[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190308222744.fox35ov75zxi6wis@treble>
Date: Fri, 8 Mar 2019 16:27:44 -0600
From: Josh Poimboeuf <jpoimboe@...hat.com>
To: Peter Zijlstra <peterz@...radead.org>
Cc: torvalds@...ux-foundation.org, tglx@...utronix.de, hpa@...or.com,
julien.thierry@....com, will.deacon@....com, luto@...capital.net,
mingo@...nel.org, catalin.marinas@....com, james.morse@....com,
valentin.schneider@....com, brgerst@...il.com, luto@...nel.org,
bp@...en8.de, dvlasenk@...hat.com, linux-kernel@...r.kernel.org,
dvyukov@...gle.com, rostedt@...dmis.org
Subject: Re: [PATCH 16/20] objtool: Rewrite alt->skip_orig
On Fri, Mar 08, 2019 at 10:34:18PM +0100, Peter Zijlstra wrote:
> On Fri, Mar 08, 2019 at 02:15:56PM -0600, Josh Poimboeuf wrote:
> > On Thu, Mar 07, 2019 at 12:45:27PM +0100, Peter Zijlstra wrote:
> > > Really skip the original instruction flow, instead of letting it
> > > continue with NOPs.
> > >
> > > Since the alternative code flow already continues after the original
> > > instructions, only the alt-original is skipped.
> >
> > I like this approach. I wonder if we can do something similar to get
> > rid of the nasty fake jumps.
>
> This actually hard relies on those fake jumps. Or am I missing the
> point?
I was just wondering out loud if we could somehow keep the same "fake
jump" functionality, but do it in a cleaner way that doesn't require
creating fake instructions. I'll might give it a try.
--
Josh
Powered by blists - more mailing lists