1 /*
2 * Copyright (c) 1997, 2024, 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 #include "precompiled.hpp"
26 #include "cds/archiveBuilder.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 assert(length <= max_length(), "SymbolTable should have caught this!");
58 _hash_and_refcount = pack_hash_and_refcount((short)os::random(), refcount);
59 _length = (u2)length;
60 // _body[0..1] are allocated in the header just by coincidence in the current
61 // implementation of Symbol. They are read by identity_hash(), so make sure they
62 // are initialized.
63 // No other code should assume that _body[0..1] are always allocated. E.g., do
64 // not unconditionally read base()[0] as that will be invalid for an empty Symbol.
65 _body[0] = _body[1] = 0;
66 memcpy(_body, name, length);
67 }
68
69 // This copies the symbol when it is added to the ConcurrentHashTable.
70 Symbol::Symbol(const Symbol& s1) {
71 _hash_and_refcount = s1._hash_and_refcount;
72 _length = s1._length;
73 memcpy(_body, s1._body, _length);
74 }
75
76 #if INCLUDE_CDS
77 void Symbol::update_identity_hash() {
78 assert(SafepointSynchronize::is_at_safepoint(), "must be at a safepoint");
79 _hash_and_refcount = pack_hash_and_refcount((short)ArchiveBuilder::current()->entropy(), PERM_REFCOUNT);
80 }
81
82 void Symbol::set_permanent() {
83 // This is called at a safepoint during dumping of a dynamic CDS archive.
84 assert(SafepointSynchronize::is_at_safepoint(), "must be at a safepoint");
85 _hash_and_refcount = pack_hash_and_refcount(extract_hash(_hash_and_refcount), PERM_REFCOUNT);
86 }
87 #endif
88
89 Symbol* Symbol::fundamental_name(TRAPS) {
90 if (char_at(0) == JVM_SIGNATURE_CLASS && ends_with(JVM_SIGNATURE_ENDCLASS)) {
91 return SymbolTable::new_symbol(this, 1, utf8_length() - 1);
92 } else {
93 // reference count is incremented to be consistent with the behavior with
94 // the SymbolTable::new_symbol() call above
95 this->increment_refcount();
96 return this;
97 }
98 }
99
100 bool Symbol::is_same_fundamental_type(Symbol* s) const {
101 if (this == s) return true;
102 if (utf8_length() < 3) return false;
103 int offset1, offset2, len;
104 if (ends_with(JVM_SIGNATURE_ENDCLASS)) {
105 if (char_at(0) != JVM_SIGNATURE_CLASS) return false;
106 offset1 = 1;
107 len = utf8_length() - 2;
108 } else {
109 offset1 = 0;
110 len = utf8_length();
111 }
112 if (ends_with(JVM_SIGNATURE_ENDCLASS)) {
113 if (s->char_at(0) != JVM_SIGNATURE_CLASS) return false;
114 offset2 = 1;
115 } else {
116 offset2 = 0;
117 }
118 if ((offset2 + len) > s->utf8_length()) return false;
119 if ((utf8_length() - offset1 * 2) != (s->utf8_length() - offset2 * 2))
120 return false;
121 int l = len;
122 while (l-- > 0) {
123 if (char_at(offset1 + l) != s->char_at(offset2 + l))
124 return false;
125 }
126 return true;
127 }
128
129 // ------------------------------------------------------------------
130 // Symbol::index_of
131 //
132 // Test if we have the give substring at or after the i-th char of this
133 // symbol's utf8 bytes.
134 // Return -1 on failure. Otherwise return the first index where substr occurs.
135 int Symbol::index_of_at(int i, const char* substr, int substr_len) const {
136 assert(i >= 0 && i <= utf8_length(), "oob");
137 if (substr_len <= 0) return 0;
138 char first_char = substr[0];
139 address bytes = (address) ((Symbol*)this)->base();
140 address limit = bytes + utf8_length() - substr_len; // inclusive limit
141 address scan = bytes + i;
142 if (scan > limit)
143 return -1;
144 for (; scan <= limit; scan++) {
145 scan = (address) memchr(scan, first_char, (limit + 1 - scan));
146 if (scan == nullptr)
147 return -1; // not found
148 assert(scan >= bytes+i && scan <= limit, "scan oob");
149 if (substr_len <= 2
150 ? (char) scan[substr_len-1] == substr[substr_len-1]
151 : memcmp(scan+1, substr+1, substr_len-1) == 0) {
152 return (int)(scan - bytes);
153 }
154 }
155 return -1;
156 }
157
158 bool Symbol::is_star_match(const char* pattern) const {
159 if (strchr(pattern, '*') == nullptr) {
160 return equals(pattern);
161 } else {
162 ResourceMark rm;
163 char* buf = as_C_string();
164 return StringUtils::is_star_match(pattern, buf);
165 }
166 }
167
168 char* Symbol::as_C_string(char* buf, int size) const {
169 if (size > 0) {
170 int len = MIN2(size - 1, utf8_length());
171 for (int i = 0; i < len; i++) {
172 buf[i] = char_at(i);
173 }
174 buf[len] = '\0';
175 }
176 return buf;
177 }
178
179 char* Symbol::as_C_string() const {
180 int len = utf8_length();
181 char* str = NEW_RESOURCE_ARRAY(char, len + 1);
182 return as_C_string(str, len + 1);
183 }
184
185 void Symbol::print_utf8_on(outputStream* st) const {
186 st->print("%s", as_C_string());
187 }
188
189 void Symbol::print_symbol_on(outputStream* st) const {
190 char *s;
191 st = st ? st : tty;
192 {
193 // ResourceMark may not affect st->print(). If st is a string
194 // stream it could resize, using the same resource arena.
