Skip to content
Closed
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
103 changes: 103 additions & 0 deletions device/tegu/NOTES.md
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,109 @@ hash/build ID; this dirty release is not a published kernel package.
`[V]` marks verified observations; `[I]` marks an inference still to test.
Raw captures, recordings, backups and vendor files remain off repository.

## 2026-10-31 and 11-01: every testing row on a published-style kernel

Each line is a boot of the named head on the palmtop, built with
defconfig + `kernel/configs/omamobile.config` (plus #227's options) +
`olddefconfig`, pushed with `bin/omamobile push`, results collected, then
the daily integration kernel pushed back. Each row gives the release of
its boot (`uname -r`; the hash is the built tree's commit). Details are on
each pull request.

- [V] **Kernel config gaps (#227, `7.3.0-rc6-omamobile-push1-g5628f1888904` with all four built in).**
A build from the fragment booted with no USB: the eUSB2 PHY and the ACPM, mailbox and S2MPG14 path its rails
come through were modules, and the initramfs loads none. The fragment
also lacked the MAX77779 fuel gauge and charger, the RT6160 camera
regulator, the zumapro PCIe host (as a module) and the TAS2572 driver.
- [V] **Display and GPU (kernel #96, `7.3.0-rc6-omamobile-push1-g5628f1888904`):** DSI-1 connected at 1080x2424, a
panthor render node, from the package's graphics start.
- [V] **Bluetooth (kernel #122, `7.3.0-rc6-omamobile-push1-g6ec07aeee91d`):** hci0 up, host wake 6 interrupts idle and
38 after a 12 s scan.
- [V] **Charging (kernel #161, omarchy-mobile #230,
`7.3.0-rc6-omamobile-push1-g7dbac6f4e65c`):** the charger binds at
0x69, and the charge limit holds charging off at 80 % through
`charge_behaviour`.
- [V] **AoC, speakers, microphone, sensors (kernel #162, omarchy-mobile
#234, `7.3.0-rc6-omamobile-push1-g922812d75006`):** the AoC from the device tree at boot (93 services), the sound
card, a tone heard from the speaker, the microphone at -32 dBFS, and the
sensor proxy with five sensors.
- [V] **NFC (kernel #104 rework, #105, omarchy-mobile #235,
`7.3.0-rc6-omamobile-push1-gb45a0d400e49`):** the clock pad
from clk-exynos-clkout, `nfc0` polling with no I2C errors, neard and the
link opener from the package. The pad variant's optional parent needed
`of_property_match_string()` first: `of_clk_get_by_name()` returns -EINVAL,
not -ENOENT, when a node has no `clock-names`.
- [V] **Cameras (kernel #164, `7.3.0-rc6-omamobile-push1-g4023f585ccae`):** all three stream (19 fps main, 30 fps
ultrawide and front, 1280x960) once PMU 0x3ebc, the DC-PHY's isolation,
is in the zuma PMU allowlist; without it every `phy_power_on()` failed
with -EIO.
- [V] **Fingerprint (kernel #165, `7.3.0-rc6-omamobile-push1-gb533af3f1cbc`):** goodix-fp binds from the device tree
(`/dev/goodix_fp`, irq on gpa7-6); matching was not run, since the
service writes Trusty's secure storage.
- [V] **Wi-Fi (kernel #168, omarchy-mobile #238,
`7.3.0-rc6-omamobile-push1-g07bdf006cdc2`):** the new zumapro PCIe
channel-1 host plus #56's brcmfmac: the link, the BCM4383 at
`14e4:4449`, `wlan0` connected at boot, a ping over it.

