linux-kernel - Re: [PATCH v7 1/3] mm: Shuffle initial free memory to improve memory-side-cache utilization

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: <CAGXu5jLVB6EKETqnKAwjtDYYXj9kjccb6HbFcghmxt8E1Qxq=g@mail.gmail.com>
Date:   Tue, 8 Jan 2019 15:24:11 -0800
From:   Kees Cook <keescook@...omium.org>
To:     Dan Williams <dan.j.williams@...el.com>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        Michal Hocko <mhocko@...e.com>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Mike Rapoport <rppt@...ux.ibm.com>,
        Keith Busch <keith.busch@...el.com>,
        Linux-MM <linux-mm@...ck.org>,
        LKML <linux-kernel@...r.kernel.org>, Mel Gorman <mgorman@...e.de>
Subject: Re: [PATCH v7 1/3] mm: Shuffle initial free memory to improve
 memory-side-cache utilization

On Mon, Jan 7, 2019 at 5:48 PM Dan Williams <dan.j.williams@...el.com> wrote:
>
> On Mon, Jan 7, 2019 at 4:19 PM Kees Cook <keescook@...omium.org> wrote:
> > Why does this need ACPI_NUMA? (e.g. why can't I use this on a non-ACPI
> > arm64 system?)
>
> I was thinking this would be expanded for each platform-type that will
> implement the auto-detect capability. However, there really is no
> direct dependency and if you wanted to just use the command line
> switch that should be allowed on any platform.
>
> I'll delete this dependency for v8, but I'll hold off on that posting
> awaiting feedback from mm folks.

Okay, cool. I'm glad there wasn't a real dep. :)

> > > +static bool shuffle_param;
> > > +extern int shuffle_show(char *buffer, const struct kernel_param *kp)
> > > +{
> > > +       return sprintf(buffer, "%c\n", test_bit(SHUFFLE_ENABLE, &shuffle_state)
> > > +                       ? 'Y' : 'N');
> > > +}
> > > +static int shuffle_store(const char *val, const struct kernel_param *kp)
> > > +{
> > > +       int rc = param_set_bool(val, kp);
> > > +
> > > +       if (rc < 0)
> > > +               return rc;
> > > +       if (shuffle_param)
> > > +               page_alloc_shuffle(SHUFFLE_ENABLE);
> > > +       else
> > > +               page_alloc_shuffle(SHUFFLE_FORCE_DISABLE);
> > > +       return 0;
> > > +}
> > > +module_param_call(shuffle, shuffle_store, shuffle_show, &shuffle_param, 0400);
> >
> > If this is 0400, you don't intend it to be changed after boot. If it's
> > supposed to be immutable, why not make these __init calls?
>
> It's not changeable after boot, but it's still readable after boot.
> This is there to allow interrogation of whether shuffling is in-effect
> at runtime.

In that case, can you make all the runtime-immutable things __ro_after_init?

> > > +                               ALIGN_DOWN(get_random_long() % z->spanned_pages,
> > > +                                               order_pages);
> >
> > How late in the boot process does this happen, btw?
>
> This happens early at mem_init() before the software rng is initialized.
>
> > Do we get warnings
> > from the RNG about early usage?
>
> Yes, it would trigger on some platforms. It does not on my test system
> because I'm running on an arch_get_random_long() enabled system.

Okay, cool. :)

-- 
Kees Cook