ib_srpt.c 91.7 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 34 35 36 37 38 39 40 41 42 43
/*
 * Copyright (c) 2006 - 2009 Mellanox Technology Inc.  All rights reserved.
 * Copyright (C) 2008 - 2011 Bart Van Assche <bvanassche@acm.org>.
 *
 * 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/init.h>
#include <linux/slab.h>
#include <linux/err.h>
#include <linux/ctype.h>
#include <linux/kthread.h>
#include <linux/string.h>
#include <linux/delay.h>
#include <linux/atomic.h>
44
#include <scsi/scsi_proto.h>
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
#include <scsi/scsi_tcq.h>
#include <target/target_core_base.h>
#include <target/target_core_fabric.h>
#include "ib_srpt.h"

/* Name of this kernel module. */
#define DRV_NAME		"ib_srpt"
#define DRV_VERSION		"2.0.0"
#define DRV_RELDATE		"2011-02-14"

#define SRPT_ID_STRING	"Linux SRP target"

#undef pr_fmt
#define pr_fmt(fmt) DRV_NAME " " fmt

MODULE_AUTHOR("Vu Pham and Bart Van Assche");
MODULE_DESCRIPTION("InfiniBand SCSI RDMA Protocol target "
		   "v" DRV_VERSION " (" DRV_RELDATE ")");
MODULE_LICENSE("Dual BSD/GPL");

/*
 * Global Variables
 */

static u64 srpt_service_guid;
70 71
static DEFINE_SPINLOCK(srpt_dev_lock);	/* Protects srpt_dev_list. */
static LIST_HEAD(srpt_dev_list);	/* List of srpt_device structures. */
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95

static unsigned srp_max_req_size = DEFAULT_MAX_REQ_SIZE;
module_param(srp_max_req_size, int, 0444);
MODULE_PARM_DESC(srp_max_req_size,
		 "Maximum size of SRP request messages in bytes.");

static int srpt_srq_size = DEFAULT_SRPT_SRQ_SIZE;
module_param(srpt_srq_size, int, 0444);
MODULE_PARM_DESC(srpt_srq_size,
		 "Shared receive queue (SRQ) size.");

static int srpt_get_u64_x(char *buffer, struct kernel_param *kp)
{
	return sprintf(buffer, "0x%016llx", *(u64 *)kp->arg);
}
module_param_call(srpt_service_guid, NULL, srpt_get_u64_x, &srpt_service_guid,
		  0444);
MODULE_PARM_DESC(srpt_service_guid,
		 "Using this value for ioc_guid, id_ext, and cm_listen_id"
		 " instead of using the node_guid of the first HCA.");

static struct ib_client srpt_client;
static void srpt_release_channel(struct srpt_rdma_ch *ch);
static int srpt_queue_status(struct se_cmd *cmd);
96 97
static void srpt_recv_done(struct ib_cq *cq, struct ib_wc *wc);
static void srpt_send_done(struct ib_cq *cq, struct ib_wc *wc);
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

/**
 * opposite_dma_dir() - Swap DMA_TO_DEVICE and DMA_FROM_DEVICE.
 */
static inline
enum dma_data_direction opposite_dma_dir(enum dma_data_direction dir)
{
	switch (dir) {
	case DMA_TO_DEVICE:	return DMA_FROM_DEVICE;
	case DMA_FROM_DEVICE:	return DMA_TO_DEVICE;
	default:		return dir;
	}
}

static enum rdma_ch_state srpt_get_ch_state(struct srpt_rdma_ch *ch)
{
	unsigned long flags;
	enum rdma_ch_state state;

	spin_lock_irqsave(&ch->spinlock, flags);
	state = ch->state;
	spin_unlock_irqrestore(&ch->spinlock, flags);
	return state;
}

static enum rdma_ch_state
srpt_set_ch_state(struct srpt_rdma_ch *ch, enum rdma_ch_state new_state)
{
	unsigned long flags;
	enum rdma_ch_state prev;

	spin_lock_irqsave(&ch->spinlock, flags);
	prev = ch->state;
	ch->state = new_state;
	spin_unlock_irqrestore(&ch->spinlock, flags);
	return prev;
}

/**
 * srpt_test_and_set_ch_state() - Test and set the channel state.
 *
 * Returns true if and only if the channel state has been set to the new state.
 */
static bool
srpt_test_and_set_ch_state(struct srpt_rdma_ch *ch, enum rdma_ch_state old,
			   enum rdma_ch_state new)
{
	unsigned long flags;
	enum rdma_ch_state prev;

	spin_lock_irqsave(&ch->spinlock, flags);
	prev = ch->state;
	if (prev == old)
		ch->state = new;
	spin_unlock_irqrestore(&ch->spinlock, flags);
	return prev == old;
}

/**
 * srpt_event_handler() - Asynchronous IB event callback function.
 *
 * Callback function called by the InfiniBand core when an asynchronous IB
 * event occurs. This callback may occur in interrupt context. See also
 * section 11.5.2, Set Asynchronous Event Handler in the InfiniBand
 * Architecture Specification.
 */
static void srpt_event_handler(struct ib_event_handler *handler,
			       struct ib_event *event)
{
	struct srpt_device *sdev;
	struct srpt_port *sport;

	sdev = ib_get_client_data(event->device, &srpt_client);
	if (!sdev || sdev->device != event->device)
		return;

	pr_debug("ASYNC event= %d on device= %s\n", event->event,
175
		 sdev->device->name);
176 177 178 179 180 181 182 183 184 185 186 187 188 189

	switch (event->event) {
	case IB_EVENT_PORT_ERR:
		if (event->element.port_num <= sdev->device->phys_port_cnt) {
			sport = &sdev->port[event->element.port_num - 1];
			sport->lid = 0;
			sport->sm_lid = 0;
		}
		break;
	case IB_EVENT_PORT_ACTIVE:
	case IB_EVENT_LID_CHANGE:
	case IB_EVENT_PKEY_CHANGE:
	case IB_EVENT_SM_CHANGE:
	case IB_EVENT_CLIENT_REREGISTER:
D
Doug Ledford 已提交
190
	case IB_EVENT_GID_CHANGE:
191 192 193 194 195 196 197 198
		/* Refresh port data asynchronously. */
		if (event->element.port_num <= sdev->device->phys_port_cnt) {
			sport = &sdev->port[event->element.port_num - 1];
			if (!sport->lid && !sport->sm_lid)
				schedule_work(&sport->work);
		}
		break;
	default:
199
		pr_err("received unrecognized IB event %d\n",
200 201 202 203 204 205 206 207 208 209
		       event->event);
		break;
	}
}

/**
 * srpt_srq_event() - SRQ event callback function.
 */
static void srpt_srq_event(struct ib_event *event, void *ctx)
{
210
	pr_info("SRQ event %d\n", event->event);
211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233
}

/**
 * srpt_qp_event() - QP event callback function.
 */
static void srpt_qp_event(struct ib_event *event, struct srpt_rdma_ch *ch)
{
	pr_debug("QP event %d on cm_id=%p sess_name=%s state=%d\n",
		 event->event, ch->cm_id, ch->sess_name, srpt_get_ch_state(ch));

	switch (event->event) {
	case IB_EVENT_COMM_EST:
		ib_cm_notify(ch->cm_id, event->event);
		break;
	case IB_EVENT_QP_LAST_WQE_REACHED:
		if (srpt_test_and_set_ch_state(ch, CH_DRAINING,
					       CH_RELEASING))
			srpt_release_channel(ch);
		else
			pr_debug("%s: state %d - ignored LAST_WQE.\n",
				 ch->sess_name, srpt_get_ch_state(ch));
		break;
	default:
234
		pr_err("received unrecognized IB QP event %d\n", event->event);
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
		break;
	}
}

/**
 * srpt_set_ioc() - Helper function for initializing an IOUnitInfo structure.
 *
 * @slot: one-based slot number.
 * @value: four-bit value.
 *
 * Copies the lowest four bits of value in element slot of the array of four
 * bit elements called c_list (controller list). The index slot is one-based.
 */
static void srpt_set_ioc(u8 *c_list, u32 slot, u8 value)
{
	u16 id;
	u8 tmp;

	id = (slot - 1) / 2;
	if (slot & 0x1) {
		tmp = c_list[id] & 0xf;
		c_list[id] = (value << 4) | tmp;
	} else {
		tmp = c_list[id] & 0xf0;
		c_list[id] = (value & 0xf) | tmp;
	}
}

/**
 * srpt_get_class_port_info() - Copy ClassPortInfo to a management datagram.
 *
 * See also section 16.3.3.1 ClassPortInfo in the InfiniBand Architecture
 * Specification.
 */
static void srpt_get_class_port_info(struct ib_dm_mad *mad)
{
	struct ib_class_port_info *cif;

	cif = (struct ib_class_port_info *)mad->data;
274
	memset(cif, 0, sizeof(*cif));
275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294
	cif->base_version = 1;
	cif->class_version = 1;
	cif->resp_time_value = 20;

	mad->mad_hdr.status = 0;
}

/**
 * srpt_get_iou() - Write IOUnitInfo to a management datagram.
 *
 * See also section 16.3.3.3 IOUnitInfo in the InfiniBand Architecture
 * Specification. See also section B.7, table B.6 in the SRP r16a document.
 */
static void srpt_get_iou(struct ib_dm_mad *mad)
{
	struct ib_dm_iou_info *ioui;
	u8 slot;
	int i;

	ioui = (struct ib_dm_iou_info *)mad->data;
295
	ioui->change_id = cpu_to_be16(1);
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
	ioui->max_controllers = 16;

	/* set present for slot 1 and empty for the rest */
	srpt_set_ioc(ioui->controller_list, 1, 1);
	for (i = 1, slot = 2; i < 16; i++, slot++)
		srpt_set_ioc(ioui->controller_list, slot, 0);

	mad->mad_hdr.status = 0;
}

/**
 * srpt_get_ioc() - Write IOControllerprofile to a management datagram.
 *
 * See also section 16.3.3.4 IOControllerProfile in the InfiniBand
 * Architecture Specification. See also section B.7, table B.7 in the SRP
 * r16a document.
 */
static void srpt_get_ioc(struct srpt_port *sport, u32 slot,
			 struct ib_dm_mad *mad)
{
	struct srpt_device *sdev = sport->sdev;
	struct ib_dm_ioc_profile *iocp;

	iocp = (struct ib_dm_ioc_profile *)mad->data;

	if (!slot || slot > 16) {
		mad->mad_hdr.status
323
			= cpu_to_be16(DM_MAD_STATUS_INVALID_FIELD);
324 325 326 327 328
		return;
	}

	if (slot > 2) {
		mad->mad_hdr.status
329
			= cpu_to_be16(DM_MAD_STATUS_NO_IOC);
330 331 332
		return;
	}

333
	memset(iocp, 0, sizeof(*iocp));
334 335
	strcpy(iocp->id_string, SRPT_ID_STRING);
	iocp->guid = cpu_to_be64(srpt_service_guid);
336 337 338 339
	iocp->vendor_id = cpu_to_be32(sdev->device->attrs.vendor_id);
	iocp->device_id = cpu_to_be32(sdev->device->attrs.vendor_part_id);
	iocp->device_version = cpu_to_be16(sdev->device->attrs.hw_ver);
	iocp->subsys_vendor_id = cpu_to_be32(sdev->device->attrs.vendor_id);
340
	iocp->subsys_device_id = 0x0;
341 342 343 344
	iocp->io_class = cpu_to_be16(SRP_REV16A_IB_IO_CLASS);
	iocp->io_subclass = cpu_to_be16(SRP_IO_SUBCLASS);
	iocp->protocol = cpu_to_be16(SRP_PROTOCOL);
	iocp->protocol_version = cpu_to_be16(SRP_PROTOCOL_VERSION);
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
	iocp->send_queue_depth = cpu_to_be16(sdev->srq_size);
	iocp->rdma_read_depth = 4;
	iocp->send_size = cpu_to_be32(srp_max_req_size);
	iocp->rdma_size = cpu_to_be32(min(sport->port_attrib.srp_max_rdma_size,
					  1U << 24));
	iocp->num_svc_entries = 1;
	iocp->op_cap_mask = SRP_SEND_TO_IOC | SRP_SEND_FROM_IOC |
		SRP_RDMA_READ_FROM_IOC | SRP_RDMA_WRITE_FROM_IOC;

	mad->mad_hdr.status = 0;
}

/**
 * srpt_get_svc_entries() - Write ServiceEntries to a management datagram.
 *
 * See also section 16.3.3.5 ServiceEntries in the InfiniBand Architecture
 * Specification. See also section B.7, table B.8 in the SRP r16a document.
 */
static void srpt_get_svc_entries(u64 ioc_guid,
				 u16 slot, u8 hi, u8 lo, struct ib_dm_mad *mad)
{
	struct ib_dm_svc_entries *svc_entries;

	WARN_ON(!ioc_guid);

	if (!slot || slot > 16) {
		mad->mad_hdr.status
372
			= cpu_to_be16(DM_MAD_STATUS_INVALID_FIELD);
373 374 375 376 377
		return;
	}

	if (slot > 2 || lo > hi || hi > 1) {
		mad->mad_hdr.status
378
			= cpu_to_be16(DM_MAD_STATUS_NO_IOC);
379 380 381 382
		return;
	}

	svc_entries = (struct ib_dm_svc_entries *)mad->data;
383
	memset(svc_entries, 0, sizeof(*svc_entries));
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
	svc_entries->service_entries[0].id = cpu_to_be64(ioc_guid);
	snprintf(svc_entries->service_entries[0].name,
		 sizeof(svc_entries->service_entries[0].name),
		 "%s%016llx",
		 SRP_SERVICE_NAME_PREFIX,
		 ioc_guid);

	mad->mad_hdr.status = 0;
}

/**
 * srpt_mgmt_method_get() - Process a received management datagram.
 * @sp:      source port through which the MAD has been received.
 * @rq_mad:  received MAD.
 * @rsp_mad: response MAD.
 */
static void srpt_mgmt_method_get(struct srpt_port *sp, struct ib_mad *rq_mad,
				 struct ib_dm_mad *rsp_mad)
{
	u16 attr_id;
	u32 slot;
	u8 hi, lo;

	attr_id = be16_to_cpu(rq_mad->mad_hdr.attr_id);
	switch (attr_id) {
	case DM_ATTR_CLASS_PORT_INFO:
		srpt_get_class_port_info(rsp_mad);
		break;
	case DM_ATTR_IOU_INFO:
		srpt_get_iou(rsp_mad);
		break;
	case DM_ATTR_IOC_PROFILE:
		slot = be32_to_cpu(rq_mad->mad_hdr.attr_mod);
		srpt_get_ioc(sp, slot, rsp_mad);
		break;
	case DM_ATTR_SVC_ENTRIES:
		slot = be32_to_cpu(rq_mad->mad_hdr.attr_mod);
		hi = (u8) ((slot >> 8) & 0xff);
		lo = (u8) (slot & 0xff);
		slot = (u16) ((slot >> 16) & 0xffff);
		srpt_get_svc_entries(srpt_service_guid,
				     slot, hi, lo, rsp_mad);
		break;
	default:
		rsp_mad->mad_hdr.status =
429
		    cpu_to_be16(DM_MAD_STATUS_UNSUP_METHOD_ATTR);
430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447
		break;
	}
}

/**
 * srpt_mad_send_handler() - Post MAD-send callback function.
 */
static void srpt_mad_send_handler(struct ib_mad_agent *mad_agent,
				  struct ib_mad_send_wc *mad_wc)
{
	ib_destroy_ah(mad_wc->send_buf->ah);
	ib_free_send_mad(mad_wc->send_buf);
}

/**
 * srpt_mad_recv_handler() - MAD reception callback function.
 */
static void srpt_mad_recv_handler(struct ib_mad_agent *mad_agent,
448
				  struct ib_mad_send_buf *send_buf,
449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468
				  struct ib_mad_recv_wc *mad_wc)
{
	struct srpt_port *sport = (struct srpt_port *)mad_agent->context;
	struct ib_ah *ah;
	struct ib_mad_send_buf *rsp;
	struct ib_dm_mad *dm_mad;

	if (!mad_wc || !mad_wc->recv_buf.mad)
		return;

	ah = ib_create_ah_from_wc(mad_agent->qp->pd, mad_wc->wc,
				  mad_wc->recv_buf.grh, mad_agent->port_num);
	if (IS_ERR(ah))
		goto err;

	BUILD_BUG_ON(offsetof(struct ib_dm_mad, data) != IB_MGMT_DEVICE_HDR);

	rsp = ib_create_send_mad(mad_agent, mad_wc->wc->src_qp,
				 mad_wc->wc->pkey_index, 0,
				 IB_MGMT_DEVICE_HDR, IB_MGMT_DEVICE_DATA,
469 470
				 GFP_KERNEL,
				 IB_MGMT_BASE_VERSION);
471 472 473 474 475 476
	if (IS_ERR(rsp))
		goto err_rsp;

	rsp->ah = ah;

	dm_mad = rsp->mad;
477
	memcpy(dm_mad, mad_wc->recv_buf.mad, sizeof(*dm_mad));
478 479 480 481 482 483 484 485 486
	dm_mad->mad_hdr.method = IB_MGMT_METHOD_GET_RESP;
	dm_mad->mad_hdr.status = 0;

	switch (mad_wc->recv_buf.mad->mad_hdr.method) {
	case IB_MGMT_METHOD_GET:
		srpt_mgmt_method_get(sport, mad_wc->recv_buf.mad, dm_mad);
		break;
	case IB_MGMT_METHOD_SET:
		dm_mad->mad_hdr.status =
487
		    cpu_to_be16(DM_MAD_STATUS_UNSUP_METHOD_ATTR);
488 489 490
		break;
	default:
		dm_mad->mad_hdr.status =
491
		    cpu_to_be16(DM_MAD_STATUS_UNSUP_METHOD);
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
		break;
	}

	if (!ib_post_send_mad(rsp, NULL)) {
		ib_free_recv_mad(mad_wc);
		/* will destroy_ah & free_send_mad in send completion */
		return;
	}

	ib_free_send_mad(rsp);

err_rsp:
	ib_destroy_ah(ah);
err:
	ib_free_recv_mad(mad_wc);
}

/**
 * srpt_refresh_port() - Configure a HCA port.
 *
 * Enable InfiniBand management datagram processing, update the cached sm_lid,
 * lid and gid values, and register a callback function for processing MADs
 * on the specified port.
 *
 * Note: It is safe to call this function more than once for the same port.
 */
static int srpt_refresh_port(struct srpt_port *sport)
{
	struct ib_mad_reg_req reg_req;
	struct ib_port_modify port_modify;
	struct ib_port_attr port_attr;
	int ret;

525
	memset(&port_modify, 0, sizeof(port_modify));
526 527 528 529 530 531 532 533 534 535 536 537 538 539
	port_modify.set_port_cap_mask = IB_PORT_DEVICE_MGMT_SUP;
	port_modify.clr_port_cap_mask = 0;

	ret = ib_modify_port(sport->sdev->device, sport->port, 0, &port_modify);
	if (ret)
		goto err_mod_port;

	ret = ib_query_port(sport->sdev->device, sport->port, &port_attr);
	if (ret)
		goto err_query_port;

	sport->sm_lid = port_attr.sm_lid;
	sport->lid = port_attr.lid;

540 541
	ret = ib_query_gid(sport->sdev->device, sport->port, 0, &sport->gid,
			   NULL);
542 543 544 545
	if (ret)
		goto err_query_port;

	if (!sport->mad_agent) {
546
		memset(&reg_req, 0, sizeof(reg_req));
547 548 549 550 551 552 553 554 555 556 557
		reg_req.mgmt_class = IB_MGMT_CLASS_DEVICE_MGMT;
		reg_req.mgmt_class_version = IB_MGMT_BASE_VERSION;
		set_bit(IB_MGMT_METHOD_GET, reg_req.method_mask);
		set_bit(IB_MGMT_METHOD_SET, reg_req.method_mask);

		sport->mad_agent = ib_register_mad_agent(sport->sdev->device,
							 sport->port,
							 IB_QPT_GSI,
							 &reg_req, 0,
							 srpt_mad_send_handler,
							 srpt_mad_recv_handler,
558
							 sport, 0);
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
		if (IS_ERR(sport->mad_agent)) {
			ret = PTR_ERR(sport->mad_agent);
			sport->mad_agent = NULL;
			goto err_query_port;
		}
	}

	return 0;

err_query_port:

	port_modify.set_port_cap_mask = 0;
	port_modify.clr_port_cap_mask = IB_PORT_DEVICE_MGMT_SUP;
	ib_modify_port(sport->sdev->device, sport->port, 0, &port_modify);

err_mod_port:

	return ret;
}

/**
 * srpt_unregister_mad_agent() - Unregister MAD callback functions.
 *
 * Note: It is safe to call this function more than once for the same device.
 */
static void srpt_unregister_mad_agent(struct srpt_device *sdev)
{
	struct ib_port_modify port_modify = {
		.clr_port_cap_mask = IB_PORT_DEVICE_MGMT_SUP,
	};
	struct srpt_port *sport;
	int i;

	for (i = 1; i <= sdev->device->phys_port_cnt; i++) {
		sport = &sdev->port[i - 1];
		WARN_ON(sport->port != i);
		if (ib_modify_port(sdev->device, i, 0, &port_modify) < 0)
596
			pr_err("disabling MAD processing failed.\n");
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
		if (sport->mad_agent) {
			ib_unregister_mad_agent(sport->mad_agent);
			sport->mad_agent = NULL;
		}
	}
}

/**
 * srpt_alloc_ioctx() - Allocate an SRPT I/O context structure.
 */
static struct srpt_ioctx *srpt_alloc_ioctx(struct srpt_device *sdev,
					   int ioctx_size, int dma_size,
					   enum dma_data_direction dir)
{
	struct srpt_ioctx *ioctx;

	ioctx = kmalloc(ioctx_size, GFP_KERNEL);
	if (!ioctx)
		goto err;

	ioctx->buf = kmalloc(dma_size, GFP_KERNEL);
	if (!ioctx->buf)
		goto err_free_ioctx;

	ioctx->dma = ib_dma_map_single(sdev->device, ioctx->buf, dma_size, dir);
	if (ib_dma_mapping_error(sdev->device, ioctx->dma))
		goto err_free_buf;

	return ioctx;

err_free_buf:
	kfree(ioctx->buf);
err_free_ioctx:
	kfree(ioctx);
err:
	return NULL;
}

/**
 * srpt_free_ioctx() - Free an SRPT I/O context structure.
 */
static void srpt_free_ioctx(struct srpt_device *sdev, struct srpt_ioctx *ioctx,
			    int dma_size, enum dma_data_direction dir)
{
	if (!ioctx)
		return;

	ib_dma_unmap_single(sdev->device, ioctx->dma, dma_size, dir);
	kfree(ioctx->buf);
	kfree(ioctx);
}

/**
 * srpt_alloc_ioctx_ring() - Allocate a ring of SRPT I/O context structures.
 * @sdev:       Device to allocate the I/O context ring for.
 * @ring_size:  Number of elements in the I/O context ring.
 * @ioctx_size: I/O context size.
 * @dma_size:   DMA buffer size.
 * @dir:        DMA data direction.
 */
static struct srpt_ioctx **srpt_alloc_ioctx_ring(struct srpt_device *sdev,
				int ring_size, int ioctx_size,
				int dma_size, enum dma_data_direction dir)
{
	struct srpt_ioctx **ring;
	int i;

	WARN_ON(ioctx_size != sizeof(struct srpt_recv_ioctx)
		&& ioctx_size != sizeof(struct srpt_send_ioctx));

	ring = kmalloc(ring_size * sizeof(ring[0]), GFP_KERNEL);
	if (!ring)
		goto out;
	for (i = 0; i < ring_size; ++i) {
		ring[i] = srpt_alloc_ioctx(sdev, ioctx_size, dma_size, dir);
		if (!ring[i])
			goto err;
		ring[i]->index = i;
	}
	goto out;

err:
	while (--i >= 0)
		srpt_free_ioctx(sdev, ring[i], dma_size, dir);
	kfree(ring);
682
	ring = NULL;
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
out:
	return ring;
}

/**
 * srpt_free_ioctx_ring() - Free the ring of SRPT I/O context structures.
 */
static void srpt_free_ioctx_ring(struct srpt_ioctx **ioctx_ring,
				 struct srpt_device *sdev, int ring_size,
				 int dma_size, enum dma_data_direction dir)
{
	int i;

	for (i = 0; i < ring_size; ++i)
		srpt_free_ioctx(sdev, ioctx_ring[i], dma_size, dir);
	kfree(ioctx_ring);
}

/**
 * srpt_get_cmd_state() - Get the state of a SCSI command.
 */
static enum srpt_command_state srpt_get_cmd_state(struct srpt_send_ioctx *ioctx)
{
	enum srpt_command_state state;
	unsigned long flags;

	BUG_ON(!ioctx);

	spin_lock_irqsave(&ioctx->spinlock, flags);
	state = ioctx->state;
	spin_unlock_irqrestore(&ioctx->spinlock, flags);
	return state;
}

/**
 * srpt_set_cmd_state() - Set the state of a SCSI command.
 *
 * Does not modify the state of aborted commands. Returns the previous command
 * state.
 */
static enum srpt_command_state srpt_set_cmd_state(struct srpt_send_ioctx *ioctx,
						  enum srpt_command_state new)
{
	enum srpt_command_state previous;
	unsigned long flags;

	BUG_ON(!ioctx);

	spin_lock_irqsave(&ioctx->spinlock, flags);
	previous = ioctx->state;
	if (previous != SRPT_STATE_DONE)
		ioctx->state = new;
	spin_unlock_irqrestore(&ioctx->spinlock, flags);

	return previous;
}

/**
 * srpt_test_and_set_cmd_state() - Test and set the state of a command.
 *
 * Returns true if and only if the previous command state was equal to 'old'.
 */
static bool srpt_test_and_set_cmd_state(struct srpt_send_ioctx *ioctx,
					enum srpt_command_state old,
					enum srpt_command_state new)
{
	enum srpt_command_state previous;
	unsigned long flags;

	WARN_ON(!ioctx);
	WARN_ON(old == SRPT_STATE_DONE);
	WARN_ON(new == SRPT_STATE_NEW);

	spin_lock_irqsave(&ioctx->spinlock, flags);
	previous = ioctx->state;
	if (previous == old)
		ioctx->state = new;
	spin_unlock_irqrestore(&ioctx->spinlock, flags);
	return previous == old;
}

/**
 * srpt_post_recv() - Post an IB receive request.
 */
static int srpt_post_recv(struct srpt_device *sdev,
			  struct srpt_recv_ioctx *ioctx)
{
	struct ib_sge list;
	struct ib_recv_wr wr, *bad_wr;

	BUG_ON(!sdev);
	list.addr = ioctx->ioctx.dma;
	list.length = srp_max_req_size;
776
	list.lkey = sdev->pd->local_dma_lkey;
777

778 779
	ioctx->ioctx.cqe.done = srpt_recv_done;
	wr.wr_cqe = &ioctx->ioctx.cqe;
780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803
	wr.next = NULL;
	wr.sg_list = &list;
	wr.num_sge = 1;

	return ib_post_srq_recv(sdev->srq, &wr, &bad_wr);
}

/**
 * srpt_post_send() - Post an IB send request.
 *
 * Returns zero upon success and a non-zero value upon failure.
 */
static int srpt_post_send(struct srpt_rdma_ch *ch,
			  struct srpt_send_ioctx *ioctx, int len)
{
	struct ib_sge list;
	struct ib_send_wr wr, *bad_wr;
	struct srpt_device *sdev = ch->sport->sdev;
	int ret;

	atomic_inc(&ch->req_lim);

	ret = -ENOMEM;
	if (unlikely(atomic_dec_return(&ch->sq_wr_avail) < 0)) {
804
		pr_warn("IB send queue full (needed 1)\n");
805 806 807 808 809 810 811 812
		goto out;
	}

	ib_dma_sync_single_for_device(sdev->device, ioctx->ioctx.dma, len,
				      DMA_TO_DEVICE);

	list.addr = ioctx->ioctx.dma;
	list.length = len;
813
	list.lkey = sdev->pd->local_dma_lkey;
814

815
	ioctx->ioctx.cqe.done = srpt_send_done;
816
	wr.next = NULL;
817
	wr.wr_cqe = &ioctx->ioctx.cqe;
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
	wr.sg_list = &list;
	wr.num_sge = 1;
	wr.opcode = IB_WR_SEND;
	wr.send_flags = IB_SEND_SIGNALED;

	ret = ib_post_send(ch->qp, &wr, &bad_wr);

out:
	if (ret < 0) {
		atomic_inc(&ch->sq_wr_avail);
		atomic_dec(&ch->req_lim);
	}
	return ret;
}

/**
 * srpt_get_desc_tbl() - Parse the data descriptors of an SRP_CMD request.
 * @ioctx: Pointer to the I/O context associated with the request.
 * @srp_cmd: Pointer to the SRP_CMD request data.
 * @dir: Pointer to the variable to which the transfer direction will be
 *   written.
 * @data_len: Pointer to the variable to which the total data length of all
 *   descriptors in the SRP_CMD request will be written.
 *
 * This function initializes ioctx->nrbuf and ioctx->r_bufs.
 *
 * Returns -EINVAL when the SRP_CMD request contains inconsistent descriptors;
 * -ENOMEM when memory allocation fails and zero upon success.
 */
static int srpt_get_desc_tbl(struct srpt_send_ioctx *ioctx,
			     struct srp_cmd *srp_cmd,
			     enum dma_data_direction *dir, u64 *data_len)
{
	struct srp_indirect_buf *idb;
	struct srp_direct_buf *db;
	unsigned add_cdb_offset;
	int ret;

	/*
	 * The pointer computations below will only be compiled correctly
	 * if srp_cmd::add_data is declared as s8*, u8*, s8[] or u8[], so check
	 * whether srp_cmd::add_data has been declared as a byte pointer.
	 */
	BUILD_BUG_ON(!__same_type(srp_cmd->add_data[0], (s8)0)
		     && !__same_type(srp_cmd->add_data[0], (u8)0));

	BUG_ON(!dir);
	BUG_ON(!data_len);

	ret = 0;
	*data_len = 0;

	/*
	 * The lower four bits of the buffer format field contain the DATA-IN
	 * buffer descriptor format, and the highest four bits contain the
	 * DATA-OUT buffer descriptor format.
	 */
	*dir = DMA_NONE;
	if (srp_cmd->buf_fmt & 0xf)
		/* DATA-IN: transfer data from target to initiator (read). */
		*dir = DMA_FROM_DEVICE;
	else if (srp_cmd->buf_fmt >> 4)
		/* DATA-OUT: transfer data from initiator to target (write). */
		*dir = DMA_TO_DEVICE;

	/*
	 * According to the SRP spec, the lower two bits of the 'ADDITIONAL
	 * CDB LENGTH' field are reserved and the size in bytes of this field
	 * is four times the value specified in bits 3..7. Hence the "& ~3".
	 */
	add_cdb_offset = srp_cmd->add_cdb_len & ~3;
	if (((srp_cmd->buf_fmt & 0xf) == SRP_DATA_DESC_DIRECT) ||
	    ((srp_cmd->buf_fmt >> 4) == SRP_DATA_DESC_DIRECT)) {
		ioctx->n_rbuf = 1;
		ioctx->rbufs = &ioctx->single_rbuf;

		db = (struct srp_direct_buf *)(srp_cmd->add_data
					       + add_cdb_offset);
896
		memcpy(ioctx->rbufs, db, sizeof(*db));
897 898 899 900 901 902
		*data_len = be32_to_cpu(db->len);
	} else if (((srp_cmd->buf_fmt & 0xf) == SRP_DATA_DESC_INDIRECT) ||
		   ((srp_cmd->buf_fmt >> 4) == SRP_DATA_DESC_INDIRECT)) {
		idb = (struct srp_indirect_buf *)(srp_cmd->add_data
						  + add_cdb_offset);

903
		ioctx->n_rbuf = be32_to_cpu(idb->table_desc.len) / sizeof(*db);
904 905 906

		if (ioctx->n_rbuf >
		    (srp_cmd->data_out_desc_cnt + srp_cmd->data_in_desc_cnt)) {
907
			pr_err("received unsupported SRP_CMD request"
908 909 910 911 912 913 914 915 916 917 918 919 920 921
			       " type (%u out + %u in != %u / %zu)\n",
			       srp_cmd->data_out_desc_cnt,
			       srp_cmd->data_in_desc_cnt,
			       be32_to_cpu(idb->table_desc.len),
			       sizeof(*db));
			ioctx->n_rbuf = 0;
			ret = -EINVAL;
			goto out;
		}

		if (ioctx->n_rbuf == 1)
			ioctx->rbufs = &ioctx->single_rbuf;
		else {
			ioctx->rbufs =
922
				kmalloc(ioctx->n_rbuf * sizeof(*db), GFP_ATOMIC);
923 924 925 926 927 928 929 930
			if (!ioctx->rbufs) {
				ioctx->n_rbuf = 0;
				ret = -ENOMEM;
				goto out;
			}
		}

		db = idb->desc_list;
931
		memcpy(ioctx->rbufs, db, ioctx->n_rbuf * sizeof(*db));
932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948
		*data_len = be32_to_cpu(idb->len);
	}
out:
	return ret;
}

