Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion device/panther/BRINGUP-M1.md
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,9 @@ Hyprland and the Omarchy lock screen, on simpledrm over the bootloader's
framebuffer. SSH answers at `172.16.42.1`, and `bin/omamobile push`
rewrites slot B and reboots into it with no fastboot. The power and
volume keys work (the power key wakes the screen); there was no touch yet:
the FocalTech FT3658 on SPI0 had no mainline driver. Kernel 7.3.0-rc6,
the FocalTech FT3658 on SPI0 never answered in boots 6 to 8, with the
driver loaded, because its supplies were off; [TOUCH-PLAN.md](TOUCH-PLAN.md)
switched them on, and touch works from its T4 on. Kernel 7.3.0-rc6,
local branch `panther-first-light` on `omamobile-int`; the board file was
merged into `omamobile` as
[linux-omarchy-mobile#108](https://github.com/derekross/linux-omarchy-mobile/pull/108), and the FT3658 driver as
Expand Down
81 changes: 81 additions & 0 deletions device/panther/NOTES.md
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,15 @@ Findings from real hardware, newest first. [BRINGUP-M1.md](BRINGUP-M1.md) is the

The board file (`gs201-panther.dts`) was merged as [linux-omarchy-mobile#108](https://github.com/derekross/linux-omarchy-mobile/pull/108), and the FT3658 driver as [#109](https://github.com/derekross/linux-omarchy-mobile/pull/109); the test kernels below add one local commit each on top.

## 2026-10-29: the touch node, suspend-to-idle and LDO25M (T6)

One boot of kernel `7.3.0-rc6-omamobile-push1-g8d46192a301e`: the board branch of [linux-omarchy-mobile#134](https://github.com/derekross/linux-omarchy-mobile/pull/134) (the GPU, the haptics, `mem_sleep_default=s2idle`, LDO25M fixed at 1.8 V and the final touch node, since merged) merged with the driver's review fixes, [#133](https://github.com/derekross/linux-omarchy-mobile/pull/133) (DMA-safe SPI buffers, `fsleep()`), with `CONFIG_TOUCHSCREEN_FOCALTECH_FT3658=m` from `kernel/configs/omamobile.config`. The build's local merge, `8d46192a3`, predates the commit messages' test paragraphs and sign-offs; its tree is the same as the merge of the PR heads, #134 at `ace1ba57fbe7` and #133 at `fdc4f4403103` (both `0eb5d6eb`). The modules were pushed before the boot.

- `/sys/power/mem_sleep` reads `[s2idle] deep`.
- `vdd-ts` reads 1.8 V and `avdd-ts` 3.0 V, each with the touch controller as its only consumer.
- `focaltech_ft3658` loads at boot by itself: chip ID 0x56, firmware 0x16. Touches across the panel, the four corners included, drive Hyprland and unlock the PIN screen, with no SPI or CRC errors.
- The GPU and the haptics keep working (a 300 ms rumble was felt), and ACPM has 14 of its 15 channels usable.

## 2026-10-28: the GPU and the haptics, power off, and USB after a replug

Test kernels from local branches on slot B, each one change on top of the touch test kernel; one boot each.
Expand All @@ -15,6 +24,78 @@ Test kernels from local branches on slot B, each one change on top of the touch
- **Power off reboots the phone.** Omarchy's "Power off" (or `systemctl poweroff`) restarts the phone within a second, which spends a slot B retry. With the reboot watchdog off, systemd-shutdown runs to the end. A shutdown hook that saved the kernel log to `userdata` ten times a second, then powered off itself (sysrq `o`, the same `kernel_power_off()`), saved nothing after the trigger: no `sysrq: Power Off`, no device `shutdown` lines, no `reboot: Power down`. So the reset comes in the first tenth of a second of the kernel's power off, or before UFS can write another frame. With `syscon-poweroff` disabled in the test DT that leaves PSCI SYSTEM_OFF, which the firmware seems to turn into a restart (the boot reason reads `watchdog/apc`). The Pixel 7 Pro has never powered off from Linux either ([SUSPEND-PLAN](../cheetah/SUSPEND-PLAN.md) P7). To get to fastboot, hold Power and Volume Down from a running phone instead; that costs no retry.
- **USB doesn't come back after a replug.** A replug while Linux runs leaves the phone off the bus, as on the other Tensor phones (`AGENTS.md`): the MAX77759 opens its USB2 data switch, and nothing closes it. The Pixel 7 Pro's `omamobile-usb-rescue` needs the MAX77759's I2C bus (USI13, `10d60000`), which this board file doesn't enable yet, so for now a replug means Power and Volume Down, then a flash.

## 2026-10-27: modules-push and linux-modules-cleanup

`kernel-modules-hook` (in the root image) enables
`linux-modules-cleanup.service`, which on every boot moves each
`/usr/lib/modules/<release>` that is not the running kernel and that no
package owns into `/usr/lib/modules/.old`. A tree from
`bin/omamobile modules-push` is owned by no package, so a boot of any
other kernel before the new one (here, the unplanned reboot above) loses
it, and the new kernel then boots with no modules. `.old` kept only the
old release, so the pushed tree was not recoverable from it. Fix used:
`modules-push` again from the new kernel. Loading the missing modules
afterwards with `udevadm trigger` raced (ACPM probed before its mailbox,
`15 channels, 0 usable`, so `sec-pmic-acpm` failed with -ENODEV before
any PMIC write); a reboot loaded everything in order. To raise with the
tooling: push right before the reboot, or have `modules-push` warn.

## 2026-10-27: Omarchy's power off reboots the phone

"Shut down" in the Omarchy menu, with the USB cable to the laptop
plugged in, rebooted into slot B instead of powering off (and spent a
slot B retry). Holding the power key forced it off. Not looked into yet:
the bootloader may be powering the phone back on because the cable is in.
Until then, reach fastboot from a forced power off. (2026-10-28: it
reboots without the cable too, from the kernel's power off; see above.)

## 2026-10-27: a second stock fallback, and a new backup

Getting to fastboot from Arch went the same way twice: the phone came back
up in slot B instead of powering off, spent slot B's last retry, and the
next boot fell to slot A, whose stock recovery started. The data reset was
declined both times; `adb reboot bootloader` from recovery reached
fastboot (getvar: current slot A, B retry 0, not unbootable), and
`bin/omamobile flash` brought slot B back. A fresh read-only backup was
taken from Linux right after (each partition `ro=1`, found by `PARTNAME`,
checksum-matched against the phone): against the 2026-10-27 baseline,
`modem_a` and `modem_b` match, `persist`, `efs`, `efs_backup` and
`modem_userdata` differ in 100-750 bytes each, `devinfo` in 2. Those bytes
sit in the same places as after boot 6 (compared block by block, both
backups read-only): in each F2FS partition, the checkpoint (the other of
its two copies this time) and the root inode's node block written just
before it; in `persist`, the ext4 superblock, the first five journal blocks
and one inode table block, where only the ctime of `/sensors` and
`/sensors/registry` moved, the same two inodes as after boot 6. No data
block changed in any of them, so recovery's mount explains it. That copy is
the baseline now. To reach fastboot without spending a retry, hold Power
and Volume Down on the running phone until the bootloader shows, without
shutting down first.

## 2026-10-27: devinfo slot bytes against ABL (L-36 evidence)

devinfo is version **3.19** here (cheetah's bootctl is verified on 3.15
only). Only the version and the slot bytes (0x30-0x37) were read, from
Linux, read-only. Byte layout per slot: retry count, then flags (bit 0
unbootable, bit 1 successful, bit 2 active, bit 3 fastboot_ok).

| When | `fastboot getvar` | devinfo 0x30-0x37 |
| --- | --- | --- |
| stock capture, before any of our boots | A: retry 2, successful, bootable; B: retry 3, not successful, bootable | not read |
| one boot after `--set-active=b` (boot 7) | not read | `02 0a 00 00 02 04 00 00` |
| before the T2 flash (after boots 7 and 8) | B: retry 1, not successful, bootable | not read |
| after the T2 flash (`--set-active=b`) and two boots (T2, T2b) | not read | `02 0a 00 00 01 04 00 00` |
| one more boot of slot B (an unplanned reboot), then fastboot | current slot **A**; A: retry 2, successful, bootable; B: retry 0, not successful, bootable | not read |

Every reading agrees with the layout: slot A `02 0a` is retry 2,
successful, fastboot_ok, as getvar says; slot B counts down from 3 by one
per boot, ABL's own change, and `--set-active` resets it. The last row is
the next step of the count (1 to 0), and shows ABL's choice once B has no
retries left: A becomes current, B is not marked unbootable. Still
missing for L-36: a getvar and a byte read of the same state with no boot
in between; a byte read needs Linux, so it has to be taken right before a
planned reboot to fastboot.

## 2026-10-27: boot 6's fallback was a stock boot, and a new backup

When slot B ran out of retries (boot 6, [BRINGUP-M1.md](BRINGUP-M1.md)), ABL booted stock Android on slot A, which went to recovery because its `userdata` is now Arch. The data reset that recovery offers was **declined**: the phone was taken to fastboot with `adb reboot bootloader` and slot B flashed again.
Expand Down
Loading