From 4dd0483a9302ec79e7d87bf45434b6718924ba03 Mon Sep 17 00:00:00 2001 From: "Sergey Semyonov (Serge)" Date: Sun, 27 Feb 2011 17:46:46 +0000 Subject: [PATCH] pixman-1.0 git-svn-id: svn://kolibrios.org@1891 a494cfbc-eb01-0410-851d-a64ba20cac60 --- programs/develop/libraries/pixman/Makefile | 54 + .../pixman/pixman-access-accessors.c | 3 + .../develop/libraries/pixman/pixman-access.c | 2989 +++++++++++++++ .../libraries/pixman/pixman-accessor.h | 40 + .../libraries/pixman/pixman-bits-image.c | 1331 +++++++ .../libraries/pixman/pixman-combine32.c | 2465 ++++++++++++ .../libraries/pixman/pixman-combine32.h | 230 ++ .../libraries/pixman/pixman-combine64.c | 2465 ++++++++++++ .../libraries/pixman/pixman-combine64.h | 230 ++ .../libraries/pixman/pixman-compiler.h | 204 + .../pixman/pixman-conical-gradient.c | 198 + .../develop/libraries/pixman/pixman-cpu.c | 598 +++ .../libraries/pixman/pixman-edge-accessors.c | 4 + .../libraries/pixman/pixman-edge-imp.h | 182 + .../develop/libraries/pixman/pixman-edge.c | 384 ++ .../libraries/pixman/pixman-fast-path.c | 1852 +++++++++ .../libraries/pixman/pixman-fast-path.h | 451 +++ .../develop/libraries/pixman/pixman-general.c | 315 ++ .../libraries/pixman/pixman-gradient-walker.c | 254 ++ .../develop/libraries/pixman/pixman-image.c | 846 +++++ .../libraries/pixman/pixman-implementation.c | 227 ++ .../libraries/pixman/pixman-linear-gradient.c | 262 ++ .../develop/libraries/pixman/pixman-matrix.c | 768 ++++ .../develop/libraries/pixman/pixman-mmx.c | 3378 +++++++++++++++++ .../develop/libraries/pixman/pixman-private.h | 873 +++++ .../libraries/pixman/pixman-radial-gradient.c | 448 +++ .../develop/libraries/pixman/pixman-region.c | 2769 ++++++++++++++ .../libraries/pixman/pixman-region16.c | 67 + .../libraries/pixman/pixman-region32.c | 47 + .../libraries/pixman/pixman-solid-fill.c | 117 + .../develop/libraries/pixman/pixman-timer.c | 66 + .../develop/libraries/pixman/pixman-trap.c | 392 ++ .../develop/libraries/pixman/pixman-utils.c | 258 ++ .../develop/libraries/pixman/pixman-version.h | 50 + .../pixman/pixman-x64-mmx-emulation.h | 263 ++ programs/develop/libraries/pixman/pixman.c | 1274 +++++++ programs/develop/libraries/pixman/pixman.h | 950 +++++ .../libraries/pixman/test/window-test.c | 173 + 38 files changed, 27477 insertions(+) create mode 100644 programs/develop/libraries/pixman/Makefile create mode 100644 programs/develop/libraries/pixman/pixman-access-accessors.c create mode 100644 programs/develop/libraries/pixman/pixman-access.c create mode 100644 programs/develop/libraries/pixman/pixman-accessor.h create mode 100644 programs/develop/libraries/pixman/pixman-bits-image.c create mode 100644 programs/develop/libraries/pixman/pixman-combine32.c create mode 100644 programs/develop/libraries/pixman/pixman-combine32.h create mode 100644 programs/develop/libraries/pixman/pixman-combine64.c create mode 100644 programs/develop/libraries/pixman/pixman-combine64.h create mode 100644 programs/develop/libraries/pixman/pixman-compiler.h create mode 100644 programs/develop/libraries/pixman/pixman-conical-gradient.c create mode 100644 programs/develop/libraries/pixman/pixman-cpu.c create mode 100644 programs/develop/libraries/pixman/pixman-edge-accessors.c create mode 100644 programs/develop/libraries/pixman/pixman-edge-imp.h create mode 100644 programs/develop/libraries/pixman/pixman-edge.c create mode 100644 programs/develop/libraries/pixman/pixman-fast-path.c create mode 100644 programs/develop/libraries/pixman/pixman-fast-path.h create mode 100644 programs/develop/libraries/pixman/pixman-general.c create mode 100644 programs/develop/libraries/pixman/pixman-gradient-walker.c create mode 100644 programs/develop/libraries/pixman/pixman-image.c create mode 100644 programs/develop/libraries/pixman/pixman-implementation.c create mode 100644 programs/develop/libraries/pixman/pixman-linear-gradient.c create mode 100644 programs/develop/libraries/pixman/pixman-matrix.c create mode 100644 programs/develop/libraries/pixman/pixman-mmx.c create mode 100644 programs/develop/libraries/pixman/pixman-private.h create mode 100644 programs/develop/libraries/pixman/pixman-radial-gradient.c create mode 100644 programs/develop/libraries/pixman/pixman-region.c create mode 100644 programs/develop/libraries/pixman/pixman-region16.c create mode 100644 programs/develop/libraries/pixman/pixman-region32.c create mode 100644 programs/develop/libraries/pixman/pixman-solid-fill.c create mode 100644 programs/develop/libraries/pixman/pixman-timer.c create mode 100644 programs/develop/libraries/pixman/pixman-trap.c create mode 100644 programs/develop/libraries/pixman/pixman-utils.c create mode 100644 programs/develop/libraries/pixman/pixman-version.h create mode 100644 programs/develop/libraries/pixman/pixman-x64-mmx-emulation.h create mode 100644 programs/develop/libraries/pixman/pixman.c create mode 100644 programs/develop/libraries/pixman/pixman.h create mode 100644 programs/develop/libraries/pixman/test/window-test.c diff --git a/programs/develop/libraries/pixman/Makefile b/programs/develop/libraries/pixman/Makefile new file mode 100644 index 0000000000..211b7bd5e3 --- /dev/null +++ b/programs/develop/libraries/pixman/Makefile @@ -0,0 +1,54 @@ + +LIBRARY = pixman-1 + +CC = gcc + +CFLAGS = -c -O2 -mmmx -Winline + +DEFINES = -DHAVE_CONFIG_H -DPACKAGE -DPIXMAN_NO_TLS -DUSE_MMX + +INCLUDES = -I../pixman -I../newlib/include -I../newlib/include/sys + +SOURCES = \ + pixman-image.c \ + pixman-access.c \ + pixman-access-accessors.c \ + pixman-region16.c \ + pixman-region32.c \ + pixman-combine32.c \ + pixman-combine64.c \ + pixman-utils.c \ + pixman-edge.c \ + pixman-edge-accessors.c \ + pixman-trap.c \ + pixman-timer.c \ + pixman-matrix.c \ + pixman-gradient-walker.c \ + pixman-linear-gradient.c \ + pixman-radial-gradient.c \ + pixman-bits-image.c \ + pixman.c \ + pixman-cpu.c \ + pixman-fast-path.c \ + pixman-implementation.c \ + pixman-solid-fill.c \ + pixman-general.c \ + pixman-mmx.c \ + $(NULL) + +OBJECTS = $(patsubst %.c, %.o, $(SOURCES)) + +# targets + +all:$(LIBRARY).a + + +$(LIBRARY).a: $(OBJECTS) Makefile + ar cvrs $(LIBRARY).a $(OBJECTS) + +%.o: %.c $(SOURCES) Makefile + $(CC) $(CFLAGS) $(DEFINES) $(INCLUDES) -o $@ $< + + + + \ No newline at end of file diff --git a/programs/develop/libraries/pixman/pixman-access-accessors.c b/programs/develop/libraries/pixman/pixman-access-accessors.c new file mode 100644 index 0000000000..3263582f18 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-access-accessors.c @@ -0,0 +1,3 @@ +#define PIXMAN_FB_ACCESSORS + +#include "pixman-access.c" diff --git a/programs/develop/libraries/pixman/pixman-access.c b/programs/develop/libraries/pixman/pixman-access.c new file mode 100644 index 0000000000..f1ce0ba400 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-access.c @@ -0,0 +1,2989 @@ +/* + * + * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc. + * 2005 Lars Knoll & Zack Rusin, Trolltech + * 2008 Aaron Plattner, NVIDIA Corporation + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif + +#include +#include +#include + +#include "pixman-private.h" +#include "pixman-accessor.h" + +#define CONVERT_RGB24_TO_Y15(s) \ + (((((s) >> 16) & 0xff) * 153 + \ + (((s) >> 8) & 0xff) * 301 + \ + (((s) ) & 0xff) * 58) >> 2) + +#define CONVERT_RGB24_TO_RGB15(s) \ + ((((s) >> 3) & 0x001f) | \ + (((s) >> 6) & 0x03e0) | \ + (((s) >> 9) & 0x7c00)) + +#define RGB15_TO_ENTRY(mif,rgb15) \ + ((mif)->ent[rgb15]) + +#define RGB24_TO_ENTRY(mif,rgb24) \ + RGB15_TO_ENTRY (mif,CONVERT_RGB24_TO_RGB15 (rgb24)) + +#define RGB24_TO_ENTRY_Y(mif,rgb24) \ + ((mif)->ent[CONVERT_RGB24_TO_Y15 (rgb24)]) + +/* + * YV12 setup and access macros + */ + +#define YV12_SETUP(image) \ + bits_image_t *__bits_image = (bits_image_t *)image; \ + uint32_t *bits = __bits_image->bits; \ + int stride = __bits_image->rowstride; \ + int offset0 = stride < 0 ? \ + ((-stride) >> 1) * ((__bits_image->height - 1) >> 1) - stride : \ + stride * __bits_image->height; \ + int offset1 = stride < 0 ? \ + offset0 + ((-stride) >> 1) * ((__bits_image->height) >> 1) : \ + offset0 + (offset0 >> 2) + +/* Note no trailing semicolon on the above macro; if it's there, then + * the typical usage of YV12_SETUP(image); will have an extra trailing ; + * that some compilers will interpret as a statement -- and then any further + * variable declarations will cause an error. + */ + +#define YV12_Y(line) \ + ((uint8_t *) ((bits) + (stride) * (line))) + +#define YV12_U(line) \ + ((uint8_t *) ((bits) + offset1 + \ + ((stride) >> 1) * ((line) >> 1))) + +#define YV12_V(line) \ + ((uint8_t *) ((bits) + offset0 + \ + ((stride) >> 1) * ((line) >> 1))) + +/********************************** Fetch ************************************/ + +static void +fetch_scanline_a8r8g8b8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + + MEMCPY_WRAPPED (image, + buffer, (const uint32_t *)bits + x, + width * sizeof(uint32_t)); +} + +static void +fetch_scanline_x8r8g8b8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = (const uint32_t *)bits + x; + const uint32_t *end = pixel + width; + + while (pixel < end) + *buffer++ = READ (image, pixel++) | 0xff000000; +} + +static void +fetch_scanline_a8b8g8r8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = (uint32_t *)bits + x; + const uint32_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + + *buffer++ = (p & 0xff00ff00) | + ((p >> 16) & 0xff) | + ((p & 0xff) << 16); + } +} + +static void +fetch_scanline_x8b8g8r8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = (uint32_t *)bits + x; + const uint32_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + + *buffer++ = 0xff000000 | + (p & 0x0000ff00) | + ((p >> 16) & 0xff) | + ((p & 0xff) << 16); + } +} + +static void +fetch_scanline_b8g8r8a8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = (uint32_t *)bits + x; + const uint32_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + + *buffer++ = (((p & 0xff000000) >> 24) | + ((p & 0x00ff0000) >> 8) | + ((p & 0x0000ff00) << 8) | + ((p & 0x000000ff) << 24)); + } +} + +static void +fetch_scanline_b8g8r8x8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = (uint32_t *)bits + x; + const uint32_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + + *buffer++ = (0xff000000 | + ((p & 0xff000000) >> 24) | + ((p & 0x00ff0000) >> 8) | + ((p & 0x0000ff00) << 8)); + } +} + +static void +fetch_scanline_x14r6g6b6 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = (const uint32_t *)bits + x; + const uint32_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b; + + r = ((p & 0x3f000) << 6) | ((p & 0x30000)); + g = ((p & 0x00fc0) << 4) | ((p & 0x00c00) >> 2); + b = ((p & 0x0003f) << 2) | ((p & 0x00030) >> 4); + + *buffer++ = 0xff000000 | r | g | b; + } +} + +/* Expects a uint64_t buffer */ +static void +fetch_scanline_a2r10g10b10 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * b, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = bits + x; + const uint32_t *end = pixel + width; + uint64_t *buffer = (uint64_t *)b; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint64_t a = p >> 30; + uint64_t r = (p >> 20) & 0x3ff; + uint64_t g = (p >> 10) & 0x3ff; + uint64_t b = p & 0x3ff; + + r = r << 6 | r >> 4; + g = g << 6 | g >> 4; + b = b << 6 | b >> 4; + + a <<= 14; + a |= a >> 2; + a |= a >> 4; + a |= a >> 8; + + *buffer++ = a << 48 | r << 32 | g << 16 | b; + } +} + +/* Expects a uint64_t buffer */ +static void +fetch_scanline_x2r10g10b10 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * b, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = (uint32_t *)bits + x; + const uint32_t *end = pixel + width; + uint64_t *buffer = (uint64_t *)b; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint64_t r = (p >> 20) & 0x3ff; + uint64_t g = (p >> 10) & 0x3ff; + uint64_t b = p & 0x3ff; + + r = r << 6 | r >> 4; + g = g << 6 | g >> 4; + b = b << 6 | b >> 4; + + *buffer++ = 0xffffULL << 48 | r << 32 | g << 16 | b; + } +} + +/* Expects a uint64_t buffer */ +static void +fetch_scanline_a2b10g10r10 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * b, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = bits + x; + const uint32_t *end = pixel + width; + uint64_t *buffer = (uint64_t *)b; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint64_t a = p >> 30; + uint64_t b = (p >> 20) & 0x3ff; + uint64_t g = (p >> 10) & 0x3ff; + uint64_t r = p & 0x3ff; + + r = r << 6 | r >> 4; + g = g << 6 | g >> 4; + b = b << 6 | b >> 4; + + a <<= 14; + a |= a >> 2; + a |= a >> 4; + a |= a >> 8; + + *buffer++ = a << 48 | r << 32 | g << 16 | b; + } +} + +/* Expects a uint64_t buffer */ +static void +fetch_scanline_x2b10g10r10 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * b, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint32_t *pixel = (uint32_t *)bits + x; + const uint32_t *end = pixel + width; + uint64_t *buffer = (uint64_t *)b; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint64_t b = (p >> 20) & 0x3ff; + uint64_t g = (p >> 10) & 0x3ff; + uint64_t r = p & 0x3ff; + + r = r << 6 | r >> 4; + g = g << 6 | g >> 4; + b = b << 6 | b >> 4; + + *buffer++ = 0xffffULL << 48 | r << 32 | g << 16 | b; + } +} + +static void +fetch_scanline_r8g8b8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint8_t *pixel = (const uint8_t *)bits + 3 * x; + const uint8_t *end = pixel + 3 * width; + + while (pixel < end) + { + uint32_t b = 0xff000000; + +#ifdef WORDS_BIGENDIAN + b |= (READ (image, pixel++) << 16); + b |= (READ (image, pixel++) << 8); + b |= (READ (image, pixel++)); +#else + b |= (READ (image, pixel++)); + b |= (READ (image, pixel++) << 8); + b |= (READ (image, pixel++) << 16); +#endif + + *buffer++ = b; + } +} + +static void +fetch_scanline_b8g8r8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint8_t *pixel = (const uint8_t *)bits + 3 * x; + const uint8_t *end = pixel + 3 * width; + + while (pixel < end) + { + uint32_t b = 0xff000000; +#ifdef WORDS_BIGENDIAN + b |= (READ (image, pixel++)); + b |= (READ (image, pixel++) << 8); + b |= (READ (image, pixel++) << 16); +#else + b |= (READ (image, pixel++) << 16); + b |= (READ (image, pixel++) << 8); + b |= (READ (image, pixel++)); +#endif + *buffer++ = b; + } +} + +static void +fetch_scanline_r5g6b5 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r = (((p) << 3) & 0xf8) | + (((p) << 5) & 0xfc00) | + (((p) << 8) & 0xf80000); + + r |= (r >> 5) & 0x70007; + r |= (r >> 6) & 0x300; + + *buffer++ = 0xff000000 | r; + } +} + +static void +fetch_scanline_b5g6r5 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b; + + b = ((p & 0xf800) | ((p & 0xe000) >> 5)) >> 8; + g = ((p & 0x07e0) | ((p & 0x0600) >> 6)) << 5; + r = ((p & 0x001c) | ((p & 0x001f) << 5)) << 14; + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_a1r5g5b5 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b, a; + + a = (uint32_t) ((uint8_t) (0 - ((p & 0x8000) >> 15))) << 24; + r = ((p & 0x7c00) | ((p & 0x7000) >> 5)) << 9; + g = ((p & 0x03e0) | ((p & 0x0380) >> 5)) << 6; + b = ((p & 0x001c) | ((p & 0x001f) << 5)) >> 2; + + *buffer++ = a | r | g | b; + } +} + +static void +fetch_scanline_x1r5g5b5 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b; + + r = ((p & 0x7c00) | ((p & 0x7000) >> 5)) << 9; + g = ((p & 0x03e0) | ((p & 0x0380) >> 5)) << 6; + b = ((p & 0x001c) | ((p & 0x001f) << 5)) >> 2; + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_a1b5g5r5 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + uint32_t r, g, b, a; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + + a = (uint32_t) ((uint8_t) (0 - ((p & 0x8000) >> 15))) << 24; + b = ((p & 0x7c00) | ((p & 0x7000) >> 5)) >> 7; + g = ((p & 0x03e0) | ((p & 0x0380) >> 5)) << 6; + r = ((p & 0x001c) | ((p & 0x001f) << 5)) << 14; + + *buffer++ = a | r | g | b; + } +} + +static void +fetch_scanline_x1b5g5r5 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b; + + b = ((p & 0x7c00) | ((p & 0x7000) >> 5)) >> 7; + g = ((p & 0x03e0) | ((p & 0x0380) >> 5)) << 6; + r = ((p & 0x001c) | ((p & 0x001f) << 5)) << 14; + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_a4r4g4b4 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b, a; + + a = ((p & 0xf000) | ((p & 0xf000) >> 4)) << 16; + r = ((p & 0x0f00) | ((p & 0x0f00) >> 4)) << 12; + g = ((p & 0x00f0) | ((p & 0x00f0) >> 4)) << 8; + b = ((p & 0x000f) | ((p & 0x000f) << 4)); + + *buffer++ = a | r | g | b; + } +} + +static void +fetch_scanline_x4r4g4b4 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b; + + r = ((p & 0x0f00) | ((p & 0x0f00) >> 4)) << 12; + g = ((p & 0x00f0) | ((p & 0x00f0) >> 4)) << 8; + b = ((p & 0x000f) | ((p & 0x000f) << 4)); + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_a4b4g4r4 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b, a; + + a = ((p & 0xf000) | ((p & 0xf000) >> 4)) << 16; + b = ((p & 0x0f00) | ((p & 0x0f00) >> 4)) >> 4; + g = ((p & 0x00f0) | ((p & 0x00f0) >> 4)) << 8; + r = ((p & 0x000f) | ((p & 0x000f) << 4)) << 16; + + *buffer++ = a | r | g | b; + } +} + +static void +fetch_scanline_x4b4g4r4 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint16_t *pixel = (const uint16_t *)bits + x; + const uint16_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b; + + b = ((p & 0x0f00) | ((p & 0x0f00) >> 4)) >> 4; + g = ((p & 0x00f0) | ((p & 0x00f0) >> 4)) << 8; + r = ((p & 0x000f) | ((p & 0x000f) << 4)) << 16; + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_a8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint8_t *pixel = (const uint8_t *)bits + x; + const uint8_t *end = pixel + width; + + while (pixel < end) + *buffer++ = READ (image, pixel++) << 24; +} + +static void +fetch_scanline_r3g3b2 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint8_t *pixel = (const uint8_t *)bits + x; + const uint8_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b; + + r = ((p & 0xe0) | ((p & 0xe0) >> 3) | ((p & 0xc0) >> 6)) << 16; + g = ((p & 0x1c) | ((p & 0x18) >> 3) | ((p & 0x1c) << 3)) << 8; + b = (((p & 0x03) ) | + ((p & 0x03) << 2) | + ((p & 0x03) << 4) | + ((p & 0x03) << 6)); + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_b2g3r3 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint8_t *pixel = (const uint8_t *)bits + x; + const uint8_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t r, g, b; + + b = p & 0xc0; + b |= b >> 2; + b |= b >> 4; + b &= 0xff; + + g = (p & 0x38) << 10; + g |= g >> 3; + g |= g >> 6; + g &= 0xff00; + + r = (p & 0x7) << 21; + r |= r >> 3; + r |= r >> 6; + r &= 0xff0000; + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_a2r2g2b2 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint8_t *pixel = (const uint8_t *)bits + x; + const uint8_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t a, r, g, b; + + a = ((p & 0xc0) * 0x55) << 18; + r = ((p & 0x30) * 0x55) << 12; + g = ((p & 0x0c) * 0x55) << 6; + b = ((p & 0x03) * 0x55); + + *buffer++ = a | r | g | b; + } +} + +static void +fetch_scanline_a2b2g2r2 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint8_t *pixel = (const uint8_t *)bits + x; + const uint8_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + uint32_t a, r, g, b; + + a = ((p & 0xc0) * 0x55) << 18; + b = ((p & 0x30) * 0x55) >> 4; + g = ((p & 0x0c) * 0x55) << 6; + r = ((p & 0x03) * 0x55) << 16; + + *buffer++ = a | r | g | b; + } +} + +static void +fetch_scanline_c8 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const pixman_indexed_t * indexed = image->bits.indexed; + const uint8_t *pixel = (const uint8_t *)bits + x; + const uint8_t *end = pixel + width; + + while (pixel < end) + { + uint32_t p = READ (image, pixel++); + + *buffer++ = indexed->rgba[p]; + } +} + +static void +fetch_scanline_x4a4 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const uint8_t *pixel = (const uint8_t *)bits + x; + const uint8_t *end = pixel + width; + + while (pixel < end) + { + uint8_t p = READ (image, pixel++) & 0xf; + + *buffer++ = (p | (p << 4)) << 24; + } +} + +#define FETCH_8(img,l,o) (READ (img, (((uint8_t *)(l)) + ((o) >> 3)))) +#ifdef WORDS_BIGENDIAN +#define FETCH_4(img,l,o) \ + (((4 * (o)) & 4) ? (FETCH_8 (img,l, 4 * (o)) & 0xf) : (FETCH_8 (img,l,(4 * (o))) >> 4)) +#else +#define FETCH_4(img,l,o) \ + (((4 * (o)) & 4) ? (FETCH_8 (img, l, 4 * (o)) >> 4) : (FETCH_8 (img, l, (4 * (o))) & 0xf)) +#endif + +static void +fetch_scanline_a4 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t p = FETCH_4 (image, bits, i + x); + + p |= p << 4; + + *buffer++ = p << 24; + } +} + +static void +fetch_scanline_r1g2b1 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t p = FETCH_4 (image, bits, i + x); + uint32_t r, g, b; + + r = ((p & 0x8) * 0xff) << 13; + g = ((p & 0x6) * 0x55) << 7; + b = ((p & 0x1) * 0xff); + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_b1g2r1 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t p = FETCH_4 (image, bits, i + x); + uint32_t r, g, b; + + b = ((p & 0x8) * 0xff) >> 3; + g = ((p & 0x6) * 0x55) << 7; + r = ((p & 0x1) * 0xff) << 16; + + *buffer++ = 0xff000000 | r | g | b; + } +} + +static void +fetch_scanline_a1r1g1b1 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + uint32_t a, r, g, b; + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t p = FETCH_4 (image, bits, i + x); + + a = ((p & 0x8) * 0xff) << 21; + r = ((p & 0x4) * 0xff) << 14; + g = ((p & 0x2) * 0xff) << 7; + b = ((p & 0x1) * 0xff); + + *buffer++ = a | r | g | b; + } +} + +static void +fetch_scanline_a1b1g1r1 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t p = FETCH_4 (image, bits, i + x); + uint32_t a, r, g, b; + + a = ((p & 0x8) * 0xff) << 21; + b = ((p & 0x4) * 0xff) >> 2; + g = ((p & 0x2) * 0xff) << 7; + r = ((p & 0x1) * 0xff) << 16; + + *buffer++ = a | r | g | b; + } +} + +static void +fetch_scanline_c4 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const pixman_indexed_t * indexed = image->bits.indexed; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t p = FETCH_4 (image, bits, i + x); + + *buffer++ = indexed->rgba[p]; + } +} + +static void +fetch_scanline_a1 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t p = READ (image, bits + ((i + x) >> 5)); + uint32_t a; + +#ifdef WORDS_BIGENDIAN + a = p >> (0x1f - ((i + x) & 0x1f)); +#else + a = p >> ((i + x) & 0x1f); +#endif + a = a & 1; + a |= a << 1; + a |= a << 2; + a |= a << 4; + + *buffer++ = a << 24; + } +} + +static void +fetch_scanline_g1 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + y * image->bits.rowstride; + const pixman_indexed_t * indexed = image->bits.indexed; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t p = READ (image, bits + ((i + x) >> 5)); + uint32_t a; + +#ifdef WORDS_BIGENDIAN + a = p >> (0x1f - ((i + x) & 0x1f)); +#else + a = p >> ((i + x) & 0x1f); +#endif + a = a & 1; + + *buffer++ = indexed->rgba[a]; + } +} + +static void +fetch_scanline_yuy2 (pixman_image_t *image, + int x, + int line, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + const uint32_t *bits = image->bits.bits + image->bits.rowstride * line; + int i; + + for (i = 0; i < width; i++) + { + int16_t y, u, v; + int32_t r, g, b; + + y = ((uint8_t *) bits)[(x + i) << 1] - 16; + u = ((uint8_t *) bits)[(((x + i) << 1) & - 4) + 1] - 128; + v = ((uint8_t *) bits)[(((x + i) << 1) & - 4) + 3] - 128; + + /* R = 1.164(Y - 16) + 1.596(V - 128) */ + r = 0x012b27 * y + 0x019a2e * v; + /* G = 1.164(Y - 16) - 0.813(V - 128) - 0.391(U - 128) */ + g = 0x012b27 * y - 0x00d0f2 * v - 0x00647e * u; + /* B = 1.164(Y - 16) + 2.018(U - 128) */ + b = 0x012b27 * y + 0x0206a2 * u; + + *buffer++ = 0xff000000 | + (r >= 0 ? r < 0x1000000 ? r & 0xff0000 : 0xff0000 : 0) | + (g >= 0 ? g < 0x1000000 ? (g >> 8) & 0x00ff00 : 0x00ff00 : 0) | + (b >= 0 ? b < 0x1000000 ? (b >> 16) & 0x0000ff : 0x0000ff : 0); + } +} + +static void +fetch_scanline_yv12 (pixman_image_t *image, + int x, + int line, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + YV12_SETUP (image); + uint8_t *y_line = YV12_Y (line); + uint8_t *u_line = YV12_U (line); + uint8_t *v_line = YV12_V (line); + int i; + + for (i = 0; i < width; i++) + { + int16_t y, u, v; + int32_t r, g, b; + + y = y_line[x + i] - 16; + u = u_line[(x + i) >> 1] - 128; + v = v_line[(x + i) >> 1] - 128; + + /* R = 1.164(Y - 16) + 1.596(V - 128) */ + r = 0x012b27 * y + 0x019a2e * v; + /* G = 1.164(Y - 16) - 0.813(V - 128) - 0.391(U - 128) */ + g = 0x012b27 * y - 0x00d0f2 * v - 0x00647e * u; + /* B = 1.164(Y - 16) + 2.018(U - 128) */ + b = 0x012b27 * y + 0x0206a2 * u; + + *buffer++ = 0xff000000 | + (r >= 0 ? r < 0x1000000 ? r & 0xff0000 : 0xff0000 : 0) | + (g >= 0 ? g < 0x1000000 ? (g >> 8) & 0x00ff00 : 0x00ff00 : 0) | + (b >= 0 ? b < 0x1000000 ? (b >> 16) & 0x0000ff : 0x0000ff : 0); + } +} + +/**************************** Pixel wise fetching *****************************/ + +/* Despite the type, expects a uint64_t buffer */ +static uint64_t +fetch_pixel_a2r10g10b10 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t p = READ (image, bits + offset); + uint64_t a = p >> 30; + uint64_t r = (p >> 20) & 0x3ff; + uint64_t g = (p >> 10) & 0x3ff; + uint64_t b = p & 0x3ff; + + r = r << 6 | r >> 4; + g = g << 6 | g >> 4; + b = b << 6 | b >> 4; + + a <<= 14; + a |= a >> 2; + a |= a >> 4; + a |= a >> 8; + + return a << 48 | r << 32 | g << 16 | b; +} + +/* Despite the type, this function expects a uint64_t buffer */ +static uint64_t +fetch_pixel_x2r10g10b10 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t p = READ (image, bits + offset); + uint64_t r = (p >> 20) & 0x3ff; + uint64_t g = (p >> 10) & 0x3ff; + uint64_t b = p & 0x3ff; + + r = r << 6 | r >> 4; + g = g << 6 | g >> 4; + b = b << 6 | b >> 4; + + return 0xffffULL << 48 | r << 32 | g << 16 | b; +} + +/* Despite the type, expects a uint64_t buffer */ +static uint64_t +fetch_pixel_a2b10g10r10 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t p = READ (image, bits + offset); + uint64_t a = p >> 30; + uint64_t b = (p >> 20) & 0x3ff; + uint64_t g = (p >> 10) & 0x3ff; + uint64_t r = p & 0x3ff; + + r = r << 6 | r >> 4; + g = g << 6 | g >> 4; + b = b << 6 | b >> 4; + + a <<= 14; + a |= a >> 2; + a |= a >> 4; + a |= a >> 8; + + return a << 48 | r << 32 | g << 16 | b; +} + +/* Despite the type, this function expects a uint64_t buffer */ +static uint64_t +fetch_pixel_x2b10g10r10 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t p = READ (image, bits + offset); + uint64_t b = (p >> 20) & 0x3ff; + uint64_t g = (p >> 10) & 0x3ff; + uint64_t r = p & 0x3ff; + + r = r << 6 | r >> 4; + g = g << 6 | g >> 4; + b = b << 6 | b >> 4; + + return 0xffffULL << 48 | r << 32 | g << 16 | b; +} + +static uint32_t +fetch_pixel_a8r8g8b8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + return READ (image, (uint32_t *)bits + offset); +} + +static uint32_t +fetch_pixel_x8r8g8b8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + + return READ (image, (uint32_t *)bits + offset) | 0xff000000; +} + +static uint32_t +fetch_pixel_a8b8g8r8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint32_t *)bits + offset); + + return ((pixel & 0xff000000) | + ((pixel >> 16) & 0xff) | + (pixel & 0x0000ff00) | + ((pixel & 0xff) << 16)); +} + +static uint32_t +fetch_pixel_x8b8g8r8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint32_t *)bits + offset); + + return ((0xff000000) | + ((pixel >> 16) & 0xff) | + (pixel & 0x0000ff00) | + ((pixel & 0xff) << 16)); +} + +static uint32_t +fetch_pixel_b8g8r8a8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint32_t *)bits + offset); + + return ((pixel & 0xff000000) >> 24 | + (pixel & 0x00ff0000) >> 8 | + (pixel & 0x0000ff00) << 8 | + (pixel & 0x000000ff) << 24); +} + +static uint32_t +fetch_pixel_b8g8r8x8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint32_t *)bits + offset); + + return ((0xff000000) | + (pixel & 0xff000000) >> 24 | + (pixel & 0x00ff0000) >> 8 | + (pixel & 0x0000ff00) << 8); +} + +static uint32_t +fetch_pixel_x14r6g6b6 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint32_t *) bits + offset); + uint32_t r, g, b; + + r = ((pixel & 0x3f000) << 6) | ((pixel & 0x30000)); + g = ((pixel & 0x00fc0) << 4) | ((pixel & 0x00c00) >> 2); + b = ((pixel & 0x0003f) << 2) | ((pixel & 0x00030) >> 4); + + return 0xff000000 | r | g | b; +} + +static uint32_t +fetch_pixel_r8g8b8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint8_t *pixel = ((uint8_t *) bits) + (offset * 3); + +#ifdef WORDS_BIGENDIAN + return (0xff000000 | + (READ (image, pixel + 0) << 16) | + (READ (image, pixel + 1) << 8) | + (READ (image, pixel + 2))); +#else + return (0xff000000 | + (READ (image, pixel + 2) << 16) | + (READ (image, pixel + 1) << 8) | + (READ (image, pixel + 0))); +#endif +} + +static uint32_t +fetch_pixel_b8g8r8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint8_t *pixel = ((uint8_t *) bits) + (offset * 3); +#ifdef WORDS_BIGENDIAN + return (0xff000000 | + (READ (image, pixel + 2) << 16) | + (READ (image, pixel + 1) << 8) | + (READ (image, pixel + 0))); +#else + return (0xff000000 | + (READ (image, pixel + 0) << 16) | + (READ (image, pixel + 1) << 8) | + (READ (image, pixel + 2))); +#endif +} + +static uint32_t +fetch_pixel_r5g6b5 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t r, g, b; + + r = ((pixel & 0xf800) | ((pixel & 0xe000) >> 5)) << 8; + g = ((pixel & 0x07e0) | ((pixel & 0x0600) >> 6)) << 5; + b = ((pixel & 0x001c) | ((pixel & 0x001f) << 5)) >> 2; + + return (0xff000000 | r | g | b); +} + +static uint32_t +fetch_pixel_b5g6r5 (bits_image_t *image, + int offset, + int line) +{ + uint32_t r, g, b; + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + + b = ((pixel & 0xf800) | ((pixel & 0xe000) >> 5)) >> 8; + g = ((pixel & 0x07e0) | ((pixel & 0x0600) >> 6)) << 5; + r = ((pixel & 0x001c) | ((pixel & 0x001f) << 5)) << 14; + + return (0xff000000 | r | g | b); +} + +static uint32_t +fetch_pixel_a1r5g5b5 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t a, r, g, b; + + a = (uint32_t) ((uint8_t) (0 - ((pixel & 0x8000) >> 15))) << 24; + r = ((pixel & 0x7c00) | ((pixel & 0x7000) >> 5)) << 9; + g = ((pixel & 0x03e0) | ((pixel & 0x0380) >> 5)) << 6; + b = ((pixel & 0x001c) | ((pixel & 0x001f) << 5)) >> 2; + + return (a | r | g | b); +} + +static uint32_t +fetch_pixel_x1r5g5b5 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t r, g, b; + + r = ((pixel & 0x7c00) | ((pixel & 0x7000) >> 5)) << 9; + g = ((pixel & 0x03e0) | ((pixel & 0x0380) >> 5)) << 6; + b = ((pixel & 0x001c) | ((pixel & 0x001f) << 5)) >> 2; + + return (0xff000000 | r | g | b); +} + +static uint32_t +fetch_pixel_a1b5g5r5 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t a, r, g, b; + + a = (uint32_t) ((uint8_t) (0 - ((pixel & 0x8000) >> 15))) << 24; + b = ((pixel & 0x7c00) | ((pixel & 0x7000) >> 5)) >> 7; + g = ((pixel & 0x03e0) | ((pixel & 0x0380) >> 5)) << 6; + r = ((pixel & 0x001c) | ((pixel & 0x001f) << 5)) << 14; + + return (a | r | g | b); +} + +static uint32_t +fetch_pixel_x1b5g5r5 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t r, g, b; + + b = ((pixel & 0x7c00) | ((pixel & 0x7000) >> 5)) >> 7; + g = ((pixel & 0x03e0) | ((pixel & 0x0380) >> 5)) << 6; + r = ((pixel & 0x001c) | ((pixel & 0x001f) << 5)) << 14; + + return (0xff000000 | r | g | b); +} + +static uint32_t +fetch_pixel_a4r4g4b4 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t a, r, g, b; + + a = ((pixel & 0xf000) | ((pixel & 0xf000) >> 4)) << 16; + r = ((pixel & 0x0f00) | ((pixel & 0x0f00) >> 4)) << 12; + g = ((pixel & 0x00f0) | ((pixel & 0x00f0) >> 4)) << 8; + b = ((pixel & 0x000f) | ((pixel & 0x000f) << 4)); + + return (a | r | g | b); +} + +static uint32_t +fetch_pixel_x4r4g4b4 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t r, g, b; + + r = ((pixel & 0x0f00) | ((pixel & 0x0f00) >> 4)) << 12; + g = ((pixel & 0x00f0) | ((pixel & 0x00f0) >> 4)) << 8; + b = ((pixel & 0x000f) | ((pixel & 0x000f) << 4)); + + return (0xff000000 | r | g | b); +} + +static uint32_t +fetch_pixel_a4b4g4r4 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t a, r, g, b; + + a = ((pixel & 0xf000) | ((pixel & 0xf000) >> 4)) << 16; + b = ((pixel & 0x0f00) | ((pixel & 0x0f00) >> 4)) >> 4; + g = ((pixel & 0x00f0) | ((pixel & 0x00f0) >> 4)) << 8; + r = ((pixel & 0x000f) | ((pixel & 0x000f) << 4)) << 16; + + return (a | r | g | b); +} + +static uint32_t +fetch_pixel_x4b4g4r4 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint16_t *) bits + offset); + uint32_t r, g, b; + + b = ((pixel & 0x0f00) | ((pixel & 0x0f00) >> 4)) >> 4; + g = ((pixel & 0x00f0) | ((pixel & 0x00f0) >> 4)) << 8; + r = ((pixel & 0x000f) | ((pixel & 0x000f) << 4)) << 16; + + return (0xff000000 | r | g | b); +} + +static uint32_t +fetch_pixel_a8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint8_t *) bits + offset); + + return pixel << 24; +} + +static uint32_t +fetch_pixel_r3g3b2 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint8_t *) bits + offset); + uint32_t r, g, b; + + r = ((pixel & 0xe0) | + ((pixel & 0xe0) >> 3) | + ((pixel & 0xc0) >> 6)) << 16; + + g = ((pixel & 0x1c) | + ((pixel & 0x18) >> 3) | + ((pixel & 0x1c) << 3)) << 8; + + b = (((pixel & 0x03) ) | + ((pixel & 0x03) << 2) | + ((pixel & 0x03) << 4) | + ((pixel & 0x03) << 6)); + + return (0xff000000 | r | g | b); +} + +static uint32_t +fetch_pixel_b2g3r3 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t p = READ (image, (uint8_t *) bits + offset); + uint32_t r, g, b; + + b = p & 0xc0; + b |= b >> 2; + b |= b >> 4; + b &= 0xff; + + g = (p & 0x38) << 10; + g |= g >> 3; + g |= g >> 6; + g &= 0xff00; + + r = (p & 0x7) << 21; + r |= r >> 3; + r |= r >> 6; + r &= 0xff0000; + + return 0xff000000 | r | g | b; +} + +static uint32_t +fetch_pixel_a2r2g2b2 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint8_t *) bits + offset); + uint32_t a, r, g, b; + + a = ((pixel & 0xc0) * 0x55) << 18; + r = ((pixel & 0x30) * 0x55) << 12; + g = ((pixel & 0x0c) * 0x55) << 6; + b = ((pixel & 0x03) * 0x55); + + return a | r | g | b; +} + +static uint32_t +fetch_pixel_a2b2g2r2 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint8_t *) bits + offset); + uint32_t a, r, g, b; + + a = ((pixel & 0xc0) * 0x55) << 18; + b = ((pixel & 0x30) * 0x55) >> 4; + g = ((pixel & 0x0c) * 0x55) << 6; + r = ((pixel & 0x03) * 0x55) << 16; + + return a | r | g | b; +} + +static uint32_t +fetch_pixel_c8 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint8_t *) bits + offset); + const pixman_indexed_t * indexed = image->indexed; + + return indexed->rgba[pixel]; +} + +static uint32_t +fetch_pixel_x4a4 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, (uint8_t *) bits + offset); + + return ((pixel & 0xf) | ((pixel & 0xf) << 4)) << 24; +} + +static uint32_t +fetch_pixel_a4 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = FETCH_4 (image, bits, offset); + + pixel |= pixel << 4; + return pixel << 24; +} + +static uint32_t +fetch_pixel_r1g2b1 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = FETCH_4 (image, bits, offset); + uint32_t r, g, b; + + r = ((pixel & 0x8) * 0xff) << 13; + g = ((pixel & 0x6) * 0x55) << 7; + b = ((pixel & 0x1) * 0xff); + + return 0xff000000 | r | g | b; +} + +static uint32_t +fetch_pixel_b1g2r1 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = FETCH_4 (image, bits, offset); + uint32_t r, g, b; + + b = ((pixel & 0x8) * 0xff) >> 3; + g = ((pixel & 0x6) * 0x55) << 7; + r = ((pixel & 0x1) * 0xff) << 16; + + return 0xff000000 | r | g | b; +} + +static uint32_t +fetch_pixel_a1r1g1b1 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = FETCH_4 (image, bits, offset); + uint32_t a, r, g, b; + + a = ((pixel & 0x8) * 0xff) << 21; + r = ((pixel & 0x4) * 0xff) << 14; + g = ((pixel & 0x2) * 0xff) << 7; + b = ((pixel & 0x1) * 0xff); + + return a | r | g | b; +} + +static uint32_t +fetch_pixel_a1b1g1r1 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = FETCH_4 (image, bits, offset); + uint32_t a, r, g, b; + + a = ((pixel & 0x8) * 0xff) << 21; + b = ((pixel & 0x4) * 0xff) >> 2; + g = ((pixel & 0x2) * 0xff) << 7; + r = ((pixel & 0x1) * 0xff) << 16; + + return a | r | g | b; +} + +static uint32_t +fetch_pixel_c4 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = FETCH_4 (image, bits, offset); + const pixman_indexed_t * indexed = image->indexed; + + return indexed->rgba[pixel]; +} + +static uint32_t +fetch_pixel_a1 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, bits + (offset >> 5)); + uint32_t a; + +#ifdef WORDS_BIGENDIAN + a = pixel >> (0x1f - (offset & 0x1f)); +#else + a = pixel >> (offset & 0x1f); +#endif + a = a & 1; + a |= a << 1; + a |= a << 2; + a |= a << 4; + + return a << 24; +} + +static uint32_t +fetch_pixel_g1 (bits_image_t *image, + int offset, + int line) +{ + uint32_t *bits = image->bits + line * image->rowstride; + uint32_t pixel = READ (image, bits + (offset >> 5)); + const pixman_indexed_t * indexed = image->indexed; + uint32_t a; + +#ifdef WORDS_BIGENDIAN + a = pixel >> (0x1f - (offset & 0x1f)); +#else + a = pixel >> (offset & 0x1f); +#endif + a = a & 1; + + return indexed->rgba[a]; +} + +static uint32_t +fetch_pixel_yuy2 (bits_image_t *image, + int offset, + int line) +{ + const uint32_t *bits = image->bits + image->rowstride * line; + + int16_t y, u, v; + int32_t r, g, b; + + y = ((uint8_t *) bits)[offset << 1] - 16; + u = ((uint8_t *) bits)[((offset << 1) & - 4) + 1] - 128; + v = ((uint8_t *) bits)[((offset << 1) & - 4) + 3] - 128; + + /* R = 1.164(Y - 16) + 1.596(V - 128) */ + r = 0x012b27 * y + 0x019a2e * v; + + /* G = 1.164(Y - 16) - 0.813(V - 128) - 0.391(U - 128) */ + g = 0x012b27 * y - 0x00d0f2 * v - 0x00647e * u; + + /* B = 1.164(Y - 16) + 2.018(U - 128) */ + b = 0x012b27 * y + 0x0206a2 * u; + + return 0xff000000 | + (r >= 0 ? r < 0x1000000 ? r & 0xff0000 : 0xff0000 : 0) | + (g >= 0 ? g < 0x1000000 ? (g >> 8) & 0x00ff00 : 0x00ff00 : 0) | + (b >= 0 ? b < 0x1000000 ? (b >> 16) & 0x0000ff : 0x0000ff : 0); +} + +static uint32_t +fetch_pixel_yv12 (bits_image_t *image, + int offset, + int line) +{ + YV12_SETUP (image); + int16_t y = YV12_Y (line)[offset] - 16; + int16_t u = YV12_U (line)[offset >> 1] - 128; + int16_t v = YV12_V (line)[offset >> 1] - 128; + int32_t r, g, b; + + /* R = 1.164(Y - 16) + 1.596(V - 128) */ + r = 0x012b27 * y + 0x019a2e * v; + + /* G = 1.164(Y - 16) - 0.813(V - 128) - 0.391(U - 128) */ + g = 0x012b27 * y - 0x00d0f2 * v - 0x00647e * u; + + /* B = 1.164(Y - 16) + 2.018(U - 128) */ + b = 0x012b27 * y + 0x0206a2 * u; + + return 0xff000000 | + (r >= 0 ? r < 0x1000000 ? r & 0xff0000 : 0xff0000 : 0) | + (g >= 0 ? g < 0x1000000 ? (g >> 8) & 0x00ff00 : 0x00ff00 : 0) | + (b >= 0 ? b < 0x1000000 ? (b >> 16) & 0x0000ff : 0x0000ff : 0); +} + +/*********************************** Store ************************************/ + +#define SPLIT_A(v) \ + uint32_t a = ((v) >> 24), \ + r = ((v) >> 16) & 0xff, \ + g = ((v) >> 8) & 0xff, \ + b = (v) & 0xff + +#define SPLIT(v) \ + uint32_t r = ((v) >> 16) & 0xff, \ + g = ((v) >> 8) & 0xff, \ + b = (v) & 0xff + +static void +store_scanline_a2r10g10b10 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *v) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint32_t *pixel = bits + x; + uint64_t *values = (uint64_t *)v; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, + ((values[i] >> 32) & 0xc0000000) | + ((values[i] >> 18) & 0x3ff00000) | + ((values[i] >> 12) & 0xffc00) | + ((values[i] >> 6) & 0x3ff)); + } +} + +static void +store_scanline_x2r10g10b10 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *v) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint64_t *values = (uint64_t *)v; + uint32_t *pixel = bits + x; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, + ((values[i] >> 18) & 0x3ff00000) | + ((values[i] >> 12) & 0xffc00) | + ((values[i] >> 6) & 0x3ff)); + } +} + +static void +store_scanline_a2b10g10r10 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *v) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint32_t *pixel = bits + x; + uint64_t *values = (uint64_t *)v; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, + ((values[i] >> 32) & 0xc0000000) | + ((values[i] >> 38) & 0x3ff) | + ((values[i] >> 12) & 0xffc00) | + ((values[i] << 14) & 0x3ff00000)); + } +} + +static void +store_scanline_x2b10g10r10 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *v) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint64_t *values = (uint64_t *)v; + uint32_t *pixel = bits + x; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, + ((values[i] >> 38) & 0x3ff) | + ((values[i] >> 12) & 0xffc00) | + ((values[i] << 14) & 0x3ff00000)); + } +} + +static void +store_scanline_a8r8g8b8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + + MEMCPY_WRAPPED (image, ((uint32_t *)bits) + x, values, + width * sizeof(uint32_t)); +} + +static void +store_scanline_x8r8g8b8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint32_t *pixel = (uint32_t *)bits + x; + int i; + + for (i = 0; i < width; ++i) + WRITE (image, pixel++, values[i] & 0xffffff); +} + +static void +store_scanline_a8b8g8r8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint32_t *pixel = (uint32_t *)bits + x; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, + (values[i] & 0xff00ff00) | + ((values[i] >> 16) & 0xff) | + ((values[i] & 0xff) << 16)); + } +} + +static void +store_scanline_x8b8g8r8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint32_t *pixel = (uint32_t *)bits + x; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, + (values[i] & 0x0000ff00) | + ((values[i] >> 16) & 0xff) | + ((values[i] & 0xff) << 16)); + } +} + +static void +store_scanline_b8g8r8a8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint32_t *pixel = (uint32_t *)bits + x; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, + ((values[i] >> 24) & 0x000000ff) | + ((values[i] >> 8) & 0x0000ff00) | + ((values[i] << 8) & 0x00ff0000) | + ((values[i] << 24) & 0xff000000)); + } +} + +static void +store_scanline_b8g8r8x8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint32_t *pixel = (uint32_t *)bits + x; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, + ((values[i] >> 8) & 0x0000ff00) | + ((values[i] << 8) & 0x00ff0000) | + ((values[i] << 24) & 0xff000000)); + } +} + +static void +store_scanline_x14r6g6b6 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint32_t *pixel = ((uint32_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = values[i]; + uint32_t r, g, b; + + r = (s & 0xfc0000) >> 6; + g = (s & 0x00fc00) >> 4; + b = (s & 0x0000fc) >> 2; + + WRITE (image, pixel++, r | g | b); + } +} + +static void +store_scanline_r8g8b8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + 3 * x; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t val = values[i]; + +#ifdef WORDS_BIGENDIAN + WRITE (image, pixel++, (val & 0x00ff0000) >> 16); + WRITE (image, pixel++, (val & 0x0000ff00) >> 8); + WRITE (image, pixel++, (val & 0x000000ff) >> 0); +#else + WRITE (image, pixel++, (val & 0x000000ff) >> 0); + WRITE (image, pixel++, (val & 0x0000ff00) >> 8); + WRITE (image, pixel++, (val & 0x00ff0000) >> 16); +#endif + } +} + +static void +store_scanline_b8g8r8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + 3 * x; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t val = values[i]; + +#ifdef WORDS_BIGENDIAN + WRITE (image, pixel++, (val & 0x000000ff) >> 0); + WRITE (image, pixel++, (val & 0x0000ff00) >> 8); + WRITE (image, pixel++, (val & 0x00ff0000) >> 16); +#else + WRITE (image, pixel++, (val & 0x00ff0000) >> 16); + WRITE (image, pixel++, (val & 0x0000ff00) >> 8); + WRITE (image, pixel++, (val & 0x000000ff) >> 0); +#endif + } +} + +static void +store_scanline_r5g6b5 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = values[i]; + + WRITE (image, pixel++, + ((s >> 3) & 0x001f) | + ((s >> 5) & 0x07e0) | + ((s >> 8) & 0xf800)); + } +} + +static void +store_scanline_b5g6r5 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT (values[i]); + + WRITE (image, pixel++, + ((b << 8) & 0xf800) | + ((g << 3) & 0x07e0) | + ((r >> 3) )); + } +} + +static void +store_scanline_a1r5g5b5 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT_A (values[i]); + + WRITE (image, pixel++, + ((a << 8) & 0x8000) | + ((r << 7) & 0x7c00) | + ((g << 2) & 0x03e0) | + ((b >> 3) )); + } +} + +static void +store_scanline_x1r5g5b5 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT (values[i]); + + WRITE (image, pixel++, + ((r << 7) & 0x7c00) | + ((g << 2) & 0x03e0) | + ((b >> 3) )); + } +} + +static void +store_scanline_a1b5g5r5 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT_A (values[i]); + + WRITE (image, pixel++, + ((a << 8) & 0x8000) | + ((b << 7) & 0x7c00) | + ((g << 2) & 0x03e0) | + ((r >> 3) )); + } +} + +static void +store_scanline_x1b5g5r5 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT (values[i]); + + WRITE (image, pixel++, ((b << 7) & 0x7c00) | + ((g << 2) & 0x03e0) | + ((r >> 3) )); + } +} + +static void +store_scanline_a4r4g4b4 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT_A (values[i]); + + WRITE (image, pixel++, + ((a << 8) & 0xf000) | + ((r << 4) & 0x0f00) | + ((g ) & 0x00f0) | + ((b >> 4) )); + } +} + +static void +store_scanline_x4r4g4b4 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT (values[i]); + + WRITE (image, pixel++, + ((r << 4) & 0x0f00) | + ((g ) & 0x00f0) | + ((b >> 4) )); + } +} + +static void +store_scanline_a4b4g4r4 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT_A (values[i]); + WRITE (image, pixel++, ((a << 8) & 0xf000) | + ((b << 4) & 0x0f00) | + ((g ) & 0x00f0) | + ((r >> 4) )); + } +} + +static void +store_scanline_x4b4g4r4 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint16_t *pixel = ((uint16_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT (values[i]); + + WRITE (image, pixel++, + ((b << 4) & 0x0f00) | + ((g ) & 0x00f0) | + ((r >> 4) )); + } +} + +static void +store_scanline_a8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + WRITE (image, pixel++, values[i] >> 24); + } +} + +static void +store_scanline_r3g3b2 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT (values[i]); + + WRITE (image, pixel++, + ((r ) & 0xe0) | + ((g >> 3) & 0x1c) | + ((b >> 6) )); + } +} + +static void +store_scanline_b2g3r3 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT (values[i]); + + WRITE (image, pixel++, + ((b ) & 0xc0) | + ((g >> 2) & 0x38) | + ((r >> 5) )); + } +} + +static void +store_scanline_a2r2g2b2 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT_A (values[i]); + + WRITE (image, pixel++, + ((a ) & 0xc0) | + ((r >> 2) & 0x30) | + ((g >> 4) & 0x0c) | + ((b >> 6) )); + } +} + +static void +store_scanline_a2b2g2r2 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + { + SPLIT_A (values[i]); + + WRITE (image, pixel++, + ((a ) & 0xc0) | + ((b >> 2) & 0x30) | + ((g >> 4) & 0x0c) | + ((r >> 6) )); + } +} + +static void +store_scanline_c8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + x; + const pixman_indexed_t *indexed = image->indexed; + int i; + + for (i = 0; i < width; ++i) + WRITE (image, pixel++, RGB24_TO_ENTRY (indexed,values[i])); +} + +static void +store_scanline_g8 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + x; + const pixman_indexed_t *indexed = image->indexed; + int i; + + for (i = 0; i < width; ++i) + WRITE (image, pixel++, RGB24_TO_ENTRY_Y (indexed,values[i])); +} + +static void +store_scanline_x4a4 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + uint8_t *pixel = ((uint8_t *) bits) + x; + int i; + + for (i = 0; i < width; ++i) + WRITE (image, pixel++, values[i] >> 28); +} + +#define STORE_8(img,l,o,v) (WRITE (img, (uint8_t *)(l) + ((o) >> 3), (v))) +#ifdef WORDS_BIGENDIAN + +#define STORE_4(img,l,o,v) \ + do \ + { \ + int bo = 4 * (o); \ + int v4 = (v) & 0x0f; \ + \ + STORE_8 (img, l, bo, ( \ + bo & 4 ? \ + (FETCH_8 (img, l, bo) & 0xf0) | (v4) : \ + (FETCH_8 (img, l, bo) & 0x0f) | (v4 << 4))); \ + } while (0) +#else + +#define STORE_4(img,l,o,v) \ + do \ + { \ + int bo = 4 * (o); \ + int v4 = (v) & 0x0f; \ + \ + STORE_8 (img, l, bo, ( \ + bo & 4 ? \ + (FETCH_8 (img, l, bo) & 0x0f) | (v4 << 4) : \ + (FETCH_8 (img, l, bo) & 0xf0) | (v4))); \ + } while (0) +#endif + +static void +store_scanline_a4 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + int i; + + for (i = 0; i < width; ++i) + STORE_4 (image, bits, i + x, values[i] >> 28); +} + +static void +store_scanline_r1g2b1 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t pixel; + + SPLIT (values[i]); + pixel = (((r >> 4) & 0x8) | + ((g >> 5) & 0x6) | + ((b >> 7) )); + STORE_4 (image, bits, i + x, pixel); + } +} + +static void +store_scanline_b1g2r1 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t pixel; + + SPLIT (values[i]); + pixel = (((b >> 4) & 0x8) | + ((g >> 5) & 0x6) | + ((r >> 7) )); + STORE_4 (image, bits, i + x, pixel); + } +} + +static void +store_scanline_a1r1g1b1 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t pixel; + + SPLIT_A (values[i]); + pixel = (((a >> 4) & 0x8) | + ((r >> 5) & 0x4) | + ((g >> 6) & 0x2) | + ((b >> 7) )); + + STORE_4 (image, bits, i + x, pixel); + } +} + +static void +store_scanline_a1b1g1r1 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t pixel; + + SPLIT_A (values[i]); + pixel = (((a >> 4) & 0x8) | + ((b >> 5) & 0x4) | + ((g >> 6) & 0x2) | + ((r >> 7) )); + + STORE_4 (image, bits, i + x, pixel); + } +} + +static void +store_scanline_c4 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + const pixman_indexed_t *indexed = image->indexed; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t pixel; + + pixel = RGB24_TO_ENTRY (indexed, values[i]); + STORE_4 (image, bits, i + x, pixel); + } +} + +static void +store_scanline_g4 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + const pixman_indexed_t *indexed = image->indexed; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t pixel; + + pixel = RGB24_TO_ENTRY_Y (indexed, values[i]); + STORE_4 (image, bits, i + x, pixel); + } +} + +static void +store_scanline_a1 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t *pixel = ((uint32_t *) bits) + ((i + x) >> 5); + uint32_t mask, v; + +#ifdef WORDS_BIGENDIAN + mask = 1 << (0x1f - ((i + x) & 0x1f)); +#else + mask = 1 << ((i + x) & 0x1f); +#endif + v = values[i] & 0x80000000 ? mask : 0; + + WRITE (image, pixel, (READ (image, pixel) & ~mask) | v); + } +} + +static void +store_scanline_g1 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *bits = image->bits + image->rowstride * y; + const pixman_indexed_t *indexed = image->indexed; + int i; + + for (i = 0; i < width; ++i) + { + uint32_t *pixel = ((uint32_t *) bits) + ((i + x) >> 5); + uint32_t mask, v; + +#ifdef WORDS_BIGENDIAN + mask = 1 << (0x1f - ((i + x) & 0x1f)); +#else + mask = 1 << ((i + x) & 0x1f); +#endif + v = RGB24_TO_ENTRY_Y (indexed, values[i]) & 0x1 ? mask : 0; + + WRITE (image, pixel, (READ (image, pixel) & ~mask) | v); + } +} + +/* + * Contracts a 64bpp image to 32bpp and then stores it using a regular 32-bit + * store proc. Despite the type, this function expects a uint64_t buffer. + */ +static void +store_scanline_generic_64 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values) +{ + uint32_t *argb8_pixels; + + assert (image->common.type == BITS); + + argb8_pixels = pixman_malloc_ab (width, sizeof(uint32_t)); + if (!argb8_pixels) + return; + + /* Contract the scanline. We could do this in place if values weren't + * const. + */ + pixman_contract (argb8_pixels, (uint64_t *)values, width); + + image->store_scanline_32 (image, x, y, width, argb8_pixels); + + free (argb8_pixels); +} + +/* Despite the type, this function expects both buffer + * and mask to be uint64_t + */ +static void +fetch_scanline_generic_64 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + pixman_format_code_t format; + + /* Fetch the pixels into the first half of buffer and then expand them in + * place. + */ + image->bits.fetch_scanline_32 (image, x, y, width, buffer, NULL); + + format = image->bits.format; + if (PIXMAN_FORMAT_TYPE (format) == PIXMAN_TYPE_COLOR || + PIXMAN_FORMAT_TYPE (format) == PIXMAN_TYPE_GRAY) + { + /* Indexed formats are mapped to a8r8g8b8 with full + * precision, so when expanding we shouldn't correct + * for the width of the channels + */ + + format = PIXMAN_a8r8g8b8; + } + + pixman_expand ((uint64_t *)buffer, buffer, format, width); +} + +/* Despite the type, this function expects a uint64_t *buffer */ +static uint64_t +fetch_pixel_generic_64 (bits_image_t *image, + int offset, + int line) +{ + uint32_t pixel32 = image->fetch_pixel_32 (image, offset, line); + uint64_t result; + pixman_format_code_t format; + + format = image->format; + if (PIXMAN_FORMAT_TYPE (format) == PIXMAN_TYPE_COLOR || + PIXMAN_FORMAT_TYPE (format) == PIXMAN_TYPE_GRAY) + { + /* Indexed formats are mapped to a8r8g8b8 with full + * precision, so when expanding we shouldn't correct + * for the width of the channels + */ + + format = PIXMAN_a8r8g8b8; + } + + pixman_expand ((uint64_t *)&result, &pixel32, format, 1); + + return result; +} + +/* + * XXX: The transformed fetch path only works at 32-bpp so far. When all + * paths have wide versions, this can be removed. + * + * WARNING: This function loses precision! + */ +static uint32_t +fetch_pixel_generic_lossy_32 (bits_image_t *image, + int offset, + int line) +{ + uint64_t pixel64 = image->fetch_pixel_64 (image, offset, line); + uint32_t result; + + pixman_contract (&result, &pixel64, 1); + + return result; +} + +typedef struct +{ + pixman_format_code_t format; + fetch_scanline_t fetch_scanline_32; + fetch_scanline_t fetch_scanline_64; + fetch_pixel_32_t fetch_pixel_32; + fetch_pixel_64_t fetch_pixel_64; + store_scanline_t store_scanline_32; + store_scanline_t store_scanline_64; +} format_info_t; + +#define FORMAT_INFO(format) \ + { \ + PIXMAN_ ## format, \ + fetch_scanline_ ## format, \ + fetch_scanline_generic_64, \ + fetch_pixel_ ## format, fetch_pixel_generic_64, \ + store_scanline_ ## format, store_scanline_generic_64 \ + } + +static const format_info_t accessors[] = +{ +/* 32 bpp formats */ + FORMAT_INFO (a8r8g8b8), + FORMAT_INFO (x8r8g8b8), + FORMAT_INFO (a8b8g8r8), + FORMAT_INFO (x8b8g8r8), + FORMAT_INFO (b8g8r8a8), + FORMAT_INFO (b8g8r8x8), + FORMAT_INFO (x14r6g6b6), + +/* 24bpp formats */ + FORMAT_INFO (r8g8b8), + FORMAT_INFO (b8g8r8), + +/* 16bpp formats */ + FORMAT_INFO (r5g6b5), + FORMAT_INFO (b5g6r5), + + FORMAT_INFO (a1r5g5b5), + FORMAT_INFO (x1r5g5b5), + FORMAT_INFO (a1b5g5r5), + FORMAT_INFO (x1b5g5r5), + FORMAT_INFO (a4r4g4b4), + FORMAT_INFO (x4r4g4b4), + FORMAT_INFO (a4b4g4r4), + FORMAT_INFO (x4b4g4r4), + +/* 8bpp formats */ + FORMAT_INFO (a8), + FORMAT_INFO (r3g3b2), + FORMAT_INFO (b2g3r3), + FORMAT_INFO (a2r2g2b2), + FORMAT_INFO (a2b2g2r2), + + FORMAT_INFO (c8), + +#define fetch_scanline_g8 fetch_scanline_c8 +#define fetch_pixel_g8 fetch_pixel_c8 + FORMAT_INFO (g8), + +#define fetch_scanline_x4c4 fetch_scanline_c8 +#define fetch_pixel_x4c4 fetch_pixel_c8 +#define store_scanline_x4c4 store_scanline_c8 + FORMAT_INFO (x4c4), + +#define fetch_scanline_x4g4 fetch_scanline_c8 +#define fetch_pixel_x4g4 fetch_pixel_c8 +#define store_scanline_x4g4 store_scanline_g8 + FORMAT_INFO (x4g4), + + FORMAT_INFO (x4a4), + +/* 4bpp formats */ + FORMAT_INFO (a4), + FORMAT_INFO (r1g2b1), + FORMAT_INFO (b1g2r1), + FORMAT_INFO (a1r1g1b1), + FORMAT_INFO (a1b1g1r1), + + FORMAT_INFO (c4), + +#define fetch_scanline_g4 fetch_scanline_c4 +#define fetch_pixel_g4 fetch_pixel_c4 + FORMAT_INFO (g4), + +/* 1bpp formats */ + FORMAT_INFO (a1), + FORMAT_INFO (g1), + +/* Wide formats */ + + { PIXMAN_a2r10g10b10, + NULL, fetch_scanline_a2r10g10b10, + fetch_pixel_generic_lossy_32, fetch_pixel_a2r10g10b10, + NULL, store_scanline_a2r10g10b10 }, + + { PIXMAN_x2r10g10b10, + NULL, fetch_scanline_x2r10g10b10, + fetch_pixel_generic_lossy_32, fetch_pixel_x2r10g10b10, + NULL, store_scanline_x2r10g10b10 }, + + { PIXMAN_a2b10g10r10, + NULL, fetch_scanline_a2b10g10r10, + fetch_pixel_generic_lossy_32, fetch_pixel_a2b10g10r10, + NULL, store_scanline_a2b10g10r10 }, + + { PIXMAN_x2b10g10r10, + NULL, fetch_scanline_x2b10g10r10, + fetch_pixel_generic_lossy_32, fetch_pixel_x2b10g10r10, + NULL, store_scanline_x2b10g10r10 }, + +/* YUV formats */ + { PIXMAN_yuy2, + fetch_scanline_yuy2, fetch_scanline_generic_64, + fetch_pixel_yuy2, fetch_pixel_generic_64, + NULL, NULL }, + + { PIXMAN_yv12, + fetch_scanline_yv12, fetch_scanline_generic_64, + fetch_pixel_yv12, fetch_pixel_generic_64, + NULL, NULL }, + + { PIXMAN_null }, +}; + +static void +setup_accessors (bits_image_t *image) +{ + const format_info_t *info = accessors; + + while (info->format != PIXMAN_null) + { + if (info->format == image->format) + { + image->fetch_scanline_32 = info->fetch_scanline_32; + image->fetch_scanline_64 = info->fetch_scanline_64; + image->fetch_pixel_32 = info->fetch_pixel_32; + image->fetch_pixel_64 = info->fetch_pixel_64; + image->store_scanline_32 = info->store_scanline_32; + image->store_scanline_64 = info->store_scanline_64; + + return; + } + + info++; + } +} + +#ifndef PIXMAN_FB_ACCESSORS +void +_pixman_bits_image_setup_accessors_accessors (bits_image_t *image); + +void +_pixman_bits_image_setup_accessors (bits_image_t *image) +{ + if (image->read_func || image->write_func) + _pixman_bits_image_setup_accessors_accessors (image); + else + setup_accessors (image); +} + +#else + +void +_pixman_bits_image_setup_accessors_accessors (bits_image_t *image) +{ + setup_accessors (image); +} + +#endif diff --git a/programs/develop/libraries/pixman/pixman-accessor.h b/programs/develop/libraries/pixman/pixman-accessor.h new file mode 100644 index 0000000000..90c8ea7b77 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-accessor.h @@ -0,0 +1,40 @@ +#ifdef PIXMAN_FB_ACCESSORS + +#define ACCESS(sym) sym##_accessors + +#define READ(img, ptr) \ + (((bits_image_t *)(img))->read_func ((ptr), sizeof(*(ptr)))) +#define WRITE(img, ptr,val) \ + (((bits_image_t *)(img))->write_func ((ptr), (val), sizeof (*(ptr)))) + +#define MEMCPY_WRAPPED(img, dst, src, size) \ + do { \ + size_t _i; \ + uint8_t *_dst = (uint8_t*)(dst), *_src = (uint8_t*)(src); \ + for(_i = 0; _i < size; _i++) { \ + WRITE((img), _dst +_i, READ((img), _src + _i)); \ + } \ + } while (0) + +#define MEMSET_WRAPPED(img, dst, val, size) \ + do { \ + size_t _i; \ + uint8_t *_dst = (uint8_t*)(dst); \ + for(_i = 0; _i < (size_t) size; _i++) { \ + WRITE((img), _dst +_i, (val)); \ + } \ + } while (0) + +#else + +#define ACCESS(sym) sym + +#define READ(img, ptr) (*(ptr)) +#define WRITE(img, ptr, val) (*(ptr) = (val)) +#define MEMCPY_WRAPPED(img, dst, src, size) \ + memcpy(dst, src, size) +#define MEMSET_WRAPPED(img, dst, val, size) \ + memset(dst, val, size) + +#endif + diff --git a/programs/develop/libraries/pixman/pixman-bits-image.c b/programs/develop/libraries/pixman/pixman-bits-image.c new file mode 100644 index 0000000000..b27a732848 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-bits-image.c @@ -0,0 +1,1331 @@ +/* + * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc. + * 2005 Lars Knoll & Zack Rusin, Trolltech + * 2008 Aaron Plattner, NVIDIA Corporation + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007, 2009 Red Hat, Inc. + * Copyright © 2008 André Tupinambá + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include +#include +#include +#include "pixman-private.h" +#include "pixman-combine32.h" + +/* Store functions */ +void +_pixman_image_store_scanline_32 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *buffer) +{ + image->store_scanline_32 (image, x, y, width, buffer); + + if (image->common.alpha_map) + { + x -= image->common.alpha_origin_x; + y -= image->common.alpha_origin_y; + + image->common.alpha_map->store_scanline_32 ( + image->common.alpha_map, x, y, width, buffer); + } +} + +void +_pixman_image_store_scanline_64 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *buffer) +{ + image->store_scanline_64 (image, x, y, width, buffer); + + if (image->common.alpha_map) + { + x -= image->common.alpha_origin_x; + y -= image->common.alpha_origin_y; + + image->common.alpha_map->store_scanline_64 ( + image->common.alpha_map, x, y, width, buffer); + } +} + +/* Fetch functions */ + +static force_inline uint32_t +fetch_pixel_no_alpha (bits_image_t *image, + int x, int y, pixman_bool_t check_bounds) +{ + if (check_bounds && + (x < 0 || x >= image->width || y < 0 || y >= image->height)) + { + return 0; + } + + return image->fetch_pixel_32 (image, x, y); +} + +typedef uint32_t (* get_pixel_t) (bits_image_t *image, + int x, int y, pixman_bool_t check_bounds); + +static force_inline void +repeat (pixman_repeat_t repeat, int size, int *coord) +{ + switch (repeat) + { + case PIXMAN_REPEAT_NORMAL: + *coord = MOD (*coord, size); + break; + + case PIXMAN_REPEAT_PAD: + *coord = CLIP (*coord, 0, size - 1); + break; + + case PIXMAN_REPEAT_REFLECT: + *coord = MOD (*coord, size * 2); + + if (*coord >= size) + *coord = size * 2 - *coord - 1; + break; + + case PIXMAN_REPEAT_NONE: + break; + + default: + break; + } +} + +static force_inline uint32_t +bits_image_fetch_pixel_nearest (bits_image_t *image, + pixman_fixed_t x, + pixman_fixed_t y, + get_pixel_t get_pixel) +{ + int x0 = pixman_fixed_to_int (x - pixman_fixed_e); + int y0 = pixman_fixed_to_int (y - pixman_fixed_e); + + if (image->common.repeat != PIXMAN_REPEAT_NONE) + { + repeat (image->common.repeat, image->width, &x0); + repeat (image->common.repeat, image->height, &y0); + + return get_pixel (image, x0, y0, FALSE); + } + else + { + return get_pixel (image, x0, y0, TRUE); + } +} + +#if SIZEOF_LONG > 4 + +static force_inline uint32_t +bilinear_interpolation (uint32_t tl, uint32_t tr, + uint32_t bl, uint32_t br, + int distx, int disty) +{ + uint64_t distxy, distxiy, distixy, distixiy; + uint64_t tl64, tr64, bl64, br64; + uint64_t f, r; + + distxy = distx * disty; + distxiy = distx * (256 - disty); + distixy = (256 - distx) * disty; + distixiy = (256 - distx) * (256 - disty); + + /* Alpha and Blue */ + tl64 = tl & 0xff0000ff; + tr64 = tr & 0xff0000ff; + bl64 = bl & 0xff0000ff; + br64 = br & 0xff0000ff; + + f = tl64 * distixiy + tr64 * distxiy + bl64 * distixy + br64 * distxy; + r = f & 0x0000ff0000ff0000ull; + + /* Red and Green */ + tl64 = tl; + tl64 = ((tl64 << 16) & 0x000000ff00000000ull) | (tl64 & 0x0000ff00ull); + + tr64 = tr; + tr64 = ((tr64 << 16) & 0x000000ff00000000ull) | (tr64 & 0x0000ff00ull); + + bl64 = bl; + bl64 = ((bl64 << 16) & 0x000000ff00000000ull) | (bl64 & 0x0000ff00ull); + + br64 = br; + br64 = ((br64 << 16) & 0x000000ff00000000ull) | (br64 & 0x0000ff00ull); + + f = tl64 * distixiy + tr64 * distxiy + bl64 * distixy + br64 * distxy; + r |= ((f >> 16) & 0x000000ff00000000ull) | (f & 0xff000000ull); + + return (uint32_t)(r >> 16); +} + +#else + +static force_inline uint32_t +bilinear_interpolation (uint32_t tl, uint32_t tr, + uint32_t bl, uint32_t br, + int distx, int disty) +{ + int distxy, distxiy, distixy, distixiy; + uint32_t f, r; + + distxy = distx * disty; + distxiy = (distx << 8) - distxy; /* distx * (256 - disty) */ + distixy = (disty << 8) - distxy; /* disty * (256 - distx) */ + distixiy = + 256 * 256 - (disty << 8) - + (distx << 8) + distxy; /* (256 - distx) * (256 - disty) */ + + /* Blue */ + r = (tl & 0x000000ff) * distixiy + (tr & 0x000000ff) * distxiy + + (bl & 0x000000ff) * distixy + (br & 0x000000ff) * distxy; + + /* Green */ + f = (tl & 0x0000ff00) * distixiy + (tr & 0x0000ff00) * distxiy + + (bl & 0x0000ff00) * distixy + (br & 0x0000ff00) * distxy; + r |= f & 0xff000000; + + tl >>= 16; + tr >>= 16; + bl >>= 16; + br >>= 16; + r >>= 16; + + /* Red */ + f = (tl & 0x000000ff) * distixiy + (tr & 0x000000ff) * distxiy + + (bl & 0x000000ff) * distixy + (br & 0x000000ff) * distxy; + r |= f & 0x00ff0000; + + /* Alpha */ + f = (tl & 0x0000ff00) * distixiy + (tr & 0x0000ff00) * distxiy + + (bl & 0x0000ff00) * distixy + (br & 0x0000ff00) * distxy; + r |= f & 0xff000000; + + return r; +} + +#endif + +static force_inline uint32_t +bits_image_fetch_pixel_bilinear (bits_image_t *image, + pixman_fixed_t x, + pixman_fixed_t y, + get_pixel_t get_pixel) +{ + pixman_repeat_t repeat_mode = image->common.repeat; + int width = image->width; + int height = image->height; + int x1, y1, x2, y2; + uint32_t tl, tr, bl, br; + int32_t distx, disty; + + x1 = x - pixman_fixed_1 / 2; + y1 = y - pixman_fixed_1 / 2; + + distx = (x1 >> 8) & 0xff; + disty = (y1 >> 8) & 0xff; + + x1 = pixman_fixed_to_int (x1); + y1 = pixman_fixed_to_int (y1); + x2 = x1 + 1; + y2 = y1 + 1; + + if (repeat_mode != PIXMAN_REPEAT_NONE) + { + repeat (repeat_mode, width, &x1); + repeat (repeat_mode, height, &y1); + repeat (repeat_mode, width, &x2); + repeat (repeat_mode, height, &y2); + + tl = get_pixel (image, x1, y1, FALSE); + bl = get_pixel (image, x1, y2, FALSE); + tr = get_pixel (image, x2, y1, FALSE); + br = get_pixel (image, x2, y2, FALSE); + } + else + { + tl = get_pixel (image, x1, y1, TRUE); + tr = get_pixel (image, x2, y1, TRUE); + bl = get_pixel (image, x1, y2, TRUE); + br = get_pixel (image, x2, y2, TRUE); + } + + return bilinear_interpolation (tl, tr, bl, br, distx, disty); +} + +static void +bits_image_fetch_bilinear_no_repeat_8888 (pixman_image_t * ima, + int offset, + int line, + int width, + uint32_t * buffer, + const uint32_t * mask) +{ + bits_image_t *bits = &ima->bits; + pixman_fixed_t x_top, x_bottom, x; + pixman_fixed_t ux_top, ux_bottom, ux; + pixman_vector_t v; + uint32_t top_mask, bottom_mask; + uint32_t *top_row; + uint32_t *bottom_row; + uint32_t *end; + uint32_t zero[2] = { 0, 0 }; + uint32_t one = 1; + int y, y1, y2; + int disty; + int mask_inc; + int w; + + /* reference point is the center of the pixel */ + v.vector[0] = pixman_int_to_fixed (offset) + pixman_fixed_1 / 2; + v.vector[1] = pixman_int_to_fixed (line) + pixman_fixed_1 / 2; + v.vector[2] = pixman_fixed_1; + + if (!pixman_transform_point_3d (bits->common.transform, &v)) + return; + + ux = ux_top = ux_bottom = bits->common.transform->matrix[0][0]; + x = x_top = x_bottom = v.vector[0] - pixman_fixed_1/2; + + y = v.vector[1] - pixman_fixed_1/2; + disty = (y >> 8) & 0xff; + + /* Load the pointers to the first and second lines from the source + * image that bilinear code must read. + * + * The main trick in this code is about the check if any line are + * outside of the image; + * + * When I realize that a line (any one) is outside, I change + * the pointer to a dummy area with zeros. Once I change this, I + * must be sure the pointer will not change, so I set the + * variables to each pointer increments inside the loop. + */ + y1 = pixman_fixed_to_int (y); + y2 = y1 + 1; + + if (y1 < 0 || y1 >= bits->height) + { + top_row = zero; + x_top = 0; + ux_top = 0; + } + else + { + top_row = bits->bits + y1 * bits->rowstride; + x_top = x; + ux_top = ux; + } + + if (y2 < 0 || y2 >= bits->height) + { + bottom_row = zero; + x_bottom = 0; + ux_bottom = 0; + } + else + { + bottom_row = bits->bits + y2 * bits->rowstride; + x_bottom = x; + ux_bottom = ux; + } + + /* Instead of checking whether the operation uses the mast in + * each loop iteration, verify this only once and prepare the + * variables to make the code smaller inside the loop. + */ + if (!mask) + { + mask_inc = 0; + mask = &one; + } + else + { + /* If have a mask, prepare the variables to check it */ + mask_inc = 1; + } + + /* If both are zero, then the whole thing is zero */ + if (top_row == zero && bottom_row == zero) + { + memset (buffer, 0, width * sizeof (uint32_t)); + return; + } + else if (bits->format == PIXMAN_x8r8g8b8) + { + if (top_row == zero) + { + top_mask = 0; + bottom_mask = 0xff000000; + } + else if (bottom_row == zero) + { + top_mask = 0xff000000; + bottom_mask = 0; + } + else + { + top_mask = 0xff000000; + bottom_mask = 0xff000000; + } + } + else + { + top_mask = 0; + bottom_mask = 0; + } + + end = buffer + width; + + /* Zero fill to the left of the image */ + while (buffer < end && x < pixman_fixed_minus_1) + { + *buffer++ = 0; + x += ux; + x_top += ux_top; + x_bottom += ux_bottom; + mask += mask_inc; + } + + /* Left edge + */ + while (buffer < end && x < 0) + { + uint32_t tr, br; + int32_t distx; + + tr = top_row[pixman_fixed_to_int (x_top) + 1] | top_mask; + br = bottom_row[pixman_fixed_to_int (x_bottom) + 1] | bottom_mask; + + distx = (x >> 8) & 0xff; + + *buffer++ = bilinear_interpolation (0, tr, 0, br, distx, disty); + + x += ux; + x_top += ux_top; + x_bottom += ux_bottom; + mask += mask_inc; + } + + /* Main part */ + w = pixman_int_to_fixed (bits->width - 1); + + while (buffer < end && x < w) + { + if (*mask) + { + uint32_t tl, tr, bl, br; + int32_t distx; + + tl = top_row [pixman_fixed_to_int (x_top)] | top_mask; + tr = top_row [pixman_fixed_to_int (x_top) + 1] | top_mask; + bl = bottom_row [pixman_fixed_to_int (x_bottom)] | bottom_mask; + br = bottom_row [pixman_fixed_to_int (x_bottom) + 1] | bottom_mask; + + distx = (x >> 8) & 0xff; + + *buffer = bilinear_interpolation (tl, tr, bl, br, distx, disty); + } + + buffer++; + x += ux; + x_top += ux_top; + x_bottom += ux_bottom; + mask += mask_inc; + } + + /* Right Edge */ + w = pixman_int_to_fixed (bits->width); + while (buffer < end && x < w) + { + if (*mask) + { + uint32_t tl, bl; + int32_t distx; + + tl = top_row [pixman_fixed_to_int (x_top)] | top_mask; + bl = bottom_row [pixman_fixed_to_int (x_bottom)] | bottom_mask; + + distx = (x >> 8) & 0xff; + + *buffer = bilinear_interpolation (tl, 0, bl, 0, distx, disty); + } + + buffer++; + x += ux; + x_top += ux_top; + x_bottom += ux_bottom; + mask += mask_inc; + } + + /* Zero fill to the left of the image */ + while (buffer < end) + *buffer++ = 0; +} + +static force_inline uint32_t +bits_image_fetch_pixel_convolution (bits_image_t *image, + pixman_fixed_t x, + pixman_fixed_t y, + get_pixel_t get_pixel) +{ + pixman_fixed_t *params = image->common.filter_params; + int x_off = (params[0] - pixman_fixed_1) >> 1; + int y_off = (params[1] - pixman_fixed_1) >> 1; + int32_t cwidth = pixman_fixed_to_int (params[0]); + int32_t cheight = pixman_fixed_to_int (params[1]); + int32_t srtot, sgtot, sbtot, satot; + int32_t i, j, x1, x2, y1, y2; + pixman_repeat_t repeat_mode = image->common.repeat; + int width = image->width; + int height = image->height; + + params += 2; + + x1 = pixman_fixed_to_int (x - pixman_fixed_e - x_off); + y1 = pixman_fixed_to_int (y - pixman_fixed_e - y_off); + x2 = x1 + cwidth; + y2 = y1 + cheight; + + srtot = sgtot = sbtot = satot = 0; + + for (i = y1; i < y2; ++i) + { + for (j = x1; j < x2; ++j) + { + int rx = j; + int ry = i; + + pixman_fixed_t f = *params; + + if (f) + { + uint32_t pixel; + + if (repeat_mode != PIXMAN_REPEAT_NONE) + { + repeat (repeat_mode, width, &rx); + repeat (repeat_mode, height, &ry); + + pixel = get_pixel (image, rx, ry, FALSE); + } + else + { + pixel = get_pixel (image, rx, ry, TRUE); + } + + srtot += RED_8 (pixel) * f; + sgtot += GREEN_8 (pixel) * f; + sbtot += BLUE_8 (pixel) * f; + satot += ALPHA_8 (pixel) * f; + } + + params++; + } + } + + satot >>= 16; + srtot >>= 16; + sgtot >>= 16; + sbtot >>= 16; + + satot = CLIP (satot, 0, 0xff); + srtot = CLIP (srtot, 0, 0xff); + sgtot = CLIP (sgtot, 0, 0xff); + sbtot = CLIP (sbtot, 0, 0xff); + + return ((satot << 24) | (srtot << 16) | (sgtot << 8) | (sbtot)); +} + +static force_inline uint32_t +bits_image_fetch_pixel_filtered (bits_image_t *image, + pixman_fixed_t x, + pixman_fixed_t y, + get_pixel_t get_pixel) +{ + switch (image->common.filter) + { + case PIXMAN_FILTER_NEAREST: + case PIXMAN_FILTER_FAST: + return bits_image_fetch_pixel_nearest (image, x, y, get_pixel); + break; + + case PIXMAN_FILTER_BILINEAR: + case PIXMAN_FILTER_GOOD: + case PIXMAN_FILTER_BEST: + return bits_image_fetch_pixel_bilinear (image, x, y, get_pixel); + break; + + case PIXMAN_FILTER_CONVOLUTION: + return bits_image_fetch_pixel_convolution (image, x, y, get_pixel); + break; + + default: + break; + } + + return 0; +} + +static void +bits_image_fetch_affine_no_alpha (pixman_image_t * image, + int offset, + int line, + int width, + uint32_t * buffer, + const uint32_t * mask) +{ + pixman_fixed_t x, y; + pixman_fixed_t ux, uy; + pixman_vector_t v; + int i; + + /* reference point is the center of the pixel */ + v.vector[0] = pixman_int_to_fixed (offset) + pixman_fixed_1 / 2; + v.vector[1] = pixman_int_to_fixed (line) + pixman_fixed_1 / 2; + v.vector[2] = pixman_fixed_1; + + if (image->common.transform) + { + if (!pixman_transform_point_3d (image->common.transform, &v)) + return; + + ux = image->common.transform->matrix[0][0]; + uy = image->common.transform->matrix[1][0]; + } + else + { + ux = pixman_fixed_1; + uy = 0; + } + + x = v.vector[0]; + y = v.vector[1]; + + for (i = 0; i < width; ++i) + { + if (!mask || mask[i]) + { + buffer[i] = bits_image_fetch_pixel_filtered ( + &image->bits, x, y, fetch_pixel_no_alpha); + } + + x += ux; + y += uy; + } +} + +/* General fetcher */ +static force_inline uint32_t +fetch_pixel_general (bits_image_t *image, int x, int y, pixman_bool_t check_bounds) +{ + uint32_t pixel; + + if (check_bounds && + (x < 0 || x >= image->width || y < 0 || y >= image->height)) + { + return 0; + } + + pixel = image->fetch_pixel_32 (image, x, y); + + if (image->common.alpha_map) + { + uint32_t pixel_a; + + x -= image->common.alpha_origin_x; + y -= image->common.alpha_origin_y; + + if (x < 0 || x >= image->common.alpha_map->width || + y < 0 || y >= image->common.alpha_map->height) + { + pixel_a = 0; + } + else + { + pixel_a = image->common.alpha_map->fetch_pixel_32 ( + image->common.alpha_map, x, y); + + pixel_a = ALPHA_8 (pixel_a); + } + + pixel &= 0x00ffffff; + pixel |= (pixel_a << 24); + } + + return pixel; +} + +static void +bits_image_fetch_general (pixman_image_t * image, + int offset, + int line, + int width, + uint32_t * buffer, + const uint32_t * mask) +{ + pixman_fixed_t x, y, w; + pixman_fixed_t ux, uy, uw; + pixman_vector_t v; + int i; + + /* reference point is the center of the pixel */ + v.vector[0] = pixman_int_to_fixed (offset) + pixman_fixed_1 / 2; + v.vector[1] = pixman_int_to_fixed (line) + pixman_fixed_1 / 2; + v.vector[2] = pixman_fixed_1; + + if (image->common.transform) + { + if (!pixman_transform_point_3d (image->common.transform, &v)) + return; + + ux = image->common.transform->matrix[0][0]; + uy = image->common.transform->matrix[1][0]; + uw = image->common.transform->matrix[2][0]; + } + else + { + ux = pixman_fixed_1; + uy = 0; + uw = 0; + } + + x = v.vector[0]; + y = v.vector[1]; + w = v.vector[2]; + + for (i = 0; i < width; ++i) + { + pixman_fixed_t x0, y0; + + if (!mask || mask[i]) + { + if (w != 0) + { + x0 = ((pixman_fixed_48_16_t)x << 16) / w; + y0 = ((pixman_fixed_48_16_t)y << 16) / w; + } + else + { + x0 = 0; + y0 = 0; + } + + buffer[i] = bits_image_fetch_pixel_filtered ( + &image->bits, x0, y0, fetch_pixel_general); + } + + x += ux; + y += uy; + w += uw; + } +} + +static const uint8_t zero[8] = { 0, 0, 0, 0, 0, 0, 0, 0 }; + +typedef uint32_t (* convert_pixel_t) (const uint8_t *row, int x); + +static force_inline void +bits_image_fetch_bilinear_affine (pixman_image_t * image, + int offset, + int line, + int width, + uint32_t * buffer, + const uint32_t * mask, + + convert_pixel_t convert_pixel, + pixman_format_code_t format, + pixman_repeat_t repeat_mode) +{ + pixman_fixed_t x, y; + pixman_fixed_t ux, uy; + pixman_vector_t v; + bits_image_t *bits = &image->bits; + int i; + + /* reference point is the center of the pixel */ + v.vector[0] = pixman_int_to_fixed (offset) + pixman_fixed_1 / 2; + v.vector[1] = pixman_int_to_fixed (line) + pixman_fixed_1 / 2; + v.vector[2] = pixman_fixed_1; + + if (!pixman_transform_point_3d (image->common.transform, &v)) + return; + + ux = image->common.transform->matrix[0][0]; + uy = image->common.transform->matrix[1][0]; + + x = v.vector[0]; + y = v.vector[1]; + + for (i = 0; i < width; ++i) + { + int x1, y1, x2, y2; + uint32_t tl, tr, bl, br; + int32_t distx, disty; + int width = image->bits.width; + int height = image->bits.height; + const uint8_t *row1; + const uint8_t *row2; + + if (mask && !mask[i]) + goto next; + + x1 = x - pixman_fixed_1 / 2; + y1 = y - pixman_fixed_1 / 2; + + distx = (x1 >> 8) & 0xff; + disty = (y1 >> 8) & 0xff; + + y1 = pixman_fixed_to_int (y1); + y2 = y1 + 1; + x1 = pixman_fixed_to_int (x1); + x2 = x1 + 1; + + if (repeat_mode != PIXMAN_REPEAT_NONE) + { + uint32_t mask; + + mask = PIXMAN_FORMAT_A (format)? 