/**
 * srpt_init_ch_qp() - Initialize queue pair attributes.
 *
 * Initialized the attributes of queue pair 'qp' by allowing local write,
 * remote read and remote write. Also transitions 'qp' to state IB_QPS_INIT.
 */
static int srpt_init_ch_qp(struct srpt_rdma_ch *ch, struct ib_qp *qp)
{
	struct ib_qp_attr *attr;
	int ret;

949
	attr = kzalloc(sizeof(*attr), GFP_KERNEL);
950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048
	if (!attr)
		return -ENOMEM;

	attr->qp_state = IB_QPS_INIT;
	attr->qp_access_flags = IB_ACCESS_LOCAL_WRITE | IB_ACCESS_REMOTE_READ |
	    IB_ACCESS_REMOTE_WRITE;
	attr->port_num = ch->sport->port;
	attr->pkey_index = 0;

	ret = ib_modify_qp(qp, attr,
			   IB_QP_STATE | IB_QP_ACCESS_FLAGS | IB_QP_PORT |
			   IB_QP_PKEY_INDEX);

	kfree(attr);
	return ret;
}

/**
 * srpt_ch_qp_rtr() - Change the state of a channel to 'ready to receive' (RTR).
 * @ch: channel of the queue pair.
 * @qp: queue pair to change the state of.
 *
 * Returns zero upon success and a negative value upon failure.
 *
 * Note: currently a struct ib_qp_attr takes 136 bytes on a 64-bit system.
 * If this structure ever becomes larger, it might be necessary to allocate
 * it dynamically instead of on the stack.
 */
static int srpt_ch_qp_rtr(struct srpt_rdma_ch *ch, struct ib_qp *qp)
{
	struct ib_qp_attr qp_attr;
	int attr_mask;
	int ret;

	qp_attr.qp_state = IB_QPS_RTR;
	ret = ib_cm_init_qp_attr(ch->cm_id, &qp_attr, &attr_mask);
	if (ret)
		goto out;

	qp_attr.max_dest_rd_atomic = 4;

	ret = ib_modify_qp(qp, &qp_attr, attr_mask);

out:
	return ret;
}

/**
 * srpt_ch_qp_rts() - Change the state of a channel to 'ready to send' (RTS).
 * @ch: channel of the queue pair.
 * @qp: queue pair to change the state of.
 *
 * Returns zero upon success and a negative value upon failure.
 *
 * Note: currently a struct ib_qp_attr takes 136 bytes on a 64-bit system.
 * If this structure ever becomes larger, it might be necessary to allocate
 * it dynamically instead of on the stack.
 */
static int srpt_ch_qp_rts(struct srpt_rdma_ch *ch, struct ib_qp *qp)
{
	struct ib_qp_attr qp_attr;
	int attr_mask;
	int ret;

	qp_attr.qp_state = IB_QPS_RTS;
	ret = ib_cm_init_qp_attr(ch->cm_id, &qp_attr, &attr_mask);
	if (ret)
		goto out;

	qp_attr.max_rd_atomic = 4;

	ret = ib_modify_qp(qp, &qp_attr, attr_mask);

out:
	return ret;
}

/**
 * srpt_ch_qp_err() - Set the channel queue pair state to 'error'.
 */
static int srpt_ch_qp_err(struct srpt_rdma_ch *ch)
{
	struct ib_qp_attr qp_attr;

	qp_attr.qp_state = IB_QPS_ERR;
	return ib_modify_qp(ch->qp, &qp_attr, IB_QP_STATE);
}

/**
 * srpt_unmap_sg_to_ib_sge() - Unmap an IB SGE list.
 */
static void srpt_unmap_sg_to_ib_sge(struct srpt_rdma_ch *ch,
				    struct srpt_send_ioctx *ioctx)
{
	struct scatterlist *sg;
	enum dma_data_direction dir;

	BUG_ON(!ch);
	BUG_ON(!ioctx);
1049
	BUG_ON(ioctx->n_rdma && !ioctx->rdma_wrs);
1050 1051

	while (ioctx->n_rdma)
1052
		kfree(ioctx->rdma_wrs[--ioctx->n_rdma].wr.sg_list);
1053

1054 1055
	kfree(ioctx->rdma_wrs);
	ioctx->rdma_wrs = NULL;
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073

	if (ioctx->mapped_sg_count) {
		sg = ioctx->sg;
		WARN_ON(!sg);
		dir = ioctx->cmd.data_direction;
		BUG_ON(dir == DMA_NONE);
		ib_dma_unmap_sg(ch->sport->sdev->device, sg, ioctx->sg_cnt,
				opposite_dma_dir(dir));
		ioctx->mapped_sg_count = 0;
	}
}

/**
 * srpt_map_sg_to_ib_sge() - Map an SG list to an IB SGE list.
 */
static int srpt_map_sg_to_ib_sge(struct srpt_rdma_ch *ch,
				 struct srpt_send_ioctx *ioctx)
{
1074
	struct ib_device *dev = ch->sport->sdev->device;
1075 1076 1077 1078
	struct se_cmd *cmd;
	struct scatterlist *sg, *sg_orig;
	int sg_cnt;
	enum dma_data_direction dir;
1079
	struct ib_rdma_wr *riu;
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095
	struct srp_direct_buf *db;
	dma_addr_t dma_addr;
	struct ib_sge *sge;
	u64 raddr;
	u32 rsize;
	u32 tsize;
	u32 dma_len;
	int count, nrdma;
	int i, j, k;

	BUG_ON(!ch);
	BUG_ON(!ioctx);
	cmd = &ioctx->cmd;
	dir = cmd->data_direction;
	BUG_ON(dir == DMA_NONE);

1096 1097
	ioctx->sg = sg = sg_orig = cmd->t_data_sg;
	ioctx->sg_cnt = sg_cnt = cmd->t_data_nents;
1098 1099 1100 1101 1102 1103 1104 1105

	count = ib_dma_map_sg(ch->sport->sdev->device, sg, sg_cnt,
			      opposite_dma_dir(dir));
	if (unlikely(!count))
		return -EAGAIN;

	ioctx->mapped_sg_count = count;

1106 1107
	if (ioctx->rdma_wrs && ioctx->n_rdma_wrs)
		nrdma = ioctx->n_rdma_wrs;
1108 1109 1110 1111
	else {
		nrdma = (count + SRPT_DEF_SG_PER_WQE - 1) / SRPT_DEF_SG_PER_WQE
			+ ioctx->n_rbuf;

1112 1113 1114
		ioctx->rdma_wrs = kcalloc(nrdma, sizeof(*ioctx->rdma_wrs),
				GFP_KERNEL);
		if (!ioctx->rdma_wrs)
1115 1116
			goto free_mem;

1117
		ioctx->n_rdma_wrs = nrdma;
1118 1119 1120 1121
	}

	db = ioctx->rbufs;
	tsize = cmd->data_length;
1122
	dma_len = ib_sg_dma_len(dev, &sg[0]);
1123
	riu = ioctx->rdma_wrs;
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136

	/*
	 * For each remote desc - calculate the #ib_sge.
	 * If #ib_sge < SRPT_DEF_SG_PER_WQE per rdma operation then
	 *      each remote desc rdma_iu is required a rdma wr;
	 * else
	 *      we need to allocate extra rdma_iu to carry extra #ib_sge in
	 *      another rdma wr
	 */
	for (i = 0, j = 0;
	     j < count && i < ioctx->n_rbuf && tsize > 0; ++i, ++riu, ++db) {
		rsize = be32_to_cpu(db->len);
		raddr = be64_to_cpu(db->va);
1137
		riu->remote_addr = raddr;
1138
		riu->rkey = be32_to_cpu(db->key);
1139
		riu->wr.num_sge = 0;
1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152

		/* calculate how many sge required for this remote_buf */
		while (rsize > 0 && tsize > 0) {

			if (rsize >= dma_len) {
				tsize -= dma_len;
				rsize -= dma_len;
				raddr += dma_len;

				if (tsize > 0) {
					++j;
					if (j < count) {
						sg = sg_next(sg);
1153 1154
						dma_len = ib_sg_dma_len(
								dev, sg);
1155 1156 1157 1158 1159 1160 1161 1162
					}
				}
			} else {
				tsize -= rsize;
				dma_len -= rsize;
				rsize = 0;
			}

1163
			++riu->wr.num_sge;
1164

1165 1166
			if (rsize > 0 &&
			    riu->wr.num_sge == SRPT_DEF_SG_PER_WQE) {
1167
				++ioctx->n_rdma;
1168 1169 1170 1171
				riu->wr.sg_list = kmalloc_array(riu->wr.num_sge,
						sizeof(*riu->wr.sg_list),
						GFP_KERNEL);
				if (!riu->wr.sg_list)
1172 1173 1174
					goto free_mem;

				++riu;
1175 1176
				riu->wr.num_sge = 0;
				riu->remote_addr = raddr;
1177 1178 1179 1180 1181
				riu->rkey = be32_to_cpu(db->key);
			}
		}

		++ioctx->n_rdma;
1182 1183 1184 1185
		riu->wr.sg_list = kmalloc_array(riu->wr.num_sge,
					sizeof(*riu->wr.sg_list),
					GFP_KERNEL);
		if (!riu->wr.sg_list)
1186 1187 1188 1189 1190
			goto free_mem;
	}

	db = ioctx->rbufs;
	tsize = cmd->data_length;
1191
	riu = ioctx->rdma_wrs;
1192
	sg = sg_orig;
1193 1194
	dma_len = ib_sg_dma_len(dev, &sg[0]);
	dma_addr = ib_sg_dma_address(dev, &sg[0]);
1195 1196 1197 1198 1199

	/* this second loop is really mapped sg_addres to rdma_iu->ib_sge */
	for (i = 0, j = 0;
	     j < count && i < ioctx->n_rbuf && tsize > 0; ++i, ++riu, ++db) {
		rsize = be32_to_cpu(db->len);
1200
		sge = riu->wr.sg_list;
1201 1202 1203 1204
		k = 0;

		while (rsize > 0 && tsize > 0) {
			sge->addr = dma_addr;
1205
			sge->lkey = ch->sport->sdev->pd->local_dma_lkey;
1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216

			if (rsize >= dma_len) {
				sge->length =
					(tsize < dma_len) ? tsize : dma_len;
				tsize -= dma_len;
				rsize -= dma_len;

				if (tsize > 0) {
					++j;
					if (j < count) {
						sg = sg_next(sg);
1217 1218 1219 1220
						dma_len = ib_sg_dma_len(
								dev, sg);
						dma_addr = ib_sg_dma_address(
								dev, sg);
1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231
					}
				}
			} else {
				sge->length = (tsize < rsize) ? tsize : rsize;
				tsize -= rsize;
				dma_len -= rsize;
				dma_addr += rsize;
				rsize = 0;
			}

			++k;
1232
			if (k == riu->wr.num_sge && rsize > 0 && tsize > 0) {
1233
				++riu;
1234
				sge = riu->wr.sg_list;
1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276
				k = 0;
			} else if (rsize > 0 && tsize > 0)
				++sge;
		}
	}

	return 0;

free_mem:
	srpt_unmap_sg_to_ib_sge(ch, ioctx);

	return -ENOMEM;
}

/**
 * srpt_get_send_ioctx() - Obtain an I/O context for sending to the initiator.
 */
static struct srpt_send_ioctx *srpt_get_send_ioctx(struct srpt_rdma_ch *ch)
{
	struct srpt_send_ioctx *ioctx;
	unsigned long flags;

	BUG_ON(!ch);

	ioctx = NULL;
	spin_lock_irqsave(&ch->spinlock, flags);
	if (!list_empty(&ch->free_list)) {
		ioctx = list_first_entry(&ch->free_list,
					 struct srpt_send_ioctx, free_list);
		list_del(&ioctx->free_list);
	}
	spin_unlock_irqrestore(&ch->spinlock, flags);

	if (!ioctx)
		return ioctx;

	BUG_ON(ioctx->ch != ch);
	spin_lock_init(&ioctx->spinlock);
	ioctx->state = SRPT_STATE_NEW;
	ioctx->n_rbuf = 0;
	ioctx->rbufs = NULL;
	ioctx->n_rdma = 0;
1277 1278
	ioctx->n_rdma_wrs = 0;
	ioctx->rdma_wrs = NULL;
1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327
	ioctx->mapped_sg_count = 0;
	init_completion(&ioctx->tx_done);
	ioctx->queue_status_only = false;
	/*
	 * transport_init_se_cmd() does not initialize all fields, so do it
	 * here.
	 */
	memset(&ioctx->cmd, 0, sizeof(ioctx->cmd));
	memset(&ioctx->sense_data, 0, sizeof(ioctx->sense_data));

	return ioctx;
}

/**
 * srpt_abort_cmd() - Abort a SCSI command.
 * @ioctx:   I/O context associated with the SCSI command.
 * @context: Preferred execution context.
 */
static int srpt_abort_cmd(struct srpt_send_ioctx *ioctx)
{
	enum srpt_command_state state;
	unsigned long flags;

	BUG_ON(!ioctx);

	/*
	 * If the command is in a state where the target core is waiting for
	 * the ib_srpt driver, change the state to the next state. Changing
	 * the state of the command from SRPT_STATE_NEED_DATA to
	 * SRPT_STATE_DATA_IN ensures that srpt_xmit_response() will call this
	 * function a second time.
	 */

	spin_lock_irqsave(&ioctx->spinlock, flags);
	state = ioctx->state;
	switch (state) {
	case SRPT_STATE_NEED_DATA:
		ioctx->state = SRPT_STATE_DATA_IN;
		break;
	case SRPT_STATE_DATA_IN:
	case SRPT_STATE_CMD_RSP_SENT:
	case SRPT_STATE_MGMT_RSP_SENT:
		ioctx->state = SRPT_STATE_DONE;
		break;
	default:
		break;
	}
	spin_unlock_irqrestore(&ioctx->spinlock, flags);

1328 1329 1330 1331 1332
	if (state == SRPT_STATE_DONE) {
		struct srpt_rdma_ch *ch = ioctx->ch;

		BUG_ON(ch->sess == NULL);

1333
		target_put_sess_cmd(&ioctx->cmd);
1334
		goto out;
1335
	}
1336 1337

	pr_debug("Aborting cmd with state %d and tag %lld\n", state,
1338
		 ioctx->cmd.tag);
1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351

	switch (state) {
	case SRPT_STATE_NEW:
	case SRPT_STATE_DATA_IN:
	case SRPT_STATE_MGMT:
		/*
		 * Do nothing - defer abort processing until
		 * srpt_queue_response() is invoked.
		 */
		WARN_ON(!transport_check_aborted_status(&ioctx->cmd, false));
		break;
	case SRPT_STATE_NEED_DATA:
		/* DMA_TO_DEVICE (write) - RDMA read error. */
1352 1353

		/* XXX(hch): this is a horrible layering violation.. */
1354
		spin_lock_irqsave(&ioctx->cmd.t_state_lock, flags);
1355
		ioctx->cmd.transport_state &= ~CMD_T_ACTIVE;
1356
		spin_unlock_irqrestore(&ioctx->cmd.t_state_lock, flags);
1357 1358 1359 1360 1361 1362 1363
		break;
	case SRPT_STATE_CMD_RSP_SENT:
		/*
		 * SRP_RSP sending failed or the SRP_RSP send completion has
		 * not been received in time.
		 */
		srpt_unmap_sg_to_ib_sge(ioctx->ch, ioctx);
1364
		target_put_sess_cmd(&ioctx->cmd);
1365 1366 1367
		break;
	case SRPT_STATE_MGMT_RSP_SENT:
		srpt_set_cmd_state(ioctx, SRPT_STATE_DONE);
1368
		target_put_sess_cmd(&ioctx->cmd);
1369 1370
		break;
	default:
G
Grant Grundler 已提交
1371
		WARN(1, "Unexpected command state (%d)", state);
1372 1373 1374 1375 1376 1377 1378 1379
		break;
	}

out:
	return state;
}

/**
1380 1381
 * XXX: what is now target_execute_cmd used to be asynchronous, and unmapping
 * the data that has been transferred via IB RDMA had to be postponed until the
1382
 * check_stop_free() callback.  None of this is necessary anymore and needs to
1383
 * be cleaned up.
1384
 */
1385
static void srpt_rdma_read_done(struct ib_cq *cq, struct ib_wc *wc)
1386
{
1387 1388
	struct srpt_rdma_ch *ch = cq->cq_context;
	struct srpt_send_ioctx *ioctx =
1389
		container_of(wc->wr_cqe, struct srpt_send_ioctx, rdma_cqe);
1390

1391 1392 1393
	WARN_ON(ioctx->n_rdma <= 0);
	atomic_add(ioctx->n_rdma, &ch->sq_wr_avail);

1394 1395 1396 1397 1398
	if (unlikely(wc->status != IB_WC_SUCCESS)) {
		pr_info("RDMA_READ for ioctx 0x%p failed with status %d\n",
			ioctx, wc->status);
		srpt_abort_cmd(ioctx);
		return;
1399
	}
1400 1401 1402 1403 1404 1405 1406

	if (srpt_test_and_set_cmd_state(ioctx, SRPT_STATE_NEED_DATA,
					SRPT_STATE_DATA_IN))
		target_execute_cmd(&ioctx->cmd);
	else
		pr_err("%s[%d]: wrong state = %d\n", __func__,
		       __LINE__, srpt_get_cmd_state(ioctx));
1407 1408
}

1409
static void srpt_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
1410
{
1411
	struct srpt_send_ioctx *ioctx =
1412
		container_of(wc->wr_cqe, struct srpt_send_ioctx, rdma_cqe);
1413

1414 1415 1416 1417
	if (unlikely(wc->status != IB_WC_SUCCESS)) {
		pr_info("RDMA_WRITE for ioctx 0x%p failed with status %d\n",
			ioctx, wc->status);
		srpt_abort_cmd(ioctx);
1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456
	}
}

/**
 * srpt_build_cmd_rsp() - Build an SRP_RSP response.
 * @ch: RDMA channel through which the request has been received.
 * @ioctx: I/O context associated with the SRP_CMD request. The response will
 *   be built in the buffer ioctx->buf points at and hence this function will
 *   overwrite the request data.
 * @tag: tag of the request for which this response is being generated.
 * @status: value for the STATUS field of the SRP_RSP information unit.
 *
 * Returns the size in bytes of the SRP_RSP response.
 *
 * An SRP_RSP response contains a SCSI status or service response. See also
 * section 6.9 in the SRP r16a document for the format of an SRP_RSP
 * response. See also SPC-2 for more information about sense data.
 */
static int srpt_build_cmd_rsp(struct srpt_rdma_ch *ch,
			      struct srpt_send_ioctx *ioctx, u64 tag,
			      int status)
{
	struct srp_rsp *srp_rsp;
	const u8 *sense_data;
	int sense_data_len, max_sense_len;

	/*
	 * The lowest bit of all SAM-3 status codes is zero (see also
	 * paragraph 5.3 in SAM-3).
	 */
	WARN_ON(status & 1);

	srp_rsp = ioctx->ioctx.buf;
	BUG_ON(!srp_rsp);

	sense_data = ioctx->sense_data;
	sense_data_len = ioctx->cmd.scsi_sense_length;
	WARN_ON(sense_data_len > sizeof(ioctx->sense_data));

1457
	memset(srp_rsp, 0, sizeof(*srp_rsp));
1458 1459
	srp_rsp->opcode = SRP_RSP;
	srp_rsp->req_lim_delta =
1460
		cpu_to_be32(1 + atomic_xchg(&ch->req_lim_delta, 0));
1461 1462 1463 1464 1465 1466 1467
	srp_rsp->tag = tag;
	srp_rsp->status = status;

	if (sense_data_len) {
		BUILD_BUG_ON(MIN_MAX_RSP_SIZE <= sizeof(*srp_rsp));
		max_sense_len = ch->max_ti_iu_len - sizeof(*srp_rsp);
		if (sense_data_len > max_sense_len) {
1468 1469
			pr_warn("truncated sense data from %d to %d"
				" bytes\n", sense_data_len, max_sense_len);
1470 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
			sense_data_len = max_sense_len;
		}

		srp_rsp->flags |= SRP_RSP_FLAG_SNSVALID;
		srp_rsp->sense_data_len = cpu_to_be32(sense_data_len);
		memcpy(srp_rsp + 1, sense_data, sense_data_len);
	}

	return sizeof(*srp_rsp) + sense_data_len;
}

/**
 * srpt_build_tskmgmt_rsp() - Build a task management response.
 * @ch:       RDMA channel through which the request has been received.
 * @ioctx:    I/O context in which the SRP_RSP response will be built.
 * @rsp_code: RSP_CODE that will be stored in the response.
 * @tag:      Tag of the request for which this response is being generated.
 *
 * Returns the size in bytes of the SRP_RSP response.
 *
 * An SRP_RSP response contains a SCSI status or service response. See also
 * section 6.9 in the SRP r16a document for the format of an SRP_RSP
 * response.
 */
static int srpt_build_tskmgmt_rsp(struct srpt_rdma_ch *ch,
				  struct srpt_send_ioctx *ioctx,
				  u8 rsp_code, u64 tag)
{
	struct srp_rsp *srp_rsp;
	int resp_data_len;
	int resp_len;

1502
	resp_data_len = 4;
1503 1504 1505 1506
	resp_len = sizeof(*srp_rsp) + resp_data_len;

	srp_rsp = ioctx->ioctx.buf;
	BUG_ON(!srp_rsp);
1507
	memset(srp_rsp, 0, sizeof(*srp_rsp));
1508 1509

	srp_rsp->opcode = SRP_RSP;
1510 1511
	srp_rsp->req_lim_delta =
		cpu_to_be32(1 + atomic_xchg(&ch->req_lim_delta, 0));
1512 1513
	srp_rsp->tag = tag;

1514 1515 1516
	srp_rsp->flags |= SRP_RSP_FLAG_RSPVALID;
	srp_rsp->resp_data_len = cpu_to_be32(resp_data_len);
	srp_rsp->data[3] = rsp_code;
1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546

	return resp_len;
}

#define NO_SUCH_LUN ((uint64_t)-1LL)

/*
 * SCSI LUN addressing method. See also SAM-2 and the section about
 * eight byte LUNs.
 */
enum scsi_lun_addr_method {
	SCSI_LUN_ADDR_METHOD_PERIPHERAL   = 0,
	SCSI_LUN_ADDR_METHOD_FLAT         = 1,
	SCSI_LUN_ADDR_METHOD_LUN          = 2,
	SCSI_LUN_ADDR_METHOD_EXTENDED_LUN = 3,
};

/*
 * srpt_unpack_lun() - Convert from network LUN to linear LUN.
 *
 * Convert an 2-byte, 4-byte, 6-byte or 8-byte LUN structure in network byte
 * order (big endian) to a linear LUN. Supports three LUN addressing methods:
 * peripheral, flat and logical unit. See also SAM-2, section 4.9.4 (page 40).
 */
static uint64_t srpt_unpack_lun(const uint8_t *lun, int len)
{
	uint64_t res = NO_SUCH_LUN;
	int addressing_method;

	if (unlikely(len < 2)) {
1547 1548
		pr_err("Illegal LUN length %d, expected 2 bytes or more\n",
		       len);
1549 1550 1551 1552 1553 1554
		goto out;
	}

	switch (len) {
	case 8:
		if ((*((__be64 *)lun) &
1555
		     cpu_to_be64(0x0000FFFFFFFFFFFFLL)) != 0)
1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581
			goto out_err;
		break;
	case 4:
		if (*((__be16 *)&lun[2]) != 0)
			goto out_err;
		break;
	case 6:
		if (*((__be32 *)&lun[2]) != 0)
			goto out_err;
		break;
	case 2:
		break;
	default:
		goto out_err;
	}

	addressing_method = (*lun) >> 6; /* highest two bits of byte 0 */
	switch (addressing_method) {
	case SCSI_LUN_ADDR_METHOD_PERIPHERAL:
	case SCSI_LUN_ADDR_METHOD_FLAT:
	case SCSI_LUN_ADDR_METHOD_LUN:
		res = *(lun + 1) | (((*lun) & 0x3f) << 8);
		break;

	case SCSI_LUN_ADDR_METHOD_EXTENDED_LUN:
	default:
1582
		pr_err("Unimplemented LUN addressing method %u\n",
1583 1584 1585 1586 1587 1588 1589 1590
		       addressing_method);
		break;
	}

out:
	return res;

out_err:
1591
	pr_err("Support for multi-level LUNs has not yet been implemented\n");
1592 1593 1594 1595 1596
	goto out;
}

static int srpt_check_stop_free(struct se_cmd *cmd)
{
1597 1598
	struct srpt_send_ioctx *ioctx = container_of(cmd,
				struct srpt_send_ioctx, cmd);
1599

1600
	return target_put_sess_cmd(&ioctx->cmd);
1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614
}

/**
 * srpt_handle_cmd() - Process SRP_CMD.
 */
static int srpt_handle_cmd(struct srpt_rdma_ch *ch,
			   struct srpt_recv_ioctx *recv_ioctx,
			   struct srpt_send_ioctx *send_ioctx)
{
	struct se_cmd *cmd;
	struct srp_cmd *srp_cmd;
	uint64_t unpacked_lun;
	u64 data_len;
	enum dma_data_direction dir;
1615
	sense_reason_t ret;
1616
	int rc;
1617 1618 1619 1620 1621

	BUG_ON(!send_ioctx);

	srp_cmd = recv_ioctx->ioctx.buf;
	cmd = &send_ioctx->cmd;
1622
	cmd->tag = srp_cmd->tag;
1623 1624 1625

	switch (srp_cmd->task_attr) {
	case SRP_CMD_SIMPLE_Q:
C
Christoph Hellwig 已提交
1626
		cmd->sam_task_attr = TCM_SIMPLE_TAG;
1627 1628 1629
		break;
	case SRP_CMD_ORDERED_Q:
	default:
C
Christoph Hellwig 已提交
1630
		cmd->sam_task_attr = TCM_ORDERED_TAG;
1631 1632
		break;
	case SRP_CMD_HEAD_OF_Q:
C
Christoph Hellwig 已提交
1633
		cmd->sam_task_attr = TCM_HEAD_TAG;
1634 1635
		break;
	case SRP_CMD_ACA:
C
Christoph Hellwig 已提交
1636
		cmd->sam_task_attr = TCM_ACA_TAG;
1637 1638 1639
		break;
	}

1640
	if (srpt_get_desc_tbl(send_ioctx, srp_cmd, &dir, &data_len)) {
1641
		pr_err("0x%llx: parsing SRP descriptor table failed.\n",
1642
		       srp_cmd->tag);
1643
		ret = TCM_INVALID_CDB_FIELD;
1644 1645 1646 1647 1648
		goto send_sense;
	}

	unpacked_lun = srpt_unpack_lun((uint8_t *)&srp_cmd->lun,
				       sizeof(srp_cmd->lun));
1649 1650
	rc = target_submit_cmd(cmd, ch->sess, srp_cmd->cdb,
			&send_ioctx->sense_data[0], unpacked_lun, data_len,
C
Christoph Hellwig 已提交
1651
			TCM_SIMPLE_TAG, dir, TARGET_SCF_ACK_KREF);
1652 1653
	if (rc != 0) {
		ret = TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
1654
		goto send_sense;
1655
	}
1656 1657 1658
	return 0;

send_sense:
1659
	transport_send_check_condition_and_sense(cmd, ret, 0);
1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694
	return -1;
}

static int srp_tmr_to_tcm(int fn)
{
	switch (fn) {
	case SRP_TSK_ABORT_TASK:
		return TMR_ABORT_TASK;
	case SRP_TSK_ABORT_TASK_SET:
		return TMR_ABORT_TASK_SET;
	case SRP_TSK_CLEAR_TASK_SET:
		return TMR_CLEAR_TASK_SET;
	case SRP_TSK_LUN_RESET:
		return TMR_LUN_RESET;
	case SRP_TSK_CLEAR_ACA:
		return TMR_CLEAR_ACA;
	default:
		return -1;
	}
}

/**
 * srpt_handle_tsk_mgmt() - Process an SRP_TSK_MGMT information unit.
 *
 * Returns 0 if and only if the request will be processed by the target core.
 *
 * For more information about SRP_TSK_MGMT information units, see also section
 * 6.7 in the SRP r16a document.
 */
static void srpt_handle_tsk_mgmt(struct srpt_rdma_ch *ch,
				 struct srpt_recv_ioctx *recv_ioctx,
				 struct srpt_send_ioctx *send_ioctx)
{
	struct srp_tsk_mgmt *srp_tsk;
	struct se_cmd *cmd;
1695
	struct se_session *sess = ch->sess;
1696 1697
	uint64_t unpacked_lun;
	int tcm_tmr;
1698
	int rc;
1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709

	BUG_ON(!send_ioctx);

	srp_tsk = recv_ioctx->ioctx.buf;
	cmd = &send_ioctx->cmd;

	pr_debug("recv tsk_mgmt fn %d for task_tag %lld and cmd tag %lld"
		 " cm_id %p sess %p\n", srp_tsk->tsk_mgmt_func,
		 srp_tsk->task_tag, srp_tsk->tag, ch->cm_id, ch->sess);

	srpt_set_cmd_state(send_ioctx, SRPT_STATE_MGMT);
1710
	send_ioctx->cmd.tag = srp_tsk->tag;
1711 1712 1713
	tcm_tmr = srp_tmr_to_tcm(srp_tsk->tsk_mgmt_func);
	unpacked_lun = srpt_unpack_lun((uint8_t *)&srp_tsk->lun,
				       sizeof(srp_tsk->lun));
1714
	rc = target_submit_tmr(&send_ioctx->cmd, sess, NULL, unpacked_lun,
1715
				srp_tsk, tcm_tmr, GFP_KERNEL, srp_tsk->task_tag,
1716 1717 1718
				TARGET_SCF_ACK_KREF);
	if (rc != 0) {
		send_ioctx->cmd.se_tmr_req->response = TMR_FUNCTION_REJECTED;
1719
		goto fail;
1720
	}
1721 1722 1723
	return;
fail:
	transport_send_check_condition_and_sense(cmd, 0, 0); // XXX:
1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772
}

/**
 * srpt_handle_new_iu() - Process a newly received information unit.
 * @ch:    RDMA channel through which the information unit has been received.
 * @ioctx: SRPT I/O context associated with the information unit.
 */
static void srpt_handle_new_iu(struct srpt_rdma_ch *ch,
			       struct srpt_recv_ioctx *recv_ioctx,
			       struct srpt_send_ioctx *send_ioctx)
{
	struct srp_cmd *srp_cmd;
	enum rdma_ch_state ch_state;

	BUG_ON(!ch);
	BUG_ON(!recv_ioctx);

	ib_dma_sync_single_for_cpu(ch->sport->sdev->device,
				   recv_ioctx->ioctx.dma, srp_max_req_size,
				   DMA_FROM_DEVICE);

	ch_state = srpt_get_ch_state(ch);
	if (unlikely(ch_state == CH_CONNECTING)) {
		list_add_tail(&recv_ioctx->wait_list, &ch->cmd_wait_list);
		goto out;
	}

	if (unlikely(ch_state != CH_LIVE))
		goto out;

	srp_cmd = recv_ioctx->ioctx.buf;
	if (srp_cmd->opcode == SRP_CMD || srp_cmd->opcode == SRP_TSK_MGMT) {
		if (!send_ioctx)
			send_ioctx = srpt_get_send_ioctx(ch);
		if (unlikely(!send_ioctx)) {
			list_add_tail(&recv_ioctx->wait_list,
				      &ch->cmd_wait_list);
			goto out;
		}
	}

	switch (srp_cmd->opcode) {
	case SRP_CMD:
		srpt_handle_cmd(ch, recv_ioctx, send_ioctx);
		break;
	case SRP_TSK_MGMT:
		srpt_handle_tsk_mgmt(ch, recv_ioctx, send_ioctx);
		break;
	case SRP_I_LOGOUT:
1773
		pr_err("Not yet implemented: SRP_I_LOGOUT\n");
1774 1775 1776 1777 1778 1779 1780 1781
		break;
	case SRP_CRED_RSP:
		pr_debug("received SRP_CRED_RSP\n");
		break;
	case SRP_AER_RSP:
		pr_debug("received SRP_AER_RSP\n");
		break;
	case SRP_RSP:
1782
		pr_err("Received SRP_RSP\n");
1783 1784
		break;
	default:
1785
		pr_err("received IU with unknown opcode 0x%x\n",
1786 1787 1788 1789 1790 1791 1792 1793 1794
		       srp_cmd->opcode);
		break;
	}

	srpt_post_recv(ch->sport->sdev, recv_ioctx);
out:
	return;
}

1795
static void srpt_recv_done(struct ib_cq *cq, struct ib_wc *wc)
1796
{
1797 1798 1799
	struct srpt_rdma_ch *ch = cq->cq_context;
	struct srpt_recv_ioctx *ioctx =
		container_of(wc->wr_cqe, struct srpt_recv_ioctx, ioctx.cqe);
1800 1801 1802 1803 1804 1805

	if (wc->status == IB_WC_SUCCESS) {
		int req_lim;

		req_lim = atomic_dec_return(&ch->req_lim);
		if (unlikely(req_lim < 0))
1806
			pr_err("req_lim = %d < 0\n", req_lim);
1807 1808
		srpt_handle_new_iu(ch, ioctx, NULL);
	} else {
1809 1810
		pr_info("receiving failed for ioctx %p with status %d\n",
			ioctx, wc->status);
1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
	}
}

/**
 * Note: Although this has not yet been observed during tests, at least in
 * theory it is possible that the srpt_get_send_ioctx() call invoked by
 * srpt_handle_new_iu() fails. This is possible because the req_lim_delta
 * value in each response is set to one, and it is possible that this response
 * makes the initiator send a new request before the send completion for that
 * response has been processed. This could e.g. happen if the call to
 * srpt_put_send_iotcx() is delayed because of a higher priority interrupt or
 * if IB retransmission causes generation of the send completion to be
 * delayed. Incoming information units for which srpt_get_send_ioctx() fails
 * are queued on cmd_wait_list. The code below processes these delayed
 * requests one at a time.
 */
1827
static void srpt_send_done(struct ib_cq *cq, struct ib_wc *wc)
1828
{
1829 1830 1831 1832
	struct srpt_rdma_ch *ch = cq->cq_context;
	struct srpt_send_ioctx *ioctx =
		container_of(wc->wr_cqe, struct srpt_send_ioctx, ioctx.cqe);
	enum srpt_command_state state;
1833

1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
	state = srpt_set_cmd_state(ioctx, SRPT_STATE_DONE);

	WARN_ON(state != SRPT_STATE_CMD_RSP_SENT &&
		state != SRPT_STATE_MGMT_RSP_SENT);

	atomic_inc(&ch->sq_wr_avail);

	if (wc->status != IB_WC_SUCCESS) {
		pr_info("sending response for ioctx 0x%p failed"
			" with status %d\n", ioctx, wc->status);

		atomic_dec(&ch->req_lim);
		srpt_abort_cmd(ioctx);
		goto out;
	}

	if (state != SRPT_STATE_DONE) {
		srpt_unmap_sg_to_ib_sge(ch, ioctx);
		transport_generic_free_cmd(&ioctx->cmd, 0);
1853
	} else {
1854 1855
		pr_err("IB completion has been received too late for"
		       " wr_id = %u.\n", ioctx->ioctx.index);
1856 1857
	}

1858 1859 1860 1861
out:
	while (!list_empty(&ch->cmd_wait_list) &&
	       srpt_get_ch_state(ch) == CH_LIVE &&
	       (ioctx = srpt_get_send_ioctx(ch)) != NULL) {
1862 1863 1864 1865 1866 1867
		struct srpt_recv_ioctx *recv_ioctx;

		recv_ioctx = list_first_entry(&ch->cmd_wait_list,
					      struct srpt_recv_ioctx,
					      wait_list);
		list_del(&recv_ioctx->wait_list);
1868
		srpt_handle_new_iu(ch, recv_ioctx, ioctx);
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885
	}
}

/**
 * srpt_create_ch_ib() - Create receive and send completion queues.
 */
static int srpt_create_ch_ib(struct srpt_rdma_ch *ch)
{
	struct ib_qp_init_attr *qp_init;
	struct srpt_port *sport = ch->sport;
	struct srpt_device *sdev = sport->sdev;
	u32 srp_sq_size = sport->port_attrib.srp_sq_size;
	int ret;

	WARN_ON(ch->rq_size < 1);

	ret = -ENOMEM;
1886
	qp_init = kzalloc(sizeof(*qp_init), GFP_KERNEL);
1887 1888 1889
	if (!qp_init)
		goto out;

1890
retry:
1891 1892
	ch->cq = ib_alloc_cq(sdev->device, ch, ch->rq_size + srp_sq_size,
			0 /* XXX: spread CQs */, IB_POLL_WORKQUEUE);
1893 1894
	if (IS_ERR(ch->cq)) {
		ret = PTR_ERR(ch->cq);
1895
		pr_err("failed to create CQ cqe= %d ret= %d\n",
1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913
		       ch->rq_size + srp_sq_size, ret);
		goto out;
	}

	qp_init->qp_context = (void *)ch;
	qp_init->event_handler
		= (void(*)(struct ib_event *, void*))srpt_qp_event;
	qp_init->send_cq = ch->cq;
	qp_init->recv_cq = ch->cq;
	qp_init->srq = sdev->srq;
	qp_init->sq_sig_type = IB_SIGNAL_REQ_WR;
	qp_init->qp_type = IB_QPT_RC;
	qp_init->cap.max_send_wr = srp_sq_size;
	qp_init->cap.max_send_sge = SRPT_DEF_SG_PER_WQE;

	ch->qp = ib_create_qp(sdev->pd, qp_init);
	if (IS_ERR(ch->qp)) {
		ret = PTR_ERR(ch->qp);
1914 1915 1916 1917 1918 1919 1920
		if (ret == -ENOMEM) {
			srp_sq_size /= 2;
			if (srp_sq_size >= MIN_SRPT_SQ_SIZE) {
				ib_destroy_cq(ch->cq);
				goto retry;
			}
		}
1921
		pr_err("failed to create_qp ret= %d\n", ret);
1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941
		goto err_destroy_cq;
	}

	atomic_set(&ch->sq_wr_avail, qp_init->cap.max_send_wr);

	pr_debug("%s: max_cqe= %d max_sge= %d sq_size = %d cm_id= %p\n",
		 __func__, ch->cq->cqe, qp_init->cap.max_send_sge,
		 qp_init->cap.max_send_wr, ch->cm_id);

	ret = srpt_init_ch_qp(ch, ch->qp);
	if (ret)
		goto err_destroy_qp;

out:
	kfree(qp_init);
	return ret;

err_destroy_qp:
	ib_destroy_qp(ch->qp);
err_destroy_cq:
1942
	ib_free_cq(ch->cq);
1943 1944 1945 1946 1947 1948
	goto out;
}

static void srpt_destroy_ch_ib(struct srpt_rdma_ch *ch)
{
	ib_destroy_qp(ch->qp);
1949
	ib_free_cq(ch->cq);
1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983
}

/**
 * __srpt_close_ch() - Close an RDMA channel by setting the QP error state.
 *
 * Reset the QP and make sure all resources associated with the channel will
 * be deallocated at an appropriate time.
 *
 * Note: The caller must hold ch->sport->sdev->spinlock.
 */
static void __srpt_close_ch(struct srpt_rdma_ch *ch)
{
	enum rdma_ch_state prev_state;
	unsigned long flags;

	spin_lock_irqsave(&ch->spinlock, flags);
	prev_state = ch->state;
	switch (prev_state) {
	case CH_CONNECTING:
	case CH_LIVE:
		ch->state = CH_DISCONNECTING;
		break;
	default:
		break;
	}
	spin_unlock_irqrestore(&ch->spinlock, flags);

	switch (prev_state) {
	case CH_CONNECTING:
		ib_send_cm_rej(ch->cm_id, IB_CM_REJ_NO_RESOURCES, NULL, 0,
			       NULL, 0);
		/* fall through */
	case CH_LIVE:
		if (ib_send_cm_dreq(ch->cm_id, NULL, 0) < 0)
1984
			pr_err("sending CM DREQ failed.\n");
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
		break;
	case CH_DISCONNECTING:
		break;
	case CH_DRAINING:
	case CH_RELEASING:
		break;
	}
}

/**
 * srpt_close_ch() - Close an RDMA channel.
 */
static void srpt_close_ch(struct srpt_rdma_ch *ch)
{
	struct srpt_device *sdev;

	sdev = ch->sport->sdev;
	spin_lock_irq(&sdev->spinlock);
	__srpt_close_ch(ch);
	spin_unlock_irq(&sdev->spinlock);
}

2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
/**
 * srpt_shutdown_session() - Whether or not a session may be shut down.
 */
static int srpt_shutdown_session(struct se_session *se_sess)
{
	struct srpt_rdma_ch *ch = se_sess->fabric_sess_ptr;
	unsigned long flags;

	spin_lock_irqsave(&ch->spinlock, flags);
	if (ch->in_shutdown) {
		spin_unlock_irqrestore(&ch->spinlock, flags);
		return true;
	}

	ch->in_shutdown = true;
	target_sess_cmd_list_set_waiting(se_sess);
	spin_unlock_irqrestore(&ch->spinlock, flags);

	return true;
}

2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065
/**
 * srpt_drain_channel() - Drain a channel by resetting the IB queue pair.
 * @cm_id: Pointer to the CM ID of the channel to be drained.
 *
 * Note: Must be called from inside srpt_cm_handler to avoid a race between
 * accessing sdev->spinlock and the call to kfree(sdev) in srpt_remove_one()
 * (the caller of srpt_cm_handler holds the cm_id spinlock; srpt_remove_one()
 * waits until all target sessions for the associated IB device have been
 * unregistered and target session registration involves a call to
 * ib_destroy_cm_id(), which locks the cm_id spinlock and hence waits until
 * this function has finished).
 */
static void srpt_drain_channel(struct ib_cm_id *cm_id)
{
	struct srpt_device *sdev;
	struct srpt_rdma_ch *ch;
	int ret;
	bool do_reset = false;

	WARN_ON_ONCE(irqs_disabled());

	sdev = cm_id->context;
	BUG_ON(!sdev);
	spin_lock_irq(&sdev->spinlock);
	list_for_each_entry(ch, &sdev->rch_list, list) {
		if (ch->cm_id == cm_id) {
			do_reset = srpt_test_and_set_ch_state(ch,
					CH_CONNECTING, CH_DRAINING) ||
				   srpt_test_and_set_ch_state(ch,
					CH_LIVE, CH_DRAINING) ||
				   srpt_test_and_set_ch_state(ch,
					CH_DISCONNECTING, CH_DRAINING);
			break;
		}
	}
	spin_unlock_irq(&sdev->spinlock);

	if (do_reset) {
2066 2067 2068
		if (ch->sess)
			srpt_shutdown_session(ch->sess);

2069 2070
		ret = srpt_ch_qp_err(ch);
		if (ret < 0)
2071
			pr_err("Setting queue pair in error state"
2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120
			       " failed: %d\n", ret);
	}
}

/**
 * srpt_find_channel() - Look up an RDMA channel.
 * @cm_id: Pointer to the CM ID of the channel to be looked up.
 *
 * Return NULL if no matching RDMA channel has been found.
 */
static struct srpt_rdma_ch *srpt_find_channel(struct srpt_device *sdev,
					      struct ib_cm_id *cm_id)
{
	struct srpt_rdma_ch *ch;
	bool found;

	WARN_ON_ONCE(irqs_disabled());
	BUG_ON(!sdev);

	found = false;
	spin_lock_irq(&sdev->spinlock);
	list_for_each_entry(ch, &sdev->rch_list, list) {
		if (ch->cm_id == cm_id) {
			found = true;
			break;
		}
	}
	spin_unlock_irq(&sdev->spinlock);

	return found ? ch : NULL;
}

/**
 * srpt_release_channel() - Release channel resources.
 *
 * Schedules the actual release because:
 * - Calling the ib_destroy_cm_id() call from inside an IB CM callback would
 *   trigger a deadlock.
 * - It is not safe to call TCM transport_* functions from interrupt context.
 */
static void srpt_release_channel(struct srpt_rdma_ch *ch)
{
	schedule_work(&ch->release_work);
}

static void srpt_release_channel_work(struct work_struct *w)
{
	struct srpt_rdma_ch *ch;
	struct srpt_device *sdev;
2121
	struct se_session *se_sess;
2122 2123 2124 2125 2126 2127 2128 2129

	ch = container_of(w, struct srpt_rdma_ch, release_work);
	pr_debug("ch = %p; ch->sess = %p; release_done = %p\n", ch, ch->sess,
		 ch->release_done);

	sdev = ch->sport->sdev;
	BUG_ON(!sdev);

2130 2131 2132
	se_sess = ch->sess;
	BUG_ON(!se_sess);

2133
	target_wait_for_sess_cmds(se_sess);
2134 2135 2136

	transport_deregister_session_configfs(se_sess);
	transport_deregister_session(se_sess);
2137 2138
	ch->sess = NULL;

2139 2140
	ib_destroy_cm_id(ch->cm_id);

2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175
	srpt_destroy_ch_ib(ch);

	srpt_free_ioctx_ring((struct srpt_ioctx **)ch->ioctx_ring,
			     ch->sport->sdev, ch->rq_size,
			     ch->rsp_size, DMA_TO_DEVICE);

	spin_lock_irq(&sdev->spinlock);
	list_del(&ch->list);
	spin_unlock_irq(&sdev->spinlock);

	if (ch->release_done)
		complete(ch->release_done);

	wake_up(&sdev->ch_releaseQ);

	kfree(ch);
}

/**
 * srpt_cm_req_recv() - Process the event IB_CM_REQ_RECEIVED.
 *
 * Ownership of the cm_id is transferred to the target session if this
 * functions returns zero. Otherwise the caller remains the owner of cm_id.
 */
static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
			    struct ib_cm_req_event_param *param,
			    void *private_data)
{
	struct srpt_device *sdev = cm_id->context;
	struct srpt_port *sport = &sdev->port[param->port - 1];
	struct srp_login_req *req;
	struct srp_login_rsp *rsp;
	struct srp_login_rej *rej;
	struct ib_cm_rep_param *rep_param;
	struct srpt_rdma_ch *ch, *tmp_ch;
2176
	struct se_node_acl *se_acl;
2177
	u32 it_iu_len;
2178 2179
	int i, ret = 0;
	unsigned char *p;
2180 2181 2182 2183 2184 2185 2186 2187 2188 2189

	WARN_ON_ONCE(irqs_disabled());

	if (WARN_ON(!sdev || !private_data))
		return -EINVAL;

	req = (struct srp_login_req *)private_data;

	it_iu_len = be32_to_cpu(req->req_it_iu_len);

2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200
	pr_info("Received SRP_LOGIN_REQ with i_port_id 0x%llx:0x%llx,"
		" t_port_id 0x%llx:0x%llx and it_iu_len %d on port %d"
		" (guid=0x%llx:0x%llx)\n",
		be64_to_cpu(*(__be64 *)&req->initiator_port_id[0]),
		be64_to_cpu(*(__be64 *)&req->initiator_port_id[8]),
		be64_to_cpu(*(__be64 *)&req->target_port_id[0]),
		be64_to_cpu(*(__be64 *)&req->target_port_id[8]),
		it_iu_len,
		param->port,
		be64_to_cpu(*(__be64 *)&sdev->port[param->port - 1].gid.raw[0]),
		be64_to_cpu(*(__be64 *)&sdev->port[param->port - 1].gid.raw[8]));
2201

2202 2203 2204
	rsp = kzalloc(sizeof(*rsp), GFP_KERNEL);
	rej = kzalloc(sizeof(*rej), GFP_KERNEL);
	rep_param = kzalloc(sizeof(*rep_param), GFP_KERNEL);
2205 2206 2207 2208 2209 2210 2211

	if (!rsp || !rej || !rep_param) {
		ret = -ENOMEM;
		goto out;
	}

	if (it_iu_len > srp_max_req_size || it_iu_len < 64) {
2212 2213
		rej->reason = cpu_to_be32(
			      SRP_LOGIN_REJ_REQ_IT_IU_LENGTH_TOO_LARGE);
2214
		ret = -EINVAL;
2215
		pr_err("rejected SRP_LOGIN_REQ because its"
2216 2217 2218 2219 2220 2221
		       " length (%d bytes) is out of range (%d .. %d)\n",
		       it_iu_len, 64, srp_max_req_size);
		goto reject;
	}

	if (!sport->enabled) {
2222 2223
		rej->reason = cpu_to_be32(
			      SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
2224
		ret = -EINVAL;
2225
		pr_err("rejected SRP_LOGIN_REQ because the target port"
2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267
		       " has not yet been enabled\n");
		goto reject;
	}

	if ((req->req_flags & SRP_MTCH_ACTION) == SRP_MULTICHAN_SINGLE) {
		rsp->rsp_flags = SRP_LOGIN_RSP_MULTICHAN_NO_CHAN;

		spin_lock_irq(&sdev->spinlock);

		list_for_each_entry_safe(ch, tmp_ch, &sdev->rch_list, list) {
			if (!memcmp(ch->i_port_id, req->initiator_port_id, 16)
			    && !memcmp(ch->t_port_id, req->target_port_id, 16)
			    && param->port == ch->sport->port
			    && param->listen_id == ch->sport->sdev->cm_id
			    && ch->cm_id) {
				enum rdma_ch_state ch_state;

				ch_state = srpt_get_ch_state(ch);
				if (ch_state != CH_CONNECTING
				    && ch_state != CH_LIVE)
					continue;

