Message-ID: <CAH5fLgj8WkjDs6qCmzxYsYTH=_knWyZDrwqLopvGZEBk931aXQ@mail.gmail.com>
Date: Sat, 31 May 2025 23:09:43 +0200
From: Alice Ryhl <aliceryhl@...gle.com>
To: Benno Lossin <lossin@...nel.org>
Cc: Miguel Ojeda <ojeda@...nel.org>, Andrew Morton <akpm@...ux-foundation.org>, 
	Alexander Viro <viro@...iv.linux.org.uk>, Greg Kroah-Hartman <gregkh@...uxfoundation.org>, 
	Boqun Feng <boqun.feng@...il.com>, Gary Guo <gary@...yguo.net>, 
	Björn Roy Baron <bjorn3_gh@...tonmail.com>, 
	Benno Lossin <benno.lossin@...ton.me>, Andreas Hindborg <a.hindborg@...nel.org>, 
	Trevor Gross <tmgross@...ch.edu>, Danilo Krummrich <dakr@...nel.org>, rust-for-linux@...r.kernel.org, 
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH v4 2/2] uaccess: rust: add UserSliceReader::strcpy_into_buf

On Sat, May 31, 2025 at 10:38 PM Benno Lossin <lossin@...nel.org> wrote:
>
> On Sat May 31, 2025 at 7:38 PM CEST, Alice Ryhl wrote:
> > On Sat, May 31, 2025 at 5:25 PM Benno Lossin <lossin@...nel.org> wrote:
> >> On Sat May 31, 2025 at 3:25 PM CEST, Alice Ryhl wrote:
> >> > On Fri, May 30, 2025 at 8:16 PM Benno Lossin <lossin@...nel.org> wrote:
> >> >> On Tue May 27, 2025 at 2:34 PM CEST, Alice Ryhl wrote:
> >> >> > This patch adds a more convenient method for reading C strings from
> >> >> > userspace. Logic is added to NUL-terminate the buffer when necessary so
> >> >> > that a &CStr can be returned.
> >> >> >
> >> >> > Note that we treat attempts to read past `self.length` as a fault, so
> >> >> > this returns EFAULT if that limit is exceeded before `buf.len()` is
> >> >> > reached.
> >> >> >
> >> >> > Reviewed-by: Danilo Krummrich <dakr@...nel.org>
> >> >> > Signed-off-by: Alice Ryhl <aliceryhl@...gle.com>
> >> >> > ---
> >> >> >  rust/kernel/uaccess.rs | 56 +++++++++++++++++++++++++++++++++++++++++++++++++-
> >> >> >  1 file changed, 55 insertions(+), 1 deletion(-)
> >> >> >
> >> >> > diff --git a/rust/kernel/uaccess.rs b/rust/kernel/uaccess.rs
> >> >> > index 9b1e4016fca2c25a44a8417c7e35e0fcf08aa959..e6534b52a1920254d61f8349426d4cdb38286089 100644
> >> >> > --- a/rust/kernel/uaccess.rs
> >> >> > +++ b/rust/kernel/uaccess.rs
> >> >> > @@ -293,6 +293,61 @@ pub fn read_all<A: Allocator>(mut self, buf: &mut Vec<u8, A>, flags: Flags) -> R
> >> >> >          unsafe { buf.set_len(buf.len() + len) };
> >> >> >          Ok(())
> >> >> >      }
> >> >> > +
> >> >> > +    /// Read a NUL-terminated string from userspace and return it.
> >> >> > +    ///
> >> >> > +    /// The string is read into `buf` and a NUL-terminator is added if the end of `buf` is reached.
> >> >> > +    /// Since there must be space to add a NUL-terminator, the buffer must not be empty. The
> >> >> > +    /// returned `&CStr` points into `buf`.
> >> >> > +    ///
> >> >> > +    /// Fails with [`EFAULT`] if the read happens on a bad address (some data may have been
> >> >> > +    /// copied).
> >> >> > +    #[doc(alias = "strncpy_from_user")]
> >> >> > +    pub fn strcpy_into_buf<'buf>(self, buf: &'buf mut [u8]) -> Result<&'buf CStr> {
> >> >> > +        if buf.is_empty() {
> >> >> > +            return Err(EINVAL);
> >> >> > +        }
> >> >> > +
> >> >> > +        // SAFETY: The types are compatible and `strncpy_from_user` doesn't write uninitialized
> >> >> > +        // bytes to `buf`.
> >> >> > +        let mut dst = unsafe { &mut *(buf as *mut [u8] as *mut [MaybeUninit<u8>]) };
> >> >> > +
> >> >> > +        // We never read more than `self.length` bytes.
> >> >> > +        if dst.len() > self.length {
> >> >> > +            dst = &mut dst[..self.length];
> >> >> > +        }
> >> >> > +
> >> >> > +        let mut len = raw_strncpy_from_user(dst, self.ptr)?;
> >> >> > +        if len < dst.len() {
> >> >> > +            // Add one to include the NUL-terminator.
> >> >> > +            //
> >> >> > +            // This means that we could not fill the entire buffer, but we had to stop reading
> >> >> > +            // because we hit the `self.length` limit of this `UserSliceReader`. Since we did not
> >> >> > +            // fill the buffer, we treat this case as if we tried to read past the `self.length`
> >> >> > +            // limit and received a page fault, which is consistent with other `UserSliceReader`
> >> >> > +            // methods that also return page faults when you exceed `self.length`.
> >> >> > +            return Err(EFAULT);
> >> >> > +        } else {
> >> >> > +            // This implies that len == buf.len().
> >> >> > +            //
> >> >> > +            // This means that we filled the buffer exactly. In this case, we add a NUL-terminator
> >> >> > +            // and return it. Unlike the `len < dst.len()` branch, don't modify `len` because it
> >> >> > +            // already represents the length including the NUL-terminator.
> >> >> > +            //
> >> >> > +            // SAFETY: Due to the check at the beginning, the buffer is not empty.
> >> >> > +            unsafe { *buf.last_mut().unwrap_unchecked() = 0 };
> >> >>
> >> >> In this case you're overwriting the last character read. Should we give
> >> >> `raw_strncpy_from_user` access to one less byte and then write NUL into
> >> >> that?
> >> >
> >> > Why? I'm not interested in changing the implementation just because.
> >> > It needs to be significantly simpler, and I do not think it is.
> >>
> >> Sure, but then I think we should document this behavior.
> >
> > Document what? I understood your suggestion as a change to the
> > implementation of strcpy_into_buf that would not change its behavior.
> > Did I misunderstand?
>
> Maybe I misunderstood the code, but if you do this:
>
>     let slice = UserSlice::new(ptr, 1024);
>     let mut buf = [0; 42];
>     let s = slice.strcpy_into_buf(&mut buf)?;
>
> Then it will read 42 characters from userspace and (if there was no nul
> byte) overwrite the last character with `\0`. If we now do
>
>     let mut buf2 = [0; 42];
>     let s2 = slice.strcpy_into_buf(&mut buf2)?;
>
> Then that will continue the read at index 42, but effectively one
> character will get skipped.
>
> (Now it's not possible to call `strcpy_into_buf` multiple times, but I
> see no real reason why it isn't a `&mut self` method. Also a user could
> call `clone_reader` and then manually `skip` 42 bytes. Although they
> might only skip 41 bytes, since that's the length of the CStr. But that
> runs into the problem that if there was a `\0` at index 41, then
> repeated uses of the pattern above will yield empty strings.)

I removed the ability to call it multiple times to avoid dealing with
this kind of question. You may submit a follow-up patch to change it
if you have a use-case.

Alice

Powered by blists - more mailing lists