test(edr_provider): broaden solx provider-path stack-trace coverage

[nebasuke]

Adding wide test coverage for stack trace tests, along with some test refactoring for more re-use. This is a follow-up of #1572. I've tried to put tests into blocks to make it a bit easier to follow what's currently covered. I could split into files instead if that is easier.

Note that I've removed two tests as they were already covered by a more generic test, see also the test folding Claude section below.

This is the first PR of a planned few. I'm trying to do a structured expansion of stack-trace coverage. The idea here is that this functions both as increased coverage for solx/solc, and some of the pre-work for being able to migrate away from hardhat-tests.

Planned next, roughly in order:

1. Failing deployments: sending ether to a non-payable constructor, or deploying with wrong or truncated constructor arguments. The error is attributed by looking at where in the bytecode execution stopped, so solx's debug info can place it differently than solc's. This is currently untested.
2. Solc-specific pattern checks running on solx bytecode: a few error classifications work by recognizing short, hard-coded instruction sequences that solc happens to emit (for example "you called an address with no contract on it"). solx compiles the same Solidity to different instructions, but these checks still run on its traces. Nothing today verifies whether they match correctly, match wrongly, or simply never fire. Includes the plain "call an address with no contract behind it" case itself.
3. Two rarer error kinds: a revert with a custom error the decoder doesn't recognize, and panic 0x51 (calling a function variable that was never assigned). Both are reported through fallback code paths no test reaches today.
4. Errors travelling between contracts: a nested call failing because it ran out of gas, errors bubbling up through proxy contracts, calls that fail before the callee even runs, and deployments rejected for exceeding the contract size limit.
5. eth_call: the tests so far only cover failing transactions; failing read-only calls take a separate path that we should either cover or explicitly leave out of scope.

Claude summary

Rebased onto main now that #1577 (the DWARF attribution fixes + fixture infrastructure) is merged. This PR is tests only: it broadens the provider-path (JSON-RPC) stack-trace coverage for solx artifacts from 11 tests to 36, all against fixtures already on main — and organizes the file so the coverage is auditable against the inference pipeline instead of reading as a blob.

The provider path is what Hardhat 3's JS-test flow (hardhat test + ethers/viem) hits. #1425 covers the solidity-test-runner path via the JS parity sweep, but the sweep is CI-skipped until @nomicfoundation/hardhat-solx is published — so these are the trace assertions that actually run in CI today.

File layout mirrors the inference pipeline

solx_stack_trace.rs puts all helpers in one block up front, then one test section per edr_solidity inference stage, so reviewing the file is walking the pipeline:

Section	Inference path pinned	Tests
Provider plumbing	end-to-end smoke on the minimal Counter fixture	1
Pre-execution guards	infer_before_tracing_call_message: payability, missing function/fallback/receive, direct library calls	6
Calldata decoding	check_last_instruction: InvalidParamsError for truncated calldata	1
Revert/panic/custom	check_revert_or_invalid_opcode: six panic codes with statement anchors, custom-error argument decoding, revert-line discrimination between requires	8
Callstack reconstruction	frame push/pop, submessage splicing, filter_redundant_frames: cross-contract frames, external/internal/mutual recursion, internal helpers and libraries, linked external library (linkReferences + DELEGATECALL decode), fallback/receive bodies	9
Modifiers	fix_initial_modifier + SolxTraceStrategy attribution: plain, nested, cross-contract, bare-revert	4
Create path	creation-code debugInfo: reverting constructors, plain and via internal helper	2
Submessage checks	check_last_submessage: returndata-size errors (real callee and EOA target)	2
-O3 twins (the mode3_* tests)	three modifier scenarios recompiled at solx optimizer mode 3 (-O3; hardhat-solx defaults to -O1). Since solx 0.1.6 the -O1 DWARF maps compiler helpers to real statement lines, so only these more aggressively optimized artifacts still reach the fallback inference paths: walking back to the last statement when a revert is attributed to a declaration line, and recovering the failing function for unmapped bare reverts	3

Every test now asserts a source anchor (function/contract/statement line), not just the entry variant — e.g. each panic code pins the statement it's attributed to, and the dispatch guards pin the declaration they anchor at. The two call shapes where frame gains/losses are the point (cross-contract call, external recursion) pin the full rendered trace via assert_trace_shape. Line pins are goldens: a shifted anchor after a solx upgrade is expected drift (update the pin), a lost frame is a regression.

Deliberately not covered — location-free classification paths already pinned by the solc corpus, plus follow-up material: message-kind dispatch (precompiles, unrecognized contracts, contract-too-large), create-side guards, the solc-opcode-pattern rewrites in mapped_inline_internal_functions_heuristics, out-of-gas rewrites, proxy propagation, and the eth_call channel. bail_on_call_failure was dropped from the test provider config since nothing exercised it.

The external-recursion test already caught one real regression: a late refactor in #1425 dropped the per-strategy recursion carve-out in filter_redundant_frames, collapsing any external-recursion depth to a single frame. That fix was upstreamed separately as #1569.

