super.c

linux 内核源代码

/* -*- mode: c; c-basic-offset: 8; -*-
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * super.c
 *
 * load/unload driver, mount/dismount volumes
 *
 * 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/module.h>
#include <linux/fs.h>
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/highmem.h>
#include <linux/utsname.h>
#include <linux/init.h>
#include <linux/random.h>
#include <linux/statfs.h>
#include <linux/moduleparam.h>
#include <linux/blkdev.h>
#include <linux/socket.h>
#include <linux/inet.h>
#include <linux/parser.h>
#include <linux/crc32.h>
#include <linux/debugfs.h>
#include <linux/mount.h>

#include <cluster/nodemanager.h>

#define MLOG_MASK_PREFIX ML_SUPER
#include <cluster/masklog.h>

#include "ocfs2.h"

/* this should be the only file to include a version 1 header */
#include "ocfs1_fs_compat.h"

#include "alloc.h"
#include "dlmglue.h"
#include "export.h"
#include "extent_map.h"
#include "heartbeat.h"
#include "inode.h"
#include "journal.h"
#include "localalloc.h"
#include "namei.h"
#include "slot_map.h"
#include "super.h"
#include "sysfile.h"
#include "uptodate.h"
#include "ver.h"
#include "vote.h"

#include "buffer_head_io.h"

static struct kmem_cache *ocfs2_inode_cachep = NULL;

/* OCFS2 needs to schedule several differnt types of work which
 * require cluster locking, disk I/O, recovery waits, etc. Since these
 * types of work tend to be heavy we avoid using the kernel events
 * workqueue and schedule on our own. */
struct workqueue_struct *ocfs2_wq = NULL;

static struct dentry *ocfs2_debugfs_root = NULL;

MODULE_AUTHOR("Oracle");
MODULE_LICENSE("GPL");

struct mount_options
{
	unsigned long	mount_opt;
	unsigned int	atime_quantum;
	signed short	slot;
};

static int ocfs2_parse_options(struct super_block *sb, char *options,
			       struct mount_options *mopt,
			       int is_remount);
static int ocfs2_show_options(struct seq_file *s, struct vfsmount *mnt);
static void ocfs2_put_super(struct super_block *sb);
static int ocfs2_mount_volume(struct super_block *sb);
static int ocfs2_remount(struct super_block *sb, int *flags, char *data);
static void ocfs2_dismount_volume(struct super_block *sb, int mnt_err);
static int ocfs2_initialize_mem_caches(void);
static void ocfs2_free_mem_caches(void);
static void ocfs2_delete_osb(struct ocfs2_super *osb);

static int ocfs2_statfs(struct dentry *dentry, struct kstatfs *buf);

static int ocfs2_sync_fs(struct super_block *sb, int wait);

static int ocfs2_init_global_system_inodes(struct ocfs2_super *osb);
static int ocfs2_init_local_system_inodes(struct ocfs2_super *osb);
static void ocfs2_release_system_inodes(struct ocfs2_super *osb);
static int ocfs2_fill_local_node_info(struct ocfs2_super *osb);
static int ocfs2_check_volume(struct ocfs2_super *osb);
static int ocfs2_verify_volume(struct ocfs2_dinode *di,
			       struct buffer_head *bh,
			       u32 sectsize);
static int ocfs2_initialize_super(struct super_block *sb,
				  struct buffer_head *bh,
				  int sector_size);
static int ocfs2_get_sector(struct super_block *sb,
			    struct buffer_head **bh,
			    int block,
			    int sect_size);
static void ocfs2_write_super(struct super_block *sb);
static struct inode *ocfs2_alloc_inode(struct super_block *sb);
static void ocfs2_destroy_inode(struct inode *inode);

static const struct super_operations ocfs2_sops = {
	.statfs		= ocfs2_statfs,
	.alloc_inode	= ocfs2_alloc_inode,
	.destroy_inode	= ocfs2_destroy_inode,
	.drop_inode	= ocfs2_drop_inode,
	.clear_inode	= ocfs2_clear_inode,
	.delete_inode	= ocfs2_delete_inode,
	.sync_fs	= ocfs2_sync_fs,
	.write_super	= ocfs2_write_super,
	.put_super	= ocfs2_put_super,
	.remount_fs	= ocfs2_remount,
	.show_options   = ocfs2_show_options,
};

enum {
	Opt_barrier,
	Opt_err_panic,
	Opt_err_ro,
	Opt_intr,
	Opt_nointr,
	Opt_hb_none,
	Opt_hb_local,
	Opt_data_ordered,
	Opt_data_writeback,
	Opt_atime_quantum,
	Opt_slot,
	Opt_err,
};

static match_table_t tokens = {
	{Opt_barrier, "barrier=%u"},
	{Opt_err_panic, "errors=panic"},
	{Opt_err_ro, "errors=remount-ro"},
	{Opt_intr, "intr"},
	{Opt_nointr, "nointr"},
	{Opt_hb_none, OCFS2_HB_NONE},
	{Opt_hb_local, OCFS2_HB_LOCAL},
	{Opt_data_ordered, "data=ordered"},
	{Opt_data_writeback, "data=writeback"},
	{Opt_atime_quantum, "atime_quantum=%u"},
	{Opt_slot, "preferred_slot=%u"},
	{Opt_err, NULL}
};

/*
 * write_super and sync_fs ripped right out of ext3.
 */
static void ocfs2_write_super(struct super_block *sb)
{
	if (mutex_trylock(&sb->s_lock) != 0)
		BUG();
	sb->s_dirt = 0;
}

static int ocfs2_sync_fs(struct super_block *sb, int wait)
{
	int status;
	tid_t target;
	struct ocfs2_super *osb = OCFS2_SB(sb);

	sb->s_dirt = 0;

	if (ocfs2_is_hard_readonly(osb))
		return -EROFS;

	if (wait) {
		status = ocfs2_flush_truncate_log(osb);
		if (status < 0)
			mlog_errno(status);
	} else {
		ocfs2_schedule_truncate_log_flush(osb, 0);
	}

	if (journal_start_commit(OCFS2_SB(sb)->journal->j_journal, &target)) {
		if (wait)
			log_wait_commit(OCFS2_SB(sb)->journal->j_journal,
					target);
	}
	return 0;
}

static int ocfs2_init_global_system_inodes(struct ocfs2_super *osb)
{
	struct inode *new = NULL;
	int status = 0;
	int i;

	mlog_entry_void();

	new = ocfs2_iget(osb, osb->root_blkno, OCFS2_FI_FLAG_SYSFILE);
	if (IS_ERR(new)) {
		status = PTR_ERR(new);
		mlog_errno(status);
		goto bail;
	}
	osb->root_inode = new;

	new = ocfs2_iget(osb, osb->system_dir_blkno, OCFS2_FI_FLAG_SYSFILE);
	if (IS_ERR(new)) {
		status = PTR_ERR(new);
		mlog_errno(status);
		goto bail;
	}
	osb->sys_root_inode = new;

	for (i = OCFS2_FIRST_ONLINE_SYSTEM_INODE;
	     i <= OCFS2_LAST_GLOBAL_SYSTEM_INODE; i++) {
		new = ocfs2_get_system_file_inode(osb, i, osb->slot_num);
		if (!new) {
			ocfs2_release_system_inodes(osb);
			status = -EINVAL;
			mlog_errno(status);
			/* FIXME: Should ERROR_RO_FS */
			mlog(ML_ERROR, "Unable to load system inode %d, "
			     "possibly corrupt fs?", i);
			goto bail;
		}
		// the array now has one ref, so drop this one
		iput(new);
	}

bail:
	mlog_exit(status);
	return status;
}

static int ocfs2_init_local_system_inodes(struct ocfs2_super *osb)
{
	struct inode *new = NULL;
	int status = 0;
	int i;

	mlog_entry_void();

	for (i = OCFS2_LAST_GLOBAL_SYSTEM_INODE + 1;
	     i < NUM_SYSTEM_INODES;
	     i++) {
		new = ocfs2_get_system_file_inode(osb, i, osb->slot_num);
		if (!new) {
			ocfs2_release_system_inodes(osb);
			status = -EINVAL;
			mlog(ML_ERROR, "status=%d, sysfile=%d, slot=%d\n",
			     status, i, osb->slot_num);
			goto bail;
		}
		/* the array now has one ref, so drop this one */
		iput(new);
	}

bail:
	mlog_exit(status);
	return status;
}

static void ocfs2_release_system_inodes(struct ocfs2_super *osb)
{
	int i;
	struct inode *inode;

	mlog_entry_void();

	for (i = 0; i < NUM_SYSTEM_INODES; i++) {
		inode = osb->system_inodes[i];
		if (inode) {
			iput(inode);
			osb->system_inodes[i] = NULL;
		}
	}

	inode = osb->sys_root_inode;
	if (inode) {
		iput(inode);
		osb->sys_root_inode = NULL;
	}

	inode = osb->root_inode;
	if (inode) {
		iput(inode);
		osb->root_inode = NULL;
	}

	mlog_exit(0);
}

/* We're allocating fs objects, use GFP_NOFS */
static struct inode *ocfs2_alloc_inode(struct super_block *sb)
{
	struct ocfs2_inode_info *oi;

	oi = kmem_cache_alloc(ocfs2_inode_cachep, GFP_NOFS);
	if (!oi)
		return NULL;

	return &oi->vfs_inode;
}

static void ocfs2_destroy_inode(struct inode *inode)
{
	kmem_cache_free(ocfs2_inode_cachep, OCFS2_I(inode));
}

static unsigned long long ocfs2_max_file_offset(unsigned int bbits,
						unsigned int cbits)
{
	unsigned int bytes = 1 << cbits;
	unsigned int trim = bytes;
	unsigned int bitshift = 32;

	/*
	 * i_size and all block offsets in ocfs2 are always 64 bits
	 * wide. i_clusters is 32 bits, in cluster-sized units. So on
	 * 64 bit platforms, cluster size will be the limiting factor.
	 */

#if BITS_PER_LONG == 32
# if defined(CONFIG_LBD)
	BUILD_BUG_ON(sizeof(sector_t) != 8);
	/*
	 * We might be limited by page cache size.
	 */
	if (bytes > PAGE_CACHE_SIZE) {
		bytes = PAGE_CACHE_SIZE;
		trim = 1;
		/*
		 * Shift by 31 here so that we don't get larger than
		 * MAX_LFS_FILESIZE
		 */
		bitshift = 31;
	}
# else
	/*
	 * We are limited by the size of sector_t. Use block size, as
	 * that's what we expose to the VFS.
	 */
	bytes = 1 << bbits;
	trim = 1;
	bitshift = 31;
# endif
#endif

	/*
	 * Trim by a whole cluster when we can actually approach the
	 * on-disk limits. Otherwise we can overflow i_clusters when
	 * an extent start is at the max offset.
	 */
	return (((unsigned long long)bytes) << bitshift) - trim;
}

static int ocfs2_remount(struct super_block *sb, int *flags, char *data)
{
	int incompat_features;
	int ret = 0;
	struct mount_options parsed_options;
	struct ocfs2_super *osb = OCFS2_SB(sb);

	if (!ocfs2_parse_options(sb, data, &parsed_options, 1)) {
		ret = -EINVAL;
		goto out;
	}

	if ((osb->s_mount_opt & OCFS2_MOUNT_HB_LOCAL) !=
	    (parsed_options.mount_opt & OCFS2_MOUNT_HB_LOCAL)) {
		ret = -EINVAL;
		mlog(ML_ERROR, "Cannot change heartbeat mode on remount\n");
		goto out;
	}

	if ((osb->s_mount_opt & OCFS2_MOUNT_DATA_WRITEBACK) !=
	    (parsed_options.mount_opt & OCFS2_MOUNT_DATA_WRITEBACK)) {
		ret = -EINVAL;
		mlog(ML_ERROR, "Cannot change data mode on remount\n");
		goto out;
	}

