checkpoint.c

linux 内核源代码

/*
 * linux/fs/jbd2/checkpoint.c
 *
 * Written by Stephen C. Tweedie <[e-mail]>, 1999
 *
 * Copyright 1999 Red Hat Software --- All Rights Reserved
 *
 * This file is part of the Linux kernel and is made available under
 * the terms of the GNU General Public License, version 2, or at your
 * option, any later version, incorporated herein by reference.
 *
 * Checkpoint routines for the generic filesystem journaling code.
 * Part of the ext2fs journaling system.
 *
 * Checkpointing is the process of ensuring that a section of the log is
 * committed fully to disk, so that that portion of the log can be
 * reused.
 */

#include <linux/time.h>
#include <linux/fs.h>
#include <linux/jbd2.h>
#include <linux/errno.h>
#include <linux/slab.h>

/*
 * Unlink a buffer from a transaction checkpoint list.
 *
 * Called with j_list_lock held.
 */
static inline void __buffer_unlink_first(struct journal_head *jh)
{
	transaction_t *transaction = jh->b_cp_transaction;

	jh->b_cpnext->b_cpprev = jh->b_cpprev;
	jh->b_cpprev->b_cpnext = jh->b_cpnext;
	if (transaction->t_checkpoint_list == jh) {
		transaction->t_checkpoint_list = jh->b_cpnext;
		if (transaction->t_checkpoint_list == jh)
			transaction->t_checkpoint_list = NULL;
	}
}

/*
 * Unlink a buffer from a transaction checkpoint(io) list.
 *
 * Called with j_list_lock held.
 */
static inline void __buffer_unlink(struct journal_head *jh)
{
	transaction_t *transaction = jh->b_cp_transaction;

	__buffer_unlink_first(jh);
	if (transaction->t_checkpoint_io_list == jh) {
		transaction->t_checkpoint_io_list = jh->b_cpnext;
		if (transaction->t_checkpoint_io_list == jh)
			transaction->t_checkpoint_io_list = NULL;
	}
}

/*
 * Move a buffer from the checkpoint list to the checkpoint io list
 *
 * Called with j_list_lock held
 */
static inline void __buffer_relink_io(struct journal_head *jh)
{
	transaction_t *transaction = jh->b_cp_transaction;

	__buffer_unlink_first(jh);

	if (!transaction->t_checkpoint_io_list) {
		jh->b_cpnext = jh->b_cpprev = jh;
	} else {
		jh->b_cpnext = transaction->t_checkpoint_io_list;
		jh->b_cpprev = transaction->t_checkpoint_io_list->b_cpprev;
		jh->b_cpprev->b_cpnext = jh;
		jh->b_cpnext->b_cpprev = jh;
	}
	transaction->t_checkpoint_io_list = jh;
}

/*
 * Try to release a checkpointed buffer from its transaction.
 * Returns 1 if we released it and 2 if we also released the
 * whole transaction.
 *
 * Requires j_list_lock
 * Called under jbd_lock_bh_state(jh2bh(jh)), and drops it
 */
static int __try_to_free_cp_buf(struct journal_head *jh)
{
	int ret = 0;
	struct buffer_head *bh = jh2bh(jh);

	if (jh->b_jlist == BJ_None && !buffer_locked(bh) && !buffer_dirty(bh)) {
		JBUFFER_TRACE(jh, "remove from checkpoint list");
		ret = __jbd2_journal_remove_checkpoint(jh) + 1;
		jbd_unlock_bh_state(bh);
		jbd2_journal_remove_journal_head(bh);
		BUFFER_TRACE(bh, "release");
		__brelse(bh);
	} else {
		jbd_unlock_bh_state(bh);
	}
	return ret;
}

/*
 * __jbd2_log_wait_for_space: wait until there is space in the journal.
 *
 * Called under j-state_lock *only*.  It will be unlocked if we have to wait
 * for a checkpoint to free up some space in the log.
 */
void __jbd2_log_wait_for_space(journal_t *journal)
{
	int nblocks;
	assert_spin_locked(&journal->j_state_lock);

	nblocks = jbd_space_needed(journal);
	while (__jbd2_log_space_left(journal) < nblocks) {
		if (journal->j_flags & JBD2_ABORT)
			return;
		spin_unlock(&journal->j_state_lock);
		mutex_lock(&journal->j_checkpoint_mutex);

		/*
		 * Test again, another process may have checkpointed while we
		 * were waiting for the checkpoint lock
		 */
		spin_lock(&journal->j_state_lock);
		nblocks = jbd_space_needed(journal);
		if (__jbd2_log_space_left(journal) < nblocks) {
			spin_unlock(&journal->j_state_lock);
			jbd2_log_do_checkpoint(journal);
			spin_lock(&journal->j_state_lock);
		}
		mutex_unlock(&journal->j_checkpoint_mutex);
	}
}

/*
 * We were unable to perform jbd_trylock_bh_state() inside j_list_lock.
 * The caller must restart a list walk.  Wait for someone else to run
 * jbd_unlock_bh_state().
 */
static void jbd_sync_bh(journal_t *journal, struct buffer_head *bh)
	__releases(journal->j_list_lock)
{
	get_bh(bh);
	spin_unlock(&journal->j_list_lock);
	jbd_lock_bh_state(bh);
	jbd_unlock_bh_state(bh);
	put_bh(bh);
}

/*
 * Clean up transaction's list of buffers submitted for io.
 * We wait for any pending IO to complete and remove any clean
 * buffers. Note that we take the buffers in the opposite ordering
 * from the one in which they were submitted for IO.
 *
 * Called with j_list_lock held.
 */
static void __wait_cp_io(journal_t *journal, transaction_t *transaction)
{
	struct journal_head *jh;
	struct buffer_head *bh;
	tid_t this_tid;
	int released = 0;

	this_tid = transaction->t_tid;
restart:
	/* Did somebody clean up the transaction in the meanwhile? */
	if (journal->j_checkpoint_transactions != transaction ||
			transaction->t_tid != this_tid)
		return;
	while (!released && transaction->t_checkpoint_io_list) {
		jh = transaction->t_checkpoint_io_list;
		bh = jh2bh(jh);
		if (!jbd_trylock_bh_state(bh)) {
			jbd_sync_bh(journal, bh);
			spin_lock(&journal->j_list_lock);
			goto restart;
		}
		if (buffer_locked(bh)) {
			atomic_inc(&bh->b_count);
			spin_unlock(&journal->j_list_lock);
			jbd_unlock_bh_state(bh);
			wait_on_buffer(bh);
			/* the journal_head may have gone by now */
			BUFFER_TRACE(bh, "brelse");
			__brelse(bh);
			spin_lock(&journal->j_list_lock);
			goto restart;
		}
		/*
		 * Now in whatever state the buffer currently is, we know that
		 * it has been written out and so we can drop it from the list
		 */
		released = __jbd2_journal_remove_checkpoint(jh);
		jbd_unlock_bh_state(bh);
		jbd2_journal_remove_journal_head(bh);
		__brelse(bh);
	}
}

#define NR_BATCH	64

static void
__flush_batch(journal_t *journal, struct buffer_head **bhs, int *batch_count)
{
	int i;

	ll_rw_block(SWRITE, *batch_count, bhs);
	for (i = 0; i < *batch_count; i++) {
		struct buffer_head *bh = bhs[i];
		clear_buffer_jwrite(bh);
		BUFFER_TRACE(bh, "brelse");
		__brelse(bh);
	}
	*batch_count = 0;
}

