Separate the notion of which blocks are stored on disk, and what data is in our
block index, from what tip a chainstate might be able to get to. We can use
chainstate-agnostic data to determine when to store a block on disk (primarily,
an anti-DoS set of criteria) and let the chainstates figure out for themselves
when a block is of interest for being a candidate tip.
Note: some of the invariants in CheckBlockIndex are modified, but more work is
needed (ie to move CheckBlockIndex to ChainstateManager, as most of what
CheckBlockIndex is doing is checking the consistency of the block index, which
is outside of Chainstate).
If a height is missing we are facing a non-contiguous block index db, and could previously
hit an assert in GetAncestor() called from BuildSkip() instead of returning an error.
Block arrival information (and the preciousblock RPC, a related concept) are
both chainstate-agnostic, so these are moved to ChainstateManager. This should
just be a refactor, without any observable behavior changes.
When writing a new block to disk, if we have filled up the current block file,
then we flush and truncate that block file (to free allocated but unused
space) before advancing to the next one. When this happens, we have to
determine whether to also flush and truncate the corresponding undo file.
Undo data is only written when blocks are connected, not when blocks are
received. Thus it's possible that the corresponding undo file already has all
the data it will ever have, and we should flush/truncate it as we advance
files; or it's possible that there is more data we expect to write, and should
therefore defer flush/truncation until undo data is later written.
Prior to this commit, we made the determination of whether the undo file was
full of all requisite data by comparing against the chain tip. This patch
replaces that dependence on validation data structures by instead just tracking
the highest height of any block written in the undo file as we go.
This change drops the last kernel dependency on shutdown.cpp. It also adds new
hooks for libbitcoinkernel applications to be able to interrupt kernel
operations when the chain tip changes.
This is a refactoring that does not affect behavior. (Looking at the code it
can appear like the new break statement in the ActivateBestChain function is a
change in behavior, but actually the previous StartShutdown call was indirectly
triggering a break before, because it was causing m_chainman.m_interrupt to be
true. The new code just makes the break more obvious.)
This has the benefit of moving the StartShutdown call out of the
blockstorage file and thus out of the kernel's responsibility. The user
can now decide if he wants to start shutdown / interrupt after a block
import or not.
At present, during init, we traverse the chain (once per index)
to confirm that all necessary blocks to sync each index up to
the current tip are present.
To make the process more efficient, we can fetch the oldest block
from the indexers and perform the chain data existence check from
that point only once.
This also moves the pruning violation check to the end of the
'loadinit' thread, which is where the reindex, block loading and
chain activation processes happen.
Making the node's startup process faster, allowing us to remove
the global g_indexes_ready_to_sync flag, and enabling the
execution of the pruning violation verification even when the
reindex or reindex-chainstate flags are enabled (which has being
skipped so far).
By generalizing 'GetFirstStoredBlock' and implementing
'CheckBlockDataAvailability' we can dedup code and
avoid repeating work when multiple indexes are enabled.
E.g. get the oldest block across all indexes and
perform the pruning violation check from that point
up to the tip only once (this feature is being introduced
in a follow-up commit).
This commit shouldn't change behavior in any way.
Co-authored-by: Ryan Ofsky <ryan@ofsky.org>
And transfer the responsibility of verifying whether 'start_block'
has data or not to the caller.
This is because the 'GetFirstStoredBlock' function responsibility
is to return the first block containing data. And the current
implementation can return 'start_block' when it has no data!. Which
is misleading at least.
Edge case behavior change:
Previously, if the block tip lacked data but all preceding blocks
contained data, there was no prune violation. And now, such
scenario will result in a prune violation.
No behavior change.
The goal here is to group indexes, so we can perform the same
initialization and verification process equally for all of them.
The checks performed inside `StartIndexes` will be expanded
in the subsequent commits.
The thread does not only load blocks, it loads the mempool and,
in a future commit, will start the indexes as well.
Also, renamed the 'ThreadImport' function to 'ImportBlocks'
And the 'm_load_block' class member to 'm_thread_load'.
-BEGIN VERIFY SCRIPT-
sed -i "s/ThreadImport/ImportBlocks/g" $(git grep -l ThreadImport -- ':!/doc/')
sed -i "s/loadblk/initload/g" $(git grep -l loadblk -- ':!/doc/release-notes/')
sed -i "s/m_load_block/m_thread_load/g" $(git grep -l m_load_block)
-END VERIFY SCRIPT-
The mempool load can take a while, and it is not
needed for the indexes' synchronization.
Also, having the mempool load function call
inside 'blockstorage.cpp' wasn't structurally
correct.
FatalError replaces what previously was the AbortNode function in
shutdown.cpp.
This commit is part of the libbitcoinkernel project and further removes
the shutdown's and, more generally, the kernel library's dependency on
interface_ui with a kernel notification method. By removing interface_ui
from the kernel library, its dependency on boost is reduced to just
boost::multi_index. At the same time it also takes a step towards
de-globalising the interrupt infrastructure.
