File: | solenv/bin/concat-deps.c |
Location: | line 927, column 12 |
Description: | Memory is never released; potential leak of memory pointed to by 'in_list_base' |
1 | /* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */ | |||
2 | /* | |||
3 | * Copyright (C) 2011 Norbert Thiebaud | |||
4 | * License: GPLv3 | |||
5 | */ | |||
6 | ||||
7 | /* define to activate stats reporting on hash usage*/ | |||
8 | /* #define HASH_STAT */ | |||
9 | ||||
10 | /* =============================================== | |||
11 | * Set-up: defines to identify the system and system related properties | |||
12 | * =============================================== | |||
13 | */ | |||
14 | ||||
15 | #ifdef __APPLE__ | |||
16 | #ifdef __x86_64__ | |||
17 | #define CORE_BIG_ENDIAN0 0 | |||
18 | #define CORE_LITTLE_ENDIAN1 1 | |||
19 | #define USE_MEMORY_ALIGNMENT64 64 /* big value -> no alignment */ | |||
20 | #else | |||
21 | #define CORE_BIG_ENDIAN0 1 | |||
22 | #define CORE_LITTLE_ENDIAN1 0 | |||
23 | #define USE_MEMORY_ALIGNMENT64 4 | |||
24 | #endif | |||
25 | ||||
26 | #endif | |||
27 | #ifdef _AIX | |||
28 | #define CORE_BIG_ENDIAN0 1 | |||
29 | #define CORE_LITTLE_ENDIAN1 0 | |||
30 | #define USE_MEMORY_ALIGNMENT64 4 | |||
31 | #endif /* Def _AIX */ | |||
32 | ||||
33 | #ifdef __CYGWIN__ | |||
34 | #define __windows | |||
35 | #define CORE_BIG_ENDIAN0 0 | |||
36 | #define CORE_LITTLE_ENDIAN1 1 | |||
37 | #define USE_MEMORY_ALIGNMENT64 64 /* big value -> no alignment */ | |||
38 | #endif /* Def __CYGWIN__ */ | |||
39 | ||||
40 | #if defined(__linux1) || defined(__OpenBSD__) || \ | |||
41 | defined(__FreeBSD__) || defined(__NetBSD__) || \ | |||
42 | defined(__DragonFly__) | |||
43 | #if __BYTE_ORDER1234 == __LITTLE_ENDIAN1234 | |||
44 | #define CORE_BIG_ENDIAN0 0 | |||
45 | #define CORE_LITTLE_ENDIAN1 1 | |||
46 | #if defined(__x86_64) || defined(__i3861) | |||
47 | #define USE_MEMORY_ALIGNMENT64 64 | |||
48 | #else | |||
49 | #define USE_MEMORY_ALIGNMENT64 4 | |||
50 | #endif | |||
51 | #else /* !(__BYTE_ORDER == __LITTLE_ENDIAN) */ | |||
52 | #if __BYTE_ORDER1234 == __BIG_ENDIAN4321 | |||
53 | #define CORE_BIG_ENDIAN0 1 | |||
54 | #define CORE_LITTLE_ENDIAN1 0 | |||
55 | #define USE_MEMORY_ALIGNMENT64 4 | |||
56 | #endif /* __BYTE_ORDER == __BIG_ENDIAN */ | |||
57 | #endif /* !(__BYTE_ORDER == __LITTLE_ENDIAN) */ | |||
58 | #endif /* Def __linux || Def *BSD */ | |||
59 | ||||
60 | #ifdef __sun | |||
61 | #ifdef __sparc | |||
62 | #define CORE_BIG_ENDIAN0 1 | |||
63 | #define CORE_LITTLE_ENDIAN1 0 | |||
64 | #define USE_MEMORY_ALIGNMENT64 4 | |||
65 | #else /* Ndef __sparc */ | |||
66 | #define CORE_BIG_ENDIAN0 0 | |||
67 | #define CORE_LITTLE_ENDIAN1 1 | |||
68 | #define USE_MEMORY_ALIGNMENT64 4 | |||
69 | #endif /* Ndef __sparc */ | |||
70 | #endif /* Def __sun */ | |||
71 | ||||
72 | /* Note USE_MEMORY_ALIGNMENT is 4 for platform that allow short non-aligned but required int access to be aligned (e.g sparc, ppc, zos..) | |||
73 | * USE_MEMORY_ALIGNMENT is 2 for platform that require short and int access to be aligned (e.g hppa ) | |||
74 | * if the platform does not have alignment requirement (x86/amd64) use a big value (i.e > 16) | |||
75 | */ | |||
76 | #ifndef USE_MEMORY_ALIGNMENT64 | |||
77 | #error "USE_MEMORY_ALIGNMENT must be defined to the proper alignment value for the platform" | |||
78 | #endif | |||
79 | ||||
80 | #include <assert.h> | |||
81 | #include <stdio.h> | |||
82 | #include <stdlib.h> | |||
83 | #include <sys/types.h> | |||
84 | #include <sys/stat.h> | |||
85 | #include <errno(*__errno_location ()).h> | |||
86 | #include <fcntl.h> | |||
87 | #include <string.h> | |||
88 | #include <ctype.h> | |||
89 | ||||
90 | #ifdef __windows | |||
91 | #include <io.h> | |||
92 | #else | |||
93 | #include <unistd.h> | |||
94 | #endif | |||
95 | ||||
96 | /* modes */ | |||
97 | #ifdef __windows | |||
98 | #define FILE_O_RDONLY00 _O_RDONLY | |||
99 | #define FILE_O_BINARY0 _O_BINARY | |||
100 | #else /* not windaube */ | |||
101 | #define FILE_O_RDONLY00 O_RDONLY00 | |||
102 | #define FILE_O_BINARY0 0 | |||
103 | #endif /* not windaube */ | |||
104 | ||||
105 | #ifndef TRUE1 | |||
106 | #define TRUE1 1 | |||
107 | #endif | |||
108 | #ifndef FALSE0 | |||
109 | #define FALSE0 0 | |||
110 | #endif | |||
111 | ||||
112 | static char* base_dir_var = "$(SRCDIR)"; | |||
113 | #define kBASE_DIR_VAR_LENGTH9 9 | |||
114 | static char* base_dir; | |||
115 | ||||
116 | #ifdef __GNUC__4 | |||
117 | #define clz__builtin_clz __builtin_clz | |||
118 | #else | |||
119 | static inline int clz__builtin_clz(unsigned int value) | |||
120 | { | |||
121 | int result = 32; | |||
122 | ||||
123 | while(value) | |||
124 | { | |||
125 | value >>= 1; | |||
126 | result -= 1; | |||
127 | } | |||
128 | return result; | |||
129 | } | |||
130 | #endif | |||
131 | ||||
132 | #if (USE_MEMORY_ALIGNMENT64 > 4) | |||
133 | #define get_unaligned_uint(str)(*(unsigned int*)(str)) (*(unsigned int*)(str)) | |||
134 | #else | |||
135 | static inline unsigned int get_unaligned_uint(const unsigned char* cursor)(*(unsigned int*)(const unsigned char* cursor)) | |||
136 | { | |||
137 | unsigned int result; | |||
138 | ||||
139 | memcpy(&result, cursor, sizeof(unsigned int)); | |||
140 | return result; | |||
141 | } | |||
142 | #endif | |||
143 | ||||
144 | /* =============================================== | |||
145 | * memory pool for fast fix-size allocation (non-tread-safe) | |||
146 | * =============================================== | |||
147 | */ | |||
148 | struct pool | |||
149 | { | |||
150 | void* head_free; /**< head of a linked list of freed element */ | |||
151 | char* fresh; /**< top of a memory block to dig new element */ | |||
152 | char* tail; /**< to detect end of extent... when fresh pass tail */ | |||
153 | void* extent; /**< pointer to the primary extent block */ | |||
154 | int size_elem; /**< size of an element. */ | |||
155 | int primary; /**< primary allocation in bytes */ | |||
156 | int secondary; /**< secondary allocation in bytes */ | |||
157 | }; | |||
158 | #define POOL_ALIGN_INCREMENT8 8 /**< Alignement, must be a power of 2 and of size > to sizeof(void*) */ | |||
159 | ||||
160 | ||||
161 | static void* pool_take_extent(struct pool* pool, int allocate) | |||
162 | { | |||
163 | unsigned int size = 0; | |||
164 | void* extent; | |||
165 | void* data = NULL((void*)0); | |||
166 | ||||
167 | if(pool->extent) | |||
168 | { | |||
169 | fputs("taking a pool extent\n", stderrstderr); | |||
170 | /* we already have an extent, so this is a secondary */ | |||
171 | if(pool->secondary) | |||
172 | { | |||
173 | size = pool->secondary; | |||
174 | } | |||
175 | } | |||
176 | else | |||
177 | { | |||
178 | assert(pool->primary)((pool->primary) ? (void) (0) : __assert_fail ("pool->primary" , "/usr/local/src/libreoffice/solenv/bin/concat-deps.c", 178, __PRETTY_FUNCTION__)); | |||
179 | size = pool->primary; | |||
180 | } | |||
181 | if(size) | |||
182 | { | |||
183 | extent = malloc(size); | |||
184 | if(extent) | |||
185 | { | |||
186 | *(void**)extent = pool->extent; | |||
187 | pool->extent = extent; | |||
188 | if(allocate) | |||
189 | { | |||
190 | data = ((char*)extent) + POOL_ALIGN_INCREMENT8; | |||
191 | pool->fresh = ((char*)data) + pool->size_elem; | |||
192 | pool->tail = pool->fresh + (size - pool->size_elem); | |||
193 | } | |||
194 | else | |||
195 | { | |||
196 | pool->fresh = ((char*)extent) + POOL_ALIGN_INCREMENT8; | |||
197 | pool->tail = pool->fresh + (size - pool->size_elem); | |||
198 | } | |||
199 | } | |||
200 | } | |||
201 | return data; | |||
202 | } | |||
203 | ||||
204 | /* Create a memory pool for fix size objects | |||
205 | * this is a simplified implementation that | |||
206 | * is _not_ thread safe. | |||
207 | */ | |||
208 | struct pool* pool_create(int size_elem, int flags, int primary, int secondary) | |||
209 | { | |||
210 | struct pool* pool; | |||
211 | ||||
212 | assert(primary > 0)((primary > 0) ? (void) (0) : __assert_fail ("primary > 0" , "/usr/local/src/libreoffice/solenv/bin/concat-deps.c", 212, __PRETTY_FUNCTION__)); | |||
213 | assert(secondary >= 0)((secondary >= 0) ? (void) (0) : __assert_fail ("secondary >= 0" , "/usr/local/src/libreoffice/solenv/bin/concat-deps.c", 213, __PRETTY_FUNCTION__)); | |||
214 | assert(size_elem > 0)((size_elem > 0) ? (void) (0) : __assert_fail ("size_elem > 0" , "/usr/local/src/libreoffice/solenv/bin/concat-deps.c", 214, __PRETTY_FUNCTION__)); | |||
215 | ||||
216 | pool = (struct pool*)calloc(1, sizeof(struct pool)); | |||
217 | if(!pool) return NULL((void*)0); | |||
218 | /* Adjust the element size so that it be aligned, and so that an element could | |||
219 | * at least contain a void* | |||