0 : 0xff000000; + + repeat (repeat_mode, width, &x1); + repeat (repeat_mode, height, &y1); + repeat (repeat_mode, width, &x2); + repeat (repeat_mode, height, &y2); + + row1 = (uint8_t *)bits->bits + bits->rowstride * 4 * y1; + row2 = (uint8_t *)bits->bits + bits->rowstride * 4 * y2; + + tl = convert_pixel (row1, x1) | mask; + tr = convert_pixel (row1, x2) | mask; + bl = convert_pixel (row2, x1) | mask; + br = convert_pixel (row2, x2) | mask; + } + else + { + uint32_t mask1, mask2; + int bpp; + + /* Note: PIXMAN_FORMAT_BPP() returns an unsigned value, + * which means if you use it in expressions, those + * expressions become unsigned themselves. Since + * the variables below can be negative in some cases, + * that will lead to crashes on 64 bit architectures. + * + * So this line makes sure bpp is signed + */ + bpp = PIXMAN_FORMAT_BPP (format); + + if (x1 >= width || x2 < 0 || y1 >= height || y2 < 0) + { + buffer[i] = 0; + goto next; + } + + if (y2 == 0) + { + row1 = zero; + mask1 = 0; + } + else + { + row1 = (uint8_t *)bits->bits + bits->rowstride * 4 * y1; + row1 += bpp / 8 * x1; + + mask1 = PIXMAN_FORMAT_A (format)? 0 : 0xff000000; + } + + if (y1 == height - 1) + { + row2 = zero; + mask2 = 0; + } + else + { + row2 = (uint8_t *)bits->bits + bits->rowstride * 4 * y2; + row2 += bpp / 8 * x1; + + mask2 = PIXMAN_FORMAT_A (format)? 0 : 0xff000000; + } + + if (x2 == 0) + { + tl = 0; + bl = 0; + } + else + { + tl = convert_pixel (row1, 0) | mask1; + bl = convert_pixel (row2, 0) | mask2; + } + + if (x1 == width - 1) + { + tr = 0; + br = 0; + } + else + { + tr = convert_pixel (row1, 1) | mask1; + br = convert_pixel (row2, 1) | mask2; + } + } + + buffer[i] = bilinear_interpolation ( + tl, tr, bl, br, distx, disty); + + next: + x += ux; + y += uy; + } +} + +static force_inline uint32_t +convert_a8r8g8b8 (const uint8_t *row, int x) +{ + return *(((uint32_t *)row) + x); +} + +static force_inline uint32_t +convert_x8r8g8b8 (const uint8_t *row, int x) +{ + return *(((uint32_t *)row) + x); +} + +static force_inline uint32_t +convert_a8 (const uint8_t *row, int x) +{ + return *(row + x) << 24; +} + +static force_inline uint32_t +convert_r5g6b5 (const uint8_t *row, int x) +{ + return CONVERT_0565_TO_0888 (*((uint16_t *)row + x)); +} + +#define MAKE_BILINEAR_FETCHER(name, format, repeat_mode) \ + static void \ + bits_image_fetch_bilinear_affine_ ## name (pixman_image_t *image, \ + int offset, \ + int line, \ + int width, \ + uint32_t * buffer, \ + const uint32_t * mask) \ + { \ + bits_image_fetch_bilinear_affine (image, offset, line, width, buffer, mask, \ + convert_ ## format, \ + PIXMAN_ ## format, \ + repeat_mode); \ + } \ + extern int no_such_variable + +MAKE_BILINEAR_FETCHER (pad_a8r8g8b8, a8r8g8b8, PIXMAN_REPEAT_PAD); +MAKE_BILINEAR_FETCHER (none_a8r8g8b8, a8r8g8b8, PIXMAN_REPEAT_NONE); +MAKE_BILINEAR_FETCHER (reflect_a8r8g8b8, a8r8g8b8, PIXMAN_REPEAT_REFLECT); +MAKE_BILINEAR_FETCHER (normal_a8r8g8b8, a8r8g8b8, PIXMAN_REPEAT_NORMAL); +MAKE_BILINEAR_FETCHER (pad_x8r8g8b8, x8r8g8b8, PIXMAN_REPEAT_PAD); +MAKE_BILINEAR_FETCHER (none_x8r8g8b8, x8r8g8b8, PIXMAN_REPEAT_NONE); +MAKE_BILINEAR_FETCHER (reflect_x8r8g8b8, x8r8g8b8, PIXMAN_REPEAT_REFLECT); +MAKE_BILINEAR_FETCHER (normal_x8r8g8b8, x8r8g8b8, PIXMAN_REPEAT_NORMAL); +MAKE_BILINEAR_FETCHER (pad_a8, a8, PIXMAN_REPEAT_PAD); +MAKE_BILINEAR_FETCHER (none_a8, a8, PIXMAN_REPEAT_NONE); +MAKE_BILINEAR_FETCHER (reflect_a8, a8, PIXMAN_REPEAT_REFLECT); +MAKE_BILINEAR_FETCHER (normal_a8, a8, PIXMAN_REPEAT_NORMAL); +MAKE_BILINEAR_FETCHER (pad_r5g6b5, r5g6b5, PIXMAN_REPEAT_PAD); +MAKE_BILINEAR_FETCHER (none_r5g6b5, r5g6b5, PIXMAN_REPEAT_NONE); +MAKE_BILINEAR_FETCHER (reflect_r5g6b5, r5g6b5, PIXMAN_REPEAT_REFLECT); +MAKE_BILINEAR_FETCHER (normal_r5g6b5, r5g6b5, PIXMAN_REPEAT_NORMAL); + +static void +bits_image_fetch_solid_32 (pixman_image_t * image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t * mask) +{ + uint32_t color; + uint32_t *end; + + color = image->bits.fetch_pixel_32 (&image->bits, 0, 0); + + end = buffer + width; + while (buffer < end) + *(buffer++) = color; +} + +static void +bits_image_fetch_solid_64 (pixman_image_t * image, + int x, + int y, + int width, + uint32_t * b, + const uint32_t * unused) +{ + uint64_t color; + uint64_t *buffer = (uint64_t *)b; + uint64_t *end; + + color = image->bits.fetch_pixel_64 (&image->bits, 0, 0); + + end = buffer + width; + while (buffer < end) + *(buffer++) = color; +} + +static void +bits_image_fetch_untransformed_repeat_none (bits_image_t *image, + pixman_bool_t wide, + int x, + int y, + int width, + uint32_t * buffer) +{ + uint32_t w; + + if (y < 0 || y >= image->height) + { + memset (buffer, 0, width * (wide? 8 : 4)); + return; + } + + if (x < 0) + { + w = MIN (width, -x); + + memset (buffer, 0, w * (wide ? 8 : 4)); + + width -= w; + buffer += w * (wide? 2 : 1); + x += w; + } + + if (x < image->width) + { + w = MIN (width, image->width - x); + + if (wide) + image->fetch_scanline_64 ((pixman_image_t *)image, x, y, w, buffer, NULL); + else + image->fetch_scanline_32 ((pixman_image_t *)image, x, y, w, buffer, NULL); + + width -= w; + buffer += w * (wide? 2 : 1); + x += w; + } + + memset (buffer, 0, width * (wide ? 8 : 4)); +} + +static void +bits_image_fetch_untransformed_repeat_normal (bits_image_t *image, + pixman_bool_t wide, + int x, + int y, + int width, + uint32_t * buffer) +{ + uint32_t w; + + while (y < 0) + y += image->height; + + while (y >= image->height) + y -= image->height; + + while (width) + { + while (x < 0) + x += image->width; + while (x >= image->width) + x -= image->width; + + w = MIN (width, image->width - x); + + if (wide) + image->fetch_scanline_64 ((pixman_image_t *)image, x, y, w, buffer, NULL); + else + image->fetch_scanline_32 ((pixman_image_t *)image, x, y, w, buffer, NULL); + + buffer += w * (wide? 2 : 1); + x += w; + width -= w; + } +} + +static void +bits_image_fetch_untransformed_32 (pixman_image_t * image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t * mask) +{ + if (image->common.repeat == PIXMAN_REPEAT_NONE) + { + bits_image_fetch_untransformed_repeat_none ( + &image->bits, FALSE, x, y, width, buffer); + } + else + { + bits_image_fetch_untransformed_repeat_normal ( + &image->bits, FALSE, x, y, width, buffer); + } +} + +static void +bits_image_fetch_untransformed_64 (pixman_image_t * image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t * unused) +{ + if (image->common.repeat == PIXMAN_REPEAT_NONE) + { + bits_image_fetch_untransformed_repeat_none ( + &image->bits, TRUE, x, y, width, buffer); + } + else + { + bits_image_fetch_untransformed_repeat_normal ( + &image->bits, TRUE, x, y, width, buffer); + } +} + +typedef struct +{ + pixman_format_code_t format; + uint32_t flags; + fetch_scanline_t fetch_32; + fetch_scanline_t fetch_64; +} fetcher_info_t; + +static const fetcher_info_t fetcher_info[] = +{ + { PIXMAN_solid, + FAST_PATH_NO_ALPHA_MAP, + bits_image_fetch_solid_32, + bits_image_fetch_solid_64 + }, + + { PIXMAN_any, + (FAST_PATH_NO_ALPHA_MAP | + FAST_PATH_ID_TRANSFORM | + FAST_PATH_NO_CONVOLUTION_FILTER | + FAST_PATH_NO_PAD_REPEAT | + FAST_PATH_NO_REFLECT_REPEAT), + bits_image_fetch_untransformed_32, + bits_image_fetch_untransformed_64 + }, + +#define FAST_BILINEAR_FLAGS \ + (FAST_PATH_NO_ALPHA_MAP | \ + FAST_PATH_NO_ACCESSORS | \ + FAST_PATH_HAS_TRANSFORM | \ + FAST_PATH_AFFINE_TRANSFORM | \ + FAST_PATH_X_UNIT_POSITIVE | \ + FAST_PATH_Y_UNIT_ZERO | \ + FAST_PATH_NONE_REPEAT | \ + FAST_PATH_BILINEAR_FILTER) + + { PIXMAN_a8r8g8b8, + FAST_BILINEAR_FLAGS, + bits_image_fetch_bilinear_no_repeat_8888, + _pixman_image_get_scanline_generic_64 + }, + + { PIXMAN_x8r8g8b8, + FAST_BILINEAR_FLAGS, + bits_image_fetch_bilinear_no_repeat_8888, + _pixman_image_get_scanline_generic_64 + }, + +#define GENERAL_BILINEAR_FLAGS \ + (FAST_PATH_NO_ALPHA_MAP | \ + FAST_PATH_NO_ACCESSORS | \ + FAST_PATH_HAS_TRANSFORM | \ + FAST_PATH_AFFINE_TRANSFORM | \ + FAST_PATH_BILINEAR_FILTER) + +#define BILINEAR_AFFINE_FAST_PATH(name, format, repeat) \ + { PIXMAN_ ## format, \ + GENERAL_BILINEAR_FLAGS | FAST_PATH_ ## repeat ## _REPEAT, \ + bits_image_fetch_bilinear_affine_ ## name, \ + _pixman_image_get_scanline_generic_64 \ + }, + + BILINEAR_AFFINE_FAST_PATH (pad_a8r8g8b8, a8r8g8b8, PAD) + BILINEAR_AFFINE_FAST_PATH (none_a8r8g8b8, a8r8g8b8, NONE) + BILINEAR_AFFINE_FAST_PATH (reflect_a8r8g8b8, a8r8g8b8, REFLECT) + BILINEAR_AFFINE_FAST_PATH (normal_a8r8g8b8, a8r8g8b8, NORMAL) + BILINEAR_AFFINE_FAST_PATH (pad_x8r8g8b8, x8r8g8b8, PAD) + BILINEAR_AFFINE_FAST_PATH (none_x8r8g8b8, x8r8g8b8, NONE) + BILINEAR_AFFINE_FAST_PATH (reflect_x8r8g8b8, x8r8g8b8, REFLECT) + BILINEAR_AFFINE_FAST_PATH (normal_x8r8g8b8, x8r8g8b8, NORMAL) + BILINEAR_AFFINE_FAST_PATH (pad_a8, a8, PAD) + BILINEAR_AFFINE_FAST_PATH (none_a8, a8, NONE) + BILINEAR_AFFINE_FAST_PATH (reflect_a8, a8, REFLECT) + BILINEAR_AFFINE_FAST_PATH (normal_a8, a8, NORMAL) + BILINEAR_AFFINE_FAST_PATH (pad_r5g6b5, r5g6b5, PAD) + BILINEAR_AFFINE_FAST_PATH (none_r5g6b5, r5g6b5, NONE) + BILINEAR_AFFINE_FAST_PATH (reflect_r5g6b5, r5g6b5, REFLECT) + BILINEAR_AFFINE_FAST_PATH (normal_r5g6b5, r5g6b5, NORMAL) + + /* Affine, no alpha */ + { PIXMAN_any, + (FAST_PATH_NO_ALPHA_MAP | FAST_PATH_HAS_TRANSFORM | FAST_PATH_AFFINE_TRANSFORM), + bits_image_fetch_affine_no_alpha, + _pixman_image_get_scanline_generic_64 + }, + + /* General */ + { PIXMAN_any, 0, bits_image_fetch_general, _pixman_image_get_scanline_generic_64 }, + + { PIXMAN_null }, +}; + +static void +bits_image_property_changed (pixman_image_t *image) +{ + uint32_t flags = image->common.flags; + pixman_format_code_t format = image->common.extended_format_code; + const fetcher_info_t *info; + + _pixman_bits_image_setup_accessors (&image->bits); + + info = fetcher_info; + while (info->format != PIXMAN_null) + { + if ((info->format == format || info->format == PIXMAN_any) && + (info->flags & flags) == info->flags) + { + image->common.get_scanline_32 = info->fetch_32; + image->common.get_scanline_64 = info->fetch_64; + break; + } + + info++; + } +} + +static uint32_t * +create_bits (pixman_format_code_t format, + int width, + int height, + int * rowstride_bytes) +{ + int stride; + int buf_size; + int bpp; + + /* what follows is a long-winded way, avoiding any possibility of integer + * overflows, of saying: + * stride = ((width * bpp + 0x1f) >> 5) * sizeof (uint32_t); + */ + + bpp = PIXMAN_FORMAT_BPP (format); + if (pixman_multiply_overflows_int (width, bpp)) + return NULL; + + stride = width * bpp; + if (pixman_addition_overflows_int (stride, 0x1f)) + return NULL; + + stride += 0x1f; + stride >>= 5; + + stride *= sizeof (uint32_t); + + if (pixman_multiply_overflows_int (height, stride)) + return NULL; + + buf_size = height * stride; + + if (rowstride_bytes) + *rowstride_bytes = stride; + + return calloc (buf_size, 1); +} + +PIXMAN_EXPORT pixman_image_t * +pixman_image_create_bits (pixman_format_code_t format, + int width, + int height, + uint32_t * bits, + int rowstride_bytes) +{ + pixman_image_t *image; + uint32_t *free_me = NULL; + + /* must be a whole number of uint32_t's + */ + return_val_if_fail ( + bits == NULL || (rowstride_bytes % sizeof (uint32_t)) == 0, NULL); + + return_val_if_fail (PIXMAN_FORMAT_BPP (format) >= PIXMAN_FORMAT_DEPTH (format), NULL); + + if (!bits && width && height) + { + free_me = bits = create_bits (format, width, height, &rowstride_bytes); + if (!bits) + return NULL; + } + + image = _pixman_image_allocate (); + + if (!image) + { + if (free_me) + free (free_me); + + return NULL; + } + + image->type = BITS; + image->bits.format = format; + image->bits.width = width; + image->bits.height = height; + image->bits.bits = bits; + image->bits.free_me = free_me; + image->bits.read_func = NULL; + image->bits.write_func = NULL; + + /* The rowstride is stored in number of uint32_t */ + image->bits.rowstride = rowstride_bytes / (int) sizeof (uint32_t); + + image->bits.indexed = NULL; + + image->common.property_changed = bits_image_property_changed; + + _pixman_image_reset_clip_region (image); + + return image; +} diff --git a/programs/develop/libraries/pixman/pixman-combine32.c b/programs/develop/libraries/pixman/pixman-combine32.c new file mode 100644 index 0000000000..d98d3f8383 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-combine32.c @@ -0,0 +1,2465 @@ +/* WARNING: This file is generated by combine.pl from combine.inc. + Please edit one of those files rather than this one. */ + +#line 1 "pixman-combine.c.template" +#ifdef HAVE_CONFIG_H +#include +#endif + +#include +#include + +#include "pixman-private.h" + +#include "pixman-combine32.h" + +/*** per channel helper functions ***/ + +static void +combine_mask_ca (uint32_t *src, uint32_t *mask) +{ + uint32_t a = *mask; + + uint32_t x; + uint16_t xa; + + if (!a) + { + *(src) = 0; + return; + } + + x = *(src); + if (a == ~0) + { + x = x >> A_SHIFT; + x |= x << G_SHIFT; + x |= x << R_SHIFT; + *(mask) = x; + return; + } + + xa = x >> A_SHIFT; + UN8x4_MUL_UN8x4 (x, a); + *(src) = x; + + UN8x4_MUL_UN8 (a, xa); + *(mask) = a; +} + +static void +combine_mask_value_ca (uint32_t *src, const uint32_t *mask) +{ + uint32_t a = *mask; + uint32_t x; + + if (!a) + { + *(src) = 0; + return; + } + + if (a == ~0) + return; + + x = *(src); + UN8x4_MUL_UN8x4 (x, a); + *(src) = x; +} + +static void +combine_mask_alpha_ca (const uint32_t *src, uint32_t *mask) +{ + uint32_t a = *(mask); + uint32_t x; + + if (!a) + return; + + x = *(src) >> A_SHIFT; + if (x == MASK) + return; + + if (a == ~0) + { + x |= x << G_SHIFT; + x |= x << R_SHIFT; + *(mask) = x; + return; + } + + UN8x4_MUL_UN8 (a, x); + *(mask) = a; +} + +/* + * There are two ways of handling alpha -- either as a single unified value or + * a separate value for each component, hence each macro must have two + * versions. The unified alpha version has a 'U' at the end of the name, + * the component version has a 'C'. Similarly, functions which deal with + * this difference will have two versions using the same convention. + */ + +/* + * All of the composing functions + */ + +static force_inline uint32_t +combine_mask (const uint32_t *src, const uint32_t *mask, int i) +{ + uint32_t s, m; + + if (mask) + { + m = *(mask + i) >> A_SHIFT; + + if (!m) + return 0; + } + + s = *(src + i); + + if (mask) + UN8x4_MUL_UN8 (s, m); + + return s; +} + +static void +combine_clear (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + memset (dest, 0, width * sizeof(uint32_t)); +} + +static void +combine_dst (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + return; +} + +static void +combine_src_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + if (!mask) + memcpy (dest, src, width * sizeof (uint32_t)); + else + { + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + + *(dest + i) = s; + } + } +} + +/* if the Src is opaque, call combine_src_u */ +static void +combine_over_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t ia = ALPHA_8 (~s); + + UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, s); + *(dest + i) = d; + } +} + +/* if the Dst is opaque, this is a noop */ +static void +combine_over_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t ia = ALPHA_8 (~*(dest + i)); + UN8x4_MUL_UN8_ADD_UN8x4 (s, ia, d); + *(dest + i) = s; + } +} + +/* if the Dst is opaque, call combine_src_u */ +static void +combine_in_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t a = ALPHA_8 (*(dest + i)); + UN8x4_MUL_UN8 (s, a); + *(dest + i) = s; + } +} + +/* if the Src is opaque, this is a noop */ +static void +combine_in_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t a = ALPHA_8 (s); + UN8x4_MUL_UN8 (d, a); + *(dest + i) = d; + } +} + +/* if the Dst is opaque, call combine_clear */ +static void +combine_out_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t a = ALPHA_8 (~*(dest + i)); + UN8x4_MUL_UN8 (s, a); + *(dest + i) = s; + } +} + +/* if the Src is opaque, call combine_clear */ +static void +combine_out_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t a = ALPHA_8 (~s); + UN8x4_MUL_UN8 (d, a); + *(dest + i) = d; + } +} + +/* if the Src is opaque, call combine_in_u */ +/* if the Dst is opaque, call combine_over_u */ +/* if both the Src and Dst are opaque, call combine_src_u */ +static void +combine_atop_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t dest_a = ALPHA_8 (d); + uint32_t src_ia = ALPHA_8 (~s); + + UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (s, dest_a, d, src_ia); + *(dest + i) = s; + } +} + +/* if the Src is opaque, call combine_over_reverse_u */ +/* if the Dst is opaque, call combine_in_reverse_u */ +/* if both the Src and Dst are opaque, call combine_dst_u */ +static void +combine_atop_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t src_a = ALPHA_8 (s); + uint32_t dest_ia = ALPHA_8 (~d); + + UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (s, dest_ia, d, src_a); + *(dest + i) = s; + } +} + +/* if the Src is opaque, call combine_over_u */ +/* if the Dst is opaque, call combine_over_reverse_u */ +/* if both the Src and Dst are opaque, call combine_clear */ +static void +combine_xor_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t src_ia = ALPHA_8 (~s); + uint32_t dest_ia = ALPHA_8 (~d); + + UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (s, dest_ia, d, src_ia); + *(dest + i) = s; + } +} + +static void +combine_add_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + UN8x4_ADD_UN8x4 (d, s); + *(dest + i) = d; + } +} + +/* if the Src is opaque, call combine_add_u */ +/* if the Dst is opaque, call combine_add_u */ +/* if both the Src and Dst are opaque, call combine_add_u */ +static void +combine_saturate_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint16_t sa, da; + + sa = s >> A_SHIFT; + da = ~d >> A_SHIFT; + if (sa > da) + { + sa = DIV_UN8 (da, sa); + UN8x4_MUL_UN8 (s, sa); + } + ; + UN8x4_ADD_UN8x4 (d, s); + *(dest + i) = d; + } +} + +/* + * PDF blend modes: + * The following blend modes have been taken from the PDF ISO 32000 + * specification, which at this point in time is available from + * http://www.adobe.com/devnet/acrobat/pdfs/PDF32000_2008.pdf + * The relevant chapters are 11.3.5 and 11.3.6. + * The formula for computing the final pixel color given in 11.3.6 is: + * αr × Cr = (1 – αs) × αb × Cb + (1 – αb) × αs × Cs + αb × αs × B(Cb, Cs) + * with B() being the blend function. + * Note that OVER is a special case of this operation, using B(Cb, Cs) = Cs + * + * These blend modes should match the SVG filter draft specification, as + * it has been designed to mirror ISO 32000. Note that at the current point + * no released draft exists that shows this, as the formulas have not been + * updated yet after the release of ISO 32000. + * + * The default implementation here uses the PDF_SEPARABLE_BLEND_MODE and + * PDF_NON_SEPARABLE_BLEND_MODE macros, which take the blend function as an + * argument. Note that this implementation operates on premultiplied colors, + * while the PDF specification does not. Therefore the code uses the formula + * ar.Cra = (1 – as) . Dca + (1 – ad) . Sca + B(Dca, ad, Sca, as) + */ + +/* + * Multiply + * B(Dca, ad, Sca, as) = Dca.Sca + */ + +static void +combine_multiply_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t ss = s; + uint32_t src_ia = ALPHA_8 (~s); + uint32_t dest_ia = ALPHA_8 (~d); + + UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (ss, dest_ia, d, src_ia); + UN8x4_MUL_UN8x4 (d, s); + UN8x4_ADD_UN8x4 (d, ss); + + *(dest + i) = d; + } +} + +static void +combine_multiply_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t m = *(mask + i); + uint32_t s = *(src + i); + uint32_t d = *(dest + i); + uint32_t r = d; + uint32_t dest_ia = ALPHA_8 (~d); + + combine_mask_value_ca (&s, &m); + + UN8x4_MUL_UN8x4_ADD_UN8x4_MUL_UN8 (r, ~m, s, dest_ia); + UN8x4_MUL_UN8x4 (d, s); + UN8x4_ADD_UN8x4 (r, d); + + *(dest + i) = r; + } +} + +#define PDF_SEPARABLE_BLEND_MODE(name) \ + static void \ + combine_ ## name ## _u (pixman_implementation_t *imp, \ + pixman_op_t op, \ + uint32_t * dest, \ + const uint32_t * src, \ + const uint32_t * mask, \ + int width) \ + { \ + int i; \ + for (i = 0; i < width; ++i) { \ + uint32_t s = combine_mask (src, mask, i); \ + uint32_t d = *(dest + i); \ + uint8_t sa = ALPHA_8 (s); \ + uint8_t isa = ~sa; \ + uint8_t da = ALPHA_8 (d); \ + uint8_t ida = ~da; \ + uint32_t result; \ + \ + result = d; \ + UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (result, isa, s, ida); \ + \ + *(dest + i) = result + \ + (DIV_ONE_UN8 (sa * da) << A_SHIFT) + \ + (blend_ ## name (RED_8 (d), da, RED_8 (s), sa) << R_SHIFT) + \ + (blend_ ## name (GREEN_8 (d), da, GREEN_8 (s), sa) << G_SHIFT) + \ + (blend_ ## name (BLUE_8 (d), da, BLUE_8 (s), sa)); \ + } \ + } \ + \ + static void \ + combine_ ## name ## _ca (pixman_implementation_t *imp, \ + pixman_op_t op, \ + uint32_t * dest, \ + const uint32_t * src, \ + const uint32_t * mask, \ + int width) \ + { \ + int i; \ + for (i = 0; i < width; ++i) { \ + uint32_t m = *(mask + i); \ + uint32_t s = *(src + i); \ + uint32_t d = *(dest + i); \ + uint8_t da = ALPHA_8 (d); \ + uint8_t ida = ~da; \ + uint32_t result; \ + \ + combine_mask_value_ca (&s, &m); \ + \ + result = d; \ + UN8x4_MUL_UN8x4_ADD_UN8x4_MUL_UN8 (result, ~m, s, ida); \ + \ + result += \ + (DIV_ONE_UN8 (ALPHA_8 (m) * da) << A_SHIFT) + \ + (blend_ ## name (RED_8 (d), da, RED_8 (s), RED_8 (m)) << R_SHIFT) + \ + (blend_ ## name (GREEN_8 (d), da, GREEN_8 (s), GREEN_8 (m)) << G_SHIFT) + \ + (blend_ ## name (BLUE_8 (d), da, BLUE_8 (s), BLUE_8 (m))); \ + \ + *(dest + i) = result; \ + } \ + } + +/* + * Screen + * B(Dca, ad, Sca, as) = Dca.sa + Sca.da - Dca.Sca + */ +static inline uint32_t +blend_screen (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + return DIV_ONE_UN8 (sca * da + dca * sa - sca * dca); +} + +PDF_SEPARABLE_BLEND_MODE (screen) + +/* + * Overlay + * B(Dca, Da, Sca, Sa) = + * if 2.Dca < Da + * 2.Sca.Dca + * otherwise + * Sa.Da - 2.(Da - Dca).(Sa - Sca) + */ +static inline uint32_t +blend_overlay (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + uint32_t rca; + + if (2 * dca < da) + rca = 2 * sca * dca; + else + rca = sa * da - 2 * (da - dca) * (sa - sca); + return DIV_ONE_UN8 (rca); +} + +PDF_SEPARABLE_BLEND_MODE (overlay) + +/* + * Darken + * B(Dca, Da, Sca, Sa) = min (Sca.Da, Dca.Sa) + */ +static inline uint32_t +blend_darken (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + uint32_t s, d; + + s = sca * da; + d = dca * sa; + return DIV_ONE_UN8 (s > d ? d : s); +} + +PDF_SEPARABLE_BLEND_MODE (darken) + +/* + * Lighten + * B(Dca, Da, Sca, Sa) = max (Sca.Da, Dca.Sa) + */ +static inline uint32_t +blend_lighten (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + uint32_t s, d; + + s = sca * da; + d = dca * sa; + return DIV_ONE_UN8 (s > d ? s : d); +} + +PDF_SEPARABLE_BLEND_MODE (lighten) + +/* + * Color dodge + * B(Dca, Da, Sca, Sa) = + * if Dca == 0 + * 0 + * if Sca == Sa + * Sa.Da + * otherwise + * Sa.Da. min (1, Dca / Da / (1 - Sca/Sa)) + */ +static inline uint32_t +blend_color_dodge (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + if (sca >= sa) + { + return dca == 0 ? 0 : DIV_ONE_UN8 (sa * da); + } + else + { + uint32_t rca = dca * sa / (sa - sca); + return DIV_ONE_UN8 (sa * MIN (rca, da)); + } +} + +PDF_SEPARABLE_BLEND_MODE (color_dodge) + +/* + * Color burn + * B(Dca, Da, Sca, Sa) = + * if Dca == Da + * Sa.Da + * if Sca == 0 + * 0 + * otherwise + * Sa.Da.(1 - min (1, (1 - Dca/Da).Sa / Sca)) + */ +static inline uint32_t +blend_color_burn (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + if (sca == 0) + { + return dca < da ? 0 : DIV_ONE_UN8 (sa * da); + } + else + { + uint32_t rca = (da - dca) * sa / sca; + return DIV_ONE_UN8 (sa * (MAX (rca, da) - rca)); + } +} + +PDF_SEPARABLE_BLEND_MODE (color_burn) + +/* + * Hard light + * B(Dca, Da, Sca, Sa) = + * if 2.Sca < Sa + * 2.Sca.Dca + * otherwise + * Sa.Da - 2.(Da - Dca).(Sa - Sca) + */ +static inline uint32_t +blend_hard_light (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + if (2 * sca < sa) + return DIV_ONE_UN8 (2 * sca * dca); + else + return DIV_ONE_UN8 (sa * da - 2 * (da - dca) * (sa - sca)); +} + +PDF_SEPARABLE_BLEND_MODE (hard_light) + +/* + * Soft light + * B(Dca, Da, Sca, Sa) = + * if (2.Sca <= Sa) + * Dca.(Sa - (1 - Dca/Da).(2.Sca - Sa)) + * otherwise if Dca.4 <= Da + * Dca.(Sa + (2.Sca - Sa).((16.Dca/Da - 12).Dca/Da + 3) + * otherwise + * (Dca.Sa + (SQRT (Dca/Da).Da - Dca).(2.Sca - Sa)) + */ +static inline uint32_t +blend_soft_light (uint32_t dca_org, + uint32_t da_org, + uint32_t sca_org, + uint32_t sa_org) +{ + double dca = dca_org * (1.0 / MASK); + double da = da_org * (1.0 / MASK); + double sca = sca_org * (1.0 / MASK); + double sa = sa_org * (1.0 / MASK); + double rca; + + if (2 * sca < sa) + { + if (da == 0) + rca = dca * sa; + else + rca = dca * sa - dca * (da - dca) * (sa - 2 * sca) / da; + } + else if (da == 0) + { + rca = 0; + } + else if (4 * dca <= da) + { + rca = dca * sa + + (2 * sca - sa) * dca * ((16 * dca / da - 12) * dca / da + 3); + } + else + { + rca = dca * sa + (sqrt (dca * da) - dca) * (2 * sca - sa); + } + return rca * MASK + 0.5; +} + +PDF_SEPARABLE_BLEND_MODE (soft_light) + +/* + * Difference + * B(Dca, Da, Sca, Sa) = abs (Dca.Sa - Sca.Da) + */ +static inline uint32_t +blend_difference (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + uint32_t dcasa = dca * sa; + uint32_t scada = sca * da; + + if (scada < dcasa) + return DIV_ONE_UN8 (dcasa - scada); + else + return DIV_ONE_UN8 (scada - dcasa); +} + +PDF_SEPARABLE_BLEND_MODE (difference) + +/* + * Exclusion + * B(Dca, Da, Sca, Sa) = (Sca.Da + Dca.Sa - 2.Sca.Dca) + */ + +/* This can be made faster by writing it directly and not using + * PDF_SEPARABLE_BLEND_MODE, but that's a performance optimization */ + +static inline uint32_t +blend_exclusion (uint32_t dca, uint32_t da, uint32_t sca, uint32_t sa) +{ + return DIV_ONE_UN8 (sca * da + dca * sa - 2 * dca * sca); +} + +PDF_SEPARABLE_BLEND_MODE (exclusion) + +#undef PDF_SEPARABLE_BLEND_MODE + +/* + * PDF nonseperable blend modes are implemented using the following functions + * to operate in Hsl space, with Cmax, Cmid, Cmin referring to the max, mid + * and min value of the red, green and blue components. + * + * LUM (C) = 0.3 × Cred + 0.59 × Cgreen + 0.11 × Cblue + * + * clip_color (C): + * l = LUM (C) + * min = Cmin + * max = Cmax + * if n < 0.0 + * C = l + ( ( ( C – l ) × l ) ⁄ ( l – min ) ) + * if x > 1.0 + * C = l + ( ( ( C – l ) × ( 1 – l ) ) ⁄ ( max – l ) ) + * return C + * + * set_lum (C, l): + * d = l – LUM (C) + * C += d + * return clip_color (C) + * + * SAT (C) = CH_MAX (C) - CH_MIN (C) + * + * set_sat (C, s): + * if Cmax > Cmin + * Cmid = ( ( ( Cmid – Cmin ) × s ) ⁄ ( Cmax – Cmin ) ) + * Cmax = s + * else + * Cmid = Cmax = 0.0 + * Cmin = 0.0 + * return C + */ + +/* For premultiplied colors, we need to know what happens when C is + * multiplied by a real number. LUM and SAT are linear: + * + * LUM (r × C) = r × LUM (C) SAT (r * C) = r * SAT (C) + * + * If we extend clip_color with an extra argument a and change + * + * if x >= 1.0 + * + * into + * + * if x >= a + * + * then clip_color is also linear: + * + * r * clip_color (C, a) = clip_color (r_c, ra); + * + * for positive r. + * + * Similarly, we can extend set_lum with an extra argument that is just passed + * on to clip_color: + * + * r * set_lum ( C, l, a) + * + * = r × clip_color ( C + l - LUM (C), a) + * + * = clip_color ( r * C + r × l - r * LUM (C), r * a) + * + * = set_lum ( r * C, r * l, r * a) + * + * Finally, set_sat: + * + * r * set_sat (C, s) = set_sat (x * C, r * s) + * + * The above holds for all non-zero x, because they x'es in the fraction for + * C_mid cancel out. Specifically, it holds for x = r: + * + * r * set_sat (C, s) = set_sat (r_c, rs) + * + */ + +/* So, for the non-separable PDF blend modes, we have (using s, d for + * non-premultiplied colors, and S, D for premultiplied: + * + * Color: + * + * a_s * a_d * B(s, d) + * = a_s * a_d * set_lum (S/a_s, LUM (D/a_d), 1) + * = set_lum (S * a_d, a_s * LUM (D), a_s * a_d) + * + * + * Luminosity: + * + * a_s * a_d * B(s, d) + * = a_s * a_d * set_lum (D/a_d, LUM(S/a_s), 1) + * = set_lum (a_s * D, a_d * LUM(S), a_s * a_d) + * + * + * Saturation: + * + * a_s * a_d * B(s, d) + * = a_s * a_d * set_lum (set_sat (D/a_d, SAT (S/a_s)), LUM (D/a_d), 1) + * = set_lum (a_s * a_d * set_sat (D/a_d, SAT (S/a_s)), + * a_s * LUM (D), a_s * a_d) + * = set_lum (set_sat (a_s * D, a_d * SAT (S), a_s * LUM (D), a_s * a_d)) + * + * Hue: + * + * a_s * a_d * B(s, d) + * = a_s * a_d * set_lum (set_sat (S/a_s, SAT (D/a_d)), LUM (D/a_d), 1) + * = a_s * a_d * set_lum (set_sat (a_d * S, a_s * SAT (D)), + * a_s * LUM (D), a_s * a_d) + * + */ + +#define CH_MIN(c) (c[0] < c[1] ? (c[0] < c[2] ? c[0] : c[2]) : (c[1] < c[2] ? c[1] : c[2])) +#define CH_MAX(c) (c[0] > c[1] ? (c[0] > c[2] ? c[0] : c[2]) : (c[1] > c[2] ? c[1] : c[2])) +#define LUM(c) ((c[0] * 30 + c[1] * 59 + c[2] * 11) / 100) +#define SAT(c) (CH_MAX (c) - CH_MIN (c)) + +#define PDF_NON_SEPARABLE_BLEND_MODE(name) \ + static void \ + combine_ ## name ## _u (pixman_implementation_t *imp, \ + pixman_op_t op, \ + uint32_t *dest, \ + const uint32_t *src, \ + const uint32_t *mask, \ + int width) \ + { \ + int i; \ + for (i = 0; i < width; ++i) \ + { \ + uint32_t s = combine_mask (src, mask, i); \ + uint32_t d = *(dest + i); \ + uint8_t sa = ALPHA_8 (s); \ + uint8_t isa = ~sa; \ + uint8_t da = ALPHA_8 (d); \ + uint8_t ida = ~da; \ + uint32_t result; \ + uint32_t sc[3], dc[3], c[3]; \ + \ + result = d; \ + UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (result, isa, s, ida); \ + dc[0] = RED_8 (d); \ + sc[0] = RED_8 (s); \ + dc[1] = GREEN_8 (d); \ + sc[1] = GREEN_8 (s); \ + dc[2] = BLUE_8 (d); \ + sc[2] = BLUE_8 (s); \ + blend_ ## name (c, dc, da, sc, sa); \ + \ + *(dest + i) = result + \ + (DIV_ONE_UN8 (sa * da) << A_SHIFT) + \ + (DIV_ONE_UN8 (c[0]) << R_SHIFT) + \ + (DIV_ONE_UN8 (c[1]) << G_SHIFT) + \ + (DIV_ONE_UN8 (c[2])); \ + } \ + } + +static void +set_lum (uint32_t dest[3], uint32_t src[3], uint32_t sa, uint32_t lum) +{ + double a, l, min, max; + double tmp[3]; + + a = sa * (1.0 / MASK); + + l = lum * (1.0 / MASK); + tmp[0] = src[0] * (1.0 / MASK); + tmp[1] = src[1] * (1.0 / MASK); + tmp[2] = src[2] * (1.0 / MASK); + + l = l - LUM (tmp); + tmp[0] += l; + tmp[1] += l; + tmp[2] += l; + + /* clip_color */ + l = LUM (tmp); + min = CH_MIN (tmp); + max = CH_MAX (tmp); + + if (min < 0) + { + if (l - min == 0.0) + { + tmp[0] = 0; + tmp[1] = 0; + tmp[2] = 0; + } + else + { + tmp[0] = l + (tmp[0] - l) * l / (l - min); + tmp[1] = l + (tmp[1] - l) * l / (l - min); + tmp[2] = l + (tmp[2] - l) * l / (l - min); + } + } + if (max > a) + { + if (max - l == 0.0) + { + tmp[0] = a; + tmp[1] = a; + tmp[2] = a; + } + else + { + tmp[0] = l + (tmp[0] - l) * (a - l) / (max - l); + tmp[1] = l + (tmp[1] - l) * (a - l) / (max - l); + tmp[2] = l + (tmp[2] - l) * (a - l) / (max - l); + } + } + + dest[0] = tmp[0] * MASK + 0.5; + dest[1] = tmp[1] * MASK + 0.5; + dest[2] = tmp[2] * MASK + 0.5; +} + +static void +set_sat (uint32_t dest[3], uint32_t src[3], uint32_t sat) +{ + int id[3]; + uint32_t min, max; + + if (src[0] > src[1]) + { + if (src[0] > src[2]) + { + id[0] = 0; + if (src[1] > src[2]) + { + id[1] = 1; + id[2] = 2; + } + else + { + id[1] = 2; + id[2] = 1; + } + } + else + { + id[0] = 2; + id[1] = 0; + id[2] = 1; + } + } + else + { + if (src[0] > src[2]) + { + id[0] = 1; + id[1] = 0; + id[2] = 2; + } + else + { + id[2] = 0; + if (src[1] > src[2]) + { + id[0] = 1; + id[1] = 2; + } + else + { + id[0] = 2; + id[1] = 1; + } + } + } + + max = dest[id[0]]; + min = dest[id[2]]; + if (max > min) + { + dest[id[1]] = (dest[id[1]] - min) * sat / (max - min); + dest[id[0]] = sat; + dest[id[2]] = 0; + } + else + { + dest[0] = dest[1] = dest[2] = 0; + } +} + +/* + * Hue: + * B(Cb, Cs) = set_lum (set_sat (Cs, SAT (Cb)), LUM (Cb)) + */ +static inline void +blend_hsl_hue (uint32_t c[3], + uint32_t dc[3], + uint32_t da, + uint32_t sc[3], + uint32_t sa) +{ + c[0] = sc[0] * da; + c[1] = sc[1] * da; + c[2] = sc[2] * da; + set_sat (c, c, SAT (dc) * sa); + set_lum (c, c, sa * da, LUM (dc) * sa); +} + +PDF_NON_SEPARABLE_BLEND_MODE (hsl_hue) + +/* + * Saturation: + * B(Cb, Cs) = set_lum (set_sat (Cb, SAT (Cs)), LUM (Cb)) + */ +static inline void +blend_hsl_saturation (uint32_t c[3], + uint32_t dc[3], + uint32_t da, + uint32_t sc[3], + uint32_t sa) +{ + c[0] = dc[0] * sa; + c[1] = dc[1] * sa; + c[2] = dc[2] * sa; + set_sat (c, c, SAT (sc) * da); + set_lum (c, c, sa * da, LUM (dc) * sa); +} + +PDF_NON_SEPARABLE_BLEND_MODE (hsl_saturation) + +/* + * Color: + * B(Cb, Cs) = set_lum (Cs, LUM (Cb)) + */ +static inline void +blend_hsl_color (uint32_t c[3], + uint32_t dc[3], + uint32_t da, + uint32_t sc[3], + uint32_t sa) +{ + c[0] = sc[0] * da; + c[1] = sc[1] * da; + c[2] = sc[2] * da; + set_lum (c, c, sa * da, LUM (dc) * sa); +} + +PDF_NON_SEPARABLE_BLEND_MODE (hsl_color) + +/* + * Luminosity: + * B(Cb, Cs) = set_lum (Cb, LUM (Cs)) + */ +static inline void +blend_hsl_luminosity (uint32_t c[3], + uint32_t dc[3], + uint32_t da, + uint32_t sc[3], + uint32_t sa) +{ + c[0] = dc[0] * sa; + c[1] = dc[1] * sa; + c[2] = dc[2] * sa; + set_lum (c, c, sa * da, LUM (sc) * da); +} + +PDF_NON_SEPARABLE_BLEND_MODE (hsl_luminosity) + +#undef SAT +#undef LUM +#undef CH_MAX +#undef CH_MIN +#undef PDF_NON_SEPARABLE_BLEND_MODE + +/* Overlay + * + * All of the disjoint composing functions + * + * The four entries in the first column indicate what source contributions + * come from each of the four areas of the picture -- areas covered by neither + * A nor B, areas covered only by A, areas covered only by B and finally + * areas covered by both A and B. + * + * Disjoint Conjoint + * Fa Fb Fa Fb + * (0,0,0,0) 0 0 0 0 + * (0,A,0,A) 1 0 1 0 + * (0,0,B,B) 0 1 0 1 + * (0,A,B,A) 1 min((1-a)/b,1) 1 max(1-a/b,0) + * (0,A,B,B) min((1-b)/a,1) 1 max(1-b/a,0) 1 + * (0,0,0,A) max(1-(1-b)/a,0) 0 min(1,b/a) 0 + * (0,0,0,B) 0 max(1-(1-a)/b,0) 0 min(a/b,1) + * (0,A,0,0) min(1,(1-b)/a) 0 max(1-b/a,0) 0 + * (0,0,B,0) 0 min(1,(1-a)/b) 0 max(1-a/b,0) + * (0,0,B,A) max(1-(1-b)/a,0) min(1,(1-a)/b) min(1,b/a) max(1-a/b,0) + * (0,A,0,B) min(1,(1-b)/a) max(1-(1-a)/b,0) max(1-b/a,0) min(1,a/b) + * (0,A,B,0) min(1,(1-b)/a) min(1,(1-a)/b) max(1-b/a,0) max(1-a/b,0) + */ + +#define COMBINE_A_OUT 1 +#define COMBINE_A_IN 2 +#define COMBINE_B_OUT 4 +#define COMBINE_B_IN 8 + +#define COMBINE_CLEAR 0 +#define COMBINE_A (COMBINE_A_OUT | COMBINE_A_IN) +#define COMBINE_B (COMBINE_B_OUT | COMBINE_B_IN) +#define COMBINE_A_OVER (COMBINE_A_OUT | COMBINE_B_OUT | COMBINE_A_IN) +#define COMBINE_B_OVER (COMBINE_A_OUT | COMBINE_B_OUT | COMBINE_B_IN) +#define COMBINE_A_ATOP (COMBINE_B_OUT | COMBINE_A_IN) +#define COMBINE_B_ATOP (COMBINE_A_OUT | COMBINE_B_IN) +#define COMBINE_XOR (COMBINE_A_OUT | COMBINE_B_OUT) + +/* portion covered by a but not b */ +static uint8_t +combine_disjoint_out_part (uint8_t a, uint8_t b) +{ + /* min (1, (1-b) / a) */ + + b = ~b; /* 1 - b */ + if (b >= a) /* 1 - b >= a -> (1-b)/a >= 1 */ + return MASK; /* 1 */ + return DIV_UN8 (b, a); /* (1-b) / a */ +} + +/* portion covered by both a and b */ +static uint8_t +combine_disjoint_in_part (uint8_t a, uint8_t b) +{ + /* max (1-(1-b)/a,0) */ + /* = - min ((1-b)/a - 1, 0) */ + /* = 1 - min (1, (1-b)/a) */ + + b = ~b; /* 1 - b */ + if (b >= a) /* 1 - b >= a -> (1-b)/a >= 1 */ + return 0; /* 1 - 1 */ + return ~DIV_UN8(b, a); /* 1 - (1-b) / a */ +} + +/* portion covered by a but not b */ +static uint8_t +combine_conjoint_out_part (uint8_t a, uint8_t b) +{ + /* max (1-b/a,0) */ + /* = 1-min(b/a,1) */ + + /* min (1, (1-b) / a) */ + + if (b >= a) /* b >= a -> b/a >= 1 */ + return 0x00; /* 0 */ + return ~DIV_UN8(b, a); /* 1 - b/a */ +} + +/* portion covered by both a and b */ +static uint8_t +combine_conjoint_in_part (uint8_t a, uint8_t b) +{ + /* min (1,b/a) */ + + if (b >= a) /* b >= a -> b/a >= 1 */ + return MASK; /* 1 */ + return DIV_UN8 (b, a); /* b/a */ +} + +#define GET_COMP(v, i) ((uint16_t) (uint8_t) ((v) >> i)) + +#define ADD(x, y, i, t) \ + ((t) = GET_COMP (x, i) + GET_COMP (y, i), \ + (uint32_t) ((uint8_t) ((t) | (0 - ((t) >> G_SHIFT)))) << (i)) + +#define GENERIC(x, y, i, ax, ay, t, u, v) \ + ((t) = (MUL_UN8 (GET_COMP (y, i), ay, (u)) + \ + MUL_UN8 (GET_COMP (x, i), ax, (v))), \ + (uint32_t) ((uint8_t) ((t) | \ + (0 - ((t) >> G_SHIFT)))) << (i)) + +static void +combine_disjoint_general_u (uint32_t * dest, + const uint32_t *src, + const uint32_t *mask, + int width, + uint8_t combine) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t m, n, o, p; + uint16_t Fa, Fb, t, u, v; + uint8_t sa = s >> A_SHIFT; + uint8_t da = d >> A_SHIFT; + + switch (combine & COMBINE_A) + { + default: + Fa = 0; + break; + + case COMBINE_A_OUT: + Fa = combine_disjoint_out_part (sa, da); + break; + + case COMBINE_A_IN: + Fa = combine_disjoint_in_part (sa, da); + break; + + case COMBINE_A: + Fa = MASK; + break; + } + + switch (combine & COMBINE_B) + { + default: + Fb = 0; + break; + + case COMBINE_B_OUT: + Fb = combine_disjoint_out_part (da, sa); + break; + + case COMBINE_B_IN: + Fb = combine_disjoint_in_part (da, sa); + break; + + case COMBINE_B: + Fb = MASK; + break; + } + m = GENERIC (s, d, 0, Fa, Fb, t, u, v); + n = GENERIC (s, d, G_SHIFT, Fa, Fb, t, u, v); + o = GENERIC (s, d, R_SHIFT, Fa, Fb, t, u, v); + p = GENERIC (s, d, A_SHIFT, Fa, Fb, t, u, v); + s = m | n | o | p; + *(dest + i) = s; + } +} + +static void +combine_disjoint_over_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint16_t a = s >> A_SHIFT; + + if (s != 0x00) + { + uint32_t d = *(dest + i); + a = combine_disjoint_out_part (d >> A_SHIFT, a); + UN8x4_MUL_UN8_ADD_UN8x4 (d, a, s); + + *(dest + i) = d; + } + } +} + +static void +combine_disjoint_in_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_IN); +} + +static void +combine_disjoint_in_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_IN); +} + +static void +combine_disjoint_out_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_OUT); +} + +static void +combine_disjoint_out_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_OUT); +} + +static void +combine_disjoint_atop_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_ATOP); +} + +static void +combine_disjoint_atop_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_ATOP); +} + +static void +combine_disjoint_xor_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_XOR); +} + +static void +combine_conjoint_general_u (uint32_t * dest, + const uint32_t *src, + const uint32_t *mask, + int width, + uint8_t combine) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = combine_mask (src, mask, i); + uint32_t d = *(dest + i); + uint32_t m, n, o, p; + uint16_t Fa, Fb, t, u, v; + uint8_t sa = s >> A_SHIFT; + uint8_t da = d >> A_SHIFT; + + switch (combine & COMBINE_A) + { + default: + Fa = 0; + break; + + case COMBINE_A_OUT: + Fa = combine_conjoint_out_part (sa, da); + break; + + case COMBINE_A_IN: + Fa = combine_conjoint_in_part (sa, da); + break; + + case COMBINE_A: + Fa = MASK; + break; + } + + switch (combine & COMBINE_B) + { + default: + Fb = 0; + break; + + case COMBINE_B_OUT: + Fb = combine_conjoint_out_part (da, sa); + break; + + case COMBINE_B_IN: + Fb = combine_conjoint_in_part (da, sa); + break; + + case COMBINE_B: + Fb = MASK; + break; + } + + m = GENERIC (s, d, 0, Fa, Fb, t, u, v); + n = GENERIC (s, d, G_SHIFT, Fa, Fb, t, u, v); + o = GENERIC (s, d, R_SHIFT, Fa, Fb, t, u, v); + p = GENERIC (s, d, A_SHIFT, Fa, Fb, t, u, v); + + s = m | n | o | p; + + *(dest + i) = s; + } +} + +static void +combine_conjoint_over_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_OVER); +} + +static void +combine_conjoint_over_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_OVER); +} + +static void +combine_conjoint_in_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_IN); +} + +static void +combine_conjoint_in_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_IN); +} + +static void +combine_conjoint_out_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_OUT); +} + +static void +combine_conjoint_out_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_OUT); +} + +static void +combine_conjoint_atop_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_ATOP); +} + +static void +combine_conjoint_atop_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_ATOP); +} + +static void +combine_conjoint_xor_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_XOR); +} + +/************************************************************************/ +/*********************** Per Channel functions **************************/ +/************************************************************************/ + +static void +combine_clear_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + memset (dest, 0, width * sizeof(uint32_t)); +} + +static void +combine_src_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + + combine_mask_value_ca (&s, &m); + + *(dest + i) = s; + } +} + +static void +combine_over_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + uint32_t a; + + combine_mask_ca (&s, &m); + + a = ~m; + if (a) + { + uint32_t d = *(dest + i); + UN8x4_MUL_UN8x4_ADD_UN8x4 (d, a, s); + s = d; + } + + *(dest + i) = s; + } +} + +static void +combine_over_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t d = *(dest + i); + uint32_t a = ~d >> A_SHIFT; + + if (a) + { + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + + UN8x4_MUL_UN8x4 (s, m); + UN8x4_MUL_UN8_ADD_UN8x4 (s, a, d); + + *(dest + i) = s; + } + } +} + +static void +combine_in_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t d = *(dest + i); + uint16_t a = d >> A_SHIFT; + uint32_t s = 0; + + if (a) + { + uint32_t m = *(mask + i); + + s = *(src + i); + combine_mask_value_ca (&s, &m); + + if (a != MASK) + UN8x4_MUL_UN8 (s, a); + } + + *(dest + i) = s; + } +} + +static void +combine_in_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + uint32_t a; + + combine_mask_alpha_ca (&s, &m); + + a = m; + if (a != ~0) + { + uint32_t d = 0; + + if (a) + { + d = *(dest + i); + UN8x4_MUL_UN8x4 (d, a); + } + + *(dest + i) = d; + } + } +} + +static void +combine_out_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t d = *(dest + i); + uint16_t a = ~d >> A_SHIFT; + uint32_t s = 0; + + if (a) + { + uint32_t m = *(mask + i); + + s = *(src + i); + combine_mask_value_ca (&s, &m); + + if (a != MASK) + UN8x4_MUL_UN8 (s, a); + } + + *(dest + i) = s; + } +} + +static void +combine_out_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + uint32_t a; + + combine_mask_alpha_ca (&s, &m); + + a = ~m; + if (a != ~0) + { + uint32_t d = 0; + + if (a) + { + d = *(dest + i); + UN8x4_MUL_UN8x4 (d, a); + } + + *(dest + i) = d; + } + } +} + +static void +combine_atop_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t d = *(dest + i); + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + uint32_t ad; + uint16_t as = d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + ad = ~m; + + UN8x4_MUL_UN8x4_ADD_UN8x4_MUL_UN8 (d, ad, s, as); + + *(dest + i) = d; + } +} + +static void +combine_atop_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t d = *(dest + i); + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + uint32_t ad; + uint16_t as = ~d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + ad = m; + + UN8x4_MUL_UN8x4_ADD_UN8x4_MUL_UN8 (d, ad, s, as); + + *(dest + i) = d; + } +} + +static void +combine_xor_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t d = *(dest + i); + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + uint32_t ad; + uint16_t as = ~d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + ad = ~m; + + UN8x4_MUL_UN8x4_ADD_UN8x4_MUL_UN8 (d, ad, s, as); + + *(dest + i) = d; + } +} + +static void +combine_add_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s = *(src + i); + uint32_t m = *(mask + i); + uint32_t d = *(dest + i); + + combine_mask_value_ca (&s, &m); + + UN8x4_ADD_UN8x4 (d, s); + + *(dest + i) = d; + } +} + +static void +combine_saturate_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s, d; + uint16_t sa, sr, sg, sb, da; + uint16_t t, u, v; + uint32_t m, n, o, p; + + d = *(dest + i); + s = *(src + i); + m = *(mask + i); + + combine_mask_ca (&s, &m); + + sa = (m >> A_SHIFT); + sr = (m >> R_SHIFT) & MASK; + sg = (m >> G_SHIFT) & MASK; + sb = m & MASK; + da = ~d >> A_SHIFT; + + if (sb <= da) + m = ADD (s, d, 0, t); + else + m = GENERIC (s, d, 0, (da << G_SHIFT) / sb, MASK, t, u, v); + + if (sg <= da) + n = ADD (s, d, G_SHIFT, t); + else + n = GENERIC (s, d, G_SHIFT, (da << G_SHIFT) / sg, MASK, t, u, v); + + if (sr <= da) + o = ADD (s, d, R_SHIFT, t); + else + o = GENERIC (s, d, R_SHIFT, (da << G_SHIFT) / sr, MASK, t, u, v); + + if (sa <= da) + p = ADD (s, d, A_SHIFT, t); + else + p = GENERIC (s, d, A_SHIFT, (da << G_SHIFT) / sa, MASK, t, u, v); + + *(dest + i) = m | n | o | p; + } +} + +static void +combine_disjoint_general_ca (uint32_t * dest, + const uint32_t *src, + const uint32_t *mask, + int width, + uint8_t combine) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s, d; + uint32_t m, n, o, p; + uint32_t Fa, Fb; + uint16_t t, u, v; + uint32_t sa; + uint8_t da; + + s = *(src + i); + m = *(mask + i); + d = *(dest + i); + da = d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + sa = m; + + switch (combine & COMBINE_A) + { + default: + Fa = 0; + break; + + case COMBINE_A_OUT: + m = (uint32_t)combine_disjoint_out_part ((uint8_t) (sa >> 0), da); + n = (uint32_t)combine_disjoint_out_part ((uint8_t) (sa >> G_SHIFT), da) << G_SHIFT; + o = (uint32_t)combine_disjoint_out_part ((uint8_t) (sa >> R_SHIFT), da) << R_SHIFT; + p = (uint32_t)combine_disjoint_out_part ((uint8_t) (sa >> A_SHIFT), da) << A_SHIFT; + Fa = m | n | o | p; + break; + + case COMBINE_A_IN: + m = (uint32_t)combine_disjoint_in_part ((uint8_t) (sa >> 0), da); + n = (uint32_t)combine_disjoint_in_part ((uint8_t) (sa >> G_SHIFT), da) << G_SHIFT; + o = (uint32_t)combine_disjoint_in_part ((uint8_t) (sa >> R_SHIFT), da) << R_SHIFT; + p = (uint32_t)combine_disjoint_in_part ((uint8_t) (sa >> A_SHIFT), da) << A_SHIFT; + Fa = m | n | o | p; + break; + + case COMBINE_A: + Fa = ~0; + break; + } + + switch (combine & COMBINE_B) + { + default: + Fb = 0; + break; + + case COMBINE_B_OUT: + m = (uint32_t)combine_disjoint_out_part (da, (uint8_t) (sa >> 0)); + n = (uint32_t)combine_disjoint_out_part (da, (uint8_t) (sa >> G_SHIFT)) << G_SHIFT; + o = (uint32_t)combine_disjoint_out_part (da, (uint8_t) (sa >> R_SHIFT)) << R_SHIFT; + p = (uint32_t)combine_disjoint_out_part (da, (uint8_t) (sa >> A_SHIFT)) << A_SHIFT; + Fb = m | n | o | p; + break; + + case COMBINE_B_IN: + m = (uint32_t)combine_disjoint_in_part (da, (uint8_t) (sa >> 0)); + n = (uint32_t)combine_disjoint_in_part (da, (uint8_t) (sa >> G_SHIFT)) << G_SHIFT; + o = (uint32_t)combine_disjoint_in_part (da, (uint8_t) (sa >> R_SHIFT)) << R_SHIFT; + p = (uint32_t)combine_disjoint_in_part (da, (uint8_t) (sa >> A_SHIFT)) << A_SHIFT; + Fb = m | n | o | p; + break; + + case COMBINE_B: + Fb = ~0; + break; + } + m = GENERIC (s, d, 0, GET_COMP (Fa, 0), GET_COMP (Fb, 0), t, u, v); + n = GENERIC (s, d, G_SHIFT, GET_COMP (Fa, G_SHIFT), GET_COMP (Fb, G_SHIFT), t, u, v); + o = GENERIC (s, d, R_SHIFT, GET_COMP (Fa, R_SHIFT), GET_COMP (Fb, R_SHIFT), t, u, v); + p = GENERIC (s, d, A_SHIFT, GET_COMP (Fa, A_SHIFT), GET_COMP (Fb, A_SHIFT), t, u, v); + + s = m | n | o | p; + + *(dest + i) = s; + } +} + +static void +combine_disjoint_over_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_OVER); +} + +static void +combine_disjoint_in_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_IN); +} + +static void +combine_disjoint_in_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_IN); +} + +static void +combine_disjoint_out_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_OUT); +} + +static void +combine_disjoint_out_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_OUT); +} + +static void +combine_disjoint_atop_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_ATOP); +} + +static void +combine_disjoint_atop_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_ATOP); +} + +static void +combine_disjoint_xor_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_XOR); +} + +static void +combine_conjoint_general_ca (uint32_t * dest, + const uint32_t *src, + const uint32_t *mask, + int width, + uint8_t combine) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint32_t s, d; + uint32_t m, n, o, p; + uint32_t Fa, Fb; + uint16_t t, u, v; + uint32_t sa; + uint8_t da; + + s = *(src + i); + m = *(mask + i); + d = *(dest + i); + da = d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + sa = m; + + switch (combine & COMBINE_A) + { + default: + Fa = 0; + break; + + case COMBINE_A_OUT: + m = (uint32_t)combine_conjoint_out_part ((uint8_t) (sa >> 0), da); + n = (uint32_t)combine_conjoint_out_part ((uint8_t) (sa >> G_SHIFT), da) << G_SHIFT; + o = (uint32_t)combine_conjoint_out_part ((uint8_t) (sa >> R_SHIFT), da) << R_SHIFT; + p = (uint32_t)combine_conjoint_out_part ((uint8_t) (sa >> A_SHIFT), da) << A_SHIFT; + Fa = m | n | o | p; + break; + + case COMBINE_A_IN: + m = (uint32_t)combine_conjoint_in_part ((uint8_t) (sa >> 0), da); + n = (uint32_t)combine_conjoint_in_part ((uint8_t) (sa >> G_SHIFT), da) << G_SHIFT; + o = (uint32_t)combine_conjoint_in_part ((uint8_t) (sa >> R_SHIFT), da) << R_SHIFT; + p = (uint32_t)combine_conjoint_in_part ((uint8_t) (sa >> A_SHIFT), da) << A_SHIFT; + Fa = m | n | o | p; + break; + + case COMBINE_A: + Fa = ~0; + break; + } + + switch (combine & COMBINE_B) + { + default: + Fb = 0; + break; + + case COMBINE_B_OUT: + m = (uint32_t)combine_conjoint_out_part (da, (uint8_t) (sa >> 0)); + n = (uint32_t)combine_conjoint_out_part (da, (uint8_t) (sa >> G_SHIFT)) << G_SHIFT; + o = (uint32_t)combine_conjoint_out_part (da, (uint8_t) (sa >> R_SHIFT)) << R_SHIFT; + p = (uint32_t)combine_conjoint_out_part (da, (uint8_t) (sa >> A_SHIFT)) << A_SHIFT; + Fb = m | n | o | p; + break; + + case COMBINE_B_IN: + m = (uint32_t)combine_conjoint_in_part (da, (uint8_t) (sa >> 0)); + n = (uint32_t)combine_conjoint_in_part (da, (uint8_t) (sa >> G_SHIFT)) << G_SHIFT; + o = (uint32_t)combine_conjoint_in_part (da, (uint8_t) (sa >> R_SHIFT)) << R_SHIFT; + p = (uint32_t)combine_conjoint_in_part (da, (uint8_t) (sa >> A_SHIFT)) << A_SHIFT; + Fb = m | n | o | p; + break; + + case COMBINE_B: + Fb = ~0; + break; + } + m = GENERIC (s, d, 0, GET_COMP (Fa, 0), GET_COMP (Fb, 0), t, u, v); + n = GENERIC (s, d, G_SHIFT, GET_COMP (Fa, G_SHIFT), GET_COMP (Fb, G_SHIFT), t, u, v); + o = GENERIC (s, d, R_SHIFT, GET_COMP (Fa, R_SHIFT), GET_COMP (Fb, R_SHIFT), t, u, v); + p = GENERIC (s, d, A_SHIFT, GET_COMP (Fa, A_SHIFT), GET_COMP (Fb, A_SHIFT), t, u, v); + + s = m | n | o | p; + + *(dest + i) = s; + } +} + +static void +combine_conjoint_over_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_OVER); +} + +static void +combine_conjoint_over_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_OVER); +} + +static void +combine_conjoint_in_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_IN); +} + +static void +combine_conjoint_in_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_IN); +} + +static void +combine_conjoint_out_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_OUT); +} + +static void +combine_conjoint_out_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_OUT); +} + +static void +combine_conjoint_atop_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_ATOP); +} + +static void +combine_conjoint_atop_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_ATOP); +} + +static void +combine_conjoint_xor_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_XOR); +} + +void +_pixman_setup_combiner_functions_32 (pixman_implementation_t *imp) +{ + /* Unified alpha */ + imp->combine_32[PIXMAN_OP_CLEAR] = combine_clear; + imp->combine_32[PIXMAN_OP_SRC] = combine_src_u; + imp->combine_32[PIXMAN_OP_DST] = combine_dst; + imp->combine_32[PIXMAN_OP_OVER] = combine_over_u; + imp->combine_32[PIXMAN_OP_OVER_REVERSE] = combine_over_reverse_u; + imp->combine_32[PIXMAN_OP_IN] = combine_in_u; + imp->combine_32[PIXMAN_OP_IN_REVERSE] = combine_in_reverse_u; + imp->combine_32[PIXMAN_OP_OUT] = combine_out_u; + imp->combine_32[PIXMAN_OP_OUT_REVERSE] = combine_out_reverse_u; + imp->combine_32[PIXMAN_OP_ATOP] = combine_atop_u; + imp->combine_32[PIXMAN_OP_ATOP_REVERSE] = combine_atop_reverse_u; + imp->combine_32[PIXMAN_OP_XOR] = combine_xor_u; + imp->combine_32[PIXMAN_OP_ADD] = combine_add_u; + imp->combine_32[PIXMAN_OP_SATURATE] = combine_saturate_u; + + /* Disjoint, unified */ + imp->combine_32[PIXMAN_OP_DISJOINT_CLEAR] = combine_clear; + imp->combine_32[PIXMAN_OP_DISJOINT_SRC] = combine_src_u; + imp->combine_32[PIXMAN_OP_DISJOINT_DST] = combine_dst; + imp->combine_32[PIXMAN_OP_DISJOINT_OVER] = combine_disjoint_over_u; + imp->combine_32[PIXMAN_OP_DISJOINT_OVER_REVERSE] = combine_saturate_u; + imp->combine_32[PIXMAN_OP_DISJOINT_IN] = combine_disjoint_in_u; + imp->combine_32[PIXMAN_OP_DISJOINT_IN_REVERSE] = combine_disjoint_in_reverse_u; + imp->combine_32[PIXMAN_OP_DISJOINT_OUT] = combine_disjoint_out_u; + imp->combine_32[PIXMAN_OP_DISJOINT_OUT_REVERSE] = combine_disjoint_out_reverse_u; + imp->combine_32[PIXMAN_OP_DISJOINT_ATOP] = combine_disjoint_atop_u; + imp->combine_32[PIXMAN_OP_DISJOINT_ATOP_REVERSE] = combine_disjoint_atop_reverse_u; + imp->combine_32[PIXMAN_OP_DISJOINT_XOR] = combine_disjoint_xor_u; + + /* Conjoint, unified */ + imp->combine_32[PIXMAN_OP_CONJOINT_CLEAR] = combine_clear; + imp->combine_32[PIXMAN_OP_CONJOINT_SRC] = combine_src_u; + imp->combine_32[PIXMAN_OP_CONJOINT_DST] = combine_dst; + imp->combine_32[PIXMAN_OP_CONJOINT_OVER] = combine_conjoint_over_u; + imp->combine_32[PIXMAN_OP_CONJOINT_OVER_REVERSE] = combine_conjoint_over_reverse_u; + imp->combine_32[PIXMAN_OP_CONJOINT_IN] = combine_conjoint_in_u; + imp->combine_32[PIXMAN_OP_CONJOINT_IN_REVERSE] = combine_conjoint_in_reverse_u; + imp->combine_32[PIXMAN_OP_CONJOINT_OUT] = combine_conjoint_out_u; + imp->combine_32[PIXMAN_OP_CONJOINT_OUT_REVERSE] = combine_conjoint_out_reverse_u; + imp->combine_32[PIXMAN_OP_CONJOINT_ATOP] = combine_conjoint_atop_u; + imp->combine_32[PIXMAN_OP_CONJOINT_ATOP_REVERSE] = combine_conjoint_atop_reverse_u; + imp->combine_32[PIXMAN_OP_CONJOINT_XOR] = combine_conjoint_xor_u; + + imp->combine_32[PIXMAN_OP_MULTIPLY] = combine_multiply_u; + imp->combine_32[PIXMAN_OP_SCREEN] = combine_screen_u; + imp->combine_32[PIXMAN_OP_OVERLAY] = combine_overlay_u; + imp->combine_32[PIXMAN_OP_DARKEN] = combine_darken_u; + imp->combine_32[PIXMAN_OP_LIGHTEN] = combine_lighten_u; + imp->combine_32[PIXMAN_OP_COLOR_DODGE] = combine_color_dodge_u; + imp->combine_32[PIXMAN_OP_COLOR_BURN] = combine_color_burn_u; + imp->combine_32[PIXMAN_OP_HARD_LIGHT] = combine_hard_light_u; + imp->combine_32[PIXMAN_OP_SOFT_LIGHT] = combine_soft_light_u; + imp->combine_32[PIXMAN_OP_DIFFERENCE] = combine_difference_u; + imp->combine_32[PIXMAN_OP_EXCLUSION] = combine_exclusion_u; + imp->combine_32[PIXMAN_OP_HSL_HUE] = combine_hsl_hue_u; + imp->combine_32[PIXMAN_OP_HSL_SATURATION] = combine_hsl_saturation_u; + imp->combine_32[PIXMAN_OP_HSL_COLOR] = combine_hsl_color_u; + imp->combine_32[PIXMAN_OP_HSL_LUMINOSITY] = combine_hsl_luminosity_u; + + /* Component alpha combiners */ + imp->combine_32_ca[PIXMAN_OP_CLEAR] = combine_clear_ca; + imp->combine_32_ca[PIXMAN_OP_SRC] = combine_src_ca; + /* dest */ + imp->combine_32_ca[PIXMAN_OP_OVER] = combine_over_ca; + imp->combine_32_ca[PIXMAN_OP_OVER_REVERSE] = combine_over_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_IN] = combine_in_ca; + imp->combine_32_ca[PIXMAN_OP_IN_REVERSE] = combine_in_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_OUT] = combine_out_ca; + imp->combine_32_ca[PIXMAN_OP_OUT_REVERSE] = combine_out_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_ATOP] = combine_atop_ca; + imp->combine_32_ca[PIXMAN_OP_ATOP_REVERSE] = combine_atop_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_XOR] = combine_xor_ca; + imp->combine_32_ca[PIXMAN_OP_ADD] = combine_add_ca; + imp->combine_32_ca[PIXMAN_OP_SATURATE] = combine_saturate_ca; + + /* Disjoint CA */ + imp->combine_32_ca[PIXMAN_OP_DISJOINT_CLEAR] = combine_clear_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_SRC] = combine_src_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_DST] = combine_dst; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_OVER] = combine_disjoint_over_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_OVER_REVERSE] = combine_saturate_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_IN] = combine_disjoint_in_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_IN_REVERSE] = combine_disjoint_in_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_OUT] = combine_disjoint_out_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_OUT_REVERSE] = combine_disjoint_out_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_ATOP] = combine_disjoint_atop_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_ATOP_REVERSE] = combine_disjoint_atop_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_DISJOINT_XOR] = combine_disjoint_xor_ca; + + /* Conjoint CA */ + imp->combine_32_ca[PIXMAN_OP_CONJOINT_CLEAR] = combine_clear_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_SRC] = combine_src_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_DST] = combine_dst; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_OVER] = combine_conjoint_over_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_OVER_REVERSE] = combine_conjoint_over_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_IN] = combine_conjoint_in_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_IN_REVERSE] = combine_conjoint_in_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_OUT] = combine_conjoint_out_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_OUT_REVERSE] = combine_conjoint_out_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_ATOP] = combine_conjoint_atop_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_ATOP_REVERSE] = combine_conjoint_atop_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_CONJOINT_XOR] = combine_conjoint_xor_ca; + + imp->combine_32_ca[PIXMAN_OP_MULTIPLY] = combine_multiply_ca; + imp->combine_32_ca[PIXMAN_OP_SCREEN] = combine_screen_ca; + imp->combine_32_ca[PIXMAN_OP_OVERLAY] = combine_overlay_ca; + imp->combine_32_ca[PIXMAN_OP_DARKEN] = combine_darken_ca; + imp->combine_32_ca[PIXMAN_OP_LIGHTEN] = combine_lighten_ca; + imp->combine_32_ca[PIXMAN_OP_COLOR_DODGE] = combine_color_dodge_ca; + imp->combine_32_ca[PIXMAN_OP_COLOR_BURN] = combine_color_burn_ca; + imp->combine_32_ca[PIXMAN_OP_HARD_LIGHT] = combine_hard_light_ca; + imp->combine_32_ca[PIXMAN_OP_SOFT_LIGHT] = combine_soft_light_ca; + imp->combine_32_ca[PIXMAN_OP_DIFFERENCE] = combine_difference_ca; + imp->combine_32_ca[PIXMAN_OP_EXCLUSION] = combine_exclusion_ca; + + /* It is not clear that these make sense, so make them noops for now */ + imp->combine_32_ca[PIXMAN_OP_HSL_HUE] = combine_dst; + imp->combine_32_ca[PIXMAN_OP_HSL_SATURATION] = combine_dst; + imp->combine_32_ca[PIXMAN_OP_HSL_COLOR] = combine_dst; + imp->combine_32_ca[PIXMAN_OP_HSL_LUMINOSITY] = combine_dst; +} + diff --git a/programs/develop/libraries/pixman/pixman-combine32.h b/programs/develop/libraries/pixman/pixman-combine32.h new file mode 100644 index 0000000000..68bde4281a --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-combine32.h @@ -0,0 +1,230 @@ +/* WARNING: This file is generated by combine.pl from combine.inc. + Please edit one of those files rather than this one. */ + +#line 1 "pixman-combine.c.template" + +#define COMPONENT_SIZE 8 +#define MASK 0xff +#define ONE_HALF 0x80 + +#define A_SHIFT 8 * 3 +#define R_SHIFT 8 * 2 +#define G_SHIFT 8 +#define A_MASK 0xff000000 +#define R_MASK 0xff0000 +#define G_MASK 0xff00 + +#define RB_MASK 0xff00ff +#define AG_MASK 0xff00ff00 +#define RB_ONE_HALF 0x800080 +#define RB_MASK_PLUS_ONE 0x10000100 + +#define ALPHA_8(x) ((x) >> A_SHIFT) +#define RED_8(x) (((x) >> R_SHIFT) & MASK) +#define GREEN_8(x) (((x) >> G_SHIFT) & MASK) +#define BLUE_8(x) ((x) & MASK) + +/* + * Helper macros. + */ + +#define MUL_UN8(a, b, t) \ + ((t) = (a) * (b) + ONE_HALF, ((((t) >> G_SHIFT ) + (t) ) >> G_SHIFT )) + +#define DIV_UN8(a, b) \ + (((uint16_t) (a) * MASK) / (b)) + +#define ADD_UN8(x, y, t) \ + ((t) = (x) + (y), \ + (uint32_t) (uint8_t) ((t) | (0 - ((t) >> G_SHIFT)))) + +#define DIV_ONE_UN8(x) \ + (((x) + ONE_HALF + (((x) + ONE_HALF) >> G_SHIFT)) >> G_SHIFT) + +/* + * The methods below use some tricks to be able to do two color + * components at the same time. + */ + +/* + * x_rb = (x_rb * a) / 255 + */ +#define UN8_rb_MUL_UN8(x, a, t) \ + do \ + { \ + t = ((x) & RB_MASK) * (a); \ + t += RB_ONE_HALF; \ + x = (t + ((t >> G_SHIFT) & RB_MASK)) >> G_SHIFT; \ + x &= RB_MASK; \ + } while (0) + +/* + * x_rb = min (x_rb + y_rb, 255) + */ +#define UN8_rb_ADD_UN8_rb(x, y, t) \ + do \ + { \ + t = ((x) + (y)); \ + t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK); \ + x = (t & RB_MASK); \ + } while (0) + +/* + * x_rb = (x_rb * a_rb) / 255 + */ +#define UN8_rb_MUL_UN8_rb(x, a, t) \ + do \ + { \ + t = (x & MASK) * (a & MASK); \ + t |= (x & R_MASK) * ((a >> R_SHIFT) & MASK); \ + t += RB_ONE_HALF; \ + t = (t + ((t >> G_SHIFT) & RB_MASK)) >> G_SHIFT; \ + x = t & RB_MASK; \ + } while (0) + +/* + * x_c = (x_c * a) / 255 + */ +#define UN8x4_MUL_UN8(x, a) \ + do \ + { \ + uint32_t r1__, r2__, t__; \ + \ + r1__ = (x); \ + UN8_rb_MUL_UN8 (r1__, (a), t__); \ + \ + r2__ = (x) >> G_SHIFT; \ + UN8_rb_MUL_UN8 (r2__, (a), t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a) / 255 + y_c + */ +#define UN8x4_MUL_UN8_ADD_UN8x4(x, a, y) \ + do \ + { \ + uint32_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (y) & RB_MASK; \ + UN8_rb_MUL_UN8 (r1__, (a), t__); \ + UN8_rb_ADD_UN8_rb (r1__, r2__, t__); \ + \ + r2__ = (x) >> G_SHIFT; \ + r3__ = ((y) >> G_SHIFT) & RB_MASK; \ + UN8_rb_MUL_UN8 (r2__, (a), t__); \ + UN8_rb_ADD_UN8_rb (r2__, r3__, t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a + y_c * b) / 255 + */ +#define UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8(x, a, y, b) \ + do \ + { \ + uint32_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (y); \ + UN8_rb_MUL_UN8 (r1__, (a), t__); \ + UN8_rb_MUL_UN8 (r2__, (b), t__); \ + UN8_rb_ADD_UN8_rb (r1__, r2__, t__); \ + \ + r2__ = ((x) >> G_SHIFT); \ + r3__ = ((y) >> G_SHIFT); \ + UN8_rb_MUL_UN8 (r2__, (a), t__); \ + UN8_rb_MUL_UN8 (r3__, (b), t__); \ + UN8_rb_ADD_UN8_rb (r2__, r3__, t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a_c) / 255 + */ +#define UN8x4_MUL_UN8x4(x, a) \ + do \ + { \ + uint32_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (a); \ + UN8_rb_MUL_UN8_rb (r1__, r2__, t__); \ + \ + r2__ = (x) >> G_SHIFT; \ + r3__ = (a) >> G_SHIFT; \ + UN8_rb_MUL_UN8_rb (r2__, r3__, t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a_c) / 255 + y_c + */ +#define UN8x4_MUL_UN8x4_ADD_UN8x4(x, a, y) \ + do \ + { \ + uint32_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (a); \ + UN8_rb_MUL_UN8_rb (r1__, r2__, t__); \ + r2__ = (y) & RB_MASK; \ + UN8_rb_ADD_UN8_rb (r1__, r2__, t__); \ + \ + r2__ = ((x) >> G_SHIFT); \ + r3__ = ((a) >> G_SHIFT); \ + UN8_rb_MUL_UN8_rb (r2__, r3__, t__); \ + r3__ = ((y) >> G_SHIFT) & RB_MASK; \ + UN8_rb_ADD_UN8_rb (r2__, r3__, t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a_c + y_c * b) / 255 + */ +#define UN8x4_MUL_UN8x4_ADD_UN8x4_MUL_UN8(x, a, y, b) \ + do \ + { \ + uint32_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (a); \ + UN8_rb_MUL_UN8_rb (r1__, r2__, t__); \ + r2__ = (y); \ + UN8_rb_MUL_UN8 (r2__, (b), t__); \ + UN8_rb_ADD_UN8_rb (r1__, r2__, t__); \ + \ + r2__ = (x) >> G_SHIFT; \ + r3__ = (a) >> G_SHIFT; \ + UN8_rb_MUL_UN8_rb (r2__, r3__, t__); \ + r3__ = (y) >> G_SHIFT; \ + UN8_rb_MUL_UN8 (r3__, (b), t__); \ + UN8_rb_ADD_UN8_rb (r2__, r3__, t__); \ + \ + x = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + x_c = min(x_c + y_c, 255) +*/ +#define UN8x4_ADD_UN8x4(x, y) \ + do \ + { \ + uint32_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x) & RB_MASK; \ + r2__ = (y) & RB_MASK; \ + UN8_rb_ADD_UN8_rb (r1__, r2__, t__); \ + \ + r2__ = ((x) >> G_SHIFT) & RB_MASK; \ + r3__ = ((y) >> G_SHIFT) & RB_MASK; \ + UN8_rb_ADD_UN8_rb (r2__, r3__, t__); \ + \ + x = r1__ | (r2__ << G_SHIFT); \ + } while (0) diff --git a/programs/develop/libraries/pixman/pixman-combine64.c b/programs/develop/libraries/pixman/pixman-combine64.c new file mode 100644 index 0000000000..850afba6ad --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-combine64.c @@ -0,0 +1,2465 @@ +/* WARNING: This file is generated by combine.pl from combine.inc. + Please edit one of those files rather than this one. */ + +#line 1 "pixman-combine.c.template" +#ifdef HAVE_CONFIG_H +#include +#endif + +#include +#include + +#include "pixman-private.h" + +#include "pixman-combine64.h" + +/*** per channel helper functions ***/ + +static void +combine_mask_ca (uint64_t *src, uint64_t *mask) +{ + uint64_t a = *mask; + + uint64_t x; + uint32_t xa; + + if (!a) + { + *(src) = 0; + return; + } + + x = *(src); + if (a == ~0) + { + x = x >> A_SHIFT; + x |= x << G_SHIFT; + x |= x << R_SHIFT; + *(mask) = x; + return; + } + + xa = x >> A_SHIFT; + UN16x4_MUL_UN16x4 (x, a); + *(src) = x; + + UN16x4_MUL_UN16 (a, xa); + *(mask) = a; +} + +static void +combine_mask_value_ca (uint64_t *src, const uint64_t *mask) +{ + uint64_t a = *mask; + uint64_t x; + + if (!a) + { + *(src) = 0; + return; + } + + if (a == ~0) + return; + + x = *(src); + UN16x4_MUL_UN16x4 (x, a); + *(src) = x; +} + +static void +combine_mask_alpha_ca (const uint64_t *src, uint64_t *mask) +{ + uint64_t a = *(mask); + uint64_t x; + + if (!a) + return; + + x = *(src) >> A_SHIFT; + if (x == MASK) + return; + + if (a == ~0) + { + x |= x << G_SHIFT; + x |= x << R_SHIFT; + *(mask) = x; + return; + } + + UN16x4_MUL_UN16 (a, x); + *(mask) = a; +} + +/* + * There are two ways of handling alpha -- either as a single unified value or + * a separate value for each component, hence each macro must have two + * versions. The unified alpha version has a 'U' at the end of the name, + * the component version has a 'C'. Similarly, functions which deal with + * this difference will have two versions using the same convention. + */ + +/* + * All of the composing functions + */ + +static force_inline uint64_t +combine_mask (const uint64_t *src, const uint64_t *mask, int i) +{ + uint64_t s, m; + + if (mask) + { + m = *(mask + i) >> A_SHIFT; + + if (!m) + return 0; + } + + s = *(src + i); + + if (mask) + UN16x4_MUL_UN16 (s, m); + + return s; +} + +static void +combine_clear (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + memset (dest, 0, width * sizeof(uint64_t)); +} + +static void +combine_dst (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + return; +} + +static void +combine_src_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + if (!mask) + memcpy (dest, src, width * sizeof (uint64_t)); + else + { + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + + *(dest + i) = s; + } + } +} + +/* if the Src is opaque, call combine_src_u */ +static void +combine_over_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t ia = ALPHA_16 (~s); + + UN16x4_MUL_UN16_ADD_UN16x4 (d, ia, s); + *(dest + i) = d; + } +} + +/* if the Dst is opaque, this is a noop */ +static void +combine_over_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t ia = ALPHA_16 (~*(dest + i)); + UN16x4_MUL_UN16_ADD_UN16x4 (s, ia, d); + *(dest + i) = s; + } +} + +/* if the Dst is opaque, call combine_src_u */ +static void +combine_in_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t a = ALPHA_16 (*(dest + i)); + UN16x4_MUL_UN16 (s, a); + *(dest + i) = s; + } +} + +/* if the Src is opaque, this is a noop */ +static void +combine_in_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t a = ALPHA_16 (s); + UN16x4_MUL_UN16 (d, a); + *(dest + i) = d; + } +} + +/* if the Dst is opaque, call combine_clear */ +static void +combine_out_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t a = ALPHA_16 (~*(dest + i)); + UN16x4_MUL_UN16 (s, a); + *(dest + i) = s; + } +} + +/* if the Src is opaque, call combine_clear */ +static void +combine_out_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t a = ALPHA_16 (~s); + UN16x4_MUL_UN16 (d, a); + *(dest + i) = d; + } +} + +/* if the Src is opaque, call combine_in_u */ +/* if the Dst is opaque, call combine_over_u */ +/* if both the Src and Dst are opaque, call combine_src_u */ +static void +combine_atop_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t dest_a = ALPHA_16 (d); + uint64_t src_ia = ALPHA_16 (~s); + + UN16x4_MUL_UN16_ADD_UN16x4_MUL_UN16 (s, dest_a, d, src_ia); + *(dest + i) = s; + } +} + +/* if the Src is opaque, call combine_over_reverse_u */ +/* if the Dst is opaque, call combine_in_reverse_u */ +/* if both the Src and Dst are opaque, call combine_dst_u */ +static void +combine_atop_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t src_a = ALPHA_16 (s); + uint64_t dest_ia = ALPHA_16 (~d); + + UN16x4_MUL_UN16_ADD_UN16x4_MUL_UN16 (s, dest_ia, d, src_a); + *(dest + i) = s; + } +} + +/* if the Src is opaque, call combine_over_u */ +/* if the Dst is opaque, call combine_over_reverse_u */ +/* if both the Src and Dst are opaque, call combine_clear */ +static void +combine_xor_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t src_ia = ALPHA_16 (~s); + uint64_t dest_ia = ALPHA_16 (~d); + + UN16x4_MUL_UN16_ADD_UN16x4_MUL_UN16 (s, dest_ia, d, src_ia); + *(dest + i) = s; + } +} + +static void +combine_add_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + UN16x4_ADD_UN16x4 (d, s); + *(dest + i) = d; + } +} + +/* if the Src is opaque, call combine_add_u */ +/* if the Dst is opaque, call combine_add_u */ +/* if both the Src and Dst are opaque, call combine_add_u */ +static void +combine_saturate_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint32_t sa, da; + + sa = s >> A_SHIFT; + da = ~d >> A_SHIFT; + if (sa > da) + { + sa = DIV_UN16 (da, sa); + UN16x4_MUL_UN16 (s, sa); + } + ; + UN16x4_ADD_UN16x4 (d, s); + *(dest + i) = d; + } +} + +/* + * PDF blend modes: + * The following blend modes have been taken from the PDF ISO 32000 + * specification, which at this point in time is available from + * http://www.adobe.com/devnet/acrobat/pdfs/PDF32000_2008.pdf + * The relevant chapters are 11.3.5 and 11.3.6. + * The formula for computing the final pixel color given in 11.3.6 is: + * αr × Cr = (1 – αs) × αb × Cb + (1 – αb) × αs × Cs + αb × αs × B(Cb, Cs) + * with B() being the blend function. + * Note that OVER is a special case of this operation, using B(Cb, Cs) = Cs + * + * These blend modes should match the SVG filter draft specification, as + * it has been designed to mirror ISO 32000. Note that at the current point + * no released draft exists that shows this, as the formulas have not been + * updated yet after the release of ISO 32000. + * + * The default implementation here uses the PDF_SEPARABLE_BLEND_MODE and + * PDF_NON_SEPARABLE_BLEND_MODE macros, which take the blend function as an + * argument. Note that this implementation operates on premultiplied colors, + * while the PDF specification does not. Therefore the code uses the formula + * ar.Cra = (1 – as) . Dca + (1 – ad) . Sca + B(Dca, ad, Sca, as) + */ + +/* + * Multiply + * B(Dca, ad, Sca, as) = Dca.Sca + */ + +static void +combine_multiply_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t ss = s; + uint64_t src_ia = ALPHA_16 (~s); + uint64_t dest_ia = ALPHA_16 (~d); + + UN16x4_MUL_UN16_ADD_UN16x4_MUL_UN16 (ss, dest_ia, d, src_ia); + UN16x4_MUL_UN16x4 (d, s); + UN16x4_ADD_UN16x4 (d, ss); + + *(dest + i) = d; + } +} + +static void +combine_multiply_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t m = *(mask + i); + uint64_t s = *(src + i); + uint64_t d = *(dest + i); + uint64_t r = d; + uint64_t dest_ia = ALPHA_16 (~d); + + combine_mask_value_ca (&s, &m); + + UN16x4_MUL_UN16x4_ADD_UN16x4_MUL_UN16 (r, ~m, s, dest_ia); + UN16x4_MUL_UN16x4 (d, s); + UN16x4_ADD_UN16x4 (r, d); + + *(dest + i) = r; + } +} + +#define PDF_SEPARABLE_BLEND_MODE(name) \ + static void \ + combine_ ## name ## _u (pixman_implementation_t *imp, \ + pixman_op_t op, \ + uint64_t * dest, \ + const uint64_t * src, \ + const uint64_t * mask, \ + int width) \ + { \ + int i; \ + for (i = 0; i < width; ++i) { \ + uint64_t s = combine_mask (src, mask, i); \ + uint64_t d = *(dest + i); \ + uint16_t sa = ALPHA_16 (s); \ + uint16_t isa = ~sa; \ + uint16_t da = ALPHA_16 (d); \ + uint16_t ida = ~da; \ + uint64_t result; \ + \ + result = d; \ + UN16x4_MUL_UN16_ADD_UN16x4_MUL_UN16 (result, isa, s, ida); \ + \ + *(dest + i) = result + \ + (DIV_ONE_UN16 (sa * da) << A_SHIFT) + \ + (blend_ ## name (RED_16 (d), da, RED_16 (s), sa) << R_SHIFT) + \ + (blend_ ## name (GREEN_16 (d), da, GREEN_16 (s), sa) << G_SHIFT) + \ + (blend_ ## name (BLUE_16 (d), da, BLUE_16 (s), sa)); \ + } \ + } \ + \ + static void \ + combine_ ## name ## _ca (pixman_implementation_t *imp, \ + pixman_op_t op, \ + uint64_t * dest, \ + const uint64_t * src, \ + const uint64_t * mask, \ + int width) \ + { \ + int i; \ + for (i = 0; i < width; ++i) { \ + uint64_t m = *(mask + i); \ + uint64_t s = *(src + i); \ + uint64_t d = *(dest + i); \ + uint16_t da = ALPHA_16 (d); \ + uint16_t ida = ~da; \ + uint64_t result; \ + \ + combine_mask_value_ca (&s, &m); \ + \ + result = d; \ + UN16x4_MUL_UN16x4_ADD_UN16x4_MUL_UN16 (result, ~m, s, ida); \ + \ + result += \ + (DIV_ONE_UN16 (ALPHA_16 (m) * da) << A_SHIFT) + \ + (blend_ ## name (RED_16 (d), da, RED_16 (s), RED_16 (m)) << R_SHIFT) + \ + (blend_ ## name (GREEN_16 (d), da, GREEN_16 (s), GREEN_16 (m)) << G_SHIFT) + \ + (blend_ ## name (BLUE_16 (d), da, BLUE_16 (s), BLUE_16 (m))); \ + \ + *(dest + i) = result; \ + } \ + } + +/* + * Screen + * B(Dca, ad, Sca, as) = Dca.sa + Sca.da - Dca.Sca + */ +static inline uint64_t +blend_screen (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + return DIV_ONE_UN16 (sca * da + dca * sa - sca * dca); +} + +PDF_SEPARABLE_BLEND_MODE (screen) + +/* + * Overlay + * B(Dca, Da, Sca, Sa) = + * if 2.Dca < Da + * 2.Sca.Dca + * otherwise + * Sa.Da - 2.(Da - Dca).(Sa - Sca) + */ +static inline uint64_t +blend_overlay (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + uint64_t rca; + + if (2 * dca < da) + rca = 2 * sca * dca; + else + rca = sa * da - 2 * (da - dca) * (sa - sca); + return DIV_ONE_UN16 (rca); +} + +PDF_SEPARABLE_BLEND_MODE (overlay) + +/* + * Darken + * B(Dca, Da, Sca, Sa) = min (Sca.Da, Dca.Sa) + */ +static inline uint64_t +blend_darken (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + uint64_t s, d; + + s = sca * da; + d = dca * sa; + return DIV_ONE_UN16 (s > d ? d : s); +} + +PDF_SEPARABLE_BLEND_MODE (darken) + +/* + * Lighten + * B(Dca, Da, Sca, Sa) = max (Sca.Da, Dca.Sa) + */ +static inline uint64_t +blend_lighten (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + uint64_t s, d; + + s = sca * da; + d = dca * sa; + return DIV_ONE_UN16 (s > d ? s : d); +} + +PDF_SEPARABLE_BLEND_MODE (lighten) + +/* + * Color dodge + * B(Dca, Da, Sca, Sa) = + * if Dca == 0 + * 0 + * if Sca == Sa + * Sa.Da + * otherwise + * Sa.Da. min (1, Dca / Da / (1 - Sca/Sa)) + */ +static inline uint64_t +blend_color_dodge (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + if (sca >= sa) + { + return dca == 0 ? 0 : DIV_ONE_UN16 (sa * da); + } + else + { + uint64_t rca = dca * sa / (sa - sca); + return DIV_ONE_UN16 (sa * MIN (rca, da)); + } +} + +PDF_SEPARABLE_BLEND_MODE (color_dodge) + +/* + * Color burn + * B(Dca, Da, Sca, Sa) = + * if Dca == Da + * Sa.Da + * if Sca == 0 + * 0 + * otherwise + * Sa.Da.(1 - min (1, (1 - Dca/Da).Sa / Sca)) + */ +static inline uint64_t +blend_color_burn (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + if (sca == 0) + { + return dca < da ? 0 : DIV_ONE_UN16 (sa * da); + } + else + { + uint64_t rca = (da - dca) * sa / sca; + return DIV_ONE_UN16 (sa * (MAX (rca, da) - rca)); + } +} + +PDF_SEPARABLE_BLEND_MODE (color_burn) + +/* + * Hard light + * B(Dca, Da, Sca, Sa) = + * if 2.Sca < Sa + * 2.Sca.Dca + * otherwise + * Sa.Da - 2.(Da - Dca).(Sa - Sca) + */ +static inline uint64_t +blend_hard_light (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + if (2 * sca < sa) + return DIV_ONE_UN16 (2 * sca * dca); + else + return DIV_ONE_UN16 (sa * da - 2 * (da - dca) * (sa - sca)); +} + +PDF_SEPARABLE_BLEND_MODE (hard_light) + +/* + * Soft light + * B(Dca, Da, Sca, Sa) = + * if (2.Sca <= Sa) + * Dca.(Sa - (1 - Dca/Da).(2.Sca - Sa)) + * otherwise if Dca.4 <= Da + * Dca.(Sa + (2.Sca - Sa).((16.Dca/Da - 12).Dca/Da + 3) + * otherwise + * (Dca.Sa + (SQRT (Dca/Da).Da - Dca).(2.Sca - Sa)) + */ +static inline uint64_t +blend_soft_light (uint64_t dca_org, + uint64_t da_org, + uint64_t sca_org, + uint64_t sa_org) +{ + double dca = dca_org * (1.0 / MASK); + double da = da_org * (1.0 / MASK); + double sca = sca_org * (1.0 / MASK); + double sa = sa_org * (1.0 / MASK); + double rca; + + if (2 * sca < sa) + { + if (da == 0) + rca = dca * sa; + else + rca = dca * sa - dca * (da - dca) * (sa - 2 * sca) / da; + } + else if (da == 0) + { + rca = 0; + } + else if (4 * dca <= da) + { + rca = dca * sa + + (2 * sca - sa) * dca * ((16 * dca / da - 12) * dca / da + 3); + } + else + { + rca = dca * sa + (sqrt (dca * da) - dca) * (2 * sca - sa); + } + return rca * MASK + 0.5; +} + +PDF_SEPARABLE_BLEND_MODE (soft_light) + +/* + * Difference + * B(Dca, Da, Sca, Sa) = abs (Dca.Sa - Sca.Da) + */ +static inline uint64_t +blend_difference (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + uint64_t dcasa = dca * sa; + uint64_t scada = sca * da; + + if (scada < dcasa) + return DIV_ONE_UN16 (dcasa - scada); + else + return DIV_ONE_UN16 (scada - dcasa); +} + +PDF_SEPARABLE_BLEND_MODE (difference) + +/* + * Exclusion + * B(Dca, Da, Sca, Sa) = (Sca.Da + Dca.Sa - 2.Sca.Dca) + */ + +/* This can be made faster by writing it directly and not using + * PDF_SEPARABLE_BLEND_MODE, but that's a performance optimization */ + +static inline uint64_t +blend_exclusion (uint64_t dca, uint64_t da, uint64_t sca, uint64_t sa) +{ + return DIV_ONE_UN16 (sca * da + dca * sa - 2 * dca * sca); +} + +PDF_SEPARABLE_BLEND_MODE (exclusion) + +#undef PDF_SEPARABLE_BLEND_MODE + +/* + * PDF nonseperable blend modes are implemented using the following functions + * to operate in Hsl space, with Cmax, Cmid, Cmin referring to the max, mid + * and min value of the red, green and blue components. + * + * LUM (C) = 0.3 × Cred + 0.59 × Cgreen + 0.11 × Cblue + * + * clip_color (C): + * l = LUM (C) + * min = Cmin + * max = Cmax + * if n < 0.0 + * C = l + ( ( ( C – l ) × l ) ⁄ ( l – min ) ) + * if x > 1.0 + * C = l + ( ( ( C – l ) × ( 1 – l ) ) ⁄ ( max – l ) ) + * return C + * + * set_lum (C, l): + * d = l – LUM (C) + * C += d + * return clip_color (C) + * + * SAT (C) = CH_MAX (C) - CH_MIN (C) + * + * set_sat (C, s): + * if Cmax > Cmin + * Cmid = ( ( ( Cmid – Cmin ) × s ) ⁄ ( Cmax – Cmin ) ) + * Cmax = s + * else + * Cmid = Cmax = 0.0 + * Cmin = 0.0 + * return C + */ + +/* For premultiplied colors, we need to know what happens when C is + * multiplied by a real number. LUM and SAT are linear: + * + * LUM (r × C) = r × LUM (C) SAT (r * C) = r * SAT (C) + * + * If we extend clip_color with an extra argument a and change + * + * if x >= 1.0 + * + * into + * + * if x >= a + * + * then clip_color is also linear: + * + * r * clip_color (C, a) = clip_color (r_c, ra); + * + * for positive r. + * + * Similarly, we can extend set_lum with an extra argument that is just passed + * on to clip_color: + * + * r * set_lum ( C, l, a) + * + * = r × clip_color ( C + l - LUM (C), a) + * + * = clip_color ( r * C + r × l - r * LUM (C), r * a) + * + * = set_lum ( r * C, r * l, r * a) + * + * Finally, set_sat: + * + * r * set_sat (C, s) = set_sat (x * C, r * s) + * + * The above holds for all non-zero x, because they x'es in the fraction for + * C_mid cancel out. Specifically, it holds for x = r: + * + * r * set_sat (C, s) = set_sat (r_c, rs) + * + */ + +/* So, for the non-separable PDF blend modes, we have (using s, d for + * non-premultiplied colors, and S, D for premultiplied: + * + * Color: + * + * a_s * a_d * B(s, d) + * = a_s * a_d * set_lum (S/a_s, LUM (D/a_d), 1) + * = set_lum (S * a_d, a_s * LUM (D), a_s * a_d) + * + * + * Luminosity: + * + * a_s * a_d * B(s, d) + * = a_s * a_d * set_lum (D/a_d, LUM(S/a_s), 1) + * = set_lum (a_s * D, a_d * LUM(S), a_s * a_d) + * + * + * Saturation: + * + * a_s * a_d * B(s, d) + * = a_s * a_d * set_lum (set_sat (D/a_d, SAT (S/a_s)), LUM (D/a_d), 1) + * = set_lum (a_s * a_d * set_sat (D/a_d, SAT (S/a_s)), + * a_s * LUM (D), a_s * a_d) + * = set_lum (set_sat (a_s * D, a_d * SAT (S), a_s * LUM (D), a_s * a_d)) + * + * Hue: + * + * a_s * a_d * B(s, d) + * = a_s * a_d * set_lum (set_sat (S/a_s, SAT (D/a_d)), LUM (D/a_d), 1) + * = a_s * a_d * set_lum (set_sat (a_d * S, a_s * SAT (D)), + * a_s * LUM (D), a_s * a_d) + * + */ + +#define CH_MIN(c) (c[0] < c[1] ? (c[0] < c[2] ? c[0] : c[2]) : (c[1] < c[2] ? c[1] : c[2])) +#define CH_MAX(c) (c[0] > c[1] ? (c[0] > c[2] ? c[0] : c[2]) : (c[1] > c[2] ? c[1] : c[2])) +#define LUM(c) ((c[0] * 30 + c[1] * 59 + c[2] * 11) / 100) +#define SAT(c) (CH_MAX (c) - CH_MIN (c)) + +#define PDF_NON_SEPARABLE_BLEND_MODE(name) \ + static void \ + combine_ ## name ## _u (pixman_implementation_t *imp, \ + pixman_op_t op, \ + uint64_t *dest, \ + const uint64_t *src, \ + const uint64_t *mask, \ + int width) \ + { \ + int i; \ + for (i = 0; i < width; ++i) \ + { \ + uint64_t s = combine_mask (src, mask, i); \ + uint64_t d = *(dest + i); \ + uint16_t sa = ALPHA_16 (s); \ + uint16_t isa = ~sa; \ + uint16_t da = ALPHA_16 (d); \ + uint16_t ida = ~da; \ + uint64_t result; \ + uint64_t sc[3], dc[3], c[3]; \ + \ + result = d; \ + UN16x4_MUL_UN16_ADD_UN16x4_MUL_UN16 (result, isa, s, ida); \ + dc[0] = RED_16 (d); \ + sc[0] = RED_16 (s); \ + dc[1] = GREEN_16 (d); \ + sc[1] = GREEN_16 (s); \ + dc[2] = BLUE_16 (d); \ + sc[2] = BLUE_16 (s); \ + blend_ ## name (c, dc, da, sc, sa); \ + \ + *(dest + i) = result + \ + (DIV_ONE_UN16 (sa * da) << A_SHIFT) + \ + (DIV_ONE_UN16 (c[0]) << R_SHIFT) + \ + (DIV_ONE_UN16 (c[1]) << G_SHIFT) + \ + (DIV_ONE_UN16 (c[2])); \ + } \ + } + +static void +set_lum (uint64_t dest[3], uint64_t src[3], uint64_t sa, uint64_t lum) +{ + double a, l, min, max; + double tmp[3]; + + a = sa * (1.0 / MASK); + + l = lum * (1.0 / MASK); + tmp[0] = src[0] * (1.0 / MASK); + tmp[1] = src[1] * (1.0 / MASK); + tmp[2] = src[2] * (1.0 / MASK); + + l = l - LUM (tmp); + tmp[0] += l; + tmp[1] += l; + tmp[2] += l; + + /* clip_color */ + l = LUM (tmp); + min = CH_MIN (tmp); + max = CH_MAX (tmp); + + if (min < 0) + { + if (l - min == 0.0) + { + tmp[0] = 0; + tmp[1] = 0; + tmp[2] = 0; + } + else + { + tmp[0] = l + (tmp[0] - l) * l / (l - min); + tmp[1] = l + (tmp[1] - l) * l / (l - min); + tmp[2] = l + (tmp[2] - l) * l / (l - min); + } + } + if (max > a) + { + if (max - l == 0.0) + { + tmp[0] = a; + tmp[1] = a; + tmp[2] = a; + } + else + { + tmp[0] = l + (tmp[0] - l) * (a - l) / (max - l); + tmp[1] = l + (tmp[1] - l) * (a - l) / (max - l); + tmp[2] = l + (tmp[2] - l) * (a - l) / (max - l); + } + } + + dest[0] = tmp[0] * MASK + 0.5; + dest[1] = tmp[1] * MASK + 0.5; + dest[2] = tmp[2] * MASK + 0.5; +} + +static void +set_sat (uint64_t dest[3], uint64_t src[3], uint64_t sat) +{ + int id[3]; + uint64_t min, max; + + if (src[0] > src[1]) + { + if (src[0] > src[2]) + { + id[0] = 0; + if (src[1] > src[2]) + { + id[1] = 1; + id[2] = 2; + } + else + { + id[1] = 2; + id[2] = 1; + } + } + else + { + id[0] = 2; + id[1] = 0; + id[2] = 1; + } + } + else + { + if (src[0] > src[2]) + { + id[0] = 1; + id[1] = 0; + id[2] = 2; + } + else + { + id[2] = 0; + if (src[1] > src[2]) + { + id[0] = 1; + id[1] = 2; + } + else + { + id[0] = 2; + id[1] = 1; + } + } + } + + max = dest[id[0]]; + min = dest[id[2]]; + if (max > min) + { + dest[id[1]] = (dest[id[1]] - min) * sat / (max - min); + dest[id[0]] = sat; + dest[id[2]] = 0; + } + else + { + dest[0] = dest[1] = dest[2] = 0; + } +} + +/* + * Hue: + * B(Cb, Cs) = set_lum (set_sat (Cs, SAT (Cb)), LUM (Cb)) + */ +static inline void +blend_hsl_hue (uint64_t c[3], + uint64_t dc[3], + uint64_t da, + uint64_t sc[3], + uint64_t sa) +{ + c[0] = sc[0] * da; + c[1] = sc[1] * da; + c[2] = sc[2] * da; + set_sat (c, c, SAT (dc) * sa); + set_lum (c, c, sa * da, LUM (dc) * sa); +} + +PDF_NON_SEPARABLE_BLEND_MODE (hsl_hue) + +/* + * Saturation: + * B(Cb, Cs) = set_lum (set_sat (Cb, SAT (Cs)), LUM (Cb)) + */ +static inline void +blend_hsl_saturation (uint64_t c[3], + uint64_t dc[3], + uint64_t da, + uint64_t sc[3], + uint64_t sa) +{ + c[0] = dc[0] * sa; + c[1] = dc[1] * sa; + c[2] = dc[2] * sa; + set_sat (c, c, SAT (sc) * da); + set_lum (c, c, sa * da, LUM (dc) * sa); +} + +PDF_NON_SEPARABLE_BLEND_MODE (hsl_saturation) + +/* + * Color: + * B(Cb, Cs) = set_lum (Cs, LUM (Cb)) + */ +static inline void +blend_hsl_color (uint64_t c[3], + uint64_t dc[3], + uint64_t da, + uint64_t sc[3], + uint64_t sa) +{ + c[0] = sc[0] * da; + c[1] = sc[1] * da; + c[2] = sc[2] * da; + set_lum (c, c, sa * da, LUM (dc) * sa); +} + +PDF_NON_SEPARABLE_BLEND_MODE (hsl_color) + +/* + * Luminosity: + * B(Cb, Cs) = set_lum (Cb, LUM (Cs)) + */ +static inline void +blend_hsl_luminosity (uint64_t c[3], + uint64_t dc[3], + uint64_t da, + uint64_t sc[3], + uint64_t sa) +{ + c[0] = dc[0] * sa; + c[1] = dc[1] * sa; + c[2] = dc[2] * sa; + set_lum (c, c, sa * da, LUM (sc) * da); +} + +PDF_NON_SEPARABLE_BLEND_MODE (hsl_luminosity) + +#undef SAT +#undef LUM +#undef CH_MAX +#undef CH_MIN +#undef PDF_NON_SEPARABLE_BLEND_MODE + +/* Overlay + * + * All of the disjoint composing functions + * + * The four entries in the first column indicate what source contributions + * come from each of the four areas of the picture -- areas covered by neither + * A nor B, areas covered only by A, areas covered only by B and finally + * areas covered by both A and B. + * + * Disjoint Conjoint + * Fa Fb Fa Fb + * (0,0,0,0) 0 0 0 0 + * (0,A,0,A) 1 0 1 0 + * (0,0,B,B) 0 1 0 1 + * (0,A,B,A) 1 min((1-a)/b,1) 1 max(1-a/b,0) + * (0,A,B,B) min((1-b)/a,1) 1 max(1-b/a,0) 1 + * (0,0,0,A) max(1-(1-b)/a,0) 0 min(1,b/a) 0 + * (0,0,0,B) 0 max(1-(1-a)/b,0) 0 min(a/b,1) + * (0,A,0,0) min(1,(1-b)/a) 0 max(1-b/a,0) 0 + * (0,0,B,0) 0 min(1,(1-a)/b) 0 max(1-a/b,0) + * (0,0,B,A) max(1-(1-b)/a,0) min(1,(1-a)/b) min(1,b/a) max(1-a/b,0) + * (0,A,0,B) min(1,(1-b)/a) max(1-(1-a)/b,0) max(1-b/a,0) min(1,a/b) + * (0,A,B,0) min(1,(1-b)/a) min(1,(1-a)/b) max(1-b/a,0) max(1-a/b,0) + */ + +#define COMBINE_A_OUT 1 +#define COMBINE_A_IN 2 +#define COMBINE_B_OUT 4 +#define COMBINE_B_IN 8 + +#define COMBINE_CLEAR 0 +#define COMBINE_A (COMBINE_A_OUT | COMBINE_A_IN) +#define COMBINE_B (COMBINE_B_OUT | COMBINE_B_IN) +#define COMBINE_A_OVER (COMBINE_A_OUT | COMBINE_B_OUT | COMBINE_A_IN) +#define COMBINE_B_OVER (COMBINE_A_OUT | COMBINE_B_OUT | COMBINE_B_IN) +#define COMBINE_A_ATOP (COMBINE_B_OUT | COMBINE_A_IN) +#define COMBINE_B_ATOP (COMBINE_A_OUT | COMBINE_B_IN) +#define COMBINE_XOR (COMBINE_A_OUT | COMBINE_B_OUT) + +/* portion covered by a but not b */ +static uint16_t +combine_disjoint_out_part (uint16_t a, uint16_t b) +{ + /* min (1, (1-b) / a) */ + + b = ~b; /* 1 - b */ + if (b >= a) /* 1 - b >= a -> (1-b)/a >= 1 */ + return MASK; /* 1 */ + return DIV_UN16 (b, a); /* (1-b) / a */ +} + +/* portion covered by both a and b */ +static uint16_t +combine_disjoint_in_part (uint16_t a, uint16_t b) +{ + /* max (1-(1-b)/a,0) */ + /* = - min ((1-b)/a - 1, 0) */ + /* = 1 - min (1, (1-b)/a) */ + + b = ~b; /* 1 - b */ + if (b >= a) /* 1 - b >= a -> (1-b)/a >= 1 */ + return 0; /* 1 - 1 */ + return ~DIV_UN16(b, a); /* 1 - (1-b) / a */ +} + +/* portion covered by a but not b */ +static uint16_t +combine_conjoint_out_part (uint16_t a, uint16_t b) +{ + /* max (1-b/a,0) */ + /* = 1-min(b/a,1) */ + + /* min (1, (1-b) / a) */ + + if (b >= a) /* b >= a -> b/a >= 1 */ + return 0x00; /* 0 */ + return ~DIV_UN16(b, a); /* 1 - b/a */ +} + +/* portion covered by both a and b */ +static uint16_t +combine_conjoint_in_part (uint16_t a, uint16_t b) +{ + /* min (1,b/a) */ + + if (b >= a) /* b >= a -> b/a >= 1 */ + return MASK; /* 1 */ + return DIV_UN16 (b, a); /* b/a */ +} + +#define GET_COMP(v, i) ((uint32_t) (uint16_t) ((v) >> i)) + +#define ADD(x, y, i, t) \ + ((t) = GET_COMP (x, i) + GET_COMP (y, i), \ + (uint64_t) ((uint16_t) ((t) | (0 - ((t) >> G_SHIFT)))) << (i)) + +#define GENERIC(x, y, i, ax, ay, t, u, v) \ + ((t) = (MUL_UN16 (GET_COMP (y, i), ay, (u)) + \ + MUL_UN16 (GET_COMP (x, i), ax, (v))), \ + (uint64_t) ((uint16_t) ((t) | \ + (0 - ((t) >> G_SHIFT)))) << (i)) + +static void +combine_disjoint_general_u (uint64_t * dest, + const uint64_t *src, + const uint64_t *mask, + int width, + uint16_t combine) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t m, n, o, p; + uint32_t Fa, Fb, t, u, v; + uint16_t sa = s >> A_SHIFT; + uint16_t da = d >> A_SHIFT; + + switch (combine & COMBINE_A) + { + default: + Fa = 0; + break; + + case COMBINE_A_OUT: + Fa = combine_disjoint_out_part (sa, da); + break; + + case COMBINE_A_IN: + Fa = combine_disjoint_in_part (sa, da); + break; + + case COMBINE_A: + Fa = MASK; + break; + } + + switch (combine & COMBINE_B) + { + default: + Fb = 0; + break; + + case COMBINE_B_OUT: + Fb = combine_disjoint_out_part (da, sa); + break; + + case COMBINE_B_IN: + Fb = combine_disjoint_in_part (da, sa); + break; + + case COMBINE_B: + Fb = MASK; + break; + } + m = GENERIC (s, d, 0, Fa, Fb, t, u, v); + n = GENERIC (s, d, G_SHIFT, Fa, Fb, t, u, v); + o = GENERIC (s, d, R_SHIFT, Fa, Fb, t, u, v); + p = GENERIC (s, d, A_SHIFT, Fa, Fb, t, u, v); + s = m | n | o | p; + *(dest + i) = s; + } +} + +static void +combine_disjoint_over_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint32_t a = s >> A_SHIFT; + + if (s != 0x00) + { + uint64_t d = *(dest + i); + a = combine_disjoint_out_part (d >> A_SHIFT, a); + UN16x4_MUL_UN16_ADD_UN16x4 (d, a, s); + + *(dest + i) = d; + } + } +} + +static void +combine_disjoint_in_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_IN); +} + +static void +combine_disjoint_in_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_IN); +} + +static void +combine_disjoint_out_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_OUT); +} + +static void +combine_disjoint_out_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_OUT); +} + +static void +combine_disjoint_atop_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_A_ATOP); +} + +static void +combine_disjoint_atop_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_B_ATOP); +} + +static void +combine_disjoint_xor_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_u (dest, src, mask, width, COMBINE_XOR); +} + +static void +combine_conjoint_general_u (uint64_t * dest, + const uint64_t *src, + const uint64_t *mask, + int width, + uint16_t combine) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = combine_mask (src, mask, i); + uint64_t d = *(dest + i); + uint64_t m, n, o, p; + uint32_t Fa, Fb, t, u, v; + uint16_t sa = s >> A_SHIFT; + uint16_t da = d >> A_SHIFT; + + switch (combine & COMBINE_A) + { + default: + Fa = 0; + break; + + case COMBINE_A_OUT: + Fa = combine_conjoint_out_part (sa, da); + break; + + case COMBINE_A_IN: + Fa = combine_conjoint_in_part (sa, da); + break; + + case COMBINE_A: + Fa = MASK; + break; + } + + switch (combine & COMBINE_B) + { + default: + Fb = 0; + break; + + case COMBINE_B_OUT: + Fb = combine_conjoint_out_part (da, sa); + break; + + case COMBINE_B_IN: + Fb = combine_conjoint_in_part (da, sa); + break; + + case COMBINE_B: + Fb = MASK; + break; + } + + m = GENERIC (s, d, 0, Fa, Fb, t, u, v); + n = GENERIC (s, d, G_SHIFT, Fa, Fb, t, u, v); + o = GENERIC (s, d, R_SHIFT, Fa, Fb, t, u, v); + p = GENERIC (s, d, A_SHIFT, Fa, Fb, t, u, v); + + s = m | n | o | p; + + *(dest + i) = s; + } +} + +static void +combine_conjoint_over_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_OVER); +} + +static void +combine_conjoint_over_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_OVER); +} + +static void +combine_conjoint_in_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_IN); +} + +static void +combine_conjoint_in_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_IN); +} + +static void +combine_conjoint_out_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_OUT); +} + +static void +combine_conjoint_out_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_OUT); +} + +static void +combine_conjoint_atop_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_A_ATOP); +} + +static void +combine_conjoint_atop_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_B_ATOP); +} + +static void +combine_conjoint_xor_u (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_u (dest, src, mask, width, COMBINE_XOR); +} + +/************************************************************************/ +/*********************** Per Channel functions **************************/ +/************************************************************************/ + +static void +combine_clear_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + memset (dest, 0, width * sizeof(uint64_t)); +} + +static void +combine_src_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + + combine_mask_value_ca (&s, &m); + + *(dest + i) = s; + } +} + +static void +combine_over_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + uint64_t a; + + combine_mask_ca (&s, &m); + + a = ~m; + if (a) + { + uint64_t d = *(dest + i); + UN16x4_MUL_UN16x4_ADD_UN16x4 (d, a, s); + s = d; + } + + *(dest + i) = s; + } +} + +static void +combine_over_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t d = *(dest + i); + uint64_t a = ~d >> A_SHIFT; + + if (a) + { + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + + UN16x4_MUL_UN16x4 (s, m); + UN16x4_MUL_UN16_ADD_UN16x4 (s, a, d); + + *(dest + i) = s; + } + } +} + +static void +combine_in_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t d = *(dest + i); + uint32_t a = d >> A_SHIFT; + uint64_t s = 0; + + if (a) + { + uint64_t m = *(mask + i); + + s = *(src + i); + combine_mask_value_ca (&s, &m); + + if (a != MASK) + UN16x4_MUL_UN16 (s, a); + } + + *(dest + i) = s; + } +} + +static void +combine_in_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + uint64_t a; + + combine_mask_alpha_ca (&s, &m); + + a = m; + if (a != ~0) + { + uint64_t d = 0; + + if (a) + { + d = *(dest + i); + UN16x4_MUL_UN16x4 (d, a); + } + + *(dest + i) = d; + } + } +} + +static void +combine_out_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t d = *(dest + i); + uint32_t a = ~d >> A_SHIFT; + uint64_t s = 0; + + if (a) + { + uint64_t m = *(mask + i); + + s = *(src + i); + combine_mask_value_ca (&s, &m); + + if (a != MASK) + UN16x4_MUL_UN16 (s, a); + } + + *(dest + i) = s; + } +} + +static void +combine_out_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + uint64_t a; + + combine_mask_alpha_ca (&s, &m); + + a = ~m; + if (a != ~0) + { + uint64_t d = 0; + + if (a) + { + d = *(dest + i); + UN16x4_MUL_UN16x4 (d, a); + } + + *(dest + i) = d; + } + } +} + +static void +combine_atop_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t d = *(dest + i); + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + uint64_t ad; + uint32_t as = d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + ad = ~m; + + UN16x4_MUL_UN16x4_ADD_UN16x4_MUL_UN16 (d, ad, s, as); + + *(dest + i) = d; + } +} + +static void +combine_atop_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t d = *(dest + i); + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + uint64_t ad; + uint32_t as = ~d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + ad = m; + + UN16x4_MUL_UN16x4_ADD_UN16x4_MUL_UN16 (d, ad, s, as); + + *(dest + i) = d; + } +} + +static void +combine_xor_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t d = *(dest + i); + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + uint64_t ad; + uint32_t as = ~d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + ad = ~m; + + UN16x4_MUL_UN16x4_ADD_UN16x4_MUL_UN16 (d, ad, s, as); + + *(dest + i) = d; + } +} + +static void +combine_add_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s = *(src + i); + uint64_t m = *(mask + i); + uint64_t d = *(dest + i); + + combine_mask_value_ca (&s, &m); + + UN16x4_ADD_UN16x4 (d, s); + + *(dest + i) = d; + } +} + +static void +combine_saturate_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s, d; + uint32_t sa, sr, sg, sb, da; + uint32_t t, u, v; + uint64_t m, n, o, p; + + d = *(dest + i); + s = *(src + i); + m = *(mask + i); + + combine_mask_ca (&s, &m); + + sa = (m >> A_SHIFT); + sr = (m >> R_SHIFT) & MASK; + sg = (m >> G_SHIFT) & MASK; + sb = m & MASK; + da = ~d >> A_SHIFT; + + if (sb <= da) + m = ADD (s, d, 0, t); + else + m = GENERIC (s, d, 0, (da << G_SHIFT) / sb, MASK, t, u, v); + + if (sg <= da) + n = ADD (s, d, G_SHIFT, t); + else + n = GENERIC (s, d, G_SHIFT, (da << G_SHIFT) / sg, MASK, t, u, v); + + if (sr <= da) + o = ADD (s, d, R_SHIFT, t); + else + o = GENERIC (s, d, R_SHIFT, (da << G_SHIFT) / sr, MASK, t, u, v); + + if (sa <= da) + p = ADD (s, d, A_SHIFT, t); + else + p = GENERIC (s, d, A_SHIFT, (da << G_SHIFT) / sa, MASK, t, u, v); + + *(dest + i) = m | n | o | p; + } +} + +static void +combine_disjoint_general_ca (uint64_t * dest, + const uint64_t *src, + const uint64_t *mask, + int width, + uint16_t combine) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s, d; + uint64_t m, n, o, p; + uint64_t Fa, Fb; + uint32_t t, u, v; + uint64_t sa; + uint16_t da; + + s = *(src + i); + m = *(mask + i); + d = *(dest + i); + da = d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + sa = m; + + switch (combine & COMBINE_A) + { + default: + Fa = 0; + break; + + case COMBINE_A_OUT: + m = (uint64_t)combine_disjoint_out_part ((uint16_t) (sa >> 0), da); + n = (uint64_t)combine_disjoint_out_part ((uint16_t) (sa >> G_SHIFT), da) << G_SHIFT; + o = (uint64_t)combine_disjoint_out_part ((uint16_t) (sa >> R_SHIFT), da) << R_SHIFT; + p = (uint64_t)combine_disjoint_out_part ((uint16_t) (sa >> A_SHIFT), da) << A_SHIFT; + Fa = m | n | o | p; + break; + + case COMBINE_A_IN: + m = (uint64_t)combine_disjoint_in_part ((uint16_t) (sa >> 0), da); + n = (uint64_t)combine_disjoint_in_part ((uint16_t) (sa >> G_SHIFT), da) << G_SHIFT; + o = (uint64_t)combine_disjoint_in_part ((uint16_t) (sa >> R_SHIFT), da) << R_SHIFT; + p = (uint64_t)combine_disjoint_in_part ((uint16_t) (sa >> A_SHIFT), da) << A_SHIFT; + Fa = m | n | o | p; + break; + + case COMBINE_A: + Fa = ~0; + break; + } + + switch (combine & COMBINE_B) + { + default: + Fb = 0; + break; + + case COMBINE_B_OUT: + m = (uint64_t)combine_disjoint_out_part (da, (uint16_t) (sa >> 0)); + n = (uint64_t)combine_disjoint_out_part (da, (uint16_t) (sa >> G_SHIFT)) << G_SHIFT; + o = (uint64_t)combine_disjoint_out_part (da, (uint16_t) (sa >> R_SHIFT)) << R_SHIFT; + p = (uint64_t)combine_disjoint_out_part (da, (uint16_t) (sa >> A_SHIFT)) << A_SHIFT; + Fb = m | n | o | p; + break; + + case COMBINE_B_IN: + m = (uint64_t)combine_disjoint_in_part (da, (uint16_t) (sa >> 0)); + n = (uint64_t)combine_disjoint_in_part (da, (uint16_t) (sa >> G_SHIFT)) << G_SHIFT; + o = (uint64_t)combine_disjoint_in_part (da, (uint16_t) (sa >> R_SHIFT)) << R_SHIFT; + p = (uint64_t)combine_disjoint_in_part (da, (uint16_t) (sa >> A_SHIFT)) << A_SHIFT; + Fb = m | n | o | p; + break; + + case COMBINE_B: + Fb = ~0; + break; + } + m = GENERIC (s, d, 0, GET_COMP (Fa, 0), GET_COMP (Fb, 0), t, u, v); + n = GENERIC (s, d, G_SHIFT, GET_COMP (Fa, G_SHIFT), GET_COMP (Fb, G_SHIFT), t, u, v); + o = GENERIC (s, d, R_SHIFT, GET_COMP (Fa, R_SHIFT), GET_COMP (Fb, R_SHIFT), t, u, v); + p = GENERIC (s, d, A_SHIFT, GET_COMP (Fa, A_SHIFT), GET_COMP (Fb, A_SHIFT), t, u, v); + + s = m | n | o | p; + + *(dest + i) = s; + } +} + +static void +combine_disjoint_over_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_OVER); +} + +static void +combine_disjoint_in_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_IN); +} + +static void +combine_disjoint_in_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_IN); +} + +static void +combine_disjoint_out_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_OUT); +} + +static void +combine_disjoint_out_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_OUT); +} + +static void +combine_disjoint_atop_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_A_ATOP); +} + +static void +combine_disjoint_atop_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_B_ATOP); +} + +static void +combine_disjoint_xor_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_disjoint_general_ca (dest, src, mask, width, COMBINE_XOR); +} + +static void +combine_conjoint_general_ca (uint64_t * dest, + const uint64_t *src, + const uint64_t *mask, + int width, + uint16_t combine) +{ + int i; + + for (i = 0; i < width; ++i) + { + uint64_t s, d; + uint64_t m, n, o, p; + uint64_t Fa, Fb; + uint32_t t, u, v; + uint64_t sa; + uint16_t da; + + s = *(src + i); + m = *(mask + i); + d = *(dest + i); + da = d >> A_SHIFT; + + combine_mask_ca (&s, &m); + + sa = m; + + switch (combine & COMBINE_A) + { + default: + Fa = 0; + break; + + case COMBINE_A_OUT: + m = (uint64_t)combine_conjoint_out_part ((uint16_t) (sa >> 0), da); + n = (uint64_t)combine_conjoint_out_part ((uint16_t) (sa >> G_SHIFT), da) << G_SHIFT; + o = (uint64_t)combine_conjoint_out_part ((uint16_t) (sa >> R_SHIFT), da) << R_SHIFT; + p = (uint64_t)combine_conjoint_out_part ((uint16_t) (sa >> A_SHIFT), da) << A_SHIFT; + Fa = m | n | o | p; + break; + + case COMBINE_A_IN: + m = (uint64_t)combine_conjoint_in_part ((uint16_t) (sa >> 0), da); + n = (uint64_t)combine_conjoint_in_part ((uint16_t) (sa >> G_SHIFT), da) << G_SHIFT; + o = (uint64_t)combine_conjoint_in_part ((uint16_t) (sa >> R_SHIFT), da) << R_SHIFT; + p = (uint64_t)combine_conjoint_in_part ((uint16_t) (sa >> A_SHIFT), da) << A_SHIFT; + Fa = m | n | o | p; + break; + + case COMBINE_A: + Fa = ~0; + break; + } + + switch (combine & COMBINE_B) + { + default: + Fb = 0; + break; + + case COMBINE_B_OUT: + m = (uint64_t)combine_conjoint_out_part (da, (uint16_t) (sa >> 0)); + n = (uint64_t)combine_conjoint_out_part (da, (uint16_t) (sa >> G_SHIFT)) << G_SHIFT; + o = (uint64_t)combine_conjoint_out_part (da, (uint16_t) (sa >> R_SHIFT)) << R_SHIFT; + p = (uint64_t)combine_conjoint_out_part (da, (uint16_t) (sa >> A_SHIFT)) << A_SHIFT; + Fb = m | n | o | p; + break; + + case COMBINE_B_IN: + m = (uint64_t)combine_conjoint_in_part (da, (uint16_t) (sa >> 0)); + n = (uint64_t)combine_conjoint_in_part (da, (uint16_t) (sa >> G_SHIFT)) << G_SHIFT; + o = (uint64_t)combine_conjoint_in_part (da, (uint16_t) (sa >> R_SHIFT)) << R_SHIFT; + p = (uint64_t)combine_conjoint_in_part (da, (uint16_t) (sa >> A_SHIFT)) << A_SHIFT; + Fb = m | n | o | p; + break; + + case COMBINE_B: + Fb = ~0; + break; + } + m = GENERIC (s, d, 0, GET_COMP (Fa, 0), GET_COMP (Fb, 0), t, u, v); + n = GENERIC (s, d, G_SHIFT, GET_COMP (Fa, G_SHIFT), GET_COMP (Fb, G_SHIFT), t, u, v); + o = GENERIC (s, d, R_SHIFT, GET_COMP (Fa, R_SHIFT), GET_COMP (Fb, R_SHIFT), t, u, v); + p = GENERIC (s, d, A_SHIFT, GET_COMP (Fa, A_SHIFT), GET_COMP (Fb, A_SHIFT), t, u, v); + + s = m | n | o | p; + + *(dest + i) = s; + } +} + +static void +combine_conjoint_over_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_OVER); +} + +static void +combine_conjoint_over_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_OVER); +} + +static void +combine_conjoint_in_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_IN); +} + +static void +combine_conjoint_in_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_IN); +} + +static void +combine_conjoint_out_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_OUT); +} + +static void +combine_conjoint_out_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_OUT); +} + +static void +combine_conjoint_atop_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_A_ATOP); +} + +static void +combine_conjoint_atop_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_B_ATOP); +} + +static void +combine_conjoint_xor_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + combine_conjoint_general_ca (dest, src, mask, width, COMBINE_XOR); +} + +void +_pixman_setup_combiner_functions_64 (pixman_implementation_t *imp) +{ + /* Unified alpha */ + imp->combine_64[PIXMAN_OP_CLEAR] = combine_clear; + imp->combine_64[PIXMAN_OP_SRC] = combine_src_u; + imp->combine_64[PIXMAN_OP_DST] = combine_dst; + imp->combine_64[PIXMAN_OP_OVER] = combine_over_u; + imp->combine_64[PIXMAN_OP_OVER_REVERSE] = combine_over_reverse_u; + imp->combine_64[PIXMAN_OP_IN] = combine_in_u; + imp->combine_64[PIXMAN_OP_IN_REVERSE] = combine_in_reverse_u; + imp->combine_64[PIXMAN_OP_OUT] = combine_out_u; + imp->combine_64[PIXMAN_OP_OUT_REVERSE] = combine_out_reverse_u; + imp->combine_64[PIXMAN_OP_ATOP] = combine_atop_u; + imp->combine_64[PIXMAN_OP_ATOP_REVERSE] = combine_atop_reverse_u; + imp->combine_64[PIXMAN_OP_XOR] = combine_xor_u; + imp->combine_64[PIXMAN_OP_ADD] = combine_add_u; + imp->combine_64[PIXMAN_OP_SATURATE] = combine_saturate_u; + + /* Disjoint, unified */ + imp->combine_64[PIXMAN_OP_DISJOINT_CLEAR] = combine_clear; + imp->combine_64[PIXMAN_OP_DISJOINT_SRC] = combine_src_u; + imp->combine_64[PIXMAN_OP_DISJOINT_DST] = combine_dst; + imp->combine_64[PIXMAN_OP_DISJOINT_OVER] = combine_disjoint_over_u; + imp->combine_64[PIXMAN_OP_DISJOINT_OVER_REVERSE] = combine_saturate_u; + imp->combine_64[PIXMAN_OP_DISJOINT_IN] = combine_disjoint_in_u; + imp->combine_64[PIXMAN_OP_DISJOINT_IN_REVERSE] = combine_disjoint_in_reverse_u; + imp->combine_64[PIXMAN_OP_DISJOINT_OUT] = combine_disjoint_out_u; + imp->combine_64[PIXMAN_OP_DISJOINT_OUT_REVERSE] = combine_disjoint_out_reverse_u; + imp->combine_64[PIXMAN_OP_DISJOINT_ATOP] = combine_disjoint_atop_u; + imp->combine_64[PIXMAN_OP_DISJOINT_ATOP_REVERSE] = combine_disjoint_atop_reverse_u; + imp->combine_64[PIXMAN_OP_DISJOINT_XOR] = combine_disjoint_xor_u; + + /* Conjoint, unified */ + imp->combine_64[PIXMAN_OP_CONJOINT_CLEAR] = combine_clear; + imp->combine_64[PIXMAN_OP_CONJOINT_SRC] = combine_src_u; + imp->combine_64[PIXMAN_OP_CONJOINT_DST] = combine_dst; + imp->combine_64[PIXMAN_OP_CONJOINT_OVER] = combine_conjoint_over_u; + imp->combine_64[PIXMAN_OP_CONJOINT_OVER_REVERSE] = combine_conjoint_over_reverse_u; + imp->combine_64[PIXMAN_OP_CONJOINT_IN] = combine_conjoint_in_u; + imp->combine_64[PIXMAN_OP_CONJOINT_IN_REVERSE] = combine_conjoint_in_reverse_u; + imp->combine_64[PIXMAN_OP_CONJOINT_OUT] = combine_conjoint_out_u; + imp->combine_64[PIXMAN_OP_CONJOINT_OUT_REVERSE] = combine_conjoint_out_reverse_u; + imp->combine_64[PIXMAN_OP_CONJOINT_ATOP] = combine_conjoint_atop_u; + imp->combine_64[PIXMAN_OP_CONJOINT_ATOP_REVERSE] = combine_conjoint_atop_reverse_u; + imp->combine_64[PIXMAN_OP_CONJOINT_XOR] = combine_conjoint_xor_u; + + imp->combine_64[PIXMAN_OP_MULTIPLY] = combine_multiply_u; + imp->combine_64[PIXMAN_OP_SCREEN] = combine_screen_u; + imp->combine_64[PIXMAN_OP_OVERLAY] = combine_overlay_u; + imp->combine_64[PIXMAN_OP_DARKEN] = combine_darken_u; + imp->combine_64[PIXMAN_OP_LIGHTEN] = combine_lighten_u; + imp->combine_64[PIXMAN_OP_COLOR_DODGE] = combine_color_dodge_u; + imp->combine_64[PIXMAN_OP_COLOR_BURN] = combine_color_burn_u; + imp->combine_64[PIXMAN_OP_HARD_LIGHT] = combine_hard_light_u; + imp->combine_64[PIXMAN_OP_SOFT_LIGHT] = combine_soft_light_u; + imp->combine_64[PIXMAN_OP_DIFFERENCE] = combine_difference_u; + imp->combine_64[PIXMAN_OP_EXCLUSION] = combine_exclusion_u; + imp->combine_64[PIXMAN_OP_HSL_HUE] = combine_hsl_hue_u; + imp->combine_64[PIXMAN_OP_HSL_SATURATION] = combine_hsl_saturation_u; + imp->combine_64[PIXMAN_OP_HSL_COLOR] = combine_hsl_color_u; + imp->combine_64[PIXMAN_OP_HSL_LUMINOSITY] = combine_hsl_luminosity_u; + + /* Component alpha combiners */ + imp->combine_64_ca[PIXMAN_OP_CLEAR] = combine_clear_ca; + imp->combine_64_ca[PIXMAN_OP_SRC] = combine_src_ca; + /* dest */ + imp->combine_64_ca[PIXMAN_OP_OVER] = combine_over_ca; + imp->combine_64_ca[PIXMAN_OP_OVER_REVERSE] = combine_over_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_IN] = combine_in_ca; + imp->combine_64_ca[PIXMAN_OP_IN_REVERSE] = combine_in_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_OUT] = combine_out_ca; + imp->combine_64_ca[PIXMAN_OP_OUT_REVERSE] = combine_out_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_ATOP] = combine_atop_ca; + imp->combine_64_ca[PIXMAN_OP_ATOP_REVERSE] = combine_atop_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_XOR] = combine_xor_ca; + imp->combine_64_ca[PIXMAN_OP_ADD] = combine_add_ca; + imp->combine_64_ca[PIXMAN_OP_SATURATE] = combine_saturate_ca; + + /* Disjoint CA */ + imp->combine_64_ca[PIXMAN_OP_DISJOINT_CLEAR] = combine_clear_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_SRC] = combine_src_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_DST] = combine_dst; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_OVER] = combine_disjoint_over_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_OVER_REVERSE] = combine_saturate_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_IN] = combine_disjoint_in_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_IN_REVERSE] = combine_disjoint_in_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_OUT] = combine_disjoint_out_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_OUT_REVERSE] = combine_disjoint_out_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_ATOP] = combine_disjoint_atop_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_ATOP_REVERSE] = combine_disjoint_atop_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_DISJOINT_XOR] = combine_disjoint_xor_ca; + + /* Conjoint CA */ + imp->combine_64_ca[PIXMAN_OP_CONJOINT_CLEAR] = combine_clear_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_SRC] = combine_src_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_DST] = combine_dst; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_OVER] = combine_conjoint_over_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_OVER_REVERSE] = combine_conjoint_over_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_IN] = combine_conjoint_in_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_IN_REVERSE] = combine_conjoint_in_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_OUT] = combine_conjoint_out_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_OUT_REVERSE] = combine_conjoint_out_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_ATOP] = combine_conjoint_atop_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_ATOP_REVERSE] = combine_conjoint_atop_reverse_ca; + imp->combine_64_ca[PIXMAN_OP_CONJOINT_XOR] = combine_conjoint_xor_ca; + + imp->combine_64_ca[PIXMAN_OP_MULTIPLY] = combine_multiply_ca; + imp->combine_64_ca[PIXMAN_OP_SCREEN] = combine_screen_ca; + imp->combine_64_ca[PIXMAN_OP_OVERLAY] = combine_overlay_ca; + imp->combine_64_ca[PIXMAN_OP_DARKEN] = combine_darken_ca; + imp->combine_64_ca[PIXMAN_OP_LIGHTEN] = combine_lighten_ca; + imp->combine_64_ca[PIXMAN_OP_COLOR_DODGE] = combine_color_dodge_ca; + imp->combine_64_ca[PIXMAN_OP_COLOR_BURN] = combine_color_burn_ca; + imp->combine_64_ca[PIXMAN_OP_HARD_LIGHT] = combine_hard_light_ca; + imp->combine_64_ca[PIXMAN_OP_SOFT_LIGHT] = combine_soft_light_ca; + imp->combine_64_ca[PIXMAN_OP_DIFFERENCE] = combine_difference_ca; + imp->combine_64_ca[PIXMAN_OP_EXCLUSION] = combine_exclusion_ca; + + /* It is not clear that these make sense, so make them noops for now */ + imp->combine_64_ca[PIXMAN_OP_HSL_HUE] = combine_dst; + imp->combine_64_ca[PIXMAN_OP_HSL_SATURATION] = combine_dst; + imp->combine_64_ca[PIXMAN_OP_HSL_COLOR] = combine_dst; + imp->combine_64_ca[PIXMAN_OP_HSL_LUMINOSITY] = combine_dst; +} + diff --git a/programs/develop/libraries/pixman/pixman-combine64.h b/programs/develop/libraries/pixman/pixman-combine64.h new file mode 100644 index 0000000000..ed9ffebf6c --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-combine64.h @@ -0,0 +1,230 @@ +/* WARNING: This file is generated by combine.pl from combine.inc. + Please edit one of those files rather than this one. */ + +#line 1 "pixman-combine.c.template" + +#define COMPONENT_SIZE 16 +#define MASK 0xffffULL +#define ONE_HALF 0x8000ULL + +#define A_SHIFT 16 * 3 +#define R_SHIFT 16 * 2 +#define G_SHIFT 16 +#define A_MASK 0xffff000000000000ULL +#define R_MASK 0xffff00000000ULL +#define G_MASK 0xffff0000ULL + +#define RB_MASK 0xffff0000ffffULL +#define AG_MASK 0xffff0000ffff0000ULL +#define RB_ONE_HALF 0x800000008000ULL +#define RB_MASK_PLUS_ONE 0x10000000010000ULL + +#define ALPHA_16(x) ((x) >> A_SHIFT) +#define RED_16(x) (((x) >> R_SHIFT) & MASK) +#define GREEN_16(x) (((x) >> G_SHIFT) & MASK) +#define BLUE_16(x) ((x) & MASK) + +/* + * Helper macros. + */ + +#define MUL_UN16(a, b, t) \ + ((t) = (a) * (b) + ONE_HALF, ((((t) >> G_SHIFT ) + (t) ) >> G_SHIFT )) + +#define DIV_UN16(a, b) \ + (((uint32_t) (a) * MASK) / (b)) + +#define ADD_UN16(x, y, t) \ + ((t) = (x) + (y), \ + (uint64_t) (uint16_t) ((t) | (0 - ((t) >> G_SHIFT)))) + +#define DIV_ONE_UN16(x) \ + (((x) + ONE_HALF + (((x) + ONE_HALF) >> G_SHIFT)) >> G_SHIFT) + +/* + * The methods below use some tricks to be able to do two color + * components at the same time. + */ + +/* + * x_rb = (x_rb * a) / 255 + */ +#define UN16_rb_MUL_UN16(x, a, t) \ + do \ + { \ + t = ((x) & RB_MASK) * (a); \ + t += RB_ONE_HALF; \ + x = (t + ((t >> G_SHIFT) & RB_MASK)) >> G_SHIFT; \ + x &= RB_MASK; \ + } while (0) + +/* + * x_rb = min (x_rb + y_rb, 255) + */ +#define UN16_rb_ADD_UN16_rb(x, y, t) \ + do \ + { \ + t = ((x) + (y)); \ + t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK); \ + x = (t & RB_MASK); \ + } while (0) + +/* + * x_rb = (x_rb * a_rb) / 255 + */ +#define UN16_rb_MUL_UN16_rb(x, a, t) \ + do \ + { \ + t = (x & MASK) * (a & MASK); \ + t |= (x & R_MASK) * ((a >> R_SHIFT) & MASK); \ + t += RB_ONE_HALF; \ + t = (t + ((t >> G_SHIFT) & RB_MASK)) >> G_SHIFT; \ + x = t & RB_MASK; \ + } while (0) + +/* + * x_c = (x_c * a) / 255 + */ +#define UN16x4_MUL_UN16(x, a) \ + do \ + { \ + uint64_t r1__, r2__, t__; \ + \ + r1__ = (x); \ + UN16_rb_MUL_UN16 (r1__, (a), t__); \ + \ + r2__ = (x) >> G_SHIFT; \ + UN16_rb_MUL_UN16 (r2__, (a), t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a) / 255 + y_c + */ +#define UN16x4_MUL_UN16_ADD_UN16x4(x, a, y) \ + do \ + { \ + uint64_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (y) & RB_MASK; \ + UN16_rb_MUL_UN16 (r1__, (a), t__); \ + UN16_rb_ADD_UN16_rb (r1__, r2__, t__); \ + \ + r2__ = (x) >> G_SHIFT; \ + r3__ = ((y) >> G_SHIFT) & RB_MASK; \ + UN16_rb_MUL_UN16 (r2__, (a), t__); \ + UN16_rb_ADD_UN16_rb (r2__, r3__, t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a + y_c * b) / 255 + */ +#define UN16x4_MUL_UN16_ADD_UN16x4_MUL_UN16(x, a, y, b) \ + do \ + { \ + uint64_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (y); \ + UN16_rb_MUL_UN16 (r1__, (a), t__); \ + UN16_rb_MUL_UN16 (r2__, (b), t__); \ + UN16_rb_ADD_UN16_rb (r1__, r2__, t__); \ + \ + r2__ = ((x) >> G_SHIFT); \ + r3__ = ((y) >> G_SHIFT); \ + UN16_rb_MUL_UN16 (r2__, (a), t__); \ + UN16_rb_MUL_UN16 (r3__, (b), t__); \ + UN16_rb_ADD_UN16_rb (r2__, r3__, t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a_c) / 255 + */ +#define UN16x4_MUL_UN16x4(x, a) \ + do \ + { \ + uint64_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (a); \ + UN16_rb_MUL_UN16_rb (r1__, r2__, t__); \ + \ + r2__ = (x) >> G_SHIFT; \ + r3__ = (a) >> G_SHIFT; \ + UN16_rb_MUL_UN16_rb (r2__, r3__, t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a_c) / 255 + y_c + */ +#define UN16x4_MUL_UN16x4_ADD_UN16x4(x, a, y) \ + do \ + { \ + uint64_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (a); \ + UN16_rb_MUL_UN16_rb (r1__, r2__, t__); \ + r2__ = (y) & RB_MASK; \ + UN16_rb_ADD_UN16_rb (r1__, r2__, t__); \ + \ + r2__ = ((x) >> G_SHIFT); \ + r3__ = ((a) >> G_SHIFT); \ + UN16_rb_MUL_UN16_rb (r2__, r3__, t__); \ + r3__ = ((y) >> G_SHIFT) & RB_MASK; \ + UN16_rb_ADD_UN16_rb (r2__, r3__, t__); \ + \ + (x) = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + * x_c = (x_c * a_c + y_c * b) / 255 + */ +#define UN16x4_MUL_UN16x4_ADD_UN16x4_MUL_UN16(x, a, y, b) \ + do \ + { \ + uint64_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x); \ + r2__ = (a); \ + UN16_rb_MUL_UN16_rb (r1__, r2__, t__); \ + r2__ = (y); \ + UN16_rb_MUL_UN16 (r2__, (b), t__); \ + UN16_rb_ADD_UN16_rb (r1__, r2__, t__); \ + \ + r2__ = (x) >> G_SHIFT; \ + r3__ = (a) >> G_SHIFT; \ + UN16_rb_MUL_UN16_rb (r2__, r3__, t__); \ + r3__ = (y) >> G_SHIFT; \ + UN16_rb_MUL_UN16 (r3__, (b), t__); \ + UN16_rb_ADD_UN16_rb (r2__, r3__, t__); \ + \ + x = r1__ | (r2__ << G_SHIFT); \ + } while (0) + +/* + x_c = min(x_c + y_c, 255) +*/ +#define UN16x4_ADD_UN16x4(x, y) \ + do \ + { \ + uint64_t r1__, r2__, r3__, t__; \ + \ + r1__ = (x) & RB_MASK; \ + r2__ = (y) & RB_MASK; \ + UN16_rb_ADD_UN16_rb (r1__, r2__, t__); \ + \ + r2__ = ((x) >> G_SHIFT) & RB_MASK; \ + r3__ = ((y) >> G_SHIFT) & RB_MASK; \ + UN16_rb_ADD_UN16_rb (r2__, r3__, t__); \ + \ + x = r1__ | (r2__ << G_SHIFT); \ + } while (0) diff --git a/programs/develop/libraries/pixman/pixman-compiler.h b/programs/develop/libraries/pixman/pixman-compiler.h new file mode 100644 index 0000000000..0e08589c83 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-compiler.h @@ -0,0 +1,204 @@ +/* Pixman uses some non-standard compiler features. This file ensures + * they exist + * + * The features are: + * + * FUNC must be defined to expand to the current function + * PIXMAN_EXPORT should be defined to whatever is required to + * export functions from a shared library + * limits limits for various types must be defined + * inline must be defined + * force_inline must be defined + */ +#if defined (__GNUC__) +# define FUNC ((const char*) (__PRETTY_FUNCTION__)) +#elif defined (__sun) || (defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) +# define FUNC ((const char*) (__func__)) +#else +# define FUNC ((const char*) ("???")) +#endif + +#ifndef INT16_MIN +# define INT16_MIN (-32767-1) +#endif + +#ifndef INT16_MAX +# define INT16_MAX (32767) +#endif + +#ifndef INT32_MIN +# define INT32_MIN (-2147483647-1) +#endif + +#ifndef INT32_MAX +# define INT32_MAX (2147483647) +#endif + +#ifndef UINT32_MIN +# define UINT32_MIN (0) +#endif + +#ifndef UINT32_MAX +# define UINT32_MAX (4294967295U) +#endif + +#ifndef M_PI +# define M_PI 3.14159265358979323846 +#endif + +#ifdef _MSC_VER +/* 'inline' is available only in C++ in MSVC */ +# define inline __inline +# define force_inline __forceinline +# define noinline __declspec(noinline) +#elif defined __GNUC__ || (defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590)) +# define inline __inline__ +# define force_inline __inline__ __attribute__ ((__always_inline__)) +# define noinline __attribute__((noinline)) +#else +# ifndef force_inline +# define force_inline inline +# endif +# ifndef noinline +# define noinline +# endif +#endif + +/* GCC visibility */ +#if defined(__GNUC__) && __GNUC__ >= 4 && !defined(_WIN32) +# define PIXMAN_EXPORT __attribute__ ((visibility("default"))) +/* Sun Studio 8 visibility */ +#elif defined(__SUNPRO_C) && (__SUNPRO_C >= 0x550) +# define PIXMAN_EXPORT __global +#else +# define PIXMAN_EXPORT +#endif + +/* TLS */ +#if defined(PIXMAN_NO_TLS) + +# define PIXMAN_DEFINE_THREAD_LOCAL(type, name) \ + static type name +# define PIXMAN_GET_THREAD_LOCAL(name) \ + (&name) + +#elif defined(TOOLCHAIN_SUPPORTS__THREAD) + +# define PIXMAN_DEFINE_THREAD_LOCAL(type, name) \ + static __thread type name +# define PIXMAN_GET_THREAD_LOCAL(name) \ + (&name) + +#elif defined(__MINGW32__) + +# define _NO_W32_PSEUDO_MODIFIERS +# include + +# define PIXMAN_DEFINE_THREAD_LOCAL(type, name) \ + static volatile int tls_ ## name ## _initialized = 0; \ + static void *tls_ ## name ## _mutex = NULL; \ + static unsigned tls_ ## name ## _index; \ + \ + static type * \ + tls_ ## name ## _alloc (void) \ + { \ + type *value = calloc (1, sizeof (type)); \ + if (value) \ + TlsSetValue (tls_ ## name ## _index, value); \ + return value; \ + } \ + \ + static force_inline type * \ + tls_ ## name ## _get (void) \ + { \ + type *value; \ + if (!tls_ ## name ## _initialized) \ + { \ + if (!tls_ ## name ## _mutex) \ + { \ + void *mutex = CreateMutexA (NULL, 0, NULL); \ + if (InterlockedCompareExchangePointer ( \ + &tls_ ## name ## _mutex, mutex, NULL) != NULL) \ + { \ + CloseHandle (mutex); \ + } \ + } \ + WaitForSingleObject (tls_ ## name ## _mutex, 0xFFFFFFFF); \ + if (!tls_ ## name ## _initialized) \ + { \ + tls_ ## name ## _index = TlsAlloc (); \ + tls_ ## name ## _initialized = 1; \ + } \ + ReleaseMutex (tls_ ## name ## _mutex); \ + } \ + if (tls_ ## name ## _index == 0xFFFFFFFF) \ + return NULL; \ + value = TlsGetValue (tls_ ## name ## _index); \ + if (!value) \ + value = tls_ ## name ## _alloc (); \ + return value; \ + } + +# define PIXMAN_GET_THREAD_LOCAL(name) \ + tls_ ## name ## _get () + +#elif defined(_MSC_VER) + +# define PIXMAN_DEFINE_THREAD_LOCAL(type, name) \ + static __declspec(thread) type name +# define PIXMAN_GET_THREAD_LOCAL(name) \ + (&name) + +#elif defined(HAVE_PTHREAD_SETSPECIFIC) + +#include + +# define PIXMAN_DEFINE_THREAD_LOCAL(type, name) \ + static pthread_once_t tls_ ## name ## _once_control = PTHREAD_ONCE_INIT; \ + static pthread_key_t tls_ ## name ## _key; \ + \ + static void \ + tls_ ## name ## _destroy_value (void *value) \ + { \ + free (value); \ + } \ + \ + static void \ + tls_ ## name ## _make_key (void) \ + { \ + pthread_key_create (&tls_ ## name ## _key, \ + tls_ ## name ## _destroy_value); \ + } \ + \ + static type * \ + tls_ ## name ## _alloc (void) \ + { \ + type *value = calloc (1, sizeof (type)); \ + if (value) \ + pthread_setspecific (tls_ ## name ## _key, value); \ + return value; \ + } \ + \ + static force_inline type * \ + tls_ ## name ## _get (void) \ + { \ + type *value = NULL; \ + if (pthread_once (&tls_ ## name ## _once_control, \ + tls_ ## name ## _make_key) == 0) \ + { \ + value = pthread_getspecific (tls_ ## name ## _key); \ + if (!value) \ + value = tls_ ## name ## _alloc (); \ + } \ + return value; \ + } \ + extern int no_such_variable + +# define PIXMAN_GET_THREAD_LOCAL(name) \ + tls_ ## name ## _get () + +#else + +# error "Unknown thread local support for this system. Pixman will not work with multiple threads. Define PIXMAN_NO_TLS to acknowledge and accept this limitation and compile pixman without thread-safety support." + +#endif diff --git a/programs/develop/libraries/pixman/pixman-conical-gradient.c b/programs/develop/libraries/pixman/pixman-conical-gradient.c new file mode 100644 index 0000000000..897948be4f --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-conical-gradient.c @@ -0,0 +1,198 @@ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007 Red Hat, Inc. + * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc. + * 2005 Lars Knoll & Zack Rusin, Trolltech + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif + +#include +#include +#include "pixman-private.h" + +static force_inline double +coordinates_to_parameter (double x, double y, double angle) +{ + double t; + + t = atan2 (y, x) + angle; + + while (t < 0) + t += 2 * M_PI; + + while (t >= 2 * M_PI) + t -= 2 * M_PI; + + return 1 - t * (1 / (2 * M_PI)); /* Scale t to [0, 1] and + * make rotation CCW + */ +} + +static void +conical_gradient_get_scanline_32 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + source_image_t *source = (source_image_t *)image; + gradient_t *gradient = (gradient_t *)source; + conical_gradient_t *conical = (conical_gradient_t *)image; + uint32_t *end = buffer + width; + pixman_gradient_walker_t walker; + pixman_bool_t affine = TRUE; + double cx = 1.; + double cy = 0.; + double cz = 0.; + double rx = x + 0.5; + double ry = y + 0.5; + double rz = 1.; + + _pixman_gradient_walker_init (&walker, gradient, source->common.repeat); + + if (source->common.transform) + { + pixman_vector_t v; + + /* reference point is the center of the pixel */ + v.vector[0] = pixman_int_to_fixed (x) + pixman_fixed_1 / 2; + v.vector[1] = pixman_int_to_fixed (y) + pixman_fixed_1 / 2; + v.vector[2] = pixman_fixed_1; + + if (!pixman_transform_point_3d (source->common.transform, &v)) + return; + + cx = source->common.transform->matrix[0][0] / 65536.; + cy = source->common.transform->matrix[1][0] / 65536.; + cz = source->common.transform->matrix[2][0] / 65536.; + + rx = v.vector[0] / 65536.; + ry = v.vector[1] / 65536.; + rz = v.vector[2] / 65536.; + + affine = + source->common.transform->matrix[2][0] == 0 && + v.vector[2] == pixman_fixed_1; + } + + if (affine) + { + rx -= conical->center.x / 65536.; + ry -= conical->center.y / 65536.; + + while (buffer < end) + { + if (!mask || *mask++) + { + double t = coordinates_to_parameter (rx, ry, conical->angle); + + *buffer = _pixman_gradient_walker_pixel ( + &walker, (pixman_fixed_48_16_t)pixman_double_to_fixed (t)); + } + + ++buffer; + + rx += cx; + ry += cy; + } + } + else + { + while (buffer < end) + { + double x, y; + + if (!mask || *mask++) + { + double t; + + if (rz != 0) + { + x = rx / rz; + y = ry / rz; + } + else + { + x = y = 0.; + } + + x -= conical->center.x / 65536.; + y -= conical->center.y / 65536.; + + t = coordinates_to_parameter (x, y, conical->angle); + + *buffer = _pixman_gradient_walker_pixel ( + &walker, (pixman_fixed_48_16_t)pixman_double_to_fixed (t)); + } + + ++buffer; + + rx += cx; + ry += cy; + rz += cz; + } + } +} + +static void +conical_gradient_property_changed (pixman_image_t *image) +{ + image->common.get_scanline_32 = conical_gradient_get_scanline_32; + image->common.get_scanline_64 = _pixman_image_get_scanline_generic_64; +} + +PIXMAN_EXPORT pixman_image_t * +pixman_image_create_conical_gradient (pixman_point_fixed_t * center, + pixman_fixed_t angle, + const pixman_gradient_stop_t *stops, + int n_stops) +{ + pixman_image_t *image = _pixman_image_allocate (); + conical_gradient_t *conical; + + if (!image) + return NULL; + + conical = &image->conical; + + if (!_pixman_init_gradient (&conical->common, stops, n_stops)) + { + free (image); + return NULL; + } + + angle = MOD (angle, pixman_int_to_fixed (360)); + + image->type = CONICAL; + + conical->center = *center; + conical->angle = (pixman_fixed_to_double (angle) / 180.0) * M_PI; + + image->common.property_changed = conical_gradient_property_changed; + + return image; +} + diff --git a/programs/develop/libraries/pixman/pixman-cpu.c b/programs/develop/libraries/pixman/pixman-cpu.c new file mode 100644 index 0000000000..e4fb1e4097 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-cpu.c @@ -0,0 +1,598 @@ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of SuSE not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. SuSE makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * SuSE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL SuSE + * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + */ +#ifdef HAVE_CONFIG_H +#include +#endif + +#include + +#if defined(USE_ARM_SIMD) && defined(_MSC_VER) +/* Needed for EXCEPTION_ILLEGAL_INSTRUCTION */ +#include +#endif + +#include "pixman-private.h" + +#ifdef USE_VMX + +/* The CPU detection code needs to be in a file not compiled with + * "-maltivec -mabi=altivec", as gcc would try to save vector register + * across function calls causing SIGILL on cpus without Altivec/vmx. + */ +static pixman_bool_t initialized = FALSE; +static volatile pixman_bool_t have_vmx = TRUE; + +#ifdef __APPLE__ +#include + +static pixman_bool_t +pixman_have_vmx (void) +{ + if (!initialized) + { + size_t length = sizeof(have_vmx); + int error = + sysctlbyname ("hw.optional.altivec", &have_vmx, &length, NULL, 0); + + if (error) + have_vmx = FALSE; + + initialized = TRUE; + } + return have_vmx; +} + +#elif defined (__OpenBSD__) +#include +#include +#include + +static pixman_bool_t +pixman_have_vmx (void) +{ + if (!initialized) + { + int mib[2] = { CTL_MACHDEP, CPU_ALTIVEC }; + size_t length = sizeof(have_vmx); + int error = + sysctl (mib, 2, &have_vmx, &length, NULL, 0); + + if (error != 0) + have_vmx = FALSE; + + initialized = TRUE; + } + return have_vmx; +} + +#elif defined (__linux__) +#include +#include +#include +#include +#include +#include +#include + +static pixman_bool_t +pixman_have_vmx (void) +{ + if (!initialized) + { + char fname[64]; + unsigned long buf[64]; + ssize_t count = 0; + pid_t pid; + int fd, i; + + pid = getpid (); + snprintf (fname, sizeof(fname) - 1, "/proc/%d/auxv", pid); + + fd = open (fname, O_RDONLY); + if (fd >= 0) + { + for (i = 0; i <= (count / sizeof(unsigned long)); i += 2) + { + /* Read more if buf is empty... */ + if (i == (count / sizeof(unsigned long))) + { + count = read (fd, buf, sizeof(buf)); + if (count <= 0) + break; + i = 0; + } + + if (buf[i] == AT_HWCAP) + { + have_vmx = !!(buf[i + 1] & PPC_FEATURE_HAS_ALTIVEC); + initialized = TRUE; + break; + } + else if (buf[i] == AT_NULL) + { + break; + } + } + close (fd); + } + } + if (!initialized) + { + /* Something went wrong. Assume 'no' rather than playing + fragile tricks with catching SIGILL. */ + have_vmx = FALSE; + initialized = TRUE; + } + + return have_vmx; +} + +#else /* !__APPLE__ && !__OpenBSD__ && !__linux__ */ +#include +#include + +static jmp_buf jump_env; + +static void +vmx_test (int sig, + siginfo_t *si, + void * unused) +{ + longjmp (jump_env, 1); +} + +static pixman_bool_t +pixman_have_vmx (void) +{ + struct sigaction sa, osa; + int jmp_result; + + if (!initialized) + { + sa.sa_flags = SA_SIGINFO; + sigemptyset (&sa.sa_mask); + sa.sa_sigaction = vmx_test; + sigaction (SIGILL, &sa, &osa); + jmp_result = setjmp (jump_env); + if (jmp_result == 0) + { + asm volatile ( "vor 0, 0, 0" ); + } + sigaction (SIGILL, &osa, NULL); + have_vmx = (jmp_result == 0); + initialized = TRUE; + } + return have_vmx; +} + +#endif /* __APPLE__ */ +#endif /* USE_VMX */ + +#if defined(USE_ARM_SIMD) || defined(USE_ARM_NEON) + +#if defined(_MSC_VER) + +#if defined(USE_ARM_SIMD) +extern int pixman_msvc_try_arm_simd_op (); + +pixman_bool_t +pixman_have_arm_simd (void) +{ + static pixman_bool_t initialized = FALSE; + static pixman_bool_t have_arm_simd = FALSE; + + if (!initialized) + { + __try { + pixman_msvc_try_arm_simd_op (); + have_arm_simd = TRUE; + } __except (GetExceptionCode () == EXCEPTION_ILLEGAL_INSTRUCTION) { + have_arm_simd = FALSE; + } + initialized = TRUE; + } + + return have_arm_simd; +} + +#endif /* USE_ARM_SIMD */ + +#if defined(USE_ARM_NEON) +extern int pixman_msvc_try_arm_neon_op (); + +pixman_bool_t +pixman_have_arm_neon (void) +{ + static pixman_bool_t initialized = FALSE; + static pixman_bool_t have_arm_neon = FALSE; + + if (!initialized) + { + __try + { + pixman_msvc_try_arm_neon_op (); + have_arm_neon = TRUE; + } + __except (GetExceptionCode () == EXCEPTION_ILLEGAL_INSTRUCTION) + { + have_arm_neon = FALSE; + } + initialized = TRUE; + } + + return have_arm_neon; +} + +#endif /* USE_ARM_NEON */ + +#else /* linux ELF */ + +#include +#include +#include +#include +#include +#include +#include +#include + +static pixman_bool_t arm_has_v7 = FALSE; +static pixman_bool_t arm_has_v6 = FALSE; +static pixman_bool_t arm_has_vfp = FALSE; +static pixman_bool_t arm_has_neon = FALSE; +static pixman_bool_t arm_has_iwmmxt = FALSE; +static pixman_bool_t arm_tests_initialized = FALSE; + +static void +pixman_arm_read_auxv () +{ + int fd; + Elf32_auxv_t aux; + + fd = open ("/proc/self/auxv", O_RDONLY); + if (fd >= 0) + { + while (read (fd, &aux, sizeof(Elf32_auxv_t)) == sizeof(Elf32_auxv_t)) + { + if (aux.a_type == AT_HWCAP) + { + uint32_t hwcap = aux.a_un.a_val; + /* hardcode these values to avoid depending on specific + * versions of the hwcap header, e.g. HWCAP_NEON + */ + arm_has_vfp = (hwcap & 64) != 0; + arm_has_iwmmxt = (hwcap & 512) != 0; + /* this flag is only present on kernel 2.6.29 */ + arm_has_neon = (hwcap & 4096) != 0; + } + else if (aux.a_type == AT_PLATFORM) + { + const char *plat = (const char*) aux.a_un.a_val; + if (strncmp (plat, "v7l", 3) == 0) + { + arm_has_v7 = TRUE; + arm_has_v6 = TRUE; + } + else if (strncmp (plat, "v6l", 3) == 0) + { + arm_has_v6 = TRUE; + } + } + } + close (fd); + } + + arm_tests_initialized = TRUE; +} + +#if defined(USE_ARM_SIMD) +pixman_bool_t +pixman_have_arm_simd (void) +{ + if (!arm_tests_initialized) + pixman_arm_read_auxv (); + + return arm_has_v6; +} + +#endif /* USE_ARM_SIMD */ + +#if defined(USE_ARM_NEON) +pixman_bool_t +pixman_have_arm_neon (void) +{ + if (!arm_tests_initialized) + pixman_arm_read_auxv (); + + return arm_has_neon; +} + +#endif /* USE_ARM_NEON */ + +#endif /* linux */ + +#endif /* USE_ARM_SIMD || USE_ARM_NEON */ + +#if defined(USE_MMX) || defined(USE_SSE2) +/* The CPU detection code needs to be in a file not compiled with + * "-mmmx -msse", as gcc would generate CMOV instructions otherwise + * that would lead to SIGILL instructions on old CPUs that don't have + * it. + */ +#if !defined(__amd64__) && !defined(__x86_64__) && !defined(_M_AMD64) + +#ifdef HAVE_GETISAX +#include +#endif + +typedef enum +{ + NO_FEATURES = 0, + MMX = 0x1, + MMX_EXTENSIONS = 0x2, + SSE = 0x6, + SSE2 = 0x8, + CMOV = 0x10 +} cpu_features_t; + + +static unsigned int +detect_cpu_features (void) +{ + unsigned int features = 0; + unsigned int result = 0; + +#ifdef HAVE_GETISAX + if (getisax (&result, 1)) + { + if (result & AV_386_CMOV) + features |= CMOV; + if (result & AV_386_MMX) + features |= MMX; + if (result & AV_386_AMD_MMX) + features |= MMX_EXTENSIONS; + if (result & AV_386_SSE) + features |= SSE; + if (result & AV_386_SSE2) + features |= SSE2; + } +#else + char vendor[13]; +#ifdef _MSC_VER + int vendor0 = 0, vendor1, vendor2; +#endif + vendor[0] = 0; + vendor[12] = 0; + +#ifdef __GNUC__ + /* see p. 118 of amd64 instruction set manual Vol3 */ + /* We need to be careful about the handling of %ebx and + * %esp here. We can't declare either one as clobbered + * since they are special registers (%ebx is the "PIC + * register" holding an offset to global data, %esp the + * stack pointer), so we need to make sure they have their + * original values when we access the output operands. + */ + __asm__ ( + "pushf\n" + "pop %%eax\n" + "mov %%eax, %%ecx\n" + "xor $0x00200000, %%eax\n" + "push %%eax\n" + "popf\n" + "pushf\n" + "pop %%eax\n" + "mov $0x0, %%edx\n" + "xor %%ecx, %%eax\n" + "jz 1f\n" + + "mov $0x00000000, %%eax\n" + "push %%ebx\n" + "cpuid\n" + "mov %%ebx, %%eax\n" + "pop %%ebx\n" + "mov %%eax, %1\n" + "mov %%edx, %2\n" + "mov %%ecx, %3\n" + "mov $0x00000001, %%eax\n" + "push %%ebx\n" + "cpuid\n" + "pop %%ebx\n" + "1:\n" + "mov %%edx, %0\n" + : "=r" (result), + "=m" (vendor[0]), + "=m" (vendor[4]), + "=m" (vendor[8]) + : + : "%eax", "%ecx", "%edx" + ); + +#elif defined (_MSC_VER) + + _asm { + pushfd + pop eax + mov ecx, eax + xor eax, 00200000h + push eax + popfd + pushfd + pop eax + mov edx, 0 + xor eax, ecx + jz nocpuid + + mov eax, 0 + push ebx + cpuid + mov eax, ebx + pop ebx + mov vendor0, eax + mov vendor1, edx + mov vendor2, ecx + mov eax, 1 + push ebx + cpuid + pop ebx + nocpuid: + mov result, edx + } + memmove (vendor + 0, &vendor0, 4); + memmove (vendor + 4, &vendor1, 4); + memmove (vendor + 8, &vendor2, 4); + +#else +# error unsupported compiler +#endif + + features = 0; + if (result) + { + /* result now contains the standard feature bits */ + if (result & (1 << 15)) + features |= CMOV; + if (result & (1 << 23)) + features |= MMX; + if (result & (1 << 25)) + features |= SSE; + if (result & (1 << 26)) + features |= SSE2; + if ((features & MMX) && !