DXR is a code search and navigation tool aimed at making sense of large projects. It supports full-text and regex searches as well as structural queries.

Mercurial (31ec81b5d7bb)

VCS Links

Line Code
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
changeset:   94061:73a9b24d863a
tag:         bilin
tag:         qbase
tag:         qtip
tag:         tip
user:        Jeff Muizelaar <jmuizelaar@mozilla.com>
date:        Tue May 15 18:26:16 2012 -0400
summary:     Bug 754364. Add bilinear non-repeat and repeat fast paths. r=joe

diff --git a/gfx/cairo/libpixman/src/pixman-fast-path.c b/gfx/cairo/libpixman/src/pixman-fast-path.c
--- a/gfx/cairo/libpixman/src/pixman-fast-path.c
+++ b/gfx/cairo/libpixman/src/pixman-fast-path.c
@@ -1186,16 +1186,228 @@ FAST_NEAREST (8888_565_none, 8888, 0565,
 FAST_NEAREST (8888_565_pad, 8888, 0565, uint32_t, uint16_t, SRC, PAD)
 FAST_NEAREST (8888_565_normal, 8888, 0565, uint32_t, uint16_t, SRC, NORMAL)
 FAST_NEAREST (565_565_normal, 0565, 0565, uint16_t, uint16_t, SRC, NORMAL)
 FAST_NEAREST (8888_565_cover, 8888, 0565, uint32_t, uint16_t, OVER, COVER)
 FAST_NEAREST (8888_565_none, 8888, 0565, uint32_t, uint16_t, OVER, NONE)
 FAST_NEAREST (8888_565_pad, 8888, 0565, uint32_t, uint16_t, OVER, PAD)
 FAST_NEAREST (8888_565_normal, 8888, 0565, uint32_t, uint16_t, OVER, NORMAL)
 
+static force_inline void
+scaled_bilinear_scanline_8888_565_OVER (uint16_t *       dst,
+                                        const uint32_t * mask,
+                                        const uint32_t * src_top,
+                                        const uint32_t * src_bottom,
+                                        int32_t          w,
+                                        int              wt,
+                                        int              wb,
+                                        pixman_fixed_t   vx,
+                                        pixman_fixed_t   unit_x,
+                                        pixman_fixed_t   max_vx,
+                                        pixman_bool_t    zero_src)
+{
+    while ((w -= 1) >= 0)
+    {
+	uint32_t tl = src_top [pixman_fixed_to_int (vx)];
+	uint32_t tr = src_top [pixman_fixed_to_int (vx) + 1];
+	uint32_t bl = src_bottom [pixman_fixed_to_int (vx)];
+	uint32_t br = src_bottom [pixman_fixed_to_int (vx) + 1];
+	uint32_t src, result;
+	uint16_t d;
+	d = *dst;
+	src = bilinear_interpolation (tl, tr,
+				      bl, br,
+				      interpolation_coord(vx),
+				      wb >> (8 - INTERPOLATION_PRECISION_BITS));
+	vx += unit_x;
+	result = over (src, CONVERT_0565_TO_0888 (d));
+	*dst++ = CONVERT_8888_TO_0565(result);
+    }
+}
+
+static force_inline void
+scaled_bilinear_scanline_8888_8888_OVER (uint32_t *       dst,
+                                         const uint32_t * mask,
+                                         const uint32_t * src_top,
+                                         const uint32_t * src_bottom,
+                                         int32_t          w,
+                                         int              wt,
+                                         int              wb,
+                                         pixman_fixed_t   vx,
+                                         pixman_fixed_t   unit_x,
+                                         pixman_fixed_t   max_vx,
+                                         pixman_bool_t    zero_src)
+{
+    while ((w -= 1) >= 0)
+    {
+	uint32_t tl = src_top [pixman_fixed_to_int (vx)];
+	uint32_t tr = src_top [pixman_fixed_to_int (vx) + 1];
+	uint32_t bl = src_bottom [pixman_fixed_to_int (vx)];
+	uint32_t br = src_bottom [pixman_fixed_to_int (vx) + 1];
+	uint32_t src;
+	uint32_t d;
+	uint32_t result;
+	d = *dst;
+	src = bilinear_interpolation (tl, tr,
+				      bl, br,
+				      interpolation_coord(vx),
+				      wb >> (8 - INTERPOLATION_PRECISION_BITS));
+	vx += unit_x;
+	*dst++ = over (src, d);
+    }
+}
+
+#if 1
+
+static force_inline void
+scaled_bilinear_scanline_565_565_SRC (uint16_t *       dst,
+				      const uint32_t * mask,
+				      const uint16_t * src_top,
+				      const uint16_t * src_bottom,
+				      int32_t          w,
+				      int              wt,
+				      int              wb,
+				      pixman_fixed_t   vx,
+				      pixman_fixed_t   unit_x,
+				      pixman_fixed_t   max_vx,
+				      pixman_bool_t    zero_src)
+{
+    while ((w -= 1) >= 0)
+    {
+	uint16_t tl = src_top [pixman_fixed_to_int (vx)];
+	uint16_t tr = src_top [pixman_fixed_to_int (vx) + 1];
+	uint16_t bl = src_bottom [pixman_fixed_to_int (vx)];
+	uint16_t br = src_bottom [pixman_fixed_to_int (vx) + 1];
+	uint32_t d;
+	d = bilinear_interpolation(CONVERT_0565_TO_8888(tl),
+				   CONVERT_0565_TO_8888(tr),
+				   CONVERT_0565_TO_8888(bl),
+				   CONVERT_0565_TO_8888(br),
+				   interpolation_coord(vx),
+				   wb >> (8 - INTERPOLATION_PRECISION_BITS));
+	vx += unit_x;
+	*dst++ = CONVERT_8888_TO_0565(d);
+    }
+}
+
+#else
+
+#define SK_G16_MASK_IN_PLACE 0xfc0
+
+static inline uint32_t SkExpand_rgb_16(uint16_t c) {
+
+    return ((c & SK_G16_MASK_IN_PLACE) << 16) | (c & ~SK_G16_MASK_IN_PLACE);
+}
+
+/** Compress an expanded value (from SkExpand_rgb_16) back down to a 16bit
+    color value. The computation yields only 16bits of valid data, but we claim
+    to return 32bits, so that the compiler won't generate extra instructions to
+    "clean" the top 16bits. However, the top 16 can contain garbage, so it is
+    up to the caller to safely ignore them.
+*/
+static inline uint16_t SkCompact_rgb_16(uint32_t c) {
+    return ((c >> 16) & SK_G16_MASK_IN_PLACE) | (c & ~SK_G16_MASK_IN_PLACE);
+}
+// returns expanded * 5bits
+static inline uint32_t Filter_565_Expanded(unsigned x, unsigned y,
+                                           uint32_t a00, uint32_t a01,
+                                           uint32_t a10, uint32_t a11) {
+    a00 = SkExpand_rgb_16(a00);
+    a01 = SkExpand_rgb_16(a01);
+    a10 = SkExpand_rgb_16(a10);
+    a11 = SkExpand_rgb_16(a11);
+    
+    int xy = x * y >> 3;
+    return  a00 * (32 - 2*y - 2*x + xy) +
+            a01 * (2*x - xy) +
+            a10 * (2*y - xy) +
+            a11 * xy;
+}
+
+
+
+static force_inline void
+scaled_bilinear_scanline_565_565_SRC (uint16_t *       dst,
+				      const uint32_t * mask,
+				      const uint16_t * src_top,
+				      const uint16_t * src_bottom,
+				      int32_t          w,
+				      int              wt,
+				      int              wb,
+				      pixman_fixed_t   vx,
+				      pixman_fixed_t   unit_x,
+				      pixman_fixed_t   max_vx,
+				      pixman_bool_t    zero_src)
+{
+    while ((w -= 1) >= 0)
+    {
+	uint16_t tl = src_top [pixman_fixed_to_int (vx)];
+	uint16_t tr = src_top [pixman_fixed_to_int (vx) + 1];
+	uint16_t bl = src_bottom [pixman_fixed_to_int (vx)];
+	uint16_t br = src_bottom [pixman_fixed_to_int (vx) + 1];
+
+        uint32_t tmp = Filter_565_Expanded((vx>>12)&0xf, wb>>4, tl, tr, bl, br);
+        vx += unit_x;
+        *dst++ = SkCompact_rgb_16((tmp) >> 5);
+    }
+}
+
+
+#endif
+FAST_BILINEAR_MAINLOOP_COMMON (565_565_cover_SRC,
+			       scaled_bilinear_scanline_565_565_SRC,
+			       uint16_t, uint32_t, uint16_t,
+			       COVER, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (565_565_pad_SRC,
+			       scaled_bilinear_scanline_565_565_SRC,
+			       uint16_t, uint32_t, uint16_t,
+			       PAD, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (565_565_none_SRC,
+			       scaled_bilinear_scanline_565_565_SRC,
+			       uint16_t, uint32_t, uint16_t,
+			       NONE, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (565_565_normal_SRC,
+			       scaled_bilinear_scanline_565_565_SRC,
+			       uint16_t, uint32_t, uint16_t,
+			       NORMAL, FLAG_NONE)
+
+FAST_BILINEAR_MAINLOOP_COMMON (8888_565_cover_OVER,
+			       scaled_bilinear_scanline_8888_565_OVER,
+			       uint32_t, uint32_t, uint16_t,
+			       COVER, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (8888_565_pad_OVER,
+			       scaled_bilinear_scanline_8888_565_OVER,
+			       uint32_t, uint32_t, uint16_t,
+			       PAD, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (8888_565_none_OVER,
+			       scaled_bilinear_scanline_8888_565_OVER,
+			       uint32_t, uint32_t, uint16_t,
+			       NONE, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (8888_565_normal_OVER,
+			       scaled_bilinear_scanline_8888_565_OVER,
+			       uint32_t, uint32_t, uint16_t,
+			       NORMAL, FLAG_NONE)
+
+FAST_BILINEAR_MAINLOOP_COMMON (8888_8888_cover_OVER,
+			       scaled_bilinear_scanline_8888_8888_OVER,
+			       uint32_t, uint32_t, uint32_t,
+			       COVER, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (8888_8888_pad_OVER,
+			       scaled_bilinear_scanline_8888_8888_OVER,
+			       uint32_t, uint32_t, uint32_t,
+			       PAD, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (8888_8888_none_OVER,
+			       scaled_bilinear_scanline_8888_8888_OVER,
+			       uint32_t, uint32_t, uint32_t,
+			       NONE, FLAG_NONE)
+FAST_BILINEAR_MAINLOOP_COMMON (8888_8888_normal_OVER,
+			       scaled_bilinear_scanline_8888_8888_OVER,
+			       uint32_t, uint32_t, uint32_t,
+			       NORMAL, FLAG_NONE)
+
 #define REPEAT_MIN_WIDTH    32
 
