1 /*
   2  * Copyright (c) 1997, 2018, 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 
  26 #include "precompiled.hpp"
  27 #include "classfile/altHashing.hpp"
  28 #include "classfile/classLoaderData.hpp"
  29 #include "gc/shared/collectedHeap.hpp"
  30 #include "logging/log.hpp"
  31 #include "logging/logStream.hpp"
  32 #include "memory/allocation.inline.hpp"
  33 #include "memory/resourceArea.hpp"
  34 #include "oops/symbol.hpp"
  35 #include "runtime/atomic.hpp"
  36 #include "runtime/os.hpp"
  37 
  38 uint32_t Symbol::pack_length_and_refcount(int length, int refcount) {
  39   STATIC_ASSERT(max_symbol_length == ((1 << 16) - 1));
  40   STATIC_ASSERT(PERM_REFCOUNT == ((1 << 16) - 1));
  41   assert(length >= 0, "negative length");
  42   assert(length <= max_symbol_length, "too long symbol");
  43   assert(refcount >= 0, "negative refcount");
  44   assert(refcount <= PERM_REFCOUNT, "invalid refcount");
  45   uint32_t hi = length;
  46   uint32_t lo = refcount;
  47   return (hi << 16) | lo;
  48 }
  49 
  50 Symbol::Symbol(const u1* name, int length, int refcount) {
  51   _length_and_refcount =  pack_length_and_refcount(length, refcount);
  52   _identity_hash = (short)os::random();
  53   for (int i = 0; i < length; i++) {
  54     byte_at_put(i, name[i]);
  55   }
  56 }
  57 
  58 void* Symbol::operator new(size_t sz, int len, TRAPS) throw() {
  59   int alloc_size = size(len)*wordSize;
  60   address res = (address) AllocateHeap(alloc_size, mtSymbol);
  61   return res;
  62 }
  63 
  64 void* Symbol::operator new(size_t sz, int len, Arena* arena, TRAPS) throw() {
  65   int alloc_size = size(len)*wordSize;
  66   address res = (address)arena->Amalloc_4(alloc_size);
  67   return res;
  68 }
  69 
  70 void Symbol::operator delete(void *p) {
  71   assert(((Symbol*)p)->refcount() == 0, "should not call this");
  72   FreeHeap(p);
  73 }
  74 
  75 // ------------------------------------------------------------------
  76 // Symbol::starts_with
  77 //
  78 // Tests if the symbol starts with the specified prefix of the given
  79 // length.
  80 bool Symbol::starts_with(const char* prefix, int len) const {
  81   if (len > utf8_length()) return false;
  82   while (len-- > 0) {
  83     if (prefix[len] != char_at(len))
  84       return false;
  85   }
  86   assert(len == -1, "we should be at the beginning");
  87   return true;
  88 }
  89 
  90 
  91 // ------------------------------------------------------------------
  92 // Symbol::index_of
  93 //
  94 // Finds if the given string is a substring of this symbol's utf8 bytes.
  95 // Return -1 on failure.  Otherwise return the first index where str occurs.
  96 int Symbol::index_of_at(int i, const char* str, int len) const {
  97   assert(i >= 0 && i <= utf8_length(), "oob");
  98   if (len <= 0)  return 0;
  99   char first_char = str[0];
 100   address bytes = (address) ((Symbol*)this)->base();
 101   address limit = bytes + utf8_length() - len;  // inclusive limit
 102   address scan = bytes + i;
 103   if (scan > limit)
 104     return -1;
 105   for (; scan <= limit; scan++) {
 106     scan = (address) memchr(scan, first_char, (limit + 1 - scan));
 107     if (scan == NULL)
 108       return -1;  // not found
 109     assert(scan >= bytes+i && scan <= limit, "scan oob");
 110     if (memcmp(scan, str, len) == 0)
 111       return (int)(scan - bytes);
 112   }
 113   return -1;
 114 }
 115 
 116 
 117 char* Symbol::as_C_string(char* buf, int size) const {
 118   if (size > 0) {
 119     int len = MIN2(size - 1, utf8_length());
 120     for (int i = 0; i < len; i++) {
 121       buf[i] = char_at(i);
 122     }
 123     buf[len] = '\0';
 124   }
 125   return buf;
 126 }
 127 
 128 char* Symbol::as_C_string() const {
 129   int len = utf8_length();
 130   char* str = NEW_RESOURCE_ARRAY(char, len + 1);
 131   return as_C_string(str, len + 1);
 132 }
 133 
 134 char* Symbol::as_C_string_flexible_buffer(Thread* t,
 135                                                  char* buf, int size) const {
 136   char* str;
 137   int len = utf8_length();
 138   int buf_len = len + 1;
 139   if (size < buf_len) {
 140     str = NEW_RESOURCE_ARRAY(char, buf_len);
 141   } else {
 142     str = buf;
 143   }
 144   return as_C_string(str, buf_len);
 145 }
 146 
 147 void Symbol::print_utf8_on(outputStream* st) const {
 148   st->print("%s", as_C_string());
 149 }
 150 
 151 void Symbol::print_symbol_on(outputStream* st) const {
 152   char *s;
 153   st = st ? st : tty;
 154   {
 155     // ResourceMark may not affect st->print(). If st is a string
 156     // stream it could resize, using the same resource arena.
 157     ResourceMark rm;
 158     s = as_quoted_ascii();
 159     s = os::strdup(s);
 160   }
 161   if (s == NULL) {
 162     st->print("(null)");
 163   } else {
 164     st->print("%s", s);
 165     os::free(s);
 166   }
 167 }
 168 
 169 char* Symbol::as_quoted_ascii() const {
 170   const char *ptr = (const char *)&_body[0];
 171   int quoted_length = UTF8::quoted_ascii_length(ptr, utf8_length());
 172   char* result = NEW_RESOURCE_ARRAY(char, quoted_length + 1);
 173   UTF8::as_quoted_ascii(ptr, utf8_length(), result, quoted_length + 1);
 174   return result;
 175 }
 176 
 177 jchar* Symbol::as_unicode(int& length) const {
 178   Symbol* this_ptr = (Symbol*)this;
 179   length = UTF8::unicode_length((char*)this_ptr->bytes(), utf8_length());
 180   jchar* result = NEW_RESOURCE_ARRAY(jchar, length);
 181   if (length > 0) {
 182     UTF8::convert_to_unicode((char*)this_ptr->bytes(), result, length);
 183   }
 184   return result;
 185 }
 186 
 187 const char* Symbol::as_klass_external_name(char* buf, int size) const {
 188   if (size > 0) {
 189     char* str    = as_C_string(buf, size);
 190     int   length = (int)strlen(str);
 191     // Turn all '/'s into '.'s (also for array klasses)
 192     for (int index = 0; index < length; index++) {
 193       if (str[index] == '/') {
 194         str[index] = '.';
 195       }
 196     }
 197     return str;
 198   } else {
 199     return buf;
 200   }
 201 }
 202 
 203 const char* Symbol::as_klass_external_name() const {
 204   char* str    = as_C_string();
 205   int   length = (int)strlen(str);
 206   // Turn all '/'s into '.'s (also for array klasses)
 207   for (int index = 0; index < length; index++) {
 208     if (str[index] == '/') {
 209       str[index] = '.';
 210     }
 211   }
 212   return str;
 213 }
 214 
 215 // Alternate hashing for unbalanced symbol tables.
 216 unsigned int Symbol::new_hash(juint seed) {
 217   ResourceMark rm;
 218   // Use alternate hashing algorithm on this symbol.
 219   return AltHashing::murmur3_32(seed, (const jbyte*)as_C_string(), utf8_length());
 220 }
 221 
 222 // Increment refcount while checking for zero.  If the Symbol's refcount becomes zero
 223 // a thread could be concurrently removing the Symbol.  This is used during SymbolTable
 224 // lookup to avoid reviving a dead Symbol.
 225 bool Symbol::try_increment_refcount() {
 226   uint32_t found = _length_and_refcount;
 227   while (true) {
 228     uint32_t old_value = found;
 229     int refc = extract_refcount(old_value);
 230     if (refc == PERM_REFCOUNT) {
 231       return true;  // sticky max or created permanent
 232     } else if (refc == 0) {
 233       return false; // dead, can't revive.
 234     } else {
 235       found = Atomic::cmpxchg(old_value + 1, &_length_and_refcount, old_value);
 236       if (found == old_value) {
 237         return true; // successfully updated.
 238       }
 239       // refcount changed, try again.
 240     }
 241   }
 242 }
 243 
 244 // The increment_refcount() is called when not doing lookup. It is assumed that you
 245 // have a symbol with a non-zero refcount and it can't become zero while referenced by
 246 // this caller.
 247 void Symbol::increment_refcount() {
 248   if (!try_increment_refcount()) {
 249 #ifdef ASSERT
 250     print();
 251     fatal("refcount has gone to zero");
 252 #endif
 253   }
 254 #ifndef PRODUCT
 255   if (refcount() != PERM_REFCOUNT) { // not a permanent symbol
 256     NOT_PRODUCT(Atomic::inc(&_total_count);)
 257   }
 258 #endif
 259 }
 260 
 261 // Decrement refcount potentially while racing increment, so we need
 262 // to check the value after attempting to decrement so that if another
 263 // thread increments to PERM_REFCOUNT the value is not decremented.
 264 void Symbol::decrement_refcount() {
 265   uint32_t found = _length_and_refcount;
 266   while (true) {
 267     uint32_t old_value = found;
 268     int refc = extract_refcount(old_value);
 269     if (refc == PERM_REFCOUNT) {
 270       return;  // refcount is permanent, permanent is sticky
 271     } else if (refc == 0) {
 272 #ifdef ASSERT
 273       print();
 274       fatal("refcount underflow");
 275 #endif
 276       return;
 277     } else {
 278       found = Atomic::cmpxchg(old_value - 1, &_length_and_refcount, old_value);
 279       if (found == old_value) {
 280         return;  // successfully updated.
 281       }
 282       // refcount changed, try again.
 283     }
 284   }
 285 }
 286 
 287 void Symbol::metaspace_pointers_do(MetaspaceClosure* it) {
 288   if (log_is_enabled(Trace, cds)) {
 289     LogStream trace_stream(Log(cds)::trace());
 290     trace_stream.print("Iter(Symbol): %p ", this);
 291     print_value_on(&trace_stream);
 292     trace_stream.cr();
 293   }
 294 }
 295 
 296 void Symbol::print_on(outputStream* st) const {
 297   if (this == NULL) {
 298     st->print_cr("NULL");
 299   } else {
 300     st->print("Symbol: '");
 301     print_symbol_on(st);
 302     st->print("'");
 303     st->print(" count %d", refcount());
 304   }
 305 }
 306 
 307 // The print_value functions are present in all builds, to support the
 308 // disassembler and error reporting.
 309 void Symbol::print_value_on(outputStream* st) const {
 310   if (this == NULL) {
 311     st->print("NULL");
 312   } else {
 313     st->print("'");
 314     for (int i = 0; i < utf8_length(); i++) {
 315       st->print("%c", char_at(i));
 316     }
 317     st->print("'");
 318   }
 319 }
 320 
 321 bool Symbol::is_valid(Symbol* s) {
 322   if (!is_aligned(s, sizeof(MetaWord))) return false;
 323   if ((size_t)s < os::min_page_size()) return false;
 324 
 325   if (!os::is_readable_range(s, s + 1)) return false;
 326 
 327   // Symbols are not allocated in Java heap.
 328   if (Universe::heap()->is_in_reserved(s)) return false;
 329 
 330   int len = s->utf8_length();
 331   if (len < 0) return false;
 332 
 333   jbyte* bytes = (jbyte*) s->bytes();
 334   return os::is_readable_range(bytes, bytes + len);
 335 }
 336 
 337 // SymbolTable prints this in its statistics
 338 NOT_PRODUCT(size_t Symbol::_total_count = 0;)
--- EOF ---