backport: bitcoin#24322, #25064, #25065, #25168, #25281, #25290, #25307, #25487, #25607, #26003 (kernel backports)

[knst]

Issue being fixed or feature implemented

This PR is a batch of backports of the libdashkernel project, see for details: bitcoin#24303

What was done?

This PR introduces a libdashkernel static library linking in the minimal list of files necessary to use our consensus engine as-is. dash-chainstate introduced in #24304 now will link against libdashkernel.

Beside that other backports are done:

• kernel 2c/n ([kernel 2c/n] Introduce kernel::Context, encapsulate global init/teardown bitcoin/bitcoin#25065)
• kernel 3a/n ([kernel 3a/n] Decouple CTxMemPool from ArgsManager bitcoin/bitcoin#25290)
• kernel 3d/n ([kernel 3d/n] Misc ChainstateManager::Options fixups bitcoin/bitcoin#25607)
• kernel 3b/n ([kernel 3b/n] Decouple {Dump,Load}Mempool from ArgsManager bitcoin/bitcoin#25487)

How Has This Been Tested?

src/dash-chainstate -datadir=/.dashcore still succeed.

Unit & functional tests succeed.

Breaking Changes

N/A

Checklist:

• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have added or updated relevant unit/integration/functional/e2e tests
• I have made corresponding changes to the documentation
• I have assigned this pull request to a milestone

[github-actions]

⚠️ Potential Merge Conflicts Detected

This PR has potential conflicts with the following open PRs:

• backport: bitcoin#25717, #25960, #25963, #25978, #26012, #26172, #26184, #26355, #26387, #26954, #30761 (anti-DOS headers sync) #7318 - backport: bitcoin#25717, #25960, #25963, #25978, #26012, #26172, #26184, #26355, #26387, #26954, #30761 (anti-DOS headers sync)
• refactor: break circular dependency over net_processing and dkgsessionhandler #7314 - refactor: break circular dependency over net_processing and dkgsessionhandler
• backport: Merge bitcoin/bitcoin#28857 , 29458, (partial) 29236, 29459, 29479, 29615, 29433 #7292 - backport: Merge bitcoin/bitcoin#28857 , 29458, (partial) 29236, 29459, 29479, 29615, 29433
• backport: bitcoin#18933, #23123, #23147, #23549, #26074, #26207, #26307, #26341, #26352, #26656, #26886, #26996, #27035 #7192 - backport: bitcoin#18933, #23123, #23147, #23549, #26074, #26207, #26307, #26341, #26352, #26656, #26886, #26996, #27035
• backport: Merge bitcoin#30170, 30026, 30017, 29961, 29904, 29910, 29849 #7257 - backport: Merge bitcoin#30170, 30026, 30017, 29961, 29904, 29910, 29849
• backport: Merge bitcoin#28486, 28969, 27921, 27177, 26331 #7187 - backport: Merge bitcoin#28486, 28969, 27921, 27177, 26331

Please coordinate with the authors of these PRs to avoid merge conflicts.

[thepastaclaw]

✅ Review complete (commit 2d14616)

[coderabbitai]

Walkthrough

This PR adds kernel context, kernel sanity checks, and mempool persistence APIs. It refactors mempool, chainstate, and validation code to use kernel-backed options, explicit paths, and adjusted-time callbacks. It also updates node startup, peer management, mining, RPC, tests, build wiring, and lint/functional test inputs to match the new interfaces.

Estimated code review effort

🎯 5 (Critical) | ⏱️ ~120+ minutes

Possibly related PRs

• dashpay/dash#7234: Touches the same dash-chainstate build wiring in configure.ac and src/Makefile.am.
• dashpay/dash#7330: Refactors PeerManager::make and PeerManagerImpl in src/net_processing.{h,cpp}.

Suggested reviewers

• PastaPastaPasta
• UdjinM6

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

[coderabbitai]

Actionable comments posted: 5

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

src/rpc/rawtransaction.cpp (1)

379-407: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Incomplete migration from global Params() to chainman parameters across RPC endpoints. The refactoring updates getrawtransaction() to use the active chainman's consensus parameters instead of the global singleton, but similar operations in getrawtransactionmulti() and gettxchainlocks() still use the global Params(). This creates inconsistent consensus parameter sourcing across functionally identical RPC endpoints and conflicts with the PR's stated goal to decouple from global singletons.

• src/rpc/rawtransaction.cpp#L379-L407: correctly migrated to chainman.GetParams() and chainman.GetConsensus().
• src/rpc/rawtransaction.cpp#L524-L524: replace Params().GetConsensus() with chainman.GetConsensus() in the GetTransaction call within getrawtransactionmulti().
• src/rpc/rawtransaction.cpp#L670-L670: replace Params().GenesisBlock().hashMerkleRoot with chainman.GetParams().GenesisBlock().hashMerkleRoot in the genesis check within gettxchainlocks().
• src/rpc/rawtransaction.cpp#L679-L679: replace Params().GetConsensus() with chainman.GetConsensus() in the GetTransaction call within gettxchainlocks().

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/rpc/rawtransaction.cpp` around lines 379 - 407, The code mixes global
Params()/GetConsensus() with chainman; unify to use chainman as done in the
anchor region: in src/rpc/rawtransaction.cpp (lines 379-407) the pattern is
correct; for src/rpc/rawtransaction.cpp (line 524) inside
getrawtransactionmulti() replace Params().GetConsensus() with
chainman.GetConsensus() in the GetTransaction call; for
src/rpc/rawtransaction.cpp (line 670) inside gettxchainlocks() replace
Params().GenesisBlock().hashMerkleRoot with
chainman.GetParams().GenesisBlock().hashMerkleRoot for the genesis coinbase
check; and for src/rpc/rawtransaction.cpp (line 679) inside gettxchainlocks()
replace Params().GetConsensus() with chainman.GetConsensus() in the
GetTransaction call so all RPCs consistently use chainman for params/consensus
(adjust include/use of chainman variable if needed).

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@configure.ac`:
- Around line 1800-1804: The AC_MSG_ERROR currently triggers whenever
build_experimental_kernel_lib = "no" even if chainstate isn't requested; change
the logic so the error is emitted only when both build_bitcoin_chainstate =
"yes" and build_experimental_kernel_lib = "no". Concretely, replace the
top-level if that checks build_experimental_kernel_lib with a conditional that
tests both variables (e.g., if test "$build_bitcoin_chainstate" = "yes" && test
"$build_experimental_kernel_lib" = "no"; then AC_MSG_ERROR(...); else
AM_CONDITIONAL([BUILD_BITCOIN_CHAINSTATE], [test $build_bitcoin_chainstate =
"yes"]) fi), leaving the AM_CONDITIONAL for BUILD_BITCOIN_CHAINSTATE unchanged.

In `@src/kernel/checks.h`:
- Around line 14-19: The enum value ERROR_CHRONO is unreachable because
kernel::SanityChecks() never returns it while AppInitSanityChecks still handles
it; fix by either implementing the missing chrono sanity check in
kernel::SanityChecks() so it can return SanityCheckError::ERROR_CHRONO (add a
chrono validity test and return that enum on failure), or remove ERROR_CHRONO
from the SanityCheckError enum and delete the corresponding case in
AppInitSanityChecks (and update any switch/default logic). Ensure the chosen
change keeps enum and switch consistent and compile-clean by updating any
related tests/comments referencing ERROR_CHRONO.

In `@src/node/chainstate.cpp`:
- Line 251: The comparison in VerifyLoadedChainstate uses GetTime() directly,
bypassing the injected time callback; replace the hardcoded GetTime() call with
the provided get_unix_time_seconds() callback so the check if (tip && tip->nTime
> ...) becomes consistent with the injected time source (update the code in
VerifyLoadedChainstate where tip->nTime is compared to GetTime() +
MAX_FUTURE_BLOCK_TIME to use get_unix_time_seconds() instead).