(features & SSE) && + (strcmp (vendor, "AuthenticAMD") == 0 || + strcmp (vendor, "Geode by NSC") == 0)) + { + /* check for AMD MMX extensions */ +#ifdef __GNUC__ + __asm__ ( + " push %%ebx\n" + " mov $0x80000000, %%eax\n" + " cpuid\n" + " xor %%edx, %%edx\n" + " cmp $0x1, %%eax\n" + " jge 2f\n" + " mov $0x80000001, %%eax\n" + " cpuid\n" + "2:\n" + " pop %%ebx\n" + " mov %%edx, %0\n" + : "=r" (result) + : + : "%eax", "%ecx", "%edx" + ); +#elif defined _MSC_VER + _asm { + push ebx + mov eax, 80000000h + cpuid + xor edx, edx + cmp eax, 1 + jge notamd + mov eax, 80000001h + cpuid + notamd: + pop ebx + mov result, edx + } +#endif + if (result & (1 << 22)) + features |= MMX_EXTENSIONS; + } + } +#endif /* HAVE_GETISAX */ + + return features; +} + +static pixman_bool_t +pixman_have_mmx (void) +{ + static pixman_bool_t initialized = FALSE; + static pixman_bool_t mmx_present; + + if (!initialized) + { + unsigned int features = detect_cpu_features (); + mmx_present = (features & (MMX | MMX_EXTENSIONS)) == (MMX | MMX_EXTENSIONS); + initialized = TRUE; + } + + return mmx_present; +} + +#ifdef USE_SSE2 +static pixman_bool_t +pixman_have_sse2 (void) +{ + static pixman_bool_t initialized = FALSE; + static pixman_bool_t sse2_present; + + if (!initialized) + { + unsigned int features = detect_cpu_features (); + sse2_present = (features & (MMX | MMX_EXTENSIONS | SSE | SSE2)) == (MMX | MMX_EXTENSIONS | SSE | SSE2); + initialized = TRUE; + } + + return sse2_present; +} + +#endif + +#else /* __amd64__ */ +#ifdef USE_MMX +#define pixman_have_mmx() TRUE +#endif +#ifdef USE_SSE2 +#define pixman_have_sse2() TRUE +#endif +#endif /* __amd64__ */ +#endif + +pixman_implementation_t * +_pixman_choose_implementation (void) +{ +#ifdef USE_SSE2 + if (pixman_have_sse2 ()) + return _pixman_implementation_create_sse2 (); +#endif +#ifdef USE_MMX + if (pixman_have_mmx ()) + return _pixman_implementation_create_mmx (); +#endif + +#ifdef USE_ARM_NEON + if (pixman_have_arm_neon ()) + return _pixman_implementation_create_arm_neon (); +#endif +#ifdef USE_ARM_SIMD + if (pixman_have_arm_simd ()) + return _pixman_implementation_create_arm_simd (); +#endif +#ifdef USE_VMX + if (pixman_have_vmx ()) + return _pixman_implementation_create_vmx (); +#endif + + return _pixman_implementation_create_fast_path (); +} + diff --git a/programs/develop/libraries/pixman/pixman-edge-accessors.c b/programs/develop/libraries/pixman/pixman-edge-accessors.c new file mode 100644 index 0000000000..ea3a31e2f5 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-edge-accessors.c @@ -0,0 +1,4 @@ + +#define PIXMAN_FB_ACCESSORS + +#include "pixman-edge.c" diff --git a/programs/develop/libraries/pixman/pixman-edge-imp.h b/programs/develop/libraries/pixman/pixman-edge-imp.h new file mode 100644 index 0000000000..a4698eddb2 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-edge-imp.h @@ -0,0 +1,182 @@ +/* + * Copyright © 2004 Keith Packard + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * KEITH PACKARD DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL KEITH PACKARD BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR + * PERFORMANCE OF THIS SOFTWARE. + */ + +#ifndef rasterize_span +#endif + +static void +RASTERIZE_EDGES (pixman_image_t *image, + pixman_edge_t *l, + pixman_edge_t *r, + pixman_fixed_t t, + pixman_fixed_t b) +{ + pixman_fixed_t y = t; + uint32_t *line; + uint32_t *buf = (image)->bits.bits; + int stride = (image)->bits.rowstride; + int width = (image)->bits.width; + + line = buf + pixman_fixed_to_int (y) * stride; + + for (;;) + { + pixman_fixed_t lx; + pixman_fixed_t rx; + int lxi; + int rxi; + + lx = l->x; + rx = r->x; +#if N_BITS == 1 + /* For the non-antialiased case, round the coordinates up, in effect + * sampling just slightly to the left of the pixel. This is so that + * when the sample point lies exactly on the line, we round towards + * north-west. + * + * (The AA case does a similar adjustment in RENDER_SAMPLES_X) + */ + lx += X_FRAC_FIRST(1) - pixman_fixed_e; + rx += X_FRAC_FIRST(1) - pixman_fixed_e; +#endif + /* clip X */ + if (lx < 0) + lx = 0; + if (pixman_fixed_to_int (rx) >= width) +#if N_BITS == 1 + rx = pixman_int_to_fixed (width); +#else + /* Use the last pixel of the scanline, covered 100%. + * We can't use the first pixel following the scanline, + * because accessing it could result in a buffer overrun. + */ + rx = pixman_int_to_fixed (width) - 1; +#endif + + /* Skip empty (or backwards) sections */ + if (rx > lx) + { + + /* Find pixel bounds for span */ + lxi = pixman_fixed_to_int (lx); + rxi = pixman_fixed_to_int (rx); + +#if N_BITS == 1 + { + +#define LEFT_MASK(x) \ + (((x) & 0x1f) ? \ + SCREEN_SHIFT_RIGHT (0xffffffff, (x) & 0x1f) : 0) +#define RIGHT_MASK(x) \ + (((32 - (x)) & 0x1f) ? \ + SCREEN_SHIFT_LEFT (0xffffffff, (32 - (x)) & 0x1f) : 0) + +#define MASK_BITS(x,w,l,n,r) { \ + n = (w); \ + r = RIGHT_MASK ((x) + n); \ + l = LEFT_MASK (x); \ + if (l) { \ + n -= 32 - ((x) & 0x1f); \ + if (n < 0) { \ + n = 0; \ + l &= r; \ + r = 0; \ + } \ + } \ + n >>= 5; \ + } + + uint32_t *a = line; + uint32_t startmask; + uint32_t endmask; + int nmiddle; + int width = rxi - lxi; + int x = lxi; + + a += x >> 5; + x &= 0x1f; + + MASK_BITS (x, width, startmask, nmiddle, endmask); + + if (startmask) { + WRITE(image, a, READ(image, a) | startmask); + a++; + } + while (nmiddle--) + WRITE(image, a++, 0xffffffff); + if (endmask) + WRITE(image, a, READ(image, a) | endmask); + } +#else + { + DEFINE_ALPHA(line,lxi); + int lxs; + int rxs; + + /* Sample coverage for edge pixels */ + lxs = RENDER_SAMPLES_X (lx, N_BITS); + rxs = RENDER_SAMPLES_X (rx, N_BITS); + + /* Add coverage across row */ + if (lxi == rxi) + { + ADD_ALPHA (rxs - lxs); + } + else + { + int xi; + + ADD_ALPHA (N_X_FRAC(N_BITS) - lxs); + STEP_ALPHA; + for (xi = lxi + 1; xi < rxi; xi++) + { + ADD_ALPHA (N_X_FRAC(N_BITS)); + STEP_ALPHA; + } + ADD_ALPHA (rxs); + } + } +#endif + } + + if (y == b) + break; + +#if N_BITS > 1 + if (pixman_fixed_frac (y) != Y_FRAC_LAST(N_BITS)) + { + RENDER_EDGE_STEP_SMALL (l); + RENDER_EDGE_STEP_SMALL (r); + y += STEP_Y_SMALL(N_BITS); + } + else +#endif + { + RENDER_EDGE_STEP_BIG (l); + RENDER_EDGE_STEP_BIG (r); + y += STEP_Y_BIG(N_BITS); + line += stride; + } + } +} + +#undef rasterize_span diff --git a/programs/develop/libraries/pixman/pixman-edge.c b/programs/develop/libraries/pixman/pixman-edge.c new file mode 100644 index 0000000000..8d498ab445 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-edge.c @@ -0,0 +1,384 @@ +/* + * Copyright © 2004 Keith Packard + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * KEITH PACKARD DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL KEITH PACKARD BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR + * PERFORMANCE OF THIS SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif + +#include + +#include "pixman-private.h" +#include "pixman-accessor.h" + +/* + * Step across a small sample grid gap + */ +#define RENDER_EDGE_STEP_SMALL(edge) \ + { \ + edge->x += edge->stepx_small; \ + edge->e += edge->dx_small; \ + if (edge->e > 0) \ + { \ + edge->e -= edge->dy; \ + edge->x += edge->signdx; \ + } \ + } + +/* + * Step across a large sample grid gap + */ +#define RENDER_EDGE_STEP_BIG(edge) \ + { \ + edge->x += edge->stepx_big; \ + edge->e += edge->dx_big; \ + if (edge->e > 0) \ + { \ + edge->e -= edge->dy; \ + edge->x += edge->signdx; \ + } \ + } + +#ifdef PIXMAN_FB_ACCESSORS +#define PIXMAN_RASTERIZE_EDGES pixman_rasterize_edges_accessors +#else +#define PIXMAN_RASTERIZE_EDGES pixman_rasterize_edges_no_accessors +#endif + +/* + * 4 bit alpha + */ + +#define N_BITS 4 +#define RASTERIZE_EDGES rasterize_edges_4 + +#ifndef WORDS_BIGENDIAN +#define SHIFT_4(o) ((o) << 2) +#else +#define SHIFT_4(o) ((1 - (o)) << 2) +#endif + +#define GET_4(x, o) (((x) >> SHIFT_4 (o)) & 0xf) +#define PUT_4(x, o, v) \ + (((x) & ~(0xf << SHIFT_4 (o))) | (((v) & 0xf) << SHIFT_4 (o))) + +#define DEFINE_ALPHA(line, x) \ + uint8_t *__ap = (uint8_t *) line + ((x) >> 1); \ + int __ao = (x) & 1 + +#define STEP_ALPHA ((__ap += __ao), (__ao ^= 1)) + +#define ADD_ALPHA(a) \ + { \ + uint8_t __o = READ (image, __ap); \ + uint8_t __a = (a) + GET_4 (__o, __ao); \ + WRITE (image, __ap, PUT_4 (__o, __ao, __a | (0 - ((__a) >> 4)))); \ + } + +#include "pixman-edge-imp.h" + +#undef ADD_ALPHA +#undef STEP_ALPHA +#undef DEFINE_ALPHA +#undef RASTERIZE_EDGES +#undef N_BITS + + +/* + * 1 bit alpha + */ + +#define N_BITS 1 +#define RASTERIZE_EDGES rasterize_edges_1 + +#include "pixman-edge-imp.h" + +#undef RASTERIZE_EDGES +#undef N_BITS + +/* + * 8 bit alpha + */ + +static force_inline uint8_t +clip255 (int x) +{ + if (x > 255) + return 255; + + return x; +} + +#define ADD_SATURATE_8(buf, val, length) \ + do \ + { \ + int i__ = (length); \ + uint8_t *buf__ = (buf); \ + int val__ = (val); \ + \ + while (i__--) \ + { \ + WRITE (image, (buf__), clip255 (READ (image, (buf__)) + (val__))); \ + (buf__)++; \ + } \ + } while (0) + +/* + * We want to detect the case where we add the same value to a long + * span of pixels. The triangles on the end are filled in while we + * count how many sub-pixel scanlines contribute to the middle section. + * + * +--------------------------+ + * fill_height =| \ / + * +------------------+ + * |================| + * fill_start fill_end + */ +static void +rasterize_edges_8 (pixman_image_t *image, + pixman_edge_t * l, + pixman_edge_t * r, + pixman_fixed_t t, + pixman_fixed_t b) +{ + pixman_fixed_t y = t; + uint32_t *line; + int fill_start = -1, fill_end = -1; + int fill_size = 0; + uint32_t *buf = (image)->bits.bits; + int stride = (image)->bits.rowstride; + int width = (image)->bits.width; + + line = buf + pixman_fixed_to_int (y) * stride; + + for (;;) + { + uint8_t *ap = (uint8_t *) line; + pixman_fixed_t lx, rx; + int lxi, rxi; + + /* clip X */ + lx = l->x; + if (lx < 0) + lx = 0; + + rx = r->x; + + if (pixman_fixed_to_int (rx) >= width) + { + /* Use the last pixel of the scanline, covered 100%. + * We can't use the first pixel following the scanline, + * because accessing it could result in a buffer overrun. + */ + rx = pixman_int_to_fixed (width) - 1; + } + + /* Skip empty (or backwards) sections */ + if (rx > lx) + { + int lxs, rxs; + + /* Find pixel bounds for span. */ + lxi = pixman_fixed_to_int (lx); + rxi = pixman_fixed_to_int (rx); + + /* Sample coverage for edge pixels */ + lxs = RENDER_SAMPLES_X (lx, 8); + rxs = RENDER_SAMPLES_X (rx, 8); + + /* Add coverage across row */ + if (lxi == rxi) + { + WRITE (image, ap + lxi, + clip255 (READ (image, ap + lxi) + rxs - lxs)); + } + else + { + WRITE (image, ap + lxi, + clip255 (READ (image, ap + lxi) + N_X_FRAC (8) - lxs)); + + /* Move forward so that lxi/rxi is the pixel span */ + lxi++; + + /* Don't bother trying to optimize the fill unless + * the span is longer than 4 pixels. */ + if (rxi - lxi > 4) + { + if (fill_start < 0) + { + fill_start = lxi; + fill_end = rxi; + fill_size++; + } + else + { + if (lxi >= fill_end || rxi < fill_start) + { + /* We're beyond what we saved, just fill it */ + ADD_SATURATE_8 (ap + fill_start, + fill_size * N_X_FRAC (8), + fill_end - fill_start); + fill_start = lxi; + fill_end = rxi; + fill_size = 1; + } + else + { + /* Update fill_start */ + if (lxi > fill_start) + { + ADD_SATURATE_8 (ap + fill_start, + fill_size * N_X_FRAC (8), + lxi - fill_start); + fill_start = lxi; + } + else if (lxi < fill_start) + { + ADD_SATURATE_8 (ap + lxi, N_X_FRAC (8), + fill_start - lxi); + } + + /* Update fill_end */ + if (rxi < fill_end) + { + ADD_SATURATE_8 (ap + rxi, + fill_size * N_X_FRAC (8), + fill_end - rxi); + fill_end = rxi; + } + else if (fill_end < rxi) + { + ADD_SATURATE_8 (ap + fill_end, + N_X_FRAC (8), + rxi - fill_end); + } + fill_size++; + } + } + } + else + { + ADD_SATURATE_8 (ap + lxi, N_X_FRAC (8), rxi - lxi); + } + + WRITE (image, ap + rxi, clip255 (READ (image, ap + rxi) + rxs)); + } + } + + if (y == b) + { + /* We're done, make sure we clean up any remaining fill. */ + if (fill_start != fill_end) + { + if (fill_size == N_Y_FRAC (8)) + { + MEMSET_WRAPPED (image, ap + fill_start, + 0xff, fill_end - fill_start); + } + else + { + ADD_SATURATE_8 (ap + fill_start, fill_size * N_X_FRAC (8), + fill_end - fill_start); + } + } + break; + } + + if (pixman_fixed_frac (y) != Y_FRAC_LAST (8)) + { + RENDER_EDGE_STEP_SMALL (l); + RENDER_EDGE_STEP_SMALL (r); + y += STEP_Y_SMALL (8); + } + else + { + RENDER_EDGE_STEP_BIG (l); + RENDER_EDGE_STEP_BIG (r); + y += STEP_Y_BIG (8); + if (fill_start != fill_end) + { + if (fill_size == N_Y_FRAC (8)) + { + MEMSET_WRAPPED (image, ap + fill_start, + 0xff, fill_end - fill_start); + } + else + { + ADD_SATURATE_8 (ap + fill_start, fill_size * N_X_FRAC (8), + fill_end - fill_start); + } + + fill_start = fill_end = -1; + fill_size = 0; + } + + line += stride; + } + } +} + +#ifndef PIXMAN_FB_ACCESSORS +static +#endif +void +PIXMAN_RASTERIZE_EDGES (pixman_image_t *image, + pixman_edge_t * l, + pixman_edge_t * r, + pixman_fixed_t t, + pixman_fixed_t b) +{ + switch (PIXMAN_FORMAT_BPP (image->bits.format)) + { + case 1: + rasterize_edges_1 (image, l, r, t, b); + break; + + case 4: + rasterize_edges_4 (image, l, r, t, b); + break; + + case 8: + rasterize_edges_8 (image, l, r, t, b); + break; + + default: + break; + } +} + +#ifndef PIXMAN_FB_ACCESSORS + +PIXMAN_EXPORT void +pixman_rasterize_edges (pixman_image_t *image, + pixman_edge_t * l, + pixman_edge_t * r, + pixman_fixed_t t, + pixman_fixed_t b) +{ + return_if_fail (image->type == BITS); + + if (image->bits.read_func || image->bits.write_func) + pixman_rasterize_edges_accessors (image, l, r, t, b); + else + pixman_rasterize_edges_no_accessors (image, l, r, t, b); +} + +#endif diff --git a/programs/develop/libraries/pixman/pixman-fast-path.c b/programs/develop/libraries/pixman/pixman-fast-path.c new file mode 100644 index 0000000000..5d5fa956c8 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-fast-path.c @@ -0,0 +1,1852 @@ +/* -*- Mode: c; c-basic-offset: 4; tab-width: 8; indent-tabs-mode: t; -*- */ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of SuSE not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. SuSE makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * SuSE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL SuSE + * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + * + * Author: Keith Packard, SuSE, Inc. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include +#include +#include "pixman-private.h" +#include "pixman-combine32.h" +#include "pixman-fast-path.h" + +static force_inline uint32_t +fetch_24 (uint8_t *a) +{ + if (((unsigned long)a) & 1) + { +#ifdef WORDS_BIGENDIAN + return (*a << 16) | (*(uint16_t *)(a + 1)); +#else + return *a | (*(uint16_t *)(a + 1) << 8); +#endif + } + else + { +#ifdef WORDS_BIGENDIAN + return (*(uint16_t *)a << 8) | *(a + 2); +#else + return *(uint16_t *)a | (*(a + 2) << 16); +#endif + } +} + +static force_inline void +store_24 (uint8_t *a, + uint32_t v) +{ + if (((unsigned long)a) & 1) + { +#ifdef WORDS_BIGENDIAN + *a = (uint8_t) (v >> 16); + *(uint16_t *)(a + 1) = (uint16_t) (v); +#else + *a = (uint8_t) (v); + *(uint16_t *)(a + 1) = (uint16_t) (v >> 8); +#endif + } + else + { +#ifdef WORDS_BIGENDIAN + *(uint16_t *)a = (uint16_t)(v >> 8); + *(a + 2) = (uint8_t)v; +#else + *(uint16_t *)a = (uint16_t)v; + *(a + 2) = (uint8_t)(v >> 16); +#endif + } +} + +static force_inline uint32_t +over (uint32_t src, + uint32_t dest) +{ + uint32_t a = ~src >> 24; + + UN8x4_MUL_UN8_ADD_UN8x4 (dest, a, src); + + return dest; +} + +static uint32_t +in (uint32_t x, + uint8_t y) +{ + uint16_t a = y; + + UN8x4_MUL_UN8 (x, a); + + return x; +} + +/* + * Naming convention: + * + * op_src_mask_dest + */ +static void +fast_composite_over_x888_8_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *src, *src_line; + uint32_t *dst, *dst_line; + uint8_t *mask, *mask_line; + int src_stride, mask_stride, dst_stride; + uint8_t m; + uint32_t s, d; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + while (height--) + { + src = src_line; + src_line += src_stride; + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + + w = width; + while (w--) + { + m = *mask++; + if (m) + { + s = *src | 0xff000000; + + if (m == 0xff) + { + *dst = s; + } + else + { + d = in (s, m); + *dst = over (d, *dst); + } + } + src++; + dst++; + } + } +} + +static void +fast_composite_in_n_8_8 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dest_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint8_t *dst_line, *dst; + uint8_t *mask_line, *mask, m; + int dst_stride, mask_stride; + int32_t w; + uint16_t t; + + src = _pixman_image_get_solid (src_image, dest_image->bits.format); + + srca = src >> 24; + + PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + if (srca == 0xff) + { + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + m = *mask++; + + if (m == 0) + *dst = 0; + else if (m != 0xff) + *dst = MUL_UN8 (m, *dst, t); + + dst++; + } + } + } + else + { + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + m = *mask++; + m = MUL_UN8 (m, srca, t); + + if (m == 0) + *dst = 0; + else if (m != 0xff) + *dst = MUL_UN8 (m, *dst, t); + + dst++; + } + } + } +} + +static void +fast_composite_in_8_8 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dest_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint8_t *dst_line, *dst; + uint8_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + uint8_t s; + uint16_t t; + + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint8_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + s = *src++; + + if (s == 0) + *dst = 0; + else if (s != 0xff) + *dst = MUL_UN8 (s, *dst, t); + + dst++; + } + } +} + +static void +fast_composite_over_n_8_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint32_t *dst_line, *dst, d; + uint8_t *mask_line, *mask, m; + int dst_stride, mask_stride; + int32_t w; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + m = *mask++; + if (m == 0xff) + { + if (srca == 0xff) + *dst = src; + else + *dst = over (src, *dst); + } + else if (m) + { + d = in (src, m); + *dst = over (d, *dst); + } + dst++; + } + } +} + +static void +fast_composite_add_n_8888_8888_ca (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca, s; + uint32_t *dst_line, *dst, d; + uint32_t *mask_line, *mask, ma; + int dst_stride, mask_stride; + int32_t w; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + ma = *mask++; + + if (ma) + { + d = *dst; + s = src; + + UN8x4_MUL_UN8x4_ADD_UN8x4 (s, ma, d); + + *dst = s; + } + + dst++; + } + } +} + +static void +fast_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca, s; + uint32_t *dst_line, *dst, d; + uint32_t *mask_line, *mask, ma; + int dst_stride, mask_stride; + int32_t w; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + ma = *mask++; + if (ma == 0xffffffff) + { + if (srca == 0xff) + *dst = src; + else + *dst = over (src, *dst); + } + else if (ma) + { + d = *dst; + s = src; + + UN8x4_MUL_UN8x4 (s, ma); + UN8x4_MUL_UN8 (ma, srca); + ma = ~ma; + UN8x4_MUL_UN8x4_ADD_UN8x4 (d, ma, s); + + *dst = d; + } + + dst++; + } + } +} + +static void +fast_composite_over_n_8_0888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint8_t *dst_line, *dst; + uint32_t d; + uint8_t *mask_line, *mask, m; + int dst_stride, mask_stride; + int32_t w; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 3); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + m = *mask++; + if (m == 0xff) + { + if (srca == 0xff) + { + d = src; + } + else + { + d = fetch_24 (dst); + d = over (src, d); + } + store_24 (dst, d); + } + else if (m) + { + d = over (in (src, m), fetch_24 (dst)); + store_24 (dst, d); + } + dst += 3; + } + } +} + +static void +fast_composite_over_n_8_0565 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint16_t *dst_line, *dst; + uint32_t d; + uint8_t *mask_line, *mask, m; + int dst_stride, mask_stride; + int32_t w; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + m = *mask++; + if (m == 0xff) + { + if (srca == 0xff) + { + d = src; + } + else + { + d = *dst; + d = over (src, CONVERT_0565_TO_0888 (d)); + } + *dst = CONVERT_8888_TO_0565 (d); + } + else if (m) + { + d = *dst; + d = over (in (src, m), CONVERT_0565_TO_0888 (d)); + *dst = CONVERT_8888_TO_0565 (d); + } + dst++; + } + } +} + +static void +fast_composite_over_n_8888_0565_ca (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca, s; + uint16_t src16; + uint16_t *dst_line, *dst; + uint32_t d; + uint32_t *mask_line, *mask, ma; + int dst_stride, mask_stride; + int32_t w; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + src16 = CONVERT_8888_TO_0565 (src); + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + ma = *mask++; + if (ma == 0xffffffff) + { + if (srca == 0xff) + { + *dst = src16; + } + else + { + d = *dst; + d = over (src, CONVERT_0565_TO_0888 (d)); + *dst = CONVERT_8888_TO_0565 (d); + } + } + else if (ma) + { + d = *dst; + d = CONVERT_0565_TO_0888 (d); + + s = src; + + UN8x4_MUL_UN8x4 (s, ma); + UN8x4_MUL_UN8 (ma, srca); + ma = ~ma; + UN8x4_MUL_UN8x4_ADD_UN8x4 (d, ma, s); + + *dst = CONVERT_8888_TO_0565 (d); + } + dst++; + } + } +} + +static void +fast_composite_over_8888_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line, *dst; + uint32_t *src_line, *src, s; + int dst_stride, src_stride; + uint8_t a; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + s = *src++; + a = s >> 24; + if (a == 0xff) + *dst = s; + else if (s) + *dst = over (s, *dst); + dst++; + } + } +} + +static void +fast_composite_src_x888_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + *dst++ = (*src++) | 0xff000000; + } +} + +#if 0 +static void +fast_composite_over_8888_0888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint8_t *dst_line, *dst; + uint32_t d; + uint32_t *src_line, *src, s; + uint8_t a; + int dst_stride, src_stride; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 3); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + s = *src++; + a = s >> 24; + if (a) + { + if (a == 0xff) + d = s; + else + d = over (s, fetch_24 (dst)); + + store_24 (dst, d); + } + dst += 3; + } + } +} +#endif + +static void +fast_composite_over_8888_0565 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint16_t *dst_line, *dst; + uint32_t d; + uint32_t *src_line, *src, s; + uint8_t a; + int dst_stride, src_stride; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + s = *src++; + a = s >> 24; + if (s) + { + if (a == 0xff) + { + d = s; + } + else + { + d = *dst; + d = over (s, CONVERT_0565_TO_0888 (d)); + } + *dst = CONVERT_8888_TO_0565 (d); + } + dst++; + } + } +} + +static void +fast_composite_src_x888_0565 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint16_t *dst_line, *dst; + uint32_t *src_line, *src, s; + int dst_stride, src_stride; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + s = *src++; + *dst = CONVERT_8888_TO_0565 (s); + dst++; + } + } +} + +static void +fast_composite_add_8_8 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint8_t *dst_line, *dst; + uint8_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + uint8_t s, d; + uint16_t t; + + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint8_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + s = *src++; + if (s) + { + if (s != 0xff) + { + d = *dst; + t = d + s; + s = t | (0 - (t >> 8)); + } + *dst = s; + } + dst++; + } + } +} + +static void +fast_composite_add_8888_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + uint32_t s, d; + + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + s = *src++; + if (s) + { + if (s != 0xffffffff) + { + d = *dst; + if (d) + UN8x4_ADD_UN8x4 (s, d); + } + *dst = s; + } + dst++; + } + } +} + +static void +fast_composite_add_n_8_8 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint8_t *dst_line, *dst; + uint8_t *mask_line, *mask; + int dst_stride, mask_stride; + int32_t w; + uint32_t src; + uint8_t sa; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + sa = (src >> 24); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + while (w--) + { + uint16_t tmp; + uint16_t a; + uint32_t m, d; + uint32_t r; + + a = *mask++; + d = *dst; + + m = MUL_UN8 (sa, a, tmp); + r = ADD_UN8 (m, d, tmp); + + *dst++ = r; + } + } +} + +#ifdef WORDS_BIGENDIAN +#define CREATE_BITMASK(n) (0x80000000 >> (n)) +#define UPDATE_BITMASK(n) ((n) >> 1) +#else +#define CREATE_BITMASK(n) (1 << (n)) +#define UPDATE_BITMASK(n) ((n) << 1) +#endif + +#define TEST_BIT(p, n) \ + (*((p) + ((n) >> 5)) & CREATE_BITMASK ((n) & 31)) +#define SET_BIT(p, n) \ + do { *((p) + ((n) >> 5)) |= CREATE_BITMASK ((n) & 31); } while (0); + +static void +fast_composite_add_1000_1000 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (src_image, 0, src_y, uint32_t, + src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE (dst_image, 0, dest_y, uint32_t, + dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + /* + * TODO: improve performance by processing uint32_t data instead + * of individual bits + */ + if (TEST_BIT (src, src_x + w)) + SET_BIT (dst, dest_x + w); + } + } +} + +static void +fast_composite_over_n_1_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint32_t *dst, *dst_line; + uint32_t *mask, *mask_line; + int mask_stride, dst_stride; + uint32_t bitcache, bitmask; + int32_t w; + + if (width <= 0) + return; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, + dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, 0, mask_y, uint32_t, + mask_stride, mask_line, 1); + mask_line += mask_x >> 5; + + if (srca == 0xff) + { + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + bitcache = *mask++; + bitmask = CREATE_BITMASK (mask_x & 31); + + while (w--) + { + if (bitmask == 0) + { + bitcache = *mask++; + bitmask = CREATE_BITMASK (0); + } + if (bitcache & bitmask) + *dst = src; + bitmask = UPDATE_BITMASK (bitmask); + dst++; + } + } + } + else + { + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + bitcache = *mask++; + bitmask = CREATE_BITMASK (mask_x & 31); + + while (w--) + { + if (bitmask == 0) + { + bitcache = *mask++; + bitmask = CREATE_BITMASK (0); + } + if (bitcache & bitmask) + *dst = over (src, *dst); + bitmask = UPDATE_BITMASK (bitmask); + dst++; + } + } + } +} + +static void +fast_composite_over_n_1_0565 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint16_t *dst, *dst_line; + uint32_t *mask, *mask_line; + int mask_stride, dst_stride; + uint32_t bitcache, bitmask; + int32_t w; + uint32_t d; + uint16_t src565; + + if (width <= 0) + return; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, + dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, 0, mask_y, uint32_t, + mask_stride, mask_line, 1); + mask_line += mask_x >> 5; + + if (srca == 0xff) + { + src565 = CONVERT_8888_TO_0565 (src); + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + bitcache = *mask++; + bitmask = CREATE_BITMASK (mask_x & 31); + + while (w--) + { + if (bitmask == 0) + { + bitcache = *mask++; + bitmask = CREATE_BITMASK (0); + } + if (bitcache & bitmask) + *dst = src565; + bitmask = UPDATE_BITMASK (bitmask); + dst++; + } + } + } + else + { + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + bitcache = *mask++; + bitmask = CREATE_BITMASK (mask_x & 31); + + while (w--) + { + if (bitmask == 0) + { + bitcache = *mask++; + bitmask = CREATE_BITMASK (0); + } + if (bitcache & bitmask) + { + d = over (src, CONVERT_0565_TO_0888 (*dst)); + *dst = CONVERT_8888_TO_0565 (d); + } + bitmask = UPDATE_BITMASK (bitmask); + dst++; + } + } + } +} + +/* + * Simple bitblt + */ + +static void +fast_composite_solid_fill (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src; + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + if (dst_image->bits.format == PIXMAN_a8) + { + src = src >> 24; + } + else if (dst_image->bits.format == PIXMAN_r5g6b5 || + dst_image->bits.format == PIXMAN_b5g6r5) + { + src = CONVERT_8888_TO_0565 (src); + } + + pixman_fill (dst_image->bits.bits, dst_image->bits.rowstride, + PIXMAN_FORMAT_BPP (dst_image->bits.format), + dest_x, dest_y, + width, height, + src); +} + +static void +fast_composite_src_memcpy (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + int bpp = PIXMAN_FORMAT_BPP (dst_image->bits.format) / 8; + uint32_t n_bytes = width * bpp; + int dst_stride, src_stride; + uint8_t *dst; + uint8_t *src; + + src_stride = src_image->bits.rowstride * 4; + dst_stride = dst_image->bits.rowstride * 4; + + src = (uint8_t *)src_image->bits.bits + src_y * src_stride + src_x * bpp; + dst = (uint8_t *)dst_image->bits.bits + dest_y * dst_stride + dest_x * bpp; + + while (height--) + { + memcpy (dst, src, n_bytes); + + dst += dst_stride; + src += src_stride; + } +} + +FAST_NEAREST (8888_8888_cover, 8888, 8888, uint32_t, uint32_t, SRC, COVER); +FAST_NEAREST (8888_8888_none, 8888, 8888, uint32_t, uint32_t, SRC, NONE); +FAST_NEAREST (8888_8888_pad, 8888, 8888, uint32_t, uint32_t, SRC, PAD); +FAST_NEAREST (8888_8888_normal, 8888, 8888, uint32_t, uint32_t, SRC, NORMAL); +FAST_NEAREST (8888_8888_cover, 8888, 8888, uint32_t, uint32_t, OVER, COVER); +FAST_NEAREST (8888_8888_none, 8888, 8888, uint32_t, uint32_t, OVER, NONE); +FAST_NEAREST (8888_8888_pad, 8888, 8888, uint32_t, uint32_t, OVER, PAD); +FAST_NEAREST (8888_8888_normal, 8888, 8888, uint32_t, uint32_t, OVER, NORMAL); +FAST_NEAREST (8888_565_cover, 8888, 0565, uint32_t, uint16_t, SRC, COVER); +FAST_NEAREST (8888_565_none, 8888, 0565, uint32_t, uint16_t, SRC, NONE); +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); + +/* Use more unrolling for src_0565_0565 because it is typically CPU bound */ +static force_inline void +scaled_nearest_scanline_565_565_SRC (uint16_t * dst, + uint16_t * src, + int32_t w, + pixman_fixed_t vx, + pixman_fixed_t unit_x, + pixman_fixed_t max_vx) +{ + uint16_t tmp1, tmp2, tmp3, tmp4; + while ((w -= 4) >= 0) + { + tmp1 = src[pixman_fixed_to_int (vx)]; + vx += unit_x; + tmp2 = src[pixman_fixed_to_int (vx)]; + vx += unit_x; + tmp3 = src[pixman_fixed_to_int (vx)]; + vx += unit_x; + tmp4 = src[pixman_fixed_to_int (vx)]; + vx += unit_x; + *dst++ = tmp1; + *dst++ = tmp2; + *dst++ = tmp3; + *dst++ = tmp4; + } + if (w & 2) + { + tmp1 = src[pixman_fixed_to_int (vx)]; + vx += unit_x; + tmp2 = src[pixman_fixed_to_int (vx)]; + vx += unit_x; + *dst++ = tmp1; + *dst++ = tmp2; + } + if (w & 1) + *dst++ = src[pixman_fixed_to_int (vx)]; +} + +FAST_NEAREST_MAINLOOP (565_565_cover_SRC, + scaled_nearest_scanline_565_565_SRC, + uint16_t, uint16_t, COVER); +FAST_NEAREST_MAINLOOP (565_565_none_SRC, + scaled_nearest_scanline_565_565_SRC, + uint16_t, uint16_t, NONE); +FAST_NEAREST_MAINLOOP (565_565_pad_SRC, + scaled_nearest_scanline_565_565_SRC, + uint16_t, uint16_t, PAD); + +static force_inline uint32_t +fetch_nearest (pixman_repeat_t src_repeat, + pixman_format_code_t format, + uint32_t *src, int x, int src_width) +{ + if (repeat (src_repeat, &x, src_width)) + { + if (format == PIXMAN_x8r8g8b8) + return *(src + x) | 0xff000000; + else + return *(src + x); + } + else + { + return 0; + } +} + +static force_inline void +combine_over (uint32_t s, uint32_t *dst) +{ + if (s) + { + uint8_t ia = 0xff - (s >> 24); + + if (ia) + UN8x4_MUL_UN8_ADD_UN8x4 (*dst, ia, s); + else + *dst = s; + } +} + +static force_inline void +combine_src (uint32_t s, uint32_t *dst) +{ + *dst = s; +} + +static void +fast_composite_scaled_nearest (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line; + uint32_t *src_line; + int dst_stride, src_stride; + int src_width, src_height; + pixman_repeat_t src_repeat; + pixman_fixed_t unit_x, unit_y; + pixman_format_code_t src_format; + pixman_vector_t v; + pixman_fixed_t vy; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + /* pass in 0 instead of src_x and src_y because src_x and src_y need to be + * transformed from destination space to source space + */ + PIXMAN_IMAGE_GET_LINE (src_image, 0, 0, uint32_t, src_stride, src_line, 1); + + /* reference point is the center of the pixel */ + v.vector[0] = pixman_int_to_fixed (src_x) + pixman_fixed_1 / 2; + v.vector[1] = pixman_int_to_fixed (src_y) + pixman_fixed_1 / 2; + v.vector[2] = pixman_fixed_1; + + if (!pixman_transform_point_3d (src_image->common.transform, &v)) + return; + + unit_x = src_image->common.transform->matrix[0][0]; + unit_y = src_image->common.transform->matrix[1][1]; + + /* Round down to closest integer, ensuring that 0.5 rounds to 0, not 1 */ + v.vector[0] -= pixman_fixed_e; + v.vector[1] -= pixman_fixed_e; + + src_height = src_image->bits.height; + src_width = src_image->bits.width; + src_repeat = src_image->common.repeat; + src_format = src_image->bits.format; + + vy = v.vector[1]; + while (height--) + { + pixman_fixed_t vx = v.vector[0]; + int y = pixman_fixed_to_int (vy); + uint32_t *dst = dst_line; + + dst_line += dst_stride; + + /* adjust the y location by a unit vector in the y direction + * this is equivalent to transforming y+1 of the destination point to source space */ + vy += unit_y; + + if (!repeat (src_repeat, &y, src_height)) + { + if (op == PIXMAN_OP_SRC) + memset (dst, 0, sizeof (*dst) * width); + } + else + { + int w = width; + + uint32_t *src = src_line + y * src_stride; + + while (w >= 2) + { + uint32_t s1, s2; + int x1, x2; + + x1 = pixman_fixed_to_int (vx); + vx += unit_x; + + x2 = pixman_fixed_to_int (vx); + vx += unit_x; + + w -= 2; + + s1 = fetch_nearest (src_repeat, src_format, src, x1, src_width); + s2 = fetch_nearest (src_repeat, src_format, src, x2, src_width); + + if (op == PIXMAN_OP_OVER) + { + combine_over (s1, dst++); + combine_over (s2, dst++); + } + else + { + combine_src (s1, dst++); + combine_src (s2, dst++); + } + } + + while (w--) + { + uint32_t s; + int x; + + x = pixman_fixed_to_int (vx); + vx += unit_x; + + s = fetch_nearest (src_repeat, src_format, src, x, src_width); + + if (op == PIXMAN_OP_OVER) + combine_over (s, dst++); + else + combine_src (s, dst++); + } + } + } +} + +static const pixman_fast_path_t c_fast_paths[] = +{ + PIXMAN_STD_FAST_PATH (OVER, solid, a8, r5g6b5, fast_composite_over_n_8_0565), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, b5g6r5, fast_composite_over_n_8_0565), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, r8g8b8, fast_composite_over_n_8_0888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, b8g8r8, fast_composite_over_n_8_0888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8r8g8b8, fast_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8r8g8b8, fast_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8b8g8r8, fast_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8b8g8r8, fast_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a1, a8r8g8b8, fast_composite_over_n_1_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a1, x8r8g8b8, fast_composite_over_n_1_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a1, a8b8g8r8, fast_composite_over_n_1_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a1, x8b8g8r8, fast_composite_over_n_1_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a1, r5g6b5, fast_composite_over_n_1_0565), + PIXMAN_STD_FAST_PATH (OVER, solid, a1, b5g6r5, fast_composite_over_n_1_0565), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, a8r8g8b8, fast_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, x8r8g8b8, fast_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, r5g6b5, fast_composite_over_n_8888_0565_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, a8b8g8r8, fast_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, x8b8g8r8, fast_composite_over_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, b5g6r5, fast_composite_over_n_8888_0565_ca), + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, a8, x8r8g8b8, fast_composite_over_x888_8_8888), + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, a8, a8r8g8b8, fast_composite_over_x888_8_8888), + PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, a8, x8b8g8r8, fast_composite_over_x888_8_8888), + PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, a8, a8b8g8r8, fast_composite_over_x888_8_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, a8r8g8b8, fast_composite_over_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, x8r8g8b8, fast_composite_over_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, r5g6b5, fast_composite_over_8888_0565), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, a8b8g8r8, fast_composite_over_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, x8b8g8r8, fast_composite_over_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, b5g6r5, fast_composite_over_8888_0565), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, null, a8r8g8b8, fast_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, null, a8b8g8r8, fast_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8, null, a8, fast_composite_add_8_8), + PIXMAN_STD_FAST_PATH (ADD, a1, null, a1, fast_composite_add_1000_1000), + PIXMAN_STD_FAST_PATH_CA (ADD, solid, a8r8g8b8, a8r8g8b8, fast_composite_add_n_8888_8888_ca), + PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8, fast_composite_add_n_8_8), + PIXMAN_STD_FAST_PATH (SRC, solid, null, a8r8g8b8, fast_composite_solid_fill), + PIXMAN_STD_FAST_PATH (SRC, solid, null, x8r8g8b8, fast_composite_solid_fill), + PIXMAN_STD_FAST_PATH (SRC, solid, null, a8b8g8r8, fast_composite_solid_fill), + PIXMAN_STD_FAST_PATH (SRC, solid, null, x8b8g8r8, fast_composite_solid_fill), + PIXMAN_STD_FAST_PATH (SRC, solid, null, a8, fast_composite_solid_fill), + PIXMAN_STD_FAST_PATH (SRC, solid, null, r5g6b5, fast_composite_solid_fill), + PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, a8r8g8b8, fast_composite_src_x888_8888), + PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, a8b8g8r8, fast_composite_src_x888_8888), + PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, x8r8g8b8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, a8r8g8b8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, x8r8g8b8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, x8b8g8r8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, a8b8g8r8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, x8b8g8r8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, b8g8r8a8, null, b8g8r8x8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, b8g8r8a8, null, b8g8r8a8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, b8g8r8x8, null, b8g8r8x8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, r5g6b5, null, r5g6b5, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, b5g6r5, null, b5g6r5, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, r8g8b8, null, r8g8b8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, b8g8r8, null, b8g8r8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, x1r5g5b5, null, x1r5g5b5, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, a1r5g5b5, null, x1r5g5b5, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, a8, null, a8, fast_composite_src_memcpy), + PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, r5g6b5, fast_composite_src_x888_0565), + PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, r5g6b5, fast_composite_src_x888_0565), + PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, b5g6r5, fast_composite_src_x888_0565), + PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, b5g6r5, fast_composite_src_x888_0565), + PIXMAN_STD_FAST_PATH (IN, a8, null, a8, fast_composite_in_8_8), + PIXMAN_STD_FAST_PATH (IN, solid, a8, a8, fast_composite_in_n_8_8), + + SIMPLE_NEAREST_FAST_PATH (SRC, x8r8g8b8, x8r8g8b8, 8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, 8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, x8b8g8r8, x8b8g8r8, 8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, x8b8g8r8, 8888_8888), + + SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, 8888_8888), + SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, a8b8g8r8, 8888_8888), + + SIMPLE_NEAREST_FAST_PATH (SRC, x8r8g8b8, r5g6b5, 8888_565), + SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, r5g6b5, 8888_565), + + SIMPLE_NEAREST_FAST_PATH (SRC, r5g6b5, r5g6b5, 565_565), + + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8, 8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8, 8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, 8888_8888), + SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8, 8888_8888), + + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, r5g6b5, 8888_565), + +#define NEAREST_FAST_PATH(op,s,d) \ + { PIXMAN_OP_ ## op, \ + PIXMAN_ ## s, SCALED_NEAREST_FLAGS, \ + PIXMAN_null, 0, \ + PIXMAN_ ## d, FAST_PATH_STD_DEST_FLAGS, \ + fast_composite_scaled_nearest, \ + } + + NEAREST_FAST_PATH (SRC, x8r8g8b8, x8r8g8b8), + NEAREST_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8), + NEAREST_FAST_PATH (SRC, x8b8g8r8, x8b8g8r8), + NEAREST_FAST_PATH (SRC, a8b8g8r8, x8b8g8r8), + + NEAREST_FAST_PATH (SRC, x8r8g8b8, a8r8g8b8), + NEAREST_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8), + NEAREST_FAST_PATH (SRC, x8b8g8r8, a8b8g8r8), + NEAREST_FAST_PATH (SRC, a8b8g8r8, a8b8g8r8), + + NEAREST_FAST_PATH (OVER, x8r8g8b8, x8r8g8b8), + NEAREST_FAST_PATH (OVER, a8r8g8b8, x8r8g8b8), + NEAREST_FAST_PATH (OVER, x8b8g8r8, x8b8g8r8), + NEAREST_FAST_PATH (OVER, a8b8g8r8, x8b8g8r8), + + NEAREST_FAST_PATH (OVER, x8r8g8b8, a8r8g8b8), + NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8), + NEAREST_FAST_PATH (OVER, x8b8g8r8, a8b8g8r8), + NEAREST_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8), + + { PIXMAN_OP_NONE }, +}; + +static void +pixman_fill8 (uint32_t *bits, + int stride, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + int byte_stride = stride * (int) sizeof (uint32_t); + uint8_t *dst = (uint8_t *) bits; + uint8_t v = xor & 0xff; + int i; + + dst = dst + y * byte_stride + x; + + while (height--) + { + for (i = 0; i < width; ++i) + dst[i] = v; + + dst += byte_stride; + } +} + +static void +pixman_fill16 (uint32_t *bits, + int stride, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + int short_stride = + (stride * (int)sizeof (uint32_t)) / (int)sizeof (uint16_t); + uint16_t *dst = (uint16_t *)bits; + uint16_t v = xor & 0xffff; + int i; + + dst = dst + y * short_stride + x; + + while (height--) + { + for (i = 0; i < width; ++i) + dst[i] = v; + + dst += short_stride; + } +} + +static void +pixman_fill32 (uint32_t *bits, + int stride, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + int i; + + bits = bits + y * stride + x; + + while (height--) + { + for (i = 0; i < width; ++i) + bits[i] = xor; + + bits += stride; + } +} + +static pixman_bool_t +fast_path_fill (pixman_implementation_t *imp, + uint32_t * bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + switch (bpp) + { + case 8: + pixman_fill8 (bits, stride, x, y, width, height, xor); + break; + + case 16: + pixman_fill16 (bits, stride, x, y, width, height, xor); + break; + + case 32: + pixman_fill32 (bits, stride, x, y, width, height, xor); + break; + + default: + return _pixman_implementation_fill ( + imp->delegate, bits, stride, bpp, x, y, width, height, xor); + break; + } + + return TRUE; +} + +pixman_implementation_t * +_pixman_implementation_create_fast_path (void) +{ + pixman_implementation_t *general = _pixman_implementation_create_general (); + pixman_implementation_t *imp = _pixman_implementation_create (general, c_fast_paths); + + imp->fill = fast_path_fill; + + return imp; +} diff --git a/programs/develop/libraries/pixman/pixman-fast-path.h b/programs/develop/libraries/pixman/pixman-fast-path.h new file mode 100644 index 0000000000..98ef81ea7d --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-fast-path.h @@ -0,0 +1,451 @@ +/* -*- Mode: c; c-basic-offset: 4; tab-width: 8; indent-tabs-mode: t; -*- */ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of SuSE not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. SuSE makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * SuSE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL SuSE + * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + * + * Author: Keith Packard, SuSE, Inc. + */ + +#ifndef PIXMAN_FAST_PATH_H__ +#define PIXMAN_FAST_PATH_H__ + +#include "pixman-private.h" + +#define PIXMAN_REPEAT_COVER -1 + +static force_inline pixman_bool_t +repeat (pixman_repeat_t repeat, int *c, int size) +{ + if (repeat == PIXMAN_REPEAT_NONE) + { + if (*c < 0 || *c >= size) + return FALSE; + } + else if (repeat == PIXMAN_REPEAT_NORMAL) + { + while (*c >= size) + *c -= size; + while (*c < 0) + *c += size; + } + else if (repeat == PIXMAN_REPEAT_PAD) + { + *c = CLIP (*c, 0, size - 1); + } + else /* REFLECT */ + { + *c = MOD (*c, size * 2); + if (*c >= size) + *c = size * 2 - *c - 1; + } + return TRUE; +} + +/* + * For each scanline fetched from source image with PAD repeat: + * - calculate how many pixels need to be padded on the left side + * - calculate how many pixels need to be padded on the right side + * - update width to only count pixels which are fetched from the image + * All this information is returned via 'width', 'left_pad', 'right_pad' + * arguments. The code is assuming that 'unit_x' is positive. + * + * Note: 64-bit math is used in order to avoid potential overflows, which + * is probably excessive in many cases. This particular function + * may need its own correctness test and performance tuning. + */ +static force_inline void +pad_repeat_get_scanline_bounds (int32_t source_image_width, + pixman_fixed_t vx, + pixman_fixed_t unit_x, + int32_t * width, + int32_t * left_pad, + int32_t * right_pad) +{ + int64_t max_vx = (int64_t) source_image_width << 16; + int64_t tmp; + if (vx < 0) + { + tmp = ((int64_t) unit_x - 1 - vx) / unit_x; + if (tmp > *width) + { + *left_pad = *width; + *width = 0; + } + else + { + *left_pad = (int32_t) tmp; + *width -= (int32_t) tmp; + } + } + else + { + *left_pad = 0; + } + tmp = ((int64_t) unit_x - 1 - vx + max_vx) / unit_x - *left_pad; + if (tmp < 0) + { + *right_pad = *width; + *width = 0; + } + else if (tmp >= *width) + { + *right_pad = 0; + } + else + { + *right_pad = *width - (int32_t) tmp; + *width = (int32_t) tmp; + } +} + +/* A macroified version of specialized nearest scalers for some + * common 8888 and 565 formats. It supports SRC and OVER ops. + * + * There are two repeat versions, one that handles repeat normal, + * and one without repeat handling that only works if the src region + * used is completely covered by the pre-repeated source samples. + * + * The loops are unrolled to process two pixels per iteration for better + * performance on most CPU architectures (superscalar processors + * can issue several operations simultaneously, other processors can hide + * instructions latencies by pipelining operations). Unrolling more + * does not make much sense because the compiler will start running out + * of spare registers soon. + */ + +#define GET_8888_ALPHA(s) ((s) >> 24) + /* This is not actually used since we don't have an OVER with + 565 source, but it is needed to build. */ +#define GET_0565_ALPHA(s) 0xff + +#define FAST_NEAREST_SCANLINE(scanline_func_name, SRC_FORMAT, DST_FORMAT, \ + src_type_t, dst_type_t, OP, repeat_mode) \ +static force_inline void \ +scanline_func_name (dst_type_t *dst, \ + src_type_t *src, \ + int32_t w, \ + pixman_fixed_t vx, \ + pixman_fixed_t unit_x, \ + pixman_fixed_t max_vx) \ +{ \ + uint32_t d; \ + src_type_t s1, s2; \ + uint8_t a1, a2; \ + int x1, x2; \ + \ + if (PIXMAN_OP_ ## OP != PIXMAN_OP_SRC && PIXMAN_OP_ ## OP != PIXMAN_OP_OVER) \ + abort(); \ + \ + while ((w -= 2) >= 0) \ + { \ + x1 = vx >> 16; \ + vx += unit_x; \ + if (PIXMAN_REPEAT_ ## repeat_mode == PIXMAN_REPEAT_NORMAL) \ + { \ + /* This works because we know that unit_x is positive */ \ + while (vx >= max_vx) \ + vx -= max_vx; \ + } \ + s1 = src[x1]; \ + \ + x2 = vx >> 16; \ + vx += unit_x; \ + if (PIXMAN_REPEAT_ ## repeat_mode == PIXMAN_REPEAT_NORMAL) \ + { \ + /* This works because we know that unit_x is positive */ \ + while (vx >= max_vx) \ + vx -= max_vx; \ + } \ + s2 = src[x2]; \ + \ + if (PIXMAN_OP_ ## OP == PIXMAN_OP_OVER) \ + { \ + a1 = GET_ ## SRC_FORMAT ## _ALPHA(s1); \ + a2 = GET_ ## SRC_FORMAT ## _ALPHA(s2); \ + \ + if (a1 == 0xff) \ + { \ + *dst = CONVERT_ ## SRC_FORMAT ## _TO_ ## DST_FORMAT (s1); \ + } \ + else if (s1) \ + { \ + d = CONVERT_ ## DST_FORMAT ## _TO_8888 (*dst); \ + s1 = CONVERT_ ## SRC_FORMAT ## _TO_8888 (s1); \ + a1 ^= 0xff; \ + UN8x4_MUL_UN8_ADD_UN8x4 (d, a1, s1); \ + *dst = CONVERT_8888_TO_ ## DST_FORMAT (d); \ + } \ + dst++; \ + \ + if (a2 == 0xff) \ + { \ + *dst = CONVERT_ ## SRC_FORMAT ## _TO_ ## DST_FORMAT (s2); \ + } \ + else if (s2) \ + { \ + d = CONVERT_## DST_FORMAT ## _TO_8888 (*dst); \ + s2 = CONVERT_## SRC_FORMAT ## _TO_8888 (s2); \ + a2 ^= 0xff; \ + UN8x4_MUL_UN8_ADD_UN8x4 (d, a2, s2); \ + *dst = CONVERT_8888_TO_ ## DST_FORMAT (d); \ + } \ + dst++; \ + } \ + else /* PIXMAN_OP_SRC */ \ + { \ + *dst++ = CONVERT_ ## SRC_FORMAT ## _TO_ ## DST_FORMAT (s1); \ + *dst++ = CONVERT_ ## SRC_FORMAT ## _TO_ ## DST_FORMAT (s2); \ + } \ + } \ + \ + if (w & 1) \ + { \ + x1 = vx >> 16; \ + s1 = src[x1]; \ + \ + if (PIXMAN_OP_ ## OP == PIXMAN_OP_OVER) \ + { \ + a1 = GET_ ## SRC_FORMAT ## _ALPHA(s1); \ + \ + if (a1 == 0xff) \ + { \ + *dst = CONVERT_ ## SRC_FORMAT ## _TO_ ## DST_FORMAT (s1); \ + } \ + else if (s1) \ + { \ + d = CONVERT_## DST_FORMAT ## _TO_8888 (*dst); \ + s1 = CONVERT_ ## SRC_FORMAT ## _TO_8888 (s1); \ + a1 ^= 0xff; \ + UN8x4_MUL_UN8_ADD_UN8x4 (d, a1, s1); \ + *dst = CONVERT_8888_TO_ ## DST_FORMAT (d); \ + } \ + dst++; \ + } \ + else /* PIXMAN_OP_SRC */ \ + { \ + *dst++ = CONVERT_ ## SRC_FORMAT ## _TO_ ## DST_FORMAT (s1); \ + } \ + } \ +} + +#define FAST_NEAREST_MAINLOOP_INT(scale_func_name, scanline_func, src_type_t, dst_type_t, \ + repeat_mode) \ +static void \ +fast_composite_scaled_nearest ## scale_func_name (pixman_implementation_t *imp, \ + pixman_op_t op, \ + pixman_image_t * src_image, \ + pixman_image_t * mask_image, \ + pixman_image_t * dst_image, \ + int32_t src_x, \ + int32_t src_y, \ + int32_t mask_x, \ + int32_t mask_y, \ + int32_t dst_x, \ + int32_t dst_y, \ + int32_t width, \ + int32_t height) \ +{ \ + dst_type_t *dst_line; \ + src_type_t *src_first_line; \ + int y; \ + pixman_fixed_t max_vx = max_vx; /* suppress uninitialized variable warning */ \ + pixman_fixed_t max_vy; \ + pixman_vector_t v; \ + pixman_fixed_t vx, vy; \ + pixman_fixed_t unit_x, unit_y; \ + int32_t left_pad, right_pad; \ + \ + src_type_t *src; \ + dst_type_t *dst; \ + int src_stride, dst_stride; \ + \ + PIXMAN_IMAGE_GET_LINE (dst_image, dst_x, dst_y, dst_type_t, dst_stride, dst_line, 1); \ + /* pass in 0 instead of src_x and src_y because src_x and src_y need to be \ + * transformed from destination space to source space */ \ + PIXMAN_IMAGE_GET_LINE (src_image, 0, 0, src_type_t, src_stride, src_first_line, 1); \ + \ + /* reference point is the center of the pixel */ \ + v.vector[0] = pixman_int_to_fixed (src_x) + pixman_fixed_1 / 2; \ + v.vector[1] = pixman_int_to_fixed (src_y) + pixman_fixed_1 / 2; \ + v.vector[2] = pixman_fixed_1; \ + \ + if (!pixman_transform_point_3d (src_image->common.transform, &v)) \ + return; \ + \ + unit_x = src_image->common.transform->matrix[0][0]; \ + unit_y = src_image->common.transform->matrix[1][1]; \ + \ + /* Round down to closest integer, ensuring that 0.5 rounds to 0, not 1 */ \ + v.vector[0] -= pixman_fixed_e; \ + v.vector[1] -= pixman_fixed_e; \ + \ + vx = v.vector[0]; \ + vy = v.vector[1]; \ + \ + if (PIXMAN_REPEAT_ ## repeat_mode == PIXMAN_REPEAT_NORMAL) \ + { \ + /* Clamp repeating positions inside the actual samples */ \ + max_vx = src_image->bits.width << 16; \ + max_vy = src_image->bits.height << 16; \ + \ + repeat (PIXMAN_REPEAT_NORMAL, &vx, max_vx); \ + repeat (PIXMAN_REPEAT_NORMAL, &vy, max_vy); \ + } \ + \ + if (PIXMAN_REPEAT_ ## repeat_mode == PIXMAN_REPEAT_PAD || \ + PIXMAN_REPEAT_ ## repeat_mode == PIXMAN_REPEAT_NONE) \ + { \ + pad_repeat_get_scanline_bounds (src_image->bits.width, vx, unit_x, \ + &width, &left_pad, &right_pad); \ + vx += left_pad * unit_x; \ + } \ + \ + while (--height >= 0) \ + { \ + dst = dst_line; \ + dst_line += dst_stride; \ + \ + y = vy >> 16; \ + vy += unit_y; \ + if (PIXMAN_REPEAT_ ## repeat_mode == PIXMAN_REPEAT_NORMAL) \ + repeat (PIXMAN_REPEAT_NORMAL, &vy, max_vy); \ + if (PIXMAN_REPEAT_ ## repeat_mode == PIXMAN_REPEAT_PAD) \ + { \ + repeat (PIXMAN_REPEAT_PAD, &y, src_image->bits.height); \ + src = src_first_line + src_stride * y; \ + if (left_pad > 0) \ + { \ + scanline_func (dst, src, left_pad, 0, 0, 0); \ + } \ + if (width > 0) \ + { \ + scanline_func (dst + left_pad, src, width, vx, unit_x, 0); \ + } \ + if (right_pad > 0) \ + { \ + scanline_func (dst + left_pad + width, src + src_image->bits.width - 1, \ + right_pad, 0, 0, 0); \ + } \ + } \ + else if (PIXMAN_REPEAT_ ## repeat_mode == PIXMAN_REPEAT_NONE) \ + { \ + static src_type_t zero = 0; \ + if (y < 0 || y >= src_image->bits.height) \ + { \ + scanline_func (dst, &zero, left_pad + width + right_pad, 0, 0, 0); \ + continue; \ + } \ + src = src_first_line + src_stride * y; \ + if (left_pad > 0) \ + { \ + scanline_func (dst, &zero, left_pad, 0, 0, 0); \ + } \ + if (width > 0) \ + { \ + scanline_func (dst + left_pad, src, width, vx, unit_x, 0); \ + } \ + if (right_pad > 0) \ + { \ + scanline_func (dst + left_pad + width, &zero, right_pad, 0, 0, 0); \ + } \ + } \ + else \ + { \ + src = src_first_line + src_stride * y; \ + scanline_func (dst, src, width, vx, unit_x, max_vx); \ + } \ + } \ +} + +/* A workaround for old sun studio, see: https://bugs.freedesktop.org/show_bug.cgi?id=32764 */ +#define FAST_NEAREST_MAINLOOP(scale_func_name, scanline_func, src_type_t, dst_type_t, \ + repeat_mode) \ + FAST_NEAREST_MAINLOOP_INT(_ ## scale_func_name, scanline_func, src_type_t, dst_type_t, \ + repeat_mode) \ + +#define FAST_NEAREST(scale_func_name, SRC_FORMAT, DST_FORMAT, \ + src_type_t, dst_type_t, OP, repeat_mode) \ + FAST_NEAREST_SCANLINE(scaled_nearest_scanline_ ## scale_func_name ## _ ## OP, \ + SRC_FORMAT, DST_FORMAT, src_type_t, dst_type_t, \ + OP, repeat_mode) \ + FAST_NEAREST_MAINLOOP_INT(_ ## scale_func_name ## _ ## OP, \ + scaled_nearest_scanline_ ## scale_func_name ## _ ## OP, \ + src_type_t, dst_type_t, repeat_mode) \ + \ + extern int no_such_variable + + +#define SCALED_NEAREST_FLAGS \ + (FAST_PATH_SCALE_TRANSFORM | \ + FAST_PATH_NO_ALPHA_MAP | \ + FAST_PATH_NEAREST_FILTER | \ + FAST_PATH_NO_ACCESSORS | \ + FAST_PATH_NARROW_FORMAT) + +#define SIMPLE_NEAREST_FAST_PATH_NORMAL(op,s,d,func) \ + { PIXMAN_OP_ ## op, \ + PIXMAN_ ## s, \ + (SCALED_NEAREST_FLAGS | \ + FAST_PATH_NORMAL_REPEAT | \ + FAST_PATH_X_UNIT_POSITIVE), \ + PIXMAN_null, 0, \ + PIXMAN_ ## d, FAST_PATH_STD_DEST_FLAGS, \ + fast_composite_scaled_nearest_ ## func ## _normal ## _ ## op, \ + } + +#define SIMPLE_NEAREST_FAST_PATH_PAD(op,s,d,func) \ + { PIXMAN_OP_ ## op, \ + PIXMAN_ ## s, \ + (SCALED_NEAREST_FLAGS | \ + FAST_PATH_PAD_REPEAT | \ + FAST_PATH_X_UNIT_POSITIVE), \ + PIXMAN_null, 0, \ + PIXMAN_ ## d, FAST_PATH_STD_DEST_FLAGS, \ + fast_composite_scaled_nearest_ ## func ## _pad ## _ ## op, \ + } + +#define SIMPLE_NEAREST_FAST_PATH_NONE(op,s,d,func) \ + { PIXMAN_OP_ ## op, \ + PIXMAN_ ## s, \ + (SCALED_NEAREST_FLAGS | \ + FAST_PATH_NONE_REPEAT | \ + FAST_PATH_X_UNIT_POSITIVE), \ + PIXMAN_null, 0, \ + PIXMAN_ ## d, FAST_PATH_STD_DEST_FLAGS, \ + fast_composite_scaled_nearest_ ## func ## _none ## _ ## op, \ + } + +#define SIMPLE_NEAREST_FAST_PATH_COVER(op,s,d,func) \ + { PIXMAN_OP_ ## op, \ + PIXMAN_ ## s, \ + SCALED_NEAREST_FLAGS | FAST_PATH_SAMPLES_COVER_CLIP, \ + PIXMAN_null, 0, \ + PIXMAN_ ## d, FAST_PATH_STD_DEST_FLAGS, \ + fast_composite_scaled_nearest_ ## func ## _cover ## _ ## op, \ + } + +/* Prefer the use of 'cover' variant, because it is faster */ +#define SIMPLE_NEAREST_FAST_PATH(op,s,d,func) \ + SIMPLE_NEAREST_FAST_PATH_COVER (op,s,d,func), \ + SIMPLE_NEAREST_FAST_PATH_NONE (op,s,d,func), \ + SIMPLE_NEAREST_FAST_PATH_PAD (op,s,d,func), \ + SIMPLE_NEAREST_FAST_PATH_NORMAL (op,s,d,func) + +#endif diff --git a/programs/develop/libraries/pixman/pixman-general.c b/programs/develop/libraries/pixman/pixman-general.c new file mode 100644 index 0000000000..8130f166ef --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-general.c @@ -0,0 +1,315 @@ +/* + * Copyright © 2009 Red Hat, Inc. + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007 Red Hat, Inc. + * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc. + * 2005 Lars Knoll & Zack Rusin, Trolltech + * 2008 Aaron Plattner, NVIDIA Corporation + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Red Hat not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. Red Hat makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ +#ifdef HAVE_CONFIG_H +#include +#endif +#include +#include +#include +#include +#include +#include +#include +#include "pixman-private.h" +#include "pixman-combine32.h" +#include "pixman-private.h" + +#define SCANLINE_BUFFER_LENGTH 8192 + +static void +general_composite_rect (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src, + pixman_image_t * mask, + pixman_image_t * dest, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint64_t stack_scanline_buffer[(SCANLINE_BUFFER_LENGTH * 3 + 7) / 8]; + uint8_t *scanline_buffer = (uint8_t *) stack_scanline_buffer; + uint8_t *src_buffer, *mask_buffer, *dest_buffer; + fetch_scanline_t fetch_src = NULL, fetch_mask = NULL, fetch_dest = NULL; + pixman_combine_32_func_t compose; + store_scanline_t store; + source_image_class_t src_class, mask_class; + pixman_bool_t component_alpha; + uint32_t *bits; + int32_t stride; + int narrow, Bpp; + int i; + + narrow = + (src->common.flags & FAST_PATH_NARROW_FORMAT) && + (!mask || mask->common.flags & FAST_PATH_NARROW_FORMAT) && + (dest->common.flags & FAST_PATH_NARROW_FORMAT); + Bpp = narrow ? 4 : 8; + + if (width * Bpp > SCANLINE_BUFFER_LENGTH) + { + scanline_buffer = pixman_malloc_abc (width, 3, Bpp); + + if (!scanline_buffer) + return; + } + + src_buffer = scanline_buffer; + mask_buffer = src_buffer + width * Bpp; + dest_buffer = mask_buffer + width * Bpp; + + src_class = _pixman_image_classify (src, + src_x, src_y, + width, height); + + mask_class = SOURCE_IMAGE_CLASS_UNKNOWN; + + if (mask) + { + mask_class = _pixman_image_classify (mask, + src_x, src_y, + width, height); + } + + if (op == PIXMAN_OP_CLEAR) + fetch_src = NULL; + else if (narrow) + fetch_src = _pixman_image_get_scanline_32; + else + fetch_src = _pixman_image_get_scanline_64; + + if (!mask || op == PIXMAN_OP_CLEAR) + fetch_mask = NULL; + else if (narrow) + fetch_mask = _pixman_image_get_scanline_32; + else + fetch_mask = _pixman_image_get_scanline_64; + + if (op == PIXMAN_OP_CLEAR || op == PIXMAN_OP_SRC) + fetch_dest = NULL; + else if (narrow) + fetch_dest = _pixman_image_get_scanline_32; + else + fetch_dest = _pixman_image_get_scanline_64; + + if (narrow) + store = _pixman_image_store_scanline_32; + else + store = _pixman_image_store_scanline_64; + + /* Skip the store step and composite directly into the + * destination if the output format of the compose func matches + * the destination format. + * + * If the destination format is a8r8g8b8 then we can always do + * this. If it is x8r8g8b8, then we can only do it if the + * operator doesn't make use of destination alpha. + */ + if ((dest->bits.format == PIXMAN_a8r8g8b8) || + (dest->bits.format == PIXMAN_x8r8g8b8 && + (op == PIXMAN_OP_OVER || + op == PIXMAN_OP_ADD || + op == PIXMAN_OP_SRC || + op == PIXMAN_OP_CLEAR || + op == PIXMAN_OP_IN_REVERSE || + op == PIXMAN_OP_OUT_REVERSE || + op == PIXMAN_OP_DST))) + { + if (narrow && + !dest->common.alpha_map && + !dest->bits.write_func) + { + store = NULL; + } + } + + if (!store) + { + bits = dest->bits.bits; + stride = dest->bits.rowstride; + } + else + { + bits = NULL; + stride = 0; + } + + component_alpha = + fetch_src && + fetch_mask && + mask && + mask->common.type == BITS && + mask->common.component_alpha && + PIXMAN_FORMAT_RGB (mask->bits.format); + + if (narrow) + { + if (component_alpha) + compose = _pixman_implementation_combine_32_ca; + else + compose = _pixman_implementation_combine_32; + } + else + { + if (component_alpha) + compose = (pixman_combine_32_func_t)_pixman_implementation_combine_64_ca; + else + compose = (pixman_combine_32_func_t)_pixman_implementation_combine_64; + } + + if (!compose) + return; + + if (!fetch_mask) + mask_buffer = NULL; + + for (i = 0; i < height; ++i) + { + /* fill first half of scanline with source */ + if (fetch_src) + { + if (fetch_mask) + { + /* fetch mask before source so that fetching of + source can be optimized */ + fetch_mask (mask, mask_x, mask_y + i, + width, (void *)mask_buffer, 0); + + if (mask_class == SOURCE_IMAGE_CLASS_HORIZONTAL) + fetch_mask = NULL; + } + + if (src_class == SOURCE_IMAGE_CLASS_HORIZONTAL) + { + fetch_src (src, src_x, src_y + i, + width, (void *)src_buffer, 0); + fetch_src = NULL; + } + else + { + fetch_src (src, src_x, src_y + i, + width, (void *)src_buffer, (void *)mask_buffer); + } + } + else if (fetch_mask) + { + fetch_mask (mask, mask_x, mask_y + i, + width, (void *)mask_buffer, 0); + } + + if (store) + { + /* fill dest into second half of scanline */ + if (fetch_dest) + { + fetch_dest (dest, dest_x, dest_y + i, + width, (void *)dest_buffer, 0); + } + + /* blend */ + compose (imp->toplevel, op, + (void *)dest_buffer, + (void *)src_buffer, + (void *)mask_buffer, + width); + + /* write back */ + store (&(dest->bits), dest_x, dest_y + i, width, + (void *)dest_buffer); + } + else + { + /* blend */ + compose (imp->toplevel, op, + bits + (dest_y + i) * stride + dest_x, + (void *)src_buffer, (void *)mask_buffer, width); + } + } + + if (scanline_buffer != (uint8_t *) stack_scanline_buffer) + free (scanline_buffer); +} + +static const pixman_fast_path_t general_fast_path[] = +{ + { PIXMAN_OP_any, PIXMAN_any, 0, PIXMAN_any, 0, PIXMAN_any, 0, general_composite_rect }, + { PIXMAN_OP_NONE } +}; + +static pixman_bool_t +general_blt (pixman_implementation_t *imp, + uint32_t * src_bits, + uint32_t * dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height) +{ + /* We can't blit unless we have sse2 or mmx */ + + return FALSE; +} + +static pixman_bool_t +general_fill (pixman_implementation_t *imp, + uint32_t * bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + return FALSE; +} + +pixman_implementation_t * +_pixman_implementation_create_general (void) +{ + pixman_implementation_t *imp = _pixman_implementation_create (NULL, general_fast_path); + + _pixman_setup_combiner_functions_32 (imp); + _pixman_setup_combiner_functions_64 (imp); + + imp->blt = general_blt; + imp->fill = general_fill; + + return imp; +} + diff --git a/programs/develop/libraries/pixman/pixman-gradient-walker.c b/programs/develop/libraries/pixman/pixman-gradient-walker.c new file mode 100644 index 0000000000..dd666b4120 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-gradient-walker.c @@ -0,0 +1,254 @@ +/* + * + * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc. + * 2005 Lars Knoll & Zack Rusin, Trolltech + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include "pixman-private.h" + +void +_pixman_gradient_walker_init (pixman_gradient_walker_t *walker, + gradient_t * gradient, + unsigned int spread) +{ + walker->num_stops = gradient->n_stops; + walker->stops = gradient->stops; + walker->left_x = 0; + walker->right_x = 0x10000; + walker->stepper = 0; + walker->left_ag = 0; + walker->left_rb = 0; + walker->right_ag = 0; + walker->right_rb = 0; + walker->spread = spread; + + walker->need_reset = TRUE; +} + +void +_pixman_gradient_walker_reset (pixman_gradient_walker_t *walker, + pixman_fixed_32_32_t pos) +{ + int32_t x, left_x, right_x; + pixman_color_t *left_c, *right_c; + int n, count = walker->num_stops; + pixman_gradient_stop_t * stops = walker->stops; + + static const pixman_color_t transparent_black = { 0, 0, 0, 0 }; + + switch (walker->spread) + { + case PIXMAN_REPEAT_NORMAL: + x = (int32_t)pos & 0xFFFF; + for (n = 0; n < count; n++) + if (x < stops[n].x) + break; + if (n == 0) + { + left_x = stops[count - 1].x - 0x10000; + left_c = &stops[count - 1].color; + } + else + { + left_x = stops[n - 1].x; + left_c = &stops[n - 1].color; + } + + if (n == count) + { + right_x = stops[0].x + 0x10000; + right_c = &stops[0].color; + } + else + { + right_x = stops[n].x; + right_c = &stops[n].color; + } + left_x += (pos - x); + right_x += (pos - x); + break; + + case PIXMAN_REPEAT_PAD: + for (n = 0; n < count; n++) + if (pos < stops[n].x) + break; + + if (n == 0) + { + left_x = INT32_MIN; + left_c = &stops[0].color; + } + else + { + left_x = stops[n - 1].x; + left_c = &stops[n - 1].color; + } + + if (n == count) + { + right_x = INT32_MAX; + right_c = &stops[n - 1].color; + } + else + { + right_x = stops[n].x; + right_c = &stops[n].color; + } + break; + + case PIXMAN_REPEAT_REFLECT: + x = (int32_t)pos & 0xFFFF; + if ((int32_t)pos & 0x10000) + x = 0x10000 - x; + for (n = 0; n < count; n++) + if (x < stops[n].x) + break; + + if (n == 0) + { + left_x = -stops[0].x; + left_c = &stops[0].color; + } + else + { + left_x = stops[n - 1].x; + left_c = &stops[n - 1].color; + } + + if (n == count) + { + right_x = 0x20000 - stops[n - 1].x; + right_c = &stops[n - 1].color; + } + else + { + right_x = stops[n].x; + right_c = &stops[n].color; + } + + if ((int32_t)pos & 0x10000) + { + pixman_color_t *tmp_c; + int32_t tmp_x; + + tmp_x = 0x10000 - right_x; + right_x = 0x10000 - left_x; + left_x = tmp_x; + + tmp_c = right_c; + right_c = left_c; + left_c = tmp_c; + + x = 0x10000 - x; + } + left_x += (pos - x); + right_x += (pos - x); + break; + + default: /* REPEAT_NONE */ + for (n = 0; n < count; n++) + if (pos < stops[n].x) + break; + + if (n == 0) + { + left_x = INT32_MIN; + right_x = stops[0].x; + left_c = right_c = (pixman_color_t*) &transparent_black; + } + else if (n == count) + { + left_x = stops[n - 1].x; + right_x = INT32_MAX; + left_c = right_c = (pixman_color_t*) &transparent_black; + } + else + { + left_x = stops[n - 1].x; + right_x = stops[n].x; + left_c = &stops[n - 1].color; + right_c = &stops[n].color; + } + } + + walker->left_x = left_x; + walker->right_x = right_x; + walker->left_ag = ((left_c->alpha >> 8) << 16) | (left_c->green >> 8); + walker->left_rb = ((left_c->red & 0xff00) << 8) | (left_c->blue >> 8); + walker->right_ag = ((right_c->alpha >> 8) << 16) | (right_c->green >> 8); + walker->right_rb = ((right_c->red & 0xff00) << 8) | (right_c->blue >> 8); + + if (walker->left_x == walker->right_x || + ( walker->left_ag == walker->right_ag && + walker->left_rb == walker->right_rb ) ) + { + walker->stepper = 0; + } + else + { + int32_t width = right_x - left_x; + walker->stepper = ((1 << 24) + width / 2) / width; + } + + walker->need_reset = FALSE; +} + +#define PIXMAN_GRADIENT_WALKER_NEED_RESET(w, x) \ + ( (w)->need_reset || (x) < (w)->left_x || (x) >= (w)->right_x) + + +/* the following assumes that PIXMAN_GRADIENT_WALKER_NEED_RESET(w,x) is FALSE */ +uint32_t +_pixman_gradient_walker_pixel (pixman_gradient_walker_t *walker, + pixman_fixed_32_32_t x) +{ + int dist, idist; + uint32_t t1, t2, a, color; + + if (PIXMAN_GRADIENT_WALKER_NEED_RESET (walker, x)) + _pixman_gradient_walker_reset (walker, x); + + dist = ((int)(x - walker->left_x) * walker->stepper) >> 16; + idist = 256 - dist; + + /* combined INTERPOLATE and premultiply */ + t1 = walker->left_rb * idist + walker->right_rb * dist; + t1 = (t1 >> 8) & 0xff00ff; + + t2 = walker->left_ag * idist + walker->right_ag * dist; + t2 &= 0xff00ff00; + + color = t2 & 0xff000000; + a = t2 >> 24; + + t1 = t1 * a + 0x800080; + t1 = (t1 + ((t1 >> 8) & 0xff00ff)) >> 8; + + t2 = (t2 >> 8) * a + 0x800080; + t2 = (t2 + ((t2 >> 8) & 0xff00ff)); + + return (color | (t1 & 0xff00ff) | (t2 & 0xff00)); +} + diff --git a/programs/develop/libraries/pixman/pixman-image.c b/programs/develop/libraries/pixman/pixman-image.c new file mode 100644 index 0000000000..c646471930 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-image.c @@ -0,0 +1,846 @@ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of SuSE not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. SuSE makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * SuSE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL SuSE + * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif + +#include +#include +#include +#include + +#include "pixman-private.h" +#include "pixman-combine32.h" + +pixman_bool_t +_pixman_init_gradient (gradient_t * gradient, + const pixman_gradient_stop_t *stops, + int n_stops) +{ + return_val_if_fail (n_stops > 0, FALSE); + + gradient->stops = pixman_malloc_ab (n_stops, sizeof (pixman_gradient_stop_t)); + if (!gradient->stops) + return FALSE; + + memcpy (gradient->stops, stops, n_stops * sizeof (pixman_gradient_stop_t)); + + gradient->n_stops = n_stops; + + gradient->stop_range = 0xffff; + + return TRUE; +} + +/* + * By default, just evaluate the image at 32bpp and expand. Individual image + * types can plug in a better scanline getter if they want to. For example + * we could produce smoother gradients by evaluating them at higher color + * depth, but that's a project for the future. + */ +void +_pixman_image_get_scanline_generic_64 (pixman_image_t * image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t * mask) +{ + uint32_t *mask8 = NULL; + + /* Contract the mask image, if one exists, so that the 32-bit fetch + * function can use it. + */ + if (mask) + { + mask8 = pixman_malloc_ab (width, sizeof(uint32_t)); + if (!mask8) + return; + + pixman_contract (mask8, (uint64_t *)mask, width); + } + + /* Fetch the source image into the first half of buffer. */ + _pixman_image_get_scanline_32 (image, x, y, width, (uint32_t*)buffer, mask8); + + /* Expand from 32bpp to 64bpp in place. */ + pixman_expand ((uint64_t *)buffer, buffer, PIXMAN_a8r8g8b8, width); + + free (mask8); +} + +pixman_image_t * +_pixman_image_allocate (void) +{ + pixman_image_t *image = malloc (sizeof (pixman_image_t)); + + if (image) + { + image_common_t *common = &image->common; + + pixman_region32_init (&common->clip_region); + + common->alpha_count = 0; + common->have_clip_region = FALSE; + common->clip_sources = FALSE; + common->transform = NULL; + common->repeat = PIXMAN_REPEAT_NONE; + common->filter = PIXMAN_FILTER_NEAREST; + common->filter_params = NULL; + common->n_filter_params = 0; + common->alpha_map = NULL; + common->component_alpha = FALSE; + common->ref_count = 1; + common->classify = NULL; + common->client_clip = FALSE; + common->destroy_func = NULL; + common->destroy_data = NULL; + common->dirty = TRUE; + } + + return image; +} + +source_image_class_t +_pixman_image_classify (pixman_image_t *image, + int x, + int y, + int width, + int height) +{ + if (image->common.classify) + return image->common.classify (image, x, y, width, height); + else + return SOURCE_IMAGE_CLASS_UNKNOWN; +} + +void +_pixman_image_get_scanline_32 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + image->common.get_scanline_32 (image, x, y, width, buffer, mask); +} + +/* Even thought the type of buffer is uint32_t *, the function actually expects + * a uint64_t *buffer. + */ +void +_pixman_image_get_scanline_64 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *unused) +{ + image->common.get_scanline_64 (image, x, y, width, buffer, unused); +} + +static void +image_property_changed (pixman_image_t *image) +{ + image->common.dirty = TRUE; +} + +/* Ref Counting */ +PIXMAN_EXPORT pixman_image_t * +pixman_image_ref (pixman_image_t *image) +{ + image->common.ref_count++; + + return image; +} + +/* returns TRUE when the image is freed */ +PIXMAN_EXPORT pixman_bool_t +pixman_image_unref (pixman_image_t *image) +{ + image_common_t *common = (image_common_t *)image; + + common->ref_count--; + + if (common->ref_count == 0) + { + if (image->common.destroy_func) + image->common.destroy_func (image, image->common.destroy_data); + + pixman_region32_fini (&common->clip_region); + + if (common->transform) + free (common->transform); + + if (common->filter_params) + free (common->filter_params); + + if (common->alpha_map) + pixman_image_unref ((pixman_image_t *)common->alpha_map); + + if (image->type == LINEAR || + image->type == RADIAL || + image->type == CONICAL) + { + if (image->gradient.stops) + free (image->gradient.stops); + } + + if (image->type == BITS && image->bits.free_me) + free (image->bits.free_me); + + free (image); + + return TRUE; + } + + return FALSE; +} + +PIXMAN_EXPORT void +pixman_image_set_destroy_function (pixman_image_t * image, + pixman_image_destroy_func_t func, + void * data) +{ + image->common.destroy_func = func; + image->common.destroy_data = data; +} + +PIXMAN_EXPORT void * +pixman_image_get_destroy_data (pixman_image_t *image) +{ + return image->common.destroy_data; +} + +void +_pixman_image_reset_clip_region (pixman_image_t *image) +{ + image->common.have_clip_region = FALSE; +} + +static pixman_bool_t out_of_bounds_workaround = TRUE; + +/* Old X servers rely on out-of-bounds accesses when they are asked + * to composite with a window as the source. They create a pixman image + * pointing to some bogus position in memory, but then they set a clip + * region to the position where the actual bits are. + * + * Due to a bug in old versions of pixman, where it would not clip + * against the image bounds when a clip region was set, this would + * actually work. So by default we allow certain out-of-bound access + * to happen unless explicitly disabled. + * + * Fixed X