/*
 * Try to flush one buffer from the checkpoint list to disk.
 *
 * Return 1 if something happened which requires us to abort the current
 * scan of the checkpoint list.
 *
 * Called with j_list_lock held and drops it if 1 is returned
 * Called under jbd_lock_bh_state(jh2bh(jh)), and drops it
 */
static int __process_buffer(journal_t *journal, struct journal_head *jh,
			struct buffer_head **bhs, int *batch_count)
{
	struct buffer_head *bh = jh2bh(jh);
	int ret = 0;

	if (buffer_locked(bh)) {
		atomic_inc(&bh->b_count);
		spin_unlock(&journal->j_list_lock);
		jbd_unlock_bh_state(bh);
		wait_on_buffer(bh);
		/* the journal_head may have gone by now */
		BUFFER_TRACE(bh, "brelse");
		__brelse(bh);
		ret = 1;
	} else if (jh->b_transaction != NULL) {
		transaction_t *t = jh->b_transaction;
		tid_t tid = t->t_tid;

		spin_unlock(&journal->j_list_lock);
		jbd_unlock_bh_state(bh);
		jbd2_log_start_commit(journal, tid);
		jbd2_log_wait_commit(journal, tid);
		ret = 1;
	} else if (!buffer_dirty(bh)) {
		J_ASSERT_JH(jh, !buffer_jbddirty(bh));
		BUFFER_TRACE(bh, "remove from checkpoint");
		__jbd2_journal_remove_checkpoint(jh);
		spin_unlock(&journal->j_list_lock);
		jbd_unlock_bh_state(bh);
		jbd2_journal_remove_journal_head(bh);
		__brelse(bh);
		ret = 1;
	} else {
		/*
		 * Important: we are about to write the buffer, and
		 * possibly block, while still holding the journal lock.
		 * We cannot afford to let the transaction logic start
		 * messing around with this buffer before we write it to
		 * disk, as that would break recoverability.
		 */
		BUFFER_TRACE(bh, "queue");
		get_bh(bh);
		J_ASSERT_BH(bh, !buffer_jwrite(bh));
		set_buffer_jwrite(bh);
		bhs[*batch_count] = bh;
		__buffer_relink_io(jh);
		jbd_unlock_bh_state(bh);
		(*batch_count)++;
		if (*batch_count == NR_BATCH) {
			spin_unlock(&journal->j_list_lock);
			__flush_batch(journal, bhs, batch_count);
			ret = 1;
		}
	}
	return ret;
}

/*
 * Perform an actual checkpoint. We take the first transaction on the
 * list of transactions to be checkpointed and send all its buffers
 * to disk. We submit larger chunks of data at once.
 *
 * The journal should be locked before calling this function.
 */
int jbd2_log_do_checkpoint(journal_t *journal)
{
	transaction_t *transaction;
	tid_t this_tid;
	int result;

	jbd_debug(1, "Start checkpoint\n");

	/*
	 * First thing: if there are any transactions in the log which
	 * don't need checkpointing, just eliminate them from the
	 * journal straight away.
	 */
	result = jbd2_cleanup_journal_tail(journal);
	jbd_debug(1, "cleanup_journal_tail returned %d\n", result);
	if (result <= 0)
		return result;

	/*
	 * OK, we need to start writing disk blocks.  Take one transaction
	 * and write it.
	 */
	spin_lock(&journal->j_list_lock);
	if (!journal->j_checkpoint_transactions)
		goto out;
	transaction = journal->j_checkpoint_transactions;
	this_tid = transaction->t_tid;
restart:
	/*
	 * If someone cleaned up this transaction while we slept, we're
	 * done (maybe it's a new transaction, but it fell at the same
	 * address).
	 */
	if (journal->j_checkpoint_transactions == transaction &&
			transaction->t_tid == this_tid) {
		int batch_count = 0;
		struct buffer_head *bhs[NR_BATCH];
		struct journal_head *jh;
		int retry = 0;

		while (!retry && transaction->t_checkpoint_list) {
			struct buffer_head *bh;

			jh = transaction->t_checkpoint_list;
			bh = jh2bh(jh);
			if (!jbd_trylock_bh_state(bh)) {
				jbd_sync_bh(journal, bh);
				retry = 1;
				break;
			}
			retry = __process_buffer(journal, jh, bhs,&batch_count);