Record::read : Custom Read on uninitialized buffer may cause UB

[JOE1994]

Hello 🦀,
we (Rust group @sslab-gatech) found a memory-safety/soundness issue in this crate while scanning Rust code on crates.io for potential vulnerabilities.

Issue Description

rust-marc/src/lib.rs

Lines 122 to 129 in e94b984

	let len = try!(misc::read_dec_5(&*data));
	if len < 5 {
	return Err(ErrorKind::RecordTooShort(len).into());
	}
	data.reserve(len - 5);
	unsafe { data.set_len(len) };
	try!(input.read_exact(&mut data[5..len]).chain_err(|| ErrorKind::UnexpectedEof));

Record::read method creates an uninitialized buffer and passes it to user-provided Read implementation. This is unsound, because it allows safe Rust code to exhibit an undefined behavior (read from uninitialized memory).

This part from the Read trait documentation explains the issue:

It is your responsibility to make sure that buf is initialized before calling read. Calling read with an uninitialized buf (of the kind one obtains via MaybeUninit<T>) is not safe, and can lead to undefined behavior.

Another question I have regarding Record::read():
Right before data.set_len(len), why data.reserve(len - 5) instead of data.reserve(len)? At the momment it seems unsound to me :(

How to fix the issue?

The Naive & safe way to fix the issue is to always zero-initialize a buffer before lending it to a user-provided Read implementation. Note that this approach will add runtime performance overhead of zero-initializing the buffer.

As of Jan 2021, there is not yet an ideal fix that works in stable Rust with no performance overhead. Below are links to relevant discussions & suggestions for the fix.

• Well-written document regarding the issue
• Rust RFC 2930
• nightly feature std::io::Initializer
• Discussion in Rust Internals Forum

[blackbeam]

Hi, @JOE1994.

Record::read method creates an uninitialized buffer and passes it to user-provided Read implementation. This is unsound, because it allows safe Rust code to exhibit an undefined behavior (read from uninitialized memory).

This is definitely an issue. Thanks!

Another question I have regarding Record::read():
Right before data.set_len(len), why data.reserve(len - 5) instead of data.reserve(len)?

Total length of a record is len and 5 bytes of capacity we already have, therefore we need to reserve space for len - 5 additional bytes (see std::vec::Vec::reserve).