Re: [PATCH v2 2/2] fs/resctrl: Give a monitor domain the counters the groups already use

From: Babu Moger

Date: Wed Nov 18 2026 - 10:58:07 EST


Hi Jinjie,

On 11/15/26 22:02, Jinjie Ruan wrote:


在 2026/11/14 0:14, Ben Horgan 写道:
Hi Jinjie,

Hi Ben,


On 13/11/2026 08:35, tizi4590 wrote:
From: Jinjie Ruan <ruanjinjie@xxxxxxxxxx>

With assignable MBM counters, resctrl can be mounted while all CPUs of
a monitor domain go offline and one of them comes back, either through
CPU hotplug or because hibernation takes the non-boot CPUs offline. The
domain is destroyed and re-created, and the new domain starts with no
counters assigned. Counters are otherwise only assigned when a group is
created, when rdtgroup_assign_cntrs() takes a free counter in each
monitor domain for each MBM event, so no code revisits the groups that
already exist.

Those groups read "Unassigned" on the new domain, on any system with
assignable MBM counters. A counter assigned to them later through
mbm_L3_assignments is the lowest free one in that domain, so the same
group can end up with a different counter ID on different domains. For
example, with the counter ID each group uses for mbm_total_bytes and
"_" for none:

domain 0 domain 1
mount, mkdir g1 default=0 g1=1 default=0 g1=1
all CPUs of domain 1 offline default=0 g1=1 (destroyed)
one CPU of domain 1 online default=0 g1=1 default=_ g1=_
assign g1 on domain 1 default=0 g1=1 default=_ g1=0

Each domain has its own counters, so g1 can use counter 0 on domain 1
and counter 1 on domain 0 without affecting the other domain, but until
it is assigned on domain 1 it reads Unassigned there.

I brought this up before and it was assessed that no change was required:
https://lore.kernel.org/lkml/171f92a3-fa2e-44fb-b0d1-1250c72e31d1@xxxxxxx/

Do you have any particular reason for wanting this behaviour?

I discovered this issue during the testing process and feel it is not
very reasonable.

This patch doesn't assign anything new - it only restores the counter ID
the group already uses on a domain that stayed online.

So with the patch, Babu's two cases (`2=e;3=e` and `2=_;3=e`) both come
back exactly as they were. The only case it can't know is a group that
used a different ID on the domain that went away, and there it leaves
the event unassigned rather than guessing.

I see this patch is trying to address two separate issues:

1. Automatically assigning counters when a domain comes online

I understand the motivation here. It is reasonable to expect counters to be assigned when a monitor domain comes online and counters are available. However, monitor domain online/offline events are typically driven by CPU hotplug, hibernation, or other user-initiated actions. In these cases, the user is generally aware of the state transition and can explicitly assign counters after the domain is online if monitoring is required.

2. Retaining the same counter ID across domains

I do not see why this is necessary. The design of assignable MBM counters intentionally treats counters as domain-local resources and provides flexibility to maximize counter availability across groups. Counter IDs are an implementation detail and are not visible to users. A group using a different counter ID in another domain does not affect functionality or monitoring results. Therefore, I do not see a strong justification for preserving the same counter ID when a domain is recreated. Given the above, I am not convinced the added complexity is justified.

Thanks,

Babu