servers should call this function to disable the workaround. + */ +PIXMAN_EXPORT void +pixman_disable_out_of_bounds_workaround (void) +{ + out_of_bounds_workaround = FALSE; +} + +static pixman_bool_t +source_image_needs_out_of_bounds_workaround (bits_image_t *image) +{ + if (image->common.clip_sources && + image->common.repeat == PIXMAN_REPEAT_NONE && + image->common.have_clip_region && + out_of_bounds_workaround) + { + if (!image->common.client_clip) + { + /* There is no client clip, so if the clip region extends beyond the + * drawable geometry, it must be because the X server generated the + * bogus clip region. + */ + const pixman_box32_t *extents = + pixman_region32_extents (&image->common.clip_region); + + if (extents->x1 >= 0 && extents->x2 <= image->width && + extents->y1 >= 0 && extents->y2 <= image->height) + { + return FALSE; + } + } + + return TRUE; + } + + return FALSE; +} + +static void +compute_image_info (pixman_image_t *image) +{ + pixman_format_code_t code; + uint32_t flags = 0; + + /* Transform */ + if (!image->common.transform) + { + flags |= (FAST_PATH_ID_TRANSFORM | + FAST_PATH_X_UNIT_POSITIVE | + FAST_PATH_Y_UNIT_ZERO | + FAST_PATH_AFFINE_TRANSFORM); + } + else + { + flags |= FAST_PATH_HAS_TRANSFORM; + + if (image->common.transform->matrix[2][0] == 0 && + image->common.transform->matrix[2][1] == 0 && + image->common.transform->matrix[2][2] == pixman_fixed_1) + { + flags |= FAST_PATH_AFFINE_TRANSFORM; + + if (image->common.transform->matrix[0][1] == 0 && + image->common.transform->matrix[1][0] == 0) + { + flags |= FAST_PATH_SCALE_TRANSFORM; + } + } + + if (image->common.transform->matrix[0][0] > 0) + flags |= FAST_PATH_X_UNIT_POSITIVE; + + if (image->common.transform->matrix[1][0] == 0) + flags |= FAST_PATH_Y_UNIT_ZERO; + } + + /* Filter */ + switch (image->common.filter) + { + case PIXMAN_FILTER_NEAREST: + case PIXMAN_FILTER_FAST: + flags |= (FAST_PATH_NEAREST_FILTER | FAST_PATH_NO_CONVOLUTION_FILTER); + break; + + case PIXMAN_FILTER_BILINEAR: + case PIXMAN_FILTER_GOOD: + case PIXMAN_FILTER_BEST: + flags |= (FAST_PATH_BILINEAR_FILTER | FAST_PATH_NO_CONVOLUTION_FILTER); + break; + + case PIXMAN_FILTER_CONVOLUTION: + break; + + default: + flags |= FAST_PATH_NO_CONVOLUTION_FILTER; + break; + } + + /* Repeat mode */ + switch (image->common.repeat) + { + case PIXMAN_REPEAT_NONE: + flags |= + FAST_PATH_NO_REFLECT_REPEAT | + FAST_PATH_NO_PAD_REPEAT | + FAST_PATH_NO_NORMAL_REPEAT; + break; + + case PIXMAN_REPEAT_REFLECT: + flags |= + FAST_PATH_NO_PAD_REPEAT | + FAST_PATH_NO_NONE_REPEAT | + FAST_PATH_NO_NORMAL_REPEAT; + break; + + case PIXMAN_REPEAT_PAD: + flags |= + FAST_PATH_NO_REFLECT_REPEAT | + FAST_PATH_NO_NONE_REPEAT | + FAST_PATH_NO_NORMAL_REPEAT; + break; + + default: + flags |= + FAST_PATH_NO_REFLECT_REPEAT | + FAST_PATH_NO_PAD_REPEAT | + FAST_PATH_NO_NONE_REPEAT; + break; + } + + /* Component alpha */ + if (image->common.component_alpha) + flags |= FAST_PATH_COMPONENT_ALPHA; + else + flags |= FAST_PATH_UNIFIED_ALPHA; + + flags |= (FAST_PATH_NO_ACCESSORS | FAST_PATH_NARROW_FORMAT); + + /* Type specific checks */ + switch (image->type) + { + case SOLID: + code = PIXMAN_solid; + + if (image->solid.color.alpha == 0xffff) + flags |= FAST_PATH_IS_OPAQUE; + break; + + case BITS: + if (image->bits.width == 1 && + image->bits.height == 1 && + image->common.repeat != PIXMAN_REPEAT_NONE) + { + code = PIXMAN_solid; + } + else + { + code = image->bits.format; + } + + if (!PIXMAN_FORMAT_A (image->bits.format) && + PIXMAN_FORMAT_TYPE (image->bits.format) != PIXMAN_TYPE_GRAY && + PIXMAN_FORMAT_TYPE (image->bits.format) != PIXMAN_TYPE_COLOR) + { + flags |= FAST_PATH_SAMPLES_OPAQUE; + + if (image->common.repeat != PIXMAN_REPEAT_NONE) + flags |= FAST_PATH_IS_OPAQUE; + } + + if (source_image_needs_out_of_bounds_workaround (&image->bits)) + flags |= FAST_PATH_NEEDS_WORKAROUND; + + if (image->bits.read_func || image->bits.write_func) + flags &= ~FAST_PATH_NO_ACCESSORS; + + if (PIXMAN_FORMAT_IS_WIDE (image->bits.format)) + flags &= ~FAST_PATH_NARROW_FORMAT; + break; + + case RADIAL: + code = PIXMAN_unknown; + + /* + * As explained in pixman-radial-gradient.c, every point of + * the plane has a valid associated radius (and thus will be + * colored) if and only if a is negative (i.e. one of the two + * circles contains the other one). + */ + + if (image->radial.a >= 0) + break; + + /* Fall through */ + + case LINEAR: + code = PIXMAN_unknown; + + if (image->common.repeat != PIXMAN_REPEAT_NONE) + { + int i; + + flags |= FAST_PATH_IS_OPAQUE; + for (i = 0; i < image->gradient.n_stops; ++i) + { + if (image->gradient.stops[i].color.alpha != 0xffff) + { + flags &= ~FAST_PATH_IS_OPAQUE; + break; + } + } + } + break; + + default: + code = PIXMAN_unknown; + break; + } + + /* Alpha map */ + if (!image->common.alpha_map) + { + flags |= FAST_PATH_NO_ALPHA_MAP; + } + else + { + if (PIXMAN_FORMAT_IS_WIDE (image->common.alpha_map->format)) + flags &= ~FAST_PATH_NARROW_FORMAT; + } + + /* Both alpha maps and convolution filters can introduce + * non-opaqueness in otherwise opaque images. Also + * an image with component alpha turned on is only opaque + * if all channels are opaque, so we simply turn it off + * unconditionally for those images. + */ + if (image->common.alpha_map || + image->common.filter == PIXMAN_FILTER_CONVOLUTION || + image->common.component_alpha) + { + flags &= ~(FAST_PATH_IS_OPAQUE | FAST_PATH_SAMPLES_OPAQUE); + } + + image->common.flags = flags; + image->common.extended_format_code = code; +} + +void +_pixman_image_validate (pixman_image_t *image) +{ + if (image->common.dirty) + { + compute_image_info (image); + + /* It is important that property_changed is + * called *after* compute_image_info() because + * property_changed() can make use of the flags + * to set up accessors etc. + */ + image->common.property_changed (image); + + image->common.dirty = FALSE; + } + + if (image->common.alpha_map) + _pixman_image_validate ((pixman_image_t *)image->common.alpha_map); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_image_set_clip_region32 (pixman_image_t * image, + pixman_region32_t *region) +{ + image_common_t *common = (image_common_t *)image; + pixman_bool_t result; + + if (region) + { + if ((result = pixman_region32_copy (&common->clip_region, region))) + image->common.have_clip_region = TRUE; + } + else + { + _pixman_image_reset_clip_region (image); + + result = TRUE; + } + + image_property_changed (image); + + return result; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_image_set_clip_region (pixman_image_t * image, + pixman_region16_t *region) +{ + image_common_t *common = (image_common_t *)image; + pixman_bool_t result; + + if (region) + { + if ((result = pixman_region32_copy_from_region16 (&common->clip_region, region))) + image->common.have_clip_region = TRUE; + } + else + { + _pixman_image_reset_clip_region (image); + + result = TRUE; + } + + image_property_changed (image); + + return result; +} + +PIXMAN_EXPORT void +pixman_image_set_has_client_clip (pixman_image_t *image, + pixman_bool_t client_clip) +{ + image->common.client_clip = client_clip; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_image_set_transform (pixman_image_t * image, + const pixman_transform_t *transform) +{ + static const pixman_transform_t id = + { + { { pixman_fixed_1, 0, 0 }, + { 0, pixman_fixed_1, 0 }, + { 0, 0, pixman_fixed_1 } } + }; + + image_common_t *common = (image_common_t *)image; + pixman_bool_t result; + + if (common->transform == transform) + return TRUE; + + if (memcmp (&id, transform, sizeof (pixman_transform_t)) == 0) + { + free (common->transform); + common->transform = NULL; + result = TRUE; + + goto out; + } + + if (common->transform == NULL) + common->transform = malloc (sizeof (pixman_transform_t)); + + if (common->transform == NULL) + { + result = FALSE; + + goto out; + } + + memcpy (common->transform, transform, sizeof(pixman_transform_t)); + + result = TRUE; + +out: + image_property_changed (image); + + return result; +} + +PIXMAN_EXPORT void +pixman_image_set_repeat (pixman_image_t *image, + pixman_repeat_t repeat) +{ + image->common.repeat = repeat; + + image_property_changed (image); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_image_set_filter (pixman_image_t * image, + pixman_filter_t filter, + const pixman_fixed_t *params, + int n_params) +{ + image_common_t *common = (image_common_t *)image; + pixman_fixed_t *new_params; + + if (params == common->filter_params && filter == common->filter) + return TRUE; + + new_params = NULL; + if (params) + { + new_params = pixman_malloc_ab (n_params, sizeof (pixman_fixed_t)); + if (!new_params) + return FALSE; + + memcpy (new_params, + params, n_params * sizeof (pixman_fixed_t)); + } + + common->filter = filter; + + if (common->filter_params) + free (common->filter_params); + + common->filter_params = new_params; + common->n_filter_params = n_params; + + image_property_changed (image); + return TRUE; +} + +PIXMAN_EXPORT void +pixman_image_set_source_clipping (pixman_image_t *image, + pixman_bool_t clip_sources) +{ + image->common.clip_sources = clip_sources; + + image_property_changed (image); +} + +/* Unlike all the other property setters, this function does not + * copy the content of indexed. Doing this copying is simply + * way, way too expensive. + */ +PIXMAN_EXPORT void +pixman_image_set_indexed (pixman_image_t * image, + const pixman_indexed_t *indexed) +{ + bits_image_t *bits = (bits_image_t *)image; + + bits->indexed = indexed; + + image_property_changed (image); +} + +PIXMAN_EXPORT void +pixman_image_set_alpha_map (pixman_image_t *image, + pixman_image_t *alpha_map, + int16_t x, + int16_t y) +{ + image_common_t *common = (image_common_t *)image; + + return_if_fail (!alpha_map || alpha_map->type == BITS); + + if (alpha_map && common->alpha_count > 0) + { + /* If this image is being used as an alpha map itself, + * then you can't give it an alpha map of its own. + */ + return; + } + + if (alpha_map && alpha_map->common.alpha_map) + { + /* If the image has an alpha map of its own, + * then it can't be used as an alpha map itself + */ + return; + } + + if (common->alpha_map != (bits_image_t *)alpha_map) + { + if (common->alpha_map) + { + common->alpha_map->common.alpha_count--; + + pixman_image_unref ((pixman_image_t *)common->alpha_map); + } + + if (alpha_map) + { + common->alpha_map = (bits_image_t *)pixman_image_ref (alpha_map); + + common->alpha_map->common.alpha_count++; + } + else + { + common->alpha_map = NULL; + } + } + + common->alpha_origin_x = x; + common->alpha_origin_y = y; + + image_property_changed (image); +} + +PIXMAN_EXPORT void +pixman_image_set_component_alpha (pixman_image_t *image, + pixman_bool_t component_alpha) +{ + image->common.component_alpha = component_alpha; + + image_property_changed (image); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_image_get_component_alpha (pixman_image_t *image) +{ + return image->common.component_alpha; +} + +PIXMAN_EXPORT void +pixman_image_set_accessors (pixman_image_t * image, + pixman_read_memory_func_t read_func, + pixman_write_memory_func_t write_func) +{ + return_if_fail (image != NULL); + + if (image->type == BITS) + { + image->bits.read_func = read_func; + image->bits.write_func = write_func; + + image_property_changed (image); + } +} + +PIXMAN_EXPORT uint32_t * +pixman_image_get_data (pixman_image_t *image) +{ + if (image->type == BITS) + return image->bits.bits; + + return NULL; +} + +PIXMAN_EXPORT int +pixman_image_get_width (pixman_image_t *image) +{ + if (image->type == BITS) + return image->bits.width; + + return 0; +} + +PIXMAN_EXPORT int +pixman_image_get_height (pixman_image_t *image) +{ + if (image->type == BITS) + return image->bits.height; + + return 0; +} + +PIXMAN_EXPORT int +pixman_image_get_stride (pixman_image_t *image) +{ + if (image->type == BITS) + return image->bits.rowstride * (int) sizeof (uint32_t); + + return 0; +} + +PIXMAN_EXPORT int +pixman_image_get_depth (pixman_image_t *image) +{ + if (image->type == BITS) + return PIXMAN_FORMAT_DEPTH (image->bits.format); + + return 0; +} + +PIXMAN_EXPORT pixman_format_code_t +pixman_image_get_format (pixman_image_t *image) +{ + if (image->type == BITS) + return image->bits.format; + + return 0; +} + +uint32_t +_pixman_image_get_solid (pixman_image_t * image, + pixman_format_code_t format) +{ + uint32_t result; + + _pixman_image_get_scanline_32 (image, 0, 0, 1, &result, NULL); + + /* If necessary, convert RGB <--> BGR. */ + if (PIXMAN_FORMAT_TYPE (format) != PIXMAN_TYPE_ARGB) + { + result = (((result & 0xff000000) >> 0) | + ((result & 0x00ff0000) >> 16) | + ((result & 0x0000ff00) >> 0) | + ((result & 0x000000ff) << 16)); + } + + return result; +} diff --git a/programs/develop/libraries/pixman/pixman-implementation.c b/programs/develop/libraries/pixman/pixman-implementation.c new file mode 100644 index 0000000000..bc3749ef59 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-implementation.c @@ -0,0 +1,227 @@ +/* + * Copyright © 2009 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Red Hat not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. Red Hat makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include +#include "pixman-private.h" + +static void +delegate_combine_32 (pixman_implementation_t * imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + _pixman_implementation_combine_32 (imp->delegate, + op, dest, src, mask, width); +} + +static void +delegate_combine_64 (pixman_implementation_t * imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + _pixman_implementation_combine_64 (imp->delegate, + op, dest, src, mask, width); +} + +static void +delegate_combine_32_ca (pixman_implementation_t * imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + _pixman_implementation_combine_32_ca (imp->delegate, + op, dest, src, mask, width); +} + +static void +delegate_combine_64_ca (pixman_implementation_t * imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + _pixman_implementation_combine_64_ca (imp->delegate, + op, dest, src, mask, width); +} + +static pixman_bool_t +delegate_blt (pixman_implementation_t * imp, + uint32_t * src_bits, + uint32_t * dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height) +{ + return _pixman_implementation_blt ( + imp->delegate, src_bits, dst_bits, src_stride, dst_stride, + src_bpp, dst_bpp, src_x, src_y, dst_x, dst_y, + width, height); +} + +static pixman_bool_t +delegate_fill (pixman_implementation_t *imp, + uint32_t * bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + return _pixman_implementation_fill ( + imp->delegate, bits, stride, bpp, x, y, width, height, xor); +} + +pixman_implementation_t * +_pixman_implementation_create (pixman_implementation_t *delegate, + const pixman_fast_path_t *fast_paths) +{ + pixman_implementation_t *imp = malloc (sizeof (pixman_implementation_t)); + pixman_implementation_t *d; + int i; + + if (!imp) + return NULL; + + assert (fast_paths); + + /* Make sure the whole delegate chain has the right toplevel */ + imp->delegate = delegate; + for (d = imp; d != NULL; d = d->delegate) + d->toplevel = imp; + + /* Fill out function pointers with ones that just delegate + */ + imp->blt = delegate_blt; + imp->fill = delegate_fill; + + for (i = 0; i < PIXMAN_N_OPERATORS; ++i) + { + imp->combine_32[i] = delegate_combine_32; + imp->combine_64[i] = delegate_combine_64; + imp->combine_32_ca[i] = delegate_combine_32_ca; + imp->combine_64_ca[i] = delegate_combine_64_ca; + } + + imp->fast_paths = fast_paths; + + return imp; +} + +void +_pixman_implementation_combine_32 (pixman_implementation_t * imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + (*imp->combine_32[op]) (imp, op, dest, src, mask, width); +} + +void +_pixman_implementation_combine_64 (pixman_implementation_t * imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + (*imp->combine_64[op]) (imp, op, dest, src, mask, width); +} + +void +_pixman_implementation_combine_32_ca (pixman_implementation_t * imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + (*imp->combine_32_ca[op]) (imp, op, dest, src, mask, width); +} + +void +_pixman_implementation_combine_64_ca (pixman_implementation_t * imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width) +{ + (*imp->combine_64_ca[op]) (imp, op, dest, src, mask, width); +} + +pixman_bool_t +_pixman_implementation_blt (pixman_implementation_t * imp, + uint32_t * src_bits, + uint32_t * dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height) +{ + return (*imp->blt) (imp, src_bits, dst_bits, src_stride, dst_stride, + src_bpp, dst_bpp, src_x, src_y, dst_x, dst_y, + width, height); +} + +pixman_bool_t +_pixman_implementation_fill (pixman_implementation_t *imp, + uint32_t * bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + return (*imp->fill) (imp, bits, stride, bpp, x, y, width, height, xor); +} + diff --git a/programs/develop/libraries/pixman/pixman-linear-gradient.c b/programs/develop/libraries/pixman/pixman-linear-gradient.c new file mode 100644 index 0000000000..b048e7b065 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-linear-gradient.c @@ -0,0 +1,262 @@ +/* -*- Mode: c; c-basic-offset: 4; tab-width: 8; indent-tabs-mode: t; -*- */ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007 Red Hat, Inc. + * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc. + * 2005 Lars Knoll & Zack Rusin, Trolltech + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include +#include "pixman-private.h" + +static source_image_class_t +linear_gradient_classify (pixman_image_t *image, + int x, + int y, + int width, + int height) +{ + source_image_t *source = (source_image_t *)image; + linear_gradient_t *linear = (linear_gradient_t *)image; + pixman_vector_t v; + pixman_fixed_32_32_t l; + pixman_fixed_48_16_t dx, dy; + double inc; + source_image_class_t class; + + class = SOURCE_IMAGE_CLASS_UNKNOWN; + + if (source->common.transform) + { + /* projective transformation */ + if (source->common.transform->matrix[2][0] != 0 || + source->common.transform->matrix[2][1] != 0 || + source->common.transform->matrix[2][2] == 0) + { + return class; + } + + v.vector[0] = source->common.transform->matrix[0][1]; + v.vector[1] = source->common.transform->matrix[1][1]; + v.vector[2] = source->common.transform->matrix[2][2]; + } + else + { + v.vector[0] = 0; + v.vector[1] = pixman_fixed_1; + v.vector[2] = pixman_fixed_1; + } + + dx = linear->p2.x - linear->p1.x; + dy = linear->p2.y - linear->p1.y; + + l = dx * dx + dy * dy; + + if (l == 0) + return class; + + /* + * compute how much the input of the gradient walked changes + * when moving vertically through the whole image + */ + inc = height * (double) pixman_fixed_1 * pixman_fixed_1 * + (dx * v.vector[0] + dy * v.vector[1]) / + (v.vector[2] * (double) l); + + /* check that casting to integer would result in 0 */ + if (-1 < inc && inc < 1) + class = SOURCE_IMAGE_CLASS_HORIZONTAL; + + return class; +} + +static void +linear_gradient_get_scanline_32 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + pixman_vector_t v, unit; + pixman_fixed_32_32_t l; + pixman_fixed_48_16_t dx, dy; + gradient_t *gradient = (gradient_t *)image; + source_image_t *source = (source_image_t *)image; + linear_gradient_t *linear = (linear_gradient_t *)image; + uint32_t *end = buffer + width; + pixman_gradient_walker_t walker; + + _pixman_gradient_walker_init (&walker, gradient, source->common.repeat); + + /* reference point is the center of the pixel */ + v.vector[0] = pixman_int_to_fixed (x) + pixman_fixed_1 / 2; + v.vector[1] = pixman_int_to_fixed (y) + pixman_fixed_1 / 2; + v.vector[2] = pixman_fixed_1; + + if (source->common.transform) + { + if (!pixman_transform_point_3d (source->common.transform, &v)) + return; + + unit.vector[0] = source->common.transform->matrix[0][0]; + unit.vector[1] = source->common.transform->matrix[1][0]; + unit.vector[2] = source->common.transform->matrix[2][0]; + } + else + { + unit.vector[0] = pixman_fixed_1; + unit.vector[1] = 0; + unit.vector[2] = 0; + } + + dx = linear->p2.x - linear->p1.x; + dy = linear->p2.y - linear->p1.y; + + l = dx * dx + dy * dy; + + if (l == 0 || unit.vector[2] == 0) + { + /* affine transformation only */ + pixman_fixed_32_32_t t, next_inc; + double inc; + + if (l == 0 || v.vector[2] == 0) + { + t = 0; + inc = 0; + } + else + { + double invden, v2; + + invden = pixman_fixed_1 * (double) pixman_fixed_1 / + (l * (double) v.vector[2]); + v2 = v.vector[2] * (1. / pixman_fixed_1); + t = ((dx * v.vector[0] + dy * v.vector[1]) - + (dx * linear->p1.x + dy * linear->p1.y) * v2) * invden; + inc = (dx * unit.vector[0] + dy * unit.vector[1]) * invden; + } + next_inc = 0; + + if (((pixman_fixed_32_32_t )(inc * width)) == 0) + { + register uint32_t color; + + color = _pixman_gradient_walker_pixel (&walker, t); + while (buffer < end) + *buffer++ = color; + } + else + { + int i; + + i = 0; + while (buffer < end) + { + if (!mask || *mask++) + { + *buffer = _pixman_gradient_walker_pixel (&walker, + t + next_inc); + } + i++; + next_inc = inc * i; + buffer++; + } + } + } + else + { + /* projective transformation */ + double t; + + t = 0; + + while (buffer < end) + { + if (!mask || *mask++) + { + if (v.vector[2] != 0) + { + double invden, v2; + + invden = pixman_fixed_1 * (double) pixman_fixed_1 / + (l * (double) v.vector[2]); + v2 = v.vector[2] * (1. / pixman_fixed_1); + t = ((dx * v.vector[0] + dy * v.vector[1]) - + (dx * linear->p1.x + dy * linear->p1.y) * v2) * invden; + } + + *buffer = _pixman_gradient_walker_pixel (&walker, t); + } + + ++buffer; + + v.vector[0] += unit.vector[0]; + v.vector[1] += unit.vector[1]; + v.vector[2] += unit.vector[2]; + } + } +} + +static void +linear_gradient_property_changed (pixman_image_t *image) +{ + image->common.get_scanline_32 = linear_gradient_get_scanline_32; + image->common.get_scanline_64 = _pixman_image_get_scanline_generic_64; +} + +PIXMAN_EXPORT pixman_image_t * +pixman_image_create_linear_gradient (pixman_point_fixed_t * p1, + pixman_point_fixed_t * p2, + const pixman_gradient_stop_t *stops, + int n_stops) +{ + pixman_image_t *image; + linear_gradient_t *linear; + + image = _pixman_image_allocate (); + + if (!image) + return NULL; + + linear = &image->linear; + + if (!_pixman_init_gradient (&linear->common, stops, n_stops)) + { + free (image); + return NULL; + } + + linear->p1 = *p1; + linear->p2 = *p2; + + image->type = LINEAR; + image->common.classify = linear_gradient_classify; + image->common.property_changed = linear_gradient_property_changed; + + return image; +} + diff --git a/programs/develop/libraries/pixman/pixman-matrix.c b/programs/develop/libraries/pixman/pixman-matrix.c new file mode 100644 index 0000000000..abdfa05258 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-matrix.c @@ -0,0 +1,768 @@ +/* + * Copyright © 2008 Keith Packard + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that copyright + * notice and this permission notice appear in supporting documentation, and + * that the name of the copyright holders not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. The copyright holders make no representations + * about the suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE + * OF THIS SOFTWARE. + */ + +/* + * Matrix interfaces + */ + +#ifdef HAVE_CONFIG_H +#include "config.h" +#endif + +#include +#include +#include "pixman-private.h" + +#define F(x) pixman_int_to_fixed (x) + +PIXMAN_EXPORT void +pixman_transform_init_identity (struct pixman_transform *matrix) +{ + int i; + + memset (matrix, '\0', sizeof (struct pixman_transform)); + for (i = 0; i < 3; i++) + matrix->matrix[i][i] = F (1); +} + +typedef pixman_fixed_32_32_t pixman_fixed_34_30_t; + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_point_3d (const struct pixman_transform *transform, + struct pixman_vector * vector) +{ + struct pixman_vector result; + pixman_fixed_32_32_t partial; + pixman_fixed_48_16_t v; + int i, j; + + for (j = 0; j < 3; j++) + { + v = 0; + for (i = 0; i < 3; i++) + { + partial = ((pixman_fixed_48_16_t) transform->matrix[j][i] * + (pixman_fixed_48_16_t) vector->vector[i]); + v += partial >> 16; + } + + if (v > pixman_max_fixed_48_16 || v < pixman_min_fixed_48_16) + return FALSE; + + result.vector[j] = (pixman_fixed_t) v; + } + + *vector = result; + + if (!result.vector[2]) + return FALSE; + + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_point (const struct pixman_transform *transform, + struct pixman_vector * vector) +{ + pixman_fixed_32_32_t partial; + pixman_fixed_34_30_t v[3]; + pixman_fixed_48_16_t quo; + int i, j; + + for (j = 0; j < 3; j++) + { + v[j] = 0; + + for (i = 0; i < 3; i++) + { + partial = ((pixman_fixed_32_32_t) transform->matrix[j][i] * + (pixman_fixed_32_32_t) vector->vector[i]); + v[j] += partial >> 2; + } + } + + if (!(v[2] >> 16)) + return FALSE; + + for (j = 0; j < 2; j++) + { + quo = v[j] / (v[2] >> 16); + if (quo > pixman_max_fixed_48_16 || quo < pixman_min_fixed_48_16) + return FALSE; + vector->vector[j] = (pixman_fixed_t) quo; + } + + vector->vector[2] = pixman_fixed_1; + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_multiply (struct pixman_transform * dst, + const struct pixman_transform *l, + const struct pixman_transform *r) +{ + struct pixman_transform d; + int dx, dy; + int o; + + for (dy = 0; dy < 3; dy++) + { + for (dx = 0; dx < 3; dx++) + { + pixman_fixed_48_16_t v; + pixman_fixed_32_32_t partial; + + v = 0; + for (o = 0; o < 3; o++) + { + partial = + (pixman_fixed_32_32_t) l->matrix[dy][o] * + (pixman_fixed_32_32_t) r->matrix[o][dx]; + + v += partial >> 16; + } + + if (v > pixman_max_fixed_48_16 || v < pixman_min_fixed_48_16) + return FALSE; + + d.matrix[dy][dx] = (pixman_fixed_t) v; + } + } + + *dst = d; + return TRUE; +} + +PIXMAN_EXPORT void +pixman_transform_init_scale (struct pixman_transform *t, + pixman_fixed_t sx, + pixman_fixed_t sy) +{ + memset (t, '\0', sizeof (struct pixman_transform)); + + t->matrix[0][0] = sx; + t->matrix[1][1] = sy; + t->matrix[2][2] = F (1); +} + +static pixman_fixed_t +fixed_inverse (pixman_fixed_t x) +{ + return (pixman_fixed_t) ((((pixman_fixed_48_16_t) F (1)) * F (1)) / x); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_scale (struct pixman_transform *forward, + struct pixman_transform *reverse, + pixman_fixed_t sx, + pixman_fixed_t sy) +{ + struct pixman_transform t; + + if (sx == 0 || sy == 0) + return FALSE; + + if (forward) + { + pixman_transform_init_scale (&t, sx, sy); + if (!pixman_transform_multiply (forward, &t, forward)) + return FALSE; + } + + if (reverse) + { + pixman_transform_init_scale (&t, fixed_inverse (sx), + fixed_inverse (sy)); + if (!pixman_transform_multiply (reverse, reverse, &t)) + return FALSE; + } + + return TRUE; +} + +PIXMAN_EXPORT void +pixman_transform_init_rotate (struct pixman_transform *t, + pixman_fixed_t c, + pixman_fixed_t s) +{ + memset (t, '\0', sizeof (struct pixman_transform)); + + t->matrix[0][0] = c; + t->matrix[0][1] = -s; + t->matrix[1][0] = s; + t->matrix[1][1] = c; + t->matrix[2][2] = F (1); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_rotate (struct pixman_transform *forward, + struct pixman_transform *reverse, + pixman_fixed_t c, + pixman_fixed_t s) +{ + struct pixman_transform t; + + if (forward) + { + pixman_transform_init_rotate (&t, c, s); + if (!pixman_transform_multiply (forward, &t, forward)) + return FALSE; + } + + if (reverse) + { + pixman_transform_init_rotate (&t, c, -s); + if (!pixman_transform_multiply (reverse, reverse, &t)) + return FALSE; + } + + return TRUE; +} + +PIXMAN_EXPORT void +pixman_transform_init_translate (struct pixman_transform *t, + pixman_fixed_t tx, + pixman_fixed_t ty) +{ + memset (t, '\0', sizeof (struct pixman_transform)); + + t->matrix[0][0] = F (1); + t->matrix[0][2] = tx; + t->matrix[1][1] = F (1); + t->matrix[1][2] = ty; + t->matrix[2][2] = F (1); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_translate (struct pixman_transform *forward, + struct pixman_transform *reverse, + pixman_fixed_t tx, + pixman_fixed_t ty) +{ + struct pixman_transform t; + + if (forward) + { + pixman_transform_init_translate (&t, tx, ty); + + if (!pixman_transform_multiply (forward, &t, forward)) + return FALSE; + } + + if (reverse) + { + pixman_transform_init_translate (&t, -tx, -ty); + + if (!pixman_transform_multiply (reverse, reverse, &t)) + return FALSE; + } + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_bounds (const struct pixman_transform *matrix, + struct pixman_box16 * b) + +{ + struct pixman_vector v[4]; + int i; + int x1, y1, x2, y2; + + v[0].vector[0] = F (b->x1); + v[0].vector[1] = F (b->y1); + v[0].vector[2] = F (1); + + v[1].vector[0] = F (b->x2); + v[1].vector[1] = F (b->y1); + v[1].vector[2] = F (1); + + v[2].vector[0] = F (b->x2); + v[2].vector[1] = F (b->y2); + v[2].vector[2] = F (1); + + v[3].vector[0] = F (b->x1); + v[3].vector[1] = F (b->y2); + v[3].vector[2] = F (1); + + for (i = 0; i < 4; i++) + { + if (!pixman_transform_point (matrix, &v[i])) + return FALSE; + + x1 = pixman_fixed_to_int (v[i].vector[0]); + y1 = pixman_fixed_to_int (v[i].vector[1]); + x2 = pixman_fixed_to_int (pixman_fixed_ceil (v[i].vector[0])); + y2 = pixman_fixed_to_int (pixman_fixed_ceil (v[i].vector[1])); + + if (i == 0) + { + b->x1 = x1; + b->y1 = y1; + b->x2 = x2; + b->y2 = y2; + } + else + { + if (x1 < b->x1) b->x1 = x1; + if (y1 < b->y1) b->y1 = y1; + if (x2 > b->x2) b->x2 = x2; + if (y2 > b->y2) b->y2 = y2; + } + } + + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_invert (struct pixman_transform * dst, + const struct pixman_transform *src) +{ + struct pixman_f_transform m, r; + + pixman_f_transform_from_pixman_transform (&m, src); + + if (!pixman_f_transform_invert (&r, &m)) + return FALSE; + + if (!pixman_transform_from_pixman_f_transform (dst, &r)) + return FALSE; + + return TRUE; +} + +static pixman_bool_t +within_epsilon (pixman_fixed_t a, + pixman_fixed_t b, + pixman_fixed_t epsilon) +{ + pixman_fixed_t t = a - b; + + if (t < 0) + t = -t; + + return t <= epsilon; +} + +#define EPSILON (pixman_fixed_t) (2) + +#define IS_SAME(a, b) (within_epsilon (a, b, EPSILON)) +#define IS_ZERO(a) (within_epsilon (a, 0, EPSILON)) +#define IS_ONE(a) (within_epsilon (a, F (1), EPSILON)) +#define IS_UNIT(a) \ + (within_epsilon (a, F (1), EPSILON) || \ + within_epsilon (a, F (-1), EPSILON) || \ + IS_ZERO (a)) +#define IS_INT(a) (IS_ZERO (pixman_fixed_frac (a))) + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_is_identity (const struct pixman_transform *t) +{ + return (IS_SAME (t->matrix[0][0], t->matrix[1][1]) && + IS_SAME (t->matrix[0][0], t->matrix[2][2]) && + !IS_ZERO (t->matrix[0][0]) && + IS_ZERO (t->matrix[0][1]) && + IS_ZERO (t->matrix[0][2]) && + IS_ZERO (t->matrix[1][0]) && + IS_ZERO (t->matrix[1][2]) && + IS_ZERO (t->matrix[2][0]) && + IS_ZERO (t->matrix[2][1])); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_is_scale (const struct pixman_transform *t) +{ + return (!IS_ZERO (t->matrix[0][0]) && + IS_ZERO (t->matrix[0][1]) && + IS_ZERO (t->matrix[0][2]) && + + IS_ZERO (t->matrix[1][0]) && + !IS_ZERO (t->matrix[1][1]) && + IS_ZERO (t->matrix[1][2]) && + + IS_ZERO (t->matrix[2][0]) && + IS_ZERO (t->matrix[2][1]) && + !IS_ZERO (t->matrix[2][2])); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_is_int_translate (const struct pixman_transform *t) +{ + return (IS_ONE (t->matrix[0][0]) && + IS_ZERO (t->matrix[0][1]) && + IS_INT (t->matrix[0][2]) && + + IS_ZERO (t->matrix[1][0]) && + IS_ONE (t->matrix[1][1]) && + IS_INT (t->matrix[1][2]) && + + IS_ZERO (t->matrix[2][0]) && + IS_ZERO (t->matrix[2][1]) && + IS_ONE (t->matrix[2][2])); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_is_inverse (const struct pixman_transform *a, + const struct pixman_transform *b) +{ + struct pixman_transform t; + + pixman_transform_multiply (&t, a, b); + + return pixman_transform_is_identity (&t); +} + +PIXMAN_EXPORT void +pixman_f_transform_from_pixman_transform (struct pixman_f_transform * ft, + const struct pixman_transform *t) +{ + int i, j; + + for (j = 0; j < 3; j++) + { + for (i = 0; i < 3; i++) + ft->m[j][i] = pixman_fixed_to_double (t->matrix[j][i]); + } +} + +PIXMAN_EXPORT pixman_bool_t +pixman_transform_from_pixman_f_transform (struct pixman_transform * t, + const struct pixman_f_transform *ft) +{ + int i, j; + + for (j = 0; j < 3; j++) + { + for (i = 0; i < 3; i++) + { + double d = ft->m[j][i]; + if (d < -32767.0 || d > 32767.0) + return FALSE; + d = d * 65536.0 + 0.5; + t->matrix[j][i] = (pixman_fixed_t) floor (d); + } + } + + return TRUE; +} + +static const int a[3] = { 3, 3, 2 }; +static const int b[3] = { 2, 1, 1 }; + +PIXMAN_EXPORT pixman_bool_t +pixman_f_transform_invert (struct pixman_f_transform * dst, + const struct pixman_f_transform *src) +{ + double det; + int i, j; + static int a[3] = { 2, 2, 1 }; + static int b[3] = { 1, 0, 0 }; + + det = 0; + for (i = 0; i < 3; i++) + { + double p; + int ai = a[i]; + int bi = b[i]; + p = src->m[i][0] * (src->m[ai][2] * src->m[bi][1] - + src->m[ai][1] * src->m[bi][2]); + if (i == 1) + p = -p; + det += p; + } + + if (det == 0) + return FALSE; + + det = 1 / det; + for (j = 0; j < 3; j++) + { + for (i = 0; i < 3; i++) + { + double p; + int ai = a[i]; + int aj = a[j]; + int bi = b[i]; + int bj = b[j]; + + p = (src->m[ai][aj] * src->m[bi][bj] - + src->m[ai][bj] * src->m[bi][aj]); + + if (((i + j) & 1) != 0) + p = -p; + + dst->m[j][i] = det * p; + } + } + + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_f_transform_point (const struct pixman_f_transform *t, + struct pixman_f_vector * v) +{ + struct pixman_f_vector result; + int i, j; + double a; + + for (j = 0; j < 3; j++) + { + a = 0; + for (i = 0; i < 3; i++) + a += t->m[j][i] * v->v[i]; + result.v[j] = a; + } + + if (!result.v[2]) + return FALSE; + + for (j = 0; j < 2; j++) + v->v[j] = result.v[j] / result.v[2]; + + v->v[2] = 1; + + return TRUE; +} + +PIXMAN_EXPORT void +pixman_f_transform_point_3d (const struct pixman_f_transform *t, + struct pixman_f_vector * v) +{ + struct pixman_f_vector result; + int i, j; + double a; + + for (j = 0; j < 3; j++) + { + a = 0; + for (i = 0; i < 3; i++) + a += t->m[j][i] * v->v[i]; + result.v[j] = a; + } + + *v = result; +} + +PIXMAN_EXPORT void +pixman_f_transform_multiply (struct pixman_f_transform * dst, + const struct pixman_f_transform *l, + const struct pixman_f_transform *r) +{ + struct pixman_f_transform d; + int dx, dy; + int o; + + for (dy = 0; dy < 3; dy++) + { + for (dx = 0; dx < 3; dx++) + { + double v = 0; + for (o = 0; o < 3; o++) + v += l->m[dy][o] * r->m[o][dx]; + d.m[dy][dx] = v; + } + } + + *dst = d; +} + +PIXMAN_EXPORT void +pixman_f_transform_init_scale (struct pixman_f_transform *t, + double sx, + double sy) +{ + t->m[0][0] = sx; + t->m[0][1] = 0; + t->m[0][2] = 0; + t->m[1][0] = 0; + t->m[1][1] = sy; + t->m[1][2] = 0; + t->m[2][0] = 0; + t->m[2][1] = 0; + t->m[2][2] = 1; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_f_transform_scale (struct pixman_f_transform *forward, + struct pixman_f_transform *reverse, + double sx, + double sy) +{ + struct pixman_f_transform t; + + if (sx == 0 || sy == 0) + return FALSE; + + if (forward) + { + pixman_f_transform_init_scale (&t, sx, sy); + pixman_f_transform_multiply (forward, &t, forward); + } + + if (reverse) + { + pixman_f_transform_init_scale (&t, 1 / sx, 1 / sy); + pixman_f_transform_multiply (reverse, reverse, &t); + } + + return TRUE; +} + +PIXMAN_EXPORT void +pixman_f_transform_init_rotate (struct pixman_f_transform *t, + double c, + double s) +{ + t->m[0][0] = c; + t->m[0][1] = -s; + t->m[0][2] = 0; + t->m[1][0] = s; + t->m[1][1] = c; + t->m[1][2] = 0; + t->m[2][0] = 0; + t->m[2][1] = 0; + t->m[2][2] = 1; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_f_transform_rotate (struct pixman_f_transform *forward, + struct pixman_f_transform *reverse, + double c, + double s) +{ + struct pixman_f_transform t; + + if (forward) + { + pixman_f_transform_init_rotate (&t, c, s); + pixman_f_transform_multiply (forward, &t, forward); + } + + if (reverse) + { + pixman_f_transform_init_rotate (&t, c, -s); + pixman_f_transform_multiply (reverse, reverse, &t); + } + + return TRUE; +} + +PIXMAN_EXPORT void +pixman_f_transform_init_translate (struct pixman_f_transform *t, + double tx, + double ty) +{ + t->m[0][0] = 1; + t->m[0][1] = 0; + t->m[0][2] = tx; + t->m[1][0] = 0; + t->m[1][1] = 1; + t->m[1][2] = ty; + t->m[2][0] = 0; + t->m[2][1] = 0; + t->m[2][2] = 1; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_f_transform_translate (struct pixman_f_transform *forward, + struct pixman_f_transform *reverse, + double tx, + double ty) +{ + struct pixman_f_transform t; + + if (forward) + { + pixman_f_transform_init_translate (&t, tx, ty); + pixman_f_transform_multiply (forward, &t, forward); + } + + if (reverse) + { + pixman_f_transform_init_translate (&t, -tx, -ty); + pixman_f_transform_multiply (reverse, reverse, &t); + } + + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_f_transform_bounds (const struct pixman_f_transform *t, + struct pixman_box16 * b) +{ + struct pixman_f_vector v[4]; + int i; + int x1, y1, x2, y2; + + v[0].v[0] = b->x1; + v[0].v[1] = b->y1; + v[0].v[2] = 1; + v[1].v[0] = b->x2; + v[1].v[1] = b->y1; + v[1].v[2] = 1; + v[2].v[0] = b->x2; + v[2].v[1] = b->y2; + v[2].v[2] = 1; + v[3].v[0] = b->x1; + v[3].v[1] = b->y2; + v[3].v[2] = 1; + + for (i = 0; i < 4; i++) + { + if (!pixman_f_transform_point (t, &v[i])) + return FALSE; + + x1 = floor (v[i].v[0]); + y1 = floor (v[i].v[1]); + x2 = ceil (v[i].v[0]); + y2 = ceil (v[i].v[1]); + + if (i == 0) + { + b->x1 = x1; + b->y1 = y1; + b->x2 = x2; + b->y2 = y2; + } + else + { + if (x1 < b->x1) b->x1 = x1; + if (y1 < b->y1) b->y1 = y1; + if (x2 > b->x2) b->x2 = x2; + if (y2 > b->y2) b->y2 = y2; + } + } + + return TRUE; +} + +PIXMAN_EXPORT void +pixman_f_transform_init_identity (struct pixman_f_transform *t) +{ + int i, j; + + for (j = 0; j < 3; j++) + { + for (i = 0; i < 3; i++) + t->m[j][i] = i == j ? 1 : 0; + } +} diff --git a/programs/develop/libraries/pixman/pixman-mmx.c b/programs/develop/libraries/pixman/pixman-mmx.c new file mode 100644 index 0000000000..34637a4fe3 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-mmx.c @@ -0,0 +1,3378 @@ +/* + * Copyright © 2004, 2005 Red Hat, Inc. + * Copyright © 2004 Nicholas Miell + * Copyright © 2005 Trolltech AS + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Red Hat not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. Red Hat makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + * + * Author: Søren Sandmann (sandmann@redhat.com) + * Minor Improvements: Nicholas Miell (nmiell@gmail.com) + * MMX code paths for fbcompose.c by Lars Knoll (lars@trolltech.com) + * + * Based on work by Owen Taylor + */ + +#ifdef HAVE_CONFIG_H +#include +#endif + +#ifdef USE_MMX + +#include +#include "pixman-private.h" +#include "pixman-combine32.h" + +#define no_vERBOSE + +#ifdef VERBOSE +#define CHECKPOINT() error_f ("at %s %d\n", __FUNCTION__, __LINE__) +#else +#define CHECKPOINT() +#endif + +/* Notes about writing mmx code + * + * give memory operands as the second operand. If you give it as the + * first, gcc will first load it into a register, then use that + * register + * + * ie. use + * + * _mm_mullo_pi16 (x, mmx_constant); + * + * not + * + * _mm_mullo_pi16 (mmx_constant, x); + * + * Also try to minimize dependencies. i.e. when you need a value, try + * to calculate it from a value that was calculated as early as + * possible. + */ + +/* --------------- MMX primitives ------------------------------------- */ + +#ifdef __GNUC__ +typedef uint64_t mmxdatafield; +#else +typedef __m64 mmxdatafield; +/* If __m64 is defined as a struct or union, define M64_MEMBER to be the + name of the member used to access the data */ +# ifdef _MSC_VER +# define M64_MEMBER m64_u64 +# elif defined(__SUNPRO_C) +# define M64_MEMBER l_ +# endif +#endif + +typedef struct +{ + mmxdatafield mmx_4x00ff; + mmxdatafield mmx_4x0080; + mmxdatafield mmx_565_rgb; + mmxdatafield mmx_565_unpack_multiplier; + mmxdatafield mmx_565_r; + mmxdatafield mmx_565_g; + mmxdatafield mmx_565_b; + mmxdatafield mmx_mask_0; + mmxdatafield mmx_mask_1; + mmxdatafield mmx_mask_2; + mmxdatafield mmx_mask_3; + mmxdatafield mmx_full_alpha; + mmxdatafield mmx_ffff0000ffff0000; + mmxdatafield mmx_0000ffff00000000; + mmxdatafield mmx_000000000000ffff; +} mmx_data_t; + +#if defined(_MSC_VER) +# define MMXDATA_INIT(field, val) { val ## UI64 } +#elif defined(M64_MEMBER) /* __m64 is a struct, not an integral type */ +# define MMXDATA_INIT(field, val) field = { val ## ULL } +#else /* __m64 is an integral type */ +# define MMXDATA_INIT(field, val) field = val ## ULL +#endif + +static const mmx_data_t c = +{ + MMXDATA_INIT (.mmx_4x00ff, 0x00ff00ff00ff00ff), + MMXDATA_INIT (.mmx_4x0080, 0x0080008000800080), + MMXDATA_INIT (.mmx_565_rgb, 0x000001f0003f001f), + MMXDATA_INIT (.mmx_565_unpack_multiplier, 0x0000008404100840), + MMXDATA_INIT (.mmx_565_r, 0x000000f800000000), + MMXDATA_INIT (.mmx_565_g, 0x0000000000fc0000), + MMXDATA_INIT (.mmx_565_b, 0x00000000000000f8), + MMXDATA_INIT (.mmx_mask_0, 0xffffffffffff0000), + MMXDATA_INIT (.mmx_mask_1, 0xffffffff0000ffff), + MMXDATA_INIT (.mmx_mask_2, 0xffff0000ffffffff), + MMXDATA_INIT (.mmx_mask_3, 0x0000ffffffffffff), + MMXDATA_INIT (.mmx_full_alpha, 0x00ff000000000000), + MMXDATA_INIT (.mmx_ffff0000ffff0000, 0xffff0000ffff0000), + MMXDATA_INIT (.mmx_0000ffff00000000, 0x0000ffff00000000), + MMXDATA_INIT (.mmx_000000000000ffff, 0x000000000000ffff), +}; + +#ifdef __GNUC__ +# ifdef __ICC +# define MC(x) to_m64 (c.mmx_ ## x) +# else +# define MC(x) ((__m64)c.mmx_ ## x) +# endif +#else +# define MC(x) c.mmx_ ## x +#endif + +static force_inline __m64 +to_m64 (uint64_t x) +{ +#ifdef __ICC + return _mm_cvtsi64_m64 (x); +#elif defined M64_MEMBER /* __m64 is a struct, not an integral type */ + __m64 res; + + res.M64_MEMBER = x; + return res; +#else /* __m64 is an integral type */ + return (__m64)x; +#endif +} + +static force_inline uint64_t +to_uint64 (__m64 x) +{ +#ifdef __ICC + return _mm_cvtm64_si64 (x); +#elif defined M64_MEMBER /* __m64 is a struct, not an integral type */ + uint64_t res = x.M64_MEMBER; + return res; +#else /* __m64 is an integral type */ + return (uint64_t)x; +#endif +} + +static force_inline __m64 +shift (__m64 v, + int s) +{ + if (s > 0) + return _mm_slli_si64 (v, s); + else if (s < 0) + return _mm_srli_si64 (v, -s); + else + return v; +} + +static force_inline __m64 +negate (__m64 mask) +{ + return _mm_xor_si64 (mask, MC (4x00ff)); +} + +static force_inline __m64 +pix_multiply (__m64 a, __m64 b) +{ + __m64 res; + + res = _mm_mullo_pi16 (a, b); + res = _mm_adds_pu16 (res, MC (4x0080)); + res = _mm_adds_pu16 (res, _mm_srli_pi16 (res, 8)); + res = _mm_srli_pi16 (res, 8); + + return res; +} + +static force_inline __m64 +pix_add (__m64 a, __m64 b) +{ + return _mm_adds_pu8 (a, b); +} + +static force_inline __m64 +expand_alpha (__m64 pixel) +{ + __m64 t1, t2; + + t1 = shift (pixel, -48); + t2 = shift (t1, 16); + t1 = _mm_or_si64 (t1, t2); + t2 = shift (t1, 32); + t1 = _mm_or_si64 (t1, t2); + + return t1; +} + +static force_inline __m64 +expand_alpha_rev (__m64 pixel) +{ + __m64 t1, t2; + + /* move alpha to low 16 bits and zero the rest */ + t1 = shift (pixel, 48); + t1 = shift (t1, -48); + + t2 = shift (t1, 16); + t1 = _mm_or_si64 (t1, t2); + t2 = shift (t1, 32); + t1 = _mm_or_si64 (t1, t2); + + return t1; +} + +static force_inline __m64 +invert_colors (__m64 pixel) +{ + __m64 x, y, z; + + x = y = z = pixel; + + x = _mm_and_si64 (x, MC (ffff0000ffff0000)); + y = _mm_and_si64 (y, MC (000000000000ffff)); + z = _mm_and_si64 (z, MC (0000ffff00000000)); + + y = shift (y, 32); + z = shift (z, -32); + + x = _mm_or_si64 (x, y); + x = _mm_or_si64 (x, z); + + return x; +} + +static force_inline __m64 +over (__m64 src, + __m64 srca, + __m64 dest) +{ + return _mm_adds_pu8 (src, pix_multiply (dest, negate (srca))); +} + +static force_inline __m64 +over_rev_non_pre (__m64 src, __m64 dest) +{ + __m64 srca = expand_alpha (src); + __m64 srcfaaa = _mm_or_si64 (srca, MC (full_alpha)); + + return over (pix_multiply (invert_colors (src), srcfaaa), srca, dest); +} + +static force_inline __m64 +in (__m64 src, __m64 mask) +{ + return pix_multiply (src, mask); +} + +static force_inline __m64 +in_over_full_src_alpha (__m64 src, __m64 mask, __m64 dest) +{ + src = _mm_or_si64 (src, MC (full_alpha)); + + return over (in (src, mask), mask, dest); +} + +#ifndef _MSC_VER +static force_inline __m64 +in_over (__m64 src, __m64 srca, __m64 mask, __m64 dest) +{ + return over (in (src, mask), pix_multiply (srca, mask), dest); +} + +#else + +#define in_over(src, srca, mask, dest) \ + over (in (src, mask), pix_multiply (srca, mask), dest) + +#endif + +static force_inline __m64 +load8888 (uint32_t v) +{ + return _mm_unpacklo_pi8 (_mm_cvtsi32_si64 (v), _mm_setzero_si64 ()); +} + +static force_inline __m64 +pack8888 (__m64 lo, __m64 hi) +{ + return _mm_packs_pu16 (lo, hi); +} + +static force_inline uint32_t +store8888 (__m64 v) +{ + return _mm_cvtsi64_si32 (pack8888 (v, _mm_setzero_si64 ())); +} + +/* Expand 16 bits positioned at @pos (0-3) of a mmx register into + * + * 00RR00GG00BB + * + * --- Expanding 565 in the low word --- + * + * m = (m << (32 - 3)) | (m << (16 - 5)) | m; + * m = m & (01f0003f001f); + * m = m * (008404100840); + * m = m >> 8; + * + * Note the trick here - the top word is shifted by another nibble to + * avoid it bumping into the middle word + */ +static force_inline __m64 +expand565 (__m64 pixel, int pos) +{ + __m64 p = pixel; + __m64 t1, t2; + + /* move pixel to low 16 bit and zero the rest */ + p = shift (shift (p, (3 - pos) * 16), -48); + + t1 = shift (p, 36 - 11); + t2 = shift (p, 16 - 5); + + p = _mm_or_si64 (t1, p); + p = _mm_or_si64 (t2, p); + p = _mm_and_si64 (p, MC (565_rgb)); + + pixel = _mm_mullo_pi16 (p, MC (565_unpack_multiplier)); + return _mm_srli_pi16 (pixel, 8); +} + +static force_inline __m64 +expand8888 (__m64 in, int pos) +{ + if (pos == 0) + return _mm_unpacklo_pi8 (in, _mm_setzero_si64 ()); + else + return _mm_unpackhi_pi8 (in, _mm_setzero_si64 ()); +} + +static force_inline __m64 +expandx888 (__m64 in, int pos) +{ + return _mm_or_si64 (expand8888 (in, pos), MC (full_alpha)); +} + +static force_inline __m64 +pack_565 (__m64 pixel, __m64 target, int pos) +{ + __m64 p = pixel; + __m64 t = target; + __m64 r, g, b; + + r = _mm_and_si64 (p, MC (565_r)); + g = _mm_and_si64 (p, MC (565_g)); + b = _mm_and_si64 (p, MC (565_b)); + + r = shift (r, -(32 - 8) + pos * 16); + g = shift (g, -(16 - 3) + pos * 16); + b = shift (b, -(0 + 3) + pos * 16); + + if (pos == 0) + t = _mm_and_si64 (t, MC (mask_0)); + else if (pos == 1) + t = _mm_and_si64 (t, MC (mask_1)); + else if (pos == 2) + t = _mm_and_si64 (t, MC (mask_2)); + else if (pos == 3) + t = _mm_and_si64 (t, MC (mask_3)); + + p = _mm_or_si64 (r, t); + p = _mm_or_si64 (g, p); + + return _mm_or_si64 (b, p); +} + +#ifndef _MSC_VER + +static force_inline __m64 +pix_add_mul (__m64 x, __m64 a, __m64 y, __m64 b) +{ + x = pix_multiply (x, a); + y = pix_multiply (y, b); + + return pix_add (x, y); +} + +#else + +#define pix_add_mul(x, a, y, b) \ + ( x = pix_multiply (x, a), \ + y = pix_multiply (y, a), \ + pix_add (x, y) ) + +#endif + +/* --------------- MMX code patch for fbcompose.c --------------------- */ + +static force_inline uint32_t +combine (const uint32_t *src, const uint32_t *mask) +{ + uint32_t ssrc = *src; + + if (mask) + { + __m64 m = load8888 (*mask); + __m64 s = load8888 (ssrc); + + m = expand_alpha (m); + s = pix_multiply (s, m); + + ssrc = store8888 (s); + } + + return ssrc; +} + +static void +mmx_combine_over_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + uint32_t ssrc = combine (src, mask); + uint32_t a = ssrc >> 24; + + if (a == 0xff) + { + *dest = ssrc; + } + else if (ssrc) + { + __m64 s, sa; + s = load8888 (ssrc); + sa = expand_alpha (s); + *dest = store8888 (over (s, sa, load8888 (*dest))); + } + + ++dest; + ++src; + if (mask) + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_over_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + __m64 d, da; + uint32_t s = combine (src, mask); + + d = load8888 (*dest); + da = expand_alpha (d); + *dest = store8888 (over (d, da, load8888 (s))); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_in_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + __m64 x, a; + + x = load8888 (combine (src, mask)); + a = load8888 (*dest); + a = expand_alpha (a); + x = pix_multiply (x, a); + + *dest = store8888 (x); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_in_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + __m64 x, a; + + x = load8888 (*dest); + a = load8888 (combine (src, mask)); + a = expand_alpha (a); + x = pix_multiply (x, a); + *dest = store8888 (x); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_out_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + __m64 x, a; + + x = load8888 (combine (src, mask)); + a = load8888 (*dest); + a = expand_alpha (a); + a = negate (a); + x = pix_multiply (x, a); + *dest = store8888 (x); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_out_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + __m64 x, a; + + x = load8888 (*dest); + a = load8888 (combine (src, mask)); + a = expand_alpha (a); + a = negate (a); + x = pix_multiply (x, a); + + *dest = store8888 (x); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_atop_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + __m64 s, da, d, sia; + + s = load8888 (combine (src, mask)); + d = load8888 (*dest); + sia = expand_alpha (s); + sia = negate (sia); + da = expand_alpha (d); + s = pix_add_mul (s, da, d, sia); + *dest = store8888 (s); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_atop_reverse_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end; + + end = dest + width; + + while (dest < end) + { + __m64 s, dia, d, sa; + + s = load8888 (combine (src, mask)); + d = load8888 (*dest); + sa = expand_alpha (s); + dia = expand_alpha (d); + dia = negate (dia); + s = pix_add_mul (s, dia, d, sa); + *dest = store8888 (s); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_xor_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + __m64 s, dia, d, sia; + + s = load8888 (combine (src, mask)); + d = load8888 (*dest); + sia = expand_alpha (s); + dia = expand_alpha (d); + sia = negate (sia); + dia = negate (dia); + s = pix_add_mul (s, dia, d, sia); + *dest = store8888 (s); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_add_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + __m64 s, d; + + s = load8888 (combine (src, mask)); + d = load8888 (*dest); + s = pix_add (s, d); + *dest = store8888 (s); + + ++dest; + ++src; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_saturate_u (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = dest + width; + + while (dest < end) + { + uint32_t s = combine (src, mask); + uint32_t d = *dest; + __m64 ms = load8888 (s); + __m64 md = load8888 (d); + uint32_t sa = s >> 24; + uint32_t da = ~d >> 24; + + if (sa > da) + { + __m64 msa = load8888 (DIV_UN8 (da, sa) << 24); + msa = expand_alpha (msa); + ms = pix_multiply (ms, msa); + } + + md = pix_add (md, ms); + *dest = store8888 (md); + + ++src; + ++dest; + if (mask) + mask++; + } + _mm_empty (); +} + +static void +mmx_combine_src_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + + s = pix_multiply (s, a); + *dest = store8888 (s); + + ++src; + ++mask; + ++dest; + } + _mm_empty (); +} + +static void +mmx_combine_over_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 sa = expand_alpha (s); + + *dest = store8888 (in_over (s, sa, a, d)); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_over_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 da = expand_alpha (d); + + *dest = store8888 (over (d, da, in (s, a))); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_in_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 da = expand_alpha (d); + + s = pix_multiply (s, a); + s = pix_multiply (s, da); + *dest = store8888 (s); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_in_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 sa = expand_alpha (s); + + a = pix_multiply (a, sa); + d = pix_multiply (d, a); + *dest = store8888 (d); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_out_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 da = expand_alpha (d); + + da = negate (da); + s = pix_multiply (s, a); + s = pix_multiply (s, da); + *dest = store8888 (s); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_out_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 sa = expand_alpha (s); + + a = pix_multiply (a, sa); + a = negate (a); + d = pix_multiply (d, a); + *dest = store8888 (d); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_atop_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 da = expand_alpha (d); + __m64 sa = expand_alpha (s); + + s = pix_multiply (s, a); + a = pix_multiply (a, sa); + a = negate (a); + d = pix_add_mul (d, a, s, da); + *dest = store8888 (d); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_atop_reverse_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 da = expand_alpha (d); + __m64 sa = expand_alpha (s); + + s = pix_multiply (s, a); + a = pix_multiply (a, sa); + da = negate (da); + d = pix_add_mul (d, a, s, da); + *dest = store8888 (d); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_xor_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + __m64 da = expand_alpha (d); + __m64 sa = expand_alpha (s); + + s = pix_multiply (s, a); + a = pix_multiply (a, sa); + da = negate (da); + a = negate (a); + d = pix_add_mul (d, a, s, da); + *dest = store8888 (d); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +static void +mmx_combine_add_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width) +{ + const uint32_t *end = src + width; + + while (src < end) + { + __m64 a = load8888 (*mask); + __m64 s = load8888 (*src); + __m64 d = load8888 (*dest); + + s = pix_multiply (s, a); + d = pix_add (s, d); + *dest = store8888 (d); + + ++src; + ++dest; + ++mask; + } + _mm_empty (); +} + +/* ------------- MMX code paths called from fbpict.c -------------------- */ + +static void +mmx_composite_over_n_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src; + uint32_t *dst_line, *dst; + int32_t w; + int dst_stride; + __m64 vsrc, vsrca; + + CHECKPOINT (); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + w = width; + + CHECKPOINT (); + + while (w && (unsigned long)dst & 7) + { + *dst = store8888 (over (vsrc, vsrca, load8888 (*dst))); + + w--; + dst++; + } + + while (w >= 2) + { + __m64 vdest; + __m64 dest0, dest1; + + vdest = *(__m64 *)dst; + + dest0 = over (vsrc, vsrca, expand8888 (vdest, 0)); + dest1 = over (vsrc, vsrca, expand8888 (vdest, 1)); + + *(__m64 *)dst = pack8888 (dest0, dest1); + + dst += 2; + w -= 2; + } + + CHECKPOINT (); + + while (w) + { + *dst = store8888 (over (vsrc, vsrca, load8888 (*dst))); + + w--; + dst++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_n_0565 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src; + uint16_t *dst_line, *dst; + int32_t w; + int dst_stride; + __m64 vsrc, vsrca; + + CHECKPOINT (); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + w = width; + + CHECKPOINT (); + + while (w && (unsigned long)dst & 7) + { + uint64_t d = *dst; + __m64 vdest = expand565 (to_m64 (d), 0); + + vdest = pack_565 (over (vsrc, vsrca, vdest), vdest, 0); + *dst = to_uint64 (vdest); + + w--; + dst++; + } + + while (w >= 4) + { + __m64 vdest; + + vdest = *(__m64 *)dst; + + vdest = pack_565 (over (vsrc, vsrca, expand565 (vdest, 0)), vdest, 0); + vdest = pack_565 (over (vsrc, vsrca, expand565 (vdest, 1)), vdest, 1); + vdest = pack_565 (over (vsrc, vsrca, expand565 (vdest, 2)), vdest, 2); + vdest = pack_565 (over (vsrc, vsrca, expand565 (vdest, 3)), vdest, 3); + + *(__m64 *)dst = vdest; + + dst += 4; + w -= 4; + } + + CHECKPOINT (); + + while (w) + { + uint64_t d = *dst; + __m64 vdest = expand565 (to_m64 (d), 0); + + vdest = pack_565 (over (vsrc, vsrca, vdest), vdest, 0); + *dst = to_uint64 (vdest); + + w--; + dst++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_n_8888_8888_ca (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint32_t *dst_line; + uint32_t *mask_line; + int dst_stride, mask_stride; + __m64 vsrc, vsrca; + + CHECKPOINT (); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1); + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + while (height--) + { + int twidth = width; + uint32_t *p = (uint32_t *)mask_line; + uint32_t *q = (uint32_t *)dst_line; + + while (twidth && (unsigned long)q & 7) + { + uint32_t m = *(uint32_t *)p; + + if (m) + { + __m64 vdest = load8888 (*q); + vdest = in_over (vsrc, vsrca, load8888 (m), vdest); + *q = store8888 (vdest); + } + + twidth--; + p++; + q++; + } + + while (twidth >= 2) + { + uint32_t m0, m1; + m0 = *p; + m1 = *(p + 1); + + if (m0 | m1) + { + __m64 dest0, dest1; + __m64 vdest = *(__m64 *)q; + + dest0 = in_over (vsrc, vsrca, load8888 (m0), + expand8888 (vdest, 0)); + dest1 = in_over (vsrc, vsrca, load8888 (m1), + expand8888 (vdest, 1)); + + *(__m64 *)q = pack8888 (dest0, dest1); + } + + p += 2; + q += 2; + twidth -= 2; + } + + while (twidth) + { + uint32_t m = *(uint32_t *)p; + + if (m) + { + __m64 vdest = load8888 (*q); + vdest = in_over (vsrc, vsrca, load8888 (m), vdest); + *q = store8888 (vdest); + } + + twidth--; + p++; + q++; + } + + dst_line += dst_stride; + mask_line += mask_stride; + } + + _mm_empty (); +} + +static void +mmx_composite_over_8888_n_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + uint32_t mask; + __m64 vmask; + int dst_stride, src_stride; + int32_t w; + __m64 srca; + + CHECKPOINT (); + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + mask = _pixman_image_get_solid (mask_image, dst_image->bits.format); + mask &= 0xff000000; + mask = mask | mask >> 8 | mask >> 16 | mask >> 24; + vmask = load8888 (mask); + srca = MC (4x00ff); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w && (unsigned long)dst & 7) + { + __m64 s = load8888 (*src); + __m64 d = load8888 (*dst); + + *dst = store8888 (in_over (s, expand_alpha (s), vmask, d)); + + w--; + dst++; + src++; + } + + while (w >= 2) + { + __m64 vs = *(__m64 *)src; + __m64 vd = *(__m64 *)dst; + __m64 vsrc0 = expand8888 (vs, 0); + __m64 vsrc1 = expand8888 (vs, 1); + + *(__m64 *)dst = pack8888 ( + in_over (vsrc0, expand_alpha (vsrc0), vmask, expand8888 (vd, 0)), + in_over (vsrc1, expand_alpha (vsrc1), vmask, expand8888 (vd, 1))); + + w -= 2; + dst += 2; + src += 2; + } + + while (w) + { + __m64 s = load8888 (*src); + __m64 d = load8888 (*dst); + + *dst = store8888 (in_over (s, expand_alpha (s), vmask, d)); + + w--; + dst++; + src++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_x888_n_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + uint32_t mask; + __m64 vmask; + int dst_stride, src_stride; + int32_t w; + __m64 srca; + + CHECKPOINT (); + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + mask = _pixman_image_get_solid (mask_image, dst_image->bits.format); + + mask &= 0xff000000; + mask = mask | mask >> 8 | mask >> 16 | mask >> 24; + vmask = load8888 (mask); + srca = MC (4x00ff); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w && (unsigned long)dst & 7) + { + __m64 s = load8888 (*src | 0xff000000); + __m64 d = load8888 (*dst); + + *dst = store8888 (in_over (s, srca, vmask, d)); + + w--; + dst++; + src++; + } + + while (w >= 16) + { + __m64 vd0 = *(__m64 *)(dst + 0); + __m64 vd1 = *(__m64 *)(dst + 2); + __m64 vd2 = *(__m64 *)(dst + 4); + __m64 vd3 = *(__m64 *)(dst + 6); + __m64 vd4 = *(__m64 *)(dst + 8); + __m64 vd5 = *(__m64 *)(dst + 10); + __m64 vd6 = *(__m64 *)(dst + 12); + __m64 vd7 = *(__m64 *)(dst + 14); + + __m64 vs0 = *(__m64 *)(src + 0); + __m64 vs1 = *(__m64 *)(src + 2); + __m64 vs2 = *(__m64 *)(src + 4); + __m64 vs3 = *(__m64 *)(src + 6); + __m64 vs4 = *(__m64 *)(src + 8); + __m64 vs5 = *(__m64 *)(src + 10); + __m64 vs6 = *(__m64 *)(src + 12); + __m64 vs7 = *(__m64 *)(src + 14); + + vd0 = pack8888 ( + in_over (expandx888 (vs0, 0), srca, vmask, expand8888 (vd0, 0)), + in_over (expandx888 (vs0, 1), srca, vmask, expand8888 (vd0, 1))); + + vd1 = pack8888 ( + in_over (expandx888 (vs1, 0), srca, vmask, expand8888 (vd1, 0)), + in_over (expandx888 (vs1, 1), srca, vmask, expand8888 (vd1, 1))); + + vd2 = pack8888 ( + in_over (expandx888 (vs2, 0), srca, vmask, expand8888 (vd2, 0)), + in_over (expandx888 (vs2, 1), srca, vmask, expand8888 (vd2, 1))); + + vd3 = pack8888 ( + in_over (expandx888 (vs3, 0), srca, vmask, expand8888 (vd3, 0)), + in_over (expandx888 (vs3, 1), srca, vmask, expand8888 (vd3, 1))); + + vd4 = pack8888 ( + in_over (expandx888 (vs4, 0), srca, vmask, expand8888 (vd4, 0)), + in_over (expandx888 (vs4, 1), srca, vmask, expand8888 (vd4, 1))); + + vd5 = pack8888 ( + in_over (expandx888 (vs5, 0), srca, vmask, expand8888 (vd5, 0)), + in_over (expandx888 (vs5, 1), srca, vmask, expand8888 (vd5, 1))); + + vd6 = pack8888 ( + in_over (expandx888 (vs6, 0), srca, vmask, expand8888 (vd6, 0)), + in_over (expandx888 (vs6, 1), srca, vmask, expand8888 (vd6, 1))); + + vd7 = pack8888 ( + in_over (expandx888 (vs7, 0), srca, vmask, expand8888 (vd7, 0)), + in_over (expandx888 (vs7, 1), srca, vmask, expand8888 (vd7, 1))); + + *(__m64 *)(dst + 0) = vd0; + *(__m64 *)(dst + 2) = vd1; + *(__m64 *)(dst + 4) = vd2; + *(__m64 *)(dst + 6) = vd3; + *(__m64 *)(dst + 8) = vd4; + *(__m64 *)(dst + 10) = vd5; + *(__m64 *)(dst + 12) = vd6; + *(__m64 *)(dst + 14) = vd7; + + w -= 16; + dst += 16; + src += 16; + } + + while (w) + { + __m64 s = load8888 (*src | 0xff000000); + __m64 d = load8888 (*dst); + + *dst = store8888 (in_over (s, srca, vmask, d)); + + w--; + dst++; + src++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_8888_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + uint32_t s; + int dst_stride, src_stride; + uint8_t a; + int32_t w; + + CHECKPOINT (); + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w--) + { + s = *src++; + a = s >> 24; + + if (a == 0xff) + { + *dst = s; + } + else if (s) + { + __m64 ms, sa; + ms = load8888 (s); + sa = expand_alpha (ms); + *dst = store8888 (over (ms, sa, load8888 (*dst))); + } + + dst++; + } + } + _mm_empty (); +} + +static void +mmx_composite_over_8888_0565 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint16_t *dst_line, *dst; + uint32_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + + CHECKPOINT (); + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + +#if 0 + /* FIXME */ + assert (src_image->drawable == mask_image->drawable); +#endif + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + CHECKPOINT (); + + while (w && (unsigned long)dst & 7) + { + __m64 vsrc = load8888 (*src); + uint64_t d = *dst; + __m64 vdest = expand565 (to_m64 (d), 0); + + vdest = pack_565 ( + over (vsrc, expand_alpha (vsrc), vdest), vdest, 0); + + *dst = to_uint64 (vdest); + + w--; + dst++; + src++; + } + + CHECKPOINT (); + + while (w >= 4) + { + __m64 vsrc0, vsrc1, vsrc2, vsrc3; + __m64 vdest; + + vsrc0 = load8888 (*(src + 0)); + vsrc1 = load8888 (*(src + 1)); + vsrc2 = load8888 (*(src + 2)); + vsrc3 = load8888 (*(src + 3)); + + vdest = *(__m64 *)dst; + + vdest = pack_565 (over (vsrc0, expand_alpha (vsrc0), expand565 (vdest, 0)), vdest, 0); + vdest = pack_565 (over (vsrc1, expand_alpha (vsrc1), expand565 (vdest, 1)), vdest, 1); + vdest = pack_565 (over (vsrc2, expand_alpha (vsrc2), expand565 (vdest, 2)), vdest, 2); + vdest = pack_565 (over (vsrc3, expand_alpha (vsrc3), expand565 (vdest, 3)), vdest, 3); + + *(__m64 *)dst = vdest; + + w -= 4; + dst += 4; + src += 4; + } + + CHECKPOINT (); + + while (w) + { + __m64 vsrc = load8888 (*src); + uint64_t d = *dst; + __m64 vdest = expand565 (to_m64 (d), 0); + + vdest = pack_565 (over (vsrc, expand_alpha (vsrc), vdest), vdest, 0); + + *dst = to_uint64 (vdest); + + w--; + dst++; + src++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_n_8_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint32_t *dst_line, *dst; + uint8_t *mask_line, *mask; + int dst_stride, mask_stride; + int32_t w; + __m64 vsrc, vsrca; + uint64_t srcsrc; + + CHECKPOINT (); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + srcsrc = (uint64_t)src << 32 | src; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + CHECKPOINT (); + + while (w && (unsigned long)dst & 7) + { + uint64_t m = *mask; + + if (m) + { + __m64 vdest = in_over (vsrc, vsrca, + expand_alpha_rev (to_m64 (m)), + load8888 (*dst)); + + *dst = store8888 (vdest); + } + + w--; + mask++; + dst++; + } + + CHECKPOINT (); + + while (w >= 2) + { + uint64_t m0, m1; + + m0 = *mask; + m1 = *(mask + 1); + + if (srca == 0xff && (m0 & m1) == 0xff) + { + *(uint64_t *)dst = srcsrc; + } + else if (m0 | m1) + { + __m64 vdest; + __m64 dest0, dest1; + + vdest = *(__m64 *)dst; + + dest0 = in_over (vsrc, vsrca, expand_alpha_rev (to_m64 (m0)), + expand8888 (vdest, 0)); + dest1 = in_over (vsrc, vsrca, expand_alpha_rev (to_m64 (m1)), + expand8888 (vdest, 1)); + + *(__m64 *)dst = pack8888 (dest0, dest1); + } + + mask += 2; + dst += 2; + w -= 2; + } + + CHECKPOINT (); + + while (w) + { + uint64_t m = *mask; + + if (m) + { + __m64 vdest = load8888 (*dst); + + vdest = in_over ( + vsrc, vsrca, expand_alpha_rev (to_m64 (m)), vdest); + *dst = store8888 (vdest); + } + + w--; + mask++; + dst++; + } + } + + _mm_empty (); +} + +pixman_bool_t +pixman_fill_mmx (uint32_t *bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + uint64_t fill; + __m64 vfill; + uint32_t byte_width; + uint8_t *byte_line; + +#ifdef __GNUC__ + __m64 v1, v2, v3, v4, v5, v6, v7; +#endif + + if (bpp != 16 && bpp != 32 && bpp != 8) + return FALSE; + + if (bpp == 8) + { + stride = stride * (int) sizeof (uint32_t) / 1; + byte_line = (uint8_t *)(((uint8_t *)bits) + stride * y + x); + byte_width = width; + stride *= 1; + xor = (xor & 0xff) * 0x01010101; + } + else if (bpp == 16) + { + stride = stride * (int) sizeof (uint32_t) / 2; + byte_line = (uint8_t *)(((uint16_t *)bits) + stride * y + x); + byte_width = 2 * width; + stride *= 2; + xor = (xor & 0xffff) * 0x00010001; + } + else + { + stride = stride * (int) sizeof (uint32_t) / 4; + byte_line = (uint8_t *)(((uint32_t *)bits) + stride * y + x); + byte_width = 4 * width; + stride *= 4; + } + + fill = ((uint64_t)xor << 32) | xor; + vfill = to_m64 (fill); + +#ifdef __GNUC__ + __asm__ ( + "movq %7, %0\n" + "movq %7, %1\n" + "movq %7, %2\n" + "movq %7, %3\n" + "movq %7, %4\n" + "movq %7, %5\n" + "movq %7, %6\n" + : "=&y" (v1), "=&y" (v2), "=&y" (v3), + "=&y" (v4), "=&y" (v5), "=&y" (v6), "=y" (v7) + : "y" (vfill)); +#endif + + while (height--) + { + int w; + uint8_t *d = byte_line; + + byte_line += stride; + w = byte_width; + + while (w >= 1 && ((unsigned long)d & 1)) + { + *(uint8_t *)d = (xor & 0xff); + w--; + d++; + } + + while (w >= 2 && ((unsigned long)d & 3)) + { + *(uint16_t *)d = xor; + w -= 2; + d += 2; + } + + while (w >= 4 && ((unsigned long)d & 7)) + { + *(uint32_t *)d = xor; + + w -= 4; + d += 4; + } + + while (w >= 64) + { +#ifdef __GNUC__ + __asm__ ( + "movq %1, (%0)\n" + "movq %2, 8(%0)\n" + "movq %3, 16(%0)\n" + "movq %4, 24(%0)\n" + "movq %5, 32(%0)\n" + "movq %6, 40(%0)\n" + "movq %7, 48(%0)\n" + "movq %8, 56(%0)\n" + : + : "r" (d), + "y" (vfill), "y" (v1), "y" (v2), "y" (v3), + "y" (v4), "y" (v5), "y" (v6), "y" (v7) + : "memory"); +#else + *(__m64*) (d + 0) = vfill; + *(__m64*) (d + 8) = vfill; + *(__m64*) (d + 16) = vfill; + *(__m64*) (d + 24) = vfill; + *(__m64*) (d + 32) = vfill; + *(__m64*) (d + 40) = vfill; + *(__m64*) (d + 48) = vfill; + *(__m64*) (d + 56) = vfill; +#endif + w -= 64; + d += 64; + } + + while (w >= 4) + { + *(uint32_t *)d = xor; + + w -= 4; + d += 4; + } + while (w >= 2) + { + *(uint16_t *)d = xor; + w -= 2; + d += 2; + } + while (w >= 1) + { + *(uint8_t *)d = (xor & 0xff); + w--; + d++; + } + + } + + _mm_empty (); + return TRUE; +} + +static void +mmx_composite_src_n_8_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint32_t *dst_line, *dst; + uint8_t *mask_line, *mask; + int dst_stride, mask_stride; + int32_t w; + __m64 vsrc, vsrca; + uint64_t srcsrc; + + CHECKPOINT (); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + { + pixman_fill_mmx (dst_image->bits.bits, dst_image->bits.rowstride, + PIXMAN_FORMAT_BPP (dst_image->bits.format), + dest_x, dest_y, width, height, 0); + return; + } + + srcsrc = (uint64_t)src << 32 | src; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + CHECKPOINT (); + + while (w && (unsigned long)dst & 7) + { + uint64_t m = *mask; + + if (m) + { + __m64 vdest = in (vsrc, expand_alpha_rev (to_m64 (m))); + + *dst = store8888 (vdest); + } + else + { + *dst = 0; + } + + w--; + mask++; + dst++; + } + + CHECKPOINT (); + + while (w >= 2) + { + uint64_t m0, m1; + m0 = *mask; + m1 = *(mask + 1); + + if (srca == 0xff && (m0 & m1) == 0xff) + { + *(uint64_t *)dst = srcsrc; + } + else if (m0 | m1) + { + __m64 vdest; + __m64 dest0, dest1; + + vdest = *(__m64 *)dst; + + dest0 = in (vsrc, expand_alpha_rev (to_m64 (m0))); + dest1 = in (vsrc, expand_alpha_rev (to_m64 (m1))); + + *(__m64 *)dst = pack8888 (dest0, dest1); + } + else + { + *(uint64_t *)dst = 0; + } + + mask += 2; + dst += 2; + w -= 2; + } + + CHECKPOINT (); + + while (w) + { + uint64_t m = *mask; + + if (m) + { + __m64 vdest = load8888 (*dst); + + vdest = in (vsrc, expand_alpha_rev (to_m64 (m))); + *dst = store8888 (vdest); + } + else + { + *dst = 0; + } + + w--; + mask++; + dst++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_n_8_0565 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint16_t *dst_line, *dst; + uint8_t *mask_line, *mask; + int dst_stride, mask_stride; + int32_t w; + __m64 vsrc, vsrca, tmp; + uint64_t srcsrcsrcsrc, src16; + + CHECKPOINT (); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + tmp = pack_565 (vsrc, _mm_setzero_si64 (), 0); + src16 = to_uint64 (tmp); + + srcsrcsrcsrc = + (uint64_t)src16 << 48 | (uint64_t)src16 << 32 | + (uint64_t)src16 << 16 | (uint64_t)src16; + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + CHECKPOINT (); + + while (w && (unsigned long)dst & 7) + { + uint64_t m = *mask; + + if (m) + { + uint64_t d = *dst; + __m64 vd = to_m64 (d); + __m64 vdest = in_over ( + vsrc, vsrca, expand_alpha_rev (to_m64 (m)), expand565 (vd, 0)); + + vd = pack_565 (vdest, _mm_setzero_si64 (), 0); + *dst = to_uint64 (vd); + } + + w--; + mask++; + dst++; + } + + CHECKPOINT (); + + while (w >= 4) + { + uint64_t m0, m1, m2, m3; + m0 = *mask; + m1 = *(mask + 1); + m2 = *(mask + 2); + m3 = *(mask + 3); + + if (srca == 0xff && (m0 & m1 & m2 & m3) == 0xff) + { + *(uint64_t *)dst = srcsrcsrcsrc; + } + else if (m0 | m1 | m2 | m3) + { + __m64 vdest; + __m64 vm0, vm1, vm2, vm3; + + vdest = *(__m64 *)dst; + + vm0 = to_m64 (m0); + vdest = pack_565 (in_over (vsrc, vsrca, expand_alpha_rev (vm0), + expand565 (vdest, 0)), vdest, 0); + vm1 = to_m64 (m1); + vdest = pack_565 (in_over (vsrc, vsrca, expand_alpha_rev (vm1), + expand565 (vdest, 1)), vdest, 1); + vm2 = to_m64 (m2); + vdest = pack_565 (in_over (vsrc, vsrca, expand_alpha_rev (vm2), + expand565 (vdest, 2)), vdest, 2); + vm3 = to_m64 (m3); + vdest = pack_565 (in_over (vsrc, vsrca, expand_alpha_rev (vm3), + expand565 (vdest, 3)), vdest, 3); + + *(__m64 *)dst = vdest; + } + + w -= 4; + mask += 