intel-gfx.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
To: John.C.Harrison@Intel.com, Intel-GFX@Lists.FreeDesktop.Org
Subject: Re: [PATCH v5 12/35] drm/i915: Added deferred work handler for scheduler
Date: Tue, 01 Mar 2016 11:16:10 +0200	[thread overview]
Message-ID: <1456823770.6867.9.camel@linux.intel.com> (raw)
In-Reply-To: <1455805644-6450-13-git-send-email-John.C.Harrison@Intel.com>

On to, 2016-02-18 at 14:27 +0000, John.C.Harrison@Intel.com wrote:
> From: John Harrison <John.C.Harrison@Intel.com>
> 
> The scheduler needs to do interrupt triggered work that is too complex
> to do in the interrupt handler. Thus it requires a deferred work
> handler to process such tasks asynchronously.
> 
> v2: Updated to reduce mutex lock usage. The lock is now only held for
> the minimum time within the remove function rather than for the whole
> of the worker thread's operation.
> 
> v5: Removed objectionable white space and added some documentation.
> [Joonas Lahtinen]
> 
> For: VIZ-1587
> Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
> ---
>  drivers/gpu/drm/i915/i915_dma.c       |  3 +++
>  drivers/gpu/drm/i915/i915_drv.h       | 10 ++++++++++
>  drivers/gpu/drm/i915/i915_gem.c       |  2 ++
>  drivers/gpu/drm/i915/i915_scheduler.c | 29 +++++++++++++++++++++++++++--
>  drivers/gpu/drm/i915/i915_scheduler.h |  1 +
>  5 files changed, 43 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_dma.c b/drivers/gpu/drm/i915/i915_dma.c
> index 678adc7..c3d382d 100644
> --- a/drivers/gpu/drm/i915/i915_dma.c
> +++ b/drivers/gpu/drm/i915/i915_dma.c
> @@ -1158,6 +1158,9 @@ int i915_driver_unload(struct drm_device *dev)
>  	WARN_ON(unregister_oom_notifier(&dev_priv->mm.oom_notifier));
>  	unregister_shrinker(&dev_priv->mm.shrinker);
>  
> +	/* Cancel the scheduler work handler, which should be idle now. */
> +	cancel_work_sync(&dev_priv->mm.scheduler_work);
> +
>  	io_mapping_free(dev_priv->gtt.mappable);
>  	arch_phys_wc_del(dev_priv->gtt.mtrr);
>  
> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> index 03add1a..4d544f1 100644
> --- a/drivers/gpu/drm/i915/i915_drv.h
> +++ b/drivers/gpu/drm/i915/i915_drv.h
> @@ -1291,6 +1291,16 @@ struct i915_gem_mm {
>  	struct delayed_work retire_work;
>  
>  	/**
> +	 * New scheme is to get an interrupt after every work packet
> +	 * in order to allow the low latency scheduling of pending
> +	 * packets. The idea behind adding new packets to a pending
> +	 * queue rather than directly into the hardware ring buffer
> +	 * is to allow high priority packets to over take low priority
> +	 * ones.
> +	 */
> +	struct work_struct scheduler_work;
> +
> +	/**
>  	 * When we detect an idle GPU, we want to turn on
>  	 * powersaving features. So once we see that there
>  	 * are no more requests outstanding and no more
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index c3b7def..1ab7256 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -5427,6 +5427,8 @@ i915_gem_load(struct drm_device *dev)
>  			  i915_gem_retire_work_handler);
>  	INIT_DELAYED_WORK(&dev_priv->mm.idle_work,
>  			  i915_gem_idle_work_handler);
> +	INIT_WORK(&dev_priv->mm.scheduler_work,
> +				i915_scheduler_work_handler);
>  	init_waitqueue_head(&dev_priv->gpu_error.reset_queue);
>  
>  	dev_priv->relative_constants_mode = I915_EXEC_CONSTANTS_REL_GENERAL;
> diff --git a/drivers/gpu/drm/i915/i915_scheduler.c b/drivers/gpu/drm/i915/i915_scheduler.c
> index ab5007a..3986890 100644
> --- a/drivers/gpu/drm/i915/i915_scheduler.c
> +++ b/drivers/gpu/drm/i915/i915_scheduler.c
> @@ -697,7 +697,9 @@ static int i915_scheduler_remove_dependent(struct i915_scheduler *scheduler,
>   */
>  void i915_scheduler_wakeup(struct drm_device *dev)
>  {
> -	/* XXX: Need to call i915_scheduler_remove() via work handler. */
> +	struct drm_i915_private *dev_priv = to_i915(dev);
> +
> +	queue_work(dev_priv->wq, &dev_priv->mm.scheduler_work);
>  }
>  
>  /**
> @@ -827,7 +829,7 @@ static bool i915_scheduler_remove(struct i915_scheduler *scheduler,
>  	return do_submit;
>  }
>  
> -void i915_scheduler_process_work(struct intel_engine_cs *ring)
> +static void i915_scheduler_process_work(struct intel_engine_cs *ring)

This is an odd change, maybe should have been static void to begin
with?

