[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20180112225714.GQ6718@tassilo.jf.intel.com>
Date: Fri, 12 Jan 2018 14:57:14 -0800
From: Andi Kleen <ak@...ux.intel.com>
To: Dominik Brodowski <law@...inikbrodowski.net>
Cc: Andi Kleen <andi@...stfloor.org>, tglx@...utronix.de,
x86@...nel.org, dwmw@...zon.co.uk, linux-kernel@...r.kernel.org,
pjt@...gle.com, torvalds@...ux-foundation.org,
gregkh@...ux-foundation.org, peterz@...radead.org,
luto@...capital.net, thomas.lendacky@....com,
arjan.van.de.ven@...el.com
Subject: Re: [PATCH 1/4] x86/retpoline: Add new mode RETPOLINE_UNDERFLOW
On Fri, Jan 12, 2018 at 11:13:13PM +0100, Dominik Brodowski wrote:
> > @@ -209,6 +234,11 @@ static void __init spectre_v2_select_mitigation(void)
> > mode = retp_compiler() ? SPECTRE_V2_RETPOLINE_GENERIC :
> > SPECTRE_V2_RETPOLINE_MINIMAL;
> > setup_force_cpu_cap(X86_FEATURE_RETPOLINE);
> > + if (mode == SPECTRE_V2_RETPOLINE_GENERIC &&
> > + cpu_needs_underflow_protection()) {
> > + mode = SPECTRE_V2_RETPOLINE_UNDERFLOW;
> > + setup_force_cpu_cap(X86_FEATURE_RETURN_UNDERFLOW);
> > + }
> > }
>
> Why does this require a retp-compatible compiler to be enabled?
Without the compiler you have a lot of other problems already,
so it's fairly pointless.
-Andi
Powered by blists - more mailing lists