[deliverable/linux.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/config.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(u32 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(u32 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, __u32 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, u32 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;
}

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 = kmalloc(sizeof(*fl), GFP_KERNEL);
	if (fl == NULL)
		goto done;
	memset(fl, 0, sizeof(*fl));

	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)
		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];
	}
	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 void ip6fl_fl_seq_show(struct seq_file *seq, struct ip6_flowlabel *fl)
{
	while(fl) {
		seq_printf(seq,
			   "%05X %-1d %-6d %-6d %-6ld %-8ld "
			   "%02x%02x%02x%02x%02x%02x%02x%02x "
			   "%-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);
		fl = fl->next;
	}
}

static int ip6fl_seq_show(struct seq_file *seq, void *v)
{
	if (v == SEQ_START_TOKEN)
		seq_puts(seq, "Label S Owner  Users  Linger Expires  "
			      "Dst                              Opt\n");
	else
		ip6fl_fl_seq_show(seq, v);
	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 = kmalloc(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;
	memset(s, 0, sizeof(*s));
out:
	return rc;
out_kfree:
	kfree(s);
	goto out;
}

static 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
	12	#include <linux/config.h>
	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
	64	static __inline__ struct ip6_flowlabel * __fl_lookup(u32 label)
	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
	75	static struct ip6_flowlabel * fl_lookup(u32 label)
	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
156	static int fl_intern(struct ip6_flowlabel *fl, __u32 label)
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
185	struct ip6_flowlabel * fl6_sock_lookup(struct sock *sk, u32 label)
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
203	void fl6_free_socklist(struct sock *sk)
204	{
205	struct ipv6_pinfo *np = inet6_sk(sk);
206	struct ipv6_fl_socklist *sfl;
207
208	while ((sfl = np->ipv6_fl_list) != NULL) {
209	np->ipv6_fl_list = sfl->next;
210	fl_release(sfl->fl);
211	kfree(sfl);
212	}
213	}
214
215	/* Service routines */
216
217
218	/*
219	It is the only difficult place. flowlabel enforces equal headers
220	before and including routing header, however user may supply options
221	following rthdr.
222	*/
223
224	struct ipv6_txoptions fl6_merge_options(struct ipv6_txoptions opt_space,
225	struct ip6_flowlabel * fl,
226	struct ipv6_txoptions * fopt)
227	{
df9890c3 YH	228	struct ipv6_txoptions * fl_opt = fl->opt;
	229
	230	if (fopt == NULL \|\| fopt->opt_flen == 0)
	231	return fl_opt;
	232
1da177e4 LT	233	if (fl_opt != NULL) {
1da177e4 LT	234	opt_space->hopopt = fl_opt->hopopt;
df9890c3	235	opt_space->dst0opt = fl_opt->dst0opt;
1da177e4 LT	236	opt_space->srcrt = fl_opt->srcrt;
	237	opt_space->opt_nflen = fl_opt->opt_nflen;
	238	} else {
	239	if (fopt->opt_nflen == 0)
	240	return fopt;
	241	opt_space->hopopt = NULL;
	242	opt_space->dst0opt = NULL;
	243	opt_space->srcrt = NULL;
	244	opt_space->opt_nflen = 0;
	245	}
	246	opt_space->dst1opt = fopt->dst1opt;
1da177e4 LT	247	opt_space->opt_flen = fopt->opt_flen;
	248	return opt_space;
	249	}
	250
	251	static unsigned long check_linger(unsigned long ttl)
	252	{
	253	if (ttl < FL_MIN_LINGER)
	254	return FL_MIN_LINGER*HZ;
	255	if (ttl > FL_MAX_LINGER && !capable(CAP_NET_ADMIN))
	256	return 0;
	257	return ttl*HZ;
	258	}
	259
	260	static int fl6_renew(struct ip6_flowlabel *fl, unsigned long linger, unsigned long expires)
	261	{
	262	linger = check_linger(linger);
	263	if (!linger)
	264	return -EPERM;
	265	expires = check_linger(expires);
	266	if (!expires)
	267	return -EPERM;
	268	fl->lastuse = jiffies;
	269	if (time_before(fl->linger, linger))
	270	fl->linger = linger;
	271	if (time_before(expires, fl->linger))
	272	expires = fl->linger;
	273	if (time_before(fl->expires, fl->lastuse + expires))
	274	fl->expires = fl->lastuse + expires;
	275	return 0;
	276	}
	277
	278	static struct ip6_flowlabel *
	279	fl_create(struct in6_flowlabel_req freq, char __user optval, int optlen, int *err_p)
	280	{
	281	struct ip6_flowlabel *fl;
	282	int olen;
	283	int addr_type;
	284	int err;
	285
	286	err = -ENOMEM;
	287	fl = kmalloc(sizeof(*fl), GFP_KERNEL);
	288	if (fl == NULL)
	289	goto done;
	290	memset(fl, 0, sizeof(*fl));
	291
	292	olen = optlen - CMSG_ALIGN(sizeof(*freq));
	293	if (olen > 0) {
	294	struct msghdr msg;
	295	struct flowi flowi;
	296	int junk;
	297
	298	err = -ENOMEM;
	299	fl->opt = kmalloc(sizeof(*fl->opt) + olen, GFP_KERNEL);
	300	if (fl->opt == NULL)
	301	goto done;
	302
	303	memset(fl->opt, 0, sizeof(*fl->opt));
	304	fl->opt->tot_len = sizeof(*fl->opt) + olen;
	305	err = -EFAULT;
	306	if (copy_from_user(fl->opt+1, optval+CMSG_ALIGN(sizeof(*freq)), olen))
	307	goto done;
	308
	309	msg.msg_controllen = olen;
	310	msg.msg_control = (void*)(fl->opt+1);
311	flowi.oif = 0;
312
41a1f8ea	313	err = datagram_send_ctl(&msg, &flowi, fl->opt, &junk, &junk);
1da177e4 LT	314	if (err)
	315	goto done;
	316	err = -EINVAL;
	317	if (fl->opt->opt_flen)
	318	goto done;
	319	if (fl->opt->opt_nflen == 0) {
	320	kfree(fl->opt);
	321	fl->opt = NULL;
	322	}
	323	}
	324
	325	fl->expires = jiffies;
	326	err = fl6_renew(fl, freq->flr_linger, freq->flr_expires);
	327	if (err)
	328	goto done;
	329	fl->share = freq->flr_share;
	330	addr_type = ipv6_addr_type(&freq->flr_dst);
	331	if ((addr_type&IPV6_ADDR_MAPPED)
	332	\|\| addr_type == IPV6_ADDR_ANY)
	333	goto done;
	334	ipv6_addr_copy(&fl->dst, &freq->flr_dst);
	335	atomic_set(&fl->users, 1);
	336	switch (fl->share) {
	337	case IPV6_FL_S_EXCL:
	338	case IPV6_FL_S_ANY:
	339	break;
	340	case IPV6_FL_S_PROCESS:
	341	fl->owner = current->pid;
	342	break;
	343	case IPV6_FL_S_USER:
	344	fl->owner = current->euid;
	345	break;
	346	default:
	347	err = -EINVAL;
	348	goto done;
	349	}
	350	return fl;
	351
	352	done:
	353	fl_free(fl);
	354	*err_p = err;
	355	return NULL;
	356	}
	357
	358	static int mem_check(struct sock *sk)
	359	{
	360	struct ipv6_pinfo *np = inet6_sk(sk);
	361	struct ipv6_fl_socklist *sfl;
	362	int room = FL_MAX_SIZE - atomic_read(&fl_size);
	363	int count = 0;
	364
	365	if (room > FL_MAX_SIZE - FL_MAX_PER_SOCK)
	366	return 0;
	367
	368	for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next)
	369	count++;
	370
	371	if (room <= 0 \|\|
	372	((count >= FL_MAX_PER_SOCK \|\|
	373	(count > 0 && room < FL_MAX_SIZE/2) \|\| room < FL_MAX_SIZE/4)
	374	&& !capable(CAP_NET_ADMIN)))
	375	return -ENOBUFS;
	376
	377	return 0;
378	}
379
380	static int ipv6_hdr_cmp(struct ipv6_opt_hdr h1, struct ipv6_opt_hdr h2)
381	{
382	if (h1 == h2)
383	return 0;
384	if (h1 == NULL \|\| h2 == NULL)
385	return 1;
386	if (h1->hdrlen != h2->hdrlen)
387	return 1;
388	return memcmp(h1+1, h2+1, ((h1->hdrlen+1)<<3) - sizeof(*h1));
389	}
390
391	static int ipv6_opt_cmp(struct ipv6_txoptions o1, struct ipv6_txoptions o2)
392	{
393	if (o1 == o2)
394	return 0;
395	if (o1 == NULL \|\| o2 == NULL)
396	return 1;
397	if (o1->opt_nflen != o2->opt_nflen)
398	return 1;
399	if (ipv6_hdr_cmp(o1->hopopt, o2->hopopt))
400	return 1;
401	if (ipv6_hdr_cmp(o1->dst0opt, o2->dst0opt))
402	return 1;
403	if (ipv6_hdr_cmp((struct ipv6_opt_hdr )o1->srcrt, (struct ipv6_opt_hdr )o2->srcrt))
404	return 1;
405	return 0;
406	}
407
408	int ipv6_flowlabel_opt(struct sock sk, char __user optval, int optlen)
409	{
410	int err;
411	struct ipv6_pinfo *np = inet6_sk(sk);
412	struct in6_flowlabel_req freq;
413	struct ipv6_fl_socklist *sfl1=NULL;
414	struct ipv6_fl_socklist sfl, *sflp;
415	struct ip6_flowlabel *fl;
416
417	if (optlen < sizeof(freq))
418	return -EINVAL;
419
420	if (copy_from_user(&freq, optval, sizeof(freq)))
421	return -EFAULT;
422
423	switch (freq.flr_action) {
424	case IPV6_FL_A_PUT:
425	write_lock_bh(&ip6_sk_fl_lock);
426	for (sflp = &np->ipv6_fl_list; (sfl=*sflp)!=NULL; sflp = &sfl->next) {
427	if (sfl->fl->label == freq.flr_label) {
428	if (freq.flr_label == (np->flow_label&IPV6_FLOWLABEL_MASK))
429	np->flow_label &= ~IPV6_FLOWLABEL_MASK;
430	*sflp = sfl->next;
431	write_unlock_bh(&ip6_sk_fl_lock);
432	fl_release(sfl->fl);
433	kfree(sfl);
434	return 0;
435	}
436	}
437	write_unlock_bh(&ip6_sk_fl_lock);
438	return -ESRCH;
439
440	case IPV6_FL_A_RENEW:
441	read_lock_bh(&ip6_sk_fl_lock);
442	for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) {
443	if (sfl->fl->label == freq.flr_label) {
444	err = fl6_renew(sfl->fl, freq.flr_linger, freq.flr_expires);
445	read_unlock_bh(&ip6_sk_fl_lock);
446	return err;
447	}
448	}
449	read_unlock_bh(&ip6_sk_fl_lock);
450
451	if (freq.flr_share == IPV6_FL_S_NONE && capable(CAP_NET_ADMIN)) {
452	fl = fl_lookup(freq.flr_label);
453	if (fl) {
454	err = fl6_renew(fl, freq.flr_linger, freq.flr_expires);
455	fl_release(fl);
