[Intel-gfx] [PATCH v2] drm/i915: Flush logical context image out to memory upon suspend

[Ville Syrjälä]

On Fri, Jul 15, 2016 at 01:03:40PM +0100, Chris Wilson wrote:
> Before suspend, and especially before building the hibernation image, we
> need to context image to be coherent in memory. To do this we require
> that we perform a context switch to a disposable context (i.e. the
> dev_priv->kernel_context) - when that switch is complete, all other
> context images will be complete. This leaves the kernel_context image as
> incomplete, but fortunately that is disposable and we can do a quick
> fixup of the logical state after resuming.
> 
> v2: Share the nearly identical code to switch to the kernel context with
> eviction.
> 
> Testcase: igt/gem_exec_suspend # bsw
> References: https://bugs.freedesktop.org/show_bug.cgi?id=96526
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Ville Syrjälä <ville.syrjala at linux.intel.com>
> ---
>  drivers/gpu/drm/i915/i915_drv.c |  4 +---
>  drivers/gpu/drm/i915/i915_drv.h |  1 +
>  drivers/gpu/drm/i915/i915_gem.c | 17 +++++++++++++++++
>  3 files changed, 19 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
> index 15440123e38d..c5b7b8e0678a 100644
> --- a/drivers/gpu/drm/i915/i915_drv.c
> +++ b/drivers/gpu/drm/i915/i915_drv.c
> @@ -1590,9 +1590,7 @@ static int i915_drm_resume(struct drm_device *dev)
>  
>  	intel_csr_ucode_resume(dev_priv);
>  
> -	mutex_lock(&dev->struct_mutex);
> -	i915_gem_restore_gtt_mappings(dev);
> -	mutex_unlock(&dev->struct_mutex);
> +	i915_gem_resume(dev);
>  
>  	i915_restore_state(dev);
>  	intel_opregion_setup(dev_priv);
> diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> index cd1ccc47ee8b..27d9b2c374b3 100644
> --- a/drivers/gpu/drm/i915/i915_drv.h
> +++ b/drivers/gpu/drm/i915/i915_drv.h
> @@ -3384,6 +3384,7 @@ void i915_gem_init_swizzling(struct drm_device *dev);
>  void i915_gem_cleanup_engines(struct drm_device *dev);
>  int __must_check i915_gem_wait_for_idle(struct drm_i915_private *dev_priv);
>  int __must_check i915_gem_suspend(struct drm_device *dev);
> +void i915_gem_resume(struct drm_device *dev);
>  void __i915_add_request(struct drm_i915_gem_request *req,
>  			struct drm_i915_gem_object *batch_obj,
>  			bool flush_caches);
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index cf0e8aa8035c..5a44ed9f4a33 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -4983,6 +4983,10 @@ i915_gem_suspend(struct drm_device *dev)
>  	intel_suspend_gt_powersave(dev_priv);
>  
>  	mutex_lock(&dev->struct_mutex);
> +	ret = i915_gem_switch_to_kernel_context(dev_priv);
> +	if (ret)
> +		goto err;
> +
>  	ret = i915_gem_wait_for_idle(dev_priv);
>  	if (ret)
>  		goto err;
> @@ -5009,6 +5013,19 @@ err:
>  	return ret;
>  }
>  
> +void i915_gem_resume(struct drm_device *dev)
> +{
> +	struct drm_i915_private *dev_priv = to_i915(dev);
> +
> +	mutex_lock(&dev->struct_mutex);
> +	i915_gem_restore_gtt_mappings(dev);
> +
> +	if (i915.enable_execlists)
> +		intel_lr_context_reset(dev_priv, dev_priv->kernel_context);

Maybe toss in a few comments as to why we do the switch and reset?

Should we do a sanity check here to make sure all contexts have
in fact HEAD==TAIL at this point? Might catch accidental breakage.

Anyways, patches look good to me, and my BSW seems happy as well, so
Reviewed-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
Tested-by: Ville Syrjälä <ville.syrjala at linux.intel.com>


> +
> +	mutex_unlock(&dev->struct_mutex);
> +}
> +
>  void i915_gem_init_swizzling(struct drm_device *dev)
>  {
>  	struct drm_i915_private *dev_priv = to_i915(dev);
> -- 
> 2.8.1

-- 
Ville Syrjälä
Intel OTC