diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index fece922..88cc793c4 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -2300,15 +2300,6 @@
 	struct drm_device *dev = node->minor->dev;
 	struct drm_i915_private *dev_priv = dev->dev_private;
 	struct drm_file *file;
-	int ret;
-
-	ret = mutex_lock_interruptible(&dev->struct_mutex);
-	if (ret)
-		return ret;
-
-	ret = mutex_lock_interruptible(&dev_priv->rps.hw_lock);
-	if (ret)
-		goto unlock;
 
 	seq_printf(m, "RPS enabled? %d\n", dev_priv->rps.enabled);
 	seq_printf(m, "GPU busy? %d\n", dev_priv->mm.busy);
@@ -2319,6 +2310,7 @@
 		   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));
+	spin_lock(&dev_priv->rps.client_lock);
 	list_for_each_entry_reverse(file, &dev->filelist, lhead) {
 		struct drm_i915_file_private *file_priv = file->driver_priv;
 		struct task_struct *task;
@@ -2339,12 +2331,9 @@
 		   dev_priv->rps.mmioflips.boosts,
 		   list_empty(&dev_priv->rps.mmioflips.link) ? "" : ", active");
 	seq_printf(m, "Kernel boosts: %d\n", dev_priv->rps.boosts);
+	spin_unlock(&dev_priv->rps.client_lock);
 
-	mutex_unlock(&dev_priv->rps.hw_lock);
-unlock:
-	mutex_unlock(&dev->struct_mutex);
-
-	return ret;
+	return 0;
 }
 
 static int i915_llc(struct seq_file *m, void *data)
diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 9adfd12..3f6cad9 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1089,9 +1089,12 @@
 	int last_adj;
 	enum { LOW_POWER, BETWEEN, HIGH_POWER } power;
 
+	spinlock_t client_lock;
+	struct list_head clients;
+	bool client_boost;
+
 	bool enabled;
 	struct delayed_work delayed_resume_work;
-	struct list_head clients;
 	unsigned boosts;
 
 	struct intel_rps_client semaphores, mmioflips;
@@ -1101,7 +1104,9 @@
 
 	/*
 	 * Protects RPS/RC6 register access and PCU communication.
-	 * Must be taken after struct_mutex if nested.
+	 * 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;
 };
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index cc206f1..be35f04 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -5223,9 +5223,9 @@
 	spin_unlock(&file_priv->mm.lock);
 
 	if (!list_empty(&file_priv->rps.link)) {
-		mutex_lock(&to_i915(dev)->rps.hw_lock);
+		spin_lock(&to_i915(dev)->rps.client_lock);
 		list_del(&file_priv->rps.link);
-		mutex_unlock(&to_i915(dev)->rps.hw_lock);
+		spin_unlock(&to_i915(dev)->rps.client_lock);
 	}
 }
 
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 707e2ca..e6bb72d 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -1086,8 +1086,9 @@
 {
 	struct drm_i915_private *dev_priv =
 		container_of(work, struct drm_i915_private, rps.work);
+	bool client_boost;
+	int new_delay, adj, min, max;
 	u32 pm_iir;
-	int new_delay, adj;
 
 	spin_lock_irq(&dev_priv->irq_lock);
 	/* Speed up work cancelation during disabling rps interrupts. */
@@ -1099,12 +1100,14 @@
 	dev_priv->rps.pm_iir = 0;
 	/* Make sure not to corrupt PMIMR state used by ringbuffer on GEN6 */
 	gen6_enable_pm_irq(dev_priv, dev_priv->pm_rps_events);
+	client_boost = dev_priv->rps.client_boost;
+	dev_priv->rps.client_boost = false;
 	spin_unlock_irq(&dev_priv->irq_lock);
 
 	/* Make sure we didn't queue anything we're not going to process. */
 	WARN_ON(pm_iir & ~dev_priv->pm_rps_events);
 
