[PATCH -next v2 0/4] mm: per-node proactive reclaim

[PATCH -next v2 0/4] mm: per-node proactive reclaim

[Davidlohr Bueso]

Hello,

This is a tardy follow up to v1:
https://lore.kernel.org/linux-mm/20240904162740.1043168-1-dave@stgolabs.net/

Changes:
 - Not a change perse, but further discussed with mhocko potential usecases
   to justify upstreaming this interface. Nowadays NUMA represents the common
   abstraction for memory tiering representing devices of various performance
   characteristics. This interface makes a lot of sense given memcg's lack
   of NUMA awareness.
   
 - Consolidate both memcg and per-node flavors into a common helper. (Yosry)

Patch 1 is a small fixlet independent of the rest of the series.
Patches 2-3 make some of the machinery more generic.
Patch 4 adds the sysfs interface (which has further been deemed ok albeit
not following the one value per file "rule").

Please consider for v6.16.

Thanks!

Davidlohr Bueso (4):
  mm/vmscan: respect psi_memstall region in node reclaim
  mm/memcg: make memory.reclaim interface generic
  mm/vmscan: make __node_reclaim() more generic
  mm: introduce per-node proactive reclaim interface

 Documentation/ABI/stable/sysfs-devices-node |   9 +
 drivers/base/node.c                         |   2 +
 include/linux/swap.h                        |  16 ++
 mm/internal.h                               |   2 +
 mm/memcontrol.c                             |  77 +-------
 mm/vmscan.c                                 | 195 +++++++++++++++++---
 6 files changed, 201 insertions(+), 100 deletions(-)

--
2.39.5

[PATCH 1/4] mm/vmscan: respect psi_memstall region in node reclaim

[Davidlohr Bueso]

... rather benign but keep proper ending order.

Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 mm/vmscan.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index a93a1ba9009e..c13c01eb0b42 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -7653,8 +7653,8 @@ static int __node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned in
 	set_task_reclaim_state(p, NULL);
 	memalloc_noreclaim_restore(noreclaim_flag);
 	fs_reclaim_release(sc.gfp_mask);
-	psi_memstall_leave(&pflags);
 	delayacct_freepages_end();
+	psi_memstall_leave(&pflags);
 
 	trace_mm_vmscan_node_reclaim_end(sc.nr_reclaimed);
 
-- 
2.39.5

[PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Davidlohr Bueso]

This adds a general call for both parsing as well as the
common reclaim semantics. memcg is still the only user and
no change in semantics.

Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 mm/internal.h   |  2 +
 mm/memcontrol.c | 77 ++------------------------------------
 mm/vmscan.c     | 98 +++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 104 insertions(+), 73 deletions(-)

diff --git a/mm/internal.h b/mm/internal.h
index 3823fb356d3b..fc4262262b31 100644
--- a/mm/internal.h
+++ b/mm/internal.h
@@ -517,6 +517,8 @@ extern unsigned long highest_memmap_pfn;
 bool folio_isolate_lru(struct folio *folio);
 void folio_putback_lru(struct folio *folio);
 extern void reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason);
+int user_proactive_reclaim(char *buf,
+			   struct mem_cgroup *memcg, pg_data_t *pgdat);
 
 /*
  * in mm/rmap.c:
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 902da8a9c643..015e406eadfa 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -51,7 +51,6 @@
 #include <linux/spinlock.h>
 #include <linux/fs.h>
 #include <linux/seq_file.h>
-#include <linux/parser.h>
 #include <linux/vmpressure.h>
 #include <linux/memremap.h>
 #include <linux/mm_inline.h>
@@ -4566,83 +4565,15 @@ static ssize_t memory_oom_group_write(struct kernfs_open_file *of,
 	return nbytes;
 }
 
-enum {
-	MEMORY_RECLAIM_SWAPPINESS = 0,
-	MEMORY_RECLAIM_SWAPPINESS_MAX,
-	MEMORY_RECLAIM_NULL,
-};
-
-static const match_table_t tokens = {
-	{ MEMORY_RECLAIM_SWAPPINESS, "swappiness=%d"},
-	{ MEMORY_RECLAIM_SWAPPINESS_MAX, "swappiness=max"},
-	{ MEMORY_RECLAIM_NULL, NULL },
-};
-
 static ssize_t memory_reclaim(struct kernfs_open_file *of, char *buf,
 			      size_t nbytes, loff_t off)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_css(of_css(of));
-	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
-	unsigned long nr_to_reclaim, nr_reclaimed = 0;
-	int swappiness = -1;
-	unsigned int reclaim_options;
-	char *old_buf, *start;
-	substring_t args[MAX_OPT_ARGS];
-
-	buf = strstrip(buf);
-
-	old_buf = buf;
-	nr_to_reclaim = memparse(buf, &buf) / PAGE_SIZE;
-	if (buf == old_buf)
-		return -EINVAL;
-
-	buf = strstrip(buf);
-
-	while ((start = strsep(&buf, " ")) != NULL) {
-		if (!strlen(start))
-			continue;
-		switch (match_token(start, tokens, args)) {
-		case MEMORY_RECLAIM_SWAPPINESS:
-			if (match_int(&args[0], &swappiness))
-				return -EINVAL;
-			if (swappiness < MIN_SWAPPINESS || swappiness > MAX_SWAPPINESS)
-				return -EINVAL;
-			break;
-		case MEMORY_RECLAIM_SWAPPINESS_MAX:
-			swappiness = SWAPPINESS_ANON_ONLY;
-			break;
-		default:
-			return -EINVAL;
-		}
-	}
-
-	reclaim_options	= MEMCG_RECLAIM_MAY_SWAP | MEMCG_RECLAIM_PROACTIVE;
-	while (nr_reclaimed < nr_to_reclaim) {
-		/* Will converge on zero, but reclaim enforces a minimum */
-		unsigned long batch_size = (nr_to_reclaim - nr_reclaimed) / 4;
-		unsigned long reclaimed;
-
-		if (signal_pending(current))
-			return -EINTR;
-
-		/*
-		 * This is the final attempt, drain percpu lru caches in the
-		 * hope of introducing more evictable pages for
-		 * try_to_free_mem_cgroup_pages().
-		 */
-		if (!nr_retries)
-			lru_add_drain_all();
-
-		reclaimed = try_to_free_mem_cgroup_pages(memcg,
-					batch_size, GFP_KERNEL,
-					reclaim_options,
-					swappiness == -1 ? NULL : &swappiness);
-
-		if (!reclaimed && !nr_retries--)
-			return -EAGAIN;
+	int ret;
 
-		nr_reclaimed += reclaimed;
-	}
+	ret = user_proactive_reclaim(buf, memcg, NULL);
+	if (ret)
+		return ret;
 
 	return nbytes;
 }
diff --git a/mm/vmscan.c b/mm/vmscan.c
index c13c01eb0b42..63ddec550c3b 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -57,6 +57,7 @@
 #include <linux/rculist_nulls.h>
 #include <linux/random.h>
 #include <linux/mmu_notifier.h>
+#include <linux/parser.h>
 
 #include <asm/tlbflush.h>
 #include <asm/div64.h>
@@ -6714,6 +6715,15 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
 
 	return nr_reclaimed;
 }
+#else
+unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
+					   unsigned long nr_pages,
+					   gfp_t gfp_mask,
+					   unsigned int reclaim_options,
+					   int *swappiness)
+{
+	return 0;
+}
 #endif
 
 static void kswapd_age_node(struct pglist_data *pgdat, struct scan_control *sc)
@@ -7708,6 +7718,94 @@ int node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order)
 
 	return ret;
 }
+
+enum {
+	MEMORY_RECLAIM_SWAPPINESS = 0,
+	MEMORY_RECLAIM_SWAPPINESS_MAX,
+	MEMORY_RECLAIM_NULL,
+};
+static const match_table_t tokens = {
+	{ MEMORY_RECLAIM_SWAPPINESS, "swappiness=%d"},
+	{ MEMORY_RECLAIM_SWAPPINESS_MAX, "swappiness=max"},
+	{ MEMORY_RECLAIM_NULL, NULL },
+};
+
+int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
+{
+	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
+	unsigned long nr_to_reclaim, nr_reclaimed = 0;
+	int swappiness = -1;
+	char *old_buf, *start;
+	substring_t args[MAX_OPT_ARGS];
+
+	if (!buf || (!memcg && !pgdat))
+		return -EINVAL;
+
+	buf = strstrip(buf);
+
+	old_buf = buf;
+	nr_to_reclaim = memparse(buf, &buf) / PAGE_SIZE;
+	if (buf == old_buf)
+		return -EINVAL;
+
+	buf = strstrip(buf);
+
+	while ((start = strsep(&buf, " ")) != NULL) {
+		if (!strlen(start))
+			continue;
+		switch (match_token(start, tokens, args)) {
+		case MEMORY_RECLAIM_SWAPPINESS:
+			if (match_int(&args[0], &swappiness))
+				return -EINVAL;
+			if (swappiness < MIN_SWAPPINESS ||
+			    swappiness > MAX_SWAPPINESS)
+				return -EINVAL;
+			break;
+		case MEMORY_RECLAIM_SWAPPINESS_MAX:
+			swappiness = SWAPPINESS_ANON_ONLY;
+			break;
+		default:
+			return -EINVAL;
+		}
+	}
+
+	while (nr_reclaimed < nr_to_reclaim) {
+		/* Will converge on zero, but reclaim enforces a minimum */
+		unsigned long batch_size = (nr_to_reclaim - nr_reclaimed) / 4;
+		unsigned long reclaimed;
+
+		if (signal_pending(current))
+			return -EINTR;
+
+		/*
+		 * This is the final attempt, drain percpu lru caches in the
+		 * hope of introducing more evictable pages.
+		 */
+		if (!nr_retries)
+			lru_add_drain_all();
+
+		if (memcg) {
+			unsigned int reclaim_options;
+
+			reclaim_options = MEMCG_RECLAIM_MAY_SWAP |
+					  MEMCG_RECLAIM_PROACTIVE;
+			reclaimed = try_to_free_mem_cgroup_pages(memcg,
+						 batch_size, GFP_KERNEL,
+						 reclaim_options,
+						 swappiness == -1 ? NULL : &swappiness);
+		} else {
+			return -EINVAL;
+		}
+
+		if (!reclaimed && !nr_retries--)
+			return -EAGAIN;
+
+		nr_reclaimed += reclaimed;
+	}
+
+	return 0;
+}
+
 #endif
 
 /**
-- 
2.39.5

[PATCH 3/4] mm/vmscan: make __node_reclaim() more generic

[Davidlohr Bueso]

As this will be called from non page allocator paths for
proactive reclaim, allow users to pass the sc and nr of
pages, and adjust the return value as well. No change in
semantics.

Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 mm/vmscan.c | 48 +++++++++++++++++++++++++-----------------------
 1 file changed, 25 insertions(+), 23 deletions(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 63ddec550c3b..cdd9cb97fb79 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -7618,36 +7618,26 @@ static unsigned long node_pagecache_reclaimable(struct pglist_data *pgdat)
 /*
  * Try to free up some pages from this node through reclaim.
  */
-static int __node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order)
+static unsigned long __node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask,
+				    unsigned long nr_pages,
+				    struct scan_control *sc)
 {
-	/* Minimum pages needed in order to stay on node */
-	const unsigned long nr_pages = 1 << order;
 	struct task_struct *p = current;
 	unsigned int noreclaim_flag;
-	struct scan_control sc = {
-		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
-		.gfp_mask = current_gfp_context(gfp_mask),
-		.order = order,
-		.priority = NODE_RECLAIM_PRIORITY,
-		.may_writepage = !!(node_reclaim_mode & RECLAIM_WRITE),
-		.may_unmap = !!(node_reclaim_mode & RECLAIM_UNMAP),
-		.may_swap = 1,
-		.reclaim_idx = gfp_zone(gfp_mask),
-	};
 	unsigned long pflags;
 
-	trace_mm_vmscan_node_reclaim_begin(pgdat->node_id, order,
-					   sc.gfp_mask);
+	trace_mm_vmscan_node_reclaim_begin(pgdat->node_id, sc->order,
+					   sc->gfp_mask);
 
 	cond_resched();
 	psi_memstall_enter(&pflags);
 	delayacct_freepages_start();
-	fs_reclaim_acquire(sc.gfp_mask);
+	fs_reclaim_acquire(sc->gfp_mask);
 	/*
 	 * We need to be able to allocate from the reserves for RECLAIM_UNMAP
 	 */
 	noreclaim_flag = memalloc_noreclaim_save();
-	set_task_reclaim_state(p, &sc.reclaim_state);
+	set_task_reclaim_state(p, &sc->reclaim_state);
 
 	if (node_pagecache_reclaimable(pgdat) > pgdat->min_unmapped_pages ||
 	    node_page_state_pages(pgdat, NR_SLAB_RECLAIMABLE_B) > pgdat->min_slab_pages) {
@@ -7656,24 +7646,36 @@ static int __node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned in
 		 * priorities until we have enough memory freed.
 		 */
 		do {
-			shrink_node(pgdat, &sc);
-		} while (sc.nr_reclaimed < nr_pages && --sc.priority >= 0);
+			shrink_node(pgdat, sc);
+		} while (sc->nr_reclaimed < nr_pages && --sc->priority >= 0);
 	}
 
 	set_task_reclaim_state(p, NULL);
 	memalloc_noreclaim_restore(noreclaim_flag);
-	fs_reclaim_release(sc.gfp_mask);
+	fs_reclaim_release(sc->gfp_mask);
 	delayacct_freepages_end();
 	psi_memstall_leave(&pflags);
 
-	trace_mm_vmscan_node_reclaim_end(sc.nr_reclaimed);
+	trace_mm_vmscan_node_reclaim_end(sc->nr_reclaimed);
 
-	return sc.nr_reclaimed >= nr_pages;
+	return sc->nr_reclaimed;
 }
 
 int node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order)
 {
 	int ret;
+	/* Minimum pages needed in order to stay on node */
+	const unsigned long nr_pages = 1 << order;
+	struct scan_control sc = {
+		.nr_to_reclaim = max(nr_pages, SWAP_CLUSTER_MAX),
+		.gfp_mask = current_gfp_context(gfp_mask),
+		.order = order,
+		.priority = NODE_RECLAIM_PRIORITY,
+		.may_writepage = !!(node_reclaim_mode & RECLAIM_WRITE),
+		.may_unmap = !!(node_reclaim_mode & RECLAIM_UNMAP),
+		.may_swap = 1,
+		.reclaim_idx = gfp_zone(gfp_mask),
+	};
 
 	/*
 	 * Node reclaim reclaims unmapped file backed pages and
@@ -7708,7 +7710,7 @@ int node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order)
 	if (test_and_set_bit_lock(PGDAT_RECLAIM_LOCKED, &pgdat->flags))
 		return NODE_RECLAIM_NOSCAN;
 
-	ret = __node_reclaim(pgdat, gfp_mask, order);
+	ret = __node_reclaim(pgdat, gfp_mask, nr_pages, &sc) >= nr_pages;
 	clear_bit_unlock(PGDAT_RECLAIM_LOCKED, &pgdat->flags);
 
 	if (ret)
-- 
2.39.5

[PATCH 4/4] mm: introduce per-node proactive reclaim interface

[Davidlohr Bueso]

This adds support for allowing proactive reclaim in general on a
NUMA system. A per-node interface extends support for beyond a
memcg-specific interface, respecting the current semantics of
memory.reclaim: respecting aging LRU and not supporting
artificially triggering eviction on nodes belonging to non-bottom
tiers.

This patch allows userspace to do:

     echo "512M swappiness=10" > /sys/devices/system/node/nodeX/reclaim

One of the premises for this is to semantically align as best as
possible with memory.reclaim. During a brief time memcg did
support nodemask until 55ab834a86a9 (Revert "mm: add nodes=
arg to memory.reclaim"), for which semantics around reclaim
(eviction) vs demotion were not clear, rendering charging
expectations to be broken.

With this approach:

1. Users who do not use memcg can benefit from proactive reclaim.
The memcg interface is not NUMA aware and there are usecases that
are focusing on NUMA balancing rather than workload memory footprint.

2. Proactive reclaim on top tiers will trigger demotion, for which
memory is still byte-addressable. Reclaiming on the bottom nodes
will trigger evicting to swap (the traditional sense of reclaim).
This follows the semantics of what is today part of the aging process
on tiered memory, mirroring what every other form of reclaim does
(reactive and memcg proactive reclaim). Furthermore per-node proactive
reclaim is not as susceptible to the memcg charging problem mentioned
above.

3. Unlike the nodes= arg, this interface avoids confusing semantics,
such as what exactly the user wants when mixing top-tier and low-tier
nodes in the nodemask. Further per-node interface is less exposed to
"free up memory in my container" usecases, where eviction is intended.

4. Users that *really* want to free up memory can use proactive reclaim
on nodes knowingly to be on the bottom tiers to force eviction in a
natural way - higher access latencies are still better than swap.
If compelled, while no guarantees and perhaps not worth the effort,
users could also also potentially follow a ladder-like approach to
eventually free up the memory. Alternatively, perhaps an 'evict' option
could be added to the parameters for both memory.reclaim and per-node
interfaces to force this action unconditionally.

Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 Documentation/ABI/stable/sysfs-devices-node |  9 ++++
 drivers/base/node.c                         |  2 +
 include/linux/swap.h                        | 16 +++++++
 mm/vmscan.c                                 | 53 ++++++++++++++++++---
 4 files changed, 74 insertions(+), 6 deletions(-)

diff --git a/Documentation/ABI/stable/sysfs-devices-node b/Documentation/ABI/stable/sysfs-devices-node
index a02707cb7cbc..2d0e023f22a7 100644
--- a/Documentation/ABI/stable/sysfs-devices-node
+++ b/Documentation/ABI/stable/sysfs-devices-node
@@ -227,3 +227,12 @@ Contact:	Jiaqi Yan <jiaqiyan@google.com>
 Description:
 		Of the raw poisoned pages on a NUMA node, how many pages are
 		recovered by memory error recovery attempt.
+
+What:		/sys/devices/system/node/nodeX/reclaim
+Date:		June 2025
+Contact:	Linux Memory Management list <linux-mm@kvack.org>
+Description:
+		Perform user-triggered proactive reclaim on a NUMA node.
+		This interface is equivalent to the memcg variant.
+
+		See Documentation/admin-guide/cgroup-v2.rst
diff --git a/drivers/base/node.c b/drivers/base/node.c
index 6d66382dae65..548b532a2129 100644
--- a/drivers/base/node.c
+++ b/drivers/base/node.c
@@ -659,6 +659,7 @@ static int register_node(struct node *node, int num)
 	} else {
 		hugetlb_register_node(node);
 		compaction_register_node(node);
+		reclaim_register_node(node);
 	}
 
 	return error;
@@ -675,6 +676,7 @@ void unregister_node(struct node *node)
 {
 	hugetlb_unregister_node(node);
 	compaction_unregister_node(node);
+	reclaim_unregister_node(node);
 	node_remove_accesses(node);
 	node_remove_caches(node);
 	device_unregister(&node->dev);
diff --git a/include/linux/swap.h b/include/linux/swap.h
index bc0e1c275fc0..dac7ba98783d 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -431,6 +431,22 @@ extern unsigned long shrink_all_memory(unsigned long nr_pages);
 extern int vm_swappiness;
 long remove_mapping(struct address_space *mapping, struct folio *folio);
 
+#if defined(CONFIG_SYSFS) && defined(CONFIG_NUMA)
+extern int reclaim_register_node(struct node *node);
+extern void reclaim_unregister_node(struct node *node);
+
+#else
+
+static inline int reclaim_register_node(struct node *node)
+{
+	return 0;
+}
+
+static inline void reclaim_unregister_node(struct node *node)
+{
+}
+#endif /* CONFIG_SYSFS && CONFIG_NUMA */
+
 #ifdef CONFIG_NUMA
 extern int sysctl_min_unmapped_ratio;
 extern int sysctl_min_slab_ratio;
diff --git a/mm/vmscan.c b/mm/vmscan.c
index cdd9cb97fb79..f77feb75c678 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -94,10 +94,8 @@ struct scan_control {
 	unsigned long	anon_cost;
 	unsigned long	file_cost;
 
-#ifdef CONFIG_MEMCG
 	/* Swappiness value for proactive reclaim. Always use sc_swappiness()! */
 	int *proactive_swappiness;
-#endif
 
 	/* Can active folios be deactivated as part of reclaim? */
 #define DEACTIVATE_ANON 1
@@ -121,7 +119,7 @@ struct scan_control {
 	/* Has cache_trim_mode failed at least once? */
 	unsigned int cache_trim_mode_failed:1;
 
-	/* Proactive reclaim invoked by userspace through memory.reclaim */
+	/* Proactive reclaim invoked by userspace */
 	unsigned int proactive:1;
 
 	/*
@@ -7732,13 +7730,15 @@ static const match_table_t tokens = {
 	{ MEMORY_RECLAIM_NULL, NULL },
 };
 
-int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
+int user_proactive_reclaim(char *buf,
+			   struct mem_cgroup *memcg, pg_data_t *pgdat)
 {
 	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
 	unsigned long nr_to_reclaim, nr_reclaimed = 0;
 	int swappiness = -1;
 	char *old_buf, *start;
 	substring_t args[MAX_OPT_ARGS];
+	gfp_t gfp_mask = GFP_KERNEL;
 
 	if (!buf || (!memcg && !pgdat))
 		return -EINVAL;
@@ -7792,11 +7792,29 @@ int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat
 			reclaim_options = MEMCG_RECLAIM_MAY_SWAP |
 					  MEMCG_RECLAIM_PROACTIVE;
 			reclaimed = try_to_free_mem_cgroup_pages(memcg,
-						 batch_size, GFP_KERNEL,
+						 batch_size, gfp_mask,
 						 reclaim_options,
 						 swappiness == -1 ? NULL : &swappiness);
 		} else {
-			return -EINVAL;
+			struct scan_control sc = {
+				.gfp_mask = current_gfp_context(gfp_mask),
+				.reclaim_idx = gfp_zone(gfp_mask),
+				.proactive_swappiness = swappiness == -1 ? NULL : &swappiness,
+				.priority = DEF_PRIORITY,
+				.may_writepage = !laptop_mode,
+				.nr_to_reclaim = max(batch_size, SWAP_CLUSTER_MAX),
+				.may_unmap = 1,
+				.may_swap = 1,
+				.proactive = 1,
+			};
+
+			if (test_and_set_bit_lock(PGDAT_RECLAIM_LOCKED,
+						  &pgdat->flags))
+				return -EAGAIN;
+
+			reclaimed = __node_reclaim(pgdat, gfp_mask,
+						   batch_size, &sc);
+			clear_bit_unlock(PGDAT_RECLAIM_LOCKED, &pgdat->flags);
 		}
 
 		if (!reclaimed && !nr_retries--)
@@ -7855,3 +7873,26 @@ void check_move_unevictable_folios(struct folio_batch *fbatch)
 	}
 }
 EXPORT_SYMBOL_GPL(check_move_unevictable_folios);
+
+#if defined(CONFIG_SYSFS) && defined(CONFIG_NUMA)
+static ssize_t reclaim_store(struct device *dev,
+			     struct device_attribute *attr,
+			     const char *buf, size_t count)
+{
+	int ret, nid = dev->id;
+
+	ret = user_proactive_reclaim((char *)buf, NULL, NODE_DATA(nid));
+	return ret ? -EAGAIN : count;
+}
+
+static DEVICE_ATTR_WO(reclaim);
+int reclaim_register_node(struct node *node)
+{
+	return device_create_file(&node->dev, &dev_attr_reclaim);
+}
+
+void reclaim_unregister_node(struct node *node)
+{
+	return device_remove_file(&node->dev, &dev_attr_reclaim);
+}
+#endif
-- 
2.39.5

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Andrew Morton]

On Mon, 23 Jun 2025 11:58:49 -0700 Davidlohr Bueso <dave@stgolabs.net> wrote:

> This adds a general call for both parsing as well as the
> common reclaim semantics. memcg is still the only user and
> no change in semantics.
> 
> +int user_proactive_reclaim(char *buf,
> +			   struct mem_cgroup *memcg, pg_data_t *pgdat);

Feeling nitty, is this a good name for it?  It's hard to imagine what a
function called "user_proactive_reclaim" actually does.

That it isn't documented isn't helpful either!

Re: [PATCH -next v2 0/4] mm: per-node proactive reclaim

[Andrew Morton]

On Mon, 23 Jun 2025 11:58:47 -0700 Davidlohr Bueso <dave@stgolabs.net> wrote:

> This is a tardy follow up to v1:
> https://lore.kernel.org/linux-mm/20240904162740.1043168-1-dave@stgolabs.net/

Cool, I'll add it to mm-new for testing.

The v2 series didn't have a [0/N] so I scraped the words from the v1
series.  These were almost a copy of the [4/4] patch changelog but
there seem to have been some changes.  Please do maintain and resend the
cover letter verbiage.

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Davidlohr Bueso]

On Mon, 23 Jun 2025, Andrew Morton wrote:

>On Mon, 23 Jun 2025 11:58:49 -0700 Davidlohr Bueso <dave@stgolabs.net> wrote:
>
>> This adds a general call for both parsing as well as the
>> common reclaim semantics. memcg is still the only user and
>> no change in semantics.
>>
>> +int user_proactive_reclaim(char *buf,
>> +			   struct mem_cgroup *memcg, pg_data_t *pgdat);
>
>Feeling nitty, is this a good name for it?  It's hard to imagine what a
>function called "user_proactive_reclaim" actually does.

I'm open to another name, sure. But imo the chosen one is actually pretty
descriptive: you know it's coming from userspace (justifying the 'buf'),
you know this is not about memory pressure and the memcg/pgdat parameters
tell the possible interfaces. Would prefixing a 'do_' be any better?

>That it isn't documented isn't helpful either!

I had done this but felt rather redundant and unnecessary, and further
don't expect for it go gain any other users. But ok, will add.

Thanks,
Davidlohr

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Klara Modin]

Hi,

On 2025-06-23 11:58:49 -0700, Davidlohr Bueso wrote:
> This adds a general call for both parsing as well as the
> common reclaim semantics. memcg is still the only user and
> no change in semantics.
> 
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
> ---
>  mm/internal.h   |  2 +
>  mm/memcontrol.c | 77 ++------------------------------------
>  mm/vmscan.c     | 98 +++++++++++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 104 insertions(+), 73 deletions(-)
> 
> diff --git a/mm/internal.h b/mm/internal.h
> index 3823fb356d3b..fc4262262b31 100644
> --- a/mm/internal.h
> +++ b/mm/internal.h
> @@ -517,6 +517,8 @@ extern unsigned long highest_memmap_pfn;
>  bool folio_isolate_lru(struct folio *folio);
>  void folio_putback_lru(struct folio *folio);
>  extern void reclaim_throttle(pg_data_t *pgdat, enum vmscan_throttle_state reason);
> +int user_proactive_reclaim(char *buf,
> +			   struct mem_cgroup *memcg, pg_data_t *pgdat);
>  
>  /*
>   * in mm/rmap.c:
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 902da8a9c643..015e406eadfa 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -51,7 +51,6 @@
>  #include <linux/spinlock.h>
>  #include <linux/fs.h>
>  #include <linux/seq_file.h>
> -#include <linux/parser.h>
>  #include <linux/vmpressure.h>
>  #include <linux/memremap.h>
>  #include <linux/mm_inline.h>
> @@ -4566,83 +4565,15 @@ static ssize_t memory_oom_group_write(struct kernfs_open_file *of,
>  	return nbytes;
>  }
>  
> -enum {
> -	MEMORY_RECLAIM_SWAPPINESS = 0,
> -	MEMORY_RECLAIM_SWAPPINESS_MAX,
> -	MEMORY_RECLAIM_NULL,
> -};
> -
> -static const match_table_t tokens = {
> -	{ MEMORY_RECLAIM_SWAPPINESS, "swappiness=%d"},
> -	{ MEMORY_RECLAIM_SWAPPINESS_MAX, "swappiness=max"},
> -	{ MEMORY_RECLAIM_NULL, NULL },
> -};
> -
>  static ssize_t memory_reclaim(struct kernfs_open_file *of, char *buf,
>  			      size_t nbytes, loff_t off)
>  {
>  	struct mem_cgroup *memcg = mem_cgroup_from_css(of_css(of));
> -	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
> -	unsigned long nr_to_reclaim, nr_reclaimed = 0;
> -	int swappiness = -1;
> -	unsigned int reclaim_options;
> -	char *old_buf, *start;
> -	substring_t args[MAX_OPT_ARGS];
> -
> -	buf = strstrip(buf);
> -
> -	old_buf = buf;
> -	nr_to_reclaim = memparse(buf, &buf) / PAGE_SIZE;
> -	if (buf == old_buf)
> -		return -EINVAL;
> -
> -	buf = strstrip(buf);
> -
> -	while ((start = strsep(&buf, " ")) != NULL) {
> -		if (!strlen(start))
> -			continue;
> -		switch (match_token(start, tokens, args)) {
> -		case MEMORY_RECLAIM_SWAPPINESS:
> -			if (match_int(&args[0], &swappiness))
> -				return -EINVAL;
> -			if (swappiness < MIN_SWAPPINESS || swappiness > MAX_SWAPPINESS)
> -				return -EINVAL;
> -			break;
> -		case MEMORY_RECLAIM_SWAPPINESS_MAX:
> -			swappiness = SWAPPINESS_ANON_ONLY;
> -			break;
> -		default:
> -			return -EINVAL;
> -		}
> -	}
> -
> -	reclaim_options	= MEMCG_RECLAIM_MAY_SWAP | MEMCG_RECLAIM_PROACTIVE;
> -	while (nr_reclaimed < nr_to_reclaim) {
> -		/* Will converge on zero, but reclaim enforces a minimum */
> -		unsigned long batch_size = (nr_to_reclaim - nr_reclaimed) / 4;
> -		unsigned long reclaimed;
> -
> -		if (signal_pending(current))
> -			return -EINTR;
> -
> -		/*
> -		 * This is the final attempt, drain percpu lru caches in the
> -		 * hope of introducing more evictable pages for
> -		 * try_to_free_mem_cgroup_pages().
> -		 */
> -		if (!nr_retries)
> -			lru_add_drain_all();
> -
> -		reclaimed = try_to_free_mem_cgroup_pages(memcg,
> -					batch_size, GFP_KERNEL,
> -					reclaim_options,
> -					swappiness == -1 ? NULL : &swappiness);
> -
> -		if (!reclaimed && !nr_retries--)
> -			return -EAGAIN;
> +	int ret;
>  
> -		nr_reclaimed += reclaimed;
> -	}

> +	ret = user_proactive_reclaim(buf, memcg, NULL);

This is outside CONFIG_NUMA.

> +	if (ret)
> +		return ret;
>  
>  	return nbytes;
>  }
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index c13c01eb0b42..63ddec550c3b 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -57,6 +57,7 @@
>  #include <linux/rculist_nulls.h>
>  #include <linux/random.h>
>  #include <linux/mmu_notifier.h>
> +#include <linux/parser.h>
>  
>  #include <asm/tlbflush.h>
>  #include <asm/div64.h>
> @@ -6714,6 +6715,15 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>  
>  	return nr_reclaimed;
>  }
> +#else
> +unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
> +					   unsigned long nr_pages,
> +					   gfp_t gfp_mask,
> +					   unsigned int reclaim_options,
> +					   int *swappiness)
> +{
> +	return 0;
> +}
>  #endif
>  
>  static void kswapd_age_node(struct pglist_data *pgdat, struct scan_control *sc)
> @@ -7708,6 +7718,94 @@ int node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order)
>  
>  	return ret;
>  }
> +
> +enum {
> +	MEMORY_RECLAIM_SWAPPINESS = 0,
> +	MEMORY_RECLAIM_SWAPPINESS_MAX,
> +	MEMORY_RECLAIM_NULL,
> +};
> +static const match_table_t tokens = {
> +	{ MEMORY_RECLAIM_SWAPPINESS, "swappiness=%d"},
> +	{ MEMORY_RECLAIM_SWAPPINESS_MAX, "swappiness=max"},
> +	{ MEMORY_RECLAIM_NULL, NULL },
> +};
> +
> +int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
> +{
> +	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
> +	unsigned long nr_to_reclaim, nr_reclaimed = 0;
> +	int swappiness = -1;
> +	char *old_buf, *start;
> +	substring_t args[MAX_OPT_ARGS];
> +
> +	if (!buf || (!memcg && !pgdat))
> +		return -EINVAL;
> +
> +	buf = strstrip(buf);
> +
> +	old_buf = buf;
> +	nr_to_reclaim = memparse(buf, &buf) / PAGE_SIZE;
> +	if (buf == old_buf)
> +		return -EINVAL;
> +
> +	buf = strstrip(buf);
> +
> +	while ((start = strsep(&buf, " ")) != NULL) {
> +		if (!strlen(start))
> +			continue;
> +		switch (match_token(start, tokens, args)) {
> +		case MEMORY_RECLAIM_SWAPPINESS:
> +			if (match_int(&args[0], &swappiness))
> +				return -EINVAL;
> +			if (swappiness < MIN_SWAPPINESS ||
> +			    swappiness > MAX_SWAPPINESS)
> +				return -EINVAL;
> +			break;
> +		case MEMORY_RECLAIM_SWAPPINESS_MAX:
> +			swappiness = SWAPPINESS_ANON_ONLY;
> +			break;
> +		default:
> +			return -EINVAL;
> +		}
> +	}
> +
> +	while (nr_reclaimed < nr_to_reclaim) {
> +		/* Will converge on zero, but reclaim enforces a minimum */
> +		unsigned long batch_size = (nr_to_reclaim - nr_reclaimed) / 4;
> +		unsigned long reclaimed;
> +
> +		if (signal_pending(current))
> +			return -EINTR;
> +
> +		/*
> +		 * This is the final attempt, drain percpu lru caches in the
> +		 * hope of introducing more evictable pages.
> +		 */
> +		if (!nr_retries)
> +			lru_add_drain_all();
> +
> +		if (memcg) {
> +			unsigned int reclaim_options;
> +
> +			reclaim_options = MEMCG_RECLAIM_MAY_SWAP |
> +					  MEMCG_RECLAIM_PROACTIVE;
> +			reclaimed = try_to_free_mem_cgroup_pages(memcg,
> +						 batch_size, GFP_KERNEL,
> +						 reclaim_options,
> +						 swappiness == -1 ? NULL : &swappiness);
> +		} else {
> +			return -EINVAL;
> +		}
> +
> +		if (!reclaimed && !nr_retries--)
> +			return -EAGAIN;
> +
> +		nr_reclaimed += reclaimed;
> +	}
> +
> +	return 0;
> +}
> +
>  #endif

Should this really be inside CONFIG_NUMA? It was moved from outside of
CONFIG_NUMA where it's now called which results in a build failure if
it's disabled. Or is there a stub missing?

