Subversion Repositories Kolibri OS

Compare Revisions

Regard whitespace Rev 1891 → Rev 3931

/programs/develop/libraries/pixman/pixman-fast-path.c
30,12 → 30,12
#include <stdlib.h>
#include "pixman-private.h"
#include "pixman-combine32.h"
#include "pixman-fast-path.h"
#include "pixman-inlines.h"
 
static force_inline uint32_t
fetch_24 (uint8_t *a)
{
if (((unsigned long)a) & 1)
if (((uintptr_t)a) & 1)
{
#ifdef WORDS_BIGENDIAN
return (*a << 16) | (*(uint16_t *)(a + 1));
57,7 → 57,7
store_24 (uint8_t *a,
uint32_t v)
{
if (((unsigned long)a) & 1)
if (((uintptr_t)a) & 1)
{
#ifdef WORDS_BIGENDIAN
*a = (uint8_t) (v >> 16);
90,7 → 90,7
return dest;
}
 
static uint32_t
static force_inline uint32_t
in (uint32_t x,
uint8_t y)
{
108,19 → 108,9
*/
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t *src, *src_line;
uint32_t *dst, *dst_line;
uint8_t *mask, *mask_line;
129,7 → 119,7
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 (dest_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);
 
168,19 → 158,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, srca;
uint8_t *dst_line, *dst;
uint8_t *mask_line, *mask, m;
188,7 → 168,7
int32_t w;
uint16_t t;
 
src = _pixman_image_get_solid (src_image, dest_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format);
 
srca = src >> 24;
 
246,19 → 226,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint8_t *dst_line, *dst;
uint8_t *src_line, *src;
int dst_stride, src_stride;
293,19 → 263,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, srca;
uint32_t *dst_line, *dst, d;
uint8_t *mask_line, *mask, m;
312,13 → 272,13
int dst_stride, mask_stride;
int32_t w;
 
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_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 (dest_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--)
351,32 → 311,21
 
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)
pixman_composite_info_t *info)
{
uint32_t src, srca, s;
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, 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);
src = _pixman_image_get_solid (imp, src_image, dest_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 (dest_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--)
408,19 → 357,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, srca, s;
uint32_t *dst_line, *dst, d;
uint32_t *mask_line, *mask, ma;
427,13 → 366,13
int dst_stride, mask_stride;
int32_t w;
 
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_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 (dest_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--)
474,19 → 413,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, srca;
uint8_t *dst_line, *dst;
uint32_t d;
494,13 → 423,13
int dst_stride, mask_stride;
int32_t w;
 
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_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 (dest_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--)
539,19 → 468,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, srca;
uint16_t *dst_line, *dst;
uint32_t d;
559,13 → 478,13
int dst_stride, mask_stride;
int32_t w;
 
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_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 (dest_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--)
588,15 → 507,15
else
{
d = *dst;
d = over (src, CONVERT_0565_TO_0888 (d));
d = over (src, convert_0565_to_0888 (d));
}
*dst = CONVERT_8888_TO_0565 (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);
d = over (in (src, m), convert_0565_to_0888 (d));
*dst = convert_8888_to_0565 (d);
}
dst++;
}
605,19 → 524,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, srca, s;
uint16_t src16;
uint16_t *dst_line, *dst;
626,15 → 535,15
int dst_stride, mask_stride;
int32_t w;
 
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format);
 
srca = src >> 24;
if (src == 0)
return;
 
src16 = CONVERT_8888_TO_0565 (src);
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 (dest_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--)
657,14 → 566,14
else
{
d = *dst;
d = over (src, CONVERT_0565_TO_0888 (d));
*dst = CONVERT_8888_TO_0565 (d);
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);
d = convert_0565_to_0888 (d);
 
s = src;
 
673,7 → 582,7
ma = ~ma;
UN8x4_MUL_UN8x4_ADD_UN8x4 (d, ma, s);
 
*dst = CONVERT_8888_TO_0565 (d);
*dst = convert_8888_to_0565 (d);
}
dst++;
}
682,19 → 591,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t *dst_line, *dst;
uint32_t *src_line, *src, s;
int dst_stride, src_stride;
701,7 → 600,7
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 (dest_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--)
727,25 → 626,15
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
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 (dest_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--)
764,19 → 653,9
#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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint8_t *dst_line, *dst;
uint32_t d;
uint32_t *src_line, *src, s;
784,7 → 663,7
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 (dest_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--)
816,19 → 695,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint16_t *dst_line, *dst;
uint32_t d;
uint32_t *src_line, *src, s;
837,7 → 706,7
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);
PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1);
 
while (height--)
{
860,9 → 729,9
else
{
d = *dst;
d = over (s, CONVERT_0565_TO_0888 (d));
d = over (s, convert_0565_to_0888 (d));
}
*dst = CONVERT_8888_TO_0565 (d);
*dst = convert_8888_to_0565 (d);
}
dst++;
}
870,27 → 739,19
}
 
