[PATCH mm-unstable v1 0/5] mm/mglru: code cleanup and

[PATCH mm-unstable v1 0/5] mm/mglru: code cleanup and

[Kinsey Ho]

This series is a follow up to the comment:
https://lore.kernel.org/7baea98a-90ff-4c2d-9d00-d3ddc715c235@linux.vnet.ibm.com/

This provides MGLRU code cleanup and refactoring for better readability.

Kinsey Ho (5):
  mm/mglru: drop unused parameter
  mm/mglru: improve should_run_aging()
  mm/mglru: improve reset_mm_stats()
  mm/mglru: improve struct lru_gen_mm_walk
  mm/mglru: improve swappiness handling

 include/linux/mmzone.h |   6 +-
 mm/vmscan.c            | 142 ++++++++++++++++++++---------------------
 2 files changed, 74 insertions(+), 74 deletions(-)

-- 
2.43.0.687.g38aa6559b0-goog

[PATCH mm-unstable v1 1/5] mm/mglru: drop unused parameter

[Kinsey Ho]

struct scan_control *sc is currently passed into try_to_inc_max_seq()
and run_aging(). This parameter is not used.

Drop the unused parameter struct scan_control *sc. No functional change.

Signed-off-by: Kinsey Ho <kinseyho@google.com>
---
 mm/vmscan.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 54cb7685beba..9d15648561ec 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -3834,7 +3834,7 @@ static bool inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
 }
 
 static bool try_to_inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
-			       struct scan_control *sc, bool can_swap, bool force_scan)
+			       bool can_swap, bool force_scan)
 {
 	bool success;
 	struct lru_gen_mm_walk *walk;
@@ -4690,7 +4690,7 @@ static long get_nr_to_scan(struct lruvec *lruvec, struct scan_control *sc, bool
 		return nr_to_scan;
 
 	/* skip this lruvec as it's low on cold folios */
-	return try_to_inc_max_seq(lruvec, max_seq, sc, can_swap, false) ? -1 : 0;
+	return try_to_inc_max_seq(lruvec, max_seq, can_swap, false) ? -1 : 0;
 }
 
 static bool should_abort_scan(struct lruvec *lruvec, struct scan_control *sc)
@@ -5350,7 +5350,7 @@ static const struct seq_operations lru_gen_seq_ops = {
 	.show = lru_gen_seq_show,
 };
 
-static int run_aging(struct lruvec *lruvec, unsigned long seq, struct scan_control *sc,
+static int run_aging(struct lruvec *lruvec, unsigned long seq,
 		     bool can_swap, bool force_scan)
 {
 	DEFINE_MAX_SEQ(lruvec);
@@ -5365,7 +5365,7 @@ static int run_aging(struct lruvec *lruvec, unsigned long seq, struct scan_contr
 	if (!force_scan && min_seq[!can_swap] + MAX_NR_GENS - 1 <= max_seq)
 		return -ERANGE;
 
-	try_to_inc_max_seq(lruvec, max_seq, sc, can_swap, force_scan);
+	try_to_inc_max_seq(lruvec, max_seq, can_swap, force_scan);
 
 	return 0;
 }
@@ -5433,7 +5433,7 @@ static int run_cmd(char cmd, int memcg_id, int nid, unsigned long seq,
 
 	switch (cmd) {
 	case '+':
-		err = run_aging(lruvec, seq, sc, swappiness, opt);
+		err = run_aging(lruvec, seq, swappiness, opt);
 		break;
 	case '-':
 		err = run_eviction(lruvec, seq, sc, swappiness, opt);
-- 
2.43.0.687.g38aa6559b0-goog

[PATCH mm-unstable v1 2/5] mm/mglru: improve should_run_aging()

[Kinsey Ho]

scan_control *sc does not need to be passed into should_run_aging(), as
it provides only the reclaim priority. This can be moved to
get_nr_to_scan().

Refactor should_run_aging() and get_nr_to_scan() to improve code
readability. No functional changes.

Signed-off-by: Kinsey Ho <kinseyho@google.com>
---
 mm/vmscan.c | 25 +++++++++++--------------
 1 file changed, 11 insertions(+), 14 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 9d15648561ec..87de39e7410b 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -4602,14 +4602,13 @@ static int evict_folios(struct lruvec *lruvec, struct scan_control *sc, int swap
 }
 
 static bool should_run_aging(struct lruvec *lruvec, unsigned long max_seq,
-			     struct scan_control *sc, bool can_swap, unsigned long *nr_to_scan)
+			     bool can_swap, unsigned long *nr_to_scan)
 {
 	int gen, type, zone;
 	unsigned long old = 0;
 	unsigned long young = 0;
 	unsigned long total = 0;
 	struct lru_gen_folio *lrugen = &lruvec->lrugen;
-	struct mem_cgroup *memcg = lruvec_memcg(lruvec);
 	DEFINE_MIN_SEQ(lruvec);
 
 	/* whether this lruvec is completely out of cold folios */
@@ -4637,13 +4636,7 @@ static bool should_run_aging(struct lruvec *lruvec, unsigned long max_seq,
 		}
 	}
 
-	/* try to scrape all its memory if this memcg was deleted */
-	if (!mem_cgroup_online(memcg)) {
-		*nr_to_scan = total;
-		return false;
-	}
-
-	*nr_to_scan = total >> sc->priority;
+	*nr_to_scan = total;
 
 	/*
 	 * The aging tries to be lazy to reduce the overhead, while the eviction
@@ -4675,6 +4668,7 @@ static bool should_run_aging(struct lruvec *lruvec, unsigned long max_seq,
  */
 static long get_nr_to_scan(struct lruvec *lruvec, struct scan_control *sc, bool can_swap)
 {
+	bool success;
 	unsigned long nr_to_scan;
 	struct mem_cgroup *memcg = lruvec_memcg(lruvec);
 	DEFINE_MAX_SEQ(lruvec);
@@ -4682,14 +4676,17 @@ static long get_nr_to_scan(struct lruvec *lruvec, struct scan_control *sc, bool
 	if (mem_cgroup_below_min(sc->target_mem_cgroup, memcg))
 		return -1;
 
-	if (!should_run_aging(lruvec, max_seq, sc, can_swap, &nr_to_scan))
-		return nr_to_scan;
+	success = should_run_aging(lruvec, max_seq, can_swap, &nr_to_scan);
 
-	/* skip the aging path at the default priority */
-	if (sc->priority == DEF_PRIORITY)
+	/* try to scrape all its memory if this memcg was deleted */
+	if (nr_to_scan && !mem_cgroup_online(memcg))
 		return nr_to_scan;
 
-	/* skip this lruvec as it's low on cold folios */
+	/* try to get away with not aging at the default priority */
+	if (!success || sc->priority == DEF_PRIORITY)
+		return nr_to_scan >> sc->priority;
+
+	/* stop scanning this lruvec as it's low on cold folios */
 	return try_to_inc_max_seq(lruvec, max_seq, can_swap, false) ? -1 : 0;
 }
 
-- 
2.43.0.687.g38aa6559b0-goog

[PATCH mm-unstable v1 3/5] mm/mglru: improve reset_mm_stats()

[Kinsey Ho]

struct lruvec* is already a field of struct lru_gen_mm_walk.
Remove the parameter struct lruvec* into functions that already have
access to struct lru_gen_mm_walk*.

Also, we do not need to handle reset histogram stats when
!should_walk_mmu(). Remove the call to reset_mm_stats()
in iterate_mm_list_nowalk().

Signed-off-by: Kinsey Ho <kinseyho@google.com>
---
 mm/vmscan.c | 42 ++++++++++++++++++++++--------------------
 1 file changed, 22 insertions(+), 20 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 87de39e7410b..8c77c332a67a 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2897,38 +2897,37 @@ static struct mm_struct *get_next_mm(struct lru_gen_mm_walk *walk)
 
 #endif
 
-static void reset_mm_stats(struct lruvec *lruvec, struct lru_gen_mm_walk *walk, bool last)
+static void reset_mm_stats(struct lru_gen_mm_walk *walk, bool last)
 {
 	int i;
 	int hist;
+	struct lruvec *lruvec = walk->lruvec;
 	struct lru_gen_mm_state *mm_state = get_mm_state(lruvec);
 
 	lockdep_assert_held(&get_mm_list(lruvec_memcg(lruvec))->lock);
 
-	if (walk) {
-		hist = lru_hist_from_seq(walk->max_seq);
+	hist = lru_hist_from_seq(walk->max_seq);
 
-		for (i = 0; i < NR_MM_STATS; i++) {
-			WRITE_ONCE(mm_state->stats[hist][i],
-				   mm_state->stats[hist][i] + walk->mm_stats[i]);
-			walk->mm_stats[i] = 0;
-		}
+	for (i = 0; i < NR_MM_STATS; i++) {
+		WRITE_ONCE(mm_state->stats[hist][i],
+			   mm_state->stats[hist][i] + walk->mm_stats[i]);
+		walk->mm_stats[i] = 0;
 	}
 
 	if (NR_HIST_GENS > 1 && last) {
-		hist = lru_hist_from_seq(mm_state->seq + 1);
+		hist = lru_hist_from_seq(walk->max_seq + 1);
 
 		for (i = 0; i < NR_MM_STATS; i++)
 			WRITE_ONCE(mm_state->stats[hist][i], 0);
 	}
 }
 
-static bool iterate_mm_list(struct lruvec *lruvec, struct lru_gen_mm_walk *walk,
-			    struct mm_struct **iter)
+static bool iterate_mm_list(struct lru_gen_mm_walk *walk, struct mm_struct **iter)
 {
 	bool first = false;
 	bool last = false;
 	struct mm_struct *mm = NULL;
+	struct lruvec *lruvec = walk->lruvec;
 	struct mem_cgroup *memcg = lruvec_memcg(lruvec);
 	struct lru_gen_mm_list *mm_list = get_mm_list(memcg);
 	struct lru_gen_mm_state *mm_state = get_mm_state(lruvec);
@@ -2972,7 +2971,7 @@ static bool iterate_mm_list(struct lruvec *lruvec, struct lru_gen_mm_walk *walk,
 	} while (!(mm = get_next_mm(walk)));
 done:
 	if (*iter || last)
-		reset_mm_stats(lruvec, walk, last);
+		reset_mm_stats(walk, last);
 
 	spin_unlock(&mm_list->lock);
 
@@ -3002,7 +3001,6 @@ static bool iterate_mm_list_nowalk(struct lruvec *lruvec, unsigned long max_seq)
 		mm_state->head = NULL;
 		mm_state->tail = NULL;
 		WRITE_ONCE(mm_state->seq, mm_state->seq + 1);
-		reset_mm_stats(lruvec, NULL, true);
 		success = true;
 	}
 
@@ -3177,9 +3175,10 @@ static void update_batch_size(struct lru_gen_mm_walk *walk, struct folio *folio,
 	walk->nr_pages[new_gen][type][zone] += delta;
 }
 
-static void reset_batch_size(struct lruvec *lruvec, struct lru_gen_mm_walk *walk)
+static void reset_batch_size(struct lru_gen_mm_walk *walk)
 {
 	int gen, type, zone;
+	struct lruvec *lruvec = walk->lruvec;
 	struct lru_gen_folio *lrugen = &lruvec->lrugen;
 
 	walk->batched = 0;
@@ -3609,7 +3608,7 @@ static int walk_pud_range(p4d_t *p4d, unsigned long start, unsigned long end,
 	return -EAGAIN;
 }
 
