diff mbox

[1/9] drm: Always reject drm_vblank_get() after drm_vblank_off()

Message ID 1401104792-26560-2-git-send-email-ville.syrjala@linux.intel.com (mailing list archive)
State New, archived
Headers show

Commit Message

Ville Syrjälä May 26, 2014, 11:46 a.m. UTC 
From: Ville Syrjälä <ville.syrjala@linux.intel.com>

Make sure drm_vblank_get() never succeeds when called between
drm_vblank_off() and drm_vblank_on(). Borrow a trick from the
old drm_vblank_{pre,post}_modeset() functions and just bump
the refcount in drm_vblank_off() and drop it in drm_vblank_on().

Hopefully the use of inmodeset won't conflict badly with
drm_vblank_{pre,post}_modeset().

For i915 there's a window between drm_vblank_off() and marking the
crtc as inactive where the current code still allows drm_vblank_get().

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
---
 drivers/gpu/drm/drm_irq.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

Comments

Daniel Vetter May 26, 2014, 1:21 p.m. UTC | #1 
On Mon, May 26, 2014 at 02:46:24PM +0300, ville.syrjala@linux.intel.com wrote:
> From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> 
> Make sure drm_vblank_get() never succeeds when called between
> drm_vblank_off() and drm_vblank_on(). Borrow a trick from the
> old drm_vblank_{pre,post}_modeset() functions and just bump
> the refcount in drm_vblank_off() and drop it in drm_vblank_on().
> 
> Hopefully the use of inmodeset won't conflict badly with
> drm_vblank_{pre,post}_modeset().
> 
> For i915 there's a window between drm_vblank_off() and marking the
> crtc as inactive where the current code still allows drm_vblank_get().
> 
> Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>

Afaics we never check anywhere for inmodeset in drm_vblank_get, so how
does this work?
-Daniel

> ---
>  drivers/gpu/drm/drm_irq.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c
> index b565372..7b1e08c 100644
> --- a/drivers/gpu/drm/drm_irq.c
> +++ b/drivers/gpu/drm/drm_irq.c
> @@ -1039,6 +1039,15 @@ void drm_vblank_off(struct drm_device *dev, int crtc)
>  	}
>  	spin_unlock(&dev->event_lock);
>  
> +	/*
> +	 * Prevent subsequent drm_vblank_get() from re-enabling
> +	 * the vblank interrupt by bumping the refcount.
> +	 */
> +	if (!dev->vblank[crtc].inmodeset) {
> +		atomic_inc(&dev->vblank[crtc].refcount);
> +		dev->vblank[crtc].inmodeset = 1;
> +	}
> +
>  	spin_unlock_irqrestore(&dev->vbl_lock, irqflags);
>  }
>  EXPORT_SYMBOL(drm_vblank_off);
> @@ -1079,6 +1088,11 @@ void drm_vblank_on(struct drm_device *dev, int crtc)
>  	unsigned long irqflags;
>  
>  	spin_lock_irqsave(&dev->vbl_lock, irqflags);
> +	/* Drop our private "prevent drm_vblank_get" refcount */
> +	if (dev->vblank[crtc].inmodeset) {
> +		atomic_dec(&dev->vblank[crtc].refcount);
> +		dev->vblank[crtc].inmodeset = 0;
> +	}
>  	/* re-enable interrupts if there's are users left */
>  	if (atomic_read(&dev->vblank[crtc].refcount) != 0)
>  		WARN_ON(drm_vblank_enable(dev, crtc));
> -- 
> 1.8.5.5
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Ville Syrjälä May 26, 2014, 1:32 p.m. UTC | #2 
On Mon, May 26, 2014 at 03:21:41PM +0200, Daniel Vetter wrote:
> On Mon, May 26, 2014 at 02:46:24PM +0300, ville.syrjala@linux.intel.com wrote:
> > From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > 
> > Make sure drm_vblank_get() never succeeds when called between
> > drm_vblank_off() and drm_vblank_on(). Borrow a trick from the
> > old drm_vblank_{pre,post}_modeset() functions and just bump
> > the refcount in drm_vblank_off() and drop it in drm_vblank_on().
> > 
> > Hopefully the use of inmodeset won't conflict badly with
> > drm_vblank_{pre,post}_modeset().
> > 
> > For i915 there's a window between drm_vblank_off() and marking the
> > crtc as inactive where the current code still allows drm_vblank_get().
> > 
> > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> 
> Afaics we never check anywhere for inmodeset in drm_vblank_get, so how
> does this work?