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)
fast_composite_add_8_8 (pixman_implementation_t *imp,
pixman_composite_info_t *info)
{
uint16_t *dst_line, *dst;
uint32_t *src_line, *src, s;
PIXMAN_COMPOSITE_ARGS (info);
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, uint32_t, src_stride, src_line, 1);
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, 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--)
{
903,7 → 764,16
while (w--)
{
s = *src++;
*dst = CONVERT_8888_TO_0565 (s);
if (s)
{
if (s != 0xff)
{
d = *dst;
t = d + s;
s = t | (0 - (t >> 8));
}
*dst = s;
}
dst++;
}
}
910,29 → 780,19
}
 
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)
fast_composite_add_0565_0565 (pixman_implementation_t *imp,
pixman_composite_info_t *info)
{
uint8_t *dst_line, *dst;
uint8_t *src_line, *src;
PIXMAN_COMPOSITE_ARGS (info);
uint16_t *dst_line, *dst;
uint32_t d;
uint16_t *src_line, *src;
uint32_t s;
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);
PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, uint16_t, src_stride, src_line, 1);
PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1);
 
while (height--)
{
947,13 → 807,14
s = *src++;
if (s)
{
if (s != 0xff)
d = *dst;
s = convert_0565_to_8888 (s);
if (d)
{
d = *dst;
t = d + s;
s = t | (0 - (t >> 8));
d = convert_0565_to_8888 (d);
UN8x4_ADD_UN8x4 (s, d);
}
*dst = s;
*dst = convert_8888_to_0565 (s);
}
dst++;
}
962,19 → 823,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t *dst_line, *dst;
uint32_t *src_line, *src;
int dst_stride, src_stride;
982,7 → 833,7
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);
PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1);
 
while (height--)
{
1012,19 → 863,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint8_t *dst_line, *dst;
uint8_t *mask_line, *mask;
int dst_stride, mask_stride;
1032,9 → 873,9
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 (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);
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format);
sa = (src >> 24);
 
while (height--)
1077,20 → 918,10
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)
fast_composite_add_1_1 (pixman_implementation_t *imp,
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t *dst_line, *dst;
uint32_t *src_line, *src;
int dst_stride, src_stride;
1098,7 → 929,7
 
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,
PIXMAN_IMAGE_GET_LINE (dest_image, 0, dest_y, uint32_t,
dst_stride, dst_line, 1);
 
while (height--)
1123,19 → 954,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, srca;
uint32_t *dst, *dst_line;
uint32_t *mask, *mask_line;
1146,12 → 967,12
if (width <= 0)
return;
 
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format);
srca = src >> 24;
if (src == 0)
return;
 
PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint32_t,
PIXMAN_IMAGE_GET_LINE (dest_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);
1215,19 → 1036,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src, srca;
uint16_t *dst, *dst_line;
uint32_t *mask, *mask_line;
1240,12 → 1051,12
if (width <= 0)
return;
 
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format);
srca = src >> 24;
if (src == 0)
return;
 
PIXMAN_IMAGE_GET_LINE (dst_image, dest_x, dest_y, uint16_t,
PIXMAN_IMAGE_GET_LINE (dest_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);
1253,7 → 1064,7
 
if (srca == 0xff)
{
src565 = CONVERT_8888_TO_0565 (src);
src565 = convert_8888_to_0565 (src);
while (height--)
{
dst = dst_line;
1301,8 → 1112,8
}
if (bitcache & bitmask)
{
d = over (src, CONVERT_0565_TO_0888 (*dst));
*dst = CONVERT_8888_TO_0565 (d);
d = over (src, convert_0565_to_0888 (*dst));
*dst = convert_8888_to_0565 (d);
}
bitmask = UPDATE_BITMASK (bitmask);
dst++;
1317,35 → 1128,29
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t src;
 
src = _pixman_image_get_solid (src_image, dst_image->bits.format);
src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format);
 
if (dst_image->bits.format == PIXMAN_a8)
if (dest_image->bits.format == PIXMAN_a1)
{
src = src >> 31;
}
else if (dest_image->bits.format == PIXMAN_a8)
{
src = src >> 24;
}
else if (dst_image->bits.format == PIXMAN_r5g6b5 ||
dst_image->bits.format == PIXMAN_b5g6r5)
else if (dest_image->bits.format == PIXMAN_r5g6b5 ||
dest_image->bits.format == PIXMAN_b5g6r5)
{
src = CONVERT_8888_TO_0565 (src);
src = convert_8888_to_0565 (src);
}
 
pixman_fill (dst_image->bits.bits, dst_image->bits.rowstride,
PIXMAN_FORMAT_BPP (dst_image->bits.format),
pixman_fill (dest_image->bits.bits, dest_image->bits.rowstride,
PIXMAN_FORMAT_BPP (dest_image->bits.format),
dest_x, dest_y,
width, height,
src);
1353,20 → 1158,10
 
