+
+	case INTEL_UC_FW_TYPE_HUC:
+		uc_fw->major_ver_found = css->huc.sw_version >> 16;
+		uc_fw->minor_ver_found = css->huc.sw_version & 0xFFFF;
+		break;
+
+	default:
+		DRM_ERROR("Unknown firmware type %d\n", uc_fw->type);
+		err = -ENOEXEC;
+		goto fail;
+	}
+
+	if (uc_fw->major_ver_wanted == 0 && uc_fw->minor_ver_wanted == 0) {
+		DRM_NOTE("Skipping %s firmware version check\n",
+			 intel_uc_fw_type_repr(uc_fw->type));
+	} else if (uc_fw->major_ver_found != uc_fw->major_ver_wanted ||
+		   uc_fw->minor_ver_found < uc_fw->minor_ver_wanted) {
+		DRM_NOTE("%s firmware version %d.%d, required %d.%d\n",
+			 intel_uc_fw_type_repr(uc_fw->type),
+			 uc_fw->major_ver_found, uc_fw->minor_ver_found,
+			 uc_fw->major_ver_wanted, uc_fw->minor_ver_wanted);
+		err = -ENOEXEC;
+		goto fail;
+	}
+
+	DRM_DEBUG_DRIVER("firmware version %d.%d OK (minimum %d.%d)\n",
+			 uc_fw->major_ver_found, uc_fw->minor_ver_found,
+			 uc_fw->major_ver_wanted, uc_fw->minor_ver_wanted);
+
+	obj = i915_gem_object_create_from_data(dev_priv, fw->data, fw->size);
+	if (IS_ERR(obj)) {
+		err = PTR_ERR(obj);
+		goto fail;
+	}
+
+	uc_fw->obj = obj;
+	uc_fw->size = fw->size;
+
+	DRM_DEBUG_DRIVER("uC fw fetch status SUCCESS, obj %p\n",
+			 uc_fw->obj);
+
+	release_firmware(fw);
+	uc_fw->fetch_status = INTEL_UC_FIRMWARE_SUCCESS;
+	return;
+
+fail:
+	DRM_WARN("Failed to fetch valid uC firmware from %s (error %d)\n",
+		 uc_fw->path, err);
+	DRM_DEBUG_DRIVER("uC fw fetch status FAIL; err %d, fw %p, obj %p\n",
+			 err, fw, uc_fw->obj);
+
+	release_firmware(fw);		/* OK even if fw is NULL */
+	uc_fw->fetch_status = INTEL_UC_FIRMWARE_FAIL;
+}
+
+/**
+ * intel_uc_fw_fini - cleanup uC firmware
+ *
+ * @uc_fw: uC firmware
+ *
+ * Cleans up uC firmware by releasing the firmware GEM obj.
+ */
+void intel_uc_fw_fini(struct intel_uc_fw *uc_fw)
+{
+	struct drm_i915_gem_object *obj;
+
+	obj = fetch_and_zero(&uc_fw->obj);
+	if (obj)
+		i915_gem_object_put(obj);
+
+	uc_fw->fetch_status = INTEL_UC_FIRMWARE_NONE;
+}
diff --git a/drivers/gpu/drm/i915/intel_uc_fw.h b/drivers/gpu/drm/i915/intel_uc_fw.h
new file mode 100644
index 000000000000..c3e9af4b9bf0
--- /dev/null
+++ b/drivers/gpu/drm/i915/intel_uc_fw.h
@@ -0,0 +1,107 @@
+/*
+ * Copyright © 2014-2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#ifndef _INTEL_UC_FW_H_
+#define _INTEL_UC_FW_H_
+
+struct drm_i915_private;
+
+enum intel_uc_fw_status {
+	INTEL_UC_FIRMWARE_FAIL = -1,
+	INTEL_UC_FIRMWARE_NONE = 0,
+	INTEL_UC_FIRMWARE_PENDING,
+	INTEL_UC_FIRMWARE_SUCCESS
+};
+
+enum intel_uc_fw_type {
+	INTEL_UC_FW_TYPE_GUC,
+	INTEL_UC_FW_TYPE_HUC
+};
+
+/*
+ * This structure encapsulates all the data needed during the process
+ * of fetching, caching, and loading the firmware image into the uC.
+ */
+struct intel_uc_fw {
+	const char *path;
+	size_t size;
+	struct drm_i915_gem_object *obj;
+	enum intel_uc_fw_status fetch_status;
+	enum intel_uc_fw_status load_status;
+
+	u16 major_ver_wanted;
+	u16 minor_ver_wanted;
+	u16 major_ver_found;
+	u16 minor_ver_found;
+
+	enum intel_uc_fw_type type;
+	u32 header_size;
+	u32 header_offset;
+	u32 rsa_size;
+	u32 rsa_offset;
+	u32 ucode_size;
+	u32 ucode_offset;
+};
+
+static inline
+const char *intel_uc_fw_status_repr(enum intel_uc_fw_status status)
+{
+	switch (status) {
+	case INTEL_UC_FIRMWARE_FAIL:
+		return "FAIL";
+	case INTEL_UC_FIRMWARE_NONE:
+		return "NONE";
+	case INTEL_UC_FIRMWARE_PENDING:
+		return "PENDING";
+	case INTEL_UC_FIRMWARE_SUCCESS:
+		return "SUCCESS";
+	}
+	return "<invalid>";
+}
+
+static inline const char *intel_uc_fw_type_repr(enum intel_uc_fw_type type)
+{
+	switch (type) {
+	case INTEL_UC_FW_TYPE_GUC:
+		return "GuC";
+	case INTEL_UC_FW_TYPE_HUC:
+		return "HuC";
+	}
+	return "uC";
+}
+
+static inline
+void intel_uc_fw_init(struct intel_uc_fw *uc_fw, enum intel_uc_fw_type type)
+{
+	uc_fw->path = NULL;
+	uc_fw->fetch_status = INTEL_UC_FIRMWARE_NONE;
+	uc_fw->load_status = INTEL_UC_FIRMWARE_NONE;
+	uc_fw->type = type;
+}
+
+void intel_uc_fw_fetch(struct drm_i915_private *dev_priv,
+		       struct intel_uc_fw *uc_fw);
+void intel_uc_fw_fini(struct intel_uc_fw *uc_fw);
+
+#endif
diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c
index b3c3f94fc7e4..983617b5b338 100644
--- a/drivers/gpu/drm/i915/intel_uncore.c
+++ b/drivers/gpu/drm/i915/intel_uncore.c
@@ -626,7 +626,23 @@ void assert_forcewakes_inactive(struct drm_i915_private *dev_priv)
 	if (!dev_priv->uncore.funcs.force_wake_get)
 		return;
 
-	WARN_ON(dev_priv->uncore.fw_domains_active);
+	WARN(dev_priv->uncore.fw_domains_active,
+	     "Expected all fw_domains to be inactive, but %08x are still on\n",
+	     dev_priv->uncore.fw_domains_active);
+}
+
+void assert_forcewakes_active(struct drm_i915_private *dev_priv,
+			      enum forcewake_domains fw_domains)
+{
+	if (!dev_priv->uncore.funcs.force_wake_get)
+		return;
+
+	assert_rpm_wakelock_held(dev_priv);
+
+	fw_domains &= dev_priv->uncore.fw_domains;
+	WARN(fw_domains & ~dev_priv->uncore.fw_domains_active,
+	     "Expected %08x fw_domains to be active, but %08x are off\n",
+	     fw_domains, fw_domains & ~dev_priv->uncore.fw_domains_active);
 }
 
 /* We give fast paths for the really cool registers */
diff --git a/drivers/gpu/drm/i915/intel_uncore.h b/drivers/gpu/drm/i915/intel_uncore.h
index 03786f931905..582771251b57 100644
--- a/drivers/gpu/drm/i915/intel_uncore.h
+++ b/drivers/gpu/drm/i915/intel_uncore.h
@@ -25,6 +25,12 @@
 #ifndef __INTEL_UNCORE_H__
 #define __INTEL_UNCORE_H__
 
+#include <linux/spinlock.h>
+#include <linux/notifier.h>
+#include <linux/hrtimer.h>
+
+#include "i915_reg.h"
+
 struct drm_i915_private;
 
 enum forcewake_domain_id {
@@ -131,6 +137,8 @@ void intel_uncore_resume_early(struct drm_i915_private *dev_priv);
 
 u64 intel_uncore_edram_size(struct drm_i915_private *dev_priv);
 void assert_forcewakes_inactive(struct drm_i915_private *dev_priv);
+void assert_forcewakes_active(struct drm_i915_private *dev_priv,
+			      enum forcewake_domains fw_domains);
 const char *intel_uncore_forcewake_domain_to_str(const enum forcewake_domain_id id);
 
 enum forcewake_domains
diff --git a/drivers/gpu/drm/i915/selftests/huge_gem_object.c b/drivers/gpu/drm/i915/selftests/huge_gem_object.c
index c5c7e8efbdd3..a2632df39173 100644
--- a/drivers/gpu/drm/i915/selftests/huge_gem_object.c
+++ b/drivers/gpu/drm/i915/selftests/huge_gem_object.c
@@ -37,8 +37,7 @@ static void huge_free_pages(struct drm_i915_gem_object *obj,
 	kfree(pages);
 }
 
-static struct sg_table *
-huge_get_pages(struct drm_i915_gem_object *obj)
+static int huge_get_pages(struct drm_i915_gem_object *obj)
 {
 #define GFP (GFP_KERNEL | __GFP_NOWARN | __GFP_NORETRY)
 	const unsigned long nreal = obj->scratch / PAGE_SIZE;
@@ -49,11 +48,11 @@ huge_get_pages(struct drm_i915_gem_object *obj)
 
 	pages = kmalloc(sizeof(*pages), GFP);
 	if (!pages)
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 
 	if (sg_alloc_table(pages, npages, GFP)) {
 		kfree(pages);
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 	}
 
 	sg = pages->sgl;
@@ -81,11 +80,14 @@ huge_get_pages(struct drm_i915_gem_object *obj)
 	if (i915_gem_gtt_prepare_pages(obj, pages))
 		goto err;
 
-	return pages;
+	__i915_gem_object_set_pages(obj, pages, PAGE_SIZE);
+
+	return 0;
 
 err:
 	huge_free_pages(obj, pages);
-	return ERR_PTR(-ENOMEM);
+
+	return -ENOMEM;
 #undef GFP
 }
 
diff --git a/drivers/gpu/drm/i915/selftests/huge_pages.c b/drivers/gpu/drm/i915/selftests/huge_pages.c
new file mode 100644
index 000000000000..c53f8474113a
--- /dev/null
+++ b/drivers/gpu/drm/i915/selftests/huge_pages.c
@@ -0,0 +1,1734 @@
+/*
+ * Copyright © 2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+
+#include "../i915_selftest.h"
+
+#include <linux/prime_numbers.h>
+
+#include "mock_drm.h"
+
+static const unsigned int page_sizes[] = {
+	I915_GTT_PAGE_SIZE_2M,
+	I915_GTT_PAGE_SIZE_64K,
+	I915_GTT_PAGE_SIZE_4K,
+};
+
+static unsigned int get_largest_page_size(struct drm_i915_private *i915,
+					  u64 rem)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(page_sizes); ++i) {
+		unsigned int page_size = page_sizes[i];
+
+		if (HAS_PAGE_SIZES(i915, page_size) && rem >= page_size)
+			return page_size;
+	}
+
+	return 0;
+}
+
+static void huge_pages_free_pages(struct sg_table *st)
+{
+	struct scatterlist *sg;
+
+	for (sg = st->sgl; sg; sg = __sg_next(sg)) {
+		if (sg_page(sg))
+			__free_pages(sg_page(sg), get_order(sg->length));
+	}
+
+	sg_free_table(st);
+	kfree(st);
+}
+
+static int get_huge_pages(struct drm_i915_gem_object *obj)
+{
+#define GFP (GFP_KERNEL | __GFP_NOWARN | __GFP_NORETRY)
+	unsigned int page_mask = obj->mm.page_mask;
+	struct sg_table *st;
+	struct scatterlist *sg;
+	unsigned int sg_page_sizes;
+	u64 rem;
+
+	st = kmalloc(sizeof(*st), GFP);
+	if (!st)
+		return -ENOMEM;
+
+	if (sg_alloc_table(st, obj->base.size >> PAGE_SHIFT, GFP)) {
+		kfree(st);
+		return -ENOMEM;
+	}
+
+	rem = obj->base.size;
+	sg = st->sgl;
+	st->nents = 0;
+	sg_page_sizes = 0;
+
+	/*
+	 * Our goal here is simple, we want to greedily fill the object from
+	 * largest to smallest page-size, while ensuring that we use *every*
+	 * page-size as per the given page-mask.
+	 */
+	do {
+		unsigned int bit = ilog2(page_mask);
+		unsigned int page_size = BIT(bit);
+		int order = get_order(page_size);
+
+		do {
+			struct page *page;
+
+			GEM_BUG_ON(order >= MAX_ORDER);
+			page = alloc_pages(GFP | __GFP_ZERO, order);
+			if (!page)
+				goto err;
+
+			sg_set_page(sg, page, page_size, 0);
+			sg_page_sizes |= page_size;
+			st->nents++;
+
+			rem -= page_size;
+			if (!rem) {
+				sg_mark_end(sg);
+				break;
+			}
+
+			sg = __sg_next(sg);
+		} while ((rem - ((page_size-1) & page_mask)) >= page_size);
+
+		page_mask &= (page_size-1);
+	} while (page_mask);
+
+	if (i915_gem_gtt_prepare_pages(obj, st))
+		goto err;
+
+	obj->mm.madv = I915_MADV_DONTNEED;
+
+	GEM_BUG_ON(sg_page_sizes != obj->mm.page_mask);
+	__i915_gem_object_set_pages(obj, st, sg_page_sizes);
+
+	return 0;
+
+err:
+	sg_set_page(sg, NULL, 0, 0);
+	sg_mark_end(sg);
+	huge_pages_free_pages(st);
+
+	return -ENOMEM;
+}
+
+static void put_huge_pages(struct drm_i915_gem_object *obj,
+			   struct sg_table *pages)
+{
+	i915_gem_gtt_finish_pages(obj, pages);
+	huge_pages_free_pages(pages);
+
+	obj->mm.dirty = false;
+	obj->mm.madv = I915_MADV_WILLNEED;
+}
+
+static const struct drm_i915_gem_object_ops huge_page_ops = {
+	.flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE |
+		 I915_GEM_OBJECT_IS_SHRINKABLE,
+	.get_pages = get_huge_pages,
+	.put_pages = put_huge_pages,
+};
+
+static struct drm_i915_gem_object *
+huge_pages_object(struct drm_i915_private *i915,
+		  u64 size,
+		  unsigned int page_mask)
+{
+	struct drm_i915_gem_object *obj;
+
+	GEM_BUG_ON(!size);
+	GEM_BUG_ON(!IS_ALIGNED(size, BIT(__ffs(page_mask))));
+
+	if (size >> PAGE_SHIFT > INT_MAX)
+		return ERR_PTR(-E2BIG);
+
+	if (overflows_type(size, obj->base.size))
+		return ERR_PTR(-E2BIG);
+
+	obj = i915_gem_object_alloc(i915);
+	if (!obj)
+		return ERR_PTR(-ENOMEM);
+
+	drm_gem_private_object_init(&i915->drm, &obj->base, size);
+	i915_gem_object_init(obj, &huge_page_ops);
+
+	obj->base.write_domain = I915_GEM_DOMAIN_CPU;
+	obj->base.read_domains = I915_GEM_DOMAIN_CPU;
+	obj->cache_level = I915_CACHE_NONE;
+
+	obj->mm.page_mask = page_mask;
+
+	return obj;
+}
+
+static int fake_get_huge_pages(struct drm_i915_gem_object *obj)
+{
+	struct drm_i915_private *i915 = to_i915(obj->base.dev);
+	const u64 max_len = rounddown_pow_of_two(UINT_MAX);
+	struct sg_table *st;
+	struct scatterlist *sg;
+	unsigned int sg_page_sizes;
+	u64 rem;
+
+	st = kmalloc(sizeof(*st), GFP);
+	if (!st)
+		return -ENOMEM;
+
+	if (sg_alloc_table(st, obj->base.size >> PAGE_SHIFT, GFP)) {
+		kfree(st);
+		return -ENOMEM;
+	}
+
+	/* Use optimal page sized chunks to fill in the sg table */
+	rem = obj->base.size;
+	sg = st->sgl;
+	st->nents = 0;
+	sg_page_sizes = 0;
+	do {
+		unsigned int page_size = get_largest_page_size(i915, rem);
+		unsigned int len = min(page_size * div_u64(rem, page_size),
+				       max_len);
+
+		GEM_BUG_ON(!page_size);
+
+		sg->offset = 0;
+		sg->length = len;
+		sg_dma_len(sg) = len;
+		sg_dma_address(sg) = page_size;
+
+		sg_page_sizes |= len;
+
+		st->nents++;
+
+		rem -= len;
+		if (!rem) {
+			sg_mark_end(sg);
+			break;
+		}
+
+		sg = sg_next(sg);
+	} while (1);
+
+	obj->mm.madv = I915_MADV_DONTNEED;
+
+	__i915_gem_object_set_pages(obj, st, sg_page_sizes);
+
+	return 0;
+}
+
+static int fake_get_huge_pages_single(struct drm_i915_gem_object *obj)
+{
+	struct drm_i915_private *i915 = to_i915(obj->base.dev);
+	struct sg_table *st;
+	struct scatterlist *sg;
+	unsigned int page_size;
+
+	st = kmalloc(sizeof(*st), GFP);
+	if (!st)
+		return -ENOMEM;
+
+	if (sg_alloc_table(st, 1, GFP)) {
+		kfree(st);
+		return -ENOMEM;
+	}
+
+	sg = st->sgl;
+	st->nents = 1;
+
+	page_size = get_largest_page_size(i915, obj->base.size);
+	GEM_BUG_ON(!page_size);
+
+	sg->offset = 0;
+	sg->length = obj->base.size;
+	sg_dma_len(sg) = obj->base.size;
+	sg_dma_address(sg) = page_size;
+
+	obj->mm.madv = I915_MADV_DONTNEED;
+
+	__i915_gem_object_set_pages(obj, st, sg->length);
+
+	return 0;
+#undef GFP
+}
+
+static void fake_free_huge_pages(struct drm_i915_gem_object *obj,
+				 struct sg_table *pages)
+{
+	sg_free_table(pages);
+	kfree(pages);
+}
+
+static void fake_put_huge_pages(struct drm_i915_gem_object *obj,
+				struct sg_table *pages)
+{
+	fake_free_huge_pages(obj, pages);
+	obj->mm.dirty = false;
+	obj->mm.madv = I915_MADV_WILLNEED;
+}
+
+static const struct drm_i915_gem_object_ops fake_ops = {
+	.flags = I915_GEM_OBJECT_IS_SHRINKABLE,
+	.get_pages = fake_get_huge_pages,
+	.put_pages = fake_put_huge_pages,
+};
+
+static const struct drm_i915_gem_object_ops fake_ops_single = {
+	.flags = I915_GEM_OBJECT_IS_SHRINKABLE,
+	.get_pages = fake_get_huge_pages_single,
+	.put_pages = fake_put_huge_pages,
+};
+
+static struct drm_i915_gem_object *
+fake_huge_pages_object(struct drm_i915_private *i915, u64 size, bool single)
+{
+	struct drm_i915_gem_object *obj;
+
+	GEM_BUG_ON(!size);
+	GEM_BUG_ON(!IS_ALIGNED(size, I915_GTT_PAGE_SIZE));
+
+	if (size >> PAGE_SHIFT > UINT_MAX)
+		return ERR_PTR(-E2BIG);
+
+	if (overflows_type(size, obj->base.size))
+		return ERR_PTR(-E2BIG);
+
+	obj = i915_gem_object_alloc(i915);
+	if (!obj)
+		return ERR_PTR(-ENOMEM);
+
+	drm_gem_private_object_init(&i915->drm, &obj->base, size);
+
+	if (single)
+		i915_gem_object_init(obj, &fake_ops_single);
+	else
+		i915_gem_object_init(obj, &fake_ops);
+
+	obj->base.write_domain = I915_GEM_DOMAIN_CPU;
+	obj->base.read_domains = I915_GEM_DOMAIN_CPU;
+	obj->cache_level = I915_CACHE_NONE;
+
+	return obj;
+}
+
+static int igt_check_page_sizes(struct i915_vma *vma)
+{
+	struct drm_i915_private *i915 = to_i915(vma->obj->base.dev);
+	unsigned int supported = INTEL_INFO(i915)->page_sizes;
+	struct drm_i915_gem_object *obj = vma->obj;
+	int err = 0;
+
+	if (!HAS_PAGE_SIZES(i915, vma->page_sizes.sg)) {
+		pr_err("unsupported page_sizes.sg=%u, supported=%u\n",
+		       vma->page_sizes.sg & ~supported, supported);
+		err = -EINVAL;
+	}
+
+	if (!HAS_PAGE_SIZES(i915, vma->page_sizes.gtt)) {
+		pr_err("unsupported page_sizes.gtt=%u, supported=%u\n",
+		       vma->page_sizes.gtt & ~supported, supported);
+		err = -EINVAL;
+	}
+
+	if (vma->page_sizes.phys != obj->mm.page_sizes.phys) {
+		pr_err("vma->page_sizes.phys(%u) != obj->mm.page_sizes.phys(%u)\n",
+		       vma->page_sizes.phys, obj->mm.page_sizes.phys);
+		err = -EINVAL;
+	}
+
+	if (vma->page_sizes.sg != obj->mm.page_sizes.sg) {
+		pr_err("vma->page_sizes.sg(%u) != obj->mm.page_sizes.sg(%u)\n",
+		       vma->page_sizes.sg, obj->mm.page_sizes.sg);
+		err = -EINVAL;
+	}
+
+	if (obj->mm.page_sizes.gtt) {
+		pr_err("obj->page_sizes.gtt(%u) should never be set\n",
+		       obj->mm.page_sizes.gtt);
+		err = -EINVAL;
+	}
+
+	return err;
+}
+
+static int igt_mock_exhaust_device_supported_pages(void *arg)
+{
+	struct i915_hw_ppgtt *ppgtt = arg;
+	struct drm_i915_private *i915 = ppgtt->base.i915;
+	unsigned int saved_mask = INTEL_INFO(i915)->page_sizes;
+	struct drm_i915_gem_object *obj;
+	struct i915_vma *vma;
+	int i, j, single;
+	int err;
+
+	/*
+	 * Sanity check creating objects with every valid page support
+	 * combination for our mock device.
+	 */
+
+	for (i = 1; i < BIT(ARRAY_SIZE(page_sizes)); i++) {
+		unsigned int combination = 0;
+
+		for (j = 0; j < ARRAY_SIZE(page_sizes); j++) {
+			if (i & BIT(j))
+				combination |= page_sizes[j];
+		}
+
+		mkwrite_device_info(i915)->page_sizes = combination;
+
+		for (single = 0; single <= 1; ++single) {
+			obj = fake_huge_pages_object(i915, combination, !!single);
+			if (IS_ERR(obj)) {
+				err = PTR_ERR(obj);
+				goto out_device;
+			}
+
+			if (obj->base.size != combination) {
+				pr_err("obj->base.size=%zu, expected=%u\n",
+				       obj->base.size, combination);
+				err = -EINVAL;
+				goto out_put;
+			}
+
+			vma = i915_vma_instance(obj, &ppgtt->base, NULL);
+			if (IS_ERR(vma)) {
+				err = PTR_ERR(vma);
+				goto out_put;
+			}
+
+			err = i915_vma_pin(vma, 0, 0, PIN_USER);
+			if (err)
+				goto out_close;
+
+			err = igt_check_page_sizes(vma);
+
+			if (vma->page_sizes.sg != combination) {
+				pr_err("page_sizes.sg=%u, expected=%u\n",
+				       vma->page_sizes.sg, combination);
+				err = -EINVAL;
+			}
+
+			i915_vma_unpin(vma);
+			i915_vma_close(vma);
+
+			i915_gem_object_put(obj);
+
+			if (err)
+				goto out_device;
+		}
+	}
+
+	goto out_device;
+
+out_close:
+	i915_vma_close(vma);
+out_put:
+	i915_gem_object_put(obj);
+out_device:
+	mkwrite_device_info(i915)->page_sizes = saved_mask;
+
+	return err;
+}
+
+static int igt_mock_ppgtt_misaligned_dma(void *arg)
+{
+	struct i915_hw_ppgtt *ppgtt = arg;
+	struct drm_i915_private *i915 = ppgtt->base.i915;
+	unsigned long supported = INTEL_INFO(i915)->page_sizes;
+	struct drm_i915_gem_object *obj;
+	int bit;
+	int err;
+
+	/*
+	 * Sanity check dma misalignment for huge pages -- the dma addresses we
+	 * insert into the paging structures need to always respect the page
+	 * size alignment.
+	 */
+
+	bit = ilog2(I915_GTT_PAGE_SIZE_64K);
+
+	for_each_set_bit_from(bit, &supported,
+			      ilog2(I915_GTT_MAX_PAGE_SIZE) + 1) {
+		IGT_TIMEOUT(end_time);
+		unsigned int page_size = BIT(bit);
+		unsigned int flags = PIN_USER | PIN_OFFSET_FIXED;
+		unsigned int offset;
+		unsigned int size =
+			round_up(page_size, I915_GTT_PAGE_SIZE_2M) << 1;
+		struct i915_vma *vma;
+
+		obj = fake_huge_pages_object(i915, size, true);
+		if (IS_ERR(obj))
+			return PTR_ERR(obj);
+
+		if (obj->base.size != size) {
+			pr_err("obj->base.size=%zu, expected=%u\n",
+			       obj->base.size, size);
+			err = -EINVAL;
+			goto out_put;
+		}
+
+		err = i915_gem_object_pin_pages(obj);
+		if (err)
+			goto out_put;
+
+		/* Force the page size for this object */
+		obj->mm.page_sizes.sg = page_size;
+
+		vma = i915_vma_instance(obj, &ppgtt->base, NULL);
+		if (IS_ERR(vma)) {
+			err = PTR_ERR(vma);
+			goto out_unpin;
+		}
+
+		err = i915_vma_pin(vma, 0, 0, flags);
+		if (err) {
+			i915_vma_close(vma);
+			goto out_unpin;
+		}
+
+
+		err = igt_check_page_sizes(vma);
+
+		if (vma->page_sizes.gtt != page_size) {
+			pr_err("page_sizes.gtt=%u, expected %u\n",
+			       vma->page_sizes.gtt, page_size);
+			err = -EINVAL;
+		}
+
+		i915_vma_unpin(vma);
+
+		if (err) {
+			i915_vma_close(vma);
+			goto out_unpin;
+		}
+
+		/*
+		 * Try all the other valid offsets until the next
+		 * boundary -- should always fall back to using 4K
+		 * pages.
+		 */
+		for (offset = 4096; offset < page_size; offset += 4096) {
+			err = i915_vma_unbind(vma);
+			if (err) {
+				i915_vma_close(vma);
+				goto out_unpin;
+			}
+
+			err = i915_vma_pin(vma, 0, 0, flags | offset);
+			if (err) {
+				i915_vma_close(vma);
+				goto out_unpin;
+			}
+
+			err = igt_check_page_sizes(vma);
+
+			if (vma->page_sizes.gtt != I915_GTT_PAGE_SIZE_4K) {
+				pr_err("page_sizes.gtt=%u, expected %lu\n",
+				       vma->page_sizes.gtt, I915_GTT_PAGE_SIZE_4K);
+				err = -EINVAL;
+			}
+
+			i915_vma_unpin(vma);
+
+			if (err) {
+				i915_vma_close(vma);
+				goto out_unpin;
+			}
+
+			if (igt_timeout(end_time,
+					"%s timed out at offset %x with page-size %x\n",
+					__func__, offset, page_size))
+				break;
+		}
+
+		i915_vma_close(vma);
+
+		i915_gem_object_unpin_pages(obj);
+		i915_gem_object_put(obj);
+	}
+
+	return 0;
+
+out_unpin:
+	i915_gem_object_unpin_pages(obj);
+out_put:
+	i915_gem_object_put(obj);
+
+	return err;
+}
+
+static void close_object_list(struct list_head *objects,
+			      struct i915_hw_ppgtt *ppgtt)
+{
+	struct drm_i915_gem_object *obj, *on;
+
+	list_for_each_entry_safe(obj, on, objects, st_link) {
+		struct i915_vma *vma;
+
+		vma = i915_vma_instance(obj, &ppgtt->base, NULL);
+		if (!IS_ERR(vma))
+			i915_vma_close(vma);
+
+		list_del(&obj->st_link);
+		i915_gem_object_unpin_pages(obj);
+		i915_gem_object_put(obj);
+	}
+}
+
+static int igt_mock_ppgtt_huge_fill(void *arg)
+{
+	struct i915_hw_ppgtt *ppgtt = arg;
+	struct drm_i915_private *i915 = ppgtt->base.i915;
+	unsigned long max_pages = ppgtt->base.total >> PAGE_SHIFT;
+	unsigned long page_num;
+	bool single = false;
+	LIST_HEAD(objects);
+	IGT_TIMEOUT(end_time);
+	int err;
+
+	for_each_prime_number_from(page_num, 1, max_pages) {
+		struct drm_i915_gem_object *obj;
+		u64 size = page_num << PAGE_SHIFT;
+		struct i915_vma *vma;
+		unsigned int expected_gtt = 0;
+		int i;
+
+		obj = fake_huge_pages_object(i915, size, single);
+		if (IS_ERR(obj)) {
+			err = PTR_ERR(obj);
+			break;
+		}
+
+		if (obj->base.size != size) {
+			pr_err("obj->base.size=%zd, expected=%llu\n",
+			       obj->base.size, size);
+			i915_gem_object_put(obj);
+			err = -EINVAL;
+			break;
+		}
+
+		err = i915_gem_object_pin_pages(obj);
+		if (err) {
+			i915_gem_object_put(obj);
+			break;
+		}
+
+		list_add(&obj->st_link, &objects);
+
+		vma = i915_vma_instance(obj, &ppgtt->base, NULL);
+		if (IS_ERR(vma)) {
+			err = PTR_ERR(vma);
+			break;
+		}
+
+		err = i915_vma_pin(vma, 0, 0, PIN_USER);
+		if (err)
+			break;
+
+		err = igt_check_page_sizes(vma);
+		if (err) {
+			i915_vma_unpin(vma);
+			break;
+		}
+
+		/*
+		 * Figure out the expected gtt page size knowing that we go from
+		 * largest to smallest page size sg chunks, and that we align to
+		 * the largest page size.
+		 */
+		for (i = 0; i < ARRAY_SIZE(page_sizes); ++i) {
+			unsigned int page_size = page_sizes[i];
+
+			if (HAS_PAGE_SIZES(i915, page_size) &&
+			    size >= page_size) {
+				expected_gtt |= page_size;
+				size &= page_size-1;
+			}
+		}
+
+		GEM_BUG_ON(!expected_gtt);
+		GEM_BUG_ON(size);
+
+		if (expected_gtt & I915_GTT_PAGE_SIZE_4K)
+			expected_gtt &= ~I915_GTT_PAGE_SIZE_64K;
+
+		i915_vma_unpin(vma);
+
+		if (vma->page_sizes.sg & I915_GTT_PAGE_SIZE_64K) {
+			if (!IS_ALIGNED(vma->node.start,
+					I915_GTT_PAGE_SIZE_2M)) {
+				pr_err("node.start(%llx) not aligned to 2M\n",
+				       vma->node.start);
+				err = -EINVAL;
+				break;
+			}
+
+			if (!IS_ALIGNED(vma->node.size,
+					I915_GTT_PAGE_SIZE_2M)) {
+				pr_err("node.size(%llx) not aligned to 2M\n",
+				       vma->node.size);
+				err = -EINVAL;
+				break;
+			}
+		}
+
+		if (vma->page_sizes.gtt != expected_gtt) {
+			pr_err("gtt=%u, expected=%u, size=%zd, single=%s\n",
+			       vma->page_sizes.gtt, expected_gtt,
+			       obj->base.size, yesno(!!single));
+			err = -EINVAL;
+			break;
+		}
+
+		if (igt_timeout(end_time,
+				"%s timed out at size %zd\n",
+				__func__, obj->base.size))
+			break;
+
+		single = !single;
+	}
+
+	close_object_list(&objects, ppgtt);
+
+	if (err == -ENOMEM || err == -ENOSPC)
+		err = 0;
+
+	return err;
+}
+
+static int igt_mock_ppgtt_64K(void *arg)
+{
+	struct i915_hw_ppgtt *ppgtt = arg;
+	struct drm_i915_private *i915 = ppgtt->base.i915;
+	struct drm_i915_gem_object *obj;
+	const struct object_info {
+		unsigned int size;
+		unsigned int gtt;
+		unsigned int offset;
+	} objects[] = {
+		/* Cases with forced padding/alignment */
+		{
+			.size = SZ_64K,
+			.gtt = I915_GTT_PAGE_SIZE_64K,
+			.offset = 0,
+		},
+		{
+			.size = SZ_64K + SZ_4K,
+			.gtt = I915_GTT_PAGE_SIZE_4K,
+			.offset = 0,
+		},
+		{
+			.size = SZ_64K - SZ_4K,
+			.gtt = I915_GTT_PAGE_SIZE_4K,
+			.offset = 0,
+		},
+		{
+			.size = SZ_2M,
+			.gtt = I915_GTT_PAGE_SIZE_64K,
+			.offset = 0,
+		},
+		{
+			.size = SZ_2M - SZ_4K,
+			.gtt = I915_GTT_PAGE_SIZE_4K,
+			.offset = 0,
+		},
+		{
+			.size = SZ_2M + SZ_4K,
+			.gtt = I915_GTT_PAGE_SIZE_64K | I915_GTT_PAGE_SIZE_4K,
+			.offset = 0,
+		},
+		{
+			.size = SZ_2M + SZ_64K,
+			.gtt = I915_GTT_PAGE_SIZE_64K,
+			.offset = 0,
+		},
+		{
+			.size = SZ_2M - SZ_64K,
+			.gtt = I915_GTT_PAGE_SIZE_64K,
+			.offset = 0,
+		},
+		/* Try without any forced padding/alignment */
+		{
+			.size = SZ_64K,
+			.offset = SZ_2M,
+			.gtt = I915_GTT_PAGE_SIZE_4K,
+		},
+		{
+			.size = SZ_128K,
+			.offset = SZ_2M - SZ_64K,
+			.gtt = I915_GTT_PAGE_SIZE_4K,
+		},
+	};
+	struct i915_vma *vma;
+	int i, single;
+	int err;
+
+	/*
+	 * Sanity check some of the trickiness with 64K pages -- either we can
+	 * safely mark the whole page-table(2M block) as 64K, or we have to
+	 * always fallback to 4K.
+	 */
+
+	if (!HAS_PAGE_SIZES(i915, I915_GTT_PAGE_SIZE_64K))
+		return 0;
+
+	for (i = 0; i < ARRAY_SIZE(objects); ++i) {
+		unsigned int size = objects[i].size;
+		unsigned int expected_gtt = objects[i].gtt;
+		unsigned int offset = objects[i].offset;
+		unsigned int flags = PIN_USER;
+
+		for (single = 0; single <= 1; single++) {
+			obj = fake_huge_pages_object(i915, size, !!single);
+			if (IS_ERR(obj))
+				return PTR_ERR(obj);
+
+			err = i915_gem_object_pin_pages(obj);
+			if (err)
+				goto out_object_put;
+
+			/*
+			 * Disable 2M pages -- We only want to use 64K/4K pages
+			 * for this test.
+			 */
+			obj->mm.page_sizes.sg &= ~I915_GTT_PAGE_SIZE_2M;
+
+			vma = i915_vma_instance(obj, &ppgtt->base, NULL);
+			if (IS_ERR(vma)) {
+				err = PTR_ERR(vma);
+				goto out_object_unpin;
+			}
+
+			if (offset)
+				flags |= PIN_OFFSET_FIXED | offset;
+
+			err = i915_vma_pin(vma, 0, 0, flags);
+			if (err)
+				goto out_vma_close;
+
+			err = igt_check_page_sizes(vma);
+			if (err)
+				goto out_vma_unpin;
+
+			if (!offset && vma->page_sizes.sg & I915_GTT_PAGE_SIZE_64K) {
+				if (!IS_ALIGNED(vma->node.start,
+						I915_GTT_PAGE_SIZE_2M)) {
+					pr_err("node.start(%llx) not aligned to 2M\n",
+					       vma->node.start);
+					err = -EINVAL;
+					goto out_vma_unpin;
+				}
+
+				if (!IS_ALIGNED(vma->node.size,
+						I915_GTT_PAGE_SIZE_2M)) {
+					pr_err("node.size(%llx) not aligned to 2M\n",
+					       vma->node.size);
+					err = -EINVAL;
+					goto out_vma_unpin;
+				}
+			}
+
+			if (vma->page_sizes.gtt != expected_gtt) {
+				pr_err("gtt=%u, expected=%u, i=%d, single=%s\n",
+				       vma->page_sizes.gtt, expected_gtt, i,
+				       yesno(!!single));
+				err = -EINVAL;
+				goto out_vma_unpin;
+			}
+
+			i915_vma_unpin(vma);
+			i915_vma_close(vma);
+
+			i915_gem_object_unpin_pages(obj);
+			i915_gem_object_put(obj);
+		}
+	}
+
+	return 0;
+
+out_vma_unpin:
+	i915_vma_unpin(vma);
