Re: [PATCH v2 2/2] uaccess: rust: add UserSliceReader::strcpy_into_buf

[Boqun Feng <boqun.feng@gmail.com>]

On Tue, Apr 29, 2025 at 11:02:07AM -0700, Boqun Feng wrote:
> On Tue, Apr 29, 2025 at 09:02:23AM +0000, 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.
> > 
> > Signed-off-by: Alice Ryhl <aliceryhl@google.com>
> > ---
> >  rust/kernel/uaccess.rs | 35 +++++++++++++++++++++++++++++++++++
> >  1 file changed, 35 insertions(+)
> > 
> > diff --git a/rust/kernel/uaccess.rs b/rust/kernel/uaccess.rs
> > index acb703f074a30e60d42a222dd26aed80d8bdb76a..7cec1b62bd8b816f523c8be12cb29905740789fc 100644
> > --- a/rust/kernel/uaccess.rs
> > +++ b/rust/kernel/uaccess.rs
> > @@ -293,6 +293,41 @@ 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 append it to `dst`.
> 
> s/`dst`/`buf`
> 
> ?
> 
> > +    ///
> > +    /// Fails with [`EFAULT`] if the read happens on a bad address.
> > +    pub fn strcpy_into_buf<'buf>(&mut 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>]) };
> 
> maybe:
> 
> 	let mut dst = unsafe { &mut *(ptr::from_mut(buf).cast() };
> 
> ? To align with:
> 
> 	https://lore.kernel.org/rust-for-linux/20250418-ptr-as-ptr-v10-0-3d63d27907aa@gmail.com/	
> 
> > +
> > +        // 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(self.ptr, dst)?;
> > +        if len < dst.len() {
> > +            // Add one to include the NUL-terminator.
> > +            len += 1;
> > +        } else if len < buf.len() {
> > +            // We hit the `self.length` limit before `buf.len()`.
> > +            return Err(EFAULT);
> > +        } else {
> > +            // SAFETY: Due to the check at the beginning, the buffer is not empty.
> > +            unsafe { *buf.last_mut().unwrap_unchecked() = 0 };
> > +        }
> > +        self.skip(len)?;
> > +
> 
> So if the UserSlice content is "abcdefg" (not tailing NUL), and the buf
> size is 4, after a strcpy_into_buf(), the return would be a CStr "abc"
> (with a tailing NUL), and the UserSlice would move 4 bytes and become
> "edg" (not tailing NUL), is this a desired behavior?
> 
> Alternatively, we can make `dst` always 1 byte less then `buf`, so that

Hmm.. this part is not correct, what we should do is:

        // 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(self.ptr, dst)?;
        if len < dst.len() {
            // Add one to include the NUL-terminator.
            len += 1;
	    self.skip(len)?;
        } else if len < buf.len() {
            // We hit the `self.length` limit before `buf.len()`.
            return Err(EFAULT);
        } else {
            // SAFETY: Due to the check at the beginning, the buffer is not empty.
            unsafe { *buf.last_mut().unwrap_unchecked() = 0 };

	    // if any copy really happened, and we don't find a NUL char
	    // until the end of the buf/dst, we will add a NUL char as
	    // above, but in this case, we need to not skip the last
	    // char in `self` (because it's overwritten in the returning
	    // string by a NUL char).
	    if dst.len() != 0 { 
	        self.skip(len - 1)?;
	    }
        }

Of course, the code can be re-organized, but this is the idea.

Regards,
Boqun

> in the above case, UserSlice will only move 3 bytes and become "defg",
> and the return CStr is still "abc" (with a tailing NUL).
> 
> The current behavior makes me feel like we can lose some information,
> for example, if the user-kernel protocol is that "a userslice that
> contains 4 64-byte strings which don't have a tailing NUL", we cannot do
> 4 strcpy_into_buf() to get them, right? But of course, the scenario is
> completely made up, just food for thoughts.
> 
> Regards,
> Boqun
> 
> > +        // SAFETY: `raw_strncpy_from_user` guarantees that this range of bytes represents a
> > +        // NUL-terminated string with the only NUL byte being at the end.
> > +        Ok(unsafe { CStr::from_bytes_with_nul_unchecked(&buf[..len]) })
> > +    }
> >  }
> >  
> >  /// A writer for [`UserSlice`].
> > 
> > -- 
> > 2.49.0.901.g37484f566f-goog
> >