static int packet_release(struct socket *sock)
{
struct sock *sk = sock->sk;
struct sock **skp;
if (!sk)
return 0;
write_lock_bh(&packet_sklist_lock);
for (skp = &packet_sklist; *skp; skp = &(*skp)->next) {
if (*skp == sk) {
*skp = sk->next;
__sock_put(sk);
break;
}
}
write_unlock_bh(&packet_sklist_lock);
/*
* Unhook packet receive handler.
*/
if (sk->protinfo.af_packet->running) {
/*
* Remove the protocol hook
*/
dev_remove_pack(&sk->protinfo.af_packet->prot_hook);
sk->protinfo.af_packet->running = 0;
__sock_put(sk);
}
#ifdef CONFIG_PACKET_MULTICAST
packet_flush_mclist(sk);
#endif
#ifdef CONFIG_PACKET_MMAP
if (sk->protinfo.af_packet->pg_vec) {
struct tpacket_req req;
memset(&req, 0, sizeof(req));
packet_set_ring(sk, &req, 1);
}
#endif
/*
* Now the socket is dead. No more input will appear.
*/
sock_orphan(sk);
sock->sk = NULL;
/* Purge queues */
skb_queue_purge(&sk->receive_queue);
sock_put(sk);
return 0;
}
static void l2cap_sock_clear_timer(struct sock *sk)
{
BT_DBG("sock %p state %d", sk, sk->state);
if (timer_pending(&sk->timer) && del_timer(&sk->timer))
__sock_put(sk);
}
static int netlink_release(struct socket *sock)
{
struct sock *sk = sock->sk;
if (!sk)
return 0;
netlink_remove(sk);
spin_lock(&sk->protinfo.af_netlink->cb_lock);
if (sk->protinfo.af_netlink->cb) {
sk->protinfo.af_netlink->cb->done(sk->protinfo.af_netlink->cb);
netlink_destroy_callback(sk->protinfo.af_netlink->cb);
sk->protinfo.af_netlink->cb = NULL;
__sock_put(sk);
}
spin_unlock(&sk->protinfo.af_netlink->cb_lock);
/* OK. Socket is unlinked, and, therefore,
no new packets will arrive */
sock_orphan(sk);
sock->sk = NULL;
wake_up_interruptible_all(&sk->protinfo.af_netlink->wait);
skb_queue_purge(&sk->write_queue);
sock_put(sk);
return 0;
}
static void sco_sock_clear_timer(struct sock *sk)
{
BT_DBG("sock %p state %d", sk, sk->state);
cyg_alarm_disable(sk->timeout_alarm_handle);
cyg_alarm_delete(sk->timeout_alarm_handle);
// if (timer_pending(&sk->timer) && del_timer(&sk->timer))
__sock_put(sk);
}
void tcp_clear_xmit_timers(struct sock *sk)
{
struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
tp->pending = 0;
if (timer_pending(&tp->retransmit_timer) &&
del_timer(&tp->retransmit_timer))
__sock_put(sk);
tp->ack.pending = 0;
tp->ack.blocked = 0;
if (timer_pending(&tp->delack_timer) &&
del_timer(&tp->delack_timer))
__sock_put(sk);
if(timer_pending(&sk->timer) && del_timer(&sk->timer))
__sock_put(sk);
}
static int packet_release(struct socket *sock)
{
struct sock *sk = sock->sk;
struct packet_sock *po;
if (!sk)
return 0;
po = pkt_sk(sk);
write_lock_bh(&packet_sklist_lock);
sk_del_node_init(sk);
write_unlock_bh(&packet_sklist_lock);
/*
* Unhook packet receive handler.
*/
if (po->running) {
/*
* Remove the protocol hook
*/
dev_remove_pack(&po->prot_hook);
po->running = 0;
po->num = 0;
__sock_put(sk);
}
#ifdef CONFIG_PACKET_MULTICAST
packet_flush_mclist(sk);
#endif
#ifdef CONFIG_PACKET_MMAP
if (po->pg_vec) {
struct tpacket_req req;
memset(&req, 0, sizeof(req));
packet_set_ring(sk, &req, 1);
}
#endif
/*
* Now the socket is dead. No more input will appear.
*/
sock_orphan(sk);
sock->sk = NULL;
/* Purge queues */
skb_queue_purge(&sk->sk_receive_queue);
sock_put(sk);
return 0;
}
static void raw_v6_unhash(struct sock *sk)
{
write_lock_bh(&raw_v6_lock);
if (sk->pprev) {
if (sk->next)
sk->next->pprev = sk->pprev;
*sk->pprev = sk->next;
sk->pprev = NULL;
sock_prot_dec_use(sk->prot);
__sock_put(sk);
}
write_unlock_bh(&raw_v6_lock);
}
static void udp_v4_unhash(struct sock *sk)
{
write_lock_bh(&udp_hash_lock);
if (sk->pprev) {
if (sk->next)
sk->next->pprev = sk->pprev;
*sk->pprev = sk->next;
sk->pprev = NULL;
sk->num = 0;
sock_prot_dec_use(sk->prot);
__sock_put(sk);
}
write_unlock_bh(&udp_hash_lock);
}
static void netlink_remove(struct sock *sk)
{
struct sock **skp;
netlink_table_grab();
for (skp = &nl_table[sk->protocol]; *skp; skp = &((*skp)->next)) {
if (*skp == sk) {
*skp = sk->next;
__sock_put(sk);
break;
}
}
netlink_table_ungrab();
}
void bluez_sock_unlink(struct bluez_sock_list *l, struct sock *sk)
{
struct sock **skp;
write_lock(&l->lock);
for (skp = &l->head; *skp; skp = &((*skp)->next)) {
if (*skp == sk) {
*skp = sk->next;
__sock_put(sk);
break;
}
}
write_unlock(&l->lock);
}
static int packet_do_bind(struct sock *sk, struct net_device *dev, int protocol)
{
struct packet_sock *po = pkt_sk(sk);
/*
* Detach an existing hook if present.
*/
lock_sock(sk);
spin_lock(&po->bind_lock);
if (po->running) {
__sock_put(sk);
po->running = 0;
po->num = 0;
spin_unlock(&po->bind_lock);
dev_remove_pack(&po->prot_hook);
spin_lock(&po->bind_lock);
}
po->num = protocol;
po->prot_hook.type = protocol;
po->prot_hook.dev = dev;
po->ifindex = dev ? dev->ifindex : 0;
if (protocol == 0)
goto out_unlock;
if (dev) {
if (dev->flags&IFF_UP) {
dev_add_pack(&po->prot_hook);
sock_hold(sk);
po->running = 1;
} else {
sk->sk_err = ENETDOWN;
if (!sock_flag(sk, SOCK_DEAD))
sk->sk_error_report(sk);
}
} else {
dev_add_pack(&po->prot_hook);
sock_hold(sk);
po->running = 1;
}
out_unlock:
spin_unlock(&po->bind_lock);
release_sock(sk);
return 0;
}
static inline void l2cap_chan_unlink(struct l2cap_chan_list *l, struct sock *sk)
{
struct sock *next = l2cap_pi(sk)->next_c, *prev = l2cap_pi(sk)->prev_c;
write_lock(&l->lock);
if (sk == l->head)
l->head = next;
if (next)
l2cap_pi(next)->prev_c = prev;
if (prev)
l2cap_pi(prev)->next_c = next;
write_unlock(&l->lock);
__sock_put(sk);
}
static int packet_do_bind(struct sock *sk, struct net_device *dev, int protocol)
{
/*
* Detach an existing hook if present.
*/
lock_sock(sk);
spin_lock(&sk->protinfo.af_packet->bind_lock);
if (sk->protinfo.af_packet->running) {
dev_remove_pack(&sk->protinfo.af_packet->prot_hook);
__sock_put(sk);
sk->protinfo.af_packet->running = 0;
}
sk->num = protocol;
sk->protinfo.af_packet->prot_hook.type = protocol;
sk->protinfo.af_packet->prot_hook.dev = dev;
sk->protinfo.af_packet->ifindex = dev ? dev->ifindex : 0;
if (protocol == 0)
goto out_unlock;
if (dev) {
if (dev->flags&IFF_UP) {
dev_add_pack(&sk->protinfo.af_packet->prot_hook);
sock_hold(sk);
sk->protinfo.af_packet->running = 1;
} else {
sk->err = ENETDOWN;
if (!sk->dead)
sk->error_report(sk);
}
} else {
dev_add_pack(&sk->protinfo.af_packet->prot_hook);
sock_hold(sk);
sk->protinfo.af_packet->running = 1;
}
out_unlock:
spin_unlock(&sk->protinfo.af_packet->bind_lock);
release_sock(sk);
return 0;
}
void pn_sock_unbind_all_res(struct sock *sk)
{
unsigned res, match = 0;
mutex_lock(&resource_mutex);
for (res = 0; res < 256; res++) {
if (pnres.sk[res] == sk) {
RCU_INIT_POINTER(pnres.sk[res], NULL);
match++;
}
}
mutex_unlock(&resource_mutex);
while (match > 0) {
__sock_put(sk);
match--;
}
/* Caller is responsible for RCU sync before final sock_put() */
}
static struct sock *pep_sock_accept(struct sock *sk, int flags, int *errp)
{
struct pep_sock *pn = pep_sk(sk);
struct sock *newsk = NULL;
struct sk_buff *oskb;
int err;
lock_sock(sk);
err = pep_wait_connreq(sk, flags & O_NONBLOCK);
if (err)
goto out;
newsk = __sk_head(&pn->ackq);
oskb = skb_dequeue(&newsk->sk_receive_queue);
err = pep_accept_conn(newsk, oskb);
if (err) {
skb_queue_head(&newsk->sk_receive_queue, oskb);
newsk = NULL;
goto out;
}
kfree_skb(oskb);
sock_hold(sk);
pep_sk(newsk)->listener = sk;
sock_hold(newsk);
sk_del_node_init(newsk);
sk_acceptq_removed(sk);
sk_add_node(newsk, &pn->hlist);
__sock_put(newsk);
out:
release_sock(sk);
*errp = err;
return newsk;
}
static int netlink_release(struct socket *sock)
{
struct sock *sk = sock->sk;
if (!sk)
return 0;
netlink_remove(sk);
spin_lock(&sk->protinfo.af_netlink->cb_lock);
if (sk->protinfo.af_netlink->cb) {
sk->protinfo.af_netlink->cb->done(sk->protinfo.af_netlink->cb);
netlink_destroy_callback(sk->protinfo.af_netlink->cb);
sk->protinfo.af_netlink->cb = NULL;
__sock_put(sk);
}
spin_unlock(&sk->protinfo.af_netlink->cb_lock);
/* OK. Socket is unlinked, and, therefore,
no new packets will arrive */
sock_orphan(sk);
sock->sk = NULL;
wake_up_interruptible_all(&sk->protinfo.af_netlink->wait);
skb_queue_purge(&sk->write_queue);
if (sk->protinfo.af_netlink->pid && !sk->protinfo.af_netlink->groups) {
struct netlink_notify n = { protocol:sk->protocol,
pid:sk->protinfo.af_netlink->pid };
notifier_call_chain(&netlink_chain, NETLINK_URELEASE, &n);
}
sock_put(sk);
return 0;
}
static int netlink_release(struct socket *sock)
{
struct sock *sk = sock->sk;
struct netlink_opt *nlk;
if (!sk)
return 0;
netlink_remove(sk);