195 ResourceMark rm;
196 s = as_quoted_ascii();
197 s = os::strdup(s);
198 }
199 if (s == nullptr) {
200 st->print("(null)");
201 } else {
202 st->print("%s", s);
203 os::free(s);
204 }
205 }
206
207 char* Symbol::as_quoted_ascii() const {
208 const char *ptr = (const char *)&_body[0];
209 size_t quoted_length = UTF8::quoted_ascii_length(ptr, utf8_length());
210 char* result = NEW_RESOURCE_ARRAY(char, quoted_length + 1);
211 UTF8::as_quoted_ascii(ptr, utf8_length(), result, quoted_length + 1);
212 return result;
213 }
214
215 jchar* Symbol::as_unicode(int& length) const {
216 Symbol* this_ptr = (Symbol*)this;
217 length = UTF8::unicode_length((char*)this_ptr->bytes(), utf8_length());
218 jchar* result = NEW_RESOURCE_ARRAY(jchar, length);
219 if (length > 0) {
220 UTF8::convert_to_unicode((char*)this_ptr->bytes(), result, length);
221 }
222 return result;
223 }
224
225 const char* Symbol::as_klass_external_name(char* buf, int size) const {
226 if (size > 0) {
227 char* str = as_C_string(buf, size);
228 int length = (int)strlen(str);
229 // Turn all '/'s into '.'s (also for array klasses)
230 for (int index = 0; index < length; index++) {
231 if (str[index] == JVM_SIGNATURE_SLASH) {
232 str[index] = JVM_SIGNATURE_DOT;
233 }
234 }
235 return str;
236 } else {
237 return buf;
238 }
239 }
240
241 const char* Symbol::as_klass_external_name() const {
242 char* str = as_C_string();
243 int length = (int)strlen(str);
244 // Turn all '/'s into '.'s (also for array klasses)
245 for (int index = 0; index < length; index++) {
246 if (str[index] == JVM_SIGNATURE_SLASH) {
247 str[index] = JVM_SIGNATURE_DOT;
248 }
249 }
250 return str;
251 }
252
253 static void print_class(outputStream *os, const SignatureStream& ss) {
254 int sb = ss.raw_symbol_begin(), se = ss.raw_symbol_end();
255 for (int i = sb; i < se; ++i) {
256 char ch = ss.raw_char_at(i);
257 if (ch == JVM_SIGNATURE_SLASH) {
258 os->put(JVM_SIGNATURE_DOT);
259 } else {
260 os->put(ch);
261 }
262 }
263 }
264
265 static void print_array(outputStream *os, SignatureStream& ss) {
266 int dimensions = ss.skip_array_prefix();
267 assert(dimensions > 0, "");
268 if (ss.is_reference()) {
269 print_class(os, ss);
270 } else {
271 os->print("%s", type2name(ss.type()));
272 }
273 for (int i = 0; i < dimensions; ++i) {
274 os->print("[]");
275 }
276 }
277
278 void Symbol::print_as_signature_external_return_type(outputStream *os) {
279 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
280 if (ss.at_return_type()) {
281 if (ss.is_array()) {
282 print_array(os, ss);
283 } else if (ss.is_reference()) {
284 print_class(os, ss);
285 } else {
286 os->print("%s", type2name(ss.type()));
287 }
288 }
289 }
290 }
291
292 void Symbol::print_as_signature_external_parameters(outputStream *os) {
293 bool first = true;
294 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
295 if (ss.at_return_type()) break;
296 if (!first) { os->print(", "); }
297 if (ss.is_array()) {
298 print_array(os, ss);
299 } else if (ss.is_reference()) {
300 print_class(os, ss);
301 } else {
302 os->print("%s", type2name(ss.type()));
303 }
304 first = false;
305 }
306 }
307
308 void Symbol::print_as_field_external_type(outputStream *os) {
309 SignatureStream ss(this, false);
310 assert(!ss.is_done(), "must have at least one element in field ref");
311 assert(!ss.at_return_type(), "field ref cannot be a return type");
312 assert(!Signature::is_method(this), "field ref cannot be a method");
313
314 if (ss.is_array()) {
315 print_array(os, ss);
316 } else if (ss.is_reference()) {
317 print_class(os, ss);
318 } else {
319 os->print("%s", type2name(ss.type()));
320 }
321 #ifdef ASSERT
322 ss.next();
323 assert(ss.is_done(), "must have at most one element in field ref");
324 #endif
325 }
326
327 // Increment refcount while checking for zero. If the Symbol's refcount becomes zero
328 // a thread could be concurrently removing the Symbol. This is used during SymbolTable
329 // lookup to avoid reviving a dead Symbol.
330 bool Symbol::try_increment_refcount() {
331 uint32_t found = _hash_and_refcount;
332 while (true) {
333 uint32_t old_value = found;
334 int refc = extract_refcount(old_value);
335 if (refc == PERM_REFCOUNT) {
336 return true; // sticky max or created permanent
337 } else if (refc == 0) {
338 return false; // dead, can't revive.
339 } else {
340 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value + 1);
341 if (found == old_value) {
342 return true; // successfully updated.
343 }
344 // refcount changed, try again.
345 }
346 }
347 }
348
349 // The increment_refcount() is called when not doing lookup. It is assumed that you
350 // have a symbol with a non-zero refcount and it can't become zero while referenced by
351 // this caller.
352 void Symbol::increment_refcount() {
353 if (!try_increment_refcount()) {
354 print();
355 fatal("refcount has gone to zero");
356 }
357 #ifndef PRODUCT
358 if (refcount() != PERM_REFCOUNT) { // not a permanent symbol
359 NOT_PRODUCT(Atomic::inc(&_total_count);)
360 }
361 #endif
362 }
363
364 // Decrement refcount potentially while racing increment, so we need
365 // to check the value after attempting to decrement so that if another
366 // thread increments to PERM_REFCOUNT the value is not decremented.
367 void Symbol::decrement_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; // refcount is permanent, permanent is sticky
374 } else if (refc == 0) {
375 print();
376 fatal("refcount underflow");
377 return;
378 } else {
379 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value - 1);
380 if (found == old_value) {
381 return; // successfully updated.
382 }
383 // refcount changed, try again.
384 }
385 }
386 }
387
388 void Symbol::make_permanent() {
389 uint32_t found = _hash_and_refcount;
390 while (true) {
391 uint32_t old_value = found;
392 int refc = extract_refcount(old_value);
393 if (refc == PERM_REFCOUNT) {
394 return; // refcount is permanent, permanent is sticky
395 } else if (refc == 0) {
396 print();
397 fatal("refcount underflow");
398 return;
399 } else {
400 short hash = extract_hash(old_value);
401 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, pack_hash_and_refcount(hash, PERM_REFCOUNT));
402 if (found == old_value) {
403 return; // successfully updated.
404 }
405 // refcount changed, try again.
406 }
407 }
408 }
409
410 void Symbol::metaspace_pointers_do(MetaspaceClosure* it) {
411 if (log_is_enabled(Trace, cds)) {
412 LogStream trace_stream(Log(cds)::trace());
413 trace_stream.print("Iter(Symbol): %p ", this);
414 print_value_on(&trace_stream);
415 trace_stream.cr();
416 }
417 }
418
419 void Symbol::print_on(outputStream* st) const {
420 st->print("Symbol: '");
421 print_symbol_on(st);
422 st->print("'");
423 st->print(" count %d", refcount());
424 }
425
426 void Symbol::print() const { print_on(tty); }
427
428 // The print_value functions are present in all builds, to support the
429 // disassembler and error reporting.
430 void Symbol::print_value_on(outputStream* st) const {
431 st->print_raw("'", 1);
432 st->print_raw((const char*)base(), utf8_length());
433 st->print_raw("'", 1);
434 }
435
436 void Symbol::print_value() const { print_value_on(tty); }
437
438 bool Symbol::is_valid(Symbol* s) {
439 if (!is_aligned(s, sizeof(MetaWord))) return false;
440 if ((size_t)s < os::min_page_size()) return false;
441
442 if (!os::is_readable_range(s, s + 1)) return false;
443
444 // Symbols are not allocated in Java heap.
445 if (Universe::heap()->is_in(s)) return false;
446
447 int len = s->utf8_length();
448 if (len < 0) return false;
449
450 jbyte* bytes = (jbyte*) s->bytes();
451 return os::is_readable_range(bytes, bytes + len);
452 }
453
454 // SymbolTable prints this in its statistics
455 NOT_PRODUCT(size_t Symbol::_total_count = 0;)
456
457 #ifndef PRODUCT
458 bool Symbol::is_valid_id(vmSymbolID vm_symbol_id) {
459 return vmSymbols::is_valid_id(vm_symbol_id);
460 }
461 #endif