[NETNS][IPV6] flowlabels - make flowlabels per namespace
authorBenjamin Thery <benjamin.thery@bull.net>
Wed, 26 Mar 2008 23:53:08 +0000 (16:53 -0700)
committerDavid S. Miller <davem@davemloft.net>
Wed, 26 Mar 2008 23:53:08 +0000 (16:53 -0700)
This patch introduces a new member, fl_net, in struct ip6_flowlabel.
This allows to create labels with the same value in different namespaces.

Signed-off-by: Benjamin Thery <benjamin.thery@bull.net>
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
include/net/ipv6.h
net/ipv6/ip6_flowlabel.c

index 1c98e737dbd016d799dac55738bcba6059a32fa4..296f61d8470965ff809019bf974e4edaa79a96fa 100644 (file)
@@ -202,6 +202,7 @@ struct ip6_flowlabel
        u32                     owner;
        unsigned long           lastuse;
        unsigned long           expires;
+       struct net              *fl_net;
 };
 
 #define IPV6_FLOWINFO_MASK     __constant_htonl(0x0FFFFFFF)
index 2b7d9ee98832f5d1a0afce7095bc8f8e10dea66f..78d1d913e36c7ff39e30efa69b5f8bfec4b0b426 100644 (file)
@@ -62,23 +62,23 @@ static DEFINE_RWLOCK(ip6_fl_lock);
 static DEFINE_RWLOCK(ip6_sk_fl_lock);
 
 
-static __inline__ struct ip6_flowlabel * __fl_lookup(__be32 label)
+static inline struct ip6_flowlabel *__fl_lookup(struct net *net, __be32 label)
 {
        struct ip6_flowlabel *fl;
 
        for (fl=fl_ht[FL_HASH(label)]; fl; fl = fl->next) {
-               if (fl->label == label)
+               if (fl->label == label && fl->fl_net == net)
                        return fl;
        }
        return NULL;
 }
 
-static struct ip6_flowlabel * fl_lookup(__be32 label)
+static struct ip6_flowlabel *fl_lookup(struct net *net, __be32 label)
 {
        struct ip6_flowlabel *fl;
 
        read_lock_bh(&ip6_fl_lock);
-       fl = __fl_lookup(label);
+       fl = __fl_lookup(net, label);
        if (fl)
                atomic_inc(&fl->users);
        read_unlock_bh(&ip6_fl_lock);
@@ -88,8 +88,10 @@ static struct ip6_flowlabel * fl_lookup(__be32 label)
 
 static void fl_free(struct ip6_flowlabel *fl)
 {
-       if (fl)
+       if (fl) {
+               release_net(fl->fl_net);
                kfree(fl->opt);
+       }
        kfree(fl);
 }
 
@@ -112,7 +114,6 @@ static void fl_release(struct ip6_flowlabel *fl)
                    time_after(ip6_fl_gc_timer.expires, ttd))
                        mod_timer(&ip6_fl_gc_timer, ttd);
        }
-
        write_unlock_bh(&ip6_fl_lock);
 }
 
@@ -148,13 +149,34 @@ static void ip6_fl_gc(unsigned long dummy)
        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);
+               mod_timer(&ip6_fl_gc_timer, sched);
        }
        write_unlock(&ip6_fl_lock);
 }
 
-static struct ip6_flowlabel *fl_intern(struct ip6_flowlabel *fl, __be32 label)
+static void ip6_fl_purge(struct net *net)
+{
+       int i;
+
+       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 (fl->fl_net == net && atomic_read(&fl->users) == 0) {
+                               *flp = fl->next;
+                               fl_free(fl);
+                               atomic_dec(&fl_size);
+                               continue;
+                       }
+                       flp = &fl->next;
+               }
+       }
+       write_unlock(&ip6_fl_lock);
+}
+
+static struct ip6_flowlabel *fl_intern(struct net *net,
+                                      struct ip6_flowlabel *fl, __be32 label)
 {
        struct ip6_flowlabel *lfl;
 
@@ -165,7 +187,7 @@ static struct ip6_flowlabel *fl_intern(struct ip6_flowlabel *fl, __be32 label)
                for (;;) {
                        fl->label = htonl(net_random())&IPV6_FLOWLABEL_MASK;
                        if (fl->label) {
-                               lfl = __fl_lookup(fl->label);
+                               lfl = __fl_lookup(net, fl->label);
                                if (lfl == NULL)
                                        break;
                        }
@@ -179,7 +201,7 @@ static struct ip6_flowlabel *fl_intern(struct ip6_flowlabel *fl, __be32 label)
                 * done in ipv6_flowlabel_opt - sock is locked, so new entry
                 * with the same label can only appear on another sock
                 */
-               lfl = __fl_lookup(fl->label);
+               lfl = __fl_lookup(net, fl->label);
                if (lfl != NULL) {
                        atomic_inc(&lfl->users);
                        write_unlock_bh(&ip6_fl_lock);
@@ -298,7 +320,8 @@ static int fl6_renew(struct ip6_flowlabel *fl, unsigned long linger, unsigned lo
 }
 
 static struct ip6_flowlabel *
-fl_create(struct in6_flowlabel_req *freq, char __user *optval, int optlen, int *err_p)
+fl_create(struct net *net, struct in6_flowlabel_req *freq, char __user *optval,
+         int optlen, int *err_p)
 {
        struct ip6_flowlabel *fl;
        int olen;
@@ -343,6 +366,7 @@ fl_create(struct in6_flowlabel_req *freq, char __user *optval, int optlen, int *
                }
        }
 
+       fl->fl_net = hold_net(net);
        fl->expires = jiffies;
        err = fl6_renew(fl, freq->flr_linger, freq->flr_expires);
        if (err)
@@ -441,6 +465,7 @@ static inline void fl_link(struct ipv6_pinfo *np, struct ipv6_fl_socklist *sfl,
 int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen)
 {
        int err;
+       struct net *net = sock_net(sk);
        struct ipv6_pinfo *np = inet6_sk(sk);
        struct in6_flowlabel_req freq;
        struct ipv6_fl_socklist *sfl1=NULL;
@@ -483,7 +508,7 @@ int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen)
                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);
+                       fl = fl_lookup(net, freq.flr_label);
                        if (fl) {
                                err = fl6_renew(fl, freq.flr_linger, freq.flr_expires);
                                fl_release(fl);
@@ -496,7 +521,7 @@ int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen)
                if (freq.flr_label & ~IPV6_FLOWLABEL_MASK)
                        return -EINVAL;
 
-               fl = fl_create(&freq, optval, optlen, &err);
+               fl = fl_create(net, &freq, optval, optlen, &err);
                if (fl == NULL)
                        return err;
                sfl1 = kmalloc(sizeof(*sfl1), GFP_KERNEL);
@@ -518,7 +543,7 @@ int ipv6_flowlabel_opt(struct sock *sk, char __user *optval, int optlen)
                        read_unlock_bh(&ip6_sk_fl_lock);
 
                        if (fl1 == NULL)
-                               fl1 = fl_lookup(freq.flr_label);
+                               fl1 = fl_lookup(net, freq.flr_label);
                        if (fl1) {
 recheck:
                                err = -EEXIST;
@@ -559,7 +584,7 @@ release:
                if (sfl1 == NULL || (err = mem_check(sk)) != 0)
                        goto done;
 
-               fl1 = fl_intern(fl, freq.flr_label);
+               fl1 = fl_intern(net, fl, freq.flr_label);
                if (fl1 != NULL)
                        goto recheck;
 
@@ -717,13 +742,28 @@ static inline void ip6_flowlabel_proc_fini(struct net *net)
 }
 #endif
 
+static inline void ip6_flowlabel_net_exit(struct net *net)
+{
+       ip6_fl_purge(net);
+}
+
+static struct pernet_operations ip6_flowlabel_net_ops = {
+       .exit = ip6_flowlabel_net_exit,
+};
+
 int ip6_flowlabel_init(void)
 {
+       int err;
+
+       err = register_pernet_subsys(&ip6_flowlabel_net_ops);
+       if (err)
+               return err;
        return ip6_flowlabel_proc_init(&init_net);
 }
 
 void ip6_flowlabel_cleanup(void)
 {
        del_timer(&ip6_fl_gc_timer);
+       unregister_pernet_subsys(&ip6_flowlabel_net_ops);
        ip6_flowlabel_proc_fini(&init_net);
 }