1 /*
2 * Copyright (c) 1997, 2023, 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 "cds/metaspaceShared.hpp"
28 #include "classfile/altHashing.hpp"
29 #include "classfile/classLoaderData.hpp"
30 #include "classfile/vmSymbols.hpp"
31 #include "gc/shared/collectedHeap.hpp"
32 #include "logging/log.hpp"
33 #include "logging/logStream.hpp"
34 #include "memory/allocation.inline.hpp"
35 #include "memory/resourceArea.hpp"
36 #include "memory/universe.hpp"
37 #include "oops/symbol.hpp"
38 #include "runtime/atomic.hpp"
39 #include "runtime/mutexLocker.hpp"
40 #include "runtime/os.hpp"
41 #include "runtime/signature.hpp"
42 #include "utilities/stringUtils.hpp"
43 #include "utilities/utf8.hpp"
44
45 Symbol* Symbol::_vm_symbols[vmSymbols::number_of_symbols()];
46
47 uint32_t Symbol::pack_hash_and_refcount(short hash, int refcount) {
48 STATIC_ASSERT(PERM_REFCOUNT == ((1 << 16) - 1));
49 assert(refcount >= 0, "negative refcount");
50 assert(refcount <= PERM_REFCOUNT, "invalid refcount");
51 uint32_t hi = hash;
52 uint32_t lo = refcount;
53 return (hi << 16) | lo;
54 }
55
56 Symbol::Symbol(const u1* name, int length, int refcount) {
57 _hash_and_refcount = pack_hash_and_refcount((short)os::random(), refcount);
58 _length = (u2)length;
59 // _body[0..1] are allocated in the header just by coincidence in the current
60 // implementation of Symbol. They are read by identity_hash(), so make sure they
61 // are initialized.
62 // No other code should assume that _body[0..1] are always allocated. E.g., do
63 // not unconditionally read base()[0] as that will be invalid for an empty Symbol.
64 _body[0] = _body[1] = 0;
65 memcpy(_body, name, length);
66 }
67
68 // This copies the symbol when it is added to the ConcurrentHashTable.
69 Symbol::Symbol(const Symbol& s1) {
70 _hash_and_refcount = s1._hash_and_refcount;
71 _length = s1._length;
72 memcpy(_body, s1._body, _length);
73 }
74
75 #if INCLUDE_CDS
76 void Symbol::update_identity_hash() {
77 // This is called at a safepoint during dumping of a static CDS archive. The caller should have
78 // called os::init_random() with a deterministic seed and then iterate all archived Symbols in
79 // a deterministic order.
80 assert(SafepointSynchronize::is_at_safepoint(), "must be at a safepoint");
81 _hash_and_refcount = pack_hash_and_refcount((short)os::random(), PERM_REFCOUNT);
82 }
83
84 void Symbol::set_permanent() {
85 // This is called at a safepoint during dumping of a dynamic CDS archive.
86 assert(SafepointSynchronize::is_at_safepoint(), "must be at a safepoint");
87 _hash_and_refcount = pack_hash_and_refcount(extract_hash(_hash_and_refcount), PERM_REFCOUNT);
88 }
89 #endif
90
91 bool Symbol::is_Q_signature() const {
92 int len = utf8_length();
93 return len > 2 && char_at(0) == JVM_SIGNATURE_PRIMITIVE_OBJECT && char_at(len - 1) == JVM_SIGNATURE_ENDCLASS;
94 }
95
96 bool Symbol::is_Q_array_signature() const {
97 int l = utf8_length();
98 if (l < 2 || char_at(0) != JVM_SIGNATURE_ARRAY || char_at(l - 1) != JVM_SIGNATURE_ENDCLASS) {
99 return false;
100 }
101 for (int i = 1; i < (l - 2); i++) {
102 char c = char_at(i);
103 if (c == JVM_SIGNATURE_PRIMITIVE_OBJECT) {
104 return true;
105 }
106 if (c != JVM_SIGNATURE_ARRAY) {
107 return false;
108 }
109 }
110 return false;
111 }
112
113 bool Symbol::is_Q_method_signature() const {
114 assert(SignatureVerifier::is_valid_method_signature(this), "must be");
115 int len = utf8_length();
116 if (len > 4 && char_at(0) == JVM_SIGNATURE_FUNC) {
117 for (int i=1; i<len-3; i++) { // Must end with ")Qx;", where x is at least one character or more.
118 if (char_at(i) == JVM_SIGNATURE_ENDFUNC && char_at(i+1) == JVM_SIGNATURE_PRIMITIVE_OBJECT) {
119 return true;
120 }
121 }
122 }
123 return false;
124 }
125
126 Symbol* Symbol::fundamental_name(TRAPS) {
127 if ((char_at(0) == JVM_SIGNATURE_PRIMITIVE_OBJECT || char_at(0) == JVM_SIGNATURE_CLASS) && ends_with(JVM_SIGNATURE_ENDCLASS)) {
128 return SymbolTable::new_symbol(this, 1, utf8_length() - 1);
129 } else {
130 // reference count is incremented to be consistent with the behavior with
131 // the SymbolTable::new_symbol() call above
132 this->increment_refcount();
133 return this;
134 }
135 }
136
137 bool Symbol::is_same_fundamental_type(Symbol* s) const {
138 if (this == s) return true;
139 if (utf8_length() < 3) return false;
140 int offset1, offset2, len;
141 if (ends_with(JVM_SIGNATURE_ENDCLASS)) {
142 if (char_at(0) != JVM_SIGNATURE_PRIMITIVE_OBJECT && char_at(0) != JVM_SIGNATURE_CLASS) return false;
143 offset1 = 1;
144 len = utf8_length() - 2;
145 } else {
146 offset1 = 0;
147 len = utf8_length();
148 }
149 if (ends_with(JVM_SIGNATURE_ENDCLASS)) {
150 if (s->char_at(0) != JVM_SIGNATURE_PRIMITIVE_OBJECT && s->char_at(0) != JVM_SIGNATURE_CLASS) return false;
151 offset2 = 1;
152 } else {
153 offset2 = 0;
154 }
155 if ((offset2 + len) > s->utf8_length()) return false;
156 if ((utf8_length() - offset1 * 2) != (s->utf8_length() - offset2 * 2))
157 return false;
158 int l = len;
159 while (l-- > 0) {
160 if (char_at(offset1 + l) != s->char_at(offset2 + l))
161 return false;
162 }
163 return true;
164 }
165
166 // ------------------------------------------------------------------
167 // Symbol::index_of
168 //
169 // Test if we have the give substring at or after the i-th char of this
170 // symbol's utf8 bytes.
171 // Return -1 on failure. Otherwise return the first index where substr occurs.
172 int Symbol::index_of_at(int i, const char* substr, int substr_len) const {
173 assert(i >= 0 && i <= utf8_length(), "oob");
174 if (substr_len <= 0) return 0;
175 char first_char = substr[0];
176 address bytes = (address) ((Symbol*)this)->base();
177 address limit = bytes + utf8_length() - substr_len; // inclusive limit
178 address scan = bytes + i;
179 if (scan > limit)
180 return -1;
181 for (; scan <= limit; scan++) {
182 scan = (address) memchr(scan, first_char, (limit + 1 - scan));
183 if (scan == nullptr)
184 return -1; // not found
185 assert(scan >= bytes+i && scan <= limit, "scan oob");
186 if (substr_len <= 2
187 ? (char) scan[substr_len-1] == substr[substr_len-1]
188 : memcmp(scan+1, substr+1, substr_len-1) == 0) {
189 return (int)(scan - bytes);
190 }
191 }
192 return -1;
193 }
194
195 bool Symbol::is_star_match(const char* pattern) const {
196 if (strchr(pattern, '*') == nullptr) {
197 return equals(pattern);
198 } else {
199 ResourceMark rm;
200 char* buf = as_C_string();
201 return StringUtils::is_star_match(pattern, buf);
202 }
203 }
204
205 char* Symbol::as_C_string(char* buf, int size) const {
206 if (size > 0) {
207 int len = MIN2(size - 1, utf8_length());
208 for (int i = 0; i < len; i++) {
209 buf[i] = char_at(i);
210 }
211 buf[len] = '\0';
212 }
213 return buf;
214 }
215
216 char* Symbol::as_C_string() const {
217 int len = utf8_length();
218 char* str = NEW_RESOURCE_ARRAY(char, len + 1);
219 return as_C_string(str, len + 1);
220 }
221
222 void Symbol::print_utf8_on(outputStream* st) const {
223 st->print("%s", as_C_string());
224 }
225
226 void Symbol::print_symbol_on(outputStream* st) const {
227 char *s;
228 st = st ? st : tty;
229 {
230 // ResourceMark may not affect st->print(). If st is a string
231 // stream it could resize, using the same resource arena.
