drm/i915: Mask USER interrupts on gen6 (until required)

Otherwise we may consume 20% of the CPU just handling IRQs whilst
rendering. Ouch.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

5 files changed, 113 insertions, 62 deletions

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 972e08e4e054..1f77d8c6c6a2 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1007,12 +1007,6 @@ extern u32 i915_get_vblank_counter(struct drm_device *dev, int crtc);
 extern u32 gm45_get_vblank_counter(struct drm_device *dev, int crtc);
 extern int i915_vblank_swap(struct drm_device *dev, void *data,
 			    struct drm_file *file_priv);
-extern void i915_enable_irq(drm_i915_private_t *dev_priv, u32 mask);
-extern void i915_disable_irq(drm_i915_private_t *dev_priv, u32 mask);
-extern void ironlake_enable_graphics_irq(drm_i915_private_t *dev_priv,
-		u32 mask);
-extern void ironlake_disable_graphics_irq(drm_i915_private_t *dev_priv,
-		u32 mask);
 
 void
 i915_enable_pipestat(drm_i915_private_t *dev_priv, int pipe, u32 mask);
diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
index 0dadc025b77b..826873a23db0 100644
--- a/drivers/gpu/drm/i915/i915_irq.c
+++ b/drivers/gpu/drm/i915/i915_irq.c
@@ -64,26 +64,6 @@
 #define DRM_I915_VBLANK_PIPE_ALL	(DRM_I915_VBLANK_PIPE_A | \
 					 DRM_I915_VBLANK_PIPE_B)
 
-void
-ironlake_enable_graphics_irq(drm_i915_private_t *dev_priv, u32 mask)
-{
-	if ((dev_priv->gt_irq_mask & mask) != 0) {
-		dev_priv->gt_irq_mask &= ~mask;
-		I915_WRITE(GTIMR, dev_priv->gt_irq_mask);
-		POSTING_READ(GTIMR);
-	}
-}
-
-void
-ironlake_disable_graphics_irq(drm_i915_private_t *dev_priv, u32 mask)
-{
-	if ((dev_priv->gt_irq_mask & mask) != mask) {
-		dev_priv->gt_irq_mask |= mask;
-		I915_WRITE(GTIMR, dev_priv->gt_irq_mask);
-		POSTING_READ(GTIMR);
-	}
-}
-
 /* For display hotplug interrupt */
 static void
 ironlake_enable_display_irq(drm_i915_private_t *dev_priv, u32 mask)
@@ -105,26 +85,6 @@ ironlake_disable_display_irq(drm_i915_private_t *dev_priv, u32 mask)
 	}
 }
 
-void
-i915_enable_irq(drm_i915_private_t *dev_priv, u32 mask)
-{
-	if ((dev_priv->irq_mask & mask) != 0) {
-		dev_priv->irq_mask &= ~mask;
-		I915_WRITE(IMR, dev_priv->irq_mask);
-		POSTING_READ(IMR);
-	}
-}
-
-void
-i915_disable_irq(drm_i915_private_t *dev_priv, u32 mask)
-{
-	if ((dev_priv->irq_mask & mask) != mask) {
-		dev_priv->irq_mask |= mask;
-		I915_WRITE(IMR, dev_priv->irq_mask);
-		POSTING_READ(IMR);
-	}
-}
-
 static inline u32
 i915_pipestat(int pipe)
 {
@@ -1673,11 +1633,6 @@ static int ironlake_irq_postinstall(struct drm_device *dev)
 
 	I915_WRITE(GTIIR, I915_READ(GTIIR));
 	I915_WRITE(GTIMR, dev_priv->gt_irq_mask);
-	if (IS_GEN6(dev)) {
-		I915_WRITE(GEN6_RENDER_IMR, ~GEN6_RENDER_USER_INTERRUPT);
-		I915_WRITE(GEN6_BSD_IMR, ~GEN6_BSD_USER_INTERRUPT);
-		I915_WRITE(GEN6_BLITTER_IMR, ~GEN6_BLITTER_USER_INTERRUPT);
-	}
 
 	if (IS_GEN6(dev))
 		render_irqs =
diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h
index ecfb0023f60d..b0ab4247ce48 100644
--- a/drivers/gpu/drm/i915/i915_reg.h
+++ b/drivers/gpu/drm/i915/i915_reg.h
@@ -288,6 +288,7 @@
 #define RING_HWS_PGA_GEN6(base)	((base)+0x2080)
 #define RING_ACTHD(base)	((base)+0x74)
 #define RING_NOPID(base)	((base)+0x94)
+#define RING_IMR(base)		((base)+0xa8)
 #define   TAIL_ADDR		0x001FFFF8
 #define   HEAD_WRAP_COUNT	0xFFE00000
 #define   HEAD_WRAP_ONE		0x00200000
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index aa8f6abf16f2..3bff7fb72341 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -485,6 +485,38 @@ pc_render_get_seqno(struct intel_ring_buffer *ring)
 	return pc->cpu_page[0];
 }
 
+static void
+ironlake_enable_irq(drm_i915_private_t *dev_priv, u32 mask)
+{
+	dev_priv->gt_irq_mask &= ~mask;
+	I915_WRITE(GTIMR, dev_priv->gt_irq_mask);
+	POSTING_READ(GTIMR);
+}
+
+static void
+ironlake_disable_irq(drm_i915_private_t *dev_priv, u32 mask)
+{
+	dev_priv->gt_irq_mask |= mask;
+	I915_WRITE(GTIMR, dev_priv->gt_irq_mask);
+	POSTING_READ(GTIMR);
+}
+
+static void
+i915_enable_irq(drm_i915_private_t *dev_priv, u32 mask)
+{
+	dev_priv->irq_mask &= ~mask;
+	I915_WRITE(IMR, dev_priv->irq_mask);
+	POSTING_READ(IMR);
+}
+
+static void
+i915_disable_irq(drm_i915_private_t *dev_priv, u32 mask)
+{
+	dev_priv->irq_mask |= mask;
+	I915_WRITE(IMR, dev_priv->irq_mask);
+	POSTING_READ(IMR);
+}
+
 static bool
 render_ring_get_irq(struct intel_ring_buffer *ring)
 {
@@ -499,8 +531,8 @@ render_ring_get_irq(struct intel_ring_buffer *ring)
 
 		spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 		if (HAS_PCH_SPLIT(dev))
-			ironlake_enable_graphics_irq(dev_priv,
-						     GT_PIPE_NOTIFY | GT_USER_INTERRUPT);
+			ironlake_enable_irq(dev_priv,
+					    GT_PIPE_NOTIFY | GT_USER_INTERRUPT);
 		else
 			i915_enable_irq(dev_priv, I915_USER_INTERRUPT);
 		spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
@@ -520,9 +552,9 @@ render_ring_put_irq(struct intel_ring_buffer *ring)
 
 		spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
 		if (HAS_PCH_SPLIT(dev))
-			ironlake_disable_graphics_irq(dev_priv,
-						      GT_USER_INTERRUPT |
-						      GT_PIPE_NOTIFY);
+			ironlake_disable_irq(dev_priv,
+					     GT_USER_INTERRUPT |
+					     GT_PIPE_NOTIFY);
 		else
 			i915_disable_irq(dev_priv, I915_USER_INTERRUPT);
 		spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
@@ -596,7 +628,7 @@ ring_get_irq(struct intel_ring_buffer *ring, u32 flag)
 		unsigned long irqflags;
 
 		spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
-		ironlake_enable_graphics_irq(dev_priv, flag);
+		ironlake_enable_irq(dev_priv, flag);
 		spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 	}
 
@@ -613,7 +645,46 @@ ring_put_irq(struct intel_ring_buffer *ring, u32 flag)
 		unsigned long irqflags;
 
 		spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
-		ironlake_disable_graphics_irq(dev_priv, flag);
+		ironlake_disable_irq(dev_priv, flag);
+		spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	}
+}
+
+static bool
+gen6_ring_get_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag)
+{
+	struct drm_device *dev = ring->dev;
+
+	if (!dev->irq_enabled)
+	       return false;
+
+	if (atomic_inc_return(&ring->irq_refcount) == 1) {
+		drm_i915_private_t *dev_priv = dev->dev_private;
+		unsigned long irqflags;
+
+		spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+		ring->irq_mask &= ~rflag;
+		I915_WRITE_IMR(ring, ring->irq_mask);
+		ironlake_enable_irq(dev_priv, gflag);
+		spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
+	}
+
+	return true;
+}
+
+static void
+gen6_ring_put_irq(struct intel_ring_buffer *ring, u32 gflag, u32 rflag)
+{
+	struct drm_device *dev = ring->dev;
+
+	if (atomic_dec_and_test(&ring->irq_refcount)) {
+		drm_i915_private_t *dev_priv = dev->dev_private;
+		unsigned long irqflags;
+
+		spin_lock_irqsave(&dev_priv->irq_lock, irqflags);
+		ring->irq_mask |= rflag;
+		I915_WRITE_IMR(ring, ring->irq_mask);
+		ironlake_disable_irq(dev_priv, gflag);
 		spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags);
 	}
 }
@@ -757,6 +828,7 @@ int intel_init_ring_buffer(struct drm_device *dev,
 	INIT_LIST_HEAD(&ring->active_list);
 	INIT_LIST_HEAD(&ring->request_list);
 	INIT_LIST_HEAD(&ring->gpu_write_list);
+	ring->irq_mask = ~0;
 