+out_vma_close:
+	i915_vma_close(vma);
+out_object_unpin:
+	i915_gem_object_unpin_pages(obj);
+out_object_put:
+	i915_gem_object_put(obj);
+
+	return err;
+}
+
+static struct i915_vma *
+gpu_write_dw(struct i915_vma *vma, u64 offset, u32 val)
+{
+	struct drm_i915_private *i915 = to_i915(vma->obj->base.dev);
+	const int gen = INTEL_GEN(vma->vm->i915);
+	unsigned int count = vma->size >> PAGE_SHIFT;
+	struct drm_i915_gem_object *obj;
+	struct i915_vma *batch;
+	unsigned int size;
+	u32 *cmd;
+	int n;
+	int err;
+
+	size = (1 + 4 * count) * sizeof(u32);
+	size = round_up(size, PAGE_SIZE);
+	obj = i915_gem_object_create_internal(i915, size);
+	if (IS_ERR(obj))
+		return ERR_CAST(obj);
+
+	cmd = i915_gem_object_pin_map(obj, I915_MAP_WB);
+	if (IS_ERR(cmd)) {
+		err = PTR_ERR(cmd);
+		goto err;
+	}
+
+	offset += vma->node.start;
+
+	for (n = 0; n < count; n++) {
+		if (gen >= 8) {
+			*cmd++ = MI_STORE_DWORD_IMM_GEN4;
+			*cmd++ = lower_32_bits(offset);
+			*cmd++ = upper_32_bits(offset);
+			*cmd++ = val;
+		} else if (gen >= 4) {
+			*cmd++ = MI_STORE_DWORD_IMM_GEN4 |
+				(gen < 6 ? 1 << 22 : 0);
+			*cmd++ = 0;
+			*cmd++ = offset;
+			*cmd++ = val;
+		} else {
+			*cmd++ = MI_STORE_DWORD_IMM | 1 << 22;
+			*cmd++ = offset;
+			*cmd++ = val;
+		}
+
+		offset += PAGE_SIZE;
+	}
+
+	*cmd = MI_BATCH_BUFFER_END;
+
+	i915_gem_object_unpin_map(obj);
+
+	err = i915_gem_object_set_to_gtt_domain(obj, false);
+	if (err)
+		goto err;
+
+	batch = i915_vma_instance(obj, vma->vm, NULL);
+	if (IS_ERR(batch)) {
+		err = PTR_ERR(batch);
+		goto err;
+	}
+
+	err = i915_vma_pin(batch, 0, 0, PIN_USER);
+	if (err)
+		goto err;
+
+	return batch;
+
+err:
+	i915_gem_object_put(obj);
+
+	return ERR_PTR(err);
+}
+
+static int gpu_write(struct i915_vma *vma,
+		     struct i915_gem_context *ctx,
+		     struct intel_engine_cs *engine,
+		     u32 dword,
+		     u32 value)
+{
+	struct drm_i915_gem_request *rq;
+	struct i915_vma *batch;
+	int flags = 0;
+	int err;
+
+	GEM_BUG_ON(!intel_engine_can_store_dword(engine));
+
+	err = i915_gem_object_set_to_gtt_domain(vma->obj, true);
+	if (err)
+		return err;
+
+	rq = i915_gem_request_alloc(engine, ctx);
+	if (IS_ERR(rq))
+		return PTR_ERR(rq);
+
+	batch = gpu_write_dw(vma, dword * sizeof(u32), value);
+	if (IS_ERR(batch)) {
+		err = PTR_ERR(batch);
+		goto err_request;
+	}
+
+	i915_vma_move_to_active(batch, rq, 0);
+	i915_gem_object_set_active_reference(batch->obj);
+	i915_vma_unpin(batch);
+	i915_vma_close(batch);
+
+	err = rq->engine->emit_flush(rq, EMIT_INVALIDATE);
+	if (err)
+		goto err_request;
+
+	err = i915_switch_context(rq);
+	if (err)
+		goto err_request;
+
+	err = rq->engine->emit_bb_start(rq,
+					batch->node.start, batch->node.size,
+					flags);
+	if (err)
+		goto err_request;
+
+	i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE);
+
+	reservation_object_lock(vma->resv, NULL);
+	reservation_object_add_excl_fence(vma->resv, &rq->fence);
+	reservation_object_unlock(vma->resv);
+
+err_request:
+	__i915_add_request(rq, err == 0);
+
+	return err;
+}
+
+static int cpu_check(struct drm_i915_gem_object *obj, u32 dword, u32 val)
+{
+	unsigned int needs_flush;
+	unsigned long n;
+	int err;
+
+	err = i915_gem_obj_prepare_shmem_read(obj, &needs_flush);
+	if (err)
+		return err;
+
+	for (n = 0; n < obj->base.size >> PAGE_SHIFT; ++n) {
+		u32 *ptr = kmap_atomic(i915_gem_object_get_page(obj, n));
+
+		if (needs_flush & CLFLUSH_BEFORE)
+			drm_clflush_virt_range(ptr, PAGE_SIZE);
+
+		if (ptr[dword] != val) {
+			pr_err("n=%lu ptr[%u]=%u, val=%u\n",
+			       n, dword, ptr[dword], val);
+			kunmap_atomic(ptr);
+			err = -EINVAL;
+			break;
+		}
+
+		kunmap_atomic(ptr);
+	}
+
+	i915_gem_obj_finish_shmem_access(obj);
+
+	return err;
+}
+
+static int igt_write_huge(struct i915_gem_context *ctx,
+			  struct drm_i915_gem_object *obj)
+{
+	struct drm_i915_private *i915 = to_i915(obj->base.dev);
+	struct i915_address_space *vm = ctx->ppgtt ? &ctx->ppgtt->base : &i915->ggtt.base;
+	struct intel_engine_cs *engine;
+	struct i915_vma *vma;
+	unsigned int flags = PIN_USER | PIN_OFFSET_FIXED;
+	unsigned int max_page_size;
+	unsigned int id;
+	u64 max;
+	u64 num;
+	u64 size;
+	int err = 0;
+
+	GEM_BUG_ON(!i915_gem_object_has_pinned_pages(obj));
+
+	size = obj->base.size;
+	if (obj->mm.page_sizes.sg & I915_GTT_PAGE_SIZE_64K)
+		size = round_up(size, I915_GTT_PAGE_SIZE_2M);
+
+	max_page_size = rounddown_pow_of_two(obj->mm.page_sizes.sg);
+	max = div_u64((vm->total - size), max_page_size);
+
+	vma = i915_vma_instance(obj, vm, NULL);
+	if (IS_ERR(vma))
+		return PTR_ERR(vma);
+
+	for_each_engine(engine, i915, id) {
+		IGT_TIMEOUT(end_time);
+
+		if (!intel_engine_can_store_dword(engine)) {
+			pr_info("store-dword-imm not supported on engine=%u\n",
+				id);
+			continue;
+		}
+
+		/*
+		 * Try various offsets until we timeout -- we want to avoid
+		 * issues hidden by effectively always using offset = 0.
+		 */
+		for_each_prime_number_from(num, 0, max) {
+			u64 offset = num * max_page_size;
+			u32 dword;
+
+			err = i915_vma_unbind(vma);
+			if (err)
+				goto out_vma_close;
+
+			err = i915_vma_pin(vma, size, max_page_size, flags | offset);
+			if (err) {
+				/*
+				 * The ggtt may have some pages reserved so
+				 * refrain from erroring out.
+				 */
+				if (err == -ENOSPC && i915_is_ggtt(vm)) {
+					err = 0;
+					continue;
+				}
+
+				goto out_vma_close;
+			}
+
+			err = igt_check_page_sizes(vma);
+			if (err)
+				goto out_vma_unpin;
+
+			dword = offset_in_page(num) / 4;
+
+			err = gpu_write(vma, ctx, engine, dword, num + 1);
+			if (err) {
+				pr_err("gpu-write failed at offset=%llx", offset);
+				goto out_vma_unpin;
+			}
+
+			err = cpu_check(obj, dword, num + 1);
+			if (err) {
+				pr_err("cpu-check failed at offset=%llx", offset);
+				goto out_vma_unpin;
+			}
+
+			i915_vma_unpin(vma);
+
+			if (num > 0 &&
+			    igt_timeout(end_time,
+					"%s timed out on engine=%u at offset=%llx, max_page_size=%x\n",
+					__func__, id, offset, max_page_size))
+				break;
+		}
+	}
+
+out_vma_unpin:
+	if (i915_vma_is_pinned(vma))
+		i915_vma_unpin(vma);
+out_vma_close:
+	i915_vma_close(vma);
+
+	return err;
+}
+
+static int igt_ppgtt_exhaust_huge(void *arg)
+{
+	struct i915_gem_context *ctx = arg;
+	struct drm_i915_private *i915 = ctx->i915;
+	unsigned long supported = INTEL_INFO(i915)->page_sizes;
+	static unsigned int pages[ARRAY_SIZE(page_sizes)];
+	struct drm_i915_gem_object *obj;
+	unsigned int size_mask;
+	unsigned int page_mask;
+	int n, i;
+	int err;
+
+	/*
+	 * Sanity check creating objects with a varying mix of page sizes --
+	 * ensuring that our writes lands in the right place.
+	 */
+
+	n = 0;
+	for_each_set_bit(i, &supported, ilog2(I915_GTT_MAX_PAGE_SIZE) + 1)
+		pages[n++] = BIT(i);
+
+	for (size_mask = 2; size_mask < BIT(n); size_mask++) {
+		unsigned int size = 0;
+
+		for (i = 0; i < n; i++) {
+			if (size_mask & BIT(i))
+				size |= pages[i];
+		}
+
+		/*
+		 * For our page mask we want to enumerate all the page-size
+		 * combinations which will fit into our chosen object size.
+		 */
+		for (page_mask = 2; page_mask <= size_mask; page_mask++) {
+			unsigned int page_sizes = 0;
+
+			for (i = 0; i < n; i++) {
+				if (page_mask & BIT(i))
+					page_sizes |= pages[i];
+			}
+
+			/*
+			 * Ensure that we can actually fill the given object
+			 * with our chosen page mask.
+			 */
+			if (!IS_ALIGNED(size, BIT(__ffs(page_sizes))))
+				continue;
+
+			obj = huge_pages_object(i915, size, page_sizes);
+			if (IS_ERR(obj)) {
+				err = PTR_ERR(obj);
+				goto out_device;
+			}
+
+			err = i915_gem_object_pin_pages(obj);
+			if (err) {
+				i915_gem_object_put(obj);
+
+				if (err == -ENOMEM) {
+					pr_info("unable to get pages, size=%u, pages=%u\n",
+						size, page_sizes);
+					err = 0;
+					break;
+				}
+
+				pr_err("pin_pages failed, size=%u, pages=%u\n",
+				       size_mask, page_mask);
+
+				goto out_device;
+			}
+
+			/* Force the page-size for the gtt insertion */
+			obj->mm.page_sizes.sg = page_sizes;
+
+			err = igt_write_huge(ctx, obj);
+			if (err) {
+				pr_err("exhaust write-huge failed with size=%u\n",
+				       size);
+				goto out_unpin;
+			}
+
+			i915_gem_object_unpin_pages(obj);
+			i915_gem_object_put(obj);
+		}
+	}
+
+	goto out_device;
+
+out_unpin:
+	i915_gem_object_unpin_pages(obj);
+	i915_gem_object_put(obj);
+out_device:
+	mkwrite_device_info(i915)->page_sizes = supported;
+
+	return err;
+}
+
+static int igt_ppgtt_internal_huge(void *arg)
+{
+	struct i915_gem_context *ctx = arg;
+	struct drm_i915_private *i915 = ctx->i915;
+	struct drm_i915_gem_object *obj;
+	static const unsigned int sizes[] = {
+		SZ_64K,
+		SZ_128K,
+		SZ_256K,
+		SZ_512K,
+		SZ_1M,
+		SZ_2M,
+	};
+	int i;
+	int err;
+
+	/*
+	 * Sanity check that the HW uses huge pages correctly through internal
+	 * -- ensure that our writes land in the right place.
+	 */
+
+	for (i = 0; i < ARRAY_SIZE(sizes); ++i) {
+		unsigned int size = sizes[i];
+
+		obj = i915_gem_object_create_internal(i915, size);
+		if (IS_ERR(obj))
+			return PTR_ERR(obj);
+
+		err = i915_gem_object_pin_pages(obj);
+		if (err)
+			goto out_put;
+
+		if (obj->mm.page_sizes.phys < I915_GTT_PAGE_SIZE_64K) {
+			pr_info("internal unable to allocate huge-page(s) with size=%u\n",
+				size);
+			goto out_unpin;
+		}
+
+		err = igt_write_huge(ctx, obj);
+		if (err) {
+			pr_err("internal write-huge failed with size=%u\n",
+			       size);
+			goto out_unpin;
+		}
+
+		i915_gem_object_unpin_pages(obj);
+		i915_gem_object_put(obj);
+	}
+
+	return 0;
+
+out_unpin:
+	i915_gem_object_unpin_pages(obj);
+out_put:
+	i915_gem_object_put(obj);
+
+	return err;
+}
+
+static inline bool igt_can_allocate_thp(struct drm_i915_private *i915)
+{
+	return i915->mm.gemfs && has_transparent_hugepage();
+}
+
+static int igt_ppgtt_gemfs_huge(void *arg)
+{
+	struct i915_gem_context *ctx = arg;
+	struct drm_i915_private *i915 = ctx->i915;
+	struct drm_i915_gem_object *obj;
+	static const unsigned int sizes[] = {
+		SZ_2M,
+		SZ_4M,
+		SZ_8M,
+		SZ_16M,
+		SZ_32M,
+	};
+	int i;
+	int err;
+
+	/*
+	 * Sanity check that the HW uses huge pages correctly through gemfs --
+	 * ensure that our writes land in the right place.
+	 */
+
+	if (!igt_can_allocate_thp(i915)) {
+		pr_info("missing THP support, skipping\n");
+		return 0;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(sizes); ++i) {
+		unsigned int size = sizes[i];
+
+		obj = i915_gem_object_create(i915, size);
+		if (IS_ERR(obj))
+			return PTR_ERR(obj);
+
+		err = i915_gem_object_pin_pages(obj);
+		if (err)
+			goto out_put;
+
+		if (obj->mm.page_sizes.phys < I915_GTT_PAGE_SIZE_2M) {
+			pr_info("finishing test early, gemfs unable to allocate huge-page(s) with size=%u\n",
+				size);
+			goto out_unpin;
+		}
+
+		err = igt_write_huge(ctx, obj);
+		if (err) {
+			pr_err("gemfs write-huge failed with size=%u\n",
+			       size);
+			goto out_unpin;
+		}
+
+		i915_gem_object_unpin_pages(obj);
+		i915_gem_object_put(obj);
+	}
+
+	return 0;
+
+out_unpin:
+	i915_gem_object_unpin_pages(obj);
+out_put:
+	i915_gem_object_put(obj);
+
+	return err;
+}
+
+static int igt_ppgtt_pin_update(void *arg)
+{
+	struct i915_gem_context *ctx = arg;
+	struct drm_i915_private *dev_priv = ctx->i915;
+	unsigned long supported = INTEL_INFO(dev_priv)->page_sizes;
+	struct i915_hw_ppgtt *ppgtt = ctx->ppgtt;
+	struct drm_i915_gem_object *obj;
+	struct i915_vma *vma;
+	unsigned int flags = PIN_USER | PIN_OFFSET_FIXED;
+	int first, last;
+	int err;
+
+	/*
+	 * Make sure there's no funny business when doing a PIN_UPDATE -- in the
+	 * past we had a subtle issue with being able to incorrectly do multiple
+	 * alloc va ranges on the same object when doing a PIN_UPDATE, which
+	 * resulted in some pretty nasty bugs, though only when using
+	 * huge-gtt-pages.
+	 */
+
+	if (!USES_FULL_48BIT_PPGTT(dev_priv)) {
+		pr_info("48b PPGTT not supported, skipping\n");
+		return 0;
+	}
+
+	first = ilog2(I915_GTT_PAGE_SIZE_64K);
+	last = ilog2(I915_GTT_PAGE_SIZE_2M);
+
+	for_each_set_bit_from(first, &supported, last + 1) {
+		unsigned int page_size = BIT(first);
+
+		obj = i915_gem_object_create_internal(dev_priv, page_size);
+		if (IS_ERR(obj))
+			return PTR_ERR(obj);
+
+		vma = i915_vma_instance(obj, &ppgtt->base, NULL);
+		if (IS_ERR(vma)) {
+			err = PTR_ERR(vma);
+			goto out_put;
+		}
+
+		err = i915_vma_pin(vma, SZ_2M, 0, flags);
+		if (err)
+			goto out_close;
+
+		if (vma->page_sizes.sg < page_size) {
+			pr_info("Unable to allocate page-size %x, finishing test early\n",
+				page_size);
+			goto out_unpin;
+		}
+
+		err = igt_check_page_sizes(vma);
+		if (err)
+			goto out_unpin;
+
+		if (vma->page_sizes.gtt != page_size) {
+			dma_addr_t addr = i915_gem_object_get_dma_address(obj, 0);
+
+			/*
+			 * The only valid reason for this to ever fail would be
+			 * if the dma-mapper screwed us over when we did the
+			 * dma_map_sg(), since it has the final say over the dma
+			 * address.
+			 */
+			if (IS_ALIGNED(addr, page_size)) {
+				pr_err("page_sizes.gtt=%u, expected=%u\n",
+				       vma->page_sizes.gtt, page_size);
+				err = -EINVAL;
+			} else {
+				pr_info("dma address misaligned, finishing test early\n");
+			}
+
+			goto out_unpin;
+		}
+
+		err = i915_vma_bind(vma, I915_CACHE_NONE, PIN_UPDATE);
+		if (err)
+			goto out_unpin;
+
+		i915_vma_unpin(vma);
+		i915_vma_close(vma);
+
+		i915_gem_object_put(obj);
+	}
+
+	obj = i915_gem_object_create_internal(dev_priv, PAGE_SIZE);
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
+
+	vma = i915_vma_instance(obj, &ppgtt->base, NULL);
+	if (IS_ERR(vma)) {
+		err = PTR_ERR(vma);
+		goto out_put;
+	}
+
+	err = i915_vma_pin(vma, 0, 0, flags);
+	if (err)
+		goto out_close;
+
+	/*
+	 * Make sure we don't end up with something like where the pde is still
+	 * pointing to the 2M page, and the pt we just filled-in is dangling --
+	 * we can check this by writing to the first page where it would then
+	 * land in the now stale 2M page.
+	 */
+
+	err = gpu_write(vma, ctx, dev_priv->engine[RCS], 0, 0xdeadbeaf);
+	if (err)
+		goto out_unpin;
+
+	err = cpu_check(obj, 0, 0xdeadbeaf);
+
+out_unpin:
+	i915_vma_unpin(vma);
+out_close:
+	i915_vma_close(vma);
+out_put:
+	i915_gem_object_put(obj);
+
+	return err;
+}
+
+static int igt_tmpfs_fallback(void *arg)
+{
+	struct i915_gem_context *ctx = arg;
+	struct drm_i915_private *i915 = ctx->i915;
+	struct vfsmount *gemfs = i915->mm.gemfs;
+	struct i915_address_space *vm = ctx->ppgtt ? &ctx->ppgtt->base : &i915->ggtt.base;
+	struct drm_i915_gem_object *obj;
+	struct i915_vma *vma;
+	u32 *vaddr;
+	int err = 0;
+
+	/*
+	 * Make sure that we don't burst into a ball of flames upon falling back
+	 * to tmpfs, which we rely on if on the off-chance we encouter a failure
+	 * when setting up gemfs.
+	 */
+
+	i915->mm.gemfs = NULL;
+
+	obj = i915_gem_object_create(i915, PAGE_SIZE);
+	if (IS_ERR(obj)) {
+		err = PTR_ERR(obj);
+		goto out_restore;
+	}
+
+	vaddr = i915_gem_object_pin_map(obj, I915_MAP_WB);
+	if (IS_ERR(vaddr)) {
+		err = PTR_ERR(vaddr);
+		goto out_put;
+	}
+	*vaddr = 0xdeadbeaf;
+
+	i915_gem_object_unpin_map(obj);
+
+	vma = i915_vma_instance(obj, vm, NULL);
+	if (IS_ERR(vma)) {
+		err = PTR_ERR(vma);
+		goto out_put;
+	}
+
+	err = i915_vma_pin(vma, 0, 0, PIN_USER);
+	if (err)
+		goto out_close;
+
+	err = igt_check_page_sizes(vma);
+
+	i915_vma_unpin(vma);
+out_close:
+	i915_vma_close(vma);
+out_put:
+	i915_gem_object_put(obj);
+out_restore:
+	i915->mm.gemfs = gemfs;
+
+	return err;
+}
+
+static int igt_shrink_thp(void *arg)
+{
+	struct i915_gem_context *ctx = arg;
+	struct drm_i915_private *i915 = ctx->i915;
+	struct i915_address_space *vm = ctx->ppgtt ? &ctx->ppgtt->base : &i915->ggtt.base;
+	struct drm_i915_gem_object *obj;
+	struct i915_vma *vma;
+	unsigned int flags = PIN_USER;
+	int err;
+
+	/*
+	 * Sanity check shrinking huge-paged object -- make sure nothing blows
+	 * up.
+	 */
+
+	if (!igt_can_allocate_thp(i915)) {
+		pr_info("missing THP support, skipping\n");
+		return 0;
+	}
+
+	obj = i915_gem_object_create(i915, SZ_2M);
+	if (IS_ERR(obj))
+		return PTR_ERR(obj);
+
+	vma = i915_vma_instance(obj, vm, NULL);
+	if (IS_ERR(vma)) {
+		err = PTR_ERR(vma);
+		goto out_put;
+	}
+
+	err = i915_vma_pin(vma, 0, 0, flags);
+	if (err)
+		goto out_close;
+
+	if (obj->mm.page_sizes.phys < I915_GTT_PAGE_SIZE_2M) {
+		pr_info("failed to allocate THP, finishing test early\n");
+		goto out_unpin;
+	}
+
+	err = igt_check_page_sizes(vma);
+	if (err)
+		goto out_unpin;
+
+	err = gpu_write(vma, ctx, i915->engine[RCS], 0, 0xdeadbeaf);
+	if (err)
+		goto out_unpin;
+
+	i915_vma_unpin(vma);
+
+	/*
+	 * Now that the pages are *unpinned* shrink-all should invoke
+	 * shmem to truncate our pages.
+	 */
+	i915_gem_shrink_all(i915);
+	if (!IS_ERR_OR_NULL(obj->mm.pages)) {
+		pr_err("shrink-all didn't truncate the pages\n");
+		err = -EINVAL;
+		goto out_close;
+	}
+
+	if (obj->mm.page_sizes.sg || obj->mm.page_sizes.phys) {
+		pr_err("residual page-size bits left\n");
+		err = -EINVAL;
+		goto out_close;
+	}
+
+	err = i915_vma_pin(vma, 0, 0, flags);
+	if (err)
+		goto out_close;
+
+	err = cpu_check(obj, 0, 0xdeadbeaf);
+
+out_unpin:
+	i915_vma_unpin(vma);
+out_close:
+	i915_vma_close(vma);
+out_put:
+	i915_gem_object_put(obj);
+
+	return err;
+}
+
+int i915_gem_huge_page_mock_selftests(void)
+{
+	static const struct i915_subtest tests[] = {
+		SUBTEST(igt_mock_exhaust_device_supported_pages),
+		SUBTEST(igt_mock_ppgtt_misaligned_dma),
+		SUBTEST(igt_mock_ppgtt_huge_fill),
+		SUBTEST(igt_mock_ppgtt_64K),
+	};
+	int saved_ppgtt = i915_modparams.enable_ppgtt;
+	struct drm_i915_private *dev_priv;
+	struct pci_dev *pdev;
+	struct i915_hw_ppgtt *ppgtt;
+	int err;
+
+	dev_priv = mock_gem_device();
+	if (!dev_priv)
+		return -ENOMEM;
+
+	/* Pretend to be a device which supports the 48b PPGTT */
+	i915_modparams.enable_ppgtt = 3;
+
+	pdev = dev_priv->drm.pdev;
+	dma_coerce_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(39));
+
+	mutex_lock(&dev_priv->drm.struct_mutex);
+	ppgtt = i915_ppgtt_create(dev_priv, ERR_PTR(-ENODEV), "mock");
+	if (IS_ERR(ppgtt)) {
+		err = PTR_ERR(ppgtt);
+		goto out_unlock;
+	}
+
+	if (!i915_vm_is_48bit(&ppgtt->base)) {
+		pr_err("failed to create 48b PPGTT\n");
+		err = -EINVAL;
+		goto out_close;
+	}
+
+	/* If we were ever hit this then it's time to mock the 64K scratch */
+	if (!i915_vm_has_scratch_64K(&ppgtt->base)) {
+		pr_err("PPGTT missing 64K scratch page\n");
+		err = -EINVAL;
+		goto out_close;
+	}
+
+	err = i915_subtests(tests, ppgtt);
+
+out_close:
+	i915_ppgtt_close(&ppgtt->base);
+	i915_ppgtt_put(ppgtt);
+
+out_unlock:
+	mutex_unlock(&dev_priv->drm.struct_mutex);
+
+	i915_modparams.enable_ppgtt = saved_ppgtt;
+
+	drm_dev_unref(&dev_priv->drm);
+
+	return err;
+}
+
+int i915_gem_huge_page_live_selftests(struct drm_i915_private *dev_priv)
+{
+	static const struct i915_subtest tests[] = {
+		SUBTEST(igt_shrink_thp),
+		SUBTEST(igt_ppgtt_pin_update),
+		SUBTEST(igt_tmpfs_fallback),
+		SUBTEST(igt_ppgtt_exhaust_huge),
+		SUBTEST(igt_ppgtt_gemfs_huge),
+		SUBTEST(igt_ppgtt_internal_huge),
+	};
+	struct drm_file *file;
+	struct i915_gem_context *ctx;
+	int err;
+
+	if (!USES_PPGTT(dev_priv)) {
+		pr_info("PPGTT not supported, skipping live-selftests\n");
+		return 0;
+	}
+
+	file = mock_file(dev_priv);
+	if (IS_ERR(file))
+		return PTR_ERR(file);
+
+	mutex_lock(&dev_priv->drm.struct_mutex);
+
+	ctx = live_context(dev_priv, file);
+	if (IS_ERR(ctx)) {
+		err = PTR_ERR(ctx);
+		goto out_unlock;
+	}
+
+	err = i915_subtests(tests, ctx);
+
+out_unlock:
+	mutex_unlock(&dev_priv->drm.struct_mutex);
+
+	mock_file_free(dev_priv, file);
+
+	return err;
+}
diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
index 6b132caffa18..9da0c9f99916 100644
--- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
+++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c
@@ -39,25 +39,26 @@ static void fake_free_pages(struct drm_i915_gem_object *obj,
 	kfree(pages);
 }
 
