1 | /* |
2 | * Copyright 2009 Red Hat Inc. |
3 | * |
4 | * Permission is hereby granted, free of charge, to any person obtaining a |
5 | * copy of this software and associated documentation files (the "Software"), |
6 | * to deal in the Software without restriction, including without limitation |
7 | * the rights to use, copy, modify, merge, publish, distribute, sublicense, |
8 | * and/or sell copies of the Software, and to permit persons to whom the |
9 | * Software is furnished to do so, subject to the following conditions: |
10 | * |
11 | * The above copyright notice and this permission notice shall be included in |
12 | * all copies or substantial portions of the Software. |
13 | * |
14 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
15 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
16 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL |
17 | * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR |
18 | * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, |
19 | * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR |
20 | * OTHER DEALINGS IN THE SOFTWARE. |
21 | * |
22 | * Authors: Dave Airlie |
23 | * Christian König |
24 | */ |
25 | #ifndef DRM_FIXED_H |
26 | #define DRM_FIXED_H |
27 | |
28 | #include <linux/kernel.h> |
29 | #include <linux/math64.h> |
30 | |
31 | typedef union dfixed { |
32 | u32 full; |
33 | } fixed20_12; |
34 | |
35 | |
36 | #define dfixed_const(A) (u32)(((A) << 12))/* + ((B + 0.000122)*4096)) */ |
37 | #define dfixed_const_half(A) (u32)(((A) << 12) + 2048) |
38 | #define dfixed_const_666(A) (u32)(((A) << 12) + 2731) |
39 | #define dfixed_const_8(A) (u32)(((A) << 12) + 3277) |
40 | #define dfixed_mul(A, B) ((u64)((u64)(A).full * (B).full + 2048) >> 12) |
41 | #define dfixed_init(A) { .full = dfixed_const((A)) } |
42 | #define dfixed_init_half(A) { .full = dfixed_const_half((A)) } |
43 | #define dfixed_trunc(A) ((A).full >> 12) |
44 | #define dfixed_frac(A) ((A).full & ((1 << 12) - 1)) |
45 | |
46 | static inline u32 dfixed_floor(fixed20_12 A) |
47 | { |
48 | u32 non_frac = dfixed_trunc(A); |
49 | |
50 | return dfixed_const(non_frac); |
51 | } |
52 | |
53 | static inline u32 dfixed_ceil(fixed20_12 A) |
54 | { |
55 | u32 non_frac = dfixed_trunc(A); |
56 | |
57 | if (A.full > dfixed_const(non_frac)) |
58 | return dfixed_const(non_frac + 1); |
59 | else |
60 | return dfixed_const(non_frac); |
61 | } |
62 | |
63 | static inline u32 dfixed_div(fixed20_12 A, fixed20_12 B) |
64 | { |
65 | u64 tmp = ((u64)A.full << 13); |
66 | |
67 | do_div(tmp, B.full); |
68 | tmp += 1; |
69 | tmp /= 2; |
70 | return lower_32_bits(tmp); |
71 | } |
72 | |
73 | #define DRM_FIXED_POINT 32 |
74 | #define DRM_FIXED_POINT_HALF 16 |
75 | #define DRM_FIXED_ONE (1ULL << DRM_FIXED_POINT) |
76 | #define DRM_FIXED_DECIMAL_MASK (DRM_FIXED_ONE - 1) |
77 | #define DRM_FIXED_DIGITS_MASK (~DRM_FIXED_DECIMAL_MASK) |
78 | #define DRM_FIXED_EPSILON 1LL |
79 | #define DRM_FIXED_ALMOST_ONE (DRM_FIXED_ONE - DRM_FIXED_EPSILON) |
80 | |
81 | static inline s64 drm_int2fixp(int a) |
82 | { |
83 | return ((s64)a) << DRM_FIXED_POINT; |
84 | } |
85 | |
86 | static inline int drm_fixp2int(s64 a) |
87 | { |
88 | return ((s64)a) >> DRM_FIXED_POINT; |
89 | } |
90 | |
91 | static inline int drm_fixp2int_round(s64 a) |
92 | { |
93 | return drm_fixp2int(a: a + (1 << (DRM_FIXED_POINT_HALF - 1))); |
94 | } |
95 | |
96 | static inline int drm_fixp2int_ceil(s64 a) |
97 | { |
98 | if (a > 0) |
99 | return drm_fixp2int(a: a + DRM_FIXED_ALMOST_ONE); |
100 | else |
101 | return drm_fixp2int(a: a - DRM_FIXED_ALMOST_ONE); |
102 | } |
103 | |
104 | static inline unsigned drm_fixp_msbset(s64 a) |
105 | { |
106 | unsigned shift, sign = (a >> 63) & 1; |
107 | |
108 | for (shift = 62; shift > 0; --shift) |
109 | if (((a >> shift) & 1) != sign) |
110 | return shift; |
111 | |
112 | return 0; |
113 | } |
114 | |
115 | static inline s64 drm_fixp_mul(s64 a, s64 b) |
116 | { |
117 | unsigned shift = drm_fixp_msbset(a) + drm_fixp_msbset(a: b); |
118 | s64 result; |
119 | |
120 | if (shift > 61) { |
121 | shift = shift - 61; |
122 | a >>= (shift >> 1) + (shift & 1); |
123 | b >>= shift >> 1; |
124 | } else |
125 | shift = 0; |
126 | |
127 | result = a * b; |
128 | |
129 | if (shift > DRM_FIXED_POINT) |
130 | return result << (shift - DRM_FIXED_POINT); |
131 | |
132 | if (shift < DRM_FIXED_POINT) |
133 | return result >> (DRM_FIXED_POINT - shift); |
134 | |
135 | return result; |
136 | } |
137 | |
138 | static inline s64 drm_fixp_div(s64 a, s64 b) |
139 | { |
140 | unsigned shift = 62 - drm_fixp_msbset(a); |
141 | s64 result; |
142 | |
143 | a <<= shift; |
144 | |
145 | if (shift < DRM_FIXED_POINT) |
146 | b >>= (DRM_FIXED_POINT - shift); |
147 | |
148 | result = div64_s64(dividend: a, divisor: b); |
149 | |
150 | if (shift > DRM_FIXED_POINT) |
151 | return result >> (shift - DRM_FIXED_POINT); |
152 | |
153 | return result; |
154 | } |
155 | |
156 | static inline s64 drm_fixp_from_fraction(s64 a, s64 b) |
157 | { |
158 | s64 res; |
159 | bool a_neg = a < 0; |
160 | bool b_neg = b < 0; |
161 | u64 a_abs = a_neg ? -a : a; |
162 | u64 b_abs = b_neg ? -b : b; |
163 | u64 rem; |
164 | |
165 | /* determine integer part */ |
166 | u64 res_abs = div64_u64_rem(dividend: a_abs, divisor: b_abs, remainder: &rem); |
167 | |
168 | /* determine fractional part */ |
169 | { |
170 | u32 i = DRM_FIXED_POINT; |
171 | |
172 | do { |
173 | rem <<= 1; |
174 | res_abs <<= 1; |
175 | if (rem >= b_abs) { |
176 | res_abs |= 1; |
177 | rem -= b_abs; |
178 | } |
179 | } while (--i != 0); |
180 | } |
181 | |
182 | /* round up LSB */ |
183 | { |
184 | u64 summand = (rem << 1) >= b_abs; |
185 | |
186 | res_abs += summand; |
187 | } |
188 | |
189 | res = (s64) res_abs; |
190 | if (a_neg ^ b_neg) |
191 | res = -res; |
192 | return res; |
193 | } |
194 | |
195 | static inline s64 drm_fixp_exp(s64 x) |
196 | { |
197 | s64 tolerance = div64_s64(DRM_FIXED_ONE, divisor: 1000000); |
198 | s64 sum = DRM_FIXED_ONE, term, y = x; |
199 | u64 count = 1; |
200 | |
201 | if (x < 0) |
202 | y = -1 * x; |
203 | |
204 | term = y; |
205 | |
206 | while (term >= tolerance) { |
207 | sum = sum + term; |
208 | count = count + 1; |
209 | term = drm_fixp_mul(a: term, b: div64_s64(dividend: y, divisor: count)); |
210 | } |
211 | |
212 | if (x < 0) |
213 | sum = drm_fixp_div(DRM_FIXED_ONE, b: sum); |
214 | |
215 | return sum; |
216 | } |
217 | |
218 | #endif |
219 | |