linux-kernel - Re: [PATCH RFC v2 4/6] x86: Disable PTI on compatibility mode

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:   Fri, 16 Feb 2018 16:25:33 +0000
From:   Dmitry Safonov <0x7f454c46@...il.com>
To:     Andy Lutomirski <luto@...nel.org>
Cc:     Nadav Amit <namit@...are.com>, Ingo Molnar <mingo@...hat.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Peter Zijlstra <peterz@...radead.org>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Willy Tarreau <w@....eu>, Nadav Amit <nadav.amit@...il.com>,
        X86 ML <x86@...nel.org>, LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH RFC v2 4/6] x86: Disable PTI on compatibility mode

2018-02-15 20:02 GMT+00:00 Andy Lutomirski <luto@...nel.org>:
> On Thu, Feb 15, 2018 at 4:36 PM, Nadav Amit <namit@...are.com> wrote:
>> Based on the understanding that there should be no way for userspace to
>> address the kernel-space from compatibility mode, disable it while
>> running in compatibility mode as long as the 64-bit code segment of the
>> user is not used.
>>
>> Reenabling PTI is performed by restoring NX-bits to the userspace
>> mappings, flushing the TLBs, and notifying all the CPUs that use the
>> affected mm to disable PTI. Each core responds by removing the present
>> bit for the 64-bit code-segment, and marking that PTI is disabled on
>> that core.
>>
>
> I dislike this patch because it's conflating two things.  The patch
> claims to merely disable PTI for compat tasks, whatever those are.
> But it's also introducing a much stronger concept of what a compat
> task is.  The kernel currently mostly doesn't care whether a task is
> "compat" or not, and I think that most remaining code paths that do
> care are buggy and should be removed.

Yes, please, don't do a stronger concept..
Speaking from CRIU side, it C/R ia32 tasks with x86_64 binaries.

-- 
             Dmitry