journal.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/journal.c
 *
 * Written by Stephen C. Tweedie <[e-mail]>, 1998
 *
 * Copyright 1998 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.
 *
 * Generic filesystem journal-writing code; part of the ext2fs
 * journaling system.
 *
 * This file manages journals: areas of disk reserved for logging
 * transactional updates.  This includes the kernel journaling thread
 * which is responsible for scheduling updates to the log.
 *
 * We do not actually manage the physical storage of the journal in this
 * file: that is left to a per-journal policy function, which allows us
 * to store the journal within a filesystem-specified area for ext2
 * journaling (ext2 can use a reserved inode for storing the log).
 */

#include <linux/module.h>
#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/smp_lock.h>
#include <linux/sched.h>
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <asm/uaccess.h>
#include <linux/proc_fs.h>

EXPORT_SYMBOL(journal_start);
EXPORT_SYMBOL(journal_try_start);
EXPORT_SYMBOL(journal_restart);
EXPORT_SYMBOL(journal_extend);
EXPORT_SYMBOL(journal_stop);
EXPORT_SYMBOL(journal_lock_updates);
EXPORT_SYMBOL(journal_unlock_updates);
EXPORT_SYMBOL(journal_get_write_access);
EXPORT_SYMBOL(journal_get_create_access);
EXPORT_SYMBOL(journal_get_undo_access);
EXPORT_SYMBOL(journal_dirty_data);
EXPORT_SYMBOL(journal_dirty_metadata);
#if 0
EXPORT_SYMBOL(journal_release_buffer);
#endif
EXPORT_SYMBOL(journal_forget);
#if 0
EXPORT_SYMBOL(journal_sync_buffer);
#endif
EXPORT_SYMBOL(journal_flush);
EXPORT_SYMBOL(journal_revoke);

EXPORT_SYMBOL(journal_init_dev);
EXPORT_SYMBOL(journal_init_inode);
EXPORT_SYMBOL(journal_update_format);
EXPORT_SYMBOL(journal_check_used_features);
EXPORT_SYMBOL(journal_check_available_features);
EXPORT_SYMBOL(journal_set_features);
EXPORT_SYMBOL(journal_create);
EXPORT_SYMBOL(journal_load);
EXPORT_SYMBOL(journal_destroy);
EXPORT_SYMBOL(journal_recover);
EXPORT_SYMBOL(journal_update_superblock);
EXPORT_SYMBOL(journal_abort);
EXPORT_SYMBOL(journal_errno);
EXPORT_SYMBOL(journal_ack_err);
EXPORT_SYMBOL(journal_clear_err);
EXPORT_SYMBOL(log_wait_commit);
EXPORT_SYMBOL(log_start_commit);
EXPORT_SYMBOL(journal_wipe);
EXPORT_SYMBOL(journal_blocks_per_page);
EXPORT_SYMBOL(journal_flushpage);
EXPORT_SYMBOL(journal_try_to_free_buffers);
EXPORT_SYMBOL(journal_bmap);
EXPORT_SYMBOL(journal_force_commit);

static int journal_convert_superblock_v1(journal_t *, journal_superblock_t *);

/*
 * journal_datalist_lock is used to protect data buffers:
 *
 *	bh->b_transaction
 *	bh->b_tprev
 *	bh->b_tnext
 *
 * journal_free_buffer() is called from journal_try_to_free_buffer(), and is
 * async wrt everything else.
 *
 * It is also used for checkpoint data, also to protect against
 * journal_try_to_free_buffer():
 *
 *	bh->b_cp_transaction
 *	bh->b_cpnext
 *	bh->b_cpprev
 *	transaction->t_checkpoint_list
 *	transaction->t_cpnext
 *	transaction->t_cpprev
 *	journal->j_checkpoint_transactions
 *
 * It is global at this time rather than per-journal because it's
 * impossible for __journal_free_buffer to go from a buffer_head
 * back to a journal_t unracily (well, not true.  Fix later)
 *
 *
 * The `datalist' and `checkpoint list' functions are quite
 * separate and we could use two spinlocks here.
 *
 * lru_list_lock nests inside journal_datalist_lock.
 */
spinlock_t journal_datalist_lock = SPIN_LOCK_UNLOCKED;

/*
 * jh_splice_lock needs explantion.
 *
 * In a number of places we want to do things like:
 *
 *	if (buffer_jbd(bh) && bh2jh(bh)->foo)
 *
 * This is racy on SMP, because another CPU could remove the journal_head
 * in the middle of this expression.  We need locking.
 *
 * But we can greatly optimise the locking cost by testing BH_JBD
 * outside the lock.  So, effectively:
 *
 *	ret = 0;
 *	if (buffer_jbd(bh)) {
 *		spin_lock(&jh_splice_lock);
 *		if (buffer_jbd(bh)) {	 (* Still there? *)
 *			ret = bh2jh(bh)->foo;
 *		}
 *		spin_unlock(&jh_splice_lock);
 *	}
 *	return ret;
 *
 * Now, that protects us from races where another CPU can remove the
 * journal_head.  But it doesn't defend us from the situation where another
 * CPU can *add* a journal_head.  This is a correctness issue.  But it's not
 * a problem because a) the calling code was *already* racy and b) it often
 * can't happen at the call site and c) the places where we add journal_heads
 * tend to be under external locking.
 */
spinlock_t jh_splice_lock = SPIN_LOCK_UNLOCKED;

/*
 * List of all journals in the system.  Protected by the BKL.
 */
static LIST_HEAD(all_journals);

/*
 * Helper function used to manage commit timeouts
 */

static void commit_timeout(unsigned long __data)
{
	struct task_struct * p = (struct task_struct *) __data;

	wake_up_process(p);
}

/* Static check for data structure consistency.  There's no code
 * invoked --- we'll just get a linker failure if things aren't right.
 */
void __journal_internal_check(void)
{
	extern void journal_bad_superblock_size(void);
	if (sizeof(struct journal_superblock_s) != 1024)
		journal_bad_superblock_size();
}

/*
 * kjournald: The main thread function used to manage a logging device
 * journal.
 *
 * This kernel thread is responsible for two things:
 *
 * 1) COMMIT:  Every so often we need to commit the current state of the
 *    filesystem to disk.  The journal thread is responsible for writing
 *    all of the metadata buffers to disk.
 *
 * 2) CHECKPOINT: We cannot reuse a used section of the log file until all
 *    of the data in that part of the log has been rewritten elsewhere on
 *    the disk.  Flushing these old buffers to reclaim space in the log is
 *    known as checkpointing, and this thread is responsible for that job.
 */

journal_t *current_journal;		// AKPM: debug

int kjournald(void *arg)
{
	journal_t *journal = (journal_t *) arg;
	transaction_t *transaction;
	struct timer_list timer;

	current_journal = journal;

	lock_kernel();
	daemonize();
	spin_lock_irq(&current->sigmask_lock);
	sigfillset(&current->blocked);
	recalc_sigpending(current);
	spin_unlock_irq(&current->sigmask_lock);

	sprintf(current->comm, "kjournald");

	/* Set up an interval timer which can be used to trigger a
           commit wakeup after the commit interval expires */
	init_timer(&timer);
	timer.data = (unsigned long) current;
	timer.function = commit_timeout;
	journal->j_commit_timer = &timer;

	/* Record that the journal thread is running */
	journal->j_task = current;
	wake_up(&journal->j_wait_done_commit);

	printk(KERN_INFO "kjournald starting.  Commit interval %ld seconds\n",
			journal->j_commit_interval / HZ);
	list_add(&journal->j_all_journals, &all_journals);

	/* And now, wait forever for commit wakeup events. */
	while (1) {
		if (journal->j_flags & JFS_UNMOUNT)
			break;

		jbd_debug(1, "commit_sequence=%d, commit_request=%d\n",
			journal->j_commit_sequence, journal->j_commit_request);

		if (journal->j_commit_sequence != journal->j_commit_request) {
			jbd_debug(1, "OK, requests differ\n");
			if (journal->j_commit_timer_active) {
				journal->j_commit_timer_active = 0;
				del_timer(journal->j_commit_timer);
			}

			journal_commit_transaction(journal);
			continue;
		}

		wake_up(&journal->j_wait_done_commit);
		interruptible_sleep_on(&journal->j_wait_commit);

		jbd_debug(1, "kjournald wakes\n");

		/* Were we woken up by a commit wakeup event? */
		if ((transaction = journal->j_running_transaction) != NULL &&
		    time_after_eq(jiffies, transaction->t_expires)) {
			journal->j_commit_request = transaction->t_tid;
			jbd_debug(1, "woke because of timeout\n");
		}
	}

	if (journal->j_commit_timer_active) {
		journal->j_commit_timer_active = 0;
		del_timer_sync(journal->j_commit_timer);
	}

	list_del(&journal->j_all_journals);

	journal->j_task = NULL;
	wake_up(&journal->j_wait_done_commit);
	jbd_debug(1, "Journal thread exiting.\n");
	return 0;
}

static void journal_start_thread(journal_t *journal)
{
	kernel_thread(kjournald, (void *) journal,
		      CLONE_VM | CLONE_FS | CLONE_FILES);
	while (!journal->j_task)
		sleep_on(&journal->j_wait_done_commit);
}

static void journal_kill_thread(journal_t *journal)
{
	journal->j_flags |= JFS_UNMOUNT;

	while (journal->j_task) {
		wake_up(&journal->j_wait_commit);
		sleep_on(&journal->j_wait_done_commit);
	}
}

#if 0

This is no longer needed - we do it in commit quite efficiently.
Note that if this function is resurrected, the loop needs to
be reorganised into the next_jh/last_jh algorithm.

/*
 * journal_clean_data_list: cleanup after data IO.
 *
 * Once the IO system has finished writing the buffers on the transaction's
 * data list, we can remove those buffers from the list.  This function
 * scans the list for such buffers and removes them cleanly.
 *
 * We assume that the journal is already locked.
 * We are called with journal_datalist_lock held.
 *
 * AKPM: This function looks inefficient.  Approximately O(n^2)
 * for potentially thousands of buffers.  It no longer shows on profiles
 * because these buffers are mainly dropped in journal_commit_transaction().
 */

void __journal_clean_data_list(transaction_t *transaction)
{
	struct journal_head *jh, *next;

	assert_spin_locked(&journal_datalist_lock);

restart:
	jh = transaction->t_sync_datalist;
	if (!jh)
		goto out;
	do {
		next = jh->b_tnext;
		if (!buffer_locked(jh2bh(jh)) && !buffer_dirty(jh2bh(jh))) {
			struct buffer_head *bh = jh2bh(jh);
			BUFFER_TRACE(bh, "data writeout complete: unfile");
			__journal_unfile_buffer(jh);
			jh->b_transaction = NULL;
			__journal_remove_journal_head(bh);
			refile_buffer(bh);
			__brelse(bh);
			goto restart;
		}
		jh = next;
	} while (transaction->t_sync_datalist &&
			jh != transaction->t_sync_datalist);
out:
	return;
}
#endif