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)
pixman_composite_info_t *info)
{
int bpp = PIXMAN_FORMAT_BPP (dst_image->bits.format) / 8;
PIXMAN_COMPOSITE_ARGS (info);
int bpp = PIXMAN_FORMAT_BPP (dest_image->bits.format) / 8;
uint32_t n_bytes = width * bpp;
int dst_stride, src_stride;
uint8_t *dst;
1373,10 → 1168,10
uint8_t *src;
 
src_stride = src_image->bits.rowstride * 4;
dst_stride = dst_image->bits.rowstride * 4;
dst_stride = dest_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;
dst = (uint8_t *)dest_image->bits.bits + dest_y * dst_stride + dest_x * bpp;
 
while (height--)
{
1387,43 → 1182,211
}
}
 
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);
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 (x888_8888_cover, x888, 8888, uint32_t, uint32_t, SRC, COVER)
FAST_NEAREST (x888_8888_pad, x888, 8888, uint32_t, uint32_t, SRC, PAD)
FAST_NEAREST (x888_8888_normal, x888, 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)
 
#define REPEAT_MIN_WIDTH 32
 
static void
fast_composite_tiled_repeat (pixman_implementation_t *imp,
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
pixman_composite_func_t func;
pixman_format_code_t mask_format;
uint32_t src_flags, mask_flags;
int32_t sx, sy;
int32_t width_remain;
int32_t num_pixels;
int32_t src_width;
int32_t i, j;
pixman_image_t extended_src_image;
uint32_t extended_src[REPEAT_MIN_WIDTH * 2];
pixman_bool_t need_src_extension;
uint32_t *src_line;
int32_t src_stride;
int32_t src_bpp;
pixman_composite_info_t info2 = *info;
 
src_flags = (info->src_flags & ~FAST_PATH_NORMAL_REPEAT) |
FAST_PATH_SAMPLES_COVER_CLIP_NEAREST;
 
if (mask_image)
{
mask_format = mask_image->common.extended_format_code;
mask_flags = info->mask_flags;
}
else
{
mask_format = PIXMAN_null;
mask_flags = FAST_PATH_IS_OPAQUE;
}
 
_pixman_implementation_lookup_composite (
imp->toplevel, info->op,
src_image->common.extended_format_code, src_flags,
mask_format, mask_flags,
dest_image->common.extended_format_code, info->dest_flags,
&imp, &func);
 
src_bpp = PIXMAN_FORMAT_BPP (src_image->bits.format);
 
if (src_image->bits.width < REPEAT_MIN_WIDTH &&
(src_bpp == 32 || src_bpp == 16 || src_bpp == 8) &&
!src_image->bits.indexed)
{
sx = src_x;
sx = MOD (sx, src_image->bits.width);
sx += width;
src_width = 0;
 
while (src_width < REPEAT_MIN_WIDTH && src_width <= sx)
src_width += src_image->bits.width;
 
src_stride = (src_width * (src_bpp >> 3) + 3) / (int) sizeof (uint32_t);
 
/* Initialize/validate stack-allocated temporary image */
_pixman_bits_image_init (&extended_src_image, src_image->bits.format,
src_width, 1, &extended_src[0], src_stride,
FALSE);
_pixman_image_validate (&extended_src_image);
 
info2.src_image = &extended_src_image;
need_src_extension = TRUE;
}
else
{
src_width = src_image->bits.width;
need_src_extension = FALSE;
}
 
sx = src_x;
sy = src_y;
 
while (--height >= 0)
{
sx = MOD (sx, src_width);
sy = MOD (sy, src_image->bits.height);
 
if (need_src_extension)
{
if (src_bpp == 32)
{
PIXMAN_IMAGE_GET_LINE (src_image, 0, sy, uint32_t, src_stride, src_line, 1);
 
for (i = 0; i < src_width; )
{
for (j = 0; j < src_image->bits.width; j++, i++)
extended_src[i] = src_line[j];
}
}
else if (src_bpp == 16)
{
uint16_t *src_line_16;
 
PIXMAN_IMAGE_GET_LINE (src_image, 0, sy, uint16_t, src_stride,
src_line_16, 1);
src_line = (uint32_t*)src_line_16;
 
for (i = 0; i < src_width; )
{
for (j = 0; j < src_image->bits.width; j++, i++)
((uint16_t*)extended_src)[i] = ((uint16_t*)src_line)[j];
}
}
else if (src_bpp == 8)
{
uint8_t *src_line_8;
 
PIXMAN_IMAGE_GET_LINE (src_image, 0, sy, uint8_t, src_stride,
src_line_8, 1);
src_line = (uint32_t*)src_line_8;
 
for (i = 0; i < src_width; )
{
for (j = 0; j < src_image->bits.width; j++, i++)
((uint8_t*)extended_src)[i] = ((uint8_t*)src_line)[j];
}
}
 
info2.src_y = 0;
}
else
{
info2.src_y = sy;
}
 
width_remain = width;
 
while (width_remain > 0)
{
num_pixels = src_width - sx;
 
if (num_pixels > width_remain)
num_pixels = width_remain;
 
info2.src_x = sx;
info2.width = num_pixels;
info2.height = 1;
 
func (imp, &info2);
 
width_remain -= num_pixels;
info2.mask_x += num_pixels;
info2.dest_x += num_pixels;
sx = 0;
}
 
sx = src_x;
sy++;
info2.mask_x = info->mask_x;
info2.mask_y++;
info2.dest_x = info->dest_x;
info2.dest_y++;
}
 
if (need_src_extension)
_pixman_image_fini (&extended_src_image);
}
 
/* 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,
const uint16_t * src,
int32_t w,
pixman_fixed_t vx,
pixman_fixed_t unit_x,
pixman_fixed_t max_vx)
pixman_fixed_t max_vx,
pixman_bool_t fully_transparent_src)
{
uint16_t tmp1, tmp2, tmp3, tmp4;
while ((w -= 4) >= 0)
{
tmp1 = src[pixman_fixed_to_int (vx)];
tmp1 = *(src + pixman_fixed_to_int (vx));
vx += unit_x;
tmp2 = src[pixman_fixed_to_int (vx)];
tmp2 = *(src + pixman_fixed_to_int (vx));
vx += unit_x;
tmp3 = src[pixman_fixed_to_int (vx)];
tmp3 = *(src + pixman_fixed_to_int (vx));
vx += unit_x;
tmp4 = src[pixman_fixed_to_int (vx)];
tmp4 = *(src + pixman_fixed_to_int (vx));
vx += unit_x;
*dst++ = tmp1;
*dst++ = tmp2;
1432,26 → 1395,26
}
if (w & 2)
{
tmp1 = src[pixman_fixed_to_int (vx)];
tmp1 = *(src + pixman_fixed_to_int (vx));
vx += unit_x;
tmp2 = src[pixman_fixed_to_int (vx)];
tmp2 = *(src + pixman_fixed_to_int (vx));
vx += unit_x;
*dst++ = tmp1;
*dst++ = tmp2;
}
if (w & 1)
*dst++ = src[pixman_fixed_to_int (vx)];
*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);
uint16_t, uint16_t, COVER)
FAST_NEAREST_MAINLOOP (565_565_none_SRC,
scaled_nearest_scanline_565_565_SRC,
uint16_t, uint16_t, NONE);
uint16_t, uint16_t, NONE)
FAST_NEAREST_MAINLOOP (565_565_pad_SRC,
scaled_nearest_scanline_565_565_SRC,
uint16_t, uint16_t, PAD);
uint16_t, uint16_t, PAD)
 
static force_inline uint32_t
fetch_nearest (pixman_repeat_t src_repeat,
1460,7 → 1423,7
{
if (repeat (src_repeat, &x, src_width))
{
if (format == PIXMAN_x8r8g8b8)
if (format == PIXMAN_x8r8g8b8 || format == PIXMAN_x8b8g8r8)
return *(src + x) | 0xff000000;
else
return *(src + x);
1493,19 → 1456,9
 
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)
pixman_composite_info_t *info)
{
PIXMAN_COMPOSITE_ARGS (info);
uint32_t *dst_line;
uint32_t *src_line;
int dst_stride, src_stride;
1516,7 → 1469,7
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);
PIXMAN_IMAGE_GET_LINE (dest_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
*/
1613,6 → 1566,252
}
}
 
#define CACHE_LINE_SIZE 64
 
#define FAST_SIMPLE_ROTATE(suffix, pix_type) \
\
static void \
blt_rotated_90_trivial_##suffix (pix_type *dst, \
int dst_stride, \
const pix_type *src, \
int src_stride, \
int w, \
int h) \
{ \
int x, y; \
for (y = 0; y < h; y++) \
{ \
const pix_type *s = src + (h - y - 1); \
pix_type *d = dst + dst_stride * y; \
for (x = 0; x < w; x++) \
{ \
*d++ = *s; \
s += src_stride; \
} \
} \
} \
\
static void \
blt_rotated_270_trivial_##suffix (pix_type *dst, \
int dst_stride, \
const pix_type *src, \
int src_stride, \
int w, \
int h) \
{ \
int x, y; \
for (y = 0; y < h; y++) \
{ \
const pix_type *s = src + src_stride * (w - 1) + y; \
pix_type *d = dst + dst_stride * y; \
for (x = 0; x < w; x++) \
{ \
*d++ = *s; \
s -= src_stride; \
} \
} \
} \
\
static void \
blt_rotated_90_##suffix (pix_type *dst, \
int dst_stride, \
const pix_type *src, \
int src_stride, \
int W, \
int H) \
{ \
int x; \
int leading_pixels = 0, trailing_pixels = 0; \
const int TILE_SIZE = CACHE_LINE_SIZE / sizeof(pix_type); \
\
/* \
* split processing into handling destination as TILE_SIZExH cache line \
* aligned vertical stripes (optimistically assuming that destination \
* stride is a multiple of cache line, if not - it will be just a bit \
* slower) \
*/ \
\
if ((uintptr_t)dst & (CACHE_LINE_SIZE - 1)) \
{ \
leading_pixels = TILE_SIZE - (((uintptr_t)dst & \
(CACHE_LINE_SIZE - 1)) / sizeof(pix_type)); \
if (leading_pixels > W) \
leading_pixels = W; \
\
/* unaligned leading part NxH (where N < TILE_SIZE) */ \
blt_rotated_90_trivial_##suffix ( \
dst, \
dst_stride, \
src, \
src_stride, \
leading_pixels, \
H); \
\
dst += leading_pixels; \
src += leading_pixels * src_stride; \
W -= leading_pixels; \
} \
\
if ((uintptr_t)(dst + W) & (CACHE_LINE_SIZE - 1)) \
{ \
trailing_pixels = (((uintptr_t)(dst + W) & \
(CACHE_LINE_SIZE - 1)) / sizeof(pix_type)); \
if (trailing_pixels > W) \
trailing_pixels = W; \
W -= trailing_pixels; \
} \
\
for (x = 0; x < W; x += TILE_SIZE) \
{ \
/* aligned middle part TILE_SIZExH */ \
blt_rotated_90_trivial_##suffix ( \
dst + x, \
dst_stride, \
src + src_stride * x, \
src_stride, \
TILE_SIZE, \
H); \
} \
\
if (trailing_pixels) \
{ \
/* unaligned trailing part NxH (where N < TILE_SIZE) */ \
blt_rotated_90_trivial_##suffix ( \
dst + W, \
dst_stride, \
src + W * src_stride, \
src_stride, \
trailing_pixels, \
H); \
} \
} \
\
static void \
blt_rotated_270_##suffix (pix_type *dst, \
int dst_stride, \
const pix_type *src, \
int src_stride, \
int W, \
int H) \
{ \
int x; \
int leading_pixels = 0, trailing_pixels = 0; \
const int TILE_SIZE = CACHE_LINE_SIZE / sizeof(pix_type); \
\
/* \
* split processing into handling destination as TILE_SIZExH cache line \
* aligned vertical stripes (optimistically assuming that destination \
* stride is a multiple of cache line, if not - it will be just a bit \
* slower) \
*/ \
\
if ((uintptr_t)dst & (CACHE_LINE_SIZE - 1)) \
{ \
leading_pixels = TILE_SIZE - (((uintptr_t)dst & \
(CACHE_LINE_SIZE - 1)) / sizeof(pix_type)); \
if (leading_pixels > W) \
leading_pixels = W; \
\
/* unaligned leading part NxH (where N < TILE_SIZE) */ \
blt_rotated_270_trivial_##suffix ( \
dst, \
dst_stride, \
src + src_stride * (W - leading_pixels), \
src_stride, \
leading_pixels, \
H); \
\
dst += leading_pixels; \
W -= leading_pixels; \
} \
\
if ((uintptr_t)(dst + W) & (CACHE_LINE_SIZE - 1)) \
{ \
trailing_pixels = (((uintptr_t)(dst + W) & \
(CACHE_LINE_SIZE - 1)) / sizeof(pix_type)); \
if (trailing_pixels > W) \
trailing_pixels = W; \
W -= trailing_pixels; \
src += trailing_pixels * src_stride; \
} \
\
for (x = 0; x < W; x += TILE_SIZE) \
{ \
/* aligned middle part TILE_SIZExH */ \
blt_rotated_270_trivial_##suffix ( \
dst + x, \
dst_stride, \
src + src_stride * (W - x - TILE_SIZE), \
src_stride, \
TILE_SIZE, \
H); \
} \
\
if (trailing_pixels) \
{ \
/* unaligned trailing part NxH (where N < TILE_SIZE) */ \
blt_rotated_270_trivial_##suffix ( \
dst + W, \
dst_stride, \
src - trailing_pixels * src_stride, \
src_stride, \
trailing_pixels, \
H); \
} \
} \
\
static void \
fast_composite_rotate_90_##suffix (pixman_implementation_t *imp, \
pixman_composite_info_t *info) \
{ \
PIXMAN_COMPOSITE_ARGS (info); \
pix_type *dst_line; \
pix_type *src_line; \
int dst_stride, src_stride; \
int src_x_t, src_y_t; \
\
PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, pix_type, \
dst_stride, dst_line, 1); \
src_x_t = -src_y + pixman_fixed_to_int ( \
src_image->common.transform->matrix[0][2] + \
pixman_fixed_1 / 2 - pixman_fixed_e) - height;\
src_y_t = src_x + pixman_fixed_to_int ( \
src_image->common.transform->matrix[1][2] + \
pixman_fixed_1 / 2 - pixman_fixed_e); \
PIXMAN_IMAGE_GET_LINE (src_image, src_x_t, src_y_t, pix_type, \
src_stride, src_line, 1); \
blt_rotated_90_##suffix (dst_line, dst_stride, src_line, src_stride, \
width, height); \
} \
\
static void \
fast_composite_rotate_270_##suffix (pixman_implementation_t *imp, \
pixman_composite_info_t *info) \
{ \
PIXMAN_COMPOSITE_ARGS (info); \
pix_type *dst_line; \
pix_type *src_line; \
int dst_stride, src_stride; \
int src_x_t, src_y_t; \
\
PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, pix_type, \
dst_stride, dst_line, 1); \
src_x_t = src_y + pixman_fixed_to_int ( \
src_image->common.transform->matrix[0][2] + \
pixman_fixed_1 / 2 - pixman_fixed_e); \
src_y_t = -src_x + pixman_fixed_to_int ( \
src_image->common.transform->matrix[1][2] + \
pixman_fixed_1 / 2 - pixman_fixed_e) - width; \
PIXMAN_IMAGE_GET_LINE (src_image, src_x_t, src_y_t, pix_type, \
src_stride, src_line, 1); \
blt_rotated_270_##suffix (dst_line, dst_stride, src_line, src_stride, \
width, height); \
}
 
