linux-kernel - Re: [PATCH v2 1/2] x86/speculation: Allow enabling STIBP with legacy IBRS

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: <Y/ZVaBKwbWUbF7u+@zn.tnic>
Date:   Wed, 22 Feb 2023 18:48:41 +0100
From:   Borislav Petkov <bp@...en8.de>
To:     KP Singh <kpsingh@...nel.org>
Cc:     linux-kernel@...r.kernel.org, pjt@...gle.com, evn@...gle.com,
        jpoimboe@...nel.org, tglx@...utronix.de, x86@...nel.org,
        hpa@...or.com, peterz@...radead.org,
        pawan.kumar.gupta@...ux.intel.com, kim.phillips@....com,
        alexandre.chartre@...cle.com, daniel.sneddon@...ux.intel.com,
        corbet@....net, bp@...e.de, linyujun809@...wei.com,
        jmattson@...gle.com,
        José Oliveira <joseloliveira11@...il.com>,
        Rodrigo Branco <rodrigo@...nelhacking.com>,
        Alexandra Sandulescu <aesa@...gle.com>, stable@...r.kernel.org
Subject: Re: [PATCH v2 1/2] x86/speculation: Allow enabling STIBP with legacy
 IBRS

On Wed, Feb 22, 2023 at 09:16:21AM -0800, KP Singh wrote:
> Thanks for iterating. I think your commit description and rewrite
> omits a few key subtleties which I have tried to reinforce in both the
> commit log and the comments.
> 
> Q: What does STIBP have to do with IBRS?
> A: Setting the IBRS bit implicitly enables STIBP / some form of cross
> thread protection.

That belongs in the docs, if you want to explain this properly.

> Q: Why does it work with eIBRS?
> A: Because we set the IBRS bit once and leave it set when using eIBRS

Also docs.

> I think this subtlety should be reinforced in the commit description
> and code comments so that we don't get it wrong again. Your commit
> does answer this one (thanks!)

Commit messages are fine when explaining *why* a change is being done.
What is even finer is when you put a lenghtier explanation in our
documentation so that people can actually find it. Finding text in
commit messages is harder...

> Q: Why does it not work with the way the kernel currently implements
> legacy IBRS?
> A: Because the kernel clears the bit on returning to user space.

>From the commit message:

    However, on return to userspace, the IBRS bit is cleared for performance
    reasons. That leaves userspace threads vulnerable to cross-thread
    predictions influence against which STIBP protects.

> The reason why I refactored this into a separate helper was to
> document the subtleties I mentioned above and anchor them to one place
> as the function is used in 2 places. But this is a maintainer's
> choice, so it's your call :)

The less code gets added in that thing, the better. Not yet another
helper pls.
 
> I do agree with Pawan that it's worth adding a pr_info about what the
> kernel is doing about STIBP.

STIBP status gets dumped through stibp_state().

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette