1 /*
2 * Copyright (c) 1997, 2019, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation.
8 *
9 * This code is distributed in the hope that it will be useful, but WITHOUT
10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 * version 2 for more details (a copy is included in the LICENSE file that
13 * accompanied this code).
14 *
15 * You should have received a copy of the GNU General Public License version
16 * 2 along with this work; if not, write to the Free Software Foundation,
17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 * or visit www.oracle.com if you need additional information or have any
21 * questions.
22 *
23 */
24
25
26 #include "precompiled.hpp"
27 #include "classfile/altHashing.hpp"
28 #include "classfile/classLoaderData.hpp"
29 #include "gc/shared/collectedHeap.hpp"
30 #include "logging/log.hpp"
31 #include "logging/logStream.hpp"
32 #include "memory/allocation.inline.hpp"
33 #include "memory/resourceArea.hpp"
34 #include "oops/symbol.hpp"
35 #include "runtime/atomic.hpp"
36 #include "runtime/os.hpp"
37 #include "utilities/utf8.hpp"
38
39 uint32_t Symbol::pack_length_and_refcount(int length, int refcount) {
40 STATIC_ASSERT(max_symbol_length == ((1 << 16) - 1));
41 STATIC_ASSERT(PERM_REFCOUNT == ((1 << 16) - 1));
42 assert(length >= 0, "negative length");
43 assert(length <= max_symbol_length, "too long symbol");
44 assert(refcount >= 0, "negative refcount");
45 assert(refcount <= PERM_REFCOUNT, "invalid refcount");
46 uint32_t hi = length;
47 uint32_t lo = refcount;
48 return (hi << 16) | lo;
49 }
50
51 Symbol::Symbol(const u1* name, int length, int refcount) {
52 _length_and_refcount = pack_length_and_refcount(length, refcount);
53 _identity_hash = (short)os::random();
54 for (int i = 0; i < length; i++) {
55 byte_at_put(i, name[i]);
56 }
57 }
58
59 void* Symbol::operator new(size_t sz, int len, TRAPS) throw() {
60 int alloc_size = size(len)*wordSize;
61 address res = (address) AllocateHeap(alloc_size, mtSymbol);
62 return res;
63 }
64
65 void* Symbol::operator new(size_t sz, int len, Arena* arena, TRAPS) throw() {
66 int alloc_size = size(len)*wordSize;
67 address res = (address)arena->Amalloc_4(alloc_size);
68 return res;
69 }
70
71 void Symbol::operator delete(void *p) {
72 assert(((Symbol*)p)->refcount() == 0, "should not call this");
73 FreeHeap(p);
74 }
75
76 // ------------------------------------------------------------------
77 // Symbol::contains_byte_at
78 //
79 // Tests if the symbol contains the given byte at the given position.
80 bool Symbol::contains_byte_at(int position, char code_byte) const {
81 if (position < 0) return false; // can happen with ends_with
82 if (position >= utf8_length()) return false;
83 return code_byte == char_at(position);
84 }
85
86 // ------------------------------------------------------------------
87 // Symbol::contains_utf8_at
88 //
89 // Tests if the symbol contains the given utf8 substring
90 // at the given byte position.
91 bool Symbol::contains_utf8_at(int position, const char* substring, int len) const {
92 assert(len > 0 && substring != NULL && (int) strlen(substring) >= len,
93 "substring must be valid");
94 if (len == 1) return contains_byte_at(position, substring[0]);
95 if (position < 0) return false; // can happen with ends_with
96 if (position + len > utf8_length()) return false;
97 while (len-- > 0) {
98 if (substring[len] != char_at(position + len))
99 return false;
100 }
101 assert(len == -1, "we should be at the beginning");
102 return true;
103 }
104
105 bool Symbol::is_Q_signature() const {
106 return utf8_length() > 2 && char_at(0) == 'Q' && ends_with(';');
107 }
108
109 Symbol* Symbol::fundamental_name(TRAPS) {
110 if ((char_at(0) == 'Q' || char_at(0) == 'L') && ends_with(';')) {
111 return SymbolTable::lookup(this, 1, utf8_length() - 1, CHECK_NULL);
112 } else {
113 // reference count is incremented to be consistent with the behavior with
114 // the SymbolTable::lookup() call above
115 this->increment_refcount();
116 return this;
117 }
118 }
119
120 bool Symbol::is_same_fundamental_type(Symbol* s) const {
121 if (this == s) return true;
122 if (utf8_length() < 3) return false;
123 int offset1, offset2, len;
124 if (ends_with(';')) {
125 if (char_at(0) != 'Q' && char_at(0) != 'L') return false;
126 offset1 = 1;
127 len = utf8_length() - 2;
128 } else {
129 offset1 = 0;
130 len = utf8_length();
131 }
132 if (ends_with(';')) {
133 if (s->char_at(0) != 'Q' && s->char_at(0) != 'L') return false;
134 offset2 = 1;
135 } else {
136 offset2 = 0;
137 }
138 if ((offset2 + len) > s->utf8_length()) return false;
139 if ((utf8_length() - offset1 * 2) != (s->utf8_length() - offset2 * 2))
140 return false;
141 int l = len;
142 while (l-- > 0) {
143 if (char_at(offset1 + l) != s->char_at(offset2 + l))
144 return false;
145 }
146 return true;
147 }
148
149 // ------------------------------------------------------------------
150 // Symbol::index_of
151 //
152 // Finds if the given string is a substring of this symbol's utf8 bytes.
153 // Return -1 on failure. Otherwise return the first index where str occurs.
154 int Symbol::index_of_at(int i, const char* str, int len) const {
155 assert(i >= 0 && i <= utf8_length(), "oob");
156 if (len <= 0) return 0;
157 char first_char = str[0];
158 address bytes = (address) ((Symbol*)this)->base();
159 address limit = bytes + utf8_length() - len; // inclusive limit
160 address scan = bytes + i;
161 if (scan > limit)
162 return -1;
163 for (; scan <= limit; scan++) {
164 scan = (address) memchr(scan, first_char, (limit + 1 - scan));
165 if (scan == NULL)
166 return -1; // not found
167 assert(scan >= bytes+i && scan <= limit, "scan oob");
168 if (memcmp(scan, str, len) == 0)
169 return (int)(scan - bytes);
170 }
171 return -1;
172 }
173
174
175 char* Symbol::as_C_string(char* buf, int size) const {
176 if (size > 0) {
177 int len = MIN2(size - 1, utf8_length());
178 for (int i = 0; i < len; i++) {
179 buf[i] = char_at(i);
180 }
181 buf[len] = '\0';
182 }
183 return buf;
184 }
185
186 char* Symbol::as_C_string() const {
187 int len = utf8_length();
188 char* str = NEW_RESOURCE_ARRAY(char, len + 1);
189 return as_C_string(str, len + 1);
190 }
191
192 void Symbol::print_utf8_on(outputStream* st) const {
193 st->print("%s", as_C_string());
194 }
195
196 void Symbol::print_symbol_on(outputStream* st) const {
197 char *s;
198 st = st ? st : tty;
199 {
200 // ResourceMark may not affect st->print(). If st is a string
201 // stream it could resize, using the same resource arena.
202 ResourceMark rm;
203 s = as_quoted_ascii();
204 s = os::strdup(s);
205 }
206 if (s == NULL) {
207 st->print("(null)");
208 } else {
209 st->print("%s", s);
210 os::free(s);
211 }
212 }
213
214 char* Symbol::as_quoted_ascii() const {
215 const char *ptr = (const char *)&_body[0];
216 int quoted_length = UTF8::quoted_ascii_length(ptr, utf8_length());
217 char* result = NEW_RESOURCE_ARRAY(char, quoted_length + 1);
218 UTF8::as_quoted_ascii(ptr, utf8_length(), result, quoted_length + 1);
219 return result;
220 }
221
222 jchar* Symbol::as_unicode(int& length) const {
223 Symbol* this_ptr = (Symbol*)this;
224 length = UTF8::unicode_length((char*)this_ptr->bytes(), utf8_length());
225 jchar* result = NEW_RESOURCE_ARRAY(jchar, length);
226 if (length > 0) {
227 UTF8::convert_to_unicode((char*)this_ptr->bytes(), result, length);
228 }
229 return result;
230 }
231
232 const char* Symbol::as_klass_external_name(char* buf, int size) const {
233 if (size > 0) {
234 char* str = as_C_string(buf, size);
235 int length = (int)strlen(str);
236 // Turn all '/'s into '.'s (also for array klasses)
237 for (int index = 0; index < length; index++) {
238 if (str[index] == '/') {
239 str[index] = '.';
240 }
241 }
242 return str;
243 } else {
244 return buf;
245 }
246 }
247
248 const char* Symbol::as_klass_external_name() const {
249 char* str = as_C_string();
250 int length = (int)strlen(str);
251 // Turn all '/'s into '.'s (also for array klasses)
252 for (int index = 0; index < length; index++) {
253 if (str[index] == '/') {
254 str[index] = '.';
255 }
256 }
257 return str;
258 }
259
260 static void print_class(outputStream *os, char *class_str, int len) {
261 for (int i = 0; i < len; ++i) {
262 if (class_str[i] == '/') {
263 os->put('.');
264 } else {
265 os->put(class_str[i]);
266 }
267 }
268 }
269
270 static void print_array(outputStream *os, char *array_str, int len) {
271 int dimensions = 0;
272 for (int i = 0; i < len; ++i) {
273 if (array_str[i] == '[') {
274 dimensions++;
275 } else if (array_str[i] == 'L' || array_str[i] == 'Q') {
276 // Expected format: L<type name>;. Skip 'L' and ';' delimiting the type name.
277 print_class(os, array_str+i+1, len-i-2);
278 break;
279 } else {
280 os->print("%s", type2name(char2type(array_str[i])));
281 }
282 }
283 for (int i = 0; i < dimensions; ++i) {
284 os->print("[]");
285 }
286 }
287
288 void Symbol::print_as_signature_external_return_type(outputStream *os) {
289 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
290 if (ss.at_return_type()) {
291 if (ss.is_array()) {
292 print_array(os, (char*)ss.raw_bytes(), (int)ss.raw_length());
293 } else if (ss.is_object()) {
294 // Expected format: L<type name>;. Skip 'L' and ';' delimiting the class name.
295 print_class(os, (char*)ss.raw_bytes()+1, (int)ss.raw_length()-2);
296 } else {
297 os->print("%s", type2name(ss.type()));
298 }
299 }
300 }
301 }
302
303 void Symbol::print_as_signature_external_parameters(outputStream *os) {
304 bool first = true;
305 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
306 if (ss.at_return_type()) break;
307 if (!first) { os->print(", "); }
308 if (ss.is_array()) {
309 print_array(os, (char*)ss.raw_bytes(), (int)ss.raw_length());
310 } else if (ss.is_object()) {
311 // Skip 'L' and ';'.
312 print_class(os, (char*)ss.raw_bytes()+1, (int)ss.raw_length()-2);
313 } else {
314 os->print("%s", type2name(ss.type()));
315 }
316 first = false;
317 }
318 }
319
320 // Increment refcount while checking for zero. If the Symbol's refcount becomes zero
321 // a thread could be concurrently removing the Symbol. This is used during SymbolTable
322 // lookup to avoid reviving a dead Symbol.
323 bool Symbol::try_increment_refcount() {
324 uint32_t found = _length_and_refcount;
325 while (true) {
326 uint32_t old_value = found;
327 int refc = extract_refcount(old_value);
328 if (refc == PERM_REFCOUNT) {
329 return true; // sticky max or created permanent
330 } else if (refc == 0) {
331 return false; // dead, can't revive.
332 } else {
333 found = Atomic::cmpxchg(old_value + 1, &_length_and_refcount, old_value);
334 if (found == old_value) {
335 return true; // successfully updated.
336 }
337 // refcount changed, try again.
338 }
339 }
340 }
341
342 // The increment_refcount() is called when not doing lookup. It is assumed that you
343 // have a symbol with a non-zero refcount and it can't become zero while referenced by
344 // this caller.
345 void Symbol::increment_refcount() {
346 if (!try_increment_refcount()) {
347 #ifdef ASSERT
348 print();
349 fatal("refcount has gone to zero");
350 #endif
351 }
352 #ifndef PRODUCT
353 if (refcount() != PERM_REFCOUNT) { // not a permanent symbol
354 NOT_PRODUCT(Atomic::inc(&_total_count);)
355 }
356 #endif
357 }
358
359 // Decrement refcount potentially while racing increment, so we need
360 // to check the value after attempting to decrement so that if another
361 // thread increments to PERM_REFCOUNT the value is not decremented.
362 void Symbol::decrement_refcount() {
363 uint32_t found = _length_and_refcount;
364 while (true) {
365 uint32_t old_value = found;
366 int refc = extract_refcount(old_value);
367 if (refc == PERM_REFCOUNT) {
368 return; // refcount is permanent, permanent is sticky
369 } else if (refc == 0) {
370 #ifdef ASSERT
371 print();
372 fatal("refcount underflow");
373 #endif
374 return;
375 } else {
376 found = Atomic::cmpxchg(old_value - 1, &_length_and_refcount, old_value);
377 if (found == old_value) {
378 return; // successfully updated.
379 }
380 // refcount changed, try again.
381 }
382 }
383 }
384
385 void Symbol::make_permanent() {
386 uint32_t found = _length_and_refcount;
387 while (true) {
388 uint32_t old_value = found;
389 int refc = extract_refcount(old_value);
390 if (refc == PERM_REFCOUNT) {
391 return; // refcount is permanent, permanent is sticky
392 } else if (refc == 0) {
393 #ifdef ASSERT
394 print();
395 fatal("refcount underflow");
396 #endif
397 return;
398 } else {
399 int len = extract_length(old_value);
400 found = Atomic::cmpxchg(pack_length_and_refcount(len, PERM_REFCOUNT), &_length_and_refcount, old_value);
401 if (found == old_value) {
402 return; // successfully updated.
403 }
404 // refcount changed, try again.
405 }
406 }
407 }
408
409 void Symbol::metaspace_pointers_do(MetaspaceClosure* it) {
410 if (log_is_enabled(Trace, cds)) {
411 LogStream trace_stream(Log(cds)::trace());
412 trace_stream.print("Iter(Symbol): %p ", this);
413 print_value_on(&trace_stream);
414 trace_stream.cr();
415 }
416 }
417
418 void Symbol::print_on(outputStream* st) const {
419 st->print("Symbol: '");
420 print_symbol_on(st);
421 st->print("'");
422 st->print(" count %d", refcount());
423 }
424
425 // The print_value functions are present in all builds, to support the
426 // disassembler and error reporting.
427 void Symbol::print_value_on(outputStream* st) const {
428 st->print("'");
429 for (int i = 0; i < utf8_length(); i++) {
430 st->print("%c", char_at(i));
431 }
432 st->print("'");
433 }
434
435 bool Symbol::is_valid(Symbol* s) {
436 if (!is_aligned(s, sizeof(MetaWord))) return false;
437 if ((size_t)s < os::min_page_size()) return false;
438
439 if (!os::is_readable_range(s, s + 1)) return false;
440
441 // Symbols are not allocated in Java heap.
442 if (Universe::heap()->is_in_reserved(s)) return false;
443
444 int len = s->utf8_length();
445 if (len < 0) return false;
446
447 jbyte* bytes = (jbyte*) s->bytes();
448 return os::is_readable_range(bytes, bytes + len);
449 }
450
451 void Symbol::print_Qvalue_on(outputStream* st) const {
452 if (this == NULL) {
453 st->print("NULL");
454 } else {
455 st->print("'Q");
456 for (int i = 0; i < utf8_length(); i++) {
457 st->print("%c", char_at(i));
458 }
459 st->print(";'");
460 }
461 }
462
463 // SymbolTable prints this in its statistics
464 NOT_PRODUCT(size_t Symbol::_total_count = 0;)