Thank you, Krister. Your explanation is consistent with what I observed
when upgrading from 6.8.0-1061-oracle to 6.8.0-1062-oracle.

Regarding the performance on 1061, I don't have a quantitative baseline
from before the incident, just my subjective observation of routine
operations like apt update.

This OCI Ampere A1 instance has been running a small production
Discourse forum for several years, and its responsiveness had previously
been more than satisfactory.

The significant performance problems became apparent during a recent
incident on Nov 5, 2026. Rolling back to 1061 restored service, but the
instance still feels slower than it did historically.

I only started benchmarking process creation during the investigation,
so I cannot establish whether the approximately 8 seconds per 100
process launches on 1061 represents a new regression or whether that
overhead was already present.

The perf results suggest that the remaining overhead is concentrated
around page-table updates and TLB synchronization, but I agree this may
be a separate issue.

I'll investigate whether I can recover the kernel version and
configuration from before the incident through the system's package
history.

Thank you for helping identify the cause of the 1062 regression.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2170644

Title:
  arm64: munmap performance regression

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2170644/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to