Skip to content

DEVICES: the matrix re-read against the records on 2026-11-03 - #406

Closed
derekross wants to merge 1 commit into
mainfrom
devices-audit-2026-11-03
Closed

derekross wants to merge 1 commit into
mainfrom
devices-audit-2026-11-03

Conversation

@derekross

Copy link
Copy Markdown
Owner

An audit of docs/DEVICES.md against the repository's own records: each device's NOTES.md and plan status lines, its omamobile-device-<codename> package, and the pull requests merged in the last two weeks in this repository and in linux-omarchy-mobile. A cell moves only where a merged pull request or the device's notes say so; "works" stays where the owner's report proved it on the phone, WIP/partial wording stays where the evidence is thin, and every cell keeps its plan link. The Pixel 7 Pro rows, rewritten today (#388, #398, #402), are untouched. No phone was touched.

Changed cells

Row Cell Was Now Why
intro - - a dated line: re-read on 2026-11-03 asked for
Pixel 9a GPU kernel #85 and #87 "in review" both landed through kernel #96, merged; the package's 890 MHz ceiling; no boot scaling from that merged form recorded yet kernel #85/#87 closed as "landed unchanged through #96"; tegu NOTES 10-29 (fixed 160 MHz) is still the last scaling record
Pixel 9a GNSS kernel #172 "in review" #172 merged, kernel #185 (the GNSS link) merged with its 11-02 first boot, #114 (Location on the 9a) merged, CPIF kernel #187 in review; still testing kernel #172 (11-02), #185 (11-03), repo #114 (11-03); tegu NOTES 11-02
Pixel 9a Modem "calls have no audio route" calls with speakerphone audio through #367 (in review), Phone and Messages via #371 (draft), boot start by #372 (opt-in, in review), #364 merged, echo and MMS open; still WIP tegu NOTES 11-02 evening (#368), TELEPHONY-PLAN (#360), #364
Pixel 9a Fingerprint "works (opt-in)" works; omamobile-fingerprint 0.1.14 starts it by default after provisioning on both phones, the 9a's guard only that the TDP image exists, no 9a record of that default start yet #358: a vendor wants link in the package, the chroot script's tegu branch
Pixel 8 Pro Display WIP (simple-framebuffer) testing: gs-decon, HK3 at 120 Hz with brightness on a test kernel (kernel #160 in review) from device-husky 0.1.3 #228, tested on the first husky by its owner
Pixel 8 Pro GPU no testing: Panthor on a test kernel (kernel #157 in review), the clock to 890 MHz from device-husky 0.1.2 #221
Pixel 8 Pro USB net testing (USB and SSH) plus the replug rescue, device-husky 0.1.4 #250, tested on an 8 Pro
Pixel 8 Pro Battery, Charging kernel #163 kernel #163, merged kernel #163 merged 11-01
Pixel 7 Display kernel #177 "in review" merged; the flip-only KMS path is kernel #181 in review kernel #177 merged 11-01
Pixel 7 Battery kernel #171 and #170 "in review" merged both merged 11-01
Pixel 6 Pro GPU DVFS on an unpublished kernel plus "in review as kernel #189" kernel #189 open
Pixel 6 Pro Modem the package doesn't enable it plus "the runtime is in review as #389" #389 open
Wanted, Pixel 9 family - "the board DTS isn't public yet" in the published kernel (kernel #81); most rows at works on the owner's kernel kernel #81 merged 10-30
Wanted, Pixel 6a - GPL tree facts only plus the factory image's AVB facts (#350) #350 merged 11-03

Checked and left alone: Fold, 7a, 6, 4, 4 XL, OnePlus 13, Tablet, 8/8a, 10 and 11 rows, the 9a's Boot, Storage, Display, Touch, Buttons, USB, Wi-Fi, Bluetooth, Battery, Charging, Haptics, Torch, AoC, Speakers, Sensors, NFC, Cameras and Suspend cells (#347 already updated Suspend), the 7's Boot, Touch, USB and Suspend cells, and raven's Display (kernel #116 open). All local links and anchors resolve.

Stale on the evidence, but left for the maintainer or an open pull request:

tools/pr-check origin/main and tests/all pass.

🤖 Generated with Claude Code

https://claude.ai/code/session_01BeEcdVeSfyLCbijF8XQF1d

Each cell was checked against its device's notes and plans, its device
package, and the pull requests merged in the last two weeks. Only cells
with a merged pull request or a notes entry behind the change move; the
Pixel 7 Pro rows, rewritten today (#388, #398, #402), are untouched.

Pixel 9a (tegu):
- GPU: kernel #85 and #87 landed through kernel #96 (merged 11-01), so
  frequency scaling and the GPU thermal zone are no longer in review;
  the package sets the 890 MHz ceiling, and the 10-29 boot at a fixed
  160 MHz stays the last record, so the cell says no scaling boot from
  the merged form is recorded yet.
- GNSS: kernel #172 (the PERIC0 clock) is merged, kernel #185 (the
  GNSS link) is merged with its 11-02 first boot, #114 (Location on
  the 9a) is merged; CPIF stays kernel #187 in review. Still testing:
  the modem starts by hand.
- Modem: the 11-02 evening record (#368) and TELEPHONY-PLAN (#360)
  say calls ran with speakerphone audio both ways through #367's
  route, from Phone and Messages (#371), with the modem started at
  boot by #372's opt-in unit; #364 is merged. Those PRs are in review
  or draft, so the cell stays WIP and says which part each one is.
- Fingerprint: omamobile-fingerprint 0.1.14 (#358) starts the service
  by default on both phones (a vendor wants link; the chroot script
  has the 9a's guard), so "opt-in" no longer describes the package.
  The cell keeps the 10-29 evidence and says the default start has
  no 9a record yet.

Pixel 8 Pro (husky): the display (gs-decon, HK3 at 120 Hz, brightness,
device-husky 0.1.3, #228) and the GPU (Panthor, the clock to 890 MHz,
device-husky 0.1.2, #221) were tested on the first phone on test
kernels (kernel #160 and #157 in review), so both move from WIP/no to
testing as the other husky cells are worded. The USB replug rescue
(device-husky 0.1.4, #250) is noted, and kernel #163 is merged.

Pixel 7 (panther): kernel #177 (the panel and DCS brightness) and
kernel #171 and #170 (the fuel gauge) are merged, not in review; the
flip-only KMS path is kernel #181 in review.

Pixel 6 Pro (raven): the GPU DVFS series is in review as kernel #189
and the modem runtime as #389; the cells now point at them.

Wanted: the 9a's board device tree is in the published kernel (kernel
#81), so "isn't public yet" goes; the 6a row records the factory
image's AVB facts from #350.

A dated line after the introduction records the re-read.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BeEcdVeSfyLCbijF8XQF1d
@derekross

Copy link
Copy Markdown
Owner Author

Self-review notes before merging. The pass is careful, and the stale list is useful. Three cells need tightening against their records:

I'll also ask the husky owners for a NOTES entry for the display and GPU runs, and consider noting the 11-02 Wi-Fi re-attach failure in the 9a's Wi-Fi cell. This lands before the other open DEVICES PRs, which then resolve their row at merge.

@derekross

Copy link
Copy Markdown
Owner Author

Closing my own audit from 11-03 as superseded by #474, which re-read the matrix today against the same records and everything merged since. Any cell from here that #474 didn't carry over can come back in a small follow-up.

@derekross derekross closed this Nov 7, 2026
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.

1 participant