Re: DAMON reporting more hot memory on huge pages (was: "Re: [RFC PATCH v2 0/9] mm/damon: hardware-sampled access reports")

On Thu, 17 Sep 2026 10:06:37 +0800 Lian Wang [off-list ref] wrote:

Hi SJ,

Thank you for thinking further about this and for suggesting the damon_prep
direction.  Let me first clarify the real-world use case and correct one
expectation that I created in my previous mail.

SXF develops a hyperconverged infrastructure platform.  The workloads running
in the guests are customer workloads, and their types, memory sizes and access
distributions are not controlled by us.  Oracle is the workload in the case
that exposed the problem, but we cannot define one Oracle access distribution
as representative of the platform or provide one generally applicable
workload-specific THP performance number.

The platform has two requirements:

  1. Host THP is enabled by default.  Large mappings generally provide useful
     translation and virtualization benefits, so disabling THP merely to make
     monitoring easier is not an acceptable general solution.

  2. We also need an estimate of each VM's active memory as an input to later
     memory-tiering configuration.  If a small number of distributed 4 KiB
     accesses makes many 2 MiB mappings appear fully active, the hot-tier
     requirement can be substantially overestimated.  VM sizes vary, so the
     absolute error can become large on a large VM when accesses are spread
     over many huge mappings.

Thank you for sharing these.  Especially the second requirement is very
interesting to me.

As you may already know, DAMON can provide not only the observability but also
such tiering like operation.  SK hynix and Micron are developing their tiering
solution using DAMON.  I also developed [1] TPP-based solution.  I'm wondering
if they considered directly using tiering solution instead of doing observation
and operation separately.

The reported setup is a KVM/QEMU VM whose large guest-RAM allocation is backed
by a shared tmpfs file.  Oracle runs in the guest, host THP is enabled, and
host DAMON vaddr monitoring targets the QEMU process.  The field observation
is that, for the same business memory-use case, the measured hot proportion is
much higher with THP.  The guest-side tmpfs 4 KiB/2 MiB writer is a controlled
diagnostic for isolating this mechanism, not a model of the complete Oracle
access distribution.

The DAMON setup is:

  operations=vaddr
  monitoring_attrs/nr_regions/min=500
  monitoring_attrs/nr_regions/max=2000
  monitoring_attrs/intervals/sample_us=500000
  monitoring_attrs/intervals/aggr_us=20000000
  monitoring_attrs/intervals/update_us=60000000
  schemes/nr_schemes=1
  schemes/0/action=stat
  schemes/0/access_pattern/age/min=0
  schemes/0/access_pattern/age/max=18446744073709551615
  schemes/0/access_pattern/nr_accesses/min=1
  schemes/0/access_pattern/nr_accesses/max=18446744073709551615
  schemes/0/access_pattern/sz/min=0
  schemes/0/access_pattern/sz/max=18446744073709551615

The hot proportion is derived from the bytes selected by this stat-only scheme
(at least one observed access in a 20-second aggregation window) relative to
the monitored QEMU scope.  It is used for later configuration, not as a
request for DAMOS to split or migrate memory.

Thank you very much for sharing these details.  I now get much clearer picture.

I think the DAMON setup may have rooms to improve.  When someone asks me
recommeded setup for access pattern observability, I recommend to use intervals
auto-tuning and read idle time percentiles, similar to DAMON_STAT.  If you
haven't tried that, I'd recommend to try it.  I presented [2] this basic idea
at Lst year LPC refereed track.  Hopefully that could give you more ideas.  If
you tried it and found some problems, and that resulted you in this setup,
knowing the detailed context would be helpful for me.

[...]

quoted

I still want to better understand the real world use case before digging into
the specific solution.  But, I was thinking perf event based monitoring might
not feasible for the case.

We agree that perf-event feasibility should be established rather than
assumed.  We will test the address space reported while a KVM vCPU is running,
per-VM attribution, coverage and loss before relying on that path.  If you had
a specific feasibility concern in mind, please let us know so that we can
include it in the test.

I'm still in a very early stage of perf event based monitoring prototyping.
Hence I have no expertise to find a real concern at the moment.  I was just
thinking it _might_.

quoted

That is, we can request DAMON to do some preparation action for the data
attribute sampling, using damon_prep.  At the moment, set_pgidle prep action is
supported.  Maybe we could add a new prep action, say, split_pmd?  Then, the
following sampling memory access check probe (would be a kind of "allow
pidle_unset" probe filter) will be able to show the access on exactly the pte
accessed bit, not the pmd accessed bit.

Breaking pmd would cause the overhead, but it would be capped by the
max_nr_regions.  So the user could at least consider about the tradeoff.  I
think this is simpler and more intuitive than DAMOS-based pmd breaking, at
least.

This placement makes sense to us.  A split_pmd preparation for access sampling
is clearer than exposing PMD breaking as a DAMOS action.  We understand it as
refining the mapping used for observation, without splitting the underlying
huge folio, and then letting the following probe observe PTE accessed bits.

We will follow this idea and validate it.  In particular, we will measure the
active-memory result and mapping cost, check whether max_nr_regions also bounds
the cumulative number of mappings that remain PTE-mapped over time, and verify
whether splitting the host process PMD is sufficient in the QEMU case or KVM's
large secondary mapping remains the limiting layer.  We will compare it with
the perf-event path, stat-only first, without proposing a user-visible DAMOS
split action or physical folio split.

Sounds good, looking forward to the progress.

Please rest assured that we see this as a long-term technical discussion and
collaboration, not an attempt to rush one mechanism into the tree.  We will
continue sharing both positive and negative results.  I also plan to present
the scenario, evidence and tradeoffs in more detail at LPC.  Please feel free
to share any concerns or ideas at any time; we are happy to keep iterating on
them together.

No worry.  Nothing needs to be rushed.

This topic has now become broader than Ravi's hardware-sampled report series.
If you think it would keep the discussions clearer, I can start a separate
thread focused on huge-page observation and the damon_prep approach, while
keeping perf-event implementation and testing comments on Ravi's thread.

Indeed we might hijacking Ravi's thread and adding unexpected mails to
recipients' inbox.  Please feel free to drop recipients or start a new thread,
whatever you prefer.

Please let me know whether this clarifies the intended use and whether our
reading of the split_pmd damon_prep proposal matches what you had in mind.

Yes, this is very helpful.  I have raised followup questions above, though ;)
No pressure.  I understand not every detail can always be revealed, and
sometimes it takes time.  Please answer only if you want, and when you get the
chance.

[1] https://lkml.kernel.org/r/20250420194030.75838-1-sj@kernel.org
[2] https://lpc.events/event/19/contributions/2075/


Thanks,
SJ

[...]