FAST_SIMPLE_ROTATE (8, uint8_t)
FAST_SIMPLE_ROTATE (565, uint16_t)
FAST_SIMPLE_ROTATE (8888, uint32_t)
 
static const pixman_fast_path_t c_fast_paths[] =
{
PIXMAN_STD_FAST_PATH (OVER, solid, a8, r5g6b5, fast_composite_over_n_8_0565),
1645,10 → 1844,12
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, r5g6b5, null, r5g6b5, fast_composite_add_0565_0565),
PIXMAN_STD_FAST_PATH (ADD, b5g6r5, null, b5g6r5, fast_composite_add_0565_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 (ADD, a1, null, a1, fast_composite_add_1_1),
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),
1655,6 → 1856,7
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, a1, 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),
1675,10 → 1877,6
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),
 
1695,6 → 1893,13
 
SIMPLE_NEAREST_FAST_PATH (SRC, r5g6b5, r5g6b5, 565_565),
 
SIMPLE_NEAREST_FAST_PATH_COVER (SRC, x8r8g8b8, a8r8g8b8, x888_8888),
SIMPLE_NEAREST_FAST_PATH_COVER (SRC, x8b8g8r8, a8b8g8r8, x888_8888),
SIMPLE_NEAREST_FAST_PATH_PAD (SRC, x8r8g8b8, a8r8g8b8, x888_8888),
SIMPLE_NEAREST_FAST_PATH_PAD (SRC, x8b8g8r8, a8b8g8r8, x888_8888),
SIMPLE_NEAREST_FAST_PATH_NORMAL (SRC, x8r8g8b8, a8r8g8b8, x888_8888),
SIMPLE_NEAREST_FAST_PATH_NORMAL (SRC, x8b8g8r8, a8b8g8r8, x888_8888),
 
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),
1730,10 → 1935,122
NEAREST_FAST_PATH (OVER, x8b8g8r8, a8b8g8r8),
NEAREST_FAST_PATH (OVER, a8b8g8r8, a8b8g8r8),
 
#define SIMPLE_ROTATE_FLAGS(angle) \
(FAST_PATH_ROTATE_ ## angle ## _TRANSFORM | \
FAST_PATH_NEAREST_FILTER | \
FAST_PATH_SAMPLES_COVER_CLIP_NEAREST | \
FAST_PATH_STANDARD_FLAGS)
 
#define SIMPLE_ROTATE_FAST_PATH(op,s,d,suffix) \
{ PIXMAN_OP_ ## op, \
PIXMAN_ ## s, SIMPLE_ROTATE_FLAGS (90), \
PIXMAN_null, 0, \
PIXMAN_ ## d, FAST_PATH_STD_DEST_FLAGS, \
fast_composite_rotate_90_##suffix, \
}, \
{ PIXMAN_OP_ ## op, \
PIXMAN_ ## s, SIMPLE_ROTATE_FLAGS (270), \
PIXMAN_null, 0, \
PIXMAN_ ## d, FAST_PATH_STD_DEST_FLAGS, \
fast_composite_rotate_270_##suffix, \
}
 
SIMPLE_ROTATE_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, 8888),
SIMPLE_ROTATE_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, 8888),
SIMPLE_ROTATE_FAST_PATH (SRC, x8r8g8b8, x8r8g8b8, 8888),
SIMPLE_ROTATE_FAST_PATH (SRC, r5g6b5, r5g6b5, 565),
SIMPLE_ROTATE_FAST_PATH (SRC, a8, a8, 8),
 
/* Simple repeat fast path entry. */
{ PIXMAN_OP_any,
PIXMAN_any,
(FAST_PATH_STANDARD_FLAGS | FAST_PATH_ID_TRANSFORM | FAST_PATH_BITS_IMAGE |
FAST_PATH_NORMAL_REPEAT),
PIXMAN_any, 0,
PIXMAN_any, FAST_PATH_STD_DEST_FLAGS,
fast_composite_tiled_repeat
},
 
{ PIXMAN_OP_NONE },
};
 
