linux-kernel - Re: [PATCH v4 3/5] security: Allow all LSMs to provide xattrs for inode_init

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 21 Nov 2022 14:29:20 +0100
From:   Roberto Sassu <roberto.sassu@...weicloud.com>
To:     Casey Schaufler <casey@...aufler-ca.com>,
        Mimi Zohar <zohar@...ux.ibm.com>, dmitry.kasatkin@...il.com,
        paul@...l-moore.com, jmorris@...ei.org, serge@...lyn.com,
        stephen.smalley.work@...il.com, eparis@...isplace.org
Cc:     linux-integrity@...r.kernel.org,
        linux-security-module@...r.kernel.org, selinux@...r.kernel.org,
        reiserfs-devel@...r.kernel.org, linux-kernel@...r.kernel.org,
        keescook@...omium.org, nicolas.bouchinet@...p-os.org,
        Roberto Sassu <roberto.sassu@...wei.com>
Subject: Re: [PATCH v4 3/5] security: Allow all LSMs to provide xattrs for
 inode_init_security hook

On Fri, 2022-11-18 at 09:31 -0800, Casey Schaufler wrote:
> On 11/18/2022 7:10 AM, Mimi Zohar wrote:
> > On Fri, 2022-11-18 at 10:14 +0100, Roberto Sassu wrote:
> > > > > +static int security_check_compact_xattrs(struct xattr *xattrs,
> > > > > +                                     int num_xattrs, int *checked_xattrs)
> > > > Perhaps the variable naming is off, making it difficult to read.   So
> > > > although this is a static function, which normally doesn't require a
> > > > comment, it's definitely needs one.
> > > Ok, will improve it.
> > > 
> > > > > +{
> > > > > +    int i;
> > > > > +
> > > > > +    for (i = *checked_xattrs; i < num_xattrs; i++) {
> > > > If the number of "checked" xattrs was kept up to date, removing the
> > > > empty xattr gaps wouldn't require a loop.  Is the purpose of this loop
> > > > to support multiple per LSM xattrs?
> > > An LSM might reserve one or more xattrs, but not set it/them (for 
> > > example because it is not initialized). In this case, removing the gaps 
> > > is needed for all subsequent LSMs.
> > Including this sort of info in the function description or as a comment
> > in the code would definitely simplify review.
> > 
> > security_check_compact_xattrs() is called in the loop after getting
> > each LSM's xattr(s).  Only the current LSMs xattrs need to be
> > compressed, yet the loop goes to the maximum number of xattrs each
> > time. Just wondering if there is a way of improving it.
> 
> At security module registration each module could identify how
> many xattrs it uses. That number could be used to limit the range
> of the loop. I have to do similar things for the forthcoming LSM
> syscalls and module stacking beyond that.

Yes, blob_sizes.lbs_xattr contains the total number of xattrs requested
by LSMs. To stop the loop earlier, at the offset of the next LSM, we
would need to search the LSM's lsm_info, using the LSM name in
the security_hook_list structure. Although it is not optimal, not doing
it makes the code simpler. I could do that, if preferred.

Roberto