[linux.git] / fs / ocfs2 / resize.c

/* -*- mode: c; c-basic-offset: 8; -*-
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * resize.c
 *
 * volume resize.
 * Inspired by ext3/resize.c.
 *
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#include <linux/fs.h>
#include <linux/types.h>

#include <cluster/masklog.h>

#include "ocfs2.h"

#include "alloc.h"
#include "dlmglue.h"
#include "inode.h"
#include "journal.h"
#include "super.h"
#include "sysfile.h"
#include "uptodate.h"
#include "ocfs2_trace.h"

#include "buffer_head_io.h"
#include "suballoc.h"
#include "resize.h"

/*
 * Check whether there are new backup superblocks exist
 * in the last group. If there are some, mark them or clear
 * them in the bitmap.
 *
 * Return how many backups we find in the last group.
 */
static u16 ocfs2_calc_new_backup_super(struct inode *inode,
				       struct ocfs2_group_desc *gd,
				       u16 cl_cpg,
				       int set)
{
	int i;
	u16 backups = 0;
	u32 cluster;
	u64 blkno, gd_blkno, lgd_blkno = le64_to_cpu(gd->bg_blkno);

	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
		blkno = ocfs2_backup_super_blkno(inode->i_sb, i);
		cluster = ocfs2_blocks_to_clusters(inode->i_sb, blkno);

		gd_blkno = ocfs2_which_cluster_group(inode, cluster);
		if (gd_blkno < lgd_blkno)
			continue;
		else if (gd_blkno > lgd_blkno)
			break;

		if (set)
			ocfs2_set_bit(cluster % cl_cpg,
				      (unsigned long *)gd->bg_bitmap);
		else
			ocfs2_clear_bit(cluster % cl_cpg,
					(unsigned long *)gd->bg_bitmap);
		backups++;
	}

	return backups;
}

static int ocfs2_update_last_group_and_inode(handle_t *handle,
					     struct inode *bm_inode,
					     struct buffer_head *bm_bh,
					     struct buffer_head *group_bh,
					     u32 first_new_cluster,
					     int new_clusters)
{
	int ret = 0;
	struct ocfs2_super *osb = OCFS2_SB(bm_inode->i_sb);
	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) bm_bh->b_data;
	struct ocfs2_chain_list *cl = &fe->id2.i_chain;
	struct ocfs2_chain_rec *cr;
	struct ocfs2_group_desc *group;
	u16 chain, num_bits, backups = 0;
	u16 cl_bpc = le16_to_cpu(cl->cl_bpc);
	u16 cl_cpg = le16_to_cpu(cl->cl_cpg);

	trace_ocfs2_update_last_group_and_inode(new_clusters,
						first_new_cluster);

	ret = ocfs2_journal_access_gd(handle, INODE_CACHE(bm_inode),
				      group_bh, OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret < 0) {
		mlog_errno(ret);
		goto out;
	}

	group = (struct ocfs2_group_desc *)group_bh->b_data;

	/* update the group first. */
	num_bits = new_clusters * cl_bpc;
	le16_add_cpu(&group->bg_bits, num_bits);
	le16_add_cpu(&group->bg_free_bits_count, num_bits);

	/*
	 * check whether there are some new backup superblocks exist in
	 * this group and update the group bitmap accordingly.
	 */
	if (OCFS2_HAS_COMPAT_FEATURE(osb->sb,
				     OCFS2_FEATURE_COMPAT_BACKUP_SB)) {
		backups = ocfs2_calc_new_backup_super(bm_inode,
						     group,
						     cl_cpg, 1);
		le16_add_cpu(&group->bg_free_bits_count, -1 * backups);
	}

	ocfs2_journal_dirty(handle, group_bh);

	/* update the inode accordingly. */
	ret = ocfs2_journal_access_di(handle, INODE_CACHE(bm_inode), bm_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret < 0) {
		mlog_errno(ret);
		goto out_rollback;
	}

	chain = le16_to_cpu(group->bg_chain);
	cr = (&cl->cl_recs[chain]);
	le32_add_cpu(&cr->c_total, num_bits);
	le32_add_cpu(&cr->c_free, num_bits);
	le32_add_cpu(&fe->id1.bitmap1.i_total, num_bits);
	le32_add_cpu(&fe->i_clusters, new_clusters);

	if (backups) {
		le32_add_cpu(&cr->c_free, -1 * backups);
		le32_add_cpu(&fe->id1.bitmap1.i_used, backups);
	}

	spin_lock(&OCFS2_I(bm_inode)->ip_lock);
	OCFS2_I(bm_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
	le64_add_cpu(&fe->i_size, (u64)new_clusters << osb->s_clustersize_bits);
	spin_unlock(&OCFS2_I(bm_inode)->ip_lock);
	i_size_write(bm_inode, le64_to_cpu(fe->i_size));

	ocfs2_journal_dirty(handle, bm_bh);

out_rollback:
	if (ret < 0) {
		ocfs2_calc_new_backup_super(bm_inode,
					    group,
					    cl_cpg, 0);
		le16_add_cpu(&group->bg_free_bits_count, backups);
		le16_add_cpu(&group->bg_bits, -1 * num_bits);
		le16_add_cpu(&group->bg_free_bits_count, -1 * num_bits);
	}
out:
	if (ret)
		mlog_errno(ret);
	return ret;
}

static int update_backups(struct inode * inode, u32 clusters, char *data)
{
	int i, ret = 0;
	u32 cluster;
	u64 blkno;
	struct buffer_head *backup = NULL;
	struct ocfs2_dinode *backup_di = NULL;
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);

	/* calculate the real backups we need to update. */
	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
		blkno = ocfs2_backup_super_blkno(inode->i_sb, i);
		cluster = ocfs2_blocks_to_clusters(inode->i_sb, blkno);
		if (cluster > clusters)
			break;

		ret = ocfs2_read_blocks_sync(osb, blkno, 1, &backup);
		if (ret < 0) {
			mlog_errno(ret);
			break;
		}

		memcpy(backup->b_data, data, inode->i_sb->s_blocksize);

		backup_di = (struct ocfs2_dinode *)backup->b_data;
		backup_di->i_blkno = cpu_to_le64(blkno);

		ret = ocfs2_write_super_or_backup(osb, backup);
		brelse(backup);
		backup = NULL;
		if (ret < 0) {
			mlog_errno(ret);
			break;
		}
	}

	return ret;
}

static void ocfs2_update_super_and_backups(struct inode *inode,
					   int new_clusters)
{
	int ret;
	u32 clusters = 0;
	struct buffer_head *super_bh = NULL;
	struct ocfs2_dinode *super_di = NULL;
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);

	/*
	 * update the superblock last.
	 * It doesn't matter if the write failed.
	 */
	ret = ocfs2_read_blocks_sync(osb, OCFS2_SUPER_BLOCK_BLKNO, 1,
				     &super_bh);
	if (ret < 0) {
		mlog_errno(ret);
		goto out;
	}

	super_di = (struct ocfs2_dinode *)super_bh->b_data;
	le32_add_cpu(&super_di->i_clusters, new_clusters);
	clusters = le32_to_cpu(super_di->i_clusters);

	ret = ocfs2_write_super_or_backup(osb, super_bh);
	if (ret < 0) {
		mlog_errno(ret);
		goto out;
	}

	if (OCFS2_HAS_COMPAT_FEATURE(osb->sb, OCFS2_FEATURE_COMPAT_BACKUP_SB))
		ret = update_backups(inode, clusters, super_bh->b_data);

out:
	brelse(super_bh);
	if (ret)
		printk(KERN_WARNING "ocfs2: Failed to update super blocks on %s"
			" during fs resize. This condition is not fatal,"
			" but fsck.ocfs2 should be run to fix it\n",
			osb->dev_str);
	return;
}

/*
 * Extend the filesystem to the new number of clusters specified.  This entry
 * point is only used to extend the current filesystem to the end of the last
 * existing group.
 */
int ocfs2_group_extend(struct inode * inode, int new_clusters)
{
	int ret;
	handle_t *handle;
	struct buffer_head *main_bm_bh = NULL;
	struct buffer_head *group_bh = NULL;
	struct inode *main_bm_inode = NULL;
	struct ocfs2_dinode *fe = NULL;
	struct ocfs2_group_desc *group = NULL;
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	u16 cl_bpc;
	u32 first_new_cluster;
	u64 lgd_blkno;

	if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb))
		return -EROFS;

	if (new_clusters < 0)
		return -EINVAL;
	else if (new_clusters == 0)
		return 0;

	main_bm_inode = ocfs2_get_system_file_inode(osb,
						    GLOBAL_BITMAP_SYSTEM_INODE,
						    OCFS2_INVALID_SLOT);
	if (!main_bm_inode) {
		ret = -EINVAL;
		mlog_errno(ret);
		goto out;
	}

	mutex_lock(&main_bm_inode->i_mutex);

	ret = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
	if (ret < 0) {
		mlog_errno(ret);
		goto out_mutex;
	}

	fe = (struct ocfs2_dinode *)main_bm_bh->b_data;

	/* main_bm_bh is validated by inode read inside ocfs2_inode_lock(),
	 * so any corruption is a code bug. */
	BUG_ON(!OCFS2_IS_VALID_DINODE(fe));

	if (le16_to_cpu(fe->id2.i_chain.cl_cpg) !=
		ocfs2_group_bitmap_size(osb->sb, 0,
					osb->s_feature_incompat) * 8) {
		mlog(ML_ERROR, "The disk is too old and small. "
		     "Force to do offline resize.");
		ret = -EINVAL;
		goto out_unlock;
	}

	first_new_cluster = le32_to_cpu(fe->i_clusters);
	lgd_blkno = ocfs2_which_cluster_group(main_bm_inode,
					      first_new_cluster - 1);

	ret = ocfs2_read_group_descriptor(main_bm_inode, fe, lgd_blkno,
					  &group_bh);
	if (ret < 0) {
		mlog_errno(ret);
		goto out_unlock;
	}
	group = (struct ocfs2_group_desc *)group_bh->b_data;

	cl_bpc = le16_to_cpu(fe->id2.i_chain.cl_bpc);
	if (le16_to_cpu(group->bg_bits) / cl_bpc + new_clusters >
		le16_to_cpu(fe->id2.i_chain.cl_cpg)) {
		ret = -EINVAL;
		goto out_unlock;
	}


	trace_ocfs2_group_extend(
	     (unsigned long long)le64_to_cpu(group->bg_blkno), new_clusters);

	handle = ocfs2_start_trans(osb, OCFS2_GROUP_EXTEND_CREDITS);
	if (IS_ERR(handle)) {
		mlog_errno(PTR_ERR(handle));
		ret = -EINVAL;
		goto out_unlock;
	}

	/* update the last group descriptor and inode. */
	ret = ocfs2_update_last_group_and_inode(handle, main_bm_inode,
						main_bm_bh, group_bh,
						first_new_cluster,
						new_clusters);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	ocfs2_update_super_and_backups(main_bm_inode, new_clusters);

out_commit:
	ocfs2_commit_trans(osb, handle);
out_unlock:
	brelse(group_bh);
	brelse(main_bm_bh);

	ocfs2_inode_unlock(main_bm_inode, 1);

out_mutex:
	mutex_unlock(&main_bm_inode->i_mutex);
	iput(main_bm_inode);

out:
	return ret;
}

static int ocfs2_check_new_group(struct inode *inode,
				 struct ocfs2_dinode *di,
				 struct ocfs2_new_group_input *input,
				 struct buffer_head *group_bh)
{
	int ret;
	struct ocfs2_group_desc *gd =
		(struct ocfs2_group_desc *)group_bh->b_data;
	u16 cl_bpc = le16_to_cpu(di->id2.i_chain.cl_bpc);

	ret = ocfs2_check_group_descriptor(inode->i_sb, di, group_bh);
	if (ret)
		goto out;

	ret = -EINVAL;
	if (le16_to_cpu(gd->bg_chain) != input->chain)
		mlog(ML_ERROR, "Group descriptor # %llu has bad chain %u "
		     "while input has %u set.\n",
		     (unsigned long long)le64_to_cpu(gd->bg_blkno),
		     le16_to_cpu(gd->bg_chain), input->chain);
	else if (le16_to_cpu(gd->bg_bits) != input->clusters * cl_bpc)
		mlog(ML_ERROR, "Group descriptor # %llu has bit count %u but "
		     "input has %u clusters set\n",
		     (unsigned long long)le64_to_cpu(gd->bg_blkno),
		     le16_to_cpu(gd->bg_bits), input->clusters);
	else if (le16_to_cpu(gd->bg_free_bits_count) != input->frees * cl_bpc)
		mlog(ML_ERROR, "Group descriptor # %llu has free bit count %u "
		     "but it should have %u set\n",
		     (unsigned long long)le64_to_cpu(gd->bg_blkno),
		     le16_to_cpu(gd->bg_bits),
		     input->frees * cl_bpc);
	else
		ret = 0;

out:
	return ret;
}

static int ocfs2_verify_group_and_input(struct inode *inode,
					struct ocfs2_dinode *di,
					struct ocfs2_new_group_input *input,
					struct buffer_head *group_bh)
{
	u16 cl_count = le16_to_cpu(di->id2.i_chain.cl_count);
	u16 cl_cpg = le16_to_cpu(di->id2.i_chain.cl_cpg);
	u16 next_free = le16_to_cpu(di->id2.i_chain.cl_next_free_rec);
	u32 cluster = ocfs2_blocks_to_clusters(inode->i_sb, input->group);
	u32 total_clusters = le32_to_cpu(di->i_clusters);
	int ret = -EINVAL;

	if (cluster < total_clusters)
		mlog(ML_ERROR, "add a group which is in the current volume.\n");
	else if (input->chain >= cl_count)
		mlog(ML_ERROR, "input chain exceeds the limit.\n");
	else if (next_free != cl_count && next_free != input->chain)
		mlog(ML_ERROR,
		     "the add group should be in chain %u\n", next_free);
	else if (total_clusters + input->clusters < total_clusters)
		mlog(ML_ERROR, "add group's clusters overflow.\n");
	else if (input->clusters > cl_cpg)
		mlog(ML_ERROR, "the cluster exceeds the maximum of a group\n");
	else if (input->frees > input->clusters)
		mlog(ML_ERROR, "the free cluster exceeds the total clusters\n");
	else if (total_clusters % cl_cpg != 0)
		mlog(ML_ERROR,
		     "the last group isn't full. Use group extend first.\n");
	else if (input->group != ocfs2_which_cluster_group(inode, cluster))
		mlog(ML_ERROR, "group blkno is invalid\n");
	else if ((ret = ocfs2_check_new_group(inode, di, input, group_bh)))
		mlog(ML_ERROR, "group descriptor check failed.\n");
	else
		ret = 0;

	return ret;
}

/* Add a new group descriptor to global_bitmap. */
int ocfs2_group_add(struct inode *inode, struct ocfs2_new_group_input *input)
{
	int ret;
	handle_t *handle;
	struct buffer_head *main_bm_bh = NULL;
	struct inode *main_bm_inode = NULL;
	struct ocfs2_dinode *fe = NULL;
	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	struct buffer_head *group_bh = NULL;
	struct ocfs2_group_desc *group = NULL;
	struct ocfs2_chain_list *cl;
	struct ocfs2_chain_rec *cr;
	u16 cl_bpc;
	u64 bg_ptr;

	if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb))
		return -EROFS;

	main_bm_inode = ocfs2_get_system_file_inode(osb,
						    GLOBAL_BITMAP_SYSTEM_INODE,
						    OCFS2_INVALID_SLOT);
	if (!main_bm_inode) {
		ret = -EINVAL;
		mlog_errno(ret);
		goto out;
	}

	mutex_lock(&main_bm_inode->i_mutex);

	ret = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
	if (ret < 0) {
		mlog_errno(ret);
		goto out_mutex;
	}

	fe = (struct ocfs2_dinode *)main_bm_bh->b_data;

	if (le16_to_cpu(fe->id2.i_chain.cl_cpg) !=
		ocfs2_group_bitmap_size(osb->sb, 0,
					osb->s_feature_incompat) * 8) {
		mlog(ML_ERROR, "The disk is too old and small."
		     " Force to do offline resize.");
		ret = -EINVAL;
		goto out_unlock;
	}

	ret = ocfs2_read_blocks_sync(osb, input->group, 1, &group_bh);
	if (ret < 0) {
		mlog(ML_ERROR, "Can't read the group descriptor # %llu "
		     "from the device.", (unsigned long long)input->group);
		goto out_unlock;
	}

	ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), group_bh);

	ret = ocfs2_verify_group_and_input(main_bm_inode, fe, input, group_bh);
	if (ret) {
		mlog_errno(ret);
		goto out_free_group_bh;
	}

	trace_ocfs2_group_add((unsigned long long)input->group,
			       input->chain, input->clusters, input->frees);

	handle = ocfs2_start_trans(osb, OCFS2_GROUP_ADD_CREDITS);
	if (IS_ERR(handle)) {
		mlog_errno(PTR_ERR(handle));
		ret = -EINVAL;
		goto out_free_group_bh;
	}

	cl_bpc = le16_to_cpu(fe->id2.i_chain.cl_bpc);
	cl = &fe->id2.i_chain;
	cr = &cl->cl_recs[input->chain];

	ret = ocfs2_journal_access_gd(handle, INODE_CACHE(main_bm_inode),
				      group_bh, OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret < 0) {
		mlog_errno(ret);
		goto out_commit;
	}

	group = (struct ocfs2_group_desc *)group_bh->b_data;
	bg_ptr = le64_to_cpu(group->bg_next_group);
	group->bg_next_group = cr->c_blkno;
	ocfs2_journal_dirty(handle, group_bh);

	ret = ocfs2_journal_access_di(handle, INODE_CACHE(main_bm_inode),
				      main_bm_bh, OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret < 0) {
		group->bg_next_group = cpu_to_le64(bg_ptr);
		mlog_errno(ret);
		goto out_commit;
	}

	if (input->chain == le16_to_cpu(cl->cl_next_free_rec)) {
		le16_add_cpu(&cl->cl_next_free_rec, 1);
		memset(cr, 0, sizeof(struct ocfs2_chain_rec));
	}

	cr->c_blkno = cpu_to_le64(input->group);
	le32_add_cpu(&cr->c_total, input->clusters * cl_bpc);
	le32_add_cpu(&cr->c_free, input->frees * cl_bpc);

	le32_add_cpu(&fe->id1.bitmap1.i_total, input->clusters *cl_bpc);
	le32_add_cpu(&fe->id1.bitmap1.i_used,
		     (input->clusters - input->frees) * cl_bpc);
	le32_add_cpu(&fe->i_clusters, input->clusters);

	ocfs2_journal_dirty(handle, main_bm_bh);

	spin_lock(&OCFS2_I(main_bm_inode)->ip_lock);
	OCFS2_I(main_bm_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
	le64_add_cpu(&fe->i_size, (u64)input->clusters << osb->s_clustersize_bits);
	spin_unlock(&OCFS2_I(main_bm_inode)->ip_lock);
	i_size_write(main_bm_inode, le64_to_cpu(fe->i_size));

	ocfs2_update_super_and_backups(main_bm_inode, input->clusters);

out_commit:
	ocfs2_commit_trans(osb, handle);

out_free_group_bh:
	brelse(group_bh);

out_unlock:
	brelse(main_bm_bh);

	ocfs2_inode_unlock(main_bm_inode, 1);

out_mutex:
	mutex_unlock(&main_bm_inode->i_mutex);
	iput(main_bm_inode);

out:
	return ret;
}
Commit	Line	Data
d659072f TM	1	/* -- mode: c; c-basic-offset: 8; --
	2	* vim: noexpandtab sw=8 ts=8 sts=0:
	3	*
	4	* resize.c
	5	*
	6	* volume resize.
	7	* Inspired by ext3/resize.c.
	8	*
	9	* Copyright (C) 2007 Oracle. All rights reserved.
	10	*
	11	* This program is free software; you can redistribute it and/or
	12	* modify it under the terms of the GNU General Public
	13	* License as published by the Free Software Foundation; either
	14	* version 2 of the License, or (at your option) any later version.
	15	*
	16	* This program is distributed in the hope that it will be useful,
	17	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	18	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	19	* General Public License for more details.
	20	*
	21	* You should have received a copy of the GNU General Public
	22	* License along with this program; if not, write to the
	23	* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
	24	* Boston, MA 021110-1307, USA.
	25	*/
	26
	27	#include <linux/fs.h>
	28	#include <linux/types.h>
	29
d659072f TM	30	#include <cluster/masklog.h>
	31
	32	#include "ocfs2.h"
	33
	34	#include "alloc.h"
	35	#include "dlmglue.h"
	36	#include "inode.h"
	37	#include "journal.h"
	38	#include "super.h"
	39	#include "sysfile.h"
	40	#include "uptodate.h"
a5438702	41	#include "ocfs2_trace.h"
d659072f TM	42
	43	#include "buffer_head_io.h"
	44	#include "suballoc.h"
	45	#include "resize.h"
	46
	47	/*
	48	* Check whether there are new backup superblocks exist
	49	* in the last group. If there are some, mark them or clear
	50	* them in the bitmap.
	51	*
	52	* Return how many backups we find in the last group.
	53	*/
	54	static u16 ocfs2_calc_new_backup_super(struct inode *inode,
	55	struct ocfs2_group_desc *gd,
d659072f TM	56	u16 cl_cpg,
	57	int set)
	58	{
	59	int i;
	60	u16 backups = 0;
	61	u32 cluster;
	62	u64 blkno, gd_blkno, lgd_blkno = le64_to_cpu(gd->bg_blkno);
	63
	64	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
	65	blkno = ocfs2_backup_super_blkno(inode->i_sb, i);
	66	cluster = ocfs2_blocks_to_clusters(inode->i_sb, blkno);
	67
	68	gd_blkno = ocfs2_which_cluster_group(inode, cluster);
	69	if (gd_blkno < lgd_blkno)
	70	continue;
	71	else if (gd_blkno > lgd_blkno)
	72	break;
	73
	74	if (set)
	75	ocfs2_set_bit(cluster % cl_cpg,
	76	(unsigned long *)gd->bg_bitmap);
	77	else
	78	ocfs2_clear_bit(cluster % cl_cpg,
	79	(unsigned long *)gd->bg_bitmap);
	80	backups++;
	81	}
	82
d659072f TM	83	return backups;
	84	}
	85
	86	static int ocfs2_update_last_group_and_inode(handle_t *handle,
	87	struct inode *bm_inode,
	88	struct buffer_head *bm_bh,
	89	struct buffer_head *group_bh,
	90	u32 first_new_cluster,
	91	int new_clusters)
	92	{
	93	int ret = 0;
	94	struct ocfs2_super *osb = OCFS2_SB(bm_inode->i_sb);
	95	struct ocfs2_dinode fe = (struct ocfs2_dinode ) bm_bh->b_data;
	96	struct ocfs2_chain_list *cl = &fe->id2.i_chain;
	97	struct ocfs2_chain_rec *cr;
	98	struct ocfs2_group_desc *group;
	99	u16 chain, num_bits, backups = 0;
	100	u16 cl_bpc = le16_to_cpu(cl->cl_bpc);
	101	u16 cl_cpg = le16_to_cpu(cl->cl_cpg);
	102
a5438702 TM	103	trace_ocfs2_update_last_group_and_inode(new_clusters,
a5438702 TM	104	first_new_cluster);
d659072f	105
0cf2f763 JB	106	ret = ocfs2_journal_access_gd(handle, INODE_CACHE(bm_inode),
0cf2f763 JB	107	group_bh, OCFS2_JOURNAL_ACCESS_WRITE);
d659072f TM	108	if (ret < 0) {
	109	mlog_errno(ret);
	110	goto out;
	111	}
	112
	113	group = (struct ocfs2_group_desc *)group_bh->b_data;
	114
	115	/* update the group first. */
	116	num_bits = new_clusters * cl_bpc;
	117	le16_add_cpu(&group->bg_bits, num_bits);
	118	le16_add_cpu(&group->bg_free_bits_count, num_bits);
	119
	120	/*
	121	* check whether there are some new backup superblocks exist in
	122	* this group and update the group bitmap accordingly.
	123	*/
	124	if (OCFS2_HAS_COMPAT_FEATURE(osb->sb,
	125	OCFS2_FEATURE_COMPAT_BACKUP_SB)) {
	126	backups = ocfs2_calc_new_backup_super(bm_inode,
	127	group,
d659072f TM	128	cl_cpg, 1);
	129	le16_add_cpu(&group->bg_free_bits_count, -1 * backups);
	130	}
	131
ec20cec7	132	ocfs2_journal_dirty(handle, group_bh);
d659072f TM	133
d659072f TM	134	/* update the inode accordingly. */
0cf2f763	135	ret = ocfs2_journal_access_di(handle, INODE_CACHE(bm_inode), bm_bh,
13723d00	136	OCFS2_JOURNAL_ACCESS_WRITE);
d659072f TM	137	if (ret < 0) {
	138	mlog_errno(ret);
	139	goto out_rollback;
	140	}
	141
	142	chain = le16_to_cpu(group->bg_chain);
	143	cr = (&cl->cl_recs[chain]);
	144	le32_add_cpu(&cr->c_total, num_bits);
	145	le32_add_cpu(&cr->c_free, num_bits);
	146	le32_add_cpu(&fe->id1.bitmap1.i_total, num_bits);
	147	le32_add_cpu(&fe->i_clusters, new_clusters);
	148
	149	if (backups) {
	150	le32_add_cpu(&cr->c_free, -1 * backups);
	151	le32_add_cpu(&fe->id1.bitmap1.i_used, backups);
	152	}
	153
	154	spin_lock(&OCFS2_I(bm_inode)->ip_lock);
	155	OCFS2_I(bm_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
17bf1418	156	le64_add_cpu(&fe->i_size, (u64)new_clusters << osb->s_clustersize_bits);
d659072f TM	157	spin_unlock(&OCFS2_I(bm_inode)->ip_lock);
	158	i_size_write(bm_inode, le64_to_cpu(fe->i_size));
	159
	160	ocfs2_journal_dirty(handle, bm_bh);
	161
	162	out_rollback:
	163	if (ret < 0) {
	164	ocfs2_calc_new_backup_super(bm_inode,
	165	group,
d659072f TM	166	cl_cpg, 0);
	167	le16_add_cpu(&group->bg_free_bits_count, backups);
	168	le16_add_cpu(&group->bg_bits, -1 * num_bits);
	169	le16_add_cpu(&group->bg_free_bits_count, -1 * num_bits);
	170	}
	171	out:
c1e8d35e TM	172	if (ret)
c1e8d35e TM	173	mlog_errno(ret);
d659072f TM	174	return ret;
	175	}
	176
	177	static int update_backups(struct inode * inode, u32 clusters, char *data)
	178	{
	179	int i, ret = 0;
	180	u32 cluster;
	181	u64 blkno;
	182	struct buffer_head *backup = NULL;
	183	struct ocfs2_dinode *backup_di = NULL;
	184	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	185
	186	/* calculate the real backups we need to update. */
	187	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
	188	blkno = ocfs2_backup_super_blkno(inode->i_sb, i);
	189	cluster = ocfs2_blocks_to_clusters(inode->i_sb, blkno);
	190	if (cluster > clusters)
	191	break;
	192
da1e9098	193	ret = ocfs2_read_blocks_sync(osb, blkno, 1, &backup);
d659072f TM	194	if (ret < 0) {
	195	mlog_errno(ret);
	196	break;
	197	}
	198
	199	memcpy(backup->b_data, data, inode->i_sb->s_blocksize);
	200
	201	backup_di = (struct ocfs2_dinode *)backup->b_data;
	202	backup_di->i_blkno = cpu_to_le64(blkno);
	203
	204	ret = ocfs2_write_super_or_backup(osb, backup);
	205	brelse(backup);
	206	backup = NULL;
	207	if (ret < 0) {
	208	mlog_errno(ret);
	209	break;
	210	}
	211	}
	212
	213	return ret;
	214	}
	215
	216	static void ocfs2_update_super_and_backups(struct inode *inode,
	217	int new_clusters)
	218	{
	219	int ret;
	220	u32 clusters = 0;
	221	struct buffer_head *super_bh = NULL;
	222	struct ocfs2_dinode *super_di = NULL;
	223	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	224
	225	/*
	226	* update the superblock last.
	227	* It doesn't matter if the write failed.
	228	*/
da1e9098 JB	229	ret = ocfs2_read_blocks_sync(osb, OCFS2_SUPER_BLOCK_BLKNO, 1,
da1e9098 JB	230	&super_bh);
d659072f TM	231	if (ret < 0) {
	232	mlog_errno(ret);
	233	goto out;
	234	}
	235
	236	super_di = (struct ocfs2_dinode *)super_bh->b_data;
	237	le32_add_cpu(&super_di->i_clusters, new_clusters);
	238	clusters = le32_to_cpu(super_di->i_clusters);
	239
	240	ret = ocfs2_write_super_or_backup(osb, super_bh);
	241	if (ret < 0) {
	242	mlog_errno(ret);
	243	goto out;
	244	}
	245
	246	if (OCFS2_HAS_COMPAT_FEATURE(osb->sb, OCFS2_FEATURE_COMPAT_BACKUP_SB))
	247	ret = update_backups(inode, clusters, super_bh->b_data);
	248
	249	out:
2fe5c1d7	250	brelse(super_bh);
d659072f TM	251	if (ret)
	252	printk(KERN_WARNING "ocfs2: Failed to update super blocks on %s"
	253	" during fs resize. This condition is not fatal,"
	254	" but fsck.ocfs2 should be run to fix it\n",
	255	osb->dev_str);
	256	return;
	257	}
	258
	259	/*
	260	* Extend the filesystem to the new number of clusters specified. This entry
	261	* point is only used to extend the current filesystem to the end of the last
	262	* existing group.
	263	*/
	264	int ocfs2_group_extend(struct inode * inode, int new_clusters)
	265	{
	266	int ret;
	267	handle_t *handle;
	268	struct buffer_head *main_bm_bh = NULL;
	269	struct buffer_head *group_bh = NULL;
	270	struct inode *main_bm_inode = NULL;
	271	struct ocfs2_dinode *fe = NULL;
	272	struct ocfs2_group_desc *group = NULL;
	273	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	274	u16 cl_bpc;
	275	u32 first_new_cluster;
	276	u64 lgd_blkno;
	277
d659072f TM	278	if (ocfs2_is_hard_readonly(osb) \|\| ocfs2_is_soft_readonly(osb))
	279	return -EROFS;
	280
	281	if (new_clusters < 0)
	282	return -EINVAL;
	283	else if (new_clusters == 0)
	284	return 0;
	285
	286	main_bm_inode = ocfs2_get_system_file_inode(osb,
	287	GLOBAL_BITMAP_SYSTEM_INODE,
	288	OCFS2_INVALID_SLOT);
	289	if (!main_bm_inode) {
	290	ret = -EINVAL;
	291	mlog_errno(ret);
	292	goto out;
	293	}
	294
	295	mutex_lock(&main_bm_inode->i_mutex);
	296
	297	ret = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
	298	if (ret < 0) {
	299	mlog_errno(ret);
	300	goto out_mutex;
	301	}
	302
	303	fe = (struct ocfs2_dinode *)main_bm_bh->b_data;
	304
10995aa2 JB	305	/* main_bm_bh is validated by inode read inside ocfs2_inode_lock(),
	306	* so any corruption is a code bug. */
	307	BUG_ON(!OCFS2_IS_VALID_DINODE(fe));
	308
d659072f	309	if (le16_to_cpu(fe->id2.i_chain.cl_cpg) !=
8571882c TM	310	ocfs2_group_bitmap_size(osb->sb, 0,
8571882c TM	311	osb->s_feature_incompat) * 8) {
d659072f TM	312	mlog(ML_ERROR, "The disk is too old and small. "
	313	"Force to do offline resize.");
	314	ret = -EINVAL;
	315	goto out_unlock;
d659072f TM	316	}
	317
	318	first_new_cluster = le32_to_cpu(fe->i_clusters);
	319	lgd_blkno = ocfs2_which_cluster_group(main_bm_inode,
	320	first_new_cluster - 1);
	321
68f64d47 JB	322	ret = ocfs2_read_group_descriptor(main_bm_inode, fe, lgd_blkno,
68f64d47 JB	323	&group_bh);
d659072f TM	324	if (ret < 0) {
	325	mlog_errno(ret);
	326	goto out_unlock;
	327	}
d659072f TM	328	group = (struct ocfs2_group_desc *)group_bh->b_data;
d659072f TM	329
d659072f TM	330	cl_bpc = le16_to_cpu(fe->id2.i_chain.cl_bpc);
	331	if (le16_to_cpu(group->bg_bits) / cl_bpc + new_clusters >
	332	le16_to_cpu(fe->id2.i_chain.cl_cpg)) {
	333	ret = -EINVAL;
	334	goto out_unlock;
	335	}
	336
a5438702 TM	337
a5438702 TM	338	trace_ocfs2_group_extend(
7909f2bf	339	(unsigned long long)le64_to_cpu(group->bg_blkno), new_clusters);
d659072f TM	340
	341	handle = ocfs2_start_trans(osb, OCFS2_GROUP_EXTEND_CREDITS);
	342	if (IS_ERR(handle)) {
	343	mlog_errno(PTR_ERR(handle));
	344	ret = -EINVAL;
	345	goto out_unlock;
	346	}
	347
	348	/* update the last group descriptor and inode. */
	349	ret = ocfs2_update_last_group_and_inode(handle, main_bm_inode,
	350	main_bm_bh, group_bh,
	351	first_new_cluster,
	352	new_clusters);
	353	if (ret) {
	354	mlog_errno(ret);
	355	goto out_commit;
	356	}
	357
	358	ocfs2_update_super_and_backups(main_bm_inode, new_clusters);
	359
	360	out_commit:
	361	ocfs2_commit_trans(osb, handle);
	362	out_unlock:
2fe5c1d7 MF	363	brelse(group_bh);
2fe5c1d7 MF	364	brelse(main_bm_bh);
d659072f TM	365
	366	ocfs2_inode_unlock(main_bm_inode, 1);
	367
	368	out_mutex:
	369	mutex_unlock(&main_bm_inode->i_mutex);
	370	iput(main_bm_inode);
	371
	372	out:
d659072f TM	373	return ret;
d659072f TM	374	}
7909f2bf TM	375
	376	static int ocfs2_check_new_group(struct inode *inode,
	377	struct ocfs2_dinode *di,
	378	struct ocfs2_new_group_input *input,
	379	struct buffer_head *group_bh)
	380	{
	381	int ret;
57e3e797 JB	382	struct ocfs2_group_desc *gd =
57e3e797 JB	383	(struct ocfs2_group_desc *)group_bh->b_data;
7909f2bf	384	u16 cl_bpc = le16_to_cpu(di->id2.i_chain.cl_bpc);
7909f2bf	385
970e4936	386	ret = ocfs2_check_group_descriptor(inode->i_sb, di, group_bh);
57e3e797 JB	387	if (ret)
57e3e797 JB	388	goto out;
7909f2bf	389
57e3e797 JB	390	ret = -EINVAL;
57e3e797 JB	391	if (le16_to_cpu(gd->bg_chain) != input->chain)
7909f2bf TM	392	mlog(ML_ERROR, "Group descriptor # %llu has bad chain %u "
	393	"while input has %u set.\n",
	394	(unsigned long long)le64_to_cpu(gd->bg_blkno),
	395	le16_to_cpu(gd->bg_chain), input->chain);
	396	else if (le16_to_cpu(gd->bg_bits) != input->clusters * cl_bpc)
	397	mlog(ML_ERROR, "Group descriptor # %llu has bit count %u but "
	398	"input has %u clusters set\n",
	399	(unsigned long long)le64_to_cpu(gd->bg_blkno),
	400	le16_to_cpu(gd->bg_bits), input->clusters);
	401	else if (le16_to_cpu(gd->bg_free_bits_count) != input->frees * cl_bpc)
	402	mlog(ML_ERROR, "Group descriptor # %llu has free bit count %u "
	403	"but it should have %u set\n",
	404	(unsigned long long)le64_to_cpu(gd->bg_blkno),
	405	le16_to_cpu(gd->bg_bits),
	406	input->frees * cl_bpc);
	407	else
	408	ret = 0;
	409
57e3e797	410	out:
7909f2bf TM	411	return ret;
	412	}
	413
	414	static int ocfs2_verify_group_and_input(struct inode *inode,
	415	struct ocfs2_dinode *di,
	416	struct ocfs2_new_group_input *input,
	417	struct buffer_head *group_bh)
	418	{
	419	u16 cl_count = le16_to_cpu(di->id2.i_chain.cl_count);
	420	u16 cl_cpg = le16_to_cpu(di->id2.i_chain.cl_cpg);
	421	u16 next_free = le16_to_cpu(di->id2.i_chain.cl_next_free_rec);
	422	u32 cluster = ocfs2_blocks_to_clusters(inode->i_sb, input->group);
	423	u32 total_clusters = le32_to_cpu(di->i_clusters);
	424	int ret = -EINVAL;
	425
	426	if (cluster < total_clusters)
	427	mlog(ML_ERROR, "add a group which is in the current volume.\n");
	428	else if (input->chain >= cl_count)
	429	mlog(ML_ERROR, "input chain exceeds the limit.\n");
	430	else if (next_free != cl_count && next_free != input->chain)
	431	mlog(ML_ERROR,
	432	"the add group should be in chain %u\n", next_free);
	433	else if (total_clusters + input->clusters < total_clusters)
	434	mlog(ML_ERROR, "add group's clusters overflow.\n");
	435	else if (input->clusters > cl_cpg)
	436	mlog(ML_ERROR, "the cluster exceeds the maximum of a group\n");
	437	else if (input->frees > input->clusters)
	438	mlog(ML_ERROR, "the free cluster exceeds the total clusters\n");
	439	else if (total_clusters % cl_cpg != 0)
	440	mlog(ML_ERROR,
	441	"the last group isn't full. Use group extend first.\n");
	442	else if (input->group != ocfs2_which_cluster_group(inode, cluster))
	443	mlog(ML_ERROR, "group blkno is invalid\n");
	444	else if ((ret = ocfs2_check_new_group(inode, di, input, group_bh)))
	445	mlog(ML_ERROR, "group descriptor check failed.\n");
	446	else
	447	ret = 0;
	448
	449	return ret;
	450	}
	451
	452	/* Add a new group descriptor to global_bitmap. */
	453	int ocfs2_group_add(struct inode inode, struct ocfs2_new_group_input input)
	454	{
	455	int ret;
	456	handle_t *handle;
	457	struct buffer_head *main_bm_bh = NULL;
	458	struct inode *main_bm_inode = NULL;
	459	struct ocfs2_dinode *fe = NULL;
	460	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	461	struct buffer_head *group_bh = NULL;
	462	struct ocfs2_group_desc *group = NULL;
	463	struct ocfs2_chain_list *cl;
	464	struct ocfs2_chain_rec *cr;
	465	u16 cl_bpc;
eedd40e1	466	u64 bg_ptr;
7909f2bf	467
7909f2bf TM	468	if (ocfs2_is_hard_readonly(osb) \|\| ocfs2_is_soft_readonly(osb))
	469	return -EROFS;
	470
	471	main_bm_inode = ocfs2_get_system_file_inode(osb,
	472	GLOBAL_BITMAP_SYSTEM_INODE,
	473	OCFS2_INVALID_SLOT);
	474	if (!main_bm_inode) {
	475	ret = -EINVAL;
	476	mlog_errno(ret);
	477	goto out;
	478	}
	479
	480	mutex_lock(&main_bm_inode->i_mutex);
	481
	482	ret = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
	483	if (ret < 0) {
	484	mlog_errno(ret);
	485	goto out_mutex;
	486	}
	487
	488	fe = (struct ocfs2_dinode *)main_bm_bh->b_data;
	489
	490	if (le16_to_cpu(fe->id2.i_chain.cl_cpg) !=
8571882c TM	491	ocfs2_group_bitmap_size(osb->sb, 0,
8571882c TM	492	osb->s_feature_incompat) * 8) {
7909f2bf TM	493	mlog(ML_ERROR, "The disk is too old and small."
	494	" Force to do offline resize.");
	495	ret = -EINVAL;
	496	goto out_unlock;
	497	}
	498
da1e9098	499	ret = ocfs2_read_blocks_sync(osb, input->group, 1, &group_bh);
7909f2bf TM	500	if (ret < 0) {
	501	mlog(ML_ERROR, "Can't read the group descriptor # %llu "
	502	"from the device.", (unsigned long long)input->group);
	503	goto out_unlock;
	504	}
	505
8cb471e8	506	ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), group_bh);
7909f2bf TM	507
	508	ret = ocfs2_verify_group_and_input(main_bm_inode, fe, input, group_bh);
	509	if (ret) {
	510	mlog_errno(ret);
8abaae8d	511	goto out_free_group_bh;
7909f2bf TM	512	}
7909f2bf TM	513
a5438702 TM	514	trace_ocfs2_group_add((unsigned long long)input->group,
a5438702 TM	515	input->chain, input->clusters, input->frees);
7909f2bf TM	516
	517	handle = ocfs2_start_trans(osb, OCFS2_GROUP_ADD_CREDITS);
	518	if (IS_ERR(handle)) {
	519	mlog_errno(PTR_ERR(handle));
	520	ret = -EINVAL;
8abaae8d	521	goto out_free_group_bh;
7909f2bf TM	522	}
	523
	524	cl_bpc = le16_to_cpu(fe->id2.i_chain.cl_bpc);
	525	cl = &fe->id2.i_chain;
	526	cr = &cl->cl_recs[input->chain];
	527
0cf2f763 JB	528	ret = ocfs2_journal_access_gd(handle, INODE_CACHE(main_bm_inode),
0cf2f763 JB	529	group_bh, OCFS2_JOURNAL_ACCESS_WRITE);
7909f2bf TM	530	if (ret < 0) {
	531	mlog_errno(ret);
	532	goto out_commit;
	533	}
	534
	535	group = (struct ocfs2_group_desc *)group_bh->b_data;
eedd40e1	536	bg_ptr = le64_to_cpu(group->bg_next_group);
7909f2bf	537	group->bg_next_group = cr->c_blkno;
ec20cec7	538	ocfs2_journal_dirty(handle, group_bh);
7909f2bf	539
0cf2f763 JB	540	ret = ocfs2_journal_access_di(handle, INODE_CACHE(main_bm_inode),
0cf2f763 JB	541	main_bm_bh, OCFS2_JOURNAL_ACCESS_WRITE);
7909f2bf	542	if (ret < 0) {
eedd40e1	543	group->bg_next_group = cpu_to_le64(bg_ptr);
7909f2bf TM	544	mlog_errno(ret);
	545	goto out_commit;
	546	}
	547
	548	if (input->chain == le16_to_cpu(cl->cl_next_free_rec)) {
	549	le16_add_cpu(&cl->cl_next_free_rec, 1);
	550	memset(cr, 0, sizeof(struct ocfs2_chain_rec));
	551	}
	552
4338ab6a	553	cr->c_blkno = cpu_to_le64(input->group);
7909f2bf TM	554	le32_add_cpu(&cr->c_total, input->clusters * cl_bpc);
	555	le32_add_cpu(&cr->c_free, input->frees * cl_bpc);
	556
	557	le32_add_cpu(&fe->id1.bitmap1.i_total, input->clusters *cl_bpc);
	558	le32_add_cpu(&fe->id1.bitmap1.i_used,
	559	(input->clusters - input->frees) * cl_bpc);
	560	le32_add_cpu(&fe->i_clusters, input->clusters);
	561
	562	ocfs2_journal_dirty(handle, main_bm_bh);
	563
	564	spin_lock(&OCFS2_I(main_bm_inode)->ip_lock);
	565	OCFS2_I(main_bm_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
17bf1418	566	le64_add_cpu(&fe->i_size, (u64)input->clusters << osb->s_clustersize_bits);
7909f2bf TM	567	spin_unlock(&OCFS2_I(main_bm_inode)->ip_lock);
	568	i_size_write(main_bm_inode, le64_to_cpu(fe->i_size));
	569
	570	ocfs2_update_super_and_backups(main_bm_inode, input->clusters);
	571
	572	out_commit:
	573	ocfs2_commit_trans(osb, handle);
8abaae8d YL	574
8abaae8d YL	575	out_free_group_bh:
2fe5c1d7	576	brelse(group_bh);
8abaae8d YL	577
8abaae8d YL	578	out_unlock:
2fe5c1d7	579	brelse(main_bm_bh);
7909f2bf TM	580
	581	ocfs2_inode_unlock(main_bm_inode, 1);
	582
	583	out_mutex:
	584	mutex_unlock(&main_bm_inode->i_mutex);
	585	iput(main_bm_inode);
	586
	587	out:
7909f2bf TM	588	return ret;
7909f2bf TM	589	}