Two tests folded, one renamed (no coverage removed)

The diff deletes two tests by name; both were variant-only assertions strictly subsumed by a stronger replacement:

• panic_error_variant_surfaces_for_overflow_scenario asserted only that some PanicError entry appears for OverflowTest — a weaker duplicate of the panic-code family right next to it. It's now panic_code_surfaces_for_arithmetic_overflow in that family, additionally pinning the code (0x11) and the statement the panic is attributed to (x = x + 1, line 26).
• custom_error_variant_surfaces_for_custom_error_scenario asserted only that some CustomError entry appears. It's now custom_error_decodes_name_and_args, additionally pinning the decoded message (reverted with custom error 'MyError(42, "custom error")') and the revert statement line — the known-selector argument-decoding path had no coverage before.
• The third test from the same old "variance-axis" block, revert_error_variant_surfaces_for_counter, is kept as revert_error_surfaces_end_to_end_for_counter: it's the only test on the minimal Counter fixture (a second, independent artifact assembly), so it stays as the plumbing smoke test.

Accepted gaps

Two SolxTraceStrategy fallbacks remain untested: unresolved_callstack_entry and the panic-helper selector fallback. Both are last resorts for instructions whose DWARF location doesn't resolve to a known function, and no fixture produces that — not even at -O3 (MutualRecursionTest, the closest candidate, reached parity without them). Auditing that solx covers every reachable PC with a DW_TAG_subprogram belongs in solx's own test suite, not EDR scenarios.

Verification

• solx_stack_trace.rs: 36/36 against the 0.1.6 fixtures (including the -O3 twins from fix(edr_solidity): solx stack-trace attribution for declaration-level and line-0 DWARF #1577).
• edr_provider integration harness: 216/216 (fork tests require ALCHEMY_URL).
• cargo +nightly fmt --check and cargo clippy -p edr_provider --tests --features test-utils clean.
• Test-only change: no changeset.

[changeset-bot]

⚠️ No Changeset found

Latest commit: 6ee957d

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 79.89%. Comparing base (bdcf6fb) to head (6ee957d).
⚠️ Report is 14 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1552      +/-   ##
==========================================
+ Coverage   79.77%   79.89%   +0.11%     
==========================================
  Files         453      452       -1     
  Lines       79038    78956      -82     
  Branches    79038    78956      -82     
==========================================
+ Hits        63050    63079      +29     
+ Misses      13810    13694     -116     
- Partials     2178     2183       +5     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 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.

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Improves solx stack-trace attribution/parity (especially around compiler-generated “declaration-attributed” locations) and adds tooling/fixtures to regenerate and test these behaviors against updated solx versions.

Changes:

• Extend solx trace strategy + DWARF decoding to better handle declaration-padding / unmapped locations and line-0 call sites.
• Add a CLI command to regenerate solx compiler-output fixtures, plus a new StackTraceScenarios fixture corpus.
• Expand provider integration tests and update sweep/docs to reflect improved parity and solx 0.1.6 behavior.

Reviewed changes

Copilot reviewed 15 out of 17 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
js/integration-tests/solx-parity-sweep/test/sweep.ts	Removes scenarios that no longer diverge from solc, tightening parity checks.
js/integration-tests/solx-parity-sweep/README.md	Updates pinned-divergence documentation to reflect solx 0.1.6 parity changes.
crates/tool/cli/src/update.rs	Fixes repo-root discovery by adjusting ancestor depth.
crates/tool/cli/src/solx_fixtures.rs	Adds fixture-regeneration helper that runs solx --standard-json and rewrites outputs.
crates/tool/cli/src/main.rs	Exposes fixture regeneration as a CLI subcommand.
crates/edr_solidity/src/trace_strategy.rs	Adds strategy hooks for location equivalence and declaration-attributed inference; improves revert attribution.
crates/edr_solidity/src/error_inferrer.rs	Uses new strategy hooks to keep stack traces stable and avoid declaration-location regressions.
crates/edr_solidity/src/debug_info/dwarf.rs	Handles solx 0.1.6 call_line=0 by falling back to decl sites; adds targeted tests.
crates/edr_solidity/fixtures/sources/StackTraceScenarios.sol	Adds new Solidity scenario corpus targeting provider-path stack-trace variants.
crates/edr_solidity/fixtures/solx_compiler_output_stack_trace_scenarios.json	Adds compiled solx output fixture for StackTraceScenarios.
crates/edr_solidity/fixtures/solx_compiler_output.json	Updates Counter fixture output to new solx/toolchain settings.
crates/edr_solidity/fixtures/solx_compiler_input_stack_trace_scenarios.json	Adds committed standard-json input for StackTraceScenarios (with contents scrubbed).
crates/edr_solidity/fixtures/solx_compiler_input.json	Pins optimizer mode in Counter standard-json input for fixture stability.
crates/edr_solidity/fixtures/README.md	Documents new gen-solx-fixtures regeneration path and the new fixture pair.
crates/edr_provider/tests/integration/solx_stack_trace.rs	Broadens provider-level integration tests for more stack-trace variants and new fixture.
.changeset/solx-declaration-attributed-inference.md	Adds changelog entry for the solx stack-trace inference fixes.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated no new comments.

Suppressed comments (1)

crates/edr_provider/tests/integration/solx_stack_trace.rs:74

• assemble_build_info uses unwrap() when replacing the source content. If the fixture format changes or the source_key is wrong, this will panic without a helpful error message, making failures harder to diagnose in CI. Prefer returning a contextual anyhow error instead.

    input.sources.get_mut(source_key).unwrap().content = source_content.to_owned();