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 "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_length_and_refcount(int length, int refcount) {
41 STATIC_ASSERT(max_symbol_length == ((1 << 16) - 1));
42 STATIC_ASSERT(PERM_REFCOUNT == ((1 << 16) - 1));
43 assert(length >= 0, "negative length");
44 assert(length <= max_symbol_length, "too long symbol");
45 assert(refcount >= 0, "negative refcount");
46 assert(refcount <= PERM_REFCOUNT, "invalid refcount");
47 uint32_t hi = length;
48 uint32_t lo = refcount;
49 return (hi << 16) | lo;
50 }
51
52 Symbol::Symbol(const u1* name, int length, int refcount) {
53 _length_and_refcount = pack_length_and_refcount(length, refcount);
54 _identity_hash = (short)os::random();
55 for (int i = 0; i < length; i++) {
56 byte_at_put(i, name[i]);
57 }
58 }
59
60 void* Symbol::operator new(size_t sz, int len) throw() {
61 int alloc_size = size(len)*wordSize;
62 address res = (address) AllocateHeap(alloc_size, mtSymbol);
63 return res;
64 }
65
66 void* Symbol::operator new(size_t sz, int len, Arena* arena) throw() {
67 int alloc_size = size(len)*wordSize;
68 address res = (address)arena->Amalloc_4(alloc_size);
69 return res;
70 }
71
72 void Symbol::operator delete(void *p) {
73 assert(((Symbol*)p)->refcount() == 0, "should not call this");
74 FreeHeap(p);
75 }
76
77 void Symbol::set_permanent() {
78 // This is called at a safepoint during dumping of a dynamic CDS archive.
79 assert(SafepointSynchronize::is_at_safepoint(), "must be at a safepoint");
80 _length_and_refcount = pack_length_and_refcount(length(), PERM_REFCOUNT);
81 }
82
83
84 // ------------------------------------------------------------------
85 // Symbol::contains_byte_at
86 //
87 // Tests if the symbol contains the given byte at the given position.
88 bool Symbol::contains_byte_at(int position, char code_byte) const {
89 if (position < 0) return false; // can happen with ends_with
90 if (position >= utf8_length()) return false;
91 return code_byte == char_at(position);
92 }
93
94 // ------------------------------------------------------------------
95 // Symbol::contains_utf8_at
96 //
97 // Tests if the symbol contains the given utf8 substring
98 // at the given byte position.
99 bool Symbol::contains_utf8_at(int position, const char* substring, int len) const {
100 assert(len > 0 && substring != NULL && (int) strlen(substring) >= len,
101 "substring must be valid");
102 if (len == 1) return contains_byte_at(position, substring[0]);
103 if (position < 0) return false; // can happen with ends_with
104 if (position + len > utf8_length()) return false;
105 while (len-- > 0) {
106 if (substring[len] != char_at(position + len))
107 return false;
108 }
109 assert(len == -1, "we should be at the beginning");
110 return true;
111 }
112
113 bool Symbol::is_Q_signature() const {
114 return utf8_length() > 2 && char_at(0) == 'Q' && ends_with(';');
115 }
116
117 bool Symbol::is_Q_array_signature() const {
118 int l = utf8_length();
119 if (l < 2 || char_at(0) != '[' || char_at(l - 1) != ';') {
120 return false;
121 }
122 for (int i = 1; i < (l - 2); i++) {
123 char c = char_at(i);
124 if (c == 'Q') {
125 return true;
126 }
127 if (c != '[') {
128 return false;
129 }
130 }
131 return false;
132 }
133
134 bool Symbol::is_Q_method_signature() const {
135 assert(SignatureVerifier::is_valid_method_signature(this), "must be");
136 int len = utf8_length();
137 if (len > 4 && char_at(0) == '(') {
138 for (int i=1; i<len-3; i++) { // Must end with ")Qx;", where x is at least one character or more.
