Date: Wed, 10 Jan 2018 10:22:34 +0100
From: Peter Zijlstra <peterz@infradead.org>
To: Dave Hansen <dave.hansen@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>,
        LKML <linux-kernel@vger.kernel.org>,
        Linus Torvalds <torvalds@linuxfoundation.org>, x86@kernel.org,
        Borislav Petkov <bp@alien8.de>,
        David Woodhouse <dwmw@amazon.co.uk>,
        Tim Chen <tim.c.chen@linux.intel.com>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Andi Kleen <ak@linux.intel.com>,
        Greg KH <gregkh@linuxfoundation.org>,
        Andy Lutomirski <luto@kernel.org>,
        Arjan Van De Ven <arjan.van.de.ven@intel.com>
Subject: Re: [patch RFC 5/5] x86/speculation: Add basic speculation control
 code
Message-ID: <20180110092234.GY29822@worktop.programming.kicks-ass.net>
References: <20180110010652.404145126@linutronix.de>
 <20180110011350.855878109@linutronix.de>
 <ce6c99be-b8ea-f789-ca7c-625c99b2a7a2@intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ce6c99be-b8ea-f789-ca7c-625c99b2a7a2@intel.com>
User-Agent: Mutt/1.5.22.1 (2013-10-16)
Sender: linux-kernel-owner@vger.kernel.org

On Tue, Jan 09, 2018 at 06:02:53PM -0800, Dave Hansen wrote:
> On 01/09/2018 05:06 PM, Thomas Gleixner wrote:
> > --- a/arch/x86/kernel/cpu/bugs.c
> > +++ b/arch/x86/kernel/cpu/bugs.c
> > @@ -79,6 +79,7 @@ enum spectre_v2_mitigation_cmd {
> >  	SPECTRE_V2_CMD_RETPOLINE,
> >  	SPECTRE_V2_CMD_RETPOLINE_GENERIC,
> >  	SPECTRE_V2_CMD_RETPOLINE_AMD,
> > +	SPECTRE_V2_CMD_IBRS,
> >  };
> 
> A few nits on this:
> 
> IBRS should not default on anywhere, which goes double when retpolines
> are available.
> 
> I think I'd also prefer that we separate the IBRS and retpoline enabling
> so that you can do both if you want.  They do nearly the same thing in
> practice, but I can't convince myself that you never ever need IBRS once
> retpolines are in place.

As per:

https://lkml.kernel.org/r/1515460999.4423.104.camel@amazon.co.uk

IBRS=2 (always on) is preferred for SKL+ over retpoline.

And from what I gather IBRS=1 is never better than retpoline, IBRS=1 is
both slower and covers less AFAIU (please educate if I'm wrong).

>From this point, I would prefer to not even have the IBRS=1 code.

The only question I have is if retpoline works at all on SKL (with ucode
update); BDW needs the ucode update for retpoline to work because of the
RSB fallback.