[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20231128085025.GA3818@noisy.programming.kicks-ass.net>
Date: Tue, 28 Nov 2023 09:50:25 +0100
From: Peter Zijlstra <peterz@...radead.org>
To: "Edgecombe, Rick P" <rick.p.edgecombe@...el.com>
Cc: "Yang, Weijiang" <weijiang.yang@...el.com>,
"Gao, Chao" <chao.gao@...el.com>,
"Hansen, Dave" <dave.hansen@...el.com>,
"seanjc@...gle.com" <seanjc@...gle.com>,
"john.allen@....com" <john.allen@....com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"kvm@...r.kernel.org" <kvm@...r.kernel.org>,
"pbonzini@...hat.com" <pbonzini@...hat.com>,
"mlevitsk@...hat.com" <mlevitsk@...hat.com>
Subject: Re: [PATCH v7 02/26] x86/fpu/xstate: Refine CET user xstate bit
enabling
On Tue, Nov 28, 2023 at 01:31:14AM +0000, Edgecombe, Rick P wrote:
> On Fri, 2023-11-24 at 10:40 +0100, Peter Zijlstra wrote:
> > So booting a host with "ibt=off" will clear the FEATURE_IBT, this was
> > fine before this patch-set, but possibly not with.
> >
> > That kernel argument really only wants to tell the kernel not to use
> > IBT
> > itself, but not inhibit IBT from being used by guests.
>
> Should we add a SW bit for it then?
Don't think we need a feature flag for this, some boolean state should
be enough.
> ibt=off sounds like it should be
> turning off the whole feature though. It doesn't sound like kernel IBT
> specifically.
Well, the intent really was to just not enable IBT, clearing the feature
was the simplest way to make that happen.
Powered by blists - more mailing lists