On Mon, Jan 16, 2017 at 10:05:21PM +0000, Matthew Auld wrote:
> On 11 January 2017 at 21:09, Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> wrote:
> > Create partial mappings to cover a large object, investigating tiling
> > (fenced regions) and VMA reuse.
> >
> > Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
> > ---
> > drivers/gpu/drm/i915/selftests/i915_gem_object.c | 252 +++++++++++++++++++++++
> > 1 file changed, 252 insertions(+)
> >
> > diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_object.c b/drivers/gpu/drm/i915/selftests/i915_gem_object.c
> > index 08e6b49b1e77..df3625f551aa 100644
> > --- a/drivers/gpu/drm/i915/selftests/i915_gem_object.c
> > +++ b/drivers/gpu/drm/i915/selftests/i915_gem_object.c
> > @@ -132,6 +132,257 @@ static int igt_gem_huge(void *arg)
> > return err;
> > }
> >
> > +struct tile {
> > + unsigned int width;
> > + unsigned int height;
> > + unsigned int stride;
> > + unsigned int size;
> > + unsigned int tiling;
> > + unsigned int swizzle;
> > +};
> > +
> > +static u64 swizzle_bit(unsigned bit, u64 offset)
> > +{
> > + return (offset & BIT_ULL(bit)) >> (bit - 6);
> > +}
> > +
> > +static u64 tiled_offset(const struct tile *tile, u64 v)
> > +{
> > + u64 x, y;
> > +
> > + if (tile->tiling == I915_TILING_NONE)
> > + return v;
> > +
> > + switch (tile->swizzle) {
> > + case I915_BIT_6_SWIZZLE_9:
> > + v ^= swizzle_bit(9, v);
> > + break;
> > + case I915_BIT_6_SWIZZLE_9_10:
> > + v ^= swizzle_bit(9, v) ^ swizzle_bit(10, v);
> > + break;
> > + case I915_BIT_6_SWIZZLE_9_11:
> > + v ^= swizzle_bit(9, v) ^ swizzle_bit(11, v);
> > + break;
> > + case I915_BIT_6_SWIZZLE_9_10_11:
> > + v ^= swizzle_bit(9, v) ^ swizzle_bit(10, v) ^ swizzle_bit(11, v);
> > + break;
> > + }
> > +
> > + y = div64_u64_rem(v, tile->stride, &x);
> > + v = div64_u64_rem(y, tile->height, &y) * tile->stride * tile->height;
> > +
> > + if (tile->tiling == I915_TILING_X) {
> > + v += y * tile->width;
> > + v += div64_u64_rem(x, tile->width, &x) << tile->size;
> > + v += x;
> > + } else {
> > + const unsigned int ytile_span = 16;
> > + const unsigned int ytile_height = 32 * ytile_span;
> I don't follow this part, is this a smaller tile within a tile ?
Yes, that is one way of thinking about y-tiling, cacheline tiles within
page tiles.
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
