mesa: add ASTC 2D LDR decoder

Tested-by: Mike Lothian <mike@fireburn.co.uk>
Tested-By: Gert Wollny <gert.wollny@collabora.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>

src/mesa/Makefile.sources

@@ -214,6 +214,8 @@ MAIN_FILES = \
 	main/syncobj.c \
 	main/syncobj.h \
 	main/texcompress.c \
+	main/texcompress_astc.cpp \
+	main/texcompress_astc.h \
 	main/texcompress_bptc.c \
 	main/texcompress_bptc.h \
 	main/texcompress_bptc_tmp.h \

src/mesa/main/formats.c

@@ -627,6 +627,48 @@ _mesa_is_format_etc2(mesa_format format)
 }
 
 
+/**
+ * Return TRUE if format is an ASTC 2D compressed format.
+ */
+bool
+_mesa_is_format_astc_2d(mesa_format format)
+{
+   switch (format) {
+   case MESA_FORMAT_RGBA_ASTC_4x4:
+   case MESA_FORMAT_RGBA_ASTC_5x4:
+   case MESA_FORMAT_RGBA_ASTC_5x5:
+   case MESA_FORMAT_RGBA_ASTC_6x5:
+   case MESA_FORMAT_RGBA_ASTC_6x6:
+   case MESA_FORMAT_RGBA_ASTC_8x5:
+   case MESA_FORMAT_RGBA_ASTC_8x6:
+   case MESA_FORMAT_RGBA_ASTC_8x8:
+   case MESA_FORMAT_RGBA_ASTC_10x5:
+   case MESA_FORMAT_RGBA_ASTC_10x6:
+   case MESA_FORMAT_RGBA_ASTC_10x8:
+   case MESA_FORMAT_RGBA_ASTC_10x10:
+   case MESA_FORMAT_RGBA_ASTC_12x10:
+   case MESA_FORMAT_RGBA_ASTC_12x12:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_4x4:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_5x4:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_5x5:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_6x5:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_6x6:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_8x5:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_8x6:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_8x8:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_10x5:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_10x6:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_10x8:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_10x10:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_12x10:
+   case MESA_FORMAT_SRGB8_ALPHA8_ASTC_12x12:
+      return true;
+   default:
+      return false;
+   }
+}
+
+
 /**
  * If the given format is a compressed format, return a corresponding
  * uncompressed format.

src/mesa/main/formats.h

@@ -721,6 +721,9 @@ _mesa_is_format_integer(mesa_format format);
 extern bool
 _mesa_is_format_etc2(mesa_format format);
 
+bool
+_mesa_is_format_astc_2d(mesa_format format);
+
 GLenum
 _mesa_is_format_color_format(mesa_format format);
 

src/mesa/main/texcompress_astc.h

@@ -0,0 +1,47 @@
+/*
+ * Copyright 2018 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
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * 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 NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS 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.
+ */
+
+#ifndef TEXCOMPRESS_ASTC_H
+#define TEXCOMPRESS_ASTC_H
+
+#include <inttypes.h>
+#include "texcompress.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void
+_mesa_unpack_astc_2d_ldr(uint8_t *dst_row,
+                         unsigned dst_stride,
+                         const uint8_t *src_row,
+                         unsigned src_stride,
+                         unsigned src_width,
+                         unsigned src_height,
+                         mesa_format format);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

src/mesa/meson.build

@@ -254,6 +254,8 @@ files_libmesa_common = files(
   'main/syncobj.c',
   'main/syncobj.h',
   'main/texcompress.c',
+  'main/texcompress_astc.cpp',
+  'main/texcompress_astc.h',
   'main/texcompress_bptc.c',
   'main/texcompress_bptc.h',
   'main/texcompress_cpal.c',

src/util/half_float.c

@@ -2,6 +2,8 @@
  * Mesa 3-D graphics library
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright 2015 Philip Taylor <philip@zaynar.co.uk>
+ * Copyright 2018 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"),
@@ -175,3 +177,60 @@ _mesa_half_to_float(uint16_t val)
    result = fi.f;
    return result;
 }
+
+/**
+  * Convert 0.0 to 0x00, 1.0 to 0xff.
+  * Values outside the range [0.0, 1.0] will give undefined results.
+  */
+uint8_t _mesa_half_to_unorm8(uint16_t val)
+{
+   const int m = val & 0x3ff;
+   const int e = (val >> 10) & 0x1f;
+   const int s = (val >> 15) & 0x1;
+
+   /* v = round_to_nearest(1.mmmmmmmmmm * 2^(e-15) * 255)
+    *   = round_to_nearest((1.mmmmmmmmmm * 255) * 2^(e-15))
+    *   = round_to_nearest((1mmmmmmmmmm * 255) * 2^(e-25))
+    *   = round_to_zero((1mmmmmmmmmm * 255) * 2^(e-25) + 0.5)
+    *   = round_to_zero(((1mmmmmmmmmm * 255) * 2^(e-24) + 1) / 2)
+    *
+    * This happens to give the correct answer for zero/subnormals too
+    */
+   assert(s == 0 && val <= FP16_ONE); /* check 0 <= this <= 1 */
+   /* (implies e <= 15, which means the bit-shifts below are safe) */
+
+   uint32_t v = ((1 << 10) | m) * 255;
+   v = ((v >> (24 - e)) + 1) >> 1;
+   return v;
+}
+
+/**
+  * Takes a uint16_t, divides by 65536, converts the infinite-precision
+  * result to fp16 with round-to-zero. Used by the ASTC decoder.
+  */
+uint16_t _mesa_uint16_div_64k_to_half(uint16_t v)
+{
+   /* Zero or subnormal. Set the mantissa to (v << 8) and return. */
+   if (v < 4)
+      return v << 8;
+
+   /* Count the leading 0s in the uint16_t */
+   int n = __builtin_clz(v) - (sizeof(unsigned int) - sizeof(uint16_t)) * 8;
+
+   /* Shift the mantissa up so bit 16 is the hidden 1 bit,
+    * mask it off, then shift back down to 10 bits
+    */
+   int m = ( ((uint32_t)v << (n + 1)) & 0xffff ) >> 6;
+
+   /*  (0{n} 1 X{15-n}) * 2^-16
+    * = 1.X * 2^(15-n-16)
+    * = 1.X * 2^(14-n - 15)
+    * which is the FP16 form with e = 14 - n
+    */
+   int e = 14 - n;
+
+   assert(e >= 1 && e <= 30);
+   assert(m >= 0 && m < 0x400);
+
+   return (e << 10) | m;
+}

src/util/half_float.h

@@ -32,8 +32,13 @@
 extern "C" {
 #endif
 
+#define FP16_ONE     0x3C00
+#define FP16_ZERO    0
+
 uint16_t _mesa_float_to_half(float val);
 float _mesa_half_to_float(uint16_t val);
+uint8_t _mesa_half_to_unorm8(uint16_t v);
+uint16_t _mesa_uint16_div_64k_to_half(uint16_t v);
 
 static inline bool
 _mesa_half_is_negative(uint16_t h)