4; + dst += 4; + } + + CHECKPOINT (); + + while (w) + { + uint64_t m = *mask; + + if (m) + { + uint64_t d = *dst; + __m64 vd = to_m64 (d); + __m64 vdest = in_over (vsrc, vsrca, expand_alpha_rev (to_m64 (m)), + expand565 (vd, 0)); + vd = pack_565 (vdest, _mm_setzero_si64 (), 0); + *dst = to_uint64 (vd); + } + + w--; + mask++; + dst++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_pixbuf_0565 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint16_t *dst_line, *dst; + uint32_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + + CHECKPOINT (); + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + +#if 0 + /* FIXME */ + assert (src_image->drawable == mask_image->drawable); +#endif + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + CHECKPOINT (); + + while (w && (unsigned long)dst & 7) + { + __m64 vsrc = load8888 (*src); + uint64_t d = *dst; + __m64 vdest = expand565 (to_m64 (d), 0); + + vdest = pack_565 (over_rev_non_pre (vsrc, vdest), vdest, 0); + + *dst = to_uint64 (vdest); + + w--; + dst++; + src++; + } + + CHECKPOINT (); + + while (w >= 4) + { + uint32_t s0, s1, s2, s3; + unsigned char a0, a1, a2, a3; + + s0 = *src; + s1 = *(src + 1); + s2 = *(src + 2); + s3 = *(src + 3); + + a0 = (s0 >> 24); + a1 = (s1 >> 24); + a2 = (s2 >> 24); + a3 = (s3 >> 24); + + if ((a0 & a1 & a2 & a3) == 0xFF) + { + __m64 vdest; + vdest = pack_565 (invert_colors (load8888 (s0)), _mm_setzero_si64 (), 0); + vdest = pack_565 (invert_colors (load8888 (s1)), vdest, 1); + vdest = pack_565 (invert_colors (load8888 (s2)), vdest, 2); + vdest = pack_565 (invert_colors (load8888 (s3)), vdest, 3); + + *(__m64 *)dst = vdest; + } + else if (s0 | s1 | s2 | s3) + { + __m64 vdest = *(__m64 *)dst; + + vdest = pack_565 (over_rev_non_pre (load8888 (s0), expand565 (vdest, 0)), vdest, 0); + vdest = pack_565 (over_rev_non_pre (load8888 (s1), expand565 (vdest, 1)), vdest, 1); + vdest = pack_565 (over_rev_non_pre (load8888 (s2), expand565 (vdest, 2)), vdest, 2); + vdest = pack_565 (over_rev_non_pre (load8888 (s3), expand565 (vdest, 3)), vdest, 3); + + *(__m64 *)dst = vdest; + } + + w -= 4; + dst += 4; + src += 4; + } + + CHECKPOINT (); + + while (w) + { + __m64 vsrc = load8888 (*src); + uint64_t d = *dst; + __m64 vdest = expand565 (to_m64 (d), 0); + + vdest = pack_565 (over_rev_non_pre (vsrc, vdest), vdest, 0); + + *dst = to_uint64 (vdest); + + w--; + dst++; + src++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_pixbuf_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + + CHECKPOINT (); + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + +#if 0 + /* FIXME */ + assert (src_image->drawable == mask_image->drawable); +#endif + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w && (unsigned long)dst & 7) + { + __m64 s = load8888 (*src); + __m64 d = load8888 (*dst); + + *dst = store8888 (over_rev_non_pre (s, d)); + + w--; + dst++; + src++; + } + + while (w >= 2) + { + uint64_t s0, s1; + unsigned char a0, a1; + __m64 d0, d1; + + s0 = *src; + s1 = *(src + 1); + + a0 = (s0 >> 24); + a1 = (s1 >> 24); + + if ((a0 & a1) == 0xFF) + { + d0 = invert_colors (load8888 (s0)); + d1 = invert_colors (load8888 (s1)); + + *(__m64 *)dst = pack8888 (d0, d1); + } + else if (s0 | s1) + { + __m64 vdest = *(__m64 *)dst; + + d0 = over_rev_non_pre (load8888 (s0), expand8888 (vdest, 0)); + d1 = over_rev_non_pre (load8888 (s1), expand8888 (vdest, 1)); + + *(__m64 *)dst = pack8888 (d0, d1); + } + + w -= 2; + dst += 2; + src += 2; + } + + while (w) + { + __m64 s = load8888 (*src); + __m64 d = load8888 (*dst); + + *dst = store8888 (over_rev_non_pre (s, d)); + + w--; + dst++; + src++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_over_n_8888_0565_ca (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t src, srca; + uint16_t *dst_line; + uint32_t *mask_line; + int dst_stride, mask_stride; + __m64 vsrc, vsrca; + + CHECKPOINT (); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + srca = src >> 24; + if (src == 0) + return; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1); + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + while (height--) + { + int twidth = width; + uint32_t *p = (uint32_t *)mask_line; + uint16_t *q = (uint16_t *)dst_line; + + while (twidth && ((unsigned long)q & 7)) + { + uint32_t m = *(uint32_t *)p; + + if (m) + { + uint64_t d = *q; + __m64 vdest = expand565 (to_m64 (d), 0); + vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m), vdest), vdest, 0); + *q = to_uint64 (vdest); + } + + twidth--; + p++; + q++; + } + + while (twidth >= 4) + { + uint32_t m0, m1, m2, m3; + + m0 = *p; + m1 = *(p + 1); + m2 = *(p + 2); + m3 = *(p + 3); + + if ((m0 | m1 | m2 | m3)) + { + __m64 vdest = *(__m64 *)q; + + vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m0), expand565 (vdest, 0)), vdest, 0); + vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m1), expand565 (vdest, 1)), vdest, 1); + vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m2), expand565 (vdest, 2)), vdest, 2); + vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m3), expand565 (vdest, 3)), vdest, 3); + + *(__m64 *)q = vdest; + } + twidth -= 4; + p += 4; + q += 4; + } + + while (twidth) + { + uint32_t m; + + m = *(uint32_t *)p; + if (m) + { + uint64_t d = *q; + __m64 vdest = expand565 (to_m64 (d), 0); + vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m), vdest), vdest, 0); + *q = to_uint64 (vdest); + } + + twidth--; + p++; + q++; + } + + mask_line += mask_stride; + dst_line += dst_stride; + } + + _mm_empty (); +} + +static void +mmx_composite_in_n_8_8 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint8_t *dst_line, *dst; + uint8_t *mask_line, *mask; + int dst_stride, mask_stride; + int32_t w; + uint32_t src; + uint8_t sa; + __m64 vsrc, vsrca; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + sa = src >> 24; + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + if ((((unsigned long)dst_image & 3) == 0) && + (((unsigned long)src_image & 3) == 0)) + { + while (w >= 4) + { + uint32_t m; + __m64 vmask; + __m64 vdest; + + m = 0; + + vmask = load8888 (*(uint32_t *)mask); + vdest = load8888 (*(uint32_t *)dst); + + *(uint32_t *)dst = store8888 (in (in (vsrca, vmask), vdest)); + + dst += 4; + mask += 4; + w -= 4; + } + } + + while (w--) + { + uint16_t tmp; + uint8_t a; + uint32_t m, d; + + a = *mask++; + d = *dst; + + m = MUL_UN8 (sa, a, tmp); + d = MUL_UN8 (m, d, tmp); + + *dst++ = d; + } + } + + _mm_empty (); +} + +static void +mmx_composite_in_8_8 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint8_t *dst_line, *dst; + uint8_t *src_line, *src; + int src_stride, dst_stride; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint8_t, src_stride, src_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + if ((((unsigned long)dst_image & 3) == 0) && + (((unsigned long)src_image & 3) == 0)) + { + while (w >= 4) + { + uint32_t *s = (uint32_t *)src; + uint32_t *d = (uint32_t *)dst; + + *d = store8888 (in (load8888 (*s), load8888 (*d))); + + w -= 4; + dst += 4; + src += 4; + } + } + + while (w--) + { + uint8_t s, d; + uint16_t tmp; + + s = *src; + d = *dst; + + *dst = MUL_UN8 (s, d, tmp); + + src++; + dst++; + } + } + + _mm_empty (); +} + +static void +mmx_composite_add_n_8_8 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint8_t *dst_line, *dst; + uint8_t *mask_line, *mask; + int dst_stride, mask_stride; + int32_t w; + uint32_t src; + uint8_t sa; + __m64 vsrc, vsrca; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + + src = _pixman_image_get_solid (src_image, dst_image->bits.format); + + sa = src >> 24; + + if (src == 0) + return; + + vsrc = load8888 (src); + vsrca = expand_alpha (vsrc); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + w = width; + + if ((((unsigned long)mask_image & 3) == 0) && + (((unsigned long)dst_image & 3) == 0)) + { + while (w >= 4) + { + __m64 vmask = load8888 (*(uint32_t *)mask); + __m64 vdest = load8888 (*(uint32_t *)dst); + + *(uint32_t *)dst = store8888 (_mm_adds_pu8 (in (vsrca, vmask), vdest)); + + w -= 4; + dst += 4; + mask += 4; + } + } + + while (w--) + { + uint16_t tmp; + uint16_t a; + uint32_t m, d; + uint32_t r; + + a = *mask++; + d = *dst; + + m = MUL_UN8 (sa, a, tmp); + r = ADD_UN8 (m, d, tmp); + + *dst++ = r; + } + } + + _mm_empty (); +} + +static void +mmx_composite_add_8_8 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint8_t *dst_line, *dst; + uint8_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + uint8_t s, d; + uint16_t t; + + CHECKPOINT (); + + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint8_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint8_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w && (unsigned long)dst & 7) + { + s = *src; + d = *dst; + t = d + s; + s = t | (0 - (t >> 8)); + *dst = s; + + dst++; + src++; + w--; + } + + while (w >= 8) + { + *(__m64*)dst = _mm_adds_pu8 (*(__m64*)src, *(__m64*)dst); + dst += 8; + src += 8; + w -= 8; + } + + while (w) + { + s = *src; + d = *dst; + t = d + s; + s = t | (0 - (t >> 8)); + *dst = s; + + dst++; + src++; + w--; + } + } + + _mm_empty (); +} + +static void +mmx_composite_add_8888_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + __m64 dst64; + uint32_t *dst_line, *dst; + uint32_t *src_line, *src; + int dst_stride, src_stride; + int32_t w; + + CHECKPOINT (); + + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + + while (height--) + { + dst = dst_line; + dst_line += dst_stride; + src = src_line; + src_line += src_stride; + w = width; + + while (w && (unsigned long)dst & 7) + { + *dst = _mm_cvtsi64_si32 (_mm_adds_pu8 (_mm_cvtsi32_si64 (*src), + _mm_cvtsi32_si64 (*dst))); + dst++; + src++; + w--; + } + + while (w >= 2) + { + dst64 = _mm_adds_pu8 (*(__m64*)src, *(__m64*)dst); + *(uint64_t*)dst = to_uint64 (dst64); + dst += 2; + src += 2; + w -= 2; + } + + if (w) + { + *dst = _mm_cvtsi64_si32 (_mm_adds_pu8 (_mm_cvtsi32_si64 (*src), + _mm_cvtsi32_si64 (*dst))); + + } + } + + _mm_empty (); +} + +static pixman_bool_t +pixman_blt_mmx (uint32_t *src_bits, + uint32_t *dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height) +{ + uint8_t * src_bytes; + uint8_t * dst_bytes; + int byte_width; + + if (src_bpp != dst_bpp) + return FALSE; + + if (src_bpp == 16) + { + src_stride = src_stride * (int) sizeof (uint32_t) / 2; + dst_stride = dst_stride * (int) sizeof (uint32_t) / 2; + src_bytes = (uint8_t *)(((uint16_t *)src_bits) + src_stride * (src_y) + (src_x)); + dst_bytes = (uint8_t *)(((uint16_t *)dst_bits) + dst_stride * (dst_y) + (dst_x)); + byte_width = 2 * width; + src_stride *= 2; + dst_stride *= 2; + } + else if (src_bpp == 32) + { + src_stride = src_stride * (int) sizeof (uint32_t) / 4; + dst_stride = dst_stride * (int) sizeof (uint32_t) / 4; + src_bytes = (uint8_t *)(((uint32_t *)src_bits) + src_stride * (src_y) + (src_x)); + dst_bytes = (uint8_t *)(((uint32_t *)dst_bits) + dst_stride * (dst_y) + (dst_x)); + byte_width = 4 * width; + src_stride *= 4; + dst_stride *= 4; + } + else + { + return FALSE; + } + + while (height--) + { + int w; + uint8_t *s = src_bytes; + uint8_t *d = dst_bytes; + src_bytes += src_stride; + dst_bytes += dst_stride; + w = byte_width; + + while (w >= 2 && ((unsigned long)d & 3)) + { + *(uint16_t *)d = *(uint16_t *)s; + w -= 2; + s += 2; + d += 2; + } + + while (w >= 4 && ((unsigned long)d & 7)) + { + *(uint32_t *)d = *(uint32_t *)s; + + w -= 4; + s += 4; + d += 4; + } + + while (w >= 64) + { +#if defined (__GNUC__) || (defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590)) + __asm__ ( + "movq (%1), %%mm0\n" + "movq 8(%1), %%mm1\n" + "movq 16(%1), %%mm2\n" + "movq 24(%1), %%mm3\n" + "movq 32(%1), %%mm4\n" + "movq 40(%1), %%mm5\n" + "movq 48(%1), %%mm6\n" + "movq 56(%1), %%mm7\n" + + "movq %%mm0, (%0)\n" + "movq %%mm1, 8(%0)\n" + "movq %%mm2, 16(%0)\n" + "movq %%mm3, 24(%0)\n" + "movq %%mm4, 32(%0)\n" + "movq %%mm5, 40(%0)\n" + "movq %%mm6, 48(%0)\n" + "movq %%mm7, 56(%0)\n" + : + : "r" (d), "r" (s) + : "memory", + "%mm0", "%mm1", "%mm2", "%mm3", + "%mm4", "%mm5", "%mm6", "%mm7"); +#else + __m64 v0 = *(__m64 *)(s + 0); + __m64 v1 = *(__m64 *)(s + 8); + __m64 v2 = *(__m64 *)(s + 16); + __m64 v3 = *(__m64 *)(s + 24); + __m64 v4 = *(__m64 *)(s + 32); + __m64 v5 = *(__m64 *)(s + 40); + __m64 v6 = *(__m64 *)(s + 48); + __m64 v7 = *(__m64 *)(s + 56); + *(__m64 *)(d + 0) = v0; + *(__m64 *)(d + 8) = v1; + *(__m64 *)(d + 16) = v2; + *(__m64 *)(d + 24) = v3; + *(__m64 *)(d + 32) = v4; + *(__m64 *)(d + 40) = v5; + *(__m64 *)(d + 48) = v6; + *(__m64 *)(d + 56) = v7; +#endif + + w -= 64; + s += 64; + d += 64; + } + while (w >= 4) + { + *(uint32_t *)d = *(uint32_t *)s; + + w -= 4; + s += 4; + d += 4; + } + if (w >= 2) + { + *(uint16_t *)d = *(uint16_t *)s; + w -= 2; + s += 2; + d += 2; + } + } + + _mm_empty (); + + return TRUE; +} + +static void +mmx_composite_copy_area (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + pixman_blt_mmx (src_image->bits.bits, + dst_image->bits.bits, + src_image->bits.rowstride, + dst_image->bits.rowstride, + PIXMAN_FORMAT_BPP (src_image->bits.format), + PIXMAN_FORMAT_BPP (dst_image->bits.format), + src_x, src_y, dest_x, dest_y, width, height); +} + +#if 0 +static void +mmx_composite_over_x888_8_8888 (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + uint32_t *src, *src_line; + uint32_t *dst, *dst_line; + uint8_t *mask, *mask_line; + int src_stride, mask_stride, dst_stride; + int32_t w; + + PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1); + PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1); + PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint32_t, src_stride, src_line, 1); + + while (height--) + { + src = src_line; + src_line += src_stride; + dst = dst_line; + dst_line += dst_stride; + mask = mask_line; + mask_line += mask_stride; + + w = width; + + while (w--) + { + uint64_t m = *mask; + + if (m) + { + __m64 s = load8888 (*src | 0xff000000); + + if (m == 0xff) + { + *dst = store8888 (s); + } + else + { + __m64 sa = expand_alpha (s); + __m64 vm = expand_alpha_rev (to_m64 (m)); + __m64 vdest = in_over (s, sa, vm, load8888 (*dst)); + + *dst = store8888 (vdest); + } + } + + mask++; + dst++; + src++; + } + } + + _mm_empty (); +} +#endif + +static const pixman_fast_path_t mmx_fast_paths[] = +{ + PIXMAN_STD_FAST_PATH (OVER, solid, a8, r5g6b5, mmx_composite_over_n_8_0565 ), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, b5g6r5, mmx_composite_over_n_8_0565 ), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8r8g8b8, mmx_composite_over_n_8_8888 ), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8r8g8b8, mmx_composite_over_n_8_8888 ), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8b8g8r8, mmx_composite_over_n_8_8888 ), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8b8g8r8, mmx_composite_over_n_8_8888 ), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, a8r8g8b8, mmx_composite_over_n_8888_8888_ca ), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, x8r8g8b8, mmx_composite_over_n_8888_8888_ca ), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, r5g6b5, mmx_composite_over_n_8888_0565_ca ), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, a8b8g8r8, mmx_composite_over_n_8888_8888_ca ), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, x8b8g8r8, mmx_composite_over_n_8888_8888_ca ), + PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, b5g6r5, mmx_composite_over_n_8888_0565_ca ), + PIXMAN_STD_FAST_PATH (OVER, pixbuf, pixbuf, a8r8g8b8, mmx_composite_over_pixbuf_8888 ), + PIXMAN_STD_FAST_PATH (OVER, pixbuf, pixbuf, x8r8g8b8, mmx_composite_over_pixbuf_8888 ), + PIXMAN_STD_FAST_PATH (OVER, pixbuf, pixbuf, r5g6b5, mmx_composite_over_pixbuf_0565 ), + PIXMAN_STD_FAST_PATH (OVER, rpixbuf, rpixbuf, a8b8g8r8, mmx_composite_over_pixbuf_8888 ), + PIXMAN_STD_FAST_PATH (OVER, rpixbuf, rpixbuf, x8b8g8r8, mmx_composite_over_pixbuf_8888 ), + PIXMAN_STD_FAST_PATH (OVER, rpixbuf, rpixbuf, b5g6r5, mmx_composite_over_pixbuf_0565 ), + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, solid, a8r8g8b8, mmx_composite_over_x888_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, solid, x8r8g8b8, mmx_composite_over_x888_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, solid, a8b8g8r8, mmx_composite_over_x888_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, solid, x8b8g8r8, mmx_composite_over_x888_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, solid, a8r8g8b8, mmx_composite_over_8888_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, solid, x8r8g8b8, mmx_composite_over_8888_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, solid, a8b8g8r8, mmx_composite_over_8888_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, solid, x8b8g8r8, mmx_composite_over_8888_n_8888 ), +#if 0 + /* FIXME: This code is commented out since it's apparently + * not actually faster than the generic code. + */ + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, a8, x8r8g8b8, mmx_composite_over_x888_8_8888 ), + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, a8, a8r8g8b8, mmx_composite_over_x888_8_8888 ), + PIXMAN_STD_FAST_PATH (OVER, x8b8r8g8, a8, x8b8g8r8, mmx_composite_over_x888_8_8888 ), + PIXMAN_STD_FAST_PATH (OVER, x8b8r8g8, a8, a8r8g8b8, mmx_composite_over_x888_8_8888 ), +#endif + PIXMAN_STD_FAST_PATH (OVER, solid, null, a8r8g8b8, mmx_composite_over_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, solid, null, x8r8g8b8, mmx_composite_over_n_8888 ), + PIXMAN_STD_FAST_PATH (OVER, solid, null, r5g6b5, mmx_composite_over_n_0565 ), + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, null, x8r8g8b8, mmx_composite_copy_area ), + PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, null, x8b8g8r8, mmx_composite_copy_area ), + + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, a8r8g8b8, mmx_composite_over_8888_8888 ), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, x8r8g8b8, mmx_composite_over_8888_8888 ), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, r5g6b5, mmx_composite_over_8888_0565 ), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, a8b8g8r8, mmx_composite_over_8888_8888 ), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, x8b8g8r8, mmx_composite_over_8888_8888 ), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, b5g6r5, mmx_composite_over_8888_0565 ), + + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, null, a8r8g8b8, mmx_composite_add_8888_8888 ), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, null, a8b8g8r8, mmx_composite_add_8888_8888 ), + PIXMAN_STD_FAST_PATH (ADD, a8, null, a8, mmx_composite_add_8_8 ), + PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8, mmx_composite_add_n_8_8 ), + + PIXMAN_STD_FAST_PATH (SRC, solid, a8, a8r8g8b8, mmx_composite_src_n_8_8888 ), + PIXMAN_STD_FAST_PATH (SRC, solid, a8, x8r8g8b8, mmx_composite_src_n_8_8888 ), + PIXMAN_STD_FAST_PATH (SRC, solid, a8, a8b8g8r8, mmx_composite_src_n_8_8888 ), + PIXMAN_STD_FAST_PATH (SRC, solid, a8, x8b8g8r8, mmx_composite_src_n_8_8888 ), + PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, a8r8g8b8, mmx_composite_copy_area ), + PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, a8b8g8r8, mmx_composite_copy_area ), + PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, x8r8g8b8, mmx_composite_copy_area ), + PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, x8b8g8r8, mmx_composite_copy_area ), + PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, x8r8g8b8, mmx_composite_copy_area ), + PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, x8b8g8r8, mmx_composite_copy_area ), + PIXMAN_STD_FAST_PATH (SRC, r5g6b5, null, r5g6b5, mmx_composite_copy_area ), + PIXMAN_STD_FAST_PATH (SRC, b5g6r5, null, b5g6r5, mmx_composite_copy_area ), + + PIXMAN_STD_FAST_PATH (IN, a8, null, a8, mmx_composite_in_8_8 ), + PIXMAN_STD_FAST_PATH (IN, solid, a8, a8, mmx_composite_in_n_8_8 ), + + { PIXMAN_OP_NONE }, +}; + +static pixman_bool_t +mmx_blt (pixman_implementation_t *imp, + uint32_t * src_bits, + uint32_t * dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height) +{ + if (!pixman_blt_mmx ( + src_bits, dst_bits, src_stride, dst_stride, src_bpp, dst_bpp, + src_x, src_y, dst_x, dst_y, width, height)) + + { + return _pixman_implementation_blt ( + imp->delegate, + src_bits, dst_bits, src_stride, dst_stride, src_bpp, dst_bpp, + src_x, src_y, dst_x, dst_y, width, height); + } + + return TRUE; +} + +static pixman_bool_t +mmx_fill (pixman_implementation_t *imp, + uint32_t * bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + if (!pixman_fill_mmx (bits, stride, bpp, x, y, width, height, xor)) + { + return _pixman_implementation_fill ( + imp->delegate, bits, stride, bpp, x, y, width, height, xor); + } + + return TRUE; +} + +pixman_implementation_t * +_pixman_implementation_create_mmx (void) +{ + pixman_implementation_t *general = _pixman_implementation_create_fast_path (); + pixman_implementation_t *imp = _pixman_implementation_create (general, mmx_fast_paths); + + imp->combine_32[PIXMAN_OP_OVER] = mmx_combine_over_u; + imp->combine_32[PIXMAN_OP_OVER_REVERSE] = mmx_combine_over_reverse_u; + imp->combine_32[PIXMAN_OP_IN] = mmx_combine_in_u; + imp->combine_32[PIXMAN_OP_IN_REVERSE] = mmx_combine_in_reverse_u; + imp->combine_32[PIXMAN_OP_OUT] = mmx_combine_out_u; + imp->combine_32[PIXMAN_OP_OUT_REVERSE] = mmx_combine_out_reverse_u; + imp->combine_32[PIXMAN_OP_ATOP] = mmx_combine_atop_u; + imp->combine_32[PIXMAN_OP_ATOP_REVERSE] = mmx_combine_atop_reverse_u; + imp->combine_32[PIXMAN_OP_XOR] = mmx_combine_xor_u; + imp->combine_32[PIXMAN_OP_ADD] = mmx_combine_add_u; + imp->combine_32[PIXMAN_OP_SATURATE] = mmx_combine_saturate_u; + + imp->combine_32_ca[PIXMAN_OP_SRC] = mmx_combine_src_ca; + imp->combine_32_ca[PIXMAN_OP_OVER] = mmx_combine_over_ca; + imp->combine_32_ca[PIXMAN_OP_OVER_REVERSE] = mmx_combine_over_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_IN] = mmx_combine_in_ca; + imp->combine_32_ca[PIXMAN_OP_IN_REVERSE] = mmx_combine_in_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_OUT] = mmx_combine_out_ca; + imp->combine_32_ca[PIXMAN_OP_OUT_REVERSE] = mmx_combine_out_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_ATOP] = mmx_combine_atop_ca; + imp->combine_32_ca[PIXMAN_OP_ATOP_REVERSE] = mmx_combine_atop_reverse_ca; + imp->combine_32_ca[PIXMAN_OP_XOR] = mmx_combine_xor_ca; + imp->combine_32_ca[PIXMAN_OP_ADD] = mmx_combine_add_ca; + + imp->blt = mmx_blt; + imp->fill = mmx_fill; + + return imp; +} + +#endif /* USE_MMX */ diff --git a/programs/develop/libraries/pixman/pixman-private.h b/programs/develop/libraries/pixman/pixman-private.h new file mode 100644 index 0000000000..e7eaca7649 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-private.h @@ -0,0 +1,873 @@ + +#ifndef PIXMAN_PRIVATE_H +#define PIXMAN_PRIVATE_H + +#define PIXMAN_DISABLE_DEPRECATED +#define PIXMAN_USE_INTERNAL_API + +#include "pixman.h" +#include +#include +#include +#include + +#include "pixman-compiler.h" + +/* + * Images + */ +typedef struct image_common image_common_t; +typedef struct source_image source_image_t; +typedef struct solid_fill solid_fill_t; +typedef struct gradient gradient_t; +typedef struct linear_gradient linear_gradient_t; +typedef struct horizontal_gradient horizontal_gradient_t; +typedef struct vertical_gradient vertical_gradient_t; +typedef struct conical_gradient conical_gradient_t; +typedef struct radial_gradient radial_gradient_t; +typedef struct bits_image bits_image_t; +typedef struct circle circle_t; + +typedef void (*fetch_scanline_t) (pixman_image_t *image, + int x, + int y, + int width, + uint32_t *buffer, + const uint32_t *mask); + +typedef uint32_t (*fetch_pixel_32_t) (bits_image_t *image, + int x, + int y); + +typedef uint64_t (*fetch_pixel_64_t) (bits_image_t *image, + int x, + int y); + +typedef void (*store_scanline_t) (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *values); + +typedef enum +{ + BITS, + LINEAR, + CONICAL, + RADIAL, + SOLID +} image_type_t; + +typedef enum +{ + SOURCE_IMAGE_CLASS_UNKNOWN, + SOURCE_IMAGE_CLASS_HORIZONTAL, +} source_image_class_t; + +typedef source_image_class_t (*classify_func_t) (pixman_image_t *image, + int x, + int y, + int width, + int height); +typedef void (*property_changed_func_t) (pixman_image_t *image); + +struct image_common +{ + image_type_t type; + int32_t ref_count; + pixman_region32_t clip_region; + int32_t alpha_count; /* How many times this image is being used as an alpha map */ + pixman_bool_t have_clip_region; /* FALSE if there is no clip */ + pixman_bool_t client_clip; /* Whether the source clip was + set by a client */ + pixman_bool_t clip_sources; /* Whether the clip applies when + * the image is used as a source + */ + pixman_bool_t dirty; + pixman_transform_t * transform; + pixman_repeat_t repeat; + pixman_filter_t filter; + pixman_fixed_t * filter_params; + int n_filter_params; + bits_image_t * alpha_map; + int alpha_origin_x; + int alpha_origin_y; + pixman_bool_t component_alpha; + classify_func_t classify; + property_changed_func_t property_changed; + fetch_scanline_t get_scanline_32; + fetch_scanline_t get_scanline_64; + + pixman_image_destroy_func_t destroy_func; + void * destroy_data; + + uint32_t flags; + pixman_format_code_t extended_format_code; +}; + +struct source_image +{ + image_common_t common; +}; + +struct solid_fill +{ + source_image_t common; + pixman_color_t color; + + uint32_t color_32; + uint64_t color_64; +}; + +struct gradient +{ + source_image_t common; + int n_stops; + pixman_gradient_stop_t *stops; + int stop_range; +}; + +struct linear_gradient +{ + gradient_t common; + pixman_point_fixed_t p1; + pixman_point_fixed_t p2; +}; + +struct circle +{ + pixman_fixed_t x; + pixman_fixed_t y; + pixman_fixed_t radius; +}; + +struct radial_gradient +{ + gradient_t common; + + circle_t c1; + circle_t c2; + + circle_t delta; + double a; + double inva; + double mindr; +}; + +struct conical_gradient +{ + gradient_t common; + pixman_point_fixed_t center; + double angle; +}; + +struct bits_image +{ + image_common_t common; + pixman_format_code_t format; + const pixman_indexed_t * indexed; + int width; + int height; + uint32_t * bits; + uint32_t * free_me; + int rowstride; /* in number of uint32_t's */ + + fetch_scanline_t fetch_scanline_32; + fetch_pixel_32_t fetch_pixel_32; + store_scanline_t store_scanline_32; + + fetch_scanline_t fetch_scanline_64; + fetch_pixel_64_t fetch_pixel_64; + store_scanline_t store_scanline_64; + + /* Used for indirect access to the bits */ + pixman_read_memory_func_t read_func; + pixman_write_memory_func_t write_func; +}; + +union pixman_image +{ + image_type_t type; + image_common_t common; + bits_image_t bits; + source_image_t source; + gradient_t gradient; + linear_gradient_t linear; + conical_gradient_t conical; + radial_gradient_t radial; + solid_fill_t solid; +}; + +void +_pixman_bits_image_setup_accessors (bits_image_t *image); + +void +_pixman_image_get_scanline_generic_64 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask); + +source_image_class_t +_pixman_image_classify (pixman_image_t *image, + int x, + int y, + int width, + int height); + +void +_pixman_image_get_scanline_32 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask); + +/* Even thought the type of buffer is uint32_t *, the function actually expects + * a uint64_t *buffer. + */ +void +_pixman_image_get_scanline_64 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *unused); + +void +_pixman_image_store_scanline_32 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *buffer); + +/* Even though the type of buffer is uint32_t *, the function + * actually expects a uint64_t *buffer. + */ +void +_pixman_image_store_scanline_64 (bits_image_t * image, + int x, + int y, + int width, + const uint32_t *buffer); + +pixman_image_t * +_pixman_image_allocate (void); + +pixman_bool_t +_pixman_init_gradient (gradient_t * gradient, + const pixman_gradient_stop_t *stops, + int n_stops); +void +_pixman_image_reset_clip_region (pixman_image_t *image); + +void +_pixman_image_validate (pixman_image_t *image); + +uint32_t +_pixman_image_get_solid (pixman_image_t * image, + pixman_format_code_t format); + +#define PIXMAN_IMAGE_GET_LINE(image, x, y, type, out_stride, line, mul) \ + do \ + { \ + uint32_t *__bits__; \ + int __stride__; \ + \ + __bits__ = image->bits.bits; \ + __stride__ = image->bits.rowstride; \ + (out_stride) = \ + __stride__ * (int) sizeof (uint32_t) / (int) sizeof (type); \ + (line) = \ + ((type *) __bits__) + (out_stride) * (y) + (mul) * (x); \ + } while (0) + +/* + * Gradient walker + */ +typedef struct +{ + uint32_t left_ag; + uint32_t left_rb; + uint32_t right_ag; + uint32_t right_rb; + int32_t left_x; + int32_t right_x; + int32_t stepper; + + pixman_gradient_stop_t *stops; + int num_stops; + unsigned int spread; + + int need_reset; +} pixman_gradient_walker_t; + +void +_pixman_gradient_walker_init (pixman_gradient_walker_t *walker, + gradient_t * gradient, + unsigned int spread); + +void +_pixman_gradient_walker_reset (pixman_gradient_walker_t *walker, + pixman_fixed_32_32_t pos); + +uint32_t +_pixman_gradient_walker_pixel (pixman_gradient_walker_t *walker, + pixman_fixed_32_32_t x); + +/* + * Edges + */ + +#define MAX_ALPHA(n) ((1 << (n)) - 1) +#define N_Y_FRAC(n) ((n) == 1 ? 1 : (1 << ((n) / 2)) - 1) +#define N_X_FRAC(n) ((n) == 1 ? 1 : (1 << ((n) / 2)) + 1) + +#define STEP_Y_SMALL(n) (pixman_fixed_1 / N_Y_FRAC (n)) +#define STEP_Y_BIG(n) (pixman_fixed_1 - (N_Y_FRAC (n) - 1) * STEP_Y_SMALL (n)) + +#define Y_FRAC_FIRST(n) (STEP_Y_BIG (n) / 2) +#define Y_FRAC_LAST(n) (Y_FRAC_FIRST (n) + (N_Y_FRAC (n) - 1) * STEP_Y_SMALL (n)) + +#define STEP_X_SMALL(n) (pixman_fixed_1 / N_X_FRAC (n)) +#define STEP_X_BIG(n) (pixman_fixed_1 - (N_X_FRAC (n) - 1) * STEP_X_SMALL (n)) + +#define X_FRAC_FIRST(n) (STEP_X_BIG (n) / 2) +#define X_FRAC_LAST(n) (X_FRAC_FIRST (n) + (N_X_FRAC (n) - 1) * STEP_X_SMALL (n)) + +#define RENDER_SAMPLES_X(x, n) \ + ((n) == 1? 0 : (pixman_fixed_frac (x) + \ + X_FRAC_FIRST (n)) / STEP_X_SMALL (n)) + +void +pixman_rasterize_edges_accessors (pixman_image_t *image, + pixman_edge_t * l, + pixman_edge_t * r, + pixman_fixed_t t, + pixman_fixed_t b); + +/* + * Implementations + */ +typedef struct pixman_implementation_t pixman_implementation_t; + +typedef void (*pixman_combine_32_func_t) (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width); + +typedef void (*pixman_combine_64_func_t) (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width); + +typedef void (*pixman_composite_func_t) (pixman_implementation_t *imp, + pixman_op_t op, + pixman_image_t * src, + pixman_image_t * mask, + pixman_image_t * dest, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height); +typedef pixman_bool_t (*pixman_blt_func_t) (pixman_implementation_t *imp, + uint32_t * src_bits, + uint32_t * dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height); +typedef pixman_bool_t (*pixman_fill_func_t) (pixman_implementation_t *imp, + uint32_t * bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor); + +void _pixman_setup_combiner_functions_32 (pixman_implementation_t *imp); +void _pixman_setup_combiner_functions_64 (pixman_implementation_t *imp); + +typedef struct +{ + pixman_op_t op; + pixman_format_code_t src_format; + uint32_t src_flags; + pixman_format_code_t mask_format; + uint32_t mask_flags; + pixman_format_code_t dest_format; + uint32_t dest_flags; + pixman_composite_func_t func; +} pixman_fast_path_t; + +struct pixman_implementation_t +{ + pixman_implementation_t * toplevel; + pixman_implementation_t * delegate; + const pixman_fast_path_t * fast_paths; + + pixman_blt_func_t blt; + pixman_fill_func_t fill; + + pixman_combine_32_func_t combine_32[PIXMAN_N_OPERATORS]; + pixman_combine_32_func_t combine_32_ca[PIXMAN_N_OPERATORS]; + pixman_combine_64_func_t combine_64[PIXMAN_N_OPERATORS]; + pixman_combine_64_func_t combine_64_ca[PIXMAN_N_OPERATORS]; +}; + +pixman_implementation_t * +_pixman_implementation_create (pixman_implementation_t *delegate, + const pixman_fast_path_t *fast_paths); + +void +_pixman_implementation_combine_32 (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width); +void +_pixman_implementation_combine_64 (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width); +void +_pixman_implementation_combine_32_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint32_t * dest, + const uint32_t * src, + const uint32_t * mask, + int width); +void +_pixman_implementation_combine_64_ca (pixman_implementation_t *imp, + pixman_op_t op, + uint64_t * dest, + const uint64_t * src, + const uint64_t * mask, + int width); + +pixman_bool_t +_pixman_implementation_blt (pixman_implementation_t *imp, + uint32_t * src_bits, + uint32_t * dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height); + +pixman_bool_t +_pixman_implementation_fill (pixman_implementation_t *imp, + uint32_t * bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor); + +/* Specific implementations */ +pixman_implementation_t * +_pixman_implementation_create_general (void); + +pixman_implementation_t * +_pixman_implementation_create_fast_path (void); + +#ifdef USE_MMX +pixman_implementation_t * +_pixman_implementation_create_mmx (void); +#endif + +#ifdef USE_SSE2 +pixman_implementation_t * +_pixman_implementation_create_sse2 (void); +#endif + +#ifdef USE_ARM_SIMD +pixman_implementation_t * +_pixman_implementation_create_arm_simd (void); +#endif + +#ifdef USE_ARM_NEON +pixman_implementation_t * +_pixman_implementation_create_arm_neon (void); +#endif + +#ifdef USE_VMX +pixman_implementation_t * +_pixman_implementation_create_vmx (void); +#endif + +pixman_implementation_t * +_pixman_choose_implementation (void); + + + +/* + * Utilities + */ + +/* These "formats" all have depth 0, so they + * will never clash with any real ones + */ +#define PIXMAN_null PIXMAN_FORMAT (0, 0, 0, 0, 0, 0) +#define PIXMAN_solid PIXMAN_FORMAT (0, 1, 0, 0, 0, 0) +#define PIXMAN_pixbuf PIXMAN_FORMAT (0, 2, 0, 0, 0, 0) +#define PIXMAN_rpixbuf PIXMAN_FORMAT (0, 3, 0, 0, 0, 0) +#define PIXMAN_unknown PIXMAN_FORMAT (0, 4, 0, 0, 0, 0) +#define PIXMAN_any PIXMAN_FORMAT (0, 5, 0, 0, 0, 0) + +#define PIXMAN_OP_any (PIXMAN_N_OPERATORS + 1) + +#define FAST_PATH_ID_TRANSFORM (1 << 0) +#define FAST_PATH_NO_ALPHA_MAP (1 << 1) +#define FAST_PATH_NO_CONVOLUTION_FILTER (1 << 2) +#define FAST_PATH_NO_PAD_REPEAT (1 << 3) +#define FAST_PATH_NO_REFLECT_REPEAT (1 << 4) +#define FAST_PATH_NO_ACCESSORS (1 << 5) +#define FAST_PATH_NARROW_FORMAT (1 << 6) +#define FAST_PATH_COMPONENT_ALPHA (1 << 8) +#define FAST_PATH_SAMPLES_OPAQUE (1 << 7) +#define FAST_PATH_UNIFIED_ALPHA (1 << 9) +#define FAST_PATH_SCALE_TRANSFORM (1 << 10) +#define FAST_PATH_NEAREST_FILTER (1 << 11) +#define FAST_PATH_HAS_TRANSFORM (1 << 12) +#define FAST_PATH_IS_OPAQUE (1 << 13) +#define FAST_PATH_NEEDS_WORKAROUND (1 << 14) +#define FAST_PATH_NO_NONE_REPEAT (1 << 15) +#define FAST_PATH_SAMPLES_COVER_CLIP (1 << 16) +#define FAST_PATH_X_UNIT_POSITIVE (1 << 17) +#define FAST_PATH_AFFINE_TRANSFORM (1 << 18) +#define FAST_PATH_Y_UNIT_ZERO (1 << 19) +#define FAST_PATH_BILINEAR_FILTER (1 << 20) +#define FAST_PATH_NO_NORMAL_REPEAT (1 << 21) + +#define FAST_PATH_PAD_REPEAT \ + (FAST_PATH_NO_NONE_REPEAT | \ + FAST_PATH_NO_NORMAL_REPEAT | \ + FAST_PATH_NO_REFLECT_REPEAT) + +#define FAST_PATH_NORMAL_REPEAT \ + (FAST_PATH_NO_NONE_REPEAT | \ + FAST_PATH_NO_PAD_REPEAT | \ + FAST_PATH_NO_REFLECT_REPEAT) + +#define FAST_PATH_NONE_REPEAT \ + (FAST_PATH_NO_NORMAL_REPEAT | \ + FAST_PATH_NO_PAD_REPEAT | \ + FAST_PATH_NO_REFLECT_REPEAT) + +#define FAST_PATH_REFLECT_REPEAT \ + (FAST_PATH_NO_NONE_REPEAT | \ + FAST_PATH_NO_NORMAL_REPEAT | \ + FAST_PATH_NO_PAD_REPEAT) + +#define FAST_PATH_STANDARD_FLAGS \ + (FAST_PATH_NO_CONVOLUTION_FILTER | \ + FAST_PATH_NO_ACCESSORS | \ + FAST_PATH_NO_ALPHA_MAP | \ + FAST_PATH_NARROW_FORMAT) + +#define FAST_PATH_STD_DEST_FLAGS \ + (FAST_PATH_NO_ACCESSORS | \ + FAST_PATH_NO_ALPHA_MAP | \ + FAST_PATH_NARROW_FORMAT) + +#define SOURCE_FLAGS(format) \ + (FAST_PATH_STANDARD_FLAGS | \ + ((PIXMAN_ ## format == PIXMAN_solid) ? \ + 0 : (FAST_PATH_SAMPLES_COVER_CLIP | FAST_PATH_ID_TRANSFORM))) + +#define MASK_FLAGS(format, extra) \ + ((PIXMAN_ ## format == PIXMAN_null) ? 0 : (SOURCE_FLAGS (format) | extra)) + +#define FAST_PATH(op, src, src_flags, mask, mask_flags, dest, dest_flags, func) \ + PIXMAN_OP_ ## op, \ + PIXMAN_ ## src, \ + src_flags, \ + PIXMAN_ ## mask, \ + mask_flags, \ + PIXMAN_ ## dest, \ + dest_flags, \ + func + +#define PIXMAN_STD_FAST_PATH(op, src, mask, dest, func) \ + { FAST_PATH ( \ + op, \ + src, SOURCE_FLAGS (src), \ + mask, MASK_FLAGS (mask, FAST_PATH_UNIFIED_ALPHA), \ + dest, FAST_PATH_STD_DEST_FLAGS, \ + func) } + +#define PIXMAN_STD_FAST_PATH_CA(op, src, mask, dest, func) \ + { FAST_PATH ( \ + op, \ + src, SOURCE_FLAGS (src), \ + mask, MASK_FLAGS (mask, FAST_PATH_COMPONENT_ALPHA), \ + dest, FAST_PATH_STD_DEST_FLAGS, \ + func) } + +/* Memory allocation helpers */ +void * +pixman_malloc_ab (unsigned int n, unsigned int b); + +void * +pixman_malloc_abc (unsigned int a, unsigned int b, unsigned int c); + +pixman_bool_t +pixman_multiply_overflows_int (unsigned int a, unsigned int b); + +pixman_bool_t +pixman_addition_overflows_int (unsigned int a, unsigned int b); + +/* Compositing utilities */ +void +pixman_expand (uint64_t * dst, + const uint32_t * src, + pixman_format_code_t format, + int width); + +void +pixman_contract (uint32_t * dst, + const uint64_t *src, + int width); + + +/* Region Helpers */ +pixman_bool_t +pixman_region32_copy_from_region16 (pixman_region32_t *dst, + pixman_region16_t *src); + +pixman_bool_t +pixman_region16_copy_from_region32 (pixman_region16_t *dst, + pixman_region32_t *src); + + +/* Misc macros */ + +#ifndef FALSE +# define FALSE 0 +#endif + +#ifndef TRUE +# define TRUE 1 +#endif + +#ifndef MIN +# define MIN(a, b) ((a < b) ? a : b) +#endif + +#ifndef MAX +# define MAX(a, b) ((a > b) ? a : b) +#endif + +/* Integer division that rounds towards -infinity */ +#define DIV(a, b) \ + ((((a) < 0) == ((b) < 0)) ? (a) / (b) : \ + ((a) - (b) + 1 - (((b) < 0) << 1)) / (b)) + +/* Modulus that produces the remainder wrt. DIV */ +#define MOD(a, b) ((a) < 0 ? ((b) - ((-(a) - 1) % (b))) - 1 : (a) % (b)) + +#define CLIP(v, low, high) ((v) < (low) ? (low) : ((v) > (high) ? (high) : (v))) + +/* Conversion between 8888 and 0565 */ + +#define CONVERT_8888_TO_0565(s) \ + ((((s) >> 3) & 0x001f) | \ + (((s) >> 5) & 0x07e0) | \ + (((s) >> 8) & 0xf800)) + +#define CONVERT_0565_TO_0888(s) \ + (((((s) << 3) & 0xf8) | (((s) >> 2) & 0x7)) | \ + ((((s) << 5) & 0xfc00) | (((s) >> 1) & 0x300)) | \ + ((((s) << 8) & 0xf80000) | (((s) << 3) & 0x70000))) + +#define CONVERT_0565_TO_8888(s) (CONVERT_0565_TO_0888(s) | 0xff000000) + +/* Trivial versions that are useful in macros */ +#define CONVERT_8888_TO_8888(s) (s) +#define CONVERT_0565_TO_0565(s) (s) + +#define PIXMAN_FORMAT_IS_WIDE(f) \ + (PIXMAN_FORMAT_A (f) > 8 || \ + PIXMAN_FORMAT_R (f) > 8 || \ + PIXMAN_FORMAT_G (f) > 8 || \ + PIXMAN_FORMAT_B (f) > 8) + +#ifdef WORDS_BIGENDIAN +# define SCREEN_SHIFT_LEFT(x,n) ((x) << (n)) +# define SCREEN_SHIFT_RIGHT(x,n) ((x) >> (n)) +#else +# define SCREEN_SHIFT_LEFT(x,n) ((x) >> (n)) +# define SCREEN_SHIFT_RIGHT(x,n) ((x) << (n)) +#endif + +/* + * Various debugging code + */ + +#undef DEBUG + +#define COMPILE_TIME_ASSERT(x) \ + do { typedef int compile_time_assertion [(x)?1:-1]; } while (0) + +/* Turn on debugging depending on what type of release this is + */ +#if (((PIXMAN_VERSION_MICRO % 2) == 0) && ((PIXMAN_VERSION_MINOR % 2) == 1)) + +/* Debugging gets turned on for development releases because these + * are the things that end up in bleeding edge distributions such + * as Rawhide etc. + * + * For performance reasons we don't turn it on for stable releases or + * random git checkouts. (Random git checkouts are often used for + * performance work). + */ + +# define DEBUG + +#endif + +#ifdef DEBUG + +void +_pixman_log_error (const char *function, const char *message); + +#define return_if_fail(expr) \ + do \ + { \ + if (!(expr)) \ + { \ + _pixman_log_error (FUNC, "The expression " # expr " was false"); \ + return; \ + } \ + } \ + while (0) + +#define return_val_if_fail(expr, retval) \ + do \ + { \ + if (!(expr)) \ + { \ + _pixman_log_error (FUNC, "The expression " # expr " was false"); \ + return (retval); \ + } \ + } \ + while (0) + +#define critical_if_fail(expr) \ + do \ + { \ + if (!(expr)) \ + _pixman_log_error (FUNC, "The expression " # expr " was false"); \ + } \ + while (0) + + +#else + +#define _pixman_log_error(f,m) do { } while (0) \ + +#define return_if_fail(expr) \ + do \ + { \ + if (!(expr)) \ + return; \ + } \ + while (0) + +#define return_val_if_fail(expr, retval) \ + do \ + { \ + if (!(expr)) \ + return (retval); \ + } \ + while (0) + +#define critical_if_fail(expr) \ + do \ + { \ + } \ + while (0) +#endif + +/* + * Timers + */ + +#ifdef PIXMAN_TIMERS + +static inline uint64_t +oil_profile_stamp_rdtsc (void) +{ + uint64_t ts; + + __asm__ __volatile__ ("rdtsc\n" : "=A" (ts)); + return ts; +} + +#define OIL_STAMP oil_profile_stamp_rdtsc + +typedef struct pixman_timer_t pixman_timer_t; + +struct pixman_timer_t +{ + int initialized; + const char * name; + uint64_t n_times; + uint64_t total; + pixman_timer_t *next; +}; + +extern int timer_defined; + +void pixman_timer_register (pixman_timer_t *timer); + +#define TIMER_BEGIN(tname) \ + { \ + static pixman_timer_t timer ## tname; \ + uint64_t begin ## tname; \ + \ + if (!timer ## tname.initialized) \ + { \ + timer ## tname.initialized = 1; \ + timer ## tname.name = # tname; \ + pixman_timer_register (&timer ## tname); \ + } \ + \ + timer ## tname.n_times++; \ + begin ## tname = OIL_STAMP (); + +#define TIMER_END(tname) \ + timer ## tname.total += OIL_STAMP () - begin ## tname; \ + } + +#endif /* PIXMAN_TIMERS */ + +#endif /* PIXMAN_PRIVATE_H */ diff --git a/programs/develop/libraries/pixman/pixman-radial-gradient.c b/programs/develop/libraries/pixman/pixman-radial-gradient.c new file mode 100644 index 0000000000..fa5725840f --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-radial-gradient.c @@ -0,0 +1,448 @@ +/* -*- Mode: c; c-basic-offset: 4; tab-width: 8; indent-tabs-mode: t; -*- */ +/* + * + * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc. + * Copyright © 2000 SuSE, Inc. + * 2005 Lars Knoll & Zack Rusin, Trolltech + * Copyright © 2007 Red Hat, Inc. + * + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include +#include +#include "pixman-private.h" + +static inline pixman_fixed_32_32_t +dot (pixman_fixed_48_16_t x1, + pixman_fixed_48_16_t y1, + pixman_fixed_48_16_t z1, + pixman_fixed_48_16_t x2, + pixman_fixed_48_16_t y2, + pixman_fixed_48_16_t z2) +{ + /* + * Exact computation, assuming that the input values can + * be represented as pixman_fixed_16_16_t + */ + return x1 * x2 + y1 * y2 + z1 * z2; +} + +static inline double +fdot (double x1, + double y1, + double z1, + double x2, + double y2, + double z2) +{ + /* + * Error can be unbound in some special cases. + * Using clever dot product algorithms (for example compensated + * dot product) would improve this but make the code much less + * obvious + */ + return x1 * x2 + y1 * y2 + z1 * z2; +} + +static uint32_t +radial_compute_color (double a, + double b, + double c, + double inva, + double dr, + double mindr, + pixman_gradient_walker_t *walker, + pixman_repeat_t repeat) +{ + /* + * In this function error propagation can lead to bad results: + * - det can have an unbound error (if b*b-a*c is very small), + * potentially making it the opposite sign of what it should have been + * (thus clearing a pixel that would have been colored or vice-versa) + * or propagating the error to sqrtdet; + * if det has the wrong sign or b is very small, this can lead to bad + * results + * + * - the algorithm used to compute the solutions of the quadratic + * equation is not numerically stable (but saves one division compared + * to the numerically stable one); + * this can be a problem if a*c is much smaller than b*b + * + * - the above problems are worse if a is small (as inva becomes bigger) + */ + double det; + + if (a == 0) + { + double t; + + if (b == 0) + return 0; + + t = pixman_fixed_1 / 2 * c / b; + if (repeat == PIXMAN_REPEAT_NONE) + { + if (0 <= t && t <= pixman_fixed_1) + return _pixman_gradient_walker_pixel (walker, t); + } + else + { + if (t * dr > mindr) + return _pixman_gradient_walker_pixel (walker, t); + } + + return 0; + } + + det = fdot (b, a, 0, b, -c, 0); + if (det >= 0) + { + double sqrtdet, t0, t1; + + sqrtdet = sqrt (det); + t0 = (b + sqrtdet) * inva; + t1 = (b - sqrtdet) * inva; + + if (repeat == PIXMAN_REPEAT_NONE) + { + if (0 <= t0 && t0 <= pixman_fixed_1) + return _pixman_gradient_walker_pixel (walker, t0); + else if (0 <= t1 && t1 <= pixman_fixed_1) + return _pixman_gradient_walker_pixel (walker, t1); + } + else + { + if (t0 * dr > mindr) + return _pixman_gradient_walker_pixel (walker, t0); + else if (t1 * dr > mindr) + return _pixman_gradient_walker_pixel (walker, t1); + } + } + + return 0; +} + +static void +radial_gradient_get_scanline_32 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + /* + * Implementation of radial gradients following the PDF specification. + * See section 8.7.4.5.4 Type 3 (Radial) Shadings of the PDF Reference + * Manual (PDF 32000-1:2008 at the time of this writing). + * + * In the radial gradient problem we are given two circles (c₁,r₁) and + * (c₂,r₂) that define the gradient itself. + * + * Mathematically the gradient can be defined as the family of circles + * + * ((1-t)·c₁ + t·(c₂), (1-t)·r₁ + t·r₂) + * + * excluding those circles whose radius would be < 0. When a point + * belongs to more than one circle, the one with a bigger t is the only + * one that contributes to its color. When a point does not belong + * to any of the circles, it is transparent black, i.e. RGBA (0, 0, 0, 0). + * Further limitations on the range of values for t are imposed when + * the gradient is not repeated, namely t must belong to [0,1]. + * + * The graphical result is the same as drawing the valid (radius > 0) + * circles with increasing t in [-inf, +inf] (or in [0,1] if the gradient + * is not repeated) using SOURCE operatior composition. + * + * It looks like a cone pointing towards the viewer if the ending circle + * is smaller than the starting one, a cone pointing inside the page if + * the starting circle is the smaller one and like a cylinder if they + * have the same radius. + * + * What we actually do is, given the point whose color we are interested + * in, compute the t values for that point, solving for t in: + * + * length((1-t)·c₁ + t·(c₂) - p) = (1-t)·r₁ + t·r₂ + * + * Let's rewrite it in a simpler way, by defining some auxiliary + * variables: + * + * cd = c₂ - c₁ + * pd = p - c₁ + * dr = r₂ - r₁ + * lenght(t·cd - pd) = r₁ + t·dr + * + * which actually means + * + * hypot(t·cdx - pdx, t·cdy - pdy) = r₁ + t·dr + * + * or + * + * ⎷((t·cdx - pdx)² + (t·cdy - pdy)²) = r₁ + t·dr. + * + * If we impose (as stated earlier) that r₁ + t·dr >= 0, it becomes: + * + * (t·cdx - pdx)² + (t·cdy - pdy)² = (r₁ + t·dr)² + * + * where we can actually expand the squares and solve for t: + * + * t²cdx² - 2t·cdx·pdx + pdx² + t²cdy² - 2t·cdy·pdy + pdy² = + * = r₁² + 2·r₁·t·dr + t²·dr² + * + * (cdx² + cdy² - dr²)t² - 2(cdx·pdx + cdy·pdy + r₁·dr)t + + * (pdx² + pdy² - r₁²) = 0 + * + * A = cdx² + cdy² - dr² + * B = pdx·cdx + pdy·cdy + r₁·dr + * C = pdx² + pdy² - r₁² + * At² - 2Bt + C = 0 + * + * The solutions (unless the equation degenerates because of A = 0) are: + * + * t = (B ± ⎷(B² - A·C)) / A + * + * The solution we are going to prefer is the bigger one, unless the + * radius associated to it is negative (or it falls outside the valid t + * range). + * + * Additional observations (useful for optimizations): + * A does not depend on p + * + * A < 0 <=> one of the two circles completely contains the other one + * <=> for every p, the radiuses associated with the two t solutions + * have opposite sign + */ + + gradient_t *gradient = (gradient_t *)image; + source_image_t *source = (source_image_t *)image; + radial_gradient_t *radial = (radial_gradient_t *)image; + uint32_t *end = buffer + width; + pixman_gradient_walker_t walker; + pixman_vector_t v, unit; + + /* reference point is the center of the pixel */ + v.vector[0] = pixman_int_to_fixed (x) + pixman_fixed_1 / 2; + v.vector[1] = pixman_int_to_fixed (y) + pixman_fixed_1 / 2; + v.vector[2] = pixman_fixed_1; + + _pixman_gradient_walker_init (&walker, gradient, source->common.repeat); + + if (source->common.transform) + { + if (!pixman_transform_point_3d (source->common.transform, &v)) + return; + + unit.vector[0] = source->common.transform->matrix[0][0]; + unit.vector[1] = source->common.transform->matrix[1][0]; + unit.vector[2] = source->common.transform->matrix[2][0]; + } + else + { + unit.vector[0] = pixman_fixed_1; + unit.vector[1] = 0; + unit.vector[2] = 0; + } + + if (unit.vector[2] == 0 && v.vector[2] == pixman_fixed_1) + { + /* + * Given: + * + * t = (B ± ⎷(B² - A·C)) / A + * + * where + * + * A = cdx² + cdy² - dr² + * B = pdx·cdx + pdy·cdy + r₁·dr + * C = pdx² + pdy² - r₁² + * det = B² - A·C + * + * Since we have an affine transformation, we know that (pdx, pdy) + * increase linearly with each pixel, + * + * pdx = pdx₀ + n·ux, + * pdy = pdy₀ + n·uy, + * + * we can then express B, C and det through multiple differentiation. + */ + pixman_fixed_32_32_t b, db, c, dc, ddc; + + /* warning: this computation may overflow */ + v.vector[0] -= radial->c1.x; + v.vector[1] -= radial->c1.y; + + /* + * B and C are computed and updated exactly. + * If fdot was used instead of dot, in the worst case it would + * lose 11 bits of precision in each of the multiplication and + * summing up would zero out all the bit that were preserved, + * thus making the result 0 instead of the correct one. + * This would mean a worst case of unbound relative error or + * about 2^10 absolute error + */ + b = dot (v.vector[0], v.vector[1], radial->c1.radius, + radial->delta.x, radial->delta.y, radial->delta.radius); + db = dot (unit.vector[0], unit.vector[1], 0, + radial->delta.x, radial->delta.y, 0); + + c = dot (v.vector[0], v.vector[1], + -((pixman_fixed_48_16_t) radial->c1.radius), + v.vector[0], v.vector[1], radial->c1.radius); + dc = dot (2 * (pixman_fixed_48_16_t) v.vector[0] + unit.vector[0], + 2 * (pixman_fixed_48_16_t) v.vector[1] + unit.vector[1], + 0, + unit.vector[0], unit.vector[1], 0); + ddc = 2 * dot (unit.vector[0], unit.vector[1], 0, + unit.vector[0], unit.vector[1], 0); + + while (buffer < end) + { + if (!mask || *mask++) + { + *buffer = radial_compute_color (radial->a, b, c, + radial->inva, + radial->delta.radius, + radial->mindr, + &walker, + source->common.repeat); + } + + b += db; + c += dc; + dc += ddc; + ++buffer; + } + } + else + { + /* projective */ + /* Warning: + * error propagation guarantees are much looser than in the affine case + */ + while (buffer < end) + { + if (!mask || *mask++) + { + if (v.vector[2] != 0) + { + double pdx, pdy, invv2, b, c; + + invv2 = 1. * pixman_fixed_1 / v.vector[2]; + + pdx = v.vector[0] * invv2 - radial->c1.x; + /* / pixman_fixed_1 */ + + pdy = v.vector[1] * invv2 - radial->c1.y; + /* / pixman_fixed_1 */ + + b = fdot (pdx, pdy, radial->c1.radius, + radial->delta.x, radial->delta.y, + radial->delta.radius); + /* / pixman_fixed_1 / pixman_fixed_1 */ + + c = fdot (pdx, pdy, -radial->c1.radius, + pdx, pdy, radial->c1.radius); + /* / pixman_fixed_1 / pixman_fixed_1 */ + + *buffer = radial_compute_color (radial->a, b, c, + radial->inva, + radial->delta.radius, + radial->mindr, + &walker, + source->common.repeat); + } + else + { + *buffer = 0; + } + } + + ++buffer; + + v.vector[0] += unit.vector[0]; + v.vector[1] += unit.vector[1]; + v.vector[2] += unit.vector[2]; + } + } +} + +static void +radial_gradient_property_changed (pixman_image_t *image) +{ + image->common.get_scanline_32 = radial_gradient_get_scanline_32; + image->common.get_scanline_64 = _pixman_image_get_scanline_generic_64; +} + +PIXMAN_EXPORT pixman_image_t * +pixman_image_create_radial_gradient (pixman_point_fixed_t * inner, + pixman_point_fixed_t * outer, + pixman_fixed_t inner_radius, + pixman_fixed_t outer_radius, + const pixman_gradient_stop_t *stops, + int n_stops) +{ + pixman_image_t *image; + radial_gradient_t *radial; + + image = _pixman_image_allocate (); + + if (!image) + return NULL; + + radial = &image->radial; + + if (!_pixman_init_gradient (&radial->common, stops, n_stops)) + { + free (image); + return NULL; + } + + image->type = RADIAL; + + radial->c1.x = inner->x; + radial->c1.y = inner->y; + radial->c1.radius = inner_radius; + radial->c2.x = outer->x; + radial->c2.y = outer->y; + radial->c2.radius = outer_radius; + + /* warning: this computations may overflow */ + radial->delta.x = radial->c2.x - radial->c1.x; + radial->delta.y = radial->c2.y - radial->c1.y; + radial->delta.radius = radial->c2.radius - radial->c1.radius; + + /* computed exactly, then cast to double -> every bit of the double + representation is correct (53 bits) */ + radial->a = dot (radial->delta.x, radial->delta.y, -radial->delta.radius, + radial->delta.x, radial->delta.y, radial->delta.radius); + if (radial->a != 0) + radial->inva = 1. * pixman_fixed_1 / radial->a; + + radial->mindr = -1. * pixman_fixed_1 * radial->c1.radius; + + image->common.property_changed = radial_gradient_property_changed; + + return image; +} + diff --git a/programs/develop/libraries/pixman/pixman-region.c b/programs/develop/libraries/pixman/pixman-region.c new file mode 100644 index 0000000000..4e7c8db75e --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-region.c @@ -0,0 +1,2769 @@ +/* + * Copyright 1987, 1988, 1989, 1998 The Open Group + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation. + * + * The above copyright notice and this permission notice 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 + * OPEN GROUP 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. + * + * Except as contained in this notice, the name of The Open Group shall not be + * used in advertising or otherwise to promote the sale, use or other dealings + * in this Software without prior written authorization from The Open Group. + * + * Copyright 1987, 1988, 1989 by + * Digital Equipment Corporation, Maynard, Massachusetts. + * + * All Rights Reserved + * + * Permission to use, copy, modify, and distribute this software and its + * documentation for any purpose and without fee is hereby granted, + * provided that the above copyright notice appear in all copies and that + * both that copyright notice and this permission notice appear in + * supporting documentation, and that the name of Digital not be + * used in advertising or publicity pertaining to distribution of the + * software without specific, written prior permission. + * + * DIGITAL DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING + * ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL + * DIGITAL BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR + * ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, + * WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, + * ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + * + * Copyright © 1998 Keith Packard + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * KEITH PACKARD DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL KEITH PACKARD BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR + * PERFORMANCE OF THIS SOFTWARE. + */ + +#include +#include +#include +#include +#include "pixman-private.h" + +#define PIXREGION_NIL(reg) ((reg)->data && !(reg)->data->numRects) +/* not a region */ +#define PIXREGION_NAR(reg) ((reg)->data == pixman_broken_data) +#define PIXREGION_NUMRECTS(reg) ((reg)->data ? (reg)->data->numRects : 1) +#define PIXREGION_SIZE(reg) ((reg)->data ? (reg)->data->size : 0) +#define PIXREGION_RECTS(reg) \ + ((reg)->data ? (box_type_t *)((reg)->data + 1) \ + : &(reg)->extents) +#define PIXREGION_BOXPTR(reg) ((box_type_t *)((reg)->data + 1)) +#define PIXREGION_BOX(reg, i) (&PIXREGION_BOXPTR (reg)[i]) +#define PIXREGION_TOP(reg) PIXREGION_BOX (reg, (reg)->data->numRects) +#define PIXREGION_END(reg) PIXREGION_BOX (reg, (reg)->data->numRects - 1) + +#define GOOD_RECT(rect) ((rect)->x1 < (rect)->x2 && (rect)->y1 < (rect)->y2) +#define BAD_RECT(rect) ((rect)->x1 > (rect)->x2 || (rect)->y1 > (rect)->y2) + +#ifdef DEBUG + +#define GOOD(reg) \ + do \ + { \ + if (!PREFIX (_selfcheck (reg))) \ + _pixman_log_error (FUNC, "Malformed region " # reg); \ + } while (0) + +#else + +#define GOOD(reg) + +#endif + +static const box_type_t PREFIX (_empty_box_) = { 0, 0, 0, 0 }; +static const region_data_type_t PREFIX (_empty_data_) = { 0, 0 }; +static const region_data_type_t PREFIX (_broken_data_) = { 0, 0 }; + +static box_type_t *pixman_region_empty_box = + (box_type_t *)&PREFIX (_empty_box_); +static region_data_type_t *pixman_region_empty_data = + (region_data_type_t *)&PREFIX (_empty_data_); +static region_data_type_t *pixman_broken_data = + (region_data_type_t *)&PREFIX (_broken_data_); + +static pixman_bool_t +pixman_break (region_type_t *region); + +/* + * The functions in this file implement the Region abstraction used extensively + * throughout the X11 sample server. A Region is simply a set of disjoint + * (non-overlapping) rectangles, plus an "extent" rectangle which is the + * smallest single rectangle that contains all the non-overlapping rectangles. + * + * A Region is implemented as a "y-x-banded" array of rectangles. This array + * imposes two degrees of order. First, all rectangles are sorted by top side + * y coordinate first (y1), and then by left side x coordinate (x1). + * + * Furthermore, the rectangles are grouped into "bands". Each rectangle in a + * band has the same top y coordinate (y1), and each has the same bottom y + * coordinate (y2). Thus all rectangles in a band differ only in their left + * and right side (x1 and x2). Bands are implicit in the array of rectangles: + * there is no separate list of band start pointers. + * + * The y-x band representation does not minimize rectangles. In particular, + * if a rectangle vertically crosses a band (the rectangle has scanlines in + * the y1 to y2 area spanned by the band), then the rectangle may be broken + * down into two or more smaller rectangles stacked one atop the other. + * + * ----------- ----------- + * | | | | band 0 + * | | -------- ----------- -------- + * | | | | in y-x banded | | | | band 1 + * | | | | form is | | | | + * ----------- | | ----------- -------- + * | | | | band 2 + * -------- -------- + * + * An added constraint on the rectangles is that they must cover as much + * horizontal area as possible: no two rectangles within a band are allowed + * to touch. + * + * Whenever possible, bands will be merged together to cover a greater vertical + * distance (and thus reduce the number of rectangles). Two bands can be merged + * only if the bottom of one touches the top of the other and they have + * rectangles in the same places (of the same width, of course). + * + * Adam de Boor wrote most of the original region code. Joel McCormack + * substantially modified or rewrote most of the core arithmetic routines, and + * added pixman_region_validate in order to support several speed improvements + * to pixman_region_validate_tree. Bob Scheifler changed the representation + * to be more compact when empty or a single rectangle, and did a bunch of + * gratuitous reformatting. Carl Worth did further gratuitous reformatting + * while re-merging the server and client region code into libpixregion. + * Soren Sandmann did even more gratuitous reformatting. + */ + +/* true iff two Boxes overlap */ +#define EXTENTCHECK(r1, r2) \ + (!( ((r1)->x2 <= (r2)->x1) || \ + ((r1)->x1 >= (r2)->x2) || \ + ((r1)->y2 <= (r2)->y1) || \ + ((r1)->y1 >= (r2)->y2) ) ) + +/* true iff (x,y) is in Box */ +#define INBOX(r, x, y) \ + ( ((r)->x2 > x) && \ + ((r)->x1 <= x) && \ + ((r)->y2 > y) && \ + ((r)->y1 <= y) ) + +/* true iff Box r1 contains Box r2 */ +#define SUBSUMES(r1, r2) \ + ( ((r1)->x1 <= (r2)->x1) && \ + ((r1)->x2 >= (r2)->x2) && \ + ((r1)->y1 <= (r2)->y1) && \ + ((r1)->y2 >= (r2)->y2) ) + +static size_t +PIXREGION_SZOF (size_t n) +{ + size_t size = n * sizeof(box_type_t); + + if (n > UINT32_MAX / sizeof(box_type_t)) + return 0; + + if (sizeof(region_data_type_t) > UINT32_MAX - size) + return 0; + + return size + sizeof(region_data_type_t); +} + +static void * +alloc_data (size_t n) +{ + size_t sz = PIXREGION_SZOF (n); + + if (!sz) + return NULL; + + return malloc (sz); +} + +#define FREE_DATA(reg) if ((reg)->data && (reg)->data->size) free ((reg)->data) + +#define RECTALLOC_BAIL(region, n, bail) \ + do \ + { \ + if (!(region)->data || \ + (((region)->data->numRects + (n)) > (region)->data->size)) \ + { \ + if (!pixman_rect_alloc (region, n)) \ + goto bail; \ + } \ + } while (0) + +#define RECTALLOC(region, n) \ + do \ + { \ + if (!(region)->data || \ + (((region)->data->numRects + (n)) > (region)->data->size)) \ + { \ + if (!pixman_rect_alloc (region, n)) { \ + return FALSE; \ + } \ + } \ + } while (0) + +#define ADDRECT(next_rect, nx1, ny1, nx2, ny2) \ + do \ + { \ + next_rect->x1 = nx1; \ + next_rect->y1 = ny1; \ + next_rect->x2 = nx2; \ + next_rect->y2 = ny2; \ + next_rect++; \ + } \ + while (0) + +#define NEWRECT(region, next_rect, nx1, ny1, nx2, ny2) \ + do \ + { \ + if (!(region)->data || \ + ((region)->data->numRects == (region)->data->size)) \ + { \ + if (!pixman_rect_alloc (region, 1)) \ + return FALSE; \ + next_rect = PIXREGION_TOP (region); \ + } \ + ADDRECT (next_rect, nx1, ny1, nx2, ny2); \ + region->data->numRects++; \ + critical_if_fail (region->data->numRects <= region->data->size); \ + } while (0) + +#define DOWNSIZE(reg, numRects) \ + do \ + { \ + if (((numRects) < ((reg)->data->size >> 1)) && \ + ((reg)->data->size > 50)) \ + { \ + region_data_type_t * new_data; \ + size_t data_size = PIXREGION_SZOF (numRects); \ + \ + if (!data_size) \ + { \ + new_data = NULL; \ + } \ + else \ + { \ + new_data = (region_data_type_t *) \ + realloc ((reg)->data, data_size); \ + } \ + \ + if (new_data) \ + { \ + new_data->size = (numRects); \ + (reg)->data = new_data; \ + } \ + } \ + } while (0) + +PIXMAN_EXPORT pixman_bool_t +PREFIX (_equal) (region_type_t *reg1, region_type_t *reg2) +{ + int i; + box_type_t *rects1; + box_type_t *rects2; + + if (reg1->extents.x1 != reg2->extents.x1) + return FALSE; + + if (reg1->extents.x2 != reg2->extents.x2) + return FALSE; + + if (reg1->extents.y1 != reg2->extents.y1) + return FALSE; + + if (reg1->extents.y2 != reg2->extents.y2) + return FALSE; + + if (PIXREGION_NUMRECTS (reg1) != PIXREGION_NUMRECTS (reg2)) + return FALSE; + + rects1 = PIXREGION_RECTS (reg1); + rects2 = PIXREGION_RECTS (reg2); + + for (i = 0; i != PIXREGION_NUMRECTS (reg1); i++) + { + if (rects1[i].x1 != rects2[i].x1) + return FALSE; + + if (rects1[i].x2 != rects2[i].x2) + return FALSE; + + if (rects1[i].y1 != rects2[i].y1) + return FALSE; + + if (rects1[i].y2 != rects2[i].y2) + return FALSE; + } + + return TRUE; +} + +int +PREFIX (_print) (region_type_t *rgn) +{ + int num, size; + int i; + box_type_t * rects; + + num = PIXREGION_NUMRECTS (rgn); + size = PIXREGION_SIZE (rgn); + rects = PIXREGION_RECTS (rgn); + + fprintf (stderr, "num: %d size: %d\n", num, size); + fprintf (stderr, "extents: %d %d %d %d\n", + rgn->extents.x1, + rgn->extents.y1, + rgn->extents.x2, + rgn->extents.y2); + + for (i = 0; i < num; i++) + { + fprintf (stderr, "%d %d %d %d \n", + rects[i].x1, rects[i].y1, rects[i].x2, rects[i].y2); + } + + fprintf (stderr, "\n"); + + return(num); +} + + +PIXMAN_EXPORT void +PREFIX (_init) (region_type_t *region) +{ + region->extents = *pixman_region_empty_box; + region->data = pixman_region_empty_data; +} + +PIXMAN_EXPORT void +PREFIX (_init_rect) (region_type_t * region, + int x, + int y, + unsigned int width, + unsigned int height) +{ + region->extents.x1 = x; + region->extents.y1 = y; + region->extents.x2 = x + width; + region->extents.y2 = y + height; + + if (!GOOD_RECT (®ion->extents)) + { + if (BAD_RECT (®ion->extents)) + _pixman_log_error (FUNC, "Invalid rectangle passed"); + PREFIX (_init) (region); + return; + } + + region->data = NULL; +} + +PIXMAN_EXPORT void +PREFIX (_init_with_extents) (region_type_t *region, box_type_t *extents) +{ + if (!GOOD_RECT (extents)) + { + if (BAD_RECT (extents)) + _pixman_log_error (FUNC, "Invalid rectangle passed"); + PREFIX (_init) (region); + return; + } + region->extents = *extents; + + region->data = NULL; +} + +PIXMAN_EXPORT void +PREFIX (_fini) (region_type_t *region) +{ + GOOD (region); + FREE_DATA (region); +} + +PIXMAN_EXPORT int +PREFIX (_n_rects) (region_type_t *region) +{ + return PIXREGION_NUMRECTS (region); +} + +PIXMAN_EXPORT box_type_t * +PREFIX (_rectangles) (region_type_t *region, + int *n_rects) +{ + if (n_rects) + *n_rects = PIXREGION_NUMRECTS (region); + + return PIXREGION_RECTS (region); +} + +static pixman_bool_t +pixman_break (region_type_t *region) +{ + FREE_DATA (region); + + region->extents = *pixman_region_empty_box; + region->data = pixman_broken_data; + + return FALSE; +} + +static pixman_bool_t +pixman_rect_alloc (region_type_t * region, + int n) +{ + region_data_type_t *data; + + if (!region->data) + { + n++; + region->data = alloc_data (n); + + if (!region->data) + return pixman_break (region); + + region->data->numRects = 1; + *PIXREGION_BOXPTR (region) = region->extents; + } + else if (!region->data->size) + { + region->data = alloc_data (n); + + if (!region->data) + return pixman_break (region); + + region->data->numRects = 0; + } + else + { + size_t data_size; + + if (n == 1) + { + n = region->data->numRects; + if (n > 500) /* XXX pick numbers out of a hat */ + n = 250; + } + + n += region->data->numRects; + data_size = PIXREGION_SZOF (n); + + if (!data_size) + { + data = NULL; + } + else + { + data = (region_data_type_t *) + realloc (region->data, PIXREGION_SZOF (n)); + } + + if (!data) + return pixman_break (region); + + region->data = data; + } + + region->data->size = n; + + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +PREFIX (_copy) (region_type_t *dst, region_type_t *src) +{ + GOOD (dst); + GOOD (src); + + if (dst == src) + return TRUE; + + dst->extents = src->extents; + + if (!src->data || !src->data->size) + { + FREE_DATA (dst); + dst->data = src->data; + return TRUE; + } + + if (!dst->data || (dst->data->size < src->data->numRects)) + { + FREE_DATA (dst); + + dst->data = alloc_data (src->data->numRects); + + if (!dst->data) + return pixman_break (dst); + + dst->data->size = src->data->numRects; + } + + dst->data->numRects = src->data->numRects; + + memmove ((char *)PIXREGION_BOXPTR (dst), (char *)PIXREGION_BOXPTR (src), + dst->data->numRects * sizeof(box_type_t)); + + return TRUE; +} + +/*====================================================================== + * Generic Region Operator + *====================================================================*/ + +/*- + *----------------------------------------------------------------------- + * pixman_coalesce -- + * Attempt to merge the boxes in the current band with those in the + * previous one. We are guaranteed that the current band extends to + * the end of the rects array. Used only by pixman_op. + * + * Results: + * The new index for the previous band. + * + * Side Effects: + * If coalescing takes place: + * - rectangles in the previous band will have their y2 fields + * altered. + * - region->data->numRects will be decreased. + * + *----------------------------------------------------------------------- + */ +static inline int +pixman_coalesce (region_type_t * region, /* Region to coalesce */ + int prev_start, /* Index of start of previous band */ + int cur_start) /* Index of start of current band */ +{ + box_type_t *prev_box; /* Current box in previous band */ + box_type_t *cur_box; /* Current box in current band */ + int numRects; /* Number rectangles in both bands */ + int y2; /* Bottom of current band */ + + /* + * Figure out how many rectangles are in the band. + */ + numRects = cur_start - prev_start; + critical_if_fail (numRects == region->data->numRects - cur_start); + + if (!numRects) return cur_start; + + /* + * The bands may only be coalesced if the bottom of the previous + * matches the top scanline of the current. + */ + prev_box = PIXREGION_BOX (region, prev_start); + cur_box = PIXREGION_BOX (region, cur_start); + if (prev_box->y2 != cur_box->y1) return cur_start; + + /* + * Make sure the bands have boxes in the same places. This + * assumes that boxes have been added in such a way that they + * cover the most area possible. I.e. two boxes in a band must + * have some horizontal space between them. + */ + y2 = cur_box->y2; + + do + { + if ((prev_box->x1 != cur_box->x1) || (prev_box->x2 != cur_box->x2)) + return (cur_start); + + prev_box++; + cur_box++; + numRects--; + } + while (numRects); + + /* + * The bands may be merged, so set the bottom y of each box + * in the previous band to the bottom y of the current band. + */ + numRects = cur_start - prev_start; + region->data->numRects -= numRects; + + do + { + prev_box--; + prev_box->y2 = y2; + numRects--; + } + while (numRects); + + return prev_start; +} + +/* Quicky macro to avoid trivial reject procedure calls to pixman_coalesce */ + +#define COALESCE(new_reg, prev_band, cur_band) \ + do \ + { \ + if (cur_band - prev_band == new_reg->data->numRects - cur_band) \ + prev_band = pixman_coalesce (new_reg, prev_band, cur_band); \ + else \ + prev_band = cur_band; \ + } while (0) + +/*- + *----------------------------------------------------------------------- + * pixman_region_append_non_o -- + * Handle a non-overlapping band for the union and subtract operations. + * Just adds the (top/bottom-clipped) rectangles into the region. + * Doesn't have to check for subsumption or anything. + * + * Results: + * None. + * + * Side Effects: + * region->data->numRects is incremented and the rectangles overwritten + * with the rectangles we're passed. + * + *----------------------------------------------------------------------- + */ +static inline pixman_bool_t +pixman_region_append_non_o (region_type_t * region, + box_type_t * r, + box_type_t * r_end, + int y1, + int y2) +{ + box_type_t *next_rect; + int new_rects; + + new_rects = r_end - r; + + critical_if_fail (y1 < y2); + critical_if_fail (new_rects != 0); + + /* Make sure we have enough space for all rectangles to be added */ + RECTALLOC (region, new_rects); + next_rect = PIXREGION_TOP (region); + region->data->numRects += new_rects; + + do + { + critical_if_fail (r->x1 < r->x2); + ADDRECT (next_rect, r->x1, y1, r->x2, y2); + r++; + } + while (r != r_end); + + return TRUE; +} + +#define FIND_BAND(r, r_band_end, r_end, ry1) \ + do \ + { \ + ry1 = r->y1; \ + r_band_end = r + 1; \ + while ((r_band_end != r_end) && (r_band_end->y1 == ry1)) { \ + r_band_end++; \ + } \ + } while (0) + +#define APPEND_REGIONS(new_reg, r, r_end) \ + do \ + { \ + int new_rects; \ + if ((new_rects = r_end - r)) { \ + RECTALLOC_BAIL (new_reg, new_rects, bail); \ + memmove ((char *)PIXREGION_TOP (new_reg), (char *)r, \ + new_rects * sizeof(box_type_t)); \ + new_reg->data->numRects += new_rects; \ + } \ + } while (0) + +/*- + *----------------------------------------------------------------------- + * pixman_op -- + * Apply an operation to two regions. Called by pixman_region_union, pixman_region_inverse, + * pixman_region_subtract, pixman_region_intersect.... Both regions MUST have at least one + * rectangle, and cannot be the same object. + * + * Results: + * TRUE if successful. + * + * Side Effects: + * The new region is overwritten. + * overlap set to TRUE if overlap_func ever returns TRUE. + * + * Notes: + * The idea behind this function is to view the two regions as sets. + * Together they cover a rectangle of area that this function divides + * into horizontal bands where points are covered only by one region + * or by both. For the first case, the non_overlap_func is called with + * each the band and the band's upper and lower extents. For the + * second, the overlap_func is called to process the entire band. It + * is responsible for clipping the rectangles in the band, though + * this function provides the boundaries. + * At the end of each band, the new region is coalesced, if possible, + * to reduce the number of rectangles in the region. + * + *----------------------------------------------------------------------- + */ + +typedef pixman_bool_t (*overlap_proc_ptr) (region_type_t *region, + box_type_t * r1, + box_type_t * r1_end, + box_type_t * r2, + box_type_t * r2_end, + int y1, + int y2, + int * overlap); + +static pixman_bool_t +pixman_op (region_type_t * new_reg, /* Place to store result */ + region_type_t * reg1, /* First region in operation */ + region_type_t * reg2, /* 2d region in operation */ + overlap_proc_ptr overlap_func, /* Function to call for over- + * lapping bands */ + int append_non1, /* Append non-overlapping bands + * in region 1 ? + */ + int append_non2, /* Append non-overlapping bands + * in region 2 ? + */ + int * overlap) +{ + box_type_t *r1; /* Pointer into first region */ + box_type_t *r2; /* Pointer into 2d region */ + box_type_t *r1_end; /* End of 1st region */ + box_type_t *r2_end; /* End of 2d region */ + int ybot; /* Bottom of intersection */ + int ytop; /* Top of intersection */ + region_data_type_t *old_data; /* Old data for new_reg */ + int prev_band; /* Index of start of + * previous band in new_reg */ + int cur_band; /* Index of start of current + * band in new_reg */ + box_type_t * r1_band_end; /* End of current band in r1 */ + box_type_t * r2_band_end; /* End of current band in r2 */ + int top; /* Top of non-overlapping band */ + int bot; /* Bottom of non-overlapping band*/ + int r1y1; /* Temps for r1->y1 and r2->y1 */ + int r2y1; + int new_size; + int numRects; + + /* + * Break any region computed from a broken region + */ + if (PIXREGION_NAR (reg1) || PIXREGION_NAR (reg2)) + return pixman_break (new_reg); + + /* + * Initialization: + * set r1, r2, r1_end and r2_end appropriately, save the rectangles + * of the destination region until the end in case it's one of + * the two source regions, then mark the "new" region empty, allocating + * another array of rectangles for it to use. + */ + + r1 = PIXREGION_RECTS (reg1); + new_size = PIXREGION_NUMRECTS (reg1); + r1_end = r1 + new_size; + + numRects = PIXREGION_NUMRECTS (reg2); + r2 = PIXREGION_RECTS (reg2); + r2_end = r2 + numRects; + + critical_if_fail (r1 != r1_end); + critical_if_fail (r2 != r2_end); + + old_data = (region_data_type_t *)NULL; + + if (((new_reg == reg1) && (new_size > 1)) || + ((new_reg == reg2) && (numRects > 1))) + { + old_data = new_reg->data; + new_reg->data = pixman_region_empty_data; + } + + /* guess at new size */ + if (numRects > new_size) + new_size = numRects; + + new_size <<= 1; + + if (!new_reg->data) + new_reg->data = pixman_region_empty_data; + else if (new_reg->data->size) + new_reg->data->numRects = 0; + + if (new_size > new_reg->data->size) + { + if (!pixman_rect_alloc (new_reg, new_size)) + { + if (old_data) + free (old_data); + return FALSE; + } + } + + /* + * Initialize ybot. + * In the upcoming loop, ybot and ytop serve different functions depending + * on whether the band being handled is an overlapping or non-overlapping + * band. + * In the case of a non-overlapping band (only one of the regions + * has points in the band), ybot is the bottom of the most recent + * intersection and thus clips the top of the rectangles in that band. + * ytop is the top of the next intersection between the two regions and + * serves to clip the bottom of the rectangles in the current band. + * For an overlapping band (where the two regions intersect), ytop clips + * the top of the rectangles of both regions and ybot clips the bottoms. + */ + + ybot = MIN (r1->y1, r2->y1); + + /* + * prev_band serves to mark the start of the previous band so rectangles + * can be coalesced into larger rectangles. qv. pixman_coalesce, above. + * In the beginning, there is no previous band, so prev_band == cur_band + * (cur_band is set later on, of course, but the first band will always + * start at index 0). prev_band and cur_band must be indices because of + * the possible expansion, and resultant moving, of the new region's + * array of rectangles. + */ + prev_band = 0; + + do + { + /* + * This algorithm proceeds one source-band (as opposed to a + * destination band, which is determined by where the two regions + * intersect) at a time. r1_band_end and r2_band_end serve to mark the + * rectangle after the last one in the current band for their + * respective regions. + */ + critical_if_fail (r1 != r1_end); + critical_if_fail (r2 != r2_end); + + FIND_BAND (r1, r1_band_end, r1_end, r1y1); + FIND_BAND (r2, r2_band_end, r2_end, r2y1); + + /* + * First handle the band that doesn't intersect, if any. + * + * Note that attention is restricted to one band in the + * non-intersecting region at once, so if a region has n + * bands between the current position and the next place it overlaps + * the other, this entire loop will be passed through n times. + */ + if (r1y1 < r2y1) + { + if (append_non1) + { + top = MAX (r1y1, ybot); + bot = MIN (r1->y2, r2y1); + if (top != bot) + { + cur_band = new_reg->data->numRects; + if (!pixman_region_append_non_o (new_reg, r1, r1_band_end, top, bot)) + goto bail; + COALESCE (new_reg, prev_band, cur_band); + } + } + ytop = r2y1; + } + else if (r2y1 < r1y1) + { + if (append_non2) + { + top = MAX (r2y1, ybot); + bot = MIN (r2->y2, r1y1); + + if (top != bot) + { + cur_band = new_reg->data->numRects; + + if (!pixman_region_append_non_o (new_reg, r2, r2_band_end, top, bot)) + goto bail; + + COALESCE (new_reg, prev_band, cur_band); + } + } + ytop = r1y1; + } + else + { + ytop = r1y1; + } + + /* + * Now see if we've hit an intersecting band. The two bands only + * intersect if ybot > ytop + */ + ybot = MIN (r1->y2, r2->y2); + if (ybot > ytop) + { + cur_band = new_reg->data->numRects; + + if (!