Re: [PATCH 1/2] security: Handle dentries without inode in security_path_post_mknod()

[Paul Moore <paul@paul-moore.com>]

On Mar 29, 2024 Roberto Sassu <roberto.sassu@huaweicloud.com> wrote:
> 
> Commit 08abce60d63fi ("security: Introduce path_post_mknod hook")
> introduced security_path_post_mknod(), to replace the IMA-specific call to
> ima_post_path_mknod().
> 
> For symmetry with security_path_mknod(), security_path_post_mknod() is
> called after a successful mknod operation, for any file type, rather than
> only for regular files at the time there was the IMA call.
> 
> However, as reported by VFS maintainers, successful mknod operation does
> not mean that the dentry always has an inode attached to it (for example,
> not for FIFOs on a SAMBA mount).
> 
> If that condition happens, the kernel crashes when
> security_path_post_mknod() attempts to verify if the inode associated to
> the dentry is private.
> 
> Add an extra check to first verify if there is an inode attached to the
> dentry, before checking if the inode is private. Also add the same check to
> the current users of the path_post_mknod hook, ima_post_path_mknod() and
> evm_post_path_mknod().
> 
> Finally, use the proper helper, d_backing_inode(), to retrieve the inode
> from the dentry in ima_post_path_mknod().
> 
> Cc: stable@vger.kernel.org # 6.8.x
> Reported-by: Steve French <smfrench@gmail.com>
> Closes: https://lore.kernel.org/linux-kernel/CAH2r5msAVzxCUHHG8VKrMPUKQHmBpE6K9_vjhgDa1uAvwx4ppw@mail.gmail.com/
> Fixes: 08abce60d63fi ("security: Introduce path_post_mknod hook")
> Signed-off-by: Roberto Sassu <roberto.sassu@huawei.com>
> Acked-by: Mimi Zohar <zohar@linux.ibm.com>
> ---
>  security/integrity/evm/evm_main.c | 6 ++++--
>  security/integrity/ima/ima_main.c | 5 +++--
>  security/security.c               | 4 +++-
>  3 files changed, 10 insertions(+), 5 deletions(-)

In addition to the stable marking that Mimi already pointed out, I've
got one small comment below, but otherwise this looks fine to me.
Also, just to confirm, you're going to send patch 1/2 up to Linus during
the v6.9-rc1 phase and hold patch 2/2 for the next merge window, right?

Acked-by: Paul Moore <paul@paul-moore.com>

> diff --git a/security/security.c b/security/security.c
> index 7e118858b545..455f0749e1b0 100644
> --- a/security/security.c
> +++ b/security/security.c
> @@ -1801,7 +1801,9 @@ EXPORT_SYMBOL(security_path_mknod);
>   */
>  void security_path_post_mknod(struct mnt_idmap *idmap, struct dentry *dentry)
>  {
> -	if (unlikely(IS_PRIVATE(d_backing_inode(dentry))))
> +	/* Not all dentries have an inode attached after mknod. */
> +	if (d_backing_inode(dentry) &&
> +	    unlikely(IS_PRIVATE(d_backing_inode(dentry))))
>  		return;

I don't know how much impact this would have on the compiled code, but
you could save yourself a call into d_backing_inode() by saving it to
a local variable:

  struct inode *inode = d_backing_inode(dentry);
  if (inode && unlikely(IS_PRIVATE(inode)))
    return;

>  	call_void_hook(path_post_mknod, idmap, dentry);
>  }
> -- 
> 2.34.1

--
paul-moore.com