* Don't delete/add values to the unique property map when it's null
This happens when revoking operator keys, as the revoked MN will then have
the address and operator keys set to the null representation. We shouldn't
add the null value to the unique property map as otherwise future revokes
crash in an assert.
* Assert that no null values are passed to Add/DeleteUniqueProperty
* Check for null values before calling Add/DeleteUniqueProperty
* Apply suggestions from code review
Co-Authored-By: codablock <ablock84@gmail.com>
* Add user generated default constructors to BLS primitives
Fixes build issues on Mac:
In file included from evo/deterministicmns.cpp:5:
./evo/deterministicmns.h:375:24: error: default initialization of an object of const type 'const CBLSPublicKey' without a user-provided default constructor
static const T nullValue;
^
All logging that happens in BuildNewListFromBlock is currently printed twice.
This is because BuildNewListFromBlock is called while processing the block
and also while calculating the DIP4 MN list commitment.
This commit adds the debugLogs to this method to allow omitting logs while
called from the DIP4 code.
* Allow to skip sig verification for CFinalCommitment::Verify
* Add CFinalCommitmentTxPayload and CheckLLMQCommitment and use it
As described in https://github.com/dashpay/dips/pull/31 (see discussion).
* Properly ban nodes for invalid commitments
* Add SPORK_17_QUORUM_DKG_ENABLED spork
* Implement CDummyDKG and CDummyCommitment until we have the real DKG merged
This is only used on testnet/devnet/regtest and will NEVER be used on
mainnet. It is NOT SECURE AT ALL!
See comment in quorums_dummydkg.h for more details.
* Test simple PoSe in DIP3 tests
* Generate 2 instead of 4 blocks per iteration in PoSe tests
4 was based on old chainparams where I used larger phases.
* Only sleep when necessary in PoSe tests
* Fix typo in comment
* Give PoSe tests more time and sync after fast-forward
Members which are not in the validMembers bitsets are now PoSe punished.
The maximum PoSe score is dynamic and mostly identical to the number of
registered MNs. Added PoSe scores for failures are then a percentage of
the maximum score, so that we can better control how often failures are
allowed per payment cycle. For LLMQ failures, this is 66% to allow
approximately 2 failures per payment cycle.
* Add LLMQ parameters to consensus params
* Add DIP6 quorum commitment special TX
* Implement CQuorumBlockProcessor which validates and handles commitments
* Add quorum commitments to new blocks
* Propagate QFCOMMITMENT messages to all nodes
* Allow special transactions in blocks which have no inputs/outputs
But only for TRANSACTION_QUORUM_COMMITMENT for now.
* Add quorum commitments to self-crafted blocks in DIP3 tests
* Add simple fork logic for current testnet
This should avoid a fork on the current testnet. It only applies to the
current chain which activated DIP3 at height 264000 and block
00000048e6e71d4bd90e7c456dcb94683ae832fcad13e1760d8283f7e89f332f.
When we revert the chain to retest the DIP3 deployment, this fork logic
can be removed again.
* Use quorumVvecHash instead of quorumHash to make null commitments unique
Implementation of https://github.com/dashpay/dips/pull/31
* Re-add quorum commitments after pruning mempool selected blocks
* Refactor CQuorumBlockProcessor::ProcessBlock to have less nested if/else statements
Also add BEGIN/END markers for temporary code.
* Add comments/documentation to LLMQParams
* Move code which determines if a commitment is required into IsCommitmentRequired
This should make the code easier to read and also removes some duplication.
The also changes the error types that are possible from 3 to 2 now. Instead
of having "bad-qc-already-mined" and "bad-qc-not-mining-phase", there is
only "bad-qc-not-allowed" now.
* Use new parameter from consensus parames for the temporary fork
* Properly initialize confirmedHash in CSimplifiedMNListEntry
* Add consensus params for temporary hard-fork on testnet
This hard-fork is required on testnet as we have to fix a few consensus
bugs in DIP3/DIP4. It will also be required to introduce LLMQ commitments
to testnet.
* Add temporary fork-logic for testnet, which sets confirmedHash to null
...before fork activation. Otherwise we'd invalidate all DIP3 blocks on
testnet.
* Bump nTemporaryTestnetForkHeight to 273000
* Track confirmedHash of masternodes
* Also include confirmedHash in DIP4
* Implement CalculateQuorum and CalculateScores in CDeterministicMNList
* Use new ProRegTx based scoring function in GetMasternodeScores
* Update test vectors for simplifiedmns_merkleroots tests
* Allow to reuse collaterals and addresses in ProRegTx
This also requires to move out the duplication check from CheckService to
the 2 individual ProTx check methods.
* Implement replacement logic for reused collaterals
* Add tests for new replacement logic
* Remove reference to now removed "protocolVersion"
* Allow to use non-DIP3 collaterals after DIP3 activation
This removes a rule which only made sense when we assumed all collaterals
would remain in DIP3 ProRegTx transactions after DIP3 activation.
As we allow to refer external collaterals now, this rule should vanish as
well.
* Sleep in case block time is ahead of real time
See comment. This was not a problem before as we had enough time passed
between the last block that caused this and the start-alias call.
Now we added another start-alias call which is directly after we generated
many blocks.
* Pass CCoinsView reference to special TX handling methods
* Allow referencing other TX outputs for ProRegTx collateral
* Remove "collateralAmount" from "protx register"
* Rename "protx register" to "protx fund_register"
* Remove UpdateSpork15Value from CDeterministicMNManager
Was not used/implemented anymore
* Lock masternode collaterals after chain/DIP3 is fully initialized
Otherwise detection of collaterals does not work.
* Implement new "protx register" RPC which uses existing collaterals
* Remove "masternode info" RPC
It is not consistent with other "masternode" RPCs anymore as it requires
the ProRegTx hash while all other RPCs work with the collateral.
* Load sporks from disk cache before initializing the chain
Otherwise spork15 is not loaded when we check it for the first time.
* Implement "protx info" RPC
* Use "protx info" instead of "masternode info" in DIP3 tests
* Test external collaterals for ProTx
* Handle review comments
* Don't pass CCoinView reference when it's not used
* Revert "Pass CCoinsView reference to special TX handling methods"
This reverts commit 28688724e1.
* Use GetUTXOCoin instead of now removed coinsView
Also remove collateral height check as GetUTXOCoin only returns confirmed
coins.
* Add conflict handling for external collaterals to mempool
* Handle review comments (squashed Github suggestions)
Co-Authored-By: codablock <ablock84@gmail.com>
* Remove nProtocolVersion fields from deterministic masternode lists
This field was part of my initial implementation from DIP3. One of the last
changes of DIP3 was then to remove this field, which I did not bring back
into code yet. This commit removes it now.
We use PROTOCOL_VERSION now in cases were compatibility in the the pre-DIP3
list is needed.
* Add type and mode fields in DIP3 special TXs
These were added to DIP3 but not in-code yet.
* Add DMN_PROTO_VERSION
* Use BLS keys for operator keys
* Add "bls generate" RPC to generate BLS keys
* Use unique_ptr to store blsKeyOperator and blsPubKeyOperator
Needed because the Chia BLS library crashes when keys are created before
the library is initialized, even if keys are not used. This is the case here
as we have static instances here.
* Remove unnecessary CheckSignature calls
This seems to be some garbage I left in by mistake.
* Fixed review comments
* Fix rpc help for operator keys
All keys that are used as examples are random. None of the secret keys
belongs to any of the public keys.
* Use .GetHash() instead of ::SerializeHash() for BLS pubkeys in txmempool.cpp
* Rename mapProTxBlsPubKeys to mapProTxBlsPubKeyHashes
These cause crashes when used in for loops. Reason is very likely that
temporary CDeterministicMNList objects are destroyed before entering the
loop while the adaptors rely on these objects.
Discovered in one of our devnets while calling "protx list"
* Don't use GetTransaction in CheckProUpRegTx
GetTransaction locks mempool.cs internally. This brings messes up the locking
order of cs_main, mempool.cs and CCheckQueue::ControlMutex.
* Lock cs_main in TestChainSetup::CreateBlock when updating CbTx
* Split keyIDMasternode into keyIDOwner/keyIDOperator/keyIDVoting
keyIDOwner is the key used for things which should stay in control of the
collateral owner, like proposal voting.
keyIDOperator is the key used for operational things, like signing network
messages, signing trigger/watchdog objects and trigger votes.
keyIDVoting is the key used for proposal voting
Legacy masternodes will always have the same key for all 3 to keep
compatibility.
Using different keys is only allowed after spork15 activation.
* Forbid reusing collateral keys for operator/owner keys and vice versa
* Bump SERIALIZATION_VERSION_STRING in CMasternodeMan