	/* We're going to/from readonly mode. */
	if ((*flags & MS_RDONLY) != (sb->s_flags & MS_RDONLY)) {
		/* Lock here so the check of HARD_RO and the potential
		 * setting of SOFT_RO is atomic. */
		spin_lock(&osb->osb_lock);
		if (osb->osb_flags & OCFS2_OSB_HARD_RO) {
			mlog(ML_ERROR, "Remount on readonly device is forbidden.\n");
			ret = -EROFS;
			goto unlock_osb;
		}

		if (*flags & MS_RDONLY) {
			mlog(0, "Going to ro mode.\n");
			sb->s_flags |= MS_RDONLY;
			osb->osb_flags |= OCFS2_OSB_SOFT_RO;
		} else {
			mlog(0, "Making ro filesystem writeable.\n");

			if (osb->osb_flags & OCFS2_OSB_ERROR_FS) {
				mlog(ML_ERROR, "Cannot remount RDWR "
				     "filesystem due to previous errors.\n");
				ret = -EROFS;
				goto unlock_osb;
			}
			incompat_features = OCFS2_HAS_RO_COMPAT_FEATURE(sb, ~OCFS2_FEATURE_RO_COMPAT_SUPP);
			if (incompat_features) {
				mlog(ML_ERROR, "Cannot remount RDWR because "
				     "of unsupported optional features "
				     "(%x).\n", incompat_features);
				ret = -EINVAL;
				goto unlock_osb;
			}
			sb->s_flags &= ~MS_RDONLY;
			osb->osb_flags &= ~OCFS2_OSB_SOFT_RO;
		}
unlock_osb:
		spin_unlock(&osb->osb_lock);
	}

	if (!ret) {
		/* Only save off the new mount options in case of a successful
		 * remount. */
		osb->s_mount_opt = parsed_options.mount_opt;
		osb->s_atime_quantum = parsed_options.atime_quantum;
		osb->preferred_slot = parsed_options.slot;

		if (!ocfs2_is_hard_readonly(osb))
			ocfs2_set_journal_params(osb);
	}
out:
	return ret;
}

static int ocfs2_sb_probe(struct super_block *sb,
			  struct buffer_head **bh,
			  int *sector_size)
{
	int status, tmpstat;
	struct ocfs1_vol_disk_hdr *hdr;
	struct ocfs2_dinode *di;
	int blksize;

	*bh = NULL;

	/* may be > 512 */
	*sector_size = bdev_hardsect_size(sb->s_bdev);
	if (*sector_size > OCFS2_MAX_BLOCKSIZE) {
		mlog(ML_ERROR, "Hardware sector size too large: %d (max=%d)\n",
		     *sector_size, OCFS2_MAX_BLOCKSIZE);
		status = -EINVAL;
		goto bail;
	}

	/* Can this really happen? */
	if (*sector_size < OCFS2_MIN_BLOCKSIZE)
		*sector_size = OCFS2_MIN_BLOCKSIZE;

	/* check block zero for old format */
	status = ocfs2_get_sector(sb, bh, 0, *sector_size);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}
	hdr = (struct ocfs1_vol_disk_hdr *) (*bh)->b_data;
	if (hdr->major_version == OCFS1_MAJOR_VERSION) {
		mlog(ML_ERROR, "incompatible version: %u.%u\n",
		     hdr->major_version, hdr->minor_version);
		status = -EINVAL;
	}
	if (memcmp(hdr->signature, OCFS1_VOLUME_SIGNATURE,
		   strlen(OCFS1_VOLUME_SIGNATURE)) == 0) {
		mlog(ML_ERROR, "incompatible volume signature: %8s\n",
		     hdr->signature);
		status = -EINVAL;
	}
	brelse(*bh);
	*bh = NULL;
	if (status < 0) {
		mlog(ML_ERROR, "This is an ocfs v1 filesystem which must be "
		     "upgraded before mounting with ocfs v2\n");
		goto bail;
	}

	/*
	 * Now check at magic offset for 512, 1024, 2048, 4096
	 * blocksizes.  4096 is the maximum blocksize because it is
	 * the minimum clustersize.
	 */
	status = -EINVAL;
	for (blksize = *sector_size;
	     blksize <= OCFS2_MAX_BLOCKSIZE;
	     blksize <<= 1) {
		tmpstat = ocfs2_get_sector(sb, bh,
					   OCFS2_SUPER_BLOCK_BLKNO,
					   blksize);
		if (tmpstat < 0) {
			status = tmpstat;
			mlog_errno(status);
			goto bail;
		}
		di = (struct ocfs2_dinode *) (*bh)->b_data;
		status = ocfs2_verify_volume(di, *bh, blksize);
		if (status >= 0)
			goto bail;
		brelse(*bh);
		*bh = NULL;
		if (status != -EAGAIN)
			break;
	}

bail:
	return status;
}

static int ocfs2_verify_heartbeat(struct ocfs2_super *osb)
{
	if (ocfs2_mount_local(osb)) {
		if (osb->s_mount_opt & OCFS2_MOUNT_HB_LOCAL) {
			mlog(ML_ERROR, "Cannot heartbeat on a locally "
			     "mounted device.\n");
			return -EINVAL;
		}
	}

	if (!(osb->s_mount_opt & OCFS2_MOUNT_HB_LOCAL)) {
		if (!ocfs2_mount_local(osb) && !ocfs2_is_hard_readonly(osb)) {
			mlog(ML_ERROR, "Heartbeat has to be started to mount "
			     "a read-write clustered device.\n");
			return -EINVAL;
		}
	}

	return 0;
}

static int ocfs2_fill_super(struct super_block *sb, void *data, int silent)
{
	struct dentry *root;
	int status, sector_size;
	struct mount_options parsed_options;
	struct inode *inode = NULL;
	struct ocfs2_super *osb = NULL;
	struct buffer_head *bh = NULL;
	char nodestr[8];

	mlog_entry("%p, %p, %i", sb, data, silent);

	if (!ocfs2_parse_options(sb, data, &parsed_options, 0)) {
		status = -EINVAL;
		goto read_super_error;
	}

	/* for now we only have one cluster/node, make sure we see it
	 * in the heartbeat universe */
	if (parsed_options.mount_opt & OCFS2_MOUNT_HB_LOCAL) {
		if (!o2hb_check_local_node_heartbeating()) {
			status = -EINVAL;
			goto read_super_error;
		}
	}

	/* probe for superblock */
	status = ocfs2_sb_probe(sb, &bh, &sector_size);
	if (status < 0) {
		mlog(ML_ERROR, "superblock probe failed!\n");
		goto read_super_error;
	}

	status = ocfs2_initialize_super(sb, bh, sector_size);
	osb = OCFS2_SB(sb);
	if (status < 0) {
		mlog_errno(status);
		goto read_super_error;
	}
	brelse(bh);
	bh = NULL;
	osb->s_mount_opt = parsed_options.mount_opt;
	osb->s_atime_quantum = parsed_options.atime_quantum;
	osb->preferred_slot = parsed_options.slot;

	sb->s_magic = OCFS2_SUPER_MAGIC;

	/* Hard readonly mode only if: bdev_read_only, MS_RDONLY,
	 * heartbeat=none */
	if (bdev_read_only(sb->s_bdev)) {
		if (!(sb->s_flags & MS_RDONLY)) {
			status = -EACCES;
			mlog(ML_ERROR, "Readonly device detected but readonly "