mad.c 139.7 KB
Newer Older
M
Mike Marciniszyn 已提交
1
/*
2
 * Copyright(c) 2015-2017 Intel Corporation.
M
Mike Marciniszyn 已提交
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
 *
 * This file is provided under a dual BSD/GPLv2 license.  When using or
 * redistributing this file, you may do so under either license.
 *
 * GPL LICENSE SUMMARY
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of version 2 of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * BSD LICENSE
 *
 * 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.
 *  - Neither the name of Intel Corporation nor the names of its
 *    contributors may be used to endorse or promote products derived
 *    from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 */

#include <linux/net.h>
49
#include <rdma/opa_addr.h>
M
Mike Marciniszyn 已提交
50 51 52 53 54 55
#define OPA_NUM_PKEY_BLOCKS_PER_SMP (OPA_SMP_DR_DATA_SIZE \
			/ (OPA_PARTITION_TABLE_BLK_SIZE * sizeof(u16)))

#include "hfi.h"
#include "mad.h"
#include "trace.h"
56
#include "qp.h"
57
#include "vnic.h"
M
Mike Marciniszyn 已提交
58 59 60 61 62

/* the reset value from the FM is supposed to be 0xffff, handle both */
#define OPA_LINK_WIDTH_RESET_OLD 0x0fff
#define OPA_LINK_WIDTH_RESET 0xffff

63 64 65 66 67 68 69 70 71 72
struct trap_node {
	struct list_head list;
	struct opa_mad_notice_attr data;
	__be64 tid;
	int len;
	u32 retry;
	u8 in_use;
	u8 repress;
};

73 74 75 76 77 78 79 80
static int smp_length_check(u32 data_size, u32 request_len)
{
	if (unlikely(request_len < data_size))
		return -EINVAL;

	return 0;
}

M
Mike Marciniszyn 已提交
81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100
static int reply(struct ib_mad_hdr *smp)
{
	/*
	 * The verbs framework will handle the directed/LID route
	 * packet changes.
	 */
	smp->method = IB_MGMT_METHOD_GET_RESP;
	if (smp->mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE)
		smp->status |= IB_SMP_DIRECTION;
	return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY;
}

static inline void clear_opa_smp_data(struct opa_smp *smp)
{
	void *data = opa_get_smp_data(smp);
	size_t size = opa_get_smp_data_size(smp);

	memset(data, 0, size);
}

101 102 103 104 105 106 107 108 109 110
static u16 hfi1_lookup_pkey_value(struct hfi1_ibport *ibp, int pkey_idx)
{
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);

	if (pkey_idx < ARRAY_SIZE(ppd->pkeys))
		return ppd->pkeys[pkey_idx];

	return 0;
}

111 112 113 114 115 116 117 118 119 120
void hfi1_event_pkey_change(struct hfi1_devdata *dd, u8 port)
{
	struct ib_event event;

	event.event = IB_EVENT_PKEY_CHANGE;
	event.device = &dd->verbs_dev.rdi.ibdev;
	event.element.port_num = port;
	ib_dispatch_event(&event);
}

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
/*
 * If the port is down, clean up all pending traps.  We need to be careful
 * with the given trap, because it may be queued.
 */
static void cleanup_traps(struct hfi1_ibport *ibp, struct trap_node *trap)
{
	struct trap_node *node, *q;
	unsigned long flags;
	struct list_head trap_list;
	int i;

	for (i = 0; i < RVT_MAX_TRAP_LISTS; i++) {
		spin_lock_irqsave(&ibp->rvp.lock, flags);
		list_replace_init(&ibp->rvp.trap_lists[i].list, &trap_list);
		ibp->rvp.trap_lists[i].list_len = 0;
		spin_unlock_irqrestore(&ibp->rvp.lock, flags);

		/*
		 * Remove all items from the list, freeing all the non-given
		 * traps.
		 */
		list_for_each_entry_safe(node, q, &trap_list, list) {
			list_del(&node->list);
			if (node != trap)
				kfree(node);
		}
	}

	/*
	 * If this wasn't on one of the lists it would not be freed.  If it
	 * was on the list, it is now safe to free.
	 */
	kfree(trap);
}

static struct trap_node *check_and_add_trap(struct hfi1_ibport *ibp,
					    struct trap_node *trap)
{
	struct trap_node *node;
	struct trap_list *trap_list;
	unsigned long flags;
	unsigned long timeout;
	int found = 0;
164 165 166 167 168 169 170
	unsigned int queue_id;
	static int trap_count;

	queue_id = trap->data.generic_type & 0x0F;
	if (queue_id >= RVT_MAX_TRAP_LISTS) {
		trap_count++;
		pr_err_ratelimited("hfi1: Invalid trap 0x%0x dropped. Total dropped: %d\n",
171
				   trap->data.generic_type, trap_count);
172 173 174
		kfree(trap);
		return NULL;
	}
175 176 177 178 179 180

	/*
	 * Since the retry (handle timeout) does not remove a trap request
	 * from the list, all we have to do is compare the node.
	 */
	spin_lock_irqsave(&ibp->rvp.lock, flags);
181
	trap_list = &ibp->rvp.trap_lists[queue_id];
182 183 184 185 186 187 188 189 190 191 192 193 194 195 196

	list_for_each_entry(node, &trap_list->list, list) {
		if (node == trap) {
			node->retry++;
			found = 1;
			break;
		}
	}

	/* If it is not on the list, add it, limited to RVT-MAX_TRAP_LEN. */
	if (!found) {
		if (trap_list->list_len < RVT_MAX_TRAP_LEN) {
			trap_list->list_len++;
			list_add_tail(&trap->list, &trap_list->list);
		} else {
197
			pr_warn_ratelimited("hfi1: Maximum trap limit reached for 0x%0x traps\n",
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
					    trap->data.generic_type);
			kfree(trap);
		}
	}

	/*
	 * Next check to see if there is a timer pending.  If not, set it up
	 * and get the first trap from the list.
	 */
	node = NULL;
	if (!timer_pending(&ibp->rvp.trap_timer)) {
		/*
		 * o14-2
		 * If the time out is set we have to wait until it expires
		 * before the trap can be sent.
		 * This should be > RVT_TRAP_TIMEOUT
		 */
		timeout = (RVT_TRAP_TIMEOUT *
			   (1UL << ibp->rvp.subnet_timeout)) / 1000;
		mod_timer(&ibp->rvp.trap_timer,
			  jiffies + usecs_to_jiffies(timeout));
		node = list_first_entry(&trap_list->list, struct trap_node,
					list);
		node->in_use = 1;
	}
	spin_unlock_irqrestore(&ibp->rvp.lock, flags);

	return node;
}

static void subn_handle_opa_trap_repress(struct hfi1_ibport *ibp,
					 struct opa_smp *smp)
{
	struct trap_list *trap_list;
	struct trap_node *trap;
	unsigned long flags;
	int i;

	if (smp->attr_id != IB_SMP_ATTR_NOTICE)
		return;

	spin_lock_irqsave(&ibp->rvp.lock, flags);
	for (i = 0; i < RVT_MAX_TRAP_LISTS; i++) {
		trap_list = &ibp->rvp.trap_lists[i];
		trap = list_first_entry_or_null(&trap_list->list,
						struct trap_node, list);
		if (trap && trap->tid == smp->tid) {
			if (trap->in_use) {
				trap->repress = 1;
			} else {
				trap_list->list_len--;
				list_del(&trap->list);
				kfree(trap);
			}
			break;
		}
	}
	spin_unlock_irqrestore(&ibp->rvp.lock, flags);
}

258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312
static void hfi1_update_sm_ah_attr(struct hfi1_ibport *ibp,
				   struct rdma_ah_attr *attr, u32 dlid)
{
	rdma_ah_set_dlid(attr, dlid);
	rdma_ah_set_port_num(attr, ppd_from_ibp(ibp)->port);
	if (dlid >= be16_to_cpu(IB_MULTICAST_LID_BASE)) {
		struct ib_global_route *grh = rdma_ah_retrieve_grh(attr);

		rdma_ah_set_ah_flags(attr, IB_AH_GRH);
		grh->sgid_index = 0;
		grh->hop_limit = 1;
		grh->dgid.global.subnet_prefix =
			ibp->rvp.gid_prefix;
		grh->dgid.global.interface_id = OPA_MAKE_ID(dlid);
	}
}

static int hfi1_modify_qp0_ah(struct hfi1_ibport *ibp,
			      struct rvt_ah *ah, u32 dlid)
{
	struct rdma_ah_attr attr;
	struct rvt_qp *qp0;
	int ret = -EINVAL;

	memset(&attr, 0, sizeof(attr));
	attr.type = ah->ibah.type;
	hfi1_update_sm_ah_attr(ibp, &attr, dlid);
	rcu_read_lock();
	qp0 = rcu_dereference(ibp->rvp.qp[0]);
	if (qp0)
		ret = rdma_modify_ah(&ah->ibah, &attr);
	rcu_read_unlock();
	return ret;
}

static struct ib_ah *hfi1_create_qp0_ah(struct hfi1_ibport *ibp, u32 dlid)
{
	struct rdma_ah_attr attr;
	struct ib_ah *ah = ERR_PTR(-EINVAL);
	struct rvt_qp *qp0;
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	struct hfi1_devdata *dd = dd_from_ppd(ppd);
	u8 port_num = ppd->port;

	memset(&attr, 0, sizeof(attr));
	attr.type = rdma_ah_find_type(&dd->verbs_dev.rdi.ibdev, port_num);
	hfi1_update_sm_ah_attr(ibp, &attr, dlid);
	rcu_read_lock();
	qp0 = rcu_dereference(ibp->rvp.qp[0]);
	if (qp0)
		ah = rdma_create_ah(qp0->ibqp.pd, &attr);
	rcu_read_unlock();
	return ah;
}

313
static void send_trap(struct hfi1_ibport *ibp, struct trap_node *trap)
M
Mike Marciniszyn 已提交
314 315 316
{
	struct ib_mad_send_buf *send_buf;
	struct ib_mad_agent *agent;
317
	struct opa_smp *smp;
M
Mike Marciniszyn 已提交
318 319 320 321
	unsigned long flags;
	int pkey_idx;
	u32 qpn = ppd_from_ibp(ibp)->sm_trap_qp;

322
	agent = ibp->rvp.send_agent;
323 324
	if (!agent) {
		cleanup_traps(ibp, trap);
M
Mike Marciniszyn 已提交
325
		return;
326
	}
M
Mike Marciniszyn 已提交
327 328

	/* o14-3.2.1 */
329 330
	if (driver_lstate(ppd_from_ibp(ibp)) != IB_PORT_ACTIVE) {
		cleanup_traps(ibp, trap);
M
Mike Marciniszyn 已提交
331
		return;
332
	}
M
Mike Marciniszyn 已提交
333

334 335 336
	/* Add the trap to the list if necessary and see if we can send it */
	trap = check_and_add_trap(ibp, trap);
	if (!trap)
M
Mike Marciniszyn 已提交
337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352
		return;

	pkey_idx = hfi1_lookup_pkey_idx(ibp, LIM_MGMT_P_KEY);
	if (pkey_idx < 0) {
		pr_warn("%s: failed to find limited mgmt pkey, defaulting 0x%x\n",
			__func__, hfi1_get_pkey(ibp, 1));
		pkey_idx = 1;
	}

	send_buf = ib_create_send_mad(agent, qpn, pkey_idx, 0,
				      IB_MGMT_MAD_HDR, IB_MGMT_MAD_DATA,
				      GFP_ATOMIC, IB_MGMT_BASE_VERSION);
	if (IS_ERR(send_buf))
		return;

	smp = send_buf->mad;
353
	smp->base_version = OPA_MGMT_BASE_VERSION;
M
Mike Marciniszyn 已提交
354
	smp->mgmt_class = IB_MGMT_CLASS_SUBN_LID_ROUTED;
355
	smp->class_version = OPA_SM_CLASS_VERSION;
M
Mike Marciniszyn 已提交
356
	smp->method = IB_MGMT_METHOD_TRAP;
357 358 359 360 361 362 363 364 365 366 367

	/* Only update the transaction ID for new traps (o13-5). */
	if (trap->tid == 0) {
		ibp->rvp.tid++;
		/* make sure that tid != 0 */
		if (ibp->rvp.tid == 0)
			ibp->rvp.tid++;
		trap->tid = cpu_to_be64(ibp->rvp.tid);
	}
	smp->tid = trap->tid;

M
Mike Marciniszyn 已提交
368 369
	smp->attr_id = IB_SMP_ATTR_NOTICE;
	/* o14-1: smp->mkey = 0; */
370 371

	memcpy(smp->route.lid.data, &trap->data, trap->len);
M
Mike Marciniszyn 已提交
372

373
	spin_lock_irqsave(&ibp->rvp.lock, flags);
374
	if (!ibp->rvp.sm_ah) {
375
		if (ibp->rvp.sm_lid != be16_to_cpu(IB_LID_PERMISSIVE)) {
M
Mike Marciniszyn 已提交
376 377
			struct ib_ah *ah;

378
			ah = hfi1_create_qp0_ah(ibp, ibp->rvp.sm_lid);
379
			if (IS_ERR(ah)) {
380 381
				spin_unlock_irqrestore(&ibp->rvp.lock, flags);
				return;
M
Mike Marciniszyn 已提交
382
			}
383 384
			send_buf->ah = ah;
			ibp->rvp.sm_ah = ibah_to_rvtah(ah);
385
		} else {
386 387
			spin_unlock_irqrestore(&ibp->rvp.lock, flags);
			return;
388
		}
M
Mike Marciniszyn 已提交
389
	} else {
390
		send_buf->ah = &ibp->rvp.sm_ah->ibah;
M
Mike Marciniszyn 已提交
391
	}
392 393 394 395 396 397 398 399 400 401 402 403 404 405

	/*
	 * If the trap was repressed while things were getting set up, don't
	 * bother sending it. This could happen for a retry.
	 */
	if (trap->repress) {
		list_del(&trap->list);
		spin_unlock_irqrestore(&ibp->rvp.lock, flags);
		kfree(trap);
		ib_free_send_mad(send_buf);
		return;
	}

	trap->in_use = 0;
406
	spin_unlock_irqrestore(&ibp->rvp.lock, flags);
M
Mike Marciniszyn 已提交
407

408
	if (ib_post_send_mad(send_buf, NULL))
M
Mike Marciniszyn 已提交
409
		ib_free_send_mad(send_buf);
410 411
}

412
void hfi1_handle_trap_timer(struct timer_list *t)
413
{
414
	struct hfi1_ibport *ibp = from_timer(ibp, t, rvp.trap_timer);
415 416 417 418 419 420 421 422 423
	struct trap_node *trap = NULL;
	unsigned long flags;
	int i;

	/* Find the trap with the highest priority */
	spin_lock_irqsave(&ibp->rvp.lock, flags);
	for (i = 0; !trap && i < RVT_MAX_TRAP_LISTS; i++) {
		trap = list_first_entry_or_null(&ibp->rvp.trap_lists[i].list,
						struct trap_node, list);
M
Mike Marciniszyn 已提交
424
	}
425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445
	spin_unlock_irqrestore(&ibp->rvp.lock, flags);

	if (trap)
		send_trap(ibp, trap);
}

static struct trap_node *create_trap_node(u8 type, __be16 trap_num, u32 lid)
{
	struct trap_node *trap;

	trap = kzalloc(sizeof(*trap), GFP_ATOMIC);
	if (!trap)
		return NULL;

