iwch_cm.c 56.9 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
/*
 * Copyright (c) 2006 Chelsio, 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>
34
#include <linux/slab.h>
35
36
37
38
#include <linux/workqueue.h>
#include <linux/skbuff.h>
#include <linux/timer.h>
#include <linux/notifier.h>
39
#include <linux/inetdevice.h>
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66

#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,
};

67
68
69
70
int peer2peer = 0;
module_param(peer2peer, int, 0644);
MODULE_PARM_DESC(peer2peer, "Support peer2peer ULPs (default=0)");

71
static int ep_timeout_secs = 60;
72
module_param(ep_timeout_secs, int, 0644);
73
MODULE_PARM_DESC(ep_timeout_secs, "CM Endpoint operation timeout "
74
				   "in seconds (default=60)");
75
76

static int mpa_rev = 1;
77
module_param(mpa_rev, int, 0644);
78
79
80
81
MODULE_PARM_DESC(mpa_rev, "MPA Revision, 0 supports amso1100, "
		 "1 is spec compliant. (default=1)");

static int markers_enabled = 0;
82
module_param(markers_enabled, int, 0644);
83
84
85
MODULE_PARM_DESC(markers_enabled, "Enable MPA MARKERS (default(0)=disabled)");

static int crc_enabled = 1;
86
module_param(crc_enabled, int, 0644);
87
88
89
MODULE_PARM_DESC(crc_enabled, "Enable MPA CRC (default(1)=enabled)");

static int rcv_win = 256 * 1024;
90
module_param(rcv_win, int, 0644);
91
92
93
MODULE_PARM_DESC(rcv_win, "TCP receive window in bytes (default=256)");

static int snd_win = 32 * 1024;
94
module_param(snd_win, int, 0644);
95
96
97
MODULE_PARM_DESC(snd_win, "TCP send window in bytes (default=32KB)");

static unsigned int nocong = 0;
98
module_param(nocong, uint, 0644);
99
100
101
MODULE_PARM_DESC(nocong, "Turn off congestion control (default=0)");

static unsigned int cong_flavor = 1;
102
module_param(cong_flavor, uint, 0644);
103
104
105
106
107
108
109
110
111
112
113
114
MODULE_PARM_DESC(cong_flavor, "TCP Congestion control flavor (default=1)");

static struct workqueue_struct *workq;

static struct sk_buff_head rxq;

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)
{
115
	PDBG("%s ep %p\n", __func__, ep);
116
	if (timer_pending(&ep->timer)) {
117
		PDBG("%s stopped / restarted timer ep %p\n", __func__, ep);
118
119
120
121
122
123
124
125
126
127
128
		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)
{
129
	PDBG("%s ep %p\n", __func__, ep);
130
131
132
133
134
135
	if (!timer_pending(&ep->timer)) {
		printk(KERN_ERR "%s timer stopped when its not running!  ep %p state %u\n",
			__func__, ep, ep->com.state);
		WARN_ON(1);
		return;
	}
136
137
138
139
	del_timer_sync(&ep->timer);
	put_ep(&ep->com);
}

140
static int iwch_l2t_send(struct t3cdev *tdev, struct sk_buff *skb, struct l2t_entry *l2e)
Steve Wise's avatar
Steve Wise committed
141
142
143
144
145
146
147
148
149
150
{
	int	error = 0;
	struct cxio_rdev *rdev;

	rdev = (struct cxio_rdev *)tdev->ulp;
	if (cxio_fatal_error(rdev)) {
		kfree_skb(skb);
		return -EIO;
	}
	error = l2t_send(tdev, skb, l2e);
151
	if (error < 0)
Steve Wise's avatar
Steve Wise committed
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
		kfree_skb(skb);
	return error;
}

int iwch_cxgb3_ofld_send(struct t3cdev *tdev, struct sk_buff *skb)
{
	int	error = 0;
	struct cxio_rdev *rdev;

	rdev = (struct cxio_rdev *)tdev->ulp;
	if (cxio_fatal_error(rdev)) {
		kfree_skb(skb);
		return -EIO;
	}
	error = cxgb3_ofld_send(tdev, skb);
167
	if (error < 0)
Steve Wise's avatar
Steve Wise committed
168
169
170
171
		kfree_skb(skb);
	return error;
}

172
173
174
175
176
177
178
179
180
181
182
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;
Steve Wise's avatar
Steve Wise committed
183
	iwch_cxgb3_ofld_send(tdev, skb);
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
	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;
Steve Wise's avatar
Steve Wise committed
205
	return iwch_cxgb3_ofld_send(ep->com.tdev, skb);
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
}

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;
Steve Wise's avatar
Steve Wise committed
226
	return iwch_cxgb3_ofld_send(ep->com.tdev, skb);
227
228
229
230
}

static void set_emss(struct iwch_ep *ep, u16 opt)
{
231
	PDBG("%s ep %p opt %u\n", __func__, ep, opt);
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
	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;
}

Adrian Bunk's avatar
Adrian Bunk committed
251
static void __state_set(struct iwch_ep_common *epc, enum iwch_ep_state new)
252
253
254
255
256
257
258
259
260
{
	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);
261
	PDBG("%s - %s -> %s\n", __func__, states[epc->state], states[new]);
262
263
264
265
266
267
268
269
270
	__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;

271
	epc = kzalloc(size, gfp);
272
273
274
275
276
	if (epc) {
		kref_init(&epc->kref);
		spin_lock_init(&epc->lock);
		init_waitqueue_head(&epc->waitq);
	}
277
	PDBG("%s alloc ep %p\n", __func__, epc);
278
279
280
281
282
	return epc;
}

void __free_ep(struct kref *kref)
{
283
284
285
286
	struct iwch_ep *ep;
	ep = container_of(container_of(kref, struct iwch_ep_common, kref),
			  struct iwch_ep, com);
	PDBG("%s ep %p state %s\n", __func__, ep, states[state_read(&ep->com)]);
287
	if (test_bit(RELEASE_RESOURCES, &ep->com.flags)) {
288
289
290
291
292
		cxgb3_remove_tid(ep->com.tdev, (void *)ep, ep->hwtid);
		dst_release(ep->dst);
		l2t_release(L2DATA(ep->com.tdev), ep->l2t);
	}
	kfree(ep);
293
294
295
296
}

static void release_ep_resources(struct iwch_ep *ep)
{
297
	PDBG("%s ep %p tid %d\n", __func__, ep, ep->hwtid);
298
	set_bit(RELEASE_RESOURCES, &ep->com.flags);
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
	put_ep(&ep->com);
}

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)
{
327
	if (skb && !skb_is_nonlinear(skb) && !skb_cloned(skb)) {
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
		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}
			  }
	};

357
	if (ip_route_output_flow(&init_net, &rt, &fl, NULL, false))
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
		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)
{
373
	PDBG("%s t3cdev %p\n", __func__, dev);
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
	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);

394
	PDBG("%s t3cdev %p\n", __func__, dev);
395
	req->cmd = CPL_ABORT_NO_RST;
Steve Wise's avatar
Steve Wise committed
396
	iwch_cxgb3_ofld_send(dev, skb);
397
398
399
400
401
402
403
}

static int send_halfclose(struct iwch_ep *ep, gfp_t gfp)
{
	struct cpl_close_con_req *req;
	struct sk_buff *skb;

404
	PDBG("%s ep %p\n", __func__, ep);
405
406
	skb = get_skb(NULL, sizeof(*req), gfp);
	if (!skb) {
407
		printk(KERN_ERR MOD "%s - failed to alloc skb\n", __func__);
408
409
410
411
412
413
414
415
		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));
Steve Wise's avatar
Steve Wise committed
416
	return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
417
418
419
420
421
422
}

static int send_abort(struct iwch_ep *ep, struct sk_buff *skb, gfp_t gfp)
{
	struct cpl_abort_req *req;

423
	PDBG("%s ep %p\n", __func__, ep);
424
425
426
	skb = get_skb(skb, sizeof(*req), gfp);
	if (!skb) {
		printk(KERN_ERR MOD "%s - failed to alloc skb.\n",
427
		       __func__);
428
429
430
431
432
433
434
435
436
		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;
Steve Wise's avatar
Steve Wise committed
437
	return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
438
439
440
441
442
443
444
445
446
447
}

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;

448
	PDBG("%s ep %p\n", __func__, ep);
449
450
451
452

	skb = get_skb(NULL, sizeof(*req), GFP_KERNEL);
	if (!skb) {
		printk(KERN_ERR MOD "%s - failed to alloc skb.\n",
453
		       __func__);
454
455
456
457
458
459
460
461
462
463
464
465
		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);
466
467
	opt2 = F_RX_COALESCE_VALID | V_RX_COALESCE(0) | V_FLAVORS_VALID(1) |
	       V_CONG_CONTROL_FLAVOR(cong_flavor);
468
469
470
471
472
473
474
475
476
477
478
479
480
481
	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);
Steve Wise's avatar
Steve Wise committed
482
	return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
483
484
485
486
487
488
489
490
491
}

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;

492
	PDBG("%s ep %p pd_len %d\n", __func__, ep, ep->plen);
493
494
495
496

	BUG_ON(skb_cloned(skb));

	mpalen = sizeof(*mpa) + ep->plen;
497
	if (skb->data + mpalen + sizeof(*req) > skb_end_pointer(skb)) {
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
		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);
527
	skb_reset_transport_header(skb);
528
529
	len = skb->len;
	req = (struct tx_data_wr *) skb_push(skb, sizeof(*req));
530
	req->wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_TX_DATA)|F_WR_COMPL);
531
532
533
534
	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));
535
	req->flags = htonl(F_TX_INIT);
536
537
538
	req->sndseq = htonl(ep->snd_seq);
	BUG_ON(ep->mpa_skb);
	ep->mpa_skb = skb;
Steve Wise's avatar
Steve Wise committed
539
	iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
540
541
542
543
544
545
546
547
548
549
550
551
	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;

552
	PDBG("%s ep %p plen %d\n", __func__, ep, plen);
553
554
555
556
557

	mpalen = sizeof(*mpa) + plen;

	skb = get_skb(NULL, mpalen + sizeof(*req), GFP_KERNEL);
	if (!skb) {
558
		printk(KERN_ERR MOD "%s - cannot alloc skb!\n", __func__);
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
		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);
579
	skb_reset_transport_header(skb);
580
	req = (struct tx_data_wr *) skb_push(skb, sizeof(*req));
581
	req->wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_TX_DATA)|F_WR_COMPL);
582
583
584
585
	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));
586
	req->flags = htonl(F_TX_INIT);
587
588
589
	req->sndseq = htonl(ep->snd_seq);
	BUG_ON(ep->mpa_skb);
	ep->mpa_skb = skb;
Steve Wise's avatar
Steve Wise committed
590
	return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
591
592
593
594
595
596
597
598
599
600
}

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;

601
	PDBG("%s ep %p plen %d\n", __func__, ep, plen);
602
603
604
605
606

	mpalen = sizeof(*mpa) + plen;

	skb = get_skb(NULL, mpalen + sizeof(*req), GFP_KERNEL);
	if (!skb) {
607
		printk(KERN_ERR MOD "%s - cannot alloc skb!\n", __func__);
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
		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);
629
	skb_reset_transport_header(skb);
630
631
	len = skb->len;
	req = (struct tx_data_wr *) skb_push(skb, sizeof(*req));
632
	req->wr_hi = htonl(V_WR_OP(FW_WROPCODE_OFLD_TX_DATA)|F_WR_COMPL);
633
634
635
636
	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));
637
	req->flags = htonl(F_TX_INIT);
638
639
640
	req->sndseq = htonl(ep->snd_seq);
	ep->mpa_skb = skb;
	state_set(&ep->com, MPA_REP_SENT);
Steve Wise's avatar
Steve Wise committed
641
	return iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
642
643
644
645
646
647
648
649
}

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);

650
	PDBG("%s ep %p tid %d\n", __func__, ep, tid);
651
652
653
654
655
656
657
658

	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);
659
	ep->rcv_seq = ntohl(req->rcv_isn);
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682

	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;

683
	PDBG("%s ep %p\n", __func__, ep);
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
	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;

700
	PDBG("%s ep %p\n", __func__, ep);
701
702
703
704
705
706
707
708
709
710
711
712
713
	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;

714
	PDBG("%s ep %p\n", __func__, ep);
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
	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;

732
	PDBG("%s ep %p status %d\n", __func__, ep, status);
733
734
735
736
737
738
739
740
741
742
743
	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) {
744
		PDBG("%s ep %p tid %d status %d\n", __func__, ep,
745
746
747
748
749
750
751
752
753
754
755
756
757
758
		     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;

759
	PDBG("%s ep %p tid %d\n", __func__, ep, ep->hwtid);
760
761
762
763
764
765
766
	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;
767
768
	if (state_read(&ep->parent_ep->com) != DEAD) {
		get_ep(&ep->com);
769
770
771
		ep->parent_ep->com.cm_id->event_handler(
						ep->parent_ep->com.cm_id,
						&event);
772
	}
773
774
775
776
777
778
779
780
	put_ep(&ep->parent_ep->com);
	ep->parent_ep = NULL;
}

static void established_upcall(struct iwch_ep *ep)
{
	struct iw_cm_event event;

781
	PDBG("%s ep %p\n", __func__, ep);
782
783
784
	memset(&event, 0, sizeof(event));
	event.event = IW_CM_EVENT_ESTABLISHED;
	if (ep->com.cm_id) {
785
		PDBG("%s ep %p tid %d\n", __func__, ep, ep->hwtid);
786
787
788
789
790
791
792
793
794
		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;

795
	PDBG("%s ep %p credits %u\n", __func__, ep, credits);
796
797
798
799
800
801
802
803
804
805
806
	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;
Steve Wise's avatar
Steve Wise committed
807
	iwch_cxgb3_ofld_send(ep->com.tdev, skb);
808
809
810
811
812
813
814
815
816
817
818
	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;

819
	PDBG("%s ep %p\n", __func__, ep);
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841

	/*
	 * 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.
	 */
842
843
	skb_copy_from_linear_data(skb, &(ep->mpa_pkt[ep->mpa_pkt_len]),
				  skb->len);
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
	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);
901
	ep->mpa_attr.initiator = 1;
902
903
904
905
906
	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, "
907
	     "xmit_marker_enabled=%d, version=%d\n", __func__,
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
	     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);
924
925
926
927
928
929
930
931
	if (err)
		goto err;

	if (peer2peer && iwch_rqes_posted(ep->com.qp) == 0) {
		iwch_post_zb_read(ep->com.qp);
	}

	goto out;
932
933
934
935
936
937
938
939
940
941
942
943
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;

944
	PDBG("%s ep %p\n", __func__, ep);
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963

	/*
	 * 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;
	}

964
	PDBG("%s enter (%s line %u)\n", __func__, __FILE__, __LINE__);
965
966
967
968

	/*
	 * Copy the new data into our accumulation buffer.
	 */
969
970
	skb_copy_from_linear_data(skb, &(ep->mpa_pkt[ep->mpa_pkt_len]),
				  skb->len);
971
972
973
974
975
976
977
978
	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;
979
	PDBG("%s enter (%s line %u)\n", __func__, __FILE__, __LINE__);
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
	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.
	 */
1024
	ep->mpa_attr.initiator = 0;
1025
1026
1027
1028
1029
	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, "
1030
	     "xmit_marker_enabled=%d, version=%d\n", __func__,
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
	     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);

1047
	PDBG("%s ep %p dlen %u\n", __func__, ep, dlen);
1048
1049
1050
1051

	skb_pull(skb, sizeof(*hdr));
	skb_trim(skb, dlen);

1052
1053
1054
	ep->rcv_seq += dlen;
	BUG_ON(ep->rcv_seq != (ntohl(hdr->seq) + dlen));

1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
	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",
1067
		       __func__, ep, state_read(&ep->com), ep->hwtid);
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092

		/*
		 * 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);

1093
	PDBG("%s ep %p credits %u\n", __func__, ep, credits);
1094

1095
	if (credits == 0) {
1096
1097
		PDBG("%s 0 credit ack  ep %p state %u\n",
		     __func__, ep, state_read(&ep->com));
1098
		return CPL_RET_BUF_DONE;
1099
1100
	}

1101
1102
	BUG_ON(credits != 1);
	dst_confirm(ep->dst);
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
	if (!ep->mpa_skb) {
		PDBG("%s rdma_init wr_ack ep %p state %u\n",
			__func__, ep, state_read(&ep->com));
		if (ep->mpa_attr.initiator) {
			PDBG("%s initiator ep %p state %u\n",
				__func__, ep, state_read(&ep->com));
			if (peer2peer)
				iwch_post_zb_read(ep->com.qp);
		} else {
			PDBG("%s responder ep %p state %u\n",
				__func__, ep, state_read(&ep->com));
			ep->com.rpl_done = 1;
			wake_up(&ep->com.waitq);
		}
	} else {
		PDBG("%s lsm ack ep %p state %u freeing skb\n",
			__func__, ep, state_read(&ep->com));
		kfree_skb(ep->mpa_skb);
		ep->mpa_skb = NULL;
1122
1123
1124
1125
1126
1127
1128
	}
	return CPL_RET_BUF_DONE;
}

static int abort_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
{
	struct iwch_ep *ep = ctx;
1129
1130
	unsigned long flags;
	int release = 0;
1131

