1 /*
2 * Copyright (c) 1997, 2025, 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 "cds/archiveBuilder.hpp"
26 #include "cds/metaspaceShared.hpp"
27 #include "classfile/altHashing.hpp"
28 #include "classfile/classLoaderData.hpp"
29 #include "classfile/vmSymbols.hpp"
30 #include "gc/shared/collectedHeap.hpp"
31 #include "logging/log.hpp"
32 #include "logging/logStream.hpp"
33 #include "memory/allocation.inline.hpp"
34 #include "memory/resourceArea.hpp"
35 #include "memory/universe.hpp"
36 #include "oops/symbol.hpp"
37 #include "runtime/atomic.hpp"
38 #include "runtime/mutexLocker.hpp"
39 #include "runtime/os.hpp"
40 #include "runtime/signature.hpp"
41 #include "utilities/stringUtils.hpp"
42 #include "utilities/utf8.hpp"
43
44 Symbol* Symbol::_vm_symbols[vmSymbols::number_of_symbols()];
45
46 uint32_t Symbol::pack_hash_and_refcount(short hash, int refcount) {
47 STATIC_ASSERT(PERM_REFCOUNT == ((1 << 16) - 1));
48 assert(refcount >= 0, "negative refcount");
49 assert(refcount <= PERM_REFCOUNT, "invalid refcount");
50 uint32_t hi = hash;
51 uint32_t lo = refcount;
52 return (hi << 16) | lo;
53 }
54
55 Symbol::Symbol(const u1* name, int length, int refcount) {
56 assert(length <= max_length(), "SymbolTable should have caught this!");
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 assert(SafepointSynchronize::is_at_safepoint(), "must be at a safepoint");
78 _hash_and_refcount = pack_hash_and_refcount((short)ArchiveBuilder::current()->entropy(), PERM_REFCOUNT);
79 }
80
81 void Symbol::set_permanent() {
82 // This is called at a safepoint during dumping of a dynamic CDS archive.
83 assert(SafepointSynchronize::is_at_safepoint(), "must be at a safepoint");
84 _hash_and_refcount = pack_hash_and_refcount(extract_hash(_hash_and_refcount), PERM_REFCOUNT);
85 }
86 #endif
87
88 // ------------------------------------------------------------------
89 // Symbol::index_of
90 //
91 // Test if we have the give substring at or after the i-th char of this
92 // symbol's utf8 bytes.
93 // Return -1 on failure. Otherwise return the first index where substr occurs.
94 int Symbol::index_of_at(int i, const char* substr, int substr_len) const {
95 assert(i >= 0 && i <= utf8_length(), "oob");
96 if (substr_len <= 0) return 0;
97 char first_char = substr[0];
98 address bytes = (address) ((Symbol*)this)->base();
99 address limit = bytes + utf8_length() - substr_len; // inclusive limit
100 address scan = bytes + i;
101 if (scan > limit)
102 return -1;
103 for (; scan <= limit; scan++) {
104 scan = (address) memchr(scan, first_char, (limit + 1 - scan));
105 if (scan == nullptr)
106 return -1; // not found
107 assert(scan >= bytes+i && scan <= limit, "scan oob");
108 if (substr_len <= 2
109 ? (char) scan[substr_len-1] == substr[substr_len-1]
110 : memcmp(scan+1, substr+1, substr_len-1) == 0) {
111 return (int)(scan - bytes);
112 }
113 }
114 return -1;
115 }
116
117 bool Symbol::is_star_match(const char* pattern) const {
118 if (strchr(pattern, '*') == nullptr) {
119 return equals(pattern);
120 } else {
121 ResourceMark rm;
122 char* buf = as_C_string();
123 return StringUtils::is_star_match(pattern, buf);
124 }
125 }
126
127 char* Symbol::as_C_string(char* buf, int size) const {
128 if (size > 0) {
129 int len = MIN2(size - 1, utf8_length());
130 for (int i = 0; i < len; i++) {
131 buf[i] = char_at(i);
132 }
133 buf[len] = '\0';
134 }
135 return buf;
136 }
137
138 char* Symbol::as_C_string() const {
139 int len = utf8_length();
140 char* str = NEW_RESOURCE_ARRAY(char, len + 1);
141 return as_C_string(str, len + 1);
142 }
143
144 void Symbol::print_utf8_on(outputStream* st) const {
145 st->print("%s", as_C_string());
146 }
147
148 void Symbol::print_symbol_on(outputStream* st) const {
149 char *s;
150 st = st ? st : tty;
151 {
152 // ResourceMark may not affect st->print(). If st is a string
153 // stream it could resize, using the same resource arena.
