super.c

linux 内核源代码

			     "mount was not specified.\n");
			goto read_super_error;
		}

		/* You should not be able to start a local heartbeat
		 * on a readonly device. */
		if (osb->s_mount_opt & OCFS2_MOUNT_HB_LOCAL) {
			status = -EROFS;
			mlog(ML_ERROR, "Local heartbeat specified on readonly "
			     "device.\n");
			goto read_super_error;
		}

		status = ocfs2_check_journals_nolocks(osb);
		if (status < 0) {
			if (status == -EROFS)
				mlog(ML_ERROR, "Recovery required on readonly "
				     "file system, but write access is "
				     "unavailable.\n");
			else
				mlog_errno(status);			
			goto read_super_error;
		}

		ocfs2_set_ro_flag(osb, 1);

		printk(KERN_NOTICE "Readonly device detected. No cluster "
		       "services will be utilized for this mount. Recovery "
		       "will be skipped.\n");
	}

	if (!ocfs2_is_hard_readonly(osb)) {
		if (sb->s_flags & MS_RDONLY)
			ocfs2_set_ro_flag(osb, 0);
	}

	status = ocfs2_verify_heartbeat(osb);
	if (status < 0) {
		mlog_errno(status);
		goto read_super_error;
	}

	osb->osb_debug_root = debugfs_create_dir(osb->uuid_str,
						 ocfs2_debugfs_root);
	if (!osb->osb_debug_root) {
		status = -EINVAL;
		mlog(ML_ERROR, "Unable to create per-mount debugfs root.\n");
		goto read_super_error;
	}

	status = ocfs2_mount_volume(sb);
	if (osb->root_inode)
		inode = igrab(osb->root_inode);

	if (status < 0)
		goto read_super_error;

	if (!inode) {
		status = -EIO;
		mlog_errno(status);
		goto read_super_error;
	}

	root = d_alloc_root(inode);
	if (!root) {
		status = -ENOMEM;
		mlog_errno(status);
		goto read_super_error;
	}

	sb->s_root = root;

	ocfs2_complete_mount_recovery(osb);

	if (ocfs2_mount_local(osb))
		snprintf(nodestr, sizeof(nodestr), "local");
	else
		snprintf(nodestr, sizeof(nodestr), "%d", osb->node_num);

	printk(KERN_INFO "ocfs2: Mounting device (%s) on (node %s, slot %d) "
	       "with %s data mode.\n",
	       osb->dev_str, nodestr, osb->slot_num,
	       osb->s_mount_opt & OCFS2_MOUNT_DATA_WRITEBACK ? "writeback" :
	       "ordered");

	atomic_set(&osb->vol_state, VOLUME_MOUNTED);
	wake_up(&osb->osb_mount_event);

	mlog_exit(status);
	return status;

read_super_error:
	if (bh != NULL)
		brelse(bh);

	if (inode)
		iput(inode);

	if (osb) {
		atomic_set(&osb->vol_state, VOLUME_DISABLED);
		wake_up(&osb->osb_mount_event);
		ocfs2_dismount_volume(sb, 1);
	}

	mlog_exit(status);
	return status;
}

static int ocfs2_get_sb(struct file_system_type *fs_type,
			int flags,
			const char *dev_name,
			void *data,
			struct vfsmount *mnt)
{
	return get_sb_bdev(fs_type, flags, dev_name, data, ocfs2_fill_super,
			   mnt);
}

static struct file_system_type ocfs2_fs_type = {
	.owner          = THIS_MODULE,
	.name           = "ocfs2",
	.get_sb         = ocfs2_get_sb, /* is this called when we mount
					* the fs? */
	.kill_sb        = kill_block_super, /* set to the generic one
					     * right now, but do we
					     * need to change that? */
	.fs_flags       = FS_REQUIRES_DEV|FS_RENAME_DOES_D_MOVE,
	.next           = NULL
};

static int ocfs2_parse_options(struct super_block *sb,
			       char *options,
			       struct mount_options *mopt,
			       int is_remount)
{
	int status;
	char *p;

	mlog_entry("remount: %d, options: \"%s\"\n", is_remount,
		   options ? options : "(none)");

	mopt->mount_opt = 0;
	mopt->atime_quantum = OCFS2_DEFAULT_ATIME_QUANTUM;
	mopt->slot = OCFS2_INVALID_SLOT;

	if (!options) {
		status = 1;
		goto bail;
	}

	while ((p = strsep(&options, ",")) != NULL) {
		int token, option;
		substring_t args[MAX_OPT_ARGS];

		if (!*p)
			continue;

		token = match_token(p, tokens, args);
		switch (token) {
		case Opt_hb_local:
			mopt->mount_opt |= OCFS2_MOUNT_HB_LOCAL;
			break;
		case Opt_hb_none:
			mopt->mount_opt &= ~OCFS2_MOUNT_HB_LOCAL;
			break;
		case Opt_barrier:
			if (match_int(&args[0], &option)) {
				status = 0;
				goto bail;
			}
			if (option)
				mopt->mount_opt |= OCFS2_MOUNT_BARRIER;
			else
				mopt->mount_opt &= ~OCFS2_MOUNT_BARRIER;
			break;
		case Opt_intr:
			mopt->mount_opt &= ~OCFS2_MOUNT_NOINTR;
			break;
		case Opt_nointr:
			mopt->mount_opt |= OCFS2_MOUNT_NOINTR;
			break;
		case Opt_err_panic:
			mopt->mount_opt |= OCFS2_MOUNT_ERRORS_PANIC;
			break;
		case Opt_err_ro:
			mopt->mount_opt &= ~OCFS2_MOUNT_ERRORS_PANIC;
			break;
		case Opt_data_ordered:
			mopt->mount_opt &= ~OCFS2_MOUNT_DATA_WRITEBACK;
			break;
		case Opt_data_writeback:
			mopt->mount_opt |= OCFS2_MOUNT_DATA_WRITEBACK;
			break;
		case Opt_atime_quantum:
			if (match_int(&args[0], &option)) {
				status = 0;
				goto bail;
			}
			if (option >= 0)
				mopt->atime_quantum = option;
			break;
		case Opt_slot:
			option = 0;
			if (match_int(&args[0], &option)) {
				status = 0;
				goto bail;
			}
			if (option)
				mopt->slot = (s16)option;
			break;
		default:
			mlog(ML_ERROR,
			     "Unrecognized mount option \"%s\" "
			     "or missing value\n", p);
			status = 0;
			goto bail;
		}
	}

	status = 1;

bail:
	mlog_exit(status);
	return status;
}

static int ocfs2_show_options(struct seq_file *s, struct vfsmount *mnt)
{
	struct ocfs2_super *osb = OCFS2_SB(mnt->mnt_sb);
	unsigned long opts = osb->s_mount_opt;

	if (opts & OCFS2_MOUNT_HB_LOCAL)
		seq_printf(s, ",_netdev,heartbeat=local");
	else
		seq_printf(s, ",heartbeat=none");

	if (opts & OCFS2_MOUNT_NOINTR)
		seq_printf(s, ",nointr");

	if (opts & OCFS2_MOUNT_DATA_WRITEBACK)
		seq_printf(s, ",data=writeback");
	else
		seq_printf(s, ",data=ordered");

	if (opts & OCFS2_MOUNT_BARRIER)
		seq_printf(s, ",barrier=1");

	if (opts & OCFS2_MOUNT_ERRORS_PANIC)
		seq_printf(s, ",errors=panic");
	else
		seq_printf(s, ",errors=remount-ro");

	if (osb->preferred_slot != OCFS2_INVALID_SLOT)
		seq_printf(s, ",preferred_slot=%d", osb->preferred_slot);

	if (osb->s_atime_quantum != OCFS2_DEFAULT_ATIME_QUANTUM)
		seq_printf(s, ",atime_quantum=%u", osb->s_atime_quantum);

	return 0;
}

static int __init ocfs2_init(void)
{
	int status;

	mlog_entry_void();

	ocfs2_print_version();

	status = init_ocfs2_uptodate_cache();
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_initialize_mem_caches();
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	ocfs2_wq = create_singlethread_workqueue("ocfs2_wq");
	if (!ocfs2_wq) {
		status = -ENOMEM;
		goto leave;
	}

