From: ebiederm@xmission.com (Eric W. Biederman)
To: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Dmitry Kasatkin <dmitry.kasatkin@gmail.com>,
        Oleg Nesterov <oleg@redhat.com>,
        Dmitry Kasatkin <d.kasatkin@samsung.com>,
        linux-security-module <linux-security-module@vger.kernel.org>,
        John Johansen <john.johansen@canonical.com>,
        Mimi Zohar <zohar@linux.vnet.ibm.com>,
        James Morris <jmorris@namei.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        kernel-team <kernel-team@lists.ubuntu.com>
References: <535A5C78.1070901@samsung.com> <535A75C1.3050901@samsung.com>
	<20140425182310.GA9128@redhat.com>
	<87sip15iy5.fsf@x220.int.ebiederm.org>
	<20140425192543.GA11908@redhat.com>
	<878uqt42q7.fsf@x220.int.ebiederm.org>
	<20140425200156.GA13727@redhat.com>
	<CACE9dm9gs64C9JfSYKboUOwT_uXC7Ha1kuXsMWOE-0isQuzfmg@mail.gmail.com>
	<874n1h16le.fsf@x220.int.ebiederm.org>
	<20140425212128.GB18016@ZenIV.linux.org.uk>
Date: Fri, 25 Apr 2014 14:43:42 -0700
In-Reply-To: <20140425212128.GB18016@ZenIV.linux.org.uk> (Al Viro's message of
	"Fri, 25 Apr 2014 22:21:28 +0100")
Message-ID: <87ppk5w0dt.fsf@x220.int.ebiederm.org>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/24.3 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain
Subject: Re: Kernel panic at Ubuntu: IMA + Apparmor
Sender: linux-kernel-owner@vger.kernel.org

Al Viro <viro@ZenIV.linux.org.uk> writes:

> On Fri, Apr 25, 2014 at 01:45:17PM -0700, Eric W. Biederman wrote:
>
>> IMA-appraisal is fundamentally broken because I can take a mandatory
>> file lock and prevent IMA-apprasial.
>> 
>> Using kernel_read is what allows this.
>> 
>> > Isn't it a clear motivating case???
>> 
>> kernel_read is not appropriate for IMA use.  The rest of this is just
>> the messenger.
>> 
>> IMA needs to use a cousin of kernel_read that operates at a lower level
>> than vfs_read.  A function that all of the permission checks and the
>> fsnotify work.
>
> It's worse than that, actually ;-/  IMA hooks in __fput() have interesting
> interplay with revoke-related stuff as well.  Another very messy thing in
> the same area is that it actually does ->read() from under ->i_mutex, leading
> to all kinds of interesting locking issues...
>
> I doubt that your "let's open-code vfs_read() guts" would be a good idea;
> if nothing else, it might make more sense to make rw_verify_area() skip
> the mandlock and security theatre when called in such situation.
>
> What a mess... ;-/

Agreed.

All I really meant is that vfs_read does too much, so it probably needs
to be refactored for this case.  But fsnotify_read, add_rchar, and
inc_syscr all seem inappropriate.  

So I think we might be able to get away with something like this:

ssize_t __vfs_read(struct file *file, char __user *buf, size_t count, loff_t *pos)
{
	ssize_t ret;

	if (!(file->f_mode & FMODE_READ))
		return -EBADF;
	if (!file->f_op->read && !file->f_op->aio_read)
		return -EINVAL;
	if (unlikely(!access_ok(VERIFY_WRITE, buf, count)))
		return -EFAULT;

	if (ret >= 0) {
		count = ret;
		if (file->f_op->read)
			ret = file->f_op->read(file, buf, count, pos);
		else
			ret = do_sync_read(file, buf, count, pos);
	}

	return ret;
}


How much of the rest we do really would seem to depend on how valuable
the sanity checks are.

This area of code keeps evolving enough that I don't see how we could
possibly avoid going through helper functions to figure out which file
ops we want to use this week.

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