pixman-combine64.h [plain text]
#line 1 "combine.inc"
#define COMPONENT_SIZE 16
#define MASK 0xffffULL
#define ONE_HALF 0x8000ULL
#define G_SHIFT 16
#define B_SHIFT 16 * 2
#define A_SHIFT 16 * 3
#define G_MASK 0xffff0000ULL
#define B_MASK 0xffff00000000ULL
#define A_MASK 0xffff000000000000ULL
#define RB_MASK 0xffff0000ffffULL
#define AG_MASK 0xffff0000ffff0000ULL
#define RB_ONE_HALF 0x800000008000ULL
#define RB_MASK_PLUS_ONE 0x10000000010000ULL
#define Alpha(x) ((x) >> A_SHIFT)
#define IntMult(a,b,t) ( (t) = (a) * (b) + ONE_HALF, ( ( ( (t)>>G_SHIFT ) + (t) )>>G_SHIFT ) )
#define IntDiv(a,b) (((uint32_t) (a) * MASK) / (b))
#define GetComp(v,i) ((uint32_t) (uint16_t) ((v) >> i))
#define Add(x,y,i,t) ((t) = GetComp(x,i) + GetComp(y,i), \
(uint64_t) ((uint16_t) ((t) | (0 - ((t) >> G_SHIFT)))) << (i))
#define FbGen(x,y,i,ax,ay,t,u,v) ((t) = (IntMult(GetComp(y,i),ay,(u)) + \
IntMult(GetComp(x,i),ax,(v))), \
(uint64_t) ((uint16_t) ((t) | \
(0 - ((t) >> G_SHIFT)))) << (i))
#define FbByteMul(x, a) do { \
uint64_t t = ((x & RB_MASK) * a) + RB_ONE_HALF; \
t = (t + ((t >> COMPONENT_SIZE) & RB_MASK)) >> COMPONENT_SIZE; \
t &= RB_MASK; \
\
x = (((x >> COMPONENT_SIZE) & RB_MASK) * a) + RB_ONE_HALF; \
x = (x + ((x >> COMPONENT_SIZE) & RB_MASK)); \
x &= RB_MASK << COMPONENT_SIZE; \
x += t; \
} while (0)
#define FbByteMulAdd(x, a, y) do { \
\
uint64_t t = ((x & RB_MASK) * a) + RB_ONE_HALF; \
t = (t + ((t >> COMPONENT_SIZE) & RB_MASK)) >> COMPONENT_SIZE; \
t &= RB_MASK; \
\
\
t += y & RB_MASK; \
\
\
t |= RB_MASK_PLUS_ONE - ((t >> COMPONENT_SIZE) & RB_MASK); \
t &= RB_MASK; \
\
\
x = (((x >> COMPONENT_SIZE) & RB_MASK) * a) + RB_ONE_HALF; \
x = (x + ((x >> COMPONENT_SIZE) & RB_MASK)) >> COMPONENT_SIZE; \
x &= RB_MASK; \
\
\
x += (y >> COMPONENT_SIZE) & RB_MASK; \
\
\
x |= RB_MASK_PLUS_ONE - ((x >> COMPONENT_SIZE) & RB_MASK); \
x &= RB_MASK; \
\
\
x <<= COMPONENT_SIZE; \
x += t; \
} while (0)
#define FbByteAddMul(x, a, y, b) do { \
uint64_t t; \
uint64_t r = (x >> A_SHIFT) * a + (y >> A_SHIFT) * b + ONE_HALF; \
r += (r >> G_SHIFT); \
r >>= G_SHIFT; \
\
t = (x & G_MASK) * a + (y & G_MASK) * b; \
t += (t >> G_SHIFT) + (ONE_HALF << G_SHIFT); \
t >>= B_SHIFT; \
\
t |= r << B_SHIFT; \
t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK); \
t &= RB_MASK; \
t <<= G_SHIFT; \
\
r = ((x >> B_SHIFT) & MASK) * a + \
((y >> B_SHIFT) & MASK) * b + ONE_HALF; \
r += (r >> G_SHIFT); \
r >>= G_SHIFT; \
\
x = (x & MASK) * a + (y & MASK) * b + ONE_HALF; \
x += (x >> G_SHIFT); \
x >>= G_SHIFT; \
x |= r << B_SHIFT; \
x |= RB_MASK_PLUS_ONE - ((x >> G_SHIFT) & RB_MASK); \
x &= RB_MASK; \
x |= t; \
} while (0)
#define FbByteAddMul_256(x, a, y, b) do { \
uint64_t t = (x & RB_MASK) * a + (y & RB_MASK) * b; \
t >>= G_SHIFT; \
t &= RB_MASK; \
\
x = ((x >> G_SHIFT) & RB_MASK) * a + \
((y >> G_SHIFT) & RB_MASK) * b; \
x &= AG_MASK; \
x += t; \
} while (0)
#define FbByteMulC(x, a) do { \
uint64_t t; \
uint64_t r = (x & MASK) * (a & MASK); \
r |= (x & B_MASK) * ((a >> B_SHIFT) & MASK); \
r += RB_ONE_HALF; \
r = (r + ((r >> G_SHIFT) & RB_MASK)) >> G_SHIFT; \
r &= RB_MASK; \
\
x >>= G_SHIFT; \
t = (x & MASK) * ((a >> G_SHIFT) & MASK); \
t |= (x & B_MASK) * (a >> A_SHIFT); \
t += RB_ONE_HALF; \
t = t + ((t >> G_SHIFT) & RB_MASK); \
x = r | (t & AG_MASK); \
} while (0)
#define FbByteMulAddC(x, a, y) do { \
uint64_t t; \
uint64_t r = (x & MASK) * (a & MASK); \
r |= (x & B_MASK) * ((a >> B_SHIFT) & MASK); \
r += RB_ONE_HALF; \
r = (r + ((r >> G_SHIFT) & RB_MASK)) >> G_SHIFT; \
r &= RB_MASK; \
r += y & RB_MASK; \
r |= RB_MASK_PLUS_ONE - ((r >> G_SHIFT) & RB_MASK); \
r &= RB_MASK; \
\
x >>= G_SHIFT; \
t = (x & MASK) * ((a >> G_SHIFT) & MASK); \
t |= (x & B_MASK) * (a >> A_SHIFT); \
t += RB_ONE_HALF; \
t = (t + ((t >> G_SHIFT) & RB_MASK)) >> G_SHIFT; \
t &= RB_MASK; \
t += (y >> G_SHIFT) & RB_MASK; \
t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK); \
t &= RB_MASK; \
x = r | (t << G_SHIFT); \
} while (0)
#define FbByteAddMulC(x, a, y, b) do { \
uint64_t t; \
uint64_t r = (x >> A_SHIFT) * (a >> A_SHIFT) + \
(y >> A_SHIFT) * b; \
r += (r >> G_SHIFT) + ONE_HALF; \
r >>= G_SHIFT; \
\
t = (x & G_MASK) * ((a >> G_SHIFT) & MASK) + (y & G_MASK) * b; \
t += (t >> G_SHIFT) + (ONE_HALF << G_SHIFT); \
t >>= B_SHIFT; \
\
t |= r << B_SHIFT; \
t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK); \
t &= RB_MASK; \
t <<= G_SHIFT; \
\
r = ((x >> B_SHIFT) & MASK) * ((a >> B_SHIFT) & MASK) + \
((y >> B_SHIFT) & MASK) * b + ONE_HALF; \
r += (r >> G_SHIFT); \
r >>= G_SHIFT; \
\
x = (x & MASK) * (a & MASK) + (y & MASK) * b + ONE_HALF; \
x += (x >> G_SHIFT); \
x >>= G_SHIFT; \
x |= r << B_SHIFT; \
x |= RB_MASK_PLUS_ONE - ((x >> G_SHIFT) & RB_MASK); \
x &= RB_MASK; \
x |= t; \
} while (0)
#define FbByteAdd(x, y) do { \
uint64_t t; \
uint64_t r = (x & RB_MASK) + (y & RB_MASK); \
r |= RB_MASK_PLUS_ONE - ((r >> G_SHIFT) & RB_MASK); \
r &= RB_MASK; \
\
t = ((x >> G_SHIFT) & RB_MASK) + ((y >> G_SHIFT) & RB_MASK); \
t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK); \
r |= (t & RB_MASK) << G_SHIFT; \
x = r; \
} while (0)