[deliverable/linux.git] / net / ipv4 / tcp_memcontrol.c

#include <net/tcp.h>
#include <net/tcp_memcontrol.h>
#include <net/sock.h>
#include <net/ip.h>
#include <linux/nsproxy.h>
#include <linux/memcontrol.h>
#include <linux/module.h>

static void memcg_tcp_enter_memory_pressure(struct sock *sk)
{
	if (sk->sk_cgrp->memory_pressure)
		sk->sk_cgrp->memory_pressure = 1;
}
EXPORT_SYMBOL(memcg_tcp_enter_memory_pressure);

int tcp_init_cgroup(struct mem_cgroup *memcg, struct cgroup_subsys *ss)
{
	/*
	 * The root cgroup does not use res_counters, but rather,
	 * rely on the data already collected by the network
	 * subsystem
	 */
	struct res_counter *res_parent = NULL;
	struct cg_proto *cg_proto, *parent_cg;
	struct mem_cgroup *parent = parent_mem_cgroup(memcg);

	cg_proto = tcp_prot.proto_cgroup(memcg);
	if (!cg_proto)
		return 0;

	cg_proto->sysctl_mem[0] = sysctl_tcp_mem[0];
	cg_proto->sysctl_mem[1] = sysctl_tcp_mem[1];
	cg_proto->sysctl_mem[2] = sysctl_tcp_mem[2];
	cg_proto->memory_pressure = 0;
	cg_proto->memcg = memcg;

	parent_cg = tcp_prot.proto_cgroup(parent);
	if (parent_cg)
		res_parent = &parent_cg->memory_allocated;

	res_counter_init(&cg_proto->memory_allocated, res_parent);
	percpu_counter_init(&cg_proto->sockets_allocated, 0);

	return 0;
}
EXPORT_SYMBOL(tcp_init_cgroup);

void tcp_destroy_cgroup(struct mem_cgroup *memcg)
{
	struct cg_proto *cg_proto;

	cg_proto = tcp_prot.proto_cgroup(memcg);
	if (!cg_proto)
		return;

	percpu_counter_destroy(&cg_proto->sockets_allocated);
}
EXPORT_SYMBOL(tcp_destroy_cgroup);

static int tcp_update_limit(struct mem_cgroup *memcg, u64 val)
{
	struct cg_proto *cg_proto;
	u64 old_lim;
	int i;
	int ret;

	cg_proto = tcp_prot.proto_cgroup(memcg);
	if (!cg_proto)
		return -EINVAL;

	if (val > RES_COUNTER_MAX)
		val = RES_COUNTER_MAX;

	old_lim = res_counter_read_u64(&cg_proto->memory_allocated, RES_LIMIT);
	ret = res_counter_set_limit(&cg_proto->memory_allocated, val);
	if (ret)
		return ret;

	for (i = 0; i < 3; i++)
		cg_proto->sysctl_mem[i] = min_t(long, val >> PAGE_SHIFT,
						sysctl_tcp_mem[i]);

	if (val == RES_COUNTER_MAX)
		clear_bit(MEMCG_SOCK_ACTIVE, &cg_proto->flags);
	else if (val != RES_COUNTER_MAX) {
		/*
		 * The active bit needs to be written after the static_key
		 * update. This is what guarantees that the socket activation
		 * function is the last one to run. See sock_update_memcg() for
		 * details, and note that we don't mark any socket as belonging
		 * to this memcg until that flag is up.
		 *
		 * We need to do this, because static_keys will span multiple
		 * sites, but we can't control their order. If we mark a socket
		 * as accounted, but the accounting functions are not patched in
		 * yet, we'll lose accounting.
		 *
		 * We never race with the readers in sock_update_memcg(),
		 * because when this value change, the code to process it is not
		 * patched in yet.
		 *
		 * The activated bit is used to guarantee that no two writers
		 * will do the update in the same memcg. Without that, we can't
		 * properly shutdown the static key.
		 */
		if (!test_and_set_bit(MEMCG_SOCK_ACTIVATED, &cg_proto->flags))
			static_key_slow_inc(&memcg_socket_limit_enabled);
		set_bit(MEMCG_SOCK_ACTIVE, &cg_proto->flags);
	}

	return 0;
}

static int tcp_cgroup_write(struct cgroup_subsys_state *css, struct cftype *cft,
			    const char *buffer)
{
	struct mem_cgroup *memcg = mem_cgroup_from_css(css);
	unsigned long long val;
	int ret = 0;

	switch (cft->private) {
	case RES_LIMIT:
		/* see memcontrol.c */
		ret = res_counter_memparse_write_strategy(buffer, &val);
		if (ret)
			break;
		ret = tcp_update_limit(memcg, val);
		break;
	default:
		ret = -EINVAL;
		break;
	}
	return ret;
}

