fix(chip): classify missing RISC-V evidence as blocked

[NubsCarson]

Summary

• Treat absent generated Chipyard/AP Linux artifacts as BLOCKED evidence work instead of source-contract FAIL in the generated Linux contract gate.
• Let the minimum Linux target gate consume companion blocked Verilator smoke JSON when the transcript has not been captured yet.
• Fix QEMU toolchain detection so auto-detected clang without a working ld.lld is reported as BLOCKED, not a build regression.
• Add a hermetic qemu-status regression test that isolates the clang path with RISCV_CLANG_CANDIDATES, without falling through to host RISC-V GCC toolchains.
• Refresh the prototype status dashboard to match current MVP gate output and remove silent placeholder wording from the AI EDA coverage-bin file.
• Address Greptile follow-ups: collapse the duplicate GEN_FIR.is_file() check and print blocked follow-up artifacts/evidence alongside hard failures.

Validation

• python3 verify/check_stub_audit.py
• python3 scripts/check_prototype_status_dashboard.py
• make qemu-status-test
• make chipyard-generated-linux-contract-check
• .venv/bin/ruff format --check packages/chip/scripts/test_qemu_smoke_status.py packages/chip/scripts/check_chipyard_generated_linux_contract.py packages/chip/scripts/check_minimum_linux_target.py
• .venv/bin/ruff check packages/chip/scripts/test_qemu_smoke_status.py packages/chip/scripts/check_chipyard_generated_linux_contract.py packages/chip/scripts/check_minimum_linux_target.py
• sh -n packages/chip/scripts/run_qemu.sh
• python3 -m py_compile packages/chip/scripts/test_qemu_smoke_status.py packages/chip/scripts/check_chipyard_generated_linux_contract.py
• git diff --check
• python3 scripts/test_minimum_linux_npu_target.py
• python3 scripts/test_cpu_ap_boot_readiness.py

CI / Remaining Noise

• PR is mergeable and the production Docker image smoke now passes on this branch.
• The current broad CI failures are not caused by this diff: repo-wide server/client/plugin/Electrobun failures, homepage Biome formatting drift, and docker-regression broad packages/chip ruff/format failures in existing chip scripts outside this PR's changed files.
• A local broad readiness rollup on this host reports PASS=41 FAIL=1 BLOCKED=29; the lone FAIL is os-rv64-release-check failing before gate logic because the active interpreter lacks jsonschema. That is a separate reproducible-deps cleanup candidate, not a RISC-V evidence-classification claim from this PR.

Claim Boundary

This PR does not claim RISC-V Linux, AOSP, generated AP, board, or silicon boot. The remaining items are still blocked on real generated artifacts, toolchains, and boot transcripts; this only fixes classification so missing evidence is not reported as checked-in source failure.

Greptile Summary

This PR corrects the classification of missing Chipyard/AP Linux artifacts from FAIL (source contract violation) to BLOCKED (absent evidence) across three Python gate scripts, and fixes QEMU toolchain detection so a clang-without-lld situation is also reported as BLOCKED rather than a build regression.

• check_chipyard_generated_linux_contract.py: All four artifact-presence checks now funnel into a shared blockers list; when failures and blockers coexist the FAIL branch additionally prints the blocked items so engineers see the full picture in one run.
• check_minimum_linux_target.py: check_evidence gains a companion-JSON fallback so a sibling .json status report with status: "blocked" is consumed instead of reporting the log file as "missing" (an error).
• run_qemu.sh / test_qemu_smoke_status.py: clang_can_link_riscv_elf is extracted into its own function, RISCV_CLANG_CANDIDATES allows test injection, and a new test isolates the detection path by restricting PATH to a minimal fake bindir.

Confidence Score: 5/5

Safe to merge — all changes narrow the classification of missing evidence from FAIL to BLOCKED, which is strictly less alarming and matches the stated claim boundary.

The logic changes are additive: a new blockers path is threaded through existing check functions without altering the success or real-failure paths. The companion-JSON check in check_minimum_linux_target is guarded by an explicit status comparison and falls through to the previous "missing" result for any other status value. The clang toolchain refactor is exercised by a new test that correctly isolates the detection path. No data is lost and no check is silently weakened.

No files require special attention; the style-only redundancies in check_chipyard_generated_linux_contract.py and run_qemu.sh do not affect runtime behaviour.

Important Files Changed

Filename	Overview
packages/chip/scripts/check_chipyard_generated_linux_contract.py	Refactored check_* functions to accept a blockers list; missing generated artifacts now route to BLOCKED instead of FAIL. Double is_file() pattern in check_import_state is a minor redundancy.
packages/chip/scripts/check_minimum_linux_target.py	Adds companion .json status-report check so verilator-smoke and similar log evidence can be classified BLOCKED (not missing/error) when the report file says blocked.
packages/chip/scripts/run_qemu.sh	Extracts clang_can_link_riscv_elf into its own function and adds RISCV_CLANG_CANDIDATES env-var override. Word-splitting of the variable is intentional but undocumented.
packages/chip/scripts/test_qemu_smoke_status.py	Adds link_host_tools helper and test_autodetected_clang_without_lld_is_blocked; PATH is restricted to bindir, correctly isolating the clang detection path from system GCC tools.
packages/chip/docs/project/prototype-status-dashboard.md	Dashboard refreshed to align with current gate output — several items corrected from PASS to BLOCK with updated reason codes and commands.
packages/chip/verify/ai_eda/coverage_bins/e1_interrupt_reset_edges.yaml	Minor wording cleanup: removes placeholder language from the coverage-bin scope and CLINT bin description.

_{Reviews (3): Last reviewed commit: "fix(chip): classify missing riscv eviden..." | Re-trigger Greptile}

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 0e9d2b21-dd57-4a69-b25e-fb8122bbbb5c

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch codex/riscv-shaw-help-20260520

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[claude]

Claude encountered an error —— View job

---

I'll analyze this and get back to you.

[claude]

Claude encountered an error —— View job

---

I'll analyze this and get back to you.

[claude]

Claude encountered an error —— View job

---

I'll analyze this and get back to you.

[NubsCarson]

Update after latest amend (6856e829af):

• Addressed Greptile P1 by keeping the clang/lld regression test on a hermetic PATH and narrowing auto-detection with RISCV_CLANG_CANDIDATES=clang, so hosts with RISC-V GCC installed cannot bypass the clang path.
• Addressed the two P2 notes in check_chipyard_generated_linux_contract.py: collapsed the duplicate GEN_FIR.is_file() check and now prints blocked follow-up artifact/evidence diagnostics even when hard failures are present.
• Re-ran targeted validation: ruff format/check on changed chip scripts, make qemu-status-test, make chipyard-generated-linux-contract-check, py_compile, and git diff --check.

Remaining red checks are broad repo/shared failures or existing chip-wide ruff drift outside this PR's changed files; PR body has the current breakdown.