[PATCH v3] mm/ksm: introduce ksm_force for each process

[PATCH v3] mm/ksm: introduce ksm_force for each process

[cgel.zte]

From: xu xin <xu.xin16@zte.com.cn>

To use KSM, we must explicitly call madvise() in application code,
which means installed apps on OS needs to be uninstall and source
code needs to be modified. It is inconvenient.

In order to change this situation, We add a new proc 'ksm_force'
under /proc/<pid>/ to support turning on/off KSM scanning of a
process's mm dynamically.

If ksm_force is set as 1, force all anonymous and 'qualified' vma
of this mm to be involved in KSM scanning without explicitly
calling madvise to make vma MADV_MERGEABLE. But It is effctive only
when the klob of '/sys/kernel/mm/ksm/run' is set as 1.

If ksm_enale is set as 0, cancel the feature of ksm_force of this
process and unmerge those merged pages which is not madvised as
MERGEABLE of this process, but leave MERGEABLE areas merged.

Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Reviewed-by: Ran Xiaokai <ran.xiaokai@zte.com.cn>
Reviewed-by: wangyong <wang.yong12@zte.com.cn>
Reviewed-by: Yunkai Zhang <zhang.yunkai@zte.com.cn>
---
v3:
- fix compile error of mm/ksm.c
v2:
- fix a spelling error in commit log.
- remove a redundant condition check in ksm_force_write().
---
 fs/proc/base.c           | 99 ++++++++++++++++++++++++++++++++++++++++
 include/linux/mm_types.h |  9 ++++
 mm/ksm.c                 | 32 ++++++++++++-
 3 files changed, 138 insertions(+), 2 deletions(-)

diff --git a/fs/proc/base.c b/fs/proc/base.c
index 8dfa36a99c74..3115ffa4c9fb 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -96,6 +96,7 @@
 #include <linux/time_namespace.h>
 #include <linux/resctrl.h>
 #include <linux/cn_proc.h>
+#include <linux/ksm.h>
 #include <trace/events/oom.h>
 #include "internal.h"
 #include "fd.h"
@@ -3168,6 +3169,102 @@ static int proc_pid_ksm_merging_pages(struct seq_file *m, struct pid_namespace *
 
 	return 0;
 }
+
+static ssize_t ksm_force_read(struct file *file, char __user *buf, size_t count,
+				loff_t *ppos)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	char buffer[PROC_NUMBUF];
+	ssize_t len;
+	int ret;
+
+	task = get_proc_task(file_inode(file));
+	if (!task)
+		return -ESRCH;
+
+	mm = get_task_mm(task);
+	ret = 0;
+	if (mm) {
+		len = snprintf(buffer, sizeof(buffer), "%d\n", mm->ksm_force);
+		ret =  simple_read_from_buffer(buf, count, ppos, buffer, len);
+		mmput(mm);
+	}
+
+	return ret;
+}
+
+static ssize_t ksm_force_write(struct file *file, const char __user *buf,
+				size_t count, loff_t *ppos)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	char buffer[PROC_NUMBUF];
+	int force;
+	int err = 0;
+
+	memset(buffer, 0, sizeof(buffer));
+	if (count > sizeof(buffer) - 1)
+		count = sizeof(buffer) - 1;
+	if (copy_from_user(buffer, buf, count)) {
+		err = -EFAULT;
+		goto out_return;
+	}
+
+	err = kstrtoint(strstrip(buffer), 0, &force);
+
+	if (err)
+		goto out_return;
+	if (force != 0 && force != 1) {
+		err = -EINVAL;
+		goto out_return;
+	}
+
+	task = get_proc_task(file_inode(file));
+	if (!task) {
+		err = -ESRCH;
+		goto out_return;
+	}
+
+	mm = get_task_mm(task);
+	if (!mm)
+		goto out_put_task;
+
+	if (mm->ksm_force != force) {
+		if (mmap_write_lock_killable(mm)) {
+			err = -EINTR;
+			goto out_mmput;
+		}
+
+		if (force == 0)
+			mm->ksm_force = force;
+		else {
+			/*
+			 * Force anonymous pages of this mm to be involved in KSM merging
+			 * without explicitly calling madvise.
+			 */
+			if (!test_bit(MMF_VM_MERGEABLE, &mm->flags))
+				err = __ksm_enter(mm);
+			if (!err)
+				mm->ksm_force = force;
+		}
+
+		mmap_write_unlock(mm);
+	}
+
+out_mmput:
+	mmput(mm);
+out_put_task:
+	put_task_struct(task);
+out_return:
+	return err < 0 ? err : count;
+}
+
+static const struct file_operations proc_pid_ksm_force_operations = {
+	.read		= ksm_force_read,
+	.write		= ksm_force_write,
+	.llseek		= generic_file_llseek,
+};
 #endif /* CONFIG_KSM */
 
 #ifdef CONFIG_STACKLEAK_METRICS
@@ -3303,6 +3400,7 @@ static const struct pid_entry tgid_base_stuff[] = {
 #endif
 #ifdef CONFIG_KSM
 	ONE("ksm_merging_pages",  S_IRUSR, proc_pid_ksm_merging_pages),
+	REG("ksm_force", S_IRUSR|S_IWUSR, proc_pid_ksm_force_operations),
 #endif
 };
 
@@ -3639,6 +3737,7 @@ static const struct pid_entry tid_base_stuff[] = {
 #endif
 #ifdef CONFIG_KSM
 	ONE("ksm_merging_pages",  S_IRUSR, proc_pid_ksm_merging_pages),
+	REG("ksm_force", S_IRUSR|S_IWUSR, proc_pid_ksm_force_operations),
 #endif
 };
 
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index b34ff2cdbc4f..1b1592c2f5cf 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -661,6 +661,15 @@ struct mm_struct {
 		 * merging.
 		 */
 		unsigned long ksm_merging_pages;
+		/*
+		 * If true, force anonymous pages of this mm to be involved in KSM
+		 * merging without explicitly calling madvise. It is effctive only
+		 * when the klob of '/sys/kernel/mm/ksm/run' is set as 1. If false,
+		 * cancel the feature of ksm_force of this process and unmerge
+		 * those merged pages which is not madvised as MERGEABLE of this
+		 * process, but leave MERGEABLE areas merged.
+		 */
+		bool ksm_force;
 #endif
 	} __randomize_layout;
 
diff --git a/mm/ksm.c b/mm/ksm.c
index 38360285497a..c9f672dcc72e 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -334,6 +334,34 @@ static void __init ksm_slab_free(void)
 	mm_slot_cache = NULL;
 }
 
+/* Check if vma is qualified for ksmd scanning */
+static bool ksm_vma_check(struct vm_area_struct *vma)
+{
+	unsigned long vm_flags = vma->vm_flags;
+
+	if (!(vma->vm_flags & VM_MERGEABLE) && !(vma->vm_mm->ksm_force))
+		return false;
+
+	if (vm_flags & (VM_SHARED	| VM_MAYSHARE	|
+			VM_PFNMAP	| VM_IO | VM_DONTEXPAND |
+			VM_HUGETLB	| VM_MIXEDMAP))
+		return false;       /* just ignore this vma*/
+
+	if (vma_is_dax(vma))
+		return false;
+
+#ifdef VM_SAO
+	if (vm_flags & VM_SAO)
+		return false;
+#endif
+#ifdef VM_SPARC_ADI
+	if (vm_flags & VM_SPARC_ADI)
+		return false;
+#endif
+
+	return true;
+}
+
 static __always_inline bool is_stable_node_chain(struct stable_node *chain)
 {
 	return chain->rmap_hlist_len == STABLE_NODE_CHAIN;
@@ -523,7 +551,7 @@ static struct vm_area_struct *find_mergeable_vma(struct mm_struct *mm,
 	if (ksm_test_exit(mm))
 		return NULL;
 	vma = vma_lookup(mm, addr);
-	if (!vma || !(vma->vm_flags & VM_MERGEABLE) || !vma->anon_vma)
+	if (!vma || !ksm_vma_check(vma) || !vma->anon_vma)
 		return NULL;
 	return vma;
 }
@@ -2297,7 +2325,7 @@ static struct rmap_item *scan_get_next_rmap_item(struct page **page)
 		vma = find_vma(mm, ksm_scan.address);
 
 	for (; vma; vma = vma->vm_next) {
-		if (!(vma->vm_flags & VM_MERGEABLE))
+		if (!ksm_vma_check(vma))
 			continue;
 		if (ksm_scan.address < vma->vm_start)
 			ksm_scan.address = vma->vm_start;
-- 
2.25.1

Re: [PATCH v3] mm/ksm: introduce ksm_force for each process

[Andrew Morton]

On Sat,  7 May 2022 05:47:02 +0000 cgel.zte@gmail.com wrote:

> To use KSM, we must explicitly call madvise() in application code,
> which means installed apps on OS needs to be uninstall and source
> code needs to be modified. It is inconvenient.
> 
> In order to change this situation, We add a new proc 'ksm_force'
> under /proc/<pid>/ to support turning on/off KSM scanning of a
> process's mm dynamically.
> 
> If ksm_force is set as 1, force all anonymous and 'qualified' vma
> of this mm to be involved in KSM scanning without explicitly
> calling madvise to make vma MADV_MERGEABLE. But It is effctive only
> when the klob of '/sys/kernel/mm/ksm/run' is set as 1.
> 
> If ksm_enale is set as 0, cancel the feature of ksm_force of this
> process and unmerge those merged pages which is not madvised as
> MERGEABLE of this process, but leave MERGEABLE areas merged.
> 

There are quite a lot of typos here.  Please proof-read it.

>  fs/proc/base.c           | 99 ++++++++++++++++++++++++++++++++++++++++
>  include/linux/mm_types.h |  9 ++++
>  mm/ksm.c                 | 32 ++++++++++++-

And please update the appropriate places under Documentation/ - all
user-facing interfaces should be well documented.

[PATCH v4] mm/ksm: introduce ksm_force for each process

[cgel.zte]

From: xu xin <xu.xin16@zte.com.cn>

To use KSM, we have to explicitly call madvise() in application code,
which means installed apps on OS needs to be uninstall and source code
needs to be modified. It is inconvenient.

In order to change this situation, We add a new proc file ksm_force
under /proc/<pid>/ to support turning on/off KSM scanning of a
process's mm dynamically.

If ksm_force is set to 1, force all anonymous and 'qualified' VMAs
of this mm to be involved in KSM scanning without explicitly calling
madvise to mark VMA as MADV_MERGEABLE. But It is effective only when
the klob of /sys/kernel/mm/ksm/run is set as 1.

If ksm_force is set to 0, cancel the feature of ksm_force of this
process and unmerge those merged pages belonging to VMAs which is not
madvised as MADV_MERGEABLE of this process, but leave MADV_MERGEABLE
areas merged.

Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Reviewed-by: Ran Xiaokai <ran.xiaokai@zte.com.cn>
Reviewed-by: wangyong <wang.yong12@zte.com.cn>
Reviewed-by: Yunkai Zhang <zhang.yunkai@zte.com.cn>
---
v4:
- fix typos in commit log
- add interface descriptions under Documentation/
v3:
- fix compile error of mm/ksm.c
v2:
- fix a spelling error in commit log.
- remove a redundant condition check in ksm_force_write().
---
 Documentation/admin-guide/mm/ksm.rst | 20 +++++-
 Documentation/filesystems/proc.rst   | 17 +++++
 fs/proc/base.c                       | 99 ++++++++++++++++++++++++++++
 include/linux/mm_types.h             |  9 +++
 mm/ksm.c                             | 32 ++++++++-
 5 files changed, 174 insertions(+), 3 deletions(-)

diff --git a/Documentation/admin-guide/mm/ksm.rst b/Documentation/admin-guide/mm/ksm.rst
index b244f0202a03..e42cffa42463 100644
--- a/Documentation/admin-guide/mm/ksm.rst
+++ b/Documentation/admin-guide/mm/ksm.rst
@@ -32,7 +32,7 @@ are swapped back in: ksmd must rediscover their identity and merge again).
 Controlling KSM with madvise
 ============================
 
-KSM only operates on those areas of address space which an application
+KSM can operates on those areas of address space which an application
 has advised to be likely candidates for merging, by using the madvise(2)
 system call::
 
@@ -70,6 +70,24 @@ Applications should be considerate in their use of MADV_MERGEABLE,
 restricting its use to areas likely to benefit.  KSM's scans may use a lot
 of processing power: some installations will disable KSM for that reason.
 
+Controlling KSM with procfs
+===========================
+
+KSM can also operate on anonymous areas of address space of those processes's
+knob ``/proc/<pid>/ksm_force`` is on, even if app codes doesn't call madvise()
+explicitly to advise specific areas as MADV_MERGEABLE.
+
+You can set ksm_force to 1 to force all anonymous and qualified VMAs of
+this process to be involved in KSM scanning. But It is effective only when the
+klob of ``/sys/kernel/mm/ksm/run`` is set as 1.
+	e.g. ``echo 1 > /proc/<pid>/ksm_force``
+
+You can also set ksm_force to 0 to cancel that force feature of this process
+and unmerge those merged pages which belongs to those VMAs not marked as
+MADV_MERGEABLE of this process. But that will leave those pages belonging to
+VMAs marked as MADV_MERGEABLE merged.
+	e.g. ``echo 0 > /proc/<pid>/ksm_force``
+
 .. _ksm_sysfs:
 
 KSM daemon sysfs interface
diff --git a/Documentation/filesystems/proc.rst b/Documentation/filesystems/proc.rst
index 061744c436d9..414977885e6e 100644
--- a/Documentation/filesystems/proc.rst
+++ b/Documentation/filesystems/proc.rst
@@ -47,6 +47,7 @@ fixes/update part 1.1  Stefani Seibold <stefani@seibold.net>    June 9 2009
   3.10  /proc/<pid>/timerslack_ns - Task timerslack value
   3.11	/proc/<pid>/patch_state - Livepatch patch operation state
   3.12	/proc/<pid>/arch_status - Task architecture specific information
+  3.13	/proc/<pid>/ksm_force - Setting of mandatory involvement in KSM
 
   4	Configuring procfs
   4.1	Mount options
@@ -2176,6 +2177,22 @@ AVX512_elapsed_ms
   the task is unlikely an AVX512 user, but depends on the workload and the
   scheduling scenario, it also could be a false negative mentioned above.
 
+3.11    /proc/<pid>/ksm_force - Setting of mandatory involvement in KSM
+-----------------------------------------------------------------------
+When CONFIG_KSM is enabled, this file can be used to specify if this
+process's anonymous memory can be involved in KSM scanning without app codes
+explicitly calling madvise to mark memory address as MADV_MERGEABLE.
+
+If writing 1 to this file, the kernel will force all anonymous and qualified
+memory to be involved in KSM scanning without explicitly calling madvise to
+mark memory address as MADV_MERGEABLE. But that is effective only when the
+klob of '/sys/kernel/mm/ksm/run' is set as 1.
+
+If writing 0 to this file, the mandatory KSM feature of this process's will
+be cancelled and unmerge those merged pages which belongs to those areas not
+marked as MADV_MERGEABLE of this process, but leave those pages belonging to
+areas marked as MADV_MERGEABLE merged.
+
 Chapter 4: Configuring procfs
 =============================
 
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 8dfa36a99c74..3115ffa4c9fb 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -96,6 +96,7 @@
 #include <linux/time_namespace.h>
 #include <linux/resctrl.h>
 #include <linux/cn_proc.h>
+#include <linux/ksm.h>
 #include <trace/events/oom.h>
 #include "internal.h"
 #include "fd.h"
@@ -3168,6 +3169,102 @@ static int proc_pid_ksm_merging_pages(struct seq_file *m, struct pid_namespace *
 
 	return 0;
 }
+
+static ssize_t ksm_force_read(struct file *file, char __user *buf, size_t count,
+				loff_t *ppos)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	char buffer[PROC_NUMBUF];
+	ssize_t len;
+	int ret;
+
+	task = get_proc_task(file_inode(file));
+	if (!task)
+		return -ESRCH;
+
+	mm = get_task_mm(task);
+	ret = 0;
+	if (mm) {
+		len = snprintf(buffer, sizeof(buffer), "%d\n", mm->ksm_force);
+		ret =  simple_read_from_buffer(buf, count, ppos, buffer, len);
+		mmput(mm);
+	}
+
+	return ret;
+}
+
+static ssize_t ksm_force_write(struct file *file, const char __user *buf,
+				size_t count, loff_t *ppos)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	char buffer[PROC_NUMBUF];
+	int force;
+	int err = 0;
+
+	memset(buffer, 0, sizeof(buffer));
+	if (count > sizeof(buffer) - 1)
+		count = sizeof(buffer) - 1;
+	if (copy_from_user(buffer, buf, count)) {
+		err = -EFAULT;
+		goto out_return;
+	}
+
+	err = kstrtoint(strstrip(buffer), 0, &force);
+
+	if (err)
+		goto out_return;
+	if (force != 0 && force != 1) {
+		err = -EINVAL;
+		goto out_return;
+	}
+
+	task = get_proc_task(file_inode(file));
+	if (!task) {
+		err = -ESRCH;
+		goto out_return;
+	}
+
+	mm = get_task_mm(task);
+	if (!mm)
+		goto out_put_task;
+
+	if (mm->ksm_force != force) {
+		if (mmap_write_lock_killable(mm)) {
+			err = -EINTR;
+			goto out_mmput;
+		}
+
+		if (force == 0)
+			mm->ksm_force = force;
+		else {
+			/*
+			 * Force anonymous pages of this mm to be involved in KSM merging
+			 * without explicitly calling madvise.
+			 */
+			if (!test_bit(MMF_VM_MERGEABLE, &mm->flags))
+				err = __ksm_enter(mm);
+			if (!err)
+				mm->ksm_force = force;
+		}
+
+		mmap_write_unlock(mm);
+	}
+
+out_mmput:
+	mmput(mm);
+out_put_task:
+	put_task_struct(task);
+out_return:
+	return err < 0 ? err : count;
+}
+
+static const struct file_operations proc_pid_ksm_force_operations = {
+	.read		= ksm_force_read,
+	.write		= ksm_force_write,
+	.llseek		= generic_file_llseek,
+};
 #endif /* CONFIG_KSM */
 
 #ifdef CONFIG_STACKLEAK_METRICS
@@ -3303,6 +3400,7 @@ static const struct pid_entry tgid_base_stuff[] = {
 #endif
 #ifdef CONFIG_KSM
 	ONE("ksm_merging_pages",  S_IRUSR, proc_pid_ksm_merging_pages),
+	REG("ksm_force", S_IRUSR|S_IWUSR, proc_pid_ksm_force_operations),
 #endif
 };
 
@@ -3639,6 +3737,7 @@ static const struct pid_entry tid_base_stuff[] = {
 #endif
 #ifdef CONFIG_KSM
 	ONE("ksm_merging_pages",  S_IRUSR, proc_pid_ksm_merging_pages),
+	REG("ksm_force", S_IRUSR|S_IWUSR, proc_pid_ksm_force_operations),
 #endif
 };
 
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index b34ff2cdbc4f..1b1592c2f5cf 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -661,6 +661,15 @@ struct mm_struct {
 		 * merging.
 		 */
 		unsigned long ksm_merging_pages;
+		/*
+		 * If true, force anonymous pages of this mm to be involved in KSM
+		 * merging without explicitly calling madvise. It is effctive only
+		 * when the klob of '/sys/kernel/mm/ksm/run' is set as 1. If false,
+		 * cancel the feature of ksm_force of this process and unmerge
+		 * those merged pages which is not madvised as MERGEABLE of this
+		 * process, but leave MERGEABLE areas merged.
+		 */
+		bool ksm_force;
 #endif
 	} __randomize_layout;
 
diff --git a/mm/ksm.c b/mm/ksm.c
index 38360285497a..c9f672dcc72e 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -334,6 +334,34 @@ static void __init ksm_slab_free(void)
 	mm_slot_cache = NULL;
 }
 
+/* Check if vma is qualified for ksmd scanning */
+static bool ksm_vma_check(struct vm_area_struct *vma)
+{
+	unsigned long vm_flags = vma->vm_flags;
+
+	if (!(vma->vm_flags & VM_MERGEABLE) && !(vma->vm_mm->ksm_force))
+		return false;
+
+	if (vm_flags & (VM_SHARED	| VM_MAYSHARE	|
+			VM_PFNMAP	| VM_IO | VM_DONTEXPAND |
+			VM_HUGETLB	| VM_MIXEDMAP))
+		return false;       /* just ignore this vma*/
+
+	if (vma_is_dax(vma))
+		return false;
+
+#ifdef VM_SAO
+	if (vm_flags & VM_SAO)
+		return false;
+#endif
+#ifdef VM_SPARC_ADI
+	if (vm_flags & VM_SPARC_ADI)
+		return false;
+#endif
+
+	return true;
+}
+
 static __always_inline bool is_stable_node_chain(struct stable_node *chain)
 {
 	return chain->rmap_hlist_len == STABLE_NODE_CHAIN;
@@ -523,7 +551,7 @@ static struct vm_area_struct *find_mergeable_vma(struct mm_struct *mm,
 	if (ksm_test_exit(mm))
 		return NULL;
 	vma = vma_lookup(mm, addr);
-	if (!vma || !(vma->vm_flags & VM_MERGEABLE) || !vma->anon_vma)
+	if (!vma || !ksm_vma_check(vma) || !vma->anon_vma)
 		return NULL;
 	return vma;
 }
@@ -2297,7 +2325,7 @@ static struct rmap_item *scan_get_next_rmap_item(struct page **page)
 		vma = find_vma(mm, ksm_scan.address);
 
 	for (; vma; vma = vma->vm_next) {
-		if (!(vma->vm_flags & VM_MERGEABLE))
+		if (!ksm_vma_check(vma))
 			continue;
 		if (ksm_scan.address < vma->vm_start)
 			ksm_scan.address = vma->vm_start;
-- 
2.25.1

[PATCH v5] mm/ksm: introduce ksm_force for each process

[cgel.zte]

From: xu xin <xu.xin16@zte.com.cn>

To use KSM, we have to explicitly call madvise() in application code,
which means installed apps on OS needs to be uninstall and source code
needs to be modified. It is inconvenient.

In order to change this situation, We add a new proc file ksm_force
under /proc/<pid>/ to support turning on/off KSM scanning of a
process's mm dynamically.

If ksm_force is set to 1, force all anonymous and 'qualified' VMAs
of this mm to be involved in KSM scanning without explicitly calling
madvise to mark VMA as MADV_MERGEABLE. But It is effective only when
the klob of /sys/kernel/mm/ksm/run is set as 1.

If ksm_force is set to 0, cancel the feature of ksm_force of this
process and unmerge those merged pages belonging to VMAs which is not
madvised as MADV_MERGEABLE of this process, but leave MADV_MERGEABLE
areas merged.

Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Reviewed-by: Ran Xiaokai <ran.xiaokai@zte.com.cn>
Reviewed-by: wangyong <wang.yong12@zte.com.cn>
Reviewed-by: Yunkai Zhang <zhang.yunkai@zte.com.cn>
---
v5:
- fix typos in Documentation/filesystems/proc.rst
v4:
- fix typos in commit log
- add interface descriptions under Documentation/
v3:
- fix compile error of mm/ksm.c
v2:
- fix a spelling error in commit log.
- remove a redundant condition check in ksm_force_write().
---
 Documentation/admin-guide/mm/ksm.rst | 20 +++++-
 Documentation/filesystems/proc.rst   | 17 +++++
 fs/proc/base.c                       | 99 ++++++++++++++++++++++++++++
 include/linux/mm_types.h             |  9 +++
 mm/ksm.c                             | 32 ++++++++-
 5 files changed, 174 insertions(+), 3 deletions(-)

diff --git a/Documentation/admin-guide/mm/ksm.rst b/Documentation/admin-guide/mm/ksm.rst
index b244f0202a03..e42cffa42463 100644
--- a/Documentation/admin-guide/mm/ksm.rst
+++ b/Documentation/admin-guide/mm/ksm.rst
@@ -32,7 +32,7 @@ are swapped back in: ksmd must rediscover their identity and merge again).
 Controlling KSM with madvise
 ============================
 
-KSM only operates on those areas of address space which an application
+KSM can operates on those areas of address space which an application
 has advised to be likely candidates for merging, by using the madvise(2)
 system call::
 
@@ -70,6 +70,24 @@ Applications should be considerate in their use of MADV_MERGEABLE,
 restricting its use to areas likely to benefit.  KSM's scans may use a lot
 of processing power: some installations will disable KSM for that reason.
 
