[PATCH v3 0/2] get rid of populate for memcg

[PATCH v3 0/2] get rid of populate for memcg

[Glauber Costa]

[ v3: fix bug with failed initialization, seen by Tejun ]

Hi Tejun,

Let me know if the following is acceptable. Turns out the only
operation with a side effect we do after the last failure spot,
is a get on the parent memcg, and the setup of our very own refcnt.

Both of them can be dealt with a mem_cgroup_put in the beginning
of the failure path. So I felt no need for any code reordering.

People from the memcg side, please verify if this is indeed
okay.

I actually tested it with some test code to force memcg_kmem_init()
to fail, and it seems to go all right.

I'll be happy to address any other problems you spot.

Glauber Costa (2):
  cgroup: pass struct mem_cgroup instead of struct cgroup to socket
    memcg
  cgroup: get rid of populate for memcg

 include/net/sock.h           |   12 ++++++------
 include/net/tcp_memcontrol.h |    4 ++--
 mm/memcontrol.c              |   39 ++++++++++++++++++---------------------
 net/core/sock.c              |   10 +++++-----
 net/ipv4/tcp_memcontrol.c    |    6 ++----
 5 files changed, 33 insertions(+), 38 deletions(-)

-- 
1.7.7.6

[PATCH v3 0/2] get rid of populate for memcg

[Glauber Costa]

[ v3: fix bug with failed initialization, seen by Tejun ]

Hi Tejun,

Let me know if the following is acceptable. Turns out the only
operation with a side effect we do after the last failure spot,
is a get on the parent memcg, and the setup of our very own refcnt.

Both of them can be dealt with a mem_cgroup_put in the beginning
of the failure path. So I felt no need for any code reordering.

People from the memcg side, please verify if this is indeed
okay.

I actually tested it with some test code to force memcg_kmem_init()
to fail, and it seems to go all right.

I'll be happy to address any other problems you spot.

Glauber Costa (2):
  cgroup: pass struct mem_cgroup instead of struct cgroup to socket
    memcg
  cgroup: get rid of populate for memcg

 include/net/sock.h           |   12 ++++++------
 include/net/tcp_memcontrol.h |    4 ++--
 mm/memcontrol.c              |   39 ++++++++++++++++++---------------------
 net/core/sock.c              |   10 +++++-----
 net/ipv4/tcp_memcontrol.c    |    6 ++----
 5 files changed, 33 insertions(+), 38 deletions(-)

-- 
1.7.7.6

[PATCH v3 1/2] cgroup: pass struct mem_cgroup instead of struct cgroup to socket memcg

[Glauber Costa]

The only reason cgroup was used, was to be consistent with the populate()
interface. Now that we're getting rid of it, not only we no longer need
it, but we also *can't* call it this way.

Since we will no longer rely on populate(), this will be called from
create(). During create, the association between struct mem_cgroup
and struct cgroup does not yet exist, since cgroup internals hasn't
yet initialized its bookkeeping. This means we would not be able
to draw the memcg pointer from the cgroup pointer in these
functions, which is highly undesirable.

Signed-off-by: Glauber Costa <glommer-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org>
CC: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
CC: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
CC: Kamezawa Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>
CC: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
CC: Michal Hocko <mhocko-AlSwsSmVLrQ@public.gmane.org>
---
 include/net/sock.h           |   12 ++++++------
 include/net/tcp_memcontrol.h |    4 ++--
 mm/memcontrol.c              |   24 +++++++++---------------
 net/core/sock.c              |   10 +++++-----
 net/ipv4/tcp_memcontrol.c    |    6 ++----
 5 files changed, 24 insertions(+), 32 deletions(-)

diff --git a/include/net/sock.h b/include/net/sock.h
index a6ba1f8..b3ebe6b 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -70,16 +70,16 @@
 struct cgroup;
 struct cgroup_subsys;
 #ifdef CONFIG_NET
-int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss);
-void mem_cgroup_sockets_destroy(struct cgroup *cgrp);
+int mem_cgroup_sockets_init(struct mem_cgroup *memcg, struct cgroup_subsys *ss);
+void mem_cgroup_sockets_destroy(struct mem_cgroup *memcg);
 #else
 static inline
-int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
+int mem_cgroup_sockets_init(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	return 0;
 }
 static inline
-void mem_cgroup_sockets_destroy(struct cgroup *cgrp)
+void mem_cgroup_sockets_destroy(struct mem_cgroup *memcg)
 {
 }
 #endif
@@ -900,9 +900,9 @@ struct proto {
 	 * This function has to setup any files the protocol want to
 	 * appear in the kmem cgroup filesystem.
 	 */
-	int			(*init_cgroup)(struct cgroup *cgrp,
+	int			(*init_cgroup)(struct mem_cgroup *memcg,
 					       struct cgroup_subsys *ss);
-	void			(*destroy_cgroup)(struct cgroup *cgrp);
+	void			(*destroy_cgroup)(struct mem_cgroup *memcg);
 	struct cg_proto		*(*proto_cgroup)(struct mem_cgroup *memcg);
 #endif
 };
diff --git a/include/net/tcp_memcontrol.h b/include/net/tcp_memcontrol.h
index 48410ff..7df18bc 100644
--- a/include/net/tcp_memcontrol.h
+++ b/include/net/tcp_memcontrol.h
@@ -12,8 +12,8 @@ struct tcp_memcontrol {
 };
 
 struct cg_proto *tcp_proto_cgroup(struct mem_cgroup *memcg);
-int tcp_init_cgroup(struct cgroup *cgrp, struct cgroup_subsys *ss);
-void tcp_destroy_cgroup(struct cgroup *cgrp);
+int tcp_init_cgroup(struct mem_cgroup *memcg, struct cgroup_subsys *ss);
+void tcp_destroy_cgroup(struct mem_cgroup *memcg);
 unsigned long long tcp_max_memory(const struct mem_cgroup *memcg);
 void tcp_prot_mem(struct mem_cgroup *memcg, long val, int idx);
 #endif /* _TCP_MEMCG_H */
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index bef1142..704054d 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4640,29 +4640,22 @@ static int mem_control_numa_stat_open(struct inode *unused, struct file *file)
 #endif /* CONFIG_NUMA */
 
 #ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
-static int register_kmem_files(struct cgroup *cont, struct cgroup_subsys *ss)
+static int register_kmem_files(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
-	/*
-	 * Part of this would be better living in a separate allocation
-	 * function, leaving us with just the cgroup tree population work.
-	 * We, however, depend on state such as network's proto_list that
-	 * is only initialized after cgroup creation. I found the less
-	 * cumbersome way to deal with it to defer it all to populate time
-	 */
-	return mem_cgroup_sockets_init(cont, ss);
+	return mem_cgroup_sockets_init(memcg, ss);
 };
 
-static void kmem_cgroup_destroy(struct cgroup *cont)
+static void kmem_cgroup_destroy(struct mem_cgroup *memcg)
 {
-	mem_cgroup_sockets_destroy(cont);
+	mem_cgroup_sockets_destroy(memcg);
 }
 #else
-static int register_kmem_files(struct cgroup *cont, struct cgroup_subsys *ss)
+static int register_kmem_files(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	return 0;
 }
 
-static void kmem_cgroup_destroy(struct cgroup *cont)
+static void kmem_cgroup_destroy(struct mem_cgroup *memcg)
 {
 }
 #endif
@@ -5034,7 +5027,7 @@ static void mem_cgroup_destroy(struct cgroup *cont)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
 
-	kmem_cgroup_destroy(cont);
+	kmem_cgroup_destroy(memcg);
 
 	mem_cgroup_put(memcg);
 }
@@ -5042,7 +5035,8 @@ static void mem_cgroup_destroy(struct cgroup *cont)
 static int mem_cgroup_populate(struct cgroup_subsys *ss,
 				struct cgroup *cont)
 {
-	return register_kmem_files(cont, ss);
+	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
+	return register_kmem_files(memcg, ss);
 }
 
 #ifdef CONFIG_MMU
diff --git a/net/core/sock.c b/net/core/sock.c
index b2e14c0..878f744 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -140,7 +140,7 @@ static DEFINE_MUTEX(proto_list_mutex);
 static LIST_HEAD(proto_list);
 
 #ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
-int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
+int mem_cgroup_sockets_init(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	struct proto *proto;
 	int ret = 0;
@@ -148,7 +148,7 @@ int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
 	mutex_lock(&proto_list_mutex);
 	list_for_each_entry(proto, &proto_list, node) {
 		if (proto->init_cgroup) {
-			ret = proto->init_cgroup(cgrp, ss);
+			ret = proto->init_cgroup(memcg, ss);
 			if (ret)
 				goto out;
 		}
@@ -159,19 +159,19 @@ int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
 out:
 	list_for_each_entry_continue_reverse(proto, &proto_list, node)
 		if (proto->destroy_cgroup)
-			proto->destroy_cgroup(cgrp);
+			proto->destroy_cgroup(memcg);
 	mutex_unlock(&proto_list_mutex);
 	return ret;
 }
 
-void mem_cgroup_sockets_destroy(struct cgroup *cgrp)
+void mem_cgroup_sockets_destroy(struct mem_cgroup *memcg)
 {
 	struct proto *proto;
 
 	mutex_lock(&proto_list_mutex);
 	list_for_each_entry_reverse(proto, &proto_list, node)
 		if (proto->destroy_cgroup)
-			proto->destroy_cgroup(cgrp);
+			proto->destroy_cgroup(memcg);
 	mutex_unlock(&proto_list_mutex);
 }
 #endif
diff --git a/net/ipv4/tcp_memcontrol.c b/net/ipv4/tcp_memcontrol.c
index 8f1753d..1517037 100644
--- a/net/ipv4/tcp_memcontrol.c
+++ b/net/ipv4/tcp_memcontrol.c
@@ -18,7 +18,7 @@ static void memcg_tcp_enter_memory_pressure(struct sock *sk)
 }
 EXPORT_SYMBOL(memcg_tcp_enter_memory_pressure);
 
-int tcp_init_cgroup(struct cgroup *cgrp, struct cgroup_subsys *ss)
+int tcp_init_cgroup(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	/*
 	 * The root cgroup does not use res_counters, but rather,
@@ -28,7 +28,6 @@ int tcp_init_cgroup(struct cgroup *cgrp, struct cgroup_subsys *ss)
 	struct res_counter *res_parent = NULL;
 	struct cg_proto *cg_proto, *parent_cg;
 	struct tcp_memcontrol *tcp;
-	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
 	struct mem_cgroup *parent = parent_mem_cgroup(memcg);
 	struct net *net = current->nsproxy->net_ns;
 
@@ -61,9 +60,8 @@ int tcp_init_cgroup(struct cgroup *cgrp, struct cgroup_subsys *ss)
 }
 EXPORT_SYMBOL(tcp_init_cgroup);
 
-void tcp_destroy_cgroup(struct cgroup *cgrp)
+void tcp_destroy_cgroup(struct mem_cgroup *memcg)
 {
-	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
 	struct cg_proto *cg_proto;
 	struct tcp_memcontrol *tcp;
 	u64 val;
-- 
1.7.7.6

[PATCH v3 1/2] cgroup: pass struct mem_cgroup instead of struct cgroup to socket memcg

[Glauber Costa]

The only reason cgroup was used, was to be consistent with the populate()
interface. Now that we're getting rid of it, not only we no longer need
it, but we also *can't* call it this way.

Since we will no longer rely on populate(), this will be called from
create(). During create, the association between struct mem_cgroup
and struct cgroup does not yet exist, since cgroup internals hasn't
yet initialized its bookkeeping. This means we would not be able
to draw the memcg pointer from the cgroup pointer in these
functions, which is highly undesirable.

Signed-off-by: Glauber Costa <glommer-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org>
CC: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
CC: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
CC: Kamezawa Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>
CC: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
CC: Michal Hocko <mhocko-AlSwsSmVLrQ@public.gmane.org>
---
 include/net/sock.h           |   12 ++++++------
 include/net/tcp_memcontrol.h |    4 ++--
 mm/memcontrol.c              |   24 +++++++++---------------
 net/core/sock.c              |   10 +++++-----
 net/ipv4/tcp_memcontrol.c    |    6 ++----
 5 files changed, 24 insertions(+), 32 deletions(-)

diff --git a/include/net/sock.h b/include/net/sock.h
index a6ba1f8..b3ebe6b 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -70,16 +70,16 @@
 struct cgroup;
 struct cgroup_subsys;
 #ifdef CONFIG_NET
-int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss);
-void mem_cgroup_sockets_destroy(struct cgroup *cgrp);
+int mem_cgroup_sockets_init(struct mem_cgroup *memcg, struct cgroup_subsys *ss);
+void mem_cgroup_sockets_destroy(struct mem_cgroup *memcg);
 #else
 static inline
-int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
+int mem_cgroup_sockets_init(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	return 0;
 }
 static inline
-void mem_cgroup_sockets_destroy(struct cgroup *cgrp)
+void mem_cgroup_sockets_destroy(struct mem_cgroup *memcg)
 {
 }
 #endif
@@ -900,9 +900,9 @@ struct proto {
 	 * This function has to setup any files the protocol want to
 	 * appear in the kmem cgroup filesystem.
 	 */
-	int			(*init_cgroup)(struct cgroup *cgrp,
+	int			(*init_cgroup)(struct mem_cgroup *memcg,
 					       struct cgroup_subsys *ss);
-	void			(*destroy_cgroup)(struct cgroup *cgrp);
+	void			(*destroy_cgroup)(struct mem_cgroup *memcg);
 	struct cg_proto		*(*proto_cgroup)(struct mem_cgroup *memcg);
 #endif
 };
diff --git a/include/net/tcp_memcontrol.h b/include/net/tcp_memcontrol.h
index 48410ff..7df18bc 100644
--- a/include/net/tcp_memcontrol.h
+++ b/include/net/tcp_memcontrol.h
@@ -12,8 +12,8 @@ struct tcp_memcontrol {
 };
 
 struct cg_proto *tcp_proto_cgroup(struct mem_cgroup *memcg);
-int tcp_init_cgroup(struct cgroup *cgrp, struct cgroup_subsys *ss);
-void tcp_destroy_cgroup(struct cgroup *cgrp);
+int tcp_init_cgroup(struct mem_cgroup *memcg, struct cgroup_subsys *ss);
+void tcp_destroy_cgroup(struct mem_cgroup *memcg);
 unsigned long long tcp_max_memory(const struct mem_cgroup *memcg);
 void tcp_prot_mem(struct mem_cgroup *memcg, long val, int idx);
 #endif /* _TCP_MEMCG_H */
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index bef1142..704054d 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4640,29 +4640,22 @@ static int mem_control_numa_stat_open(struct inode *unused, struct file *file)
 #endif /* CONFIG_NUMA */
 
 #ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
-static int register_kmem_files(struct cgroup *cont, struct cgroup_subsys *ss)
+static int register_kmem_files(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
-	/*
-	 * Part of this would be better living in a separate allocation
-	 * function, leaving us with just the cgroup tree population work.
-	 * We, however, depend on state such as network's proto_list that
-	 * is only initialized after cgroup creation. I found the less
-	 * cumbersome way to deal with it to defer it all to populate time
-	 */
-	return mem_cgroup_sockets_init(cont, ss);
+	return mem_cgroup_sockets_init(memcg, ss);
 };
 
-static void kmem_cgroup_destroy(struct cgroup *cont)
+static void kmem_cgroup_destroy(struct mem_cgroup *memcg)
 {
-	mem_cgroup_sockets_destroy(cont);
+	mem_cgroup_sockets_destroy(memcg);
 }
 #else
-static int register_kmem_files(struct cgroup *cont, struct cgroup_subsys *ss)
+static int register_kmem_files(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	return 0;
 }
 
-static void kmem_cgroup_destroy(struct cgroup *cont)
+static void kmem_cgroup_destroy(struct mem_cgroup *memcg)
 {
 }
 #endif
@@ -5034,7 +5027,7 @@ static void mem_cgroup_destroy(struct cgroup *cont)
 {
 	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
 
-	kmem_cgroup_destroy(cont);
+	kmem_cgroup_destroy(memcg);
 
 	mem_cgroup_put(memcg);
 }
@@ -5042,7 +5035,8 @@ static void mem_cgroup_destroy(struct cgroup *cont)
 static int mem_cgroup_populate(struct cgroup_subsys *ss,
 				struct cgroup *cont)
 {
-	return register_kmem_files(cont, ss);
+	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
+	return register_kmem_files(memcg, ss);
 }
 
 #ifdef CONFIG_MMU
diff --git a/net/core/sock.c b/net/core/sock.c
index b2e14c0..878f744 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -140,7 +140,7 @@ static DEFINE_MUTEX(proto_list_mutex);
 static LIST_HEAD(proto_list);
 
 #ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
-int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
+int mem_cgroup_sockets_init(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	struct proto *proto;
 	int ret = 0;
@@ -148,7 +148,7 @@ int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
 	mutex_lock(&proto_list_mutex);
 	list_for_each_entry(proto, &proto_list, node) {
 		if (proto->init_cgroup) {
-			ret = proto->init_cgroup(cgrp, ss);
+			ret = proto->init_cgroup(memcg, ss);
 			if (ret)
 				goto out;
 		}
@@ -159,19 +159,19 @@ int mem_cgroup_sockets_init(struct cgroup *cgrp, struct cgroup_subsys *ss)
 out:
 	list_for_each_entry_continue_reverse(proto, &proto_list, node)
 		if (proto->destroy_cgroup)
-			proto->destroy_cgroup(cgrp);
+			proto->destroy_cgroup(memcg);
 	mutex_unlock(&proto_list_mutex);
 	return ret;
 }
 
-void mem_cgroup_sockets_destroy(struct cgroup *cgrp)
+void mem_cgroup_sockets_destroy(struct mem_cgroup *memcg)
 {
 	struct proto *proto;
 
 	mutex_lock(&proto_list_mutex);
 	list_for_each_entry_reverse(proto, &proto_list, node)
 		if (proto->destroy_cgroup)
-			proto->destroy_cgroup(cgrp);
+			proto->destroy_cgroup(memcg);
 	mutex_unlock(&proto_list_mutex);
 }
 #endif
diff --git a/net/ipv4/tcp_memcontrol.c b/net/ipv4/tcp_memcontrol.c
index 8f1753d..1517037 100644
--- a/net/ipv4/tcp_memcontrol.c
+++ b/net/ipv4/tcp_memcontrol.c
@@ -18,7 +18,7 @@ static void memcg_tcp_enter_memory_pressure(struct sock *sk)
 }
 EXPORT_SYMBOL(memcg_tcp_enter_memory_pressure);
 
-int tcp_init_cgroup(struct cgroup *cgrp, struct cgroup_subsys *ss)
+int tcp_init_cgroup(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	/*
 	 * The root cgroup does not use res_counters, but rather,
@@ -28,7 +28,6 @@ int tcp_init_cgroup(struct cgroup *cgrp, struct cgroup_subsys *ss)
 	struct res_counter *res_parent = NULL;
 	struct cg_proto *cg_proto, *parent_cg;
 	struct tcp_memcontrol *tcp;
-	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
 	struct mem_cgroup *parent = parent_mem_cgroup(memcg);
 	struct net *net = current->nsproxy->net_ns;
 
@@ -61,9 +60,8 @@ int tcp_init_cgroup(struct cgroup *cgrp, struct cgroup_subsys *ss)
 }
 EXPORT_SYMBOL(tcp_init_cgroup);
 
-void tcp_destroy_cgroup(struct cgroup *cgrp)
+void tcp_destroy_cgroup(struct mem_cgroup *memcg)
 {
-	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
 	struct cg_proto *cg_proto;
 	struct tcp_memcontrol *tcp;
 	u64 val;
-- 
1.7.7.6

[PATCH v3 2/2] cgroup: get rid of populate for memcg

[Glauber Costa]

The last man standing justifying the need for populate() is the
sock memcg initialization functions. Now that we are able to pass
a struct mem_cgroup instead of a struct cgroup to the socket
initialization, there is nothing that stops us from initializing
everything in create().

Signed-off-by: Glauber Costa <glommer-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org>
CC: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
CC: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
CC: Kamezawa Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>
CC: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
CC: Michal Hocko <mhocko-AlSwsSmVLrQ@public.gmane.org>
---
 mm/memcontrol.c |   23 +++++++++++++----------
 1 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 704054d..02b01d2 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4640,7 +4640,7 @@ static int mem_control_numa_stat_open(struct inode *unused, struct file *file)
 #endif /* CONFIG_NUMA */
 
 #ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
-static int register_kmem_files(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
+static int memcg_init_kmem(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	return mem_cgroup_sockets_init(memcg, ss);
 };
@@ -4650,7 +4650,7 @@ static void kmem_cgroup_destroy(struct mem_cgroup *memcg)
 	mem_cgroup_sockets_destroy(memcg);
 }
 #else
-static int register_kmem_files(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
+static int memcg_init_kmem(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	return 0;
 }
@@ -5010,6 +5010,17 @@ mem_cgroup_create(struct cgroup *cont)
 	memcg->move_charge_at_immigrate = 0;
 	mutex_init(&memcg->thresholds_lock);
 	spin_lock_init(&memcg->move_lock);
+
+	error = memcg_init_kmem(memcg, &mem_cgroup_subsys);
+	if (error) {
+		/*
+		 * We call put now because our (and parent's) refcnts
+		 * are already in place. mem_cgroup_put() will internally
+		 * call __mem_cgroup_free, so return directly
+		 */
+		mem_cgroup_put(memcg);
+		return ERR_PTR(error);
+	}
 	return &memcg->css;
 free_out:
 	__mem_cgroup_free(memcg);
@@ -5032,13 +5043,6 @@ static void mem_cgroup_destroy(struct cgroup *cont)
 	mem_cgroup_put(memcg);
 }
 
-static int mem_cgroup_populate(struct cgroup_subsys *ss,
-				struct cgroup *cont)
-{
-	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
-	return register_kmem_files(memcg, ss);
-}
-
 #ifdef CONFIG_MMU
 /* Handlers for move charge at task migration. */
 #define PRECHARGE_COUNT_AT_ONCE	256
@@ -5622,7 +5626,6 @@ struct cgroup_subsys mem_cgroup_subsys = {
 	.create = mem_cgroup_create,
 	.pre_destroy = mem_cgroup_pre_destroy,
 	.destroy = mem_cgroup_destroy,
-	.populate = mem_cgroup_populate,
 	.can_attach = mem_cgroup_can_attach,
 	.cancel_attach = mem_cgroup_cancel_attach,
 	.attach = mem_cgroup_move_task,
-- 
1.7.7.6

[PATCH v3 2/2] cgroup: get rid of populate for memcg

[Glauber Costa]

The last man standing justifying the need for populate() is the
sock memcg initialization functions. Now that we are able to pass
a struct mem_cgroup instead of a struct cgroup to the socket
initialization, there is nothing that stops us from initializing
everything in create().

Signed-off-by: Glauber Costa <glommer-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org>
CC: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
CC: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
CC: Kamezawa Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>
CC: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
CC: Michal Hocko <mhocko-AlSwsSmVLrQ@public.gmane.org>
---
 mm/memcontrol.c |   23 +++++++++++++----------
 1 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 704054d..02b01d2 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -4640,7 +4640,7 @@ static int mem_control_numa_stat_open(struct inode *unused, struct file *file)
 #endif /* CONFIG_NUMA */
 
 #ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
-static int register_kmem_files(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
+static int memcg_init_kmem(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	return mem_cgroup_sockets_init(memcg, ss);
 };
@@ -4650,7 +4650,7 @@ static void kmem_cgroup_destroy(struct mem_cgroup *memcg)
 	mem_cgroup_sockets_destroy(memcg);
 }
 #else
-static int register_kmem_files(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
+static int memcg_init_kmem(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
 {
 	return 0;
 }
@@ -5010,6 +5010,17 @@ mem_cgroup_create(struct cgroup *cont)
 	memcg->move_charge_at_immigrate = 0;
 	mutex_init(&memcg->thresholds_lock);
 	spin_lock_init(&memcg->move_lock);
+
+	error = memcg_init_kmem(memcg, &mem_cgroup_subsys);
+	if (error) {
+		/*
+		 * We call put now because our (and parent's) refcnts
+		 * are already in place. mem_cgroup_put() will internally
+		 * call __mem_cgroup_free, so return directly
+		 */
+		mem_cgroup_put(memcg);
+		return ERR_PTR(error);
+	}
 	return &memcg->css;
 free_out:
 	__mem_cgroup_free(memcg);
@@ -5032,13 +5043,6 @@ static void mem_cgroup_destroy(struct cgroup *cont)
 	mem_cgroup_put(memcg);
 }
 
-static int mem_cgroup_populate(struct cgroup_subsys *ss,
-				struct cgroup *cont)
-{
-	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
-	return register_kmem_files(memcg, ss);
-}
-
 #ifdef CONFIG_MMU
 /* Handlers for move charge at task migration. */
 #define PRECHARGE_COUNT_AT_ONCE	256
@@ -5622,7 +5626,6 @@ struct cgroup_subsys mem_cgroup_subsys = {
 	.create = mem_cgroup_create,
 	.pre_destroy = mem_cgroup_pre_destroy,
 	.destroy = mem_cgroup_destroy,
-	.populate = mem_cgroup_populate,
 	.can_attach = mem_cgroup_can_attach,
 	.cancel_attach = mem_cgroup_cancel_attach,
 	.attach = mem_cgroup_move_task,
-- 
1.7.7.6

Re: [PATCH v3 2/2] cgroup: get rid of populate for memcg

[Tejun Heo]

On Mon, Apr 09, 2012 at 07:36:34PM -0300, Glauber Costa wrote:
> The last man standing justifying the need for populate() is the
> sock memcg initialization functions. Now that we are able to pass
> a struct mem_cgroup instead of a struct cgroup to the socket
> initialization, there is nothing that stops us from initializing
> everything in create().
> 
> Signed-off-by: Glauber Costa <glommer@parallels.com>
> CC: Tejun Heo <tj@kernel.org>
> CC: Li Zefan <lizefan@huawei.com>
> CC: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> CC: Johannes Weiner <hannes@cmpxchg.org>
> CC: Michal Hocko <mhocko@suse.cz>

Will apply once memcg maintainers ack.

Thanks.

-- 
tejun

Re: [PATCH v3 1/2] cgroup: pass struct mem_cgroup instead of struct cgroup to socket memcg

[KAMEZAWA Hiroyuki]

(2012/04/10 7:36), Glauber Costa wrote:

> The only reason cgroup was used, was to be consistent with the populate()
> interface. Now that we're getting rid of it, not only we no longer need
> it, but we also *can't* call it this way.
> 
> Since we will no longer rely on populate(), this will be called from
> create(). During create, the association between struct mem_cgroup
> and struct cgroup does not yet exist, since cgroup internals hasn't
> yet initialized its bookkeeping. This means we would not be able
> to draw the memcg pointer from the cgroup pointer in these
> functions, which is highly undesirable.
> 
> Signed-off-by: Glauber Costa <glommer-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org>
> CC: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
> CC: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
> CC: Kamezawa Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>
> CC: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
> CC: Michal Hocko <mhocko-AlSwsSmVLrQ@public.gmane.org>


Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>

Re: [PATCH v3 2/2] cgroup: get rid of populate for memcg

[KAMEZAWA Hiroyuki]

(2012/04/10 7:36), Glauber Costa wrote:

> The last man standing justifying the need for populate() is the
> sock memcg initialization functions. Now that we are able to pass
> a struct mem_cgroup instead of a struct cgroup to the socket
> initialization, there is nothing that stops us from initializing
> everything in create().
> 
> Signed-off-by: Glauber Costa <glommer-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org>
> CC: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
> CC: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
> CC: Kamezawa Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>
> CC: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
> CC: Michal Hocko <mhocko-AlSwsSmVLrQ@public.gmane.org>


Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>

Re: [PATCH v3 2/2] cgroup: get rid of populate for memcg

[Tejun Heo]

On Tue, Apr 10, 2012 at 11:44:03AM +0900, KAMEZAWA Hiroyuki wrote:
> (2012/04/10 7:36), Glauber Costa wrote:
> 
> > The last man standing justifying the need for populate() is the
> > sock memcg initialization functions. Now that we are able to pass
> > a struct mem_cgroup instead of a struct cgroup to the socket
> > initialization, there is nothing that stops us from initializing
> > everything in create().
> > 
> > Signed-off-by: Glauber Costa <glommer-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org>
> > CC: Tejun Heo <tj-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
> > CC: Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
> > CC: Kamezawa Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>
> > CC: Johannes Weiner <hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org>
> > CC: Michal Hocko <mhocko-AlSwsSmVLrQ@public.gmane.org>
> 
> Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu-+CUm20s59erQFUHtdCDX3A@public.gmane.org>

Applied to cgroup/for-3.5.  Thanks.

-- 
tejun