[RFC PATCH -next 00/16] mm/damon: support ARM32 with LPAE

[RFC PATCH -next 00/16] mm/damon: support ARM32 with LPAE

[Quanmin Yan]

Previously, DAMON's physical address space monitoring only supported
memory ranges below 4GB on LPAE-enabled systems. This was due to
the use of 'unsigned long' in 'struct damon_addr_range', which is
32-bit on ARM32 even with LPAE enabled.

Implements DAMON compatibility for ARM32 with LPAE enabled.

Patches 01/16 through 10/16 are from the mailing list[1], add a new core
layer parameter called 'addr_unit'. Operations set layer can translate a
core layer address to the real address by multiplying the parameter value
to the core layer address.

Patches 11/16 through 14/16 extend and complement patches 01~10, addressing
various issues introduced by the addr_unit implementation.

Patches 15/16 and 16/16 complete native DAMON support for 32-bit systems.

[1] https://lore.kernel.org/all/20250416042551.158131-1-sj@kernel.org/

Quanmin Yan (6):
  mm/damon: add addr_unit for DAMON_RECLAIM and LRU_SORT
  mm/damon: add damon_ctx->min_region and damon_target->min_region
  mm/damon/sysfs: ensure valid addr_unit setting in
    damon_sysfs_apply_inputs()
  mm/damon/core: convert sz to byte units when updating state
  mm/damon: the byte statistics data type in damos_stat uses unsigned
    long long
  mm/damon/core: handle quota->esz overflow issues

SeongJae Park (10):
  mm/damon/core: add damon_ctx->addr_unit
  mm/damon/paddr: support addr_unit for access monitoring
  mm/damon/paddr: support addr_unit for DAMOS_PAGEOUT
  mm/damon/paddr: support addr_unit for DAMOS_LRU_[DE]PRIO
  mm/damon/paddr: support addr_unit for MIGRATE_{HOT,COLD}
  mm/damon/paddr: support addr_unit for DAMOS_STAT
  mm/damon/sysfs: implement addr_unit file under context dir
  Docs/mm/damon/design: document 'address unit' parameter
  Docs/admin-guide/mm/damon/usage: document addr_unit file
  Docs/ABI/damon: document addr_unit file

 .../ABI/testing/sysfs-kernel-mm-damon         |  7 ++
 Documentation/admin-guide/mm/damon/usage.rst  | 11 ++-
 Documentation/mm/damon/design.rst             | 16 +++-
 include/linux/damon.h                         | 24 +++--
 mm/damon/core.c                               | 80 +++++++++++-----
 mm/damon/lru_sort.c                           | 16 +++-
 mm/damon/modules-common.c                     |  5 +-
 mm/damon/modules-common.h                     |  6 +-
 mm/damon/paddr.c                              | 95 ++++++++++++-------
 mm/damon/reclaim.c                            | 16 +++-
 mm/damon/stat.c                               |  2 +-
 mm/damon/sysfs-schemes.c                      | 12 +--
 mm/damon/sysfs.c                              | 27 ++++++
 13 files changed, 226 insertions(+), 91 deletions(-)

-- 
2.34.1

[RFC PATCH -next 01/16] mm/damon/core: add damon_ctx->addr_unit

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

In some cases, some of the real address that handled by the underlying
operations set cannot be handled by DAMON since it uses only 'unsinged
long' as the address type.  Using DAMON for physical address space
monitoring of 32 bit ARM devices with large physical address extension
(LPAE) is one example[1].

Add a parameter name 'addr_unit' to core layer to help such cases.
DAMON core API callers can set it as the scale factor that will be used
by the operations set for translating the core layer's addresses to the
real address by multiplying the parameter value to the core layer
address.  Support of the parameter is up to each operations set layer.
The support from the physical address space operations set (paddr) will
be added with following commits.

[1] https://lore.kernel.org/20250408075553.959388-1-zuoze1@huawei.com

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 include/linux/damon.h | 3 ++-
 mm/damon/core.c       | 3 +++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/include/linux/damon.h b/include/linux/damon.h
index f13664c62ddd..b85c6c669cd0 100644
--- a/include/linux/damon.h
+++ b/include/linux/damon.h
@@ -746,7 +746,7 @@ struct damon_attrs {
  * Accesses to other fields must be protected by themselves.
  *
  * @ops:	Set of monitoring operations for given use cases.
- *
+ * @addr_unit:	Scale factor for core to ops address conversion.
  * @adaptive_targets:	Head of monitoring targets (&damon_target) list.
  * @schemes:		Head of schemes (&damos) list.
  */
@@ -788,6 +788,7 @@ struct damon_ctx {
 	struct mutex kdamond_lock;
 
 	struct damon_operations ops;
+	unsigned long addr_unit;
 
 	struct list_head adaptive_targets;
 	struct list_head schemes;
diff --git a/mm/damon/core.c b/mm/damon/core.c
index 52a48c9316bc..1a8d3009d606 100644
--- a/mm/damon/core.c
+++ b/mm/damon/core.c
@@ -544,6 +544,8 @@ struct damon_ctx *damon_new_ctx(void)
 	ctx->attrs.min_nr_regions = 10;
 	ctx->attrs.max_nr_regions = 1000;
 
+	ctx->addr_unit = 1;
+
 	INIT_LIST_HEAD(&ctx->adaptive_targets);
 	INIT_LIST_HEAD(&ctx->schemes);
 
@@ -1213,6 +1215,7 @@ int damon_commit_ctx(struct damon_ctx *dst, struct damon_ctx *src)
 	if (err)
 		return err;
 	dst->ops = src->ops;
+	dst->addr_unit = src->addr_unit ? : 1;
 
 	return 0;
 }
-- 
2.34.1

[RFC PATCH -next 02/16] mm/damon/paddr: support addr_unit for access monitoring

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Add support of addr_unit paramer for access monitoing operations of
paddr.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 32 +++++++++++++++++++++-----------
 1 file changed, 21 insertions(+), 11 deletions(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index 53a55c5114fb..345a0ae63b19 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -18,7 +18,13 @@
 #include "../internal.h"
 #include "ops-common.h"
 
-static void damon_pa_mkold(unsigned long paddr)
+static phys_addr_t damon_pa_phys_addr(
+		unsigned long addr, unsigned long addr_unit)
+{
+	return (phys_addr_t)addr * addr_unit;
+}
+
+static void damon_pa_mkold(phys_addr_t paddr)
 {
 	struct folio *folio = damon_get_folio(PHYS_PFN(paddr));
 
@@ -29,11 +35,12 @@ static void damon_pa_mkold(unsigned long paddr)
 	folio_put(folio);
 }
 
-static void __damon_pa_prepare_access_check(struct damon_region *r)
+static void __damon_pa_prepare_access_check(struct damon_region *r,
+		unsigned long addr_unit)
 {
 	r->sampling_addr = damon_rand(r->ar.start, r->ar.end);
 
-	damon_pa_mkold(r->sampling_addr);
+	damon_pa_mkold(damon_pa_phys_addr(r->sampling_addr, addr_unit));
 }
 
 static void damon_pa_prepare_access_checks(struct damon_ctx *ctx)
@@ -43,11 +50,11 @@ static void damon_pa_prepare_access_checks(struct damon_ctx *ctx)
 
 	damon_for_each_target(t, ctx) {
 		damon_for_each_region(r, t)
-			__damon_pa_prepare_access_check(r);
+			__damon_pa_prepare_access_check(r, ctx->addr_unit);
 	}
 }
 
-static bool damon_pa_young(unsigned long paddr, unsigned long *folio_sz)
+static bool damon_pa_young(phys_addr_t paddr, unsigned long *folio_sz)
 {
 	struct folio *folio = damon_get_folio(PHYS_PFN(paddr));
 	bool accessed;
@@ -62,23 +69,25 @@ static bool damon_pa_young(unsigned long paddr, unsigned long *folio_sz)
 }
 
 static void __damon_pa_check_access(struct damon_region *r,
-		struct damon_attrs *attrs)
+		struct damon_attrs *attrs, unsigned long addr_unit)
 {
-	static unsigned long last_addr;
+	static phys_addr_t last_addr;
 	static unsigned long last_folio_sz = PAGE_SIZE;
 	static bool last_accessed;
+	phys_addr_t sampling_addr = damon_pa_phys_addr(
+			r->sampling_addr, addr_unit);
 
 	/* If the region is in the last checked page, reuse the result */
 	if (ALIGN_DOWN(last_addr, last_folio_sz) ==
-				ALIGN_DOWN(r->sampling_addr, last_folio_sz)) {
+				ALIGN_DOWN(sampling_addr, last_folio_sz)) {
 		damon_update_region_access_rate(r, last_accessed, attrs);
 		return;
 	}
 
-	last_accessed = damon_pa_young(r->sampling_addr, &last_folio_sz);
+	last_accessed = damon_pa_young(sampling_addr, &last_folio_sz);
 	damon_update_region_access_rate(r, last_accessed, attrs);
 
-	last_addr = r->sampling_addr;
+	last_addr = sampling_addr;
 }
 
 static unsigned int damon_pa_check_accesses(struct damon_ctx *ctx)
@@ -89,7 +98,8 @@ static unsigned int damon_pa_check_accesses(struct damon_ctx *ctx)
 
 	damon_for_each_target(t, ctx) {
 		damon_for_each_region(r, t) {
-			__damon_pa_check_access(r, &ctx->attrs);
+			__damon_pa_check_access(
+					r, &ctx->attrs, ctx->addr_unit);
 			max_nr_accesses = max(r->nr_accesses, max_nr_accesses);
 		}
 	}
-- 
2.34.1

[RFC PATCH -next 03/16] mm/damon/paddr: support addr_unit for DAMOS_PAGEOUT

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Add support of addr_unit for DAMOS_PAGEOUT action handling from the
DAMOS operation implementation for the physical address space.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index 345a0ae63b19..b548813a0472 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -135,10 +135,12 @@ static bool damon_pa_invalid_damos_folio(struct folio *folio, struct damos *s)
 	return false;
 }
 
-static unsigned long damon_pa_pageout(struct damon_region *r, struct damos *s,
+static unsigned long damon_pa_pageout(struct damon_region *r,
+		unsigned long addr_unit, struct damos *s,
 		unsigned long *sz_filter_passed)
 {
-	unsigned long addr, applied;
+	phys_addr_t addr;
+	unsigned long applied;
 	LIST_HEAD(folio_list);
 	bool install_young_filter = true;
 	struct damos_filter *filter;
@@ -159,8 +161,8 @@ static unsigned long damon_pa_pageout(struct damon_region *r, struct damos *s,
 		damos_add_filter(s, filter);
 	}
 
-	addr = r->ar.start;
-	while (addr < r->ar.end) {
+	addr = damon_pa_phys_addr(r->ar.start, addr_unit);
+	while (addr < damon_pa_phys_addr(r->ar.end, addr_unit)) {
 		folio = damon_get_folio(PHYS_PFN(addr));
 		if (damon_pa_invalid_damos_folio(folio, s)) {
 			addr += PAGE_SIZE;
@@ -311,9 +313,11 @@ static unsigned long damon_pa_apply_scheme(struct damon_ctx *ctx,
 		struct damon_target *t, struct damon_region *r,
 		struct damos *scheme, unsigned long *sz_filter_passed)
 {
+	unsigned long aunit = ctx->addr_unit;
+
 	switch (scheme->action) {
 	case DAMOS_PAGEOUT:
-		return damon_pa_pageout(r, scheme, sz_filter_passed);
+		return damon_pa_pageout(r, aunit, scheme, sz_filter_passed);
 	case DAMOS_LRU_PRIO:
 		return damon_pa_mark_accessed(r, scheme, sz_filter_passed);
 	case DAMOS_LRU_DEPRIO:
-- 
2.34.1

[RFC PATCH -next 04/16] mm/damon/paddr: support addr_unit for DAMOS_LRU_[DE]PRIO

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Add support of addr_unit for DAMOS_LRU_PRIO and DAMOS_LRU_DEPRIO action
handling from the DAMOS operation implementation for the physical
address space.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index b548813a0472..b8a7f462967b 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -195,14 +195,16 @@ static unsigned long damon_pa_pageout(struct damon_region *r,
 }
 
 static inline unsigned long damon_pa_mark_accessed_or_deactivate(
-		struct damon_region *r, struct damos *s, bool mark_accessed,
+		struct damon_region *r, unsigned long addr_unit,
+		struct damos *s, bool mark_accessed,
 		unsigned long *sz_filter_passed)
 {
-	unsigned long addr, applied = 0;
+	phys_addr_t addr;
+	unsigned long applied = 0;
 	struct folio *folio;
 
-	addr = r->ar.start;
-	while (addr < r->ar.end) {
+	addr = damon_pa_phys_addr(r->ar.start, addr_unit);
+	while (addr < damon_pa_phys_addr(r->ar.end, addr_unit)) {
 		folio = damon_get_folio(PHYS_PFN(addr));
 		if (damon_pa_invalid_damos_folio(folio, s)) {
 			addr += PAGE_SIZE;
@@ -228,16 +230,18 @@ static inline unsigned long damon_pa_mark_accessed_or_deactivate(
 }
 
 static unsigned long damon_pa_mark_accessed(struct damon_region *r,
-	struct damos *s, unsigned long *sz_filter_passed)
+		unsigned long addr_unit, struct damos *s,
+		unsigned long *sz_filter_passed)
 {
-	return damon_pa_mark_accessed_or_deactivate(r, s, true,
+	return damon_pa_mark_accessed_or_deactivate(r, addr_unit, s, true,
 			sz_filter_passed);
 }
 
 static unsigned long damon_pa_deactivate_pages(struct damon_region *r,
-	struct damos *s, unsigned long *sz_filter_passed)
+		unsigned long addr_unit, struct damos *s,
+		unsigned long *sz_filter_passed)
 {
-	return damon_pa_mark_accessed_or_deactivate(r, s, false,
+	return damon_pa_mark_accessed_or_deactivate(r, addr_unit, s, false,
 			sz_filter_passed);
 }
 
@@ -319,9 +323,11 @@ static unsigned long damon_pa_apply_scheme(struct damon_ctx *ctx,
 	case DAMOS_PAGEOUT:
 		return damon_pa_pageout(r, aunit, scheme, sz_filter_passed);
 	case DAMOS_LRU_PRIO:
-		return damon_pa_mark_accessed(r, scheme, sz_filter_passed);
+		return damon_pa_mark_accessed(r, aunit, scheme,
+				sz_filter_passed);
 	case DAMOS_LRU_DEPRIO:
-		return damon_pa_deactivate_pages(r, scheme, sz_filter_passed);
+		return damon_pa_deactivate_pages(r, aunit, scheme,
+				sz_filter_passed);
 	case DAMOS_MIGRATE_HOT:
 	case DAMOS_MIGRATE_COLD:
 		return damon_pa_migrate(r, scheme, sz_filter_passed);
-- 
2.34.1

[RFC PATCH -next 05/16] mm/damon/paddr: support addr_unit for MIGRATE_{HOT,COLD}

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Add support of addr_unit for DAMOS_MIGRATE_HOT and DAMOS_MIGRATE_COLD
action handling from the DAMOS operation implementation for the physical
address space.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index b8a7f462967b..76e1ee82b441 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -245,15 +245,17 @@ static unsigned long damon_pa_deactivate_pages(struct damon_region *r,
 			sz_filter_passed);
 }
 
