xref: /PHP-7.4/Zend/zend_string.c (revision 3a04adce)
1 /*
2    +----------------------------------------------------------------------+
3    | Zend Engine                                                          |
4    +----------------------------------------------------------------------+
5    | Copyright (c) Zend Technologies Ltd. (http://www.zend.com)           |
6    +----------------------------------------------------------------------+
7    | This source file is subject to version 2.00 of the Zend license,     |
8    | that is bundled with this package in the file LICENSE, and is        |
9    | available through the world-wide-web at the following url:           |
10    | http://www.zend.com/license/2_00.txt.                                |
11    | If you did not receive a copy of the Zend license and are unable to  |
12    | obtain it through the world-wide-web, please send a note to          |
13    | license@zend.com so we can mail you a copy immediately.              |
14    +----------------------------------------------------------------------+
15    | Authors: Dmitry Stogov <dmitry@php.net>                              |
16    +----------------------------------------------------------------------+
17 */
18 
19 #include "zend.h"
20 #include "zend_globals.h"
21 
22 #ifdef HAVE_VALGRIND
23 # include "valgrind/callgrind.h"
24 #endif
25 
26 ZEND_API zend_new_interned_string_func_t zend_new_interned_string;
27 ZEND_API zend_string_init_interned_func_t zend_string_init_interned;
28 
29 static zend_string* ZEND_FASTCALL zend_new_interned_string_permanent(zend_string *str);
30 static zend_string* ZEND_FASTCALL zend_new_interned_string_request(zend_string *str);
31 static zend_string* ZEND_FASTCALL zend_string_init_interned_permanent(const char *str, size_t size, int permanent);
32 static zend_string* ZEND_FASTCALL zend_string_init_interned_request(const char *str, size_t size, int permanent);
33 
34 /* Any strings interned in the startup phase. Common to all the threads,
35    won't be free'd until process exit. If we want an ability to
36    add permanent strings even after startup, it would be still
37    possible on costs of locking in the thread safe builds. */
38 static HashTable interned_strings_permanent;
39 
40 static zend_new_interned_string_func_t interned_string_request_handler = zend_new_interned_string_request;
41 static zend_string_init_interned_func_t interned_string_init_request_handler = zend_string_init_interned_request;
42 
43 ZEND_API zend_string  *zend_empty_string = NULL;
44 ZEND_API zend_string  *zend_one_char_string[256];
45 ZEND_API zend_string **zend_known_strings = NULL;
46 
zend_string_hash_func(zend_string * str)47 ZEND_API zend_ulong ZEND_FASTCALL zend_string_hash_func(zend_string *str)
48 {
49 	return ZSTR_H(str) = zend_hash_func(ZSTR_VAL(str), ZSTR_LEN(str));
50 }
51 
zend_hash_func(const char * str,size_t len)52 ZEND_API zend_ulong ZEND_FASTCALL zend_hash_func(const char *str, size_t len)
53 {
54 	return zend_inline_hash_func(str, len);
55 }
56 
_str_dtor(zval * zv)57 static void _str_dtor(zval *zv)
58 {
59 	zend_string *str = Z_STR_P(zv);
60 	pefree(str, GC_FLAGS(str) & IS_STR_PERSISTENT);
61 }
62 
63 static const char *known_strings[] = {
64 #define _ZEND_STR_DSC(id, str) str,
65 ZEND_KNOWN_STRINGS(_ZEND_STR_DSC)
66 #undef _ZEND_STR_DSC
67 	NULL
68 };
69 
zend_init_interned_strings_ht(HashTable * interned_strings,int permanent)70 static zend_always_inline void zend_init_interned_strings_ht(HashTable *interned_strings, int permanent)
71 {
72 	zend_hash_init(interned_strings, 1024, NULL, _str_dtor, permanent);
73 	if (permanent) {
74 		zend_hash_real_init_mixed(interned_strings);
75 	}
76 }
77 
zend_interned_strings_init(void)78 ZEND_API void zend_interned_strings_init(void)
79 {
80 	char s[2];
81 	unsigned int i;
82 	zend_string *str;
83 
84 	interned_string_request_handler = zend_new_interned_string_request;
85 	interned_string_init_request_handler = zend_string_init_interned_request;
86 
87 	zend_empty_string = NULL;
88 	zend_known_strings = NULL;
89 
90 	zend_init_interned_strings_ht(&interned_strings_permanent, 1);
91 
92 	zend_new_interned_string = zend_new_interned_string_permanent;
93 	zend_string_init_interned = zend_string_init_interned_permanent;
94 
95 	/* interned empty string */
96 	str = zend_string_alloc(sizeof("")-1, 1);
97 	ZSTR_VAL(str)[0] = '\000';
98 	zend_empty_string = zend_new_interned_string_permanent(str);
99 
100 	s[1] = 0;
101 	for (i = 0; i < 256; i++) {
102 		s[0] = i;
103 		zend_one_char_string[i] = zend_new_interned_string_permanent(zend_string_init(s, 1, 1));
104 	}
105 
106 	/* known strings */
107 	zend_known_strings = pemalloc(sizeof(zend_string*) * ((sizeof(known_strings) / sizeof(known_strings[0]) - 1)), 1);
108 	for (i = 0; i < (sizeof(known_strings) / sizeof(known_strings[0])) - 1; i++) {
109 		str = zend_string_init(known_strings[i], strlen(known_strings[i]), 1);
110 		zend_known_strings[i] = zend_new_interned_string_permanent(str);
111 	}
112 }
113 
zend_interned_strings_dtor(void)114 ZEND_API void zend_interned_strings_dtor(void)
115 {
116 	zend_hash_destroy(&interned_strings_permanent);
117 
118 	free(zend_known_strings);
119 	zend_known_strings = NULL;
120 }
121 
zend_interned_string_ht_lookup_ex(zend_ulong h,const char * str,size_t size,HashTable * interned_strings)122 static zend_always_inline zend_string *zend_interned_string_ht_lookup_ex(zend_ulong h, const char *str, size_t size, HashTable *interned_strings)
123 {
124 	uint32_t nIndex;
125 	uint32_t idx;
126 	Bucket *p;
127 
128 	nIndex = h | interned_strings->nTableMask;
129 	idx = HT_HASH(interned_strings, nIndex);
130 	while (idx != HT_INVALID_IDX) {
131 		p = HT_HASH_TO_BUCKET(interned_strings, idx);
132 		if ((p->h == h) && (ZSTR_LEN(p->key) == size)) {
133 			if (!memcmp(ZSTR_VAL(p->key), str, size)) {
134 				return p->key;
135 			}
136 		}
137 		idx = Z_NEXT(p->val);
138 	}
139 
140 	return NULL;
141 }
142 
zend_interned_string_ht_lookup(zend_string * str,HashTable * interned_strings)143 static zend_always_inline zend_string *zend_interned_string_ht_lookup(zend_string *str, HashTable *interned_strings)
144 {
145 	zend_ulong h = ZSTR_H(str);
146 	uint32_t nIndex;
147 	uint32_t idx;
148 	Bucket *p;
149 
150 	nIndex = h | interned_strings->nTableMask;
151 	idx = HT_HASH(interned_strings, nIndex);
152 	while (idx != HT_INVALID_IDX) {
153 		p = HT_HASH_TO_BUCKET(interned_strings, idx);
154 		if ((p->h == h) && zend_string_equal_content(p->key, str)) {
155 			return p->key;
156 		}
157 		idx = Z_NEXT(p->val);
158 	}
159 
160 	return NULL;
161 }
162 
163 /* This function might be not thread safe at least because it would update the
164    hash val in the passed string. Be sure it is called in the appropriate context. */
zend_add_interned_string(zend_string * str,HashTable * interned_strings,uint32_t flags)165 static zend_always_inline zend_string *zend_add_interned_string(zend_string *str, HashTable *interned_strings, uint32_t flags)
166 {
167 	zval val;
168 
169 	GC_SET_REFCOUNT(str, 1);
170 	GC_ADD_FLAGS(str, IS_STR_INTERNED | flags);
171 
172 	ZVAL_INTERNED_STR(&val, str);
173 
174 	zend_hash_add_new(interned_strings, str, &val);
175 
176 	return str;
177 }
178 
zend_interned_string_find_permanent(zend_string * str)179 ZEND_API zend_string* ZEND_FASTCALL zend_interned_string_find_permanent(zend_string *str)
180 {
181 	zend_string_hash_val(str);
182 	return zend_interned_string_ht_lookup(str, &interned_strings_permanent);
183 }
184 
zend_new_interned_string_permanent(zend_string * str)185 static zend_string* ZEND_FASTCALL zend_new_interned_string_permanent(zend_string *str)
186 {
187 	zend_string *ret;
188 
189 	if (ZSTR_IS_INTERNED(str)) {
190 		return str;
191 	}
192 
193 	zend_string_hash_val(str);
194 	ret = zend_interned_string_ht_lookup(str, &interned_strings_permanent);
195 	if (ret) {
196 		zend_string_release(str);
197 		return ret;
198 	}
199 
200 	ZEND_ASSERT(GC_FLAGS(str) & GC_PERSISTENT);
201 	if (GC_REFCOUNT(str) > 1) {
202 		zend_ulong h = ZSTR_H(str);
203 		zend_string_delref(str);
204 		str = zend_string_init(ZSTR_VAL(str), ZSTR_LEN(str), 1);
205 		ZSTR_H(str) = h;
206 	}
207 
208 	return zend_add_interned_string(str, &interned_strings_permanent, IS_STR_PERMANENT);
209 }
210 
zend_new_interned_string_request(zend_string * str)211 static zend_string* ZEND_FASTCALL zend_new_interned_string_request(zend_string *str)
212 {
213 	zend_string *ret;
214 
215 	if (ZSTR_IS_INTERNED(str)) {
216 		return str;
217 	}
218 
219 	zend_string_hash_val(str);
220 
221 	/* Check for permanent strings, the table is readonly at this point. */
222 	ret = zend_interned_string_ht_lookup(str, &interned_strings_permanent);
223 	if (ret) {
224 		zend_string_release(str);
225 		return ret;
226 	}
227 
228 	ret = zend_interned_string_ht_lookup(str, &CG(interned_strings));
229 	if (ret) {
230 		zend_string_release(str);
231 		return ret;
232 	}
233 
234 	/* Create a short living interned, freed after the request. */
235 #if ZEND_RC_DEBUG
236 	if (zend_rc_debug) {
237 		/* PHP shouldn't create persistent interned string during request,
238 		 * but at least dl() may do this */
239 		ZEND_ASSERT(!(GC_FLAGS(str) & GC_PERSISTENT));
240 	}
241 #endif
242 	if (GC_REFCOUNT(str) > 1) {
243 		zend_ulong h = ZSTR_H(str);
244 		zend_string_delref(str);
245 		str = zend_string_init(ZSTR_VAL(str), ZSTR_LEN(str), 0);
246 		ZSTR_H(str) = h;
247 	}
248 
249 	ret = zend_add_interned_string(str, &CG(interned_strings), 0);
