[Intel-gfx] [PATCH v2] drm/i915/fbdev: Limit the global async-domain synchronization

[Chris Wilson]

On Thu, May 19, 2016 at 10:14:35PM +0200, Lukas Wunner wrote:
> Hi Chris,
> 
> On Thu, May 19, 2016 at 09:28:10AM +0100, Chris Wilson wrote:
> > During cleanup we have to synchronise with the async task we are using
> > to initialise and register our fbdev. Currently, we are using a full
> > synchronisation on the global domain, but we can restrict this to just
> > synchronising up to our task if we remember our cookie.
> > 
> > v2: async_synchronize_cookie() takes an exclusive upper bound, to
> > synchronize with our task we have to pass in the next cookie.
> 
> Oops, good catch, missed that in my own version of this patch:
> https://lists.freedesktop.org/archives/intel-gfx/2016-March/091257.html
> 
> > 
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> > ---
> >  drivers/gpu/drm/i915/intel_drv.h   |  1 +
> >  drivers/gpu/drm/i915/intel_fbdev.c | 31 ++++++++++++++++++-------------
> >  2 files changed, 19 insertions(+), 13 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
> > index 3536292babe0..5bb045ba608e 100644
> > --- a/drivers/gpu/drm/i915/intel_drv.h
> > +++ b/drivers/gpu/drm/i915/intel_drv.h
> > @@ -159,6 +159,7 @@ struct intel_framebuffer {
> >  struct intel_fbdev {
> >  	struct drm_fb_helper helper;
> >  	struct intel_framebuffer *fb;
> > +	async_cookie_t cookie;
> >  	int preferred_bpp;
> >  };
> >  
> > diff --git a/drivers/gpu/drm/i915/intel_fbdev.c b/drivers/gpu/drm/i915/intel_fbdev.c
> > index 99e27530e264..2e9c3f38c023 100644
> > --- a/drivers/gpu/drm/i915/intel_fbdev.c
> > +++ b/drivers/gpu/drm/i915/intel_fbdev.c
> > @@ -538,8 +538,7 @@ static const struct drm_fb_helper_funcs intel_fb_helper_funcs = {
> >  	.fb_probe = intelfb_create,
> >  };
> >  
> > -static void intel_fbdev_destroy(struct drm_device *dev,
> > -				struct intel_fbdev *ifbdev)
> > +static void intel_fbdev_destroy(struct intel_fbdev *ifbdev)
> >  {
> >  	/* We rely on the object-free to release the VMA pinning for
> >  	 * the info->screen_base mmaping. Leaking the VMA is simpler than
> > @@ -554,12 +553,14 @@ static void intel_fbdev_destroy(struct drm_device *dev,
> >  	if (ifbdev->fb) {
> >  		drm_framebuffer_unregister_private(&ifbdev->fb->base);
> >  
> > -		mutex_lock(&dev->struct_mutex);
> > +		mutex_lock(&ifbdev->helper.dev->struct_mutex);
> >  		intel_unpin_fb_obj(&ifbdev->fb->base, BIT(DRM_ROTATE_0));
> > -		mutex_unlock(&dev->struct_mutex);
> > +		mutex_unlock(&ifbdev->helper.dev->struct_mutex);
> >  
> >  		drm_framebuffer_remove(&ifbdev->fb->base);
> >  	}
> > +
> > +	kfree(ifbdev);
> >  }
> >  
> >  /*
> > @@ -736,32 +737,36 @@ int intel_fbdev_init(struct drm_device *dev)
> >  
> >  static void intel_fbdev_initial_config(void *data, async_cookie_t cookie)
> >  {
> > -	struct drm_i915_private *dev_priv = data;
> > -	struct intel_fbdev *ifbdev = dev_priv->fbdev;
> > +	struct intel_fbdev *ifbdev = data;
> > +
> > +	ifbdev->cookie = 0;
> 
> Hm, why are you setting this to 0 here? IIUC the effect is that
> async_synchronize_cookie() will wait until intel_fbdev_initial_config()
> has been *entered*, but isn't the desired effect that it has *finished*?

True, it's also an unserialised write. Premature optimisation to not
block the teardown waiting for the out-of-order completion of earlier
tasks.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre