Re: [PATCH 0/1] mm: restore full accuracy in COW page reuse

From: Andy Lutomirski <luto@kernel.org>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: John Hubbard <jhubbard@nvidia.com>,
	Andrea Arcangeli <aarcange@redhat.com>,
	 Andrew Morton <akpm@linux-foundation.org>,
	Linux-MM <linux-mm@kvack.org>,
	 Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Yu Zhao <yuzhao@google.com>,  Andy Lutomirski <luto@kernel.org>,
	Peter Xu <peterx@redhat.com>, Pavel Emelyanov <xemul@openvz.org>,
	 Mike Kravetz <mike.kravetz@oracle.com>,
	Mike Rapoport <rppt@linux.vnet.ibm.com>,
	 Minchan Kim <minchan@kernel.org>, Will Deacon <will@kernel.org>,
	 Peter Zijlstra <peterz@infradead.org>,
	Hugh Dickins <hughd@google.com>,
	 "Kirill A. Shutemov" <kirill@shutemov.name>,
	Matthew Wilcox <willy@infradead.org>,
	 Oleg Nesterov <oleg@redhat.com>, Jann Horn <jannh@google.com>,
	Kees Cook <keescook@chromium.org>,
	 Leon Romanovsky <leonro@nvidia.com>,
	Jason Gunthorpe <jgg@ziepe.ca>, Jan Kara <jack@suse.cz>,
	 Kirill Tkhai <ktkhai@virtuozzo.com>,
	Nadav Amit <nadav.amit@gmail.com>,  Jens Axboe <axboe@kernel.dk>
Subject: Re: [PATCH 0/1] mm: restore full accuracy in COW page reuse
Date: Tue, 12 Jan 2021 09:07:31 -0800	[thread overview]
Message-ID: <CALCETrWEAOj28Y6SaBVuxahuRmk9exp1fqcJO0ibGKbFw4HQ4A@mail.gmail.com> (raw)
In-Reply-To: <CAHk-=wje9r3fREBdZcOu=NihGczBtkqkhXRPDhY-ZkNVv=thiQ@mail.gmail.com>

On Mon, Jan 11, 2021 at 2:18 PM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> On Mon, Jan 11, 2021 at 11:19 AM Linus Torvalds
> <torvalds@linux-foundation.org> wrote:

> Actually, what I think might be a better model is to actually
> strengthen the rules even more, and get rid of GUP_PIN_COUNTING_BIAS
> entirely.
>
> What we could do is just make a few clear rules explicit (most of
> which we already basically hold to). Starting from that basic
>
>  (a) Anonymous pages are made writable (ie COW) only when they have a
> page_count() of 1

Seems reasonable to me.

>
> That very simple rule then automatically results in the corollary
>
>  (b) a writable page in a COW mapping always starts out reachable
> _only_ from the page tables

Seems reasonable.  I guess that if the COW is triggered by GUP, then
it starts out reachable only from the page tables but then because
reachable through GUP very soon thereafter.

>
> and now we could have a couple of really simple new rules:
>
>  (c) we never ever make a writable page in a COW mapping read-only
> _unless_ it has a page_count() of 1

I don't love this.  Having mprotect() fail in a multithreaded process
because another thread happens to be doing a short-lived IO seems like
it may result in annoying intermittent bugs.

As I understand it, the issue is that the way we determine that we
need to COW a COWable page is that we see that it's read-only.  It
would be nice if we could separately track "the VMA allows writes" and
"this PTE points to a page that is private to the owning VMA", but
maybe there's no bit available for the latter other than looking at RO
vs RW directly.

>
>  (d) we never create a swap cache page out of a writable COW mapping page
>
> Now, if you combine these rules, the whole need for the
> GUP_PIN_COUNTING_BIAS basically goes away.
>
> Why? Because we know that the _only_ thing that can elevate the
> refcount of a writable COW page is GUP - we'll just make sure nothing
> else touches it.

How common is !FOLL_WRITE GUP?  We could potentially say that a
short-term !FOLL_WRITE GUP is permitted on an RO COW page and that a
subsequent COW on the page will wait for the GUP to go away.  This
might be too big a can of worms for the benefit it would provide,
though.

--Andy