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: <20230708191212.4147700-1-surenb@google.com> <20230708191212.4147700-3-surenb@google.com>
 <20230804214620.btgwhsszsd7rh6nf@f> <CAHk-=wiy125k1dBmQFTGpHwiOqEyrD6xnd4xKWfe97H_HodgDA@mail.gmail.com>
 <CAGudoHFsAU_BDCOuz8UgDBLGEM8xg=aUGjaVoqkM_Zvxo2Re_g@mail.gmail.com>
 <CAHk-=wiG9xaVvBJXHqTxtop0=mW9KxPS9C54ED23p59VNEKdWg@mail.gmail.com>
 <CAJuCfpGWGsh2BRgwcJ7oVHnqZfrtiesvhzomK0ZmxE_KK=R7FA@mail.gmail.com>
 <CAJuCfpG6BBP+fjV9oyBx3SNiKhiafPzM9vV9bx_goO2aZzAptg@mail.gmail.com> <CAGudoHFrDG6-u-XXEmQoPS2CJ2Wpo4ETwhXc2R=jy78RSYw-Zg@mail.gmail.com>
In-Reply-To: <CAGudoHFrDG6-u-XXEmQoPS2CJ2Wpo4ETwhXc2R=jy78RSYw-Zg@mail.gmail.com>
From:   Suren Baghdasaryan <surenb@google.com>
Date:   Fri, 4 Aug 2023 18:06:22 -0700
Message-ID: <CAJuCfpHYBqULvwNELO3Gkc0bkKDV7VJxMjvBru4zaAz4WKQNhw@mail.gmail.com>
Subject: Re: [PATCH v2 3/3] fork: lock VMAs of the parent process when forking
To:     Mateusz Guzik <mjguzik@gmail.com>
Cc:     Linus Torvalds <torvalds@linux-foundation.org>,
        akpm@linux-foundation.org, regressions@leemhuis.info,
        bagasdotme@gmail.com, jacobly.alt@gmail.com, willy@infradead.org,
        liam.howlett@oracle.com, david@redhat.com, peterx@redhat.com,
        ldufour@linux.ibm.com, linux-kernel@vger.kernel.org,
        linux-mm@kvack.org, linuxppc-dev@lists.ozlabs.org,
        linux-arm-kernel@lists.infradead.org, gregkh@linuxfoundation.org,
        regressions@lists.linux.dev, Jiri Slaby <jirislaby@kernel.org>,
        =?UTF-8?Q?Holger_Hoffst=C3=A4tte?= <holger@applied-asynchrony.com>,
        stable@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Fri, Aug 4, 2023 at 5:49=E2=80=AFPM Mateusz Guzik <mjguzik@gmail.com> wr=
ote:
>
> On 8/5/23, Suren Baghdasaryan <surenb@google.com> wrote:
> > On Fri, Aug 4, 2023 at 5:26=E2=80=AFPM Suren Baghdasaryan <surenb@googl=
e.com>
> > wrote:
> >>
> >> On Fri, Aug 4, 2023 at 5:15=E2=80=AFPM Linus Torvalds
> >> <torvalds@linux-foundation.org> wrote:
> >> >
> >> > On Fri, 4 Aug 2023 at 16:25, Mateusz Guzik <mjguzik@gmail.com> wrote=
:
> >> > >
> >> > > I know of these guys, I think they are excluded as is -- they go
> >> > > through access_remote_vm, starting with:
> >> > >         if (mmap_read_lock_killable(mm))
> >> > >                 return 0;
> >> > >
> >> > > while dup_mmap already write locks the parent's mm.
> >> >
> >> > Oh, you're only worried about vma_start_write()?
> >> >
> >> > That's a non-issue. It doesn't take the lock normally, since it star=
ts
> >> > off with
> >> >
> >> >         if (__is_vma_write_locked(vma, &mm_lock_seq))
> >> >                 return;
> >> >
> >> > which catches on the lock sequence number already being set.
> >>
> >> That check will prevent re-locking but if vma is not already locked
> >> then the call will proceed with obtaining the lock and setting
> >> vma->vm_lock_seq to mm->mm_lock_seq.
> >
> > The optimization Mateusz describes looks valid to me. If there is
> > nobody else to fault a page and mm_users is stable (which I think it
> > is because we are holding mmap_lock for write) then we can skip vma
> > locking, I think.
> >
>
> mm_users is definitely *not* stable -- it can be bumped by
> get_task_mm, which is only synchronized with task lock.

Ugh, you are of course correct. Poor choice for saying no new users
(threads) can appear from under us.

>
> However, the other users (that I know of ) go through the mmap
> semaphore to mess with anything which means they will wait for
> dup_mmap to finish (or do their work first). I would be surprised if
> there were any cases which don't take the semaphore, given that it was
> a requirement prior to the vma patchset (unless you patched some to no
> longer need it?). I would guess worst case the semaphore can be added
> if missing.

No, the only mmap_lock read-lock that is affected is during the page
fault, which is expected.

>
> What is guaranteed is that if the forking process is single-threaded,
> there will be no threads added out of nowhere -- the only thread which
> could do it is busy creating one in dup_mmap. If multithreaded
> operation of the forking process was the only problem, that's it.
>
> >>
> >> >
> >> > So no extra locking there.
> >> >
> >> > Well, technically there's extra locking because the code stupidly
> >> > doesn't initialize new vma allocations to the right sequence number,
> >> > but that was talked about here:
> >> >
> >> >
> >> > https://lore.kernel.org/all/CAHk-=3DwiCrWAoEesBuoGoqqufvesicbGp3cX0L=
yKgEvsFaZNpDA@mail.gmail.com/
> >> >
> >> > and it's a separate issue.
> >> >
> >> >           Linus
> >
>
>
> --
> Mateusz Guzik <mjguzik gmail.com>