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

/* -*- mode: c; c-basic-offset: 8; -*-
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * io.c
 *
 * Buffer cache handling
 *
 * Copyright (C) 2002, 2004 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 <linux/highmem.h>

#include <cluster/masklog.h>

#include "ocfs2.h"

#include "alloc.h"
#include "inode.h"
#include "journal.h"
#include "uptodate.h"
#include "buffer_head_io.h"
#include "ocfs2_trace.h"

/*
 * Bits on bh->b_state used by ocfs2.
 *
 * These MUST be after the JBD2 bits.  Hence, we use BH_JBDPrivateStart.
 */
enum ocfs2_state_bits {
	BH_NeedsValidate = BH_JBDPrivateStart,
};

/* Expand the magic b_state functions */
BUFFER_FNS(NeedsValidate, needs_validate);

int ocfs2_write_block(struct ocfs2_super *osb, struct buffer_head *bh,
		      struct ocfs2_caching_info *ci)
{
	int ret = 0;

	trace_ocfs2_write_block((unsigned long long)bh->b_blocknr, ci);

	BUG_ON(bh->b_blocknr < OCFS2_SUPER_BLOCK_BLKNO);
	BUG_ON(buffer_jbd(bh));

	/* No need to check for a soft readonly file system here. non
	 * journalled writes are only ever done on system files which
	 * can get modified during recovery even if read-only. */
	if (ocfs2_is_hard_readonly(osb)) {
		ret = -EROFS;
		mlog_errno(ret);
		goto out;
	}

	ocfs2_metadata_cache_io_lock(ci);

	lock_buffer(bh);
	set_buffer_uptodate(bh);

	/* remove from dirty list before I/O. */
	clear_buffer_dirty(bh);

	get_bh(bh); /* for end_buffer_write_sync() */
	bh->b_end_io = end_buffer_write_sync;
	submit_bh(WRITE, bh);

	wait_on_buffer(bh);

	if (buffer_uptodate(bh)) {
		ocfs2_set_buffer_uptodate(ci, bh);
	} else {
		/* We don't need to remove the clustered uptodate
		 * information for this bh as it's not marked locally
		 * uptodate. */
		ret = -EIO;
		put_bh(bh);
		mlog_errno(ret);
	}

	ocfs2_metadata_cache_io_unlock(ci);
out:
	return ret;
}

int ocfs2_read_blocks_sync(struct ocfs2_super *osb, u64 block,
			   unsigned int nr, struct buffer_head *bhs[])
{
	int status = 0;
	unsigned int i;
	struct buffer_head *bh;

	trace_ocfs2_read_blocks_sync((unsigned long long)block, nr);

	if (!nr)
		goto bail;

	for (i = 0 ; i < nr ; i++) {
		if (bhs[i] == NULL) {
			bhs[i] = sb_getblk(osb->sb, block++);
			if (bhs[i] == NULL) {
				status = -EIO;
				mlog_errno(status);
				goto bail;
			}
		}
		bh = bhs[i];

		if (buffer_jbd(bh)) {
			trace_ocfs2_read_blocks_sync_jbd(
					(unsigned long long)bh->b_blocknr);
			continue;
		}

		if (buffer_dirty(bh)) {
			/* This should probably be a BUG, or
			 * at least return an error. */
			mlog(ML_ERROR,
			     "trying to sync read a dirty "
			     "buffer! (blocknr = %llu), skipping\n",
			     (unsigned long long)bh->b_blocknr);
			continue;
		}

		lock_buffer(bh);
		if (buffer_jbd(bh)) {
			mlog(ML_ERROR,
			     "block %llu had the JBD bit set "
			     "while I was in lock_buffer!",
			     (unsigned long long)bh->b_blocknr);
			BUG();
		}

		clear_buffer_uptodate(bh);
		get_bh(bh); /* for end_buffer_read_sync() */
		bh->b_end_io = end_buffer_read_sync;
		submit_bh(READ, bh);
	}

	for (i = nr; i > 0; i--) {
		bh = bhs[i - 1];

