summaryrefslogtreecommitdiffstats
path: root/drivers/infiniband/ulp/iser/iser_verbs.c
blob: 5278916c31038cdea28ad9a524d2aa18d653e21d (plain)
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
34
35
36
37
38
39
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
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
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
327
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
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
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
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
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
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
/*
 * Copyright (c) 2004, 2005, 2006 Voltaire, Inc. All rights reserved.
 * Copyright (c) 2005, 2006 Cisco Systems.  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/kernel.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/delay.h>

#include "iscsi_iser.h"

#define ISCSI_ISER_MAX_CONN	8
#define ISER_MAX_RX_CQ_LEN	(ISER_QP_MAX_RECV_DTOS * ISCSI_ISER_MAX_CONN)
#define ISER_MAX_TX_CQ_LEN	(ISER_QP_MAX_REQ_DTOS  * ISCSI_ISER_MAX_CONN)

static void iser_cq_tasklet_fn(unsigned long data);
static void iser_cq_callback(struct ib_cq *cq, void *cq_context);

static void iser_cq_event_callback(struct ib_event *cause, void *context)
{
	iser_err("got cq event %d \n", cause->event);
}

static void iser_qp_event_callback(struct ib_event *cause, void *context)
{
	iser_err("got qp event %d\n",cause->event);
}

static void iser_event_handler(struct ib_event_handler *handler,
				struct ib_event *event)
{
	iser_err("async event %d on device %s port %d\n", event->event,
		event->device->name, event->element.port_num);
}

/**
 * iser_create_device_ib_res - creates Protection Domain (PD), Completion
 * Queue (CQ), DMA Memory Region (DMA MR) with the device associated with
 * the adapator.
 *
 * returns 0 on success, -1 on failure
 */
static int iser_create_device_ib_res(struct iser_device *device)
{
	int i, j;
	struct iser_cq_desc *cq_desc;

	device->cqs_used = min(ISER_MAX_CQ, device->ib_device->num_comp_vectors);
	iser_info("using %d CQs, device %s supports %d vectors\n",
		  device->cqs_used, device->ib_device->name,
		  device->ib_device->num_comp_vectors);

	device->cq_desc = kmalloc(sizeof(struct iser_cq_desc) * device->cqs_used,
				  GFP_KERNEL);
	if (device->cq_desc == NULL)
		goto cq_desc_err;
	cq_desc = device->cq_desc;

	device->pd = ib_alloc_pd(device->ib_device);
	if (IS_ERR(device->pd))
		goto pd_err;

	for (i = 0; i < device->cqs_used; i++) {
		cq_desc[i].device   = device;
		cq_desc[i].cq_index = i;

		device->rx_cq[i] = ib_create_cq(device->ib_device,
					  iser_cq_callback,
					  iser_cq_event_callback,
					  (void *)&cq_desc[i],
					  ISER_MAX_RX_CQ_LEN, i);
		if (IS_ERR(device->rx_cq[i]))
			goto cq_err;

		device->tx_cq[i] = ib_create_cq(device->ib_device,
					  NULL, iser_cq_event_callback,
					  (void *)&cq_desc[i],
					  ISER_MAX_TX_CQ_LEN, i);

		if (IS_ERR(device->tx_cq[i]))
			goto cq_err;

		if (ib_req_notify_cq(device->rx_cq[i], IB_CQ_NEXT_COMP))
			goto cq_err;

		tasklet_init(&device->cq_tasklet[i],
			     iser_cq_tasklet_fn,
			(unsigned long)&cq_desc[i]);
	}

	device->mr = ib_get_dma_mr(device->pd, IB_ACCESS_LOCAL_WRITE |
				   IB_ACCESS_REMOTE_WRITE |
				   IB_ACCESS_REMOTE_READ);
	if (IS_ERR(device->mr))
		goto dma_mr_err;

	INIT_IB_EVENT_HANDLER(&device->event_handler, device->ib_device,
				iser_event_handler);
	if (ib_register_event_handler(&device->event_handler))
		goto handler_err;

	return 0;

handler_err:
	ib_dereg_mr(device->mr);
dma_mr_err:
	for (j = 0; j < device->cqs_used; j++)
		tasklet_kill(&device->cq_tasklet[j]);
cq_err:
	for (j = 0; j < i; j++) {
		if (device->tx_cq[j])
			ib_destroy_cq(device->tx_cq[j]);
		if (device->rx_cq[j])
			ib_destroy_cq(device->rx_cq[j]);
	}
	ib_dealloc_pd(device->pd);
pd_err:
	kfree(device->cq_desc);
cq_desc_err:
	iser_err("failed to allocate an IB resource\n");
	return -1;
}

/**
 * iser_free_device_ib_res - destroy/dealloc/dereg the DMA MR,
 * CQ and PD created with the device associated with the adapator.
 */
static void iser_free_device_ib_res(struct iser_device *device)
{
	int i;
	BUG_ON(device->mr == NULL);

	for (i = 0; i < device->cqs_used; i++) {
		tasklet_kill(&device->cq_tasklet[i]);
		(void)ib_destroy_cq(device->tx_cq[i]);
		(void)ib_destroy_cq(device->rx_cq[i]);
		device->tx_cq[i] = NULL;
		device->rx_cq[i] = NULL;
	}

	(void)ib_unregister_event_handler(&device->event_handler);
	(void)ib_dereg_mr(device->mr);
	(void)ib_dealloc_pd(device->pd);

	kfree(device->cq_desc);

	device->mr = NULL;
	device->pd = NULL;
}

/**
 * iser_create_ib_conn_res - Creates FMR pool and Queue-Pair (QP)
 *
 * returns 0 on success, -1 on failure
 */
static int iser_create_ib_conn_res(struct iser_conn *ib_conn)
{
	struct iser_device	*device;
	struct ib_qp_init_attr	init_attr;
	int			req_err, resp_err, ret = -ENOMEM;
	struct ib_fmr_pool_param params;
	int index, min_index = 0;

	BUG_ON(ib_conn->device == NULL);

	device = ib_conn->device;

	ib_conn->login_buf = kmalloc(ISCSI_DEF_MAX_RECV_SEG_LEN +
					ISER_RX_LOGIN_SIZE, GFP_KERNEL);
	if (!ib_conn->login_buf)
		goto out_err;

	ib_conn->login_req_buf  = ib_conn->login_buf;
	ib_conn->login_resp_buf = ib_conn->login_buf + ISCSI_DEF_MAX_RECV_SEG_LEN;

	ib_conn->login_req_dma = ib_dma_map_single(ib_conn->device->ib_device,
				(void *)ib_conn->login_req_buf,
				ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_TO_DEVICE);

	ib_conn->login_resp_dma = ib_dma_map_single(ib_conn->device->ib_device,
				(void *)ib_conn->login_resp_buf,
				ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE);

	req_err  = ib_dma_mapping_error(device->ib_device, ib_conn->login_req_dma);
	resp_err = ib_dma_mapping_error(device->ib_device, ib_conn->login_resp_dma);

	if (req_err || resp_err) {
		if (req_err)
			ib_conn->login_req_dma = 0;
		if (resp_err)
			ib_conn->login_resp_dma = 0;
		goto out_err;
	}

	ib_conn->page_vec = kmalloc(sizeof(struct iser_page_vec) +
				    (sizeof(u64) * (ISCSI_ISER_SG_TABLESIZE +1)),
				    GFP_KERNEL);
	if (!ib_conn->page_vec)
		goto out_err;

	ib_conn->page_vec->pages = (u64 *) (ib_conn->page_vec + 1);

	params.page_shift        = SHIFT_4K;
	/* when the first/last SG element are not start/end *
	 * page aligned, the map whould be of N+1 pages     */
	params.max_pages_per_fmr = ISCSI_ISER_SG_TABLESIZE + 1;
	/* make the pool size twice the max number of SCSI commands *
	 * the ML is expected to queue, watermark for unmap at 50%  */
	params.pool_size	 = ISCSI_DEF_XMIT_CMDS_MAX * 2;
	params.dirty_watermark	 = ISCSI_DEF_XMIT_CMDS_MAX;
	params.cache		 = 0;
	params.flush_function	 = NULL;
	params.access		 = (IB_ACCESS_LOCAL_WRITE  |
				    IB_ACCESS_REMOTE_WRITE |
				    IB_ACCESS_REMOTE_READ);

	ib_conn->fmr_pool = ib_create_fmr_pool(device->pd, &params);
	ret = PTR_ERR(ib_conn->fmr_pool);
	if (IS_ERR(ib_conn->fmr_pool) && ret != -ENOSYS) {
		ib_conn->fmr_pool = NULL;
		goto out_err;
	} else if (ret == -ENOSYS) {
		ib_conn->fmr_pool = NULL;
		iser_warn("FMRs are not supported, using unaligned mode\n");
		ret = 0;
	}

	memset(&init_attr, 0, sizeof init_attr);

	mutex_lock(&ig.connlist_mutex);
	/* select the CQ with the minimal number of usages */
	for (index = 0; index < device->cqs_used; index++)
		if (device->cq_active_qps[index] <
		    device->cq_active_qps[min_index])
			min_index = index;
	device->cq_active_qps[min_index]++;
	mutex_unlock(&ig.connlist_mutex);
	iser_info("cq index %d used for ib_conn %p\n", min_index, ib_conn);

	init_attr.event_handler = iser_qp_event_callback;
	init_attr.qp_context	= (void *)ib_conn;
	init_attr.send_cq	= device->tx_cq[min_index];
	init_attr.recv_cq	= device->rx_cq[min_index];
	init_attr.cap.max_send_wr  = ISER_QP_MAX_REQ_DTOS;
	init_attr.cap.max_recv_wr  = ISER_QP_MAX_RECV_DTOS;
	init_attr.cap.max_send_sge = 2;
	init_attr.cap.max_recv_sge = 1;
	init_attr.sq_sig_type	= IB_SIGNAL_REQ_WR;
	init_attr.qp_type	= IB_QPT_RC;

	ret = rdma_create_qp(ib_conn->cma_id, device->pd, &init_attr);
	if (ret)
		goto out_err;

	ib_conn->qp = ib_conn->cma_id->qp;
	iser_info("setting conn %p cma_id %p: fmr_pool %p qp %p\n",
		  ib_conn, ib_conn->cma_id,
		  ib_conn->fmr_pool, ib_conn->cma_id->qp);
	return ret;

out_err:
	iser_err("unable to alloc mem or create resource, err %d\n", ret);
	return ret;
}

/**
 * releases the FMR pool, QP and CMA ID objects, returns 0 on success,
 * -1 on failure
 */
static int iser_free_ib_conn_res(struct iser_conn *ib_conn, int can_destroy_id)
{
	int cq_index;
	BUG_ON(ib_conn == NULL);

	iser_info("freeing conn %p cma_id %p fmr pool %p qp %p\n",
		  ib_conn, ib_conn->cma_id,
		  ib_conn->fmr_pool, ib_conn->qp);

	/* qp is created only once both addr & route are resolved */
	if (ib_conn->fmr_pool != NULL)
		ib_destroy_fmr_pool(ib_conn->fmr_pool);

	if (ib_conn->qp != NULL) {
		cq_index = ((struct iser_cq_desc *)ib_conn->qp->recv_cq->cq_context)->cq_index;
		ib_conn->device->cq_active_qps[cq_index]--;

		rdma_destroy_qp(ib_conn->cma_id);
	}
	/* if cma handler context, the caller acts s.t the cma destroy the id */
	if (ib_conn->cma_id != NULL && can_destroy_id)
		rdma_destroy_id(ib_conn->cma_id);

	ib_conn->fmr_pool = NULL;
	ib_conn->qp	  = NULL;
	ib_conn->cma_id   = NULL;
	kfree(ib_conn->page_vec);

	if (ib_conn->login_buf) {
		if (ib_conn->login_req_dma)
			ib_dma_unmap_single(ib_conn->device->ib_device,
				ib_conn->login_req_dma,
				ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_TO_DEVICE);
		if (ib_conn->login_resp_dma)
			ib_dma_unmap_single(ib_conn->device->ib_device,
				ib_conn->login_resp_dma,
				ISER_RX_LOGIN_SIZE, DMA_FROM_DEVICE);
		kfree(ib_conn->login_buf);
	}

	return 0;
}

/**
 * based on the resolved device node GUID see if there already allocated
 * device for this device. If there's no such, create one.
 */
static
struct iser_device *iser_device_find_by_ib_device(struct rdma_cm_id *cma_id)
{
	struct iser_device *device;

	mutex_lock(&ig.device_list_mutex);

	list_for_each_entry(device, &ig.device_list, ig_list)
		/* find if there's a match using the node GUID */
		if (device->ib_device->node_guid == cma_id->device->node_guid)
			goto inc_refcnt;

	device = kzalloc(sizeof *device, GFP_KERNEL);
	if (device == NULL)
		goto out;

	/* assign this device to the device */
	device->ib_device = cma_id->device;
	/* init the device and link it into ig device list */
	if (iser_create_device_ib_res(device)) {
		kfree(device);
		device = NULL;
		goto out;
	}
	list_add(&device->ig_list, &ig.device_list);

inc_refcnt:
	device->refcount++;
out:
	mutex_unlock(&ig.device_list_mutex);
	return device;
}

/* if there's no demand for this device, release it */
static void iser_device_try_release(struct iser_device *device)
{
	mutex_lock(&ig.device_list_mutex);
	device->refcount--;
	iser_info("device %p refcount %d\n", device, device->refcount);
	if (!device->refcount) {
		iser_free_device_ib_res(device);
		list_del(&device->ig_list);
		kfree(device);
	}
	mutex_unlock(&ig.device_list_mutex);
}

static int iser_conn_state_comp_exch(struct iser_conn *ib_conn,
				     enum iser_ib_conn_state comp,
				     enum iser_ib_conn_state exch)
{
	int ret;

	spin_lock_bh(&ib_conn->lock);
	if ((ret = (ib_conn->state == comp)))
		ib_conn->state = exch;
	spin_unlock_bh(&ib_conn->lock);
	return ret;
}

/**
 * Frees all conn objects and deallocs conn descriptor
 */
static void iser_conn_release(struct iser_conn *ib_conn, int can_destroy_id)
{
	struct iser_device  *device = ib_conn->device;

	BUG_ON(ib_conn->state != ISER_CONN_DOWN);

	mutex_lock(&ig.connlist_mutex);
	list_del(&ib_conn->conn_list);
	mutex_unlock(&ig.connlist_mutex);
	iser_free_rx_descriptors(ib_conn);
	iser_free_ib_conn_res(ib_conn, can_destroy_id);
	ib_conn->device = NULL;
	/* on EVENT_ADDR_ERROR there's no device yet for this conn */
	if (device != NULL)
		iser_device_try_release(device);
	iscsi_destroy_endpoint(ib_conn->ep);
}

void iser_conn_get(struct iser_conn *ib_conn)
{
	atomic_inc(&ib_conn->refcount);
}

int iser_conn_put(struct iser_conn *ib_conn, int can_destroy_id)
{
	if (atomic_dec_and_test(&ib_conn->refcount)) {
		iser_conn_release(ib_conn, can_destroy_id);
		return 1;
	}
	return 0;
}

/**
 * triggers start of the disconnect procedures and wait for them to be done
 */
void iser_conn_terminate(struct iser_conn *ib_conn)
{
	int err = 0;

	/* change the ib conn state only if the conn is UP, however always call
	 * rdma_disconnect since this is the only way to cause the CMA to change
	 * the QP state to ERROR
	 */

	iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP, ISER_CONN_TERMINATING);
	err = rdma_disconnect(ib_conn->cma_id);
	if (err)
		iser_err("Failed to disconnect, conn: 0x%p err %d\n",
			 ib_conn,err);

	wait_event_interruptible(ib_conn->wait,
				 ib_conn->state == ISER_CONN_DOWN);

	iser_conn_put(ib_conn, 1); /* deref ib conn deallocate */
}

static int iser_connect_error(struct rdma_cm_id *cma_id)
{
	struct iser_conn *ib_conn;
	ib_conn = (struct iser_conn *)cma_id->context;

	ib_conn->state = ISER_CONN_DOWN;
	wake_up_interruptible(&ib_conn->wait);
	return iser_conn_put(ib_conn, 0); /* deref ib conn's cma id */
}

static int iser_addr_handler(struct rdma_cm_id *cma_id)
{
	struct iser_device *device;
	struct iser_conn   *ib_conn;
	int    ret;

	device = iser_device_find_by_ib_device(cma_id);
	if (!device) {
		iser_err("device lookup/creation failed\n");
		return iser_connect_error(cma_id);
	}

	ib_conn = (struct iser_conn *)cma_id->context;
	ib_conn->device = device;

	ret = rdma_resolve_route(cma_id, 1000);
	if (ret) {
		iser_err("resolve route failed: %d\n", ret);
		return iser_connect_error(cma_id);
	}

	return 0;
}

static int iser_route_handler(struct rdma_cm_id *cma_id)
{
	struct rdma_conn_param conn_param;
	int    ret;
	struct iser_cm_hdr req_hdr;

	ret = iser_create_ib_conn_res((struct iser_conn *)cma_id->context);
	if (ret)
		goto failure;

	memset(&conn_param, 0, sizeof conn_param);
	conn_param.responder_resources = 4;
	conn_param.initiator_depth     = 1;
	conn_param.retry_count	       = 7;
	conn_param.rnr_retry_count     = 6;

	memset(&req_hdr, 0, sizeof(req_hdr));
	req_hdr.flags = (ISER_ZBVA_NOT_SUPPORTED |
			ISER_SEND_W_INV_NOT_SUPPORTED);
	conn_param.private_data		= (void *)&req_hdr;
	conn_param.private_data_len	= sizeof(struct iser_cm_hdr);

	ret = rdma_connect(cma_id, &conn_param);
	if (ret) {
		iser_err("failure connecting: %d\n", ret);
		goto failure;
	}

	return 0;
failure:
	return iser_connect_error(cma_id);
}

static void iser_connected_handler(struct rdma_cm_id *cma_id)
{
	struct iser_conn *ib_conn;

	ib_conn = (struct iser_conn *)cma_id->context;
	ib_conn->state = ISER_CONN_UP;
	wake_up_interruptible(&ib_conn->wait);
}

static int iser_disconnected_handler(struct rdma_cm_id *cma_id)
{
	struct iser_conn *ib_conn;
	int ret;

	ib_conn = (struct iser_conn *)cma_id->context;

	/* getting here when the state is UP means that the conn is being *
	 * terminated asynchronously from the iSCSI layer's perspective.  */
	if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP,
				      ISER_CONN_TERMINATING))
		iscsi_conn_failure(ib_conn->iser_conn->iscsi_conn,
				   ISCSI_ERR_CONN_FAILED);

	/* Complete the termination process if no posts are pending */
	if (ib_conn->post_recv_buf_count == 0 &&
	    (atomic_read(&ib_conn->post_send_buf_count) == 0)) {
		ib_conn->state = ISER_CONN_DOWN;
		wake_up_interruptible(&ib_conn->wait);
	}

	ret = iser_conn_put(ib_conn, 0); /* deref ib conn's cma id */
	return ret;
}

static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
{
	int ret = 0;

	iser_info("event %d status %d conn %p id %p\n",
		  event->event, event->status, cma_id->context, cma_id);

	switch (event->event) {
	case RDMA_CM_EVENT_ADDR_RESOLVED:
		ret = iser_addr_handler(cma_id);
		break;
	case RDMA_CM_EVENT_ROUTE_RESOLVED:
		ret = iser_route_handler(cma_id);
		break;
	case RDMA_CM_EVENT_ESTABLISHED:
		iser_connected_handler(cma_id);
		break;
	case RDMA_CM_EVENT_ADDR_ERROR:
	case RDMA_CM_EVENT_ROUTE_ERROR:
	case RDMA_CM_EVENT_CONNECT_ERROR:
	case RDMA_CM_EVENT_UNREACHABLE:
	case RDMA_CM_EVENT_REJECTED:
		ret = iser_connect_error(cma_id);
		break;
	case RDMA_CM_EVENT_DISCONNECTED:
	case RDMA_CM_EVENT_DEVICE_REMOVAL:
	case RDMA_CM_EVENT_ADDR_CHANGE:
		ret = iser_disconnected_handler(cma_id);
		break;
	default:
		iser_err("Unexpected RDMA CM event (%d)\n", event->event);
		break;
	}
	return ret;
}

void iser_conn_init(struct iser_conn *ib_conn)
{
	ib_conn->state = ISER_CONN_INIT;
	init_waitqueue_head(&ib_conn->wait);
	ib_conn->post_recv_buf_count = 0;
	atomic_set(&ib_conn->post_send_buf_count, 0);
	atomic_set(&ib_conn->refcount, 1); /* ref ib conn allocation */
	INIT_LIST_HEAD(&ib_conn->conn_list);
	spin_lock_init(&ib_conn->lock);
}

 /**
 * starts the process of connecting to the target
 * sleeps until the connection is established or rejected
 */
int iser_connect(struct iser_conn   *ib_conn,
		 struct sockaddr_in *src_addr,
		 struct sockaddr_in *dst_addr,
		 int                 non_blocking)
{
	struct sockaddr *src, *dst;
	int err = 0;

	sprintf(ib_conn->name, "%pI4:%d",
		&dst_addr->sin_addr.s_addr, dst_addr->sin_port);

	/* the device is known only --after-- address resolution */
	ib_conn->device = NULL;

	iser_info("connecting to: %pI4, port 0x%x\n",
		  &dst_addr->sin_addr, dst_addr->sin_port);

	ib_conn->state = ISER_CONN_PENDING;

	iser_conn_get(ib_conn); /* ref ib conn's cma id */
	ib_conn->cma_id = rdma_create_id(iser_cma_handler,
					     (void *)ib_conn,
					     RDMA_PS_TCP, IB_QPT_RC);
	if (IS_ERR(ib_conn->cma_id)) {
		err = PTR_ERR(ib_conn->cma_id);
		iser_err("rdma_create_id failed: %d\n", err);
		goto id_failure;
	}

	src = (struct sockaddr *)src_addr;
	dst = (struct sockaddr *)dst_addr;
	err = rdma_resolve_addr(ib_conn->cma_id, src, dst, 1000);
	if (err) {
		iser_err("rdma_resolve_addr failed: %d\n", err);
		goto addr_failure;
	}

	if (!non_blocking) {
		wait_event_interruptible(ib_conn->wait,
					 (ib_conn->state != ISER_CONN_PENDING));

		if (ib_conn->state != ISER_CONN_UP) {
			err =  -EIO;
			goto connect_failure;
		}
	}

	mutex_lock(&ig.connlist_mutex);
	list_add(&ib_conn->conn_list, &ig.connlist);
	mutex_unlock(&ig.connlist_mutex);
	return 0;

id_failure:
	ib_conn->cma_id = NULL;
addr_failure:
	ib_conn->state = ISER_CONN_DOWN;
	iser_conn_put(ib_conn, 1); /* deref ib conn's cma id */
connect_failure:
	iser_conn_put(ib_conn, 1); /* deref ib conn deallocate */
	return err;
}

/**
 * iser_reg_page_vec - Register physical memory
 *
 * returns: 0 on success, errno code on failure
 */
int iser_reg_page_vec(struct iser_conn     *ib_conn,
		      struct iser_page_vec *page_vec,
		      struct iser_mem_reg  *mem_reg)
{
	struct ib_pool_fmr *mem;
	u64		   io_addr;
	u64		   *page_list;
	int		   status;

	page_list = page_vec->pages;
	io_addr	  = page_list[0];

	mem  = ib_fmr_pool_map_phys(ib_conn->fmr_pool,
				    page_list,
				    page_vec->length,
				    io_addr);

	if (IS_ERR(mem)) {
		status = (int)PTR_ERR(mem);
		iser_err("ib_fmr_pool_map_phys failed: %d\n", status);
		return status;
	}

	mem_reg->lkey  = mem->fmr->lkey;
	mem_reg->rkey  = mem->fmr->rkey;
	mem_reg->len   = page_vec->length * SIZE_4K;
	mem_reg->va    = io_addr;
	mem_reg->is_fmr = 1;
	mem_reg->mem_h = (void *)mem;

	mem_reg->va   += page_vec->offset;
	mem_reg->len   = page_vec->data_size;

	iser_dbg("PHYSICAL Mem.register, [PHYS p_array: 0x%p, sz: %d, "
		 "entry[0]: (0x%08lx,%ld)] -> "
		 "[lkey: 0x%08X mem_h: 0x%p va: 0x%08lX sz: %ld]\n",
		 page_vec, page_vec->length,
		 (unsigned long)page_vec->pages[0],
		 (unsigned long)page_vec->data_size,
		 (unsigned int)mem_reg->lkey, mem_reg->mem_h,
		 (unsigned long)mem_reg->va, (unsigned long)mem_reg->len);
	return 0;
}

/**
 * Unregister (previosuly registered) memory.
 */
void iser_unreg_mem(struct iser_mem_reg *reg)
{
	int ret;

	iser_dbg("PHYSICAL Mem.Unregister mem_h %p\n",reg->mem_h);

	ret = ib_fmr_pool_unmap((struct ib_pool_fmr *)reg->mem_h);
	if (ret)
		iser_err("ib_fmr_pool_unmap failed %d\n", ret);

	reg->mem_h = NULL;
}

int iser_post_recvl(struct iser_conn *ib_conn)
{
	struct ib_recv_wr rx_wr, *rx_wr_failed;
	struct ib_sge	  sge;
	int ib_ret;

	sge.addr   = ib_conn->login_resp_dma;
	sge.length = ISER_RX_LOGIN_SIZE;
	sge.lkey   = ib_conn->device->mr->lkey;

	rx_wr.wr_id   = (unsigned long)ib_conn->login_resp_buf;
	rx_wr.sg_list = &sge;
	rx_wr.num_sge = 1;
	rx_wr.next    = NULL;

	ib_conn->post_recv_buf_count++;
	ib_ret	= ib_post_recv(ib_conn->qp, &rx_wr, &rx_wr_failed);
	if (ib_ret) {
		iser_err("ib_post_recv failed ret=%d\n", ib_ret);
		ib_conn->post_recv_buf_count--;
	}
	return ib_ret;
}

int iser_post_recvm(struct iser_conn *ib_conn, int count)
{
	struct ib_recv_wr *rx_wr, *rx_wr_failed;
	int i, ib_ret;
	unsigned int my_rx_head = ib_conn->rx_desc_head;
	struct iser_rx_desc *rx_desc;

	for (rx_wr = ib_conn->rx_wr, i = 0; i < count; i++, rx_wr++) {
		rx_desc		= &ib_conn->rx_descs[my_rx_head];
		rx_wr->wr_id	= (unsigned long)rx_desc;
		rx_wr->sg_list	= &rx_desc->rx_sg;
		rx_wr->num_sge	= 1;
		rx_wr->next	= rx_wr + 1;
		my_rx_head = (my_rx_head + 1) & (ISER_QP_MAX_RECV_DTOS - 1);
	}

	rx_wr--;
	rx_wr->next = NULL; /* mark end of work requests list */

	ib_conn->post_recv_buf_count += count;
	ib_ret	= ib_post_recv(ib_conn->qp, ib_conn->rx_wr, &rx_wr_failed);
	if (ib_ret) {
		iser_err("ib_post_recv failed ret=%d\n", ib_ret);
		ib_conn->post_recv_buf_count -= count;
	} else
		ib_conn->rx_desc_head = my_rx_head;
	return ib_ret;
}


/**
 * iser_start_send - Initiate a Send DTO operation
 *
 * returns 0 on success, -1 on failure
 */
int iser_post_send(struct iser_conn *ib_conn, struct iser_tx_desc *tx_desc)
{
	int		  ib_ret;
	struct ib_send_wr send_wr, *send_wr_failed;

	ib_dma_sync_single_for_device(ib_conn->device->ib_device,
		tx_desc->dma_addr, ISER_HEADERS_LEN, DMA_TO_DEVICE);

	send_wr.next	   = NULL;
	send_wr.wr_id	   = (unsigned long)tx_desc;
	send_wr.sg_list	   = tx_desc->tx_sg;
	send_wr.num_sge	   = tx_desc->num_sge;
	send_wr.opcode	   = IB_WR_SEND;
	send_wr.send_flags = IB_SEND_SIGNALED;

	atomic_inc(&ib_conn->post_send_buf_count);

	ib_ret = ib_post_send(ib_conn->qp, &send_wr, &send_wr_failed);
	if (ib_ret) {
		iser_err("ib_post_send failed, ret:%d\n", ib_ret);
		atomic_dec(&ib_conn->post_send_buf_count);
	}
	return ib_ret;
}

static void iser_handle_comp_error(struct iser_tx_desc *desc,
				struct iser_conn *ib_conn)
{
	if (desc && desc->type == ISCSI_TX_DATAOUT)
		kmem_cache_free(ig.desc_cache, desc);

	if (ib_conn->post_recv_buf_count == 0 &&
	    atomic_read(&ib_conn->post_send_buf_count) == 0) {
		/* getting here when the state is UP means that the conn is *
		 * being terminated asynchronously from the iSCSI layer's   *
		 * perspective.                                             */
		if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP,
		    ISER_CONN_TERMINATING))
			iscsi_conn_failure(ib_conn->iser_conn->iscsi_conn,
					   ISCSI_ERR_CONN_FAILED);

		/* no more non completed posts to the QP, complete the
		 * termination process w.o worrying on disconnect event */
		ib_conn->state = ISER_CONN_DOWN;
		wake_up_interruptible(&ib_conn->wait);
	}
}

static int iser_drain_tx_cq(struct iser_device  *device, int cq_index)
{
	struct ib_cq  *cq = device->tx_cq[cq_index];
	struct ib_wc  wc;
	struct iser_tx_desc *tx_desc;
	struct iser_conn *ib_conn;
	int completed_tx = 0;

	while (ib_poll_cq(cq, 1, &wc) == 1) {
		tx_desc	= (struct iser_tx_desc *) (unsigned long) wc.wr_id;
		ib_conn = wc.qp->qp_context;
		if (wc.status == IB_WC_SUCCESS) {
			if (wc.opcode == IB_WC_SEND)
				iser_snd_completion(tx_desc, ib_conn);
			else
				iser_err("expected opcode %d got %d\n",
					IB_WC_SEND, wc.opcode);
		} else {
			iser_err("tx id %llx status %d vend_err %x\n",
				wc.wr_id, wc.status, wc.vendor_err);
			atomic_dec(&ib_conn->post_send_buf_count);
			iser_handle_comp_error(tx_desc, ib_conn);
		}
		completed_tx++;
	}
	return completed_tx;
}


static void iser_cq_tasklet_fn(unsigned long data)
{
	struct iser_cq_desc *cq_desc = (struct iser_cq_desc *)data;
	struct iser_device  *device = cq_desc->device;
	int cq_index = cq_desc->cq_index;
	struct ib_cq	     *cq = device->rx_cq[cq_index];
	 struct ib_wc	     wc;
	 struct iser_rx_desc *desc;
	 unsigned long	     xfer_len;
	struct iser_conn *ib_conn;
	int completed_tx, completed_rx;
	completed_tx = completed_rx = 0;

	while (ib_poll_cq(cq, 1, &wc) == 1) {
		desc	 = (struct iser_rx_desc *) (unsigned long) wc.wr_id;
		BUG_ON(desc == NULL);
		ib_conn = wc.qp->qp_context;
		if (wc.status == IB_WC_SUCCESS) {
			if (wc.opcode == IB_WC_RECV) {
				xfer_len = (unsigned long)wc.byte_len;
				iser_rcv_completion(desc, xfer_len, ib_conn);
			} else
				iser_err("expected opcode %d got %d\n",
					IB_WC_RECV, wc.opcode);
		} else {
			if (wc.status != IB_WC_WR_FLUSH_ERR)
				iser_err("rx id %llx status %d vend_err %x\n",
					wc.wr_id, wc.status, wc.vendor_err);
			ib_conn->post_recv_buf_count--;
			iser_handle_comp_error(NULL, ib_conn);
		}
		completed_rx++;
		if (!(completed_rx & 63))
			completed_tx += iser_drain_tx_cq(device, cq_index);
	}
	/* #warning "it is assumed here that arming CQ only once its empty" *
	 * " would not cause interrupts to be missed"                       */
	ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);

	completed_tx += iser_drain_tx_cq(device, cq_index);
	iser_dbg("got %d rx %d tx completions\n", completed_rx, completed_tx);
}

static void iser_cq_callback(struct ib_cq *cq, void *cq_context)
{
	struct iser_cq_desc *cq_desc = (struct iser_cq_desc *)cq_context;
	struct iser_device  *device = cq_desc->device;
	int cq_index = cq_desc->cq_index;

	tasklet_schedule(&device->cq_tasklet[cq_index]);
}