[Intel-gfx] [PATCH 4/4] drm/i915: optional fewer warning patch

[Ben Widawsky]

This patch will likely produce much fewer warnings, but perhaps hide
some bugs in the driver. Any warnings while using this patch are
extremely likely to cause problems, while warnings without this patch
are also driver bugs, but much less likely to be causing issues. Without
this patch we may also get false warnings for intelligent users of the
new refcount mechanism.

Signed-off-by: Ben Widawsky <ben at bwidawsk.net>
---
 drivers/gpu/drm/i915/i915_drv.c |   10 ++++++++--
 1 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 96b3bfc..6da7079 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -286,8 +286,13 @@ void gen6_gt_force_wake_get(struct drm_i915_private *dev_priv)
 	 * immediately. Not having the lock causes a race, but all bets are off
 	 * when using forced forcewake, which should only be touched through
 	 * root-only entry in debugfs.
+	 *
+	 * Intelligent users of the interface may do a force_wake_get() followed
+	 * by many register reads and writes, knowing that the reference count
+	 * is already incremented. So we do not want to warn on those.
 	 */
-	WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex));
+	WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex) &&
+		!dev_priv->forcewake_count);
 
 	if (dev_priv->forcewake_count++ == 0)
 		__gen6_gt_force_wake_get(dev_priv);
@@ -301,7 +306,8 @@ static void __gen6_gt_force_wake_put(struct drm_i915_private *dev_priv)
 
 void gen6_gt_force_wake_put(struct drm_i915_private *dev_priv)
 {
-	WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex));
+	WARN_ON(!mutex_is_locked(&dev_priv->dev->struct_mutex) &&
+		!dev_priv->forcewake_count);
 
 	if (--dev_priv->forcewake_count == 0)
 		__gen6_gt_force_wake_put(dev_priv);
-- 
1.7.3.4