[Mesa-dev] [PATCH 14/33] intel/blorp: Add an entrypoint for doing bit-for-bit copies
Jason Ekstrand
jason at jlekstrand.net
Wed Sep 7 16:50:19 UTC 2016
On Wed, Sep 7, 2016 at 9:36 AM, Nanley Chery <nanleychery at gmail.com> wrote:
> On Tue, Sep 06, 2016 at 05:02:55PM -0700, Jason Ekstrand wrote:
> > On Tue, Sep 6, 2016 at 4:12 PM, Nanley Chery <nanleychery at gmail.com>
> wrote:
> >
> > > On Wed, Aug 31, 2016 at 02:22:33PM -0700, Jason Ekstrand wrote:
> > > > ---
> > > > src/intel/blorp/blorp.h | 10 ++++
> > > > src/intel/blorp/blorp_blit.c | 133 ++++++++++++++++++++++++++++++
> > > +++++++++++++
> > > > 2 files changed, 143 insertions(+)
> > > >
> > > > diff --git a/src/intel/blorp/blorp.h b/src/intel/blorp/blorp.h
> > > > index c1e93fd..6574124 100644
> > > > --- a/src/intel/blorp/blorp.h
> > > > +++ b/src/intel/blorp/blorp.h
> > > > @@ -109,6 +109,16 @@ blorp_blit(struct blorp_batch *batch,
> > > > uint32_t filter, bool mirror_x, bool mirror_y);
> > > >
> > > > void
> > > > +blorp_copy(struct blorp_batch *batch,
> > > > + const struct blorp_surf *src_surf,
> > > > + unsigned src_level, unsigned src_layer,
> > > > + const struct blorp_surf *dst_surf,
> > > > + unsigned dst_level, unsigned dst_layer,
> > > > + uint32_t src_x, uint32_t src_y,
> > > > + uint32_t dst_x, uint32_t dst_y,
> > > > + uint32_t src_width, uint32_t src_height);
> > > > +
> > > > +void
> > > > blorp_fast_clear(struct blorp_batch *batch,
> > > > const struct blorp_surf *surf,
> > > > uint32_t level, uint32_t layer, enum isl_format
> format,
> > > > diff --git a/src/intel/blorp/blorp_blit.c
> b/src/intel/blorp/blorp_blit.c
> > > > index 3ab39a3..42a502c 100644
> > > > --- a/src/intel/blorp/blorp_blit.c
> > > > +++ b/src/intel/blorp/blorp_blit.c
> > > > @@ -1685,3 +1685,136 @@ blorp_blit(struct blorp_batch *batch,
> > > > dst_x0, dst_y0, dst_x1, dst_y1,
> > > > mirror_x, mirror_y);
> > > > }
> > > > +
> > > > +static enum isl_format
> > > > +get_copy_format_for_bpb(unsigned bpb)
> > > > +{
> > > > + /* The choice of UNORM and UINT formats is very intentional here.
> > > Most of
> > > > + * the time, we want to use a UINT format to avoid any rounding
> > > error in
> > > > + * the blit. For stencil blits, R8_UINT is required by the
> hardware.
> > > > + * (It's the only format allowed in conjunction with W-tiling.)
> > > Also we
> > > > + * intentionally use the 4-channel formats whenever we can.
> This is
> > > so
> > > > + * that, when we do a RGB <-> RGBX copy, the two formats will
> line
> > > up even
> > > > + * though one of them is 3/4 the size of the other. The choice
> of
> > > UNORM
> > > > + * vs. UINT is also very intentional because Haswell doesn't
> handle
> > > 8 or
> > > > + * 16-bit RGB UINT formats at all so we have to use UNORM there.
> > > > + * Fortunately, the only time we should ever use two different
> > > formats in
> > > > + * the table below is for RGB -> RGBA blits and so we will never
> > > have any
> > > > + * UNORM/UINT mismatch.
