drm/i915: move wedged to the other gpu error handling stuff

And to make Ben Widawsky happier, use the gpu_error instead of
the entire device as the argument in some functions.

Drop the outdated comment on ->wedged for now, a follow-up patch will
change the semantics and add a proper comment again.

Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

drivers/gpu/drm/i915/i915_debugfs.c

@@ -1672,7 +1672,7 @@ i915_wedged_read(struct file *filp,
 
 	len = snprintf(buf, sizeof(buf),
 		       "wedged :  %d\n",
-		       atomic_read(&dev_priv->mm.wedged));
+		       atomic_read(&dev_priv->gpu_error.wedged));
 
 	if (len > sizeof(buf))
 		len = sizeof(buf);

drivers/gpu/drm/i915/i915_drv.h

@@ -744,15 +744,6 @@ struct i915_gem_mm {
 	 */
 	int suspended;
 
-	/**
-	 * Flag if the hardware appears to be wedged.
-	 *
-	 * This is set when attempts to idle the device timeout.
-	 * It prevents command submission from occurring and makes
-	 * every pending request fail
-	 */
-	atomic_t wedged;
-
 	/** Bit 6 swizzling required for X tiling */
 	uint32_t bit_6_swizzle_x;
 	/** Bit 6 swizzling required for Y tiling */
@@ -784,6 +775,8 @@ struct i915_gpu_error {
 
 	unsigned long last_reset;
 
+	atomic_t wedged;
+
 	/* For gpu hang simulation. */
 	unsigned int stop_rings;
 };
@@ -1548,7 +1541,7 @@ i915_gem_object_unpin_fence(struct drm_i915_gem_object *obj)
 
 void i915_gem_retire_requests(struct drm_device *dev);
 void i915_gem_retire_requests_ring(struct intel_ring_buffer *ring);
-int __must_check i915_gem_check_wedge(struct drm_i915_private *dev_priv,
+int __must_check i915_gem_check_wedge(struct i915_gpu_error *error,
 				      bool interruptible);
 
 void i915_gem_reset(struct drm_device *dev);

drivers/gpu/drm/i915/i915_gem.c

@@ -87,14 +87,13 @@ static void i915_gem_info_remove_obj(struct drm_i915_private *dev_priv,
 }
 
 static int
-i915_gem_wait_for_error(struct drm_device *dev)
+i915_gem_wait_for_error(struct i915_gpu_error *error)
 {
-	struct drm_i915_private *dev_priv = dev->dev_private;
+	struct completion *x = &error->completion;
-	struct completion *x = &dev_priv->gpu_error.completion;
 	unsigned long flags;
 	int ret;
 
-	if (!atomic_read(&dev_priv->mm.wedged))
+	if (!atomic_read(&error->wedged))
 		return 0;
 
 	/*
@@ -110,7 +109,7 @@ i915_gem_wait_for_error(struct drm_device *dev)
 		return ret;
 	}
 
-	if (atomic_read(&dev_priv->mm.wedged)) {
+	if (atomic_read(&error->wedged)) {
 		/* GPU is hung, bump the completion count to account for
 		 * the token we just consumed so that we never hit zero and
 		 * end up waiting upon a subsequent completion event that
@@ -125,9 +124,10 @@ i915_gem_wait_for_error(struct drm_device *dev)
 
 int i915_mutex_lock_interruptible(struct drm_device *dev)
 {
+	struct drm_i915_private *dev_priv = dev->dev_private;
 	int ret;
 
-	ret = i915_gem_wait_for_error(dev);
+	ret = i915_gem_wait_for_error(&dev_priv->gpu_error);
 	if (ret)
 		return ret;
 
@@ -939,11 +939,11 @@ unlock:
 }
 
 int
-i915_gem_check_wedge(struct drm_i915_private *dev_priv,
+i915_gem_check_wedge(struct i915_gpu_error *error,
 		     bool interruptible)
 {
-	if (atomic_read(&dev_priv->mm.wedged)) {
+	if (atomic_read(&error->wedged)) {
-		struct completion *x = &dev_priv->gpu_error.completion;
+		struct completion *x = &error->completion;
 		bool recovery_complete;
 		unsigned long flags;
 
@@ -1025,7 +1025,7 @@ static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno,
 
 #define EXIT_COND \
 	(i915_seqno_passed(ring->get_seqno(ring, false), seqno) || \
-	atomic_read(&dev_priv->mm.wedged))
+	atomic_read(&dev_priv->gpu_error.wedged))
 	do {
 		if (interruptible)
 			end = wait_event_interruptible_timeout(ring->irq_queue,
@@ -1035,7 +1035,7 @@ static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno,
 			end = wait_event_timeout(ring->irq_queue, EXIT_COND,
 						 timeout_jiffies);
 
-		ret = i915_gem_check_wedge(dev_priv, interruptible);
+		ret = i915_gem_check_wedge(&dev_priv->gpu_error, interruptible);
 		if (ret)
 			end = ret;
 	} while (end == 0 && wait_forever);
@@ -1081,7 +1081,7 @@ i915_wait_seqno(struct intel_ring_buffer *ring, uint32_t seqno)
 	BUG_ON(!mutex_is_locked(&dev->struct_mutex));
 	BUG_ON(seqno == 0);
 
