[PATCH v3 0/4] mm/ksm: keep smart scan from losing duplicate pages

[PATCH v3 0/4] mm/ksm: keep smart scan from losing duplicate pages

[Qiliang Yuan]

With smart scan, which is on by default, two identical pages that become
mergeable some scans apart can stay unmerged for as long as they live:
each page times its skipped scans on its own, and once both are old they
may never be looked at in the same scan, which merging through the
unstable tree requires. This was first noticed as intermittent failures
of ksm_tests "KSM NUMA merging" in run_vmtests.sh.

Patch 1 picks the skipped scans by the global scan number, so all old
pages are looked at together every eighth scan. Over 420 pairs made
mergeable 0 to 20 scans apart, 361 and 334 merged before on two boots
and all of them after, at the cost of about 10% more ksmd time on memory
that never merges.

Patch 2 documents how smart scan picks the scans to skip.

Patch 3 moves waiting for KSM full scans into a vm_util helper, and
patch 4 adds a ksm_functional_tests case that fails 20/20 without patch
1 and 0/20 with it, on both boots.

Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
---
V2 -> V3:
- Document smart scan in Documentation/mm/ksm.rst, as Xu Xin suggested
- Wait for full scans and switch smart scan through vm_util helpers, as
  David suggested
- Rebase onto v7.4-rc1-345-geedaa7d61f1e and retest there
- Measure ksmd CPU time with schedstat, which puts the cost at about 10%

V1 -> V2:
- Rebase onto v7.4-rc1-132-g05441d697f7f and retest there
- Drop the ksm_tests -N -m 1 numbers, as it no longer fails on this base
  without the fix
- Update the ksmd cost to 10-12%, measured on this base

v2: https://lore.kernel.org/r/20261105-bug-mm-ksm-smart-scan-align-skips-v2-0-8a9780b6844e@gmail.com
v1: https://lore.kernel.org/r/20261016-bug-mm-ksm-smart-scan-align-skips-v1-0-2f91d76416be@gmail.com

---
Qiliang Yuan (4):
      mm/ksm: align smart scan skips with the scan number
      docs/mm: ksm: describe how smart scan picks the scans to skip
      selftests/mm: move waiting for KSM full scans into a helper
      selftests/mm: cover merging KSM pages aged by smart scan

 Documentation/mm/ksm.rst                          | 16 ++++
 mm/ksm.c                                          | 24 +++---
 tools/testing/selftests/mm/ksm_functional_tests.c | 89 ++++++++++++++++++++++-
 tools/testing/selftests/mm/vm_util.c              | 63 ++++++++++++----
 tools/testing/selftests/mm/vm_util.h              |  3 +
 5 files changed, 167 insertions(+), 28 deletions(-)
---
base-commit: eedaa7d61f1ec8ae053a9ece0d553785350ac486
change-id: 20261016-bug-mm-ksm-smart-scan-align-skips-cd01e5a9889a

Best regards,
-- 
Qiliang Yuan <odys.yuan@gmail.com>

[PATCH v3 1/4] mm/ksm: align smart scan skips with the scan number

[Qiliang Yuan]

Smart scan skips pages that KSM has failed to de-duplicate for a while.
Once a page is three scans old, its rmap_item counts down its own number
of scans to skip, 1, 2, 4 and then 8 as it ages, so an old page is looked
at every ninth scan.

The unstable tree is rebuilt on every scan, so two identical pages that
are not in the stable tree yet can only be merged in a scan that looks
at both. Each countdown runs from when its own page became mergeable, so
two pages that became mergeable at different times end up on different
phases and, once both are old, may never be looked at in the same scan
again. They then stay unmerged for as long as they live. The pairs below
are counted by making a page mergeable, waiting for 0 to 20 full scans
before doing the same with an identical page, 20 times for each wait,
and checking the pair after another 64 full scans, on two boots per
kernel.