+Controlling KSM with procfs
+===========================
+
+KSM can also operate on anonymous areas of address space of those processes's
+knob ``/proc/<pid>/ksm_force`` is on, even if app codes doesn't call madvise()
+explicitly to advise specific areas as MADV_MERGEABLE.
+
+You can set ksm_force to 1 to force all anonymous and qualified VMAs of
+this process to be involved in KSM scanning. But It is effective only when the
+klob of ``/sys/kernel/mm/ksm/run`` is set as 1.
+	e.g. ``echo 1 > /proc/<pid>/ksm_force``
+
+You can also set ksm_force to 0 to cancel that force feature of this process
+and unmerge those merged pages which belongs to those VMAs not marked as
+MADV_MERGEABLE of this process. But that will leave those pages belonging to
+VMAs marked as MADV_MERGEABLE merged.
+	e.g. ``echo 0 > /proc/<pid>/ksm_force``
+
 .. _ksm_sysfs:
 
 KSM daemon sysfs interface
diff --git a/Documentation/filesystems/proc.rst b/Documentation/filesystems/proc.rst
index 061744c436d9..8f959697ae1e 100644
--- a/Documentation/filesystems/proc.rst
+++ b/Documentation/filesystems/proc.rst
@@ -47,6 +47,7 @@ fixes/update part 1.1  Stefani Seibold <stefani@seibold.net>    June 9 2009
   3.10  /proc/<pid>/timerslack_ns - Task timerslack value
   3.11	/proc/<pid>/patch_state - Livepatch patch operation state
   3.12	/proc/<pid>/arch_status - Task architecture specific information
+  3.13	/proc/<pid>/ksm_force - Setting of mandatory involvement in KSM
 
   4	Configuring procfs
   4.1	Mount options
@@ -2176,6 +2177,22 @@ AVX512_elapsed_ms
   the task is unlikely an AVX512 user, but depends on the workload and the
   scheduling scenario, it also could be a false negative mentioned above.
 
+3.13	/proc/<pid>/ksm_force - Setting of mandatory involvement in KSM
+-----------------------------------------------------------------------
+When CONFIG_KSM is enabled, this file can be used to specify if this
+process's anonymous memory can be involved in KSM scanning without app codes
+explicitly calling madvise to mark memory address as MADV_MERGEABLE.
+
+If writing 1 to this file, the kernel will force all anonymous and qualified
+memory to be involved in KSM scanning without explicitly calling madvise to
+mark memory address as MADV_MERGEABLE. But that is effective only when the
+klob of '/sys/kernel/mm/ksm/run' is set as 1.
+
+If writing 0 to this file, the mandatory KSM feature of this process's will
+be cancelled and unmerge those merged pages which belongs to those areas not
+marked as MADV_MERGEABLE of this process, but leave those pages belonging to
+areas marked as MADV_MERGEABLE merged.
+
 Chapter 4: Configuring procfs
 =============================
 
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 8dfa36a99c74..3115ffa4c9fb 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -96,6 +96,7 @@
 #include <linux/time_namespace.h>
 #include <linux/resctrl.h>
 #include <linux/cn_proc.h>
+#include <linux/ksm.h>
 #include <trace/events/oom.h>
 #include "internal.h"
 #include "fd.h"
@@ -3168,6 +3169,102 @@ static int proc_pid_ksm_merging_pages(struct seq_file *m, struct pid_namespace *
 
 	return 0;
 }
+
+static ssize_t ksm_force_read(struct file *file, char __user *buf, size_t count,
+				loff_t *ppos)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	char buffer[PROC_NUMBUF];
+	ssize_t len;
+	int ret;
+
+	task = get_proc_task(file_inode(file));
+	if (!task)
+		return -ESRCH;
+
+	mm = get_task_mm(task);
+	ret = 0;
+	if (mm) {
+		len = snprintf(buffer, sizeof(buffer), "%d\n", mm->ksm_force);
+		ret =  simple_read_from_buffer(buf, count, ppos, buffer, len);
+		mmput(mm);
+	}
+
+	return ret;
+}
+
+static ssize_t ksm_force_write(struct file *file, const char __user *buf,
+				size_t count, loff_t *ppos)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	char buffer[PROC_NUMBUF];
+	int force;
+	int err = 0;
+
+	memset(buffer, 0, sizeof(buffer));
+	if (count > sizeof(buffer) - 1)
+		count = sizeof(buffer) - 1;
+	if (copy_from_user(buffer, buf, count)) {
+		err = -EFAULT;
+		goto out_return;
+	}
+
+	err = kstrtoint(strstrip(buffer), 0, &force);
+
+	if (err)
+		goto out_return;
+	if (force != 0 && force != 1) {
+		err = -EINVAL;
+		goto out_return;
+	}
+
+	task = get_proc_task(file_inode(file));
+	if (!task) {
+		err = -ESRCH;
+		goto out_return;
+	}
+
+	mm = get_task_mm(task);
+	if (!mm)
+		goto out_put_task;
+
+	if (mm->ksm_force != force) {
+		if (mmap_write_lock_killable(mm)) {
+			err = -EINTR;
+			goto out_mmput;
+		}
+
+		if (force == 0)
+			mm->ksm_force = force;
+		else {
+			/*
+			 * Force anonymous pages of this mm to be involved in KSM merging
+			 * without explicitly calling madvise.
+			 */
+			if (!test_bit(MMF_VM_MERGEABLE, &mm->flags))
+				err = __ksm_enter(mm);
+			if (!err)
+				mm->ksm_force = force;
+		}
+
+		mmap_write_unlock(mm);
+	}
+
+out_mmput:
+	mmput(mm);
+out_put_task:
+	put_task_struct(task);
+out_return:
+	return err < 0 ? err : count;
+}
+
+static const struct file_operations proc_pid_ksm_force_operations = {
+	.read		= ksm_force_read,
+	.write		= ksm_force_write,
+	.llseek		= generic_file_llseek,
+};
 #endif /* CONFIG_KSM */
 
 #ifdef CONFIG_STACKLEAK_METRICS
@@ -3303,6 +3400,7 @@ static const struct pid_entry tgid_base_stuff[] = {
 #endif
 #ifdef CONFIG_KSM
 	ONE("ksm_merging_pages",  S_IRUSR, proc_pid_ksm_merging_pages),
+	REG("ksm_force", S_IRUSR|S_IWUSR, proc_pid_ksm_force_operations),
 #endif
 };
 
@@ -3639,6 +3737,7 @@ static const struct pid_entry tid_base_stuff[] = {
 #endif
 #ifdef CONFIG_KSM
 	ONE("ksm_merging_pages",  S_IRUSR, proc_pid_ksm_merging_pages),
+	REG("ksm_force", S_IRUSR|S_IWUSR, proc_pid_ksm_force_operations),
 #endif
 };
 
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index b34ff2cdbc4f..1b1592c2f5cf 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -661,6 +661,15 @@ struct mm_struct {
 		 * merging.
 		 */
 		unsigned long ksm_merging_pages;
+		/*
+		 * If true, force anonymous pages of this mm to be involved in KSM
+		 * merging without explicitly calling madvise. It is effctive only
+		 * when the klob of '/sys/kernel/mm/ksm/run' is set as 1. If false,
+		 * cancel the feature of ksm_force of this process and unmerge
+		 * those merged pages which is not madvised as MERGEABLE of this
+		 * process, but leave MERGEABLE areas merged.
+		 */
+		bool ksm_force;
 #endif
 	} __randomize_layout;
 
diff --git a/mm/ksm.c b/mm/ksm.c
index 38360285497a..c9f672dcc72e 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -334,6 +334,34 @@ static void __init ksm_slab_free(void)
 	mm_slot_cache = NULL;
 }
 
