Skip to content

arm64: dts: exynos: google: zuma-husky: Default to suspend-to-idle - #244

Merged
derekross merged 1 commit into
derekross:omamobilefrom
gitColinX:omamobile-zuma-husky-s2idle-default
Nov 12, 2026
Merged

derekross merged 1 commit into
derekross:omamobilefrom
gitColinX:omamobile-zuma-husky-s2idle-default

Conversation

@gitColinX

@gitColinX gitColinX commented Nov 11, 2026 •

Copy link
Copy Markdown
Contributor

The kernel line @derekross asked for in derekross/omarchy-mobile#656: mem_sleep_default=s2idle in zuma-husky.dts, as #238 has it for the 8a. One commit, one file: the argument on the end of bootargs and the 8a's three comment lines above it.

Why it matters on this phone

  • /sys/power/mem_sleep reads s2idle [deep] on my Pixel 8 Pro (G1MNW), on husky-test-20261109-t1 plus my RTC test commits.
  • A deep suspend has now happened on it, unplanned. The phone was off its charger overnight and the battery fell to about 4 %. upower then asked logind for a suspend (suspend requested from client PID 1340 ('upowerd')), and the last line of that boot's journal is PM: suspend entry (deep). It did not wake, and came back only when it was restarted. The root still had omamobile-device-husky 0.1.8, so there was no drop-in either.
  • The 0.1.10 drop-in covers systemctl suspend. This line covers what writes mem itself, rtcwake -m mem among them.

Not tested: I have not booted a kernel with this line. I can build it on the t1 tree and report /proc/cmdline and /sys/power/mem_sleep from the phone if that is wanted before merge. Suspend to idle itself is still not something I have run on this unit; the runbook Derek described in #656 comes first.

Checks: tools/kernel-pr-check --build --merge passes: it merges cleanly onto fc7086a68, checkpatch --strict has no errors (its warnings are the long bootargs line, as on the 8a, and the Co-Authored-By trailer's spelling), and 10 device trees build with W=1 and pass dtbs_check with no new warnings. No C files are touched.

The firmware advertises PSCI SYSTEM_SUSPEND, so the kernel offers deep
sleep and makes it the default: /sys/power/mem_sleep reads
"s2idle [deep]" on the Pixel 8 Pro. Linux has no zuma sleep sequence
yet, so a deep suspend does not wake: on a Pixel 8 Pro (G1MNW) whose
battery ran low, upower asked for a suspend, the log ends at
"PM: suspend entry (deep)" and the phone stayed down until it was
restarted. Pass mem_sleep_default=s2idle, as the Pixel 8a does, so a
suspend that names no state stays in s2idle. The omamobile-device-husky
package asks systemd for s2idle since 0.1.10; this covers anything that
writes "mem" directly, rtcwake -m mem among them.

Not booted yet with this line; the same argument is on the Pixel 8a,
the Pixel 7 Pro, the Pixel 7 and the Pixel 9a.

Suggested-by: Derek Ross <derekross@gmail.com>
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
Signed-off-by: Colin Lundholm <265166644+gitColinX@users.noreply.github.com>
@derekross

Copy link
Copy Markdown
Owner

Thank you, @gitColinX, for following up on #656 the same day, and for the account of what happened on your phone. A battery-critical suspend that went to deep and needed a restart is exactly the case this line is for, and it's useful to have it on record for zuma rather than only inferred from the Pixel 7 Pro.

I compared it with #238: the argument is in the same place, the comment is the 8a's word for word, and the built device tree differs from today's in that one string and nothing else. The gate passes on today's omamobile (clean merge onto fc7086a68fca, signed off, the prefix is the file's own, checkpatch with only the long bootargs line and the trailer spelling, device trees build with W=1 and no new warnings). CI is waiting for my approval, and I ran the schema check by hand in the meantime: nothing new.

The commit says plainly that it hasn't booted yet, and I'm fine merging it that way once CI is green: it is one kernel parameter on the end of a string, the 8a showed it arriving in /proc/cmdline on the same SoC, and without it anything that writes mem directly still sends this phone into a sleep it can't leave. When you next boot a tree with it, please read /proc/cmdline and /sys/power/mem_sleep (it should say [s2idle] deep) and put both in device/husky/NOTES.md.

Two things to know, neither for this PR:

  • device/husky/overlays/klog.dtso replaces the whole bootargs string and doesn't have the argument, so a boot with that overlay is back on deep for a direct write. It wants the same few words the next time someone touches it.
  • This doesn't make suspend-to-idle tested on the 8 Pro, as you say. The short runbook from #656 still comes first (Wi-Fi has no sleep hook in the package, spm_lvl at 5, NFC not polling, attended, on the cable). Once arm64: dts: exynos: google: zuma-husky: the main PMIC's interrupt, so the Pixel 8 Pro has an RTC alarm #243 is in, that runbook also has an RTC alarm to wake the phone with.

#243 touches the same file in another place; the two merge cleanly in either order.

@derekross
derekross merged commit 5c8b9e5 into derekross:omamobile Nov 12, 2026
1 check passed
@gitColinX

gitColinX commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor Author

@derekross, the readings you asked for, late: on published omamobile at 82f31d3a7 on my unit, /proc/cmdline has mem_sleep_default=s2idle and /sys/power/mem_sleep reads [s2idle] deep. Both are going into device/husky/NOTES.md through derekross/omarchy-mobile#835. klog.dtso still lacks the argument; I haven't touched it.

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