linux-kernel - Re: [PATCH 3/3] exec: Transform exec_update_mutex into a rw

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: <20201204214836.3rncqw5kox42b4i2@linux-p48b.lan>
Date:   Fri, 4 Dec 2020 13:48:36 -0800
From:   Davidlohr Bueso <dave@...olabs.net>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Bernd Edlinger <bernd.edlinger@...mail.de>,
        "Eric W. Biederman" <ebiederm@...ssion.com>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>, Will Deacon <will@...nel.org>,
        Jann Horn <jannh@...gle.com>,
        Vasiliy Kulikov <segoon@...nwall.com>,
        Al Viro <viro@...iv.linux.org.uk>,
        Oleg Nesterov <oleg@...hat.com>,
        Cyrill Gorcunov <gorcunov@...il.com>,
        Sargun Dhillon <sargun@...gun.me>,
        Christian Brauner <christian.brauner@...ntu.com>,
        Arnd Bergmann <arnd@...db.de>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Waiman Long <longman@...hat.com>
Subject: Re: [PATCH 3/3] exec: Transform exec_update_mutex into a rw_semaphore

On Fri, 04 Dec 2020, Linus Torvalds wrote:

>On Fri, Dec 4, 2020 at 12:30 PM Bernd Edlinger
><bernd.edlinger@...mail.de> wrote:
>>>
>> >    perf_event_open  (exec_update_mutex -> ovl_i_mutex)
>
>Side note: this one looks like it should be easy to fix.
>
>Is there any real reason why exec_update_mutex is actually gotten that
>early, and held for that long in the perf event code?

afaict just to validate the whole operation early. Per 79c9ce57eb2 the
mutex will guard the check and the perf_install_in_context vs exec.

>
>I _think_ we could move the ptrace check to be much later, to _just_ before that
>
>         * This is the point on no return; we cannot fail hereafter.
>
>point in the perf event install chain..

Peter had the idea of doing the ptrace_may_access() check twice: first
lockless and early, then under exec_update_mutex when it mattered right
before perf_install_in_context():

https://lore.kernel.org/linux-fsdevel/20200828123720.GZ1362448@hirez.programming.kicks-ass.net/

>
>I don't think it needs to be moved down even that much, I think it
>would be sufficient to move it down below the "perf_event_alloc()",
>but I didn't check very much.

Yeah we could just keep a single ptrace_may_access() check just further
down until it won't deadlock.

Thanks,
Davidlohr