linux-kernel - Re: [PATCH] x86_64: test that userspace stack is in fact NX

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: <f972d59c-40dd-2a68-ff13-a2658513a25b@intel.com>
Date:   Tue, 3 Oct 2023 07:23:56 -0700
From:   Dave Hansen <dave.hansen@...el.com>
To:     Alexey Dobriyan <adobriyan@...il.com>
Cc:     Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        linux-kernel@...r.kernel.org, "H. Peter Anvin" <hpa@...or.com>
Subject: Re: [PATCH] x86_64: test that userspace stack is in fact NX

On 10/3/23 06:00, Alexey Dobriyan wrote:
> On Mon, Oct 02, 2023 at 07:23:10AM -0700, Dave Hansen wrote:
>> Basically, could you spend a moment in the changelog to talk about:
>>
>> 1. 32-bit kernels on NX hardware
>> and
>> 2. 64-bit kernels on non-NX hardware
> 
> Sure. My logic whas that i386 is dead arch, but this test is easy to
> port to i386, only 2 simple functions.

I honestly don't feel strongly about it one way or the other.  But
whatever we do, let's explain it, please.

> I don't want to parse /proc/cpuinfo. If someone knows they're shipping
> NX-incapable hardware, just let them disable the test.

Other than clearcpuid=nx, I don't _think_ we have any way to clear the
X86_FEATURE_NX bit right now.  That should mean that you can use regular
old CPUID to see if the booted kernel supports NX.  Perhaps something
like what:

	tools/testing/selftests/x86/amx.c

does with CPUID_LEAF1_ECX_XSAVE_MASK.  That should be quite a bit easier
than parsing /proc/cpuinfo.

If someone does use clearcpuid, then I think it's perfectly reasonable
to fail the selftest.