Skip to content

device/husky SUSPEND-PLAN: H0b and H1 passed on the third phone; NOTES and ROADMAP leftovers - #835

Merged
derekross merged 4 commits into
derekross:mainfrom
gitColinX:husky-suspend-h0b
Nov 18, 2026
Merged

derekross merged 4 commits into
derekross:mainfrom
gitColinX:husky-suspend-h0b

Conversation

@gitColinX

@gitColinX gitColinX commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor

@derekross, H0b from the merged runbook ran on my unit today, attended, over USB. It passed. Documentation only: the result goes into section 4, and the status line now says H0b passed and H1 to H4 have not run.

The run

Published omamobile at 82f31d3a7, omamobile-device-husky 0.1.13, 85 % and charging, slot B refilled to 3 earlier the same day. No sleep: suspend_stats read 0 / 0 before and after.

  • Unload: pre passed (spm_lvl 5, no /sys/class/nfc). modprobe -r brcmfmac_wcc brcmfmac pcie_gs201 returned 0, none of the three was left in lsmod, the PCI bus was empty and no wlp interface remained. brcmutil and phy_zuma_pcie stayed loaded with no users.
  • Reload, 5 s later: modprobe pcie_gs201 && modprobe brcmfmac returned 0. Both PCI functions came back in D0, the firmware loaded, and wlp1s0 was connected to the same 5 GHz network 14 s after the reload, with three pings out of three.
  • Kernel log: only the brcmfmac lines this kernel also prints at boot (Ignoring unexpected firmware channel, one p2pdev timeout, one join ... firmware error -14 before it associated, Received unexpected monitor pkt). No PCIe, AER or UFS line; running, no failed units.
  • One thing to know for later scripts: the interface came back on phy1 where it had been phy0.

Next

H1 (pm_test freezer, devices and platform through systemctl suspend) is the next stage. I'll run it attended and on the cable, with slot B at two or more retries. If you'd rather see anything else first, tell me.

tools/pr-check origin/main: all ok.

@gitColinX

gitColinX commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor Author

@derekross, H1 ran 20 minutes after H0b, same kernel and package, and passed too. I've added it as a second commit so the two results sit together; the title now undersells it, so feel free to retitle at merge.

  • pm_test freezer, devices and platform, each through systemctl suspend: suspend_stats 0 to 3 successes, 0 failures, three suspend entry (s2idle)/exit pairs with the 5 s debug wait.
  • platform: suspend of devices 43 ms, late 2.6 ms, noirq 1.9 ms; resume 69 ms, the UFS device W-LUN the slowest at 41 ms.
  • UFS printed one dme-set: attr-id 0x4021 val 0x1 failed 0 retries after the devices and the platform resume, the line Lef's resumes show. No error handler run, no hibern8 or I/O line, root still read-write.
  • Afterwards: running, no failed units, Bluetooth, haptics and the GPU there, Wi-Fi back 14 s after the reload. Not checked: touch after resume (nobody touched the screen).

H2, the first real s2idle with the RTC wake, is next. I'll wait for your word on these two before running it.

gitColinX and others added 3 commits November 15, 2026 22:36
The Wi-Fi and PCIe unload and reload on their own, no sleep, on the
published omamobile at 82f31d3a7 with device-husky 0.1.13: all three
modules gone and the bus empty after the unload; after the reload the
BCM4398 back in D0 and wlp1s0 connected again in 14 s, with only the
brcmfmac noise this kernel also prints at boot. H1 to H4 have not run.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
pm_test freezer, devices and platform through systemctl suspend:
three successes, no failures, devices back in 69 ms, and UFS's one
"dme-set: attr-id 0x4021" line after the devices and platform
resumes, as on Lef's phone, with no error handler run. H2 to H4 have
not run.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
…ekross#243 and derekross#244

- NOTES no longer calls SUSPEND-PLAN a draft (derekross#802).
- The ROADMAP Suspend row says mem_sleep read "s2idle [deep]" before
  kernel derekross#244 and reads "[s2idle] deep" on the published kernel (derekross#802).
- The ROADMAP Time row no longer says "No RTC": the S2MPG14's RTC is
  rtc0 since kernel derekross#243 (asked for on that PR).
- The 2026-11-15 NOTES entry records /proc/cmdline's
  mem_sleep_default=s2idle with mem_sleep (asked for on kernel derekross#244),
  and the RTC's probe lines, wakealarm and idle interrupt count (asked
  for on kernel derekross#243).

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
@gitColinX gitColinX changed the title device/husky SUSPEND-PLAN: H0b passed on the third phone device/husky SUSPEND-PLAN: H0b and H1 passed on the third phone; NOTES and ROADMAP leftovers Nov 16, 2026
@gitColinX

gitColinX commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor Author

@derekross, a third commit, documentation only, picking up leftovers so they don't get lost:

I've retitled the PR to match. H2 still waits for your word.

@derekross

Copy link
Copy Markdown
Owner

Thank you, Colin. This is a clean record: the published kernel by commit, the package version, the counts before and after, the timings from pm_print_times, and what wasn't checked said in so many words. I checked the kernel commit against the branch, #243 and #244 against the board file, and the two #802 leftovers; all right. The phy1 note will save someone a script bug.

Three small changes before I merge, all so that "passed" means what the runbook's Pass lines say:

  1. H1. Its Pass line lists display, touch, GPU, Bluetooth and haptics working afterwards. Your result says touch wasn't checked, doesn't mention the display, and for haptics has the input device present. Please make the status line and the heading "H1 passed (touch, display and a played haptic effect not checked; carried into H2)", or check the three on a later boot and add them with the date.

  2. The UFS line. Section 2 says any UFS error on resume stops everything, and H1 logged dme-set: attr-id 0x4021 val 0x1 failed 0 retries twice and went on. I agree with going on: 0x4021 is the peer device ID that ufs-exynos sets when the link comes back, and the 9a's SLEEP-PLAN has the same line on every healthy wake. But the plan should say so before H2, in one sentence in section 2: that line at resume is expected and not a stop, anything else from ufshcd is. H2's after helper doesn't grep for it, so it will pass by unseen otherwise.

  3. H0b's noise list. The Pass line allows two brcmfmac lines and the run had four. Since the other two (brcmf_join_v1 … firmware error -14, Received unexpected monitor pkt) also print at boot on this kernel, please add them to the list in section 3, where H1 to H4 reuse it.

Smaller, in the same push if you like: the two result paragraphs say "over USB" but not "attended" (your PR text does); the Kernels paragraph and section 2 still say device-husky 0.1.10 where both stages ran on 0.1.13; and the H1 result says 25 s between levels where the runbook's loop sleeps 20.

On the rule "one stage per go-ahead": I had said H0b could run, and H1 followed twenty minutes later in the same sitting. It was attended, on the cable, with no real sleep, and you stopped before H2, so no harm; but I meant my go-ahead per stage, and the plan should say whose it is. Please add that to section 2.

For H2, two things first. #806 is merging as device-husky 0.1.14 (SuspendState=freeze); I'd like H2 to H4 on that package once it is in the development channel, named in the plan. And Lef found that a hand copy of 10-omamobile.conf in /etc/systemd/sleep.conf.d/ had been hiding the package's file on his phone, so please run systemd-analyze cat-config systemd/sleep.conf once and put what it shows in the record. With this PR merged and those two done, I'll give the word for H2 here.

From the review on derekross#835:

- H1's status line and heading say what was not checked: touch, the
  display and a played haptic effect, carried into H2, whose Pass line
  now names them.
- Section 2 names the one UFS line that is expected at resume
  (dme-set: attr-id 0x4021, the peer device ID ufs-exynos sets when
  the link comes back) and says any other UFS error is a stop. The
  recovery bullet, the stop conditions and H1's Pass line say the same.
- Section 3's brcmfmac noise list has the two other lines the H0b run
  printed, which this kernel also prints at boot.
- The two results say "attended"; the package versions are the ones
  the stages ran on (0.1.13) and will run on (0.1.14 or later, derekross#806);
  the H1 loop sleeps 25 s, as the run's script did, and the reconnect
  comment matches.
- Section 2 says whose go-ahead a stage waits for.
- Section 4 records systemd-analyze cat-config systemd/sleep.conf on
  this unit: no /etc/systemd/sleep.conf.d/, so nothing hides the
  package's drop-in.

Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
@gitColinX

Copy link
Copy Markdown
Contributor Author

@derekross, thank you. All of it is in one commit on top:

  1. H1. The status line and the heading say "H1 passed (touch, display and a played haptic effect not checked; carried into H2)", and H2's Pass line names the played effect.
  2. The UFS line. Section 2 has it as expected and not a stop, with the reason (0x4021, T_PEERDEVICEID, and the 9a's healthy wakes). The recovery bullet, the stop conditions and H1's Pass line now say the same. I left the after helper alone; say if you want dme-set added to its grep so the line lands in each stage's record.
  3. H0b's noise list. Section 3 has all four lines.

The smaller ones: both results say "attended"; the Kernels paragraph says 0.1.13 for H0b and H1 and 0.1.14 or later for H2 to H4; the H1 loop says 25 s, which is what the run's script used, and the reconnect comment matches.

On the go-ahead: understood, and sorry for running ahead. Section 2 now says it is yours, one per stage, and records that H1 ran on H0b's.

For H2: section 4 has systemd-analyze cat-config systemd/sleep.conf from today on 0.1.13. The package's drop-in is the only one, and there is no /etc/systemd/sleep.conf.d/. I'll read it again once 0.1.14 is installed, and wait for your word here.

@derekross

Copy link
Copy Markdown
Owner

Thank you, Colin. All of it is there: the H1 status with what wasn't checked, the dme-set line as expected with its reason in section 2 and the same words in the recovery bullet, the stop conditions and H1's Pass line, the four noise lines, "attended" in both results, the package versions, the 25 s, and whose go-ahead it is. Thank you also for the sleep configuration read. Merging this.

On the after helper: leave its grep as it is, since the rule is that it comes back empty and the expected line would break that. If you'd like the line on record for each stage, add a separate count next to it (dmesg | grep -c 'dme-set: attr-id 0x4021') when you write up H2.

For H2: 0.1.14 is on main now and will be in the development channel after the next nightly build. Once it is installed, please read systemd-analyze cat-config systemd/sleep.conf again as you planned, and I'll give the word for H2 here.

@derekross
derekross merged commit 7b9f41b into derekross:main Nov 18, 2026
2 checks passed
@gitColinX

Copy link
Copy Markdown
Contributor Author

@derekross, installed and read again.

  • Package: omamobile-device-husky 0.1.16-1, built on the phone from main (17f9a3cd) with tools/phone-install-package, since the development channel still has 0.1.13. No restart: same boot, kernel 82f31d3a7.
  • systemd-analyze cat-config systemd/sleep.conf shows two files: /etc/systemd/sleep.conf with nothing set, and the package's /usr/lib/systemd/sleep.conf.d/10-omamobile.conf (owned by 0.1.16-1) with SuspendState=freeze, MemorySleepMode=s2idle, AllowHibernation=no, AllowHybridSleep=no and AllowSuspendThenHibernate=no. There is still no /etc/systemd/sleep.conf.d/.
  • Wi-Fi power save still reads off: 0.1.15's file is in place, but NetworkManager hasn't reloaded and the phone hasn't restarted. H0b and H1 ran with it off. Tell me if you want it on for H2.

Ready for H2 on your word. I'll add the dme-set count to the write-up as you suggested.

One more question: on 11-13 you asked me to hold the Omarchy 4.5 update on this phone. Is that still your advice?

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