-static unsigned long damon_pa_migrate(struct damon_region *r, struct damos *s,
+static unsigned long damon_pa_migrate(struct damon_region *r,
+		unsigned long addr_unit, struct damos *s,
 		unsigned long *sz_filter_passed)
 {
-	unsigned long addr, applied;
+	phys_addr_t addr;
+	unsigned long applied;
 	LIST_HEAD(folio_list);
 	struct folio *folio;
 
-	addr = r->ar.start;
-	while (addr < r->ar.end) {
+	addr = damon_pa_phys_addr(r->ar.start, addr_unit);
+	while (addr < damon_pa_phys_addr(r->ar.end, addr_unit)) {
 		folio = damon_get_folio(PHYS_PFN(addr));
 		if (damon_pa_invalid_damos_folio(folio, s)) {
 			addr += PAGE_SIZE;
@@ -330,7 +332,7 @@ static unsigned long damon_pa_apply_scheme(struct damon_ctx *ctx,
 				sz_filter_passed);
 	case DAMOS_MIGRATE_HOT:
 	case DAMOS_MIGRATE_COLD:
-		return damon_pa_migrate(r, scheme, sz_filter_passed);
+		return damon_pa_migrate(r, aunit, scheme, sz_filter_passed);
 	case DAMOS_STAT:
 		return damon_pa_stat(r, scheme, sz_filter_passed);
 	default:
-- 
2.34.1

[RFC PATCH -next 06/16] mm/damon/paddr: support addr_unit for DAMOS_STAT

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Add support of addr_unit for DAMOS_STAT action handling from the DAMOS
operation implementation for the physical address space.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index 76e1ee82b441..530bc9d3ce3b 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -289,17 +289,18 @@ static bool damon_pa_scheme_has_filter(struct damos *s)
 	return false;
 }
 
-static unsigned long damon_pa_stat(struct damon_region *r, struct damos *s,
+static unsigned long damon_pa_stat(struct damon_region *r,
+		unsigned long addr_unit, struct damos *s,
 		unsigned long *sz_filter_passed)
 {
-	unsigned long addr;
+	phys_addr_t addr;
 	struct folio *folio;
 
 	if (!damon_pa_scheme_has_filter(s))
 		return 0;
 
-	addr = r->ar.start;
-	while (addr < r->ar.end) {
+	addr = damon_pa_phys_addr(r->ar.start, addr_unit);
+	while (addr < damon_pa_phys_addr(r->ar.end, addr_unit)) {
 		folio = damon_get_folio(PHYS_PFN(addr));
 		if (damon_pa_invalid_damos_folio(folio, s)) {
 			addr += PAGE_SIZE;
@@ -334,7 +335,7 @@ static unsigned long damon_pa_apply_scheme(struct damon_ctx *ctx,
 	case DAMOS_MIGRATE_COLD:
 		return damon_pa_migrate(r, aunit, scheme, sz_filter_passed);
 	case DAMOS_STAT:
-		return damon_pa_stat(r, scheme, sz_filter_passed);
+		return damon_pa_stat(r, aunit, scheme, sz_filter_passed);
 	default:
 		/* DAMOS actions that not yet supported by 'paddr'. */
 		break;
-- 
2.34.1

[RFC PATCH -next 07/16] mm/damon/sysfs: implement addr_unit file under context dir

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Only DAMON kernel API callers can use addr_unit parameter.  Implement a
sysfs file to let DAMON sysfs ABI users use it.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/sysfs.c | 26 ++++++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/mm/damon/sysfs.c b/mm/damon/sysfs.c
index 6d2b0dab50cb..bea782b0a711 100644
--- a/mm/damon/sysfs.c
+++ b/mm/damon/sysfs.c
@@ -834,6 +834,7 @@ static const struct damon_sysfs_ops_name damon_sysfs_ops_names[] = {
 struct damon_sysfs_context {
 	struct kobject kobj;
 	enum damon_ops_id ops_id;
+	unsigned long addr_unit;
 	struct damon_sysfs_attrs *attrs;
 	struct damon_sysfs_targets *targets;
 	struct damon_sysfs_schemes *schemes;
@@ -849,6 +850,7 @@ static struct damon_sysfs_context *damon_sysfs_context_alloc(
 		return NULL;
 	context->kobj = (struct kobject){};
 	context->ops_id = ops_id;
+	context->addr_unit = 1;
 	return context;
 }
 
@@ -997,6 +999,25 @@ static ssize_t operations_store(struct kobject *kobj,
 	return -EINVAL;
 }
 
+static ssize_t addr_unit_show(struct kobject *kobj,
+		struct kobj_attribute *attr, char *buf)
+{
+	struct damon_sysfs_context *context = container_of(kobj,
+			struct damon_sysfs_context, kobj);
+
+	return sysfs_emit(buf, "%lu\n", context->addr_unit);
+}
+
+static ssize_t addr_unit_store(struct kobject *kobj,
+		struct kobj_attribute *attr, const char *buf, size_t count)
+{
+	struct damon_sysfs_context *context = container_of(kobj,
+			struct damon_sysfs_context, kobj);
+	int err = kstrtoul(buf, 0, &context->addr_unit);
+
+	return err ? err : count;
+}
+
 static void damon_sysfs_context_release(struct kobject *kobj)
 {
 	kfree(container_of(kobj, struct damon_sysfs_context, kobj));
@@ -1008,9 +1029,13 @@ static struct kobj_attribute damon_sysfs_context_avail_operations_attr =
 static struct kobj_attribute damon_sysfs_context_operations_attr =
 		__ATTR_RW_MODE(operations, 0600);
 
+static struct kobj_attribute damon_sysfs_context_addr_unit_attr =
+		__ATTR_RW_MODE(addr_unit, 0600);
+
 static struct attribute *damon_sysfs_context_attrs[] = {
 	&damon_sysfs_context_avail_operations_attr.attr,
 	&damon_sysfs_context_operations_attr.attr,
+	&damon_sysfs_context_addr_unit_attr.attr,
 	NULL,
 };
 ATTRIBUTE_GROUPS(damon_sysfs_context);
@@ -1397,6 +1422,7 @@ static int damon_sysfs_apply_inputs(struct damon_ctx *ctx,
 	err = damon_select_ops(ctx, sys_ctx->ops_id);
 	if (err)
 		return err;
+	ctx->addr_unit = sys_ctx->addr_unit;
 	err = damon_sysfs_set_attrs(ctx, sys_ctx->attrs);
 	if (err)
 		return err;
-- 
2.34.1

[RFC PATCH -next 08/16] Docs/mm/damon/design: document 'address unit' parameter

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Add 'addr_unit' parameter description on DAMON design document.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 Documentation/mm/damon/design.rst | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/Documentation/mm/damon/design.rst b/Documentation/mm/damon/design.rst
index 03f8137256f5..cffa320ead88 100644
--- a/Documentation/mm/damon/design.rst
+++ b/Documentation/mm/damon/design.rst
@@ -67,7 +67,7 @@ processes, NUMA nodes, files, and backing memory devices would be supportable.
 Also, if some architectures or devices support special optimized access check
 features, those will be easily configurable.
 
-DAMON currently provides below three operation sets.  Below two subsections
+DAMON currently provides below three operation sets.  Below three subsections
 describe how those work.
 
  - vaddr: Monitor virtual address spaces of specific processes
@@ -135,6 +135,18 @@ the interference is the responsibility of sysadmins.  However, it solves the
 conflict with the reclaim logic using ``PG_idle`` and ``PG_young`` page flags,
 as Idle page tracking does.
 
+Address Unit
+------------
+
+DAMON core layer uses ``unsinged long`` type for monitoring target address
+ranges.  In some cases, the address space for a given operations set could be
+too large to be handled with the type.  ARM (32-bit) with large physical
+address extension is an example.  For such cases, a per-operations set
+parameter called ``address unit`` is provided.  It represents the scale factor
+that need to be multiplied to the core layer's address for calculating real
+address on the given address space.  Support of ``address unit`` parameter is
+up to each operations set implementation.  ``paddr`` is the only operations set
+implementation that supports the parameter.
 
 .. _damon_core_logic:
 
-- 
2.34.1

[RFC PATCH -next 09/16] Docs/admin-guide/mm/damon/usage: document addr_unit file

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Document addr_unit DAMON sysfs file on DAMON usage document.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 Documentation/admin-guide/mm/damon/usage.rst | 11 +++++++----
 Documentation/mm/damon/design.rst            |  2 ++
 2 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/Documentation/admin-guide/mm/damon/usage.rst b/Documentation/admin-guide/mm/damon/usage.rst
index ff3a2dda1f02..2cae60b6f3ca 100644
--- a/Documentation/admin-guide/mm/damon/usage.rst
+++ b/Documentation/admin-guide/mm/damon/usage.rst
@@ -61,7 +61,7 @@ comma (",").
     │ :ref:`kdamonds <sysfs_kdamonds>`/nr_kdamonds
     │ │ :ref:`0 <sysfs_kdamond>`/state,pid,refresh_ms
     │ │ │ :ref:`contexts <sysfs_contexts>`/nr_contexts
-    │ │ │ │ :ref:`0 <sysfs_context>`/avail_operations,operations
+    │ │ │ │ :ref:`0 <sysfs_context>`/avail_operations,operations,addr_unit
     │ │ │ │ │ :ref:`monitoring_attrs <sysfs_monitoring_attrs>`/
     │ │ │ │ │ │ intervals/sample_us,aggr_us,update_us
     │ │ │ │ │ │ │ intervals_goal/access_bp,aggrs,min_sample_us,max_sample_us
@@ -188,9 +188,9 @@ details).  At the moment, only one context per kdamond is supported, so only
 contexts/<N>/
 -------------
 
-In each context directory, two files (``avail_operations`` and ``operations``)
-and three directories (``monitoring_attrs``, ``targets``, and ``schemes``)
-exist.
+In each context directory, three files (``avail_operations``, ``operations``
+and ``addr_unit``) and three directories (``monitoring_attrs``, ``targets``,
+and ``schemes``) exist.
 
 DAMON supports multiple types of :ref:`monitoring operations
 <damon_design_configurable_operations_set>`, including those for virtual address
@@ -205,6 +205,9 @@ You can set and get what type of monitoring operations DAMON will use for the
 context by writing one of the keywords listed in ``avail_operations`` file and
 reading from the ``operations`` file.
 
+``addr_unit`` file is for setting and getting the :ref:`address unit
+<damon_design_addr_unit>` parameter of the operations set.
+
 .. _sysfs_monitoring_attrs:
 
 contexts/<N>/monitoring_attrs/
diff --git a/Documentation/mm/damon/design.rst b/Documentation/mm/damon/design.rst
index cffa320ead88..fd183cd09ef2 100644
--- a/Documentation/mm/damon/design.rst
+++ b/Documentation/mm/damon/design.rst
@@ -135,6 +135,8 @@ the interference is the responsibility of sysadmins.  However, it solves the
 conflict with the reclaim logic using ``PG_idle`` and ``PG_young`` page flags,
 as Idle page tracking does.
 
+.. _damon_design_addr_unit:
+
 Address Unit
 ------------
 
-- 
2.34.1

[RFC PATCH -next 10/16] Docs/ABI/damon: document addr_unit file

[Quanmin Yan]

From: SeongJae Park <sj@kernel.org>

Document addr_unit DAMON sysfs file on DAMON ABI document.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 Documentation/ABI/testing/sysfs-kernel-mm-damon | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/Documentation/ABI/testing/sysfs-kernel-mm-damon b/Documentation/ABI/testing/sysfs-kernel-mm-damon
index 6791d879759e..cf4d66bd119d 100644
--- a/Documentation/ABI/testing/sysfs-kernel-mm-damon
+++ b/Documentation/ABI/testing/sysfs-kernel-mm-damon
@@ -77,6 +77,13 @@ Description:	Writing a keyword for a monitoring operations set ('vaddr' for
 		Note that only the operations sets that listed in
 		'avail_operations' file are valid inputs.
 
+What:		/sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/addr_unit
+Date:		Apr 2025
+Contact:	SeongJae Park <sj@kernel.org>
+Description:	Writing an integer to this file sets the 'address unit'
+		parameter of the given operations set of the context.  Reading
+		the file returns the last-written 'address unit' value.
+
 What:		/sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/monitoring_attrs/intervals/sample_us
 Date:		Mar 2022
 Contact:	SeongJae Park <sj@kernel.org>
-- 
2.34.1

[RFC PATCH -next 11/16] mm/damon: add addr_unit for DAMON_RECLAIM and LRU_SORT

[Quanmin Yan]

In module DAMON_RECLAIM and DAMON_LRU_SORT, the damon_ctx is
independent of the core, necessitating dedicated addr_unit
integration for these features.
Additionally, if the input monitor_region_start and monitor_region_end
are both 0 while addr_unit is set to a non-zero valuethe default
system RAM range should be divided by addr_unit.

Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
---
 include/linux/damon.h     |  4 +++-
 mm/damon/core.c           | 14 ++++++++++----
 mm/damon/lru_sort.c       | 16 +++++++++++++---
 mm/damon/modules-common.c |  5 ++++-
 mm/damon/modules-common.h |  2 +-
 mm/damon/reclaim.c        | 16 +++++++++++++---
 mm/damon/stat.c           |  2 +-
 7 files changed, 45 insertions(+), 14 deletions(-)

diff --git a/include/linux/damon.h b/include/linux/damon.h
index b85c6c669cd0..1b7b4cf1a3c5 100644
--- a/include/linux/damon.h
+++ b/include/linux/damon.h
@@ -942,7 +942,9 @@ int damon_call(struct damon_ctx *ctx, struct damon_call_control *control);
 int damos_walk(struct damon_ctx *ctx, struct damos_walk_control *control);
 
 int damon_set_region_biggest_system_ram_default(struct damon_target *t,
-				unsigned long *start, unsigned long *end);
+						unsigned long *start,
+						unsigned long *end,
+						unsigned long addr_unit);
 
 #endif	/* CONFIG_DAMON */
 
diff --git a/mm/damon/core.c b/mm/damon/core.c
index 1a8d3009d606..803c30f64b94 100644
--- a/mm/damon/core.c
+++ b/mm/damon/core.c
@@ -2714,6 +2714,7 @@ static bool damon_find_biggest_system_ram(unsigned long *start,
  * @t:		The monitoring target to set the region.
  * @start:	The pointer to the start address of the region.
  * @end:	The pointer to the end address of the region.
+ * @addr_unit:	Scale factor for core to ops address conversion.
  *
  * This function sets the region of @t as requested by @start and @end.  If the
  * values of @start and @end are zero, however, this function finds the biggest
@@ -2724,16 +2725,21 @@ static bool damon_find_biggest_system_ram(unsigned long *start,
  * Return: 0 on success, negative error code otherwise.
  */
 int damon_set_region_biggest_system_ram_default(struct damon_target *t,
-			unsigned long *start, unsigned long *end)
+						unsigned long *start,
+						unsigned long *end,
+						unsigned long addr_unit)
 {
 	struct damon_addr_range addr_range;
 
 	if (*start > *end)
 		return -EINVAL;
 
-	if (!*start && !*end &&
-		!damon_find_biggest_system_ram(start, end))
-		return -EINVAL;
+	if (!*start && !*end) {
+		if (!damon_find_biggest_system_ram(start, end) || !addr_unit)
+			return -EINVAL;
+		*start /= addr_unit;
+		*end /= addr_unit;
+	}
 
 	addr_range.start = *start;
 	addr_range.end = *end;
diff --git a/mm/damon/lru_sort.c b/mm/damon/lru_sort.c
index 151a9de5ad8b..8107d08c5e4b 100644
--- a/mm/damon/lru_sort.c
+++ b/mm/damon/lru_sort.c
@@ -111,6 +111,14 @@ module_param(monitor_region_start, ulong, 0600);
 static unsigned long monitor_region_end __read_mostly;
 module_param(monitor_region_end, ulong, 0600);
 
+/*
+ * Scale factor for DAMON_LRU_SORT to ops address conversion.
+ *
+ * This parameter is used to convert to the actual physical address.
+ */
+static unsigned long addr_unit __read_mostly = 1;
+module_param(addr_unit, ulong, 0600);
+
 /*
  * PID of the DAMON thread
  *
@@ -194,7 +202,8 @@ static int damon_lru_sort_apply_parameters(void)
 	unsigned int hot_thres, cold_thres;
 	int err;
 
-	err = damon_modules_new_paddr_ctx_target(&param_ctx, &param_target);
+	err = damon_modules_new_paddr_ctx_target(&param_ctx, &param_target,
+			addr_unit);
 	if (err)
 		return err;
 
@@ -221,7 +230,8 @@ static int damon_lru_sort_apply_parameters(void)
 
 	err = damon_set_region_biggest_system_ram_default(param_target,
 					&monitor_region_start,
-					&monitor_region_end);
+					&monitor_region_end,
+					addr_unit);
 	if (err)
 		goto out;
 	err = damon_commit_ctx(ctx, param_ctx);
@@ -323,7 +333,7 @@ MODULE_PARM_DESC(enabled,
 
 static int __init damon_lru_sort_init(void)
 {
-	int err = damon_modules_new_paddr_ctx_target(&ctx, &target);
+	int err = damon_modules_new_paddr_ctx_target(&ctx, &target, 1);
 
 	if (err)
 		goto out;
diff --git a/mm/damon/modules-common.c b/mm/damon/modules-common.c
index 86d58f8c4f63..613b7cc99368 100644
--- a/mm/damon/modules-common.c
+++ b/mm/damon/modules-common.c
@@ -13,9 +13,10 @@
  * Allocate, set, and return a DAMON context for the physical address space.
  * @ctxp:	Pointer to save the point to the newly created context
  * @targetp:	Pointer to save the point to the newly created target
+ * @addr_unit:	Scale factor for modules to ops address conversion.
  */
 int damon_modules_new_paddr_ctx_target(struct damon_ctx **ctxp,
-		struct damon_target **targetp)
+		struct damon_target **targetp, unsigned long addr_unit)
 {
 	struct damon_ctx *ctx;
 	struct damon_target *target;
@@ -24,6 +25,8 @@ int damon_modules_new_paddr_ctx_target(struct damon_ctx **ctxp,
 	if (!ctx)
 		return -ENOMEM;
 
+	ctx->addr_unit = addr_unit;
+
 	if (damon_select_ops(ctx, DAMON_OPS_PADDR)) {
 		damon_destroy_ctx(ctx);
 		return -EINVAL;
diff --git a/mm/damon/modules-common.h b/mm/damon/modules-common.h
index f103ad556368..c7048a449321 100644
--- a/mm/damon/modules-common.h
+++ b/mm/damon/modules-common.h
@@ -46,4 +46,4 @@
 			0400);
 
 int damon_modules_new_paddr_ctx_target(struct damon_ctx **ctxp,
-		struct damon_target **targetp);
+		struct damon_target **targetp, unsigned long addr_unit);
diff --git a/mm/damon/reclaim.c b/mm/damon/reclaim.c
index 3c71b4596676..0e11b121d693 100644
--- a/mm/damon/reclaim.c
+++ b/mm/damon/reclaim.c
@@ -128,6 +128,14 @@ module_param(monitor_region_start, ulong, 0600);
 static unsigned long monitor_region_end __read_mostly;
 module_param(monitor_region_end, ulong, 0600);
 
+/*
+ * Scale factor for DAMON_RECLAIM to ops address conversion.
+ *
+ * This parameter is used to convert to the actual physical address.
+ */
+static unsigned long addr_unit __read_mostly = 1;
+module_param(addr_unit, ulong, 0600);
+
 /*
  * Skip anonymous pages reclamation.
  *
@@ -190,7 +198,8 @@ static int damon_reclaim_apply_parameters(void)
 	struct damos_filter *filter;
 	int err;
 
-	err = damon_modules_new_paddr_ctx_target(&param_ctx, &param_target);
+	err = damon_modules_new_paddr_ctx_target(&param_ctx, &param_target,
+			addr_unit);
 	if (err)
 		return err;
 
@@ -229,7 +238,8 @@ static int damon_reclaim_apply_parameters(void)
 
 	err = damon_set_region_biggest_system_ram_default(param_target,
 					&monitor_region_start,
-					&monitor_region_end);
+					&monitor_region_end,
+					addr_unit);
 	if (err)
 		goto out;
 	err = damon_commit_ctx(ctx, param_ctx);
@@ -327,7 +337,7 @@ MODULE_PARM_DESC(enabled,
 
 static int __init damon_reclaim_init(void)
 {
-	int err = damon_modules_new_paddr_ctx_target(&ctx, &target);
+	int err = damon_modules_new_paddr_ctx_target(&ctx, &target, 1);
 
 	if (err)
 		goto out;
diff --git a/mm/damon/stat.c b/mm/damon/stat.c
index 87bcd8866d4b..ae7377e7409f 100644
--- a/mm/damon/stat.c
+++ b/mm/damon/stat.c
@@ -181,7 +181,7 @@ static struct damon_ctx *damon_stat_build_ctx(void)
 	if (!target)
 		goto free_out;
 	damon_add_target(ctx, target);
-	if (damon_set_region_biggest_system_ram_default(target, &start, &end))
+	if (damon_set_region_biggest_system_ram_default(target, &start, &end, ctx->addr_unit))
 		goto free_out;
 	return ctx;
 free_out:
-- 
2.34.1

[RFC PATCH -next 12/16] mm/damon: add damon_ctx->min_region and damon_target->min_region

[Quanmin Yan]

Adopting addr_unit would make DAMON_MINREGION 'addr_unit * 4096'
bytes and cause data alignment issues[1].

Add damon_ctx->min_region to change DAMON_MIN_REGION from a global
macro value to per-context variable, let target inherit the min_region
from its associated ctx to avoid excessive passing of ctx.

[1] https://lore.kernel.org/all/527714dd-0e33-43ab-bbbd-d89670ba79e7@huawei.com

Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
---
 include/linux/damon.h |  7 ++++++-
 mm/damon/core.c       | 41 +++++++++++++++++++++++++++--------------
 2 files changed, 33 insertions(+), 15 deletions(-)

diff --git a/include/linux/damon.h b/include/linux/damon.h
index 1b7b4cf1a3c5..aa045dcb5b5d 100644
--- a/include/linux/damon.h
+++ b/include/linux/damon.h
@@ -88,6 +88,7 @@ struct damon_region {
 /**
  * struct damon_target - Represents a monitoring target.
  * @pid:		The PID of the virtual address space to monitor.
+ * @min_region:		Minimum Region Size.
  * @nr_regions:		Number of monitoring target regions of this target.
  * @regions_list:	Head of the monitoring target regions of this target.
  * @list:		List head for siblings.
@@ -95,10 +96,12 @@ struct damon_region {
  * Each monitoring context could have multiple targets.  For example, a context
  * for virtual memory address spaces could have multiple target processes.  The
  * @pid should be set for appropriate &struct damon_operations including the
- * virtual address spaces monitoring operations.
+ * virtual address spaces monitoring operations. The @min_region Keeps consistent
+ * with the associated monitoring context.
  */
 struct damon_target {
 	struct pid *pid;
+	unsigned long min_region;
 	unsigned int nr_regions;
 	struct list_head regions_list;
 	struct list_head list;
@@ -747,6 +750,7 @@ struct damon_attrs {
  *
  * @ops:	Set of monitoring operations for given use cases.
  * @addr_unit:	Scale factor for core to ops address conversion.
+ * @min_region:		Minimum Region Size.
  * @adaptive_targets:	Head of monitoring targets (&damon_target) list.
  * @schemes:		Head of schemes (&damos) list.
  */
@@ -789,6 +793,7 @@ struct damon_ctx {
 
 	struct damon_operations ops;
 	unsigned long addr_unit;
+	unsigned long min_region;
 
 	struct list_head adaptive_targets;
 	struct list_head schemes;
diff --git a/mm/damon/core.c b/mm/damon/core.c
index 803c30f64b94..b162aa1156fc 100644
--- a/mm/damon/core.c
+++ b/mm/damon/core.c
@@ -245,16 +245,16 @@ int damon_set_regions(struct damon_target *t, struct damon_addr_range *ranges,
 			/* no region intersects with this range */
 			newr = damon_new_region(
 					ALIGN_DOWN(range->start,
-						DAMON_MIN_REGION),
-					ALIGN(range->end, DAMON_MIN_REGION));
+						t->min_region),
+					ALIGN(range->end, t->min_region));
 			if (!newr)
 				return -ENOMEM;
 			damon_insert_region(newr, damon_prev_region(r), r, t);
 		} else {
 			/* resize intersecting regions to fit in this range */
 			first->ar.start = ALIGN_DOWN(range->start,
-					DAMON_MIN_REGION);
-			last->ar.end = ALIGN(range->end, DAMON_MIN_REGION);
+					t->min_region);
+			last->ar.end = ALIGN(range->end, t->min_region);
 
 			/* fill possible holes in the range */
 			err = damon_fill_regions_holes(first, last, t);
@@ -472,6 +472,7 @@ struct damon_target *damon_new_target(void)
 
 	t->pid = NULL;
 	t->nr_regions = 0;
+	t->min_region = DAMON_MIN_REGION;
 	INIT_LIST_HEAD(&t->regions_list);
 	INIT_LIST_HEAD(&t->list);
 
@@ -480,6 +481,7 @@ struct damon_target *damon_new_target(void)
 
 void damon_add_target(struct damon_ctx *ctx, struct damon_target *t)
 {
+	t->min_region = ctx->min_region;
 	list_add_tail(&t->list, &ctx->adaptive_targets);
 }
 
@@ -545,6 +547,7 @@ struct damon_ctx *damon_new_ctx(void)
 	ctx->attrs.max_nr_regions = 1000;
 
 	ctx->addr_unit = 1;
+	ctx->min_region = DAMON_MIN_REGION;
 
 	INIT_LIST_HEAD(&ctx->adaptive_targets);
 	INIT_LIST_HEAD(&ctx->schemes);
@@ -1181,6 +1184,14 @@ static int damon_commit_targets(
 	return 0;
 }
 
+static void damon_sync_target_min_region(struct damon_ctx *ctx)
+{
+	struct damon_target *t;
+
+	damon_for_each_target(t, ctx)
+		t->min_region = ctx->min_region;
+}
+
 /**
  * damon_commit_ctx() - Commit parameters of a DAMON context to another.
  * @dst:	The commit destination DAMON context.
@@ -1216,6 +1227,8 @@ int damon_commit_ctx(struct damon_ctx *dst, struct damon_ctx *src)
 		return err;
 	dst->ops = src->ops;
 	dst->addr_unit = src->addr_unit ? : 1;
+	dst->min_region = max(DAMON_MIN_REGION / dst->addr_unit, 1);
+	damon_sync_target_min_region(dst);
 
 	return 0;
 }
@@ -1248,8 +1261,8 @@ static unsigned long damon_region_sz_limit(struct damon_ctx *ctx)
 
 	if (ctx->attrs.min_nr_regions)
 		sz /= ctx->attrs.min_nr_regions;
-	if (sz < DAMON_MIN_REGION)
-		sz = DAMON_MIN_REGION;
+	if (sz < ctx->min_region)
+		sz = ctx->min_region;
 
 	return sz;
 }
@@ -1632,11 +1645,11 @@ static bool damos_skip_charged_region(struct damon_target *t,
 		if (quota->charge_addr_from && r->ar.start <
 				quota->charge_addr_from) {
 			sz_to_skip = ALIGN_DOWN(quota->charge_addr_from -
-					r->ar.start, DAMON_MIN_REGION);
+					r->ar.start, t->min_region);
 			if (!sz_to_skip) {
-				if (damon_sz_region(r) <= DAMON_MIN_REGION)
+				if (damon_sz_region(r) <= t->min_region)
 					return true;
-				sz_to_skip = DAMON_MIN_REGION;
+				sz_to_skip = t->min_region;
 			}
 			damon_split_region_at(t, r, sz_to_skip);
 			r = damon_next_region(r);
@@ -1678,8 +1691,8 @@ static bool damos_filter_match(struct damon_ctx *ctx, struct damon_target *t,
 		matched = target_idx == filter->target_idx;
 		break;
 	case DAMOS_FILTER_TYPE_ADDR:
-		start = ALIGN_DOWN(filter->addr_range.start, DAMON_MIN_REGION);
-		end = ALIGN_DOWN(filter->addr_range.end, DAMON_MIN_REGION);
+		start = ALIGN_DOWN(filter->addr_range.start, t->min_region);
+		end = ALIGN_DOWN(filter->addr_range.end, t->min_region);
 
 		/* inside the range */
 		if (start <= r->ar.start && r->ar.end <= end) {
@@ -1850,7 +1863,7 @@ static void damos_apply_scheme(struct damon_ctx *c, struct damon_target *t,
 	if (c->ops.apply_scheme) {
 		if (quota->esz && quota->charged_sz + sz > quota->esz) {
 			sz = ALIGN_DOWN(quota->esz - quota->charged_sz,
-					DAMON_MIN_REGION);
+					c->min_region);
 			if (!sz)
 				goto update_stat;
 			damon_split_region_at(t, r, sz);
@@ -2302,13 +2315,13 @@ static void damon_split_regions_of(struct damon_target *t, int nr_subs)
 		sz_region = damon_sz_region(r);
 
 		for (i = 0; i < nr_subs - 1 &&
-				sz_region > 2 * DAMON_MIN_REGION; i++) {
+				sz_region > 2 * t->min_region; i++) {
 			/*
 			 * Randomly select size of left sub-region to be at
 			 * least 10 percent and at most 90% of original region
 			 */
 			sz_sub = ALIGN_DOWN(damon_rand(1, 10) *
-					sz_region / 10, DAMON_MIN_REGION);
+					sz_region / 10, t->min_region);
 			/* Do not allow blank region */
 			if (sz_sub == 0 || sz_sub >= sz_region)
 				continue;
-- 
2.34.1

[RFC PATCH -next 13/16] mm/damon/sysfs: ensure valid addr_unit setting in damon_sysfs_apply_inputs()

[Quanmin Yan]

By calling damon_sysfs_turn_damon_on(), the execution of damon_commit_ctx()
can be bypassed. Therefore, it is necessary to prevent ctx->addr_unit from
being set to 0 in damon_sysfs_apply_inputs() and update min_region to avoid
potential issues.

Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
---
 mm/damon/sysfs.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/mm/damon/sysfs.c b/mm/damon/sysfs.c
index bea782b0a711..122824776c1d 100644
--- a/mm/damon/sysfs.c
+++ b/mm/damon/sysfs.c
@@ -1422,7 +1422,8 @@ static int damon_sysfs_apply_inputs(struct damon_ctx *ctx,
 	err = damon_select_ops(ctx, sys_ctx->ops_id);
 	if (err)
 		return err;
-	ctx->addr_unit = sys_ctx->addr_unit;
+	ctx->addr_unit = sys_ctx->addr_unit ? : 1;
+	ctx->min_region = max(DAMON_MIN_REGION / ctx->addr_unit, 1);
 	err = damon_sysfs_set_attrs(ctx, sys_ctx->attrs);
 	if (err)
 		return err;
-- 
2.34.1

[RFC PATCH -next 14/16] mm/damon/core: convert sz to byte units when updating state

[Quanmin Yan]

After introducing ctx->addr_unit, the unit of sz might not
be in bytes. However, sz_applied is returned in bytes after
processing by paddr. To maintain external consistency, sz
is converted to byte units when updating the state.

Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
---
 mm/damon/core.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/mm/damon/core.c b/mm/damon/core.c
index b162aa1156fc..bc764f9dc5c5 100644
--- a/mm/damon/core.c
+++ b/mm/damon/core.c
@@ -1889,7 +1889,9 @@ static void damos_apply_scheme(struct damon_ctx *c, struct damon_target *t,
 		r->age = 0;
 
 update_stat:
-	damos_update_stat(s, sz, sz_applied, sz_ops_filter_passed);
+	damos_update_stat(s,
+			sz * (c->ops.id == DAMON_OPS_PADDR ? c->addr_unit : 1),
+			sz_applied, sz_ops_filter_passed);
 }
 
 static void damon_do_apply_schemes(struct damon_ctx *c,
-- 
2.34.1

[RFC PATCH -next 15/16] mm/damon: the byte statistics data type in damos_stat uses unsigned long long

[Quanmin Yan]

For 32-bit systems, damos_stat now uses unsigned long long for byte
statistics data to avoid integer overflow risks inherent in the
previous design.

Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
---
 include/linux/damon.h     |  6 +++---
 mm/damon/modules-common.h |  4 ++--
 mm/damon/sysfs-schemes.c  | 12 ++++++------
 3 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/include/linux/damon.h b/include/linux/damon.h
index aa045dcb5b5d..d85850cf06c5 100644
--- a/include/linux/damon.h
+++ b/include/linux/damon.h
@@ -333,10 +333,10 @@ struct damos_watermarks {
  */
 struct damos_stat {
 	unsigned long nr_tried;
-	unsigned long sz_tried;
+	unsigned long long sz_tried;
 	unsigned long nr_applied;
-	unsigned long sz_applied;
-	unsigned long sz_ops_filter_passed;
+	unsigned long long sz_applied;
+	unsigned long long sz_ops_filter_passed;
 	unsigned long qt_exceeds;
 };
 
diff --git a/mm/damon/modules-common.h b/mm/damon/modules-common.h
index c7048a449321..ae45d0eb960e 100644
--- a/mm/damon/modules-common.h
+++ b/mm/damon/modules-common.h
@@ -36,11 +36,11 @@
 #define DEFINE_DAMON_MODULES_DAMOS_STATS_PARAMS(stat, try_name,		\
 		succ_name, qt_exceed_name)				\
 	module_param_named(nr_##try_name, stat.nr_tried, ulong, 0400);	\
-	module_param_named(bytes_##try_name, stat.sz_tried, ulong,	\
+	module_param_named(bytes_##try_name, stat.sz_tried, ullong,	\
 			0400);						\
 	module_param_named(nr_##succ_name, stat.nr_applied, ulong,	\
 			0400);						\
-	module_param_named(bytes_##succ_name, stat.sz_applied, ulong,	\
+	module_param_named(bytes_##succ_name, stat.sz_applied, ullong,	\
 			0400);						\
 	module_param_named(nr_##qt_exceed_name, stat.qt_exceeds, ulong,	\
 			0400);
diff --git a/mm/damon/sysfs-schemes.c b/mm/damon/sysfs-schemes.c
index 74056bcd6a2c..3c4882549a28 100644
--- a/mm/damon/sysfs-schemes.c
+++ b/mm/damon/sysfs-schemes.c
@@ -199,10 +199,10 @@ static const struct kobj_type damon_sysfs_scheme_regions_ktype = {
 struct damon_sysfs_stats {
 	struct kobject kobj;
 	unsigned long nr_tried;
-	unsigned long sz_tried;
+	unsigned long long sz_tried;
 	unsigned long nr_applied;
-	unsigned long sz_applied;
-	unsigned long sz_ops_filter_passed;
+	unsigned long long sz_applied;
+	unsigned long long sz_ops_filter_passed;
 	unsigned long qt_exceeds;
 };
 
@@ -226,7 +226,7 @@ static ssize_t sz_tried_show(struct kobject *kobj, struct kobj_attribute *attr,
 	struct damon_sysfs_stats *stats = container_of(kobj,
 			struct damon_sysfs_stats, kobj);
 
-	return sysfs_emit(buf, "%lu\n", stats->sz_tried);
+	return sysfs_emit(buf, "%llu\n", stats->sz_tried);
 }
 
 static ssize_t nr_applied_show(struct kobject *kobj,
@@ -244,7 +244,7 @@ static ssize_t sz_applied_show(struct kobject *kobj,
 	struct damon_sysfs_stats *stats = container_of(kobj,
 			struct damon_sysfs_stats, kobj);
 
-	return sysfs_emit(buf, "%lu\n", stats->sz_applied);
+	return sysfs_emit(buf, "%llu\n", stats->sz_applied);
 }
 
 static ssize_t sz_ops_filter_passed_show(struct kobject *kobj,
@@ -253,7 +253,7 @@ static ssize_t sz_ops_filter_passed_show(struct kobject *kobj,
 	struct damon_sysfs_stats *stats = container_of(kobj,
 			struct damon_sysfs_stats, kobj);
 
-	return sysfs_emit(buf, "%lu\n", stats->sz_ops_filter_passed);
+	return sysfs_emit(buf, "%llu\n", stats->sz_ops_filter_passed);
 }
 
 static ssize_t qt_exceeds_show(struct kobject *kobj,
-- 
2.34.1

[RFC PATCH -next 16/16] mm/damon/core: handle quota->esz overflow issues

[Quanmin Yan]

In the original quota enforcement implementation, the traffic
calculation multiplied A by 1000000 due to time unit conversion,
making it highly prone to overflow on 32-bit systems:

damos_set_effective_quota
  if (quota->total_charged_ns)
    throughput = quota->total_charged_sz * 1000000 /
		quota->total_charged_ns;

Requiring total_charged_sz to be less than 4GB/1000000 is unreasonable.
Additionally, when overflow occurs and causes quota->esz to become
extremely small, the subsequent damos_apply_scheme logic permanently
sets sz to 0, while quota stop updating, ultimately leading to complete
functional failure:

damos_apply_scheme
  if (quota->esz && quota->charged_sz + sz > quota->esz)
    sz = ALIGN_DOWN(quota->esz - quota->charged_sz, DAMON_MIN_REGION);

Total charged stats use the unsigned long long data type to reduce
overflow risk, with data reset capability after overflow occurs.

Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
---
 include/linux/damon.h |  4 ++--
 mm/damon/core.c       | 18 ++++++++++++------
 2 files changed, 14 insertions(+), 8 deletions(-)

diff --git a/include/linux/damon.h b/include/linux/damon.h
index d85850cf06c5..45aab331dfb7 100644
--- a/include/linux/damon.h
+++ b/include/linux/damon.h
@@ -247,8 +247,8 @@ struct damos_quota {
 
 /* private: */
 	/* For throughput estimation */
-	unsigned long total_charged_sz;
-	unsigned long total_charged_ns;
+	unsigned long long total_charged_sz;
+	unsigned long long total_charged_ns;
 
 	/* For charging the quota */
 	unsigned long charged_sz;
diff --git a/mm/damon/core.c b/mm/damon/core.c
index bc764f9dc5c5..5e05fdd91c12 100644
--- a/mm/damon/core.c
+++ b/mm/damon/core.c
@@ -15,6 +15,7 @@
 #include <linux/slab.h>
 #include <linux/string.h>
 #include <linux/string_choices.h>
+#include <linux/math64.h>
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/damon.h>
@@ -2059,8 +2060,8 @@ static unsigned long damos_quota_score(struct damos_quota *quota)
  */
 static void damos_set_effective_quota(struct damos_quota *quota)
 {
-	unsigned long throughput;
-	unsigned long esz = ULONG_MAX;
+	unsigned long long throughput;
+	unsigned long long esz = ULLONG_MAX;
 
 	if (!quota->ms && list_empty(&quota->goals)) {
 		quota->esz = quota->sz;
@@ -2077,11 +2078,16 @@ static void damos_set_effective_quota(struct damos_quota *quota)
 	}
 
 	if (quota->ms) {
-		if (quota->total_charged_ns)
-			throughput = quota->total_charged_sz * 1000000 /
-				quota->total_charged_ns;
-		else
+		if (quota->total_charged_ns &&
+			likely(quota->total_charged_sz < ULLONG_MAX / 1000000)) {
+			throughput = div64_u64(quota->total_charged_sz * 1000000,
+					quota->total_charged_ns);
+		} else {
 			throughput = PAGE_SIZE * 1024;
+			/* Reset the variable when an overflow occurs */
+			quota->total_charged_ns = 0;
+			quota->total_charged_sz = 0;
+		}
 		esz = min(throughput * quota->ms, esz);
 	}
 
-- 
2.34.1

Re: [RFC PATCH -next 11/16] mm/damon: add addr_unit for DAMON_RECLAIM and LRU_SORT

[SeongJae Park]

On Wed, 13 Aug 2025 13:07:01 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> In module DAMON_RECLAIM and DAMON_LRU_SORT, the damon_ctx is
> independent of the core, necessitating dedicated addr_unit
> integration for these features.
> Additionally, if the input monitor_region_start and monitor_region_end
> are both 0 while addr_unit is set to a non-zero valuethe default
> system RAM range should be divided by addr_unit.

Do you plan to, and need to use DAMON_RECLAIM and DAMON_LRU_SORT on LPAE-ARM32
environments?  Can't you use DAMON sysfs interface instead?  If need to use the
modules, this change looks good to me in high level.  But if not, I'd like to
skip this change, and wait until someone requests it.

I'll review the code change in depth after the above question is answered.


Thanks,
SJ

[...]

Re: [RFC PATCH -next 12/16] mm/damon: add damon_ctx->min_region and damon_target->min_region

[SeongJae Park]

On Wed, 13 Aug 2025 13:07:02 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> Adopting addr_unit would make DAMON_MINREGION 'addr_unit * 4096'
> bytes and cause data alignment issues[1].
> 
> Add damon_ctx->min_region to change DAMON_MIN_REGION from a global
> macro value to per-context variable,

Nice!

> let target inherit the min_region
> from its associated ctx to avoid excessive passing of ctx.

To me, keeping the synchronization seems more complicated than just passing the
addr_unit value to relevant functions.  I'd prefer passing the addr_unit to
relevant functions.  It could require more changes and might make code uglier,
but I think that's easier to maintain, and we can cleanup/refactor the code
later.

Could you please do so in the next version?


Thanks,
SJ

[...]

Re: [RFC PATCH -next 13/16] mm/damon/sysfs: ensure valid addr_unit setting in damon_sysfs_apply_inputs()

[SeongJae Park]

On Wed, 13 Aug 2025 13:07:03 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> By calling damon_sysfs_turn_damon_on(), the execution of damon_commit_ctx()
> can be bypassed. Therefore, it is necessary to prevent ctx->addr_unit from
> being set to 0 in damon_sysfs_apply_inputs() and update min_region to avoid
> potential issues.

Nice catch!

> 
> Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
> ---
>  mm/damon/sysfs.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/mm/damon/sysfs.c b/mm/damon/sysfs.c
> index bea782b0a711..122824776c1d 100644
> --- a/mm/damon/sysfs.c
> +++ b/mm/damon/sysfs.c
> @@ -1422,7 +1422,8 @@ static int damon_sysfs_apply_inputs(struct damon_ctx *ctx,
>  	err = damon_select_ops(ctx, sys_ctx->ops_id);
>  	if (err)
>  		return err;
> -	ctx->addr_unit = sys_ctx->addr_unit;
> +	ctx->addr_unit = sys_ctx->addr_unit ? : 1;

So this is fixing a bug of the seventh patch ("mm/damon/sysfs: implement
addr_unit file under context dir") of this series, right?  It is better to not
add a broken patch, and then fixing it in the same series.  Let's squash the
fix of the problem into the patch.  Don't forget adding your Signed-off-by on
the patch.

Also, since sys_ctx->addr_unit is initialized as 1, the value being zero is
user's wrong input.  Let's return -EINVAL instead of making it silently
success.

> +	ctx->min_region = max(DAMON_MIN_REGION / ctx->addr_unit, 1);

Seems this is a fix of an issue in the 12th patch ("mm/damon: add
damon_ctx->min_region and damon_target->min_region") of this series?  Let's fix
it on the patch.

>  	err = damon_sysfs_set_attrs(ctx, sys_ctx->attrs);
>  	if (err)
>  		return err;
> -- 
> 2.34.1


Thanks,
SJ

Re: [RFC PATCH -next 14/16] mm/damon/core: convert sz to byte units when updating state

[SeongJae Park]

On Wed, 13 Aug 2025 13:07:04 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> After introducing ctx->addr_unit, the unit of sz might not
> be in bytes. However, sz_applied is returned in bytes after
> processing by paddr.

This is not an intended behavior, but a bug of my code.  Let's update patches
3-6 of this series to return sz_applied as core-layer address unit, instead of
bytes.

> To maintain external consistency, sz
> is converted to byte units when updating the state.

Users could keep the consistency by multiplying the addr_unit, which they set
themselves.

> 
> Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
> ---
>  mm/damon/core.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/mm/damon/core.c b/mm/damon/core.c
> index b162aa1156fc..bc764f9dc5c5 100644
> --- a/mm/damon/core.c
> +++ b/mm/damon/core.c
> @@ -1889,7 +1889,9 @@ static void damos_apply_scheme(struct damon_ctx *c, struct damon_target *t,
>  		r->age = 0;
>  
>  update_stat:
> -	damos_update_stat(s, sz, sz_applied, sz_ops_filter_passed);
> +	damos_update_stat(s,
> +			sz * (c->ops.id == DAMON_OPS_PADDR ? c->addr_unit : 1),
> +			sz_applied, sz_ops_filter_passed);
>  }
>  
>  static void damon_do_apply_schemes(struct damon_ctx *c,
> -- 
> 2.34.1


Thanks,
SJ

Re: [RFC PATCH -next 15/16] mm/damon: the byte statistics data type in damos_stat uses unsigned long long

[SeongJae Park]

On Wed, 13 Aug 2025 13:07:05 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> For 32-bit systems, damos_stat now uses unsigned long long for byte
> statistics data to avoid integer overflow risks inherent in the
> previous design.