232 ResourceMark rm;
233 s = as_quoted_ascii();
234 s = os::strdup(s);
235 }
236 if (s == nullptr) {
237 st->print("(null)");
238 } else {
239 st->print("%s", s);
240 os::free(s);
241 }
242 }
243
244 char* Symbol::as_quoted_ascii() const {
245 const char *ptr = (const char *)&_body[0];
246 int quoted_length = UTF8::quoted_ascii_length(ptr, utf8_length());
247 char* result = NEW_RESOURCE_ARRAY(char, quoted_length + 1);
248 UTF8::as_quoted_ascii(ptr, utf8_length(), result, quoted_length + 1);
249 return result;
250 }
251
252 jchar* Symbol::as_unicode(int& length) const {
253 Symbol* this_ptr = (Symbol*)this;
254 length = UTF8::unicode_length((char*)this_ptr->bytes(), utf8_length());
255 jchar* result = NEW_RESOURCE_ARRAY(jchar, length);
256 if (length > 0) {
257 UTF8::convert_to_unicode((char*)this_ptr->bytes(), result, length);
258 }
259 return result;
260 }
261
262 const char* Symbol::as_klass_external_name(char* buf, int size) const {
263 if (size > 0) {
264 char* str = as_C_string(buf, size);
265 int length = (int)strlen(str);
266 // Turn all '/'s into '.'s (also for array klasses)
267 for (int index = 0; index < length; index++) {
268 if (str[index] == JVM_SIGNATURE_SLASH) {
269 str[index] = JVM_SIGNATURE_DOT;
270 }
271 }
272 return str;
273 } else {
274 return buf;
275 }
276 }
277
278 const char* Symbol::as_klass_external_name() const {
279 char* str = as_C_string();
280 int length = (int)strlen(str);
281 // Turn all '/'s into '.'s (also for array klasses)
282 for (int index = 0; index < length; index++) {
283 if (str[index] == JVM_SIGNATURE_SLASH) {
284 str[index] = JVM_SIGNATURE_DOT;
285 }
286 }
287 return str;
288 }
289
290 static void print_class(outputStream *os, const SignatureStream& ss) {
291 int sb = ss.raw_symbol_begin(), se = ss.raw_symbol_end();
292 for (int i = sb; i < se; ++i) {
293 char ch = ss.raw_char_at(i);
294 if (ch == JVM_SIGNATURE_SLASH) {
295 os->put(JVM_SIGNATURE_DOT);
296 } else {
297 os->put(ch);
298 }
299 }
300 }
301
302 static void print_array(outputStream *os, SignatureStream& ss) {
303 int dimensions = ss.skip_array_prefix();
304 assert(dimensions > 0, "");
305 if (ss.is_reference()) {
306 print_class(os, ss);
307 } else {
308 os->print("%s", type2name(ss.type()));
309 }
310 for (int i = 0; i < dimensions; ++i) {
311 os->print("[]");
312 }
313 }
314
315 void Symbol::print_as_signature_external_return_type(outputStream *os) {
316 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
317 if (ss.at_return_type()) {
318 if (ss.is_array()) {
319 print_array(os, ss);
320 } else if (ss.is_reference()) {
321 print_class(os, ss);
322 } else {
323 os->print("%s", type2name(ss.type()));
324 }
325 }
326 }
327 }
328
329 void Symbol::print_as_signature_external_parameters(outputStream *os) {
330 bool first = true;
331 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
332 if (ss.at_return_type()) break;
333 if (!first) { os->print(", "); }
334 if (ss.is_array()) {
335 print_array(os, ss);
336 } else if (ss.is_reference()) {
337 print_class(os, ss);
338 } else {
339 os->print("%s", type2name(ss.type()));
340 }
341 first = false;
342 }
343 }
344
345 void Symbol::print_as_field_external_type(outputStream *os) {
346 SignatureStream ss(this, false);
347 assert(!ss.is_done(), "must have at least one element in field ref");
348 assert(!ss.at_return_type(), "field ref cannot be a return type");
349 assert(!Signature::is_method(this), "field ref cannot be a method");
350
351 if (ss.is_array()) {
352 print_array(os, ss);
353 } else if (ss.is_reference()) {
354 print_class(os, ss);
355 } else {
356 os->print("%s", type2name(ss.type()));
357 }
358 #ifdef ASSERT
359 ss.next();
360 assert(ss.is_done(), "must have at most one element in field ref");
361 #endif
362 }
363
364 // Increment refcount while checking for zero. If the Symbol's refcount becomes zero
365 // a thread could be concurrently removing the Symbol. This is used during SymbolTable
366 // lookup to avoid reviving a dead Symbol.
