[v8,2/6] ksm: count all zero pages placed by KSM

Message ID 20230522105229.4066-1-yang.yang29@zte.com.cn
State New
Headers
Series ksm: support tracking KSM-placed zero-pages |

Commit Message

Yang Yang May 22, 2023, 10:52 a.m. UTC
  From: xu xin <xu.xin16@zte.com.cn>

As pages_sharing and pages_shared don't include the number of zero pages
merged by KSM, we cannot know how many pages are zero pages placed by KSM
when enabling use_zero_pages, which leads to KSM not being transparent with
all actual merged pages by KSM. In the early days of use_zero_pages,
zero-pages was unable to get unshared by the ways like MADV_UNMERGEABLE so
it's hard to count how many times one of those zeropages was then unmerged.

But now, unsharing KSM-placed zero page accurately has been achieved, so we
can easily count both how many times a page full of zeroes was merged with
zero-page and how many times one of those pages was then unmerged. and so,
it helps to estimate memory demands when each and every shared page could
get unshared.

So we add ksm_zero_pages under /sys/kernel/mm/ksm/ to show the number
of all zero pages placed by KSM.

v7->v8:
Handle the case when khugepaged replaces a shared zeropage by a THP.

Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Suggested-by: David Hildenbrand <david@redhat.com>
Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>
Cc: Xuexin Jiang <jiang.xuexin@zte.com.cn>
Reviewed-by: Xiaokai Ran <ran.xiaokai@zte.com.cn>
Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
---
 include/linux/ksm.h | 17 +++++++++++++++++
 mm/khugepaged.c     |  3 +++
 mm/ksm.c            | 12 ++++++++++++
 mm/memory.c         |  7 ++++++-
 4 files changed, 38 insertions(+), 1 deletion(-)
  

Comments

David Hildenbrand May 23, 2023, 9:47 a.m. UTC | #1
On 22.05.23 12:52, Yang Yang wrote:
> From: xu xin <xu.xin16@zte.com.cn>
> 
> As pages_sharing and pages_shared don't include the number of zero pages
> merged by KSM, we cannot know how many pages are zero pages placed by KSM
> when enabling use_zero_pages, which leads to KSM not being transparent with
> all actual merged pages by KSM. In the early days of use_zero_pages,
> zero-pages was unable to get unshared by the ways like MADV_UNMERGEABLE so
> it's hard to count how many times one of those zeropages was then unmerged.
> 
> But now, unsharing KSM-placed zero page accurately has been achieved, so we
> can easily count both how many times a page full of zeroes was merged with
> zero-page and how many times one of those pages was then unmerged. and so,
> it helps to estimate memory demands when each and every shared page could
> get unshared.
> 
> So we add ksm_zero_pages under /sys/kernel/mm/ksm/ to show the number
> of all zero pages placed by KSM.
> 
> v7->v8:
> Handle the case when khugepaged replaces a shared zeropage by a THP.
> 
> Signed-off-by: xu xin <xu.xin16@zte.com.cn>
> Suggested-by: David Hildenbrand <david@redhat.com>
> Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>
> Cc: Xuexin Jiang <jiang.xuexin@zte.com.cn>
> Reviewed-by: Xiaokai Ran <ran.xiaokai@zte.com.cn>
> Reviewed-by: Yang Yang <yang.yang29@zte.com.cn>
> ---
>   include/linux/ksm.h | 17 +++++++++++++++++
>   mm/khugepaged.c     |  3 +++
>   mm/ksm.c            | 12 ++++++++++++
>   mm/memory.c         |  7 ++++++-
>   4 files changed, 38 insertions(+), 1 deletion(-)
> 
> diff --git a/include/linux/ksm.h b/include/linux/ksm.h
> index 7989200cdbb7..1adcae0205e3 100644
> --- a/include/linux/ksm.h
> +++ b/include/linux/ksm.h
> @@ -29,6 +29,16 @@ void __ksm_exit(struct mm_struct *mm);
>   /* use pte_mkdirty to track a KSM-placed zero page */
>   #define set_pte_ksm_zero(pte)	pte_mkdirty(pte_mkspecial(pte))
>   #define is_ksm_zero_pte(pte)	(is_zero_pfn(pte_pfn(pte)) && pte_dirty(pte))
> +extern unsigned long ksm_zero_pages;
> +static inline void inc_ksm_zero_pages(void)
> +{
> +	ksm_zero_pages++;
> +}
> +

No need to export the inc, just inline this.