+/* Check if vma is qualified for ksmd scanning */
+static bool ksm_vma_check(struct vm_area_struct *vma)
+{
+	unsigned long vm_flags = vma->vm_flags;
+
+	if (!(vma->vm_flags & VM_MERGEABLE) && !(vma->vm_mm->ksm_force))
+		return false;
+
+	if (vm_flags & (VM_SHARED	| VM_MAYSHARE	|
+			VM_PFNMAP	| VM_IO | VM_DONTEXPAND |
+			VM_HUGETLB	| VM_MIXEDMAP))
+		return false;       /* just ignore this vma*/
+
+	if (vma_is_dax(vma))
+		return false;
+
+#ifdef VM_SAO
+	if (vm_flags & VM_SAO)
+		return false;
+#endif
+#ifdef VM_SPARC_ADI
+	if (vm_flags & VM_SPARC_ADI)
+		return false;
+#endif
+
+	return true;
+}
+
 static __always_inline bool is_stable_node_chain(struct stable_node *chain)
 {
 	return chain->rmap_hlist_len == STABLE_NODE_CHAIN;
@@ -523,7 +551,7 @@ static struct vm_area_struct *find_mergeable_vma(struct mm_struct *mm,
 	if (ksm_test_exit(mm))
 		return NULL;
 	vma = vma_lookup(mm, addr);
-	if (!vma || !(vma->vm_flags & VM_MERGEABLE) || !vma->anon_vma)
+	if (!vma || !ksm_vma_check(vma) || !vma->anon_vma)
 		return NULL;
 	return vma;
 }
@@ -2297,7 +2325,7 @@ static struct rmap_item *scan_get_next_rmap_item(struct page **page)
 		vma = find_vma(mm, ksm_scan.address);
 
 	for (; vma; vma = vma->vm_next) {
-		if (!(vma->vm_flags & VM_MERGEABLE))
+		if (!ksm_vma_check(vma))
 			continue;
 		if (ksm_scan.address < vma->vm_start)
 			ksm_scan.address = vma->vm_start;
-- 
2.25.1

Re: [PATCH v5] mm/ksm: introduce ksm_force for each process

[Matthew Wilcox]

On Sun, May 08, 2022 at 09:27:10AM +0000, cgel.zte@gmail.com wrote:
> If ksm_force is set to 0, cancel the feature of ksm_force of this
> process and unmerge those merged pages belonging to VMAs which is not
> madvised as MADV_MERGEABLE of this process, but leave MADV_MERGEABLE
> areas merged.

Is that actually a useful feature?  Otherwise, we could simply turn
on/off the existing MMF_VM_MERGEABLE flag instead of introducing this
new bool.

> +Controlling KSM with procfs
> +===========================
> +
> +KSM can also operate on anonymous areas of address space of those processes's
> +knob ``/proc/<pid>/ksm_force`` is on, even if app codes doesn't call madvise()
> +explicitly to advise specific areas as MADV_MERGEABLE.
> +
> +You can set ksm_force to 1 to force all anonymous and qualified VMAs of
> +this process to be involved in KSM scanning. But It is effective only when the
> +klob of ``/sys/kernel/mm/ksm/run`` is set as 1.

I think that last sentence doesn't really add any value.

> +	memset(buffer, 0, sizeof(buffer));
> +	if (count > sizeof(buffer) - 1)
> +		count = sizeof(buffer) - 1;
> +	if (copy_from_user(buffer, buf, count)) {
> +		err = -EFAULT;
> +		goto out_return;

This feels a bit unnecessary.  Just 'return -EFAULT' here.

> +	}
> +
> +	err = kstrtoint(strstrip(buffer), 0, &force);
> +
> +	if (err)
> +		goto out_return;

'return err'

> +	if (force != 0 && force != 1) {
> +		err = -EINVAL;
> +		goto out_return;

'return -EINVAL'

> +	}
> +
> +	task = get_proc_task(file_inode(file));
> +	if (!task) {
> +		err = -ESRCH;
> +		goto out_return;

'return -ESRCH'

Re: [PATCH v5] mm/ksm: introduce ksm_force for each process

[CGEL]

On Sun, May 08, 2022 at 07:03:36PM +0100, Matthew Wilcox wrote:
> On Sun, May 08, 2022 at 09:27:10AM +0000, cgel.zte@gmail.com wrote:
> > If ksm_force is set to 0, cancel the feature of ksm_force of this
> > process and unmerge those merged pages belonging to VMAs which is not
> > madvised as MADV_MERGEABLE of this process, but leave MADV_MERGEABLE
> > areas merged.
> 
> Is that actually a useful feature?  Otherwise, we could simply turn
> on/off the existing MMF_VM_MERGEABLE flag instead of introducing this
> new bool.
> 
I think this will be very useful for those apps which are very likely to
cause Same Pages in memory and users and operators are not willing to
modified the source codes for any reasons.

Besides, simply turning of/off the existing MMF_VM_MERGEABLE flag may be
not feasible because madvise will also turn on the MMF_VM_MERGEABLE
flag.

I think the following suggestions is good, and I will resend a patch.
> > +Controlling KSM with procfs
> > +===========================
> > +
> > +KSM can also operate on anonymous areas of address space of those processes's
> > +knob ``/proc/<pid>/ksm_force`` is on, even if app codes doesn't call madvise()
> > +explicitly to advise specific areas as MADV_MERGEABLE.
> > +
> > +You can set ksm_force to 1 to force all anonymous and qualified VMAs of
> > +this process to be involved in KSM scanning. But It is effective only when the
> > +klob of ``/sys/kernel/mm/ksm/run`` is set as 1.
> 
> I think that last sentence doesn't really add any value.
> 
> > +	memset(buffer, 0, sizeof(buffer));
> > +	if (count > sizeof(buffer) - 1)
> > +		count = sizeof(buffer) - 1;
> > +	if (copy_from_user(buffer, buf, count)) {
> > +		err = -EFAULT;
> > +		goto out_return;
> 
> This feels a bit unnecessary.  Just 'return -EFAULT' here.
> 
> > +	}
> > +
> > +	err = kstrtoint(strstrip(buffer), 0, &force);
> > +
> > +	if (err)
> > +		goto out_return;
> 
> 'return err'
> 
> > +	if (force != 0 && force != 1) {
> > +		err = -EINVAL;
> > +		goto out_return;
> 
> 'return -EINVAL'
> 
> > +	}
> > +
> > +	task = get_proc_task(file_inode(file));
> > +	if (!task) {
> > +		err = -ESRCH;
> > +		goto out_return;
> 
> 'return -ESRCH'

Re: [PATCH v5] mm/ksm: introduce ksm_force for each process

[Matthew Wilcox]

On Mon, May 09, 2022 at 06:57:33AM +0000, CGEL wrote:
> On Sun, May 08, 2022 at 07:03:36PM +0100, Matthew Wilcox wrote:
> > On Sun, May 08, 2022 at 09:27:10AM +0000, cgel.zte@gmail.com wrote:
> > > If ksm_force is set to 0, cancel the feature of ksm_force of this
> > > process and unmerge those merged pages belonging to VMAs which is not
> > > madvised as MADV_MERGEABLE of this process, but leave MADV_MERGEABLE
> > > areas merged.
> > 
> > Is that actually a useful feature?  Otherwise, we could simply turn
> > on/off the existing MMF_VM_MERGEABLE flag instead of introducing this
> > new bool.
> > 
> I think this will be very useful for those apps which are very likely to
> cause Same Pages in memory and users and operators are not willing to
> modified the source codes for any reasons.

No, you misunderstand.  Is it useful to have the "force KSM off"
functionality?  ie code which has been modified to allow KSM, but
then overridden by an admin?

> Besides, simply turning of/off the existing MMF_VM_MERGEABLE flag may be
> not feasible because madvise will also turn on the MMF_VM_MERGEABLE
> flag.
> 
> I think the following suggestions is good, and I will resend a patch.
> > > +Controlling KSM with procfs
> > > +===========================
> > > +
> > > +KSM can also operate on anonymous areas of address space of those processes's
> > > +knob ``/proc/<pid>/ksm_force`` is on, even if app codes doesn't call madvise()
> > > +explicitly to advise specific areas as MADV_MERGEABLE.
> > > +
> > > +You can set ksm_force to 1 to force all anonymous and qualified VMAs of
> > > +this process to be involved in KSM scanning. But It is effective only when the
> > > +klob of ``/sys/kernel/mm/ksm/run`` is set as 1.
> > 
> > I think that last sentence doesn't really add any value.
> > 
> > > +	memset(buffer, 0, sizeof(buffer));
> > > +	if (count > sizeof(buffer) - 1)
> > > +		count = sizeof(buffer) - 1;
> > > +	if (copy_from_user(buffer, buf, count)) {
> > > +		err = -EFAULT;
> > > +		goto out_return;
> > 
> > This feels a bit unnecessary.  Just 'return -EFAULT' here.
> > 
> > > +	}
> > > +
> > > +	err = kstrtoint(strstrip(buffer), 0, &force);
> > > +
> > > +	if (err)
> > > +		goto out_return;
> > 
> > 'return err'
> > 
> > > +	if (force != 0 && force != 1) {
> > > +		err = -EINVAL;
> > > +		goto out_return;
> > 
> > 'return -EINVAL'
> > 
> > > +	}
> > > +
> > > +	task = get_proc_task(file_inode(file));
> > > +	if (!task) {
> > > +		err = -ESRCH;
> > > +		goto out_return;
> > 
> > 'return -ESRCH'

Re: [PATCH v5] mm/ksm: introduce ksm_force for each process

[CGEL]

On Mon, May 09, 2022 at 04:40:50PM +0100, Matthew Wilcox wrote:
> On Mon, May 09, 2022 at 06:57:33AM +0000, CGEL wrote:
> > On Sun, May 08, 2022 at 07:03:36PM +0100, Matthew Wilcox wrote:
> > > On Sun, May 08, 2022 at 09:27:10AM +0000, cgel.zte@gmail.com wrote:
> > > > If ksm_force is set to 0, cancel the feature of ksm_force of this
> > > > process and unmerge those merged pages belonging to VMAs which is not
> > > > madvised as MADV_MERGEABLE of this process, but leave MADV_MERGEABLE
> > > > areas merged.
> > > 
> > > Is that actually a useful feature?  Otherwise, we could simply turn
> > > on/off the existing MMF_VM_MERGEABLE flag instead of introducing this
> > > new bool.
> > > 
> > I think this will be very useful for those apps which are very likely to
> > cause Same Pages in memory and users and operators are not willing to
> > modified the source codes for any reasons.
> 
> No, you misunderstand.  Is it useful to have the "force KSM off"
> functionality?  ie code which has been modified to allow KSM, but
> then overridden by an admin?
> 
Oh, I see what you mean. It should be mentioned that "force KSM off" is not
implemented for the current patch. In this patch, setting ksm_force to 0 just
restores the system to the default state (the state before patching)

> > Besides, simply turning of/off the existing MMF_VM_MERGEABLE flag may be
> > not feasible because madvise will also turn on the MMF_VM_MERGEABLE
> > flag.
> > 
> > I think the following suggestions is good, and I will resend a patch.
> > > > +Controlling KSM with procfs
> > > > +===========================
> > > > +
> > > > +KSM can also operate on anonymous areas of address space of those processes's
> > > > +knob ``/proc/<pid>/ksm_force`` is on, even if app codes doesn't call madvise()
> > > > +explicitly to advise specific areas as MADV_MERGEABLE.
> > > > +
> > > > +You can set ksm_force to 1 to force all anonymous and qualified VMAs of
> > > > +this process to be involved in KSM scanning. But It is effective only when the
> > > > +klob of ``/sys/kernel/mm/ksm/run`` is set as 1.
> > > 
> > > I think that last sentence doesn't really add any value.
> > > 
> > > > +	memset(buffer, 0, sizeof(buffer));
> > > > +	if (count > sizeof(buffer) - 1)
> > > > +		count = sizeof(buffer) - 1;
> > > > +	if (copy_from_user(buffer, buf, count)) {
> > > > +		err = -EFAULT;
> > > > +		goto out_return;
> > > 
> > > This feels a bit unnecessary.  Just 'return -EFAULT' here.
> > > 
> > > > +	}
> > > > +
> > > > +	err = kstrtoint(strstrip(buffer), 0, &force);
> > > > +
> > > > +	if (err)
> > > > +		goto out_return;
> > > 
> > > 'return err'
> > > 
> > > > +	if (force != 0 && force != 1) {
> > > > +		err = -EINVAL;
> > > > +		goto out_return;
> > > 
> > > 'return -EINVAL'
> > > 
> > > > +	}
> > > > +
> > > > +	task = get_proc_task(file_inode(file));
> > > > +	if (!task) {
> > > > +		err = -ESRCH;
> > > > +		goto out_return;
> > > 
> > > 'return -ESRCH'

Re: [PATCH v5] mm/ksm: introduce ksm_force for each process

[Ammar Faizi]

On 5/8/22 4:27 PM, cgel.zte@gmail.com wrote:
> +static ssize_t ksm_force_write(struct file *file, const char __user *buf,
> +				size_t count, loff_t *ppos)
> +{
> +	struct task_struct *task;
> +	struct mm_struct *mm;
> +	char buffer[PROC_NUMBUF];
> +	int force;
> +	int err = 0;
> +
> +	memset(buffer, 0, sizeof(buffer));
> +	if (count > sizeof(buffer) - 1)
> +		count = sizeof(buffer) - 1;
> +	if (copy_from_user(buffer, buf, count)) {
> +		err = -EFAULT;
> +		goto out_return;
> +	}

This one looks like over-zeroing to me. You don't need to zero
all elements in the array. You're going to overwrite it with
`copy_from_user()` anyway.

Just zero the last potentially useful element by using @count
as the index. It can be like this:

```
	char buffer[PROC_NUMBUF];

	if (count > sizeof(buffer) - 1)
		count = sizeof(buffer) - 1;
	if (copy_from_user(buffer, buf, count))
		return -EFAULT;
	buffer[count] = '\0';
```

-- 
Ammar Faizi

Re: [PATCH v5] mm/ksm: introduce ksm_force for each process

[Andrew Morton]

On Wed, 11 May 2022 03:10:31 +0700 Ammar Faizi <ammarfaizi2@gnuweeb.org> wrote:

> On 5/8/22 4:27 PM, cgel.zte@gmail.com wrote:
> > +static ssize_t ksm_force_write(struct file *file, const char __user *buf,
> > +				size_t count, loff_t *ppos)
> > +{
> > +	struct task_struct *task;
> > +	struct mm_struct *mm;
> > +	char buffer[PROC_NUMBUF];
> > +	int force;
> > +	int err = 0;
> > +
> > +	memset(buffer, 0, sizeof(buffer));
> > +	if (count > sizeof(buffer) - 1)
> > +		count = sizeof(buffer) - 1;
> > +	if (copy_from_user(buffer, buf, count)) {
> > +		err = -EFAULT;
> > +		goto out_return;
> > +	}
> 
> This one looks like over-zeroing to me. You don't need to zero
> all elements in the array. You're going to overwrite it with
> `copy_from_user()` anyway.
> 
> Just zero the last potentially useful element by using @count
> as the index. It can be like this:
> 
> ```
> 	char buffer[PROC_NUMBUF];
> 
> 	if (count > sizeof(buffer) - 1)
> 		count = sizeof(buffer) - 1;
> 	if (copy_from_user(buffer, buf, count))
> 		return -EFAULT;
> 	buffer[count] = '\0';
> ```

Use strncpy_from_user()?

Can this code use proc_dointvec_minmax() or similar?

Re: [PATCH v5] mm/ksm: introduce ksm_force for each process

[Ammar Faizi]

On 5/11/22 3:30 AM, Andrew Morton wrote:
> On Wed, 11 May 2022 03:10:31 +0700 Ammar Faizi <ammarfaizi2@gnuweeb.org> wrote:
> 
>> On 5/8/22 4:27 PM, cgel.zte@gmail.com wrote:
>>> +static ssize_t ksm_force_write(struct file *file, const char __user *buf,
>>> +				size_t count, loff_t *ppos)
>>> +{
>>> +	struct task_struct *task;
>>> +	struct mm_struct *mm;
>>> +	char buffer[PROC_NUMBUF];
>>> +	int force;
>>> +	int err = 0;
>>> +
>>> +	memset(buffer, 0, sizeof(buffer));
>>> +	if (count > sizeof(buffer) - 1)
>>> +		count = sizeof(buffer) - 1;
>>> +	if (copy_from_user(buffer, buf, count)) {
>>> +		err = -EFAULT;
>>> +		goto out_return;
>>> +	}
>>
>> This one looks like over-zeroing to me. You don't need to zero
>> all elements in the array. You're going to overwrite it with
>> `copy_from_user()` anyway.
>>
>> Just zero the last potentially useful element by using @count
>> as the index. It can be like this:
>>
>> ```
>> 	char buffer[PROC_NUMBUF];
>>
>> 	if (count > sizeof(buffer) - 1)
>> 		count = sizeof(buffer) - 1;
>> 	if (copy_from_user(buffer, buf, count))
>> 		return -EFAULT;
>> 	buffer[count] = '\0';
>> ```
> 
> Use strncpy_from_user()?

Sounds better.

> Can this code use proc_dointvec_minmax() or similar?

Not familiar with that API at all. Leaving it to other participants...

-- 
Ammar Faizi

[PATCH v7] mm/ksm: introduce ksm_force for each process

[cgel.zte]

From: xu xin <xu.xin16@zte.com.cn>

To use KSM, we have to explicitly call madvise() in application code,
which means installed apps on OS needs to be uninstall and source code
needs to be modified. It is inconvenient.

In order to change this situation, We add a new proc file ksm_force
under /proc/<pid>/ to support turning on/off KSM scanning of a
process's mm dynamically.

If ksm_force is set to 1, force all anonymous and 'qualified' VMAs
of this mm to be involved in KSM scanning without explicitly calling
madvise to mark VMA as MADV_MERGEABLE. But It is effective only when
the klob of /sys/kernel/mm/ksm/run is set as 1.

If ksm_force is set to 0, cancel the feature of ksm_force of this
process and unmerge those merged pages belonging to VMAs which is not
madvised as MADV_MERGEABLE of this process, but leave MADV_MERGEABLE
areas merged.

Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
Reviewed-by: Ran Xiaokai <ran.xiaokai@zte.com.cn>
Reviewed-by: wangyong <wang.yong12@zte.com.cn>
Reviewed-by: Yunkai Zhang <zhang.yunkai@zte.com.cn>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Suggested-by: Ammar Faizi <ammarfaizi2@gnuweeb.org>
---
v7:
- remove over-zeroing in ksm_force_write() and using strncpy_from_user
instead of copy_from_user.
v6:
- modify the way of "return"
- remove unnecessary words in Documentation/admin-guide/mm/ksm.rst
- add additional notes to "set 0 to ksm_force" in Documentation/../ksm.rst
and Documentation/../proc.rst
v5:
- fix typos in Documentation/filesystem/proc.rst
v4:
- fix typos in commit log
- add interface descriptions under Documentation/
v3:
- fix compile error of mm/ksm.c
v2:
- fix a spelling error in commit log.
- remove a redundant condition check in ksm_force_write().
---
 Documentation/admin-guide/mm/ksm.rst | 19 +++++-
 Documentation/filesystems/proc.rst   | 17 +++++
 fs/proc/base.c                       | 97 ++++++++++++++++++++++++++++
 include/linux/mm_types.h             |  9 +++
 mm/ksm.c                             | 32 ++++++++-
 5 files changed, 171 insertions(+), 3 deletions(-)

diff --git a/Documentation/admin-guide/mm/ksm.rst b/Documentation/admin-guide/mm/ksm.rst
index b244f0202a03..8cabc2504005 100644
--- a/Documentation/admin-guide/mm/ksm.rst
+++ b/Documentation/admin-guide/mm/ksm.rst
@@ -32,7 +32,7 @@ are swapped back in: ksmd must rediscover their identity and merge again).
 Controlling KSM with madvise
 ============================
 
-KSM only operates on those areas of address space which an application
+KSM can operates on those areas of address space which an application
 has advised to be likely candidates for merging, by using the madvise(2)
 system call::
 
@@ -70,6 +70,23 @@ Applications should be considerate in their use of MADV_MERGEABLE,
 restricting its use to areas likely to benefit.  KSM's scans may use a lot
 of processing power: some installations will disable KSM for that reason.
 
+Controlling KSM with procfs
+===========================
+
+KSM can also operate on anonymous areas of address space of those processes's
+knob ``/proc/<pid>/ksm_force`` is on, even if app codes doesn't call madvise()
+explicitly to advise specific areas as MADV_MERGEABLE.
+
+You can set ksm_force to 1 to force all anonymous and qualified VMAs of
+this process to be involved in KSM scanning.
+	e.g. ``echo 1 > /proc/<pid>/ksm_force``
+
+You can also set ksm_force to 0 to cancel that force feature of this process
+and unmerge those merged pages which belongs to those VMAs not marked as
+MADV_MERGEABLE of this process. But that still leave those pages belonging to
+VMAs marked as MADV_MERGEABLE merged (fallback to the default state).
+	e.g. ``echo 0 > /proc/<pid>/ksm_force``
+
 .. _ksm_sysfs:
 
 KSM daemon sysfs interface
diff --git a/Documentation/filesystems/proc.rst b/Documentation/filesystems/proc.rst
index 061744c436d9..8890b8b457a4 100644
--- a/Documentation/filesystems/proc.rst
+++ b/Documentation/filesystems/proc.rst
@@ -47,6 +47,7 @@ fixes/update part 1.1  Stefani Seibold <stefani@seibold.net>    June 9 2009
   3.10  /proc/<pid>/timerslack_ns - Task timerslack value
   3.11	/proc/<pid>/patch_state - Livepatch patch operation state
   3.12	/proc/<pid>/arch_status - Task architecture specific information
+  3.13	/proc/<pid>/ksm_force - Setting of mandatory involvement in KSM
 
   4	Configuring procfs
   4.1	Mount options
@@ -2176,6 +2177,22 @@ AVX512_elapsed_ms
   the task is unlikely an AVX512 user, but depends on the workload and the
   scheduling scenario, it also could be a false negative mentioned above.
 
+3.13	/proc/<pid>/ksm_force - Setting of mandatory involvement in KSM
+-----------------------------------------------------------------------
+When CONFIG_KSM is enabled, this file can be used to specify if this
+process's anonymous memory can be involved in KSM scanning without app codes
+explicitly calling madvise to mark memory address as MADV_MERGEABLE.
+
+If writing 1 to this file, the kernel will force all anonymous and qualified
+memory to be involved in KSM scanning without explicitly calling madvise to
+mark memory address as MADV_MERGEABLE. But that is effective only when the
+klob of '/sys/kernel/mm/ksm/run' is set as 1.
+
+If writing 0 to this file, the mandatory KSM feature of this process's will
+be cancelled and unmerge those merged pages which belongs to those areas not
+marked as MADV_MERGEABLE of this process, but leave those pages belonging to
+areas marked as MADV_MERGEABLE merged (fallback to the default state).
+
 Chapter 4: Configuring procfs
 =============================
 
diff --git a/fs/proc/base.c b/fs/proc/base.c
index 8dfa36a99c74..99ab0e8cdcbc 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -96,6 +96,7 @@
 #include <linux/time_namespace.h>
 #include <linux/resctrl.h>
 #include <linux/cn_proc.h>
+#include <linux/ksm.h>
 #include <trace/events/oom.h>
 #include "internal.h"
 #include "fd.h"
@@ -3168,6 +3169,100 @@ static int proc_pid_ksm_merging_pages(struct seq_file *m, struct pid_namespace *
 
 	return 0;
 }
+
+static ssize_t ksm_force_read(struct file *file, char __user *buf, size_t count,
+				loff_t *ppos)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	char buffer[PROC_NUMBUF];
+	ssize_t len;
+	int ret;
+
+	task = get_proc_task(file_inode(file));
+	if (!task)
+		return -ESRCH;
+
+	mm = get_task_mm(task);
+	ret = 0;
+	if (mm) {
+		len = snprintf(buffer, sizeof(buffer), "%d\n", mm->ksm_force);
+		ret =  simple_read_from_buffer(buf, count, ppos, buffer, len);
+		mmput(mm);
+	}
+
+	return ret;
+}
+
+static ssize_t ksm_force_write(struct file *file, const char __user *buf,
+				size_t count, loff_t *ppos)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	char buffer[PROC_NUMBUF];
+	int force;
+	int err = 0;
+	int str_len;
+
+	if (count > sizeof(buffer) - 1) {
+		count = sizeof(buffer) - 1;
+	}
+
+	str_len = strncpy_from_user(buffer, buf, count);
+	if (str_len < 0)
+		return -EFAULT;
+	buffer[str_len] = '\0';
+
+	err = kstrtoint(strstrip(buffer), 0, &force);
+	if (err)
+		return err;
+
+	if (force != 0 && force != 1)
+		return -EINVAL;
+
+	task = get_proc_task(file_inode(file));
+	if (!task)
+		return -ESRCH;
+
+	mm = get_task_mm(task);
+	if (!mm)
+		goto out_put_task;
+
+	if (mm->ksm_force != force) {
+		if (mmap_write_lock_killable(mm)) {
+			err = -EINTR;
+			goto out_mmput;
+		}
+
+		if (force == 0)
+			mm->ksm_force = force;
+		else {
+			/*
+			 * Force anonymous pages of this mm to be involved in KSM merging
+			 * without explicitly calling madvise.
+			 */
+			if (!test_bit(MMF_VM_MERGEABLE, &mm->flags))
+				err = __ksm_enter(mm);
+			if (!err)
+				mm->ksm_force = force;
+		}
+
+		mmap_write_unlock(mm);
+	}
+
+out_mmput:
+	mmput(mm);
+out_put_task:
+	put_task_struct(task);
+
+	return err < 0 ? err : count;
+}
+
+static const struct file_operations proc_pid_ksm_force_operations = {
+	.read		= ksm_force_read,
+	.write		= ksm_force_write,
+	.llseek		= generic_file_llseek,
+};
 #endif /* CONFIG_KSM */
 
