Skip to content

device/tegu, DEVICES, AGENTS: the 9a on 2026-10-27 and 10-28 - #100

Merged
derekross merged 8 commits into
derekross:mainfrom
W4HO-ham:tegu-notes-20261027
Oct 30, 2026
Merged

derekross merged 8 commits into
derekross:mainfrom
W4HO-ham:tegu-notes-20261027

Conversation

@W4HO-ham

Copy link
Copy Markdown
Contributor

This is the tegu notes PR you asked for on kernel #63 and on #72.

pr-check passes. No identifiers: the camera lines are DT paths only.

@W4HO-ham W4HO-ham changed the title device/tegu, DEVICES, AGENTS: the 9a on 2026-10-27 device/tegu, DEVICES, AGENTS: the 9a on 2026-10-27 and 10-28 Oct 28, 2026
@W4HO-ham

Copy link
Copy Markdown
Contributor Author

The red CI here was the flaky audio watchdog test (a race in the test's fake mixer, nothing in this PR). The fix is #107 on its own; I've put the same commit on this branch so CI can pass. Whichever lands first, the other merges clean.

@derekross

Copy link
Copy Markdown
Owner

Thanks, @W4HO-ham, this is a big, useful set of notes: the LoCH abort fix for camera switching, the cluster 0 power-down hint resetting the phone (and Google never using it), the MIF numbers in s2idle, and the honest account of where the extra reserved regions come from. The AGENTS.md change checks out against MODEM-PLAN's suspend section.

A few things before it merges:

  1. Some matrix cells claim results that no notes entry shows. The matrix header asks for the notes entry behind each claim, and I couldn't find one in device/tegu for: Display "at 120 Hz" (the 120 Hz PR is still open, and the published mode table is 60 Hz), Touch "sleep fix", Buttons "Power wakes s2idle", Bluetooth "BCM4383" (the notes have A2DP and hci0 powered, but not the chip work), Charging "status and inhibit", Haptics "CS40L26, VPBR", Torch "LM3644, 25 steps", and Suspend "Power … wake". The kernel PRs may well have this evidence; please add a short dated line to NOTES for each (what ran, on which release), or cut those cells back to "in review, kernel #N".
  2. Fingerprint: "Trusty's stores mapped as stock maps them" isn't quite right. Stock maps 0 and persist/0 onto the real trusty_userdata and trusty_persist partitions; omamobile-fingerprint 0.1.10: map Trusty's tamper-detect stores as block devices #101 maps loop devices of image files under /var/lib/omamobile-fp. Please say that, and record the step itself in FINGERPRINT-PLAN (details on omamobile-fingerprint 0.1.10: map Trusty's tamper-detect stores as block devices #101).
  3. The Status cell lost its link on "Pixel 9a (tegu / Tensor G4): native bring-up #28"; please restore it.

This branch also carries #107's commits. I'll merge #107 first, after which they drop out of this PR's diff by themselves.

W4HO-ham added a commit to W4HO-ham/omarchy-mobile that referenced this pull request Oct 28, 2026
…ff slot A

As asked on derekross#100: a dated NOTES line for each claim in the 9a's row that had none (touch sleep fix, Power and RTC wakes, the BCM4383, charger status and inhibit, VPBR, the torch), the display cell back to 60 Hz with 120 Hz in review, the fingerprint wording saying the stores are loop devices of images in userdata, and derekross#28 linked again. MODEM-PLAN no longer sends the SIM to slot A for activation (issue derekross#51).
@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Thanks, all three are in 5c6b52f.

  1. Evidence for the row: NOTES has a new dated entry, "what the device table's claims rest on", with one line per claim you listed. Each says what ran, on which release, and which kernel PR has the details: touch's sleep fix (initramfs, device-felix 0.1.4: unmask both display triggers #82, 20 RTC cycles), Power waking s2idle (by hand, 10-27), the BCM4383 (docs/review: 18 lessons and 2 corrections from the twenty-second round #94), charger status and inhibit (CONTRIBUTING: shared drivers leave the Pixel 7 Pro as it is #92), VPBR (DEVICES: status after the 2026-10-27 evening merges #95), the torch's 25 steps, and the Power and RTC wakes. I cut Display back: the cell now says 60 Hz with 120 Hz in review, and the note says 120 Hz runs on the palmtop with device-lynx 0.2.0: the device package, the Arch root and the Wi-Fi plan #88's files but isn't in the published kernel.
  2. Fingerprint wording: both the cell and the notes now say the stores sit at stock's paths but on loop devices of Linux-owned images under /var/lib/omamobile-fp, not on trusty_userdata and trusty_persist. FINGERPRINT-PLAN records the step itself in omamobile-fingerprint 0.1.10: map Trusty's tamper-detect stores as block devices #101 (D1 and F4b).
  3. Pixel 9a (tegu / Tensor G4): native bring-up #28 is linked again.

