linux-kernel - Re: [PATCH] LSM: Allow syzbot to ignore security= parameter.

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] [day] [month] [year] [list]

Message-ID: <0dd46edc-d3ee-75ca-5e50-fee70bd83ddb@i-love.sakura.ne.jp>
Date:   Sat, 9 Feb 2019 10:40:17 +0900
From:   Tetsuo Handa <penguin-kernel@...ove.sakura.ne.jp>
To:     Casey Schaufler <casey@...aufler-ca.com>,
        Kees Cook <keescook@...omium.org>
Cc:     Dmitry Vyukov <dvyukov@...gle.com>,
        Paul Moore <paul@...l-moore.com>,
        Stephen Smalley <sds@...ho.nsa.gov>,
        syzbot <syzbot+21016130b0580a9de3b5@...kaller.appspotmail.com>,
        tyhicks@...onical.com, John Johansen <john.johansen@...onical.com>,
        James Morris <jmorris@...ei.org>,
        LKML <linux-kernel@...r.kernel.org>,
        linux-security-module@...r.kernel.org,
        Serge Hallyn <serge@...lyn.com>,
        syzkaller-bugs <syzkaller-bugs@...glegroups.com>,
        Jeffrey Vander Stoep <jeffv@...gle.com>,
        SELinux <selinux@...r.kernel.org>,
        Russell Coker <russell@...er.com.au>,
        Laurent Bigonville <bigon@...ian.org>,
        syzkaller <syzkaller@...glegroups.com>,
        Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [PATCH] LSM: Allow syzbot to ignore security= parameter.

On 2019/02/09 9:28, Tetsuo Handa wrote:
> On 2019/02/09 1:23, Casey Schaufler wrote:
>> On 2/8/2019 2:52 AM, Tetsuo Handa wrote:
>>> To help administrators easily understand what LSM modules are possibly enabled by default (which
>>> have to be fetched from e.g. /boot/config-`uname -r`)
>>
>> $ cat /sys/kernel/security/lsm
>>
> 
> /sys/kernel/security/lsm is list of "actually" enabled modules, isn't it?
> What I want is "possibly" enabled modules. Ubuntu would chose from either
> 
>   (a) explicitly add security=apparmor to kernel command line
> 
> or
> 
>   (b) explicitly remove tomoyo from CONFIG_LSM at kernel config
> 
> in order not to enable TOMOYO for those who want to enable only one of 
> SELinux/Smack/AppArmor. And for those who want to enable TOMOYO, I think
> that (b) (in other words, add
> 
>   lsm="modules listed in CONFIG_LSM" + ",tomoyo"
> 
> ) will retain compatibility when it becomes possible to enable more than
> one of SELinux/Smack/AppArmor at the same time.
> 
> If we can know "possibly" enabled modules from dmesg, users don't need to
> look at e.g. /boot/config-`uname -r`. It is not essential, but it's handy.
> 

Well, thinking again, specifying

lsm="modules listed in /sys/kernel/security/lsm" + ",tomoyo"

makes sense, for there is no need to care about disabled modules when
enabling TOMOYO. Therefore,

+	pr_info("Security Framework initializing: %s\n", order);
-	pr_info("Security Framework initializing\n");

won't be needed.

On 2019/02/09 6:33, Kees Cook wrote:
> On Thu, Feb 7, 2019 at 8:24 AM Casey Schaufler <casey@...aufler-ca.com> wrote:
>> I added Kees to the CC list. Kees, what to you think about
>> ignoring security= if lsm= is specified? I'm ambivalent.
> 
> This was one of many earlier suggestions, and the consensus seemed to
> be "don't mix security= and lsm=". Why would anyone use both?
> 

Then, can we add this change?

+	if (chosen_lsm_order) {
+		if (chosen_major_lsm) {
+			pr_info("security= is ignored because of lsm=\n");
+			chosen_major_lsm = NULL;
+		}
+	}