	ocfs2_debugfs_root = debugfs_create_dir("ocfs2", NULL);
	if (!ocfs2_debugfs_root) {
		status = -EFAULT;
		mlog(ML_ERROR, "Unable to create ocfs2 debugfs root.\n");
	}

leave:
	if (status < 0) {
		ocfs2_free_mem_caches();
		exit_ocfs2_uptodate_cache();
	}

	mlog_exit(status);

	if (status >= 0) {
		return register_filesystem(&ocfs2_fs_type);
	} else
		return -1;
}

static void __exit ocfs2_exit(void)
{
	mlog_entry_void();

	if (ocfs2_wq) {
		flush_workqueue(ocfs2_wq);
		destroy_workqueue(ocfs2_wq);
	}

	debugfs_remove(ocfs2_debugfs_root);

	ocfs2_free_mem_caches();

	unregister_filesystem(&ocfs2_fs_type);

	exit_ocfs2_uptodate_cache();

	mlog_exit_void();
}

static void ocfs2_put_super(struct super_block *sb)
{
	mlog_entry("(0x%p)\n", sb);

	ocfs2_sync_blockdev(sb);
	ocfs2_dismount_volume(sb, 0);

	mlog_exit_void();
}

static int ocfs2_statfs(struct dentry *dentry, struct kstatfs *buf)
{
	struct ocfs2_super *osb;
	u32 numbits, freebits;
	int status;
	struct ocfs2_dinode *bm_lock;
	struct buffer_head *bh = NULL;
	struct inode *inode = NULL;

	mlog_entry("(%p, %p)\n", dentry->d_sb, buf);

	osb = OCFS2_SB(dentry->d_sb);

	inode = ocfs2_get_system_file_inode(osb,
					    GLOBAL_BITMAP_SYSTEM_INODE,
					    OCFS2_INVALID_SLOT);
	if (!inode) {
		mlog(ML_ERROR, "failed to get bitmap inode\n");
		status = -EIO;
		goto bail;
	}

	status = ocfs2_meta_lock(inode, &bh, 0);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	bm_lock = (struct ocfs2_dinode *) bh->b_data;

	numbits = le32_to_cpu(bm_lock->id1.bitmap1.i_total);
	freebits = numbits - le32_to_cpu(bm_lock->id1.bitmap1.i_used);

	buf->f_type = OCFS2_SUPER_MAGIC;
	buf->f_bsize = dentry->d_sb->s_blocksize;
	buf->f_namelen = OCFS2_MAX_FILENAME_LEN;
	buf->f_blocks = ((sector_t) numbits) *
			(osb->s_clustersize >> osb->sb->s_blocksize_bits);
	buf->f_bfree = ((sector_t) freebits) *
		       (osb->s_clustersize >> osb->sb->s_blocksize_bits);
	buf->f_bavail = buf->f_bfree;
	buf->f_files = numbits;
	buf->f_ffree = freebits;

	brelse(bh);

	ocfs2_meta_unlock(inode, 0);
	status = 0;
bail:
	if (inode)
		iput(inode);

	mlog_exit(status);

	return status;
}

static void ocfs2_inode_init_once(struct kmem_cache *cachep, void *data)
{
	struct ocfs2_inode_info *oi = data;

	oi->ip_flags = 0;
	oi->ip_open_count = 0;
	spin_lock_init(&oi->ip_lock);
	ocfs2_extent_map_init(&oi->vfs_inode);
	INIT_LIST_HEAD(&oi->ip_io_markers);
	oi->ip_created_trans = 0;
	oi->ip_last_trans = 0;
	oi->ip_dir_start_lookup = 0;

	init_rwsem(&oi->ip_alloc_sem);
	mutex_init(&oi->ip_io_mutex);

	oi->ip_blkno = 0ULL;
	oi->ip_clusters = 0;

	ocfs2_lock_res_init_once(&oi->ip_rw_lockres);
	ocfs2_lock_res_init_once(&oi->ip_meta_lockres);
	ocfs2_lock_res_init_once(&oi->ip_data_lockres);
	ocfs2_lock_res_init_once(&oi->ip_open_lockres);

	ocfs2_metadata_cache_init(&oi->vfs_inode);

	inode_init_once(&oi->vfs_inode);
}

static int ocfs2_initialize_mem_caches(void)
{
	ocfs2_inode_cachep = kmem_cache_create("ocfs2_inode_cache",
				       sizeof(struct ocfs2_inode_info),
				       0,
				       (SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|
						SLAB_MEM_SPREAD),
				       ocfs2_inode_init_once);
	if (!ocfs2_inode_cachep)
		return -ENOMEM;

	return 0;
}

static void ocfs2_free_mem_caches(void)
{
	if (ocfs2_inode_cachep)
		kmem_cache_destroy(ocfs2_inode_cachep);

	ocfs2_inode_cachep = NULL;
}

static int ocfs2_get_sector(struct super_block *sb,
			    struct buffer_head **bh,
			    int block,
			    int sect_size)
{
	if (!sb_set_blocksize(sb, sect_size)) {
		mlog(ML_ERROR, "unable to set blocksize\n");
		return -EIO;
	}

	*bh = sb_getblk(sb, block);
	if (!*bh) {
		mlog_errno(-EIO);
		return -EIO;
	}
	lock_buffer(*bh);
	if (!buffer_dirty(*bh))
		clear_buffer_uptodate(*bh);
	unlock_buffer(*bh);
	ll_rw_block(READ, 1, bh);
	wait_on_buffer(*bh);
	return 0;
}

/* ocfs2 1.0 only allows one cluster and node identity per kernel image. */
static int ocfs2_fill_local_node_info(struct ocfs2_super *osb)
{
	int status;

	/* XXX hold a ref on the node while mounte?  easy enough, if
	 * desirable. */
	if (ocfs2_mount_local(osb))
		osb->node_num = 0;
	else
		osb->node_num = o2nm_this_node();

	if (osb->node_num == O2NM_MAX_NODES) {
		mlog(ML_ERROR, "could not find this host's node number\n");
		status = -ENOENT;
		goto bail;
	}

	mlog(0, "I am node %d\n", osb->node_num);

	status = 0;
bail:
	return status;
}

static int ocfs2_mount_volume(struct super_block *sb)
{
	int status = 0;
	int unlock_super = 0;
	struct ocfs2_super *osb = OCFS2_SB(sb);

	mlog_entry_void();

	if (ocfs2_is_hard_readonly(osb))
		goto leave;

	status = ocfs2_fill_local_node_info(osb);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_register_hb_callbacks(osb);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_dlm_init(osb);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	/* requires vote_thread to be running. */
	status = ocfs2_register_net_handlers(osb);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_super_lock(osb, 1);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}
	unlock_super = 1;

	/* This will load up the node map and add ourselves to it. */
	status = ocfs2_find_slot(osb);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	ocfs2_populate_mounted_map(osb);

	/* load all node-local system inodes */
	status = ocfs2_init_local_system_inodes(osb);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_check_volume(osb);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_truncate_log_init(osb);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	if (ocfs2_mount_local(osb))
		goto leave;

	/* This should be sent *after* we recovered our journal as it
	 * will cause other nodes to unmark us as needing
	 * recovery. However, we need to send it *before* dropping the
	 * super block lock as otherwise their recovery threads might
	 * try to clean us up while we're live! */
	status = ocfs2_request_mount_vote(osb);
	if (status < 0)
		mlog_errno(status);

leave:
	if (unlock_super)
		ocfs2_super_unlock(osb, 1);

	mlog_exit(status);
	return status;
}

/* we can't grab the goofy sem lock from inside wait_event, so we use
 * memory barriers to make sure that we'll see the null task before
 * being woken up */
static int ocfs2_recovery_thread_running(struct ocfs2_super *osb)
{
	mb();
	return osb->recovery_thread_task != NULL;
}

static void ocfs2_dismount_volume(struct super_block *sb, int mnt_err)
{
	int tmp;
	struct ocfs2_super *osb = NULL;
	char nodestr[8];

	mlog_entry("(0x%p)\n", sb);

	BUG_ON(!sb);
	osb = OCFS2_SB(sb);
	BUG_ON(!osb);

	ocfs2_shutdown_local_alloc(osb);