-static void walk_mm(struct lruvec *lruvec, struct mm_struct *mm, struct lru_gen_mm_walk *walk)
+static void walk_mm(struct mm_struct *mm, struct lru_gen_mm_walk *walk)
 {
 	static const struct mm_walk_ops mm_walk_ops = {
 		.test_walk = should_skip_vma,
@@ -3618,6 +3617,7 @@ static void walk_mm(struct lruvec *lruvec, struct mm_struct *mm, struct lru_gen_
 	};
 
 	int err;
+	struct lruvec *lruvec = walk->lruvec;
 	struct mem_cgroup *memcg = lruvec_memcg(lruvec);
 
 	walk->next_addr = FIRST_USER_ADDRESS;
@@ -3646,7 +3646,7 @@ static void walk_mm(struct lruvec *lruvec, struct mm_struct *mm, struct lru_gen_
 
 		if (walk->batched) {
 			spin_lock_irq(&lruvec->lru_lock);
-			reset_batch_size(lruvec, walk);
+			reset_batch_size(walk);
 			spin_unlock_irq(&lruvec->lru_lock);
 		}
 
@@ -3874,9 +3874,9 @@ static bool try_to_inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
 	walk->force_scan = force_scan;
 
 	do {
-		success = iterate_mm_list(lruvec, walk, &mm);
+		success = iterate_mm_list(walk, &mm);
 		if (mm)
-			walk_mm(lruvec, mm, walk);
+			walk_mm(mm, walk);
 	} while (mm);
 done:
 	if (success) {
@@ -4576,8 +4576,10 @@ static int evict_folios(struct lruvec *lruvec, struct scan_control *sc, int swap
 	move_folios_to_lru(lruvec, &list);
 
 	walk = current->reclaim_state->mm_walk;
-	if (walk && walk->batched)
-		reset_batch_size(lruvec, walk);
+	if (walk && walk->batched) {
+		walk->lruvec = lruvec;
+		reset_batch_size(walk);
+	}
 
 	item = PGSTEAL_KSWAPD + reclaimer_offset();
 	if (!cgroup_reclaim(sc))
-- 
2.43.0.687.g38aa6559b0-goog

[PATCH mm-unstable v1 4/5] mm/mglru: improve struct lru_gen_mm_walk

[Kinsey Ho]

Rename max_seq to seq in struct lru_gen_mm_walk to keep consistent with
struct lru_gen_mm_state. Note that seq is not always up to date with
max_seq from lru_gen_folio.

No functional changes.

Signed-off-by: Kinsey Ho <kinseyho@google.com>
---
 include/linux/mmzone.h |  6 ++---
 mm/vmscan.c            | 50 ++++++++++++++++++++++--------------------
 2 files changed, 29 insertions(+), 27 deletions(-)

diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 22fd46c77840..e51c75270e82 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -468,7 +468,7 @@ enum {
 #define NR_BLOOM_FILTERS	2
 
 struct lru_gen_mm_state {
-	/* set to max_seq after each iteration */
+	/* synced with max_seq after each iteration */
 	unsigned long seq;
 	/* where the current iteration continues after */
 	struct list_head *head;
@@ -483,8 +483,8 @@ struct lru_gen_mm_state {
 struct lru_gen_mm_walk {
 	/* the lruvec under reclaim */
 	struct lruvec *lruvec;
-	/* unstable max_seq from lru_gen_folio */
-	unsigned long max_seq;
+	/* max_seq from lru_gen_folio: can be out of date */
+	unsigned long seq;
 	/* the next address within an mm to scan */
 	unsigned long next_addr;
 	/* to batch promoted pages */
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 8c77c332a67a..f1019afa7813 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2906,7 +2906,7 @@ static void reset_mm_stats(struct lru_gen_mm_walk *walk, bool last)
 
 	lockdep_assert_held(&get_mm_list(lruvec_memcg(lruvec))->lock);
 
-	hist = lru_hist_from_seq(walk->max_seq);
+	hist = lru_hist_from_seq(walk->seq);
 
 	for (i = 0; i < NR_MM_STATS; i++) {
 		WRITE_ONCE(mm_state->stats[hist][i],
@@ -2915,7 +2915,7 @@ static void reset_mm_stats(struct lru_gen_mm_walk *walk, bool last)
 	}
 
 	if (NR_HIST_GENS > 1 && last) {
-		hist = lru_hist_from_seq(walk->max_seq + 1);
+		hist = lru_hist_from_seq(walk->seq + 1);
 
 		for (i = 0; i < NR_MM_STATS; i++)
 			WRITE_ONCE(mm_state->stats[hist][i], 0);
@@ -2944,9 +2944,9 @@ static bool iterate_mm_list(struct lru_gen_mm_walk *walk, struct mm_struct **ite
 	 */
 	spin_lock(&mm_list->lock);
 
-	VM_WARN_ON_ONCE(mm_state->seq + 1 < walk->max_seq);
+	VM_WARN_ON_ONCE(mm_state->seq + 1 < walk->seq);
 
-	if (walk->max_seq <= mm_state->seq)
+	if (walk->seq <= mm_state->seq)
 		goto done;
 
 	if (!mm_state->head)
@@ -2976,7 +2976,7 @@ static bool iterate_mm_list(struct lru_gen_mm_walk *walk, struct mm_struct **ite
 	spin_unlock(&mm_list->lock);
 
 	if (mm && first)
-		reset_bloom_filter(mm_state, walk->max_seq + 1);
+		reset_bloom_filter(mm_state, walk->seq + 1);
 
 	if (*iter)
 		mmput_async(*iter);
@@ -2986,7 +2986,7 @@ static bool iterate_mm_list(struct lru_gen_mm_walk *walk, struct mm_struct **ite
 	return last;
 }
 
-static bool iterate_mm_list_nowalk(struct lruvec *lruvec, unsigned long max_seq)
+static bool iterate_mm_list_nowalk(struct lruvec *lruvec, unsigned long seq)
 {
 	bool success = false;
 	struct mem_cgroup *memcg = lruvec_memcg(lruvec);
@@ -2995,9 +2995,9 @@ static bool iterate_mm_list_nowalk(struct lruvec *lruvec, unsigned long max_seq)
 
 	spin_lock(&mm_list->lock);
 
-	VM_WARN_ON_ONCE(mm_state->seq + 1 < max_seq);
+	VM_WARN_ON_ONCE(mm_state->seq + 1 < seq);
 
-	if (max_seq > mm_state->seq) {
+	if (seq > mm_state->seq) {
 		mm_state->head = NULL;
 		mm_state->tail = NULL;
 		WRITE_ONCE(mm_state->seq, mm_state->seq + 1);
@@ -3348,7 +3348,8 @@ static bool walk_pte_range(pmd_t *pmd, unsigned long start, unsigned long end,
 	struct lru_gen_mm_walk *walk = args->private;
 	struct mem_cgroup *memcg = lruvec_memcg(walk->lruvec);
 	struct pglist_data *pgdat = lruvec_pgdat(walk->lruvec);
-	int old_gen, new_gen = lru_gen_from_seq(walk->max_seq);
+	DEFINE_MAX_SEQ(walk->lruvec);
+	int old_gen, new_gen = lru_gen_from_seq(max_seq);
 
 	pte = pte_offset_map_nolock(args->mm, pmd, start & PMD_MASK, &ptl);
 	if (!pte)
@@ -3415,7 +3416,8 @@ static void walk_pmd_range_locked(pud_t *pud, unsigned long addr, struct vm_area
 	struct lru_gen_mm_walk *walk = args->private;
 	struct mem_cgroup *memcg = lruvec_memcg(walk->lruvec);
 	struct pglist_data *pgdat = lruvec_pgdat(walk->lruvec);
-	int old_gen, new_gen = lru_gen_from_seq(walk->max_seq);
+	DEFINE_MAX_SEQ(walk->lruvec);
+	int old_gen, new_gen = lru_gen_from_seq(max_seq);
 
 	VM_WARN_ON_ONCE(pud_leaf(*pud));
 
@@ -3546,7 +3548,7 @@ static void walk_pmd_range(pud_t *pud, unsigned long start, unsigned long end,
 			walk_pmd_range_locked(pud, addr, vma, args, bitmap, &first);
 		}
 