220 | */ | |||
221 | pool->size_elem = size_elem = (size_elem + POOL_ALIGN_INCREMENT8 - 1) & ~(POOL_ALIGN_INCREMENT8 - 1); | |||
222 | ||||
223 | pool->primary = (size_elem * primary) + POOL_ALIGN_INCREMENT8; | |||
224 | pool->secondary = secondary > 0 ? (size_elem * secondary) + POOL_ALIGN_INCREMENT8 : 0; | |||
225 | pool_take_extent(pool, FALSE0); | |||
226 | ||||
227 | return pool; | |||
228 | ||||
229 | } | |||
230 | ||||
231 | void pool_destroy(struct pool* pool) | |||
232 | { | |||
233 | void* extent; | |||
234 | void* next; | |||
235 | ||||
236 | if(pool != NULL((void*)0)) | |||
237 | { | |||
238 | extent = pool->extent; | |||
239 | while(extent) | |||
240 | { | |||
241 | next = *(void**)extent; | |||
242 | free(extent); | |||
243 | extent = next; | |||
244 | } | |||
245 | free(pool); | |||
246 | } | |||
247 | } | |||
248 | ||||
249 | static inline void* pool_alloc(struct pool* pool) | |||
250 | { | |||
251 | void* data; | |||
252 | ||||
253 | data = pool->head_free; | |||
254 | if(data == NULL((void*)0)) | |||
255 | { | |||
256 | /* we have no old-freed elem */ | |||
257 | if(pool->fresh <= pool->tail) | |||
258 | { | |||
259 | /* pick a slice of the current extent */ | |||
260 | data = (void*)pool->fresh; | |||
261 | pool->fresh += pool->size_elem; | |||
262 | } | |||
263 | else | |||
264 | { | |||
265 | /* allocate a new extent */ | |||
266 | data = pool_take_extent(pool, TRUE1); | |||
267 | } | |||
268 | } | |||
269 | else | |||
270 | { | |||
271 | /* re-used old freed element by chopipng the head of the free list */ | |||
272 | pool->head_free = *(void**)data; | |||
273 | } | |||
274 | ||||
275 | return data; | |||
276 | } | |||
277 | ||||
278 | ||||
279 | static inline void pool_free(struct pool* pool, void* data) | |||
280 | { | |||
281 | assert(pool && data)((pool && data) ? (void) (0) : __assert_fail ("pool && data" , "/usr/local/src/libreoffice/solenv/bin/concat-deps.c", 281, __PRETTY_FUNCTION__)); | |||
282 | ||||
283 | /* stack on top of the free list */ | |||
284 | *(void**)data = pool->head_free; | |||
285 | pool->head_free = data; | |||
286 | } | |||
287 | ||||
288 | ||||
289 | /* =============================================== | |||
290 | * Hash implementation custumized to be just tracking | |||
291 | * a unique list of string (i.e no data associated | |||
292 | * with the key, no need for retrieval, etc.. | |||
293 | * | |||
294 | * This is tuned for the particular use-case we have here | |||
295 | * measures in tail_build showed that | |||
296 | * we can get north of 4000 distinct values stored in a hash | |||
297 | * the collision rate is at worse around 2% | |||
298 | * the collision needing an expensive memcmp to resolve | |||
299 | * have a rate typically at 1 per 1000 | |||
300 | * for tail_build we register 37229 unique key | |||
301 | * with a total of 377 extra memcmp needed | |||
302 | * which is completely negligible compared to the | |||
303 | * number of memcmp required to eliminate duplicate | |||
304 | * entry (north of 2.5 millions for tail_build) | |||
305 | * =============================================== | |||
306 | */ | |||
307 | ||||
308 | struct hash_elem | |||
309 | { | |||
310 | struct hash_elem* next; | |||
311 | const char* key; | |||
312 | int key_len; | |||
313 | }; | |||
314 | ||||
315 | struct hash | |||
316 | { | |||
317 | struct hash_elem** array; | |||
318 | struct pool* elems_pool; | |||
319 | int flags; | |||
320 | unsigned int used; | |||
321 | unsigned int size; | |||
322 | unsigned int load_limit; | |||
323 | #ifdef HASH_STAT | |||
324 | int stored; | |||
325 | int collisions; | |||
326 | int cost; | |||
327 | int memcmp; | |||
328 | #endif | |||
329 | }; | |||
330 | #define HASH_F_NO_RESIZE(1<<0) (1<<0) | |||
331 | ||||
332 | /* The following hash_compute function was adapted from : | |||
333 | * lookup3.c, by Bob Jenkins, May 2006, Public Domain. | |||
334 | * | |||
335 | * The changes from the original are mostly cosmetic | |||
336 | */ | |||
337 | #define hashsize(n)(1<<(n)) (1<<(n)) | |||
338 | #define hashmask(n)((1<<(n))-1) (hashsize(n)(1<<(n))-1) | |||
339 | #define rot(x,k)(((x)<<(k)) | ((x)>>(32 -(k)))) (((x)<<(k)) | ((x)>>(32-(k)))) | |||
340 | ||||
341 | ||||
342 | #if CORE_BIG_ENDIAN0 | |||
343 | #define MASK_C10xFFFFFF 0xFFFFFF00 | |||
344 | #define MASK_C20xFFFF 0xFFFF0000 | |||