static u64 tcp_read_stat(struct mem_cgroup *memcg, int type, u64 default_val)
{
	struct cg_proto *cg_proto;

	cg_proto = tcp_prot.proto_cgroup(memcg);
	if (!cg_proto)
		return default_val;

	return res_counter_read_u64(&cg_proto->memory_allocated, type);
}

static u64 tcp_read_usage(struct mem_cgroup *memcg)
{
	struct cg_proto *cg_proto;

	cg_proto = tcp_prot.proto_cgroup(memcg);
	if (!cg_proto)
		return atomic_long_read(&tcp_memory_allocated) << PAGE_SHIFT;

	return res_counter_read_u64(&cg_proto->memory_allocated, RES_USAGE);
}

static u64 tcp_cgroup_read(struct cgroup_subsys_state *css, struct cftype *cft)
{
	struct mem_cgroup *memcg = mem_cgroup_from_css(css);
	u64 val;

	switch (cft->private) {
	case RES_LIMIT:
		val = tcp_read_stat(memcg, RES_LIMIT, RES_COUNTER_MAX);
		break;
	case RES_USAGE:
		val = tcp_read_usage(memcg);
		break;
	case RES_FAILCNT:
	case RES_MAX_USAGE:
		val = tcp_read_stat(memcg, cft->private, 0);
		break;
	default:
		BUG();
	}
	return val;
}

static int tcp_cgroup_reset(struct cgroup_subsys_state *css, unsigned int event)
{
	struct mem_cgroup *memcg;
	struct cg_proto *cg_proto;

	memcg = mem_cgroup_from_css(css);
	cg_proto = tcp_prot.proto_cgroup(memcg);
	if (!cg_proto)
		return 0;

	switch (event) {
	case RES_MAX_USAGE:
		res_counter_reset_max(&cg_proto->memory_allocated);
		break;
	case RES_FAILCNT:
		res_counter_reset_failcnt(&cg_proto->memory_allocated);
		break;
	}

	return 0;
}

static struct cftype tcp_files[] = {
	{
		.name = "kmem.tcp.limit_in_bytes",
		.write_string = tcp_cgroup_write,
		.read_u64 = tcp_cgroup_read,
		.private = RES_LIMIT,
	},
	{
		.name = "kmem.tcp.usage_in_bytes",
		.read_u64 = tcp_cgroup_read,
		.private = RES_USAGE,
	},
	{
		.name = "kmem.tcp.failcnt",
		.private = RES_FAILCNT,
		.trigger = tcp_cgroup_reset,
		.read_u64 = tcp_cgroup_read,
	},
	{
		.name = "kmem.tcp.max_usage_in_bytes",
		.private = RES_MAX_USAGE,
		.trigger = tcp_cgroup_reset,
		.read_u64 = tcp_cgroup_read,
	},
	{ }	/* terminate */
};

static int __init tcp_memcontrol_init(void)
{
	WARN_ON(cgroup_add_cftypes(&mem_cgroup_subsys, tcp_files));
	return 0;
}
__initcall(tcp_memcontrol_init);
Commit	Line	Data
d1a4c0b3 GC	1	#include <net/tcp.h>
	2	#include <net/tcp_memcontrol.h>
	3	#include <net/sock.h>
3dc43e3e GC	4	#include <net/ip.h>
3dc43e3e GC	5	#include <linux/nsproxy.h>
d1a4c0b3 GC	6	#include <linux/memcontrol.h>
	7	#include <linux/module.h>
	8
d1a4c0b3 GC	9	static void memcg_tcp_enter_memory_pressure(struct sock *sk)
d1a4c0b3 GC	10	{
c48e074c	11	if (sk->sk_cgrp->memory_pressure)
2e685cad	12	sk->sk_cgrp->memory_pressure = 1;
d1a4c0b3 GC	13	}
	14	EXPORT_SYMBOL(memcg_tcp_enter_memory_pressure);
	15
1d62e436	16	int tcp_init_cgroup(struct mem_cgroup memcg, struct cgroup_subsys ss)
d1a4c0b3 GC	17	{
	18	/*
	19	* The root cgroup does not use res_counters, but rather,
	20	* rely on the data already collected by the network
	21	* subsystem
	22	*/
	23	struct res_counter *res_parent = NULL;
	24	struct cg_proto cg_proto, parent_cg;
d1a4c0b3 GC	25	struct mem_cgroup *parent = parent_mem_cgroup(memcg);
	26
	27	cg_proto = tcp_prot.proto_cgroup(memcg);
	28	if (!cg_proto)
6bc10349	29	return 0;
d1a4c0b3	30
2e685cad EB	31	cg_proto->sysctl_mem[0] = sysctl_tcp_mem[0];
	32	cg_proto->sysctl_mem[1] = sysctl_tcp_mem[1];
	33	cg_proto->sysctl_mem[2] = sysctl_tcp_mem[2];
	34	cg_proto->memory_pressure = 0;
	35	cg_proto->memcg = memcg;
d1a4c0b3 GC	36
	37	parent_cg = tcp_prot.proto_cgroup(parent);
	38	if (parent_cg)
2e685cad	39	res_parent = &parent_cg->memory_allocated;
d1a4c0b3	40
2e685cad EB	41	res_counter_init(&cg_proto->memory_allocated, res_parent);
2e685cad EB	42	percpu_counter_init(&cg_proto->sockets_allocated, 0);
d1a4c0b3	43
6bc10349	44	return 0;
d1a4c0b3 GC	45	}
	46	EXPORT_SYMBOL(tcp_init_cgroup);
	47
1d62e436	48	void tcp_destroy_cgroup(struct mem_cgroup *memcg)
d1a4c0b3	49	{
d1a4c0b3	50	struct cg_proto *cg_proto;
d1a4c0b3 GC	51
	52	cg_proto = tcp_prot.proto_cgroup(memcg);
	53	if (!cg_proto)
	54	return;
	55
2e685cad	56	percpu_counter_destroy(&cg_proto->sockets_allocated);
d1a4c0b3 GC	57	}
d1a4c0b3 GC	58	EXPORT_SYMBOL(tcp_destroy_cgroup);
3aaabe23 GC	59
	60	static int tcp_update_limit(struct mem_cgroup *memcg, u64 val)
	61	{
3aaabe23 GC	62	struct cg_proto *cg_proto;
	63	u64 old_lim;
	64	int i;
	65	int ret;
	66
	67	cg_proto = tcp_prot.proto_cgroup(memcg);
	68	if (!cg_proto)
	69	return -EINVAL;
	70
6de5a8bf SZ	71	if (val > RES_COUNTER_MAX)
6de5a8bf SZ	72	val = RES_COUNTER_MAX;
3aaabe23	73
2e685cad EB	74	old_lim = res_counter_read_u64(&cg_proto->memory_allocated, RES_LIMIT);
2e685cad EB	75	ret = res_counter_set_limit(&cg_proto->memory_allocated, val);
3aaabe23 GC	76	if (ret)
	77	return ret;
	78
	79	for (i = 0; i < 3; i++)
2e685cad EB	80	cg_proto->sysctl_mem[i] = min_t(long, val >> PAGE_SHIFT,
2e685cad EB	81	sysctl_tcp_mem[i]);
3aaabe23	82
6de5a8bf	83	if (val == RES_COUNTER_MAX)
3f134619	84	clear_bit(MEMCG_SOCK_ACTIVE, &cg_proto->flags);
6de5a8bf	85	else if (val != RES_COUNTER_MAX) {
3f134619 GC	86	/*
	87	* The active bit needs to be written after the static_key
	88	* update. This is what guarantees that the socket activation
	89	* function is the last one to run. See sock_update_memcg() for
	90	* details, and note that we don't mark any socket as belonging
	91	* to this memcg until that flag is up.
	92	*
	93	* We need to do this, because static_keys will span multiple
	94	* sites, but we can't control their order. If we mark a socket
	95	* as accounted, but the accounting functions are not patched in
	96	* yet, we'll lose accounting.
	97	*
	98	* We never race with the readers in sock_update_memcg(),
	99	* because when this value change, the code to process it is not
	100	* patched in yet.
	101	*
	102	* The activated bit is used to guarantee that no two writers
	103	* will do the update in the same memcg. Without that, we can't
