1 /*
2 * Copyright (c) 1997, 2013, 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
46 // Class to hold a newly created or referenced Symbol* temporarily in scope.
47 // new_symbol() and lookup_and_add() will create a Symbol* if not already in the
48 // symbol table and add to the symbol's reference count.
49 // lookup_and_ignore_hash() and lookup_and_hash() will increment the refcount if symbol is found.
50 class TempNewSymbol : public StackObj {
51 Symbol* _temp;
52
53 public:
54 TempNewSymbol() : _temp(NULL) {}
55 // Creating or looking up a symbol increments the symbol's reference count
56 TempNewSymbol(Symbol *s) : _temp(s) {}
57
58 // Operator= increments reference count.
59 void operator=(const TempNewSymbol &s) {
60 //clear(); //FIXME
61 _temp = s._temp;
62 if (_temp !=NULL) _temp->increment_refcount();
63 }
64
65 // Decrement reference counter so it can go away if it's unique
66 void clear() { if (_temp != NULL) _temp->decrement_refcount(); _temp = NULL; }
67
68 ~TempNewSymbol() { clear(); }
69
70 // Operators so they can be used like Symbols
71 Symbol* operator -> () const { return _temp; }
72 bool operator == (Symbol* o) const { return _temp == o; }
73 // Sneaky conversion function
74 operator Symbol*() { return _temp; }
75 };
76
77 class SymbolTable : public Hashtable<Symbol*, mtSymbol> {
78 friend class VMStructs;
79 friend class ClassFileParser;
80
81 private:
82 // The symbol table
83 static SymbolTable* _the_table;
84
85 // Set if one bucket is out of balance due to hash algorithm deficiency
86 static bool _needs_rehashing;
87
88 // For statistics
89 static int _symbols_removed;
90 static int _symbols_counted;
91
92 Symbol* allocate_symbol(const u1* name, int len, bool c_heap, TRAPS); // Assumes no characters larger than 0x7F
93
94 // Adding elements
95 Symbol* basic_add(int index, u1* name, int len, unsigned int hashValue,
96 bool c_heap, TRAPS);
97 bool basic_add(ClassLoaderData* loader_data,
98 constantPoolHandle cp, int names_count,
99 const char** names, int* lengths, int* cp_indices,
100 unsigned int* hashValues, TRAPS);
101
102 static void new_symbols(ClassLoaderData* loader_data,
103 constantPoolHandle cp, int names_count,
104 const char** name, int* lengths,
105 int* cp_indices, unsigned int* hashValues,
106 TRAPS) {
107 add(loader_data, cp, names_count, name, lengths, cp_indices, hashValues, THREAD);
108 }
109
110 Symbol* lookup(int index, const char* name, int len, unsigned int hash);
111
112 SymbolTable()
113 : Hashtable<Symbol*, mtSymbol>(SymbolTableSize, sizeof (HashtableEntry<Symbol*, mtSymbol>)) {}
114
115 SymbolTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
116 : Hashtable<Symbol*, mtSymbol>(SymbolTableSize, sizeof (HashtableEntry<Symbol*, mtSymbol>), t,
117 number_of_entries) {}
118
119 // Arena for permanent symbols (null class loader) that are never unloaded
120 static Arena* _arena;
121 static Arena* arena() { return _arena; } // called for statistics
122
123 static void initialize_symbols(int arena_alloc_size = 0);
124
125 static volatile int _parallel_claimed_idx;
126
127 // Release any dead symbols
128 static void buckets_unlink(int start_idx, int end_idx, int* processed, int* removed, size_t* memory_total);
129 public:
130 enum {
131 symbol_alloc_batch_size = 8,
132 // Pick initial size based on java -version size measurements
133 symbol_alloc_arena_size = 360*K
134 };
135
136 // The symbol table
137 static SymbolTable* the_table() { return _the_table; }
138
139 // Size of one bucket in the string table. Used when checking for rollover.
140 static uint bucket_size() { return sizeof(HashtableBucket<mtSymbol>); }
141
142 static void create_table() {
143 assert(_the_table == NULL, "One symbol table allowed.");
144 _the_table = new SymbolTable();
145 initialize_symbols(symbol_alloc_arena_size);
146 }
147
148 static void create_table(HashtableBucket<mtSymbol>* t, int length,
149 int number_of_entries) {
150 assert(_the_table == NULL, "One symbol table allowed.");
151
152 // If CDS archive used a different symbol table size, use that size instead
153 // which is better than giving an error.
154 SymbolTableSize = length/bucket_size();
155
156 _the_table = new SymbolTable(t, number_of_entries);
157 // if CDS give symbol table a default arena size since most symbols
158 // are already allocated in the shared misc section.
159 initialize_symbols();
160 }
161
162 static unsigned int hash_symbol(const char* s, int len);
163
164 // lookup only, won't add. Also calculate hash.
