linux-kernel - Re: [PATCH 1/9] Unionfs: security convert lsm into a static interface fix

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, 22 Oct 2007 20:48:04 -0400
From:	Erez Zadok <ezk@...sunysb.edu>
To:	Christoph Hellwig <hch@...radead.org>
Cc:	Erez Zadok <ezk@...sunysb.edu>, akpm@...ux-foundation.org,
	linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org,
	viro@....linux.org.uk, "Serge E. Hallyn" <serue@...ibm.com>,
	Arjan van de Ven <arjan@...radead.org>,
	Chris Wright <chrisw@...s-sol.org>,
	James Morris <jmorris@...ei.org>,
	Stephen Smalley <sds@...ho.nsa.gov>,
	"Josef 'Jeff' Sipek" <jsipek@...sunysb.edu>
Subject: Re: [PATCH 1/9] Unionfs: security convert lsm into a static interface fix 

In message <20071022082231.GA15132@...radead.org>, Christoph Hellwig writes:
> On Sun, Oct 21, 2007 at 07:51:14PM -0400, Erez Zadok wrote:
> > From: Andrew Morton <akpm@...ux-foundation.org>
> > 
> > ERROR: "security_inode_permission" [fs/unionfs/unionfs.ko] undefined!
> > ERROR: "security_file_ioctl" [fs/unionfs/unionfs.ko] undefined!
> > 
> > Need these back.
> 
> These should never used by modules.

Why?  Are you concerned that the security policy may change after a module
is loaded?  My understanding of the security code is that it should handle
this, even if people call security_*() functions directly.  When I look at
the security_* functions in security.c, to me they very much smell like
global wrappers that others can call, b/c they refer to private/global ops
vectors that one should not be referencing directly.  For example:

int security_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
	return security_ops->file_ioctl(file, cmd, arg);
}

> You'll need to use and/or introduce
> vfs_ helpers that do proper security checks before calling the methods.

I can probably get rid of having unionfs call security_inode_permission, by
calling permission() myself and carefully post-process its return code
(unionfs needs to "ignore" EROFS initially, to allow copyup to take place).

But security_file_ioctl doesn't have any existing helper I can call.  I can
introduce a trivial vfs_security_file_ioctl wrapper to security_file_ioctl,
but what about the already existing *19* exported security_* functions in
security/security.c?  Do you want to see simple wrappers for all of them?
It seems redundant to add a one-line wrapper around an already one-line
function around security_ops->XXX.  Plus, some of the existing exported
security_* functions are file-system related, others are networking, etc.  So
we'll need wrappers whose names are prefixed appropriately: vfs_*, net_*,
etc.

Thanks,
Erez.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/