bf95976061 doc: add note about snapshot chainstate init (James O'Beirne)
e4d7995286 test: add testcases for snapshot initialization (James O'Beirne)
cced4e7336 test: move-only-ish: factor out LoadVerifyActivateChainstate() (James O'Beirne)
51fc9241c0 test: allow on-disk coins and block tree dbs in tests (James O'Beirne)
3c361391b8 test: add reset_chainstate parameter for snapshot unittests (James O'Beirne)
00b357c215 validation: add ResetChainstates() (James O'Beirne)
3a29dfbfb2 move-only: test: make snapshot chainstate setup reusable (James O'Beirne)
8153bd9247 blockmanager: avoid undefined behavior during FlushBlockFile (James O'Beirne)
ad67ff377c validation: remove snapshot datadirs upon validation failure (James O'Beirne)
34d1590331 add utilities for deleting on-disk leveldb data (James O'Beirne)
252abd1e8b init: add utxo snapshot detection (James O'Beirne)
f9f1735f13 validation: rename snapshot chainstate dir (James O'Beirne)
d14bebf100 db: add StoragePath to CDBWrapper/CCoinsViewDB (James O'Beirne)
Pull request description:
This is part of the [assumeutxo project](https://github.com/bitcoin/bitcoin/projects/11) (parent PR: https://github.com/bitcoin/bitcoin/pull/15606)
---
Half of the replacement for #24232. The original PR grew larger than expected throughout the review process.
This change adds the ability to initialize a snapshot-based chainstate during init if one is detected on disk. This is of course unused as of now (aside from in unittests) given that we haven't yet enabled actually loading snapshots.
Don't be scared! There are some big move-only commits in here.
Accompanying changes include:
- moving the snapshot coinsdb directory from being called `chainstate_[base blockhash]` to `chainstate_snapshot`, since we only support one snapshot in use at a time. This simplifies some logic, but it necessitates writing that base blockhash out to a file within the coinsdb dir. See [discussion here](https://github.com/bitcoin/bitcoin/pull/24232#discussion_r832762880).
- adding a simple fix in `FlushBlockFile()` that avoids a crash when attemping to flush to disk before `LoadBlockIndexDB()` is called, which happens when calling `MaybeRebalanceCaches()` during multiple chainstate init.
- improving the unittest to allow testing with on-disk chainstates - necessary to test a simulated restart and re-initialization.
ACKs for top commit:
naumenkogs:
utACK bf95976061
ariard:
Code Review ACK bf9597606
ryanofsky:
Code review ACK bf95976061. Changes since last review: rebasing, switching from CAutoFile to AutoFile, adding comments, switching from BOOST_CHECK to Assert in test util, using chainman.GetMutex() in tests, destroying one ChainstateManager before creating a new one in tests
fjahr:
utACK bf95976061
aureleoules:
ACK bf95976061
Tree-SHA512: 15ae75caf19f8d12a12d2647c52897904d27b265a7af6b4ae7b858592eeadb8f9da6c2394b6baebec90adc28742c053e3eb506119577dae7c1e722ebb3b7bcc0
bcb0cacac2 reindex, log, test: fixes #21379 (mruddy)
Pull request description:
Fixes #21379.
The blocks/blk?????.dat files are mutated and become increasingly malformed, or corrupt, as a result of running the re-indexing process.
The mutations occur after the re-indexing process has finished, as new blocks are appended, but are a result of a re-indexing process miscalculation that lingers in the block manager's `m_blockfile_info` `nSize` data until node restart.
These additions to the blk files are non-fatal, but also not desirable.
That is, this is a form of data corruption that the reading code is lenient enough to process (it skips the extra bytes), but it adds some scary looking log messages as it encounters them.
The summary of the problem is that the re-index process double counts the size of the serialization header (magic message start bytes [4 bytes] + length [4 bytes] = 8 bytes) while calculating the blk data file size (both values already account for the serialization header's size, hence why it is over accounted).
This bug manifests itself in a few different ways, after re-indexing, when a new block from a peer is processed:
1. If the new block will not fit into the last blk file processed while re-indexing, while remaining under the 128MiB limit, then the blk file is flushed to disk and truncated to a size that is 8 greater than it should be. The truncation adds zero bytes (see `FlatFileSeq::Flush` and `TruncateFile`).
1. If the last blk file processed while re-indexing has logical space for the new block under the 128 MiB limit:
1. If the blk file was not already large enough to hold the new block, then the zeros are, in effect, added by `fseek` when the file is opened for writing. Eight zero bytes are added to the end of the last blk file just before the new block is written. This happens because the write offset is 8 too great due to the miscalculation. The result is 8 zero bytes between the end of the last block and the beginning of the next block's magic + length + block.
1. If the blk file was already large enough to hold the new block, then the current existing file contents remain in the 8 byte gap between the end of the last block and the beginning of the next block's magic + length + block. Commonly, when this occcurs, it is due to the blk file containing blocks that are not connected to the block tree during reindex and are thus left behind by the reindex process and later overwritten when new blocks are added. The orphaned blocks can be valid blocks, but due to the nature of concurrent block download, the parent may not have been retrieved and written by the time the node was previously shutdown.
ACKs for top commit:
LarryRuane:
tested code-review ACK bcb0cacac2
ryanofsky:
Code review ACK bcb0cacac2. This is a disturbing bug with an easy fix which seems well-worth merging.
mzumsande:
ACK bcb0cacac2 (reviewed code and did some testing, I agree that it fixes the bug).
w0xlt:
tACK bcb0cacac2
Tree-SHA512: acc97927ea712916506772550451136b0f1e5404e92df24cc05e405bb09eb6fe7c3011af3dd34a7723c3db17fda657ae85fa314387e43833791e9169c0febe51
If we call FlushBlockFile() without having intitialized the block index
with LoadBlockIndexDB(), we may be indexing into an empty vector.
Specifically this is an issue when we call MaybeRebalanceCaches() during
chainstate init before the block index has been loaded, which calls
FlushBlockFile().
Also add an assert to avoid undefined behavior.
Co-authored-by: Russell Yanofsky <russ@yanofsky.org>
dd065dae9f refactor: Make mapBlocksUnknownParent local, and rename it (Hennadii Stepanov)
Pull request description:
This PR is a second attempt at #19594. This PR has two motivations:
- Improve code hygiene by eliminating a global variable, `mapBlocksUnknownParent`
- Fix fuzz test OOM when running too long ([see #19594 comment](https://github.com/bitcoin/bitcoin/pull/19594#issuecomment-958801638))
A minor added advantage is to release `mapBlocksUnknownParent` memory when the reindexing phase is done. The current situation is somewhat similar to a memory leak because this map exists unused for the remaining lifetime of the process. It's true that this map should be empty of data elements after use, but its internal metadata (indexing structures, etc.) can have non-trivial size because there can be many thousands of simultaneous elements in this map.
This PR helps our efforts to reduce the use of global variables. This variable isn't just global, it's hidden inside a function (it looks like a local variable but has the `static` attribute).
This global variable exists because the `-reindex` processing code calls `LoadExternalBlockFile()` multiple times (once for each block file), but that function must preserve some state between calls (the `mapBlocksUnknownParent` map). This PR fixes this by allocating this map as a local variable in the caller's scope and passing it in on each call. When reindexing completes, the map goes out of scope and is deallocated.
I tested this manually by reindexing on mainnet and signet. Also, the existing `feature_reindex.py` functional test passes.
ACKs for top commit:
mzumsande:
re-ACK dd065dae9f
theStack:
re-ACK dd065dae9f
shaavan:
reACK dd065dae9f
Tree-SHA512: 9cd20e44d2fa1096dd405bc107bc065ea8f904f5b3f63080341b08d8cf57b790df565f58815c2f331377d044d5306708b4bf6bdfc5ef8d0ed85d8e97d744732c
Also:
1. Have CChainState::LoadMempool and ::ThreadImport take in paths and
pass it through untouched to LoadMempool.
2. Make LoadMempool exit early if the load_path is empty.
3. Adjust the call to ::ThreadImport in ::AppInitMain to correctly pass
in an empty path if mempool persistence is disabled.
This fixes a blk file size calculation made during reindex that results in increased blk file malformity.
The fix is to avoid double counting the size of the serialization header during reindex.
This adds a unit test to reproduce the bug before the fix and to ensure that it does not recur.
These changes include a log message change also so as to not be as alarming. This is a common and recoverable
data corruption. These messages can now be filtered by the debug log reindex category.
In previous commits in this patchset, we've made sure that every
Unload/UnloadBlockIndex member function resets its own members, and does
not reach out to globals.
This means that their corresponding classes' default destructors can now
replace them, and do an even more thorough job without the need to be
updated for every new member variable.
Therefore, we can remove them, and also remove UnloadBlockIndex since
that's not used anymore.
Unfortunately, chainstatemanager_loadblockindex relies on
CChainState::UnloadBlockIndex, so that needs to stay for now.
This change also introduces an aditional buffer of 10 blocks (PRUNE_LOCK_BUFFER) that will not be pruned before the best block.
Co-authored-by: Luke Dashjr <luke-jr+git@utopios.org>
-----
Code Reviewer Notes
Call graph of relevant functions:
UnloadBlockIndex() <-- Moved from
calls ChainstateManager::Unload()
which calls BlockManager::Unload() <-- Moved to
So calling UnloadBlockIndex() would still run this moved code. The code
will also now run when ~BlockManager gets called, which makes sense.
Now BlockManager::LoadBlockIndex() will ACTUALLY only load BlockMan
members.
[META] In a later commit, pindexBestHeader will be moved to ChainMan as
a member
-----
Code Reviewer Notes
Call graph of relevant functions:
ChainstateManager::LoadBlockIndex() <-- Moved to
calls BlockManager::LoadBlockIndexDB()
which calls BlockManager::LoadBlockIndex() <-- Moved from
There is only one call to each of inner functions, meaning that no
behavior is changing.
This commit effectively splits the "load block index itself" logic from
"derive Chainstate variables from loaded block index" logic.
This means that BlockManager::LoadBlockIndex{,DB} will only load what's
relevant to the BlockManager.
I strongly recommend reviewing with the following git-diff flags:
--color-moved=dimmed_zebra --color-moved-ws=allow-indentation-change
Instead of having CBlockIndex's live on the heap, which requires manual
memory management, have them be owned by m_block_index. This means that
they will live and die with BlockManager.
A change to BlockManager::LookupBlockIndex:
- Previously, it was a const member function returning a non-const CBlockIndex*
- Now, there's are const and non-const versions of
BlockManager::LookupBlockIndex returning a CBlockIndex with the same
const-ness as the member function:
(e.g. const CBlockIndex* LookupBlockIndex(...) const)
See next commit for some weirdness that this eliminates.
The range based for-loops are modernize (using auto + destructuring) in
a future commit.
fa5d2e678c Remove unused char serialize (MarcoFalke)
fa24493d63 Use spans of std::byte in serialize (MarcoFalke)
fa65bbf217 span: Add BytePtr helper (MarcoFalke)
Pull request description:
This changes the serialize code (`.read()` and `.write()` functions) to take a `Span` instead of a pointer and size. This is a breaking change for the serialize interface, so at no additional cost we can also switch to `std::byte` (instead of using `char`).
The benefits of using `Span`:
* Less verbose and less fragile code when passing an already existing `Span`(-like) object to or from serialization
The benefits of using `std::byte`:
* `std::byte` can't accidentally be mistaken for an integer
The goal here is to only change serialize to use spans of `std::byte`. If needed, `AsBytes`, `MakeUCharSpan`, ... can be used (temporarily) to pass spans of the right type.
Other changes that are included here:
* [#22167](https://github.com/bitcoin/bitcoin/pull/22167) (refactor: Remove char serialize by MarcoFalke)
* [#21906](https://github.com/bitcoin/bitcoin/pull/21906) (Preserve const in cast on CTransactionSignatureSerializer by promag)
ACKs for top commit:
laanwj:
Concept and code review ACK fa5d2e678c
sipa:
re-utACK fa5d2e678c
Tree-SHA512: 08ee9eced5fb777cedae593b11e33660bed9a3e1711a7451a87b835089a96c99ce0632918bb4666a4e859c4d020f88fb50f2dd734216b0c3d1a9a704967ece6f
Mutex cs_main is already held by the caller of WriteUndoDataForBlock().
This change is needed to require CBlockIndex::GetUndoPos() to hold
cs_main and CBlockIndex::nStatus to be guarded by cs_main in the
following commits without adding 2 unnecessary cs_main locks to
WriteUndoDataForBlock().
e5b6aef612 Move CBlockFileInfo::ToString method where class is declared (Russell Yanofsky)
f7086fd8ff Add src/wallet/* code to wallet:: namespace (Russell Yanofsky)
90fc8b089d Add src/node/* code to node:: namespace (Russell Yanofsky)
Pull request description:
There are no code changes, this is just adding `namespace` and `using` declarations and `node::` or `wallet::` qualifiers in some places.
Motivations for this change are:
- To make it easier to see when node and wallet code is being accessed places where it shouldn't be. For example if GUI code is accessing node and wallet internals or if wallet and node code are referencing each other.
- To make source code organization clearer ([#15732](https://github.com/bitcoin/bitcoin/issues/15732)), being able to know that `wallet::` code is in `src/wallet/`, `node::` code is in `src/node/`, `init::` code is in `src/init/`, `util::` code is in `src/util/`, etc.
Reviewing with `git log -p -n1 -U0 --word-diff-regex=.` can be helpful to verify this is only updating declarations, not changing code.
ACKs for top commit:
achow101:
ACK e5b6aef612
MarcoFalke:
Concept ACK e5b6aef612🍨
Tree-SHA512: 3797745c90246794e2d55a2ee6e8b0ad5c811e4e03a242d3fdfeb68032f8787f0d48ed4097f6b7730f540220c0af99ef423cd9dbe7f76b2ec12e769a757a2c8d
The new helper function, BlockManager::WriteBlockIndexDB(),
has a thread safety lock annotation in its declaration but is
missing the corresponding run-time lock assertion in its definition.
Per doc/developer-notes.md: "Combine annotations in function
declarations with run-time asserts in function definitions."
fa68a6c2fc scripted-diff: Rename touched member variables (MarcoFalke)
facd3df21f Make blockstorage globals private members of BlockManager (MarcoFalke)
faa8c2d7d7 doc: Clarify nPruneAfterHeight for signet (MarcoFalke)
fad381b2f8 test: Load genesis block to allow flush (MarcoFalke)
fab262174b Move blockstorage-related unload to BlockManager::Unload (MarcoFalke)
fa467f3913 move-only: Create WriteBlockIndexDB helper (MarcoFalke)
fa88cfd3f9 Move functions to BlockManager (MarcoFalke)
Pull request description:
Globals aren't too nice because they hide dependencies, also they make testing harder.
Fix that by removing some.
ACKs for top commit:
Sjors:
ACK fa68a6c2fc
ryanofsky:
Code review ACK fa68a6c2fc. Nice changes!
Tree-SHA512: 6abc5929a5e43a05e238276721d46a64a44f23dca18c2caa9775437a32351d6815d88b88757254686421531d0df13861bbd3a202e13a3192798d87a96abef65d
This is a refactor and safe to do because:
* UnloadBlockIndex calls ChainstateManager::Unload, which calls
BlockManager::Unload
* Only unit tests call Unload directly