 	if (I915_NEED_GFX_HWS(dev)) {
 		ret = init_status_page(ring);
@@ -1030,15 +1102,35 @@ gen6_ring_dispatch_execbuffer(struct intel_ring_buffer *ring,
 }
 
 static bool
+gen6_render_ring_get_irq(struct intel_ring_buffer *ring)
+{
+	return gen6_ring_get_irq(ring,
+				 GT_USER_INTERRUPT,
+				 GEN6_RENDER_USER_INTERRUPT);
+}
+
+static void
+gen6_render_ring_put_irq(struct intel_ring_buffer *ring)
+{
+	return gen6_ring_put_irq(ring,
+				 GT_USER_INTERRUPT,
+				 GEN6_RENDER_USER_INTERRUPT);
+}
+
+static bool
 gen6_bsd_ring_get_irq(struct intel_ring_buffer *ring)
 {
-	return ring_get_irq(ring, GT_GEN6_BSD_USER_INTERRUPT);
+	return gen6_ring_get_irq(ring,
+				 GT_GEN6_BSD_USER_INTERRUPT,
+				 GEN6_BSD_USER_INTERRUPT);
 }
 
 static void
 gen6_bsd_ring_put_irq(struct intel_ring_buffer *ring)
 {
-	ring_put_irq(ring, GT_GEN6_BSD_USER_INTERRUPT);
+	return gen6_ring_put_irq(ring,
+				 GT_GEN6_BSD_USER_INTERRUPT,
+				 GEN6_BSD_USER_INTERRUPT);
 }
 
 /* ring buffer for Video Codec for Gen6+ */
@@ -1062,13 +1154,17 @@ static const struct intel_ring_buffer gen6_bsd_ring = {
 static bool
 blt_ring_get_irq(struct intel_ring_buffer *ring)
 {
-	return ring_get_irq(ring, GT_BLT_USER_INTERRUPT);
+	return gen6_ring_get_irq(ring,
+				 GT_BLT_USER_INTERRUPT,
+				 GEN6_BLITTER_USER_INTERRUPT);
 }
 
 static void
 blt_ring_put_irq(struct intel_ring_buffer *ring)
 {
-	ring_put_irq(ring, GT_BLT_USER_INTERRUPT);
+	gen6_ring_put_irq(ring,
+			  GT_BLT_USER_INTERRUPT,
+			  GEN6_BLITTER_USER_INTERRUPT);
 }
 
 
@@ -1192,6 +1288,8 @@ int intel_init_render_ring_buffer(struct drm_device *dev)
 	*ring = render_ring;
 	if (INTEL_INFO(dev)->gen >= 6) {
 		ring->add_request = gen6_add_request;
+		ring->irq_get = gen6_render_ring_get_irq;
+		ring->irq_put = gen6_render_ring_put_irq;
 	} else if (IS_GEN5(dev)) {
 		ring->add_request = pc_render_add_request;
 		ring->get_seqno = pc_render_get_seqno;
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index 5969c2ed1028..634f6f84cb57 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -28,6 +28,8 @@ struct  intel_hw_status_page {
 #define I915_READ_CTL(ring) I915_RING_READ(RING_CTL(ring->mmio_base))
 #define I915_WRITE_CTL(ring, val) I915_WRITE(RING_CTL(ring->mmio_base), val)
 
+#define I915_WRITE_IMR(ring, val) I915_WRITE(RING_IMR(ring->mmio_base), val)
+
 #define I915_READ_NOPID(ring) I915_RING_READ(RING_NOPID(ring->mmio_base))
 #define I915_READ_SYNC_0(ring) I915_RING_READ(RING_SYNC_0(ring->mmio_base))
 #define I915_READ_SYNC_1(ring) I915_RING_READ(RING_SYNC_1(ring->mmio_base))
@@ -52,6 +54,7 @@ struct  intel_ring_buffer {
 	int		effective_size;
 	struct intel_hw_status_page status_page;
 
+	u32		irq_mask;
 	u32		irq_seqno;		/* last seq seem at irq time */
 	u32		waiting_seqno;
 	u32		sync_seqno[I915_NUM_RINGS-1];