 static void
 fast_composite_tiled_repeat (pixman_implementation_t *imp,
 			     pixman_composite_info_t *info)
 {
     PIXMAN_COMPOSITE_ARGS (info);
     pixman_composite_func_t func;
@@ -1960,16 +2172,20 @@ static const pixman_fast_path_t c_fast_p
 	PIXMAN_any,
 	(FAST_PATH_STANDARD_FLAGS | FAST_PATH_ID_TRANSFORM | FAST_PATH_BITS_IMAGE |
 	 FAST_PATH_NORMAL_REPEAT),
 	PIXMAN_any, 0,
 	PIXMAN_any, FAST_PATH_STD_DEST_FLAGS,
 	fast_composite_tiled_repeat
     },
 
+    SIMPLE_BILINEAR_FAST_PATH (SRC, r5g6b5, r5g6b5, 565_565),
+    SIMPLE_BILINEAR_FAST_PATH (OVER, a8r8g8b8, r5g6b5, 8888_565),
+    SIMPLE_BILINEAR_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, 8888_8888),
+
     {   PIXMAN_OP_NONE	},
 };
 
 #ifdef WORDS_BIGENDIAN
 #define A1_FILL_MASK(n, offs) (((1U << (n)) - 1) << (32 - (offs) - (n)))
 #else
 #define A1_FILL_MASK(n, offs) (((1U << (n)) - 1) << (offs))
 #endif
diff --git a/gfx/cairo/libpixman/src/pixman-inlines.h b/gfx/cairo/libpixman/src/pixman-inlines.h
--- a/gfx/cairo/libpixman/src/pixman-inlines.h
+++ b/gfx/cairo/libpixman/src/pixman-inlines.h
@@ -80,16 +80,21 @@ repeat (pixman_repeat_t repeat, int *c, 
     }
     return TRUE;
 }
 
 #ifdef MOZ_GFX_OPTIMIZE_MOBILE
 #define LOW_QUALITY_INTERPOLATION
 #endif
 
+#ifdef LOW_QUALITY_INTERPOLATION
+#define INTERPOLATION_PRECISION_BITS 4
+#else
+#define INTERPOLATION_PRECISION_BITS 8
+#endif
 static force_inline int32_t
 interpolation_coord(pixman_fixed_t t)
 {
 #ifdef LOW_QUALITY_INTERPOLATION
     return (t >> 12) & 0xf;
 #else
     return (t >> 8) & 0xff;
 #endif