linux-kernel - Re: [PATCH v2] kernel/module.c: heuristic enhancement when INSTALL_MOD

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: <YeBEAavbBh/MnbEF@bombadil.infradead.org>
Date:   Thu, 13 Jan 2022 07:23:45 -0800
From:   Luis Chamberlain <mcgrof@...nel.org>
To:     Vimal Agrawal <avimalin@...il.com>
Cc:     Christoph Hellwig <hch@...radead.org>,
        Vimal Agrawal <vimal.Agrawal@...hos.com>,
        Masahiro Yamada <masahiroy@...nel.org>,
        Michal Marek <michal.lkml@...kovi.net>,
        Nick Desaulniers <ndesaulniers@...gle.com>,
        Jan Beulich <JBeulich@...e.com>, Jeff Mahoney <jeffm@...e.com>,
        Sam Ravnborg <sam@...nborg.org>, linux-kbuild@...r.kernel.org,
        jeyu@...nel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] kernel/module.c: heuristic enhancement when
 INSTALL_MOD_STRIP= "--strip-unneeded" is used

On Wed, Jan 12, 2022 at 02:06:48PM +0530, Vimal Agrawal wrote:
> On Tue, Jan 11, 2022 at 9:19 PM Luis Chamberlain <mcgrof@...nel.org> wrote:
> > Yes but the point here is the heuristic you are adding for
> > when "--strip-unneeded" is used is now *always* being used and
> > we have no way of knowing this. So I'd agree with Christoph that
> > if we want to support this it might make sense to make a kconfig
> > option for enabling "--strip-unneeded" and then another for this
> > heuristic.
> 
> This heuristic is applicable to any case when an address inside a
> module can not be decoded to any known symbol. e.g. 

You mean it is safe for that case too? If so can you add this to the
commit log as well?

> anyone can still
> build with ----strip-all though module load fails with this option.
> but one can add or remove symbols manually or use objcopy or some
> other utility to play with symbols. It does not matter for
> functionality much if symbols are available or not and it is just that
> symbol decodes in traces are not providing help as it displays
> absolute address in such cases.
> 
> There are several options in strip command and we can't have kconfig
> for each such option. 

This is a good point. Pointing out it is safe regardless of the
situation in the commit log I think does this heuristic more justice.

Can you resend with that being clarified? I'll still have to test this.
Can you also use the modules-next tree? I'll still have to test this.

[0] https://git.kernel.org/pub/scm/linux/kernel/git/mcgrof/linux.git/ branch modules-next

  Luis

> All options are supported currently unless the
> module is so broken that it can even be loaded ( e.g. --strip-all
> option).
> 
> Vimal