[net-next-2.6.git] / net / ipv6 / ip6_flowlabel.c

/*
 *	ip6_flowlabel.c		IPv6 flowlabel manager.
 *
 *	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:	Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
 */

#include <linux/capability.h>
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/net.h>
#include <linux/netdevice.h>
#include <linux/if_arp.h>
#include <linux/in6.h>
#include <linux/route.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>

#include <net/sock.h>

#include <net/ipv6.h>
#include <net/ndisc.h>
#include <net/protocol.h>
#include <net/ip6_route.h>
#include <net/addrconf.h>
#include <net/rawv6.h>
#include <net/icmp.h>
#include <net/transp_v6.h>

#include <asm/uaccess.h>

#define FL_MIN_LINGER	6	/* Minimal linger. It is set to 6sec specified
				   in old IPv6 RFC. Well, it was reasonable value.
				 */
#define FL_MAX_LINGER	60	/* Maximal linger timeout */

/* FL hash table */

#define FL_MAX_PER_SOCK	32
#define FL_MAX_SIZE	4096
#define FL_HASH_MASK	255
#define FL_HASH(l)	(ntohl(l)&FL_HASH_MASK)

static atomic_t fl_size = ATOMIC_INIT(0);
static struct ip6_flowlabel *fl_ht[FL_HASH_MASK+1];

static void ip6_fl_gc(unsigned long dummy);
static DEFINE_TIMER(ip6_fl_gc_timer, ip6_fl_gc, 0, 0);

/* FL hash table lock: it protects only of GC */

static DEFINE_RWLOCK(ip6_fl_lock);

/* Big socket sock */

static DEFINE_RWLOCK(ip6_sk_fl_lock);


static __inline__ struct ip6_flowlabel * __fl_lookup(__be32 label)
{
	struct ip6_flowlabel *fl;

	for (fl=fl_ht[FL_HASH(label)]; fl; fl = fl->next) {
		if (fl->label == label)
			return fl;
	}
	return NULL;
}

static struct ip6_flowlabel * fl_lookup(__be32 label)
{
	struct ip6_flowlabel *fl;

	read_lock_bh(&ip6_fl_lock);
	fl = __fl_lookup(label);
	if (fl)
		atomic_inc(&fl->users);
	read_unlock_bh(&ip6_fl_lock);
	return fl;
}


static void fl_free(struct ip6_flowlabel *fl)
{
	if (fl)
		kfree(fl->opt);
	kfree(fl);
}

static void fl_release(struct ip6_flowlabel *fl)
{
	write_lock_bh(&ip6_fl_lock);

	fl->lastuse = jiffies;
	if (atomic_dec_and_test(&fl->users)) {
		unsigned long ttd = fl->lastuse + fl->linger;
		if (time_after(ttd, fl->expires))
			fl->expires = ttd;
		ttd = fl->expires;
		if (fl->opt && fl->share == IPV6_FL_S_EXCL) {
			struct ipv6_txoptions *opt = fl->opt;
			fl->opt = NULL;
			kfree(opt);
		}
		if (!timer_pending(&ip6_fl_gc_timer) ||
		    time_after(ip6_fl_gc_timer.expires, ttd))
			mod_timer(&ip6_fl_gc_timer, ttd);
	}

	write_unlock_bh(&ip6_fl_lock);
}

static void ip6_fl_gc(unsigned long dummy)
{
	int i;
	unsigned long now = jiffies;
	unsigned long sched = 0;

	write_lock(&ip6_fl_lock);

	for (i=0; i<=FL_HASH_MASK; i++) {
		struct ip6_flowlabel *fl, **flp;
		flp = &fl_ht[i];
		while ((fl=*flp) != NULL) {
			if (atomic_read(&fl->users) == 0) {
				unsigned long ttd = fl->lastuse + fl->linger;
				if (time_after(ttd, fl->expires))
					fl->expires = ttd;
				ttd = fl->expires;
				if (time_after_eq(now, ttd)) {
					*flp = fl->next;
					fl_free(fl);
					atomic_dec(&fl_size);
					continue;
				}
				if (!sched || time_before(ttd, sched))
					sched = ttd;
			}
			flp = &fl->next;
		}
	}
	if (!sched && atomic_read(&fl_size))
		sched = now + FL_MAX_LINGER;
	if (sched) {
		ip6_fl_gc_timer.expires = sched;
		add_timer(&ip6_fl_gc_timer);
	}
	write_unlock(&ip6_fl_lock);
}

static int fl_intern(struct ip6_flowlabel *fl, __be32 label)
{
	fl->label = label & IPV6_FLOWLABEL_MASK;

	write_lock_bh(&ip6_fl_lock);
	if (label == 0) {
		for (;;) {
			fl->label = htonl(net_random())&IPV6_FLOWLABEL_MASK;
			if (fl->label) {
				struct ip6_flowlabel *lfl;
				lfl = __fl_lookup(fl->label);
				if (lfl == NULL)
					break;
			}
		}
	}

	fl->lastuse = jiffies;
	fl->next = fl_ht[FL_HASH(fl->label)];
	fl_ht[FL_HASH(fl->label)] = fl;
	atomic_inc(&fl_size);
	write_unlock_bh(&ip6_fl_lock);
	return 0;
}


/* Socket flowlabel lists */

struct ip6_flowlabel * fl6_sock_lookup(struct sock *sk, __be32 label)
{
	struct ipv6_fl_socklist *sfl;
	struct ipv6_pinfo *np = inet6_sk(sk);

	label &= IPV6_FLOWLABEL_MASK;

	for (sfl=np->ipv6_fl_list; sfl; sfl = sfl->next) {
		struct ip6_flowlabel *fl = sfl->fl;
		if (fl->label == label) {
			fl->lastuse = jiffies;
			atomic_inc(&fl->users);
			return fl;
		}
	}
	return NULL;
}

EXPORT_SYMBOL_GPL(fl6_sock_lookup);

void fl6_free_socklist(struct sock *sk)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct ipv6_fl_socklist *sfl;

	while ((sfl = np->ipv6_fl_list) != NULL) {
		np->ipv6_fl_list = sfl->next;
		fl_release(sfl->fl);
		kfree(sfl);
	}
}

/* Service routines */


/*
   It is the only difficult place. flowlabel enforces equal headers
   before and including routing header, however user may supply options
   following rthdr.
 */

struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions * opt_space,
					 struct ip6_flowlabel * fl,
					 struct ipv6_txoptions * fopt)
{
	struct ipv6_txoptions * fl_opt = fl->opt;

	if (fopt == NULL || fopt->opt_flen == 0)
		return fl_opt;

	if (fl_opt != NULL) {
		opt_space->hopopt = fl_opt->hopopt;
		opt_space->dst0opt = fl_opt->dst0opt;
		opt_space->srcrt = fl_opt->srcrt;
		opt_space->opt_nflen = fl_opt->opt_nflen;
	} else {
		if (fopt->opt_nflen == 0)
			return fopt;
		opt_space->hopopt = NULL;
		opt_space->dst0opt = NULL;
		opt_space->srcrt = NULL;
		opt_space->opt_nflen = 0;
	}
	opt_space->dst1opt = fopt->dst1opt;
	opt_space->opt_flen = fopt->opt_flen;
	return opt_space;
}

static unsigned long check_linger(unsigned long ttl)
{
	if (ttl < FL_MIN_LINGER)
		return FL_MIN_LINGER*HZ;
	if (ttl > FL_MAX_LINGER && !capable(CAP_NET_ADMIN))
		return 0;
	return ttl*HZ;
}

static int fl6_renew(struct ip6_flowlabel *fl, unsigned long linger, unsigned long expires)
{
	linger = check_linger(linger);
	if (!linger)
		return -EPERM;
	expires = check_linger(expires);
	if (!expires)
		return -EPERM;
	fl->lastuse = jiffies;
	if (time_before(fl->linger, linger))
		fl->linger = linger;
	if (time_before(expires, fl->linger))
		expires = fl->linger;
	if (time_before(fl->expires, fl->lastuse + expires))
		fl->expires = fl->lastuse + expires;
	return 0;
}

static struct ip6_flowlabel *
fl_create(struct in6_flowlabel_req *freq, char __user *optval, int optlen, int *err_p)
{
	struct ip6_flowlabel *fl;
	int olen;
	int addr_type;
	int err;

	err = -ENOMEM;
	fl = kzalloc(sizeof(*fl), GFP_KERNEL);
	if (fl == NULL)
		goto done;

	olen = optlen - CMSG_ALIGN(sizeof(*freq));
	if (olen > 0) {
		struct msghdr msg;
		struct flowi flowi;
		int junk;

		err = -ENOMEM;
		fl->opt = kmalloc(sizeof(*fl->opt) + olen, GFP_KERNEL);
		if (fl->opt == NULL)
			goto done;

		memset(fl->opt, 0, sizeof(*fl->opt));
		fl->opt->tot_len = sizeof(*fl->opt) + olen;
		err = -EFAULT;
		if (copy_from_user(fl->opt+1, optval+CMSG_ALIGN(sizeof(*freq)), olen))
			goto done;

		msg.msg_controllen = olen;
		msg.msg_control = (void*)(fl->opt+1);
		flowi.oif = 0;

		err = datagram_send_ctl(&msg, &flowi, fl->opt, &junk, &junk);
		if (err)
			goto done;
		err = -EINVAL;
		if (fl->opt->opt_flen)
			goto done;
		if (fl->opt->opt_nflen == 0) {
			kfree(fl->opt);
			fl->opt = NULL;
		}
	}

	fl->expires = jiffies;
	err = fl6_renew(fl, freq->flr_linger, freq->flr_expires);
	if (err)
		goto done;
	fl->share = freq->flr_share;
	addr_type = ipv6_addr_type(&freq->flr_dst);
	if ((addr_type&IPV6_ADDR_MAPPED)
	    || addr_type == IPV6_ADDR_ANY) {
		err = -EINVAL;
		goto done;
	}
	ipv6_addr_copy(&fl->dst, &freq->flr_dst);
	atomic_set(&fl->users, 1);
	switch (fl->share) {
	case IPV6_FL_S_EXCL:
	case IPV6_FL_S_ANY:
		break;
	case IPV6_FL_S_PROCESS:
		fl->owner = current->pid;
		break;
	case IPV6_FL_S_USER:
		fl->owner = current->euid;
		break;
	default:
		err = -EINVAL;
		goto done;
	}
	return fl;

done:
	fl_free(fl);
	*err_p = err;
	return NULL;
}

static int mem_check(struct sock *sk)
{
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct ipv6_fl_socklist *sfl;
	int room = FL_MAX_SIZE - atomic_read(&fl_size);
	int count = 0;

	if (room > FL_MAX_SIZE - FL_MAX_PER_SOCK)
		return 0;

	for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next)
		count++;

	if (room <= 0 ||
	    ((count >= FL_MAX_PER_SOCK ||
	     (count > 0 && room < FL_MAX_SIZE/2) || room < FL_MAX_SIZE/4)
	     && !capable(CAP_NET_ADMIN)))
		return -ENOBUFS;

	return 0;
}

static int ipv6_hdr_cmp(struct ipv6_opt_hdr *h1, struct ipv6_opt_hdr *h2)
{
	if (h1 == h2)
		return 0;
	if (h1 == NULL || h2 == NULL)
		return 1;
	if (h1->hdrlen != h2->hdrlen)
		return 1;
	return memcmp(h1+1, h2+1, ((h1->hdrlen+1)<<3) - sizeof(*h1));
}

static int ipv6_opt_cmp(struct ipv6_txoptions *o1, struct ipv6_txoptions *o2)
{
	if (o1 == o2)
		return 0;
	if (o1 == NULL || o2 == NULL)
		return 1;
	if (o1->opt_nflen != o2->opt_nflen)
		return 1;
	if (ipv6_hdr_cmp(o1->hopopt, o2->hopopt))
		return 1;
	if (ipv6_hdr_cmp(o1->dst0opt, o2->dst0opt))
		return 1;
	if (ipv6_hdr_cmp((struct ipv6_opt_hdr *)o1->srcrt, (struct ipv6_opt_hdr *)o2->srcrt))
		return 1;
	return 0;
}

int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen)
{
	int err;
	struct ipv6_pinfo *np = inet6_sk(sk);
	struct in6_flowlabel_req freq;
	struct ipv6_fl_socklist *sfl1=NULL;
	struct ipv6_fl_socklist *sfl, **sflp;
	struct ip6_flowlabel *fl;

	if (optlen < sizeof(freq))
		return -EINVAL;

	if (copy_from_user(&freq, optval, sizeof(freq)))
		return -EFAULT;

	switch (freq.flr_action) {
	case IPV6_FL_A_PUT:
		write_lock_bh(&ip6_sk_fl_lock);
		for (sflp = &np->ipv6_fl_list; (sfl=*sflp)!=NULL; sflp = &sfl->next) {
			if (sfl->fl->label == freq.flr_label) {
				if (freq.flr_label == (np->flow_label&IPV6_FLOWLABEL_MASK))
					np->flow_label &= ~IPV6_FLOWLABEL_MASK;
				*sflp = sfl->next;
				write_unlock_bh(&ip6_sk_fl_lock);
				fl_release(sfl->fl);
				kfree(sfl);
				return 0;
			}
		}
		write_unlock_bh(&ip6_sk_fl_lock);
		return -ESRCH;

	case IPV6_FL_A_RENEW:
		read_lock_bh(&ip6_sk_fl_lock);
		for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) {
			if (sfl->fl->label == freq.flr_label) {
				err = fl6_renew(sfl->fl, freq.flr_linger, freq.flr_expires);
				read_unlock_bh(&ip6_sk_fl_lock);
				return err;
			}
		}
		read_unlock_bh(&ip6_sk_fl_lock);

		if (freq.flr_share == IPV6_FL_S_NONE && capable(CAP_NET_ADMIN)) {
			fl = fl_lookup(freq.flr_label);
			if (fl) {
				err = fl6_renew(fl, freq.flr_linger, freq.flr_expires);
				fl_release(fl);
				return err;
			}
		}
		return -ESRCH;

	case IPV6_FL_A_GET:
		if (freq.flr_label & ~IPV6_FLOWLABEL_MASK)
			return -EINVAL;

		fl = fl_create(&freq, optval, optlen, &err);
		if (fl == NULL)
			return err;
		sfl1 = kmalloc(sizeof(*sfl1), GFP_KERNEL);

		if (freq.flr_label) {
			struct ip6_flowlabel *fl1 = NULL;

			err = -EEXIST;
			read_lock_bh(&ip6_sk_fl_lock);
			for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) {
				if (sfl->fl->label == freq.flr_label) {
					if (freq.flr_flags&IPV6_FL_F_EXCL) {
						read_unlock_bh(&ip6_sk_fl_lock);
						goto done;
					}
					fl1 = sfl->fl;
					atomic_inc(&fl1->users);
					break;
				}
			}
			read_unlock_bh(&ip6_sk_fl_lock);

			if (fl1 == NULL)
				fl1 = fl_lookup(freq.flr_label);
			if (fl1) {
				err = -EEXIST;
				if (freq.flr_flags&IPV6_FL_F_EXCL)
					goto release;
				err = -EPERM;
				if (fl1->share == IPV6_FL_S_EXCL ||
				    fl1->share != fl->share ||
				    fl1->owner != fl->owner)
					goto release;

				err = -EINVAL;
				if (!ipv6_addr_equal(&fl1->dst, &fl->dst) ||
				    ipv6_opt_cmp(fl1->opt, fl->opt))
					goto release;

				err = -ENOMEM;
				if (sfl1 == NULL)
					goto release;
				if (fl->linger > fl1->linger)
					fl1->linger = fl->linger;
				if ((long)(fl->expires - fl1->expires) > 0)
					fl1->expires = fl->expires;
				write_lock_bh(&ip6_sk_fl_lock);
				sfl1->fl = fl1;
				sfl1->next = np->ipv6_fl_list;
				np->ipv6_fl_list = sfl1;
				write_unlock_bh(&ip6_sk_fl_lock);
				fl_free(fl);
				return 0;

release:
				fl_release(fl1);
				goto done;
			}
		}
		err = -ENOENT;
		if (!(freq.flr_flags&IPV6_FL_F_CREATE))
			goto done;

		err = -ENOMEM;
		if (sfl1 == NULL || (err = mem_check(sk)) != 0)
			goto done;

		err = fl_intern(fl, freq.flr_label);
		if (err)
			goto done;

		if (!freq.flr_label) {
			if (copy_to_user(&((struct in6_flowlabel_req __user *) optval)->flr_label,
					 &fl->label, sizeof(fl->label))) {
				/* Intentionally ignore fault. */
			}
		}

		sfl1->fl = fl;
		sfl1->next = np->ipv6_fl_list;
		np->ipv6_fl_list = sfl1;
		return 0;

	default:
		return -EINVAL;
	}

done:
	fl_free(fl);
	kfree(sfl1);
	return err;
}

#ifdef CONFIG_PROC_FS

struct ip6fl_iter_state {
	int bucket;
};

#define ip6fl_seq_private(seq)	((struct ip6fl_iter_state *)(seq)->private)

static struct ip6_flowlabel *ip6fl_get_first(struct seq_file *seq)
{
	struct ip6_flowlabel *fl = NULL;
	struct ip6fl_iter_state *state = ip6fl_seq_private(seq);

	for (state->bucket = 0; state->bucket <= FL_HASH_MASK; ++state->bucket) {
		if (fl_ht[state->bucket]) {
			fl = fl_ht[state->bucket];
			break;
		}
	}
	return fl;
}

static struct ip6_flowlabel *ip6fl_get_next(struct seq_file *seq, struct ip6_flowlabel *fl)
{
	struct ip6fl_iter_state *state = ip6fl_seq_private(seq);

	fl = fl->next;
	while (!fl) {
		if (++state->bucket <= FL_HASH_MASK)
			fl = fl_ht[state->bucket];
		else
			break;
	}
	return fl;
}

static struct ip6_flowlabel *ip6fl_get_idx(struct seq_file *seq, loff_t pos)
{
	struct ip6_flowlabel *fl = ip6fl_get_first(seq);
	if (fl)
		while (pos && (fl = ip6fl_get_next(seq, fl)) != NULL)
			--pos;
	return pos ? NULL : fl;
}

static void *ip6fl_seq_start(struct seq_file *seq, loff_t *pos)
{
	read_lock_bh(&ip6_fl_lock);
	return *pos ? ip6fl_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
}

static void *ip6fl_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
	struct ip6_flowlabel *fl;

	if (v == SEQ_START_TOKEN)
		fl = ip6fl_get_first(seq);
	else
		fl = ip6fl_get_next(seq, v);
	++*pos;
	return fl;
}

static void ip6fl_seq_stop(struct seq_file *seq, void *v)
{
	read_unlock_bh(&ip6_fl_lock);
}

static int ip6fl_seq_show(struct seq_file *seq, void *v)
{
	if (v == SEQ_START_TOKEN)
		seq_printf(seq, "%-5s %-1s %-6s %-6s %-6s %-8s %-32s %s\n",
			   "Label", "S", "Owner", "Users", "Linger", "Expires", "Dst", "Opt");
	else {
		struct ip6_flowlabel *fl = v;
		seq_printf(seq,
			   "%05X %-1d %-6d %-6d %-6ld %-8ld " NIP6_SEQFMT " %-4d\n",
			   (unsigned)ntohl(fl->label),
			   fl->share,
			   (unsigned)fl->owner,
			   atomic_read(&fl->users),
			   fl->linger/HZ,
			   (long)(fl->expires - jiffies)/HZ,
			   NIP6(fl->dst),
			   fl->opt ? fl->opt->opt_nflen : 0);
	}
	return 0;
}

static struct seq_operations ip6fl_seq_ops = {
	.start	=	ip6fl_seq_start,
	.next	=	ip6fl_seq_next,
	.stop	=	ip6fl_seq_stop,
	.show	=	ip6fl_seq_show,
};

static int ip6fl_seq_open(struct inode *inode, struct file *file)
{
	struct seq_file *seq;
	int rc = -ENOMEM;
	struct ip6fl_iter_state *s = kzalloc(sizeof(*s), GFP_KERNEL);

	if (!s)
		goto out;

	rc = seq_open(file, &ip6fl_seq_ops);
	if (rc)
		goto out_kfree;

	seq = file->private_data;
	seq->private = s;
out:
	return rc;
out_kfree:
	kfree(s);
	goto out;
}

static const struct file_operations ip6fl_seq_fops = {
	.owner		=	THIS_MODULE,
	.open		=	ip6fl_seq_open,
	.read		=	seq_read,
	.llseek		=	seq_lseek,
	.release	=	seq_release_private,
};
#endif


void ip6_flowlabel_init(void)
{
#ifdef CONFIG_PROC_FS
	proc_net_fops_create("ip6_flowlabel", S_IRUGO, &ip6fl_seq_fops);
#endif
}

void ip6_flowlabel_cleanup(void)
{
	del_timer(&ip6_fl_gc_timer);
#ifdef CONFIG_PROC_FS
	proc_net_remove("ip6_flowlabel");
#endif
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* ip6_flowlabel.c IPv6 flowlabel manager.
	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: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
	10	*/
	11
4fc268d2	12	#include <linux/capability.h>
1da177e4 LT	13	#include <linux/errno.h>
	14	#include <linux/types.h>
	15	#include <linux/socket.h>
	16	#include <linux/net.h>
	17	#include <linux/netdevice.h>
	18	#include <linux/if_arp.h>
	19	#include <linux/in6.h>
	20	#include <linux/route.h>
	21	#include <linux/proc_fs.h>
	22	#include <linux/seq_file.h>
	23
	24	#include <net/sock.h>
	25
	26	#include <net/ipv6.h>
	27	#include <net/ndisc.h>
	28	#include <net/protocol.h>
	29	#include <net/ip6_route.h>
	30	#include <net/addrconf.h>
	31	#include <net/rawv6.h>
	32	#include <net/icmp.h>
	33	#include <net/transp_v6.h>
	34
	35	#include <asm/uaccess.h>
	36
	37	#define FL_MIN_LINGER 6 /* Minimal linger. It is set to 6sec specified
	38	in old IPv6 RFC. Well, it was reasonable value.
	39	*/
	40	#define FL_MAX_LINGER 60 /* Maximal linger timeout */
	41
	42	/* FL hash table */
	43
	44	#define FL_MAX_PER_SOCK 32
	45	#define FL_MAX_SIZE 4096
	46	#define FL_HASH_MASK 255
	47	#define FL_HASH(l) (ntohl(l)&FL_HASH_MASK)
	48
	49	static atomic_t fl_size = ATOMIC_INIT(0);
	50	static struct ip6_flowlabel *fl_ht[FL_HASH_MASK+1];
	51
	52	static void ip6_fl_gc(unsigned long dummy);
8d06afab	53	static DEFINE_TIMER(ip6_fl_gc_timer, ip6_fl_gc, 0, 0);
1da177e4 LT	54
	55	/* FL hash table lock: it protects only of GC */
	56
	57	static DEFINE_RWLOCK(ip6_fl_lock);
	58
	59	/* Big socket sock */
	60
	61	static DEFINE_RWLOCK(ip6_sk_fl_lock);
	62
	63
90bcaf7b	64	static __inline__ struct ip6_flowlabel * __fl_lookup(__be32 label)
1da177e4 LT	65	{
	66	struct ip6_flowlabel *fl;
	67
	68	for (fl=fl_ht[FL_HASH(label)]; fl; fl = fl->next) {
	69	if (fl->label == label)
	70	return fl;
	71	}
	72	return NULL;
	73	}
	74
90bcaf7b	75	static struct ip6_flowlabel * fl_lookup(__be32 label)
1da177e4 LT	76	{
	77	struct ip6_flowlabel *fl;
	78
	79	read_lock_bh(&ip6_fl_lock);
	80	fl = __fl_lookup(label);
	81	if (fl)
	82	atomic_inc(&fl->users);
	83	read_unlock_bh(&ip6_fl_lock);
	84	return fl;
	85	}
	86
	87
	88	static void fl_free(struct ip6_flowlabel *fl)
	89	{
	90	if (fl)
	91	kfree(fl->opt);
	92	kfree(fl);
	93	}
	94
	95	static void fl_release(struct ip6_flowlabel *fl)
	96	{
	97	write_lock_bh(&ip6_fl_lock);
	98
	99	fl->lastuse = jiffies;
	100	if (atomic_dec_and_test(&fl->users)) {
	101	unsigned long ttd = fl->lastuse + fl->linger;
	102	if (time_after(ttd, fl->expires))
	103	fl->expires = ttd;
	104	ttd = fl->expires;
	105	if (fl->opt && fl->share == IPV6_FL_S_EXCL) {
	106	struct ipv6_txoptions *opt = fl->opt;
	107	fl->opt = NULL;
	108	kfree(opt);
	109	}
	110	if (!timer_pending(&ip6_fl_gc_timer) \|\|
	111	time_after(ip6_fl_gc_timer.expires, ttd))
	112	mod_timer(&ip6_fl_gc_timer, ttd);
	113	}
	114
	115	write_unlock_bh(&ip6_fl_lock);
	116	}
	117
	118	static void ip6_fl_gc(unsigned long dummy)
	119	{
	120	int i;
	121	unsigned long now = jiffies;
	122	unsigned long sched = 0;
	123
	124	write_lock(&ip6_fl_lock);
	125
	126	for (i=0; i<=FL_HASH_MASK; i++) {
	127	struct ip6_flowlabel fl, *flp;
	128	flp = &fl_ht[i];
	129	while ((fl=*flp) != NULL) {
	130	if (atomic_read(&fl->users) == 0) {
	131	unsigned long ttd = fl->lastuse + fl->linger;
	132	if (time_after(ttd, fl->expires))
	133	fl->expires = ttd;
	134	ttd = fl->expires;
	135	if (time_after_eq(now, ttd)) {
	136	*flp = fl->next;
	137	fl_free(fl);
	138	atomic_dec(&fl_size);
	139	continue;
140	}
141	if (!sched \|\| time_before(ttd, sched))
142	sched = ttd;
143	}
144	flp = &fl->next;
145	}
146	}
147	if (!sched && atomic_read(&fl_size))
148	sched = now + FL_MAX_LINGER;
149	if (sched) {
150	ip6_fl_gc_timer.expires = sched;
151	add_timer(&ip6_fl_gc_timer);
152	}
153	write_unlock(&ip6_fl_lock);
154	}
155
90bcaf7b	156	static int fl_intern(struct ip6_flowlabel *fl, __be32 label)
1da177e4 LT	157	{
	158	fl->label = label & IPV6_FLOWLABEL_MASK;
	159
	160	write_lock_bh(&ip6_fl_lock);
	161	if (label == 0) {
	162	for (;;) {
	163	fl->label = htonl(net_random())&IPV6_FLOWLABEL_MASK;
	164	if (fl->label) {
	165	struct ip6_flowlabel *lfl;
	166	lfl = __fl_lookup(fl->label);
	167	if (lfl == NULL)
	168	break;
	169	}
	170	}
	171	}
	172
	173	fl->lastuse = jiffies;
	174	fl->next = fl_ht[FL_HASH(fl->label)];
	175	fl_ht[FL_HASH(fl->label)] = fl;
	176	atomic_inc(&fl_size);
	177	write_unlock_bh(&ip6_fl_lock);
	178	return 0;
	179	}
	180
	181
	182
	183	/* Socket flowlabel lists */
	184
90bcaf7b	185	struct ip6_flowlabel * fl6_sock_lookup(struct sock *sk, __be32 label)
1da177e4 LT	186	{
	187	struct ipv6_fl_socklist *sfl;
	188	struct ipv6_pinfo *np = inet6_sk(sk);
	189
	190	label &= IPV6_FLOWLABEL_MASK;
	191
	192	for (sfl=np->ipv6_fl_list; sfl; sfl = sfl->next) {
	193	struct ip6_flowlabel *fl = sfl->fl;
	194	if (fl->label == label) {
	195	fl->lastuse = jiffies;
	196	atomic_inc(&fl->users);
	197	return fl;
	198	}
	199	}
	200	return NULL;
	201	}
	202
3cf3dc6c ACM	203	EXPORT_SYMBOL_GPL(fl6_sock_lookup);
3cf3dc6c ACM	204
1da177e4 LT	205	void fl6_free_socklist(struct sock *sk)
	206	{
	207	struct ipv6_pinfo *np = inet6_sk(sk);
	208	struct ipv6_fl_socklist *sfl;
	209
	210	while ((sfl = np->ipv6_fl_list) != NULL) {
	211	np->ipv6_fl_list = sfl->next;
	212	fl_release(sfl->fl);
	213	kfree(sfl);
	214	}
	215	}
	216
	217	/* Service routines */
	218
	219
	220	/*
	221	It is the only difficult place. flowlabel enforces equal headers
	222	before and including routing header, however user may supply options
	223	following rthdr.
	224	*/
	225
	226	struct ipv6_txoptions fl6_merge_options(struct ipv6_txoptions opt_space,
	227	struct ip6_flowlabel * fl,
	228	struct ipv6_txoptions * fopt)
	229	{
df9890c3	230	struct ipv6_txoptions * fl_opt = fl->opt;
1ab1457c	231
df9890c3 YH	232	if (fopt == NULL \|\| fopt->opt_flen == 0)
df9890c3 YH	233	return fl_opt;
1ab1457c	234
1da177e4 LT	235	if (fl_opt != NULL) {
1da177e4 LT	236	opt_space->hopopt = fl_opt->hopopt;
df9890c3	237	opt_space->dst0opt = fl_opt->dst0opt;
1da177e4 LT	238	opt_space->srcrt = fl_opt->srcrt;
	239	opt_space->opt_nflen = fl_opt->opt_nflen;
	240	} else {
	241	if (fopt->opt_nflen == 0)
	242	return fopt;
	243	opt_space->hopopt = NULL;
	244	opt_space->dst0opt = NULL;
	245	opt_space->srcrt = NULL;
	246	opt_space->opt_nflen = 0;
	247	}
	248	opt_space->dst1opt = fopt->dst1opt;
1da177e4 LT	249	opt_space->opt_flen = fopt->opt_flen;
	250	return opt_space;
	251	}
	252
	253	static unsigned long check_linger(unsigned long ttl)
	254	{
	255	if (ttl < FL_MIN_LINGER)
	256	return FL_MIN_LINGER*HZ;
	257	if (ttl > FL_MAX_LINGER && !capable(CAP_NET_ADMIN))
	258	return 0;
	259	return ttl*HZ;
	260	}
	261
	262	static int fl6_renew(struct ip6_flowlabel *fl, unsigned long linger, unsigned long expires)
	263	{
	264	linger = check_linger(linger);
	265	if (!linger)
	266	return -EPERM;
	267	expires = check_linger(expires);
	268	if (!expires)
	269	return -EPERM;
	270	fl->lastuse = jiffies;
	271	if (time_before(fl->linger, linger))
	272	fl->linger = linger;
	273	if (time_before(expires, fl->linger))
	274	expires = fl->linger;
	275	if (time_before(fl->expires, fl->lastuse + expires))
	276	fl->expires = fl->lastuse + expires;
	277	return 0;
	278	}
	279
	280	static struct ip6_flowlabel *
	281	fl_create(struct in6_flowlabel_req freq, char __user optval, int optlen, int *err_p)
	282	{
	283	struct ip6_flowlabel *fl;
	284	int olen;
	285	int addr_type;
	286	int err;
	287
	288	err = -ENOMEM;
0c600eda	289	fl = kzalloc(sizeof(*fl), GFP_KERNEL);
1da177e4 LT	290	if (fl == NULL)
1da177e4 LT	291	goto done;
1da177e4 LT	292
	293	olen = optlen - CMSG_ALIGN(sizeof(*freq));
	294	if (olen > 0) {
	295	struct msghdr msg;
	296	struct flowi flowi;
	297	int junk;
	298
	299	err = -ENOMEM;
	300	fl->opt = kmalloc(sizeof(*fl->opt) + olen, GFP_KERNEL);
	301	if (fl->opt == NULL)
	302	goto done;
	303
	304	memset(fl->opt, 0, sizeof(*fl->opt));
	305	fl->opt->tot_len = sizeof(*fl->opt) + olen;
	306	err = -EFAULT;
	307	if (copy_from_user(fl->opt+1, optval+CMSG_ALIGN(sizeof(*freq)), olen))
	308	goto done;
	309
	310	msg.msg_controllen = olen;
	311	msg.msg_control = (void*)(fl->opt+1);
	312	flowi.oif = 0;
	313
41a1f8ea	314	err = datagram_send_ctl(&msg, &flowi, fl->opt, &junk, &junk);
1da177e4 LT	315	if (err)
	316	goto done;
	317	err = -EINVAL;
	318	if (fl->opt->opt_flen)
	319	goto done;
	320	if (fl->opt->opt_nflen == 0) {
	321	kfree(fl->opt);
	322	fl->opt = NULL;
	323	}
	324	}
	325
	326	fl->expires = jiffies;
	327	err = fl6_renew(fl, freq->flr_linger, freq->flr_expires);
	328	if (err)
	329	goto done;
	330	fl->share = freq->flr_share;
	331	addr_type = ipv6_addr_type(&freq->flr_dst);
	332	if ((addr_type&IPV6_ADDR_MAPPED)
c6817e4c JM	333	\|\| addr_type == IPV6_ADDR_ANY) {
c6817e4c JM	334	err = -EINVAL;
1da177e4	335	goto done;
c6817e4c	336	}
1da177e4 LT	337	ipv6_addr_copy(&fl->dst, &freq->flr_dst);
	338	atomic_set(&fl->users, 1);
	339	switch (fl->share) {
	340	case IPV6_FL_S_EXCL:
	341	case IPV6_FL_S_ANY:
	342	break;
	343	case IPV6_FL_S_PROCESS:
	344	fl->owner = current->pid;
	345	break;
	346	case IPV6_FL_S_USER:
	347	fl->owner = current->euid;
	348	break;
	349	default:
	350	err = -EINVAL;
	351	goto done;
	352	}
	353	return fl;
	354
	355	done:
	356	fl_free(fl);
	357	*err_p = err;
	358	return NULL;
	359	}
	360
	361	static int mem_check(struct sock *sk)
	362	{
	363	struct ipv6_pinfo *np = inet6_sk(sk);
	364	struct ipv6_fl_socklist *sfl;
	365	int room = FL_MAX_SIZE - atomic_read(&fl_size);
	366	int count = 0;
	367
	368	if (room > FL_MAX_SIZE - FL_MAX_PER_SOCK)
	369	return 0;
	370
	371	for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next)
	372	count++;
	373
	374	if (room <= 0 \|\|
	375	((count >= FL_MAX_PER_SOCK \|\|
	376	(count > 0 && room < FL_MAX_SIZE/2) \|\| room < FL_MAX_SIZE/4)
	377	&& !capable(CAP_NET_ADMIN)))
	378	return -ENOBUFS;
	379
	380	return 0;
	381	}
	382
	383	static int ipv6_hdr_cmp(struct ipv6_opt_hdr h1, struct ipv6_opt_hdr h2)
	384	{
	385	if (h1 == h2)
	386	return 0;
	387	if (h1 == NULL \|\| h2 == NULL)
	388	return 1;
	389	if (h1->hdrlen != h2->hdrlen)
	390	return 1;
	391	return memcmp(h1+1, h2+1, ((h1->hdrlen+1)<<3) - sizeof(*h1));
	392	}
	393
	394	static int ipv6_opt_cmp(struct ipv6_txoptions o1, struct ipv6_txoptions o2)
	395	{
	396	if (o1 == o2)
	397	return 0;
	398	if (o1 == NULL \|\| o2 == NULL)
	399	return 1;
	400	if (o1->opt_nflen != o2->opt_nflen)
401	return 1;
402	if (ipv6_hdr_cmp(o1->hopopt, o2->hopopt))
403	return 1;
404	if (ipv6_hdr_cmp(o1->dst0opt, o2->dst0opt))
405	return 1;
406	if (ipv6_hdr_cmp((struct ipv6_opt_hdr )o1->srcrt, (struct ipv6_opt_hdr )o2->srcrt))
407	return 1;
408	return 0;
409	}
410
411	int ipv6_flowlabel_opt(struct sock sk, char __user optval, int optlen)
412	{
413	int err;
414	struct ipv6_pinfo *np = inet6_sk(sk);
415	struct in6_flowlabel_req freq;
416	struct ipv6_fl_socklist *sfl1=NULL;
417	struct ipv6_fl_socklist sfl, *sflp;
418	struct ip6_flowlabel *fl;
419
420	if (optlen < sizeof(freq))
421	return -EINVAL;
422
423	if (copy_from_user(&freq, optval, sizeof(freq)))
424	return -EFAULT;
425
426	switch (freq.flr_action) {
427	case IPV6_FL_A_PUT:
428	write_lock_bh(&ip6_sk_fl_lock);
429	for (sflp = &np->ipv6_fl_list; (sfl=*sflp)!=NULL; sflp = &sfl->next) {
430	if (sfl->fl->label == freq.flr_label) {
431	if (freq.flr_label == (np->flow_label&IPV6_FLOWLABEL_MASK))
432	np->flow_label &= ~IPV6_FLOWLABEL_MASK;
433	*sflp = sfl->next;
434	write_unlock_bh(&ip6_sk_fl_lock);
435	fl_release(sfl->fl);
436	kfree(sfl);
437	return 0;
438	}
439	}
440	write_unlock_bh(&ip6_sk_fl_lock);
441	return -ESRCH;
442
443	case IPV6_FL_A_RENEW:
444	read_lock_bh(&ip6_sk_fl_lock);
445	for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) {
446	if (sfl->fl->label == freq.flr_label) {
447	err = fl6_renew(sfl->fl, freq.flr_linger, freq.flr_expires);
448	read_unlock_bh(&ip6_sk_fl_lock);
449	return err;
450	}
451	}
452	read_unlock_bh(&ip6_sk_fl_lock);
453
454	if (freq.flr_share == IPV6_FL_S_NONE && capable(CAP_NET_ADMIN)) {
455	fl = fl_lookup(freq.flr_label);
456	if (fl) {
457	err = fl6_renew(fl, freq.flr_linger, freq.flr_expires);
458	fl_release(fl);
459	return err;
460	}
461	}
462	return -ESRCH;
463
464	case IPV6_FL_A_GET:
465	if (freq.flr_label & ~IPV6_FLOWLABEL_MASK)
466	return -EINVAL;
467
468	fl = fl_create(&freq, optval, optlen, &err);
469	if (fl == NULL)
470	return err;
471	sfl1 = kmalloc(sizeof(*sfl1), GFP_KERNEL);
472
473	if (freq.flr_label) {
474	struct ip6_flowlabel *fl1 = NULL;
475
476	err = -EEXIST;
477	read_lock_bh(&ip6_sk_fl_lock);
478	for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) {
479	if (sfl->fl->label == freq.flr_label) {
480	if (freq.flr_flags&IPV6_FL_F_EXCL) {
481	read_unlock_bh(&ip6_sk_fl_lock);
482	goto done;
483	}
484	fl1 = sfl->fl;
4ea6a804	485	atomic_inc(&fl1->users);
1da177e4 LT	486	break;
	487	}
	488	}
	489	read_unlock_bh(&ip6_sk_fl_lock);
	490
	491	if (fl1 == NULL)
	492	fl1 = fl_lookup(freq.flr_label);
	493	if (fl1) {
	494	err = -EEXIST;
	495	if (freq.flr_flags&IPV6_FL_F_EXCL)
	496	goto release;
	497	err = -EPERM;
	498	if (fl1->share == IPV6_FL_S_EXCL \|\|
	499	fl1->share != fl->share \|\|
	500	fl1->owner != fl->owner)
	501	goto release;
	502
	503	err = -EINVAL;
	504	if (!ipv6_addr_equal(&fl1->dst, &fl->dst) \|\|
	505	ipv6_opt_cmp(fl1->opt, fl->opt))
	506	goto release;
	507
	508	err = -ENOMEM;
	509	if (sfl1 == NULL)
	510	goto release;
	511	if (fl->linger > fl1->linger)
	512	fl1->linger = fl->linger;
	513	if ((long)(fl->expires - fl1->expires) > 0)
	514	fl1->expires = fl->expires;
	515	write_lock_bh(&ip6_sk_fl_lock);
	516	sfl1->fl = fl1;
	517	sfl1->next = np->ipv6_fl_list;
	518	np->ipv6_fl_list = sfl1;
	519	write_unlock_bh(&ip6_sk_fl_lock);
	520	fl_free(fl);
	521	return 0;
	522
	523	release:
	524	fl_release(fl1);
	525	goto done;
	526	}
	527	}
	528	err = -ENOENT;
	529	if (!(freq.flr_flags&IPV6_FL_F_CREATE))
	530	goto done;
	531
	532	err = -ENOMEM;
	533	if (sfl1 == NULL \|\| (err = mem_check(sk)) != 0)
	534	goto done;
	535
	536	err = fl_intern(fl, freq.flr_label);
	537	if (err)
	538	goto done;
	539
6c94d361 DM	540	if (!freq.flr_label) {
	541	if (copy_to_user(&((struct in6_flowlabel_req __user *) optval)->flr_label,
	542	&fl->label, sizeof(fl->label))) {
	543	/* Intentionally ignore fault. */
	544	}
	545	}
1da177e4 LT	546
	547	sfl1->fl = fl;
	548	sfl1->next = np->ipv6_fl_list;
	549	np->ipv6_fl_list = sfl1;
	550	return 0;
	551
	552	default:
	553	return -EINVAL;
	554	}
	555
	556	done:
	557	fl_free(fl);
	558	kfree(sfl1);
	559	return err;
	560	}
	561
	562	#ifdef CONFIG_PROC_FS
	563
	564	struct ip6fl_iter_state {
	565	int bucket;
	566	};
	567
	568	#define ip6fl_seq_private(seq) ((struct ip6fl_iter_state *)(seq)->private)
	569
	570	static struct ip6_flowlabel ip6fl_get_first(struct seq_file seq)
	571	{
	572	struct ip6_flowlabel *fl = NULL;
	573	struct ip6fl_iter_state *state = ip6fl_seq_private(seq);
	574
	575	for (state->bucket = 0; state->bucket <= FL_HASH_MASK; ++state->bucket) {
	576	if (fl_ht[state->bucket]) {
	577	fl = fl_ht[state->bucket];
	578	break;
	579	}
	580	}
	581	return fl;
	582	}
	583
	584	static struct ip6_flowlabel ip6fl_get_next(struct seq_file seq, struct ip6_flowlabel *fl)
	585	{
	586	struct ip6fl_iter_state *state = ip6fl_seq_private(seq);
	587
	588	fl = fl->next;
	589	while (!fl) {
	590	if (++state->bucket <= FL_HASH_MASK)
	591	fl = fl_ht[state->bucket];
bcd62075 JM	592	else
bcd62075 JM	593	break;
1da177e4 LT	594	}
	595	return fl;
	596	}
	597
	598	static struct ip6_flowlabel ip6fl_get_idx(struct seq_file seq, loff_t pos)
	599	{
	600	struct ip6_flowlabel *fl = ip6fl_get_first(seq);
	601	if (fl)
	602	while (pos && (fl = ip6fl_get_next(seq, fl)) != NULL)
	603	--pos;
	604	return pos ? NULL : fl;
	605	}
	606
	607	static void ip6fl_seq_start(struct seq_file seq, loff_t *pos)
	608	{
	609	read_lock_bh(&ip6_fl_lock);
	610	return pos ? ip6fl_get_idx(seq, pos - 1) : SEQ_START_TOKEN;
	611	}
	612
	613	static void ip6fl_seq_next(struct seq_file seq, void v, loff_t pos)
	614	{
	615	struct ip6_flowlabel *fl;
	616
	617	if (v == SEQ_START_TOKEN)
	618	fl = ip6fl_get_first(seq);
	619	else
	620	fl = ip6fl_get_next(seq, v);
	621	++*pos;
	622	return fl;
	623	}
	624
	625	static void ip6fl_seq_stop(struct seq_file seq, void v)
	626	{
	627	read_unlock_bh(&ip6_fl_lock);
	628	}
	629
1b7c2dbc	630	static int ip6fl_seq_show(struct seq_file seq, void v)
1da177e4	631	{
1b7c2dbc JM	632	if (v == SEQ_START_TOKEN)
	633	seq_printf(seq, "%-5s %-1s %-6s %-6s %-6s %-8s %-32s %s\n",
	634	"Label", "S", "Owner", "Users", "Linger", "Expires", "Dst", "Opt");
	635	else {
	636	struct ip6_flowlabel *fl = v;
1da177e4	637	seq_printf(seq,
9343e79a	638	"%05X %-1d %-6d %-6d %-6ld %-8ld " NIP6_SEQFMT " %-4d\n",
1da177e4 LT	639	(unsigned)ntohl(fl->label),
	640	fl->share,
	641	(unsigned)fl->owner,
	642	atomic_read(&fl->users),
	643	fl->linger/HZ,
	644	(long)(fl->expires - jiffies)/HZ,
	645	NIP6(fl->dst),
	646	fl->opt ? fl->opt->opt_nflen : 0);
1da177e4	647	}
1da177e4 LT	648	return 0;
	649	}
	650
	651	static struct seq_operations ip6fl_seq_ops = {
	652	.start = ip6fl_seq_start,
	653	.next = ip6fl_seq_next,
	654	.stop = ip6fl_seq_stop,
	655	.show = ip6fl_seq_show,
	656	};
	657
	658	static int ip6fl_seq_open(struct inode inode, struct file file)
	659	{
	660	struct seq_file *seq;
	661	int rc = -ENOMEM;
0c600eda	662	struct ip6fl_iter_state s = kzalloc(sizeof(s), GFP_KERNEL);
1da177e4 LT	663
	664	if (!s)
	665	goto out;
	666
	667	rc = seq_open(file, &ip6fl_seq_ops);
	668	if (rc)
	669	goto out_kfree;
	670
	671	seq = file->private_data;
	672	seq->private = s;
1da177e4 LT	673	out:
	674	return rc;
	675	out_kfree:
	676	kfree(s);
	677	goto out;
	678	}
	679
9a32144e	680	static const struct file_operations ip6fl_seq_fops = {
1da177e4 LT	681	.owner = THIS_MODULE,
	682	.open = ip6fl_seq_open,
	683	.read = seq_read,
	684	.llseek = seq_lseek,
	685	.release = seq_release_private,
	686	};
	687	#endif
	688
	689
	690	void ip6_flowlabel_init(void)
	691	{
	692	#ifdef CONFIG_PROC_FS
	693	proc_net_fops_create("ip6_flowlabel", S_IRUGO, &ip6fl_seq_fops);
	694	#endif
	695	}
	696
	697	void ip6_flowlabel_cleanup(void)
	698	{
	699	del_timer(&ip6_fl_gc_timer);
	700	#ifdef CONFIG_PROC_FS
	701	proc_net_remove("ip6_flowlabel");
	702	#endif
	703	}