345 | #define MASK_C30xFF 0xFF000000 | |||
346 | #else | |||
347 | #if CORE_LITTLE_ENDIAN1 | |||
348 | #define MASK_C10xFFFFFF 0xFFFFFF | |||
349 | #define MASK_C20xFFFF 0xFFFF | |||
350 | #define MASK_C30xFF 0xFF | |||
351 | #else | |||
352 | #error "Missing Endianness definition" | |||
353 | #endif | |||
354 | #endif | |||
355 | ||||
356 | ||||
357 | #define mix(a,b,c){ a -= c; a ^= (((c)<<(4)) | ((c)>>(32 -(4)))); c += b; b -= a; b ^= (((a)<<(6)) | ((a)>>(32 -(6)) )); a += c; c -= b; c ^= (((b)<<(8)) | ((b)>>(32 - (8)))); b += a; a -= c; a ^= (((c)<<(16)) | ((c)>> (32 -(16)))); c += b; b -= a; b ^= (((a)<<(19)) | ((a)>> (32 -(19)))); a += c; c -= b; c ^= (((b)<<(4)) | ((b)>> (32 -(4)))); b += a; } \ | |||
358 | { \ | |||
359 | a -= c; a ^= rot(c, 4)(((c)<<(4)) | ((c)>>(32 -(4)))); c += b; \ | |||
360 | b -= a; b ^= rot(a, 6)(((a)<<(6)) | ((a)>>(32 -(6)))); a += c; \ | |||
361 | c -= b; c ^= rot(b, 8)(((b)<<(8)) | ((b)>>(32 -(8)))); b += a; \ | |||
362 | a -= c; a ^= rot(c,16)(((c)<<(16)) | ((c)>>(32 -(16)))); c += b; \ | |||
363 | b -= a; b ^= rot(a,19)(((a)<<(19)) | ((a)>>(32 -(19)))); a += c; \ | |||
364 | c -= b; c ^= rot(b, 4)(((b)<<(4)) | ((b)>>(32 -(4)))); b += a; \ | |||
365 | } | |||
366 | #define final(a,b,c){ c ^= b; c -= (((b)<<(14)) | ((b)>>(32 -(14)))); a ^= c; a -= (((c)<<(11)) | ((c)>>(32 -(11)))); b ^= a; b -= (((a)<<(25)) | ((a)>>(32 -(25)))); c ^= b; c -= (((b)<<(16)) | ((b)>>(32 -(16)))); a ^= c ; a -= (((c)<<(4)) | ((c)>>(32 -(4)))); b ^= a; b -= (((a)<<(14)) | ((a)>>(32 -(14)))); c ^= b; c -= (((b)<<(24)) | ((b)>>(32 -(24)))); } \ | |||
367 | { \ | |||
368 | c ^= b; c -= rot(b,14)(((b)<<(14)) | ((b)>>(32 -(14)))); \ | |||
369 | a ^= c; a -= rot(c,11)(((c)<<(11)) | ((c)>>(32 -(11)))); \ | |||
370 | b ^= a; b -= rot(a,25)(((a)<<(25)) | ((a)>>(32 -(25)))); \ | |||
371 | c ^= b; c -= rot(b,16)(((b)<<(16)) | ((b)>>(32 -(16)))); \ | |||
372 | a ^= c; a -= rot(c,4)(((c)<<(4)) | ((c)>>(32 -(4)))); \ | |||
373 | b ^= a; b -= rot(a,14)(((a)<<(14)) | ((a)>>(32 -(14)))); \ | |||
374 | c ^= b; c -= rot(b,24)(((b)<<(24)) | ((b)>>(32 -(24)))); \ | |||
375 | } | |||
376 | ||||
377 | static unsigned int hash_compute( struct hash* hash, const char* key, int length) | |||
378 | { | |||
379 | unsigned int a; | |||
380 | unsigned int b; | |||
381 | unsigned int c; /* internal state */ | |||
382 | const unsigned char* uk = (const unsigned char*)key; | |||
383 | ||||
384 | /* Set up the internal state */ | |||
385 | a = b = c = 0xdeadbeef + (length << 2); | |||
386 | ||||
387 | /* we use this to 'hash' full path with mostly a common root | |||
388 | * let's now waste too much cycles hashing mostly constant stuff | |||
389 | */ | |||
390 | if(length > 36) | |||
391 | { | |||
392 | uk += length - 36; | |||
393 | length = 36; | |||
394 | } | |||
395 | /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */ | |||
396 | while (length > 12) | |||
397 | { | |||
398 | a += get_unaligned_uint(uk)(*(unsigned int*)(uk)); | |||
399 | b += get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)); | |||
400 | c += get_unaligned_uint(uk+8)(*(unsigned int*)(uk+8)); | |||
401 | mix(a,b,c){ a -= c; a ^= (((c)<<(4)) | ((c)>>(32 -(4)))); c += b; b -= a; b ^= (((a)<<(6)) | ((a)>>(32 -(6)) )); a += c; c -= b; c ^= (((b)<<(8)) | ((b)>>(32 - (8)))); b += a; a -= c; a ^= (((c)<<(16)) | ((c)>> (32 -(16)))); c += b; b -= a; b ^= (((a)<<(19)) | ((a)>> (32 -(19)))); a += c; c -= b; c ^= (((b)<<(4)) | ((b)>> (32 -(4)))); b += a; }; | |||
402 | length -= 12; | |||
403 | uk += 12; | |||
404 | } | |||
405 | ||||
406 | /*----------------------------- handle the last (probably partial) block */ | |||
407 | /* Note: we possibly over-read, which would trigger complaint from VALGRIND | |||
408 | * but we mask the undefined stuff if any, so we are still good, thanks | |||
409 | * to alignment of memory allocation and tail-memory managment overhead | |||
410 | * we always can read 3 bytes past the official end without triggering | |||
411 | * a segfault -- if you find a platform/compiler couple for which that postulat | |||
412 | * is false, then you just need to over-allocate by 2 more bytes in file_load() | |||
413 | * file_load already over-allocate by 1 to sitck a \0 at the end of the buffer. | |||
414 | */ | |||
415 | switch(length) | |||
416 | { | |||