165 static Symbol* lookup_and_hash(const char* name, int len, unsigned int& hash);
166 // lookup and add if lookup failed
167 static Symbol* lookup_and_add(const char* name, int len, TRAPS);
168 // Only copy to C string to be added if lookup failed.
169 static Symbol* lookup_and_add(const Symbol* sym, int begin, int end, TRAPS);
170
171 static void release(Symbol* sym);
172
173 // Look up the address of the literal in the SymbolTable for this Symbol*
174 static Symbol** lookup_symbol_addr(Symbol* sym);
175
176 // jchar (utf16) version of lookups
177 static Symbol* lookup_and_add_unicode(const jchar* name, int len, TRAPS);
178 static Symbol* lookup_and_hash_unicode(const jchar* name, int len, unsigned int& hash);
179
180 static void add(ClassLoaderData* loader_data,
181 constantPoolHandle cp, int names_count,
182 const char** names, int* lengths, int* cp_indices,
183 unsigned int* hashValues, TRAPS);
184
185 // Release any dead symbols
186 static void unlink() {
187 int processed = 0;
188 int removed = 0;
189 unlink(&processed, &removed);
190 }
191 static void unlink(int* processed, int* removed);
192 // Release any dead symbols, possibly parallel version
193 static void possibly_parallel_unlink(int* processed, int* removed);
194
195 // iterate over symbols
196 static void symbols_do(SymbolClosure *cl);
197
198 // Symbol creation
199 static Symbol* new_symbol(const char* utf8_buffer, int length, TRAPS) {
200 assert(utf8_buffer != NULL, "just checking");
201 return lookup_and_add(utf8_buffer, length, THREAD);
202 }
203 static Symbol* new_symbol(const char* name, TRAPS) {
204 return new_symbol(name, (int)strlen(name), THREAD);
205 }
206 static Symbol* new_symbol(const Symbol* sym, int begin, int end, TRAPS) {
207 assert(begin <= end && end <= sym->utf8_length(), "just checking");
208 return lookup_and_add(sym, begin, end, THREAD);
209 }
210
211 // Create a symbol in the arena for symbols that are not deleted
212 static Symbol* new_permanent_symbol(const char* name, TRAPS);
213
214 // Symbol lookup
215 static Symbol* lookup(int index, const char* name, int len, TRAPS);
216
217 // Needed for preloading classes in signatures when compiling.
218 // Returns the symbol is already present in symbol table, otherwise
219 // NULL. NO ALLOCATION IS GUARANTEED!
220 static Symbol* lookup_and_ignore_hash(const char* name, int len) {
221 unsigned int ignore_hash;
222 return lookup_and_hash(name, len, ignore_hash);
223 }
224 static Symbol* lookup_and_ignore_hash_unicode(const jchar* name, int len) {
225 unsigned int ignore_hash;
226 return lookup_and_hash_unicode(name, len, ignore_hash);
227 }
228
229 // Histogram
230 static void print_histogram() PRODUCT_RETURN;
231 static void print() PRODUCT_RETURN;
232
233 // Debugging
234 static void verify();
235 static void dump(outputStream* st);
236
237 // Sharing
238 static void copy_buckets(char** top, char*end) {
239 the_table()->Hashtable<Symbol*, mtSymbol>::copy_buckets(top, end);
240 }
241 static void copy_table(char** top, char*end) {
242 the_table()->Hashtable<Symbol*, mtSymbol>::copy_table(top, end);
243 }
244 static void reverse(void* boundary = NULL) {
245 the_table()->Hashtable<Symbol*, mtSymbol>::reverse(boundary);
246 }
247
248 // Rehash the symbol table if it gets out of balance
249 static void rehash_table();
250 static bool needs_rehashing() { return _needs_rehashing; }
251 // Parallel chunked scanning
252 static void clear_parallel_claimed_index() { _parallel_claimed_idx = 0; }
253 static int parallel_claimed_index() { return _parallel_claimed_idx; }
254 };
255
256 class StringTable : public Hashtable<oop, mtSymbol> {
257 friend class VMStructs;
258
259 private:
260 // The string table
261 static StringTable* _the_table;
262
263 // Set if one bucket is out of balance due to hash algorithm deficiency
264 static bool _needs_rehashing;
265
266 // Claimed high water mark for parallel chunked scanning
267 static volatile int _parallel_claimed_idx;
268
269 static oop intern(Handle string_or_null, jchar* chars, int length, TRAPS);
270 oop basic_add(int index, Handle string_or_null, jchar* name, int len,
271 unsigned int hashValue, TRAPS);
272
273 oop lookup(int index, jchar* chars, int length, unsigned int hashValue);
274
275 // Apply the give oop closure to the entries to the buckets
276 // in the range [start_idx, end_idx).
277 static void buckets_oops_do(OopClosure* f, int start_idx, int end_idx);
278 // Unlink or apply the give oop closure to the entries to the buckets
279 // in the range [start_idx, end_idx).
280 static void buckets_unlink_or_oops_do(BoolObjectClosure* is_alive, OopClosure* f, int start_idx, int end_idx, int* processed, int* removed);
281
282 StringTable() : Hashtable<oop, mtSymbol>((int)StringTableSize,
283 sizeof (HashtableEntry<oop, mtSymbol>)) {}
284
285 StringTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
286 : Hashtable<oop, mtSymbol>((int)StringTableSize, sizeof (HashtableEntry<oop, mtSymbol>), t,
287 number_of_entries) {}
288 public:
289 // The string table
290 static StringTable* the_table() { return _the_table; }
291
292 // Size of one bucket in the string table. Used when checking for rollover.
293 static uint bucket_size() { return sizeof(HashtableBucket<mtSymbol>); }
294
295 static void create_table() {
296 assert(_the_table == NULL, "One string table allowed.");
297 _the_table = new StringTable();
298 }
299
300 // GC support
301 // Delete pointers to otherwise-unreachable objects.
302 static void unlink_or_oops_do(BoolObjectClosure* cl, OopClosure* f) {
303 int processed = 0;
304 int removed = 0;
305 unlink_or_oops_do(cl, f, &processed, &removed);
306 }
307 static void unlink(BoolObjectClosure* cl) {
308 int processed = 0;
309 int removed = 0;
310 unlink_or_oops_do(cl, NULL, &processed, &removed);
311 }
312 static void unlink_or_oops_do(BoolObjectClosure* cl, OopClosure* f, int* processed, int* removed);
313 static void unlink(BoolObjectClosure* cl, int* processed, int* removed) {
314 unlink_or_oops_do(cl, NULL, processed, removed);
315 }
316 // Serially invoke "f->do_oop" on the locations of all oops in the table.
317 static void oops_do(OopClosure* f);
318
319 // Possibly parallel versions of the above
320 static void possibly_parallel_unlink_or_oops_do(BoolObjectClosure* cl, OopClosure* f, int* processed, int* removed);
321 static void possibly_parallel_unlink(BoolObjectClosure* cl, int* processed, int* removed) {
322 possibly_parallel_unlink_or_oops_do(cl, NULL, processed, removed);
323 }
324 static void possibly_parallel_oops_do(OopClosure* f);
325
326 // Hashing algorithm, used as the hash value used by the
327 // StringTable for bucket selection and comparison (stored in the
328 // HashtableEntry structures). This is used in the String.intern() method.
329 static unsigned int hash_string(const jchar* s, int len);
330
331 // Internal test.
332 static void test_alt_hash() PRODUCT_RETURN;
333
334 // Probing
335 static oop lookup(Symbol* symbol);
336 static oop lookup(jchar* chars, int length);
337
338 // Interning
339 static oop intern(Symbol* symbol, TRAPS);
340 static oop intern(oop string, TRAPS);
341 static oop intern(const char *utf8_string, TRAPS);
342
343 // Debugging
344 static void verify();
345 static void dump(outputStream* st);
346
347 enum VerifyMesgModes {
348 _verify_quietly = 0,
349 _verify_with_mesgs = 1
350 };
351
352 enum VerifyRetTypes {
353 _verify_pass = 0,
354 _verify_fail_continue = 1,
355 _verify_fail_done = 2
356 };
357
358 static VerifyRetTypes compare_entries(int bkt1, int e_cnt1,
359 HashtableEntry<oop, mtSymbol>* e_ptr1,
360 int bkt2, int e_cnt2,
361 HashtableEntry<oop, mtSymbol>* e_ptr2);
362 static VerifyRetTypes verify_entry(int bkt, int e_cnt,
363 HashtableEntry<oop, mtSymbol>* e_ptr,
364 VerifyMesgModes mesg_mode);
365 static int verify_and_compare_entries();
366
367 // Sharing
368 static void copy_buckets(char** top, char*end) {
369 the_table()->Hashtable<oop, mtSymbol>::copy_buckets(top, end);
370 }
371 static void copy_table(char** top, char*end) {
372 the_table()->Hashtable<oop, mtSymbol>::copy_table(top, end);
373 }
374 static void reverse() {
375 the_table()->Hashtable<oop, mtSymbol>::reverse();
376 }
377
378 // Rehash the symbol table if it gets out of balance
379 static void rehash_table();
380 static bool needs_rehashing() { return _needs_rehashing; }
381
382 // Parallel chunked scanning
383 static void clear_parallel_claimed_index() { _parallel_claimed_idx = 0; }
384 static int parallel_claimed_index() { return _parallel_claimed_idx; }
385 };
386 #endif // SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP