aboutsummaryrefslogtreecommitdiffhomepage
path: root/src/core/SkMathPriv.h
blob: 30c5912a96c69ed040149c81e25ecb5e5e70922a (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
/*
 * Copyright 2012 Google Inc.
 *
 * Use of this source code is governed by a BSD-style license that can be
 * found in the LICENSE file.
 */

#ifndef SkMathPriv_DEFINED
#define SkMathPriv_DEFINED

#include "SkMath.h"

#if defined(SK_BUILD_FOR_IOS) && (defined(SK_BUILD_FOR_ARM32) || defined(SK_BUILD_FOR_ARM64))
// iOS on ARM starts processes with the Flush-To-Zero (FTZ) and
// Denormals-Are-Zero (DAZ) bits in the fpscr register set.
// Algorithms that rely on denormalized numbers need alternative implementations.
// This can also be controlled in SSE with the MXCSR register,
// x87 with FSTCW/FLDCW, and mips with FCSR. This should be detected at runtime,
// or the library built one way or the other more generally (by the build).
#define SK_CPU_FLUSH_TO_ZERO
#endif

/** Returns -1 if n < 0, else returns 0
 */
#define SkExtractSign(n)    ((int32_t)(n) >> 31)

/** If sign == -1, returns -n, else sign must be 0, and returns n.
 Typically used in conjunction with SkExtractSign().
 */
static inline int32_t SkApplySign(int32_t n, int32_t sign) {
    SkASSERT(sign == 0 || sign == -1);
    return (n ^ sign) - sign;
}

/** Return x with the sign of y */
static inline int32_t SkCopySign32(int32_t x, int32_t y) {
    return SkApplySign(x, SkExtractSign(x ^ y));
}

/** Given a positive value and a positive max, return the value
 pinned against max.
 Note: only works as long as max - value doesn't wrap around
 @return max if value >= max, else value
 */
static inline unsigned SkClampUMax(unsigned value, unsigned max) {
    if (value > max) {
        value = max;
    }
    return value;
}

// If a signed int holds min_int (e.g. 0x80000000) it is undefined what happens when
// we negate it (even though we *know* we're 2's complement and we'll get the same
// value back). So we create this helper function that casts to size_t (unsigned) first,
// to avoid the complaint.
static inline size_t sk_negate_to_size_t(int32_t value) {
#if defined(_MSC_VER)
#pragma warning(push)
#pragma warning(disable : 4146)  // Thanks MSVC, we know what we're negating an unsigned
#endif
    return -static_cast<size_t>(value);
#if defined(_MSC_VER)
#pragma warning(pop)
#endif
}

///////////////////////////////////////////////////////////////////////////////

/** Return a*b/255, truncating away any fractional bits. Only valid if both
 a and b are 0..255
 */
static inline U8CPU SkMulDiv255Trunc(U8CPU a, U8CPU b) {
    SkASSERT((uint8_t)a == a);
    SkASSERT((uint8_t)b == b);
    unsigned prod = a*b + 1;
    return (prod + (prod >> 8)) >> 8;
}

/** Return (a*b)/255, taking the ceiling of any fractional bits. Only valid if
 both a and b are 0..255. The expected result equals (a * b + 254) / 255.
 */
static inline U8CPU SkMulDiv255Ceiling(U8CPU a, U8CPU b) {
    SkASSERT((uint8_t)a == a);
    SkASSERT((uint8_t)b == b);
    unsigned prod = a*b + 255;
    return (prod + (prod >> 8)) >> 8;
}

/** Just the rounding step in SkDiv255Round: round(value / 255)
 */
static inline unsigned SkDiv255Round(unsigned prod) {
    prod += 128;
    return (prod + (prod >> 8)) >> 8;
}

static inline float SkPinToUnitFloat(float x) {
    return SkTMin(SkTMax(x, 0.0f), 1.0f);
}

