linux-kernel - Re: [RFC][PATCH 5/7] x86/speculation: Change __FILL_RETURN

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200419185838.GX20696@hirez.programming.kicks-ass.net>
Date:   Sun, 19 Apr 2020 20:58:38 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Josh Poimboeuf <jpoimboe@...hat.com>
Cc:     alexandre.chartre@...cle.com, linux-kernel@...r.kernel.org,
        jthierry@...hat.com, tglx@...utronix.de, x86@...nel.org
Subject: Re: [RFC][PATCH 5/7] x86/speculation: Change __FILL_RETURN_BUFFER to
 work with objtool

On Sun, Apr 19, 2020 at 11:55:19AM -0500, Josh Poimboeuf wrote:
> On Sun, Apr 19, 2020 at 11:52:00AM -0500, Josh Poimboeuf wrote:
> > Are we still planning to warn about stack changes inside an alternative?
> > If so then this would still fail...
> > 
> > In this case I think it should be safe, but I'm not sure how we can
> > ensure that will always be the case for other alternatives.
> > 
> > And do the ORC entries actually work for this?  As far as I can tell,
> > they would be associated with the .altinstructions section and not
> > .text, so it wouldn't work.
> 
> My preference would be to move RSB stuffing out-of-line too, like you
> did the retpolines.  Or use static branches.  Then we could add an
> objtool warning to prevent stack changes in alternatives.

I effectively did the static_branch thing, but with an alternative, it's
in the last patch, due to me being a moron and not refreshing the stack
before sending it out.