[Intel-gfx] [PATCH] drm/i915: Accurately track when we mark the hardware as idle/busy
Paulo Zanoni
przanoni at gmail.com
Fri Feb 21 20:33:07 CET 2014
2014-02-21 14:55 GMT-03:00 Chris Wilson <chris at chris-wilson.co.uk>:
> 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.
>
> v3: We do still need to cancel the pending idle task so that it is sent
> again after the current busy load completes (not in the middle of it).
>
> 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>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni at intel.com>
Tested-by: Paulo Zanoni <paulo.r.zanoni at intel.com> (ran the pm_pc8
test suite on HSW)
> ---
> drivers/gpu/drm/i915/i915_drv.h | 8 ++++++++
> drivers/gpu/drm/i915/i915_gem.c | 14 +++++---------
> drivers/gpu/drm/i915/intel_display.c | 9 +++++++++
> 3 files changed, 22 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..293018392ac1 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 +2325,11 @@ 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);
> - }
> + 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);
> }
>
> 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
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
--
Paulo Zanoni
More information about the Intel-gfx
mailing list