417 | case 12: c+=get_unaligned_uint(uk+8)(*(unsigned int*)(uk+8)); b+=get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)); a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
418 | case 11: c+=get_unaligned_uint(uk+8)(*(unsigned int*)(uk+8)) & MASK_C10xFFFFFF; b+=get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)); a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
419 | case 10: c+=get_unaligned_uint(uk+8)(*(unsigned int*)(uk+8)) & MASK_C20xFFFF; b+=get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)); a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
420 | case 9 : c+=get_unaligned_uint(uk+8)(*(unsigned int*)(uk+8)) & MASK_C30xFF; b+=get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)); a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
421 | case 8 : b+=get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)); a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
422 | case 7 : b+=get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)) & MASK_C10xFFFFFF; a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
423 | case 6 : b+=get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)) & MASK_C20xFFFF; a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
424 | case 5 : b+=get_unaligned_uint(uk+4)(*(unsigned int*)(uk+4)) & MASK_C30xFF; a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
425 | case 4 : a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)); break; | |||
426 | case 3 : a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)) & MASK_C10xFFFFFF; break; | |||
427 | case 2 : a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)) & MASK_C20xFFFF; break; | |||
428 | case 1 : a+=get_unaligned_uint(uk)(*(unsigned int*)(uk)) & MASK_C30xFF; break; | |||
429 | case 0 : return c & hash->size; /* zero length strings require no mixing */ | |||
430 | } | |||
431 | ||||
432 | final(a,b,c){ c ^= b; c -= (((b)<<(14)) | ((b)>>(32 -(14)))); a ^= c; a -= (((c)<<(11)) | ((c)>>(32 -(11)))); b ^= a; b -= (((a)<<(25)) | ((a)>>(32 -(25)))); c ^= b; c -= (((b)<<(16)) | ((b)>>(32 -(16)))); a ^= c ; a -= (((c)<<(4)) | ((c)>>(32 -(4)))); b ^= a; b -= (((a)<<(14)) | ((a)>>(32 -(14)))); c ^= b; c -= (((b)<<(24)) | ((b)>>(32 -(24)))); }; | |||
433 | return c & hash->size; | |||
434 | } | |||
435 | ||||
436 | static void hash_destroy(struct hash* hash) | |||
437 | { | |||
438 | if(hash) | |||
439 | { | |||
440 | if(hash->array) | |||
441 | { | |||
442 | free(hash->array); | |||
443 | } | |||
444 | if(hash->elems_pool) | |||
445 | { | |||
446 | pool_destroy(hash->elems_pool); | |||
447 | } | |||
448 | free(hash); | |||
449 | } | |||
450 | } | |||
451 | ||||
452 | static struct hash* hash_create(unsigned int size) | |||
453 | { | |||
454 | struct hash* hash; | |||
455 | ||||
456 | assert(size > 0)((size > 0) ? (void) (0) : __assert_fail ("size > 0", "/usr/local/src/libreoffice/solenv/bin/concat-deps.c" , 456, __PRETTY_FUNCTION__)); | |||
457 | hash = calloc(1, sizeof(struct hash)); | |||
458 | if(hash) | |||
459 | { | |||
460 | size += (size >> 2) + 1; /* ~ 75% load factor */ | |||
461 | if(size >= 15) | |||
462 | { | |||
463 | hash->size = (((unsigned int)0xFFFFFFFF) >> clz__builtin_clz((unsigned int)size)); | |||
464 | } | |||
465 | else | |||
466 | { | |||
467 | hash->size = size = 15; | |||
468 | } | |||
469 | hash->load_limit = hash->size - (hash->size >> 2); | |||
470 | hash->used = 0; | |||
471 | hash->array = (struct hash_elem**)calloc(hash->size + 1, sizeof(struct hash_elem*)); | |||
472 | if(hash->array == NULL((void*)0)) | |||
473 | { | |||
474 | hash_destroy(hash); | |||
475 | hash = NULL((void*)0); | |||
476 | } | |||
477 | } | |||
478 | if(hash) | |||
479 | { | |||
480 | hash->elems_pool = pool_create(sizeof(struct hash_elem), | |||
481 | 0, size, size << 1); | |||
482 | if(!hash->elems_pool) | |||
483 | { | |||
484 | hash_destroy(hash); | |||
485 | hash = NULL((void*)0); | |||
486 | } | |||
487 | } | |||
488 | return hash; | |||
489 | } | |||
490 | ||||
491 | static void hash_resize(struct hash* hash) | |||
492 | { | |||
493 | unsigned int old_size = hash->size; | |||
494 | unsigned int hashed; | |||
495 | struct hash_elem* hash_elem; | |||
496 | struct hash_elem* next; | |||
497 | struct hash_elem** array; | |||
498 | int i; | |||
499 | ||||
500 | hash->size = (old_size << 1) + 1; | |||
501 | /* we really should avoid to get there... so print a message to alert of the condition */ | |||