Pick the skipped scans by the global scan number instead, with periods
of 2, 4 and 8 scans. The periods being powers of two, every old page is
looked at on each scan that is a multiple of 8, whatever its age, so any
two pages meet at least every eighth scan. The longest period goes from
nine scans to eight, so ksmd looks at more of the pages that never
merge. The rest of the table is measured on 64M of such pages, median of
five runs of 256 scans on each of three boots per kernel, CPU time from
schedstat:

                              before     after      after/before
  pairs merged (of 420)       361        420        116%
                              334        420        126%
  pages looked at per scan    1792       2048       114%
  ksmd CPU time per scan      6.30 ms    6.88 ms    109%
                              6.40 ms    6.96 ms    109%
                              6.15 ms    6.88 ms    112%

Fixes: 5e924ff54d08 ("mm/ksm: add "smart" page scanning mode")
Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
Reviewed-by: Xu Xin <xu.xin@linux.dev>
---
 mm/ksm.c | 24 ++++++++++--------------
 1 file changed, 10 insertions(+), 14 deletions(-)

diff --git a/mm/ksm.c b/mm/ksm.c
index d24b129932bf7..3053b438e9c10 100644
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -197,7 +197,6 @@ struct ksm_stable_node {
  * @head: pointer to stable_node heading this list in the stable tree
  * @hlist: link into hlist of rmap_items hanging off that stable_node
  * @age: number of scan iterations since creation (unstable node)
- * @remaining_skips: how many scans to skip (unstable node)
  * @linear_page_index: the original page's index before merged by KSM (stable node)
  */
 struct ksm_rmap_item {
@@ -214,7 +213,6 @@ struct ksm_rmap_item {
 		struct {
 			unsigned int oldchecksum;
 			rmap_age_t age;
-			rmap_age_t remaining_skips;
 		};			/* when unstable */
 		unsigned long linear_page_index;    /* for reverse mapping, when stable */
 	};
@@ -2457,16 +2455,14 @@ static struct ksm_rmap_item *get_next_rmap_item(struct ksm_mm_slot *mm_slot,
 }
 
 /*
- * Calculate skip age for the ksm page age. The age determines how often
- * de-duplicating has already been tried unsuccessfully. If the age is
- * smaller, the scanning of this page is skipped for less scans.
+ * Calculate how often a page of the given ksm page age is scanned. The age
+ * determines how often de-duplicating has already been tried
+ * unsuccessfully. The older the page, the fewer scans look at it.
  *
  * @age: rmap_item age of page
  */
-static unsigned int skip_age(rmap_age_t age)
+static unsigned int scan_period(rmap_age_t age)
 {
-	if (age <= 3)
-		return 1;
 	if (age <= 5)
 		return 2;
 	if (age <= 8)
@@ -2516,17 +2512,17 @@ static bool should_skip_rmap_item(struct folio *folio,
 		return false;
 
 	/*
-	 * Are we still allowed to skip? If not, then don't skip it
-	 * and determine how much more often we are allowed to skip next.
+	 * The unstable tree is rebuilt on every scan, so two identical pages
+	 * can only be merged if both are looked at in the same scan. Pick the
+	 * scans by the global scan number rather than per page: the periods
+	 * are powers of two, so pages of any age meet on every scan that is a
+	 * multiple of the longest one.
 	 */
-	if (!rmap_item->remaining_skips) {
-		rmap_item->remaining_skips = skip_age(age);
+	if (!(ksm_scan.seqnr & (scan_period(age) - 1)))
 		return false;
-	}
 
 	/* Skip this page */
 	ksm_pages_skipped++;
-	rmap_item->remaining_skips--;
 	remove_rmap_item_from_tree(rmap_item);
 	return true;
 }

-- 
2.43.0

[PATCH v3 2/4] docs/mm: ksm: describe how smart scan picks the scans to skip

[Qiliang Yuan]

Documentation/mm/ksm.rst covers the stable tree and its reverse mapping,
and Documentation/admin-guide/mm/ksm.rst only says that smart scan skips
pages KSM has failed to merge before.

Neither says how often an unmerged page is looked at, or that the scans
are picked by the global scan number so that identical pages of
different ages still meet in some scan, which merging through the
unstable tree depends on.

Add a section on smart scan that describes both.

Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
---
 Documentation/mm/ksm.rst | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/Documentation/mm/ksm.rst b/Documentation/mm/ksm.rst
index 5a80573162def..f954d221e8017 100644
--- a/Documentation/mm/ksm.rst
+++ b/Documentation/mm/ksm.rst
@@ -75,6 +75,22 @@ The whole list of stable_node "dups" linked in the stable_node
 The frequency of such scans is defined by
 ``stable_node_chains_prune_millisecs`` sysfs tunable.
 
