diff --git a/drivers/gpu/drm/i915/gem/i915_gem_mman.c b/drivers/gpu/drm/i915/gem/i915_gem_mman.c
index b55befda33877968dc5997f04479f41409bda78b..73d9eda1d6b7a6fdd237da6bff783e67495648cc 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_mman.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_mman.c
@@ -550,6 +550,20 @@ void i915_gem_object_release_mmap_gtt(struct drm_i915_gem_object *obj)
 	intel_runtime_pm_put(&i915->runtime_pm, wakeref);
 }
 
+void i915_gem_object_runtime_pm_release_mmap_offset(struct drm_i915_gem_object *obj)
+{
+	struct ttm_buffer_object *bo = i915_gem_to_ttm(obj);
+	struct ttm_device *bdev = bo->bdev;
+
+	drm_vma_node_unmap(&bo->base.vma_node, bdev->dev_mapping);
+
+	if (obj->userfault_count) {
+		/* rpm wakeref provide exclusive access */
+		list_del(&obj->userfault_link);
+		obj->userfault_count = 0;
+	}
+}
+
 void i915_gem_object_release_mmap_offset(struct drm_i915_gem_object *obj)
 {
 	struct i915_mmap_offset *mmo, *mn;
@@ -573,6 +587,13 @@ void i915_gem_object_release_mmap_offset(struct drm_i915_gem_object *obj)
 		spin_lock(&obj->mmo.lock);
 	}
 	spin_unlock(&obj->mmo.lock);
+
+	if (obj->userfault_count) {
+		mutex_lock(&to_gt(to_i915(obj->base.dev))->lmem_userfault_lock);
+		list_del(&obj->userfault_link);
+		mutex_unlock(&to_gt(to_i915(obj->base.dev))->lmem_userfault_lock);
+		obj->userfault_count = 0;
+	}
 }
 
 static struct i915_mmap_offset *
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_mman.h b/drivers/gpu/drm/i915/gem/i915_gem_mman.h
index efee9e0d2508678f435e2ac36c75a320cb3fdf63..1fa91b3033b35aa44abc3665516e635b2506658b 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_mman.h
+++ b/drivers/gpu/drm/i915/gem/i915_gem_mman.h
@@ -27,6 +27,7 @@ int i915_gem_dumb_mmap_offset(struct drm_file *file_priv,
 void __i915_gem_object_release_mmap_gtt(struct drm_i915_gem_object *obj);
 void i915_gem_object_release_mmap_gtt(struct drm_i915_gem_object *obj);
 
+void i915_gem_object_runtime_pm_release_mmap_offset(struct drm_i915_gem_object *obj);
 void i915_gem_object_release_mmap_offset(struct drm_i915_gem_object *obj);
 
 #endif
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object.c b/drivers/gpu/drm/i915/gem/i915_gem_object.c
index 85482a04d15840646131263a89892e14b07605bd..7ff9c7877becf977d4b6e39fb8aa5dccb95b8ef7 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_object.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_object.c
@@ -238,7 +238,7 @@ static void __i915_gem_object_free_mmaps(struct drm_i915_gem_object *obj)
 {
 	/* Skip serialisation and waking the device if known to be not used. */
 
-	if (obj->userfault_count)
+	if (obj->userfault_count && !IS_DGFX(to_i915(obj->base.dev)))
 		i915_gem_object_release_mmap_gtt(obj);
 
 	if (!RB_EMPTY_ROOT(&obj->mmo.offsets)) {
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object_types.h b/drivers/gpu/drm/i915/gem/i915_gem_object_types.h
index 9f6b14ec189a2e33d62be44958a494724ae7b189..40305e2bcd49b9f24e24a4fa1f4aaa63164c20a6 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_object_types.h
+++ b/drivers/gpu/drm/i915/gem/i915_gem_object_types.h
@@ -298,7 +298,8 @@ struct drm_i915_gem_object {
 	};
 
 	/**
-	 * Whether the object is currently in the GGTT mmap.
+	 * Whether the object is currently in the GGTT or any other supported
+	 * fake offset mmap backed by lmem.
 	 */
 	unsigned int userfault_count;
 	struct list_head userfault_link;
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
index 6f3ab7ade41ad9410a11eeb7da6f517d1871d4f5..dea6e378946f738eb45b1ddbfcc95a05a245ae59 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c
@@ -548,9 +548,18 @@ static int i915_ttm_shrink(struct drm_i915_gem_object *obj, unsigned int flags)
 static void i915_ttm_delete_mem_notify(struct ttm_buffer_object *bo)
 {
 	struct drm_i915_gem_object *obj = i915_ttm_to_gem(bo);
+	intel_wakeref_t wakeref = 0;
 
 	if (likely(obj)) {
+		/* ttm_bo_release() already has dma_resv_lock */
+		if (i915_ttm_cpu_maps_iomem(bo->resource))
+			wakeref = intel_runtime_pm_get(&to_i915(obj->base.dev)->runtime_pm);
+
 		__i915_gem_object_pages_fini(obj);
+
+		if (wakeref)
+			intel_runtime_pm_put(&to_i915(obj->base.dev)->runtime_pm, wakeref);
+
 		i915_ttm_free_cached_io_rsgt(obj);
 	}
 }
@@ -1020,6 +1029,7 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf)
 	struct ttm_buffer_object *bo = area->vm_private_data;
 	struct drm_device *dev = bo->base.dev;
 	struct drm_i915_gem_object *obj;
+	intel_wakeref_t wakeref = 0;
 	vm_fault_t ret;
 	int idx;
 
@@ -1041,6 +1051,9 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf)
 		return VM_FAULT_SIGBUS;
 	}
 
+	if (i915_ttm_cpu_maps_iomem(bo->resource))
+		wakeref = intel_runtime_pm_get(&to_i915(obj->base.dev)->runtime_pm);
+
 	if (!i915_ttm_resource_mappable(bo->resource)) {
 		int err = -ENODEV;
 		int i;
@@ -1062,7 +1075,8 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf)
 		if (err) {
 			drm_dbg(dev, "Unable to make resource CPU accessible\n");
 			dma_resv_unlock(bo->base.resv);
-			return VM_FAULT_SIGBUS;
+			ret = VM_FAULT_SIGBUS;
+			goto out_rpm;
 		}
 	}
 
@@ -1073,12 +1087,30 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf)
 	} else {
 		ret = ttm_bo_vm_dummy_page(vmf, vmf->vma->vm_page_prot);
 	}
+
 	if (ret == VM_FAULT_RETRY && !(vmf->flags & FAULT_FLAG_RETRY_NOWAIT))
-		return ret;
+		goto out_rpm;
+
+	/* ttm_bo_vm_reserve() already has dma_resv_lock */
+	if (ret == VM_FAULT_NOPAGE && wakeref && !obj->userfault_count) {
+		obj->userfault_count = 1;
+		mutex_lock(&to_gt(to_i915(obj->base.dev))->lmem_userfault_lock);
+		list_add(&obj->userfault_link, &to_gt(to_i915(obj->base.dev))->lmem_userfault_list);
+		mutex_unlock(&to_gt(to_i915(obj->base.dev))->lmem_userfault_lock);
+	}
+
+	if (wakeref & CONFIG_DRM_I915_USERFAULT_AUTOSUSPEND)
+		intel_wakeref_auto(&to_gt(to_i915(obj->base.dev))->userfault_wakeref,
+				   msecs_to_jiffies_timeout(CONFIG_DRM_I915_USERFAULT_AUTOSUSPEND));
 
 	i915_ttm_adjust_lru(obj);
 
 	dma_resv_unlock(bo->base.resv);
+
+out_rpm:
+	if (wakeref)
+		intel_runtime_pm_put(&to_i915(obj->base.dev)->runtime_pm, wakeref);
+
 	return ret;
 }
 
diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c
index f7a918f9de0deb859f7734d93e992aafbaddc634..926a1515cd3b09bce69531327b77e2e84050da9f 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt.c
+++ b/drivers/gpu/drm/i915/gt/intel_gt.c
@@ -39,6 +39,8 @@ void intel_gt_common_init_early(struct intel_gt *gt)
 {
 	spin_lock_init(gt->irq_lock);
 
+	INIT_LIST_HEAD(&gt->lmem_userfault_list);
+	mutex_init(&gt->lmem_userfault_lock);
 	INIT_LIST_HEAD(&gt->closed_vma);
 	spin_lock_init(&gt->closed_lock);
 
diff --git a/drivers/gpu/drm/i915/gt/intel_gt_types.h b/drivers/gpu/drm/i915/gt/intel_gt_types.h
index 0757d9577551ae87ac5eafd14f817647e6fa20e0..f19c2de77ff667ecd99cd242ef78d3f18a97c8ba 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt_types.h
+++ b/drivers/gpu/drm/i915/gt/intel_gt_types.h
@@ -141,6 +141,20 @@ struct intel_gt {
 	struct intel_wakeref wakeref;
 	atomic_t user_wakeref;
 
+	/**
+	 *  Protects access to lmem usefault list.
+	 *  It is required, if we are outside of the runtime suspend path,
+	 *  access to @lmem_userfault_list requires always first grabbing the
+	 *  runtime pm, to ensure we can't race against runtime suspend.
+	 *  Once we have that we also need to grab @lmem_userfault_lock,
+	 *  at which point we have exclusive access.
+	 *  The runtime suspend path is special since it doesn't really hold any locks,
+	 *  but instead has exclusive access by virtue of all other accesses requiring
+	 *  holding the runtime pm wakeref.
+	 */
+	struct mutex lmem_userfault_lock;
+	struct list_head lmem_userfault_list;
+
 	struct list_head closed_vma;
 	spinlock_t closed_lock; /* guards the list of closed_vma */
 
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index cc1117c7e352940de6c3c8a71a7db4bc361f4a5f..c8e14ed9c2a96e35f88ca34ccc681cbd6b5270bf 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -842,6 +842,10 @@ void i915_gem_runtime_suspend(struct drm_i915_private *i915)
 				 &to_gt(i915)->ggtt->userfault_list, userfault_link)
 		__i915_gem_object_release_mmap_gtt(obj);
 
+	list_for_each_entry_safe(obj, on,
+				 &to_gt(i915)->lmem_userfault_list, userfault_link)
+		i915_gem_object_runtime_pm_release_mmap_offset(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