In `@src/test/fuzz/tx_pool.cpp`:
- Around line 120-131: The code currently constructs a CTxMemPool via
MakeMempool but then UB-ly downcasts tx_pool_ to MockedTxPool with
*static_cast<MockedTxPool*>(&tx_pool_); fix this by actually constructing a
MockedTxPool instead of a CTxMemPool where the cast is used: either change the
declaration tx_pool_ to be MockedTxPool (and update MakeMempool to return a
MockedTxPool or add a MakeMockedMempool helper that builds the same Options and
constructs a MockedTxPool), and remove the static_cast usages; do not leave a
CTxMemPool and cast it to MockedTxPool. Ensure references to CTxMemPool,
MockedTxPool, tx_pool_, MakeMempool and the static_cast sites are updated
consistently.

In `@src/test/mempool_tests.cpp`:
- Around line 20-24: The test currently static_casts m_node.mempool to
MemPoolTest&, which is UB because m_node.mempool is actually a
std::unique_ptr<CTxMemPool>; fix by constructing a real MemPoolTest object
instead of a CTxMemPool: replace the
std::make_unique<CTxMemPool>(MemPoolOptionsForTest(m_node)) allocation with
std::make_unique<MemPoolTest>(MemPoolOptionsForTest(m_node)) so the dynamic type
matches MemPoolTest, or alternatively change the fixture/member type that holds
m_node.mempool from std::unique_ptr<CTxMemPool> to std::unique_ptr<MemPoolTest>
and initialize it with MemPoolOptionsForTest(m_node); ensure all usages that
rely on CTxMemPool still compile against the MemPoolTest subclass.

---

Outside diff comments:
In `@src/rpc/rawtransaction.cpp`:
- Around line 379-407: The code mixes global Params()/GetConsensus() with
chainman; unify to use chainman as done in the anchor region: in
src/rpc/rawtransaction.cpp (lines 379-407) the pattern is correct; for
src/rpc/rawtransaction.cpp (line 524) inside getrawtransactionmulti() replace
Params().GetConsensus() with chainman.GetConsensus() in the GetTransaction call;
for src/rpc/rawtransaction.cpp (line 670) inside gettxchainlocks() replace
Params().GenesisBlock().hashMerkleRoot with
chainman.GetParams().GenesisBlock().hashMerkleRoot for the genesis coinbase
check; and for src/rpc/rawtransaction.cpp (line 679) inside gettxchainlocks()
replace Params().GetConsensus() with chainman.GetConsensus() in the
GetTransaction call so all RPCs consistently use chainman for params/consensus
(adjust include/use of chainman variable if needed).

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: e36e1775-c57a-44f6-ab45-54cb8d77291c

📥 Commits

Reviewing files that changed from the base of the PR and between 7fcc54a and a1acc28.

📒 Files selected for processing (87)

• ci/dash/lint-tidy.sh
• ci/test/00_setup_env_native_nowallet_libbitcoinkernel.sh
• configure.ac
• contrib/devtools/iwyu/bitcoin.core.imp
• src/Makefile.am
• src/Makefile.test_fuzz.include
• src/bitcoin-chainstate.cpp
• src/bitcoind.cpp
• src/bls/bls.cpp
• src/bls/bls.h
• src/deploymentstatus.h
• src/evo/mnhftx.cpp
• src/fs.h
• src/init.cpp
• src/init.h
• src/init/common.cpp
• src/init/common.h
• src/kernel/bitcoinkernel.cpp
• src/kernel/chainstatemanager_opts.h
• src/kernel/checks.cpp
• src/kernel/checks.h
• src/kernel/context.cpp
• src/kernel/context.h
• src/kernel/mempool_limits.h
• src/kernel/mempool_options.h
• src/kernel/mempool_persist.cpp
• src/kernel/mempool_persist.h
• src/mempool_args.cpp
• src/mempool_args.h
• src/net_processing.cpp
• src/net_processing.h
• src/node/blockstorage.cpp
• src/node/blockstorage.h
• src/node/chainstate.cpp
• src/node/chainstate.h
• src/node/context.cpp
• src/node/context.h
• src/node/interfaces.cpp
• src/node/mempool_persist_args.cpp
• src/node/mempool_persist_args.h
• src/node/miner.cpp
• src/node/miner.h
• src/policy/fees.cpp
• src/policy/fees.h
• src/policy/fees_args.cpp
• src/policy/fees_args.h
• src/policy/packages.h
• src/policy/policy.h
• src/rest.cpp
• src/rpc/blockchain.cpp
• src/rpc/fees.cpp
• src/rpc/mempool.cpp
• src/rpc/mining.cpp
• src/rpc/rawtransaction.cpp
• src/rpc/txoutproof.cpp
• src/test/block_reward_reallocation_tests.cpp
• src/test/blockfilter_index_tests.cpp
• src/test/denialofservice_tests.cpp
• src/test/dynamic_activation_thresholds_tests.cpp
• src/test/evo_deterministicmns_tests.cpp
• src/test/fuzz/mempool_utils.h
• src/test/fuzz/policy_estimator.cpp
• src/test/fuzz/policy_estimator_io.cpp
• src/test/fuzz/tx_pool.cpp
• src/test/fuzz/validation_load_mempool.cpp
• src/test/mempool_tests.cpp
• src/test/miner_tests.cpp
• src/test/net_peer_connection_tests.cpp
• src/test/txpackage_tests.cpp
• src/test/util/mining.cpp
• src/test/util/setup_common.cpp
• src/test/util/setup_common.h
• src/test/util/txmempool.cpp
• src/test/util/txmempool.h
• src/test/validation_block_tests.cpp
• src/test/validation_chainstate_tests.cpp
• src/test/validation_chainstatemanager_tests.cpp
• src/txmempool.cpp
• src/txmempool.h
• src/util/system.cpp
• src/util/system.h
• src/validation.cpp
• src/validation.h
• test/functional/mempool_expiry.py
• test/functional/mempool_persist.py
• test/lint/lint-circular-dependencies.py
• test/lint/spelling.ignore-words.txt

💤 Files with no reviewable changes (3)

• src/init/common.h
• src/init/common.cpp
• src/bls/bls.cpp

[thepastaclaw]

Code Review

Verified all six agent findings against the worktree at 9a2c79d. One blocking configure regression in the new --with-experimental-kernel-lib check: the guard fires whenever the kernel lib is disabled even though dash-chainstate is opt-in and defaults to no, so a plain --with-experimental-kernel-lib=no aborts configure. Three small backport-cleanup leftovers (stale Dump/LoadMempool declarations in validation.h, duplicate common/bloom.h line in Makefile.am, dead init::UnsetGlobals comment in bitcoin-chainstate.cpp) and two harmless adaptation residues (unused get_unix_time_seconds parameter, unreachable ERROR_CHRONO enum value) are kept as suggestions/nitpicks.

🔴 1 blocking | 🟡 1 suggestion(s) | 💬 4 nitpick(s)

1 additional finding(s) omitted (not in diff).

🤖 Prompt for all review comments with AI agents

These findings are from an automated code review. Verify each finding against the current code and only fix it if needed.

In `configure.ac`:
- [BLOCKING] configure.ac:1799-1805: --with-experimental-kernel-lib=no aborts configure even when dash-chainstate is not requested
  The new guard at configure.ac:1800 fires on `build_experimental_kernel_lib = no` unconditionally, but `build_bitcoin_chainstate` defaults to `no` (line 777) and is also forced to `no` at line 1460. The result: any invocation of `./configure --with-experimental-kernel-lib=no` (or any future path that disables the kernel library) bails out with an error about dash-chainstate, even though dash-chainstate was never requested. The check must be gated on `build_bitcoin_chainstate = yes` so it only triggers when the user actually asked for dash-chainstate while explicitly disabling the kernel library. As a bonus, the `AM_CONDITIONAL` must be emitted unconditionally so automake does not see a conditionally-defined conditional.

In `src/validation.h`:
- [SUGGESTION] src/validation.h:1110-1116: Stale global LoadMempool/DumpMempool declarations left behind after move to kernel::
  bitcoin#25487 moves DumpMempool/LoadMempool into the kernel:: namespace with new signatures (taking dump_path / load_path) and the upstream hunk removes the old declarations from validation.h. This Dash backport correctly relocates the definitions to src/kernel/mempool_persist.cpp and updates callers via `using kernel::DumpMempool; using kernel::LoadMempool;`, but validation.h still advertises the old global-namespace declarations together with the `FopenFn` alias:

      using FopenFn = std::function<FILE*(const fs::path&, const char*)>;
      bool DumpMempool(const CTxMemPool& pool, FopenFn = fsbridge::fopen, bool skip_file_commit = false);
      bool LoadMempool(CTxMemPool& pool, CChainState& active_chainstate, FopenFn = fsbridge::fopen);

  No definitions match these signatures anywhere in the tree, so any future TU that includes validation.h and calls `LoadMempool(pool, chainstate)` will compile and then fail at link time with an undefined reference instead of the clear compile error it should get. Drop the dead declarations (and the now-unused `FopenFn` alias if nothing else uses it) to match the upstream cleanup.

[thepastaclaw]

Code Review

Delta between previously reviewed SHA 9a2c79d and current head a1acc28 is empty — no new source changes. Both reviewer agents independently re-validated all 6 prior findings against the current head and found them STILL VALID; I verified each one in the worktree. One blocking issue (configure.ac --with-experimental-kernel-lib=no guard), one suggestion (stale validation.h declarations), and four nitpicks are carried forward.

🟡 1 suggestion(s)

1 additional finding(s) omitted (not in diff).

5 carried-forward finding(s) already raised on this PR; not re-posting as new inline comments.

🤖 Prompt for all review comments with AI agents

These findings are from an automated code review. Verify each finding against the current code and only fix it if needed.

In `src/validation.h`:
- [SUGGESTION] src/validation.h:1110-1116: Stale global LoadMempool/DumpMempool declarations left behind after move to kernel::
  Verified at a1acc28a — validation.h:1110-1116 still declares `using FopenFn = std::function<FILE*(const fs::path&, const char*)>;` plus `DumpMempool(const CTxMemPool&, FopenFn, bool)` and `LoadMempool(CTxMemPool&, CChainState&, FopenFn)` in the global namespace. bitcoin#25487 moved these into `kernel::` with new path-taking signatures and the upstream hunk removed these declarations from validation.h. This backport correctly relocated the definitions to src/kernel/mempool_persist.cpp and updates callers via `using kernel::DumpMempool; using kernel::LoadMempool;`, but the old global declarations remain advertised with no matching definitions. A future TU that includes validation.h and calls `LoadMempool(pool, chainstate)` will compile and then fail at link time with an undefined reference instead of producing a clear compile error. Drop the dead declarations and the now-unused `FopenFn` alias to match the upstream cleanup.

[thepastaclaw]

Code Review — correction / carry-forward

Correction: the review system briefly posted auto-resolve replies on several prior threads even though this recheck verified those findings are still valid. I removed those incorrect auto-resolve replies.

I re-reviewed current head a1acc28a4ba1737d8bba27498b3a9ea1dd288fe0 cumulatively and checked the requested delta git diff 9a2c79db..a1acc28a4ba1737d8bba27498b3a9ea1dd288fe0. The delta is empty, and all six prior findings remain valid at the current head:

• STILL VALID / blocking: configure.ac — --with-experimental-kernel-lib=no still aborts configure even when dash-chainstate is not requested.
• STILL VALID / suggestion: src/validation.h — stale global LoadMempool / DumpMempool declarations remain after the move to kernel::.
• STILL VALID / nitpick: src/Makefile.am — duplicate common/bloom.h remains in BITCOIN_CORE_H.
• STILL VALID / nitpick: src/bitcoin-chainstate.cpp — stale comment still references removed init::UnsetGlobals().
• STILL VALID / nitpick: src/node/chainstate.cpp — unused get_unix_time_seconds parameter remains from the partial bitcoin#25168 adaptation.
• STILL VALID / nitpick: src/kernel/checks.h — unreachable SanityCheckError::ERROR_CHRONO remains after the BLS-substitution adaptation of bitcoin#25065.

No new latest-delta findings were found. Requesting changes is still warranted because the carried-forward configure issue is blocking.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 461435620a

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[coderabbitai]

Actionable comments posted: 1

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

src/validation.cpp (1)

4342-4379: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Restore bls_legacy_scheme on every exit path.

TestBlockValidity() snapshots the global flag, but every failure return before the tail restore skips putting it back. If any of the validation calls in this function flip the scheme and then fail, later validations inherit the wrong global state. Please move the restore into a scope-exit/RAII guard so it runs on both success and error paths.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/validation.cpp` around lines 4342 - 4379, TestBlockValidity currently
snapshots bls::bls_legacy_scheme into a local bls_legacy_scheme but only
restores it at the end, skipping all early returns; wrap that snapshot in an
RAII/scope-exit guard so the original value is restored on every exit (success
or any early return) — e.g. after auto bls_legacy_scheme =
bls::bls_legacy_scheme.load(); create a small scope_guard that calls
bls::bls_legacy_scheme.store(bls_legacy_scheme) in its destructor, remove the
manual tail restore block that checks and stores the value, and leave all
existing checks (ContextualCheckBlockHeader, CheckBlock, ContextualCheckBlock,
chainstate.ConnectBlock, etc.) unchanged so the guard handles restoration even
when those calls return early.

♻️ Duplicate comments (1)

src/test/fuzz/tx_pool.cpp (1)

120-131: ⚠️ Potential issue | 🔴 Critical | ⚡ Quick win

Constructing CTxMemPool then downcasting to MockedTxPool is UB.

Line 120 returns a base CTxMemPool, but Line 155 and Line 331 cast those objects to MockedTxPool&. That downcast is invalid unless the object was originally constructed as MockedTxPool.

Suggested fix

 struct MockedTxPool : public CTxMemPool {
+    using CTxMemPool::CTxMemPool;
+
     void RollingFeeUpdate() EXCLUSIVE_LOCKS_REQUIRED(!cs)
     {
         LOCK(cs);
         lastRollingFeeUpdate = GetTime();
         blockSinceLastRollingFeeBump = true;
     }
 };
@@
-CTxMemPool MakeMempool(const NodeContext& node)
+MockedTxPool MakeMempool(const NodeContext& node)
 {
@@
-    return CTxMemPool{mempool_opts};
+    return MockedTxPool{mempool_opts};
 }
@@
-    CTxMemPool tx_pool_{MakeMempool(node)};
-    MockedTxPool& tx_pool = *static_cast<MockedTxPool*>(&tx_pool_);
+    MockedTxPool tx_pool{MakeMempool(node)};
@@
-    CTxMemPool tx_pool_{MakeMempool(node)};
-    MockedTxPool& tx_pool = *static_cast<MockedTxPool*>(&tx_pool_);
+    MockedTxPool tx_pool{MakeMempool(node)};

#!/bin/bash
set -euo pipefail
# Verify the base construction + derived downcast pattern is present.
rg -n -C2 'MockedTxPool|MakeMempool\(|static_cast<MockedTxPool\*>\(&tx_pool_\)' src/test/fuzz/tx_pool.cpp

Also applies to: 154-156, 330-332

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/test/fuzz/tx_pool.cpp` around lines 120 - 131, Make MakeMempool construct
and return a polymorphic heap object (avoid slicing) instead of returning
CTxMemPool by value: change the signature of MakeMempool to return
std::unique_ptr<CTxMemPool> and return
std::make_unique<MockedTxPool>(mempool_opts) (use MockedTxPool's constructor
that accepts mempool_opts); then update call sites that currently take the
returned value (and that do static_cast<MockedTxPool*>(&tx_pool_) or
static_cast<MockedTxPool&>(...)) to hold the std::unique_ptr and either
dynamic_cast the pointer to MockedTxPool* (and check for null) or otherwise use
the CTxMemPool virtual interface so no downcast is needed—this ensures
MakeMempool, CTxMemPool, MockedTxPool, and places referencing tx_pool_ are
consistent and avoid undefined behavior.

🧹 Nitpick comments (1)

src/kernel/context.cpp (1)

25-27: BLSInit() is safe to call twice (allocator callback assignment), but the sanity-check error path is redundant

• kernel::Context::Context() calls BLSInit() (src/kernel/context.cpp:25-27), and kernel::SanityChecks() calls it again (src/kernel/checks.cpp:19-20).
• BLSInit() (src/bls/bls.cpp:431-437) only invokes bls::BLS::SetSecureAllocator(secure_allocate, secure_free) and unconditionally returns true; it does not perform the one-time relic/BLS library initialization.
• bls::BLS::SetSecureAllocator (src/dashbls/src/bls.cpp:72-78) only assigns global callback function pointers, so repeated calls just overwrite the same callbacks.
• One-time BLS library init happens separately via bool BLSInitResult = BLS::Init(); (src/dashbls/src/bls.cpp:29), so the kernel startup path isn’t duplicating that init.

Optional cleanup: since BLSInit() can’t fail, the if (!BLSInit()) { ... ERROR_BLS } branch in src/kernel/checks.cpp is unnecessary.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/kernel/context.cpp` around lines 25 - 27, The constructor call to
BLSInit() in src/kernel/context.cpp (lines 25-27) is fine and requires no
change; remove the redundant BLSInit() invocation and its failing-branch in
src/kernel/checks.cpp (lines 19-20) — delete the if (!BLSInit()) { ... ERROR_BLS
} block (or at minimum the BLSInit() call and related error handling) since
BLSInit() cannot fail and the one-time library init occurs elsewhere.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/kernel/checks.cpp`:
- Around line 13-27: SanityChecks currently never returns
SanityCheckError::ERROR_CHRONO; update the SanityChecks(const Context&)
implementation to invoke the chrono sanity check and return ERROR_CHRONO on
failure (e.g., call the existing chrono-check function — name it
ChronoSanityCheck() or the actual project function that validates
monotonic/chrono behavior — and if it returns false, return
SanityCheckError::ERROR_CHRONO); alternatively, if no chrono check exists,
remove the ERROR_CHRONO enum and any caller handling (e.g., in src/init.cpp) so
the public contract and callers stay consistent.

---

Outside diff comments:
In `@src/validation.cpp`:
- Around line 4342-4379: TestBlockValidity currently snapshots
bls::bls_legacy_scheme into a local bls_legacy_scheme but only restores it at
the end, skipping all early returns; wrap that snapshot in an RAII/scope-exit
guard so the original value is restored on every exit (success or any early
return) — e.g. after auto bls_legacy_scheme = bls::bls_legacy_scheme.load();
create a small scope_guard that calls
bls::bls_legacy_scheme.store(bls_legacy_scheme) in its destructor, remove the
manual tail restore block that checks and stores the value, and leave all
existing checks (ContextualCheckBlockHeader, CheckBlock, ContextualCheckBlock,
chainstate.ConnectBlock, etc.) unchanged so the guard handles restoration even
when those calls return early.

---

Duplicate comments:
In `@src/test/fuzz/tx_pool.cpp`:
- Around line 120-131: Make MakeMempool construct and return a polymorphic heap
object (avoid slicing) instead of returning CTxMemPool by value: change the
signature of MakeMempool to return std::unique_ptr<CTxMemPool> and return
std::make_unique<MockedTxPool>(mempool_opts) (use MockedTxPool's constructor
that accepts mempool_opts); then update call sites that currently take the
returned value (and that do static_cast<MockedTxPool*>(&tx_pool_) or
static_cast<MockedTxPool&>(...)) to hold the std::unique_ptr and either
dynamic_cast the pointer to MockedTxPool* (and check for null) or otherwise use
the CTxMemPool virtual interface so no downcast is needed—this ensures
MakeMempool, CTxMemPool, MockedTxPool, and places referencing tx_pool_ are
consistent and avoid undefined behavior.

---

Nitpick comments:
In `@src/kernel/context.cpp`:
- Around line 25-27: The constructor call to BLSInit() in src/kernel/context.cpp
(lines 25-27) is fine and requires no change; remove the redundant BLSInit()
invocation and its failing-branch in src/kernel/checks.cpp (lines 19-20) —
delete the if (!BLSInit()) { ... ERROR_BLS } block (or at minimum the BLSInit()
call and related error handling) since BLSInit() cannot fail and the one-time
library init occurs elsewhere.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: a24de521-92cd-4f11-ac40-1d970d8854a3

📥 Commits

Reviewing files that changed from the base of the PR and between a1acc28 and 4614356.

📒 Files selected for processing (88)

• ci/dash/lint-tidy.sh
• ci/test/00_setup_env_native_nowallet_libbitcoinkernel.sh
• configure.ac
• contrib/devtools/iwyu/bitcoin.core.imp
• src/Makefile.am
• src/Makefile.test_fuzz.include
• src/bitcoin-chainstate.cpp
• src/bitcoind.cpp
• src/bls/bls.h
• src/bls/bls_worker.cpp
• src/bls/bls_worker.h
• src/deploymentstatus.h
• src/evo/mnhftx.cpp
• src/fs.h
• src/init.cpp
• src/init.h
• src/init/common.cpp
• src/init/common.h
• src/kernel/bitcoinkernel.cpp
• src/kernel/chainstatemanager_opts.h
• src/kernel/checks.cpp
• src/kernel/checks.h
• src/kernel/context.cpp
• src/kernel/context.h
• src/kernel/mempool_limits.h
• src/kernel/mempool_options.h
• src/kernel/mempool_persist.cpp
• src/kernel/mempool_persist.h
• src/mempool_args.cpp
• src/mempool_args.h
• src/net_processing.cpp
• src/net_processing.h
• src/node/blockstorage.cpp
• src/node/blockstorage.h
• src/node/chainstate.cpp
• src/node/chainstate.h
• src/node/context.cpp
• src/node/context.h
• src/node/interfaces.cpp
• src/node/mempool_persist_args.cpp
• src/node/mempool_persist_args.h
• src/node/miner.cpp
• src/node/miner.h
• src/policy/fees.cpp
• src/policy/fees.h
• src/policy/fees_args.cpp
• src/policy/fees_args.h
• src/policy/packages.h
• src/policy/policy.h
• src/rest.cpp
• src/rpc/blockchain.cpp
• src/rpc/fees.cpp
• src/rpc/mempool.cpp
• src/rpc/mining.cpp
• src/rpc/rawtransaction.cpp
• src/rpc/txoutproof.cpp
• src/test/block_reward_reallocation_tests.cpp
• src/test/blockfilter_index_tests.cpp
• src/test/denialofservice_tests.cpp
• src/test/dynamic_activation_thresholds_tests.cpp
• src/test/evo_deterministicmns_tests.cpp
• src/test/fuzz/mempool_utils.h
• src/test/fuzz/policy_estimator.cpp
• src/test/fuzz/policy_estimator_io.cpp
• src/test/fuzz/tx_pool.cpp
• src/test/fuzz/validation_load_mempool.cpp
• src/test/mempool_tests.cpp
• src/test/miner_tests.cpp
• src/test/net_peer_connection_tests.cpp
• src/test/txpackage_tests.cpp
• src/test/util/mining.cpp
• src/test/util/setup_common.cpp
• src/test/util/setup_common.h
• src/test/util/txmempool.cpp
• src/test/util/txmempool.h
• src/test/validation_block_tests.cpp
• src/test/validation_chainstate_tests.cpp
• src/test/validation_chainstatemanager_tests.cpp
• src/txmempool.cpp
• src/txmempool.h
• src/util/system.cpp
• src/util/system.h
• src/validation.cpp
• src/validation.h
• test/functional/mempool_expiry.py
• test/functional/mempool_persist.py
• test/lint/lint-circular-dependencies.py
• test/lint/spelling.ignore-words.txt

💤 Files with no reviewable changes (3)

• src/init/common.h
• src/init/common.cpp
• src/bls/bls.h

✅ Files skipped from review due to trivial changes (10)

• test/lint/spelling.ignore-words.txt
• src/node/context.cpp
• ci/test/00_setup_env_native_nowallet_libbitcoinkernel.sh
• ci/dash/lint-tidy.sh
• contrib/devtools/iwyu/bitcoin.core.imp
• src/evo/mnhftx.cpp
• src/policy/packages.h
• src/test/validation_chainstate_tests.cpp
• src/policy/fees_args.cpp
• test/functional/mempool_expiry.py

🚧 Files skipped from review as they are similar to previous changes (50)

• src/test/fuzz/mempool_utils.h
• src/kernel/checks.h
• src/test/mempool_tests.cpp
• src/test/util/mining.cpp
• src/test/miner_tests.cpp
• src/kernel/mempool_limits.h
• src/Makefile.test_fuzz.include
• src/rpc/txoutproof.cpp
• src/rest.cpp
• src/mempool_args.h
• src/node/blockstorage.cpp
• src/node/blockstorage.h
• src/kernel/context.h
• src/test/txpackage_tests.cpp
• src/test/validation_chainstatemanager_tests.cpp
• src/bitcoind.cpp
• src/test/validation_block_tests.cpp
• test/lint/lint-circular-dependencies.py
• src/init.h
• src/deploymentstatus.h
• src/test/util/txmempool.h
• src/net_processing.h
• src/policy/fees.h
• src/node/miner.h
• src/test/denialofservice_tests.cpp
• src/fs.h
• src/util/system.h
• src/kernel/mempool_persist.h
• src/kernel/chainstatemanager_opts.h
• src/node/miner.cpp
• src/policy/fees_args.h
• src/rpc/blockchain.cpp
• src/kernel/mempool_options.h
• src/node/context.h
• src/test/evo_deterministicmns_tests.cpp
• src/util/system.cpp
• src/policy/policy.h
• src/node/chainstate.h
• configure.ac
• src/net_processing.cpp
• src/policy/fees.cpp
• src/txmempool.h
• src/txmempool.cpp
• src/node/interfaces.cpp
• src/node/chainstate.cpp
• src/validation.h
• src/rpc/fees.cpp
• src/rpc/mining.cpp
• src/init.cpp
• src/Makefile.am

[thepastaclaw]

Code Review

Cumulative review of PR #7361 at 4614356. The latest delta (a1acc28..4614356) is a small housekeeping pass — wrapping Dash BLS APIs in #ifndef BUILD_BITCOIN_INTERNAL for libdashkernel exclusion and removing one stale using node::ShouldPersistMempool in rpc/mempool.cpp. No new defects in the delta. All six prior findings from a1acc28 remain unaddressed and are carried forward verbatim. The blocking issue remains configure.ac aborting when --with-experimental-kernel-lib=no is passed even without dash-chainstate being requested.

🟡 1 suggestion(s)

1 additional finding(s) omitted (not in diff).

5 carried-forward finding(s) already raised on this PR; not re-posting as new inline comments.

🤖 Prompt for all review comments with AI agents

These findings are from an automated code review. Verify each finding against the current code and only fix it if needed.

In `src/validation.h`:
- [SUGGESTION] src/validation.h:1110-1116: Stale global LoadMempool/DumpMempool declarations left behind after move to kernel::
  Carried forward — STILL VALID at 461435620a. Verified: validation.h:1110-1116 still declares `using FopenFn = std::function<FILE*(const fs::path&, const char*)>;` plus `DumpMempool(const CTxMemPool&, FopenFn, bool)` and `LoadMempool(CTxMemPool&, CChainState&, FopenFn)` in the global namespace. bitcoin#25487 moved these into kernel:: with new path-taking signatures (definitions now live at src/kernel/mempool_persist.cpp); the upstream hunk removed these prototypes from validation.h. Callers in-tree correctly use `using kernel::DumpMempool; using kernel::LoadMempool;`, but the old globals are still advertised with no matching definitions. A future translation unit including validation.h that calls `LoadMempool(pool, chainstate)` will compile successfully and then fail at link time with an undefined reference instead of producing a clear compile error. Drop the dead declarations and the now-unused FopenFn alias, and have callers include kernel/mempool_persist.h directly.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 586000ee34

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Remove the unavailable chrono sanity case

This switch now handles kernel::SanityCheckError::ERROR_CHRONO, but SanityCheckError only defines ERROR_ECC, ERROR_BLS, and ERROR_RANDOM in src/kernel/checks.h, so any build compiling init.cpp fails because ERROR_CHRONO is not a member. Either add the enum value and corresponding check in kernel::SanityChecks or drop this case.

Useful? React with 👍 / 👎.

[coderabbitai]

Actionable comments posted: 1

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

src/Makefile.am (1)

1368-1368: ⚠️ Potential issue | 🔴 Critical | ⚡ Quick win

Typo in dependency rule will cause build failure.

The object file extension .l$(OBJEXT) should be .$(OBJEXT). The extra l character means this dependency rule won't match the actual object file, so obj/build.h won't be rebuilt before compiling clientversion.cpp for the kernel library.

Proposed fix

-libdashkernel_la-clientversion.l$(OBJEXT): obj/build.h
+libdashkernel_la-clientversion.$(OBJEXT): obj/build.h

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/Makefile.am` at line 1368, In the Makefile.am dependency rule for the
libdashkernel_la-clientversion object file, there is a typo where the object
file extension is written as `.l$(OBJEXT)` instead of `.$(OBJEXT)`. Remove the
extra `l` character from the dependency rule so that the actual object file name
matches what the build system creates, ensuring that the obj/build.h
prerequisite is properly recognized and rebuilt before clientversion.cpp is
compiled for the kernel library.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/rpc/mempool.cpp`:
- Around line 694-695: The field name "loaded" in the mempool RPC response at
the pushKV call does not match the semantics of GetLoadTried(), which returns
whether an attempt to load was made, not whether it succeeded. Either rename the
field from "loaded" to a name that reflects attempt status such as
"load_attempted", or update the RPC schema documentation to clarify that this
field indicates an attempt was made to load the mempool regardless of success.
Choose the approach that aligns with your API documentation standards.

---

Outside diff comments:
In `@src/Makefile.am`:
- Line 1368: In the Makefile.am dependency rule for the
libdashkernel_la-clientversion object file, there is a typo where the object
file extension is written as `.l$(OBJEXT)` instead of `.$(OBJEXT)`. Remove the
extra `l` character from the dependency rule so that the actual object file name
matches what the build system creates, ensuring that the obj/build.h
prerequisite is properly recognized and rebuilt before clientversion.cpp is
compiled for the kernel library.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: a5c75688-6d32-48c9-9f68-fe308b760c2e

📥 Commits

Reviewing files that changed from the base of the PR and between 4614356 and 586000e.

📒 Files selected for processing (83)

• ci/dash/lint-tidy.sh
• configure.ac
• contrib/devtools/iwyu/bitcoin.core.imp
• src/Makefile.am
• src/Makefile.test_fuzz.include
• src/bitcoin-chainstate.cpp
• src/bitcoind.cpp
• src/deploymentstatus.h
• src/evo/mnhftx.cpp
• src/fs.h
• src/init.cpp
• src/init.h
• src/init/common.cpp
• src/init/common.h
• src/kernel/chainstatemanager_opts.h
• src/kernel/checks.cpp
• src/kernel/checks.h
• src/kernel/context.cpp
• src/kernel/context.h
• src/kernel/mempool_limits.h
• src/kernel/mempool_options.h
• src/kernel/mempool_persist.cpp
• src/kernel/mempool_persist.h
• src/mempool_args.cpp
• src/mempool_args.h
• src/net_processing.cpp
• src/net_processing.h
• src/node/blockstorage.cpp
• src/node/blockstorage.h
• src/node/chainstate.cpp
• src/node/chainstate.h
• src/node/context.cpp
• src/node/context.h
• src/node/interfaces.cpp
• src/node/mempool_persist_args.cpp
• src/node/mempool_persist_args.h
• src/node/miner.cpp
• src/node/miner.h
• src/policy/fees.cpp
• src/policy/fees.h
• src/policy/fees_args.cpp
• src/policy/fees_args.h
• src/policy/packages.h
• src/policy/policy.h
• src/rest.cpp
• src/rpc/blockchain.cpp
• src/rpc/fees.cpp
• src/rpc/mempool.cpp
• src/rpc/mining.cpp
• src/rpc/rawtransaction.cpp
• src/rpc/txoutproof.cpp
• src/test/block_reward_reallocation_tests.cpp
• src/test/blockfilter_index_tests.cpp
• src/test/denialofservice_tests.cpp
• src/test/dynamic_activation_thresholds_tests.cpp
• src/test/evo_deterministicmns_tests.cpp
• src/test/fuzz/mempool_utils.h
• src/test/fuzz/policy_estimator.cpp
• src/test/fuzz/policy_estimator_io.cpp
• src/test/fuzz/tx_pool.cpp
• src/test/fuzz/validation_load_mempool.cpp
• src/test/mempool_tests.cpp
• src/test/miner_tests.cpp
• src/test/net_peer_connection_tests.cpp
• src/test/txpackage_tests.cpp
• src/test/util/mining.cpp
• src/test/util/setup_common.cpp
• src/test/util/setup_common.h
• src/test/util/txmempool.cpp
• src/test/util/txmempool.h
• src/test/validation_block_tests.cpp
• src/test/validation_chainstate_tests.cpp
• src/test/validation_chainstatemanager_tests.cpp
• src/txmempool.cpp
• src/txmempool.h
• src/util/system.cpp
• src/util/system.h
• src/validation.cpp
• src/validation.h
• test/functional/mempool_expiry.py
• test/functional/mempool_persist.py
• test/lint/lint-circular-dependencies.py
• test/lint/spelling.ignore-words.txt

💤 Files with no reviewable changes (2)

• src/init/common.cpp
• src/init/common.h

✅ Files skipped from review due to trivial changes (9)

• test/lint/spelling.ignore-words.txt
• src/Makefile.test_fuzz.include
• contrib/devtools/iwyu/bitcoin.core.imp
• src/mempool_args.h
• src/test/validation_chainstate_tests.cpp
• src/kernel/mempool_persist.h
• ci/dash/lint-tidy.sh
• src/node/context.cpp
• src/evo/mnhftx.cpp

🚧 Files skipped from review as they are similar to previous changes (52)

• src/policy/packages.h
• src/test/miner_tests.cpp
• src/policy/fees_args.cpp
• src/node/blockstorage.h
• src/kernel/context.h
• src/rpc/fees.cpp
• src/rpc/rawtransaction.cpp
• src/test/net_peer_connection_tests.cpp
• src/test/blockfilter_index_tests.cpp
• src/test/util/txmempool.h
• src/test/block_reward_reallocation_tests.cpp
• src/test/validation_block_tests.cpp
• src/test/evo_deterministicmns_tests.cpp
• src/fs.h
• test/functional/mempool_expiry.py
• src/net_processing.h
• src/kernel/mempool_options.h
• src/policy/fees.h
• src/rest.cpp
• src/test/txpackage_tests.cpp
• src/node/miner.h
• src/bitcoind.cpp
• src/test/util/mining.cpp
• src/test/mempool_tests.cpp
• src/kernel/chainstatemanager_opts.h
• src/deploymentstatus.h
• src/test/validation_chainstatemanager_tests.cpp
• src/rpc/txoutproof.cpp
• src/init.h
• src/test/denialofservice_tests.cpp
• test/lint/lint-circular-dependencies.py
• src/node/mempool_persist_args.h
• test/functional/mempool_persist.py
• src/node/miner.cpp
• src/util/system.cpp
• src/test/fuzz/mempool_utils.h
• src/validation.h
• src/test/util/setup_common.h
• configure.ac
• src/util/system.h
• src/rpc/blockchain.cpp
• src/policy/fees.cpp
• src/txmempool.cpp
• src/node/chainstate.h
• src/node/chainstate.cpp
• src/net_processing.cpp
• src/policy/policy.h
• src/bitcoin-chainstate.cpp
• src/rpc/mining.cpp
• src/node/interfaces.cpp
• src/txmempool.h
• src/init.cpp

[coderabbitai]

⚠️ Potential issue | 🟡 Minor

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Verify whether GetLoadTried means "attempted load" or "successfully loaded",
# and whether getmempoolinfo docs match that behavior.

rg -n -C3 '\bGetLoadTried\b|\bSetLoadTried\b|\bIsLoaded\b|\bLoadMempool\b' src/txmempool.cpp src/txmempool.h src/validation.cpp src/validation.h src/rpc/mempool.cpp

Repository: dashpay/dash

Length of output: 4007

---

Semantic mismatch: loaded field reports load attempt status, not successful completion.

The loaded field in the RPC response now returns GetLoadTried(), which is documented as "true if we've made an attempt to load the mempool regardless of whether the attempt was successful or not." This semantics differs from the field name—consider renaming to load_attempted or updating the RPC schema documentation to clarify that it reports attempt status rather than success.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/rpc/mempool.cpp` around lines 694 - 695, The field name "loaded" in the
mempool RPC response at the pushKV call does not match the semantics of
GetLoadTried(), which returns whether an attempt to load was made, not whether
it succeeded. Either rename the field from "loaded" to a name that reflects
attempt status such as "load_attempted", or update the RPC schema documentation
to clarify that this field indicates an attempt was made to load the mempool
regardless of success. Choose the approach that aligns with your API
documentation standards.

[thepastaclaw]

Code Review

Latest delta (461435..04afe98) resolves three of the six prior findings (configure.ac conditional regression, stale init::UnsetGlobals comment, unused get_unix_time_seconds parameter). The three remaining carry-forward findings still reproduce at 04afe98: stale global LoadMempool/DumpMempool declarations in validation.h, duplicate common/bloom.h line in Makefile.am, and unreachable SanityCheckError::ERROR_CHRONO enum value. No new defects introduced by this push.

🟡 1 suggestion(s)

1 additional finding(s) omitted (not in diff).

2 carried-forward finding(s) already raised on this PR; not re-posting as new inline comments.

🤖 Prompt for all review comments with AI agents

These findings are from an automated code review. Verify each finding against the current code and only fix it if needed.

In `src/validation.h`:
- [SUGGESTION] src/validation.h:1110-1116: Stale global LoadMempool/DumpMempool declarations left behind after move to kernel::
  Carried forward — STILL VALID at 04afe98956. validation.h:1110-1116 still declares `using FopenFn = std::function<FILE*(const fs::path&, const char*)>;` plus `DumpMempool(const CTxMemPool&, FopenFn, bool)` and `LoadMempool(CTxMemPool&, CChainState&, FopenFn)` in the global namespace. bitcoin#25487 moved these into kernel:: with new path-taking signatures (definitions now live in src/kernel/mempool_persist.cpp); the upstream hunk removed these prototypes from validation.h. In-tree callers correctly use `kernel::DumpMempool` / `kernel::LoadMempool`, so the build is currently fine, but the old globals remain advertised with no matching definitions. A future translation unit that includes validation.h and calls `LoadMempool(pool, chainstate)` will compile cleanly against the stale prototypes and fail at link time with an undefined reference instead of producing a clear compile error. Drop the dead declarations plus the now-unused `FopenFn` alias, and have any callers include kernel/mempool_persist.h directly.

