[Bug 2051835] Re: [24.04 FEAT] Memory hotplug vmem pages (s390x)

This bug is awaiting verification that the linux-azure/6.8.0-1008.8
kernel in -proposed solves the problem. Please test the kernel and
update this bug with the results. If the problem is solved, change the
tag 'verification-needed-noble-linux-azure' to 'verification-done-noble-
linux-azure'. If the problem still exists, change the tag 'verification-
needed-noble-linux-azure' to 'verification-failed-noble-linux-azure'.


If verification is not done by 5 working days from today, this fix will
be dropped from the source code, and this bug will be closed.


See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how
to enable and use -proposed. Thank you!


** Tags added: kernel-spammed-noble-linux-azure-v2 verification-needed-noble-linux-azure

--
You received this bug notification because you are subscribed to linux
in Ubuntu.
Matching subscriptions: Bgg, Bmail, Nb
https://bugs.launchpad.net/bugs/2051835

Title:
[24.04 FEAT] Memory hotplug vmem pages (s390x)

Status in Ubuntu on IBM z Systems:
Fix Released
Status in linux package in Ubuntu:
Fix Released
Status in linux source package in Noble:
Fix Released

Bug description:
The current s390 specific memory hotplug backend allocates 'struct page' management structures for all standby memory regions, when those standby regions are detected at ipl time.
The reason for this is, that setting standby online memory is supposed to succeed especially in memory constrained environments, since lack of free memory is likely the reason why additional memory is brought online.
If in such cases 'struct pages' would have to be allocated before memory could be brought up, this would likely fail, and contradict the whole rationale of memory hotplug.

However pre-allocating memory for 'struct pages' comes with the
downside that for highly unbalanced ratios of online/standby memory a
system might even fail to ipl, because there is not enough memory
available for all possible struct pages which are required for standby
memory.

The idea is to improve the situation: when standby memory is brought
online, the memory for struct pages (and maybe other management
structures) required for this new memory area should be taken from the
online memory, instead of pre-allocating them.

This would solve the problems with unbalanced ratios as described
above.

Note: there are intentions to tell customers that they should always
define the maximum size of standby memory for their LPAR activation
profiles. This would allow for maximum flexibility for all LPARs
during runtime, given that the amount of standby memory cannot be
changed during runtime.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-z-systems/+bug/2051835/+subscriptions