linux-kernel - Re: [PATCH] userfaultfd: prevent non-cooperative events vs mcopy

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:   Fri, 25 May 2018 17:05:42 +0300
From:   Pavel Emelyanov <xemul@...tuozzo.com>
To:     Mike Rapoport <rppt@...ux.vnet.ibm.com>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        linux-mm <linux-mm@...ck.org>,
        lkml <linux-kernel@...r.kernel.org>,
        Andrea Arcangeli <aarcange@...hat.com>,
        Mike Kravetz <mike.kravetz@...cle.com>,
        Andrei Vagin <avagin@...tuozzo.com>
Subject: Re: [PATCH] userfaultfd: prevent non-cooperative events vs
 mcopy_atomic races


>> But doesn't it race even with regular PF handling, not only the fork? How
>> do we handle this race?
> 
> With the regular #PF handing, the faulting thread patiently waits until
> page fault is resolved. With fork(), mremap() etc the thread that caused
> the event resumes once the uffd message is read by the monitor. That's
> surely way before monitor had chance to somehow process that message.

Ouch, yes. This is nasty :( So having no better solution in mind, let's
move forward with this.

Acked-by: Pavel Emelyanov <xemul@...tuozzo.com>