Received-SPF: pass (google.com: domain of linux-kernel+bounces-172148-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Message-ID: <cc11e15bdbfbb36415b36a796387e130b764fd6a.camel@linux.ibm.com>
Subject: Re: [RFC][PATCH] ima: Use sequence number to wait for policy updates
From: Mimi Zohar <zohar@linux.ibm.com>
To: Roberto Sassu <roberto.sassu@huaweicloud.com>, dmitry.kasatkin@gmail.com,
        eric.snowberg@oracle.com, paul@paul-moore.com, jmorris@namei.org,
        serge@hallyn.com, john.johansen@canonical.com,
        stephen.smalley.work@gmail.com, casey@schaufler-ca.com,
        eparis@redhat.com
Cc: linux-integrity@vger.kernel.org, linux-security-module@vger.kernel.org,
        linux-kernel@vger.kernel.org, guozihua@huawei.com, omosnace@redhat.com,
        audit@vger.kernel.org, apparmor@lists.ubuntu.com,
        selinux@vger.kernel.org, Roberto Sassu <roberto.sassu@huawei.com>
Date: Tue, 07 May 2024 16:27:33 -0400
In-Reply-To: <e6e2b31983610bf566d8046edbac0e00c8453233.camel@huaweicloud.com>
References: <20240507092831.3590793-1-roberto.sassu@huaweicloud.com>
	 <e6e2b31983610bf566d8046edbac0e00c8453233.camel@huaweicloud.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit

Hi Roberto,

On Tue, 2024-05-07 at 11:32 +0200, Roberto Sassu wrote:
> On Tue, 2024-05-07 at 11:28 +0200, Roberto Sassu wrote:
> > From: Roberto Sassu <roberto.sassu@huawei.com>
> > 
> > Maintain a global sequence number, and set it to individual policy rules,
> > when they are created.
> 
> Just did an attempt, to see if this path is viable.
> 
> This patch would be an alternative to:
> 
> [PATCH v3] ima: Avoid blocking in RCU read-side critical section

Stephen had said,
   "Sidebar: the refactoring of the SELinux policy loading logic may have
   made it possible to revisit the approaches here to permit holding a
   reference to the policy from which the rule was derived so that we
   don't have to return -ESTALE in this scenario."

Removing -ESTALE would be the best solution.  We could then remove the -ESTALE
detection.

I assume the change would be in selinux_policy_commit(). Instead of freeing the
old policy, define and increment a per policy reference count for each
registered notifier callback.

        /* Free the old policy */
        synchronize_rcu();
        selinux_policy_free(oldpolicy);
        kfree(load_state->convert_data);

        /* Notify others of the policy change */
        selinux_notify_policy_change(seqno);

Mimi