[PATCH -v4 next 0/4] Some cleanup for memcg

[PATCH -v4 next 0/4] Some cleanup for memcg

[Chen Ridong]

From: Chen Ridong <chenridong@huawei.com>

Some cleanup for memcg.

In this series, patches 1-3 have been reviewed.
Patch 4 is new, which adds CONFIG_MEMCG_V1 for the local functions,
suggested by Johannes Weiner.

---

v3->v4:
 - keep the 'local' functions in the memcontrol-v1.c and add
   CONFIG_MEMCG_V1 for them.

v2->v3:
 - move the wrapper function definitions to header files.
 - add a patch to move the 'local' functions to the memcontrol-v1.c.

v1->v2:
 - drop the patch 'simplify the mem_cgroup_update_lru_size function'.
 - for patch 3, rename '__refill_obj_stock' to replace_stock_objcg, and
   keep the 'objcg equal' check in the calling functions.

Chen Ridong (4):
  memcg: use OFP_PEAK_UNSET instead of -1
  memcg: call the free function when allocation of pn fails
  memcg: factor out the replace_stock_objcg function
  memcg: add CONFIG_MEMCG_V1 for 'local' functions

 mm/memcontrol-v1.h |  6 ++---
 mm/memcontrol.c    | 61 ++++++++++++++++++++++++++--------------------
 2 files changed, 38 insertions(+), 29 deletions(-)

-- 
2.34.1

[PATCH -v4 next 1/4] memcg: use OFP_PEAK_UNSET instead of -1

[Chen Ridong]

From: Chen Ridong <chenridong@huawei.com>

The 'OFP_PEAK_UNSET' has been defined, use it instead of '-1'.

Signed-off-by: Chen Ridong <chenridong@huawei.com>
Reviewed-by: Michal Koutný <mkoutny@suse.com>
Acked-by: David Finkel <davidf@vimeo.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/memcontrol.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 46f8b372d212..05a32c860554 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4004,7 +4004,7 @@ static ssize_t peak_write(struct kernfs_open_file *of, char *buf, size_t nbytes,
 			WRITE_ONCE(peer_ctx->value, usage);
 
 	/* initial write, register watcher */
-	if (ofp->value == -1)
+	if (ofp->value == OFP_PEAK_UNSET)
 		list_add(&ofp->list, watchers);
 
 	WRITE_ONCE(ofp->value, usage);
-- 
2.34.1

Re: [PATCH -v4 next 1/4] memcg: use OFP_PEAK_UNSET instead of -1

[Roman Gushchin]

On Fri, Jan 24, 2025 at 07:35:11AM +0000, Chen Ridong wrote:
> From: Chen Ridong <chenridong@huawei.com>
> 
> The 'OFP_PEAK_UNSET' has been defined, use it instead of '-1'.
> 
> Signed-off-by: Chen Ridong <chenridong@huawei.com>
> Reviewed-by: Michal Koutný <mkoutny@suse.com>
> Acked-by: David Finkel <davidf@vimeo.com>
> Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>

I think I acked it previously?

Anyway,

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

Thanks!

> ---
>  mm/memcontrol.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 46f8b372d212..05a32c860554 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -4004,7 +4004,7 @@ static ssize_t peak_write(struct kernfs_open_file *of, char *buf, size_t nbytes,
>  			WRITE_ONCE(peer_ctx->value, usage);
>  
>  	/* initial write, register watcher */
> -	if (ofp->value == -1)
> +	if (ofp->value == OFP_PEAK_UNSET)
>  		list_add(&ofp->list, watchers);
>  
>  	WRITE_ONCE(ofp->value, usage);
> -- 
> 2.34.1
> 

[PATCH -v4 next 2/4] memcg: call the free function when allocation of pn fails

[Chen Ridong]

From: Chen Ridong <chenridong@huawei.com>

The 'free_mem_cgroup_per_node_info' function is used to free
the 'mem_cgroup_per_node' struct. Using 'pn' as the input for the
free_mem_cgroup_per_node_info function will be much clearer.
Call 'free_mem_cgroup_per_node_info' when 'alloc_mem_cgroup_per_node_info'
fails, to free 'pn' as a whole, which makes the code more cohesive.

Signed-off-by: Chen Ridong <chenridong@huawei.com>
Reviewed-by: Michal Koutný <mkoutny@suse.com>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/memcontrol.c | 27 ++++++++++++---------------
 1 file changed, 12 insertions(+), 15 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 05a32c860554..98f84a9fa228 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -3424,6 +3424,16 @@ struct mem_cgroup *mem_cgroup_get_from_ino(unsigned long ino)
 }
 #endif
 
+static void free_mem_cgroup_per_node_info(struct mem_cgroup_per_node *pn)
+{
+	if (!pn)
+		return;
+
+	free_percpu(pn->lruvec_stats_percpu);
+	kfree(pn->lruvec_stats);
+	kfree(pn);
+}
+
 static bool alloc_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node)
 {
 	struct mem_cgroup_per_node *pn;
@@ -3448,23 +3458,10 @@ static bool alloc_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node)
 	memcg->nodeinfo[node] = pn;
 	return true;
 fail:
-	kfree(pn->lruvec_stats);
-	kfree(pn);
+	free_mem_cgroup_per_node_info(pn);
 	return false;
 }
 
-static void free_mem_cgroup_per_node_info(struct mem_cgroup *memcg, int node)
-{
-	struct mem_cgroup_per_node *pn = memcg->nodeinfo[node];
-
-	if (!pn)
-		return;
-
-	free_percpu(pn->lruvec_stats_percpu);
-	kfree(pn->lruvec_stats);
-	kfree(pn);
-}
-
 static void __mem_cgroup_free(struct mem_cgroup *memcg)
 {
 	int node;
@@ -3472,7 +3469,7 @@ static void __mem_cgroup_free(struct mem_cgroup *memcg)
 	obj_cgroup_put(memcg->orig_objcg);
 
 	for_each_node(node)
-		free_mem_cgroup_per_node_info(memcg, node);
+		free_mem_cgroup_per_node_info(memcg->nodeinfo[node]);
 	memcg1_free_events(memcg);
 	kfree(memcg->vmstats);
 	free_percpu(memcg->vmstats_percpu);
-- 
2.34.1

Re: [PATCH -v4 next 2/4] memcg: call the free function when allocation of pn fails

[Roman Gushchin]

On Fri, Jan 24, 2025 at 07:35:12AM +0000, Chen Ridong wrote:
> From: Chen Ridong <chenridong@huawei.com>
> 
> The 'free_mem_cgroup_per_node_info' function is used to free
> the 'mem_cgroup_per_node' struct. Using 'pn' as the input for the
> free_mem_cgroup_per_node_info function will be much clearer.
> Call 'free_mem_cgroup_per_node_info' when 'alloc_mem_cgroup_per_node_info'
> fails, to free 'pn' as a whole, which makes the code more cohesive.
> 
> Signed-off-by: Chen Ridong <chenridong@huawei.com>
> Reviewed-by: Michal Koutný <mkoutny@suse.com>
> Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
> Acked-by: Johannes Weiner <hannes@cmpxchg.org>

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

[PATCH -v4 next 3/4] memcg: factor out the replace_stock_objcg function

[Chen Ridong]

From: Chen Ridong <chenridong@huawei.com>

Factor out the 'replace_stock_objcg' function to make the code more
cohesive.

Signed-off-by: Chen Ridong <chenridong@huawei.com>
Reviewed-by: Roman Gushchin <roman.gushchin@linux.dev>
Acked-by: Shakeel Butt <shakeel.butt@linux.dev>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
---
 mm/memcontrol.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 98f84a9fa228..b10e0a8f3375 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -2691,6 +2691,20 @@ void __memcg_kmem_uncharge_page(struct page *page, int order)
 	obj_cgroup_put(objcg);
 }
 
+/* Replace the stock objcg with objcg, return the old objcg */
+static struct obj_cgroup *replace_stock_objcg(struct memcg_stock_pcp *stock,
+					     struct obj_cgroup *objcg)
+{
+	struct obj_cgroup *old = NULL;
+
+	old = drain_obj_stock(stock);
+	obj_cgroup_get(objcg);
+	stock->nr_bytes = atomic_read(&objcg->nr_charged_bytes)
+			? atomic_xchg(&objcg->nr_charged_bytes, 0) : 0;
+	WRITE_ONCE(stock->cached_objcg, objcg);
+	return old;
+}
+
 static void mod_objcg_state(struct obj_cgroup *objcg, struct pglist_data *pgdat,
 		     enum node_stat_item idx, int nr)
 {
@@ -2708,11 +2722,7 @@ static void mod_objcg_state(struct obj_cgroup *objcg, struct pglist_data *pgdat,
 	 * changes.
 	 */
 	if (READ_ONCE(stock->cached_objcg) != objcg) {
-		old = drain_obj_stock(stock);
-		obj_cgroup_get(objcg);
-		stock->nr_bytes = atomic_read(&objcg->nr_charged_bytes)
-				? atomic_xchg(&objcg->nr_charged_bytes, 0) : 0;
-		WRITE_ONCE(stock->cached_objcg, objcg);
+		old = replace_stock_objcg(stock, objcg);
 		stock->cached_pgdat = pgdat;
 	} else if (stock->cached_pgdat != pgdat) {
 		/* Flush the existing cached vmstat data */
@@ -2866,11 +2876,7 @@ static void refill_obj_stock(struct obj_cgroup *objcg, unsigned int nr_bytes,
 
 	stock = this_cpu_ptr(&memcg_stock);
 	if (READ_ONCE(stock->cached_objcg) != objcg) { /* reset if necessary */
-		old = drain_obj_stock(stock);
-		obj_cgroup_get(objcg);
-		WRITE_ONCE(stock->cached_objcg, objcg);
-		stock->nr_bytes = atomic_read(&objcg->nr_charged_bytes)
-				? atomic_xchg(&objcg->nr_charged_bytes, 0) : 0;
+		old = replace_stock_objcg(stock, objcg);
 		allow_uncharge = true;	/* Allow uncharge when objcg changes */
 	}
 	stock->nr_bytes += nr_bytes;
-- 
2.34.1

[PATCH -v4 next 4/4] memcg: add CONFIG_MEMCG_V1 for 'local' functions

[Chen Ridong]

From: Chen Ridong <chenridong@huawei.com>

Add CONFIG_MEMCG_V1 for the 'local' functions, which are only used in
memcg v1, so that they won't be built for v2.

Signed-off-by: Chen Ridong <chenridong@huawei.com>
---
 mm/memcontrol-v1.h | 6 +++---
 mm/memcontrol.c    | 6 ++++++
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/mm/memcontrol-v1.h b/mm/memcontrol-v1.h
index 144d71b65907..ecff454373e2 100644
--- a/mm/memcontrol-v1.h
+++ b/mm/memcontrol-v1.h
@@ -60,15 +60,15 @@ unsigned long mem_cgroup_usage(struct mem_cgroup *memcg, bool swap);
 void drain_all_stock(struct mem_cgroup *root_memcg);
 
 unsigned long memcg_events(struct mem_cgroup *memcg, int event);
-unsigned long memcg_events_local(struct mem_cgroup *memcg, int event);
-unsigned long memcg_page_state_local(struct mem_cgroup *memcg, int idx);
 unsigned long memcg_page_state_output(struct mem_cgroup *memcg, int item);
-unsigned long memcg_page_state_local_output(struct mem_cgroup *memcg, int item);
 int memory_stat_show(struct seq_file *m, void *v);
 
 /* Cgroup v1-specific declarations */
 #ifdef CONFIG_MEMCG_V1
 
+unsigned long memcg_events_local(struct mem_cgroup *memcg, int event);
+unsigned long memcg_page_state_local(struct mem_cgroup *memcg, int idx);
+unsigned long memcg_page_state_local_output(struct mem_cgroup *memcg, int item);
 bool memcg1_alloc_events(struct mem_cgroup *memcg);
 void memcg1_free_events(struct mem_cgroup *memcg);
 
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index b10e0a8f3375..daa74af15b05 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -706,6 +706,7 @@ void __mod_memcg_state(struct mem_cgroup *memcg, enum memcg_stat_item idx,
 	trace_mod_memcg_state(memcg, idx, val);
 }
 
+#ifdef CONFIG_MEMCG_V1
 /* idx can be of type enum memcg_stat_item or node_stat_item. */
 unsigned long memcg_page_state_local(struct mem_cgroup *memcg, int idx)
 {
@@ -722,6 +723,7 @@ unsigned long memcg_page_state_local(struct mem_cgroup *memcg, int idx)
 #endif
 	return x;
 }
+#endif
 
 static void __mod_memcg_lruvec_state(struct lruvec *lruvec,
 				     enum node_stat_item idx,
@@ -869,6 +871,7 @@ unsigned long memcg_events(struct mem_cgroup *memcg, int event)
 	return READ_ONCE(memcg->vmstats->events[i]);
 }
 
+#ifdef CONFIG_MEMCG_V1
 unsigned long memcg_events_local(struct mem_cgroup *memcg, int event)
 {
 	int i = memcg_events_index(event);
@@ -878,6 +881,7 @@ unsigned long memcg_events_local(struct mem_cgroup *memcg, int event)
 
 	return READ_ONCE(memcg->vmstats->events_local[i]);
 }
+#endif
 
 struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p)
 {
@@ -1447,11 +1451,13 @@ unsigned long memcg_page_state_output(struct mem_cgroup *memcg, int item)
 		memcg_page_state_output_unit(item);
 }
 
+#ifdef CONFIG_MEMCG_V1
 unsigned long memcg_page_state_local_output(struct mem_cgroup *memcg, int item)
 {
 	return memcg_page_state_local(memcg, item) *
 		memcg_page_state_output_unit(item);
 }
+#endif
 
 #ifdef CONFIG_HUGETLB_PAGE
 static bool memcg_accounts_hugetlb(void)
-- 
2.34.1

Re: [PATCH -v4 next 4/4] memcg: add CONFIG_MEMCG_V1 for 'local' functions

[Johannes Weiner]

On Fri, Jan 24, 2025 at 07:35:14AM +0000, Chen Ridong wrote:
> From: Chen Ridong <chenridong@huawei.com>
> 
> Add CONFIG_MEMCG_V1 for the 'local' functions, which are only used in
> memcg v1, so that they won't be built for v2.
> 
> Signed-off-by: Chen Ridong <chenridong@huawei.com>

Acked-by: Johannes Weiner <hannes@cmpxchg.org>

Thanks!

Re: [PATCH -v4 next 4/4] memcg: add CONFIG_MEMCG_V1 for 'local' functions

[Shakeel Butt]

On Fri, Jan 24, 2025 at 07:35:14AM +0000, Chen Ridong wrote:
> From: Chen Ridong <chenridong@huawei.com>
> 
> Add CONFIG_MEMCG_V1 for the 'local' functions, which are only used in
> memcg v1, so that they won't be built for v2.
> 
> Signed-off-by: Chen Ridong <chenridong@huawei.com>

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

Re: [PATCH -v4 next 4/4] memcg: add CONFIG_MEMCG_V1 for 'local' functions

[Roman Gushchin]

On Fri, Jan 24, 2025 at 07:35:14AM +0000, Chen Ridong wrote:
> From: Chen Ridong <chenridong@huawei.com>
> 
> Add CONFIG_MEMCG_V1 for the 'local' functions, which are only used in
> memcg v1, so that they won't be built for v2.
> 
> Signed-off-by: Chen Ridong <chenridong@huawei.com>

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