The same commit also fixes MODEM-PLAN's SIM line for issue #51: the SIM is activated elsewhere, not on slot A.

@derekross

Copy link
Copy Markdown
Owner

Thanks, @W4HO-ham, all three are done, and the "what the device table's claims rest on" entry is exactly what the table needed: one line per claim, with the release and the kernel PR. I checked the cells against the kernel PRs' states and they match.

Two small things:

  1. This branch now conflicts with main in docs/DEVICES.md, so CI didn't run on your last push. While this was open, another PR updated the Pixel 4 XL row, the line above yours. The resolution is simple: take main's file and replace only the Pixel 9a row with yours. That row is the only thing this PR changes in the table. A rebase onto main will do it.
  2. The 10-28 Fingerprint line in NOTES still ends "the HAL reads the panel's id, likely that key". On omamobile-fingerprint 0.1.10: map Trusty's tamper-detect stores as block devices #101 you found that it isn't the key, since the bootloader names the file, so please update that sentence.

I'll merge #101 first, since the fingerprint cell links its D1 and F4b steps, and then this one.

Cameras (the evidence asked for on kernel derekross#63), NFC reading cards,
USB-C and host-mode status, UFS clock gating, the fingerprint sensor's
rail, the cable-out drain numbers and the recovery buttons, on one tested
release. The 9a's DEVICES row now links the kernel PRs each cell rests on,
and the NFC gotcha names the release of its 35 s2idle cycles.
Camera switching fixed, pstore, the idle-hint and sleep memory-rate measurements, the cameras' link errors, fingerprint storage and where its calibration lives, and the GNSS rails, with the pull requests in the 9a's row.
… come from

As asked on kernel derekross#80: abl@f8800000 is from stock's factory base DTB, the two gaps from the early bring-up's private DRAM map, and what would free them.
…ff slot A

As asked on derekross#100: a dated NOTES line for each claim in the 9a's row that had none (touch sleep fix, Power and RTC wakes, the BCM4383, charger status and inhibit, VPBR, the torch), the display cell back to 60 Hz with 120 Hz in review, the fingerprint wording saying the stores are loop devices of images in userdata, and derekross#28 linked again. MODEM-PLAN no longer sends the SIM to slot A for activation (issue derekross#51).
@W4HO-ham
W4HO-ham force-pushed the tegu-notes-20261027 branch from 5c6b52f to 2f9db1d Compare October 28, 2026 19:51
@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Rebased onto main: docs/DEVICES.md is main's file with only the Pixel 9a row replaced, and #107's commits dropped out. The NOTES fingerprint line now says the bootloader names the calibration file (its get-cal-url answer) and that the panel's id isn't the key.

@derekross

Copy link
Copy Markdown
Owner

Thanks, @W4HO-ham. The rebase is right: docs/DEVICES.md now differs from main only in the Pixel 9a row, #107's commits have dropped out, CI ran on this head and passed, and the fingerprint line now says the bootloader names the calibration file. Both points from my last comment are done.

Your record on #125 brings up one more thing for this PR. Two lines here still say stock on slot A "stops in recovery" with Arch on userdata:

  • device/tegu/NOTES.md, in the reserved-memory section: "this palmtop can't produce that (stock on slot A stops in recovery with Arch on userdata)";
  • device/tegu/MODEM-PLAN.md, in the new SIM sentence: "Not on slot A: stock there stops in recovery with Arch on userdata (see the 10-27 correction below), and the factory reset it would need wipes the Arch install."