367 bool Symbol::try_increment_refcount() {
368 uint32_t found = _hash_and_refcount;
369 while (true) {
370 uint32_t old_value = found;
371 int refc = extract_refcount(old_value);
372 if (refc == PERM_REFCOUNT) {
373 return true; // sticky max or created permanent
374 } else if (refc == 0) {
375 return false; // dead, can't revive.
376 } else {
377 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value + 1);
378 if (found == old_value) {
379 return true; // successfully updated.
380 }
381 // refcount changed, try again.
382 }
383 }
384 }
385
386 // The increment_refcount() is called when not doing lookup. It is assumed that you
387 // have a symbol with a non-zero refcount and it can't become zero while referenced by
388 // this caller.
389 void Symbol::increment_refcount() {
390 if (!try_increment_refcount()) {
391 print();
392 fatal("refcount has gone to zero");
393 }
394 #ifndef PRODUCT
395 if (refcount() != PERM_REFCOUNT) { // not a permanent symbol
396 NOT_PRODUCT(Atomic::inc(&_total_count);)
397 }
398 #endif
399 }
400
401 // Decrement refcount potentially while racing increment, so we need
402 // to check the value after attempting to decrement so that if another
403 // thread increments to PERM_REFCOUNT the value is not decremented.
404 void Symbol::decrement_refcount() {
405 uint32_t found = _hash_and_refcount;
406 while (true) {
407 uint32_t old_value = found;
408 int refc = extract_refcount(old_value);
409 if (refc == PERM_REFCOUNT) {
410 return; // refcount is permanent, permanent is sticky
411 } else if (refc == 0) {
412 print();
413 fatal("refcount underflow");
414 return;
415 } else {
416 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value - 1);
417 if (found == old_value) {
418 return; // successfully updated.
419 }
420 // refcount changed, try again.
421 }
422 }
423 }
424
425 void Symbol::make_permanent() {
426 uint32_t found = _hash_and_refcount;
427 while (true) {
428 uint32_t old_value = found;
429 int refc = extract_refcount(old_value);
430 if (refc == PERM_REFCOUNT) {
431 return; // refcount is permanent, permanent is sticky
432 } else if (refc == 0) {
433 print();
434 fatal("refcount underflow");
435 return;
436 } else {
437 short hash = extract_hash(old_value);
438 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, pack_hash_and_refcount(hash, PERM_REFCOUNT));
439 if (found == old_value) {
440 return; // successfully updated.
441 }
442 // refcount changed, try again.
443 }
444 }
445 }
446
447 void Symbol::metaspace_pointers_do(MetaspaceClosure* it) {
448 if (log_is_enabled(Trace, cds)) {
449 LogStream trace_stream(Log(cds)::trace());
450 trace_stream.print("Iter(Symbol): %p ", this);
451 print_value_on(&trace_stream);
452 trace_stream.cr();
453 }
454 }
455
456 void Symbol::print_on(outputStream* st) const {
457 st->print("Symbol: '");
458 print_symbol_on(st);
459 st->print("'");
460 st->print(" count %d", refcount());
461 }
462
463 void Symbol::print() const { print_on(tty); }
464
465 // The print_value functions are present in all builds, to support the
466 // disassembler and error reporting.
467 void Symbol::print_value_on(outputStream* st) const {
468 st->print("'");
469 for (int i = 0; i < utf8_length(); i++) {
470 st->print("%c", char_at(i));
471 }
472 st->print("'");
473 }
474
475 void Symbol::print_value() const { print_value_on(tty); }
476
477 bool Symbol::is_valid(Symbol* s) {
478 if (!is_aligned(s, sizeof(MetaWord))) return false;
479 if ((size_t)s < os::min_page_size()) return false;
480
481 if (!os::is_readable_range(s, s + 1)) return false;
482
483 // Symbols are not allocated in Java heap.
484 if (Universe::heap()->is_in(s)) return false;
485
486 int len = s->utf8_length();
487 if (len < 0) return false;
488
489 jbyte* bytes = (jbyte*) s->bytes();
490 return os::is_readable_range(bytes, bytes + len);
491 }
492
493 void Symbol::print_Qvalue_on(outputStream* st) const {
494 st->print("'Q");
495 for (int i = 0; i < utf8_length(); i++) {
496 st->print("%c", char_at(i));
497 }
498 st->print(";'");
499 }
500
501 // SymbolTable prints this in its statistics
502 NOT_PRODUCT(size_t Symbol::_total_count = 0;)
503
504 #ifndef PRODUCT
505 bool Symbol::is_valid_id(vmSymbolID vm_symbol_id) {
506 return vmSymbols::is_valid_id(vm_symbol_id);
507 }
508 #endif