Re: [RFC][PATCH] overlayfs: Redirect xattr ops on security.evm to security.evm_overlayfs

[Seth Forshee <sforshee@kernel.org>]

On Mon, Dec 11, 2023 at 04:41:46PM +0100, Roberto Sassu wrote:
> On Mon, 2023-12-11 at 09:36 -0600, Seth Forshee wrote:
> > On Mon, Dec 11, 2023 at 03:56:06PM +0100, Roberto Sassu wrote:
> > > Ok, I will try.
> > > 
> > > I explain first how EVM works in general, and then why EVM does not
> > > work with overlayfs.
> > > 
> > > EVM gets called before there is a set/removexattr operation, and after,
> > > if that operation is successful. Before the set/removexattr operation
> > > EVM calculates the HMAC on current inode metadata (i_ino, i_generation,
> > > i_uid, i_gid, i_mode, POSIX ACLs, protected xattrs). Finally, it
> > > compares the calculated HMAC with the one in security.evm.
> > > 
> > > If the verification and the set/removexattr operation are successful,
> > > EVM calculates again the HMAC (in the post hooks) based on the updated
> > > inode metadata, and sets security.evm with the new HMAC.
> > > 
> > > The problem is the combination of: overlayfs inodes have different
> > > metadata than the lower/upper inodes; overlayfs calls the VFS to
> > > set/remove xattrs.
> > 
> > I don't know all of the inner workings of overlayfs in detail, but is it
> > not true that whatever metadata an overlayfs mount presents for a given
> > inode is stored in the lower and/or upper filesystem inodes? If the
> > metadata for those inodes is verified with EVM, why is it also necessary
> > to verify the metadata at the overlayfs level? If some overlayfs
> > metadata is currently omitted from the checks on the lower/upper inodes,
> > is there any reason EVM couldn't start including that its checksums?
> 
> Currently, the metadata where there is a misalignment are:
> i_generation, s_uuid, (i_ino?). Maybe there is more?
> 
> If metadata are aligned, there is no need to store two separate HMACs.

I can only think of three possible sources for the metadata overlayfs
presents:

 1. It comes directly from the underlying filesystems
 2. overlayfs synthesizes if from the underlying filesystem data
 3. It's purely generated at runtime

Are there others?

1 and 2 should be covered by EVM on the underlying filesystems. If 3 is
happening then it seems like hashing that data is just confirming that
overlayfs consistently generates the same values for that data, and
verifying code behavior doesn't seem in-scope for EVM.