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 // ------------------------------------------------------------------
82 // Symbol::index_of
83 //
84 // Finds if the given string is a substring of this symbol's utf8 bytes.
85 // Return -1 on failure. Otherwise return the first index where str occurs.
86 int Symbol::index_of_at(int i, const char* str, int len) const {
87 assert(i >= 0 && i <= utf8_length(), "oob");
88 if (len <= 0) return 0;
89 char first_char = str[0];
90 address bytes = (address) ((Symbol*)this)->base();
91 address limit = bytes + utf8_length() - len; // inclusive limit
92 address scan = bytes + i;
93 if (scan > limit)
94 return -1;
95 for (; scan <= limit; scan++) {
96 scan = (address) memchr(scan, first_char, (limit + 1 - scan));
97 if (scan == NULL)
98 return -1; // not found
99 assert(scan >= bytes+i && scan <= limit, "scan oob");
100 if (len <= 2
101 ? (char) scan[len-1] == str[len-1]
102 : memcmp(scan+1, str+1, len-1) == 0) {
103 return (int)(scan - bytes);
104 }
105 }
106 return -1;
107 }
108
109
110 char* Symbol::as_C_string(char* buf, int size) const {
111 if (size > 0) {
112 int len = MIN2(size - 1, utf8_length());
113 for (int i = 0; i < len; i++) {
114 buf[i] = char_at(i);
115 }
116 buf[len] = '\0';
117 }
118 return buf;
119 }
120
121 char* Symbol::as_C_string() const {
122 int len = utf8_length();
123 char* str = NEW_RESOURCE_ARRAY(char, len + 1);
124 return as_C_string(str, len + 1);
125 }
126
127 void Symbol::print_utf8_on(outputStream* st) const {
128 st->print("%s", as_C_string());
129 }
130
131 void Symbol::print_symbol_on(outputStream* st) const {
132 char *s;
133 st = st ? st : tty;
134 {
135 // ResourceMark may not affect st->print(). If st is a string
136 // stream it could resize, using the same resource arena.
137 ResourceMark rm;
138 s = as_quoted_ascii();
139 s = os::strdup(s);
140 }
141 if (s == NULL) {
142 st->print("(null)");
143 } else {
144 st->print("%s", s);
145 os::free(s);
146 }
147 }
148
149 char* Symbol::as_quoted_ascii() const {
150 const char *ptr = (const char *)&_body[0];
151 int quoted_length = UTF8::quoted_ascii_length(ptr, utf8_length());
152 char* result = NEW_RESOURCE_ARRAY(char, quoted_length + 1);
153 UTF8::as_quoted_ascii(ptr, utf8_length(), result, quoted_length + 1);
154 return result;
155 }
156
157 jchar* Symbol::as_unicode(int& length) const {
158 Symbol* this_ptr = (Symbol*)this;
159 length = UTF8::unicode_length((char*)this_ptr->bytes(), utf8_length());
160 jchar* result = NEW_RESOURCE_ARRAY(jchar, length);
161 if (length > 0) {
162 UTF8::convert_to_unicode((char*)this_ptr->bytes(), result, length);
163 }
164 return result;
165 }
166
167 const char* Symbol::as_klass_external_name(char* buf, int size) const {
168 if (size > 0) {
169 char* str = as_C_string(buf, size);
170 int length = (int)strlen(str);
171 // Turn all '/'s into '.'s (also for array klasses)
172 for (int index = 0; index < length; index++) {
173 if (str[index] == JVM_SIGNATURE_SLASH) {
174 str[index] = JVM_SIGNATURE_DOT;
175 }
176 }
177 return str;
178 } else {
179 return buf;
180 }
181 }
182
183 const char* Symbol::as_klass_external_name() const {
184 char* str = as_C_string();
185 int length = (int)strlen(str);
186 // Turn all '/'s into '.'s (also for array klasses)
187 for (int index = 0; index < length; index++) {
188 if (str[index] == JVM_SIGNATURE_SLASH) {
189 str[index] = JVM_SIGNATURE_DOT;
190 }
191 }
192 return str;
193 }
194
195 static void print_class(outputStream *os, const SignatureStream& ss) {
196 int sb = ss.raw_symbol_begin(), se = ss.raw_symbol_end();
197 for (int i = sb; i < se; ++i) {
198 int ch = ss.raw_char_at(i);
199 if (ch == JVM_SIGNATURE_SLASH) {
200 os->put(JVM_SIGNATURE_DOT);
201 } else {
202 os->put(ch);
203 }
204 }
205 }
206
207 static void print_array(outputStream *os, SignatureStream& ss) {
208 int dimensions = ss.skip_array_prefix();
209 assert(dimensions > 0, "");
210 if (ss.is_reference()) {
211 print_class(os, ss);
212 } else {
213 os->print("%s", type2name(ss.type()));
214 }
215 for (int i = 0; i < dimensions; ++i) {
216 os->print("[]");
217 }
218 }
219
220 void Symbol::print_as_signature_external_return_type(outputStream *os) {
221 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
222 if (ss.at_return_type()) {
223 if (ss.is_array()) {
224 print_array(os, ss);
225 } else if (ss.is_reference()) {
226 print_class(os, ss);
227 } else {
228 os->print("%s", type2name(ss.type()));
229 }
230 }
231 }
232 }
233
234 void Symbol::print_as_signature_external_parameters(outputStream *os) {
235 bool first = true;
236 for (SignatureStream ss(this); !ss.is_done(); ss.next()) {
237 if (ss.at_return_type()) break;
238 if (!first) { os->print(", "); }
239 if (ss.is_array()) {
240 print_array(os, ss);
241 } else if (ss.is_reference()) {
242 print_class(os, ss);
243 } else {
244 os->print("%s", type2name(ss.type()));
245 }
246 first = false;
247 }
248 }
249
250 // Increment refcount while checking for zero. If the Symbol's refcount becomes zero
251 // a thread could be concurrently removing the Symbol. This is used during SymbolTable
252 // lookup to avoid reviving a dead Symbol.
253 bool Symbol::try_increment_refcount() {
254 uint32_t found = _hash_and_refcount;
255 while (true) {
256 uint32_t old_value = found;
257 int refc = extract_refcount(old_value);
258 if (refc == PERM_REFCOUNT) {
259 return true; // sticky max or created permanent
260 } else if (refc == 0) {
261 return false; // dead, can't revive.
262 } else {
263 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value + 1);
264 if (found == old_value) {
265 return true; // successfully updated.
266 }
267 // refcount changed, try again.
268 }
269 }
270 }
271
272 // The increment_refcount() is called when not doing lookup. It is assumed that you
273 // have a symbol with a non-zero refcount and it can't become zero while referenced by
274 // this caller.
275 void Symbol::increment_refcount() {
276 if (!try_increment_refcount()) {
277 #ifdef ASSERT
278 print();
279 fatal("refcount has gone to zero");
280 #endif
281 }
282 #ifndef PRODUCT
283 if (refcount() != PERM_REFCOUNT) { // not a permanent symbol
284 NOT_PRODUCT(Atomic::inc(&_total_count);)
285 }
286 #endif
287 }
288
289 // Decrement refcount potentially while racing increment, so we need
290 // to check the value after attempting to decrement so that if another
291 // thread increments to PERM_REFCOUNT the value is not decremented.
292 void Symbol::decrement_refcount() {
293 uint32_t found = _hash_and_refcount;
294 while (true) {
295 uint32_t old_value = found;
296 int refc = extract_refcount(old_value);
297 if (refc == PERM_REFCOUNT) {
298 return; // refcount is permanent, permanent is sticky
299 } else if (refc == 0) {
300 #ifdef ASSERT
301 print();
302 fatal("refcount underflow");
303 #endif
304 return;
305 } else {
306 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, old_value - 1);
307 if (found == old_value) {
308 return; // successfully updated.
309 }
310 // refcount changed, try again.
311 }
312 }
313 }
314
315 void Symbol::make_permanent() {
316 uint32_t found = _hash_and_refcount;
317 while (true) {
318 uint32_t old_value = found;
319 int refc = extract_refcount(old_value);
320 if (refc == PERM_REFCOUNT) {
321 return; // refcount is permanent, permanent is sticky
322 } else if (refc == 0) {
323 #ifdef ASSERT
324 print();
325 fatal("refcount underflow");
326 #endif
327 return;
328 } else {
329 int hash = extract_hash(old_value);
330 found = Atomic::cmpxchg(&_hash_and_refcount, old_value, pack_hash_and_refcount(hash, PERM_REFCOUNT));
331 if (found == old_value) {
332 return; // successfully updated.
333 }
334 // refcount changed, try again.
335 }
336 }
337 }
338
339 void Symbol::metaspace_pointers_do(MetaspaceClosure* it) {
340 if (log_is_enabled(Trace, cds)) {
341 LogStream trace_stream(Log(cds)::trace());
342 trace_stream.print("Iter(Symbol): %p ", this);
343 print_value_on(&trace_stream);
344 trace_stream.cr();
345 }
346 }
347
348 void Symbol::print_on(outputStream* st) const {
349 st->print("Symbol: '");
350 print_symbol_on(st);
351 st->print("'");
352 st->print(" count %d", refcount());
353 }
354
355 void Symbol::print() const { print_on(tty); }
356
357 // The print_value functions are present in all builds, to support the
358 // disassembler and error reporting.
359 void Symbol::print_value_on(outputStream* st) const {
360 st->print("'");
361 for (int i = 0; i < utf8_length(); i++) {
362 st->print("%c", char_at(i));
363 }
364 st->print("'");
365 }
366
367 void Symbol::print_value() const { print_value_on(tty); }
368
369 bool Symbol::is_valid(Symbol* s) {
370 if (!is_aligned(s, sizeof(MetaWord))) return false;
371 if ((size_t)s < os::min_page_size()) return false;
372
373 if (!os::is_readable_range(s, s + 1)) return false;
374
375 // Symbols are not allocated in Java heap.
376 if (Universe::heap()->is_in(s)) return false;
377
378 int len = s->utf8_length();
379 if (len < 0) return false;
380
381 jbyte* bytes = (jbyte*) s->bytes();
382 return os::is_readable_range(bytes, bytes + len);
383 }
384
385 // SymbolTable prints this in its statistics
386 NOT_PRODUCT(size_t Symbol::_total_count = 0;)