		/* No need to wait on the buffer if it's managed by JBD. */
		if (!buffer_jbd(bh))
			wait_on_buffer(bh);

		if (!buffer_uptodate(bh)) {
			/* Status won't be cleared from here on out,
			 * so we can safely record this and loop back
			 * to cleanup the other buffers. */
			status = -EIO;
			put_bh(bh);
			bhs[i - 1] = NULL;
		}
	}

bail:
	return status;
}

int ocfs2_read_blocks(struct ocfs2_caching_info *ci, u64 block, int nr,
		      struct buffer_head *bhs[], int flags,
		      int (*validate)(struct super_block *sb,
				      struct buffer_head *bh))
{
	int status = 0;
	int i, ignore_cache = 0;
	struct buffer_head *bh;
	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);

	trace_ocfs2_read_blocks_begin(ci, (unsigned long long)block, nr, flags);

	BUG_ON(!ci);
	BUG_ON((flags & OCFS2_BH_READAHEAD) &&
	       (flags & OCFS2_BH_IGNORE_CACHE));

	if (bhs == NULL) {
		status = -EINVAL;
		mlog_errno(status);
		goto bail;
	}

	if (nr < 0) {
		mlog(ML_ERROR, "asked to read %d blocks!\n", nr);
		status = -EINVAL;
		mlog_errno(status);
		goto bail;
	}

	if (nr == 0) {
		status = 0;
		goto bail;
	}

	ocfs2_metadata_cache_io_lock(ci);
	for (i = 0 ; i < nr ; i++) {
		if (bhs[i] == NULL) {
			bhs[i] = sb_getblk(sb, block++);
			if (bhs[i] == NULL) {
				ocfs2_metadata_cache_io_unlock(ci);
				status = -EIO;
				mlog_errno(status);
				goto bail;
			}
		}
		bh = bhs[i];
		ignore_cache = (flags & OCFS2_BH_IGNORE_CACHE);

		/* There are three read-ahead cases here which we need to
		 * be concerned with. All three assume a buffer has
		 * previously been submitted with OCFS2_BH_READAHEAD
		 * and it hasn't yet completed I/O.
		 *
		 * 1) The current request is sync to disk. This rarely
		 *    happens these days, and never when performance
		 *    matters - the code can just wait on the buffer
		 *    lock and re-submit.
		 *
		 * 2) The current request is cached, but not
		 *    readahead. ocfs2_buffer_uptodate() will return
		 *    false anyway, so we'll wind up waiting on the
		 *    buffer lock to do I/O. We re-check the request
		 *    with after getting the lock to avoid a re-submit.
		 *
		 * 3) The current request is readahead (and so must
		 *    also be a caching one). We short circuit if the
		 *    buffer is locked (under I/O) and if it's in the
		 *    uptodate cache. The re-check from #2 catches the
		 *    case that the previous read-ahead completes just
		 *    before our is-it-in-flight check.
		 */

		if (!ignore_cache && !ocfs2_buffer_uptodate(ci, bh)) {
			trace_ocfs2_read_blocks_from_disk(
			     (unsigned long long)bh->b_blocknr,
			     (unsigned long long)ocfs2_metadata_cache_owner(ci));
			/* We're using ignore_cache here to say
			 * "go to disk" */
			ignore_cache = 1;
		}

		trace_ocfs2_read_blocks_bh((unsigned long long)bh->b_blocknr,
			ignore_cache, buffer_jbd(bh), buffer_dirty(bh));

		if (buffer_jbd(bh)) {
			continue;
		}

		if (ignore_cache) {
			if (buffer_dirty(bh)) {
				/* This should probably be a BUG, or
				 * at least return an error. */
				continue;
			}

			/* A read-ahead request was made - if the
			 * buffer is already under read-ahead from a
			 * previously submitted request than we are
			 * done here. */
			if ((flags & OCFS2_BH_READAHEAD)
			    && ocfs2_buffer_read_ahead(ci, bh))
				continue;

			lock_buffer(bh);
			if (buffer_jbd(bh)) {
#ifdef CATCH_BH_JBD_RACES
				mlog(ML_ERROR, "block %llu had the JBD bit set "
					       "while I was in lock_buffer!",
				     (unsigned long long)bh->b_blocknr);
				BUG();
#else
				unlock_buffer(bh);
				continue;
#endif
			}

