revoke.c

嵌入式系统设计与实例开发实验教材二源码 多线程应用程序设计 串行端口程序设计 AD接口实验 CAN总线通信实验 GPS通信实验 Linux内核移植与编译实验 IC卡读写实验 SD驱动使

/*
 * linux/fs/revoke.c
 * 
 * Written by Stephen C. Tweedie <[e-mail]>, 2000
 *
 * Copyright 2000 Red Hat corp --- 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.
 *
 * 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.
 *
 * 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 *			buffer has not been revoked, and cancel_revoke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.  
 */

#ifndef __KERNEL__
#include "jfs_user.h"
#else
#include <linux/sched.h>
#include <linux/fs.h>
#include <linux/jbd.h>
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/locks.h>
#include <linux/list.h>
#include <linux/smp_lock.h>
#include <linux/init.h>
#endif

static kmem_cache_t *revoke_record_cache;
static kmem_cache_t *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 jbd_revoke_record_s 
{
	struct list_head  hash;
	tid_t		  sequence;	/* Used for recovery only */
	unsigned long	  blocknr;	
};


/* The revoke table is just a simple hash table of revoke records. */
struct jbd_revoke_table_s
{
	/* It is conceivable that we might want a larger hash table
	 * for recovery.  Must be a power of two. */
	int		  hash_size; 
	int		  hash_shift; 
	struct list_head *hash_table;
};


#ifdef __KERNEL__
static void write_one_revoke_record(journal_t *, transaction_t *,
				    struct journal_head **, int *,
				    struct jbd_revoke_record_s *);
static void flush_descriptor(journal_t *, struct journal_head *, int);
#endif

/* Utility functions to maintain the revoke table */

/* Borrowed from buffer.c: this is a tried and tested block hash function */
static inline int hash(journal_t *journal, unsigned long block)
{
	struct jbd_revoke_table_s *table = journal->j_revoke;
	int hash_shift = table->hash_shift;
	
	return ((block << (hash_shift - 6)) ^
		(block >> 13) ^
		(block << (hash_shift - 12))) & (table->hash_size - 1);
}

int insert_revoke_hash(journal_t *journal, unsigned long blocknr, tid_t seq)
{
	struct list_head *hash_list;
	struct jbd_revoke_record_s *record;

repeat:
	record = kmem_cache_alloc(revoke_record_cache, GFP_NOFS);
	if (!record)
		goto oom;

	record->sequence = seq;
	record->blocknr = blocknr;
	hash_list = &journal->j_revoke->hash_table[hash(journal, blocknr)];
	list_add(&record->hash, hash_list);
	return 0;

oom:
	if (!journal_oom_retry)
		return -ENOMEM;
	jbd_debug(1, "ENOMEM in " __FUNCTION__ ", retrying.\n");
	current->policy |= SCHED_YIELD;
	schedule();
	goto repeat;
}

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

static struct jbd_revoke_record_s *find_revoke_record(journal_t *journal,
						      unsigned long blocknr)
{
	struct list_head *hash_list;
	struct jbd_revoke_record_s *record;
	
	hash_list = &journal->j_revoke->hash_table[hash(journal, blocknr)];

	record = (struct jbd_revoke_record_s *) hash_list->next;
	while (&(record->hash) != hash_list) {
		if (record->blocknr == blocknr)
			return record;
		record = (struct jbd_revoke_record_s *) record->hash.next;
	}
	return NULL;
}

int __init journal_init_revoke_caches(void)
{
	revoke_record_cache = kmem_cache_create("revoke_record",
					   sizeof(struct jbd_revoke_record_s),
					   0, SLAB_HWCACHE_ALIGN, NULL, NULL);
	if (revoke_record_cache == 0)
		return -ENOMEM;

	revoke_table_cache = kmem_cache_create("revoke_table",
					   sizeof(struct jbd_revoke_table_s),
					   0, 0, NULL, NULL);
	if (revoke_table_cache == 0) {
		kmem_cache_destroy(revoke_record_cache);
		revoke_record_cache = NULL;
		return -ENOMEM;
	}
	return 0;
}	

void journal_destroy_revoke_caches(void)
{
	kmem_cache_destroy(revoke_record_cache);
	revoke_record_cache = 0;
	kmem_cache_destroy(revoke_table_cache);
	revoke_table_cache = 0;
}

/* Initialise the revoke table for a given journal to a given size. */

int journal_init_revoke(journal_t *journal, int hash_size)
{
	int shift, tmp;
	
	J_ASSERT (journal->j_revoke == NULL);
	
	journal->j_revoke = kmem_cache_alloc(revoke_table_cache, GFP_KERNEL);
	if (!journal->j_revoke)
		return -ENOMEM;
	
	/* Check that the hash_size is a power of two */
	J_ASSERT ((hash_size & (hash_size-1)) == 0);

	journal->j_revoke->hash_size = hash_size;

	shift = 0;
	tmp = hash_size;
	while((tmp >>= 1UL) != 0UL)
		shift++;
	journal->j_revoke->hash_shift = shift;

	journal->j_revoke->hash_table =
		kmalloc(hash_size * sizeof(struct list_head), GFP_KERNEL);
	if (!journal->j_revoke->hash_table) {
		kmem_cache_free(revoke_table_cache, journal->j_revoke);
		journal->j_revoke = NULL;
		return -ENOMEM;
	}
	
	for (tmp = 0; tmp < hash_size; tmp++)
		INIT_LIST_HEAD(&journal->j_revoke->hash_table[tmp]);
	
	return 0;
}

/* Destoy a journal's revoke table.  The table must already be empty! */

void journal_destroy_revoke(journal_t *journal)
{
	struct jbd_revoke_table_s *table;
	struct list_head *hash_list;
	int i;
	
	table = journal->j_revoke;
	if (!table)
		return;
	
	for (i=0; i<table->hash_size; i++) {
		hash_list = &table->hash_table[i];
		J_ASSERT (list_empty(hash_list));
	}
	
	kfree(table->hash_table);
	kmem_cache_free(revoke_table_cache, table);
	journal->j_revoke = NULL;
}


#ifdef __KERNEL__

/* 
 * 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 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, journal_revoke() will decrement its b_count
 * by one.
 */

int journal_revoke(handle_t *handle, unsigned long blocknr, 
		   struct buffer_head *bh_in)
{
	struct buffer_head *bh = NULL;
	journal_t *journal;
	kdev_t dev;
	int err;

	if (bh_in)
		BUFFER_TRACE(bh_in, "enter");

	journal = handle->h_transaction->t_journal;
	if (!journal_set_features(journal, 0, 0, JFS_FEATURE_INCOMPAT_REVOKE)){
		J_ASSERT (!"Cannot set revoke feature!");
		return -EINVAL;
	}

	dev = journal->j_fs_dev;
	bh = bh_in;

	if (!bh) {
		bh = get_hash_table(dev, blocknr, journal->j_blocksize);
		if (bh)
			BUFFER_TRACE(bh, "found on hash");
	}
#ifdef JBD_EXPENSIVE_CHECKING
	else {
		struct buffer_head *bh2;

		/* If there is a different buffer_head lying around in
		 * memory anywhere... */
		bh2 = get_hash_table(dev, blocknr, journal->j_blocksize);
		if (bh2) {
			/* ... and it has RevokeValid status... */
			if ((bh2 != bh) &&
			    test_bit(BH_RevokeValid, &bh2->b_state))
				/* ...then it better be revoked too,
				 * since it's illegal to create a revoke
				 * record against a buffer_head which is
				 * not marked revoked --- that would
				 * risk missing a subsequent revoke
				 * cancel. */
				J_ASSERT_BH(bh2, test_bit(BH_Revoked, &