nlk = nlk_sk(sk);
spin_lock(&nlk->cb_lock);
if (nlk->cb) {
nlk->cb->done(nlk->cb);
netlink_destroy_callback(nlk->cb);
nlk->cb = NULL;
__sock_put(sk);
}
spin_unlock(&nlk->cb_lock);
/* OK. Socket is unlinked, and, therefore,
no new packets will arrive */
sock_orphan(sk);
sock->sk = NULL;
wake_up_interruptible_all(&nlk->wait);
skb_queue_purge(&sk->sk_write_queue);
if (nlk->pid && !nlk->groups) {
struct netlink_notify n = {
.protocol = sk->sk_protocol,
.pid = nlk->pid,
};
notifier_call_chain(&netlink_chain, NETLINK_URELEASE, &n);
}
sock_put(sk);
return 0;
}
static int netlink_autobind(struct socket *sock)
{
struct sock *sk = sock->sk;
struct nl_pid_hash *hash = &nl_table[sk->sk_protocol].hash;
struct hlist_head *head;
struct sock *osk;
struct hlist_node *node;
s32 pid = current->pid;
int err;
static s32 rover = -4097;
retry:
cond_resched();
netlink_table_grab();
head = nl_pid_hashfn(hash, pid);
sk_for_each(osk, node, head) {
if (nlk_sk(osk)->pid == pid) {
/* Bind collision, search negative pid values. */
pid = rover--;
if (rover > -4097)
rover = -4097;
netlink_table_ungrab();
goto retry;
}
}
netlink_table_ungrab();
err = netlink_insert(sk, pid);
if (err == -EADDRINUSE)
goto retry;
nlk_sk(sk)->groups = 0;
return 0;
}
static inline int netlink_capable(struct socket *sock, unsigned int flag)
{
return (nl_nonroot[sock->sk->sk_protocol] & flag) ||
capable(CAP_NET_ADMIN);
}
static int netlink_bind(struct socket *sock, struct sockaddr *addr, int addr_len)
{
struct sock *sk = sock->sk;
struct netlink_opt *nlk = nlk_sk(sk);
struct sockaddr_nl *nladdr = (struct sockaddr_nl *)addr;
int err;
if (nladdr->nl_family != AF_NETLINK)
return -EINVAL;
/* Only superuser is allowed to listen multicasts */
if (nladdr->nl_groups && !netlink_capable(sock, NL_NONROOT_RECV))
return -EPERM;
if (nlk->pid) {
if (nladdr->nl_pid != nlk->pid)
return -EINVAL;
} else {
err = nladdr->nl_pid ?
netlink_insert(sk, nladdr->nl_pid) :
netlink_autobind(sock);
if (err)
return err;
}
if (!nladdr->nl_groups && !nlk->groups)
return 0;
netlink_table_grab();
if (nlk->groups && !nladdr->nl_groups)
__sk_del_bind_node(sk);
else if (!nlk->groups && nladdr->nl_groups)
sk_add_bind_node(sk, &nl_table[sk->sk_protocol].mc_list);
nlk->groups = nladdr->nl_groups;
netlink_table_ungrab();
return 0;
}
static int netlink_connect(struct socket *sock, struct sockaddr *addr,
int alen, int flags)
{
int err = 0;
struct sock *sk = sock->sk;
struct netlink_opt *nlk = nlk_sk(sk);
struct sockaddr_nl *nladdr=(struct sockaddr_nl*)addr;
if (addr->sa_family == AF_UNSPEC) {
sk->sk_state = NETLINK_UNCONNECTED;
nlk->dst_pid = 0;
nlk->dst_groups = 0;
return 0;
}
if (addr->sa_family != AF_NETLINK)
return -EINVAL;
/* Only superuser is allowed to send multicasts */
if (nladdr->nl_groups && !netlink_capable(sock, NL_NONROOT_SEND))
return -EPERM;
if (!nlk->pid)
err = netlink_autobind(sock);
if (err == 0) {
sk->sk_state = NETLINK_CONNECTED;
nlk->dst_pid = nladdr->nl_pid;
nlk->dst_groups = nladdr->nl_groups;
}
return err;
}
int ip6_push_pending_frames(struct sock *sk)
{
struct sk_buff *skb, *tmp_skb;
struct sk_buff **tail_skb;
struct in6_addr final_dst_buf, *final_dst = &final_dst_buf;
struct inet_sock *inet = inet_sk(sk);
struct ipv6_pinfo *np = inet6_sk(sk);
struct ipv6hdr *hdr;
struct ipv6_txoptions *opt = np->cork.opt;
struct rt6_info *rt = (struct rt6_info *)inet->cork.dst;
struct flowi *fl = &inet->cork.fl;
unsigned char proto = fl->proto;
int err = 0;
if ((skb = __skb_dequeue(&sk->sk_write_queue)) == NULL)
goto out;
tail_skb = &(skb_shinfo(skb)->frag_list);
/* move skb->data to ip header from ext header */
if (skb->data < skb_network_header(skb))
__skb_pull(skb, skb_network_offset(skb));
while ((tmp_skb = __skb_dequeue(&sk->sk_write_queue)) != NULL) {
__skb_pull(tmp_skb, skb_network_header_len(skb));
*tail_skb = tmp_skb;
tail_skb = &(tmp_skb->next);
skb->len += tmp_skb->len;
skb->data_len += tmp_skb->len;
skb->truesize += tmp_skb->truesize;
__sock_put(tmp_skb->sk);
tmp_skb->destructor = NULL;
tmp_skb->sk = NULL;
}
/* Allow local fragmentation. */
if (np->pmtudisc < IPV6_PMTUDISC_DO)
skb->local_df = 1;
ipv6_addr_copy(final_dst, &fl->fl6_dst);
__skb_pull(skb, skb_network_header_len(skb));
if (opt && opt->opt_flen)
ipv6_push_frag_opts(skb, opt, &proto);
if (opt && opt->opt_nflen)
ipv6_push_nfrag_opts(skb, opt, &proto, &final_dst);
skb_push(skb, sizeof(struct ipv6hdr));
skb_reset_network_header(skb);
hdr = ipv6_hdr(skb);
*(__be32*)hdr = fl->fl6_flowlabel |
htonl(0x60000000 | ((int)np->cork.tclass << 20));
hdr->hop_limit = np->cork.hop_limit;
hdr->nexthdr = proto;
ipv6_addr_copy(&hdr->saddr, &fl->fl6_src);
ipv6_addr_copy(&hdr->daddr, final_dst);
skb->priority = sk->sk_priority;
skb->mark = sk->sk_mark;
skb->dst = dst_clone(&rt->u.dst);
IP6_INC_STATS(rt->rt6i_idev, IPSTATS_MIB_OUTREQUESTS);
if (proto == IPPROTO_ICMPV6) {
struct inet6_dev *idev = ip6_dst_idev(skb->dst);
ICMP6MSGOUT_INC_STATS_BH(idev, icmp6_hdr(skb)->icmp6_type);
ICMP6_INC_STATS_BH(idev, ICMP6_MIB_OUTMSGS);
}
err = ip6_local_out(skb);
if (err) {
if (err > 0)
err = np->recverr ? net_xmit_errno(err) : 0;
if (err)
goto error;
}
out:
ip6_cork_release(inet, np);
return err;
error:
goto out;
}
int ip6_push_pending_frames(struct sock *sk)
{
struct sk_buff *skb, *tmp_skb;
struct sk_buff **tail_skb;
struct in6_addr final_dst_buf, *final_dst = &final_dst_buf;
struct inet_sock *inet = inet_sk(sk);
struct ipv6_pinfo *np = inet6_sk(sk);
struct ipv6hdr *hdr;
struct ipv6_txoptions *opt = np->cork.opt;
struct rt6_info *rt = np->cork.rt;
struct flowi *fl = &inet->cork.fl;
unsigned char proto = fl->proto;
int err = 0;
if ((skb = __skb_dequeue(&sk->sk_write_queue)) == NULL)
goto out;
tail_skb = &(skb_shinfo(skb)->frag_list);
/* move skb->data to ip header from ext header */
if (skb->data < skb->nh.raw)
__skb_pull(skb, skb->nh.raw - skb->data);
while ((tmp_skb = __skb_dequeue(&sk->sk_write_queue)) != NULL) {
__skb_pull(tmp_skb, skb->h.raw - skb->nh.raw);
*tail_skb = tmp_skb;
tail_skb = &(tmp_skb->next);
skb->len += tmp_skb->len;
skb->data_len += tmp_skb->len;
skb->truesize += tmp_skb->truesize;
__sock_put(tmp_skb->sk);
tmp_skb->destructor = NULL;
tmp_skb->sk = NULL;
}
ipv6_addr_copy(final_dst, &fl->fl6_dst);
__skb_pull(skb, skb->h.raw - skb->nh.raw);
if (opt && opt->opt_flen)
ipv6_push_frag_opts(skb, opt, &proto);
if (opt && opt->opt_nflen)
ipv6_push_nfrag_opts(skb, opt, &proto, &final_dst);
skb->nh.ipv6h = hdr = (struct ipv6hdr*) skb_push(skb, sizeof(struct ipv6hdr));
*(u32*)hdr = fl->fl6_flowlabel | htonl(0x60000000);
if (skb->len <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN)
hdr->payload_len = htons(skb->len - sizeof(struct ipv6hdr));
else
hdr->payload_len = 0;
hdr->hop_limit = np->cork.hop_limit;
hdr->nexthdr = proto;
ipv6_addr_copy(&hdr->saddr, &fl->fl6_src);
ipv6_addr_copy(&hdr->daddr, final_dst);
skb->dst = dst_clone(&rt->u.dst);
IP6_INC_STATS(IPSTATS_MIB_OUTREQUESTS);
err = NF_HOOK(PF_INET6, NF_IP6_LOCAL_OUT, skb, NULL, skb->dst->dev, dst_output);
if (err) {
if (err > 0)
err = np->recverr ? net_xmit_errno(err) : 0;
if (err)
goto error;
}
out:
inet->cork.flags &= ~IPCORK_OPT;
if (np->cork.opt) {
kfree(np->cork.opt);
np->cork.opt = NULL;
}
if (np->cork.rt) {
dst_release(&np->cork.rt->u.dst);
np->cork.rt = NULL;
inet->cork.flags &= ~IPCORK_ALLFRAG;
}
memset(&inet->cork.fl, 0, sizeof(inet->cork.fl));
return err;
error:
goto out;
}
static struct sock *pep_sock_accept(struct sock *sk, int flags, int *errp)
{
struct pep_sock *pn = pep_sk(sk), *newpn;
struct sock *newsk = NULL;
struct sk_buff *skb;
struct pnpipehdr *hdr;
struct sockaddr_pn dst, src;
int err;
u16 peer_type;
u8 pipe_handle, enabled, n_sb;
u8 aligned = 0;
skb = skb_recv_datagram(sk, 0, flags & O_NONBLOCK, errp);
if (!skb)
return NULL;
lock_sock(sk);
if (sk->sk_state != TCP_LISTEN) {
err = -EINVAL;
goto drop;
}
sk_acceptq_removed(sk);
err = -EPROTO;
if (!pskb_may_pull(skb, sizeof(*hdr) + 4))
goto drop;
hdr = pnp_hdr(skb);
pipe_handle = hdr->pipe_handle;
switch (hdr->state_after_connect) {
case PN_PIPE_DISABLE:
enabled = 0;
break;
case PN_PIPE_ENABLE:
enabled = 1;
break;
default:
pep_reject_conn(sk, skb, PN_PIPE_ERR_INVALID_PARAM,
GFP_KERNEL);
goto drop;
}
peer_type = hdr->other_pep_type << 8;
/* Parse sub-blocks (options) */
n_sb = hdr->data[4];
while (n_sb > 0) {
u8 type, buf[1], len = sizeof(buf);
const u8 *data = pep_get_sb(skb, &type, &len, buf);
if (data == NULL)
goto drop;
switch (type) {
case PN_PIPE_SB_CONNECT_REQ_PEP_SUB_TYPE:
if (len < 1)
goto drop;
peer_type = (peer_type & 0xff00) | data[0];
break;
case PN_PIPE_SB_ALIGNED_DATA:
aligned = data[0] != 0;
break;
}
n_sb--;
}
/* Check for duplicate pipe handle */
newsk = pep_find_pipe(&pn->hlist, &dst, pipe_handle);
if (unlikely(newsk)) {
__sock_put(newsk);
newsk = NULL;
pep_reject_conn(sk, skb, PN_PIPE_ERR_PEP_IN_USE, GFP_KERNEL);
goto drop;
}
/* Create a new to-be-accepted sock */
newsk = sk_alloc(sock_net(sk), PF_PHONET, GFP_KERNEL, sk->sk_prot);
if (!newsk) {
pep_reject_conn(sk, skb, PN_PIPE_ERR_OVERLOAD, GFP_KERNEL);
err = -ENOBUFS;
goto drop;
}
sock_init_data(NULL, newsk);
newsk->sk_state = TCP_SYN_RECV;
newsk->sk_backlog_rcv = pipe_do_rcv;
newsk->sk_protocol = sk->sk_protocol;
newsk->sk_destruct = pipe_destruct;
newpn = pep_sk(newsk);
pn_skb_get_dst_sockaddr(skb, &dst);
pn_skb_get_src_sockaddr(skb, &src);
newpn->pn_sk.sobject = pn_sockaddr_get_object(&dst);
newpn->pn_sk.dobject = pn_sockaddr_get_object(&src);
newpn->pn_sk.resource = pn_sockaddr_get_resource(&dst);
sock_hold(sk);
newpn->listener = sk;
skb_queue_head_init(&newpn->ctrlreq_queue);
newpn->pipe_handle = pipe_handle;
atomic_set(&newpn->tx_credits, 0);
newpn->ifindex = 0;
newpn->peer_type = peer_type;
newpn->rx_credits = 0;
newpn->rx_fc = newpn->tx_fc = PN_LEGACY_FLOW_CONTROL;
newpn->init_enable = enabled;
newpn->aligned = aligned;
err = pep_accept_conn(newsk, skb);
if (err) {
sock_put(newsk);
newsk = NULL;
goto drop;
}
sk_add_node(newsk, &pn->hlist);
drop:
release_sock(sk);
kfree_skb(skb);
*errp = err;
return newsk;
}
static int sdp_init_qp(struct sock *sk, struct rdma_cm_id *id)
{
struct ib_qp_init_attr qp_init_attr = {
.event_handler = sdp_qp_event_handler,
.cap.max_send_wr = SDP_TX_SIZE,
.cap.max_recv_wr = sdp_rx_size,
.cap.max_inline_data = sdp_inline_thresh,
.sq_sig_type = IB_SIGNAL_REQ_WR,
.qp_type = IB_QPT_RC,
};
struct ib_device *device = id->device;
int rc;
sdp_dbg(sk, "%s\n", __func__);
sdp_sk(sk)->max_sge = sdp_get_max_dev_sge(device);
sdp_dbg(sk, "Max sges: %d\n", sdp_sk(sk)->max_sge);
qp_init_attr.cap.max_send_sge = MIN(sdp_sk(sk)->max_sge, SDP_MAX_SEND_SGES);
sdp_dbg(sk, "Setting max send sge to: %d\n", qp_init_attr.cap.max_send_sge);
qp_init_attr.cap.max_recv_sge = MIN(sdp_sk(sk)->max_sge, SDP_MAX_RECV_SGES);
sdp_dbg(sk, "Setting max recv sge to: %d\n", qp_init_attr.cap.max_recv_sge);
sdp_sk(sk)->sdp_dev = ib_get_client_data(device, &sdp_client);
if (!sdp_sk(sk)->sdp_dev) {
sdp_warn(sk, "SDP not available on device %s\n", device->name);
rc = -ENODEV;
goto err_rx;
}
rc = sdp_rx_ring_create(sdp_sk(sk), device);
if (rc)
goto err_rx;
rc = sdp_tx_ring_create(sdp_sk(sk), device);
if (rc)
goto err_tx;
qp_init_attr.recv_cq = sdp_sk(sk)->rx_ring.cq;
qp_init_attr.send_cq = sdp_sk(sk)->tx_ring.cq;
rc = rdma_create_qp(id, sdp_sk(sk)->sdp_dev->pd, &qp_init_attr);
if (rc) {
sdp_warn(sk, "Unable to create QP: %d.\n", rc);
goto err_qp;
}
sdp_sk(sk)->qp = id->qp;
sdp_sk(sk)->ib_device = device;
sdp_sk(sk)->qp_active = 1;
sdp_sk(sk)->context.device = device;
sdp_sk(sk)->inline_thresh = qp_init_attr.cap.max_inline_data;
sdp_dbg(sk, "%s done\n", __func__);
return 0;
err_qp:
sdp_tx_ring_destroy(sdp_sk(sk));
err_tx:
sdp_rx_ring_destroy(sdp_sk(sk));
err_rx:
return rc;
}
static int sdp_get_max_send_frags(u32 buf_size)
{
return MIN(
/* +1 to conpensate on not aligned buffers */
(PAGE_ALIGN(buf_size) >> PAGE_SHIFT) + 1,
SDP_MAX_SEND_SGES - 1);
}
static int sdp_connect_handler(struct sock *sk, struct rdma_cm_id *id,
struct rdma_cm_event *event)
{
struct sockaddr_in *dst_addr;
struct sock *child;
const struct sdp_hh *h;
int rc = 0;
sdp_dbg(sk, "%s %p -> %p\n", __func__, sdp_sk(sk)->id, id);