-	ret = i915_gem_check_wedge(dev_priv, interruptible);
+	ret = i915_gem_check_wedge(&dev_priv->gpu_error, interruptible);
 	if (ret)
 		return ret;
 
@@ -1146,7 +1146,7 @@ i915_gem_object_wait_rendering__nonblocking(struct drm_i915_gem_object *obj,
 	if (seqno == 0)
 		return 0;
 
-	ret = i915_gem_check_wedge(dev_priv, true);
+	ret = i915_gem_check_wedge(&dev_priv->gpu_error, true);
 	if (ret)
 		return ret;
 
@@ -1379,7 +1379,7 @@ out:
 		/* If this -EIO is due to a gpu hang, give the reset code a
 		 * chance to clean up the mess. Otherwise return the proper
 		 * SIGBUS. */
-		if (!atomic_read(&dev_priv->mm.wedged))
+		if (!atomic_read(&dev_priv->gpu_error.wedged))
 			return VM_FAULT_SIGBUS;
 	case -EAGAIN:
 		/* Give the error handler a chance to run and move the
@@ -3390,7 +3390,7 @@ i915_gem_ring_throttle(struct drm_device *dev, struct drm_file *file)
 	u32 seqno = 0;
 	int ret;
 
-	if (atomic_read(&dev_priv->mm.wedged))
+	if (atomic_read(&dev_priv->gpu_error.wedged))
 		return -EIO;
 
 	spin_lock(&file_priv->mm.lock);
@@ -3978,9 +3978,9 @@ i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
 	if (drm_core_check_feature(dev, DRIVER_MODESET))
 		return 0;
 
-	if (atomic_read(&dev_priv->mm.wedged)) {
+	if (atomic_read(&dev_priv->gpu_error.wedged)) {
 		DRM_ERROR("Reenabling wedged hardware, good luck\n");
-		atomic_set(&dev_priv->mm.wedged, 0);
+		atomic_set(&dev_priv->gpu_error.wedged, 0);
 	}
 
 	mutex_lock(&dev->struct_mutex);

drivers/gpu/drm/i915/i915_irq.c

@@ -871,11 +871,11 @@ static void i915_error_work_func(struct work_struct *work)
 
 	kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, error_event);
 
-	if (atomic_read(&dev_priv->mm.wedged)) {
+	if (atomic_read(&dev_priv->gpu_error.wedged)) {
 		DRM_DEBUG_DRIVER("resetting chip\n");
 		kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, reset_event);
 		if (!i915_reset(dev)) {
-			atomic_set(&dev_priv->mm.wedged, 0);
+			atomic_set(&dev_priv->gpu_error.wedged, 0);
 			kobject_uevent_env(&dev->primary->kdev.kobj, KOBJ_CHANGE, reset_done_event);
 		}
 		complete_all(&dev_priv->gpu_error.completion);
@@ -1483,7 +1483,7 @@ void i915_handle_error(struct drm_device *dev, bool wedged)
 
 	if (wedged) {
 		INIT_COMPLETION(dev_priv->gpu_error.completion);
-		atomic_set(&dev_priv->mm.wedged, 1);
+		atomic_set(&dev_priv->gpu_error.wedged, 1);
 
 		/*
 		 * Wakeup waiting processes so they don't hang

drivers/gpu/drm/i915/intel_display.c

@@ -2223,7 +2223,7 @@ intel_finish_fb(struct drm_framebuffer *old_fb)
 	WARN_ON(waitqueue_active(&dev_priv->pending_flip_queue));
 
 	wait_event(dev_priv->pending_flip_queue,
-		   atomic_read(&dev_priv->mm.wedged) ||
+		   atomic_read(&dev_priv->gpu_error.wedged) ||
 		   atomic_read(&obj->pending_flip) == 0);
 
 	/* Big Hammer, we also need to ensure that any pending
@@ -2871,7 +2871,7 @@ static bool intel_crtc_has_pending_flip(struct drm_crtc *crtc)
 	unsigned long flags;
 	bool pending;
 
-	if (atomic_read(&dev_priv->mm.wedged))
+	if (atomic_read(&dev_priv->gpu_error.wedged))
 		return false;
 
 	spin_lock_irqsave(&dev->event_lock, flags);

drivers/gpu/drm/i915/intel_ringbuffer.c

@@ -1371,7 +1371,8 @@ static int ring_wait_for_space(struct intel_ring_buffer *ring, int n)
 
 		msleep(1);
 
-		ret = i915_gem_check_wedge(dev_priv, dev_priv->mm.interruptible);
+		ret = i915_gem_check_wedge(&dev_priv->gpu_error,
+					   dev_priv->mm.interruptible);
 		if (ret)
 			return ret;
 	} while (!time_after(jiffies, end));
@@ -1460,7 +1461,8 @@ int intel_ring_begin(struct intel_ring_buffer *ring,
 	drm_i915_private_t *dev_priv = ring->dev->dev_private;
 	int ret;
 
-	ret = i915_gem_check_wedge(dev_priv, dev_priv->mm.interruptible);
+	ret = i915_gem_check_wedge(&dev_priv->gpu_error,
+				   dev_priv->mm.interruptible);
 	if (ret)
 		return ret;