#ifdef WORDS_BIGENDIAN
#define A1_FILL_MASK(n, offs) (((1U << (n)) - 1) << (32 - (offs) - (n)))
#else
#define A1_FILL_MASK(n, offs) (((1U << (n)) - 1) << (offs))
#endif
 
static force_inline void
pixman_fill1_line (uint32_t *dst, int offs, int width, int v)
{
if (offs)
{
int leading_pixels = 32 - offs;
if (leading_pixels >= width)
{
if (v)
*dst |= A1_FILL_MASK (width, offs);
else
*dst &= ~A1_FILL_MASK (width, offs);
return;
}
else
{
if (v)
*dst++ |= A1_FILL_MASK (leading_pixels, offs);
else
*dst++ &= ~A1_FILL_MASK (leading_pixels, offs);
width -= leading_pixels;
}
}
while (width >= 32)
{
if (v)
*dst++ = 0xFFFFFFFF;
else
*dst++ = 0;
width -= 32;
}
if (width > 0)
{
if (v)
*dst |= A1_FILL_MASK (width, 0);
else
*dst &= ~A1_FILL_MASK (width, 0);
}
}
 
static void
pixman_fill1 (uint32_t *bits,
int stride,
int x,
int y,
int width,
int height,
uint32_t filler)
{
uint32_t *dst = bits + y * stride + (x >> 5);
int offs = x & 31;
 
if (filler & 1)
{
while (height--)
{
pixman_fill1_line (dst, offs, width, 1);
dst += stride;
}
}
else
{
while (height--)
{
pixman_fill1_line (dst, offs, width, 0);
dst += stride;
}
}
}
 
static void
pixman_fill8 (uint32_t *bits,
int stride,
int x,
1740,11 → 2057,11
int y,
int width,
int height,
uint32_t xor)
uint32_t filler)
{
int byte_stride = stride * (int) sizeof (uint32_t);
uint8_t *dst = (uint8_t *) bits;
uint8_t v = xor & 0xff;
uint8_t v = filler & 0xff;
int i;
 
dst = dst + y * byte_stride + x;
1765,12 → 2082,12
int y,
int width,
int height,
uint32_t xor)
uint32_t filler)
{
int short_stride =
(stride * (int)sizeof (uint32_t)) / (int)sizeof (uint16_t);
uint16_t *dst = (uint16_t *)bits;
uint16_t v = xor & 0xffff;
uint16_t v = filler & 0xffff;
int i;
 
dst = dst + y * short_stride + x;
1791,7 → 2108,7
int y,
int width,
int height,
uint32_t xor)
uint32_t filler)
{
int i;
 
1800,7 → 2117,7
while (height--)
{
for (i = 0; i < width; ++i)
bits[i] = xor;
bits[i] = filler;
 
bits += stride;
}
1815,38 → 2132,227
int y,
int width,
int height,
uint32_t xor)
uint32_t filler)
{
switch (bpp)
{
case 1:
pixman_fill1 (bits, stride, x, y, width, height, filler);
break;
 
case 8:
pixman_fill8 (bits, stride, x, y, width, height, xor);
pixman_fill8 (bits, stride, x, y, width, height, filler);
break;
 
case 16:
pixman_fill16 (bits, stride, x, y, width, height, xor);
pixman_fill16 (bits, stride, x, y, width, height, filler);
break;
 
case 32:
pixman_fill32 (bits, stride, x, y, width, height, xor);
pixman_fill32 (bits, stride, x, y, width, height, filler);
break;
 
default:
return _pixman_implementation_fill (
imp->delegate, bits, stride, bpp, x, y, width, height, xor);
break;
return FALSE;
}
 
return TRUE;
}
 
/*****************************************************************************/
 
static uint32_t *
fast_fetch_r5g6b5 (pixman_iter_t *iter, const uint32_t *mask)
{
int32_t w = iter->width;
uint32_t *dst = iter->buffer;
const uint16_t *src = (const uint16_t *)iter->bits;
 
iter->bits += iter->stride;
 
/* Align the source buffer at 4 bytes boundary */
if (w > 0 && ((uintptr_t)src & 3))
{
*dst++ = convert_0565_to_8888 (*src++);
w--;
}
/* Process two pixels per iteration */
while ((w -= 2) >= 0)
{
uint32_t sr, sb, sg, t0, t1;
uint32_t s = *(const uint32_t *)src;
src += 2;
sr = (s >> 8) & 0x00F800F8;
sb = (s << 3) & 0x00F800F8;
sg = (s >> 3) & 0x00FC00FC;
sr |= sr >> 5;
sb |= sb >> 5;
sg |= sg >> 6;
t0 = ((sr << 16) & 0x00FF0000) | ((sg << 8) & 0x0000FF00) |
(sb & 0xFF) | 0xFF000000;
t1 = (sr & 0x00FF0000) | ((sg >> 8) & 0x0000FF00) |
(sb >> 16) | 0xFF000000;
#ifdef WORDS_BIGENDIAN
*dst++ = t1;
*dst++ = t0;
#else
*dst++ = t0;
*dst++ = t1;
#endif
}
if (w & 1)
{
*dst = convert_0565_to_8888 (*src);
}
 
return iter->buffer;
}
 
static uint32_t *
fast_dest_fetch_noop (pixman_iter_t *iter, const uint32_t *mask)
{
iter->bits += iter->stride;
return iter->buffer;
}
 
