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() 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;
  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   // Table size
 111   enum {
 112     symbol_table_size = 20011
 113   };
 114 
 115   Symbol* lookup(int index, const char* name, int len, unsigned int hash);
 116 
 117   SymbolTable()
 118     : Hashtable<Symbol*, mtSymbol>(symbol_table_size, sizeof (HashtableEntry<Symbol*, mtSymbol>)) {}
 119 
 120   SymbolTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
 121     : Hashtable<Symbol*, mtSymbol>(symbol_table_size, sizeof (HashtableEntry<Symbol*, mtSymbol>), t,
 122                 number_of_entries) {}
 123 
 124   // Arena for permanent symbols (null class loader) that are never unloaded
 125   static Arena*  _arena;
 126   static Arena* arena() { return _arena; }  // called for statistics
 127 
 128   static void initialize_symbols(int arena_alloc_size = 0);
 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   static void create_table() {
 140     assert(_the_table == NULL, "One symbol table allowed.");
 141     _the_table = new SymbolTable();
 142     initialize_symbols(symbol_alloc_arena_size);
 143   }
 144 
 145   static void create_table(HashtableBucket<mtSymbol>* t, int length,
 146                            int number_of_entries) {
 147     assert(_the_table == NULL, "One symbol table allowed.");
 148     assert(length == symbol_table_size * sizeof(HashtableBucket<mtSymbol>),
 149            "bad shared symbol size.");
 150     _the_table = new SymbolTable(t, number_of_entries);
 151     // if CDS give symbol table a default arena size since most symbols
 152     // are already allocated in the shared misc section.
 153     initialize_symbols();
 154   }
 155 
 156   static unsigned int hash_symbol(const char* s, int len);
 157 
 158   static Symbol* lookup(const char* name, int len, TRAPS);
 159   // lookup only, won't add. Also calculate hash.
 160   static Symbol* lookup_only(const char* name, int len, unsigned int& hash);
 161   // Only copy to C string to be added if lookup failed.
 162   static Symbol* lookup(const Symbol* sym, int begin, int end, TRAPS);
 163 
 164   static void release(Symbol* sym);
 165 
 166   // Look up the address of the literal in the SymbolTable for this Symbol*
 167   static Symbol** lookup_symbol_addr(Symbol* sym);
 168 
 169   // jchar (utf16) version of lookups
 170   static Symbol* lookup_unicode(const jchar* name, int len, TRAPS);
 171   static Symbol* lookup_only_unicode(const jchar* name, int len, unsigned int& hash);
 172 
 173   static void add(ClassLoaderData* loader_data,
 174                   constantPoolHandle cp, int names_count,
 175                   const char** names, int* lengths, int* cp_indices,
 176                   unsigned int* hashValues, TRAPS);
 177 
 178   // Release any dead symbols
 179   static void unlink();
 180 
 181   // iterate over symbols
 182   static void symbols_do(SymbolClosure *cl);
 183 
 184   // Symbol creation
 185   static Symbol* new_symbol(const char* utf8_buffer, int length, TRAPS) {
 186     assert(utf8_buffer != NULL, "just checking");
 187     return lookup(utf8_buffer, length, THREAD);
 188   }
 189   static Symbol*       new_symbol(const char* name, TRAPS) {
 190     return new_symbol(name, (int)strlen(name), THREAD);
 191   }
 192   static Symbol*       new_symbol(const Symbol* sym, int begin, int end, TRAPS) {
 193     assert(begin <= end && end <= sym->utf8_length(), "just checking");
 194     return lookup(sym, begin, end, THREAD);
 195   }
 196 
 197   // Create a symbol in the arena for symbols that are not deleted
 198   static Symbol* new_permanent_symbol(const char* name, TRAPS);
 199 
 200   // Symbol lookup
 201   static Symbol* lookup(int index, const char* name, int len, TRAPS);
 202 
 203   // Needed for preloading classes in signatures when compiling.
 204   // Returns the symbol is already present in symbol table, otherwise
 205   // NULL.  NO ALLOCATION IS GUARANTEED!
 206   static Symbol* probe(const char* name, int len) {
 207     unsigned int ignore_hash;
 208     return lookup_only(name, len, ignore_hash);
 209   }
 210   static Symbol* probe_unicode(const jchar* name, int len) {
 211     unsigned int ignore_hash;
 212     return lookup_only_unicode(name, len, ignore_hash);
 213   }
 214 
 215   // Histogram
 216   static void print_histogram()     PRODUCT_RETURN;
 217   static void print()     PRODUCT_RETURN;
 218 
 219   // Debugging
 220   static void verify();
 221   static void dump(outputStream* st);
 222 
 223   // Sharing
 224   static void copy_buckets(char** top, char*end) {
 225     the_table()->Hashtable<Symbol*, mtSymbol>::copy_buckets(top, end);
 226   }
 227   static void copy_table(char** top, char*end) {
 228     the_table()->Hashtable<Symbol*, mtSymbol>::copy_table(top, end);
 229   }
 230   static void reverse(void* boundary = NULL) {
 231     the_table()->Hashtable<Symbol*, mtSymbol>::reverse(boundary);
 232   }
 233 
 234   // Rehash the symbol table if it gets out of balance
 235   static void rehash_table();
 236   static bool needs_rehashing()         { return _needs_rehashing; }
 237 };
 238 
 239 class StringTable : public Hashtable<oop, mtSymbol> {
 240   friend class VMStructs;
 241 
 242 private:
 243   // The string table
 244   static StringTable* _the_table;
 245 
 246   // Set if one bucket is out of balance due to hash algorithm deficiency
 247   static bool _needs_rehashing;
 248 
 249   // Claimed high water mark for parallel chunked scanning
 250   static volatile int _par_claimed_idx;
 251 
 252   static oop intern(Handle string_or_null, jchar* chars, int length, TRAPS);
 253   oop basic_add(int index, Handle string_or_null, jchar* name, int len,
 254                 unsigned int hashValue, TRAPS);
 255 
 256   oop lookup(int index, jchar* chars, int length, unsigned int hashValue);
 257 
 258   // Apply the give oop closure to the entries to the buckets
 259   // in the range [start_idx, end_idx).
 260   static void buckets_do(OopClosure* f, int start_idx, int end_idx);
 261 
 262   StringTable() : Hashtable<oop, mtSymbol>((int)StringTableSize,
 263                               sizeof (HashtableEntry<oop, mtSymbol>)) {}
 264 
 265   StringTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
 266     : Hashtable<oop, mtSymbol>((int)StringTableSize, sizeof (HashtableEntry<oop, mtSymbol>), t,
 267                      number_of_entries) {}
 268 public:
 269   // The string table
 270   static StringTable* the_table() { return _the_table; }
 271 
 272   // Size of one bucket in the string table.  Used when checking for rollover.
 273   static uint bucket_size() { return sizeof(HashtableBucket<mtSymbol>); }
 274 
 275   static void create_table() {
 276     assert(_the_table == NULL, "One string table allowed.");
 277     _the_table = new StringTable();
 278   }
 279 
 280   // GC support
 281   //   Delete pointers to otherwise-unreachable objects.
 282   static void unlink_or_oops_do(BoolObjectClosure* cl, OopClosure* f);
 283   static void unlink(BoolObjectClosure* cl) {
 284     unlink_or_oops_do(cl, NULL);
 285   }
 286 
 287   // Serially invoke "f->do_oop" on the locations of all oops in the table.
 288   static void oops_do(OopClosure* f);
 289 
 290   // Possibly parallel version of the above
 291   static void possibly_parallel_oops_do(OopClosure* f);
 292 
 293   // Hashing algorithm, used as the hash value used by the
 294   //     StringTable for bucket selection and comparison (stored in the
 295   //     HashtableEntry structures).  This is used in the String.intern() method.
 296   static unsigned int hash_string(const jchar* s, int len);
 297 
 298   // Internal test.
 299   static void test_alt_hash() PRODUCT_RETURN;
 300 
 301   // Probing
 302   static oop lookup(Symbol* symbol);
 303   static oop lookup(jchar* chars, int length);
 304 
 305   // Interning
 306   static oop intern(Symbol* symbol, TRAPS);
 307   static oop intern(oop string, TRAPS);
 308   static oop intern(const char *utf8_string, TRAPS);
 309 
 310   // Debugging
 311   static void verify();
 312   static void dump(outputStream* st);
 313 
 314   // Sharing
 315   static void copy_buckets(char** top, char*end) {
 316     the_table()->Hashtable<oop, mtSymbol>::copy_buckets(top, end);
 317   }
 318   static void copy_table(char** top, char*end) {
 319     the_table()->Hashtable<oop, mtSymbol>::copy_table(top, end);
 320   }
 321   static void reverse() {
 322     the_table()->Hashtable<oop, mtSymbol>::reverse();
 323   }
 324 
 325   // Rehash the symbol table if it gets out of balance
 326   static void rehash_table();
 327   static bool needs_rehashing() { return _needs_rehashing; }
 328 
 329   // Parallel chunked scanning
 330   static void clear_par_claimed_index() { _par_claimed_idx = 0; }
 331 };
 332 #endif // SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP