util: Use stubs for the dynamically loaded S3TC functions.

Loosely based on Luca Barbieri's commit
52e9b990a192a9329006d5f7dd2ac222effea5a5.

1 files changed, 349 insertions, 334 deletions

diff --git a/src/gallium/auxiliary/util/u_format_s3tc.c b/src/gallium/auxiliary/util/u_format_s3tc.c
index 268f0e8dfab..06cfdcfa8e7 100644
--- a/src/gallium/auxiliary/util/u_format_s3tc.c
+++ b/src/gallium/auxiliary/util/u_format_s3tc.c
@@ -35,58 +35,121 @@
 #endif
 
 
+static void
+util_format_dxt1_rgb_fetch_stub(int src_stride,
+                                const uint8_t *src,
+                                int col, int row,
+                                uint8_t *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 )
+{
+   assert(0);
+}
+
+
+static void
+util_format_dxt3_rgba_fetch_stub(int src_stride,
+                                 const uint8_t *src,
+                                 int col, int row,
+                                 uint8_t *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 )
+{
+   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)
+{
+   assert(0);
+}
+
+
 boolean util_format_s3tc_enabled = FALSE;
 
-util_format_dxtn_fetch_t util_format_dxt1_rgb_fetch = NULL;
-util_format_dxtn_fetch_t util_format_dxt1_rgba_fetch = NULL;
-util_format_dxtn_fetch_t util_format_dxt3_rgba_fetch = NULL;
-util_format_dxtn_fetch_t util_format_dxt5_rgba_fetch = NULL;
+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;
+util_format_dxtn_fetch_t util_format_dxt3_rgba_fetch = util_format_dxt3_rgba_fetch_stub;
+util_format_dxtn_fetch_t util_format_dxt5_rgba_fetch = util_format_dxt5_rgba_fetch_stub;
 
-util_format_dxtn_pack_t util_format_dxtn_pack = NULL;
+util_format_dxtn_pack_t util_format_dxtn_pack = util_format_dxtn_pack_stub;
 
 
 void
 util_format_s3tc_init(void)
 {
-   static struct util_dl_library *
-   library = NULL;
+   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);
    if (!library) {
-      library = util_dl_open(DXTN_LIBNAME);
-      if (!library) {
-	 debug_printf("couldn't open " DXTN_LIBNAME ", software DXTn "
-	    "compression/decompression unavailable");
-      }
-      else {
-         util_format_dxt1_rgb_fetch = (util_format_dxtn_fetch_t)
-            util_dl_get_proc_address(library, "fetch_2d_texel_rgb_dxt1");
-         util_format_dxt1_rgba_fetch = (util_format_dxtn_fetch_t)
-            util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt1");
-         util_format_dxt3_rgba_fetch = (util_format_dxtn_fetch_t)
-            util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt3");
-         util_format_dxt5_rgba_fetch = (util_format_dxtn_fetch_t)
-            util_dl_get_proc_address(library, "fetch_2d_texel_rgba_dxt5");
-         util_format_dxtn_pack = (util_format_dxtn_pack_t)
-            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) {
-            util_format_s3tc_enabled = TRUE;
-            debug_printf("software DXTn compression/decompression available");
-
-         } else {
-	    debug_printf("couldn't reference all symbols in "
-	       DXTN_LIBNAME ", software DXTn compression/decompression "
-	       "unavailable");
-         }
-      }
+      debug_printf("couldn't open " DXTN_LIBNAME ", software DXTn "
+         "compression/decompression unavailable");
+      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");
+      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;
 }
 