				/* found an existing channel */
				pr_debug("Found existing channel %s"
					 " cm_id= %p state= %d\n",
					 ch->sess_name, ch->cm_id, ch_state);

				__srpt_close_ch(ch);

				rsp->rsp_flags =
					SRP_LOGIN_RSP_MULTICHAN_TERMINATED;
			}
		}

		spin_unlock_irq(&sdev->spinlock);

	} else
		rsp->rsp_flags = SRP_LOGIN_RSP_MULTICHAN_MAINTAINED;

	if (*(__be64 *)req->target_port_id != cpu_to_be64(srpt_service_guid)
	    || *(__be64 *)(req->target_port_id + 8) !=
	       cpu_to_be64(srpt_service_guid)) {
2268 2269
		rej->reason = cpu_to_be32(
			      SRP_LOGIN_REJ_UNABLE_ASSOCIATE_CHANNEL);
2270
		ret = -ENOMEM;
2271
		pr_err("rejected SRP_LOGIN_REQ because it"
2272 2273 2274 2275
		       " has an invalid target port identifier.\n");
		goto reject;
	}

2276
	ch = kzalloc(sizeof(*ch), GFP_KERNEL);
2277
	if (!ch) {
2278 2279
		rej->reason = cpu_to_be32(
			      SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
2280
		pr_err("rejected SRP_LOGIN_REQ because no memory.\n");
2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314
		ret = -ENOMEM;
		goto reject;
	}

	INIT_WORK(&ch->release_work, srpt_release_channel_work);
	memcpy(ch->i_port_id, req->initiator_port_id, 16);
	memcpy(ch->t_port_id, req->target_port_id, 16);
	ch->sport = &sdev->port[param->port - 1];
	ch->cm_id = cm_id;
	/*
	 * Avoid QUEUE_FULL conditions by limiting the number of buffers used
	 * for the SRP protocol to the command queue size.
	 */
	ch->rq_size = SRPT_RQ_SIZE;
	spin_lock_init(&ch->spinlock);
	ch->state = CH_CONNECTING;
	INIT_LIST_HEAD(&ch->cmd_wait_list);
	ch->rsp_size = ch->sport->port_attrib.srp_max_rsp_size;

	ch->ioctx_ring = (struct srpt_send_ioctx **)
		srpt_alloc_ioctx_ring(ch->sport->sdev, ch->rq_size,
				      sizeof(*ch->ioctx_ring[0]),
				      ch->rsp_size, DMA_TO_DEVICE);
	if (!ch->ioctx_ring)
		goto free_ch;

	INIT_LIST_HEAD(&ch->free_list);
	for (i = 0; i < ch->rq_size; i++) {
		ch->ioctx_ring[i]->ch = ch;
		list_add_tail(&ch->ioctx_ring[i]->free_list, &ch->free_list);
	}

	ret = srpt_create_ch_ib(ch);
	if (ret) {
2315 2316
		rej->reason = cpu_to_be32(
			      SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
2317
		pr_err("rejected SRP_LOGIN_REQ because creating"
2318 2319 2320 2321 2322 2323
		       " a new RDMA channel failed.\n");
		goto free_ring;
	}

	ret = srpt_ch_qp_rtr(ch, ch->qp);
	if (ret) {
2324
		rej->reason = cpu_to_be32(SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
2325
		pr_err("rejected SRP_LOGIN_REQ because enabling"
2326 2327 2328
		       " RTR failed (error code = %d)\n", ret);
		goto destroy_ib;
	}
2329

2330
	/*
2331 2332 2333
	 * Use the initator port identifier as the session name, when
	 * checking against se_node_acl->initiatorname[] this can be
	 * with or without preceeding '0x'.
2334 2335 2336 2337 2338 2339
	 */
	snprintf(ch->sess_name, sizeof(ch->sess_name), "0x%016llx%016llx",
			be64_to_cpu(*(__be64 *)ch->i_port_id),
			be64_to_cpu(*(__be64 *)(ch->i_port_id + 8)));

	pr_debug("registering session %s\n", ch->sess_name);
2340
	p = &ch->sess_name[0];
2341

2342 2343
	ch->sess = transport_init_session(TARGET_PROT_NORMAL);
	if (IS_ERR(ch->sess)) {
2344
		rej->reason = cpu_to_be32(
2345 2346
				SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
		pr_debug("Failed to create session\n");
2347 2348 2349
		goto destroy_ib;
	}

2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361
try_again:
	se_acl = core_tpg_get_initiator_node_acl(&sport->port_tpg_1, p);
	if (!se_acl) {
		pr_info("Rejected login because no ACL has been"
			" configured yet for initiator %s.\n", ch->sess_name);
		/*
		 * XXX: Hack to retry of ch->i_port_id without leading '0x'
		 */
		if (p == &ch->sess_name[0]) {
			p += 2;
			goto try_again;
		}
2362
		rej->reason = cpu_to_be32(
2363 2364 2365
				SRP_LOGIN_REJ_CHANNEL_LIMIT_REACHED);
		transport_free_session(ch->sess);
		goto destroy_ib;
2366
	}
2367 2368 2369
	ch->sess->se_node_acl = se_acl;

	transport_register_session(&sport->port_tpg_1, se_acl, ch->sess, ch);
2370 2371 2372 2373 2374 2375 2376 2377 2378 2379

	pr_debug("Establish connection sess=%p name=%s cm_id=%p\n", ch->sess,
		 ch->sess_name, ch->cm_id);

	/* create srp_login_response */
	rsp->opcode = SRP_LOGIN_RSP;
	rsp->tag = req->tag;
	rsp->max_it_iu_len = req->req_it_iu_len;
	rsp->max_ti_iu_len = req->req_it_iu_len;
	ch->max_ti_iu_len = it_iu_len;
2380 2381
	rsp->buf_fmt = cpu_to_be16(SRP_BUF_FORMAT_DIRECT
				   | SRP_BUF_FORMAT_INDIRECT);
2382 2383 2384 2385 2386 2387 2388
	rsp->req_lim_delta = cpu_to_be32(ch->rq_size);
	atomic_set(&ch->req_lim, ch->rq_size);
	atomic_set(&ch->req_lim_delta, 0);

	/* create cm reply */
	rep_param->qp_num = ch->qp->qp_num;
	rep_param->private_data = (void *)rsp;
2389
	rep_param->private_data_len = sizeof(*rsp);
2390 2391 2392 2393 2394 2395 2396 2397 2398
	rep_param->rnr_retry_count = 7;
	rep_param->flow_control = 1;
	rep_param->failover_accepted = 0;
	rep_param->srq = 1;
	rep_param->responder_resources = 4;
	rep_param->initiator_depth = 4;

	ret = ib_send_cm_rep(cm_id, rep_param);
	if (ret) {
2399
		pr_err("sending SRP_LOGIN_REQ response failed"
2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428
		       " (error code = %d)\n", ret);
		goto release_channel;
	}

	spin_lock_irq(&sdev->spinlock);
	list_add_tail(&ch->list, &sdev->rch_list);
	spin_unlock_irq(&sdev->spinlock);

	goto out;

release_channel:
	srpt_set_ch_state(ch, CH_RELEASING);
	transport_deregister_session_configfs(ch->sess);
	transport_deregister_session(ch->sess);
	ch->sess = NULL;

destroy_ib:
	srpt_destroy_ch_ib(ch);

free_ring:
	srpt_free_ioctx_ring((struct srpt_ioctx **)ch->ioctx_ring,
			     ch->sport->sdev, ch->rq_size,
			     ch->rsp_size, DMA_TO_DEVICE);
free_ch:
	kfree(ch);

reject:
	rej->opcode = SRP_LOGIN_REJ;
	rej->tag = req->tag;
2429 2430
	rej->buf_fmt = cpu_to_be16(SRP_BUF_FORMAT_DIRECT
				   | SRP_BUF_FORMAT_INDIRECT);
2431 2432

	ib_send_cm_rej(cm_id, IB_CM_REJ_CONSUMER_DEFINED, NULL, 0,
2433
			     (void *)rej, sizeof(*rej));
2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444

out:
	kfree(rep_param);
	kfree(rsp);
	kfree(rej);

	return ret;
}

static void srpt_cm_rej_recv(struct ib_cm_id *cm_id)
{
2445
	pr_info("Received IB REJ for cm_id %p.\n", cm_id);
2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479
	srpt_drain_channel(cm_id);
}

/**
 * srpt_cm_rtu_recv() - Process an IB_CM_RTU_RECEIVED or USER_ESTABLISHED event.
 *
 * An IB_CM_RTU_RECEIVED message indicates that the connection is established
 * and that the recipient may begin transmitting (RTU = ready to use).
 */
static void srpt_cm_rtu_recv(struct ib_cm_id *cm_id)
{
	struct srpt_rdma_ch *ch;
	int ret;

	ch = srpt_find_channel(cm_id->context, cm_id);
	BUG_ON(!ch);

	if (srpt_test_and_set_ch_state(ch, CH_CONNECTING, CH_LIVE)) {
		struct srpt_recv_ioctx *ioctx, *ioctx_tmp;

		ret = srpt_ch_qp_rts(ch, ch->qp);

		list_for_each_entry_safe(ioctx, ioctx_tmp, &ch->cmd_wait_list,
					 wait_list) {
			list_del(&ioctx->wait_list);
			srpt_handle_new_iu(ch, ioctx, NULL);
		}
		if (ret)
			srpt_close_ch(ch);
	}
}

static void srpt_cm_timewait_exit(struct ib_cm_id *cm_id)
{
2480
	pr_info("Received IB TimeWait exit for cm_id %p.\n", cm_id);
2481 2482 2483 2484 2485
	srpt_drain_channel(cm_id);
}

static void srpt_cm_rep_error(struct ib_cm_id *cm_id)
{
2486
	pr_info("Received IB REP error for cm_id %p.\n", cm_id);
2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520
	srpt_drain_channel(cm_id);
}

/**
 * srpt_cm_dreq_recv() - Process reception of a DREQ message.
 */
static void srpt_cm_dreq_recv(struct ib_cm_id *cm_id)
{
	struct srpt_rdma_ch *ch;
	unsigned long flags;
	bool send_drep = false;

	ch = srpt_find_channel(cm_id->context, cm_id);
	BUG_ON(!ch);

	pr_debug("cm_id= %p ch->state= %d\n", cm_id, srpt_get_ch_state(ch));

	spin_lock_irqsave(&ch->spinlock, flags);
	switch (ch->state) {
	case CH_CONNECTING:
	case CH_LIVE:
		send_drep = true;
		ch->state = CH_DISCONNECTING;
		break;
	case CH_DISCONNECTING:
	case CH_DRAINING:
	case CH_RELEASING:
		WARN(true, "unexpected channel state %d\n", ch->state);
		break;
	}
	spin_unlock_irqrestore(&ch->spinlock, flags);

	if (send_drep) {
		if (ib_send_cm_drep(ch->cm_id, NULL, 0) < 0)
2521 2522 2523
			pr_err("Sending IB DREP failed.\n");
		pr_info("Received DREQ and sent DREP for session %s.\n",
			ch->sess_name);
2524 2525 2526 2527 2528 2529 2530 2531
	}
}

/**
 * srpt_cm_drep_recv() - Process reception of a DREP message.
 */
static void srpt_cm_drep_recv(struct ib_cm_id *cm_id)
{
2532
	pr_info("Received InfiniBand DREP message for cm_id %p.\n", cm_id);
2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575
	srpt_drain_channel(cm_id);
}

/**
 * srpt_cm_handler() - IB connection manager callback function.
 *
 * A non-zero return value will cause the caller destroy the CM ID.
 *
 * Note: srpt_cm_handler() must only return a non-zero value when transferring
 * ownership of the cm_id to a channel by srpt_cm_req_recv() failed. Returning
 * a non-zero value in any other case will trigger a race with the
 * ib_destroy_cm_id() call in srpt_release_channel().
 */
static int srpt_cm_handler(struct ib_cm_id *cm_id, struct ib_cm_event *event)
{
	int ret;

	ret = 0;
	switch (event->event) {
	case IB_CM_REQ_RECEIVED:
		ret = srpt_cm_req_recv(cm_id, &event->param.req_rcvd,
				       event->private_data);
		break;
	case IB_CM_REJ_RECEIVED:
		srpt_cm_rej_recv(cm_id);
		break;
	case IB_CM_RTU_RECEIVED:
	case IB_CM_USER_ESTABLISHED:
		srpt_cm_rtu_recv(cm_id);
		break;
	case IB_CM_DREQ_RECEIVED:
		srpt_cm_dreq_recv(cm_id);
		break;
	case IB_CM_DREP_RECEIVED:
		srpt_cm_drep_recv(cm_id);
		break;
	case IB_CM_TIMEWAIT_EXIT:
		srpt_cm_timewait_exit(cm_id);
		break;
	case IB_CM_REP_ERROR:
		srpt_cm_rep_error(cm_id);
		break;
	case IB_CM_DREQ_ERROR:
2576
		pr_info("Received IB DREQ ERROR event.\n");
2577 2578
		break;
	case IB_CM_MRA_RECEIVED:
2579
		pr_info("Received IB MRA event\n");
2580 2581
		break;
	default:
2582
		pr_err("received unrecognized IB CM event %d\n", event->event);
2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597
		break;
	}

	return ret;
}

/**
 * srpt_perform_rdmas() - Perform IB RDMA.
 *
 * Returns zero upon success or a negative number upon failure.
 */
static int srpt_perform_rdmas(struct srpt_rdma_ch *ch,
			      struct srpt_send_ioctx *ioctx)
{
	struct ib_send_wr *bad_wr;
2598
	int sq_wr_avail, ret, i;
2599 2600 2601 2602 2603 2604 2605 2606 2607
	enum dma_data_direction dir;
	const int n_rdma = ioctx->n_rdma;

	dir = ioctx->cmd.data_direction;
	if (dir == DMA_TO_DEVICE) {
		/* write */
		ret = -ENOMEM;
		sq_wr_avail = atomic_sub_return(n_rdma, &ch->sq_wr_avail);
		if (sq_wr_avail < 0) {
2608 2609
			pr_warn("IB send queue full (needed %d)\n",
				n_rdma);
2610 2611 2612 2613
			goto out;
		}
	}

2614 2615
	for (i = 0; i < n_rdma; i++) {
		struct ib_send_wr *wr = &ioctx->rdma_wrs[i].wr;
2616

2617 2618
		wr->opcode = (dir == DMA_FROM_DEVICE) ?
				IB_WR_RDMA_WRITE : IB_WR_RDMA_READ;
2619

2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633
		if (i == n_rdma - 1) {
			/* only get completion event for the last rdma read */
			if (dir == DMA_TO_DEVICE) {
				wr->send_flags = IB_SEND_SIGNALED;
				ioctx->rdma_cqe.done = srpt_rdma_read_done;
			} else {
				ioctx->rdma_cqe.done = srpt_rdma_write_done;
			}
			wr->wr_cqe = &ioctx->rdma_cqe;
			wr->next = NULL;
		} else {
			wr->wr_cqe = NULL;
			wr->next = &ioctx->rdma_wrs[i + 1].wr;
		}
2634 2635
	}

2636
	ret = ib_post_send(ch->qp, &ioctx->rdma_wrs->wr, &bad_wr);
2637
	if (ret)
2638
		pr_err("%s[%d]: ib_post_send() returned %d for %d/%d\n",
2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655
				 __func__, __LINE__, ret, i, n_rdma);
out:
	if (unlikely(dir == DMA_TO_DEVICE && ret < 0))
		atomic_add(n_rdma, &ch->sq_wr_avail);
	return ret;
}

/**
 * srpt_xfer_data() - Start data transfer from initiator to target.
 */
static int srpt_xfer_data(struct srpt_rdma_ch *ch,
			  struct srpt_send_ioctx *ioctx)
{
	int ret;

	ret = srpt_map_sg_to_ib_sge(ch, ioctx);
	if (ret) {
2656
		pr_err("%s[%d] ret=%d\n", __func__, __LINE__, ret);
2657 2658 2659 2660 2661 2662
		goto out;
	}

	ret = srpt_perform_rdmas(ch, ioctx);
	if (ret) {
		if (ret == -EAGAIN || ret == -ENOMEM)
2663 2664
			pr_info("%s[%d] queue full -- ret=%d\n",
				__func__, __LINE__, ret);
2665
		else
2666
			pr_err("%s[%d] fatal error -- ret=%d\n",
2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716
			       __func__, __LINE__, ret);
		goto out_unmap;
	}

out:
	return ret;
out_unmap:
	srpt_unmap_sg_to_ib_sge(ch, ioctx);
	goto out;
}

static int srpt_write_pending_status(struct se_cmd *se_cmd)
{
	struct srpt_send_ioctx *ioctx;

	ioctx = container_of(se_cmd, struct srpt_send_ioctx, cmd);
	return srpt_get_cmd_state(ioctx) == SRPT_STATE_NEED_DATA;
}

/*
 * srpt_write_pending() - Start data transfer from initiator to target (write).
 */
static int srpt_write_pending(struct se_cmd *se_cmd)
{
	struct srpt_rdma_ch *ch;
	struct srpt_send_ioctx *ioctx;
	enum srpt_command_state new_state;
	enum rdma_ch_state ch_state;
	int ret;

	ioctx = container_of(se_cmd, struct srpt_send_ioctx, cmd);

	new_state = srpt_set_cmd_state(ioctx, SRPT_STATE_NEED_DATA);
	WARN_ON(new_state == SRPT_STATE_DONE);

	ch = ioctx->ch;
	BUG_ON(!ch);

	ch_state = srpt_get_ch_state(ch);
	switch (ch_state) {
	case CH_CONNECTING:
		WARN(true, "unexpected channel state %d\n", ch_state);
		ret = -EINVAL;
		goto out;
	case CH_LIVE:
		break;
	case CH_DISCONNECTING:
	case CH_DRAINING:
	case CH_RELEASING:
		pr_debug("cmd with tag %lld: channel disconnecting\n",
2717
			 ioctx->cmd.tag);
2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744
		srpt_set_cmd_state(ioctx, SRPT_STATE_DATA_IN);
		ret = -EINVAL;
		goto out;
	}
	ret = srpt_xfer_data(ch, ioctx);

out:
	return ret;
}

static u8 tcm_to_srp_tsk_mgmt_status(const int tcm_mgmt_status)
{
	switch (tcm_mgmt_status) {
	case TMR_FUNCTION_COMPLETE:
		return SRP_TSK_MGMT_SUCCESS;
	case TMR_FUNCTION_REJECTED:
		return SRP_TSK_MGMT_FUNC_NOT_SUPP;
	}
	return SRP_TSK_MGMT_FAILED;
}

/**
 * srpt_queue_response() - Transmits the response to a SCSI command.
 *
 * Callback function called by the TCM core. Must not block since it can be
 * invoked on the context of the IB completion handler.
 */
2745
static void srpt_queue_response(struct se_cmd *cmd)
2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780
{
	struct srpt_rdma_ch *ch;
	struct srpt_send_ioctx *ioctx;
	enum srpt_command_state state;
	unsigned long flags;
	int ret;
	enum dma_data_direction dir;
	int resp_len;
	u8 srp_tm_status;

	ioctx = container_of(cmd, struct srpt_send_ioctx, cmd);
	ch = ioctx->ch;
	BUG_ON(!ch);

	spin_lock_irqsave(&ioctx->spinlock, flags);
	state = ioctx->state;
	switch (state) {
	case SRPT_STATE_NEW:
	case SRPT_STATE_DATA_IN:
		ioctx->state = SRPT_STATE_CMD_RSP_SENT;
		break;
	case SRPT_STATE_MGMT:
		ioctx->state = SRPT_STATE_MGMT_RSP_SENT;
		break;
	default:
		WARN(true, "ch %p; cmd %d: unexpected command state %d\n",
			ch, ioctx->ioctx.index, ioctx->state);
		break;
	}
	spin_unlock_irqrestore(&ioctx->spinlock, flags);

	if (unlikely(transport_check_aborted_status(&ioctx->cmd, false)
		     || WARN_ON_ONCE(state == SRPT_STATE_CMD_RSP_SENT))) {
		atomic_inc(&ch->req_lim_delta);
		srpt_abort_cmd(ioctx);
2781
		return;
2782 2783 2784 2785 2786 2787 2788 2789 2790
	}

	dir = ioctx->cmd.data_direction;

	/* For read commands, transfer the data to the initiator. */
	if (dir == DMA_FROM_DEVICE && ioctx->cmd.data_length &&
	    !ioctx->queue_status_only) {
		ret = srpt_xfer_data(ch, ioctx);
		if (ret) {
2791
			pr_err("xfer_data failed for tag %llu\n",
2792
			       ioctx->cmd.tag);
2793
			return;
2794 2795 2796 2797
		}
	}

	if (state != SRPT_STATE_MGMT)
2798
		resp_len = srpt_build_cmd_rsp(ch, ioctx, ioctx->cmd.tag,
2799 2800 2801 2802 2803
					      cmd->scsi_status);
	else {
		srp_tm_status
			= tcm_to_srp_tsk_mgmt_status(cmd->se_tmr_req->response);
		resp_len = srpt_build_tskmgmt_rsp(ch, ioctx, srp_tm_status,
2804
						 ioctx->cmd.tag);
2805 2806 2807
	}
	ret = srpt_post_send(ch, ioctx, resp_len);
	if (ret) {
2808
		pr_err("sending cmd response failed for tag %llu\n",
2809
		       ioctx->cmd.tag);
2810 2811
		srpt_unmap_sg_to_ib_sge(ch, ioctx);
		srpt_set_cmd_state(ioctx, SRPT_STATE_DONE);
2812
		target_put_sess_cmd(&ioctx->cmd);
2813
	}
2814
}
2815

2816 2817 2818 2819 2820 2821 2822 2823 2824
static int srpt_queue_data_in(struct se_cmd *cmd)
{
	srpt_queue_response(cmd);
	return 0;
}

static void srpt_queue_tm_rsp(struct se_cmd *cmd)
{
	srpt_queue_response(cmd);
2825 2826
}

2827 2828 2829 2830 2831 2832 2833 2834
static void srpt_aborted_task(struct se_cmd *cmd)
{
	struct srpt_send_ioctx *ioctx = container_of(cmd,
				struct srpt_send_ioctx, cmd);

	srpt_unmap_sg_to_ib_sge(ioctx->ch, ioctx);
}

2835 2836 2837 2838 2839 2840 2841 2842 2843 2844
static int srpt_queue_status(struct se_cmd *cmd)
{
	struct srpt_send_ioctx *ioctx;

	ioctx = container_of(cmd, struct srpt_send_ioctx, cmd);
	BUG_ON(ioctx->sense_data != cmd->sense_buffer);
	if (cmd->se_cmd_flags &
	    (SCF_TRANSPORT_TASK_SENSE | SCF_EMULATED_TASK_SENSE))
		WARN_ON(cmd->scsi_status != SAM_STAT_CHECK_CONDITION);
	ioctx->queue_status_only = true;
2845 2846
	srpt_queue_response(cmd);
	return 0;
2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886
}

static void srpt_refresh_port_work(struct work_struct *work)
{
	struct srpt_port *sport = container_of(work, struct srpt_port, work);

	srpt_refresh_port(sport);
}

static int srpt_ch_list_empty(struct srpt_device *sdev)
{
	int res;

	spin_lock_irq(&sdev->spinlock);
	res = list_empty(&sdev->rch_list);
	spin_unlock_irq(&sdev->spinlock);

	return res;
}

/**
 * srpt_release_sdev() - Free the channel resources associated with a target.
 */
static int srpt_release_sdev(struct srpt_device *sdev)
{
	struct srpt_rdma_ch *ch, *tmp_ch;
	int res;

	WARN_ON_ONCE(irqs_disabled());

	BUG_ON(!sdev);

	spin_lock_irq(&sdev->spinlock);
	list_for_each_entry_safe(ch, tmp_ch, &sdev->rch_list, list)
		__srpt_close_ch(ch);
	spin_unlock_irq(&sdev->spinlock);

	res = wait_event_interruptible(sdev->ch_releaseQ,
				       srpt_ch_list_empty(sdev));
	if (res)
2887
		pr_err("%s: interrupted.\n", __func__);
2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938

	return 0;
}

static struct srpt_port *__srpt_lookup_port(const char *name)
{
	struct ib_device *dev;
	struct srpt_device *sdev;
	struct srpt_port *sport;
	int i;

	list_for_each_entry(sdev, &srpt_dev_list, list) {
		dev = sdev->device;
		if (!dev)
			continue;

		for (i = 0; i < dev->phys_port_cnt; i++) {
			sport = &sdev->port[i];

			if (!strcmp(sport->port_guid, name))
				return sport;
		}
	}

	return NULL;
}

static struct srpt_port *srpt_lookup_port(const char *name)
{
	struct srpt_port *sport;

	spin_lock(&srpt_dev_lock);
	sport = __srpt_lookup_port(name);
	spin_unlock(&srpt_dev_lock);

	return sport;
}

/**
 * srpt_add_one() - Infiniband device addition callback function.
 */
static void srpt_add_one(struct ib_device *device)
{
	struct srpt_device *sdev;
	struct srpt_port *sport;
	struct ib_srq_init_attr srq_attr;
	int i;

	pr_debug("device = %p, device->dma_ops = %p\n", device,
		 device->dma_ops);

2939
	sdev = kzalloc(sizeof(*sdev), GFP_KERNEL);
2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951
	if (!sdev)
		goto err;

	sdev->device = device;
	INIT_LIST_HEAD(&sdev->rch_list);
	init_waitqueue_head(&sdev->ch_releaseQ);
	spin_lock_init(&sdev->spinlock);

	sdev->pd = ib_alloc_pd(device);
	if (IS_ERR(sdev->pd))
		goto free_dev;

2952
	sdev->srq_size = min(srpt_srq_size, sdev->device->attrs.max_srq_wr);
2953 2954 2955 2956 2957 2958

	srq_attr.event_handler = srpt_srq_event;
	srq_attr.srq_context = (void *)sdev;
	srq_attr.attr.max_wr = sdev->srq_size;
	srq_attr.attr.max_sge = 1;
	srq_attr.attr.srq_limit = 0;
2959
	srq_attr.srq_type = IB_SRQT_BASIC;
2960 2961 2962

	sdev->srq = ib_create_srq(sdev->pd, &srq_attr);
	if (IS_ERR(sdev->srq))
2963
		goto err_pd;
2964 2965

	pr_debug("%s: create SRQ #wr= %d max_allow=%d dev= %s\n",
2966
		 __func__, sdev->srq_size, sdev->device->attrs.max_srq_wr,
2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986
		 device->name);

	if (!srpt_service_guid)
		srpt_service_guid = be64_to_cpu(device->node_guid);

	sdev->cm_id = ib_create_cm_id(device, srpt_cm_handler, sdev);
	if (IS_ERR(sdev->cm_id))
		goto err_srq;

	/* print out target login information */
	pr_debug("Target login info: id_ext=%016llx,ioc_guid=%016llx,"
		 "pkey=ffff,service_id=%016llx\n", srpt_service_guid,
		 srpt_service_guid, srpt_service_guid);

	/*
	 * We do not have a consistent service_id (ie. also id_ext of target_id)
	 * to identify this target. We currently use the guid of the first HCA
	 * in the system as service_id; therefore, the target_id will change
	 * if this HCA is gone bad and replaced by different HCA
	 */
H
Haggai Eran 已提交
2987
	if (ib_cm_listen(sdev->cm_id, cpu_to_be64(srpt_service_guid), 0))
2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004
		goto err_cm;

	INIT_IB_EVENT_HANDLER(&sdev->event_handler, sdev->device,
			      srpt_event_handler);
	if (ib_register_event_handler(&sdev->event_handler))
		goto err_cm;

	sdev->ioctx_ring = (struct srpt_recv_ioctx **)
		srpt_alloc_ioctx_ring(sdev, sdev->srq_size,
				      sizeof(*sdev->ioctx_ring[0]),
				      srp_max_req_size, DMA_FROM_DEVICE);
	if (!sdev->ioctx_ring)
		goto err_event;

	for (i = 0; i < sdev->srq_size; ++i)
		srpt_post_recv(sdev, sdev->ioctx_ring[i]);

3005
	WARN_ON(sdev->device->phys_port_cnt > ARRAY_SIZE(sdev->port));
3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016

	for (i = 1; i <= sdev->device->phys_port_cnt; i++) {
		sport = &sdev->port[i - 1];
		sport->sdev = sdev;
		sport->port = i;
		sport->port_attrib.srp_max_rdma_size = DEFAULT_MAX_RDMA_SIZE;
		sport->port_attrib.srp_max_rsp_size = DEFAULT_MAX_RSP_SIZE;
		sport->port_attrib.srp_sq_size = DEF_SRPT_SQ_SIZE;
		INIT_WORK(&sport->work, srpt_refresh_port_work);

		if (srpt_refresh_port(sport)) {
3017
			pr_err("MAD registration failed for %s-%d.\n",
3018
			       sdev->device->name, i);
3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051
			goto err_ring;
		}
		snprintf(sport->port_guid, sizeof(sport->port_guid),
			"0x%016llx%016llx",
			be64_to_cpu(sport->gid.global.subnet_prefix),
			be64_to_cpu(sport->gid.global.interface_id));
	}

	spin_lock(&srpt_dev_lock);
	list_add_tail(&sdev->list, &srpt_dev_list);
	spin_unlock(&srpt_dev_lock);

out:
	ib_set_client_data(device, &srpt_client, sdev);
	pr_debug("added %s.\n", device->name);
	return;

err_ring:
	srpt_free_ioctx_ring((struct srpt_ioctx **)sdev->ioctx_ring, sdev,
			     sdev->srq_size, srp_max_req_size,
			     DMA_FROM_DEVICE);
err_event:
	ib_unregister_event_handler(&sdev->event_handler);
err_cm:
	ib_destroy_cm_id(sdev->cm_id);
err_srq:
	ib_destroy_srq(sdev->srq);
err_pd:
	ib_dealloc_pd(sdev->pd);
free_dev:
	kfree(sdev);
err:
	sdev = NULL;
3052
	pr_info("%s(%s) failed.\n", __func__, device->name);
3053 3054 3055 3056 3057 3058
	goto out;
}

/**
 * srpt_remove_one() - InfiniBand device removal callback function.
 */
3059
static void srpt_remove_one(struct ib_device *device, void *client_data)
3060
{
3061
	struct srpt_device *sdev = client_data;
3062 3063 3064
	int i;

	if (!sdev) {
3065
		pr_info("%s(%s): nothing to do.\n", __func__, device->name);
3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137
		return;
	}

	srpt_unregister_mad_agent(sdev);

	ib_unregister_event_handler(&sdev->event_handler);

	/* Cancel any work queued by the just unregistered IB event handler. */
	for (i = 0; i < sdev->device->phys_port_cnt; i++)
		cancel_work_sync(&sdev->port[i].work);

	ib_destroy_cm_id(sdev->cm_id);

	/*
	 * Unregistering a target must happen after destroying sdev->cm_id
	 * such that no new SRP_LOGIN_REQ information units can arrive while
	 * destroying the target.
	 */
	spin_lock(&srpt_dev_lock);
	list_del(&sdev->list);
	spin_unlock(&srpt_dev_lock);
	srpt_release_sdev(sdev);

	ib_destroy_srq(sdev->srq);
	ib_dealloc_pd(sdev->pd);

	srpt_free_ioctx_ring((struct srpt_ioctx **)sdev->ioctx_ring, sdev,
			     sdev->srq_size, srp_max_req_size, DMA_FROM_DEVICE);
	sdev->ioctx_ring = NULL;
	kfree(sdev);
}

static struct ib_client srpt_client = {
	.name = DRV_NAME,
	.add = srpt_add_one,
	.remove = srpt_remove_one
};

static int srpt_check_true(struct se_portal_group *se_tpg)
{
	return 1;
}

static int srpt_check_false(struct se_portal_group *se_tpg)
{
	return 0;
}

static char *srpt_get_fabric_name(void)
{
	return "srpt";
}

static char *srpt_get_fabric_wwn(struct se_portal_group *tpg)
{
	struct srpt_port *sport = container_of(tpg, struct srpt_port, port_tpg_1);

	return sport->port_guid;
}

static u16 srpt_get_tag(struct se_portal_group *tpg)
{
	return 1;
}

static u32 srpt_tpg_get_inst_index(struct se_portal_group *se_tpg)
{
	return 1;
}

static void srpt_release_cmd(struct se_cmd *se_cmd)
{
3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154
	struct srpt_send_ioctx *ioctx = container_of(se_cmd,
				struct srpt_send_ioctx, cmd);
	struct srpt_rdma_ch *ch = ioctx->ch;
	unsigned long flags;

	WARN_ON(ioctx->state != SRPT_STATE_DONE);
	WARN_ON(ioctx->mapped_sg_count != 0);

	if (ioctx->n_rbuf > 1) {
		kfree(ioctx->rbufs);
		ioctx->rbufs = NULL;
		ioctx->n_rbuf = 0;
	}

	spin_lock_irqsave(&ch->spinlock, flags);
	list_add(&ioctx->free_list, &ch->free_list);
	spin_unlock_irqrestore(&ch->spinlock, flags);
3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168
}

/**
 * srpt_close_session() - Forcibly close a session.
 *
 * Callback function invoked by the TCM core to clean up sessions associated
 * with a node ACL when the user invokes
 * rmdir /sys/kernel/config/target/$driver/$port/$tpg/acls/$i_port_id
 */
static void srpt_close_session(struct se_session *se_sess)
{
	DECLARE_COMPLETION_ONSTACK(release_done);
	struct srpt_rdma_ch *ch;
	struct srpt_device *sdev;
3169
	unsigned long res;
3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183

	ch = se_sess->fabric_sess_ptr;
	WARN_ON(ch->sess != se_sess);

	pr_debug("ch %p state %d\n", ch, srpt_get_ch_state(ch));

	sdev = ch->sport->sdev;
	spin_lock_irq(&sdev->spinlock);
	BUG_ON(ch->release_done);
	ch->release_done = &release_done;
	__srpt_close_ch(ch);
	spin_unlock_irq(&sdev->spinlock);

	res = wait_for_completion_timeout(&release_done, 60 * HZ);
3184
	WARN_ON(res == 0);
3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224
}

/**
 * srpt_sess_get_index() - Return the value of scsiAttIntrPortIndex (SCSI-MIB).
 *
 * A quote from RFC 4455 (SCSI-MIB) about this MIB object:
 * This object represents an arbitrary integer used to uniquely identify a
 * particular attached remote initiator port to a particular SCSI target port
 * within a particular SCSI target device within a particular SCSI instance.
 */
static u32 srpt_sess_get_index(struct se_session *se_sess)
{
	return 0;
}

static void srpt_set_default_node_attrs(struct se_node_acl *nacl)
{
}

/* Note: only used from inside debug printk's by the TCM core. */
static int srpt_get_tcm_cmd_state(struct se_cmd *se_cmd)
{
	struct srpt_send_ioctx *ioctx;

	ioctx = container_of(se_cmd, struct srpt_send_ioctx, cmd);
	return srpt_get_cmd_state(ioctx);
}

/**
 * srpt_parse_i_port_id() - Parse an initiator port ID.
 * @name: ASCII representation of a 128-bit initiator port ID.
 * @i_port_id: Binary 128-bit port ID.
 */
static int srpt_parse_i_port_id(u8 i_port_id[16], const char *name)
{
	const char *p;
	unsigned len, count, leading_zero_bytes;
	int ret, rc;

	p = name;
3225
	if (strncasecmp(p, "0x", 2) == 0)
3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245
		p += 2;
	ret = -EINVAL;
	len = strlen(p);
	if (len % 2)
		goto out;
	count = min(len / 2, 16U);
	leading_zero_bytes = 16 - count;
	memset(i_port_id, 0, leading_zero_bytes);
	rc = hex2bin(i_port_id + leading_zero_bytes, p, count);
	if (rc < 0)
		pr_debug("hex2bin failed for srpt_parse_i_port_id: %d\n", rc);
	ret = 0;
out:
	return ret;
}

/*
 * configfs callback function invoked for
 * mkdir /sys/kernel/config/target/$driver/$port/$tpg/acls/$i_port_id
 */
3246
static int srpt_init_nodeacl(struct se_node_acl *se_nacl, const char *name)
3247 3248 3249 3250
{
	u8 i_port_id[16];

	if (srpt_parse_i_port_id(i_port_id, name) < 0) {
3251
		pr_err("invalid initiator port ID %s\n", name);
3252
		return -EINVAL;
3253
	}
3254
	return 0;
3255 3256
}

3257 3258
static ssize_t srpt_tpg_attrib_srp_max_rdma_size_show(struct config_item *item,
		char *page)
3259
{
3260
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
3261 3262 3263 3264 3265
	struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);

	return sprintf(page, "%u\n", sport->port_attrib.srp_max_rdma_size);
}

3266 3267
static ssize_t srpt_tpg_attrib_srp_max_rdma_size_store(struct config_item *item,
		const char *page, size_t count)
3268
{
3269
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
3270 3271 3272 3273
	struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);
	unsigned long val;
	int ret;

3274
	ret = kstrtoul(page, 0, &val);
3275
	if (ret < 0) {
3276
		pr_err("kstrtoul() failed with ret: %d\n", ret);
3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293
		return -EINVAL;
	}
	if (val > MAX_SRPT_RDMA_SIZE) {
		pr_err("val: %lu exceeds MAX_SRPT_RDMA_SIZE: %d\n", val,
			MAX_SRPT_RDMA_SIZE);
		return -EINVAL;
	}
	if (val < DEFAULT_MAX_RDMA_SIZE) {
		pr_err("val: %lu smaller than DEFAULT_MAX_RDMA_SIZE: %d\n",
			val, DEFAULT_MAX_RDMA_SIZE);
		return -EINVAL;
	}
	sport->port_attrib.srp_max_rdma_size = val;

	return count;
}

3294 3295
static ssize_t srpt_tpg_attrib_srp_max_rsp_size_show(struct config_item *item,
		char *page)
3296
{
3297
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
3298 3299 3300 3301 3302
	struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);

	return sprintf(page, "%u\n", sport->port_attrib.srp_max_rsp_size);
}

3303 3304
static ssize_t srpt_tpg_attrib_srp_max_rsp_size_store(struct config_item *item,
		const char *page, size_t count)
3305
{
3306
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
3307 3308 3309 3310
	struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);
	unsigned long val;
	int ret;

3311
	ret = kstrtoul(page, 0, &val);
3312
	if (ret < 0) {
3313
		pr_err("kstrtoul() failed with ret: %d\n", ret);
3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330
		return -EINVAL;
	}
	if (val > MAX_SRPT_RSP_SIZE) {
		pr_err("val: %lu exceeds MAX_SRPT_RSP_SIZE: %d\n", val,
			MAX_SRPT_RSP_SIZE);
		return -EINVAL;
	}
	if (val < MIN_MAX_RSP_SIZE) {
		pr_err("val: %lu smaller than MIN_MAX_RSP_SIZE: %d\n", val,
			MIN_MAX_RSP_SIZE);
		return -EINVAL;
	}
	sport->port_attrib.srp_max_rsp_size = val;

	return count;
}

3331 3332
static ssize_t srpt_tpg_attrib_srp_sq_size_show(struct config_item *item,
		char *page)
3333
{
3334
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
3335 3336 3337 3338 3339
	struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);

	return sprintf(page, "%u\n", sport->port_attrib.srp_sq_size);
}

3340 3341
static ssize_t srpt_tpg_attrib_srp_sq_size_store(struct config_item *item,
		const char *page, size_t count)
3342
{
3343
	struct se_portal_group *se_tpg = attrib_to_tpg(item);
3344 3345 3346 3347
	struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);
	unsigned long val;
	int ret;

3348
	ret = kstrtoul(page, 0, &val);
3349
	if (ret < 0) {
3350
		pr_err("kstrtoul() failed with ret: %d\n", ret);
3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367
		return -EINVAL;
	}
	if (val > MAX_SRPT_SRQ_SIZE) {
		pr_err("val: %lu exceeds MAX_SRPT_SRQ_SIZE: %d\n", val,
			MAX_SRPT_SRQ_SIZE);
		return -EINVAL;
	}
	if (val < MIN_SRPT_SRQ_SIZE) {
		pr_err("val: %lu smaller than MIN_SRPT_SRQ_SIZE: %d\n", val,
			MIN_SRPT_SRQ_SIZE);
		return -EINVAL;
	}
	sport->port_attrib.srp_sq_size = val;

	return count;
}

3368 3369 3370
CONFIGFS_ATTR(srpt_tpg_attrib_,  srp_max_rdma_size);
CONFIGFS_ATTR(srpt_tpg_attrib_,  srp_max_rsp_size);
CONFIGFS_ATTR(srpt_tpg_attrib_,  srp_sq_size);
3371 3372

static struct configfs_attribute *srpt_tpg_attrib_attrs[] = {
3373 3374 3375
	&srpt_tpg_attrib_attr_srp_max_rdma_size,
	&srpt_tpg_attrib_attr_srp_max_rsp_size,
	&srpt_tpg_attrib_attr_srp_sq_size,
3376 3377 3378
	NULL,
};

3379
static ssize_t srpt_tpg_enable_show(struct config_item *item, char *page)
3380
{
3381
	struct se_portal_group *se_tpg = to_tpg(item);
3382 3383 3384 3385 3386
	struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);

	return snprintf(page, PAGE_SIZE, "%d\n", (sport->enabled) ? 1: 0);
}

3387 3388
static ssize_t srpt_tpg_enable_store(struct config_item *item,
		const char *page, size_t count)
3389
{
3390
	struct se_portal_group *se_tpg = to_tpg(item);
3391 3392 3393 3394
	struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);
	unsigned long tmp;
        int ret;

3395
	ret = kstrtoul(page, 0, &tmp);
3396
	if (ret < 0) {
3397
		pr_err("Unable to extract srpt_tpg_store_enable\n");
3398 3399 3400 3401
		return -EINVAL;
	}

	if ((tmp != 0) && (tmp != 1)) {
3402
		pr_err("Illegal value for srpt_tpg_store_enable: %lu\n", tmp);
3403 3404 3405 3406 3407 3408 3409 3410 3411 3412
		return -EINVAL;
	}
	if (tmp == 1)
		sport->enabled = true;
	else
		sport->enabled = false;

	return count;
}

3413
CONFIGFS_ATTR(srpt_tpg_, enable);
3414 3415

static struct configfs_attribute *srpt_tpg_attrs[] = {
3416
	&srpt_tpg_attr_enable,
3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431
	NULL,
};

/**
 * configfs callback invoked for
 * mkdir /sys/kernel/config/target/$driver/$port/$tpg
 */
static struct se_portal_group *srpt_make_tpg(struct se_wwn *wwn,
					     struct config_group *group,
					     const char *name)
{
	struct srpt_port *sport = container_of(wwn, struct srpt_port, port_wwn);
	int res;

	/* Initialize sport->port_wwn and sport->port_tpg_1 */
3432
	res = core_tpg_register(&sport->port_wwn, &sport->port_tpg_1, SCSI_PROTOCOL_SRP);
3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485
	if (res)
		return ERR_PTR(res);

	return &sport->port_tpg_1;
}

/**
 * configfs callback invoked for
 * rmdir /sys/kernel/config/target/$driver/$port/$tpg
 */
static void srpt_drop_tpg(struct se_portal_group *tpg)
{
	struct srpt_port *sport = container_of(tpg,
				struct srpt_port, port_tpg_1);

	sport->enabled = false;
	core_tpg_deregister(&sport->port_tpg_1);
}

/**
 * configfs callback invoked for
 * mkdir /sys/kernel/config/target/$driver/$port
 */
static struct se_wwn *srpt_make_tport(struct target_fabric_configfs *tf,
				      struct config_group *group,
				      const char *name)
{
	struct srpt_port *sport;
	int ret;

	sport = srpt_lookup_port(name);
	pr_debug("make_tport(%s)\n", name);
	ret = -EINVAL;
	if (!sport)
		goto err;

	return &sport->port_wwn;

err:
	return ERR_PTR(ret);
}

/**
 * configfs callback invoked for
 * rmdir /sys/kernel/config/target/$driver/$port
 */
static void srpt_drop_tport(struct se_wwn *wwn)
{
	struct srpt_port *sport = container_of(wwn, struct srpt_port, port_wwn);

	pr_debug("drop_tport(%s\n", config_item_name(&sport->port_wwn.wwn_group.cg_item));
}

3486
static ssize_t srpt_wwn_version_show(struct config_item *item, char *buf)
3487 3488 3489 3490
{
	return scnprintf(buf, PAGE_SIZE, "%s\n", DRV_VERSION);
}

3491
CONFIGFS_ATTR_RO(srpt_wwn_, version);
3492 3493

static struct configfs_attribute *srpt_wwn_attrs[] = {
3494
	&srpt_wwn_attr_version,
3495 3496 3497
	NULL,
};

3498 3499 3500
static const struct target_core_fabric_ops srpt_template = {
	.module				= THIS_MODULE,
	.name				= "srpt",
3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518
	.get_fabric_name		= srpt_get_fabric_name,
	.tpg_get_wwn			= srpt_get_fabric_wwn,
	.tpg_get_tag			= srpt_get_tag,
	.tpg_check_demo_mode		= srpt_check_false,
	.tpg_check_demo_mode_cache	= srpt_check_true,
	.tpg_check_demo_mode_write_protect = srpt_check_true,
	.tpg_check_prod_mode_write_protect = srpt_check_false,
	.tpg_get_inst_index		= srpt_tpg_get_inst_index,
	.release_cmd			= srpt_release_cmd,
	.check_stop_free		= srpt_check_stop_free,
	.shutdown_session		= srpt_shutdown_session,
	.close_session			= srpt_close_session,
	.sess_get_index			= srpt_sess_get_index,
	.sess_get_initiator_sid		= NULL,
	.write_pending			= srpt_write_pending,
	.write_pending_status		= srpt_write_pending_status,
	.set_default_node_attributes	= srpt_set_default_node_attrs,
	.get_cmd_state			= srpt_get_tcm_cmd_state,
3519
	.queue_data_in			= srpt_queue_data_in,
3520
	.queue_status			= srpt_queue_status,
3521
	.queue_tm_rsp			= srpt_queue_tm_rsp,
3522
	.aborted_task			= srpt_aborted_task,
3523 3524 3525 3526 3527 3528 3529 3530
	/*
	 * Setup function pointers for generic logic in
	 * target_core_fabric_configfs.c
	 */
	.fabric_make_wwn		= srpt_make_tport,
	.fabric_drop_wwn		= srpt_drop_tport,
	.fabric_make_tpg		= srpt_make_tpg,
	.fabric_drop_tpg		= srpt_drop_tpg,
3531
	.fabric_init_nodeacl		= srpt_init_nodeacl,
3532 3533 3534 3535

	.tfc_wwn_attrs			= srpt_wwn_attrs,
	.tfc_tpg_base_attrs		= srpt_tpg_attrs,
	.tfc_tpg_attrib_attrs		= srpt_tpg_attrib_attrs,
3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551
};

/**
 * srpt_init_module() - Kernel module initialization.
 *
 * Note: Since ib_register_client() registers callback functions, and since at
 * least one of these callback functions (srpt_add_one()) calls target core
 * functions, this driver must be registered with the target core before
 * ib_register_client() is called.
 */
static int __init srpt_init_module(void)
{
	int ret;

	ret = -EINVAL;
	if (srp_max_req_size < MIN_MAX_REQ_SIZE) {
3552
		pr_err("invalid value %d for kernel module parameter"
3553 3554 3555 3556 3557 3558 3559
		       " srp_max_req_size -- must be at least %d.\n",
		       srp_max_req_size, MIN_MAX_REQ_SIZE);
		goto out;
	}

	if (srpt_srq_size < MIN_SRPT_SRQ_SIZE
	    || srpt_srq_size > MAX_SRPT_SRQ_SIZE) {
3560
		pr_err("invalid value %d for kernel module parameter"
3561 3562 3563 3564 3565
		       " srpt_srq_size -- must be in the range [%d..%d].\n",
		       srpt_srq_size, MIN_SRPT_SRQ_SIZE, MAX_SRPT_SRQ_SIZE);
		goto out;
	}

3566 3567
	ret = target_register_template(&srpt_template);
	if (ret)
3568 3569 3570 3571
		goto out;

	ret = ib_register_client(&srpt_client);
	if (ret) {
3572
		pr_err("couldn't register IB client\n");
3573 3574 3575 3576 3577 3578
		goto out_unregister_target;
	}

	return 0;

out_unregister_target:
3579
	target_unregister_template(&srpt_template);
3580 3581 3582 3583 3584 3585 3586
out:
	return ret;
}

static void __exit srpt_cleanup_module(void)
{
	ib_unregister_client(&srpt_client);
3587
	target_unregister_template(&srpt_template);
3588 3589 3590 3591
}

module_init(srpt_init_module);
module_exit(srpt_cleanup_module);