			/* Re-check ocfs2_buffer_uptodate() as a
			 * previously read-ahead buffer may have
			 * completed I/O while we were waiting for the
			 * buffer lock. */
			if (!(flags & OCFS2_BH_IGNORE_CACHE)
			    && !(flags & OCFS2_BH_READAHEAD)
			    && ocfs2_buffer_uptodate(ci, bh)) {
				unlock_buffer(bh);
				continue;
			}

			clear_buffer_uptodate(bh);
			get_bh(bh); /* for end_buffer_read_sync() */
			if (validate)
				set_buffer_needs_validate(bh);
			bh->b_end_io = end_buffer_read_sync;
			submit_bh(READ, bh);
			continue;
		}
	}

	status = 0;

	for (i = (nr - 1); i >= 0; i--) {
		bh = bhs[i];

		if (!(flags & OCFS2_BH_READAHEAD)) {
			/* We know this can't have changed as we hold the
			 * owner sem. Avoid doing any work on the bh if the
			 * journal has it. */
			if (!buffer_jbd(bh))
				wait_on_buffer(bh);

			if (!buffer_uptodate(bh)) {
				/* Status won't be cleared from here on out,
				 * so we can safely record this and loop back
				 * to cleanup the other buffers. Don't need to
				 * remove the clustered uptodate information
				 * for this bh as it's not marked locally
				 * uptodate. */
				status = -EIO;
				put_bh(bh);
				bhs[i] = NULL;
				continue;
			}

			if (buffer_needs_validate(bh)) {
				/* We never set NeedsValidate if the
				 * buffer was held by the journal, so
				 * that better not have changed */
				BUG_ON(buffer_jbd(bh));
				clear_buffer_needs_validate(bh);
				status = validate(sb, bh);
				if (status) {
					put_bh(bh);
					bhs[i] = NULL;
					continue;
				}
			}
		}

		/* Always set the buffer in the cache, even if it was
		 * a forced read, or read-ahead which hasn't yet
		 * completed. */
		ocfs2_set_buffer_uptodate(ci, bh);
	}
	ocfs2_metadata_cache_io_unlock(ci);

	trace_ocfs2_read_blocks_end((unsigned long long)block, nr,
				    flags, ignore_cache);

bail:

	return status;
}

/* Check whether the blkno is the super block or one of the backups. */
static void ocfs2_check_super_or_backup(struct super_block *sb,
					sector_t blkno)
{
	int i;
	u64 backup_blkno;

	if (blkno == OCFS2_SUPER_BLOCK_BLKNO)
		return;

	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
		backup_blkno = ocfs2_backup_super_blkno(sb, i);
		if (backup_blkno == blkno)
			return;
	}

	BUG();
}

/*
 * Write super block and backups doesn't need to collaborate with journal,
 * so we don't need to lock ip_io_mutex and ci doesn't need to bea passed
 * into this function.
 */
int ocfs2_write_super_or_backup(struct ocfs2_super *osb,
				struct buffer_head *bh)
{
	int ret = 0;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)bh->b_data;

	BUG_ON(buffer_jbd(bh));
	ocfs2_check_super_or_backup(osb->sb, bh->b_blocknr);

	if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb)) {
		ret = -EROFS;
		mlog_errno(ret);
		goto out;
	}

	lock_buffer(bh);
	set_buffer_uptodate(bh);

	/* remove from dirty list before I/O. */
	clear_buffer_dirty(bh);

	get_bh(bh); /* for end_buffer_write_sync() */
	bh->b_end_io = end_buffer_write_sync;
	ocfs2_compute_meta_ecc(osb->sb, bh->b_data, &di->i_check);
	submit_bh(WRITE, bh);

	wait_on_buffer(bh);

	if (!buffer_uptodate(bh)) {
		ret = -EIO;
		put_bh(bh);
		mlog_errno(ret);
	}

out:
	return ret;
}
Commit	Line	Data
ccd979bd MF	1	/* -- mode: c; c-basic-offset: 8; --
	2	* vim: noexpandtab sw=8 ts=8 sts=0:
	3	*
	4	* io.c
	5	*
	6	* Buffer cache handling
	7	*
	8	* Copyright (C) 2002, 2004 Oracle. All rights reserved.
	9	*
	10	* This program is free software; you can redistribute it and/or
	11	* modify it under the terms of the GNU General Public
	12	* License as published by the Free Software Foundation; either
	13	* version 2 of the License, or (at your option) any later version.
	14	*
	15	* This program is distributed in the hope that it will be useful,
	16	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	17	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	18	* General Public License for more details.
	19	*
	20	* You should have received a copy of the GNU General Public
	21	* License along with this program; if not, write to the
	22	* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
	23	* Boston, MA 021110-1307, USA.
	24	*/
	25
	26	#include <linux/fs.h>
	27	#include <linux/types.h>
ccd979bd MF	28	#include <linux/highmem.h>
	29
	30	#include <cluster/masklog.h>
	31
	32	#include "ocfs2.h"
	33
	34	#include "alloc.h"
	35	#include "inode.h"
	36	#include "journal.h"
	37	#include "uptodate.h"
ccd979bd	38	#include "buffer_head_io.h"
15057e98	39	#include "ocfs2_trace.h"
ccd979bd	40
970e4936 JB	41	/*
	42	* Bits on bh->b_state used by ocfs2.
	43	*
b86c86fa	44	* These MUST be after the JBD2 bits. Hence, we use BH_JBDPrivateStart.
970e4936 JB	45	*/
970e4936 JB	46	enum ocfs2_state_bits {
b86c86fa	47	BH_NeedsValidate = BH_JBDPrivateStart,
970e4936 JB	48	};
	49
	50	/* Expand the magic b_state functions */
	51	BUFFER_FNS(NeedsValidate, needs_validate);
	52
ccd979bd	53	int ocfs2_write_block(struct ocfs2_super osb, struct buffer_head bh,
8cb471e8	54	struct ocfs2_caching_info *ci)
ccd979bd MF	55	{
	56	int ret = 0;
	57
15057e98	58	trace_ocfs2_write_block((unsigned long long)bh->b_blocknr, ci);
ccd979bd MF	59
	60	BUG_ON(bh->b_blocknr < OCFS2_SUPER_BLOCK_BLKNO);
	61	BUG_ON(buffer_jbd(bh));
	62
	63	/* No need to check for a soft readonly file system here. non
	64	* journalled writes are only ever done on system files which
	65	* can get modified during recovery even if read-only. */
	66	if (ocfs2_is_hard_readonly(osb)) {
	67	ret = -EROFS;
c1e8d35e	68	mlog_errno(ret);
ccd979bd MF	69	goto out;
	70	}
	71
8cb471e8	72	ocfs2_metadata_cache_io_lock(ci);
ccd979bd MF	73
	74	lock_buffer(bh);
	75	set_buffer_uptodate(bh);
	76
	77	/* remove from dirty list before I/O. */
	78	clear_buffer_dirty(bh);
	79
da1e9098	80	get_bh(bh); /* for end_buffer_write_sync() */
ccd979bd MF	81	bh->b_end_io = end_buffer_write_sync;
	82	submit_bh(WRITE, bh);
	83
	84	wait_on_buffer(bh);
	85
	86	if (buffer_uptodate(bh)) {
8cb471e8	87	ocfs2_set_buffer_uptodate(ci, bh);
ccd979bd MF	88	} else {
	89	/* We don't need to remove the clustered uptodate
	90	* information for this bh as it's not marked locally
	91	* uptodate. */
	92	ret = -EIO;
2fe5c1d7	93	put_bh(bh);
c1e8d35e	94	mlog_errno(ret);
ccd979bd MF	95	}
ccd979bd MF	96
8cb471e8	97	ocfs2_metadata_cache_io_unlock(ci);
ccd979bd	98	out:
ccd979bd MF	99	return ret;
	100	}
	101
da1e9098 JB	102	int ocfs2_read_blocks_sync(struct ocfs2_super *osb, u64 block,
	103	unsigned int nr, struct buffer_head *bhs[])
	104	{
	105	int status = 0;
	106	unsigned int i;
	107	struct buffer_head *bh;
	108
15057e98 TM	109	trace_ocfs2_read_blocks_sync((unsigned long long)block, nr);
	110
	111	if (!nr)
da1e9098	112	goto bail;
da1e9098 JB	113
	114	for (i = 0 ; i < nr ; i++) {
	115	if (bhs[i] == NULL) {
	116	bhs[i] = sb_getblk(osb->sb, block++);
	117	if (bhs[i] == NULL) {
	118	status = -EIO;
	119	mlog_errno(status);
	120	goto bail;
	121	}
	122	}
	123	bh = bhs[i];
	124
	125	if (buffer_jbd(bh)) {
15057e98 TM	126	trace_ocfs2_read_blocks_sync_jbd(
15057e98 TM	127	(unsigned long long)bh->b_blocknr);
da1e9098 JB	128	continue;
	129	}
	130
	131	if (buffer_dirty(bh)) {
	132	/* This should probably be a BUG, or
	133	* at least return an error. */
	134	mlog(ML_ERROR,
	135	"trying to sync read a dirty "
	136	"buffer! (blocknr = %llu), skipping\n",
	137	(unsigned long long)bh->b_blocknr);
	138	continue;
	139	}
	140
	141	lock_buffer(bh);
	142	if (buffer_jbd(bh)) {
	143	mlog(ML_ERROR,
	144	"block %llu had the JBD bit set "
	145	"while I was in lock_buffer!",
	146	(unsigned long long)bh->b_blocknr);
	147	BUG();
	148	}
	149
	150	clear_buffer_uptodate(bh);
	151	get_bh(bh); /* for end_buffer_read_sync() */
	152	bh->b_end_io = end_buffer_read_sync;
	153	submit_bh(READ, bh);
	154	}
	155
	156	for (i = nr; i > 0; i--) {
	157	bh = bhs[i - 1];
	158
d6b58f89 MF	159	/* No need to wait on the buffer if it's managed by JBD. */
	160	if (!buffer_jbd(bh))
	161	wait_on_buffer(bh);
da1e9098	162
da1e9098 JB	163	if (!buffer_uptodate(bh)) {
	164	/* Status won't be cleared from here on out,
	165	* so we can safely record this and loop back
	166	* to cleanup the other buffers. */
	167	status = -EIO;
	168	put_bh(bh);
	169	bhs[i - 1] = NULL;
	170	}
	171	}
	172
	173	bail:
	174	return status;
	175	}
	176
8cb471e8	177	int ocfs2_read_blocks(struct ocfs2_caching_info *ci, u64 block, int nr,
970e4936 JB	178	struct buffer_head *bhs[], int flags,
	179	int (validate)(struct super_block sb,
	180	struct buffer_head *bh))
ccd979bd MF	181	{
ccd979bd MF	182	int status = 0;
ccd979bd MF	183	int i, ignore_cache = 0;
ccd979bd MF	184	struct buffer_head *bh;
8cb471e8	185	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
ccd979bd	186
15057e98	187	trace_ocfs2_read_blocks_begin(ci, (unsigned long long)block, nr, flags);
ccd979bd	188
8cb471e8	189	BUG_ON(!ci);
d4a8c93c JB	190	BUG_ON((flags & OCFS2_BH_READAHEAD) &&
d4a8c93c JB	191	(flags & OCFS2_BH_IGNORE_CACHE));
aa958874	192
31d33073	193	if (bhs == NULL) {
ccd979bd MF	194	status = -EINVAL;
	195	mlog_errno(status);
	196	goto bail;
	197	}
	198
	199	if (nr < 0) {
	200	mlog(ML_ERROR, "asked to read %d blocks!\n", nr);
	201	status = -EINVAL;
	202	mlog_errno(status);
	203	goto bail;
	204	}
	205
	206	if (nr == 0) {
ccd979bd MF	207	status = 0;
	208	goto bail;
	209	}
	210
8cb471e8	211	ocfs2_metadata_cache_io_lock(ci);
ccd979bd MF	212	for (i = 0 ; i < nr ; i++) {
ccd979bd MF	213	if (bhs[i] == NULL) {
8cb471e8	214	bhs[i] = sb_getblk(sb, block++);
ccd979bd	215	if (bhs[i] == NULL) {
8cb471e8	216	ocfs2_metadata_cache_io_unlock(ci);
ccd979bd MF	217	status = -EIO;
	218	mlog_errno(status);
	219	goto bail;
	220	}
	221	}
	222	bh = bhs[i];
d4a8c93c	223	ignore_cache = (flags & OCFS2_BH_IGNORE_CACHE);
ccd979bd	224
aa958874 MF	225	/* There are three read-ahead cases here which we need to
	226	* be concerned with. All three assume a buffer has
	227	* previously been submitted with OCFS2_BH_READAHEAD
	228	* and it hasn't yet completed I/O.
	229	*
	230	* 1) The current request is sync to disk. This rarely
	231	* happens these days, and never when performance
	232	* matters - the code can just wait on the buffer
	233	* lock and re-submit.
	234	*
	235	* 2) The current request is cached, but not
	236	* readahead. ocfs2_buffer_uptodate() will return
	237	* false anyway, so we'll wind up waiting on the
	238	* buffer lock to do I/O. We re-check the request
	239	* with after getting the lock to avoid a re-submit.
	240	*
	241	* 3) The current request is readahead (and so must
	242	* also be a caching one). We short circuit if the
	243	* buffer is locked (under I/O) and if it's in the
	244	* uptodate cache. The re-check from #2 catches the
	245	* case that the previous read-ahead completes just
	246	* before our is-it-in-flight check.
	247	*/
	248
8cb471e8	249	if (!ignore_cache && !ocfs2_buffer_uptodate(ci, bh)) {
d701485a	250	trace_ocfs2_read_blocks_from_disk(
ccd979bd	251	(unsigned long long)bh->b_blocknr,
8cb471e8	252	(unsigned long long)ocfs2_metadata_cache_owner(ci));
d4a8c93c JB	253	/* We're using ignore_cache here to say
d4a8c93c JB	254	* "go to disk" */
ccd979bd MF	255	ignore_cache = 1;
	256	}
	257
15057e98 TM	258	trace_ocfs2_read_blocks_bh((unsigned long long)bh->b_blocknr,
	259	ignore_cache, buffer_jbd(bh), buffer_dirty(bh));
	260
ccd979bd	261	if (buffer_jbd(bh)) {
ccd979bd MF	262	continue;
	263	}
	264
d4a8c93c	265	if (ignore_cache) {
ccd979bd MF	266	if (buffer_dirty(bh)) {
	267	/* This should probably be a BUG, or
	268	* at least return an error. */
ccd979bd MF	269	continue;
	270	}
	271
aa958874 MF	272	/* A read-ahead request was made - if the
	273	* buffer is already under read-ahead from a
	274	* previously submitted request than we are
	275	* done here. */
	276	if ((flags & OCFS2_BH_READAHEAD)
8cb471e8	277	&& ocfs2_buffer_read_ahead(ci, bh))
aa958874 MF	278	continue;
aa958874 MF	279
ccd979bd MF	280	lock_buffer(bh);
	281	if (buffer_jbd(bh)) {
	282	#ifdef CATCH_BH_JBD_RACES
	283	mlog(ML_ERROR, "block %llu had the JBD bit set "
	284	"while I was in lock_buffer!",
	285	(unsigned long long)bh->b_blocknr);
	286	BUG();
	287	#else
	288	unlock_buffer(bh);
	289	continue;
	290	#endif
	291	}
aa958874 MF	292
	293	/* Re-check ocfs2_buffer_uptodate() as a
	294	* previously read-ahead buffer may have
	295	* completed I/O while we were waiting for the
	296	* buffer lock. */
d4a8c93c	297	if (!(flags & OCFS2_BH_IGNORE_CACHE)
aa958874	298	&& !(flags & OCFS2_BH_READAHEAD)
8cb471e8	299	&& ocfs2_buffer_uptodate(ci, bh)) {
aa958874 MF	300	unlock_buffer(bh);
	301	continue;
	302	}
	303
ccd979bd MF	304	clear_buffer_uptodate(bh);
ccd979bd MF	305	get_bh(bh); /* for end_buffer_read_sync() */
970e4936 JB	306	if (validate)
970e4936 JB	307	set_buffer_needs_validate(bh);
ccd979bd	308	bh->b_end_io = end_buffer_read_sync;
aa958874	309	submit_bh(READ, bh);
ccd979bd MF	310	continue;
	311	}
	312	}
	313
	314	status = 0;
	315
	316	for (i = (nr - 1); i >= 0; i--) {
	317	bh = bhs[i];
	318
aa958874 MF	319	if (!(flags & OCFS2_BH_READAHEAD)) {
aa958874 MF	320	/* We know this can't have changed as we hold the
8cb471e8	321	* owner sem. Avoid doing any work on the bh if the
aa958874 MF	322	* journal has it. */
	323	if (!buffer_jbd(bh))
	324	wait_on_buffer(bh);
	325
	326	if (!buffer_uptodate(bh)) {
	327	/* Status won't be cleared from here on out,
	328	* so we can safely record this and loop back
	329	* to cleanup the other buffers. Don't need to
	330	* remove the clustered uptodate information
	331	* for this bh as it's not marked locally
	332	* uptodate. */
	333	status = -EIO;
2fe5c1d7	334	put_bh(bh);
aa958874 MF	335	bhs[i] = NULL;
	336	continue;
	337	}
970e4936 JB	338
	339	if (buffer_needs_validate(bh)) {
	340	/* We never set NeedsValidate if the
	341	* buffer was held by the journal, so
	342	* that better not have changed */
	343	BUG_ON(buffer_jbd(bh));
	344	clear_buffer_needs_validate(bh);
8cb471e8	345	status = validate(sb, bh);
970e4936 JB	346	if (status) {
	347	put_bh(bh);
	348	bhs[i] = NULL;
	349	continue;
	350	}
	351	}
ccd979bd MF	352	}
ccd979bd MF	353
aa958874 MF	354	/* Always set the buffer in the cache, even if it was
	355	* a forced read, or read-ahead which hasn't yet
	356	* completed. */
8cb471e8	357	ocfs2_set_buffer_uptodate(ci, bh);
ccd979bd	358	}
8cb471e8	359	ocfs2_metadata_cache_io_unlock(ci);
ccd979bd	360
15057e98 TM	361	trace_ocfs2_read_blocks_end((unsigned long long)block, nr,
15057e98 TM	362	flags, ignore_cache);
ccd979bd MF	363
	364	bail:
	365
ccd979bd MF	366	return status;
ccd979bd MF	367	}
d659072f TM	368
	369	/* Check whether the blkno is the super block or one of the backups. */
	370	static void ocfs2_check_super_or_backup(struct super_block *sb,
	371	sector_t blkno)
	372	{
	373	int i;
	374	u64 backup_blkno;
	375
	376	if (blkno == OCFS2_SUPER_BLOCK_BLKNO)
	377	return;
	378
	379	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
	380	backup_blkno = ocfs2_backup_super_blkno(sb, i);
	381	if (backup_blkno == blkno)
	382	return;
	383	}
	384
	385	BUG();
	386	}
	387
	388	/*
	389	* Write super block and backups doesn't need to collaborate with journal,
8cb471e8	390	* so we don't need to lock ip_io_mutex and ci doesn't need to bea passed
d659072f TM	391	* into this function.
	392	*/
	393	int ocfs2_write_super_or_backup(struct ocfs2_super *osb,
	394	struct buffer_head *bh)
	395	{
	396	int ret = 0;
a42ab8e1	397	struct ocfs2_dinode di = (struct ocfs2_dinode )bh->b_data;
d659072f	398
d659072f TM	399	BUG_ON(buffer_jbd(bh));
	400	ocfs2_check_super_or_backup(osb->sb, bh->b_blocknr);
	401
	402	if (ocfs2_is_hard_readonly(osb) \|\| ocfs2_is_soft_readonly(osb)) {
	403	ret = -EROFS;
c1e8d35e	404	mlog_errno(ret);
d659072f TM	405	goto out;
	406	}
	407
	408	lock_buffer(bh);
	409	set_buffer_uptodate(bh);
	410
	411	/* remove from dirty list before I/O. */
	412	clear_buffer_dirty(bh);
	413
	414	get_bh(bh); /* for end_buffer_write_sync() */
	415	bh->b_end_io = end_buffer_write_sync;
a42ab8e1	416	ocfs2_compute_meta_ecc(osb->sb, bh->b_data, &di->i_check);
d659072f TM	417	submit_bh(WRITE, bh);
	418
	419	wait_on_buffer(bh);
	420
	421	if (!buffer_uptodate(bh)) {
	422	ret = -EIO;
2fe5c1d7	423	put_bh(bh);
c1e8d35e	424	mlog_errno(ret);
d659072f TM	425	}
	426
	427	out:
d659072f TM	428	return ret;
d659072f TM	429	}