When #3303 was merged, a cyclic dependency chain was generated:
libdevice <- libcncrypto <- libringct <- libdevice
This was because libdevice needs access to a set of basic crypto operations
implemented in libringct such as scalarmultBase(), while libringct also needs
access to abstracted crypto operations implemented in libdevice such as
ecdhEncode(). To untangle this cyclic dependency chain, this patch splits libringct
into libringct_basic and libringct, where the basic crypto ops previously in
libringct are moved into libringct_basic. The cyclic dependency is now resolved
thanks to this separation:
libcncrypto <- libringct_basic <- libdevice <- libcryptonote_basic <- libringct
This eliminates the need for crypto_device.cpp and rctOps_device.cpp.
Also, many abstracted interfaces of hw::device such as encrypt_payment_id() and
get_subaddress_secret_key() were previously implemented in libcryptonote_basic
(cryptonote_format_utils.cpp) and were then called from hw::core::device_default,
which is odd because libdevice is supposed to be independent of libcryptonote_basic.
Therefore, those functions were moved to device_default.cpp.
The basic approach it to delegate all sensitive data (master key, secret
ephemeral key, key derivation, ....) and related operations to the device.
As device has low memory, it does not keep itself the values
(except for view/spend keys) but once computed there are encrypted (with AES
are equivalent) and return back to monero-wallet-cli. When they need to be
manipulated by the device, they are decrypted on receive.
Moreover, using the client for storing the value in encrypted form limits
the modification in the client code. Those values are transfered from one
C-structure to another one as previously.
The code modification has been done with the wishes to be open to any
other hardware wallet. To achieve that a C++ class hw::Device has been
introduced. Two initial implementations are provided: the "default", which
remaps all calls to initial Monero code, and the "Ledger", which delegates
all calls to Ledger device.
Scheme by luigi1111:
Multisig for RingCT on Monero
2 of 2
User A (coordinator):
Spendkey b,B
Viewkey a,A (shared)
User B:
Spendkey c,C
Viewkey a,A (shared)
Public Address: C+B, A
Both have their own watch only wallet via C+B, a
A will coordinate spending process (though B could easily as well, coordinator is more needed for more participants)
A and B watch for incoming outputs
B creates "half" key images for discovered output D:
I2_D = (Hs(aR)+c) * Hp(D)
B also creates 1.5 random keypairs (one scalar and 2 pubkeys; one on base G and one on base Hp(D)) for each output, storing the scalar(k) (linked to D),
and sending the pubkeys with I2_D.
A also creates "half" key images:
I1_D = (Hs(aR)+b) * Hp(D)
Then I_D = I1_D + I2_D
Having I_D allows A to check spent status of course, but more importantly allows A to actually build a transaction prefix (and thus transaction).
A builds the transaction until most of the way through MLSAG_Gen, adding the 2 pubkeys (per input) provided with I2_D
to his own generated ones where they are needed (secret row L, R).
At this point, A has a mostly completed transaction (but with an invalid/incomplete signature). A sends over the tx and includes r,
which allows B (with the recipient's address) to verify the destination and amount (by reconstructing the stealth address and decoding ecdhInfo).
B then finishes the signature by computing ss[secret_index][0] = ss[secret_index][0] + k - cc[secret_index]*c (secret indices need to be passed as well).
B can then broadcast the tx, or send it back to A for broadcasting. Once B has completed the signing (and verified the tx to be valid), he can add the full I_D
to his cache, allowing him to verify spent status as well.
NOTE:
A and B *must* present key A and B to each other with a valid signature proving they know a and b respectively.
Otherwise, trickery like the following becomes possible:
A creates viewkey a,A, spendkey b,B, and sends a,A,B to B.
B creates a fake key C = zG - B. B sends C back to A.
The combined spendkey C+B then equals zG, allowing B to spend funds at any time!
The signature fixes this, because B does not know a c corresponding to C (and thus can't produce a signature).
2 of 3
User A (coordinator)
Shared viewkey a,A
"spendkey" j,J
User B
"spendkey" k,K
User C
"spendkey" m,M
A collects K and M from B and C
B collects J and M from A and C
C collects J and K from A and B
A computes N = nG, n = Hs(jK)
A computes O = oG, o = Hs(jM)
B anc C compute P = pG, p = Hs(kM) || Hs(mK)
B and C can also compute N and O respectively if they wish to be able to coordinate
Address: N+O+P, A
The rest follows as above. The coordinator possesses 2 of 3 needed keys; he can get the other
needed part of the signature/key images from either of the other two.
Alternatively, if secure communication exists between parties:
A gives j to B
B gives k to C
C gives m to A
Address: J+K+M, A
3 of 3
Identical to 2 of 2, except the coordinator must collect the key images from both of the others.
The transaction must also be passed an additional hop: A -> B -> C (or A -> C -> B), who can then broadcast it
or send it back to A.
N-1 of N
Generally the same as 2 of 3, except participants need to be arranged in a ring to pass their keys around
(using either the secure or insecure method).
For example (ignoring viewkey so letters line up):
[4 of 5]
User: spendkey
A: a
B: b
C: c
D: d
E: e
a -> B, b -> C, c -> D, d -> E, e -> A
Order of signing does not matter, it just must reach n-1 users. A "remaining keys" list must be passed around with
the transaction so the signers know if they should use 1 or both keys.
Collecting key image parts becomes a little messy, but basically every wallet sends over both of their parts with a tag for each.
Thia way the coordinating wallet can keep track of which images have been added and which wallet they come from. Reasoning:
1. The key images must be added only once (coordinator will get key images for key a from both A and B, he must add only one to get the proper key actual key image)
2. The coordinator must keep track of which helper pubkeys came from which wallet (discussed in 2 of 2 section). The coordinator
must choose only one set to use, then include his choice in the "remaining keys" list so the other wallets know which of their keys to use.
You can generalize it further to N-2 of N or even M of N, but I'm not sure there's legitimate demand to justify the complexity. It might
also be straightforward enough to support with minimal changes from N-1 format.
You basically just give each user additional keys for each additional "-1" you desire. N-2 would be 3 keys per user, N-3 4 keys, etc.
The process is somewhat cumbersome:
To create a N/N multisig wallet:
- each participant creates a normal wallet
- each participant runs "prepare_multisig", and sends the resulting string to every other participant
- each participant runs "make_multisig N A B C D...", with N being the threshold and A B C D... being the strings received from other participants (the threshold must currently equal N)
As txes are received, participants' wallets will need to synchronize so that those new outputs may be spent:
- each participant runs "export_multisig FILENAME", and sends the FILENAME file to every other participant
- each participant runs "import_multisig A B C D...", with A B C D... being the filenames received from other participants
Then, a transaction may be initiated:
- one of the participants runs "transfer ADDRESS AMOUNT"
- this partly signed transaction will be written to the "multisig_monero_tx" file
- the initiator sends this file to another participant
- that other participant runs "sign_multisig multisig_monero_tx"
- the resulting transaction is written to the "multisig_monero_tx" file again
- if the threshold was not reached, the file must be sent to another participant, until enough have signed
- the last participant to sign runs "submit_multisig multisig_monero_tx" to relay the transaction to the Monero network
This allows the key to be not the same for two outputs sent to
the same address (eg, if you pay yourself, and also get change
back). Also remove the key amounts lists and return parameters
since we don't actually generate random ones, so we don't need
to save them as we can recalculate them when needed if we have
the correct keys.
The whole rct data apart from the MLSAGs is now included in
the signed message, to avoid malleability issues.
Instead of passing the data that's not serialized as extra
parameters to the verification API, the transaction is modified
to fill all that information. This means the transaction can
not be const anymore, but it cleaner in other ways.
The mixRing (output keys and commitments) and II fields (key images)
can be reconstructed from vin data.
This saves some modest amount of space in the tx.
A new version of genRct takes the mixRing as parameter, instead
of the inPk. inPk are part of the mixRing, and it is cleaner to
pass the mixRing data than to fetch it from the RingCT code.
A new version of decodeRct also returns the mask.
Also, failure to decode throws, so errors are properly detected.
