Re: [PATCH 10/18] drm/irq: track the irq installed in drm_irq_install in dev->irq

From: Daniel Vetter <daniel@ffwll.ch>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>, dri-devel@lists.freedesktop.org
Subject: Re: [PATCH 10/18] drm/irq: track the irq installed in drm_irq_install in dev->irq
Date: Tue, 22 Apr 2014 11:46:45 +0200	[thread overview]
Message-ID: <20140422094645.GN10722@phenom.ffwll.local> (raw)
In-Reply-To: <1978319.Njcif9FTVM@avalon>

On Thu, Apr 17, 2014 at 12:02:07AM +0200, Laurent Pinchart wrote:
> And another comment...
> 
> On Friday 11 April 2014 23:36:07 Daniel Vetter wrote:
> > To get rid of the dev->bus->get_irq callback we need to pass in the
> > desired irq explicitly into drm_irq_install. To avoid having to do the
> > same for drm_irq_unistall just track it internally. That leaves
> > drivers with less room to botch things up.
> > 
> > v2: Add the hunk lost in an earlier patch to this one (Thierry).
> > 
> > Cc: Thierry Reding <thierry.reding@gmail.com>
> > Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
> > ---
> >  drivers/gpu/drm/drm_irq.c | 18 +++++++++++-------
> >  include/drm/drmP.h        |  2 ++
> >  2 files changed, 13 insertions(+), 7 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/drm_irq.c b/drivers/gpu/drm/drm_irq.c
> > index 330e85b19115..1c3b6229363d 100644
> > --- a/drivers/gpu/drm/drm_irq.c
> > +++ b/drivers/gpu/drm/drm_irq.c
> > @@ -249,14 +249,16 @@ static inline int drm_dev_to_irq(struct drm_device
> > *dev) */
> >  int drm_irq_install(struct drm_device *dev)
> >  {
> > -	int ret;
> > +	int ret, irq;
> >  	unsigned long sh_flags = 0;
> >  	char *irqname;
> > 
> > +	irq = drm_dev_to_irq(dev);
> > +
> >  	if (!drm_core_check_feature(dev, DRIVER_HAVE_IRQ))
> >  		return -EINVAL;
> > 
> > -	if (drm_dev_to_irq(dev) == 0)
> > +	if (irq == 0)
> 
> Isn't 0 a valid IRQ number ? Shouldn't you check for irq < 0 instead ? At 
> least platform_get_irq() returns a negative error value on failure.

tbh I'm not really clear on how this works. If we want to change this then
I think that should be a separate patch. Also it might be better to
extract this check into drm_control, which is the only function that
really needs it since it is called by userspace.

For all other places it is a simple driver bug and I'm not sure how much
we should care - request_irq should catch any abuse already.

In any case this should be a separate patch.
-Daniel

> 
> >  		return -EINVAL;
> > 
> >  	/* Driver must have been initialized */
> > @@ -267,7 +269,7 @@ int drm_irq_install(struct drm_device *dev)
> >  		return -EBUSY;
> >  	dev->irq_enabled = true;
> > 
> > -	DRM_DEBUG("irq=%d\n", drm_dev_to_irq(dev));
> > +	DRM_DEBUG("irq=%d\n", dev->irq);
> > 
> >  	/* Before installing handler */
> >  	if (dev->driver->irq_preinstall)
> > @@ -282,7 +284,7 @@ int drm_irq_install(struct drm_device *dev)
> >  	else
> >  		irqname = dev->driver->name;
> > 
> > -	ret = request_irq(drm_dev_to_irq(dev), dev->driver->irq_handler,
> > +	ret = request_irq(dev->irq, dev->driver->irq_handler,
> >  			  sh_flags, irqname, dev);
> > 
> >  	if (ret < 0) {
> > @@ -301,7 +303,9 @@ int drm_irq_install(struct drm_device *dev)
> >  		dev->irq_enabled = false;
> >  		if (!drm_core_check_feature(dev, DRIVER_MODESET))
> >  			vga_client_register(dev->pdev, NULL, NULL, NULL);
> > -		free_irq(drm_dev_to_irq(dev), dev);
> > +		free_irq(dev->irq, dev);
> > +	} else {
> > +		dev->irq = irq;
> >  	}
> > 
> >  	return ret;
> > @@ -344,7 +348,7 @@ int drm_irq_uninstall(struct drm_device *dev)
> >  	if (!irq_enabled)
> >  		return -EINVAL;
> > 
> > -	DRM_DEBUG("irq=%d\n", drm_dev_to_irq(dev));
> > +	DRM_DEBUG("irq=%d\n", dev->irq);
> > 
> >  	if (!drm_core_check_feature(dev, DRIVER_MODESET))
> >  		vga_client_register(dev->pdev, NULL, NULL, NULL);
> > @@ -352,7 +356,7 @@ int drm_irq_uninstall(struct drm_device *dev)
> >  	if (dev->driver->irq_uninstall)
> >  		dev->driver->irq_uninstall(dev);
> > 
> > -	free_irq(drm_dev_to_irq(dev), dev);
> > +	free_irq(dev->irq, dev);
> > 
> >  	return 0;
> >  }
> > diff --git a/include/drm/drmP.h b/include/drm/drmP.h
> > index 8b23a34a103e..6f512cd97cd5 100644
> > --- a/include/drm/drmP.h
> > +++ b/include/drm/drmP.h
> > @@ -1107,6 +1107,8 @@ struct drm_device {
> >  	/** \name Context support */
> >  	/*@{ */
> >  	bool irq_enabled;		/**< True if irq handler is enabled */
> > +	int irq;
> > +
> >  	__volatile__ long context_flag;	/**< Context swapping flag */
> >  	int last_context;		/**< Last current context */
> >  	/*@} */
> 
> -- 
> Regards,
> 
> Laurent Pinchart
> 

-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch