[igt-dev] [PATCH i-g-t 4/4] lib/igt_fb: Generalize the slow read from gtt mmap handling

[Ville Syrjälä]

On Fri, Nov 02, 2018 at 07:17:03PM +0000, Chris Wilson wrote:
> Quoting Ville Syrjala (2018-11-02 19:06:51)
> > +static void *convert_src_get(const struct fb_convert *cvt)
> > +{
> > +       void *buf;
> > +
> > +       if (!cvt->src.slow_reads)
> > +               return cvt->src.ptr;
> > +
> > +       /*
> > +        * Reading from the BO is awfully slow because of lack of read caching,
> > +        * it's faster to copy the whole BO to a temporary buffer and convert
> > +        * from there.
> > +        */
> > +       buf = malloc(cvt->src.fb->size);
> 
> if (!buf)
> 	return cvt->src.ptr;

Sure, why not.

> 
> > +       igt_memcpy_from_wc(buf, cvt->src.ptr, cvt->src.fb->size);
> > +
> > +       return buf;
> > +}
> > +
> > +static void convert_src_put(const struct fb_convert *cvt,
> > +                           void *src_buf)
> > +{
> > +       /*
> > +        * Reading from the BO is awfully slow because of lack of read caching,
> > +        * it's faster to copy the whole BO to a temporary buffer and convert
> > +        * from there.
> > +        */
> > +       if (cvt->src.slow_reads)
> 
> 	if (src_buf != cvt->src.ptr)

Yes, that is better (tm).

> 
> > +               free(src_buf);
> > +}

-- 
Ville Syrjälä
Intel