-static struct sg_table *
-fake_get_pages(struct drm_i915_gem_object *obj)
+static int fake_get_pages(struct drm_i915_gem_object *obj)
 {
 #define GFP (GFP_KERNEL | __GFP_NOWARN | __GFP_NORETRY)
 #define PFN_BIAS 0x1000
 	struct sg_table *pages;
 	struct scatterlist *sg;
+	unsigned int sg_page_sizes;
 	typeof(obj->base.size) rem;
 
 	pages = kmalloc(sizeof(*pages), GFP);
 	if (!pages)
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 
 	rem = round_up(obj->base.size, BIT(31)) >> 31;
 	if (sg_alloc_table(pages, rem, GFP)) {
 		kfree(pages);
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 	}
 
+	sg_page_sizes = 0;
 	rem = obj->base.size;
 	for (sg = pages->sgl; sg; sg = sg_next(sg)) {
 		unsigned long len = min_t(typeof(rem), rem, BIT(31));
@@ -66,13 +67,17 @@ fake_get_pages(struct drm_i915_gem_object *obj)
 		sg_set_page(sg, pfn_to_page(PFN_BIAS), len, 0);
 		sg_dma_address(sg) = page_to_phys(sg_page(sg));
 		sg_dma_len(sg) = len;
+		sg_page_sizes |= len;
 
