st/mesa: implement ARB_copy_image

I wonder if the craziness was worth it.

Reviewed-by: Brian Paul <brianp@vmware.com>

1 files changed, 578 insertions, 0 deletions

diff --git a/src/mesa/state_tracker/st_cb_copyimage.c b/src/mesa/state_tracker/st_cb_copyimage.c
new file mode 100644
index 0000000..c94a0b2
--- /dev/null
+++ b/src/mesa/state_tracker/st_cb_copyimage.c
@@ -0,0 +1,578 @@
+/*
+ * Copyright 2015 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#include "state_tracker/st_context.h"
+#include "state_tracker/st_cb_copyimage.h"
+#include "state_tracker/st_cb_fbo.h"
+#include "state_tracker/st_texture.h"
+
+#include "util/u_box.h"
+#include "util/u_format.h"
+#include "util/u_inlines.h"
+
+
+/**
+ * Return an equivalent canonical format without "X" channels.
+ *
+ * Copying between incompatible formats is easier when the format is
+ * canonicalized, meaning that it is in a standard form.
+ *
+ * The returned format has the same component sizes and swizzles as
+ * the source format, the type is changed to UINT or UNORM, depending on
+ * which one has the most swizzle combinations in their group.
+ *
+ * If it's not an array format, return a memcpy-equivalent array format.
+ *
+ * The key feature is that swizzled versions of formats of the same
+ * component size always return the same component type.
+ *
+ * X returns A.
+ * Luminance, intensity, alpha, depth, stencil, and 8-bit and 16-bit packed
+ * formats are not supported. (same as ARB_copy_image)
+ */
+static enum pipe_format
+get_canonical_format(enum pipe_format format)
+{
+   const struct util_format_description *desc =
+      util_format_description(format);
+
+   /* Packed formats. Return the equivalent array format. */
+   if (format == PIPE_FORMAT_R11G11B10_FLOAT ||
+       format == PIPE_FORMAT_R9G9B9E5_FLOAT)
+      return get_canonical_format(PIPE_FORMAT_R8G8B8A8_UINT);
+
+   if (desc->nr_channels == 4 &&
+       desc->channel[0].size == 10 &&
+       desc->channel[1].size == 10 &&
+       desc->channel[2].size == 10 &&
+       desc->channel[3].size == 2) {
+      if (desc->swizzle[0] == UTIL_FORMAT_SWIZZLE_X &&
+          desc->swizzle[1] == UTIL_FORMAT_SWIZZLE_Y &&
+          desc->swizzle[2] == UTIL_FORMAT_SWIZZLE_Z)
+         return get_canonical_format(PIPE_FORMAT_R8G8B8A8_UINT);
+
+      return PIPE_FORMAT_NONE;
+   }
+
+#define RETURN_FOR_SWIZZLE1(x, format) \
+   if (desc->swizzle[0] == UTIL_FORMAT_SWIZZLE_##x) \
+      return format
+
+#define RETURN_FOR_SWIZZLE2(x, y, format) \
+   if (desc->swizzle[0] == UTIL_FORMAT_SWIZZLE_##x && \
+       desc->swizzle[1] == UTIL_FORMAT_SWIZZLE_##y) \
+      return format
+
+#define RETURN_FOR_SWIZZLE3(x, y, z, format) \
+   if (desc->swizzle[0] == UTIL_FORMAT_SWIZZLE_##x && \
+       desc->swizzle[1] == UTIL_FORMAT_SWIZZLE_##y && \
+       desc->swizzle[2] == UTIL_FORMAT_SWIZZLE_##z) \
+      return format
+
+#define RETURN_FOR_SWIZZLE4(x, y, z, w, format) \
+   if (desc->swizzle[0] == UTIL_FORMAT_SWIZZLE_##x && \
+       desc->swizzle[1] == UTIL_FORMAT_SWIZZLE_##y && \
+       desc->swizzle[2] == UTIL_FORMAT_SWIZZLE_##z && \
+       desc->swizzle[3] == UTIL_FORMAT_SWIZZLE_##w) \
+      return format
+
+   /* Array formats. */
+   if (desc->is_array) {
+      switch (desc->nr_channels) {
+      case 1:
+         switch (desc->channel[0].size) {
+         case 8:
+            RETURN_FOR_SWIZZLE1(X, PIPE_FORMAT_R8_UINT);
+            break;
+
+         case 16:
+            RETURN_FOR_SWIZZLE1(X, PIPE_FORMAT_R16_UINT);
+            break;
+
+         case 32:
+            RETURN_FOR_SWIZZLE1(X, PIPE_FORMAT_R32_UINT);
+            break;
+         }
+         break;
+
+      case 2:
+         switch (desc->channel[0].size) {
+         case 8:
+            /* All formats in each group must be of the same type.
+             * We can't use UINT for R8G8 while using UNORM for G8R8.
+             */
+            RETURN_FOR_SWIZZLE2(X, Y, PIPE_FORMAT_R8G8_UNORM);
+            RETURN_FOR_SWIZZLE2(Y, X, PIPE_FORMAT_G8R8_UNORM);
+            break;
+
+         case 16:
+            RETURN_FOR_SWIZZLE2(X, Y, PIPE_FORMAT_R16G16_UNORM);
+            RETURN_FOR_SWIZZLE2(Y, X, PIPE_FORMAT_G16R16_UNORM);
+            break;
+
+         case 32:
+            RETURN_FOR_SWIZZLE2(X, Y, PIPE_FORMAT_R32G32_UINT);
+            break;
+         }
+         break;
+
+      case 3:
+         switch (desc->channel[0].size) {
+         case 8:
+            RETURN_FOR_SWIZZLE3(X, Y, Z, PIPE_FORMAT_R8G8B8_UINT);
+            break;
+
+         case 16:
+            RETURN_FOR_SWIZZLE3(X, Y, Z, PIPE_FORMAT_R16G16B16_UINT);
+            break;
+
+         case 32:
+            RETURN_FOR_SWIZZLE3(X, Y, Z, PIPE_FORMAT_R32G32B32_UINT);
+            break;
+         }
+         break;
+
+      case 4:
+         switch (desc->channel[0].size) {
+         case 8:
+            RETURN_FOR_SWIZZLE4(X, Y, Z, W, PIPE_FORMAT_R8G8B8A8_UNORM);
+            RETURN_FOR_SWIZZLE4(X, Y, Z, 1, PIPE_FORMAT_R8G8B8A8_UNORM);
+            RETURN_FOR_SWIZZLE4(Z, Y, X, W, PIPE_FORMAT_B8G8R8A8_UNORM);
+            RETURN_FOR_SWIZZLE4(Z, Y, X, 1, PIPE_FORMAT_B8G8R8A8_UNORM);
+            RETURN_FOR_SWIZZLE4(W, Z, Y, X, PIPE_FORMAT_A8B8G8R8_UNORM);
+            RETURN_FOR_SWIZZLE4(1, Z, Y, X, PIPE_FORMAT_A8B8G8R8_UNORM);
+            RETURN_FOR_SWIZZLE4(W, X, Y, Z, PIPE_FORMAT_A8R8G8B8_UNORM);
+            RETURN_FOR_SWIZZLE4(1, X, Y, Z, PIPE_FORMAT_A8R8G8B8_UNORM);
+            break;
+
+         case 16:
+            RETURN_FOR_SWIZZLE4(X, Y, Z, W, PIPE_FORMAT_R16G16B16A16_UINT);
+            RETURN_FOR_SWIZZLE4(X, Y, Z, 1, PIPE_FORMAT_R16G16B16A16_UINT);
+            break;
+
+         case 32:
+            RETURN_FOR_SWIZZLE4(X, Y, Z, W, PIPE_FORMAT_R32G32B32A32_UINT);
+            RETURN_FOR_SWIZZLE4(X, Y, Z, 1, PIPE_FORMAT_R32G32B32A32_UINT);
+            break;
+         }
+      }
+
+      assert(!"unknown array format");
+      return PIPE_FORMAT_NONE;
+   }
+
+   assert(!"unknown packed format");
+   return PIPE_FORMAT_NONE;
+}
+
+/**
+ * Return true if the swizzle is XYZW in case of a 4-channel format,
+ * XY in case of a 2-channel format, or X in case of a 1-channel format.
+ */
+static bool
+has_identity_swizzle(const struct util_format_description *desc)
+{
+   int i;
+
+   for (i = 0; i < desc->nr_channels; i++)
+      if (desc->swizzle[i] != UTIL_FORMAT_SWIZZLE_X + i)
+         return false;
+
+   return true;
+}
+
+/**
+ * Return a canonical format for the given bits and channel size.
+ */
+static enum pipe_format
+canonical_format_from_bits(unsigned bits, unsigned channel_size)
+{
+   switch (bits) {
+   case 8:
+      if (channel_size == 8)
+         return get_canonical_format(PIPE_FORMAT_R8_UINT);
+      break;
+
+   case 16:
+      if (channel_size == 8)
+         return get_canonical_format(PIPE_FORMAT_R8G8_UINT);
+      if (channel_size == 16)
+         return get_canonical_format(PIPE_FORMAT_R16_UINT);
+      break;
+
+   case 32:
+      if (channel_size == 8)
+         return get_canonical_format(PIPE_FORMAT_R8G8B8A8_UINT);
+      if (channel_size == 16)
+         return get_canonical_format(PIPE_FORMAT_R16G16_UINT);
+      if (channel_size == 32)
+         return get_canonical_format(PIPE_FORMAT_R32_UINT);
+      break;
+
+   case 64:
+      if (channel_size == 16)
+         return get_canonical_format(PIPE_FORMAT_R16G16B16A16_UINT);
+      if (channel_size == 32)
+         return get_canonical_format(PIPE_FORMAT_R32G32_UINT);
+      break;
+
+   case 128:
+      if (channel_size == 32)
+         return get_canonical_format(PIPE_FORMAT_R32G32B32A32_UINT);
+      break;
+   }
+
+   assert(!"impossible format");
+   return PIPE_FORMAT_NONE;
+}
+
+static void
+blit(struct pipe_context *pipe,
+     struct pipe_resource *dst,
+     enum pipe_format dst_format,
+     unsigned dst_level,
+     unsigned dstx, unsigned dsty, unsigned dstz,
+     struct pipe_resource *src,
+     enum pipe_format src_format,
+     unsigned src_level,
+     const struct pipe_box *src_box)
+{
+   struct pipe_blit_info blit = {{0}};
+
+   blit.src.resource = src;
+   blit.dst.resource = dst;
+   blit.src.format = src_format;
+   blit.dst.format = dst_format;
+   blit.src.level = src_level;
+   blit.dst.level = dst_level;
+   blit.src.box = *src_box;
+   u_box_3d(dstx, dsty, dstz, src_box->width, src_box->height,
+            src_box->depth, &blit.dst.box);
+   blit.mask = PIPE_MASK_RGBA;
+   blit.filter = PIPE_TEX_FILTER_NEAREST;
+
+   pipe->blit(pipe, &blit);
+}
+
+static void
+swizzled_copy(struct pipe_context *pipe,
+              struct pipe_resource *dst,
+              unsigned dst_level,
+              unsigned dstx, unsigned dsty, unsigned dstz,
+              struct pipe_resource *src,
+              unsigned src_level,
+              const struct pipe_box *src_box)
+{
+   const struct util_format_description *src_desc, *dst_desc;
+   unsigned bits;
+   enum pipe_format blit_src_format, blit_dst_format;
+
+   /* Get equivalent canonical formats. Those are always array formats and
+    * copying between compatible canonical formats behaves either like
+    * memcpy or like swizzled memcpy. The idea is that we won't have to care
+    * about the channel type from this point on.
+    * Only the swizzle and channel size.
+    */
+   blit_src_format = get_canonical_format(src->format);
+   blit_dst_format = get_canonical_format(dst->format);
+
+   assert(blit_src_format != PIPE_FORMAT_NONE);
+   assert(blit_dst_format != PIPE_FORMAT_NONE);
+
+   src_desc = util_format_description(blit_src_format);
+   dst_desc = util_format_description(blit_dst_format);
+
+   assert(src_desc->block.bits == dst_desc->block.bits);
+   bits = src_desc->block.bits;
+
+   if (dst_desc->channel[0].size == src_desc->channel[0].size) {
+      /* Only the swizzle is different, which means we can just blit,
+       * e.g. RGBA -> BGRA.
+       */
+   } else if (has_identity_swizzle(src_desc)) {
+      /* Src is unswizzled and dst can be swizzled, so src is typecast
+       * to an equivalent dst-compatible format.
+       * e.g. R32 -> BGRA8 is realized as RGBA8 -> BGRA8
+       */
+      blit_src_format =
+         canonical_format_from_bits(bits, dst_desc->channel[0].size);
+   } else if (has_identity_swizzle(dst_desc)) {
+      /* Dst is unswizzled and src can be swizzled, so dst is typecast
+       * to an equivalent src-compatible format.
+       * e.g. BGRA8 -> R32 is realized as BGRA8 -> RGBA8
+       */
+      blit_dst_format =
+         canonical_format_from_bits(bits, src_desc->channel[0].size);
+   } else {
+      assert(!"This should have been handled by handle_complex_copy.");
+      return;
+   }
+
+   blit(pipe, dst, blit_dst_format, dst_level, dstx, dsty, dstz,
+        src, blit_src_format, src_level, src_box);
+}
+
+static bool
+same_size_and_swizzle(const struct util_format_description *d1,
+                      const struct util_format_description *d2)
+{
+   int i;
+
+   if (d1->layout != d2->layout ||
+       d1->nr_channels != d2->nr_channels ||
+       d1->is_array != d2->is_array)
+      return false;
+
+   for (i = 0; i < d1->nr_channels; i++) {
+      if (d1->channel[i].size != d2->channel[i].size)
+         return false;
+
+      if (d1->swizzle[i] <= UTIL_FORMAT_SWIZZLE_W &&
+          d2->swizzle[i] <= UTIL_FORMAT_SWIZZLE_W &&
+          d1->swizzle[i] != d2->swizzle[i])
+         return false;
+   }
+
+   return true;
+}
+
+static struct pipe_resource *
+create_texture(struct pipe_screen *screen, enum pipe_format format,
+               unsigned width, unsigned height, unsigned depth)
+{
+   struct pipe_resource templ;
+
+   memset(&templ, 0, sizeof(templ));
+   templ.format = format;
+   templ.width0 = width;
+   templ.height0 = height;
+   templ.depth0 = 1;
+   templ.array_size = depth;
+   templ.usage = PIPE_USAGE_DEFAULT;
+   templ.bind = PIPE_BIND_SAMPLER_VIEW | PIPE_BIND_RENDER_TARGET;
+
+   if (depth > 1)
+      templ.target = PIPE_TEXTURE_2D_ARRAY;
+   else
+      templ.target = PIPE_TEXTURE_2D;
+
+   return screen->resource_create(screen, &templ);
+}
+
+/**
+ * Handle complex format conversions using 2 blits with a temporary texture
+ * in between, e.g. blitting from B10G10R10A2 to G16R16.
+ *
+ * This example is implemented this way:
+ * 1) First, blit from B10G10R10A2 to R10G10B10A2, which is canonical, so it
+ *    can be reinterpreted as a different canonical format of the same bpp,
+ *    such as R16G16. This blit only swaps R and B 10-bit components.
+ * 2) Finally, blit the result, which is R10G10B10A2, as R16G16 to G16R16.
+ *    This blit only swaps R and G 16-bit components.
+ */
+static bool
+handle_complex_copy(struct pipe_context *pipe,
+                    struct pipe_resource *dst,
+                    unsigned dst_level,
+                    unsigned dstx, unsigned dsty, unsigned dstz,
+                    struct pipe_resource *src,
+                    unsigned src_level,
+                    const struct pipe_box *src_box,
+                    enum pipe_format noncanon_format,
+                    enum pipe_format canon_format)
+{
+   struct pipe_box temp_box;
+   struct pipe_resource *temp = NULL;
+   const struct util_format_description *src_desc, *dst_desc;
+   const struct util_format_description *canon_desc, *noncanon_desc;
+   bool src_is_canon;
+   bool src_is_noncanon;
+   bool dst_is_canon;
+   bool dst_is_noncanon;
+
+   src_desc = util_format_description(src->format);
+   dst_desc = util_format_description(dst->format);
+   canon_desc = util_format_description(canon_format);
+   noncanon_desc = util_format_description(noncanon_format);
+
+   src_is_canon = same_size_and_swizzle(src_desc, canon_desc);
+   dst_is_canon = same_size_and_swizzle(dst_desc, canon_desc);
+   src_is_noncanon = same_size_and_swizzle(src_desc, noncanon_desc);
+   dst_is_noncanon = same_size_and_swizzle(dst_desc, noncanon_desc);
+
+   if (src_is_noncanon) {
+      /* Simple case - only types differ (e.g. UNORM and UINT). */
+      if (dst_is_noncanon) {
+         blit(pipe, dst, noncanon_format, dst_level, dstx, dsty, dstz, src,
+              noncanon_format, src_level, src_box);
+         return true;
+      }
+
+      /* Simple case - only types and swizzles differ. */
+      if (dst_is_canon) {
+         blit(pipe, dst, canon_format, dst_level, dstx, dsty, dstz, src,
+              noncanon_format, src_level, src_box);
+         return true;
+      }
+
+      /* Use the temporary texture. Src is converted to a canonical format,
+       * then proceed the generic swizzled_copy.
+       */
+      temp = create_texture(pipe->screen, canon_format, src_box->width,
+                            src_box->height, src_box->depth);
+
+      u_box_3d(0, 0, 0, src_box->width, src_box->height, src_box->depth,
+               &temp_box);
+
+      blit(pipe, temp, canon_format, 0, 0, 0, 0, src, noncanon_format,
+           src_level, src_box);
+      swizzled_copy(pipe, dst, dst_level, dstx, dsty, dstz, temp, 0,
+                    &temp_box);
+      pipe_resource_reference(&temp, NULL);
+      return true;
+   }
+
+   if (dst_is_noncanon) {
+      /* Simple case - only types and swizzles differ. */
+      if (src_is_canon) {
+         blit(pipe, dst, noncanon_format, dst_level, dstx, dsty, dstz, src,
+              canon_format, src_level, src_box);
+         return true;
+      }
+
+      /* Use the temporary texture. First, use the generic copy, but use
+       * a canonical format in the destination. Then convert */
+      temp = create_texture(pipe->screen, canon_format, src_box->width,
+                            src_box->height, src_box->depth);
+
+      u_box_3d(0, 0, 0, src_box->width, src_box->height, src_box->depth,
+               &temp_box);
+
+      swizzled_copy(pipe, temp, 0, 0, 0, 0, src, src_level, src_box);
+      blit(pipe, dst, noncanon_format, dst_level, dstx, dsty, dstz, temp,
+           canon_format, 0, &temp_box);
+      pipe_resource_reference(&temp, NULL);
+      return true;
+   }
+
+   return false;
+}
+
+static void
+copy_image(struct pipe_context *pipe,
+           struct pipe_resource *dst,
+           unsigned dst_level,
+           unsigned dstx, unsigned dsty, unsigned dstz,
+           struct pipe_resource *src,
+           unsigned src_level,
+           const struct pipe_box *src_box)
+{
+   if (src->format == dst->format ||
+       util_format_is_compressed(src->format) ||
+       util_format_is_compressed(dst->format)) {
+      pipe->resource_copy_region(pipe, dst, dst_level, dstx, dsty, dstz,
+                                 src, src_level, src_box);
+      return;
+   }
+
+   /* Copying to/from B10G10R10*2 needs 2 blits with R10G10B10A2
+    * as a temporary texture in between.
+    */
+   if (handle_complex_copy(pipe, dst, dst_level, dstx, dsty, dstz, src,
+                           src_level, src_box, PIPE_FORMAT_B10G10R10A2_UINT,
+                           PIPE_FORMAT_R10G10B10A2_UINT))
+      return;
+
+   /* Copying to/from G8R8 needs 2 blits with R8G8 as a temporary texture
+    * in between.
+    */
+   if (handle_complex_copy(pipe, dst, dst_level, dstx, dsty, dstz, src,
+                           src_level, src_box, PIPE_FORMAT_G8R8_UNORM,
+                           PIPE_FORMAT_R8G8_UNORM))
+      return;
+
+   /* Copying to/from G16R16 needs 2 blits with R16G16 as a temporary texture
+    * in between.
+    */
+   if (handle_complex_copy(pipe, dst, dst_level, dstx, dsty, dstz, src,
+                           src_level, src_box, PIPE_FORMAT_G16R16_UNORM,
+                           PIPE_FORMAT_R16G16_UNORM))
+      return;
+
+   /* Only allow non-identity swizzling on RGBA8 formats. */
+
+   /* Simple copy, memcpy with swizzling, no format conversion. */
+   swizzled_copy(pipe, dst, dst_level, dstx, dsty, dstz, src, src_level,
+                 src_box);
+}
+
+static void
+st_CopyImageSubData(struct gl_context *ctx,
+                    struct gl_texture_image *src_image,
+                    struct gl_renderbuffer *src_renderbuffer,
+                    int src_x, int src_y, int src_z,
+                    struct gl_texture_image *dst_image,
+                    struct gl_renderbuffer *dst_renderbuffer,
+                    int dst_x, int dst_y, int dst_z,
+                    int src_width, int src_height)
+{
+   struct st_context *st = st_context(ctx);
+   struct pipe_context *pipe = st->pipe;
+   struct pipe_resource *src_res, *dst_res;
+   struct pipe_box box;
+   int src_level, dst_level;
+
+   if (src_image) {
+      struct st_texture_image *src = st_texture_image(src_image);
+      src_res = src->pt;
+      src_level = src_image->Level;
+      src_z += src_image->Face;
+   } else {
+      struct st_renderbuffer *src = st_renderbuffer(src_renderbuffer);
+      src_res = src->texture;
+      src_level = 0;
+   }
+
+   if (dst_image) {
+      struct st_texture_image *dst = st_texture_image(dst_image);
+      dst_res = dst->pt;
+      dst_level = dst_image->Level;
+      dst_z += dst_image->Face;
+   } else {
+      struct st_renderbuffer *dst = st_renderbuffer(dst_renderbuffer);
+      dst_res = dst->texture;
+      dst_level = 0;
+   }
+
+   u_box_2d_zslice(src_x, src_y, src_z, src_width, src_height, &box);
+
+   copy_image(pipe, dst_res, dst_level, dst_x, dst_y, dst_z,
+              src_res, src_level, &box);
+}
+
+void
+st_init_copy_image_functions(struct dd_function_table *functions)
+{
+   functions->CopyImageSubData = st_CopyImageSubData;
+}