Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 24 Oct 2019 13:31:15 -0500
From:   Josh Poimboeuf <jpoimboe@redhat.com>
To:     Peter Zijlstra <peterz@infradead.org>
Cc:     x86@kernel.org, linux-kernel@vger.kernel.org, rostedt@goodmis.org,
        mhiramat@kernel.org, bristot@redhat.com, jbaron@akamai.com,
        torvalds@linux-foundation.org, tglx@linutronix.de,
        mingo@kernel.org, namit@vmware.com, hpa@zytor.com, luto@kernel.org,
        ard.biesheuvel@linaro.org, jeyu@kernel.org
Subject: Re: [PATCH v4 15/16] module: Move where we mark modules RO,X
Message-ID: <20191024183115.jomddpijq5u453qc@treble>
References: <20191018073525.768931536@infradead.org>
 <20191018074634.801435443@infradead.org>
 <20191021135312.jbbxsuipxldocdjk@treble>
 <20191021141402.GI1817@hirez.programming.kicks-ass.net>
 <20191023114835.GT1817@hirez.programming.kicks-ass.net>
 <20191023151654.GF19358@hirez.programming.kicks-ass.net>
 <20191023171514.7hkhtvfcj5vluwcg@treble>
 <20191024105904.GB4131@hirez.programming.kicks-ass.net>
MIME-Version: 1.0
In-Reply-To: <20191024105904.GB4131@hirez.programming.kicks-ass.net>
User-Agent: NeoMutt/20180716
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Oct 24, 2019 at 12:59:04PM +0200, Peter Zijlstra wrote:
> On Wed, Oct 23, 2019 at 12:15:14PM -0500, Josh Poimboeuf wrote:
> > On Wed, Oct 23, 2019 at 05:16:54PM +0200, Peter Zijlstra wrote:
> > > @@ -157,6 +158,14 @@ static int __apply_relocate_add(Elf64_Sh
> > > =20
> > >  =09=09val =3D sym->st_value + rel[i].r_addend;
> > > =20
> > > +=09=09/*
> > > +=09=09 * .klp.rela.* sections should only contain module
> > > +=09=09 * related RELAs. All core-kernel RELAs should be in
> > > +=09=09 * normal .rela.* sections and be applied when loading
> > > +=09=09 * the patch module itself.
> > > +=09=09 */
> > > +=09=09WARN_ON_ONCE(klp && core_kernel_text(val));
> > > +
> >=20
> > This isn't quite true, we also use .klp.rela sections to access
> > unexported vmlinux symbols.
>=20
> Yes, you said in that earlier email. That all makes it really hard to
> validate this. But unless we validate it, it will stay buggy :/
>=20
> Hmmm.... /me ponders
>=20
> The alternative would be to apply the .klp.rela.* sections twice, once
> at patch-module load time and then apply those core_kernel_text()
> entries, and then once later and skip over them.
>=20
> How's this?

How about something like this?  Completely untested, but if you agree
with this approach I could hack up kpatch-build to test it properly.

diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
index ab4a4606d19b..597bf32bc591 100644
--- a/kernel/livepatch/core.c
+++ b/kernel/livepatch/core.c
@@ -239,6 +239,17 @@ static int klp_resolve_symbols(Elf_Shdr *relasec, stru=
ct module *pmod)
 =09=09if (ret)
 =09=09=09return ret;
=20
+=09=09/*
+=09=09 * Prevent module patches from using livepatch relas for
+=09=09 * vmlinux symbols.  Presumably such symbols are exported and
+=09=09 * normal relas can instead be used at patch module loading
+=09=09 * time.
+=09=09 */
+=09=09if (!vmlinux && core_kernel_text(addr)) {
+=09=09=09pr_err("unsupported livepatch symbol\n");
+=09=09=09return -EINVAL;
+=09=09}
+
 =09=09sym->st_value =3D addr;
 =09}
=20