mm/list_lru: Optimize memcg_reparent_list_lru_node()
Subsystems (2)
PATCH v2 1/1 mm/list_lru: Optimize memcg_reparent_list_lru_node()
Since commit 2c80cd57c743 ("mm/list_lru.c: fix list_lru_count_node() to be race free"), we are tracking the total number of lru entries in a list_lru_node in its nr_items field. In the case of memcg_reparent_list_lru_node(), there is nothing to be done if nr_items is 0. We don't even need to take the nlru->lock as no new lru entry could be added by a racing list_lru_add() to the draining src_idx memcg at this point.
--- mm/list_lru.c | 6 ++++++ 1 file changed, 6 insertions(+)
mm/list_lru.c +6 -0
@@ -394,6 +394,12 @@ static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid,int dst_idx = dst_memcg->kmemcg_id;struct list_lru_one *src, *dst; +/* +* If there is no lru entry in this nlru, we can skip it immediately. +*/ +if (!READ_ONCE (nlru->nr_items)) +return ; +/* * Since list_lru_{add,del} may be called under an IRQ-safe lock, * we have to use IRQ-safe primitives here to avoid deadlock.
Signature · 1 line
2.27.0
Roman Gushchin
reviewerQuoted 18 lines · Waiman Long
Since commit
2c80cd57c743("mm/list_lru.c: fixlist_lru_count_node()to be race free"), we are tracking the total number of lru entries in alist_lru_nodein itsnr_itemsfield. In the case ofmemcg_reparent_list_lru_node(), there is nothing to be done ifnr_itemsis 0. We don't even need to take thenlru->lockas no new lru entry could be added by a racinglist_lru_add()to the drainingsrc_idxmemcg at this point.Signed-off-by Waiman Long <[email protected]>--- mm/list_lru.c | 6 ++++++ 1 file changed, 6 insertions(+)mm/list_lru.c +0
-0
mm/list_lru.c static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid, · line 401
Quoted 8 lines
mm/list_lru.c +5
-0@@ -394,6 +394,12 @@ static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid,int dst_idx = dst_memcg->kmemcg_id;struct list_lru_one *src, *dst; +/* +* If there is no lru entry in this nlru, we can skip it immediately. +*/ +if (!READ_ONCE (nlru->nr_items)) +return ;
This is a per-node counter, not a per-memcg, right?
If so, do we optimize for the case when all lru items belong to one node and others are empty?
Quoted 22 lines · Roman Gushchin
Quoted 21 lines · Waiman Long
Since commit
2c80cd57c743("mm/list_lru.c: fixlist_lru_count_node()to be race free"), we are tracking the total number of lru entries in alist_lru_nodein itsnr_itemsfield. In the case ofmemcg_reparent_list_lru_node(), there is nothing to be done ifnr_itemsis 0. We don't even need to take thenlru->lockas no new lru entry could be added by a racinglist_lru_add()to the drainingsrc_idxmemcg at this point.Signed-off-by Waiman Long <[email protected]>--- mm/list_lru.c | 6 ++++++ 1 file changed, 6 insertions(+)mm/list_lru.c +0
-0@@ -394,6 +394,12 @@ static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid,int dst_idx = dst_memcg->kmemcg_id;struct list_lru_one *src, *dst;
Quoted 6 lines
mm/list_lru.c static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid, · line 401
Quoted 5 lines
mm/list_lru.c +5
-0+/* +* If there is no lru entry in this nlru, we can skip it immediately. +*/ +if (!READ_ONCE (nlru->nr_items)) +return ;This is a per-node counter, not a per-memcg, right?
Right. list_lru_node is a per-node structure inside list_lru.
Quoted 2 lines
If so, do we optimize for the case when all lru items belong to one node and others are empty?
That is actually the case that I am trying to optimize for.
If a system has many containers. It is also likely each container may mount one or more container specific filesystems. Since a container likely use just a few cpus, it is highly that only the list_lru_node that contains those cpus will be utilized while the rests may be empty.
I got the idea of doing this patch when I was looking at a crash dump related to the list_lru code. That particular crash dump has more than 13k list_lru's and thousands of mount points. I had notice even if nr_items of a list_lru_node is 0, it still tries to transfer lru entries from source idx to dest idx. Without doing an lock/unlock and loading a cacheline from the memcg_lrus, it can save some time. That can be substantial saving if we are talking about thousands of list_lru's.
Cheers,
Longman
Quoted 39 lines · Waiman Long
Quoted 14 lines · Roman Gushchin
Quoted 13 lines · Waiman Long
Since commit
2c80cd57c743("mm/list_lru.c: fixlist_lru_count_node()to be race free"), we are tracking the total number of lru entries in alist_lru_nodein itsnr_itemsfield. In the case ofmemcg_reparent_list_lru_node(), there is nothing to be done ifnr_itemsis 0. We don't even need to take thenlru->lockas no new lru entry could be added by a racinglist_lru_add()to the drainingsrc_idxmemcg at this point.Signed-off-by Waiman Long <[email protected]>--- mm/list_lru.c | 6 ++++++ 1 file changed, 6 insertions(+)Quoted 13 lines
mm/list_lru.c static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid, · line 400
Quoted 12 lines
mm/list_lru.c +5
-0@@ -394,6 +394,12 @@ static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid,int dst_idx = dst_memcg->kmemcg_id;struct list_lru_one *src, *dst; +/* +* If there is no lru entry in this nlru, we can skip it immediately. +*/ +if (!READ_ONCE (nlru->nr_items)) +return ;This is a per-node counter, not a per-memcg, right?
Right.
list_lru_nodeis a per-node structure insidelist_lru.Quoted 2 lines
If so, do we optimize for the case when all lru items belong to one node and others are empty?
That is actually the case that I am trying to optimize for.
If a system has many containers. It is also likely each container may mount one or more container specific filesystems. Since a container likely use just a few cpus, it is highly that only the
list_lru_nodethat contains those cpus will be utilized while the rests may be empty.
Quoted 7 lines
I got the idea of doing this patch when I was looking at a crash dump related to the
list_lrucode. That particular crash dump has more than 13k list_lru's and thousands of mount points. I had notice even ifnr_itemsof alist_lru_nodeis 0, it still tries to transfer lru entries from source idx to dest idx. Without doing an lock/unlock and loading a cacheline from thememcg_lrus, it can save some time. That can be substantial saving if we are talking about thousands of list_lru's.
Cool! Makes total sense to me. Thanks for the explanation! Would you mind to add this text to the commit log?
Thanks!
Quoted 46 lines · Roman Gushchin
Quoted 38 lines · Waiman Long
Quoted 14 lines · Roman Gushchin
Quoted 13 lines · Waiman Long
Since commit
2c80cd57c743("mm/list_lru.c: fixlist_lru_count_node()to be race free"), we are tracking the total number of lru entries in alist_lru_nodein itsnr_itemsfield. In the case ofmemcg_reparent_list_lru_node(), there is nothing to be done ifnr_itemsis 0. We don't even need to take thenlru->lockas no new lru entry could be added by a racinglist_lru_add()to the drainingsrc_idxmemcg at this point.Signed-off-by Waiman Long <[email protected]>--- mm/list_lru.c | 6 ++++++ 1 file changed, 6 insertions(+)Quoted 13 lines
mm/list_lru.c static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid, · line 400
Quoted 12 lines
mm/list_lru.c +5
-0@@ -394,6 +394,12 @@ static void memcg_reparent_list_lru_node(struct list_lru *lru, int nid,int dst_idx = dst_memcg->kmemcg_id;struct list_lru_one *src, *dst; +/* +* If there is no lru entry in this nlru, we can skip it immediately. +*/ +if (!READ_ONCE (nlru->nr_items)) +return ;This is a per-node counter, not a per-memcg, right?
Right.
list_lru_nodeis a per-node structure insidelist_lru.Quoted 2 lines
If so, do we optimize for the case when all lru items belong to one node and others are empty?
That is actually the case that I am trying to optimize for.
If a system has many containers. It is also likely each container may mount one or more container specific filesystems. Since a container likely use just a few cpus, it is highly that only the
list_lru_nodethat contains those cpus will be utilized while the rests may be empty.Quoted 7 lines
I got the idea of doing this patch when I was looking at a crash dump related to the
list_lrucode. That particular crash dump has more than 13k list_lru's and thousands of mount points. I had notice even ifnr_itemsof alist_lru_nodeis 0, it still tries to transfer lru entries from source idx to dest idx. Without doing an lock/unlock and loading a cacheline from thememcg_lrus, it can save some time. That can be substantial saving if we are talking about thousands of list_lru's.
Quoted 4 lines
Cool! Makes total sense to me. Thanks for the explanation! Would you mind to add this text to the commit log?
Reviewed-by Roman Gushchin <[email protected]>
Sure. I will add some of it into the commit log.
Cheers,
Longman