fa6fe43207 net: Clarify that m_addr_local is only set once (MarcoFalke)
Pull request description:
The function is supposed to be only called once when the version msg arrives (a single time). Calling it twice would be an internal logic bug. However, the `LogError` in this function has many issues:
* If the error happens in tests, as is the case for the buggy fuzz test, it will go unnoticed
* It is dead code, unless a bug is introduced to execute it
Fix all issues by using `Assume(!m_addr_local.IsValid())` instead. Idea taken from https://github.com/bitcoin/bitcoin/pull/30364#discussion_r1680530382
ACKs for top commit:
achow101:
ACK fa6fe43207
mzumsande:
utACK fa6fe43207
glozow:
ACK fa6fe43207
Tree-SHA512: 8c1e8c524768f4f36cc50110ae54ee423e057a963ff78f736f3bf92df1ce5af28e3e0149153780897944e1d5c22ddbca9dac9865d9f4d44afffa152bc8559405
It is expensive to construct, and only one test uses it.
Fix both issues by disallowing the construction and moving it to the
single test that uses it.
Blank legacy wallets do not have active SPKM. They can
only be detected by checking the descriptors' flag or
the db format.
This enables the migration of blank legacy wallets in
the GUI.
The following bitcoind parameters / RPC calls missed the "testnet4"
network string:
- `-chain=` parameter
- `getblockchaininfo` RPC, "chain" result
- `getmininginfo` RPC, "chain" result
5215c925d1 Compare ASMaps with respect to specific addresses (virtu)
Pull request description:
Right now, we have no way to quantify the "degradation" of an ASMap over time in the context of Bitcoin's P2P network in a meaningful way. However, such data would be useful for:
1. Making sure the minimum shelf life of ASMaps is compatible with the release cycle (we wouldn't want to start shipping ASMaps with releases before making sure ASMaps typically do not become obsolete before the time of the next release)
2. Node operators eager to keep their ASMaps up-to-date between releases.
While `asmap-tool.py` has a `diff` command to perform a prefix-based comparison of two ASMaps, it is hard to reason about whether an old ASMap still is "good enough" or should be replaced with a newer one based on a prefix-based diff such as the following:
```shell
$ ./asmap-tool.py diff 1704463200_asmap.dat 1710770400_asmap.dat
[...]
# 2c0f:fc98::/32 was AS37282
# 2c0f:fcb8::/32 was AS37323
2c0f:ff18::/32 AS37044 # was unassigned
2c0f:ff98::/32 AS37113 # was unassigned
2c0f:ffa0::/32 AS37273 # was unassigned
# 76082350 (2^26.18) IPv4 addresses changed; 834271985742505274886878979424260 (2^109.36) IPv6 addresses changed
```
One option for a more Bitcoin-centric ASMap comparison comprises comparing ASNs for the addresses of Bitcoin nodes and reporting on the number/share of addresses of nodes with disagreeing ASNs. By applying this approach to a node's set of known peers, a node operator can estimate how many of the node's peers are mapped to out-of-date AS when using the currently deployed and an up-to-date ASMap as input.
This PR adds this functionality to `asmap-tool.py` by introducing a `diff_addrs` subcommand. In addition to two ASMaps, the subcommand reads addresses from a (`getnodeaddresses`-compatible) file, and computes statistics for those addresses:
```bash
$ ./asmap-tool.py diff_addrs 1704463200_asmap.dat 1710770400_asmap.dat <(bitcoin-cli getnodeaddresses 0)
275 address(es) reassigned from unassigned to AS51167
84 address(es) reassigned from AS198949 to AS15557
66 address(es) reassigned from AS45758 to AS45629
33 address(es) reassigned from AS174 to AS212238
[...]
1 address(es) reassigned from unassigned to AS399619
Summary: 919 (1.67%) of 54,901 addresses were reassigned.
```
When the `-s / --show-addresses` flag is used, addresses subject to reassignment are included in the output.
ACKs for top commit:
fjahr:
tACK 5215c925d1
achow101:
ACK 5215c925d1
brunoerg:
reACK 5215c925d1
Tree-SHA512: ebcf47754bce92794fad9f4c3bfc1c5e9daf077db5975f444c5135092eb6a26ecaa1eca6748a03ae0c87d9e45532426966fe8f3c17249b17f9dcad490d6dd3bf
86b38529d5 qa: a fuzz target for the block index database (Antoine Poinsot)
Pull request description:
This introduces a small fuzz target for `CBlockTreeDB` which asserts a few invariants by using an in-memory LevelDb.
ACKs for top commit:
achow101:
ACK 86b38529d5
TheCharlatan:
Re-ACK 86b38529d5
maflcko:
review ACK 86b38529d5🥒
brunoerg:
utACK 86b38529d5
Tree-SHA512: ab75b4ae1c7e0a4b15f8a6ceffdf509fbc79833e6ea073ecef68558d53b83663d1b30362aaa2d77c22b8890a572f5b1d4b1c5abbca483c8c8f9b1fb5b276a59a
8fee5355ee guix: fix suggested fake date for openssl -1.1.1l (Sjors Provoost)
Pull request description:
Using `2020-10-01` as the fake timestamp will cause many test failures with `/gnu/store/bfirgq65ndhf63nn4q6vlkbha9zd931q-openssl-1.1.1l.drv`. I didn't investigate why, but I guess because it's _before_ the test certificates were created. They expired in June 2022. I tried a month before that, which worked.
Also fixes layout of instructions.
ACKs for top commit:
achow101:
ACK 8fee5355ee
maflcko:
review ACK 8fee5355ee
Tree-SHA512: df5dd3aa961e25bd57d0b8b73daeb3ec76856b06e35277f24b6b19be81774512228f75e2b779afa8ea92fcc39beb869f43e0c57fba19ad16a82812e7c0bea38b
77ff0ec1f1 contrib: support reading XORed blocks in linearize-data.py script (Sebastian Falbesoner)
Pull request description:
This PR is a small follow-up for #28052, adding support for the block linearization script to handle XORed blocksdir *.dat files. Note that if no xor.dat file exists, the XOR pattern is set to all-zeros, in order to still support blockdirs that have been created with versions earlier than 28.x.
Partly fixes issue #30599.
ACKs for top commit:
achow101:
ACK 77ff0ec1f1
tdb3:
ACK 77ff0ec1f1
hodlinator:
ACK 77ff0ec1f1
Tree-SHA512: 011eb02e2411de373cbbf4b26db4640fc693a20be8c2430529fba6e36a3a3abfdfdc3b005d330f9ec2846bfad9bfbf34231c574ba99289ef37dd51a68e6e7f3d
401cc4ec70 fuzz: improve scriptpubkeyman target (brunoerg)
Pull request description:
Fixes#30541
This PR aims to improve `scriptpubkeyman` target to avoid timeouts. The input provided in #30541 takes too much time to run because it basically calls only `MarkUnusedAddresses` (300 times * number of spks). The following changes were made to improve it:
- Reduce keypool size.
- When calling `MarkUnusedAddresses`, do it with one of the spks per iteration.
- Remove the specific `AddDescriptorKey` call since it is already covered with `AddWalletDescriptor`.
- Limit number of iterations to a reasonable value.
ACKs for top commit:
maflcko:
lgtm ACK 401cc4ec70
achow101:
ACK 401cc4ec70
Tree-SHA512: 941812bc6d991dd03675a2974ce1b839494ca7f6e6d8a22c689d4bf4fed2dac5491246998f19cb15dbff516fdd8eeda27e7628c3206d45f57dc292bc05624a5c
15aa7d0236 gui, qt: brintToFront workaround for Wayland (pablomartin4btc)
Pull request description:
There are known issues around handling windows focus in `Wayland` ([this one specific](https://bugs.kde.org/show_bug.cgi?id=462574) in KDE but also in [gnome](https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/730)).
The idea is that the workaround will be executed if `bitcoin-qt` is running using `Wayland` platform (e.g.: `QT_QPA_PLATFORM=wayland ./src/qt/bitcoin-qt -regtest`), since the workaround behaviour looks like re-opening the window again (which I tried to fix by moving the window to the original position and/ or re-setting the original geometry without success) while in `X11` (not sure in Mac) the current `GUIUtil::brintToFront` actually sets the focus to the desired window, keeping its original position as expected, and I didn't want to change that (`X11` behaviour).
The solution was [initially discussed](https://github.com/bitcoin-core/gui/pull/817#issuecomment-2256158902) with hebasto in #817.
ACKs for top commit:
hebasto:
ACK 15aa7d0236.
Tree-SHA512: 141d6cc4a618026e551627b9f4cc284285980db02a54a7b19c7de91e8c5adccf0c1d67380625146b5413e58c59f39c9e944ed5ba68cb8644f67647518918b6f7
6b2dcba076 wallet: List sqlite wallets with empty string name (Ava Chow)
3ddbdd1815 wallet: Ignore .bak files when listing wallet files (Ava Chow)
Pull request description:
When the default wallet is migrated, we do not rename the wallet so we end up having a descriptor wallet with the empty string as its name and the wallet.dat file in the root of the walletdir. This is supposed to be an unsupported configuration and there is no other way to achieve this (other than file copying), but the wallet loading code does not disallow loading such wallets. However `listwalletdir` does not currently list the default wallet if it is sqlite. This is confusing to users, so change `listwalletdir` to include these wallets.
Additionally, the migration of the default wallet, and of any plain wallet files in the walletdir, produces a backup file in the walletdir itself. Since these backups are a BDB file, `listwalletdir` will detect them as being another wallet that we could open, but this is erroneous and could lead to confusion and potentially funds loss if both the backup and the migrated wallet are in use simultaneously. To reduce the likelihood of this issue, don't list these wallets in `listwalletdir`.
***
Possibly we could have more stringent checks on loading to resolve these issues, but I'm concerned that that will just confuse users and gratuitously break things that already worked.
Since the original intent was to disallow default wallets for sqlite/descriptors, a possible alternative would be to prevent people from loading such wallets and change migration to rename those wallets. However, given that this behavior with migrating default wallets has existed since default wallet migration was fixed, I think that making such a change would be confusing and break things for no good reason. Although perhaps we should still do the renaming.
For the backups, we could also change loading to refuse to load any wallet named with `.bak` (or `.legacy.bak`) as such wallets can still be loaded by giving the path to them directly, which some users may do to "restore" the backup. However restricting what can be loaded based on filename seems a little heavyhanded. It wouldn't be funds loss though since the correct way to restore the backup is with `restorewallet`.
ACKs for top commit:
fjahr:
Code review ACK 6b2dcba076
furszy:
Code ACK 6b2dcba076
glozow:
ACK 6b2dcba076
Tree-SHA512: 0b033f6ed55830f8a054afea3fb2cf1fa82a94040053ebfaf123bda36c99f45d3f01a2aec4ed02fed9c61bb3d320b047ed892d7f6644b5a356a7bc5974b10cff
055bc05792 policy/feerate.h: avoid constraint self-dependency (Matt Whitlock)
138f867156 add missing #include <cstdint> for GCC 15 (Matt Whitlock)
Pull request description:
#30612 with changes made.
GCC 15 introduces three build failures:
* Two are related to missing includes. You can't use `uint16_t` et al. without including `<cstdint>`.
* The third is harder to understand but easy to fix. GCC changed something about the way templates are instantiated when checking type constraints, and now there is a dependency loop while checking `std::optional<CFeeRate>`. This manifests as the following compile-time mess:
```
In file included from /usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/format:48,
from /usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/bits/chrono_io.h:39,
from /usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/chrono:3362,
from ./util/time.h:9,
from ./primitives/block.h:12,
from ./blockencodings.h:8,
from blockencodings.cpp:5:
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/type_traits: In substitution of 'template<class _Up> requires !(is_same_v<std::optional<_Tp>, typename std::remove_cvref<_It2>::type>) && (is_constructible_v<_Tp, const _Up&>) && (__construct_from_contained_value<_Up, typename std::remove_cv< <template-parameter-1-1> >::type>) constexpr std::optional<CFeeRate>::optional(const std::optional<_Tp>&) [with _Up = CFeeRate]':
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/type_traits:1140:25: required by substitution of 'template<class _Tp, class ... _Args> using std::__is_constructible_impl = std::__bool_constant<__is_constructible(_Tp, _Args ...)> [with _Tp = CFeeRate; _Args = {std::optional<CFeeRate>&}]'
1140 | = __bool_constant<__is_constructible(_Tp, _Args...)>;
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/type_traits:1145:12: required from 'struct std::is_constructible<CFeeRate, std::optional<CFeeRate>&>'
1145 | struct is_constructible
| ^~~~~~~~~~~~~~~~
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/type_traits:178:35: required by substitution of 'template<class ... _Bn> std::__detail::__first_t<std::integral_constant<bool, false>, typename std::enable_if<(!(bool)(_Bn::value)), void>::type ...> std::__detail::__or_fn(int) [with _Bn = {std::is_constructible<CFeeRate, std::optional<CFeeRate>&>, std::is_convertible<std::optional<CFeeRate>&, CFeeRate>, std::is_constructible<CFeeRate, std::optional<CFeeRate> >, std::is_convertible<std::optional<CFeeRate>, CFeeRate>, std::is_constructible<CFeeRate, const std::optional<CFeeRate>&>, std::is_convertible<const std::optional<CFeeRate>&, CFeeRate>, std::is_constructible<CFeeRate, const std::optional<CFeeRate> >, std::is_convertible<const std::optional<CFeeRate>, CFeeRate>}]'
178 | __enable_if_t<!bool(_Bn::value)>...>;
| ^~~~~
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/type_traits:196:41: required from 'struct std::__or_<std::is_constructible<CFeeRate, std::optional<CFeeRate>&>, std::is_convertible<std::optional<CFeeRate>&, CFeeRate>, std::is_constructible<CFeeRate, std::optional<CFeeRate> >, std::is_convertible<std::optional<CFeeRate>, CFeeRate>, std::is_constructible<CFeeRate, const std::optional<CFeeRate>&>, std::is_convertible<const std::optional<CFeeRate>&, CFeeRate>, std::is_constructible<CFeeRate, const std::optional<CFeeRate> >, std::is_convertible<const std::optional<CFeeRate>, CFeeRate> >'
196 | : decltype(__detail::__or_fn<_Bn...>(0))
| ~~~~~~~~~~~~~~~~~~~~~~~~~^~~
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/optional:824:45: required from 'constexpr const bool std::optional<CFeeRate>::__construct_from_contained_value<CFeeRate, CFeeRate>'
824 | = !__converts_from_optional<_Tp, _From>::value;
| ^~~~~
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/optional:884:7: required by substitution of 'template<class _Up> requires !(is_same_v<std::optional<_Tp>, typename std::remove_cvref<_It2>::type>) && (is_constructible_v<_Tp, const _Up&>) && (__construct_from_contained_value<_Up, typename std::remove_cv< <template-parameter-1-1> >::type>) constexpr std::optional<CFeeRate>::optional(const std::optional<_Tp>&) [with _Up = CFeeRate]'
884 | && __construct_from_contained_value<_Up>
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
./validation.h:164:41: required from here
164 | return MempoolAcceptResult(state);
| ^
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/optional:886:2: required by the constraints of 'template<class _Tp> template<class _Up> requires !(is_same_v<std::optional<_Tp>, typename std::remove_cvref<_It2>::type>) && (is_constructible_v<_Tp, const _Up&>) && (__construct_from_contained_value<_Up, typename std::remove_cv< <template-parameter-1-1> >::type>) constexpr std::optional<_Tp>::optional(const std::optional<_From>&)'
/usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/optional:884:14: error: satisfaction of atomic constraint '__construct_from_contained_value<_Up, typename std::remove_cv< <template-parameter-1-1> >::type> [with _Tp = _Tp; _Up = _Up]' depends on itself
884 | && __construct_from_contained_value<_Up>
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
```
It is easiest to solve this by changing the `static_assert` in the explicit `CFeeRate` constructor to a SFINAE by using a type constraint on the function template parameter.