456	return err;
457	}
458	}
459	return -ESRCH;
460
461	case IPV6_FL_A_GET:
462	if (freq.flr_label & ~IPV6_FLOWLABEL_MASK)
463	return -EINVAL;
464
465	fl = fl_create(&freq, optval, optlen, &err);
466	if (fl == NULL)
467	return err;
468	sfl1 = kmalloc(sizeof(*sfl1), GFP_KERNEL);
469
470	if (freq.flr_label) {
471	struct ip6_flowlabel *fl1 = NULL;
472
473	err = -EEXIST;
474	read_lock_bh(&ip6_sk_fl_lock);
475	for (sfl = np->ipv6_fl_list; sfl; sfl = sfl->next) {
476	if (sfl->fl->label == freq.flr_label) {
477	if (freq.flr_flags&IPV6_FL_F_EXCL) {
478	read_unlock_bh(&ip6_sk_fl_lock);
479	goto done;
480	}
481	fl1 = sfl->fl;
4ea6a804	482	atomic_inc(&fl1->users);
1da177e4 LT	483	break;
	484	}
	485	}
	486	read_unlock_bh(&ip6_sk_fl_lock);
	487
	488	if (fl1 == NULL)
	489	fl1 = fl_lookup(freq.flr_label);
	490	if (fl1) {
	491	err = -EEXIST;
	492	if (freq.flr_flags&IPV6_FL_F_EXCL)
	493	goto release;
	494	err = -EPERM;
	495	if (fl1->share == IPV6_FL_S_EXCL \|\|
	496	fl1->share != fl->share \|\|
	497	fl1->owner != fl->owner)
	498	goto release;
	499
	500	err = -EINVAL;
	501	if (!ipv6_addr_equal(&fl1->dst, &fl->dst) \|\|
	502	ipv6_opt_cmp(fl1->opt, fl->opt))
	503	goto release;
	504
	505	err = -ENOMEM;
	506	if (sfl1 == NULL)
	507	goto release;
	508	if (fl->linger > fl1->linger)
	509	fl1->linger = fl->linger;
	510	if ((long)(fl->expires - fl1->expires) > 0)
	511	fl1->expires = fl->expires;
	512	write_lock_bh(&ip6_sk_fl_lock);
	513	sfl1->fl = fl1;
	514	sfl1->next = np->ipv6_fl_list;
	515	np->ipv6_fl_list = sfl1;
	516	write_unlock_bh(&ip6_sk_fl_lock);
	517	fl_free(fl);
	518	return 0;
	519
	520	release:
	521	fl_release(fl1);
	522	goto done;
	523	}
	524	}
	525	err = -ENOENT;
	526	if (!(freq.flr_flags&IPV6_FL_F_CREATE))
	527	goto done;
	528
	529	err = -ENOMEM;
	530	if (sfl1 == NULL \|\| (err = mem_check(sk)) != 0)
	531	goto done;
	532
	533	err = fl_intern(fl, freq.flr_label);
	534	if (err)
	535	goto done;
	536
6c94d361 DM	537	if (!freq.flr_label) {
	538	if (copy_to_user(&((struct in6_flowlabel_req __user *) optval)->flr_label,
	539	&fl->label, sizeof(fl->label))) {
	540	/* Intentionally ignore fault. */
	541	}
	542	}
1da177e4 LT	543
	544	sfl1->fl = fl;
	545	sfl1->next = np->ipv6_fl_list;
	546	np->ipv6_fl_list = sfl1;
	547	return 0;
	548
	549	default:
	550	return -EINVAL;
	551	}
	552
	553	done:
	554	fl_free(fl);
	555	kfree(sfl1);
	556	return err;
	557	}
	558
	559	#ifdef CONFIG_PROC_FS
	560
	561	struct ip6fl_iter_state {
	562	int bucket;
	563	};
	564
	565	#define ip6fl_seq_private(seq) ((struct ip6fl_iter_state *)(seq)->private)
	566
	567	static struct ip6_flowlabel ip6fl_get_first(struct seq_file seq)
	568	{
	569	struct ip6_flowlabel *fl = NULL;
	570	struct ip6fl_iter_state *state = ip6fl_seq_private(seq);
	571
	572	for (state->bucket = 0; state->bucket <= FL_HASH_MASK; ++state->bucket) {
	573	if (fl_ht[state->bucket]) {
	574	fl = fl_ht[state->bucket];
	575	break;
	576	}
	577	}
	578	return fl;
	579	}
	580
	581	static struct ip6_flowlabel ip6fl_get_next(struct seq_file seq, struct ip6_flowlabel *fl)
	582	{
	583	struct ip6fl_iter_state *state = ip6fl_seq_private(seq);
	584
	585	fl = fl->next;
	586	while (!fl) {
	587	if (++state->bucket <= FL_HASH_MASK)
	588	fl = fl_ht[state->bucket];
	589	}
	590	return fl;
	591	}
	592
	593	static struct ip6_flowlabel ip6fl_get_idx(struct seq_file seq, loff_t pos)
	594	{
	595	struct ip6_flowlabel *fl = ip6fl_get_first(seq);
	596	if (fl)
	597	while (pos && (fl = ip6fl_get_next(seq, fl)) != NULL)
	598	--pos;
	599	return pos ? NULL : fl;
	600	}
	601
	602	static void ip6fl_seq_start(struct seq_file seq, loff_t *pos)
	603	{
	604	read_lock_bh(&ip6_fl_lock);
	605	return pos ? ip6fl_get_idx(seq, pos - 1) : SEQ_START_TOKEN;
	606	}
607
608	static void ip6fl_seq_next(struct seq_file seq, void v, loff_t pos)
609	{
610	struct ip6_flowlabel *fl;
611
612	if (v == SEQ_START_TOKEN)
613	fl = ip6fl_get_first(seq);
614	else
615	fl = ip6fl_get_next(seq, v);
616	++*pos;
617	return fl;
618	}
619
620	static void ip6fl_seq_stop(struct seq_file seq, void v)
621	{
622	read_unlock_bh(&ip6_fl_lock);
623	}
624
625	static void ip6fl_fl_seq_show(struct seq_file seq, struct ip6_flowlabel fl)
626	{
627	while(fl) {
628	seq_printf(seq,
629	"%05X %-1d %-6d %-6d %-6ld %-8ld "
630	"%02x%02x%02x%02x%02x%02x%02x%02x "
631	"%-4d\n",
632	(unsigned)ntohl(fl->label),
633	fl->share,
634	(unsigned)fl->owner,
635	atomic_read(&fl->users),
636	fl->linger/HZ,
637	(long)(fl->expires - jiffies)/HZ,
638	NIP6(fl->dst),
639	fl->opt ? fl->opt->opt_nflen : 0);
640	fl = fl->next;
641	}
642	}
643
644	static int ip6fl_seq_show(struct seq_file seq, void v)
645	{
646	if (v == SEQ_START_TOKEN)
647	seq_puts(seq, "Label S Owner Users Linger Expires "
648	"Dst Opt\n");
649	else
650	ip6fl_fl_seq_show(seq, v);
651	return 0;
652	}
653
654	static struct seq_operations ip6fl_seq_ops = {
655	.start = ip6fl_seq_start,
656	.next = ip6fl_seq_next,
657	.stop = ip6fl_seq_stop,
658	.show = ip6fl_seq_show,
659	};
660
661	static int ip6fl_seq_open(struct inode inode, struct file file)
662	{
663	struct seq_file *seq;
664	int rc = -ENOMEM;
665	struct ip6fl_iter_state s = kmalloc(sizeof(s), GFP_KERNEL);
666
667	if (!s)
668	goto out;
669
670	rc = seq_open(file, &ip6fl_seq_ops);
671	if (rc)
672	goto out_kfree;
673
674	seq = file->private_data;
675	seq->private = s;
676	memset(s, 0, sizeof(*s));
677	out:
678	return rc;
679	out_kfree:
680	kfree(s);
681	goto out;
682	}
683
684	static struct file_operations ip6fl_seq_fops = {
685	.owner = THIS_MODULE,
686	.open = ip6fl_seq_open,
687	.read = seq_read,
688	.llseek = seq_lseek,
689	.release = seq_release_private,
690	};
691	#endif
692
693
694	void ip6_flowlabel_init(void)
695	{
696	#ifdef CONFIG_PROC_FS
697	proc_net_fops_create("ip6_flowlabel", S_IRUGO, &ip6fl_seq_fops);
698	#endif
699	}
700
701	void ip6_flowlabel_cleanup(void)
702	{
703	del_timer(&ip6_fl_gc_timer);
704	#ifdef CONFIG_PROC_FS
705	proc_net_remove("ip6_flowlabel");
706	#endif
707	}