17cf9ff7ef Use cluster size limit for -maxmempool bound, and allow -maxmempool=0 in general (Suhas Daftuar)
315e43e5d8 Sanity check `GetFeerateDiagram()` in CTxMemPool::check() (Suhas Daftuar)
de2e9a24c4 test: extend package rbf functional test to larger clusters (Suhas Daftuar)
4ef4ddb504 doc: update policy/packages.md for new package acceptance logic (Suhas Daftuar)
79f73ad713 Add check that GetSortedScoreWithTopology() agrees with CompareMiningScoreWithTopology() (Suhas Daftuar)
a86ac11768 Update comments for CTxMemPool class (Suhas Daftuar)
9567eaa66d Invoke TxGraph::DoWork() at appropriate times (Suhas Daftuar)
6c5c44f774 test: add functional test for new cluster mempool RPCs (Suhas Daftuar)
72f60c877e doc: Update mempool_replacements.md to reflect feerate diagram checks (Suhas Daftuar)
21693f031a Expose cluster information via rpc (Suhas Daftuar)
72e74e0d42 fuzz: try to add more code coverage for mempool fuzzing (Suhas Daftuar)
f107417490 bench: add more mempool benchmarks (Suhas Daftuar)
7976eb1ae7 Avoid violating mempool policy limits in tests (Suhas Daftuar)
84de685cf7 Stop tracking parents/children outside of txgraph (Suhas Daftuar)
88672e205b Rewrite GatherClusters to use the txgraph implementation (Suhas Daftuar)
1ca4f01090 Fix miniminer_tests to work with cluster limits (Suhas Daftuar)
1902111e0f Eliminate CheckPackageLimits, which no longer does anything (Suhas Daftuar)
3a646ec462 Rework RBF and TRUC validation (Suhas Daftuar)
19b8479868 Make getting parents/children a function of the mempool, not a mempool entry (Suhas Daftuar)
5560913e51 Rework truc_policy to use descendants, not children (Suhas Daftuar)
a4458d6c40 Use txgraph to calculate descendants (Suhas Daftuar)
c8b6f70d64 Use txgraph to calculate ancestors (Suhas Daftuar)
241a3e666b Simplify ancestor calculation functions (Suhas Daftuar)
b9cec7f0a1 Make removeConflicts private (Suhas Daftuar)
0402e6c780 Remove unused limits from CalculateMemPoolAncestors (Suhas Daftuar)
08be765ac2 Remove mempool logic designed to maintain ancestor/descendant state (Suhas Daftuar)
fc4e3e6bc1 Remove unused members from CTxMemPoolEntry (Suhas Daftuar)
ff3b398d12 mempool: eliminate accessors to mempool entry ancestor/descendant cached state (Suhas Daftuar)
b9a2039f51 Eliminate use of cached ancestor data in miniminer_tests and truc_policy (Suhas Daftuar)
ba09fc9774 mempool: Remove unused function CalculateDescendantMaximum (Suhas Daftuar)
8e49477e86 wallet: Replace max descendant count with cluster_count (Suhas Daftuar)
e031085fd4 Eliminate Single-Conflict RBF Carve Out (Suhas Daftuar)
cf3ab8e1d0 Stop enforcing descendant size/count limits (Suhas Daftuar)
89ae38f489 test: remove rbf carveout test from mempool_limit.py (Suhas Daftuar)
c0bd04d18f Calculate descendant information for mempool RPC output on-the-fly (Suhas Daftuar)
bdcefb8a8b Use mempool/txgraph to determine if a tx has descendants (Suhas Daftuar)
69e1eaa6ed Add test case for cluster size limits to TRUC logic (Suhas Daftuar)
9cda64b86c Stop enforcing ancestor size/count limits (Suhas Daftuar)
1f93227a84 Remove dependency on cached ancestor data in mini-miner (Suhas Daftuar)
9fbe0a4ac2 rpc: Calculate ancestor data from scratch for mempool rpc calls (Suhas Daftuar)
7961496dda Reimplement GetTransactionAncestry() to not rely on cached data (Suhas Daftuar)
feceaa42e8 Remove CTxMemPool::GetSortedDepthAndScore (Suhas Daftuar)
21b5cea588 Use cluster linearization for transaction relay sort order (Suhas Daftuar)
6445aa7d97 Remove the ancestor and descendant indices from the mempool (Suhas Daftuar)
216e693729 Implement new RBF logic for cluster mempool (Suhas Daftuar)
ff8f115dec policy: Remove CPFP carveout rule (Suhas Daftuar)
c3f1afc934 test: rewrite PopulateMempool to not violate mempool policy (cluster size) limits (Suhas Daftuar)
47ab32fdb1 Select transactions for blocks based on chunk feerate (Suhas Daftuar)
dec138d1dd fuzz: remove comparison between mini_miner block construction and miner (Suhas Daftuar)
6c2bceb200 bench: rewrite ComplexMemPool to not create oversized clusters (Suhas Daftuar)
1ad4590f63 Limit mempool size based on chunk feerate (Suhas Daftuar)
b11c89cab2 Rework miner_tests to not require large cluster limit (Suhas Daftuar)
95a8297d48 Check cluster limits when using -walletrejectlongchains (Suhas Daftuar)
95762e6759 Do not allow mempool clusters to exceed configured limits (Suhas Daftuar)
edb3e7cdf6 [test] rework/delete feature_rbf tests requiring large clusters (glozow)
435fd56711 test: update feature_rbf.py replacement test (Suhas Daftuar)
34e32985e8 Add new (unused) limits for cluster size/count (Suhas Daftuar)
838d7e3553 Add transactions to txgraph, but without cluster dependencies (Suhas Daftuar)
d5ed9cb3eb Add accessor for sigops-adjusted weight (Suhas Daftuar)
1bf3b51396 Add sigops adjusted weight calculator (Suhas Daftuar)
c18c68a950 Create a txgraph inside CTxMemPool (Suhas Daftuar)
29a94d5b2f Make CTxMemPoolEntry derive from TxGraph::Ref (Suhas Daftuar)
92b0079fe3 Allow moving CTxMemPoolEntry objects, disallow copying (Suhas Daftuar)
6c73e47448 mempool: Store iterators into mapTx in mapNextTx (Suhas Daftuar)
51430680ec Allow moving an Epoch::Marker (Suhas Daftuar)
Pull request description:
[Reopening #28676 here as a new PR, because GitHub is slow to load the page making it hard to scroll through and see comments. Also, that PR was originally opened with a prototype implementation which has changed significantly with the introduction of `TxGraph`.]
This is an implementation of the [cluster mempool proposal](https://delvingbitcoin.org/t/an-overview-of-the-cluster-mempool-proposal/393).
This branch implements the following observable behavior changes:
- Maintains a partitioning of the mempool into connected clusters (via the `txgraph` class), which are limited in vsize to 101 kvB by default, and limited in count to 64 by default.
- Each cluster is sorted ("linearized") to try to optimize for selecting highest-feerate-subsets of a cluster first
- Transaction selection for mining is updated to use the cluster linearizations, selecting highest feerate "chunks" first for inclusion in a block template.
- Mempool eviction is updated to use the cluster linearizations, selecting lowest feerate "chunks" first for removal.
- The RBF rules are updated to: (a) drop the requirement that no new inputs are introduced; (b) change the feerate requirement to instead check that the feerate diagram of the mempool will strictly improve; (c) replace the direct conflicts limit with a directly-conflicting-clusters limit.
- The CPFP carveout rule is eliminated (it doesn't make sense in a cluster-limited mempool)
- The ancestor and descendant limits are no longer enforced.
- New cluster count/cluster vsize limits are now enforced instead.
- Transaction relay now uses chunk feerate comparisons to determine the order that newly received transactions are announced to peers.
Additionally, the cached ancestor and descendant data are dropped from the mempool, along with the multi_index indices that were maintained to sort the mempool by ancestor and descendant feerates. For compatibility (eg with wallet behavior or RPCs exposing this), this information is now calculated dynamically instead.
ACKs for top commit:
instagibbs:
reACK 17cf9ff7ef
glozow:
reACK 17cf9ff7ef
sipa:
ACK 17cf9ff7ef
Tree-SHA512: bbde46d913d56f8d9c0426cb0a6c4fa80b01b0a4c2299500769921f886082fb4f51f1694e0ee1bc318c52e1976d7ebed8134a64eda0b8044f3a708c04938eee7
Using std::ranges::copy from the C++ standard library has a few benefits
here:
* It has the additional benefit of being a bit more type safe and
document the byte cast explicitly.
* The compiler will likely optimize it to the same asm, but performance
doesn't really matter here anyway.
* It works around an UB-Sanitizer bug, when the source range is empty.
Fixes https://github.com/bitcoin/bitcoin/issues/33643
0f7d4ee4e8 p2p: Use different inbound inv timer per network (Martin Zumsande)
94db966a3b net: use generic network key for addrcache (Martin Zumsande)
Pull request description:
Currently, `NextInvToInbounds` schedules each round of `inv` at the same time for all inbound peers. It's being done this way because with a separate timer per peer (like it's done for outbounds), an attacker could do multiple connections to learn about the time a transaction arrived. (#13298).
However, having a single timer for inbounds of all networks is also an obvious fingerprinting vector: Connecting to a suspected pair of privacy-network and clearnet addresses and observing the `inv` pattern makes it trivial to confirm or refute that they are the same node.
This PR changes it such that a separate timer is used for each network.
It uses the existing method from `getaddr` caching and generalizes it to be saved in a new field `m_network_key` in `CNode` which will be used for both `getaddr` caching and `inv` scheduling, and can also be used for any future anti-fingerprinting measures.
ACKs for top commit:
sipa:
utACK 0f7d4ee4e8
stratospher:
reACK 0f7d4ee.
naiyoma:
Tested ACK 0f7d4ee4e8
danielabrozzoni:
reACK 0f7d4ee4e8
Tree-SHA512: e197c3005b2522051db432948874320b74c23e01e66988ee1ee11917dac0923f58c1252fa47da24e68b08d7a355d8e5e0a3ccdfa6e4324cb901f21dfa880cd9c
0802398e74 fuzz: make it possible to mock (fuzz) CThreadInterrupt (Vasil Dimov)
6d9e5d130d fuzz: add CConnman::SocketHandler() to the tests (Vasil Dimov)
3265df63a4 fuzz: add CConnman::InitBinds() to the tests (Vasil Dimov)
91cbf4dbd8 fuzz: add CConnman::CreateNodeFromAcceptedSocket() to the tests (Vasil Dimov)
50da7432ec fuzz: add CConnman::OpenNetworkConnection() to the tests (Vasil Dimov)
e6a917c8f8 fuzz: add Fuzzed NetEventsInterface and use it in connman tests (Vasil Dimov)
e883b37768 fuzz: set the output argument of FuzzedSock::Accept() (Vasil Dimov)
Pull request description:
Extend `CConnman` fuzz tests to also exercise the methods `OpenNetworkConnection()`, `CreateNodeFromAcceptedSocket()`, `InitBinds()` and `SocketHandler()`.
Previously fuzzing those methods would have resulted in real socket functions being called in the operating system which is undesirable during fuzzing. Now that https://github.com/bitcoin/bitcoin/pull/21878 is complete all those are mocked to a fuzzed socket and a fuzzed DNS resolver (see how `CreateSock` and `g_dns_lookup` are replaced in the first commit).
ACKs for top commit:
achow101:
ACK 0802398e74
jonatack:
Review re-ACK 0802398e74
dergoegge:
Code review ACK 0802398e74
Tree-SHA512: a717d4e79f42bacf2b029c821fdc265e10e4e5c41af77cd4cb452cc5720ec83c62789d5b3dfafd39a22cc8c0500b18169aa7864d497dded729a32ab863dd6c4d
The generic key can also be used in other places
where behavior between different network identities should
be uncorrelated to avoid fingerprinting.
