replay: verify Alpenglow block sequence

[ripatel-fd]

Agave has a state machine verifying that the pieces of an
Alpenglow block appear in a particular order

[github-actions]

Performance Measurements ⏳

Suite	Baseline	New	Change
backtest mainnet-424669000-perf per slot	0.049702 s	0.049703 s	0.002% ✅
backtest mainnet-424669000-perf snapshot load	2.617 s	2.638 s	0.802% ✅
backtest mainnet-424669000-perf total elapsed	64.413451 s	64.414617 s	0.002% ✅
firedancer mem usage with mainnet.toml	173.41 GiB	173.41 GiB	0.000% ✅

[Copilot]

🟡 Changes recommended

Critical handling gaps reject or crash on protocol-valid Alpenglow components.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

Adds replay validation for Alpenglow block-component ordering and reports malformed sequences through replay events.

Changes:

• Adds scheduler state-machine validation and a new dead reason.
• Extends block-completion event schemas and generated constants.
• Adds valid and invalid sequence tests.

File summaries

File	Review
src/discof/replay/test_sched.c	Adds ordering tests. Nit (1 vote): Missing coverage for rejecting transaction-bearing alpenticks.
src/discof/replay/fd_sched.h	Defines the new dead reason. No issues found.
src/discof/replay/fd_sched.c	Implements sequence validation. Critical (2 votes): Valid genesis-certificate markers cannot be accepted. Critical (1 vote): Valid UPDATE_PARENT components terminate replay instead of reparenting/resetting state.
src/discof/replay/fd_replay_tile.c	Maps the dead reason to replay events. No issues found.
src/disco/events/schema/events.proto	Extends the event enum. No issues found.
src/disco/events/schema/block_completed.json	Documents the new completion reason. No issues found.
src/disco/events/generated/fd_event_gen.h	Adds the generated event constant. No issues found.

Review details

Suppressed comments (1)

src/discof/replay/test_sched.c:768

• The invalid-alpentick cases cover mblk_cnt!=1 and hash_cnt!=1, but not the new txn_cnt!=0 rejection. Add a one-microblock transaction-entry component after the footer so a regression that accepts a transaction-bearing “alpentick” is caught.

  CASE( "wide entry after footer", BAD, AG_HEADER, AG_FOOTER, AG_WIDE_ENTRY );
  CASE( "fat entry after footer",  BAD, AG_HEADER, AG_FOOTER, AG_FAT_ENTRY  );

• Files reviewed: 6/7 changed files
• Comments generated: 2
• Review effort level: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.