diff options
Diffstat (limited to 'drivers/infiniband/hw/cxgb3/iwch_cm.c')
-rw-r--r-- | drivers/infiniband/hw/cxgb3/iwch_cm.c | 2081 |
1 files changed, 2081 insertions, 0 deletions
diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c new file mode 100644 index 00000000000..a522b1baa3b --- /dev/null +++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c @@ -0,0 +1,2081 @@ +/* + * Copyright (c) 2006 Chelsio, Inc. All rights reserved. + * Copyright (c) 2006 Open Grid Computing, Inc. All rights reserved. + * + * This software is available to you under a choice of one of two + * licenses. You may choose to be licensed under the terms of the GNU + * General Public License (GPL) Version 2, available from the file + * COPYING in the main directory of this source tree, or the + * OpenIB.org BSD license below: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * - Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * - Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#include <linux/module.h> +#include <linux/list.h> +#include <linux/workqueue.h> +#include <linux/skbuff.h> +#include <linux/timer.h> +#include <linux/notifier.h> + +#include <net/neighbour.h> +#include <net/netevent.h> +#include <net/route.h> + +#include "tcb.h" +#include "cxgb3_offload.h" +#include "iwch.h" +#include "iwch_provider.h" +#include "iwch_cm.h" + +static char *states[] = { + "idle", + "listen", + "connecting", + "mpa_wait_req", + "mpa_req_sent", + "mpa_req_rcvd", + "mpa_rep_sent", + "fpdu_mode", + "aborting", + "closing", + "moribund", + "dead", + NULL, +}; + +static int ep_timeout_secs = 10; +module_param(ep_timeout_secs, int, 0444); +MODULE_PARM_DESC(ep_timeout_secs, "CM Endpoint operation timeout " + "in seconds (default=10)"); + +static int mpa_rev = 1; +module_param(mpa_rev, int, 0444); +MODULE_PARM_DESC(mpa_rev, "MPA Revision, 0 supports amso1100, " + "1 is spec compliant. (default=1)"); + +static int markers_enabled = 0; +module_param(markers_enabled, int, 0444); +MODULE_PARM_DESC(markers_enabled, "Enable MPA MARKERS (default(0)=disabled)"); + +static int crc_enabled = 1; +module_param(crc_enabled, int, 0444); +MODULE_PARM_DESC(crc_enabled, "Enable MPA CRC (default(1)=enabled)"); + +static int rcv_win = 256 * 1024; +module_param(rcv_win, int, 0444); +MODULE_PARM_DESC(rcv_win, "TCP receive window in bytes (default=256)"); + +static int snd_win = 32 * 1024; +module_param(snd_win, int, 0444); +MODULE_PARM_DESC(snd_win, "TCP send window in bytes (default=32KB)"); + +static unsigned int nocong = 0; +module_param(nocong, uint, 0444); +MODULE_PARM_DESC(nocong, "Turn off congestion control (default=0)"); + +static unsigned int cong_flavor = 1; +module_param(cong_flavor, uint, 0444); +MODULE_PARM_DESC(cong_flavor, "TCP Congestion control flavor (default=1)"); + +static void process_work(struct work_struct *work); +static struct workqueue_struct *workq; +static DECLARE_WORK(skb_work, process_work); + +static struct sk_buff_head rxq; +static cxgb3_cpl_handler_func work_handlers[NUM_CPL_CMDS]; + +static struct sk_buff *get_skb(struct sk_buff *skb, int len, gfp_t gfp); +static void ep_timeout(unsigned long arg); +static void connect_reply_upcall(struct iwch_ep *ep, int status); + +static void start_ep_timer(struct iwch_ep *ep) +{ + PDBG("%s ep %p\n", __FUNCTION__, ep); + if (timer_pending(&ep->timer)) { + PDBG("%s stopped / restarted timer ep %p\n", __FUNCTION__, ep); + del_timer_sync(&ep->timer); + } else + get_ep(&ep->com); + ep->timer.expires = jiffies + ep_timeout_secs * HZ; + ep->timer.data = (unsigned long)ep; + ep->timer.function = ep_timeout; + add_timer(&ep->timer); +} + +static void stop_ep_timer(struct iwch_ep *ep) +{ + PDBG("%s ep %p\n", __FUNCTION__, ep); + del_timer_sync(&ep->timer); + put_ep(&ep->com); +} + +static void release_tid(struct t3cdev *tdev, u32 hwtid, struct sk_buff *skb) +{ + struct cpl_tid_release *req; + + skb = get_skb(skb, sizeof *req, GFP_KERNEL); + if (!skb) + return; + req = (struct cpl_tid_release *) skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_TID_RELEASE, hwtid)); + skb->priority = CPL_PRIORITY_SETUP; + tdev->send(tdev, skb); + return; +} + +int iwch_quiesce_tid(struct iwch_ep *ep) +{ + struct cpl_set_tcb_field *req; + struct sk_buff *skb = get_skb(NULL, sizeof(*req), GFP_KERNEL); + + if (!skb) + return -ENOMEM; + req = (struct cpl_set_tcb_field *) skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD)); + req->wr.wr_lo = htonl(V_WR_TID(ep->hwtid)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_SET_TCB_FIELD, ep->hwtid)); + req->reply = 0; + req->cpu_idx = 0; + req->word = htons(W_TCB_RX_QUIESCE); + req->mask = cpu_to_be64(1ULL << S_TCB_RX_QUIESCE); + req->val = cpu_to_be64(1 << S_TCB_RX_QUIESCE); + + skb->priority = CPL_PRIORITY_DATA; + ep->com.tdev->send(ep->com.tdev, skb); + return 0; +} + +int iwch_resume_tid(struct iwch_ep *ep) +{ + struct cpl_set_tcb_field *req; + struct sk_buff *skb = get_skb(NULL, sizeof(*req), GFP_KERNEL); + + if (!skb) + return -ENOMEM; + req = (struct cpl_set_tcb_field *) skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD)); + req->wr.wr_lo = htonl(V_WR_TID(ep->hwtid)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_SET_TCB_FIELD, ep->hwtid)); + req->reply = 0; + req->cpu_idx = 0; + req->word = htons(W_TCB_RX_QUIESCE); + req->mask = cpu_to_be64(1ULL << S_TCB_RX_QUIESCE); + req->val = 0; + + skb->priority = CPL_PRIORITY_DATA; + ep->com.tdev->send(ep->com.tdev, skb); + return 0; +} + +static void set_emss(struct iwch_ep *ep, u16 opt) +{ + PDBG("%s ep %p opt %u\n", __FUNCTION__, ep, opt); + ep->emss = T3C_DATA(ep->com.tdev)->mtus[G_TCPOPT_MSS(opt)] - 40; + if (G_TCPOPT_TSTAMP(opt)) + ep->emss -= 12; + if (ep->emss < 128) + ep->emss = 128; + PDBG("emss=%d\n", ep->emss); +} + +static enum iwch_ep_state state_read(struct iwch_ep_common *epc) +{ + unsigned long flags; + enum iwch_ep_state state; + + spin_lock_irqsave(&epc->lock, flags); + state = epc->state; + spin_unlock_irqrestore(&epc->lock, flags); + return state; +} + +static inline void __state_set(struct iwch_ep_common *epc, + enum iwch_ep_state new) +{ + epc->state = new; +} + +static void state_set(struct iwch_ep_common *epc, enum iwch_ep_state new) +{ + unsigned long flags; + + spin_lock_irqsave(&epc->lock, flags); + PDBG("%s - %s -> %s\n", __FUNCTION__, states[epc->state], states[new]); + __state_set(epc, new); + spin_unlock_irqrestore(&epc->lock, flags); + return; +} + +static void *alloc_ep(int size, gfp_t gfp) +{ + struct iwch_ep_common *epc; + + epc = kmalloc(size, gfp); + if (epc) { + memset(epc, 0, size); + kref_init(&epc->kref); + spin_lock_init(&epc->lock); + init_waitqueue_head(&epc->waitq); + } + PDBG("%s alloc ep %p\n", __FUNCTION__, epc); + return epc; +} + +void __free_ep(struct kref *kref) +{ + struct iwch_ep_common *epc; + epc = container_of(kref, struct iwch_ep_common, kref); + PDBG("%s ep %p state %s\n", __FUNCTION__, epc, states[state_read(epc)]); + kfree(epc); +} + +static void release_ep_resources(struct iwch_ep *ep) +{ + PDBG("%s ep %p tid %d\n", __FUNCTION__, ep, ep->hwtid); + cxgb3_remove_tid(ep->com.tdev, (void *)ep, ep->hwtid); + dst_release(ep->dst); + l2t_release(L2DATA(ep->com.tdev), ep->l2t); + if (ep->com.tdev->type == T3B) + release_tid(ep->com.tdev, ep->hwtid, NULL); + put_ep(&ep->com); +} + +static void process_work(struct work_struct *work) +{ + struct sk_buff *skb = NULL; + void *ep; + struct t3cdev *tdev; + int ret; + + while ((skb = skb_dequeue(&rxq))) { + ep = *((void **) (skb->cb)); + tdev = *((struct t3cdev **) (skb->cb + sizeof(void *))); + ret = work_handlers[G_OPCODE(ntohl((__force __be32)skb->csum))](tdev, skb, ep); + if (ret & CPL_RET_BUF_DONE) + kfree_skb(skb); + + /* + * ep was referenced in sched(), and is freed here. + */ + put_ep((struct iwch_ep_common *)ep); + } +} + +static int status2errno(int status) +{ + switch (status) { + case CPL_ERR_NONE: + return 0; + case CPL_ERR_CONN_RESET: + return -ECONNRESET; + case CPL_ERR_ARP_MISS: + return -EHOSTUNREACH; + case CPL_ERR_CONN_TIMEDOUT: + return -ETIMEDOUT; + case CPL_ERR_TCAM_FULL: + return -ENOMEM; + case CPL_ERR_CONN_EXIST: + return -EADDRINUSE; + default: + return -EIO; + } +} + +/* + * Try and reuse skbs already allocated... + */ +static struct sk_buff *get_skb(struct sk_buff *skb, int len, gfp_t gfp) +{ + if (skb) { + BUG_ON(skb_cloned(skb)); + skb_trim(skb, 0); + skb_get(skb); + } else { + skb = alloc_skb(len, gfp); + } + return skb; +} + +static struct rtable *find_route(struct t3cdev *dev, __be32 local_ip, + __be32 peer_ip, __be16 local_port, + __be16 peer_port, u8 tos) +{ + struct rtable *rt; + struct flowi fl = { + .oif = 0, + .nl_u = { + .ip4_u = { + .daddr = peer_ip, + .saddr = local_ip, + .tos = tos} + }, + .proto = IPPROTO_TCP, + .uli_u = { + .ports = { + .sport = local_port, + .dport = peer_port} + } + }; + + if (ip_route_output_flow(&rt, &fl, NULL, 0)) + return NULL; + return rt; +} + +static unsigned int find_best_mtu(const struct t3c_data *d, unsigned short mtu) +{ + int i = 0; + + while (i < d->nmtus - 1 && d->mtus[i + 1] <= mtu) + ++i; + return i; +} + +static void arp_failure_discard(struct t3cdev *dev, struct sk_buff *skb) +{ + PDBG("%s t3cdev %p\n", __FUNCTION__, dev); + kfree_skb(skb); +} + +/* + * Handle an ARP failure for an active open. + */ +static void act_open_req_arp_failure(struct t3cdev *dev, struct sk_buff *skb) +{ + printk(KERN_ERR MOD "ARP failure duing connect\n"); + kfree_skb(skb); +} + +/* + * Handle an ARP failure for a CPL_ABORT_REQ. Change it into a no RST variant + * and send it along. + */ +static void abort_arp_failure(struct t3cdev *dev, struct sk_buff *skb) +{ + struct cpl_abort_req *req = cplhdr(skb); + + PDBG("%s t3cdev %p\n", __FUNCTION__, dev); + req->cmd = CPL_ABORT_NO_RST; + cxgb3_ofld_send(dev, skb); +} + +static int send_halfclose(struct iwch_ep *ep, gfp_t gfp) +{ + struct cpl_close_con_req *req; + struct sk_buff *skb; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + skb = get_skb(NULL, sizeof(*req), gfp); + if (!skb) { + printk(KERN_ERR MOD "%s - failed to alloc skb\n", __FUNCTION__); + return -ENOMEM; + } + skb->priority = CPL_PRIORITY_DATA; + set_arp_failure_handler(skb, arp_failure_discard); + req = (struct cpl_close_con_req *) skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_CLOSE_CON)); + req->wr.wr_lo = htonl(V_WR_TID(ep->hwtid)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_CLOSE_CON_REQ, ep->hwtid)); + l2t_send(ep->com.tdev, skb, ep->l2t); + return 0; +} + +static int send_abort(struct iwch_ep *ep, struct sk_buff *skb, gfp_t gfp) +{ + struct cpl_abort_req *req; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + skb = get_skb(skb, sizeof(*req), gfp); + if (!skb) { + printk(KERN_ERR MOD "%s - failed to alloc skb.\n", + __FUNCTION__); + return -ENOMEM; + } + skb->priority = CPL_PRIORITY_DATA; + set_arp_failure_handler(skb, abort_arp_failure); + req = (struct cpl_abort_req *) skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_HOST_ABORT_CON_REQ)); + req->wr.wr_lo = htonl(V_WR_TID(ep->hwtid)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_ABORT_REQ, ep->hwtid)); + req->cmd = CPL_ABORT_SEND_RST; + l2t_send(ep->com.tdev, skb, ep->l2t); + return 0; +} + +static int send_connect(struct iwch_ep *ep) +{ + struct cpl_act_open_req *req; + struct sk_buff *skb; + u32 opt0h, opt0l, opt2; + unsigned int mtu_idx; + int wscale; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + + skb = get_skb(NULL, sizeof(*req), GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "%s - failed to alloc skb.\n", + __FUNCTION__); + return -ENOMEM; + } + mtu_idx = find_best_mtu(T3C_DATA(ep->com.tdev), dst_mtu(ep->dst)); + wscale = compute_wscale(rcv_win); + opt0h = V_NAGLE(0) | + V_NO_CONG(nocong) | + V_KEEP_ALIVE(1) | + F_TCAM_BYPASS | + V_WND_SCALE(wscale) | + V_MSS_IDX(mtu_idx) | + V_L2T_IDX(ep->l2t->idx) | V_TX_CHANNEL(ep->l2t->smt_idx); + opt0l = V_TOS((ep->tos >> 2) & M_TOS) | V_RCV_BUFSIZ(rcv_win>>10); + opt2 = V_FLAVORS_VALID(1) | V_CONG_CONTROL_FLAVOR(cong_flavor); + skb->priority = CPL_PRIORITY_SETUP; + set_arp_failure_handler(skb, act_open_req_arp_failure); + + req = (struct cpl_act_open_req *) skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_ACT_OPEN_REQ, ep->atid)); + req->local_port = ep->com.local_addr.sin_port; + req->peer_port = ep->com.remote_addr.sin_port; + req->local_ip = ep->com.local_addr.sin_addr.s_addr; + req->peer_ip = ep->com.remote_addr.sin_addr.s_addr; + req->opt0h = htonl(opt0h); + req->opt0l = htonl(opt0l); + req->params = 0; + req->opt2 = htonl(opt2); + l2t_send(ep->com.tdev, skb, ep->l2t); + return 0; +} + +static void send_mpa_req(struct iwch_ep *ep, struct sk_buff *skb) +{ + int mpalen; + struct tx_data_wr *req; + struct mpa_message *mpa; + int len; + + PDBG("%s ep %p pd_len %d\n", __FUNCTION__, ep, ep->plen); + + BUG_ON(skb_cloned(skb)); + + mpalen = sizeof(*mpa) + ep->plen; + if (skb->data + mpalen + sizeof(*req) > skb->end) { + kfree_skb(skb); + skb=alloc_skb(mpalen + sizeof(*req), GFP_KERNEL); + if (!skb) { + connect_reply_upcall(ep, -ENOMEM); + return; + } + } + skb_trim(skb, 0); + skb_reserve(skb, sizeof(*req)); + skb_put(skb, mpalen); + skb->priority = CPL_PRIORITY_DATA; + mpa = (struct mpa_message *) skb->data; + memset(mpa, 0, sizeof(*mpa)); + memcpy(mpa->key, MPA_KEY_REQ, sizeof(mpa->key)); + mpa->flags = (crc_enabled ? MPA_CRC : 0) | + (markers_enabled ? MPA_MARKERS : 0); + mpa->private_data_size = htons(ep->plen); + mpa->revision = mpa_rev; + + if (ep->plen) + memcpy(mpa->private_data, ep->mpa_pkt + sizeof(*mpa), ep->plen); + + /* + * Reference the mpa skb. This ensures the data area + * will remain in memory until the hw acks the tx. + * Function tx_ack() will deref it. + */ + skb_get(skb); + set_arp_failure_handler(skb, arp_failure_discard); + skb->h.raw = skb->data; + len = skb->len; + req = (struct tx_data_wr *) skb_push(skb, sizeof(*req)); + req->wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_TX_DATA)); + req->wr_lo = htonl(V_WR_TID(ep->hwtid)); + req->len = htonl(len); + req->param = htonl(V_TX_PORT(ep->l2t->smt_idx) | + V_TX_SNDBUF(snd_win>>15)); + req->flags = htonl(F_TX_IMM_ACK|F_TX_INIT); + req->sndseq = htonl(ep->snd_seq); + BUG_ON(ep->mpa_skb); + ep->mpa_skb = skb; + l2t_send(ep->com.tdev, skb, ep->l2t); + start_ep_timer(ep); + state_set(&ep->com, MPA_REQ_SENT); + return; +} + +static int send_mpa_reject(struct iwch_ep *ep, const void *pdata, u8 plen) +{ + int mpalen; + struct tx_data_wr *req; + struct mpa_message *mpa; + struct sk_buff *skb; + + PDBG("%s ep %p plen %d\n", __FUNCTION__, ep, plen); + + mpalen = sizeof(*mpa) + plen; + + skb = get_skb(NULL, mpalen + sizeof(*req), GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "%s - cannot alloc skb!\n", __FUNCTION__); + return -ENOMEM; + } + skb_reserve(skb, sizeof(*req)); + mpa = (struct mpa_message *) skb_put(skb, mpalen); + memset(mpa, 0, sizeof(*mpa)); + memcpy(mpa->key, MPA_KEY_REP, sizeof(mpa->key)); + mpa->flags = MPA_REJECT; + mpa->revision = mpa_rev; + mpa->private_data_size = htons(plen); + if (plen) + memcpy(mpa->private_data, pdata, plen); + + /* + * Reference the mpa skb again. This ensures the data area + * will remain in memory until the hw acks the tx. + * Function tx_ack() will deref it. + */ + skb_get(skb); + skb->priority = CPL_PRIORITY_DATA; + set_arp_failure_handler(skb, arp_failure_discard); + skb->h.raw = skb->data; + req = (struct tx_data_wr *) skb_push(skb, sizeof(*req)); + req->wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_TX_DATA)); + req->wr_lo = htonl(V_WR_TID(ep->hwtid)); + req->len = htonl(mpalen); + req->param = htonl(V_TX_PORT(ep->l2t->smt_idx) | + V_TX_SNDBUF(snd_win>>15)); + req->flags = htonl(F_TX_IMM_ACK|F_TX_INIT); + req->sndseq = htonl(ep->snd_seq); + BUG_ON(ep->mpa_skb); + ep->mpa_skb = skb; + l2t_send(ep->com.tdev, skb, ep->l2t); + return 0; +} + +static int send_mpa_reply(struct iwch_ep *ep, const void *pdata, u8 plen) +{ + int mpalen; + struct tx_data_wr *req; + struct mpa_message *mpa; + int len; + struct sk_buff *skb; + + PDBG("%s ep %p plen %d\n", __FUNCTION__, ep, plen); + + mpalen = sizeof(*mpa) + plen; + + skb = get_skb(NULL, mpalen + sizeof(*req), GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "%s - cannot alloc skb!\n", __FUNCTION__); + return -ENOMEM; + } + skb->priority = CPL_PRIORITY_DATA; + skb_reserve(skb, sizeof(*req)); + mpa = (struct mpa_message *) skb_put(skb, mpalen); + memset(mpa, 0, sizeof(*mpa)); + memcpy(mpa->key, MPA_KEY_REP, sizeof(mpa->key)); + mpa->flags = (ep->mpa_attr.crc_enabled ? MPA_CRC : 0) | + (markers_enabled ? MPA_MARKERS : 0); + mpa->revision = mpa_rev; + mpa->private_data_size = htons(plen); + if (plen) + memcpy(mpa->private_data, pdata, plen); + + /* + * Reference the mpa skb. This ensures the data area + * will remain in memory until the hw acks the tx. + * Function tx_ack() will deref it. + */ + skb_get(skb); + set_arp_failure_handler(skb, arp_failure_discard); + skb->h.raw = skb->data; + len = skb->len; + req = (struct tx_data_wr *) skb_push(skb, sizeof(*req)); + req->wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_TX_DATA)); + req->wr_lo = htonl(V_WR_TID(ep->hwtid)); + req->len = htonl(len); + req->param = htonl(V_TX_PORT(ep->l2t->smt_idx) | + V_TX_SNDBUF(snd_win>>15)); + req->flags = htonl(F_TX_MORE | F_TX_IMM_ACK | F_TX_INIT); + req->sndseq = htonl(ep->snd_seq); + ep->mpa_skb = skb; + state_set(&ep->com, MPA_REP_SENT); + l2t_send(ep->com.tdev, skb, ep->l2t); + return 0; +} + +static int act_establish(struct t3cdev *tdev, struct sk_buff *skb, void *ctx) +{ + struct iwch_ep *ep = ctx; + struct cpl_act_establish *req = cplhdr(skb); + unsigned int tid = GET_TID(req); + + PDBG("%s ep %p tid %d\n", __FUNCTION__, ep, tid); + + dst_confirm(ep->dst); + + /* setup the hwtid for this connection */ + ep->hwtid = tid; + cxgb3_insert_tid(ep->com.tdev, &t3c_client, ep, tid); + + ep->snd_seq = ntohl(req->snd_isn); + + set_emss(ep, ntohs(req->tcp_opt)); + + /* dealloc the atid */ + cxgb3_free_atid(ep->com.tdev, ep->atid); + + /* start MPA negotiation */ + send_mpa_req(ep, skb); + + return 0; +} + +static void abort_connection(struct iwch_ep *ep, struct sk_buff *skb, gfp_t gfp) +{ + PDBG("%s ep %p\n", __FILE__, ep); + state_set(&ep->com, ABORTING); + send_abort(ep, skb, gfp); +} + +static void close_complete_upcall(struct iwch_ep *ep) +{ + struct iw_cm_event event; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_CLOSE; + if (ep->com.cm_id) { + PDBG("close complete delivered ep %p cm_id %p tid %d\n", + ep, ep->com.cm_id, ep->hwtid); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + ep->com.cm_id->rem_ref(ep->com.cm_id); + ep->com.cm_id = NULL; + ep->com.qp = NULL; + } +} + +static void peer_close_upcall(struct iwch_ep *ep) +{ + struct iw_cm_event event; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_DISCONNECT; + if (ep->com.cm_id) { + PDBG("peer close delivered ep %p cm_id %p tid %d\n", + ep, ep->com.cm_id, ep->hwtid); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + } +} + +static void peer_abort_upcall(struct iwch_ep *ep) +{ + struct iw_cm_event event; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_CLOSE; + event.status = -ECONNRESET; + if (ep->com.cm_id) { + PDBG("abort delivered ep %p cm_id %p tid %d\n", ep, + ep->com.cm_id, ep->hwtid); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + ep->com.cm_id->rem_ref(ep->com.cm_id); + ep->com.cm_id = NULL; + ep->com.qp = NULL; + } +} + +static void connect_reply_upcall(struct iwch_ep *ep, int status) +{ + struct iw_cm_event event; + + PDBG("%s ep %p status %d\n", __FUNCTION__, ep, status); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_CONNECT_REPLY; + event.status = status; + event.local_addr = ep->com.local_addr; + event.remote_addr = ep->com.remote_addr; + + if ((status == 0) || (status == -ECONNREFUSED)) { + event.private_data_len = ep->plen; + event.private_data = ep->mpa_pkt + sizeof(struct mpa_message); + } + if (ep->com.cm_id) { + PDBG("%s ep %p tid %d status %d\n", __FUNCTION__, ep, + ep->hwtid, status); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + } + if (status < 0) { + ep->com.cm_id->rem_ref(ep->com.cm_id); + ep->com.cm_id = NULL; + ep->com.qp = NULL; + } +} + +static void connect_request_upcall(struct iwch_ep *ep) +{ + struct iw_cm_event event; + + PDBG("%s ep %p tid %d\n", __FUNCTION__, ep, ep->hwtid); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_CONNECT_REQUEST; + event.local_addr = ep->com.local_addr; + event.remote_addr = ep->com.remote_addr; + event.private_data_len = ep->plen; + event.private_data = ep->mpa_pkt + sizeof(struct mpa_message); + event.provider_data = ep; + if (state_read(&ep->parent_ep->com) != DEAD) + ep->parent_ep->com.cm_id->event_handler( + ep->parent_ep->com.cm_id, + &event); + put_ep(&ep->parent_ep->com); + ep->parent_ep = NULL; +} + +static void established_upcall(struct iwch_ep *ep) +{ + struct iw_cm_event event; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + memset(&event, 0, sizeof(event)); + event.event = IW_CM_EVENT_ESTABLISHED; + if (ep->com.cm_id) { + PDBG("%s ep %p tid %d\n", __FUNCTION__, ep, ep->hwtid); + ep->com.cm_id->event_handler(ep->com.cm_id, &event); + } +} + +static int update_rx_credits(struct iwch_ep *ep, u32 credits) +{ + struct cpl_rx_data_ack *req; + struct sk_buff *skb; + + PDBG("%s ep %p credits %u\n", __FUNCTION__, ep, credits); + skb = get_skb(NULL, sizeof(*req), GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "update_rx_credits - cannot alloc skb!\n"); + return 0; + } + + req = (struct cpl_rx_data_ack *) skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_RX_DATA_ACK, ep->hwtid)); + req->credit_dack = htonl(V_RX_CREDITS(credits) | V_RX_FORCE_ACK(1)); + skb->priority = CPL_PRIORITY_ACK; + ep->com.tdev->send(ep->com.tdev, skb); + return credits; +} + +static void process_mpa_reply(struct iwch_ep *ep, struct sk_buff *skb) +{ + struct mpa_message *mpa; + u16 plen; + struct iwch_qp_attributes attrs; + enum iwch_qp_attr_mask mask; + int err; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + + /* + * Stop mpa timer. If it expired, then the state has + * changed and we bail since ep_timeout already aborted + * the connection. + */ + stop_ep_timer(ep); + if (state_read(&ep->com) != MPA_REQ_SENT) + return; + + /* + * If we get more than the supported amount of private data + * then we must fail this connection. + */ + if (ep->mpa_pkt_len + skb->len > sizeof(ep->mpa_pkt)) { + err = -EINVAL; + goto err; + } + + /* + * copy the new data into our accumulation buffer. + */ + memcpy(&(ep->mpa_pkt[ep->mpa_pkt_len]), skb->data, skb->len); + ep->mpa_pkt_len += skb->len; + + /* + * if we don't even have the mpa message, then bail. + */ + if (ep->mpa_pkt_len < sizeof(*mpa)) + return; + mpa = (struct mpa_message *) ep->mpa_pkt; + + /* Validate MPA header. */ + if (mpa->revision != mpa_rev) { + err = -EPROTO; + goto err; + } + if (memcmp(mpa->key, MPA_KEY_REP, sizeof(mpa->key))) { + err = -EPROTO; + goto err; + } + + plen = ntohs(mpa->private_data_size); + + /* + * Fail if there's too much private data. + */ + if (plen > MPA_MAX_PRIVATE_DATA) { + err = -EPROTO; + goto err; + } + + /* + * If plen does not account for pkt size + */ + if (ep->mpa_pkt_len > (sizeof(*mpa) + plen)) { + err = -EPROTO; + goto err; + } + + ep->plen = (u8) plen; + + /* + * If we don't have all the pdata yet, then bail. + * We'll continue process when more data arrives. + */ + if (ep->mpa_pkt_len < (sizeof(*mpa) + plen)) + return; + + if (mpa->flags & MPA_REJECT) { + err = -ECONNREFUSED; + goto err; + } + + /* + * If we get here we have accumulated the entire mpa + * start reply message including private data. And + * the MPA header is valid. + */ + state_set(&ep->com, FPDU_MODE); + ep->mpa_attr.crc_enabled = (mpa->flags & MPA_CRC) | crc_enabled ? 1 : 0; + ep->mpa_attr.recv_marker_enabled = markers_enabled; + ep->mpa_attr.xmit_marker_enabled = mpa->flags & MPA_MARKERS ? 1 : 0; + ep->mpa_attr.version = mpa_rev; + PDBG("%s - crc_enabled=%d, recv_marker_enabled=%d, " + "xmit_marker_enabled=%d, version=%d\n", __FUNCTION__, + ep->mpa_attr.crc_enabled, ep->mpa_attr.recv_marker_enabled, + ep->mpa_attr.xmit_marker_enabled, ep->mpa_attr.version); + + attrs.mpa_attr = ep->mpa_attr; + attrs.max_ird = ep->ird; + attrs.max_ord = ep->ord; + attrs.llp_stream_handle = ep; + attrs.next_state = IWCH_QP_STATE_RTS; + + mask = IWCH_QP_ATTR_NEXT_STATE | + IWCH_QP_ATTR_LLP_STREAM_HANDLE | IWCH_QP_ATTR_MPA_ATTR | + IWCH_QP_ATTR_MAX_IRD | IWCH_QP_ATTR_MAX_ORD; + + /* bind QP and TID with INIT_WR */ + err = iwch_modify_qp(ep->com.qp->rhp, + ep->com.qp, mask, &attrs, 1); + if (!err) + goto out; +err: + abort_connection(ep, skb, GFP_KERNEL); +out: + connect_reply_upcall(ep, err); + return; +} + +static void process_mpa_request(struct iwch_ep *ep, struct sk_buff *skb) +{ + struct mpa_message *mpa; + u16 plen; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + + /* + * Stop mpa timer. If it expired, then the state has + * changed and we bail since ep_timeout already aborted + * the connection. + */ + stop_ep_timer(ep); + if (state_read(&ep->com) != MPA_REQ_WAIT) + return; + + /* + * If we get more than the supported amount of private data + * then we must fail this connection. + */ + if (ep->mpa_pkt_len + skb->len > sizeof(ep->mpa_pkt)) { + abort_connection(ep, skb, GFP_KERNEL); + return; + } + + PDBG("%s enter (%s line %u)\n", __FUNCTION__, __FILE__, __LINE__); + + /* + * Copy the new data into our accumulation buffer. + */ + memcpy(&(ep->mpa_pkt[ep->mpa_pkt_len]), skb->data, skb->len); + ep->mpa_pkt_len += skb->len; + + /* + * If we don't even have the mpa message, then bail. + * We'll continue process when more data arrives. + */ + if (ep->mpa_pkt_len < sizeof(*mpa)) + return; + PDBG("%s enter (%s line %u)\n", __FUNCTION__, __FILE__, __LINE__); + mpa = (struct mpa_message *) ep->mpa_pkt; + + /* + * Validate MPA Header. + */ + if (mpa->revision != mpa_rev) { + abort_connection(ep, skb, GFP_KERNEL); + return; + } + + if (memcmp(mpa->key, MPA_KEY_REQ, sizeof(mpa->key))) { + abort_connection(ep, skb, GFP_KERNEL); + return; + } + + plen = ntohs(mpa->private_data_size); + + /* + * Fail if there's too much private data. + */ + if (plen > MPA_MAX_PRIVATE_DATA) { + abort_connection(ep, skb, GFP_KERNEL); + return; + } + + /* + * If plen does not account for pkt size + */ + if (ep->mpa_pkt_len > (sizeof(*mpa) + plen)) { + abort_connection(ep, skb, GFP_KERNEL); + return; + } + ep->plen = (u8) plen; + + /* + * If we don't have all the pdata yet, then bail. + */ + if (ep->mpa_pkt_len < (sizeof(*mpa) + plen)) + return; + + /* + * If we get here we have accumulated the entire mpa + * start reply message including private data. + */ + ep->mpa_attr.crc_enabled = (mpa->flags & MPA_CRC) | crc_enabled ? 1 : 0; + ep->mpa_attr.recv_marker_enabled = markers_enabled; + ep->mpa_attr.xmit_marker_enabled = mpa->flags & MPA_MARKERS ? 1 : 0; + ep->mpa_attr.version = mpa_rev; + PDBG("%s - crc_enabled=%d, recv_marker_enabled=%d, " + "xmit_marker_enabled=%d, version=%d\n", __FUNCTION__, + ep->mpa_attr.crc_enabled, ep->mpa_attr.recv_marker_enabled, + ep->mpa_attr.xmit_marker_enabled, ep->mpa_attr.version); + + state_set(&ep->com, MPA_REQ_RCVD); + + /* drive upcall */ + connect_request_upcall(ep); + return; +} + +static int rx_data(struct t3cdev *tdev, struct sk_buff *skb, void *ctx) +{ + struct iwch_ep *ep = ctx; + struct cpl_rx_data *hdr = cplhdr(skb); + unsigned int dlen = ntohs(hdr->len); + + PDBG("%s ep %p dlen %u\n", __FUNCTION__, ep, dlen); + + skb_pull(skb, sizeof(*hdr)); + skb_trim(skb, dlen); + + switch (state_read(&ep->com)) { + case MPA_REQ_SENT: + process_mpa_reply(ep, skb); + break; + case MPA_REQ_WAIT: + process_mpa_request(ep, skb); + break; + case MPA_REP_SENT: + break; + default: + printk(KERN_ERR MOD "%s Unexpected streaming data." + " ep %p state %d tid %d\n", + __FUNCTION__, ep, state_read(&ep->com), ep->hwtid); + + /* + * The ep will timeout and inform the ULP of the failure. + * See ep_timeout(). + */ + break; + } + + /* update RX credits */ + update_rx_credits(ep, dlen); + + return CPL_RET_BUF_DONE; +} + +/* + * Upcall from the adapter indicating data has been transmitted. + * For us its just the single MPA request or reply. We can now free + * the skb holding the mpa message. + */ +static int tx_ack(struct t3cdev *tdev, struct sk_buff *skb, void *ctx) +{ + struct iwch_ep *ep = ctx; + struct cpl_wr_ack *hdr = cplhdr(skb); + unsigned int credits = ntohs(hdr->credits); + enum iwch_qp_attr_mask mask; + + PDBG("%s ep %p credits %u\n", __FUNCTION__, ep, credits); + + if (credits == 0) + return CPL_RET_BUF_DONE; + BUG_ON(credits != 1); + BUG_ON(ep->mpa_skb == NULL); + kfree_skb(ep->mpa_skb); + ep->mpa_skb = NULL; + dst_confirm(ep->dst); + if (state_read(&ep->com) == MPA_REP_SENT) { + struct iwch_qp_attributes attrs; + + /* bind QP to EP and move to RTS */ + attrs.mpa_attr = ep->mpa_attr; + attrs.max_ird = ep->ord; + attrs.max_ord = ep->ord; + attrs.llp_stream_handle = ep; + attrs.next_state = IWCH_QP_STATE_RTS; + + /* bind QP and TID with INIT_WR */ + mask = IWCH_QP_ATTR_NEXT_STATE | + IWCH_QP_ATTR_LLP_STREAM_HANDLE | + IWCH_QP_ATTR_MPA_ATTR | + IWCH_QP_ATTR_MAX_IRD | + IWCH_QP_ATTR_MAX_ORD; + + ep->com.rpl_err = iwch_modify_qp(ep->com.qp->rhp, + ep->com.qp, mask, &attrs, 1); + + if (!ep->com.rpl_err) { + state_set(&ep->com, FPDU_MODE); + established_upcall(ep); + } + + ep->com.rpl_done = 1; + PDBG("waking up ep %p\n", ep); + wake_up(&ep->com.waitq); + } + return CPL_RET_BUF_DONE; +} + +static int abort_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx) +{ + struct iwch_ep *ep = ctx; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + + close_complete_upcall(ep); + state_set(&ep->com, DEAD); + release_ep_resources(ep); + return CPL_RET_BUF_DONE; +} + +static int act_open_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx) +{ + struct iwch_ep *ep = ctx; + struct cpl_act_open_rpl *rpl = cplhdr(skb); + + PDBG("%s ep %p status %u errno %d\n", __FUNCTION__, ep, rpl->status, + status2errno(rpl->status)); + connect_reply_upcall(ep, status2errno(rpl->status)); + state_set(&ep->com, DEAD); + if (ep->com.tdev->type == T3B) + release_tid(ep->com.tdev, GET_TID(rpl), NULL); + cxgb3_free_atid(ep->com.tdev, ep->atid); + dst_release(ep->dst); + l2t_release(L2DATA(ep->com.tdev), ep->l2t); + put_ep(&ep->com); + return CPL_RET_BUF_DONE; +} + +static int listen_start(struct iwch_listen_ep *ep) +{ + struct sk_buff *skb; + struct cpl_pass_open_req *req; + + PDBG("%s ep %p\n", __FUNCTION__, ep); + skb = get_skb(NULL, sizeof(*req), GFP_KERNEL); + if (!skb) { + printk(KERN_ERR MOD "t3c_listen_start failed to alloc skb!\n"); + return -ENOMEM; + } + + req = (struct cpl_pass_open_req *) skb_put(skb, sizeof(*req)); + req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD)); + OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_PASS_OPEN_REQ, ep->stid)); + req->local_port = ep->com.local_addr.sin_port; + req->local_ip = ep->com.local_addr.sin_addr.s_addr; + req->peer_port = 0; + req->peer_ip = 0; + req->peer_netmask = 0; + req->opt0h = htonl(F_DELACK | F_TCAM_BYPASS); + req->opt0l = htonl(V_RCV_BUFSIZ(rcv_win>>10)); + req->opt1 = htonl(V_CONN_PO |