jvm安全点(四)openjdk17 c++源码垃圾回收之安全点轮询页内存设置不可访问

轮询页的不可访问性是在 ​​JVM 启动初始化阶段静态设置​​ 的，具体由以下函数完成：

​​核心函数：os::protect_memory​​

bool os::protect_memory(char* addr, size_t bytes, ProtType prot, bool is_committed) { unsigned int p = 0; switch (prot) { case MEM_PROT_NONE: p = PROT_NONE; break; // 设置为不可访问 case MEM_PROT_READ: p = PROT_READ; break; // ...其他权限设置 } return linux_mprotect(addr, bytes, p); // 调用系统调用修改内存权限 }

• ​​功能​​：通过 mprotect 系统调用修改指定内存区域的权限。
• ​​调用链​​：在 JVM 启动时由 SafepointMechanism::default_initialize() 调用。

​​初始化阶段设置轮询页权限​​

在 SafepointMechanism::default_initialize() 中：

// 分配两个连续页：bad_page（不可访问）和 good_page（可读） os::protect_memory(bad_page, page_size, os::MEM_PROT_NONE); // 设置 bad_page 不可访问 os::protect_memory(good_page, page_size, os::MEM_PROT_READ); // 设置 good_page 可读

• ​​静态配置​​：此时 bad_page 被永久设置为 ​​不可访问​​，good_page 设置为可读。
• ​​后续使用​​：
  • 当需要进入安全点时，JVM 让线程访问 bad_page（通过修改线程的轮询地址），触发 ​​SIGSEGV 信号​​。
  • 安全点结束后，线程恢复访问 good_page。

​​动态触发安全点的机制​​

在安全点开始时，通过以下函数 ​​切换线程的轮询地址​​ 到 bad_page（而不是修改内存权限）：

void SafepointMechanism::arm_local_poll(JavaThread* thread) { // 将线程的轮询地址设为 _poll_page_armed_value（即 bad_page） thread->set_polling_page(_poll_page_armed_value); }

• ​​不修改内存权限​​：轮询页的权限在初始化时已固定，JVM 通过切换线程访问的目标地址（bad_page 或 good_page）控制安全点触发。

​​总结​​

• ​​设置轮询页不可访问的函数​​：os::protect_memory（在 JVM 启动时调用）。
• ​​触发安全点的机制​​：通过 arm_local_poll() 将线程的轮询地址指向预先不可访问的 bad_page，而非动态修改内存权限。

##源码

/ Set protections specified
bool os::protect_memory(char* addr, size_t bytes, ProtType prot,
                        bool is_committed) {
  unsigned int p = 0;
  switch (prot) {
  case MEM_PROT_NONE: p = PROT_NONE; break;
  case MEM_PROT_READ: p = PROT_READ; break;
  case MEM_PROT_RW:   p = PROT_READ|PROT_WRITE; break;
  case MEM_PROT_RWX:  p = PROT_READ|PROT_WRITE|PROT_EXEC; break;
  default:
    ShouldNotReachHere();
  }
  // is_committed is unused.
  return linux_mprotect(addr, bytes, p);
}


int SafepointSynchronize::synchronize_threads(jlong safepoint_limit_time, int nof_threads, int* initial_running)
{
  JavaThreadIteratorWithHandle jtiwh;

#ifdef ASSERT
  for (; JavaThread *cur = jtiwh.next(); ) {
    assert(cur->safepoint_state()->is_running(), "Illegal initial state");
  }
  jtiwh.rewind();
#endif // ASSERT

  // Iterate through all threads until it has been determined how to stop them all at a safepoint.
  int still_running = nof_threads;
  ThreadSafepointState *tss_head = NULL;
  ThreadSafepointState **p_prev = &tss_head;
  for (; JavaThread *cur = jtiwh.next(); ) {
    ThreadSafepointState *cur_tss = cur->safepoint_state();
    assert(cur_tss->get_next() == NULL, "Must be NULL");
    if (thread_not_running(cur_tss)) {
      --still_running;
    } else {
      *p_prev = cur_tss;
      p_prev = cur_tss->next_ptr();
    }
  }
  *p_prev = NULL;

  DEBUG_ONLY(assert_list_is_valid(tss_head, still_running);)

  *initial_running = still_running;

  // If there is no thread still running, we are already done.
  if (still_running <= 0) {
    assert(tss_head == NULL, "Must be empty");
    return 1;
  }

  int iterations = 1; // The first iteration is above.
  int64_t start_time = os::javaTimeNanos();

  do {
    // Check if this has taken too long:
    if (SafepointTimeout && safepoint_limit_time < os::javaTimeNanos()) {
      print_safepoint_timeout();
    }

    p_prev = &tss_head;
    ThreadSafepointState *cur_tss = tss_head;
    while (cur_tss != NULL) {
      assert(cur_tss->is_running(), "Illegal initial state");
      if (thread_not_running(cur_tss)) {
        --still_running;
        *p_prev = NULL;
        ThreadSafepointState *tmp = cur_tss;
        cur_tss = cur_tss->get_next();
        tmp->set_next(NULL);
      } else {
        *p_prev = cur_tss;
        p_prev = cur_tss->next_ptr();
        cur_tss = cur_tss->get_next();
      }
    }

    DEBUG_ONLY(assert_list_is_valid(tss_head, still_running);)

    if (still_running > 0) {
      back_off(start_time);
    }

    iterations++;
  } while (still_running > 0);

  assert(tss_head == NULL, "Must be empty");

  return iterations;
}

void SafepointSynchronize::arm_safepoint() {
  // Begin the process of bringing the system to a safepoint.
  // Java threads can be in several different states and are
  // stopped by different mechanisms:
  //
  //  1. Running interpreted
  //     When executing branching/returning byte codes interpreter
  //     checks if the poll is armed, if so blocks in SS::block().
  //  2. Running in native code
  //     When returning from the native code, a Java thread must check
  //     the safepoint _state to see if we must block.  If the
  //     VM thread sees a Java thread in native, it does
  //     not wait for this thread to block.  The order of the memory
  //     writes and reads of both the safepoint state and the Java
  //     threads state is critical.  In order to guarantee that the
  //     memory writes are serialized with respect to each other,
  //     the VM thread issues a memory barrier instruction.
  //  3. Running compiled Code
  //     Compiled code reads the local polling page that
  //     is set to fault if we are trying to get to a safepoint.
  //  4. Blocked
  //     A thread which is blocked will not be allowed to return from the
  //     block condition until the safepoint operation is complete.
  //  5. In VM or Transitioning between states
  //     If a Java thread is currently running in the VM or transitioning
  //     between states, the safepointing code will poll the thread state
  //     until the thread blocks itself when it attempts transitions to a
  //     new state or locking a safepoint checked monitor.

  // We must never miss a thread with correct safepoint id, so we must make sure we arm
  // the wait barrier for the next safepoint id/counter.
  // Arming must be done after resetting _current_jni_active_count, _waiting_to_block.
  _wait_barrier->arm(static_cast<int>(_safepoint_counter + 1));

  assert((_safepoint_counter & 0x1) == 0, "must be even");
  // The store to _safepoint_counter must happen after any stores in arming.
  Atomic::release_store(&_safepoint_counter, _safepoint_counter + 1);

  // We are synchronizing
  OrderAccess::storestore(); // Ordered with _safepoint_counter
  _state = _synchronizing;

  // Arming the per thread poll while having _state != _not_synchronized means safepointing
  log_trace(safepoint)("Setting thread local yield flag for threads");
  OrderAccess::storestore(); // storestore, global state -> local state
  for (JavaThreadIteratorWithHandle jtiwh; JavaThread *cur = jtiwh.next(); ) {
    // Make sure the threads start polling, it is time to yield.
    SafepointMechanism::arm_local_poll(cur);
  }

  OrderAccess::fence(); // storestore|storeload, global state -> local state
}

// Roll all threads forward to a safepoint and suspend them all
void SafepointSynchronize::begin() {
  assert(Thread::current()->is_VM_thread(), "Only VM thread may execute a safepoint");

  EventSafepointBegin begin_event;
  SafepointTracing::begin(VMThread::vm_op_type());

  Universe::heap()->safepoint_synchronize_begin();

  // By getting the Threads_lock, we assure that no threads are about to start or
  // exit. It is released again in SafepointSynchronize::end().
  Threads_lock->lock();

  assert( _state == _not_synchronized, "trying to safepoint synchronize with wrong state");

  int nof_threads = Threads::number_of_threads();

  _nof_threads_hit_polling_page = 0;

  log_debug(safepoint)("Safepoint synchronization initiated using %s wait barrier. (%d threads)", _wait_barrier->description(), nof_threads);

  // Reset the count of active JNI critical threads
  _current_jni_active_count = 0;

  // Set number of threads to wait for
  _waiting_to_block = nof_threads;

  jlong safepoint_limit_time = 0;
  if (SafepointTimeout) {
    // Set the limit time, so that it can be compared to see if this has taken
    // too long to complete.
    safepoint_limit_time = SafepointTracing::start_of_safepoint() + (jlong)SafepointTimeoutDelay * (NANOUNITS / MILLIUNITS);
    timeout_error_printed = false;
  }

  EventSafepointStateSynchronization sync_event;
  int initial_running = 0;

  // Arms the safepoint, _current_jni_active_count and _waiting_to_block must be set before.
  arm_safepoint();

  // Will spin until all threads are safe.
  int iterations = synchronize_threads(safepoint_limit_time, nof_threads, &initial_running);
  assert(_waiting_to_block == 0, "No thread should be running");

#ifndef PRODUCT
  // Mark all threads
  if (VerifyCrossModifyFence) {
    JavaThreadIteratorWithHandle jtiwh;
    for (; JavaThread *cur = jtiwh.next(); ) {
      cur->set_requires_cross_modify_fence(true);
    }
  }

  if (safepoint_limit_time != 0) {
    jlong current_time = os::javaTimeNanos();
    if (safepoint_limit_time < current_time) {
      log_warning(safepoint)("# SafepointSynchronize: Finished after "
                    INT64_FORMAT_W(6) " ms",
                    (int64_t)(current_time - SafepointTracing::start_of_safepoint()) / (NANOUNITS / MILLIUNITS));
    }
  }
#endif

  assert(Threads_lock->owned_by_self(), "must hold Threads_lock");

  // Record state
  _state = _synchronized;

  OrderAccess::fence();

  // Set the new id
  ++_safepoint_id;

#ifdef ASSERT
  // Make sure all the threads were visited.
  for (JavaThreadIteratorWithHandle jtiwh; JavaThread *cur = jtiwh.next(); ) {
    assert(cur->was_visited_for_critical_count(_safepoint_counter), "missed a thread");
  }
#endif // ASSERT

  // Update the count of active JNI critical regions
  GCLocker::set_jni_lock_count(_current_jni_active_count);

  post_safepoint_synchronize_event(sync_event,
                                   _safepoint_id,
                                   initial_running,
                                   _waiting_to_block, iterations);

  SafepointTracing::synchronized(nof_threads, initial_running, _nof_threads_hit_polling_page);

  // We do the safepoint cleanup first since a GC related safepoint
  // needs cleanup to be completed before running the GC op.
  EventSafepointCleanup cleanup_event;
  do_cleanup_tasks();
  post_safepoint_cleanup_event(cleanup_event, _safepoint_id);

  post_safepoint_begin_event(begin_event, _safepoint_id, nof_threads, _current_jni_active_count);
  SafepointTracing::cleanup();
}

void SafepointMechanism::default_initialize() {
  // Poll bit values
  _poll_word_armed_value    = poll_bit();
  _poll_word_disarmed_value = ~_poll_word_armed_value;

  bool poll_bit_only = false;

#ifdef USE_POLL_BIT_ONLY
  poll_bit_only = USE_POLL_BIT_ONLY;
#endif

  if (poll_bit_only) {
    _poll_page_armed_value    = poll_bit();
    _poll_page_disarmed_value = 0;
  } else {
    // Polling page
    const size_t page_size = os::vm_page_size();
    const size_t allocation_size = 2 * page_size;
    char* polling_page = os::reserve_memory(allocation_size);
    os::commit_memory_or_exit(polling_page, allocation_size, false, "Unable to commit Safepoint polling page");
    MemTracker::record_virtual_memory_type((address)polling_page, mtSafepoint);

    char* bad_page  = polling_page;
    char* good_page = polling_page + page_size;

    os::protect_memory(bad_page,  page_size, os::MEM_PROT_NONE);
    os::protect_memory(good_page, page_size, os::MEM_PROT_READ);

    log_info(os)("SafePoint Polling address, bad (protected) page:" INTPTR_FORMAT ", good (unprotected) page:" INTPTR_FORMAT, p2i(bad_page), p2i(good_page));

    // Poll address values
    _poll_page_armed_value    = reinterpret_cast<uintptr_t>(bad_page);
    _poll_page_disarmed_value = reinterpret_cast<uintptr_t>(good_page);
    _polling_page = (address)bad_page;
  }
}

void SafepointMechanism::arm_local_poll(JavaThread* thread) {
  thread->poll_data()->set_polling_word(_poll_word_armed_value);
  thread->poll_data()->set_polling_page(_poll_page_armed_value);
}

// Caller is responsible for using a memory barrier if needed.
inline void SafepointMechanism::ThreadData::set_polling_page(uintptr_t poll_value) {
  Atomic::store(&_polling_page, poll_value);
}