[deliverable/linux.git] / net / core / netprio_cgroup.c

/*
 * net/core/netprio_cgroup.c	Priority Control Group
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 *
 * Authors:	Neil Horman <nhorman@tuxdriver.com>
 */

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

#include <linux/module.h>
#include <linux/slab.h>
#include <linux/types.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/skbuff.h>
#include <linux/cgroup.h>
#include <linux/rcupdate.h>
#include <linux/atomic.h>
#include <net/rtnetlink.h>
#include <net/pkt_cls.h>
#include <net/sock.h>
#include <net/netprio_cgroup.h>

#define PRIOIDX_SZ 128

static unsigned long prioidx_map[PRIOIDX_SZ];
static DEFINE_SPINLOCK(prioidx_map_lock);
static atomic_t max_prioidx = ATOMIC_INIT(0);

static inline struct cgroup_netprio_state *cgrp_netprio_state(struct cgroup *cgrp)
{
	return container_of(cgroup_subsys_state(cgrp, net_prio_subsys_id),
			    struct cgroup_netprio_state, css);
}

static int get_prioidx(u32 *prio)
{
	unsigned long flags;
	u32 prioidx;

	spin_lock_irqsave(&prioidx_map_lock, flags);
	prioidx = find_first_zero_bit(prioidx_map, sizeof(unsigned long) * PRIOIDX_SZ);
	if (prioidx == sizeof(unsigned long) * PRIOIDX_SZ) {
		spin_unlock_irqrestore(&prioidx_map_lock, flags);
		return -ENOSPC;
	}
	set_bit(prioidx, prioidx_map);
	if (atomic_read(&max_prioidx) < prioidx)
		atomic_set(&max_prioidx, prioidx);
	spin_unlock_irqrestore(&prioidx_map_lock, flags);
	*prio = prioidx;
	return 0;
}

static void put_prioidx(u32 idx)
{
	unsigned long flags;

	spin_lock_irqsave(&prioidx_map_lock, flags);
	clear_bit(idx, prioidx_map);
	spin_unlock_irqrestore(&prioidx_map_lock, flags);
}

static int extend_netdev_table(struct net_device *dev, u32 new_len)
{
	size_t new_size = sizeof(struct netprio_map) +
			   ((sizeof(u32) * new_len));
	struct netprio_map *new_priomap = kzalloc(new_size, GFP_KERNEL);
	struct netprio_map *old_priomap;
	int i;

	old_priomap  = rtnl_dereference(dev->priomap);

	if (!new_priomap) {
		pr_warn("Unable to alloc new priomap!\n");
		return -ENOMEM;
	}

	for (i = 0;
	     old_priomap && (i < old_priomap->priomap_len);
	     i++)
		new_priomap->priomap[i] = old_priomap->priomap[i];

	new_priomap->priomap_len = new_len;

	rcu_assign_pointer(dev->priomap, new_priomap);
	if (old_priomap)
		kfree_rcu(old_priomap, rcu);
	return 0;
}

static int write_update_netdev_table(struct net_device *dev)
{
	int ret = 0;
	u32 max_len;
	struct netprio_map *map;

	rtnl_lock();
	max_len = atomic_read(&max_prioidx) + 1;
	map = rtnl_dereference(dev->priomap);
	if (!map || map->priomap_len < max_len)
		ret = extend_netdev_table(dev, max_len);
	rtnl_unlock();

	return ret;
}

static int update_netdev_tables(void)
{
	int ret = 0;
	struct net_device *dev;
	u32 max_len;
	struct netprio_map *map;

	rtnl_lock();
	max_len = atomic_read(&max_prioidx) + 1;
	for_each_netdev(&init_net, dev) {
		map = rtnl_dereference(dev->priomap);
		/*
		 * don't allocate priomap if we didn't
		 * change net_prio.ifpriomap (map == NULL),
		 * this will speed up skb_update_prio.
		 */
		if (map && map->priomap_len < max_len) {
			ret = extend_netdev_table(dev, max_len);
			if (ret < 0)
				break;
		}
	}
	rtnl_unlock();
	return ret;
}

static struct cgroup_subsys_state *cgrp_create(struct cgroup *cgrp)
{
	struct cgroup_netprio_state *cs;
	int ret = -EINVAL;

	cs = kzalloc(sizeof(*cs), GFP_KERNEL);
	if (!cs)
		return ERR_PTR(-ENOMEM);

	if (cgrp->parent && cgrp_netprio_state(cgrp->parent)->prioidx)
		goto out;

	ret = get_prioidx(&cs->prioidx);
	if (ret < 0) {
		pr_warn("No space in priority index array\n");
		goto out;
	}

	ret = update_netdev_tables();
	if (ret < 0) {
		put_prioidx(cs->prioidx);
		goto out;
	}

	return &cs->css;
out:
	kfree(cs);
	return ERR_PTR(ret);
}

static void cgrp_destroy(struct cgroup *cgrp)
{
	struct cgroup_netprio_state *cs;
	struct net_device *dev;
	struct netprio_map *map;

	cs = cgrp_netprio_state(cgrp);
	rtnl_lock();
	for_each_netdev(&init_net, dev) {
		map = rtnl_dereference(dev->priomap);
		if (map && cs->prioidx < map->priomap_len)
			map->priomap[cs->prioidx] = 0;
	}
	rtnl_unlock();
	put_prioidx(cs->prioidx);
	kfree(cs);
}

static u64 read_prioidx(struct cgroup *cgrp, struct cftype *cft)
{
	return (u64)cgrp_netprio_state(cgrp)->prioidx;
}

static int read_priomap(struct cgroup *cont, struct cftype *cft,
			struct cgroup_map_cb *cb)
{
	struct net_device *dev;
	u32 prioidx = cgrp_netprio_state(cont)->prioidx;
	u32 priority;
	struct netprio_map *map;

	rcu_read_lock();
	for_each_netdev_rcu(&init_net, dev) {
		map = rcu_dereference(dev->priomap);
		priority = (map && prioidx < map->priomap_len) ? map->priomap[prioidx] : 0;
		cb->fill(cb, dev->name, priority);
	}
	rcu_read_unlock();
	return 0;
}

static int write_priomap(struct cgroup *cgrp, struct cftype *cft,
			 const char *buffer)
{
	char *devname = kstrdup(buffer, GFP_KERNEL);
	int ret = -EINVAL;
	u32 prioidx = cgrp_netprio_state(cgrp)->prioidx;
	unsigned long priority;
	char *priostr;
	struct net_device *dev;
	struct netprio_map *map;

	if (!devname)
		return -ENOMEM;

	/*
	 * Minimally sized valid priomap string
	 */
	if (strlen(devname) < 3)
		goto out_free_devname;

	priostr = strstr(devname, " ");
	if (!priostr)
		goto out_free_devname;

	/*
	 *Separate the devname from the associated priority
	 *and advance the priostr poitner to the priority value
	 */
	*priostr = '\0';
	priostr++;

	/*
	 * If the priostr points to NULL, we're at the end of the passed
	 * in string, and its not a valid write
	 */
	if (*priostr == '\0')
		goto out_free_devname;

	ret = kstrtoul(priostr, 10, &priority);
	if (ret < 0)
		goto out_free_devname;

	ret = -ENODEV;

	dev = dev_get_by_name(&init_net, devname);
	if (!dev)
		goto out_free_devname;

	ret = write_update_netdev_table(dev);
	if (ret < 0)
		goto out_put_dev;

	rcu_read_lock();
	map = rcu_dereference(dev->priomap);
	if (map)
		map->priomap[prioidx] = priority;
	rcu_read_unlock();

out_put_dev:
	dev_put(dev);

out_free_devname:
	kfree(devname);
	return ret;
}

static struct cftype ss_files[] = {
	{
		.name = "prioidx",
		.read_u64 = read_prioidx,
	},
	{
		.name = "ifpriomap",
		.read_map = read_priomap,
		.write_string = write_priomap,
	},
	{ }	/* terminate */
};

struct cgroup_subsys net_prio_subsys = {
	.name		= "net_prio",
	.create		= cgrp_create,
	.destroy	= cgrp_destroy,
#ifdef CONFIG_NETPRIO_CGROUP
	.subsys_id	= net_prio_subsys_id,
#endif
	.base_cftypes	= ss_files,
	.module		= THIS_MODULE
};

static int netprio_device_event(struct notifier_block *unused,
				unsigned long event, void *ptr)
{
	struct net_device *dev = ptr;
	struct netprio_map *old;

	/*
	 * Note this is called with rtnl_lock held so we have update side
	 * protection on our rcu assignments
	 */

	switch (event) {
	case NETDEV_UNREGISTER:
		old = rtnl_dereference(dev->priomap);
		RCU_INIT_POINTER(dev->priomap, NULL);
		if (old)
			kfree_rcu(old, rcu);
		break;
	}
	return NOTIFY_DONE;
}

static struct notifier_block netprio_device_notifier = {
	.notifier_call = netprio_device_event
};

static int __init init_cgroup_netprio(void)
{
	int ret;

	ret = cgroup_load_subsys(&net_prio_subsys);
	if (ret)
		goto out;
#ifndef CONFIG_NETPRIO_CGROUP
	smp_wmb();
	net_prio_subsys_id = net_prio_subsys.subsys_id;
#endif

	register_netdevice_notifier(&netprio_device_notifier);

out:
	return ret;
}

static void __exit exit_cgroup_netprio(void)
{
	struct netprio_map *old;
	struct net_device *dev;

	unregister_netdevice_notifier(&netprio_device_notifier);

	cgroup_unload_subsys(&net_prio_subsys);

#ifndef CONFIG_NETPRIO_CGROUP
	net_prio_subsys_id = -1;
	synchronize_rcu();
#endif

	rtnl_lock();
	for_each_netdev(&init_net, dev) {
		old = rtnl_dereference(dev->priomap);
		RCU_INIT_POINTER(dev->priomap, NULL);
		if (old)
			kfree_rcu(old, rcu);
	}
	rtnl_unlock();
}

module_init(init_cgroup_netprio);
module_exit(exit_cgroup_netprio);
MODULE_LICENSE("GPL v2");
Commit	Line	Data
5bc1421e NH	1	/*
	2	* net/core/netprio_cgroup.c Priority Control Group
	3	*
	4	* This program is free software; you can redistribute it and/or
	5	* modify it under the terms of the GNU General Public License
	6	* as published by the Free Software Foundation; either version
	7	* 2 of the License, or (at your option) any later version.
	8	*
	9	* Authors: Neil Horman <nhorman@tuxdriver.com>
	10	*/
	11
e005d193 JP	12	#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
e005d193 JP	13
5bc1421e NH	14	#include <linux/module.h>
	15	#include <linux/slab.h>
	16	#include <linux/types.h>
	17	#include <linux/string.h>
	18	#include <linux/errno.h>
	19	#include <linux/skbuff.h>
	20	#include <linux/cgroup.h>
	21	#include <linux/rcupdate.h>
	22	#include <linux/atomic.h>
	23	#include <net/rtnetlink.h>
	24	#include <net/pkt_cls.h>
	25	#include <net/sock.h>
	26	#include <net/netprio_cgroup.h>
	27
5bc1421e NH	28	#define PRIOIDX_SZ 128
	29
	30	static unsigned long prioidx_map[PRIOIDX_SZ];
	31	static DEFINE_SPINLOCK(prioidx_map_lock);
	32	static atomic_t max_prioidx = ATOMIC_INIT(0);
	33
	34	static inline struct cgroup_netprio_state cgrp_netprio_state(struct cgroup cgrp)
	35	{
	36	return container_of(cgroup_subsys_state(cgrp, net_prio_subsys_id),
	37	struct cgroup_netprio_state, css);
	38	}
	39
	40	static int get_prioidx(u32 *prio)
	41	{
	42	unsigned long flags;
	43	u32 prioidx;
	44
	45	spin_lock_irqsave(&prioidx_map_lock, flags);
	46	prioidx = find_first_zero_bit(prioidx_map, sizeof(unsigned long) * PRIOIDX_SZ);
5962b35c NH	47	if (prioidx == sizeof(unsigned long) * PRIOIDX_SZ) {
	48	spin_unlock_irqrestore(&prioidx_map_lock, flags);
	49	return -ENOSPC;
	50	}
5bc1421e	51	set_bit(prioidx, prioidx_map);
b761c9b1 G	52	if (atomic_read(&max_prioidx) < prioidx)
b761c9b1 G	53	atomic_set(&max_prioidx, prioidx);
5bc1421e	54	spin_unlock_irqrestore(&prioidx_map_lock, flags);
5bc1421e NH	55	*prio = prioidx;
	56	return 0;
	57	}
	58
	59	static void put_prioidx(u32 idx)
	60	{
	61	unsigned long flags;
	62
	63	spin_lock_irqsave(&prioidx_map_lock, flags);
	64	clear_bit(idx, prioidx_map);
	65	spin_unlock_irqrestore(&prioidx_map_lock, flags);
	66	}
	67
ef209f15	68	static int extend_netdev_table(struct net_device *dev, u32 new_len)
5bc1421e NH	69	{
	70	size_t new_size = sizeof(struct netprio_map) +
	71	((sizeof(u32) * new_len));
	72	struct netprio_map *new_priomap = kzalloc(new_size, GFP_KERNEL);
	73	struct netprio_map *old_priomap;
	74	int i;
	75
	76	old_priomap = rtnl_dereference(dev->priomap);
	77
	78	if (!new_priomap) {
e005d193	79	pr_warn("Unable to alloc new priomap!\n");
ef209f15	80	return -ENOMEM;
5bc1421e NH	81	}
	82
	83	for (i = 0;
	84	old_priomap && (i < old_priomap->priomap_len);
	85	i++)
	86	new_priomap->priomap[i] = old_priomap->priomap[i];
	87
	88	new_priomap->priomap_len = new_len;
	89
	90	rcu_assign_pointer(dev->priomap, new_priomap);
	91	if (old_priomap)
	92	kfree_rcu(old_priomap, rcu);
ef209f15 G	93	return 0;
	94	}
	95
	96	static int write_update_netdev_table(struct net_device *dev)
	97	{
	98	int ret = 0;
	99	u32 max_len;
	100	struct netprio_map *map;
	101
	102	rtnl_lock();
	103	max_len = atomic_read(&max_prioidx) + 1;
	104	map = rtnl_dereference(dev->priomap);
	105	if (!map \|\| map->priomap_len < max_len)
	106	ret = extend_netdev_table(dev, max_len);
	107	rtnl_unlock();
	108
	109	return ret;
5bc1421e NH	110	}
5bc1421e NH	111
ef209f15	112	static int update_netdev_tables(void)
5bc1421e	113	{
ef209f15	114	int ret = 0;
5bc1421e	115	struct net_device *dev;
ef209f15	116	u32 max_len;
5bc1421e NH	117	struct netprio_map *map;
	118
	119	rtnl_lock();
ef209f15	120	max_len = atomic_read(&max_prioidx) + 1;
5bc1421e NH	121	for_each_netdev(&init_net, dev) {
5bc1421e NH	122	map = rtnl_dereference(dev->priomap);
ef209f15 G	123	/*
	124	* don't allocate priomap if we didn't
	125	* change net_prio.ifpriomap (map == NULL),
	126	* this will speed up skb_update_prio.
	127	*/
	128	if (map && map->priomap_len < max_len) {
	129	ret = extend_netdev_table(dev, max_len);
	130	if (ret < 0)
	131	break;
	132	}
5bc1421e NH	133	}
5bc1421e NH	134	rtnl_unlock();
ef209f15	135	return ret;
5bc1421e NH	136	}
5bc1421e NH	137
761b3ef5	138	static struct cgroup_subsys_state cgrp_create(struct cgroup cgrp)
5bc1421e NH	139	{
5bc1421e NH	140	struct cgroup_netprio_state *cs;
ef209f15	141	int ret = -EINVAL;
5bc1421e NH	142
	143	cs = kzalloc(sizeof(*cs), GFP_KERNEL);
	144	if (!cs)
	145	return ERR_PTR(-ENOMEM);
	146
ef209f15 G	147	if (cgrp->parent && cgrp_netprio_state(cgrp->parent)->prioidx)
ef209f15 G	148	goto out;
5bc1421e NH	149
5bc1421e NH	150	ret = get_prioidx(&cs->prioidx);
ef209f15	151	if (ret < 0) {
e005d193	152	pr_warn("No space in priority index array\n");
ef209f15 G	153	goto out;
	154	}
	155
	156	ret = update_netdev_tables();
	157	if (ret < 0) {
	158	put_prioidx(cs->prioidx);
	159	goto out;
5bc1421e NH	160	}
	161
	162	return &cs->css;
ef209f15 G	163	out:
	164	kfree(cs);
	165	return ERR_PTR(ret);
5bc1421e NH	166	}
5bc1421e NH	167
761b3ef5	168	static void cgrp_destroy(struct cgroup *cgrp)
5bc1421e NH	169	{
	170	struct cgroup_netprio_state *cs;
	171	struct net_device *dev;
	172	struct netprio_map *map;
	173
	174	cs = cgrp_netprio_state(cgrp);
	175	rtnl_lock();
	176	for_each_netdev(&init_net, dev) {
	177	map = rtnl_dereference(dev->priomap);
91c68ce2	178	if (map && cs->prioidx < map->priomap_len)
5bc1421e NH	179	map->priomap[cs->prioidx] = 0;
	180	}
	181	rtnl_unlock();
	182	put_prioidx(cs->prioidx);
	183	kfree(cs);
	184	}
	185
	186	static u64 read_prioidx(struct cgroup cgrp, struct cftype cft)
	187	{
	188	return (u64)cgrp_netprio_state(cgrp)->prioidx;
	189	}
	190
	191	static int read_priomap(struct cgroup cont, struct cftype cft,
	192	struct cgroup_map_cb *cb)
	193	{
	194	struct net_device *dev;
	195	u32 prioidx = cgrp_netprio_state(cont)->prioidx;
	196	u32 priority;
	197	struct netprio_map *map;
	198
	199	rcu_read_lock();
	200	for_each_netdev_rcu(&init_net, dev) {
	201	map = rcu_dereference(dev->priomap);
91c68ce2	202	priority = (map && prioidx < map->priomap_len) ? map->priomap[prioidx] : 0;
5bc1421e NH	203	cb->fill(cb, dev->name, priority);
	204	}
	205	rcu_read_unlock();
	206	return 0;
	207	}
	208
	209	static int write_priomap(struct cgroup cgrp, struct cftype cft,
	210	const char *buffer)
	211	{
	212	char *devname = kstrdup(buffer, GFP_KERNEL);
	213	int ret = -EINVAL;
	214	u32 prioidx = cgrp_netprio_state(cgrp)->prioidx;
	215	unsigned long priority;
	216	char *priostr;
	217	struct net_device *dev;
	218	struct netprio_map *map;
	219
	220	if (!devname)
	221	return -ENOMEM;
	222
	223	/*
	224	* Minimally sized valid priomap string
	225	*/
	226	if (strlen(devname) < 3)
	227	goto out_free_devname;
	228
	229	priostr = strstr(devname, " ");
	230	if (!priostr)
	231	goto out_free_devname;
	232
	233	/*
	234	*Separate the devname from the associated priority
	235	*and advance the priostr poitner to the priority value
	236	*/
	237	*priostr = '\0';
	238	priostr++;
	239
	240	/*
	241	* If the priostr points to NULL, we're at the end of the passed
	242	* in string, and its not a valid write
	243	*/
	244	if (*priostr == '\0')
	245	goto out_free_devname;
	246
	247	ret = kstrtoul(priostr, 10, &priority);
	248	if (ret < 0)
	249	goto out_free_devname;
	250
	251	ret = -ENODEV;
	252
	253	dev = dev_get_by_name(&init_net, devname);
	254	if (!dev)
	255	goto out_free_devname;
	256
ef209f15 G	257	ret = write_update_netdev_table(dev);
	258	if (ret < 0)
	259	goto out_put_dev;
	260
5bc1421e NH	261	rcu_read_lock();
	262	map = rcu_dereference(dev->priomap);
	263	if (map)
	264	map->priomap[prioidx] = priority;
	265	rcu_read_unlock();
ef209f15 G	266
ef209f15 G	267	out_put_dev:
5bc1421e NH	268	dev_put(dev);
	269
	270	out_free_devname:
	271	kfree(devname);
	272	return ret;
	273	}
	274
	275	static struct cftype ss_files[] = {
	276	{
	277	.name = "prioidx",
	278	.read_u64 = read_prioidx,
	279	},
	280	{
	281	.name = "ifpriomap",
	282	.read_map = read_priomap,
	283	.write_string = write_priomap,
	284	},
4baf6e33	285	{ } /* terminate */
5bc1421e NH	286	};
5bc1421e NH	287
676f7c8f TH	288	struct cgroup_subsys net_prio_subsys = {
	289	.name = "net_prio",
	290	.create = cgrp_create,
	291	.destroy = cgrp_destroy,
676f7c8f TH	292	#ifdef CONFIG_NETPRIO_CGROUP
	293	.subsys_id = net_prio_subsys_id,
	294	#endif
4baf6e33	295	.base_cftypes = ss_files,
676f7c8f TH	296	.module = THIS_MODULE
676f7c8f TH	297	};
5bc1421e NH	298
	299	static int netprio_device_event(struct notifier_block *unused,
	300	unsigned long event, void *ptr)
	301	{
	302	struct net_device *dev = ptr;
	303	struct netprio_map *old;
5bc1421e NH	304
	305	/*
	306	* Note this is called with rtnl_lock held so we have update side
	307	* protection on our rcu assignments
	308	*/
	309
	310	switch (event) {
5bc1421e NH	311	case NETDEV_UNREGISTER:
5bc1421e NH	312	old = rtnl_dereference(dev->priomap);
2cfa5a04	313	RCU_INIT_POINTER(dev->priomap, NULL);
5bc1421e NH	314	if (old)
	315	kfree_rcu(old, rcu);
	316	break;
	317	}
	318	return NOTIFY_DONE;
	319	}
	320
	321	static struct notifier_block netprio_device_notifier = {
	322	.notifier_call = netprio_device_event
	323	};
	324
	325	static int __init init_cgroup_netprio(void)
	326	{
	327	int ret;
	328
	329	ret = cgroup_load_subsys(&net_prio_subsys);
	330	if (ret)
	331	goto out;
	332	#ifndef CONFIG_NETPRIO_CGROUP
	333	smp_wmb();
	334	net_prio_subsys_id = net_prio_subsys.subsys_id;
	335	#endif
	336
	337	register_netdevice_notifier(&netprio_device_notifier);
	338
	339	out:
	340	return ret;
	341	}
	342
	343	static void __exit exit_cgroup_netprio(void)
	344	{
	345	struct netprio_map *old;
	346	struct net_device *dev;
	347
	348	unregister_netdevice_notifier(&netprio_device_notifier);
	349
	350	cgroup_unload_subsys(&net_prio_subsys);
	351
	352	#ifndef CONFIG_NETPRIO_CGROUP
	353	net_prio_subsys_id = -1;
	354	synchronize_rcu();
	355	#endif
	356
	357	rtnl_lock();
	358	for_each_netdev(&init_net, dev) {
	359	old = rtnl_dereference(dev->priomap);
2cfa5a04	360	RCU_INIT_POINTER(dev->priomap, NULL);
5bc1421e NH	361	if (old)
	362	kfree_rcu(old, rcu);
	363	}
	364	rtnl_unlock();
	365	}
	366
	367	module_init(init_cgroup_netprio);
	368	module_exit(exit_cgroup_netprio);
	369	MODULE_LICENSE("GPL v2");