>  {
>  	struct drm_i915_private *dev_priv = ring->dev->dev_private;
>  	struct i915_scheduler *scheduler = dev_priv->scheduler;
> @@ -874,6 +876,29 @@ void i915_scheduler_process_work(struct intel_engine_cs *ring)
>  }
>  
>  /**
> + * i915_scheduler_work_handler - scheduler's work handler callback.
> + * @work: Work structure
> + * A lot of the scheduler's work must be done asynchronously in response to
> + * an interrupt or other event. However, that work cannot be done at
> + * interrupt time or in the context of the event signaller (which might in
> + * fact be an interrupt). Thus a worker thread is required. This function
> + * will cause the thread to wake up and do its processing.
> + */
> +void i915_scheduler_work_handler(struct work_struct *work)
> +{
> +	struct intel_engine_cs *ring;
> +	struct drm_i915_private *dev_priv;
> +	struct drm_device *dev;
> +	int i;
> +
> +	dev_priv = container_of(work, struct drm_i915_private, mm.scheduler_work);
> +	dev = dev_priv->dev;

dev is unused variable?

With those changes,

Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>

> +
> +	for_each_ring(ring, dev_priv, i)
> +		i915_scheduler_process_work(ring);
> +}
> +
> +/**
>   * i915_scheduler_closefile - notify the scheduler that a DRM file handle
>   * has been closed.
>   * @dev: DRM device
> diff --git a/drivers/gpu/drm/i915/i915_scheduler.h b/drivers/gpu/drm/i915/i915_scheduler.h
> index 0e8b6a9..180d75f 100644
> --- a/drivers/gpu/drm/i915/i915_scheduler.h
> +++ b/drivers/gpu/drm/i915/i915_scheduler.h
> @@ -93,5 +93,6 @@ void i915_scheduler_clean_node(struct i915_scheduler_queue_entry *node);
>  int i915_scheduler_queue_execbuffer(struct i915_scheduler_queue_entry *qe);
>  bool i915_scheduler_notify_request(struct drm_i915_gem_request *req);
>  void i915_scheduler_wakeup(struct drm_device *dev);
> +void i915_scheduler_work_handler(struct work_struct *work);
>  
>  #endif  /* _I915_SCHEDULER_H_ */
-- 
Joonas Lahtinen
Open Source Technology Center
Intel Corporation
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

  reply	other threads:[~2016-03-01  9:16 UTC|newest]