/*
 * journal_write_metadata_buffer: write a metadata buffer to the journal.
 *
 * Writes a metadata buffer to a given disk block.  The actual IO is not
 * performed but a new buffer_head is constructed which labels the data
 * to be written with the correct destination disk block.
 *
 * Any magic-number escaping which needs to be done will cause a
 * copy-out here.  If the buffer happens to start with the
 * JFS_MAGIC_NUMBER, then we can't write it to the log directly: the
 * magic number is only written to the log for descripter blocks.  In
 * this case, we copy the data and replace the first word with 0, and we
 * return a result code which indicates that this buffer needs to be
 * marked as an escaped buffer in the corresponding log descriptor
 * block.  The missing word can then be restored when the block is read
 * during recovery.
 *
 * If the source buffer has already been modified by a new transaction
 * since we took the last commit snapshot, we use the frozen copy of
 * that data for IO.  If we end up using the existing buffer_head's data
 * for the write, then we *have* to lock the buffer to prevent anyone
 * else from using and possibly modifying it while the IO is in
 * progress.
 *
 * The function returns a pointer to the buffer_heads to be used for IO.
 *
 * We assume that the journal has already been locked in this function.
 *
 * Return value:
 *  <0: Error
 * >=0: Finished OK
 *
 * On success:
 * Bit 0 set == escape performed on the data
 * Bit 1 set == buffer copy-out performed (kfree the data after IO)
 */

static inline unsigned long virt_to_offset(void *p) 
{return ((unsigned long) p) & ~PAGE_MASK;}
					       
int journal_write_metadata_buffer(transaction_t *transaction,
				  struct journal_head  *jh_in,
				  struct journal_head **jh_out,
				  int blocknr)
{
	int need_copy_out = 0;
	int done_copy_out = 0;
	int do_escape = 0;
	char *mapped_data;
	struct buffer_head *new_bh;
	struct journal_head * new_jh;
	struct page *new_page;
	unsigned int new_offset;

	/*
	 * The buffer really shouldn't be locked: only the current committing
	 * transaction is allowed to write it, so nobody else is allowed
	 * to do any IO.
	 *
	 * akpm: except if we're journalling data, and write() output is
	 * also part of a shared mapping, and another thread has
	 * decided to launch a writepage() against this buffer.
	 */
	J_ASSERT_JH(jh_in, buffer_jdirty(jh2bh(jh_in)));

	/*
	 * If a new transaction has already done a buffer copy-out, then
	 * we use that version of the data for the commit.
	 */

	if (jh_in->b_frozen_data) {
		done_copy_out = 1;
		new_page = virt_to_page(jh_in->b_frozen_data);
		new_offset = virt_to_offset(jh_in->b_frozen_data);
	} else {
		new_page = jh2bh(jh_in)->b_page;
		new_offset = virt_to_offset(jh2bh(jh_in)->b_data);
	}

	mapped_data = ((char *) kmap(new_page)) + new_offset;

	/*
	 * Check for escaping
	 */
	if (* ((unsigned int *) mapped_data) == htonl(JFS_MAGIC_NUMBER)) {
		need_copy_out = 1;
		do_escape = 1;
	}

	/*
	 * Do we need to do a data copy?
	 */

	if (need_copy_out && !done_copy_out) {
		char *tmp;
		tmp = jbd_rep_kmalloc(jh2bh(jh_in)->b_size, GFP_NOFS);

		jh_in->b_frozen_data = tmp;
		memcpy (tmp, mapped_data, jh2bh(jh_in)->b_size);
		
		/* If we get to this path, we'll always need the new
		   address kmapped so that we can clear the escaped
		   magic number below. */
		kunmap(new_page);
		new_page = virt_to_page(tmp);
		new_offset = virt_to_offset(tmp);
		mapped_data = ((char *) kmap(new_page)) + new_offset;
		
		done_copy_out = 1;
	}

	/*
	 * Right, time to make up the new buffer_head.
	 */
	do {
		new_bh = get_unused_buffer_head(0);
		if (!new_bh) {
			printk (KERN_NOTICE __FUNCTION__
				": ENOMEM at get_unused_buffer_head, "
				"trying again.\n");
			current->policy |= SCHED_YIELD;
			schedule();
		}
	} while (!new_bh);
	/* keep subsequent assertions sane */
	new_bh->b_prev_free = 0;
	new_bh->b_next_free = 0;