154 ResourceMark rm;
155 s = as_quoted_ascii();
156 s = os::strdup(s);
157 }
158 if (s == nullptr) {
159 st->print("(null)");
160 } else {
161 st->print("%s", s);
162 os::free(s);
163 }
164 }
165
166 char* Symbol::as_quoted_ascii() const {
167 const char *ptr = (const char *)&_body[0];
168 size_t quoted_length = UTF8::quoted_ascii_length(ptr, utf8_length());
169 char* result = NEW_RESOURCE_ARRAY(char, quoted_length + 1);
170 UTF8::as_quoted_ascii(ptr, utf8_length(), result, quoted_length + 1);
171 return result;
172 }
173
174 jchar* Symbol::as_unicode(int& length) const {
175 Symbol* this_ptr = (Symbol*)this;
176 length = UTF8::unicode_length((char*)this_ptr->bytes(), utf8_length());
177 jchar* result = NEW_RESOURCE_ARRAY(jchar, length);
178 if (length > 0) {
179 UTF8::convert_to_unicode((char*)this_ptr->bytes(), result, length);
180 }
181 return result;
182 }
183
184 const char* Symbol::as_klass_external_name(char* buf, int size) const {
185 if (size > 0) {
186 char* str = as_C_string(buf, size);
187 int length = (int)strlen(str);
188 // Turn all '/'s into '.'s (also for array klasses)
189 for (int index = 0; index < length; index++) {
190 if (str[index] == JVM_SIGNATURE_SLASH) {
191 str[index] = JVM_SIGNATURE_DOT;
192 }
193 }
194 return str;
195 } else {
196 return buf;
197 }
198 }
199
200 const char* Symbol::as_klass_external_name() const {
201 char* str = as_C_string();
202 int length = (int)strlen(str);
203 // Turn all '/'s into '.'s (also for array klasses)
204 for (int index = 0; index < length; index++) {
205 if (str[index] == JVM_SIGNATURE_SLASH) {
206 str[index] = JVM_SIGNATURE_DOT;
207 }
208 }
209 return str;
210 }
211
212 static void print_class(outputStream *os, const SignatureStream& ss) {
213 int sb = ss.raw_symbol_begin(), se = ss.raw_symbol_end();
214 for (int i = sb; i < se; ++i) {
215 char ch = ss.raw_char_at(i);
216 if (ch == JVM_SIGNATURE_SLASH) {
217 os->put(JVM_SIGNATURE_DOT);
218 } else {
219 os->put(ch);
220 }
221 }
222 }
223
224 static void print_array(outputStream *os, SignatureStream& ss) {
225 int dimensions = ss.skip_array_prefix();
226 assert(dimensions > 0, "");
227 if (ss.is_reference()) {
228 print_class(os, ss);
229 } else {
230 os->print("%s", type2name(ss.type()));
231 }
232 for (int i = 0; i < dimensions; ++i) {
233 os->print("[]");
234 }
235 }
236
237 void Symbol::print_as_signature_external_return_type(outputStream *os) {
238 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
239 if (ss.at_return_type()) {
240 if (ss.is_array()) {
241 print_array(os, ss);
242 } else if (ss.is_reference()) {
243 print_class(os, ss);
244 } else {
245 os->print("%s", type2name(ss.type()));
246 }
247 }
248 }
249 }
250
251 void Symbol::print_as_signature_external_parameters(outputStream *os) {
252 bool first = true;
253 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
254 if (ss.at_return_type()) break;
255 if (!first) { os->print(", "); }
256 if (ss.is_array()) {
257 print_array(os, ss);
258 } else if (ss.is_reference()) {
259 print_class(os, ss);
260 } else {
261 os->print("%s", type2name(ss.type()));
262 }
263 first = false;
264 }
265 }
266
267 void Symbol::print_as_field_external_type(outputStream *os) {
268 SignatureStream ss(this, false);
269 assert(!ss.is_done(), "must have at least one element in field ref");
270 assert(!ss.at_return_type(), "field ref cannot be a return type");
271 assert(!Signature::is_method(this), "field ref cannot be a method");
272
273 if (ss.is_array()) {
274 print_array(os, ss);
275 } else if (ss.is_reference()) {
276 print_class(os, ss);
277 } else {
278 os->print("%s", type2name(ss.type()));
279 }
280 #ifdef ASSERT
281 ss.next();
282 assert(ss.is_done(), "must have at most one element in field ref");
283 #endif
284 }
285
286 // Increment refcount while checking for zero. If the Symbol's refcount becomes zero
287 // a thread could be concurrently removing the Symbol. This is used during SymbolTable
288 // lookup to avoid reviving a dead Symbol.
289 bool Symbol::try_increment_refcount() {
290 uint32_t found = _hash_and_refcount;
291 while (true) {
292 uint32_t old_value = found;
293 int refc = extract_refcount(old_value);
294 if (refc == PERM_REFCOUNT) {
295 return true; // sticky max or created permanent
296 } else if (refc == 0) {
297 return false; // dead, can't revive.
298 } else {
299 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value + 1);
300 if (found == old_value) {
301 return true; // successfully updated.
302 }
303 // refcount changed, try again.
304 }
305 }
306 }
307
308 // The increment_refcount() is called when not doing lookup. It is assumed that you
309 // have a symbol with a non-zero refcount and it can't become zero while referenced by
310 // this caller.
311 void Symbol::increment_refcount() {
312 if (!try_increment_refcount()) {
313 print();
314 fatal("refcount has gone to zero");
315 }
316 #ifndef PRODUCT
317 if (refcount() != PERM_REFCOUNT) { // not a permanent symbol
318 NOT_PRODUCT(Atomic::inc(&_total_count);)
319 }
320 #endif
321 }
322
323 // Decrement refcount potentially while racing increment, so we need
324 // to check the value after attempting to decrement so that if another
325 // thread increments to PERM_REFCOUNT the value is not decremented.
326 void Symbol::decrement_refcount() {
327 uint32_t found = _hash_and_refcount;
328 while (true) {
329 uint32_t old_value = found;
330 int refc = extract_refcount(old_value);
331 if (refc == PERM_REFCOUNT) {
332 return; // refcount is permanent, permanent is sticky
333 } else if (refc == 0) {
334 print();
335 fatal("refcount underflow");
336 return;
337 } else {
338 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value - 1);
339 if (found == old_value) {
340 return; // successfully updated.
341 }
342 // refcount changed, try again.
343 }
344 }
345 }
346
347 void Symbol::make_permanent() {
348 uint32_t found = _hash_and_refcount;
349 while (true) {
350 uint32_t old_value = found;
351 int refc = extract_refcount(old_value);
352 if (refc == PERM_REFCOUNT) {
353 return; // refcount is permanent, permanent is sticky
354 } else if (refc == 0) {
355 print();
356 fatal("refcount underflow");
357 return;
358 } else {
359 short hash = extract_hash(old_value);
360 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, pack_hash_and_refcount(hash, PERM_REFCOUNT));
361 if (found == old_value) {
362 return; // successfully updated.
363 }
364 // refcount changed, try again.
365 }
366 }
367 }
368
369 void Symbol::metaspace_pointers_do(MetaspaceClosure* it) {
370 if (log_is_enabled(Trace, cds)) {
371 LogStream trace_stream(Log(cds)::trace());
372 trace_stream.print("Iter(Symbol): %p ", this);
373 print_value_on(&trace_stream);
374 trace_stream.cr();
375 }
376 }
377
378 void Symbol::print_on(outputStream* st) const {
379 st->print("Symbol: '");
380 print_symbol_on(st);
381 st->print("'");
382 st->print(" count %d", refcount());
383 }
384
385 void Symbol::print() const { print_on(tty); }
386
387 // The print_value functions are present in all builds, to support the
388 // disassembler and error reporting.
389 void Symbol::print_value_on(outputStream* st) const {
390 st->print_raw("'", 1);
391 st->print_raw((const char*)base(), utf8_length());
392 st->print_raw("'", 1);
393 }
394
395 void Symbol::print_value() const { print_value_on(tty); }
396
397 bool Symbol::is_valid(Symbol* s) {
398 if (!is_aligned(s, sizeof(MetaWord))) return false;
399 if ((size_t)s < os::min_page_size()) return false;
400
401 if (!os::is_readable_range(s, s + 1)) return false;
402
403 // Symbols are not allocated in Java heap.
404 if (Universe::heap()->is_in(s)) return false;
405
406 int len = s->utf8_length();
407 if (len < 0) return false;
408
409 jbyte* bytes = (jbyte*) s->bytes();
410 return os::is_readable_range(bytes, bytes + len);
411 }
412
413 // SymbolTable prints this in its statistics
414 NOT_PRODUCT(size_t Symbol::_total_count = 0;)
415
416 #ifndef PRODUCT
417 bool Symbol::is_valid_id(vmSymbolID vm_symbol_id) {
418 return vmSymbols::is_valid_id(vm_symbol_id);
419 }
420 #endif