/**
 * Swap byte order of a 4-byte value, e.g. 0xaarrggbb -> 0xbbggrraa.
 */
#if defined(_MSC_VER)
    #include <intrin.h>
    static inline uint32_t SkBSwap32(uint32_t v) { return _byteswap_ulong(v); }
#else
    static inline uint32_t SkBSwap32(uint32_t v) { return __builtin_bswap32(v); }
#endif

//! Returns the number of leading zero bits (0...32)
int SkCLZ_portable(uint32_t);

#ifndef SkCLZ
    #if defined(SK_BUILD_FOR_WIN)
        #include <intrin.h>

        static inline int SkCLZ(uint32_t mask) {
            if (mask) {
                unsigned long index;
                _BitScanReverse(&index, mask);
                // Suppress this bogus /analyze warning. The check for non-zero
                // guarantees that _BitScanReverse will succeed.
#pragma warning(suppress : 6102) // Using 'index' from failed function call
                return index ^ 0x1F;
            } else {
                return 32;
            }
        }
    #elif defined(SK_CPU_ARM32) || defined(__GNUC__) || defined(__clang__)
        static inline int SkCLZ(uint32_t mask) {
            // __builtin_clz(0) is undefined, so we have to detect that case.
            return mask ? __builtin_clz(mask) : 32;
        }
    #else
        #define SkCLZ(x)    SkCLZ_portable(x)
    #endif
#endif

/**
 *  Returns the smallest power-of-2 that is >= the specified value. If value
 *  is already a power of 2, then it is returned unchanged. It is undefined
 *  if value is <= 0.
 */
static inline int SkNextPow2(int value) {
    SkASSERT(value > 0);
    return 1 << (32 - SkCLZ(value - 1));
}

/**
*  Returns the largest power-of-2 that is <= the specified value. If value
*  is already a power of 2, then it is returned unchanged. It is undefined
*  if value is <= 0.
*/
static inline int SkPrevPow2(int value) {
    SkASSERT(value > 0);
    return 1 << (32 - SkCLZ(value >> 1));
}

/**
 *  Returns the log2 of the specified value, were that value to be rounded up
 *  to the next power of 2. It is undefined to pass 0. Examples:
 *  SkNextLog2(1) -> 0
 *  SkNextLog2(2) -> 1
 *  SkNextLog2(3) -> 2
 *  SkNextLog2(4) -> 2
 *  SkNextLog2(5) -> 3
 */
static inline int SkNextLog2(uint32_t value) {
    SkASSERT(value != 0);
    return 32 - SkCLZ(value - 1);
}

/**
*  Returns the log2 of the specified value, were that value to be rounded down
*  to the previous power of 2. It is undefined to pass 0. Examples:
*  SkPrevLog2(1) -> 0
*  SkPrevLog2(2) -> 1
*  SkPrevLog2(3) -> 1
*  SkPrevLog2(4) -> 2
*  SkPrevLog2(5) -> 2
*/
static inline int SkPrevLog2(uint32_t value) {
    SkASSERT(value != 0);
    return 32 - SkCLZ(value >> 1);
}

///////////////////////////////////////////////////////////////////////////////

/**
 *  Return the next power of 2 >= n.
 */
static inline uint32_t GrNextPow2(uint32_t n) {
    return n ? (1 << (32 - SkCLZ(n - 1))) : 1;
}

/**
 * Returns the next power of 2 >= n or n if the next power of 2 can't be represented by size_t.
 */
static inline size_t GrNextSizePow2(size_t n) {
    constexpr int kNumSizeTBits = 8 * sizeof(size_t);
    constexpr size_t kHighBitSet = size_t(1) << (kNumSizeTBits - 1);

    if (!n) {
        return 1;
    } else if (n >= kHighBitSet) {
        return n;
    }

    n--;
    uint32_t shift = 1;
    while (shift < kNumSizeTBits) {
        n |= n >> shift;
        shift <<= 1;
    }
    return n + 1;
}

// conservative check. will return false for very large values that "could" fit
template <typename T> static inline bool SkFitsInFixed(T x) {
    return SkTAbs(x) <= 32767.0f;
}

#endif