This also changes RANDOMIZER_ID - since it is not
being persisted to disk, there are no compatibility issues.
* Make the methods of `CThreadInterrupt` virtual and store a pointer to
it in `CConnman`, thus making it possible to override with a mocked
instance.
* Initialize `CConnman::m_interrupt_net` from the constructor, making it
possible for callers to supply mocked version.
* Introduce `FuzzedThreadInterrupt` and `ConsumeThreadInterrupt()` and
use them in `src/test/fuzz/connman.cpp` and `src/test/fuzz/i2p.cpp`.
This improves the CPU utilization of the `connman` fuzz test.
As a nice side effect, the `std::shared_ptr` used for
`CConnman::m_interrupt_net` resolves the possible lifetime issues with
it (see the removed comment for that variable).
`FuzzedSock::Accept()` properly returns a new socket, but it forgot to
set the output argument `addr`, like `accept(2)` is expected to.
This could lead to reading uninitialized data during testing when we
read it, e.g. from `CService::SetSockAddr()` which reads the `sa_family`
member.
Set `addr` to a fuzzed IPv4 or IPv6 address.
32023 changed AddWalletDescriptor to return util::Error, but did not
change all of the failure cases to do so. This may result in some
callers continuing when there was actually an error. Unify all of the
failure cases to use util::Error so that all callers handle
AddWalletDescriptor errors in the same way.
The encapsulated return type is changed from ScriptPubKeyMan* to
std::reference_wrapper<DescriptorScriptPubKeyMan>. This avoids having a
value that can be interpreted as a bool, and also removes the need to
constantly dynamic_cast the returned value. The only kind of
ScriptPubKeyMan that can come out of AddWalletDescriptor is a
DescriptorScriptPubKeyMan anyways.
55b931934a removed duplicate calling of GetDescriptorScriptPubKeyMan (Saikiran)
Pull request description:
Removed duplicate call to GetDescriptorScriptPubKeyMan and
Instead of checking linearly I have used find method so time complexity reduced significantly for GetDescriptorScriptPubKeyMan
after this fix improved performance of importdescriptor part refs https://github.com/bitcoin/bitcoin/issues/32013.
**Steps to reproduce in testnet environment**
**Input size:** 2 million address in the wallet
**Step1:** call importaddresdescriptor rpc method
observe the time it has taken.
**With the provided fix:**
Do the same steps again
observe the time it has taken.
There is a huge improvement in the performance. (previously it may take 5 to 6 seconds now it will take 1 seconds or less)
main changes i've made during this pr:
1. remove duplicate call to GetDescriptorScriptPubKeyMan method
2. And inside GetDescriptorScriptPubKeyMan method previously we checking **each address linearly** so each time it is calling HasWallet method which has aquired lock.
3. Now i've modified this logic call **find method on the map (O(logn)**) time it is taking, so only once we calling HasWallet method.
**Note:** Smaller inputs in the wallet you may not see the issue but huge wallet size it will definitely impact the performance.
ACKs for top commit:
achow101:
ACK 55b931934a
w0xlt:
ACK 55b931934a
Tree-SHA512: 4a7fdbcbb4e55bd034e9cf28ab4e7ee3fb1745fc8847adb388c98a19c952a1fb66d7b54f0f28b4c2a75a42473923742b4a99fb26771577183a98e0bcbf87a8ca
Removed duplicate call to GetDescriptorScriptPubKeyMan and
Instead of checking linearly I have used find method so time complexity reduced significantly for GetDescriptorScriptPubKeyMan
after this fix improved performance of importdescriptor part refs #32013.
ConsumeData() will always try to return a name as long as the requested size. It is more useful, and
closer to how `getsockname` would actually behave in reality, to return a random length name
instead.
This was hindering coverage in the PCP fuzz target as the addr len was set to the size of the
sockaddr_in struct and would exhaust all the provided data from the fuzzer.
