Re: [PATCH v2 1/3] mm,slub: Do not special case N_NORMAL nodes for slab_nodes

[David Hildenbrand <david@redhat.com>]

On 08.04.25 10:41, Oscar Salvador wrote:
> Currently, slab_mem_going_going_callback() checks whether the node has
> N_NORMAL memory in order to be set in slab_nodes.
> While it is true that gettind rid of that enforcing would mean
> ending up with movables nodes in slab_nodes, the memory waste that comes
> with that is negligible.
> 
> So stop checking for status_change_nid_normal and just use status_change_nid
> instead which works for both types of memory.
> 
> Also, once we allocate the kmem_cache_node cache  for the node in
> slab_mem_online_callback(), we never deallocate it in
> slab_mem_off_callback() when the node goes memoryless, so we can just
> get rid of it.
> 
> The only side effect is that we will stop clearing the node from slab_nodes.
> 

Feel free to add a Suggested-by: if you think it applies.


Do we have to take care of the N_NORMAL_MEMORY check in kmem_cache_init() ? Likely it
would have to be a N_MEMORY check.


But, I was wondering if we could get rid of the "slab_nodes" thingy as a first step?

 From 518a2b83a9c5bd85d74ddabbc36ce5d181a88ed6 Mon Sep 17 00:00:00 2001
From: David Hildenbrand <david@redhat.com>
Date: Tue, 8 Apr 2025 12:16:13 +0200
Subject: [PATCH] tmp

Signed-off-by: David Hildenbrand <david@redhat.com>
---
  mm/slub.c | 56 ++++---------------------------------------------------
  1 file changed, 4 insertions(+), 52 deletions(-)

diff --git a/mm/slub.c b/mm/slub.c
index b46f87662e71d..afe31149e7f4e 100644
--- a/mm/slub.c
+++ b/mm/slub.c
@@ -445,14 +445,6 @@ static inline struct kmem_cache_node *get_node(struct kmem_cache *s, int node)
  	for (__node = 0; __node < nr_node_ids; __node++) \
  		 if ((__n = get_node(__s, __node)))
  
-/*
- * Tracks for which NUMA nodes we have kmem_cache_nodes allocated.
- * Corresponds to node_state[N_NORMAL_MEMORY], but can temporarily
- * differ during memory hotplug/hotremove operations.
- * Protected by slab_mutex.
- */
-static nodemask_t slab_nodes;
-
  #ifndef CONFIG_SLUB_TINY
  /*
   * Workqueue used for flush_cpu_slab().
@@ -3706,10 +3698,9 @@ static void *___slab_alloc(struct kmem_cache *s, gfp_t gfpflags, int node,
  	if (!slab) {
  		/*
  		 * if the node is not online or has no normal memory, just
-		 * ignore the node constraint
+		 * ignore the node constraint.
  		 */
-		if (unlikely(node != NUMA_NO_NODE &&
-			     !node_isset(node, slab_nodes)))
+		if (unlikely(node != NUMA_NO_NODE && !node_state(node, N_NORMAL_MEMORY)))
  			node = NUMA_NO_NODE;
  		goto new_slab;
  	}
@@ -3719,7 +3710,7 @@ static void *___slab_alloc(struct kmem_cache *s, gfp_t gfpflags, int node,
  		 * same as above but node_match() being false already
  		 * implies node != NUMA_NO_NODE
  		 */
-		if (!node_isset(node, slab_nodes)) {
+		if (!node_state(node, N_NORMAL_MEMORY)) {
  			node = NUMA_NO_NODE;
  		} else {
  			stat(s, ALLOC_NODE_MISMATCH);
@@ -5623,7 +5614,7 @@ static int init_kmem_cache_nodes(struct kmem_cache *s)
  {
  	int node;
  
-	for_each_node_mask(node, slab_nodes) {
+	for_each_node_state(node, N_NORMAL_MEMORY) {
  		struct kmem_cache_node *n;
  
  		if (slab_state == DOWN) {
@@ -6164,30 +6155,6 @@ static int slab_mem_going_offline_callback(void *arg)
  	return 0;
  }
  
-static void slab_mem_offline_callback(void *arg)
-{
-	struct memory_notify *marg = arg;
-	int offline_node;
-
-	offline_node = marg->status_change_nid_normal;
-
-	/*
-	 * If the node still has available memory. we need kmem_cache_node
-	 * for it yet.
-	 */
-	if (offline_node < 0)
-		return;
-
-	mutex_lock(&slab_mutex);
-	node_clear(offline_node, slab_nodes);
-	/*
-	 * We no longer free kmem_cache_node structures here, as it would be
-	 * racy with all get_node() users, and infeasible to protect them with
-	 * slab_mutex.
-	 */
-	mutex_unlock(&slab_mutex);
-}
-
  static int slab_mem_going_online_callback(void *arg)
  {
  	struct kmem_cache_node *n;
@@ -6229,11 +6196,6 @@ static int slab_mem_going_online_callback(void *arg)
  		init_kmem_cache_node(n);
  		s->node[nid] = n;
  	}
-	/*
-	 * Any cache created after this point will also have kmem_cache_node
-	 * initialized for the new node.
-	 */
-	node_set(nid, slab_nodes);
  out:
  	mutex_unlock(&slab_mutex);
  	return ret;
@@ -6253,8 +6215,6 @@ static int slab_memory_callback(struct notifier_block *self,
  		break;
  	case MEM_OFFLINE:
  	case MEM_CANCEL_ONLINE:
-		slab_mem_offline_callback(arg);
-		break;
  	case MEM_ONLINE:
  	case MEM_CANCEL_OFFLINE:
  		break;
@@ -6309,7 +6269,6 @@ void __init kmem_cache_init(void)
  {
  	static __initdata struct kmem_cache boot_kmem_cache,
  		boot_kmem_cache_node;
-	int node;
  
  	if (debug_guardpage_minorder())
  		slub_max_order = 0;
@@ -6321,13 +6280,6 @@ void __init kmem_cache_init(void)
  	kmem_cache_node = &boot_kmem_cache_node;
  	kmem_cache = &boot_kmem_cache;
  
-	/*
-	 * Initialize the nodemask for which we will allocate per node
-	 * structures. Here we don't need taking slab_mutex yet.
-	 */
-	for_each_node_state(node, N_NORMAL_MEMORY)
-		node_set(node, slab_nodes);
-
  	create_boot_cache(kmem_cache_node, "kmem_cache_node",
  			sizeof(struct kmem_cache_node),
  			SLAB_HWCACHE_ALIGN | SLAB_NO_OBJ_EXT, 0, 0);
-- 
2.48.1


Not sure if there are any races to consider ... just an idea.

-- 
Cheers,

David / dhildenb