Dear Maintainer,

* What led up to the situation?

The machine (ThinkPad T14 Gen 4 AMD, 21K3CTO1WW, BIOS R2FET58W 1.38) was
hibernated (suspend-to-disk) after roughly 8.8 hours of uptime. On the
following boot I resumed from the hibernation image.

* What exactly did you do (or not do) that was effective (or ineffective)?

Nothing unusual: a regular hibernate/resume cycle, as I do routinely on
this machine. Resume from hibernation has otherwise been working with
this kernel.

* What was the outcome of this action?

The resume itself completed, but shortly afterwards the kernel panicked
("Fatal exception in interrupt") in an EFI runtime services call and the
machine froze. The Oops shows a supervisor instruction fetch from an
unmapped page (PTE 0) at RIP 0xfffffffed6fe0468, i.e. inside the EFI
runtime services virtual mapping region rather than kernel text. The
faulting task was a kworker on the efi_rts_wq workqueue running
efi_call_rts, and immediately afterwards efi_pstore failed with
"pstore: backend (efi_pstore) writing error (-16)", so the crashing call
was most likely a SetVariable invocation on behalf of efi-pstore.

* What outcome did you expect instead?

A normal resume from hibernation without a kernel panic; at worst a
disabled EFI runtime services interface (efi=noruntime-like behaviour)
rather than a fatal exception.

Additional notes:

- This has happened once so far. I will report back if it recurs.
- Note the panic occurred on linux-image-6.19.8+deb13-amd64
  (6.19.8-1~bpo13+1, trixie-backports); the reportbug run below was made
  from a newer kernel (7.0.9+deb13-amd64), so the "Kernel:" line in the
  system information does not reflect the kernel that crashed.
- Hibernation is set up with swap on dm-crypt (dm_crypt/md in use, see
  module list in the Oops).

Control: tags -1 + moreinfo

If things are happening only once then we cannot really help. But two
things:

- Upgrade to the most recent kernel possible. This is for now
  7.0.12-2~bpo13+1 in trixie-backports.

- Run a memory test, as there might be some indication of defective
  memory.

For now I tag the bug moreinfo, please report back if you re-encounter
the issue.

Regards,
Salvatore