Thread overview: 82+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-02-18 14:26 [PATCH v5 00/35] GPU scheduler for i915 driver John.C.Harrison
2016-02-18 14:26 ` [PATCH v5 01/35] drm/i915: Add total count to context status debugfs output John.C.Harrison
2016-02-18 14:26 ` [PATCH v5 02/35] drm/i915: Prelude to splitting i915_gem_do_execbuffer in two John.C.Harrison
2016-02-18 14:26 ` [PATCH v5 03/35] drm/i915: Split i915_dem_do_execbuffer() in half John.C.Harrison
2016-02-18 14:26 ` [PATCH v5 04/35] drm/i915: Cache request pointer in *_submission_final() John.C.Harrison
2016-02-18 14:26 ` [PATCH v5 05/35] drm/i915: Re-instate request->uniq because it is extremely useful John.C.Harrison
2016-02-18 14:26 ` [PATCH v5 06/35] drm/i915: Start of GPU scheduler John.C.Harrison
2016-02-19 13:03   ` Joonas Lahtinen
2016-02-19 17:03     ` John Harrison
2016-02-26  9:13       ` Joonas Lahtinen
2016-02-26 14:18         ` John Harrison
2016-02-18 14:26 ` [PATCH v5 07/35] drm/i915: Prepare retire_requests to handle out-of-order seqnos John.C.Harrison
2016-02-19 19:23   ` Jesse Barnes
2016-02-18 14:26 ` [PATCH v5 08/35] drm/i915: Disable hardware semaphores when GPU scheduler is enabled John.C.Harrison
2016-02-19 19:27   ` Jesse Barnes
2016-02-18 14:26 ` [PATCH v5 09/35] drm/i915: Force MMIO flips when scheduler enabled John.C.Harrison
2016-02-19 19:28   ` Jesse Barnes
2016-02-19 19:53     ` Ville Syrjälä
2016-02-19 20:01       ` Jesse Barnes
2016-02-22  9:41         ` Lankhorst, Maarten
2016-02-22 12:53           ` John Harrison
2016-02-20  9:22     ` Chris Wilson
2016-02-22 20:42       ` Jesse Barnes
2016-02-23 11:16         ` Chris Wilson
2016-02-18 14:26 ` [PATCH v5 10/35] drm/i915: Added scheduler hook when closing DRM file handles John.C.Harrison
2016-03-01  8:59   ` Joonas Lahtinen
2016-03-01 14:52     ` John Harrison
2016-02-18 14:26 ` [PATCH v5 11/35] drm/i915: Added scheduler hook into i915_gem_request_notify() John.C.Harrison
2016-03-01  9:10   ` Joonas Lahtinen
2016-02-18 14:27 ` [PATCH v5 12/35] drm/i915: Added deferred work handler for scheduler John.C.Harrison
2016-03-01  9:16   ` Joonas Lahtinen [this message]
2016-03-01 15:12     ` John Harrison
2016-02-18 14:27 ` [PATCH v5 13/35] drm/i915: Redirect execbuffer_final() via scheduler John.C.Harrison
2016-02-19 19:33   ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 14/35] drm/i915: Keep the reserved space mechanism happy John.C.Harrison
2016-02-19 19:36   ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 15/35] drm/i915: Added tracking/locking of batch buffer objects John.C.Harrison
2016-02-19 19:42   ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 16/35] drm/i915: Hook scheduler node clean up into retire requests John.C.Harrison
2016-02-19 19:44   ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 17/35] drm/i915: Added scheduler support to __wait_request() calls John.C.Harrison
2016-03-01 10:02   ` Joonas Lahtinen
2016-03-11 11:47     ` John Harrison
2016-02-18 14:27 ` [PATCH v5 18/35] drm/i915: Added scheduler support to page fault handler John.C.Harrison
2016-02-19 19:45   ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 19/35] drm/i915: Added scheduler flush calls to ring throttle and idle functions John.C.Harrison
2016-03-07 11:31   ` Joonas Lahtinen
2016-03-11 16:22     ` John Harrison
2016-02-18 14:27 ` [PATCH v5 20/35] drm/i915: Add scheduler hook to GPU reset John.C.Harrison
2016-02-23 20:27   ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 21/35] drm/i915: Added a module parameter to allow the scheduler to be disabled John.C.Harrison
2016-02-23 20:29   ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 22/35] drm/i915: Support for 'unflushed' ring idle John.C.Harrison
2016-02-23 20:35   ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 23/35] drm/i915: Defer seqno allocation until actual hardware submission time John.C.Harrison
2016-03-07 12:16   ` Joonas Lahtinen
2016-02-18 14:27 ` [PATCH v5 24/35] drm/i915: Added trace points to scheduler John.C.Harrison
2016-02-23 20:42   ` Jesse Barnes
2016-02-23 20:42   ` Jesse Barnes
2016-02-26 15:55     ` John Harrison
2016-02-26 17:12       ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 25/35] drm/i915: Added scheduler queue throttling by DRM file handle John.C.Harrison
2016-02-23 21:02   ` Jesse Barnes
2016-03-01 15:52     ` John Harrison
2016-02-18 14:27 ` [PATCH v5 26/35] drm/i915: Added debugfs interface to scheduler tuning parameters John.C.Harrison
2016-02-23 21:06   ` Jesse Barnes
2016-03-11 16:28     ` John Harrison
2016-03-11 17:25       ` Jesse Barnes
2016-02-18 14:27 ` [PATCH v5 27/35] drm/i915: Added debug state dump facilities to scheduler John.C.Harrison
2016-03-07 12:31   ` Joonas Lahtinen
2016-03-11 16:38     ` John Harrison
2016-03-15 10:53       ` Joonas Lahtinen
2016-02-18 14:27 ` [PATCH v5 28/35] drm/i915: Add early exit to execbuff_final() if insufficient ring space John.C.Harrison
2016-02-18 14:27 ` [PATCH v5 29/35] drm/i915: Added scheduler statistic reporting to debugfs John.C.Harrison
2016-02-18 14:27 ` [PATCH v5 30/35] drm/i915: Add scheduler support functions for TDR John.C.Harrison
2016-02-18 14:27 ` [PATCH v5 31/35] drm/i915: Scheduler state dump via debugfs John.C.Harrison
2016-02-18 14:27 ` [PATCH v5 32/35] drm/i915: Enable GPU scheduler by default John.C.Harrison
2016-02-18 14:27 ` [PATCH v5 33/35] drm/i915: Add scheduling priority to per-context parameters John.C.Harrison
2016-02-18 14:27 ` [PATCH v5 34/35] drm/i915: Add support for retro-actively banning batch buffers John.C.Harrison
2016-02-18 14:27 ` [PATCH v5 35/35] drm/i915: Allow scheduler to manage inter-ring object synchronisation John.C.Harrison
2016-02-18 14:27 ` [PATCH 01/20] igt/gem_ctx_param_basic: Updated to support scheduler priority interface John.C.Harrison
2016-02-18 15:30 ` ✗ Fi.CI.BAT: failure for GPU scheduler for i915 driver Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1456823770.6867.9.camel@linux.intel.com \
    --to=joonas.lahtinen@linux.intel.com \
    --cc=Intel-GFX@Lists.FreeDesktop.Org \
    --cc=John.C.Harrison@Intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).