nilfs2: always lookup disk block address before reading metadata block (14351104) · Commits · 戴 / test

The current metadata file code skips disk address lookup for its data block if the buffer has a mapped flag. This has a potential risk to cause read request to be performed against the stale block address that GC moved, and it may lead to meta data corruption. The mapped flag is safe if the buffer has an uptodate flag, otherwise it may prevent necessary update of disk address in the next read. This will avoid the potential problem by ensuring disk address lookup before reading metadata block even for buffers with the mapped flag. Signed-off-by:

Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>

fs/nilfs2/mdt.c

+10 −15

Original line number	Diff line number	Diff line
		@@ -103,15 +103,12 @@ static int nilfs_mdt_create_block(struct inode *inode, unsigned long block,
		goto failed_unlock;

		err = -EEXIST;
		if (buffer_uptodate(bh) \|\| buffer_mapped(bh))
		if (buffer_uptodate(bh))
		goto failed_bh;
		#if 0
		/* The uptodate flag is not protected by the page lock, but
		the mapped flag is. Thus, we don't have to wait the buffer. */

		wait_on_buffer(bh);
		if (buffer_uptodate(bh))
		goto failed_bh;
		#endif

		bh->b_bdev = nilfs->ns_bdev;
		err = nilfs_mdt_insert_new_block(inode, block, bh, init_block);
		@@ -162,9 +159,8 @@ nilfs_mdt_submit_block(struct inode *inode, unsigned long blkoff,
		unlock_buffer(bh);
		goto out;
		}
		if (!buffer_mapped(bh)) { /* unused buffer */
		ret = nilfs_bmap_lookup(NILFS_I(inode)->i_bmap, blkoff,
		&blknum);

		ret = nilfs_bmap_lookup(NILFS_I(inode)->i_bmap, blkoff, &blknum);
		if (unlikely(ret)) {
		unlock_buffer(bh);
		goto failed_bh;
		@@ -172,7 +168,6 @@ nilfs_mdt_submit_block(struct inode *inode, unsigned long blkoff,
		bh->b_bdev = NILFS_MDT(inode)->mi_nilfs->ns_bdev;
		bh->b_blocknr = blknum;
		set_buffer_mapped(bh);
		}

		bh->b_end_io = end_buffer_read_sync;
		get_bh(bh);

Admin message