[GitHub/mt8127/android_kernel_alcatel_ttab.git] / net / netfilter / ipvs / ip_vs_app.c

/*
 * ip_vs_app.c: Application module support for IPVS
 *
 * Authors:     Wensong Zhang <wensong@linuxvirtualserver.org>
 *
 *              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.
 *
 * Most code here is taken from ip_masq_app.c in kernel 2.2. The difference
 * is that ip_vs_app module handles the reverse direction (incoming requests
 * and outgoing responses).
 *
 *		IP_MASQ_APP application masquerading module
 *
 * Author:	Juan Jose Ciarlante, <jjciarla@raiz.uncu.edu.ar>
 *
 */

#define KMSG_COMPONENT "IPVS"
#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt

#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/skbuff.h>
#include <linux/in.h>
#include <linux/ip.h>
#include <linux/netfilter.h>
#include <linux/slab.h>
#include <net/net_namespace.h>
#include <net/protocol.h>
#include <net/tcp.h>
#include <linux/stat.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/mutex.h>

#include <net/ip_vs.h>

EXPORT_SYMBOL(register_ip_vs_app);
EXPORT_SYMBOL(unregister_ip_vs_app);
EXPORT_SYMBOL(register_ip_vs_app_inc);

static DEFINE_MUTEX(__ip_vs_app_mutex);

/*
 *	Get an ip_vs_app object
 */
static inline int ip_vs_app_get(struct ip_vs_app *app)
{
	return try_module_get(app->module);
}


static inline void ip_vs_app_put(struct ip_vs_app *app)
{
	module_put(app->module);
}


/*
 *	Allocate/initialize app incarnation and register it in proto apps.
 */
static int
ip_vs_app_inc_new(struct net *net, struct ip_vs_app *app, __u16 proto,
		  __u16 port)
{
	struct ip_vs_protocol *pp;
	struct ip_vs_app *inc;
	int ret;

	if (!(pp = ip_vs_proto_get(proto)))
		return -EPROTONOSUPPORT;

	if (!pp->unregister_app)
		return -EOPNOTSUPP;

	inc = kmemdup(app, sizeof(*inc), GFP_KERNEL);
	if (!inc)
		return -ENOMEM;
	INIT_LIST_HEAD(&inc->p_list);
	INIT_LIST_HEAD(&inc->incs_list);
	inc->app = app;
	inc->port = htons(port);
	atomic_set(&inc->usecnt, 0);

	if (app->timeouts) {
		inc->timeout_table =
			ip_vs_create_timeout_table(app->timeouts,
						   app->timeouts_size);
		if (!inc->timeout_table) {
			ret = -ENOMEM;
			goto out;
		}
	}

	ret = pp->register_app(net, inc);
	if (ret)
		goto out;

	list_add(&inc->a_list, &app->incs_list);
	IP_VS_DBG(9, "%s App %s:%u registered\n",
		  pp->name, inc->name, ntohs(inc->port));

	return 0;

  out:
	kfree(inc->timeout_table);
	kfree(inc);
	return ret;
}


/*
 *	Release app incarnation
 */
static void
ip_vs_app_inc_release(struct net *net, struct ip_vs_app *inc)
{
	struct ip_vs_protocol *pp;

	if (!(pp = ip_vs_proto_get(inc->protocol)))
		return;

	if (pp->unregister_app)
		pp->unregister_app(net, inc);

	IP_VS_DBG(9, "%s App %s:%u unregistered\n",
		  pp->name, inc->name, ntohs(inc->port));

	list_del(&inc->a_list);

	kfree(inc->timeout_table);
	kfree(inc);
}


/*
 *	Get reference to app inc (only called from softirq)
 *
 */
int ip_vs_app_inc_get(struct ip_vs_app *inc)
{
	int result;

	atomic_inc(&inc->usecnt);
	if (unlikely((result = ip_vs_app_get(inc->app)) != 1))
		atomic_dec(&inc->usecnt);
	return result;
}


/*
 *	Put the app inc (only called from timer or net softirq)
 */
void ip_vs_app_inc_put(struct ip_vs_app *inc)
{
	ip_vs_app_put(inc->app);
	atomic_dec(&inc->usecnt);
}


/*
 *	Register an application incarnation in protocol applications
 */
int
register_ip_vs_app_inc(struct net *net, struct ip_vs_app *app, __u16 proto,
		       __u16 port)
{
	int result;

	mutex_lock(&__ip_vs_app_mutex);

	result = ip_vs_app_inc_new(net, app, proto, port);

	mutex_unlock(&__ip_vs_app_mutex);

	return result;
}


/*
 *	ip_vs_app registration routine
 */
int register_ip_vs_app(struct net *net, struct ip_vs_app *app)
{
	struct netns_ipvs *ipvs = net_ipvs(net);
	/* increase the module use count */
	ip_vs_use_count_inc();

	mutex_lock(&__ip_vs_app_mutex);

	list_add(&app->a_list, &ipvs->app_list);

	mutex_unlock(&__ip_vs_app_mutex);

	return 0;
}


/*
 *	ip_vs_app unregistration routine
 *	We are sure there are no app incarnations attached to services
 */
void unregister_ip_vs_app(struct net *net, struct ip_vs_app *app)
{
	struct ip_vs_app *inc, *nxt;

	mutex_lock(&__ip_vs_app_mutex);

	list_for_each_entry_safe(inc, nxt, &app->incs_list, a_list) {
		ip_vs_app_inc_release(net, inc);
	}

	list_del(&app->a_list);

	mutex_unlock(&__ip_vs_app_mutex);

	/* decrease the module use count */
	ip_vs_use_count_dec();
}


/*
 *	Bind ip_vs_conn to its ip_vs_app (called by cp constructor)
 */
int ip_vs_bind_app(struct ip_vs_conn *cp,
		   struct ip_vs_protocol *pp)
{
	return pp->app_conn_bind(cp);
}


/*
 *	Unbind cp from application incarnation (called by cp destructor)
 */
void ip_vs_unbind_app(struct ip_vs_conn *cp)
{
	struct ip_vs_app *inc = cp->app;

	if (!inc)
		return;

	if (inc->unbind_conn)
		inc->unbind_conn(inc, cp);
	if (inc->done_conn)
		inc->done_conn(inc, cp);
	ip_vs_app_inc_put(inc);
	cp->app = NULL;
}


/*
 *	Fixes th->seq based on ip_vs_seq info.
 */
static inline void vs_fix_seq(const struct ip_vs_seq *vseq, struct tcphdr *th)
{
	__u32 seq = ntohl(th->seq);

	/*
	 *	Adjust seq with delta-offset for all packets after
	 *	the most recent resized pkt seq and with previous_delta offset
	 *	for all packets	before most recent resized pkt seq.
	 */
	if (vseq->delta || vseq->previous_delta) {
		if(after(seq, vseq->init_seq)) {
			th->seq = htonl(seq + vseq->delta);
			IP_VS_DBG(9, "%s(): added delta (%d) to seq\n",
				  __func__, vseq->delta);
		} else {
			th->seq = htonl(seq + vseq->previous_delta);
			IP_VS_DBG(9, "%s(): added previous_delta (%d) to seq\n",
				  __func__, vseq->previous_delta);
		}
	}
}


/*
 *	Fixes th->ack_seq based on ip_vs_seq info.
 */
static inline void
vs_fix_ack_seq(const struct ip_vs_seq *vseq, struct tcphdr *th)
{
	__u32 ack_seq = ntohl(th->ack_seq);

	/*
	 * Adjust ack_seq with delta-offset for
	 * the packets AFTER most recent resized pkt has caused a shift
	 * for packets before most recent resized pkt, use previous_delta
	 */
	if (vseq->delta || vseq->previous_delta) {
		/* since ack_seq is the number of octet that is expected
		   to receive next, so compare it with init_seq+delta */
		if(after(ack_seq, vseq->init_seq+vseq->delta)) {
			th->ack_seq = htonl(ack_seq - vseq->delta);
			IP_VS_DBG(9, "%s(): subtracted delta "
				  "(%d) from ack_seq\n", __func__, vseq->delta);

		} else {
			th->ack_seq = htonl(ack_seq - vseq->previous_delta);
			IP_VS_DBG(9, "%s(): subtracted "
				  "previous_delta (%d) from ack_seq\n",
				  __func__, vseq->previous_delta);
		}
	}
}


/*
 *	Updates ip_vs_seq if pkt has been resized
 *	Assumes already checked proto==IPPROTO_TCP and diff!=0.
 */
static inline void vs_seq_update(struct ip_vs_conn *cp, struct ip_vs_seq *vseq,
				 unsigned int flag, __u32 seq, int diff)
{
	/* spinlock is to keep updating cp->flags atomic */
	spin_lock(&cp->lock);
	if (!(cp->flags & flag) || after(seq, vseq->init_seq)) {
		vseq->previous_delta = vseq->delta;
		vseq->delta += diff;
		vseq->init_seq = seq;
		cp->flags |= flag;
	}
	spin_unlock(&cp->lock);
}

static inline int app_tcp_pkt_out(struct ip_vs_conn *cp, struct sk_buff *skb,
				  struct ip_vs_app *app)
{
	int diff;
	const unsigned int tcp_offset = ip_hdrlen(skb);
	struct tcphdr *th;
	__u32 seq;

	if (!skb_make_writable(skb, tcp_offset + sizeof(*th)))
		return 0;

	th = (struct tcphdr *)(skb_network_header(skb) + tcp_offset);

	/*
	 *	Remember seq number in case this pkt gets resized
	 */
	seq = ntohl(th->seq);

	/*
	 *	Fix seq stuff if flagged as so.
	 */
	if (cp->flags & IP_VS_CONN_F_OUT_SEQ)
		vs_fix_seq(&cp->out_seq, th);
	if (cp->flags & IP_VS_CONN_F_IN_SEQ)
		vs_fix_ack_seq(&cp->in_seq, th);

	/*
	 *	Call private output hook function
	 */
	if (app->pkt_out == NULL)
		return 1;

	if (!app->pkt_out(app, cp, skb, &diff))
		return 0;

	/*
	 *	Update ip_vs seq stuff if len has changed.
	 */
	if (diff != 0)
		vs_seq_update(cp, &cp->out_seq,
			      IP_VS_CONN_F_OUT_SEQ, seq, diff);

	return 1;
}

/*
 *	Output pkt hook. Will call bound ip_vs_app specific function
 *	called by ipvs packet handler, assumes previously checked cp!=NULL
 *	returns false if it can't handle packet (oom)
 */
int ip_vs_app_pkt_out(struct ip_vs_conn *cp, struct sk_buff *skb)
{
	struct ip_vs_app *app;

	/*
	 *	check if application module is bound to
	 *	this ip_vs_conn.
	 */
	if ((app = cp->app) == NULL)
		return 1;

	/* TCP is complicated */
	if (cp->protocol == IPPROTO_TCP)
		return app_tcp_pkt_out(cp, skb, app);

	/*
	 *	Call private output hook function
	 */
	if (app->pkt_out == NULL)
		return 1;

	return app->pkt_out(app, cp, skb, NULL);
}


static inline int app_tcp_pkt_in(struct ip_vs_conn *cp, struct sk_buff *skb,
				 struct ip_vs_app *app)
{
	int diff;
	const unsigned int tcp_offset = ip_hdrlen(skb);
	struct tcphdr *th;
	__u32 seq;

	if (!skb_make_writable(skb, tcp_offset + sizeof(*th)))
		return 0;

	th = (struct tcphdr *)(skb_network_header(skb) + tcp_offset);

	/*
	 *	Remember seq number in case this pkt gets resized
	 */
	seq = ntohl(th->seq);

	/*
	 *	Fix seq stuff if flagged as so.
	 */
	if (cp->flags & IP_VS_CONN_F_IN_SEQ)
		vs_fix_seq(&cp->in_seq, th);
	if (cp->flags & IP_VS_CONN_F_OUT_SEQ)
		vs_fix_ack_seq(&cp->out_seq, th);

	/*
	 *	Call private input hook function
	 */
	if (app->pkt_in == NULL)
		return 1;

	if (!app->pkt_in(app, cp, skb, &diff))
		return 0;

	/*
	 *	Update ip_vs seq stuff if len has changed.
	 */
	if (diff != 0)
		vs_seq_update(cp, &cp->in_seq,
			      IP_VS_CONN_F_IN_SEQ, seq, diff);

	return 1;
}

/*
 *	Input pkt hook. Will call bound ip_vs_app specific function
 *	called by ipvs packet handler, assumes previously checked cp!=NULL.
 *	returns false if can't handle packet (oom).
 */
int ip_vs_app_pkt_in(struct ip_vs_conn *cp, struct sk_buff *skb)
{
	struct ip_vs_app *app;

	/*
	 *	check if application module is bound to
	 *	this ip_vs_conn.
	 */
	if ((app = cp->app) == NULL)
		return 1;

	/* TCP is complicated */
	if (cp->protocol == IPPROTO_TCP)
		return app_tcp_pkt_in(cp, skb, app);

	/*
	 *	Call private input hook function
	 */
	if (app->pkt_in == NULL)
		return 1;

	return app->pkt_in(app, cp, skb, NULL);
}


#ifdef CONFIG_PROC_FS
/*
 *	/proc/net/ip_vs_app entry function
 */

static struct ip_vs_app *ip_vs_app_idx(struct netns_ipvs *ipvs, loff_t pos)
{
	struct ip_vs_app *app, *inc;

	list_for_each_entry(app, &ipvs->app_list, a_list) {
		list_for_each_entry(inc, &app->incs_list, a_list) {
			if (pos-- == 0)
				return inc;
		}
	}
	return NULL;

}

static void *ip_vs_app_seq_start(struct seq_file *seq, loff_t *pos)
{
	struct net *net = seq_file_net(seq);
	struct netns_ipvs *ipvs = net_ipvs(net);

	mutex_lock(&__ip_vs_app_mutex);

	return *pos ? ip_vs_app_idx(ipvs, *pos - 1) : SEQ_START_TOKEN;
}

static void *ip_vs_app_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	struct ip_vs_app *inc, *app;
	struct list_head *e;
	struct net *net = seq_file_net(seq);
	struct netns_ipvs *ipvs = net_ipvs(net);

	++*pos;
	if (v == SEQ_START_TOKEN)
		return ip_vs_app_idx(ipvs, 0);

	inc = v;
	app = inc->app;

	if ((e = inc->a_list.next) != &app->incs_list)
		return list_entry(e, struct ip_vs_app, a_list);

	/* go on to next application */
	for (e = app->a_list.next; e != &ipvs->app_list; e = e->next) {
		app = list_entry(e, struct ip_vs_app, a_list);
		list_for_each_entry(inc, &app->incs_list, a_list) {
			return inc;
		}
	}
	return NULL;
}

static void ip_vs_app_seq_stop(struct seq_file *seq, void *v)
{
	mutex_unlock(&__ip_vs_app_mutex);
}

static int ip_vs_app_seq_show(struct seq_file *seq, void *v)
{
	if (v == SEQ_START_TOKEN)
		seq_puts(seq, "prot port    usecnt name\n");
	else {
		const struct ip_vs_app *inc = v;

		seq_printf(seq, "%-3s  %-7u %-6d %-17s\n",
			   ip_vs_proto_name(inc->protocol),
			   ntohs(inc->port),
			   atomic_read(&inc->usecnt),
			   inc->name);
	}
	return 0;
}

static const struct seq_operations ip_vs_app_seq_ops = {
	.start = ip_vs_app_seq_start,
	.next  = ip_vs_app_seq_next,
	.stop  = ip_vs_app_seq_stop,
	.show  = ip_vs_app_seq_show,
};

static int ip_vs_app_open(struct inode *inode, struct file *file)
{
	return seq_open_net(inode, file, &ip_vs_app_seq_ops,
			    sizeof(struct seq_net_private));
}

static const struct file_operations ip_vs_app_fops = {
	.owner	 = THIS_MODULE,
	.open	 = ip_vs_app_open,
	.read	 = seq_read,
	.llseek  = seq_lseek,
	.release = seq_release_net,
};
#endif

int __net_init ip_vs_app_net_init(struct net *net)
{
	struct netns_ipvs *ipvs = net_ipvs(net);

	INIT_LIST_HEAD(&ipvs->app_list);
	proc_net_fops_create(net, "ip_vs_app", 0, &ip_vs_app_fops);
	return 0;
}

void __net_exit ip_vs_app_net_cleanup(struct net *net)
{
	proc_net_remove(net, "ip_vs_app");
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* ip_vs_app.c: Application module support for IPVS
	3	*
1da177e4 LT	4	* Authors: Wensong Zhang <wensong@linuxvirtualserver.org>
	5	*
	6	* This program is free software; you can redistribute it and/or
	7	* modify it under the terms of the GNU General Public License
	8	* as published by the Free Software Foundation; either version
	9	* 2 of the License, or (at your option) any later version.
	10	*
	11	* Most code here is taken from ip_masq_app.c in kernel 2.2. The difference
	12	* is that ip_vs_app module handles the reverse direction (incoming requests
	13	* and outgoing responses).
	14	*
	15	* IP_MASQ_APP application masquerading module
	16	*
	17	* Author: Juan Jose Ciarlante, <jjciarla@raiz.uncu.edu.ar>
	18	*
	19	*/
	20
9aada7ac HE	21	#define KMSG_COMPONENT "IPVS"
	22	#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
	23
1da177e4 LT	24	#include <linux/module.h>
	25	#include <linux/kernel.h>
	26	#include <linux/skbuff.h>
	27	#include <linux/in.h>
	28	#include <linux/ip.h>
af1e1cf0	29	#include <linux/netfilter.h>
5a0e3ad6	30	#include <linux/slab.h>
457c4cbc	31	#include <net/net_namespace.h>
1da177e4	32	#include <net/protocol.h>
c752f073	33	#include <net/tcp.h>
1da177e4 LT	34	#include <linux/stat.h>
	35	#include <linux/proc_fs.h>
	36	#include <linux/seq_file.h>
57b47a53	37	#include <linux/mutex.h>
1da177e4 LT	38
	39	#include <net/ip_vs.h>
	40
	41	EXPORT_SYMBOL(register_ip_vs_app);
	42	EXPORT_SYMBOL(unregister_ip_vs_app);
	43	EXPORT_SYMBOL(register_ip_vs_app_inc);
	44
736561a0 SH	45	static DEFINE_MUTEX(__ip_vs_app_mutex);
736561a0 SH	46
1da177e4 LT	47	/*
	48	* Get an ip_vs_app object
	49	*/
	50	static inline int ip_vs_app_get(struct ip_vs_app *app)
	51	{
85b60680	52	return try_module_get(app->module);
1da177e4 LT	53	}
	54
	55
	56	static inline void ip_vs_app_put(struct ip_vs_app *app)
	57	{
85b60680	58	module_put(app->module);
1da177e4 LT	59	}
	60
	61
	62	/*
	63	* Allocate/initialize app incarnation and register it in proto apps.
	64	*/
	65	static int
ab8a5e84 HS	66	ip_vs_app_inc_new(struct net net, struct ip_vs_app app, __u16 proto,
ab8a5e84 HS	67	__u16 port)
1da177e4 LT	68	{
	69	struct ip_vs_protocol *pp;
	70	struct ip_vs_app *inc;
	71	int ret;
	72
	73	if (!(pp = ip_vs_proto_get(proto)))
	74	return -EPROTONOSUPPORT;
	75
	76	if (!pp->unregister_app)
	77	return -EOPNOTSUPP;
	78
8b2ed4bb	79	inc = kmemdup(app, sizeof(*inc), GFP_KERNEL);
1da177e4 LT	80	if (!inc)
1da177e4 LT	81	return -ENOMEM;
1da177e4 LT	82	INIT_LIST_HEAD(&inc->p_list);
	83	INIT_LIST_HEAD(&inc->incs_list);
	84	inc->app = app;
	85	inc->port = htons(port);
	86	atomic_set(&inc->usecnt, 0);
	87
	88	if (app->timeouts) {
	89	inc->timeout_table =
	90	ip_vs_create_timeout_table(app->timeouts,
	91	app->timeouts_size);
	92	if (!inc->timeout_table) {
	93	ret = -ENOMEM;
	94	goto out;
	95	}
	96	}
	97
ab8a5e84	98	ret = pp->register_app(net, inc);
1da177e4 LT	99	if (ret)
	100	goto out;
	101
	102	list_add(&inc->a_list, &app->incs_list);
26c15cfd JA	103	IP_VS_DBG(9, "%s App %s:%u registered\n",
26c15cfd JA	104	pp->name, inc->name, ntohs(inc->port));
1da177e4 LT	105
	106	return 0;
	107
	108	out:
a51482bd	109	kfree(inc->timeout_table);
1da177e4 LT	110	kfree(inc);
	111	return ret;
	112	}
	113
	114
	115	/*
	116	* Release app incarnation
	117	*/
	118	static void
ab8a5e84	119	ip_vs_app_inc_release(struct net net, struct ip_vs_app inc)
1da177e4 LT	120	{
	121	struct ip_vs_protocol *pp;
	122
	123	if (!(pp = ip_vs_proto_get(inc->protocol)))
	124	return;
	125
	126	if (pp->unregister_app)
ab8a5e84	127	pp->unregister_app(net, inc);
1da177e4 LT	128
1da177e4 LT	129	IP_VS_DBG(9, "%s App %s:%u unregistered\n",
26c15cfd	130	pp->name, inc->name, ntohs(inc->port));
1da177e4 LT	131
	132	list_del(&inc->a_list);
	133
a51482bd	134	kfree(inc->timeout_table);
1da177e4 LT	135	kfree(inc);
	136	}
	137
	138
	139	/*
	140	* Get reference to app inc (only called from softirq)
	141	*
	142	*/
	143	int ip_vs_app_inc_get(struct ip_vs_app *inc)
	144	{
	145	int result;
	146
	147	atomic_inc(&inc->usecnt);
	148	if (unlikely((result = ip_vs_app_get(inc->app)) != 1))
	149	atomic_dec(&inc->usecnt);
	150	return result;
	151	}
	152
	153
	154	/*
	155	* Put the app inc (only called from timer or net softirq)
	156	*/
	157	void ip_vs_app_inc_put(struct ip_vs_app *inc)
	158	{
	159	ip_vs_app_put(inc->app);
	160	atomic_dec(&inc->usecnt);
	161	}
	162
	163
	164	/*
	165	* Register an application incarnation in protocol applications
	166	*/
	167	int
ab8a5e84 HS	168	register_ip_vs_app_inc(struct net net, struct ip_vs_app app, __u16 proto,
ab8a5e84 HS	169	__u16 port)
1da177e4 LT	170	{
	171	int result;
	172
736561a0	173	mutex_lock(&__ip_vs_app_mutex);
1da177e4	174
ab8a5e84	175	result = ip_vs_app_inc_new(net, app, proto, port);
1da177e4	176
736561a0	177	mutex_unlock(&__ip_vs_app_mutex);
1da177e4 LT	178
	179	return result;
	180	}
	181
	182
	183	/*
	184	* ip_vs_app registration routine
	185	*/
ab8a5e84	186	int register_ip_vs_app(struct net net, struct ip_vs_app app)
1da177e4	187	{
ab8a5e84	188	struct netns_ipvs *ipvs = net_ipvs(net);
1da177e4 LT	189	/* increase the module use count */
	190	ip_vs_use_count_inc();
	191
736561a0	192	mutex_lock(&__ip_vs_app_mutex);
1da177e4	193
ab8a5e84	194	list_add(&app->a_list, &ipvs->app_list);
1da177e4	195
736561a0	196	mutex_unlock(&__ip_vs_app_mutex);
1da177e4 LT	197
	198	return 0;
	199	}
	200
	201
	202	/*
	203	* ip_vs_app unregistration routine
	204	* We are sure there are no app incarnations attached to services
	205	*/
ab8a5e84	206	void unregister_ip_vs_app(struct net net, struct ip_vs_app app)
1da177e4 LT	207	{
	208	struct ip_vs_app inc, nxt;
	209
736561a0	210	mutex_lock(&__ip_vs_app_mutex);
1da177e4 LT	211
1da177e4 LT	212	list_for_each_entry_safe(inc, nxt, &app->incs_list, a_list) {
ab8a5e84	213	ip_vs_app_inc_release(net, inc);
1da177e4 LT	214	}
	215
	216	list_del(&app->a_list);
	217
736561a0	218	mutex_unlock(&__ip_vs_app_mutex);
1da177e4 LT	219
	220	/* decrease the module use count */
	221	ip_vs_use_count_dec();
	222	}
	223
	224
1da177e4 LT	225	/*
	226	* Bind ip_vs_conn to its ip_vs_app (called by cp constructor)
	227	*/
ab8a5e84 HS	228	int ip_vs_bind_app(struct ip_vs_conn *cp,
ab8a5e84 HS	229	struct ip_vs_protocol *pp)
1da177e4 LT	230	{
	231	return pp->app_conn_bind(cp);
	232	}
	233
	234
	235	/*
	236	* Unbind cp from application incarnation (called by cp destructor)
	237	*/
	238	void ip_vs_unbind_app(struct ip_vs_conn *cp)
	239	{
	240	struct ip_vs_app *inc = cp->app;
	241
	242	if (!inc)
	243	return;
	244
	245	if (inc->unbind_conn)
	246	inc->unbind_conn(inc, cp);
	247	if (inc->done_conn)
	248	inc->done_conn(inc, cp);
	249	ip_vs_app_inc_put(inc);
	250	cp->app = NULL;
	251	}
	252
	253
	254	/*
	255	* Fixes th->seq based on ip_vs_seq info.
	256	*/
	257	static inline void vs_fix_seq(const struct ip_vs_seq vseq, struct tcphdr th)
	258	{
	259	__u32 seq = ntohl(th->seq);
	260
	261	/*
	262	* Adjust seq with delta-offset for all packets after
	263	* the most recent resized pkt seq and with previous_delta offset
	264	* for all packets before most recent resized pkt seq.
	265	*/
	266	if (vseq->delta \|\| vseq->previous_delta) {
	267	if(after(seq, vseq->init_seq)) {
	268	th->seq = htonl(seq + vseq->delta);
1e3e238e HE	269	IP_VS_DBG(9, "%s(): added delta (%d) to seq\n",
1e3e238e HE	270	__func__, vseq->delta);
1da177e4 LT	271	} else {
1da177e4 LT	272	th->seq = htonl(seq + vseq->previous_delta);
1e3e238e HE	273	IP_VS_DBG(9, "%s(): added previous_delta (%d) to seq\n",
1e3e238e HE	274	__func__, vseq->previous_delta);
1da177e4 LT	275	}
	276	}
	277	}
	278
	279
	280	/*
	281	* Fixes th->ack_seq based on ip_vs_seq info.
	282	*/
	283	static inline void
	284	vs_fix_ack_seq(const struct ip_vs_seq vseq, struct tcphdr th)
	285	{
	286	__u32 ack_seq = ntohl(th->ack_seq);
	287
	288	/*
	289	* Adjust ack_seq with delta-offset for
	290	* the packets AFTER most recent resized pkt has caused a shift
	291	* for packets before most recent resized pkt, use previous_delta
	292	*/
	293	if (vseq->delta \|\| vseq->previous_delta) {
	294	/* since ack_seq is the number of octet that is expected
	295	to receive next, so compare it with init_seq+delta */
	296	if(after(ack_seq, vseq->init_seq+vseq->delta)) {
	297	th->ack_seq = htonl(ack_seq - vseq->delta);
1e3e238e HE	298	IP_VS_DBG(9, "%s(): subtracted delta "
1e3e238e HE	299	"(%d) from ack_seq\n", __func__, vseq->delta);
1da177e4 LT	300
	301	} else {
	302	th->ack_seq = htonl(ack_seq - vseq->previous_delta);
1e3e238e	303	IP_VS_DBG(9, "%s(): subtracted "
1da177e4	304	"previous_delta (%d) from ack_seq\n",
1e3e238e	305	__func__, vseq->previous_delta);
1da177e4 LT	306	}
	307	}
	308	}
	309
	310
	311	/*
	312	* Updates ip_vs_seq if pkt has been resized
	313	* Assumes already checked proto==IPPROTO_TCP and diff!=0.
	314	*/
	315	static inline void vs_seq_update(struct ip_vs_conn cp, struct ip_vs_seq vseq,
95c96174	316	unsigned int flag, __u32 seq, int diff)
1da177e4 LT	317	{
	318	/* spinlock is to keep updating cp->flags atomic */
	319	spin_lock(&cp->lock);
	320	if (!(cp->flags & flag) \|\| after(seq, vseq->init_seq)) {
	321	vseq->previous_delta = vseq->delta;
	322	vseq->delta += diff;
	323	vseq->init_seq = seq;
	324	cp->flags \|= flag;
	325	}
	326	spin_unlock(&cp->lock);
	327	}
	328
3db05fea	329	static inline int app_tcp_pkt_out(struct ip_vs_conn cp, struct sk_buff skb,
1da177e4 LT	330	struct ip_vs_app *app)
	331	{
	332	int diff;
3db05fea	333	const unsigned int tcp_offset = ip_hdrlen(skb);
1da177e4 LT	334	struct tcphdr *th;
	335	__u32 seq;
	336
3db05fea	337	if (!skb_make_writable(skb, tcp_offset + sizeof(*th)))
1da177e4 LT	338	return 0;
1da177e4 LT	339
3db05fea	340	th = (struct tcphdr *)(skb_network_header(skb) + tcp_offset);
1da177e4 LT	341
	342	/*
	343	* Remember seq number in case this pkt gets resized
	344	*/
	345	seq = ntohl(th->seq);
	346
	347	/*
	348	* Fix seq stuff if flagged as so.
	349	*/
	350	if (cp->flags & IP_VS_CONN_F_OUT_SEQ)
	351	vs_fix_seq(&cp->out_seq, th);
	352	if (cp->flags & IP_VS_CONN_F_IN_SEQ)
	353	vs_fix_ack_seq(&cp->in_seq, th);
	354
	355	/*
	356	* Call private output hook function
	357	*/
	358	if (app->pkt_out == NULL)
	359	return 1;
	360
3db05fea	361	if (!app->pkt_out(app, cp, skb, &diff))
1da177e4 LT	362	return 0;
	363
	364	/*
	365	* Update ip_vs seq stuff if len has changed.
	366	*/
	367	if (diff != 0)
	368	vs_seq_update(cp, &cp->out_seq,
	369	IP_VS_CONN_F_OUT_SEQ, seq, diff);
	370
	371	return 1;
	372	}
	373
	374	/*
	375	* Output pkt hook. Will call bound ip_vs_app specific function
	376	* called by ipvs packet handler, assumes previously checked cp!=NULL
	377	* returns false if it can't handle packet (oom)
	378	*/
3db05fea	379	int ip_vs_app_pkt_out(struct ip_vs_conn cp, struct sk_buff skb)
1da177e4 LT	380	{
	381	struct ip_vs_app *app;
	382
	383	/*
	384	* check if application module is bound to
	385	* this ip_vs_conn.
	386	*/
	387	if ((app = cp->app) == NULL)
	388	return 1;
	389
	390	/* TCP is complicated */
	391	if (cp->protocol == IPPROTO_TCP)
3db05fea	392	return app_tcp_pkt_out(cp, skb, app);
1da177e4 LT	393
	394	/*
	395	* Call private output hook function
	396	*/
	397	if (app->pkt_out == NULL)
	398	return 1;
	399
3db05fea	400	return app->pkt_out(app, cp, skb, NULL);
1da177e4 LT	401	}
	402
	403
3db05fea	404	static inline int app_tcp_pkt_in(struct ip_vs_conn cp, struct sk_buff skb,
1da177e4 LT	405	struct ip_vs_app *app)
	406	{
	407	int diff;
3db05fea	408	const unsigned int tcp_offset = ip_hdrlen(skb);
1da177e4 LT	409	struct tcphdr *th;
	410	__u32 seq;
	411
3db05fea	412	if (!skb_make_writable(skb, tcp_offset + sizeof(*th)))
1da177e4 LT	413	return 0;
1da177e4 LT	414
3db05fea	415	th = (struct tcphdr *)(skb_network_header(skb) + tcp_offset);
1da177e4 LT	416
	417	/*
	418	* Remember seq number in case this pkt gets resized
	419	*/
	420	seq = ntohl(th->seq);
	421
	422	/*
	423	* Fix seq stuff if flagged as so.
	424	*/
	425	if (cp->flags & IP_VS_CONN_F_IN_SEQ)
	426	vs_fix_seq(&cp->in_seq, th);
	427	if (cp->flags & IP_VS_CONN_F_OUT_SEQ)
	428	vs_fix_ack_seq(&cp->out_seq, th);
	429
	430	/*
	431	* Call private input hook function
	432	*/
	433	if (app->pkt_in == NULL)
	434	return 1;
	435
3db05fea	436	if (!app->pkt_in(app, cp, skb, &diff))
1da177e4 LT	437	return 0;
	438
	439	/*
	440	* Update ip_vs seq stuff if len has changed.
	441	*/
	442	if (diff != 0)
	443	vs_seq_update(cp, &cp->in_seq,
	444	IP_VS_CONN_F_IN_SEQ, seq, diff);
	445
	446	return 1;
	447	}
	448
	449	/*
	450	* Input pkt hook. Will call bound ip_vs_app specific function
	451	* called by ipvs packet handler, assumes previously checked cp!=NULL.
	452	* returns false if can't handle packet (oom).
	453	*/
3db05fea	454	int ip_vs_app_pkt_in(struct ip_vs_conn cp, struct sk_buff skb)
1da177e4 LT	455	{
	456	struct ip_vs_app *app;
	457
	458	/*
	459	* check if application module is bound to
	460	* this ip_vs_conn.
	461	*/
	462	if ((app = cp->app) == NULL)
	463	return 1;
	464
	465	/* TCP is complicated */
	466	if (cp->protocol == IPPROTO_TCP)
3db05fea	467	return app_tcp_pkt_in(cp, skb, app);
1da177e4 LT	468
	469	/*
	470	* Call private input hook function
	471	*/
	472	if (app->pkt_in == NULL)
	473	return 1;
	474
3db05fea	475	return app->pkt_in(app, cp, skb, NULL);
1da177e4 LT	476	}
	477
	478
	479	#ifdef CONFIG_PROC_FS
	480	/*
	481	* /proc/net/ip_vs_app entry function
	482	*/
	483
ab8a5e84	484	static struct ip_vs_app ip_vs_app_idx(struct netns_ipvs ipvs, loff_t pos)
1da177e4 LT	485	{
	486	struct ip_vs_app app, inc;
	487
ab8a5e84	488	list_for_each_entry(app, &ipvs->app_list, a_list) {
1da177e4 LT	489	list_for_each_entry(inc, &app->incs_list, a_list) {
	490	if (pos-- == 0)
	491	return inc;
	492	}
	493	}
	494	return NULL;
	495
	496	}
	497
	498	static void ip_vs_app_seq_start(struct seq_file seq, loff_t *pos)
	499	{
ab8a5e84 HS	500	struct net *net = seq_file_net(seq);
	501	struct netns_ipvs *ipvs = net_ipvs(net);
	502
736561a0	503	mutex_lock(&__ip_vs_app_mutex);
1da177e4	504
ab8a5e84	505	return pos ? ip_vs_app_idx(ipvs, pos - 1) : SEQ_START_TOKEN;
1da177e4 LT	506	}
	507
	508	static void ip_vs_app_seq_next(struct seq_file seq, void v, loff_t pos)
	509	{
	510	struct ip_vs_app inc, app;
	511	struct list_head *e;
ab8a5e84 HS	512	struct net *net = seq_file_net(seq);
ab8a5e84 HS	513	struct netns_ipvs *ipvs = net_ipvs(net);
1da177e4 LT	514
	515	++*pos;
	516	if (v == SEQ_START_TOKEN)
ab8a5e84	517	return ip_vs_app_idx(ipvs, 0);
1da177e4 LT	518
	519	inc = v;
	520	app = inc->app;
	521
	522	if ((e = inc->a_list.next) != &app->incs_list)
	523	return list_entry(e, struct ip_vs_app, a_list);
	524
	525	/* go on to next application */
ab8a5e84	526	for (e = app->a_list.next; e != &ipvs->app_list; e = e->next) {
1da177e4 LT	527	app = list_entry(e, struct ip_vs_app, a_list);
	528	list_for_each_entry(inc, &app->incs_list, a_list) {
	529	return inc;
	530	}
	531	}
	532	return NULL;
	533	}
	534
	535	static void ip_vs_app_seq_stop(struct seq_file seq, void v)
	536	{
736561a0	537	mutex_unlock(&__ip_vs_app_mutex);
1da177e4 LT	538	}
	539
	540	static int ip_vs_app_seq_show(struct seq_file seq, void v)
	541	{
	542	if (v == SEQ_START_TOKEN)
	543	seq_puts(seq, "prot port usecnt name\n");
	544	else {
	545	const struct ip_vs_app *inc = v;
	546
	547	seq_printf(seq, "%-3s %-7u %-6d %-17s\n",
	548	ip_vs_proto_name(inc->protocol),
	549	ntohs(inc->port),
	550	atomic_read(&inc->usecnt),
	551	inc->name);
	552	}
	553	return 0;
	554	}
	555
56b3d975	556	static const struct seq_operations ip_vs_app_seq_ops = {
1da177e4 LT	557	.start = ip_vs_app_seq_start,
	558	.next = ip_vs_app_seq_next,
	559	.stop = ip_vs_app_seq_stop,
	560	.show = ip_vs_app_seq_show,
	561	};
	562
	563	static int ip_vs_app_open(struct inode inode, struct file file)
	564	{
ab8a5e84 HS	565	return seq_open_net(inode, file, &ip_vs_app_seq_ops,
ab8a5e84 HS	566	sizeof(struct seq_net_private));
1da177e4 LT	567	}
1da177e4 LT	568
9a32144e	569	static const struct file_operations ip_vs_app_fops = {
1da177e4 LT	570	.owner = THIS_MODULE,
	571	.open = ip_vs_app_open,
	572	.read = seq_read,
	573	.llseek = seq_lseek,
0f08190f	574	.release = seq_release_net,
1da177e4 LT	575	};
	576	#endif
	577
503cf15a	578	int __net_init ip_vs_app_net_init(struct net *net)
1da177e4	579	{
ab8a5e84 HS	580	struct netns_ipvs *ipvs = net_ipvs(net);
ab8a5e84 HS	581
ab8a5e84	582	INIT_LIST_HEAD(&ipvs->app_list);
61b1ab45	583	proc_net_fops_create(net, "ip_vs_app", 0, &ip_vs_app_fops);
1da177e4 LT	584	return 0;
	585	}
	586
503cf15a	587	void __net_exit ip_vs_app_net_cleanup(struct net *net)
61b1ab45 HS	588	{
	589	proc_net_remove(net, "ip_vs_app");
	590	}