> +static inline void dec_ksm_zero_pages(void)
> +{
> +	ksm_zero_pages--;
> +}
>   
>   static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm)
>   {
> @@ -100,6 +110,13 @@ static inline void ksm_exit(struct mm_struct *mm)
>   
>   #define set_pte_ksm_zero(pte)	pte_mkspecial(pte)
>   #define is_ksm_zero_pte(pte)	0
> +static inline void inc_ksm_zero_pages(void)
> +{
> +}
> +
> +static inline void dec_ksm_zero_pages(void)
> +{
> +}
>   
>   #ifdef CONFIG_MEMORY_FAILURE
>   static inline void collect_procs_ksm(struct page *page,
> diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> index 6b9d39d65b73..ba0d077b6951 100644
> --- a/mm/khugepaged.c
> +++ b/mm/khugepaged.c
> @@ -19,6 +19,7 @@
>   #include <linux/page_table_check.h>
>   #include <linux/swapops.h>
>   #include <linux/shmem_fs.h>
> +#include <linux/ksm.h>
>   
>   #include <asm/tlb.h>
>   #include <asm/pgalloc.h>
> @@ -711,6 +712,8 @@ static void __collapse_huge_page_copy_succeeded(pte_t *pte,
>   				spin_lock(ptl);
>   				ptep_clear(vma->vm_mm, address, _pte);
>   				spin_unlock(ptl);
> +				if (is_ksm_zero_pte(pteval))
> +					dec_ksm_zero_pages();
>   			}
>   		} else {
>   			src_page = pte_page(pteval);
> diff --git a/mm/ksm.c b/mm/ksm.c
> index 9962f5962afd..2ca7e8860faa 100644
> --- a/mm/ksm.c
> +++ b/mm/ksm.c
> @@ -278,6 +278,9 @@ static unsigned int zero_checksum __read_mostly;
>   /* Whether to merge empty (zeroed) pages with actual zero pages */
>   static bool ksm_use_zero_pages __read_mostly;
>   
> +/* The number of zero pages which is placed by KSM */
> +unsigned long ksm_zero_pages;
> +
>   #ifdef CONFIG_NUMA
>   /* Zeroed when merging across nodes is not allowed */
>   static unsigned int ksm_merge_across_nodes = 1;
> @@ -1223,6 +1226,7 @@ static int replace_page(struct vm_area_struct *vma, struct page *page,
>   	} else {
>   		newpte = set_pte_ksm_zero(pfn_pte(page_to_pfn(kpage),
>   					       vma->vm_page_prot));
> +		inc_ksm_zero_pages();
>   		/*
>   		 * We're replacing an anonymous page with a zero page, which is
>   		 * not anonymous. We need to do proper accounting otherwise we
> @@ -3350,6 +3354,13 @@ static ssize_t pages_volatile_show(struct kobject *kobj,
>   }
>   KSM_ATTR_RO(pages_volatile);
>   
> +static ssize_t ksm_zero_pages_show(struct kobject *kobj,
> +				struct kobj_attribute *attr, char *buf)
> +{
> +	return sysfs_emit(buf, "%ld\n", ksm_zero_pages);
> +}
> +KSM_ATTR_RO(ksm_zero_pages);
> +
>   static ssize_t general_profit_show(struct kobject *kobj,
>   				   struct kobj_attribute *attr, char *buf)
>   {
> @@ -3417,6 +3428,7 @@ static struct attribute *ksm_attrs[] = {
>   	&pages_sharing_attr.attr,
>   	&pages_unshared_attr.attr,
>   	&pages_volatile_attr.attr,
> +	&ksm_zero_pages_attr.attr,
>   	&full_scans_attr.attr,
>   #ifdef CONFIG_NUMA
>   	&merge_across_nodes_attr.attr,
> diff --git a/mm/memory.c b/mm/memory.c
> index 8358f3b853f2..058b416adf24 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -1415,8 +1415,11 @@ static unsigned long zap_pte_range(struct mmu_gather *tlb,
>   			tlb_remove_tlb_entry(tlb, pte, addr);
>   			zap_install_uffd_wp_if_needed(vma, addr, pte, details,
>   						      ptent);
> -			if (unlikely(!page))
> +			if (unlikely(!page)) {
> +				if (is_ksm_zero_pte(ptent))
> +					dec_ksm_zero_pages();
>   				continue;
> +			}
>   
>   			delay_rmap = 0;
>   			if (!PageAnon(page)) {
> @@ -3120,6 +3123,8 @@ static vm_fault_t wp_page_copy(struct vm_fault *vmf)
>   				inc_mm_counter(mm, MM_ANONPAGES);
>   			}
>   		} else {
> +			if (is_ksm_zero_pte(vmf->orig_pte))
> +				dec_ksm_zero_pages();
>   			inc_mm_counter(mm, MM_ANONPAGES);
>   		}
>   		flush_cache_page(vma, vmf->address, pte_pfn(vmf->orig_pte));

Can we maybe avoid exporting the dec semantics and rather add a callback 
to KSM? Ideally, we'd even distill that down to a single call, and 
handle the details in ksm.h. Maybe simply:

ksm_notify_unmap_zero_page(vmf->orig_pte);

and then just have in ksm.h

static inline void ksm_notify_unmap_zero_page(pte_t pte)
{
	if (is_ksm_zero_pte(pte))
		ksm_zero_pages--;
}
  
David Hildenbrand May 23, 2023, 9:51 a.m. UTC | #2
On 22.05.23 12:52, Yang Yang wrote:
> From: xu xin <xu.xin16@zte.com.cn>
> 
> As pages_sharing and pages_shared don't include the number of zero pages
> merged by KSM, we cannot know how many pages are zero pages placed by KSM
> when enabling use_zero_pages, which leads to KSM not being transparent with
> all actual merged pages by KSM. In the early days of use_zero_pages,
> zero-pages was unable to get unshared by the ways like MADV_UNMERGEABLE so
> it's hard to count how many times one of those zeropages was then unmerged.
> 
> But now, unsharing KSM-placed zero page accurately has been achieved, so we
> can easily count both how many times a page full of zeroes was merged with
> zero-page and how many times one of those pages was then unmerged. and so,
> it helps to estimate memory demands when each and every shared page could
> get unshared.
> 
> So we add ksm_zero_pages under /sys/kernel/mm/ksm/ to show the number
> of all zero pages placed by KSM.
> 
> v7->v8:
> Handle the case when khugepaged replaces a shared zeropage by a THP.
> 

Oh, and just a note, such version comments should go below the "--", 
such that they will automatically get dropped when applying the patch.

(Usually, version information in the cover letter is sufficient :) )
  

Patch

diff --git a/include/linux/ksm.h b/include/linux/ksm.h
index 7989200cdbb7..1adcae0205e3 100644
--- a/include/linux/ksm.h
+++ b/include/linux/ksm.h
@@ -29,6 +29,16 @@  void __ksm_exit(struct mm_struct *mm);
 /* use pte_mkdirty to track a KSM-placed zero page */
 #define set_pte_ksm_zero(pte)	pte_mkdirty(pte_mkspecial(pte))
 #define is_ksm_zero_pte(pte)	(is_zero_pfn(pte_pfn(pte)) && pte_dirty(pte))
+extern unsigned long ksm_zero_pages;
+static inline void inc_ksm_zero_pages(void)
+{
+	ksm_zero_pages++;
+}
+
+static inline void dec_ksm_zero_pages(void)
+{
+	ksm_zero_pages--;
+}
 
 static inline int ksm_fork(struct mm_struct *mm, struct mm_struct *oldmm)
 {
@@ -100,6 +110,13 @@  static inline void ksm_exit(struct mm_struct *mm)
 
 #define set_pte_ksm_zero(pte)	pte_mkspecial(pte)
 #define is_ksm_zero_pte(pte)	0
+static inline void inc_ksm_zero_pages(void)
+{
+}
+
+static inline void dec_ksm_zero_pages(void)
+{
+}
 
 #ifdef CONFIG_MEMORY_FAILURE
 static inline void collect_procs_ksm(struct page *page,
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index 6b9d39d65b73..ba0d077b6951 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c
@@ -19,6 +19,7 @@ 
 #include <linux/page_table_check.h>
 #include <linux/swapops.h>
 #include <linux/shmem_fs.h>
+#include <linux/ksm.h>
 
 #include <asm/tlb.h>
 #include <asm/pgalloc.h>
@@ -711,6 +712,8 @@  static void __collapse_huge_page_copy_succeeded(pte_t *pte,
 				spin_lock(ptl);
 				ptep_clear(vma->vm_mm, address, _pte);
 				spin_unlock(ptl);
+				if (is_ksm_zero_pte(pteval))
+					dec_ksm_zero_pages();
 			}
 		} else {
 			src_page = pte_page(pteval);
diff --git a/mm/ksm.c b/mm/ksm.c
index 9962f5962afd..2ca7e8860faa 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -278,6 +278,9 @@  static unsigned int zero_checksum __read_mostly;
 /* Whether to merge empty (zeroed) pages with actual zero pages */
 static bool ksm_use_zero_pages __read_mostly;
 
+/* The number of zero pages which is placed by KSM */
+unsigned long ksm_zero_pages;
+
 #ifdef CONFIG_NUMA
 /* Zeroed when merging across nodes is not allowed */
 static unsigned int ksm_merge_across_nodes = 1;
@@ -1223,6 +1226,7 @@  static int replace_page(struct vm_area_struct *vma, struct page *page,
 	} else {
 		newpte = set_pte_ksm_zero(pfn_pte(page_to_pfn(kpage),
 					       vma->vm_page_prot));
+		inc_ksm_zero_pages();
 		/*
 		 * We're replacing an anonymous page with a zero page, which is
 		 * not anonymous. We need to do proper accounting otherwise we
@@ -3350,6 +3354,13 @@  static ssize_t pages_volatile_show(struct kobject *kobj,
 }
 KSM_ATTR_RO(pages_volatile);
 
+static ssize_t ksm_zero_pages_show(struct kobject *kobj,
+				struct kobj_attribute *attr, char *buf)
+{
+	return sysfs_emit(buf, "%ld\n", ksm_zero_pages);
+}
+KSM_ATTR_RO(ksm_zero_pages);
+
 static ssize_t general_profit_show(struct kobject *kobj,
 				   struct kobj_attribute *attr, char *buf)
 {
@@ -3417,6 +3428,7 @@  static struct attribute *ksm_attrs[] = {
 	&pages_sharing_attr.attr,
 	&pages_unshared_attr.attr,
 	&pages_volatile_attr.attr,
+	&ksm_zero_pages_attr.attr,
 	&full_scans_attr.attr,
 #ifdef CONFIG_NUMA
 	&merge_across_nodes_attr.attr,
diff --git a/mm/memory.c b/mm/memory.c
index 8358f3b853f2..058b416adf24 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -1415,8 +1415,11 @@  static unsigned long zap_pte_range(struct mmu_gather *tlb,
 			tlb_remove_tlb_entry(tlb, pte, addr);
 			zap_install_uffd_wp_if_needed(vma, addr, pte, details,
 						      ptent);
-			if (unlikely(!page))
+			if (unlikely(!page)) {
+				if (is_ksm_zero_pte(ptent))
+					dec_ksm_zero_pages();
 				continue;
+			}
 
 			delay_rmap = 0;
 			if (!PageAnon(page)) {
@@ -3120,6 +3123,8 @@  static vm_fault_t wp_page_copy(struct vm_fault *vmf)
 				inc_mm_counter(mm, MM_ANONPAGES);
 			}
 		} else {
+			if (is_ksm_zero_pte(vmf->orig_pte))
+				dec_ksm_zero_pages();
 			inc_mm_counter(mm, MM_ANONPAGES);
 		}
 		flush_cache_page(vma, vmf->address, pte_pfn(vmf->orig_pte));