#include "vlan.h"
static struct kmem_cache *flow_cache;
-static unsigned int hash_seed __read_mostly;
static int check_header(struct sk_buff *skb, int len)
{
sizeof(struct icmphdr));
}
-u64 flow_used_time(unsigned long flow_jiffies)
+u64 ovs_flow_used_time(unsigned long flow_jiffies)
{
struct timespec cur_ts;
u64 cur_ms, idle_ms;
(offsetof(struct sw_flow_key, field) + \
FIELD_SIZEOF(struct sw_flow_key, field))
-/**
- * skip_exthdr - skip any IPv6 extension headers
- * @skb: skbuff to parse
- * @start: offset of first extension header
- * @nexthdrp: Initially, points to the type of the extension header at @start.
- * This function updates it to point to the extension header at the final
- * offset.
- * @frag: Points to the @frag member in a &struct sw_flow_key. This
- * function sets an appropriate %OVS_FRAG_TYPE_* value.
- *
- * This is based on ipv6_skip_exthdr() but adds the updates to *@frag.
- *
- * When there is more than one fragment header, this version reports whether
- * the final fragment header that it examines is a first fragment.
- *
- * Returns the final payload offset, or -1 on error.
- */
-static int skip_exthdr(const struct sk_buff *skb, int start, u8 *nexthdrp,
- u8 *frag)
-{
- u8 nexthdr = *nexthdrp;
-
- while (ipv6_ext_hdr(nexthdr)) {
- struct ipv6_opt_hdr _hdr, *hp;
- int hdrlen;
-
- if (nexthdr == NEXTHDR_NONE)
- return -1;
- hp = skb_header_pointer(skb, start, sizeof(_hdr), &_hdr);
- if (hp == NULL)
- return -1;
- if (nexthdr == NEXTHDR_FRAGMENT) {
- __be16 _frag_off, *fp;
- fp = skb_header_pointer(skb,
- start+offsetof(struct frag_hdr,
- frag_off),
- sizeof(_frag_off),
- &_frag_off);
- if (fp == NULL)
- return -1;
-
- if (ntohs(*fp) & ~0x7) {
- *frag = OVS_FRAG_TYPE_LATER;
- break;
- }
- *frag = OVS_FRAG_TYPE_FIRST;
- hdrlen = 8;
- } else if (nexthdr == NEXTHDR_AUTH)
- hdrlen = (hp->hdrlen+2)<<2;
- else
- hdrlen = ipv6_optlen(hp);
-
- nexthdr = hp->nexthdr;
- start += hdrlen;
- }
-
- *nexthdrp = nexthdr;
- return start;
-}
-
static int parse_ipv6hdr(struct sk_buff *skb, struct sw_flow_key *key,
int *key_lenp)
{
int payload_ofs;
struct ipv6hdr *nh;
uint8_t nexthdr;
+ __be16 frag_off;
int err;
*key_lenp = SW_FLOW_KEY_OFFSET(ipv6.label);
key->ip.tos = ipv6_get_dsfield(nh);
key->ip.ttl = nh->hop_limit;
key->ipv6.label = *(__be32 *)nh & htonl(IPV6_FLOWINFO_FLOWLABEL);
- ipv6_addr_copy(&key->ipv6.addr.src, &nh->saddr);
- ipv6_addr_copy(&key->ipv6.addr.dst, &nh->daddr);
+ key->ipv6.addr.src = nh->saddr;
+ key->ipv6.addr.dst = nh->daddr;
- payload_ofs = skip_exthdr(skb, payload_ofs, &nexthdr, &key->ip.frag);
+ payload_ofs = ipv6_skip_exthdr(skb, payload_ofs, &nexthdr, &frag_off);
if (unlikely(payload_ofs < 0))
return -EINVAL;
+ if (frag_off) {
+ if (frag_off & htons(~0x7))
+ key->ip.frag = OVS_FRAG_TYPE_LATER;
+ else
+ key->ip.frag = OVS_FRAG_TYPE_FIRST;
+ }
+
nh_len = payload_ofs - nh_ofs;
skb_set_transport_header(skb, nh_ofs + nh_len);
key->ip.proto = nexthdr;
#define TCP_FLAGS_OFFSET 13
#define TCP_FLAG_MASK 0x3f
-void flow_used(struct sw_flow *flow, struct sk_buff *skb)
+void ovs_flow_used(struct sw_flow *flow, struct sk_buff *skb)
{
u8 tcp_flags = 0;
spin_unlock(&flow->lock);
}
-struct sw_flow_actions *flow_actions_alloc(const struct nlattr *actions)
+struct sw_flow_actions *ovs_flow_actions_alloc(const struct nlattr *actions)
{
int actions_len = nla_len(actions);
struct sw_flow_actions *sfa;
return sfa;
}
-struct sw_flow *flow_alloc(void)
+struct sw_flow *ovs_flow_alloc(void)
{
struct sw_flow *flow;
return flow;
}
-static struct hlist_head __rcu *find_bucket(struct flow_table * table, u32 hash)
+static struct hlist_head *find_bucket(struct flow_table *table, u32 hash)
{
+ hash = jhash_1word(hash, table->hash_seed);
return flex_array_get(table->buckets,
(hash & (table->n_buckets - 1)));
}
-static struct flex_array __rcu *alloc_buckets(unsigned int n_buckets)
+static struct flex_array *alloc_buckets(unsigned int n_buckets)
{
- struct flex_array __rcu *buckets;
+ struct flex_array *buckets;
int i, err;
buckets = flex_array_alloc(sizeof(struct hlist_head *),
flex_array_free(buckets);
}
-struct flow_table *flow_tbl_alloc(int new_size)
+struct flow_table *ovs_flow_tbl_alloc(int new_size)
{
struct flow_table *table = kmalloc(sizeof(*table), GFP_KERNEL);
}
table->n_buckets = new_size;
table->count = 0;
+ table->node_ver = 0;
+ table->keep_flows = false;
+ get_random_bytes(&table->hash_seed, sizeof(u32));
return table;
}
static void flow_free(struct sw_flow *flow)
{
flow->dead = true;
- flow_put(flow);
+ ovs_flow_put(flow);
}
-void flow_tbl_destroy(struct flow_table *table)
+void ovs_flow_tbl_destroy(struct flow_table *table)
{
int i;
if (!table)
return;
+ if (table->keep_flows)
+ goto skip_flows;
+
for (i = 0; i < table->n_buckets; i++) {
struct sw_flow *flow;
struct hlist_head *head = flex_array_get(table->buckets, i);
struct hlist_node *node, *n;
+ int ver = table->node_ver;
- hlist_for_each_entry_safe(flow, node, n, head, hash_node) {
- hlist_del_init_rcu(&flow->hash_node);
+ hlist_for_each_entry_safe(flow, node, n, head, hash_node[ver]) {
+ hlist_del_rcu(&flow->hash_node[ver]);
flow_free(flow);
}
}
+skip_flows:
free_buckets(table->buckets);
kfree(table);
}
{
struct flow_table *table = container_of(rcu, struct flow_table, rcu);
- flow_tbl_destroy(table);
+ ovs_flow_tbl_destroy(table);
}
-void flow_tbl_deferred_destroy(struct flow_table *table)
+void ovs_flow_tbl_deferred_destroy(struct flow_table *table)
{
if (!table)
return;
call_rcu(&table->rcu, flow_tbl_destroy_rcu_cb);
}
-struct sw_flow *flow_tbl_next(struct flow_table *table, u32 *bucket, u32 *last)
+struct sw_flow *ovs_flow_tbl_next(struct flow_table *table, u32 *bucket, u32 *last)
{
struct sw_flow *flow;
struct hlist_head *head;
struct hlist_node *n;
+ int ver;
int i;
+ ver = table->node_ver;
while (*bucket < table->n_buckets) {
i = 0;
head = flex_array_get(table->buckets, *bucket);
- hlist_for_each_entry_rcu(flow, n, head, hash_node) {
+ hlist_for_each_entry_rcu(flow, n, head, hash_node[ver]) {
if (i < *last) {
i++;
continue;
return NULL;
}
-struct flow_table *flow_tbl_expand(struct flow_table *table)
+static void flow_table_copy_flows(struct flow_table *old, struct flow_table *new)
{
- struct flow_table *new_table;
- int n_buckets = table->n_buckets * 2;
+ int old_ver;
int i;
- new_table = flow_tbl_alloc(n_buckets);
- if (!new_table)
- return ERR_PTR(-ENOMEM);
+ old_ver = old->node_ver;
+ new->node_ver = !old_ver;
- for (i = 0; i < table->n_buckets; i++) {
+ /* Insert in new table. */
+ for (i = 0; i < old->n_buckets; i++) {
struct sw_flow *flow;
struct hlist_head *head;
- struct hlist_node *n, *pos;
+ struct hlist_node *n;
- head = flex_array_get(table->buckets, i);
+ head = flex_array_get(old->buckets, i);
- hlist_for_each_entry_safe(flow, n, pos, head, hash_node) {
- hlist_del_init_rcu(&flow->hash_node);
- flow_tbl_insert(new_table, flow);
- }
+ hlist_for_each_entry(flow, n, head, hash_node[old_ver])
+ ovs_flow_tbl_insert(new, flow);
}
+ old->keep_flows = true;
+}
+
+static struct flow_table *__flow_tbl_rehash(struct flow_table *table, int n_buckets)
+{
+ struct flow_table *new_table;
+
+ new_table = ovs_flow_tbl_alloc(n_buckets);
+ if (!new_table)
+ return ERR_PTR(-ENOMEM);
+
+ flow_table_copy_flows(table, new_table);
return new_table;
}
-/* RCU callback used by flow_deferred_free. */
+struct flow_table *ovs_flow_tbl_rehash(struct flow_table *table)
+{
+ return __flow_tbl_rehash(table, table->n_buckets);
+}
+
+struct flow_table *ovs_flow_tbl_expand(struct flow_table *table)
+{
+ return __flow_tbl_rehash(table, table->n_buckets * 2);
+}
+
+/* RCU callback used by ovs_flow_deferred_free. */
static void rcu_free_flow_callback(struct rcu_head *rcu)
{
struct sw_flow *flow = container_of(rcu, struct sw_flow, rcu);
flow->dead = true;
- flow_put(flow);
+ ovs_flow_put(flow);
}
/* Schedules 'flow' to be freed after the next RCU grace period.
* The caller must hold rcu_read_lock for this to be sensible. */
-void flow_deferred_free(struct sw_flow *flow)
+void ovs_flow_deferred_free(struct sw_flow *flow)
{
call_rcu(&flow->rcu, rcu_free_flow_callback);
}
-void flow_hold(struct sw_flow *flow)
+void ovs_flow_hold(struct sw_flow *flow)
{
atomic_inc(&flow->refcnt);
}
-void flow_put(struct sw_flow *flow)
+void ovs_flow_put(struct sw_flow *flow)
{
if (unlikely(!flow))
return;
}
}
-/* RCU callback used by flow_deferred_free_acts. */
+/* RCU callback used by ovs_flow_deferred_free_acts. */
static void rcu_free_acts_callback(struct rcu_head *rcu)
{
struct sw_flow_actions *sf_acts = container_of(rcu,
/* Schedules 'sf_acts' to be freed after the next RCU grace period.
* The caller must hold rcu_read_lock for this to be sensible. */
-void flow_deferred_free_acts(struct sw_flow_actions *sf_acts)
+void ovs_flow_deferred_free_acts(struct sw_flow_actions *sf_acts)
{
call_rcu(&sf_acts->rcu, rcu_free_acts_callback);
}
}
nd = (struct nd_msg *)skb_transport_header(skb);
- ipv6_addr_copy(&key->ipv6.nd.target, &nd->target);
+ key->ipv6.nd.target = nd->target;
key_len = SW_FLOW_KEY_OFFSET(ipv6.nd);
icmp_len -= sizeof(*nd);
}
/**
- * flow_extract - extracts a flow key from an Ethernet frame.
+ * ovs_flow_extract - extracts a flow key from an Ethernet frame.
* @skb: sk_buff that contains the frame, with skb->data pointing to the
* Ethernet header
* @in_port: port number on which @skb was received.
* of a correct length, otherwise the same as skb->network_header.
* For other key->dl_type values it is left untouched.
*/
-int flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key,
+int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key,
int *key_lenp)
{
int error = 0;
return error;
}
-u32 flow_hash(const struct sw_flow_key *key, int key_len)
+u32 ovs_flow_hash(const struct sw_flow_key *key, int key_len)
{
- return jhash2((u32 *)key, DIV_ROUND_UP(key_len, sizeof(u32)), hash_seed);
+ return jhash2((u32 *)key, DIV_ROUND_UP(key_len, sizeof(u32)), 0);
}
-struct sw_flow *flow_tbl_lookup(struct flow_table *table,
+struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *table,
struct sw_flow_key *key, int key_len)
{
struct sw_flow *flow;
struct hlist_head *head;
u32 hash;
- hash = flow_hash(key, key_len);
+ hash = ovs_flow_hash(key, key_len);
head = find_bucket(table, hash);
- hlist_for_each_entry_rcu(flow, n, head, hash_node) {
+ hlist_for_each_entry_rcu(flow, n, head, hash_node[table->node_ver]) {
if (flow->hash == hash &&
!memcmp(&flow->key, key, key_len)) {
return NULL;
}
-void flow_tbl_insert(struct flow_table *table, struct sw_flow *flow)
+void ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow)
{
struct hlist_head *head;
head = find_bucket(table, flow->hash);
- hlist_add_head_rcu(&flow->hash_node, head);
+ hlist_add_head_rcu(&flow->hash_node[table->node_ver], head);
table->count++;
}
-void flow_tbl_remove(struct flow_table *table, struct sw_flow *flow)
+void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow)
{
- if (!hlist_unhashed(&flow->hash_node)) {
- hlist_del_init_rcu(&flow->hash_node);
- table->count--;
- BUG_ON(table->count < 0);
- }
+ hlist_del_rcu(&flow->hash_node[table->node_ver]);
+ table->count--;
+ BUG_ON(table->count < 0);
}
/* The size of the argument for each %OVS_KEY_ATTR_* Netlink attribute. */
}
/**
- * flow_from_nlattrs - parses Netlink attributes into a flow key.
+ * ovs_flow_from_nlattrs - parses Netlink attributes into a flow key.
* @swkey: receives the extracted flow key.
* @key_lenp: number of bytes used in @swkey.
* @attr: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute
* sequence.
*/
-int flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
+int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp,
const struct nlattr *attr)
{
const struct nlattr *a[OVS_KEY_ATTR_MAX + 1];
}
/**
- * flow_metadata_from_nlattrs - parses Netlink attributes into a flow key.
+ * ovs_flow_metadata_from_nlattrs - parses Netlink attributes into a flow key.
* @in_port: receives the extracted input port.
* @tun_id: receives the extracted tunnel ID.
* @key: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute
* get the metadata, that is, the parts of the flow key that cannot be
* extracted from the packet itself.
*/
-int flow_metadata_from_nlattrs(u32 *priority, u16 *in_port, __be64 *tun_id,
- const struct nlattr *attr)
+int ovs_flow_metadata_from_nlattrs(u32 *priority, u16 *in_port, __be64 *tun_id,
+ const struct nlattr *attr)
{
const struct nlattr *nla;
int rem;
return 0;
}
-int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
+int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb)
{
struct ovs_key_ethernet *eth_key;
struct nlattr *nla, *encap;
/* Initializes the flow module.
* Returns zero if successful or a negative error code. */
-int flow_init(void)
+int ovs_flow_init(void)
{
flow_cache = kmem_cache_create("sw_flow", sizeof(struct sw_flow), 0,
0, NULL);
if (flow_cache == NULL)
return -ENOMEM;
- get_random_bytes(&hash_seed, sizeof(hash_seed));
-
return 0;
}
/* Uninitializes the flow module. */
-void flow_exit(void)
+void ovs_flow_exit(void)
{
kmem_cache_destroy(flow_cache);
}
git://git.onelab.eu / sliver-openvswitch.git / blobdiff