 
@@ -97,85 +160,69 @@ util_format_s3tc_init(void)
 void
 util_format_dxt1_rgb_fetch_8unorm(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   if (util_format_dxt1_rgb_fetch) {
-      util_format_dxt1_rgb_fetch(0, src, i, j, dst);
-   }
+   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)
 {
-   if (util_format_dxt1_rgba_fetch) {
-      util_format_dxt1_rgba_fetch(0, src, i, j, dst);
-   }
+   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)
 {
-   if (util_format_dxt3_rgba_fetch) {
-      util_format_dxt3_rgba_fetch(0, src, i, j, dst);
-   }
+   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)
 {
-   if (util_format_dxt5_rgba_fetch) {
-      util_format_dxt5_rgba_fetch(0, src, i, j, dst);
-   }
+   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)
 {
-   if (util_format_dxt1_rgb_fetch) {
-      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;
-   }
+   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;
 }
 
 void
 util_format_dxt1_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   if (util_format_dxt1_rgba_fetch) {
-      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]);
-   }
+   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]);
 }
 
 void
 util_format_dxt3_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   if (util_format_dxt3_rgba_fetch) {
-      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]);
-   }
+   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]);
 }
 
 void
 util_format_dxt5_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, unsigned j)
 {
-   if (util_format_dxt5_rgba_fetch) {
-      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]);
-   }
+   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]);
 }
 
 
@@ -186,188 +233,172 @@ util_format_dxt5_rgba_fetch_float(float *dst, const uint8_t *src, unsigned i, un
 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 (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);
-               }
+   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);
             }
-            src += 8;
          }
-         src_row += src_stride;
+         src += 8;
       }
+      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)
 {
-   if (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);
-               }
+   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;
+         src += 8;
       }
+      src_row += src_stride;
    }
 }
 
 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)
 {
-   if (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);
-               }
+   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;
+         src += 16;
       }
+      src_row += src_stride;
    }
 }
 
 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)
 {
-   if (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);
-               }
+   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;
+         src += 16;
       }
+      src_row += src_stride;
    }
 }
 
 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)
 {
-   if (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;
-               }
+   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;
             }
-            src += 8;
          }
-         src_row += src_stride;
+         src += 8;
       }
+      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)
 {
-   if (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]);
-               }
+   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;
+         src += 8;
       }
+      src_row += src_stride;
    }
 }
 
 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)
 {
-   if (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]);
-               }
+   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;
+         src += 16;
       }
+      src_row += src_stride;
    }
 }
 
 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)
 {
-   if (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]);
-               }
+   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;
+         src += 16;
       }
+      src_row += src_stride;
    }
 }
 
@@ -379,216 +410,200 @@ 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 (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];
-                  }
+   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_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, dst_stride);
+         src += 4*4;
+         dst += 8;
       }
+      src_row += src_stride;
+      dst_row += 4*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)
 {
-   if (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];
-                  }
+   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];
                }
             }
-            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, dst_stride);
+         src += 4*4;
+         dst += 8;
       }
+      src_row += src_stride;
+      dst_row += 4*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)
 {
-   if (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];
-                  }
+   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];
                }
             }
-            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, dst_stride);
+         src += 4*4;
+         dst += 16;
       }
+      src_row += src_stride;
+      dst_row += 4*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)
 {
-   if (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];
-                  }
+   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];
                }
             }
-            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, dst_stride);
+         src += 4*4;
+         dst += 16;
       }
+      src_row += src_stride;
+      dst_row += 4*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)
 {
-   if (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) {
+      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]);
                }
             }
-            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, dst_stride);
+         src += 4*4;
+         dst += 8;
       }
+      src_row += src_stride;
+      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)
 {
-   if (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) {
+      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]);
                }
             }
-            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, dst_stride);
+         src += 4*4;
+         dst += 8;
       }
+      src_row += src_stride;
+      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)
 {
-   if (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) {
+      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]);
                }
             }
-            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, dst_stride);
+         src += 4*4;
+         dst += 16;
       }
+      src_row += src_stride;
+      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)
 {
-   if (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) {
+      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]);
                }
             }
-            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, dst_stride);
+         src += 4*4;
+         dst += 16;
       }
+      src_row += src_stride;
+      dst_row += 4*dst_stride/sizeof(*dst_row);
    }
 }