Same failure on different AMD hardware, with power figures.
ASUS ExpertBook BM1503CDA, Ryzen 5 7535U (Rembrandt-R), BIOS BM1503CDA.314.
Kubuntu 26.04, 7.0.0-30-generic, Plasma/Wayland. External monitor permanently
attached to HDMI-A-1 (amdgpu). Clean install 15 Aug 2026; the same hardware
with the same peripherals ran for months on earlier releases without this.
After a normal shutdown the machine keeps drawing about 3 W. It is
intermittent - twice in five nights, with an identical configuration in
between that stayed clean, which fits the race you described.
19->20 Aug, reproduced
22:13:40 33.2 Wh remaining (upower history-rate 11.174 W x
history-time-empty 10689 s; sample 30 s earlier agrees, 33.26 Wh)
22:14:27 systemd-poweroff.service finished, poweroff.target reached
~08:47 AC connected, battery reads 5% while already charging
~10 h 33 min, ~33 Wh => ~3.15 W average. RTC had reset (booted
stamped 2026-07-27 23:44).
18->19 Aug, control, did not reproduce
23:52:52 energy_now 33.357 Wh, poweroff 23:56:50, boot 07:43:15 at
32.599 Wh. 0.758 Wh over 7 h 46 min, most of it the ~4 min awake before
poweroff (~7.8 W idle) plus POST. Residual leakage <= 0.01 W.
17->18 Aug, reproduced
100% at poweroff 22:34:24, pack empty by morning, RTC likewise reset.
~3.3 W over ~11 h.
Ruled out: not suspend - grepping the affected boot for "PM: suspend entry",
s2idle, "Freezing user space", systemd-suspend and sleep.target returns zero
hits, and the full poweroff sequence is present ending at "Journal stopped".
Not a wake - journalctl --list-boots shows no boot in between. Not the USB
peripherals - the two dongles declare 100 mA and 50 mA, and the clean night
had the same two attached. Not AC state at shutdown - 17 Aug was on AC, 18
and 19 Aug on battery, one clean and two not.
Not verified: I did not check the power LED or the fan on either night, so I
can report the draw but not the "stays powered on" observation directly. No
A/B against 7.0.0-14 yet.
apport-collect data attached above, and the kernel log of the boot that ended
in the failed poweroff (19 Aug, ends 22:14:27) is attached separately -
apport-collect only covers the current boot. Happy to test 7.0.0-14 and any
-proposed kernel with the ISM fix; given the race I would run several nights
per kernel and report per-night energy rather than a single pass/fail
** Attachment added: "dmesg-2026-08-19-failed-poweroff.txt"
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2163121/+attachment/5993786/+files/dmesg-2026-08-19-failed-poweroff.txt
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2163121
Title:
[Regression] Laptop fails to power off completely when HDMI is
connected in kernel 7.0.0-28
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2163121/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