#125's 10-26 and 10-28 entry shows that userdata was overwritten before recovery appeared, with nobody choosing anything. Booting slot A loses the root by itself, not only through a reset. The conclusion in both places stays the same (no stock reference from this palmtop, and the SIM goes in another phone), so only the reason needs to change. Something like "stock on slot A overwrites userdata before it reaches recovery (NOTES, 2026-10-26 and 2026-10-28)", linked to #125's section, would do.

Merge order: #125 and this PR both add a section at the top of tegu's NOTES, so they conflict there. I'll merge #125 first. When you rebase this one onto main after that, keep both sections, and the two lines above can link #125's entry. #101 still goes before this one too, as planned.

@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Both lines now say stock on slot A overwrites userdata before it reaches recovery, and link #125's NOTES section (the MODEM-PLAN correction bullet too, matching #125). Once #125 is merged I'll rebase this one onto main and keep both NOTES sections.

@derekross

Copy link
Copy Markdown
Owner

Thanks, @W4HO-ham. Both lines are right now. The reserved-memory note and the SIM sentence say stock on slot A overwrites userdata before it reaches recovery, and both link #125's entry. The MODEM-PLAN correction bullet matches #125's word for word, so the two branches merge cleanly there. Policy and tests pass on this head.

The plan stays the same: #125 first, then this one. I tried that order locally. The only conflict is the top of device/tegu/NOTES.md, where both PRs add sections. Keep both, with a blank line before ## 2026-10-28: what the device table's claims rest on. Everything else merges by itself, and the policy check passes on the result. Your links to #2026-10-26-and-2026-10-28-slot-a-fallbacks-that-cost-userdata only work once #125's section is on main, which is one more reason to keep that order. Once you've rebased, I'll merge.

W4HO-ham added a commit to W4HO-ham/omarchy-mobile that referenced this pull request Oct 29, 2026
…fter, the code by commit

The runbook stops if slot B isn't marked successful and doesn't mark it (a devinfo write on the 9a, derekross#125). S1 needs the NFC stop-poll hook active or polling stopped by hand, with the kernel's derekross#131 state and the rule that stays, a Rollback line and pstore after a cycle that doesn't come back. S2 snapshots regulator_summary before and diffs it after, stops if more than the four rails changed, records stock's consumers of them (none in its DT: no phandle on any of the four, the GNSS three on the PMIC's channel mux, LDO15S a spare), keeps the modem off in that boot and stays away from a G3 session. The push wording says what bin/omamobile push writes and that only vendor_kernel_boot differs. The module, the bring-up DT and the S2 change are linked by commit on the fork's tegu-sleep-bringup branch, and the CPU_INFORM hints of 10-28 are credited to the bring-up module that wrote them over the PMU SMC. The measurements link derekross#100's NOTES entry.
derekross pushed a commit that referenced this pull request Oct 29, 2026
* device/tegu: SLEEP-PLAN, an audit of the s2idle drain

The 9a sleeps and wakes, but its drain is high. This audits why from stock's device tree and a snapshot of the palmtop: no cluster or SICD idle states, the GNSS and fingerprint rails always on in the bring-up DTB, and MIF, with a measurement runbook.

* device/tegu SLEEP-PLAN: the firmware's PSCI has no OS-initiated mode

So cluster states on zumapro would be platform-coordinated, with a composite StateID; and the core state's rejection rate, read from cpuidle.

* device/tegu SLEEP-PLAN: start from what's measured

The cable-out s2idle figure (about 140 mA), MIF in sleep (about 57 mA saved), and the cluster power-down tries of 10-28 (cluster 0 hangs; clusters 1-2 save nothing measurable) were already measured; the candidates are re-ranked on them.

* device/tegu SLEEP-PLAN: S2's image, its expectation and rollback

* device/tegu SLEEP-PLAN: the review's preconditions, S2's before and after, the code by commit

The runbook stops if slot B isn't marked successful and doesn't mark it (a devinfo write on the 9a, #125). S1 needs the NFC stop-poll hook active or polling stopped by hand, with the kernel's #131 state and the rule that stays, a Rollback line and pstore after a cycle that doesn't come back. S2 snapshots regulator_summary before and diffs it after, stops if more than the four rails changed, records stock's consumers of them (none in its DT: no phandle on any of the four, the GNSS three on the PMIC's channel mux, LDO15S a spare), keeps the modem off in that boot and stays away from a G3 session. The push wording says what bin/omamobile push writes and that only vendor_kernel_boot differs. The module, the bring-up DT and the S2 change are linked by commit on the fork's tegu-sleep-bringup branch, and the CPU_INFORM hints of 10-28 are credited to the bring-up module that wrote them over the PMU SMC. The measurements link #100's NOTES entry.

* device/tegu SLEEP-PLAN: the fingerprint stack stays stopped in S2's boot, S1's command and column

S2 says it turns LDO15S off on purpose for one boot against FINGERPRINT-PLAN's rule (that plan gets the matching note once #138 is on main), and step 3 keeps omamobile-fp, the trusted app and storageproxyd stopped before the push and until the rollback, since F6c still writes secure objects. S1 names its command, power-ab with #157's systemd suspend mode (rtcwake -m no plus systemctl suspend, so the NFC hook runs), reads charge_now, column 7, because the 9a's gauge has no charge_counter, and waits for #157. Also: #125's rules are being settled; the four rails read disabled with or without the unused-rail cleanup; stock's source grepped for the four rail names (DT sources and the register header only, no module asks by name); a Polling read before each run; nothing starts the modem at boot, so nothing to mask.
docs/DEVICES.md: the Pixel 9a row conflicted; this branch's row is the newer status, so it stays.
@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Main moved under this one today (#140 and the afternoon's merges), and DEVICES.md's Pixel 9a row conflicted. I merged main into the branch rather than rebase (e83701a): the only conflict was that row, and this branch's row is the newer status, so it stays; nothing else changed. It should merge cleanly now, and #125, #131 and #136 behind it.

@derekross

Copy link
Copy Markdown
Owner

Thanks, @W4HO-ham. Merging main in was fine. I checked e83701a: against today's main it only touches this PR's four files, and CI passed on it. All my earlier points are done.

One correction to "it should merge cleanly now, and #125, #131 and #136 behind it": that's true against today's main, but not once #125 lands. #125 and this PR both add a section right under the NOTES header, so they'll conflict there again after #125's squash. The fix is the one I gave before: keep both, #125's section first, then a blank line, then ## 2026-10-28: what the device table's claims rest on and the rest of yours. I've tried that order locally. That's the only conflict, the MODEM-PLAN bullet you matched to #125 merges by itself, and the policy check passes on the result. You've allowed maintainer edits, so I'll push that resolution to your branch myself right after #125 merges, wait for CI, and merge, unless you'd rather do it.

One thing for my side, nothing for you to do: taking your 9a row whole was right for everything you'd changed. But it also undid the one cell #140 had touched (Fingerprint, with its link to FINGERPRINT-PLAN). Since #138 landed, both versions are behind the plan anyway: matching and the lock screen work on your kernel now. I'll fix that cell, with the plan link, in my next DEVICES status update.

I'll write the squash message myself, so the "Merge main into …" line doesn't land in it. Heads-up for #137: once this is in, it conflicts with this PR's NFC line in AGENTS. The resolution is #137's line with your "(7.3.0-rc6-omamobile-tegu-s2mpg15-m1)" after "all woke". I'll sort that out there.

@W4HO-ham

Copy link
Copy Markdown
Contributor Author

Please push the resolution yourself after #125 (both sections, #125's first) and merge. Understood on the Fingerprint cell; #165 carries the whole row again and I'll keep it behind this one and #166 as you asked there.

# Conflicts:
#	device/tegu/NOTES.md
@derekross
derekross merged commit aff16a4 into derekross:main Oct 30, 2026
2 checks passed
@derekross

Copy link
Copy Markdown
Owner

Thanks, @W4HO-ham. Merged, with the resolution we agreed: #153's 10-29 section on top (it landed last night), then #125's, then yours. I'll fix the Fingerprint cell, and bring the GNSS cell in line with the parking in #153, in my next device-table update.