+Smart scan
+----------
+With ``smart_scan`` enabled, ksmd skips pages that it has already tried
+to merge without success. The ``age`` of a page's ksm_rmap_item counts
+the scans that went by while the page stayed unmerged. A page younger
+than three scans is looked at on every scan; after that it is looked at
+every second scan up to an age of five, every fourth scan up to an age
+of eight, and every eighth scan from then on.
+
+The scans that look at a page are picked by the global scan number
+``ksm_scan.seqnr``, not counted per page. The unstable tree is rebuilt
+on every scan, so two identical pages can only be merged in a scan that
+looks at both. As the periods are powers of two, every page is looked
+at on each scan whose number is a multiple of eight, so two identical
+pages meet at least every eighth scan whatever their ages.
+
 Reference
 ---------
 .. kernel-doc:: mm/ksm.c

-- 
2.43.0

[PATCH v3 3/4] selftests/mm: move waiting for KSM full scans into a helper

[Qiliang Yuan]

ksm_start() enables ksmd and then polls full_scans until two more full
scans have completed, so that any merging that can happen has happened.

There is no helper to wait for a number of full scans while ksmd is
already running, so a test that needs one has to copy that loop.

Move the loop into ksm_wait_full_scans() and make ksm_start() call it
after enabling ksmd. The count is now read after ksmd starts rather than
before, which can only make ksm_start() wait longer.

Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
---
 tools/testing/selftests/mm/vm_util.c | 31 ++++++++++++++++++-------------
 tools/testing/selftests/mm/vm_util.h |  1 +
 2 files changed, 19 insertions(+), 13 deletions(-)

diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c
index 515031b8283da..1c2bac54696dd 100644
--- a/tools/testing/selftests/mm/vm_util.c
+++ b/tools/testing/selftests/mm/vm_util.c
@@ -984,33 +984,38 @@ int ksm_use_zero_pages(void)
 	return ret == 1 ? 0 : -errno;
 }
 
+int ksm_wait_full_scans(long nr)
+{
+	long start_scans, end_scans;
+
+	start_scans = ksm_get_full_scans();
+	if (start_scans < 0)
+		return start_scans;
+	do {
+		end_scans = ksm_get_full_scans();
+		if (end_scans < 0)
+			return end_scans;
+	} while (end_scans < start_scans + nr);
+
+	return 0;
+}
+
 int ksm_start(void)
 {
 	int ksm_fd;
 	ssize_t ret;
-	long start_scans, end_scans;
 
 	ksm_fd = open("/sys/kernel/mm/ksm/run", O_RDWR);
 	if (ksm_fd < 0)
 		return -errno;
 
-	/* Wait for two full scans such that any possible merging happened. */
-	start_scans = ksm_get_full_scans();
-	if (start_scans < 0) {
-		close(ksm_fd);
-		return start_scans;
-	}
 	ret = write(ksm_fd, "1", 1);
 	close(ksm_fd);
 	if (ret != 1)
 		return -errno;
-	do {
-		end_scans = ksm_get_full_scans();
-		if (end_scans < 0)
-			return end_scans;
-	} while (end_scans < start_scans + 2);
 
-	return 0;
+	/* Wait for two full scans such that any possible merging happened. */
+	return ksm_wait_full_scans(2);
 }
 
 int ksm_stop(void)
diff --git a/tools/testing/selftests/mm/vm_util.h b/tools/testing/selftests/mm/vm_util.h
index b5d59729d4327..7e00030077d2f 100644
--- a/tools/testing/selftests/mm/vm_util.h
+++ b/tools/testing/selftests/mm/vm_util.h
@@ -162,6 +162,7 @@ void *sys_mremap(void *old_address, unsigned long old_size,
 long ksm_get_self_zero_pages(void);
 long ksm_get_self_merging_pages(void);
 long ksm_get_full_scans(void);
+int ksm_wait_full_scans(long nr);
 int ksm_use_zero_pages(void);
 int ksm_start(void);
 int ksm_stop(void);

-- 
2.43.0

[PATCH v3 4/4] selftests/mm: cover merging KSM pages aged by smart scan

[Qiliang Yuan]

ksm_functional_tests checks merging and unmerging with ksmd's default
settings, but nothing there exercises smart scan, which skips pages KSM
has failed to merge for a while.