I suggested using the core-layer address unit on stat, and ask users to
multiply the addr_unit value to stat values if they want bytes value.  If we
agree on it, I think this patch wouldn't really be required.

> 
> Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
> ---
>  include/linux/damon.h     |  6 +++---
>  mm/damon/modules-common.h |  4 ++--
>  mm/damon/sysfs-schemes.c  | 12 ++++++------
>  3 files changed, 11 insertions(+), 11 deletions(-)
> 
> diff --git a/include/linux/damon.h b/include/linux/damon.h
> index aa045dcb5b5d..d85850cf06c5 100644
> --- a/include/linux/damon.h
> +++ b/include/linux/damon.h
> @@ -333,10 +333,10 @@ struct damos_watermarks {
>   */
>  struct damos_stat {
>  	unsigned long nr_tried;
> -	unsigned long sz_tried;
> +	unsigned long long sz_tried;
>  	unsigned long nr_applied;
> -	unsigned long sz_applied;
> -	unsigned long sz_ops_filter_passed;
> +	unsigned long long sz_applied;
> +	unsigned long long sz_ops_filter_passed;
>  	unsigned long qt_exceeds;
>  };
>  
> diff --git a/mm/damon/modules-common.h b/mm/damon/modules-common.h
> index c7048a449321..ae45d0eb960e 100644
> --- a/mm/damon/modules-common.h
> +++ b/mm/damon/modules-common.h
> @@ -36,11 +36,11 @@
>  #define DEFINE_DAMON_MODULES_DAMOS_STATS_PARAMS(stat, try_name,		\
>  		succ_name, qt_exceed_name)				\
>  	module_param_named(nr_##try_name, stat.nr_tried, ulong, 0400);	\
> -	module_param_named(bytes_##try_name, stat.sz_tried, ulong,	\
> +	module_param_named(bytes_##try_name, stat.sz_tried, ullong,	\
>  			0400);						\
>  	module_param_named(nr_##succ_name, stat.nr_applied, ulong,	\
>  			0400);						\
> -	module_param_named(bytes_##succ_name, stat.sz_applied, ulong,	\
> +	module_param_named(bytes_##succ_name, stat.sz_applied, ullong,	\
>  			0400);						\
>  	module_param_named(nr_##qt_exceed_name, stat.qt_exceeds, ulong,	\
>  			0400);
> diff --git a/mm/damon/sysfs-schemes.c b/mm/damon/sysfs-schemes.c
> index 74056bcd6a2c..3c4882549a28 100644
> --- a/mm/damon/sysfs-schemes.c
> +++ b/mm/damon/sysfs-schemes.c
> @@ -199,10 +199,10 @@ static const struct kobj_type damon_sysfs_scheme_regions_ktype = {
>  struct damon_sysfs_stats {
>  	struct kobject kobj;
>  	unsigned long nr_tried;
> -	unsigned long sz_tried;
> +	unsigned long long sz_tried;
>  	unsigned long nr_applied;
> -	unsigned long sz_applied;
> -	unsigned long sz_ops_filter_passed;
> +	unsigned long long sz_applied;
> +	unsigned long long sz_ops_filter_passed;
>  	unsigned long qt_exceeds;
>  };
>  
> @@ -226,7 +226,7 @@ static ssize_t sz_tried_show(struct kobject *kobj, struct kobj_attribute *attr,
>  	struct damon_sysfs_stats *stats = container_of(kobj,
>  			struct damon_sysfs_stats, kobj);
>  
> -	return sysfs_emit(buf, "%lu\n", stats->sz_tried);
> +	return sysfs_emit(buf, "%llu\n", stats->sz_tried);
>  }
>  
>  static ssize_t nr_applied_show(struct kobject *kobj,
> @@ -244,7 +244,7 @@ static ssize_t sz_applied_show(struct kobject *kobj,
>  	struct damon_sysfs_stats *stats = container_of(kobj,
>  			struct damon_sysfs_stats, kobj);
>  
> -	return sysfs_emit(buf, "%lu\n", stats->sz_applied);
> +	return sysfs_emit(buf, "%llu\n", stats->sz_applied);
>  }
>  
>  static ssize_t sz_ops_filter_passed_show(struct kobject *kobj,
> @@ -253,7 +253,7 @@ static ssize_t sz_ops_filter_passed_show(struct kobject *kobj,
>  	struct damon_sysfs_stats *stats = container_of(kobj,
>  			struct damon_sysfs_stats, kobj);
>  
> -	return sysfs_emit(buf, "%lu\n", stats->sz_ops_filter_passed);
> +	return sysfs_emit(buf, "%llu\n", stats->sz_ops_filter_passed);
>  }
>  
>  static ssize_t qt_exceeds_show(struct kobject *kobj,
> -- 
> 2.34.1

Re: [RFC PATCH -next 16/16] mm/damon/core: handle quota->esz overflow issues

[SeongJae Park]

On Wed, 13 Aug 2025 13:07:06 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> In the original quota enforcement implementation, the traffic
> calculation multiplied A by 1000000 due to time unit conversion,
> making it highly prone to overflow on 32-bit systems:
> 
> damos_set_effective_quota
>   if (quota->total_charged_ns)
>     throughput = quota->total_charged_sz * 1000000 /
> 		quota->total_charged_ns;
> 
> Requiring total_charged_sz to be less than 4GB/1000000 is unreasonable.
> Additionally, when overflow occurs and causes quota->esz to become
> extremely small, the subsequent damos_apply_scheme logic permanently
> sets sz to 0, while quota stop updating, ultimately leading to complete
> functional failure:
> 
> damos_apply_scheme
>   if (quota->esz && quota->charged_sz + sz > quota->esz)
>     sz = ALIGN_DOWN(quota->esz - quota->charged_sz, DAMON_MIN_REGION);
> 
> Total charged stats use the unsigned long long data type to reduce
> overflow risk, with data reset capability after overflow occurs.

Thank you for finding this issue!  I don't want to change the data type if
possible, though.  Could replacing the easily-overflowing throughput
calculation with mult_frac() fix the issue?


Thanks,
SJ

[...]

Re: [RFC PATCH -next 00/16] mm/damon: support ARM32 with LPAE

[SeongJae Park]

Hello Quanmin,

On Wed, 13 Aug 2025 13:06:50 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> Previously, DAMON's physical address space monitoring only supported
> memory ranges below 4GB on LPAE-enabled systems. This was due to
> the use of 'unsigned long' in 'struct damon_addr_range', which is
> 32-bit on ARM32 even with LPAE enabled.
> 
> Implements DAMON compatibility for ARM32 with LPAE enabled.

Thank you for working on this, Quanmin!

> 
> Patches 01/16 through 10/16 are from the mailing list[1], add a new core
> layer parameter called 'addr_unit'. Operations set layer can translate a
> core layer address to the real address by multiplying the parameter value
> to the core layer address.
> 
> Patches 11/16 through 14/16 extend and complement patches 01~10, addressing
> various issues introduced by the addr_unit implementation.
> 
> Patches 15/16 and 16/16 complete native DAMON support for 32-bit systems.

Overall, looks good to me.  I have a few change requests including below major
ones, though.

First, let's squash patches for fixing problems made with patches 1-10 into
patches 1-10.  If you don't mind, I will post RFC v2 of those so that you can
pick into your series.

Second, let's keep DAMOS stats in 'unsigned long' type.  This require fixups of
patches 1-10.  If you don't mind, I will also do this in RFC v2 of those.

Please let me know what do you think.

I left a few more comments to patches, let's discuss on the replies.


Thanks,
SJ

[...]

Re: [RFC PATCH -next 00/16] mm/damon: support ARM32 with LPAE

[SeongJae Park]

On Wed, 13 Aug 2025 13:06:50 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> Previously, DAMON's physical address space monitoring only supported
> memory ranges below 4GB on LPAE-enabled systems. This was due to
> the use of 'unsigned long' in 'struct damon_addr_range', which is
> 32-bit on ARM32 even with LPAE enabled.
> 
> Implements DAMON compatibility for ARM32 with LPAE enabled.
> 
> Patches 01/16 through 10/16 are from the mailing list[1], add a new core
> layer parameter called 'addr_unit'. Operations set layer can translate a
> core layer address to the real address by multiplying the parameter value
> to the core layer address.
> 
> Patches 11/16 through 14/16 extend and complement patches 01~10, addressing
> various issues introduced by the addr_unit implementation.
> 
> Patches 15/16 and 16/16 complete native DAMON support for 32-bit systems.
> 
> [1] https://lore.kernel.org/all/20250416042551.158131-1-sj@kernel.org/

It would be nice to add more history about this nice work here, with links.


Thanks,
SJ

[...]

Re: [RFC PATCH -next 00/16] mm/damon: support ARM32 with LPAE

[SeongJae Park]

On Wed, 13 Aug 2025 10:25:44 -0700 SeongJae Park <sj@kernel.org> wrote:

> Hello Quanmin,
> 
> On Wed, 13 Aug 2025 13:06:50 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
> 
> > Previously, DAMON's physical address space monitoring only supported
> > memory ranges below 4GB on LPAE-enabled systems. This was due to
> > the use of 'unsigned long' in 'struct damon_addr_range', which is
> > 32-bit on ARM32 even with LPAE enabled.
> > 
> > Implements DAMON compatibility for ARM32 with LPAE enabled.
> 
> Thank you for working on this, Quanmin!
> 
> > 
> > Patches 01/16 through 10/16 are from the mailing list[1], add a new core
> > layer parameter called 'addr_unit'. Operations set layer can translate a
> > core layer address to the real address by multiplying the parameter value
> > to the core layer address.
> > 
> > Patches 11/16 through 14/16 extend and complement patches 01~10, addressing
> > various issues introduced by the addr_unit implementation.
> > 
> > Patches 15/16 and 16/16 complete native DAMON support for 32-bit systems.
> 
> Overall, looks good to me.  I have a few change requests including below major
> ones, though.
> 
> First, let's squash patches for fixing problems made with patches 1-10 into
> patches 1-10.  If you don't mind, I will post RFC v2 of those so that you can
> pick into your series.
> 
> Second, let's keep DAMOS stats in 'unsigned long' type.  This require fixups of
> patches 1-10.  If you don't mind, I will also do this in RFC v2 of those.

Instead of posting completely new RFC v2 of the ten patches, I think posting
fixup patches as replies to this thread might be a better approach.  I will
make fixups first, see what looks easier for working together with you, and
either post entirely new version of the patch series, or send individual fixups
as replies to each patch of this thread.

And one more questions.  What is the baseline if this series?  I cannot simply
apply these patches on mm-unstable or mm-new.  It would be nice if you could
share a git tree having these patches fully applied, since 'cherry-pick' is
easier than 'am' for me.


Thanks,
SJ

[...]

Re: [RFC PATCH -next 11/16] mm/damon: add addr_unit for DAMON_RECLAIM and LRU_SORT

[Quanmin Yan]

在 2025/8/14 0:36, SeongJae Park 写道:
> On Wed, 13 Aug 2025 13:07:01 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>
>> In module DAMON_RECLAIM and DAMON_LRU_SORT, the damon_ctx is
>> independent of the core, necessitating dedicated addr_unit
>> integration for these features.
>> Additionally, if the input monitor_region_start and monitor_region_end
>> are both 0 while addr_unit is set to a non-zero valuethe default
>> system RAM range should be divided by addr_unit.
> Do you plan to, and need to use DAMON_RECLAIM and DAMON_LRU_SORT on LPAE-ARM32
> environments?  Can't you use DAMON sysfs interface instead?  If need to use the
> modules, this change looks good to me in high level.  But if not, I'd like to
> skip this change, and wait until someone requests it.
>
> I'll review the code change in depth after the above question is answered.
>
Hi SJ,

Yes, we need to use these modules in an LPAE-ARM32 environment. The modular
approach often provides more flexibility in our workflow, so we would greatly
appreciate it if you could take some time to review the code!🙂

Thanks,
Quanmin Yan

>
> Thanks,
> SJ
>
> [...]
>

Re: [RFC PATCH -next 00/16] mm/damon: support ARM32 with LPAE

[Quanmin Yan]

在 2025/8/14 8:57, SeongJae Park 写道:
> On Wed, 13 Aug 2025 10:25:44 -0700 SeongJae Park <sj@kernel.org> wrote:
>
>> Hello Quanmin,
>>
>> On Wed, 13 Aug 2025 13:06:50 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>>
>>> Previously, DAMON's physical address space monitoring only supported
>>> memory ranges below 4GB on LPAE-enabled systems. This was due to
>>> the use of 'unsigned long' in 'struct damon_addr_range', which is
>>> 32-bit on ARM32 even with LPAE enabled.
>>>
>>> Implements DAMON compatibility for ARM32 with LPAE enabled.
>> Thank you for working on this, Quanmin!
>>
>>> Patches 01/16 through 10/16 are from the mailing list[1], add a new core
>>> layer parameter called 'addr_unit'. Operations set layer can translate a
>>> core layer address to the real address by multiplying the parameter value
>>> to the core layer address.
>>>
>>> Patches 11/16 through 14/16 extend and complement patches 01~10, addressing
>>> various issues introduced by the addr_unit implementation.
>>>
>>> Patches 15/16 and 16/16 complete native DAMON support for 32-bit systems.
>> Overall, looks good to me.  I have a few change requests including below major
>> ones, though.
>>
>> First, let's squash patches for fixing problems made with patches 1-10 into
>> patches 1-10.  If you don't mind, I will post RFC v2 of those so that you can
>> pick into your series.
>>
>> Second, let's keep DAMOS stats in 'unsigned long' type.  This require fixups of
>> patches 1-10.  If you don't mind, I will also do this in RFC v2 of those.
> Instead of posting completely new RFC v2 of the ten patches, I think posting
> fixup patches as replies to this thread might be a better approach.  I will
> make fixups first, see what looks easier for working together with you, and
> either post entirely new version of the patch series, or send individual fixups
> as replies to each patch of this thread.
>
> And one more questions.  What is the baseline if this series?  I cannot simply
> apply these patches on mm-unstable or mm-new.  It would be nice if you could
> share a git tree having these patches fully applied, since 'cherry-pick' is
> easier than 'am' for me.

Hi SJ,

Thank you for your detailed suggestions on the patch series. Please allow me
some time to thoroughly review each of your recommendations. I haven’t responded
to every point immediately because I’d like to first attempt updating the patches
accordingly. If I encounter any questions or issues during the process, I’ll promptly
reach out to discuss them with you, very appreciate your patience and guidance.

By the way, this patch series is based on linux-next(commit:2674d1eadaa2).