/* Helper function for a workaround, which tries to ensure that 0x1F001F
* constant is always allocated in a register on RISC architectures.
*/
static force_inline uint32_t
convert_8888_to_0565_workaround (uint32_t s, uint32_t x1F001F)
{
uint32_t a, b;
a = (s >> 3) & x1F001F;
b = s & 0xFC00;
a |= a >> 5;
a |= b >> 5;
return a;
}
 
static void
fast_write_back_r5g6b5 (pixman_iter_t *iter)
{
int32_t w = iter->width;
uint16_t *dst = (uint16_t *)(iter->bits - iter->stride);
const uint32_t *src = iter->buffer;
/* Workaround to ensure that x1F001F variable is allocated in a register */
static volatile uint32_t volatile_x1F001F = 0x1F001F;
uint32_t x1F001F = volatile_x1F001F;
 
while ((w -= 4) >= 0)
{
uint32_t s1 = *src++;
uint32_t s2 = *src++;
uint32_t s3 = *src++;
uint32_t s4 = *src++;
*dst++ = convert_8888_to_0565_workaround (s1, x1F001F);
*dst++ = convert_8888_to_0565_workaround (s2, x1F001F);
*dst++ = convert_8888_to_0565_workaround (s3, x1F001F);
*dst++ = convert_8888_to_0565_workaround (s4, x1F001F);
}
if (w & 2)
{
*dst++ = convert_8888_to_0565_workaround (*src++, x1F001F);
*dst++ = convert_8888_to_0565_workaround (*src++, x1F001F);
}
if (w & 1)
{
*dst = convert_8888_to_0565_workaround (*src, x1F001F);
}
}
 
typedef struct
{
pixman_format_code_t format;
pixman_iter_get_scanline_t get_scanline;
pixman_iter_write_back_t write_back;
} fetcher_info_t;
 
static const fetcher_info_t fetchers[] =
{
{ PIXMAN_r5g6b5, fast_fetch_r5g6b5, fast_write_back_r5g6b5 },
{ PIXMAN_null }
};
 
static pixman_bool_t
fast_src_iter_init (pixman_implementation_t *imp, pixman_iter_t *iter)
{
pixman_image_t *image = iter->image;
 
#define FLAGS \
(FAST_PATH_STANDARD_FLAGS | FAST_PATH_ID_TRANSFORM | \
FAST_PATH_BITS_IMAGE | FAST_PATH_SAMPLES_COVER_CLIP_NEAREST)
 
if ((iter->iter_flags & ITER_NARROW) &&
(iter->image_flags & FLAGS) == FLAGS)
{
const fetcher_info_t *f;
 
for (f = &fetchers[0]; f->format != PIXMAN_null; f++)
{
if (image->common.extended_format_code == f->format)
{
uint8_t *b = (uint8_t *)image->bits.bits;
int s = image->bits.rowstride * 4;
 
iter->bits = b + s * iter->y + iter->x * PIXMAN_FORMAT_BPP (f->format) / 8;
iter->stride = s;
 
iter->get_scanline = f->get_scanline;
return TRUE;
}
}
}
 
return FALSE;
}
 
static pixman_bool_t
fast_dest_iter_init (pixman_implementation_t *imp, pixman_iter_t *iter)
{
pixman_image_t *image = iter->image;
 
if ((iter->iter_flags & ITER_NARROW) &&
(iter->image_flags & FAST_PATH_STD_DEST_FLAGS) == FAST_PATH_STD_DEST_FLAGS)
{
const fetcher_info_t *f;
 
for (f = &fetchers[0]; f->format != PIXMAN_null; f++)
{
if (image->common.extended_format_code == f->format)
{
uint8_t *b = (uint8_t *)image->bits.bits;
int s = image->bits.rowstride * 4;
 
iter->bits = b + s * iter->y + iter->x * PIXMAN_FORMAT_BPP (f->format) / 8;
iter->stride = s;
 
if ((iter->iter_flags & (ITER_IGNORE_RGB | ITER_IGNORE_ALPHA)) ==
(ITER_IGNORE_RGB | ITER_IGNORE_ALPHA))
{
iter->get_scanline = fast_dest_fetch_noop;
}
else
{
iter->get_scanline = f->get_scanline;
}
iter->write_back = f->write_back;
return TRUE;
}
}
}
return FALSE;
}
 
 
pixman_implementation_t *
_pixman_implementation_create_fast_path (void)
_pixman_implementation_create_fast_path (pixman_implementation_t *fallback)
{
pixman_implementation_t *general = _pixman_implementation_create_general ();
pixman_implementation_t *imp = _pixman_implementation_create (general, c_fast_paths);
pixman_implementation_t *imp = _pixman_implementation_create (fallback, c_fast_paths);
 
imp->fill = fast_path_fill;
imp->src_iter_init = fast_src_iter_init;
imp->dest_iter_init = fast_dest_iter_init;
 
return imp;
}