1 /*
   2  * Copyright (c) 1997, 2017, 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 RehashableHashtable<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     : RehashableHashtable<Symbol*, mtSymbol>(SymbolTableSize, sizeof (HashtableEntry<Symbol*, mtSymbol>)) {}
 114 
 115   SymbolTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
 116     : RehashableHashtable<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   typedef SymbolTable::BucketUnlinkContext BucketUnlinkContext;
 128   // Release any dead symbols. Unlinked bucket entries are collected in the given
 129   // context to be freed later.
 130   // This allows multiple threads to work on the table at once.
 131   static void buckets_unlink(int start_idx, int end_idx, BucketUnlinkContext* context, size_t* memory_total);
 132 public:
 133   enum {
 134     symbol_alloc_batch_size = 8,
 135     // Pick initial size based on java -version size measurements
 136     symbol_alloc_arena_size = 360*K
 137   };
 138 
 139   // The symbol table
 140   static SymbolTable* the_table() { return _the_table; }
 141 
 142   // Size of one bucket in the string table.  Used when checking for rollover.
 143   static uint bucket_size() { return sizeof(HashtableBucket<mtSymbol>); }
 144 
 145   static void create_table() {
 146     assert(_the_table == NULL, "One symbol table allowed.");
 147     _the_table = new SymbolTable();
 148     initialize_symbols(symbol_alloc_arena_size);
 149   }
 150 
 151   static void create_table(HashtableBucket<mtSymbol>* t, int length,
 152                            int number_of_entries) {
 153     assert(_the_table == NULL, "One symbol table allowed.");
 154 
 155     // If CDS archive used a different symbol table size, use that size instead
 156     // which is better than giving an error.
 157     SymbolTableSize = length/bucket_size();
 158 
 159     _the_table = new SymbolTable(t, number_of_entries);
 160     // if CDS give symbol table a default arena size since most symbols
 161     // are already allocated in the shared misc section.
 162     initialize_symbols();
 163   }
 164 
 165   static unsigned int hash_symbol(const char* s, int len);
 166 
 167   static Symbol* lookup(const char* name, int len, TRAPS);
 168   // lookup only, won't add. Also calculate hash.
 169   static Symbol* lookup_only(const char* name, int len, unsigned int& hash);
 170   // Only copy to C string to be added if lookup failed.
 171   static Symbol* lookup(const Symbol* sym, int begin, int end, TRAPS);
 172 
 173   static void release(Symbol* sym);
 174 
 175   // Look up the address of the literal in the SymbolTable for this Symbol*
 176   static Symbol** lookup_symbol_addr(Symbol* sym);
 177 
 178   // jchar (utf16) version of lookups
 179   static Symbol* lookup_unicode(const jchar* name, int len, TRAPS);
 180   static Symbol* lookup_only_unicode(const jchar* name, int len, unsigned int& hash);
 181 
 182   static void add(ClassLoaderData* loader_data,
 183                   constantPoolHandle cp, int names_count,
 184                   const char** names, int* lengths, int* cp_indices,
 185                   unsigned int* hashValues, TRAPS);
 186 
 187   // Release any dead symbols
 188   static void unlink() {
 189     int processed = 0;
 190     int removed = 0;
 191     unlink(&processed, &removed);
 192   }
 193   static void unlink(int* processed, int* removed);
 194   // Release any dead symbols, possibly parallel version
 195   static void possibly_parallel_unlink(int* processed, int* removed);
 196 
 197   // iterate over symbols
 198   static void symbols_do(SymbolClosure *cl);
 199 
 200   // Symbol creation
 201   static Symbol* new_symbol(const char* utf8_buffer, int length, TRAPS) {
 202     assert(utf8_buffer != NULL, "just checking");
 203     return lookup(utf8_buffer, length, THREAD);
 204   }
 205   static Symbol*       new_symbol(const char* name, TRAPS) {
 206     return new_symbol(name, (int)strlen(name), THREAD);
 207   }
 208   static Symbol*       new_symbol(const Symbol* sym, int begin, int end, TRAPS) {
 209     assert(begin <= end && end <= sym->utf8_length(), "just checking");
 210     return lookup(sym, begin, end, THREAD);
 211   }
 212 
 213   // Create a symbol in the arena for symbols that are not deleted
 214   static Symbol* new_permanent_symbol(const char* name, TRAPS);
 215 
 216   // Symbol lookup
 217   static Symbol* lookup(int index, const char* name, int len, TRAPS);
 218 
 219   // Needed for preloading classes in signatures when compiling.
 220   // Returns the symbol is already present in symbol table, otherwise
 221   // NULL.  NO ALLOCATION IS GUARANTEED!
 222   static Symbol* probe(const char* name, int len) {
 223     unsigned int ignore_hash;
 224     return lookup_only(name, len, ignore_hash);
 225   }
 226   static Symbol* probe_unicode(const jchar* name, int len) {
 227     unsigned int ignore_hash;
 228     return lookup_only_unicode(name, len, ignore_hash);
 229   }
 230 
 231   // Histogram
 232   static void print_histogram()     PRODUCT_RETURN;
 233   static void print()     PRODUCT_RETURN;
 234 
 235   // Debugging
 236   static void verify();
 237   static void dump(outputStream* st);
 238 
 239   // Sharing
 240   static void copy_buckets(char** top, char*end) {
 241     the_table()->Hashtable<Symbol*, mtSymbol>::copy_buckets(top, end);
 242   }
 243   static void copy_table(char** top, char*end) {
 244     the_table()->Hashtable<Symbol*, mtSymbol>::copy_table(top, end);
 245   }
 246   static void reverse(void* boundary = NULL) {
 247     the_table()->Hashtable<Symbol*, mtSymbol>::reverse(boundary);
 248   }
 249 
 250   // Rehash the symbol table if it gets out of balance
 251   static void rehash_table();
 252   static bool needs_rehashing()         { return _needs_rehashing; }
 253   // Parallel chunked scanning
 254   static void clear_parallel_claimed_index() { _parallel_claimed_idx = 0; }
 255   static int parallel_claimed_index()        { return _parallel_claimed_idx; }
 256 };
 257 
 258 class StringTable : public RehashableHashtable<oop, mtSymbol> {
 259   friend class VMStructs;
 260 
 261 private:
 262   // The string table
 263   static StringTable* _the_table;
 264 
 265   // Set if one bucket is out of balance due to hash algorithm deficiency
 266   static bool _needs_rehashing;
 267 
 268   // Claimed high water mark for parallel chunked scanning
 269   static volatile int _parallel_claimed_idx;
 270 
 271   static oop intern(Handle string_or_null, jchar* chars, int length, TRAPS);
 272   oop basic_add(int index, Handle string_or_null, jchar* name, int len,
 273                 unsigned int hashValue, TRAPS);
 274 
 275   oop lookup(int index, jchar* chars, int length, unsigned int hashValue);
 276 
 277   // Apply the give oop closure to the entries to the buckets
 278   // in the range [start_idx, end_idx).
 279   static void buckets_oops_do(OopClosure* f, int start_idx, int end_idx);
 280 
 281   typedef StringTable::BucketUnlinkContext BucketUnlinkContext;
 282   // Unlink or apply the give oop closure to the entries to the buckets
 283   // in the range [start_idx, end_idx). Unlinked bucket entries are collected in the given
 284   // context to be freed later.
 285   // This allows multiple threads to work on the table at once.
 286   static void buckets_unlink_or_oops_do(BoolObjectClosure* is_alive, OopClosure* f, int start_idx, int end_idx, BucketUnlinkContext* context);
 287 
 288   StringTable() : RehashableHashtable<oop, mtSymbol>((int)StringTableSize,
 289                               sizeof (HashtableEntry<oop, mtSymbol>)) {}
 290 
 291   StringTable(HashtableBucket<mtSymbol>* t, int number_of_entries)
 292     : RehashableHashtable<oop, mtSymbol>((int)StringTableSize, sizeof (HashtableEntry<oop, mtSymbol>), t,
 293                      number_of_entries) {}
 294 public:
 295   // The string table
 296   static StringTable* the_table() { return _the_table; }
 297 
 298   // Size of one bucket in the string table.  Used when checking for rollover.
 299   static uint bucket_size() { return sizeof(HashtableBucket<mtSymbol>); }
 300 
 301   static void create_table() {
 302     assert(_the_table == NULL, "One string table allowed.");
 303     _the_table = new StringTable();
 304   }
 305 
 306   // GC support
 307   //   Delete pointers to otherwise-unreachable objects.
 308   static void unlink_or_oops_do(BoolObjectClosure* cl, OopClosure* f) {
 309     int processed = 0;
 310     int removed = 0;
 311     unlink_or_oops_do(cl, f, &processed, &removed);
 312   }
 313   static void unlink(BoolObjectClosure* cl) {
 314     int processed = 0;
 315     int removed = 0;
 316     unlink_or_oops_do(cl, NULL, &processed, &removed);
 317   }
 318   static void unlink_or_oops_do(BoolObjectClosure* cl, OopClosure* f, int* processed, int* removed);
 319   static void unlink(BoolObjectClosure* cl, int* processed, int* removed) {
 320     unlink_or_oops_do(cl, NULL, processed, removed);
 321   }
 322   // Serially invoke "f->do_oop" on the locations of all oops in the table.
 323   static void oops_do(OopClosure* f);
 324 
 325   // Possibly parallel versions of the above
 326   static void possibly_parallel_unlink_or_oops_do(BoolObjectClosure* cl, OopClosure* f, int* processed, int* removed);
 327   static void possibly_parallel_unlink(BoolObjectClosure* cl, int* processed, int* removed) {
 328     possibly_parallel_unlink_or_oops_do(cl, NULL, processed, removed);
 329   }
 330   static void possibly_parallel_oops_do(OopClosure* f);
 331   static void possibly_parallel_oops_do_shenandoah(OopClosure* f);
 332 
 333   // Hashing algorithm, used as the hash value used by the
 334   //     StringTable for bucket selection and comparison (stored in the
 335   //     HashtableEntry structures).  This is used in the String.intern() method.
 336   static unsigned int hash_string(const jchar* s, int len);
 337 
 338   // Internal test.
 339   static void test_alt_hash() PRODUCT_RETURN;
 340 
 341   // Probing
 342   static oop lookup(Symbol* symbol);
 343   static oop lookup(jchar* chars, int length);
 344 
 345   // Interning
 346   static oop intern(Symbol* symbol, TRAPS);
 347   static oop intern(oop string, TRAPS);
 348   static oop intern(const char *utf8_string, TRAPS);
 349 
 350   // Debugging
 351   static void verify();
 352   static void dump(outputStream* st);
 353 
 354   enum VerifyMesgModes {
 355     _verify_quietly    = 0,
 356     _verify_with_mesgs = 1
 357   };
 358 
 359   enum VerifyRetTypes {
 360     _verify_pass          = 0,
 361     _verify_fail_continue = 1,
 362     _verify_fail_done     = 2
 363   };
 364 
 365   static VerifyRetTypes compare_entries(int bkt1, int e_cnt1,
 366                                         HashtableEntry<oop, mtSymbol>* e_ptr1,
 367                                         int bkt2, int e_cnt2,
 368                                         HashtableEntry<oop, mtSymbol>* e_ptr2);
 369   static VerifyRetTypes verify_entry(int bkt, int e_cnt,
 370                                      HashtableEntry<oop, mtSymbol>* e_ptr,
 371                                      VerifyMesgModes mesg_mode);
 372   static int verify_and_compare_entries();
 373 
 374   // Sharing
 375   static void copy_buckets(char** top, char*end) {
 376     the_table()->Hashtable<oop, mtSymbol>::copy_buckets(top, end);
 377   }
 378   static void copy_table(char** top, char*end) {
 379     the_table()->Hashtable<oop, mtSymbol>::copy_table(top, end);
 380   }
 381   static void reverse() {
 382     the_table()->Hashtable<oop, mtSymbol>::reverse();
 383   }
 384 
 385   // Rehash the symbol table if it gets out of balance
 386   static void rehash_table();
 387   static bool needs_rehashing() { return _needs_rehashing; }
 388 
 389   // Parallel chunked scanning
 390   static void clear_parallel_claimed_index() { _parallel_claimed_idx = 0; }
 391   static int parallel_claimed_index() { return _parallel_claimed_idx; }
 392 };
 393 #endif // SHARE_VM_CLASSFILE_SYMBOLTABLE_HPP