1 /*
   2  * Copyright (c) 1998, 2020, 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_RUNTIME_OBJECTMONITOR_HPP
  26 #define SHARE_RUNTIME_OBJECTMONITOR_HPP
  27 
  28 #include "memory/allocation.hpp"
  29 #include "memory/padded.hpp"
  30 #include "oops/markWord.hpp"
  31 #include "runtime/os.hpp"
  32 #include "runtime/park.hpp"
  33 #include "runtime/perfData.hpp"
  34 
  35 class ObjectMonitor;
  36 
  37 // ObjectWaiter serves as a "proxy" or surrogate thread.
  38 // TODO-FIXME: Eliminate ObjectWaiter and use the thread-specific
  39 // ParkEvent instead.  Beware, however, that the JVMTI code
  40 // knows about ObjectWaiters, so we'll have to reconcile that code.
  41 // See next_waiter(), first_waiter(), etc.
  42 
  43 class ObjectWaiter : public StackObj {
  44  public:
  45   enum TStates { TS_UNDEF, TS_READY, TS_RUN, TS_WAIT, TS_ENTER, TS_CXQ };
  46   ObjectWaiter* volatile _next;
  47   ObjectWaiter* volatile _prev;
  48   Thread*       _thread;
  49   jlong         _notifier_tid;
  50   ParkEvent *   _event;
  51   volatile int  _notified;
  52   volatile TStates TState;
  53   bool          _active;           // Contention monitoring is enabled
  54  public:
  55   ObjectWaiter(Thread* thread);
  56 
  57   void wait_reenter_begin(ObjectMonitor *mon);
  58   void wait_reenter_end(ObjectMonitor *mon);
  59 };
  60 
  61 // The ObjectMonitor class implements the heavyweight version of a
  62 // JavaMonitor. The lightweight BasicLock/stack lock version has been
  63 // inflated into an ObjectMonitor. This inflation is typically due to
  64 // contention or use of Object.wait().
  65 //
  66 // WARNING: This is a very sensitive and fragile class. DO NOT make any
  67 // changes unless you are fully aware of the underlying semantics.
  68 //
  69 // ObjectMonitor Layout Overview/Highlights/Restrictions:
  70 //
  71 // - The _header field must be at offset 0 because the displaced header
  72 //   from markWord is stored there. We do not want markWord.hpp to include
  73 //   ObjectMonitor.hpp to avoid exposing ObjectMonitor everywhere. This
  74 //   means that ObjectMonitor cannot inherit from any other class nor can
  75 //   it use any virtual member functions. This restriction is critical to
  76 //   the proper functioning of the VM.
  77 // - The _header and _owner fields should be separated by enough space
  78 //   to avoid false sharing due to parallel access by different threads.
  79 //   This is an advisory recommendation.
  80 // - The general layout of the fields in ObjectMonitor is:
  81 //     _header
  82 //     <lightly_used_fields>
  83 //     <optional padding>
  84 //     _owner
  85 //     <remaining_fields>
  86 // - The VM assumes write ordering and machine word alignment with
  87 //   respect to the _owner field and the <remaining_fields> that can
  88 //   be read in parallel by other threads.
  89 // - Generally fields that are accessed closely together in time should
  90 //   be placed proximally in space to promote data cache locality. That
  91 //   is, temporal locality should condition spatial locality.
  92 // - We have to balance avoiding false sharing with excessive invalidation
  93 //   from coherence traffic. As such, we try to cluster fields that tend
  94 //   to be _written_ at approximately the same time onto the same data
  95 //   cache line.
  96 // - We also have to balance the natural tension between minimizing
  97 //   single threaded capacity misses with excessive multi-threaded
  98 //   coherency misses. There is no single optimal layout for both
  99 //   single-threaded and multi-threaded environments.
 100 //
 101 // - See TEST_VM(ObjectMonitor, sanity) gtest for how critical restrictions are
 102 //   enforced.
 103 // - Adjacent ObjectMonitors should be separated by enough space to avoid
 104 //   false sharing. This is handled by the ObjectMonitor allocation code
 105 //   in synchronizer.cpp. Also see TEST_VM(SynchronizerTest, sanity) gtest.
 106 //
 107 // Futures notes:
 108 //   - Separating _owner from the <remaining_fields> by enough space to
 109 //     avoid false sharing might be profitable. Given
 110 //     http://blogs.oracle.com/dave/entry/cas_and_cache_trivia_invalidate
 111 //     we know that the CAS in monitorenter will invalidate the line
 112 //     underlying _owner. We want to avoid an L1 data cache miss on that
 113 //     same line for monitorexit. Putting these <remaining_fields>:
 114 //     _recursions, _EntryList, _cxq, and _succ, all of which may be
 115 //     fetched in the inflated unlock path, on a different cache line
 116 //     would make them immune to CAS-based invalidation from the _owner
 117 //     field.
 118 //
 119 //   - The _recursions field should be of type int, or int32_t but not
 120 //     intptr_t. There's no reason to use a 64-bit type for this field
 121 //     in a 64-bit JVM.
 122 
 123 #ifndef OM_CACHE_LINE_SIZE
 124 // Use DEFAULT_CACHE_LINE_SIZE if not already specified for
 125 // the current build platform.
 126 #define OM_CACHE_LINE_SIZE DEFAULT_CACHE_LINE_SIZE
 127 #endif
 128 
 129 class ObjectMonitor {
 130   friend class ObjectMonitorHandle;
 131   friend class ObjectSynchronizer;
 132   friend class ObjectWaiter;
 133   friend class VMStructs;
 134   JVMCI_ONLY(friend class JVMCIVMStructs;)
 135 
 136   // The sync code expects the header field to be at offset zero (0).
 137   // Enforced by the assert() in header_addr().
 138   volatile markWord _header;        // displaced object header word - mark
 139   void* volatile _object;           // backward object pointer - strong root
 140   typedef enum {
 141     Free = 0,  // Free must be 0 for monitor to be free after memset(..,0,..).
 142     New,
 143     Old
 144   } AllocationState;
 145   AllocationState _allocation_state;
 146   // Separate _header and _owner on different cache lines since both can
 147   // have busy multi-threaded access. _header, _object and _allocation_state
 148   // are set at initial inflation. _object and _allocation_state don't
 149   // change until deflation so _object and _allocation_state are good
 150   // choices to share the cache line with _header.
 151   DEFINE_PAD_MINUS_SIZE(0, OM_CACHE_LINE_SIZE, sizeof(volatile markWord) +
 152                         sizeof(void* volatile) + sizeof(AllocationState));
 153   // Used by async deflation as a marker in the _owner field:
 154   #define DEFLATER_MARKER reinterpret_cast<void*>(-1)
 155   void* volatile _owner;            // pointer to owning thread OR BasicLock
 156   volatile jlong _previous_owner_tid;  // thread id of the previous owner of the monitor
 157   // Separate _owner and _ref_count on different cache lines since both
 158   // can have busy multi-threaded access. _previous_owner_tid is only
 159   // changed by ObjectMonitor::exit() so it is a good choice to share the
 160   // cache line with _owner.
 161   DEFINE_PAD_MINUS_SIZE(1, OM_CACHE_LINE_SIZE, sizeof(void* volatile) +
 162                         sizeof(volatile jlong));
 163   jint _ref_count;                  // ref count for ObjectMonitor* and used by the async deflation
 164                                     // protocol. See ObjectSynchronizer::deflate_monitor_using_JT().
 165   // Separate _ref_count and _next_om on different cache lines since
 166   // both can have busy multi-threaded access.
 167   DEFINE_PAD_MINUS_SIZE(2, OM_CACHE_LINE_SIZE, sizeof(volatile jint));
 168   ObjectMonitor* _next_om;          // Next ObjectMonitor* linkage
 169   volatile intx _recursions;        // recursion count, 0 for first entry
 170   ObjectWaiter* volatile _EntryList;  // Threads blocked on entry or reentry.
 171                                       // The list is actually composed of WaitNodes,
 172                                       // acting as proxies for Threads.
 173 
 174   ObjectWaiter* volatile _cxq;      // LL of recently-arrived threads blocked on entry.
 175   Thread* volatile _succ;           // Heir presumptive thread - used for futile wakeup throttling
 176   Thread* volatile _Responsible;
 177 
 178   volatile int _Spinner;            // for exit->spinner handoff optimization
 179   volatile int _SpinDuration;
 180 
 181   volatile jint  _contentions;      // Number of active contentions in enter(). It is used by is_busy()
 182                                     // along with other fields to determine if an ObjectMonitor can be
 183                                     // deflated. See ObjectSynchronizer::deflate_monitor() and
 184                                     // ObjectSynchronizer::deflate_monitor_using_JT().
 185  protected:
 186   ObjectWaiter* volatile _WaitSet;  // LL of threads wait()ing on the monitor
 187   volatile jint  _waiters;          // number of waiting threads
 188  private:
 189   volatile int _WaitSetLock;        // protects Wait Queue - simple spinlock
 190 
 191  public:
 192   static void Initialize();
 193 
 194   // Only perform a PerfData operation if the PerfData object has been
 195   // allocated and if the PerfDataManager has not freed the PerfData
 196   // objects which can happen at normal VM shutdown.
 197   //
 198   #define OM_PERFDATA_OP(f, op_str)              \
 199     do {                                         \
 200       if (ObjectMonitor::_sync_ ## f != NULL &&  \
 201           PerfDataManager::has_PerfData()) {     \
 202         ObjectMonitor::_sync_ ## f->op_str;      \
 203       }                                          \
 204     } while (0)
 205 
 206   static PerfCounter * _sync_ContendedLockAttempts;
 207   static PerfCounter * _sync_FutileWakeups;
 208   static PerfCounter * _sync_Parks;
 209   static PerfCounter * _sync_Notifications;
 210   static PerfCounter * _sync_Inflations;
 211   static PerfCounter * _sync_Deflations;
 212   static PerfLongVariable * _sync_MonExtant;
 213 
 214   static int Knob_SpinLimit;
 215 
 216   void* operator new (size_t size) throw();
 217   void* operator new[] (size_t size) throw();
 218   void operator delete(void* p);
 219   void operator delete[] (void* p);
 220 
 221   // TODO-FIXME: the "offset" routines should return a type of off_t instead of int ...
 222   // ByteSize would also be an appropriate type.
 223   static int header_offset_in_bytes()      { return offset_of(ObjectMonitor, _header); }
 224   static int object_offset_in_bytes()      { return offset_of(ObjectMonitor, _object); }
 225   static int owner_offset_in_bytes()       { return offset_of(ObjectMonitor, _owner); }
 226   static int recursions_offset_in_bytes()  { return offset_of(ObjectMonitor, _recursions); }
 227   static int cxq_offset_in_bytes()         { return offset_of(ObjectMonitor, _cxq); }
 228   static int succ_offset_in_bytes()        { return offset_of(ObjectMonitor, _succ); }
 229   static int EntryList_offset_in_bytes()   { return offset_of(ObjectMonitor, _EntryList); }
 230 
 231   // ObjectMonitor references can be ORed with markWord::monitor_value
 232   // as part of the ObjectMonitor tagging mechanism. When we combine an
 233   // ObjectMonitor reference with an offset, we need to remove the tag
 234   // value in order to generate the proper address.
 235   //
 236   // We can either adjust the ObjectMonitor reference and then add the
 237   // offset or we can adjust the offset that is added to the ObjectMonitor
 238   // reference. The latter avoids an AGI (Address Generation Interlock)
 239   // stall so the helper macro adjusts the offset value that is returned
 240   // to the ObjectMonitor reference manipulation code:
 241   //
 242   #define OM_OFFSET_NO_MONITOR_VALUE_TAG(f) \
 243     ((ObjectMonitor::f ## _offset_in_bytes()) - markWord::monitor_value)
 244 
 245   markWord           header() const;
 246   volatile markWord* header_addr();
 247   void               set_header(markWord hdr);
 248 
 249   intptr_t is_busy() const {
 250     // TODO-FIXME: assert _owner == null implies _recursions = 0
 251     // We do not include _ref_count in the is_busy() check because
 252     // _ref_count is for indicating that the ObjectMonitor* is in
 253     // use which is orthogonal to whether the ObjectMonitor itself
 254     // is in use for a locking operation.
 255     intptr_t ret_code = _contentions | _waiters | intptr_t(_cxq) | intptr_t(_EntryList);
 256     if (!AsyncDeflateIdleMonitors) {
 257       ret_code |= intptr_t(_owner);
 258     } else {
 259       if (_owner != DEFLATER_MARKER) {
 260         ret_code |= intptr_t(_owner);
 261       }
 262     }
 263     return ret_code;
 264   }
 265   const char* is_busy_to_string(stringStream* ss);
 266 
 267   intptr_t  is_entered(Thread* current) const;
 268 
 269   void*     owner() const;  // Returns NULL if DEFLATER_MARKER is observed.
 270   // Returns true if owner field == DEFLATER_MARKER and false otherwise.
 271   bool      owner_is_DEFLATER_MARKER();
 272   // Clear _owner field; current value must match old_value.
 273   void      release_clear_owner(void* old_value);
 274   // Simply set _owner field to new_value; current value must match old_value.
 275   void      set_owner_from(void* old_value, void* new_value);
 276   // Simply set _owner field to new_value; current value must match old_value1 or old_value2.
 277   void      set_owner_from(void* old_value1, void* old_value2, void* new_value);
 278   // Simply set _owner field to self; current value must match basic_lock_p.
 279   void      set_owner_from_BasicLock(void* basic_lock_p, Thread* self);
 280   // Try to set _owner field to new_value if the current value matches
 281   // old_value, using Atomic::cmpxchg(). Otherwise, does not change the
 282   // _owner field. Returns the prior value of the _owner field.
 283   void*     try_set_owner_from(void* old_value, void* new_value);
 284 
 285   ObjectMonitor* next_om() const;
 286   // Simply set _next_om field to new_value.
 287   void set_next_om(ObjectMonitor* new_value);
 288   // Try to set _next_om field to new_value if the current value matches
 289   // old_value, using Atomic::cmpxchg(). Otherwise, does not change the
 290   // _next_om field. Returns the prior value of the _next_om field.
 291   ObjectMonitor* try_set_next_om(ObjectMonitor* old_value, ObjectMonitor* new_value);
 292 
 293   jint      waiters() const;
 294 
 295   jint      contentions() const;
 296   intx      recursions() const                                         { return _recursions; }
 297 
 298   // JVM/TI GetObjectMonitorUsage() needs this:
 299   ObjectWaiter* first_waiter()                                         { return _WaitSet; }
 300   ObjectWaiter* next_waiter(ObjectWaiter* o)                           { return o->_next; }
 301   Thread* thread_of_waiter(ObjectWaiter* o)                            { return o->_thread; }
 302 
 303  protected:
 304   // We don't typically expect or want the ctors or dtors to run.
 305   // normal ObjectMonitors are type-stable and immortal.
 306   ObjectMonitor() { ::memset((void*)this, 0, sizeof(*this)); }
 307 
 308   ~ObjectMonitor() {
 309     // TODO: Add asserts ...
 310     // _cxq == 0 _succ == NULL _owner == NULL _waiters == 0
 311     // _contentions == 0 _EntryList  == NULL etc
 312   }
 313 
 314  private:
 315   void Recycle() {
 316     // TODO: add stronger asserts ...
 317     // _cxq == 0 _succ == NULL _owner == NULL _waiters == 0
 318     // _contentions == 0 EntryList  == NULL
 319     // _recursions == 0 _WaitSet == NULL
 320 #ifdef ASSERT
 321     stringStream ss;
 322 #endif
 323     assert((is_busy() | _recursions) == 0, "freeing in-use monitor: %s, "
 324            "recursions=" INTX_FORMAT, is_busy_to_string(&ss), _recursions);
 325     _succ          = NULL;
 326     _EntryList     = NULL;
 327     _cxq           = NULL;
 328     _WaitSet       = NULL;
 329     _recursions    = 0;
 330   }
 331 
 332  public:
 333 
 334   void*     object() const;
 335   void*     object_addr();
 336   void      set_object(void* obj);
 337   void      set_allocation_state(AllocationState s);
 338   AllocationState allocation_state() const;
 339   bool      is_free() const;
 340   bool      is_old() const;
 341   bool      is_new() const;
 342   void      dec_ref_count();
 343   void      inc_ref_count();
 344   jint      ref_count() const;
 345 
 346   // Returns true if the specified thread owns the ObjectMonitor. Otherwise
 347   // returns false and throws IllegalMonitorStateException (IMSE).
 348   bool      check_owner(Thread* THREAD);
 349   void      clear();
 350   void      clear_using_JT();
 351 
 352   void      enter(TRAPS);
 353   void      exit(bool not_suspended, TRAPS);
 354   void      wait(jlong millis, bool interruptable, TRAPS);
 355   void      notify(TRAPS);
 356   void      notifyAll(TRAPS);
 357 
 358   void      print() const;
 359 #ifdef ASSERT
 360   void      print_debug_style_on(outputStream* st) const;
 361 #endif
 362   void      print_on(outputStream* st) const;
 363 
 364 // Use the following at your own risk
 365   intx      complete_exit(TRAPS);
 366   void      reenter(intx recursions, TRAPS);
 367 
 368  private:
 369   void      AddWaiter(ObjectWaiter* waiter);
 370   void      INotify(Thread* self);
 371   ObjectWaiter* DequeueWaiter();
 372   void      DequeueSpecificWaiter(ObjectWaiter* waiter);
 373   void      EnterI(TRAPS);
 374   void      ReenterI(Thread* self, ObjectWaiter* self_node);
 375   void      UnlinkAfterAcquire(Thread* self, ObjectWaiter* self_node);
 376   int       TryLock(Thread* self);
 377   int       NotRunnable(Thread* self, Thread* Owner);
 378   int       TrySpin(Thread* self);
 379   void      ExitEpilog(Thread* self, ObjectWaiter* Wakee);
 380   bool      ExitSuspendEquivalent(JavaThread* self);
 381   void      install_displaced_markword_in_object(const oop obj);
 382 };
 383 
 384 // A helper object for managing an ObjectMonitor*'s ref_count. There
 385 // are special safety considerations when async deflation is used.
 386 class ObjectMonitorHandle : public StackObj {
 387  private:
 388   ObjectMonitor* _om_ptr;
 389  public:
 390   ObjectMonitorHandle() { _om_ptr = NULL; }
 391   ~ObjectMonitorHandle();
 392 
 393   ObjectMonitor* om_ptr() const { return _om_ptr; }
 394   // Save the ObjectMonitor* associated with the specified markWord and
 395   // increment the ref_count.
 396   bool save_om_ptr(oop object, markWord mark);
 397   // Save the specified ObjectMonitor* if safe and increment the ref_count.
 398   bool save_om_ptr_if_safe(ObjectMonitor* om_ptr);
 399   // Unset the _om_ptr field and decrement the ref_count.
 400   void unset_om_ptr();
 401 
 402   // For internal use by ObjectSynchronizer::inflate().
 403   void set_om_ptr(ObjectMonitor* om_ptr);
 404 };
 405 
 406 // Macro to use guarantee() for more strict AsyncDeflateIdleMonitors
 407 // checks and assert() otherwise.
 408 #define ADIM_guarantee(p, ...)       \
 409   do {                               \
 410     if (AsyncDeflateIdleMonitors) {  \
 411       guarantee(p, __VA_ARGS__);     \
 412     } else {                         \
 413       assert(p, __VA_ARGS__);        \
 414     }                                \
 415   } while (0)
 416 
 417 #endif // SHARE_RUNTIME_OBJECTMONITOR_HPP