linux-kernel - Re: [PATCH] lkdtm: don't move ctors to .rodata

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: <YCla7cNQxBoG2KCr@kroah.com>
Date:   Sun, 14 Feb 2021 18:16:29 +0100
From:   Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To:     Stephen Boyd <swboyd@...omium.org>
Cc:     Kees Cook <keescook@...omium.org>,
        Mark Rutland <mark.rutland@....com>,
        linux-kernel@...r.kernel.org, Arnd Bergmann <arnd@...db.de>,
        stable@...r.kernel.org
Subject: Re: [PATCH] lkdtm: don't move ctors to .rodata

On Thu, Feb 11, 2021 at 10:53:10AM -0800, Stephen Boyd wrote:
> Quoting Greg Kroah-Hartman (2021-02-11 06:23:10)
> > On Wed, Feb 10, 2021 at 04:36:08PM -0800, Stephen Boyd wrote:
> > > Quoting Greg Kroah-Hartman (2020-12-09 06:51:33)
> > > > On Tue, Dec 08, 2020 at 01:20:56PM -0800, Kees Cook wrote:
> > > > > On Mon, Dec 07, 2020 at 05:05:33PM +0000, Mark Rutland wrote:
> > > > > > [    0.969110] Code: 00000003 00000000 00000000 00000000 (00000000)
> > > > > > [    0.970815] ---[ end trace b5339784e20d015c ]---
> > > > > > 
> > > > > > Signed-off-by: Mark Rutland <mark.rutland@....com>
> > > > > 
> > > > > Oh, eek. Why was a ctor generated at all? But yes, this looks good.
> > > > > Greg, can you pick this up please?
> > > > > 
> > > > > Acked-by: Kees Cook <keescook@...omium.org>
> > > > 
> > > > Now picked up, thanks.
> > > > 
> > > 
> > > Can this be backported to 5.4 and 5.10 stable trees? I just ran across
> > > this trying to use kasan on 5.4 with lkdtm and it blows up early. This
> > > patch applies on 5.4 cleanly but doesn't compile because it's missing
> > > noinstr. Here's a version of the patch that introduces noinstr on 5.4.97
> > > so this patch can be picked to 5.4 stable trees.
> > 
> > Why 5.10?  This showed up in 5.8, so how would it be needed there?
> > 
> 
> Sorry for the confusion. Can commit 655389666643 ("vmlinux.lds.h: Create
> section for protection against instrumentation") and commit 3f618ab33234
> ("lkdtm: don't move ctors to .rodata") be backported to 5.4.y and only
> commit 3f618ab3323407ee4c6a6734a37eb6e9663ebfb9 be backported to 5.10.y?

655389666643 ("vmlinux.lds.h: Create section for protection against
instrumentation") does not apply cleanly to 5.4.y, so can you provide a
working backport for both of those patches to 5.4.y that you have
tested?

thanks,

greg k-h