namei.c

ocfs1.2.7 源码

		BUG();
	}

	ocfs2_inode_set_new(osb, inode);
	if (!ocfs2_mount_local(osb)) {
		status = ocfs2_create_new_inode_locks(inode);
		if (status < 0)
			mlog_errno(status);
	}

	status = 0; /* error in ocfs2_create_new_inode_locks is not
		     * critical */

	*ret_inode = inode;
leave:
	if (status < 0) {
		if (*new_fe_bh) {
			brelse(*new_fe_bh);
			*new_fe_bh = NULL;
		}
		if (inode)
			iput(inode);
	}

	mlog_exit(status);
	return status;
}

static int ocfs2_mkdir(struct inode *dir,
		       struct dentry *dentry,
		       int mode)
{
	int ret;

	mlog_entry("(0x%p, 0x%p, %d, '%.*s')\n", dir, dentry, mode,
		   dentry->d_name.len, dentry->d_name.name);
	ret = ocfs2_mknod(dir, dentry, mode | S_IFDIR, 0);
	mlog_exit(ret);

	return ret;
}

static int ocfs2_create(struct inode *dir,
			struct dentry *dentry,
			int mode,
			struct nameidata *nd)
{
	int ret;

	mlog_entry("(0x%p, 0x%p, %d, '%.*s')\n", dir, dentry, mode,
		   dentry->d_name.len, dentry->d_name.name);
	ret = ocfs2_mknod(dir, dentry, mode | S_IFREG, 0);
	mlog_exit(ret);

	return ret;
}

static int ocfs2_link(struct dentry *old_dentry,
		      struct inode *dir,
		      struct dentry *dentry)
{
	struct ocfs2_journal_handle *handle = NULL;
	struct inode *inode = old_dentry->d_inode;
	int err;
	struct buffer_head *fe_bh = NULL;
	struct buffer_head *parent_fe_bh = NULL;
	struct buffer_head *de_bh = NULL;
	struct ocfs2_dinode *fe = NULL;
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);

	mlog_entry("(inode=%lu, old='%.*s' new='%.*s')\n", inode->i_ino,
		   old_dentry->d_name.len, old_dentry->d_name.name,
		   dentry->d_name.len, dentry->d_name.name);

	if (S_ISDIR(inode->i_mode)) {
		err = -EPERM;
		goto bail;
	}

	if (inode->i_nlink >= OCFS2_LINK_MAX) {
		err = -EMLINK;
		goto bail;
	}

	handle = ocfs2_alloc_handle(osb);
	if (handle == NULL) {
		err = -ENOMEM;
		goto bail;
	}

	err = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1);
	if (err < 0) {
		if (err != -ENOENT)
			mlog_errno(err);
		goto bail;
	}

	err = ocfs2_check_dir_for_entry(dir, dentry->d_name.name,
					dentry->d_name.len);
	if (err)
		goto bail;

	err = ocfs2_prepare_dir_for_insert(osb, dir, parent_fe_bh,
					   dentry->d_name.name,
					   dentry->d_name.len, &de_bh);
	if (err < 0) {
		mlog_errno(err);
		goto bail;
	}

	err = ocfs2_meta_lock(inode, handle, &fe_bh, 1);
	if (err < 0) {
		if (err != -ENOENT)
			mlog_errno(err);
		goto bail;
	}

	fe = (struct ocfs2_dinode *) fe_bh->b_data;
	if (le16_to_cpu(fe->i_links_count) >= OCFS2_LINK_MAX) {
		err = -EMLINK;
		goto bail;
	}

	handle = ocfs2_start_trans(osb, handle, OCFS2_LINK_CREDITS);
	if (IS_ERR(handle)) {
		err = PTR_ERR(handle);
		handle = NULL;
		mlog_errno(err);
		goto bail;
	}

	err = ocfs2_journal_access(handle, inode, fe_bh,
				   OCFS2_JOURNAL_ACCESS_WRITE);
	if (err < 0) {
		mlog_errno(err);
		goto bail;
	}

	inode->i_nlink++;
	inode->i_ctime = CURRENT_TIME;
	fe->i_links_count = cpu_to_le16(inode->i_nlink);
	fe->i_ctime = cpu_to_le64(inode->i_ctime.tv_sec);
	fe->i_ctime_nsec = cpu_to_le32(inode->i_ctime.tv_nsec);

	err = ocfs2_journal_dirty(handle, fe_bh);
	if (err < 0) {
		le16_add_cpu(&fe->i_links_count, -1);
		inode->i_nlink--;
		mlog_errno(err);
		goto bail;
	}

	err = ocfs2_add_entry(handle, dentry, inode,
			      OCFS2_I(inode)->ip_blkno,
			      parent_fe_bh, de_bh);
	if (err) {
		le16_add_cpu(&fe->i_links_count, -1);
		inode->i_nlink--;
		mlog_errno(err);
		goto bail;
	}

	atomic_inc(&inode->i_count);
	dentry->d_op = &ocfs2_dentry_ops;
	d_instantiate(dentry, inode);
bail:
	if (handle)
		ocfs2_commit_trans(handle);
	if (de_bh)
		brelse(de_bh);
	if (fe_bh)
		brelse(fe_bh);
	if (parent_fe_bh)
		brelse(parent_fe_bh);

	mlog_exit(err);

	return err;
}

static int ocfs2_unlink(struct inode *dir,
			struct dentry *dentry)
{
	int status;
	unsigned int saved_nlink = 0;
	struct inode *inode = dentry->d_inode;
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
	u64 blkno;
	struct ocfs2_dinode *fe = NULL;
	struct buffer_head *fe_bh = NULL;
	struct buffer_head *parent_node_bh = NULL;
	struct ocfs2_journal_handle *handle = NULL;
	struct ocfs2_dir_entry *dirent = NULL;
	struct buffer_head *dirent_bh = NULL;
	char orphan_name[OCFS2_ORPHAN_NAMELEN + 1];
	struct buffer_head *orphan_entry_bh = NULL;

	mlog_entry("(0x%p, 0x%p, '%.*s')\n", dir, dentry,
		   dentry->d_name.len, dentry->d_name.name);

	BUG_ON(dentry->d_parent->d_inode != dir);

	mlog(0, "ino = %"MLFu64"\n", OCFS2_I(inode)->ip_blkno);

	if (inode == osb->root_inode) {
		mlog(0, "Cannot delete the root directory\n");
		status = -EPERM;
		goto leave;
	}

	handle = ocfs2_alloc_handle(osb);
	if (handle == NULL) {
		status = -ENOMEM;
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_meta_lock(dir, handle, &parent_node_bh, 1);
	if (status < 0) {
		if (status != -ENOENT)
			mlog_errno(status);
		goto leave;
	}

	status = ocfs2_find_files_on_disk(dentry->d_name.name,
					  dentry->d_name.len, &blkno,
					  dir, &dirent_bh, &dirent);
	if (status < 0) {
		if (status != -ENOENT)
			mlog_errno(status);
		goto leave;
	}

	if (OCFS2_I(inode)->ip_blkno != blkno) {
		status = -ENOENT;

		mlog(0, "ip_blkno (%"MLFu64") != dirent blkno (%"MLFu64") "
		     "ip_flags = %x\n", OCFS2_I(inode)->ip_blkno, blkno,
		     OCFS2_I(inode)->ip_flags);
		goto leave;
	}

	status = ocfs2_meta_lock(inode, handle, &fe_bh, 1);
	if (status < 0) {
		if (status != -ENOENT)
			mlog_errno(status);
		goto leave;
	}

	if (S_ISDIR(inode->i_mode)) {
	       	if (!ocfs2_empty_dir(inode)) {
			status = -ENOTEMPTY;
			goto leave;
		} else if (inode->i_nlink != 2) {
			status = -ENOTEMPTY;
			goto leave;
		}
	}

	/* There are still a few steps left until we can consider the
	 * unlink to have succeeded. Save off nlink here before
	 * modification so we can set it back in case we hit an issue
	 * before commit. */
	saved_nlink = inode->i_nlink;
	if (S_ISDIR(inode->i_mode))
		inode->i_nlink = 0;
	else
		inode->i_nlink--;

	status = ocfs2_request_unlink_vote(inode, dentry,
					   (unsigned int) inode->i_nlink);
	if (status < 0) {
		/* This vote should succeed under all normal
		 * circumstances. */
		mlog_errno(status);
		goto leave;
	}

	if (!inode->i_nlink) {
		status = ocfs2_prepare_orphan_dir(osb, handle, inode,
						  orphan_name,
						  &orphan_entry_bh);
		if (status < 0) {
			mlog_errno(status);
			goto leave;
		}
	}

	handle = ocfs2_start_trans(osb, handle, OCFS2_UNLINK_CREDITS);
	if (IS_ERR(handle)) {
		status = PTR_ERR(handle);
		handle = NULL;
		mlog_errno(status);
		goto leave;
	}

	status = ocfs2_journal_access(handle, inode, fe_bh,
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	fe = (struct ocfs2_dinode *) fe_bh->b_data;

	if (!inode->i_nlink) {
		status = ocfs2_orphan_add(osb, handle, inode, fe, orphan_name,
					  orphan_entry_bh);
		if (status < 0) {
			mlog_errno(status);
			goto leave;
		}
	}

	/* delete the name from the parent dir */
	status = ocfs2_delete_entry(handle, dir, dirent, dirent_bh);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	/* We can set nlink on the dinode now. clear the saved version
	 * so that it doesn't get set later. */
	fe->i_links_count = cpu_to_le16(inode->i_nlink);
	saved_nlink = 0;

	status = ocfs2_journal_dirty(handle, fe_bh);
	if (status < 0) {
		mlog_errno(status);
		goto leave;
	}

	dir->i_ctime = dir->i_mtime = CURRENT_TIME;
	if (S_ISDIR(inode->i_mode))
		dir->i_nlink--;

	status = ocfs2_mark_inode_dirty(handle, dir,
					parent_node_bh);
	if (status < 0) {
		mlog_errno(status);
		if (S_ISDIR(inode->i_mode))
			dir->i_nlink++;
	}

leave:
	if (status < 0 && saved_nlink)
		inode->i_nlink = saved_nlink;

	if (handle)
		ocfs2_commit_trans(handle);

	if (fe_bh)
		brelse(fe_bh);

	if (dirent_bh)
		brelse(dirent_bh);

	if (parent_node_bh)
		brelse(parent_node_bh);

	if (orphan_entry_bh)
		brelse(orphan_entry_bh);

	mlog_exit(status);

	return status;
}

/*
 * The only place this should be used is rename!
 * if they have the same id, then the 1st one is the only one locked.
 */
static int ocfs2_double_lock(struct ocfs2_super *osb,
			     struct ocfs2_journal_handle *handle,
			     struct buffer_head **bh1,
			     struct inode *inode1,
			     struct buffer_head **bh2,
			     struct inode *inode2)
{
	int status;
	struct ocfs2_inode_info *oi1 = OCFS2_I(inode1);
	struct ocfs2_inode_info *oi2 = OCFS2_I(inode2);
	struct buffer_head **tmpbh;
	struct inode *tmpinode;

	mlog_entry("(inode1 = %"MLFu64", inode2 = %"MLFu64")\n",
		   oi1->ip_blkno, oi2->ip_blkno);

	BUG_ON(!handle);

	if (*bh1)
		*bh1 = NULL;
	if (*bh2)
		*bh2 = NULL;

	/* we always want to lock the one with the lower lockid first. */
	if (oi1->ip_blkno != oi2->ip_blkno) {
		if (oi1->ip_blkno < oi2->ip_blkno) {
			/* switch id1 and id2 around */
			mlog(0, "switching them around...\n");
			tmpbh = bh2;
			bh2 = bh1;
			bh1 = tmpbh;

			tmpinode = inode2;
			inode2 = inode1;
			inode1 = tmpinode;
		}
		/* lock id2 */
		status = ocfs2_meta_lock(inode2, handle, bh2, 1);
		if (status < 0) {
			if (status != -ENOENT)
				mlog_errno(status);
			goto bail;
		}
	}
	/* lock id1 */
	status = ocfs2_meta_lock(inode1, handle, bh1, 1);
	if (status < 0) {
		if (status != -ENOENT)
			mlog_errno(status);
		goto bail;
	}
bail:
	mlog_exit(status);
	return status;
}