 #ifdef CONFIG_STACKLEAK_METRICS
@@ -3303,6 +3398,7 @@ static const struct pid_entry tgid_base_stuff[] = {
 #endif
 #ifdef CONFIG_KSM
 	ONE("ksm_merging_pages",  S_IRUSR, proc_pid_ksm_merging_pages),
+	REG("ksm_force", S_IRUSR|S_IWUSR, proc_pid_ksm_force_operations),
 #endif
 };
 
@@ -3639,6 +3735,7 @@ static const struct pid_entry tid_base_stuff[] = {
 #endif
 #ifdef CONFIG_KSM
 	ONE("ksm_merging_pages",  S_IRUSR, proc_pid_ksm_merging_pages),
+	REG("ksm_force", S_IRUSR|S_IWUSR, proc_pid_ksm_force_operations),
 #endif
 };
 
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index b34ff2cdbc4f..1b1592c2f5cf 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -661,6 +661,15 @@ struct mm_struct {
 		 * merging.
 		 */
 		unsigned long ksm_merging_pages;
+		/*
+		 * If true, force anonymous pages of this mm to be involved in KSM
+		 * merging without explicitly calling madvise. It is effctive only
+		 * when the klob of '/sys/kernel/mm/ksm/run' is set as 1. If false,
+		 * cancel the feature of ksm_force of this process and unmerge
+		 * those merged pages which is not madvised as MERGEABLE of this
+		 * process, but leave MERGEABLE areas merged.
+		 */
+		bool ksm_force;
 #endif
 	} __randomize_layout;
 
diff --git a/mm/ksm.c b/mm/ksm.c
index 38360285497a..c9f672dcc72e 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -334,6 +334,34 @@ static void __init ksm_slab_free(void)
 	mm_slot_cache = NULL;
 }
 
+/* Check if vma is qualified for ksmd scanning */
+static bool ksm_vma_check(struct vm_area_struct *vma)
+{
+	unsigned long vm_flags = vma->vm_flags;
+
+	if (!(vma->vm_flags & VM_MERGEABLE) && !(vma->vm_mm->ksm_force))
+		return false;
+
+	if (vm_flags & (VM_SHARED	| VM_MAYSHARE	|
+			VM_PFNMAP	| VM_IO | VM_DONTEXPAND |
+			VM_HUGETLB	| VM_MIXEDMAP))
+		return false;       /* just ignore this vma*/
+
+	if (vma_is_dax(vma))
+		return false;
+
+#ifdef VM_SAO
+	if (vm_flags & VM_SAO)
+		return false;
+#endif
+#ifdef VM_SPARC_ADI
+	if (vm_flags & VM_SPARC_ADI)
+		return false;
+#endif
+
+	return true;
+}
+
 static __always_inline bool is_stable_node_chain(struct stable_node *chain)
 {
 	return chain->rmap_hlist_len == STABLE_NODE_CHAIN;
@@ -523,7 +551,7 @@ static struct vm_area_struct *find_mergeable_vma(struct mm_struct *mm,
 	if (ksm_test_exit(mm))
 		return NULL;
 	vma = vma_lookup(mm, addr);
-	if (!vma || !(vma->vm_flags & VM_MERGEABLE) || !vma->anon_vma)
+	if (!vma || !ksm_vma_check(vma) || !vma->anon_vma)
 		return NULL;
 	return vma;
 }
@@ -2297,7 +2325,7 @@ static struct rmap_item *scan_get_next_rmap_item(struct page **page)
 		vma = find_vma(mm, ksm_scan.address);
 
 	for (; vma; vma = vma->vm_next) {
-		if (!(vma->vm_flags & VM_MERGEABLE))
+		if (!ksm_vma_check(vma))
 			continue;
 		if (ksm_scan.address < vma->vm_start)
 			ksm_scan.address = vma->vm_start;
-- 
2.25.1

Re: [PATCH v7] mm/ksm: introduce ksm_force for each process

[Andrew Morton]

On Thu, 12 May 2022 07:03:47 +0000 cgel.zte@gmail.com wrote:

> From: xu xin <xu.xin16@zte.com.cn>
> 
> To use KSM, we have to explicitly call madvise() in application code,
> which means installed apps on OS needs to be uninstall and source code
> needs to be modified. It is inconvenient.
> 
> In order to change this situation, We add a new proc file ksm_force
> under /proc/<pid>/ to support turning on/off KSM scanning of a
> process's mm dynamically.
> 
> If ksm_force is set to 1, force all anonymous and 'qualified' VMAs
> of this mm to be involved in KSM scanning without explicitly calling
> madvise to mark VMA as MADV_MERGEABLE. But It is effective only when
> the klob of /sys/kernel/mm/ksm/run is set as 1.
> 
> If ksm_force is set to 0, cancel the feature of ksm_force of this
> process and unmerge those merged pages belonging to VMAs which is not
> madvised as MADV_MERGEABLE of this process, but leave MADV_MERGEABLE
> areas merged.

It certainly seems like a useful feature.

> Signed-off-by: xu xin <xu.xin16@zte.com.cn>
> Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
> Reviewed-by: Ran Xiaokai <ran.xiaokai@zte.com.cn>
> Reviewed-by: wangyong <wang.yong12@zte.com.cn>
> Reviewed-by: Yunkai Zhang <zhang.yunkai@zte.com.cn>
> Suggested-by: Matthew Wilcox <willy@infradead.org>
> Suggested-by: Ammar Faizi <ammarfaizi2@gnuweeb.org>

This patch doesn't have your Signed-off-by:.  It should, because you
were on the delivery path.  This is described in
Documentation/process/submitting-patches.rst, "Developer's Certificate
of Origin".

I'll queue it for some testing but please do resend with that tag.


> +/* Check if vma is qualified for ksmd scanning */
> +static bool ksm_vma_check(struct vm_area_struct *vma)

I have trouble with "check" names, because the name doesn't convey what
is being checked, nor does the name convey whether it's checking for
truth or for falsity.

I suggest that "vma_scannable" is a more informative name.  It doesn't
need the "ksm_" prefix as this is a static file-local function.

See, with the name "vma_scannable", that comment which you added is
barely needed.

--- a/mm/ksm.c~mm-ksm-introduce-ksm_force-for-each-process-fix
+++ a/mm/ksm.c
@@ -335,7 +335,7 @@ static void __init ksm_slab_free(void)
 }
 
 /* Check if vma is qualified for ksmd scanning */
-static bool ksm_vma_check(struct vm_area_struct *vma)
+static bool vma_scannable(struct vm_area_struct *vma)
 {
 	unsigned long vm_flags = vma->vm_flags;
 
@@ -551,7 +551,7 @@ static struct vm_area_struct *find_merge
 	if (ksm_test_exit(mm))
 		return NULL;
 	vma = vma_lookup(mm, addr);
-	if (!vma || !ksm_vma_check(vma) || !vma->anon_vma)
+	if (!vma || !vma_scannable(vma) || !vma->anon_vma)
 		return NULL;
 	return vma;
 }
@@ -2328,7 +2328,7 @@ next_mm:
 		goto no_vmas;
 
 	for_each_vma(vmi, vma) {
-		if (!ksm_vma_check(vma))
+		if (!vma_scannable(vma))
 			continue;
 		if (ksm_scan.address < vma->vm_start)
 			ksm_scan.address = vma->vm_start;
_