## 2026-10-31 and 11-01: leftovers on the palmtop from bring-up

- [V] **A plugin link in `~/.config/omarchy/plugins/` outranks the package.**
`omamobile.quicksettings` pointed at a 10-17 hand copy (no notification
shade) and `omamobile.switcher` at a 10-22 copy; four plugins were not
linked at all. Relinked with `omamobile-setup-user`'s loop (AGENTS, #243).
- [V] **NetworkManager was masked in `/etc` and iwd ran Wi-Fi,** so the
quick-settings Wi-Fi tile had no device. The switch, under a two-minute
dead-man unit that would have masked NetworkManager and started iwd
again: remove `/etc/systemd/system/NetworkManager.service` (the mask, a
link to `/dev/null`), `systemctl daemon-reload`, `systemctl disable --now
iwd`, `systemctl enable --now NetworkManager`, then `nmcli device` showed
`wlan0` managed, and a fresh SSH login stopped the dead-man. The
bring-up start script on the palmtop that used to start iwd now leaves
Wi-Fi to NetworkManager when it is enabled.
- [V] **`omamobile-aoc`, `omamobile-sensor-proxy` and `omamobile-nfc-open`
were masked in `/etc`;** the three masks are gone, recorded in
`/root/bringup-backup/etc-masks/`.
- [V] **A fastboot flash of slot B's images clears slot B's successful
flag** (#224): the evidence, and the `devinfo` write that marked B again,
are in the next entry.

## 2026-10-31: a fastboot flash of slot B cleared its successful mark, and B was marked again

[V] Linux-omarchy-mobile #96's test kernel came up without USB (the config
gaps in #227), so the palmtop went to fastboot by the buttons and got the
daily images back. Before the flashes, `fastboot getvar` read
`current-slot: b`, `slot-successful:b: yes`, `slot-unbootable:b: no`,
`slot-retry-count:b: 2`. The five commands, suffixed, with no
`--set-active`:

```sh
fastboot -s <serial> flash boot_b boot.img
fastboot -s <serial> flash init_boot_b init_boot.img
fastboot -s <serial> flash vendor_boot_b vendor_boot.img
fastboot -s <serial> flash vendor_kernel_boot_b vendor_kernel_boot.img
fastboot -s <serial> flash dtbo_b dtbo.img
```

After them, `slot-successful:b: no` and `current-slot: b`. The retry count
wasn't read again in fastboot; on the boot that followed, `devinfo` read B
active, not successful, `retry-count` 2.

[V] B was then marked successful again, a `devinfo` write against the 9a
rule's hold, made under the 10-21 approval that the hold had replaced. The tool
was `omamobile-bootctl` (`packages/omamobile-device-cheetah/src/omamobile-bootctl.c`,
a static arm64 build with its source and binary hashes pinned in the
runner). The steps: the live 8192 bytes copied off the palmtop and
compared with the backup taken after the 10-28 stock boot (only the
bootloader-owned slot bytes and the record at 0x6d-0x70 differed); a dry
run that planned one byte, 0x35 `1c` to `1e` (B's successful bit); a
detached relock armed for 30 s; then `blockdev --setrw` on the disk and on
`devinfo`, the write, and an EXIT trap that set both read-only again. The
read-back matched the predicted image in all 8192 bytes, so 0x35 is the
only byte that changed, slot A's bytes are unchanged, and both read-only
flags read 1 afterwards. No other backup was taken first. Until the
maintainer decides how B is marked on the 9a, there are no more `devinfo`
writes: with B unmarked, reboots stay few and images are tried with `push`.

## 2026-10-30: the hands-on morning (fingerprint, card, USB-C host, GNSS)

Release `7.3.0-rc6-omamobile-tegu-s2mpg15-m1-g991042e2eff3-dirty`, device
Expand Down
117 changes: 117 additions & 0 deletions device/tegu/SLEEP-PLAN.md
Original file line number Diff line number Diff line change
Expand Up @@ -427,6 +427,123 @@ with the cable in.
no ACPM timeouts. With gating on, 10-31's earlier run was also ten of ten,
so this does not yet separate the two; the count both ways goes on until
one side fails.

**S1 on the clock-fix kernel, 2026-11-01 (same release, the daily kernel
with the PERIC0 GPIO clock kept on, linux-omarchy-mobile #172).** That
gate hung the SoC under the modem's GPIO traffic (MODEM-PLAN G3-L1), so
it was a candidate for these no-wakes too. Cable in, modem never loaded,
NFC polling off before every cycle, UFS clock gating on (the default),
thirty 30-second RTC cycles planned: cycles 1-20 woke (47-80 s from
suspend to SSH, `suspend_stats` success 20 and fail 0, no UFS or ACPM
lines, wake IRQ 56 each time), and cycle 21 did not wake: no SSH 160 s
after the suspend, and it needed Power + Volume Up. So the clock gate is
not the sleep hang; one in 21 here, against about one in 40 before.
After the reset (2026-11-01 morning), nothing from the hung sleep was
kept: `/sys/fs/pstore` was empty, and the persistent journal of that
boot ends on `PM: suspend entry (s2idle)` with nothing after it. That
run didn't keep the console through sleep (`printk.console_suspend=N`
went into the script after it), and a Power + Volume Up reset may not
keep the ramoops region anyway. The next run keeps the console, and
whether the SoC watchdog still runs in s2idle (a watchdog reset keeps
ramoops; the buttons were needed every time so far) is the next thing
to check.

**A watchdog through s2idle, tried 2026-11-01 (same release): it breaks
the sleep, so it can't catch the hang.** `s3c2410_wdt` stops both
watchdogs in its suspend callback, so a hang in s2idle never resets the
palmtop. A bring-up module (`tegu-wdt-s2idle`, in the owner's logs)
restarted the cluster 0 watchdog in the noirq phase with the PMU reset
path unmasked and the counter enabled, as the driver's own enable does,
and stopped it in resume_noirq. Armed, three sleeps of three ended in a
reset: the s2idle wake came within 60 ms of arming, and after resume
ACPM stopped answering (`Timeout! ch:2`, then `RX front can not keep up
with TX front` and `sec-pmic-acpm … Failed to read IRQ status: -110`
every 0.5 s) until the restored 10 s watchdog reset the SoC; ramoops
kept all of it. Loaded but disarmed, two sleeps of two woke (after 1.5
and 1.6 s, wake IRQ 56, no ACPM timeouts). So a running cluster 0
watchdog (or its PMU counter enable) during s2idle upsets the APM, and
this module is not a way to record the real hang. The real hang's
console is still unknown; what this run does show is what an APM that
stops answering after resume looks like, which matches the stall seen
in G3-L1's CPU-sleep run (MODEM-PLAN).

**The sleeps that woke were short: the PMIC wakes s2idle about a second
in (2026-11-01, same release).** In the disarmed sleeps above,
`PM: suspend exit` came 1.5 and 1.6 s after `PM: suspend entry`, not
at the 30 s RTC alarm, with `/sys/power/pm_wakeup_irq` 56, the
S2MPG14's own line (GICv3 106, `s2mpg14-pmic`). One more 30 s sleep
with `pm_debug_messages` on: "Triggering wakeup from IRQ 56" 0.6 s
after entry, exit 1.5 s after it. Across that sleep and the 40 s after
it, the PMIC line counted 2 and `rtc-alarm0`, its only mapped child,
1, which is the real alarm going off after the wake; awake, the same
`rtcwake -m no -s 30` alarm fired at 30 s as it should. So another PMIC
source, one without a mapped child, raises the line as the palmtop
goes to sleep. Earlier cycle logs only recorded wake IRQ 56 and the
time to SSH, which can't tell a short sleep from a full one, so S1's
wake counts may describe sleeps of about a second. Next: read the
S2MPG14 interrupt registers (INT1-INT5 through the ACPM regmap, from
the kernel, never from userspace) at that wake, and check the 30-minute
sleeps' records for their real length.

