Tue, 08 Oct 2013 09:33:51 +0100
8019375: Internal symbol table size should be tunable.
Reviewed-by: coleenp, kamg
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 */
25 #ifndef SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP
26 #define SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP
28 #include "memory/allocation.inline.hpp"
29 #include "oops/symbol.hpp"
30 #include "utilities/hashtable.hpp"
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.
42 class BoolObjectClosure;
43 class outputStream;
46 // Class to hold a newly created or referenced Symbol* temporarily in scope.
47 // new_symbol() and lookup() will create a Symbol* if not already in the
48 // symbol table and add to the symbol's reference count.
49 // probe() and lookup_only() will increment the refcount if symbol is found.
50 class TempNewSymbol : public StackObj {
51 Symbol* _temp;
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) {}
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 }
65 // Decrement reference counter so it can go away if it's unique
66 void clear() { if (_temp != NULL) _temp->decrement_refcount(); _temp = NULL; }
68 ~TempNewSymbol() { clear(); }
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 };
77 class SymbolTable : public Hashtable<Symbol*, mtSymbol> {
78 friend class VMStructs;
79 friend class ClassFileParser;
81 private:
82 // The symbol table
83 static SymbolTable* _the_table;
85 // Set if one bucket is out of balance due to hash algorithm deficiency
86 static bool _needs_rehashing;
88 // For statistics
89 static int symbols_removed;
90 static int symbols_counted;
92 Symbol* allocate_symbol(const u1* name, int len, bool c_heap, TRAPS); // Assumes no characters larger than 0x7F
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);
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 }
110 Symbol* lookup(int index, const char* name, int len, unsigned int hash);
112 SymbolTable()
113 : Hashtable<Symbol*, mtSymbol>(SymbolTableSize, sizeof (HashtableEntry<Symbol*, mtSymbol>)) {}
115 SymbolTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
116 : Hashtable<Symbol*, mtSymbol>(SymbolTableSize, sizeof (HashtableEntry<Symbol*, mtSymbol>), t,
117 number_of_entries) {}
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
123 static void initialize_symbols(int arena_alloc_size = 0);
124 public:
125 enum {
126 symbol_alloc_batch_size = 8,
127 // Pick initial size based on java -version size measurements
128 symbol_alloc_arena_size = 360*K
129 };
131 // The symbol table
132 static SymbolTable* the_table() { return _the_table; }
134 // Size of one bucket in the string table. Used when checking for rollover.
135 static uint bucket_size() { return sizeof(HashtableBucket<mtSymbol>); }
137 static void create_table() {
138 assert(_the_table == NULL, "One symbol table allowed.");
139 _the_table = new SymbolTable();
140 initialize_symbols(symbol_alloc_arena_size);
141 }
143 static void create_table(HashtableBucket<mtSymbol>* t, int length,
144 int number_of_entries) {
145 assert(_the_table == NULL, "One symbol table allowed.");
147 // If CDS archive used a different symbol table size, use that size instead
148 // which is better than giving an error.
149 SymbolTableSize = length/bucket_size();
151 _the_table = new SymbolTable(t, number_of_entries);
152 // if CDS give symbol table a default arena size since most symbols
153 // are already allocated in the shared misc section.
154 initialize_symbols();
155 }
157 static unsigned int hash_symbol(const char* s, int len);
159 static Symbol* lookup(const char* name, int len, TRAPS);
160 // lookup only, won't add. Also calculate hash.
161 static Symbol* lookup_only(const char* name, int len, unsigned int& hash);
162 // Only copy to C string to be added if lookup failed.
163 static Symbol* lookup(const Symbol* sym, int begin, int end, TRAPS);
165 static void release(Symbol* sym);
167 // Look up the address of the literal in the SymbolTable for this Symbol*
168 static Symbol** lookup_symbol_addr(Symbol* sym);
170 // jchar (utf16) version of lookups
171 static Symbol* lookup_unicode(const jchar* name, int len, TRAPS);
172 static Symbol* lookup_only_unicode(const jchar* name, int len, unsigned int& hash);
174 static void add(ClassLoaderData* loader_data,
175 constantPoolHandle cp, int names_count,
176 const char** names, int* lengths, int* cp_indices,
177 unsigned int* hashValues, TRAPS);
179 // Release any dead symbols
180 static void unlink();
182 // iterate over symbols
183 static void symbols_do(SymbolClosure *cl);
185 // Symbol creation
186 static Symbol* new_symbol(const char* utf8_buffer, int length, TRAPS) {
187 assert(utf8_buffer != NULL, "just checking");
188 return lookup(utf8_buffer, length, THREAD);
189 }
190 static Symbol* new_symbol(const char* name, TRAPS) {
191 return new_symbol(name, (int)strlen(name), THREAD);
192 }
193 static Symbol* new_symbol(const Symbol* sym, int begin, int end, TRAPS) {
194 assert(begin <= end && end <= sym->utf8_length(), "just checking");
195 return lookup(sym, begin, end, THREAD);
196 }
198 // Create a symbol in the arena for symbols that are not deleted
199 static Symbol* new_permanent_symbol(const char* name, TRAPS);
201 // Symbol lookup
202 static Symbol* lookup(int index, const char* name, int len, TRAPS);
204 // Needed for preloading classes in signatures when compiling.
205 // Returns the symbol is already present in symbol table, otherwise
206 // NULL. NO ALLOCATION IS GUARANTEED!
207 static Symbol* probe(const char* name, int len) {
208 unsigned int ignore_hash;
209 return lookup_only(name, len, ignore_hash);
210 }
211 static Symbol* probe_unicode(const jchar* name, int len) {
212 unsigned int ignore_hash;
213 return lookup_only_unicode(name, len, ignore_hash);
214 }
216 // Histogram
217 static void print_histogram() PRODUCT_RETURN;
218 static void print() PRODUCT_RETURN;
220 // Debugging
221 static void verify();
222 static void dump(outputStream* st);
224 // Sharing
225 static void copy_buckets(char** top, char*end) {
226 the_table()->Hashtable<Symbol*, mtSymbol>::copy_buckets(top, end);
227 }
228 static void copy_table(char** top, char*end) {
229 the_table()->Hashtable<Symbol*, mtSymbol>::copy_table(top, end);
230 }
231 static void reverse(void* boundary = NULL) {
232 the_table()->Hashtable<Symbol*, mtSymbol>::reverse(boundary);
233 }
235 // Rehash the symbol table if it gets out of balance
236 static void rehash_table();
237 static bool needs_rehashing() { return _needs_rehashing; }
238 };
240 class StringTable : public Hashtable<oop, mtSymbol> {
241 friend class VMStructs;
243 private:
244 // The string table
245 static StringTable* _the_table;
247 // Set if one bucket is out of balance due to hash algorithm deficiency
248 static bool _needs_rehashing;
250 // Claimed high water mark for parallel chunked scanning
251 static volatile int _parallel_claimed_idx;
253 static oop intern(Handle string_or_null, jchar* chars, int length, TRAPS);
254 oop basic_add(int index, Handle string_or_null, jchar* name, int len,
255 unsigned int hashValue, TRAPS);
257 oop lookup(int index, jchar* chars, int length, unsigned int hashValue);
259 // Apply the give oop closure to the entries to the buckets
260 // in the range [start_idx, end_idx).
261 static void buckets_do(OopClosure* f, int start_idx, int end_idx);
263 StringTable() : Hashtable<oop, mtSymbol>((int)StringTableSize,
264 sizeof (HashtableEntry<oop, mtSymbol>)) {}
266 StringTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
267 : Hashtable<oop, mtSymbol>((int)StringTableSize, sizeof (HashtableEntry<oop, mtSymbol>), t,
268 number_of_entries) {}
269 public:
270 // The string table
271 static StringTable* the_table() { return _the_table; }
273 // Size of one bucket in the string table. Used when checking for rollover.
274 static uint bucket_size() { return sizeof(HashtableBucket<mtSymbol>); }
276 static void create_table() {
277 assert(_the_table == NULL, "One string table allowed.");
278 _the_table = new StringTable();
279 }
281 // GC support
282 // Delete pointers to otherwise-unreachable objects.
283 static void unlink_or_oops_do(BoolObjectClosure* cl, OopClosure* f);
284 static void unlink(BoolObjectClosure* cl) {
285 unlink_or_oops_do(cl, NULL);
286 }
288 // Serially invoke "f->do_oop" on the locations of all oops in the table.
289 static void oops_do(OopClosure* f);
291 // Possibly parallel version of the above
292 static void possibly_parallel_oops_do(OopClosure* f);
294 // Hashing algorithm, used as the hash value used by the
295 // StringTable for bucket selection and comparison (stored in the
296 // HashtableEntry structures). This is used in the String.intern() method.
297 static unsigned int hash_string(const jchar* s, int len);
299 // Internal test.
300 static void test_alt_hash() PRODUCT_RETURN;
302 // Probing
303 static oop lookup(Symbol* symbol);
304 static oop lookup(jchar* chars, int length);
306 // Interning
307 static oop intern(Symbol* symbol, TRAPS);
308 static oop intern(oop string, TRAPS);
309 static oop intern(const char *utf8_string, TRAPS);
311 // Debugging
312 static void verify();
313 static void dump(outputStream* st);
315 enum VerifyMesgModes {
316 _verify_quietly = 0,
317 _verify_with_mesgs = 1
318 };
320 enum VerifyRetTypes {
321 _verify_pass = 0,
322 _verify_fail_continue = 1,
323 _verify_fail_done = 2
324 };
326 static VerifyRetTypes compare_entries(int bkt1, int e_cnt1,
327 HashtableEntry<oop, mtSymbol>* e_ptr1,
328 int bkt2, int e_cnt2,
329 HashtableEntry<oop, mtSymbol>* e_ptr2);
330 static VerifyRetTypes verify_entry(int bkt, int e_cnt,
331 HashtableEntry<oop, mtSymbol>* e_ptr,
332 VerifyMesgModes mesg_mode);
333 static int verify_and_compare_entries();
335 // Sharing
336 static void copy_buckets(char** top, char*end) {
337 the_table()->Hashtable<oop, mtSymbol>::copy_buckets(top, end);
338 }
339 static void copy_table(char** top, char*end) {
340 the_table()->Hashtable<oop, mtSymbol>::copy_table(top, end);
341 }
342 static void reverse() {
343 the_table()->Hashtable<oop, mtSymbol>::reverse();
344 }
346 // Rehash the symbol table if it gets out of balance
347 static void rehash_table();
348 static bool needs_rehashing() { return _needs_rehashing; }
350 // Parallel chunked scanning
351 static void clear_parallel_claimed_index() { _parallel_claimed_idx = 0; }
352 };
353 #endif // SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP