proc.c

postgresql8.3.4源码,开源数据库

			pfree(buf.data);
		}
	} while (myWaitStatus == STATUS_WAITING);

	/*
	 * Disable the timer, if it's still running
	 */
	if (!disable_sig_alarm(false))
		elog(FATAL, "could not disable timer for process wakeup");

	/*
	 * Re-acquire the lock table's partition lock.  We have to do this to hold
	 * off cancel/die interrupts before we can mess with lockAwaited (else we
	 * might have a missed or duplicated locallock update).
	 */
	LWLockAcquire(partitionLock, LW_EXCLUSIVE);

	/*
	 * We no longer want LockWaitCancel to do anything.
	 */
	lockAwaited = NULL;

	/*
	 * If we got the lock, be sure to remember it in the locallock table.
	 */
	if (MyProc->waitStatus == STATUS_OK)
		GrantAwaitedLock();

	/*
	 * We don't have to do anything else, because the awaker did all the
	 * necessary update of the lock table and MyProc.
	 */
	return MyProc->waitStatus;
}


/*
 * ProcWakeup -- wake up a process by releasing its private semaphore.
 *
 *	 Also remove the process from the wait queue and set its links invalid.
 *	 RETURN: the next process in the wait queue.
 *
 * The appropriate lock partition lock must be held by caller.
 *
 * XXX: presently, this code is only used for the "success" case, and only
 * works correctly for that case.  To clean up in failure case, would need
 * to twiddle the lock's request counts too --- see RemoveFromWaitQueue.
 * Hence, in practice the waitStatus parameter must be STATUS_OK.
 */
PGPROC *
ProcWakeup(PGPROC *proc, int waitStatus)
{
	PGPROC	   *retProc;

	/* Proc should be sleeping ... */
	if (proc->links.prev == INVALID_OFFSET ||
		proc->links.next == INVALID_OFFSET)
		return NULL;
	Assert(proc->waitStatus == STATUS_WAITING);

	/* Save next process before we zap the list link */
	retProc = (PGPROC *) MAKE_PTR(proc->links.next);

	/* Remove process from wait queue */
	SHMQueueDelete(&(proc->links));
	(proc->waitLock->waitProcs.size)--;

	/* Clean up process' state and pass it the ok/fail signal */
	proc->waitLock = NULL;
	proc->waitProcLock = NULL;
	proc->waitStatus = waitStatus;

	/* And awaken it */
	PGSemaphoreUnlock(&proc->sem);

	return retProc;
}

/*
 * ProcLockWakeup -- routine for waking up processes when a lock is
 *		released (or a prior waiter is aborted).  Scan all waiters
 *		for lock, waken any that are no longer blocked.
 *
 * The appropriate lock partition lock must be held by caller.
 */
void
ProcLockWakeup(LockMethod lockMethodTable, LOCK *lock)
{
	PROC_QUEUE *waitQueue = &(lock->waitProcs);
	int			queue_size = waitQueue->size;
	PGPROC	   *proc;
	LOCKMASK	aheadRequests = 0;

	Assert(queue_size >= 0);

	if (queue_size == 0)
		return;

	proc = (PGPROC *) MAKE_PTR(waitQueue->links.next);

	while (queue_size-- > 0)
	{
		LOCKMODE	lockmode = proc->waitLockMode;

		/*
		 * Waken if (a) doesn't conflict with requests of earlier waiters, and
		 * (b) doesn't conflict with already-held locks.
		 */
		if ((lockMethodTable->conflictTab[lockmode] & aheadRequests) == 0 &&
			LockCheckConflicts(lockMethodTable,
							   lockmode,
							   lock,
							   proc->waitProcLock,
							   proc) == STATUS_OK)
		{
			/* OK to waken */
			GrantLock(lock, proc->waitProcLock, lockmode);
			proc = ProcWakeup(proc, STATUS_OK);

			/*
			 * ProcWakeup removes proc from the lock's waiting process queue
			 * and returns the next proc in chain; don't use proc's next-link,
			 * because it's been cleared.
			 */
		}
		else
		{
			/*
			 * Cannot wake this guy. Remember his request for later checks.
			 */
			aheadRequests |= LOCKBIT_ON(lockmode);
			proc = (PGPROC *) MAKE_PTR(proc->links.next);
		}
	}

	Assert(waitQueue->size >= 0);
}

/*
 * CheckDeadLock
 *
 * We only get to this routine if we got SIGALRM after DeadlockTimeout
 * while waiting for a lock to be released by some other process.  Look
 * to see if there's a deadlock; if not, just return and continue waiting.
 * (But signal ProcSleep to log a message, if log_lock_waits is true.)
 * If we have a real deadlock, remove ourselves from the lock's wait queue
 * and signal an error to ProcSleep.
 *
 * NB: this is run inside a signal handler, so be very wary about what is done
 * here or in called routines.
 */
static void
CheckDeadLock(void)
{
	int			i;

	/*
	 * Acquire exclusive lock on the entire shared lock data structures. Must
	 * grab LWLocks in partition-number order to avoid LWLock deadlock.
	 *
	 * Note that the deadlock check interrupt had better not be enabled
	 * anywhere that this process itself holds lock partition locks, else this
	 * will wait forever.  Also note that LWLockAcquire creates a critical
	 * section, so that this routine cannot be interrupted by cancel/die
	 * interrupts.
	 */
	for (i = 0; i < NUM_LOCK_PARTITIONS; i++)
		LWLockAcquire(FirstLockMgrLock + i, LW_EXCLUSIVE);

	/*
	 * Check to see if we've been awoken by anyone in the interim.
	 *
	 * If we have we can return and resume our transaction -- happy day.
	 * Before we are awoken the process releasing the lock grants it to us so
	 * we know that we don't have to wait anymore.
	 *
	 * We check by looking to see if we've been unlinked from the wait queue.
	 * This is quicker than checking our semaphore's state, since no kernel
	 * call is needed, and it is safe because we hold the lock partition lock.
	 */
	if (MyProc->links.prev == INVALID_OFFSET ||
		MyProc->links.next == INVALID_OFFSET)
		goto check_done;

#ifdef LOCK_DEBUG
	if (Debug_deadlocks)
		DumpAllLocks();
#endif

	/* Run the deadlock check, and set deadlock_state for use by ProcSleep */
	deadlock_state = DeadLockCheck(MyProc);

	if (deadlock_state == DS_HARD_DEADLOCK)
	{
		/*
		 * Oops.  We have a deadlock.
		 *
		 * Get this process out of wait state. (Note: we could do this more
		 * efficiently by relying on lockAwaited, but use this coding to
		 * preserve the flexibility to kill some other transaction than the
		 * one detecting the deadlock.)
		 *
		 * RemoveFromWaitQueue sets MyProc->waitStatus to STATUS_ERROR, so
		 * ProcSleep will report an error after we return from the signal
		 * handler.
		 */
		Assert(MyProc->waitLock != NULL);
		RemoveFromWaitQueue(MyProc, LockTagHashCode(&(MyProc->waitLock->tag)));

		/*
		 * Unlock my semaphore so that the interrupted ProcSleep() call can
		 * finish.
		 */
		PGSemaphoreUnlock(&MyProc->sem);

		/*
		 * We're done here.  Transaction abort caused by the error that
		 * ProcSleep will raise will cause any other locks we hold to be
		 * released, thus allowing other processes to wake up; we don't need
		 * to do that here.  NOTE: an exception is that releasing locks we
		 * hold doesn't consider the possibility of waiters that were blocked
		 * behind us on the lock we just failed to get, and might now be
		 * wakable because we're not in front of them anymore.  However,
		 * RemoveFromWaitQueue took care of waking up any such processes.
		 */
	}
	else if (log_lock_waits || deadlock_state == DS_BLOCKED_BY_AUTOVACUUM)
	{
		/*
		 * Unlock my semaphore so that the interrupted ProcSleep() call can
		 * print the log message (we daren't do it here because we are inside
		 * a signal handler).  It will then sleep again until someone releases
		 * the lock.
		 *
		 * If blocked by autovacuum, this wakeup will enable ProcSleep to send
		 * the cancelling signal to the autovacuum worker.
		 */
		PGSemaphoreUnlock(&MyProc->sem);
	}

	/*
	 * And release locks.  We do this in reverse order for two reasons: (1)
	 * Anyone else who needs more than one of the locks will be trying to lock
	 * them in increasing order; we don't want to release the other process
	 * until it can get all the locks it needs. (2) This avoids O(N^2)
	 * behavior inside LWLockRelease.
	 */
check_done:
	for (i = NUM_LOCK_PARTITIONS; --i >= 0;)
		LWLockRelease(FirstLockMgrLock + i);
}


/*
 * ProcWaitForSignal - wait for a signal from another backend.
 *
 * This can share the semaphore normally used for waiting for locks,
 * since a backend could never be waiting for a lock and a signal at
 * the same time.  As with locks, it's OK if the signal arrives just
 * before we actually reach the waiting state.	Also as with locks,
 * it's necessary that the caller be robust against bogus wakeups:
 * always check that the desired state has occurred, and wait again
 * if not.	This copes with possible "leftover" wakeups.
 */
void
ProcWaitForSignal(void)
{
	PGSemaphoreLock(&MyProc->sem, true);
}

/*
 * ProcSendSignal - send a signal to a backend identified by PID
 */
void
ProcSendSignal(int pid)
{
	PGPROC	   *proc = BackendPidGetProc(pid);

	if (proc != NULL)
		PGSemaphoreUnlock(&proc->sem);
}


/*****************************************************************************
 * SIGALRM interrupt support
 *
 * Maybe these should be in pqsignal.c?
 *****************************************************************************/

/*
 * Enable the SIGALRM interrupt to fire after the specified delay
 *
 * Delay is given in milliseconds.	Caller should be sure a SIGALRM
 * signal handler is installed before this is called.
 *
 * This code properly handles nesting of deadlock timeout alarms within
 * statement timeout alarms.
 *
 * Returns TRUE if okay, FALSE on failure.
 */
bool
enable_sig_alarm(int delayms, bool is_statement_timeout)
{
	TimestampTz fin_time;
	struct itimerval timeval;

	if (is_statement_timeout)
	{
		/*
		 * Begin statement-level timeout
		 *
		 * Note that we compute statement_fin_time with reference to the
		 * statement_timestamp, but apply the specified delay without any
		 * correction; that is, we ignore whatever time has elapsed since
		 * statement_timestamp was set.  In the normal case only a small
		 * interval will have elapsed and so this doesn't matter, but there
		 * are corner cases (involving multi-statement query strings with
		 * embedded COMMIT or ROLLBACK) where we might re-initialize the
		 * statement timeout long after initial receipt of the message. In
		 * such cases the enforcement of the statement timeout will be a bit
		 * inconsistent.  This annoyance is judged not worth the cost of
		 * performing an additional gettimeofday() here.
		 */
		Assert(!deadlock_timeout_active);
		fin_time = GetCurrentStatementStartTimestamp();
		fin_time = TimestampTzPlusMilliseconds(fin_time, delayms);
		statement_fin_time = fin_time;
		cancel_from_timeout = false;
		statement_timeout_active = true;
	}
	else if (statement_timeout_active)
	{
		/*
		 * Begin deadlock timeout with statement-level timeout active
		 *
		 * Here, we want to interrupt at the closer of the two timeout times.
		 * If fin_time >= statement_fin_time then we need not touch the
		 * existing timer setting; else set up to interrupt at the deadlock
		 * timeout time.
		 *
		 * NOTE: in this case it is possible that this routine will be
		 * interrupted by the previously-set timer alarm.  This is okay
		 * because the signal handler will do only what it should do according
		 * to the state variables.	The deadlock checker may get run earlier
		 * than normal, but that does no harm.
		 */
		timeout_start_time = GetCurrentTimestamp();
		fin_time = TimestampTzPlusMilliseconds(timeout_start_time, delayms);
		deadlock_timeout_active = true;
		if (fin_time >= statement_fin_time)
			return true;
	}
	else
	{
		/* Begin deadlock timeout with no statement-level timeout */
		deadlock_timeout_active = true;
		/* GetCurrentTimestamp can be expensive, so only do it if we must */
		if (log_lock_waits)
			timeout_start_time = GetCurrentTimestamp();
	}

	/* If we reach here, okay to set the timer interrupt */
	MemSet(&timeval, 0, sizeof(struct itimerval));
	timeval.it_value.tv_sec = delayms / 1000;
	timeval.it_value.tv_usec = (delayms % 1000) * 1000;
	if (setitimer(ITIMER_REAL, &timeval, NULL))
		return false;
	return true;
}

/*
 * Cancel the SIGALRM timer, either for a deadlock timeout or a statement
 * timeout.  If a deadlock timeout is canceled, any active statement timeout
 * remains in force.
 *
 * Returns TRUE if okay, FALSE on failure.
 */
bool
disable_sig_alarm(bool is_statement_timeout)
{
	/*
	 * Always disable the interrupt if it is active; this avoids being
	 * interrupted by the signal handler and thereby possibly getting
	 * confused.
	 *
	 * We will re-enable the interrupt if necessary in CheckStatementTimeout.
	 */
	if (statement_timeout_active || deadlock_timeout_active)
	{
		struct itimerval timeval;

		MemSet(&timeval, 0, sizeof(struct itimerval));
		if (setitimer(ITIMER_REAL, &timeval, NULL))
		{
			statement_timeout_active = false;
			cancel_from_timeout = false;
			deadlock_timeout_active = false;
			return false;
		}
	}

	/* Always cancel deadlock timeout, in case this is error cleanup */
	deadlock_timeout_active = false;

	/* Cancel or reschedule statement timeout */
	if (is_statement_timeout)
	{
		statement_timeout_active = false;
		cancel_from_timeout = false;
	}
	else if (statement_timeout_active)
	{
		if (!CheckStatementTimeout())
			return false;
	}
	return true;
}


/*
 * Check for statement timeout.  If the timeout time has come,
 * trigger a query-cancel interrupt; if not, reschedule the SIGALRM
 * interrupt to occur at the right time.
 *
 * Returns true if okay, false if failed to set the interrupt.
 */
static bool
CheckStatementTimeout(void)
{
	TimestampTz now;

	if (!statement_timeout_active)
		return true;			/* do nothing if not active */

	now = GetCurrentTimestamp();

	if (now >= statement_fin_time)
	{
		/* Time to die */
		statement_timeout_active = false;
		cancel_from_timeout = true;
#ifdef HAVE_SETSID
		/* try to signal whole process group */
		kill(-MyProcPid, SIGINT);
#endif
		kill(MyProcPid, SIGINT);
	}
	else
	{
		/* Not time yet, so (re)schedule the interrupt */
		long		secs;
		int			usecs;
		struct itimerval timeval;

		TimestampDifference(now, statement_fin_time,
							&secs, &usecs);

		/*
		 * It's possible that the difference is less than a microsecond;
		 * ensure we don't cancel, rather than set, the interrupt.
		 */
		if (secs == 0 && usecs == 0)
			usecs = 1;
		MemSet(&timeval, 0, sizeof(struct itimerval));
		timeval.it_value.tv_sec = secs;
		timeval.it_value.tv_usec = usecs;
		if (setitimer(ITIMER_REAL, &timeval, NULL))
			return false;
	}

	return true;
}


/*
 * Signal handler for SIGALRM
 *
 * Process deadlock check and/or statement timeout check, as needed.
 * To avoid various edge cases, we must be careful to do nothing
 * when there is nothing to be done.  We also need to be able to
 * reschedule the timer interrupt if called before end of statement.
 */
void
handle_sig_alarm(SIGNAL_ARGS)
{
	int			save_errno = errno;

	if (deadlock_timeout_active)
	{
		deadlock_timeout_active = false;
		CheckDeadLock();
	}

	if (statement_timeout_active)
		(void) CheckStatementTimeout();

	errno = save_errno;
}