r600g: fix linear and flat interpolation
[mesa.git] / src / gallium / auxiliary / util / u_format_s3tc.c
index 8a5e6f0c43744452ed52d4bbfa2fa45c55dae2ea..d8a7c0d453fdc37790873bf534bcd926ea2fdeb2 100644 (file)
 
 #if defined(_WIN32) || defined(WIN32)
 #define DXTN_LIBNAME "dxtn.dll"
+#elif defined(__APPLE__)
+#define DXTN_LIBNAME "libtxc_dxtn.dylib"
 #else
 #define DXTN_LIBNAME "libtxc_dxtn.so"
 #endif
 
+
 static void
-util_format_dxt1_rgb_fetch_stub( int src_stride,
-                             const uint8_t *src,
-                             int col, int row,
-                             uint8_t *dst )
+util_format_dxt1_rgb_fetch_stub(int src_stride,
+                                const uint8_t *src,
+                                int col, int row,
+                                uint8_t *dst)
 {
-   util_format_s3tc_init();
-   util_format_dxt1_rgb_fetch(src_stride, src, col, row, dst);
+   assert(0);
 }
 
+
 static void
-util_format_dxt1_rgba_fetch_stub( int src_stride,
-                             const uint8_t *src,
-                             int col, int row,
-                             uint8_t *dst )
+util_format_dxt1_rgba_fetch_stub(int src_stride,
+                                 const uint8_t *src,
+                                 int col, int row,
+                                 uint8_t *dst )
 {
-   util_format_s3tc_init();
-   util_format_dxt1_rgba_fetch(src_stride, src, col, row, dst);
+   assert(0);
 }
 
+
 static void
-util_format_dxt3_rgba_fetch_stub( int src_stride,
-                             const uint8_t *src,
-                             int col, int row,
-                             uint8_t *dst )
+util_format_dxt3_rgba_fetch_stub(int src_stride,
+                                 const uint8_t *src,
+                                 int col, int row,
+                                 uint8_t *dst )
 {
-   util_format_s3tc_init();
-   util_format_dxt3_rgba_fetch(src_stride, src, col, row, dst);
+   assert(0);
 }
 
+
 static void
-util_format_dxt5_rgba_fetch_stub( int src_stride,
-                             const uint8_t *src,
-                             int col, int row,
-                             uint8_t *dst )
+util_format_dxt5_rgba_fetch_stub(int src_stride,
+                                 const uint8_t *src,
+                                 int col, int row,
+                                 uint8_t *dst )
 {
-   util_format_s3tc_init();
-   util_format_dxt5_rgba_fetch(src_stride, src, col, row, dst);
+   assert(0);
 }
 
-static
-void util_format_dxtn_pack_stub( int src_comps,
-                            int width, int height,
-                            const uint8_t *src,
-                            enum util_format_dxtn dst_format,
-                            uint8_t *dst,
-                            int dst_stride)
+
+static void
+util_format_dxtn_pack_stub(int src_comps,
+                           int width, int height,
+                           const uint8_t *src,
+                           enum util_format_dxtn dst_format,
+                           uint8_t *dst,
+                           int dst_stride)
 {
-   util_format_s3tc_init();
-   util_format_dxtn_pack_stub(src_comps, width, height, src, dst_format, dst, dst_stride);
+   assert(0);
 }
 
+
 boolean util_format_s3tc_enabled = FALSE;
-boolean util_format_s3tc_inited = FALSE;
 
 util_format_dxtn_fetch_t util_format_dxt1_rgb_fetch = util_format_dxt1_rgb_fetch_stub;
 util_format_dxtn_fetch_t util_format_dxt1_rgba_fetch = util_format_dxt1_rgba_fetch_stub;
@@ -96,57 +98,68 @@ util_format_dxtn_fetch_t util_format_dxt5_rgba_fetch = util_format_dxt5_rgba_fet
 
 util_format_dxtn_pack_t util_format_dxtn_pack = util_format_dxtn_pack_stub;
 
-static void
-nop(void)
-{}
-
-#define is_nop(f) ((void*)(f) == (void*)nop)
-
-static util_dl_proc
-get_proc_address_or_nop(struct util_dl_library *library,
-                         const char *procname)
-{
-   if(library) {
-      util_dl_proc proc = util_dl_get_proc_address(library, procname);
-      if(proc)
-         return proc;
-   }
-   return (util_dl_proc)nop;
-}
 
 void
-util_format_s3tc_do_init(void)
+util_format_s3tc_init(void)
 {
-   struct util_dl_library *library;
+   static boolean first_time = TRUE;
+   struct util_dl_library *library = NULL;
+   util_dl_proc fetch_2d_texel_rgb_dxt1;
+   util_dl_proc fetch_2d_texel_rgba_dxt1;
+   util_dl_proc fetch_2d_texel_rgba_dxt3;
+   util_dl_proc fetch_2d_texel_rgba_dxt5;
+   util_dl_proc tx_compress_dxtn;
+
+   if (!first_time)
+      return;
+   first_time = FALSE;
+
+   if (util_format_s3tc_enabled)
+      return;
 
    library = util_dl_open(DXTN_LIBNAME);
-   util_format_dxt1_rgb_fetch = (util_format_dxtn_fetch_t)
-      get_proc_address_or_nop(library, "fetch_2d_texel_rgb_dxt1");
-   util_format_dxt1_rgba_fetch = (util_format_dxtn_fetch_t)
-      get_proc_address_or_nop(library, "fetch_2d_texel_rgba_dxt1");
-   util_format_dxt3_rgba_fetch = (util_format_dxtn_fetch_t)
-      get_proc_address_or_nop(library, "fetch_2d_texel_rgba_dxt3");
-   util_format_dxt5_rgba_fetch = (util_format_dxtn_fetch_t)
-      get_proc_address_or_nop(library, "fetch_2d_texel_rgba_dxt5");
-   util_format_dxtn_pack = (util_format_dxtn_pack_t)
-      get_proc_address_or_nop(library, "tx_compress_dxtn");
-
-   if (!library)
-      debug_printf("couldn't open " DXTN_LIBNAME ", software DXTn "
-         "compression/decompression unavailable");
-   else {
-      if (!is_nop(util_format_dxt1_rgb_fetch) &&
-         !is_nop(util_format_dxt1_rgba_fetch) &&
-         !is_nop(util_format_dxt3_rgba_fetch) &&
-         !is_nop(util_format_dxt5_rgba_fetch) &&
-         !is_nop(util_format_dxtn_pack)) {
-         debug_printf("software DXTn compression/decompression available");
+   if (!library) {
+      if (getenv("force_s3tc_enable") &&
+          !strcmp(getenv("force_s3tc_enable"), "true")) {
+         debug_printf("couldn't open " DXTN_LIBNAME ", enabling DXTn due to "
+            "force_s3tc_enable=true environment variable\n");
          util_format_s3tc_enabled = TRUE;
-      } else
-         debug_printf("couldn't reference all symbols in "
-            DXTN_LIBNAME ", software DXTn compression/decompression "
-            "unavailable");
+      } else {
+         debug_printf("couldn't open " DXTN_LIBNAME ", software DXTn "
+            "compression/decompression unavailable\n");
+      }
+      return;
    }
+
+   fetch_2d_texel_rgb_dxt1 =
+         util_dl_get_proc_address(library, "fetch_2d_texel_rgb_dxt1");
+   fetch_2d_texel_rgba_dxt1 =
+         util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt1");
+   fetch_2d_texel_rgba_dxt3 =
+         util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt3");
+   fetch_2d_texel_rgba_dxt5 =
+         util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt5");
+   tx_compress_dxtn =
+         util_dl_get_proc_address(library, "tx_compress_dxtn");
+
+   if (!util_format_dxt1_rgb_fetch ||
+       !util_format_dxt1_rgba_fetch ||
+       !util_format_dxt3_rgba_fetch ||
+       !util_format_dxt5_rgba_fetch ||
+       !util_format_dxtn_pack) {
+      debug_printf("couldn't reference all symbols in " DXTN_LIBNAME
+                   ", software DXTn compression/decompression "
+                   "unavailable\n");
+      util_dl_close(library);
+      return;
+   }
+
+   util_format_dxt1_rgb_fetch = (util_format_dxtn_fetch_t)fetch_2d_texel_rgb_dxt1;
+   util_format_dxt1_rgba_fetch = (util_format_dxtn_fetch_t)fetch_2d_texel_rgba_dxt1;
+   util_format_dxt3_rgba_fetch = (util_format_dxtn_fetch_t)fetch_2d_texel_rgba_dxt3;
+   util_format_dxt5_rgba_fetch = (util_format_dxtn_fetch_t)fetch_2d_texel_rgba_dxt5;
+   util_format_dxtn_pack = (util_format_dxtn_pack_t)tx_compress_dxtn;
+   util_format_s3tc_enabled = TRUE;
 }
 
 
@@ -155,31 +168,31 @@ util_format_s3tc_do_init(void)
  */
 
 void
-util_format_dxt1_rgb_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt1_rgb_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
    util_format_dxt1_rgb_fetch(0, src, i, j, dst);
 }
 
 void
-util_format_dxt1_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt1_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
    util_format_dxt1_rgba_fetch(0, src, i, j, dst);
 }
 
 void
-util_format_dxt3_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt3_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
    util_format_dxt3_rgba_fetch(0, src, i, j, dst);
 }
 
 void
-util_format_dxt5_rgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt5_rgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
    util_format_dxt5_rgba_fetch(0, src, i, j, dst);
 }
 
 void
-util_format_dxt1_rgb_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt1_rgb_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
    uint8_t tmp[4];
    util_format_dxt1_rgb_fetch(0, src, i, j, tmp);
@@ -190,7 +203,7 @@ util_format_dxt1_rgb_fetch_float(float *dst, const uint8_t *src, unsigned i, uns
 }
 
 void
-util_format_dxt1_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt1_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
    uint8_t tmp[4];
    util_format_dxt1_rgba_fetch(0, src, i, j, tmp);
@@ -201,7 +214,7 @@ util_format_dxt1_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, un
 }
 
 void
-util_format_dxt3_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt3_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
    uint8_t tmp[4];
    util_format_dxt3_rgba_fetch(0, src, i, j, tmp);
@@ -212,7 +225,7 @@ util_format_dxt3_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, un
 }
 
 void
-util_format_dxt5_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt5_rgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
    uint8_t tmp[4];
    util_format_dxt5_rgba_fetch(0, src, i, j, tmp);
@@ -227,192 +240,144 @@ util_format_dxt5_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, un
  * Block decompression.
  */
 
-void
-util_format_dxt1_rgb_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
-{
-   if (!is_nop(util_format_dxt1_rgb_fetch)) {
-      unsigned x, y, i, j;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         for(x = 0; x < width; x += 4) {
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  uint8_t *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
-                  util_format_dxt1_rgb_fetch(0, src, i, j, dst);
-               }
+static INLINE void
+util_format_dxtn_rgb_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height,
+                                        util_format_dxtn_fetch_t fetch,
+                                        unsigned block_size)
+{
+   const unsigned bw = 4, bh = 4, comps = 4;
+   unsigned x, y, i, j;
+   for(y = 0; y < height; y += bh) {
+      const uint8_t *src = src_row;
+      for(x = 0; x < width; x += bw) {
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               uint8_t *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*comps;
+               fetch(0, src, i, j, dst);
             }
-            src += 8;
          }
-         src_row += src_stride;
+         src += block_size;
       }
+      src_row += src_stride;
    }
 }
 
 void
-util_format_dxt1_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_rgb_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxt1_rgba_fetch)) {
-      unsigned x, y, i, j;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         for(x = 0; x < width; x += 4) {
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  uint8_t *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
-                  util_format_dxt1_rgba_fetch(0, src, i, j, dst);
-               }
-            }
-            src += 8;
-         }
-         src_row += src_stride;
-      }
-   }
+   util_format_dxtn_rgb_unpack_rgba_8unorm(dst_row, dst_stride,
+                                           src_row, src_stride,
+                                           width, height,
+                                           util_format_dxt1_rgb_fetch, 8);
 }
 
 void
-util_format_dxt3_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxt3_rgba_fetch)) {
-      unsigned x, y, i, j;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         for(x = 0; x < width; x += 4) {
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  uint8_t *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
-                  util_format_dxt3_rgba_fetch(0, src, i, j, dst);
-               }
-            }
-            src += 16;
-         }
-         src_row += src_stride;
-      }
-   }
+   util_format_dxtn_rgb_unpack_rgba_8unorm(dst_row, dst_stride,
+                                           src_row, src_stride,
+                                           width, height,
+                                           util_format_dxt1_rgba_fetch, 8);
 }
 
 void
-util_format_dxt5_rgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt3_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
 {
-   if (is_nop(util_format_dxt5_rgba_fetch)) {
-      unsigned x, y, i, j;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         for(x = 0; x < width; x += 4) {
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  uint8_t *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
-                  util_format_dxt5_rgba_fetch(0, src, i, j, dst);
-               }
-            }
-            src += 16;
-         }
-         src_row += src_stride;
-      }
-   }
+   util_format_dxtn_rgb_unpack_rgba_8unorm(dst_row, dst_stride,
+                                           src_row, src_stride,
+                                           width, height,
+                                           util_format_dxt3_rgba_fetch, 16);
 }
 
 void
-util_format_dxt1_rgb_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt5_rgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                         const uint8_t *src_row, unsigned src_stride,
+                                         unsigned width, unsigned height)
 {
-   if (is_nop(util_format_dxt1_rgb_fetch)) {
-      unsigned x, y, i, j;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         for(x = 0; x < width; x += 4) {
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
-                  uint8_t tmp[4];
-                  util_format_dxt1_rgb_fetch(0, src, i, j, tmp);
-                  dst[0] = ubyte_to_float(tmp[0]);
-                  dst[1] = ubyte_to_float(tmp[1]);
-                  dst[2] = ubyte_to_float(tmp[2]);
-                  dst[3] = 1.0;
-               }
+   util_format_dxtn_rgb_unpack_rgba_8unorm(dst_row, dst_stride,
+                                           src_row, src_stride,
+                                           width, height,
+                                           util_format_dxt5_rgba_fetch, 16);
+}
+
+static INLINE void
+util_format_dxtn_rgb_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height,
+                                       util_format_dxtn_fetch_t fetch,
+                                       unsigned block_size)
+{
+   unsigned x, y, i, j;
+   for(y = 0; y < height; y += 4) {
+      const uint8_t *src = src_row;
+      for(x = 0; x < width; x += 4) {
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
+               uint8_t tmp[4];
+               fetch(0, src, i, j, tmp);
+               dst[0] = ubyte_to_float(tmp[0]);
+               dst[1] = ubyte_to_float(tmp[1]);
+               dst[2] = ubyte_to_float(tmp[2]);
+               dst[3] = ubyte_to_float(tmp[3]);
             }
-            src += 8;
          }
-         src_row += src_stride;
+         src += block_size;
       }
+      src_row += src_stride;
    }
 }
 
 void
-util_format_dxt1_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_rgb_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                       const uint8_t *src_row, unsigned src_stride,
+                                       unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxt1_rgba_fetch)) {
-      unsigned x, y, i, j;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         for(x = 0; x < width; x += 4) {
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
-                  uint8_t tmp[4];
-                  util_format_dxt1_rgba_fetch(0, src, i, j, tmp);
-                  dst[0] = ubyte_to_float(tmp[0]);
-                  dst[1] = ubyte_to_float(tmp[1]);
-                  dst[2] = ubyte_to_float(tmp[2]);
-                  dst[3] = ubyte_to_float(tmp[3]);
-               }
-            }
-            src += 8;
-         }
-         src_row += src_stride;
-      }
-   }
+   util_format_dxtn_rgb_unpack_rgba_float(dst_row, dst_stride,
+                                          src_row, src_stride,
+                                          width, height,
+                                          util_format_dxt1_rgb_fetch, 8);
 }
 
 void
-util_format_dxt3_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxt3_rgba_fetch)) {
-      unsigned x, y, i, j;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         for(x = 0; x < width; x += 4) {
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
-                  uint8_t tmp[4];
-                  util_format_dxt3_rgba_fetch(0, src, i, j, tmp);
-                  dst[0] = ubyte_to_float(tmp[0]);
-                  dst[1] = ubyte_to_float(tmp[1]);
-                  dst[2] = ubyte_to_float(tmp[2]);
-                  dst[3] = ubyte_to_float(tmp[3]);
-               }
-            }
-            src += 16;
-         }
-         src_row += src_stride;
-      }
-   }
+   util_format_dxtn_rgb_unpack_rgba_float(dst_row, dst_stride,
+                                          src_row, src_stride,
+                                          width, height,
+                                          util_format_dxt1_rgba_fetch, 8);
 }
 
 void
-util_format_dxt5_rgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt3_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxt5_rgba_fetch)) {
-      unsigned x, y, i, j;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         for(x = 0; x < width; x += 4) {
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  float *dst = dst_row + (y + j)*dst_stride/sizeof(*dst_row) + (x + i)*4;
-                  uint8_t tmp[4];
-                  util_format_dxt5_rgba_fetch(0, src, i, j, tmp);
-                  dst[0] = ubyte_to_float(tmp[0]);
-                  dst[1] = ubyte_to_float(tmp[1]);
-                  dst[2] = ubyte_to_float(tmp[2]);
-                  dst[3] = ubyte_to_float(tmp[3]);
-               }
-            }
-            src += 16;
-         }
-         src_row += src_stride;
-      }
-   }
+   util_format_dxtn_rgb_unpack_rgba_float(dst_row, dst_stride,
+                                          src_row, src_stride,
+                                          width, height,
+                                          util_format_dxt3_rgba_fetch, 16);
+}
+
+void
+util_format_dxt5_rgba_unpack_rgba_float(float *dst_row, unsigned dst_stride,
+                                        const uint8_t *src_row, unsigned src_stride,
+                                        unsigned width, unsigned height)
+{
+   util_format_dxtn_rgb_unpack_rgba_float(dst_row, dst_stride,
+                                          src_row, src_stride,
+                                          width, height,
+                                          util_format_dxt5_rgba_fetch, 16);
 }
 
 
@@ -421,218 +386,199 @@ util_format_dxt5_rgba_unpack_float(float *dst_row, unsigned dst_stride, const ui
  */
 
 void
-util_format_dxt1_rgb_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
-{
-   if (!is_nop(util_format_dxtn_pack)) {
-      unsigned x, y, i, j, k;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         uint8_t *dst = dst_row;
-         for(x = 0; x < width; x += 4) {
-            uint8_t tmp[4][4][3];
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  for(k = 0; k < 3; ++k) {
-                     tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + i*4 + k];
-                  }
+util_format_dxt1_rgb_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                      const uint8_t *src, unsigned src_stride,
+                                      unsigned width, unsigned height)
+{
+   const unsigned bw = 4, bh = 4, bytes_per_block = 8;
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += bh) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += bw) {
+         uint8_t tmp[4][4][3];  /* [bh][bw][comps] */
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               for(k = 0; k < 3; ++k) {
+                  tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + (x + i)*4 + k];
                }
             }
-            util_format_dxtn_pack(3, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGB, dst, dst_stride);
-            src += 4*4;
-            dst += 8;
          }
-         src_row += src_stride;
-         dst_row += 4*dst_stride/sizeof(*dst_row);
+         util_format_dxtn_pack(3, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGB, dst, 0);
+         dst += bytes_per_block;
       }
+      dst_row += dst_stride / sizeof(*dst_row);
    }
 }
 
 void
-util_format_dxt1_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src, unsigned src_stride,
+                                       unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxtn_pack)) {
-      unsigned x, y, i, j, k;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         uint8_t *dst = dst_row;
-         for(x = 0; x < width; x += 4) {
-            uint8_t tmp[4][4][4];
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  for(k = 0; k < 4; ++k) {
-                     tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + i*4 + k];
-                  }
+   const unsigned bw = 4, bh = 4, comps = 4, bytes_per_block = 8;
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += bh) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += bw) {
+         uint8_t tmp[4][4][4];  /* [bh][bw][comps] */
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               for(k = 0; k < comps; ++k) {
+                  tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + (x + i)*comps + k];
                }
             }
-            util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGBA, dst, dst_stride);
-            src += 4*4;
-            dst += 8;
          }
-         src_row += src_stride;
-         dst_row += 4*dst_stride/sizeof(*dst_row);
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGBA, dst, 0);
+         dst += bytes_per_block;
       }
+      dst_row += dst_stride / sizeof(*dst_row);
    }
 }
 
 void
-util_format_dxt3_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt3_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src, unsigned src_stride,
+                                       unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxtn_pack)) {
-      unsigned x, y, i, j, k;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         uint8_t *dst = dst_row;
-         for(x = 0; x < width; x += 4) {
-            uint8_t tmp[4][4][4];
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  for(k = 0; k < 4; ++k) {
-                     tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + i*4 + k];
-                  }
+   const unsigned bw = 4, bh = 4, comps = 4, bytes_per_block = 16;
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += bh) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += bw) {
+         uint8_t tmp[4][4][4];  /* [bh][bw][comps] */
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               for(k = 0; k < comps; ++k) {
+                  tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + (x + i)*comps + k];
                }
             }
-            util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT3_RGBA, dst, dst_stride);
-            src += 4*4;
-            dst += 16;
          }
-         src_row += src_stride;
-         dst_row += 4*dst_stride/sizeof(*dst_row);
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT3_RGBA, dst, 0);
+         dst += bytes_per_block;
       }
+      dst_row += dst_stride / sizeof(*dst_row);
    }
 }
 
 void
-util_format_dxt5_rgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt5_rgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride,
+                                       const uint8_t *src, unsigned src_stride,
+                                       unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxtn_pack)) {
-      unsigned x, y, i, j, k;
-      for(y = 0; y < height; y += 4) {
-         const uint8_t *src = src_row;
-         uint8_t *dst = dst_row;
-         for(x = 0; x < width; x += 4) {
-            uint8_t tmp[4][4][4];
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  for(k = 0; k < 4; ++k) {
-                     tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + i*4 + k];
-                  }
+   const unsigned bw = 4, bh = 4, comps = 4, bytes_per_block = 16;
+   unsigned x, y, i, j, k;
+
+   for(y = 0; y < height; y += bh) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += bw) {
+         uint8_t tmp[4][4][4];  /* [bh][bw][comps] */
+         for(j = 0; j < bh; ++j) {
+            for(i = 0; i < bw; ++i) {
+               for(k = 0; k < comps; ++k) {
+                  tmp[j][i][k] = src[(y + j)*src_stride/sizeof(*src) + (x + i)*comps + k];
                }
             }
-            util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT5_RGBA, dst, dst_stride);
-            src += 4*4;
-            dst += 16;
          }
-         src_row += src_stride;
-         dst_row += 4*dst_stride/sizeof(*dst_row);
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT5_RGBA, dst, 0);
+         dst += bytes_per_block;
       }
+      dst_row += dst_stride / sizeof(*dst_row);
    }
 }
 
 void
-util_format_dxt1_rgb_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_rgb_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                     const float *src, unsigned src_stride,
+                                     unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxtn_pack)) {
-      unsigned x, y, i, j, k;
-      for(y = 0; y < height; y += 4) {
-         const float *src = src_row;
-         uint8_t *dst = dst_row;
-         for(x = 0; x < width; x += 4) {
-            uint8_t tmp[4][4][3];
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  for(k = 0; k < 3; ++k) {
-                     tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + i*4 + k]);
-                  }
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += 4) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += 4) {
+         uint8_t tmp[4][4][3];
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               for(k = 0; k < 3; ++k) {
+                  tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + (x+i)*4 + k]);
                }
             }
-            util_format_dxtn_pack(3, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGB, dst, dst_stride);
-            src += 4*4;
-            dst += 8;
          }
-         src_row += src_stride;
-         dst_row += 4*dst_stride/sizeof(*dst_row);
+         util_format_dxtn_pack(3, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGB, dst, 0);
+         dst += 8;
       }
+      dst_row += 4*dst_stride/sizeof(*dst_row);
    }
 }
 
 void
-util_format_dxt1_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                      const float *src, unsigned src_stride,
+                                      unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxtn_pack)) {
-      unsigned x, y, i, j, k;
-      for(y = 0; y < height; y += 4) {
-         const float *src = src_row;
-         uint8_t *dst = dst_row;
-         for(x = 0; x < width; x += 4) {
-            uint8_t tmp[4][4][4];
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  for(k = 0; k < 4; ++k) {
-                     tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + i*4 + k]);
-                  }
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += 4) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += 4) {
+         uint8_t tmp[4][4][4];
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               for(k = 0; k < 4; ++k) {
+                  tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + (x+i)*4 + k]);
                }
             }
-            util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGBA, dst, dst_stride);
-            src += 4*4;
-            dst += 8;
          }
-         src_row += src_stride;
-         dst_row += 4*dst_stride/sizeof(*dst_row);
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT1_RGBA, dst, 0);
+         dst += 8;
       }
+      dst_row += 4*dst_stride/sizeof(*dst_row);
    }
 }
 
 void
-util_format_dxt3_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt3_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                      const float *src, unsigned src_stride,
+                                      unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxtn_pack)) {
-      unsigned x, y, i, j, k;
-      for(y = 0; y < height; y += 4) {
-         const float *src = src_row;
-         uint8_t *dst = dst_row;
-         for(x = 0; x < width; x += 4) {
-            uint8_t tmp[4][4][4];
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  for(k = 0; k < 4; ++k) {
-                     tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + i*4 + k]);
-                  }
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += 4) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += 4) {
+         uint8_t tmp[4][4][4];
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               for(k = 0; k < 4; ++k) {
+                  tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + (x+i)*4 + k]);
                }
             }
-            util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT3_RGBA, dst, dst_stride);
-            src += 4*4;
-            dst += 16;
          }
-         src_row += src_stride;
-         dst_row += 4*dst_stride/sizeof(*dst_row);
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT3_RGBA, dst, 0);
+         dst += 16;
       }
+      dst_row += 4*dst_stride/sizeof(*dst_row);
    }
 }
 
 void
-util_format_dxt5_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt5_rgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride,
+                                      const float *src, unsigned src_stride,
+                                      unsigned width, unsigned height)
 {
-   if (!is_nop(util_format_dxtn_pack)) {
-      unsigned x, y, i, j, k;
-      for(y = 0; y < height; y += 4) {
-         const float *src = src_row;
-         uint8_t *dst = dst_row;
-         for(x = 0; x < width; x += 4) {
-            uint8_t tmp[4][4][4];
-            for(j = 0; j < 4; ++j) {
-               for(i = 0; i < 4; ++i) {
-                  for(k = 0; k < 4; ++k) {
-                     tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + i*4 + k]);
-                  }
+   unsigned x, y, i, j, k;
+   for(y = 0; y < height; y += 4) {
+      uint8_t *dst = dst_row;
+      for(x = 0; x < width; x += 4) {
+         uint8_t tmp[4][4][4];
+         for(j = 0; j < 4; ++j) {
+            for(i = 0; i < 4; ++i) {
+               for(k = 0; k < 4; ++k) {
+                  tmp[j][i][k] = float_to_ubyte(src[(y + j)*src_stride/sizeof(*src) + (x+i)*4 + k]);
                }
             }
-            util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT5_RGBA, dst, dst_stride);
-            src += 4*4;
-            dst += 16;
          }
-         src_row += src_stride;
-         dst_row += 4*dst_stride/sizeof(*dst_row);
+         util_format_dxtn_pack(4, 4, 4, &tmp[0][0][0], UTIL_FORMAT_DXT5_RGBA, dst, 0);
+         dst += 16;
       }
+      dst_row += 4*dst_stride/sizeof(*dst_row);
    }
 }
 
@@ -644,146 +590,146 @@ util_format_dxt5_rgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const fl
  */
 
 void
-util_format_dxt1_srgb_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_srgb_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt1_rgb_unpack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt1_rgb_unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt1_srgb_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_srgb_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt1_rgb_pack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt1_rgb_pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt1_srgb_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt1_srgb_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   util_format_dxt1_rgb_fetch_8unorm(dst, src, i, j);
+   util_format_dxt1_rgb_fetch_rgba_8unorm(dst, src, i, j);
 }
 
 void
-util_format_dxt1_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt1_rgba_unpack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt1_rgba_unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt1_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt1_rgba_pack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt1_rgba_pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt1_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt1_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   util_format_dxt1_rgba_fetch_8unorm(dst, src, i, j);
+   util_format_dxt1_rgba_fetch_rgba_8unorm(dst, src, i, j);
 }
 
 void
-util_format_dxt3_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt3_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt3_rgba_unpack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt3_rgba_unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt3_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt3_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt3_rgba_pack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt3_rgba_pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt3_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt3_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   util_format_dxt3_rgba_fetch_8unorm(dst, src, i, j);
+   util_format_dxt3_rgba_fetch_rgba_8unorm(dst, src, i, j);
 }
 
 void
-util_format_dxt5_srgba_unpack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt5_srgba_unpack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt5_rgba_unpack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt5_rgba_unpack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt5_srgba_pack_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt5_srgba_pack_rgba_8unorm(uint8_t *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt5_rgba_pack_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt5_rgba_pack_rgba_8unorm(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt5_srgba_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt5_srgba_fetch_rgba_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   util_format_dxt5_rgba_fetch_8unorm(dst, src, i, j);
+   util_format_dxt5_rgba_fetch_rgba_8unorm(dst, src, i, j);
 }
 
 void
-util_format_dxt1_srgb_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_srgb_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt1_rgb_unpack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt1_rgb_unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt1_srgb_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_srgb_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt1_rgb_pack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt1_rgb_pack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt1_srgb_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt1_srgb_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   util_format_dxt1_rgb_fetch_float(dst, src, i, j);
+   util_format_dxt1_rgb_fetch_rgba_float(dst, src, i, j);
 }
 
 void
-util_format_dxt1_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt1_rgba_unpack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt1_rgba_unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt1_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt1_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt1_rgba_pack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt1_rgba_pack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt1_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt1_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   util_format_dxt1_rgba_fetch_float(dst, src, i, j);
+   util_format_dxt1_rgba_fetch_rgba_float(dst, src, i, j);
 }
 
 void
-util_format_dxt3_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt3_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt3_rgba_unpack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt3_rgba_unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt3_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt3_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt3_rgba_pack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt3_rgba_pack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt3_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt3_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   util_format_dxt3_rgba_fetch_float(dst, src, i, j);
+   util_format_dxt3_rgba_fetch_rgba_float(dst, src, i, j);
 }
 
 void
-util_format_dxt5_srgba_unpack_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt5_srgba_unpack_rgba_float(float *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt5_rgba_unpack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt5_rgba_unpack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt5_srgba_pack_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
+util_format_dxt5_srgba_pack_rgba_float(uint8_t *dst_row, unsigned dst_stride, const float *src_row, unsigned src_stride, unsigned width, unsigned height)
 {
-   util_format_dxt5_rgba_pack_float(dst_row, dst_stride, src_row, src_stride, width, height);
+   util_format_dxt5_rgba_pack_rgba_float(dst_row, dst_stride, src_row, src_stride, width, height);
 }
 
 void
-util_format_dxt5_srgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
+util_format_dxt5_srgba_fetch_rgba_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   util_format_dxt5_rgba_fetch_float(dst, src, i, j);
+   util_format_dxt5_rgba_fetch_rgba_float(dst, src, i, j);
 }