Re: [PATCH v1 0/3] mm/damon: Introduce a huge page collapsing mechanism using auto tuning

[Gutierrez Asier]

Hi SJ,

On 6/20/2026 8:11 PM, Gutierrez Asier wrote:
> Hi SJ,
> 
> So sorry, I missed your email. I just found it. Sorry for the late answer.
> 
> On 6/17/2026 4:44 AM, SeongJae Park wrote:
>> On Tue, 16 Jun 2026 15:03:13 +0000 <gutierrez.asier@xxxxxxxxxxxxxxxxxxx> wrote:
>>
>>> From: Asier Gutierrez <gutierrez.asier@xxxxxxxxxxxxxxxxxxx>
>>>
>>> Overview
>>> ========
>>>
>>> This patch set introduces a new autotuning which allows to collapse
>>> hot regions into hugepages.
>>>
>>> Motivation
>>> ==========
>>>
>>> Since TLB is a bottleneck for many systems[1], a way to optimize TLB
>>> misses (or hits) is to use huge pages. Unfortunately, using "always"
>>> in THP leads to memory fragmentation and memory waste. For this reason,
>>> most application guides and system administrators suggest to disable THP.
>>>
>>> Currently DAMON has DAMOS_HUGEPAGE, DAMOS_NONHUGEPAGE and DAMOS_COLLAPSE.
>>> However, there is no way to tune the settings. It will collapse all the
>>> hot regions that meet the access pattern. If the server is a bare metal
>>> database or big data server, this will also lead to eventual fragmentation.
>>>
>>> Additionally, currently THP is set globally. Ideally, there should be a
>>> way to control which tasks can use huge pages.
>>
>> Could you please reword for prctl(PR_SET_THP_DISABLE) like per-process control
>> cases, as we discussed [1] on RFC v3?
>>
>>>
>>> Solution
>>> ========
>>>
>>> DAMON has now a way to autotune some of the variables and adjust quotas
>>> automatically, so that DAMON is fired only under the right circumstances.
>>> It would be nice to have something similar, but for huge pages.
>>>
>>> A new autotuning quota goal[2], damos_hugepage_mem_bp, is introduced,
>>> which checks the huge page consumption to total memory consumption. This
>>> new quota mechanism reuses current autotuning architecture.
>>>
>>> A new sample module (SAMPLE_DAMON_HPAGE) is introduced to demonstrate
>>> the use of huge pages collapse autotuning. The goal is to collapse hot
>>> regions of a given process into huge pages. The sample module launches
>>> a kdamond thread for a certain task provided by the user through
>>> taget_pid module argument. Hugepage goal autotuning will automatically
>>> adjust the aggressiveness of hot region collapses.
>>>
>>> This sample module also has a user autotuning knob which allows the
>>> user to adjust the aggressiveness of page collapsing.
>>>
>>> Benchmarks
>>> ==========
>>>
>>> Huge page collapse autotuning was tested in a physicial machine with
>>> MariaDB 10.5.29 and sysbench as the benchmark framework.
>>>
>>> The hugepage module was set up in the following way:
>>>
>>> # echo 1000 > min_age
>>> # echo 1000 > quota_percentage_hugepage
>>
>> I guess this is the quota goal?  What is the unit?  I guess it is aparently not
>> percentage?  The name doesn't sound like very consistent or intuitive.  How
>> about hugepage_mem_bp or target_hugepage_mem_bp?
> Right, we agreed to change the name. I will correct it.
>>> # echo $(pidof mariadbd) > taget_pid
>>> # echo on > enabled
>>>
>>> The goal was to achieve 5% of the total memory used as hugepage.
>>
>> I guess this is what the above example is setting using
>> 'quotta_percentage_hugepage'?  If so, it means the unit is 1/20000 ?  Is this
>> correct...?
> I actually set it to 500. I will update the cover letter.
>>> Since the database was not very big, we may not be able to achieve
>>> high amount of huge pages per total memory consumption ratio.
>>
>> I believe this patch series will work as you explained.  But, it seems bit
>> weird to show a test result that doesn't demonstrate what this patch is aimed
>> to achive.  Could you increase the size of the database?  IIRC, you were able
>> to show the percentage is over-achived case in an early version.
> Actually, this is what I got using the TEMPORAL quota goals. With the regular
> quota goals, it actually over-achieves the goal.
> 
> Is this an actual bug in the TEMPORAL quota goal?
>>>
>>> The table below shows the memory consumption over time. Timestamp is in
>>> second and the memory usage in is MBytes. Gaps in the timestamp means
>>> that no changes in the hugepage consumption happened over that period
>>> of time in MB. The total used memory is calculated as
>>> mem_total - mem free. The huge page used is calculated as
>>> huge_page_anon + huge_page_shmem + huge_page_file. The table also
>>> shows the huge pages to total memory ratio.
>>>
>>> Hugepage autotune benchmark:
>>> +-----------+----------------+----------------+----------------------+
>>> | timestamp | total mem used | huge page used | percentage hugepage  |
>>> +-----------+----------------+----------------+----------------------+
>>> | 0         | 3044.988281    | 0              | 0%                   |
>>> | 22        | 3160.207031    | 2              | 0.06%                |
>>> | 30        | 3250.90625     | 4              | 0.12%                |
>>> | 69        | 3781.238281    | 6              | 0.16%                |
>>> | 71        | 3822.226563    | 8              | 0.21%                |
>>> | 72        | 3846.578125    | 10             | 0.26%                |
>>> | 73        | 3852.402344    | 12             | 0.31%                |
>>> | 74        | 3868           | 14             | 0.36%                |
>>> | 75        | 3881.84375     | 104            | 2.68%                |
>>> | 275       | 4194.175781    | 106            | 2.52%                |
>>> +-----------+----------------+----------------+----------------------+
>>> After second 275, no more pages are collapsed into hugepages
>>>
>>>
>>> THP (always) benchmark:
>>> +-----------+----------------+----------------+---------------------+
>>> | timestamp | total mem used | huge page used | percentage hugepage |
>>> +-----------+----------------+----------------+---------------------+
>>> |         1 |    4489.320313 |            184 |         4.098615986 |
>>> |        15 |    4581.871094 |            214 |         4.670580984 |
>>> |        30 |    4757.742188 |            376 |         7.902908253 |
>>> |        45 |    4937.574219 |            558 |         11.30109595 |
>>> |        60 |    5147.867188 |            728 |         14.14177898 |
>>> |        75 |      5407.0625 |            918 |         16.97779524 |
>>> |        95 |    5668.796875 |           1040 |         18.34604455 |
>>> |       105 |    5723.839844 |           1056 |         18.44915352 |
>>> |       115 |     5736.84375 |           1072 |         18.68623317 |
>>> |       125 |    5732.042969 |           1088 |         18.98101612 |
>>> |       186 |    5753.601563 |           1184 |         20.57841488 |
>>> |       246 |    5746.398438 |           1280 |         22.27482159 |
>>> |       306 |    5752.128906 |           1376 |         23.92157795 |
>>> |       367 |      5772.5625 |           1472 |         25.49994045 |
>>> |       427 |    5832.019531 |           1568 |         26.88605536 |
>>> |       488 |    5813.246094 |           1664 |         28.62428277 |
>>> |       548 |    5807.621094 |           1760 |         30.30500736 |
>>> |       598 |    5841.253906 |           1822 |         31.19193292 |
>>> |       669 |    5982.160156 |           1854 |         30.99214918 |
>>> |       931 |    5946.605469 |           1868 |         31.41287933 |
>>> |       981 |    6020.207031 |           1896 |         31.49393352 |
>>> |       991 |    5988.445313 |           1910 |         31.89475566 |
>>> |      1011 |    5988.570313 |           1926 |         32.16126554 |
>>> |      1032 |    6016.039063 |           1936 |         32.18064211 |
>>> |      1575 |    6057.289063 |           1968 |         32.48978181 |
>>> |      1606 |    6026.167969 |           2000 |         33.18858702 |
>>> +-----------+----------------+----------------+---------------------+
>>> I ignored some points to make the table shorter. Anyway, the amount
>>> of memory consumption, total and huge pages, is a lot higher than
>>> with DAMON hugepage autotuning.
>>
>> Could you further clarify why it is, and what this means
> Memory fragmentation. I will add information about memory fragmentation
> in the next cover letter. >>
>>>
>>> Performance:
>>> Baseline (no THP, module off) -> 18,162.45 transactions per second
>>> Hugepage autotune -> 18,211.82 transactions per second (+0.27% improvement)
>>> THP always -> 18,388.3 (+1.24%)
>>> THP madvise -> 18,179.25 (+0.09%)
>>>
>>> Improvement is due to lower TLB misses
>>
>> So this result says THP always is much better than the Hugepage autotune in
>> terms of the performance.  Maybe you want to claim Hugepage autotune is better
>> in terms of the memory efficiency?  Could you please clarify further?
> It's better than THP "never", but worse than THP "always". THP "always" is worse
> in terms of memory consumption, "always" is worse.
>>>
>>> Patches Sequence
>>> ================
>>> Patch 1 -> Introduce DAMOS_QUOTA_HUGEPAGE_MEM_BP and autotuning
>>> Patch 2 -> Module that demonstrates how to use
>>>            DAMOS_QUOTA_HUGEPAGE_MEM_BP and DAMOS_QUOTA_GOAL_TUNER_TEMPORAL
>>> Patch 3 -> Support for DAMOS_QUOTA_HUGEPAGE_MEM_BP in sysfs-schemes
>>>
>>> Changes from previous versions
>>> ==============================
>>> RFC 4[3] -> v1
>>>   - Renamed config to SAMPLE_DAMON_HPAGE, file to hpage.c and
>>>     functions to damon_sample_hpage_...
>>>   - Make the module depend on TRANSPARENT_HUGEPAGE, since
>>>     the module will need some THP functions anyway
>>>   - Removed documentation, since this is just a sample module
>>>   - Removed DAMOS_QUOTA_HUGEPAGE_MEM_BP from
>>>     damos_sysfs_add_quota_score
>>>   - Added a short description of the module in Kconfig
>>
>> Thank you for continuing this work!
>>
>> [...]
>>
>>> [1] https://dl.acm.org/doi/pdf/10.1145/3307650.3322227
>>> [2] https://lore.kernel.org/e67f05ad-dbb9-45e6-ba30-b167a99ac67d@xxxxxxxxxxxxxxxxxxx
>>> [3] https://lore.kernel.org/20260611150244.3454699-1-gutierrez.asier@xxxxxxxxxxxxxxxxxxx
>>> [4] https://lore.kernel.org/20260604150338.501128-1-gutierrez.asier@xxxxxxxxxxxxxxxxxxx
>>> [5] https://lore.kernel.org/20260522145518.158910-1-gutierrez.asier@xxxxxxxxxxxxxxxxxxx
>>> [6] https://lore.kernel.org/20260522171210.900B11F00A3D@xxxxxxxxxxxxxxx
>>> [7] https://lore.kernel.org/20260522171633.AAF5B1F000E9@xxxxxxxxxxxxxxx
>>> [8] https://lore.kernel.org/20260430134139.2446417-1-gutierrez.asier@xxxxxxxxxxxxxxxxxxx
>>> [9] https://lore.kernel.org/all/20260430154338.E22E6C2BCB3@xxxxxxxxxxxxxxx/
>>
>> [1] https://lore.kernel.org/9f9e2159-5a6b-496f-9633-fa06c0217948@xxxxxxxxxxxxxxxxxxx
>>
>>
>> Thanks,
>> SJ
>>
>> [...]
>>
> 
> SJ, once again, sorry for the late answer. Please, disregard my new patch set, I will fix
> it with your feedback.
> 

I didn't get a reply to this email.

I will update the cover letter and submit a new patch set soon.

-- 
Asier Gutierrez
Huawei