If refcount is already >0 and vblank irq is disabled, drm_vblank_get() will
return -EINVAL.

> -Daniel
> 
> > ---
> >  drivers/gpu/drm/drm_irq.c | 14 ++++++++++++++
> >  1 file changed, 14 insertions(+)
> > 
> > diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c
> > index b565372..7b1e08c 100644
> > --- a/drivers/gpu/drm/drm_irq.c
> > +++ b/drivers/gpu/drm/drm_irq.c
> > @@ -1039,6 +1039,15 @@ void drm_vblank_off(struct drm_device *dev, int crtc)
> >  	}
> >  	spin_unlock(&dev->event_lock);
> >  
> > +	/*
> > +	 * Prevent subsequent drm_vblank_get() from re-enabling
> > +	 * the vblank interrupt by bumping the refcount.
> > +	 */
> > +	if (!dev->vblank[crtc].inmodeset) {
> > +		atomic_inc(&dev->vblank[crtc].refcount);
> > +		dev->vblank[crtc].inmodeset = 1;
> > +	}
> > +
> >  	spin_unlock_irqrestore(&dev->vbl_lock, irqflags);
> >  }
> >  EXPORT_SYMBOL(drm_vblank_off);
> > @@ -1079,6 +1088,11 @@ void drm_vblank_on(struct drm_device *dev, int crtc)
> >  	unsigned long irqflags;
> >  
> >  	spin_lock_irqsave(&dev->vbl_lock, irqflags);
> > +	/* Drop our private "prevent drm_vblank_get" refcount */
> > +	if (dev->vblank[crtc].inmodeset) {
> > +		atomic_dec(&dev->vblank[crtc].refcount);
> > +		dev->vblank[crtc].inmodeset = 0;
> > +	}
> >  	/* re-enable interrupts if there's are users left */
> >  	if (atomic_read(&dev->vblank[crtc].refcount) != 0)
> >  		WARN_ON(drm_vblank_enable(dev, crtc));
> > -- 
> > 1.8.5.5
> > 
> > _______________________________________________
> > Intel-gfx mailing list
> > Intel-gfx@lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/intel-gfx
> 
> -- 
> Daniel Vetter
> Software Engineer, Intel Corporation
> +41 (0) 79 365 57 48 - http://blog.ffwll.ch
Daniel Vetter May 26, 2014, 1:47 p.m. UTC | #3 
On Mon, May 26, 2014 at 04:32:02PM +0300, Ville Syrjälä wrote:
> On Mon, May 26, 2014 at 03:21:41PM +0200, Daniel Vetter wrote:
> > On Mon, May 26, 2014 at 02:46:24PM +0300, ville.syrjala@linux.intel.com wrote:
> > > From: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > > 
> > > Make sure drm_vblank_get() never succeeds when called between
> > > drm_vblank_off() and drm_vblank_on(). Borrow a trick from the
> > > old drm_vblank_{pre,post}_modeset() functions and just bump
> > > the refcount in drm_vblank_off() and drop it in drm_vblank_on().
> > > 
> > > Hopefully the use of inmodeset won't conflict badly with
> > > drm_vblank_{pre,post}_modeset().
> > > 
> > > For i915 there's a window between drm_vblank_off() and marking the
> > > crtc as inactive where the current code still allows drm_vblank_get().
> > > 
> > > Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
> > 
> > Afaics we never check anywhere for inmodeset in drm_vblank_get, so how
> > does this work?
> 
> If refcount is already >0 and vblank irq is disabled, drm_vblank_get() will
> return -EINVAL.

