[net-next-2.6.git] / fs / gfs2 / log.c

/*
 * Copyright (C) Sistina Software, Inc.  1997-2003 All rights reserved.
 * Copyright (C) 2004-2005 Red Hat, Inc.  All rights reserved.
 *
 * This copyrighted material is made available to anyone wishing to use,
 * modify, copy, or redistribute it subject to the terms and conditions
 * of the GNU General Public License v.2.
 */

#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/completion.h>
#include <linux/buffer_head.h>
#include <linux/gfs2_ondisk.h>
#include <linux/crc32.h>
#include <asm/semaphore.h>

#include "gfs2.h"
#include "lm_interface.h"
#include "incore.h"
#include "bmap.h"
#include "glock.h"
#include "log.h"
#include "lops.h"
#include "meta_io.h"
#include "util.h"
#include "dir.h"

#define PULL 1

static void lock_for_trans(struct gfs2_sbd *sdp)
{
	wait_event(sdp->sd_log_trans_wq, atomic_read(&sdp->sd_log_flush_count) ? 0 : 1);
	atomic_inc(&sdp->sd_log_trans_count);
}

static void unlock_from_trans(struct gfs2_sbd *sdp)
{
	gfs2_assert_warn(sdp, atomic_read(&sdp->sd_log_trans_count));
	if (atomic_dec_and_test(&sdp->sd_log_trans_count))
		wake_up(&sdp->sd_log_flush_wq);
}

static void gfs2_lock_for_flush(struct gfs2_sbd *sdp)
{
	atomic_inc(&sdp->sd_log_flush_count);
	wait_event(sdp->sd_log_flush_wq, atomic_read(&sdp->sd_log_trans_count) ? 0 : 1);
}

static void gfs2_unlock_from_flush(struct gfs2_sbd *sdp)
{
	gfs2_assert_warn(sdp, atomic_read(&sdp->sd_log_flush_count));
	if (atomic_dec_and_test(&sdp->sd_log_flush_count))
		wake_up(&sdp->sd_log_trans_wq);
}

/**
 * gfs2_struct2blk - compute stuff
 * @sdp: the filesystem
 * @nstruct: the number of structures
 * @ssize: the size of the structures
 *
 * Compute the number of log descriptor blocks needed to hold a certain number
 * of structures of a certain size.
 *
 * Returns: the number of blocks needed (minimum is always 1)
 */

unsigned int gfs2_struct2blk(struct gfs2_sbd *sdp, unsigned int nstruct,
			     unsigned int ssize)
{
	unsigned int blks;
	unsigned int first, second;

	blks = 1;
	first = (sdp->sd_sb.sb_bsize - sizeof(struct gfs2_log_descriptor)) /
		ssize;

	if (nstruct > first) {
		second = (sdp->sd_sb.sb_bsize -
			  sizeof(struct gfs2_meta_header)) / ssize;
		blks += DIV_ROUND_UP(nstruct - first, second);
	}

	return blks;
}

void gfs2_ail1_start(struct gfs2_sbd *sdp, int flags)
{
	struct list_head *head = &sdp->sd_ail1_list;
	uint64_t sync_gen;
	struct list_head *first, *tmp;
	struct gfs2_ail *first_ai, *ai;

	gfs2_log_lock(sdp);
	if (list_empty(head)) {
		gfs2_log_unlock(sdp);
		return;
	}
	sync_gen = sdp->sd_ail_sync_gen++;

	first = head->prev;
	first_ai = list_entry(first, struct gfs2_ail, ai_list);
	first_ai->ai_sync_gen = sync_gen;
	gfs2_ail1_start_one(sdp, first_ai);

	if (flags & DIO_ALL)
		first = NULL;

	for (;;) {
		if (first &&
		    (head->prev != first ||
		     gfs2_ail1_empty_one(sdp, first_ai, 0)))
			break;

		for (tmp = head->prev; tmp != head; tmp = tmp->prev) {
			ai = list_entry(tmp, struct gfs2_ail, ai_list);
			if (ai->ai_sync_gen >= sync_gen)
				continue;
			ai->ai_sync_gen = sync_gen;
			gfs2_ail1_start_one(sdp, ai);
			break;
		}

		if (tmp == head)
			break;
	}

	gfs2_log_unlock(sdp);
}

int gfs2_ail1_empty(struct gfs2_sbd *sdp, int flags)
{
	struct gfs2_ail *ai, *s;
	int ret;

	gfs2_log_lock(sdp);

	list_for_each_entry_safe_reverse(ai, s, &sdp->sd_ail1_list, ai_list) {
		if (gfs2_ail1_empty_one(sdp, ai, flags))
			list_move(&ai->ai_list, &sdp->sd_ail2_list);
		else if (!(flags & DIO_ALL))
			break;
	}

	ret = list_empty(&sdp->sd_ail1_list);

	gfs2_log_unlock(sdp);

	return ret;
}

static void ail2_empty(struct gfs2_sbd *sdp, unsigned int new_tail)
{
	struct gfs2_ail *ai, *safe;
	unsigned int old_tail = sdp->sd_log_tail;
	int wrap = (new_tail < old_tail);
	int a, b, rm;

	gfs2_log_lock(sdp);

	list_for_each_entry_safe(ai, safe, &sdp->sd_ail2_list, ai_list) {
		a = (old_tail <= ai->ai_first);
		b = (ai->ai_first < new_tail);
		rm = (wrap) ? (a || b) : (a && b);
		if (!rm)
			continue;

		gfs2_ail2_empty_one(sdp, ai);
		list_del(&ai->ai_list);
		gfs2_assert_warn(sdp, list_empty(&ai->ai_ail1_list));
		gfs2_assert_warn(sdp, list_empty(&ai->ai_ail2_list));
		kfree(ai);
	}

	gfs2_log_unlock(sdp);
}

/**
 * gfs2_log_reserve - Make a log reservation
 * @sdp: The GFS2 superblock
 * @blks: The number of blocks to reserve
 *
 * Returns: errno
 */

int gfs2_log_reserve(struct gfs2_sbd *sdp, unsigned int blks)
{
	unsigned int try = 0;

	if (gfs2_assert_warn(sdp, blks) ||
	    gfs2_assert_warn(sdp, blks <= sdp->sd_jdesc->jd_blocks))
		return -EINVAL;

	mutex_lock(&sdp->sd_log_reserve_mutex);
	for (;;) {
		gfs2_log_lock(sdp);
		if (sdp->sd_log_blks_free > blks) {
			sdp->sd_log_blks_free -= blks;
			gfs2_log_unlock(sdp);
			mutex_unlock(&sdp->sd_log_reserve_mutex);
			break;
		}

		gfs2_log_unlock(sdp);
		gfs2_ail1_empty(sdp, 0);
		gfs2_log_flush(sdp);

		if (try++)
			gfs2_ail1_start(sdp, 0);
	}
	lock_for_trans(sdp);

	return 0;
}

/**
 * gfs2_log_release - Release a given number of log blocks
 * @sdp: The GFS2 superblock
 * @blks: The number of blocks
 *
 */

void gfs2_log_release(struct gfs2_sbd *sdp, unsigned int blks)
{
	unlock_from_trans(sdp);

	gfs2_log_lock(sdp);
	sdp->sd_log_blks_free += blks;
	gfs2_assert_withdraw(sdp,
			     sdp->sd_log_blks_free <= sdp->sd_jdesc->jd_blocks);
	gfs2_log_unlock(sdp);
}

static uint64_t log_bmap(struct gfs2_sbd *sdp, unsigned int lbn)
{
	int new = 0;
	uint64_t dbn;
	int error;

	error = gfs2_block_map(sdp->sd_jdesc->jd_inode->u.generic_ip,
			       lbn, &new, &dbn, NULL);
	gfs2_assert_withdraw(sdp, !error && dbn);

	return dbn;
}

/**
 * log_distance - Compute distance between two journal blocks
 * @sdp: The GFS2 superblock
 * @newer: The most recent journal block of the pair
 * @older: The older journal block of the pair
 *
 *   Compute the distance (in the journal direction) between two
 *   blocks in the journal
 *
 * Returns: the distance in blocks
 */

static inline unsigned int log_distance(struct gfs2_sbd *sdp,
					unsigned int newer,
					unsigned int older)
{
	int dist;

	dist = newer - older;
	if (dist < 0)
		dist += sdp->sd_jdesc->jd_blocks;

	return dist;
}

static unsigned int current_tail(struct gfs2_sbd *sdp)
{
	struct gfs2_ail *ai;
	unsigned int tail;

	gfs2_log_lock(sdp);

	if (list_empty(&sdp->sd_ail1_list))
		tail = sdp->sd_log_head;
	else {
		ai = list_entry(sdp->sd_ail1_list.prev,
				struct gfs2_ail, ai_list);
		tail = ai->ai_first;
	}

	gfs2_log_unlock(sdp);

	return tail;
}

static inline void log_incr_head(struct gfs2_sbd *sdp)
{
	if (sdp->sd_log_flush_head == sdp->sd_log_tail)
		gfs2_assert_withdraw(sdp,
				sdp->sd_log_flush_head == sdp->sd_log_head);

	if (++sdp->sd_log_flush_head == sdp->sd_jdesc->jd_blocks) {
		sdp->sd_log_flush_head = 0;
		sdp->sd_log_flush_wrapped = 1;
	}
}

/**
 * gfs2_log_get_buf - Get and initialize a buffer to use for log control data
 * @sdp: The GFS2 superblock
 *
 * Returns: the buffer_head
 */

struct buffer_head *gfs2_log_get_buf(struct gfs2_sbd *sdp)
{
	uint64_t blkno = log_bmap(sdp, sdp->sd_log_flush_head);
	struct gfs2_log_buf *lb;
	struct buffer_head *bh;

	lb = kzalloc(sizeof(struct gfs2_log_buf), GFP_NOFS | __GFP_NOFAIL);
	list_add(&lb->lb_list, &sdp->sd_log_flush_list);

	bh = lb->lb_bh = sb_getblk(sdp->sd_vfs, blkno);
	lock_buffer(bh);
	memset(bh->b_data, 0, bh->b_size);
	set_buffer_uptodate(bh);
	clear_buffer_dirty(bh);
	unlock_buffer(bh);

	log_incr_head(sdp);

	return bh;
}

/**
 * gfs2_log_fake_buf - Build a fake buffer head to write metadata buffer to log
 * @sdp: the filesystem
 * @data: the data the buffer_head should point to
 *
 * Returns: the log buffer descriptor
 */

struct buffer_head *gfs2_log_fake_buf(struct gfs2_sbd *sdp,
				      struct buffer_head *real)
{
	uint64_t blkno = log_bmap(sdp, sdp->sd_log_flush_head);
	struct gfs2_log_buf *lb;
	struct buffer_head *bh;

	lb = kzalloc(sizeof(struct gfs2_log_buf), GFP_NOFS | __GFP_NOFAIL);
	list_add(&lb->lb_list, &sdp->sd_log_flush_list);
	lb->lb_real = real;

	bh = lb->lb_bh = alloc_buffer_head(GFP_NOFS | __GFP_NOFAIL);
	atomic_set(&bh->b_count, 1);
	bh->b_state = (1 << BH_Mapped) | (1 << BH_Uptodate);
	set_bh_page(bh, real->b_page, bh_offset(real));
	bh->b_blocknr = blkno;
	bh->b_size = sdp->sd_sb.sb_bsize;
	bh->b_bdev = sdp->sd_vfs->s_bdev;

	log_incr_head(sdp);

	return bh;
}

static void log_pull_tail(struct gfs2_sbd *sdp, unsigned int new_tail, int pull)
{
	unsigned int dist = log_distance(sdp, new_tail, sdp->sd_log_tail);

	ail2_empty(sdp, new_tail);

	gfs2_log_lock(sdp);
	sdp->sd_log_blks_free += dist - ((pull) ? 1 : 0);
	gfs2_assert_withdraw(sdp,
			     sdp->sd_log_blks_free <= sdp->sd_jdesc->jd_blocks);
	gfs2_log_unlock(sdp);

	sdp->sd_log_tail = new_tail;
}

/**
 * log_write_header - Get and initialize a journal header buffer
 * @sdp: The GFS2 superblock
 *
 * Returns: the initialized log buffer descriptor
 */

static void log_write_header(struct gfs2_sbd *sdp, uint32_t flags, int pull)
{
	uint64_t blkno = log_bmap(sdp, sdp->sd_log_flush_head);
	struct buffer_head *bh;
	struct gfs2_log_header *lh;
	unsigned int tail;
	uint32_t hash;

	bh = sb_getblk(sdp->sd_vfs, blkno);
	lock_buffer(bh);
	memset(bh->b_data, 0, bh->b_size);
	set_buffer_uptodate(bh);
	clear_buffer_dirty(bh);
	unlock_buffer(bh);

	gfs2_ail1_empty(sdp, 0);
	tail = current_tail(sdp);

	lh = (struct gfs2_log_header *)bh->b_data;
	memset(lh, 0, sizeof(struct gfs2_log_header));
	lh->lh_header.mh_magic = cpu_to_be32(GFS2_MAGIC);
	lh->lh_header.mh_type = cpu_to_be16(GFS2_METATYPE_LH);
	lh->lh_header.mh_format = cpu_to_be16(GFS2_FORMAT_LH);
	lh->lh_sequence = be64_to_cpu(sdp->sd_log_sequence++);
	lh->lh_flags = be32_to_cpu(flags);
	lh->lh_tail = be32_to_cpu(tail);
	lh->lh_blkno = be32_to_cpu(sdp->sd_log_flush_head);
	hash = gfs2_disk_hash(bh->b_data, sizeof(struct gfs2_log_header));
	lh->lh_hash = cpu_to_be32(hash);

	set_buffer_dirty(bh);
	if (sync_dirty_buffer(bh))
		gfs2_io_error_bh(sdp, bh);
	brelse(bh);

	if (sdp->sd_log_tail != tail)
		log_pull_tail(sdp, tail, pull);
	else
		gfs2_assert_withdraw(sdp, !pull);

	sdp->sd_log_idle = (tail == sdp->sd_log_flush_head);
	log_incr_head(sdp);
}

static void log_flush_commit(struct gfs2_sbd *sdp)
{
	struct list_head *head = &sdp->sd_log_flush_list;
	struct gfs2_log_buf *lb;
	struct buffer_head *bh;
	unsigned int d;

	d = log_distance(sdp, sdp->sd_log_flush_head, sdp->sd_log_head);

	gfs2_assert_withdraw(sdp, d + 1 == sdp->sd_log_blks_reserved);

	while (!list_empty(head)) {
		lb = list_entry(head->next, struct gfs2_log_buf, lb_list);
		list_del(&lb->lb_list);
		bh = lb->lb_bh;

		wait_on_buffer(bh);
		if (!buffer_uptodate(bh))
			gfs2_io_error_bh(sdp, bh);
		if (lb->lb_real) {
			while (atomic_read(&bh->b_count) != 1)  /* Grrrr... */
				schedule();
			free_buffer_head(bh);
		} else
			brelse(bh);
		kfree(lb);
	}

	log_write_header(sdp, 0, 0);
}

/**
 * gfs2_log_flush_i - flush incore transaction(s)
 * @sdp: the filesystem
 * @gl: The glock structure to flush.  If NULL, flush the whole incore log
 *
 */

void gfs2_log_flush_i(struct gfs2_sbd *sdp, struct gfs2_glock *gl)
{
	struct gfs2_ail *ai;

	ai = kzalloc(sizeof(struct gfs2_ail), GFP_NOFS | __GFP_NOFAIL);
	INIT_LIST_HEAD(&ai->ai_ail1_list);
	INIT_LIST_HEAD(&ai->ai_ail2_list);
	gfs2_lock_for_flush(sdp);

	if (gl) {
		gfs2_log_lock(sdp);
		if (list_empty(&gl->gl_le.le_list)) {
			gfs2_log_unlock(sdp);
			gfs2_unlock_from_flush(sdp);
			kfree(ai);
			return;
		}
		gfs2_log_unlock(sdp);
	}

	mutex_lock(&sdp->sd_log_flush_lock);

	gfs2_assert_withdraw(sdp,
			sdp->sd_log_num_buf == sdp->sd_log_commited_buf);
	gfs2_assert_withdraw(sdp,
			sdp->sd_log_num_revoke == sdp->sd_log_commited_revoke);

	sdp->sd_log_flush_head = sdp->sd_log_head;
	sdp->sd_log_flush_wrapped = 0;
	ai->ai_first = sdp->sd_log_flush_head;

	lops_before_commit(sdp);
	if (!list_empty(&sdp->sd_log_flush_list))
		log_flush_commit(sdp);
	else if (sdp->sd_log_tail != current_tail(sdp) && !sdp->sd_log_idle)
		log_write_header(sdp, 0, PULL);
	lops_after_commit(sdp, ai);
	sdp->sd_log_head = sdp->sd_log_flush_head;
	if (sdp->sd_log_flush_wrapped)
		sdp->sd_log_wraps++;

	sdp->sd_log_blks_reserved =
		sdp->sd_log_commited_buf =
		sdp->sd_log_commited_revoke = 0;

	gfs2_log_lock(sdp);
	if (!list_empty(&ai->ai_ail1_list)) {
		list_add(&ai->ai_list, &sdp->sd_ail1_list);
		ai = NULL;
	}
	gfs2_log_unlock(sdp);

	mutex_unlock(&sdp->sd_log_flush_lock);
	sdp->sd_vfs->s_dirt = 0;
	gfs2_unlock_from_flush(sdp);

	kfree(ai);
}

static void log_refund(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
{
	unsigned int reserved = 1;
	unsigned int old;

	gfs2_log_lock(sdp);

	sdp->sd_log_commited_buf += tr->tr_num_buf_new - tr->tr_num_buf_rm;
	gfs2_assert_withdraw(sdp, ((int)sdp->sd_log_commited_buf) >= 0);
	sdp->sd_log_commited_revoke += tr->tr_num_revoke - tr->tr_num_revoke_rm;
	gfs2_assert_withdraw(sdp, ((int)sdp->sd_log_commited_revoke) >= 0);

	if (sdp->sd_log_commited_buf)
		reserved += 1 + sdp->sd_log_commited_buf +
			    sdp->sd_log_commited_buf/503;
	if (sdp->sd_log_commited_revoke)
		reserved += gfs2_struct2blk(sdp, sdp->sd_log_commited_revoke,
					    sizeof(uint64_t));

	old = sdp->sd_log_blks_free;
	sdp->sd_log_blks_free += tr->tr_reserved -
				 (reserved - sdp->sd_log_blks_reserved);

	gfs2_assert_withdraw(sdp,
			     sdp->sd_log_blks_free >= old);
	gfs2_assert_withdraw(sdp,
			     sdp->sd_log_blks_free <= sdp->sd_jdesc->jd_blocks);

	sdp->sd_log_blks_reserved = reserved;

	gfs2_log_unlock(sdp);
}

/**
 * gfs2_log_commit - Commit a transaction to the log
 * @sdp: the filesystem
 * @tr: the transaction
 *
 * Returns: errno
 */

void gfs2_log_commit(struct gfs2_sbd *sdp, struct gfs2_trans *tr)
{
	log_refund(sdp, tr);
	lops_incore_commit(sdp, tr);

	sdp->sd_vfs->s_dirt = 1;
	unlock_from_trans(sdp);

	gfs2_log_lock(sdp);
	if (sdp->sd_log_num_buf > gfs2_tune_get(sdp, gt_incore_log_blocks)) {
		gfs2_log_unlock(sdp);
		gfs2_log_flush(sdp);
	} else
		gfs2_log_unlock(sdp);
}

/**
 * gfs2_log_shutdown - write a shutdown header into a journal
 * @sdp: the filesystem
 *
 */

void gfs2_log_shutdown(struct gfs2_sbd *sdp)
{
	mutex_lock(&sdp->sd_log_flush_lock);

	gfs2_assert_withdraw(sdp, !atomic_read(&sdp->sd_log_trans_count));
	gfs2_assert_withdraw(sdp, !sdp->sd_log_blks_reserved);
	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_gl);
	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_buf);
	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_jdata);
	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_revoke);
	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_rg);
	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_databuf);
	gfs2_assert_withdraw(sdp, list_empty(&sdp->sd_ail1_list));

	sdp->sd_log_flush_head = sdp->sd_log_head;
	sdp->sd_log_flush_wrapped = 0;

	log_write_header(sdp, GFS2_LOG_HEAD_UNMOUNT, 0);

	gfs2_assert_withdraw(sdp, sdp->sd_log_blks_free ==
			     sdp->sd_jdesc->jd_blocks);
	gfs2_assert_withdraw(sdp, sdp->sd_log_head == sdp->sd_log_tail);
	gfs2_assert_withdraw(sdp, list_empty(&sdp->sd_ail2_list));

	sdp->sd_log_head = sdp->sd_log_flush_head;
	if (sdp->sd_log_flush_wrapped)
		sdp->sd_log_wraps++;
	sdp->sd_log_tail = sdp->sd_log_head;

	mutex_unlock(&sdp->sd_log_flush_lock);
}
Commit	Line	Data
b3b94faa DT	1	/*
	2	* Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
	3	* Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
	4	*
	5	* This copyrighted material is made available to anyone wishing to use,
	6	* modify, copy, or redistribute it subject to the terms and conditions
	7	* of the GNU General Public License v.2.
	8	*/
	9
	10	#include <linux/sched.h>
	11	#include <linux/slab.h>
	12	#include <linux/spinlock.h>
	13	#include <linux/completion.h>
	14	#include <linux/buffer_head.h>
5c676f6d	15	#include <linux/gfs2_ondisk.h>
71b86f56	16	#include <linux/crc32.h>
b3b94faa DT	17	#include <asm/semaphore.h>
	18
	19	#include "gfs2.h"
5c676f6d SW	20	#include "lm_interface.h"
5c676f6d SW	21	#include "incore.h"
b3b94faa DT	22	#include "bmap.h"
	23	#include "glock.h"
	24	#include "log.h"
	25	#include "lops.h"
	26	#include "meta_io.h"
5c676f6d	27	#include "util.h"
71b86f56	28	#include "dir.h"
b3b94faa DT	29
	30	#define PULL 1
	31
b3b94faa DT	32	static void lock_for_trans(struct gfs2_sbd *sdp)
b3b94faa DT	33	{
71b86f56	34	wait_event(sdp->sd_log_trans_wq, atomic_read(&sdp->sd_log_flush_count) ? 0 : 1);
b3b94faa	35	atomic_inc(&sdp->sd_log_trans_count);
b3b94faa DT	36	}
	37
	38	static void unlock_from_trans(struct gfs2_sbd *sdp)
	39	{
	40	gfs2_assert_warn(sdp, atomic_read(&sdp->sd_log_trans_count));
	41	if (atomic_dec_and_test(&sdp->sd_log_trans_count))
	42	wake_up(&sdp->sd_log_flush_wq);
	43	}
	44
f55ab26a	45	static void gfs2_lock_for_flush(struct gfs2_sbd *sdp)
b3b94faa	46	{
b3b94faa	47	atomic_inc(&sdp->sd_log_flush_count);
71b86f56	48	wait_event(sdp->sd_log_flush_wq, atomic_read(&sdp->sd_log_trans_count) ? 0 : 1);
b3b94faa DT	49	}
b3b94faa DT	50
f55ab26a	51	static void gfs2_unlock_from_flush(struct gfs2_sbd *sdp)
b3b94faa DT	52	{
	53	gfs2_assert_warn(sdp, atomic_read(&sdp->sd_log_flush_count));
	54	if (atomic_dec_and_test(&sdp->sd_log_flush_count))
	55	wake_up(&sdp->sd_log_trans_wq);
	56	}
	57
	58	/**
	59	* gfs2_struct2blk - compute stuff
	60	* @sdp: the filesystem
	61	* @nstruct: the number of structures
	62	* @ssize: the size of the structures
	63	*
	64	* Compute the number of log descriptor blocks needed to hold a certain number
	65	* of structures of a certain size.
	66	*
	67	* Returns: the number of blocks needed (minimum is always 1)
	68	*/
	69
	70	unsigned int gfs2_struct2blk(struct gfs2_sbd *sdp, unsigned int nstruct,
	71	unsigned int ssize)
	72	{
	73	unsigned int blks;
	74	unsigned int first, second;
	75
	76	blks = 1;
568f4c96 SW	77	first = (sdp->sd_sb.sb_bsize - sizeof(struct gfs2_log_descriptor)) /
568f4c96 SW	78	ssize;
b3b94faa DT	79
b3b94faa DT	80	if (nstruct > first) {
568f4c96 SW	81	second = (sdp->sd_sb.sb_bsize -
568f4c96 SW	82	sizeof(struct gfs2_meta_header)) / ssize;
5c676f6d	83	blks += DIV_ROUND_UP(nstruct - first, second);
b3b94faa DT	84	}
	85
	86	return blks;
	87	}
	88
	89	void gfs2_ail1_start(struct gfs2_sbd *sdp, int flags)
	90	{
	91	struct list_head *head = &sdp->sd_ail1_list;
	92	uint64_t sync_gen;
	93	struct list_head first, tmp;
	94	struct gfs2_ail first_ai, ai;
	95
	96	gfs2_log_lock(sdp);
	97	if (list_empty(head)) {
	98	gfs2_log_unlock(sdp);
	99	return;
	100	}
	101	sync_gen = sdp->sd_ail_sync_gen++;
	102
	103	first = head->prev;
	104	first_ai = list_entry(first, struct gfs2_ail, ai_list);
	105	first_ai->ai_sync_gen = sync_gen;
	106	gfs2_ail1_start_one(sdp, first_ai);
	107
	108	if (flags & DIO_ALL)
	109	first = NULL;
	110
	111	for (;;) {
	112	if (first &&
	113	(head->prev != first \|\|
	114	gfs2_ail1_empty_one(sdp, first_ai, 0)))
	115	break;
	116
	117	for (tmp = head->prev; tmp != head; tmp = tmp->prev) {
	118	ai = list_entry(tmp, struct gfs2_ail, ai_list);
	119	if (ai->ai_sync_gen >= sync_gen)
	120	continue;
	121	ai->ai_sync_gen = sync_gen;
	122	gfs2_ail1_start_one(sdp, ai);
	123	break;
	124	}
	125
	126	if (tmp == head)
	127	break;
	128	}
	129
	130	gfs2_log_unlock(sdp);
	131	}
	132
	133	int gfs2_ail1_empty(struct gfs2_sbd *sdp, int flags)
	134	{
	135	struct gfs2_ail ai, s;
	136	int ret;
	137
	138	gfs2_log_lock(sdp);
	139
	140	list_for_each_entry_safe_reverse(ai, s, &sdp->sd_ail1_list, ai_list) {
	141	if (gfs2_ail1_empty_one(sdp, ai, flags))
	142	list_move(&ai->ai_list, &sdp->sd_ail2_list);
	143	else if (!(flags & DIO_ALL))
	144	break;
	145	}
	146
	147	ret = list_empty(&sdp->sd_ail1_list);
148
149	gfs2_log_unlock(sdp);
150
151	return ret;
152	}
153
154	static void ail2_empty(struct gfs2_sbd *sdp, unsigned int new_tail)
155	{
156	struct gfs2_ail ai, safe;
157	unsigned int old_tail = sdp->sd_log_tail;
158	int wrap = (new_tail < old_tail);
159	int a, b, rm;
160
161	gfs2_log_lock(sdp);
162
163	list_for_each_entry_safe(ai, safe, &sdp->sd_ail2_list, ai_list) {
164	a = (old_tail <= ai->ai_first);
165	b = (ai->ai_first < new_tail);
166	rm = (wrap) ? (a \|\| b) : (a && b);
167	if (!rm)
168	continue;
169
170	gfs2_ail2_empty_one(sdp, ai);
171	list_del(&ai->ai_list);
172	gfs2_assert_warn(sdp, list_empty(&ai->ai_ail1_list));
173	gfs2_assert_warn(sdp, list_empty(&ai->ai_ail2_list));
174	kfree(ai);
175	}
176
177	gfs2_log_unlock(sdp);
178	}
179
180	/**
181	* gfs2_log_reserve - Make a log reservation
182	* @sdp: The GFS2 superblock
183	* @blks: The number of blocks to reserve
184	*
185	* Returns: errno
186	*/
187
188	int gfs2_log_reserve(struct gfs2_sbd *sdp, unsigned int blks)
189	{
b3b94faa DT	190	unsigned int try = 0;
	191
	192	if (gfs2_assert_warn(sdp, blks) \|\|
	193	gfs2_assert_warn(sdp, blks <= sdp->sd_jdesc->jd_blocks))
	194	return -EINVAL;
	195
71b86f56	196	mutex_lock(&sdp->sd_log_reserve_mutex);
b3b94faa DT	197	for (;;) {
b3b94faa DT	198	gfs2_log_lock(sdp);
b3b94faa DT	199	if (sdp->sd_log_blks_free > blks) {
b3b94faa DT	200	sdp->sd_log_blks_free -= blks;
b3b94faa	201	gfs2_log_unlock(sdp);
71b86f56	202	mutex_unlock(&sdp->sd_log_reserve_mutex);
b3b94faa DT	203	break;
	204	}
	205
	206	gfs2_log_unlock(sdp);
b3b94faa DT	207	gfs2_ail1_empty(sdp, 0);
	208	gfs2_log_flush(sdp);
	209
	210	if (try++)
	211	gfs2_ail1_start(sdp, 0);
	212	}
b3b94faa DT	213	lock_for_trans(sdp);
	214
	215	return 0;
	216	}
	217
	218	/**
	219	* gfs2_log_release - Release a given number of log blocks
	220	* @sdp: The GFS2 superblock
	221	* @blks: The number of blocks
	222	*
	223	*/
	224
	225	void gfs2_log_release(struct gfs2_sbd *sdp, unsigned int blks)
	226	{
	227	unlock_from_trans(sdp);
	228
	229	gfs2_log_lock(sdp);
	230	sdp->sd_log_blks_free += blks;
	231	gfs2_assert_withdraw(sdp,
	232	sdp->sd_log_blks_free <= sdp->sd_jdesc->jd_blocks);
	233	gfs2_log_unlock(sdp);
	234	}
	235
	236	static uint64_t log_bmap(struct gfs2_sbd *sdp, unsigned int lbn)
	237	{
	238	int new = 0;
	239	uint64_t dbn;
	240	int error;
	241
5c676f6d	242	error = gfs2_block_map(sdp->sd_jdesc->jd_inode->u.generic_ip,
568f4c96	243	lbn, &new, &dbn, NULL);
b3b94faa DT	244	gfs2_assert_withdraw(sdp, !error && dbn);
	245
	246	return dbn;
	247	}
	248
	249	/**
	250	* log_distance - Compute distance between two journal blocks
	251	* @sdp: The GFS2 superblock
	252	* @newer: The most recent journal block of the pair
	253	* @older: The older journal block of the pair
	254	*
	255	* Compute the distance (in the journal direction) between two
	256	* blocks in the journal
	257	*
	258	* Returns: the distance in blocks
	259	*/
	260
	261	static inline unsigned int log_distance(struct gfs2_sbd *sdp,
	262	unsigned int newer,
	263	unsigned int older)
	264	{
	265	int dist;
	266
	267	dist = newer - older;
	268	if (dist < 0)
	269	dist += sdp->sd_jdesc->jd_blocks;
	270
	271	return dist;
	272	}
	273
	274	static unsigned int current_tail(struct gfs2_sbd *sdp)
	275	{
	276	struct gfs2_ail *ai;
	277	unsigned int tail;
	278
	279	gfs2_log_lock(sdp);
	280
	281	if (list_empty(&sdp->sd_ail1_list))
	282	tail = sdp->sd_log_head;
	283	else {
	284	ai = list_entry(sdp->sd_ail1_list.prev,
	285	struct gfs2_ail, ai_list);
	286	tail = ai->ai_first;
	287	}
	288
	289	gfs2_log_unlock(sdp);
	290
	291	return tail;
	292	}
	293
	294	static inline void log_incr_head(struct gfs2_sbd *sdp)
	295	{
	296	if (sdp->sd_log_flush_head == sdp->sd_log_tail)
	297	gfs2_assert_withdraw(sdp,
	298	sdp->sd_log_flush_head == sdp->sd_log_head);
	299
	300	if (++sdp->sd_log_flush_head == sdp->sd_jdesc->jd_blocks) {
	301	sdp->sd_log_flush_head = 0;
	302	sdp->sd_log_flush_wrapped = 1;
	303	}
	304	}
	305
	306	/**
	307	* gfs2_log_get_buf - Get and initialize a buffer to use for log control data
308	* @sdp: The GFS2 superblock
309	*
310	* Returns: the buffer_head
311	*/
312
313	struct buffer_head gfs2_log_get_buf(struct gfs2_sbd sdp)
314	{
315	uint64_t blkno = log_bmap(sdp, sdp->sd_log_flush_head);
316	struct gfs2_log_buf *lb;
317	struct buffer_head *bh;
318
4f3df041	319	lb = kzalloc(sizeof(struct gfs2_log_buf), GFP_NOFS \| __GFP_NOFAIL);
b3b94faa DT	320	list_add(&lb->lb_list, &sdp->sd_log_flush_list);
	321
	322	bh = lb->lb_bh = sb_getblk(sdp->sd_vfs, blkno);
	323	lock_buffer(bh);
	324	memset(bh->b_data, 0, bh->b_size);
	325	set_buffer_uptodate(bh);
	326	clear_buffer_dirty(bh);
	327	unlock_buffer(bh);
	328
	329	log_incr_head(sdp);
	330
	331	return bh;
	332	}
	333
