[PATCH 19/32] drm: Store a pointer to drm_format_info under drm_framebuffer

Ville Syrjälä ville.syrjala at linux.intel.com
Thu Nov 17 18:06:10 UTC 2016


On Thu, Nov 17, 2016 at 07:57:27PM +0200, Laurent Pinchart wrote:
> Hi Ville,
> 
> Thank you for the patch.
> 
> On Thursday 17 Nov 2016 18:14:18 ville.syrjala at linux.intel.com wrote:
> > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > 
> > To avoid having to look up the format information struct every time,
> > let's just store a pointer to it under drm_framebuffer.
> > 
> > Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> > ---
> >  drivers/gpu/drm/drm_framebuffer.c    | 4 +++-
> >  drivers/gpu/drm/drm_modeset_helper.c | 1 +
> >  include/drm/drm_framebuffer.h        | 4 ++++
> >  3 files changed, 8 insertions(+), 1 deletion(-)
> > 
> > diff --git a/drivers/gpu/drm/drm_framebuffer.c
> > b/drivers/gpu/drm/drm_framebuffer.c index 527220c08f9b..47478678d609 100644
> > --- a/drivers/gpu/drm/drm_framebuffer.c
> > +++ b/drivers/gpu/drm/drm_framebuffer.c
> > @@ -632,8 +632,10 @@ int drm_framebuffer_init(struct drm_device *dev, struct
> > drm_framebuffer *fb, int ret;
> > 
> >  	INIT_LIST_HEAD(&fb->filp_head);
> > -	if (WARN_ON_ONCE(fb->dev != dev))
> > +	if (WARN_ON_ONCE(fb->dev != dev)) {
> >  		fb->dev = dev;
> > +		fb->format = drm_format_info(fb->pixel_format);
> > +	}
> 
> 
> With this drivers will start relying on fb->format. I believe we should commit 
> to provide a non-NULL format info to avoid having to sprinkle NULL checks in 
> all drivers. Should drm_framebuffer_init() return an error in that case ? Or 
> do we guarantee already through another mean that the pixel format is valid 
> and will always lead to a non-NULL format info here ?

I think for normal cases we already reject the NULL format info. I am a
little worried that I might have missed some special case where that
might not hold. But I'm not sure what would be the best way to guard
against it. I really don't want to sprinkle NULL checks everywhere, and
in fact I didn't.

> 
> >  	fb->funcs = funcs;
> > 
> >  	ret = drm_mode_object_get_reg(dev, &fb->base, DRM_MODE_OBJECT_FB,
> > diff --git a/drivers/gpu/drm/drm_modeset_helper.c
> > b/drivers/gpu/drm/drm_modeset_helper.c index 57a319e3f780..1aa5e3bcc8a1
> > 100644
> > --- a/drivers/gpu/drm/drm_modeset_helper.c
> > +++ b/drivers/gpu/drm/drm_modeset_helper.c
> > @@ -91,6 +91,7 @@ void drm_helper_mode_fill_fb_struct(struct drm_device
> > *dev, }
> > 
> >  	fb->dev = dev;
> > +	fb->format = info;
> >  	fb->width = mode_cmd->width;
> >  	fb->height = mode_cmd->height;
> >  	for (i = 0; i < 4; i++) {
> > diff --git a/include/drm/drm_framebuffer.h b/include/drm/drm_framebuffer.h
> > index b3141a0e609b..741b3a994d1f 100644
> > --- a/include/drm/drm_framebuffer.h
> > +++ b/include/drm/drm_framebuffer.h
> > @@ -122,6 +122,10 @@ struct drm_framebuffer {
> >  	 */
> >  	struct drm_mode_object base;
> >  	/**
> > +	 * @format: framebuffer format information
> > +	 */
> > +	const struct drm_format_info *format;
> > +	/**
> >  	 * @funcs: framebuffer vfunc table
> >  	 */
> >  	const struct drm_framebuffer_funcs *funcs;
> 
> -- 
> Regards,
> 
> Laurent Pinchart

-- 
Ville Syrjälä
Intel OTC


More information about the dri-devel mailing list