Oh right, I've missed that. drm_irq.c is never short of suprises. With
that enlightenment (and maybe a pimped commit message for blind people
like me) this is Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> 
> > -Daniel
> > 
> > > ---
> > >  drivers/gpu/drm/drm_irq.c | 14 ++++++++++++++
> > >  1 file changed, 14 insertions(+)
> > > 
> > > diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c
> > > index b565372..7b1e08c 100644
> > > --- a/drivers/gpu/drm/drm_irq.c
> > > +++ b/drivers/gpu/drm/drm_irq.c
> > > @@ -1039,6 +1039,15 @@ void drm_vblank_off(struct drm_device *dev, int crtc)
> > >  	}
> > >  	spin_unlock(&dev->event_lock);
> > >  
> > > +	/*
> > > +	 * Prevent subsequent drm_vblank_get() from re-enabling
> > > +	 * the vblank interrupt by bumping the refcount.
> > > +	 */
> > > +	if (!dev->vblank[crtc].inmodeset) {
> > > +		atomic_inc(&dev->vblank[crtc].refcount);
> > > +		dev->vblank[crtc].inmodeset = 1;
> > > +	}
> > > +
> > >  	spin_unlock_irqrestore(&dev->vbl_lock, irqflags);
> > >  }
> > >  EXPORT_SYMBOL(drm_vblank_off);
> > > @@ -1079,6 +1088,11 @@ void drm_vblank_on(struct drm_device *dev, int crtc)
> > >  	unsigned long irqflags;
> > >  
> > >  	spin_lock_irqsave(&dev->vbl_lock, irqflags);
> > > +	/* Drop our private "prevent drm_vblank_get" refcount */
> > > +	if (dev->vblank[crtc].inmodeset) {
> > > +		atomic_dec(&dev->vblank[crtc].refcount);
> > > +		dev->vblank[crtc].inmodeset = 0;
> > > +	}
> > >  	/* re-enable interrupts if there's are users left */
> > >  	if (atomic_read(&dev->vblank[crtc].refcount) != 0)
> > >  		WARN_ON(drm_vblank_enable(dev, crtc));
> > > -- 
> > > 1.8.5.5
> > > 
> > > _______________________________________________
> > > Intel-gfx mailing list
> > > Intel-gfx@lists.freedesktop.org
> > > http://lists.freedesktop.org/mailman/listinfo/intel-gfx
> > 
> > -- 
> > Daniel Vetter
> > Software Engineer, Intel Corporation
> > +41 (0) 79 365 57 48 - http://blog.ffwll.ch
> 
> -- 
> Ville Syrjälä
> Intel OTC
diff mbox

Patch

diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c
index b565372..7b1e08c 100644
--- a/drivers/gpu/drm/drm_irq.c
+++ b/drivers/gpu/drm/drm_irq.c
@@ -1039,6 +1039,15 @@  void drm_vblank_off(struct drm_device *dev, int crtc)
 	}
 	spin_unlock(&dev->event_lock);
 
+	/*
+	 * Prevent subsequent drm_vblank_get() from re-enabling
+	 * the vblank interrupt by bumping the refcount.
+	 */
+	if (!dev->vblank[crtc].inmodeset) {
+		atomic_inc(&dev->vblank[crtc].refcount);
+		dev->vblank[crtc].inmodeset = 1;
+	}
+
 	spin_unlock_irqrestore(&dev->vbl_lock, irqflags);
 }
 EXPORT_SYMBOL(drm_vblank_off);
@@ -1079,6 +1088,11 @@  void drm_vblank_on(struct drm_device *dev, int crtc)
 	unsigned long irqflags;
 
 	spin_lock_irqsave(&dev->vbl_lock, irqflags);
+	/* Drop our private "prevent drm_vblank_get" refcount */
+	if (dev->vblank[crtc].inmodeset) {
+		atomic_dec(&dev->vblank[crtc].refcount);
+		dev->vblank[crtc].inmodeset = 0;
+	}
 	/* re-enable interrupts if there's are users left */
 	if (atomic_read(&dev->vblank[crtc].refcount) != 0)
 		WARN_ON(drm_vblank_enable(dev, crtc));