[bitcoin-dev] Safer sighashes and more granular SIGHASH_NOINPUT

[Christian Decker]

Pieter Wuille via bitcoin-dev <bitcoin-dev at lists.linuxfoundation.org>
writes:

> On Mon, 19 Nov 2018 at 14:37, Pieter Wuille <pieter.wuille at gmail.com> wrote:
> * It's probably better to sign the amounts of all inputs, as suggested
> by Johnson Lau. As that would cause default sighashes to sign all
> input and output amounts, is there still a need to sign the tx fee
> explicitly? Or in other words, are there situations where changing the
> set of inputs or outputs after signing is desired, but the net
> difference between them cannot change? If not, that would remove the
> need for NOFEE.

So the final proposal would be to append a new `hashValues` field to the
hashed representation, with `hashValues` just being the double SHA256 of
all values? In that case SINGLE needs to blank that hash, otherwise we'd
be committing to all inputs again.

Once we have that detail, we can start thinking about what it means to
commit to the fee vs. committing to the values. Since the fee is given
by the output values and the input values we only need to consider the
cases in which they can be modified.

 - NOINPUT (as in BIP118) commits to the value (and I can't think of a
   usecase where we'd want to change that), and that transparently
   extends to all other inputs.
 - For ANYONECANPAY can't really commit to a fee anyway so ANYONECANPAY
   would likely imply NOFEE.
 - With NONE all bets are off anyway, so no need to consider that :-)
 - SINGLE is a bit special, and for value commitments it reduces to the
   current commitment to its own value, for fee commitment it's hard to
   see a use of that commitment at all afaik (I think the combination
   SINGLE|NOFEE would always be used).

> * Do we need to keep the rule that sequence values of other inputs are
> only signed with default sighash? It feels cleaner to always sign the
> sequence values of all inputs that are included in the sighash anyway
> (so all of them, unless ANYONECANPAY or NOINPUT, which would make it
> sign only the current input's sequence value). If NOINPUT also blanks
> the sequence values (as currently specified by BIP118), and all input
> amounts are signed, that would make amounts/sequence values always be
> treated identically.

Single cannot commit to other the sequence of other inputs, otherwise
we're breaking SINGLE completely. As mentioned before NOINPUT doesn't
need to blank `hashSequence`, but I'm happy to make it match if that
makes implementations handle fewer cases.

> * If MASK implies NOINPUT, and NOINPUT implies ANYONECANPAY, the 3 of
> them can be encoded in just 2 bits using the
> PARTIALSCRIPT/KNOWNSCRIPT/KNOWNTX/ALL_INPUTS encoding Anthony Towns
> suggested.

So we'd end up enumerating the combinations rather than having
independent bits for each of them? This might save us storage bits, but
it'd also result in uglier code imho, not a strong feeling but might
come back to haunt us if we ever come up with something new :-)

> So a combined proposal:
> * All existing sighash flags, plus NOINPUT and MASK
> (ANYONECANPAY/NOINPUT/MASK are encoded in 2 bits).
> * A new opcode called OP_MASKEDPUSH, whose only runtime behaviour is
> failing if not immediately followed by a push, or when appearing as
> last opcode in the script.
> * Signatures are 64 plus an optional sighash byte. A missing sighash
> byte implies ALL, and ALL cannot be specified explicitly.
> * The sighash is computed from the following:
>   * A 64-byte constant tag
>   * Data about the spending transaction:
>     * The transaction version number
>     * The hash of txins' prevouts+amounts+sequences (or nothing if ANYONECANPAY)

This needs to be partially blanked for SINGLE as well, otherwise we
break SINGLE.

>     * The hash of all txouts (or just the corresponding txout if
> SINGLE; nothing if NONE)
>     * The transaction locktime
>   * Data about the output being spent:
>     * The prevout (or nothing if NOINPUT)
>     * The amount
>     * The sequence number

I assume the sequence number here refers to the input being signed, not
the sequence number of the transaction output being spent :-) Might be
easier if we consider 3 parts: the spending transaction, the input being
signed, and the output (or TX) being spent.

Cheers,
Christian