This PR implements a `non_secure_erase()` method on SecretKey,
KeyPair, SharedSecret, Scalar, and DisplaySecret. The purpose
of this method is to (attempt to) overwrite secret data with
valid default values. This method can be used by libraries
to implement Zeroize on structs containing secret values.
`non_secure_erase()` attempts to avoid being optimized away or
reordered using the same mechanism as the zeroize crate: first,
using `std::ptr::write_volatile` (which will not be optimized
away) to overwrite the memory, then using a memory fence to
prevent subtle issues due to load or store reordering.
Note, however, that this method is *very unlikely* to do anything
useful on its own. Effective use involves carefully placing these
values inside non-Copy structs and pinning those structs in place.
See the [`zeroize`](https://docs.rs/zeroize) documentation for tips
and tricks, and for further discussion.
[this commit includes a squashed-in commit from tcharding to fix docs
and helpful suggestions from apoelstra and Kixunil]
Currently we have a few problems with our feature gating, attempt to
audit all feature gating and fix it by doing:
1. Do not enable features on optional dependencies (`rand` and
`bitcoin-hashes`) in dev-dependencies, doing so hides broken feature
gating in unit tests.
2. Do not use unnecessary feature combinations when one feature enables
another e.g. `any(feature = "std", feature = "alloc")`.
3. Enable "std" from "rand-std" and "bitcoin-std" (and fix features
gating as for point 2).
4. Clean up code around `rand::thread_rng`, this is part of this patch
because `thread_rng` requires the "rand-std" feature.
5. Clean up CI test script to test each feature individually now that
"rand-std" and "bitcoin-hashes-std" enable "std".
Run the command `cargo +nightly fmt` to fix formatting issues.
The formatter got confused in one place, adding an incorrect
indentation, this was manually fixed.
In multiple places we use array constants for zero and one. Add two
constants and use them throughout the codebase. Note the endian-ness of
`ONE` in the docs.
This adds `Scalar` newtype to better represent values accepted by
tweaking functions. This type is always 32-bytes and guarantees being
within curve order.
kwantam
Arik Sosman
Tobin C. Harding
Martin Habovstiak