	334	/**
	335	* gfs2_log_fake_buf - Build a fake buffer head to write metadata buffer to log
	336	* @sdp: the filesystem
	337	* @data: the data the buffer_head should point to
	338	*
	339	* Returns: the log buffer descriptor
	340	*/
	341
	342	struct buffer_head gfs2_log_fake_buf(struct gfs2_sbd sdp,
	343	struct buffer_head *real)
	344	{
	345	uint64_t blkno = log_bmap(sdp, sdp->sd_log_flush_head);
	346	struct gfs2_log_buf *lb;
	347	struct buffer_head *bh;
	348
4f3df041	349	lb = kzalloc(sizeof(struct gfs2_log_buf), GFP_NOFS \| __GFP_NOFAIL);
b3b94faa DT	350	list_add(&lb->lb_list, &sdp->sd_log_flush_list);
	351	lb->lb_real = real;
	352
	353	bh = lb->lb_bh = alloc_buffer_head(GFP_NOFS \| __GFP_NOFAIL);
	354	atomic_set(&bh->b_count, 1);
	355	bh->b_state = (1 << BH_Mapped) \| (1 << BH_Uptodate);
18ec7d5c	356	set_bh_page(bh, real->b_page, bh_offset(real));
b3b94faa DT	357	bh->b_blocknr = blkno;
	358	bh->b_size = sdp->sd_sb.sb_bsize;
	359	bh->b_bdev = sdp->sd_vfs->s_bdev;
	360
	361	log_incr_head(sdp);
	362
	363	return bh;
	364	}
	365
	366	static void log_pull_tail(struct gfs2_sbd *sdp, unsigned int new_tail, int pull)
	367	{
	368	unsigned int dist = log_distance(sdp, new_tail, sdp->sd_log_tail);
	369
	370	ail2_empty(sdp, new_tail);
	371
	372	gfs2_log_lock(sdp);
	373	sdp->sd_log_blks_free += dist - ((pull) ? 1 : 0);
	374	gfs2_assert_withdraw(sdp,
	375	sdp->sd_log_blks_free <= sdp->sd_jdesc->jd_blocks);
	376	gfs2_log_unlock(sdp);
	377
	378	sdp->sd_log_tail = new_tail;
	379	}
	380
	381	/**
	382	* log_write_header - Get and initialize a journal header buffer
	383	* @sdp: The GFS2 superblock
	384	*
	385	* Returns: the initialized log buffer descriptor
	386	*/
	387
	388	static void log_write_header(struct gfs2_sbd *sdp, uint32_t flags, int pull)
	389	{
	390	uint64_t blkno = log_bmap(sdp, sdp->sd_log_flush_head);
	391	struct buffer_head *bh;
	392	struct gfs2_log_header *lh;
	393	unsigned int tail;
	394	uint32_t hash;
	395
b3b94faa DT	396	bh = sb_getblk(sdp->sd_vfs, blkno);
	397	lock_buffer(bh);
	398	memset(bh->b_data, 0, bh->b_size);
	399	set_buffer_uptodate(bh);
	400	clear_buffer_dirty(bh);
	401	unlock_buffer(bh);
	402
	403	gfs2_ail1_empty(sdp, 0);
	404	tail = current_tail(sdp);
	405
	406	lh = (struct gfs2_log_header *)bh->b_data;
	407	memset(lh, 0, sizeof(struct gfs2_log_header));
	408	lh->lh_header.mh_magic = cpu_to_be32(GFS2_MAGIC);
	409	lh->lh_header.mh_type = cpu_to_be16(GFS2_METATYPE_LH);
	410	lh->lh_header.mh_format = cpu_to_be16(GFS2_FORMAT_LH);
	411	lh->lh_sequence = be64_to_cpu(sdp->sd_log_sequence++);
	412	lh->lh_flags = be32_to_cpu(flags);
	413	lh->lh_tail = be32_to_cpu(tail);
	414	lh->lh_blkno = be32_to_cpu(sdp->sd_log_flush_head);
	415	hash = gfs2_disk_hash(bh->b_data, sizeof(struct gfs2_log_header));
	416	lh->lh_hash = cpu_to_be32(hash);
	417
	418	set_buffer_dirty(bh);
	419	if (sync_dirty_buffer(bh))
	420	gfs2_io_error_bh(sdp, bh);
	421	brelse(bh);
	422
	423	if (sdp->sd_log_tail != tail)
	424	log_pull_tail(sdp, tail, pull);
	425	else
	426	gfs2_assert_withdraw(sdp, !pull);
	427
	428	sdp->sd_log_idle = (tail == sdp->sd_log_flush_head);
	429	log_incr_head(sdp);
	430	}
	431
	432	static void log_flush_commit(struct gfs2_sbd *sdp)
	433	{
	434	struct list_head *head = &sdp->sd_log_flush_list;
	435	struct gfs2_log_buf *lb;
	436	struct buffer_head *bh;
	437	unsigned int d;
	438
	439	d = log_distance(sdp, sdp->sd_log_flush_head, sdp->sd_log_head);
	440
	441	gfs2_assert_withdraw(sdp, d + 1 == sdp->sd_log_blks_reserved);
	442
	443	while (!list_empty(head)) {
	444	lb = list_entry(head->next, struct gfs2_log_buf, lb_list);
	445	list_del(&lb->lb_list);
	446	bh = lb->lb_bh;
	447
	448	wait_on_buffer(bh);
	449	if (!buffer_uptodate(bh))
	450	gfs2_io_error_bh(sdp, bh);
	451	if (lb->lb_real) {
	452	while (atomic_read(&bh->b_count) != 1) /* Grrrr... */
	453	schedule();
	454	free_buffer_head(bh);
	455	} else
	456	brelse(bh);
	457	kfree(lb);
	458	}
	459
460	log_write_header(sdp, 0, 0);
461	}
462
463	/**
464	* gfs2_log_flush_i - flush incore transaction(s)
465	* @sdp: the filesystem
466	* @gl: The glock structure to flush. If NULL, flush the whole incore log
467	*
468	*/
469
470	void gfs2_log_flush_i(struct gfs2_sbd sdp, struct gfs2_glock gl)
471	{
472	struct gfs2_ail *ai;
473
4f3df041	474	ai = kzalloc(sizeof(struct gfs2_ail), GFP_NOFS \| __GFP_NOFAIL);
b3b94faa DT	475	INIT_LIST_HEAD(&ai->ai_ail1_list);
b3b94faa DT	476	INIT_LIST_HEAD(&ai->ai_ail2_list);
b3b94faa	477	gfs2_lock_for_flush(sdp);
f55ab26a SW	478
	479	if (gl) {
	480	gfs2_log_lock(sdp);
	481	if (list_empty(&gl->gl_le.le_list)) {
	482	gfs2_log_unlock(sdp);
	483	gfs2_unlock_from_flush(sdp);
	484	kfree(ai);
	485	return;
	486	}
	487	gfs2_log_unlock(sdp);
	488	}
	489
	490	mutex_lock(&sdp->sd_log_flush_lock);
b3b94faa DT	491
	492	gfs2_assert_withdraw(sdp,
	493	sdp->sd_log_num_buf == sdp->sd_log_commited_buf);
	494	gfs2_assert_withdraw(sdp,
	495	sdp->sd_log_num_revoke == sdp->sd_log_commited_revoke);
	496
b3b94faa DT	497	sdp->sd_log_flush_head = sdp->sd_log_head;
	498	sdp->sd_log_flush_wrapped = 0;
	499	ai->ai_first = sdp->sd_log_flush_head;
	500
	501	lops_before_commit(sdp);
	502	if (!list_empty(&sdp->sd_log_flush_list))
	503	log_flush_commit(sdp);
	504	else if (sdp->sd_log_tail != current_tail(sdp) && !sdp->sd_log_idle)
	505	log_write_header(sdp, 0, PULL);
	506	lops_after_commit(sdp, ai);
b3b94faa DT	507	sdp->sd_log_head = sdp->sd_log_flush_head;
	508	if (sdp->sd_log_flush_wrapped)
	509	sdp->sd_log_wraps++;
	510
	511	sdp->sd_log_blks_reserved =
	512	sdp->sd_log_commited_buf =
	513	sdp->sd_log_commited_revoke = 0;
	514
	515	gfs2_log_lock(sdp);
	516	if (!list_empty(&ai->ai_ail1_list)) {
	517	list_add(&ai->ai_list, &sdp->sd_ail1_list);
	518	ai = NULL;
	519	}
	520	gfs2_log_unlock(sdp);
	521
f55ab26a	522	mutex_unlock(&sdp->sd_log_flush_lock);
b3b94faa DT	523	sdp->sd_vfs->s_dirt = 0;
	524	gfs2_unlock_from_flush(sdp);
	525
	526	kfree(ai);
	527	}
	528
	529	static void log_refund(struct gfs2_sbd sdp, struct gfs2_trans tr)
	530	{
	531	unsigned int reserved = 1;
	532	unsigned int old;
	533
	534	gfs2_log_lock(sdp);
	535
	536	sdp->sd_log_commited_buf += tr->tr_num_buf_new - tr->tr_num_buf_rm;
	537	gfs2_assert_withdraw(sdp, ((int)sdp->sd_log_commited_buf) >= 0);
	538	sdp->sd_log_commited_revoke += tr->tr_num_revoke - tr->tr_num_revoke_rm;
	539	gfs2_assert_withdraw(sdp, ((int)sdp->sd_log_commited_revoke) >= 0);
	540
	541	if (sdp->sd_log_commited_buf)
568f4c96 SW	542	reserved += 1 + sdp->sd_log_commited_buf +
568f4c96 SW	543	sdp->sd_log_commited_buf/503;
b3b94faa DT	544	if (sdp->sd_log_commited_revoke)
	545	reserved += gfs2_struct2blk(sdp, sdp->sd_log_commited_revoke,
	546	sizeof(uint64_t));
	547
	548	old = sdp->sd_log_blks_free;
	549	sdp->sd_log_blks_free += tr->tr_reserved -
	550	(reserved - sdp->sd_log_blks_reserved);
	551
	552	gfs2_assert_withdraw(sdp,
	553	sdp->sd_log_blks_free >= old);
	554	gfs2_assert_withdraw(sdp,
	555	sdp->sd_log_blks_free <= sdp->sd_jdesc->jd_blocks);
	556
	557	sdp->sd_log_blks_reserved = reserved;
	558
	559	gfs2_log_unlock(sdp);
	560	}
	561
	562	/**
	563	* gfs2_log_commit - Commit a transaction to the log
	564	* @sdp: the filesystem
	565	* @tr: the transaction
	566	*
	567	* Returns: errno
	568	*/
	569
	570	void gfs2_log_commit(struct gfs2_sbd sdp, struct gfs2_trans tr)
	571	{
	572	log_refund(sdp, tr);
	573	lops_incore_commit(sdp, tr);
	574
	575	sdp->sd_vfs->s_dirt = 1;
	576	unlock_from_trans(sdp);
	577
b3b94faa DT	578	gfs2_log_lock(sdp);
	579	if (sdp->sd_log_num_buf > gfs2_tune_get(sdp, gt_incore_log_blocks)) {
	580	gfs2_log_unlock(sdp);
	581	gfs2_log_flush(sdp);
	582	} else
	583	gfs2_log_unlock(sdp);
	584	}
	585
	586	/**
	587	* gfs2_log_shutdown - write a shutdown header into a journal
	588	* @sdp: the filesystem
	589	*
	590	*/
	591
	592	void gfs2_log_shutdown(struct gfs2_sbd *sdp)
	593	{
f55ab26a	594	mutex_lock(&sdp->sd_log_flush_lock);
b3b94faa DT	595
	596	gfs2_assert_withdraw(sdp, !atomic_read(&sdp->sd_log_trans_count));
	597	gfs2_assert_withdraw(sdp, !sdp->sd_log_blks_reserved);
	598	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_gl);
	599	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_buf);
18ec7d5c	600	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_jdata);
b3b94faa DT	601	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_revoke);
	602	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_rg);
	603	gfs2_assert_withdraw(sdp, !sdp->sd_log_num_databuf);
	604	gfs2_assert_withdraw(sdp, list_empty(&sdp->sd_ail1_list));
	605
	606	sdp->sd_log_flush_head = sdp->sd_log_head;
	607	sdp->sd_log_flush_wrapped = 0;
	608
	609	log_write_header(sdp, GFS2_LOG_HEAD_UNMOUNT, 0);
	610
	611	gfs2_assert_withdraw(sdp, sdp->sd_log_blks_free ==
	612	sdp->sd_jdesc->jd_blocks);
	613	gfs2_assert_withdraw(sdp, sdp->sd_log_head == sdp->sd_log_tail);
	614	gfs2_assert_withdraw(sdp, list_empty(&sdp->sd_ail2_list));
	615
	616	sdp->sd_log_head = sdp->sd_log_flush_head;
	617	if (sdp->sd_log_flush_wrapped)
	618	sdp->sd_log_wraps++;
	619	sdp->sd_log_tail = sdp->sd_log_head;
	620
f55ab26a	621	mutex_unlock(&sdp->sd_log_flush_lock);
b3b94faa DT	622	}
b3b94faa DT	623