h = event->param.conn.private_data;
SDP_DUMP_PACKET(sk, "RX", NULL, &h->bsdh);
if (h->ipv_cap & HH_IPV_MASK & ~(HH_IPV4 | HH_IPV6)) {
sdp_warn(sk, "Bad IPV field in SDP Hello header: 0x%x\n",
h->ipv_cap & HH_IPV_MASK);
return -EINVAL;
}
if (!h->max_adverts)
return -EINVAL;
#if (LINUX_VERSION_CODE < KERNEL_VERSION(3, 3, 0))
child = sk_clone(sk, GFP_KERNEL);
#else
child = sk_clone_lock(sk, GFP_KERNEL);
#endif
if (!child)
return -ENOMEM;
sdp_init_sock(child);
dst_addr = (struct sockaddr_in *)&id->route.addr.dst_addr;
sdp_inet_dport(child) = dst_addr->sin_port;
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
if (inet6_sk(sk)) {
struct ipv6_pinfo *newnp;
newnp = inet_sk(child)->pinet6 = sdp_inet6_sk_generic(child);
memcpy(newnp, inet6_sk(sk), sizeof(struct ipv6_pinfo));
if ((h->ipv_cap & HH_IPV_MASK) == HH_IPV4) {
/* V6 mapped */
sdp_inet_daddr(child) = dst_addr->sin_addr.s_addr;
ipv6_addr_set(&child->sk_v6_daddr, 0, 0, htonl(0x0000FFFF),
h->src_addr.ip4.addr);
ipv6_addr_set(&child->sk_v6_rcv_saddr, 0, 0, htonl(0x0000FFFF),
h->dst_addr.ip4.addr);
ipv6_addr_copy(&child->sk_v6_rcv_saddr, &child->sk_v6_daddr);
} else if ((h->ipv_cap & HH_IPV_MASK) == HH_IPV6) {
struct sockaddr_in6 *dst_addr6 = (struct sockaddr_in6 *)dst_addr;
struct sockaddr_in6 *src_addr6 =
(struct sockaddr_in6 *)&id->route.addr.src_addr;
ipv6_addr_copy(&child->sk_v6_daddr, &dst_addr6->sin6_addr);
ipv6_addr_copy(&child->sk_v6_rcv_saddr, &src_addr6->sin6_addr);
ipv6_addr_copy(&newnp->saddr, &src_addr6->sin6_addr);
} else {
sdp_warn(child, "Bad IPV field: 0x%x\n", h->ipv_cap & HH_IPV_MASK);
}
sdp_inet_daddr(child) = sdp_inet_saddr(child) =
sdp_inet_rcv_saddr(child) = LOOPBACK4_IPV6;
} else
#endif
{
sdp_inet_daddr(child) = dst_addr->sin_addr.s_addr;
}
#ifdef SDP_SOCK_HISTORY
sdp_ssk_hist_rename(sk);
#endif
__sock_put(child, SOCK_REF_CLONE);
down_read(&device_removal_lock);
rc = sdp_init_qp(child, id);
if (rc) {
bh_unlock_sock(child);
up_read(&device_removal_lock);
sdp_sk(child)->destructed_already = 1;
#ifdef SDP_SOCK_HISTORY
sdp_ssk_hist_close(child);
#endif
sk_free(child);
return rc;
}
sdp_sk(child)->max_bufs = ntohs(h->bsdh.bufs);
atomic_set(&sdp_sk(child)->tx_ring.credits, sdp_sk(child)->max_bufs);
sdp_sk(child)->min_bufs = tx_credits(sdp_sk(child)) / 4;
sdp_sk(child)->xmit_size_goal = ntohl(h->localrcvsz) -
sizeof(struct sdp_bsdh);
sdp_sk(child)->send_frags = sdp_get_max_send_frags(sdp_sk(child)->xmit_size_goal);
sdp_init_buffers(sdp_sk(child), rcvbuf_initial_size);
id->context = child;
sdp_sk(child)->id = id;
list_add_tail(&sdp_sk(child)->backlog_queue,
&sdp_sk(sk)->backlog_queue);
sdp_sk(child)->parent = sk;
bh_unlock_sock(child);
sdp_add_sock(sdp_sk(child));
up_read(&device_removal_lock);
sdp_exch_state(child, TCPF_LISTEN | TCPF_CLOSE, TCP_SYN_RECV);
/* child->sk_write_space(child); */
/* child->sk_data_ready(child, 0); */
sk->sk_data_ready(sk);
return 0;
}
static int sdp_response_handler(struct sock *sk, struct rdma_cm_id *id,
struct rdma_cm_event *event)
{
const struct sdp_hah *h;
struct sockaddr_in *dst_addr;
sdp_dbg(sk, "%s\n", __func__);
sdp_exch_state(sk, TCPF_SYN_SENT, TCP_ESTABLISHED);
sdp_set_default_moderation(sdp_sk(sk));
if (sock_flag(sk, SOCK_KEEPOPEN))
sdp_start_keepalive_timer(sk);
if (sock_flag(sk, SOCK_DEAD))
return 0;
h = event->param.conn.private_data;
SDP_DUMP_PACKET(sk, "RX", NULL, &h->bsdh);
sdp_sk(sk)->max_bufs = ntohs(h->bsdh.bufs);
atomic_set(&sdp_sk(sk)->tx_ring.credits, sdp_sk(sk)->max_bufs);
sdp_sk(sk)->min_bufs = tx_credits(sdp_sk(sk)) / 4;
sdp_sk(sk)->xmit_size_goal =
ntohl(h->actrcvsz) - sizeof(struct sdp_bsdh);
sdp_sk(sk)->send_frags = sdp_get_max_send_frags(sdp_sk(sk)->xmit_size_goal);
sdp_sk(sk)->xmit_size_goal = MIN(sdp_sk(sk)->xmit_size_goal,
sdp_sk(sk)->send_frags * PAGE_SIZE);
sdp_sk(sk)->poll_cq = 1;
sk->sk_state_change(sk);
sk_wake_async(sk, 0, POLL_OUT);
dst_addr = (struct sockaddr_in *)&id->route.addr.dst_addr;
sdp_inet_dport(sk) = dst_addr->sin_port;
sdp_inet_daddr(sk) = dst_addr->sin_addr.s_addr;
#ifdef SDP_SOCK_HISTORY
sdp_ssk_hist_rename(sk);
