refactor: msg pool to make more structured

[akaladarshi]

Summary of changes

Changes introduced in this pull request:

• Part 1 one of the refactor of the msg_pool, this mostly contains the structural refactor such as combining methods and creating new files etc.:
  • Introduced a new pending_store that holds all the pending related data for the mempool
  • Moved the mset related changes into it's own file to reduce the bloat on msgpool

Reference issue to close (if applicable)

Closes

Other information and links

Change checklist

• I have performed a self-review of my own code,
• I have made corresponding changes to the documentation. All new code adheres to the team's documentation standards,
• I have added tests that prove my fix is effective or that my feature works (if possible),
• I have made sure the CHANGELOG is up-to-date. All user-facing changes should be reflected in this document.

Outside contributions

• I have read and agree to the CONTRIBUTING document.
• I have read and agree to the AI Policy document. I understand that failure to comply with the guidelines will lead to rejection of the pull request.

Summary by CodeRabbit

• New Features

  • Real-time pending-message events and a subscription API to receive add/remove notifications.

• Improvements

  • Replaced in-memory pending map with a snapshot-capable store for safer concurrent access and more efficient reads.
  • Tighter, clearer replace-by-fee and nonce-gap handling for more robust transaction behavior and capacity enforcement.

• Tests

  • Updated and expanded tests for snapshots, event emission, capacity/RBF, and nonce-gap scenarios.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: cd0c8ac7-e135-416b-9500-194e66a696ef

📥 Commits

Reviewing files that changed from the base of the PR and between 48cdcbe and 0d66af3.

📒 Files selected for processing (3)

• src/message_pool/msgpool/events.rs
• src/message_pool/msgpool/mod.rs
• src/message_pool/msgpool/msg_set.rs

🚧 Files skipped from review as they are similar to previous changes (1)

• src/message_pool/msgpool/events.rs

---

Walkthrough

Replaces the per-address in-memory pending HashMap with a crate-scoped PendingStore, extracts per-sender logic into MsgSet, and adds MpoolUpdate broadcast events; updates insertion/removal/read paths, threads pending_store through selection/republish/head-change, and adjusts tests to use snapshot APIs and event subscriptions.

Changes

Cohort / File(s)	Summary
Event Infrastructure src/message_pool/msgpool/events.rs	Adds MpoolUpdate enum (Add, Remove) and MPOOL_UPDATE_CHANNEL_CAPACITY constant for pending-pool broadcast events.
Pending Store src/message_pool/msgpool/pending_store.rs	Introduces PendingStore (crate-visible) with new, insert, remove, snapshot, snapshot_for, has_subscribers, and subscribe/broadcast emission semantics; emits MpoolUpdate on real mutations.
Per-sender Message Set src/message_pool/msgpool/msg_set.rs	Adds MsgSet, MsgSetLimits, StrictnessPolicy, MAX_NONCE_GAP, and add/rm logic implementing nonce-gap rules, gap-filling, replace-by-fee, capacity limits, and metrics updates.
MessagePool Core src/message_pool/msgpool/msg_pool.rs, src/message_pool/msgpool/mod.rs	Replaces pending: SyncRwLock<HashMap> with pending_store: PendingStore; threads pending_store through contexts; adds subscribe_to_updates(); tightens head_change visibility; updates callers to use pending_store APIs.
Selection & Republish src/message_pool/msgpool/selection.rs, src/message_pool/msgpool/mod.rs	Refactors selection and republish paths to use pending_store.snapshot() / snapshot_for(...); constructs MpoolCtx with pending_store; updates tests to assert via snapshots.
Exports & Tests src/message_pool/msgpool/...	Adds crate-visible submodules events, msg_set, pending_store; re-exports MpoolUpdate; updates tests to use PendingStore snapshots and subscription behavior; removes some MsgSet-specific tests now validated via store snapshots.

Sequence Diagram(s)

sequenceDiagram
    participant Client as "MessagePool API"
    participant Store as "PendingStore"
    participant MsgSet as "MsgSet (per-sender)"
    participant Sub as "Subscriber (broadcast::Receiver)"
    Client->>Store: insert(resolved_from, msg, sequence, trust, strictness)
    Store->>MsgSet: get_or_create(resolved_from)
    MsgSet-->>Store: add(msg, strictness, trust, limits) -> Result
    alt insert succeeded
        Store->>Store: persist into map
        Store->>Sub: broadcast MpoolUpdate::Add(msg)
    else insert failed
        Store-->>Client: return Error
    end

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~60 minutes

Possibly related issues

• fix: make run_head_change() side-effect free in message selection #6975 — head-change/selection now threads and mutates live PendingStore; this PR aligns with that concern by switching head-change to use PendingStore.

Possibly related PRs

• fix: match Lotus's behaviour with replace by fee in mpool #6535 — closely related changes to MsgSet::add RBF and capacity semantics.
• fix(mpool): fix flaky eth_newPendingTransactionFilter test #6946 — related refactor of MessagePool pending access patterns and snapshot behavior.
• fix(mpool): resolve_to_key error handling #6890 — related changes to address resolution and republish/pending handling paths.

Suggested reviewers

• sudo-shashank
• LesnyRumcajs
• hanabi1224

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately reflects the main change: a structural refactor of the message pool codebase that improves organization by introducing pending_store and separating concerns into new modules.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch akaladarshi/msgpool-refactor

✨ Simplify code

• Create PR with simplified code
• Commit simplified code in branch akaladarshi/msgpool-refactor

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 3

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@src/message_pool/msgpool/msg_pool.rs`:
- Around line 449-453: The public method subscribe_to_updates currently returns
broadcast::Receiver<MpoolUpdate> but MpoolUpdate is crate-private (declared
pub(in crate::message_pool) in events.rs), so callers outside
crate::message_pool cannot name the type; either re-export MpoolUpdate from a
public module (add a pub use crate::message_pool::msgpool::events::MpoolUpdate
in a public mod) so the type is publicly reachable, or reduce
subscribe_to_updates() visibility to match MpoolUpdate (make
subscribe_to_updates pub(crate) or pub(in crate::message_pool)) so the declared
signature and type visibility align; update the signature or re-export
accordingly and adjust any callers.

In `@src/message_pool/msgpool/msg_set.rs`:
- Around line 183-188: The applied=true branch in msg_set.rs only sets
self.next_sequence = sequence + 1 which can leave already-buffered higher nonces
unaccounted for; change the branch to advance next_sequence past any buffered
successors the same way the unknown-message branch does: set next_sequence =
sequence + 1 and then loop while self.msgs.contains_key(&self.next_sequence) (or
the equivalent buffered map/set used in this struct) incrementing next_sequence
until no buffered message exists, so callers are not told they can reuse a
still-pending nonce.

In `@src/message_pool/msgpool/selection.rs`:
- Around line 853-857: The simulated path run_head_change() is mutating the live
PendingStore because MpoolCtx is constructed with the real pending_store; change
the helper to be side-effect free by giving MpoolCtx a simulation-only context
or snapshot: either construct MpoolCtx with a read-only/snapshot PendingStore
clone or introduce a SimulationPendingStore (or a flag on MpoolCtx) that makes
remove_from_selected_msgs() operate only on the local rmsgs/result map rather
than calling pending_store.remove(...). Update the MpoolCtx construction in
run_head_change() and ensure MpoolCtx::remove_from_selected_msgs() checks the
simulation mode or uses the snapshot store so select_messages() against
non-current tipsets cannot delete entries from the live mempool.

🪄 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: 3914d600-4744-47e8-8b77-311223afc188

📥 Commits

Reviewing files that changed from the base of the PR and between 65c39a0 and c515b49.

📒 Files selected for processing (6)

• src/message_pool/msgpool/events.rs
• src/message_pool/msgpool/mod.rs
• src/message_pool/msgpool/msg_pool.rs
• src/message_pool/msgpool/msg_set.rs
• src/message_pool/msgpool/pending_store.rs
• src/message_pool/msgpool/selection.rs

[codecov]

❌ 1 Tests Failed:

Tests completed	Failed	Passed	Skipped
2044	1	2043	0

View the top 1 failed test(s) by shortest run time

forest-filecoin::lint::lint

Stack Traces | 1.29s run time

thread 'lint' (47948) panicked at tests/lint.rs:174:17:
found 1 violations in 633 files
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}

[LesnyRumcajs]

Let's ease up on the comments; ideally documentation should explain things high-level, and not repeat the implementation.

[akaladarshi]

Done.

[LesnyRumcajs]

@akaladarshi Anything with 🐰 comments to address? If not, mark them as resolved.