drm/i915: Free RPS boosts for all laggards

If the client stalls on a congested request, chosen to be 20ms old to
match throttling, allow the client a free RPS boost.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: s/rq/req/]
[danvet: s/0/NULL/ reported by 0-day build]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

3 files changed, 19 insertions, 6 deletions

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 289178b7e684..50e49a3d7e51 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -1245,7 +1245,7 @@ int __i915_wait_request(struct drm_i915_gem_request *req,
                 jiffies + nsecs_to_jiffies_timeout((u64)*timeout) : 0;
 
         if (INTEL_INFO(dev_priv)->gen >= 6)
-                gen6_rps_boost(dev_priv, rps);
+                gen6_rps_boost(dev_priv, rps, req->emitted_jiffies);
 
         /* Record current time in case interrupted by signal, or wedged */
         trace_i915_gem_request_wait_begin(req);
diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
index 6968e7483863..e94cb6e16456 100644
--- a/drivers/gpu/drm/i915/intel_drv.h
+++ b/drivers/gpu/drm/i915/intel_drv.h
@@ -1365,7 +1365,8 @@ void gen6_rps_busy(struct drm_i915_private *dev_priv);
 void gen6_rps_reset_ei(struct drm_i915_private *dev_priv);
 void gen6_rps_idle(struct drm_i915_private *dev_priv);
 void gen6_rps_boost(struct drm_i915_private *dev_priv,
-                    struct intel_rps_client *rps);
+                    struct intel_rps_client *rps,
+                    unsigned long submitted);
 void intel_queue_rps_boost_for_request(struct drm_device *dev,
                                        struct drm_i915_gem_request *req);
 void ilk_wm_get_hw_state(struct drm_device *dev);
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index b676fe81c563..6414b27692d4 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -4150,10 +4150,17 @@ void gen6_rps_idle(struct drm_i915_private *dev_priv)
 }
 
 void gen6_rps_boost(struct drm_i915_private *dev_priv,
-                    struct intel_rps_client *rps)
+                    struct intel_rps_client *rps,
+                    unsigned long submitted)
 {
         u32 val;
 
+        /* Force a RPS boost (and don't count it against the client) if
+         * the GPU is severely congested.
+         */
+        if (rps && time_after(jiffies, submitted + msecs_to_jiffies(20)))
+                rps = NULL;
+
         mutex_lock(&dev_priv->rps.hw_lock);
         val = dev_priv->rps.max_freq_softlimit;
         if (dev_priv->rps.enabled &&
@@ -6848,11 +6855,13 @@ struct request_boost {
 static void __intel_rps_boost_work(struct work_struct *work)
 {
         struct request_boost *boost = container_of(work, struct request_boost, work);
+        struct drm_i915_gem_request *req = boost->req;
 
-        if (!i915_gem_request_completed(boost->req, true))
-                gen6_rps_boost(to_i915(boost->req->ring->dev), NULL);
+        if (!i915_gem_request_completed(req, true))
+                gen6_rps_boost(to_i915(req->ring->dev), NULL,
+                               req->emitted_jiffies);
 
-        i915_gem_request_unreference__unlocked(boost->req);
+        i915_gem_request_unreference__unlocked(req);
         kfree(boost);
 }
 
@@ -6864,6 +6873,9 @@ void intel_queue_rps_boost_for_request(struct drm_device *dev,
         if (req == NULL || INTEL_INFO(dev)->gen < 6)
                 return;
 
+        if (i915_gem_request_completed(req, true))
+                return;
+
         boost = kmalloc(sizeof(*boost), GFP_ATOMIC);
         if (boost == NULL)
                 return;