Re: interaction of MADV_PAGEOUT with CoW anonymous mappings?

From: Daniel Colascione <dancol@google.com>
To: Jann Horn <jannh@google.com>
Cc: Michal Hocko <mhocko@kernel.org>,
	Minchan Kim <minchan@kernel.org>, Linux-MM <linux-mm@kvack.org>,
	kernel list <linux-kernel@vger.kernel.org>,
	Dave Hansen <dave.hansen@intel.com>,
	"Joel Fernandes (Google)" <joel@joelfernandes.org>
Subject: Re: interaction of MADV_PAGEOUT with CoW anonymous mappings?
Date: Tue, 10 Mar 2020 14:52:55 -0700	[thread overview]
Message-ID: <CAKOZues-ejm8kKOp9YaZwDy_HNhvWK12x2cqtqHBYj4DCQtXpQ@mail.gmail.com> (raw)
In-Reply-To: <CAG48ez2EWzn_xm0Q3YaPjSXrhVAAOUUvNDyDQ+jpgavmN6wLvQ@mail.gmail.com>

On Tue, Mar 10, 2020 at 2:40 PM Jann Horn <jannh@google.com> wrote:
>
> On Tue, Mar 10, 2020 at 9:19 PM Daniel Colascione <dancol@google.com> wrote:
> > On Tue, Mar 10, 2020 at 11:48 AM Michal Hocko <mhocko@kernel.org> wrote:
> > > On Tue 10-03-20 19:08:28, Jann Horn wrote:
> > > > >From looking at the source code, it looks to me as if using
> > > > MADV_PAGEOUT on a CoW anonymous mapping will page out the page if
> > > > possible, even if other processes still have the same page mapped. Is
> > > > that correct?
> > > >
> > > > If so, that's probably bad in environments where many processes (with
> > > > different privileges) are forked from a single zygote process (like
> > > > Android and Chrome), I think? If you accidentally call it on a CoW
> > > > anonymous mapping with shared pages, you'll degrade the performance of
> > > > other processes. And if an attacker does it intentionally, they could
> > > > use that to aid with exploiting race conditions or weird
> > > > microarchitectural stuff (e.g. the new https://lviattack.eu/lvi.pdf
> > > > talks about "the assumption that attackers can provoke page faults or
> > > > microcode assists for (arbitrary) load operations in the victim
> > > > domain").
> > > >
> > > > Should madvise_cold_or_pageout_pte_range() maybe refuse to operate on
> > > > pages with mapcount>1, or something like that? Or does it already do
> > > > that, and I just missed the check?
> > >
> > > I have brought up side channel attacks earlier [1] but only in the
> > > context of shared page cache pages. I didn't really consider shared
> > > anonymous pages to be a real problem. I was under impression that CoW
> > > pages shouldn't be a real problem because any security sensible
> > > applications shouldn't allow untrusted code to be forked and CoW
> > > anything really important. I believe we have made this assumption
> > > in other places - IIRC on gup with FOLL_FORCE but I admit I have
> > > very happily forgot most details.
> >
> > I'm more worried about the performance implications. Consider
> > libc.so's data section: that's a COW mapping, and we COW it during
> > zygote initialization as we load and relocate libc.so. Child processes
> > shouldn't be dirtying and re-COWing those relocated pages. If I
> > understand Jann's message correctly, MADV_PAGEOUT would force the
> > pages corresponding to the libc.so data segment out to zram just
> > because we MADV_PAGEOUT-ed a single process that happened to use libc.
> > We should leave those pages in memory, IMHO.
>
> Actually, the libc.so data section is a file mapping, so I think
> can_do_pageout() would decide whether the caller is allowed to force
> pageout based on whether the caller is the owner of (or capable over)
> libc (in other words, root, basically). But I think the bss section,
> as well as heap memory, could have pageout forced by anyone.

lmkd would have that capability though, right? But the point stands
regardless. It sounds like both security and performance suggest a
behavior change here. Thanks for bringing it up!