139 if (char_at(i) == ')' && char_at(i+1) == 'Q') {
140 return true;
141 }
142 }
143 }
144 return false;
145 }
146
147 bool Symbol::is_Q_singledim_array_signature() const {
148 return utf8_length() > 3 && char_at(0) == '[' && char_at(1) == 'Q' && ends_with(';');
149 }
150
151 Symbol* Symbol::fundamental_name(TRAPS) {
152 if ((char_at(0) == 'Q' || char_at(0) == 'L') && ends_with(';')) {
153 return SymbolTable::new_symbol(this, 1, utf8_length() - 1);
154 } else {
155 // reference count is incremented to be consistent with the behavior with
156 // the SymbolTable::new_symbol() call above
157 this->increment_refcount();
158 return this;
159 }
160 }
161
162 bool Symbol::is_same_fundamental_type(Symbol* s) const {
163 if (this == s) return true;
164 if (utf8_length() < 3) return false;
165 int offset1, offset2, len;
166 if (ends_with(';')) {
167 if (char_at(0) != 'Q' && char_at(0) != 'L') return false;
168 offset1 = 1;
169 len = utf8_length() - 2;
170 } else {
171 offset1 = 0;
172 len = utf8_length();
173 }
174 if (ends_with(';')) {
175 if (s->char_at(0) != 'Q' && s->char_at(0) != 'L') return false;
176 offset2 = 1;
177 } else {
178 offset2 = 0;
179 }
180 if ((offset2 + len) > s->utf8_length()) return false;
181 if ((utf8_length() - offset1 * 2) != (s->utf8_length() - offset2 * 2))
182 return false;
183 int l = len;
184 while (l-- > 0) {
185 if (char_at(offset1 + l) != s->char_at(offset2 + l))
186 return false;
187 }
188 return true;
189 }
190
191 // ------------------------------------------------------------------
192 // Symbol::index_of
193 //
194 // Finds if the given string is a substring of this symbol's utf8 bytes.
195 // Return -1 on failure. Otherwise return the first index where str occurs.
196 int Symbol::index_of_at(int i, const char* str, int len) const {
197 assert(i >= 0 && i <= utf8_length(), "oob");
198 if (len <= 0) return 0;
199 char first_char = str[0];
200 address bytes = (address) ((Symbol*)this)->base();
201 address limit = bytes + utf8_length() - len; // inclusive limit
202 address scan = bytes + i;
203 if (scan > limit)
204 return -1;
205 for (; scan <= limit; scan++) {
206 scan = (address) memchr(scan, first_char, (limit + 1 - scan));
207 if (scan == NULL)
208 return -1; // not found
209 assert(scan >= bytes+i && scan <= limit, "scan oob");
210 if (memcmp(scan, str, len) == 0)
211 return (int)(scan - bytes);
212 }
213 return -1;
214 }
215
216
217 char* Symbol::as_C_string(char* buf, int size) const {
218 if (size > 0) {
219 int len = MIN2(size - 1, utf8_length());
220 for (int i = 0; i < len; i++) {
221 buf[i] = char_at(i);
222 }
223 buf[len] = '\0';
224 }
225 return buf;
226 }
227
228 char* Symbol::as_C_string() const {
229 int len = utf8_length();
230 char* str = NEW_RESOURCE_ARRAY(char, len + 1);
231 return as_C_string(str, len + 1);
232 }
233
234 void Symbol::print_utf8_on(outputStream* st) const {
235 st->print("%s", as_C_string());
236 }
237
238 void Symbol::print_symbol_on(outputStream* st) const {
239 char *s;
240 st = st ? st : tty;
241 {
242 // ResourceMark may not affect st->print(). If st is a string
243 // stream it could resize, using the same resource arena.
244 ResourceMark rm;
245 s = as_quoted_ascii();
246 s = os::strdup(s);
247 }
248 if (s == NULL) {
249 st->print("(null)");
250 } else {
251 st->print("%s", s);
252 os::free(s);
253 }
254 }
255
256 char* Symbol::as_quoted_ascii() const {
257 const char *ptr = (const char *)&_body[0];
258 int quoted_length = UTF8::quoted_ascii_length(ptr, utf8_length());
259 char* result = NEW_RESOURCE_ARRAY(char, quoted_length + 1);
260 UTF8::as_quoted_ascii(ptr, utf8_length(), result, quoted_length + 1);
261 return result;
262 }
263
264 jchar* Symbol::as_unicode(int& length) const {
265 Symbol* this_ptr = (Symbol*)this;
266 length = UTF8::unicode_length((char*)this_ptr->bytes(), utf8_length());
267 jchar* result = NEW_RESOURCE_ARRAY(jchar, length);
268 if (length > 0) {
269 UTF8::convert_to_unicode((char*)this_ptr->bytes(), result, length);
270 }
271 return result;
272 }
273
274 const char* Symbol::as_klass_external_name(char* buf, int size) const {
275 if (size > 0) {
276 char* str = as_C_string(buf, size);
277 int length = (int)strlen(str);
278 // Turn all '/'s into '.'s (also for array klasses)
279 for (int index = 0; index < length; index++) {
280 if (str[index] == '/') {
281 str[index] = '.';
282 }
283 }
284 return str;
285 } else {
286 return buf;
287 }
288 }
289
290 const char* Symbol::as_klass_external_name() const {
291 char* str = as_C_string();
292 int length = (int)strlen(str);
293 // Turn all '/'s into '.'s (also for array klasses)
294 for (int index = 0; index < length; index++) {
295 if (str[index] == '/') {
296 str[index] = '.';
297 }
298 }
299 return str;
300 }
301
302 static void print_class(outputStream *os, char *class_str, int len) {
303 for (int i = 0; i < len; ++i) {
304 if (class_str[i] == '/') {
305 os->put('.');
306 } else {
307 os->put(class_str[i]);
308 }
309 }
310 }
311
312 static void print_array(outputStream *os, char *array_str, int len) {
313 int dimensions = 0;
314 for (int i = 0; i < len; ++i) {
315 if (array_str[i] == '[') {
316 dimensions++;
317 } else if (array_str[i] == 'L' || array_str[i] == 'Q') {
318 // Expected format: L<type name>;. Skip 'L' and ';' delimiting the type name.
319 print_class(os, array_str+i+1, len-i-2);
320 break;
321 } else {
322 os->print("%s", type2name(char2type(array_str[i])));
323 }
324 }
325 for (int i = 0; i < dimensions; ++i) {
326 os->print("[]");
327 }
328 }
329
330 void Symbol::print_as_signature_external_return_type(outputStream *os) {
331 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
332 if (ss.at_return_type()) {
333 if (ss.is_array()) {
334 print_array(os, (char*)ss.raw_bytes(), (int)ss.raw_length());
335 } else if (ss.is_object()) {
336 // Expected format: L<type name>;. Skip 'L' and ';' delimiting the class name.
337 print_class(os, (char*)ss.raw_bytes()+1, (int)ss.raw_length()-2);
338 } else {
339 os->print("%s", type2name(ss.type()));
340 }
341 }
342 }
343 }
344
345 void Symbol::print_as_signature_external_parameters(outputStream *os) {
346 bool first = true;
347 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
348 if (ss.at_return_type()) break;
349 if (!first) { os->print(", "); }
350 if (ss.is_array()) {
351 print_array(os, (char*)ss.raw_bytes(), (int)ss.raw_length());
352 } else if (ss.is_object()) {
353 // Skip 'L' and ';'.
354 print_class(os, (char*)ss.raw_bytes()+1, (int)ss.raw_length()-2);
355 } else {
356 os->print("%s", type2name(ss.type()));
357 }
358 first = false;
359 }
360 }
361
362 // Increment refcount while checking for zero. If the Symbol's refcount becomes zero
363 // a thread could be concurrently removing the Symbol. This is used during SymbolTable
364 // lookup to avoid reviving a dead Symbol.
365 bool Symbol::try_increment_refcount() {
366 uint32_t found = _length_and_refcount;
367 while (true) {
368 uint32_t old_value = found;
369 int refc = extract_refcount(old_value);
370 if (refc == PERM_REFCOUNT) {
371 return true; // sticky max or created permanent
372 } else if (refc == 0) {
373 return false; // dead, can't revive.
374 } else {
375 found = Atomic::cmpxchg(old_value + 1, &_length_and_refcount, old_value);
376 if (found == old_value) {
377 return true; // successfully updated.
378 }
379 // refcount changed, try again.
380 }
381 }
382 }
383
384 // The increment_refcount() is called when not doing lookup. It is assumed that you
385 // have a symbol with a non-zero refcount and it can't become zero while referenced by
386 // this caller.
387 void Symbol::increment_refcount() {
388 if (!try_increment_refcount()) {
389 #ifdef ASSERT
390 print();
391 fatal("refcount has gone to zero");
392 #endif
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 = _length_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 #ifdef ASSERT
413 print();
414 fatal("refcount underflow");
415 #endif
416 return;
417 } else {
418 found = Atomic::cmpxchg(old_value - 1, &_length_and_refcount, old_value);
419 if (found == old_value) {
420 return; // successfully updated.
421 }
422 // refcount changed, try again.
423 }
424 }
425 }
426
427 void Symbol::make_permanent() {
428 uint32_t found = _length_and_refcount;
429 while (true) {
430 uint32_t old_value = found;
431 int refc = extract_refcount(old_value);
432 if (refc == PERM_REFCOUNT) {
433 return; // refcount is permanent, permanent is sticky
434 } else if (refc == 0) {
435 #ifdef ASSERT
436 print();
437 fatal("refcount underflow");
438 #endif
439 return;
440 } else {
441 int len = extract_length(old_value);
442 found = Atomic::cmpxchg(pack_length_and_refcount(len, PERM_REFCOUNT), &_length_and_refcount, old_value);
443 if (found == old_value) {
444 return; // successfully updated.
445 }
446 // refcount changed, try again.
447 }
448 }
449 }
450
451 void Symbol::metaspace_pointers_do(MetaspaceClosure* it) {
452 if (log_is_enabled(Trace, cds)) {
453 LogStream trace_stream(Log(cds)::trace());
454 trace_stream.print("Iter(Symbol): %p ", this);
455 print_value_on(&trace_stream);
456 trace_stream.cr();
457 }
458 }
459
460 void Symbol::print_on(outputStream* st) const {
461 st->print("Symbol: '");
462 print_symbol_on(st);
463 st->print("'");
464 st->print(" count %d", refcount());
465 }
466
467 void Symbol::print() const { print_on(tty); }
468
469 // The print_value functions are present in all builds, to support the
470 // disassembler and error reporting.
471 void Symbol::print_value_on(outputStream* st) const {
472 st->print("'");
473 for (int i = 0; i < utf8_length(); i++) {
474 st->print("%c", char_at(i));
475 }
476 st->print("'");
477 }
478
479 void Symbol::print_value() const { print_value_on(tty); }
480
481 bool Symbol::is_valid(Symbol* s) {
482 if (!is_aligned(s, sizeof(MetaWord))) return false;
483 if ((size_t)s < os::min_page_size()) return false;
484
485 if (!os::is_readable_range(s, s + 1)) return false;
486
487 // Symbols are not allocated in Java heap.
488 if (Universe::heap()->is_in_reserved(s)) return false;
489
490 int len = s->utf8_length();
491 if (len < 0) return false;
492
493 jbyte* bytes = (jbyte*) s->bytes();
494 return os::is_readable_range(bytes, bytes + len);
495 }
496
497 void Symbol::print_Qvalue_on(outputStream* st) const {
498 if (this == NULL) {
499 st->print("NULL");
500 } else {
501 st->print("'Q");
502 for (int i = 0; i < utf8_length(); i++) {
503 st->print("%c", char_at(i));
504 }
505 st->print(";'");
506 }
507 }
508
509 // SymbolTable prints this in its statistics
510 NOT_PRODUCT(size_t Symbol::_total_count = 0;)