#endif
return 0;
}
static int sdp_connected_handler(struct sock *sk)
{
struct sock *parent;
sdp_dbg(sk, "%s\n", __func__);
parent = sdp_sk(sk)->parent;
BUG_ON(!parent);
sdp_exch_state(sk, TCPF_SYN_RECV, TCP_ESTABLISHED);
#ifdef SDP_SOCK_HISTORY
sdp_ssk_hist_rename(sk);
#endif
sdp_set_default_moderation(sdp_sk(sk));
if (sock_flag(sk, SOCK_KEEPOPEN))
sdp_start_keepalive_timer(sk);
if (sock_flag(sk, SOCK_DEAD))
return 0;
lock_sock(parent);
if (!sdp_sk(parent)->id) { /* TODO: look at SOCK_DEAD? */
sdp_dbg(sk, "parent is going away.\n");
goto done;
}
sk_acceptq_added(parent);
sdp_dbg(parent, "%s child connection established\n", __func__);
list_del_init(&sdp_sk(sk)->backlog_queue);
list_add_tail(&sdp_sk(sk)->accept_queue,
&sdp_sk(parent)->accept_queue);
parent->sk_state_change(parent);
sk_wake_async(parent, 0, POLL_OUT);
done:
release_sock(parent);
return 0;
}
static int sdp_disconnected_handler(struct sock *sk)
{
struct sdp_sock *ssk = sdp_sk(sk);
sdp_dbg(sk, "%s\n", __func__);
if (ssk->tx_ring.cq)
if (sdp_xmit_poll(ssk, 1))
sdp_post_sends(ssk, 0);
if (sk->sk_state == TCP_SYN_RECV) {
sdp_connected_handler(sk);
if (rcv_nxt(ssk))
return 0;
}
return -ECONNRESET;
}
int sdp_cma_handler(struct rdma_cm_id *id, struct rdma_cm_event *event)
{
struct rdma_conn_param conn_param;
struct sock *parent = NULL;
struct sock *child = NULL;
struct sock *sk;
struct sdp_hah hah;
struct sdp_hh hh;
int rc = 0, rc2;
sk = id->context;
if (!sk) {
sdp_dbg(NULL, "cm_id is being torn down, event %s\n",
rdma_cm_event_str(event->event));
return event->event == RDMA_CM_EVENT_CONNECT_REQUEST ?
-EINVAL : 0;
}
sdp_add_to_history(sk, rdma_cm_event_str(event->event));
lock_sock_nested(sk, SINGLE_DEPTH_NESTING);
sdp_dbg(sk, "event: %s\n", rdma_cm_event_str(event->event));
if (!sdp_sk(sk)->id) {
sdp_dbg(sk, "socket is being torn down\n");
rc = event->event == RDMA_CM_EVENT_CONNECT_REQUEST ?
-EINVAL : 0;
release_sock(sk);
return rc;
}
switch (event->event) {
case RDMA_CM_EVENT_ADDR_RESOLVED:
if (sdp_link_layer_ib_only &&
rdma_node_get_transport(id->device->node_type) ==
RDMA_TRANSPORT_IB &&
rdma_port_get_link_layer(id->device, id->port_num) !=
IB_LINK_LAYER_INFINIBAND) {
sdp_dbg(sk, "Link layer is: %d. Only IB link layer "
"is allowed\n",
rdma_port_get_link_layer(id->device,
id->port_num));
rc = -ENETUNREACH;
break;
}
rc = rdma_resolve_route(id, SDP_ROUTE_TIMEOUT);
break;
case RDMA_CM_EVENT_ADDR_ERROR:
rc = -ENETUNREACH;
break;
case RDMA_CM_EVENT_ROUTE_RESOLVED:
rc = sdp_init_qp(sk, id);
if (rc)
break;
memset(&hh, 0, sizeof hh);
hh.bsdh.mid = SDP_MID_HELLO;
hh.bsdh.len = htonl(sizeof(struct sdp_hh));
hh.max_adverts = 1;
hh.majv_minv = SDP_MAJV_MINV;
sdp_init_buffers(sdp_sk(sk), rcvbuf_initial_size);
hh.bsdh.bufs = htons(rx_ring_posted(sdp_sk(sk)));
atomic_set(&sdp_sk(sk)->remote_credits,
rx_ring_posted(sdp_sk(sk)));
hh.localrcvsz = hh.desremrcvsz = htonl(sdp_sk(sk)->recv_frags *
PAGE_SIZE + sizeof(struct sdp_bsdh));
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
if (inet6_sk(sk)) {
struct sockaddr *src_addr = (struct sockaddr *)&id->route.addr.src_addr;
struct sockaddr_in *addr4 = (struct sockaddr_in *)src_addr;
struct sockaddr_in6 *addr6 = (struct sockaddr_in6 *)src_addr;
if (src_addr->sa_family == AF_INET) {
/* IPv4 over IPv6 */
ipv6_addr_set(&sk->sk_v6_rcv_saddr, 0, 0, htonl(0xFFFF),
addr4->sin_addr.s_addr);
} else {
sk->sk_v6_rcv_saddr = addr6->sin6_addr;
}
inet6_sk(sk)->saddr = sk->sk_v6_rcv_saddr;
}
else
#endif
{
sdp_inet_saddr(sk) = sdp_inet_rcv_saddr(sk) =
((struct sockaddr_in *)&id->route.addr.src_addr)->sin_addr.s_addr;
}
memset(&conn_param, 0, sizeof conn_param);
conn_param.private_data_len = sizeof hh;
conn_param.private_data = &hh;
conn_param.responder_resources = 4 /* TODO */;
conn_param.initiator_depth = 4 /* TODO */;
conn_param.retry_count = sdp_retry_count;
SDP_DUMP_PACKET(sk, "TX", NULL, &hh.bsdh);
if (sdp_apm_enable) {
rc = rdma_enable_apm(id, RDMA_ALT_PATH_BEST);
if (rc)
sdp_warn(sk, "APM couldn't be enabled for active side: %d\n", rc);
}
rc = rdma_connect(id, &conn_param);
break;
case RDMA_CM_EVENT_ALT_ROUTE_RESOLVED:
sdp_dbg(sk, "alt route was resolved slid=%d, dlid=%d\n",
id->route.path_rec[1].slid, id->route.path_rec[1].dlid);
break;
case RDMA_CM_EVENT_ALT_PATH_LOADED:
sdp_dbg(sk, "alt route path loaded\n");
break;
case RDMA_CM_EVENT_ALT_ROUTE_ERROR:
sdp_warn(sk, "alt route resolve error\n");
break;
case RDMA_CM_EVENT_ROUTE_ERROR:
rc = -ETIMEDOUT;
break;
case RDMA_CM_EVENT_CONNECT_REQUEST:
rc = sdp_connect_handler(sk, id, event);
if (rc) {
sdp_dbg(sk, "Destroying qp\n");
rdma_reject(id, NULL, 0);
break;
}
child = id->context;
atomic_set(&sdp_sk(child)->remote_credits,
rx_ring_posted(sdp_sk(child)));
memset(&hah, 0, sizeof hah);
hah.bsdh.mid = SDP_MID_HELLO_ACK;
hah.bsdh.bufs = htons(rx_ring_posted(sdp_sk(child)));
hah.bsdh.len = htonl(sizeof(struct sdp_hah));
hah.majv_minv = SDP_MAJV_MINV;
hah.ext_max_adverts = 1; /* Doesn't seem to be mandated by spec,
but just in case */
hah.actrcvsz = htonl(sdp_sk(child)->recv_frags * PAGE_SIZE +
sizeof(struct sdp_bsdh));
memset(&conn_param, 0, sizeof conn_param);
conn_param.private_data_len = sizeof hah;
conn_param.private_data = &hah;
conn_param.responder_resources = 4 /* TODO */;
conn_param.initiator_depth = 4 /* TODO */;
conn_param.retry_count = sdp_retry_count;
SDP_DUMP_PACKET(sk, "TX", NULL, &hah.bsdh);
rc = rdma_accept(id, &conn_param);
if (rc) {
sdp_sk(child)->id = NULL;
id->qp = NULL;
id->context = NULL;
parent = sdp_sk(child)->parent; /* TODO: hold ? */
} else if (sdp_apm_enable) {
rc2 = rdma_enable_apm(id, RDMA_ALT_PATH_BEST);
if (rc2)
sdp_warn(sk, "APM couldn't be enabled for passive side: %d\n", rc2);
}
break;
case RDMA_CM_EVENT_CONNECT_RESPONSE:
rc = sdp_response_handler(sk, id, event);
if (rc) {
sdp_dbg(sk, "Destroying qp\n");
rdma_reject(id, NULL, 0);
} else {
rc = rdma_accept(id, NULL);
if (!rc && sdp_apm_enable) {
rc2 = rdma_enable_apm(id, RDMA_ALT_PATH_BEST);
if (rc2)
sdp_warn(sk, "APM couldn't be enabled for passive side:%d \n", rc2);
}
}
break;
case RDMA_CM_EVENT_CONNECT_ERROR:
rc = -ETIMEDOUT;
break;
case RDMA_CM_EVENT_UNREACHABLE:
rc = -ENETUNREACH;
break;
case RDMA_CM_EVENT_REJECTED:
rc = -ECONNREFUSED;
break;
case RDMA_CM_EVENT_ESTABLISHED:
sdp_inet_saddr(sk) = sdp_inet_rcv_saddr(sk) =
((struct sockaddr_in *)&id->route.addr.src_addr)->sin_addr.s_addr;
rc = sdp_connected_handler(sk);
break;
case RDMA_CM_EVENT_DISCONNECTED: /* This means DREQ/DREP received */
if (sk->sk_state == TCP_LAST_ACK) {
sdp_cancel_dreq_wait_timeout(sdp_sk(sk));
sdp_exch_state(sk, TCPF_LAST_ACK, TCP_TIME_WAIT);
sdp_dbg(sk, "%s: waiting for Infiniband tear down\n",
__func__);
}
sdp_sk(sk)->qp_active = 0;
rdma_disconnect(id);
if (sk->sk_state != TCP_TIME_WAIT) {
if (sk->sk_state == TCP_CLOSE_WAIT) {
sdp_dbg(sk, "IB teardown while in "
"TCP_CLOSE_WAIT taking reference to "
"let close() finish the work\n");
sock_hold(sk, SOCK_REF_CMA);
sdp_start_cma_timewait_timeout(sdp_sk(sk),
SDP_CMA_TIMEWAIT_TIMEOUT);
}
sdp_set_error(sk, -EPIPE);
rc = sdp_disconnected_handler(sk);
}
break;
case RDMA_CM_EVENT_TIMEWAIT_EXIT:
rc = sdp_disconnected_handler(sk);
break;
case RDMA_CM_EVENT_DEVICE_REMOVAL:
rc = -ENETRESET;
break;
case RDMA_CM_EVENT_ADDR_CHANGE:
sdp_dbg(sk, "Got Address change event\n");
rc = 0;
break;
default:
printk(KERN_ERR "SDP: Unexpected CMA event: %d\n",
event->event);
rc = -ECONNABORTED;
break;
}
sdp_dbg(sk, "event: %s handled\n", rdma_cm_event_str(event->event));
if (rc && sdp_sk(sk)->id == id) {
child = sk;
sdp_sk(sk)->id = NULL;
id->qp = NULL;
id->context = NULL;
parent = sdp_sk(sk)->parent;
sdp_reset_sk(sk, rc);
}
release_sock(sk);
sdp_dbg(sk, "event: %s done. status %d\n",
rdma_cm_event_str(event->event), rc);
if (parent) {
lock_sock(parent);
if (!sdp_sk(parent)->id) { /* TODO: look at SOCK_DEAD? */
sdp_dbg(sk, "parent is going away.\n");
child = NULL;
goto done;
}
if (!list_empty(&sdp_sk(child)->backlog_queue))
list_del_init(&sdp_sk(child)->backlog_queue);
else
child = NULL;
done:
release_sock(parent);
if (child)
sdp_common_release(child);
}
return rc;
}
