linux-kernel - Re: [RFC][PATCH 06/17] x86/cpu: Add SRSO untrain to retbleed=

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: <20230812112404.GB749618@hirez.programming.kicks-ass.net>
Date:   Sat, 12 Aug 2023 13:24:04 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Josh Poimboeuf <jpoimboe@...nel.org>
Cc:     Borislav Petkov <bp@...en8.de>, x86@...nel.org,
        linux-kernel@...r.kernel.org, David.Kaplan@....com,
        Andrew.Cooper3@...rix.com, gregkh@...uxfoundation.org
Subject: Re: [RFC][PATCH 06/17] x86/cpu: Add SRSO untrain to retbleed=

On Thu, Aug 10, 2023 at 12:10:03PM -0400, Josh Poimboeuf wrote:
> On Thu, Aug 10, 2023 at 05:44:04PM +0200, Borislav Petkov wrote:
> > On Wed, Aug 09, 2023 at 09:12:24AM +0200, Peter Zijlstra wrote:
> > > Since it is now readily apparent that the two SRSO
> > > untrain_ret+return_thunk variants are exactly the same mechanism as
> > > the existing (retbleed) zen untrain_ret+return_thunk, add them to the
> > > existing retbleed options.
> > 
> > Except that Zen3/4 are not affected by retbleed, according to my current
> > state of information.
> > 
> > I don't like this lumping together of the issues even if their
> > mitigations are more or less similar.
> 
> I tend to agree that SRSO is a new issue and should have its own sysfs
> and cmdline options (though a separate CONFIG option is overkill IMO).
> 
> The mitigations are unfortunately intertwined, but we've been in that
> situation several times before (e.g., spectre_v2 + intel retbleed).

That very experience wants me to avoid doing it again :-/ But you all
really want to keep the parameter, can we at least rename it something
you can remember how to type, like 'srso=' instead of this horrific
'spec_rstack_overflow=' thing?