Thanks,
Quanmin Yan

>
> Thanks,
> SJ
>
> [...]
>

Re: [RFC PATCH -next 00/16] mm/damon: support ARM32 with LPAE

[SeongJae Park]

On Thu, 14 Aug 2025 22:07:12 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> 
> 在 2025/8/14 8:57, SeongJae Park 写道:
> > On Wed, 13 Aug 2025 10:25:44 -0700 SeongJae Park <sj@kernel.org> wrote:
> >
> >> Hello Quanmin,
> >>
> >> On Wed, 13 Aug 2025 13:06:50 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
> >>
> >>> Previously, DAMON's physical address space monitoring only supported
> >>> memory ranges below 4GB on LPAE-enabled systems. This was due to
> >>> the use of 'unsigned long' in 'struct damon_addr_range', which is
> >>> 32-bit on ARM32 even with LPAE enabled.
> >>>
> >>> Implements DAMON compatibility for ARM32 with LPAE enabled.
> >> Thank you for working on this, Quanmin!
> >>
> >>> Patches 01/16 through 10/16 are from the mailing list[1], add a new core
> >>> layer parameter called 'addr_unit'. Operations set layer can translate a
> >>> core layer address to the real address by multiplying the parameter value
> >>> to the core layer address.
> >>>
> >>> Patches 11/16 through 14/16 extend and complement patches 01~10, addressing
> >>> various issues introduced by the addr_unit implementation.
> >>>
> >>> Patches 15/16 and 16/16 complete native DAMON support for 32-bit systems.
> >> Overall, looks good to me.  I have a few change requests including below major
> >> ones, though.
> >>
> >> First, let's squash patches for fixing problems made with patches 1-10 into
> >> patches 1-10.  If you don't mind, I will post RFC v2 of those so that you can
> >> pick into your series.
> >>
> >> Second, let's keep DAMOS stats in 'unsigned long' type.  This require fixups of
> >> patches 1-10.  If you don't mind, I will also do this in RFC v2 of those.
> > Instead of posting completely new RFC v2 of the ten patches, I think posting
> > fixup patches as replies to this thread might be a better approach.  I will
> > make fixups first, see what looks easier for working together with you, and
> > either post entirely new version of the patch series, or send individual fixups
> > as replies to each patch of this thread.
> >
> > And one more questions.  What is the baseline if this series?  I cannot simply
> > apply these patches on mm-unstable or mm-new.  It would be nice if you could
> > share a git tree having these patches fully applied, since 'cherry-pick' is
> > easier than 'am' for me.
> 
> Hi SJ,
> 
> Thank you for your detailed suggestions on the patch series. Please allow me
> some time to thoroughly review each of your recommendations.

No worry, please take your time :)

> I haven’t responded
> to every point immediately because I’d like to first attempt updating the patches
> accordingly. If I encounter any questions or issues during the process, I’ll promptly
> reach out to discuss them with you, very appreciate your patience and guidance.

Sounds good.

> 
> By the way, this patch series is based on linux-next(commit:2674d1eadaa2).

Thank you for sharing this.  From the next time, please use mm-new[1] as a
baseline for DAMON patches if there is no reason to not do so.

[1] https://origin.kernel.org/doc/html/latest/mm/damon/maintainer-profile.html#scm-trees


Thanks,
SJ

[...]

Re: [RFC PATCH -next 11/16] mm/damon: add addr_unit for DAMON_RECLAIM and LRU_SORT

[SeongJae Park]

On Thu, 14 Aug 2025 20:59:04 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> 
> 在 2025/8/14 0:36, SeongJae Park 写道:
> > On Wed, 13 Aug 2025 13:07:01 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
> >
> >> In module DAMON_RECLAIM and DAMON_LRU_SORT, the damon_ctx is
> >> independent of the core, necessitating dedicated addr_unit
> >> integration for these features.
> >> Additionally, if the input monitor_region_start and monitor_region_end
> >> are both 0 while addr_unit is set to a non-zero valuethe default
> >> system RAM range should be divided by addr_unit.
> > Do you plan to, and need to use DAMON_RECLAIM and DAMON_LRU_SORT on LPAE-ARM32
> > environments?  Can't you use DAMON sysfs interface instead?  If need to use the
> > modules, this change looks good to me in high level.  But if not, I'd like to
> > skip this change, and wait until someone requests it.
> >
> > I'll review the code change in depth after the above question is answered.
> >
> Hi SJ,
> 
> Yes, we need to use these modules in an LPAE-ARM32 environment. The modular
> approach often provides more flexibility in our workflow, so we would greatly
> appreciate it if you could take some time to review the code!🙂

Thank you for clarifying.  Ok, I understand this change is really required.

However, I think reviewing and revising this part may take time.  Meanwhile,
seems this part is not an essential one of this patch series, and has no
problem at be separated and merged after the essential parts.

So, could we separate this part from this patch series?  That is, let's work on
the essential part first.  After the work on the essential part is done, you
could post this part as another patch series, and then we can work together
again on it.


Thanks,
SJ

[...]

Re: [RFC PATCH -next 03/16] mm/damon/paddr: support addr_unit for DAMOS_PAGEOUT

[SeongJae Park]

Hi Quanmin,



On Wed, 13 Aug 2025 13:06:53 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> From: SeongJae Park <sj@kernel.org>
> 
> Add support of addr_unit for DAMOS_PAGEOUT action handling from the
> DAMOS operation implementation for the physical address space.

As I suggested on another reply[1], please squash attaching patch to this one
when you post next version of this series.

[1] https://lore.kernel.org/0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch

[...]


Thanks,
SJ

==== Attachment 0 (0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch) ====
From 525b3d1d9dd53f4ca6b4d2254e5cc7f99c8eae0d Mon Sep 17 00:00:00 2001
From: SeongJae Park <sj@kernel.org>
Date: Wed, 13 Aug 2025 21:24:35 -0700
Subject: [PATCH] mm/damon/paddr: set DAMOS_PAGEOUT stat in core address unit

Operations layer should set DAMOS stat in core layer address unit, but
paddr is returning that for PAGEOUT in paddr address unit.  Fix it.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index 98cb6930c183..826c2064dbfd 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -139,8 +139,7 @@ static unsigned long damon_pa_pageout(struct damon_region *r,
 		unsigned long addr_unit, struct damos *s,
 		unsigned long *sz_filter_passed)
 {
-	phys_addr_t addr;
-	unsigned long applied;
+	phys_addr_t addr, applied;
 	LIST_HEAD(folio_list);
 	bool install_young_filter = true;
 	struct damos_filter *filter;
@@ -172,7 +171,7 @@ static unsigned long damon_pa_pageout(struct damon_region *r,
 		if (damos_pa_filter_out(s, folio))
 			goto put_folio;
 		else
-			*sz_filter_passed += folio_size(folio);
+			*sz_filter_passed += folio_size(folio) / addr_unit;
 
 		folio_clear_referenced(folio);
 		folio_test_clear_young(folio);
@@ -191,7 +190,7 @@ static unsigned long damon_pa_pageout(struct damon_region *r,
 	applied = reclaim_pages(&folio_list);
 	cond_resched();
 	s->last_applied = folio;
-	return applied * PAGE_SIZE;
+	return applied * PAGE_SIZE / addr_unit;
 }
 
 static inline unsigned long damon_pa_mark_accessed_or_deactivate(
-- 
2.39.5

Re: [RFC PATCH -next 04/16] mm/damon/paddr: support addr_unit for DAMOS_LRU_[DE]PRIO

[SeongJae Park]

Hi Quanmin,

On Wed, 13 Aug 2025 13:06:54 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> From: SeongJae Park <sj@kernel.org>
> 
> Add support of addr_unit for DAMOS_LRU_PRIO and DAMOS_LRU_DEPRIO action
> handling from the DAMOS operation implementation for the physical
> address space.

As I suggested on another reply[1], please squash attaching patch to this one
when you post next version of this series.

[1] https://lore.kernel.org/0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch

[...]


Thanks,
SJ

==== Attachment 0 (0001-mm-damon-paddr-set-DAMOS_LRU_-DE-PRIO-stat-in-core-a.patch) ====
From a05d13d152443f35b7a85e0bd5ccecf294b699c0 Mon Sep 17 00:00:00 2001
From: SeongJae Park <sj@kernel.org>
Date: Wed, 13 Aug 2025 21:26:48 -0700
Subject: [PATCH] mm/damon/paddr: set DAMOS_LRU_[DE]PRIO stat in core address
 unit

Operations layer should set DAMOS stat in core layer address unit, but
paddr is returning that for LRU_[DE]PRIO in paddr address unit.  Fix it.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index 6263dd58a2e7..ed71dd0bf80e 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -198,8 +198,7 @@ static inline unsigned long damon_pa_mark_accessed_or_deactivate(
 		struct damos *s, bool mark_accessed,
 		unsigned long *sz_filter_passed)
 {
-	phys_addr_t addr;
-	unsigned long applied = 0;
+	phys_addr_t addr, applied = 0;
 	struct folio *folio;
 
 	addr = damon_pa_phys_addr(r->ar.start, addr_unit);
@@ -213,7 +212,7 @@ static inline unsigned long damon_pa_mark_accessed_or_deactivate(
 		if (damos_pa_filter_out(s, folio))
 			goto put_folio;
 		else
-			*sz_filter_passed += folio_size(folio);
+			*sz_filter_passed += folio_size(folio) / addr_unit;
 
 		if (mark_accessed)
 			folio_mark_accessed(folio);
@@ -225,7 +224,7 @@ static inline unsigned long damon_pa_mark_accessed_or_deactivate(
 		folio_put(folio);
 	}
 	s->last_applied = folio;
-	return applied * PAGE_SIZE;
+	return applied * PAGE_SIZE / addr_unit;
 }
 
 static unsigned long damon_pa_mark_accessed(struct damon_region *r,
-- 
2.39.5

Re: [RFC PATCH -next 05/16] mm/damon/paddr: support addr_unit for MIGRATE_{HOT,COLD}

[SeongJae Park]

Hi Quanmin,

On Wed, 13 Aug 2025 13:06:55 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> From: SeongJae Park <sj@kernel.org>
> 
> Add support of addr_unit for DAMOS_MIGRATE_HOT and DAMOS_MIGRATE_COLD
> action handling from the DAMOS operation implementation for the physical
> address space.

As I suggested on another reply[1], please squash attaching patch to this one
when you post next version of this series.

[1] https://lore.kernel.org/0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch

[...]


Thanks,
SJ

==== Attachment 0 (0001-mm-damon-paddr-set-DAMOS_MIGRATE_-HOT-COLD-stat-in-c.patch) ====
From f491a6a940444799dfe1e76433234d1849a21c58 Mon Sep 17 00:00:00 2001
From: SeongJae Park <sj@kernel.org>
Date: Wed, 13 Aug 2025 21:27:59 -0700
Subject: [PATCH] mm/damon/paddr: set DAMOS_MIGRATE_{HOT,COLD} stat in core
 address unit

Operations layer should set DAMOS stat in core layer address unit, but
paddr is returning that for MIGRATE_{HOT,COLD} in paddr address unit.
Fix it.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index bb3b149b75ba..0305e59818da 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -247,8 +247,7 @@ static unsigned long damon_pa_migrate(struct damon_region *r,
 		unsigned long addr_unit, struct damos *s,
 		unsigned long *sz_filter_passed)
 {
-	phys_addr_t addr;
-	unsigned long applied;
+	phys_addr_t addr, applied;
 	LIST_HEAD(folio_list);
 	struct folio *folio;
 
@@ -263,7 +262,7 @@ static unsigned long damon_pa_migrate(struct damon_region *r,
 		if (damos_pa_filter_out(s, folio))
 			goto put_folio;
 		else
-			*sz_filter_passed += folio_size(folio);
+			*sz_filter_passed += folio_size(folio) / addr_unit;
 
 		if (!folio_isolate_lru(folio))
 			goto put_folio;
@@ -275,7 +274,7 @@ static unsigned long damon_pa_migrate(struct damon_region *r,
 	applied = damon_migrate_pages(&folio_list, s->target_nid);
 	cond_resched();
 	s->last_applied = folio;
-	return applied * PAGE_SIZE;
+	return applied * PAGE_SIZE / addr_unit;
 }
 
 static unsigned long damon_pa_stat(struct damon_region *r, struct damos *s,
-- 
2.39.5

Re: [RFC PATCH -next 06/16] mm/damon/paddr: support addr_unit for DAMOS_STAT

[SeongJae Park]

Hi Quanmin,

On Wed, 13 Aug 2025 13:06:56 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> From: SeongJae Park <sj@kernel.org>
> 
> Add support of addr_unit for DAMOS_STAT action handling from the DAMOS
> operation implementation for the physical address space.

As I suggested on another reply[1], please squash attaching patch to this one
when you post next version of this series.

[1] https://lore.kernel.org/0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch

[...]


Thanks,
SJ

==== Attachment 0 (0001-mm-damon-paddr-set-DAMOS_STAT-stat-in-core-address-u.patch) ====
From 30d5bccbebe4ff5ae4d7d73ad857526f1648e786 Mon Sep 17 00:00:00 2001
From: SeongJae Park <sj@kernel.org>
Date: Wed, 13 Aug 2025 21:29:05 -0700
Subject: [PATCH] mm/damon/paddr: set DAMOS_STAT stat in core address unit

Operations layer should set DAMOS stat in core layer address unit, but
paddr is returning that for STAT in paddr address unit.  Fix it.

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/paddr.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
index beee3bfc503d..5fad2f9a99a0 100644
--- a/mm/damon/paddr.c
+++ b/mm/damon/paddr.c
@@ -296,7 +296,7 @@ static unsigned long damon_pa_stat(struct damon_region *r,
 		}
 
 		if (!damos_pa_filter_out(s, folio))
-			*sz_filter_passed += folio_size(folio);
+			*sz_filter_passed += folio_size(folio) / addr_unit;
 		addr += folio_size(folio);
 		folio_put(folio);
 	}
-- 
2.39.5

Re: [RFC PATCH -next 07/16] mm/damon/sysfs: implement addr_unit file under context dir

[SeongJae Park]

Hi Quanmin,

On Wed, 13 Aug 2025 13:06:57 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> From: SeongJae Park <sj@kernel.org>
> 
> Only DAMON kernel API callers can use addr_unit parameter.  Implement a
> sysfs file to let DAMON sysfs ABI users use it.

As I suggested on another reply[1], please squash attaching patch to this one
when you post next version of this series.

[1] https://lore.kernel.org/20250813170224.6128-1-sj@kernel.org

[...]


Thanks,
SJ

==== Attachment 0 (0001-mm-damon-sysfs-return-EINVAL-for-zero-addr_unit.patch) ====
From e0a5aa5e571ecd0f58b0914f8fc8562a60014ae8 Mon Sep 17 00:00:00 2001
From: SeongJae Park <sj@kernel.org>
Date: Wed, 13 Aug 2025 21:17:03 -0700
Subject: [PATCH] mm/damon/sysfs: return -EINVAL for zero addr_unit

Signed-off-by: SeongJae Park <sj@kernel.org>
---
 mm/damon/sysfs.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/mm/damon/sysfs.c b/mm/damon/sysfs.c
index bea782b0a711..eb4269383bae 100644
--- a/mm/damon/sysfs.c
+++ b/mm/damon/sysfs.c
@@ -1422,6 +1422,8 @@ static int damon_sysfs_apply_inputs(struct damon_ctx *ctx,
 	err = damon_select_ops(ctx, sys_ctx->ops_id);
 	if (err)
 		return err;
+	if (!sys_ctx->addr_unit)
+		return -EINVAL;
 	ctx->addr_unit = sys_ctx->addr_unit;
 	err = damon_sysfs_set_attrs(ctx, sys_ctx->attrs);
 	if (err)
-- 
2.39.5

Re: [RFC PATCH -next 03/16] mm/damon/paddr: support addr_unit for DAMOS_PAGEOUT

[SeongJae Park]

On Mon, 18 Aug 2025 23:18:26 -0700 SeongJae Park <sj@kernel.org> wrote:

> Hi Quanmin,
> 
> 
> 
> On Wed, 13 Aug 2025 13:06:53 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
> 
> > From: SeongJae Park <sj@kernel.org>
> > 
> > Add support of addr_unit for DAMOS_PAGEOUT action handling from the
> > DAMOS operation implementation for the physical address space.
> 
> As I suggested on another reply[1], please squash attaching patch to this one
> when you post next version of this series.
> 
> [1] https://lore.kernel.org/0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch

Sorry for the above wrong link.  Plese use
https://lore.kernel.org/20250813170806.6251-1-sj@kernel.org instead.

> 
> [...]
> 
> 
> Thanks,
> SJ
> 
> ==== Attachment 0 (0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch) ====
> >From 525b3d1d9dd53f4ca6b4d2254e5cc7f99c8eae0d Mon Sep 17 00:00:00 2001
> From: SeongJae Park <sj@kernel.org>
> Date: Wed, 13 Aug 2025 21:24:35 -0700
> Subject: [PATCH] mm/damon/paddr: set DAMOS_PAGEOUT stat in core address unit
> 
> Operations layer should set DAMOS stat in core layer address unit, but
> paddr is returning that for PAGEOUT in paddr address unit.  Fix it.
> 
> Signed-off-by: SeongJae Park <sj@kernel.org>
> ---
>  mm/damon/paddr.c | 7 +++----
>  1 file changed, 3 insertions(+), 4 deletions(-)
> 
> diff --git a/mm/damon/paddr.c b/mm/damon/paddr.c
> index 98cb6930c183..826c2064dbfd 100644
> --- a/mm/damon/paddr.c
> +++ b/mm/damon/paddr.c
> @@ -139,8 +139,7 @@ static unsigned long damon_pa_pageout(struct damon_region *r,
>  		unsigned long addr_unit, struct damos *s,
>  		unsigned long *sz_filter_passed)
>  {
> -	phys_addr_t addr;
> -	unsigned long applied;
> +	phys_addr_t addr, applied;
>  	LIST_HEAD(folio_list);
>  	bool install_young_filter = true;
>  	struct damos_filter *filter;
> @@ -172,7 +171,7 @@ static unsigned long damon_pa_pageout(struct damon_region *r,
>  		if (damos_pa_filter_out(s, folio))
>  			goto put_folio;
>  		else
> -			*sz_filter_passed += folio_size(folio);
> +			*sz_filter_passed += folio_size(folio) / addr_unit;
>  
>  		folio_clear_referenced(folio);
>  		folio_test_clear_young(folio);
> @@ -191,7 +190,7 @@ static unsigned long damon_pa_pageout(struct damon_region *r,
>  	applied = reclaim_pages(&folio_list);
>  	cond_resched();
>  	s->last_applied = folio;
> -	return applied * PAGE_SIZE;
> +	return applied * PAGE_SIZE / addr_unit;
>  }
>  
>  static inline unsigned long damon_pa_mark_accessed_or_deactivate(
> -- 
> 2.39.5
> 
> 

Re: [RFC PATCH -next 04/16] mm/damon/paddr: support addr_unit for DAMOS_LRU_[DE]PRIO

[SeongJae Park]

On Mon, 18 Aug 2025 23:19:52 -0700 SeongJae Park <sj@kernel.org> wrote:

> Hi Quanmin,
> 
> On Wed, 13 Aug 2025 13:06:54 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
> 
> > From: SeongJae Park <sj@kernel.org>
> > 
> > Add support of addr_unit for DAMOS_LRU_PRIO and DAMOS_LRU_DEPRIO action
> > handling from the DAMOS operation implementation for the physical
> > address space.
> 
> As I suggested on another reply[1], please squash attaching patch to this one
> when you post next version of this series.
> 
> [1] https://lore.kernel.org/0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch

Sorry for the above wrong link.  Plese use
https://lore.kernel.org/20250813170806.6251-1-sj@kernel.org instead.

Thanks,
SJ

[...]

Re: [RFC PATCH -next 05/16] mm/damon/paddr: support addr_unit for MIGRATE_{HOT,COLD}

[SeongJae Park]

On Mon, 18 Aug 2025 23:21:16 -0700 SeongJae Park <sj@kernel.org> wrote:

> Hi Quanmin,
> 
> On Wed, 13 Aug 2025 13:06:55 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
> 
> > From: SeongJae Park <sj@kernel.org>
> > 
> > Add support of addr_unit for DAMOS_MIGRATE_HOT and DAMOS_MIGRATE_COLD
> > action handling from the DAMOS operation implementation for the physical
> > address space.
> 
> As I suggested on another reply[1], please squash attaching patch to this one
> when you post next version of this series.
> 
> [1] https://lore.kernel.org/0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch

Sorry for the above wrong link.  Plese use
https://lore.kernel.org/20250813170806.6251-1-sj@kernel.org instead.


Thanks,
SJ

[...]

Re: [RFC PATCH -next 06/16] mm/damon/paddr: support addr_unit for DAMOS_STAT

[SeongJae Park]

On Mon, 18 Aug 2025 23:22:17 -0700 SeongJae Park <sj@kernel.org> wrote:

> Hi Quanmin,
> 
> On Wed, 13 Aug 2025 13:06:56 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
> 
> > From: SeongJae Park <sj@kernel.org>
> > 
> > Add support of addr_unit for DAMOS_STAT action handling from the DAMOS
> > operation implementation for the physical address space.
> 
> As I suggested on another reply[1], please squash attaching patch to this one
> when you post next version of this series.
> 
> [1] https://lore.kernel.org/0001-mm-damon-paddr-set-DAMOS_PAGEOUT-stat-in-core-addres.patch

Sorry for the above wrong link.  Plese use
https://lore.kernel.org/20250813170806.6251-1-sj@kernel.org instead.


Thanks,
SJ

[...]

Re: [RFC PATCH -next 03/16] mm/damon/paddr: support addr_unit for DAMOS_PAGEOUT

[Quanmin Yan]

Hi SJ,

在 2025/8/19 14:26, SeongJae Park 写道:
> Sorry for the above wrong link.  Plese use
> https://lore.kernel.org/20250813170806.6251-1-sj@kernel.org instead.

Okay, I've merged the corresponding changes into the original patches. So now we have v2 versions
for patches 3, 4, 5, and 6. I'll send out the patches later. That understanding is correct, right?

Thanks,
Quanmin Yan

Re: [RFC PATCH -next 07/16] mm/damon/sysfs: implement addr_unit file under context dir

[Quanmin Yan]

Hi SJ,

在 2025/8/19 14:24, SeongJae Park 写道:
> Hi Quanmin,
>
> As I suggested on another reply[1], please squash attaching patch to this one
> when you post next version of this series.
>
> [1] https://lore.kernel.org/20250813170224.6128-1-sj@kernel.org
>
> [...]
>
>
> Thanks,
> SJ
>
> ==== Attachment 0 (0001-mm-damon-sysfs-return-EINVAL-for-zero-addr_unit.patch) ====
>  From e0a5aa5e571ecd0f58b0914f8fc8562a60014ae8 Mon Sep 17 00:00:00 2001
> From: SeongJae Park <sj@kernel.org>
> Date: Wed, 13 Aug 2025 21:17:03 -0700
> Subject: [PATCH] mm/damon/sysfs: return -EINVAL for zero addr_unit
>
> Signed-off-by: SeongJae Park <sj@kernel.org>
> ---
>   mm/damon/sysfs.c | 2 ++
>   1 file changed, 2 insertions(+)
>
> diff --git a/mm/damon/sysfs.c b/mm/damon/sysfs.c
> index bea782b0a711..eb4269383bae 100644
> --- a/mm/damon/sysfs.c
> +++ b/mm/damon/sysfs.c
> @@ -1422,6 +1422,8 @@ static int damon_sysfs_apply_inputs(struct damon_ctx *ctx,
>   	err = damon_select_ops(ctx, sys_ctx->ops_id);
>   	if (err)
>   		return err;
> +	if (!sys_ctx->addr_unit)
> +		return -EINVAL;
>   	ctx->addr_unit = sys_ctx->addr_unit;
>   	err = damon_sysfs_set_attrs(ctx, sys_ctx->attrs);
>   	if (err)

Aha, returning -EINVAL when sys_ctx->addr_unit is 0 makes sense, but I wonder if it
might be better to prevent users from inputting 0 at the source instead? I've attempted
to modify patch 7 by adding a check in addr_unit_store. I'll send out the v2 version
of patch 7 later (PS: I am performing a comprehensive validation of the v2 patch series),
and we can discuss it then.

Thanks,
Quanmin Yan

Re: [RFC PATCH -next 12/16] mm/damon: add damon_ctx->min_region and damon_target->min_region

[Quanmin Yan]

Hi, SJ

在 2025/8/14 0:49, SeongJae Park 写道:
>> To me, keeping the synchronization seems more complicated than just passing the
>> addr_unit value to relevant functions.  I'd prefer passing the addr_unit to
>> relevant functions.  It could require more changes and might make code uglier,
>> but I think that's easier to maintain, and we can cleanup/refactor the code
>> later.
>>
>> Could you please do so in the next version?

Yes, I've revised patch 12 based on your suggestions and will send out the v2 version shortly.

Thanks,
Quanmin Yan

Re: [RFC PATCH -next 11/16] mm/damon: add addr_unit for DAMON_RECLAIM and LRU_SORT

[Quanmin Yan]

Hi SJ,

在 2025/8/15 0:11, SeongJae Park 写道:
> On Thu, 14 Aug 2025 20:59:04 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>
>> 在 2025/8/14 0:36, SeongJae Park 写道:
>>> On Wed, 13 Aug 2025 13:07:01 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>>>
>>>> In module DAMON_RECLAIM and DAMON_LRU_SORT, the damon_ctx is
>>>> independent of the core, necessitating dedicated addr_unit
>>>> integration for these features.
>>>> Additionally, if the input monitor_region_start and monitor_region_end
>>>> are both 0 while addr_unit is set to a non-zero valuethe default
>>>> system RAM range should be divided by addr_unit.
>>> Do you plan to, and need to use DAMON_RECLAIM and DAMON_LRU_SORT on LPAE-ARM32
>>> environments?  Can't you use DAMON sysfs interface instead?  If need to use the
>>> modules, this change looks good to me in high level.  But if not, I'd like to
>>> skip this change, and wait until someone requests it.
>>>
>>> I'll review the code change in depth after the above question is answered.
>>>
>> Hi SJ,
>>
>> Yes, we need to use these modules in an LPAE-ARM32 environment. The modular
>> approach often provides more flexibility in our workflow, so we would greatly
>> appreciate it if you could take some time to review the code!🙂
> Thank you for clarifying.  Ok, I understand this change is really required.
>
> However, I think reviewing and revising this part may take time.  Meanwhile,
> seems this part is not an essential one of this patch series, and has no
> problem at be separated and merged after the essential parts.
>
> So, could we separate this part from this patch series?  That is, let's work on
> the essential part first.  After the work on the essential part is done, you
> could post this part as another patch series, and then we can work together
> again on it.

You are right, let's focus on the essential part first.

Thanks,
Quanmin Yan

>
> Thanks,
> SJ
>
> [...]

Re: [RFC PATCH -next 03/16] mm/damon/paddr: support addr_unit for DAMOS_PAGEOUT

[SeongJae Park]

On Tue, 19 Aug 2025 22:18:06 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> Hi SJ,
> 
> 在 2025/8/19 14:26, SeongJae Park 写道:
> > Sorry for the above wrong link.  Plese use
> > https://lore.kernel.org/20250813170806.6251-1-sj@kernel.org instead.
> 
> Okay, I've merged the corresponding changes into the original patches. So now we have v2 versions
> for patches 3, 4, 5, and 6. I'll send out the patches later. That understanding is correct, right?

You're right :)


Thanks,
SJ

[...]

Re: [RFC PATCH -next 07/16] mm/damon/sysfs: implement addr_unit file under context dir

[SeongJae Park]

On Tue, 19 Aug 2025 22:45:58 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> Hi SJ,
> 
> 在 2025/8/19 14:24, SeongJae Park 写道:
> > Hi Quanmin,
> >
> > As I suggested on another reply[1], please squash attaching patch to this one
> > when you post next version of this series.
> >
> > [1] https://lore.kernel.org/20250813170224.6128-1-sj@kernel.org
> >
> > [...]
> >
> >
> > Thanks,
> > SJ
> >
> > ==== Attachment 0 (0001-mm-damon-sysfs-return-EINVAL-for-zero-addr_unit.patch) ====
> >  From e0a5aa5e571ecd0f58b0914f8fc8562a60014ae8 Mon Sep 17 00:00:00 2001
> > From: SeongJae Park <sj@kernel.org>
> > Date: Wed, 13 Aug 2025 21:17:03 -0700
> > Subject: [PATCH] mm/damon/sysfs: return -EINVAL for zero addr_unit
> >
> > Signed-off-by: SeongJae Park <sj@kernel.org>
> > ---
> >   mm/damon/sysfs.c | 2 ++
> >   1 file changed, 2 insertions(+)
> >
> > diff --git a/mm/damon/sysfs.c b/mm/damon/sysfs.c
> > index bea782b0a711..eb4269383bae 100644
> > --- a/mm/damon/sysfs.c
> > +++ b/mm/damon/sysfs.c
> > @@ -1422,6 +1422,8 @@ static int damon_sysfs_apply_inputs(struct damon_ctx *ctx,
> >   	err = damon_select_ops(ctx, sys_ctx->ops_id);
> >   	if (err)
> >   		return err;
> > +	if (!sys_ctx->addr_unit)
> > +		return -EINVAL;
> >   	ctx->addr_unit = sys_ctx->addr_unit;
> >   	err = damon_sysfs_set_attrs(ctx, sys_ctx->attrs);
> >   	if (err)
> 
> Aha, returning -EINVAL when sys_ctx->addr_unit is 0 makes sense, but I wonder if it
> might be better to prevent users from inputting 0 at the source instead? I've attempted
> to modify patch 7 by adding a check in addr_unit_store.

I agree that is better.  Please don't forget adding your Signed-off-by: to the
patch.

> I'll send out the v2 version
> of patch 7 later (PS: I am performing a comprehensive validation of the v2 patch series),
> and we can discuss it then.

Looking forwrd to the patch! :)


Thanks,
SJ

[...]

Re: [RFC PATCH -next 13/16] mm/damon/sysfs: ensure valid addr_unit setting in damon_sysfs_apply_inputs()

[Quanmin Yan]

Hi SJ,

在 2025/8/14 1:02, SeongJae Park 写道:
> On Wed, 13 Aug 2025 13:07:03 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>
>> By calling damon_sysfs_turn_damon_on(), the execution of damon_commit_ctx()
>> can be bypassed. Therefore, it is necessary to prevent ctx->addr_unit from
>> being set to 0 in damon_sysfs_apply_inputs() and update min_region to avoid
>> potential issues.
> Nice catch!
>
>> Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
>> ---
>>   mm/damon/sysfs.c | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/mm/damon/sysfs.c b/mm/damon/sysfs.c
>> index bea782b0a711..122824776c1d 100644
>> --- a/mm/damon/sysfs.c
>> +++ b/mm/damon/sysfs.c
>> @@ -1422,7 +1422,8 @@ static int damon_sysfs_apply_inputs(struct damon_ctx *ctx,
>>   	err = damon_select_ops(ctx, sys_ctx->ops_id);
>>   	if (err)
>>   		return err;
>> -	ctx->addr_unit = sys_ctx->addr_unit;
>> +	ctx->addr_unit = sys_ctx->addr_unit ? : 1;
> So this is fixing a bug of the seventh patch ("mm/damon/sysfs: implement
> addr_unit file under context dir") of this series, right?  It is better to not
> add a broken patch, and then fixing it in the same series.  Let's squash the
> fix of the problem into the patch.  Don't forget adding your Signed-off-by on
> the patch.
>
> Also, since sys_ctx->addr_unit is initialized as 1, the value being zero is
> user's wrong input.  Let's return -EINVAL instead of making it silently
> success.

Thank you for the kind reminder! The relevant changes have been integrated into
patch #7 of the v2 series[1]. It's worth noting that we've already prevented users
from inputting 0 at the source, therefore "ctx->addr_unit = sys_ctx->addr_unit;"
has been retained in the v2 version.

[1] https://lore.kernel.org/all/20250820080623.3799131-8-yanquanmin1@huawei.com/

>> +	ctx->min_region = max(DAMON_MIN_REGION / ctx->addr_unit, 1);
> Seems this is a fix of an issue in the 12th patch ("mm/damon: add
> damon_ctx->min_region and damon_target->min_region") of this series?  Let's fix
> it on the patch.

This set of changes has been integrated into patch #11 of the v2 series[2].

[2] https://lore.kernel.org/all/20250820080623.3799131-12-yanquanmin1@huawei.com/

Thanks,
Quanmin Yan

Re: [RFC PATCH -next 15/16] mm/damon: the byte statistics data type in damos_stat uses unsigned long long

[Quanmin Yan]

Hi SJ,

在 2025/8/14 1:10, SeongJae Park 写道:
> On Wed, 13 Aug 2025 13:07:05 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>
>> For 32-bit systems, damos_stat now uses unsigned long long for byte
>> statistics data to avoid integer overflow risks inherent in the
>> previous design.
> I suggested using the core-layer address unit on stat, and ask users to
> multiply the addr_unit value to stat values if they want bytes value.  If we
> agree on it, I think this patch wouldn't really be required.

Thank you for the guidance, I agree with your perspective. However, this patch doesn't actually belong in the addr_unit series, my apologies
for the confusion. It is actually intended to address potential overflow issues
in statistical data on 32-bit systems, and it is not directly related to addr_unit.
This patch has been dropped from the v2 series.

After introducing addr_unit, if it is set to a larger value, it can help mitigate
the overflow issue. However, under the default setting of addr_unit=1, statistical
data may still overflow after a sufficiently long runtime, for example, when sz_tried
exceeds 4GB.

Besides, please allow me to mention one point in advance: if addr is extended for
use in modules(e.g. DAMON_RECLAIM, LRU_SORT) in the future, the term "bytes" in
module_param_named(bytes_##try_name...), although multiplied by addr would yield
the actual byte count, might cause confusion due to its seemingly direct naming.

Overall, this patch isn’t critically important at the moment, nor does it offer a
sufficiently robust solution, but I’d still appreciate hearing your perspective on
the matter — I’m all ears.

Thanks,
Quanmin Yan

>> Signed-off-by: Quanmin Yan <yanquanmin1@huawei.com>
>> ---
>>   include/linux/damon.h     |  6 +++---
>>   mm/damon/modules-common.h |  4 ++--
>>   mm/damon/sysfs-schemes.c  | 12 ++++++------
>>   3 files changed, 11 insertions(+), 11 deletions(-)
>>
>> diff --git a/include/linux/damon.h b/include/linux/damon.h
>> index aa045dcb5b5d..d85850cf06c5 100644
>> --- a/include/linux/damon.h
>> +++ b/include/linux/damon.h
>> @@ -333,10 +333,10 @@ struct damos_watermarks {
>>    */
>>   struct damos_stat {
>>   	unsigned long nr_tried;
>> -	unsigned long sz_tried;
>> +	unsigned long long sz_tried;
>>   	unsigned long nr_applied;
>> -	unsigned long sz_applied;
>> -	unsigned long sz_ops_filter_passed;
>> +	unsigned long long sz_applied;
>> +	unsigned long long sz_ops_filter_passed;
>>   	unsigned long qt_exceeds;
>>   };
>>   
>> diff --git a/mm/damon/modules-common.h b/mm/damon/modules-common.h
>> index c7048a449321..ae45d0eb960e 100644
>> --- a/mm/damon/modules-common.h
>> +++ b/mm/damon/modules-common.h
>> @@ -36,11 +36,11 @@
>>   #define DEFINE_DAMON_MODULES_DAMOS_STATS_PARAMS(stat, try_name,		\
>>   		succ_name, qt_exceed_name)				\
>>   	module_param_named(nr_##try_name, stat.nr_tried, ulong, 0400);	\
>> -	module_param_named(bytes_##try_name, stat.sz_tried, ulong,	\
>> +	module_param_named(bytes_##try_name, stat.sz_tried, ullong,	\
>>   			0400);						\
>>   	module_param_named(nr_##succ_name, stat.nr_applied, ulong,	\
>>   			0400);						\
>> -	module_param_named(bytes_##succ_name, stat.sz_applied, ulong,	\
>> +	module_param_named(bytes_##succ_name, stat.sz_applied, ullong,	\
>>   			0400);						\
>>   	module_param_named(nr_##qt_exceed_name, stat.qt_exceeds, ulong,	\
>>   			0400);
>> diff --git a/mm/damon/sysfs-schemes.c b/mm/damon/sysfs-schemes.c
>> index 74056bcd6a2c..3c4882549a28 100644
>> --- a/mm/damon/sysfs-schemes.c
>> +++ b/mm/damon/sysfs-schemes.c
>> @@ -199,10 +199,10 @@ static const struct kobj_type damon_sysfs_scheme_regions_ktype = {
>>   struct damon_sysfs_stats {
>>   	struct kobject kobj;
>>   	unsigned long nr_tried;
>> -	unsigned long sz_tried;
>> +	unsigned long long sz_tried;
>>   	unsigned long nr_applied;
>> -	unsigned long sz_applied;
>> -	unsigned long sz_ops_filter_passed;
>> +	unsigned long long sz_applied;
>> +	unsigned long long sz_ops_filter_passed;
>>   	unsigned long qt_exceeds;
>>   };
>>   
>> @@ -226,7 +226,7 @@ static ssize_t sz_tried_show(struct kobject *kobj, struct kobj_attribute *attr,
>>   	struct damon_sysfs_stats *stats = container_of(kobj,
>>   			struct damon_sysfs_stats, kobj);
>>   
>> -	return sysfs_emit(buf, "%lu\n", stats->sz_tried);
>> +	return sysfs_emit(buf, "%llu\n", stats->sz_tried);
>>   }
>>   
>>   static ssize_t nr_applied_show(struct kobject *kobj,
>> @@ -244,7 +244,7 @@ static ssize_t sz_applied_show(struct kobject *kobj,
>>   	struct damon_sysfs_stats *stats = container_of(kobj,
>>   			struct damon_sysfs_stats, kobj);
>>   
>> -	return sysfs_emit(buf, "%lu\n", stats->sz_applied);
>> +	return sysfs_emit(buf, "%llu\n", stats->sz_applied);
>>   }
>>   
>>   static ssize_t sz_ops_filter_passed_show(struct kobject *kobj,
>> @@ -253,7 +253,7 @@ static ssize_t sz_ops_filter_passed_show(struct kobject *kobj,
>>   	struct damon_sysfs_stats *stats = container_of(kobj,
>>   			struct damon_sysfs_stats, kobj);
>>   
>> -	return sysfs_emit(buf, "%lu\n", stats->sz_ops_filter_passed);
>> +	return sysfs_emit(buf, "%llu\n", stats->sz_ops_filter_passed);
>>   }
>>   
>>   static ssize_t qt_exceeds_show(struct kobject *kobj,
>> -- 
>> 2.34.1

Re: [RFC PATCH -next 16/16] mm/damon/core: handle quota->esz overflow issues

[Quanmin Yan]

Hi SJ,

在 2025/8/14 1:15, SeongJae Park 写道:
> On Wed, 13 Aug 2025 13:07:06 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>
>> In the original quota enforcement implementation, the traffic
>> calculation multiplied A by 1000000 due to time unit conversion,
>> making it highly prone to overflow on 32-bit systems:
>>
>> damos_set_effective_quota
>>    if (quota->total_charged_ns)
>>      throughput = quota->total_charged_sz * 1000000 /
>> 		quota->total_charged_ns;
>>
>> Requiring total_charged_sz to be less than 4GB/1000000 is unreasonable.
>> Additionally, when overflow occurs and causes quota->esz to become
>> extremely small, the subsequent damos_apply_scheme logic permanently
>> sets sz to 0, while quota stop updating, ultimately leading to complete
>> functional failure:
>>
>> damos_apply_scheme
>>    if (quota->esz && quota->charged_sz + sz > quota->esz)
>>      sz = ALIGN_DOWN(quota->esz - quota->charged_sz, DAMON_MIN_REGION);
>>
>> Total charged stats use the unsigned long long data type to reduce
>> overflow risk, with data reset capability after overflow occurs.
> Thank you for finding this issue!  I don't want to change the data type if
> possible, though.  Could replacing the easily-overflowing throughput
> calculation with mult_frac() fix the issue?

Thank you for your guidance, it does work effectively. The relevant changes
have been included in patch #12 of the v2 series[1].

[1] https://lore.kernel.org/all/20250820080623.3799131-13-yanquanmin1@huawei.com/

Thanks, Quanmin Yan

Re: [RFC PATCH -next 14/16] mm/damon/core: convert sz to byte units when updating state

[Quanmin Yan]

Hi SJ,

在 2025/8/14 1:08, SeongJae Park 写道:
> On Wed, 13 Aug 2025 13:07:04 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>
>> After introducing ctx->addr_unit, the unit of sz might not
>> be in bytes. However, sz_applied is returned in bytes after
>> processing by paddr.
> This is not an intended behavior, but a bug of my code.  Let's update patches
> 3-6 of this series to return sz_applied as core-layer address unit, instead of
> bytes.
>
>> To maintain external consistency, sz
>> is converted to byte units when updating the state.
> Users could keep the consistency by multiplying the addr_unit, which they set
> themselves.

Thanks for the clarification! this patch has been dropped from the v2 series.

Thanks,
Quanmin Yan

Re: [RFC PATCH -next 00/16] mm/damon: support ARM32 with LPAE

[Quanmin Yan]

Hi SJ,

在 2025/8/15 0:04, SeongJae Park 写道:
> On Thu, 14 Aug 2025 22:07:12 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
>
>> Hi SJ,
>>
>> Thank you for your detailed suggestions on the patch series. Please allow me
>> some time to thoroughly review each of your recommendations.
> No worry, please take your time :)
>
>> I haven’t responded
>> to every point immediately because I’d like to first attempt updating the patches
>> accordingly. If I encounter any questions or issues during the process, I’ll promptly
>> reach out to discuss them with you, very appreciate your patience and guidance.
> Sounds good.
>
>> By the way, this patch series is based on linux-next(commit:2674d1eadaa2).
> Thank you for sharing this.  From the next time, please use mm-new[1] as a
> baseline for DAMON patches if there is no reason to not do so.
>
> [1] https://origin.kernel.org/doc/html/latest/mm/damon/maintainer-profile.html#scm-trees

I've prepared the v2 patch set based on the mm-new branch[1]. Your valuable
feedback would be greatly appreciated!

[1] https://lore.kernel.org/all/20250820080623.3799131-1-yanquanmin1@huawei.com/

Thanks,
Quanmin Yan

Re: [RFC PATCH -next 15/16] mm/damon: the byte statistics data type in damos_stat uses unsigned long long

[SeongJae Park]

On Wed, 20 Aug 2025 17:54:32 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:

> Hi SJ,
> 
> 在 2025/8/14 1:10, SeongJae Park 写道:
> > On Wed, 13 Aug 2025 13:07:05 +0800 Quanmin Yan <yanquanmin1@huawei.com> wrote:
> >
> >> For 32-bit systems, damos_stat now uses unsigned long long for byte
> >> statistics data to avoid integer overflow risks inherent in the
> >> previous design.
> > I suggested using the core-layer address unit on stat, and ask users to
> > multiply the addr_unit value to stat values if they want bytes value.  If we
> > agree on it, I think this patch wouldn't really be required.
> 
> Thank you for the guidance, I agree with your perspective. However, this patch doesn't actually belong in the addr_unit series, my apologies
> for the confusion. It is actually intended to address potential overflow issues
> in statistical data on 32-bit systems, and it is not directly related to addr_unit.
> This patch has been dropped from the v2 series.
> 
> After introducing addr_unit, if it is set to a larger value, it can help mitigate
> the overflow issue. However, under the default setting of addr_unit=1, statistical
> data may still overflow after a sufficiently long runtime, for example, when sz_tried
> exceeds 4GB.

Thank you for clarifying this!  My opinion is that, since we use core-layer
address unit for DAMOS stats, as long as users set appropriate addr_unit, I
think the overflow wouldn't really happen in real problematic ways?

For example, if addr_unit is 2**10 (=1024) and the scheme has tried to 4 *
2**30 bytes (4 GiB) of region, the sz_tried value will be 4 * 2**20, so far
from overflowing.

I think still the chance to overflow is higher than 64bit, but maybe the user
space tools can monitor and handle the overflow...?  Maybe we can discuss
further, but let's focus on the essential part for now.

> 
> Besides, please allow me to mention one point in advance: if addr is extended for
> use in modules(e.g. DAMON_RECLAIM, LRU_SORT) in the future, the term "bytes" in
> module_param_named(bytes_##try_name...), although multiplied by addr would yield
> the actual byte count, might cause confusion due to its seemingly direct naming.

Thank you for heasdup!  I agree it could be confusing, but I have no real good
idea at the moment, sorry.  Let's revisit after the essential part work is
done.

> 
> Overall, this patch isn’t critically important at the moment, nor does it offer a
> sufficiently robust solution, but I’d still appreciate hearing your perspective on
> the matter — I’m all ears.

Thank you again for headsup of the remaining issues.  Yes, let's keep eyes and
revisit those later.


Thanks,
SJ