-		if (!walk->force_scan && !test_bloom_filter(mm_state, walk->max_seq, pmd + i))
+		if (!walk->force_scan && !test_bloom_filter(mm_state, walk->seq, pmd + i))
 			continue;
 
 		walk->mm_stats[MM_NONLEAF_FOUND]++;
@@ -3557,7 +3559,7 @@ static void walk_pmd_range(pud_t *pud, unsigned long start, unsigned long end,
 		walk->mm_stats[MM_NONLEAF_ADDED]++;
 
 		/* carry over to the next generation */
-		update_bloom_filter(mm_state, walk->max_seq + 1, pmd + i);
+		update_bloom_filter(mm_state, walk->seq + 1, pmd + i);
 	}
 
 	walk_pmd_range_locked(pud, -1, vma, args, bitmap, &first);
@@ -3628,7 +3630,7 @@ static void walk_mm(struct mm_struct *mm, struct lru_gen_mm_walk *walk)
 		err = -EBUSY;
 
 		/* another thread might have called inc_max_seq() */
-		if (walk->max_seq != max_seq)
+		if (walk->seq != max_seq)
 			break;
 
 		/* folio_update_gen() requires stable folio_memcg() */
@@ -3765,7 +3767,7 @@ static bool try_to_inc_min_seq(struct lruvec *lruvec, bool can_swap)
 	return success;
 }
 
-static bool inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
+static bool inc_max_seq(struct lruvec *lruvec, unsigned long seq,
 			bool can_swap, bool force_scan)
 {
 	bool success;
@@ -3773,14 +3775,14 @@ static bool inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
 	int type, zone;
 	struct lru_gen_folio *lrugen = &lruvec->lrugen;
 restart:
-	if (max_seq < READ_ONCE(lrugen->max_seq))
+	if (seq < READ_ONCE(lrugen->max_seq))
 		return false;
 
 	spin_lock_irq(&lruvec->lru_lock);
 
 	VM_WARN_ON_ONCE(!seq_is_valid(lruvec));
 
-	success = max_seq == lrugen->max_seq;
+	success = seq == lrugen->max_seq;
 	if (!success)
 		goto unlock;
 
@@ -3833,7 +3835,7 @@ static bool inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
 	return success;
 }
 
-static bool try_to_inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
+static bool try_to_inc_max_seq(struct lruvec *lruvec, unsigned long seq,
 			       bool can_swap, bool force_scan)
 {
 	bool success;
@@ -3842,13 +3844,13 @@ static bool try_to_inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
 	struct lru_gen_folio *lrugen = &lruvec->lrugen;
 	struct lru_gen_mm_state *mm_state = get_mm_state(lruvec);
 
-	VM_WARN_ON_ONCE(max_seq > READ_ONCE(lrugen->max_seq));
+	VM_WARN_ON_ONCE(seq > READ_ONCE(lrugen->max_seq));
 
 	if (!mm_state)
-		return inc_max_seq(lruvec, max_seq, can_swap, force_scan);
+		return inc_max_seq(lruvec, seq, can_swap, force_scan);
 
 	/* see the comment in iterate_mm_list() */
-	if (max_seq <= READ_ONCE(mm_state->seq))
+	if (seq <= READ_ONCE(mm_state->seq))
 		return false;
 
 	/*
@@ -3858,18 +3860,18 @@ static bool try_to_inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
 	 * is less efficient, but it avoids bursty page faults.
 	 */
 	if (!should_walk_mmu()) {
-		success = iterate_mm_list_nowalk(lruvec, max_seq);
+		success = iterate_mm_list_nowalk(lruvec, seq);
 		goto done;
 	}
 
 	walk = set_mm_walk(NULL, true);
 	if (!walk) {
-		success = iterate_mm_list_nowalk(lruvec, max_seq);
+		success = iterate_mm_list_nowalk(lruvec, seq);
 		goto done;
 	}
 
 	walk->lruvec = lruvec;
-	walk->max_seq = max_seq;
+	walk->seq = seq;
 	walk->can_swap = can_swap;
 	walk->force_scan = force_scan;
 