>  
>  /**
> -- 
> 2.39.5
> 

Regards,
Klara Modin

Re: [PATCH 1/4] mm/vmscan: respect psi_memstall region in node reclaim

[Shakeel Butt]

On Mon, Jun 23, 2025 at 11:58:48AM -0700, Davidlohr Bueso wrote:
> ... rather benign but keep proper ending order.
> 
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Nit: in next version, please have a more clear commit message.

Acked-by: Shakeel Butt <shakeel.butt@linux.dev>

Re: [PATCH 4/4] mm: introduce per-node proactive reclaim interface

[Shakeel Butt]

On Mon, Jun 23, 2025 at 11:58:51AM -0700, Davidlohr Bueso wrote:
> This adds support for allowing proactive reclaim in general on a
> NUMA system. A per-node interface extends support for beyond a
> memcg-specific interface, respecting the current semantics of
> memory.reclaim: respecting aging LRU and not supporting
> artificially triggering eviction on nodes belonging to non-bottom
> tiers.
> 
> This patch allows userspace to do:
> 
>      echo "512M swappiness=10" > /sys/devices/system/node/nodeX/reclaim
> 
> One of the premises for this is to semantically align as best as
> possible with memory.reclaim. During a brief time memcg did
> support nodemask until 55ab834a86a9 (Revert "mm: add nodes=
> arg to memory.reclaim"), for which semantics around reclaim
> (eviction) vs demotion were not clear, rendering charging
> expectations to be broken.
> 
> With this approach:
> 
> 1. Users who do not use memcg can benefit from proactive reclaim.
> The memcg interface is not NUMA aware and there are usecases that
> are focusing on NUMA balancing rather than workload memory footprint.
> 
> 2. Proactive reclaim on top tiers will trigger demotion, for which
> memory is still byte-addressable. Reclaiming on the bottom nodes
> will trigger evicting to swap (the traditional sense of reclaim).
> This follows the semantics of what is today part of the aging process
> on tiered memory, mirroring what every other form of reclaim does
> (reactive and memcg proactive reclaim). Furthermore per-node proactive
> reclaim is not as susceptible to the memcg charging problem mentioned
> above.
> 
> 3. Unlike the nodes= arg, this interface avoids confusing semantics,
> such as what exactly the user wants when mixing top-tier and low-tier
> nodes in the nodemask. Further per-node interface is less exposed to
> "free up memory in my container" usecases, where eviction is intended.
> 
> 4. Users that *really* want to free up memory can use proactive reclaim
> on nodes knowingly to be on the bottom tiers to force eviction in a
> natural way - higher access latencies are still better than swap.
> If compelled, while no guarantees and perhaps not worth the effort,
> users could also also potentially follow a ladder-like approach to
> eventually free up the memory. Alternatively, perhaps an 'evict' option
> could be added to the parameters for both memory.reclaim and per-node
> interfaces to force this action unconditionally.
> 
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Overall looks good but I will try to dig deeper in next couple of days
(or weeks).

One orthogonal thought: I wonder if we want a unified aging (hotness or
generation or active/inactive) view of jobs/memcgs/system. At the moment
due to the way LRUs are implemented i.e. per-memcg per-node, we can have
different view of these LRUs even for the same memcg. For example the
hottest pages in low tier node might be colder than coldest pages in the
top tier. Not sure how to implement it in a scalable way.

Re: [PATCH 4/4] mm: introduce per-node proactive reclaim interface

[SeongJae Park]

On Wed, 25 Jun 2025 16:10:16 -0700 Shakeel Butt <shakeel.butt@linux.dev> wrote:

> On Mon, Jun 23, 2025 at 11:58:51AM -0700, Davidlohr Bueso wrote:
> > This adds support for allowing proactive reclaim in general on a
> > NUMA system. A per-node interface extends support for beyond a
> > memcg-specific interface, respecting the current semantics of
> > memory.reclaim: respecting aging LRU and not supporting
> > artificially triggering eviction on nodes belonging to non-bottom
> > tiers.
> > 
> > This patch allows userspace to do:
> > 
> >      echo "512M swappiness=10" > /sys/devices/system/node/nodeX/reclaim
[...]
> One orthogonal thought: I wonder if we want a unified aging (hotness or
> generation or active/inactive) view of jobs/memcgs/system. At the moment
> due to the way LRUs are implemented i.e. per-memcg per-node, we can have
> different view of these LRUs even for the same memcg. For example the
> hottest pages in low tier node might be colder than coldest pages in the
> top tier.

I think it would be nice to have, and DAMON could help.

DAMON can monitor access patterns on the entire physical address space and make
actions such as migrating pages to different nodes[1] or LRU-[de]activate
([anti-]aging)[2] for specific cgroups[3,4], based on the monitored access
pattern.

Such migrations and [anti-]aging would not conflict with page fault and memory
pressure based promotions and demotions, so could help existing tiering
solutions by running those together.

> Not sure how to implement it in a scalable way.

DAMON's monitoring overhead is designed to be not ruled by memory size, so
scalable in terms of memory size.  We recently found it actually shows
reasonable monitoring results on an 1 TiB memory machine[5].  DAMON incurs
minimum overhead and limited to one CPU by default.  If needed, it could also
scale out using multiple threads.

[1] https://lore.kernel.org/all/20250420194030.75838-1-sj@kernel.org
[2] https://lore.kernel.org/all/20220613192301.8817-1-sj@kernel.org
[3] https://lkml.kernel.org/r/20221205230830.144349-1-sj@kernel.org
[4] https://lore.kernel.org/20250619220023.24023-1-sj@kernel.org
[5] page 46, right side plot of
    https://static.sched.com/hosted_files/ossna2025/16/damon_ossna25.pdf?_gl=1*12x1jv*_gcl_au*OTkyNjI0NTk0LjE3NTA4Nzg1Mzg.*FPAU*OTkyNjI0NTk0LjE3NTA4Nzg1Mzg.


Thanks,
SJ

Re: [PATCH -next v2 0/4] mm: per-node proactive reclaim

[Andrew Morton]

We're a bit short of reviews of David's series.  Does anyone have
additional input?

Thanks.

Re: [PATCH -next v2 0/4] mm: per-node proactive reclaim

[Shakeel Butt]

On Tue, Jul 15, 2025 at 05:24:10PM -0700, Andrew Morton wrote:
> 
> We're a bit short of reviews of David's series.  Does anyone have
> additional input?
> 

I am on this and will respond within a day or two.

Re: [PATCH 1/4] mm/vmscan: respect psi_memstall region in node reclaim

[Roman Gushchin]

Davidlohr Bueso <dave@stgolabs.net> writes:

> ... rather benign but keep proper ending order.
>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Reviewed-by: Roman Gushchin <roman.gushchin@linux.dev>

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Roman Gushchin]

Davidlohr Bueso <dave@stgolabs.net> writes:

> This adds a general call for both parsing as well as the
> common reclaim semantics. memcg is still the only user and
> no change in semantics.
>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
> ---
>  mm/internal.h   |  2 +
>  mm/memcontrol.c | 77 ++------------------------------------
>  mm/vmscan.c     | 98 +++++++++++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 104 insertions(+), 73 deletions(-)
> ...
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index c13c01eb0b42..63ddec550c3b 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -57,6 +57,7 @@
>  #include <linux/rculist_nulls.h>
>  #include <linux/random.h>
>  #include <linux/mmu_notifier.h>
> +#include <linux/parser.h>
>  
>  #include <asm/tlbflush.h>
>  #include <asm/div64.h>
> @@ -6714,6 +6715,15 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>  
>  	return nr_reclaimed;
>  }
> +#else
> +unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
> +					   unsigned long nr_pages,
> +					   gfp_t gfp_mask,
> +					   unsigned int reclaim_options,
> +					   int *swappiness)
> +{
> +	return 0;
> +}
>  #endif
>  
>  static void kswapd_age_node(struct pglist_data *pgdat, struct scan_control *sc)
> @@ -7708,6 +7718,94 @@ int node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order)
>  
>  	return ret;
>  }
> +
> +enum {
> +	MEMORY_RECLAIM_SWAPPINESS = 0,
> +	MEMORY_RECLAIM_SWAPPINESS_MAX,
> +	MEMORY_RECLAIM_NULL,
> +};
> +static const match_table_t tokens = {
> +	{ MEMORY_RECLAIM_SWAPPINESS, "swappiness=%d"},
> +	{ MEMORY_RECLAIM_SWAPPINESS_MAX, "swappiness=max"},
> +	{ MEMORY_RECLAIM_NULL, NULL },
> +};
> +
> +int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
> +{
> +	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
> +	unsigned long nr_to_reclaim, nr_reclaimed = 0;
> +	int swappiness = -1;
> +	char *old_buf, *start;
> +	substring_t args[MAX_OPT_ARGS];
> +
> +	if (!buf || (!memcg && !pgdat))
> +		return -EINVAL;
> +
> +	buf = strstrip(buf);
> +
> +	old_buf = buf;
> +	nr_to_reclaim = memparse(buf, &buf) / PAGE_SIZE;
> +	if (buf == old_buf)
> +		return -EINVAL;
> +
> +	buf = strstrip(buf);

To be honest, not a big fan of this refactoring. Effectively parts of
the memcg user interface are moved into mm/vmscan.c. I get that you want
to use the exact same interface somewhere else, but still...

Is it possible to keep it in mm/memcontrol.c?
Also maybe split the actual reclaim mechanism and user's input parsing?

Thanks

Re: [PATCH 3/4] mm/vmscan: make __node_reclaim() more generic

[Roman Gushchin]

Davidlohr Bueso <dave@stgolabs.net> writes:

> As this will be called from non page allocator paths for
> proactive reclaim, allow users to pass the sc and nr of
> pages, and adjust the return value as well. No change in
> semantics.
>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Reviewed-by: Roman Gushchin <roman.gushchin@linux.dev>

Thanks

Re: [PATCH 4/4] mm: introduce per-node proactive reclaim interface

[Roman Gushchin]

Davidlohr Bueso <dave@stgolabs.net> writes:

> This adds support for allowing proactive reclaim in general on a
> NUMA system. A per-node interface extends support for beyond a
> memcg-specific interface, respecting the current semantics of
> memory.reclaim: respecting aging LRU and not supporting
> artificially triggering eviction on nodes belonging to non-bottom
> tiers.
>
> This patch allows userspace to do:
>
>      echo "512M swappiness=10" > /sys/devices/system/node/nodeX/reclaim
>
> One of the premises for this is to semantically align as best as
> possible with memory.reclaim. During a brief time memcg did
> support nodemask until 55ab834a86a9 (Revert "mm: add nodes=
> arg to memory.reclaim"), for which semantics around reclaim
> (eviction) vs demotion were not clear, rendering charging
> expectations to be broken.
>
> With this approach:
>
> 1. Users who do not use memcg can benefit from proactive reclaim.
> The memcg interface is not NUMA aware and there are usecases that
> are focusing on NUMA balancing rather than workload memory footprint.
>
> 2. Proactive reclaim on top tiers will trigger demotion, for which
> memory is still byte-addressable. Reclaiming on the bottom nodes
> will trigger evicting to swap (the traditional sense of reclaim).
> This follows the semantics of what is today part of the aging process
> on tiered memory, mirroring what every other form of reclaim does
> (reactive and memcg proactive reclaim). Furthermore per-node proactive
> reclaim is not as susceptible to the memcg charging problem mentioned
> above.
>
> 3. Unlike the nodes= arg, this interface avoids confusing semantics,
> such as what exactly the user wants when mixing top-tier and low-tier
> nodes in the nodemask. Further per-node interface is less exposed to
> "free up memory in my container" usecases, where eviction is intended.
>
> 4. Users that *really* want to free up memory can use proactive reclaim
> on nodes knowingly to be on the bottom tiers to force eviction in a
> natural way - higher access latencies are still better than swap.
> If compelled, while no guarantees and perhaps not worth the effort,
> users could also also potentially follow a ladder-like approach to
> eventually free up the memory. Alternatively, perhaps an 'evict' option
> could be added to the parameters for both memory.reclaim and per-node
> interfaces to force this action unconditionally.
>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Acked-by: Roman Gushchin <roman.gushchin@linux.dev>

small nit below

> ---
>  Documentation/ABI/stable/sysfs-devices-node |  9 ++++
>  drivers/base/node.c                         |  2 +
>  include/linux/swap.h                        | 16 +++++++
>  mm/vmscan.c                                 | 53 ++++++++++++++++++---
>  4 files changed, 74 insertions(+), 6 deletions(-)
>
> diff --git a/Documentation/ABI/stable/sysfs-devices-node b/Documentation/ABI/stable/sysfs-devices-node
> index a02707cb7cbc..2d0e023f22a7 100644
> --- a/Documentation/ABI/stable/sysfs-devices-node
> +++ b/Documentation/ABI/stable/sysfs-devices-node
> @@ -227,3 +227,12 @@ Contact:	Jiaqi Yan <jiaqiyan@google.com>
>  Description:
>  		Of the raw poisoned pages on a NUMA node, how many pages are
>  		recovered by memory error recovery attempt.
> +
> +What:		/sys/devices/system/node/nodeX/reclaim
> +Date:		June 2025
> +Contact:	Linux Memory Management list <linux-mm@kvack.org>
> +Description:
> +		Perform user-triggered proactive reclaim on a NUMA node.
> +		This interface is equivalent to the memcg variant.
> +
> +		See Documentation/admin-guide/cgroup-v2.rst
> diff --git a/drivers/base/node.c b/drivers/base/node.c
> index 6d66382dae65..548b532a2129 100644
> --- a/drivers/base/node.c
> +++ b/drivers/base/node.c
> @@ -659,6 +659,7 @@ static int register_node(struct node *node, int num)
>  	} else {
>  		hugetlb_register_node(node);
>  		compaction_register_node(node);
> +		reclaim_register_node(node);
>  	}
>  
>  	return error;
> @@ -675,6 +676,7 @@ void unregister_node(struct node *node)
>  {
>  	hugetlb_unregister_node(node);
>  	compaction_unregister_node(node);
> +	reclaim_unregister_node(node);
>  	node_remove_accesses(node);
>  	node_remove_caches(node);
>  	device_unregister(&node->dev);
> diff --git a/include/linux/swap.h b/include/linux/swap.h
> index bc0e1c275fc0..dac7ba98783d 100644
> --- a/include/linux/swap.h
> +++ b/include/linux/swap.h
> @@ -431,6 +431,22 @@ extern unsigned long shrink_all_memory(unsigned long nr_pages);
>  extern int vm_swappiness;
>  long remove_mapping(struct address_space *mapping, struct folio *folio);
>  
> +#if defined(CONFIG_SYSFS) && defined(CONFIG_NUMA)
> +extern int reclaim_register_node(struct node *node);
> +extern void reclaim_unregister_node(struct node *node);
> +
> +#else
> +
> +static inline int reclaim_register_node(struct node *node)
> +{
> +	return 0;
> +}
> +
> +static inline void reclaim_unregister_node(struct node *node)
> +{
> +}
> +#endif /* CONFIG_SYSFS && CONFIG_NUMA */
> +
>  #ifdef CONFIG_NUMA
>  extern int sysctl_min_unmapped_ratio;
>  extern int sysctl_min_slab_ratio;
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index cdd9cb97fb79..f77feb75c678 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -94,10 +94,8 @@ struct scan_control {
>  	unsigned long	anon_cost;
>  	unsigned long	file_cost;
>  
> -#ifdef CONFIG_MEMCG
>  	/* Swappiness value for proactive reclaim. Always use sc_swappiness()! */
>  	int *proactive_swappiness;
> -#endif
>  
>  	/* Can active folios be deactivated as part of reclaim? */
>  #define DEACTIVATE_ANON 1
> @@ -121,7 +119,7 @@ struct scan_control {
>  	/* Has cache_trim_mode failed at least once? */
>  	unsigned int cache_trim_mode_failed:1;
>  
> -	/* Proactive reclaim invoked by userspace through memory.reclaim */
> +	/* Proactive reclaim invoked by userspace */
>  	unsigned int proactive:1;
>  
>  	/*
> @@ -7732,13 +7730,15 @@ static const match_table_t tokens = {
>  	{ MEMORY_RECLAIM_NULL, NULL },
>  };
>  
> -int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
> +int user_proactive_reclaim(char *buf,
> +			   struct mem_cgroup *memcg, pg_data_t *pgdat)
>  {
>  	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
>  	unsigned long nr_to_reclaim, nr_reclaimed = 0;
>  	int swappiness = -1;
>  	char *old_buf, *start;
>  	substring_t args[MAX_OPT_ARGS];
> +	gfp_t gfp_mask = GFP_KERNEL;
>  
>  	if (!buf || (!memcg && !pgdat))
>  		return -EINVAL;
> @@ -7792,11 +7792,29 @@ int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat
>  			reclaim_options = MEMCG_RECLAIM_MAY_SWAP |
>  					  MEMCG_RECLAIM_PROACTIVE;
>  			reclaimed = try_to_free_mem_cgroup_pages(memcg,
> -						 batch_size, GFP_KERNEL,
> +						 batch_size, gfp_mask,
>  						 reclaim_options,
>  						 swappiness == -1 ? NULL : &swappiness);
>  		} else {
> -			return -EINVAL;
> +			struct scan_control sc = {
> +				.gfp_mask = current_gfp_context(gfp_mask),
> +				.reclaim_idx = gfp_zone(gfp_mask),
> +				.proactive_swappiness = swappiness == -1 ? NULL : &swappiness,
> +				.priority = DEF_PRIORITY,
> +				.may_writepage = !laptop_mode,
> +				.nr_to_reclaim = max(batch_size, SWAP_CLUSTER_MAX),
> +				.may_unmap = 1,
> +				.may_swap = 1,
> +				.proactive = 1,
> +			};
> +
> +			if (test_and_set_bit_lock(PGDAT_RECLAIM_LOCKED,
> +						  &pgdat->flags))
> +				return -EAGAIN;

