linux-kernel - Re: [PATCH 3/4] x86,module: Detect VMX vs SLD conflicts

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: <202004071611.233B0045@keescook>
Date:   Tue, 7 Apr 2020 16:15:12 -0700
From:   Kees Cook <keescook@...omium.org>
To:     David Laight <David.Laight@...LAB.COM>
Cc:     'Peter Zijlstra' <peterz@...radead.org>,
        "tglx@...utronix.de" <tglx@...utronix.de>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "hch@...radead.org" <hch@...radead.org>,
        "sean.j.christopherson@...el.com" <sean.j.christopherson@...el.com>,
        "mingo@...hat.com" <mingo@...hat.com>,
        "bp@...en8.de" <bp@...en8.de>, "hpa@...or.com" <hpa@...or.com>,
        "x86@...nel.org" <x86@...nel.org>,
        "kenny@...ix.com" <kenny@...ix.com>,
        "jeyu@...nel.org" <jeyu@...nel.org>,
        "rasmus.villemoes@...vas.dk" <rasmus.villemoes@...vas.dk>,
        "pbonzini@...hat.com" <pbonzini@...hat.com>,
        "fenghua.yu@...el.com" <fenghua.yu@...el.com>,
        "xiaoyao.li@...el.com" <xiaoyao.li@...el.com>,
        "nadav.amit@...il.com" <nadav.amit@...il.com>,
        "thellstrom@...are.com" <thellstrom@...are.com>,
        "tony.luck@...el.com" <tony.luck@...el.com>,
        "rostedt@...dmis.org" <rostedt@...dmis.org>,
        "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
        "jannh@...gle.com" <jannh@...gle.com>,
        "dcovelli@...are.com" <dcovelli@...are.com>,
        "mhiramat@...nel.org" <mhiramat@...nel.org>
Subject: Re: [PATCH 3/4] x86,module: Detect VMX vs SLD conflicts

On Tue, Apr 07, 2020 at 09:25:56PM +0000, David Laight wrote:
> From: Peter Zijlstra
> > Sent: 07 April 2020 12:03
> > 
> > It turns out that with Split-Lock-Detect enabled (default) any VMX
> > hypervisor needs at least a little modification in order to not blindly
> > inject the #AC into the guest without the guest being ready for it.
> > 
> > Since there is no telling which module implements a hypervisor, scan
> > all out-of-tree modules' text and look for VMX instructions and refuse
> > to load it when SLD is enabled (default) and the module isn't marked
> > 'sld_safe'.
> ...
> > +	while (text < text_end) {
> > +		kernel_insn_init(&insn, text, text_end - text);
> > +		insn_get_length(&insn);
> > +
> > +		if (WARN_ON_ONCE(!insn_complete(&insn))) {
> > +			pr_err("Module text malformed: %s\n", mod->name);
> > +			return -ENOEXEC;
> > +		}
> > +
> > +		if (!allow_vmx && insn_is_vmx(&insn)) {
> > +			pr_err("Module has VMX instructions and is not marked 'sld_safe', boot with:
> > 'split_lock_detect=off': %s\n", mod->name);
> > +			return -ENOEXEC;
> > +		}
> > +
> > +		text += insn.length;
> > +	}
> 
> There is a slight flaw in the above.
> A malicious module can hide the required instruction by jumping into the
> middle of a long instruction.
> 
> Even checking branch targets hit instruction barriers isn't enough,
> an indirect jump could be used.

If I understand the goals here, it's to provide feedback for good actors
doing things that they don't realize aren't safe. Trying to stop a
malicious module from doing malicious things is basically impossible:
it can just load a data blob and self-modify, etc. :)

Though, Peter, this does get me thinking: if this is meant to be helpful
for module authors tripping over things they shouldn't be touching,
perhaps every test needs to include explicit recommendations? It's
_kind_ of doing this already. Maybe the above should be:

pr_err("%s: contains VMX instructions but is not marked 'sld_safe'. Please see <url> or boot with: 'split_lock_detect=off' to ignore.\n", mod->name);

?

-- 
Kees Cook