[PATCH bpf] bpf: fs/xattr: don't assume the inode is locked in path_unlink/path_rmdir

[PATCH bpf] bpf: fs/xattr: don't assume the inode is locked in path_unlink/path_rmdir

[Andrea Parri]

bpf_lsm_has_d_inode_locked() makes the verifier rewrite
bpf_[set|remove]_dentry_xattr() to the _locked variants, which assume
that the caller already holds the inode's i_rwsem.  The path_unlink and
path_rmdir hooks are listed, but security_path_unlink() and
security_path_rmdir() run before vfs_unlink()/vfs_rmdir() take the
victim inode's i_rwsem, so a sleepable BPF LSM program attached to
either hook mutates the victim's xattrs without the lock held.

Drop the two path hooks from d_inode_locked_hooks so that the verifier
keeps the locking bpf_[set|remove]_dentry_xattr() variants, which take
the lock themselves.

Fixes: 56467292794b8 ("bpf: fs/xattr: Add BPF kfuncs to set and remove xattrs")
Cc: stable@vger.kernel.org
Assisted-by: LLM
Signed-off-by: Andrea Parri <parri.andrea@gmail.com>
---
 fs/bpf_fs_kfuncs.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/fs/bpf_fs_kfuncs.c b/fs/bpf_fs_kfuncs.c
index 6cb8772679782..357a379ef92a5 100644
--- a/fs/bpf_fs_kfuncs.c
+++ b/fs/bpf_fs_kfuncs.c
@@ -472,10 +472,6 @@ BTF_ID(func, bpf_lsm_inode_rmdir)
 BTF_ID(func, bpf_lsm_inode_setattr)
 BTF_ID(func, bpf_lsm_inode_setxattr)
 BTF_ID(func, bpf_lsm_inode_unlink)
-#ifdef CONFIG_SECURITY_PATH
-BTF_ID(func, bpf_lsm_path_unlink)
-BTF_ID(func, bpf_lsm_path_rmdir)
-#endif /* CONFIG_SECURITY_PATH */
 BTF_SET_END(d_inode_locked_hooks)
 
 bool bpf_lsm_has_d_inode_locked(const struct bpf_prog *prog)
-- 
2.53.0

Re: [PATCH bpf] bpf: fs/xattr: don't assume the inode is locked in path_unlink/path_rmdir

[Alexei Starovoitov]

On Tue Sep 22, 2026 at 2:55 PM UTC, Andrea Parri wrote:
> bpf_lsm_has_d_inode_locked() makes the verifier rewrite
> bpf_[set|remove]_dentry_xattr() to the _locked variants, which assume
> that the caller already holds the inode's i_rwsem.  The path_unlink and
> path_rmdir hooks are listed, but security_path_unlink() and
> security_path_rmdir() run before vfs_unlink()/vfs_rmdir() take the
> victim inode's i_rwsem, so a sleepable BPF LSM program attached to
> either hook mutates the victim's xattrs without the lock held.
>
> Drop the two path hooks from d_inode_locked_hooks so that the verifier
> keeps the locking bpf_[set|remove]_dentry_xattr() variants, which take
> the lock themselves.
>
> Fixes: 56467292794b8 ("bpf: fs/xattr: Add BPF kfuncs to set and remove xattrs")
> Cc: stable@vger.kernel.org
> Assisted-by: LLM
> Signed-off-by: Andrea Parri <parri.andrea@gmail.com>
> ---
>  fs/bpf_fs_kfuncs.c | 4 ----
>  1 file changed, 4 deletions(-)
>
> diff --git a/fs/bpf_fs_kfuncs.c b/fs/bpf_fs_kfuncs.c
> index 6cb8772679782..357a379ef92a5 100644
> --- a/fs/bpf_fs_kfuncs.c
> +++ b/fs/bpf_fs_kfuncs.c
> @@ -472,10 +472,6 @@ BTF_ID(func, bpf_lsm_inode_rmdir)
>  BTF_ID(func, bpf_lsm_inode_setattr)
>  BTF_ID(func, bpf_lsm_inode_setxattr)
>  BTF_ID(func, bpf_lsm_inode_unlink)
> -#ifdef CONFIG_SECURITY_PATH
> -BTF_ID(func, bpf_lsm_path_unlink)
> -BTF_ID(func, bpf_lsm_path_rmdir)
> -#endif /* CONFIG_SECURITY_PATH */
>  BTF_SET_END(d_inode_locked_hooks)

Christian,
this looks correct. Want to take it via vfs?