We already [downstreamed](https://github.com/gentoo/gentoo/pull/38015) these fixes in Gentoo.
ACKs for top commit:
stickies-v:
ACK 055bc05792
Tree-SHA512: ce9cb27bcd9b0f4bbc80951e45cf7127112dcb7f9937bcb0167b362026d35beecb1255354746de0aac82e03c41eaccbe26acbfe0ddff2ee1e5a8634673f4f4ba
4c2cc63d84 Revert "depends: Fetch miniupnpc sources from an alternative website" (sunerok)
Pull request description:
This reverts commit 21b8a14d37.
started fresh fanquake
ACKs for top commit:
fanquake:
ACK 4c2cc63d84 - please don't @ mention anyone in your pull request descriptions, otherwise they are likely to just get spammed with notifications.
Tree-SHA512: ab3e6cc65e4531b6c129a6bc663ebc9859e0de1d4b745b29834344cc8aa83ea182b0a5d1fae14e609a85bbbe24c34f42502fb25b53a15b38dd63396913ea4682
16d8261181 doc: Drop no longer needed workaround for WSL (Hennadii Stepanov)
Pull request description:
This PR effectively reverts commit 4f890ba6bc from https://github.com/bitcoin/bitcoin/pull/11437, which fixed some build issues on WSL seven years ago.
Testing the current master branch @ 31a3ff5515 on Windows 11 + WSL using Ubuntu 24.04 or Debian images, I noticed that the workaround is no longer required. Moreover, it doesn't affect the build process at all, which means the hashes of the built packages in depends remain the same and the `configure` log in the main build system remains the same as well.
ACKs for top commit:
sipsorcery:
utACK 16d8261181.
Tree-SHA512: 703a2ac4647125c91aad47131e2723fd05af30b0cfae5677a26d3e89a77c2779e0197584208f3b378ed64dd7305512d9064fd073ec06517f86e9905af4ec8838
fa5755b0a8 doc: rpc: Use "output script" consistently (2/2) (MarcoFalke)
Pull request description:
Small follow-up to https://github.com/bitcoin/bitcoin/pull/30408 to fixup the RPCs that were forgotten.
ACKs for top commit:
theStack:
lgtm ACK fa5755b0a8
Tree-SHA512: f1fc0aabb59017da216d6fe0f08a2274336d04db332ad6ce3d9608cd6f03667be1c76423f24a489ac8e7d536011a129dca752ab64b4621b7bc1d4d53f68602e4
49d569cb1f p2p: For assumeutxo, download snapshot chain before background chain (Martin Zumsande)
7a885518d5 p2p: Restrict downloading of blocks for snapshot chain (Martin Zumsande)
Pull request description:
After loading a snapshot, `pindexLastCommonBlock` is usually already set to some block for existing peers. That means we'd continue syncing the background chain from those peers instead of prioritising the snapshot chain, which defeats the purpose of doing assumeutxo in the first place. Only existing peers are affected by this bug.
ACKs for top commit:
fjahr:
re-ACK 49d569cb1f
achow101:
ACK 49d569cb1f
Sjors:
tACK 49d569cb1f
Tree-SHA512: 0eaebe1c29a8510d5ced57e14c09b128ccb34b491692815291df68bf12e2a15b52b1e7bf8d9f34808904e7f7bc20f70b0ad0f7e14df93bbdf456bd12cc02a5d2
00618e8745 assumeutxo: Drop block height from metadata (Fabian Jahr)
Pull request description:
Fixes https://github.com/bitcoin/bitcoin/issues/30514 which has more context and shows how the issue can be reproduced. Since the value in question is removed, there is no test to add to reproduce anything.
This is an alternative approach to #30516 with much of the [code being suggested there](https://github.com/bitcoin/bitcoin/pull/30516#discussion_r1689146902).