250 
251 	return ret;
252 }
253 
zend_string_init_interned_permanent(const char * str,size_t size,int permanent)254 static zend_string* ZEND_FASTCALL zend_string_init_interned_permanent(const char *str, size_t size, int permanent)
255 {
256 	zend_string *ret;
257 	zend_ulong h = zend_inline_hash_func(str, size);
258 
259 	ret = zend_interned_string_ht_lookup_ex(h, str, size, &interned_strings_permanent);
260 	if (ret) {
261 		return ret;
262 	}
263 
264 	ZEND_ASSERT(permanent);
265 	ret = zend_string_init(str, size, permanent);
266 	ZSTR_H(ret) = h;
267 	return zend_add_interned_string(ret, &interned_strings_permanent, IS_STR_PERMANENT);
268 }
269 
zend_string_init_interned_request(const char * str,size_t size,int permanent)270 static zend_string* ZEND_FASTCALL zend_string_init_interned_request(const char *str, size_t size, int permanent)
271 {
272 	zend_string *ret;
273 	zend_ulong h = zend_inline_hash_func(str, size);
274 
275 	/* Check for permanent strings, the table is readonly at this point. */
276 	ret = zend_interned_string_ht_lookup_ex(h, str, size, &interned_strings_permanent);
277 	if (ret) {
278 		return ret;
279 	}
280 
281 	ret = zend_interned_string_ht_lookup_ex(h, str, size, &CG(interned_strings));
282 	if (ret) {
283 		return ret;
284 	}
285 
286 #if ZEND_RC_DEBUG
287 	if (zend_rc_debug) {
288 		/* PHP shouldn't create persistent interned string during request,
289 		 * but at least dl() may do this */
290 		ZEND_ASSERT(!permanent);
291 	}
292 #endif
293 	ret = zend_string_init(str, size, permanent);
294 	ZSTR_H(ret) = h;
295 
296 	/* Create a short living interned, freed after the request. */
297 	return zend_add_interned_string(ret, &CG(interned_strings), 0);
298 }
299 
zend_interned_strings_activate(void)300 ZEND_API void zend_interned_strings_activate(void)
301 {
302 	zend_init_interned_strings_ht(&CG(interned_strings), 0);
303 }
304 
zend_interned_strings_deactivate(void)305 ZEND_API void zend_interned_strings_deactivate(void)
306 {
307 	zend_hash_destroy(&CG(interned_strings));
308 }
309 
zend_interned_strings_set_request_storage_handlers(zend_new_interned_string_func_t handler,zend_string_init_interned_func_t init_handler)310 ZEND_API void zend_interned_strings_set_request_storage_handlers(zend_new_interned_string_func_t handler, zend_string_init_interned_func_t init_handler)
311 {
312 	interned_string_request_handler = handler;
313 	interned_string_init_request_handler = init_handler;
314 }
315 
zend_interned_strings_switch_storage(zend_bool request)316 ZEND_API void zend_interned_strings_switch_storage(zend_bool request)
317 {
318 	if (request) {
319 		zend_new_interned_string = interned_string_request_handler;
320 		zend_string_init_interned = interned_string_init_request_handler;
321 	} else {
322 		zend_new_interned_string = zend_new_interned_string_permanent;
323 		zend_string_init_interned = zend_string_init_interned_permanent;
324 	}
325 }
326 
327 #if defined(__GNUC__) && defined(__i386__)
zend_string_equal_val(zend_string * s1,zend_string * s2)328 ZEND_API zend_bool ZEND_FASTCALL zend_string_equal_val(zend_string *s1, zend_string *s2)
329 {
330 	char *ptr = ZSTR_VAL(s1);
331 	size_t delta = (char*)s2 - (char*)s1;
332 	size_t len = ZSTR_LEN(s1);
333 	zend_ulong ret;
334 
335 	__asm__ (
336 		".LL0%=:\n\t"
337 		"movl (%2,%3), %0\n\t"
338 		"xorl (%2), %0\n\t"
339 		"jne .LL1%=\n\t"
340 		"addl $0x4, %2\n\t"
341 		"subl $0x4, %1\n\t"
342 		"ja .LL0%=\n\t"
343 		"movl $0x1, %0\n\t"
344 		"jmp .LL3%=\n\t"
345 		".LL1%=:\n\t"
346 		"cmpl $0x4,%1\n\t"
347 		"jb .LL2%=\n\t"
348 		"xorl %0, %0\n\t"
349 		"jmp .LL3%=\n\t"
350 		".LL2%=:\n\t"
351 		"negl %1\n\t"
352 		"lea 0x20(,%1,8), %1\n\t"
353 		"shll %b1, %0\n\t"
354 		"sete %b0\n\t"
355 		"movzbl %b0, %0\n\t"
356 		".LL3%=:\n"
357 		: "=&a"(ret),
358 		  "+c"(len),
359 		  "+r"(ptr)
360 		: "r"(delta)
361 		: "cc");
362 	return ret;
363 }
364 
365 #ifdef HAVE_VALGRIND
I_WRAP_SONAME_FNNAME_ZU(NONE,zend_string_equal_val)366 ZEND_API zend_bool ZEND_FASTCALL I_WRAP_SONAME_FNNAME_ZU(NONE,zend_string_equal_val)(zend_string *s1, zend_string *s2)
367 {
368 	size_t len = ZSTR_LEN(s1);
369 	char *ptr1 = ZSTR_VAL(s1);
370 	char *ptr2 = ZSTR_VAL(s2);
371 	zend_ulong ret;
372 
373 	__asm__ (
374 		"test %1, %1\n\t"
375 		"jnz .LL1%=\n\t"
376 		"movl $0x1, %0\n\t"
377 		"jmp .LL2%=\n\t"
378 		".LL1%=:\n\t"
379 		"cld\n\t"
380 		"rep\n\t"
381 		"cmpsb\n\t"
382 		"sete %b0\n\t"
383 		"movzbl %b0, %0\n\t"
384 		".LL2%=:\n"
385 		: "=a"(ret),
386 		  "+c"(len),
387 		  "+D"(ptr1),
388 		  "+S"(ptr2)
389 		:
390 		: "cc");
391 	return ret;
392 }
393 #endif
394 
395 #elif defined(__GNUC__) && defined(__x86_64__) && !defined(__ILP32__)
zend_string_equal_val(zend_string * s1,zend_string * s2)396 ZEND_API zend_bool ZEND_FASTCALL zend_string_equal_val(zend_string *s1, zend_string *s2)
397 {
398 	char *ptr = ZSTR_VAL(s1);
399 	size_t delta = (char*)s2 - (char*)s1;
400 	size_t len = ZSTR_LEN(s1);
401 	zend_ulong ret;
402 
403 	__asm__ (
404 		".LL0%=:\n\t"
405 		"movq (%2,%3), %0\n\t"
406 		"xorq (%2), %0\n\t"
407 		"jne .LL1%=\n\t"
408 		"addq $0x8, %2\n\t"
409 		"subq $0x8, %1\n\t"
410 		"ja .LL0%=\n\t"
411 		"movq $0x1, %0\n\t"
412 		"jmp .LL3%=\n\t"
413 		".LL1%=:\n\t"
414 		"cmpq $0x8,%1\n\t"
415 		"jb .LL2%=\n\t"
416 		"xorq %0, %0\n\t"
417 		"jmp .LL3%=\n\t"
418 		".LL2%=:\n\t"
419 		"negq %1\n\t"
420 		"lea 0x40(,%1,8), %1\n\t"
421 		"shlq %b1, %0\n\t"
422 		"sete %b0\n\t"
423 		"movzbq %b0, %0\n\t"
424 		".LL3%=:\n"
425 		: "=&a"(ret),
426 		  "+c"(len),
427 		  "+r"(ptr)
428 		: "r"(delta)
429 		: "cc");
430 	return ret;
431 }
432 
433 #ifdef HAVE_VALGRIND
I_WRAP_SONAME_FNNAME_ZU(NONE,zend_string_equal_val)434 ZEND_API zend_bool ZEND_FASTCALL I_WRAP_SONAME_FNNAME_ZU(NONE,zend_string_equal_val)(zend_string *s1, zend_string *s2)
435 {
436 	size_t len = ZSTR_LEN(s1);
437 	char *ptr1 = ZSTR_VAL(s1);
438 	char *ptr2 = ZSTR_VAL(s2);
439 	zend_ulong ret;
440 
441 	__asm__ (
442 		"test %1, %1\n\t"
443 		"jnz .LL1%=\n\t"
444 		"movq $0x1, %0\n\t"
445 		"jmp .LL2%=\n\t"
446 		".LL1%=:\n\t"
447 		"cld\n\t"
448 		"rep\n\t"
449 		"cmpsb\n\t"
450 		"sete %b0\n\t"
451 		"movzbq %b0, %0\n\t"
452 		".LL2%=:\n"
453 		: "=a"(ret),
454 		  "+c"(len),
455 		  "+D"(ptr1),
456 		  "+S"(ptr2)
457 		:
458 		: "cc");
459 	return ret;
460 }
461 #endif
462 
463 #endif
464