0bba55f6af spork: replace `LOCKS_EXCLUDED` with negative `EXCLUSIVE_LOCKS_REQUIRED` (Kittywhiskers Van Gogh)
d657951f90 masternode: replace `LOCKS_EXCLUDED` with negative `EXCLUSIVE_LOCKS_REQUIRED` (Kittywhiskers Van Gogh)
8b1d3b55ab llmq: replace `LOCKS_EXCLUDED` with negative `EXCLUSIVE_LOCKS_REQUIRED` (Kittywhiskers Van Gogh)
cceff152ef coinjoin: replace `LOCKS_EXCLUDED` with negative `EXCLUSIVE_LOCKS_REQUIRED` (Kittywhiskers Van Gogh)
Pull request description:
## Additional Information
With the exception some usages of `cs_main` and a few (`Recursive`)`Mutex`es, Bitcoin has replaced their usage of `LOCKS_EXCLUDED(cs)` with `EXCLUSIVE_LOCKS_REQUIRED(!cs)` due to the stricter enforcement that negative locking brings with Clang (and it having a trickle-up effect caused by needing lock annotations on calling functions as well).
Dash intensively uses `LOCKS_EXCLUDED` for Dash-specific logic and moving it over also required updating (or adding) lock annotations for calling functions.
This pull request is being opened due to an upcoming pull request that includes https://github.com/bitcoin/bitcoin/pull/25109, which requires all `AssertLockNotHeld` usage to accompany a negative lock annotation.
## Breaking Changes
None expected. (Negative) lock enforcement has been made stricter but no new locks should be introduced by changes.
## Checklist:
- [x] I have performed a self-review of my own code
- [x] I have commented my code, particularly in hard-to-understand areas **(note: N/A)**
- [x] I have added or updated relevant unit/integration/functional/e2e tests **(note: N/A)**
- [x] I have made corresponding changes to the documentation **(note: N/A)**
- [x] I have assigned this pull request to a milestone _(for repository code-owners and collaborators only)_
ACKs for top commit:
PastaPastaPasta:
utACK 0bba55f6af
Tree-SHA512: 54c6b80f6ad8dc821cbe9250f8e8f111bc49fb7b38ef0de10423110c066c3e7ea4b93dc61580f319837ac4690ed21912f3ec43aaa1d3cd954043798a675226ee
10869fff3b fix: order of locks cs_wallet and cs_main in rpc/evo (Konstantin Akimov)
Pull request description:
## Issue being fixed or feature implemented
Changing order of locking `cs_main` and `cs_wallet` to prevent a deadlock.
There's call `protx_list` -> `BuildDMNListEntry` -> `CheckWalletOwnsScript`:
```
return WITH_LOCK(pwallet->cs_wallet, return pwallet->IsMine(script)) == isminetype::ISMINE_SPENDABLE;
```
It can cause a deadlock due to wrong order of locks (cs_wallet supposed to be blocked in prior of cs_main)
## What was done?
This PR adds and extra lock of cs_wallet and reduce scope of cs_main for a bit
## How Has This Been Tested?
Deadlock warning is reproduced with this PR: https://github.com/dashpay/dash/pull/6003
## Breaking Changes
N/A
## Checklist:
- [x] I have performed a self-review of my own code
- [ ] I have commented my code, particularly in hard-to-understand areas
- [ ] I have added or updated relevant unit/integration/functional/e2e tests
- [ ] I have made corresponding changes to the documentation
- [x] I have assigned this pull request to a milestone
Top commit has no ACKs.
