DER encoded BIT STRING produces BitString with incorrect length

[benmaddison]

Decoding a DER-encoded BIT STRING with a non-zero "unused bits" byte into a types::BitString results in a value padded with 0s to the nearest byte.

This makes it impossible to tell the difference between values with "real" trailing zeros and those without.

In particular, RFC3779 and other RPKI-related standards use a BIT STRING to encode IP prefixes, indicating the length of the prefix through the bit-length of the BIT STRING. This issue makes it impossible to correctly decode IP prefixes with lengths that are not a multiple of 8.

Note that I have only tried this with the DER implementation. I do not know whether the other codecs suffer from a similar issue.

The issue can be reproduced with the following minimal example:

use bitvec::{bitvec, order::Msb0};
use rasn::{der, types::BitString};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    // DER-encoded byte-slice of BIT STRING B'0101'
    let encoded: &[u8] = &[
        0x03,        // tag = 3
        0x02,        // length = 2
        0x04,        // unused bits = 4
        0b0101_0000, // value
    ];
    let bit_string = der::decode::<BitString>(encoded)?;

    // Expected value
    let expected = bitvec![u8, Msb0; 0, 1, 0, 1];

    assert_eq!(bit_string, expected);

    Ok(())
}

I will take a look at the relevant code, and propose a fix if I can grok it easily enough.

[XAMPPRocky]

Thank you for your issue! Just to check. Have you tested with the latest version? I'm pretty sure I fixed this last week 39a9f47

[XAMPPRocky]

Ah nvm, it seems I didn't add back in the intentional bit string truncation. So it should be simple to just adjust the decode and encode to do that.

Note that I have only tried this with the DER implementation. I do not know whether the other codecs suffer from a similar issue.

This is only relevant for BER, PER bit strings don't have the concept of extra bits since it is a bit-level codec.

[benmaddison]

@XAMPPRocky have attempted a fix in #135.
Let me know your thoughts when you have a sec.
I have tested on my project, and this fixes the bug that lead me here.