Thanks to Marco Fleon for suggesting this.
Co-Authored-by: marcofleon <marleo23@proton.me>
The fuzz provider's `ConsumeData` may return less data than necessary
to fill the sockaddr struct and still return success. Fix this to avoid
the caller using uninitialized memory.
This brings the format types closer to the standard library types:
* FormatStringCheck corresponds to std::basic_format_string, with
compile-time checks done via ConstevalFormatString
* RuntimeFormat corresponds to std::runtime_format, with no compile-time
checks done.
Also, it documents where no compile-time checks are done.
552cae243a fuzz: cover `ASMapHealthCheck` in connman target (brunoerg)
33b0f3ae96 fuzz: use `ConsumeNetGroupManager` in connman target (brunoerg)
18c8a0945b fuzz: move `ConsumeNetGroupManager` to util (brunoerg)
fe624631ae fuzz: fuzz `connman` with a non-empty addrman (brunoerg)
0a12cff2a8 fuzz: move `AddrManDeterministic` to util (brunoerg)
Pull request description:
### Motivation
Currently, we fuzz connman with an addrman from `NodeContext`. However,
fuzzing connman with only empty addrman might not be effective, especially
for functions like `GetAddresses` and other ones that plays with addrman. Also,
we do not fuzz connman with ASMap, what would be good for functions that need
`GetGroup`, or even for addrman. Without it, I do not see how effective would be
fuzzing `ASMapHealthCheck`, for example.
### Changes
- Move `AddrManDeterministic` and `ConsumeNetGroupManager` to util.
- Use `ConsumeNetGroupManager` in connman target to construct a netgroupmanager
and use it for `ConnmanTestMsg`.
- Use `AddrManDeterministic` in connman target to create an addrman. It does
not slow down as "filling" the addrman (e.g. with `FillAddrman`).
- Add coverage for `ASMapHealthCheck`.
ACKs for top commit:
maflcko:
review ACK 552cae243a🏀
dergoegge:
Code review ACK 552cae243a
marcofleon:
Code review ACK 552cae243a. Changes match the PR description.
Tree-SHA512: ba861c839602054077e4bf3649763eeb48357cda83ca3ddd32b02a1b61f4e44a0c5070182f001f9bf531d0d64717876279a7de3ddb9de028b343533b89233851
01960c53c7 fuzz: make FuzzedDataProvider usage deterministic (Martin Leitner-Ankerl)
Pull request description:
There exist many usages of `fuzzed_data_provider` where it is evaluated directly in the function call.
Unfortunately, [the order of evaluation of function arguments is unspecified](https://en.cppreference.com/w/cpp/language/eval_order), and a simple example shows that it can differ e.g. between clang++ and g++: https://godbolt.org/z/jooMezWWY
When the evaluation order is not consistent, the same fuzzing/random input will produce different output, which is bad for coverage/reproducibility. This PR fixes all these cases I have found where unspecified evaluation order could be a problem.
Finding these has been manual work; I grepped the sourcecode for these patterns, and looked at each usage individually. So there is a chance I missed some.
* `fuzzed_data_provider`
* `.Consume`
* `>Consume`
* `.rand`
I first discovered this in https://github.com/bitcoin/bitcoin/pull/29013#discussion_r1420236394. Note that there is a possibility that due to this fix the evaluation order is now different in many cases than when the fuzzing corpus has been created. If that is the case, the fuzzing corpus will have worse coverage than before.
Update: In list-initialization the order of evaluation is well defined, so e.g. usages in `initializer_list` or constructors that use `{...}` is ok.
ACKs for top commit:
achow101:
ACK 01960c53c7
vasild:
ACK 01960c53c7
ismaelsadeeq:
ACK 01960c53c7
Tree-SHA512: e56d087f6f4bf79c90b972a5f0c6908d1784b3cfbb8130b6b450d5ca7d116c5a791df506b869a23bce930b2a6977558e1fb5115bb4e061969cc40f568077a1ad
The script building logic performs a quadratic number of copies in the
number of nested wrappers in the miniscript. Limit the number of nested
wrappers to avoid fuzz timeouts.
