Received-SPF: pass (google.com: domain of selinux-refpolicy-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Russell Coker <russell@coker.com.au>
To:     Dominick Grift <dominick.grift@defensec.nl>
Cc:     selinux-refpolicy@vger.kernel.org
Subject: Re: sddm issue and patch not for inclusion
Date:   Fri, 29 Jan 2021 01:00:26 +1100
Message-ID: <1944791.8TnJqL1ZiL@liv>
In-Reply-To: <ypjlzh0ti98s.fsf@defensec.nl>
References: <YBKffiAQlZZZbCix@xev> <ypjl4kj1jny7.fsf@defensec.nl> <ypjlzh0ti98s.fsf@defensec.nl>
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="us-ascii"
Precedence: bulk

On Thursday, 28 January 2021 10:36:19 PM AEDT Dominick Grift wrote:
> >> In Debian/Unstable (which will soon be frozen and become the next stable
> >> release) the sddm X login program (the one that's generally recommended
> >> and specifically known to generally work well with SE Linux) uses PAM to
> >> start a session for the "greeter" (the program that asks for a password
> >> before a new session is started).
> >> 
> >> With the policy currently in Debian that means the sddm user matches
> >> "__default__" and gets unconfined_u:unconfined_r:unconfined_t, not what
> >> is desirable for a program that takes input from unauthenticated users.
> >> 
> >> role xdm_r;
> >> role xdm_r types xdm_t;
> >> allow system_r xdm_r;
> >> allow xdm_t xdm_tmpfs_t:file execmod;
> > 
> > that looks like a bug or at least bad code

That's a design decision.  One could make a convincing argument that it's a 
good decision.

> >> corecmd_bin_entry_type(xdm_t)
> 
> Also wondering what or which bin_t file or files this applies to and if
> it instead is not possible to give these a private type

/usr/bin/sddm-greeter.  Yes I can give it a private type, might be a good idea 
in any case.

-- 
My Main Blog         http://etbe.coker.com.au/
My Documents Blog    http://doc.coker.com.au/