A bring-up module (`tegu-pmic-wakeread`) then read INT1-INT5 and their
masks through the PMIC's regmap in suspend_noirq and resume_noirq
(reads only; INT1-INT5 clear on read). Three sleeps, all woken about a
second in:

| Alarm set | INT1-5 at sleep | INT1-5 at wake | Wake IRQ |
| --- | --- | --- | --- |
| 30 s | 00 09 00 00 00 | 00 0c 00 00 00 | 56 (PMIC) |
| 120 s | 00 09 00 00 00 | 00 0d 00 00 00 | 56 (PMIC) |
| none (`rtcwake -m disable`) | 00 08 00 00 00 | 02 09 00 00 00 | 59 (Power key, gpa4-0) |

The masks read `ff fb ff ff ff` each time, so only INT2 bit 2, alarm 0,
is unmasked. With an alarm set, alarm 0 latched within 60 ms of the
suspend_noirq read, whether it was due in 30 or 120 s; with none set,
INT1 bit 1 (power key released) latched and the power key's GPIO woke
the SoC, with nobody touching the key. So something in the last stage
of the suspend makes the S2MPG14 raise events that didn't happen, and
every sleep since at least 2026-10-30 has probably lasted about a
second. The 1 s and 60 s RTC ticks (INT2 bits 3 and 0) were latched
but masked throughout. After the no-alarm sleep, SSH over USB took
about 8 minutes to come back, with nothing in the kernel log about
USB.

**With CPU core power-down off, the sleep lasted its full length.**
The only deeper cpuidle state on each CPU is state1
(`cpu-a520-sleep`, `cpu-a720-sleep`, `cpu-x4-sleep`, PSCI core
power-down; each was used 4-5 times by s2idle in the short sleeps).
With state1 disabled on all eight CPUs (WFI only) and a 30 s alarm,
`PM: suspend entry` at 472.7 s and the wake at 501.9 s, on IRQ 56 with
INT2 `0c` (alarm 0): the real alarm, 29 s in. So the events that end
every other sleep a second in come with the CPUs' power-down during
s2idle, most likely from the firmware's side of that power-down, not
from the PMIC drivers. One sleep so far; next: a series of longer
sleeps with state1 off (a sleep hook can turn it off before the sleep
and on after), their hang rate, and the drain against the short-sleep
figures.

Our idle states match stock's: the same three `c2` states with
`arm,psci-suspend-param = <0x10000>` (core power-down) and stock's
latencies. What stock has besides them is Samsung's `exynos-cpupm`
node: cluster power-down per cluster (`cpd_cl0` for CPUs 0-3, `cpd_cl1`
for 4-6, `cpd_cl2` for 7) and `sicd` across all eight (system idle
clock down), which its vendor driver coordinates with the APM. Mainline
has no such driver, so when every CPU sits in core power-down in s2idle,
the firmware sees a state stock's kernel would have announced to it
first. That the PMIC's false events come from there is an inference,
not a measurement. The device package's fix for now is a sleep hook
that allows only WFI for the length of a sleep (`omamobile-device-tegu`
0.2.5, #255).

**Ten sleeps with the hook (2026-11-01, same release, device-tegu 0.2.5
installed from its branch).** `systemctl suspend` with a 60 s RTC alarm,
ten times: nine slept 55-61 s, entry to exit, and woke on IRQ 56 (the
alarm); the seventh's resume took longer than the 8 s the script waited
before measuring, so its length isn't known, but it woke and the hook
restored the states. `suspend_stats` 6 to 15 successes, 0 failures; no
UFS errors, no ACPM timeouts. The screen stayed lit through these sleeps
(`systemctl suspend` skips the shell's blanking), so no drain figure
from this run.
- **What the five clean cycles rule out, and what they don't.** Five in a
row with the storage counters at zero says the UFS resume path is not
broken per se; the afternoon failure and this hang are both the sixth or
Expand Down