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/math64.h>
29 #include <linux/types.h>
30 #include <linux/wordpart.h>
31
32 typedef union dfixed {
33 u32 full;
34 } fixed20_12;
35
36
37 #define dfixed_const(A) (u32)(((A) << 12))/* + ((B + 0.000122)*4096)) */
38 #define dfixed_const_half(A) (u32)(((A) << 12) + 2048)
39 #define dfixed_const_666(A) (u32)(((A) << 12) + 2731)
40 #define dfixed_const_8(A) (u32)(((A) << 12) + 3277)
41 #define dfixed_mul(A, B) ((u64)((u64)(A).full * (B).full + 2048) >> 12)
42 #define dfixed_init(A) { .full = dfixed_const((A)) }
43 #define dfixed_init_half(A) { .full = dfixed_const_half((A)) }
44 #define dfixed_trunc(A) ((A).full >> 12)
45 #define dfixed_frac(A) ((A).full & ((1 << 12) - 1))
46
dfixed_floor(fixed20_12 A)47 static inline u32 dfixed_floor(fixed20_12 A)
48 {
49 u32 non_frac = dfixed_trunc(A);
50
51 return dfixed_const(non_frac);
52 }
53
dfixed_ceil(fixed20_12 A)54 static inline u32 dfixed_ceil(fixed20_12 A)
55 {
56 u32 non_frac = dfixed_trunc(A);
57
58 if (A.full > dfixed_const(non_frac))
59 return dfixed_const(non_frac + 1);
60 else
61 return dfixed_const(non_frac);
62 }
63
dfixed_div(fixed20_12 A,fixed20_12 B)64 static inline u32 dfixed_div(fixed20_12 A, fixed20_12 B)
65 {
66 u64 tmp = ((u64)A.full << 13);
67
68 do_div(tmp, B.full);
69 tmp += 1;
70 tmp /= 2;
71 return lower_32_bits(tmp);
72 }
73
74 #define DRM_FIXED_POINT 32
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 /**
82 * @drm_sm2fixp
83 *
84 * Convert a 1.31.32 signed-magnitude fixed point to 32.32
85 * 2s-complement fixed point
86 *
87 * @return s64 2s-complement fixed point
88 */
drm_sm2fixp(__u64 a)89 static inline s64 drm_sm2fixp(__u64 a)
90 {
91 if ((a & (1LL << 63))) {
92 return -(a & 0x7fffffffffffffffll);
93 } else {
94 return a;
95 }
96 }
97
drm_int2fixp(int a)98 static inline s64 drm_int2fixp(int a)
99 {
100 return ((s64)a) << DRM_FIXED_POINT;
101 }
102
drm_fixp2int(s64 a)103 static inline int drm_fixp2int(s64 a)
104 {
105 return ((s64)a) >> DRM_FIXED_POINT;
106 }
107
drm_fixp2int_round(s64 a)108 static inline int drm_fixp2int_round(s64 a)
109 {
110 return drm_fixp2int(a + DRM_FIXED_ONE / 2);
111 }
112
drm_fixp2int_ceil(s64 a)113 static inline int drm_fixp2int_ceil(s64 a)
114 {
115 if (a >= 0)
116 return drm_fixp2int(a + DRM_FIXED_ALMOST_ONE);
117 else
118 return drm_fixp2int(a - DRM_FIXED_ALMOST_ONE);
119 }
120
drm_fixp_msbset(s64 a)121 static inline unsigned drm_fixp_msbset(s64 a)
122 {
123 unsigned shift, sign = (a >> 63) & 1;
124
125 for (shift = 62; shift > 0; --shift)
126 if (((a >> shift) & 1) != sign)
127 return shift;
128
129 return 0;
130 }
131
drm_fixp_mul(s64 a,s64 b)132 static inline s64 drm_fixp_mul(s64 a, s64 b)
133 {
134 unsigned shift = drm_fixp_msbset(a) + drm_fixp_msbset(b);
135 s64 result;
136
137 if (shift > 61) {
138 shift = shift - 61;
139 a >>= (shift >> 1) + (shift & 1);
140 b >>= shift >> 1;
141 } else
142 shift = 0;
143
144 result = a * b;
145
146 if (shift > DRM_FIXED_POINT)
147 return result << (shift - DRM_FIXED_POINT);
148
149 if (shift < DRM_FIXED_POINT)
150 return result >> (DRM_FIXED_POINT - shift);
151
152 return result;
153 }
154
drm_fixp_div(s64 a,s64 b)155 static inline s64 drm_fixp_div(s64 a, s64 b)
156 {
157 unsigned shift = 62 - drm_fixp_msbset(a);
158 s64 result;
159
160 a <<= shift;
161
162 if (shift < DRM_FIXED_POINT)
163 b >>= (DRM_FIXED_POINT - shift);
164
165 result = div64_s64(a, b);
166
167 if (shift > DRM_FIXED_POINT)
168 return result >> (shift - DRM_FIXED_POINT);
169
170 return result;
171 }
172
drm_fixp_from_fraction(s64 a,s64 b)173 static inline s64 drm_fixp_from_fraction(s64 a, s64 b)
174 {
175 s64 res;
176 bool a_neg = a < 0;
177 bool b_neg = b < 0;
178 u64 a_abs = a_neg ? -a : a;
179 u64 b_abs = b_neg ? -b : b;
180 u64 rem;
181
182 /* determine integer part */
183 u64 res_abs = div64_u64_rem(a_abs, b_abs, &rem);
184
185 /* determine fractional part */
186 {
187 u32 i = DRM_FIXED_POINT;
188
189 do {
190 rem <<= 1;
191 res_abs <<= 1;
192 if (rem >= b_abs) {
193 res_abs |= 1;
194 rem -= b_abs;
195 }
196 } while (--i != 0);
197 }
198
199 /* round up LSB */
200 {
201 u64 summand = (rem << 1) >= b_abs;
202
203 res_abs += summand;
204 }
205
206 res = (s64) res_abs;
207 if (a_neg ^ b_neg)
208 res = -res;
209 return res;
210 }
211
drm_fixp_exp(s64 x)212 static inline s64 drm_fixp_exp(s64 x)
213 {
214 s64 tolerance = div64_s64(DRM_FIXED_ONE, 1000000);
215 s64 sum = DRM_FIXED_ONE, term, y = x;
216 u64 count = 1;
217
218 if (x < 0)
219 y = -1 * x;
220
221 term = y;
222
223 while (term >= tolerance) {
224 sum = sum + term;
225 count = count + 1;
226 term = drm_fixp_mul(term, div64_s64(y, count));
227 }
228
229 if (x < 0)
230 sum = drm_fixp_div(DRM_FIXED_ONE, sum);
231
232 return sum;
233 }
234
fxp_q4_from_int(int val_int)235 static inline int fxp_q4_from_int(int val_int)
236 {
237 return val_int << 4;
238 }
239
fxp_q4_to_int(int val_q4)240 static inline int fxp_q4_to_int(int val_q4)
241 {
242 return val_q4 >> 4;
243 }
244
fxp_q4_to_int_roundup(int val_q4)245 static inline int fxp_q4_to_int_roundup(int val_q4)
246 {
247 return (val_q4 + 0xf) >> 4;
248 }
249
fxp_q4_to_frac(int val_q4)250 static inline int fxp_q4_to_frac(int val_q4)
251 {
252 return val_q4 & 0xf;
253 }
254
255 #define FXP_Q4_FMT "%d.%04d"
256 #define FXP_Q4_ARGS(val_q4) fxp_q4_to_int(val_q4), (fxp_q4_to_frac(val_q4) * 625)
257
258 #endif
259