Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed, 15 Apr 2020 11:33:03 -0500
From:   Josh Poimboeuf <jpoimboe@redhat.com>
To:     Jessica Yu <jeyu@kernel.org>
Cc:     live-patching@vger.kernel.org, linux-kernel@vger.kernel.org,
        Peter Zijlstra <peterz@infradead.org>
Subject: Re: [PATCH 6/7] livepatch: Remove module_disable_ro() usage
Message-ID: <20200415163303.ubdnza6okg4h3e5a@treble>
References: <cover.1586881704.git.jpoimboe@redhat.com>
 <9f0d8229bbe79d8c13c091ed70c41d49caf598f2.1586881704.git.jpoimboe@redhat.com>
 <20200415150216.GA6164@linux-8ccs.fritz.box>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20200415150216.GA6164@linux-8ccs.fritz.box>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Apr 15, 2020 at 05:02:16PM +0200, Jessica Yu wrote:
> +++ Josh Poimboeuf [14/04/20 11:28 -0500]:
> > With arch_klp_init_object_loaded() gone, and apply_relocate_add() now
> > using text_poke(), livepatch no longer needs to use module_disable_ro().
> > 
> > The text_mutex usage can also be removed -- its purpose was to protect
> > against module permission change races.
> > 
> > Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
> > ---
> > kernel/livepatch/core.c | 8 --------
> > 1 file changed, 8 deletions(-)
> > 
> > diff --git a/kernel/livepatch/core.c b/kernel/livepatch/core.c
> > index 817676caddee..3a88639b3326 100644
> > --- a/kernel/livepatch/core.c
> > +++ b/kernel/livepatch/core.c
> > @@ -767,10 +767,6 @@ static int klp_init_object_loaded(struct klp_patch *patch,
> > 	struct klp_modinfo *info = patch->mod->klp_info;
> > 
> > 	if (klp_is_module(obj)) {
> > -
> > -		mutex_lock(&text_mutex);
> > -		module_disable_ro(patch->mod);
> > -
> 
> Don't you still need the text_mutex to use text_poke() though?
> (Through klp_write_relocations -> apply_relocate_add -> text_poke)
> At least, I see this assertion there:
> 
> void *text_poke(void *addr, const void *opcode, size_t len)
> {
> 	lockdep_assert_held(&text_mutex);
> 
> 	return __text_poke(addr, opcode, len);
> }

Hm, guess I should have tested with lockdep ;-)

-- 
Josh