	104	* properly shutdown the static key.
	105	*/
	106	if (!test_and_set_bit(MEMCG_SOCK_ACTIVATED, &cg_proto->flags))
	107	static_key_slow_inc(&memcg_socket_limit_enabled);
	108	set_bit(MEMCG_SOCK_ACTIVE, &cg_proto->flags);
	109	}
3aaabe23 GC	110
	111	return 0;
	112	}
	113
182446d0	114	static int tcp_cgroup_write(struct cgroup_subsys_state css, struct cftype cft,
3aaabe23 GC	115	const char *buffer)
3aaabe23 GC	116	{
182446d0	117	struct mem_cgroup *memcg = mem_cgroup_from_css(css);
3aaabe23 GC	118	unsigned long long val;
	119	int ret = 0;
	120
	121	switch (cft->private) {
	122	case RES_LIMIT:
	123	/* see memcontrol.c */
	124	ret = res_counter_memparse_write_strategy(buffer, &val);
	125	if (ret)
	126	break;
	127	ret = tcp_update_limit(memcg, val);
	128	break;
	129	default:
	130	ret = -EINVAL;
	131	break;
	132	}
	133	return ret;
	134	}
	135
	136	static u64 tcp_read_stat(struct mem_cgroup *memcg, int type, u64 default_val)
	137	{
3aaabe23 GC	138	struct cg_proto *cg_proto;
	139
	140	cg_proto = tcp_prot.proto_cgroup(memcg);
	141	if (!cg_proto)
	142	return default_val;
	143
2e685cad	144	return res_counter_read_u64(&cg_proto->memory_allocated, type);
3aaabe23 GC	145	}
3aaabe23 GC	146
5a6dd343 GC	147	static u64 tcp_read_usage(struct mem_cgroup *memcg)
5a6dd343 GC	148	{
5a6dd343 GC	149	struct cg_proto *cg_proto;
	150
	151	cg_proto = tcp_prot.proto_cgroup(memcg);
	152	if (!cg_proto)
	153	return atomic_long_read(&tcp_memory_allocated) << PAGE_SHIFT;
	154
2e685cad	155	return res_counter_read_u64(&cg_proto->memory_allocated, RES_USAGE);
5a6dd343 GC	156	}
5a6dd343 GC	157
182446d0	158	static u64 tcp_cgroup_read(struct cgroup_subsys_state css, struct cftype cft)
3aaabe23	159	{
182446d0	160	struct mem_cgroup *memcg = mem_cgroup_from_css(css);
3aaabe23 GC	161	u64 val;
	162
	163	switch (cft->private) {
	164	case RES_LIMIT:
6de5a8bf	165	val = tcp_read_stat(memcg, RES_LIMIT, RES_COUNTER_MAX);
3aaabe23	166	break;
5a6dd343 GC	167	case RES_USAGE:
	168	val = tcp_read_usage(memcg);
	169	break;
ffea59e5	170	case RES_FAILCNT:
0850f0f5 GC	171	case RES_MAX_USAGE:
0850f0f5 GC	172	val = tcp_read_stat(memcg, cft->private, 0);
ffea59e5	173	break;
3aaabe23 GC	174	default:
	175	BUG();
	176	}
	177	return val;
	178	}
	179
182446d0	180	static int tcp_cgroup_reset(struct cgroup_subsys_state *css, unsigned int event)
ffea59e5 GC	181	{
ffea59e5 GC	182	struct mem_cgroup *memcg;
ffea59e5 GC	183	struct cg_proto *cg_proto;
ffea59e5 GC	184
182446d0	185	memcg = mem_cgroup_from_css(css);
ffea59e5 GC	186	cg_proto = tcp_prot.proto_cgroup(memcg);
	187	if (!cg_proto)
	188	return 0;
ffea59e5 GC	189
ffea59e5 GC	190	switch (event) {
0850f0f5	191	case RES_MAX_USAGE:
2e685cad	192	res_counter_reset_max(&cg_proto->memory_allocated);
0850f0f5	193	break;
ffea59e5	194	case RES_FAILCNT:
2e685cad	195	res_counter_reset_failcnt(&cg_proto->memory_allocated);
ffea59e5 GC	196	break;
	197	}
	198
	199	return 0;
	200	}
	201
676f7c8f TH	202	static struct cftype tcp_files[] = {
	203	{
	204	.name = "kmem.tcp.limit_in_bytes",
	205	.write_string = tcp_cgroup_write,
	206	.read_u64 = tcp_cgroup_read,
	207	.private = RES_LIMIT,
	208	},
	209	{
	210	.name = "kmem.tcp.usage_in_bytes",
	211	.read_u64 = tcp_cgroup_read,
	212	.private = RES_USAGE,
	213	},
	214	{
	215	.name = "kmem.tcp.failcnt",
	216	.private = RES_FAILCNT,
	217	.trigger = tcp_cgroup_reset,
	218	.read_u64 = tcp_cgroup_read,
	219	},
	220	{
	221	.name = "kmem.tcp.max_usage_in_bytes",
	222	.private = RES_MAX_USAGE,
	223	.trigger = tcp_cgroup_reset,
	224	.read_u64 = tcp_cgroup_read,
	225	},
6bc10349	226	{ } /* terminate */
676f7c8f	227	};
6bc10349 TH	228
	229	static int __init tcp_memcontrol_init(void)
	230	{
	231	WARN_ON(cgroup_add_cftypes(&mem_cgroup_subsys, tcp_files));
	232	return 0;
	233	}
	234	__initcall(tcp_memcontrol_init);