Two more updates: another confirmed occurrence of the GPU-reset-during-
hibernate-resume signature, plus a distinct new failure mode (silently
orphaned desktop session, no suspend ever logged) — and frequency data
showing this class of bug is a regular, recurring problem rather than an
occasional edge case.
## Update 1 (2026-09-03): another GPU reset during hibernation resume,
kernel 7.0.0-27-generic
Same signature as the 2026-08-19 update above:
amdgpu 0000:04:00.0: ring sdma0 timeout, signaled seq=296951, emitted
seq=296953
amdgpu 0000:04:00.0: GPU reset begin!. Source: 1
amdgpu 0000:04:00.0: [drm] *ERROR* dpcd_set_link_settings:1122/1127/1155:
core_link_write_dpcd (...) failed
amdgpu 0000:04:00.0: GPU pre asic reset failed with err, -22
amdgpu 0000:04:00.0: MODE2 reset
amdgpu 0000:04:00.0: GPU reset succeeded, trying to resume
amdgpu 0000:04:00.0: VRAM is lost due to GPU reset!
amdgpu 0000:04:00.0: [drm] *ERROR* DPM disable vcn failed, ret = -62.
This happened immediately on hibernation exit (image had frozen
successfully, plenty of free-page margin, so not the ENOMEM case below).
The machine kept running afterward in a degraded graphics state but
became fully unresponsive roughly an hour later, with no orderly
shutdown logged — required a hard power-off.
## Update 2 (2026-09-08): new failure mode — desktop session silently
lost, suspend request never reaches logind, machine runs hot for hours,
battery drained
This is different from the amdgpu ring-timeout pattern above. Timeline:
- ~19:00 — user triggers "Suspend" from the Plasma menu. No corresponding "The
system will suspend now!" line from systemd-logind, no "PM: suspend entry" —
the request never reached logind at all. No error of any kind logged either.
- 1-2 hours later — laptop found running hot (fans presumably active, not
asleep). A second suspend attempt is made; again nothing logged.
- ~21:48 — an explicit, orderly shutdown finally happens (clean
systemd-poweroff sequence). Right as this shutdown starts, kwin_wayland logs:
kwin_core: Failed to open /dev/dri/renderD128 device (Unknown object
'/org/freedesktop/login1/session/_33'.)
kwin_core: Failed to open /dev/input/event5 device (Unknown object
'/org/freedesktop/login1/session/_33'.)
i.e. logind no longer recognized the Plasma session's own session
object as valid — consistent with the interactive session having
silently detached/orphaned itself well before the shutdown, explaining
why neither suspend attempt could have worked: the session invoking them
was already broken.
- Next morning: lid-open did not wake anything (machine was genuinely
off by then, from the 21:48 shutdown), and the battery was found drained
— consistent with several hours of full operation (screen/fans active)
between the first failed suspend (~19:00) and the eventual shutdown
(~21:48), rather than genuine low-power suspend.
No amdgpu errors, no OOM, no crash of any kind logged around the failed
suspend attempts — the session simply stopped being able to act on sleep
requests, silently.
## Frequency data (last 3 weeks, 2026-08-19 to 2026-09-09)
Grepping this machine's journal for each known failure signature:
- GPU reset during suspend/hibernate (ring timeout + MODE2 reset): 3 separate
days (Aug 19, Sep 1, Sep 3)
- Orphaned/lost logind session (this update's new pattern): 4 separate days
(Aug 20, Aug 28, Sep 1, Sep 8)
- Hibernation aborting cleanly with ENOMEM ("PM: hibernation: Error -12
creating image", safe failure, no data loss): 8 separate days (Aug 22, 25, 28,
29, 30, 31 [x2], Sep 3, Sep 6)
Combined, some form of sleep/hibernate failure has occurred on the large
majority of weeks since this was first reported, across multiple
distinct failure modes (GPU driver hang, orphaned session, memory
pressure) and across both kernel builds tested (7.0.0-27 and 7.0.0-28).
This no longer looks like an occasional edge case — power-state
transitions are broadly unreliable on this AMD Barcelo iGPU (amdgpu)
platform under current Ubuntu 25.10 kernels.
## Hardware/software (unchanged)
Dell Inspiron 14 5425, AMD Ryzen 7 5825U, AMD Barcelo iGPU (amdgpu),
Ubuntu 25.10, KDE Plasma 6.5, Wayland.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2161312
Title:
System hangs on suspend (s2idle) with kernel 7.0.0-28-generic, works
fine on 7.0.0-27
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2161312/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
