diff options
author | Faisal Latif <faisal.latif@intel.com> | 2009-03-06 15:15:01 -0800 |
---|---|---|
committer | Roland Dreier <rolandd@cisco.com> | 2009-03-06 15:15:01 -0800 |
commit | 9d5ab13325d0bb855cf856946c140a68ceb20e32 (patch) | |
tree | c226cba67009db77052aceb1c7bf885da63fa4b8 /drivers/infiniband | |
parent | 0145f341a951b998d6d0fa38992a42d2a90b5bab (diff) | |
download | blackbird-op-linux-9d5ab13325d0bb855cf856946c140a68ceb20e32.tar.gz blackbird-op-linux-9d5ab13325d0bb855cf856946c140a68ceb20e32.zip |
RDMA/nes: Handle MPA Reject message properly
While doing testing, there are failures as MPA Reject call is not
handled. To handle MPA Reject call, following changes are done:
*Handle inbound/outbound MPA Reject response message.
When nes_reject() is called for pending MPA request reply,
send the MPA Reject message to its peer (active
side)cm_node. The peer cm_node (active side) will indicate
Reject message event for the pending Connect Request.
*Handle MPA Reject response message for loopback connections and listener.
When MPA Request is rejected, check if it is a loopback
connection and if it is then it will send Reject message event
to its peer loopback node. Also when destroying listener,
check if the cm_nodes for that listener are loopback or not.
*Add gracefull connection close with the MPA Reject response message.
Send gracefull close (FIN, FIN ACK..) to terminate the cm_nodes.
*Some code re-org while making the above changes.
Removed recv_list and recv_list_lock from the cm_node
structure as there can be only one receive close entry on the
timer. Also implemented handle_recv_entry() as receive close
entry is processed from both nes_rem_ref_cm_node() as well as
nes_cm_timer_tick().
Signed-off-by: Faisal Latif <faisal.latif@intel.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Diffstat (limited to 'drivers/infiniband')
-rw-r--r-- | drivers/infiniband/hw/nes/nes_cm.c | 580 | ||||
-rw-r--r-- | drivers/infiniband/hw/nes/nes_cm.h | 10 |
2 files changed, 399 insertions, 191 deletions
diff --git a/drivers/infiniband/hw/nes/nes_cm.c b/drivers/infiniband/hw/nes/nes_cm.c index bd918df3d22c..5327f2bec6bf 100644 --- a/drivers/infiniband/hw/nes/nes_cm.c +++ b/drivers/infiniband/hw/nes/nes_cm.c @@ -103,6 +103,7 @@ static int nes_disconnect(struct nes_qp *nesqp, int abrupt); static void nes_disconnect_worker(struct work_struct *work); static int send_mpa_request(struct nes_cm_node *, struct sk_buff *); +static int send_mpa_reject(struct nes_cm_node *); static int send_syn(struct nes_cm_node *, u32, struct sk_buff *); static int send_reset(struct nes_cm_node *, struct sk_buff *); static int send_ack(struct nes_cm_node *cm_node, struct sk_buff *skb); @@ -113,8 +114,7 @@ static void process_packet(struct nes_cm_node *, struct sk_buff *, static void active_open_err(struct nes_cm_node *, struct sk_buff *, int); static void passive_open_err(struct nes_cm_node *, struct sk_buff *, int); static void cleanup_retrans_entry(struct nes_cm_node *); -static void handle_rcv_mpa(struct nes_cm_node *, struct sk_buff *, - enum nes_cm_event_type); +static void handle_rcv_mpa(struct nes_cm_node *, struct sk_buff *); static void free_retrans_entry(struct nes_cm_node *cm_node); static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph, struct sk_buff *skb, int optionsize, int passive); @@ -124,6 +124,8 @@ static void cm_event_connected(struct nes_cm_event *); static void cm_event_connect_error(struct nes_cm_event *); static void cm_event_reset(struct nes_cm_event *); static void cm_event_mpa_req(struct nes_cm_event *); +static void cm_event_mpa_reject(struct nes_cm_event *); +static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node); static void print_core(struct nes_cm_core *core); @@ -196,7 +198,6 @@ static struct nes_cm_event *create_event(struct nes_cm_node *cm_node, */ static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb) { - int ret; if (!skb) { nes_debug(NES_DBG_CM, "skb set to NULL\n"); return -1; @@ -206,11 +207,27 @@ static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb) form_cm_frame(skb, cm_node, NULL, 0, &cm_node->mpa_frame, cm_node->mpa_frame_size, SET_ACK); - ret = schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0); - if (ret < 0) - return ret; + return schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0); +} - return 0; + + +static int send_mpa_reject(struct nes_cm_node *cm_node) +{ + struct sk_buff *skb = NULL; + + skb = dev_alloc_skb(MAX_CM_BUFFER); + if (!skb) { + nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n"); + return -ENOMEM; + } + + /* send an MPA reject frame */ + form_cm_frame(skb, cm_node, NULL, 0, &cm_node->mpa_frame, + cm_node->mpa_frame_size, SET_ACK | SET_FIN); + + cm_node->state = NES_CM_STATE_FIN_WAIT1; + return schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0); } @@ -218,14 +235,17 @@ static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb) * recv_mpa - process a received TCP pkt, we are expecting an * IETF MPA frame */ -static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 len) +static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 *type, + u32 len) { struct ietf_mpa_frame *mpa_frame; + *type = NES_MPA_REQUEST_ACCEPT; + /* assume req frame is in tcp data payload */ if (len < sizeof(struct ietf_mpa_frame)) { nes_debug(NES_DBG_CM, "The received ietf buffer was too small (%x)\n", len); - return -1; + return -EINVAL; } mpa_frame = (struct ietf_mpa_frame *)buffer; @@ -234,14 +254,25 @@ static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 len) if (cm_node->mpa_frame_size + sizeof(struct ietf_mpa_frame) != len) { nes_debug(NES_DBG_CM, "The received ietf buffer was not right" " complete (%x + %x != %x)\n", - cm_node->mpa_frame_size, (u32)sizeof(struct ietf_mpa_frame), len); - return -1; + cm_node->mpa_frame_size, + (u32)sizeof(struct ietf_mpa_frame), len); + return -EINVAL; + } + /* make sure it does not exceed the max size */ + if (len > MAX_CM_BUFFER) { + nes_debug(NES_DBG_CM, "The received ietf buffer was too large" + " (%x + %x != %x)\n", + cm_node->mpa_frame_size, + (u32)sizeof(struct ietf_mpa_frame), len); + return -EINVAL; } /* copy entire MPA frame to our cm_node's frame */ memcpy(cm_node->mpa_frame_buf, buffer + sizeof(struct ietf_mpa_frame), cm_node->mpa_frame_size); + if (mpa_frame->flags & IETF_MPA_FLAGS_REJECT) + *type = NES_MPA_REQUEST_REJECT; return 0; } @@ -380,7 +411,7 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb, new_send = kzalloc(sizeof(*new_send), GFP_ATOMIC); if (!new_send) - return -1; + return -ENOMEM; /* new_send->timetosend = currenttime */ new_send->retrycount = NES_DEFAULT_RETRYS; @@ -394,9 +425,11 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb, if (type == NES_TIMER_TYPE_CLOSE) { new_send->timetosend += (HZ/10); - spin_lock_irqsave(&cm_node->recv_list_lock, flags); - list_add_tail(&new_send->list, &cm_node->recv_list); - spin_unlock_irqrestore(&cm_node->recv_list_lock, flags); + if (cm_node->recv_entry) { + WARN_ON(1); + return -EINVAL; + } + cm_node->recv_entry = new_send; } if (type == NES_TIMER_TYPE_SEND) { @@ -435,24 +468,78 @@ int schedule_nes_timer(struct nes_cm_node *cm_node, struct sk_buff *skb, return ret; } +static void nes_retrans_expired(struct nes_cm_node *cm_node) +{ + switch (cm_node->state) { + case NES_CM_STATE_SYN_RCVD: + case NES_CM_STATE_CLOSING: + rem_ref_cm_node(cm_node->cm_core, cm_node); + break; + case NES_CM_STATE_LAST_ACK: + case NES_CM_STATE_FIN_WAIT1: + case NES_CM_STATE_MPAREJ_RCVD: + send_reset(cm_node, NULL); + break; + default: + create_event(cm_node, NES_CM_EVENT_ABORTED); + } +} + +static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node) +{ + struct nes_timer_entry *recv_entry = cm_node->recv_entry; + struct iw_cm_id *cm_id = cm_node->cm_id; + struct nes_qp *nesqp; + unsigned long qplockflags; + + if (!recv_entry) + return; + nesqp = (struct nes_qp *)recv_entry->skb; + if (nesqp) { + spin_lock_irqsave(&nesqp->lock, qplockflags); + if (nesqp->cm_id) { + nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, " + "refcount = %d: HIT A " + "NES_TIMER_TYPE_CLOSE with something " + "to do!!!\n", nesqp->hwqp.qp_id, cm_id, + atomic_read(&nesqp->refcount)); + nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED; + nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT; + nesqp->ibqp_state = IB_QPS_ERR; + spin_unlock_irqrestore(&nesqp->lock, qplockflags); + nes_cm_disconn(nesqp); + } else { + spin_unlock_irqrestore(&nesqp->lock, qplockflags); + nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, " + "refcount = %d: HIT A " + "NES_TIMER_TYPE_CLOSE with nothing " + "to do!!!\n", nesqp->hwqp.qp_id, cm_id, + atomic_read(&nesqp->refcount)); + } + } else if (rem_node) { + /* TIME_WAIT state */ + rem_ref_cm_node(cm_node->cm_core, cm_node); + } + if (cm_node->cm_id) + cm_id->rem_ref(cm_id); + kfree(recv_entry); + cm_node->recv_entry = NULL; +} /** * nes_cm_timer_tick */ static void nes_cm_timer_tick(unsigned long pass) { - unsigned long flags, qplockflags; + unsigned long flags; unsigned long nexttimeout = jiffies + NES_LONG_TIME; - struct iw_cm_id *cm_id; struct nes_cm_node *cm_node; struct nes_timer_entry *send_entry, *recv_entry; - struct list_head *list_core, *list_core_temp; - struct list_head *list_node, *list_node_temp; + struct list_head *list_core_temp; + struct list_head *list_node; struct nes_cm_core *cm_core = g_cm_core; - struct nes_qp *nesqp; u32 settimer = 0; int ret = NETDEV_TX_OK; - enum nes_cm_node_state last_state; struct list_head timer_list; INIT_LIST_HEAD(&timer_list); @@ -461,7 +548,7 @@ static void nes_cm_timer_tick(unsigned long pass) list_for_each_safe(list_node, list_core_temp, &cm_core->connected_nodes) { cm_node = container_of(list_node, struct nes_cm_node, list); - if (!list_empty(&cm_node->recv_list) || (cm_node->send_entry)) { + if ((cm_node->recv_entry) || (cm_node->send_entry)) { add_ref_cm_node(cm_node); list_add(&cm_node->timer_entry, &timer_list); } @@ -471,54 +558,18 @@ static void nes_cm_timer_tick(unsigned long pass) list_for_each_safe(list_node, list_core_temp, &timer_list) { cm_node = container_of(list_node, struct nes_cm_node, timer_entry); - spin_lock_irqsave(&cm_node->recv_list_lock, flags); - list_for_each_safe(list_core, list_node_temp, - &cm_node->recv_list) { - recv_entry = container_of(list_core, - struct nes_timer_entry, list); - if (!recv_entry) - break; + recv_entry = cm_node->recv_entry; + + if (recv_entry) { if (time_after(recv_entry->timetosend, jiffies)) { if (nexttimeout > recv_entry->timetosend || - !settimer) { + !settimer) { nexttimeout = recv_entry->timetosend; settimer = 1; } - continue; - } - list_del(&recv_entry->list); - cm_id = cm_node->cm_id; - spin_unlock_irqrestore(&cm_node->recv_list_lock, flags); - nesqp = (struct nes_qp *)recv_entry->skb; - spin_lock_irqsave(&nesqp->lock, qplockflags); - if (nesqp->cm_id) { - nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, " - "refcount = %d: HIT A " - "NES_TIMER_TYPE_CLOSE with something " - "to do!!!\n", nesqp->hwqp.qp_id, cm_id, - atomic_read(&nesqp->refcount)); - nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED; - nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT; - nesqp->ibqp_state = IB_QPS_ERR; - spin_unlock_irqrestore(&nesqp->lock, - qplockflags); - nes_cm_disconn(nesqp); - } else { - spin_unlock_irqrestore(&nesqp->lock, - qplockflags); - nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, " - "refcount = %d: HIT A " - "NES_TIMER_TYPE_CLOSE with nothing " - "to do!!!\n", nesqp->hwqp.qp_id, cm_id, - atomic_read(&nesqp->refcount)); - } - if (cm_id) - cm_id->rem_ref(cm_id); - - kfree(recv_entry); - spin_lock_irqsave(&cm_node->recv_list_lock, flags); + } else + handle_recv_entry(cm_node, 1); } - spin_unlock_irqrestore(&cm_node->recv_list_lock, flags); spin_lock_irqsave(&cm_node->retrans_list_lock, flags); do { @@ -533,12 +584,11 @@ static void nes_cm_timer_tick(unsigned long pass) nexttimeout = send_entry->timetosend; settimer = 1; - break; } } else { free_retrans_entry(cm_node); - break; } + break; } if ((cm_node->state == NES_CM_STATE_TSA) || @@ -550,16 +600,12 @@ static void nes_cm_timer_tick(unsigned long pass) if (!send_entry->retranscount || !send_entry->retrycount) { cm_packets_dropped++; - last_state = cm_node->state; - cm_node->state = NES_CM_STATE_CLOSED; free_retrans_entry(cm_node); + spin_unlock_irqrestore( &cm_node->retrans_list_lock, flags); - if (last_state == NES_CM_STATE_SYN_RCVD) - rem_ref_cm_node(cm_core, cm_node); - else - create_event(cm_node, - NES_CM_EVENT_ABORTED); + nes_retrans_expired(cm_node); + cm_node->state = NES_CM_STATE_CLOSED; spin_lock_irqsave(&cm_node->retrans_list_lock, flags); break; @@ -714,7 +760,7 @@ static int send_reset(struct nes_cm_node *cm_node, struct sk_buff *skb) skb = dev_alloc_skb(MAX_CM_BUFFER); if (!skb) { nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n"); - return -1; + return -ENOMEM; } form_cm_frame(skb, cm_node, NULL, 0, NULL, 0, flags); @@ -871,7 +917,8 @@ static int add_hte_node(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, struct nes_cm_listener *listener, int free_hanging_nodes) { - int ret = 1; + int ret = -EINVAL; + int err = 0; unsigned long flags; struct list_head *list_pos = NULL; struct list_head *list_temp = NULL; @@ -900,10 +947,60 @@ static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core, list_for_each_safe(list_pos, list_temp, &reset_list) { cm_node = container_of(list_pos, struct nes_cm_node, - reset_entry); - cleanup_retrans_entry(cm_node); - send_reset(cm_node, NULL); - rem_ref_cm_node(cm_node->cm_core, cm_node); + reset_entry); + { + struct nes_cm_node *loopback = cm_node->loopbackpartner; + if (NES_CM_STATE_FIN_WAIT1 <= cm_node->state) { + rem_ref_cm_node(cm_node->cm_core, cm_node); + } else { + if (!loopback) { + cleanup_retrans_entry(cm_node); + err = send_reset(cm_node, NULL); + if (err) { + cm_node->state = + NES_CM_STATE_CLOSED; + WARN_ON(1); + } else { + cm_node->state = + NES_CM_STATE_CLOSED; + rem_ref_cm_node( + cm_node->cm_core, + cm_node); + } + } else { + struct nes_cm_event event; + + event.cm_node = loopback; + event.cm_info.rem_addr = + loopback->rem_addr; + event.cm_info.loc_addr = + loopback->loc_addr; + event.cm_info.rem_port = + loopback->rem_port; + event.cm_info.loc_port = + loopback->loc_port; + event.cm_info.cm_id = loopback->cm_id; + cm_event_connect_error(&event); + loopback->state = NES_CM_STATE_CLOSED; + + event.cm_node = cm_node; + event.cm_info.rem_addr = + cm_node->rem_addr; + event.cm_info.loc_addr = + cm_node->loc_addr; + event.cm_info.rem_port = + cm_node->rem_port; + event.cm_info.loc_port = + cm_node->loc_port; + event.cm_info.cm_id = cm_node->cm_id; + cm_event_reset(&event); + + rem_ref_cm_node(cm_node->cm_core, + cm_node); + + } + } + } } spin_lock_irqsave(&cm_core->listen_list_lock, flags); @@ -964,6 +1061,7 @@ static inline int mini_cm_accelerated(struct nes_cm_core *cm_core, if (cm_node->accept_pend) { BUG_ON(!cm_node->listener); atomic_dec(&cm_node->listener->pend_accepts_cnt); + cm_node->accept_pend = 0; BUG_ON(atomic_read(&cm_node->listener->pend_accepts_cnt) < 0); } @@ -990,7 +1088,7 @@ static int nes_addr_resolve_neigh(struct nes_vnic *nesvnic, u32 dst_ip) memset(&fl, 0, sizeof fl); fl.nl_u.ip4_u.daddr = htonl(dst_ip); if (ip_route_output_key(&init_net, &rt, &fl)) { - printk("%s: ip_route_output_key failed for 0x%08X\n", + printk(KERN_ERR "%s: ip_route_output_key failed for 0x%08X\n", __func__, dst_ip); return rc; } @@ -1053,8 +1151,6 @@ static struct nes_cm_node *make_cm_node(struct nes_cm_core *cm_core, cm_node->cm_id); spin_lock_init(&cm_node->retrans_list_lock); - INIT_LIST_HEAD(&cm_node->recv_list); - spin_lock_init(&cm_node->recv_list_lock); cm_node->loopbackpartner = NULL; atomic_set(&cm_node->ref_count, 1); @@ -1122,10 +1218,7 @@ static int add_ref_cm_node(struct nes_cm_node *cm_node) static int rem_ref_cm_node(struct nes_cm_core *cm_core, struct nes_cm_node *cm_node) { - unsigned long flags, qplockflags; - struct nes_timer_entry *recv_entry; - struct iw_cm_id *cm_id; - struct list_head *list_core, *list_node_temp; + unsigned long flags; struct nes_qp *nesqp; if (!cm_node) @@ -1146,38 +1239,9 @@ static int rem_ref_cm_node(struct nes_cm_core *cm_core, atomic_dec(&cm_node->listener->pend_accepts_cnt); BUG_ON(atomic_read(&cm_node->listener->pend_accepts_cnt) < 0); } - BUG_ON(cm_node->send_entry); - spin_lock_irqsave(&cm_node->recv_list_lock, flags); - list_for_each_safe(list_core, list_node_temp, &cm_node->recv_list) { - recv_entry = container_of(list_core, struct nes_timer_entry, - list); - list_del(&recv_entry->list); - cm_id = cm_node->cm_id; - spin_unlock_irqrestore(&cm_node->recv_list_lock, flags); - nesqp = (struct nes_qp *)recv_entry->skb; - spin_lock_irqsave(&nesqp->lock, qplockflags); - if (nesqp->cm_id) { - nes_debug(NES_DBG_CM, "QP%u: cm_id = %p: HIT A " - "NES_TIMER_TYPE_CLOSE with something to do!\n", - nesqp->hwqp.qp_id, cm_id); - nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED; - nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT; - nesqp->ibqp_state = IB_QPS_ERR; - spin_unlock_irqrestore(&nesqp->lock, qplockflags); - nes_cm_disconn(nesqp); - } else { - spin_unlock_irqrestore(&nesqp->lock, qplockflags); - nes_debug(NES_DBG_CM, "QP%u: cm_id = %p: HIT A " - "NES_TIMER_TYPE_CLOSE with nothing to do!\n", - nesqp->hwqp.qp_id, cm_id); - } - cm_id->rem_ref(cm_id); - - kfree(recv_entry); - spin_lock_irqsave(&cm_node->recv_list_lock, flags); - } - spin_unlock_irqrestore(&cm_node->recv_list_lock, flags); - + WARN_ON(cm_node->send_entry); + if (cm_node->recv_entry) + handle_recv_entry(cm_node, 0); if (cm_node->listener) { mini_cm_dec_refcnt_listen(cm_core, cm_node->listener, 0); } else { @@ -1262,8 +1326,7 @@ static void drop_packet(struct sk_buff *skb) dev_kfree_skb_any(skb); } -static void handle_fin_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, - struct tcphdr *tcph) +static void handle_fin_pkt(struct nes_cm_node *cm_node) { nes_debug(NES_DBG_CM, "Received FIN, cm_node = %p, state = %u. " "refcnt=%d\n", cm_node, cm_node->state, @@ -1275,23 +1338,30 @@ static void handle_fin_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, case NES_CM_STATE_SYN_SENT: case NES_CM_STATE_ESTABLISHED: case NES_CM_STATE_MPAREQ_SENT: + case NES_CM_STATE_MPAREJ_RCVD: cm_node->state = NES_CM_STATE_LAST_ACK; - send_fin(cm_node, skb); + send_fin(cm_node, NULL); break; case NES_CM_STATE_FIN_WAIT1: cm_node->state = NES_CM_STATE_CLOSING; - send_ack(cm_node, skb); + send_ack(cm_node, NULL); + /* Wait for ACK as this is simultanous close.. + * After we receive ACK, do not send anything.. + * Just rm the node.. Done.. */ break; case NES_CM_STATE_FIN_WAIT2: cm_node->state = NES_CM_STATE_TIME_WAIT; - send_ack(cm_node, skb); + send_ack(cm_node, NULL); + schedule_nes_timer(cm_node, NULL, NES_TIMER_TYPE_CLOSE, 1, 0); + break; + case NES_CM_STATE_TIME_WAIT: cm_node->state = NES_CM_STATE_CLOSED; + rem_ref_cm_node(cm_node->cm_core, cm_node); break; case NES_CM_STATE_TSA: default: nes_debug(NES_DBG_CM, "Error Rcvd FIN for node-%p state = %d\n", cm_node, cm_node->state); - drop_packet(skb); break; } } @@ -1337,23 +1407,35 @@ static void handle_rst_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, cleanup_retrans_entry(cm_node); drop_packet(skb); break; + case NES_CM_STATE_TIME_WAIT: + cleanup_retrans_entry(cm_node); + cm_node->state = NES_CM_STATE_CLOSED; + rem_ref_cm_node(cm_node->cm_core, cm_node); + drop_packet(skb); + break; + case NES_CM_STATE_FIN_WAIT1: + cleanup_retrans_entry(cm_node); + nes_debug(NES_DBG_CM, "Bad state %s[%u]\n", __func__, __LINE__); default: drop_packet(skb); break; } } -static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb, - enum nes_cm_event_type type) + +static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb) { - int ret; + int ret = 0; int datasize = skb->len; u8 *dataloc = skb->data; - ret = parse_mpa(cm_node, dataloc, datasize); - if (ret < 0) { + + enum nes_cm_event_type type = NES_CM_EVENT_UNKNOWN; + u32 res_type; + ret = parse_mpa(cm_node, dataloc, &res_type, datasize); + if (ret) { nes_debug(NES_DBG_CM, "didn't like MPA Request\n"); - if (type == NES_CM_EVENT_CONNECTED) { + if (cm_node->state == NES_CM_STATE_MPAREQ_SENT) { nes_debug(NES_DBG_CM, "%s[%u] create abort for " "cm_node=%p listener=%p state=%d\n", __func__, __LINE__, cm_node, cm_node->listener, @@ -1362,18 +1444,38 @@ static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb, } else { passive_open_err(cm_node, skb, 1); } - } else { - cleanup_retrans_entry(cm_node); - dev_kfree_skb_any(skb); - if (type == NES_CM_EVENT_CONNECTED) + return; + } + + switch (cm_node->state) { + case NES_CM_STATE_ESTABLISHED: + if (res_type == NES_MPA_REQUEST_REJECT) { + /*BIG problem as we are receiving the MPA.. So should + * not be REJECT.. This is Passive Open.. We can + * only receive it Reject for Active Open...*/ + WARN_ON(1); + } + cm_node->state = NES_CM_STATE_MPAREQ_RCVD; + type = NES_CM_EVENT_MPA_REQ; + atomic_set(&cm_node->passive_state, + NES_PASSIVE_STATE_INDICATED); + break; + case NES_CM_STATE_MPAREQ_SENT: + if (res_type == NES_MPA_REQUEST_REJECT) { + type = NES_CM_EVENT_MPA_REJECT; + cm_node->state = NES_CM_STATE_MPAREJ_RCVD; + } else { + type = NES_CM_EVENT_CONNECTED; cm_node->state = NES_CM_STATE_TSA; - else - atomic_set(&cm_node->passive_state, - NES_PASSIVE_STATE_INDICATED); - create_event(cm_node, type); + } + break; + default: + WARN_ON(1); + break; } - return ; + dev_kfree_skb_any(skb); + create_event(cm_node, type); } static void indicate_pkt_err(struct nes_cm_node *cm_node, struct sk_buff *skb) @@ -1461,8 +1563,6 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, break; case NES_CM_STATE_LISTENING: /* Passive OPEN */ - cm_node->accept_pend = 1; - atomic_inc(&cm_node->listener->pend_accepts_cnt); if (atomic_read(&cm_node->listener->pend_accepts_cnt) > cm_node->listener->backlog) { nes_debug(NES_DBG_CM, "drop syn due to backlog " @@ -1480,6 +1580,9 @@ static void handle_syn_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, } cm_node->tcp_cntxt.rcv_nxt = inc_sequence + 1; BUG_ON(cm_node->send_entry); + cm_node->accept_pend = 1; + atomic_inc(&cm_node->listener->pend_accepts_cnt); + cm_node->state = NES_CM_STATE_SYN_RCVD; send_syn(cm_node, 1, skb); break; @@ -1514,6 +1617,7 @@ static void handle_synack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, inc_sequence = ntohl(tcph->seq); switch (cm_node->state) { case NES_CM_STATE_SYN_SENT: + cleanup_retrans_entry(cm_node); /* active open */ if (check_syn(cm_node, tcph, skb)) return; @@ -1563,10 +1667,7 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, u32 rem_seq; int ret; int optionsize; - u32 temp_seq = cm_node->tcp_cntxt.loc_seq_num; - optionsize = (tcph->doff << 2) - sizeof(struct tcphdr); - cm_node->tcp_cntxt.loc_seq_num = ntohl(tcph->ack_seq); if (check_seq(cm_node, tcph, skb)) return; @@ -1576,7 +1677,7 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, rem_seq = ntohl(tcph->seq); rem_seq_ack = ntohl(tcph->ack_seq); datasize = skb->len; - + cleanup_retrans_entry(cm_node); switch (cm_node->state) { case NES_CM_STATE_SYN_RCVD: /* Passive OPEN */ @@ -1584,7 +1685,6 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, if (ret) break; cm_node->tcp_cntxt.rem_ack_num = ntohl(tcph->ack_seq); - cm_node->tcp_cntxt.loc_seq_num = temp_seq; if (cm_node->tcp_cntxt.rem_ack_num != cm_node->tcp_cntxt.loc_seq_num) { nes_debug(NES_DBG_CM, "rem_ack_num != loc_seq_num\n"); @@ -1593,31 +1693,30 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, return; } cm_node->state = NES_CM_STATE_ESTABLISHED; + cleanup_retrans_entry(cm_node); if (datasize) { cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize; - cm_node->state = NES_CM_STATE_MPAREQ_RCVD; - handle_rcv_mpa(cm_node, skb, NES_CM_EVENT_MPA_REQ); - } else { /* rcvd ACK only */ + handle_rcv_mpa(cm_node, skb); + } else { /* rcvd ACK only */ dev_kfree_skb_any(skb); cleanup_retrans_entry(cm_node); } break; case NES_CM_STATE_ESTABLISHED: /* Passive OPEN */ - /* We expect mpa frame to be received only */ + cleanup_retrans_entry(cm_node); if (datasize) { cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize; - cm_node->state = NES_CM_STATE_MPAREQ_RCVD; - handle_rcv_mpa(cm_node, skb, - NES_CM_EVENT_MPA_REQ); + handle_rcv_mpa(cm_node, skb); } else drop_packet(skb); break; case NES_CM_STATE_MPAREQ_SENT: + cleanup_retrans_entry(cm_node); cm_node->tcp_cntxt.rem_ack_num = ntohl(tcph->ack_seq); if (datasize) { cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize; - handle_rcv_mpa(cm_node, skb, NES_CM_EVENT_CONNECTED); + handle_rcv_mpa(cm_node, skb); } else { /* Could be just an ack pkt.. */ cleanup_retrans_entry(cm_node); dev_kfree_skb_any(skb); @@ -1628,13 +1727,24 @@ static void handle_ack_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb, cleanup_retrans_entry(cm_node); send_reset(cm_node, skb); break; + case NES_CM_STATE_LAST_ACK: + cleanup_retrans_entry(cm_node); + cm_node->state = NES_CM_STATE_CLOSED; + cm_node->cm_id->rem_ref(cm_node->cm_id); + case NES_CM_STATE_CLOSING: + cleanup_retrans_entry(cm_node); + rem_ref_cm_node(cm_node->cm_core, cm_node); + drop_packet(skb); + break; case NES_CM_STATE_FIN_WAIT1: + cleanup_retrans_entry(cm_node); + drop_packet(skb); + cm_node->state = NES_CM_STATE_FIN_WAIT2; + break; case NES_CM_STATE_SYN_SENT: case NES_CM_STATE_FIN_WAIT2: case NES_CM_STATE_TSA: case NES_CM_STATE_MPAREQ_RCVD: - case NES_CM_STATE_LAST_ACK: - case NES_CM_STATE_CLOSING: case NES_CM_STATE_UNKNOWN: default: drop_packet(skb); @@ -1744,6 +1854,7 @@ static void process_packet(struct nes_cm_node *cm_node, struct sk_buff *skb, { enum nes_tcpip_pkt_type pkt_type = NES_PKT_TYPE_UNKNOWN; struct tcphdr *tcph = tcp_hdr(skb); + u32 fin_set = 0; skb_pull(skb, ip_hdr(skb)->ihl << 2); nes_debug(NES_DBG_CM, "process_packet: cm_node=%p state =%d syn=%d " @@ -1756,10 +1867,10 @@ static void process_packet(struct nes_cm_node *cm_node, struct sk_buff *skb, pkt_type = NES_PKT_TYPE_SYN; if (tcph->ack) pkt_type = NES_PKT_TYPE_SYNACK; - } else if (tcph->fin) - pkt_type = NES_PKT_TYPE_FIN; - else if (tcph->ack) + } else if (tcph->ack) pkt_type = NES_PKT_TYPE_ACK; + if (tcph->fin) + fin_set = 1; switch (pkt_type) { case NES_PKT_TYPE_SYN: @@ -1770,15 +1881,16 @@ static void process_packet(struct nes_cm_node *cm_node, struct sk_buff *skb, break; case NES_PKT_TYPE_ACK: handle_ack_pkt(cm_node, skb, tcph); + if (fin_set) + handle_fin_pkt(cm_node); break; case NES_PKT_TYPE_RST: handle_rst_pkt(cm_node, skb, tcph); break; - case NES_PKT_TYPE_FIN: - handle_fin_pkt(cm_node, skb, tcph); - break; default: drop_packet(skb); + if (fin_set) + handle_fin_pkt(cm_node); break; } } @@ -1921,7 +2033,7 @@ static struct nes_cm_node *mini_cm_connect(struct nes_cm_core *cm_core, loopbackremotenode->tcp_cntxt.rcv_wscale; loopbackremotenode->tcp_cntxt.snd_wscale = cm_node->tcp_cntxt.rcv_wscale; - + loopbackremotenode->state = NES_CM_STATE_MPAREQ_RCVD; create_event(loopbackremotenode, NES_CM_EVENT_MPA_REQ); } return cm_node; @@ -1976,7 +2088,11 @@ static int mini_cm_reject(struct nes_cm_core *cm_core, struct ietf_mpa_frame *mpa_frame, struct nes_cm_node *cm_node) { int ret = 0; + int err = 0; int passive_state; + struct nes_cm_event event; + struct iw_cm_id *cm_id = cm_node->cm_id; + struct nes_cm_node *loopback = cm_node->loopbackpartner; nes_debug(NES_DBG_CM, "%s cm_node=%p type=%d state=%d\n", __func__, cm_node, cm_node->tcp_cntxt.client, cm_node->state); @@ -1985,12 +2101,38 @@ static int mini_cm_reject(struct nes_cm_core *cm_core, return ret; cleanup_retrans_entry(cm_node); - passive_state = atomic_add_return(1, &cm_node->passive_state); - cm_node->state = NES_CM_STATE_CLOSED; - if (passive_state == NES_SEND_RESET_EVENT) + if (!loopback) { + passive_state = atomic_add_return(1, &cm_node->passive_state); + if (passive_state == NES_SEND_RESET_EVENT) { + cm_node->state = NES_CM_STATE_CLOSED; + rem_ref_cm_node(cm_core, cm_node); + } else { + ret = send_mpa_reject(cm_node); + if (ret) { + cm_node->state = NES_CM_STATE_CLOSED; + err = send_reset(cm_node, NULL); + if (err) + WARN_ON(1); + } else + cm_id->add_ref(cm_id); + } + } else { + cm_node->cm_id = NULL; + event.cm_node = loopback; + event.cm_info.rem_addr = loopback->rem_addr; + event.cm_info.loc_addr = loopback->loc_addr; + event.cm_info.rem_port = loopback->rem_port; + event.cm_info.loc_port = loopback->loc_port; + event.cm_info.cm_id = loopback->cm_id; + cm_event_mpa_reject(&event); rem_ref_cm_node(cm_core, cm_node); - else - ret = send_reset(cm_node, NULL); + loopback->state = NES_CM_STATE_CLOSING; + + cm_id = loopback->cm_id; + rem_ref_cm_node(cm_core, loopback); + cm_id->rem_ref(cm_id); + } + return ret; } @@ -2027,6 +2169,7 @@ static int mini_cm_close(struct nes_cm_core *cm_core, struct nes_cm_node *cm_nod case NES_CM_STATE_CLOSING: ret = -1; break; + case NES_CM_STATE_MPAREJ_RCVD: case NES_CM_STATE_LISTENING: case NES_CM_STATE_UNKNOWN: case NES_CM_STATE_INITED: @@ -2223,15 +2366,15 @@ static int mini_cm_set(struct nes_cm_core *cm_core, u32 type, u32 value) int ret = 0; switch (type) { - case NES_CM_SET_PKT_SIZE: - cm_core->mtu = value; - break; - case NES_CM_SET_FREE_PKT_Q_SIZE: - cm_core->free_tx_pkt_max = value; - break; - default: - /* unknown set option */ - ret = -EINVAL; + case NES_CM_SET_PKT_SIZE: + cm_core->mtu = value; + break; + case NES_CM_SET_FREE_PKT_Q_SIZE: + cm_core->free_tx_pkt_max = value; + break; + default: + /* unknown set option */ + ret = -EINVAL; } return ret; @@ -2621,9 +2764,7 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) NES_QPCONTEXT_ORDIRD_WRPDU); } else { nesqp->nesqp_context->ird_ord_sizes |= - cpu_to_le32((NES_QPCONTEXT_ORDIRD_LSMM_PRESENT | - NES_QPCONTEXT_ORDIRD_WRPDU | - NES_QPCONTEXT_ORDIRD_ALSMM)); + cpu_to_le32(NES_QPCONTEXT_ORDIRD_WRPDU); } nesqp->skip_lsmm = 1; @@ -2745,23 +2886,35 @@ int nes_accept(struct iw_cm_id *cm_id, struct iw_cm_conn_param *conn_param) int nes_reject(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len) { struct nes_cm_node *cm_node; + struct nes_cm_node *loopback; + struct nes_cm_core *cm_core; atomic_inc(&cm_rejects); cm_node = (struct nes_cm_node *) cm_id->provider_data; + loopback = cm_node->loopbackpartner; cm_core = cm_node->cm_core; + cm_node->cm_id = cm_id; cm_node->mpa_frame_size = sizeof(struct ietf_mpa_frame) + pdata_len; + if (cm_node->mpa_frame_size > MAX_CM_BUFFER) + return -EINVAL; + strcpy(&cm_node->mpa_frame.key[0], IEFT_MPA_KEY_REP); - memcpy(&cm_node->mpa_frame.priv_data, pdata, pdata_len); + if (loopback) { + memcpy(&loopback->mpa_frame.priv_data, pdata, pdata_len); + loopback->mpa_frame.priv_data_len = pdata_len; + loopback->mpa_frame_size = sizeof(struct ietf_mpa_frame) + + pdata_len; + } else { + memcpy(&cm_node->mpa_frame.priv_data, pdata, pdata_len); + cm_node->mpa_frame.priv_data_len = cpu_to_be16(pdata_len); + } - cm_node->mpa_frame.priv_data_len = cpu_to_be16(pdata_len); cm_node->mpa_frame.rev = mpa_version; cm_node->mpa_frame.flags = IETF_MPA_FLAGS_CRC | IETF_MPA_FLAGS_REJECT; - cm_core->api->reject(cm_core, &cm_node->mpa_frame, cm_node); - - return 0; + return cm_core->api->reject(cm_core, &cm_node->mpa_frame, cm_node); } @@ -3270,13 +3423,56 @@ static void cm_event_mpa_req(struct nes_cm_event *event) cm_event.remote_addr.sin_family = AF_INET; cm_event.remote_addr.sin_port = htons(event->cm_info.rem_port); cm_event.remote_addr.sin_addr.s_addr = htonl(event->cm_info.rem_addr); + cm_event.private_data = cm_node->mpa_frame_buf; + cm_event.private_data_len = (u8) cm_node->mpa_frame_size; + + ret = cm_id->event_handler(cm_id, &cm_event); + if (ret) + printk(KERN_ERR "%s[%u] OFA CM event_handler returned, ret=%d\n", + __func__, __LINE__, ret); + return; +} + + +static void cm_event_mpa_reject(struct nes_cm_event *event) +{ + struct iw_cm_id *cm_id; + struct iw_cm_event cm_event; + struct nes_cm_node *cm_node; + int ret; + + cm_node = event->cm_node; + if (!cm_node) + return; + cm_id = cm_node->cm_id; + + atomic_inc(&cm_connect_reqs); + nes_debug(NES_DBG_CM, "cm_node = %p - cm_id = %p, jiffies = %lu\n", + cm_node, cm_id, jiffies); + + cm_event.event = IW_CM_EVENT_CONNECT_REPLY; + cm_event.status = -ECONNREFUSED; + cm_event.provider_data = cm_id->provider_data; + + cm_event.local_addr.sin_family = AF_INET; + cm_event.local_addr.sin_port = htons(event->cm_info.loc_port); + cm_event.local_addr.sin_addr.s_addr = htonl(event->cm_info.loc_addr); + + cm_event.remote_addr.sin_family = AF_INET; + cm_event.remote_addr.sin_port = htons(event->cm_info.rem_port); + cm_event.remote_addr.sin_addr.s_addr = htonl(event->cm_info.rem_addr); - cm_event.private_data = cm_node->mpa_frame_buf; - cm_event.private_data_len = (u8) cm_node->mpa_frame_size; + cm_event.private_data = cm_node->mpa_frame_buf; + cm_event.private_data_len = (u8) cm_node->mpa_frame_size; + + nes_debug(NES_DBG_CM, "call CM_EVENT_MPA_REJECTED, local_addr=%08x, " + "remove_addr=%08x\n", + cm_event.local_addr.sin_addr.s_addr, + cm_event.remote_addr.sin_addr.s_addr); ret = cm_id->event_handler(cm_id, &cm_event); if (ret) - printk("%s[%u] OFA CM event_handler returned, ret=%d\n", + printk(KERN_ERR "%s[%u] OFA CM event_handler returned, ret=%d\n", __func__, __LINE__, ret); return; @@ -3341,6 +3537,14 @@ static void nes_cm_event_handler(struct work_struct *work) cm_event_connected(event); nes_debug(NES_DBG_CM, "CM Event: CONNECTED\n"); break; + case NES_CM_EVENT_MPA_REJECT: + if ((!event->cm_node->cm_id) || + (event->cm_node->state == NES_CM_STATE_TSA)) + break; + cm_event_mpa_reject(event); + nes_debug(NES_DBG_CM, "CM Event: REJECT\n"); + break; + case NES_CM_EVENT_ABORTED: if ((!event->cm_node->cm_id) || (event->cm_node->state == NES_CM_STATE_TSA)) diff --git a/drivers/infiniband/hw/nes/nes_cm.h b/drivers/infiniband/hw/nes/nes_cm.h index 4ab2bebf12ac..d5f778202eb7 100644 --- a/drivers/infiniband/hw/nes/nes_cm.h +++ b/drivers/infiniband/hw/nes/nes_cm.h @@ -39,6 +39,9 @@ #define NES_MANAGE_APBVT_DEL 0 #define NES_MANAGE_APBVT_ADD 1 +#define NES_MPA_REQUEST_ACCEPT 1 +#define NES_MPA_REQUEST_REJECT 2 + /* IETF MPA -- defines, enums, structs */ #define IEFT_MPA_KEY_REQ "MPA ID Req Frame" #define IEFT_MPA_KEY_REP "MPA ID Rep Frame" @@ -186,6 +189,7 @@ enum nes_cm_node_state { NES_CM_STATE_ACCEPTING, NES_CM_STATE_MPAREQ_SENT, NES_CM_STATE_MPAREQ_RCVD, + NES_CM_STATE_MPAREJ_RCVD, NES_CM_STATE_TSA, NES_CM_STATE_FIN_WAIT1, NES_CM_STATE_FIN_WAIT2, @@ -278,13 +282,12 @@ struct nes_cm_node { struct nes_timer_entry *send_entry; spinlock_t retrans_list_lock; - struct list_head recv_list; - spinlock_t recv_list_lock; + struct nes_timer_entry *recv_entry; int send_write0; union { struct ietf_mpa_frame mpa_frame; - u8 mpa_frame_buf[NES_CM_DEFAULT_MTU]; + u8 mpa_frame_buf[MAX_CM_BUFFER]; }; u16 mpa_frame_size; struct iw_cm_id *cm_id; @@ -326,6 +329,7 @@ enum nes_cm_event_type { NES_CM_EVENT_MPA_REQ, NES_CM_EVENT_MPA_CONNECT, NES_CM_EVENT_MPA_ACCEPT, + NES_CM_EVENT_MPA_REJECT, NES_CM_EVENT_MPA_ESTABLISHED, NES_CM_EVENT_CONNECTED, NES_CM_EVENT_CLOSED, |