[PATCH v2 0/3] KSM: Mark new vma for deduplication

[PATCH v2 0/3] KSM: Mark new vma for deduplication

[Timofey Titovets]

Good time of day List,
this tiny series of patches implement feature for auto deduping all anonymous memory.
mark_new_vma - new ksm sysfs interface
Every time then new vma created and mark_new_vma set to 1, then will be vma marked as VM_MERGEABLE and added to ksm queue.
This can produce small overheads
(I have not catch any problems or slowdown)

This is useful for:
Android (CM) devs which implement ksm support with patching system.
Users of tiny pc.
Servers what not use KVM but use something very releated, like containers.

For tests:
I have tested it and it working very good. For testing apply it and enable ksm:
echo 1 | sudo tee /sys/kernel/mm/ksm/run
This show how much memory saved:
echo $[$(cat /sys/kernel/mm/ksm/pages_shared)*$(getconf PAGE_SIZE)/1024 ]KB

On my system i save ~1% of memory 26 Mb/2100 Mb (deduped)/(used)

Timofey Titovets (3):
  KSM: Add auto flag new VMA as VM_MERGEABLE
  KSM: Add to sysfs - mark_new_vma
  KSM: Add config to control mark_new_vma

 include/linux/ksm.h | 39 +++++++++++++++++++++++++++++++++++++++
 mm/Kconfig          |  7 +++++++
 mm/ksm.c            | 39 ++++++++++++++++++++++++++++++++++++++-
 mm/mmap.c           | 17 +++++++++++++++++
 4 files changed, 101 insertions(+), 1 deletion(-)

--
2.1.3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH v2 1/3] KSM: Add auto flag new VMA as VM_MERGEABLE

[Timofey Titovets]

Implement two functions:
ksm_vm_flags_mod() - if existing flags supported by ksm - then mark like VM_MERGEABLE
ksm_vma_add_new() - If vma marked as VM_MERGEABLE add it to ksm page queue

Signed-off-by: Timofey Titovets <nefelim4ag@gmail.com>
---
 include/linux/ksm.h | 31 +++++++++++++++++++++++++++++++
 mm/mmap.c           | 17 +++++++++++++++++
 2 files changed, 48 insertions(+)

diff --git a/include/linux/ksm.h b/include/linux/ksm.h
index 3be6bb1..c3fff64 100644
--- a/include/linux/ksm.h
+++ b/include/linux/ksm.h
@@ -76,6 +76,29 @@ struct page *ksm_might_need_to_copy(struct page *page,
 int rmap_walk_ksm(struct page *page, struct rmap_walk_control *rwc);
 void ksm_migrate_page(struct page *newpage, struct page *oldpage);
 
+/*
+ * Allow to mark new vma as VM_MERGEABLE
+ */
+#ifndef VM_SAO
+#define VM_SAO 0
+#endif
+static inline void ksm_vm_flags_mod(unsigned long *vm_flags)
+{
+	if (*vm_flags & (VM_MERGEABLE | VM_SHARED  | VM_MAYSHARE   |
+			 VM_PFNMAP    | VM_IO      | VM_DONTEXPAND |
+			 VM_HUGETLB | VM_NONLINEAR | VM_MIXEDMAP   | VM_SAO) )
+		return;
+	*vm_flags |= VM_MERGEABLE;
+}
+
+static inline void ksm_vma_add_new(struct vm_area_struct *vma)
+{
+	struct mm_struct *mm = vma->vm_mm;
+	if (!test_bit(MMF_VM_MERGEABLE, &mm->flags)) {
+		__ksm_enter(mm);
+	}
+}
+
 #else  /* !CONFIG_KSM */
 
 static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm)
@@ -92,6 +115,14 @@ static inline int PageKsm(struct page *page)
 	return 0;
 }
 
+static inline void ksm_vm_flags_mod(unsigned long *vm_flags_p)
+{
+}
+
+void ksm_vma_add_new(struct vm_area_struct *vma)
+{
+}
+
 #ifdef CONFIG_MMU
 static inline int ksm_madvise(struct vm_area_struct *vma, unsigned long start,
 		unsigned long end, int advice, unsigned long *vm_flags)
diff --git a/mm/mmap.c b/mm/mmap.c
index 7f85520..ce0073e 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -41,6 +41,7 @@
 #include <linux/notifier.h>
 #include <linux/memory.h>
 #include <linux/printk.h>
+#include <linux/ksm.h>
 
 #include <asm/uaccess.h>
 #include <asm/cacheflush.h>
@@ -911,10 +912,14 @@ again:			remove_next = 1 + (end > next->vm_end);
 			vma_gap_update(next);
 		else
 			mm->highest_vm_end = end;
+	} else {
+		if (next && !insert)
+			ksm_vma_add_new(next);
 	}
 	if (insert && file)
 		uprobe_mmap(insert);
 
+	ksm_vma_add_new(vma);
 	validate_mm(mm);
 
 	return 0;
@@ -1307,6 +1312,9 @@ unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
 	vm_flags = calc_vm_prot_bits(prot) | calc_vm_flag_bits(flags) |
 			mm->def_flags | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC;
 
+	/* If ksm is enabled, we add VM_MERGABLE to new VMAs. */
+	ksm_vm_flags_mod(&vm_flags);
+
 	if (flags & MAP_LOCKED)
 		if (!can_do_mlock())
 			return -EPERM;
@@ -1648,6 +1656,7 @@ munmap_back:
 			allow_write_access(file);
 	}
 	file = vma->vm_file;
+	ksm_vma_add_new(vma);
 out:
 	perf_event_mmap(vma);
 
@@ -2484,6 +2493,8 @@ static int __split_vma(struct mm_struct *mm, struct vm_area_struct *vma,
 	else
 		err = vma_adjust(vma, vma->vm_start, addr, vma->vm_pgoff, new);
 
+	ksm_vma_add_new(vma);
+
 	/* Success. */
 	if (!err)
 		return 0;
@@ -2659,6 +2670,9 @@ static unsigned long do_brk(unsigned long addr, unsigned long len)
 	if (error)
 		return error;
 
+	/* If ksm is enabled, we add VM_MERGABLE to new VMAs. */
+	ksm_vm_flags_mod(&flags);
+
 	/*
 	 * mm->mmap_sem is required to protect against another thread
 	 * changing the mappings in case we sleep.
@@ -2708,6 +2722,7 @@ static unsigned long do_brk(unsigned long addr, unsigned long len)
 	vma->vm_flags = flags;
 	vma->vm_page_prot = vm_get_page_prot(flags);
 	vma_link(mm, vma, prev, rb_link, rb_parent);
+	ksm_vma_add_new(vma);
 out:
 	perf_event_mmap(vma);
 	mm->total_vm += len >> PAGE_SHIFT;
@@ -2887,6 +2902,7 @@ struct vm_area_struct *copy_vma(struct vm_area_struct **vmap,
 				new_vma->vm_ops->open(new_vma);
 			vma_link(mm, new_vma, prev, rb_link, rb_parent);
 			*need_rmap_locks = false;
+			ksm_vma_add_new(new_vma);
 		}
 	}
 	return new_vma;
@@ -3004,6 +3020,7 @@ static struct vm_area_struct *__install_special_mapping(
 	mm->total_vm += len >> PAGE_SHIFT;
 
 	perf_event_mmap(vma);
+	ksm_vma_add_new(vma);
 
 	return vma;
 
-- 
2.1.3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH v2 2/3] KSM: Add to sysfs - mark_new_vma

[Timofey Titovets]

It allow for user to  control process of marking new vma as VM_MERGEABLE
Create new sysfs interface /sys/kernel/mm/ksm/mark_new_vma
1 - enabled - mark new allocated vma as VM_MERGEABLE and add it to ksm queue
0 - disable it

Signed-off-by: Timofey Titovets <nefelim4ag@gmail.com>
---
 include/linux/ksm.h | 10 +++++++++-
 mm/ksm.c            | 39 ++++++++++++++++++++++++++++++++++++++-
 2 files changed, 47 insertions(+), 2 deletions(-)

diff --git a/include/linux/ksm.h b/include/linux/ksm.h
index c3fff64..deb60fc 100644
--- a/include/linux/ksm.h
+++ b/include/linux/ksm.h
@@ -76,6 +76,12 @@ struct page *ksm_might_need_to_copy(struct page *page,
 int rmap_walk_ksm(struct page *page, struct rmap_walk_control *rwc);
 void ksm_migrate_page(struct page *newpage, struct page *oldpage);
 
+
+/* Mark new vma as mergeable */
+#define ALLOW	1
+#define DENY	0
+extern unsigned mark_new_vma __read_mostly;
+
 /*
  * Allow to mark new vma as VM_MERGEABLE
  */
@@ -84,6 +90,8 @@ void ksm_migrate_page(struct page *newpage, struct page *oldpage);
 #endif
 static inline void ksm_vm_flags_mod(unsigned long *vm_flags)
 {
+	if (!mark_new_vma)
+		return;
 	if (*vm_flags & (VM_MERGEABLE | VM_SHARED  | VM_MAYSHARE   |
 			 VM_PFNMAP    | VM_IO      | VM_DONTEXPAND |
 			 VM_HUGETLB | VM_NONLINEAR | VM_MIXEDMAP   | VM_SAO) )
@@ -94,7 +102,7 @@ static inline void ksm_vm_flags_mod(unsigned long *vm_flags)
 static inline void ksm_vma_add_new(struct vm_area_struct *vma)
 {
 	struct mm_struct *mm = vma->vm_mm;
-	if (!test_bit(MMF_VM_MERGEABLE, &mm->flags)) {
+	if (mark_new_vma && !test_bit(MMF_VM_MERGEABLE, &mm->flags)) {
 		__ksm_enter(mm);
 	}
 }
diff --git a/mm/ksm.c b/mm/ksm.c
index 6b2e337..7bfb616 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -223,6 +223,12 @@ static unsigned int ksm_thread_pages_to_scan = 100;
 /* Milliseconds ksmd should sleep between batches */
 static unsigned int ksm_thread_sleep_millisecs = 20;
 
+/* Mark new vma as mergeable */
+#ifndef CONFIG_KSM_MARK_NEW_VMA
+#define CONFIG_KSM_MARK_NEW_VMA DENY
+#endif
+unsigned mark_new_vma = CONFIG_KSM_MARK_NEW_VMA;
+
 #ifdef CONFIG_NUMA
 /* Zeroed when merging across nodes is not allowed */
 static unsigned int ksm_merge_across_nodes = 1;
@@ -2127,6 +2133,34 @@ static ssize_t pages_to_scan_store(struct kobject *kobj,
 }
 KSM_ATTR(pages_to_scan);
 
+static ssize_t mark_new_vma_show(struct kobject *kobj, struct kobj_attribute *attr,
+			char *buf)
+{
+	return sprintf(buf, "%u\n", mark_new_vma);
+}
+static ssize_t mark_new_vma_store(struct kobject *kobj,
+				  struct kobj_attribute *attr,
+				  const char *buf, size_t count)
+{
+	int err;
+	unsigned long flags;
+
+	err = kstrtoul(buf, 10, &flags);
+	if (err || flags > UINT_MAX || flags > ALLOW)
+		return -EINVAL;
+
+	/*
+	 * ALLOW = 1 - sets allow for mark new vma as
+	 * VM_MERGEABLE and adding it to ksm
+	 * DENY = 0 - disable it
+	 */
+	if (mark_new_vma != flags) {
+		mark_new_vma = flags;
+	}
+	return count;
+}
+KSM_ATTR(mark_new_vma);
+
 static ssize_t run_show(struct kobject *kobj, struct kobj_attribute *attr,
 			char *buf)
 {
@@ -2287,6 +2321,7 @@ static struct attribute *ksm_attrs[] = {
 	&pages_unshared_attr.attr,
 	&pages_volatile_attr.attr,
 	&full_scans_attr.attr,
+	&mark_new_vma_attr.attr,
 #ifdef CONFIG_NUMA
 	&merge_across_nodes_attr.attr,
 #endif
@@ -2323,7 +2358,9 @@ static int __init ksm_init(void)
 		goto out_free;
 	}
 #else
-	ksm_run = KSM_RUN_MERGE;	/* no way for user to start it */
+	/* no way for user to start it */
+	mark_new_vma = ALLOW;
+	ksm_run = KSM_RUN_MERGE;
 
 #endif /* CONFIG_SYSFS */
 
-- 
2.1.3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

[PATCH v2 3/3] KSM: Add config to control mark_new_vma

[Timofey Titovets]

Allowing to control mark_new_vma default value
Allowing work ksm on early allocated vmas

Signed-off-by: Timofey Titovets <nefelim4ag@gmail.com>
---
 mm/Kconfig | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/mm/Kconfig b/mm/Kconfig
index 1d1ae6b..90f40a6 100644
--- a/mm/Kconfig
+++ b/mm/Kconfig
@@ -340,6 +340,13 @@ config KSM
 	  until a program has madvised that an area is MADV_MERGEABLE, and
 	  root has set /sys/kernel/mm/ksm/run to 1 (if CONFIG_SYSFS is set).
 
+config KSM_MARK_NEW_VMA
+	int "Marking new vma pages as VM_MERGEABLE"
+	depends on KSM
+	default 0
+	range 0 1
+	help
+
 config DEFAULT_MMAP_MIN_ADDR
         int "Low address space to protect from user allocation"
 	depends on MMU
-- 
2.1.3

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>