Isn't EBUSY a better choice here?
At least to distinguish between no reclaimable memory left and
somebody else is abusing the same interface cases.

Re: [PATCH 4/4] mm: introduce per-node proactive reclaim interface

[Michal Hocko]

On Thu 17-07-25 14:49:24, Hillf Danton wrote:
> Davidlohr Bueso <dave@stgolabs.net> writes:
> 
> > This adds support for allowing proactive reclaim in general on a
> > NUMA system. A per-node interface extends support for beyond a
> > memcg-specific interface, respecting the current semantics of
> > memory.reclaim: respecting aging LRU and not supporting
> > artificially triggering eviction on nodes belonging to non-bottom
> > tiers.
> >
> > This patch allows userspace to do:
> >
> >      echo "512M swappiness=10" > /sys/devices/system/node/nodeX/reclaim
> >
> When kswapd is active, this is not needed.
> When kswapd is idle, why is this needed?

Usecases are described in the section of the email you haven't quoted in your reply.
-- 
Michal Hocko
SUSE Labs

Re: [PATCH 4/4] mm: introduce per-node proactive reclaim interface

[Davidlohr Bueso]

On Wed, 16 Jul 2025, Roman Gushchin wrote:

>Davidlohr Bueso <dave@stgolabs.net> writes:
>> @@ -7792,11 +7792,29 @@ int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat
>>			reclaim_options = MEMCG_RECLAIM_MAY_SWAP |
>>					  MEMCG_RECLAIM_PROACTIVE;
>>			reclaimed = try_to_free_mem_cgroup_pages(memcg,
>> -						 batch_size, GFP_KERNEL,
>> +						 batch_size, gfp_mask,
>>						 reclaim_options,
>>						 swappiness == -1 ? NULL : &swappiness);
>>		} else {
>> -			return -EINVAL;
>> +			struct scan_control sc = {
>> +				.gfp_mask = current_gfp_context(gfp_mask),
>> +				.reclaim_idx = gfp_zone(gfp_mask),
>> +				.proactive_swappiness = swappiness == -1 ? NULL : &swappiness,
>> +				.priority = DEF_PRIORITY,
>> +				.may_writepage = !laptop_mode,
>> +				.nr_to_reclaim = max(batch_size, SWAP_CLUSTER_MAX),
>> +				.may_unmap = 1,
>> +				.may_swap = 1,
>> +				.proactive = 1,
>> +			};
>> +
>> +			if (test_and_set_bit_lock(PGDAT_RECLAIM_LOCKED,
>> +						  &pgdat->flags))
>> +				return -EAGAIN;
>
>Isn't EBUSY a better choice here?
>At least to distinguish between no reclaimable memory left and
>somebody else is abusing the same interface cases.

Yes, I agree.

Thanks,
Davidlohr

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Davidlohr Bueso]

On Wed, 16 Jul 2025, Roman Gushchin wrote:

>Davidlohr Bueso <dave@stgolabs.net> writes:
>
>> This adds a general call for both parsing as well as the
>> common reclaim semantics. memcg is still the only user and
>> no change in semantics.
>>
>> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
>> ---
>>  mm/internal.h   |  2 +
>>  mm/memcontrol.c | 77 ++------------------------------------
>>  mm/vmscan.c     | 98 +++++++++++++++++++++++++++++++++++++++++++++++++
>>  3 files changed, 104 insertions(+), 73 deletions(-)
>> ...
>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>> index c13c01eb0b42..63ddec550c3b 100644
>> --- a/mm/vmscan.c
>> +++ b/mm/vmscan.c
>> @@ -57,6 +57,7 @@
>>  #include <linux/rculist_nulls.h>
>>  #include <linux/random.h>
>>  #include <linux/mmu_notifier.h>
>> +#include <linux/parser.h>
>>
>>  #include <asm/tlbflush.h>
>>  #include <asm/div64.h>
>> @@ -6714,6 +6715,15 @@ unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>>
>>  	return nr_reclaimed;
>>  }
>> +#else
>> +unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg,
>> +					   unsigned long nr_pages,
>> +					   gfp_t gfp_mask,
>> +					   unsigned int reclaim_options,
>> +					   int *swappiness)
>> +{
>> +	return 0;
>> +}
>>  #endif
>>
>>  static void kswapd_age_node(struct pglist_data *pgdat, struct scan_control *sc)
>> @@ -7708,6 +7718,94 @@ int node_reclaim(struct pglist_data *pgdat, gfp_t gfp_mask, unsigned int order)
>>
>>  	return ret;
>>  }
>> +
>> +enum {
>> +	MEMORY_RECLAIM_SWAPPINESS = 0,
>> +	MEMORY_RECLAIM_SWAPPINESS_MAX,
>> +	MEMORY_RECLAIM_NULL,
>> +};
>> +static const match_table_t tokens = {
>> +	{ MEMORY_RECLAIM_SWAPPINESS, "swappiness=%d"},
>> +	{ MEMORY_RECLAIM_SWAPPINESS_MAX, "swappiness=max"},
>> +	{ MEMORY_RECLAIM_NULL, NULL },
>> +};
>> +
>> +int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
>> +{
>> +	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
>> +	unsigned long nr_to_reclaim, nr_reclaimed = 0;
>> +	int swappiness = -1;
>> +	char *old_buf, *start;
>> +	substring_t args[MAX_OPT_ARGS];
>> +
>> +	if (!buf || (!memcg && !pgdat))
>> +		return -EINVAL;
>> +
>> +	buf = strstrip(buf);
>> +
>> +	old_buf = buf;
>> +	nr_to_reclaim = memparse(buf, &buf) / PAGE_SIZE;
>> +	if (buf == old_buf)
>> +		return -EINVAL;
>> +
>> +	buf = strstrip(buf);
>
>To be honest, not a big fan of this refactoring. Effectively parts of
>the memcg user interface are moved into mm/vmscan.c. I get that you want
>to use the exact same interface somewhere else, but still...

I disagree, further this is no different than other callers in vmscan.c
around memcg.

>Is it possible to keep it in mm/memcontrol.c?

Why? now proactive reclaim is not special to memcg... unless strong reasons
it makes little sense to keep it there.

>Also maybe split the actual reclaim mechanism and user's input parsing?

I tried something like this initially, and ended up prefering this way.

Further, this approach limits the reachability of the input parsing logic, and
the interface is already being an exception to the one-value per file "rule".

Thanks,
Davidlohr

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Shakeel Butt]

On Mon, Jun 23, 2025 at 11:58:49AM -0700, Davidlohr Bueso wrote:
> +
> +int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
> +{
> +	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
> +	unsigned long nr_to_reclaim, nr_reclaimed = 0;
> +	int swappiness = -1;
> +	char *old_buf, *start;
> +	substring_t args[MAX_OPT_ARGS];
> +
> +	if (!buf || (!memcg && !pgdat))

I don't think this series is adding a use-case where both memcg and
pgdat are non-NULL, so let's error out on that as well.

