[igt-dev] [PATCH i-g-t v5 07/13] igt: fb: Don't pass the stride when allocating a dumb, multi-planar buffer

[Lyude Paul]

On Wed, 2019-01-30 at 16:38 +0100, Maxime Ripard wrote:
> Hi!
> 
> On Tue, Jan 29, 2019 at 03:08:10PM -0500, Lyude Paul wrote:
> > On Fri, 2019-01-25 at 15:58 +0100, Maxime Ripard wrote:
> > > The dumb buffer allocation API only considers a single plane, and even
> > > though allocating multi-planar buffers through it is allowed, the stride
> > > it
> > > gives back is the the width times the bpp passed as an argument.
> > > 
> > > That doesn't work in our case, since the bpp is going to be the one we
> > > give
> > > as an argument, but split over three planes so the stride doesn't match
> > > anymore.
> > > 
> > > A proper fix for this would be to have a better dumb buffer allocation
> > > API,
> > > but for the time being, let's do it that way.
> > > 
> > > Reviewed-by: Paul Kocialkowski <paul.kocialkowski at bootlin.com>
> > > Signed-off-by: Maxime Ripard <maxime.ripard at bootlin.com>
> > > ---
> > >  lib/igt_fb.c | 17 ++++++++++++++++-
> > >  1 file changed, 16 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/lib/igt_fb.c b/lib/igt_fb.c
> > > index 1c52aebb674e..048d274e5d36 100644
> > > --- a/lib/igt_fb.c
> > > +++ b/lib/igt_fb.c
> > > @@ -530,6 +530,7 @@ static int create_bo_for_fb(struct igt_fb *fb)
> > >  {
> > >  	const struct format_desc_struct *fmt = lookup_drm_format(fb-
> > > > drm_format);
> > >  	unsigned int plane, bpp;
> > > +	unsigned *strides = &fb->strides[0];
> > >  	int fd = fb->fd;
> > >  
> > >  	if (fb->tiling || fb->size || fb->strides[0] || igt_format_is_yuv(fb-
> > > > drm_format)) {
> > > @@ -575,8 +576,22 @@ static int create_bo_for_fb(struct igt_fb *fb)
> > >  				    plane ? fmt->hsub * fmt->vsub : 1);
> > >  
> > >  	fb->is_dumb = true;
> > > +
> > > +	/*
> > > +	 * We can't really pass the stride array here since the dumb
> > > +	 * buffer allocation is assuming that it operates on one
> > > +	 * plane, and therefore will calculate the stride as if each
> > > +	 * pixels were stored on a single plane.
> > 
> > nitpick: s/each pixels were/each pixel was/
> > 
> > > +	 *
> > > +	 * This might cause issues at some point on drivers that would
> > > +	 * change the stride of YUV buffers, but we haven't
> > > +	 * encountered any yet.
> > > +	 */
> > Is it possible to add an igt_assert to check for this? Either way, with
> > those
> > changes:
> 
> I'm not sure we can test this actually. If we don't pass any pointer,
> then we don't get a stride back, and if we do then we will always get
> a stride that is wrong for what we're trying to do :/
Ah, makes sense
> 
> Maxime
> 
-- 
Cheers,
	Lyude Paul