linux/fs/jbd2/revoke.c

// SPDX-License-Identifier: GPL-2.0+
/*
 * linux/fs/jbd2/revoke.c
 *
 * Written by Stephen C. Tweedie <[email protected]>, 2000
 *
 * Copyright 2000 Red Hat corp --- All Rights Reserved
 *
 * Journal revoke routines for the generic filesystem journaling code;
 * part of the ext2fs journaling system.
 *
 * Revoke is the mechanism used to prevent old log records for deleted
 * metadata from being replayed on top of newer data using the same
 * blocks.  The revoke mechanism is used in two separate places:
 *
 * + Commit: during commit we write the entire list of the current
 *   transaction's revoked blocks to the journal
 *
 * + Recovery: during recovery we record the transaction ID of all
 *   revoked blocks.  If there are multiple revoke records in the log
 *   for a single block, only the last one counts, and if there is a log
 *   entry for a block beyond the last revoke, then that log entry still
 *   gets replayed.
 *
 * We can get interactions between revokes and new log data within a
 * single transaction:
 *
 * Block is revoked and then journaled:
 *   The desired end result is the journaling of the new block, so we
 *   cancel the revoke before the transaction commits.
 *
 * Block is journaled and then revoked:
 *   The revoke must take precedence over the write of the block, so we
 *   need either to cancel the journal entry or to write the revoke
 *   later in the log than the log block.  In this case, we choose the
 *   latter: journaling a block cancels any revoke record for that block
 *   in the current transaction, so any revoke for that block in the
 *   transaction must have happened after the block was journaled and so
 *   the revoke must take precedence.
 *
 * Block is revoked and then written as data:
 *   The data write is allowed to succeed, but the revoke is _not_
 *   cancelled.  We still need to prevent old log records from
 *   overwriting the new data.  We don't even need to clear the revoke
 *   bit here.
 *
 * We cache revoke status of a buffer in the current transaction in b_states
 * bits.  As the name says, revokevalid flag indicates that the cached revoke
 * status of a buffer is valid and we can rely on the cached status.
 *
 * Revoke information on buffers is a tri-state value:
 *
 * RevokeValid clear:	no cached revoke status, need to look it up
 * RevokeValid set, Revoked clear:
 *			buffer has not been revoked, and cancel_revoke
 *			need do nothing.
 * RevokeValid set, Revoked set:
 *			buffer has been revoked.
 *
 * Locking rules:
 * We keep two hash tables of revoke records. One hashtable belongs to the
 * running transaction (is pointed to by journal->j_revoke), the other one
 * belongs to the committing transaction. Accesses to the second hash table
 * happen only from the kjournald and no other thread touches this table.  Also
 * journal_switch_revoke_table() which switches which hashtable belongs to the
 * running and which to the committing transaction is called only from
 * kjournald. Therefore we need no locks when accessing the hashtable belonging
 * to the committing transaction.
 *
 * All users operating on the hash table belonging to the running transaction
 * have a handle to the transaction. Therefore they are safe from kjournald
 * switching hash tables under them. For operations on the lists of entries in
 * the hash table j_revoke_lock is used.
 *
 * Finally, also replay code uses the hash tables but at this moment no one else
 * can touch them (filesystem isn't mounted yet) and hence no locking is
 * needed.
 */

#ifndef __KERNEL__
#include "jfs_user.h"
#else
#include <linux/time.h>
#include <linux/fs.h>
#include <linux/jbd2.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/list.h>
#include <linux/init.h>
#include <linux/bio.h>
#include <linux/log2.h>
#include <linux/hash.h>
#endif

static struct kmem_cache *jbd2_revoke_record_cache;
static struct kmem_cache *jbd2_revoke_table_cache;

/* Each revoke record represents one single revoked block.  During
   journal replay, this involves recording the transaction ID of the
   last transaction to revoke this block. */

struct jbd2_revoke_record_s
{ … };


/* The revoke table is just a simple hash table of revoke records. */
struct jbd2_revoke_table_s
{ … };


#ifdef __KERNEL__
static void write_one_revoke_record(transaction_t *,
				    struct list_head *,
				    struct buffer_head **, int *,
				    struct jbd2_revoke_record_s *);
static void flush_descriptor(journal_t *, struct buffer_head *, int);
#endif

/* Utility functions to maintain the revoke table */

static inline int hash(journal_t *journal, unsigned long long block)
{ … }

static int insert_revoke_hash(journal_t *journal, unsigned long long blocknr,
			      tid_t seq)
{ … }

/* Find a revoke record in the journal's hash table. */

static struct jbd2_revoke_record_s *find_revoke_record(journal_t *journal,
						      unsigned long long blocknr)
{ … }

void jbd2_journal_destroy_revoke_record_cache(void)
{ … }

void jbd2_journal_destroy_revoke_table_cache(void)
{ … }

int __init jbd2_journal_init_revoke_record_cache(void)
{ … }

int __init jbd2_journal_init_revoke_table_cache(void)
{ … }

static struct jbd2_revoke_table_s *jbd2_journal_init_revoke_table(int hash_size)
{ … }

static void jbd2_journal_destroy_revoke_table(struct jbd2_revoke_table_s *table)
{ … }

/* Initialise the revoke table for a given journal to a given size. */
int jbd2_journal_init_revoke(journal_t *journal, int hash_size)
{ … }

/* Destroy a journal's revoke table.  The table must already be empty! */
void jbd2_journal_destroy_revoke(journal_t *journal)
{ … }


#ifdef __KERNEL__

/*
 * jbd2_journal_revoke: revoke a given buffer_head from the journal.  This
 * prevents the block from being replayed during recovery if we take a
 * crash after this current transaction commits.  Any subsequent
 * metadata writes of the buffer in this transaction cancel the
 * revoke.
 *
 * Note that this call may block --- it is up to the caller to make
 * sure that there are no further calls to journal_write_metadata
 * before the revoke is complete.  In ext3, this implies calling the
 * revoke before clearing the block bitmap when we are deleting
 * metadata.
 *
 * Revoke performs a jbd2_journal_forget on any buffer_head passed in as a
 * parameter, but does _not_ forget the buffer_head if the bh was only
 * found implicitly.
 *
 * bh_in may not be a journalled buffer - it may have come off
 * the hash tables without an attached journal_head.
 *
 * If bh_in is non-zero, jbd2_journal_revoke() will decrement its b_count
 * by one.
 */

int jbd2_journal_revoke(handle_t *handle, unsigned long long blocknr,
		   struct buffer_head *bh_in)
{ … }

/*
 * Cancel an outstanding revoke.  For use only internally by the
 * journaling code (called from jbd2_journal_get_write_access).
 *
 * We trust buffer_revoked() on the buffer if the buffer is already
 * being journaled: if there is no revoke pending on the buffer, then we
 * don't do anything here.
 *
 * This would break if it were possible for a buffer to be revoked and
 * discarded, and then reallocated within the same transaction.  In such
 * a case we would have lost the revoked bit, but when we arrived here
 * the second time we would still have a pending revoke to cancel.  So,
 * do not trust the Revoked bit on buffers unless RevokeValid is also
 * set.
 */
int jbd2_journal_cancel_revoke(handle_t *handle, struct journal_head *jh)
{ … }

/*
 * journal_clear_revoked_flag clears revoked flag of buffers in
 * revoke table to reflect there is no revoked buffers in the next
 * transaction which is going to be started.
 */
void jbd2_clear_buffer_revoked_flags(journal_t *journal)
{ … }

/* journal_switch_revoke table select j_revoke for next transaction
 * we do not want to suspend any processing until all revokes are
 * written -bzzz
 */
void jbd2_journal_switch_revoke_table(journal_t *journal)
{ … }

/*
 * Write revoke records to the journal for all entries in the current
 * revoke hash, deleting the entries as we go.
 */
void jbd2_journal_write_revoke_records(transaction_t *transaction,
				       struct list_head *log_bufs)
{ … }

/*
 * Write out one revoke record.  We need to create a new descriptor
 * block if the old one is full or if we have not already created one.
 */

static void write_one_revoke_record(transaction_t *transaction,
				    struct list_head *log_bufs,
				    struct buffer_head **descriptorp,
				    int *offsetp,
				    struct jbd2_revoke_record_s *record)
{ … }

/*
 * Flush a revoke descriptor out to the journal.  If we are aborting,
 * this is a noop; otherwise we are generating a buffer which needs to
 * be waited for during commit, so it has to go onto the appropriate
 * journal buffer list.
 */

static void flush_descriptor(journal_t *journal,
			     struct buffer_head *descriptor,
			     int offset)
{ … }
#endif

/*
 * Revoke support for recovery.
 *
 * Recovery needs to be able to:
 *
 *  record all revoke records, including the tid of the latest instance
 *  of each revoke in the journal
 *
 *  check whether a given block in a given transaction should be replayed
 *  (ie. has not been revoked by a revoke record in that or a subsequent
 *  transaction)
 *
 *  empty the revoke table after recovery.
 */

/*
 * First, setting revoke records.  We create a new revoke record for
 * every block ever revoked in the log as we scan it for recovery, and
 * we update the existing records if we find multiple revokes for a
 * single block.
 */

int jbd2_journal_set_revoke(journal_t *journal,
		       unsigned long long blocknr,
		       tid_t sequence)
{ … }

/*
 * Test revoke records.  For a given block referenced in the log, has
 * that block been revoked?  A revoke record with a given transaction
 * sequence number revokes all blocks in that transaction and earlier
 * ones, but later transactions still need replayed.
 */

int jbd2_journal_test_revoke(journal_t *journal,
			unsigned long long blocknr,
			tid_t sequence)
{ … }

/*
 * Finally, once recovery is over, we need to clear the revoke table so
 * that it can be reused by the running filesystem.
 */

void jbd2_journal_clear_revoke(journal_t *journal)
{ … }