-	if ((pm_iir & dev_priv->pm_rps_events) == 0)
+	if ((pm_iir & dev_priv->pm_rps_events) == 0 && !client_boost)
 		return;
 
 	mutex_lock(&dev_priv->rps.hw_lock);
@@ -1113,7 +1116,13 @@
 
 	adj = dev_priv->rps.last_adj;
 	new_delay = dev_priv->rps.cur_freq;
-	if (pm_iir & GEN6_PM_RP_UP_THRESHOLD) {
+	min = dev_priv->rps.min_freq_softlimit;
+	max = dev_priv->rps.max_freq_softlimit;
+
+	if (client_boost) {
+		new_delay = dev_priv->rps.max_freq_softlimit;
+		adj = 0;
+	} else if (pm_iir & GEN6_PM_RP_UP_THRESHOLD) {
 		if (adj > 0)
 			adj *= 2;
 		else /* CHV needs even encode values */
@@ -1149,9 +1158,7 @@
 	 * interrupt
 	 */
 	new_delay += adj;
-	new_delay = clamp_t(int, new_delay,
-			    dev_priv->rps.min_freq_softlimit,
-			    dev_priv->rps.max_freq_softlimit);
+	new_delay = clamp_t(int, new_delay, min, max);
 
 	intel_set_rps(dev_priv->dev, new_delay);
 
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index 55d9931..7984321 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -4143,17 +4143,25 @@
 		dev_priv->rps.last_adj = 0;
 		I915_WRITE(GEN6_PMINTRMSK, 0xffffffff);
 	}
+	mutex_unlock(&dev_priv->rps.hw_lock);
 
+	spin_lock(&dev_priv->rps.client_lock);
 	while (!list_empty(&dev_priv->rps.clients))
 		list_del_init(dev_priv->rps.clients.next);
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	spin_unlock(&dev_priv->rps.client_lock);
 }
 
 void gen6_rps_boost(struct drm_i915_private *dev_priv,
 		    struct intel_rps_client *rps,
 		    unsigned long submitted)
 {
-	u32 val;
+	/* This is intentionally racy! We peek at the state here, then
+	 * validate inside the RPS worker.
+	 */
+	if (!(dev_priv->mm.busy &&
+	      dev_priv->rps.enabled &&
+	      dev_priv->rps.cur_freq < dev_priv->rps.max_freq_softlimit))
+		return;
 
 	/* Force a RPS boost (and don't count it against the client) if
 	 * the GPU is severely congested.
@@ -4161,14 +4169,14 @@
 	if (rps && time_after(jiffies, submitted + DRM_I915_THROTTLE_JIFFIES))
 		rps = NULL;
 
-	mutex_lock(&dev_priv->rps.hw_lock);
-	val = dev_priv->rps.max_freq_softlimit;
-	if (dev_priv->rps.enabled &&
-	    dev_priv->mm.busy &&
-	    dev_priv->rps.cur_freq < val &&
-	    (rps == NULL || list_empty(&rps->link))) {
-		intel_set_rps(dev_priv->dev, val);
-		dev_priv->rps.last_adj = 0;
+	spin_lock(&dev_priv->rps.client_lock);
+	if (rps == NULL || list_empty(&rps->link)) {
+		spin_lock_irq(&dev_priv->irq_lock);
+		if (dev_priv->rps.interrupts_enabled) {
+			dev_priv->rps.client_boost = true;
+			queue_work(dev_priv->wq, &dev_priv->rps.work);
+		}
+		spin_unlock_irq(&dev_priv->irq_lock);
 
 		if (rps != NULL) {
 			list_add(&rps->link, &dev_priv->rps.clients);
@@ -4176,7 +4184,7 @@
 		} else
 			dev_priv->rps.boosts++;
 	}
-	mutex_unlock(&dev_priv->rps.hw_lock);
+	spin_unlock(&dev_priv->rps.client_lock);
 }
 
 void intel_set_rps(struct drm_device *dev, u8 val)
@@ -6913,6 +6921,7 @@
 	struct drm_i915_private *dev_priv = dev->dev_private;
 
 	mutex_init(&dev_priv->rps.hw_lock);
+	spin_lock_init(&dev_priv->rps.client_lock);
 
 	INIT_DELAYED_WORK(&dev_priv->rps.delayed_resume_work,
 			  intel_gen6_powersave_work);