[PATCH 0/4] [focal/impish linux-aws] AWS: Hibernate resume crashes when platform changes

Tim Gardner tim.gardner at canonical.com
Sat Mar 19 16:10:58 UTC 2022

BugLink: https://bugs.launchpad.net/bugs/1965002

Even though these are all cherry-picks, there is enough fuzz that 'git am' won't do
the job whereas 'git cherry-pick' does. Hence the 4 patches that are nearly identical.

SRU Justification


At present, if a virt instance is hibernated and restarted in a different environment,
the instance will crash. We have seen situations where customers have to enable support
cases to recover an instance. The proposed patch is supposed to enable the kernel to
detect the change and reboot the instance instead of crashing. In the ACPI specification;
there is a 'hardware signature' field in the ACPI tables, and the OS can reboot cleanly
if/when that signature changes, instead of attempting the resume and crashing.

commit f6c46b1d62f8ffbf2cf6eb43ab0d277bd3f7e948 linux-next

[Test Case]

Hibernate an AWS instance, resume on a different platform. Positive test results from Amazon engineers.

[Where things could go wrong]

Resume currently crashes when a hibernated instance is migrated to a new machine. It can't
get much worse, unless the initial hibernate fails.

[Other Info]

SF: #00330794

More information about the kernel-team mailing list