ACKs for top commit:
maflcko:
re-ACK 00618e8745🎌
achow101:
ACK 00618e8745
theStack:
Code-review ACK 00618e8745
ismaelsadeeq:
Re-ACK 00618e8745
mzumsande:
ACK 00618e8745
Tree-SHA512: db9575247bae838ad7742a27a216faaf55bb11e022f9afdd05752bb09bbf9614717d0ad64304ff5722a16bf41d8dea888af544e4ae26dcaa528c1add0269a4a8
Although it is not explicitly possible to create a default wallet with
descriptors, it is possible to migrate a default wallet and have it end
up being a default wallet with descriptors. These wallets should be
listed by ListDatabases so that it appears in wallet directory listings
to avoid user confusion.
Migration creates backup files in the wallet directory with .bak as the
extension. This pollutes the output of listwalletdir with backup files
that most users should not need to care about.
92c1d7d1f8 validation: Use MAX_TIMEWARP constant as testnet4 timewarp defense delta (Fabian Jahr)
4b2fad502e doc: Add release notes for 29775 (Fabian Jahr)
f7cc97313b doc: Align deprecation warnings (Fabian Jahr)
1163b08378 chainparams: Add initial minimum chain work for Testnet4 (Fabian Jahr)
Pull request description:
This completes follow-ups left open in #29775.
- Adds release notes
- Addresses the [misalignment](https://github.com/bitcoin/bitcoin/pull/29775#discussion_r1706982102) in deprecation warnings and hints at the intention to remove support for Testnet3.
- Adds initial minimum chainwork for Testnet4.
- Use the `MAX_TIMEWARP` constant as the timewarp defense delta, equal to `MAX_FUTURE_BLOCK_TIME`.
ACKs for top commit:
Sjors:
ACK 92c1d7d1f8
achow101:
ACK 92c1d7d1f8
tdb3:
re ACK 92c1d7d1f8
Tree-SHA512: 7ebdac7809f96231f75ca62706af59cd1ed27f713a4c7be5e2ad69fae95832b146b3ea23c712fb03b412da1deda7e8a5dae55bb2bbd2dcfd9f926e85c2a72666
2925bd537c refactor: use c++20 std::views::reverse instead of reverse_iterator.h (stickies-v)
Pull request description:
C++20 introduces [`std::ranges::views::reverse`](https://en.cppreference.com/w/cpp/ranges/reverse_view), which allows us to drop our own `reverse_iterator.h` implementation and also makes it easier to chain views (even though I think we currently don't use this).
ACKs for top commit:
achow101:
ACK 2925bd537c
maflcko:
ACK 2925bd537c🎷
Tree-SHA512: 567666ec44af5d1beb7a271836bcc89c4c577abc77f522fcc18bc6d4de516ae9b0df766d0bfa6dd217569e6878331c2aee1d9815620860375e3510dad7fed476
Enhanced efficiency and readability of CCoinsViewCache::FetchCoin by replacing separate find() and emplace() calls with a single try_emplace(), reducing map lookups and potential insertions.
589db872e1 validation: don't erase coins cache on prune flushes (Andrew Toth)
0e8918755f Add linked-list test to CCoinsViewCache::SanityCheck (Pieter Wuille)
05cf4e1875 coins: move Sync logic to CoinsViewCacheCursor (Andrew Toth)
7825b8b9ae coins: pass linked list of flagged entries to BatchWrite (Andrew Toth)
a14edada8a test: add cache entry linked list tests (Andrew Toth)
24ce37cb86 coins: track flagged cache entries in linked list (Andrew Toth)
58b7ed156d coins: call ClearFlags in CCoinsCacheEntry destructor (Andrew Toth)
8bd3959fea refactor: require self and sentinel parameters for AddFlags (Andrew Toth)
75f36d241d refactor: add CoinsCachePair alias (Andrew Toth)
f08faeade2 refactor: move flags to private uint8_t and rename to m_flags (Andrew Toth)
4e4fb4cbab refactor: disallow setting flags in CCoinsCacheEntry constructors (Andrew Toth)
8737c0cefa refactor: encapsulate flags setting with AddFlags and ClearFlags (Andrew Toth)
9715d3bf1e refactor: encapsulate flags get access for all other checks (Andrew Toth)
df34a94e57 refactor: encapsulate flags access for dirty and fresh checks (Andrew Toth)
Pull request description:
Since https://github.com/bitcoin/bitcoin/pull/17487 we no longer need to clear the coins cache when syncing to disk. A warm coins cache significantly speeds up block connection, and only needs to be fully flushed when nearing the `dbcache` limit.
For frequent pruning flushes there's no need to empty the cache and kill connect block speed. However, simply using `Sync` in place of `Flush` actually slows down a pruned full IBD with a high `dbcache` value. This is because as the cache grows, sync takes longer since every coin in the cache is scanned to check if it's dirty. For frequent prune flushes and a large cache this constant scanning starts to really slow IBD down, and just emptying the cache on every prune becomes faster.
To fix this, we can add two pointers to each cache entry and construct a doubly linked list of dirty entries. We can then only iterate through all dirty entries on each `Sync`, and simply clear the pointers after.
With this approach a full IBD with `dbcache=16384` and `prune=550` was 32% faster than master. For default `dbcache=450` speedup was ~9%. All benchmarks were run with `stopatheight=800000`.
| | prune | dbcache | time | max RSS | speedup |
|-----------:|----------:|------------:|--------:|-------------:|--------------:|
| master | 550 | 16384 | 8:52:57 | 2,417,464k | - |
| branch | 550 | 16384 | 6:01:00 | 16,216,736k | 32% |
| branch | 550 | 450 | 8:05:08 | 2,818,072k | 8.8% |
| master | 10000 | 5000 | 8:19:59 | 2,962,752k | - |
| branch | 10000 | 5000| 5:56:39 | 6,179,764k | 28.8% |
| master | 0 | 16384 | 4:51:53 | 14,726,408k | - |
| branch | 0 | 16384 | 4:43:11 | 16,526,348k | 2.7% |
| master | 0 | 450 | 7:08:07 | 3,005,892k | - |
| branch | 0 | 450 | 6:57:24 | 3,013,556k |2.6%|
While the 2 pointers add memory to each cache entry, it did not slow down IBD. For non-pruned IBD results were similar for this branch and master. When I performed the initial IBD, the full UTXO set could be held in memory when using the max `dbcache` value. For non-pruned IBD with max `dbcache` to tip ended up using 12% more memory, but it was also 2.7% faster somehow. For smaller `dbcache` values the `dbcache` limit is respected so does not consume more memory, and the potentially more frequent flushes were not significant enough to cause any slowdown.
For reviewers, the commits in order do the following:
First 4 commits encapsulate all accesses to `flags` on cache entries, and then the 5th makes `flags` private.
Commits `refactor: add CoinsCachePair alias` to `coins: call ClearFlags in CCoinsCacheEntry destructor` create the linked list head nodes and cache entry self references and pass them into `AddFlags`.
Commit `coins: track flagged cache entries in linked list` actually adds the entries into a linked list when they are flagged DIRTY or FRESH and removes them from the linked list when they are destroyed or the flags are cleared manually. However, the linked list is not yet used anywhere.
Commit `test: add cache entry linked list tests` adds unit tests for the linked list.
Commit `coins: pass linked list of flagged entries to BatchWrite` uses the linked list to iterate through DIRTY entries instead of using the entire coins cache.
Commit `validation: don't erase coins cache on prune flushes` uses `Sync` instead of `Flush` for pruning flushes, so the cache is no longer cleared.
Inspired by [this comment](https://github.com/bitcoin/bitcoin/pull/15265#issuecomment-457720636).
Fixes https://github.com/bitcoin/bitcoin/issues/11315.
ACKs for top commit:
paplorinc:
ACK 589db872e1
sipa:
reACK 589db872e1
achow101:
ACK 589db872e1
mzumsande:
re-ACK 589db872e1
Tree-SHA512: 23b2bc01c83edacb5b39aa60bb0b766de9a74ce17f0c59bf13b97b4328a7b758ad9aff6581c3ca88e2973f7658380651530d497444f48d6e22ea0bfc51cc921d
