Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20200609060137.143501-1-daeho43@gmail.com> <20200609165107.GA228564@gmail.com>
 <CACOAw_xEZ+au9yhFerq9amkRO62Dzxj7h71gEc=i16ReYu5xrg@mail.gmail.com>
 <20200610031532.GA6286@sol.localdomain> <CACOAw_wErOPC=Kf3UU8nFGhWRy84ZnCeJbsyPhSCcXv51B_XxQ@mail.gmail.com>
In-Reply-To: <CACOAw_wErOPC=Kf3UU8nFGhWRy84ZnCeJbsyPhSCcXv51B_XxQ@mail.gmail.com>
From:   Daeho Jeong <daeho43@gmail.com>
Date:   Thu, 11 Jun 2020 08:31:20 +0900
Message-ID: <CACOAw_zka6d06RxFOUTwEV7B6o8A2-_6FvqWh_A1nJ0+7FU9yQ@mail.gmail.com>
Subject: Re: [f2fs-dev] [PATCH] f2fs: add F2FS_IOC_SEC_TRIM_FILE ioctl
To:     Eric Biggers <ebiggers@kernel.org>
Cc:     linux-kernel@vger.kernel.org,
        linux-f2fs-devel@lists.sourceforge.net, kernel-team@android.com,
        Daeho Jeong <daehojeong@google.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

> > > > > +
> > > > > +     if (f2fs_readonly(sbi->sb))
> > > > > +             return -EROFS;
> > > >
> > > > Isn't this redundant with mnt_want_write_file()?
> > > >
> > > > Also, shouldn't write access to the file be required, i.e.
> > > > (filp->f_mode & FMODE_WRITE)?  Then the f2fs_readonly() and
> > > > mnt_want_write_file() checks would be unnecessary.
> > > >
> > >
> > > Using FMODE_WRITE is more proper for this case, since we're going to
> > > modify the data. But I think mnt_want_write_file() is still required
> > > to prevent the filesystem from freezing or something else.
> >
> > Right, the freezing check is actually still necessary.  But getting wri=
te access
> > to the mount is not necessary.  I think you should use file_start_write=
() and
> > file_end_write(), like vfs_write() does.

I've checked this again.

But I think mnt_want_write_file() looks better than the combination of
checking FMODE_WRITE and file_start_write(), because
mnt_want_write_file() handles all the things we need.
It checks FMODE_WRITER, which is set in do_dentry_open() when
FMODE_WRITE is already set, and does the stuff that file_start_write()
is doing. This is why the other filesystem system calls use it.

What do you think?

2020=EB=85=84 6=EC=9B=94 10=EC=9D=BC (=EC=88=98) =EC=98=A4=ED=9B=84 12:55, =
Daeho Jeong <daeho43@gmail.com>=EB=8B=98=EC=9D=B4 =EC=9E=91=EC=84=B1:
>
> > >
> > > To prevent the file data from garbage collecting, the user needs to
> > > use pinfile ioctl and fallocate system call after creating the file.
> > > The sequence is like below.
> > > 1. create an empty file
> > > 2. pinfile
> > > 3. fallocate()
> >
> > Is that persistent?  So the file will never be moved afterwards?
> >
> > Is there a place where this is (or should be) documented?
>
> Yes, this is persistent. F2FS_IOC_SET_PIN_FILE ioctl is to prevent
> file data from moving and being garbage collected, and further update
> to the file will be handled in in-place update manner.
> I don't see any document on this, but you can find the below in
> Documentation/filesystems/f2fs.rst
>
> However, once F2FS receives ioctl(fd, F2FS_IOC_SET_PIN_FILE) in prior to
> fallocate(fd, DEFAULT_MODE), it allocates on-disk blocks addresses having
> zero or random data, which is useful to the below scenario where:
>
>  1. create(fd)
>  2. ioctl(fd, F2FS_IOC_SET_PIN_FILE)
>  3. fallocate(fd, 0, 0, size)
>  4. address =3D fibmap(fd, offset)
>  5. open(blkdev)
>  6. write(blkdev, address)
>
> > Right, the freezing check is actually still necessary.  But getting wri=
te access
> > to the mount is not necessary.  I think you should use file_start_write=
() and
> > file_end_write(), like vfs_write() does.
>
> Yes, agreed.
>
> 2020=EB=85=84 6=EC=9B=94 10=EC=9D=BC (=EC=88=98) =EC=98=A4=ED=9B=84 12:15=
, Eric Biggers <ebiggers@kernel.org>=EB=8B=98=EC=9D=B4 =EC=9E=91=EC=84=B1:
> >
> > On Wed, Jun 10, 2020 at 11:05:46AM +0900, Daeho Jeong wrote:
> > > > > Added a new ioctl to send discard commands or/and zero out
> > > > > to whole data area of a regular file for security reason.
> > > >
> > > > With this ioctl available, what is the exact procedure to write and=
 then later
> > > > securely erase a file on f2fs?  In particular, how can the user pre=
vent f2fs
> > > > from making multiple copies of file data blocks as part of garbage =
collection?
> > > >
> > >
> > > To prevent the file data from garbage collecting, the user needs to
> > > use pinfile ioctl and fallocate system call after creating the file.
> > > The sequence is like below.
> > > 1. create an empty file
> > > 2. pinfile
> > > 3. fallocate()
> >
> > Is that persistent?  So the file will never be moved afterwards?
> >
> > Is there a place where this is (or should be) documented?
> >
> > > > > +
> > > > > +     if (f2fs_readonly(sbi->sb))
> > > > > +             return -EROFS;
> > > >
> > > > Isn't this redundant with mnt_want_write_file()?
> > > >
> > > > Also, shouldn't write access to the file be required, i.e.
> > > > (filp->f_mode & FMODE_WRITE)?  Then the f2fs_readonly() and
> > > > mnt_want_write_file() checks would be unnecessary.
> > > >
> > >
> > > Using FMODE_WRITE is more proper for this case, since we're going to
> > > modify the data. But I think mnt_want_write_file() is still required
> > > to prevent the filesystem from freezing or something else.
> >
> > Right, the freezing check is actually still necessary.  But getting wri=
te access
> > to the mount is not necessary.  I think you should use file_start_write=
() and
> > file_end_write(), like vfs_write() does.
> >
> > > >
> > > > > +
> > > > > +     if (get_user(flags, (u32 __user *)arg))
> > > > > +             return -EFAULT;
> > > > > +     if (!(flags & F2FS_TRIM_FILE_MASK))
> > > > > +             return -EINVAL;
> > > >
> > > > Need to reject unknown flags:
> > > >
> > > >         if (flags & ~F2FS_TRIM_FILE_MASK)
> > > >                 return -EINVAL;
> > >
> > > I needed a different thing here. This was to check neither discard no=
r
> > > zeroing out are not here. But we still need to check unknown flags,
> > > too.
> > > The below might be better.
> > > if (!flags || flags & ~F2FS_TRIM_FILE_MASK)
> > >        return -EINVAL;
> >
> > Sure, but please put parentheses around the second clause:
> >
> >         if (flags =3D=3D 0 || (flags & ~F2FS_TRIM_FILE_MASK))
> >                 return -EINVAL;
> >
> > - Eric