 		rem -= len;
 	}
 	GEM_BUG_ON(rem);
 
 	obj->mm.madv = I915_MADV_DONTNEED;
-	return pages;
+
+	__i915_gem_object_set_pages(obj, pages, sg_page_sizes);
+
+	return 0;
 #undef GFP
 }
 
diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_object.c b/drivers/gpu/drm/i915/selftests/i915_gem_object.c
index 8f011c447e41..1b8774a42e48 100644
--- a/drivers/gpu/drm/i915/selftests/i915_gem_object.c
+++ b/drivers/gpu/drm/i915/selftests/i915_gem_object.c
@@ -251,14 +251,6 @@ static int check_partial_mapping(struct drm_i915_gem_object *obj,
 			return PTR_ERR(io);
 		}
 
-		err = i915_vma_get_fence(vma);
-		if (err) {
-			pr_err("Failed to get fence for partial view: offset=%lu\n",
-			       page);
-			i915_vma_unpin_iomap(vma);
-			return err;
-		}
-
 		iowrite32(page, io + n * PAGE_SIZE/sizeof(*io));
 		i915_vma_unpin_iomap(vma);
 
diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_request.c b/drivers/gpu/drm/i915/selftests/i915_gem_request.c
index 6664cb2eb0b8..a999161e8db1 100644
--- a/drivers/gpu/drm/i915/selftests/i915_gem_request.c
+++ b/drivers/gpu/drm/i915/selftests/i915_gem_request.c
@@ -215,7 +215,9 @@ static int igt_request_rewind(void *arg)
 	}
 	i915_gem_request_get(vip);
 	i915_add_request(vip);
+	rcu_read_lock();
 	request->engine->submit_request(request);
+	rcu_read_unlock();
 
 	mutex_unlock(&i915->drm.struct_mutex);
 
@@ -418,7 +420,10 @@ static struct i915_vma *empty_batch(struct drm_i915_private *i915)
 		err = PTR_ERR(cmd);
 		goto err;
 	}
+
 	*cmd = MI_BATCH_BUFFER_END;
+	i915_gem_chipset_flush(i915);
+
 	i915_gem_object_unpin_map(obj);
 
 	err = i915_gem_object_set_to_gtt_domain(obj, false);
@@ -605,8 +610,8 @@ static struct i915_vma *recursive_batch(struct drm_i915_private *i915)
 		*cmd++ = lower_32_bits(vma->node.start);
 	}
 	*cmd++ = MI_BATCH_BUFFER_END; /* terminate early in case of error */
+	i915_gem_chipset_flush(i915);
 
-	wmb();
 	i915_gem_object_unpin_map(obj);
 
 	return vma;
@@ -625,7 +630,7 @@ static int recursive_batch_resolve(struct i915_vma *batch)
 		return PTR_ERR(cmd);
 
 	*cmd = MI_BATCH_BUFFER_END;
-	wmb();
+	i915_gem_chipset_flush(batch->vm->i915);
 
 	i915_gem_object_unpin_map(batch->obj);
 
@@ -858,7 +863,8 @@ out_request:
 					      I915_MAP_WC);
 		if (!IS_ERR(cmd)) {
 			*cmd = MI_BATCH_BUFFER_END;
-			wmb();
+			i915_gem_chipset_flush(i915);
+
 			i915_gem_object_unpin_map(request[id]->batch->obj);
 		}
 
diff --git a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
index 18b174d855ca..64acd7eccc5c 100644
--- a/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
+++ b/drivers/gpu/drm/i915/selftests/i915_live_selftests.h
@@ -15,5 +15,6 @@ selftest(objects, i915_gem_object_live_selftests)
 selftest(dmabuf, i915_gem_dmabuf_live_selftests)
 selftest(coherency, i915_gem_coherency_live_selftests)
 selftest(gtt, i915_gem_gtt_live_selftests)
+selftest(hugepages, i915_gem_huge_page_live_selftests)
 selftest(contexts, i915_gem_context_live_selftests)
 selftest(hangcheck, intel_hangcheck_live_selftests)
diff --git a/drivers/gpu/drm/i915/selftests/i915_mock_selftests.h b/drivers/gpu/drm/i915/selftests/i915_mock_selftests.h
index fc74687501ba..9961b44f76ed 100644
--- a/drivers/gpu/drm/i915/selftests/i915_mock_selftests.h
+++ b/drivers/gpu/drm/i915/selftests/i915_mock_selftests.h
@@ -21,3 +21,4 @@ selftest(dmabuf, i915_gem_dmabuf_mock_selftests)
 selftest(vma, i915_vma_mock_selftests)
 selftest(evict, i915_gem_evict_mock_selftests)
 selftest(gtt, i915_gem_gtt_mock_selftests)
+selftest(hugepages, i915_gem_huge_page_mock_selftests)
diff --git a/drivers/gpu/drm/i915/selftests/intel_hangcheck.c b/drivers/gpu/drm/i915/selftests/intel_hangcheck.c
index 377c1de766ce..71ce06680d66 100644
--- a/drivers/gpu/drm/i915/selftests/intel_hangcheck.c
+++ b/drivers/gpu/drm/i915/selftests/intel_hangcheck.c
@@ -165,6 +165,7 @@ static int emit_recurse_batch(struct hang *h,
 		*batch++ = lower_32_bits(vma->node.start);
 	}
 	*batch++ = MI_BATCH_BUFFER_END; /* not reached */
+	i915_gem_chipset_flush(h->i915);
 
 	flags = 0;
 	if (INTEL_GEN(vm->i915) <= 5)
@@ -231,7 +232,7 @@ static u32 hws_seqno(const struct hang *h,
 static void hang_fini(struct hang *h)
 {
 	*h->batch = MI_BATCH_BUFFER_END;
-	wmb();
+	i915_gem_chipset_flush(h->i915);
 
 	i915_gem_object_unpin_map(h->obj);
 	i915_gem_object_put(h->obj);
@@ -275,6 +276,8 @@ static int igt_hang_sanitycheck(void *arg)
 		i915_gem_request_get(rq);
 
 		*h.batch = MI_BATCH_BUFFER_END;
+		i915_gem_chipset_flush(i915);
+
 		__i915_add_request(rq, true);
 
 		timeout = i915_wait_request(rq,
@@ -621,8 +624,11 @@ static int igt_wait_reset(void *arg)
 	__i915_add_request(rq, true);
 
 	if (!wait_for_hang(&h, rq)) {
+		struct drm_printer p = drm_info_printer(i915->drm.dev);
+
 		pr_err("Failed to start request %x, at %x\n",
 		       rq->fence.seqno, hws_seqno(&h, rq));
+		intel_engine_dump(rq->engine, &p);
 
 		i915_reset(i915, 0);
 		i915_gem_set_wedged(i915);
@@ -713,8 +719,12 @@ static int igt_reset_queue(void *arg)
 			__i915_add_request(rq, true);
 
 			if (!wait_for_hang(&h, prev)) {
+				struct drm_printer p = drm_info_printer(i915->drm.dev);
+
 				pr_err("Failed to start request %x, at %x\n",
 				       prev->fence.seqno, hws_seqno(&h, prev));
+				intel_engine_dump(rq->engine, &p);
+
 				i915_gem_request_put(rq);
 				i915_gem_request_put(prev);
 
@@ -765,7 +775,7 @@ static int igt_reset_queue(void *arg)
 		pr_info("%s: Completed %d resets\n", engine->name, count);
 
 		*h.batch = MI_BATCH_BUFFER_END;
-		wmb();
+		i915_gem_chipset_flush(i915);
 
 		i915_gem_request_put(prev);
 	}
@@ -815,8 +825,11 @@ static int igt_handle_error(void *arg)
 	__i915_add_request(rq, true);
 
 	if (!wait_for_hang(&h, rq)) {
+		struct drm_printer p = drm_info_printer(i915->drm.dev);
+
 		pr_err("Failed to start request %x, at %x\n",
 		       rq->fence.seqno, hws_seqno(&h, rq));
+		intel_engine_dump(rq->engine, &p);
 
 		i915_reset(i915, 0);
 		i915_gem_set_wedged(i915);
@@ -865,9 +878,16 @@ int intel_hangcheck_live_selftests(struct drm_i915_private *i915)
 		SUBTEST(igt_reset_queue),
 		SUBTEST(igt_handle_error),
 	};
+	int err;
 
 	if (!intel_has_gpu_reset(i915))
 		return 0;
 
-	return i915_subtests(tests, i915);
+	intel_runtime_pm_get(i915);
+
+	err = i915_subtests(tests, i915);
+
+	intel_runtime_pm_put(i915);
+
+	return err;
 }
diff --git a/drivers/gpu/drm/i915/selftests/mock_gem_device.c b/drivers/gpu/drm/i915/selftests/mock_gem_device.c
index 2388424a14da..04eb9362f4f8 100644
--- a/drivers/gpu/drm/i915/selftests/mock_gem_device.c
+++ b/drivers/gpu/drm/i915/selftests/mock_gem_device.c
@@ -83,6 +83,8 @@ static void mock_device_release(struct drm_device *dev)
 	kmem_cache_destroy(i915->vmas);
 	kmem_cache_destroy(i915->objects);
 
+	i915_gemfs_fini(i915);
+
 	drm_dev_fini(&i915->drm);
 	put_device(&i915->drm.pdev->dev);
 }
@@ -146,7 +148,7 @@ struct drm_i915_private *mock_gem_device(void)
 	dev_set_name(&pdev->dev, "mock");
 	dma_coerce_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
 
-#if IS_ENABLED(CONFIG_IOMMU_API)
+#if IS_ENABLED(CONFIG_IOMMU_API) && defined(CONFIG_INTEL_IOMMU)
 	/* hack to disable iommu for the fake device; force identity mapping */
 	pdev->dev.archdata.iommu = (void *)-1;
 #endif
@@ -172,6 +174,11 @@ struct drm_i915_private *mock_gem_device(void)
 
 	mkwrite_device_info(i915)->gen = -1;
 
+	mkwrite_device_info(i915)->page_sizes =
+		I915_GTT_PAGE_SIZE_4K |
+		I915_GTT_PAGE_SIZE_64K |
+		I915_GTT_PAGE_SIZE_2M;
+
 	spin_lock_init(&i915->mm.object_stat_lock);
 	mock_uncore_init(i915);
 
@@ -239,8 +246,16 @@ struct drm_i915_private *mock_gem_device(void)
 	if (!i915->kernel_context)
 		goto err_engine;
 
+	i915->preempt_context = mock_context(i915, NULL);
+	if (!i915->preempt_context)
+		goto err_kernel_context;
+
+	WARN_ON(i915_gemfs_init(i915));
+
 	return i915;
 
+err_kernel_context:
+	i915_gem_context_put(i915->kernel_context);
 err_engine:
 	for_each_engine(engine, i915, id)
 		mock_engine_free(engine);
diff --git a/drivers/gpu/drm/i915/selftests/mock_gtt.c b/drivers/gpu/drm/i915/selftests/mock_gtt.c
index f2118cf535a0..336e1afb250f 100644
--- a/drivers/gpu/drm/i915/selftests/mock_gtt.c
+++ b/drivers/gpu/drm/i915/selftests/mock_gtt.c
@@ -43,7 +43,6 @@ static int mock_bind_ppgtt(struct i915_vma *vma,
 			   u32 flags)
 {
 	GEM_BUG_ON(flags & I915_VMA_GLOBAL_BIND);
-	vma->pages = vma->obj->mm.pages;
 	vma->flags |= I915_VMA_LOCAL_BIND;
 	return 0;
 }
@@ -84,6 +83,8 @@ mock_ppgtt(struct drm_i915_private *i915,
 	ppgtt->base.insert_entries = mock_insert_entries;
 	ppgtt->base.bind_vma = mock_bind_ppgtt;
 	ppgtt->base.unbind_vma = mock_unbind_ppgtt;
+	ppgtt->base.set_pages = ppgtt_set_pages;
+	ppgtt->base.clear_pages = clear_pages;
 	ppgtt->base.cleanup = mock_cleanup;
 
 	return ppgtt;
@@ -93,12 +94,6 @@ static int mock_bind_ggtt(struct i915_vma *vma,
 			  enum i915_cache_level cache_level,
 			  u32 flags)
 {
-	int err;
-
-	err = i915_get_ggtt_vma_pages(vma);
-	if (err)
-		return err;
-
 	vma->flags |= I915_VMA_GLOBAL_BIND | I915_VMA_LOCAL_BIND;
 	return 0;
 }
@@ -124,6 +119,8 @@ void mock_init_ggtt(struct drm_i915_private *i915)
 	ggtt->base.insert_entries = mock_insert_entries;
 	ggtt->base.bind_vma = mock_bind_ggtt;
 	ggtt->base.unbind_vma = mock_unbind_ggtt;
+	ggtt->base.set_pages = ggtt_set_pages;
+	ggtt->base.clear_pages = clear_pages;
 	ggtt->base.cleanup = mock_cleanup;
 
 	i915_address_space_init(&ggtt->base, i915, "global");
diff --git a/drivers/gpu/drm/i915/selftests/scatterlist.c b/drivers/gpu/drm/i915/selftests/scatterlist.c
index 1cc5d2931753..cd6d2a16071f 100644
--- a/drivers/gpu/drm/i915/selftests/scatterlist.c
+++ b/drivers/gpu/drm/i915/selftests/scatterlist.c
@@ -189,6 +189,20 @@ static unsigned int random(unsigned long n,
 	return 1 + (prandom_u32_state(rnd) % 1024);
 }
 
+static unsigned int random_page_size_pages(unsigned long n,
+					   unsigned long count,
+					   struct rnd_state *rnd)
+{
+	/* 4K, 64K, 2M */
+	static unsigned int page_count[] = {
+		BIT(12) >> PAGE_SHIFT,
+		BIT(16) >> PAGE_SHIFT,
+		BIT(21) >> PAGE_SHIFT,
+	};
+
+	return page_count[(prandom_u32_state(rnd) % 3)];
+}
+
 static inline bool page_contiguous(struct page *first,
 				   struct page *last,
 				   unsigned long npages)
@@ -252,6 +266,7 @@ static const npages_fn_t npages_funcs[] = {
 	grow,
 	shrink,
 	random,
+	random_page_size_pages,
 	NULL,
 };