[mirror_ubuntu-artful-kernel.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,
				       u16 old_bg_clusters,
				       int set)
{
	int i;
	u16 backups = 0;
	u32 cluster, lgd_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;

		/* check if already done backup super */
		lgd_cluster = ocfs2_blocks_to_clusters(inode->i_sb, lgd_blkno);
		lgd_cluster += old_bg_clusters;
		if (lgd_cluster >= cluster)
			continue;

		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);
	u16 old_bg_clusters;

	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;

	old_bg_clusters = le16_to_cpu(group->bg_bits) / cl_bpc;
	/* 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, old_bg_clusters, 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, old_bg_clusters, 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;
	}

	inode_lock(main_bm_inode);

	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:
	inode_unlock(main_bm_inode);
	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;
	}

	inode_lock(main_bm_inode);

	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:
	inode_unlock(main_bm_inode);
	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	56	u16 cl_cpg,
5c9ee4cb	57	u16 old_bg_clusters,
d659072f TM	58	int set)
	59	{
	60	int i;
	61	u16 backups = 0;
5c9ee4cb	62	u32 cluster, lgd_cluster;
d659072f TM	63	u64 blkno, gd_blkno, lgd_blkno = le64_to_cpu(gd->bg_blkno);
	64
	65	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
	66	blkno = ocfs2_backup_super_blkno(inode->i_sb, i);
	67	cluster = ocfs2_blocks_to_clusters(inode->i_sb, blkno);
	68
	69	gd_blkno = ocfs2_which_cluster_group(inode, cluster);
	70	if (gd_blkno < lgd_blkno)
	71	continue;
	72	else if (gd_blkno > lgd_blkno)
	73	break;
	74
5c9ee4cb JQ	75	/* check if already done backup super */
	76	lgd_cluster = ocfs2_blocks_to_clusters(inode->i_sb, lgd_blkno);
	77	lgd_cluster += old_bg_clusters;
	78	if (lgd_cluster >= cluster)
	79	continue;
	80
d659072f TM	81	if (set)
	82	ocfs2_set_bit(cluster % cl_cpg,
	83	(unsigned long *)gd->bg_bitmap);
	84	else
	85	ocfs2_clear_bit(cluster % cl_cpg,
	86	(unsigned long *)gd->bg_bitmap);
	87	backups++;
	88	}
	89
d659072f TM	90	return backups;
	91	}
	92
	93	static int ocfs2_update_last_group_and_inode(handle_t *handle,
	94	struct inode *bm_inode,
	95	struct buffer_head *bm_bh,
	96	struct buffer_head *group_bh,
	97	u32 first_new_cluster,
	98	int new_clusters)
	99	{
	100	int ret = 0;
	101	struct ocfs2_super *osb = OCFS2_SB(bm_inode->i_sb);
	102	struct ocfs2_dinode fe = (struct ocfs2_dinode ) bm_bh->b_data;
	103	struct ocfs2_chain_list *cl = &fe->id2.i_chain;
	104	struct ocfs2_chain_rec *cr;
	105	struct ocfs2_group_desc *group;
	106	u16 chain, num_bits, backups = 0;
	107	u16 cl_bpc = le16_to_cpu(cl->cl_bpc);
	108	u16 cl_cpg = le16_to_cpu(cl->cl_cpg);
5c9ee4cb	109	u16 old_bg_clusters;
d659072f	110
a5438702 TM	111	trace_ocfs2_update_last_group_and_inode(new_clusters,
a5438702 TM	112	first_new_cluster);
d659072f	113
0cf2f763 JB	114	ret = ocfs2_journal_access_gd(handle, INODE_CACHE(bm_inode),
0cf2f763 JB	115	group_bh, OCFS2_JOURNAL_ACCESS_WRITE);
d659072f TM	116	if (ret < 0) {
	117	mlog_errno(ret);
	118	goto out;
	119	}
	120
	121	group = (struct ocfs2_group_desc *)group_bh->b_data;
	122
5c9ee4cb	123	old_bg_clusters = le16_to_cpu(group->bg_bits) / cl_bpc;
d659072f TM	124	/* update the group first. */
	125	num_bits = new_clusters * cl_bpc;
	126	le16_add_cpu(&group->bg_bits, num_bits);
	127	le16_add_cpu(&group->bg_free_bits_count, num_bits);
	128
	129	/*
	130	* check whether there are some new backup superblocks exist in
	131	* this group and update the group bitmap accordingly.
	132	*/
	133	if (OCFS2_HAS_COMPAT_FEATURE(osb->sb,
	134	OCFS2_FEATURE_COMPAT_BACKUP_SB)) {
	135	backups = ocfs2_calc_new_backup_super(bm_inode,
	136	group,
5c9ee4cb	137	cl_cpg, old_bg_clusters, 1);
d659072f TM	138	le16_add_cpu(&group->bg_free_bits_count, -1 * backups);
	139	}
	140
ec20cec7	141	ocfs2_journal_dirty(handle, group_bh);
d659072f TM	142
d659072f TM	143	/* update the inode accordingly. */
0cf2f763	144	ret = ocfs2_journal_access_di(handle, INODE_CACHE(bm_inode), bm_bh,
13723d00	145	OCFS2_JOURNAL_ACCESS_WRITE);
d659072f TM	146	if (ret < 0) {
	147	mlog_errno(ret);
	148	goto out_rollback;
	149	}
	150
	151	chain = le16_to_cpu(group->bg_chain);
	152	cr = (&cl->cl_recs[chain]);
	153	le32_add_cpu(&cr->c_total, num_bits);
	154	le32_add_cpu(&cr->c_free, num_bits);
	155	le32_add_cpu(&fe->id1.bitmap1.i_total, num_bits);
	156	le32_add_cpu(&fe->i_clusters, new_clusters);
	157
	158	if (backups) {
	159	le32_add_cpu(&cr->c_free, -1 * backups);
	160	le32_add_cpu(&fe->id1.bitmap1.i_used, backups);
	161	}
	162
	163	spin_lock(&OCFS2_I(bm_inode)->ip_lock);
	164	OCFS2_I(bm_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
17bf1418	165	le64_add_cpu(&fe->i_size, (u64)new_clusters << osb->s_clustersize_bits);
d659072f TM	166	spin_unlock(&OCFS2_I(bm_inode)->ip_lock);
	167	i_size_write(bm_inode, le64_to_cpu(fe->i_size));
	168
	169	ocfs2_journal_dirty(handle, bm_bh);
	170
	171	out_rollback:
	172	if (ret < 0) {
	173	ocfs2_calc_new_backup_super(bm_inode,
	174	group,
5c9ee4cb	175	cl_cpg, old_bg_clusters, 0);
d659072f TM	176	le16_add_cpu(&group->bg_free_bits_count, backups);
	177	le16_add_cpu(&group->bg_bits, -1 * num_bits);
	178	le16_add_cpu(&group->bg_free_bits_count, -1 * num_bits);
	179	}
	180	out:
c1e8d35e TM	181	if (ret)
c1e8d35e TM	182	mlog_errno(ret);
d659072f TM	183	return ret;
	184	}
	185
	186	static int update_backups(struct inode * inode, u32 clusters, char *data)
	187	{
	188	int i, ret = 0;
	189	u32 cluster;
	190	u64 blkno;
	191	struct buffer_head *backup = NULL;
	192	struct ocfs2_dinode *backup_di = NULL;
	193	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	194
	195	/* calculate the real backups we need to update. */
	196	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
	197	blkno = ocfs2_backup_super_blkno(inode->i_sb, i);
	198	cluster = ocfs2_blocks_to_clusters(inode->i_sb, blkno);
584dca34	199	if (cluster >= clusters)
d659072f TM	200	break;
d659072f TM	201
da1e9098	202	ret = ocfs2_read_blocks_sync(osb, blkno, 1, &backup);
d659072f TM	203	if (ret < 0) {
	204	mlog_errno(ret);
	205	break;
	206	}
	207
	208	memcpy(backup->b_data, data, inode->i_sb->s_blocksize);
	209
	210	backup_di = (struct ocfs2_dinode *)backup->b_data;
	211	backup_di->i_blkno = cpu_to_le64(blkno);
	212
	213	ret = ocfs2_write_super_or_backup(osb, backup);
	214	brelse(backup);
	215	backup = NULL;
	216	if (ret < 0) {
	217	mlog_errno(ret);
	218	break;
	219	}
	220	}
	221
	222	return ret;
	223	}
	224
	225	static void ocfs2_update_super_and_backups(struct inode *inode,
	226	int new_clusters)
	227	{
	228	int ret;
	229	u32 clusters = 0;
	230	struct buffer_head *super_bh = NULL;
	231	struct ocfs2_dinode *super_di = NULL;
	232	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	233
	234	/*
	235	* update the superblock last.
	236	* It doesn't matter if the write failed.
	237	*/
da1e9098 JB	238	ret = ocfs2_read_blocks_sync(osb, OCFS2_SUPER_BLOCK_BLKNO, 1,
da1e9098 JB	239	&super_bh);
d659072f TM	240	if (ret < 0) {
	241	mlog_errno(ret);
	242	goto out;
	243	}
	244
	245	super_di = (struct ocfs2_dinode *)super_bh->b_data;
	246	le32_add_cpu(&super_di->i_clusters, new_clusters);
	247	clusters = le32_to_cpu(super_di->i_clusters);
	248
	249	ret = ocfs2_write_super_or_backup(osb, super_bh);
	250	if (ret < 0) {
	251	mlog_errno(ret);
	252	goto out;
	253	}
	254
	255	if (OCFS2_HAS_COMPAT_FEATURE(osb->sb, OCFS2_FEATURE_COMPAT_BACKUP_SB))
	256	ret = update_backups(inode, clusters, super_bh->b_data);
	257
	258	out:
2fe5c1d7	259	brelse(super_bh);
d659072f TM	260	if (ret)
	261	printk(KERN_WARNING "ocfs2: Failed to update super blocks on %s"
	262	" during fs resize. This condition is not fatal,"
	263	" but fsck.ocfs2 should be run to fix it\n",
	264	osb->dev_str);
	265	return;
	266	}
	267
	268	/*
	269	* Extend the filesystem to the new number of clusters specified. This entry
	270	* point is only used to extend the current filesystem to the end of the last
	271	* existing group.
	272	*/
	273	int ocfs2_group_extend(struct inode * inode, int new_clusters)
	274	{
	275	int ret;
	276	handle_t *handle;
	277	struct buffer_head *main_bm_bh = NULL;
	278	struct buffer_head *group_bh = NULL;
	279	struct inode *main_bm_inode = NULL;
	280	struct ocfs2_dinode *fe = NULL;
	281	struct ocfs2_group_desc *group = NULL;
	282	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	283	u16 cl_bpc;
	284	u32 first_new_cluster;
	285	u64 lgd_blkno;
	286
d659072f TM	287	if (ocfs2_is_hard_readonly(osb) \|\| ocfs2_is_soft_readonly(osb))
	288	return -EROFS;
	289
	290	if (new_clusters < 0)
	291	return -EINVAL;
	292	else if (new_clusters == 0)
	293	return 0;
	294
	295	main_bm_inode = ocfs2_get_system_file_inode(osb,
	296	GLOBAL_BITMAP_SYSTEM_INODE,
	297	OCFS2_INVALID_SLOT);
	298	if (!main_bm_inode) {
	299	ret = -EINVAL;
	300	mlog_errno(ret);
	301	goto out;
	302	}
	303
5955102c	304	inode_lock(main_bm_inode);
d659072f TM	305
	306	ret = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
	307	if (ret < 0) {
	308	mlog_errno(ret);
	309	goto out_mutex;
	310	}
	311
	312	fe = (struct ocfs2_dinode *)main_bm_bh->b_data;
	313
10995aa2 JB	314	/* main_bm_bh is validated by inode read inside ocfs2_inode_lock(),
	315	* so any corruption is a code bug. */
	316	BUG_ON(!OCFS2_IS_VALID_DINODE(fe));
	317
d659072f	318	if (le16_to_cpu(fe->id2.i_chain.cl_cpg) !=
8571882c TM	319	ocfs2_group_bitmap_size(osb->sb, 0,
8571882c TM	320	osb->s_feature_incompat) * 8) {
d659072f TM	321	mlog(ML_ERROR, "The disk is too old and small. "
	322	"Force to do offline resize.");
	323	ret = -EINVAL;
	324	goto out_unlock;
d659072f TM	325	}
	326
	327	first_new_cluster = le32_to_cpu(fe->i_clusters);
	328	lgd_blkno = ocfs2_which_cluster_group(main_bm_inode,
	329	first_new_cluster - 1);
	330
68f64d47 JB	331	ret = ocfs2_read_group_descriptor(main_bm_inode, fe, lgd_blkno,
68f64d47 JB	332	&group_bh);
d659072f TM	333	if (ret < 0) {
	334	mlog_errno(ret);
	335	goto out_unlock;
	336	}
d659072f TM	337	group = (struct ocfs2_group_desc *)group_bh->b_data;
d659072f TM	338
d659072f TM	339	cl_bpc = le16_to_cpu(fe->id2.i_chain.cl_bpc);
	340	if (le16_to_cpu(group->bg_bits) / cl_bpc + new_clusters >
	341	le16_to_cpu(fe->id2.i_chain.cl_cpg)) {
	342	ret = -EINVAL;
	343	goto out_unlock;
	344	}
	345
a5438702 TM	346
a5438702 TM	347	trace_ocfs2_group_extend(
7909f2bf	348	(unsigned long long)le64_to_cpu(group->bg_blkno), new_clusters);
d659072f TM	349
	350	handle = ocfs2_start_trans(osb, OCFS2_GROUP_EXTEND_CREDITS);
	351	if (IS_ERR(handle)) {
	352	mlog_errno(PTR_ERR(handle));
	353	ret = -EINVAL;
	354	goto out_unlock;
	355	}
	356
	357	/* update the last group descriptor and inode. */
	358	ret = ocfs2_update_last_group_and_inode(handle, main_bm_inode,
	359	main_bm_bh, group_bh,
	360	first_new_cluster,
	361	new_clusters);
	362	if (ret) {
	363	mlog_errno(ret);
	364	goto out_commit;
	365	}
	366
	367	ocfs2_update_super_and_backups(main_bm_inode, new_clusters);
	368
	369	out_commit:
	370	ocfs2_commit_trans(osb, handle);
	371	out_unlock:
2fe5c1d7 MF	372	brelse(group_bh);
2fe5c1d7 MF	373	brelse(main_bm_bh);
d659072f TM	374
	375	ocfs2_inode_unlock(main_bm_inode, 1);
	376
	377	out_mutex:
5955102c	378	inode_unlock(main_bm_inode);
d659072f TM	379	iput(main_bm_inode);
	380
	381	out:
d659072f TM	382	return ret;
d659072f TM	383	}
7909f2bf TM	384
	385	static int ocfs2_check_new_group(struct inode *inode,
	386	struct ocfs2_dinode *di,
	387	struct ocfs2_new_group_input *input,
	388	struct buffer_head *group_bh)
	389	{
	390	int ret;
57e3e797 JB	391	struct ocfs2_group_desc *gd =
57e3e797 JB	392	(struct ocfs2_group_desc *)group_bh->b_data;
7909f2bf	393	u16 cl_bpc = le16_to_cpu(di->id2.i_chain.cl_bpc);
7909f2bf	394
970e4936	395	ret = ocfs2_check_group_descriptor(inode->i_sb, di, group_bh);
57e3e797 JB	396	if (ret)
57e3e797 JB	397	goto out;
7909f2bf	398
57e3e797 JB	399	ret = -EINVAL;
57e3e797 JB	400	if (le16_to_cpu(gd->bg_chain) != input->chain)
7909f2bf TM	401	mlog(ML_ERROR, "Group descriptor # %llu has bad chain %u "
	402	"while input has %u set.\n",
	403	(unsigned long long)le64_to_cpu(gd->bg_blkno),
	404	le16_to_cpu(gd->bg_chain), input->chain);
	405	else if (le16_to_cpu(gd->bg_bits) != input->clusters * cl_bpc)
	406	mlog(ML_ERROR, "Group descriptor # %llu has bit count %u but "
	407	"input has %u clusters set\n",
	408	(unsigned long long)le64_to_cpu(gd->bg_blkno),
	409	le16_to_cpu(gd->bg_bits), input->clusters);
	410	else if (le16_to_cpu(gd->bg_free_bits_count) != input->frees * cl_bpc)
	411	mlog(ML_ERROR, "Group descriptor # %llu has free bit count %u "
	412	"but it should have %u set\n",
	413	(unsigned long long)le64_to_cpu(gd->bg_blkno),
	414	le16_to_cpu(gd->bg_bits),
	415	input->frees * cl_bpc);
	416	else
	417	ret = 0;
	418
57e3e797	419	out:
7909f2bf TM	420	return ret;
	421	}
	422
	423	static int ocfs2_verify_group_and_input(struct inode *inode,
	424	struct ocfs2_dinode *di,
	425	struct ocfs2_new_group_input *input,
	426	struct buffer_head *group_bh)
	427	{
	428	u16 cl_count = le16_to_cpu(di->id2.i_chain.cl_count);
	429	u16 cl_cpg = le16_to_cpu(di->id2.i_chain.cl_cpg);
	430	u16 next_free = le16_to_cpu(di->id2.i_chain.cl_next_free_rec);
	431	u32 cluster = ocfs2_blocks_to_clusters(inode->i_sb, input->group);
	432	u32 total_clusters = le32_to_cpu(di->i_clusters);
	433	int ret = -EINVAL;
	434
	435	if (cluster < total_clusters)
	436	mlog(ML_ERROR, "add a group which is in the current volume.\n");
	437	else if (input->chain >= cl_count)
	438	mlog(ML_ERROR, "input chain exceeds the limit.\n");
	439	else if (next_free != cl_count && next_free != input->chain)
	440	mlog(ML_ERROR,
	441	"the add group should be in chain %u\n", next_free);
	442	else if (total_clusters + input->clusters < total_clusters)
	443	mlog(ML_ERROR, "add group's clusters overflow.\n");
	444	else if (input->clusters > cl_cpg)
	445	mlog(ML_ERROR, "the cluster exceeds the maximum of a group\n");
	446	else if (input->frees > input->clusters)
	447	mlog(ML_ERROR, "the free cluster exceeds the total clusters\n");
	448	else if (total_clusters % cl_cpg != 0)
	449	mlog(ML_ERROR,
	450	"the last group isn't full. Use group extend first.\n");
	451	else if (input->group != ocfs2_which_cluster_group(inode, cluster))
	452	mlog(ML_ERROR, "group blkno is invalid\n");
	453	else if ((ret = ocfs2_check_new_group(inode, di, input, group_bh)))
	454	mlog(ML_ERROR, "group descriptor check failed.\n");
	455	else
	456	ret = 0;
	457
	458	return ret;
	459	}
	460
	461	/* Add a new group descriptor to global_bitmap. */
	462	int ocfs2_group_add(struct inode inode, struct ocfs2_new_group_input input)
	463	{
	464	int ret;
	465	handle_t *handle;
	466	struct buffer_head *main_bm_bh = NULL;
	467	struct inode *main_bm_inode = NULL;
	468	struct ocfs2_dinode *fe = NULL;
	469	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
	470	struct buffer_head *group_bh = NULL;
	471	struct ocfs2_group_desc *group = NULL;
	472	struct ocfs2_chain_list *cl;
	473	struct ocfs2_chain_rec *cr;
	474	u16 cl_bpc;
eedd40e1	475	u64 bg_ptr;
7909f2bf	476
7909f2bf TM	477	if (ocfs2_is_hard_readonly(osb) \|\| ocfs2_is_soft_readonly(osb))
	478	return -EROFS;
	479
	480	main_bm_inode = ocfs2_get_system_file_inode(osb,
	481	GLOBAL_BITMAP_SYSTEM_INODE,
	482	OCFS2_INVALID_SLOT);
	483	if (!main_bm_inode) {
	484	ret = -EINVAL;
	485	mlog_errno(ret);
	486	goto out;
	487	}
	488
5955102c	489	inode_lock(main_bm_inode);
7909f2bf TM	490
	491	ret = ocfs2_inode_lock(main_bm_inode, &main_bm_bh, 1);
	492	if (ret < 0) {
	493	mlog_errno(ret);
	494	goto out_mutex;
	495	}
	496
	497	fe = (struct ocfs2_dinode *)main_bm_bh->b_data;
	498
	499	if (le16_to_cpu(fe->id2.i_chain.cl_cpg) !=
8571882c TM	500	ocfs2_group_bitmap_size(osb->sb, 0,
8571882c TM	501	osb->s_feature_incompat) * 8) {
7909f2bf TM	502	mlog(ML_ERROR, "The disk is too old and small."
	503	" Force to do offline resize.");
	504	ret = -EINVAL;
	505	goto out_unlock;
	506	}
	507
da1e9098	508	ret = ocfs2_read_blocks_sync(osb, input->group, 1, &group_bh);
7909f2bf TM	509	if (ret < 0) {
	510	mlog(ML_ERROR, "Can't read the group descriptor # %llu "
	511	"from the device.", (unsigned long long)input->group);
	512	goto out_unlock;
	513	}
	514
8cb471e8	515	ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), group_bh);
7909f2bf TM	516
	517	ret = ocfs2_verify_group_and_input(main_bm_inode, fe, input, group_bh);
	518	if (ret) {
	519	mlog_errno(ret);
8abaae8d	520	goto out_free_group_bh;
7909f2bf TM	521	}
7909f2bf TM	522
a5438702 TM	523	trace_ocfs2_group_add((unsigned long long)input->group,
a5438702 TM	524	input->chain, input->clusters, input->frees);
7909f2bf TM	525
	526	handle = ocfs2_start_trans(osb, OCFS2_GROUP_ADD_CREDITS);
	527	if (IS_ERR(handle)) {
	528	mlog_errno(PTR_ERR(handle));
	529	ret = -EINVAL;
8abaae8d	530	goto out_free_group_bh;
7909f2bf TM	531	}
	532
	533	cl_bpc = le16_to_cpu(fe->id2.i_chain.cl_bpc);
	534	cl = &fe->id2.i_chain;
	535	cr = &cl->cl_recs[input->chain];
	536
0cf2f763 JB	537	ret = ocfs2_journal_access_gd(handle, INODE_CACHE(main_bm_inode),
0cf2f763 JB	538	group_bh, OCFS2_JOURNAL_ACCESS_WRITE);
7909f2bf TM	539	if (ret < 0) {
	540	mlog_errno(ret);
	541	goto out_commit;
	542	}
	543
	544	group = (struct ocfs2_group_desc *)group_bh->b_data;
eedd40e1	545	bg_ptr = le64_to_cpu(group->bg_next_group);
7909f2bf	546	group->bg_next_group = cr->c_blkno;
ec20cec7	547	ocfs2_journal_dirty(handle, group_bh);
7909f2bf	548
0cf2f763 JB	549	ret = ocfs2_journal_access_di(handle, INODE_CACHE(main_bm_inode),
0cf2f763 JB	550	main_bm_bh, OCFS2_JOURNAL_ACCESS_WRITE);
7909f2bf	551	if (ret < 0) {
eedd40e1	552	group->bg_next_group = cpu_to_le64(bg_ptr);
7909f2bf TM	553	mlog_errno(ret);
	554	goto out_commit;
	555	}
	556
	557	if (input->chain == le16_to_cpu(cl->cl_next_free_rec)) {
	558	le16_add_cpu(&cl->cl_next_free_rec, 1);
	559	memset(cr, 0, sizeof(struct ocfs2_chain_rec));
	560	}
	561
4338ab6a	562	cr->c_blkno = cpu_to_le64(input->group);
7909f2bf TM	563	le32_add_cpu(&cr->c_total, input->clusters * cl_bpc);
	564	le32_add_cpu(&cr->c_free, input->frees * cl_bpc);
	565
	566	le32_add_cpu(&fe->id1.bitmap1.i_total, input->clusters *cl_bpc);
	567	le32_add_cpu(&fe->id1.bitmap1.i_used,
	568	(input->clusters - input->frees) * cl_bpc);
	569	le32_add_cpu(&fe->i_clusters, input->clusters);
	570
	571	ocfs2_journal_dirty(handle, main_bm_bh);
	572
	573	spin_lock(&OCFS2_I(main_bm_inode)->ip_lock);
	574	OCFS2_I(main_bm_inode)->ip_clusters = le32_to_cpu(fe->i_clusters);
17bf1418	575	le64_add_cpu(&fe->i_size, (u64)input->clusters << osb->s_clustersize_bits);
7909f2bf TM	576	spin_unlock(&OCFS2_I(main_bm_inode)->ip_lock);
	577	i_size_write(main_bm_inode, le64_to_cpu(fe->i_size));
	578
	579	ocfs2_update_super_and_backups(main_bm_inode, input->clusters);
	580
	581	out_commit:
	582	ocfs2_commit_trans(osb, handle);
8abaae8d YL	583
8abaae8d YL	584	out_free_group_bh:
2fe5c1d7	585	brelse(group_bh);
8abaae8d YL	586
8abaae8d YL	587	out_unlock:
2fe5c1d7	588	brelse(main_bm_bh);
7909f2bf TM	589
	590	ocfs2_inode_unlock(main_bm_inode, 1);
	591
	592	out_mutex:
5955102c	593	inode_unlock(main_bm_inode);
7909f2bf TM	594	iput(main_bm_inode);
	595
	596	out:
7909f2bf TM	597	return ret;
7909f2bf TM	598	}