Bump aead crate dependency to v0.5

[tarcieri]

Release PR: RustCrypto/traits#1058

[codecov-commenter]

Codecov Report

Merging #444 (398134e) into master (0b24c8d) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #444   +/-   ##
=======================================
  Coverage   87.21%   87.21%           
=======================================
  Files          40       40           
  Lines        1955     1955           
=======================================
  Hits         1705     1705           
  Misses        250      250           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0b24c8d...398134e. Read the comment docs.

[roy-work]

It would be nice to get a more specific callout in the CHANGELOG about this one, specifically, what I as a dev need to do around this in order to update (to aes-siv 0.7).

This PR makes it seem like it is just a dependency update, but this is a breaking change in aes-siv's interface.

[tarcieri]

@roy-work did you follow the link in the description? RustCrypto/traits#1058

As noted there, the only major change is replacing NewAead with KeyInit. The two traits are nearly identical, so it should just be s/NewAead/KeyInit/.

[roy-work]

@tarcieri I did, and I did make that replacement; there is still other breakage:

error[E0284]: type annotations needed: cannot satisfy `<_ as KeySizeUser>::KeySize == UInt<UInt<UInt<UInt<UInt<UInt<UInt<UTerm, B1>, B0>, B0>, B0>, B0>, B0>, B0>`
   |
79 |     let key = Key::from_slice(self.key);
   |               ^^^ cannot satisfy `<_ as KeySizeUser>::KeySize == UInt<UInt<UInt<UInt<UInt<UInt<UInt<UTerm, B1>, B0>, B0>, B0>, B0>, B0>, B0>`

which was introduced at 0.7, so it isn't as simple as just updating the use statements.

KeyInit seems to call it new_from_slice, but that doesn't seem to work, either:

error[E0599]: the function or associated item `new_from_slice` exists for struct `GenericArray<u8, _>`, but its trait bounds were not satisfied
    |
79  |     let key = Key::new_from_slice(self.key);
    |                    ^^^^^^^^^^^^^^ function or associated item cannot be called on `GenericArray<u8, _>` due to unsatisfied trait bounds
    |
   ::: /[…]/.cargo/registry/src/-662ab3103e52a138/generic-array-0.14.5/src/lib.rs:172:1
    |
172 | pub struct GenericArray<T, U: ArrayLength<T>> {
    | ---------------------------------------------
    | |
    | doesn't satisfy `GenericArray<u8, _>: KeyInit`
    | doesn't satisfy `GenericArray<u8, _>: crypto_common::InnerInit`
    | doesn't satisfy `GenericArray<u8, _>: crypto_common::InnerUser`
    |
    = note: the following trait bounds were not satisfied:
            `GenericArray<u8, _>: crypto_common::InnerInit`
            which is required by `GenericArray<u8, _>: KeyInit`
            `GenericArray<u8, _>: crypto_common::InnerUser`
            which is required by `GenericArray<u8, _>: KeyInit`
            `&GenericArray<u8, _>: crypto_common::InnerInit`
            which is required by `&GenericArray<u8, _>: KeyInit`
            `&GenericArray<u8, _>: crypto_common::InnerUser`
            which is required by `&GenericArray<u8, _>: KeyInit`
            `&mut GenericArray<u8, _>: crypto_common::InnerInit`
            which is required by `&mut GenericArray<u8, _>: KeyInit`
            `&mut GenericArray<u8, _>: crypto_common::InnerUser`
            which is required by `&mut GenericArray<u8, _>: KeyInit`

(The arg here, self.key, is a &[u8]. Nothing too surpising, I think…?)

[tarcieri]

KeyInit isn't impl'd for Key. It's impl'd for the Aes*SivAead types, i.e. the types NewAead was previously impl'd for. Take a look at the example:

https://docs.rs/aes-siv/latest/aes_siv/#usage

Key is an alias for GenericArray which requires a type parameter:

https://docs.rs/aes-siv/latest/aes_siv/#usage

You need to explicitly specify the cipher the key is for, e.g. Key::<Aes256SivAead>

[roy-work]

You need to explicitly specify the cipher the key is for, e.g. Key::<Aes256SivAead>

That works, thank you! (The need for that is apparently new, though.)

Although, the two lines here are…

let key = Key::<Aes256SivAead>::from_slice(self.key);
Aes256SivAead::new(key)

Which seems equivalent to:

Aes256SivAead::new_from_slice(&self.key).unwrap()

… which I might leave it as.

(This code has been around since miscreant. This is getting dragged in with the recent git2 vuln., for me. cargo audit depends on both git2 and this crate, and so fixing the git2 vuln. pulls in a new aes-siv. That in turn conflicts with another crate of ours, due to aes-siv's MSRV requirements preventing two version from co-existing.)