Skip to contentSkip to conversation

mm/memcg: bail out early when !memcg in mem_cgroup_lruvec

PATCH In review Alex Shi 3 participants 5 messages Nov 27, 2020 – Nov 30, 2020 memcg

lore
Share & tools
lore

Alex Shi

author
#0 lore raw

PATCH 1/1 mm/memcg: bail out early when !memcg in mem_cgroup_lruvec

Sometime, we use NULL memcg in mem_cgroup_lruvec(memcg, pgdat) so we could get out early in the situation to avoid useless checking.

Also warning if both parameter are NULL.

Cc Andrew Morton <[email protected]>
Cc Johannes Weiner <[email protected]>
Cc Shakeel Butt <[email protected]>
Cc Roman Gushchin <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Stephen Rothwell <[email protected]>
Cc Alexander Duyck <[email protected]>
Cc Yafang Shao <[email protected]>
Cc Wei Yang <[email protected]>
---
 include/linux/memcontrol.h | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
include/linux/memcontrol.h +3 -4
index 3e6a1df3bdb9..4cdb110f84e0 100644
@@ -613,14 +613,13 @@ static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg,
 	struct mem_cgroup_per_node *mz;
 	struct lruvec *lruvec;
 
-	if (mem_cgroup_disabled()) {
+	VM_WARN_ON_ONCE(!memcg && !pgdat);
+
+	if (mem_cgroup_disabled() || !memcg) {
 		lruvec = &pgdat->__lruvec;
 		goto out;
 	}
 
-	if (!memcg)
-		memcg = root_mem_cgroup;
-
 	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
 	lruvec = &mz->lruvec;
 out:
Signature · 1 line
2.29.GIT
#1 lore raw
Quoted 4 lines · On Fri, 27 Nov 2020 11:08:35 +0800 Alex Shi <[email protected]> wrote:

Sometime, we use NULL memcg in mem_cgroup_lruvec(memcg, pgdat) so we could get out early in the situation to avoid useless checking.

Also warning if both parameter are NULL.

Why do you think a warning is needed here?

Quoted 7 lines
include/linux/memcontrol.h +0 -1
@@ -613,14 +613,13 @@ static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg,
 	struct mem_cgroup_per_node *mz;
 	struct lruvec *lruvec;
 
-	if (mem_cgroup_disabled()) {

include/linux/memcontrol.h static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg, · line 623

Quoted 10 lines
include/linux/memcontrol.h +3 -3
+	VM_WARN_ON_ONCE(!memcg && !pgdat);
+
+	if (mem_cgroup_disabled() || !memcg) {
 		lruvec = &pgdat->__lruvec;
 		goto out;
 	}
 
-	if (!memcg)
-		memcg = root_mem_cgroup;
-

This change isn't obviously equivalent, is it?

include/linux/memcontrol.h static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg, · line 625

Quoted 3 lines
include/linux/memcontrol.h +0 -0
 	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
 	lruvec = &mz->lruvec;
 out:

And the resulting code is awkward:

	if (mem_cgroup_disabled() || !memcg) {
		lruvec = &pgdat->__lruvec;
		goto out;
	}

	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
	lruvec = &mz->lruvec;

out:

could be

	if (mem_cgroup_disabled() || !memcg) {
		lruvec = &pgdat->__lruvec;
	} else {
		mem_cgroup_per_node mz;

		mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
		lruvec = &mz->lruvec;
	}
author
#2 lore raw
Quoted 8 lines · Andrew Morton
Quoted 4 lines · On Fri, 27 Nov 2020 11:08:35 +0800 Alex Shi <[email protected]> wrote:

Sometime, we use NULL memcg in mem_cgroup_lruvec(memcg, pgdat) so we could get out early in the situation to avoid useless checking.

Also warning if both parameter are NULL.

Why do you think a warning is needed here?

Uh, Consider there are no problem for long time, it could be saved.

Quoted 15 lines
Quoted 15 lines
include/linux/memcontrol.h +3 -1
@@ -613,14 +613,13 @@ static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg,
 	struct mem_cgroup_per_node *mz;
 	struct lruvec *lruvec;
 
-	if (mem_cgroup_disabled()) {
+	VM_WARN_ON_ONCE(!memcg && !pgdat);
+
+	if (mem_cgroup_disabled() || !memcg) {
 		lruvec = &pgdat->__lruvec;
 		goto out;
 	}
 
Quoted 5 lines

include/linux/memcontrol.h static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg, · line 623

Quoted 3 lines
include/linux/memcontrol.h +0 -3
-	if (!memcg)
-		memcg = root_mem_cgroup;
-

This change isn't obviously equivalent, is it?

If !memcg, the root_mem_cgroup will still lead the lruvec to a pgdat same as parameter.

Quoted 19 lines

include/linux/memcontrol.h static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg, · line 625

Quoted 4 lines
include/linux/memcontrol.h +0 -0
 	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
 	lruvec = &mz->lruvec;
 out:

And the resulting code is awkward:

	if (mem_cgroup_disabled() || !memcg) {
		lruvec = &pgdat->__lruvec;
		goto out;
	}

	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
	lruvec = &mz->lruvec;

out:

could be

Quoted 9 lines
	if (mem_cgroup_disabled() || !memcg) {
		lruvec = &pgdat->__lruvec;
	} else {
		mem_cgroup_per_node mz;

		mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
		lruvec = &mz->lruvec;
	}

Right. remove 'goto' is better for understander.

So, is the following patch ok?

From 225f29e03b40a7cbaeb4e3bb76f8efbcd7d648a2 Mon Sep 17 00:00:00 2001
From: Alex Shi <[email protected]>
Date: Wed, 25 Nov 2020 14:06:33 +0800
Subject: [PATCH v2] mm/memcg: bail out early when !memcg in mem_cgroup_lruvec

Sometime, we use NULL memcg in mem_cgroup_lruvec(memcg, pgdat) so we could get out early in the situation to avoid useless checking.

Polished as Andrew Morton's suggestion.

Cc Andrew Morton <[email protected]>
Cc Johannes Weiner <[email protected]>
Cc Shakeel Butt <[email protected]>
Cc Roman Gushchin <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Stephen Rothwell <[email protected]>
Cc Alexander Duyck <[email protected]>
Cc Yafang Shao <[email protected]>
Cc Wei Yang <[email protected]>
---
 include/linux/memcontrol.h | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)
include/linux/memcontrol.h +6 -9
index 3e6a1df3bdb9..4ff2ffe2b73d 100644
@@ -610,20 +610,17 @@ mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid)
 static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg,
 					       struct pglist_data *pgdat)
 {
-	struct mem_cgroup_per_node *mz;
 	struct lruvec *lruvec;
 
-	if (mem_cgroup_disabled()) {
+	if (mem_cgroup_disabled() || !memcg) {
 		lruvec = &pgdat->__lruvec;
-		goto out;
-	}
+	} else {
+		struct mem_cgroup_per_node *mz;
 
-	if (!memcg)
-		memcg = root_mem_cgroup;
+		mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
+		lruvec = &mz->lruvec;
+	}
 
-	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
-	lruvec = &mz->lruvec;
-out:
 	/*
 	 * Since a node can be onlined after the mem_cgroup was created,
 	 * we have to be prepared to initialize lruvec->pgdat here;
Signature · 1 line
2.29.GIT
#3 lore raw

29.11.2020 07:43, Alex Shi пишет:

Quoted 118 lines · Andrew Morton
Quoted 8 lines · On Fri, 27 Nov 2020 11:08:35 +0800 Alex Shi <[email protected]> wrote:
Quoted 5 lines

Sometime, we use NULL memcg in mem_cgroup_lruvec(memcg, pgdat) so we could get out early in the situation to avoid useless checking.

Also warning if both parameter are NULL.

Why do you think a warning is needed here?

Uh, Consider there are no problem for long time, it could be saved.

Quoted 15 lines
Quoted 15 lines
include/linux/memcontrol.h +3 -1
@@ -613,14 +613,13 @@ static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg,
 	struct mem_cgroup_per_node *mz;
 	struct lruvec *lruvec;
 
-	if (mem_cgroup_disabled()) {
+	VM_WARN_ON_ONCE(!memcg && !pgdat);
+
+	if (mem_cgroup_disabled() || !memcg) {
 		lruvec = &pgdat->__lruvec;
 		goto out;
 	}
 
Quoted 5 lines

include/linux/memcontrol.h mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid)

Quoted 3 lines
include/linux/memcontrol.h +0 -3
-	if (!memcg)
-		memcg = root_mem_cgroup;
-

This change isn't obviously equivalent, is it?

If !memcg, the root_mem_cgroup will still lead the lruvec to a pgdat same as parameter.

Quoted 19 lines

include/linux/memcontrol.h static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg, · line 625

Quoted 4 lines
include/linux/memcontrol.h +0 -0
 	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
 	lruvec = &mz->lruvec;
 out:

And the resulting code is awkward:

	if (mem_cgroup_disabled() || !memcg) {
		lruvec = &pgdat->__lruvec;
		goto out;
	}

	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
	lruvec = &mz->lruvec;

out:

could be

Quoted 9 lines
	if (mem_cgroup_disabled() || !memcg) {
		lruvec = &pgdat->__lruvec;
	} else {
		mem_cgroup_per_node mz;

		mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
		lruvec = &mz->lruvec;
	}

Right. remove 'goto' is better for understander.

So, is the following patch ok?

From 225f29e03b40a7cbaeb4e3bb76f8efbcd7d648a2 Mon Sep 17 00:00:00 2001
From: Alex Shi <[email protected]>
Date: Wed, 25 Nov 2020 14:06:33 +0800
Subject: [PATCH v2] mm/memcg: bail out early when !memcg in mem_cgroup_lruvec

Sometime, we use NULL memcg in mem_cgroup_lruvec(memcg, pgdat) so we could get out early in the situation to avoid useless checking.

Polished as Andrew Morton's suggestion.

Cc Andrew Morton <[email protected]>
Cc Johannes Weiner <[email protected]>
Cc Shakeel Butt <[email protected]>
Cc Roman Gushchin <[email protected]>
Cc Lorenzo Stoakes <[email protected]>
Cc Stephen Rothwell <[email protected]>
Cc Alexander Duyck <[email protected]>
Cc Yafang Shao <[email protected]>
Cc Wei Yang <[email protected]>
---
 include/linux/memcontrol.h | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)
include/linux/memcontrol.h +3 -6
index 3e6a1df3bdb9..4ff2ffe2b73d 100644
@@ -610,20 +610,17 @@ mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid)
 static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg,
 					       struct pglist_data *pgdat)
 {
-	struct mem_cgroup_per_node *mz;
 	struct lruvec *lruvec;
 
-	if (mem_cgroup_disabled()) {
+	if (mem_cgroup_disabled() || !memcg) {
 		lruvec = &pgdat->__lruvec;
-		goto out;
-	}
+	} else {
+		struct mem_cgroup_per_node *mz;
 
-	if (!memcg)
-		memcg = root_mem_cgroup;

include/linux/memcontrol.h mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid) · line 626

Quoted 10 lines
include/linux/memcontrol.h +3 -3
+		mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
+		lruvec = &mz->lruvec;
+	}
 
-	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
-	lruvec = &mz->lruvec;
-out:
 	/*
 	 * Since a node can be onlined after the mem_cgroup was created,
 	 * we have to be prepared to initialize lruvec->pgdat here;

Hi,

This patch causes a hard lock on one of my ARM32 devices using today's linux-next, please fix.

#4 lore raw
Quoted 42 lines · On Mon, 30 Nov 2020 22:44:11 +0300 Dmitry Osipenko <[email protected]> wrote:
Quoted 29 lines

From: Alex Shi <[email protected]>
Date: Wed, 25 Nov 2020 14:06:33 +0800
Subject: [PATCH v2] mm/memcg: bail out early when !memcg in mem_cgroup_lruvec

Sometime, we use NULL memcg in mem_cgroup_lruvec(memcg, pgdat) so we could get out early in the situation to avoid useless checking.

Polished as Andrew Morton's suggestion.

include/linux/memcontrol.h +3 -6
@@ -610,20 +610,17 @@ mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid)
 static inline struct lruvec *mem_cgroup_lruvec(struct mem_cgroup *memcg,
 					       struct pglist_data *pgdat)
 {
-	struct mem_cgroup_per_node *mz;
 	struct lruvec *lruvec;
 
-	if (mem_cgroup_disabled()) {
+	if (mem_cgroup_disabled() || !memcg) {
 		lruvec = &pgdat->__lruvec;
-		goto out;
-	}
+	} else {
+		struct mem_cgroup_per_node *mz;
 
-	if (!memcg)
-		memcg = root_mem_cgroup;

include/linux/memcontrol.h mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid) · line 626

Quoted 10 lines
include/linux/memcontrol.h +3 -3
+		mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
+		lruvec = &mz->lruvec;
+	}
 
-	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
-	lruvec = &mz->lruvec;
-out:
 	/*
 	 * Since a node can be onlined after the mem_cgroup was created,
 	 * we have to be prepared to initialize lruvec->pgdat here;

Hi,

Quoted 2 lines

This patch causes a hard lock on one of my ARM32 devices using today's linux-next, please fix.

Thanks.

This is unexpected. I assume you've confirmed that reverting this change from linux-next fixes things?