> > > > + */
> > > > + switch (bpb) {
> > > > + case 8: return ISL_FORMAT_R8_UINT;
> > > > + case 16: return ISL_FORMAT_R8G8_UINT;
> > > > + case 24: return ISL_FORMAT_R8G8B8_UNORM;
> > > > + case 32: return ISL_FORMAT_R8G8B8A8_UNORM;
> > > > + case 48: return ISL_FORMAT_R16G16B16_UNORM;
> > > > + case 64: return ISL_FORMAT_R16G16B16A16_UNORM;
> > > > + case 96: return ISL_FORMAT_R32G32B32_UINT;
> > > > + case 128:return ISL_FORMAT_R32G32B32A32_UINT;
> > > > + default:
> > > > + unreachable("Unknown format bpb");
> > > > + }
> > > > +}
> > > > +
> > > > +static void
> > > > +surf_convert_to_uncompressed(const struct isl_device *isl_dev,
> > > > + struct brw_blorp_surface_info *info,
> > > > + uint32_t *x, uint32_t *y,
> > > > + uint32_t *width, uint32_t *height)
> > > > +{
> > > > + const struct isl_format_layout *fmtl =
> > > > + isl_format_get_layout(info->surf.format);
> > > > +
> > > > + assert(fmtl->bw > 1 || fmtl->bh > 1);
> > > > +
> > > > + /* This is a compressed surface. We need to convert it to a
> single
> > > > + * slice (becase compressed layouts don't perfectly match
> > > uncompressed
> > > > + * ones with the same bpb) and divide x, y, width, and height by
> the
> > > > + * block size.
> > > > + */
> > > > + surf_convert_to_single_slice(isl_dev, info);
> > > > +
> > > > + if (width || height) {
> > > > + assert(*width % fmtl->bw == 0 ||
> > > > + *x + *width == info->surf.logical_level0_px.width);
> > > > + assert(*height % fmtl->bh == 0 ||
> > > > + *y + *height == info->surf.logical_level0_px.height);
> > > > + *width = DIV_ROUND_UP(*width, fmtl->bw);
> > > > + *height = DIV_ROUND_UP(*height, fmtl->bh);
> > > > + }
> > > > +
> > > > + assert(*x % fmtl->bw == 0);
> > > > + assert(*y % fmtl->bh == 0);
> > > > + *x /= fmtl->bw;
> > > > + *y /= fmtl->bh;
> > > > +
> > > > + info->surf.logical_level0_px.width =
> > > > + DIV_ROUND_UP(info->surf.logical_level0_px.width, fmtl->bw);
> > > > + info->surf.logical_level0_px.height =
> > > > + DIV_ROUND_UP(info->surf.logical_level0_px.height, fmtl->bh);
> > > > +
> > > > + assert(info->surf.phys_level0_sa.width % fmtl->bw == 0);
> > > > + assert(info->surf.phys_level0_sa.height % fmtl->bh == 0);
> > > > + info->surf.phys_level0_sa.width /= fmtl->bw;
> > > > + info->surf.phys_level0_sa.height /= fmtl->bh;
> > > > +
> > > > + assert(info->tile_x_sa % fmtl->bw == 0);
> > > > + assert(info->tile_y_sa % fmtl->bh == 0);
> > > > + info->tile_x_sa /= fmtl->bw;
> > > > + info->tile_y_sa /= fmtl->bh;
> > > > +
> > > > + /* It's now an uncompressed surface so we need an uncompressed
> > > format */
> > > > + info->surf.format = get_copy_format_for_bpb(fmtl->bpb);
> > > > +}
> > > > +
> > > > +void
> > > > +blorp_copy(struct blorp_batch *batch,
> > > > + const struct blorp_surf *src_surf,
> > > > + unsigned src_level, unsigned src_layer,
> > > > + const struct blorp_surf *dst_surf,
> > > > + unsigned dst_level, unsigned dst_layer,
> > > > + uint32_t src_x, uint32_t src_y,
> > > > + uint32_t dst_x, uint32_t dst_y,
> > > > + uint32_t src_width, uint32_t src_height)
> > > > +{
> > > > + struct blorp_params params;
> > > > + blorp_params_init(¶ms);
> > > > +
> > > > + brw_blorp_surface_info_init(batch->blorp, ¶ms.src, src_surf,
> > > src_level,
> > > > + src_layer, ISL_FORMAT_UNSUPPORTED,
> > > false);
> > > > + brw_blorp_surface_info_init(batch->blorp, ¶ms.dst, dst_surf,
> > > dst_level,
> > > > + dst_layer, ISL_FORMAT_UNSUPPORTED,
> true);
> > > > +
> > > > + struct brw_blorp_blit_prog_key wm_prog_key;