Co-authored-by: Russell Yanofsky <russ@yanofsky.org>
Co-authored-by: TheCharlatan <seb.kung@gmail.com>
This is done in addition with the following commit. Both have the goal
of getting rid of direct calls to AbortNode from kernel code. This extra
flushError method is added to notify specifically about errors that
arrise when flushing (syncing) block data to disk. Unlike other
instances, the current calls to AbortNode in the blockstorage flush
functions do not report an error to their callers.
This commit is part of the libbitcoinkernel project and further removes
the shutdown's and, more generally, the kernel library's dependency on
interface_ui with a kernel notification method. By removing interface_ui
from the kernel library, its dependency on boost is reduced to just
boost::multi_index. At the same time it also takes a step towards
de-globalising the interrupt infrastructure.
This and the following commit seek to decouple the libbitcoinkernel
library from the shutdown code. As a library, it should it should have
its own flexible interrupt infrastructure without relying on node-wide
globals.
The commit takes the first step towards this goal by de-globalising
`ShutdownRequested` calls in kernel code.
Co-authored-by: Russell Yanofsky <russ@yanofsky.org>
Co-authored-by: TheCharlatan <seb.kung@gmail.com>
32e2ffc393 Remove the syscall sandbox (fanquake)
Pull request description:
After initially being merged in #20487, it's no-longer clear that an internal syscall sandboxing mechanism is something that Bitcoin Core should have/maintain, especially when compared to better maintained/supported alterantives, i.e [firejail](https://github.com/netblue30/firejail).
There is more related discussion in #24771.
Note that given where it's used, the sandbox also gets dragged into the kernel.
If it's removed, this should not require any sort of deprecation, as this was only ever an opt-in, experimental feature.
Closes #24771.
ACKs for top commit:
davidgumberg:
crACK 32e2ffc393
achow101:
ACK 32e2ffc393
dergoegge:
ACK 32e2ffc393
Tree-SHA512: 8cf71c5623bb642cb515531d4a2545d806e503b9d57bfc15a996597632b06103d60d985fd7f843a3c1da6528bc38d0298d6b8bcf0be6f851795a8040d71faf16
1c7d08b9ac validation: Stricter assumeutxo error handling in InvalidateCoinsDBOnDisk (Ryan Ofsky)
9047337d36 validation: Stricter assumeutxo error handling in LoadChainstate (Ryan Ofsky)
Pull request description:
There are two places in assumeutxo code where it is calling `AbortNode` to trigger asynchronous shutdowns without returning errors to calling functions.
One case, in `LoadChainstate`, happens when snapshot validation succeeds, and there is an error trying to replace the background chainstate with the snapshot chainstate.
The other case, in `InvalidateCoinsDBOnDisk`, happens when snapshot validatiion fails, and there is an error trying to remove the snapshot chainstate.
In both cases the node is being forced to shut down, so it makes sense for these functions to raise errors so callers can know that an error happened without having to infer it from the shutdown state.
Noticed these cases while reviewing #27861, which replaces the `AbortNode` function with a `FatalError` function.
ACKs for top commit:
achow101:
ACK 1c7d08b9ac
TheCharlatan:
ACK 1c7d08b9ac
jamesob:
ACK 1c7d08b9ac ([`jamesob/ackr/27862.1.ryanofsky.validation_stricter_assu`](https://github.com/jamesob/bitcoin/tree/ackr/27862.1.ryanofsky.validation_stricter_assu))
Tree-SHA512: fb1dcde3fa0e77b4ba0c48507d289552b939c2866781579c8e994edc209abc3cd29cf81c89380057199323a8eec484956abb1fd3a43c957ecd0e7f7bbfd63fd8
After initially being merged in #20487, it's no-longer clear that an
internal syscall sandboxing mechanism is something that Bitcoin Core
should have/maintain, especially when compared to better
maintained/supported alterantives, i.e firejail.
Note that given where it's used, the sandbox also gets dragged into the
kernel.
There is some related discussion in #24771.
This should not require any sort of deprecation, as this was only ever
an opt-in, experimental feature.
Closes #24771.
Make LoadChainstate return an explicit error when snapshot validation succeeds,
but there is an error trying to replace the background chainstate with the
snapshot chainstate. Previously in this case LoadChainstate would trigger a
shutdown and return INTERRUPTED, now it will return an actual error code.
There's no real change to behavior other than error message being formatted a
little differently.
Motivation for this change is to replace error handling via callbacks with
error handling via return value ahead of
https://github.com/bitcoin/bitcoin/pull/27861
61c569ab60 refactor: decouple early return commands from AppInit (furszy)
4927167f85 gui: return EXIT_FAILURE on post-init fatal errors (furszy)
3b2c61e819 Return EXIT_FAILURE on post-init fatal errors (furszy)
3c06926cf2 refactor: index: use `AbortNode` in fatal error helper (Sebastian Falbesoner)
9ddf7e03a3 move ThreadImport ABC error to use AbortNode (furszy)
Pull request description:
It seems odd to return `EXIT_SUCCESS` when the node aborted execution due a fatal internal error
or any post-init problem that triggers an unrequested shutdown.
e.g. blocks or coins db I/O errors, disconnect block failure, failure during thread import (external
blocks loading process error), among others.
ACKs for top commit:
TheCharlatan:
ACK 61c569ab60
ryanofsky:
Code review ACK 61c569ab60
pinheadmz:
ACK 61c569ab60
theStack:
Code-review ACK 61c569ab60
Tree-SHA512: 18a59c3acc1c6d12cbc74a20a401e89659740c6477fccb59070c9f97922dfe588468e9e5eef56c5f395762187c34179a5e3954aa5b844787fa13da2e666c63d3
It seems odd to return `EXIT_SUCCESS` when the node aborted
execution due a fatal internal error or any post-init problem
that triggers an unrequested shutdown.
e.g. blocks or coins db I/O errors, disconnect block failure,
failure during thread import (external blocks loading process
error), among others.
Co-authored-by: Ryan Ofsky <ryan@ofsky.org>
The slow fuzz seed described in #27799 was just slower than expected,
not an endless loop. Ensuring that every anscestor is only processed
once speeds up the termination of the graph traversal.
Fixes #27799
Remove access to the global gArgs for getting the directory in
utxo_snapshot.
This is done in the context of the libbitcoinkernel project, wherein
reliance of libbitcoinkernel code on the global gArgs is incrementally
removed.
Remove access to the global gArgs for the stopatheight argument and
replace it by adding a field to the existing ChainstateManager Options
struct.
This should eventually allow users of the ChainstateManager to not rely
on the global gArgs and instead pass in their own options.
fa5680b752 fix includes for touched header files (iwyu) (MarcoFalke)
dddde27f6f Add [[nodiscard]] where ignoring a Result return type is an error (MarcoFalke)
Pull request description:
Only add it for those where it is an error to ignore. Also, fix the gcc compile warning https://github.com/bitcoin/bitcoin/pull/25977#issuecomment-1564350880. Also, fix iwyu for touched header files.
ACKs for top commit:
TheCharlatan:
ACK fa5680b752
stickies-v:
ACK fa5680b752
Tree-SHA512: c3509103bfeae246e2cf565bc561fcd68d8118515bac5431ba5304c3a63c8254b9c4f40e268b6f6d6b79405675c5a960db9b4eb3bdd14aedca333dc1c9e76415
7d3b35004b refactor: Move system from util to common library (TheCharlatan)
7eee356c0a refactor: Split util::AnyPtr into its own file (TheCharlatan)
44de325d95 refactor: Split util::insert into its own file (TheCharlatan)
9ec5da36b6 refactor: Move ScheduleBatchPriority to its own file (TheCharlatan)
f871c69191 kernel: Add warning method to notifications (TheCharlatan)
4452707ede kernel: Add progress method to notifications (TheCharlatan)
84d71457e7 kernel: Add headerTip method to notifications (TheCharlatan)
447761c822 kernel: Add notification interface (TheCharlatan)
Pull request description:
This pull request is part of the `libbitcoinkernel` project https://github.com/bitcoin/bitcoin/issues/27587https://github.com/bitcoin/bitcoin/projects/18 and more specifically its "Step 2: Decouple most non-consensus code from libbitcoinkernel".
---
It removes the kernel library's dependency on `util/system` and `interface_ui`. `util/system` contains networking and shell-related code that should not be part of the kernel library. The following pull requests prepared `util/system` for this final step: https://github.com/bitcoin/bitcoin/pull/27419https://github.com/bitcoin/bitcoin/pull/27254https://github.com/bitcoin/bitcoin/pull/27238.
`interface_ui` defines functions for a more general node interface and has a dependency on `boost/signals2`. After applying the patches from this pull request, the kernel's reliance on boost is down to `boost::multiindex`.
The approach implemented here introduces some indirection, which makes the code a bit harder to read. Any suggestions for improving or reworking this pull request to make it more concise, or even reworking it into a more proper interface, are appreciated.
ACKs for top commit:
MarcoFalke:
re-ACK 7d3b35004b (no change) 🎋
stickies-v:
Code Review ACK 7d3b35004b
hebasto:
re-ACK 7d3b35004b, only last two commits dropped since my [recent](https://github.com/bitcoin/bitcoin/pull/27636#pullrequestreview-1435394620) review.
Tree-SHA512: c8cfc698dc9d78e20191c444708f2d957501229abe95e5806106d1126fb9c5fbcee686fb55645658c0107ce71f10646f37a2fdf7fde16bbf22cbf1ac885dd08d
Since the kernel library no longer depends on the system file, move it
to the common library instead in accordance to the diagram in
doc/design/libraries.md.
With the previous move of AlertNotify out of the validation file, and
thus out of the kernel library, ScheduleBatchPriority is the last
remaining function used by the kernel library from util/system. Move it
to its own file, such that util/system can be moved out of the util
library in the following few commits.
Moving util/system out of the kernel library removes further networking
as well as shell related code from it.
This commit is part of the libbitcoinkernel project and seeks to remove
the ChainstateManager's and, more generally, the kernel library's
dependency on interface_ui with options methods in this and the following
few commits. By removing interface_ui from the kernel library, its
dependency on boost is reduced to just boost::multi_index.
The DoWarning and AlertNotify functions are moved out of the
validation.cpp file, which removes its dependency on interface_ui as
well as util/system.
This commit is part of the libbitcoinkernel project and seeks to remove
the ChainstateManager's and, more generally, the kernel library's
dependency on interface_ui with options methods in this and the
following few commits. By removing interface_ui from the kernel library,
its dependency on boost is reduced to just boost::multi_index.
This commit is part of the libbitcoinkernel project and seeks to remove
the ChainstateManager's and, more generally, the kernel library's
dependency on interface_ui with options methods in this and the following
few commits. By removing interface_ui from the kernel library, its
dependency on boost is reduced to just boost::multi_index.
This commit is part of the libbitcoinkernel project and seeks to remove
the ChainstateManager's and, more generally, the kernel library's
dependency on interface_ui with options methods in this and the following
few commits. By removing interface_ui from the kernel library, its
dependency on boost is reduced to just boost::multi_index.
Define a new kernel notification class with virtual methods for
notifying about internal kernel events. Create a new file in the node
library for defining a function creating the default set of notification
methods such that these do not need to be re-defined all over the
codebase. As a first step, add a `blockTip` method, wrapping
`uiInterface.NotifyBlockTip`.
6b605b91c1 [fuzz] Add MiniMiner target + diff fuzz against BlockAssembler (glozow)
3f3f2d59ea [unit test] GatherClusters and MiniMiner unit tests (glozow)
59afcc8354 Implement Mini version of BlockAssembler to calculate mining scores (glozow)
56484f0fdc [mempool] find connected mempool entries with GatherClusters(…) (glozow)
Pull request description:
Implement Mini version of BlockAssembler to calculate mining scores
Run the mining algorithm on a subset of the mempool, only disturbing the
mempool to copy out fee information for relevant entries. Intended to be
used by wallet to calculate amounts needed for fee-bumping unconfirmed
transactions.
From comments of sipa and glozow below:
> > In what way does the code added here differ from the real block assembly code?
>
> * Only operates on the relevant transactions rather than full mempool
> * Has the ability to remove transactions that will be replaced so they don't impact their ancestors
> * Does not hold mempool lock outside of the constructor, makes copies of the entries it needs instead (though I'm not sure if this has an effect in practice)
> * Doesn't do the sanity checks like keeping weight within max block weight and `IsFinalTx()`
> * After the block template is built, additionally calculates fees to bump remaining ancestor packages to target feerate
ACKs for top commit:
achow101:
ACK 6b605b91c1
Xekyo:
> ACK [6b605b9](6b605b91c1) modulo `miniminer_overlap` test.
furszy:
ACK 6b605b91 modulo `miniminer_overlap` test.
theStack:
Code-review ACK 6b605b91c1
Tree-SHA512: f86a8b4ae0506858a7b15d90f417ebceea5038b395c05c825e3796123ad3b6cb8a98ebb948521316802a4c6d60ebd7041093356b1e2c2922a06b3b96b3b8acb6
This is achieved by letting the index sync thread wait until
reindex-chainstate is finished.
This also disables the pruning check when reindexing the chainstate (which is
incompatible with prune mode) because there would be no chain at this point
in init.
Add a stop_after_block_import field to the BlockManager options. Use
this field instead of the global gArgs.
This should allow users of the BlockManager to not rely on the global
Args.
Add a blocks_dir field to the BlockManager options. Move functions
relying on the global gArgs to get the blocks_dir into the BlockManager
class.
This should eventually allow users of the BlockManager to not rely on
the global Args and instead pass in their own options.
Remove access to the global gArgs for the fastprune argument and
replace it by adding a field to the existing BlockManager Options
struct.
When running `clang-tidy-diff` on this commit, there is a diagnostic
error: `unknown type name 'uint64_t' [clang-diagnostic-error] uint64_t
prune_target{0};`, which is fixed by including cstdint.
This should eventually allow users of the BlockManager to not rely on
the global gArgs and instead pass in their own options.
