[RFC PATCH 3/7] memcg: make mod_memcg_state re-entrant safe against irqs

Shakeel Butt posted 7 patches 7 months, 1 week ago
There is a newer version of this series
[RFC PATCH 3/7] memcg: make mod_memcg_state re-entrant safe against irqs
Posted by Shakeel Butt 7 months, 1 week ago
Let's make mod_memcg_state re-entrant safe against irqs. The only thing
needed is to convert the usage of __this_cpu_add() to this_cpu_add().
In addition, with re-entrant safety, there is no need to disable irqs.

mod_memcg_state() is not safe against nmi, so let's add warning if
someone tries to call it in nmi context.

Signed-off-by: Shakeel Butt <shakeel.butt@linux.dev>
---
 include/linux/memcontrol.h | 20 ++------------------
 mm/memcontrol.c            | 12 ++++++++----
 2 files changed, 10 insertions(+), 22 deletions(-)

diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h
index ed9acb68652a..84e2cea7e666 100644
--- a/include/linux/memcontrol.h
+++ b/include/linux/memcontrol.h
@@ -911,19 +911,9 @@ struct mem_cgroup *mem_cgroup_get_oom_group(struct task_struct *victim,
 					    struct mem_cgroup *oom_domain);
 void mem_cgroup_print_oom_group(struct mem_cgroup *memcg);
 
-void __mod_memcg_state(struct mem_cgroup *memcg, enum memcg_stat_item idx,
-		       int val);
-
 /* idx can be of type enum memcg_stat_item or node_stat_item */
-static inline void mod_memcg_state(struct mem_cgroup *memcg,
-				   enum memcg_stat_item idx, int val)
-{
-	unsigned long flags;
-
-	local_irq_save(flags);
-	__mod_memcg_state(memcg, idx, val);
-	local_irq_restore(flags);
-}
+void mod_memcg_state(struct mem_cgroup *memcg,
+		     enum memcg_stat_item idx, int val);
 
 static inline void mod_memcg_page_state(struct page *page,
 					enum memcg_stat_item idx, int val)
@@ -1390,12 +1380,6 @@ static inline void mem_cgroup_print_oom_group(struct mem_cgroup *memcg)
 {
 }
 
-static inline void __mod_memcg_state(struct mem_cgroup *memcg,
-				     enum memcg_stat_item idx,
-				     int nr)
-{
-}
-
 static inline void mod_memcg_state(struct mem_cgroup *memcg,
 				   enum memcg_stat_item idx,
 				   int nr)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 62450e7991d8..373d36cae069 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -681,12 +681,12 @@ static int memcg_state_val_in_pages(int idx, int val)
 }
 
 /**
- * __mod_memcg_state - update cgroup memory statistics
+ * mod_memcg_state - update cgroup memory statistics
  * @memcg: the memory cgroup
  * @idx: the stat item - can be enum memcg_stat_item or enum node_stat_item
  * @val: delta to add to the counter, can be negative
  */
-void __mod_memcg_state(struct mem_cgroup *memcg, enum memcg_stat_item idx,
+void mod_memcg_state(struct mem_cgroup *memcg, enum memcg_stat_item idx,
 		       int val)
 {
 	int i = memcg_stats_index(idx);
@@ -698,9 +698,13 @@ void __mod_memcg_state(struct mem_cgroup *memcg, enum memcg_stat_item idx,
 	if (WARN_ONCE(BAD_STAT_IDX(i), "%s: missing stat item %d\n", __func__, idx))
 		return;
 
+	if (WARN_ONCE(in_nmi(), "%s: called in nmi context for stat item %d\n",
+		      __func__, idx))
+		return;
+
 	cpu = get_cpu();
 
-	__this_cpu_add(memcg->vmstats_percpu->state[i], val);
+	this_cpu_add(memcg->vmstats_percpu->state[i], val);
 	val = memcg_state_val_in_pages(idx, val);
 	memcg_rstat_updated(memcg, val, cpu);
 	trace_mod_memcg_state(memcg, idx, val);
@@ -2969,7 +2973,7 @@ static void drain_obj_stock(struct obj_stock_pcp *stock)
 
 			memcg = get_mem_cgroup_from_objcg(old);
 
-			__mod_memcg_state(memcg, MEMCG_KMEM, -nr_pages);
+			mod_memcg_state(memcg, MEMCG_KMEM, -nr_pages);
 			memcg1_account_kmem(memcg, -nr_pages);
 			if (!mem_cgroup_is_root(memcg))
 				memcg_uncharge(memcg, nr_pages);
-- 
2.47.1
Re: [RFC PATCH 3/7] memcg: make mod_memcg_state re-entrant safe against irqs
Posted by Vlastimil Babka 7 months, 1 week ago
On 5/13/25 05:13, Shakeel Butt wrote:
> Let's make mod_memcg_state re-entrant safe against irqs. The only thing
> needed is to convert the usage of __this_cpu_add() to this_cpu_add().
> In addition, with re-entrant safety, there is no need to disable irqs.
> 
> mod_memcg_state() is not safe against nmi, so let's add warning if
> someone tries to call it in nmi context.
> 
> Signed-off-by: Shakeel Butt <shakeel.butt@linux.dev>

Also a good cleanup.

Acked-by: Vlastimil Babka <vbabka@suse.cz>