[GitHub/mt8127/android_kernel_alcatel_ttab.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"

/*
 * 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;

	mlog_entry("(bh->b_blocknr = %llu, ci=%p)\n",
		   (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;
		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);
	}

	ocfs2_metadata_cache_io_unlock(ci);
out:
	mlog_exit(ret);
	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;

	if (!nr) {
		mlog(ML_BH_IO, "No buffers will be read!\n");
		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)) {
			mlog(ML_BH_IO,
			     "trying to sync read a jbd "
			     "managed bh (blocknr = %llu), skipping\n",
			     (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);

	mlog_entry("(ci=%p, block=(%llu), nr=(%d), flags=%d)\n",
		   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) {
		mlog(ML_BH_IO, "No buffers will be read!\n");
		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)) {
			mlog(ML_UPTODATE,
			     "bh (%llu), owner %llu not uptodate\n",
			     (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;
		}

		if (buffer_jbd(bh)) {
			if (ignore_cache)
				mlog(ML_BH_IO, "trying to sync read a jbd "
					       "managed bh (blocknr = %llu)\n",
				     (unsigned long long)bh->b_blocknr);
			continue;
		}

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

	mlog(ML_BH_IO, "block=(%llu), nr=(%d), cached=%s, flags=0x%x\n",
	     (unsigned long long)block, nr,
	     ((flags & OCFS2_BH_IGNORE_CACHE) || ignore_cache) ? "no" : "yes",
	     flags);

bail:

	mlog_exit(status);
	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;

	mlog_entry_void();

	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;
		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;
	submit_bh(WRITE, bh);

	wait_on_buffer(bh);

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

out:
	mlog_exit(ret);
	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"
	38
	39	#include "buffer_head_io.h"
	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
8cb471e8 JB	58	mlog_entry("(bh->b_blocknr = %llu, ci=%p)\n",
8cb471e8 JB	59	(unsigned long long)bh->b_blocknr, ci);
ccd979bd MF	60
	61	BUG_ON(bh->b_blocknr < OCFS2_SUPER_BLOCK_BLKNO);
	62	BUG_ON(buffer_jbd(bh));
	63
	64	/* No need to check for a soft readonly file system here. non
	65	* journalled writes are only ever done on system files which
	66	* can get modified during recovery even if read-only. */
	67	if (ocfs2_is_hard_readonly(osb)) {
	68	ret = -EROFS;
	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);
ccd979bd MF	94	}
ccd979bd MF	95
8cb471e8	96	ocfs2_metadata_cache_io_unlock(ci);
ccd979bd MF	97	out:
	98	mlog_exit(ret);
	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
	109	if (!nr) {
	110	mlog(ML_BH_IO, "No buffers will be read!\n");
	111	goto bail;
	112	}
	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)) {
d6b58f89	126	mlog(ML_BH_IO,
da1e9098 JB	127	"trying to sync read a jbd "
	128	"managed bh (blocknr = %llu), skipping\n",
	129	(unsigned long long)bh->b_blocknr);
	130	continue;
	131	}
	132
	133	if (buffer_dirty(bh)) {
	134	/* This should probably be a BUG, or
	135	* at least return an error. */
	136	mlog(ML_ERROR,
	137	"trying to sync read a dirty "
	138	"buffer! (blocknr = %llu), skipping\n",
	139	(unsigned long long)bh->b_blocknr);
	140	continue;
	141	}
	142
	143	lock_buffer(bh);
	144	if (buffer_jbd(bh)) {
	145	mlog(ML_ERROR,
	146	"block %llu had the JBD bit set "
	147	"while I was in lock_buffer!",
	148	(unsigned long long)bh->b_blocknr);
	149	BUG();
	150	}
	151
	152	clear_buffer_uptodate(bh);
	153	get_bh(bh); /* for end_buffer_read_sync() */
	154	bh->b_end_io = end_buffer_read_sync;
	155	submit_bh(READ, bh);
	156	}
	157
	158	for (i = nr; i > 0; i--) {
	159	bh = bhs[i - 1];
	160
d6b58f89 MF	161	/* No need to wait on the buffer if it's managed by JBD. */
	162	if (!buffer_jbd(bh))
	163	wait_on_buffer(bh);
da1e9098	164
da1e9098 JB	165	if (!buffer_uptodate(bh)) {
	166	/* Status won't be cleared from here on out,
	167	* so we can safely record this and loop back
	168	* to cleanup the other buffers. */
	169	status = -EIO;
	170	put_bh(bh);
	171	bhs[i - 1] = NULL;
	172	}
	173	}
	174
	175	bail:
	176	return status;
	177	}
	178
8cb471e8	179	int ocfs2_read_blocks(struct ocfs2_caching_info *ci, u64 block, int nr,
970e4936 JB	180	struct buffer_head *bhs[], int flags,
	181	int (validate)(struct super_block sb,
	182	struct buffer_head *bh))
ccd979bd MF	183	{
ccd979bd MF	184	int status = 0;
ccd979bd MF	185	int i, ignore_cache = 0;
ccd979bd MF	186	struct buffer_head *bh;
8cb471e8	187	struct super_block *sb = ocfs2_metadata_cache_get_super(ci);
ccd979bd	188
8cb471e8 JB	189	mlog_entry("(ci=%p, block=(%llu), nr=(%d), flags=%d)\n",
8cb471e8 JB	190	ci, (unsigned long long)block, nr, flags);
ccd979bd	191
8cb471e8	192	BUG_ON(!ci);
d4a8c93c JB	193	BUG_ON((flags & OCFS2_BH_READAHEAD) &&
d4a8c93c JB	194	(flags & OCFS2_BH_IGNORE_CACHE));
aa958874	195
31d33073	196	if (bhs == NULL) {
ccd979bd MF	197	status = -EINVAL;
	198	mlog_errno(status);
	199	goto bail;
	200	}
	201
	202	if (nr < 0) {
	203	mlog(ML_ERROR, "asked to read %d blocks!\n", nr);
	204	status = -EINVAL;
	205	mlog_errno(status);
	206	goto bail;
	207	}
	208
	209	if (nr == 0) {
	210	mlog(ML_BH_IO, "No buffers will be read!\n");
	211	status = 0;
	212	goto bail;
	213	}
	214
8cb471e8	215	ocfs2_metadata_cache_io_lock(ci);
ccd979bd MF	216	for (i = 0 ; i < nr ; i++) {
ccd979bd MF	217	if (bhs[i] == NULL) {
8cb471e8	218	bhs[i] = sb_getblk(sb, block++);
ccd979bd	219	if (bhs[i] == NULL) {
8cb471e8	220	ocfs2_metadata_cache_io_unlock(ci);
ccd979bd MF	221	status = -EIO;
	222	mlog_errno(status);
	223	goto bail;
	224	}
	225	}
	226	bh = bhs[i];
d4a8c93c	227	ignore_cache = (flags & OCFS2_BH_IGNORE_CACHE);
ccd979bd	228
aa958874 MF	229	/* There are three read-ahead cases here which we need to
	230	* be concerned with. All three assume a buffer has
	231	* previously been submitted with OCFS2_BH_READAHEAD
	232	* and it hasn't yet completed I/O.
	233	*
	234	* 1) The current request is sync to disk. This rarely
	235	* happens these days, and never when performance
	236	* matters - the code can just wait on the buffer
	237	* lock and re-submit.
	238	*
	239	* 2) The current request is cached, but not
	240	* readahead. ocfs2_buffer_uptodate() will return
	241	* false anyway, so we'll wind up waiting on the
	242	* buffer lock to do I/O. We re-check the request
	243	* with after getting the lock to avoid a re-submit.
	244	*
	245	* 3) The current request is readahead (and so must
	246	* also be a caching one). We short circuit if the
	247	* buffer is locked (under I/O) and if it's in the
	248	* uptodate cache. The re-check from #2 catches the
	249	* case that the previous read-ahead completes just
	250	* before our is-it-in-flight check.
	251	*/
	252
8cb471e8	253	if (!ignore_cache && !ocfs2_buffer_uptodate(ci, bh)) {
ccd979bd	254	mlog(ML_UPTODATE,
8cb471e8	255	"bh (%llu), owner %llu not uptodate\n",
ccd979bd	256	(unsigned long long)bh->b_blocknr,
8cb471e8	257	(unsigned long long)ocfs2_metadata_cache_owner(ci));
d4a8c93c JB	258	/* We're using ignore_cache here to say
d4a8c93c JB	259	* "go to disk" */
ccd979bd MF	260	ignore_cache = 1;
	261	}
	262
ccd979bd	263	if (buffer_jbd(bh)) {
d4a8c93c	264	if (ignore_cache)
ccd979bd MF	265	mlog(ML_BH_IO, "trying to sync read a jbd "
	266	"managed bh (blocknr = %llu)\n",
	267	(unsigned long long)bh->b_blocknr);
	268	continue;
	269	}
	270
d4a8c93c	271	if (ignore_cache) {
ccd979bd MF	272	if (buffer_dirty(bh)) {
	273	/* This should probably be a BUG, or
	274	* at least return an error. */
	275	mlog(ML_BH_IO, "asking me to sync read a dirty "
	276	"buffer! (blocknr = %llu)\n",
	277	(unsigned long long)bh->b_blocknr);
	278	continue;
	279	}
	280
aa958874 MF	281	/* A read-ahead request was made - if the
	282	* buffer is already under read-ahead from a
	283	* previously submitted request than we are
	284	* done here. */
	285	if ((flags & OCFS2_BH_READAHEAD)
8cb471e8	286	&& ocfs2_buffer_read_ahead(ci, bh))
aa958874 MF	287	continue;
aa958874 MF	288
ccd979bd MF	289	lock_buffer(bh);
	290	if (buffer_jbd(bh)) {
	291	#ifdef CATCH_BH_JBD_RACES
	292	mlog(ML_ERROR, "block %llu had the JBD bit set "
	293	"while I was in lock_buffer!",
	294	(unsigned long long)bh->b_blocknr);
	295	BUG();
	296	#else
	297	unlock_buffer(bh);
	298	continue;
	299	#endif
	300	}
aa958874 MF	301
	302	/* Re-check ocfs2_buffer_uptodate() as a
	303	* previously read-ahead buffer may have
	304	* completed I/O while we were waiting for the
	305	* buffer lock. */
d4a8c93c	306	if (!(flags & OCFS2_BH_IGNORE_CACHE)
aa958874	307	&& !(flags & OCFS2_BH_READAHEAD)
8cb471e8	308	&& ocfs2_buffer_uptodate(ci, bh)) {
aa958874 MF	309	unlock_buffer(bh);
	310	continue;
	311	}
	312
ccd979bd MF	313	clear_buffer_uptodate(bh);
ccd979bd MF	314	get_bh(bh); /* for end_buffer_read_sync() */
970e4936 JB	315	if (validate)
970e4936 JB	316	set_buffer_needs_validate(bh);
ccd979bd	317	bh->b_end_io = end_buffer_read_sync;
aa958874	318	submit_bh(READ, bh);
ccd979bd MF	319	continue;
	320	}
	321	}
	322
	323	status = 0;
	324
	325	for (i = (nr - 1); i >= 0; i--) {
	326	bh = bhs[i];
	327
aa958874 MF	328	if (!(flags & OCFS2_BH_READAHEAD)) {
aa958874 MF	329	/* We know this can't have changed as we hold the
8cb471e8	330	* owner sem. Avoid doing any work on the bh if the
aa958874 MF	331	* journal has it. */
	332	if (!buffer_jbd(bh))
	333	wait_on_buffer(bh);
	334
	335	if (!buffer_uptodate(bh)) {
	336	/* Status won't be cleared from here on out,
	337	* so we can safely record this and loop back
	338	* to cleanup the other buffers. Don't need to
	339	* remove the clustered uptodate information
	340	* for this bh as it's not marked locally
	341	* uptodate. */
	342	status = -EIO;
2fe5c1d7	343	put_bh(bh);
aa958874 MF	344	bhs[i] = NULL;
	345	continue;
	346	}
970e4936 JB	347
	348	if (buffer_needs_validate(bh)) {
	349	/* We never set NeedsValidate if the
	350	* buffer was held by the journal, so
	351	* that better not have changed */
	352	BUG_ON(buffer_jbd(bh));
	353	clear_buffer_needs_validate(bh);
8cb471e8	354	status = validate(sb, bh);
970e4936 JB	355	if (status) {
	356	put_bh(bh);
	357	bhs[i] = NULL;
	358	continue;
	359	}
	360	}
ccd979bd MF	361	}
ccd979bd MF	362
aa958874 MF	363	/* Always set the buffer in the cache, even if it was
	364	* a forced read, or read-ahead which hasn't yet
	365	* completed. */
8cb471e8	366	ocfs2_set_buffer_uptodate(ci, bh);
ccd979bd	367	}
8cb471e8	368	ocfs2_metadata_cache_io_unlock(ci);
ccd979bd	369
2bd63216	370	mlog(ML_BH_IO, "block=(%llu), nr=(%d), cached=%s, flags=0x%x\n",
b0697053	371	(unsigned long long)block, nr,
d4a8c93c JB	372	((flags & OCFS2_BH_IGNORE_CACHE) \|\| ignore_cache) ? "no" : "yes",
d4a8c93c JB	373	flags);
ccd979bd MF	374
	375	bail:
	376
	377	mlog_exit(status);
	378	return status;
	379	}
d659072f TM	380
	381	/* Check whether the blkno is the super block or one of the backups. */
	382	static void ocfs2_check_super_or_backup(struct super_block *sb,
	383	sector_t blkno)
	384	{
	385	int i;
	386	u64 backup_blkno;
	387
	388	if (blkno == OCFS2_SUPER_BLOCK_BLKNO)
	389	return;
	390
	391	for (i = 0; i < OCFS2_MAX_BACKUP_SUPERBLOCKS; i++) {
	392	backup_blkno = ocfs2_backup_super_blkno(sb, i);
	393	if (backup_blkno == blkno)
	394	return;
	395	}
	396
	397	BUG();
	398	}
	399
	400	/*
	401	* Write super block and backups doesn't need to collaborate with journal,
8cb471e8	402	* so we don't need to lock ip_io_mutex and ci doesn't need to bea passed
d659072f TM	403	* into this function.
	404	*/
	405	int ocfs2_write_super_or_backup(struct ocfs2_super *osb,
	406	struct buffer_head *bh)
	407	{
	408	int ret = 0;
	409
	410	mlog_entry_void();
	411
	412	BUG_ON(buffer_jbd(bh));
	413	ocfs2_check_super_or_backup(osb->sb, bh->b_blocknr);
	414
	415	if (ocfs2_is_hard_readonly(osb) \|\| ocfs2_is_soft_readonly(osb)) {
	416	ret = -EROFS;
	417	goto out;
	418	}
	419
	420	lock_buffer(bh);
	421	set_buffer_uptodate(bh);
	422
	423	/* remove from dirty list before I/O. */
	424	clear_buffer_dirty(bh);
	425
	426	get_bh(bh); /* for end_buffer_write_sync() */
	427	bh->b_end_io = end_buffer_write_sync;
	428	submit_bh(WRITE, bh);
	429
	430	wait_on_buffer(bh);
	431
	432	if (!buffer_uptodate(bh)) {
	433	ret = -EIO;
2fe5c1d7	434	put_bh(bh);
d659072f TM	435	}
	436
	437	out:
	438	mlog_exit(ret);
	439	return ret;
	440	}