1 /*
2 * Copyright (c) 1997, 2020, 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 "memory/universe.hpp"
35 #include "oops/symbol.hpp"
36 #include "runtime/atomic.hpp"
37 #include "runtime/os.hpp"
38 #include "utilities/utf8.hpp"
39
40 uint32_t Symbol::pack_hash_and_refcount(short hash, int refcount) {
41 STATIC_ASSERT(PERM_REFCOUNT == ((1 << 16) - 1));
42 assert(refcount >= 0, "negative refcount");
43 assert(refcount <= PERM_REFCOUNT, "invalid refcount");
44 uint32_t hi = hash;
45 uint32_t lo = refcount;
46 return (hi << 16) | lo;
47 }
48
49 Symbol::Symbol(const u1* name, int length, int refcount) {
50 _hash_and_refcount = pack_hash_and_refcount((short)os::random(), refcount);
51 _length = length;
52 _body[0] = 0; // in case length == 0
53 for (int i = 0; i < length; i++) {
54 byte_at_put(i, name[i]);
55 }
56 }
57
58 void* Symbol::operator new(size_t sz, int len) throw() {
59 int alloc_size = size(len)*wordSize;
60 address res = (address) AllocateHeap(alloc_size, mtSymbol);
61 return res;
62 }
63
64 void* Symbol::operator new(size_t sz, int len, Arena* arena) throw() {
65 int alloc_size = size(len)*wordSize;
66 address res = (address)arena->Amalloc_4(alloc_size);
67 return res;
68 }
69
70 void Symbol::operator delete(void *p) {
71 assert(((Symbol*)p)->refcount() == 0, "should not call this");
72 FreeHeap(p);
73 }
74
75 void Symbol::set_permanent() {
76 // This is called at a safepoint during dumping of a dynamic CDS archive.
77 assert(SafepointSynchronize::is_at_safepoint(), "must be at a safepoint");
78 _hash_and_refcount = pack_hash_and_refcount(extract_hash(_hash_and_refcount), PERM_REFCOUNT);
79 }
80
81 bool Symbol::is_Q_signature() const {
82 int len = utf8_length();
83 return len > 2 && char_at(0) == JVM_SIGNATURE_VALUETYPE && char_at(len - 1) == JVM_SIGNATURE_ENDCLASS;
84 }
85
86 bool Symbol::is_L_signature() const {
87 int len = utf8_length();
88 return len > 2 && char_at(0) == JVM_SIGNATURE_CLASS && char_at(len - 1) == JVM_SIGNATURE_ENDCLASS;
89 }
90
91 bool Symbol::is_Q_array_signature() const {
92 int l = utf8_length();
93 if (l < 2 || char_at(0) != JVM_SIGNATURE_ARRAY || char_at(l - 1) != JVM_SIGNATURE_ENDCLASS) {
94 return false;
95 }
96 for (int i = 1; i < (l - 2); i++) {
97 char c = char_at(i);
98 if (c == JVM_SIGNATURE_VALUETYPE) {
99 return true;
100 }
101 if (c != JVM_SIGNATURE_ARRAY) {
102 return false;
103 }
104 }
105 return false;
106 }
107
108 bool Symbol::is_Q_method_signature() const {
109 assert(SignatureVerifier::is_valid_method_signature(this), "must be");
110 int len = utf8_length();
111 if (len > 4 && char_at(0) == JVM_SIGNATURE_FUNC) {
112 for (int i=1; i<len-3; i++) { // Must end with ")Qx;", where x is at least one character or more.
113 if (char_at(i) == JVM_SIGNATURE_ENDFUNC && char_at(i+1) == JVM_SIGNATURE_VALUETYPE) {
114 return true;
115 }
116 }
117 }
118 return false;
119 }
120
121 bool Symbol::is_Q_singledim_array_signature() const {
122 int len = utf8_length();
123 return len > 3 && char_at(0) == JVM_SIGNATURE_ARRAY && char_at(1) == JVM_SIGNATURE_VALUETYPE &&
124 char_at(len - 1) == JVM_SIGNATURE_ENDCLASS;
125 }
126
127 Symbol* Symbol::fundamental_name(TRAPS) {
128 if ((char_at(0) == JVM_SIGNATURE_VALUETYPE || char_at(0) == JVM_SIGNATURE_CLASS) && ends_with(JVM_SIGNATURE_ENDCLASS)) {
129 return SymbolTable::new_symbol(this, 1, utf8_length() - 1);
130 } else {
131 // reference count is incremented to be consistent with the behavior with
132 // the SymbolTable::new_symbol() call above
133 this->increment_refcount();
134 return this;
135 }
136 }
137
138 bool Symbol::is_same_fundamental_type(Symbol* s) const {
139 if (this == s) return true;
140 if (utf8_length() < 3) return false;
141 int offset1, offset2, len;
142 if (ends_with(JVM_SIGNATURE_ENDCLASS)) {
143 if (char_at(0) != JVM_SIGNATURE_VALUETYPE && char_at(0) != JVM_SIGNATURE_CLASS) return false;
144 offset1 = 1;
145 len = utf8_length() - 2;
146 } else {
147 offset1 = 0;
148 len = utf8_length();
149 }
150 if (ends_with(JVM_SIGNATURE_ENDCLASS)) {
151 if (s->char_at(0) != JVM_SIGNATURE_VALUETYPE && s->char_at(0) != JVM_SIGNATURE_CLASS) return false;
152 offset2 = 1;
153 } else {
154 offset2 = 0;
155 }
156 if ((offset2 + len) > s->utf8_length()) return false;
157 if ((utf8_length() - offset1 * 2) != (s->utf8_length() - offset2 * 2))
158 return false;
159 int l = len;
160 while (l-- > 0) {
161 if (char_at(offset1 + l) != s->char_at(offset2 + l))
162 return false;
163 }
164 return true;
165 }
166
167 // ------------------------------------------------------------------
168 // Symbol::index_of
169 //
170 // Finds if the given string is a substring of this symbol's utf8 bytes.
171 // Return -1 on failure. Otherwise return the first index where str occurs.
172 int Symbol::index_of_at(int i, const char* str, int len) const {
173 assert(i >= 0 && i <= utf8_length(), "oob");
174 if (len <= 0) return 0;
175 char first_char = str[0];
176 address bytes = (address) ((Symbol*)this)->base();
177 address limit = bytes + utf8_length() - 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 == NULL)
184 return -1; // not found
185 assert(scan >= bytes+i && scan <= limit, "scan oob");
186 if (len <= 2
187 ? (char) scan[len-1] == str[len-1]
188 : memcmp(scan+1, str+1, len-1) == 0) {
189 return (int)(scan - bytes);
190 }
191 }
192 return -1;
193 }
194
195
196 char* Symbol::as_C_string(char* buf, int size) const {
197 if (size > 0) {
198 int len = MIN2(size - 1, utf8_length());
199 for (int i = 0; i < len; i++) {
200 buf[i] = char_at(i);
201 }
202 buf[len] = '\0';
203 }
204 return buf;
205 }
206
207 char* Symbol::as_C_string() const {
208 int len = utf8_length();
209 char* str = NEW_RESOURCE_ARRAY(char, len + 1);
210 return as_C_string(str, len + 1);
211 }
212
213 void Symbol::print_utf8_on(outputStream* st) const {
214 st->print("%s", as_C_string());
215 }
216
217 void Symbol::print_symbol_on(outputStream* st) const {
218 char *s;
219 st = st ? st : tty;
220 {
221 // ResourceMark may not affect st->print(). If st is a string
222 // stream it could resize, using the same resource arena.
223 ResourceMark rm;
224 s = as_quoted_ascii();
225 s = os::strdup(s);
226 }
227 if (s == NULL) {
228 st->print("(null)");
229 } else {
230 st->print("%s", s);
231 os::free(s);
232 }
233 }
234
235 char* Symbol::as_quoted_ascii() const {
236 const char *ptr = (const char *)&_body[0];
237 int quoted_length = UTF8::quoted_ascii_length(ptr, utf8_length());
238 char* result = NEW_RESOURCE_ARRAY(char, quoted_length + 1);
239 UTF8::as_quoted_ascii(ptr, utf8_length(), result, quoted_length + 1);
240 return result;
241 }
242
243 jchar* Symbol::as_unicode(int& length) const {
244 Symbol* this_ptr = (Symbol*)this;
245 length = UTF8::unicode_length((char*)this_ptr->bytes(), utf8_length());
246 jchar* result = NEW_RESOURCE_ARRAY(jchar, length);
247 if (length > 0) {
248 UTF8::convert_to_unicode((char*)this_ptr->bytes(), result, length);
249 }
250 return result;
251 }
252
253 const char* Symbol::as_klass_external_name(char* buf, int size) const {
254 if (size > 0) {
255 char* str = as_C_string(buf, size);
256 int length = (int)strlen(str);
257 // Turn all '/'s into '.'s (also for array klasses)
258 for (int index = 0; index < length; index++) {
259 if (str[index] == JVM_SIGNATURE_SLASH) {
260 str[index] = JVM_SIGNATURE_DOT;
261 }
262 }
263 return str;
264 } else {
265 return buf;
266 }
267 }
268
269 const char* Symbol::as_klass_external_name() const {
270 char* str = as_C_string();
271 int length = (int)strlen(str);
272 // Turn all '/'s into '.'s (also for array klasses)
273 for (int index = 0; index < length; index++) {
274 if (str[index] == JVM_SIGNATURE_SLASH) {
275 str[index] = JVM_SIGNATURE_DOT;
276 }
277 }
278 return str;
279 }
280
281 static void print_class(outputStream *os, const SignatureStream& ss) {
282 int sb = ss.raw_symbol_begin(), se = ss.raw_symbol_end();
283 for (int i = sb; i < se; ++i) {
284 int ch = ss.raw_char_at(i);
285 if (ch == JVM_SIGNATURE_SLASH) {
286 os->put(JVM_SIGNATURE_DOT);
287 } else {
288 os->put(ch);
289 }
290 }
291 }
292
293 static void print_array(outputStream *os, SignatureStream& ss) {
294 int dimensions = ss.skip_array_prefix();
295 assert(dimensions > 0, "");
296 if (ss.is_reference()) {
297 print_class(os, ss);
298 } else {
299 os->print("%s", type2name(ss.type()));
300 }
301 for (int i = 0; i < dimensions; ++i) {
302 os->print("[]");
303 }
304 }
305
306 void Symbol::print_as_signature_external_return_type(outputStream *os) {
307 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
308 if (ss.at_return_type()) {
309 if (ss.is_array()) {
310 print_array(os, ss);
311 } else if (ss.is_reference()) {
312 print_class(os, ss);
313 } else {
314 os->print("%s", type2name(ss.type()));
315 }
316 }
317 }
318 }
319
320 void Symbol::print_as_signature_external_parameters(outputStream *os) {
321 bool first = true;
322 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
323 if (ss.at_return_type()) break;
324 if (!first) { os->print(", "); }
325 if (ss.is_array()) {
326 print_array(os, ss);
327 } else if (ss.is_reference()) {
328 print_class(os, ss);
329 } else {
330 os->print("%s", type2name(ss.type()));
331 }
332 first = false;
333 }
334 }
335
336 // Increment refcount while checking for zero. If the Symbol's refcount becomes zero
337 // a thread could be concurrently removing the Symbol. This is used during SymbolTable
338 // lookup to avoid reviving a dead Symbol.
339 bool Symbol::try_increment_refcount() {
340 uint32_t found = _hash_and_refcount;
341 while (true) {
342 uint32_t old_value = found;
343 int refc = extract_refcount(old_value);
344 if (refc == PERM_REFCOUNT) {
345 return true; // sticky max or created permanent
346 } else if (refc == 0) {
347 return false; // dead, can't revive.
348 } else {
349 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value + 1);
350 if (found == old_value) {
351 return true; // successfully updated.
352 }
353 // refcount changed, try again.
354 }
355 }
356 }
357
358 // The increment_refcount() is called when not doing lookup. It is assumed that you
359 // have a symbol with a non-zero refcount and it can't become zero while referenced by
360 // this caller.
