linux-kernel - Re: [PATCH] x86/mm: Remove set_pages_x() and set_pages

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]

Message-ID: <bd5684ee-afaf-a9ba-6c32-15d2aa94733c@lwfinger.net>
Date:   Wed, 18 Sep 2019 13:17:13 -0500
From:   Larry Finger <Larry.Finger@...inger.net>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Christoph Hellwig <hch@....de>,
        Linux List Kernel Mailing <linux-kernel@...r.kernel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...nel.org>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Subject: Re: [PATCH] x86/mm: Remove set_pages_x() and set_pages_nx()

On 9/18/19 12:53 PM, Linus Torvalds wrote:
> On Wed, Sep 18, 2019 at 10:50 AM Larry Finger <Larry.Finger@...inger.net> wrote:
>>
>> Is there approved way for pages to be set to be executable by an external module
>> that would not be a security issue?
> 
> Point to what external module and why.
> 
> Honestly, the likely answer is simply "no". Why would an external
> module ever need to make something executable that isn't read-only
> code? That's pretty fundamental. Marking data executable is fairly
> questionable these days.
> 
> Instead, what might work is to have some higher-level concept that we
> actually trust, and that isn't about making data executable, but about
> doing something reasonable.
> 
> See the difference? Making things executable is not ok, but perhaps a
> "alternative runtime code sequence" is ok.
> 
>                  Linus


Linus,

Yes, I do see the difference.

The external module is vboxdrv, which is part of VirtualBox. The setting of 
pages to be executable appears to have been added in kernel 2.4.20.

I am now testing with the former calls to set_pages_x() and set_pages_nx() 
disabled. Thus far, VMs seem to be running OK. I will contact Oracle to discuss 
the matter with them and see if there is some special case that requires this 
facility. If there is one, then they will need to discuss it with you and Christoph.

Larry