verbs.c 60.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6
/*
 * Copyright (c) 2004 Mellanox Technologies Ltd.  All rights reserved.
 * Copyright (c) 2004 Infinicon Corporation.  All rights reserved.
 * Copyright (c) 2004 Intel Corporation.  All rights reserved.
 * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
 * Copyright (c) 2004 Voltaire Corporation.  All rights reserved.
7
 * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
8
 * Copyright (c) 2005, 2006 Cisco Systems.  All rights reserved.
L
Linus Torvalds 已提交
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
 *
 * 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/errno.h>
#include <linux/err.h>
41
#include <linux/export.h>
42
#include <linux/string.h>
43
#include <linux/slab.h>
44 45 46
#include <linux/in.h>
#include <linux/in6.h>
#include <net/addrconf.h>
47
#include <linux/security.h>
L
Linus Torvalds 已提交
48

49 50
#include <rdma/ib_verbs.h>
#include <rdma/ib_cache.h>
51
#include <rdma/ib_addr.h>
52
#include <rdma/rw.h>
L
Linus Torvalds 已提交
53

54
#include "core_priv.h"
L
Linus Torvalds 已提交
55

56 57 58
static int ib_resolve_eth_dmac(struct ib_device *device,
			       struct rdma_ah_attr *ah_attr);

59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80
static const char * const ib_events[] = {
	[IB_EVENT_CQ_ERR]		= "CQ error",
	[IB_EVENT_QP_FATAL]		= "QP fatal error",
	[IB_EVENT_QP_REQ_ERR]		= "QP request error",
	[IB_EVENT_QP_ACCESS_ERR]	= "QP access error",
	[IB_EVENT_COMM_EST]		= "communication established",
	[IB_EVENT_SQ_DRAINED]		= "send queue drained",
	[IB_EVENT_PATH_MIG]		= "path migration successful",
	[IB_EVENT_PATH_MIG_ERR]		= "path migration error",
	[IB_EVENT_DEVICE_FATAL]		= "device fatal error",
	[IB_EVENT_PORT_ACTIVE]		= "port active",
	[IB_EVENT_PORT_ERR]		= "port error",
	[IB_EVENT_LID_CHANGE]		= "LID change",
	[IB_EVENT_PKEY_CHANGE]		= "P_key change",
	[IB_EVENT_SM_CHANGE]		= "SM change",
	[IB_EVENT_SRQ_ERR]		= "SRQ error",
	[IB_EVENT_SRQ_LIMIT_REACHED]	= "SRQ limit reached",
	[IB_EVENT_QP_LAST_WQE_REACHED]	= "last WQE reached",
	[IB_EVENT_CLIENT_REREGISTER]	= "client reregister",
	[IB_EVENT_GID_CHANGE]		= "GID changed",
};

81
const char *__attribute_const__ ib_event_msg(enum ib_event_type event)
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114
{
	size_t index = event;

	return (index < ARRAY_SIZE(ib_events) && ib_events[index]) ?
			ib_events[index] : "unrecognized event";
}
EXPORT_SYMBOL(ib_event_msg);

static const char * const wc_statuses[] = {
	[IB_WC_SUCCESS]			= "success",
	[IB_WC_LOC_LEN_ERR]		= "local length error",
	[IB_WC_LOC_QP_OP_ERR]		= "local QP operation error",
	[IB_WC_LOC_EEC_OP_ERR]		= "local EE context operation error",
	[IB_WC_LOC_PROT_ERR]		= "local protection error",
	[IB_WC_WR_FLUSH_ERR]		= "WR flushed",
	[IB_WC_MW_BIND_ERR]		= "memory management operation error",
	[IB_WC_BAD_RESP_ERR]		= "bad response error",
	[IB_WC_LOC_ACCESS_ERR]		= "local access error",
	[IB_WC_REM_INV_REQ_ERR]		= "invalid request error",
	[IB_WC_REM_ACCESS_ERR]		= "remote access error",
	[IB_WC_REM_OP_ERR]		= "remote operation error",
	[IB_WC_RETRY_EXC_ERR]		= "transport retry counter exceeded",
	[IB_WC_RNR_RETRY_EXC_ERR]	= "RNR retry counter exceeded",
	[IB_WC_LOC_RDD_VIOL_ERR]	= "local RDD violation error",
	[IB_WC_REM_INV_RD_REQ_ERR]	= "remote invalid RD request",
	[IB_WC_REM_ABORT_ERR]		= "operation aborted",
	[IB_WC_INV_EECN_ERR]		= "invalid EE context number",
	[IB_WC_INV_EEC_STATE_ERR]	= "invalid EE context state",
	[IB_WC_FATAL_ERR]		= "fatal error",
	[IB_WC_RESP_TIMEOUT_ERR]	= "response timeout error",
	[IB_WC_GENERAL_ERR]		= "general error",
};

115
const char *__attribute_const__ ib_wc_status_msg(enum ib_wc_status status)
116 117 118 119 120 121 122 123
{
	size_t index = status;

	return (index < ARRAY_SIZE(wc_statuses) && wc_statuses[index]) ?
			wc_statuses[index] : "unrecognized status";
}
EXPORT_SYMBOL(ib_wc_status_msg);

124
__attribute_const__ int ib_rate_to_mult(enum ib_rate rate)
125 126
{
	switch (rate) {
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144
	case IB_RATE_2_5_GBPS: return   1;
	case IB_RATE_5_GBPS:   return   2;
	case IB_RATE_10_GBPS:  return   4;
	case IB_RATE_20_GBPS:  return   8;
	case IB_RATE_30_GBPS:  return  12;
	case IB_RATE_40_GBPS:  return  16;
	case IB_RATE_60_GBPS:  return  24;
	case IB_RATE_80_GBPS:  return  32;
	case IB_RATE_120_GBPS: return  48;
	case IB_RATE_14_GBPS:  return   6;
	case IB_RATE_56_GBPS:  return  22;
	case IB_RATE_112_GBPS: return  45;
	case IB_RATE_168_GBPS: return  67;
	case IB_RATE_25_GBPS:  return  10;
	case IB_RATE_100_GBPS: return  40;
	case IB_RATE_200_GBPS: return  80;
	case IB_RATE_300_GBPS: return 120;
	default:	       return  -1;
145 146 147 148
	}
}
EXPORT_SYMBOL(ib_rate_to_mult);

149
__attribute_const__ enum ib_rate mult_to_ib_rate(int mult)
150 151
{
	switch (mult) {
152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169
	case 1:   return IB_RATE_2_5_GBPS;
	case 2:   return IB_RATE_5_GBPS;
	case 4:   return IB_RATE_10_GBPS;
	case 8:   return IB_RATE_20_GBPS;
	case 12:  return IB_RATE_30_GBPS;
	case 16:  return IB_RATE_40_GBPS;
	case 24:  return IB_RATE_60_GBPS;
	case 32:  return IB_RATE_80_GBPS;
	case 48:  return IB_RATE_120_GBPS;
	case 6:   return IB_RATE_14_GBPS;
	case 22:  return IB_RATE_56_GBPS;
	case 45:  return IB_RATE_112_GBPS;
	case 67:  return IB_RATE_168_GBPS;
	case 10:  return IB_RATE_25_GBPS;
	case 40:  return IB_RATE_100_GBPS;
	case 80:  return IB_RATE_200_GBPS;
	case 120: return IB_RATE_300_GBPS;
	default:  return IB_RATE_PORT_CURRENT;
170 171 172 173
	}
}
EXPORT_SYMBOL(mult_to_ib_rate);

174
__attribute_const__ int ib_rate_to_mbps(enum ib_rate rate)
175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198
{
	switch (rate) {
	case IB_RATE_2_5_GBPS: return 2500;
	case IB_RATE_5_GBPS:   return 5000;
	case IB_RATE_10_GBPS:  return 10000;
	case IB_RATE_20_GBPS:  return 20000;
	case IB_RATE_30_GBPS:  return 30000;
	case IB_RATE_40_GBPS:  return 40000;
	case IB_RATE_60_GBPS:  return 60000;
	case IB_RATE_80_GBPS:  return 80000;
	case IB_RATE_120_GBPS: return 120000;
	case IB_RATE_14_GBPS:  return 14062;
	case IB_RATE_56_GBPS:  return 56250;
	case IB_RATE_112_GBPS: return 112500;
	case IB_RATE_168_GBPS: return 168750;
	case IB_RATE_25_GBPS:  return 25781;
	case IB_RATE_100_GBPS: return 103125;
	case IB_RATE_200_GBPS: return 206250;
	case IB_RATE_300_GBPS: return 309375;
	default:	       return -1;
	}
}
EXPORT_SYMBOL(ib_rate_to_mbps);

199
__attribute_const__ enum rdma_transport_type
T
Tom Tucker 已提交
200 201
rdma_node_get_transport(enum rdma_node_type node_type)
{
202 203

	if (node_type == RDMA_NODE_USNIC)
204
		return RDMA_TRANSPORT_USNIC;
205
	if (node_type == RDMA_NODE_USNIC_UDP)
206
		return RDMA_TRANSPORT_USNIC_UDP;
207 208 209 210
	if (node_type == RDMA_NODE_RNIC)
		return RDMA_TRANSPORT_IWARP;

	return RDMA_TRANSPORT_IB;
T
Tom Tucker 已提交
211 212 213
}
EXPORT_SYMBOL(rdma_node_get_transport);

214 215
enum rdma_link_layer rdma_port_get_link_layer(struct ib_device *device, u8 port_num)
{
216
	enum rdma_transport_type lt;
217 218 219
	if (device->get_link_layer)
		return device->get_link_layer(device, port_num);

220 221
	lt = rdma_node_get_transport(device->node_type);
	if (lt == RDMA_TRANSPORT_IB)
222
		return IB_LINK_LAYER_INFINIBAND;
223 224

	return IB_LINK_LAYER_ETHERNET;
225 226 227
}
EXPORT_SYMBOL(rdma_port_get_link_layer);

L
Linus Torvalds 已提交
228 229
/* Protection domains */

230 231 232 233 234 235 236 237 238 239
/**
 * ib_alloc_pd - Allocates an unused protection domain.
 * @device: The device on which to allocate the protection domain.
 *
 * A protection domain object provides an association between QPs, shared
 * receive queues, address handles, memory regions, and memory windows.
 *
 * Every PD has a local_dma_lkey which can be used as the lkey value for local
 * memory operations.
 */
240 241
struct ib_pd *__ib_alloc_pd(struct ib_device *device, unsigned int flags,
		const char *caller)
L
Linus Torvalds 已提交
242 243
{
	struct ib_pd *pd;
244
	int mr_access_flags = 0;
L
Linus Torvalds 已提交
245

246
	pd = device->alloc_pd(device, NULL, NULL);
247 248
	if (IS_ERR(pd))
		return pd;
L
Linus Torvalds 已提交
249

250 251
	pd->device = device;
	pd->uobject = NULL;
252
	pd->__internal_mr = NULL;
253
	atomic_set(&pd->usecnt, 0);
254
	pd->flags = flags;
L
Linus Torvalds 已提交
255

256
	if (device->attrs.device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY)
257
		pd->local_dma_lkey = device->local_dma_lkey;
258 259 260 261 262 263 264 265
	else
		mr_access_flags |= IB_ACCESS_LOCAL_WRITE;

	if (flags & IB_PD_UNSAFE_GLOBAL_RKEY) {
		pr_warn("%s: enabling unsafe global rkey\n", caller);
		mr_access_flags |= IB_ACCESS_REMOTE_READ | IB_ACCESS_REMOTE_WRITE;
	}

266 267 268 269
	pd->res.type = RDMA_RESTRACK_PD;
	pd->res.kern_name = caller;
	rdma_restrack_add(&pd->res);

270
	if (mr_access_flags) {
271 272
		struct ib_mr *mr;

273
		mr = pd->device->get_dma_mr(pd, mr_access_flags);
274 275
		if (IS_ERR(mr)) {
			ib_dealloc_pd(pd);
276
			return ERR_CAST(mr);
277
		}
L
Linus Torvalds 已提交
278

279 280 281 282 283
		mr->device	= pd->device;
		mr->pd		= pd;
		mr->uobject	= NULL;
		mr->need_inval	= false;

284
		pd->__internal_mr = mr;
285 286 287 288 289 290

		if (!(device->attrs.device_cap_flags & IB_DEVICE_LOCAL_DMA_LKEY))
			pd->local_dma_lkey = pd->__internal_mr->lkey;

		if (flags & IB_PD_UNSAFE_GLOBAL_RKEY)
			pd->unsafe_global_rkey = pd->__internal_mr->rkey;
L
Linus Torvalds 已提交
291
	}
292

L
Linus Torvalds 已提交
293 294
	return pd;
}
295
EXPORT_SYMBOL(__ib_alloc_pd);
L
Linus Torvalds 已提交
296

297 298 299 300 301 302 303 304 305
/**
 * ib_dealloc_pd - Deallocates a protection domain.
 * @pd: The protection domain to deallocate.
 *
 * It is an error to call this function while any resources in the pd still
 * exist.  The caller is responsible to synchronously destroy them and
 * guarantee no new allocations will happen.
 */
void ib_dealloc_pd(struct ib_pd *pd)
L
Linus Torvalds 已提交
306
{
307 308
	int ret;

309
	if (pd->__internal_mr) {
310
		ret = pd->device->dereg_mr(pd->__internal_mr);
311
		WARN_ON(ret);
312
		pd->__internal_mr = NULL;
313
	}
L
Linus Torvalds 已提交
314

315 316 317
	/* uverbs manipulates usecnt with proper locking, while the kabi
	   requires the caller to guarantee we can't race here. */
	WARN_ON(atomic_read(&pd->usecnt));
L
Linus Torvalds 已提交
318

319
	rdma_restrack_del(&pd->res);
320 321 322 323
	/* Making delalloc_pd a void return is a WIP, no driver should return
	   an error here. */
	ret = pd->device->dealloc_pd(pd);
	WARN_ONCE(ret, "Infiniband HW driver failed dealloc_pd");
L
Linus Torvalds 已提交
324 325 326 327 328
}
EXPORT_SYMBOL(ib_dealloc_pd);

/* Address handles */

329 330 331
static struct ib_ah *_rdma_create_ah(struct ib_pd *pd,
				     struct rdma_ah_attr *ah_attr,
				     struct ib_udata *udata)
L
Linus Torvalds 已提交
332 333 334
{
	struct ib_ah *ah;

335
	ah = pd->device->create_ah(pd, ah_attr, udata);
L
Linus Torvalds 已提交
336 337

	if (!IS_ERR(ah)) {
338 339 340
		ah->device  = pd->device;
		ah->pd      = pd;
		ah->uobject = NULL;
341
		ah->type    = ah_attr->type;
L
Linus Torvalds 已提交
342 343 344 345 346
		atomic_inc(&pd->usecnt);
	}

	return ah;
}
347 348 349 350 351

struct ib_ah *rdma_create_ah(struct ib_pd *pd, struct rdma_ah_attr *ah_attr)
{
	return _rdma_create_ah(pd, ah_attr, NULL);
}
352
EXPORT_SYMBOL(rdma_create_ah);
L
Linus Torvalds 已提交
353

354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
/**
 * rdma_create_user_ah - Creates an address handle for the
 * given address vector.
 * It resolves destination mac address for ah attribute of RoCE type.
 * @pd: The protection domain associated with the address handle.
 * @ah_attr: The attributes of the address vector.
 * @udata: pointer to user's input output buffer information need by
 *         provider driver.
 *
 * It returns 0 on success and returns appropriate error code on error.
 * The address handle is used to reference a local or global destination
 * in all UD QP post sends.
 */
struct ib_ah *rdma_create_user_ah(struct ib_pd *pd,
				  struct rdma_ah_attr *ah_attr,
				  struct ib_udata *udata)
{
	int err;

	if (ah_attr->type == RDMA_AH_ATTR_TYPE_ROCE) {
		err = ib_resolve_eth_dmac(pd->device, ah_attr);
		if (err)
			return ERR_PTR(err);
	}

	return _rdma_create_ah(pd, ah_attr, udata);
}
EXPORT_SYMBOL(rdma_create_user_ah);

383
int ib_get_rdma_header_version(const union rdma_network_hdr *hdr)
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
{
	const struct iphdr *ip4h = (struct iphdr *)&hdr->roce4grh;
	struct iphdr ip4h_checked;
	const struct ipv6hdr *ip6h = (struct ipv6hdr *)&hdr->ibgrh;

	/* If it's IPv6, the version must be 6, otherwise, the first
	 * 20 bytes (before the IPv4 header) are garbled.
	 */
	if (ip6h->version != 6)
		return (ip4h->version == 4) ? 4 : 0;
	/* version may be 6 or 4 because the first 20 bytes could be garbled */

	/* RoCE v2 requires no options, thus header length
	 * must be 5 words
	 */
	if (ip4h->ihl != 5)
		return 6;

	/* Verify checksum.
	 * We can't write on scattered buffers so we need to copy to
	 * temp buffer.
	 */
	memcpy(&ip4h_checked, ip4h, sizeof(ip4h_checked));
	ip4h_checked.check = 0;
	ip4h_checked.check = ip_fast_csum((u8 *)&ip4h_checked, 5);
	/* if IPv4 header checksum is OK, believe it */
	if (ip4h->check == ip4h_checked.check)
		return 4;
	return 6;
}
414
EXPORT_SYMBOL(ib_get_rdma_header_version);
415 416 417 418 419 420 421 422 423 424

static enum rdma_network_type ib_get_net_type_by_grh(struct ib_device *device,
						     u8 port_num,
						     const struct ib_grh *grh)
{
	int grh_version;

	if (rdma_protocol_ib(device, port_num))
		return RDMA_NETWORK_IB;

425
	grh_version = ib_get_rdma_header_version((union rdma_network_hdr *)grh);
426 427 428 429 430 431 432 433 434 435

	if (grh_version == 4)
		return RDMA_NETWORK_IPV4;

	if (grh->next_hdr == IPPROTO_UDP)
		return RDMA_NETWORK_IPV6;

	return RDMA_NETWORK_ROCE_V1;
}

436 437
struct find_gid_index_context {
	u16 vlan_id;
438
	enum ib_gid_type gid_type;
439 440 441 442 443 444
};

static bool find_gid_index(const union ib_gid *gid,
			   const struct ib_gid_attr *gid_attr,
			   void *context)
{
445
	struct find_gid_index_context *ctx = context;
446

447 448 449
	if (ctx->gid_type != gid_attr->gid_type)
		return false;

450 451 452 453 454 455 456 457 458 459
	if ((!!(ctx->vlan_id != 0xffff) == !is_vlan_dev(gid_attr->ndev)) ||
	    (is_vlan_dev(gid_attr->ndev) &&
	     vlan_dev_vlan_id(gid_attr->ndev) != ctx->vlan_id))
		return false;

	return true;
}

static int get_sgid_index_from_eth(struct ib_device *device, u8 port_num,
				   u16 vlan_id, const union ib_gid *sgid,
460
				   enum ib_gid_type gid_type,
461 462
				   u16 *gid_index)
{
463 464
	struct find_gid_index_context context = {.vlan_id = vlan_id,
						 .gid_type = gid_type};
465 466 467 468 469

	return ib_find_gid_by_filter(device, sgid, port_num, find_gid_index,
				     &context, gid_index);
}

470 471 472
int ib_get_gids_from_rdma_hdr(const union rdma_network_hdr *hdr,
			      enum rdma_network_type net_type,
			      union ib_gid *sgid, union ib_gid *dgid)
473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501
{
	struct sockaddr_in  src_in;
	struct sockaddr_in  dst_in;
	__be32 src_saddr, dst_saddr;

	if (!sgid || !dgid)
		return -EINVAL;

	if (net_type == RDMA_NETWORK_IPV4) {
		memcpy(&src_in.sin_addr.s_addr,
		       &hdr->roce4grh.saddr, 4);
		memcpy(&dst_in.sin_addr.s_addr,
		       &hdr->roce4grh.daddr, 4);
		src_saddr = src_in.sin_addr.s_addr;
		dst_saddr = dst_in.sin_addr.s_addr;
		ipv6_addr_set_v4mapped(src_saddr,
				       (struct in6_addr *)sgid);
		ipv6_addr_set_v4mapped(dst_saddr,
				       (struct in6_addr *)dgid);
		return 0;
	} else if (net_type == RDMA_NETWORK_IPV6 ||
		   net_type == RDMA_NETWORK_IB) {
		*dgid = hdr->ibgrh.dgid;
		*sgid = hdr->ibgrh.sgid;
		return 0;
	} else {
		return -EINVAL;
	}
}
502
EXPORT_SYMBOL(ib_get_gids_from_rdma_hdr);
503

504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528
/* Resolve destination mac address and hop limit for unicast destination
 * GID entry, considering the source GID entry as well.
 * ah_attribute must have have valid port_num, sgid_index.
 */
static int ib_resolve_unicast_gid_dmac(struct ib_device *device,
				       struct rdma_ah_attr *ah_attr)
{
	struct ib_gid_attr sgid_attr;
	struct ib_global_route *grh;
	int hop_limit = 0xff;
	union ib_gid sgid;
	int ret;

	grh = rdma_ah_retrieve_grh(ah_attr);

	ret = ib_query_gid(device,
			   rdma_ah_get_port_num(ah_attr),
			   grh->sgid_index,
			   &sgid, &sgid_attr);
	if (ret || !sgid_attr.ndev) {
		if (!ret)
			ret = -ENXIO;
		return ret;
	}

529 530 531 532 533 534 535 536 537 538
	/* If destination is link local and source GID is RoCEv1,
	 * IP stack is not used.
	 */
	if (rdma_link_local_addr((struct in6_addr *)grh->dgid.raw) &&
	    sgid_attr.gid_type == IB_GID_TYPE_ROCE) {
		rdma_get_ll_mac((struct in6_addr *)grh->dgid.raw,
				ah_attr->roce.dmac);
		goto done;
	}

539 540 541
	ret = rdma_addr_find_l2_eth_by_grh(&sgid, &grh->dgid,
					   ah_attr->roce.dmac,
					   sgid_attr.ndev, &hop_limit);
542
done:
543 544 545 546 547 548
	dev_put(sgid_attr.ndev);

	grh->hop_limit = hop_limit;
	return ret;
}

549
/*
550
 * This function initializes address handle attributes from the incoming packet.
551 552 553 554 555 556 557 558
 * Incoming packet has dgid of the receiver node on which this code is
 * getting executed and, sgid contains the GID of the sender.
 *
 * When resolving mac address of destination, the arrived dgid is used
 * as sgid and, sgid is used as dgid because sgid contains destinations
 * GID whom to respond to.
 *
 */
559 560 561
int ib_init_ah_attr_from_wc(struct ib_device *device, u8 port_num,
			    const struct ib_wc *wc, const struct ib_grh *grh,
			    struct rdma_ah_attr *ah_attr)
562 563 564 565
{
	u32 flow_class;
	u16 gid_index;
	int ret;
566 567
	enum rdma_network_type net_type = RDMA_NETWORK_IB;
	enum ib_gid_type gid_type = IB_GID_TYPE_IB;
568
	int hoplimit = 0xff;
569 570
	union ib_gid dgid;
	union ib_gid sgid;
571

572 573
	might_sleep();

S
Sean Hefty 已提交
574
	memset(ah_attr, 0, sizeof *ah_attr);
575
	ah_attr->type = rdma_ah_find_type(device, port_num);
576
	if (rdma_cap_eth_ah(device, port_num)) {
577 578 579 580 581 582
		if (wc->wc_flags & IB_WC_WITH_NETWORK_HDR_TYPE)
			net_type = wc->network_hdr_type;
		else
			net_type = ib_get_net_type_by_grh(device, port_num, grh);
		gid_type = ib_network_to_gid_type(net_type);
	}
583 584
	ret = ib_get_gids_from_rdma_hdr((union rdma_network_hdr *)grh, net_type,
					&sgid, &dgid);
585 586 587
	if (ret)
		return ret;

588 589 590
	rdma_ah_set_sl(ah_attr, wc->sl);
	rdma_ah_set_port_num(ah_attr, port_num);

591
	if (rdma_protocol_roce(device, port_num)) {
592 593 594
		u16 vlan_id = wc->wc_flags & IB_WC_WITH_VLAN ?
				wc->vlan_id : 0xffff;

595 596 597
		if (!(wc->wc_flags & IB_WC_GRH))
			return -EPROTOTYPE;

598 599 600
		ret = get_sgid_index_from_eth(device, port_num,
					      vlan_id, &dgid,
					      gid_type, &gid_index);
601 602
		if (ret)
			return ret;
603

604 605 606 607 608 609 610 611 612
		flow_class = be32_to_cpu(grh->version_tclass_flow);
		rdma_ah_set_grh(ah_attr, &sgid,
				flow_class & 0xFFFFF,
				(u8)gid_index, hoplimit,
				(flow_class >> 20) & 0xFF);
		return ib_resolve_unicast_gid_dmac(device, ah_attr);
	} else {
		rdma_ah_set_dlid(ah_attr, wc->slid);
		rdma_ah_set_path_bits(ah_attr, wc->dlid_path_bits);
613

614
		if (wc->wc_flags & IB_WC_GRH) {
615 616 617 618 619 620 621 622 623 624
			if (dgid.global.interface_id != cpu_to_be64(IB_SA_WELL_KNOWN_GUID)) {
				ret = ib_find_cached_gid_by_port(device, &dgid,
								 IB_GID_TYPE_IB,
								 port_num, NULL,
								 &gid_index);
				if (ret)
					return ret;
			} else {
				gid_index = 0;
			}
625

626 627 628 629 630 631 632
			flow_class = be32_to_cpu(grh->version_tclass_flow);
			rdma_ah_set_grh(ah_attr, &sgid,
					flow_class & 0xFFFFF,
					(u8)gid_index, hoplimit,
					(flow_class >> 20) & 0xFF);
		}
		return 0;
633
	}
S
Sean Hefty 已提交
634
}
635
EXPORT_SYMBOL(ib_init_ah_attr_from_wc);
S
Sean Hefty 已提交
636

637 638
struct ib_ah *ib_create_ah_from_wc(struct ib_pd *pd, const struct ib_wc *wc,
				   const struct ib_grh *grh, u8 port_num)
S
Sean Hefty 已提交
639
{
640
	struct rdma_ah_attr ah_attr;
S
Sean Hefty 已提交
641 642
	int ret;

643
	ret = ib_init_ah_attr_from_wc(pd->device, port_num, wc, grh, &ah_attr);
S
Sean Hefty 已提交
644 645
	if (ret)
		return ERR_PTR(ret);
646

647
	return rdma_create_ah(pd, &ah_attr);
648 649 650
}
EXPORT_SYMBOL(ib_create_ah_from_wc);

651
int rdma_modify_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr)
L
Linus Torvalds 已提交
652
{
653 654 655
	if (ah->type != ah_attr->type)
		return -EINVAL;

L
Linus Torvalds 已提交
656 657
	return ah->device->modify_ah ?
		ah->device->modify_ah(ah, ah_attr) :
658
		-EOPNOTSUPP;
L
Linus Torvalds 已提交
659
}
660
EXPORT_SYMBOL(rdma_modify_ah);
L
Linus Torvalds 已提交
661

662
int rdma_query_ah(struct ib_ah *ah, struct rdma_ah_attr *ah_attr)
L
Linus Torvalds 已提交
663 664 665
{
	return ah->device->query_ah ?
		ah->device->query_ah(ah, ah_attr) :
666
		-EOPNOTSUPP;
L
Linus Torvalds 已提交
667
}
668
EXPORT_SYMBOL(rdma_query_ah);
L
Linus Torvalds 已提交
669

670
int rdma_destroy_ah(struct ib_ah *ah)
L
Linus Torvalds 已提交
671 672 673 674 675 676 677 678 679 680 681
{
	struct ib_pd *pd;
	int ret;

	pd = ah->pd;
	ret = ah->device->destroy_ah(ah);
	if (!ret)
		atomic_dec(&pd->usecnt);

	return ret;
}
682
EXPORT_SYMBOL(rdma_destroy_ah);
L
Linus Torvalds 已提交
683

684 685 686 687 688 689 690 691
/* Shared receive queues */

struct ib_srq *ib_create_srq(struct ib_pd *pd,
			     struct ib_srq_init_attr *srq_init_attr)
{
	struct ib_srq *srq;

	if (!pd->device->create_srq)
692
		return ERR_PTR(-EOPNOTSUPP);
693 694 695 696 697 698 699 700 701

	srq = pd->device->create_srq(pd, srq_init_attr, NULL);

	if (!IS_ERR(srq)) {
		srq->device    	   = pd->device;
		srq->pd        	   = pd;
		srq->uobject       = NULL;
		srq->event_handler = srq_init_attr->event_handler;
		srq->srq_context   = srq_init_attr->srq_context;
S
Sean Hefty 已提交
702
		srq->srq_type      = srq_init_attr->srq_type;
703 704 705 706
		if (ib_srq_has_cq(srq->srq_type)) {
			srq->ext.cq   = srq_init_attr->ext.cq;
			atomic_inc(&srq->ext.cq->usecnt);
		}
S
Sean Hefty 已提交
707 708 709 710
		if (srq->srq_type == IB_SRQT_XRC) {
			srq->ext.xrc.xrcd = srq_init_attr->ext.xrc.xrcd;
			atomic_inc(&srq->ext.xrc.xrcd->usecnt);
		}
711 712 713 714 715 716 717 718 719 720 721 722
		atomic_inc(&pd->usecnt);
		atomic_set(&srq->usecnt, 0);
	}

	return srq;
}
EXPORT_SYMBOL(ib_create_srq);

int ib_modify_srq(struct ib_srq *srq,
		  struct ib_srq_attr *srq_attr,
		  enum ib_srq_attr_mask srq_attr_mask)
{
723 724
	return srq->device->modify_srq ?
		srq->device->modify_srq(srq, srq_attr, srq_attr_mask, NULL) :
725
		-EOPNOTSUPP;
726 727 728 729 730 731 732
}
EXPORT_SYMBOL(ib_modify_srq);

int ib_query_srq(struct ib_srq *srq,
		 struct ib_srq_attr *srq_attr)
{
	return srq->device->query_srq ?
733
		srq->device->query_srq(srq, srq_attr) : -EOPNOTSUPP;
734 735 736 737 738 739
}
EXPORT_SYMBOL(ib_query_srq);

int ib_destroy_srq(struct ib_srq *srq)
{
	struct ib_pd *pd;
S
Sean Hefty 已提交
740 741 742
	enum ib_srq_type srq_type;
	struct ib_xrcd *uninitialized_var(xrcd);
	struct ib_cq *uninitialized_var(cq);
743 744 745 746 747 748
	int ret;

	if (atomic_read(&srq->usecnt))
		return -EBUSY;

	pd = srq->pd;
S
Sean Hefty 已提交
749
	srq_type = srq->srq_type;
750 751 752
	if (ib_srq_has_cq(srq_type))
		cq = srq->ext.cq;
	if (srq_type == IB_SRQT_XRC)
S
Sean Hefty 已提交
753
		xrcd = srq->ext.xrc.xrcd;
754 755

	ret = srq->device->destroy_srq(srq);
S
Sean Hefty 已提交
756
	if (!ret) {
757
		atomic_dec(&pd->usecnt);
758
		if (srq_type == IB_SRQT_XRC)
S
Sean Hefty 已提交
759
			atomic_dec(&xrcd->usecnt);
760
		if (ib_srq_has_cq(srq_type))
S
Sean Hefty 已提交
761 762
			atomic_dec(&cq->usecnt);
	}
763 764 765 766 767

	return ret;
}
EXPORT_SYMBOL(ib_destroy_srq);

L
Linus Torvalds 已提交
768 769
/* Queue pairs */

770 771 772
static void __ib_shared_qp_event_handler(struct ib_event *event, void *context)
{
	struct ib_qp *qp = context;
773
	unsigned long flags;
774

775
	spin_lock_irqsave(&qp->device->event_handler_lock, flags);
776
	list_for_each_entry(event->element.qp, &qp->open_list, open_list)
777 778
		if (event->element.qp->event_handler)
			event->element.qp->event_handler(event, event->element.qp->qp_context);
779
	spin_unlock_irqrestore(&qp->device->event_handler_lock, flags);
780 781
}

782 783 784 785 786 787 788
static void __ib_insert_xrcd_qp(struct ib_xrcd *xrcd, struct ib_qp *qp)
{
	mutex_lock(&xrcd->tgt_qp_mutex);
	list_add(&qp->xrcd_list, &xrcd->tgt_qp_list);
	mutex_unlock(&xrcd->tgt_qp_mutex);
}

789 790 791
static struct ib_qp *__ib_open_qp(struct ib_qp *real_qp,
				  void (*event_handler)(struct ib_event *, void *),
				  void *qp_context)
792
{
793 794
	struct ib_qp *qp;
	unsigned long flags;
795
	int err;
796 797 798 799 800

	qp = kzalloc(sizeof *qp, GFP_KERNEL);
	if (!qp)
		return ERR_PTR(-ENOMEM);

801 802 803 804 805 806 807
	qp->real_qp = real_qp;
	err = ib_open_shared_qp_security(qp, real_qp->device);
	if (err) {
		kfree(qp);
		return ERR_PTR(err);
	}

808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831
	qp->real_qp = real_qp;
	atomic_inc(&real_qp->usecnt);
	qp->device = real_qp->device;
	qp->event_handler = event_handler;
	qp->qp_context = qp_context;
	qp->qp_num = real_qp->qp_num;
	qp->qp_type = real_qp->qp_type;

	spin_lock_irqsave(&real_qp->device->event_handler_lock, flags);
	list_add(&qp->open_list, &real_qp->open_list);
	spin_unlock_irqrestore(&real_qp->device->event_handler_lock, flags);

	return qp;
}

struct ib_qp *ib_open_qp(struct ib_xrcd *xrcd,
			 struct ib_qp_open_attr *qp_open_attr)
{
	struct ib_qp *qp, *real_qp;

	if (qp_open_attr->qp_type != IB_QPT_XRC_TGT)
		return ERR_PTR(-EINVAL);

	qp = ERR_PTR(-EINVAL);
832
	mutex_lock(&xrcd->tgt_qp_mutex);
833 834 835 836 837 838 839
	list_for_each_entry(real_qp, &xrcd->tgt_qp_list, xrcd_list) {
		if (real_qp->qp_num == qp_open_attr->qp_num) {
			qp = __ib_open_qp(real_qp, qp_open_attr->event_handler,
					  qp_open_attr->qp_context);
			break;
		}
	}
840
	mutex_unlock(&xrcd->tgt_qp_mutex);
841
	return qp;
842
}
843
EXPORT_SYMBOL(ib_open_qp);
844

845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867
static struct ib_qp *ib_create_xrc_qp(struct ib_qp *qp,
		struct ib_qp_init_attr *qp_init_attr)
{
	struct ib_qp *real_qp = qp;

	qp->event_handler = __ib_shared_qp_event_handler;
	qp->qp_context = qp;
	qp->pd = NULL;
	qp->send_cq = qp->recv_cq = NULL;
	qp->srq = NULL;
	qp->xrcd = qp_init_attr->xrcd;
	atomic_inc(&qp_init_attr->xrcd->usecnt);
	INIT_LIST_HEAD(&qp->open_list);

	qp = __ib_open_qp(real_qp, qp_init_attr->event_handler,
			  qp_init_attr->qp_context);
	if (!IS_ERR(qp))
		__ib_insert_xrcd_qp(qp_init_attr->xrcd, real_qp);
	else
		real_qp->device->destroy_qp(real_qp);
	return qp;
}

L
Linus Torvalds 已提交
868 869 870
struct ib_qp *ib_create_qp(struct ib_pd *pd,
			   struct ib_qp_init_attr *qp_init_attr)
{
871 872
	struct ib_device *device = pd ? pd->device : qp_init_attr->xrcd->device;
	struct ib_qp *qp;
873 874
	int ret;

875 876 877 878 879 880
	if (qp_init_attr->rwq_ind_tbl &&
	    (qp_init_attr->recv_cq ||
	    qp_init_attr->srq || qp_init_attr->cap.max_recv_wr ||
	    qp_init_attr->cap.max_recv_sge))
		return ERR_PTR(-EINVAL);

881 882 883 884 885 886 887 888
	/*
	 * If the callers is using the RDMA API calculate the resources
	 * needed for the RDMA READ/WRITE operations.
	 *
	 * Note that these callers need to pass in a port number.
	 */
	if (qp_init_attr->cap.max_rdma_ctxs)
		rdma_rw_init_qp(device, qp_init_attr);
L
Linus Torvalds 已提交
889

890
	qp = _ib_create_qp(device, pd, qp_init_attr, NULL, NULL);
891 892 893
	if (IS_ERR(qp))
		return qp;

894 895 896 897 898 899
	ret = ib_create_qp_security(qp, device);
	if (ret) {
		ib_destroy_qp(qp);
		return ERR_PTR(ret);
	}

900 901
	qp->real_qp    = qp;
	qp->qp_type    = qp_init_attr->qp_type;
902
	qp->rwq_ind_tbl = qp_init_attr->rwq_ind_tbl;
903 904

	atomic_set(&qp->usecnt, 0);
905 906
	qp->mrs_used = 0;
	spin_lock_init(&qp->mr_lock);
907
	INIT_LIST_HEAD(&qp->rdma_mrs);
908
	INIT_LIST_HEAD(&qp->sig_mrs);
909
	qp->port = 0;
910

911 912 913 914 915 916 917 918 919 920
	if (qp_init_attr->qp_type == IB_QPT_XRC_TGT)
		return ib_create_xrc_qp(qp, qp_init_attr);

	qp->event_handler = qp_init_attr->event_handler;
	qp->qp_context = qp_init_attr->qp_context;
	if (qp_init_attr->qp_type == IB_QPT_XRC_INI) {
		qp->recv_cq = NULL;
		qp->srq = NULL;
	} else {
		qp->recv_cq = qp_init_attr->recv_cq;
921 922
		if (qp_init_attr->recv_cq)
			atomic_inc(&qp_init_attr->recv_cq->usecnt);
923 924 925
		qp->srq = qp_init_attr->srq;
		if (qp->srq)
			atomic_inc(&qp_init_attr->srq->usecnt);
L
Linus Torvalds 已提交
926 927
	}

928 929 930 931
	qp->send_cq = qp_init_attr->send_cq;
	qp->xrcd    = NULL;

	atomic_inc(&pd->usecnt);
932 933 934 935
	if (qp_init_attr->send_cq)
		atomic_inc(&qp_init_attr->send_cq->usecnt);
	if (qp_init_attr->rwq_ind_tbl)
		atomic_inc(&qp->rwq_ind_tbl->usecnt);
936 937 938 939 940 941

	if (qp_init_attr->cap.max_rdma_ctxs) {
		ret = rdma_rw_init_mrs(qp, qp_init_attr);
		if (ret) {
			pr_err("failed to init MR pool ret= %d\n", ret);
			ib_destroy_qp(qp);
942
			return ERR_PTR(ret);
943 944 945
		}
	}

946 947 948 949 950 951 952 953 954
	/*
	 * Note: all hw drivers guarantee that max_send_sge is lower than
	 * the device RDMA WRITE SGE limit but not all hw drivers ensure that
	 * max_send_sge <= max_sge_rd.
	 */
	qp->max_write_sge = qp_init_attr->cap.max_send_sge;
	qp->max_read_sge = min_t(u32, qp_init_attr->cap.max_send_sge,
				 device->attrs.max_sge_rd);

L
Linus Torvalds 已提交
955 956 957 958
	return qp;
}
EXPORT_SYMBOL(ib_create_qp);

959 960
static const struct {
	int			valid;
S
Sean Hefty 已提交
961 962
	enum ib_qp_attr_mask	req_param[IB_QPT_MAX];
	enum ib_qp_attr_mask	opt_param[IB_QPT_MAX];
963 964 965 966 967 968 969 970 971
} qp_state_table[IB_QPS_ERR + 1][IB_QPS_ERR + 1] = {
	[IB_QPS_RESET] = {
		[IB_QPS_RESET] = { .valid = 1 },
		[IB_QPS_INIT]  = {
			.valid = 1,
			.req_param = {
				[IB_QPT_UD]  = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_QKEY),
O
Or Gerlitz 已提交
972
				[IB_QPT_RAW_PACKET] = IB_QP_PORT,
973 974 975 976 977 978
				[IB_QPT_UC]  = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_ACCESS_FLAGS),
				[IB_QPT_RC]  = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_ACCESS_FLAGS),
S
Sean Hefty 已提交
979 980 981 982 983 984
				[IB_QPT_XRC_INI] = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_ACCESS_FLAGS),
				[IB_QPT_XRC_TGT] = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_ACCESS_FLAGS),
985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006
				[IB_QPT_SMI] = (IB_QP_PKEY_INDEX		|
						IB_QP_QKEY),
				[IB_QPT_GSI] = (IB_QP_PKEY_INDEX		|
						IB_QP_QKEY),
			}
		},
	},
	[IB_QPS_INIT]  = {
		[IB_QPS_RESET] = { .valid = 1 },
		[IB_QPS_ERR] =   { .valid = 1 },
		[IB_QPS_INIT]  = {
			.valid = 1,
			.opt_param = {
				[IB_QPT_UD]  = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_QKEY),
				[IB_QPT_UC]  = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_ACCESS_FLAGS),
				[IB_QPT_RC]  = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_ACCESS_FLAGS),
S
Sean Hefty 已提交
1007 1008 1009 1010 1011 1012
				[IB_QPT_XRC_INI] = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_ACCESS_FLAGS),
				[IB_QPT_XRC_TGT] = (IB_QP_PKEY_INDEX		|
						IB_QP_PORT			|
						IB_QP_ACCESS_FLAGS),
1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
				[IB_QPT_SMI] = (IB_QP_PKEY_INDEX		|
						IB_QP_QKEY),
				[IB_QPT_GSI] = (IB_QP_PKEY_INDEX		|
						IB_QP_QKEY),
			}
		},
		[IB_QPS_RTR]   = {
			.valid = 1,
			.req_param = {
				[IB_QPT_UC]  = (IB_QP_AV			|
						IB_QP_PATH_MTU			|
						IB_QP_DEST_QPN			|
						IB_QP_RQ_PSN),
				[IB_QPT_RC]  = (IB_QP_AV			|
						IB_QP_PATH_MTU			|
						IB_QP_DEST_QPN			|
						IB_QP_RQ_PSN			|
						IB_QP_MAX_DEST_RD_ATOMIC	|
						IB_QP_MIN_RNR_TIMER),
S
Sean Hefty 已提交
1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
				[IB_QPT_XRC_INI] = (IB_QP_AV			|
						IB_QP_PATH_MTU			|
						IB_QP_DEST_QPN			|
						IB_QP_RQ_PSN),
				[IB_QPT_XRC_TGT] = (IB_QP_AV			|
						IB_QP_PATH_MTU			|
						IB_QP_DEST_QPN			|
						IB_QP_RQ_PSN			|
						IB_QP_MAX_DEST_RD_ATOMIC	|
						IB_QP_MIN_RNR_TIMER),
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051
			},
			.opt_param = {
				 [IB_QPT_UD]  = (IB_QP_PKEY_INDEX		|
						 IB_QP_QKEY),
				 [IB_QPT_UC]  = (IB_QP_ALT_PATH			|
						 IB_QP_ACCESS_FLAGS		|
						 IB_QP_PKEY_INDEX),
				 [IB_QPT_RC]  = (IB_QP_ALT_PATH			|
						 IB_QP_ACCESS_FLAGS		|
						 IB_QP_PKEY_INDEX),
S
Sean Hefty 已提交
1052 1053 1054 1055 1056 1057
				 [IB_QPT_XRC_INI] = (IB_QP_ALT_PATH		|
						 IB_QP_ACCESS_FLAGS		|
						 IB_QP_PKEY_INDEX),
				 [IB_QPT_XRC_TGT] = (IB_QP_ALT_PATH		|
						 IB_QP_ACCESS_FLAGS		|
						 IB_QP_PKEY_INDEX),
1058 1059 1060 1061
				 [IB_QPT_SMI] = (IB_QP_PKEY_INDEX		|
						 IB_QP_QKEY),
				 [IB_QPT_GSI] = (IB_QP_PKEY_INDEX		|
						 IB_QP_QKEY),
1062
			 },
1063
		},
1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
	},
	[IB_QPS_RTR]   = {
		[IB_QPS_RESET] = { .valid = 1 },
		[IB_QPS_ERR] =   { .valid = 1 },
		[IB_QPS_RTS]   = {
			.valid = 1,
			.req_param = {
				[IB_QPT_UD]  = IB_QP_SQ_PSN,
				[IB_QPT_UC]  = IB_QP_SQ_PSN,
				[IB_QPT_RC]  = (IB_QP_TIMEOUT			|
						IB_QP_RETRY_CNT			|
						IB_QP_RNR_RETRY			|
						IB_QP_SQ_PSN			|
						IB_QP_MAX_QP_RD_ATOMIC),
S
Sean Hefty 已提交
1078 1079 1080 1081 1082 1083 1084
				[IB_QPT_XRC_INI] = (IB_QP_TIMEOUT		|
						IB_QP_RETRY_CNT			|
						IB_QP_RNR_RETRY			|
						IB_QP_SQ_PSN			|
						IB_QP_MAX_QP_RD_ATOMIC),
				[IB_QPT_XRC_TGT] = (IB_QP_TIMEOUT		|
						IB_QP_SQ_PSN),
1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
				[IB_QPT_SMI] = IB_QP_SQ_PSN,
				[IB_QPT_GSI] = IB_QP_SQ_PSN,
			},
			.opt_param = {
				 [IB_QPT_UD]  = (IB_QP_CUR_STATE		|
						 IB_QP_QKEY),
				 [IB_QPT_UC]  = (IB_QP_CUR_STATE		|
						 IB_QP_ALT_PATH			|
						 IB_QP_ACCESS_FLAGS		|
						 IB_QP_PATH_MIG_STATE),
				 [IB_QPT_RC]  = (IB_QP_CUR_STATE		|
						 IB_QP_ALT_PATH			|
						 IB_QP_ACCESS_FLAGS		|
						 IB_QP_MIN_RNR_TIMER		|
						 IB_QP_PATH_MIG_STATE),
S
Sean Hefty 已提交
1100 1101 1102 1103 1104 1105 1106 1107 1108
				 [IB_QPT_XRC_INI] = (IB_QP_CUR_STATE		|
						 IB_QP_ALT_PATH			|
						 IB_QP_ACCESS_FLAGS		|
						 IB_QP_PATH_MIG_STATE),
				 [IB_QPT_XRC_TGT] = (IB_QP_CUR_STATE		|
						 IB_QP_ALT_PATH			|
						 IB_QP_ACCESS_FLAGS		|
						 IB_QP_MIN_RNR_TIMER		|
						 IB_QP_PATH_MIG_STATE),
1109 1110 1111 1112
				 [IB_QPT_SMI] = (IB_QP_CUR_STATE		|
						 IB_QP_QKEY),
				 [IB_QPT_GSI] = (IB_QP_CUR_STATE		|
						 IB_QP_QKEY),
1113
				 [IB_QPT_RAW_PACKET] = IB_QP_RATE_LIMIT,
1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
			 }
		}
	},
	[IB_QPS_RTS]   = {
		[IB_QPS_RESET] = { .valid = 1 },
		[IB_QPS_ERR] =   { .valid = 1 },
		[IB_QPS_RTS]   = {
			.valid = 1,
			.opt_param = {
				[IB_QPT_UD]  = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
1125 1126
				[IB_QPT_UC]  = (IB_QP_CUR_STATE			|
						IB_QP_ACCESS_FLAGS		|
1127 1128
						IB_QP_ALT_PATH			|
						IB_QP_PATH_MIG_STATE),
1129 1130
				[IB_QPT_RC]  = (IB_QP_CUR_STATE			|
						IB_QP_ACCESS_FLAGS		|
1131 1132 1133
						IB_QP_ALT_PATH			|
						IB_QP_PATH_MIG_STATE		|
						IB_QP_MIN_RNR_TIMER),
S
Sean Hefty 已提交
1134 1135 1136 1137 1138 1139 1140 1141 1142
				[IB_QPT_XRC_INI] = (IB_QP_CUR_STATE		|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_ALT_PATH			|
						IB_QP_PATH_MIG_STATE),
				[IB_QPT_XRC_TGT] = (IB_QP_CUR_STATE		|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_ALT_PATH			|
						IB_QP_PATH_MIG_STATE		|
						IB_QP_MIN_RNR_TIMER),
1143 1144 1145 1146
				[IB_QPT_SMI] = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
				[IB_QPT_GSI] = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
1147
				[IB_QPT_RAW_PACKET] = IB_QP_RATE_LIMIT,
1148 1149 1150 1151 1152 1153 1154 1155
			}
		},
		[IB_QPS_SQD]   = {
			.valid = 1,
			.opt_param = {
				[IB_QPT_UD]  = IB_QP_EN_SQD_ASYNC_NOTIFY,
				[IB_QPT_UC]  = IB_QP_EN_SQD_ASYNC_NOTIFY,
				[IB_QPT_RC]  = IB_QP_EN_SQD_ASYNC_NOTIFY,
S
Sean Hefty 已提交
1156 1157
				[IB_QPT_XRC_INI] = IB_QP_EN_SQD_ASYNC_NOTIFY,
				[IB_QPT_XRC_TGT] = IB_QP_EN_SQD_ASYNC_NOTIFY, /* ??? */
1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179
				[IB_QPT_SMI] = IB_QP_EN_SQD_ASYNC_NOTIFY,
				[IB_QPT_GSI] = IB_QP_EN_SQD_ASYNC_NOTIFY
			}
		},
	},
	[IB_QPS_SQD]   = {
		[IB_QPS_RESET] = { .valid = 1 },
		[IB_QPS_ERR] =   { .valid = 1 },
		[IB_QPS_RTS]   = {
			.valid = 1,
			.opt_param = {
				[IB_QPT_UD]  = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
				[IB_QPT_UC]  = (IB_QP_CUR_STATE			|
						IB_QP_ALT_PATH			|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_PATH_MIG_STATE),
				[IB_QPT_RC]  = (IB_QP_CUR_STATE			|
						IB_QP_ALT_PATH			|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_MIN_RNR_TIMER		|
						IB_QP_PATH_MIG_STATE),
S
Sean Hefty 已提交
1180 1181 1182 1183 1184 1185 1186 1187 1188
				[IB_QPT_XRC_INI] = (IB_QP_CUR_STATE		|
						IB_QP_ALT_PATH			|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_PATH_MIG_STATE),
				[IB_QPT_XRC_TGT] = (IB_QP_CUR_STATE		|
						IB_QP_ALT_PATH			|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_MIN_RNR_TIMER		|
						IB_QP_PATH_MIG_STATE),
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216
				[IB_QPT_SMI] = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
				[IB_QPT_GSI] = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
			}
		},
		[IB_QPS_SQD]   = {
			.valid = 1,
			.opt_param = {
				[IB_QPT_UD]  = (IB_QP_PKEY_INDEX		|
						IB_QP_QKEY),
				[IB_QPT_UC]  = (IB_QP_AV			|
						IB_QP_ALT_PATH			|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_PKEY_INDEX		|
						IB_QP_PATH_MIG_STATE),
				[IB_QPT_RC]  = (IB_QP_PORT			|
						IB_QP_AV			|
						IB_QP_TIMEOUT			|
						IB_QP_RETRY_CNT			|
						IB_QP_RNR_RETRY			|
						IB_QP_MAX_QP_RD_ATOMIC		|
						IB_QP_MAX_DEST_RD_ATOMIC	|
						IB_QP_ALT_PATH			|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_PKEY_INDEX		|
						IB_QP_MIN_RNR_TIMER		|
						IB_QP_PATH_MIG_STATE),
S
Sean Hefty 已提交
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
				[IB_QPT_XRC_INI] = (IB_QP_PORT			|
						IB_QP_AV			|
						IB_QP_TIMEOUT			|
						IB_QP_RETRY_CNT			|
						IB_QP_RNR_RETRY			|
						IB_QP_MAX_QP_RD_ATOMIC		|
						IB_QP_ALT_PATH			|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_PKEY_INDEX		|
						IB_QP_PATH_MIG_STATE),
				[IB_QPT_XRC_TGT] = (IB_QP_PORT			|
						IB_QP_AV			|
						IB_QP_TIMEOUT			|
						IB_QP_MAX_DEST_RD_ATOMIC	|
						IB_QP_ALT_PATH			|
						IB_QP_ACCESS_FLAGS		|
						IB_QP_PKEY_INDEX		|
						IB_QP_MIN_RNR_TIMER		|
						IB_QP_PATH_MIG_STATE),
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
				[IB_QPT_SMI] = (IB_QP_PKEY_INDEX		|
						IB_QP_QKEY),
				[IB_QPT_GSI] = (IB_QP_PKEY_INDEX		|
						IB_QP_QKEY),
			}
		}
	},
	[IB_QPS_SQE]   = {
		[IB_QPS_RESET] = { .valid = 1 },
		[IB_QPS_ERR] =   { .valid = 1 },
		[IB_QPS_RTS]   = {
			.valid = 1,
			.opt_param = {
				[IB_QPT_UD]  = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
				[IB_QPT_UC]  = (IB_QP_CUR_STATE			|
						IB_QP_ACCESS_FLAGS),
				[IB_QPT_SMI] = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
				[IB_QPT_GSI] = (IB_QP_CUR_STATE			|
						IB_QP_QKEY),
			}
		}
	},
	[IB_QPS_ERR] = {
		[IB_QPS_RESET] = { .valid = 1 },
		[IB_QPS_ERR] =   { .valid = 1 }
	}
};

1266 1267 1268
bool ib_modify_qp_is_ok(enum ib_qp_state cur_state, enum ib_qp_state next_state,
			enum ib_qp_type type, enum ib_qp_attr_mask mask,
			enum rdma_link_layer ll)
1269 1270 1271 1272 1273 1274
{
	enum ib_qp_attr_mask req_param, opt_param;

	if (mask & IB_QP_CUR_STATE  &&
	    cur_state != IB_QPS_RTR && cur_state != IB_QPS_RTS &&
	    cur_state != IB_QPS_SQD && cur_state != IB_QPS_SQE)
1275
		return false;
1276 1277

	if (!qp_state_table[cur_state][next_state].valid)
1278
		return false;
1279 1280 1281 1282 1283

	req_param = qp_state_table[cur_state][next_state].req_param[type];
	opt_param = qp_state_table[cur_state][next_state].opt_param[type];

	if ((mask & req_param) != req_param)
1284
		return false;
1285 1286

	if (mask & ~(req_param | opt_param | IB_QP_STATE))
1287
		return false;
1288

1289
	return true;
1290 1291 1292
}
EXPORT_SYMBOL(ib_modify_qp_is_ok);

1293 1294
static int ib_resolve_eth_dmac(struct ib_device *device,
			       struct rdma_ah_attr *ah_attr)
1295 1296
{
	int           ret = 0;
1297
	struct ib_global_route *grh;
1298

1299
	if (!rdma_is_port_valid(device, rdma_ah_get_port_num(ah_attr)))
1300
		return -EINVAL;
1301

1302 1303
	grh = rdma_ah_retrieve_grh(ah_attr);

1304 1305 1306 1307 1308 1309 1310 1311 1312 1313
	if (rdma_is_multicast_addr((struct in6_addr *)ah_attr->grh.dgid.raw)) {
		if (ipv6_addr_v4mapped((struct in6_addr *)ah_attr->grh.dgid.raw)) {
			__be32 addr = 0;

			memcpy(&addr, ah_attr->grh.dgid.raw + 12, 4);
			ip_eth_mc_map(addr, (char *)ah_attr->roce.dmac);
		} else {
			ipv6_eth_mc_map((struct in6_addr *)ah_attr->grh.dgid.raw,
					(char *)ah_attr->roce.dmac);
		}
1314
	} else {
1315
		ret = ib_resolve_unicast_gid_dmac(device, ah_attr);
1316 1317 1318 1319
	}
	return ret;
}

1320
/**
1321
 * IB core internal function to perform QP attributes modification.
1322
 */
1323 1324
static int _ib_modify_qp(struct ib_qp *qp, struct ib_qp_attr *attr,
			 int attr_mask, struct ib_udata *udata)
L
Linus Torvalds 已提交
1325
{
1326
	u8 port = attr_mask & IB_QP_PORT ? attr->port_num : qp->port;
1327
	int ret;
1328

1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342
	if (rdma_ib_or_roce(qp->device, port)) {
		if (attr_mask & IB_QP_RQ_PSN && attr->rq_psn & ~0xffffff) {
			pr_warn("%s: %s rq_psn overflow, masking to 24 bits\n",
				__func__, qp->device->name);
			attr->rq_psn &= 0xffffff;
		}

		if (attr_mask & IB_QP_SQ_PSN && attr->sq_psn & ~0xffffff) {
			pr_warn("%s: %s sq_psn overflow, masking to 24 bits\n",
				__func__, qp->device->name);
			attr->sq_psn &= 0xffffff;
		}
	}

1343 1344 1345 1346 1347
	ret = ib_security_modify_qp(qp, attr, attr_mask, udata);
	if (!ret && (attr_mask & IB_QP_PORT))
		qp->port = attr->port_num;

	return ret;
1348
}
1349

1350 1351 1352 1353 1354 1355 1356 1357
static bool is_qp_type_connected(const struct ib_qp *qp)
{
	return (qp->qp_type == IB_QPT_UC ||
		qp->qp_type == IB_QPT_RC ||
		qp->qp_type == IB_QPT_XRC_INI ||
		qp->qp_type == IB_QPT_XRC_TGT);
}

1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374
/**
 * ib_modify_qp_with_udata - Modifies the attributes for the specified QP.
 * @ib_qp: The QP to modify.
 * @attr: On input, specifies the QP attributes to modify.  On output,
 *   the current values of selected QP attributes are returned.
 * @attr_mask: A bit-mask used to specify which attributes of the QP
 *   are being modified.
 * @udata: pointer to user's input output buffer information
 *   are being modified.
 * It returns 0 on success and returns appropriate error code on error.
 */
int ib_modify_qp_with_udata(struct ib_qp *ib_qp, struct ib_qp_attr *attr,
			    int attr_mask, struct ib_udata *udata)
{
	struct ib_qp *qp = ib_qp->real_qp;
	int ret;

1375
	if (attr_mask & IB_QP_AV &&
1376 1377
	    attr->ah_attr.type == RDMA_AH_ATTR_TYPE_ROCE &&
	    is_qp_type_connected(qp)) {
1378 1379 1380 1381 1382 1383
		ret = ib_resolve_eth_dmac(qp->device, &attr->ah_attr);
		if (ret)
			return ret;
	}
	return _ib_modify_qp(qp, attr, attr_mask, udata);
}
1384
EXPORT_SYMBOL(ib_modify_qp_with_udata);
1385

