linux-kernel - Re: [PATCH 1/3] module: Fix livepatch/ftrace module text permissions race

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:   Wed, 26 Jun 2019 16:44:45 +0200 (CEST)
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Petr Mladek <pmladek@...e.com>
cc:     Miroslav Benes <mbenes@...e.cz>,
        Steven Rostedt <rostedt@...dmis.org>,
        Josh Poimboeuf <jpoimboe@...hat.com>,
        Jessica Yu <jeyu@...nel.org>, Jiri Kosina <jikos@...nel.org>,
        Joe Lawrence <joe.lawrence@...hat.com>,
        linux-kernel@...r.kernel.org, live-patching@...r.kernel.org,
        Johannes Erdfelt <johannes@...felt.com>,
        Ingo Molnar <mingo@...nel.org>, mhiramat@...nel.org,
        torvalds@...ux-foundation.org
Subject: Re: [PATCH 1/3] module: Fix livepatch/ftrace module text permissions
 race

On Wed, 26 Jun 2019, Petr Mladek wrote:
> On Wed 2019-06-26 10:22:45, Miroslav Benes wrote:
> It is similar problem that has been solved by 2d1e38f56622b9bb5af8
> ("kprobes: Cure hotplug lock ordering issues"). This commit solved
> it by always taking cpu_hotplug_lock.rw_sem before text_mutex inside.
> 
> If we follow the lock ordering then ftrace has to take text_mutex
> only when stop_machine() is not called or from code called via
> stop_machine() parameter.
> 
> This is not easy with the current design. For example, arm calls
> set_all_modules_text_rw() already in ftrace_arch_code_modify_prepare(),
> see arch/arm/kernel/ftrace.c. And it is called:
> 
>   + outside stop_machine() from ftrace_run_update_code()
>   + without stop_machine() from ftrace_module_enable()
> 
> A conservative solution for 5.2 release would be to move text_mutex
> locking from the generic kernel/trace/ftrace.c into
> arch/x86/kernel/ftrace.c:
> 
>    ftrace_arch_code_modify_prepare()
>    ftrace_arch_code_modify_post_process()
> 
> It should be enough to fix the original problem because
> x86 is the only architecture that calls set_all_modules_text_rw()
> in ftrace path and supports livepatching at the same time.

Looks correct, but I've paged out all the gory details vs. lock ordering in
that area.

Thanks,

	tglx