Make claimpegin fail in the case of a double claim #615
Conversation
This bug got introduced with the 0.17 rebase. In 0.14.1 the double pegin claim check was done in the beginning of the validation logic, where the coins view was the actual UTXO set. In 0.17, Core moved several checks into a new method called CheckTxInputs in tx_verify.cpp, where we also put the pegin check. In the mempool acceptance check, however (as opposed to the ConnectBlock check), CheckTxInputs is called with only a partial UTXO view for optimization reasons. This commit adds an extra pegin check in the beginning of the mempool code where we have a full UTXO view.
Currently it will commit to the transaction without it being able to enter the mempool. This makes the block creation code include the tx into the block and create an invalid block.
stevenroose
force-pushed
the
fix-claimpegin
branch
from
2b15ad3
to
af82ff4
Compare
- changes the expected behavior of claimpegin to fail on double spend - adds extra test to make sure a block with a double spend is also not accepted
| @@ -226,8 +232,11 @@ def run_test(self): | |||
|
|
|||
| pegtxid1 = sidechain.claimpegin(raw, proof) | |||
| # Make sure it can't get accepted twice. | |||
| pegtxid2 = sidechain.claimpegin(raw, proof) | |||
| assert(pegtxid2 not in sidechain.getrawmempool()) | |||
| try: | |||
There was a problem hiding this comment.
use assert_raises_rpc_error instead
There was a problem hiding this comment.
also make it clearer that this is a mempool-mempool conflict, versus:
block-mempool
block-block
| sidechain.generate(7) | ||
| if sidechain.gettransaction(pegtxid2)["confirmations"] > 0: | ||
| raise Exception("Pegin should not be able to get confirmed") | ||
| try: |
There was a problem hiding this comment.
use assert_raises_rpc_error instead
There was a problem hiding this comment.
note that this is block-mempool conflict
| doublespendblock.hashMerkleRoot = doublespendblock.calc_merkle_root() | ||
| add_witness_commitment(doublespendblock) | ||
| doublespendblock.solve() | ||
| try: |
There was a problem hiding this comment.
use assert_raises_rpc_error instead
|
|
|
We should try covering all the scenarios: mempool-mempool |
|
Closing, merged into #614. |
Fixes #613. Builds on #614.