linux-kernel - Re: [PATCH v3 2/4] x86/traps: Print non-canonical address on #GP

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20191120162143.GB32572@linux.intel.com>
Date:   Wed, 20 Nov 2019 08:21:43 -0800
From:   Sean Christopherson <sean.j.christopherson@...el.com>
To:     Borislav Petkov <bp@...en8.de>
Cc:     Ingo Molnar <mingo@...nel.org>, Jann Horn <jannh@...gle.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>,
        "H. Peter Anvin" <hpa@...or.com>,
        the arch/x86 maintainers <x86@...nel.org>,
        Andrey Ryabinin <aryabinin@...tuozzo.com>,
        Alexander Potapenko <glider@...gle.com>,
        Dmitry Vyukov <dvyukov@...gle.com>,
        kasan-dev <kasan-dev@...glegroups.com>,
        kernel list <linux-kernel@...r.kernel.org>,
        Andrey Konovalov <andreyknvl@...gle.com>,
        Andy Lutomirski <luto@...nel.org>,
        Andi Kleen <ak@...ux.intel.com>
Subject: Re: [PATCH v3 2/4] x86/traps: Print non-canonical address on #GP

On Wed, Nov 20, 2019 at 02:39:13PM +0100, Borislav Petkov wrote:
> On Wed, Nov 20, 2019 at 02:28:30PM +0100, Ingo Molnar wrote:
> > I'd rather we not trust the decoder and the execution environment so much 
> > that it never produces a 0 linear address in a #GP:
> 
> I was just scratching my head whether I could trigger a #GP with address
> of 0. But yeah, I agree, let's be really cautious here. I wouldn't want
> to debug a #GP with a wrong address reported.

It's definitely possible, there are a handful of non-SIMD instructions that
generate #GP(0) it CPL=0 in 64-bit mode *and* have a memory operand.  Some
of them might even be legitimately encountered in the wild.

  - CMPXCHG16B if it's not supported by the CPU.
  - VMXON if CR4 is misconfigured or VMX isn't enabled in FEATURE_CONTROL.
  - MONITOR if ECX has an invalid hint (although MONITOR hardcodes the
    address in DS:RAX and so doesn't have a ModR/M byte).

Undoudbtedly there are other instructions with similar sources of #GP.