502 | fprintf(stderrstderr, "resize hash %d -> %d\n", old_size, hash->size); | |||
503 | if(hash->size == old_size) | |||
504 | { | |||
505 | hash->flags |= HASH_F_NO_RESIZE(1<<0); | |||
506 | return; | |||
507 | } | |||
508 | array = calloc(hash->size + 1, sizeof(struct hash_elem*)); | |||
509 | if(array) | |||
510 | { | |||
511 | hash->load_limit = hash->size - (hash->size >> 2); | |||
512 | for(i=0; i <= old_size; i++) | |||
513 | { | |||
514 | hash_elem = (struct hash_elem*)hash->array[i]; | |||
515 | while(hash_elem) | |||
516 | { | |||
517 | next = hash_elem->next; | |||
518 | ||||
519 | hashed = hash_compute(hash, hash_elem->key, hash_elem->key_len); | |||
520 | hash_elem->next = array[hashed]; | |||
521 | array[hashed] = hash_elem; | |||
522 | hash_elem = next; | |||
523 | } | |||
524 | } | |||
525 | free(hash->array); | |||
526 | hash->array = (struct hash_elem**)array; | |||
527 | } | |||
528 | else | |||
529 | { | |||
530 | hash->size = old_size; | |||
531 | hash->flags |= HASH_F_NO_RESIZE(1<<0); | |||
532 | } | |||
533 | } | |||
534 | ||||
535 | #ifdef HASH_STAT | |||
536 | static inline int compare_key(struct hash* hash, const char* a, const char* b, int len, int* cost)memcmp(const char* a,const char* b,int len) | |||
537 | { | |||
538 | *cost += 1; | |||
539 | hash->memcmp += 1; | |||
540 | return memcmp(a,b, len); | |||
541 | } | |||
542 | #else | |||
543 | #define compare_key(h,a,b,l,c)memcmp(a,b,l) memcmp(a,b,l) | |||
544 | #endif | |||
545 | ||||
546 | /* a customized hash_store function that just store the key and return | |||
547 | * TRUE if the key was effectively stored, or FALSE if the key was already there | |||
548 | */ | |||
549 | static int hash_store(struct hash* hash, const char* key, int key_len) | |||
550 | { | |||
551 | unsigned int hashed; | |||
552 | struct hash_elem* hash_elem; | |||
553 | int cost = 0; | |||
554 | ||||
555 | hashed = hash_compute(hash, key, key_len); | |||
556 | #ifdef HASH_STAT | |||
557 | hash->stored += 1; | |||
558 | #endif | |||
559 | hash_elem = (struct hash_elem*)hash->array[hashed]; | |||
560 | while(hash_elem && (hash_elem->key_len != key_len || compare_key(hash, hash_elem->key, key, key_len, &cost)memcmp(hash_elem->key,key,key_len))) | |||
561 | { | |||
562 | hash_elem = hash_elem->next; | |||
563 | } | |||
564 | ||||
565 | if(!hash_elem) | |||
566 | { | |||
567 | hash_elem = pool_alloc(hash->elems_pool); | |||
568 | if(hash_elem) | |||
569 | { | |||
570 | hash_elem->key = key; | |||
571 | hash_elem->key_len = key_len; | |||
572 | hash_elem->next = hash->array[hashed]; | |||
573 | ||||
574 | #ifdef HASH_STAT | |||
575 | if(hash_elem->next) | |||
576 | { | |||
577 | hash->collisions += 1; | |||
578 | hash->cost += cost; | |||
579 | } | |||
580 | #endif | |||
581 | hash->array[hashed] = hash_elem; | |||
582 | hash->used += 1; | |||
583 | if(hash->used > hash->load_limit) | |||
584 | { | |||
585 | hash_resize(hash); | |||
586 | } | |||
587 | } | |||
588 | return TRUE1; | |||
589 | } | |||
590 | return FALSE0; | |||
591 | } | |||
592 | ||||
593 | static int file_stat(const char* name, struct stat* buffer_stat, int* rc) | |||
594 | { | |||
595 | int rc_local = 0; | |||
596 | ||||
597 | rc_local = stat(name, buffer_stat); | |||
598 | if (rc_local < 0) | |||
599 | { | |||
600 | *rc = errno(*__errno_location ()); | |||
601 | } | |||
602 | return rc_local; | |||
603 | } | |||
604 | ||||
605 | static off_t file_get_size(const char* name, int* rc) | |||
606 | { | |||
607 | struct stat buffer_stat; | |||
608 | off_t size = -1; | |||
609 | ||||
610 | if (!file_stat(name, &buffer_stat, rc)) | |||
611 | { | |||
612 | if(S_ISREG(buffer_stat.st_mode)((((buffer_stat.st_mode)) & 0170000) == (0100000))) | |||
613 | { | |||
614 | size = buffer_stat.st_size; | |||
615 | } | |||
616 | else | |||
617 | { | |||
618 | *rc = EINVAL22; | |||
619 | } | |||
620 | } | |||
621 | return size; | |||
622 | } | |||
623 | ||||
624 | static char* file_load(const char* name, off_t* size, int* return_rc) | |||
625 | { | |||
626 | off_t local_size = 0; | |||
627 | int rc = 0; | |||
628 | char* buffer = NULL((void*)0); | |||
629 | int fd; | |||
630 | ||||
631 | assert(name != NULL)((name != ((void*)0)) ? (void) (0) : __assert_fail ("name != ((void*)0)" , "/usr/local/src/libreoffice/solenv/bin/concat-deps.c", 631, __PRETTY_FUNCTION__)); | |||
632 | ||||
633 | if(!size) | |||
634 | { | |||
635 | size = &local_size; | |||
636 | } | |||
637 | *size = file_get_size(name, &rc); | |||
638 | if (!rc) | |||
639 | { | |||
640 | fd = open(name, FILE_O_RDONLY00 | FILE_O_BINARY0); | |||
641 | if (!(fd == -1)) | |||
642 | { | |||
643 | buffer = malloc((size_t)(*size + 1)); | |||
644 | if (buffer == NULL((void*)0)) | |||
645 | { | |||
646 | rc = ENOMEM12; | |||
647 | } | |||
648 | else | |||
649 | { | |||
650 | ssize_t i; | |||
651 | ||||
652 | REDO: | |||
653 | i = read(fd, buffer, (size_t)(*size)); | |||
654 | if(i == -1) | |||
655 | { | |||
656 | if(errno(*__errno_location ()) == EINTR4) | |||
657 | { | |||
658 | goto REDO; | |||
659 | } | |||
660 | else | |||
661 | { | |||
662 | rc = errno(*__errno_location ()); | |||
663 | } | |||
664 | } | |||
665 | else | |||
666 | { | |||
667 | if (i != *size) | |||
668 | { | |||
669 | rc = EIO5; | |||
670 | } | |||
671 | } | |||
672 | close(fd); | |||
673 | buffer[*size] = 0; | |||
674 | } | |||
675 | } | |||
676 | } | |||
677 | ||||
678 | if(rc && buffer) | |||
679 | { | |||
680 | free(buffer); | |||
681 | buffer = NULL((void*)0); | |||
682 | } | |||
683 | if(return_rc) | |||
684 | { | |||
685 | *return_rc = rc; | |||
686 | } | |||
687 | return buffer; | |||
688 | } | |||
689 | ||||
690 | static void _cancel_relative(char* base, char** ref_cursor, char** ref_cursor_out, char* end) | |||
691 | { | |||
692 | char* cursor = *ref_cursor; | |||
693 | char* cursor_out = *ref_cursor_out; | |||
694 | ||||
695 | do | |||
696 | { | |||
697 | cursor += 3; | |||
698 | while(cursor_out > base && cursor_out[-1] == '/') | |||
699 | cursor_out--; | |||
700 | while(cursor_out > base && *--cursor_out != '/'); | |||
701 | } | |||
702 | while(cursor + 3 < end && !memcmp(cursor, "/../", 4)); | |||
703 | *ref_cursor = cursor; | |||
704 | *ref_cursor_out = cursor_out; | |||
705 | } | |||
706 | ||||
707 | static inline void eat_space(char ** token) | |||
708 | { | |||
709 | while ((' ' == **token) || ('\t' == **token)) { | |||
710 | ++(*token); | |||
711 | } | |||
712 | } | |||
713 | ||||
714 | /* prefix paths to absolute */ | |||
715 | static inline void print_fullpaths(char* line) | |||
716 | { | |||
717 | char* token; | |||
718 | char* end; | |||
719 | ||||
720 | token = line; | |||
721 | eat_space(&token); | |||
722 | while (*token) | |||
723 | { | |||
724 | end = token; | |||
725 | while (*end && (' ' != *end) && ('\t' != *end)) { | |||
726 | ++end; | |||
727 | } | |||
728 | if(*token == ':' || *token == '\\' || *token == '/' || *token == '$' | |||
729 | || ':' == token[1]) | |||
730 | { | |||
731 | fwrite(token, end - token, 1, stdoutstdout); | |||
732 | } | |||
733 | else | |||
734 | { | |||
735 | fputs(base_dir_var, stdoutstdout); | |||
736 | fputc('/', stdoutstdout); | |||
737 | fwrite(token, end - token, 1, stdoutstdout); | |||
738 | } | |||
739 | fputc(' ', stdoutstdout); | |||
740 | token = end; | |||
741 | eat_space(&token); | |||
742 | } | |||
743 | } | |||
744 | ||||
745 | static int _process(struct hash* dep_hash, char* fn) | |||
746 | { | |||
747 | int rc; | |||
748 | char* buffer; | |||
749 | char* end; | |||
750 | char* cursor; | |||
751 | char* cursor_out; | |||
752 | char* base; | |||
753 | int continuation = 0; | |||
754 | char last_ns = 0; | |||
755 | off_t size; | |||
756 | ||||
757 | buffer = file_load(fn, &size, &rc); | |||
758 | /* Note: yes we are going to leak 'buffer' | |||
759 | * this is on purpose, to avoid cloning the 'key' out of it | |||
760 | * and our special 'hash' just store the pointer to the key | |||
761 | * inside of buffer, hence it need to remain allocated | |||
762 | */ | |||
763 | if(!rc) | |||
764 | { | |||
765 | base = cursor_out = cursor = end = buffer; | |||
766 | end += size; | |||
767 | while(cursor < end) | |||
768 | { | |||
769 | if(*cursor == '\\') | |||
770 | { | |||
771 | continuation = 1; | |||
772 | *cursor_out++ = *cursor++; | |||
773 | } | |||
774 | else if(*cursor == '/') | |||
775 | { | |||
776 | if(cursor + 3 < end) | |||
777 | { | |||
778 | if(!memcmp(cursor, "/../", 4)) | |||
779 | { | |||
780 | _cancel_relative(base, &cursor, &cursor_out, end); | |||
781 | } | |||
782 | } | |||
783 | *cursor_out++ = *cursor++; | |||
784 | } | |||
785 | else if(*cursor == '\n') | |||
786 | { | |||
787 | if(!continuation) | |||
788 | { | |||
