1 | /* |
1 | /* |
2 | * This file is part of Deliantra, the Roguelike Realtime MMORPG. |
2 | * This file is part of Deliantra, the Roguelike Realtime MMORPG. |
3 | * |
3 | * |
4 | * Copyright (©) 2005,2006,2007,2008 Marc Alexander Lehmann / Robin Redeker / the Deliantra team |
4 | * Copyright (©) 2005,2006,2007,2008,2009,2010 Marc Alexander Lehmann / Robin Redeker / the Deliantra team |
5 | * |
5 | * |
6 | * Deliantra is free software: you can redistribute it and/or modify |
6 | * Deliantra is free software: you can redistribute it and/or modify it under |
7 | * it under the terms of the GNU General Public License as published by |
7 | * the terms of the Affero GNU General Public License as published by the |
8 | * the Free Software Foundation, either version 3 of the License, or |
8 | * Free Software Foundation, either version 3 of the License, or (at your |
9 | * (at your option) any later version. |
9 | * option) any later version. |
10 | * |
10 | * |
11 | * This program is distributed in the hope that it will be useful, |
11 | * This program is distributed in the hope that it will be useful, |
12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
14 | * GNU General Public License for more details. |
14 | * GNU General Public License for more details. |
15 | * |
15 | * |
16 | * You should have received a copy of the GNU General Public License |
16 | * You should have received a copy of the Affero GNU General Public License |
17 | * along with this program. If not, see <http://www.gnu.org/licenses/>. |
17 | * and the GNU General Public License along with this program. If not, see |
|
|
18 | * <http://www.gnu.org/licenses/>. |
18 | * |
19 | * |
19 | * The authors can be reached via e-mail to <support@deliantra.net> |
20 | * The authors can be reached via e-mail to <support@deliantra.net> |
20 | */ |
21 | */ |
21 | |
22 | |
22 | #ifndef UTIL_H__ |
23 | #ifndef UTIL_H__ |
23 | #define UTIL_H__ |
24 | #define UTIL_H__ |
24 | |
25 | |
|
|
26 | #include <compiler.h> |
|
|
27 | |
25 | #define DEBUG_POISON 0x00 // poison memory before freeing it if != 0 |
28 | #define DEBUG_POISON 0x00 // poison memory before freeing it if != 0 |
26 | #define DEBUG_SALLOC 0 // add a debug wrapper around all sallocs |
29 | #define DEBUG_SALLOC 0 // add a debug wrapper around all sallocs |
27 | #define PREFER_MALLOC 0 // use malloc and not the slice allocator |
30 | #define PREFER_MALLOC 0 // use malloc and not the slice allocator |
28 | |
|
|
29 | #if __GNUC__ >= 3 |
|
|
30 | # define is_constant(c) __builtin_constant_p (c) |
|
|
31 | # define expect(expr,value) __builtin_expect ((expr),(value)) |
|
|
32 | # define prefetch(addr,rw,locality) __builtin_prefetch (addr, rw, locality) |
|
|
33 | #else |
|
|
34 | # define is_constant(c) 0 |
|
|
35 | # define expect(expr,value) (expr) |
|
|
36 | # define prefetch(addr,rw,locality) |
|
|
37 | #endif |
|
|
38 | |
|
|
39 | #if __GNUC__ < 4 || (__GNUC__ == 4 || __GNUC_MINOR__ < 4) |
|
|
40 | # define decltype(x) typeof(x) |
|
|
41 | #endif |
|
|
42 | |
|
|
43 | // put into ifs if you are very sure that the expression |
|
|
44 | // is mostly true or mosty false. note that these return |
|
|
45 | // booleans, not the expression. |
|
|
46 | #define expect_false(expr) expect ((expr) != 0, 0) |
|
|
47 | #define expect_true(expr) expect ((expr) != 0, 1) |
|
|
48 | |
31 | |
49 | #include <pthread.h> |
32 | #include <pthread.h> |
50 | |
33 | |
51 | #include <cstddef> |
34 | #include <cstddef> |
52 | #include <cmath> |
35 | #include <cmath> |
… | |
… | |
116 | |
99 | |
117 | // sign0 returns -1, 0 or +1 |
100 | // sign0 returns -1, 0 or +1 |
118 | template<typename T> |
101 | template<typename T> |
119 | static inline T sign0 (T v) { return v ? sign (v) : 0; } |
102 | static inline T sign0 (T v) { return v ? sign (v) : 0; } |
120 | |
103 | |
|
|
104 | // div* only work correctly for div > 0 |
121 | // div, with correct rounding (< 0.5 downwards, >=0.5 upwards) |
105 | // div, with correct rounding (< 0.5 downwards, >=0.5 upwards) |
122 | template<typename T> static inline T div (T val, T div) { return (val + div / 2) / div; } |
106 | template<typename T> static inline T div (T val, T div) |
|
|
107 | { |
|
|
108 | return expect_false (val < 0) ? - ((-val + (div - 1) / 2) / div) : (val + div / 2) / div; |
|
|
109 | } |
123 | // div, round-up |
110 | // div, round-up |
124 | template<typename T> static inline T div_ru (T val, T div) { return (val + div - 1) / div; } |
111 | template<typename T> static inline T div_ru (T val, T div) |
|
|
112 | { |
|
|
113 | return expect_false (val < 0) ? - ((-val ) / div) : (val + div - 1) / div; |
|
|
114 | } |
125 | // div, round-down |
115 | // div, round-down |
126 | template<typename T> static inline T div_rd (T val, T div) { return (val ) / div; } |
116 | template<typename T> static inline T div_rd (T val, T div) |
|
|
117 | { |
|
|
118 | return expect_false (val < 0) ? - ((-val + (div - 1) ) / div) : (val ) / div; |
|
|
119 | } |
127 | |
120 | |
|
|
121 | // lerp* only work correctly for min_in < max_in |
|
|
122 | // Linear intERPolate, scales val from min_in..max_in to min_out..max_out |
128 | template<typename T> |
123 | template<typename T> |
129 | static inline T |
124 | static inline T |
130 | lerp (T val, T min_in, T max_in, T min_out, T max_out) |
125 | lerp (T val, T min_in, T max_in, T min_out, T max_out) |
131 | { |
126 | { |
132 | return min_out + div <T> ((val - min_in) * (max_out - min_out), max_in - min_in); |
127 | return min_out + div <T> ((val - min_in) * (max_out - min_out), max_in - min_in); |
… | |
… | |
191 | int32_t d = b - a; |
186 | int32_t d = b - a; |
192 | d &= d >> 31; |
187 | d &= d >> 31; |
193 | return b - d; |
188 | return b - d; |
194 | } |
189 | } |
195 | |
190 | |
196 | // this is much faster than crossfires original algorithm |
191 | // this is much faster than crossfire's original algorithm |
197 | // on modern cpus |
192 | // on modern cpus |
198 | inline int |
193 | inline int |
199 | isqrt (int n) |
194 | isqrt (int n) |
200 | { |
195 | { |
201 | return (int)sqrtf ((float)n); |
196 | return (int)sqrtf ((float)n); |
|
|
197 | } |
|
|
198 | |
|
|
199 | // this is kind of like the ^^ operator, if it would exist, without sequence point. |
|
|
200 | // more handy than it looks like, due to the implicit !! done on its arguments |
|
|
201 | inline bool |
|
|
202 | logical_xor (bool a, bool b) |
|
|
203 | { |
|
|
204 | return a != b; |
|
|
205 | } |
|
|
206 | |
|
|
207 | inline bool |
|
|
208 | logical_implies (bool a, bool b) |
|
|
209 | { |
|
|
210 | return a <= b; |
202 | } |
211 | } |
203 | |
212 | |
204 | // this is only twice as fast as naive sqrtf (dx*dy+dy*dy) |
213 | // this is only twice as fast as naive sqrtf (dx*dy+dy*dy) |
205 | #if 0 |
214 | #if 0 |
206 | // and has a max. error of 6 in the range -100..+100. |
215 | // and has a max. error of 6 in the range -100..+100. |
… | |
… | |
231 | absdir (int d) |
240 | absdir (int d) |
232 | { |
241 | { |
233 | return ((d - 1) & 7) + 1; |
242 | return ((d - 1) & 7) + 1; |
234 | } |
243 | } |
235 | |
244 | |
|
|
245 | // avoid ctz name because netbsd or freebsd spams it's namespace with it |
|
|
246 | #if GCC_VERSION(3,4) |
|
|
247 | static inline int least_significant_bit (uint32_t x) |
|
|
248 | { |
|
|
249 | return __builtin_ctz (x); |
|
|
250 | } |
|
|
251 | #else |
|
|
252 | int least_significant_bit (uint32_t x); |
|
|
253 | #endif |
|
|
254 | |
|
|
255 | #define for_all_bits_sparse_32(mask, idxvar) \ |
|
|
256 | for (uint32_t idxvar, mask_ = mask; \ |
|
|
257 | mask_ && ((idxvar = least_significant_bit (mask_)), mask_ &= ~(1 << idxvar), 1);) |
|
|
258 | |
236 | extern ssize_t slice_alloc; // statistics |
259 | extern ssize_t slice_alloc; // statistics |
237 | |
260 | |
238 | void *salloc_ (int n) throw (std::bad_alloc); |
261 | void *salloc_ (int n) throw (std::bad_alloc); |
239 | void *salloc_ (int n, void *src) throw (std::bad_alloc); |
262 | void *salloc_ (int n, void *src) throw (std::bad_alloc); |
240 | |
263 | |
… | |
… | |
389 | // P. L'Ecuyer, “Maximally Equidistributed Combined Tausworthe Generators”, Mathematics of Computation, 65, 213 (1996), 203–213. |
412 | // P. L'Ecuyer, “Maximally Equidistributed Combined Tausworthe Generators”, Mathematics of Computation, 65, 213 (1996), 203–213. |
390 | // http://www.iro.umontreal.ca/~lecuyer/myftp/papers/tausme.ps |
413 | // http://www.iro.umontreal.ca/~lecuyer/myftp/papers/tausme.ps |
391 | // http://www.iro.umontreal.ca/~lecuyer/myftp/papers/tausme2.ps |
414 | // http://www.iro.umontreal.ca/~lecuyer/myftp/papers/tausme2.ps |
392 | struct tausworthe_random_generator |
415 | struct tausworthe_random_generator |
393 | { |
416 | { |
394 | // generator |
|
|
395 | uint32_t state [4]; |
417 | uint32_t state [4]; |
396 | |
418 | |
397 | void operator =(const tausworthe_random_generator &src) |
419 | void operator =(const tausworthe_random_generator &src) |
398 | { |
420 | { |
399 | state [0] = src.state [0]; |
421 | state [0] = src.state [0]; |
… | |
… | |
402 | state [3] = src.state [3]; |
424 | state [3] = src.state [3]; |
403 | } |
425 | } |
404 | |
426 | |
405 | void seed (uint32_t seed); |
427 | void seed (uint32_t seed); |
406 | uint32_t next (); |
428 | uint32_t next (); |
|
|
429 | }; |
407 | |
430 | |
|
|
431 | // Xorshift RNGs, George Marsaglia |
|
|
432 | // http://www.jstatsoft.org/v08/i14/paper |
|
|
433 | // this one is about 40% faster than the tausworthe one above (i.e. not much), |
|
|
434 | // despite the inlining, and has the issue of only creating 2**32-1 numbers. |
|
|
435 | // see also http://www.iro.umontreal.ca/~lecuyer/myftp/papers/xorshift.pdf |
|
|
436 | struct xorshift_random_generator |
|
|
437 | { |
|
|
438 | uint32_t x, y; |
|
|
439 | |
|
|
440 | void operator =(const xorshift_random_generator &src) |
