1 /*
2 * Copyright (c) 1997, 2015, 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 #ifndef SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP
26 #define SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP
27
28 #include "memory/allocation.inline.hpp"
29 #include "oops/symbol.hpp"
30 #include "utilities/hashtable.hpp"
31
32 // The symbol table holds all Symbol*s and corresponding interned strings.
33 // Symbol*s and literal strings should be canonicalized.
34 //
35 // The interned strings are created lazily.
36 //
37 // It is implemented as an open hash table with a fixed number of buckets.
38 //
39 // %note:
40 // - symbolTableEntrys are allocated in blocks to reduce the space overhead.
41
42 class BoolObjectClosure;
43 class outputStream;
44
45 // TempNewSymbol acts as a handle class in a handle/body idiom and is
46 // responsible for proper resource management of the body (which is a Symbol*).
47 // The body is resource managed by a reference counting scheme.
48 // TempNewSymbol can therefore be used to properly hold a newly created or referenced
49 // Symbol* temporarily in scope.
50 //
51 // Routines in SymbolTable will initialize the reference count of a Symbol* before
52 // it becomes "managed" by TempNewSymbol instances. As a handle class, TempNewSymbol
53 // needs to maintain proper reference counting in context of copy semantics.
54 //
55 // In SymbolTable, new_symbol() and lookup() will create a Symbol* if not already in the
56 // symbol table and add to the symbol's reference count.
57 // probe() and lookup_only() will increment the refcount if symbol is found.
58 class TempNewSymbol : public StackObj {
59 Symbol* _temp;
60
61 public:
62 TempNewSymbol() : _temp(NULL) {}
63
64 // Conversion from a Symbol* to a TempNewSymbol.
65 // Does not increment the current reference count.
66 TempNewSymbol(Symbol *s) : _temp(s) {}
67
68 // Copy constructor increments reference count.
69 TempNewSymbol(const TempNewSymbol& rhs) {
70 _temp = rhs._temp;
71 if (_temp != NULL) {
72 _temp->increment_refcount();
73 }
74 }
75
76 // Assignment operator decrements the reference count for any
77 // existing resource. It increments the reference count
78 // for the newly assigned resource.
79 const TempNewSymbol& operator=(const TempNewSymbol &rhs) {
80 if (this != &rhs && _temp != rhs._temp) {
81 clear();
82 _temp = rhs._temp;
83 if (_temp != NULL) {
84 _temp->increment_refcount();
85 }
86 }
87 return *this;
88 }
89
90 // Decrement reference counter so it can go away if it's unique
91 void clear() {
92 if (_temp != NULL) {
93 _temp->decrement_refcount();
94 }
95 }
96
97 ~TempNewSymbol() { clear(); }
98
99 // Symbol* conversion operators
100 Symbol* operator -> () const { return _temp; }
101 bool operator == (Symbol* o) const { return _temp == o; }
102 operator Symbol*() { return _temp; }
103 };
104
105 template <class T, class N> class CompactHashtable;
106
107 class SymbolTable : public RehashableHashtable<Symbol*, mtSymbol> {
108 friend class VMStructs;
109 friend class ClassFileParser;
110
111 private:
112 // The symbol table
113 static SymbolTable* _the_table;
114
115 // Set if one bucket is out of balance due to hash algorithm deficiency
116 static bool _needs_rehashing;
117 static bool _lookup_shared_first;
118
119 // For statistics
120 static int _symbols_removed;
121 static int _symbols_counted;
122
123 // shared symbol table.
124 static CompactHashtable<Symbol*, char> _shared_table;
125
126 Symbol* allocate_symbol(const u1* name, int len, bool c_heap, TRAPS); // Assumes no characters larger than 0x7F
127
128 // Adding elements
129 Symbol* basic_add(int index, u1* name, int len, unsigned int hashValue,
130 bool c_heap, TRAPS);
131 bool basic_add(ClassLoaderData* loader_data,
132 const constantPoolHandle& cp, int names_count,
133 const char** names, int* lengths, int* cp_indices,
134 unsigned int* hashValues, TRAPS);
135
136 static void new_symbols(ClassLoaderData* loader_data,
137 const constantPoolHandle& cp, int names_count,
138 const char** name, int* lengths,
139 int* cp_indices, unsigned int* hashValues,
140 TRAPS) {
141 add(loader_data, cp, names_count, name, lengths, cp_indices, hashValues, THREAD);
142 }
143
144 static Symbol* lookup_shared(const char* name, int len, unsigned int hash);
145 Symbol* lookup_dynamic(int index, const char* name, int len, unsigned int hash);
146 Symbol* lookup(int index, const char* name, int len, unsigned int hash);
147
148 SymbolTable()
149 : RehashableHashtable<Symbol*, mtSymbol>(SymbolTableSize, sizeof (HashtableEntry<Symbol*, mtSymbol>)) {}
150
151 SymbolTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
152 : RehashableHashtable<Symbol*, mtSymbol>(SymbolTableSize, sizeof (HashtableEntry<Symbol*, mtSymbol>), t,
153 number_of_entries) {}
154
155 // Arena for permanent symbols (null class loader) that are never unloaded
156 static Arena* _arena;
157 static Arena* arena() { return _arena; } // called for statistics
158
159 static void initialize_symbols(int arena_alloc_size = 0);
160
161 static volatile int _parallel_claimed_idx;
162
163 // Release any dead symbols
164 static void buckets_unlink(int start_idx, int end_idx, int* processed, int* removed);
165 public:
166 enum {
167 symbol_alloc_batch_size = 8,
168 // Pick initial size based on java -version size measurements
169 symbol_alloc_arena_size = 360*K
170 };
171
172 // The symbol table
173 static SymbolTable* the_table() { return _the_table; }
174
175 // Size of one bucket in the string table. Used when checking for rollover.
176 static uint bucket_size() { return sizeof(HashtableBucket<mtSymbol>); }
177
178 static void create_table() {
179 assert(_the_table == NULL, "One symbol table allowed.");
180 _the_table = new SymbolTable();
181 initialize_symbols(symbol_alloc_arena_size);
182 }
183
184 static unsigned int hash_symbol(const char* s, int len);
185
186 static Symbol* lookup(const char* name, int len, TRAPS);
187 // lookup only, won't add. Also calculate hash.
188 static Symbol* lookup_only(const char* name, int len, unsigned int& hash);
189 // Only copy to C string to be added if lookup failed.
190 static Symbol* lookup(const Symbol* sym, int begin, int end, TRAPS);
191
192 static void release(Symbol* sym);
193
194 // Look up the address of the literal in the SymbolTable for this Symbol*
195 static Symbol** lookup_symbol_addr(Symbol* sym);
196
197 // jchar (UTF16) version of lookups
198 static Symbol* lookup_unicode(const jchar* name, int len, TRAPS);
199 static Symbol* lookup_only_unicode(const jchar* name, int len, unsigned int& hash);
200
201 static void add(ClassLoaderData* loader_data,
202 const constantPoolHandle& cp, int names_count,
203 const char** names, int* lengths, int* cp_indices,
204 unsigned int* hashValues, TRAPS);
205
206 // Release any dead symbols
207 static void unlink() {
208 int processed = 0;
209 int removed = 0;
210 unlink(&processed, &removed);
211 }
212 static void unlink(int* processed, int* removed);
213 // Release any dead symbols, possibly parallel version
214 static void possibly_parallel_unlink(int* processed, int* removed);
215
216 // iterate over symbols
217 static void symbols_do(SymbolClosure *cl);
218
219 // Symbol creation
220 static Symbol* new_symbol(const char* utf8_buffer, int length, TRAPS) {
221 assert(utf8_buffer != NULL, "just checking");
222 return lookup(utf8_buffer, length, THREAD);
223 }
224 static Symbol* new_symbol(const char* name, TRAPS) {
225 return new_symbol(name, (int)strlen(name), THREAD);
226 }
227 static Symbol* new_symbol(const Symbol* sym, int begin, int end, TRAPS) {
228 assert(begin <= end && end <= sym->utf8_length(), "just checking");
229 return lookup(sym, begin, end, THREAD);
230 }
231
232 // Create a symbol in the arena for symbols that are not deleted
233 static Symbol* new_permanent_symbol(const char* name, TRAPS);
234
235 // Symbol lookup
236 static Symbol* lookup(int index, const char* name, int len, TRAPS);
237
238 // Needed for preloading classes in signatures when compiling.
239 // Returns the symbol is already present in symbol table, otherwise
240 // NULL. NO ALLOCATION IS GUARANTEED!
241 static Symbol* probe(const char* name, int len) {
242 unsigned int ignore_hash;
243 return lookup_only(name, len, ignore_hash);
244 }
245 static Symbol* probe_unicode(const jchar* name, int len) {
246 unsigned int ignore_hash;
247 return lookup_only_unicode(name, len, ignore_hash);
248 }
249
250 // Histogram
251 static void print_histogram() PRODUCT_RETURN;
252 static void print() PRODUCT_RETURN;
253
254 // Debugging
255 static void verify();
256 static void dump(outputStream* st, bool verbose=false);
257 static void read(const char* filename, TRAPS);
258
259 // Sharing
260 static bool copy_compact_table(char** top, char* end);
261 static const char* init_shared_table(const char* buffer);
262
263 // Rehash the symbol table if it gets out of balance
264 static void rehash_table();
265 static bool needs_rehashing() { return _needs_rehashing; }
266 // Parallel chunked scanning
267 static void clear_parallel_claimed_index() { _parallel_claimed_idx = 0; }
268 static int parallel_claimed_index() { return _parallel_claimed_idx; }
269 };
270
271 #endif // SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP