Received-SPF: pass (google.com: domain of linux-kernel+bounces-144843-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Date: Mon, 15 Apr 2024 09:36:56 +0000
To: Alice Ryhl <aliceryhl@google.com>, Miguel Ojeda <ojeda@kernel.org>, Matthew Wilcox <willy@infradead.org>, Al Viro <viro@zeniv.linux.org.uk>, Andrew Morton <akpm@linux-foundation.org>, Kees Cook <keescook@chromium.org>
From: Benno Lossin <benno.lossin@proton.me>
Cc: Alex Gaynor <alex.gaynor@gmail.com>, Wedson Almeida Filho <wedsonaf@gmail.com>, Boqun Feng <boqun.feng@gmail.com>, Gary Guo <gary@garyguo.net>, =?utf-8?Q?Bj=C3=B6rn_Roy_Baron?= <bjorn3_gh@protonmail.com>, Andreas Hindborg <a.hindborg@samsung.com>, Greg Kroah-Hartman <gregkh@linuxfoundation.org>, =?utf-8?Q?Arve_Hj=C3=B8nnev=C3=A5g?= <arve@android.com>, Todd Kjos <tkjos@android.com>, Martijn Coenen <maco@android.com>, Joel Fernandes <joel@joelfernandes.org>, Carlos Llamas <cmllamas@google.com>, Suren Baghdasaryan <surenb@google.com>, Arnd Bergmann <arnd@arndb.de>, linux-mm@kvack.org, linux-kernel@vger.kernel.org, rust-for-linux@vger.kernel.org, Christian Brauner <brauner@kernel.org>
Subject: Re: [PATCH v5 1/4] rust: uaccess: add userspace pointers
Message-ID: <2cae6fd4-906c-44ad-88be-0dfed090d07c@proton.me>
In-Reply-To: <20240415-alice-mm-v5-1-6f55e4d8ef51@google.com>
References: <20240415-alice-mm-v5-0-6f55e4d8ef51@google.com> <20240415-alice-mm-v5-1-6f55e4d8ef51@google.com>
Feedback-ID: 71780778:user:proton
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

On 15.04.24 09:13, Alice Ryhl wrote:
> +impl UserSlice {
> +    /// Constructs a user slice from a raw pointer and a length in bytes=
.
> +    ///
> +    /// Constructing a [`UserSlice`] performs no checks on the provided =
address and length, it can
> +    /// safely be constructed inside a kernel thread with no current use=
rspace process. Reads and
> +    /// writes wrap the kernel APIs `copy_from_user` and `copy_to_user`,=
 which check the memory map
> +    /// of the current process and enforce that the address range is wit=
hin the user range (no
> +    /// additional calls to `access_ok` are needed).
> +    ///
> +    /// Callers must be careful to avoid time-of-check-time-of-use (TOCT=
OU) issues. The simplest way
> +    /// is to create a single instance of [`UserSlice`] per user memory =
block as it reads each byte
> +    /// at most once.
> +    pub fn new(ptr: *mut c_void, length: usize) -> Self {

What would happen if I call this with a kernel pointer and then
read/write to it? For example

     let mut arr =3D [MaybeUninit::uninit(); 64];
     let ptr: *mut [MaybeUninit<u8>] =3D &mut arr;
     let ptr =3D ptr.cast::<c_void>();

     let slice =3D UserSlice::new(ptr, 64);
     let (mut r, mut w) =3D slice.reader_writer();

     r.read_raw(&mut arr)?;
     // SAFETY: `arr` was initialized above.
     w.write_slice(unsafe { MaybeUninit::slice_assume_init_ref(&arr) })?;

I think this would violate the exclusivity of `&mut` without any
`unsafe` code. (the `unsafe` block at the end cannot possibly be wrong)

> +        UserSlice { ptr, length }
> +    }

[...]

> +    /// Returns `true` if no data is available in the io buffer.
> +    pub fn is_empty(&self) -> bool {
> +        self.length =3D=3D 0
> +    }
> +
> +    /// Reads raw data from the user slice into a kernel buffer.
> +    ///
> +    /// After a successful call to this method, all bytes in `out` are i=
nitialized.

I think we should put things like this into a `# Guarantees` section.

--=20
Cheers,
Benno

> +    ///
> +    /// Fails with `EFAULT` if the read happens on a bad address.
> +    pub fn read_raw(&mut self, out: &mut [MaybeUninit<u8>]) -> Result {
> +        let len =3D out.len();
> +        let out_ptr =3D out.as_mut_ptr().cast::<c_void>();
> +        if len > self.length {
> +            return Err(EFAULT);
> +        }
> +        let Ok(len_ulong) =3D c_ulong::try_from(len) else {
> +            return Err(EFAULT);
> +        };
> +        // SAFETY: `out_ptr` points into a mutable slice of length `len_=
ulong`, so we may write
> +        // that many bytes to it.
> +        let res =3D unsafe { bindings::copy_from_user(out_ptr, self.ptr,=
 len_ulong) };
> +        if res !=3D 0 {
> +            return Err(EFAULT);
> +        }
> +        // Userspace pointers are not directly dereferencable by the ker=
nel, so we cannot use `add`,
> +        // which has C-style rules for defined behavior.
> +        self.ptr =3D self.ptr.wrapping_byte_add(len);
> +        self.length -=3D len;
> +        Ok(())
> +    }