Smart scan used to time those skips per page, so two identical pages
that became mergeable some scans apart could end up never being looked
at in the same scan, and were never merged. "mm/ksm: align smart scan
skips with the scan number" fixed that.

Add ksm_get_smart_scan() and ksm_set_smart_scan() to vm_util, and a
case that enables smart scan, lets sixteen unique pages age for a dozen
scans, then makes a duplicate of each mergeable one scan apart, and
expects every pair to be merged 32 scans later:

  kernel               failures, two boots
  before the fix       20/20, 20/20
  with the fix          0/20,  0/20

Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
---
 tools/testing/selftests/mm/ksm_functional_tests.c | 89 ++++++++++++++++++++++-
 tools/testing/selftests/mm/vm_util.c              | 32 ++++++++
 tools/testing/selftests/mm/vm_util.h              |  2 +
 3 files changed, 122 insertions(+), 1 deletion(-)

diff --git a/tools/testing/selftests/mm/ksm_functional_tests.c b/tools/testing/selftests/mm/ksm_functional_tests.c
index 31c06c72203fd..135191f99962d 100644
--- a/tools/testing/selftests/mm/ksm_functional_tests.c
+++ b/tools/testing/selftests/mm/ksm_functional_tests.c
@@ -695,6 +695,92 @@ static void test_fork_ksm_merging_page_count(void)
 	munmap(map, size);
 }
 
+static void test_merge_aged_pairs(void)
+{
+	const unsigned int pairs = 16;
+	const unsigned long size = 2 * pairs * pagesize;
+	int old_smart_scan;
+	unsigned int i;
+	char *map;
+
+	ksft_print_msg("[RUN] %s\n", __func__);
+
+	old_smart_scan = ksm_get_smart_scan();
+	if (old_smart_scan < 0) {
+		ksft_test_result_skip("smart_scan not supported\n");
+		return;
+	}
+	if (ksm_set_smart_scan(true) < 0) {
+		ksft_test_result_fail("Enabling smart_scan failed\n");
+		return;
+	}
+
+	if (ksm_stop() < 0) {
+		ksft_test_result_fail("Disabling (unmerging) KSM failed\n");
+		goto restore;
+	}
+
+	map = mmap(NULL, size, PROT_READ | PROT_WRITE,
+		   MAP_PRIVATE | MAP_ANON, -1, 0);
+	if (map == MAP_FAILED) {
+		ksft_test_result_fail("mmap() failed\n");
+		goto restore;
+	}
+	if (madvise(map, size, MADV_NOHUGEPAGE) && errno != EINVAL) {
+		ksft_test_result_fail("MADV_NOHUGEPAGE failed\n");
+		goto unmap;
+	}
+
+	/*
+	 * Page i of the first half and page i of the second half form a
+	 * pair with content no other page shares.
+	 */
+	for (i = 0; i < pairs; i++) {
+		memset(map + i * pagesize, i + 1, pagesize);
+		memset(map + (pairs + i) * pagesize, i + 1, pagesize);
+	}
+
+	/* Let the first page of each pair age without a duplicate. */
+	if (madvise(map, size / 2, MADV_MERGEABLE)) {
+		ksft_test_result_fail("MADV_MERGEABLE failed\n");
+		goto unmap;
+	}
+	if (ksm_start() < 0 || ksm_wait_full_scans(12) < 0) {
+		ksft_test_result_fail("Running KSM failed\n");
+		goto unmap;
+	}
+
+	/*
+	 * Make the duplicates mergeable one scan apart so that each pair
+	 * starts out with a different distance in scans.
+	 */
+	for (i = 0; i < pairs; i++) {
+		if (madvise(map + (pairs + i) * pagesize, pagesize,
+			    MADV_MERGEABLE)) {
+			ksft_test_result_fail("MADV_MERGEABLE failed\n");
+			goto unmap;
+		}
+		if (ksm_wait_full_scans(1) < 0) {
+			ksft_test_result_fail("Running KSM failed\n");
+			goto unmap;
+		}
+	}
+
+	if (ksm_wait_full_scans(32) < 0) {
+		ksft_test_result_fail("Running KSM failed\n");
+		goto unmap;
+	}
+
+	ksft_test_result(ksm_get_self_merging_pages() == 2 * pairs,
+			 "Pages aged by smart scan got merged\n");
+unmap:
+	ksm_stop();
+	munmap(map, size);
+restore:
+	if (ksm_set_smart_scan(old_smart_scan) < 0)
+		ksft_print_msg("Restoring smart_scan failed\n");
+}
+
 static void init_global_file_handles(void)
 {
 	mem_fd = open("/proc/self/mem", O_RDWR);
@@ -720,7 +806,7 @@ static void init_global_file_handles(void)
 
