fix(seidb-bench): stop the state-store bench backends fail on a nil config

[blindchaser]

Summary

runBenchmark hands NewDBImpl a nil dbConfig for every backend, but the SSComposite, SSHistoricalOffload and CompositeDual_SSComposite cases still asserted dbConfig.(*T) in single-value form. That panics on a nil interface, so BenchmarkSSCompositeWrite, BenchmarkSSHistoricalOffloadWrite and BenchmarkCombinedCompositeDualSSCompositeWrite crashed at startup with interface conversion: interface {} is nil, not *config.StateStoreConfig — the same failure #3770 fixed for FlatKV, in the same switch.

Copying that fix case-by-case wouldn't be enough: with only a comma-ok assertion the crash moves one frame later, since both SS Composite constructors dereference *ssConfig. And this is now the third instance of the same bare-assertion bug in this switch, so the fix collapses config extraction into one place rather than patching instance three.

• wrappers/db_implementations.go: backendConfig[T] becomes the single path from the untyped dbConfig to a backend's typed config. It treats nil as ordinary — each constructor supplies its own default — and rejects only a wrong type, now naming the type it wanted. All five config-taking cases route through it, so a new backend can't reintroduce the bare assertion. newSSCompositeStateStore and newCombinedCompositeDualSSComposite fall back to DefaultBenchStateStoreConfig(), matching the fallback the MemIAVL and FlatKV constructors already have. SSHistoricalOffload deliberately keeps no default: its stream needs brokers only the caller knows, so nil reaches HistoricalOffloadConfig.Validate, which already reports historical offload config is required.

Benchmark-only change; no production code path touched.

Test plan

• go test ./sei-db/state_db/bench/wrappers — wrappers_test.go gains two table tests over every config-taking backend: nil resolves to the default and opens, a wrong type errors with the backend named. Plus an offload-specific case asserting the missing config is reported rather than panicked. The two FlatKV-only tests from fix(seidb-bench): tolerate nil FlatKV config in bench wrapper factory #3770 are folded into those tables.
• Before, on main: BenchmarkSSCompositeWrite → panic: interface conversion: interface {} is nil, not *config.StateStoreConfig
• After, both benchmarks run to completion:
  BenchmarkSSCompositeWrite/100_keys_per_block-11 1 32137667 ns/op 311161 keys/sec
BenchmarkCombinedCompositeDualSSCompositeWrite/100_keys_per_block-11 1 174587333 ns/op 57278 keys/sec
• After, BenchmarkSSHistoricalOffloadWrite fails with failed to create historical offload stream: historical offload config is required instead of panicking (it needs a live Kafka broker to actually run)
• go vet / gofmt -s / goimports clean

[cursor]

PR Summary

Low Risk
Changes are confined to sei-db benchmark wrappers and tests; production DB paths are untouched.

Overview
Fixes benchmark startup panics when runBenchmark passes a nil dbConfig into NewDBImpl for state-store backends (SSComposite, CompositeDual_SSComposite, and related cases that used bare dbConfig.(*T) assertions).

Introduces generic backendConfig[T] so nil config is passed through (constructors apply defaults) and only wrong types return a named error. newSSCompositeStateStore and newCombinedCompositeDualSSComposite now use DefaultBenchStateStoreConfig() when config is nil, matching MemIAVL/FlatKV. SSHistoricalOffload still has no default; nil config surfaces historical offload config is required from validation instead of panicking.

Tests are broadened to table-driven coverage: nil opens for backends with defaults, invalid types error per backend, and offload nil is an error not a panic.

^{Reviewed by Cursor Bugbot for commit f9b74de. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	✅ passed	Aug 1, 2026, 2:28 AM

[seidroid]

Benchmark-only fix that correctly collapses three panicking bare type assertions in NewDBImpl into one generic backendConfig[T] helper, with nil-config defaults added to the two SS Composite constructors and table tests covering all five config-taking backends. I found no correctness or security issues; only a minor test-coverage gap and one consistency nit.

Findings: 0 blocking | 4 non-blocking | 1 posted inline

Blockers

• None at the file/PR level.

Non-blocking

• cursor-review.md is empty — the Cursor pass produced no output, so this review reflects only the Claude and Codex passes (Codex reported no material issues).
• The non-config backends (NoOp, CompositeDual, CompositeSplit, CompositeCosmos) still silently discard a non-nil dbConfig. Now that config extraction is centralized in backendConfig, it would be cheap and consistent to reject a supplied config for those cases too, so a misrouted config surfaces as an error instead of being ignored. Non-blocking, pre-existing behavior.
• Pre-existing, worth noting since backendConfig now returns the caller's pointer verbatim: newFlatKVCommitStore mutates the caller's config (config.DataDir = dbDir) and HistoricalOffloadConfig.Validate mutates the caller's Kafka sub-config via applyDefaults. Harmless for bench single-open usage, but a copy-on-entry would make the helper's contract fully read-only.
• 1 suggestion(s)/nit(s) flagged inline on specific lines.

[seidroid]

[suggestion] The two tables cover the nil path and the wrong-type path, but nothing asserts the third branch of backendConfig — that a correctly-typed non-nil config is passed through and actually honored rather than silently replaced by the backend default. A small case (e.g. open SSComposite with a *config.StateStoreConfig whose EVMSplit/AsyncWriteBuffer differs from DefaultBenchStateStoreConfig(), and assert the store reflects it) would lock in the behavior cryptosim.go:146 depends on when the user supplies a config.

[claude]

LGTM — benchmark-only fix, well-tested.

Extended reasoning...

Overview

Two-file change confined to sei-db/state_db/bench/wrappers: a benchmark harness that instantiates DB backends for micro-benchmarking. The PR generalizes the fix from #3770 (nil-config panic on FlatKV) into a single backendConfig[T] generic helper used by all five config-taking cases in NewDBImpl's switch, replacing bare dbConfig.(*T) type assertions that panic on nil interfaces.

Security risks

None. This code only runs under go test/go bench in the benchmark package; it is never invoked from any production binary or RPC path, and touches no auth, crypto, or user-facing input.

Level of scrutiny

Low. Non-production, test/benchmark-only code with a narrow, mechanical fix pattern (replace panic-prone assertions with a helper that treats nil as valid and wrong-type as an error). I verified the one subtlety called out in the description — HistoricalOffloadConfig.Validate() has a nil receiver check (if c == nil { return fmt.Errorf("historical offload config is required") }) — matches the code in historical_offload_wrapper.go, so the claimed non-panic behavior for that backend is correct.

Other factors

Test coverage is thorough: table-driven tests cover nil-config defaulting for every backend that has a default, wrong-type rejection for every backend, and the one backend (SSHistoricalOffload) that intentionally has no default. The diff is small, self-contained, and matches an established prior-fix pattern (#3770) rather than introducing new design decisions.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 60.40%. Comparing base (438cc04) to head (f9b74de).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3838      +/-   ##
==========================================
- Coverage   61.29%   60.40%   -0.90%     
==========================================
  Files        2351     2259      -92     
  Lines      197491   186867   -10624     
==========================================
- Hits       121060   112879    -8181     
+ Misses      65579    63988    -1591     
+ Partials    10852    10000     -852     

Flag	Coverage Δ
sei-db	70.41% <ø> (+1.04%)	⬆️
sei-db-state-db	?

Flags with carried forward coverage won't be shown. Click here to find out more.
see 97 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.