linux-kernel - Re: arm: Add livepatch arch specific code

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]

Date:   Tue, 17 Jan 2017 14:53:06 +0100 (CET)
From:   Miroslav Benes <mbenes@...e.cz>
To:     Jessica Yu <jeyu@...hat.com>
cc:     Abel Vesa <abelvesa@...ux.com>, linux@...linux.org.uk,
        jpoimboe@...hat.com, jikos@...nel.org, pmladek@...e.com,
        rostedt@...dmis.org, mingo@...hat.com, gregkh@...uxfoundation.org,
        geert+renesas@...der.be, davem@...emloft.net,
        akpm@...ux-foundation.org, emil.l.velikov@...il.com,
        mchehab@...nel.org, linux@...ck-us.net, ard.biesheuvel@...aro.org,
        jens.wiklander@...aro.org, jean-philippe.brucker@....com,
        viro@...iv.linux.org.uk, stefano.stabellini@...citrix.com,
        chris.brandt@...esas.com, linux-kernel@...r.kernel.org,
        linux-arm-kernel@...ts.infradead.org, live-patching@...r.kernel.org
Subject: Re: arm: Add livepatch arch specific code


> > But apply_relocate_add() is not implemented on arm yet. I guess it would
> > nice to have it... otherwise we could get to an unpleasant situation.
> > Livepatch module can rely on its livepatching relocations (that is, there
> > are some). apply_relocate_add() returns 0 on arm, so everything seems to
> > be nice and then boom some time later.
> 
> Doesn't it return -ENOEXEC? MODULES_USE_ELF_RELA is not defined
> on arm, I think (see moduleloader.h).

You're right. It confused me too.
 
> > The question is what happens with normal modules. There are no SHT_RELA
> > records probably.
> 
> For arm, I think there are only SHT_REL relocation sections
> (MODULES_USE_ELF_REL is set instead of MODULES_USE_ELF_RELA), so
> during apply_relocations() in load_module(), only apply_relocate()
> should be called, and not apply_relocate_add().

True.

> Hm, I guess that means if we want livepatch support for the
> architectures that don't support RELA relocations, we would have to
> check if the section is SHT_REL or SHT_RELA before calling the right
> apply_relocate* function.

Agreed.

Miroslav