 int main(int argc, char **argv)
 {
-	unsigned int tests = 9;
+	unsigned int tests = 10;
 	int err;
 
 	if (argc > 1 && !strcmp(argv[1], FORK_EXEC_CHILD_PRG_NAME)) {
@@ -753,6 +839,7 @@ int main(int argc, char **argv)
 	test_prctl_fork_exec();
 	test_prctl_unmerge();
 	test_fork_ksm_merging_page_count();
+	test_merge_aged_pairs();
 
 	err = ksft_get_fail_cnt();
 	if (err)
diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c
index 1c2bac54696dd..490e3c7161e8d 100644
--- a/tools/testing/selftests/mm/vm_util.c
+++ b/tools/testing/selftests/mm/vm_util.c
@@ -984,6 +984,38 @@ int ksm_use_zero_pages(void)
 	return ret == 1 ? 0 : -errno;
 }
 
+int ksm_get_smart_scan(void)
+{
+	int smart_scan_fd;
+	char buf;
+	ssize_t ret;
+
+	smart_scan_fd = open("/sys/kernel/mm/ksm/smart_scan", O_RDONLY);
+	if (smart_scan_fd < 0)
+		return -errno;
+
+	ret = pread(smart_scan_fd, &buf, 1, 0);
+	close(smart_scan_fd);
+	if (ret != 1)
+		return -errno;
+
+	return buf == '1';
+}
+
+int ksm_set_smart_scan(bool enable)
+{
+	int smart_scan_fd;
+	ssize_t ret;
+
+	smart_scan_fd = open("/sys/kernel/mm/ksm/smart_scan", O_WRONLY);
+	if (smart_scan_fd < 0)
+		return -errno;
+
+	ret = write(smart_scan_fd, enable ? "1" : "0", 1);
+	close(smart_scan_fd);
+	return ret == 1 ? 0 : -errno;
+}
+
 int ksm_wait_full_scans(long nr)
 {
 	long start_scans, end_scans;
diff --git a/tools/testing/selftests/mm/vm_util.h b/tools/testing/selftests/mm/vm_util.h
index 7e00030077d2f..ab1325dc65800 100644
--- a/tools/testing/selftests/mm/vm_util.h
+++ b/tools/testing/selftests/mm/vm_util.h
@@ -163,6 +163,8 @@ long ksm_get_self_zero_pages(void);
 long ksm_get_self_merging_pages(void);
 long ksm_get_full_scans(void);
 int ksm_wait_full_scans(long nr);
+int ksm_get_smart_scan(void);
+int ksm_set_smart_scan(bool enable);
 int ksm_use_zero_pages(void);
 int ksm_start(void);
 int ksm_stop(void);

-- 
2.43.0

Re: [PATCH v3 2/4] docs/mm: ksm: describe how smart scan picks the scans to skip

[David Hildenbrand (Arm)]

