Skip to contentSkip to conversation

correct the parameter type of some mm functions

PATCH Superseded Qi Zheng 7 participants 23 messages Mar 24 – Mar 26

Subsystems (3)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Qi Zheng

author
#0 lore raw

cover correct the parameter type of some mm functions

From: Qi Zheng <[email protected]>

Hi all,

As Harry Yoo pointed out [1], in the case of reparenting LRU folios, the value passed to functions such as __mod_memcg{_lruvec}_state() may exceed the upper limit of int, so this series aims to correct the parameter type of these functions.

This series is based on the next-20260323.

Comments and suggestions are welcome!

Thanks,
Qi

[1]. https://lore.kernel.org/all/acDxaEgnqPI-Z4be@hyeyoo/

Qi Zheng (3):
mm: memcontrol: correct the type of stats_updates to unsigned long
mm: memcontrol: correct the parameter type of

    __mod_memcg{_lruvec}_state()

mm: memcontrol: correct the nr_pages parameter type of
mem_cgroup_update_lru_size()

 include/linux/memcontrol.h   |  2 +-
 include/trace/events/memcg.h | 10 +++++-----
 mm/memcontrol.c              | 28 ++++++++++++++--------------
 3 files changed, 20 insertions(+), 20 deletions(-)
Signature · 1 line
2.20.1
author
#1 lore raw

PATCH 1/1 fix: mm: memcontrol: convert objcg to be per-memcg per-node type

From: Qi Zheng <[email protected]>

Reset pn->orig_objcg to NULL to prevent obj_cgroup_put() from being called agagin in __mem_cgroup_free().

---
 mm/memcontrol.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
mm/memcontrol.c +7 -1
index 992a3f5caa62b..ad32639ea5959 100644
@@ -4140,8 +4140,14 @@ static int mem_cgroup_css_online(struct cgroup_subsys_state *css)
 	for_each_node(nid) {
 		struct mem_cgroup_per_node *pn = memcg->nodeinfo[nid];
 
-		if (pn && pn->orig_objcg)
+		if (pn && pn->orig_objcg) {
 			obj_cgroup_put(pn->orig_objcg);
+			/*
+			 * Reset pn->orig_objcg to NULL to prevent obj_cgroup_put()
+			 * from being called agagin in __mem_cgroup_free().
+			 */
+			pn->orig_objcg = NULL;
+		}
 	}
 	free_shrinker_info(memcg);
 offline_kmem:
Signature · 1 line
2.20.1
#2 lore raw

Re: [PATCH] fix: mm: memcontrol: convert objcg to be per-memcg per-node type

Oops, Please ignore this patch.

author
#3 lore raw

PATCH 1/3 mm: memcontrol: correct the type of stats_updates to unsigned long

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

Now, the memcg_rstat_updated() is for vmstats_percpu->state and lruvec_stats_percpu->state, which are both of type long, so let's change the type of stats_updates to unsigned long as well.

---
 mm/memcontrol.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
mm/memcontrol.c +9 -9
index a47fb68dd65f1..7fb9cbc10dfbb 100644
@@ -608,7 +608,7 @@ static inline int memcg_events_index(enum vm_event_item idx)
 
 struct memcg_vmstats_percpu {
 	/* Stats updates since the last flush */
-	unsigned int			stats_updates;
+	unsigned long			stats_updates;
 
 	/* Cached pointers for fast iteration in memcg_rstat_updated() */
 	struct memcg_vmstats_percpu __percpu	*parent_pcpu;
@@ -639,7 +639,7 @@ struct memcg_vmstats {
 	unsigned long		events_pending[NR_MEMCG_EVENTS];
 
 	/* Stats updates since the last flush */
-	atomic_t		stats_updates;
+	atomic_long_t		stats_updates;
 };
 
 /*
@@ -665,16 +665,16 @@ static u64 flush_last_time;
 
 static bool memcg_vmstats_needs_flush(struct memcg_vmstats *vmstats)
 {
-	return atomic_read(&vmstats->stats_updates) >
+	return atomic_long_read(&vmstats->stats_updates) >
 		MEMCG_CHARGE_BATCH * num_online_cpus();
 }
 
-static inline void memcg_rstat_updated(struct mem_cgroup *memcg, int val,
+static inline void memcg_rstat_updated(struct mem_cgroup *memcg, long val,
 				       int cpu)
 {
 	struct memcg_vmstats_percpu __percpu *statc_pcpu;
 	struct memcg_vmstats_percpu *statc;
-	unsigned int stats_updates;
+	unsigned long stats_updates;
 
 	if (!val)
 		return;
@@ -697,7 +697,7 @@ static inline void memcg_rstat_updated(struct mem_cgroup *memcg, int val,
 			continue;
 
 		stats_updates = this_cpu_xchg(statc_pcpu->stats_updates, 0);
-		atomic_add(stats_updates, &statc->vmstats->stats_updates);
+		atomic_long_add(stats_updates, &statc->vmstats->stats_updates);
 	}
 }
 
@@ -705,7 +705,7 @@ static void __mem_cgroup_flush_stats(struct mem_cgroup *memcg, bool force)
 {
 	bool needs_flush = memcg_vmstats_needs_flush(memcg->vmstats);
 
-	trace_memcg_flush_stats(memcg, atomic_read(&memcg->vmstats->stats_updates),
+	trace_memcg_flush_stats(memcg, atomic_long_read(&memcg->vmstats->stats_updates),
 		force, needs_flush);
 
 	if (!force && !needs_flush)
@@ -4406,8 +4406,8 @@ static void mem_cgroup_css_rstat_flush(struct cgroup_subsys_state *css, int cpu)
 	}
 	WRITE_ONCE(statc->stats_updates, 0);
 	/* We are in a per-cpu loop here, only do the atomic write once */
-	if (atomic_read(&memcg->vmstats->stats_updates))
-		atomic_set(&memcg->vmstats->stats_updates, 0);
+	if (atomic_long_read(&memcg->vmstats->stats_updates))
+		atomic_long_set(&memcg->vmstats->stats_updates, 0);
 }
 
 static void mem_cgroup_fork(struct task_struct *task)
Signature · 1 line
2.20.1
#4 lore raw

Re: [PATCH 1/3] mm: memcontrol: correct the type of stats_updates to unsigned long

(-cc old mail)

Quoted 3 lines · Qi Zheng

From: Qi Zheng <[email protected]>

Now, the memcg_rstat_updated() is for vmstats_percpu->state and

Now? Did this change? If so, please put commit here in commit xxx ("blah") format, I don't want to have to try and look it up :P

Quoted 2 lines

lruvec_stats_percpu->state, which are both of type long, so let's change the type of stats_updates to unsigned long as well.

Is this a bug whereby before this could be overflowed?

Or are you deciding to make these long now? Because it seems that are proactively making this change _yourself_.

This comment message needs a lot more explanation.

Quoted 80 lines
---
 mm/memcontrol.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
mm/memcontrol.c +9 -10
index a47fb68dd65f1..7fb9cbc10dfbb 100644
@@ -608,7 +608,7 @@ static inline int memcg_events_index(enum vm_event_item idx)
 
 struct memcg_vmstats_percpu {
 	/* Stats updates since the last flush */
-	unsigned int			stats_updates;
+	unsigned long			stats_updates;
 
 	/* Cached pointers for fast iteration in memcg_rstat_updated() */
 	struct memcg_vmstats_percpu __percpu	*parent_pcpu;
@@ -639,7 +639,7 @@ struct memcg_vmstats {
 	unsigned long		events_pending[NR_MEMCG_EVENTS];
 
 	/* Stats updates since the last flush */
-	atomic_t		stats_updates;
+	atomic_long_t		stats_updates;
 };
 
 /*
@@ -665,16 +665,16 @@ static u64 flush_last_time;
 
 static bool memcg_vmstats_needs_flush(struct memcg_vmstats *vmstats)
 {
-	return atomic_read(&vmstats->stats_updates) >
+	return atomic_long_read(&vmstats->stats_updates) >
 		MEMCG_CHARGE_BATCH * num_online_cpus();
 }
 
-static inline void memcg_rstat_updated(struct mem_cgroup *memcg, int val,
+static inline void memcg_rstat_updated(struct mem_cgroup *memcg, long val,
 				       int cpu)
 {
 	struct memcg_vmstats_percpu __percpu *statc_pcpu;
 	struct memcg_vmstats_percpu *statc;
-	unsigned int stats_updates;
+	unsigned long stats_updates;
 
 	if (!val)
 		return;
@@ -697,7 +697,7 @@ static inline void memcg_rstat_updated(struct mem_cgroup *memcg, int val,
 			continue;
 
 		stats_updates = this_cpu_xchg(statc_pcpu->stats_updates, 0);
-		atomic_add(stats_updates, &statc->vmstats->stats_updates);
+		atomic_long_add(stats_updates, &statc->vmstats->stats_updates);
 	}
 }
 
@@ -705,7 +705,7 @@ static void __mem_cgroup_flush_stats(struct mem_cgroup *memcg, bool force)
 {
 	bool needs_flush = memcg_vmstats_needs_flush(memcg->vmstats);
 
-	trace_memcg_flush_stats(memcg, atomic_read(&memcg->vmstats->stats_updates),
+	trace_memcg_flush_stats(memcg, atomic_long_read(&memcg->vmstats->stats_updates),
 		force, needs_flush);
 
 	if (!force && !needs_flush)
@@ -4406,8 +4406,8 @@ static void mem_cgroup_css_rstat_flush(struct cgroup_subsys_state *css, int cpu)
 	}
 	WRITE_ONCE(statc->stats_updates, 0);
 	/* We are in a per-cpu loop here, only do the atomic write once */
-	if (atomic_read(&memcg->vmstats->stats_updates))
-		atomic_set(&memcg->vmstats->stats_updates, 0);
+	if (atomic_long_read(&memcg->vmstats->stats_updates))
+		atomic_long_set(&memcg->vmstats->stats_updates, 0);
 }
 
 static void mem_cgroup_fork(struct task_struct *task)
--

2.20.1

author
#5 lore raw

PATCH 2/3 mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them to long.

---
 include/trace/events/memcg.h | 10 +++++-----
 mm/memcontrol.c              |  8 ++++----
 2 files changed, 9 insertions(+), 9 deletions(-)
include/trace/events/memcg.h +5 -5
index dfe2f51019b4c..51b62c5931fc2 100644
@@ -11,14 +11,14 @@
 
 DECLARE_EVENT_CLASS(memcg_rstat_stats,
 
-	TP_PROTO(struct mem_cgroup *memcg, int item, int val),
+	TP_PROTO(struct mem_cgroup *memcg, int item, long val),
 
 	TP_ARGS(memcg, item, val),
 
 	TP_STRUCT__entry(
 		__field(u64, id)
 		__field(int, item)
-		__field(int, val)
+		__field(long, val)
 	),
 
 	TP_fast_assign(
@@ -27,20 +27,20 @@ DECLARE_EVENT_CLASS(memcg_rstat_stats,
 		__entry->val = val;
 	),
 
-	TP_printk("memcg_id=%llu item=%d val=%d",
+	TP_printk("memcg_id=%llu item=%d val=%ld",
 		  __entry->id, __entry->item, __entry->val)
 );
 
 DEFINE_EVENT(memcg_rstat_stats, mod_memcg_state,
 
-	TP_PROTO(struct mem_cgroup *memcg, int item, int val),
+	TP_PROTO(struct mem_cgroup *memcg, int item, long val),
 
 	TP_ARGS(memcg, item, val)
 );
 
 DEFINE_EVENT(memcg_rstat_stats, mod_memcg_lruvec_state,
 
-	TP_PROTO(struct mem_cgroup *memcg, int item, int val),
+	TP_PROTO(struct mem_cgroup *memcg, int item, long val),
 
 	TP_ARGS(memcg, item, val)
 );
mm/memcontrol.c +4 -4
index 7fb9cbc10dfbb..4a78550f6174e 100644
@@ -527,7 +527,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
 
 #ifdef CONFIG_MEMCG_V1
 static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val);
+				     enum node_stat_item idx, long val);
 
 void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
 				       struct mem_cgroup *parent, int idx)
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
  * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
  * up non-zero sub-page updates to 1 page as zero page updates are ignored.
  */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
 {
 	int unit = memcg_page_state_unit(idx);
 
@@ -831,7 +831,7 @@ static inline void get_non_dying_memcg_end(void)
 #endif
 
 static void __mod_memcg_state(struct mem_cgroup *memcg,
-			      enum memcg_stat_item idx, int val)
+			      enum memcg_stat_item idx, long val)
 {
 	int i = memcg_stats_index(idx);
 	int cpu;
@@ -896,7 +896,7 @@ void reparent_memcg_state_local(struct mem_cgroup *memcg,
 #endif
 
 static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val)
+				     enum node_stat_item idx, long val)
 {
 	struct mem_cgroup *memcg = pn->memcg;
 	int i = memcg_stats_index(idx);
Signature · 1 line
2.20.1
#6 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

Quoted 5 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them

Why might they? What precipitated this change?

Quoted 1 line

to long.

Why is it a signed value?

Quoted 93 lines
---
 include/trace/events/memcg.h | 10 +++++-----
 mm/memcontrol.c              |  8 ++++----
 2 files changed, 9 insertions(+), 9 deletions(-)
include/trace/events/memcg.h +5 -5
index dfe2f51019b4c..51b62c5931fc2 100644
@@ -11,14 +11,14 @@
 
 DECLARE_EVENT_CLASS(memcg_rstat_stats,
 
-	TP_PROTO(struct mem_cgroup *memcg, int item, int val),
+	TP_PROTO(struct mem_cgroup *memcg, int item, long val),
 
 	TP_ARGS(memcg, item, val),
 
 	TP_STRUCT__entry(
 		__field(u64, id)
 		__field(int, item)
-		__field(int, val)
+		__field(long, val)
 	),
 
 	TP_fast_assign(
@@ -27,20 +27,20 @@ DECLARE_EVENT_CLASS(memcg_rstat_stats,
 		__entry->val = val;
 	),
 
-	TP_printk("memcg_id=%llu item=%d val=%d",
+	TP_printk("memcg_id=%llu item=%d val=%ld",
 		  __entry->id, __entry->item, __entry->val)
 );
 
 DEFINE_EVENT(memcg_rstat_stats, mod_memcg_state,
 
-	TP_PROTO(struct mem_cgroup *memcg, int item, int val),
+	TP_PROTO(struct mem_cgroup *memcg, int item, long val),
 
 	TP_ARGS(memcg, item, val)
 );
 
 DEFINE_EVENT(memcg_rstat_stats, mod_memcg_lruvec_state,
 
-	TP_PROTO(struct mem_cgroup *memcg, int item, int val),
+	TP_PROTO(struct mem_cgroup *memcg, int item, long val),
 
 	TP_ARGS(memcg, item, val)
 );
mm/memcontrol.c +4 -5
index 7fb9cbc10dfbb..4a78550f6174e 100644
@@ -527,7 +527,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
 
 #ifdef CONFIG_MEMCG_V1
 static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val);
+				     enum node_stat_item idx, long val);
 
 void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
 				       struct mem_cgroup *parent, int idx)
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
  * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
  * up non-zero sub-page updates to 1 page as zero page updates are ignored.
  */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
 {
 	int unit = memcg_page_state_unit(idx);
 
@@ -831,7 +831,7 @@ static inline void get_non_dying_memcg_end(void)
 #endif
 
 static void __mod_memcg_state(struct mem_cgroup *memcg,
-			      enum memcg_stat_item idx, int val)
+			      enum memcg_stat_item idx, long val)
 {
 	int i = memcg_stats_index(idx);
 	int cpu;
@@ -896,7 +896,7 @@ void reparent_memcg_state_local(struct mem_cgroup *memcg,
 #endif
 
 static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val)
+				     enum node_stat_item idx, long val)
 {
 	struct mem_cgroup *memcg = pn->memcg;
 	int i = memcg_stats_index(idx);
--
Quoted 2 lines

2.20.1

Cheers, Lorenzo

#7 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

Quoted 7 lines · Lorenzo Stoakes (Oracle)
Quoted 5 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them

Quoted 5 lines

Why might they? What precipitated this change?

Quoted 1 line

to long.

Why is it a signed value?

Because 'mod' can both increase and decrease this counter. It must be signed.

#8 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

Quoted 8 lines · Matthew Wilcox
Quoted 7 lines · Lorenzo Stoakes (Oracle)
Quoted 5 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them

Quoted 8 lines
Quoted 5 lines

Why might they? What precipitated this change?

Quoted 1 line

to long.

Why is it a signed value?

Because 'mod' can both increase and decrease this counter. It must be signed.

Right yeah noticed that in another patch :)

Cheers, Lorenzo

#9 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

In reply to Qi Zheng
Quoted 27 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them to long.

---
 include/trace/events/memcg.h | 10 +++++-----
 mm/memcontrol.c              |  8 ++++----
 2 files changed, 9 insertions(+), 9 deletions(-)
mm/memcontrol.c +1 -1
index 7fb9cbc10dfbb..4a78550f6174e 100644
@@ -527,7 +527,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
 
 #ifdef CONFIG_MEMCG_V1
 static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val);
+				     enum node_stat_item idx, long val);
 
 void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
 				       struct mem_cgroup *parent, int idx)

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 789

Quoted 7 lines
mm/memcontrol.c +1 -1
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
  * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
  * up non-zero sub-page updates to 1 page as zero page updates are ignored.
  */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
 {
 	int unit = memcg_page_state_unit(idx);

Sashiko AI made an interesting argument [1] that this could lead to incorrectly returning a very large positive number. Let me verify that.

[1] https://sashiko.dev/#/patchset/cover.1774342371.git.zhengqi.arch%40bytedance.com

Quoted 9 lines · Sashiko

Does this change inadvertently break the handling of negative byte-sized updates?
Looking at the rest of the function:

	if (!val || unit == PAGE_SIZE)
		return val;
	else
		return max(val * unit / PAGE_SIZE, 1UL);

PAGE_SIZE is defined as an unsigned long.

Right, it's defined as 1UL << PAGE_SHIFT.

Quoted 2 lines

When val is negative, such as during uncharging of byte-sized stats like MEMCG_ZSWAP_B, the expression val * unit is a negative long.

Right.

Quoted 2 lines

Dividing a signed long by an unsigned long causes the signed long to be promoted to unsigned before division,

Right.

Quoted 2 lines

resulting in a massive positive
number instead of a small negative one.

Let's look at an example (assuming unit is 1).

val = val * unit = -16384 (-16 KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(0x3FFFFFFFFFFFFF, 1UL) = 0x3FFFFFFFFFF

Yeah, that's a massive positive number.

Hmm but how did it work when it was int?

val = val * unit = -16384 (-16KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(val * unit / PAGE_SIZE, 1UL) = 0x3FFFFFFFFFFFFF (int)0x3FFFFFFFFFFFFF = 0xFFFFFFFF = (-1)

That's incorrect. It should have been -4?

Quoted 3 lines

Before this change, the function returned an int, which implicitly truncated the massive unsigned 64-bit result to a 32-bit int, accidentally yielding the correct negative arithmetic value.

So... "accidentally yielding the correct negative arithemetic value" is wrong.

Sounds like it's been subtly broken even before this patch and nobody noticed.

Quoted 2 lines

By changing the return type to long, this implicit truncation is removed, and the huge positive value is returned unaltered.

That's true.

Quoted 1 line

Could this corrupt tracepoint logs when passed to trace_mod_memcg_state?

I'm not sure if that's critical but yeah that's true.

Quoted 3 lines

Also, would passing this huge positive value to memcg_rstat_updated instantly exceed the charge batch threshold and trigger endless, expensive global cgroup_rstat flushing, severely degrading system performance?

It would lead to more frequent flushes, at least.

Signature · 2 lines
Cheers,
Harry / Hyeonggon
author
#10 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

Quoted 88 lines · Harry Yoo (Oracle)
Quoted 27 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them to long.

---
  include/trace/events/memcg.h | 10 +++++-----
  mm/memcontrol.c              |  8 ++++----
  2 files changed, 9 insertions(+), 9 deletions(-)
mm/memcontrol.c +1 -1
index 7fb9cbc10dfbb..4a78550f6174e 100644
@@ -527,7 +527,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
 
  #ifdef CONFIG_MEMCG_V1
  static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val);
+				     enum node_stat_item idx, long val);
 
  void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
  				       struct mem_cgroup *parent, int idx)

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 789

Quoted 7 lines
mm/memcontrol.c +1 -1
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
   * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
   * up non-zero sub-page updates to 1 page as zero page updates are ignored.
   */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
  {
  	int unit = memcg_page_state_unit(idx);

Sashiko AI made an interesting argument [1] that this could lead to incorrectly returning a very large positive number. Let me verify that.

[1] https://sashiko.dev/#/patchset/cover.1774342371.git.zhengqi.arch%40bytedance.com

Quoted 9 lines · Sashiko

Does this change inadvertently break the handling of negative byte-sized updates?
Looking at the rest of the function:

	if (!val || unit == PAGE_SIZE)
		return val;
	else
		return max(val * unit / PAGE_SIZE, 1UL);

PAGE_SIZE is defined as an unsigned long.

Right, it's defined as 1UL << PAGE_SHIFT.

Quoted 2 lines

When val is negative, such as during uncharging of byte-sized stats like MEMCG_ZSWAP_B, the expression val * unit is a negative long.

Right.

Quoted 2 lines

Dividing a signed long by an unsigned long causes the signed long to be promoted to unsigned before division,

Right.

Quoted 2 lines

resulting in a massive positive
number instead of a small negative one.

Let's look at an example (assuming unit is 1).

val = val * unit = -16384 (-16 KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(0x3FFFFFFFFFFFFF, 1UL) = 0x3FFFFFFFFFF

Yeah, that's a massive positive number.

Hmm but how did it work when it was int?

val = val * unit = -16384 (-16KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(val * unit / PAGE_SIZE, 1UL) = 0x3FFFFFFFFFFFFF (int)0x3FFFFFFFFFFFFF = 0xFFFFFFFF = (-1)

That's incorrect. It should have been -4?

Quoted 3 lines

Before this change, the function returned an int, which implicitly truncated the massive unsigned 64-bit result to a 32-bit int, accidentally yielding the correct negative arithmetic value.

Quoted 5 lines

So... "accidentally yielding the correct negative arithemetic value" is wrong.

Sounds like it's been subtly broken even before this patch and nobody noticed.

Thank you for such a detailed analysis! And I think you are right.

The memcg_state_val_in_pages() is only to make @val to be in pages, so perhaps we can avoid the above problem by taking the absolute value first?

Thanks,
Qi

Quoted 16 lines
Quoted 3 lines

By changing the return type to long, this implicit truncation is removed, and the huge positive value is returned unaltered.

That's true.

Quoted 1 line

Could this corrupt tracepoint logs when passed to trace_mod_memcg_state?

I'm not sure if that's critical but yeah that's true.

Quoted 3 lines

Also, would passing this huge positive value to memcg_rstat_updated instantly exceed the charge batch threshold and trigger endless, expensive global cgroup_rstat flushing, severely degrading system performance?

It would lead to more frequent flushes, at least.

#11 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

Quoted 87 lines · Qi Zheng
Quoted 86 lines · Harry Yoo (Oracle)
Quoted 25 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them to long.

---
  include/trace/events/memcg.h | 10 +++++-----
  mm/memcontrol.c              |  8 ++++----
  2 files changed, 9 insertions(+), 9 deletions(-)
mm/memcontrol.c +1 -1
index 7fb9cbc10dfbb..4a78550f6174e 100644
@@ -527,7 +527,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
  #ifdef CONFIG_MEMCG_V1
  static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val);
+				     enum node_stat_item idx, long val);
  void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
  				       struct mem_cgroup *parent, int idx)

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 789

Quoted 7 lines
mm/memcontrol.c +1 -1
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
   * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
   * up non-zero sub-page updates to 1 page as zero page updates are ignored.
   */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
  {
  	int unit = memcg_page_state_unit(idx);

Sashiko AI made an interesting argument [1] that this could lead to incorrectly returning a very large positive number. Let me verify that.

[1] https://sashiko.dev/#/patchset/cover.1774342371.git.zhengqi.arch%40bytedance.com

Quoted 9 lines · Sashiko

Does this change inadvertently break the handling of negative byte-sized updates?
Looking at the rest of the function:

	if (!val || unit == PAGE_SIZE)
		return val;
	else
		return max(val * unit / PAGE_SIZE, 1UL);

PAGE_SIZE is defined as an unsigned long.

Right, it's defined as 1UL << PAGE_SHIFT.

Quoted 2 lines

When val is negative, such as during uncharging of byte-sized stats like MEMCG_ZSWAP_B, the expression val * unit is a negative long.

Right.

Quoted 2 lines

Dividing a signed long by an unsigned long causes the signed long to be promoted to unsigned before division,

Right.

Quoted 2 lines

resulting in a massive positive
number instead of a small negative one.

Let's look at an example (assuming unit is 1).

val = val * unit = -16384 (-16 KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(0x3FFFFFFFFFFFFF, 1UL) = 0x3FFFFFFFFFF

Yeah, that's a massive positive number.

Hmm but how did it work when it was int?

val = val * unit = -16384 (-16KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(val * unit / PAGE_SIZE, 1UL) = 0x3FFFFFFFFFFFFF (int)0x3FFFFFFFFFFFFF = 0xFFFFFFFF = (-1)

That's incorrect. It should have been -4?

Quoted 3 lines

Before this change, the function returned an int, which implicitly truncated the massive unsigned 64-bit result to a 32-bit int, accidentally yielding the correct negative arithmetic value.

Quoted 7 lines
Quoted 5 lines

So... "accidentally yielding the correct negative arithemetic value" is wrong.

Sounds like it's been subtly broken even before this patch and nobody noticed.

Thank you for such a detailed analysis! And I think you are right.

No problem ;)

Quoted 3 lines

The memcg_state_val_in_pages() is only to make @val to be in pages, so perhaps we can avoid the above problem by taking the absolute value first?

That would work for memcg_rstat_updated(),
but not for trace_mod_memcg_state()?

Signature · 2 lines
Cheers,
Harry / Hyeonggon
author
#12 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

Quoted 91 lines · Harry Yoo (Oracle)
Quoted 90 lines · Qi Zheng
Quoted 89 lines · Harry Yoo (Oracle)
Quoted 25 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them to long.

---
   include/trace/events/memcg.h | 10 +++++-----
   mm/memcontrol.c              |  8 ++++----
   2 files changed, 9 insertions(+), 9 deletions(-)
mm/memcontrol.c +1 -1
index 7fb9cbc10dfbb..4a78550f6174e 100644
@@ -527,7 +527,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
   #ifdef CONFIG_MEMCG_V1
   static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val);
+				     enum node_stat_item idx, long val);
   void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
   				       struct mem_cgroup *parent, int idx)

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 789

Quoted 7 lines
mm/memcontrol.c +1 -1
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
    * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
    * up non-zero sub-page updates to 1 page as zero page updates are ignored.
    */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
   {
   	int unit = memcg_page_state_unit(idx);

Sashiko AI made an interesting argument [1] that this could lead to incorrectly returning a very large positive number. Let me verify that.

[1] https://sashiko.dev/#/patchset/cover.1774342371.git.zhengqi.arch%40bytedance.com

Quoted 9 lines · Sashiko

Does this change inadvertently break the handling of negative byte-sized updates?
Looking at the rest of the function:

	if (!val || unit == PAGE_SIZE)
		return val;
	else
		return max(val * unit / PAGE_SIZE, 1UL);

PAGE_SIZE is defined as an unsigned long.

Right, it's defined as 1UL << PAGE_SHIFT.

Quoted 2 lines

When val is negative, such as during uncharging of byte-sized stats like MEMCG_ZSWAP_B, the expression val * unit is a negative long.

Right.

Quoted 2 lines

Dividing a signed long by an unsigned long causes the signed long to be promoted to unsigned before division,

Right.

Quoted 2 lines

resulting in a massive positive
number instead of a small negative one.

Let's look at an example (assuming unit is 1).

val = val * unit = -16384 (-16 KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(0x3FFFFFFFFFFFFF, 1UL) = 0x3FFFFFFFFFF

Yeah, that's a massive positive number.

Hmm but how did it work when it was int?

val = val * unit = -16384 (-16KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(val * unit / PAGE_SIZE, 1UL) = 0x3FFFFFFFFFFFFF (int)0x3FFFFFFFFFFFFF = 0xFFFFFFFF = (-1)

That's incorrect. It should have been -4?

Quoted 3 lines

Before this change, the function returned an int, which implicitly truncated the massive unsigned 64-bit result to a 32-bit int, accidentally yielding the correct negative arithmetic value.

So... "accidentally yielding the correct negative arithemetic value" is wrong.

Quoted 10 lines
Quoted 4 lines
Quoted 2 lines

Sounds like it's been subtly broken even before this patch and nobody noticed.

Thank you for such a detailed analysis! And I think you are right.

No problem ;)

Quoted 3 lines

The memcg_state_val_in_pages() is only to make @val to be in pages, so perhaps we can avoid the above problem by taking the absolute value first?

For more clearly, here is the diff:

mm/memcontrol.c +8 -2
index 6491ca74b3398..2b34805b01476 100644
@@ -787,11 +787,17 @@ static int memcg_page_state_unit(int item);
 static long memcg_state_val_in_pages(int idx, long val)
 {
        int unit = memcg_page_state_unit(idx);
+       unsigned long uval;
+       long res;
 
        if (!val || unit == PAGE_SIZE)
                return val;
-       else
-               return max(val * unit / PAGE_SIZE, 1UL);
+
+       uval = val < 0 ? -(unsigned long)val : (unsigned long)val;
+
+       res = max(mult_frac(uval, unit, PAGE_SIZE), 1UL);
+
+       return val < 0 ? -res : res;
 }
 
 #ifdef CONFIG_MEMCG_V1
Quoted 3 lines

That would work for memcg_rstat_updated(),
but not for trace_mod_memcg_state()?

Why? The trace_mod_memcg_state() seems to accept 'long' type, am I missing something?

Thanks,
Qi

Quoted 1 line
#13 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

Quoted 114 lines · Qi Zheng
Quoted 91 lines · Harry Yoo (Oracle)
Quoted 90 lines · Qi Zheng
Quoted 89 lines · Harry Yoo (Oracle)
Quoted 25 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them to long.

---
   include/trace/events/memcg.h | 10 +++++-----
   mm/memcontrol.c              |  8 ++++----
   2 files changed, 9 insertions(+), 9 deletions(-)
mm/memcontrol.c +1 -1
index 7fb9cbc10dfbb..4a78550f6174e 100644
@@ -527,7 +527,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
   #ifdef CONFIG_MEMCG_V1
   static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val);
+				     enum node_stat_item idx, long val);
   void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
   				       struct mem_cgroup *parent, int idx)

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 789

Quoted 7 lines
mm/memcontrol.c +1 -1
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
    * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
    * up non-zero sub-page updates to 1 page as zero page updates are ignored.
    */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
   {
   	int unit = memcg_page_state_unit(idx);

Sashiko AI made an interesting argument [1] that this could lead to incorrectly returning a very large positive number. Let me verify that.

[1] https://sashiko.dev/#/patchset/cover.1774342371.git.zhengqi.arch%40bytedance.com

Quoted 9 lines · Sashiko

Does this change inadvertently break the handling of negative byte-sized updates?
Looking at the rest of the function:

	if (!val || unit == PAGE_SIZE)
		return val;
	else
		return max(val * unit / PAGE_SIZE, 1UL);

PAGE_SIZE is defined as an unsigned long.

Right, it's defined as 1UL << PAGE_SHIFT.

Quoted 2 lines

When val is negative, such as during uncharging of byte-sized stats like MEMCG_ZSWAP_B, the expression val * unit is a negative long.

Right.

Quoted 2 lines

Dividing a signed long by an unsigned long causes the signed long to be promoted to unsigned before division,

Right.

Quoted 2 lines

resulting in a massive positive
number instead of a small negative one.

Let's look at an example (assuming unit is 1).

val = val * unit = -16384 (-16 KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(0x3FFFFFFFFFFFFF, 1UL) = 0x3FFFFFFFFFF

Yeah, that's a massive positive number.

Hmm but how did it work when it was int?

val = val * unit = -16384 (-16KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(val * unit / PAGE_SIZE, 1UL) = 0x3FFFFFFFFFFFFF (int)0x3FFFFFFFFFFFFF = 0xFFFFFFFF = (-1)

That's incorrect. It should have been -4?

Quoted 3 lines

Before this change, the function returned an int, which implicitly truncated the massive unsigned 64-bit result to a 32-bit int, accidentally yielding the correct negative arithmetic value.

So... "accidentally yielding the correct negative arithemetic value" is wrong.

Quoted 10 lines
Quoted 4 lines
Quoted 2 lines

Sounds like it's been subtly broken even before this patch and nobody noticed.

Thank you for such a detailed analysis! And I think you are right.

No problem ;)

Quoted 3 lines

The memcg_state_val_in_pages() is only to make @val to be in pages, so perhaps we can avoid the above problem by taking the absolute value first?

For more clearly, here is the diff:

mm/memcontrol.c +0 -0
index 6491ca74b3398..2b34805b01476 100644
@@ -787,11 +787,17 @@ static int memcg_page_state_unit(int item);
 static long memcg_state_val_in_pages(int idx, long val)
 {

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 796

Quoted 10 lines
mm/memcontrol.c +4 -2
        int unit = memcg_page_state_unit(idx);
+       unsigned long uval;
+       long res;
 
        if (!val || unit == PAGE_SIZE)
                return val;
-       else
-               return max(val * unit / PAGE_SIZE, 1UL);
+
+       uval = val < 0 ? -(unsigned long)val : (unsigned long)val;

We could use abs() macro in linux/math.h here :)

val should not be LONG_MIN, but it should be fine
to assume it won't reach LONG_MIN.

Quoted 8 lines
mm/memcontrol.c +4 -0
+
+       res = max(mult_frac(uval, unit, PAGE_SIZE), 1UL);
+
+       return val < 0 ? -res : res;
 }
 
 #ifdef CONFIG_MEMCG_V1
Quoted 5 lines
Quoted 2 lines

That would work for memcg_rstat_updated(),
but not for trace_mod_memcg_state()?

Why? The trace_mod_memcg_state() seems to accept 'long' type, am I missing something?

Nevermind. Without the diff, I misunderstood what you meant.

Thanks!

Signature · 2 lines
Cheers,
Harry / Hyeonggon
author
#14 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

Quoted 128 lines · Harry Yoo (Oracle)
Quoted 114 lines · Qi Zheng
Quoted 91 lines · Harry Yoo (Oracle)
Quoted 90 lines · Qi Zheng
Quoted 89 lines · Harry Yoo (Oracle)
Quoted 25 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The __mod_memcg_state() and __mod_memcg_lruvec_state() were used to reparent non-hierarchical stats, the values passed to them might exceed the upper limit of the type int, so correct the val parameter type of them to long.

---
    include/trace/events/memcg.h | 10 +++++-----
    mm/memcontrol.c              |  8 ++++----
    2 files changed, 9 insertions(+), 9 deletions(-)
mm/memcontrol.c +1 -1
index 7fb9cbc10dfbb..4a78550f6174e 100644
@@ -527,7 +527,7 @@ unsigned long lruvec_page_state_local(struct lruvec *lruvec,
    #ifdef CONFIG_MEMCG_V1
    static void __mod_memcg_lruvec_state(struct mem_cgroup_per_node *pn,
-				     enum node_stat_item idx, int val);
+				     enum node_stat_item idx, long val);
    void reparent_memcg_lruvec_state_local(struct mem_cgroup *memcg,
    				       struct mem_cgroup *parent, int idx)

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 789

Quoted 7 lines
mm/memcontrol.c +1 -1
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
     * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
     * up non-zero sub-page updates to 1 page as zero page updates are ignored.
     */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
    {
    	int unit = memcg_page_state_unit(idx);

Sashiko AI made an interesting argument [1] that this could lead to incorrectly returning a very large positive number. Let me verify that.

[1] https://sashiko.dev/#/patchset/cover.1774342371.git.zhengqi.arch%40bytedance.com

Quoted 9 lines · Sashiko

Does this change inadvertently break the handling of negative byte-sized updates?
Looking at the rest of the function:

	if (!val || unit == PAGE_SIZE)
		return val;
	else
		return max(val * unit / PAGE_SIZE, 1UL);

PAGE_SIZE is defined as an unsigned long.

Right, it's defined as 1UL << PAGE_SHIFT.

Quoted 2 lines

When val is negative, such as during uncharging of byte-sized stats like MEMCG_ZSWAP_B, the expression val * unit is a negative long.

Right.

Quoted 2 lines

Dividing a signed long by an unsigned long causes the signed long to be promoted to unsigned before division,

Right.

Quoted 2 lines

resulting in a massive positive
number instead of a small negative one.

Let's look at an example (assuming unit is 1).

val = val * unit = -16384 (-16 KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(0x3FFFFFFFFFFFFF, 1UL) = 0x3FFFFFFFFFF

Yeah, that's a massive positive number.

Hmm but how did it work when it was int?

val = val * unit = -16384 (-16KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(val * unit / PAGE_SIZE, 1UL) = 0x3FFFFFFFFFFFFF (int)0x3FFFFFFFFFFFFF = 0xFFFFFFFF = (-1)

That's incorrect. It should have been -4?

Quoted 3 lines

Before this change, the function returned an int, which implicitly truncated the massive unsigned 64-bit result to a 32-bit int, accidentally yielding the correct negative arithmetic value.

So... "accidentally yielding the correct negative arithemetic value" is wrong.

Quoted 10 lines
Quoted 4 lines
Quoted 2 lines

Sounds like it's been subtly broken even before this patch and nobody noticed.

Thank you for such a detailed analysis! And I think you are right.

No problem ;)

Quoted 3 lines

The memcg_state_val_in_pages() is only to make @val to be in pages, so perhaps we can avoid the above problem by taking the absolute value first?

For more clearly, here is the diff:

mm/memcontrol.c +0 -0
index 6491ca74b3398..2b34805b01476 100644
@@ -787,11 +787,17 @@ static int memcg_page_state_unit(int item);
  static long memcg_state_val_in_pages(int idx, long val)
  {

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 796

Quoted 10 lines
mm/memcontrol.c +4 -2
         int unit = memcg_page_state_unit(idx);
+       unsigned long uval;
+       long res;
 
         if (!val || unit == PAGE_SIZE)
                 return val;
-       else
-               return max(val * unit / PAGE_SIZE, 1UL);
+
+       uval = val < 0 ? -(unsigned long)val : (unsigned long)val;

We could use abs() macro in linux/math.h here :)

Quoted 2 lines

val should not be LONG_MIN, but it should be fine
to assume it won't reach LONG_MIN.

OK, will use abs() macro.

Quoted 9 lines
Quoted 9 lines
mm/memcontrol.c +4 -0
+
+       res = max(mult_frac(uval, unit, PAGE_SIZE), 1UL);
+
+       return val < 0 ? -res : res;
  }
 
  #ifdef CONFIG_MEMCG_V1
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines

That would work for memcg_rstat_updated(),
but not for trace_mod_memcg_state()?

Why? The trace_mod_memcg_state() seems to accept 'long' type, am I missing something?

Nevermind. Without the diff, I misunderstood what you meant.

OK, will apply this diff and send the v2.

Thanks,
Qi

Quoted 3 lines

Thanks!

#15 lore raw

Re: [PATCH 2/3] mm: memcontrol: correct the parameter type of __mod_memcg{_lruvec}_state()

In reply to Harry Yoo (Oracle)
Quoted 43 lines · Harry Yoo (Oracle)

mm/memcontrol.c static int memcg_page_state_unit(int item); · line 789

Quoted 8 lines · Qi Zheng
mm/memcontrol.c +1 -1
@@ -784,7 +784,7 @@ static int memcg_page_state_unit(int item);
  * Normalize the value passed into memcg_rstat_updated() to be in pages. Round
  * up non-zero sub-page updates to 1 page as zero page updates are ignored.
  */
-static int memcg_state_val_in_pages(int idx, int val)
+static long memcg_state_val_in_pages(int idx, long val)
 {
 	int unit = memcg_page_state_unit(idx);

Sashiko AI made an interesting argument [1] that this could lead to incorrectly returning a very large positive number. Let me verify that.

[1] https://sashiko.dev/#/patchset/cover.1774342371.git.zhengqi.arch%40bytedance.com

Quoted 9 lines · Sashiko

Does this change inadvertently break the handling of negative byte-sized updates?
Looking at the rest of the function:

	if (!val || unit == PAGE_SIZE)
		return val;
	else
		return max(val * unit / PAGE_SIZE, 1UL);

PAGE_SIZE is defined as an unsigned long.

Right, it's defined as 1UL << PAGE_SHIFT.

Quoted 2 lines

When val is negative, such as during uncharging of byte-sized stats like MEMCG_ZSWAP_B, the expression val * unit is a negative long.

Right.

Quoted 2 lines

Dividing a signed long by an unsigned long causes the signed long to be promoted to unsigned before division,

Right.

Quoted 2 lines

resulting in a massive positive
number instead of a small negative one.

Let's look at an example (assuming unit is 1).

Quoted 7 lines

val = val * unit = -16384 (-16 KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF max(0x3FFFFFFFFFFFFF, 1UL) = 0x3FFFFFFFFFF

Yeah, that's a massive positive number.

Hmm but how did it work when it was int?

Oops, I was about to say... "Oh, doesn't patch 4/4 in v2 need to have

Fixes 7bd5bc3ce963 ("mm: memcg: normalize the value passed into memcg_rstat_updated()")

???"

but then I realized that I made a silly mistake here.

Quoted 2 lines

val = val * unit = -16384 (-16KiB)
val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / PAGE_SIZE = 0x3FFFFFFFFFFFFF

Err, I should have divided it by 0x1000, not 0x4096.

val * unit / PAGE_SIZE = 0xFFFFFFFFFFFFC000 / 0x1000 = 0xFFFFFFFFFFFFC max(val * unit / PAGE_SIZE, 1UL) = 0xFFFFFFFFFFFFC (int)0xFFFFFFFFFFFFC = -4.

Quoted 4 lines

max(val * unit / PAGE_SIZE, 1UL) = 0x3FFFFFFFFFFFFF (int)0x3FFFFFFFFFFFFF = 0xFFFFFFFF = (-1)

That's incorrect. It should have been -4?

So that was correct.

The existing logic produces an accurate number (intended or not) as it is right-shifted to only PAGE_SHIFT bits and truncated to int.

The existing logic is fine, it'll only be a problem when it's not truncated to int.

Quoted 6 lines
Quoted 3 lines

Before this change, the function returned an int, which implicitly truncated the massive unsigned 64-bit result to a 32-bit int, accidentally yielding the correct negative arithmetic value.

So... "accidentally yielding the correct negative arithemetic value" is wrong.

I was wrong, not sashiko!

Quoted 2 lines

Sounds like it's been subtly broken even before this patch and nobody noticed.

No, it's not.

Signature · 2 lines
Cheers,
Harry / Hyeonggon
author
#16 lore raw

PATCH 3/3 mm: memcontrol: correct the nr_pages parameter type of mem_cgroup_update_lru_size()

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

The nr_pages parameter of mem_cgroup_update_lru_size() should clearly be long instead of int, let's correct it.

---
 include/linux/memcontrol.h | 2 +-
 mm/memcontrol.c            | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
include/linux/memcontrol.h +1 -1
index 0861589695298..dc3fa687759b4 100644
@@ -878,7 +878,7 @@ static inline bool mem_cgroup_online(struct mem_cgroup *memcg)
 }
 
 void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru,
-		int zid, int nr_pages);
+		int zid, long nr_pages);
 
 static inline
 unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec,
mm/memcontrol.c +1 -1
index 4a78550f6174e..6491ca74b3398 100644
@@ -1466,7 +1466,7 @@ struct lruvec *folio_lruvec_lock_irqsave(struct folio *folio,
  * to or just after a page is removed from an lru list.
  */
 void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru,
-				int zid, int nr_pages)
+				int zid, long nr_pages)
 {
 	struct mem_cgroup_per_node *mz;
 	unsigned long *lru_size;
Signature · 1 line
2.20.1
Reviewed-by on v1 3/3
#17 lore raw

Re: [PATCH 3/3] mm: memcontrol: correct the nr_pages parameter type of mem_cgroup_update_lru_size()

Quoted 4 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The nr_pages parameter of mem_cgroup_update_lru_size() should clearly be long instead of int, let's correct it.

Hmm, but are you ever going to be adding that many pages? I guess technically correct though.

You should list all the callers and how they use long or unsigned long, e.g. update_lru_sizes(), lruvec_reparent_lru(), lru_gen_reparent_memcg().

And maybe look a bit into why this was int, if there's any reason for it at all :)

I'm assuming there's no weird 'let's intentionally let this overflow' behaviour here.

Quoted 2 lines

This one looks correct overall, so, with commit message improved:

Quoted 34 lines
---
 include/linux/memcontrol.h | 2 +-
 mm/memcontrol.c            | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
include/linux/memcontrol.h +1 -1
index 0861589695298..dc3fa687759b4 100644
@@ -878,7 +878,7 @@ static inline bool mem_cgroup_online(struct mem_cgroup *memcg)
 }
 
 void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru,
-		int zid, int nr_pages);
+		int zid, long nr_pages);
 
 static inline
 unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec,
mm/memcontrol.c +1 -2
index 4a78550f6174e..6491ca74b3398 100644
@@ -1466,7 +1466,7 @@ struct lruvec *folio_lruvec_lock_irqsave(struct folio *folio,
  * to or just after a page is removed from an lru list.
  */
 void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru,
-				int zid, int nr_pages)
+				int zid, long nr_pages)
 {
 	struct mem_cgroup_per_node *mz;
 	unsigned long *lru_size;
--

2.20.1

#18 lore raw

Re: [PATCH 3/3] mm: memcontrol: correct the nr_pages parameter type of mem_cgroup_update_lru_size()

Quoted 8 lines · Lorenzo Stoakes (Oracle)
Quoted 4 lines · Qi Zheng

From: Qi Zheng <[email protected]>

The nr_pages parameter of mem_cgroup_update_lru_size() should clearly be long instead of int, let's correct it.

Hmm, but are you ever going to be adding that many pages? I guess technically correct though.

It has been fine as-is, but reparenting changes that. Reparenting in theory could add billions of pages at once!

And yeah the commit message could be improved.

Signature · 2 lines
Cheers,
Harry / Hyeonggon
author
#19 lore raw

Re: [PATCH 3/3] mm: memcontrol: correct the nr_pages parameter type of mem_cgroup_update_lru_size()

Quoted 4 lines · Harry Yoo (Oracle)
Quoted 3 lines · Lorenzo Stoakes (Oracle)
Quoted 2 lines · Qi Zheng

From: Qi Zheng <[email protected]>

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

The nr_pages parameter of mem_cgroup_update_lru_size() should clearly be long instead of int, let's correct it.

Hmm, but are you ever going to be adding that many pages? I guess technically correct though.

It has been fine as-is, but reparenting changes that. Reparenting in theory could add billions of pages at once!

Right. Thanks for the explanation!

Quoted 2 lines

And yeah the commit message could be improved.

Yes, I've omitted a lot of background information, which might seem abrupt to some reviewers unfamiliar with it. I will improve this in the next version.

Thanks,
Qi

Quoted 1 line
#20 lore raw

Re: [PATCH 3/3] mm: memcontrol: correct the nr_pages parameter type of mem_cgroup_update_lru_size()

In reply to Qi Zheng
kernel test robot noticed the following build warnings:

Hi Qi,

kernel test robot noticed the following build warnings:

[auto build test WARNING on akpm-mm/mm-everything]
[also build test WARNING on next-20260325]
[cannot apply to trace/for-next linus/master v7.0-rc5]
[If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

url:    https://github.com/intel-lab-lkp/linux/commits/Qi-Zheng/mm-memcontrol-correct-the-type-of-stats_updates-to-unsigned-long/20260325-230337
base:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything
patch link:    https://lore.kernel.org/r/2cf06f9faf51900ce6acbb4740fc60355a2842ed.1774342371.git.zhengqi.arch%40bytedance.com
patch subject: [PATCH 3/3] mm: memcontrol: correct the nr_pages parameter type of mem_cgroup_update_lru_size()
config: riscv-allnoconfig-bpf (https://download.01.org/0day-ci/archive/20260326/[email protected]/config)
compiler: riscv64-linux-gnu-gcc (Debian 14.2.0-19) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260326/[email protected]/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/

All warnings (new ones prefixed by >>):

In file included from ./arch/riscv/include/asm/bug.h:90,

                    from ./include/linux/bug.h:5,
                    from ./arch/riscv/include/asm/cmpxchg.h:9,
                    from ./arch/riscv/include/asm/barrier.h:14,
                    from ./include/linux/list.h:11,
                    from ./include/linux/cgroup-defs.h:12,
                    from mm/memcontrol.c:28:
   mm/memcontrol.c: In function 'mem_cgroup_update_lru_size':
>> mm/memcontrol.c:1486:17: warning: format '%d' expects argument of type 'int', but argument 5 has type 'long int' [-Wformat=]
    1486 |                 "%s(%p, %d, %d): lru_size %ld\n",
         |                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    1487 |                 __func__, lruvec, lru, nr_pages, size)) {
         |                                        ~~~~~~~~
         |                                        |
         |                                        long int
   ./include/asm-generic/bug.h:133:31: note: in definition of macro '__WARN_printf'
     133 |                 __warn_printk(arg);                                     \
         |                               ^~~
   ./include/linux/once_lite.h:31:25: note: in expansion of macro 'WARN'
      31 |                         func(__VA_ARGS__);                              \
         |                         ^~~~
   ./include/asm-generic/bug.h:185:9: note: in expansion of macro 'DO_ONCE_LITE_IF'
     185 |         DO_ONCE_LITE_IF(condition, WARN, 1, format)
         |         ^~~~~~~~~~~~~~~
   mm/memcontrol.c:1485:13: note: in expansion of macro 'WARN_ONCE'
    1485 |         if (WARN_ONCE(size < 0,
         |             ^~~~~~~~~
   mm/memcontrol.c:1486:30: note: format string is defined here
    1486 |                 "%s(%p, %d, %d): lru_size %ld\n",
         |                             ~^
         |                              |
         |                              int
         |                             %ld

vim +1486 mm/memcontrol.c

6168d0da2b479ce Alex Shi           2020-12-15  1457  
925b7673cce3911 Johannes Weiner    2012-01-12  1458  /**
fa9add641b1b1c5 Hugh Dickins       2012-05-29  1459   * mem_cgroup_update_lru_size - account for adding or removing an lru page
fa9add641b1b1c5 Hugh Dickins       2012-05-29  1460   * @lruvec: mem_cgroup per zone lru vector
fa9add641b1b1c5 Hugh Dickins       2012-05-29  1461   * @lru: index of lru list the page is sitting on
b4536f0c829c858 Michal Hocko       2017-01-10  1462   * @zid: zone id of the accounted pages
fa9add641b1b1c5 Hugh Dickins       2012-05-29  1463   * @nr_pages: positive when adding or negative when removing
925b7673cce3911 Johannes Weiner    2012-01-12  1464   *
ca707239e8a7958 Hugh Dickins       2016-05-19  1465   * This function must be called under lru_lock, just before a page is added
07ca760673088f2 Hugh Dickins       2022-02-14  1466   * to or just after a page is removed from an lru list.
925b7673cce3911 Johannes Weiner    2012-01-12  1467   */
fa9add641b1b1c5 Hugh Dickins       2012-05-29  1468  void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru,
6a3cda9fd4d0f0b Qi Zheng           2026-03-24  1469  				int zid, long nr_pages)
08e552c69c6930d KAMEZAWA Hiroyuki  2009-01-07  1470  {
ef8f2327996b5c2 Mel Gorman         2016-07-28  1471  	struct mem_cgroup_per_node *mz;
fa9add641b1b1c5 Hugh Dickins       2012-05-29  1472  	unsigned long *lru_size;
ca707239e8a7958 Hugh Dickins       2016-05-19  1473  	long size;
08e552c69c6930d KAMEZAWA Hiroyuki  2009-01-07  1474  
f8d665422603ee1 Hirokazu Takahashi 2009-01-07  1475  	if (mem_cgroup_disabled())
08e552c69c6930d KAMEZAWA Hiroyuki  2009-01-07  1476  		return;
6d12e2d8ddbe653 KAMEZAWA Hiroyuki  2008-02-07  1477  
ef8f2327996b5c2 Mel Gorman         2016-07-28  1478  	mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
b4536f0c829c858 Michal Hocko       2017-01-10  1479  	lru_size = &mz->lru_zone_size[zid][lru];
ca707239e8a7958 Hugh Dickins       2016-05-19  1480  
ca707239e8a7958 Hugh Dickins       2016-05-19  1481  	if (nr_pages < 0)
ca707239e8a7958 Hugh Dickins       2016-05-19  1482  		*lru_size += nr_pages;
ca707239e8a7958 Hugh Dickins       2016-05-19  1483  
ca707239e8a7958 Hugh Dickins       2016-05-19  1484  	size = *lru_size;
b4536f0c829c858 Michal Hocko       2017-01-10  1485  	if (WARN_ONCE(size < 0,
b4536f0c829c858 Michal Hocko       2017-01-10 @1486  		"%s(%p, %d, %d): lru_size %ld\n",
b4536f0c829c858 Michal Hocko       2017-01-10  1487  		__func__, lruvec, lru, nr_pages, size)) {
ca707239e8a7958 Hugh Dickins       2016-05-19  1488  		VM_BUG_ON(1);
ca707239e8a7958 Hugh Dickins       2016-05-19  1489  		*lru_size = 0;
ca707239e8a7958 Hugh Dickins       2016-05-19  1490  	}
ca707239e8a7958 Hugh Dickins       2016-05-19  1491  
ca707239e8a7958 Hugh Dickins       2016-05-19  1492  	if (nr_pages > 0)
fa9add641b1b1c5 Hugh Dickins       2012-05-29  1493  		*lru_size += nr_pages;
08e552c69c6930d KAMEZAWA Hiroyuki  2009-01-07  1494  }
544122e5e0ee27d KAMEZAWA Hiroyuki  2009-01-07  1495  
Signature · 2 lines
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
maintainer
#21 lore raw

Re: [PATCH 3/3] mm: memcontrol: correct the nr_pages parameter type of mem_cgroup_update_lru_size()

Quoted 11 lines · On Thu, 26 Mar 2026 03:35:39 +0100 kernel test robot <[email protected]> wrote:

Hi Qi,

kernel test robot noticed the following build warnings:

[auto build test WARNING on akpm-mm/mm-everything]
[also build test WARNING on next-20260325]
[cannot apply to trace/for-next linus/master v7.0-rc5]
[If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]

Quoted 7 lines
url:    https://github.com/intel-lab-lkp/linux/commits/Qi-Zheng/mm-memcontrol-correct-the-type-of-stats_updates-to-unsigned-long/20260325-230337
base:   https://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm.git mm-everything
patch link:    https://lore.kernel.org/r/2cf06f9faf51900ce6acbb4740fc60355a2842ed.1774342371.git.zhengqi.arch%40bytedance.com
patch subject: [PATCH 3/3] mm: memcontrol: correct the nr_pages parameter type of mem_cgroup_update_lru_size()
config: riscv-allnoconfig-bpf (https://download.01.org/0day-ci/archive/20260326/[email protected]/config)
compiler: riscv64-linux-gnu-gcc (Debian 14.2.0-19) 14.2.0
reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260326/[email protected]/reproduce)

Thanks. Can't reproduce. (there are meds for this!)

Quoted 1 line

b4536f0c829c858 Michal Hocko 2017-01-10 @1486 "%s(%p, %d, %d): lru_size %ld\n",

We presently have

		"%s(%p, %d, %ld): lru_size %ld\n",

so it appears we fixed this recently.

#22 lore raw
In reply to Qi Zheng

-cc old mail

(just a gentle nudge to send stuff to [email protected] instead thanks :)

Quoted 33 lines · Qi Zheng

From: Qi Zheng <[email protected]>

Hi all,

As Harry Yoo pointed out [1], in the case of reparenting LRU folios, the value passed to functions such as __mod_memcg{_lruvec}_state() may exceed the upper limit of int, so this series aims to correct the parameter type of these functions.

This series is based on the next-20260323.

Comments and suggestions are welcome!

Thanks,
Qi

[1]. https://lore.kernel.org/all/acDxaEgnqPI-Z4be@hyeyoo/

Qi Zheng (3):
mm: memcontrol: correct the type of stats_updates to unsigned long
mm: memcontrol: correct the parameter type of

    __mod_memcg{_lruvec}_state()

mm: memcontrol: correct the nr_pages parameter type of
mem_cgroup_update_lru_size()

 include/linux/memcontrol.h   |  2 +-
 include/trace/events/memcg.h | 10 +++++-----
 mm/memcontrol.c              | 28 ++++++++++++++--------------
 3 files changed, 20 insertions(+), 20 deletions(-)

--
2.20.1