Skip to content

docs: explain MacBookPro14,1 USB fallback after suspend - #18138

Open
tcholewa wants to merge 6 commits into
omacom:quattrofrom
tcholewa:docs/macbookpro14-1-suspend-validation
Open

tcholewa wants to merge 6 commits into
omacom:quattrofrom
tcholewa:docs/macbookpro14-1-suspend-validation

Conversation

@tcholewa

@tcholewa tcholewa commented Nov 8, 2026 •

Copy link
Copy Markdown

On a MacBookPro14,1, a USB controller resume failure can leave Gigabit Ethernet running through USB 2 even while the cable link still reports 1 Gbps. The Mac manual currently gives no guidance for recognizing that fallback.

Add a short manual section covering the symptom, lsusb -t checks before/after sleep, cold power-on recovery, and links to the ongoing upstream work. Add a reference report with independent module/package provenance, comparison measurements, first-suspend-after-cold-boot validation, and explicit test limits.

On the tested MacBookPro14,1, the unchanged proposed shim with s2idle and normal Thunderbolt driver resumed devices in 0.5322 seconds on the first post-boot cycle; the subsequent Ethernet download test reported roughly 925–934 Mbps. The report distinguishes the tested local module configuration from the companion PR's installer/migration, and keeps the unmerged package status explicit.

Physical lid testing isolated a separate EC wake-registration failure: acpi_set_gpe_wake_mask returned AE_TYPE for EC GPE7 because it was not marked wake-capable. The symptom persisted with both suspend setups, including a cold boot of the old hook after an SMC reset. A model-scoped local helper calling acpi_ec_mark_gpe_for_wake() changed the traced result to AE_OK, then passed one physical lid-open test. The helper was packaged locally for DKMS, built for both installed kernels, and verified to load automatically after reboot. The user completed two further lid cycles; the logs record both successful resumes, zero suspend failures and USB5000M retained. This was tested with the old hook; the EC helper plus Alpine Ridge shim combination, longer-term reliability and overnight behavior remain untested. The report distinguishes this local workaround from the unchanged upstream shim package.

Changes are limited to two Markdown files on a fresh branch from current quattro. The older hook installer is excluded.

Validation:

  • git diff --check and staged whitespace checks passed.
  • Markdown fences and relative links checked.
  • Hardware evidence covers three RTC-triggered module cycles on one machine, including one first-suspend-after-cold-power-on check, followed by repeated physical lid tests of EC wake, the D3cold guard and the restored hook; Thunderbolt peripherals, overnight battery drain, hibernation and other models remain untested.

Related work:

@greptile-apps

greptile-apps Bot commented Nov 8, 2026 •

Copy link
Copy Markdown
Contributor

RetriggerView in GreptileConfidence Score: 5/5

[Low impact] The documentation changes appear safe to merge.

Summary

Documents USB 2 fallback after suspend on MacBookPro14,1 and explains how to check it.

  • Mac owners can spot USB 2 fallback after sleep.
  • The guide explains a separate local fix for lid-open wake.

Reviews (6) · Last reviewed commit: "docs: verify EC helper boot loading and ..." · Reviewed by Greptile

@omarchybot

Copy link
Copy Markdown
Collaborator

Reviewed by GPT-6 High in Codex. I read both Markdown changes, the existing Mac support section, #10593, and the linked hardware-validation reports on #10758 and omacom/omarchy-pkgs#349. No code was run.

I classified this as documentation, but the second classification model, DeepSeek V4p1 Flash, returned no usable class. Classification remains unsettled, so verification has not proceeded and no labels were added. This needs another classification pass before verification; there is no requested change for the author at this stage.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants