mm/damon: introduce data attributes only monitoring
Subsystems (4)
SJ Park
maintainercover mm/damon: introduce data attributes only monitoring
TL;DR: Introduce a way to get DAMON's best effort accuracy monitoring of user-demanding non-access data attributes.
Background
==========
DAMON was initially designed for only access monitoring. It turned out users want to get the information together with more data attributes. For example, some users want to know how much of a hot memory region belongs to huge pages or specific cgroups. Page level properties based monitoring was introduced with commit 626ffabe67c2 ("mm/damon: clarify trying vs applying on damos_stat kernel-doc comment") to fill the gap. Because it works only at snapshot level and snapshot capturing in the mode can induce high overhead, commit 45c49d9fd608 ("mm/damon/core:
introduce struct damon_probe") introduced data attributes monitoring.
Data attributes monitoring treats the attributes as only additional and subordinate information. Data access monitoring is always turned on, and regions are adjusted for best accuracy of the access information. In some cases, users may be primarily interested in the attributes more than the access. They might even not care about the access information at all. Because DAMON treats data accesses as the only primary information, such users cannot get high quality attributes information.
Design and Implementation
=========================
Introduce another way for treating data attributes as the primary information. Add 'weight' property to each data attribute probe. When any of the weights are set, the mode is enabled. Data access monitoring is completely turned off in the mode. For region adjustment, the weighted sum of probe hit counters is used instead of the nr_accesses.
Using the weights, users can specify to what attributes they are interested in to what degree. DAMON will adjust the regions and provide the best-effort quality monitoring that is optimized for the user demands.
Extend damon_operations for efficient use of probe hits. Update regions merge and kdamond main logic to support the new mode. Add a new struct field and a sysfs file for API callers and ABI users, respectively.
Test
====
On ~7 GiB memory idle system, run a simple AI-assisted program. The program allocates and faults 2 GiB anonymous pages. Then, it does nothing but wait until the user terminates it. Hence, the system ~2 GiB of anonymous pages with no active accesses.
Monitor the distribution of the anonymous pages using DAMON attributes monitoring mode, using DAMON user-space tool, damo [1].
$ sudo ./damo start --probe_filter allow anon
$ sudo ./damo report access --dont_merge_regions
heatmap: 00000000000000000000000000000000000000000000000399999995111111146666666666666666
# min/max temperatures: -2,470,000,000, -1,620,000,000, column size: 99.800 MiB
intervals: sample 5 ms aggr 100 ms (max access hz 200)
# <start> <size> <freq> <age> <probe hits>
0 4.000 KiB 79.840 MiB 0 hz 24.700 s 2
1 79.844 MiB 718.562 MiB 0 hz 24.700 s 8
2 798.406 MiB 793.148 MiB 0 hz 24.700 s 7
3 1.554 GiB 797.828 MiB 0 hz 24.700 s 7
4 2.333 GiB 794.668 MiB 0 hz 24.600 s 8
5 3.109 GiB 791.117 MiB 0 hz 24.500 s 0
6 3.882 GiB 785.312 MiB 0 hz 24 s 2
7 4.649 GiB 787.867 MiB 0 hz 16.200 s 6
8 5.418 GiB 784.477 MiB 0 hz 23.300 s 6
9 6.184 GiB 783.820 MiB 0 hz 18.200 s 9
10 6.950 GiB 797.730 MiB 0 hz 18.900 s 7
11 7.729 GiB 69.625 MiB 0 hz 18.900 s 0
memory bw estimate: 0 B per second
total size: 7.797 GiB
record DAMON intervals: sample 5 ms, aggr 100 msNote that the line after the line starting with "intervals:" is not provided by the current version of 'damo'. I manually added the legends line for easier understanding of these results.
Each of the 12 lines after the legend line shows the DAMON-found regions. Each line shows 1) index of the region, 2) start address of the region, 3) size of the region, 4) access frequency of the region, 5) age (how long the access frequency on the region was kept) of the region, and finally 6) the probe hit count.
Because data access is the primary information that adjusts region for, and there is only nearly zero access on the system, regions are naively adjusted with the same size. Still <probe hits> show different distribution of the anonymous pages, but it is obviously very rough information.
Switch to the attributes only mode and show how it changes the picture:
$ sudo ./damo tune --probe_filter allow anon --probe_weight 100
$ sudo ./damo report access --dont_merge_regions
heatmap: 88888888888888888889888999999889999999000004888888888888889999988888898888888888
# min/max temperatures: -4,430,000,000, 0, column size: 99.800 MiB
intervals: sample 5 ms aggr 100 ms (max access hz 200)
# <start> <size> <freq> <age> <probe hits>
0 4.000 KiB 60.445 MiB 0 hz 700 ms 0
1 60.449 MiB 1.363 MiB 0 hz 600 ms 18
2 61.812 MiB 144.000 KiB 0 hz 0 ns 1
3 61.953 MiB 1.922 MiB 0 hz 2.400 s 19
4 63.875 MiB 12.133 MiB 0 hz 200 ms 0
[...]
500 5.132 GiB 8.000 KiB 0 hz 2 m 15.800 s 20
501 5.132 GiB 8.000 KiB 0 hz 2 m 16.200 s 0
502 5.132 GiB 16.000 KiB 0 hz 2 m 16.900 s 20
503 5.132 GiB 24.000 KiB 0 hz 2 m 14.200 s 0
504 5.132 GiB 8.000 KiB 0 hz 2 m 14.900 s 20
[...]
923 7.534 GiB 126.637 MiB 0 hz 0 ns 6
924 7.658 GiB 252.000 KiB 0 hz 54.800 s 20
925 7.658 GiB 142.242 MiB 0 hz 300 ms 0
memory bw estimate: 0 B per second
total size: 7.797 GiB
record DAMON intervals: sample 5 ms, aggr 100 msAs expected, regions are adjusted to provide the best accurate picture for the anonymous pages distribution (<probe hits>). The region 0 (60.445 MiB memory from the address 4.000 KiB) has nearly zero anonymous pages. The region 1 (1.363 MiB memory from the address 60.449 MiB) is nearly full with anonymous pages. Region 500 (8 KiB memory from the address 5.132 GiB) is certainly two anonymous pages.
Future Work
===========
Attributes only monitoring disables access monitoring. We will enable that in future, by extending the supported attributes to include data accesses. This patch series, and the future work are parts of the ongoing project [2] for extending DAMON. The project aims to extend DAMON with primitives other than page table accessed bits such as AMD IBS, Intel PEBS, and Arm SPE, to provide more powerful and detailed information like per-CPUs/threads/reads/writes monitoring.
Patches Sequence
================
Patch 1 introduces damon_probe->weight for specifying the weights of each attribute. Patches 2-6 extends apply_probe() damon_ops callback to efficiently support the new mode. Patch 7 extends regions merge function to work with probe hits in the mode. Patch 7 also introduces the function for detecting the mode enablement but always returns false, for safe and incremental changes. Patches 8-11 incrementally update kdamond_fn() to support the mode. Patch 12 completes the mode detection function implementation, so that the new mode really works. Patch 13 introduces a new sysfs file for ABI users. Finally, patches 14-16 respectively updates design, usage and ABI documents for the new feature and interfaces.
[1] https://github.com/damonitor/damo
[2] https://lore.kernel.org/[email protected]/
Changelog
=========
Changes from RFC
- RFC: https://lore.kernel.org/[email protected]
- Add kernel-doc comment for damon_probe->weight.
- Fix typos in kernel-doc, design and usage documents.
SJ Park (16):
mm/damon/core: introduce damon_probe->weight mm/damon/core: ask apply_probes() ops callback to set sampling address mm/damon/paddr: set samples in apply_probes() if requested mm/damon/core: ask apply_probe() to return max probe hits weighted sum mm/damon/core: implement damon_probe_hits_wsum() mm/damon/paddr: respect return_max_wsum
mm/damon/core: extend merge function to work with probe hits mm/damon/core: disable access monitoring when probe weights are set mm/damon/core: set samples in apply_probes() if probe weights are set mm/damon/core: s/max_nr_accesses/max_merge_score/ in kdamond_fn() mm/damon/core: get merge threshold from probe hits when weights are
set
mm/damon/core: implement damon_has_probe_weight() mm/damon/sysfs: implement probe/weight file Docs/mm/damon/design: document attrs-only monitoring Docs/admin-guide/mm/damon/usage: document weight sysfs file Docs/ABI/damon: document probe weight file
.../ABI/testing/sysfs-kernel-mm-damon | 6 + Documentation/admin-guide/mm/damon/usage.rst | 11 +- Documentation/mm/damon/design.rst | 25 ++++ include/linux/damon.h | 12 +- mm/damon/core.c | 107 +++++++++++++++--- mm/damon/paddr.c | 11 +- mm/damon/sysfs.c | 25 ++++ mm/damon/tests/core-kunit.h | 13 ++- 8 files changed, 185 insertions(+), 25 deletions(-)
base-commit: 6a2e3b53ae2be579022206bf5d43b87dff04fe12
Signature · 1 line
2.47.3
SJ Park
maintainerRFC PATCH 1/16 mm/damon/core: introduce damon_probe->weight
Add a new field, weight to damon_probe struct. The field is used to specify the degree of the API caller's interest to the data attribute of the probe.
--- include/linux/damon.h | 2 ++ mm/damon/core.c | 3 +++ 2 files changed, 5 insertions(+)
include/linux/damon.h +2 -0
@@ -760,10 +760,12 @@ struct damon_filter {/** * struct damon_probe - Data region attribute probe. * +* @weight: Relative priority of the attribute for this probe. * @filters: Filters for assessing if a given region is for this probe. * @list: Siblings list. */ struct damon_probe { +unsigned int weight;struct list_head filters;struct list_head list; };
mm/damon/core.c +3 -0
@@ -166,6 +166,7 @@ struct damon_probe *damon_new_probe(void) p =kmalloc_obj (*p);if (!p)return NULL ; + p->weight =0 ;INIT_LIST_HEAD (&p->filters);INIT_LIST_HEAD (&p->list);return p;
@@ -1641,6 +1642,7 @@ static int damon_commit_probes(struct damon_ctx *dst, struct damon_ctx *src)damon_for_each_probe_safe (dst_probe, next, dst) { src_probe =damon_nth_probe (i++, src);if (src_probe) { + dst_probe->weight = src_probe->weight; err =damon_commit_filters (dst_probe, src_probe);if (err)return err;
@@ -1657,6 +1659,7 @@ static int damon_commit_probes(struct damon_ctx *dst, struct damon_ctx *src)if (!new_probe)return -ENOMEM ;damon_add_probe (dst, new_probe); + new_probe->weight = src_probe->weight; err =damon_commit_filters (new_probe, src_probe);if (err)return err;
Signature · 1 line
2.47.3
SJ Park
maintainerRFC PATCH 2/16 mm/damon/core: ask apply_probes() ops callback to set sampling address
In reply to SJ Parkprepare_access_checks() DAMON ops callback sets the monitoring sampling address per region. In future, DAMON will be able to call only apply_probes(). In this case, applyy_probes() may need to do the sampling address setup, to minimize unnecessary regions iteration. Update the protocol for the request.
--- include/linux/damon.h | 5 +++-- mm/damon/core.c | 2 +- mm/damon/paddr.c | 2 +- 3 files changed, 5 insertions(+), 4 deletions(-)
include/linux/damon.h +3 -2
@@ -647,7 +647,8 @@ enum damon_ops_id {* It should also return max number of observed accesses that made as a result * of its update. The value will be used for regions adjustment threshold. * @apply_probes should apply the data attribute probes to each region and -* accordingly update the probe hits counter of the region. +* accordingly update the probe hits counter of the region. It should also +* &set sampling_addr of each region if ``set_samples`` is true. * @get_scheme_score should return the priority score of a region for a scheme * as an integer in [0, &DAMOS_MAX_SCORE]. * @apply_scheme is called from @kdamond when a region for user provided
@@ -665,7 +666,7 @@ struct damon_operations {void (*update)(struct damon_ctx *context);void (*prepare_access_checks)(struct damon_ctx *context);unsigned int (*check_accesses)(struct damon_ctx *context); -void (*apply_probes)(struct damon_ctx *context); +void (*apply_probes)(struct damon_ctx *context,bool set_samples);int (*get_scheme_score)(struct damon_ctx *context,struct damon_region *r,struct damos *scheme); unsigned long (*apply_scheme)(struct damon_ctx *context,
mm/damon/core.c +1 -1
@@ -3675,7 +3675,7 @@ static int kdamond_fn(void *data)if (ctx->ops.check_accesses) max_nr_accesses = ctx->ops.check_accesses (ctx);if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx); + ctx->ops.apply_probes (ctx, false);if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {
mm/damon/paddr.c +1 -1
@@ -166,7 +166,7 @@ static bool damon_pa_filter_pass(phys_addr_t pa, struct folio *folio,return pass; } -static void damon_pa_apply_probes (struct damon_ctx *ctx) +static void damon_pa_apply_probes (struct damon_ctx *ctx,bool set_samples) {struct damon_target *t;struct damon_region *r;
Signature · 1 line
2.47.3
SJ Park
maintainerRFC PATCH 3/16 mm/damon/paddr: set samples in apply_probes() if requested
In reply to SJ Parkapply_probe() callback implementation in DAMON_PADDR is ignoring set_samples parameter. Respect it.
--- mm/damon/paddr.c | 4 ++++ 1 file changed, 4 insertions(+)
mm/damon/paddr.c +4 -0
@@ -178,6 +178,10 @@ static void damon_pa_apply_probes(struct damon_ctx *ctx, bool set_samples)phys_addr_t pa;struct folio *folio; +if (set_samples) + r->sampling_addr =damon_rand (ctx, r->ar.start, + r->ar.end); + pa =damon_pa_phys_addr (r->sampling_addr, ctx->addr_unit); folio =damon_get_folio (PHYS_PFN (pa));
Signature · 1 line
2.47.3
SJ Park
maintainerRFC PATCH 4/16 mm/damon/core: ask apply_probe() to return max probe hits weighted sum
In reply to SJ Parkcheck_accesses() DAMON ops callback returns the maximum nr_accesses of regions. DAMON core uses it to calculate a reasonable region merge threshold. The core will need to adjust regions for not nr_accesses but probe hits weighted sum in future. For that, the core needs to know the maximum weighted sum of the regions. Update the protocol for the task.
--- include/linux/damon.h | 7 +++++-- mm/damon/core.c | 2 +- mm/damon/paddr.c | 4 +++- 3 files changed, 9 insertions(+), 4 deletions(-)
include/linux/damon.h +5 -2
@@ -648,7 +648,9 @@ enum damon_ops_id {* of its update. The value will be used for regions adjustment threshold. * @apply_probes should apply the data attribute probes to each region and * accordingly update the probe hits counter of the region. It should also -* &set sampling_addr of each region if ``set_samples`` is true. +* &set sampling_addr of each region if ``set_samples`` is true. It should +* also return maximum probe hits weighted sum of regions if +* ``return_max_wsum`` is true. * @get_scheme_score should return the priority score of a region for a scheme * as an integer in [0, &DAMOS_MAX_SCORE]. * @apply_scheme is called from @kdamond when a region for user provided
@@ -666,7 +668,8 @@ struct damon_operations {void (*update)(struct damon_ctx *context);void (*prepare_access_checks)(struct damon_ctx *context);unsigned int (*check_accesses)(struct damon_ctx *context); -void (*apply_probes)(struct damon_ctx *context,bool set_samples); +unsigned int (*apply_probes)(struct damon_ctx *context, +bool set_samples,bool return_max_wsum);int (*get_scheme_score)(struct damon_ctx *context,struct damon_region *r,struct damos *scheme); unsigned long (*apply_scheme)(struct damon_ctx *context,
mm/damon/core.c +1 -1
@@ -3675,7 +3675,7 @@ static int kdamond_fn(void *data)if (ctx->ops.check_accesses) max_nr_accesses = ctx->ops.check_accesses (ctx);if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx, false); + ctx->ops.apply_probes (ctx, false, false);if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {
mm/damon/paddr.c +3 -1
@@ -166,7 +166,8 @@ static bool damon_pa_filter_pass(phys_addr_t pa, struct folio *folio,return pass; } -static void damon_pa_apply_probes (struct damon_ctx *ctx,bool set_samples) +static unsigned int damon_pa_apply_probes (struct damon_ctx *ctx, +bool set_samples,bool return_max_wsum) {struct damon_target *t;struct damon_region *r;
@@ -194,6 +195,7 @@ static void damon_pa_apply_probes(struct damon_ctx *ctx, bool set_samples)folio_put (folio); } } +return 0 ; }/*
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 04/16] mm/damon/core: ask apply_probe() to return max probe hits weighted sum
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Pre-existing issues:
- [High] Advancing physical addresses using addr += folio_size(folio) in large folio iteration paths skips memory when addr is not aligned to the start of the folio.
- [High] Static local variables used for caching are shared across concurrent kernel threads without synchronization.
--
commit 1ecdc723ec9bfdc97493a942107d11622114a694
Author: SJ Park <[email protected]>
mm/damon/core: ask apply_probe() to return max probe hits weighted sum
check_accesses() DAMON ops callback returns the maximum nr_accesses of regions. DAMON core uses it to calculate a reasonable region merge threshold. The core will need to adjust regions for not nr_accesses but probe hits weighted sum in future. For that, the core needs to know the maximum weighted sum of the regions. Update the protocol for the task.
Quoted 15 lines
mm/damon/paddr.c +2
-1@@ -166,7 +166,8 @@ static bool damon_pa_filter_pass(phys_addr_t pa, struct folio *folio,return pass; } -static void damon_pa_apply_probes (struct damon_ctx *ctx,bool set_samples) +static unsigned int damon_pa_apply_probes (struct damon_ctx *ctx, +bool set_samples,bool return_max_wsum) {struct damon_target *t;struct damon_region *r;
mm/damon/paddr.c static void damon_pa_apply_probes(struct damon_ctx *ctx, bool set_samples) · line 199
Quoted 5 lines
mm/damon/paddr.c +1
-0@@ -194,6 +195,7 @@ static void damon_pa_apply_probes(struct damon_ctx *ctx, bool set_samples)folio_put (folio); } } +return 0 ; }
[Severity: High]
This isn't a bug introduced by this patch, but while reviewing the changes in mm/damon/paddr.c, I noticed that __damon_pa_check_access() declares static local variables for caching:
mm/damon/paddr.c:__damon_pa_check_access() {
static phys_addr_t last_addr;
static unsigned long last_folio_sz = PAGE_SIZE;
static bool last_accessed;
...
}Because DAMON allows multiple contexts to be created and run concurrently, each in its own kdamond thread, could these static variables be modified concurrently without locking? Does this data race risk corrupting the cache state and misapplying access results between different monitoring contexts?
[Severity: High]
This is a pre-existing issue, but I also noticed a potential overshooting bug in multiple physical address iteration paths in mm/damon/paddr.c, such as in damon_pa_pageout():
mm/damon/paddr.c:damon_pa_pageout() {
...
addr += folio_size(folio);
...
}If a region start address is not aligned to the size of the folio (for instance, starting in the middle of a 2MB THP), advancing the unaligned addr by the full folio size will overshoot the folio boundary. Could this silently skip valid physical memory that falls in the gap?
The same pattern appears to be used in damon_pa_de_activate(), damon_pa_migrate(), and damon_pa_stat().
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=4
SJ Park
maintainerRe: [RFC PATCH v1.1 04/16] mm/damon/core: ask apply_probe() to return max probe hits weighted sum
Quoted 51 lines · On Mon, 06 Jul 2026 14:39:37 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Pre-existing issues:
- [High] Advancing physical addresses usingaddr += folio_size(folio)in large folio iteration paths skips memory whenaddris not aligned to the start of the folio.
- [High] Static local variables used for caching are shared across concurrent kernel threads without synchronization.
--commit
1ecdc723ec9bfdc97493a942107d11622114a694
Author: SJ Park <[email protected]>
mm/damon/core: askapply_probe()to return max probe hits weighted sum
check_accesses()DAMON ops callback returns the maximumnr_accessesof regions. DAMON core uses it to calculate a reasonable region merge threshold. The core will need to adjust regions for notnr_accessesbut probe hits weighted sum in future. For that, the core needs to know the maximum weighted sum of the regions. Update the protocol for the task.Quoted 15 lines
mm/damon/paddr.c +2
-1@@ -166,7 +166,8 @@ static bool damon_pa_filter_pass(phys_addr_t pa, struct folio *folio,return pass; } -static void damon_pa_apply_probes (struct damon_ctx *ctx,bool set_samples) +static unsigned int damon_pa_apply_probes (struct damon_ctx *ctx, +bool set_samples,bool return_max_wsum) {struct damon_target *t;struct damon_region *r;mm/damon/paddr.c static void damon_pa_apply_probes(struct damon_ctx *ctx, bool set_samples) · line 199
Quoted 5 lines
mm/damon/paddr.c +1
-0@@ -194,6 +195,7 @@ static void damon_pa_apply_probes(struct damon_ctx *ctx, bool set_samples)folio_put (folio); } } +return 0 ; }[Severity: High]
This isn't a bug introduced by this patch, but while reviewing the changes inmm/damon/paddr.c, I noticed that__damon_pa_check_access()declares static local variables for caching:mm/damon/paddr.c:__damon_pa_check_access() { static phys_addr_t last_addr; static unsigned long last_folio_sz = PAGE_SIZE; static bool last_accessed; ... }
Quoted 4 lines
Because DAMON allows multiple contexts to be created and run concurrently, each in its own kdamond thread, could these static variables be modified concurrently without locking? Does this data race risk corrupting the cache state and misapplying access results between different monitoring contexts?
Correct. I'm separately working [1] on this.
Quoted 12 lines
[Severity: High]
This is a pre-existing issue, but I also noticed a potential overshooting bug in multiple physical address iteration paths inmm/damon/paddr.c, such as indamon_pa_pageout():mm/damon/paddr.c:damon_pa_pageout() { ... addr += folio_size(folio); ... }
Quoted 7 lines
If a region start address is not aligned to the size of the folio (for instance, starting in the middle of a 2MB THP), advancing the unaligned addr by the full folio size will overshoot the folio boundary. Could this silently skip valid physical memory that falls in the gap?
The same pattern appears to be used in
damon_pa_de_activate(),damon_pa_migrate(), anddamon_pa_stat().
correct. I'm separately working [2] on this.
Quoted 3 lines
Signature · 2 lines
-- Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=4
[1] https://lore.kernel.org/[email protected]
[2] https://lore.kernel.org/[email protected]
Thanks,
SJ
SJ Park
maintainerRFC PATCH 5/16 mm/damon/core: implement damon_probe_hits_wsum()
In reply to SJ ParkWhen damon_probe->weight is set, the weighted sum of probe hits will be useful. It will be useful for not only the users but also DAMON internal logics like regions merging. Implement a function for calculating it.
--- include/linux/damon.h | 2 ++ mm/damon/core.c | 24 ++++++++++++++++++++++++ 2 files changed, 26 insertions(+)
include/linux/damon.h +2 -0
@@ -1012,6 +1012,8 @@ unsigned int damon_nr_accesses_mvsum(struct damon_region *r, struct damon_ctx *ctx);unsigned char damon_probe_hits_mvsum (int probe_idx,struct damon_region *r,struct damon_ctx *ctx); +unsigned int damon_probe_hits_wsum (struct damon_region *r,bool last, +struct damon_ctx *ctx);int damon_set_regions (struct damon_target *t,struct damon_addr_range *ranges,unsigned int nr_ranges,unsigned long min_region_sz);
mm/damon/core.c +24 -0
@@ -390,6 +390,30 @@ static bool damon_is_last_region(struct damon_region *r,return list_is_last (&r->list, &t->regions_list); } +/** +* damon_probe_hits_wsum() - Returns probe hits weighted sum of a region. +* @r: region to get the weigted sum of. +* @last: if the request is for last-window agggregated probe hits. +* @ctx: context of &r. +* +* Return: the weighted sum of probe hits of the region. +*/ +unsigned int damon_probe_hits_wsum (struct damon_region *r,bool last, +struct damon_ctx *ctx) +{ +struct damon_probe *probe; +unsigned int sum =0 ; +int i =0 ; + +damon_for_each_probe (probe, ctx) { +if (last) + sum += r->last_probe_hits[i++] * probe->weight; +else + sum += r->probe_hits[i++] * probe->weight; + } +return sum; +} +/* * Check whether a region is intersecting an address range *
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 05/16] mm/damon/core: implement damon_probe_hits_wsum()
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
New issues:
- [High] The calculation of the weighted sum in damon_probe_hits_wsum() is vulnerable to integer overflow, resulting in a silent wrap-around.
- [Low] Misspelled words in the kernel-doc comment for damon_probe_hits_wsum.
Pre-existing issues:
- [High] r->probe_hits[i] is an 8-bit unsigned char but serves as a sample counter across an aggregation interval, making it highly susceptible to wrap-around. --
commit 627b12cd883dbd832e93cc6b4f4e3373c791b6d0
Author: SJ Park <[email protected]>
mm/damon/core: implement damon_probe_hits_wsum()
This commit implements damon_probe_hits_wsum() to calculate the weighted sum of probe hits for a DAMON region. This weighted sum is intended to be used by users as well as DAMON internal logic like region merging.
[ ... ]
mm/damon/core.c static bool damon_is_last_region(struct damon_region *r, · line 396
Quoted 12 lines
mm/damon/core.c +4
-0@@ -390,6 +390,30 @@ static bool damon_is_last_region(struct damon_region *r,return list_is_last (&r->list, &t->regions_list); } +/** +* damon_probe_hits_wsum() - Returns probe hits weighted sum of a region. +* @r: region to get the weigted sum of. +* @last: if the request is for last-window agggregated probe hits.
[Severity: Low]
This isn't a bug, but there are spelling errors in this kernel-doc block.
"weigted" should be "weighted" and "agggregated" should be "aggregated".
Quoted 6 lines
mm/damon/core.c +6
-0+* @ctx: context of &r. +* +* Return: the weighted sum of probe hits of the region. +*/ +unsigned int damon_probe_hits_wsum (struct damon_region *r,bool last, +struct damon_ctx *ctx)
mm/damon/core.c static bool damon_is_last_region(struct damon_region *r, · line 412
Quoted 10 lines
mm/damon/core.c +10
-0+{ +struct damon_probe *probe; +unsigned int sum =0 ; +int i =0 ; + +damon_for_each_probe (probe, ctx) { +if (last) + sum += r->last_probe_hits[i++] * probe->weight; +else + sum += r->probe_hits[i++] * probe->weight;
[Severity: High]
Can the accumulation into sum silently overflow?
If a large probe weight is configured via sysfs (e.g., > 16,843,009), the multiplication of this 32-bit unsigned int weight by r->probe_hits[i] could wrap around when accumulated into sum.
This might corrupt the DAMON region merge and split logic that relies on this score.
[Severity: High]
This is a pre-existing issue, but can r->probe_hits[i] wrap around during long aggregation intervals?
The probe_hits array is defined as an 8-bit unsigned char. If the user configures aggr_interval and sample_interval such that their ratio exceeds 255 (for example, 1s aggregation and 1ms sample), the counter will silently wrap around to 0 without any bounds checking.
This could severely degrade the accuracy of DAMON's monitoring, and is exacerbated here by relying on the potentially corrupted counter for the new weighted sum logic.
Quoted 3 lines
mm/damon/core.c +3
-0+ } +return sum; +}
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=5
SJ Park
maintainerRe: [RFC PATCH v1.1 05/16] mm/damon/core: implement damon_probe_hits_wsum()
Quoted 37 lines · On Mon, 06 Jul 2026 14:41:34 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
New issues:
- [High] The calculation of the weighted sum indamon_probe_hits_wsum()is vulnerable to integer overflow, resulting in a silent wrap-around.
- [Low] Misspelled words in the kernel-doc comment fordamon_probe_hits_wsum.Pre-existing issues:
- [High]r->probe_hits[i]is an 8-bitunsigned charbut serves as a sample counter across an aggregation interval, making it highly susceptible to wrap-around. --commit
627b12cd883dbd832e93cc6b4f4e3373c791b6d0
Author: SJ Park <[email protected]>
mm/damon/core: implementdamon_probe_hits_wsum()This commit implements
damon_probe_hits_wsum()to calculate the weighted sum of probe hits for a DAMON region. This weighted sum is intended to be used by users as well as DAMON internal logic like region merging.[ ... ]
mm/damon/core.c static bool damon_is_last_region(struct damon_region *r, · line 396
Quoted 12 lines
mm/damon/core.c +4
-0@@ -390,6 +390,30 @@ static bool damon_is_last_region(struct damon_region *r,return list_is_last (&r->list, &t->regions_list); } +/** +* damon_probe_hits_wsum() - Returns probe hits weighted sum of a region. +* @r: region to get the weigted sum of. +* @last: if the request is for last-window agggregated probe hits. [Severity: Low]
This isn't a bug, but there are spelling errors in this kernel-doc block.
Quoted 1 line
"weigted" should be "weighted" and "agggregated" should be "aggregated".
Good finding, I will fix these in the next spin.
Quoted 21 lines
Quoted 7 lines
mm/damon/core.c +6
-0+* @ctx: context of &r. +* +* Return: the weighted sum of probe hits of the region. +*/ +unsigned int damon_probe_hits_wsum (struct damon_region *r,bool last, +struct damon_ctx *ctx)mm/damon/core.c static bool damon_is_last_region(struct damon_region *r, · line 412
Quoted 10 lines
mm/damon/core.c +10
-0+{ +struct damon_probe *probe; +unsigned int sum =0 ; +int i =0 ; + +damon_for_each_probe (probe, ctx) { +if (last) + sum += r->last_probe_hits[i++] * probe->weight; +else + sum += r->probe_hits[i++] * probe->weight;[Severity: High]
Can the accumulation into sum silently overflow?
Quoted 6 lines
If a large probe weight is configured via sysfs (e.g., > 16,843,009), the multiplication of this 32-bit unsigned int weight by r->probe_hits[i] could wrap around when accumulated into sum.
This might corrupt the DAMON region merge and split logic that relies on this score.
Yes, it can. But it causes only poor monitoring results for wrong user setup. Maybe we can improve it, but not necessarily right now. No blocker.
Quoted 5 lines
[Severity: High]
This is a pre-existing issue, but can r->probe_hits[i] wrap around during long aggregation intervals?
Quoted 8 lines
The
probe_hitsarray is defined as an 8-bit unsigned char. If the user configuresaggr_intervalandsample_intervalsuch that their ratio exceeds 255 (for example, 1s aggregation and 1ms sample), the counter will silently wrap around to 0 without any bounds checking.This could severely degrade the accuracy of DAMON's monitoring, and is exacerbated here by relying on the potentially corrupted counter for the new weighted sum logic.
Known low priority issue. No blocker.
Quoted 7 lines
Quoted 4 lines
mm/damon/core.c +3
-0+ } +return sum; +}Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=5
Thanks,
SJ
SJ Park
maintainerRFC PATCH 6/16 mm/damon/paddr: respect return_max_wsum
In reply to SJ Parkapply_probes() ops implementation in DAMON_PADDR is ignoring return_max_wsum. Respect it.
--- mm/damon/paddr.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)
mm/damon/paddr.c +5 -2
@@ -172,6 +172,7 @@ static unsigned int damon_pa_apply_probes(struct damon_ctx *ctx,struct damon_target *t;struct damon_region *r;struct damon_probe *p; +unsigned int max_wsum =0 ;damon_for_each_target (t, ctx) {damon_for_each_region (r, t) {
@@ -182,7 +183,6 @@ static unsigned int damon_pa_apply_probes(struct damon_ctx *ctx,if (set_samples) r->sampling_addr =damon_rand (ctx, r->ar.start, r->ar.end); - pa =damon_pa_phys_addr (r->sampling_addr, ctx->addr_unit); folio =damon_get_folio (PHYS_PFN (pa));
@@ -193,9 +193,12 @@ static unsigned int damon_pa_apply_probes(struct damon_ctx *ctx, }if (folio)folio_put (folio); +if (return_max_wsum) + max_wsum =max (damon_probe_hits_wsum (r, false, + ctx), max_wsum); } } -return 0 ; +return max_wsum; }/*
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 06/16] mm/damon/paddr: respect return_max_wsum
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [High] The probe_hits array in struct damon_region is of type unsigned char, causing silent wrap-around if the number of sampling intervals in an aggregation interval (aggr_samples) exceeds 255. --
commit d141f3812ddd7e9395e9f362f8ddc5918e7ab555
Author: SJ Park <[email protected]>
mm/damon/paddr: respect return_max_wsum
apply_probes() ops implementation in DAMON_PADDR is ignoring return_max_wsum. Respect it.
Quoted 4 lines
mm/damon/paddr.c +0
-0
[ ... ]
mm/damon/paddr.c static unsigned int damon_pa_apply_probes(struct damon_ctx *ctx, · line 202
Quoted 12 lines
mm/damon/paddr.c +4
-1@@ -193,9 +193,12 @@ static unsigned int damon_pa_apply_probes(struct damon_ctx *ctx, }if (folio)folio_put (folio); +if (return_max_wsum) + max_wsum =max (damon_probe_hits_wsum (r, false, + ctx), max_wsum); } } -return 0 ; +return max_wsum; }
[Severity: High]
This is a pre-existing issue, but I noticed something about the probe_hits array while reviewing this area. Does this code silently wrap around if the number of samples in an aggregation interval exceeds 255?
Looking at the damon_region structure in include/linux/damon.h, probe_hits is defined as an unsigned char array:
include/linux/damon.h:struct damon_region {
...
unsigned char probe_hits[DAMON_MAX_PROBES];
...
};In mm/damon/paddr.c:damon_pa_apply_probes(), this value is incremented on every sampling interval without bounds checking:
mm/damon/paddr.c:damon_pa_apply_probes() {
...
damon_for_each_probe(p, ctx) {
if (damon_pa_filter_pass(pa, folio, p))
r->probe_hits[i]++;
i++;
}
...
}If a user configures the sysfs settings such that the aggregation interval is significantly larger than the sample interval (for example, a 1ms sample interval and a 1s aggregation interval yields 1000 samples), will r->probe_hits[i] wrap around and start over from zero?
If so, it appears this might break DAMON region merging and monitoring statistics by operating on incorrect, wrapped-around numbers.
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=6
SJ Park
maintainerRe: [RFC PATCH v1.1 06/16] mm/damon/paddr: respect return_max_wsum
Quoted 61 lines · On Mon, 06 Jul 2026 14:44:25 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [High] Theprobe_hitsarray instruct damon_regionis of typeunsigned char, causing silent wrap-around if the number of sampling intervals in an aggregation interval (aggr_samples) exceeds 255. --commit
d141f3812ddd7e9395e9f362f8ddc5918e7ab555
Author: SJ Park <[email protected]>
mm/damon/paddr: respectreturn_max_wsum
apply_probes()ops implementation inDAMON_PADDRis ignoringreturn_max_wsum. Respect it.Quoted 4 lines
mm/damon/paddr.c +0
-0[ ... ]
mm/damon/paddr.c static unsigned int damon_pa_apply_probes(struct damon_ctx *ctx, · line 202
Quoted 12 lines
mm/damon/paddr.c +4
-1@@ -193,9 +193,12 @@ static unsigned int damon_pa_apply_probes(struct damon_ctx *ctx, }if (folio)folio_put (folio); +if (return_max_wsum) + max_wsum =max (damon_probe_hits_wsum (r, false, + ctx), max_wsum); } } -return 0 ; +return max_wsum; }[Severity: High]
This is a pre-existing issue, but I noticed something about theprobe_hitsarray while reviewing this area. Does this code silently wrap around if the number of samples in an aggregation interval exceeds 255?Looking at the
damon_regionstructure ininclude/linux/damon.h,probe_hitsis defined as an unsigned char array:include/linux/damon.h:struct damon_region { ... unsigned char probe_hits[DAMON_MAX_PROBES]; ... };In mm/damon/paddr.c:damon_pa_apply_probes(), this value is incremented on every sampling interval without bounds checking:
mm/damon/paddr.c:damon_pa_apply_probes() { ... damon_for_each_probe(p, ctx) { if (damon_pa_filter_pass(pa, folio, p)) r->probe_hits[i]++; i++; } ... }
Quoted 7 lines
If a user configures the sysfs settings such that the aggregation interval is significantly larger than the sample interval (for example, a 1ms sample interval and a 1s aggregation interval yields 1000 samples), will r->probe_hits[i] wrap around and start over from zero?
If so, it appears this might break DAMON region merging and monitoring statistics by operating on incorrect, wrapped-around numbers.
Known low priority issue. No blocker.
Quoted 3 lines
Signature · 2 lines
-- Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=6
Thanks,
SJ
SJ Park
maintainerRFC PATCH 7/16 mm/damon/core: extend merge function to work with probe hits
In reply to SJ ParkWhen probe weights are set, users may want DAMON monitoring results to be optimized for the weights. For that, regions adjustment should work for the weighted sum of probe hits. Extend damon_merge_regions_of() to detect if the weights are set, and work with probe hits in the case.
The weights setup detection function is incomplete. It always returns false. It is intentional, so that more changes to completely support weights can be made in an incremental but safe way. Until the function is completed, all changes depend on it is no-op, so DAMON works in the current mode.
--- mm/damon/core.c | 48 +++++++++++++++++++++++++++++-------- mm/damon/tests/core-kunit.h | 13 ++++++++-- 2 files changed, 49 insertions(+), 12 deletions(-)
mm/damon/core.c +38 -10
@@ -209,6 +209,11 @@ static struct damon_probe *damon_nth_probe(int n, struct damon_ctx *ctx)return NULL ; } +static bool damon_has_probe_weights (struct damon_ctx *c) +{ +return false; +} +/* * damon_mvsum() - Returns pseudo moving sum value for a time window. * @current_nr: The value of the current aggregation window.
@@ -3231,6 +3236,16 @@ static void damon_merge_two_regions(struct damon_target *t,damon_destroy_region (r, t); } +static unsigned int damon_merge_score (struct damon_region *r,bool last, +struct damon_ctx *ctx,bool use_probe_hits) +{ +if (use_probe_hits) +return damon_probe_hits_wsum (r, last, ctx); +if (last) +return r->last_nr_accesses; +return r->nr_accesses; +} +/* * Merge adjacent regions having similar access frequencies *
@@ -3239,24 +3254,37 @@ static void damon_merge_two_regions(struct damon_target *t,* sz_limit size upper limit of each region */ static void damon_merge_regions_of (struct damon_target *t,unsigned int thres, -unsigned long sz_limit) +unsigned long sz_limit,struct damon_ctx *ctx) {struct damon_region *r, *prev =NULL , *next; +bool use_probe_hits =damon_has_probe_weights (ctx);damon_for_each_region_safe (r, next, t) { -if (abs (r->nr_accesses - r->last_nr_accesses) > thres) +unsigned int score, last_score; + + score =damon_merge_score (r, false, ctx, use_probe_hits); + last_score =damon_merge_score (r, true, ctx, use_probe_hits); + +if (abs (score - last_score) > thres) r->age =0 ; -else if ((r->nr_accesses ==0 ) != (r->last_nr_accesses ==0 )) +else if ((score ==0 ) != (last_score ==0 )) r->age =0 ;else r->age++; -if (prev && prev->ar.end == r->ar.start && -abs (prev->nr_accesses - r->nr_accesses) <= thres && -damon_sz_region (prev) +damon_sz_region (r) <= sz_limit) -damon_merge_two_regions (t, prev, r); -else - prev = r; +if (!prev) + goto set_prev_continue; +if (prev->ar.end != r->ar.start) + goto set_prev_continue; +if (abs (damon_merge_score (prev, false, ctx, use_probe_hits) - + score) > thres) + goto set_prev_continue; +if (damon_sz_region (prev) +damon_sz_region (r) > sz_limit) + goto set_prev_continue; +damon_merge_two_regions (t, prev, r); +continue ; +set_prev_continue: + prev = r; } }
@@ -3289,7 +3317,7 @@ static void kdamond_merge_regions(struct damon_ctx *c, unsigned int threshold, do { nr_regions = 0; damon_for_each_target(t, c) { - damon_merge_regions_of(t, threshold, sz_limit); + damon_merge_regions_of(t, threshold, sz_limit, c); nr_regions += damon_nr_regions(t); } threshold = max(1, threshold * 2);
mm/damon/tests/core-kunit.h +11 -2
@@ -226,6 +226,7 @@ static struct damon_region *__nth_region_of(struct damon_target *t, int idx)static void damon_test_merge_regions_of (struct kunit *test) { +struct damon_ctx *ctx;struct damon_target *t;struct damon_region *r;unsigned long sa[] = {0 ,100 ,114 ,122 ,130 ,156 ,170 ,184 ,230 };
@@ -236,20 +237,27 @@ static void damon_test_merge_regions_of(struct kunit *test)unsigned long eaddrs[] = {112 ,130 ,156 ,170 ,230 ,10170 };int i; + ctx =damon_new_ctx (); +if (!ctx) +kunit_skip (test,"ctx alloc fail" ); + t =damon_new_target (); -if (!t) +if (!t) { +damon_destroy_ctx (ctx);kunit_skip (test,"target alloc fail" ); + }for (i =0 ; i <ARRAY_SIZE (sa); i++) { r =damon_new_region (sa[i], ea[i]);if (!r) {damon_free_target (t); +damon_destroy_ctx (ctx);kunit_skip (test,"region alloc fail" ); } r->nr_accesses = nrs[i];damon_add_region (r, t); } -damon_merge_regions_of (t,9 ,9999 ); +damon_merge_regions_of (t,9 ,9999 , ctx);/* 0-112, 114-130, 130-156, 156-170, 170-230, 230-10170 */ KUNIT_EXPECT_EQ (test,damon_nr_regions (t),6u );for (i =0 ; i <6 ; i++) {
@@ -258,6 +266,7 @@ static void damon_test_merge_regions_of(struct kunit *test)KUNIT_EXPECT_EQ (test, r->ar.end, eaddrs[i]); }damon_free_target (t); +damon_destroy_ctx (ctx); }static void damon_test_split_regions_of (struct kunit *test)
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 07/16] mm/damon/core: extend merge function to work with probe hits
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] API Misuse / Silent Overflow: Using `abs()` on the difference of two large unsigned integers…
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] API Misuse / Silent Overflow: Using abs() on the difference of two large unsigned integers causes incorrect results.
- [High] Integer Overflow: damon_probe_hits_wsum() wraps around the 32-bit boundary when large weights are used. --
commit b2b82783307a6d10c78010502dd30f5168036627
Author: SJ Park <[email protected]>
mm/damon/core: extend merge function to work with probe hits
When probe weights are set, this commit extends damon_merge_regions_of() to work with the weighted sum of probe hits. It also introduces a placeholder detection function to allow safe incremental changes.
Quoted 4 lines
mm/damon/core.c +0
-0
[ ... ]
mm/damon/core.c static void damon_merge_two_regions(struct damon_target *t, · line 3242
Quoted 9 lines
mm/damon/core.c +5
-0@@ -3230,6 +3235,16 @@ static void damon_merge_two_regions(struct damon_target *t,damon_destroy_region (r, t); } +static unsigned int damon_merge_score (struct damon_region *r,bool last, +struct damon_ctx *ctx,bool use_probe_hits) +{ +if (use_probe_hits) +return damon_probe_hits_wsum (r, last, ctx);
^^^^^^^^^^^^^^^^^^^^^^^^^
[Severity: High]
Does this code wrap around the 32-bit boundary when large weights are used?
Looking at damon_probe_hits_wsum(), it accumulates probe_hits[i] * probe->weight into a 32-bit unsigned int sum:
mm/damon/core.c:damon_probe_hits_wsum() {
...
unsigned int sum = 0;
...
if (last)
sum += r->last_probe_hits[i++] * probe->weight;
else
sum += r->probe_hits[i++] * probe->weight;
...
return sum;
}If a user configures a large weight (e.g., > 16,843,000) for a probe via sysfs, this could wrap around (e.g., 255 hits * 40,000,000 weight = 10.2 billion, which wraps to ~1.6 billion).
Could this break the scoring mechanism, causing damon_merge_regions_of() to merge active and inactive regions incorrectly?
Quoted 5 lines
mm/damon/core.c +5
-0+if (last) +return r->last_nr_accesses; +return r->nr_accesses; +} +
[ ... ]
Quoted 5 lines
mm/damon/core.c +1
-1@@ -3238,24 +3253,37 @@ static void damon_merge_regions_of(struct damon_target *t, unsigned int thres, - unsigned long sz_limit) + unsigned long sz_limit, struct damon_ctx *ctx) { struct damon_region *r, *prev = NULL, *next;
mm/damon/core.c static void damon_merge_regions_of(struct damon_target *t, unsigned int thres, · line 3264
Quoted 10 lines
mm/damon/core.c +7
-1+bool use_probe_hits =damon_has_probe_weights (ctx);damon_for_each_region_safe (r, next, t) { -if (abs (r->nr_accesses - r->last_nr_accesses) > thres) +unsigned int score, last_score; + + score =damon_merge_score (r, false, ctx, use_probe_hits); + last_score =damon_merge_score (r, true, ctx, use_probe_hits); + +if (abs (score - last_score) > thres)
^^^^^^^^^^^^^^^^^^^^^^^^
[Severity: High]
Can this call to abs() cause incorrect results due to silent overflow?
Since score and last_score are unsigned integers, abs(score - last_score) casts the unsigned difference to a signed int.
If the actual difference exceeds the maximum signed integer value, it wraps to a negative value, which abs() then negates into a smaller positive value.
Could this falsely small difference cause the threshold check > thres to fail incorrectly, allowing vastly different regions to be merged and affecting the accuracy of the monitor?
Quoted 9 lines
mm/damon/core.c +1
-3r->age =0 ; - else if ((r->nr_accesses == 0) != (r->last_nr_accesses == 0)) + else if ((score == 0) != (last_score == 0)) r->age = 0; else r->age++; - if (prev && prev->ar.end == r->ar.start && - abs(prev->nr_accesses - r->nr_accesses) <= thres &&
mm/damon/core.c static void damon_merge_two_regions(struct damon_target *t, · line 3280
Quoted 10 lines
mm/damon/core.c +6
-4- damon_sz_region(prev) + damon_sz_region(r) <= sz_limit) -damon_merge_two_regions (t, prev, r); - else - prev = r; +if (!prev) + goto set_prev_continue; +if (prev->ar.end != r->ar.start) + goto set_prev_continue; +if (abs (damon_merge_score (prev, false, ctx, use_probe_hits) - + score) > thres)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[Severity: High]
Does this second use of abs() on the unsigned difference of the merge scores suffer from the exact same signed-integer overflow vulnerability?
Quoted 9 lines
mm/damon/core.c +7
-0+ goto set_prev_continue; +if (damon_sz_region (prev) +damon_sz_region (r) > sz_limit) + goto set_prev_continue; +damon_merge_two_regions (t, prev, r); +continue ; +set_prev_continue: + prev = r; } }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=7
SJ Park
maintainerRe: [RFC PATCH v1.1 07/16] mm/damon/core: extend merge function to work with probe hits
Quoted 50 lines · On Mon, 06 Jul 2026 14:55:38 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] API Misuse / Silent Overflow: Usingabs()on the difference of two large unsigned integers causes incorrect results.
- [High] Integer Overflow:damon_probe_hits_wsum()wraps around the 32-bit boundary when large weights are used. --commit
b2b82783307a6d10c78010502dd30f5168036627
Author: SJ Park <[email protected]>
mm/damon/core: extend merge function to work with probe hitsWhen probe weights are set, this commit extends
damon_merge_regions_of()to work with the weighted sum of probe hits. It also introduces a placeholder detection function to allow safe incremental changes.Quoted 4 lines
mm/damon/core.c +0
-0[ ... ]
mm/damon/core.c static void damon_merge_two_regions(struct damon_target *t, · line 3242
Quoted 9 lines
mm/damon/core.c +5
-0@@ -3230,6 +3235,16 @@ static void damon_merge_two_regions(struct damon_target *t,damon_destroy_region (r, t); } +static unsigned int damon_merge_score (struct damon_region *r,bool last, +struct damon_ctx *ctx,bool use_probe_hits) +{ +if (use_probe_hits) +return damon_probe_hits_wsum (r, last, ctx);^^^^^^^^^^^^^^^^^^^^^^^^^[Severity: High]
Does this code wrap around the 32-bit boundary when large weights are used?Looking at
damon_probe_hits_wsum(), it accumulates probe_hits[i] *probe->weightinto a 32-bit unsigned int sum:mm/damon/core.c:damon_probe_hits_wsum() { ... unsigned int sum = 0; ... if (last) sum += r->last_probe_hits[i++] * probe->weight; else sum += r->probe_hits[i++] * probe->weight; ... return sum; }
Quoted 6 lines
If a user configures a large weight (e.g., > 16,843,000) for a probe via sysfs, this could wrap around (e.g., 255 hits * 40,000,000 weight = 10.2 billion, which wraps to ~1.6 billion).
Could this break the scoring mechanism, causing
damon_merge_regions_of()to merge active and inactive regions incorrectly?
Known low priority issue. No blocker.
Quoted 32 lines
Quoted 6 lines
mm/damon/core.c +5
-0+if (last) +return r->last_nr_accesses; +return r->nr_accesses; +} +[ ... ]
Quoted 5 lines
mm/damon/core.c +1
-1@@ -3238,24 +3253,37 @@ static void damon_merge_regions_of(struct damon_target *t, unsigned int thres, - unsigned long sz_limit) + unsigned long sz_limit, struct damon_ctx *ctx) { struct damon_region *r, *prev = NULL, *next;mm/damon/core.c static void damon_merge_regions_of(struct damon_target *t, unsigned int thres, · line 3264
Quoted 10 lines
mm/damon/core.c +7
-1+bool use_probe_hits =damon_has_probe_weights (ctx);damon_for_each_region_safe (r, next, t) { -if (abs (r->nr_accesses - r->last_nr_accesses) > thres) +unsigned int score, last_score; + + score =damon_merge_score (r, false, ctx, use_probe_hits); + last_score =damon_merge_score (r, true, ctx, use_probe_hits); + +if (abs (score - last_score) > thres)^^^^^^^^^^^^^^^^^^^^^^^^[Severity: High]
Can this call toabs()cause incorrect results due to silent overflow?Since score and
last_scoreare unsigned integers,abs(score - last_score)casts the unsigned difference to a signed int.
Quoted 6 lines
If the actual difference exceeds the maximum signed integer value, it wraps to a negative value, which
abs()then negates into a smaller positive value.Could this falsely small difference cause the threshold check > thres to fail incorrectly, allowing vastly different regions to be merged and affecting the accuracy of the monitor?
Same issue exists for the above abs() call. User impact is minor and I'm separately working [1] on it.
Quoted 22 lines
Quoted 10 lines
mm/damon/core.c +1
-3r->age =0 ; - else if ((r->nr_accesses == 0) != (r->last_nr_accesses == 0)) + else if ((score == 0) != (last_score == 0)) r->age = 0; else r->age++; - if (prev && prev->ar.end == r->ar.start && - abs(prev->nr_accesses - r->nr_accesses) <= thres &&mm/damon/core.c static void damon_merge_two_regions(struct damon_target *t, · line 3280
Quoted 10 lines
mm/damon/core.c +6
-4- damon_sz_region(prev) + damon_sz_region(r) <= sz_limit) -damon_merge_two_regions (t, prev, r); - else - prev = r; +if (!prev) + goto set_prev_continue; +if (prev->ar.end != r->ar.start) + goto set_prev_continue; +if (abs (damon_merge_score (prev, false, ctx, use_probe_hits) - + score) > thres)^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Quoted 3 lines
[Severity: High]
Does this second use ofabs()on the unsigned difference of the merge scores suffer from the exact same signed-integer overflow vulnerability?
Ditto.
Quoted 13 lines
Quoted 10 lines
mm/damon/core.c +7
-0+ goto set_prev_continue; +if (damon_sz_region (prev) +damon_sz_region (r) > sz_limit) + goto set_prev_continue; +damon_merge_two_regions (t, prev, r); +continue ; +set_prev_continue: + prev = r; } }Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=7
[1] https://lore.kernel.org/[email protected]
Thanks,
SJ
[...]
SJ Park
maintainerRe: [RFC PATCH v1.1 07/16] mm/damon/core: extend merge function to work with probe hits
Quoted 96 lines · On Mon, 6 Jul 2026 16:52:00 -0700 SJ Park <[email protected]> wrote:
Quoted 50 lines · On Mon, 06 Jul 2026 14:55:38 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] API Misuse / Silent Overflow: Usingabs()on the difference of two large unsigned integers causes incorrect results.
- [High] Integer Overflow:damon_probe_hits_wsum()wraps around the 32-bit boundary when large weights are used. --commit
b2b82783307a6d10c78010502dd30f5168036627
Author: SJ Park <[email protected]>
mm/damon/core: extend merge function to work with probe hitsWhen probe weights are set, this commit extends
damon_merge_regions_of()to work with the weighted sum of probe hits. It also introduces a placeholder detection function to allow safe incremental changes.Quoted 4 lines
mm/damon/core.c +0
-0[ ... ]
mm/damon/core.c static void damon_merge_two_regions(struct damon_target *t, · line 3242
Quoted 9 lines
mm/damon/core.c +5
-0@@ -3230,6 +3235,16 @@ static void damon_merge_two_regions(struct damon_target *t,damon_destroy_region (r, t); } +static unsigned int damon_merge_score (struct damon_region *r,bool last, +struct damon_ctx *ctx,bool use_probe_hits) +{ +if (use_probe_hits) +return damon_probe_hits_wsum (r, last, ctx);^^^^^^^^^^^^^^^^^^^^^^^^^[Severity: High]
Does this code wrap around the 32-bit boundary when large weights are used?Looking at
damon_probe_hits_wsum(), it accumulates probe_hits[i] *probe->weightinto a 32-bit unsigned int sum:mm/damon/core.c:damon_probe_hits_wsum() { ... unsigned int sum = 0; ... if (last) sum += r->last_probe_hits[i++] * probe->weight; else sum += r->probe_hits[i++] * probe->weight; ... return sum; }Quoted 6 lines
If a user configures a large weight (e.g., > 16,843,000) for a probe via sysfs, this could wrap around (e.g., 255 hits * 40,000,000 weight = 10.2 billion, which wraps to ~1.6 billion).
Could this break the scoring mechanism, causing
damon_merge_regions_of()to merge active and inactive regions incorrectly?Known low priority issue. No blocker.
Quoted 35 lines
Quoted 6 lines
mm/damon/core.c +5
-0+if (last) +return r->last_nr_accesses; +return r->nr_accesses; +} +[ ... ]
Quoted 5 lines
mm/damon/core.c +1
-1@@ -3238,24 +3253,37 @@ static void damon_merge_regions_of(struct damon_target *t, unsigned int thres, - unsigned long sz_limit) + unsigned long sz_limit, struct damon_ctx *ctx) { struct damon_region *r, *prev = NULL, *next;mm/damon/core.c static void damon_merge_regions_of(struct damon_target *t, unsigned int thres, · line 3264
Quoted 10 lines
mm/damon/core.c +7
-1+bool use_probe_hits =damon_has_probe_weights (ctx);damon_for_each_region_safe (r, next, t) { -if (abs (r->nr_accesses - r->last_nr_accesses) > thres) +unsigned int score, last_score; + + score =damon_merge_score (r, false, ctx, use_probe_hits); + last_score =damon_merge_score (r, true, ctx, use_probe_hits); + +if (abs (score - last_score) > thres)^^^^^^^^^^^^^^^^^^^^^^^^[Severity: High]
Can this call toabs()cause incorrect results due to silent overflow?Since score and
last_scoreare unsigned integers,abs(score - last_score)casts the unsigned difference to a signed int.If the actual difference exceeds the maximum signed integer value, it wraps to a negative value, which
abs()then negates into a smaller positive value.
Quoted 6 lines
Quoted 3 lines
Could this falsely small difference cause the threshold check > thres to fail incorrectly, allowing vastly different regions to be merged and affecting the accuracy of the monitor?
Same issue exists for the above
abs()call. User impact is minor and I'm separately working [1] on it.
Now I think this has no reason to fix right now. I'll fix it in the next revision.
Thanks,
SJ
[...]
SJ Park
maintainerRFC PATCH 8/16 mm/damon/core: disable access monitoring when probe weights are set
In reply to SJ ParkWhen users set probe weights, they are interested in the data attributes. Not necessarily they are interested in data access patterns. Disable the data access monitoring in the case.
--- mm/damon/core.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)
mm/damon/core.c +3 -2
@@ -3714,17 +3714,18 @@ static int kdamond_fn(void *data)unsigned long next_aggregation_sis = ctx->next_aggregation_sis;unsigned long next_ops_update_sis = ctx->next_ops_update_sis;unsigned long sample_interval = ctx->attrs.sample_interval; +bool access_check_disabled =damon_has_probe_weights (ctx);if (kdamond_wait_activation (ctx))break ; -if (ctx->ops.prepare_access_checks) +if (!access_check_disabled && ctx->ops.prepare_access_checks) ctx->ops.prepare_access_checks (ctx);kdamond_usleep (sample_interval); ctx->passed_sample_intervals++; -if (ctx->ops.check_accesses) +if (!access_check_disabled && ctx->ops.check_accesses) max_nr_accesses = ctx->ops.check_accesses (ctx);if (ctx->ops.apply_probes) ctx->ops.apply_probes (ctx, false, false);
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 08/16] mm/damon/core: disable access monitoring when probe weights are set
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] The region merging loop termination condition is incompatible with weighted probe hits, leading to…
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The region merging loop termination condition is incompatible with weighted probe hits, leading to unbounded region growth.
- [High] Unconditionally disabling access checks forces nr_accesses to 0, breaking access-based DAMOS schemes and interval auto-tuning. --
commit d9de5e6aadab2e17a0c4428844dec0cfc05d08ef
Author: SJ Park <[email protected]>
mm/damon/core: disable access monitoring when probe weights are set
When users set probe weights, they are interested in the data attributes rather than data access patterns. This commit disables data access monitoring when probe weights are set.
mm/damon/core.c static int kdamond_fn(void *data) · line 3716
Quoted 9 lines
mm/damon/core.c +1
-0@@ -3713,17 +3713,18 @@ static int kdamond_fn(void *data)unsigned long next_aggregation_sis = ctx->next_aggregation_sis;unsigned long next_ops_update_sis = ctx->next_ops_update_sis;unsigned long sample_interval = ctx->attrs.sample_interval; +bool access_check_disabled =damon_has_probe_weights (ctx);
[Severity: High]
Can configuring probe weights lead to unbounded region growth during merging?
When damon_has_probe_weights(ctx) is true, the region merging loop in kdamond_merge_regions() compares a threshold scaled with weighted probe hits against max_thres:
mm/damon/core.c:kdamond_merge_regions() {
...
max_thres = c->attrs.aggr_interval /
(c->attrs.sample_interval ? c->attrs.sample_interval : 1);
do {
...
} while (nr_regions > c->attrs.max_nr_regions &&
threshold / 2 < max_thres);
}Since max_thres is hardcoded to the maximum unscaled accesses (aggr_interval / sample_interval) and doesn't account for probe weights, the loop could terminate prematurely.
If weighted probe hits result in score differences exceeding this unscaled cap, wouldn't this prevent regions from merging back down below the user-configured max_nr_regions limit?
Quoted 4 lines
mm/damon/core.c +0
-0if (kdamond_wait_activation (ctx))break ;
mm/damon/core.c static int kdamond_fn(void *data) · line 3729
Quoted 10 lines
mm/damon/core.c +2
-2-if (ctx->ops.prepare_access_checks) +if (!access_check_disabled && ctx->ops.prepare_access_checks) ctx->ops.prepare_access_checks (ctx);kdamond_usleep (sample_interval); ctx->passed_sample_intervals++; -if (ctx->ops.check_accesses) +if (!access_check_disabled && ctx->ops.check_accesses) max_nr_accesses = ctx->ops.check_accesses (ctx);
[Severity: High]
Does disabling the access checks here break DAMOS schemes and interval auto-tuning?
By preventing ctx->ops.check_accesses from running when probe weights are configured, r->nr_accesses will remain 0.
Since damon_nr_accesses_mvsum() consumes these 0s, it seems this would cause interval auto-tuning to exponentially increase the sampling interval until it hits the maximum allowed value.
Additionally, wouldn't this force all regions to instantly match any DAMOS scheme configured to target low-access memory?
Quoted 2 lines
mm/damon/core.c +0
-0if (ctx->ops.apply_probes) ctx->ops.apply_probes (ctx, false, false);
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=8
SJ Park
maintainerRe: [RFC PATCH v1.1 08/16] mm/damon/core: disable access monitoring when probe weights are set
Quoted 41 lines · On Mon, 06 Jul 2026 15:03:08 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The region merging loop termination condition is incompatible with weighted probe hits, leading to unbounded region growth.
- [High] Unconditionally disabling access checks forcesnr_accessesto 0, breaking access-based DAMOS schemes and interval auto-tuning. --commit
d9de5e6aadab2e17a0c4428844dec0cfc05d08ef
Author: SJ Park <[email protected]>
mm/damon/core: disable access monitoring when probe weights are setWhen users set probe weights, they are interested in the data attributes rather than data access patterns. This commit disables data access monitoring when probe weights are set.
mm/damon/core.c static int kdamond_fn(void *data) · line 3716
Quoted 9 lines
mm/damon/core.c +1
-0@@ -3713,17 +3713,18 @@ static int kdamond_fn(void *data)unsigned long next_aggregation_sis = ctx->next_aggregation_sis;unsigned long next_ops_update_sis = ctx->next_ops_update_sis;unsigned long sample_interval = ctx->attrs.sample_interval; +bool access_check_disabled =damon_has_probe_weights (ctx);[Severity: High]
Can configuring probe weights lead to unbounded region growth during merging?When
damon_has_probe_weights(ctx)is true, the region merging loop inkdamond_merge_regions()compares a threshold scaled with weighted probe hits againstmax_thres:mm/damon/core.c:kdamond_merge_regions() { ... max_thres = c->attrs.aggr_interval / (c->attrs.sample_interval ? c->attrs.sample_interval : 1); do { ... } while (nr_regions > c->attrs.max_nr_regions && threshold / 2 < max_thres); }
Quoted 7 lines
Since
max_thresis hardcoded to the maximum unscaled accesses (aggr_interval/sample_interval) and doesn't account for probe weights, the loop could terminate prematurely.If weighted probe hits result in score differences exceeding this unscaled cap, wouldn't this prevent regions from merging back down below the user-configured
max_nr_regionslimit?
No such issue can happen because damon_has_probe_weights() always returns false.
Quoted 23 lines
Quoted 5 lines
mm/damon/core.c +0
-0if (kdamond_wait_activation (ctx))break ;mm/damon/core.c static int kdamond_fn(void *data) · line 3729
Quoted 10 lines
mm/damon/core.c +2
-2-if (ctx->ops.prepare_access_checks) +if (!access_check_disabled && ctx->ops.prepare_access_checks) ctx->ops.prepare_access_checks (ctx);kdamond_usleep (sample_interval); ctx->passed_sample_intervals++; -if (ctx->ops.check_accesses) +if (!access_check_disabled && ctx->ops.check_accesses) max_nr_accesses = ctx->ops.check_accesses (ctx);[Severity: High]
Does disabling the access checks here break DAMOS schemes and interval auto-tuning?By preventing
ctx->ops.check_accessesfrom running when probe weights are configured,r->nr_accesseswill remain 0.
Quoted 6 lines
Since
damon_nr_accesses_mvsum()consumes these 0s, it seems this would cause interval auto-tuning to exponentially increase the sampling interval until it hits the maximum allowed value.Additionally, wouldn't this force all regions to instantly match any DAMOS scheme configured to target low-access memory?
That's the intended behavior. Users are informed that access monitoring is disabled in the mode.
Quoted 6 lines
Quoted 3 lines
mm/damon/core.c +0
-0if (ctx->ops.apply_probes) ctx->ops.apply_probes (ctx, false, false);Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=8
Thanks,
SJ
[...]
SJ Park
maintainerRFC PATCH 9/16 mm/damon/core: set samples in apply_probes() if probe weights are set
In reply to SJ ParkWhen probe weights are set, the prepare_access_checks() ops callback is not invoked, so sampling addresses of regions are unset. Ask apply_probes() to set those together in the case.
--- mm/damon/core.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
mm/damon/core.c +2 -1
@@ -3728,7 +3728,8 @@ static int kdamond_fn(void *data)if (!access_check_disabled && ctx->ops.check_accesses) max_nr_accesses = ctx->ops.check_accesses (ctx);if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx, false, false); + ctx->ops.apply_probes (ctx, access_check_disabled, + false);if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 09/16] mm/damon/core: set samples in apply_probes() if probe weights are set
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [High] Region merging loop bound `max_thres` is incorrectly scaled when probe weights are used, causing…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Region merging loop bound max_thres is incorrectly scaled when probe weights are used, causing premature loop termination and permanent violation of max_nr_regions limits. --
commit cc6e72a3b37a62b24190b0e42f141fba6caab8fc
Author: SJ Park <[email protected]>
mm/damon/core: set samples in apply_probes() if probe weights are set
When probe weights are set, the prepare_access_checks() ops callback is not invoked, so sampling addresses of regions are unset. Ask apply_probes() to set those together in the case.
mm/damon/core.c static int kdamond_fn(void *data) · line 3731
Quoted 11 lines
mm/damon/core.c +2
-1@@ -3727,7 +3727,8 @@ static int kdamond_fn(void *data)if (!access_check_disabled && ctx->ops.check_accesses) max_nr_accesses = ctx->ops.check_accesses (ctx);if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx, false, false); + ctx->ops.apply_probes (ctx, access_check_disabled, + false);
[Severity: High]
Does enabling probe weights here introduce a regression with the region merging loop bound in kdamond_merge_regions()?
Later in the series, a commit ("mm/damon/core: get merge threshold from probe hits when weights are set") derives the merge threshold from weighted sums when probe weights are used. However, the loop bound max_thres in kdamond_merge_regions() remains tied to the raw access frequency:
mm/damon/core.c:kdamond_merge_regions() {
...
max_thres = c->attrs.aggr_interval /
(c->attrs.sample_interval ? c->attrs.sample_interval : 1);
do {
nr_regions = 0;
damon_for_each_target(t, c) {
damon_merge_regions_of(t, threshold, sz_limit, c);
nr_regions += damon_nr_regions(t);
}
threshold = max(1, threshold * 2);
} while (nr_regions > c->attrs.max_nr_regions &&
threshold / 2 < max_thres);
}If a large weighted sum is used as the initial threshold, it seems the loop will exit immediately on the first iteration without merging enough regions because threshold / 2 will already exceed the raw max_thres.
Can this permanently violate the max_nr_regions limits?
Quoted 3 lines
mm/damon/core.c +0
-0if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=9
SJ Park
maintainerRe: [RFC PATCH v1.1 09/16] mm/damon/core: set samples in apply_probes() if probe weights are set
Quoted 49 lines · On Mon, 06 Jul 2026 15:02:05 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Region merging loop boundmax_thresis incorrectly scaled when probe weights are used, causing premature loop termination and permanent violation ofmax_nr_regionslimits. --commit
cc6e72a3b37a62b24190b0e42f141fba6caab8fc
Author: SJ Park <[email protected]>
mm/damon/core: set samples inapply_probes()if probe weights are setWhen probe weights are set, the
prepare_access_checks()ops callback is not invoked, so sampling addresses of regions are unset. Askapply_probes()to set those together in the case.mm/damon/core.c static int kdamond_fn(void *data) · line 3731
Quoted 11 lines
mm/damon/core.c +2
-1@@ -3727,7 +3727,8 @@ static int kdamond_fn(void *data)if (!access_check_disabled && ctx->ops.check_accesses) max_nr_accesses = ctx->ops.check_accesses (ctx);if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx, false, false); + ctx->ops.apply_probes (ctx, access_check_disabled, + false);[Severity: High]
Does enabling probe weights here introduce a regression with the region merging loop bound inkdamond_merge_regions()?Later in the series, a commit ("
mm/damon/core: get merge threshold from probe hits when weights are set") derives the merge threshold from weighted sums when probe weights are used. However, the loop boundmax_thresinkdamond_merge_regions()remains tied to the raw access frequency:mm/damon/core.c:kdamond_merge_regions() { ... max_thres = c->attrs.aggr_interval / (c->attrs.sample_interval ? c->attrs.sample_interval : 1); do { nr_regions = 0; damon_for_each_target(t, c) { damon_merge_regions_of(t, threshold, sz_limit, c); nr_regions += damon_nr_regions(t); } threshold = max(1, threshold * 2); } while (nr_regions > c->attrs.max_nr_regions && threshold / 2 < max_thres); }
Quoted 5 lines
If a large weighted sum is used as the initial threshold, it seems the loop will exit immediately on the first iteration without merging enough regions because threshold / 2 will already exceed the raw
max_thres.Can this permanently violate the
max_nr_regionslimits?
No. damon_has_probe_weights() always returns false at this moment, so no issue.
Quoted 7 lines
Quoted 4 lines
mm/damon/core.c +0
-0if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=9
Thanks,
SJ
[...]
SJ Park
maintainerRFC PATCH 10/16 mm/damon/core: s/max_nr_accesses/max_merge_score/ in kdamond_fn()
In reply to SJ ParkWhen probe weights are set, DAMON merge logic works with the probe hits weighted sum. The core logic is setting the merge threshold based on the access frequency, though. Before using probe hits based threshold, rename the variables to look more general.
--- mm/damon/core.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
mm/damon/core.c +3 -3
@@ -3681,7 +3681,6 @@ static void kdamond_init_ctx(struct damon_ctx *ctx)static int kdamond_fn (void *data) {struct damon_ctx *ctx = data; -unsigned int max_nr_accesses =0 ;unsigned long sz_limit =0 ;pr_debug ("kdamond (%d) starts\n" , current->pid);
@@ -3715,6 +3714,7 @@ static int kdamond_fn(void *data)unsigned long next_ops_update_sis = ctx->next_ops_update_sis;unsigned long sample_interval = ctx->attrs.sample_interval;bool access_check_disabled =damon_has_probe_weights (ctx); +unsigned int max_merge_score =0 ;if (kdamond_wait_activation (ctx))break ;
@@ -3726,7 +3726,7 @@ static int kdamond_fn(void *data) ctx->passed_sample_intervals++;if (!access_check_disabled && ctx->ops.check_accesses) - max_nr_accesses = ctx->ops.check_accesses (ctx); + max_merge_score = ctx->ops.check_accesses (ctx);if (ctx->ops.apply_probes) ctx->ops.apply_probes (ctx, access_check_disabled, false);
@@ -3734,7 +3734,7 @@ static int kdamond_fn(void *data)if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {kdamond_merge_regions (ctx, - max_nr_accesses /10 , + max_merge_score /10 , sz_limit);/* online updates might be made */ sz_limit =damon_apply_min_nr_regions (ctx);
Signature · 1 line
2.47.3
SJ Park
maintainerRFC PATCH 11/16 mm/damon/core: get merge threshold from probe hits when weights are set
In reply to SJ ParkWhen probe weights are set, DAMON merges regions based on their probe hits weighted sum. But the merge threshold is calculated based on the access frequency. Update it to retrieve the maximum probe hits weighted sum in the snapshot from apply_probes() ops callback, and generate the threshold based on it.
--- mm/damon/core.c | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-)
mm/damon/core.c +14 -4
@@ -3714,7 +3714,8 @@ static int kdamond_fn(void *data)unsigned long next_ops_update_sis = ctx->next_ops_update_sis;unsigned long sample_interval = ctx->attrs.sample_interval;bool access_check_disabled =damon_has_probe_weights (ctx); -unsigned int max_merge_score =0 ; +unsigned int max_merge_score =0 , max_wsum; +bool get_max_wsum;if (kdamond_wait_activation (ctx))break ;
@@ -3727,9 +3728,18 @@ static int kdamond_fn(void *data)if (!access_check_disabled && ctx->ops.check_accesses) max_merge_score = ctx->ops.check_accesses (ctx); -if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx, access_check_disabled, - false); +if (ctx->ops.apply_probes) { +if (time_after_eq (ctx->passed_sample_intervals, + next_aggregation_sis) && + access_check_disabled) + get_max_wsum = true; +else + get_max_wsum = false; + max_wsum = ctx->ops.apply_probes (ctx, + access_check_disabled, get_max_wsum); +if (get_max_wsum) + max_merge_score = max_wsum; + }if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 11/16] mm/damon/core: get merge threshold from probe hits when weights are set
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] Region merge loop terminates prematurely because `max_thres` is unscaled for probe hit weighted…
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Region merge loop terminates prematurely because max_thres is unscaled for probe hit weighted sums, causing a failure to respect max_nr_regions.
- [High] Adding probe weights to contexts using operations without apply_probes (e.g., vaddr) silently breaks all monitoring. --
commit f896f3e81bf75b446b16cb4f79f6673a154b512b
Author: SJ Park <[email protected]>
mm/damon/core: get merge threshold from probe hits when weights are set
When probe weights are set, DAMON merges regions based on their probe hits weighted sum. But the merge threshold is calculated based on the access frequency. Update it to retrieve the maximum probe hits weighted sum in the snapshot from apply_probes() ops callback, and generate the threshold based on it.
Quoted 4 lines
mm/damon/core.c +0
-0
[ ... ]
mm/damon/core.c static int kdamond_fn(void *data) · line 3729
Quoted 4 lines
mm/damon/core.c +0
-0@@ -3726,9 +3727,18 @@ static int kdamond_fn(void *data)if (!access_check_disabled && ctx->ops.check_accesses) max_merge_score = ctx->ops.check_accesses (ctx);
[Severity: High]
Does setting probe weights on a context that lacks an apply_probes callback (like vaddr) silently break monitoring here?
If a user sets probe weights via sysfs, access_check_disabled will be true. This skips check_accesses(), but since apply_probes is NULL, neither callback runs. The accesses are never updated, and all memory appears completely cold.
Quoted 4 lines
mm/damon/core.c +1
-3-if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx, access_check_disabled, - false); +if (ctx->ops.apply_probes) {
mm/damon/core.c static int kdamond_fn(void *data) · line 3741
Quoted 10 lines
mm/damon/core.c +10
-0+if (time_after_eq (ctx->passed_sample_intervals, + next_aggregation_sis) && + access_check_disabled) + get_max_wsum = true; +else + get_max_wsum = false; + max_wsum = ctx->ops.apply_probes (ctx, + access_check_disabled, get_max_wsum); +if (get_max_wsum) + max_merge_score = max_wsum;
[Severity: High]
Can this cause kdamond_merge_regions() to fail to reduce the number of regions below max_nr_regions?
When kdamond_fn() later passes this max_wsum / 10 as the initial threshold to kdamond_merge_regions(), the merge loop termination condition compares it against an unscaled max_thres:
mm/damon/core.c:kdamond_merge_regions() {
...
max_thres = c->attrs.aggr_interval /
(c->attrs.sample_interval ? c->attrs.sample_interval : 1);
...
} while (nr_regions > c->attrs.max_nr_regions &&
threshold / 2 < max_thres);
}Since a user-defined weighted sum can be vastly larger than the unweighted max_thres, will the loop condition fail prematurely, leading to an unbounded growth of memory regions?
Quoted 4 lines
mm/damon/core.c +1
-0+ }if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=11
SJ Park
maintainerRe: [RFC PATCH v1.1 11/16] mm/damon/core: get merge threshold from probe hits when weights are set
Quoted 30 lines · On Mon, 06 Jul 2026 15:18:01 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Region merge loop terminates prematurely becausemax_thresis unscaled for probe hit weighted sums, causing a failure to respectmax_nr_regions.
- [High] Adding probe weights to contexts using operations withoutapply_probes(e.g.,vaddr) silently breaks all monitoring. --commit
f896f3e81bf75b446b16cb4f79f6673a154b512b
Author: SJ Park <[email protected]>
mm/damon/core: get merge threshold from probe hits when weights are setWhen probe weights are set, DAMON merges regions based on their probe hits weighted sum. But the merge threshold is calculated based on the access frequency. Update it to retrieve the maximum probe hits weighted sum in the snapshot from
apply_probes()ops callback, and generate the threshold based on it.Quoted 4 lines
mm/damon/core.c +0
-0[ ... ]
mm/damon/core.c static int kdamond_fn(void *data) · line 3729
Quoted 4 lines
mm/damon/core.c +0
-0@@ -3726,9 +3727,18 @@ static int kdamond_fn(void *data)if (!access_check_disabled && ctx->ops.check_accesses) max_merge_score = ctx->ops.check_accesses (ctx);[Severity: High]
Does setting probe weights on a context that lacks anapply_probescallback (like vaddr) silently break monitoring here?
Quoted 3 lines
If a user sets probe weights via sysfs,
access_check_disabledwill be true. This skipscheck_accesses(), but sinceapply_probesis NULL, neither callback runs. The accesses are never updated, and all memory appears completely cold.
I was initially thinking this is no blocker. But, since the core layer params validation patch series is merged into mm-new, fixing this should be simple, and we don't really need to rush for this series. I will fix this in the next spin.
Quoted 33 lines
Quoted 5 lines
mm/damon/core.c +1
-3-if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx, access_check_disabled, - false); +if (ctx->ops.apply_probes) {mm/damon/core.c static int kdamond_fn(void *data) · line 3741
Quoted 10 lines
mm/damon/core.c +10
-0+if (time_after_eq (ctx->passed_sample_intervals, + next_aggregation_sis) && + access_check_disabled) + get_max_wsum = true; +else + get_max_wsum = false; + max_wsum = ctx->ops.apply_probes (ctx, + access_check_disabled, get_max_wsum); +if (get_max_wsum) + max_merge_score = max_wsum;[Severity: High]
Can this causekdamond_merge_regions()to fail to reduce the number of regions belowmax_nr_regions?When
kdamond_fn()later passes thismax_wsum/ 10 as the initial threshold tokdamond_merge_regions(), the merge loop termination condition compares it against an unscaledmax_thres:mm/damon/core.c:kdamond_merge_regions() { ... max_thres = c->attrs.aggr_interval / (c->attrs.sample_interval ? c->attrs.sample_interval : 1); ... } while (nr_regions > c->attrs.max_nr_regions && threshold / 2 < max_thres); }
Quoted 3 lines
Since a user-defined weighted sum can be vastly larger than the unweighted
max_thres, will the loop condition fail prematurely, leading to an unbounded growth of memory regions?
Yes, but only for one aggregation interval. I don't think that's an issue.
Quoted 8 lines
Quoted 5 lines
mm/damon/core.c +1
-0+ }if (time_after_eq (ctx->passed_sample_intervals, next_aggregation_sis)) {Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=11
Thanks,
SJ
[...]
SJ Park
maintainerRe: [RFC PATCH v1.1 11/16] mm/damon/core: get merge threshold from probe hits when weights are set
Quoted 32 lines · On Mon, 6 Jul 2026 17:03:28 -0700 SJ Park <[email protected]> wrote:
Quoted 30 lines · On Mon, 06 Jul 2026 15:18:01 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Region merge loop terminates prematurely becausemax_thresis unscaled for probe hit weighted sums, causing a failure to respectmax_nr_regions.
- [High] Adding probe weights to contexts using operations withoutapply_probes(e.g.,vaddr) silently breaks all monitoring. --commit
f896f3e81bf75b446b16cb4f79f6673a154b512b
Author: SJ Park <[email protected]>
mm/damon/core: get merge threshold from probe hits when weights are setWhen probe weights are set, DAMON merges regions based on their probe hits weighted sum. But the merge threshold is calculated based on the access frequency. Update it to retrieve the maximum probe hits weighted sum in the snapshot from
apply_probes()ops callback, and generate the threshold based on it.Quoted 4 lines
mm/damon/core.c +0
-0[ ... ]
mm/damon/core.c static int kdamond_fn(void *data) · line 3729
Quoted 4 lines
mm/damon/core.c +0
-0@@ -3726,9 +3727,18 @@ static int kdamond_fn(void *data)if (!access_check_disabled && ctx->ops.check_accesses) max_merge_score = ctx->ops.check_accesses (ctx);[Severity: High]
Does setting probe weights on a context that lacks anapply_probescallback (like vaddr) silently break monitoring here?
Quoted 8 lines
Quoted 3 lines
If a user sets probe weights via sysfs,
access_check_disabledwill be true. This skipscheck_accesses(), but sinceapply_probesis NULL, neither callback runs. The accesses are never updated, and all memory appears completely cold.I was initially thinking this is no blocker. But, since the core layer params validation patch series is merged into mm-new, fixing this should be simple, and we don't really need to rush for this series. I will fix this in the next spin.
I changed my mind. There are many features that supported by only specific DAMON ops implementations. We allow setting such configuration, and later add supports if needed. It is arguable, but I think keeping the consistency is the right thing for now. I will keep this behavior unless someone argues differently.
Thanks,
SJ
[...]
SJ Park
maintainerRFC PATCH 12/16 mm/damon/core: implement damon_has_probe_weight()
In reply to SJ ParkDAMON monitors only data attributes when probe weights are set. However, the function for detecting it is intentionally returning false always. The intention was to make changes for complete support of the mode in a safe and incremental way. Now all the building blocks are made. Complete the mode detection function so that the mode can really start working.
--- mm/damon/core.c | 6 ++++++ 1 file changed, 6 insertions(+)
mm/damon/core.c +6 -0
@@ -211,6 +211,12 @@ static struct damon_probe *damon_nth_probe(int n, struct damon_ctx *ctx)static bool damon_has_probe_weights (struct damon_ctx *c) { +struct damon_probe *p; + +damon_for_each_probe (p, c) { +if (p->weight) +return true; + }return false; }
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 12/16] mm/damon/core: implement damon_has_probe_weight()
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
New issues:
- [High] Setting probe weights on a DAMON context whose operations do not support apply_probes (e.g., vaddr) silently disables all monitoring.
Pre-existing issues:
- [High] The probe_hits array in struct damon_region uses unsigned char counters which wrap around at 256, corrupting monitoring data when the number of samples per aggregation interval exceeds 255. --
commit 3891f362834bef45987ebcee1cdb0abcd9df6ea2
Author: SJ Park <[email protected]>
mm/damon/core: implement damon_has_probe_weight()
DAMON monitors only data attributes when probe weights are set. However, the function for detecting it is intentionally returning false always. The intention was to make changes for complete support of the mode in a safe and incremental way. Now all the building blocks are made. Complete the mode detection function so that the mode can really start working.
mm/damon/core.c static struct damon_probe *damon_nth_probe(int n, struct damon_ctx *ctx) · line 218
Quoted 13 lines
mm/damon/core.c +5
-0@@ -211,6 +211,12 @@ static struct damon_probe *damon_nth_probe(int n, struct damon_ctx *ctx)static bool damon_has_probe_weights (struct damon_ctx *c) { +struct damon_probe *p; + +damon_for_each_probe (p, c) { +if (p->weight) +return true;
[Severity: High]
Does returning true here silently disable monitoring if the context operations do not support probes (e.g., vaddr)?
In kdamond_fn(), if this returns true, access_check_disabled bypasses standard checks. However, if ctx->ops.apply_probes is NULL, the probes are skipped:
mm/damon/core.c:kdamond_fn() {
...
bool access_check_disabled = damon_has_probe_weights(ctx);
...
if (!access_check_disabled && ctx->ops.prepare_access_checks)
ctx->ops.prepare_access_checks(ctx);
...
if (ctx->ops.apply_probes) {
...
}Would it be better to ensure ctx->ops.apply_probes is also implemented before returning true?
[Severity: High]
This is a pre-existing issue, but since this patch enables probe weights, can the 8-bit probe_hits counters overflow if the number of samples per aggregation interval exceeds 255?
Looking at struct damon_region in include/linux/damon.h, probe_hits is declared as an unsigned char array:
include/linux/damon.h:struct damon_region {
...
unsigned char probe_hits[DAMON_MAX_PROBES];
...
}In damon_pa_apply_probes(), it is incremented during every sample interval without checking for overflow:
mm/damon/paddr.c:damon_pa_apply_probes() {
...
damon_for_each_probe(p, ctx) {
if (damon_pa_filter_pass(pa, folio, p))
r->probe_hits[i]++;
...
}If a user configures an aggregation interval that is > 255 times the sample interval, could highly-hit regions wrap around to 0 and be falsely identified as cold memory?
Quoted 3 lines
mm/damon/core.c +1
-0+ }return false; }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=12
SJ Park
maintainerRe: [RFC PATCH v1.1 12/16] mm/damon/core: implement damon_has_probe_weight()
Quoted 53 lines · On Mon, 06 Jul 2026 15:16:52 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
New issues:
- [High] Setting probe weights on a DAMON context whose operations do not supportapply_probes(e.g.,vaddr) silently disables all monitoring.Pre-existing issues:
- [High] Theprobe_hitsarray instruct damon_regionusesunsigned charcounters which wrap around at 256, corrupting monitoring data when the number of samples per aggregation interval exceeds 255. --commit
3891f362834bef45987ebcee1cdb0abcd9df6ea2
Author: SJ Park <[email protected]>
mm/damon/core: implementdamon_has_probe_weight()DAMON monitors only data attributes when probe weights are set. However, the function for detecting it is intentionally returning false always. The intention was to make changes for complete support of the mode in a safe and incremental way. Now all the building blocks are made. Complete the mode detection function so that the mode can really start working.
mm/damon/core.c static struct damon_probe *damon_nth_probe(int n, struct damon_ctx *ctx) · line 218
Quoted 13 lines
mm/damon/core.c +5
-0@@ -211,6 +211,12 @@ static struct damon_probe *damon_nth_probe(int n, struct damon_ctx *ctx)static bool damon_has_probe_weights (struct damon_ctx *c) { +struct damon_probe *p; + +damon_for_each_probe (p, c) { +if (p->weight) +return true;[Severity: High]
Does returning true here silently disable monitoring if the context operations do not support probes (e.g., vaddr)?In
kdamond_fn(), if this returns true,access_check_disabledbypasses standard checks. However, ifctx->ops.apply_probesis NULL, the probes are skipped:mm/damon/core.c:kdamond_fn() { ... bool access_check_disabled = damon_has_probe_weights(ctx); ... if (!access_check_disabled && ctx->ops.prepare_access_checks) ctx->ops.prepare_access_checks(ctx); ... if (ctx->ops.apply_probes) { ... }
Quoted 2 lines
Would it be better to ensure
ctx->ops.apply_probesis also implemented before returning true?
I will fix this in the next revision.
Quoted 26 lines
[Severity: High]
This is a pre-existing issue, but since this patch enables probe weights, can the 8-bitprobe_hitscounters overflow if the number of samples per aggregation interval exceeds 255?Looking at struct
damon_regionininclude/linux/damon.h,probe_hitsis declared as an unsigned char array:include/linux/damon.h:struct damon_region { ... unsigned char probe_hits[DAMON_MAX_PROBES]; ... }In
damon_pa_apply_probes(), it is incremented during every sample interval without checking for overflow:mm/damon/paddr.c:damon_pa_apply_probes() { ... damon_for_each_probe(p, ctx) { if (damon_pa_filter_pass(pa, folio, p)) r->probe_hits[i]++; ... }
Quoted 3 lines
If a user configures an aggregation interval that is > 255 times the sample interval, could highly-hit regions wrap around to 0 and be falsely identified as cold memory?
Known low priority issue. No blocker.
Quoted 7 lines
Quoted 4 lines
mm/damon/core.c +1
-0+ }return false; }Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=12
Thanks,
SJ
[...]
SJ Park
maintainerRFC PATCH 13/16 mm/damon/sysfs: implement probe/weight file
In reply to SJ ParkImplement a new sysfs file, 'weight', under probe directory. Users will be able to set the probe weight and enable the attributes only monitoring using it.
--- mm/damon/sysfs.c | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+)
mm/damon/sysfs.c +25 -0
@@ -1065,6 +1065,7 @@ static const struct kobj_type damon_sysfs_filters_ktype = {struct damon_sysfs_probe {struct kobject kobj; +unsigned int weight;struct damon_sysfs_filters *filters; };
@@ -1100,12 +1101,35 @@ static void damon_sysfs_probe_rm_dirs(struct damon_sysfs_probe *probe) } } +static ssize_t weight_show (struct kobject *kobj,struct kobj_attribute *attr, +char *buf) +{ +struct damon_sysfs_probe *probe =container_of (kobj, + struct damon_sysfs_probe, kobj); + +return sysfs_emit (buf,"%u\n" , probe->weight); +} + +static ssize_t weight_store (struct kobject *kobj, +struct kobj_attribute *attr,const char *buf,size_t count) +{ +struct damon_sysfs_probe *probe =container_of (kobj, + struct damon_sysfs_probe, kobj); +int err =kstrtouint (buf,0 , &probe->weight); + +return err ? err : count; +} +static void damon_sysfs_probe_release (struct kobject *kobj) {kfree (container_of(kobj ,struct damon_sysfs_probe ,kobj )); } +static struct kobj_attribute damon_sysfs_probe_weight_attr = +__ATTR_RW_MODE (weight,0600 ); +static struct attribute *damon_sysfs_probe_attrs[] = { + &damon_sysfs_probe_weight_attr.attr,NULL , };ATTRIBUTE_GROUPS (damon_sysfs_probe);
@@ -1965,6 +1989,7 @@ static int damon_sysfs_set_probes(struct damon_ctx *ctx,return -ENOMEM ;damon_add_probe (ctx, p); sys_probe = sys_probes->probes_arr[i]; + p->weight = sys_probe->weight; err =damon_sysfs_set_probe (p, sys_probe);if (err)return err;
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 13/16] mm/damon/sysfs: implement probe/weight file
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider: - [High] Unbounded `weight` from sysfs causes an integer overflow in `damon_probe_hits_wsum()`, which…
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Unbounded weight from sysfs causes an integer overflow in damon_probe_hits_wsum(), which corrupts the core region scoring algorithm.
- [High] Setting a large weight via sysfs trivially bypasses the max_nr_regions limit, leading to unbounded kernel memory usage and CPU starvation. --
commit 13d5322850c993787f439a6daf8065f8717e7f80
Author: SJ Park <[email protected]>
mm/damon/sysfs: implement probe/weight file
This commit implements a new sysfs file, weight, under the probe directory. Users can set the probe weight and enable monitoring using it.
Quoted 4 lines
mm/damon/sysfs.c +0
-0
[ ... ]
Quoted 1 line
mm/damon/sysfs.c +0
-0@@ -1100,12 +1101,35 @@ static void damon_sysfs_probe_rm_dirs(struct damon_sysfs_probe *probe)
[ ... ]
mm/damon/sysfs.c static void damon_sysfs_probe_rm_dirs(struct damon_sysfs_probe *probe) · line 1118
Quoted 6 lines
mm/damon/sysfs.c +6
-0+static ssize_t weight_store (struct kobject *kobj, +struct kobj_attribute *attr,const char *buf,size_t count) +{ +struct damon_sysfs_probe *probe =container_of (kobj, + struct damon_sysfs_probe, kobj); +int err =kstrtouint (buf,0 , &probe->weight);
[Severity: High]
Does this allow an unbounded unsigned int weight to be set, potentially causing an integer overflow during calculation?
In damon_probe_hits_wsum(), the score is calculated by multiplying the hits and the weight:
mm/damon/core.c:damon_probe_hits_wsum() {
...
sum += r->last_probe_hits[i++] * probe->weight;
...
}Because weight is allowed to be up to UINT_MAX, and probe_hits can be up to 255, could this calculation wrap around? If it wraps around, regions with a higher number of probe hits might erroneously compute a lower weighted sum than colder regions, which seems like it would break the ability to accurately track memory.
[Severity: High]
Could a large weight bypass the max_nr_regions limit and lead to unbounded region growth?
In kdamond_merge_regions(), the merge loop condition checks against max_thres:
mm/damon/core.c:kdamond_merge_regions() {
...
} while (nr_regions > c->attrs.max_nr_regions &&
threshold / 2 < max_thres);
...
}The initial threshold scales with the unbounded weight. If a user configures a large weight, does threshold / 2 immediately exceed max_thres on the first loop iteration?
This would cause the loop to terminate prematurely, failing to iteratively increase the threshold to merge regions. Could this allow the number of regions to grow without bound and lead to severe CPU starvation?
Quoted 4 lines
mm/damon/sysfs.c +4
-0+ +return err ? err : count; +} +
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=13
SJ Park
maintainerRe: [RFC PATCH v1.1 13/16] mm/damon/sysfs: implement probe/weight file
Quoted 44 lines · On Mon, 06 Jul 2026 15:19:38 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] Unboundedweightfrom sysfs causes an integer overflow indamon_probe_hits_wsum(), which corrupts the core region scoring algorithm.
- [High] Setting a largeweightvia sysfs trivially bypasses themax_nr_regionslimit, leading to unbounded kernel memory usage and CPU starvation. --commit
13d5322850c993787f439a6daf8065f8717e7f80
Author: SJ Park <[email protected]>
mm/damon/sysfs: implement probe/weight fileThis commit implements a new sysfs file, weight, under the probe directory. Users can set the probe weight and enable monitoring using it.
Quoted 4 lines
mm/damon/sysfs.c +0
-0[ ... ]
Quoted 1 line
mm/damon/sysfs.c +0
-0@@ -1100,12 +1101,35 @@ static void damon_sysfs_probe_rm_dirs(struct damon_sysfs_probe *probe)[ ... ]
mm/damon/sysfs.c static void damon_sysfs_probe_rm_dirs(struct damon_sysfs_probe *probe) · line 1118
Quoted 6 lines
mm/damon/sysfs.c +6
-0+static ssize_t weight_store (struct kobject *kobj, +struct kobj_attribute *attr,const char *buf,size_t count) +{ +struct damon_sysfs_probe *probe =container_of (kobj, + struct damon_sysfs_probe, kobj); +int err =kstrtouint (buf,0 , &probe->weight);[Severity: High]
Does this allow an unbounded unsigned int weight to be set, potentially causing an integer overflow during calculation?In
damon_probe_hits_wsum(), the score is calculated by multiplying the hits and the weight:mm/damon/core.c:damon_probe_hits_wsum() { ... sum += r->last_probe_hits[i++] * probe->weight; ... }
Quoted 5 lines
Because weight is allowed to be up to
UINT_MAX, andprobe_hitscan be up to 255, could this calculation wrap around? If it wraps around, regions with a higher number of probe hits might erroneously compute a lower weighted sum than colder regions, which seems like it would break the ability to accurately track memory.
A minor issue for unlikely user setup. But I agree this is better to be fixed, and the core layer parameters validation makes this easy to fix. I will work on fixing this, in the next version of this series, or separately.
Quoted 15 lines
[Severity: High]
Could a large weight bypass themax_nr_regionslimit and lead to unbounded region growth?In
kdamond_merge_regions(), the merge loop condition checks againstmax_thres:mm/damon/core.c:kdamond_merge_regions() { ... } while (nr_regions > c->attrs.max_nr_regions && threshold / 2 < max_thres); ... }
Quoted 7 lines
The initial threshold scales with the unbounded weight. If a user configures a large weight, does threshold / 2 immediately exceed
max_threson the first loop iteration?This would cause the loop to terminate prematurely, failing to iteratively increase the threshold to merge regions. Could this allow the number of regions to grow without bound and lead to severe CPU starvation?
Yes, but only for one aggregation interval. I don't think this is an issue.
Quoted 8 lines
Quoted 5 lines
mm/damon/sysfs.c +4
-0+ +return err ? err : count; +} +Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=13
Thanks,
SJ
[...]
SJ Park
maintainerRFC PATCH 14/16 Docs/mm/damon/design: document attrs-only monitoring
In reply to SJ ParkUpdate DAMON design document for the attributes only monitoring mode.
--- Documentation/mm/damon/design.rst | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+)
Documentation/mm/damon/design.rst +20 -0
@@ -316,6 +316,26 @@ Another way to do this for higher accuracy is using :ref:`DAMOS filter information in page level. But, because it is operated in page level, the overhead is proportional to the size of the memory. +Data Attributes-only Monitoring +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Data access is the primary monitoring information for DAMON. Hence it +:ref:`adjusts <damon_design_data_attrs_monitoring>` regions using the access +:ref:`counter <damon_design_region_based_sampling>` (``nr_accesses``). In some +use cases, however, users may want some of :ref:`attributes +<damon_design_data_attrs_monitoring>` to be the primary information. + +Data attributes-only monitoring mode supports this use case. For the mode, +each attribute probe has their priority weight value. Users can describe by +what combination of the attributes the primary information is decided, by +setting the priority weight value. If the total sum of the weights is not +zero, the mode is enabled. The regions adjustment mechanism uses the weighted +sum of the :ref:`probe hit counts <damon_design_data_attrs_monitoring>` instead +of ``nr_accesses`` in the case. When the mode is enabled, access monitoring is +automatically turned off. The access counter (``nr_accesses``) will always be +zero and not updated. Hence the mode is called Data Attributes "only" +monitoring. + Dynamic Target Space Updates Handling ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Signature · 1 line
2.47.3
[email protected]
AI reviewRe: [RFC PATCH v1.1 14/16] Docs/mm/damon/design: document attrs-only monitoring
Thank you for your contribution! Sashiko AI review found 4 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 4 potential issue(s) to consider:
New issues:
- [High] Multiplication and accumulation in damon_probe_hits_wsum() can silently overflow a 32-bit unsigned int.
- [High] Silently disables all monitoring if probe weights are configured for operations sets that lack probe support (e.g., vaddr).
Pre-existing issues:
- [High] The abs() macro behaves incorrectly for large unsigned int differences, breaking region adjustment logic.
- [High] The probe_hits array relies on 8-bit unsigned char, leading to silent wraparound during long aggregation intervals. --
commit 61ffda736cfeed6cc995344cdd1d753a189751a8
Author: SJ Park <[email protected]>
Docs/mm/damon/design: document attrs-only monitoring
This patch series introduces Attributes-only Monitoring for DAMON, adding probe weights to dictate memory management scoring. When weights are configured, data access monitoring is disabled and regions are evaluated based on weighted probe hits instead.
Quoted 4 lines
mm/damon/core.c +0
-0
[ ... ]
Quoted 14 lines
mm/damon/core.c +10
-0@@ -389,6 +401,30 @@ static bool damon_is_last_region(struct damon_region *r,return list_is_last (&r->list, &t->regions_list); } +/** +* damon_probe_hits_wsum() - Returns probe hits weighted sum of a region. +* @r: region to get the weigted sum of. +* @last: if the request is for last-window agggregated probe hits. +* @ctx: context of &r. +* +* Return: the weighted sum of probe hits of the region. +*/ +unsigned int damon_probe_hits_wsum (struct damon_region *r,bool last, +struct damon_ctx *ctx)
mm/damon/core.c static bool damon_is_last_region(struct damon_region *r, · line 423
Quoted 10 lines
mm/damon/core.c +10
-0+{ +struct damon_probe *probe; +unsigned int sum =0 ; +int i =0 ; + +damon_for_each_probe (probe, ctx) { +if (last) + sum += r->last_probe_hits[i++] * probe->weight; +else + sum += r->probe_hits[i++] * probe->weight;
[Severity: High]
Does sum need protection against integer overflow here?
If a user configures a large priority weight for a probe via sysfs, this multiplication can silently overflow the 32-bit unsigned int sum (for example, 255 * 4294967295).
Could this integer overflow wrap around the score metric, making heavily weighted regions incorrectly appear as "cold" regions and corrupting the core memory management data?
[Severity: High]
This is a pre-existing issue, but since Attributes-only Monitoring makes probe hits the primary metric, could probe_hits overflow its 8-bit limit?
In include/linux/damon.h, the array is defined as an 8-bit value:
struct damon_region {
...
unsigned char probe_hits[DAMON_MAX_PROBES];
...
};If a user configures an aggregation window that encompasses more than 255 samples, the counter increment in damon_pa_apply_probes():
r->probe_hits[i]++;
will repeatedly wrap around, destroying the statistical data and corrupting the region scoring.
Quoted 3 lines
mm/damon/core.c +3
-0+ } +return sum; +}
[ ... ]
Quoted 8 lines
mm/damon/core.c +1
-1@@ -3211,24 +3259,37 @@ static void damon_merge_two_regions(struct damon_target *t,* sz_limit size upper limit of each region */ static void damon_merge_regions_of (struct damon_target *t,unsigned int thres, -unsigned long sz_limit) +unsigned long sz_limit,struct damon_ctx *ctx) {struct damon_region *r, *prev =NULL , *next;
mm/damon/core.c static void damon_merge_two_regions(struct damon_target *t, · line 3273
Quoted 10 lines
mm/damon/core.c +7
-1+bool use_probe_hits =damon_has_probe_weights (ctx);damon_for_each_region_safe (r, next, t) { -if (abs (r->nr_accesses - r->last_nr_accesses) > thres) +unsigned int score, last_score; + + score =damon_merge_score (r, false, ctx, use_probe_hits); + last_score =damon_merge_score (r, true, ctx, use_probe_hits); + +if (abs (score - last_score) > thres)
[Severity: High]
This is a pre-existing issue, but it seems exacerbated now that arbitrary weight multipliers are introduced. Does abs() behave correctly for large unsigned int differences?
The abs() macro implicitly casts its arguments to signed int. If the absolute difference between score and last_score exceeds INT_MAX, it will overflow and be evaluated incorrectly as a smaller positive number.
Because weights can now easily push scores past INT_MAX, will this cause damon_merge_regions_of() to improperly ignore large score differences?
Quoted 4 lines
mm/damon/core.c +1
-1r->age =0 ; - else if ((r->nr_accesses == 0) != (r->last_nr_accesses == 0)) + else if ((score == 0) != (last_score == 0)) r->age = 0;
[ ... ]
Quoted 15 lines
mm/damon/core.c +4
-1@@ -3658,25 +3718,38 @@ static int kdamond_fn(void *data)unsigned long next_aggregation_sis = ctx->next_aggregation_sis;unsigned long next_ops_update_sis = ctx->next_ops_update_sis;unsigned long sample_interval = ctx->attrs.sample_interval; +bool access_check_disabled =damon_has_probe_weights (ctx); +unsigned int max_merge_score =0 , max_wsum; +bool get_max_wsum;if (kdamond_wait_activation (ctx))break ; -if (ctx->ops.prepare_access_checks) +if (!access_check_disabled && ctx->ops.prepare_access_checks) ctx->ops.prepare_access_checks (ctx);
mm/damon/core.c static int kdamond_fn(void *data) · line 3736
Quoted 10 lines
mm/damon/core.c +3
-4kdamond_usleep (sample_interval); ctx->passed_sample_intervals++; -if (ctx->ops.check_accesses) - max_nr_accesses = ctx->ops.check_accesses (ctx); -if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx); +if (!access_check_disabled && ctx->ops.check_accesses) + max_merge_score = ctx->ops.check_accesses (ctx); +if (ctx->ops.apply_probes) {
[Severity: High]
If a user configures probe weights via sysfs for an operations set that doesn't implement apply_probes (e.g., DAMON_OPS_VADDR), will this silently disable all monitoring?
Because access_check_disabled becomes true unconditionally when weights are present, default access checking is skipped. Without ctx->ops.apply_probes, no probes are applied either, effectively breaking the DAMON monitoring thread for that operations set.
Quoted 3 lines
mm/damon/core.c +3
-0+if (time_after_eq (ctx->passed_sample_intervals, + next_aggregation_sis) && + access_check_disabled)
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=14
SJ Park
maintainerRe: [RFC PATCH v1.1 14/16] Docs/mm/damon/design: document attrs-only monitoring
Quoted 56 lines · On Mon, 06 Jul 2026 15:34:57 +0000 [email protected] wrote:
Thank you for your contribution! Sashiko AI review found 4 potential issue(s) to consider:
New issues:
- [High] Multiplication and accumulation indamon_probe_hits_wsum()can silently overflow a 32-bitunsigned int.
- [High] Silently disables all monitoring if probe weights are configured for operations sets that lack probe support (e.g.,vaddr).Pre-existing issues:
- [High] Theabs()macro behaves incorrectly for largeunsigned intdifferences, breaking region adjustment logic.
- [High] Theprobe_hitsarray relies on 8-bitunsigned char, leading to silent wraparound during long aggregation intervals. --commit
61ffda736cfeed6cc995344cdd1d753a189751a8
Author: SJ Park <[email protected]>Docs/mm/damon/design: document attrs-only monitoring
This patch series introduces Attributes-only Monitoring for DAMON, adding probe weights to dictate memory management scoring. When weights are configured, data access monitoring is disabled and regions are evaluated based on weighted probe hits instead.
Quoted 4 lines
mm/damon/core.c +0
-0[ ... ]
Quoted 14 lines
mm/damon/core.c +10
-0@@ -389,6 +401,30 @@ static bool damon_is_last_region(struct damon_region *r,return list_is_last (&r->list, &t->regions_list); } +/** +* damon_probe_hits_wsum() - Returns probe hits weighted sum of a region. +* @r: region to get the weigted sum of. +* @last: if the request is for last-window agggregated probe hits. +* @ctx: context of &r. +* +* Return: the weighted sum of probe hits of the region. +*/ +unsigned int damon_probe_hits_wsum (struct damon_region *r,bool last, +struct damon_ctx *ctx)mm/damon/core.c static bool damon_is_last_region(struct damon_region *r, · line 423
Quoted 10 lines
mm/damon/core.c +10
-0+{ +struct damon_probe *probe; +unsigned int sum =0 ; +int i =0 ; + +damon_for_each_probe (probe, ctx) { +if (last) + sum += r->last_probe_hits[i++] * probe->weight; +else + sum += r->probe_hits[i++] * probe->weight;[Severity: High]
Does sum need protection against integer overflow here?
Quoted 7 lines
If a user configures a large priority weight for a probe via sysfs, this multiplication can silently overflow the 32-bit unsigned int sum (for example, 255 * 4294967295).
Could this integer overflow wrap around the score metric, making heavily weighted regions incorrectly appear as "cold" regions and corrupting the core memory management data?
Will fix in the next version of this series, or separately.
Quoted 13 lines
[Severity: High]
This is a pre-existing issue, but since Attributes-only Monitoring makes probe hits the primary metric, couldprobe_hitsoverflow its 8-bit limit?In
include/linux/damon.h, the array is defined as an 8-bit value:struct damon_region { ... unsigned char probe_hits[DAMON_MAX_PROBES]; ... };
Quoted 7 lines
If a user configures an aggregation window that encompasses more than 255 samples, the counter increment in
damon_pa_apply_probes():r->probe_hits[i]++;will repeatedly wrap around, destroying the statistical data and corrupting the region scoring.
Known low priority issue. No blocker. But I will consider fixing this in the next revision or seaprately.
Quoted 31 lines
Quoted 4 lines
mm/damon/core.c +3
-0+ } +return sum; +}[ ... ]
Quoted 8 lines
mm/damon/core.c +1
-1@@ -3211,24 +3259,37 @@ static void damon_merge_two_regions(struct damon_target *t,* sz_limit size upper limit of each region */ static void damon_merge_regions_of (struct damon_target *t,unsigned int thres, -unsigned long sz_limit) +unsigned long sz_limit,struct damon_ctx *ctx) {struct damon_region *r, *prev =NULL , *next;mm/damon/core.c static void damon_merge_two_regions(struct damon_target *t, · line 3273
Quoted 10 lines
mm/damon/core.c +7
-1+bool use_probe_hits =damon_has_probe_weights (ctx);damon_for_each_region_safe (r, next, t) { -if (abs (r->nr_accesses - r->last_nr_accesses) > thres) +unsigned int score, last_score; + + score =damon_merge_score (r, false, ctx, use_probe_hits); + last_score =damon_merge_score (r, true, ctx, use_probe_hits); + +if (abs (score - last_score) > thres)[Severity: High]
This is a pre-existing issue, but it seems exacerbated now that arbitrary weight multipliers are introduced. Doesabs()behave correctly for large unsigned int differences?
Quoted 6 lines
The
abs()macro implicitly casts its arguments to signed int. If the absolute difference between score andlast_scoreexceedsINT_MAX, it will overflow and be evaluated incorrectly as a smaller positive number.Because weights can now easily push scores past
INT_MAX, will this causedamon_merge_regions_of()to improperly ignore large score differences?
Will fix in the next revision.
Quoted 39 lines
Quoted 5 lines
mm/damon/core.c +1
-1r->age =0 ; - else if ((r->nr_accesses == 0) != (r->last_nr_accesses == 0)) + else if ((score == 0) != (last_score == 0)) r->age = 0;[ ... ]
Quoted 15 lines
mm/damon/core.c +4
-1@@ -3658,25 +3718,38 @@ static int kdamond_fn(void *data)unsigned long next_aggregation_sis = ctx->next_aggregation_sis;unsigned long next_ops_update_sis = ctx->next_ops_update_sis;unsigned long sample_interval = ctx->attrs.sample_interval; +bool access_check_disabled =damon_has_probe_weights (ctx); +unsigned int max_merge_score =0 , max_wsum; +bool get_max_wsum;if (kdamond_wait_activation (ctx))break ; -if (ctx->ops.prepare_access_checks) +if (!access_check_disabled && ctx->ops.prepare_access_checks) ctx->ops.prepare_access_checks (ctx);mm/damon/core.c static int kdamond_fn(void *data) · line 3736
Quoted 10 lines
mm/damon/core.c +3
-4kdamond_usleep (sample_interval); ctx->passed_sample_intervals++; -if (ctx->ops.check_accesses) - max_nr_accesses = ctx->ops.check_accesses (ctx); -if (ctx->ops.apply_probes) - ctx->ops.apply_probes (ctx); +if (!access_check_disabled && ctx->ops.check_accesses) + max_merge_score = ctx->ops.check_accesses (ctx); +if (ctx->ops.apply_probes) {[Severity: High]
If a user configures probe weights via sysfs for an operations set that doesn't implementapply_probes(e.g.,DAMON_OPS_VADDR), will this silently disable all monitoring?
Quoted 4 lines
Because
access_check_disabledbecomes true unconditionally when weights are present, default access checking is skipped. Withoutctx->ops.apply_probes, no probes are applied either, effectively breaking the DAMON monitoring thread for that operations set.
I will fix this in the next revision, or separately.
Quoted 7 lines
Quoted 4 lines
mm/damon/core.c +3
-0+if (time_after_eq (ctx->passed_sample_intervals, + next_aggregation_sis) && + access_check_disabled)Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=14
Thanks,
SJ
[...]
SJ Park
maintainerRFC PATCH 15/16 Docs/admin-guide/mm/damon/usage: document weight sysfs file
In reply to SJ ParkUpdate DAMON usage document for newly added probe weight file.
--- Documentation/admin-guide/mm/damon/usage.rst | 11 ++++++++--- Documentation/mm/damon/design.rst | 5 +++++ 2 files changed, 13 insertions(+), 3 deletions(-)
Documentation/admin-guide/mm/damon/usage.rst +8 -3
@@ -73,9 +73,10 @@ comma (","). │ │ │ │ │ │ │ intervals_goal/access_bp,aggrs,min_sample_us,max_sample_us │ │ │ │ │ │ nr_regions/min,max │ │ │ │ │ │ :ref:`probes <damon_usage_sysfs_probes>`/nr_probes - │ │ │ │ │ │ │ 0/filters/nr_filters - │ │ │ │ │ │ │ │ 0/type,matching,allow,path - │ │ │ │ │ │ │ │ ... + │ │ │ │ │ │ │ 0/weight + │ │ │ │ │ │ │ │ filters/nr_filters + │ │ │ │ │ │ │ │ │ 0/type,matching,allow,path + │ │ │ │ │ │ │ │ │ ... │ │ │ │ │ │ │ ... │ │ │ │ │ :ref:`targets <sysfs_targets>`/nr_targets │ │ │ │ │ │ :ref:`0 <sysfs_target>`/pid_target,obsolete_target
@@ -286,6 +287,10 @@ In each probe directory, one directory, ``filters`` exists. The directory contains files for installing filters for the probe, that is used to determine the data attribute for the probe. +Each probe directory also contains ``weight`` file. Reading from and writing +to the file gets and sets the :ref:`attributes-only monitoring +<damon_design_attrs_only_monitoring>` weight for the attribute of the probe. + In the beginning, ``filters`` directory has only one file, ``nr_filters``. Writing a number (``N``) to the file creates the number of child directories named ``0`` to ``N-1``. Each directory represents each filter and works in a
Documentation/mm/damon/design.rst +5 -0
@@ -316,6 +316,8 @@ Another way to do this for higher accuracy is using :ref:`DAMOS filter information in page level. But, because it is operated in page level, the overhead is proportional to the size of the memory. +.. _damon_design_attrs_only_monitoring: + Data Attributes-only Monitoring ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
@@ -336,6 +338,9 @@ automatically turned off. The access counter (``nr_accesses``) will always be zero and not updated. Hence the mode is called Data Attributes "only" monitoring. +Refer to the :ref:`admin guide <damon_usage_sysfs_probes>` to know how users +can use the mode. + Dynamic Target Space Updates Handling ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Signature · 1 line
2.47.3
SJ Park
maintainerRFC PATCH 16/16 Docs/ABI/damon: document probe weight file
In reply to SJ ParkUpdate DAMON ABI document for the newly added probe weight sysfs file.
--- Documentation/ABI/testing/sysfs-kernel-mm-damon | 6 ++++++ 1 file changed, 6 insertions(+)
Documentation/ABI/testing/sysfs-kernel-mm-damon +6 -0
@@ -164,6 +164,12 @@ Description: Writing a number 'N' to this file creates the number of directories for each DAMON probe named '0' to 'N-1' under the probes/ directory. +What: /sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/monitoring_attrs/probes/<P>/weight +Date: Jul 2026 +Contact: SJ Park <[email protected]> +Description: Writing to and reading from this file sets and gets the + per-probe attribute weight. + What: /sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/monitoring_attrs/probes/<P>/filters/nr_filters Date: May 2026 Contact: SJ Park <[email protected]>
Signature · 1 line
2.47.3
SJ Park
maintainerQuoted 2 lines · On Mon, 6 Jul 2026 07:18:54 -0700 SJ Park <[email protected]> wrote:
TL;DR: Introduce a way to get DAMON's best effort accuracy monitoring of user-demanding non-access data attributes.
Sashiko is continuously finding a few things including integer overflows from wrong or weird user setups. Those are minor issues in my opinion. But because the core layer parameters validation patch series is merged into mm-new, fixing thosse should be straightforward.
I will fix simple ones in the way, as a part of the next series or as a separate series.
Thanks,
SJ
[...]