[Intel-gfx] [PATCH] drm/i915: Accurately track when we mark the hardware as idle/busy

[Chris Wilson]

We currently call intel_mark_idle() too often, as we do so as a
side-effect of processing the request queue. However, we the calls to
intel_mark_idle() are expected to be paired with a call to
intel_mark_busy() (or else we try to idle the hardware by accessing
registers that are already disabled). Make the idle/busy tracking
explicit to prevent the multiple calls.

v2: We can drop some of the complexity in __i915_add_request() as
queue_delayed_work() already behaves as we want (not requeuing the item
if it is already in the queue) and mark_busy/mark_idle imply that the
idle task is inactive.

Reported-by: Paulo Zanoni <paulo.r.zanoni at intel.com>
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Paulo Zanoni <paulo.r.zanoni at intel.com>
---
 drivers/gpu/drm/i915/i915_drv.h      |  8 ++++++++
 drivers/gpu/drm/i915/i915_gem.c      | 13 ++++---------
 drivers/gpu/drm/i915/intel_display.c |  9 +++++++++
 3 files changed, 21 insertions(+), 9 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
index 00222ccc0708..8441c8ac7cc8 100644
--- a/drivers/gpu/drm/i915/i915_drv.h
+++ b/drivers/gpu/drm/i915/i915_drv.h
@@ -1134,6 +1134,14 @@ struct i915_gem_mm {
 	 */
 	bool interruptible;
 
+	/**
+	 * Is the GPU currently considered idle, or busy executing userspace
+	 * requests?  Whilst idle, we attempt to power down the hardware and
+	 * display clocks. In order to reduce the effect on performance, there
+	 * is a slight delay before we do so.
+	 */
+	bool busy;
+
 	/** Bit 6 swizzling required for X tiling */
 	uint32_t bit_6_swizzle_x;
 	/** Bit 6 swizzling required for Y tiling */
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 979c6b5ac6a3..4f25c0398a49 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -2255,7 +2255,6 @@ int __i915_add_request(struct intel_ring_buffer *ring,
 	drm_i915_private_t *dev_priv = ring->dev->dev_private;
 	struct drm_i915_gem_request *request;
 	u32 request_ring_position, request_start;
-	int was_empty;
 	int ret;
 
 	request_start = intel_ring_get_tail(ring);
@@ -2306,7 +2305,6 @@ int __i915_add_request(struct intel_ring_buffer *ring,
 		i915_gem_context_reference(request->ctx);
 
 	request->emitted_jiffies = jiffies;
-	was_empty = list_empty(&ring->request_list);
 	list_add_tail(&request->list, &ring->request_list);
 	request->file_priv = NULL;
 
@@ -2327,13 +2325,10 @@ int __i915_add_request(struct intel_ring_buffer *ring,
 	if (!dev_priv->ums.mm_suspended) {
 		i915_queue_hangcheck(ring->dev);
 
-		if (was_empty) {
-			cancel_delayed_work_sync(&dev_priv->mm.idle_work);
-			queue_delayed_work(dev_priv->wq,
-					   &dev_priv->mm.retire_work,
-					   round_jiffies_up_relative(HZ));
-			intel_mark_busy(dev_priv->dev);
-		}
+		queue_delayed_work(dev_priv->wq,
+				   &dev_priv->mm.retire_work,
+				   round_jiffies_up_relative(HZ));
+		intel_mark_busy(dev_priv->dev);
 	}
 
 	if (out_seqno)
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index e127b23f0c31..b3b2576a32fc 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -8220,8 +8220,12 @@ void intel_mark_busy(struct drm_device *dev)
 {
 	struct drm_i915_private *dev_priv = dev->dev_private;
 
+	if (dev_priv->mm.busy)
+		return;
+
 	hsw_package_c8_gpu_busy(dev_priv);
 	i915_update_gfx_val(dev_priv);
+	dev_priv->mm.busy = true;
 }
 
 void intel_mark_idle(struct drm_device *dev)
@@ -8229,6 +8233,11 @@ void intel_mark_idle(struct drm_device *dev)
 	struct drm_i915_private *dev_priv = dev->dev_private;
 	struct drm_crtc *crtc;
 
+	if (!dev_priv->mm.busy)
+		return;
+
+	dev_priv->mm.busy = false;
+
 	hsw_package_c8_gpu_idle(dev_priv);
 
 	if (!i915.powersave)
-- 
1.9.0