W4HO-ham added a commit to W4HO-ham/omarchy-mobile that referenced this pull request Oct 30, 2026
Merge main (keep this row over derekross#100's, both NOTES entries). Display no longer claims brightness and the panel's sleep until they have a dated record; charging's kernel derekross#92 is merged, not in review; the Wi-Fi line says kept, not reconnected; the USB cell points at derekross#100's replug record.
W4HO-ham added a commit to W4HO-ham/omarchy-mobile that referenced this pull request Nov 9, 2026
… 6-hour auto-sleep series clean, and derekross#100

The shell's own auto-sleep behind the lock, with the fixed kernel: the power-key wake, 30-minute and one-hour sleeps, then a series of twelve half-hour sleeps woken by the RTC, all clean, 15 wakes in the boot with no UFS error. The two scripts arm the alarms and watch from the laptop. derekross#100 adds the modem fixes the test tree had lacked; the sleep path is unchanged.
W4HO-ham added a commit to W4HO-ham/omarchy-mobile that referenced this pull request Nov 9, 2026
… 6-hour auto-sleep series clean, and derekross#100

The shell's own auto-sleep behind the lock, with the fixed kernel: the power-key wake, 30-minute and one-hour sleeps, then a series of twelve half-hour sleeps woken by the RTC, all clean, 15 wakes in the boot with no UFS error. The two scripts arm the alarms and watch from the laptop. derekross#100 adds the modem fixes the test tree had lacked; the sleep path is unchanged.
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
* device/tegu: SLEEP-PLAN, an audit of the s2idle drain

The 9a sleeps and wakes, but its drain is high. This audits why from stock's device tree and a snapshot of the palmtop: no cluster or SICD idle states, the GNSS and fingerprint rails always on in the bring-up DTB, and MIF, with a measurement runbook.

* device/tegu SLEEP-PLAN: the firmware's PSCI has no OS-initiated mode

So cluster states on zumapro would be platform-coordinated, with a composite StateID; and the core state's rejection rate, read from cpuidle.

* device/tegu SLEEP-PLAN: start from what's measured

The cable-out s2idle figure (about 140 mA), MIF in sleep (about 57 mA saved), and the cluster power-down tries of 10-28 (cluster 0 hangs; clusters 1-2 save nothing measurable) were already measured; the candidates are re-ranked on them.

* device/tegu SLEEP-PLAN: S2's image, its expectation and rollback

* device/tegu SLEEP-PLAN: the review's preconditions, S2's before and after, the code by commit

The runbook stops if slot B isn't marked successful and doesn't mark it (a devinfo write on the 9a, derekross#125). S1 needs the NFC stop-poll hook active or polling stopped by hand, with the kernel's derekross#131 state and the rule that stays, a Rollback line and pstore after a cycle that doesn't come back. S2 snapshots regulator_summary before and diffs it after, stops if more than the four rails changed, records stock's consumers of them (none in its DT: no phandle on any of the four, the GNSS three on the PMIC's channel mux, LDO15S a spare), keeps the modem off in that boot and stays away from a G3 session. The push wording says what bin/omamobile push writes and that only vendor_kernel_boot differs. The module, the bring-up DT and the S2 change are linked by commit on the fork's tegu-sleep-bringup branch, and the CPU_INFORM hints of 10-28 are credited to the bring-up module that wrote them over the PMU SMC. The measurements link derekross#100's NOTES entry.

* device/tegu SLEEP-PLAN: the fingerprint stack stays stopped in S2's boot, S1's command and column

S2 says it turns LDO15S off on purpose for one boot against FINGERPRINT-PLAN's rule (that plan gets the matching note once derekross#138 is on main), and step 3 keeps omamobile-fp, the trusted app and storageproxyd stopped before the push and until the rollback, since F6c still writes secure objects. S1 names its command, power-ab with derekross#157's systemd suspend mode (rtcwake -m no plus systemctl suspend, so the NFC hook runs), reads charge_now, column 7, because the 9a's gauge has no charge_counter, and waits for derekross#157. Also: derekross#125's rules are being settled; the four rails read disabled with or without the unused-rail cleanup; stock's source grepped for the four rail names (DT sources and the register header only, no module asks by name); a Polling read before each run; nothing starts the modem at boot, so nothing to mask.
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
…ss#100)

Pixel 9a records from 2026-10-27 and 2026-10-28 in tegu NOTES: cameras,
NFC, USB-C, storage power, fingerprint and drain on 10-27; camera
switching, pstore, idle hints and sleep memory rate on 10-28; and what the
device table's claims rest on. MODEM-PLAN, AGENTS and the DEVICES row
follow those records.

NOTES keeps newest first: the 10-29 section, then the slot A fallbacks
from derekross#125, then these three.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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