gallium/util: added support for SRGB formats

Fixes glean/texture_srgb failure, bug #23449.

src/gallium/auxiliary/util/u_tile.c

@@ -596,6 +596,184 @@ r16g16b16a16_put_tile_rgba(short *dst,
 }
 
 
+/*** PIPE_FORMAT_R8G8B8A8_SRGB ***/
+
+/**
+ * Convert an 8-bit sRGB value from non-linear space to a
+ * linear RGB value in [0, 1].
+ * Implemented with a 256-entry lookup table.
+ */
+static INLINE float
+srgb_to_linear(ubyte cs8)
+{
+   static float table[256];
+   static boolean tableReady = FALSE;
+   if (!tableReady) {
+      /* compute lookup table now */
+      uint i;
+      for (i = 0; i < 256; i++) {
+         const float cs = ubyte_to_float(i);
+         if (cs <= 0.04045) {
+            table[i] = cs / 12.92f;
+         }
+         else {
+            table[i] = (float) powf((cs + 0.055) / 1.055, 2.4);
+         }
+      }
+      tableReady = TRUE;
+   }
+   return table[cs8];
+}
+
+
+/**
+ * Convert linear float in [0,1] to an srgb ubyte value in [0,255].
+ * XXX this hasn't been tested (render to srgb surface).
+ * XXX this needs optimization.
+ */
+static INLINE ubyte
+linear_to_srgb(float cl)
+{
+   if (cl >= 1.0F)
+      return 255;
+   else if (cl >= 0.0031308F)
+      return float_to_ubyte(1.055F * powf(cl, 0.41666F) - 0.055F);
+   else if (cl > 0.0F)
+      return float_to_ubyte(12.92F * cl);
+   else
+      return 0.0;
+}
+
+
+static void
+a8r8g8b8_srgb_get_tile_rgba(const unsigned *src,
+                            unsigned w, unsigned h,
+                            float *p,
+                            unsigned dst_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         const unsigned pixel = *src++;
+         pRow[0] = srgb_to_linear((pixel >> 16) & 0xff);
+         pRow[1] = srgb_to_linear((pixel >>  8) & 0xff);
+         pRow[2] = srgb_to_linear((pixel >>  0) & 0xff);
+         pRow[3] = ubyte_to_float((pixel >> 24) & 0xff);
+      }
+      p += dst_stride;
+   }
+}
+
+static void
+a8r8g8b8_srgb_put_tile_rgba(unsigned *dst,
+                            unsigned w, unsigned h,
+                            const float *p,
+                            unsigned src_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      const float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         unsigned r, g, b, a;
+         r = linear_to_srgb(pRow[0]);
+         g = linear_to_srgb(pRow[1]);
+         b = linear_to_srgb(pRow[2]);
+         a = float_to_ubyte(pRow[3]);
+         *dst++ = (a << 24) | (r << 16) | (g << 8) | b;
+      }
+      p += src_stride;
+   }
+}
+
+
+/*** PIPE_FORMAT_A8L8_SRGB ***/
+
+static void
+a8l8_srgb_get_tile_rgba(const ushort *src,
+                        unsigned w, unsigned h,
+                        float *p,
+                        unsigned dst_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         ushort p = *src++;
+         pRow[0] =
+         pRow[1] =
+         pRow[2] = srgb_to_linear(p & 0xff);
+         pRow[3] = ubyte_to_float(p >> 8);
+      }
+      p += dst_stride;
+   }
+}
+
+static void
+a8l8_srgb_put_tile_rgba(ushort *dst,
+                        unsigned w, unsigned h,
+                        const float *p,
+                        unsigned src_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      const float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         unsigned r, a;
+         r = linear_to_srgb(pRow[0]);
+         a = float_to_ubyte(pRow[3]);
+         *dst++ = (a << 8) | r;
+      }
+      p += src_stride;
+   }
+}
+
+
+/*** PIPE_FORMAT_L8_SRGB ***/
+
+static void
+l8_srgb_get_tile_rgba(const ubyte *src,
+                      unsigned w, unsigned h,
+                      float *p,
+                      unsigned dst_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      float *pRow = p;
+      for (j = 0; j < w; j++, src++, pRow += 4) {
+         pRow[0] =
+         pRow[1] =
+         pRow[2] = srgb_to_linear(*src);
+         pRow[3] = 1.0;
+      }
+      p += dst_stride;
+   }
+}
+
+static void
+l8_srgb_put_tile_rgba(ubyte *dst,
+                      unsigned w, unsigned h,
+                      const float *p,
+                      unsigned src_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      const float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         unsigned r;
+         r = linear_to_srgb(pRow[0]);
+         *dst++ = (ubyte) r;
+      }
+      p += src_stride;
+   }
+}
+
 
 /*** PIPE_FORMAT_I8_UNORM ***/
 
@@ -946,6 +1124,15 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
    case PIPE_FORMAT_R16G16B16A16_SNORM:
       r16g16b16a16_get_tile_rgba((short *) src, w, h, dst, dst_stride);
       break;
+   case PIPE_FORMAT_A8R8G8B8_SRGB:
+      a8r8g8b8_srgb_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
+      break;
+   case PIPE_FORMAT_A8L8_SRGB:
+      a8l8_srgb_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
+      break;
+   case PIPE_FORMAT_L8_SRGB:
+      l8_srgb_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
+      break;
    case PIPE_FORMAT_Z16_UNORM:
       z16_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
       break;
@@ -1059,6 +1246,15 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
    case PIPE_FORMAT_R16G16B16A16_SNORM:
       r16g16b16a16_put_tile_rgba((short *) packed, w, h, p, src_stride);
       break;
+   case PIPE_FORMAT_A8R8G8B8_SRGB:
+      a8r8g8b8_srgb_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
+      break;
+   case PIPE_FORMAT_A8L8_SRGB:
+      a8l8_srgb_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
+      break;
+   case PIPE_FORMAT_L8_SRGB:
+      l8_srgb_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
+      break;
    case PIPE_FORMAT_Z16_UNORM:
       /*z16_put_tile_rgba((ushort *) packed, w, h, p, src_stride);*/
       break;