On Feb 13, 2020, at 3:15 AM, Jan Kara <jack@suse.cz> wrote:
> 
> E2fsck directory rehashing code can fail with ENOSPC due to a bug in
> ext2fs_htree_intnode_maxrecs() which fails to take metadata checksum
> into account and thus e.g. e2fsck can decide to create 1 indirect level
> of index tree when two are actually needed. Fix the logic to account for
> metadata checksum.
> 
> Signed-off-by: Jan Kara <jack@suse.cz>

Reviewed-by: Andreas Dilger <adilger@dilger.ca>

> ---
> lib/ext2fs/ext2fs.h | 10 +++++++---
> 1 file changed, 7 insertions(+), 3 deletions(-)
> 
> diff --git a/lib/ext2fs/ext2fs.h b/lib/ext2fs/ext2fs.h
> index 93ecf29c568d..5fde3343b1f1 100644
> --- a/lib/ext2fs/ext2fs.h
> +++ b/lib/ext2fs/ext2fs.h
> @@ -1783,7 +1783,6 @@ extern blk_t ext2fs_group_first_block(ext2_filsys fs, dgrp_t group);
> extern blk_t ext2fs_group_last_block(ext2_filsys fs, dgrp_t group);
> extern blk_t ext2fs_inode_data_blocks(ext2_filsys fs,
> 				      struct ext2_inode *inode);
> -extern int ext2fs_htree_intnode_maxrecs(ext2_filsys fs, int blocks);
> extern unsigned int ext2fs_div_ceil(unsigned int a, unsigned int b);
> extern __u64 ext2fs_div64_ceil(__u64 a, __u64 b);
> extern int ext2fs_dirent_name_len(const struct ext2_dir_entry *entry);
> @@ -2015,9 +2014,14 @@ _INLINE_ blk_t ext2fs_inode_data_blocks(ext2_filsys fs,
> 	return (blk_t) ext2fs_inode_data_blocks2(fs, inode);
> }
> 
> -_INLINE_ int ext2fs_htree_intnode_maxrecs(ext2_filsys fs, int blocks)
> +static inline int ext2fs_htree_intnode_maxrecs(ext2_filsys fs, int blocks)
> {
> -	return blocks * ((fs->blocksize - 8) / sizeof(struct ext2_dx_entry));
> +	int csum_size = 0;
> +
> +	if (ext2fs_has_feature_metadata_csum(fs->super))
> +		csum_size = sizeof(struct ext2_dx_tail);
> +	return blocks * ((fs->blocksize - (8 + csum_size)) /
> +						sizeof(struct ext2_dx_entry));
> }
> 
> /*
> --
> 2.16.4
> 


Cheers, Andreas