linux-kernel - Re: [PATCH v4 2/8] mm: introduce external memory hinting API

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: <20200213161007.GA24649@google.com>
Date:   Thu, 13 Feb 2020 08:10:07 -0800
From:   Minchan Kim <minchan@...nel.org>
To:     Jann Horn <jannh@...gle.com>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        LKML <linux-kernel@...r.kernel.org>,
        linux-mm <linux-mm@...ck.org>,
        Linux API <linux-api@...r.kernel.org>,
        Oleksandr Natalenko <oleksandr@...hat.com>,
        Suren Baghdasaryan <surenb@...gle.com>,
        Tim Murray <timmurray@...gle.com>,
        Daniel Colascione <dancol@...gle.com>,
        Sandeep Patil <sspatil@...gle.com>,
        Sonny Rao <sonnyrao@...gle.com>,
        Brian Geffon <bgeffon@...gle.com>,
        Michal Hocko <mhocko@...e.com>,
        Johannes Weiner <hannes@...xchg.org>,
        Shakeel Butt <shakeelb@...gle.com>,
        John Dias <joaodias@...gle.com>,
        Joel Fernandes <joel@...lfernandes.org>,
        Alexander Duyck <alexander.h.duyck@...ux.intel.com>
Subject: Re: [PATCH v4 2/8] mm: introduce external memory hinting API

Hi Jann,

On Thu, Feb 13, 2020 at 03:08:59PM +0100, Jann Horn wrote:
> On Thu, Feb 13, 2020 at 12:40 AM Minchan Kim <minchan@...nel.org> wrote:
> > To solve the issue, this patch introduces a new syscall process_madvise(2).
> > It uses pidfd of an external process to give the hint.
> [...]
> > +       mm = mm_access(task, PTRACE_MODE_ATTACH_FSCREDS);
> > +       if (IS_ERR_OR_NULL(mm)) {
> > +               ret = IS_ERR(mm) ? PTR_ERR(mm) : -ESRCH;
> > +               goto release_task;
> > +       }
> > +
> > +       ret = do_madvise(task, start, len_in, behavior);
> 
> When you're accessing another task, you should ensure that the other
> task doesn't gain new privileges by executing a setuid binary in the
> middle of being accessed. mm_access() does that for you; it holds the
> ->cred_guard_mutex while it is looking up the task's ->mm and doing
> the security check. mm_access() then returns you an mm pointer that
> you're allowed to access without worrying about such things; an
> mm_struct never gains privileges, since a setuid execution creates a
> fresh mm_struct. However, the task may still execute setuid binaries
> and such things.
> 
> This means that after you've looked up the mm with mm_access(), you
> have to actually *use* that pointer. You're not allowed to simply read
> task->mm yourself.
> 
> Therefore, I think you should:
> 
>  - change patch 1/8 ("mm: pass task to do_madvise") to also pass an
> mm_struct* to do_madvise (but keep the task_struct* for patch 4/8)
>  - in this patch, pass the mm_struct* from mm_access() into do_madvise()
>  - drop patch 3/8 ("mm: validate mm in do_madvise"); it just papers
> over a symptom without addressing the underlying problem

Actually, it was what this patch series was doing until last version
but I changed it to reduce just *a parameter* to do_madvise.
And then, this time, I got a good advise I was not familiar.
I will fix it again.
Thanks for the review!