[techmore]

Supersedes #518 (this branch carries its commit, rebased onto current dev) and adds upstream omacom/linux-aurora #186.

Why

Dock HDMI fails on wip3 when the dock is first plugged in right after resuming from s2idle: apple-dcp 28cc00000 times out the DPTX validate call six times (-110), then the same dock works on 289c00000 after a port change. Across retained journals (Oct 20-27): no DPTX failures on .20 or wip2 (wip2 used 28cc00000 fine) or on a wip3 boot with the dock plugged at boot; all three failing boots were wip3 with the first plug within ~4 s of resume. Consistent with, not proof of, a post-resume DCP state problem.

Series (revision 6, 7.2.11-elementary-aurora72-wip4)

• 0001 Document Aurora .20 hotplug findings and old module cleanup #157 sRGB external display (unchanged)
• 0002 local: skip a DCP whose DPTX times out for 300 s; a timed-out release no longer pins the route (from Kernel wip4 candidate: free a DP tunnel route pinned to a silent dcpext (build-only) #518)
• 0003-0005 upstream Battery app: icon, remembered state, shortcuts #186 (tested head 2a37f6f2): take over a pipeline held without HPD; route a DP-alt port only once its sink asserts HPD; -ENOSPC = taken display on every SoC. Its message reports early routing leaving the DPTX service on a DCP answering -110 until reboot.
• upstream Add compact network rates, Ethernet icons and battery watts toggle #163 intentionally not carried: same tipd bug as 0005, different change, conflicts with it.

Checked

• candidate.py prepare --profile unified72: all five patches apply to the full pinned Simplify Orchard panel to icon and running-instance count #158 source (exit 0)
• candidate.py check: passes; 83 aurora tests pass (series test updated deliberately)
• the aurora-enhancements CI job here does the first native kernel build

Not done

Kernel build/KUnit (CI), any boot, hot-plugging a dock (upstream's test report did not either). 0002 and 0003/0004 both edit dcp-fabric.c; they apply in order but their semantic interaction is unreviewed. Nothing installed, staged, armed or rebooted; wip3 stays the running kernel.

Receipt: mseries/aurora/unified72/review-wip4-2026-10-27.json

[techmore]

Note

🤖 Claude Opus 5.5 (Claude Code triage) responding on behalf of techmore

Recommendation: rebase (mechanical), then land before #533 or decide the order with it. Only 3 commits behind dev. The single conflict is mseries/OS9-HANDOFF.md, which is docs-only.

• This PR supersedes Kernel wip4 candidate: free a DP tunnel route pinned to a silent dcpext (build-only) #518.
• It overlaps with Aurora wip4: port Aurora20 enhancement series #533: both claim series revision 6 for wip4 with different patch lists. Settle which one is revision 6 before merging either.

Next step: rebase onto dev and resolve the handoff doc. The owner decides the order relative to #533.

[techmore]

Native kernel build passed: CI run 40498010695, job ARM64 unified72 build and payload inspection (39m20s), all 18 checks green. Docs and receipt updated (1dfb6ec). Still not done: KUnit, any boot, hot-plugging a dock with this series (upstream's #186 report didn't either). Artifacts (293/297 MB) were not downloaded or audited. Stays a draft: booting it is the owner's decision, and the owner-run boot trial is the only real test of the post-resume dock failure on 28cc00000.