There turned out to be a mismatch in the tx output counts which caused
'ConnectBlockMixedEcdsaSchnorr' benchmark to run slower than
'ConnectBlockAllEcdsa' and 'ConnectBlockAllSchnorr'. This commit makes
the tx output counts uniform across all benchmarks.
This commit also renames the 'taproot_tx' variable to 'tx' to reflect
that this variable represents a general tx and not just a taproot tx.
Since cluster_linearize.h does not actually have a Cluster type anymore, it is more
appropriate to rename the index type to DepGraphIndex.
-BEGIN VERIFY SCRIPT-
sed -i 's/Data type to represent transaction indices in clusters./Data type to represent transaction indices in DepGraphs and the clusters they represent./' $(git grep -l 'using ClusterIndex')
sed -i 's|\<ClusterIndex\>|DepGraphIndex|g' $(git grep -l 'ClusterIndex')
-END VERIFY SCRIPT-
7edaf8b64cb2d59ada22042fee62a417e52368b8 Benchmark Chainstate::ConnectBlock duration (Eunovo)
Pull request description:
Introduce benchmarks to evaluate ConnectBlock performance for:
- Blocks containing only Schnorr signatures
- Blocks containing both Schnorr and ECDSA signatures
- Blocks containing only ECDSA signatures
The benchmarks in this PR, focus on signature validation. Additional benchmarks may be added in the future to assess other aspects of ConnectBlock.
This is the first step toward implementing Batch Verification of Schnorr Signatures in Core. It provides a way to test and measure the performance improvements of batch verification on Core.
For more details on batch validation, refer to the [batch-verify module on secp](https://github.com/bitcoin-core/secp256k1/pull/1134) and [batch-verify on core](https://github.com/bitcoin/bitcoin/pull/29491).
ACKs for top commit:
josibake:
reACK 7edaf8b64c
fjahr:
utACK 7edaf8b64cb2d59ada22042fee62a417e52368b8
l0rinc:
ACK 7edaf8b64cb2d59ada22042fee62a417e52368b8
Tree-SHA512: 883c8a5e4e4de401ffb9ac9b6789b7fe0737afefbdaf02c6d7e1645392efc4f0d2d28b423ba7e34366a33608e0835793f5e7a1312b5c8063de14446319529cc7
ffff4a293ad878494e12f8f00108cc99ee2b713e bench: Update span-serialize comment (MarcoFalke)
fa4d6ec97bcb1790a7cd4363a13fda7c80c3dd90 refactor: Avoid false-positive gcc warning (MarcoFalke)
fa942332b40c97375af0722f32f7575bca3af819 scripted-diff: Bump copyright headers after std::span changes (MarcoFalke)
fa0c6b7179c062b7ca92d120455ce02a9f4e9e19 refactor: Remove unused Span alias (MarcoFalke)
fade0b5e5e6e80e3da1ab6448b6212244bafa5d3 scripted-diff: Use std::span over Span (MarcoFalke)
fadccc26c03db00a2be3f703aa7e5eec4312bd2e refactor: Make Span an alias of std::span (MarcoFalke)
fa27e36717ec18d64b7ff7bba71b8f0c202ba31d test: Fix broken span_tests (MarcoFalke)
fadf02ef8bf96ad5b3b8e34fd425b31b555f4371 refactor: Return std::span from MakeUCharSpan (MarcoFalke)
fa720b94be17fa9e7c91188710e6a04939ceab11 refactor: Return std::span from MakeByteSpan (MarcoFalke)
Pull request description:
`Span` has some issues:
* It does not support fixed-size spans, which are available through `std::span`.
* It is confusing to have it available and in use at the same time with `std::span`.
* It does not obey the standard library iterator build hardening flags. See https://github.com/bitcoin/bitcoin/issues/31272 for a discussion. For example, this allows to catch issues like the one fixed in commit fabeca3458b38a3d8930cb0cbc866388c3f120f1.
Both types are type-safe and can even implicitly convert into each other in most contexts.
However, exclusively using `std::span` seems less confusing, so do it here with a scripted-diff.
ACKs for top commit:
l0rinc:
reACK ffff4a293ad878494e12f8f00108cc99ee2b713e
theuni:
ACK ffff4a293ad878494e12f8f00108cc99ee2b713e.
Tree-SHA512: 9cc2f1f43551e2c07cc09f38b1f27d11e57e9e9bc0c6138c8fddd0cef54b91acd8b14711205ff949be874294a121910d0aceffe0e8914c4cff07f1e0e87ad5b8
4cd95a2921805f447a8bcecc6b448a365171eb93 refactor: modernize remaining outdated trait patterns (Lőrinc)
ab2b67fce20fd7d8017f8a26425cab99e91f420d scripted-diff: modernize outdated trait patterns - values (Lőrinc)
8327889f358289f918d04ddb9469fb5562720bf4 scripted-diff: modernize outdated trait patterns - types (Lőrinc)
Pull request description:
The use of [`std::underlying_type_t<T>`](https://en.cppreference.com/w/cpp/types/underlying_type) or [`std::is_enum_v<T>`](https://en.cppreference.com/w/cpp/types/is_enum) (and similar ones, introduced in C++14) replace the `typename std::underlying_type<T>::type` and `std::is_enum<T>::value` constructs (available in C++11).
The `_t` and `_v` helper alias templates offer a more concise way to extract the type and value directly.
I've modified the instances I found in the codebase one-by-one (noticed them while investigating https://github.com/bitcoin/bitcoin/pull/31868), and afterwards extracted scripted diff commits to do the trivial ones automatically.
The last commit contains the values that were easier done manually.
I've excluded changes from `src/bench/nanobench.h`, `src/leveldb`, `src/minisketch`, `src/span.h` and `src/sync.h` - let me know if you think they should be included instead.
A few of the code changes can also be reproduced by clang-tidy (but not all of them):
```bash
cmake -B build -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DBUILD_BENCH=ON -DBUILD_FUZZ_BINARY=ON -DBUILD_FOR_FUZZING=ON && cmake --build build -j$(nproc)
run-clang-tidy -quiet -p build -j $(nproc) -checks='-*,modernize-type-traits' -fix $(git grep -lE '::(value|type)' ./src ':(exclude)src/bench/nanobench.h' ':(exclude)src/leveldb' ':(exclude)src/minisketch' ':(exclude)src/span.h' ':(exclude)src/sync.h')
```
ACKs for top commit:
laanwj:
Concept and code review ACK 4cd95a2921805f447a8bcecc6b448a365171eb93
Tree-SHA512: a4bcf0f267c0f4e02983b4d548ed6f58d464ec379ac5cd1f998b9ec0cf698b53a9f2557a05a342b661f1d94adefc9a0ce2dc8f764d49453aaea95451e2c4c581
36b6f36ac4724cb2c9ed0e25314c3bbf55e4ebb8 build: require sqlite when building the wallet (Sjors Provoost)
Pull request description:
Require that sqlite is available in order to compile the wallet. Removes instances of `USE_SQLITE` since it is no longer possible to not have sqlite available.
The `NO_SQLITE` option is dropped from depends.
This is another step towards dropping the legacy wallet, extracted from #31250.
ACKs for top commit:
m3dwards:
ACK 36b6f36ac4724cb2c9ed0e25314c3bbf55e4ebb8
davidgumberg:
crACK 36b6f36ac4
hebasto:
re-ACK 36b6f36ac4724cb2c9ed0e25314c3bbf55e4ebb8.
Tree-SHA512: 870a0135671c80c4f28602119eb8637a1ed43b51b1673bfe88425782fb62ec6ef0f3d6baf0d5984d6a243779b0f63423fd4c4dc324ef87bffba13d63e05ad793
Historically, the headers have been bumped some time after a file has
been touched. Do it now to avoid having to touch them again in the
future for that reason.
-BEGIN VERIFY SCRIPT-
sed -i --regexp-extended 's;( 20[0-2][0-9])(-20[0-2][0-9])? The Bitcoin Core developers;\1-present The Bitcoin Core developers;g' $( git show --pretty="" --name-only HEAD~1 )
-END VERIFY SCRIPT-
Require that sqlite is available in order to compile the wallet. Removes
instances of USE_SQLITE since it is no longer possible to not have
sqlite available.
The NO_SQLITE option is dropped from depends.
Co-authored-by: Ava Chow <github@achow101.com>
Co-authored-by: Hennadii Stepanov <32963518+hebasto@users.noreply.github.com>
Measure ConnectBlock performance for
- blocks containing only schnorr sigs
- blocks containing both schnorr and ecdsa sigs
- blocks containing only ecdsa sigs
This will allow testing and measurement of performance improvement for features like
batch verification of schnorr signatures
f5883286e32b625aab3dd80c74d6adb4f37f0a80 Add a fuzz test for Num3072 multiplication and inversion (Pieter Wuille)
a26ce628942243fc9848a63bfdfa5e61f5e936f3 Safegcd based modular inverse for Num3072 (Pieter Wuille)
91ce8cef2d8955d980ab7e89fbf74e8b29adf178 Add benchmark for MuHash finalization (Pieter Wuille)
Pull request description:
This implements a safegcd-based modular inverse for MuHash3072. It is a fairly straightforward translation of [the libsecp256k1 implementation](https://github.com/bitcoin-core/secp256k1/pull/831), with the following changes:
* Generic for 32-bit and 64-bit
* Specialized for the specific MuHash3072 modulus (2^3072 - 1103717).
* A bit more C++ish
* Far fewer sanity checks
A benchmark is also included for MuHash3072::Finalize. The new implementation is around 100x faster on x86_64 for me (from 5.8 ms to 57 μs); for 32-bit code the factor is likely even larger.
For more information:
* [Original paper](https://gcd.cr.yp.to/papers.html) by Daniel J. Bernstein and Bo-Yin Yang
* [Implementation](https://github.com/bitcoin-core/secp256k1/pull/767) for libsecp256k1 by Peter Dettman; and the [final](https://github.com/bitcoin-core/secp256k1/pull/831) version
* [Explanation](https://github.com/bitcoin-core/secp256k1/blob/master/doc/safegcd_implementation.md) of the algorithm using Python snippets
* [Analysis](https://github.com/sipa/safegcd-bounds) of the maximum number of iterations the algorithm needs
* [Formal proof in Coq](https://medium.com/blockstream/a-formal-proof-of-safegcd-bounds-695e1735a348) by Russell O'Connor (for the 256-bit version of the algorithm; here we use a 3072-bit one).
ACKs for top commit:
achow101:
ACK f5883286e32b625aab3dd80c74d6adb4f37f0a80
TheCharlatan:
Re-ACK f5883286e32b625aab3dd80c74d6adb4f37f0a80
dergoegge:
tACK f5883286e32b625aab3dd80c74d6adb4f37f0a80
Tree-SHA512: 275872c61d30817a82901dee93fc7153afca55c32b72a95b8768f3fd464da1b09b36f952f30e70225e766b580751cfb9b874b2feaeb73ffaa6943c8062aee19a
18619b473255786b80f27dd33b46eb26a63fffc7 wallet: remove BDB dependency from wallet migration benchmark (furszy)
Pull request description:
Part of the legacy wallet removal working path #20160.
Stops creating a bdb database in the wallet migration benchmark.
Instead, the benchmark now creates the db in memory and re-uses it for the migration process.
ACKs for top commit:
achow101:
ACK 18619b473255786b80f27dd33b46eb26a63fffc7
brunoerg:
code review ACK 18619b473255786b80f27dd33b46eb26a63fffc7
theStack:
Code-review ACK 18619b473255786b80f27dd33b46eb26a63fffc7
Tree-SHA512: a107deee3d2c00b980e3606be07d038ca524b98251442956d702a7996e2ac5e2901f656482018cacbac8ef6a628ac1fb03f677d1658aeaded4036d834a95d7e0
a4df12323c4e9230bca58562ba17ecee4233f8fe doc: add release notes (Sjors Provoost)
c75872ffdd98ce9f04fb8489515d1b63853f03b4 test: use DIFF_1_N_BITS in tool_signet_miner (tdb3)
4131f322ac0abe43639065362cd3c4ea36d2c5c3 test: check difficulty adjustment using alternate mainnet (Sjors Provoost)
c4f68c12e228818f655352d17d038dcc7ba1db3a Use OP_0 for BIP34 padding in signet and tests (Sjors Provoost)
cf0a62878be214cd4ec779aab214221b27b769b6 rpc: add next to getmininginfo (Sjors Provoost)
2d18a078a2d9eaa53b8b7acc7600141c69f0d742 rpc: add target and bits to getchainstates (Sjors Provoost)
f153f57acc9f9a6f84af161d5bed9aa9965abaa3 rpc: add target and bits to getblockchaininfo (Sjors Provoost)
baa504fdfaff4a9f61bc939035df5d5f2978cfd7 rpc: add target to getmininginfo result (Sjors Provoost)
2a7bfebd5e788e1d9e7e07a9f1b8e3625a0301cd Add target to getblock(header) in RPC and REST (Sjors Provoost)
341f93251677fee66c822f414b75499e8b3b31f6 rpc: add GetTarget helper (Sjors Provoost)
d20d96fa41ce706ccc480b4f3143438ce0720348 test: use REGTEST_N_BITS in feature_block (tdb3)
7ddbed4f9fc0c90bfed244a71194740a4a1fa1be rpc: add nBits to getmininginfo (Sjors Provoost)
ba7b9f3d7bf5a1ad395262b080e832f5c9958e4d build: move pow and chain to bitcoin_common (Sjors Provoost)
c4cc9e3e9df2733260942e0513dd8478d2a104da consensus: add DeriveTarget() to pow.h (Sjors Provoost)
Pull request description:
**tl&dr for consensus-code only reviewers**: the first commit splits `CheckProofOfWorkImpl()` in order to create a `DeriveTarget()` helper. The rest of this PR does not touch consensus code.
There are three ways to represent the proof-of-work in a block:
1. nBits
2. Difficulty
3. Target
The latter notation is useful when you want to compare share work against either the pool target (to get paid) or network difficulty (found an actual block). E.g. for difficulty 1 which corresponds to an nBits value of `0x00ffff`:
```
share hash: f6b973257df982284715b0c7a20640dad709d22b0b1a58f2f88d35886ea5ac45
target: 7fffff0000000000000000000000000000000000000000000000000000000000
```
It's immediately clear that the share is invalid because the hash is above the target.
This type of logging is mostly done by the pool software. It's a nice extra convenience, but not very important. It impacts the following RPC calls:
1. `getmininginfo` displays the `target` for the tip block
2. `getblock` and `getblockheader` display the `target` for a specific block (ditto for their REST equivalents)
The `getdifficulty` method is a bit useless in its current state, because what miners really want to know if the difficulty for the _next_ block. So I added a boolean argument `next` to `getdifficulty`. (These values are typically the same, except for the first block in a retarget period. On testnet3 / testnet4 they change when no block is found after 20 minutes).
Similarly I added a `next` object to `getmininginfo` which shows `bit`, `difficulty` and `target` for the next block.
In order to test the difficulty transition, an alternate mainnet chain with 2016 blocks was generated and used in `mining_mainnet.py`. The chain is deterministic except for its timestamp and nonce values, which are stored in `mainnet_alt.json`.
As described at the top, this PR introduces a helper method `DeriveTarget()` which is split out from `CheckProofOfWorkImpl`. The proposed `checkblock` RPC in #31564 needs this helper method internally to figure out the consensus target.
Finally, this PR moves `pow.cpp` and `chain.cpp` from `bitcoin_node` to `bitcoin_common`, in order to give `rpc/util.cpp` (which lives in `bitcoin_common`) access to `pow.h`.
ACKs for top commit:
ismaelsadeeq:
re-ACK a4df12323c4e9230bca58562ba17ecee4233f8fe
tdb3:
code review re ACK a4df12323c4e9230bca58562ba17ecee4233f8fe
ryanofsky:
Code review ACK a4df12323c4e9230bca58562ba17ecee4233f8fe. Only overall changes since last review were dropping new `gettarget` method and dropping changes to `getdifficulty`, but there were also various internal changes splitting and rearranging commits.
Tree-SHA512: edef5633590379c4be007ac96fd1deda8a5b9562ca6ff19fe377cb552b5166f3890d158554c249ab8345977a06da5df07866c9f42ac43ee83dfe3830c61cd169
Done in separate commit to simplify review.
Also renames benchmarks, since they're not strictly tests.
Co-authored-by: Hodlinator <172445034+hodlinator@users.noreply.github.com>
Stops creating a bdb database in the wallet migration benchmark.
Instead, the benchmark now creates the db in memory and re-uses
it for the migration process.
The check type function now needs to return a std::optional<R> for some type R,
and the check queue overall will return std::nullopt if all individual checks
return that, or one of the non-nullopt values if there is at least one.
For most tests, we use R=int, but for the actual validation code, we make it return
the ScriptError.
11f3bc229ccd4b20191855fb1df882cfa6145264 refactor: Reserve vectors in fuzz tests (Lőrinc)
152fefe7a22b7da3cfe2815083634bece9c5654e refactor: Preallocate PrevectorFillVector(In)Direct without vector resize (Lőrinc)
a774c7a339c26b1409c9a9572d2b52810ee64062 refactor: Fix remaining clang-tidy performance-inefficient-vector errors (Lőrinc)
Pull request description:
PR inspired by https://github.com/bitcoin/bitcoin/pull/29608#issuecomment-2437847307 (and https://github.com/bitcoin/bitcoin/pull/29458, https://github.com/bitcoin/bitcoin/pull/29606, https://github.com/bitcoin/bitcoin/pull/29607, https://github.com/bitcoin/bitcoin/pull/30093).
The `clang-tidy` check can be run via:
```bash
cmake -B build -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DBUILD_BENCH=ON -DBUILD_FUZZ_BINARY=ON -DBUILD_FOR_FUZZING=ON && cmake --build build -j$(nproc)
run-clang-tidy -quiet -p build -j $(nproc) -checks='-*,performance-inefficient-vector-operation' | grep -v 'clang-tidy'
```
which revealed 3 tests and 1 prod warning (+ fuzz and benching, found by hebasto).
Even though the tests aren't performance critical, getting rid of these warnings (for which the checks were already enabled via https://github.com/bitcoin/bitcoin/blob/master/src/.clang-tidy#L18, see below), the fix was quite simple.
<details>
<summary>clang-tidy -list-checks</summary>
```bash
cd src && clang-tidy -list-checks | grep 'vector'
performance-inefficient-vector-operation
```
</details>
<details>
<summary>Output before the change</summary>
```
src/test/rpc_tests.cpp:434:9: error: 'emplace_back' is called inside a loop; consider pre-allocating the container capacity before the loop [performance-inefficient-vector-operation,-warnings-as-errors]
433 | for (int64_t i = 0; i < 100; i++) {
434 | feerates.emplace_back(1 ,1);
| ^
src/test/checkqueue_tests.cpp:366:13: error: 'emplace_back' is called inside a loop; consider pre-allocating the container capacity before the loop [performance-inefficient-vector-operation,-warnings-as-errors]
365 | for (size_t i = 0; i < 3; ++i) {
366 | tg.emplace_back(
| ^
src/test/cuckoocache_tests.cpp:231:9: error: 'emplace_back' is called inside a loop; consider pre-allocating the container capacity before the loop [performance-inefficient-vector-operation,-warnings-as-errors]
228 | for (uint32_t x = 0; x < 3; ++x)
229 | /** Each thread is emplaced with x copy-by-value
230 | */
231 | threads.emplace_back([&, x] {
| ^
src/rpc/output_script.cpp:127:17: error: 'push_back' is called inside a loop; consider pre-allocating the container capacity before the loop [performance-inefficient-vector-operation,-warnings-as-errors]
126 | for (unsigned int i = 0; i < keys.size(); ++i) {
127 | pubkeys.push_back(HexToPubKey(keys[i].get_str()));
| ^
```
And the fuzz and benchmarks, noticed by hebasto: https://github.com/bitcoin/bitcoin/pull/31305#issuecomment-2483124499
</details>
ACKs for top commit:
maflcko:
review ACK 11f3bc229ccd4b20191855fb1df882cfa6145264 🎦
achow101:
ACK 11f3bc229ccd4b20191855fb1df882cfa6145264
theuni:
ACK 11f3bc229ccd4b20191855fb1df882cfa6145264
hebasto:
ACK 11f3bc229ccd4b20191855fb1df882cfa6145264, tested with clang 19.1.5 + clang-tidy.
Tree-SHA512: 41691c19f35c63b922a95407617a54f9bff1af3f95f99d15642064f321df038aeb1ae5f061f854ed913f69036807cc28fa6222b2ff4c24ef43b909027fa0f9b3
5736d1ddacc4019101e7a5170dd25efbc63b622a tracing: pass if replaced by tx/pkg to tracepoint (0xb10c)
a4ec07f1944999c2eead41d08d7dd4fc3aa71243 doc: add comments for CTxMemPool::ChangeSet (Suhas Daftuar)
83f814b1d1100baac9dca9c176f89b0ec2555dbc Remove m_all_conflicts from SubPackageState (Suhas Daftuar)
d3c8e7dfb63f7986a1f9654ea2393aabe3cd78da Ensure that we don't add duplicate transactions in rbf fuzz tests (Suhas Daftuar)
d7dc9fd2f7bc675256687b9c55fdbec9cc8ac781 Move CalculateChunksForRBF() to the mempool changeset (Suhas Daftuar)
284a1d33f1dcbc3b3404ea40a948ff6600239613 Move prioritisation into changeset (Suhas Daftuar)
446b08b599bc492bbec10ccc2292aee6f90c58e7 Don't distinguish between direct conflicts and all conflicts when doing cluster-size-2-rbf checks (Suhas Daftuar)
b53041021abc4f9ee7203341413e8676e2d5a7ca Duplicate transactions are not permitted within a changeset (Suhas Daftuar)
b447416fddcb8c8647391502cca3dbfd1552e02e Public mempool removal methods Assume() no changeset is outstanding (Suhas Daftuar)
2b30f4d36c86f775ac637b171d27d42a02309c5b Make RemoveStaged() private (Suhas Daftuar)
18829194ca68152ac0b38d34e94b9265ee74c410 Enforce that there is only one changeset at a time (Suhas Daftuar)
7fb62f7db60c7d793828ae45f87bc3f5c63cc989 Apply mempool changeset transactions directly into the mempool (Suhas Daftuar)
34b6c5833d11ea84fbd4b891e06408f6f4ca6fac Clean up FinalizeSubpackage to avoid workspace-specific information (Suhas Daftuar)
57983b8add72a04721d3f2050c063a3c4d8683ed Move LimitMempoolSize to take place outside FinalizeSubpackage (Suhas Daftuar)
01e145b9758f1df14a7ea18058ba9577bf88e459 Move changeset from workspace to subpackage (Suhas Daftuar)
802214c0832de00f24268183f7763fa984ba7903 Introduce mempool changesets (Suhas Daftuar)
87d92fa340195d9c87be3d023ca133b90b3b7d4e test: Add unit test coverage of package rbf + prioritisetransaction (Suhas Daftuar)
15d982f91e6b0f145c9dd4edf29827cfabb37a3f Add package hash to package-rbf log message (Suhas Daftuar)
Pull request description:
part of cluster mempool: #30289
It became clear while working on cluster mempool that it would be helpful for transaction validation if we could consider a full set of proposed changes to the mempool -- consisting of a set of transactions to add, and a set of transactions (ie conflicts) to simultaneously remove -- and perform calculations on what the mempool would look like if the proposed changes were to be applied. Two specific examples of where we'd like to do this:
- Determining if ancestor/descendant/TRUC limits would be violated (in the future, cluster limits) if either a single transaction or a package of transactions were to be accepted
- Determining if an RBF would make the mempool "better", however that idea is defined, both in the single transaction and package of transaction cases
In preparation for cluster mempool, I have pulled this reworking of the mempool interface out of #28676 so it can be reviewed on its own. I have not re-implemented ancestor/descendant limits to be run through the changeset, since with cluster mempool those limits will be going away, so this seems like wasted effort. However, I have rebased #28676 on top of this branch so reviewers can see what the new mempool interface could look like in the cluster mempool setting.
There are some minor behavior changes here, which I believe are inconsequential:
- In the package validation setting, transactions would be added to the mempool before the `ConsensusScriptChecks()` are run. In theory, `ConsensusScriptChecks()` should always pass if the `PolicyScriptChecks()` have passed and it's just a belt-and-suspenders for us, but if somehow they were to diverge then there could be some small behavior change from adding transactions and then removing them, versus never adding them at all.
- The error reporting on `CheckConflictTopology()` has slightly changed due to no longer distinguishing between direct conflicts and indirect conflicts. I believe this should be entirely inconsequential because there shouldn't be a logical difference between those two ideas from the perspective of this function, but I did have to update some error strings in some tests.
- Because, in a package setting, RBFs now happen as part of the entire package being accepted, the logging has changed slightly because we do not know which transaction specifically evicted a given removed transaction.
- Specifically, the "package hash" is now used to reference the set of transactions that are being accepted, rather than any single txid. The log message relating to package RBF that happen in the `TXPACKAGES` category has been updated as well to include the package hash, so that it's possible to see which specific set of transactions are being referenced by that package hash.
- Relatedly, the tracepoint logging in the package rbf case has been updated as well to reference the package hash, rather than a transaction hash.
ACKs for top commit:
naumenkogs:
ACK 5736d1ddac
instagibbs:
ACK 5736d1ddacc4019101e7a5170dd25efbc63b622a
ismaelsadeeq:
reACK 5736d1ddacc4019101e7a5170dd25efbc63b622a
glozow:
ACK 5736d1ddacc
Tree-SHA512: 21810872e082920d337c89ac406085aa71c5f8e5151ab07aedf41e6601f60a909b22fbf462ef3b735d5d5881e9b76142c53957158e674dd5dfe6f6aabbdf630b
42066f45ff5d48e78a317eda63c035809bd657c6 Refactor SipHash_32b benchmark to improve accuracy and avoid optimization issues (Lőrinc)
Pull request description:
This PR stems from the discussions in https://github.com/bitcoin/bitcoin/pull/30317#discussion_r1649187336
The previous benchmark for `SipHash` was slightly less accurate in representing real-world usage and allowed for potential compiler optimizations that could invalidate the benchmark.
This change aims to ensure the benchmark produces more realistic results.
By modifying the initial values and only incrementing the bytes of `val`, the benchmark should reflects a more typical usage patterns - and prevent the compiler from optimizing away the calculations.
-------
On my M1 processor the benchmark's speed changed significantly (but the CI seems to produce the same result as before):
> cmake -B build -DCMAKE_BUILD_TYPE=Release -DBUILD_BENCH=ON && cmake --build build -j10 &&
./build/src/bench/bench_bitcoin --filter=SipHash_32b --min-time=1000
Before:
| ns/op | op/s | err% | total | benchmark
|--------------------:|--------------------:|--------:|----------:|:----------
| 35.15 | 28,445,856.66 | 0.2% | 1.10 | `SipHash_32b`
After (note that only the benchmark changed):
| ns/op | op/s | err% | total | benchmark
|--------------------:|--------------------:|--------:|----------:|:----------
| 22.05 | 45,350,886.64 | 0.3% | 1.10 | `SipHash_32b`
ACKs for top commit:
maflcko:
review ACK 42066f45ff5d48e78a317eda63c035809bd657c6
achow101:
ACK 42066f45ff5d48e78a317eda63c035809bd657c6
hodlinator:
ACK 42066f45ff5d48e78a317eda63c035809bd657c6
Tree-SHA512: 6bbe9d725d4c3396642e55ce48c31baa5339e56838d6d5fb377fb1069daa9292375e7020ceff7da0d78befffc1e984f717b5232217fe911989613480adaa937e
fa66e0887ca1a1445d8b18ba1fadb12b2d911048 bench: add support for custom data directory (furszy)
ad9c2cceda9cd893c0f754e49f7fca6e417ee95f test, bench: specialize working directory name (furszy)
Pull request description:
Expands the benchmark framework with the existing `-testdatadir` arg,
enabling the ability to change the benchmark data directory.
This is useful for running benchmarks on different storage devices, and
not just under the OS `/tmp/` directory.
A good use case is #28574, where we are benchmarking the wallet
migration process on an HDD.
ACKs for top commit:
maflcko:
re-ACK fa66e0887ca1a1445d8b18ba1fadb12b2d911048
achow101:
ACK fa66e0887ca1a1445d8b18ba1fadb12b2d911048
tdb3:
re ACK fa66e0887ca1a1445d8b18ba1fadb12b2d911048
hodlinator:
re-ACK fa66e0887ca1a1445d8b18ba1fadb12b2d911048
pablomartin4btc:
re-ACK fa66e0887ca1a1445d8b18ba1fadb12b2d911048
Tree-SHA512: 4e87206c07e26fe193c07074ae9eb0cc9c70a58aeea8cf27d18fb5425d77e4b00dbe0e6d6a75c17b427744e9066458b9a84e5ef7b0420f02a4fccb9c5ef4dacc
Rather than individually calling addUnchecked for each transaction added in a
changeset (after removing all the to-be-removed transactions), instead we can
take advantage of boost::multi_index's splicing features to extract and insert
entries directly from the staging multi_index into mapTx.
This has the immediate advantage of saving allocation overhead for mempool
entries which have already been allocated once. This also means that the memory
locations of mempool entries will not change when transactions go from staging
to the main mempool.
Additionally, eliminate addUnchecked and require all new transactions to enter
the mempool via a CTxMemPoolChangeSet.
Expands the benchmark framework with the existing '-testdatadir' arg,
enabling the ability to change the benchmark data directory.
This is useful for running benchmarks on different storage devices, and
not just under the OS /tmp/ directory.
Since G_TEST_GET_FULL_NAME is not initialized in the benchmark framework,
benchmarks using the unit test setup run in the same directory without
any clear distinction between them.
This poses an extra complication for locating any specific benchmark
directory during a failure.
In master, unit tests and benchmarks run in the following path:
/<OS_tmp_dir>/test_common bitcoin/<random_uint256>/
After this commit, unit tests and benchmarks are contained within its
own directory:
/<OS_tmp_dir>/test_common bitcoin/<test_name>/<time_in_nanoseconds>/
This makes it easier to find any benchmark run when a failure occurs.
- Modify `SipHash_32b` benchmark to use `FastRandomContext` for generating initial values.
- Cycle through and modify each byte of the `uint256` value to ensure no part of it can be optimized away.
The lack of "recursion" (where the method call overwrites the used inputs partially) and the systematic modification of each input byte makes the benchmark usage more reliable and thorough.
0b3ec8c59b2b6d598531cd4e688eb276e597c825 clusterlin: remove Cluster type (Pieter Wuille)
1c24c625105cd62518d2eee7e95c3b1c74ea9923 clusterlin: merge two DepGraph fuzz tests into simulation test (Pieter Wuille)
0606e66fdbb914f984433d8950f0c32b5fb871f3 clusterlin: add DepGraph::RemoveTransactions and support for holes in DepGraph (Pieter Wuille)
75b5d42419ed1286fc9557bc97ec5bac3f9be837 clusterlin: make DepGraph::AddDependency support multiple dependencies at once (Pieter Wuille)
abf50649d13018bf40c5803730a03053737efeee clusterlin: simplify DepGraphFormatter::Ser (Pieter Wuille)
eaab55ffc8102140086297877747b7fa07b419eb clusterlin: rework DepGraphFormatter::Unser (Pieter Wuille)
5901cf7100a75c8131223e23b6c90e0e93611eae clusterlin: abstract out DepGraph::GetReduced{Parents,Children} (Pieter Wuille)
Pull request description:
Part of cluster mempool: #30289
This adds:
* `DepGraph::AddDependencies` to add 0 or more dependencies to a single transaction at once (identical to calling `DepGraph::AddDependency` once for each, but more efficient).
* `DepGraph::RemoveTransactions` to remove 0 or more transactions from a depgraph.
* `DepGraph::GetReducedParents` (and `DepGraph::GetReducedChildren`) to get the (reduced) direct parents and children of a transaction in a depgraph.
After which, the `Cluster` type is removed.
This is the result of fleshing out the design for the "intermediate layer" ("TxGraph", no PR yet) between the cluster linearization layer and the mempool layer. My earlier thinking was that TxGraph would store `Cluster` objects (vectors of pairs of `FeeFrac`s and sets of parents), and convert them to `DepGraph` on the fly whenever needed. However, after more consideration, it seems better to have TxGraph store `DepGraph` objects, and manipulate them directly without constantly re-creating them. This requires `DepGraph` to have some additional functionality.
The bulk of the complexity here is the addition of `DepGraph::RemoveTransactions`, which leaves the remaining transactions' positions within the `DepGraph` untouched (we want existing identifiers to remain valid), so this implies that graphs can now have "holes" (positions that are unused, but followed by positions that are used). To enable that, an extension of the fuzz/test serialization format `DepGraphFormatter` is included to deal with such holes.
ACKs for top commit:
sdaftuar:
reACK 0b3ec8c59b2b6d598531cd4e688eb276e597c825
instagibbs:
reACK 0b3ec8c59b2b6d598531cd4e688eb276e597c825
ismaelsadeeq:
reACK 0b3ec8c59b2b6d598531cd4e688eb276e597c825
glozow:
ACK 0b3ec8c59b2b6d598531cd4e688eb276e597c825, reviewed range-diff from aab53ddcd8fcbc3c0be0da9383f8e06abe5badda and `clusterlin_depgraph_sim`
Tree-SHA512: a804b7f26d544c5cb0847322e235c810525cb0607737be6116c3156d582da3ba3352af8ea48e74eed5268f9c3eca63b30181d01b23a6dd0be1b99191f81cceb0
This changes DepGraph::AddDependency into DepGraph::AddDependencies, which takes
in a single child, but a set of parent transactions, making them all dependencies
at once.
This is important for performance. N transactions can have O(N^2) parents combined,
so constructing a full DepGraph using just AddDependency (which is O(N) on its own)
could take O(N^3) time, while doing the same with AddDependencies (also O(N) on its
own) only takes O(N^2).
Notably, this matters for DepGraphFormatter::Unser, which goes from O(N^3) to O(N^2).
Co-Authored-By: Greg Sanders <gsanders87@gmail.com>
a240e150e837b5a95ed19765a2e8b7c5b6013f35 streams: remove AutoFile::Get() entirely (Pieter Wuille)
e624a9bef16b6335fd119c10698352b59bf2930a streams: cache file position within AutoFile (Pieter Wuille)
Pull request description:
Fixes#30833.
Instead of relying on frequent `ftell` calls (which appear to cause a significant slowdown on some systems) in XOR-enabled `AutoFile`s, cache the file position within `AutoFile` itself.
ACKs for top commit:
achow101:
ACK a240e150e837b5a95ed19765a2e8b7c5b6013f35
davidgumberg:
untested reACK a240e150e8
theStack:
Code-review ACK a240e150e837b5a95ed19765a2e8b7c5b6013f35
Tree-SHA512: fd3681edc018afaf955dc7a41a0c953ca80d46c1129e3c5b306c87c95aae93b2fe7b900794eb8b6f10491f9211645e7939918a28838295e6873eb226fca7006f
e4e3b44e9cc7227b3ad765397c884999f57bac2e net: call `Select` with reachable networks in `ThreadOpenConnections` (brunoerg)
829becd990b504a2e8a57fa8a6ff6ac6ae8ff900 addrman: change `Select` to support multiple networks (brunoerg)
f698636ec86c004ab331994559c163b7319e6423 net: add `All()` in `ReachableNets` (brunoerg)
Pull request description:
This PR changes addrman's `Select` to support multiple networks and change `ThreadOpenConnections` to call it with reachable networks. It can avoid unnecessary `Select` calls and avoid exceeding the max number of tries (100), especially when turning a clearnet + Tor/I2P/CJDNS node to Tor/I2P/CJDNS. Compared to #29330, this approach is "less aggresive". It does not add a new init flag and does not impact address relay.
I did an experiment of calling `Select` without passing a network until it finds an address from a network that compose 20% ~ 25% of the addrman (limited to 100 tries).
ACKs for top commit:
achow101:
ACK e4e3b44e9cc7227b3ad765397c884999f57bac2e
vasild:
ACK e4e3b44e9cc7227b3ad765397c884999f57bac2e
naumenkogs:
ACK e4e3b44e9cc7227b3ad765397c884999f57bac2e
Tree-SHA512: e8466b72b85bbc2ad8bfb14471eb27d2c50d4e84218f5ede2c15a6fa3653af61b488cde492dbd398f7502bd847e95bfee1abb7e01092daba2236d3ce3d6d2268