(*overlap_func)(new_reg, + r1, r1_band_end, + r2, r2_band_end, + ytop, ybot, + overlap)) + { + goto bail; + } + + COALESCE (new_reg, prev_band, cur_band); + } + + /* + * If we've finished with a band (y2 == ybot) we skip forward + * in the region to the next band. + */ + if (r1->y2 == ybot) + r1 = r1_band_end; + + if (r2->y2 == ybot) + r2 = r2_band_end; + + } + while (r1 != r1_end && r2 != r2_end); + + /* + * Deal with whichever region (if any) still has rectangles left. + * + * We only need to worry about banding and coalescing for the very first + * band left. After that, we can just group all remaining boxes, + * regardless of how many bands, into one final append to the list. + */ + + if ((r1 != r1_end) && append_non1) + { + /* Do first non_overlap1Func call, which may be able to coalesce */ + FIND_BAND (r1, r1_band_end, r1_end, r1y1); + + cur_band = new_reg->data->numRects; + + if (!pixman_region_append_non_o (new_reg, + r1, r1_band_end, + MAX (r1y1, ybot), r1->y2)) + { + goto bail; + } + + COALESCE (new_reg, prev_band, cur_band); + + /* Just append the rest of the boxes */ + APPEND_REGIONS (new_reg, r1_band_end, r1_end); + } + else if ((r2 != r2_end) && append_non2) + { + /* Do first non_overlap2Func call, which may be able to coalesce */ + FIND_BAND (r2, r2_band_end, r2_end, r2y1); + + cur_band = new_reg->data->numRects; + + if (!pixman_region_append_non_o (new_reg, + r2, r2_band_end, + MAX (r2y1, ybot), r2->y2)) + { + goto bail; + } + + COALESCE (new_reg, prev_band, cur_band); + + /* Append rest of boxes */ + APPEND_REGIONS (new_reg, r2_band_end, r2_end); + } + + if (old_data) + free (old_data); + + if (!(numRects = new_reg->data->numRects)) + { + FREE_DATA (new_reg); + new_reg->data = pixman_region_empty_data; + } + else if (numRects == 1) + { + new_reg->extents = *PIXREGION_BOXPTR (new_reg); + FREE_DATA (new_reg); + new_reg->data = (region_data_type_t *)NULL; + } + else + { + DOWNSIZE (new_reg, numRects); + } + + return TRUE; + +bail: + if (old_data) + free (old_data); + + return pixman_break (new_reg); +} + +/*- + *----------------------------------------------------------------------- + * pixman_set_extents -- + * Reset the extents of a region to what they should be. Called by + * pixman_region_subtract and pixman_region_intersect as they can't + * figure it out along the way or do so easily, as pixman_region_union can. + * + * Results: + * None. + * + * Side Effects: + * The region's 'extents' structure is overwritten. + * + *----------------------------------------------------------------------- + */ +static void +pixman_set_extents (region_type_t *region) +{ + box_type_t *box, *box_end; + + if (!region->data) + return; + + if (!region->data->size) + { + region->extents.x2 = region->extents.x1; + region->extents.y2 = region->extents.y1; + return; + } + + box = PIXREGION_BOXPTR (region); + box_end = PIXREGION_END (region); + + /* + * Since box is the first rectangle in the region, it must have the + * smallest y1 and since box_end is the last rectangle in the region, + * it must have the largest y2, because of banding. Initialize x1 and + * x2 from box and box_end, resp., as good things to initialize them + * to... + */ + region->extents.x1 = box->x1; + region->extents.y1 = box->y1; + region->extents.x2 = box_end->x2; + region->extents.y2 = box_end->y2; + + critical_if_fail (region->extents.y1 < region->extents.y2); + + while (box <= box_end) + { + if (box->x1 < region->extents.x1) + region->extents.x1 = box->x1; + if (box->x2 > region->extents.x2) + region->extents.x2 = box->x2; + box++; + } + + critical_if_fail (region->extents.x1 < region->extents.x2); +} + +/*====================================================================== + * Region Intersection + *====================================================================*/ +/*- + *----------------------------------------------------------------------- + * pixman_region_intersect_o -- + * Handle an overlapping band for pixman_region_intersect. + * + * Results: + * TRUE if successful. + * + * Side Effects: + * Rectangles may be added to the region. + * + *----------------------------------------------------------------------- + */ +/*ARGSUSED*/ +static pixman_bool_t +pixman_region_intersect_o (region_type_t *region, + box_type_t * r1, + box_type_t * r1_end, + box_type_t * r2, + box_type_t * r2_end, + int y1, + int y2, + int * overlap) +{ + int x1; + int x2; + box_type_t * next_rect; + + next_rect = PIXREGION_TOP (region); + + critical_if_fail (y1 < y2); + critical_if_fail (r1 != r1_end && r2 != r2_end); + + do + { + x1 = MAX (r1->x1, r2->x1); + x2 = MIN (r1->x2, r2->x2); + + /* + * If there's any overlap between the two rectangles, add that + * overlap to the new region. + */ + if (x1 < x2) + NEWRECT (region, next_rect, x1, y1, x2, y2); + + /* + * Advance the pointer(s) with the leftmost right side, since the next + * rectangle on that list may still overlap the other region's + * current rectangle. + */ + if (r1->x2 == x2) + { + r1++; + } + if (r2->x2 == x2) + { + r2++; + } + } + while ((r1 != r1_end) && (r2 != r2_end)); + + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +PREFIX (_intersect) (region_type_t * new_reg, + region_type_t * reg1, + region_type_t * reg2) +{ + GOOD (reg1); + GOOD (reg2); + GOOD (new_reg); + + /* check for trivial reject */ + if (PIXREGION_NIL (reg1) || PIXREGION_NIL (reg2) || + !EXTENTCHECK (®1->extents, ®2->extents)) + { + /* Covers about 20% of all cases */ + FREE_DATA (new_reg); + new_reg->extents.x2 = new_reg->extents.x1; + new_reg->extents.y2 = new_reg->extents.y1; + if (PIXREGION_NAR (reg1) || PIXREGION_NAR (reg2)) + { + new_reg->data = pixman_broken_data; + return FALSE; + } + else + { + new_reg->data = pixman_region_empty_data; + } + } + else if (!reg1->data && !reg2->data) + { + /* Covers about 80% of cases that aren't trivially rejected */ + new_reg->extents.x1 = MAX (reg1->extents.x1, reg2->extents.x1); + new_reg->extents.y1 = MAX (reg1->extents.y1, reg2->extents.y1); + new_reg->extents.x2 = MIN (reg1->extents.x2, reg2->extents.x2); + new_reg->extents.y2 = MIN (reg1->extents.y2, reg2->extents.y2); + + FREE_DATA (new_reg); + + new_reg->data = (region_data_type_t *)NULL; + } + else if (!reg2->data && SUBSUMES (®2->extents, ®1->extents)) + { + return PREFIX (_copy) (new_reg, reg1); + } + else if (!reg1->data && SUBSUMES (®1->extents, ®2->extents)) + { + return PREFIX (_copy) (new_reg, reg2); + } + else if (reg1 == reg2) + { + return PREFIX (_copy) (new_reg, reg1); + } + else + { + /* General purpose intersection */ + int overlap; /* result ignored */ + + if (!pixman_op (new_reg, reg1, reg2, pixman_region_intersect_o, FALSE, FALSE, + &overlap)) + { + return FALSE; + } + + pixman_set_extents (new_reg); + } + + GOOD (new_reg); + return(TRUE); +} + +#define MERGERECT(r) \ + do \ + { \ + if (r->x1 <= x2) \ + { \ + /* Merge with current rectangle */ \ + if (r->x1 < x2) \ + *overlap = TRUE; \ + \ + if (x2 < r->x2) \ + x2 = r->x2; \ + } \ + else \ + { \ + /* Add current rectangle, start new one */ \ + NEWRECT (region, next_rect, x1, y1, x2, y2); \ + x1 = r->x1; \ + x2 = r->x2; \ + } \ + r++; \ + } while (0) + +/*====================================================================== + * Region Union + *====================================================================*/ + +/*- + *----------------------------------------------------------------------- + * pixman_region_union_o -- + * Handle an overlapping band for the union operation. Picks the + * left-most rectangle each time and merges it into the region. + * + * Results: + * TRUE if successful. + * + * Side Effects: + * region is overwritten. + * overlap is set to TRUE if any boxes overlap. + * + *----------------------------------------------------------------------- + */ +static pixman_bool_t +pixman_region_union_o (region_type_t *region, + box_type_t * r1, + box_type_t * r1_end, + box_type_t * r2, + box_type_t * r2_end, + int y1, + int y2, + int * overlap) +{ + box_type_t *next_rect; + int x1; /* left and right side of current union */ + int x2; + + critical_if_fail (y1 < y2); + critical_if_fail (r1 != r1_end && r2 != r2_end); + + next_rect = PIXREGION_TOP (region); + + /* Start off current rectangle */ + if (r1->x1 < r2->x1) + { + x1 = r1->x1; + x2 = r1->x2; + r1++; + } + else + { + x1 = r2->x1; + x2 = r2->x2; + r2++; + } + while (r1 != r1_end && r2 != r2_end) + { + if (r1->x1 < r2->x1) + MERGERECT (r1); + else + MERGERECT (r2); + } + + /* Finish off whoever (if any) is left */ + if (r1 != r1_end) + { + do + { + MERGERECT (r1); + } + while (r1 != r1_end); + } + else if (r2 != r2_end) + { + do + { + MERGERECT (r2); + } + while (r2 != r2_end); + } + + /* Add current rectangle */ + NEWRECT (region, next_rect, x1, y1, x2, y2); + + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +PREFIX(_intersect_rect) (region_type_t *dest, + region_type_t *source, + int x, int y, + unsigned int width, + unsigned int height) +{ + region_type_t region; + + region.data = NULL; + region.extents.x1 = x; + region.extents.y1 = y; + region.extents.x2 = x + width; + region.extents.y2 = y + height; + + return PREFIX(_intersect) (dest, source, ®ion); +} + +/* Convenience function for performing union of region with a + * single rectangle + */ +PIXMAN_EXPORT pixman_bool_t +PREFIX (_union_rect) (region_type_t *dest, + region_type_t *source, + int x, + int y, + unsigned int width, + unsigned int height) +{ + region_type_t region; + + region.extents.x1 = x; + region.extents.y1 = y; + region.extents.x2 = x + width; + region.extents.y2 = y + height; + + if (!GOOD_RECT (®ion.extents)) + { + if (BAD_RECT (®ion.extents)) + _pixman_log_error (FUNC, "Invalid rectangle passed"); + return PREFIX (_copy) (dest, source); + } + + region.data = NULL; + + return PREFIX (_union) (dest, source, ®ion); +} + +PIXMAN_EXPORT pixman_bool_t +PREFIX (_union) (region_type_t *new_reg, + region_type_t *reg1, + region_type_t *reg2) +{ + int overlap; /* result ignored */ + + /* Return TRUE if some overlap + * between reg1, reg2 + */ + GOOD (reg1); + GOOD (reg2); + GOOD (new_reg); + + /* checks all the simple cases */ + + /* + * Region 1 and 2 are the same + */ + if (reg1 == reg2) + return PREFIX (_copy) (new_reg, reg1); + + /* + * Region 1 is empty + */ + if (PIXREGION_NIL (reg1)) + { + if (PIXREGION_NAR (reg1)) + return pixman_break (new_reg); + + if (new_reg != reg2) + return PREFIX (_copy) (new_reg, reg2); + + return TRUE; + } + + /* + * Region 2 is empty + */ + if (PIXREGION_NIL (reg2)) + { + if (PIXREGION_NAR (reg2)) + return pixman_break (new_reg); + + if (new_reg != reg1) + return PREFIX (_copy) (new_reg, reg1); + + return TRUE; + } + + /* + * Region 1 completely subsumes region 2 + */ + if (!reg1->data && SUBSUMES (®1->extents, ®2->extents)) + { + if (new_reg != reg1) + return PREFIX (_copy) (new_reg, reg1); + + return TRUE; + } + + /* + * Region 2 completely subsumes region 1 + */ + if (!reg2->data && SUBSUMES (®2->extents, ®1->extents)) + { + if (new_reg != reg2) + return PREFIX (_copy) (new_reg, reg2); + + return TRUE; + } + + if (!pixman_op (new_reg, reg1, reg2, pixman_region_union_o, TRUE, TRUE, &overlap)) + return FALSE; + + new_reg->extents.x1 = MIN (reg1->extents.x1, reg2->extents.x1); + new_reg->extents.y1 = MIN (reg1->extents.y1, reg2->extents.y1); + new_reg->extents.x2 = MAX (reg1->extents.x2, reg2->extents.x2); + new_reg->extents.y2 = MAX (reg1->extents.y2, reg2->extents.y2); + + GOOD (new_reg); + + return TRUE; +} + +/*====================================================================== + * Batch Rectangle Union + *====================================================================*/ + +#define EXCHANGE_RECTS(a, b) \ + { \ + box_type_t t; \ + t = rects[a]; \ + rects[a] = rects[b]; \ + rects[b] = t; \ + } + +static void +quick_sort_rects ( + box_type_t rects[], + int numRects) +{ + int y1; + int x1; + int i, j; + box_type_t *r; + + /* Always called with numRects > 1 */ + + do + { + if (numRects == 2) + { + if (rects[0].y1 > rects[1].y1 || + (rects[0].y1 == rects[1].y1 && rects[0].x1 > rects[1].x1)) + { + EXCHANGE_RECTS (0, 1); + } + + return; + } + + /* Choose partition element, stick in location 0 */ + EXCHANGE_RECTS (0, numRects >> 1); + y1 = rects[0].y1; + x1 = rects[0].x1; + + /* Partition array */ + i = 0; + j = numRects; + + do + { + r = &(rects[i]); + do + { + r++; + i++; + } + + while (i != numRects && (r->y1 < y1 || (r->y1 == y1 && r->x1 < x1))) + ; + + r = &(rects[j]); + do + { + r--; + j--; + } + while (y1 < r->y1 || (y1 == r->y1 && x1 < r->x1)); + + if (i < j) + EXCHANGE_RECTS (i, j); + } + while (i < j); + + /* Move partition element back to middle */ + EXCHANGE_RECTS (0, j); + + /* Recurse */ + if (numRects - j - 1 > 1) + quick_sort_rects (&rects[j + 1], numRects - j - 1); + + numRects = j; + } + while (numRects > 1); +} + +/*- + *----------------------------------------------------------------------- + * pixman_region_validate -- + * + * Take a ``region'' which is a non-y-x-banded random collection of + * rectangles, and compute a nice region which is the union of all the + * rectangles. + * + * Results: + * TRUE if successful. + * + * Side Effects: + * The passed-in ``region'' may be modified. + * overlap set to TRUE if any retangles overlapped, + * else FALSE; + * + * Strategy: + * Step 1. Sort the rectangles into ascending order with primary key y1 + * and secondary key x1. + * + * Step 2. Split the rectangles into the minimum number of proper y-x + * banded regions. This may require horizontally merging + * rectangles, and vertically coalescing bands. With any luck, + * this step in an identity transformation (ala the Box widget), + * or a coalescing into 1 box (ala Menus). + * + * Step 3. Merge the separate regions down to a single region by calling + * pixman_region_union. Maximize the work each pixman_region_union call does by using + * a binary merge. + * + *----------------------------------------------------------------------- + */ + +static pixman_bool_t +validate (region_type_t * badreg, + int * overlap) +{ + /* Descriptor for regions under construction in Step 2. */ + typedef struct + { + region_type_t reg; + int prev_band; + int cur_band; + } region_info_t; + + region_info_t stack_regions[64]; + + int numRects; /* Original numRects for badreg */ + region_info_t *ri; /* Array of current regions */ + int num_ri; /* Number of entries used in ri */ + int size_ri; /* Number of entries available in ri */ + int i; /* Index into rects */ + int j; /* Index into ri */ + region_info_t *rit; /* &ri[j] */ + region_type_t *reg; /* ri[j].reg */ + box_type_t *box; /* Current box in rects */ + box_type_t *ri_box; /* Last box in ri[j].reg */ + region_type_t *hreg; /* ri[j_half].reg */ + pixman_bool_t ret = TRUE; + + *overlap = FALSE; + if (!badreg->data) + { + GOOD (badreg); + return TRUE; + } + + numRects = badreg->data->numRects; + if (!numRects) + { + if (PIXREGION_NAR (badreg)) + return FALSE; + GOOD (badreg); + return TRUE; + } + + if (badreg->extents.x1 < badreg->extents.x2) + { + if ((numRects) == 1) + { + FREE_DATA (badreg); + badreg->data = (region_data_type_t *) NULL; + } + else + { + DOWNSIZE (badreg, numRects); + } + + GOOD (badreg); + + return TRUE; + } + + /* Step 1: Sort the rects array into ascending (y1, x1) order */ + quick_sort_rects (PIXREGION_BOXPTR (badreg), numRects); + + /* Step 2: Scatter the sorted array into the minimum number of regions */ + + /* Set up the first region to be the first rectangle in badreg */ + /* Note that step 2 code will never overflow the ri[0].reg rects array */ + ri = stack_regions; + size_ri = sizeof (stack_regions) / sizeof (stack_regions[0]); + num_ri = 1; + ri[0].prev_band = 0; + ri[0].cur_band = 0; + ri[0].reg = *badreg; + box = PIXREGION_BOXPTR (&ri[0].reg); + ri[0].reg.extents = *box; + ri[0].reg.data->numRects = 1; + badreg->extents = *pixman_region_empty_box; + badreg->data = pixman_region_empty_data; + + /* Now scatter rectangles into the minimum set of valid regions. If the + * next rectangle to be added to a region would force an existing rectangle + * in the region to be split up in order to maintain y-x banding, just + * forget it. Try the next region. If it doesn't fit cleanly into any + * region, make a new one. + */ + + for (i = numRects; --i > 0;) + { + box++; + /* Look for a region to append box to */ + for (j = num_ri, rit = ri; --j >= 0; rit++) + { + reg = &rit->reg; + ri_box = PIXREGION_END (reg); + + if (box->y1 == ri_box->y1 && box->y2 == ri_box->y2) + { + /* box is in same band as ri_box. Merge or append it */ + if (box->x1 <= ri_box->x2) + { + /* Merge it with ri_box */ + if (box->x1 < ri_box->x2) + *overlap = TRUE; + + if (box->x2 > ri_box->x2) + ri_box->x2 = box->x2; + } + else + { + RECTALLOC_BAIL (reg, 1, bail); + *PIXREGION_TOP (reg) = *box; + reg->data->numRects++; + } + + goto next_rect; /* So sue me */ + } + else if (box->y1 >= ri_box->y2) + { + /* Put box into new band */ + if (reg->extents.x2 < ri_box->x2) + reg->extents.x2 = ri_box->x2; + + if (reg->extents.x1 > box->x1) + reg->extents.x1 = box->x1; + + COALESCE (reg, rit->prev_band, rit->cur_band); + rit->cur_band = reg->data->numRects; + RECTALLOC_BAIL (reg, 1, bail); + *PIXREGION_TOP (reg) = *box; + reg->data->numRects++; + + goto next_rect; + } + /* Well, this region was inappropriate. Try the next one. */ + } /* for j */ + + /* Uh-oh. No regions were appropriate. Create a new one. */ + if (size_ri == num_ri) + { + size_t data_size; + + /* Oops, allocate space for new region information */ + size_ri <<= 1; + + data_size = size_ri * sizeof(region_info_t); + if (data_size / size_ri != sizeof(region_info_t)) + goto bail; + + if (ri == stack_regions) + { + rit = malloc (data_size); + if (!rit) + goto bail; + memcpy (rit, ri, num_ri * sizeof (region_info_t)); + } + else + { + rit = (region_info_t *) realloc (ri, data_size); + if (!rit) + goto bail; + } + ri = rit; + rit = &ri[num_ri]; + } + num_ri++; + rit->prev_band = 0; + rit->cur_band = 0; + rit->reg.extents = *box; + rit->reg.data = (region_data_type_t *)NULL; + + /* MUST force allocation */ + if (!pixman_rect_alloc (&rit->reg, (i + num_ri) / num_ri)) + goto bail; + + next_rect: ; + } /* for i */ + + /* Make a final pass over each region in order to COALESCE and set + * extents.x2 and extents.y2 + */ + for (j = num_ri, rit = ri; --j >= 0; rit++) + { + reg = &rit->reg; + ri_box = PIXREGION_END (reg); + reg->extents.y2 = ri_box->y2; + + if (reg->extents.x2 < ri_box->x2) + reg->extents.x2 = ri_box->x2; + + COALESCE (reg, rit->prev_band, rit->cur_band); + + if (reg->data->numRects == 1) /* keep unions happy below */ + { + FREE_DATA (reg); + reg->data = (region_data_type_t *)NULL; + } + } + + /* Step 3: Union all regions into a single region */ + while (num_ri > 1) + { + int half = num_ri / 2; + for (j = num_ri & 1; j < (half + (num_ri & 1)); j++) + { + reg = &ri[j].reg; + hreg = &ri[j + half].reg; + + if (!pixman_op (reg, reg, hreg, pixman_region_union_o, TRUE, TRUE, overlap)) + ret = FALSE; + + if (hreg->extents.x1 < reg->extents.x1) + reg->extents.x1 = hreg->extents.x1; + + if (hreg->extents.y1 < reg->extents.y1) + reg->extents.y1 = hreg->extents.y1; + + if (hreg->extents.x2 > reg->extents.x2) + reg->extents.x2 = hreg->extents.x2; + + if (hreg->extents.y2 > reg->extents.y2) + reg->extents.y2 = hreg->extents.y2; + + FREE_DATA (hreg); + } + + num_ri -= half; + + if (!ret) + goto bail; + } + + *badreg = ri[0].reg; + + if (ri != stack_regions) + free (ri); + + GOOD (badreg); + return ret; + +bail: + for (i = 0; i < num_ri; i++) + FREE_DATA (&ri[i].reg); + + if (ri != stack_regions) + free (ri); + + return pixman_break (badreg); +} + +/*====================================================================== + * Region Subtraction + *====================================================================*/ + +/*- + *----------------------------------------------------------------------- + * pixman_region_subtract_o -- + * Overlapping band subtraction. x1 is the left-most point not yet + * checked. + * + * Results: + * TRUE if successful. + * + * Side Effects: + * region may have rectangles added to it. + * + *----------------------------------------------------------------------- + */ +/*ARGSUSED*/ +static pixman_bool_t +pixman_region_subtract_o (region_type_t * region, + box_type_t * r1, + box_type_t * r1_end, + box_type_t * r2, + box_type_t * r2_end, + int y1, + int y2, + int * overlap) +{ + box_type_t * next_rect; + int x1; + + x1 = r1->x1; + + critical_if_fail (y1 < y2); + critical_if_fail (r1 != r1_end && r2 != r2_end); + + next_rect = PIXREGION_TOP (region); + + do + { + if (r2->x2 <= x1) + { + /* + * Subtrahend entirely to left of minuend: go to next subtrahend. + */ + r2++; + } + else if (r2->x1 <= x1) + { + /* + * Subtrahend preceeds minuend: nuke left edge of minuend. + */ + x1 = r2->x2; + if (x1 >= r1->x2) + { + /* + * Minuend completely covered: advance to next minuend and + * reset left fence to edge of new minuend. + */ + r1++; + if (r1 != r1_end) + x1 = r1->x1; + } + else + { + /* + * Subtrahend now used up since it doesn't extend beyond + * minuend + */ + r2++; + } + } + else if (r2->x1 < r1->x2) + { + /* + * Left part of subtrahend covers part of minuend: add uncovered + * part of minuend to region and skip to next subtrahend. + */ + critical_if_fail (x1 < r2->x1); + NEWRECT (region, next_rect, x1, y1, r2->x1, y2); + + x1 = r2->x2; + if (x1 >= r1->x2) + { + /* + * Minuend used up: advance to new... + */ + r1++; + if (r1 != r1_end) + x1 = r1->x1; + } + else + { + /* + * Subtrahend used up + */ + r2++; + } + } + else + { + /* + * Minuend used up: add any remaining piece before advancing. + */ + if (r1->x2 > x1) + NEWRECT (region, next_rect, x1, y1, r1->x2, y2); + + r1++; + + if (r1 != r1_end) + x1 = r1->x1; + } + } + while ((r1 != r1_end) && (r2 != r2_end)); + + /* + * Add remaining minuend rectangles to region. + */ + while (r1 != r1_end) + { + critical_if_fail (x1 < r1->x2); + + NEWRECT (region, next_rect, x1, y1, r1->x2, y2); + + r1++; + if (r1 != r1_end) + x1 = r1->x1; + } + return TRUE; +} + +/*- + *----------------------------------------------------------------------- + * pixman_region_subtract -- + * Subtract reg_s from reg_m and leave the result in reg_d. + * S stands for subtrahend, M for minuend and D for difference. + * + * Results: + * TRUE if successful. + * + * Side Effects: + * reg_d is overwritten. + * + *----------------------------------------------------------------------- + */ +PIXMAN_EXPORT pixman_bool_t +PREFIX (_subtract) (region_type_t *reg_d, + region_type_t *reg_m, + region_type_t *reg_s) +{ + int overlap; /* result ignored */ + + GOOD (reg_m); + GOOD (reg_s); + GOOD (reg_d); + + /* check for trivial rejects */ + if (PIXREGION_NIL (reg_m) || PIXREGION_NIL (reg_s) || + !EXTENTCHECK (®_m->extents, ®_s->extents)) + { + if (PIXREGION_NAR (reg_s)) + return pixman_break (reg_d); + + return PREFIX (_copy) (reg_d, reg_m); + } + else if (reg_m == reg_s) + { + FREE_DATA (reg_d); + reg_d->extents.x2 = reg_d->extents.x1; + reg_d->extents.y2 = reg_d->extents.y1; + reg_d->data = pixman_region_empty_data; + + return TRUE; + } + + /* Add those rectangles in region 1 that aren't in region 2, + do yucky substraction for overlaps, and + just throw away rectangles in region 2 that aren't in region 1 */ + if (!pixman_op (reg_d, reg_m, reg_s, pixman_region_subtract_o, TRUE, FALSE, &overlap)) + return FALSE; + + /* + * Can't alter reg_d's extents before we call pixman_op because + * it might be one of the source regions and pixman_op depends + * on the extents of those regions being unaltered. Besides, this + * way there's no checking against rectangles that will be nuked + * due to coalescing, so we have to examine fewer rectangles. + */ + pixman_set_extents (reg_d); + GOOD (reg_d); + return TRUE; +} + +/*====================================================================== + * Region Inversion + *====================================================================*/ + +/*- + *----------------------------------------------------------------------- + * pixman_region_inverse -- + * Take a region and a box and return a region that is everything + * in the box but not in the region. The careful reader will note + * that this is the same as subtracting the region from the box... + * + * Results: + * TRUE. + * + * Side Effects: + * new_reg is overwritten. + * + *----------------------------------------------------------------------- + */ +pixman_bool_t +PIXMAN_EXPORT PREFIX (_inverse) (region_type_t *new_reg, /* Destination region */ + region_type_t *reg1, /* Region to invert */ + box_type_t * inv_rect) /* Bounding box for inversion */ +{ + region_type_t inv_reg; /* Quick and dirty region made from the + * bounding box */ + int overlap; /* result ignored */ + + GOOD (reg1); + GOOD (new_reg); + + /* check for trivial rejects */ + if (PIXREGION_NIL (reg1) || !EXTENTCHECK (inv_rect, ®1->extents)) + { + if (PIXREGION_NAR (reg1)) + return pixman_break (new_reg); + + new_reg->extents = *inv_rect; + FREE_DATA (new_reg); + new_reg->data = (region_data_type_t *)NULL; + + return TRUE; + } + + /* Add those rectangles in region 1 that aren't in region 2, + * do yucky substraction for overlaps, and + * just throw away rectangles in region 2 that aren't in region 1 + */ + inv_reg.extents = *inv_rect; + inv_reg.data = (region_data_type_t *)NULL; + if (!pixman_op (new_reg, &inv_reg, reg1, pixman_region_subtract_o, TRUE, FALSE, &overlap)) + return FALSE; + + /* + * Can't alter new_reg's extents before we call pixman_op because + * it might be one of the source regions and pixman_op depends + * on the extents of those regions being unaltered. Besides, this + * way there's no checking against rectangles that will be nuked + * due to coalescing, so we have to examine fewer rectangles. + */ + pixman_set_extents (new_reg); + GOOD (new_reg); + return TRUE; +} + +/* + * rect_in(region, rect) + * This routine takes a pointer to a region and a pointer to a box + * and determines if the box is outside/inside/partly inside the region. + * + * The idea is to travel through the list of rectangles trying to cover the + * passed box with them. Anytime a piece of the rectangle isn't covered + * by a band of rectangles, part_out is set TRUE. Any time a rectangle in + * the region covers part of the box, part_in is set TRUE. The process ends + * when either the box has been completely covered (we reached a band that + * doesn't overlap the box, part_in is TRUE and part_out is false), the + * box has been partially covered (part_in == part_out == TRUE -- because of + * the banding, the first time this is true we know the box is only + * partially in the region) or is outside the region (we reached a band + * that doesn't overlap the box at all and part_in is false) + */ + +pixman_region_overlap_t +PIXMAN_EXPORT PREFIX (_contains_rectangle) (region_type_t * region, + box_type_t * prect) +{ + box_type_t * pbox; + box_type_t * pbox_end; + int part_in, part_out; + int numRects; + int x, y; + + GOOD (region); + + numRects = PIXREGION_NUMRECTS (region); + + /* useful optimization */ + if (!numRects || !EXTENTCHECK (®ion->extents, prect)) + return(PIXMAN_REGION_OUT); + + if (numRects == 1) + { + /* We know that it must be PIXMAN_REGION_IN or PIXMAN_REGION_PART */ + if (SUBSUMES (®ion->extents, prect)) + return(PIXMAN_REGION_IN); + else + return(PIXMAN_REGION_PART); + } + + part_out = FALSE; + part_in = FALSE; + + /* (x,y) starts at upper left of rect, moving to the right and down */ + x = prect->x1; + y = prect->y1; + + /* can stop when both part_out and part_in are TRUE, or we reach prect->y2 */ + for (pbox = PIXREGION_BOXPTR (region), pbox_end = pbox + numRects; + pbox != pbox_end; + pbox++) + { + + if (pbox->y2 <= y) + continue; /* getting up to speed or skipping remainder of band */ + + if (pbox->y1 > y) + { + part_out = TRUE; /* missed part of rectangle above */ + if (part_in || (pbox->y1 >= prect->y2)) + break; + y = pbox->y1; /* x guaranteed to be == prect->x1 */ + } + + if (pbox->x2 <= x) + continue; /* not far enough over yet */ + + if (pbox->x1 > x) + { + part_out = TRUE; /* missed part of rectangle to left */ + if (part_in) + break; + } + + if (pbox->x1 < prect->x2) + { + part_in = TRUE; /* definitely overlap */ + if (part_out) + break; + } + + if (pbox->x2 >= prect->x2) + { + y = pbox->y2; /* finished with this band */ + if (y >= prect->y2) + break; + x = prect->x1; /* reset x out to left again */ + } + else + { + /* + * Because boxes in a band are maximal width, if the first box + * to overlap the rectangle doesn't completely cover it in that + * band, the rectangle must be partially out, since some of it + * will be uncovered in that band. part_in will have been set true + * by now... + */ + part_out = TRUE; + break; + } + } + + if (part_in) + { + if (y < prect->y2) + return PIXMAN_REGION_PART; + else + return PIXMAN_REGION_IN; + } + else + { + return PIXMAN_REGION_OUT; + } +} + +/* PREFIX(_translate) (region, x, y) + * translates in place + */ + +PIXMAN_EXPORT void +PREFIX (_translate) (region_type_t *region, int x, int y) +{ + overflow_int_t x1, x2, y1, y2; + int nbox; + box_type_t * pbox; + + GOOD (region); + region->extents.x1 = x1 = region->extents.x1 + x; + region->extents.y1 = y1 = region->extents.y1 + y; + region->extents.x2 = x2 = region->extents.x2 + x; + region->extents.y2 = y2 = region->extents.y2 + y; + + if (((x1 - PIXMAN_REGION_MIN) | (y1 - PIXMAN_REGION_MIN) | (PIXMAN_REGION_MAX - x2) | (PIXMAN_REGION_MAX - y2)) >= 0) + { + if (region->data && (nbox = region->data->numRects)) + { + for (pbox = PIXREGION_BOXPTR (region); nbox--; pbox++) + { + pbox->x1 += x; + pbox->y1 += y; + pbox->x2 += x; + pbox->y2 += y; + } + } + return; + } + + if (((x2 - PIXMAN_REGION_MIN) | (y2 - PIXMAN_REGION_MIN) | (PIXMAN_REGION_MAX - x1) | (PIXMAN_REGION_MAX - y1)) <= 0) + { + region->extents.x2 = region->extents.x1; + region->extents.y2 = region->extents.y1; + FREE_DATA (region); + region->data = pixman_region_empty_data; + return; + } + + if (x1 < PIXMAN_REGION_MIN) + region->extents.x1 = PIXMAN_REGION_MIN; + else if (x2 > PIXMAN_REGION_MAX) + region->extents.x2 = PIXMAN_REGION_MAX; + + if (y1 < PIXMAN_REGION_MIN) + region->extents.y1 = PIXMAN_REGION_MIN; + else if (y2 > PIXMAN_REGION_MAX) + region->extents.y2 = PIXMAN_REGION_MAX; + + if (region->data && (nbox = region->data->numRects)) + { + box_type_t * pbox_out; + + for (pbox_out = pbox = PIXREGION_BOXPTR (region); nbox--; pbox++) + { + pbox_out->x1 = x1 = pbox->x1 + x; + pbox_out->y1 = y1 = pbox->y1 + y; + pbox_out->x2 = x2 = pbox->x2 + x; + pbox_out->y2 = y2 = pbox->y2 + y; + + if (((x2 - PIXMAN_REGION_MIN) | (y2 - PIXMAN_REGION_MIN) | + (PIXMAN_REGION_MAX - x1) | (PIXMAN_REGION_MAX - y1)) <= 0) + { + region->data->numRects--; + continue; + } + + if (x1 < PIXMAN_REGION_MIN) + pbox_out->x1 = PIXMAN_REGION_MIN; + else if (x2 > PIXMAN_REGION_MAX) + pbox_out->x2 = PIXMAN_REGION_MAX; + + if (y1 < PIXMAN_REGION_MIN) + pbox_out->y1 = PIXMAN_REGION_MIN; + else if (y2 > PIXMAN_REGION_MAX) + pbox_out->y2 = PIXMAN_REGION_MAX; + + pbox_out++; + } + + if (pbox_out != pbox) + { + if (region->data->numRects == 1) + { + region->extents = *PIXREGION_BOXPTR (region); + FREE_DATA (region); + region->data = (region_data_type_t *)NULL; + } + else + { + pixman_set_extents (region); + } + } + } + + GOOD (region); +} + +PIXMAN_EXPORT void +PREFIX (_reset) (region_type_t *region, box_type_t *box) +{ + GOOD (region); + + critical_if_fail (GOOD_RECT (box)); + + region->extents = *box; + + FREE_DATA (region); + + region->data = NULL; +} + +/* box is "return" value */ +PIXMAN_EXPORT int +PREFIX (_contains_point) (region_type_t * region, + int x, int y, + box_type_t * box) +{ + box_type_t *pbox, *pbox_end; + int numRects; + + GOOD (region); + numRects = PIXREGION_NUMRECTS (region); + + if (!numRects || !INBOX (®ion->extents, x, y)) + return(FALSE); + + if (numRects == 1) + { + if (box) + *box = region->extents; + + return(TRUE); + } + + for (pbox = PIXREGION_BOXPTR (region), pbox_end = pbox + numRects; + pbox != pbox_end; + pbox++) + { + if (y >= pbox->y2) + continue; /* not there yet */ + + if ((y < pbox->y1) || (x < pbox->x1)) + break; /* missed it */ + + if (x >= pbox->x2) + continue; /* not there yet */ + + if (box) + *box = *pbox; + + return(TRUE); + } + + return(FALSE); +} + +PIXMAN_EXPORT int +PREFIX (_not_empty) (region_type_t * region) +{ + GOOD (region); + + return(!PIXREGION_NIL (region)); +} + +PIXMAN_EXPORT box_type_t * +PREFIX (_extents) (region_type_t * region) +{ + GOOD (region); + + return(®ion->extents); +} + +/* + * Clip a list of scanlines to a region. The caller has allocated the + * space. FSorted is non-zero if the scanline origins are in ascending order. + * + * returns the number of new, clipped scanlines. + */ + +PIXMAN_EXPORT pixman_bool_t +PREFIX (_selfcheck) (region_type_t *reg) +{ + int i, numRects; + + if ((reg->extents.x1 > reg->extents.x2) || + (reg->extents.y1 > reg->extents.y2)) + { + return FALSE; + } + + numRects = PIXREGION_NUMRECTS (reg); + if (!numRects) + { + return ((reg->extents.x1 == reg->extents.x2) && + (reg->extents.y1 == reg->extents.y2) && + (reg->data->size || (reg->data == pixman_region_empty_data))); + } + else if (numRects == 1) + { + return (!reg->data); + } + else + { + box_type_t * pbox_p, * pbox_n; + box_type_t box; + + pbox_p = PIXREGION_RECTS (reg); + box = *pbox_p; + box.y2 = pbox_p[numRects - 1].y2; + pbox_n = pbox_p + 1; + + for (i = numRects; --i > 0; pbox_p++, pbox_n++) + { + if ((pbox_n->x1 >= pbox_n->x2) || + (pbox_n->y1 >= pbox_n->y2)) + { + return FALSE; + } + + if (pbox_n->x1 < box.x1) + box.x1 = pbox_n->x1; + + if (pbox_n->x2 > box.x2) + box.x2 = pbox_n->x2; + + if ((pbox_n->y1 < pbox_p->y1) || + ((pbox_n->y1 == pbox_p->y1) && + ((pbox_n->x1 < pbox_p->x2) || (pbox_n->y2 != pbox_p->y2)))) + { + return FALSE; + } + } + + return ((box.x1 == reg->extents.x1) && + (box.x2 == reg->extents.x2) && + (box.y1 == reg->extents.y1) && + (box.y2 == reg->extents.y2)); + } +} + +PIXMAN_EXPORT pixman_bool_t +PREFIX (_init_rects) (region_type_t *region, + const box_type_t *boxes, int count) +{ + box_type_t *rects; + int displacement; + int i; + + /* if it's 1, then we just want to set the extents, so call + * the existing method. */ + if (count == 1) + { + PREFIX (_init_rect) (region, + boxes[0].x1, + boxes[0].y1, + boxes[0].x2 - boxes[0].x1, + boxes[0].y2 - boxes[0].y1); + return TRUE; + } + + PREFIX (_init) (region); + + /* if it's 0, don't call pixman_rect_alloc -- 0 rectangles is + * a special case, and causing pixman_rect_alloc would cause + * us to leak memory (because the 0-rect case should be the + * static pixman_region_empty_data data). + */ + if (count == 0) + return TRUE; + + if (!pixman_rect_alloc (region, count)) + return FALSE; + + rects = PIXREGION_RECTS (region); + + /* Copy in the rects */ + memcpy (rects, boxes, sizeof(box_type_t) * count); + region->data->numRects = count; + + /* Eliminate empty and malformed rectangles */ + displacement = 0; + + for (i = 0; i < count; ++i) + { + box_type_t *box = &rects[i]; + + if (box->x1 >= box->x2 || box->y1 >= box->y2) + displacement++; + else if (displacement) + rects[i - displacement] = rects[i]; + } + + region->data->numRects -= displacement; + + /* If eliminating empty rectangles caused there + * to be only 0 or 1 rectangles, deal with that. + */ + if (region->data->numRects == 0) + { + FREE_DATA (region); + PREFIX (_init) (region); + + return TRUE; + } + + if (region->data->numRects == 1) + { + region->extents = rects[0]; + + FREE_DATA (region); + region->data = NULL; + + GOOD (region); + + return TRUE; + } + + /* Validate */ + region->extents.x1 = region->extents.x2 = 0; + + return validate (region, &i); +} + +#define READ(_ptr) (*(_ptr)) + +static inline box_type_t * +bitmap_addrect (region_type_t *reg, + box_type_t *r, + box_type_t **first_rect, + int rx1, int ry1, + int rx2, int ry2) +{ + if ((rx1 < rx2) && (ry1 < ry2) && + (!(reg->data->numRects && + ((r-1)->y1 == ry1) && ((r-1)->y2 == ry2) && + ((r-1)->x1 <= rx1) && ((r-1)->x2 >= rx2)))) + { + if (!reg->data || + reg->data->numRects == reg->data->size) + { + if (!pixman_rect_alloc (reg, 1)) + return NULL; + *first_rect = PIXREGION_BOXPTR(reg); + r = *first_rect + reg->data->numRects; + } + r->x1 = rx1; + r->y1 = ry1; + r->x2 = rx2; + r->y2 = ry2; + reg->data->numRects++; + if (r->x1 < reg->extents.x1) + reg->extents.x1 = r->x1; + if (r->x2 > reg->extents.x2) + reg->extents.x2 = r->x2; + r++; + } + return r; +} + +/* Convert bitmap clip mask into clipping region. + * First, goes through each line and makes boxes by noting the transitions + * from 0 to 1 and 1 to 0. + * Then it coalesces the current line with the previous if they have boxes + * at the same X coordinates. + * Stride is in number of uint32_t per line. + */ +PIXMAN_EXPORT void +PREFIX (_init_from_image) (region_type_t *region, + pixman_image_t *image) +{ + uint32_t mask0 = 0xffffffff & ~SCREEN_SHIFT_RIGHT(0xffffffff, 1); + box_type_t *first_rect, *rects, *prect_line_start; + box_type_t *old_rect, *new_rect; + uint32_t *pw, w, *pw_line, *pw_line_end; + int irect_prev_start, irect_line_start; + int h, base, rx1 = 0, crects; + int ib; + pixman_bool_t in_box, same; + int width, height, stride; + + PREFIX(_init) (region); + + return_if_fail (image->type == BITS); + return_if_fail (image->bits.format == PIXMAN_a1); + + pw_line = pixman_image_get_data (image); + width = pixman_image_get_width (image); + height = pixman_image_get_height (image); + stride = pixman_image_get_stride (image) / 4; + + first_rect = PIXREGION_BOXPTR(region); + rects = first_rect; + + region->extents.x1 = width - 1; + region->extents.x2 = 0; + irect_prev_start = -1; + for (h = 0; h < height; h++) + { + pw = pw_line; + pw_line += stride; + irect_line_start = rects - first_rect; + + /* If the Screen left most bit of the word is set, we're starting in + * a box */ + if (READ(pw) & mask0) + { + in_box = TRUE; + rx1 = 0; + } + else + { + in_box = FALSE; + } + + /* Process all words which are fully in the pixmap */ + pw_line_end = pw + (width >> 5); + for (base = 0; pw < pw_line_end; base += 32) + { + w = READ(pw++); + if (in_box) + { + if (!~w) + continue; + } + else + { + if (!w) + continue; + } + for (ib = 0; ib < 32; ib++) + { + /* If the Screen left most bit of the word is set, we're + * starting a box */ + if (w & mask0) + { + if (!in_box) + { + rx1 = base + ib; + /* start new box */ + in_box = TRUE; + } + } + else + { + if (in_box) + { + /* end box */ + rects = bitmap_addrect (region, rects, &first_rect, + rx1, h, base + ib, h + 1); + if (rects == NULL) + goto error; + in_box = FALSE; + } + } + /* Shift the word VISUALLY left one. */ + w = SCREEN_SHIFT_LEFT(w, 1); + } + } + + if (width & 31) + { + /* Process final partial word on line */ + w = READ(pw++); + for (ib = 0; ib < (width & 31); ib++) + { + /* If the Screen left most bit of the word is set, we're + * starting a box */ + if (w & mask0) + { + if (!in_box) + { + rx1 = base + ib; + /* start new box */ + in_box = TRUE; + } + } + else + { + if (in_box) + { + /* end box */ + rects = bitmap_addrect(region, rects, &first_rect, + rx1, h, base + ib, h + 1); + if (rects == NULL) + goto error; + in_box = FALSE; + } + } + /* Shift the word VISUALLY left one. */ + w = SCREEN_SHIFT_LEFT(w, 1); + } + } + /* If scanline ended with last bit set, end the box */ + if (in_box) + { + rects = bitmap_addrect(region, rects, &first_rect, + rx1, h, base + (width & 31), h + 1); + if (rects == NULL) + goto error; + } + /* if all rectangles on this line have the same x-coords as + * those on the previous line, then add 1 to all the previous y2s and + * throw away all the rectangles from this line + */ + same = FALSE; + if (irect_prev_start != -1) + { + crects = irect_line_start - irect_prev_start; + if (crects != 0 && + crects == ((rects - first_rect) - irect_line_start)) + { + old_rect = first_rect + irect_prev_start; + new_rect = prect_line_start = first_rect + irect_line_start; + same = TRUE; + while (old_rect < prect_line_start) + { + if ((old_rect->x1 != new_rect->x1) || + (old_rect->x2 != new_rect->x2)) + { + same = FALSE; + break; + } + old_rect++; + new_rect++; + } + if (same) + { + old_rect = first_rect + irect_prev_start; + while (old_rect < prect_line_start) + { + old_rect->y2 += 1; + old_rect++; + } + rects -= crects; + region->data->numRects -= crects; + } + } + } + if(!same) + irect_prev_start = irect_line_start; + } + if (!region->data->numRects) + { + region->extents.x1 = region->extents.x2 = 0; + } + else + { + region->extents.y1 = PIXREGION_BOXPTR(region)->y1; + region->extents.y2 = PIXREGION_END(region)->y2; + if (region->data->numRects == 1) + { + free (region->data); + region->data = NULL; + } + } + + error: + return; +} diff --git a/programs/develop/libraries/pixman/pixman-region16.c b/programs/develop/libraries/pixman/pixman-region16.c new file mode 100644 index 0000000000..d88d3380f8 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-region16.c @@ -0,0 +1,67 @@ +/* + * Copyright © 2008 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software + * and its documentation for any purpose is hereby granted without + * fee, provided that the above copyright notice appear in all copies + * and that both that copyright notice and this permission notice + * appear in supporting documentation, and that the name of + * Red Hat, Inc. not be used in advertising or publicity pertaining to + * distribution of the software without specific, written prior + * permission. Red Hat, Inc. makes no representations about the + * suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * RED HAT, INC. DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL RED HAT, INC. BE LIABLE FOR ANY SPECIAL, + * INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER + * RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR + * IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + * + * Author: Soren Sandmann + */ +#ifdef HAVE_CONFIG_H +#include +#endif + +#undef PIXMAN_DISABLE_DEPRECATED + +#include "pixman-private.h" + +#include + +typedef pixman_box16_t box_type_t; +typedef pixman_region16_data_t region_data_type_t; +typedef pixman_region16_t region_type_t; +typedef int32_t overflow_int_t; + +typedef struct { + int x, y; +} point_type_t; + +#define PREFIX(x) pixman_region##x + +#define PIXMAN_REGION_MAX INT16_MAX +#define PIXMAN_REGION_MIN INT16_MIN + +#include "pixman-region.c" + +/* This function exists only to make it possible to preserve the X ABI - + * it should go away at first opportunity. + * + * The problem is that the X ABI exports the three structs and has used + * them through macros. So the X server calls this function with + * the addresses of those structs which makes the existing code continue to + * work. + */ +PIXMAN_EXPORT void +pixman_region_set_static_pointers (pixman_box16_t *empty_box, + pixman_region16_data_t *empty_data, + pixman_region16_data_t *broken_data) +{ + pixman_region_empty_box = empty_box; + pixman_region_empty_data = empty_data; + pixman_broken_data = broken_data; +} diff --git a/programs/develop/libraries/pixman/pixman-region32.c b/programs/develop/libraries/pixman/pixman-region32.c new file mode 100644 index 0000000000..abd6b1a937 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-region32.c @@ -0,0 +1,47 @@ +/* + * Copyright © 2008 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software + * and its documentation for any purpose is hereby granted without + * fee, provided that the above copyright notice appear in all copies + * and that both that copyright notice and this permission notice + * appear in supporting documentation, and that the name of + * Red Hat, Inc. not be used in advertising or publicity pertaining to + * distribution of the software without specific, written prior + * permission. Red Hat, Inc. makes no representations about the + * suitability of this software for any purpose. It is provided "as + * is" without express or implied warranty. + * + * RED HAT, INC. DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL RED HAT, INC. BE LIABLE FOR ANY SPECIAL, + * INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER + * RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR + * IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + * + * Author: Soren Sandmann + */ +#ifdef HAVE_CONFIG_H +#include +#endif + +#include "pixman-private.h" + +#include + +typedef pixman_box32_t box_type_t; +typedef pixman_region32_data_t region_data_type_t; +typedef pixman_region32_t region_type_t; +typedef int64_t overflow_int_t; + +typedef struct { + int x, y; +} point_type_t; + +#define PREFIX(x) pixman_region32##x + +#define PIXMAN_REGION_MAX INT32_MAX +#define PIXMAN_REGION_MIN INT32_MIN + +#include "pixman-region.c" diff --git a/programs/develop/libraries/pixman/pixman-solid-fill.c b/programs/develop/libraries/pixman/pixman-solid-fill.c new file mode 100644 index 0000000000..1d911e99d9 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-solid-fill.c @@ -0,0 +1,117 @@ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007, 2009 Red Hat, Inc. + * Copyright © 2009 Soren Sandmann + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of SuSE not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. SuSE makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * SuSE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL SuSE + * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include "pixman-private.h" + +static void +solid_fill_get_scanline_32 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + uint32_t *end = buffer + width; + uint32_t color = image->solid.color_32; + + while (buffer < end) + *(buffer++) = color; + + return; +} + +static void +solid_fill_get_scanline_64 (pixman_image_t *image, + int x, + int y, + int width, + uint32_t * buffer, + const uint32_t *mask) +{ + uint64_t *b = (uint64_t *)buffer; + uint64_t *e = b + width; + uint64_t color = image->solid.color_64; + + while (b < e) + *(b++) = color; +} + +static source_image_class_t +solid_fill_classify (pixman_image_t *image, + int x, + int y, + int width, + int height) +{ + return SOURCE_IMAGE_CLASS_HORIZONTAL; +} + +static void +solid_fill_property_changed (pixman_image_t *image) +{ + image->common.get_scanline_32 = solid_fill_get_scanline_32; + image->common.get_scanline_64 = solid_fill_get_scanline_64; +} + +static uint32_t +color_to_uint32 (const pixman_color_t *color) +{ + return + (color->alpha >> 8 << 24) | + (color->red >> 8 << 16) | + (color->green & 0xff00) | + (color->blue >> 8); +} + +static uint64_t +color_to_uint64 (const pixman_color_t *color) +{ + return + ((uint64_t)color->alpha << 48) | + ((uint64_t)color->red << 32) | + ((uint64_t)color->green << 16) | + ((uint64_t)color->blue); +} + +PIXMAN_EXPORT pixman_image_t * +pixman_image_create_solid_fill (pixman_color_t *color) +{ + pixman_image_t *img = _pixman_image_allocate (); + + if (!img) + return NULL; + + img->type = SOLID; + img->solid.color = *color; + img->solid.color_32 = color_to_uint32 (color); + img->solid.color_64 = color_to_uint64 (color); + + img->common.classify = solid_fill_classify; + img->common.property_changed = solid_fill_property_changed; + + return img; +} + diff --git a/programs/develop/libraries/pixman/pixman-timer.c b/programs/develop/libraries/pixman/pixman-timer.c new file mode 100644 index 0000000000..f5ae18e89f --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-timer.c @@ -0,0 +1,66 @@ +/* + * Copyright © 2007 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Red Hat not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. Red Hat makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * RED HAT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL RED HAT + * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif + +#include +#include +#include "pixman-private.h" + +#ifdef PIXMAN_TIMERS + +static pixman_timer_t *timers; + +static void +dump_timers (void) +{ + pixman_timer_t *timer; + + for (timer = timers; timer != NULL; timer = timer->next) + { + printf ("%s: total: %llu n: %llu avg: %f\n", + timer->name, + timer->total, + timer->n_times, + timer->total / (double)timer->n_times); + } +} + +void +pixman_timer_register (pixman_timer_t *timer) +{ + static int initialized; + + int atexit (void (*function)(void)); + + if (!initialized) + { + atexit (dump_timers); + initialized = 1; + } + + timer->next = timers; + timers = timer; +} + +#endif diff --git a/programs/develop/libraries/pixman/pixman-trap.c b/programs/develop/libraries/pixman/pixman-trap.c new file mode 100644 index 0000000000..8353992c58 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-trap.c @@ -0,0 +1,392 @@ +/* + * Copyright © 2004 Keith Packard + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * KEITH PACKARD DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL KEITH PACKARD BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR + * PERFORMANCE OF THIS SOFTWARE. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif + +#include +#include "pixman-private.h" + +/* + * Compute the smallest value greater than or equal to y which is on a + * grid row. + */ + +PIXMAN_EXPORT pixman_fixed_t +pixman_sample_ceil_y (pixman_fixed_t y, int n) +{ + pixman_fixed_t f = pixman_fixed_frac (y); + pixman_fixed_t i = pixman_fixed_floor (y); + + f = DIV (f - Y_FRAC_FIRST (n) + (STEP_Y_SMALL (n) - pixman_fixed_e), STEP_Y_SMALL (n)) * STEP_Y_SMALL (n) + + Y_FRAC_FIRST (n); + + if (f > Y_FRAC_LAST (n)) + { + if (pixman_fixed_to_int (i) == 0x7fff) + { + f = 0xffff; /* saturate */ + } + else + { + f = Y_FRAC_FIRST (n); + i += pixman_fixed_1; + } + } + return (i | f); +} + +/* + * Compute the largest value strictly less than y which is on a + * grid row. + */ +PIXMAN_EXPORT pixman_fixed_t +pixman_sample_floor_y (pixman_fixed_t y, + int n) +{ + pixman_fixed_t f = pixman_fixed_frac (y); + pixman_fixed_t i = pixman_fixed_floor (y); + + f = DIV (f - pixman_fixed_e - Y_FRAC_FIRST (n), STEP_Y_SMALL (n)) * STEP_Y_SMALL (n) + + Y_FRAC_FIRST (n); + + if (f < Y_FRAC_FIRST (n)) + { + if (pixman_fixed_to_int (i) == 0x8000) + { + f = 0; /* saturate */ + } + else + { + f = Y_FRAC_LAST (n); + i -= pixman_fixed_1; + } + } + return (i | f); +} + +/* + * Step an edge by any amount (including negative values) + */ +PIXMAN_EXPORT void +pixman_edge_step (pixman_edge_t *e, + int n) +{ + pixman_fixed_48_16_t ne; + + e->x += n * e->stepx; + + ne = e->e + n * (pixman_fixed_48_16_t) e->dx; + + if (n >= 0) + { + if (ne > 0) + { + int nx = (ne + e->dy - 1) / e->dy; + e->e = ne - nx * (pixman_fixed_48_16_t) e->dy; + e->x += nx * e->signdx; + } + } + else + { + if (ne <= -e->dy) + { + int nx = (-ne) / e->dy; + e->e = ne + nx * (pixman_fixed_48_16_t) e->dy; + e->x -= nx * e->signdx; + } + } +} + +/* + * A private routine to initialize the multi-step + * elements of an edge structure + */ +static void +_pixman_edge_multi_init (pixman_edge_t * e, + int n, + pixman_fixed_t *stepx_p, + pixman_fixed_t *dx_p) +{ + pixman_fixed_t stepx; + pixman_fixed_48_16_t ne; + + ne = n * (pixman_fixed_48_16_t) e->dx; + stepx = n * e->stepx; + + if (ne > 0) + { + int nx = ne / e->dy; + ne -= nx * e->dy; + stepx += nx * e->signdx; + } + + *dx_p = ne; + *stepx_p = stepx; +} + +/* + * Initialize one edge structure given the line endpoints and a + * starting y value + */ +PIXMAN_EXPORT void +pixman_edge_init (pixman_edge_t *e, + int n, + pixman_fixed_t y_start, + pixman_fixed_t x_top, + pixman_fixed_t y_top, + pixman_fixed_t x_bot, + pixman_fixed_t y_bot) +{ + pixman_fixed_t dx, dy; + + e->x = x_top; + e->e = 0; + dx = x_bot - x_top; + dy = y_bot - y_top; + e->dy = dy; + e->dx = 0; + + if (dy) + { + if (dx >= 0) + { + e->signdx = 1; + e->stepx = dx / dy; + e->dx = dx % dy; + e->e = -dy; + } + else + { + e->signdx = -1; + e->stepx = -(-dx / dy); + e->dx = -dx % dy; + e->e = 0; + } + + _pixman_edge_multi_init (e, STEP_Y_SMALL (n), + &e->stepx_small, &e->dx_small); + + _pixman_edge_multi_init (e, STEP_Y_BIG (n), + &e->stepx_big, &e->dx_big); + } + pixman_edge_step (e, y_start - y_top); +} + +/* + * Initialize one edge structure given a line, starting y value + * and a pixel offset for the line + */ +PIXMAN_EXPORT void +pixman_line_fixed_edge_init (pixman_edge_t * e, + int n, + pixman_fixed_t y, + const pixman_line_fixed_t *line, + int x_off, + int y_off) +{ + pixman_fixed_t x_off_fixed = pixman_int_to_fixed (x_off); + pixman_fixed_t y_off_fixed = pixman_int_to_fixed (y_off); + const pixman_point_fixed_t *top, *bot; + + if (line->p1.y <= line->p2.y) + { + top = &line->p1; + bot = &line->p2; + } + else + { + top = &line->p2; + bot = &line->p1; + } + + pixman_edge_init (e, n, y, + top->x + x_off_fixed, + top->y + y_off_fixed, + bot->x + x_off_fixed, + bot->y + y_off_fixed); +} + +PIXMAN_EXPORT void +pixman_add_traps (pixman_image_t * image, + int16_t x_off, + int16_t y_off, + int ntrap, + pixman_trap_t * traps) +{ + int bpp; + int width; + int height; + + pixman_fixed_t x_off_fixed; + pixman_fixed_t y_off_fixed; + pixman_edge_t l, r; + pixman_fixed_t t, b; + + _pixman_image_validate (image); + + width = image->bits.width; + height = image->bits.height; + bpp = PIXMAN_FORMAT_BPP (image->bits.format); + + x_off_fixed = pixman_int_to_fixed (x_off); + y_off_fixed = pixman_int_to_fixed (y_off); + + while (ntrap--) + { + t = traps->top.y + y_off_fixed; + if (t < 0) + t = 0; + t = pixman_sample_ceil_y (t, bpp); + + b = traps->bot.y + y_off_fixed; + if (pixman_fixed_to_int (b) >= height) + b = pixman_int_to_fixed (height) - 1; + b = pixman_sample_floor_y (b, bpp); + + if (b >= t) + { + /* initialize edge walkers */ + pixman_edge_init (&l, bpp, t, + traps->top.l + x_off_fixed, + traps->top.y + y_off_fixed, + traps->bot.l + x_off_fixed, + traps->bot.y + y_off_fixed); + + pixman_edge_init (&r, bpp, t, + traps->top.r + x_off_fixed, + traps->top.y + y_off_fixed, + traps->bot.r + x_off_fixed, + traps->bot.y + y_off_fixed); + + pixman_rasterize_edges (image, &l, &r, t, b); + } + + traps++; + } +} + +#if 0 +static void +dump_image (pixman_image_t *image, + const char * title) +{ + int i, j; + + if (!image->type == BITS) + printf ("%s is not a regular image\n", title); + + if (!image->bits.format == PIXMAN_a8) + printf ("%s is not an alpha mask\n", title); + + printf ("\n\n\n%s: \n", title); + + for (i = 0; i < image->bits.height; ++i) + { + uint8_t *line = + (uint8_t *)&(image->bits.bits[i * image->bits.rowstride]); + + for (j = 0; j < image->bits.width; ++j) + printf ("%c", line[j] ? '#' : ' '); + + printf ("\n"); + } +} +#endif + +PIXMAN_EXPORT void +pixman_add_trapezoids (pixman_image_t * image, + int16_t x_off, + int y_off, + int ntraps, + const pixman_trapezoid_t *traps) +{ + int i; + +#if 0 + dump_image (image, "before"); +#endif + + for (i = 0; i < ntraps; ++i) + { + const pixman_trapezoid_t *trap = &(traps[i]); + + if (!pixman_trapezoid_valid (trap)) + continue; + + pixman_rasterize_trapezoid (image, trap, x_off, y_off); + } + +#if 0 + dump_image (image, "after"); +#endif +} + +PIXMAN_EXPORT void +pixman_rasterize_trapezoid (pixman_image_t * image, + const pixman_trapezoid_t *trap, + int x_off, + int y_off) +{ + int bpp; + int width; + int height; + + pixman_fixed_t x_off_fixed; + pixman_fixed_t y_off_fixed; + pixman_edge_t l, r; + pixman_fixed_t t, b; + + return_if_fail (image->type == BITS); + + _pixman_image_validate (image); + + if (!pixman_trapezoid_valid (trap)) + return; + + width = image->bits.width; + height = image->bits.height; + bpp = PIXMAN_FORMAT_BPP (image->bits.format); + + x_off_fixed = pixman_int_to_fixed (x_off); + y_off_fixed = pixman_int_to_fixed (y_off); + + t = trap->top + y_off_fixed; + if (t < 0) + t = 0; + t = pixman_sample_ceil_y (t, bpp); + + b = trap->bottom + y_off_fixed; + if (pixman_fixed_to_int (b) >= height) + b = pixman_int_to_fixed (height) - 1; + b = pixman_sample_floor_y (b, bpp); + + if (b >= t) + { + /* initialize edge walkers */ + pixman_line_fixed_edge_init (&l, bpp, t, &trap->left, x_off, y_off); + pixman_line_fixed_edge_init (&r, bpp, t, &trap->right, x_off, y_off); + + pixman_rasterize_edges (image, &l, &r, t, b); + } +} diff --git a/programs/develop/libraries/pixman/pixman-utils.c b/programs/develop/libraries/pixman/pixman-utils.c new file mode 100644 index 0000000000..3ef88b7532 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-utils.c @@ -0,0 +1,258 @@ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 1999 Keith Packard + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of SuSE not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. SuSE makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * SuSE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL SuSE + * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + * + * Author: Keith Packard, SuSE, Inc. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include +#include + +#include "pixman-private.h" + +pixman_bool_t +pixman_multiply_overflows_int (unsigned int a, + unsigned int b) +{ + return a >= INT32_MAX / b; +} + +pixman_bool_t +pixman_addition_overflows_int (unsigned int a, + unsigned int b) +{ + return a > INT32_MAX - b; +} + +void * +pixman_malloc_ab (unsigned int a, + unsigned int b) +{ + if (a >= INT32_MAX / b) + return NULL; + + return malloc (a * b); +} + +void * +pixman_malloc_abc (unsigned int a, + unsigned int b, + unsigned int c) +{ + if (a >= INT32_MAX / b) + return NULL; + else if (a * b >= INT32_MAX / c) + return NULL; + else + return malloc (a * b * c); +} + +/* + * Helper routine to expand a color component from 0 < n <= 8 bits to 16 + * bits by replication. + */ +static inline uint64_t +expand16 (const uint8_t val, int nbits) +{ + /* Start out with the high bit of val in the high bit of result. */ + uint16_t result = (uint16_t)val << (16 - nbits); + + if (nbits == 0) + return 0; + + /* Copy the bits in result, doubling the number of bits each time, until + * we fill all 16 bits. + */ + while (nbits < 16) + { + result |= result >> nbits; + nbits *= 2; + } + + return result; +} + +/* + * This function expands images from ARGB8 format to ARGB16. To preserve + * precision, it needs to know the original source format. For example, if the + * source was PIXMAN_x1r5g5b5 and the red component contained bits 12345, then + * the expanded value is 12345123. To correctly expand this to 16 bits, it + * should be 1234512345123451 and not 1234512312345123. + */ +void +pixman_expand (uint64_t * dst, + const uint32_t * src, + pixman_format_code_t format, + int width) +{ + /* + * Determine the sizes of each component and the masks and shifts + * required to extract them from the source pixel. + */ + const int a_size = PIXMAN_FORMAT_A (format), + r_size = PIXMAN_FORMAT_R (format), + g_size = PIXMAN_FORMAT_G (format), + b_size = PIXMAN_FORMAT_B (format); + const int a_shift = 32 - a_size, + r_shift = 24 - r_size, + g_shift = 16 - g_size, + b_shift = 8 - b_size; + const uint8_t a_mask = ~(~0 << a_size), + r_mask = ~(~0 << r_size), + g_mask = ~(~0 << g_size), + b_mask = ~(~0 << b_size); + int i; + + /* Start at the end so that we can do the expansion in place + * when src == dst + */ + for (i = width - 1; i >= 0; i--) + { + const uint32_t pixel = src[i]; + const uint8_t a = (pixel >> a_shift) & a_mask, + r = (pixel >> r_shift) & r_mask, + g = (pixel >> g_shift) & g_mask, + b = (pixel >> b_shift) & b_mask; + const uint64_t a16 = a_size ? expand16 (a, a_size) : 0xffff, + r16 = expand16 (r, r_size), + g16 = expand16 (g, g_size), + b16 = expand16 (b, b_size); + + dst[i] = a16 << 48 | r16 << 32 | g16 << 16 | b16; + } +} + +/* + * Contracting is easier than expanding. We just need to truncate the + * components. + */ +void +pixman_contract (uint32_t * dst, + const uint64_t *src, + int width) +{ + int i; + + /* Start at the beginning so that we can do the contraction in + * place when src == dst + */ + for (i = 0; i < width; i++) + { + const uint8_t a = src[i] >> 56, + r = src[i] >> 40, + g = src[i] >> 24, + b = src[i] >> 8; + + dst[i] = a << 24 | r << 16 | g << 8 | b; + } +} + +#define N_TMP_BOXES (16) + +pixman_bool_t +pixman_region16_copy_from_region32 (pixman_region16_t *dst, + pixman_region32_t *src) +{ + int n_boxes, i; + pixman_box32_t *boxes32; + pixman_box16_t *boxes16; + pixman_bool_t retval; + + boxes32 = pixman_region32_rectangles (src, &n_boxes); + + boxes16 = pixman_malloc_ab (n_boxes, sizeof (pixman_box16_t)); + + if (!boxes16) + return FALSE; + + for (i = 0; i < n_boxes; ++i) + { + boxes16[i].x1 = boxes32[i].x1; + boxes16[i].y1 = boxes32[i].y1; + boxes16[i].x2 = boxes32[i].x2; + boxes16[i].y2 = boxes32[i].y2; + } + + pixman_region_fini (dst); + retval = pixman_region_init_rects (dst, boxes16, n_boxes); + free (boxes16); + return retval; +} + +pixman_bool_t +pixman_region32_copy_from_region16 (pixman_region32_t *dst, + pixman_region16_t *src) +{ + int n_boxes, i; + pixman_box16_t *boxes16; + pixman_box32_t *boxes32; + pixman_box32_t tmp_boxes[N_TMP_BOXES]; + pixman_bool_t retval; + + boxes16 = pixman_region_rectangles (src, &n_boxes); + + if (n_boxes > N_TMP_BOXES) + boxes32 = pixman_malloc_ab (n_boxes, sizeof (pixman_box32_t)); + else + boxes32 = tmp_boxes; + + if (!boxes32) + return FALSE; + + for (i = 0; i < n_boxes; ++i) + { + boxes32[i].x1 = boxes16[i].x1; + boxes32[i].y1 = boxes16[i].y1; + boxes32[i].x2 = boxes16[i].x2; + boxes32[i].y2 = boxes16[i].y2; + } + + pixman_region32_fini (dst); + retval = pixman_region32_init_rects (dst, boxes32, n_boxes); + + if (boxes32 != tmp_boxes) + free (boxes32); + + return retval; +} + +#ifdef DEBUG + +void +_pixman_log_error (const char *function, const char *message) +{ + static int n_messages = 0; + + if (n_messages < 10) + { + fprintf (stderr, + "*** BUG ***\n" + "In %s: %s\n" + "Set a breakpoint on '_pixman_log_error' to debug\n\n", + function, message); + + n_messages++; + } +} + +#endif diff --git a/programs/develop/libraries/pixman/pixman-version.h b/programs/develop/libraries/pixman/pixman-version.h new file mode 100644 index 0000000000..172beea370 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-version.h @@ -0,0 +1,50 @@ +/* + * Copyright © 2008 Red Hat, 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 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. + * + * Author: Carl D. Worth + */ + +#ifndef PIXMAN_VERSION_H__ +#define PIXMAN_VERSION_H__ + +#ifndef PIXMAN_H__ +# error pixman-version.h should only be included by pixman.h +#endif + +#define PIXMAN_VERSION_MAJOR 0 +#define PIXMAN_VERSION_MINOR 20 +#define PIXMAN_VERSION_MICRO 2 + +#define PIXMAN_VERSION_STRING "0.20.2" + +#define PIXMAN_VERSION_ENCODE(major, minor, micro) ( \ + ((major) * 10000) \ + + ((minor) * 100) \ + + ((micro) * 1)) + +#define PIXMAN_VERSION PIXMAN_VERSION_ENCODE( \ + PIXMAN_VERSION_MAJOR, \ + PIXMAN_VERSION_MINOR, \ + PIXMAN_VERSION_MICRO) + +#endif /* PIXMAN_VERSION_H__ */ diff --git a/programs/develop/libraries/pixman/pixman-x64-mmx-emulation.h b/programs/develop/libraries/pixman/pixman-x64-mmx-emulation.h new file mode 100644 index 0000000000..378019cf27 --- /dev/null +++ b/programs/develop/libraries/pixman/pixman-x64-mmx-emulation.h @@ -0,0 +1,263 @@ +#ifndef MMX_X64_H_INCLUDED +#define MMX_X64_H_INCLUDED + +/* Implementation of x64 MMX substitition functions, before + * pixman is reimplemented not to use __m64 type on Visual C++ + * + * Copyright (C)2009 by George Yohng + * Released in public domain. + */ + +#include + +#define M64C(a) (*(const __m64 *)(&a)) +#define M64U(a) (*(const unsigned long long *)(&a)) + +__inline __m64 +_m_from_int (int a) +{ + long long i64 = a; + + return M64C (i64); +} + +__inline __m64 +_mm_setzero_si64 () +{ + long long i64 = 0; + + return M64C (i64); +} + +__inline __m64 +_mm_set_pi32 (int i1, int i0) +{ + unsigned long long i64 = ((unsigned)i0) + (((unsigned long long)(unsigned)i1) << 32); + + return M64C (i64); +} + +__inline void +_m_empty () +{ +} + +__inline __m64 +_mm_set1_pi16 (short w) +{ + unsigned long long i64 = ((unsigned long long)(unsigned short)(w)) * 0x0001000100010001ULL; + + return M64C (i64); +} + +__inline int +_m_to_int (__m64 m) +{ + return m.m64_i32[0]; +} + +__inline __m64 +_mm_movepi64_pi64 (__m128i a) +{ + return M64C (a.m128i_i64[0]); +} + +__inline __m64 +_m_pand (__m64 a, __m64 b) +{ + unsigned long long i64 = M64U (a) & M64U (b); + + return M64C (i64); +} + +__inline __m64 +_m_por (__m64 a, __m64 b) +{ + unsigned long long i64 = M64U (a) | M64U (b); + + return M64C (i64); +} + +__inline __m64 +_m_pxor (__m64 a, __m64 b) +{ + unsigned long long i64 = M64U (a) ^ M64U (b); + + return M64C (i64); +} + +__inline __m64 +_m_pmulhuw (__m64 a, __m64 b) /* unoptimized */ +{ + unsigned short d[4] = + { + (unsigned short)((((unsigned)a.m64_u16[0]) * b.m64_u16[0]) >> 16), + (unsigned short)((((unsigned)a.m64_u16[1]) * b.m64_u16[1]) >> 16), + (unsigned short)((((unsigned)a.m64_u16[2]) * b.m64_u16[2]) >> 16), + (unsigned short)((((unsigned)a.m64_u16[3]) * b.m64_u16[3]) >> 16) + }; + + return M64C (d[0]); +} + +__inline __m64 +_m_pmullw2 (__m64 a, __m64 b) /* unoptimized */ +{ + unsigned short d[4] = + { + (unsigned short)((((unsigned)a.m64_u16[0]) * b.m64_u16[0])), + (unsigned short)((((unsigned)a.m64_u16[1]) * b.m64_u16[1])), + (unsigned short)((((unsigned)a.m64_u16[2]) * b.m64_u16[2])), + (unsigned short)((((unsigned)a.m64_u16[3]) * b.m64_u16[3])) + }; + + return M64C (d[0]); +} + +__inline __m64 +_m_pmullw (__m64 a, __m64 b) /* unoptimized */ +{ + unsigned long long x = + ((unsigned long long)(unsigned short)((((unsigned)a.m64_u16[0]) * b.m64_u16[0]))) + + (((unsigned long long)(unsigned short)((((unsigned)a.m64_u16[1]) * b.m64_u16[1]))) << 16) + + (((unsigned long long)(unsigned short)((((unsigned)a.m64_u16[2]) * b.m64_u16[2]))) << 32) + + (((unsigned long long)(unsigned short)((((unsigned)a.m64_u16[3]) * b.m64_u16[3]))) << 48); + + return M64C (x); +} + +__inline __m64 +_m_paddusb (__m64 a, __m64 b) /* unoptimized */ +{ + unsigned long long x = (M64U (a) & 0x00FF00FF00FF00FFULL) + + (M64U (b) & 0x00FF00FF00FF00FFULL); + + unsigned long long y = ((M64U (a) >> 8) & 0x00FF00FF00FF00FFULL) + + ((M64U (b) >> 8) & 0x00FF00FF00FF00FFULL); + + x |= ((x & 0xFF00FF00FF00FF00ULL) >> 8) * 0xFF; + y |= ((y & 0xFF00FF00FF00FF00ULL) >> 8) * 0xFF; + + x = (x & 0x00FF00FF00FF00FFULL) | ((y & 0x00FF00FF00FF00FFULL) << 8); + + return M64C (x); +} + +__inline __m64 +_m_paddusw (__m64 a, __m64 b) /* unoptimized */ +{ + unsigned long long x = (M64U (a) & 0x0000FFFF0000FFFFULL) + + (M64U (b) & 0x0000FFFF0000FFFFULL); + + unsigned long long y = ((M64U (a) >> 16) & 0x0000FFFF0000FFFFULL) + + ((M64U (b) >> 16) & 0x0000FFFF0000FFFFULL); + + x |= ((x & 0xFFFF0000FFFF0000) >> 16) * 0xFFFF; + y |= ((y & 0xFFFF0000FFFF0000) >> 16) * 0xFFFF; + + x = (x & 0x0000FFFF0000FFFFULL) | ((y & 0x0000FFFF0000FFFFULL) << 16); + + return M64C (x); +} + +__inline __m64 +_m_pshufw (__m64 a, int n) /* unoptimized */ +{ + unsigned short d[4] = + { + a.m64_u16[n & 3], + a.m64_u16[(n >> 2) & 3], + a.m64_u16[(n >> 4) & 3], + a.m64_u16[(n >> 6) & 3] + }; + + return M64C (d[0]); +} + +__inline unsigned char +sat16 (unsigned short d) +{ + if (d > 0xFF) return 0xFF; + else return d & 0xFF; +} + +__inline __m64 +_m_packuswb (__m64 m1, __m64 m2) /* unoptimized */ +{ + unsigned char d[8] = + { + sat16 (m1.m64_u16[0]), + sat16 (m1.m64_u16[1]), + sat16 (m1.m64_u16[2]), + sat16 (m1.m64_u16[3]), + sat16 (m2.m64_u16[0]), + sat16 (m2.m64_u16[1]), + sat16 (m2.m64_u16[2]), + sat16 (m2.m64_u16[3]) + }; + + return M64C (d[0]); +} + +__inline __m64 _m_punpcklbw (__m64 m1, __m64 m2) /* unoptimized */ +{ + unsigned char d[8] = + { + m1.m64_u8[0], + m2.m64_u8[0], + m1.m64_u8[1], + m2.m64_u8[1], + m1.m64_u8[2], + m2.m64_u8[2], + m1.m64_u8[3], + m2.m64_u8[3], + }; + + return M64C (d[0]); +} + +__inline __m64 _m_punpckhbw (__m64 m1, __m64 m2) /* unoptimized */ +{ + unsigned char d[8] = + { + m1.m64_u8[4], + m2.m64_u8[4], + m1.m64_u8[5], + m2.m64_u8[5], + m1.m64_u8[6], + m2.m64_u8[6], + m1.m64_u8[7], + m2.m64_u8[7], + }; + + return M64C (d[0]); +} + +__inline __m64 _m_psrlwi (__m64 a, int n) /* unoptimized */ +{ + unsigned short d[4] = + { + a.m64_u16[0] >> n, + a.m64_u16[1] >> n, + a.m64_u16[2] >> n, + a.m64_u16[3] >> n + }; + + return M64C (d[0]); +} + +__inline __m64 _m_psrlqi (__m64 m, int n) +{ + unsigned long long x = M64U (m) >> n; + + return M64C (x); +} + +__inline __m64 _m_psllqi (__m64 m, int n) +{ + unsigned long long x = M64U (m) << n; + + return M64C (x); +} + +#endif /* MMX_X64_H_INCLUDED */ diff --git a/programs/develop/libraries/pixman/pixman.c b/programs/develop/libraries/pixman/pixman.c new file mode 100644 index 0000000000..3a62b2d7df --- /dev/null +++ b/programs/develop/libraries/pixman/pixman.c @@ -0,0 +1,1274 @@ +/* -*- Mode: c; c-basic-offset: 4; tab-width: 8; indent-tabs-mode: t; -*- */ +/* + * Copyright © 2000 SuSE, Inc. + * Copyright © 2007 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of SuSE not be used in advertising or + * publicity pertaining to distribution of the software without specific, + * written prior permission. SuSE makes no representations about the + * suitability of this software for any purpose. It is provided "as is" + * without express or implied warranty. + * + * SuSE DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING ALL + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL SuSE + * BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION + * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN + * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + * + * Author: Keith Packard, SuSE, Inc. + */ + +#ifdef HAVE_CONFIG_H +#include +#endif +#include "pixman-private.h" + +#include + +static force_inline pixman_implementation_t * +get_implementation (void) +{ + static pixman_implementation_t *global_implementation; + + if (!global_implementation) + global_implementation = _pixman_choose_implementation (); + + return global_implementation; +} + +typedef struct operator_info_t operator_info_t; + +struct operator_info_t +{ + uint8_t opaque_info[4]; +}; + +#define PACK(neither, src, dest, both) \ + {{ (uint8_t)PIXMAN_OP_ ## neither, \ + (uint8_t)PIXMAN_OP_ ## src, \ + (uint8_t)PIXMAN_OP_ ## dest, \ + (uint8_t)PIXMAN_OP_ ## both }} + +static const operator_info_t operator_table[] = +{ + /* Neither Opaque Src Opaque Dst Opaque Both Opaque */ + PACK (CLEAR, CLEAR, CLEAR, CLEAR), + PACK (SRC, SRC, SRC, SRC), + PACK (DST, DST, DST, DST), + PACK (OVER, SRC, OVER, SRC), + PACK (OVER_REVERSE, OVER_REVERSE, DST, DST), + PACK (IN, IN, SRC, SRC), + PACK (IN_REVERSE, DST, IN_REVERSE, DST), + PACK (OUT, OUT, CLEAR, CLEAR), + PACK (OUT_REVERSE, CLEAR, OUT_REVERSE, CLEAR), + PACK (ATOP, IN, OVER, SRC), + PACK (ATOP_REVERSE, OVER_REVERSE, IN_REVERSE, DST), + PACK (XOR, OUT, OUT_REVERSE, CLEAR), + PACK (ADD, ADD, ADD, ADD), + PACK (SATURATE, OVER_REVERSE, DST, DST), + + {{ 0 /* 0x0e */ }}, + {{ 0 /* 0x0f */ }}, + + PACK (CLEAR, CLEAR, CLEAR, CLEAR), + PACK (SRC, SRC, SRC, SRC), + PACK (DST, DST, DST, DST), + PACK (DISJOINT_OVER, DISJOINT_OVER, DISJOINT_OVER, DISJOINT_OVER), + PACK (DISJOINT_OVER_REVERSE, DISJOINT_OVER_REVERSE, DISJOINT_OVER_REVERSE, DISJOINT_OVER_REVERSE), + PACK (DISJOINT_IN, DISJOINT_IN, DISJOINT_IN, DISJOINT_IN), + PACK (DISJOINT_IN_REVERSE, DISJOINT_IN_REVERSE, DISJOINT_IN_REVERSE, DISJOINT_IN_REVERSE), + PACK (DISJOINT_OUT, DISJOINT_OUT, DISJOINT_OUT, DISJOINT_OUT), + PACK (DISJOINT_OUT_REVERSE, DISJOINT_OUT_REVERSE, DISJOINT_OUT_REVERSE, DISJOINT_OUT_REVERSE), + PACK (DISJOINT_ATOP, DISJOINT_ATOP, DISJOINT_ATOP, DISJOINT_ATOP), + PACK (DISJOINT_ATOP_REVERSE, DISJOINT_ATOP_REVERSE, DISJOINT_ATOP_REVERSE, DISJOINT_ATOP_REVERSE), + PACK (DISJOINT_XOR, DISJOINT_XOR, DISJOINT_XOR, DISJOINT_XOR), + + {{ 0 /* 0x1c */ }}, + {{ 0 /* 0x1d */ }}, + {{ 0 /* 0x1e */ }}, + {{ 0 /* 0x1f */ }}, + + PACK (CLEAR, CLEAR, CLEAR, CLEAR), + PACK (SRC, SRC, SRC, SRC), + PACK (DST, DST, DST, DST), + PACK (CONJOINT_OVER, CONJOINT_OVER, CONJOINT_OVER, CONJOINT_OVER), + PACK (CONJOINT_OVER_REVERSE, CONJOINT_OVER_REVERSE, CONJOINT_OVER_REVERSE, CONJOINT_OVER_REVERSE), + PACK (CONJOINT_IN, CONJOINT_IN, CONJOINT_IN, CONJOINT_IN), + PACK (CONJOINT_IN_REVERSE, CONJOINT_IN_REVERSE, CONJOINT_IN_REVERSE, CONJOINT_IN_REVERSE), + PACK (CONJOINT_OUT, CONJOINT_OUT, CONJOINT_OUT, CONJOINT_OUT), + PACK (CONJOINT_OUT_REVERSE, CONJOINT_OUT_REVERSE, CONJOINT_OUT_REVERSE, CONJOINT_OUT_REVERSE), + PACK (CONJOINT_ATOP, CONJOINT_ATOP, CONJOINT_ATOP, CONJOINT_ATOP), + PACK (CONJOINT_ATOP_REVERSE, CONJOINT_ATOP_REVERSE, CONJOINT_ATOP_REVERSE, CONJOINT_ATOP_REVERSE), + PACK (CONJOINT_XOR, CONJOINT_XOR, CONJOINT_XOR, CONJOINT_XOR), + + {{ 0 /* 0x2c */ }}, + {{ 0 /* 0x2d */ }}, + {{ 0 /* 0x2e */ }}, + {{ 0 /* 0x2f */ }}, + + PACK (MULTIPLY, MULTIPLY, MULTIPLY, MULTIPLY), + PACK (SCREEN, SCREEN, SCREEN, SCREEN), + PACK (OVERLAY, OVERLAY, OVERLAY, OVERLAY), + PACK (DARKEN, DARKEN, DARKEN, DARKEN), + PACK (LIGHTEN, LIGHTEN, LIGHTEN, LIGHTEN), + PACK (COLOR_DODGE, COLOR_DODGE, COLOR_DODGE, COLOR_DODGE), + PACK (COLOR_BURN, COLOR_BURN, COLOR_BURN, COLOR_BURN), + PACK (HARD_LIGHT, HARD_LIGHT, HARD_LIGHT, HARD_LIGHT), + PACK (SOFT_LIGHT, SOFT_LIGHT, SOFT_LIGHT, SOFT_LIGHT), + PACK (DIFFERENCE, DIFFERENCE, DIFFERENCE, DIFFERENCE), + PACK (EXCLUSION, EXCLUSION, EXCLUSION, EXCLUSION), + PACK (HSL_HUE, HSL_HUE, HSL_HUE, HSL_HUE), + PACK (HSL_SATURATION, HSL_SATURATION, HSL_SATURATION, HSL_SATURATION), + PACK (HSL_COLOR, HSL_COLOR, HSL_COLOR, HSL_COLOR), + PACK (HSL_LUMINOSITY, HSL_LUMINOSITY, HSL_LUMINOSITY, HSL_LUMINOSITY), +}; + +/* + * Optimize the current operator based on opacity of source or destination + * The output operator should be mathematically equivalent to the source. + */ +static pixman_op_t +optimize_operator (pixman_op_t op, + uint32_t src_flags, + uint32_t mask_flags, + uint32_t dst_flags) +{ + pixman_bool_t is_source_opaque, is_dest_opaque; + +#define OPAQUE_SHIFT 13 + + COMPILE_TIME_ASSERT (FAST_PATH_IS_OPAQUE == (1 << OPAQUE_SHIFT)); + + is_dest_opaque = (dst_flags & FAST_PATH_IS_OPAQUE); + is_source_opaque = ((src_flags & mask_flags) & FAST_PATH_IS_OPAQUE); + + is_dest_opaque >>= OPAQUE_SHIFT - 1; + is_source_opaque >>= OPAQUE_SHIFT; + + return operator_table[op].opaque_info[is_dest_opaque | is_source_opaque]; +} + +static void +apply_workaround (pixman_image_t *image, + int32_t * x, + int32_t * y, + uint32_t ** save_bits, + int * save_dx, + int * save_dy) +{ + if (image && (image->common.flags & FAST_PATH_NEEDS_WORKAROUND)) + { + /* Some X servers generate images that point to the + * wrong place in memory, but then set the clip region + * to point to the right place. Because of an old bug + * in pixman, this would actually work. + * + * Here we try and undo the damage + */ + int bpp = PIXMAN_FORMAT_BPP (image->bits.format) / 8; + pixman_box32_t *extents; + uint8_t *t; + int dx, dy; + + extents = pixman_region32_extents (&(image->common.clip_region)); + dx = extents->x1; + dy = extents->y1; + + *save_bits = image->bits.bits; + + *x -= dx; + *y -= dy; + pixman_region32_translate (&(image->common.clip_region), -dx, -dy); + + t = (uint8_t *)image->bits.bits; + t += dy * image->bits.rowstride * 4 + dx * bpp; + image->bits.bits = (uint32_t *)t; + + *save_dx = dx; + *save_dy = dy; + } +} + +static void +unapply_workaround (pixman_image_t *image, uint32_t *bits, int dx, int dy) +{ + if (image && (image->common.flags & FAST_PATH_NEEDS_WORKAROUND)) + { + image->bits.bits = bits; + pixman_region32_translate (&image->common.clip_region, dx, dy); + } +} + +/* + * Computing composite region + */ +static inline pixman_bool_t +clip_general_image (pixman_region32_t * region, + pixman_region32_t * clip, + int dx, + int dy) +{ + if (pixman_region32_n_rects (region) == 1 && + pixman_region32_n_rects (clip) == 1) + { + pixman_box32_t * rbox = pixman_region32_rectangles (region, NULL); + pixman_box32_t * cbox = pixman_region32_rectangles (clip, NULL); + int v; + + if (rbox->x1 < (v = cbox->x1 + dx)) + rbox->x1 = v; + if (rbox->x2 > (v = cbox->x2 + dx)) + rbox->x2 = v; + if (rbox->y1 < (v = cbox->y1 + dy)) + rbox->y1 = v; + if (rbox->y2 > (v = cbox->y2 + dy)) + rbox->y2 = v; + if (rbox->x1 >= rbox->x2 || rbox->y1 >= rbox->y2) + { + pixman_region32_init (region); + return FALSE; + } + } + else if (!pixman_region32_not_empty (clip)) + { + return FALSE; + } + else + { + if (dx || dy) + pixman_region32_translate (region, -dx, -dy); + + if (!pixman_region32_intersect (region, region, clip)) + return FALSE; + + if (dx || dy) + pixman_region32_translate (region, dx, dy); + } + + return pixman_region32_not_empty (region); +} + +static inline pixman_bool_t +clip_source_image (pixman_region32_t * region, + pixman_image_t * image, + int dx, + int dy) +{ + /* Source clips are ignored, unless they are explicitly turned on + * and the clip in question was set by an X client. (Because if + * the clip was not set by a client, then it is a hierarchy + * clip and those should always be ignored for sources). + */ + if (!image->common.clip_sources || !image->common.client_clip) + return TRUE; + + return clip_general_image (region, + &image->common.clip_region, + dx, dy); +} + +/* + * returns FALSE if the final region is empty. Indistinguishable from + * an allocation failure, but rendering ignores those anyways. + */ +static pixman_bool_t +pixman_compute_composite_region32 (pixman_region32_t * region, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + region->extents.x1 = dest_x; + region->extents.x2 = dest_x + width; + region->extents.y1 = dest_y; + region->extents.y2 = dest_y + height; + + region->extents.x1 = MAX (region->extents.x1, 0); + region->extents.y1 = MAX (region->extents.y1, 0); + region->extents.x2 = MIN (region->extents.x2, dst_image->bits.width); + region->extents.y2 = MIN (region->extents.y2, dst_image->bits.height); + + region->data = 0; + + /* Check for empty operation */ + if (region->extents.x1 >= region->extents.x2 || + region->extents.y1 >= region->extents.y2) + { + region->extents.x1 = 0; + region->extents.x2 = 0; + region->extents.y1 = 0; + region->extents.y2 = 0; + return FALSE; + } + + if (dst_image->common.have_clip_region) + { + if (!clip_general_image (region, &dst_image->common.clip_region, 0, 0)) + return FALSE; + } + + if (dst_image->common.alpha_map) + { + if (!pixman_region32_intersect_rect (region, region, + dst_image->common.alpha_origin_x, + dst_image->common.alpha_origin_y, + dst_image->common.alpha_map->width, + dst_image->common.alpha_map->height)) + { + return FALSE; + } + if (!pixman_region32_not_empty (region)) + return FALSE; + if (dst_image->common.alpha_map->common.have_clip_region) + { + if (!clip_general_image (region, &dst_image->common.alpha_map->common.clip_region, + -dst_image->common.alpha_origin_x, + -dst_image->common.alpha_origin_y)) + { + return FALSE; + } + } + } + + /* clip against src */ + if (src_image->common.have_clip_region) + { + if (!clip_source_image (region, src_image, dest_x - src_x, dest_y - src_y)) + return FALSE; + } + if (src_image->common.alpha_map && src_image->common.alpha_map->common.have_clip_region) + { + if (!clip_source_image (region, (pixman_image_t *)src_image->common.alpha_map, + dest_x - (src_x - src_image->common.alpha_origin_x), + dest_y - (src_y - src_image->common.alpha_origin_y))) + { + return FALSE; + } + } + /* clip against mask */ + if (mask_image && mask_image->common.have_clip_region) + { + if (!clip_source_image (region, mask_image, dest_x - mask_x, dest_y - mask_y)) + return FALSE; + + if (mask_image->common.alpha_map && mask_image->common.alpha_map->common.have_clip_region) + { + if (!clip_source_image (region, (pixman_image_t *)mask_image->common.alpha_map, + dest_x - (mask_x - mask_image->common.alpha_origin_x), + dest_y - (mask_y - mask_image->common.alpha_origin_y))) + { + return FALSE; + } + } + } + + return TRUE; +} + +#define N_CACHED_FAST_PATHS 8 + +typedef struct +{ + struct + { + pixman_implementation_t * imp; + pixman_fast_path_t fast_path; + } cache [N_CACHED_FAST_PATHS]; +} cache_t; + +PIXMAN_DEFINE_THREAD_LOCAL (cache_t, fast_path_cache); + +static force_inline pixman_bool_t +lookup_composite_function (pixman_op_t op, + pixman_format_code_t src_format, + uint32_t src_flags, + pixman_format_code_t mask_format, + uint32_t mask_flags, + pixman_format_code_t dest_format, + uint32_t dest_flags, + pixman_implementation_t **out_imp, + pixman_composite_func_t *out_func) +{ + pixman_implementation_t *imp; + cache_t *cache; + int i; + + /* Check cache for fast paths */ + cache = PIXMAN_GET_THREAD_LOCAL (fast_path_cache); + + for (i = 0; i < N_CACHED_FAST_PATHS; ++i) + { + const pixman_fast_path_t *info = &(cache->cache[i].fast_path); + + /* Note that we check for equality here, not whether + * the cached fast path matches. This is to prevent + * us from selecting an overly general fast path + * when a more specific one would work. + */ + if (info->op == op && + info->src_format == src_format && + info->mask_format == mask_format && + info->dest_format == dest_format && + info->src_flags == src_flags && + info->mask_flags == mask_flags && + info->dest_flags == dest_flags && + info->func) + { + *out_imp = cache->cache[i].imp; + *out_func = cache->cache[i].fast_path.func; + + goto update_cache; + } + } + + for (imp = get_implementation (); imp != NULL; imp = imp->delegate) + { + const pixman_fast_path_t *info = imp->fast_paths; + + while (info->op != PIXMAN_OP_NONE) + { + if ((info->op == op || info->op == PIXMAN_OP_any) && + /* Formats */ + ((info->src_format == src_format) || + (info->src_format == PIXMAN_any)) && + ((info->mask_format == mask_format) || + (info->mask_format == PIXMAN_any)) && + ((info->dest_format == dest_format) || + (info->dest_format == PIXMAN_any)) && + /* Flags */ + (info->src_flags & src_flags) == info->src_flags && + (info->mask_flags & mask_flags) == info->mask_flags && + (info->dest_flags & dest_flags) == info->dest_flags) + { + *out_imp = imp; + *out_func = info->func; + + /* Set i to the last spot in the cache so that the + * move-to-front code below will work + */ + i = N_CACHED_FAST_PATHS - 1; + + goto update_cache; + } + + ++info; + } + } + return FALSE; + +update_cache: + if (i) + { + while (i--) + cache->cache[i + 1] = cache->cache[i]; + + cache->cache[0].imp = *out_imp; + cache->cache[0].fast_path.op = op; + cache->cache[0].fast_path.src_format = src_format; + cache->cache[0].fast_path.src_flags = src_flags; + cache->cache[0].fast_path.mask_format = mask_format; + cache->cache[0].fast_path.mask_flags = mask_flags; + cache->cache[0].fast_path.dest_format = dest_format; + cache->cache[0].fast_path.dest_flags = dest_flags; + cache->cache[0].fast_path.func = *out_func; + } + + return TRUE; +} + +static pixman_bool_t +compute_sample_extents (pixman_transform_t *transform, + pixman_box32_t *extents, int x, int y, + pixman_fixed_t x_off, pixman_fixed_t y_off, + pixman_fixed_t width, pixman_fixed_t height) +{ + pixman_fixed_t x1, y1, x2, y2; + pixman_fixed_48_16_t tx1, ty1, tx2, ty2; + + /* We have checked earlier that (extents->x1 - x) etc. fit in a pixman_fixed_t */ + x1 = (pixman_fixed_48_16_t)pixman_int_to_fixed (extents->x1 - x) + pixman_fixed_1 / 2; + y1 = (pixman_fixed_48_16_t)pixman_int_to_fixed (extents->y1 - y) + pixman_fixed_1 / 2; + x2 = (pixman_fixed_48_16_t)pixman_int_to_fixed (extents->x2 - x) - pixman_fixed_1 / 2; + y2 = (pixman_fixed_48_16_t)pixman_int_to_fixed (extents->y2 - y) - pixman_fixed_1 / 2; + + if (!transform) + { + tx1 = (pixman_fixed_48_16_t)x1; + ty1 = (pixman_fixed_48_16_t)y1; + tx2 = (pixman_fixed_48_16_t)x2; + ty2 = (pixman_fixed_48_16_t)y2; + } + else + { + int i; + + /* Silence GCC */ + tx1 = ty1 = tx2 = ty2 = 0; + + for (i = 0; i < 4; ++i) + { + pixman_fixed_48_16_t tx, ty; + pixman_vector_t v; + + v.vector[0] = (i & 0x01)? x1 : x2; + v.vector[1] = (i & 0x02)? y1 : y2; + v.vector[2] = pixman_fixed_1; + + if (!pixman_transform_point (transform, &v)) + return FALSE; + + tx = (pixman_fixed_48_16_t)v.vector[0]; + ty = (pixman_fixed_48_16_t)v.vector[1]; + + if (i == 0) + { + tx1 = tx; + ty1 = ty; + tx2 = tx; + ty2 = ty; + } + else + { + if (tx < tx1) + tx1 = tx; + if (ty < ty1) + ty1 = ty; + if (tx > tx2) + tx2 = tx; + if (ty > ty2) + ty2 = ty; + } + } + } + + /* Expand the source area by a tiny bit so account of different rounding that + * may happen during sampling. Note that (8 * pixman_fixed_e) is very far from + * 0.5 so this won't cause the area computed to be overly pessimistic. + */ + tx1 += x_off - 8 * pixman_fixed_e; + ty1 += y_off - 8 * pixman_fixed_e; + tx2 += x_off + width + 8 * pixman_fixed_e; + ty2 += y_off + height + 8 * pixman_fixed_e; + + if (tx1 < pixman_min_fixed_48_16 || tx1 > pixman_max_fixed_48_16 || + ty1 < pixman_min_fixed_48_16 || ty1 > pixman_max_fixed_48_16 || + tx2 < pixman_min_fixed_48_16 || tx2 > pixman_max_fixed_48_16 || + ty2 < pixman_min_fixed_48_16 || ty2 > pixman_max_fixed_48_16) + { + return FALSE; + } + else + { + extents->x1 = pixman_fixed_to_int (tx1); + extents->y1 = pixman_fixed_to_int (ty1); + extents->x2 = pixman_fixed_to_int (tx2) + 1; + extents->y2 = pixman_fixed_to_int (ty2) + 1; + + return TRUE; + } +} + +#define IS_16BIT(x) (((x) >= INT16_MIN) && ((x) <= INT16_MAX)) + +static pixman_bool_t +analyze_extent (pixman_image_t *image, int x, int y, + const pixman_box32_t *extents, uint32_t *flags) +{ + pixman_transform_t *transform; + pixman_fixed_t *params; + pixman_fixed_t x_off, y_off; + pixman_fixed_t width, height; + pixman_box32_t ex; + + if (!image) + return TRUE; + + /* Some compositing functions walk one step + * outside the destination rectangle, so we + * check here that the expanded-by-one source + * extents in destination space fits in 16 bits + */ + if (!IS_16BIT (extents->x1 - x - 1) || + !IS_16BIT (extents->y1 - y - 1) || + !IS_16BIT (extents->x2 - x + 1) || + !IS_16BIT (extents->y2 - y + 1)) + { + return FALSE; + } + + transform = image->common.transform; + if (image->common.type == BITS) + { + /* During repeat mode calculations we might convert the + * width/height of an image to fixed 16.16, so we need + * them to be smaller than 16 bits. + */ + if (image->bits.width >= 0x7fff || image->bits.height >= 0x7fff) + return FALSE; + +#define ID_AND_NEAREST (FAST_PATH_ID_TRANSFORM | FAST_PATH_NEAREST_FILTER) + + if ((image->common.flags & ID_AND_NEAREST) == ID_AND_NEAREST && + extents->x1 - x >= 0 && + extents->y1 - y >= 0 && + extents->x2 - x <= image->bits.width && + extents->y2 - y <= image->bits.height) + { + *flags |= FAST_PATH_SAMPLES_COVER_CLIP; + return TRUE; + } + + switch (image->common.filter) + { + case PIXMAN_FILTER_CONVOLUTION: + params = image->common.filter_params; + x_off = - pixman_fixed_e - ((params[0] - pixman_fixed_1) >> 1); + y_off = - pixman_fixed_e - ((params[1] - pixman_fixed_1) >> 1); + width = params[0]; + height = params[1]; + break; + + case PIXMAN_FILTER_GOOD: + case PIXMAN_FILTER_BEST: + case PIXMAN_FILTER_BILINEAR: + x_off = - pixman_fixed_1 / 2; + y_off = - pixman_fixed_1 / 2; + width = pixman_fixed_1; + height = pixman_fixed_1; + break; + + case PIXMAN_FILTER_FAST: + case PIXMAN_FILTER_NEAREST: + x_off = - pixman_fixed_e; + y_off = - pixman_fixed_e; + width = 0; + height = 0; + break; + + default: + return FALSE; + } + + /* Check whether the non-expanded, transformed extent is entirely within + * the source image, and set the FAST_PATH_SAMPLES_COVER_CLIP if it is. + */ + ex = *extents; + if (compute_sample_extents (transform, &ex, x, y, x_off, y_off, width, height) && + ex.x1 >= 0 && ex.y1 >= 0 && + ex.x2 <= image->bits.width && ex.y2 <= image->bits.height) + { + *flags |= FAST_PATH_SAMPLES_COVER_CLIP; + } + } + else + { + x_off = 0; + y_off = 0; + width = 0; + height = 0; + } + + /* Check that the extents expanded by one don't overflow. This ensures that + * compositing functions can simply walk the source space using 16.16 + * variables without worrying about overflow. + */ + ex.x1 = extents->x1 - 1; + ex.y1 = extents->y1 - 1; + ex.x2 = extents->x2 + 1; + ex.y2 = extents->y2 + 1; + + if (!compute_sample_extents (transform, &ex, x, y, x_off, y_off, width, height)) + return FALSE; + + return TRUE; +} + +/* + * Work around GCC bug causing crashes in Mozilla with SSE2 + * + * When using -msse, gcc generates movdqa instructions assuming that + * the stack is 16 byte aligned. Unfortunately some applications, such + * as Mozilla and Mono, end up aligning the stack to 4 bytes, which + * causes the movdqa instructions to fail. + * + * The __force_align_arg_pointer__ makes gcc generate a prologue that + * realigns the stack pointer to 16 bytes. + * + * On x86-64 this is not necessary because the standard ABI already + * calls for a 16 byte aligned stack. + * + * See https://bugs.freedesktop.org/show_bug.cgi?id=15693 + */ +#if defined (USE_SSE2) && defined(__GNUC__) && !defined(__x86_64__) && !defined(__amd64__) +__attribute__((__force_align_arg_pointer__)) +#endif +PIXMAN_EXPORT void +pixman_image_composite32 (pixman_op_t op, + pixman_image_t * src, + pixman_image_t * mask, + pixman_image_t * dest, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height) +{ + pixman_format_code_t src_format, mask_format, dest_format; + uint32_t src_flags, mask_flags, dest_flags; + pixman_region32_t region; + pixman_box32_t *extents; + uint32_t *src_bits; + int src_dx, src_dy; + uint32_t *mask_bits; + int mask_dx, mask_dy; + uint32_t *dest_bits; + int dest_dx, dest_dy; + pixman_bool_t need_workaround; + pixman_implementation_t *imp; + pixman_composite_func_t func; + + _pixman_image_validate (src); + if (mask) + _pixman_image_validate (mask); + _pixman_image_validate (dest); + + src_format = src->common.extended_format_code; + src_flags = src->common.flags; + + if (mask) + { + mask_format = mask->common.extended_format_code; + mask_flags = mask->common.flags; + } + else + { + mask_format = PIXMAN_null; + mask_flags = FAST_PATH_IS_OPAQUE; + } + + dest_format = dest->common.extended_format_code; + dest_flags = dest->common.flags; + + /* Check for pixbufs */ + if ((mask_format == PIXMAN_a8r8g8b8 || mask_format == PIXMAN_a8b8g8r8) && + (src->type == BITS && src->bits.bits == mask->bits.bits) && + (src->common.repeat == mask->common.repeat) && + (src_x == mask_x && src_y == mask_y)) + { + if (src_format == PIXMAN_x8b8g8r8) + src_format = mask_format = PIXMAN_pixbuf; + else if (src_format == PIXMAN_x8r8g8b8) + src_format = mask_format = PIXMAN_rpixbuf; + } + + /* Check for workaround */ + need_workaround = (src_flags | mask_flags | dest_flags) & FAST_PATH_NEEDS_WORKAROUND; + + if (need_workaround) + { + apply_workaround (src, &src_x, &src_y, &src_bits, &src_dx, &src_dy); + apply_workaround (mask, &mask_x, &mask_y, &mask_bits, &mask_dx, &mask_dy); + apply_workaround (dest, &dest_x, &dest_y, &dest_bits, &dest_dx, &dest_dy); + } + + pixman_region32_init (®ion); + + if (!pixman_compute_composite_region32 ( + ®ion, src, mask, dest, + src_x, src_y, mask_x, mask_y, dest_x, dest_y, width, height)) + { + goto out; + } + + extents = pixman_region32_extents (®ion); + + if (!analyze_extent (src, dest_x - src_x, dest_y - src_y, extents, &src_flags)) + goto out; + + if (!analyze_extent (mask, dest_x - mask_x, dest_y - mask_y, extents, &mask_flags)) + goto out; + + /* If the clip is within the source samples, and the samples are opaque, + * then the source is effectively opaque. + */ +#define BOTH (FAST_PATH_SAMPLES_OPAQUE | FAST_PATH_SAMPLES_COVER_CLIP) + + if ((src_flags & BOTH) == BOTH) + src_flags |= FAST_PATH_IS_OPAQUE; + + if ((mask_flags & BOTH) == BOTH) + mask_flags |= FAST_PATH_IS_OPAQUE; + + /* + * Check if we can replace our operator by a simpler one + * if the src or dest are opaque. The output operator should be + * mathematically equivalent to the source. + */ + op = optimize_operator (op, src_flags, mask_flags, dest_flags); + if (op == PIXMAN_OP_DST) + goto out; + + if (lookup_composite_function (op, + src_format, src_flags, + mask_format, mask_flags, + dest_format, dest_flags, + &imp, &func)) + { + const pixman_box32_t *pbox; + int n; + + pbox = pixman_region32_rectangles (®ion, &n); + + while (n--) + { + func (imp, op, + src, mask, dest, + pbox->x1 + src_x - dest_x, + pbox->y1 + src_y - dest_y, + pbox->x1 + mask_x - dest_x, + pbox->y1 + mask_y - dest_y, + pbox->x1, + pbox->y1, + pbox->x2 - pbox->x1, + pbox->y2 - pbox->y1); + + pbox++; + } + } + +out: + if (need_workaround) + { + unapply_workaround (src, src_bits, src_dx, src_dy); + unapply_workaround (mask, mask_bits, mask_dx, mask_dy); + unapply_workaround (dest, dest_bits, dest_dx, dest_dy); + } + + pixman_region32_fini (®ion); +} + +PIXMAN_EXPORT void +pixman_image_composite (pixman_op_t op, + pixman_image_t * src, + pixman_image_t * mask, + pixman_image_t * dest, + int16_t src_x, + int16_t src_y, + int16_t mask_x, + int16_t mask_y, + int16_t dest_x, + int16_t dest_y, + uint16_t width, + uint16_t height) +{ + pixman_image_composite32 (op, src, mask, dest, src_x, src_y, + mask_x, mask_y, dest_x, dest_y, width, height); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_blt (uint32_t *src_bits, + uint32_t *dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height) +{ + return _pixman_implementation_blt (get_implementation(), + src_bits, dst_bits, src_stride, dst_stride, + src_bpp, dst_bpp, + src_x, src_y, + dst_x, dst_y, + width, height); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_fill (uint32_t *bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t xor) +{ + return _pixman_implementation_fill ( + get_implementation(), bits, stride, bpp, x, y, width, height, xor); +} + +static uint32_t +color_to_uint32 (const pixman_color_t *color) +{ + return + (color->alpha >> 8 << 24) | + (color->red >> 8 << 16) | + (color->green & 0xff00) | + (color->blue >> 8); +} + +static pixman_bool_t +color_to_pixel (pixman_color_t * color, + uint32_t * pixel, + pixman_format_code_t format) +{ + uint32_t c = color_to_uint32 (color); + + if (!(format == PIXMAN_a8r8g8b8 || + format == PIXMAN_x8r8g8b8 || + format == PIXMAN_a8b8g8r8 || + format == PIXMAN_x8b8g8r8 || + format == PIXMAN_b8g8r8a8 || + format == PIXMAN_b8g8r8x8 || + format == PIXMAN_r5g6b5 || + format == PIXMAN_b5g6r5 || + format == PIXMAN_a8)) + { + return FALSE; + } + + if (PIXMAN_FORMAT_TYPE (format) == PIXMAN_TYPE_ABGR) + { + c = ((c & 0xff000000) >> 0) | + ((c & 0x00ff0000) >> 16) | + ((c & 0x0000ff00) >> 0) | + ((c & 0x000000ff) << 16); + } + if (PIXMAN_FORMAT_TYPE (format) == PIXMAN_TYPE_BGRA) + { + c = ((c & 0xff000000) >> 24) | + ((c & 0x00ff0000) >> 8) | + ((c & 0x0000ff00) << 8) | + ((c & 0x000000ff) << 24); + } + + if (format == PIXMAN_a8) + c = c >> 24; + else if (format == PIXMAN_r5g6b5 || + format == PIXMAN_b5g6r5) + c = CONVERT_8888_TO_0565 (c); + +#if 0 + printf ("color: %x %x %x %x\n", color->alpha, color->red, color->green, color->blue); + printf ("pixel: %x\n", c); +#endif + + *pixel = c; + return TRUE; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_image_fill_rectangles (pixman_op_t op, + pixman_image_t * dest, + pixman_color_t * color, + int n_rects, + const pixman_rectangle16_t *rects) +{ + pixman_box32_t stack_boxes[6]; + pixman_box32_t *boxes; + pixman_bool_t result; + int i; + + if (n_rects > 6) + { + boxes = pixman_malloc_ab (sizeof (pixman_box32_t), n_rects); + if (boxes == NULL) + return FALSE; + } + else + { + boxes = stack_boxes; + } + + for (i = 0; i < n_rects; ++i) + { + boxes[i].x1 = rects[i].x; + boxes[i].y1 = rects[i].y; + boxes[i].x2 = boxes[i].x1 + rects[i].width; + boxes[i].y2 = boxes[i].y1 + rects[i].height; + } + + result = pixman_image_fill_boxes (op, dest, color, n_rects, boxes); + + if (boxes != stack_boxes) + free (boxes); + + return result; +} + +PIXMAN_EXPORT pixman_bool_t +pixman_image_fill_boxes (pixman_op_t op, + pixman_image_t * dest, + pixman_color_t * color, + int n_boxes, + const pixman_box32_t *boxes) +{ + pixman_image_t *solid; + pixman_color_t c; + int i; + + _pixman_image_validate (dest); + + if (color->alpha == 0xffff) + { + if (op == PIXMAN_OP_OVER) + op = PIXMAN_OP_SRC; + } + + if (op == PIXMAN_OP_CLEAR) + { + c.red = 0; + c.green = 0; + c.blue = 0; + c.alpha = 0; + + color = &c; + + op = PIXMAN_OP_SRC; + } + + if (op == PIXMAN_OP_SRC) + { + uint32_t pixel; + + if (color_to_pixel (color, &pixel, dest->bits.format)) + { + pixman_region32_t fill_region; + int n_rects, j; + pixman_box32_t *rects; + + if (!pixman_region32_init_rects (&fill_region, boxes, n_boxes)) + return FALSE; + + if (dest->common.have_clip_region) + { + if (!pixman_region32_intersect (&fill_region, + &fill_region, + &dest->common.clip_region)) + return FALSE; + } + + rects = pixman_region32_rectangles (&fill_region, &n_rects); + for (j = 0; j < n_rects; ++j) + { + const pixman_box32_t *rect = &(rects[j]); + pixman_fill (dest->bits.bits, dest->bits.rowstride, PIXMAN_FORMAT_BPP (dest->bits.format), + rect->x1, rect->y1, rect->x2 - rect->x1, rect->y2 - rect->y1, + pixel); + } + + pixman_region32_fini (&fill_region); + return TRUE; + } + } + + solid = pixman_image_create_solid_fill (color); + if (!solid) + return FALSE; + + for (i = 0; i < n_boxes; ++i) + { + const pixman_box32_t *box = &(boxes[i]); + + pixman_image_composite32 (op, solid, NULL, dest, + 0, 0, 0, 0, + box->x1, box->y1, + box->x2 - box->x1, box->y2 - box->y1); + } + + pixman_image_unref (solid); + + return TRUE; +} + +/** + * pixman_version: + * + * Returns the version of the pixman library encoded in a single + * integer as per %PIXMAN_VERSION_ENCODE. The encoding ensures that + * later versions compare greater than earlier versions. + * + * A run-time comparison to check that pixman's version is greater than + * or equal to version X.Y.Z could be performed as follows: + * + * + * if (pixman_version() >= PIXMAN_VERSION_ENCODE(X,Y,Z)) {...} + * + * + * See also pixman_version_string() as well as the compile-time + * equivalents %PIXMAN_VERSION and %PIXMAN_VERSION_STRING. + * + * Return value: the encoded version. + **/ +PIXMAN_EXPORT int +pixman_version (void) +{ + return PIXMAN_VERSION; +} + +/** + * pixman_version_string: + * + * Returns the version of the pixman library as a human-readable string + * of the form "X.Y.Z". + * + * See also pixman_version() as well as the compile-time equivalents + * %PIXMAN_VERSION_STRING and %PIXMAN_VERSION. + * + * Return value: a string containing the version. + **/ +PIXMAN_EXPORT const char* +pixman_version_string (void) +{ + return PIXMAN_VERSION_STRING; +} + +/** + * pixman_format_supported_source: + * @format: A pixman_format_code_t format + * + * Return value: whether the provided format code is a supported + * format for a pixman surface used as a source in + * rendering. + * + * Currently, all pixman_format_code_t values are supported. + **/ +PIXMAN_EXPORT pixman_bool_t +pixman_format_supported_source (pixman_format_code_t format) +{ + switch (format) + { + /* 32 bpp formats */ + case PIXMAN_a2b10g10r10: + case PIXMAN_x2b10g10r10: + case PIXMAN_a2r10g10b10: + case PIXMAN_x2r10g10b10: + case PIXMAN_a8r8g8b8: + case PIXMAN_x8r8g8b8: + case PIXMAN_a8b8g8r8: + case PIXMAN_x8b8g8r8: + case PIXMAN_b8g8r8a8: + case PIXMAN_b8g8r8x8: + case PIXMAN_r8g8b8: + case PIXMAN_b8g8r8: + case PIXMAN_r5g6b5: + case PIXMAN_b5g6r5: + case PIXMAN_x14r6g6b6: + /* 16 bpp formats */ + case PIXMAN_a1r5g5b5: + case PIXMAN_x1r5g5b5: + case PIXMAN_a1b5g5r5: + case PIXMAN_x1b5g5r5: + case PIXMAN_a4r4g4b4: + case PIXMAN_x4r4g4b4: + case PIXMAN_a4b4g4r4: + case PIXMAN_x4b4g4r4: + /* 8bpp formats */ + case PIXMAN_a8: + case PIXMAN_r3g3b2: + case PIXMAN_b2g3r3: + case PIXMAN_a2r2g2b2: + case PIXMAN_a2b2g2r2: + case PIXMAN_c8: + case PIXMAN_g8: + case PIXMAN_x4a4: + /* Collides with PIXMAN_c8 + case PIXMAN_x4c4: + */ + /* Collides with PIXMAN_g8 + case PIXMAN_x4g4: + */ + /* 4bpp formats */ + case PIXMAN_a4: + case PIXMAN_r1g2b1: + case PIXMAN_b1g2r1: + case PIXMAN_a1r1g1b1: + case PIXMAN_a1b1g1r1: + case PIXMAN_c4: + case PIXMAN_g4: + /* 1bpp formats */ + case PIXMAN_a1: + case PIXMAN_g1: + /* YUV formats */ + case PIXMAN_yuy2: + case PIXMAN_yv12: + return TRUE; + + default: + return FALSE; + } +} + +/** + * pixman_format_supported_destination: + * @format: A pixman_format_code_t format + * + * Return value: whether the provided format code is a supported + * format for a pixman surface used as a destination in + * rendering. + * + * Currently, all pixman_format_code_t values are supported + * except for the YUV formats. + **/ +PIXMAN_EXPORT pixman_bool_t +pixman_format_supported_destination (pixman_format_code_t format) +{ + /* YUV formats cannot be written to at the moment */ + if (format == PIXMAN_yuy2 || format == PIXMAN_yv12) + return FALSE; + + return pixman_format_supported_source (format); +} + +PIXMAN_EXPORT pixman_bool_t +pixman_compute_composite_region (pixman_region16_t * region, + pixman_image_t * src_image, + pixman_image_t * mask_image, + pixman_image_t * dst_image, + int16_t src_x, + int16_t src_y, + int16_t mask_x, + int16_t mask_y, + int16_t dest_x, + int16_t dest_y, + uint16_t width, + uint16_t height) +{ + pixman_region32_t r32; + pixman_bool_t retval; + + pixman_region32_init (&r32); + + retval = pixman_compute_composite_region32 ( + &r32, src_image, mask_image, dst_image, + src_x, src_y, mask_x, mask_y, dest_x, dest_y, + width, height); + + if (retval) + { + if (!pixman_region16_copy_from_region32 (region, &r32)) + retval = FALSE; + } + + pixman_region32_fini (&r32); + return retval; +} diff --git a/programs/develop/libraries/pixman/pixman.h b/programs/develop/libraries/pixman/pixman.h new file mode 100644 index 0000000000..cfffa79add --- /dev/null +++ b/programs/develop/libraries/pixman/pixman.h @@ -0,0 +1,950 @@ +/*********************************************************** + +Copyright 1987, 1998 The Open Group + +Permission to use, copy, modify, distribute, and sell this software and its +documentation for any purpose is hereby granted without fee, provided that +the above copyright notice appear in all copies and that both that +copyright notice and this permission notice appear in supporting +documentation. + +The above copyright notice and this permission notice 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 +OPEN GROUP 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. + +Except as contained in this notice, the name of The Open Group shall not be +used in advertising or otherwise to promote the sale, use or other dealings +in this Software without prior written authorization from The Open Group. + +Copyright 1987 by Digital Equipment Corporation, Maynard, Massachusetts. + + All Rights Reserved + +Permission to use, copy, modify, and distribute this software and its +documentation for any purpose and without fee is hereby granted, +provided that the above copyright notice appear in all copies and that +both that copyright notice and this permission notice appear in +supporting documentation, and that the name of Digital not be +used in advertising or publicity pertaining to distribution of the +software without specific, written prior permission. + +DIGITAL DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING +ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL +DIGITAL BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR +ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, +WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, +ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS +SOFTWARE. + +******************************************************************/ +/* + * Copyright © 1998, 2004 Keith Packard + * Copyright 2007 Red Hat, Inc. + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of Keith Packard not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. Keith Packard makes no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * KEITH PACKARD DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO + * EVENT SHALL KEITH PACKARD BE LIABLE FOR ANY SPECIAL, INDIRECT OR + * CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, + * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER + * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR + * PERFORMANCE OF THIS SOFTWARE. + */ + +#ifndef PIXMAN_H__ +#define PIXMAN_H__ + +#include + +#ifdef __cplusplus +#define PIXMAN_BEGIN_DECLS extern "C" { +#define PIXMAN_END_DECLS } +#else +#define PIXMAN_BEGIN_DECLS +#define PIXMAN_END_DECLS +#endif + +PIXMAN_BEGIN_DECLS + +/* + * Standard integers + */ + +#if !defined (PIXMAN_DONT_DEFINE_STDINT) + +#if defined (_SVR4) || defined (SVR4) || defined (__OpenBSD__) || defined (_sgi) || defined (__sun) || defined (sun) || defined (__digital__) || defined (__HP_cc) +# include +/* VS 2010 (_MSC_VER 1600) has stdint.h */ +#elif defined (_MSC_VER) && _MSC_VER < 1600 +typedef __int8 int8_t; +typedef unsigned __int8 uint8_t; +typedef __int16 int16_t; +typedef unsigned __int16 uint16_t; +typedef __int32 int32_t; +typedef unsigned __int32 uint32_t; +typedef __int64 int64_t; +typedef unsigned __int64 uint64_t; +#elif defined (_AIX) +# include +#else +# include +#endif + +#endif + +/* + * Boolean + */ +typedef int pixman_bool_t; + +/* + * Fixpoint numbers + */ +typedef int64_t pixman_fixed_32_32_t; +typedef pixman_fixed_32_32_t pixman_fixed_48_16_t; +typedef uint32_t pixman_fixed_1_31_t; +typedef uint32_t pixman_fixed_1_16_t; +typedef int32_t pixman_fixed_16_16_t; +typedef pixman_fixed_16_16_t pixman_fixed_t; + +#define pixman_fixed_e ((pixman_fixed_t) 1) +#define pixman_fixed_1 (pixman_int_to_fixed(1)) +#define pixman_fixed_1_minus_e (pixman_fixed_1 - pixman_fixed_e) +#define pixman_fixed_minus_1 (pixman_int_to_fixed(-1)) +#define pixman_fixed_to_int(f) ((int) ((f) >> 16)) +#define pixman_int_to_fixed(i) ((pixman_fixed_t) ((i) << 16)) +#define pixman_fixed_to_double(f) (double) ((f) / (double) pixman_fixed_1) +#define pixman_double_to_fixed(d) ((pixman_fixed_t) ((d) * 65536.0)) +#define pixman_fixed_frac(f) ((f) & pixman_fixed_1_minus_e) +#define pixman_fixed_floor(f) ((f) & ~pixman_fixed_1_minus_e) +#define pixman_fixed_ceil(f) pixman_fixed_floor ((f) + pixman_fixed_1_minus_e) +#define pixman_fixed_fraction(f) ((f) & pixman_fixed_1_minus_e) +#define pixman_fixed_mod_2(f) ((f) & (pixman_fixed1 | pixman_fixed_1_minus_e)) +#define pixman_max_fixed_48_16 ((pixman_fixed_48_16_t) 0x7fffffff) +#define pixman_min_fixed_48_16 (-((pixman_fixed_48_16_t) 1 << 31)) + +/* + * Misc structs + */ +typedef struct pixman_color pixman_color_t; +typedef struct pixman_point_fixed pixman_point_fixed_t; +typedef struct pixman_line_fixed pixman_line_fixed_t; +typedef struct pixman_vector pixman_vector_t; +typedef struct pixman_transform pixman_transform_t; + +struct pixman_color +{ + uint16_t red; + uint16_t green; + uint16_t blue; + uint16_t alpha; +}; + +struct pixman_point_fixed +{ + pixman_fixed_t x; + pixman_fixed_t y; +}; + +struct pixman_line_fixed +{ + pixman_point_fixed_t p1, p2; +}; + +/* + * Fixed point matrices + */ + +struct pixman_vector +{ + pixman_fixed_t vector[3]; +}; + +struct pixman_transform +{ + pixman_fixed_t matrix[3][3]; +}; + +/* forward declaration (sorry) */ +struct pixman_box16; +typedef union pixman_image pixman_image_t; + +void pixman_transform_init_identity (struct pixman_transform *matrix); +pixman_bool_t pixman_transform_point_3d (const struct pixman_transform *transform, + struct pixman_vector *vector); +pixman_bool_t pixman_transform_point (const struct pixman_transform *transform, + struct pixman_vector *vector); +pixman_bool_t pixman_transform_multiply (struct pixman_transform *dst, + const struct pixman_transform *l, + const struct pixman_transform *r); +void pixman_transform_init_scale (struct pixman_transform *t, + pixman_fixed_t sx, + pixman_fixed_t sy); +pixman_bool_t pixman_transform_scale (struct pixman_transform *forward, + struct pixman_transform *reverse, + pixman_fixed_t sx, + pixman_fixed_t sy); +void pixman_transform_init_rotate (struct pixman_transform *t, + pixman_fixed_t cos, + pixman_fixed_t sin); +pixman_bool_t pixman_transform_rotate (struct pixman_transform *forward, + struct pixman_transform *reverse, + pixman_fixed_t c, + pixman_fixed_t s); +void pixman_transform_init_translate (struct pixman_transform *t, + pixman_fixed_t tx, + pixman_fixed_t ty); +pixman_bool_t pixman_transform_translate (struct pixman_transform *forward, + struct pixman_transform *reverse, + pixman_fixed_t tx, + pixman_fixed_t ty); +pixman_bool_t pixman_transform_bounds (const struct pixman_transform *matrix, + struct pixman_box16 *b); +pixman_bool_t pixman_transform_invert (struct pixman_transform *dst, + const struct pixman_transform *src); +pixman_bool_t pixman_transform_is_identity (const struct pixman_transform *t); +pixman_bool_t pixman_transform_is_scale (const struct pixman_transform *t); +pixman_bool_t pixman_transform_is_int_translate (const struct pixman_transform *t); +pixman_bool_t pixman_transform_is_inverse (const struct pixman_transform *a, + const struct pixman_transform *b); + +/* + * Floating point matrices + */ +struct pixman_f_vector +{ + double v[3]; +}; + +struct pixman_f_transform +{ + double m[3][3]; +}; + +pixman_bool_t pixman_transform_from_pixman_f_transform (struct pixman_transform *t, + const struct pixman_f_transform *ft); +void pixman_f_transform_from_pixman_transform (struct pixman_f_transform *ft, + const struct pixman_transform *t); +pixman_bool_t pixman_f_transform_invert (struct pixman_f_transform *dst, + const struct pixman_f_transform *src); +pixman_bool_t pixman_f_transform_point (const struct pixman_f_transform *t, + struct pixman_f_vector *v); +void pixman_f_transform_point_3d (const struct pixman_f_transform *t, + struct pixman_f_vector *v); +void pixman_f_transform_multiply (struct pixman_f_transform *dst, + const struct pixman_f_transform *l, + const struct pixman_f_transform *r); +void pixman_f_transform_init_scale (struct pixman_f_transform *t, + double sx, + double sy); +pixman_bool_t pixman_f_transform_scale (struct pixman_f_transform *forward, + struct pixman_f_transform *reverse, + double sx, + double sy); +void pixman_f_transform_init_rotate (struct pixman_f_transform *t, + double cos, + double sin); +pixman_bool_t pixman_f_transform_rotate (struct pixman_f_transform *forward, + struct pixman_f_transform *reverse, + double c, + double s); +void pixman_f_transform_init_translate (struct pixman_f_transform *t, + double tx, + double ty); +pixman_bool_t pixman_f_transform_translate (struct pixman_f_transform *forward, + struct pixman_f_transform *reverse, + double tx, + double ty); +pixman_bool_t pixman_f_transform_bounds (const struct pixman_f_transform *t, + struct pixman_box16 *b); +void pixman_f_transform_init_identity (struct pixman_f_transform *t); + +typedef enum +{ + PIXMAN_REPEAT_NONE, + PIXMAN_REPEAT_NORMAL, + PIXMAN_REPEAT_PAD, + PIXMAN_REPEAT_REFLECT +} pixman_repeat_t; + +typedef enum +{ + PIXMAN_FILTER_FAST, + PIXMAN_FILTER_GOOD, + PIXMAN_FILTER_BEST, + PIXMAN_FILTER_NEAREST, + PIXMAN_FILTER_BILINEAR, + PIXMAN_FILTER_CONVOLUTION +} pixman_filter_t; + +typedef enum +{ + PIXMAN_OP_CLEAR = 0x00, + PIXMAN_OP_SRC = 0x01, + PIXMAN_OP_DST = 0x02, + PIXMAN_OP_OVER = 0x03, + PIXMAN_OP_OVER_REVERSE = 0x04, + PIXMAN_OP_IN = 0x05, + PIXMAN_OP_IN_REVERSE = 0x06, + PIXMAN_OP_OUT = 0x07, + PIXMAN_OP_OUT_REVERSE = 0x08, + PIXMAN_OP_ATOP = 0x09, + PIXMAN_OP_ATOP_REVERSE = 0x0a, + PIXMAN_OP_XOR = 0x0b, + PIXMAN_OP_ADD = 0x0c, + PIXMAN_OP_SATURATE = 0x0d, + + PIXMAN_OP_DISJOINT_CLEAR = 0x10, + PIXMAN_OP_DISJOINT_SRC = 0x11, + PIXMAN_OP_DISJOINT_DST = 0x12, + PIXMAN_OP_DISJOINT_OVER = 0x13, + PIXMAN_OP_DISJOINT_OVER_REVERSE = 0x14, + PIXMAN_OP_DISJOINT_IN = 0x15, + PIXMAN_OP_DISJOINT_IN_REVERSE = 0x16, + PIXMAN_OP_DISJOINT_OUT = 0x17, + PIXMAN_OP_DISJOINT_OUT_REVERSE = 0x18, + PIXMAN_OP_DISJOINT_ATOP = 0x19, + PIXMAN_OP_DISJOINT_ATOP_REVERSE = 0x1a, + PIXMAN_OP_DISJOINT_XOR = 0x1b, + + PIXMAN_OP_CONJOINT_CLEAR = 0x20, + PIXMAN_OP_CONJOINT_SRC = 0x21, + PIXMAN_OP_CONJOINT_DST = 0x22, + PIXMAN_OP_CONJOINT_OVER = 0x23, + PIXMAN_OP_CONJOINT_OVER_REVERSE = 0x24, + PIXMAN_OP_CONJOINT_IN = 0x25, + PIXMAN_OP_CONJOINT_IN_REVERSE = 0x26, + PIXMAN_OP_CONJOINT_OUT = 0x27, + PIXMAN_OP_CONJOINT_OUT_REVERSE = 0x28, + PIXMAN_OP_CONJOINT_ATOP = 0x29, + PIXMAN_OP_CONJOINT_ATOP_REVERSE = 0x2a, + PIXMAN_OP_CONJOINT_XOR = 0x2b, + + PIXMAN_OP_MULTIPLY = 0x30, + PIXMAN_OP_SCREEN = 0x31, + PIXMAN_OP_OVERLAY = 0x32, + PIXMAN_OP_DARKEN = 0x33, + PIXMAN_OP_LIGHTEN = 0x34, + PIXMAN_OP_COLOR_DODGE = 0x35, + PIXMAN_OP_COLOR_BURN = 0x36, + PIXMAN_OP_HARD_LIGHT = 0x37, + PIXMAN_OP_SOFT_LIGHT = 0x38, + PIXMAN_OP_DIFFERENCE = 0x39, + PIXMAN_OP_EXCLUSION = 0x3a, + PIXMAN_OP_HSL_HUE = 0x3b, + PIXMAN_OP_HSL_SATURATION = 0x3c, + PIXMAN_OP_HSL_COLOR = 0x3d, + PIXMAN_OP_HSL_LUMINOSITY = 0x3e + +#ifdef PIXMAN_USE_INTERNAL_API + , + PIXMAN_N_OPERATORS, + PIXMAN_OP_NONE = PIXMAN_N_OPERATORS +#endif +} pixman_op_t; + +/* + * Regions + */ +typedef struct pixman_region16_data pixman_region16_data_t; +typedef struct pixman_box16 pixman_box16_t; +typedef struct pixman_rectangle16 pixman_rectangle16_t; +typedef struct pixman_region16 pixman_region16_t; + +struct pixman_region16_data { + long size; + long numRects; +/* pixman_box16_t rects[size]; in memory but not explicitly declared */ +}; + +struct pixman_rectangle16 +{ + int16_t x, y; + uint16_t width, height; +}; + +struct pixman_box16 +{ + int16_t x1, y1, x2, y2; +}; + +struct pixman_region16 +{ + pixman_box16_t extents; + pixman_region16_data_t *data; +}; + +typedef enum +{ + PIXMAN_REGION_OUT, + PIXMAN_REGION_IN, + PIXMAN_REGION_PART +} pixman_region_overlap_t; + +/* This function exists only to make it possible to preserve + * the X ABI - it should go away at first opportunity. + */ +void pixman_region_set_static_pointers (pixman_box16_t *empty_box, + pixman_region16_data_t *empty_data, + pixman_region16_data_t *broken_data); + +/* creation/destruction */ +void pixman_region_init (pixman_region16_t *region); +void pixman_region_init_rect (pixman_region16_t *region, + int x, + int y, + unsigned int width, + unsigned int height); +pixman_bool_t pixman_region_init_rects (pixman_region16_t *region, + const pixman_box16_t *boxes, + int count); +void pixman_region_init_with_extents (pixman_region16_t *region, + pixman_box16_t *extents); +void pixman_region_init_from_image (pixman_region16_t *region, + pixman_image_t *image); +void pixman_region_fini (pixman_region16_t *region); + + +/* manipulation */ +void pixman_region_translate (pixman_region16_t *region, + int x, + int y); +pixman_bool_t pixman_region_copy (pixman_region16_t *dest, + pixman_region16_t *source); +pixman_bool_t pixman_region_intersect (pixman_region16_t *new_reg, + pixman_region16_t *reg1, + pixman_region16_t *reg2); +pixman_bool_t pixman_region_union (pixman_region16_t *new_reg, + pixman_region16_t *reg1, + pixman_region16_t *reg2); +pixman_bool_t pixman_region_union_rect (pixman_region16_t *dest, + pixman_region16_t *source, + int x, + int y, + unsigned int width, + unsigned int height); +pixman_bool_t pixman_region_intersect_rect (pixman_region16_t *dest, + pixman_region16_t *source, + int x, + int y, + unsigned int width, + unsigned int height); +pixman_bool_t pixman_region_subtract (pixman_region16_t *reg_d, + pixman_region16_t *reg_m, + pixman_region16_t *reg_s); +pixman_bool_t pixman_region_inverse (pixman_region16_t *new_reg, + pixman_region16_t *reg1, + pixman_box16_t *inv_rect); +pixman_bool_t pixman_region_contains_point (pixman_region16_t *region, + int x, + int y, + pixman_box16_t *box); +pixman_region_overlap_t pixman_region_contains_rectangle (pixman_region16_t *region, + pixman_box16_t *prect); +pixman_bool_t pixman_region_not_empty (pixman_region16_t *region); +pixman_box16_t * pixman_region_extents (pixman_region16_t *region); +int pixman_region_n_rects (pixman_region16_t *region); +pixman_box16_t * pixman_region_rectangles (pixman_region16_t *region, + int *n_rects); +pixman_bool_t pixman_region_equal (pixman_region16_t *region1, + pixman_region16_t *region2); +pixman_bool_t pixman_region_selfcheck (pixman_region16_t *region); +void pixman_region_reset (pixman_region16_t *region, + pixman_box16_t *box); +/* + * 32 bit regions + */ +typedef struct pixman_region32_data pixman_region32_data_t; +typedef struct pixman_box32 pixman_box32_t; +typedef struct pixman_rectangle32 pixman_rectangle32_t; +typedef struct pixman_region32 pixman_region32_t; + +struct pixman_region32_data { + long size; + long numRects; +/* pixman_box32_t rects[size]; in memory but not explicitly declared */ +}; + +struct pixman_rectangle32 +{ + int32_t x, y; + uint32_t width, height; +}; + +struct pixman_box32 +{ + int32_t x1, y1, x2, y2; +}; + +struct pixman_region32 +{ + pixman_box32_t extents; + pixman_region32_data_t *data; +}; + +/* creation/destruction */ +void pixman_region32_init (pixman_region32_t *region); +void pixman_region32_init_rect (pixman_region32_t *region, + int x, + int y, + unsigned int width, + unsigned int height); +pixman_bool_t pixman_region32_init_rects (pixman_region32_t *region, + const pixman_box32_t *boxes, + int count); +void pixman_region32_init_with_extents (pixman_region32_t *region, + pixman_box32_t *extents); +void pixman_region32_init_from_image (pixman_region32_t *region, + pixman_image_t *image); +void pixman_region32_fini (pixman_region32_t *region); + + +/* manipulation */ +void pixman_region32_translate (pixman_region32_t *region, + int x, + int y); +pixman_bool_t pixman_region32_copy (pixman_region32_t *dest, + pixman_region32_t *source); +pixman_bool_t pixman_region32_intersect (pixman_region32_t *new_reg, + pixman_region32_t *reg1, + pixman_region32_t *reg2); +pixman_bool_t pixman_region32_union (pixman_region32_t *new_reg, + pixman_region32_t *reg1, + pixman_region32_t *reg2); +pixman_bool_t pixman_region32_intersect_rect (pixman_region32_t *dest, + pixman_region32_t *source, + int x, + int y, + unsigned int width, + unsigned int height); +pixman_bool_t pixman_region32_union_rect (pixman_region32_t *dest, + pixman_region32_t *source, + int x, + int y, + unsigned int width, + unsigned int height); +pixman_bool_t pixman_region32_subtract (pixman_region32_t *reg_d, + pixman_region32_t *reg_m, + pixman_region32_t *reg_s); +pixman_bool_t pixman_region32_inverse (pixman_region32_t *new_reg, + pixman_region32_t *reg1, + pixman_box32_t *inv_rect); +pixman_bool_t pixman_region32_contains_point (pixman_region32_t *region, + int x, + int y, + pixman_box32_t *box); +pixman_region_overlap_t pixman_region32_contains_rectangle (pixman_region32_t *region, + pixman_box32_t *prect); +pixman_bool_t pixman_region32_not_empty (pixman_region32_t *region); +pixman_box32_t * pixman_region32_extents (pixman_region32_t *region); +int pixman_region32_n_rects (pixman_region32_t *region); +pixman_box32_t * pixman_region32_rectangles (pixman_region32_t *region, + int *n_rects); +pixman_bool_t pixman_region32_equal (pixman_region32_t *region1, + pixman_region32_t *region2); +pixman_bool_t pixman_region32_selfcheck (pixman_region32_t *region); +void pixman_region32_reset (pixman_region32_t *region, + pixman_box32_t *box); + + +/* Copy / Fill / Misc */ +pixman_bool_t pixman_blt (uint32_t *src_bits, + uint32_t *dst_bits, + int src_stride, + int dst_stride, + int src_bpp, + int dst_bpp, + int src_x, + int src_y, + int dst_x, + int dst_y, + int width, + int height); +pixman_bool_t pixman_fill (uint32_t *bits, + int stride, + int bpp, + int x, + int y, + int width, + int height, + uint32_t _xor); + +int pixman_version (void); +const char* pixman_version_string (void); + +/* + * Images + */ +typedef struct pixman_indexed pixman_indexed_t; +typedef struct pixman_gradient_stop pixman_gradient_stop_t; + +typedef uint32_t (* pixman_read_memory_func_t) (const void *src, int size); +typedef void (* pixman_write_memory_func_t) (void *dst, uint32_t value, int size); + +typedef void (* pixman_image_destroy_func_t) (pixman_image_t *image, void *data); + +struct pixman_gradient_stop { + pixman_fixed_t x; + pixman_color_t color; +}; + +#define PIXMAN_MAX_INDEXED 256 /* XXX depth must be <= 8 */ + +#if PIXMAN_MAX_INDEXED <= 256 +typedef uint8_t pixman_index_type; +#endif + +struct pixman_indexed +{ + pixman_bool_t color; + uint32_t rgba[PIXMAN_MAX_INDEXED]; + pixman_index_type ent[32768]; +}; + +/* + * While the protocol is generous in format support, the + * sample implementation allows only packed RGB and GBR + * representations for data to simplify software rendering, + */ +#define PIXMAN_FORMAT(bpp,type,a,r,g,b) (((bpp) << 24) | \ + ((type) << 16) | \ + ((a) << 12) | \ + ((r) << 8) | \ + ((g) << 4) | \ + ((b))) + +#define PIXMAN_FORMAT_BPP(f) (((f) >> 24) ) +#define PIXMAN_FORMAT_TYPE(f) (((f) >> 16) & 0xff) +#define PIXMAN_FORMAT_A(f) (((f) >> 12) & 0x0f) +#define PIXMAN_FORMAT_R(f) (((f) >> 8) & 0x0f) +#define PIXMAN_FORMAT_G(f) (((f) >> 4) & 0x0f) +#define PIXMAN_FORMAT_B(f) (((f) ) & 0x0f) +#define PIXMAN_FORMAT_RGB(f) (((f) ) & 0xfff) +#define PIXMAN_FORMAT_VIS(f) (((f) ) & 0xffff) +#define PIXMAN_FORMAT_DEPTH(f) (PIXMAN_FORMAT_A(f) + \ + PIXMAN_FORMAT_R(f) + \ + PIXMAN_FORMAT_G(f) + \ + PIXMAN_FORMAT_B(f)) + +#define PIXMAN_TYPE_OTHER 0 +#define PIXMAN_TYPE_A 1 +#define PIXMAN_TYPE_ARGB 2 +#define PIXMAN_TYPE_ABGR 3 +#define PIXMAN_TYPE_COLOR 4 +#define PIXMAN_TYPE_GRAY 5 +#define PIXMAN_TYPE_YUY2 6 +#define PIXMAN_TYPE_YV12 7 +#define PIXMAN_TYPE_BGRA 8 + +#define PIXMAN_FORMAT_COLOR(f) \ + (PIXMAN_FORMAT_TYPE(f) == PIXMAN_TYPE_ARGB || \ + PIXMAN_FORMAT_TYPE(f) == PIXMAN_TYPE_ABGR || \ + PIXMAN_FORMAT_TYPE(f) == PIXMAN_TYPE_BGRA) + +/* 32bpp formats */ +typedef enum { + PIXMAN_a8r8g8b8 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ARGB,8,8,8,8), + PIXMAN_x8r8g8b8 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ARGB,0,8,8,8), + PIXMAN_a8b8g8r8 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,8,8,8,8), + PIXMAN_x8b8g8r8 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,0,8,8,8), + PIXMAN_b8g8r8a8 = PIXMAN_FORMAT(32,PIXMAN_TYPE_BGRA,8,8,8,8), + PIXMAN_b8g8r8x8 = PIXMAN_FORMAT(32,PIXMAN_TYPE_BGRA,0,8,8,8), + PIXMAN_x14r6g6b6 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ARGB,0,6,6,6), + PIXMAN_x2r10g10b10 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ARGB,0,10,10,10), + PIXMAN_a2r10g10b10 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ARGB,2,10,10,10), + PIXMAN_x2b10g10r10 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,0,10,10,10), + PIXMAN_a2b10g10r10 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,2,10,10,10), + +/* 24bpp formats */ + PIXMAN_r8g8b8 = PIXMAN_FORMAT(24,PIXMAN_TYPE_ARGB,0,8,8,8), + PIXMAN_b8g8r8 = PIXMAN_FORMAT(24,PIXMAN_TYPE_ABGR,0,8,8,8), + +/* 16bpp formats */ + PIXMAN_r5g6b5 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ARGB,0,5,6,5), + PIXMAN_b5g6r5 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ABGR,0,5,6,5), + + PIXMAN_a1r5g5b5 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ARGB,1,5,5,5), + PIXMAN_x1r5g5b5 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ARGB,0,5,5,5), + PIXMAN_a1b5g5r5 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ABGR,1,5,5,5), + PIXMAN_x1b5g5r5 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ABGR,0,5,5,5), + PIXMAN_a4r4g4b4 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ARGB,4,4,4,4), + PIXMAN_x4r4g4b4 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ARGB,0,4,4,4), + PIXMAN_a4b4g4r4 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ABGR,4,4,4,4), + PIXMAN_x4b4g4r4 = PIXMAN_FORMAT(16,PIXMAN_TYPE_ABGR,0,4,4,4), + +/* 8bpp formats */ + PIXMAN_a8 = PIXMAN_FORMAT(8,PIXMAN_TYPE_A,8,0,0,0), + PIXMAN_r3g3b2 = PIXMAN_FORMAT(8,PIXMAN_TYPE_ARGB,0,3,3,2), + PIXMAN_b2g3r3 = PIXMAN_FORMAT(8,PIXMAN_TYPE_ABGR,0,3,3,2), + PIXMAN_a2r2g2b2 = PIXMAN_FORMAT(8,PIXMAN_TYPE_ARGB,2,2,2,2), + PIXMAN_a2b2g2r2 = PIXMAN_FORMAT(8,PIXMAN_TYPE_ABGR,2,2,2,2), + + PIXMAN_c8 = PIXMAN_FORMAT(8,PIXMAN_TYPE_COLOR,0,0,0,0), + PIXMAN_g8 = PIXMAN_FORMAT(8,PIXMAN_TYPE_GRAY,0,0,0,0), + + PIXMAN_x4a4 = PIXMAN_FORMAT(8,PIXMAN_TYPE_A,4,0,0,0), + + PIXMAN_x4c4 = PIXMAN_FORMAT(8,PIXMAN_TYPE_COLOR,0,0,0,0), + PIXMAN_x4g4 = PIXMAN_FORMAT(8,PIXMAN_TYPE_GRAY,0,0,0,0), + +/* 4bpp formats */ + PIXMAN_a4 = PIXMAN_FORMAT(4,PIXMAN_TYPE_A,4,0,0,0), + PIXMAN_r1g2b1 = PIXMAN_FORMAT(4,PIXMAN_TYPE_ARGB,0,1,2,1), + PIXMAN_b1g2r1 = PIXMAN_FORMAT(4,PIXMAN_TYPE_ABGR,0,1,2,1), + PIXMAN_a1r1g1b1 = PIXMAN_FORMAT(4,PIXMAN_TYPE_ARGB,1,1,1,1), + PIXMAN_a1b1g1r1 = PIXMAN_FORMAT(4,PIXMAN_TYPE_ABGR,1,1,1,1), + + PIXMAN_c4 = PIXMAN_FORMAT(4,PIXMAN_TYPE_COLOR,0,0,0,0), + PIXMAN_g4 = PIXMAN_FORMAT(4,PIXMAN_TYPE_GRAY,0,0,0,0), + +/* 1bpp formats */ + PIXMAN_a1 = PIXMAN_FORMAT(1,PIXMAN_TYPE_A,1,0,0,0), + + PIXMAN_g1 = PIXMAN_FORMAT(1,PIXMAN_TYPE_GRAY,0,0,0,0), + +/* YUV formats */ + PIXMAN_yuy2 = PIXMAN_FORMAT(16,PIXMAN_TYPE_YUY2,0,0,0,0), + PIXMAN_yv12 = PIXMAN_FORMAT(12,PIXMAN_TYPE_YV12,0,0,0,0) +} pixman_format_code_t; + +/* Querying supported format values. */ +pixman_bool_t pixman_format_supported_destination (pixman_format_code_t format); +pixman_bool_t pixman_format_supported_source (pixman_format_code_t format); + +/* Constructors */ +pixman_image_t *pixman_image_create_solid_fill (pixman_color_t *color); +pixman_image_t *pixman_image_create_linear_gradient (pixman_point_fixed_t *p1, + pixman_point_fixed_t *p2, + const pixman_gradient_stop_t *stops, + int n_stops); +pixman_image_t *pixman_image_create_radial_gradient (pixman_point_fixed_t *inner, + pixman_point_fixed_t *outer, + pixman_fixed_t inner_radius, + pixman_fixed_t outer_radius, + const pixman_gradient_stop_t *stops, + int n_stops); +pixman_image_t *pixman_image_create_conical_gradient (pixman_point_fixed_t *center, + pixman_fixed_t angle, + const pixman_gradient_stop_t *stops, + int n_stops); +pixman_image_t *pixman_image_create_bits (pixman_format_code_t format, + int width, + int height, + uint32_t *bits, + int rowstride_bytes); + +/* Destructor */ +pixman_image_t *pixman_image_ref (pixman_image_t *image); +pixman_bool_t pixman_image_unref (pixman_image_t *image); + +void pixman_image_set_destroy_function (pixman_image_t *image, + pixman_image_destroy_func_t function, + void *data); +void * pixman_image_get_destroy_data (pixman_image_t *image); + +/* Set properties */ +pixman_bool_t pixman_image_set_clip_region (pixman_image_t *image, + pixman_region16_t *region); +pixman_bool_t pixman_image_set_clip_region32 (pixman_image_t *image, + pixman_region32_t *region); +void pixman_image_set_has_client_clip (pixman_image_t *image, + pixman_bool_t clien_clip); +pixman_bool_t pixman_image_set_transform (pixman_image_t *image, + const pixman_transform_t *transform); +void pixman_image_set_repeat (pixman_image_t *image, + pixman_repeat_t repeat); +pixman_bool_t pixman_image_set_filter (pixman_image_t *image, + pixman_filter_t filter, + const pixman_fixed_t *filter_params, + int n_filter_params); +void pixman_image_set_source_clipping (pixman_image_t *image, + pixman_bool_t source_clipping); +void pixman_image_set_alpha_map (pixman_image_t *image, + pixman_image_t *alpha_map, + int16_t x, + int16_t y); +void pixman_image_set_component_alpha (pixman_image_t *image, + pixman_bool_t component_alpha); +pixman_bool_t pixman_image_get_component_alpha (pixman_image_t *image); +void pixman_image_set_accessors (pixman_image_t *image, + pixman_read_memory_func_t read_func, + pixman_write_memory_func_t write_func); +void pixman_image_set_indexed (pixman_image_t *image, + const pixman_indexed_t *indexed); +uint32_t *pixman_image_get_data (pixman_image_t *image); +int pixman_image_get_width (pixman_image_t *image); +int pixman_image_get_height (pixman_image_t *image); +int pixman_image_get_stride (pixman_image_t *image); /* in bytes */ +int pixman_image_get_depth (pixman_image_t *image); +pixman_format_code_t pixman_image_get_format (pixman_image_t *image); +pixman_bool_t pixman_image_fill_rectangles (pixman_op_t op, + pixman_image_t *image, + pixman_color_t *color, + int n_rects, + const pixman_rectangle16_t *rects); +pixman_bool_t pixman_image_fill_boxes (pixman_op_t op, + pixman_image_t *dest, + pixman_color_t *color, + int n_boxes, + const pixman_box32_t *boxes); + +/* Composite */ +pixman_bool_t pixman_compute_composite_region (pixman_region16_t *region, + pixman_image_t *src_image, + pixman_image_t *mask_image, + pixman_image_t *dst_image, + int16_t src_x, + int16_t src_y, + int16_t mask_x, + int16_t mask_y, + int16_t dest_x, + int16_t dest_y, + uint16_t width, + uint16_t height); +void pixman_image_composite (pixman_op_t op, + pixman_image_t *src, + pixman_image_t *mask, + pixman_image_t *dest, + int16_t src_x, + int16_t src_y, + int16_t mask_x, + int16_t mask_y, + int16_t dest_x, + int16_t dest_y, + uint16_t width, + uint16_t height); +void pixman_image_composite32 (pixman_op_t op, + pixman_image_t *src, + pixman_image_t *mask, + pixman_image_t *dest, + int32_t src_x, + int32_t src_y, + int32_t mask_x, + int32_t mask_y, + int32_t dest_x, + int32_t dest_y, + int32_t width, + int32_t height); + +/* Old X servers rely on out-of-bounds accesses when they are asked + * to composite with a window as the source. They create a pixman image + * pointing to some bogus position in memory, but then they set a clip + * region to the position where the actual bits are. + * + * Due to a bug in old versions of pixman, where it would not clip + * against the image bounds when a clip region was set, this would + * actually work. So by default we allow certain out-of-bound access + * to happen unless explicitly disabled. + * + * Fixed X servers should call this function to disable the workaround. + */ +void pixman_disable_out_of_bounds_workaround (void); + +/* + * Trapezoids + */ +typedef struct pixman_edge pixman_edge_t; +typedef struct pixman_trapezoid pixman_trapezoid_t; +typedef struct pixman_trap pixman_trap_t; +typedef struct pixman_span_fix pixman_span_fix_t; + +/* + * An edge structure. This represents a single polygon edge + * and can be quickly stepped across small or large gaps in the + * sample grid + */ +struct pixman_edge +{ + pixman_fixed_t x; + pixman_fixed_t e; + pixman_fixed_t stepx; + pixman_fixed_t signdx; + pixman_fixed_t dy; + pixman_fixed_t dx; + + pixman_fixed_t stepx_small; + pixman_fixed_t stepx_big; + pixman_fixed_t dx_small; + pixman_fixed_t dx_big; +}; + +struct pixman_trapezoid +{ + pixman_fixed_t top, bottom; + pixman_line_fixed_t left, right; +}; + + +/* whether 't' is a well defined not obviously empty trapezoid */ +#define pixman_trapezoid_valid(t) \ + ((t)->left.p1.y != (t)->left.p2.y && \ + (t)->right.p1.y != (t)->right.p2.y && \ + (int) ((t)->bottom - (t)->top) > 0) + +struct pixman_span_fix +{ + pixman_fixed_t l, r, y; +}; + +struct pixman_trap +{ + pixman_span_fix_t top, bot; +}; + +pixman_fixed_t pixman_sample_ceil_y (pixman_fixed_t y, + int bpp); +pixman_fixed_t pixman_sample_floor_y (pixman_fixed_t y, + int bpp); +void pixman_edge_step (pixman_edge_t *e, + int n); +void pixman_edge_init (pixman_edge_t *e, + int bpp, + pixman_fixed_t y_start, + pixman_fixed_t x_top, + pixman_fixed_t y_top, + pixman_fixed_t x_bot, + pixman_fixed_t y_bot); +void pixman_line_fixed_edge_init (pixman_edge_t *e, + int bpp, + pixman_fixed_t y, + const pixman_line_fixed_t *line, + int x_off, + int y_off); +void pixman_rasterize_edges (pixman_image_t *image, + pixman_edge_t *l, + pixman_edge_t *r, + pixman_fixed_t t, + pixman_fixed_t b); +void pixman_add_traps (pixman_image_t *image, + int16_t x_off, + int16_t y_off, + int ntrap, + pixman_trap_t *traps); +void pixman_add_trapezoids (pixman_image_t *image, + int16_t x_off, + int y_off, + int ntraps, + const pixman_trapezoid_t *traps); +void pixman_rasterize_trapezoid (pixman_image_t *image, + const pixman_trapezoid_t *trap, + int x_off, + int y_off); + +PIXMAN_END_DECLS + +#endif /* PIXMAN_H__ */ diff --git a/programs/develop/libraries/pixman/test/window-test.c b/programs/develop/libraries/pixman/test/window-test.c new file mode 100644 index 0000000000..919fc16ed2 --- /dev/null +++ b/programs/develop/libraries/pixman/test/window-test.c @@ -0,0 +1,173 @@ +#include +#include +#include +#include "pixman-private.h" +#include "pixman.h" + +#define FALSE 0 +#define TRUE 1 + +/* Randomly decide between 32 and 16 bit + * + * Allocate bits with random width, stride and height + * + * Then make up some random offset (dx, dy) + * + * Then make an image with those values. + * + * Do this for both source and destination + * + * Composite them together using OVER. + * + * The bits in the source and the destination should have + * recognizable colors so that the result can be verified. + * + * Ie., walk the bits and verify that they have been composited. + */ + +static int +get_rand (int bound) +{ + return rand () % bound; +} + +static pixman_image_t * +make_image (int width, int height, pixman_bool_t src, int *rx, int *ry) +{ + pixman_format_code_t format; + pixman_image_t *image; + pixman_region32_t region; + uint8_t *bits; + int stride; + int bpp; + int dx, dy; + int i, j; + + if (src) + format = PIXMAN_a8r8g8b8; + else + format = PIXMAN_r5g6b5; + + bpp = PIXMAN_FORMAT_BPP (format) / 8; + + stride = width + get_rand (width); + stride += (stride & 1); /* Make it an even number */ + + bits = malloc (height * stride * bpp); + + for (j = 0; j < height; ++j) + { + for (i = 0; i < width; ++i) + { + uint8_t *pixel = bits + (stride * j + i) * bpp; + + if (src) + *(uint32_t *)pixel = 0x7f00007f; + else + *(uint16_t *)pixel = 0xf100; + } + } + + dx = dy = 0; + + dx = get_rand (500); + dy = get_rand (500); + + if (!src) + { + /* Now simulate the bogus X server translations */ + bits -= (dy * stride + dx) * bpp; + } + + image = pixman_image_create_bits ( + format, width, height, (uint32_t *)bits, stride * bpp); + + if (!src) + { + /* And add the bogus clip region */ + pixman_region32_init_rect (®ion, dx, dy, dx + width, dy + height); + + pixman_image_set_clip_region32 (image, ®ion); + } + + pixman_image_set_source_clipping (image, TRUE); + + if (src) + { + pixman_transform_t trans; + + pixman_transform_init_identity (&trans); + + pixman_transform_translate (&trans, + NULL, + - pixman_int_to_fixed (width / 2), + - pixman_int_to_fixed (height / 2)); + + pixman_transform_scale (&trans, + NULL, + pixman_double_to_fixed (0.5), + pixman_double_to_fixed (0.5)); + + pixman_transform_translate (&trans, + NULL, + pixman_int_to_fixed (width / 2), + pixman_int_to_fixed (height / 2)); + + pixman_image_set_transform (image, &trans); + pixman_image_set_filter (image, PIXMAN_FILTER_BILINEAR, NULL, 0); + pixman_image_set_repeat (image, PIXMAN_REPEAT_PAD); + } + + if (!src) + { + *rx = dx; + *ry = dy; + } + else + { + *rx = *ry = 0; + } + + return image; +} + +int +main () +{ + pixman_image_t *src, *dest; + int src_x, src_y, dest_x, dest_y; + int i, j; + int width = get_rand (499) + 1; + int height = get_rand (499) + 1; + + src = make_image (width, height, TRUE, &src_x, &src_y); + dest = make_image (width, height, FALSE, &dest_x, &dest_y); + + pixman_image_composite ( + PIXMAN_OP_OVER, src, NULL, dest, + src_x, src_y, + -1, -1, + dest_x, dest_y, + width, height); + + for (i = 0; i < height; ++i) + { + for (j = 0; j < width; ++j) + { + uint8_t *bits = (uint8_t *)dest->bits.bits; + int bpp = PIXMAN_FORMAT_BPP (dest->bits.format) / 8; + int stride = dest->bits.rowstride * 4; + + uint8_t *pixel = + bits + (i + dest_y) * stride + (j + dest_x) * bpp; + + if (*(uint16_t *)pixel != 0x788f) + { + printf ("bad pixel %x\n", *(uint16_t *)pixel); + assert (*(uint16_t *)pixel == 0x788f); + } + } + } + + return 0; +}