Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 14 Jul 2022 20:11:29 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Josh Poimboeuf <jpoimboe@kernel.org>
Cc:     Thadeu Lima de Souza Cascardo <cascardo@canonical.com>,
        Pawan Gupta <pawan.kumar.gupta@linux.intel.com>,
        Thomas Gleixner <tglx@linutronix.de>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        Dave Hansen <dave.hansen@linux.intel.com>, x86@kernel.org,
        "H. Peter Anvin" <hpa@zytor.com>, linux-kernel@vger.kernel.org,
        stable@vger.kernel.org,
        Daniel Sneddon <daniel.sneddon@linux.intel.com>,
        antonio.gomez.iglesias@linux.intel.com
Subject: Re: [PATCH] x86/bugs: Switch to "auto" when "ibrs" selected on
 Enhanced IBRS parts
Message-ID: <YtBcUWIqJjUdWtro@worktop.programming.kicks-ass.net>
References: <0456b35fb9ef957d9a9138e0913fb1a3fd445dff.1657747493.git.pawan.kumar.gupta@linux.intel.com>
 <Ys/7RiC9Z++38tzq@quatroqueijos>
 <YtABEwRnWrJyIKTY@worktop.programming.kicks-ass.net>
 <20220714160106.c6efowo6ptsu72ne@treble>
 <YtBMZMaOnA8g8m0a@worktop.programming.kicks-ass.net>
 <20220714173814.p5kdyimu6ho7zjt5@treble>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220714173814.p5kdyimu6ho7zjt5@treble>
Precedence: bulk

On Thu, Jul 14, 2022 at 10:38:14AM -0700, Josh Poimboeuf wrote:
> On Thu, Jul 14, 2022 at 07:03:32PM +0200, Peter Zijlstra wrote:
> > On Thu, Jul 14, 2022 at 09:01:06AM -0700, Josh Poimboeuf wrote:
> > 
> > > > Yeah this; if the user asks for IBRS, we should give him IBRS. I hate
> > > > the 'I know better, let me change that for you' mentality.
> > > 
> > > eIBRS CPUs don't even have legacy IBRS so I don't see how this is even
> > > possible.
> > 
> > You can still WRMSR a lot on them. Might not make sense but it 'works'.
> 
> Even in Intel documentation, eIBRS is often referred to as IBRS. It
> wouldn't be surprising for a user to consider spectre_v2=ibrs to mean
> "use eIBRS".
> 
> I'm pretty sure there's nobody out there that wants spectre_v2=ibrs to
> mean "make it slower and possibly less secure because it's being used
> contrary to the spec".

Then make it print a big honking warning.

Most people will either use auto or off, the very few people that force
an option get what they ask for, not something else.

Like said upthread, it allows testing the code-paths at the very least.