On 11/7/26 17:57, Qiliang Yuan wrote:
> Documentation/mm/ksm.rst covers the stable tree and its reverse mapping,
> and Documentation/admin-guide/mm/ksm.rst only says that smart scan skips
> pages KSM has failed to merge before.
> 
> Neither says how often an unmerged page is looked at, or that the scans
> are picked by the global scan number so that identical pages of
> different ages still meet in some scan, which merging through the
> unstable tree depends on.
> 
> Add a section on smart scan that describes both.
> 
> Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
> ---
>  Documentation/mm/ksm.rst | 16 ++++++++++++++++
>  1 file changed, 16 insertions(+)
> 
> diff --git a/Documentation/mm/ksm.rst b/Documentation/mm/ksm.rst
> index 5a80573162def..f954d221e8017 100644
> --- a/Documentation/mm/ksm.rst
> +++ b/Documentation/mm/ksm.rst
> @@ -75,6 +75,22 @@ The whole list of stable_node "dups" linked in the stable_node
>  The frequency of such scans is defined by
>  ``stable_node_chains_prune_millisecs`` sysfs tunable.
>  
> +Smart scan
> +----------
> +With ``smart_scan`` enabled, ksmd skips pages that it has already tried
> +to merge without success. The ``age`` of a page's ksm_rmap_item counts
> +the scans that went by while the page stayed unmerged. A page younger
> +than three scans is looked at on every scan; after that it is looked at
> +every second scan up to an age of five, every fourth scan up to an age
> +of eight, and every eighth scan from then on.
> +
> +The scans that look at a page are picked by the global scan number
> +``ksm_scan.seqnr``, not counted per page. The unstable tree is rebuilt
> +on every scan, so two identical pages can only be merged in a scan that
> +looks at both. As the periods are powers of two, every page is looked
> +at on each scan whose number is a multiple of eight, so two identical
> +pages meet at least every eighth scan whatever their ages.

Yeah, much clearer, thanks

Reviewed-by: David Hildenbrand (Arm) <david@kernel.org>

-- 
Cheers,

David

Re: [PATCH v3 3/4] selftests/mm: move waiting for KSM full scans into a helper

[David Hildenbrand (Arm)]

On 11/7/26 17:57, Qiliang Yuan wrote:
> ksm_start() enables ksmd and then polls full_scans until two more full
> scans have completed, so that any merging that can happen has happened.
> 
> There is no helper to wait for a number of full scans while ksmd is
> already running, so a test that needs one has to copy that loop.
> 
> Move the loop into ksm_wait_full_scans() and make ksm_start() call it
> after enabling ksmd. The count is now read after ksmd starts rather than
> before, which can only make ksm_start() wait longer.

Right, we don't care about that

Reviewed-by: David Hildenbrand (Arm) <david@kernel.org>

-- 
Cheers,

David

Re: [PATCH v3 1/4] mm/ksm: align smart scan skips with the scan number

[David Hildenbrand (Arm)]

On 11/7/26 17:57, Qiliang Yuan wrote:
> Smart scan skips pages that KSM has failed to de-duplicate for a while.
> Once a page is three scans old, its rmap_item counts down its own number
> of scans to skip, 1, 2, 4 and then 8 as it ages, so an old page is looked
> at every ninth scan.
> 
> The unstable tree is rebuilt on every scan, so two identical pages that
> are not in the stable tree yet can only be merged in a scan that looks
> at both. Each countdown runs from when its own page became mergeable, so
> two pages that became mergeable at different times end up on different
> phases and, once both are old, may never be looked at in the same scan
> again. They then stay unmerged for as long as they live. The pairs below
> are counted by making a page mergeable, waiting for 0 to 20 full scans
> before doing the same with an identical page, 20 times for each wait,
> and checking the pair after another 64 full scans, on two boots per
> kernel.
> 
> Pick the skipped scans by the global scan number instead, with periods
> of 2, 4 and 8 scans. The periods being powers of two, every old page is
> looked at on each scan that is a multiple of 8, whatever its age, so any
> two pages meet at least every eighth scan. The longest period goes from
> nine scans to eight, so ksmd looks at more of the pages that never
> merge. The rest of the table is measured on 64M of such pages, median of
> five runs of 256 scans on each of three boots per kernel, CPU time from
> schedstat:
> 
>                               before     after      after/before
>   pairs merged (of 420)       361        420        116%
>                               334        420        126%
>   pages looked at per scan    1792       2048       114%
>   ksmd CPU time per scan      6.30 ms    6.88 ms    109%
>                               6.40 ms    6.96 ms    109%
>                               6.15 ms    6.88 ms    112%
> 
> Fixes: 5e924ff54d08 ("mm/ksm: add "smart" page scanning mode")
> Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
> Reviewed-by: Xu Xin <xu.xin@linux.dev>
> ---

