[nirmoy]

SHA-referenced backports such as #696 and #697 contain the same edits as upstream but fail replay and adjacent-anchor validation because their surrounding kernel code differs. Allow a single (backported from commit …) reference with an immediately attached, nonblank backporter explanation to report noted when zero-context patch IDs and exact ordered removed/inserted lines match. This waives placement proof only; substantive edits, whitespace changes within the edit payload, file-mode changes, nonregular files, and mixed cherry-pick/backport provenance remain errors.

Ordinary cherry-picks stay strict. Upstream resolution, subject matching, and sign-off checks remain independently enforced. Lore-backed validation is unchanged.

Add 26.04_linux-nvidia-7.3-next to the PR Validation workflow branch list so PRs targeting it receive automatic Patchscan + PR lint checks after merge.

Validation:

• All 40 validator self-tests pass; new negative cases were observed failing before the corresponding fixes.
• Real SHA-backed commits from [26.04_linux-nvidia-bos] veth: set the real number of queues before registering the devices #696 and [26.04_linux-nvidia] veth: set the real number of queues before registering the devices - #696 #697 return noted; [26.04_linux-nvidia-7.3-next] veth: set the real number of queues before registering the devices #698 returns match; all exit successfully.
• Workflow YAML parses with existing branch entries preserved.

Ambiguous repeated-line diff grouping still fails closed. Accepted noted entries require manual placement review; the fallback does not certify semantic equivalence.

[nirmoy]

BaseOS Kernel Review

Tip

✅ Review passed

No issues found across the reviewed commits.

Findings: none

🔍 Review artifacts

• 🟢 Boro review: no findings
• 🔵 Test plan: ready for validation
• 🟡 KTS Report: unavailable

📦 Build checks — 🔴 0/4 passed

• 🔴 amd64: build report · failure log
• 🔴 arm64: build report · failure log
• 🟡 amd64: NVIDIA driver and DOCA not configured for github-actions
• 🟡 arm64: NVIDIA driver and DOCA not configured for github-actions

Note

Build reports and debs are retained for 10 days after the PR closes.

• 🟢 PR Illustrated: ready

Review metadata

• Reviewed head: 7172ed8e88b6
• Overall status: build failed
• Build checks: 0/4 passed

This comment is maintained by BaseOS Reviewer and updated when the GitHub watcher publishes a newer review.

[ianm-nv]

Hi Nirmoy — I reproduced a case where the new noted fallback accepts a cherry-pick that fails patch-ID/replay validation.

get_upstream_commit_sha() gives (cherry picked from commit ...) priority, but the fallback only checks whether a (backported from commit ...) trailer appears anywhere in the same message. Starting with the new _build_annotated_backport_case() fixture, I added a cherry-pick trailer for the same upstream SHA and kept the backport trailer and annotation. The validator exited 0 with Patch-ID noted, although the local and upstream patch IDs differ. I expected the cherry-pick to remain a mismatch, as described in the commit message.

Could the fallback require that no cherry-pick trailer is present, and could we add this mixed-trailer case to the tests?

One related policy question: the fallback checks that both patch IDs exist, but does not compare their contents. I changed the fixture's payload to unrelated text; it also exited 0 as noted. Is noted intended to mean that a reviewer must manually verify the entire backport diff, even when the change is not equivalent with zero context?

[nirmoy]

Thanks Ian. Updated the fix to reject mixed provenance and require matching zero-context patch IDs plus exact edit payloads, including whitespace. Added regression coverage; all 40 tests and CI pass.

noted now waives placement proof only, so placement still needs manual review.