1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440
int ib_get_eth_speed(struct ib_device *dev, u8 port_num, u8 *speed, u8 *width)
{
	int rc;
	u32 netdev_speed;
	struct net_device *netdev;
	struct ethtool_link_ksettings lksettings;

	if (rdma_port_get_link_layer(dev, port_num) != IB_LINK_LAYER_ETHERNET)
		return -EINVAL;

	if (!dev->get_netdev)
		return -EOPNOTSUPP;

	netdev = dev->get_netdev(dev, port_num);
	if (!netdev)
		return -ENODEV;

	rtnl_lock();
	rc = __ethtool_get_link_ksettings(netdev, &lksettings);
	rtnl_unlock();

	dev_put(netdev);

	if (!rc) {
		netdev_speed = lksettings.base.speed;
	} else {
		netdev_speed = SPEED_1000;
		pr_warn("%s speed is unknown, defaulting to %d\n", netdev->name,
			netdev_speed);
	}

	if (netdev_speed <= SPEED_1000) {
		*width = IB_WIDTH_1X;
		*speed = IB_SPEED_SDR;
	} else if (netdev_speed <= SPEED_10000) {
		*width = IB_WIDTH_1X;
		*speed = IB_SPEED_FDR10;
	} else if (netdev_speed <= SPEED_20000) {
		*width = IB_WIDTH_4X;
		*speed = IB_SPEED_DDR;
	} else if (netdev_speed <= SPEED_25000) {
		*width = IB_WIDTH_1X;
		*speed = IB_SPEED_EDR;
	} else if (netdev_speed <= SPEED_40000) {
		*width = IB_WIDTH_4X;
		*speed = IB_SPEED_FDR10;
	} else {
		*width = IB_WIDTH_4X;
		*speed = IB_SPEED_EDR;
	}

	return 0;
}
EXPORT_SYMBOL(ib_get_eth_speed);

1441 1442 1443 1444
int ib_modify_qp(struct ib_qp *qp,
		 struct ib_qp_attr *qp_attr,
		 int qp_attr_mask)
{
1445
	return _ib_modify_qp(qp->real_qp, qp_attr, qp_attr_mask, NULL);
L
Linus Torvalds 已提交
1446 1447 1448 1449 1450 1451 1452 1453 1454
}
EXPORT_SYMBOL(ib_modify_qp);

int ib_query_qp(struct ib_qp *qp,
		struct ib_qp_attr *qp_attr,
		int qp_attr_mask,
		struct ib_qp_init_attr *qp_init_attr)
{
	return qp->device->query_qp ?
1455
		qp->device->query_qp(qp->real_qp, qp_attr, qp_attr_mask, qp_init_attr) :
1456
		-EOPNOTSUPP;
L
Linus Torvalds 已提交
1457 1458 1459
}
EXPORT_SYMBOL(ib_query_qp);

1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473
int ib_close_qp(struct ib_qp *qp)
{
	struct ib_qp *real_qp;
	unsigned long flags;

	real_qp = qp->real_qp;
	if (real_qp == qp)
		return -EINVAL;

	spin_lock_irqsave(&real_qp->device->event_handler_lock, flags);
	list_del(&qp->open_list);
	spin_unlock_irqrestore(&real_qp->device->event_handler_lock, flags);

	atomic_dec(&real_qp->usecnt);
1474 1475
	if (qp->qp_sec)
		ib_close_shared_qp_security(qp->qp_sec);
1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
	kfree(qp);

	return 0;
}
EXPORT_SYMBOL(ib_close_qp);

static int __ib_destroy_shared_qp(struct ib_qp *qp)
{
	struct ib_xrcd *xrcd;
	struct ib_qp *real_qp;
	int ret;

	real_qp = qp->real_qp;
	xrcd = real_qp->xrcd;

	mutex_lock(&xrcd->tgt_qp_mutex);
	ib_close_qp(qp);
	if (atomic_read(&real_qp->usecnt) == 0)
		list_del(&real_qp->xrcd_list);
	else
		real_qp = NULL;
	mutex_unlock(&xrcd->tgt_qp_mutex);

	if (real_qp) {
		ret = ib_destroy_qp(real_qp);
		if (!ret)
			atomic_dec(&xrcd->usecnt);
		else
			__ib_insert_xrcd_qp(xrcd, real_qp);
	}

	return 0;
}

L
Linus Torvalds 已提交
1510 1511 1512 1513 1514
int ib_destroy_qp(struct ib_qp *qp)
{
	struct ib_pd *pd;
	struct ib_cq *scq, *rcq;
	struct ib_srq *srq;
1515
	struct ib_rwq_ind_table *ind_tbl;
1516
	struct ib_qp_security *sec;
L
Linus Torvalds 已提交
1517 1518
	int ret;

1519 1520
	WARN_ON_ONCE(qp->mrs_used > 0);

1521 1522 1523 1524 1525 1526
	if (atomic_read(&qp->usecnt))
		return -EBUSY;

	if (qp->real_qp != qp)
		return __ib_destroy_shared_qp(qp);

S
Sean Hefty 已提交
1527 1528 1529 1530
	pd   = qp->pd;
	scq  = qp->send_cq;
	rcq  = qp->recv_cq;
	srq  = qp->srq;
1531
	ind_tbl = qp->rwq_ind_tbl;
1532 1533 1534
	sec  = qp->qp_sec;
	if (sec)
		ib_destroy_qp_security_begin(sec);
L
Linus Torvalds 已提交
1535

1536 1537 1538
	if (!qp->uobject)
		rdma_rw_cleanup_mrs(qp);

1539
	rdma_restrack_del(&qp->res);
L
Linus Torvalds 已提交
1540 1541
	ret = qp->device->destroy_qp(qp);
	if (!ret) {
S
Sean Hefty 已提交
1542 1543 1544 1545 1546 1547
		if (pd)
			atomic_dec(&pd->usecnt);
		if (scq)
			atomic_dec(&scq->usecnt);
		if (rcq)
			atomic_dec(&rcq->usecnt);
L
Linus Torvalds 已提交
1548 1549
		if (srq)
			atomic_dec(&srq->usecnt);
1550 1551
		if (ind_tbl)
			atomic_dec(&ind_tbl->usecnt);
1552 1553 1554 1555 1556
		if (sec)
			ib_destroy_qp_security_end(sec);
	} else {
		if (sec)
			ib_destroy_qp_security_abort(sec);
L
Linus Torvalds 已提交
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
	}

	return ret;
}
EXPORT_SYMBOL(ib_destroy_qp);

/* Completion queues */

struct ib_cq *ib_create_cq(struct ib_device *device,
			   ib_comp_handler comp_handler,
			   void (*event_handler)(struct ib_event *, void *),
1568 1569
			   void *cq_context,
			   const struct ib_cq_init_attr *cq_attr)
L
Linus Torvalds 已提交
1570 1571 1572
{
	struct ib_cq *cq;

1573
	cq = device->create_cq(device, cq_attr, NULL, NULL);
L
Linus Torvalds 已提交
1574 1575 1576

	if (!IS_ERR(cq)) {
		cq->device        = device;
1577
		cq->uobject       = NULL;
L
Linus Torvalds 已提交
1578 1579 1580 1581
		cq->comp_handler  = comp_handler;
		cq->event_handler = event_handler;
		cq->cq_context    = cq_context;
		atomic_set(&cq->usecnt, 0);
1582 1583
		cq->res.type = RDMA_RESTRACK_CQ;
		rdma_restrack_add(&cq->res);
L
Linus Torvalds 已提交
1584 1585 1586 1587 1588 1589
	}

	return cq;
}
EXPORT_SYMBOL(ib_create_cq);

1590
int rdma_set_cq_moderation(struct ib_cq *cq, u16 cq_count, u16 cq_period)
E
Eli Cohen 已提交
1591 1592
{
	return cq->device->modify_cq ?
1593
		cq->device->modify_cq(cq, cq_count, cq_period) : -EOPNOTSUPP;
E
Eli Cohen 已提交
1594
}
1595
EXPORT_SYMBOL(rdma_set_cq_moderation);
E
Eli Cohen 已提交
1596

L
Linus Torvalds 已提交
1597 1598 1599 1600 1601
int ib_destroy_cq(struct ib_cq *cq)
{
	if (atomic_read(&cq->usecnt))
		return -EBUSY;

1602
	rdma_restrack_del(&cq->res);
L
Linus Torvalds 已提交
1603 1604 1605 1606
	return cq->device->destroy_cq(cq);
}
EXPORT_SYMBOL(ib_destroy_cq);

R
Roland Dreier 已提交
1607
int ib_resize_cq(struct ib_cq *cq, int cqe)
L
Linus Torvalds 已提交
1608
{
1609
	return cq->device->resize_cq ?
1610
		cq->device->resize_cq(cq, cqe, NULL) : -EOPNOTSUPP;
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616 1617
}
EXPORT_SYMBOL(ib_resize_cq);

/* Memory regions */

int ib_dereg_mr(struct ib_mr *mr)
{
1618
	struct ib_pd *pd = mr->pd;
1619
	struct ib_dm *dm = mr->dm;
L
Linus Torvalds 已提交
1620 1621
	int ret;

1622
	rdma_restrack_del(&mr->res);
L
Linus Torvalds 已提交
1623
	ret = mr->device->dereg_mr(mr);
1624
	if (!ret) {
L
Linus Torvalds 已提交
1625
		atomic_dec(&pd->usecnt);
1626 1627 1628
		if (dm)
			atomic_dec(&dm->usecnt);
	}
L
Linus Torvalds 已提交
1629 1630 1631 1632 1633

	return ret;
}
EXPORT_SYMBOL(ib_dereg_mr);

S
Sagi Grimberg 已提交
1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648
/**
 * ib_alloc_mr() - Allocates a memory region
 * @pd:            protection domain associated with the region
 * @mr_type:       memory region type
 * @max_num_sg:    maximum sg entries available for registration.
 *
 * Notes:
 * Memory registeration page/sg lists must not exceed max_num_sg.
 * For mr_type IB_MR_TYPE_MEM_REG, the total length cannot exceed
 * max_num_sg * used_page_size.
 *
 */
struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
			  enum ib_mr_type mr_type,
			  u32 max_num_sg)
1649 1650 1651
{
	struct ib_mr *mr;

1652
	if (!pd->device->alloc_mr)
1653
		return ERR_PTR(-EOPNOTSUPP);
1654

1655
	mr = pd->device->alloc_mr(pd, mr_type, max_num_sg);
1656 1657 1658 1659 1660
	if (!IS_ERR(mr)) {
		mr->device  = pd->device;
		mr->pd      = pd;
		mr->uobject = NULL;
		atomic_inc(&pd->usecnt);
1661
		mr->need_inval = false;
1662 1663
		mr->res.type = RDMA_RESTRACK_MR;
		rdma_restrack_add(&mr->res);
1664 1665 1666 1667
	}

	return mr;
}
1668
EXPORT_SYMBOL(ib_alloc_mr);
1669

L
Linus Torvalds 已提交
1670 1671 1672 1673 1674 1675 1676 1677 1678
/* "Fast" memory regions */

struct ib_fmr *ib_alloc_fmr(struct ib_pd *pd,
			    int mr_access_flags,
			    struct ib_fmr_attr *fmr_attr)
{
	struct ib_fmr *fmr;

	if (!pd->device->alloc_fmr)
1679
		return ERR_PTR(-EOPNOTSUPP);
L
Linus Torvalds 已提交
1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719

	fmr = pd->device->alloc_fmr(pd, mr_access_flags, fmr_attr);
	if (!IS_ERR(fmr)) {
		fmr->device = pd->device;
		fmr->pd     = pd;
		atomic_inc(&pd->usecnt);
	}

	return fmr;
}
EXPORT_SYMBOL(ib_alloc_fmr);

int ib_unmap_fmr(struct list_head *fmr_list)
{
	struct ib_fmr *fmr;

	if (list_empty(fmr_list))
		return 0;

	fmr = list_entry(fmr_list->next, struct ib_fmr, list);
	return fmr->device->unmap_fmr(fmr_list);
}
EXPORT_SYMBOL(ib_unmap_fmr);

int ib_dealloc_fmr(struct ib_fmr *fmr)
{
	struct ib_pd *pd;
	int ret;

	pd = fmr->pd;
	ret = fmr->device->dealloc_fmr(fmr);
	if (!ret)
		atomic_dec(&pd->usecnt);

	return ret;
}
EXPORT_SYMBOL(ib_dealloc_fmr);

/* Multicast groups */

1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
static bool is_valid_mcast_lid(struct ib_qp *qp, u16 lid)
{
	struct ib_qp_init_attr init_attr = {};
	struct ib_qp_attr attr = {};
	int num_eth_ports = 0;
	int port;

