Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <CAG48ez0B18eh3Q1853Cug8WSip7dPb2G9fhgqsPWzr0D_TBjRQ@mail.gmail.com>
 <20221014222346.n337tvkbyr33dsdx@box.shutemov.name>
In-Reply-To: <20221014222346.n337tvkbyr33dsdx@box.shutemov.name>
From:   Jann Horn <jannh@google.com>
Date:   Sat, 15 Oct 2022 00:29:57 +0200
Message-ID: <CAG48ez1B11EFyssTi=4izy04_FBOP1qdYVhEomYRdDBXb3jHkA@mail.gmail.com>
Subject: Re: [BUG?] X86 arch_tlbbatch_flush() seems to be lacking
 mm_tlb_flush_nested() integration
To:     "Kirill A. Shutemov" <kirill@shutemov.name>
Cc:     Andy Lutomirski <luto@kernel.org>, Linux-MM <linux-mm@kvack.org>,
        Mel Gorman <mgorman@suse.de>, Rik van Riel <riel@redhat.com>,
        kernel list <linux-kernel@vger.kernel.org>,
        Kees Cook <keescook@chromium.org>,
        Ingo Molnar <mingo@kernel.org>,
        Sasha Levin <sasha.levin@oracle.com>,
        Andrew Morton <akpm@linux-foundation.org>,
        Will Deacon <will@kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Linus Torvalds <torvalds@linuxfoundation.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Sat, Oct 15, 2022 at 12:23 AM Kirill A. Shutemov
<kirill@shutemov.name> wrote:
> On Fri, Oct 14, 2022 at 08:19:42PM +0200, Jann Horn wrote:
> > Hi!
> >
> > I haven't actually managed to reproduce this behavior, so maybe I'm
> > just misunderstanding how this works; but I think the
> > arch_tlbbatch_flush() path for batched TLB flushing in vmscan ought to
> > have some kind of integration with mm_tlb_flush_nested().
> >
> > I think that currently, the following race could happen:
> >
> > [initial situation: page P is mapped into a page table of task B, but
> > the page is not referenced, the PTE's A/D bits are clear]
> > A: vmscan begins
> > A: vmscan looks at P and P's PTEs, and concludes that P is not currently in use
> > B: reads from P through the PTE, setting the Accessed bit and creating
> > a TLB entry
> > A: vmscan enters try_to_unmap_one()
> > A: try_to_unmap_one() calls should_defer_flush(), which returns true
> > A: try_to_unmap_one() removes the PTE and queues a TLB flush
> > (arch_tlbbatch_add_mm())
> > A: try_to_unmap_one() returns, try_to_unmap() returns to shrink_folio_list()
> > B: calls munmap() on the VMA that mapped P
> > B: no PTEs are removed, so no TLB flush happens
> > B: munmap() returns
>
> I think here we will serialize against anon_vma/i_mmap lock in
> __do_munmap() -> unmap_region() -> free_pgtables() that A also holds.
>
> So I believe munmap() is safe, but MADV_DONTNEED (and its flavours) is not.

shrink_folio_list() is not in a context that is operating on a
specific MM; it is operating on a list of pages that might be mapped
into different processes all over the system.

So A has temporarily held those locks somewhere inside
try_to_unmap_one(), but it will drop them before it reaches the point
where it issues the batched TLB flush.
And this batched TLB flush potentially covers multiple MMs at once; it
is not targeted towards a specific MM, but towards all of the CPUs on
which any of the touched MMs might be active.