1132
	PDBG("%s ep %p\n", __func__, ep);
1133
	BUG_ON(!ep);
1134

1135
1136
1137
1138
	/*
	 * We get 2 abort replies from the HW.  The first one must
	 * be ignored except for scribbling that we need one more.
	 */
1139
	if (!test_and_set_bit(ABORT_REQ_IN_PROGRESS, &ep->com.flags)) {
1140
1141
1142
		return CPL_RET_BUF_DONE;
	}

1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
	spin_lock_irqsave(&ep->com.lock, flags);
	switch (ep->com.state) {
	case ABORTING:
		close_complete_upcall(ep);
		__state_set(&ep->com, DEAD);
		release = 1;
		break;
	default:
		printk(KERN_ERR "%s ep %p state %d\n",
		     __func__, ep, ep->com.state);
		break;
	}
	spin_unlock_irqrestore(&ep->com.lock, flags);

	if (release)
		release_ep_resources(ep);
1159
1160
1161
	return CPL_RET_BUF_DONE;
}

1162
1163
1164
1165
1166
1167
1168
1169
1170
/*
 * Return whether a failed active open has allocated a TID
 */
static inline int act_open_has_tid(int status)
{
	return status != CPL_ERR_TCAM_FULL && status != CPL_ERR_CONN_EXIST &&
	       status != CPL_ERR_ARP_MISS;
}

1171
1172
1173
1174
1175
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);

1176
	PDBG("%s ep %p status %u errno %d\n", __func__, ep, rpl->status,
1177
1178
1179
	     status2errno(rpl->status));
	connect_reply_upcall(ep, status2errno(rpl->status));
	state_set(&ep->com, DEAD);
1180
	if (ep->com.tdev->type != T3A && act_open_has_tid(rpl->status))
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
		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;

1194
	PDBG("%s ep %p\n", __func__, ep);
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
	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_POLICY(CPL_CONN_POLICY_ASK));

	skb->priority = 1;
Steve Wise's avatar
Steve Wise committed
1214
	return iwch_cxgb3_ofld_send(ep->com.tdev, skb);
1215
1216
1217
1218
1219
1220
1221
}

static int pass_open_rpl(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
{
	struct iwch_listen_ep *ep = ctx;
	struct cpl_pass_open_rpl *rpl = cplhdr(skb);

1222
	PDBG("%s ep %p status %d error %d\n", __func__, ep,
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
	     rpl->status, status2errno(rpl->status));
	ep->com.rpl_err = status2errno(rpl->status);
	ep->com.rpl_done = 1;
	wake_up(&ep->com.waitq);

	return CPL_RET_BUF_DONE;
}

static int listen_stop(struct iwch_listen_ep *ep)
{
	struct sk_buff *skb;
	struct cpl_close_listserv_req *req;

1236
	PDBG("%s ep %p\n", __func__, ep);
1237
1238
	skb = get_skb(NULL, sizeof(*req), GFP_KERNEL);
	if (!skb) {
1239
		printk(KERN_ERR MOD "%s - failed to alloc skb\n", __func__);
1240
1241
1242
1243
		return -ENOMEM;
	}
	req = (struct cpl_close_listserv_req *) skb_put(skb, sizeof(*req));
	req->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD));
1244
	req->cpu_idx = 0;
1245
1246
	OPCODE_TID(req) = htonl(MK_OPCODE_TID(CPL_CLOSE_LISTSRV_REQ, ep->stid));
	skb->priority = 1;
Steve Wise's avatar
Steve Wise committed
1247
	return iwch_cxgb3_ofld_send(ep->com.tdev, skb);
1248
1249
1250
1251
1252
1253
1254
1255
}

static int close_listsrv_rpl(struct t3cdev *tdev, struct sk_buff *skb,
			     void *ctx)
{
	struct iwch_listen_ep *ep = ctx;
	struct cpl_close_listserv_rpl *rpl = cplhdr(skb);

1256
	PDBG("%s ep %p\n", __func__, ep);
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
	ep->com.rpl_err = status2errno(rpl->status);
	ep->com.rpl_done = 1;
	wake_up(&ep->com.waitq);
	return CPL_RET_BUF_DONE;
}

static void accept_cr(struct iwch_ep *ep, __be32 peer_ip, struct sk_buff *skb)
{
	struct cpl_pass_accept_rpl *rpl;
	unsigned int mtu_idx;
	u32 opt0h, opt0l, opt2;
	int wscale;

1270
	PDBG("%s ep %p\n", __func__, ep);
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
	BUG_ON(skb_cloned(skb));
	skb_trim(skb, sizeof(*rpl));
	skb_get(skb);
	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);
1284
1285
	opt2 = F_RX_COALESCE_VALID | V_RX_COALESCE(0) | V_FLAVORS_VALID(1) |
	       V_CONG_CONTROL_FLAVOR(cong_flavor);
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295

	rpl = cplhdr(skb);
	rpl->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD));
	OPCODE_TID(rpl) = htonl(MK_OPCODE_TID(CPL_PASS_ACCEPT_RPL, ep->hwtid));
	rpl->peer_ip = peer_ip;
	rpl->opt0h = htonl(opt0h);
	rpl->opt0l_status = htonl(opt0l | CPL_PASS_OPEN_ACCEPT);
	rpl->opt2 = htonl(opt2);
	rpl->rsvd = rpl->opt2;	/* workaround for HW bug */
	skb->priority = CPL_PRIORITY_SETUP;
Steve Wise's avatar
Steve Wise committed
1296
	iwch_l2t_send(ep->com.tdev, skb, ep->l2t);
1297
1298
1299
1300
1301
1302
1303

	return;
}

static void reject_cr(struct t3cdev *tdev, u32 hwtid, __be32 peer_ip,
		      struct sk_buff *skb)
{
1304
	PDBG("%s t3cdev %p tid %u peer_ip %x\n", __func__, tdev, hwtid,
1305
1306
1307
1308
1309
	     peer_ip);
	BUG_ON(skb_cloned(skb));
	skb_trim(skb, sizeof(struct cpl_tid_release));
	skb_get(skb);

1310
	if (tdev->type != T3A)
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
		release_tid(tdev, hwtid, skb);
	else {
		struct cpl_pass_accept_rpl *rpl;

		rpl = cplhdr(skb);
		skb->priority = CPL_PRIORITY_SETUP;
		rpl->wr.wr_hi = htonl(V_WR_OP(FW_WROPCODE_FORWARD));
		OPCODE_TID(rpl) = htonl(MK_OPCODE_TID(CPL_PASS_ACCEPT_RPL,
						      hwtid));
		rpl->peer_ip = peer_ip;
		rpl->opt0h = htonl(F_TCAM_BYPASS);
		rpl->opt0l_status = htonl(CPL_PASS_OPEN_REJECT);
		rpl->opt2 = 0;
		rpl->rsvd = rpl->opt2;
Steve Wise's avatar
Steve Wise committed
1325
		iwch_cxgb3_ofld_send(tdev, skb);
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
	}
}

static int pass_accept_req(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
{
	struct iwch_ep *child_ep, *parent_ep = ctx;
	struct cpl_pass_accept_req *req = cplhdr(skb);
	unsigned int hwtid = GET_TID(req);
	struct dst_entry *dst;
	struct l2t_entry *l2t;
	struct rtable *rt;
	struct iff_mac tim;

1339
	PDBG("%s parent ep %p tid %u\n", __func__, parent_ep, hwtid);
1340
1341
1342

	if (state_read(&parent_ep->com) != LISTEN) {
		printk(KERN_ERR "%s - listening ep not in LISTEN\n",
1343
		       __func__);
1344
1345
1346
1347
1348
1349
1350
1351
1352
		goto reject;
	}

	/*
	 * Find the netdev for this connection request.
	 */
	tim.mac_addr = req->dst_mac;
	tim.vlan_tag = ntohs(req->vlan_tag);
	if (tdev->ctl(tdev, GET_IFF_FROM_MAC, &tim) < 0 || !tim.dev) {
1353
1354
		printk(KERN_ERR "%s bad dst mac %pM\n",
			__func__, req->dst_mac);
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
		goto reject;
	}

	/* Find output route */
	rt = find_route(tdev,
			req->local_ip,
			req->peer_ip,
			req->local_port,
			req->peer_port, G_PASS_OPEN_TOS(ntohl(req->tos_tid)));
	if (!rt) {
		printk(KERN_ERR MOD "%s - failed to find dst entry!\n",
1366
		       __func__);
1367
1368
		goto reject;
	}
1369
	dst = &rt->dst;
1370
1371
1372
	l2t = t3_l2t_get(tdev, dst->neighbour, dst->neighbour->dev);
	if (!l2t) {
		printk(KERN_ERR MOD "%s - failed to allocate l2t entry!\n",
1373
		       __func__);
1374
1375
1376
1377
1378
1379
		dst_release(dst);
		goto reject;
	}
	child_ep = alloc_ep(sizeof(*child_ep), GFP_KERNEL);
	if (!child_ep) {
		printk(KERN_ERR MOD "%s - failed to allocate ep entry!\n",
1380
		       __func__);
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
		l2t_release(L2DATA(tdev), l2t);
		dst_release(dst);
		goto reject;
	}
	state_set(&child_ep->com, CONNECTING);
	child_ep->com.tdev = tdev;
	child_ep->com.cm_id = NULL;
	child_ep->com.local_addr.sin_family = PF_INET;
	child_ep->com.local_addr.sin_port = req->local_port;
	child_ep->com.local_addr.sin_addr.s_addr = req->local_ip;
	child_ep->com.remote_addr.sin_family = PF_INET;
	child_ep->com.remote_addr.sin_port = req->peer_port;
	child_ep->com.remote_addr.sin_addr.s_addr = req->peer_ip;
	get_ep(&parent_ep->com);
	child_ep->parent_ep = parent_ep;
	child_ep->tos = G_PASS_OPEN_TOS(ntohl(req->tos_tid));
	child_ep->l2t = l2t;
	child_ep->dst = dst;
	child_ep->hwtid = hwtid;
	init_timer(&child_ep->timer);
	cxgb3_insert_tid(tdev, &t3c_client, child_ep, hwtid);
	accept_cr(child_ep, req->peer_ip, skb);
	goto out;
reject:
	reject_cr(tdev, hwtid, req->peer_ip, skb);
out:
	return CPL_RET_BUF_DONE;
}

static int pass_establish(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
{
	struct iwch_ep *ep = ctx;
	struct cpl_pass_establish *req = cplhdr(skb);

1415
	PDBG("%s ep %p\n", __func__, ep);
1416
	ep->snd_seq = ntohl(req->snd_isn);
1417
	ep->rcv_seq = ntohl(req->rcv_isn);
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435

	set_emss(ep, ntohs(req->tcp_opt));

	dst_confirm(ep->dst);
	state_set(&ep->com, MPA_REQ_WAIT);
	start_ep_timer(ep);

	return CPL_RET_BUF_DONE;
}

static int peer_close(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
{
	struct iwch_ep *ep = ctx;
	struct iwch_qp_attributes attrs;
	unsigned long flags;
	int disconnect = 1;
	int release = 0;

1436
	PDBG("%s ep %p\n", __func__, ep);
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
	dst_confirm(ep->dst);

	spin_lock_irqsave(&ep->com.lock, flags);
	switch (ep->com.state) {
	case MPA_REQ_WAIT:
		__state_set(&ep->com, CLOSING);
		break;
	case MPA_REQ_SENT:
		__state_set(&ep->com, CLOSING);
		connect_reply_upcall(ep, -ECONNRESET);
		break;
	case MPA_REQ_RCVD:

		/*
		 * We're gonna mark this puppy DEAD, but keep
		 * the reference on it until the ULP accepts or
1453
1454
		 * rejects the CR. Also wake up anyone waiting
		 * in rdma connection migration (see iwch_accept_cr()).
1455
1456
		 */
		__state_set(&ep->com, CLOSING);
1457
1458
1459
1460
		ep->com.rpl_done = 1;
		ep->com.rpl_err = -ECONNRESET;
		PDBG("waking up ep %p\n", ep);
		wake_up(&ep->com.waitq);
1461
1462
1463
1464
1465
1466
1467
1468
1469
		break;
	case MPA_REP_SENT:
		__state_set(&ep->com, CLOSING);
		ep->com.rpl_done = 1;
		ep->com.rpl_err = -ECONNRESET;
		PDBG("waking up ep %p\n", ep);
		wake_up(&ep->com.waitq);
		break;
	case FPDU_MODE:
1470
		start_ep_timer(ep);
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
		__state_set(&ep->com, CLOSING);
		attrs.next_state = IWCH_QP_STATE_CLOSING;
		iwch_modify_qp(ep->com.qp->rhp, ep->com.qp,
			       IWCH_QP_ATTR_NEXT_STATE, &attrs, 1);
		peer_close_upcall(ep);
		break;
	case ABORTING:
		disconnect = 0;
		break;
	case CLOSING:
		__state_set(&ep->com, MORIBUND);
		disconnect = 0;
		break;
	case MORIBUND:
		stop_ep_timer(ep);
		if (ep->com.cm_id && ep->com.qp) {
			attrs.next_state = IWCH_QP_STATE_IDLE;
			iwch_modify_qp(ep->com.qp->rhp, ep->com.qp,
				       IWCH_QP_ATTR_NEXT_STATE, &attrs, 1);
		}
		close_complete_upcall(ep);
		__state_set(&ep->com, DEAD);
		release = 1;
		disconnect = 0;
		break;
	case DEAD:
		disconnect = 0;
		break;
	default:
		BUG_ON(1);
	}
	spin_unlock_irqrestore(&ep->com.lock, flags);
	if (disconnect)
		iwch_ep_disconnect(ep, 0, GFP_KERNEL);
	if (release)
		release_ep_resources(ep);
	return CPL_RET_BUF_DONE;
}

/*
 * Returns whether an ABORT_REQ_RSS message is a negative advice.
 */
Adrian Bunk's avatar
Adrian Bunk committed
1513
static int is_neg_adv_abort(unsigned int status)
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
{
	return status == CPL_ERR_RTX_NEG_ADVICE ||
	       status == CPL_ERR_PERSIST_NEG_ADVICE;
}

static int peer_abort(struct t3cdev *tdev, struct sk_buff *skb, void *ctx)
{
	struct cpl_abort_req_rss *req = cplhdr(skb);
	struct iwch_ep *ep = ctx;
	struct cpl_abort_rpl *rpl;
	struct sk_buff *rpl_skb;
	struct iwch_qp_attributes attrs;
	int ret;
1527
1528
	int release = 0;
	unsigned long flags;
1529

1530
	if (is_neg_adv_abort(req->status)) {
1531
		PDBG("%s neg_adv_abort ep %p tid %d\n", __func__, ep,
1532
1533
1534
1535
1536
		     ep->hwtid);
		t3_l2t_send_event(ep->com.tdev, ep->l2t);
		return CPL_RET_BUF_DONE;
	}

1537
1538
1539
1540
	/*
	 * We get 2 peer aborts from the HW.  The first one must
	 * be ignored except for scribbling that we need one more.
	 */
1541
	if (!test_and_set_bit(PEER_ABORT_IN_PROGRESS, &ep->com.flags)) {
1542
1543
1544
		return CPL_RET_BUF_DONE;
	}

1545
1546
1547
	spin_lock_irqsave(&ep->com.lock, flags);
	PDBG("%s ep %p state %u\n", __func__, ep, ep->com.state);
	switch (ep->com.state) {
1548
1549
1550
	case CONNECTING:
		break;
	case MPA_REQ_WAIT:
1551
		stop_ep_timer(ep);
1552
1553
		break;
	case MPA_REQ_SENT:
1554
		stop_ep_timer(ep);
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
		connect_reply_upcall(ep, -ECONNRESET);
		break;
	case MPA_REP_SENT:
		ep->com.rpl_done = 1;
		ep->com.rpl_err = -ECONNRESET;
		PDBG("waking up ep %p\n", ep);
		wake_up(&ep->com.waitq);
		break;
	case MPA_REQ_RCVD:

		/*
		 * We're gonna mark this puppy DEAD, but keep
		 * the reference on it until the ULP accepts or
1568
1569
		 * rejects the CR. Also wake up anyone waiting
		 * in rdma connection migration (see iwch_accept_cr()).
1570
		 */
1571
1572
1573
1574
		ep->com.rpl_done = 1;
		ep->com.rpl_err = -ECONNRESET;
		PDBG("waking up ep %p\n", ep);
		wake_up(&ep->com.waitq);
1575
1576
		break;
	case MORIBUND:
1577
	case CLOSING:
1578