From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: MIME-Version: 1.0 In-Reply-To: <20181110032005.GA22238@google.com> References: <20181108041537.39694-1-joel@joelfernandes.org> <20181110032005.GA22238@google.com> From: Daniel Colascione Date: Sat, 10 Nov 2018 04:26:46 -0800 Message-ID: Subject: Re: [PATCH v3 resend 1/2] mm: Add an F_SEAL_FUTURE_WRITE seal to memfd To: Joel Fernandes Cc: Jann Horn , kernel list , "jreck@google.com" , John Stultz , Todd Kjos , Greg Kroah-Hartman , Christoph Hellwig , Al Viro , Andrew Morton , Bruce Fields , "jlayton@kernel.org" , Khalid Aziz , "Lei.Yang@windriver.com" , "linux-fsdevel@vger.kernel.org" , "linux-kselftest@vger.kernel.org" , Linux-MM , "marcandre.lureau@redhat.com" , Mike Kravetz , "minchan@kernel.org" , "shuah@kernel.org" , "valdis.kletnieks@vt.edu" , Hugh Dickins , Linux API Content-Type: multipart/alternative; boundary="000000000000e74e36057a4e94da" Sender: owner-linux-mm@kvack.org List-ID: --000000000000e74e36057a4e94da Content-Type: text/plain; charset="UTF-8" On Friday, November 9, 2018, Joel Fernandes wrote: > On Fri, Nov 09, 2018 at 10:19:03PM +0100, Jann Horn wrote: > > On Fri, Nov 9, 2018 at 10:06 PM Jann Horn wrote: > > > On Fri, Nov 9, 2018 at 9:46 PM Joel Fernandes (Google) > > > wrote: > > > > Android uses ashmem for sharing memory regions. We are looking > forward > > > > to migrating all usecases of ashmem to memfd so that we can possibly > > > > remove the ashmem driver in the future from staging while also > > > > benefiting from using memfd and contributing to it. Note staging > drivers > > > > are also not ABI and generally can be removed at anytime. > > > > > > > > One of the main usecases Android has is the ability to create a > region > > > > and mmap it as writeable, then add protection against making any > > > > "future" writes while keeping the existing already mmap'ed > > > > writeable-region active. This allows us to implement a usecase where > > > > receivers of the shared memory buffer can get a read-only view, while > > > > the sender continues to write to the buffer. > > > > See CursorWindow documentation in Android for more details: > > > > https://developer.android.com/reference/android/database/ > CursorWindow > > > > > > > > This usecase cannot be implemented with the existing F_SEAL_WRITE > seal. > > > > To support the usecase, this patch adds a new F_SEAL_FUTURE_WRITE > seal > > > > which prevents any future mmap and write syscalls from succeeding > while > > > > keeping the existing mmap active. > > > > > > Please CC linux-api@ on patches like this. If you had done that, I > > > might have criticized your v1 patch instead of your v3 patch... > > > > > > > The following program shows the seal > > > > working in action: > > > [...] > > > > Cc: jreck@google.com > > > > Cc: john.stultz@linaro.org > > > > Cc: tkjos@google.com > > > > Cc: gregkh@linuxfoundation.org > > > > Cc: hch@infradead.org > > > > Reviewed-by: John Stultz > > > > Signed-off-by: Joel Fernandes (Google) > > > > --- > > > [...] > > > > diff --git a/mm/memfd.c b/mm/memfd.c > > > > index 2bb5e257080e..5ba9804e9515 100644 > > > > --- a/mm/memfd.c > > > > +++ b/mm/memfd.c > > > [...] > > > > @@ -219,6 +220,25 @@ static int memfd_add_seals(struct file *file, > unsigned int seals) > > > > } > > > > } > > > > > > > > + if ((seals & F_SEAL_FUTURE_WRITE) && > > > > + !(*file_seals & F_SEAL_FUTURE_WRITE)) { > > > > + /* > > > > + * The FUTURE_WRITE seal also prevents growing and > shrinking > > > > + * so we need them to be already set, or requested > now. > > > > + */ > > > > + int test_seals = (seals | *file_seals) & > > > > + (F_SEAL_GROW | F_SEAL_SHRINK); > > > > + > > > > + if (test_seals != (F_SEAL_GROW | F_SEAL_SHRINK)) { > > > > + error = -EINVAL; > > > > + goto unlock; > > > > + } > > > > + > > > > + spin_lock(&file->f_lock); > > > > + file->f_mode &= ~(FMODE_WRITE | FMODE_PWRITE); > > > > + spin_unlock(&file->f_lock); > > > > + } > > > > > > So you're fiddling around with the file, but not the inode? How are > > > you preventing code like the following from re-opening the file as > > > writable? > > > > > > $ cat memfd.c > > > #define _GNU_SOURCE > > > #include > > > #include > > > #include > > > #include > > > #include > > > #include > > > > > > int main(void) { > > > int fd = syscall(__NR_memfd_create, "testfd", 0); > > > if (fd == -1) err(1, "memfd"); > > > char path[100]; > > > sprintf(path, "/proc/self/fd/%d", fd); > > > int fd2 = open(path, O_RDWR); > > > if (fd2 == -1) err(1, "reopen"); > > > printf("reopen successful: %d\n", fd2); > > > } > > > $ gcc -o memfd memfd.c > > > $ ./memfd > > > reopen successful: 4 > > > $ > > > > > > That aside: I wonder whether a better API would be something that > > > allows you to create a new readonly file descriptor, instead of > > > fiddling with the writability of an existing fd. > > > > My favorite approach would be to forbid open() on memfds, hope that > > nobody notices the tiny API break, and then add an ioctl for "reopen > > this memfd with reduced permissions" - but that's just my personal > > opinion. > > I did something along these lines and it fixes the issue, but I forbid open > of memfd only when the F_SEAL_FUTURE_WRITE seal is in place. So then its > not > an ABI break because this is a brand new seal. That seems the least > intrusive > solution and it works. Do you mind testing it and I'll add your and > Tested-by > to the new fix? The patch is based on top of this series. > Please don't forbid reopens entirely. You're taking a feature that works generally (reopens) and breaking it in one specific case (memfd write sealed files). The open modes are available in .open in the struct file: you can deny *only* opens for write instead of denying reopens generally. --000000000000e74e36057a4e94da Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable

On Friday, November 9, 2018, Joel Fernandes <joel@joelfernandes.org> wrote:
On Fri, Nov 09, 2018 at 10:19:03PM +0100, Jann Horn wro= te:
> On Fri, Nov 9, 2018 at 10:06 PM Jann Horn <jannh@google.com> wrote:
> > On Fri, Nov 9, 2018 at 9:46 PM Joel Fernandes (Google)
> > <joel@joelfernandes.= org> wrote:
> > > Android uses ashmem for sharing memory regions. We are looki= ng forward
> > > to migrating all usecases of ashmem to memfd so that we can = possibly
> > > remove the ashmem driver in the future from staging while al= so
> > > benefiting from using memfd and contributing to it. Note sta= ging drivers
> > > are also not ABI and generally can be removed at anytime. > > >
> > > One of the main usecases Android has is the ability to creat= e a region
> > > and mmap it as writeable, then add protection against making= any
> > > "future" writes while keeping the existing already= mmap'ed
> > > writeable-region active.=C2=A0 This allows us to implement a= usecase where
> > > receivers of the shared memory buffer can get a read-only vi= ew, while
> > > the sender continues to write to the buffer.
> > > See CursorWindow documentation in Android for more details:<= br> > > > https://developer.android.com/= reference/android/database/CursorWindow
> > >
> > > This usecase cannot be implemented with the existing F_SEAL_= WRITE seal.
> > > To support the usecase, this patch adds a new F_SEAL_FUTURE_= WRITE seal
> > > which prevents any future mmap and write syscalls from succe= eding while
> > > keeping the existing mmap active.
> >
> > Please CC linux-api@ on patches like this. If you had done that, = I
> > might have criticized your v1 patch instead of your v3 patch... > >
> > > The following program shows the seal
> > > working in action:
> > [...]
> > > Cc: jreck@google.com=
> > > Cc: john.stultz@li= naro.org
> > > Cc: tkjos@google.com=
> > > Cc: gregkh@lin= uxfoundation.org
> > > Cc: hch@infradead.org
> > > Reviewed-by: John Stultz <john.stultz@linaro.org>
> > > Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
> > > ---
> > [...]
> > > diff --git a/mm/memfd.c b/mm/memfd.c
> > > index 2bb5e257080e..5ba9804e9515 100644
> > > --- a/mm/memfd.c
> > > +++ b/mm/memfd.c
> > [...]
> > > @@ -219,6 +220,25 @@ static int memfd_add_seals(struct file = *file, unsigned int seals)
> > >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0}
> > >=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0}
> > >
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0if ((seals & F_SEAL_FUTURE_W= RITE) &&
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0!(*file_seals &= ; F_SEAL_FUTURE_WRITE)) {
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0/* > > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 * T= he FUTURE_WRITE seal also prevents growing and shrinking
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 * s= o we need them to be already set, or requested now.
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 */<= br> > > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0int = test_seals =3D (seals | *file_seals) &
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 (F_SEAL_GROW | F_SEAL_= SHRINK);
> > > +
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0if (= test_seals !=3D (F_SEAL_GROW | F_SEAL_SHRINK)) {
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0error =3D -EINVAL;
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0goto unlock;
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0} > > > +
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0spin= _lock(&file->f_lock);
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0file= ->f_mode &=3D ~(FMODE_WRITE | FMODE_PWRITE);
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0spin= _unlock(&file->f_lock);
> > > +=C2=A0 =C2=A0 =C2=A0 =C2=A0}
> >
> > So you're fiddling around with the file, but not the inode? H= ow are
> > you preventing code like the following from re-opening the file a= s
> > writable?
> >
> > $ cat memfd.c
> > #define _GNU_SOURCE
> > #include <unistd.h>
> > #include <sys/syscall.h>
> > #include <printf.h>
> > #include <fcntl.h>
> > #include <err.h>
> > #include <stdio.h>
> >
> > int main(void) {
> >=C2=A0 =C2=A0int fd =3D syscall(__NR_memfd_create, "testfd&qu= ot;, 0);
> >=C2=A0 =C2=A0if (fd =3D=3D -1) err(1, "memfd");
> >=C2=A0 =C2=A0char path[100];
> >=C2=A0 =C2=A0sprintf(path, "/proc/self/fd/%d", fd);
> >=C2=A0 =C2=A0int fd2 =3D open(path, O_RDWR);
> >=C2=A0 =C2=A0if (fd2 =3D=3D -1) err(1, "reopen");
> >=C2=A0 =C2=A0printf("reopen successful: %d\n", fd2);
> > }
> > $ gcc -o memfd memfd.c
> > $ ./memfd
> > reopen successful: 4
> > $
> >
> > That aside: I wonder whether a better API would be something that=
> > allows you to create a new readonly file descriptor, instead of > > fiddling with the writability of an existing fd.
>
> My favorite approach would be to forbid open() on memfds, hope that > nobody notices the tiny API break, and then add an ioctl for "reo= pen
> this memfd with reduced permissions" - but that's just my per= sonal
> opinion.

I did something along these lines and it fixes the issue, but I forbid open=
of memfd only when the F_SEAL_FUTURE_WRITE seal is in place. So then its no= t
an ABI break because this is a brand new seal. That seems the least intrusi= ve
solution and it works. Do you mind testing it and I'll add your and Tes= ted-by
to the new fix? The patch is based on top of this series.
=C2=A0
Please don't forbid reopens entirely= . You're taking a feature that works generally (reopens) and breaking i= t in one specific case (memfd write sealed files). The open modes are avail= able in .open in the struct file: you can deny *only* opens for write inste= ad of denying reopens generally.
--000000000000e74e36057a4e94da--