Add native receiver routes and rebase OpenRazer 3.12.4

[offalexjackson777-stack]

Outcome

Add the source-bound native receiver tunnel and application routes, then rebase the narrow OpenRazer transport patch onto official OpenRazer 3.12.4 commit 1ef8a91712906a89fd332a2ed62bbd105315d6d1.

The patch preserves upstream direct-USB behavior, strscpy() cleanup, and the complete hid_report_raw_event() backport handling. HyperFlux adds only trusted feature-only UHID transport, exact receiver-route selection, capability markers, runtime detection, and package ownership/update policy.

Affected Domains

• Kernel transport or UAPI
• Bridge, protocol, or SDK
• Hardware profiles or application integrations
• Packaging or lifecycle
• Verification, assurance, or governance
• Documentation only

Contract Impact

• Canonical authority changed: native route, lifecycle, protocol, integration, packaging, and generated projection authorities.
• Generated outputs refreshed: yes; two consecutive ./hfx generate runs were clean.
• Schema, protocol, UAPI, or ABI compatibility: adds bounded native exchange and protocol v7 while retaining versioned earlier protocol projections.
• Package or activation behavior: HyperFlux owns/provides the patched openrazer-driver-dkms source, conflicts with/replaces a second driver owner, and fail-closes the relay when verified 3.12.4 runtime/capability requirements are absent.
• Performance budget impact: all declared static and package budgets pass.
• Privacy or support-bundle impact: private-data boundary passes; no raw captures, serials, host paths, or network upload are added.
• Release-gate impact: source remains unreleased; no tag, release, package channel, or supported-product claim is created.
• Evidence level: software, with previously retained route evidence remaining source-bound.
• Qualification state: exact mouse and keyboard routes are route qualified; this fresh package still requires separately authorized installation and live regression qualification.

Evidence

• Source revision: 1c1a0b30ee2a1797fe99b2013ebe0a768dbf44b6
• Upstreams: OpenRazer 1ef8a91712906a89fd332a2ed62bbd105315d6d1; OpenRGB 6fbcf62d7694e7b92fd0a5884b40b92984fbd1b0; Polychromatic 716737bf444ab7e962f4d3b7a5188b2f07c0c4a1
• Verification commands, selected nodes, and timings: pinned fast 26/26; pinned full software 34/34 in 1,234,416 ms; OpenRGB ThreadSanitizer passed inside the full graph.
• Generated freshness: clean and idempotent.
• Fast verification evidence: build/verification/openrazer-3124-final-fast-r3/evidence.json
• Full software evidence: build/verification/openrazer-3124-final-full-1c1a0b3/evidence.json
• OpenRazer host module compile: all four transformed modules compile against 6.18.37-1-lts and 7.0.14-arch1-1 headers.
• Arch package: two independent packages and manifests are byte-identical; package SHA-256 f022a0b03bf0da3e713b12d7f0e5ff7c019857d46b3ea83c389a987e6b46ec2b.
• Upgrade policy: offline pacman -Syu scenarios prove 3.12.4 release-suffix daemon upgrades cannot replace the HyperFlux-owned driver; 3.13 daemon updates also leave the relay fail-closed pending a reviewed source rebase.
• Physical evidence, when separately authorized: no fresh physical claim is made for this not-yet-installed package.

Interlocks

• Hardware authority is none, or a separate bounded authorization and evidence record is linked above.
• This change does not create a release, tag, package channel, or supported-product claim.
• User-facing links resolve inside the repository or to an explicitly reviewed external destination.
• User-facing documentation contains no internal engineering mission identifiers.

[offalexjackson777-stack]

CI-only follow-up 1e50dad6a47077ac261e8d3e46ac06cdb586ccd8 isolates exact upstream source caching from reusable build caches. The initial hosted failures restored a stale OpenRazer 3.12.2 checkout through the former broad fallback key; no product, transport, profile, package, or OpenRazer patch defect was involved.

Local correction verification: governance tests 18/18, generation idempotent, git diff --check clean, and change-selected fast verification 26/26. The source-bound installation candidate remains immutable commit 1c1a0b30ee2a1797fe99b2013ebe0a768dbf44b6 and package SHA-256 f022a0b03bf0da3e713b12d7f0e5ff7c019857d46b3ea83c389a987e6b46ec2b.

[offalexjackson777-stack]

Hosted Full software on 1e50dad exposed a second environment-specific verifier defect: Python placed the compiled OpenRazer constructor harness on a runner temporary mount that denied execution. Commit d64776a7de4fa86fd612fc3d10beee1ec7c23d09 moves only that ephemeral verifier binary to the ignored repository build/ root used by the other compiled contracts.

Direct reproduction now passes inside the pinned container with /tmp explicitly mounted noexec. Focused integration tests pass, generation is idempotent, and change-selected fast verification passes 26/26. Product, package, transport, upstream patch, and device-profile files remain unchanged.

[offalexjackson777-stack]

Final integration record:

• Verified package source commit: 1c1a0b30ee2a1797fe99b2013ebe0a768dbf44b6
• Integrated main commit: d64776a7de4fa86fd612fc3d10beee1ec7c23d09
• OpenRazer source: 1ef8a91712906a89fd332a2ed62bbd105315d6d1 (official v3.12.4)
• Local full graph: 34/34 passed, including OpenRGB ThreadSanitizer, source-bound OpenRazer routing, and package/upgrade contracts
• Exact-head hosted checks: Dependency review, Fast software, Full software, CodeQL C/C++, and CodeQL Python all passed
• Post-integration main checks: Verification and CodeQL passed
• Integration was a non-force fast-forward, preserving the verified candidate as an unchanged ancestor
• Protected-main governance was restored and verified current immediately afterward

No release, tag, binary publication, privileged installation, module reload, service mutation, or hardware claim was made.