]> git.ipfire.org Git - thirdparty/kernel/linux.git/commit
configfs_lookup(): don't leave ->s_dentry dangling on failure
authorAl Viro <viro@zeniv.linux.org.uk>
Tue, 12 May 2026 04:29:37 +0000 (00:29 -0400)
committerAl Viro <viro@zeniv.linux.org.uk>
Wed, 3 Jun 2026 07:15:19 +0000 (03:15 -0400)
commit10da12d352b7b2bb330a8609fdda9a58bf0e9856
tree308c71e2af1ffa92a0b45aeeb7dd724a27b21dab
parente43ffb69e0438cddd72aaa30898b4dc446f664f8
configfs_lookup(): don't leave ->s_dentry dangling on failure

Normally ->s_dentry is cleared when dentry it's pointing to becomes
negative (on eviction, realistically).  However, that only happens
if dentry gets to be positive in the first place; in case of inode
allocation failure dentry never becomes positive, so ->d_iput()
is not called at all.

We do part of what normally would've been done by configfs_d_iput()
(dropping the reference to configfs_dirent) manually, but we do
not clear ->s_dentry there.  Sloppy as it is, it does not matter in
case of configfs_create_{dir,link}() - there configfs_dirent does
not survive dropping the sole reference to it.

However, for configfs_lookup() it *does* survive, with a dangling
pointer to soon to be freed dentry sitting it its ->s_dentry.

Subsequent getdents(2) in that directory will end up dereferencing
that pointer in order to pick the inode number.  Use after free...

This is the minimal fix; the right approach is to set the linkage
between dentry and configfs_dirent only after we know that we have
an inode, but that takes more surgery and the bug had been there
since 2006, so...

Fixes: 3d0f89bb1694 ("configfs: Add permission and ownership to configfs objects") # 2.6.16-rc3
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Breno Leitao <leitao@debian.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
fs/configfs/dir.c