	/* If QP state >= init, it is assigned to a port and we can check this
	 * port only.
	 */
	if (!ib_query_qp(qp, &attr, IB_QP_STATE | IB_QP_PORT, &init_attr)) {
		if (attr.qp_state >= IB_QPS_INIT) {
A
Alex Estrin 已提交
1732
			if (rdma_port_get_link_layer(qp->device, attr.port_num) !=
1733 1734 1735 1736 1737 1738 1739 1740
			    IB_LINK_LAYER_INFINIBAND)
				return true;
			goto lid_check;
		}
	}

	/* Can't get a quick answer, iterate over all ports */
	for (port = 0; port < qp->device->phys_port_cnt; port++)
A
Alex Estrin 已提交
1741
		if (rdma_port_get_link_layer(qp->device, port) !=
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
		    IB_LINK_LAYER_INFINIBAND)
			num_eth_ports++;

	/* If we have at lease one Ethernet port, RoCE annex declares that
	 * multicast LID should be ignored. We can't tell at this step if the
	 * QP belongs to an IB or Ethernet port.
	 */
	if (num_eth_ports)
		return true;

	/* If all the ports are IB, we can check according to IB spec. */
lid_check:
	return !(lid < be16_to_cpu(IB_MULTICAST_LID_BASE) ||
		 lid == be16_to_cpu(IB_LID_PERMISSIVE));
}

L
Linus Torvalds 已提交
1758 1759
int ib_attach_mcast(struct ib_qp *qp, union ib_gid *gid, u16 lid)
{
1760 1761
	int ret;

1762
	if (!qp->device->attach_mcast)
1763
		return -EOPNOTSUPP;
1764 1765 1766

	if (!rdma_is_multicast_addr((struct in6_addr *)gid->raw) ||
	    qp->qp_type != IB_QPT_UD || !is_valid_mcast_lid(qp, lid))
1767 1768
		return -EINVAL;

1769 1770 1771 1772
	ret = qp->device->attach_mcast(qp, gid, lid);
	if (!ret)
		atomic_inc(&qp->usecnt);
	return ret;
L
Linus Torvalds 已提交
1773 1774 1775 1776 1777
}
EXPORT_SYMBOL(ib_attach_mcast);

int ib_detach_mcast(struct ib_qp *qp, union ib_gid *gid, u16 lid)
{
1778 1779
	int ret;

1780
	if (!qp->device->detach_mcast)
1781
		return -EOPNOTSUPP;
1782 1783 1784

	if (!rdma_is_multicast_addr((struct in6_addr *)gid->raw) ||
	    qp->qp_type != IB_QPT_UD || !is_valid_mcast_lid(qp, lid))
1785 1786
		return -EINVAL;

1787 1788 1789 1790
	ret = qp->device->detach_mcast(qp, gid, lid);
	if (!ret)
		atomic_dec(&qp->usecnt);
	return ret;
L
Linus Torvalds 已提交
1791 1792
}
EXPORT_SYMBOL(ib_detach_mcast);
S
Sean Hefty 已提交
1793

1794
struct ib_xrcd *__ib_alloc_xrcd(struct ib_device *device, const char *caller)
S
Sean Hefty 已提交
1795 1796 1797 1798
{
	struct ib_xrcd *xrcd;

	if (!device->alloc_xrcd)
1799
		return ERR_PTR(-EOPNOTSUPP);
S
Sean Hefty 已提交
1800 1801 1802 1803

	xrcd = device->alloc_xrcd(device, NULL, NULL);
	if (!IS_ERR(xrcd)) {
		xrcd->device = device;
1804
		xrcd->inode = NULL;
S
Sean Hefty 已提交
1805
		atomic_set(&xrcd->usecnt, 0);
1806 1807
		mutex_init(&xrcd->tgt_qp_mutex);
		INIT_LIST_HEAD(&xrcd->tgt_qp_list);
S
Sean Hefty 已提交
1808 1809 1810 1811
	}

	return xrcd;
}
1812
EXPORT_SYMBOL(__ib_alloc_xrcd);
S
Sean Hefty 已提交
1813 1814 1815

int ib_dealloc_xrcd(struct ib_xrcd *xrcd)
{
1816 1817 1818
	struct ib_qp *qp;
	int ret;

S
Sean Hefty 已提交
1819 1820 1821
	if (atomic_read(&xrcd->usecnt))
		return -EBUSY;

1822 1823 1824 1825 1826 1827 1828
	while (!list_empty(&xrcd->tgt_qp_list)) {
		qp = list_entry(xrcd->tgt_qp_list.next, struct ib_qp, xrcd_list);
		ret = ib_destroy_qp(qp);
		if (ret)
			return ret;
	}

S
Sean Hefty 已提交
1829 1830 1831
	return xrcd->device->dealloc_xrcd(xrcd);
}
EXPORT_SYMBOL(ib_dealloc_xrcd);
1832

1833 1834 1835 1836
/**
 * ib_create_wq - Creates a WQ associated with the specified protection
 * domain.
 * @pd: The protection domain associated with the WQ.
1837
 * @wq_attr: A list of initial attributes required to create the
1838 1839 1840
 * WQ. If WQ creation succeeds, then the attributes are updated to
 * the actual capabilities of the created WQ.
 *
1841
 * wq_attr->max_wr and wq_attr->max_sge determine
1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
 * the requested size of the WQ, and set to the actual values allocated
 * on return.
 * If ib_create_wq() succeeds, then max_wr and max_sge will always be
 * at least as large as the requested values.
 */
struct ib_wq *ib_create_wq(struct ib_pd *pd,
			   struct ib_wq_init_attr *wq_attr)
{
	struct ib_wq *wq;

	if (!pd->device->create_wq)
1853
		return ERR_PTR(-EOPNOTSUPP);
1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907

	wq = pd->device->create_wq(pd, wq_attr, NULL);
	if (!IS_ERR(wq)) {
		wq->event_handler = wq_attr->event_handler;
		wq->wq_context = wq_attr->wq_context;
		wq->wq_type = wq_attr->wq_type;
		wq->cq = wq_attr->cq;
		wq->device = pd->device;
		wq->pd = pd;
		wq->uobject = NULL;
		atomic_inc(&pd->usecnt);
		atomic_inc(&wq_attr->cq->usecnt);
		atomic_set(&wq->usecnt, 0);
	}
	return wq;
}
EXPORT_SYMBOL(ib_create_wq);

/**
 * ib_destroy_wq - Destroys the specified WQ.
 * @wq: The WQ to destroy.
 */
int ib_destroy_wq(struct ib_wq *wq)
{
	int err;
	struct ib_cq *cq = wq->cq;
	struct ib_pd *pd = wq->pd;

	if (atomic_read(&wq->usecnt))
		return -EBUSY;

	err = wq->device->destroy_wq(wq);
	if (!err) {
		atomic_dec(&pd->usecnt);
		atomic_dec(&cq->usecnt);
	}
	return err;
}
EXPORT_SYMBOL(ib_destroy_wq);

/**
 * ib_modify_wq - Modifies the specified WQ.
 * @wq: The WQ to modify.
 * @wq_attr: On input, specifies the WQ attributes to modify.
 * @wq_attr_mask: A bit-mask used to specify which attributes of the WQ
 *   are being modified.
 * On output, the current values of selected WQ attributes are returned.
 */
int ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr,
		 u32 wq_attr_mask)
{
	int err;

	if (!wq->device->modify_wq)
1908
		return -EOPNOTSUPP;
1909 1910 1911 1912 1913 1914

	err = wq->device->modify_wq(wq, wq_attr, wq_attr_mask, NULL);
	return err;
}
EXPORT_SYMBOL(ib_modify_wq);

1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932
/*
 * ib_create_rwq_ind_table - Creates a RQ Indirection Table.
 * @device: The device on which to create the rwq indirection table.
 * @ib_rwq_ind_table_init_attr: A list of initial attributes required to
 * create the Indirection Table.
 *
 * Note: The life time of ib_rwq_ind_table_init_attr->ind_tbl is not less
 *	than the created ib_rwq_ind_table object and the caller is responsible
 *	for its memory allocation/free.
 */
struct ib_rwq_ind_table *ib_create_rwq_ind_table(struct ib_device *device,
						 struct ib_rwq_ind_table_init_attr *init_attr)
{
	struct ib_rwq_ind_table *rwq_ind_table;
	int i;
	u32 table_size;

	if (!device->create_rwq_ind_table)
1933
		return ERR_PTR(-EOPNOTSUPP);
1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 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

	table_size = (1 << init_attr->log_ind_tbl_size);
	rwq_ind_table = device->create_rwq_ind_table(device,
				init_attr, NULL);
	if (IS_ERR(rwq_ind_table))
		return rwq_ind_table;

	rwq_ind_table->ind_tbl = init_attr->ind_tbl;
	rwq_ind_table->log_ind_tbl_size = init_attr->log_ind_tbl_size;
	rwq_ind_table->device = device;
	rwq_ind_table->uobject = NULL;
	atomic_set(&rwq_ind_table->usecnt, 0);

	for (i = 0; i < table_size; i++)
		atomic_inc(&rwq_ind_table->ind_tbl[i]->usecnt);

	return rwq_ind_table;
}
EXPORT_SYMBOL(ib_create_rwq_ind_table);

/*
 * ib_destroy_rwq_ind_table - Destroys the specified Indirection Table.
 * @wq_ind_table: The Indirection Table to destroy.
*/
int ib_destroy_rwq_ind_table(struct ib_rwq_ind_table *rwq_ind_table)
{
	int err, i;
	u32 table_size = (1 << rwq_ind_table->log_ind_tbl_size);
	struct ib_wq **ind_tbl = rwq_ind_table->ind_tbl;

	if (atomic_read(&rwq_ind_table->usecnt))
		return -EBUSY;

	err = rwq_ind_table->device->destroy_rwq_ind_table(rwq_ind_table);
	if (!err) {
		for (i = 0; i < table_size; i++)
			atomic_dec(&ind_tbl[i]->usecnt);
	}

	return err;
}
EXPORT_SYMBOL(ib_destroy_rwq_ind_table);

1977 1978 1979 1980 1981 1982
struct ib_flow *ib_create_flow(struct ib_qp *qp,
			       struct ib_flow_attr *flow_attr,
			       int domain)
{
	struct ib_flow *flow_id;
	if (!qp->device->create_flow)
1983
		return ERR_PTR(-EOPNOTSUPP);
1984 1985

	flow_id = qp->device->create_flow(qp, flow_attr, domain);
M
Mark Bloch 已提交
1986
	if (!IS_ERR(flow_id)) {
1987
		atomic_inc(&qp->usecnt);
M
Mark Bloch 已提交
1988 1989
		flow_id->qp = qp;
	}
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
	return flow_id;
}
EXPORT_SYMBOL(ib_create_flow);

int ib_destroy_flow(struct ib_flow *flow_id)
{
	int err;
	struct ib_qp *qp = flow_id->qp;

	err = qp->device->destroy_flow(flow_id);
	if (!err)
		atomic_dec(&qp->usecnt);
	return err;
}
EXPORT_SYMBOL(ib_destroy_flow);
2005 2006 2007 2008 2009

int ib_check_mr_status(struct ib_mr *mr, u32 check_mask,
		       struct ib_mr_status *mr_status)
{
	return mr->device->check_mr_status ?
2010
		mr->device->check_mr_status(mr, check_mask, mr_status) : -EOPNOTSUPP;
2011 2012
}
EXPORT_SYMBOL(ib_check_mr_status);
2013

2014 2015 2016 2017
int ib_set_vf_link_state(struct ib_device *device, int vf, u8 port,
			 int state)
{
	if (!device->set_vf_link_state)
2018
		return -EOPNOTSUPP;
2019 2020 2021 2022 2023 2024 2025 2026 2027

	return device->set_vf_link_state(device, vf, port, state);
}
EXPORT_SYMBOL(ib_set_vf_link_state);

int ib_get_vf_config(struct ib_device *device, int vf, u8 port,
		     struct ifla_vf_info *info)
{
	if (!device->get_vf_config)
2028
		return -EOPNOTSUPP;
2029 2030 2031 2032 2033 2034 2035 2036 2037

	return device->get_vf_config(device, vf, port, info);
}
EXPORT_SYMBOL(ib_get_vf_config);

int ib_get_vf_stats(struct ib_device *device, int vf, u8 port,
		    struct ifla_vf_stats *stats)
{
	if (!device->get_vf_stats)
2038
		return -EOPNOTSUPP;
2039 2040 2041 2042 2043 2044 2045 2046 2047

	return device->get_vf_stats(device, vf, port, stats);
}
EXPORT_SYMBOL(ib_get_vf_stats);

int ib_set_vf_guid(struct ib_device *device, int vf, u8 port, u64 guid,
		   int type)
{
	if (!device->set_vf_guid)
2048
		return -EOPNOTSUPP;
2049 2050 2051 2052 2053

	return device->set_vf_guid(device, vf, port, guid, type);
}
EXPORT_SYMBOL(ib_set_vf_guid);

2054 2055 2056 2057 2058 2059
/**
 * ib_map_mr_sg() - Map the largest prefix of a dma mapped SG list
 *     and set it the memory region.
 * @mr:            memory region
 * @sg:            dma mapped scatterlist
 * @sg_nents:      number of entries in sg
2060
 * @sg_offset:     offset in bytes into sg
2061 2062 2063 2064
 * @page_size:     page vector desired page size
 *
 * Constraints:
 * - The first sg element is allowed to have an offset.
2065 2066 2067
 * - Each sg element must either be aligned to page_size or virtually
 *   contiguous to the previous element. In case an sg element has a
 *   non-contiguous offset, the mapping prefix will not include it.
2068 2069 2070
 * - The last sg element is allowed to have length less than page_size.
 * - If sg_nents total byte length exceeds the mr max_num_sge * page_size
 *   then only max_num_sg entries will be mapped.
2071
 * - If the MR was allocated with type IB_MR_TYPE_SG_GAPS, none of these
2072
 *   constraints holds and the page_size argument is ignored.
2073 2074 2075 2076 2077 2078
 *
 * Returns the number of sg elements that were mapped to the memory region.
 *
 * After this completes successfully, the  memory region
 * is ready for registration.
 */
2079
int ib_map_mr_sg(struct ib_mr *mr, struct scatterlist *sg, int sg_nents,
2080
		 unsigned int *sg_offset, unsigned int page_size)
2081 2082
{
	if (unlikely(!mr->device->map_mr_sg))
2083
		return -EOPNOTSUPP;
2084 2085 2086

	mr->page_size = page_size;

2087
	return mr->device->map_mr_sg(mr, sg, sg_nents, sg_offset);
2088 2089 2090 2091 2092 2093 2094 2095 2096
}
EXPORT_SYMBOL(ib_map_mr_sg);

/**
 * ib_sg_to_pages() - Convert the largest prefix of a sg list
 *     to a page vector
 * @mr:            memory region
 * @sgl:           dma mapped scatterlist
 * @sg_nents:      number of entries in sg
2097 2098 2099 2100
 * @sg_offset_p:   IN:  start offset in bytes into sg
 *                 OUT: offset in bytes for element n of the sg of the first
 *                      byte that has not been processed where n is the return
 *                      value of this function.
2101 2102
 * @set_page:      driver page assignment function pointer
 *
B
Bart Van Assche 已提交
2103
 * Core service helper for drivers to convert the largest
2104 2105 2106 2107 2108 2109 2110
 * prefix of given sg list to a page vector. The sg list
 * prefix converted is the prefix that meet the requirements
 * of ib_map_mr_sg.
 *
 * Returns the number of sg elements that were assigned to
 * a page vector.
 */
2111
int ib_sg_to_pages(struct ib_mr *mr, struct scatterlist *sgl, int sg_nents,
2112
		unsigned int *sg_offset_p, int (*set_page)(struct ib_mr *, u64))
2113 2114
{
	struct scatterlist *sg;
2115
	u64 last_end_dma_addr = 0;
2116
	unsigned int sg_offset = sg_offset_p ? *sg_offset_p : 0;
2117 2118
	unsigned int last_page_off = 0;
	u64 page_mask = ~((u64)mr->page_size - 1);
B
Bart Van Assche 已提交
2119
	int i, ret;
2120

2121 2122 2123
	if (unlikely(sg_nents <= 0 || sg_offset > sg_dma_len(&sgl[0])))
		return -EINVAL;

2124
	mr->iova = sg_dma_address(&sgl[0]) + sg_offset;
2125 2126 2127
	mr->length = 0;

	for_each_sg(sgl, sg, sg_nents, i) {
2128
		u64 dma_addr = sg_dma_address(sg) + sg_offset;
2129
		u64 prev_addr = dma_addr;
2130
		unsigned int dma_len = sg_dma_len(sg) - sg_offset;
2131 2132 2133
		u64 end_dma_addr = dma_addr + dma_len;
		u64 page_addr = dma_addr & page_mask;

B
Bart Van Assche 已提交
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149
		/*
		 * For the second and later elements, check whether either the
		 * end of element i-1 or the start of element i is not aligned
		 * on a page boundary.
		 */
		if (i && (last_page_off != 0 || page_addr != dma_addr)) {
			/* Stop mapping if there is a gap. */
			if (last_end_dma_addr != dma_addr)
				break;

			/*
			 * Coalesce this element with the last. If it is small
			 * enough just update mr->length. Otherwise start
			 * mapping from the next page.
			 */
			goto next_page;
2150 2151 2152
		}

		do {
B
Bart Van Assche 已提交
2153
			ret = set_page(mr, page_addr);
2154 2155 2156 2157 2158 2159 2160 2161
			if (unlikely(ret < 0)) {
				sg_offset = prev_addr - sg_dma_address(sg);
				mr->length += prev_addr - dma_addr;
				if (sg_offset_p)
					*sg_offset_p = sg_offset;
				return i || sg_offset ? i : ret;
			}
			prev_addr = page_addr;
B
Bart Van Assche 已提交
2162
next_page:
2163 2164 2165 2166 2167 2168
			page_addr += mr->page_size;
		} while (page_addr < end_dma_addr);

		mr->length += dma_len;
		last_end_dma_addr = end_dma_addr;
		last_page_off = end_dma_addr & ~page_mask;
2169 2170

		sg_offset = 0;
2171 2172
	}

2173 2174
	if (sg_offset_p)
		*sg_offset_p = 0;
2175 2176 2177
	return i;
}
EXPORT_SYMBOL(ib_sg_to_pages);
2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196

struct ib_drain_cqe {
	struct ib_cqe cqe;
	struct completion done;
};

static void ib_drain_qp_done(struct ib_cq *cq, struct ib_wc *wc)
{
	struct ib_drain_cqe *cqe = container_of(wc->wr_cqe, struct ib_drain_cqe,
						cqe);

	complete(&cqe->done);
}

/*
 * Post a WR and block until its completion is reaped for the SQ.
 */
static void __ib_drain_sq(struct ib_qp *qp)
{
2197
	struct ib_cq *cq = qp->send_cq;
2198 2199
	struct ib_qp_attr attr = { .qp_state = IB_QPS_ERR };
	struct ib_drain_cqe sdrain;
2200 2201 2202
	struct ib_send_wr *bad_swr;
	struct ib_rdma_wr swr = {
		.wr = {
2203 2204
			.next = NULL,
			{ .wr_cqe	= &sdrain.cqe, },
2205 2206 2207
			.opcode	= IB_WR_RDMA_WRITE,
		},
	};
2208 2209 2210 2211 2212 2213 2214 2215
	int ret;

	ret = ib_modify_qp(qp, &attr, IB_QP_STATE);
	if (ret) {
		WARN_ONCE(ret, "failed to drain send queue: %d\n", ret);
		return;
	}

2216 2217 2218
	sdrain.cqe.done = ib_drain_qp_done;
	init_completion(&sdrain.done);

2219
	ret = ib_post_send(qp, &swr.wr, &bad_swr);
2220 2221 2222 2223 2224
	if (ret) {
		WARN_ONCE(ret, "failed to drain send queue: %d\n", ret);
		return;
	}

2225 2226 2227 2228 2229
	if (cq->poll_ctx == IB_POLL_DIRECT)
		while (wait_for_completion_timeout(&sdrain.done, HZ / 10) <= 0)
			ib_process_cq_direct(cq, -1);
	else
		wait_for_completion(&sdrain.done);
2230 2231 2232 2233 2234 2235 2236
}

/*
 * Post a WR and block until its completion is reaped for the RQ.
 */
static void __ib_drain_rq(struct ib_qp *qp)
{
2237
	struct ib_cq *cq = qp->recv_cq;
2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248
	struct ib_qp_attr attr = { .qp_state = IB_QPS_ERR };
	struct ib_drain_cqe rdrain;
	struct ib_recv_wr rwr = {}, *bad_rwr;
	int ret;

	ret = ib_modify_qp(qp, &attr, IB_QP_STATE);
	if (ret) {
		WARN_ONCE(ret, "failed to drain recv queue: %d\n", ret);
		return;
	}

2249 2250 2251 2252
	rwr.wr_cqe = &rdrain.cqe;
	rdrain.cqe.done = ib_drain_qp_done;
	init_completion(&rdrain.done);

2253 2254 2255 2256 2257 2258
	ret = ib_post_recv(qp, &rwr, &bad_rwr);
	if (ret) {
		WARN_ONCE(ret, "failed to drain recv queue: %d\n", ret);
		return;
	}

2259 2260 2261 2262 2263
	if (cq->poll_ctx == IB_POLL_DIRECT)
		while (wait_for_completion_timeout(&rdrain.done, HZ / 10) <= 0)
			ib_process_cq_direct(cq, -1);
	else
		wait_for_completion(&rdrain.done);
2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279
}

/**
 * ib_drain_sq() - Block until all SQ CQEs have been consumed by the
 *		   application.
 * @qp:            queue pair to drain
 *
 * If the device has a provider-specific drain function, then
 * call that.  Otherwise call the generic drain function
 * __ib_drain_sq().
 *
 * The caller must:
 *
 * ensure there is room in the CQ and SQ for the drain work request and
 * completion.
 *
2280
 * allocate the CQ using ib_alloc_cq().
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
 *
 * ensure that there are no other contexts that are posting WRs concurrently.
 * Otherwise the drain is not guaranteed.
 */
void ib_drain_sq(struct ib_qp *qp)
{
	if (qp->device->drain_sq)
		qp->device->drain_sq(qp);
	else
		__ib_drain_sq(qp);
}
EXPORT_SYMBOL(ib_drain_sq);

/**
 * ib_drain_rq() - Block until all RQ CQEs have been consumed by the
 *		   application.
 * @qp:            queue pair to drain
 *
 * If the device has a provider-specific drain function, then
 * call that.  Otherwise call the generic drain function
 * __ib_drain_rq().
 *
 * The caller must:
 *
 * ensure there is room in the CQ and RQ for the drain work request and
 * completion.
 *
2308
 * allocate the CQ using ib_alloc_cq().
2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331
 *
 * ensure that there are no other contexts that are posting WRs concurrently.
 * Otherwise the drain is not guaranteed.
 */
void ib_drain_rq(struct ib_qp *qp)
{
	if (qp->device->drain_rq)
		qp->device->drain_rq(qp);
	else
		__ib_drain_rq(qp);
}
EXPORT_SYMBOL(ib_drain_rq);

/**
 * ib_drain_qp() - Block until all CQEs have been consumed by the
 *		   application on both the RQ and SQ.
 * @qp:            queue pair to drain
 *
 * The caller must:
 *
 * ensure there is room in the CQ(s), SQ, and RQ for drain work requests
 * and completions.
 *
2332
 * allocate the CQs using ib_alloc_cq().
2333 2334 2335 2336 2337 2338 2339
 *
 * ensure that there are no other contexts that are posting WRs concurrently.
 * Otherwise the drain is not guaranteed.
 */
void ib_drain_qp(struct ib_qp *qp)
{
	ib_drain_sq(qp);
2340 2341
	if (!qp->srq)
		ib_drain_rq(qp);
2342 2343
}
EXPORT_SYMBOL(ib_drain_qp);