Tree-SHA512: 6d166fe7c47067c1c6d889d87e015ff3bc43aa9f66234341840cc8465ce00a79e7140bc09cbdb2fd08feaae5463b320e0b66bbe410422783f86cbc9d616af6b3
ab7ac1b85b partial bitcoin#25176: Fix frequent -netinfo JSON errors from null getpeerinfo#relaytxes (Kittywhiskers Van Gogh)
c89799d46c merge bitcoin#24692: Follow-ups to #21160 (Kittywhiskers Van Gogh)
33098aefff merge bitcoin#21160: Move tx inventory into net_processing (Kittywhiskers Van Gogh)
24205d94fe partial bitcoin#20196: fix GetListenPort() to derive the proper port (Kittywhiskers Van Gogh)
7f7200986b merge bitcoin-core/gui#526: Add address relay/processed/rate-limited fields to peer details (Kittywhiskers Van Gogh)
a9114f1ce8 merge bitcoin#23695: Always serialize local timestamp for version msg (Kittywhiskers Van Gogh)
d936c28b4e merge bitcoin#23575: Rework FillNode (Kittywhiskers Van Gogh)
6f8c730f35 merge bitcoin#19499: Make timeout mockable and type safe, speed up test (Kittywhiskers Van Gogh)
43a82bdd29 merge bitcoin#20769: fixes "advertised address where nobody is listening" (Kittywhiskers Van Gogh)
Pull request description:
## Additional Information
* Dependent on https://github.com/dashpay/dash/pull/5978
* Dependent on https://github.com/dashpay/dash/pull/5977
## Breaking Changes
None observed.
## Checklist:
- [x] I have performed a self-review of my own code
- [x] I have commented my code, particularly in hard-to-understand areas **(note: N/A)**
- [x] I have added or updated relevant unit/integration/functional/e2e tests
- [x] I have made corresponding changes to the documentation **(note: N/A)**
- [x] I have assigned this pull request to a milestone _(for repository code-owners and collaborators only)_
ACKs for top commit:
PastaPastaPasta:
utACK ab7ac1b85b
Tree-SHA512: 87bf5108bb80576c5bff8cd577add7800044da252fd18590e06a727f0bf70de94e2e9294b4412cdd9f1f6676472b0359902af361aaffc4c9ee299ad07d6af009
c8742057ac fix(qt): tab switching via shortcuts doesn't work after 5986 (UdjinM6)
Pull request description:
## Issue being fixed or feature implemented
Looks like 3265b54a2f (#5986) broke it
## What was done?
## How Has This Been Tested?
run dash-qt, try using shortcuts `cmd+1`, `cmd+2` etc. (on macos)
## Breaking Changes
n/a
## Checklist:
- [x] I have performed a self-review of my own code
- [ ] I have commented my code, particularly in hard-to-understand areas
- [ ] I have added or updated relevant unit/integration/functional/e2e tests
- [ ] I have made corresponding changes to the documentation
- [x] I have assigned this pull request to a milestone _(for repository code-owners and collaborators only)_
ACKs for top commit:
PastaPastaPasta:
ACK c8742057ac
kwvg:
ACK c8742057ac
Tree-SHA512: 62a593ec804d75ff8aada0ef9ea90106adbf8cd11b202a6296086f55c2a4d2181e56dc8e56193a0ed49d94e55ee3236ab441ab477c8ca6d7b0c649dff987dbbc
b4477e409c trivial: don't print `fDisableGovernance` value anymore (Kittywhiskers Van Gogh)
a42370df93 refactor: remove fDisableGovernance global, define default in variable (Kittywhiskers Van Gogh)
b1527599e4 refactor: remove fMasternodeMode and fDisableGovernance from Qt code (Kittywhiskers Van Gogh)
9402ce7171 refactor: limit usage of fDisableGovernance, use `IsValid()` instead (Kittywhiskers Van Gogh)
106f6bdd4e refactor: reduce fMasternodeMode usage in governance and mnauth (Kittywhiskers Van Gogh)
3ba293fbcc refactor: remove fMasternodeMode checks in CActiveMasternodeManager (Kittywhiskers Van Gogh)
b0216ac8a6 refactor: remove fMasternodeMode usage in rpc logic (Kittywhiskers Van Gogh)
4d629a04fb refactor: limit fMasternodeMode usage in blockstorage, init, net_processing (Kittywhiskers Van Gogh)
a9cbdfcebc refactor: remove fMasternodeMode usage from llmq logic (Kittywhiskers Van Gogh)
c62a3d5778 refactor: remove fMasternodeMode usage from coinjoin logic (Kittywhiskers Van Gogh)
Pull request description:
## Motivation
Since https://github.com/dashpay/dash/pull/5940, `CActiveMasternodeManager` ceased to be a global variable and became a conditional smart pointer initialized based on the value of `fMasternodeMode`.
Likewise, since https://github.com/dashpay/dash/pull/5555, we can tell if any `CFlatDB`-based manager has successfully loaded its database. `CGovernanceManager` is one of them and conditionally loads its database based on the value of `fGovernanceDisabled`.
`fMasternodeMode` and `fGovernanceDisabled` were (and the former to a certain degree still is) unavoidable globals due to the way the functionality they influenced was structured (i.e. decided in initialization code with no way to query from the manager itself). As we can directly ask the managers now, we can start reducing the usage of these globals and at least in this PR, get rid of one of them.
This PR was the idea of PastaPastaPasta, special thanks for the suggestion!
## Additional Information
* There are two conventions being used for checking `nullptr`-ity of a pointer, `if (mn_activeman)` and `if (mn_activeman != nullptr)`. The former is used in initialization and RPC code due to existing conventions there ([source](2dacfb08bd/src/init.cpp (L1659-L1677)), [source](2dacfb08bd/src/rpc/net.cpp (L942-L945)), [source](2dacfb08bd/src/rpc/misc.cpp (L215-L218))). The latter is used whenever the value has to be passed as a `bool` (you cannot pass the implicit conversion to a `bool` argument without explicitly casting it) and in Dash-specific code where it is the prevalent convention ([source](2dacfb08bd/src/governance/governance.cpp (L125)), [source](2dacfb08bd/src/coinjoin/client.cpp (L1064))).
Unfortunately, that means this PR expresses the same thing sometimes in two different ways but this approach was taken so that reading is consistent within the same file. Codebase-wide harmonization is outside the scope of this PR.
* Where `mn_activeman` isn't directly available, the result of the check is passed as an argument named `is_masternode` and/or set for the manager during its construction as `m_is_masternode` (`const bool`) as it is expected for the `CActiveMasternodeManager`'s presence or absence to remain as-is for the duration of the manager's lifetime.
This does mean that some parts of the codebase check for `mn_activeman` while others check for {`m_`}`is_masternode`, which does reduce clarity while reading. Suggestions on improving this are welcomed.
* One of the reasons this PR was made was to avoid having to deal the _possibility_ of `fMasternodeMode` or `fDisableGovernance` from desynchronizing from the behaviour of the managers it's suppose to influence. It's why additional assertions were placed in to make sure that `fMasternodeMode` and the existence of `mn_activeman` were always in sync ([source](2dacfb08bd/src/evo/mnauth.cpp (L137-L139)), [source](2dacfb08bd/src/rpc/governance.cpp (L319-L320))).
But removing the tracking global and relying on a manager's state itself prevents a potential desync, which is what this PR is aiming to do.
## Breaking Changes
None expected.
## Checklist:
- [x] I have performed a self-review of my own code
- [x] I have commented my code, particularly in hard-to-understand areas
- [x] I have added or updated relevant unit/integration/functional/e2e tests
- [x] I have made corresponding changes to the documentation **(note: N/A)**
- [x] I have assigned this pull request to a milestone _(for repository code-owners and collaborators only)_
Top commit has no ACKs.
Tree-SHA512: 7861afd17c83b92af4c95b2841e9b0f676116eb3f234c4d0b1dcd3c20395452893e8ca3a17c7225389c8411dac80aeb5050f06a2ae35df5ec48998a571ef120c
e5129e6c41 fix: if hdseed is wrong - do not setup random seed, user can lost his fund (Konstantin Akimov)
e52498b7d5 chore: add todoes to UpgradeToHD function (Konstantin Akimov)
0d12ea9fa6 fix: wrong if/else branches for key refill in wallet creation (Konstantin Akimov)
d2c3dcb839 refactor: move list of function in rpcwallet to the end (Konstantin Akimov)
Pull request description:
## Issue being fixed or feature implemented
Reviewed our implementation of HD wallets and compare it to bitcoin's for aiming to backport or re-implement `sethdseed` rpc.
Noticed some strange things in our implementation, which this PR is aim to fix.
## What was done?
See each commit for detailed changes.
## How Has This Been Tested?
Run unit/functional tests
## Breaking Changes
`-hdseed` doesn't assign a random seed anymore if an user provided an invalid hex string.
## Checklist:
- [x] I have performed a self-review of my own code
- [x] I have commented my code, particularly in hard-to-understand areas
- [ ] I have added or updated relevant unit/integration/functional/e2e tests
- [ ] I have made corresponding changes to the documentation
- [x] I have assigned this pull request to a milestone
ACKs for top commit:
PastaPastaPasta:
utACK e5129e6c41
Tree-SHA512: b404dbc0762777abf421a847f58cd243e0aa00151ba3f036835c9ff54c1109b6159921ec24e29455e975797f49d54832c55f7876188da90f37dd1e4a811a21e0
CGovernanceManager::IsValid() returns true only if its db is successfully
initialized. If we attempt to initialize it and fail, init logic will
report to us that it failed. If we don't attempt to initialize it at all,
it will remain false.
Since fDisableGovernance is the same as not initializing it at all and
the other case where IsValid() is false is dealt with in init, we can
use IsValid() to infer if governance is enabled.
CActiveMasternodeManager no longer exists as a global variable, it is a
conditionally initialized smart pointer. If it exists, then it's in masternode
mode, no need to check if we're in masternode mode anymore.
7cc77f3a30 Merge #21373: test: generate fewer blocks in feature_nulldummy to fix timeouts, speed up (MarcoFalke)
a933a60b1a feat: new command line argument -bip147height for bitcoin#21373 (Konstantin Akimov)
51911388f2 Merge #21377: Speedy trial support for versionbits (fanquake)
ecade9bc39 Merge bitcoin/bitcoin#21749: test: Bump shellcheck version (W. J. van der Laan)
eeec2f2799 Merge bitcoin/bitcoin#21884: fuzz: Remove unused --enable-danger-fuzz-link-all option (fanquake)
51633d70ea Merge bitcoin/bitcoin#21874: fuzz: Add WRITE_ALL_FUZZ_TARGETS_AND_ABORT (MarcoFalke)
a02a2c0322 Merge bitcoin/bitcoin#21681: validation: fix ActivateSnapshot to use hardcoded nChainTx (MarcoFalke)
71f23d6e33 Merge bitcoin/bitcoin#21814: test: Fix feature_config_args.py intermittent issue (MarcoFalke)
de4d2a839d Merge bitcoin/bitcoin#21714: refactor: Drop CCoinControl::SetNull (MarcoFalke)
a63f9c31cc Merge bitcoin-core/gui#284: refactor: Simplify SendCoinsDialog::updateCoinControlState (Hennadii Stepanov)
b2d889380c Merge bitcoin/bitcoin#21792: test: Fix intermittent issue in p2p_segwit.py (MarcoFalke)
Pull request description:
## Issue being fixed or feature implemented
Regular batch of backports from bitcoin v22
## What was done?
Implemented new commandline argument `-bip147height` for RegTest.
Backports:
- bitcoin/bitcoin#21377
- bitcoin/bitcoin#21792
- bitcoin-core/gui#284
- bitcoin/bitcoin#21714
- bitcoin/bitcoin#21373
- bitcoin/bitcoin#21814
- bitcoin/bitcoin#21681
- bitcoin/bitcoin#21874
- bitcoin/bitcoin#21884
- bitcoin/bitcoin#21749
## How Has This Been Tested?
Run unit/functional tests
## Breaking Changes
n/a
## Checklist:
- [x] I have performed a self-review of my own code
- [ ] I have commented my code, particularly in hard-to-understand areas
- [ ] I have added or updated relevant unit/integration/functional/e2e tests
- [ ] I have made corresponding changes to the documentation
- [x] I have assigned this pull request to a milestone
ACKs for top commit:
PastaPastaPasta:
utACK 7cc77f3a30
Tree-SHA512: d46218667430af19c445d67d6411b1e7f19920e85f86e6e74ae7b9062aeb3763637a8613587c203ba8d285ccc3ee755f936141010944cfae8627397e8b8584d3
f2fe39fadc trivial: add `Asserts` for `m_peerman` pointer container uses (Kittywhiskers Van Gogh)
35be4e2ebe llmq: pass PeerManager to llmq::CInstantSendManager constructor (Kittywhiskers Van Gogh)
bfd33cd2b4 net: move CConnman::RelayInv{Filtered} into PeerManager (Kittywhiskers Van Gogh)
313a7e9a50 trivial: cleanup unnecessary headers in context files (Kittywhiskers Van Gogh)
c3f1ac2291 net: retire CConnman::RelayTransaction, use PeerManager::RelayTransaction (Kittywhiskers Van Gogh)
0323c6ca17 net: move Relay{Inv, InvFiltered, Transaction} out of CConnman (Kittywhiskers Van Gogh)
d54ba447a7 net: use ForEachNode instead of manually iterating through vNodes (Kittywhiskers Van Gogh)
Pull request description:
## Additional information
* Dependency for https://github.com/dashpay/dash/pull/5982
* Unfortunately, we are stuck with using the `unique_ptr` const-ref of `PeerManager` for initializing {`CJ`, `LLMQ`}`Context` due to some unit tests depending on Dash-specific entities but based on a testing setup that doesn't initialize `PeerManager` (specifically `validation_chainstatemanager_tests`, which uses `ChainTestingSetup`, as opposed to `TestingSetup`, which initializes `PeerManager`).
Attempting to dereference it earlier will result in crashing (see CI pipeline [here](https://gitlab.com/dashpay/dash/-/pipelines/1245924304))
## Breaking Changes
None observed.
## Checklist:
- [x] I have performed a self-review of my own code
- [x] I have commented my code, particularly in hard-to-understand areas
- [x] I have added or updated relevant unit/integration/functional/e2e tests
- [x] I have made corresponding changes to the documentation **(note: N/A)**
- [x] I have assigned this pull request to a milestone _(for repository code-owners and collaborators only)_
ACKs for top commit:
PastaPastaPasta:
re-utACK [f2fe39f](f2fe39fadc)
knst:
utACK [f2fe39f](f2fe39fadc)
Tree-SHA512: 5e9ca47ba9f7f0c5fdc93bf6cfd99d91be6d134610d2bd3dccde17443a956c7d8848b3406082ee8b6ee3fd25f586ce7bac45bc89b568424479b8fc149172ae51
f5f8bd9e9a doc: replace gfd() function to recommendation to use `git diff-range` (Konstantin Akimov)
Pull request description:
## Issue being fixed or feature implemented
`gfd` is less convenient to use, `git range-diff` is more functional. Dropped recommendation to use it.
## What was done?
Updated `CONTRIBUTION.md` file
## How Has This Been Tested?
Run `git range-diff` to ensure that it indeed works
## Breaking Changes
N/A
## Checklist:
- [x] I have performed a self-review of my own code
- [ ] I have commented my code, particularly in hard-to-understand areas
- [ ] I have added or updated relevant unit/integration/functional/e2e tests
- [x] I have made corresponding changes to the documentation
- [x] I have assigned this pull request to a milestone
Top commit has no ACKs.
Tree-SHA512: 08663e5371753b6851b216da66ef72bb6bfb80acd42c31c57de9f8d52f00766e3121adf960e27236451d6520088f0c2c927f5f4f3001a40c591feb62074951e0
Required to avoid crashes when calling RelayInvFiltered in situations
where the PeerManager* atomic hasn't been set (possible when ProcessMessage
isn't called, leaving the value unset, while a separate thread traverses
the ProcessPendingInstantSendLocks > ProcessPendingInstantSendLocks[1] >
ProcessInstantSendLock > RelayInvFiltered call chain).
[1] - There is a function with the exact same name but with multiple
arguments
ccd976dd3dbb8f991dc1203ada2043f1736be5a4 test: use 327 fewer blocks in feature_nulldummy (Jon Atack)
68c280f19732fb96bc29113ce9c8007d0101868c test, refactor: abstract the feature_nulldummy blockheight values (Jon Atack)
Pull request description:
The resolved timeout issue seen in the CI can be reproduced locally by running `test/functional/feature_nulldummy.py --valgrind --loglevel=debug`
Speeds up the normal test runtime for me from 3.8 to 2.2 seconds (debug build). Thanks to Marco Falke for the approach suggestion.
ACKs for top commit:
AnthonyRonning:
reACK ccd976dd3dbb8f991dc1203ada2043f1736be5a4 - ran a few times with the rest of the tests and still passing for me with just the fewer block change.
MarcoFalke:
review ACK ccd976dd3dbb8f991dc1203ada2043f1736be5a4 🏝
Tree-SHA512: 38339dca4276d1972e3a5a5ee436da64e9e58fd3b50b186e34b07ade9523ac4c03f6c3869c5f2a59c23b43c44f87e712f8297a01a8d83202049c081e3eeb4445
ffe33dfbd4c3b11e3475b022b6c1dd077613de79 chainparams: drop versionbits threshold to 90% for mainnnet and signet (Anthony Towns)
f054f6bcd2c2ce5fea84cf8681013f85a444e7ea versionbits: simplify state transitions (Anthony Towns)
55ac5f568a3b73d6f1ef4654617fb76e8bcbccdf versionbits: Add explicit NEVER_ACTIVE deployments (Anthony Towns)
dd07e6da48040dc7eae46bc7941db48d98a669fd fuzz: test versionbits delayed activation (Anthony Towns)
dd85d5411c1702c8ae259610fe55050ba212e21e tests: test versionbits delayed activation (Anthony Towns)
73d4a706393e6dbd6b6d6b6428f8d3233ac0a2d8 versionbits: Add support for delayed activation (Anthony Towns)
9e6b65f6fa205eee5c3b99343988adcb8d320460 tests: clean up versionbits test (Anthony Towns)
593274445004506c921d5d851361aefb3434d744 tests: test ComputeBlockVersion for all deployments (Anthony Towns)
63879f0a4760c0c0f784029849cb5d21ee088abb tests: pull ComputeBlockVersion test into its own function (Anthony Towns)
Pull request description:
BIP9-based implementation of "speedy trial" activation specification, see https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2021-March/018583.html
Edge cases are tested by fuzzing added in #21380.
ACKs for top commit:
instagibbs:
tACK ffe33dfbd4
jnewbery:
utACK ffe33dfbd4c3b11e3475b022b6c1dd077613de79
MarcoFalke:
review ACK ffe33dfbd4c3b11e3475b022b6c1dd077613de79 💈
achow101:
re-ACK ffe33dfbd4c3b11e3475b022b6c1dd077613de79
gmaxwell:
ACK ffe33dfbd4c3b11e3475b022b6c1dd077613de79
benthecarman:
ACK ffe33dfbd4c3b11e3475b022b6c1dd077613de79
Sjors:
ACK ffe33dfbd4c3b11e3475b022b6c1dd077613de79
jonatack:
Initial approach ACK ffe33dfbd4c3b11e3475b022b6c1dd077613de79 after a first pass of review, building and testing each commit, mostly looking at the changes and diffs. Will do a more high-level review iteration. A few minor comments follow to pick/choose/ignore.
ariard:
Code Review ACK ffe33df
Tree-SHA512: f79a7146b2450057ee92155cbbbcec12cd64334236d9239c6bd7d31b32eec145a9781c320f178da7b44ababdb8808b84d9d22a40e0851e229ba6d224e3be747c
08f3dbb1b0cd5ca01d87e488a2fa905adf7df057 test: Bump shellcheck version (Hennadii Stepanov)
Pull request description:
The changelog for v0.7.2 is available [here](https://github.com/koalaman/shellcheck/blob/v0.7.2/CHANGELOG.md).
Only [SC2268](https://github.com/koalaman/shellcheck/wiki/SC2268) requires to update our code.
ACKs for top commit:
jarolrod:
ACK 08f3dbb1b0cd5ca01d87e488a2fa905adf7df057
Tree-SHA512: 4585cd1f4d9def2fbaafe5a2a57761288d432781eb8c6c6d37064727d7ca8fc3f35c552e6a2ffdf0820d753d4bde2c8e43e5f3f57d242f5f57591a9b1b03558d
