linux-kernel - Re: [PATCH 4/5] Make LSM Writable Hooks a command line option

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:   Tue, 6 Jun 2017 15:11:58 +0300
From:   Igor Stoppa <igor.stoppa@...wei.com>
To:     Tetsuo Handa <penguin-kernel@...ove.SAKURA.ne.jp>,
        <casey@...aufler-ca.com>, <keescook@...omium.org>,
        <mhocko@...nel.org>, <jmorris@...ei.org>
CC:     <paul@...l-moore.com>, <sds@...ho.nsa.gov>, <hch@...radead.org>,
        <labbott@...hat.com>, <linux-mm@...ck.org>,
        <linux-kernel@...r.kernel.org>,
        <kernel-hardening@...ts.openwall.com>
Subject: Re: [PATCH 4/5] Make LSM Writable Hooks a command line option

On 06/06/17 14:42, Tetsuo Handa wrote:
> Igor Stoppa wrote:
>> Who decides when enough is enough, meaning that all the needed modules
>> are loaded?
>> Should I provide an interface to user-space? A sysfs entry?
> 
> No such interface is needed. Just an API for applying set_memory_rw()
> and set_memory_ro() on LSM hooks is enough.
> 
> security_add_hooks() can call set_memory_rw() before adding hooks and
> call set_memory_ro() after adding hooks. Ditto for security_delete_hooks()
> for SELinux's unregistration.

I think this should be considered part of the 2nd phase "write seldom",
as we agreed with Kees Cook.

Right now the goal was to provide the basic API for:
- create pool
- get memory from pool
- lock the pool
- destroy the pool

And, behind the scene, verify that a memory range falls into Pmalloc pages.

Then would come the "write seldom" part.

The reason for this is that a proper implementation of write seldom
should, imho, make writable only those pages that really need to be
modified. Possibly also add some verification on the call stack about
who is requesting the unlocking.

Therefore I would feel more comfortable in splitting the work into 2 part.

For the case at hand, would it work if there was a non-API call that you
could use until the API is properly expanded?

--
igor