Reviewed-by: David Hildenbrand (Arm) <david@kernel.org>

-- 
Cheers,

David

Re: [PATCH v3 1/4] mm/ksm: align smart scan skips with the scan number

[David Hildenbrand (Arm)]

On 11/7/26 17:57, Qiliang Yuan wrote:
> Smart scan skips pages that KSM has failed to de-duplicate for a while.
> Once a page is three scans old, its rmap_item counts down its own number
> of scans to skip, 1, 2, 4 and then 8 as it ages, so an old page is looked
> at every ninth scan.
> 
> The unstable tree is rebuilt on every scan, so two identical pages that
> are not in the stable tree yet can only be merged in a scan that looks
> at both. Each countdown runs from when its own page became mergeable, so
> two pages that became mergeable at different times end up on different
> phases and, once both are old, may never be looked at in the same scan
> again. They then stay unmerged for as long as they live. The pairs below
> are counted by making a page mergeable, waiting for 0 to 20 full scans
> before doing the same with an identical page, 20 times for each wait,
> and checking the pair after another 64 full scans, on two boots per
> kernel.
> 
> Pick the skipped scans by the global scan number instead, with periods
> of 2, 4 and 8 scans. The periods being powers of two, every old page is
> looked at on each scan that is a multiple of 8, whatever its age, so any
> two pages meet at least every eighth scan. The longest period goes from
> nine scans to eight, so ksmd looks at more of the pages that never
> merge. The rest of the table is measured on 64M of such pages, median of
> five runs of 256 scans on each of three boots per kernel, CPU time from
> schedstat:
> 
>                               before     after      after/before
>   pairs merged (of 420)       361        420        116%
>                               334        420        126%
>   pages looked at per scan    1792       2048       114%
>

Oh, I was wondering whether this should actually go to stable? Do you think the
above is rather an extreme case, and most users won't really run into it that
severely?

-- 
Cheers,

David

Re: [PATCH v3 4/4] selftests/mm: cover merging KSM pages aged by smart scan

[David Hildenbrand (Arm)]

On 11/7/26 17:57, Qiliang Yuan wrote:
> ksm_functional_tests checks merging and unmerging with ksmd's default
> settings, but nothing there exercises smart scan, which skips pages KSM
> has failed to merge for a while.
> 
> Smart scan used to time those skips per page, so two identical pages
> that became mergeable some scans apart could end up never being looked
> at in the same scan, and were never merged. "mm/ksm: align smart scan
> skips with the scan number" fixed that.
> 
> Add ksm_get_smart_scan() and ksm_set_smart_scan() to vm_util, and a
> case that enables smart scan, lets sixteen unique pages age for a dozen
> scans, then makes a duplicate of each mergeable one scan apart, and
> expects every pair to be merged 32 scans later:
> 
>   kernel               failures, two boots
>   before the fix       20/20, 20/20
>   with the fix          0/20,  0/20
> 
> Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
> ---

Reviewed-by: David Hildenbrand (Arm) <david@kernel.org>

-- 
Cheers,

David

Re: [PATCH v3 0/4] mm/ksm: keep smart scan from losing duplicate pages

[David Hildenbrand (Arm)]

On 11/7/26 17:56, Qiliang Yuan wrote:
> With smart scan, which is on by default, two identical pages that become
> mergeable some scans apart can stay unmerged for as long as they live:
> each page times its skipped scans on its own, and once both are old they
> may never be looked at in the same scan, which merging through the
> unstable tree requires. This was first noticed as intermittent failures
> of ksm_tests "KSM NUMA merging" in run_vmtests.sh.
> 
> Patch 1 picks the skipped scans by the global scan number, so all old
> pages are looked at together every eighth scan. Over 420 pairs made
> mergeable 0 to 20 scans apart, 361 and 334 merged before on two boots
> and all of them after, at the cost of about 10% more ksmd time on memory
> that never merges.
> 
> Patch 2 documents how smart scan picks the scans to skip.
> 
> Patch 3 moves waiting for KSM full scans into a vm_util helper, and
> patch 4 adds a ksm_functional_tests case that fails 20/20 without patch
> 1 and 0/20 with it, on both boots.
> 
> Signed-off-by: Qiliang Yuan <odys.yuan@gmail.com>
> ---

I'll wait a bit for more review before I queue this, thanks.

-- 
Cheers,

David