> > > > + memset(&wm_prog_key, 0, sizeof(wm_prog_key));
> > > > +
> > > > + const struct isl_format_layout *src_fmtl =
> > > > + isl_format_get_layout(params.src.surf.format);
> > > > + const struct isl_format_layout *dst_fmtl =
> > > > + isl_format_get_layout(params.dst.surf.format);
> > > > +
> > > > + params.src.view.format = get_copy_format_for_bpb(src_fmtl->bpb);
> > > > + if (src_fmtl->bw > 1 || src_fmtl->bh > 1) {
> > > > + surf_convert_to_uncompressed(batch->blorp->isl_dev,
> ¶ms.src,
> > > > + &src_x, &src_y, &src_width,
> > > &src_height);
> > > > + wm_prog_key.need_dst_offset = true;
> > > > + }
> > > > +
> > > > + params.dst.view.format = get_copy_format_for_bpb(dst_fmtl->bpb);
> > > > + if (dst_fmtl->bw > 1 || dst_fmtl->bh > 1) {
> > > > + surf_convert_to_uncompressed(batch->blorp->isl_dev,
> ¶ms.dst,
> > > > + &dst_x, &dst_y, NULL, NULL);
> > > > + wm_prog_key.need_dst_offset = true;
> > > > + }
> > >
> > > When this function is later used to replace meta in vkCopyImage and the
> > > like,
> > > I suspect that copying non-zero mip levels on D16 and S8 textures can
> fail.
> > > These textures have a non-4x4 alignment on SKL and, like compressed
> > > textures,
> > > need surf_convert_to_single_slice().
> > >
> >
> > I don't think that's an issue. Depth and stencil are still valid
> multi-LOD
> > textures. You don't need 4x4, you just need an alignment that we can
> > validly express so 4x8 (stencil), for instance, is fine. The problem
> with
> > compressed is that, when you fake it as R32U32G32A32_UINT, you can end up
> > with halign/valign that can't actually be expressed. On SKL, I think we
> > can actually express them all so we may not need the stomp at all.
> >
> >
>
> My suspicion was wrong. I was under the assumption that blorp behaved like
> meta in that it always created an image from the tile nearest to the copy
> region
> of the src/dst image - that is not the case. Sorry for not simply asking.
> This
> seems like a much better way to copy images.
>
> This series currently causes 6 crucbile tests to fail in
> func.miptree.r8g8b8a8-unorm.aspect-color.view-3d. Does your local copy fix
> these? If so, could you send it out?
>
Drp... I didn't run crucible. No CTS failures but I'm seeing the same 6
crucible fails that you are. I'll look into them.
--Jason
> > > One idea is to call surf_convert_to_single_slice() when ever a copy
> occurs
> > > with an image's subresource that isn't the 1st mip level on the 1st
> array
> > > layer
> > > /depth slice and the alignment differs from what would be used the copy
> > > format.
> > > Thoughts?
> > >
> >
> > That seems a bit harsh.
> >
> >
>
> I agree.
>
> - Nanley
>
> > > - Nanley
> > >
> > > > +
> > > > + /* Once both surfaces are stompped to uncompressed as needed, the
> > > > + * destination size is the same as the source size.
> > > > + */
> > > > + uint32_t dst_width = src_width;
> > > > + uint32_t dst_height = src_height;
> > > > +
> > > > + do_blorp_blit(batch, ¶ms, &wm_prog_key,
> > > > + src_x, src_y, src_x + src_width, src_y +
> src_height,
> > > > + dst_x, dst_y, dst_x + dst_width, dst_y +
> dst_height,
> > > > + false, false);
> > > > +}
> > > > --
> > > > 2.5.0.400.gff86faf
> > > >
> > > > _______________________________________________
> > > > mesa-dev mailing list
> > > > mesa-dev at lists.freedesktop.org
> > > > https://lists.freedesktop.org/mailman/listinfo/mesa-dev
> > >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20160907/a896e016/attachment-0001.html>
More information about the mesa-dev
mailing list