361 void Symbol::increment_refcount() {
362 if (!try_increment_refcount()) {
363 #ifdef ASSERT
364 print();
365 fatal("refcount has gone to zero");
366 #endif
367 }
368 #ifndef PRODUCT
369 if (refcount() != PERM_REFCOUNT) { // not a permanent symbol
370 NOT_PRODUCT(Atomic::inc(&_total_count);)
371 }
372 #endif
373 }
374
375 // Decrement refcount potentially while racing increment, so we need
376 // to check the value after attempting to decrement so that if another
377 // thread increments to PERM_REFCOUNT the value is not decremented.
378 void Symbol::decrement_refcount() {
379 uint32_t found = _hash_and_refcount;
380 while (true) {
381 uint32_t old_value = found;
382 int refc = extract_refcount(old_value);
383 if (refc == PERM_REFCOUNT) {
384 return; // refcount is permanent, permanent is sticky
385 } else if (refc == 0) {
386 #ifdef ASSERT
387 print();
388 fatal("refcount underflow");
389 #endif
390 return;
391 } else {
392 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value - 1);
393 if (found == old_value) {
394 return; // successfully updated.
395 }
396 // refcount changed, try again.
397 }
398 }
399 }
400
401 void Symbol::make_permanent() {
402 uint32_t found = _hash_and_refcount;
403 while (true) {
404 uint32_t old_value = found;
405 int refc = extract_refcount(old_value);
406 if (refc == PERM_REFCOUNT) {
407 return; // refcount is permanent, permanent is sticky
408 } else if (refc == 0) {
409 #ifdef ASSERT
410 print();
411 fatal("refcount underflow");
412 #endif
413 return;
414 } else {
415 int hash = extract_hash(old_value);
416 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, pack_hash_and_refcount(hash, PERM_REFCOUNT));
417 if (found == old_value) {
418 return; // successfully updated.
419 }
420 // refcount changed, try again.
421 }
422 }
423 }
424
425 void Symbol::metaspace_pointers_do(MetaspaceClosure* it) {
426 if (log_is_enabled(Trace, cds)) {
427 LogStream trace_stream(Log(cds)::trace());
428 trace_stream.print("Iter(Symbol): %p ", this);
429 print_value_on(&trace_stream);
430 trace_stream.cr();
431 }
432 }
433
434 void Symbol::print_on(outputStream* st) const {
435 st->print("Symbol: '");
436 print_symbol_on(st);
437 st->print("'");
438 st->print(" count %d", refcount());
439 }
440
441 void Symbol::print() const { print_on(tty); }
442
443 // The print_value functions are present in all builds, to support the
444 // disassembler and error reporting.
445 void Symbol::print_value_on(outputStream* st) const {
446 st->print("'");
447 for (int i = 0; i < utf8_length(); i++) {
448 st->print("%c", char_at(i));
449 }
450 st->print("'");
451 }
452
453 void Symbol::print_value() const { print_value_on(tty); }
454
455 bool Symbol::is_valid(Symbol* s) {
456 if (!is_aligned(s, sizeof(MetaWord))) return false;
457 if ((size_t)s < os::min_page_size()) return false;
458
459 if (!os::is_readable_range(s, s + 1)) return false;
460
461 // Symbols are not allocated in Java heap.
462 if (Universe::heap()->is_in(s)) return false;
463
464 int len = s->utf8_length();
465 if (len < 0) return false;
466
467 jbyte* bytes = (jbyte*) s->bytes();
468 return os::is_readable_range(bytes, bytes + len);
469 }
470
471 void Symbol::print_Qvalue_on(outputStream* st) const {
472 if (this == NULL) {
473 st->print("NULL");
474 } else {
475 st->print("'Q");
476 for (int i = 0; i < utf8_length(); i++) {
477 st->print("%c", char_at(i));
478 }
479 st->print(";'");
480 }
481 }
482
483 // SymbolTable prints this in its statistics
484 NOT_PRODUCT(size_t Symbol::_total_count = 0;)