Thanks to Marco Falke for reporting the fuzz timeouts and providing a
minimal input to reproduce.
This target may call into logic quadratic over the number of
sub-fragments. Limit the number of sub-fragments to keep the runtime
reasonable.
Thanks to Marco Falke for reporting the fuzz timeouts with a minimized
input.
Problem:
If `FuzzedSock::Recv(N, MSG_PEEK)` is called then `N` bytes would be
retrieved from the fuzz provider, saved in `m_peek_data` and returned
to the caller (ok).
If after this `FuzzedSock::Recv(M, 0)` is called where `M < N`
then the first `M` bytes from `m_peek_data` would be returned
to the caller (ok), but the remaining `N - M` bytes in `m_peek_data`
would be discarded/lost (not ok). They must be returned by a subsequent
`Recv()`.
To resolve this, only remove the head `N` bytes from `m_peek_data`.
Currently, when the FuzzedDataProvider of a FuzzedSock runs out of data,
FuzzedSock::Wait and WaitMany will simulate endless waiting as the
requested events are never simulated as occured.
Fix this by simulating event occurence when ConsumeBool() returns false
(e.g. when the data provider runs out).
Co-authored-by: dergoegge <n.goeggi@gmail.com>
FuzzedSock only supports peeking at one byte at a time, which is not
fuzzer friendly when trying to receive long data.
Fix this by supporting peek data of arbitrary length instead of only one
byte.
In the process of doing so, refactor `ConsumeNetAddr()` to generate the
addresses from IPv4, IPv6, Tor, I2P and CJDNS networks in the same way -
by preparing some random stream and deserializing from it. Similar code
was already found in `RandAddr()`.
There exist many usages of `fuzzed_data_provider` where it is evaluated directly in the function call.
Unfortunately, the order of evaluation of function arguments is unspecified. This means it can differ
between compilers/version/optimization levels etc. But when the evaluation order changes, the same
fuzzing input will produce different output, which is bad for coverage/reproducibility.
This PR fixes all these cases where by moving multiple calls to `fuzzed_data_provider` out of the
function arguments.
7df4508369 test: improve sock_tests/move_assignment (Vasil Dimov)
5086a99b84 net: remove Sock default constructor, it's not necessary (Vasil Dimov)
7829272f78 net: remove now unnecessary Sock::Get() (Vasil Dimov)
944b21b70a net: don't check if the socket is valid in ConnectSocketDirectly() (Vasil Dimov)
aeac68d036 net: don't check if the socket is valid in GetBindAddress() (Vasil Dimov)
5ac1a51ee5 i2p: avoid using Sock::Get() for checking for a valid socket (Vasil Dimov)
Pull request description:
_This is a piece of #21878, chopped off to ease review._
Peeking at the underlying socket file descriptor of `Sock` and checkig if it is `INVALID_SOCKET` is bad encapsulation and stands in the way of testing/mocking/fuzzing.
Instead use an empty `unique_ptr` to denote that there is no valid socket where appropriate or outright remove such checks where they are not necessary.
The default constructor `Sock::Sock()` is unnecessary now after recent changes, thus remove it.
ACKs for top commit:
ajtowns:
ACK 7df4508369
jonatack:
ACK 7df4508369
Tree-SHA512: 9742aeeeabe8690530bf74caa6ba296787028c52f4a3342afd193b05dbbb1f6645935c33ba0a5230199a09af01c666bd3c7fb16b48692a0d185356ea59a8ddbf
This also cleans up the addrman (de)serialization code paths to only
allow `Disk` serialization. Some unit tests previously forced a
`Network` serialization, which does not make sense, because Bitcoin Core
in production will always `Disk` serialize.
This cleanup idea was suggested by Pieter Wuille and implemented by Anthony
Towns.
Co-authored-by: Pieter Wuille <pieter@wuille.net>
Co-authored-by: Anthony Towns <aj@erisian.com.au>