|
|
441 | { |
|
|
442 | x = src.x; |
|
|
443 | y = src.y; |
|
|
444 | } |
|
|
445 | |
|
|
446 | void seed (uint32_t seed) |
|
|
447 | { |
|
|
448 | x = seed; |
|
|
449 | y = seed * 69069U; |
|
|
450 | } |
|
|
451 | |
|
|
452 | uint32_t next () |
|
|
453 | { |
|
|
454 | uint32_t t = x ^ (x << 10); |
|
|
455 | x = y; |
|
|
456 | y = y ^ (y >> 13) ^ t ^ (t >> 10); |
|
|
457 | return y; |
|
|
458 | } |
|
|
459 | }; |
|
|
460 | |
|
|
461 | template<class generator> |
|
|
462 | struct random_number_generator : generator |
|
|
463 | { |
408 | // uniform distribution, 0 .. max (0, num - 1) |
464 | // uniform distribution, 0 .. max (0, num - 1) |
409 | uint32_t operator ()(uint32_t num) |
465 | uint32_t operator ()(uint32_t num) |
410 | { |
466 | { |
411 | return !is_constant (num) ? get_range (num) // non-constant |
467 | return !is_constant (num) ? get_range (num) // non-constant |
412 | : num & (num - 1) ? (next () * (uint64_t)num) >> 32U // constant, non-power-of-two |
468 | : num & (num - 1) ? (this->next () * (uint64_t)num) >> 32U // constant, non-power-of-two |
413 | : next () & (num - 1); // constant, power-of-two |
469 | : this->next () & (num - 1); // constant, power-of-two |
414 | } |
470 | } |
415 | |
471 | |
416 | // return a number within (min .. max) |
472 | // return a number within (min .. max) |
417 | int operator () (int r_min, int r_max) |
473 | int operator () (int r_min, int r_max) |
418 | { |
474 | { |
… | |
… | |
429 | protected: |
485 | protected: |
430 | uint32_t get_range (uint32_t r_max); |
486 | uint32_t get_range (uint32_t r_max); |
431 | int get_range (int r_min, int r_max); |
487 | int get_range (int r_min, int r_max); |
432 | }; |
488 | }; |
433 | |
489 | |
434 | typedef tausworthe_random_generator rand_gen; |
490 | typedef random_number_generator<tausworthe_random_generator> rand_gen; |
435 | |
491 | |
436 | extern rand_gen rndm, rmg_rndm; |
492 | extern rand_gen rndm, rmg_rndm; |
437 | |
493 | |
438 | INTERFACE_CLASS (attachable) |
494 | INTERFACE_CLASS (attachable) |
439 | struct refcnt_base |
495 | struct refcnt_base |
… | |
… | |
504 | typedef refptr<object> object_ptr; |
560 | typedef refptr<object> object_ptr; |
505 | typedef refptr<archetype> arch_ptr; |
561 | typedef refptr<archetype> arch_ptr; |
506 | typedef refptr<client> client_ptr; |
562 | typedef refptr<client> client_ptr; |
507 | typedef refptr<player> player_ptr; |
563 | typedef refptr<player> player_ptr; |
508 | |
564 | |
|
|
565 | #define STRHSH_NULL 2166136261 |
|
|
566 | |
|
|
567 | static inline uint32_t |
|
|
568 | strhsh (const char *s) |
|
|
569 | { |
|
|
570 | // use FNV-1a hash (http://isthe.com/chongo/tech/comp/fnv/) |
|
|
571 | // it is about twice as fast as the one-at-a-time one, |
|
|
572 | // with good distribution. |
|
|
573 | // FNV-1a is faster on many cpus because the multiplication |
|
|
574 | // runs concurrently with the looping logic. |
|
|
575 | uint32_t hash = STRHSH_NULL; |
|
|
576 | |
|
|
577 | while (*s) |
|
|
578 | hash = (hash ^ *s++) * 16777619; |
|
|
579 | |
|
|
580 | return hash; |
|
|
581 | } |
|
|
582 | |
|
|
583 | static inline uint32_t |
|
|
584 | memhsh (const char *s, size_t len) |
|
|
585 | { |
|
|
586 | uint32_t hash = STRHSH_NULL; |
|
|
587 | |
|
|
588 | while (len--) |
|
|
589 | hash = (hash ^ *s++) * 16777619; |
|
|
590 | |
|
|
591 | return hash; |
|
|
592 | } |
|
|
593 | |
509 | struct str_hash |
594 | struct str_hash |
510 | { |
595 | { |
511 | std::size_t operator ()(const char *s) const |
596 | std::size_t operator ()(const char *s) const |
512 | { |
597 | { |
513 | unsigned long hash = 0; |
|
|
514 | |
|
|
515 | /* use the one-at-a-time hash function, which supposedly is |
|
|
516 | * better than the djb2-like one used by perl5.005, but |
|
|
517 | * certainly is better then the bug used here before. |
|
|
518 | * see http://burtleburtle.net/bob/hash/doobs.html |
|
|
519 | */ |
|
|
520 | while (*s) |
|
|
521 | { |
|
|
522 | hash += *s++; |
|
|
523 | hash += hash << 10; |
|
|
524 | hash ^= hash >> 6; |
|
|
525 | } |
|
|
526 | |
|
|
527 | hash += hash << 3; |
|
|
528 | hash ^= hash >> 11; |
|
|
529 | hash += hash << 15; |
|
|
530 | |
|
|
531 | return hash; |
598 | return strhsh (s); |
|
|
599 | } |
|
|
600 | |
|
|
601 | std::size_t operator ()(const shstr &s) const |
|
|
602 | { |
|
|
603 | return strhsh (s); |
532 | } |
604 | } |
533 | }; |
605 | }; |
534 | |
606 | |
535 | struct str_equal |
607 | struct str_equal |
536 | { |
608 | { |
… | |
… | |
629 | erase (&obj); |
701 | erase (&obj); |
630 | } |
702 | } |
631 | }; |
703 | }; |
632 | |
704 | |
633 | // basically does what strncpy should do, but appends "..." to strings exceeding length |
705 | // basically does what strncpy should do, but appends "..." to strings exceeding length |
|
|
706 | // returns the number of bytes actually used (including \0) |
634 | void assign (char *dst, const char *src, int maxlen); |
707 | int assign (char *dst, const char *src, int maxsize); |
635 | |
708 | |
636 | // type-safe version of assign |
709 | // type-safe version of assign |
637 | template<int N> |
710 | template<int N> |
638 | inline void assign (char (&dst)[N], const char *src) |
711 | inline int assign (char (&dst)[N], const char *src) |
639 | { |
712 | { |
640 | assign ((char *)&dst, src, N); |
713 | return assign ((char *)&dst, src, N); |
641 | } |
714 | } |
642 | |
715 | |
643 | typedef double tstamp; |
716 | typedef double tstamp; |
644 | |
717 | |
645 | // return current time as timestamp |
718 | // return current time as timestamp |
646 | tstamp now (); |
719 | tstamp now (); |
647 | |
720 | |
648 | int similar_direction (int a, int b); |
721 | int similar_direction (int a, int b); |
649 | |
722 | |
650 | // like sprintf, but returns a "static" buffer |
723 | // like v?sprintf, but returns a "static" buffer |
651 | const char *format (const char *format, ...); |
724 | char *vformat (const char *format, va_list ap); |
|
|
725 | char *format (const char *format, ...) attribute ((format (printf, 1, 2))); |
|
|
726 | |
|
|
727 | // safety-check player input which will become object->msg |
|
|
728 | bool msg_is_safe (const char *msg); |
652 | |
729 | |
653 | ///////////////////////////////////////////////////////////////////////////// |
730 | ///////////////////////////////////////////////////////////////////////////// |
654 | // threads, very very thin wrappers around pthreads |
731 | // threads, very very thin wrappers around pthreads |
655 | |
732 | |
656 | struct thread |
733 | struct thread |