Re: [PATCH 3/4] mm/vmscan: make __node_reclaim() more generic

[Shakeel Butt]

On Mon, Jun 23, 2025 at 11:58:50AM -0700, Davidlohr Bueso wrote:
> As this will be called from non page allocator paths for
> proactive reclaim, allow users to pass the sc and nr of
> pages, and adjust the return value as well. No change in
> semantics.
> 
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Acked-by: Shakeel Butt <shakeel.butt@linux.dev>

Re: [PATCH 4/4] mm: introduce per-node proactive reclaim interface

[Shakeel Butt]

On Mon, Jun 23, 2025 at 11:58:51AM -0700, Davidlohr Bueso wrote:
> This adds support for allowing proactive reclaim in general on a
> NUMA system. A per-node interface extends support for beyond a
> memcg-specific interface, respecting the current semantics of
> memory.reclaim: respecting aging LRU and not supporting
> artificially triggering eviction on nodes belonging to non-bottom
> tiers.
> 
> This patch allows userspace to do:
> 
>      echo "512M swappiness=10" > /sys/devices/system/node/nodeX/reclaim
> 
> One of the premises for this is to semantically align as best as
> possible with memory.reclaim. During a brief time memcg did
> support nodemask until 55ab834a86a9 (Revert "mm: add nodes=
> arg to memory.reclaim"), for which semantics around reclaim
> (eviction) vs demotion were not clear, rendering charging
> expectations to be broken.
> 
> With this approach:
> 
> 1. Users who do not use memcg can benefit from proactive reclaim.
> The memcg interface is not NUMA aware and there are usecases that
> are focusing on NUMA balancing rather than workload memory footprint.
> 
> 2. Proactive reclaim on top tiers will trigger demotion, for which
> memory is still byte-addressable. Reclaiming on the bottom nodes
> will trigger evicting to swap (the traditional sense of reclaim).
> This follows the semantics of what is today part of the aging process
> on tiered memory, mirroring what every other form of reclaim does
> (reactive and memcg proactive reclaim). Furthermore per-node proactive
> reclaim is not as susceptible to the memcg charging problem mentioned
> above.
> 
> 3. Unlike the nodes= arg, this interface avoids confusing semantics,
> such as what exactly the user wants when mixing top-tier and low-tier
> nodes in the nodemask. Further per-node interface is less exposed to
> "free up memory in my container" usecases, where eviction is intended.
> 
> 4. Users that *really* want to free up memory can use proactive reclaim
> on nodes knowingly to be on the bottom tiers to force eviction in a
> natural way - higher access latencies are still better than swap.
> If compelled, while no guarantees and perhaps not worth the effort,
> users could also also potentially follow a ladder-like approach to
> eventually free up the memory. Alternatively, perhaps an 'evict' option
> could be added to the parameters for both memory.reclaim and per-node
> interfaces to force this action unconditionally.
> 
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

After Roman's suggestion, you can add:

Acked-by: Shakeel Butt <shakeel.butt@linux.dev>

Re: [PATCH 4/4] mm: introduce per-node proactive reclaim interface

[Andrew Morton]

On Thu, 17 Jul 2025 09:26:37 -0700 Davidlohr Bueso <dave@stgolabs.net> wrote:

> >> +			if (test_and_set_bit_lock(PGDAT_RECLAIM_LOCKED,
> >> +						  &pgdat->flags))
> >> +				return -EAGAIN;
> >
> >Isn't EBUSY a better choice here?
> >At least to distinguish between no reclaimable memory left and
> >somebody else is abusing the same interface cases.
> 
> Yes, I agree.

From: Andrew Morton <akpm@linux-foundation.org>
Subject: mm-introduce-per-node-proactive-reclaim-interface-fix
Date: Thu Jul 17 03:44:14 PM PDT 2025

user_proactive_reclaim(): return -EBUSY on PGDAT_RECLAIM_LOCKED
contention, per Roman

Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Shakeel Butt <shakeel.butt@linux.dev>
Cc: Yosry Ahmed <yosryahmed@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 mm/vmscan.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- a/mm/vmscan.c~mm-introduce-per-node-proactive-reclaim-interface-fix
+++ a/mm/vmscan.c
@@ -7818,7 +7818,7 @@ int user_proactive_reclaim(char *buf,
 
 			if (test_and_set_bit_lock(PGDAT_RECLAIM_LOCKED,
 						  &pgdat->flags))
-				return -EAGAIN;
+				return -EBUSY;
 
 			reclaimed = __node_reclaim(pgdat, gfp_mask,
 						   batch_size, &sc);
_

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Andrew Morton]

On Thu, 17 Jul 2025 15:17:09 -0700 Shakeel Butt <shakeel.butt@linux.dev> wrote:

> On Mon, Jun 23, 2025 at 11:58:49AM -0700, Davidlohr Bueso wrote:
> > +
> > +int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
> > +{
> > +	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
> > +	unsigned long nr_to_reclaim, nr_reclaimed = 0;
> > +	int swappiness = -1;
> > +	char *old_buf, *start;
> > +	substring_t args[MAX_OPT_ARGS];
> > +
> > +	if (!buf || (!memcg && !pgdat))
> 
> I don't think this series is adding a use-case where both memcg and
> pgdat are non-NULL, so let's error out on that as well.

As a followup, please.  This has been in -next for four weeks and I'd
prefer not to have to route around it (again).

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Davidlohr Bueso]

On Thu, 17 Jul 2025, Andrew Morton wrote:

>On Thu, 17 Jul 2025 15:17:09 -0700 Shakeel Butt <shakeel.butt@linux.dev> wrote:
>
>> On Mon, Jun 23, 2025 at 11:58:49AM -0700, Davidlohr Bueso wrote:
>> > +
>> > +int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
>> > +{
>> > +	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
>> > +	unsigned long nr_to_reclaim, nr_reclaimed = 0;
>> > +	int swappiness = -1;
>> > +	char *old_buf, *start;
>> > +	substring_t args[MAX_OPT_ARGS];
>> > +
>> > +	if (!buf || (!memcg && !pgdat))
>>
>> I don't think this series is adding a use-case where both memcg and
>> pgdat are non-NULL, so let's error out on that as well.
>
>As a followup, please.  This has been in -next for four weeks and I'd
>prefer not to have to route around it (again).
>

From: Davidlohr Bueso <dave@stgolabs.net>
Date: Thu, 17 Jul 2025 16:53:24 -0700
Subject: [PATCH] mm-introduce-per-node-proactive-reclaim-interface-fix

Both memcg and node is also a bogus case, per Shakeel.

Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
  mm/vmscan.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index 4598d18ba256..d5f7b1703234 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -7758,7 +7758,7 @@ int user_proactive_reclaim(char *buf,
  	substring_t args[MAX_OPT_ARGS];
  	gfp_t gfp_mask = GFP_KERNEL;
  
-	if (!buf || (!memcg && !pgdat))
+	if (!buf || (!memcg && !pgdat) || (memcg && pgdat))
  		return -EINVAL;
  
  	buf = strstrip(buf);
-- 
2.39.5

Re: [PATCH 2/4] mm/memcg: make memory.reclaim interface generic

[Shakeel Butt]

On Thu, Jul 17, 2025 at 04:56:04PM -0700, Davidlohr Bueso wrote:
> On Thu, 17 Jul 2025, Andrew Morton wrote:
> 
> > On Thu, 17 Jul 2025 15:17:09 -0700 Shakeel Butt <shakeel.butt@linux.dev> wrote:
> > 
> > > On Mon, Jun 23, 2025 at 11:58:49AM -0700, Davidlohr Bueso wrote:
> > > > +
> > > > +int user_proactive_reclaim(char *buf, struct mem_cgroup *memcg, pg_data_t *pgdat)
> > > > +{
> > > > +	unsigned int nr_retries = MAX_RECLAIM_RETRIES;
> > > > +	unsigned long nr_to_reclaim, nr_reclaimed = 0;
> > > > +	int swappiness = -1;
> > > > +	char *old_buf, *start;
> > > > +	substring_t args[MAX_OPT_ARGS];
> > > > +
> > > > +	if (!buf || (!memcg && !pgdat))
> > > 
> > > I don't think this series is adding a use-case where both memcg and
> > > pgdat are non-NULL, so let's error out on that as well.
> > 
> > As a followup, please.  This has been in -next for four weeks and I'd
> > prefer not to have to route around it (again).
> > 
> 
> From: Davidlohr Bueso <dave@stgolabs.net>
> Date: Thu, 17 Jul 2025 16:53:24 -0700
> Subject: [PATCH] mm-introduce-per-node-proactive-reclaim-interface-fix
> 
> Both memcg and node is also a bogus case, per Shakeel.
> 
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

With this, I think we are good. We can always refactor and move code
around to our taste but interface and functionality wise this is fine.

Acked-by: Shakeel Butt <shakeel.butt@linux.dev>