drm_fixed.h 5.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258
  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. #include <linux/math64.h>
  28. #include <linux/types.h>
  29. #include <linux/wordpart.h>
  30. typedef union dfixed {
  31. u32 full;
  32. } fixed20_12;
  33. #define dfixed_const(A) (u32)(((A) << 12))/* + ((B + 0.000122)*4096)) */
  34. #define dfixed_const_half(A) (u32)(((A) << 12) + 2048)
  35. #define dfixed_const_666(A) (u32)(((A) << 12) + 2731)
  36. #define dfixed_const_8(A) (u32)(((A) << 12) + 3277)
  37. #define dfixed_mul(A, B) ((u64)((u64)(A).full * (B).full + 2048) >> 12)
  38. #define dfixed_init(A) { .full = dfixed_const((A)) }
  39. #define dfixed_init_half(A) { .full = dfixed_const_half((A)) }
  40. #define dfixed_trunc(A) ((A).full >> 12)
  41. #define dfixed_frac(A) ((A).full & ((1 << 12) - 1))
  42. static inline u32 dfixed_floor(fixed20_12 A)
  43. {
  44. u32 non_frac = dfixed_trunc(A);
  45. return dfixed_const(non_frac);
  46. }
  47. static inline u32 dfixed_ceil(fixed20_12 A)
  48. {
  49. u32 non_frac = dfixed_trunc(A);
  50. if (A.full > dfixed_const(non_frac))
  51. return dfixed_const(non_frac + 1);
  52. else
  53. return dfixed_const(non_frac);
  54. }
  55. static inline u32 dfixed_div(fixed20_12 A, fixed20_12 B)
  56. {
  57. u64 tmp = ((u64)A.full << 13);
  58. do_div(tmp, B.full);
  59. tmp += 1;
  60. tmp /= 2;
  61. return lower_32_bits(tmp);
  62. }
  63. #define DRM_FIXED_POINT 32
  64. #define DRM_FIXED_ONE (1ULL << DRM_FIXED_POINT)
  65. #define DRM_FIXED_DECIMAL_MASK (DRM_FIXED_ONE - 1)
  66. #define DRM_FIXED_DIGITS_MASK (~DRM_FIXED_DECIMAL_MASK)
  67. #define DRM_FIXED_EPSILON 1LL
  68. #define DRM_FIXED_ALMOST_ONE (DRM_FIXED_ONE - DRM_FIXED_EPSILON)
  69. /**
  70. * @drm_sm2fixp
  71. *
  72. * Convert a 1.31.32 signed-magnitude fixed point to 32.32
  73. * 2s-complement fixed point
  74. *
  75. * @return s64 2s-complement fixed point
  76. */
  77. static inline s64 drm_sm2fixp(__u64 a)
  78. {
  79. if ((a & (1LL << 63))) {
  80. return -(a & 0x7fffffffffffffffll);
  81. } else {
  82. return a;
  83. }
  84. }
  85. static inline s64 drm_int2fixp(int a)
  86. {
  87. return ((s64)a) << DRM_FIXED_POINT;
  88. }
  89. static inline int drm_fixp2int(s64 a)
  90. {
  91. return ((s64)a) >> DRM_FIXED_POINT;
  92. }
  93. static inline int drm_fixp2int_round(s64 a)
  94. {
  95. return drm_fixp2int(a + DRM_FIXED_ONE / 2);
  96. }
  97. static inline int drm_fixp2int_ceil(s64 a)
  98. {
  99. if (a >= 0)
  100. return drm_fixp2int(a + DRM_FIXED_ALMOST_ONE);
  101. else
  102. return drm_fixp2int(a - DRM_FIXED_ALMOST_ONE);
  103. }
  104. static inline unsigned drm_fixp_msbset(s64 a)
  105. {
  106. unsigned shift, sign = (a >> 63) & 1;
  107. for (shift = 62; shift > 0; --shift)
  108. if (((a >> shift) & 1) != sign)
  109. return shift;
  110. return 0;
  111. }
  112. static inline s64 drm_fixp_mul(s64 a, s64 b)
  113. {
  114. unsigned shift = drm_fixp_msbset(a) + drm_fixp_msbset(b);
  115. s64 result;
  116. if (shift > 61) {
  117. shift = shift - 61;
  118. a >>= (shift >> 1) + (shift & 1);
  119. b >>= shift >> 1;
  120. } else
  121. shift = 0;
  122. result = a * b;
  123. if (shift > DRM_FIXED_POINT)
  124. return result << (shift - DRM_FIXED_POINT);
  125. if (shift < DRM_FIXED_POINT)
  126. return result >> (DRM_FIXED_POINT - shift);
  127. return result;
  128. }
  129. static inline s64 drm_fixp_div(s64 a, s64 b)
  130. {
  131. unsigned shift = 62 - drm_fixp_msbset(a);
  132. s64 result;
  133. a <<= shift;
  134. if (shift < DRM_FIXED_POINT)
  135. b >>= (DRM_FIXED_POINT - shift);
  136. result = div64_s64(a, b);
  137. if (shift > DRM_FIXED_POINT)
  138. return result >> (shift - DRM_FIXED_POINT);
  139. return result;
  140. }
  141. static inline s64 drm_fixp_from_fraction(s64 a, s64 b)
  142. {
  143. s64 res;
  144. bool a_neg = a < 0;
  145. bool b_neg = b < 0;
  146. u64 a_abs = a_neg ? -a : a;
  147. u64 b_abs = b_neg ? -b : b;
  148. u64 rem;
  149. /* determine integer part */
  150. u64 res_abs = div64_u64_rem(a_abs, b_abs, &rem);
  151. /* determine fractional part */
  152. {
  153. u32 i = DRM_FIXED_POINT;
  154. do {
  155. rem <<= 1;
  156. res_abs <<= 1;
  157. if (rem >= b_abs) {
  158. res_abs |= 1;
  159. rem -= b_abs;
  160. }
  161. } while (--i != 0);
  162. }
  163. /* round up LSB */
  164. {
  165. u64 summand = (rem << 1) >= b_abs;
  166. res_abs += summand;
  167. }
  168. res = (s64) res_abs;
  169. if (a_neg ^ b_neg)
  170. res = -res;
  171. return res;
  172. }
  173. static inline s64 drm_fixp_exp(s64 x)
  174. {
  175. s64 tolerance = div64_s64(DRM_FIXED_ONE, 1000000);
  176. s64 sum = DRM_FIXED_ONE, term, y = x;
  177. u64 count = 1;
  178. if (x < 0)
  179. y = -1 * x;
  180. term = y;
  181. while (term >= tolerance) {
  182. sum = sum + term;
  183. count = count + 1;
  184. term = drm_fixp_mul(term, div64_s64(y, count));
  185. }
  186. if (x < 0)
  187. sum = drm_fixp_div(DRM_FIXED_ONE, sum);
  188. return sum;
  189. }
  190. static inline int fxp_q4_from_int(int val_int)
  191. {
  192. return val_int << 4;
  193. }
  194. static inline int fxp_q4_to_int(int val_q4)
  195. {
  196. return val_q4 >> 4;
  197. }
  198. static inline int fxp_q4_to_int_roundup(int val_q4)
  199. {
  200. return (val_q4 + 0xf) >> 4;
  201. }
  202. static inline int fxp_q4_to_frac(int val_q4)
  203. {
  204. return val_q4 & 0xf;
  205. }
  206. #define FXP_Q4_FMT "%d.%04d"
  207. #define FXP_Q4_ARGS(val_q4) fxp_q4_to_int(val_q4), (fxp_q4_to_frac(val_q4) * 625)
  208. #endif