dd435ad Add unit tests for signals generated by ProcessNewBlock() (Jesse Cohen)
a3ae8e6 Fix concurrency-related bugs in ActivateBestChain (Jesse Cohen)
ecc3c4a Do not unlock cs_main in ABC unless we've actually made progress. (Matt Corallo)
Pull request description:
Originally this PR was just to add tests around concurrency in block validation - those tests seem to have uncovered another bug in ActivateBestChain - this now fixes that bug and adds tests.
ActivateBestChain (invoked after a new block is validated) proceeds in steps - acquiring and releasing cs_main while incrementally disconnecting and connecting blocks to sync to the most work chain known (FindMostWorkChain()). Every time cs_main is released the result of FindMostWorkChain() can change - but currently that value is cached across acquisitions of cs_main and only refreshed when an invalid chain is explored. It needs to be refreshed every time cs_main is reacquired. The test added in 6094ce7304 will occasionally fail without the commit fixing this issue 26bfdbaddb
Original description below
--
After a bug discovered where UpdatedBlockTip() notifications could be triggered out of order (#12978), these unit tests check certain invariants about these signals.
The scheduler test asserts that a SingleThreadedSchedulerClient processes callbacks fully and sequentially.
The block validation test generates a random chain and calls ProcessNewBlock from multiple threads at random and in parallel. ValidationInterface callbacks verify that the ordering of BlockConnected BlockDisconnected and UpdatedBlockTip events occur as expected.
Tree-SHA512: 4102423a03d2ea28580c7a70add8a6bdb22ef9e33b107c3aadef80d5af02644cdfaae516c44933924717599c81701e0b96fbf9cf38696e9e41372401a5ee1f3c
After a recent bug discovered in callback ordering in MainSignals,
this test checks invariants in ordering of
BlockConnected / BlockDisconnected / UpdatedChainTip signals
60f61f9 Tighten up bech32::Decode(); add tests. (murrayn)
Pull request description:
Just a few minor optimizations to bech32::Decode():
1) optimize the order and logic of the conditionals
2) get rid of subsequent '(c < 33 || c > 126)' check which is redundant (already performed above)
3) add a couple more bech32 tests (mixed-case)
Tree-SHA512: e41af834c8f6b7d34c22c28b724df42c60f72e00df616e70a12efbc4271d15d80627fe1bc36845caf29f615c238499a566298a863cbe119fef457287231053c8
chain.h does not actually depend on the methods defined in pow.h, just its
include of consensus/params.h, which is standalone and can be included instead.
Confirmed by inspection and successful build.
159c32d1f1 Add assertion to guide static analyzers. Clang Static Analyzer needs this guidance. (practicalswift)
fd447a6efe Fix dead stores. Values were stored but never read. Limit scope. (practicalswift)
Pull request description:
Fix Clang Static Analyzer warnings reported by @kallewoof in #12961:
* Fix dead stores. Values were stored but never read.
* Add assertion to guide static analyzers. See #12961 for details.
Tree-SHA512: 83dbec821f45217637316bee978e7543f2d2caeb7f7b0b3aec107fede0fff8baa756da8f6b761ae0d38537740839ac9752f6689109c38a4b05c0c041aaa3a1fb
364bae5 qa: Pad scriptPubKeys to get minimum sized txs (MarcoFalke)
7485488 Policy to reject extremely small transactions (Johnson Lau)
0f8719b Add transaction tests for constant scriptCode (Johnson Lau)
9dabfe4 Add constant scriptCode policy in non-segwit scripts (Johnson Lau)
Pull request description:
This disables `OP_CODESEPARATOR` in non-segwit scripts (even in an unexecuted branch), and makes a positive `FindAndDelete` result invalid. This ensures that the `scriptCode` serialized in `SignatureHash` is always the same as the script passing to the `EvalScript`.
Tree-SHA512: a0552cb920294d130251c48053fa2ff1fbdd26332e62b52147d918837852750f0ce35ce2cd1cbdb86588943312f8154ccb4925e850dbb7c2254bc353070cd5f8
This disables OP_CODESEPARATOR in non-segwit scripts (even in an unexecuted branch), and makes a positive FindAndDelete result invalid. This ensures that the scriptCode serialized in SignatureHash() is always the same as the script passing to the EvalScript.
7d0f80b Use anonymous namespace instead of static functions (Pieter Wuille)
b61fb71 Mention removal of bare multisig IsMine in release notes (Pieter Wuille)
9c2a8b8 Do not treat bare multisig as IsMine (Pieter Wuille)
08f3228 Optimization: only test for witness scripts at top level (Pieter Wuille)
3619735 Track difference between scriptPubKey and P2SH execution in IsMine (Pieter Wuille)
ac6ec62 Switch to a private version of SigVersion inside IsMine (Pieter Wuille)
19fc973 Do not expose SigVersion argument to IsMine (Pieter Wuille)
fb1dfbb Remove unused IsMine overload (Pieter Wuille)
952d821 Make CScript -> CScriptID conversion explicit (Pieter Wuille)
Pull request description:
Currently our wallet code will treat bare multisig outputs (meaning scriptPubKeys with multiple public keys + `OP_CHECKMULTISIG` operator in it) as ours without the user asking for it, as long as all private keys in it are in our wallet.
This is a pointless feature. As it only works when all private keys are in one place, it's useless compared to single key outputs (P2PK, P2PKH, P2WPKH, P2SH-P2WPKH), and worse in terms of space, cost, UTXO size, and ability to test (due to lack of address format for them).
Furthermore, they are problematic in that producing a list of all `scriptPubKeys` we accept is not tractable (it involves all combinations of all public keys that are ours). In further wallet changes I'd like to move to a model where all scriptPubKeys that are treated as ours are explicit, rather than defined by whatever keys we have. The current behavior of the wallet is very hard to model in such a design, so I'd like to get rid of it.
I think there are two options:
* Remove it entirely (do not ever accept bare multisig outputs as ours, unless watched)
* Only accept bare multisig outputs in situations where the P2SH version of that output would also be acceptable
This PR implements the first option. The second option was explored in #12874.
Tree-SHA512: 917ed45b3cac864cee53e27f9a3e900390c576277fbd6751b1250becea04d692b3b426fa09065a3399931013bd579c4f3dbeeb29d51d19ed0c64da75d430ad9a
54a5a21 [MOVEONLY] Turn CScript::GetOp2 into a function and move to cpp (Pieter Wuille)
6a7456a [MOVEONLY] Move CSCript::FindAndDelete to interpreter (Pieter Wuille)
33a8ecf Delete unused non-const-iterator CSCript::GetOp overloads (Pieter Wuille)
2fb168b Make iterators in CScript::FindAndDelete const (Pieter Wuille)
Pull request description:
This PR moves `FindAndDelete` and `GetOp2` out of CScript (the first is only used inside the interpreter and moved there, the second does not actually depend on any script specifics and works on any vector). Furthermore, all non-const-iterator versions of GetOp are replaced by const ones, removing a number of methods in the process.
The longer term goal here is making the script interpreter independent from the CScript representation.
Note for reviewers: both `FindAndDelete` and `GetScriptOp` are consensus critical.
Tree-SHA512: c4ccf91c0b33c37cff0d474aa8dd2dab25b5b7655e2ed69a9b15e29daf0a67b21d51c23e1defb3a72ec762bd6138de96f69c6db1fb9c1fe1e976e421261aedb7
Such outputs can still be watched, and signed for, but they aren't treated as valid payments.
That means they won't cause transactions to appear in listtransactions, their outputs to be
shown under listunspent, or affect balances.
Seems providing at least minimal visibility to the failure is a good practice.
The only remaining ignored state is in LoadExternalBlockFile, where logging
would likely be spammy.
fae58eca93 tests: Avoid copies of CTransaction (MarcoFalke)
Pull request description:
Avoid the copy (or move) constructor of `CTransaction` in test code, whereever a simple reference can be used instead.
Tree-SHA512: 8ef2077a277d6182996f4671722fdc01a90909ae7431c1e52604aab8ed028910615028caf9b4cb07a9b15fdc04939dea2209cc3189dde7d38271256d9fe1076c
c25321f Add config changes to release notes (Anthony Towns)
5e3cbe0 [tests] Unit tests for -testnet/-regtest in [test]/[regtest] sections (Anthony Towns)
005ad26 ArgsManager: special handling for -regtest and -testnet (Anthony Towns)
608415d [tests] Unit tests for network-specific config entries (Anthony Towns)
68797e2 ArgsManager: Warn when ignoring network-specific config setting (Anthony Towns)
d1fc4d9 ArgsManager: limit some options to only apply on mainnet when in default section (Anthony Towns)
8a9817d [tests] Use regtest section in functional tests configs (Anthony Towns)
30f9407 [tests] Unit tests for config file sections (Anthony Towns)
95eb66d ArgsManager: support config file sections (Anthony Towns)
4d34fcc ArgsManager: drop m_negated_args (Anthony Towns)
3673ca3 ArgsManager: keep command line and config file arguments separate (Anthony Towns)
Pull request description:
The weekly meeting on [2017-12-07](http://www.erisian.com.au/meetbot/bitcoin-core-dev/2017/bitcoin-core-dev.2017-12-07-19.00.log.html) discussed allowing options to bitcoin to have some sensitivity to what network is in use. @theuni suggested having sections in the config file:
<cfields> an alternative to that would be sections in a config file. and on the
cmdline they'd look like namespaces. so, [testnet] port=5. or -testnet::port=5.
This approach is (more or less) supported by `boost::program_options::detail::config_file_iterator` -- when it sees a `[testnet]` section with `port=5`, it will treat that the same as "testnet.port=5". So `[testnet] port=5` (or `testnet.port=5` without the section header) in bitcoin.conf and `-testnet.port=5` on the command line.
The other aspect to this question is possibly limiting some options so that there is no possibility of accidental cross-contamination across networks. For example, if you're using a particular wallet.dat on mainnet, you may not want to accidentally use the same wallet on testnet and risk reusing keys.
I've set this up so that the `-addnode` and `-wallet` options are `NETWORK_ONLY`, so that if you have a bitcoin.conf:
wallet=/secret/wallet.dat
upnp=1
and you run `bitcoind -testnet` or `bitcoind -regtest`, then the `wallet=` setting will be ignored, and should behave as if your bitcoin.conf had specified:
upnp=1
[main]
wallet=/secret/wallet.dat
For any `NETWORK_ONLY` options, if you're using `-testnet` or `-regtest`, you'll have to add the prefix to any command line options. This was necessary for `multiwallet.py` for instance.
I've left the "default" options as taking precedence over network specific ones, which might be backwards. So if you have:
maxmempool=200
[regtest]
maxmempool=100
your maxmempool will still be 200 on regtest. The advantage of doing it this way is that if you have `[regtest] maxmempool=100` in bitcoin.conf, and then say `bitcoind -regtest -maxmempool=200`, the same result is probably in line with what you expect...
The other thing to note is that I'm using the chain names from `chainparamsbase.cpp` / `ChainNameFromCommandLine`, so the sections are `[main]`, `[test]` and `[regtest]`; not `[mainnet]` or `[testnet]` as might be expected.
Thoughts? Ping @MeshCollider @laanwj @jonasschnelli @morcos
Tree-SHA512: f00b5eb75f006189987e5c15e154a42b66ee251777768c1e185d764279070fcb7c41947d8794092b912a03d985843c82e5189871416995436a6260520fb7a4db
be67831 Make DummySignatureCreator a singleton (Pieter Wuille)
190b8d2 Make BaseSignatureCreator a pure interface (Pieter Wuille)
Pull request description:
* Removes the `m_provider` field from `BaseSignatureCreator`. Instead both a `SigningProvider` (which provides keys and scripts) and a `BaseSignatureCreator` (which implements the transaction-specific (or other) signing logic) are passed into and down in `ProduceSignature`, making the two concepts orthogonal.
* Makes `BaseSignatureCreator` a pure interface without constructor, making it easier to implement new derivations of it (for example for message signing).
* As `DummySignatureCreator` now becomes a stateless object, turn it into a singleton `DUMMY_SIGNATURE_CREATOR`.
Tree-SHA512: 5f1f4512e4ea7d02a31df7b9ede55008efa716c5b74a2630ca1c2fc6599584d8bf5f5641487266127f4b3788033803539fbd22b03ef1219c83c10da2d3da3dcd
When a -nofoo option is seen, instead of adding it to a separate
set of negated args, set the arg as being an empty vector of strings.
This changes the behaviour in some ways:
- -nofoo=0 still sets foo=1 but no longer treats it as a negated arg
- -nofoo=1 -foo=2 has GetArgs() return [2] rather than [2,0]
- "foo=2 \n -nofoo=1" in a config file no longer returns [2,0], just [0]
- GetArgs returns an empty vector for negated args
c55aa4f test: Fix sign for expected values (Karl-Johan Alm)
Pull request description:
A number of `BOOST_CHECK_EQUAL` calls would result in warnings about signs.
This PR fixes signedness for all expectation values, sometimes resulting in `int` → `unsigned int`. No other code changes besides adding/removing `U` to/from values.
Running `make &> make_output_...` on master versus on this PR:
```
$ wc make_output_*
1464 5925 90357 make_output_master
613 1469 28370 make_output_signfixed
```
More than halves the output lines from compiling.
Tree-SHA512: b06c9fb81704fd32a6a61fe7b2ceb5f1bb381e9873d79e13d7e4d26bbd9b67c9725a84e6fb2903bcda775aea2a792e544b0799d36735c19f5d1c7225e8c6d14e
7ef9cd8 Increase entropy in test temp directory name (Pieter Wuille)
f6dfb0f Reorder travis builds (Pieter Wuille)
156db42 tests: run tests in parallel (Cory Fields)
66f3255 tests: split up actual tests and helper files (Cory Fields)
Pull request description:
This runs the unit tests (`src/test/test_bitcoin`) in 4 separate simultaneous processes, significantly speeding up some Travis runs (over 2x for win32).
This uses an approach by @theuni that relies on `make` as the mechanism for distributing tests over processes (through `-j`). For every test .cpp file, we search for `BOOST_FIXTURE_TEST_SUITE` or `BOOST_AUTO_TEST_SUITE`, and then invoke the test binary for just that suite (using `-t`). The (verbose) output is stored in a temporary file, and only shown in the case of failure.
Some makefile reshuffling is necessary to avoid trying to run tests from `src/test/test_bitcoin.cpp` for example, which contains framework/utility code but no real tests.
Finally, order the Travis jobs from slow to fast (apart from the arm/doc job which goes first, for fast failure). This should help reducing the total wall clock time before opening a PR and finishing Travis, in case where not all jobs are started simultaneously.
This is an alternative to #12831.
Tree-SHA512: 9f82eb4ade14ac859618da533c7d9df2aa9f5592a076dcc4939beeffd109eda33f7d5480d8f50c0d8b23bf3099759e9f3a2d4c78efb5b66b04569b39b354c185
b120f7b [test] Add tests for self usage in arith_uint256 (Karl-Johan Alm)
08b17de [arith_uint256] Do not destroy *this content if passed-in operator may reference it (Karl-Johan Alm)
Pull request description:
Before this fix (see test commit), `v *= v` would result in `0` because `operator*=` set `*this` (`==b`) to `0` at the start. This patch changes the code to use `a` as temporary for `*this`~~, with drawback that `*this` is set to `a` at the end, an extra `=` operation in other words~~.
Tree-SHA512: 8028a99880c3198a39c4bcc5056169735ba960625d553e15c0317510a52940c875f7a1fefe14e1af7fcf10c07a246411994a328cb1507bf3eaf1b6e7425390dc
