712d35bc56 wallet: Add missing cs_db lock (João Barbosa)
Pull request description:
Without this lock `BerkeleyEnvironment::~BerkeleyEnvironment` and `GetWalletEnv` would race for `g_dbenvs`. This wasn't detected before because thread safety analysis does not check constructors
and destructors.
Reference: http://releases.llvm.org/5.0.2/tools/clang/docs/ThreadSafetyAnalysis.html#no-checking-inside-constructors-and-destructors
Tree-SHA512: 350cb2b991ca699a6bca85f87c82c38f0814484c8ccb0d7d83cb3bff9afcf60dd32b2a9554a9e72eb5803bfad8b6970fe7da618b39be5889178b86faa1b74124
fae169c95e test: Make bloom tests deterministic (MarcoFalke)
Pull request description:
non-deterministic tests are useless, since a failing test could not be reproduced unless the seed is known.
Tree-SHA512: 4f634ff0c6adf663444f1ac504f6dbceaa46b78d697b840531977ba30006453ac559d5c21cc3eaef6d92b87d46008a34b0db6331ea3318001987fcfaec634acf
0dd6a8c124 Check m_internals in UnregisterValidationInterface (João Barbosa)
fd6d499bda gui: Fix m_node.startShutdown() order (João Barbosa)
07b9aadcfc gui: Expose BitcoinGUI::unsubscribeFromCoreSignals (João Barbosa)
60e190ceb3 gui: Fix WalletController deletion (João Barbosa)
Pull request description:
This PR consists in small fixes in order to have a clean shutdown from the GUI.
Tree-SHA512: a9c641f202bc810698c4a39d5c5a1f54e54bdab098c412d65418879e00764a9db9f38383813914d591e24e097e49f177942b2ae6c57bba05dcc095e8a1d0b8f4
This introduces support for autodetecting and using the RdSeed instruction.
In addition:
* In SeedFast, only 64 bits of entropy are generated through RdRand (256 was relatively slow).
* In SeedStartup, 256 bits of entropy are generated, using RdSeed (preferably) or RdRand (otherwise).
When a wallet is created it is registered in the validation interface (in
CWallet::CreateWalletFromFile) but it is not immediately added to the
wallets list. If a shutdown is requested before AddWallet (case more
evident when -rescan is set) then m_internals can be released (in
Shutdown -> UnregisterBackgroundSignalScheduler) before the wallet and
then ReleaseWallet would call UnregisterValidationInterface with
m_internals already released.
This change forwards the shutdown request on the GUI (close the
application for instace) to the node as soon as possible. This way the
GUI doesn't have to wait for long operations to complete (rescan the
wallet for instance), instead those operations detect the shutdown
request and abort/interrupt.
Move only change that makes unsubscribeFromCoreSignals public. It must be
called if the event loop is not running otherwise core signals handlers
can deadlock.
fac3a054cb test: Fix race in p2p_invalid_messages (MarcoFalke)
Pull request description:
After we change our magic bytes, the node may or may not send us messages such as feefilter or sendheaders, which would be rejected by `_on_data`.
Solve that by replacing `_on_data` with a noop.
Tree-SHA512: bd25a81f0c6e31f09155e00abab8062777d827b9210d6a9b85ef35cfe5069338f100fecf058842f41a1f134fdb3cf7ac1fe80db493e4dab7988acdacb33149df
11e0fd8d6 Descriptor expansions only need pubkey entries for PKH/WPKH (Pieter Wuille)
Pull request description:
Currently, calling `Expand` on a `Descriptor` object will populate the output FlatSigningProvider with all public keys involved in the descriptor. This is overkill, as pubkey entries are only needed when the lookup of a public key based on its hash is desired (which is the case for `pkh`, `wpkh`, and `combo` descriptors).
Fix this by pushing the population of pubkey entries down into the individual descriptor implementation's `MakeScript` function, instead of doing it generically.
This should make it easier to implement #14491 without importing P2PKH outputs for the individual public keys listed inside a multisig.
Tree-SHA512: 5bc7e9bd29f1b3bc63514803e9489b3bf126bfc177d46313aa9eeb98770ec61a97b55bd8ad4e2384154799f24b1bc4183bfdb4708b2ffa6e37ed2601a451cabc
facaae4cc4 travis: Save cache when compilation took very long (MarcoFalke)
faee6c9cac travis: Remove unused FUNCTIONAL_TESTS_CONFIG (MarcoFalke)
Pull request description:
Closes#14163
Tree-SHA512: fb3dd0b7a0a3ee7ac1c6d4e9314239cd03ab4a11f04b6345ef4a1f50ea2b79ea4248c41e99a6fb923a4b1b477758c4fa665d56a4ab703acf5a58dbc7d852ffc4
Without this lock BerkeleyEnvironment::~BerkeleyEnvironment and
GetWalletEnv would race for g_dbenvs. This wasn't detected before
because thread safety analysis does not check constructors and
destructors.
119d360aab travis: Document whether functional tests are run in the job name (Ben Woosley)
64f28545e3 Revert "travis: Compile trusty with depends for now" (Ben Woosley)
267eac00f9 Prefer boost::optional#get_value_or over #value_or (Ben Woosley)
1971f5ba04 Piecewise construct to avoid invalid construction (Ben Woosley)
Pull request description:
In light of #14979, I realized that only qt 5.5+ was being tested under CI, while compatibility lists 5.2+.
In #15276, Marco added Trusty to CI, building with depends. This changes that build to system libraries, in order to ensure ongoing compatibility with our claimed minimum required versions.
Fixes#14983, previously open as #14998
Tree-SHA512: 6cff5e28c756ecb8bf797c8f6eb77c1944ba61a8dd6d7d4984e63eef384f6429dc79c505da3241c05b9c4db31c72b2a9846c7365aba9280f2e0620e5f3998d07
fa5278a419 qa: Use wallet to retrieve raw transactions (MarcoFalke)
fa2198328e qa: Style-only fixes in touched files (MarcoFalke)
Pull request description:
Instead of asking the coin database and block storage about a transaction, pull it directly from the wallet in wallet related tests.
This refactoring only makes sense in light of #15159.
<sub>This product may contain minor stylistic cleanups
Tree-SHA512: ec34c7150d873da9f19fead3f7e3f758baba5ef10061942384c470a47a6f320690109be9c5160f0c8bc228272a729653d44c78471455337318f657d6c164ba23
e6c58d3b01 Do not import private keys to wallets with private keys disabled (Andrew Chow)
b5c5021b64 Refactor importwallet to extract data from the file and then import (Andrew Chow)
1f77f6754c tests: unify RPC argument to cli argument conversion and handle dicts and lists (Andrew Chow)
Pull request description:
Fixes a bug where private keys could be imported to wallets with private keys disabled. Now every RPC which can import private keys checks for whether the wallet has private keys are disabled and errors if it is. Also added an belt-and-suspenders check to `AddKeyPubkeyWithDB` to have it assert that the wallet has private keys enabled.
Tree-SHA512: 5cd04febce9aa2bd9bfd02f312c6ff8705e37278cae59efd3895f6d6e2f1b477aefd297e2dd0860791bdd3d4f3cad8eb1a404f8f3d4e2035b91314ad2c1028ae
In CMainSignals::RegisterWithMempoolSignals running under Ubuntu 14.04
(QT 5.2), absent piecewise construction this fails to create the pair
because the argument is a connection, which is converted into a
non-copyable scoped_connection.
validationinterface.cpp:80:186: required from here
/usr/include/boost/signals2/connection.hpp:234:7: error: ‘boost::signals2::scoped_connection::scoped_connection(const boost::signals2::scoped_connection&)’ is private
scoped_connection(const scoped_connection &other);
^
In file included from /usr/include/c++/4.8/utility:70:0,
from /usr/include/c++/4.8/algorithm:60,
from ./prevector.h:13,
from ./script/script.h:10,
from ./primitives/transaction.h:11,
from ./validationinterface.h:9,
from validationinterface.cpp:6:
/usr/include/c++/4.8/bits/stl_pair.h:134:45: error: within this context
: first(std::forward<_U1>(__x)), second(__y) { }
https://travis-ci.org/bitcoin/bitcoin/jobs/473689141#L2172
2e023419c5 tests: unify RPC argument to cli argument conversion and handle dicts and lists (Andrew Chow)
Pull request description:
When running tests with --usecli, unify the conversion from argument objects to strings using a new function arg_to_cli(). This fixes boolean arguments when using named arguments.
Also use json.dumps() to get the string values for arguments that are dicts and lists so that bitcoind's JSON parser does not become confused.
Tree-SHA512: 472bef3cd78410a8552fd342b1852bcd7c57721cfa9176b26bacda6b0791cc0b3758561a529c4117a7428242f98bb7d5482b2a2dcd06bea0ef2b15ae26183405
82dcacb822 msvc: build leveldb locally (Chun Kuan Lee)
52091066be msvc: build secp256k1 locally (Chun Kuan Lee)
Pull request description:
In current MSVC build setup, the code depends on leveldb and secp256k1 that are installed from vcpkg which is not controlled by us. If we update our code, we have to wait for vcpkg port being merged.
This PR move them from vcpkg to local branch to make it as same as autoconf.
The leveldb changes is based on bitcoin-core/leveldb#14 and bitcoin-core/leveldb#18
Tree-SHA512: aa2cc1c3191e8d9cab23d555da4be296314c46d944f452c2ec6202b1779e4cc223b603e589b38196cd2c793a03a8bb0ba128cc66256b35a58c5e7bb358475206