@@ -3880,7 +3882,7 @@ static bool try_to_inc_max_seq(struct lruvec *lruvec, unsigned long max_seq,
 	} while (mm);
 done:
 	if (success) {
-		success = inc_max_seq(lruvec, max_seq, can_swap, force_scan);
+		success = inc_max_seq(lruvec, seq, can_swap, force_scan);
 		WARN_ON_ONCE(!success);
 	}
 
-- 
2.43.0.687.g38aa6559b0-goog

[PATCH mm-unstable v1 5/5] mm/mglru: improve swappiness handling

[Kinsey Ho]

The reclaimable number of anon pages used to set initial reclaim
priority is only based on get_swappiness(). Use can_reclaim_anon_pages()
to include NUMA node demotion.

Also move the swappiness handling of when !__GFP_IO in
try_to_shrink_lruvec() into isolate_folios().

Signed-off-by: Kinsey Ho <kinseyho@google.com>
---
 mm/vmscan.c | 19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index f1019afa7813..0daf932feb86 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -4307,7 +4307,7 @@ static bool isolate_folio(struct lruvec *lruvec, struct folio *folio, struct sca
 {
 	bool success;
 
-	/* swapping inhibited */
+	/* swap constrained */
 	if (!(sc->gfp_mask & __GFP_IO) &&
 	    (folio_test_dirty(folio) ||
 	     (folio_test_anon(folio) && !folio_test_swapcache(folio))))
@@ -4476,9 +4476,11 @@ static int isolate_folios(struct lruvec *lruvec, struct scan_control *sc, int sw
 	DEFINE_MIN_SEQ(lruvec);
 
 	/*
-	 * Try to make the obvious choice first. When anon and file are both
-	 * available from the same generation, interpret swappiness 1 as file
-	 * first and MAX_SWAPPINESS as anon first.
+	 * Try to make the obvious choice first, and if anon and file are both
+	 * available from the same generation,
+	 * 1. Interpret swappiness 1 as file first and MAX_SWAPPINESS as anon first.
+	 * 2. If !__GFP_IO, file first since clean pagecache is more likely to
+	 *    exist than clean swapcache.
 	 */
 	if (!swappiness)
 		type = LRU_GEN_FILE;
@@ -4488,6 +4490,8 @@ static int isolate_folios(struct lruvec *lruvec, struct scan_control *sc, int sw
 		type = LRU_GEN_FILE;
 	else if (swappiness == MAX_SWAPPINESS)
 		type = LRU_GEN_ANON;
+	else if (!(sc->gfp_mask & __GFP_IO))
+		type = LRU_GEN_FILE;
 	else
 		type = get_type_to_scan(lruvec, swappiness, &tier);
 
@@ -4731,10 +4735,6 @@ static bool try_to_shrink_lruvec(struct lruvec *lruvec, struct scan_control *sc)
 	unsigned long scanned = 0;
 	int swappiness = get_swappiness(lruvec, sc);
 
-	/* clean file folios are more likely to exist */
-	if (swappiness && !(sc->gfp_mask & __GFP_IO))
-		swappiness = 1;
-
 	while (true) {
 		int delta;
 
@@ -4897,7 +4897,6 @@ static void set_initial_priority(struct pglist_data *pgdat, struct scan_control
 {
 	int priority;
 	unsigned long reclaimable;
-	struct lruvec *lruvec = mem_cgroup_lruvec(NULL, pgdat);
 
 	if (sc->priority != DEF_PRIORITY || sc->nr_to_reclaim < MIN_LRU_BATCH)
 		return;
@@ -4907,7 +4906,7 @@ static void set_initial_priority(struct pglist_data *pgdat, struct scan_control
 	 * where reclaimed_to_scanned_ratio = inactive / total.
 	 */
 	reclaimable = node_page_state(pgdat, NR_INACTIVE_FILE);
-	if (get_swappiness(lruvec, sc))
+	if (can_reclaim_anon_pages(NULL, pgdat->node_id, sc))
 		reclaimable += node_page_state(pgdat, NR_INACTIVE_ANON);
 
 	/* round down reclaimable and round up sc->nr_to_reclaim */
-- 
2.43.0.687.g38aa6559b0-goog