[thepastaclaw]

Code Review

One blocking build break carried forward verified: init.cpp:1489 still references kernel::SanityCheckError::ERROR_CHRONO after the latest delta dropped that enumerator from kernel/checks.h. Switch has no default and operates on a scoped enum, so compilation fails. Prior validation.h stale global LoadMempool/DumpMempool declarations (bitcoin#25487 residue) remain STILL VALID and are carried forward as a suggestion. Prior Makefile common/bloom.h duplicate finding is resolved. CodeRabbit's two items (Makefile .l$(OBJEXT) typo and 'loaded' field naming) are false positives — the libtool .lo extension is correct, and the field name preserves upstream behavior.

🔴 1 blocking | 🟡 1 suggestion(s)

1 additional finding(s) omitted (not in diff).

🤖 Prompt for all review comments with AI agents

These findings are from an automated code review. Verify each finding against the current code and only fix it if needed.

In `src/init.cpp`:
- [BLOCKING] src/init.cpp:1489-1491: Build break: switch references removed enumerator kernel::SanityCheckError::ERROR_CHRONO
  The latest push removed `ERROR_CHRONO` from `enum class SanityCheckError` in `src/kernel/checks.h` (now only `ERROR_ECC`, `ERROR_BLS`, `ERROR_RANDOM`), but `src/init.cpp:1489-1491` still contains `case kernel::SanityCheckError::ERROR_CHRONO:`. The switch is annotated `// no default case, so the compiler can warn about missing cases`, and `SanityCheckError` is a scoped enum, so the case label naming a non-existent enumerator is a hard compile error (`'ERROR_CHRONO' is not a member of 'kernel::SanityCheckError'`), not just an unreachable-code warning. The cleanup adapting bitcoin#25065's chrono residue (Dash uses `BLSInit()` rather than `ChronoSanityCheck()`) landed in the enum but not in the consumer. Remove the stale case arm to restore the build.

In `src/validation.h`:
- [SUGGESTION] src/validation.h:1110-1116: Stale global LoadMempool/DumpMempool declarations remain after move to kernel:: (bitcoin#25487)
  Carried forward — STILL VALID at 586000ee. `src/validation.h:1110-1116` still declares `using FopenFn = std::function<FILE*(const fs::path&, const char*)>;` plus the legacy `DumpMempool(const CTxMemPool&, FopenFn, bool)` and `LoadMempool(CTxMemPool&, CChainState&, FopenFn)` prototypes in the global namespace. bitcoin#25487 moved these APIs into `kernel::` with path-taking signatures (definitions in `src/kernel/mempool_persist.cpp`, declarations in `src/kernel/mempool_persist.h`), and upstream removed the old prototypes from `validation.h`. All in-tree callers use `kernel::DumpMempool` / `kernel::LoadMempool` (via `using kernel::...;` in `init.cpp`, `validation.cpp`, `rpc/mempool.cpp`, fuzz harness) or the `CChainState::LoadMempool` member, so there are no in-tree definitions matching the stale prototypes. The dangling declarations are a silent linker trap: any future TU including `validation.h` that calls `LoadMempool(pool, chainstate)` will compile against the stale prototype and fail at link with an undefined reference instead of producing a clear compile error. Drop the dead declarations and the unused `FopenFn` alias; have remaining consumers include `kernel/mempool_persist.h` directly.

[thepastaclaw]

🔴 Blocking: Build break: switch references removed enumerator kernel::SanityCheckError::ERROR_CHRONO

The latest push removed ERROR_CHRONO from enum class SanityCheckError in src/kernel/checks.h (now only ERROR_ECC, ERROR_BLS, ERROR_RANDOM), but src/init.cpp:1489-1491 still contains case kernel::SanityCheckError::ERROR_CHRONO:. The switch is annotated // no default case, so the compiler can warn about missing cases, and SanityCheckError is a scoped enum, so the case label naming a non-existent enumerator is a hard compile error ('ERROR_CHRONO' is not a member of 'kernel::SanityCheckError'), not just an unreachable-code warning. The cleanup adapting bitcoin#25065's chrono residue (Dash uses BLSInit() rather than ChronoSanityCheck()) landed in the enum but not in the consumer. Remove the stale case arm to restore the build.

Suggested change

	case kernel::SanityCheckError::ERROR_CHRONO:
	InitError(Untranslated("Clock epoch mismatch. Aborting."));
	break;
	case kernel::SanityCheckError::ERROR_RANDOM:
	InitError(Untranslated("OS cryptographic RNG sanity check failure. Aborting."));
	break;
	} // no default case, so the compiler can warn about missing cases

_{source: ['claude', 'codex']}

[thepastaclaw]

Code Review

Latest push (586000e → 2d14616) is a one-line cleanup that removes the stale case kernel::SanityCheckError::ERROR_CHRONO: arm in AppInitSanityChecks, resolving the prior blocking build break. The other prior finding — dangling pre-bitcoin#25487 global DumpMempool/LoadMempool prototypes and the FopenFn alias in src/validation.h — is still present at head and is carried forward as a suggestion. No new in-scope issues in the delta.

🟡 1 suggestion(s)

1 additional finding(s) omitted (not in diff).

🤖 Prompt for all review comments with AI agents

These findings are from an automated code review. Verify each finding against the current code and only fix it if needed.

In `src/validation.h`:
- [SUGGESTION] src/validation.h:1110-1116: Drop stale global LoadMempool/DumpMempool declarations left over from bitcoin#25487 (carried forward — STILL VALID)
  Carried forward from prior review at 586000ee and re-verified at 2d146167. `src/validation.h:1110-1116` still declares `using FopenFn = std::function<FILE*(const fs::path&, const char*)>;` plus the legacy global prototypes `bool DumpMempool(const CTxMemPool&, FopenFn, bool)` and `bool LoadMempool(CTxMemPool&, CChainState&, FopenFn)`. bitcoin#25487 moved these APIs into `namespace kernel` with path-taking signatures: declarations in `src/kernel/mempool_persist.h:16,21` and definitions in `src/kernel/mempool_persist.cpp:40,135` (both take `const fs::path&`). A repo-wide grep confirms no translation unit defines a global `::DumpMempool`/`::LoadMempool` matching the stale signatures; every caller routes through `using kernel::...;` (`src/init.cpp:155`, `src/validation.cpp:76`, `src/rpc/mempool.cpp:30`, `src/test/fuzz/validation_load_mempool.cpp:23`) or the `CChainState::LoadMempool` member. The dangling declarations are a silent linker trap: any future TU that includes only `validation.h` and calls `LoadMempool(pool, chainstate)` or `DumpMempool(pool)` without `using kernel::...;` will compile against the stale prototype and fail at link time with an undefined reference instead of getting a clear compile-time error. Upstream removed these prototypes from `validation.h` as part of bitcoin#25487; the Dash backport should do the same and drop the unused `FopenFn` alias. Remaining consumers should include `kernel/mempool_persist.h` directly. Non-blocking since current callers link correctly, but worth cleaning up before it traps a future contributor.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2d14616718

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Initialize BlockManager from owned chain params

When a ChainstateManager is created with params that do not have process-lifetime storage (for example, a kernel embedding builds a local CChainParams and then releases it), this leaves m_blockman holding a reference to opts.chainparams while m_chainparams has already copied it. Later BlockManager::GetConsensus() dereferences the stale reference even though ChainstateManager::GetParams() uses the owned copy, so initialize m_blockman from m_chainparams instead.

Useful? React with 👍 / 👎.

[coderabbitai]

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (1)

src/validation.cpp (1)

4342-4379: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Restore bls_legacy_scheme on every exit path.

The saved value is restored only on the success path. Any early return above Line 4376, including the adjusted-time contextual header failure at Line 4364, can leave the global BLS scheme changed for later validation.

Proposed RAII fix

-    auto bls_legacy_scheme = bls::bls_legacy_scheme.load();
+    const auto bls_legacy_scheme = bls::bls_legacy_scheme.load();
+    struct BLSLegacySchemeRestorer {
+        decltype(bls::bls_legacy_scheme.load()) original;
+
+        ~BLSLegacySchemeRestorer() noexcept
+        {
+            if (original != bls::bls_legacy_scheme.load()) {
+                bls::bls_legacy_scheme.store(original);
+                LogPrintf("%s: bls_legacy_scheme=%d\n", "TestBlockValidity", bls::bls_legacy_scheme.load());
+            }
+        }
+    } restore_bls_legacy_scheme{bls_legacy_scheme};
@@
-    // we could switch to another scheme while testing, switch back to the original one
-    if (bls_legacy_scheme != bls::bls_legacy_scheme.load()) {
-        bls::bls_legacy_scheme.store(bls_legacy_scheme);
-        LogPrintf("%s: bls_legacy_scheme=%d\n", __func__, bls::bls_legacy_scheme.load());
-    }
-
     return true;

If you prefer not to address it in this backport PR, I can help open a follow-up issue for the existing TODO.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/validation.cpp` around lines 4342 - 4379, The bls_legacy_scheme value is
saved at the beginning of the function but only restored at the end on the
success path. Any early return from the validation checks
(ContextualCheckBlockHeader, CheckBlock, ContextualCheckBlock, or ConnectBlock)
will exit before reaching the restoration code, leaving the global BLS scheme
modified. Use an RAII pattern (such as a scope guard or lambda-based finally
mechanism) to wrap the restoration logic so that the saved bls_legacy_scheme
value is restored on every exit path, including early returns from error cases.

♻️ Duplicate comments (1)

src/rpc/mempool.cpp (1)

759-770: ⚠️ Potential issue | 🟡 Minor

Update savemempool help text and error message to align with attempted-load semantics.

Line 759 guards on GetLoadTried(), which returns true when a load attempt has been made—regardless of success. However, the help text (line 743) says "fully loaded" and the error message (line 760) says "not loaded yet," both implying successful completion. Since no successful-load state predicate exists in the codebase, the intended contract is attempted-load completion. Update the wording:

Wording fix

-        "\nDumps the mempool to disk. It will fail until the previous dump is fully loaded.\n",
+        "\nDumps the mempool to disk. It will fail until the previous dump load attempt has completed.\n",
...
-        throw JSONRPCError(RPC_MISC_ERROR, "The mempool was not loaded yet");
+        throw JSONRPCError(RPC_MISC_ERROR, "The mempool load attempt has not completed yet");

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/rpc/mempool.cpp` around lines 759 - 770, The help text for the
savemempool RPC method and the error message in the guard condition use
misleading language about mempool loading. The code guards on GetLoadTried()
which returns true when a load attempt has been made (regardless of success),
but the help text (around line 743) says "fully loaded" and the error message at
line 760 says "not loaded yet," both implying successful completion. Update the
help text to clarify that the mempool must have been attempted to load, and
update the error message to say something like "The mempool load attempt has not
yet completed" to align with the attempted-load semantics of GetLoadTried().

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Outside diff comments:
In `@src/validation.cpp`:
- Around line 4342-4379: The bls_legacy_scheme value is saved at the beginning
of the function but only restored at the end on the success path. Any early
return from the validation checks (ContextualCheckBlockHeader, CheckBlock,
ContextualCheckBlock, or ConnectBlock) will exit before reaching the restoration
code, leaving the global BLS scheme modified. Use an RAII pattern (such as a
scope guard or lambda-based finally mechanism) to wrap the restoration logic so
that the saved bls_legacy_scheme value is restored on every exit path, including
early returns from error cases.

---

Duplicate comments:
In `@src/rpc/mempool.cpp`:
- Around line 759-770: The help text for the savemempool RPC method and the
error message in the guard condition use misleading language about mempool
loading. The code guards on GetLoadTried() which returns true when a load
attempt has been made (regardless of success), but the help text (around line
743) says "fully loaded" and the error message at line 760 says "not loaded
yet," both implying successful completion. Update the help text to clarify that
the mempool must have been attempted to load, and update the error message to
say something like "The mempool load attempt has not yet completed" to align
with the attempted-load semantics of GetLoadTried().

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 67c18cba-54f6-42b8-8b96-51389942b77e

📥 Commits

Reviewing files that changed from the base of the PR and between 586000e and 2d14616.

📒 Files selected for processing (59)

• ci/dash/lint-tidy.sh
• contrib/devtools/iwyu/bitcoin.core.imp
• src/Makefile.am
• src/Makefile.test_fuzz.include
• src/bitcoin-chainstate.cpp
• src/bitcoind.cpp
• src/fs.h
• src/init.cpp
• src/init.h
• src/init/common.cpp
• src/init/common.h
• src/kernel/chainstatemanager_opts.h
• src/kernel/checks.cpp
• src/kernel/checks.h
• src/kernel/context.cpp
• src/kernel/context.h
• src/kernel/mempool_limits.h
• src/kernel/mempool_options.h
• src/kernel/mempool_persist.cpp
• src/kernel/mempool_persist.h
• src/mempool_args.cpp
• src/mempool_args.h
• src/node/blockstorage.cpp
• src/node/blockstorage.h
• src/node/context.cpp
• src/node/context.h
• src/node/interfaces.cpp
• src/node/mempool_persist_args.cpp
• src/node/mempool_persist_args.h
• src/policy/fees.cpp
• src/policy/fees.h
• src/policy/fees_args.cpp
• src/policy/fees_args.h
• src/policy/packages.h
• src/policy/policy.h
• src/rpc/fees.cpp
• src/rpc/mempool.cpp
• src/test/evo_deterministicmns_tests.cpp
• src/test/fuzz/mempool_utils.h
• src/test/fuzz/policy_estimator.cpp
• src/test/fuzz/policy_estimator_io.cpp
• src/test/fuzz/tx_pool.cpp
• src/test/fuzz/validation_load_mempool.cpp
• src/test/mempool_tests.cpp
• src/test/txpackage_tests.cpp
• src/test/util/setup_common.cpp
• src/test/util/setup_common.h
• src/test/util/txmempool.cpp
• src/test/util/txmempool.h
• src/txmempool.cpp
• src/txmempool.h
• src/util/system.cpp
• src/util/system.h
• src/validation.cpp
• src/validation.h
• test/functional/mempool_expiry.py
• test/functional/mempool_persist.py
• test/lint/lint-circular-dependencies.py
• test/lint/spelling.ignore-words.txt

💤 Files with no reviewable changes (2)

• src/init/common.cpp
• src/init/common.h

✅ Files skipped from review due to trivial changes (5)

• contrib/devtools/iwyu/bitcoin.core.imp
• src/node/context.cpp
• src/Makefile.test_fuzz.include
• test/functional/mempool_expiry.py
• src/test/evo_deterministicmns_tests.cpp

🚧 Files skipped from review as they are similar to previous changes (37)

• ci/dash/lint-tidy.sh
• src/mempool_args.h
• test/lint/lint-circular-dependencies.py
• src/fs.h
• src/node/blockstorage.cpp
• src/kernel/mempool_persist.h
• src/rpc/fees.cpp
• src/init.h
• src/bitcoind.cpp
• src/test/fuzz/mempool_utils.h
• test/lint/spelling.ignore-words.txt
• src/test/mempool_tests.cpp
• src/node/blockstorage.h
• src/policy/packages.h
• src/test/txpackage_tests.cpp
• src/test/util/setup_common.h
• src/kernel/mempool_options.h
• src/policy/fees_args.cpp
• src/policy/fees_args.h
• src/kernel/chainstatemanager_opts.h
• src/kernel/checks.h
• src/policy/fees.h
• src/kernel/context.h
• src/policy/policy.h
• src/policy/fees.cpp
• src/util/system.h
• src/node/context.h
• src/util/system.cpp
• test/functional/mempool_persist.py
• src/test/util/txmempool.h
• src/node/mempool_persist_args.h
• src/txmempool.cpp
• src/validation.h
• src/txmempool.h
• src/init.cpp
• src/bitcoin-chainstate.cpp
• src/Makefile.am