Skip to contentSkip to conversation

mm: KSM: fix data types

PATCH Superseded Zhansaya Bagdauletkyzy 3 participants 11 messages Jul 15, 2021 – Jul 16, 2021

Subsystems (2)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

cover mm: KSM: fix data types

Change data types of some variables as there were a few discrepancies between the types and actual values that were stored.

Zhansaya Bagdauletkyzy (2):
mm: KSM: fix ksm_run data type
mm: KSM: fix data type

 mm/ksm.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
Signature · 1 line
2.25.1
#1 lore raw

PATCH 1/2 mm: KSM: fix ksm_run data type

ksm_run is declared as unsigned long but in run_store(), it is converted to unsigned int. Change its type to unsigned int.

---
 mm/ksm.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
mm/ksm.c +2 -2
index 3fa9bc8a67cf..057d0c245bf4 100644
@@ -289,7 +289,7 @@ static int ksm_nr_node_ids = 1;
 #define KSM_RUN_MERGE	1
 #define KSM_RUN_UNMERGE	2
 #define KSM_RUN_OFFLINE	4
-static unsigned long ksm_run = KSM_RUN_STOP;
+static unsigned int ksm_run = KSM_RUN_STOP;
 static void wait_while_offlining(void);
 
 static DECLARE_WAIT_QUEUE_HEAD(ksm_thread_wait);
@@ -2874,7 +2874,7 @@ KSM_ATTR(pages_to_scan);
 static ssize_t run_show(struct kobject *kobj, struct kobj_attribute *attr,
 			char *buf)
 {
-	return sysfs_emit(buf, "%lu\n", ksm_run);
+	return sysfs_emit(buf, "%u\n", ksm_run);
 }
 
 static ssize_t run_store(struct kobject *kobj, struct kobj_attribute *attr,
Signature · 1 line
2.25.1
Reviewed-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] mm: KSM: fix ksm_run data type

Quoted 23 lines · Zhansaya Bagdauletkyzy

ksm_run is declared as unsigned long but in run_store(), it is converted to unsigned int. Change its type to unsigned int.

---
 mm/ksm.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
mm/ksm.c +1 -1
index 3fa9bc8a67cf..057d0c245bf4 100644
@@ -289,7 +289,7 @@ static int ksm_nr_node_ids = 1;
 #define KSM_RUN_MERGE  1
 #define KSM_RUN_UNMERGE        2
 #define KSM_RUN_OFFLINE        4
-static unsigned long ksm_run = KSM_RUN_STOP;
+static unsigned int ksm_run = KSM_RUN_STOP;
 static void wait_while_offlining(void);
 
 static DECLARE_WAIT_QUEUE_HEAD(ksm_thread_wait);

mm/ksm.c KSM_ATTR(pages_to_scan); · line 2877

Quoted 5 lines
mm/ksm.c +1 -1
@@ -2874,7 +2874,7 @@ KSM_ATTR(pages_to_scan);
 static ssize_t run_show(struct kobject *kobj, struct kobj_attribute *attr,
                        char *buf)
 {
-       return sysfs_emit(buf, "%lu\n", ksm_run);
+       return sysfs_emit(buf, "%u\n", ksm_run);
#3 lore raw

Re: [PATCH 1/2] mm: KSM: fix ksm_run data type

In reply to Zhansaya Bagdauletkyzy

mm/ksm.c static int ksm_nr_node_ids = 1; · line 292

Quoted 7 lines · Zhansaya Bagdauletkyzy

+++ b/mm/ksm.c

mm/ksm.c +1 -1
@@ -289,7 +289,7 @@ static int ksm_nr_node_ids = 1;
 #define KSM_RUN_MERGE	1
 #define KSM_RUN_UNMERGE	2
 #define KSM_RUN_OFFLINE	4
-static unsigned long ksm_run = KSM_RUN_STOP;
+static unsigned int ksm_run = KSM_RUN_STOP;

Should this be an enum instead?

#4 lore raw

Re: [PATCH 1/2] mm: KSM: fix ksm_run data type

Quoted 11 lines · Matthew Wilcox

mm/ksm.c static int ksm_nr_node_ids = 1; · line 292

Quoted 7 lines · Zhansaya Bagdauletkyzy

+++ b/mm/ksm.c

mm/ksm.c +1 -1
@@ -289,7 +289,7 @@ static int ksm_nr_node_ids = 1;
 #define KSM_RUN_MERGE        1
 #define KSM_RUN_UNMERGE      2
 #define KSM_RUN_OFFLINE      4
-static unsigned long ksm_run = KSM_RUN_STOP;
+static unsigned int ksm_run = KSM_RUN_STOP;

Should this be an enum instead?

I think "unsigned int" is OK here, as it is exposed as uint to users:
Documentation/ABI/testing/sysfs-kernel-mm-ksm

/sys/kernel/mm/ksm/run

run: write 0 to disable ksm, read 0 while ksm is disabled.

- write 1 to run ksm, read 1 while ksm is running.
- write 2 to disable ksm and unmerge all its pages.

Pasha

#5 lore raw

Re: [PATCH 1/2] mm: KSM: fix ksm_run data type

Quoted 16 lines · Pavel Tatashin
Quoted 11 lines · Matthew Wilcox

mm/ksm.c static int ksm_nr_node_ids = 1; · line 292

Quoted 7 lines · Zhansaya Bagdauletkyzy

+++ b/mm/ksm.c

mm/ksm.c +1 -1
@@ -289,7 +289,7 @@ static int ksm_nr_node_ids = 1;
 #define KSM_RUN_MERGE        1
 #define KSM_RUN_UNMERGE      2
 #define KSM_RUN_OFFLINE      4
-static unsigned long ksm_run = KSM_RUN_STOP;
+static unsigned int ksm_run = KSM_RUN_STOP;

Should this be an enum instead?

I think "unsigned int" is OK here, as it is exposed as uint to users:
Documentation/ABI/testing/sysfs-kernel-mm-ksm

Quoted 6 lines

/sys/kernel/mm/ksm/run

run: write 0 to disable ksm, read 0 while ksm is disabled.

- write 1 to run ksm, read 1 while ksm is running.
- write 2 to disable ksm and unmerge all its pages.

The document is out of date then as it does not mention 'offline'.

Also, why does the call to kstrtouint() specify base 10? If it is a bitmap, then permitting 0x [1] is more natural. I would expect to see base 0 there.

[1] or even 0b, although I see that _parse_integer_fixup_radix does not support the 0b notation.

#6 lore raw

Re: [PATCH 1/2] mm: KSM: fix ksm_run data type

Quoted 18 lines · Matthew Wilcox
Quoted 16 lines · Pavel Tatashin
Quoted 11 lines · Matthew Wilcox

mm/ksm.c static int ksm_nr_node_ids = 1; · line 292

Quoted 7 lines · Zhansaya Bagdauletkyzy

+++ b/mm/ksm.c

mm/ksm.c +1 -1
@@ -289,7 +289,7 @@ static int ksm_nr_node_ids = 1;
 #define KSM_RUN_MERGE        1
 #define KSM_RUN_UNMERGE      2
 #define KSM_RUN_OFFLINE      4
-static unsigned long ksm_run = KSM_RUN_STOP;
+static unsigned int ksm_run = KSM_RUN_STOP;

Should this be an enum instead?

I think "unsigned int" is OK here, as it is exposed as uint to users:
Documentation/ABI/testing/sysfs-kernel-mm-ksm

Quoted 8 lines
Quoted 6 lines

/sys/kernel/mm/ksm/run

run: write 0 to disable ksm, read 0 while ksm is disabled.

- write 1 to run ksm, read 1 while ksm is running.
- write 2 to disable ksm and unmerge all its pages.

The document is out of date then as it does not mention 'offline'.

The offline mode cannot be set externally.

In run_store()
if (flags > KSM_RUN_UNMERGE)

      return -EINVAL;
Quoted 4 lines

Also, why does the call to kstrtouint() specify base 10? If it is a bitmap, then permitting 0x [1] is more natural. I would expect to see base 0 there.

Users can only write 0, 1, or 2, it is not a bitmap from the user's perspective as the user cannot write: '3' . But, I think it is somewhat weird that ksm_run is used as a bitmap internally with KSM_RUN_OFFLINE = 4.

Imo, KSM_RUN_OFFLINE should be placed in a separate boolean from "ksm_run".

Quoted 3 lines

[1] or even 0b, although I see that _parse_integer_fixup_radix does not support the 0b notation.

#7 lore raw

Re: [PATCH 1/2] mm: KSM: fix ksm_run data type

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

/sys/kernel/mm/ksm/run

run: write 0 to disable ksm, read 0 while ksm is disabled.

- write 1 to run ksm, read 1 while ksm is running.
- write 2 to disable ksm and unmerge all its pages.

Quoted 7 lines
Quoted 1 line

The document is out of date then as it does not mention 'offline'.

The offline mode cannot be set externally.

In run_store()
if (flags > KSM_RUN_UNMERGE)

      return -EINVAL;

However, I see that KSM_RUN_OFFLINE is visible via run_show(), so yes doc should be updated. And, yes it becomes a bitfield from the user perspective.

Quoted 16 lines
Quoted 5 lines

Also, why does the call to kstrtouint() specify base 10? If it is a bitmap, then permitting 0x [1] is more natural. I would expect to see base 0 there.

Users can only write 0, 1, or 2, it is not a bitmap from the user's perspective as the user cannot write: '3' . But, I think it is somewhat weird that ksm_run is used as a bitmap internally with KSM_RUN_OFFLINE = 4.

Imo, KSM_RUN_OFFLINE should be placed in a separate boolean from "ksm_run".

Quoted 3 lines

[1] or even 0b, although I see that _parse_integer_fixup_radix does not support the 0b notation.

#8 lore raw

PATCH 2/2 mm: KSM: fix data type

In reply to Zhansaya Bagdauletkyzy

ksm_stable_node_chains_prune_millisecs is declared as int, but in stable__node_chains_prune_millisecs_store(), it can store values up to UINT_MAX. Change the variable type to unsigned int.

---
 mm/ksm.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
mm/ksm.c +4 -4
index 057d0c245bf4..2e4bd7662e52 100644
@@ -259,7 +259,7 @@ static unsigned long ksm_stable_node_chains;
 static unsigned long ksm_stable_node_dups;
 
 /* Delay in pruning stale stable_node_dups in the stable_node_chains */
-static int ksm_stable_node_chains_prune_millisecs = 2000;
+static unsigned int ksm_stable_node_chains_prune_millisecs = 2000;
 
 /* Maximum number of page slots sharing a stable node */
 static int ksm_max_page_sharing = 256;
@@ -3105,11 +3105,11 @@ stable_node_chains_prune_millisecs_store(struct kobject *kobj,
 					 struct kobj_attribute *attr,
 					 const char *buf, size_t count)
 {
-	unsigned long msecs;
+	unsigned int msecs;
 	int err;
 
-	err = kstrtoul(buf, 10, &msecs);
-	if (err || msecs > UINT_MAX)
+	err = kstrtouint(buf, 10, &msecs);
+	if (err)
 		return -EINVAL;
 
 	ksm_stable_node_chains_prune_millisecs = msecs;
Signature · 1 line
2.25.1
Reviewed-by on v1 2/2
#9 lore raw

Re: [PATCH 2/2] mm: KSM: fix data type

Quoted 28 lines · Zhansaya Bagdauletkyzy

ksm_stable_node_chains_prune_millisecs is declared as int, but in stable__node_chains_prune_millisecs_store(), it can store values up to UINT_MAX. Change the variable type to unsigned int.

---
 mm/ksm.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
mm/ksm.c +1 -2
index 057d0c245bf4..2e4bd7662e52 100644
@@ -259,7 +259,7 @@ static unsigned long ksm_stable_node_chains;
 static unsigned long ksm_stable_node_dups;
 
 /* Delay in pruning stale stable_node_dups in the stable_node_chains */
-static int ksm_stable_node_chains_prune_millisecs = 2000;
+static unsigned int ksm_stable_node_chains_prune_millisecs = 2000;
 
 /* Maximum number of page slots sharing a stable node */
 static int ksm_max_page_sharing = 256;
@@ -3105,11 +3105,11 @@ stable_node_chains_prune_millisecs_store(struct kobject *kobj,
                                         struct kobj_attribute *attr,
                                         const char *buf, size_t count)
 {
-       unsigned long msecs;

mm/ksm.c stable_node_chains_prune_millisecs_store(struct kobject *kobj, · line 3115

Quoted 10 lines
mm/ksm.c +3 -2
+       unsigned int msecs;
        int err;
 
-       err = kstrtoul(buf, 10, &msecs);
-       if (err || msecs > UINT_MAX)
+       err = kstrtouint(buf, 10, &msecs);
+       if (err)
                return -EINVAL;
 
        ksm_stable_node_chains_prune_millisecs = msecs;

LGTM, but I would merge the two patches together. They both update types of sysfs tunnables in the same file.

Quoted 3 lines

--
2.25.1

#10 lore raw

Re: [PATCH 2/2] mm: KSM: fix data type

Quoted 39 lines · Pavel Tatashin
Quoted 37 lines · Zhansaya Bagdauletkyzy

ksm_stable_node_chains_prune_millisecs is declared as int, but in stable__node_chains_prune_millisecs_store(), it can store values up to UINT_MAX. Change the variable type to unsigned int.

---
 mm/ksm.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
mm/ksm.c +4 -4
index 057d0c245bf4..2e4bd7662e52 100644
@@ -259,7 +259,7 @@ static unsigned long ksm_stable_node_chains;
 static unsigned long ksm_stable_node_dups;
 
 /* Delay in pruning stale stable_node_dups in the stable_node_chains */
-static int ksm_stable_node_chains_prune_millisecs = 2000;
+static unsigned int ksm_stable_node_chains_prune_millisecs = 2000;
 
 /* Maximum number of page slots sharing a stable node */
 static int ksm_max_page_sharing = 256;
@@ -3105,11 +3105,11 @@ stable_node_chains_prune_millisecs_store(struct kobject *kobj,
                                         struct kobj_attribute *attr,
                                         const char *buf, size_t count)
 {
-       unsigned long msecs;
+       unsigned int msecs;
        int err;
 
-       err = kstrtoul(buf, 10, &msecs);
-       if (err || msecs > UINT_MAX)
+       err = kstrtouint(buf, 10, &msecs);
+       if (err)
                return -EINVAL;
 
Quoted 4 lines

mm/ksm.c stable_node_chains_prune_millisecs_store(struct kobject *kobj, · line 3115

Quoted 1 line
mm/ksm.c +0 -0
        ksm_stable_node_chains_prune_millisecs = msecs;

LGTM, but I would merge the two patches together. They both update types of sysfs tunnables in the same file.

Ok, I'll send v2 as a single patch.

Thanks,
Zhansaya