static char *sctp_hmac_alg = SCTP_COOKIE_HMAC_ALG;
extern struct kmem_cache *sctp_bucket_cachep;
+extern int sysctl_sctp_mem[3];
+extern int sysctl_sctp_rmem[3];
+extern int sysctl_sctp_wmem[3];
+
+int sctp_memory_pressure;
+atomic_t sctp_memory_allocated;
+atomic_t sctp_sockets_allocated;
+
+static void sctp_enter_memory_pressure(void)
+{
+ sctp_memory_pressure = 1;
+}
+
/* Get the sndbuf space available at the time on the association. */
static inline int sctp_wspace(struct sctp_association *asoc)
{
- struct sock *sk = asoc->base.sk;
- int amt = 0;
+ int amt;
- if (asoc->ep->sndbuf_policy) {
- /* make sure that no association uses more than sk_sndbuf */
- amt = sk->sk_sndbuf - asoc->sndbuf_used;
+ if (asoc->ep->sndbuf_policy)
+ amt = asoc->sndbuf_used;
+ else
+ amt = atomic_read(&asoc->base.sk->sk_wmem_alloc);
+
+ if (amt >= asoc->base.sk->sk_sndbuf) {
+ if (asoc->base.sk->sk_userlocks & SOCK_SNDBUF_LOCK)
+ amt = 0;
+ else {
+ amt = sk_stream_wspace(asoc->base.sk);
+ if (amt < 0)
+ amt = 0;
+ }
} else {
- /* do socket level accounting */
- amt = sk->sk_sndbuf - atomic_read(&sk->sk_wmem_alloc);
+ amt = asoc->base.sk->sk_sndbuf - amt;
}
-
- if (amt < 0)
- amt = 0;
-
return amt;
}
sizeof(struct sctp_chunk);
atomic_add(sizeof(struct sctp_chunk), &sk->sk_wmem_alloc);
+ sk_charge_skb(sk, chunk->skb);
}
/* Verify that this is a valid address. */
* The function sctp_get_port_local() does duplicate address
* detection.
*/
+ addr->v4.sin_port = htons(snum);
if ((ret = sctp_get_port_local(sk, addr))) {
if (ret == (long) sk) {
/* This endpoint has a conflicting address. */
if (!bp->port)
bp->port = inet_sk(sk)->num;
- /* Add the address to the bind address list. */
- sctp_local_bh_disable();
- sctp_write_lock(&ep->base.addr_lock);
-
- /* Use GFP_ATOMIC since BHs are disabled. */
+ /* Add the address to the bind address list.
+ * Use GFP_ATOMIC since BHs will be disabled.
+ */
ret = sctp_add_bind_addr(bp, addr, 1, GFP_ATOMIC);
- sctp_write_unlock(&ep->base.addr_lock);
- sctp_local_bh_enable();
/* Copy back into socket for getsockname() use. */
if (!ret) {
*
* Only sctp_setsockopt_bindx() is supposed to call this function.
*/
-int sctp_bindx_add(struct sock *sk, struct sockaddr *addrs, int addrcnt)
+static int sctp_bindx_add(struct sock *sk, struct sockaddr *addrs, int addrcnt)
{
int cnt;
int retval = 0;
if (i < addrcnt)
continue;
- /* Use the first address in bind addr list of association as
- * Address Parameter of ASCONF CHUNK.
+ /* Use the first valid address in bind addr list of
+ * association as Address Parameter of ASCONF CHUNK.
*/
- sctp_read_lock(&asoc->base.addr_lock);
bp = &asoc->base.bind_addr;
p = bp->address_list.next;
laddr = list_entry(p, struct sctp_sockaddr_entry, list);
- sctp_read_unlock(&asoc->base.addr_lock);
-
chunk = sctp_make_asconf_update_ip(asoc, &laddr->a, addrs,
addrcnt, SCTP_PARAM_ADD_IP);
if (!chunk) {
/* Add the new addresses to the bind address list with
* use_as_src set to 0.
*/
- sctp_local_bh_disable();
- sctp_write_lock(&asoc->base.addr_lock);
addr_buf = addrs;
for (i = 0; i < addrcnt; i++) {
addr = (union sctp_addr *)addr_buf;
GFP_ATOMIC);
addr_buf += af->sockaddr_len;
}
- sctp_write_unlock(&asoc->base.addr_lock);
- sctp_local_bh_enable();
}
out:
*
* Only sctp_setsockopt_bindx() is supposed to call this function.
*/
-int sctp_bindx_rem(struct sock *sk, struct sockaddr *addrs, int addrcnt)
+static int sctp_bindx_rem(struct sock *sk, struct sockaddr *addrs, int addrcnt)
{
struct sctp_sock *sp = sctp_sk(sk);
struct sctp_endpoint *ep = sp->ep;
* socket routing and failover schemes. Refer to comments in
* sctp_do_bind(). -daisy
*/
- sctp_local_bh_disable();
- sctp_write_lock(&ep->base.addr_lock);
-
- retval = sctp_del_bind_addr(bp, sa_addr);
-
- sctp_write_unlock(&ep->base.addr_lock);
- sctp_local_bh_enable();
+ retval = sctp_del_bind_addr(bp, sa_addr, call_rcu);
addr_buf += af->sockaddr_len;
err_bindx_rem:
* make sure that we do not delete all the addresses in the
* association.
*/
- sctp_read_lock(&asoc->base.addr_lock);
bp = &asoc->base.bind_addr;
laddr = sctp_find_unmatch_addr(bp, (union sctp_addr *)addrs,
addrcnt, sp);
- sctp_read_unlock(&asoc->base.addr_lock);
if (!laddr)
continue;
+ /* We do not need RCU protection throughout this loop
+ * because this is done under a socket lock from the
+ * setsockopt call.
+ */
chunk = sctp_make_asconf_update_ip(asoc, laddr, addrs, addrcnt,
SCTP_PARAM_DEL_IP);
if (!chunk) {
/* Reset use_as_src flag for the addresses in the bind address
* list that are to be deleted.
*/
- sctp_local_bh_disable();
- sctp_write_lock(&asoc->base.addr_lock);
addr_buf = addrs;
for (i = 0; i < addrcnt; i++) {
laddr = (union sctp_addr *)addr_buf;
af = sctp_get_af_specific(laddr->v4.sin_family);
- list_for_each(pos1, &bp->address_list) {
- saddr = list_entry(pos1,
- struct sctp_sockaddr_entry,
- list);
+ list_for_each_entry(saddr, &bp->address_list, list) {
if (sctp_cmp_addr_exact(&saddr->a, laddr))
saddr->use_as_src = 0;
}
addr_buf += af->sockaddr_len;
}
- sctp_write_unlock(&asoc->base.addr_lock);
- sctp_local_bh_enable();
/* Update the route and saddr entries for all the transports
* as some of the addresses in the bind address list are
int err = 0;
int addrcnt = 0;
int walk_size = 0;
- union sctp_addr *sa_addr;
+ union sctp_addr *sa_addr = NULL;
void *addr_buf;
unsigned short port;
unsigned int f_flags = 0;
goto out_free;
}
- err = sctp_verify_addr(sk, sa_addr, af->sockaddr_len);
+ /* Save current address so we can work with it */
+ memcpy(&to, sa_addr, af->sockaddr_len);
+
+ err = sctp_verify_addr(sk, &to, af->sockaddr_len);
if (err)
goto out_free;
if (asoc && asoc->peer.port && asoc->peer.port != port)
goto out_free;
- memcpy(&to, sa_addr, af->sockaddr_len);
/* Check if there already is a matching association on the
* endpoint (other than the one created here).
*/
- asoc2 = sctp_endpoint_lookup_assoc(ep, sa_addr, &transport);
+ asoc2 = sctp_endpoint_lookup_assoc(ep, &to, &transport);
if (asoc2 && asoc2 != asoc) {
if (asoc2->state >= SCTP_STATE_ESTABLISHED)
err = -EISCONN;
* make sure that there is no peeled-off association matching
* the peer address even on another socket.
*/
- if (sctp_endpoint_is_peeled_off(ep, sa_addr)) {
+ if (sctp_endpoint_is_peeled_off(ep, &to)) {
err = -EADDRNOTAVAIL;
goto out_free;
}
}
}
- scope = sctp_scope(sa_addr);
+ scope = sctp_scope(&to);
asoc = sctp_association_new(ep, sk, scope, GFP_KERNEL);
if (!asoc) {
err = -ENOMEM;
}
/* Prime the peer's transport structures. */
- transport = sctp_assoc_add_peer(asoc, sa_addr, GFP_KERNEL,
+ transport = sctp_assoc_add_peer(asoc, &to, GFP_KERNEL,
SCTP_UNKNOWN);
if (!transport) {
err = -ENOMEM;
/* Initialize sk's dport and daddr for getpeername() */
inet_sk(sk)->dport = htons(asoc->peer.port);
- af = sctp_get_af_specific(to.sa.sa_family);
- af->to_sk_daddr(&to, sk);
+ af = sctp_get_af_specific(sa_addr->sa.sa_family);
+ af->to_sk_daddr(sa_addr, sk);
sk->sk_err = 0;
/* in-kernel sockets don't generally have a file allocated to them
goto out_unlock;
}
if (sinfo_flags & SCTP_ABORT) {
- struct sctp_chunk *chunk;
chunk = sctp_make_abort_user(asoc, msg, msg_len);
if (!chunk) {
sp->hmac = NULL;
SCTP_DBG_OBJCNT_INC(sock);
+ atomic_inc(&sctp_sockets_allocated);
return 0;
}
/* Release our hold on the endpoint. */
ep = sctp_sk(sk)->ep;
sctp_endpoint_free(ep);
-
+ atomic_dec(&sctp_sockets_allocated);
return 0;
}
sctp_assoc_t id;
struct sctp_bind_addr *bp;
struct sctp_association *asoc;
- struct list_head *pos, *temp;
struct sctp_sockaddr_entry *addr;
- rwlock_t *addr_lock;
int cnt = 0;
if (len < sizeof(sctp_assoc_t))
*/
if (0 == id) {
bp = &sctp_sk(sk)->ep->base.bind_addr;
- addr_lock = &sctp_sk(sk)->ep->base.addr_lock;
} else {
asoc = sctp_id2assoc(sk, id);
if (!asoc)
return -EINVAL;
bp = &asoc->base.bind_addr;
- addr_lock = &asoc->base.addr_lock;
}
- sctp_read_lock(addr_lock);
-
/* If the endpoint is bound to 0.0.0.0 or ::0, count the valid
* addresses from the global local address list.
*/
addr = list_entry(bp->address_list.next,
struct sctp_sockaddr_entry, list);
if (sctp_is_any(&addr->a)) {
- list_for_each_safe(pos, temp, &sctp_local_addr_list) {
- addr = list_entry(pos,
- struct sctp_sockaddr_entry,
- list);
+ rcu_read_lock();
+ list_for_each_entry_rcu(addr,
+ &sctp_local_addr_list, list) {
+ if (!addr->valid)
+ continue;
+
if ((PF_INET == sk->sk_family) &&
(AF_INET6 == addr->a.sa.sa_family))
continue;
+
cnt++;
}
+ rcu_read_unlock();
} else {
cnt = 1;
}
goto done;
}
- list_for_each(pos, &bp->address_list) {
+ /* Protection on the bound address list is not needed,
+ * since in the socket option context we hold the socket lock,
+ * so there is no way that the bound address list can change.
+ */
+ list_for_each_entry(addr, &bp->address_list, list) {
cnt ++;
}
-
done:
- sctp_read_unlock(addr_lock);
return cnt;
}
int max_addrs, void *to,
int *bytes_copied)
{
- struct list_head *pos, *next;
struct sctp_sockaddr_entry *addr;
union sctp_addr temp;
int cnt = 0;
int addrlen;
- list_for_each_safe(pos, next, &sctp_local_addr_list) {
- addr = list_entry(pos, struct sctp_sockaddr_entry, list);
+ rcu_read_lock();
+ list_for_each_entry_rcu(addr, &sctp_local_addr_list, list) {
+ if (!addr->valid)
+ continue;
+
if ((PF_INET == sk->sk_family) &&
(AF_INET6 == addr->a.sa.sa_family))
continue;
cnt ++;
if (cnt >= max_addrs) break;
}
+ rcu_read_unlock();
return cnt;
}
static int sctp_copy_laddrs(struct sock *sk, __u16 port, void *to,
size_t space_left, int *bytes_copied)
{
- struct list_head *pos, *next;
struct sctp_sockaddr_entry *addr;
union sctp_addr temp;
int cnt = 0;
int addrlen;
- list_for_each_safe(pos, next, &sctp_local_addr_list) {
- addr = list_entry(pos, struct sctp_sockaddr_entry, list);
+ rcu_read_lock();
+ list_for_each_entry_rcu(addr, &sctp_local_addr_list, list) {
+ if (!addr->valid)
+ continue;
+
if ((PF_INET == sk->sk_family) &&
(AF_INET6 == addr->a.sa.sa_family))
continue;
sctp_get_pf_specific(sk->sk_family)->addr_v4map(sctp_sk(sk),
&temp);
addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len;
- if (space_left < addrlen)
- return -ENOMEM;
+ if (space_left < addrlen) {
+ cnt = -ENOMEM;
+ break;
+ }
memcpy(to, &temp, addrlen);
to += addrlen;
space_left -= addrlen;
*bytes_copied += addrlen;
}
+ rcu_read_unlock();
return cnt;
}
{
struct sctp_bind_addr *bp;
struct sctp_association *asoc;
- struct list_head *pos;
int cnt = 0;
struct sctp_getaddrs_old getaddrs;
struct sctp_sockaddr_entry *addr;
union sctp_addr temp;
struct sctp_sock *sp = sctp_sk(sk);
int addrlen;
- rwlock_t *addr_lock;
int err = 0;
void *addrs;
void *buf;
*/
if (0 == getaddrs.assoc_id) {
bp = &sctp_sk(sk)->ep->base.bind_addr;
- addr_lock = &sctp_sk(sk)->ep->base.addr_lock;
} else {
asoc = sctp_id2assoc(sk, getaddrs.assoc_id);
if (!asoc)
return -EINVAL;
bp = &asoc->base.bind_addr;
- addr_lock = &asoc->base.addr_lock;
}
to = getaddrs.addrs;
if (!addrs)
return -ENOMEM;
- sctp_read_lock(addr_lock);
-
/* If the endpoint is bound to 0.0.0.0 or ::0, get the valid
* addresses from the global local address list.
*/
}
buf = addrs;
- list_for_each(pos, &bp->address_list) {
- addr = list_entry(pos, struct sctp_sockaddr_entry, list);
+ /* Protection on the bound address list is not needed since
+ * in the socket option context we hold a socket lock and
+ * thus the bound address list can't change.
+ */
+ list_for_each_entry(addr, &bp->address_list, list) {
memcpy(&temp, &addr->a, sizeof(temp));
sctp_get_pf_specific(sk->sk_family)->addr_v4map(sp, &temp);
addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len;
}
copy_getaddrs:
- sctp_read_unlock(addr_lock);
-
/* copy the entire address list into the user provided space */
if (copy_to_user(to, addrs, bytes_copied)) {
err = -EFAULT;
{
struct sctp_bind_addr *bp;
struct sctp_association *asoc;
- struct list_head *pos;
int cnt = 0;
struct sctp_getaddrs getaddrs;
struct sctp_sockaddr_entry *addr;
union sctp_addr temp;
struct sctp_sock *sp = sctp_sk(sk);
int addrlen;
- rwlock_t *addr_lock;
int err = 0;
size_t space_left;
int bytes_copied = 0;
*/
if (0 == getaddrs.assoc_id) {
bp = &sctp_sk(sk)->ep->base.bind_addr;
- addr_lock = &sctp_sk(sk)->ep->base.addr_lock;
} else {
asoc = sctp_id2assoc(sk, getaddrs.assoc_id);
if (!asoc)
return -EINVAL;
bp = &asoc->base.bind_addr;
- addr_lock = &asoc->base.addr_lock;
}
to = optval + offsetof(struct sctp_getaddrs,addrs);
if (!addrs)
return -ENOMEM;
- sctp_read_lock(addr_lock);
-
/* If the endpoint is bound to 0.0.0.0 or ::0, get the valid
* addresses from the global local address list.
*/
space_left, &bytes_copied);
if (cnt < 0) {
err = cnt;
- goto error;
+ goto out;
}
goto copy_getaddrs;
}
}
buf = addrs;
- list_for_each(pos, &bp->address_list) {
- addr = list_entry(pos, struct sctp_sockaddr_entry, list);
+ /* Protection on the bound address list is not needed since
+ * in the socket option context we hold a socket lock and
+ * thus the bound address list can't change.
+ */
+ list_for_each_entry(addr, &bp->address_list, list) {
memcpy(&temp, &addr->a, sizeof(temp));
sctp_get_pf_specific(sk->sk_family)->addr_v4map(sp, &temp);
addrlen = sctp_get_af_specific(temp.sa.sa_family)->sockaddr_len;
if (space_left < addrlen) {
err = -ENOMEM; /*fixme: right error?*/
- goto error;
+ goto out;
}
memcpy(buf, &temp, addrlen);
buf += addrlen;
}
copy_getaddrs:
- sctp_read_unlock(addr_lock);
-
if (copy_to_user(to, addrs, bytes_copied)) {
err = -EFAULT;
- goto error;
+ goto out;
}
if (put_user(cnt, &((struct sctp_getaddrs __user *)optval)->addr_num)) {
err = -EFAULT;
- goto error;
+ goto out;
}
if (put_user(bytes_copied, optlen))
err = -EFAULT;
-error:
+out:
kfree(addrs);
return err;
}
char __user *optval,
int __user *optlen)
{
- u32 val;
+ u32 val;
if (len < sizeof(u32))
return -EINVAL;
char __user *optval,
int __user *optlen)
{
- int val;
+ int val;
if (len < sizeof(int))
return -EINVAL;
sctp_unhash_endpoint(ep);
sk->sk_state = SCTP_SS_CLOSED;
+ return 0;
}
/* Return if we are already listening. */
sctp_unhash_endpoint(ep);
sk->sk_state = SCTP_SS_CLOSED;
+ return 0;
}
if (sctp_sstate(sk, LISTENING))
atomic_sub(sizeof(struct sctp_chunk), &sk->sk_wmem_alloc);
+ /*
+ * This undoes what is done via sk_charge_skb
+ */
+ sk->sk_wmem_queued -= skb->truesize;
+ sk->sk_forward_alloc += skb->truesize;
+
sock_wfree(skb);
__sctp_write_space(asoc);
struct sctp_ulpevent *event = sctp_skb2event(skb);
atomic_sub(event->rmem_len, &sk->sk_rmem_alloc);
+
+ /*
+ * Mimic the behavior of sk_stream_rfree
+ */
+ sk->sk_forward_alloc += event->rmem_len;
}
return err;
}
-void sctp_wait_for_close(struct sock *sk, long timeout)
+static void sctp_wait_for_close(struct sock *sk, long timeout)
{
DEFINE_WAIT(wait);
sctp_release_sock(newsk);
}
+
/* This proto struct describes the ULP interface for SCTP. */
struct proto sctp_prot = {
.name = "SCTP",
.unhash = sctp_unhash,
.get_port = sctp_get_port,
.obj_size = sizeof(struct sctp_sock),
+ .sysctl_mem = sysctl_sctp_mem,
+ .sysctl_rmem = sysctl_sctp_rmem,
+ .sysctl_wmem = sysctl_sctp_wmem,
+ .memory_pressure = &sctp_memory_pressure,
+ .enter_memory_pressure = sctp_enter_memory_pressure,
+ .memory_allocated = &sctp_memory_allocated,
};
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
.unhash = sctp_unhash,
.get_port = sctp_get_port,
.obj_size = sizeof(struct sctp6_sock),
+ .sysctl_mem = sysctl_sctp_mem,
+ .sysctl_rmem = sysctl_sctp_rmem,
+ .sysctl_wmem = sysctl_sctp_wmem,
+ .memory_pressure = &sctp_memory_pressure,
+ .enter_memory_pressure = sctp_enter_memory_pressure,
+ .memory_allocated = &sctp_memory_allocated,
};
#endif /* defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) */
git://bbs.cooldavid.org / net-next-2.6.git / blobdiff