789 | *cursor_out = 0; | |||
790 | if(base < cursor) | |||
791 | { | |||
792 | /* here we have a complete rule */ | |||
793 | if(last_ns == ':') | |||
794 | { | |||
795 | /* if the rule ended in ':' that is a no-dep rule | |||
796 | * these are the one for which we want to filter | |||
797 | * duplicate out | |||
798 | */ | |||
799 | if(hash_store(dep_hash, base, (int)(cursor_out - base))) | |||
800 | { | |||
801 | /* DO NOT modify base after it has been added | |||
802 | as key by hash_store */ | |||
803 | print_fullpaths(base); | |||
804 | putc('\n', stdout)_IO_putc ('\n', stdout); | |||
805 | } | |||
806 | } | |||
807 | else | |||
808 | { | |||
809 | /* rule with dep, just write it */ | |||
810 | print_fullpaths(base); | |||
811 | putc('\n', stdout)_IO_putc ('\n', stdout); | |||
812 | } | |||
813 | } | |||
814 | cursor += 1; | |||
815 | base = cursor_out = cursor; | |||
816 | } | |||
817 | else | |||
818 | { | |||
819 | /* here we have a '\' followed by \n this is a continuation | |||
820 | * i.e not a complete rule yet | |||
821 | */ | |||
822 | *cursor_out++ = *cursor++; | |||
823 | } | |||
824 | } | |||
825 | else | |||
826 | { | |||
827 | continuation = 0; | |||
828 | /* not using isspace() here save 25% of I refs and 75% of D refs based on cachegrind */ | |||
829 | if(*cursor != ' ' && *cursor != '\n' && *cursor != '\t' ) | |||
830 | { | |||
831 | last_ns = *cursor; | |||
832 | } | |||
833 | *cursor_out++ = *cursor++; | |||
834 | } | |||
835 | } | |||
836 | /* just in case the file did not end with a \n, there may be a pending rule */ | |||
837 | if(base < cursor_out) | |||
838 | { | |||
839 | if(last_ns == ':') | |||
840 | { | |||
841 | if(hash_store(dep_hash, base, (int)(cursor_out - base))) | |||
842 | { | |||
843 | puts(base); | |||
844 | putc('\n', stdout)_IO_putc ('\n', stdout); | |||
845 | } | |||
846 | } | |||
847 | else | |||
848 | { | |||
849 | puts(base); | |||
850 | putc('\n', stdout)_IO_putc ('\n', stdout); | |||
851 | } | |||
852 | } | |||
853 | } | |||
854 | return rc; | |||
855 | } | |||
856 | ||||
857 | static void _usage(void) | |||
858 | { | |||
859 | fputs("Usage: concat-deps <file that contains dep_files>\n", stderrstderr); | |||
860 | } | |||
861 | ||||
862 | #define kDEFAULT_HASH_SIZE4096 4096 | |||
863 | ||||
864 | int main(int argc, char** argv) | |||
865 | { | |||
866 | int rc = 0; | |||
867 | off_t in_list_size = 0; | |||
868 | char* in_list; | |||
869 | char* in_list_cursor; | |||
870 | char* in_list_base; | |||
871 | struct hash* dep_hash; | |||
872 | ||||
873 | if(argc < 2) | |||
| ||||
874 | { | |||
875 | _usage(); | |||
876 | return 1; | |||
877 | } | |||
878 | base_dir = getenv("SRCDIR"); | |||
879 | if(!base_dir) | |||
880 | { | |||
881 | fputs("Error: SRCDIR is missing in the environement\n", stderrstderr); | |||
882 | return 1; | |||
883 | } | |||
884 | ||||
885 | in_list = file_load(argv[1], &in_list_size, &rc); | |||
886 | if(!rc) | |||
887 | { | |||
888 | dep_hash = hash_create( kDEFAULT_HASH_SIZE4096); | |||
889 | in_list_base = in_list_cursor = in_list; | |||
890 | ||||
891 | /* extract filename of dep file from a 'space' separated list */ | |||
892 | while(*in_list_cursor) | |||
893 | { | |||
894 | if(*in_list_cursor == ' ' || *in_list_cursor == '\n') | |||
895 | { | |||
896 | *in_list_cursor = 0; | |||
897 | if(in_list_base < in_list_cursor) | |||
898 | { | |||
899 | rc = _process(dep_hash, in_list_base); | |||
900 | if(rc) | |||
901 | { | |||
902 | break; | |||
903 | } | |||
904 | } | |||
905 | in_list_cursor += 1; | |||
906 | in_list_base = in_list_cursor; | |||
907 | } | |||
908 | else | |||
909 | { | |||
910 | in_list_cursor += 1; | |||
911 | } | |||
912 | } | |||
913 | if(!rc) | |||
914 | { | |||
915 | /* catch the last entry in case the input did not terminate with a 'space' */ | |||
916 | if(in_list_base < in_list_cursor) | |||
917 | { | |||
918 | rc = _process(dep_hash, in_list_base); | |||
919 | } | |||
920 | } | |||
921 | #ifdef HASH_STAT | |||
922 | fprintf(stderrstderr, "stats: u:%d s:%d l:%d t:%d c:%d m:%d $:%d\n", | |||
923 | dep_hash->used, dep_hash->size, dep_hash->load_limit, dep_hash->stored, | |||
924 | dep_hash->collisions, dep_hash->memcmp, dep_hash->cost); | |||
925 | #endif | |||
926 | } | |||
927 | return rc; | |||
| ||||
928 | } | |||
929 | ||||
930 | /* vim:set shiftwidth=4 softtabstop=4 expandtab: */ |