#define PARENT_INO(buffer) \
	((struct ocfs2_dir_entry *) \
	 ((char *)buffer + \
	  le16_to_cpu(((struct ocfs2_dir_entry *)buffer)->rec_len)))->inode

static int ocfs2_rename(struct inode *old_dir,
			struct dentry *old_dentry,
			struct inode *new_dir,
			struct dentry *new_dentry)
{
	int status = 0, rename_lock = 0;
	struct inode *old_inode = old_dentry->d_inode;
	struct inode *new_inode = new_dentry->d_inode;
	struct ocfs2_dinode *oldfe = NULL;
	struct ocfs2_dinode *newfe = NULL;
	char orphan_name[OCFS2_ORPHAN_NAMELEN + 1];
	struct buffer_head *orphan_entry_bh = NULL;
	struct buffer_head *newfe_bh = NULL;
	struct buffer_head *old_inode_bh = NULL;
	struct buffer_head *insert_entry_bh = NULL;
	struct ocfs2_super *osb = NULL;
	u64 newfe_blkno;
	struct ocfs2_journal_handle *handle = NULL;
	struct buffer_head *old_dir_bh = NULL;
	struct buffer_head *new_dir_bh = NULL;
	struct ocfs2_dir_entry *old_de = NULL, *new_de = NULL; // dirent for old_dentry
							       // and new_dentry
	struct buffer_head *new_de_bh = NULL, *old_de_bh = NULL; // bhs for above
	struct buffer_head *old_inode_de_bh = NULL; // if old_dentry is a dir,
						    // this is the 1st dirent bh
	nlink_t old_dir_nlink = old_dir->i_nlink;
	unsigned int links_count;

	/* At some point it might be nice to break this function up a
	 * bit. */

	mlog_entry("(0x%p, 0x%p, 0x%p, 0x%p, from='%.*s' to='%.*s')\n",
		   old_dir, old_dentry, new_dir, new_dentry,
		   old_dentry->d_name.len, old_dentry->d_name.name,
		   new_dentry->d_name.len, new_dentry->d_name.name);

	osb = OCFS2_SB(old_dir->i_sb);

	if (new_inode) {
		if (!igrab(new_inode))
			BUG();
	}

	/* Assume a directory heirarchy thusly:
	 * a/b/c
	 * a/d
	 * a,b,c, and d are all directories.
	 *
	 * from cwd of 'a' on both nodes:
	 * node1: mv b/c d
	 * node2: mv d   b/c
	 *
	 * And that's why, just like the VFS, we need a file system
	 * rename lock. */
	if (old_dentry != new_dentry) {
		status = ocfs2_rename_lock(osb);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
		rename_lock = 1;
	}

	handle = ocfs2_alloc_handle(osb);
	if (handle == NULL) {
		status = -ENOMEM;
		mlog_errno(status);
		goto bail;
	}

	/* if old and new are the same, this'll just do one lock. */
	status = ocfs2_double_lock(osb, handle,
				  &old_dir_bh, old_dir,
				  &new_dir_bh, new_dir);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	/* make sure both dirs have bhs
	 * get an extra ref on old_dir_bh if old==new */
	if (!new_dir_bh) {
		if (old_dir_bh) {
			new_dir_bh = old_dir_bh;
			get_bh(new_dir_bh);
		} else {
			mlog(ML_ERROR, "no old_dir_bh!\n");
			status = -EIO;
			goto bail;
		}
	}

	status = ocfs2_meta_lock(old_inode, handle, &old_inode_bh, 1);
	if (status < 0) {
		if (status != -ENOENT)
			mlog_errno(status);
		goto bail;
	}

	status = ocfs2_request_rename_vote(old_inode, old_dentry);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	if (S_ISDIR(old_inode->i_mode)) {
		status = -EIO;
		old_inode_de_bh = ocfs2_bread(old_inode, 0, &status, 0);
		if (!old_inode_de_bh)
			goto bail;

		status = -EIO;
		if (le64_to_cpu(PARENT_INO(old_inode_de_bh->b_data)) !=
		    OCFS2_I(old_dir)->ip_blkno)
			goto bail;
		status = -EMLINK;
		if (!new_inode && new_dir!=old_dir &&
		    new_dir->i_nlink >= OCFS2_LINK_MAX)
			goto bail;
	}

	status = -ENOENT;
	old_de_bh = ocfs2_find_entry(old_dentry->d_name.name,
				     old_dentry->d_name.len,
				     old_dir, &old_de);
	if (!old_de_bh)
		goto bail;

	/*
	 *  Check for inode number is _not_ due to possible IO errors.
	 *  We might rmdir the source, keep it as pwd of some process
	 *  and merrily kill the link to whatever was created under the
	 *  same name. Goodbye sticky bit ;-<
	 */
	if (le64_to_cpu(old_de->inode) != OCFS2_I(old_inode)->ip_blkno)