	INIT_LIST_HEAD(&trap->list);
	trap->data.generic_type = type;
	trap->data.prod_type_lsb = IB_NOTICE_PROD_CA;
	trap->data.trap_num = trap_num;
	trap->data.issuer_lid = cpu_to_be32(lid);

	return trap;
M
Mike Marciniszyn 已提交
446 447 448
}

/*
449
 * Send a bad P_Key trap (ch. 14.3.8).
M
Mike Marciniszyn 已提交
450
 */
451
void hfi1_bad_pkey(struct hfi1_ibport *ibp, u32 key, u32 sl,
D
Don Hiatt 已提交
452
		   u32 qp1, u32 qp2, u32 lid1, u32 lid2)
M
Mike Marciniszyn 已提交
453
{
454
	struct trap_node *trap;
455
	u32 lid = ppd_from_ibp(ibp)->lid;
M
Mike Marciniszyn 已提交
456

457
	ibp->rvp.n_pkt_drops++;
458
	ibp->rvp.pkey_violations++;
M
Mike Marciniszyn 已提交
459

460 461 462 463 464
	trap = create_trap_node(IB_NOTICE_TYPE_SECURITY, OPA_TRAP_BAD_P_KEY,
				lid);
	if (!trap)
		return;

M
Mike Marciniszyn 已提交
465
	/* Send violation trap */
D
Don Hiatt 已提交
466 467
	trap->data.ntc_257_258.lid1 = cpu_to_be32(lid1);
	trap->data.ntc_257_258.lid2 = cpu_to_be32(lid2);
468 469 470 471 472 473 474
	trap->data.ntc_257_258.key = cpu_to_be32(key);
	trap->data.ntc_257_258.sl = sl << 3;
	trap->data.ntc_257_258.qp1 = cpu_to_be32(qp1);
	trap->data.ntc_257_258.qp2 = cpu_to_be32(qp2);

	trap->len = sizeof(trap->data);
	send_trap(ibp, trap);
M
Mike Marciniszyn 已提交
475 476 477 478 479 480 481 482
}

/*
 * Send a bad M_Key trap (ch. 14.3.9).
 */
static void bad_mkey(struct hfi1_ibport *ibp, struct ib_mad_hdr *mad,
		     __be64 mkey, __be32 dr_slid, u8 return_path[], u8 hop_cnt)
{
483
	struct trap_node *trap;
484
	u32 lid = ppd_from_ibp(ibp)->lid;
M
Mike Marciniszyn 已提交
485

486 487 488 489 490
	trap = create_trap_node(IB_NOTICE_TYPE_SECURITY, OPA_TRAP_BAD_M_KEY,
				lid);
	if (!trap)
		return;

M
Mike Marciniszyn 已提交
491
	/* Send violation trap */
492 493 494 495 496
	trap->data.ntc_256.lid = trap->data.issuer_lid;
	trap->data.ntc_256.method = mad->method;
	trap->data.ntc_256.attr_id = mad->attr_id;
	trap->data.ntc_256.attr_mod = mad->attr_mod;
	trap->data.ntc_256.mkey = mkey;
M
Mike Marciniszyn 已提交
497
	if (mad->mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) {
498 499 500 501
		trap->data.ntc_256.dr_slid = dr_slid;
		trap->data.ntc_256.dr_trunc_hop = IB_NOTICE_TRAP_DR_NOTICE;
		if (hop_cnt > ARRAY_SIZE(trap->data.ntc_256.dr_rtn_path)) {
			trap->data.ntc_256.dr_trunc_hop |=
M
Mike Marciniszyn 已提交
502
				IB_NOTICE_TRAP_DR_TRUNC;
503
			hop_cnt = ARRAY_SIZE(trap->data.ntc_256.dr_rtn_path);
M
Mike Marciniszyn 已提交
504
		}
505 506
		trap->data.ntc_256.dr_trunc_hop |= hop_cnt;
		memcpy(trap->data.ntc_256.dr_rtn_path, return_path,
M
Mike Marciniszyn 已提交
507 508 509
		       hop_cnt);
	}

510 511 512
	trap->len = sizeof(trap->data);

	send_trap(ibp, trap);
M
Mike Marciniszyn 已提交
513 514 515 516 517
}

/*
 * Send a Port Capability Mask Changed trap (ch. 14.3.11).
 */
518
void hfi1_cap_mask_chg(struct rvt_dev_info *rdi, u8 port_num)
M
Mike Marciniszyn 已提交
519
{
520
	struct trap_node *trap;
521 522 523
	struct hfi1_ibdev *verbs_dev = dev_from_rdi(rdi);
	struct hfi1_devdata *dd = dd_from_dev(verbs_dev);
	struct hfi1_ibport *ibp = &dd->pport[port_num - 1].ibport_data;
524 525
	u32 lid = ppd_from_ibp(ibp)->lid;

526 527 528 529 530
	trap = create_trap_node(IB_NOTICE_TYPE_INFO,
				OPA_TRAP_CHANGE_CAPABILITY,
				lid);
	if (!trap)
		return;
M
Mike Marciniszyn 已提交
531

532 533 534
	trap->data.ntc_144.lid = trap->data.issuer_lid;
	trap->data.ntc_144.new_cap_mask = cpu_to_be32(ibp->rvp.port_cap_flags);
	trap->data.ntc_144.cap_mask3 = cpu_to_be16(ibp->rvp.port_cap3_flags);
M
Mike Marciniszyn 已提交
535

536 537
	trap->len = sizeof(trap->data);
	send_trap(ibp, trap);
M
Mike Marciniszyn 已提交
538 539 540 541 542 543 544
}

/*
 * Send a System Image GUID Changed trap (ch. 14.3.12).
 */
void hfi1_sys_guid_chg(struct hfi1_ibport *ibp)
{
545
	struct trap_node *trap;
546 547
	u32 lid = ppd_from_ibp(ibp)->lid;

548 549 550 551
	trap = create_trap_node(IB_NOTICE_TYPE_INFO, OPA_TRAP_CHANGE_SYSGUID,
				lid);
	if (!trap)
		return;
M
Mike Marciniszyn 已提交
552

553 554
	trap->data.ntc_145.new_sys_guid = ib_hfi1_sys_image_guid;
	trap->data.ntc_145.lid = trap->data.issuer_lid;
M
Mike Marciniszyn 已提交
555

556 557
	trap->len = sizeof(trap->data);
	send_trap(ibp, trap);
M
Mike Marciniszyn 已提交
558 559 560 561 562 563 564
}

/*
 * Send a Node Description Changed trap (ch. 14.3.13).
 */
void hfi1_node_desc_chg(struct hfi1_ibport *ibp)
{
565
	struct trap_node *trap;
566 567
	u32 lid = ppd_from_ibp(ibp)->lid;

568 569 570 571 572
	trap = create_trap_node(IB_NOTICE_TYPE_INFO,
				OPA_TRAP_CHANGE_CAPABILITY,
				lid);
	if (!trap)
		return;
M
Mike Marciniszyn 已提交
573

574 575
	trap->data.ntc_144.lid = trap->data.issuer_lid;
	trap->data.ntc_144.change_flags =
576
		cpu_to_be16(OPA_NOTICE_TRAP_NODE_DESC_CHG);
M
Mike Marciniszyn 已提交
577

578 579
	trap->len = sizeof(trap->data);
	send_trap(ibp, trap);
M
Mike Marciniszyn 已提交
580 581 582 583
}

static int __subn_get_opa_nodedesc(struct opa_smp *smp, u32 am,
				   u8 *data, struct ib_device *ibdev,
584
				   u8 port, u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
585 586 587
{
	struct opa_node_description *nd;

588
	if (am || smp_length_check(sizeof(*nd), max_len)) {
M
Mike Marciniszyn 已提交
589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	nd = (struct opa_node_description *)data;

	memcpy(nd->data, ibdev->node_desc, sizeof(nd->data));

	if (resp_len)
		*resp_len += sizeof(*nd);

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_get_opa_nodeinfo(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
605
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
606 607 608 609 610 611 612 613
{
	struct opa_node_info *ni;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	unsigned pidx = port - 1; /* IB number port from 1, hw from 0 */

	ni = (struct opa_node_info *)data;

	/* GUID 0 is illegal */
614
	if (am || pidx >= dd->num_pports || ibdev->node_guid == 0 ||
615
	    smp_length_check(sizeof(*ni), max_len) ||
616
	    get_sguid(to_iport(ibdev, port), HFI1_PORT_GUID_INDEX) == 0) {
M
Mike Marciniszyn 已提交
617 618 619 620
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

621
	ni->port_guid = get_sguid(to_iport(ibdev, port), HFI1_PORT_GUID_INDEX);
M
Mike Marciniszyn 已提交
622
	ni->base_version = OPA_MGMT_BASE_VERSION;
623
	ni->class_version = OPA_SM_CLASS_VERSION;
M
Mike Marciniszyn 已提交
624 625 626 627
	ni->node_type = 1;     /* channel adapter */
	ni->num_ports = ibdev->phys_port_cnt;
	/* This is already in network order */
	ni->system_image_guid = ib_hfi1_sys_image_guid;
628
	ni->node_guid = ibdev->node_guid;
M
Mike Marciniszyn 已提交
629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651
	ni->partition_cap = cpu_to_be16(hfi1_get_npkeys(dd));
	ni->device_id = cpu_to_be16(dd->pcidev->device);
	ni->revision = cpu_to_be32(dd->minrev);
	ni->local_port_num = port;
	ni->vendor_id[0] = dd->oui1;
	ni->vendor_id[1] = dd->oui2;
	ni->vendor_id[2] = dd->oui3;

	if (resp_len)
		*resp_len += sizeof(*ni);

	return reply((struct ib_mad_hdr *)smp);
}

static int subn_get_nodeinfo(struct ib_smp *smp, struct ib_device *ibdev,
			     u8 port)
{
	struct ib_node_info *nip = (struct ib_node_info *)&smp->data;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	unsigned pidx = port - 1; /* IB number port from 1, hw from 0 */

	/* GUID 0 is illegal */
	if (smp->attr_mod || pidx >= dd->num_pports ||
652 653
	    ibdev->node_guid == 0 ||
	    get_sguid(to_iport(ibdev, port), HFI1_PORT_GUID_INDEX) == 0) {
M
Mike Marciniszyn 已提交
654
		smp->status |= IB_SMP_INVALID_FIELD;
655 656
		return reply((struct ib_mad_hdr *)smp);
	}
M
Mike Marciniszyn 已提交
657

658
	nip->port_guid = get_sguid(to_iport(ibdev, port), HFI1_PORT_GUID_INDEX);
M
Mike Marciniszyn 已提交
659
	nip->base_version = OPA_MGMT_BASE_VERSION;
660
	nip->class_version = OPA_SM_CLASS_VERSION;
M
Mike Marciniszyn 已提交
661 662 663 664
	nip->node_type = 1;     /* channel adapter */
	nip->num_ports = ibdev->phys_port_cnt;
	/* This is already in network order */
	nip->sys_guid = ib_hfi1_sys_image_guid;
665
	nip->node_guid = ibdev->node_guid;
M
Mike Marciniszyn 已提交
666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
	nip->partition_cap = cpu_to_be16(hfi1_get_npkeys(dd));
	nip->device_id = cpu_to_be16(dd->pcidev->device);
	nip->revision = cpu_to_be32(dd->minrev);
	nip->local_port_num = port;
	nip->vendor_id[0] = dd->oui1;
	nip->vendor_id[1] = dd->oui2;
	nip->vendor_id[2] = dd->oui3;

	return reply((struct ib_mad_hdr *)smp);
}

static void set_link_width_enabled(struct hfi1_pportdata *ppd, u32 w)
{
	(void)hfi1_set_ib_cfg(ppd, HFI1_IB_CFG_LWID_ENB, w);
}

static void set_link_width_downgrade_enabled(struct hfi1_pportdata *ppd, u32 w)
{
	(void)hfi1_set_ib_cfg(ppd, HFI1_IB_CFG_LWID_DG_ENB, w);
}

static void set_link_speed_enabled(struct hfi1_pportdata *ppd, u32 s)
{
	(void)hfi1_set_ib_cfg(ppd, HFI1_IB_CFG_SPD_ENB, s);
}

static int check_mkey(struct hfi1_ibport *ibp, struct ib_mad_hdr *mad,
		      int mad_flags, __be64 mkey, __be32 dr_slid,
		      u8 return_path[], u8 hop_cnt)
{
	int valid_mkey = 0;
	int ret = 0;

	/* Is the mkey in the process of expiring? */
700 701
	if (ibp->rvp.mkey_lease_timeout &&
	    time_after_eq(jiffies, ibp->rvp.mkey_lease_timeout)) {
M
Mike Marciniszyn 已提交
702
		/* Clear timeout and mkey protection field. */
703 704
		ibp->rvp.mkey_lease_timeout = 0;
		ibp->rvp.mkeyprot = 0;
M
Mike Marciniszyn 已提交
705 706
	}

707 708
	if ((mad_flags & IB_MAD_IGNORE_MKEY) ||  ibp->rvp.mkey == 0 ||
	    ibp->rvp.mkey == mkey)
M
Mike Marciniszyn 已提交
709 710 711
		valid_mkey = 1;

	/* Unset lease timeout on any valid Get/Set/TrapRepress */
712
	if (valid_mkey && ibp->rvp.mkey_lease_timeout &&
M
Mike Marciniszyn 已提交
713 714 715
	    (mad->method == IB_MGMT_METHOD_GET ||
	     mad->method == IB_MGMT_METHOD_SET ||
	     mad->method == IB_MGMT_METHOD_TRAP_REPRESS))
716
		ibp->rvp.mkey_lease_timeout = 0;
M
Mike Marciniszyn 已提交
717 718 719 720 721

	if (!valid_mkey) {
		switch (mad->method) {
		case IB_MGMT_METHOD_GET:
			/* Bad mkey not a violation below level 2 */
722
			if (ibp->rvp.mkeyprot < 2)
M
Mike Marciniszyn 已提交
723
				break;
724
			/* fall through */
M
Mike Marciniszyn 已提交
725 726
		case IB_MGMT_METHOD_SET:
		case IB_MGMT_METHOD_TRAP_REPRESS:
727 728 729 730 731 732
			if (ibp->rvp.mkey_violations != 0xFFFF)
				++ibp->rvp.mkey_violations;
			if (!ibp->rvp.mkey_lease_timeout &&
			    ibp->rvp.mkey_lease_period)
				ibp->rvp.mkey_lease_timeout = jiffies +
					ibp->rvp.mkey_lease_period * HZ;
M
Mike Marciniszyn 已提交
733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797
			/* Generate a trap notice. */
			bad_mkey(ibp, mad, mkey, dr_slid, return_path,
				 hop_cnt);
			ret = 1;
		}
	}

	return ret;
}

/*
 * The SMA caches reads from LCB registers in case the LCB is unavailable.
 * (The LCB is unavailable in certain link states, for example.)
 */
struct lcb_datum {
	u32 off;
	u64 val;
};

static struct lcb_datum lcb_cache[] = {
	{ DC_LCB_STS_ROUND_TRIP_LTP_CNT, 0 },
};

static int write_lcb_cache(u32 off, u64 val)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(lcb_cache); i++) {
		if (lcb_cache[i].off == off) {
			lcb_cache[i].val = val;
			return 0;
		}
	}

	pr_warn("%s bad offset 0x%x\n", __func__, off);
	return -1;
}

static int read_lcb_cache(u32 off, u64 *val)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(lcb_cache); i++) {
		if (lcb_cache[i].off == off) {
			*val = lcb_cache[i].val;
			return 0;
		}
	}

	pr_warn("%s bad offset 0x%x\n", __func__, off);
	return -1;
}

void read_ltp_rtt(struct hfi1_devdata *dd)
{
	u64 reg;

	if (read_lcb_csr(dd, DC_LCB_STS_ROUND_TRIP_LTP_CNT, &reg))
		dd_dev_err(dd, "%s: unable to read LTP RTT\n", __func__);
	else
		write_lcb_cache(DC_LCB_STS_ROUND_TRIP_LTP_CNT, reg);
}

static int __subn_get_opa_portinfo(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
798
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
799 800 801 802 803 804 805 806
{
	int i;
	struct hfi1_devdata *dd;
	struct hfi1_pportdata *ppd;
	struct hfi1_ibport *ibp;
	struct opa_port_info *pi = (struct opa_port_info *)data;
	u8 mtu;
	u8 credit_rate;
807
	u8 is_beaconing_active;
M
Mike Marciniszyn 已提交
808 809 810 811 812 813
	u32 state;
	u32 num_ports = OPA_AM_NPORT(am);
	u32 start_of_sm_config = OPA_AM_START_SM_CFG(am);
	u32 buffer_units;
	u64 tmp = 0;

814
	if (num_ports != 1 || smp_length_check(sizeof(*pi), max_len)) {
M
Mike Marciniszyn 已提交
815 816 817 818 819 820 821 822 823
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	dd = dd_from_ibdev(ibdev);
	/* IB numbers ports from 1, hw from 0 */
	ppd = dd->pport + (port - 1);
	ibp = &ppd->ibport_data;

824
	if (ppd->vls_supported / 2 > ARRAY_SIZE(pi->neigh_mtu.pvlx_to_mtu) ||
825
	    ppd->vls_supported > ARRAY_SIZE(dd->vld)) {
M
Mike Marciniszyn 已提交
826 827 828 829 830 831 832 833
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	pi->lid = cpu_to_be32(ppd->lid);

	/* Only return the mkey if the protection field allows it. */
	if (!(smp->method == IB_MGMT_METHOD_GET &&
834 835 836 837 838 839 840 841
	      ibp->rvp.mkey != smp->mkey &&
	      ibp->rvp.mkeyprot == 1))
		pi->mkey = ibp->rvp.mkey;

	pi->subnet_prefix = ibp->rvp.gid_prefix;
	pi->sm_lid = cpu_to_be32(ibp->rvp.sm_lid);
	pi->ib_cap_mask = cpu_to_be32(ibp->rvp.port_cap_flags);
	pi->mkey_lease_period = cpu_to_be16(ibp->rvp.mkey_lease_period);
M
Mike Marciniszyn 已提交
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
	pi->sm_trap_qp = cpu_to_be32(ppd->sm_trap_qp);
	pi->sa_qp = cpu_to_be32(ppd->sa_qp);

	pi->link_width.enabled = cpu_to_be16(ppd->link_width_enabled);
	pi->link_width.supported = cpu_to_be16(ppd->link_width_supported);
	pi->link_width.active = cpu_to_be16(ppd->link_width_active);

	pi->link_width_downgrade.supported =
			cpu_to_be16(ppd->link_width_downgrade_supported);
	pi->link_width_downgrade.enabled =
			cpu_to_be16(ppd->link_width_downgrade_enabled);
	pi->link_width_downgrade.tx_active =
			cpu_to_be16(ppd->link_width_downgrade_tx_active);
	pi->link_width_downgrade.rx_active =
			cpu_to_be16(ppd->link_width_downgrade_rx_active);

	pi->link_speed.supported = cpu_to_be16(ppd->link_speed_supported);
	pi->link_speed.active = cpu_to_be16(ppd->link_speed_active);
	pi->link_speed.enabled = cpu_to_be16(ppd->link_speed_enabled);

	state = driver_lstate(ppd);

	if (start_of_sm_config && (state == IB_PORT_INIT))
		ppd->is_sm_config_started = 1;

867
	pi->port_phys_conf = (ppd->port_type & 0xf);
M
Mike Marciniszyn 已提交
868 869 870 871

	pi->port_states.ledenable_offlinereason = ppd->neighbor_normal << 4;
	pi->port_states.ledenable_offlinereason |=
		ppd->is_sm_config_started << 5;
872
	/*
873 874 875
	 * This pairs with the memory barrier in hfi1_start_led_override to
	 * ensure that we read the correct state of LED beaconing represented
	 * by led_override_timer_active
876
	 */
877
	smp_rmb();
878 879
	is_beaconing_active = !!atomic_read(&ppd->led_override_timer_active);
	pi->port_states.ledenable_offlinereason |= is_beaconing_active << 6;
M
Mike Marciniszyn 已提交
880
	pi->port_states.ledenable_offlinereason |=
881
		ppd->offline_disabled_reason;
M
Mike Marciniszyn 已提交
882 883

	pi->port_states.portphysstate_portstate =
884
		(driver_pstate(ppd) << 4) | state;
M
Mike Marciniszyn 已提交
885

886
	pi->mkeyprotect_lmc = (ibp->rvp.mkeyprot << 6) | ppd->lmc;
M
Mike Marciniszyn 已提交
887 888 889 890 891

	memset(pi->neigh_mtu.pvlx_to_mtu, 0, sizeof(pi->neigh_mtu.pvlx_to_mtu));
	for (i = 0; i < ppd->vls_supported; i++) {
		mtu = mtu_to_enum(dd->vld[i].mtu, HFI1_DEFAULT_ACTIVE_MTU);
		if ((i % 2) == 0)
892
			pi->neigh_mtu.pvlx_to_mtu[i / 2] |= (mtu << 4);
M
Mike Marciniszyn 已提交
893
		else
894
			pi->neigh_mtu.pvlx_to_mtu[i / 2] |= mtu;
M
Mike Marciniszyn 已提交
895 896 897
	}
	/* don't forget VL 15 */
	mtu = mtu_to_enum(dd->vld[15].mtu, 2048);
898
	pi->neigh_mtu.pvlx_to_mtu[15 / 2] |= mtu;
899
	pi->smsl = ibp->rvp.sm_sl & OPA_PI_MASK_SMSL;
M
Mike Marciniszyn 已提交
900 901 902 903 904 905 906
	pi->operational_vls = hfi1_get_ib_cfg(ppd, HFI1_IB_CFG_OP_VLS);
	pi->partenforce_filterraw |=
		(ppd->linkinit_reason & OPA_PI_MASK_LINKINIT_REASON);
	if (ppd->part_enforce & HFI1_PART_ENFORCE_IN)
		pi->partenforce_filterraw |= OPA_PI_MASK_PARTITION_ENFORCE_IN;
	if (ppd->part_enforce & HFI1_PART_ENFORCE_OUT)
		pi->partenforce_filterraw |= OPA_PI_MASK_PARTITION_ENFORCE_OUT;
907
	pi->mkey_violations = cpu_to_be16(ibp->rvp.mkey_violations);
M
Mike Marciniszyn 已提交
908
	/* P_KeyViolations are counted by hardware. */
909 910
	pi->pkey_violations = cpu_to_be16(ibp->rvp.pkey_violations);
	pi->qkey_violations = cpu_to_be16(ibp->rvp.qkey_violations);
M
Mike Marciniszyn 已提交
911 912

	pi->vl.cap = ppd->vls_supported;
913
	pi->vl.high_limit = cpu_to_be16(ibp->rvp.vl_high_limit);
M
Mike Marciniszyn 已提交
914 915 916
	pi->vl.arb_high_cap = (u8)hfi1_get_ib_cfg(ppd, HFI1_IB_CFG_VL_HIGH_CAP);
	pi->vl.arb_low_cap = (u8)hfi1_get_ib_cfg(ppd, HFI1_IB_CFG_VL_LOW_CAP);

917
	pi->clientrereg_subnettimeout = ibp->rvp.subnet_timeout;
M
Mike Marciniszyn 已提交
918 919 920 921 922 923 924 925 926 927 928 929

	pi->port_link_mode  = cpu_to_be16(OPA_PORT_LINK_MODE_OPA << 10 |
					  OPA_PORT_LINK_MODE_OPA << 5 |
					  OPA_PORT_LINK_MODE_OPA);

	pi->port_ltp_crc_mode = cpu_to_be16(ppd->port_ltp_crc_mode);

	pi->port_mode = cpu_to_be16(
				ppd->is_active_optimize_enabled ?
					OPA_PI_MASK_PORT_ACTIVE_OPTOMIZE : 0);

	pi->port_packet_format.supported =
930 931
		cpu_to_be16(OPA_PORT_PACKET_FORMAT_9B |
			    OPA_PORT_PACKET_FORMAT_16B);
M
Mike Marciniszyn 已提交
932
	pi->port_packet_format.enabled =
933 934
		cpu_to_be16(OPA_PORT_PACKET_FORMAT_9B |
			    OPA_PORT_PACKET_FORMAT_16B);
M
Mike Marciniszyn 已提交
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982

	/* flit_control.interleave is (OPA V1, version .76):
	 * bits		use
	 * ----		---
	 * 2		res
	 * 2		DistanceSupported
	 * 2		DistanceEnabled
	 * 5		MaxNextLevelTxEnabled
	 * 5		MaxNestLevelRxSupported
	 *
	 * HFI supports only "distance mode 1" (see OPA V1, version .76,
	 * section 9.6.2), so set DistanceSupported, DistanceEnabled
	 * to 0x1.
	 */
	pi->flit_control.interleave = cpu_to_be16(0x1400);

	pi->link_down_reason = ppd->local_link_down_reason.sma;
	pi->neigh_link_down_reason = ppd->neigh_link_down_reason.sma;
	pi->port_error_action = cpu_to_be32(ppd->port_error_action);
	pi->mtucap = mtu_to_enum(hfi1_max_mtu, IB_MTU_4096);

	/* 32.768 usec. response time (guessing) */
	pi->resptimevalue = 3;

	pi->local_port_num = port;

	/* buffer info for FM */
	pi->overall_buffer_space = cpu_to_be16(dd->link_credits);

	pi->neigh_node_guid = cpu_to_be64(ppd->neighbor_guid);
	pi->neigh_port_num = ppd->neighbor_port_number;
	pi->port_neigh_mode =
		(ppd->neighbor_type & OPA_PI_MASK_NEIGH_NODE_TYPE) |
		(ppd->mgmt_allowed ? OPA_PI_MASK_NEIGH_MGMT_ALLOWED : 0) |
		(ppd->neighbor_fm_security ?
			OPA_PI_MASK_NEIGH_FW_AUTH_BYPASS : 0);

	/* HFIs shall always return VL15 credits to their
	 * neighbor in a timely manner, without any credit return pacing.
	 */
	credit_rate = 0;
	buffer_units  = (dd->vau) & OPA_PI_MASK_BUF_UNIT_BUF_ALLOC;
	buffer_units |= (dd->vcu << 3) & OPA_PI_MASK_BUF_UNIT_CREDIT_ACK;
	buffer_units |= (credit_rate << 6) &
				OPA_PI_MASK_BUF_UNIT_VL15_CREDIT_RATE;
	buffer_units |= (dd->vl15_init << 11) & OPA_PI_MASK_BUF_UNIT_VL15_INIT;
	pi->buffer_units = cpu_to_be32(buffer_units);

983
	pi->opa_cap_mask = cpu_to_be16(ibp->rvp.port_cap3_flags);
984 985
	pi->collectivemask_multicastmask = ((OPA_COLLECTIVE_NR & 0x7)
					    << 3 | (OPA_MCAST_NR & 0x7));
M
Mike Marciniszyn 已提交
986 987 988 989 990 991

	/* HFI supports a replay buffer 128 LTPs in size */
	pi->replay_depth.buffer = 0x80;
	/* read the cached value of DC_LCB_STS_ROUND_TRIP_LTP_CNT */
	read_lcb_cache(DC_LCB_STS_ROUND_TRIP_LTP_CNT, &tmp);

992 993 994 995
	/*
	 * this counter is 16 bits wide, but the replay_depth.wire
	 * variable is only 8 bits
	 */
M
Mike Marciniszyn 已提交
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
	if (tmp > 0xff)
		tmp = 0xff;
	pi->replay_depth.wire = tmp;

	if (resp_len)
		*resp_len += sizeof(struct opa_port_info);

	return reply((struct ib_mad_hdr *)smp);
}

/**
 * get_pkeys - return the PKEY table
 * @dd: the hfi1_ib device
 * @port: the IB port number
 * @pkeys: the pkey table is placed here
 */
static int get_pkeys(struct hfi1_devdata *dd, u8 port, u16 *pkeys)
{
	struct hfi1_pportdata *ppd = dd->pport + port - 1;

	memcpy(pkeys, ppd->pkeys, sizeof(ppd->pkeys));

	return 0;
}

static int __subn_get_opa_pkeytable(struct opa_smp *smp, u32 am, u8 *data,
				    struct ib_device *ibdev, u8 port,
1023
				    u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	u32 n_blocks_req = OPA_AM_NBLK(am);
	u32 start_block = am & 0x7ff;
	__be16 *p;
	u16 *q;
	int i;
	u16 n_blocks_avail;
	unsigned npkeys = hfi1_get_npkeys(dd);
	size_t size;

	if (n_blocks_req == 0) {
		pr_warn("OPA Get PKey AM Invalid : P = %d; B = 0x%x; N = 0x%x\n",
			port, start_block, n_blocks_req);
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

1042
	n_blocks_avail = (u16)(npkeys / OPA_PARTITION_TABLE_BLK_SIZE) + 1;
M
Mike Marciniszyn 已提交
1043 1044 1045

	size = (n_blocks_req * OPA_PARTITION_TABLE_BLK_SIZE) * sizeof(u16);

1046 1047 1048 1049 1050
	if (smp_length_check(size, max_len)) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

M
Mike Marciniszyn 已提交
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
	if (start_block + n_blocks_req > n_blocks_avail ||
	    n_blocks_req > OPA_NUM_PKEY_BLOCKS_PER_SMP) {
		pr_warn("OPA Get PKey AM Invalid : s 0x%x; req 0x%x; "
			"avail 0x%x; blk/smp 0x%lx\n",
			start_block, n_blocks_req, n_blocks_avail,
			OPA_NUM_PKEY_BLOCKS_PER_SMP);
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

1061
	p = (__be16 *)data;
M
Mike Marciniszyn 已提交
1062 1063 1064 1065 1066 1067 1068 1069
	q = (u16 *)data;
	/* get the real pkeys if we are requesting the first block */
	if (start_block == 0) {
		get_pkeys(dd, port, q);
		for (i = 0; i < npkeys; i++)
			p[i] = cpu_to_be16(q[i]);
		if (resp_len)
			*resp_len += size;
1070
	} else {
M
Mike Marciniszyn 已提交
1071
		smp->status |= IB_SMP_INVALID_FIELD;
1072
	}
M
Mike Marciniszyn 已提交
1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
	return reply((struct ib_mad_hdr *)smp);
}

enum {
	HFI_TRANSITION_DISALLOWED,
	HFI_TRANSITION_IGNORED,
	HFI_TRANSITION_ALLOWED,
	HFI_TRANSITION_UNDEFINED,
};

/*
 * Use shortened names to improve readability of
 * {logical,physical}_state_transitions
 */
enum {
	__D = HFI_TRANSITION_DISALLOWED,
	__I = HFI_TRANSITION_IGNORED,
	__A = HFI_TRANSITION_ALLOWED,
	__U = HFI_TRANSITION_UNDEFINED,
};

/*
 * IB_PORTPHYSSTATE_POLLING (2) through OPA_PORTPHYSSTATE_MAX (11) are
 * represented in physical_state_transitions.
 */
#define __N_PHYSTATES (OPA_PORTPHYSSTATE_MAX - IB_PORTPHYSSTATE_POLLING + 1)

/*
 * Within physical_state_transitions, rows represent "old" states,
 * columns "new" states, and physical_state_transitions.allowed[old][new]
 * indicates if the transition from old state to new state is legal (see
 * OPAg1v1, Table 6-4).
 */
static const struct {
	u8 allowed[__N_PHYSTATES][__N_PHYSTATES];
} physical_state_transitions = {
	{
		/* 2    3    4    5    6    7    8    9   10   11 */
	/* 2 */	{ __A, __A, __D, __D, __D, __D, __D, __D, __D, __D },
	/* 3 */	{ __A, __I, __D, __D, __D, __D, __D, __D, __D, __A },
	/* 4 */	{ __U, __U, __U, __U, __U, __U, __U, __U, __U, __U },
	/* 5 */	{ __A, __A, __D, __I, __D, __D, __D, __D, __D, __D },
	/* 6 */	{ __U, __U, __U, __U, __U, __U, __U, __U, __U, __U },
	/* 7 */	{ __D, __A, __D, __D, __D, __I, __D, __D, __D, __D },
	/* 8 */	{ __U, __U, __U, __U, __U, __U, __U, __U, __U, __U },
	/* 9 */	{ __I, __A, __D, __D, __D, __D, __D, __I, __D, __D },
	/*10 */	{ __U, __U, __U, __U, __U, __U, __U, __U, __U, __U },
	/*11 */	{ __D, __A, __D, __D, __D, __D, __D, __D, __D, __I },
	}
};

/*
 * IB_PORT_DOWN (1) through IB_PORT_ACTIVE_DEFER (5) are represented
 * logical_state_transitions
 */

#define __N_LOGICAL_STATES (IB_PORT_ACTIVE_DEFER - IB_PORT_DOWN + 1)

/*
 * Within logical_state_transitions rows represent "old" states,
 * columns "new" states, and logical_state_transitions.allowed[old][new]
 * indicates if the transition from old state to new state is legal (see
 * OPAg1v1, Table 9-12).
 */
static const struct {
	u8 allowed[__N_LOGICAL_STATES][__N_LOGICAL_STATES];
} logical_state_transitions = {
	{
		/* 1    2    3    4    5 */
	/* 1 */	{ __I, __D, __D, __D, __U},
	/* 2 */	{ __D, __I, __A, __D, __U},
	/* 3 */	{ __D, __D, __I, __A, __U},
	/* 4 */	{ __D, __D, __I, __I, __U},
	/* 5 */	{ __U, __U, __U, __U, __U},
	}
};

static int logical_transition_allowed(int old, int new)
{
	if (old < IB_PORT_NOP || old > IB_PORT_ACTIVE_DEFER ||
	    new < IB_PORT_NOP || new > IB_PORT_ACTIVE_DEFER) {
		pr_warn("invalid logical state(s) (old %d new %d)\n",
			old, new);
		return HFI_TRANSITION_UNDEFINED;
	}

	if (new == IB_PORT_NOP)
		return HFI_TRANSITION_ALLOWED; /* always allowed */

	/* adjust states for indexing into logical_state_transitions */
	old -= IB_PORT_DOWN;
	new -= IB_PORT_DOWN;

	if (old < 0 || new < 0)
		return HFI_TRANSITION_UNDEFINED;
	return logical_state_transitions.allowed[old][new];
}

static int physical_transition_allowed(int old, int new)
{
	if (old < IB_PORTPHYSSTATE_NOP || old > OPA_PORTPHYSSTATE_MAX ||
	    new < IB_PORTPHYSSTATE_NOP || new > OPA_PORTPHYSSTATE_MAX) {
		pr_warn("invalid physical state(s) (old %d new %d)\n",
			old, new);
		return HFI_TRANSITION_UNDEFINED;
	}

	if (new == IB_PORTPHYSSTATE_NOP)
		return HFI_TRANSITION_ALLOWED; /* always allowed */

	/* adjust states for indexing into physical_state_transitions */
	old -= IB_PORTPHYSSTATE_POLLING;
	new -= IB_PORTPHYSSTATE_POLLING;

	if (old < 0 || new < 0)
		return HFI_TRANSITION_UNDEFINED;
	return physical_state_transitions.allowed[old][new];
}

static int port_states_transition_allowed(struct hfi1_pportdata *ppd,
					  u32 logical_new, u32 physical_new)
{
1195
	u32 physical_old = driver_pstate(ppd);
1196
	u32 logical_old = driver_lstate(ppd);
M
Mike Marciniszyn 已提交
1197 1198
	int ret, logical_allowed, physical_allowed;

1199 1200
	ret = logical_transition_allowed(logical_old, logical_new);
	logical_allowed = ret;
M
Mike Marciniszyn 已提交
1201 1202 1203 1204 1205 1206 1207 1208 1209

	if (ret == HFI_TRANSITION_DISALLOWED ||
	    ret == HFI_TRANSITION_UNDEFINED) {
		pr_warn("invalid logical state transition %s -> %s\n",
			opa_lstate_name(logical_old),
			opa_lstate_name(logical_new));
		return ret;
	}

1210 1211
	ret = physical_transition_allowed(physical_old, physical_new);
	physical_allowed = ret;
M
Mike Marciniszyn 已提交
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224

	if (ret == HFI_TRANSITION_DISALLOWED ||
	    ret == HFI_TRANSITION_UNDEFINED) {
		pr_warn("invalid physical state transition %s -> %s\n",
			opa_pstate_name(physical_old),
			opa_pstate_name(physical_new));
		return ret;
	}

	if (logical_allowed == HFI_TRANSITION_IGNORED &&
	    physical_allowed == HFI_TRANSITION_IGNORED)
		return HFI_TRANSITION_IGNORED;

1225 1226 1227 1228 1229 1230 1231 1232
	/*
	 * A change request of Physical Port State from
	 * 'Offline' to 'Polling' should be ignored.
	 */
	if ((physical_old == OPA_PORTPHYSSTATE_OFFLINE) &&
	    (physical_new == IB_PORTPHYSSTATE_POLLING))
		return HFI_TRANSITION_IGNORED;

M
Mike Marciniszyn 已提交
1233 1234 1235 1236 1237 1238 1239 1240
	/*
	 * Either physical_allowed or logical_allowed is
	 * HFI_TRANSITION_ALLOWED.
	 */
	return HFI_TRANSITION_ALLOWED;
}

static int set_port_states(struct hfi1_pportdata *ppd, struct opa_smp *smp,
1241
			   u32 logical_state, u32 phys_state, int local_mad)
M
Mike Marciniszyn 已提交
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
{
	struct hfi1_devdata *dd = ppd->dd;
	u32 link_state;
	int ret;

	ret = port_states_transition_allowed(ppd, logical_state, phys_state);
	if (ret == HFI_TRANSITION_DISALLOWED ||
	    ret == HFI_TRANSITION_UNDEFINED) {
		/* error message emitted above */
		smp->status |= IB_SMP_INVALID_FIELD;
		return 0;
	}

	if (ret == HFI_TRANSITION_IGNORED)
		return 0;

	if ((phys_state != IB_PORTPHYSSTATE_NOP) &&
	    !(logical_state == IB_PORT_DOWN ||
	      logical_state == IB_PORT_NOP)){
		pr_warn("SubnSet(OPA_PortInfo) port state invalid: logical_state 0x%x physical_state 0x%x\n",
			logical_state, phys_state);
		smp->status |= IB_SMP_INVALID_FIELD;
	}

	/*
	 * Logical state changes are summarized in OPAv1g1 spec.,
	 * Table 9-12; physical state changes are summarized in
	 * OPAv1g1 spec., Table 6.4.
	 */
	switch (logical_state) {
	case IB_PORT_NOP:
		if (phys_state == IB_PORTPHYSSTATE_NOP)
			break;
		/* FALLTHROUGH */
	case IB_PORT_DOWN:
1277
		if (phys_state == IB_PORTPHYSSTATE_NOP) {
M
Mike Marciniszyn 已提交
1278
			link_state = HLS_DN_DOWNDEF;
1279
		} else if (phys_state == IB_PORTPHYSSTATE_POLLING) {
M
Mike Marciniszyn 已提交
1280
			link_state = HLS_DN_POLL;
1281 1282
			set_link_down_reason(ppd, OPA_LINKDOWN_REASON_FM_BOUNCE,
					     0, OPA_LINKDOWN_REASON_FM_BOUNCE);
1283
		} else if (phys_state == IB_PORTPHYSSTATE_DISABLED) {
M
Mike Marciniszyn 已提交
1284
			link_state = HLS_DN_DISABLE;
1285
		} else {
M
Mike Marciniszyn 已提交
1286 1287 1288 1289 1290 1291
			pr_warn("SubnSet(OPA_PortInfo) invalid physical state 0x%x\n",
				phys_state);
			smp->status |= IB_SMP_INVALID_FIELD;
			break;
		}

1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305
		if ((link_state == HLS_DN_POLL ||
		     link_state == HLS_DN_DOWNDEF)) {
			/*
			 * Going to poll.  No matter what the current state,
			 * always move offline first, then tune and start the
			 * link.  This correctly handles a FM link bounce and
			 * a link enable.  Going offline is a no-op if already
			 * offline.
			 */
			set_link_state(ppd, HLS_DN_OFFLINE);
			start_link(ppd);
		} else {
			set_link_state(ppd, link_state);
		}
M
Mike Marciniszyn 已提交
1306 1307
		if (link_state == HLS_DN_DISABLE &&
		    (ppd->offline_disabled_reason >
1308
		     HFI1_ODR_MASK(OPA_LINKDOWN_REASON_SMA_DISABLED) ||
M
Mike Marciniszyn 已提交
1309
		     ppd->offline_disabled_reason ==
1310
		     HFI1_ODR_MASK(OPA_LINKDOWN_REASON_NONE)))
M
Mike Marciniszyn 已提交
1311
			ppd->offline_disabled_reason =
1312
			HFI1_ODR_MASK(OPA_LINKDOWN_REASON_SMA_DISABLED);
M
Mike Marciniszyn 已提交
1313 1314 1315 1316
		/*
		 * Don't send a reply if the response would be sent
		 * through the disabled port.
		 */
1317
		if (link_state == HLS_DN_DISABLE && !local_mad)
M
Mike Marciniszyn 已提交
1318 1319 1320 1321
			return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED;
		break;
	case IB_PORT_ARMED:
		ret = set_link_state(ppd, HLS_UP_ARMED);
1322
		if (!ret)
M
Mike Marciniszyn 已提交
1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352
			send_idle_sma(dd, SMA_IDLE_ARM);
		break;
	case IB_PORT_ACTIVE:
		if (ppd->neighbor_normal) {
			ret = set_link_state(ppd, HLS_UP_ACTIVE);
			if (ret == 0)
				send_idle_sma(dd, SMA_IDLE_ACTIVE);
		} else {
			pr_warn("SubnSet(OPA_PortInfo) Cannot move to Active with NeighborNormal 0\n");
			smp->status |= IB_SMP_INVALID_FIELD;
		}
		break;
	default:
		pr_warn("SubnSet(OPA_PortInfo) invalid logical state 0x%x\n",
			logical_state);
		smp->status |= IB_SMP_INVALID_FIELD;
	}

	return 0;
}

/**
 * subn_set_opa_portinfo - set port information
 * @smp: the incoming SM packet
 * @ibdev: the infiniband device
 * @port: the port on the device
 *
 */
static int __subn_set_opa_portinfo(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
1353
				   u32 *resp_len, u32 max_len, int local_mad)
M
Mike Marciniszyn 已提交
1354 1355 1356 1357 1358 1359 1360 1361
{
	struct opa_port_info *pi = (struct opa_port_info *)data;
	struct ib_event event;
	struct hfi1_devdata *dd;
	struct hfi1_pportdata *ppd;
	struct hfi1_ibport *ibp;
	u8 clientrereg;
	unsigned long flags;
1362 1363
	u32 smlid;
	u32 lid;
M
Mike Marciniszyn 已提交
1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
	u8 ls_old, ls_new, ps_new;
	u8 vls;
	u8 msl;
	u8 crc_enabled;
	u16 lse, lwe, mtu;
	u32 num_ports = OPA_AM_NPORT(am);
	u32 start_of_sm_config = OPA_AM_START_SM_CFG(am);
	int ret, i, invalid = 0, call_set_mtu = 0;
	int call_link_downgrade_policy = 0;

1374 1375
	if (num_ports != 1 ||
	    smp_length_check(sizeof(*pi), max_len)) {
M
Mike Marciniszyn 已提交
1376 1377 1378 1379
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

1380 1381 1382
	lid = be32_to_cpu(pi->lid);
	if (lid & 0xFF000000) {
		pr_warn("OPA_PortInfo lid out of range: %X\n", lid);
M
Mike Marciniszyn 已提交
1383 1384 1385 1386 1387 1388
		smp->status |= IB_SMP_INVALID_FIELD;
		goto get_only;
	}


	smlid = be32_to_cpu(pi->sm_lid);
1389
	if (smlid & 0xFF000000) {
M
Mike Marciniszyn 已提交
1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
		pr_warn("OPA_PortInfo SM lid out of range: %X\n", smlid);
		smp->status |= IB_SMP_INVALID_FIELD;
		goto get_only;
	}

	clientrereg = (pi->clientrereg_subnettimeout &
			OPA_PI_MASK_CLIENT_REREGISTER);

	dd = dd_from_ibdev(ibdev);
	/* IB numbers ports from 1, hw from 0 */
	ppd = dd->pport + (port - 1);
	ibp = &ppd->ibport_data;
	event.device = ibdev;
	event.element.port_num = port;

	ls_old = driver_lstate(ppd);

1407
	ibp->rvp.mkey = pi->mkey;
1408 1409 1410 1411 1412
	if (ibp->rvp.gid_prefix != pi->subnet_prefix) {
		ibp->rvp.gid_prefix = pi->subnet_prefix;
		event.event = IB_EVENT_GID_CHANGE;
		ib_dispatch_event(&event);
	}
1413
	ibp->rvp.mkey_lease_period = be16_to_cpu(pi->mkey_lease_period);
M
Mike Marciniszyn 已提交
1414 1415 1416

	/* Must be a valid unicast LID address. */
	if ((lid == 0 && ls_old > IB_PORT_INIT) ||
1417
	     (hfi1_is_16B_mcast(lid))) {
M
Mike Marciniszyn 已提交
1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429
		smp->status |= IB_SMP_INVALID_FIELD;
		pr_warn("SubnSet(OPA_PortInfo) lid invalid 0x%x\n",
			lid);
	} else if (ppd->lid != lid ||
		 ppd->lmc != (pi->mkeyprotect_lmc & OPA_PI_MASK_LMC)) {
		if (ppd->lid != lid)
			hfi1_set_uevent_bits(ppd, _HFI1_EVENT_LID_CHANGE_BIT);
		if (ppd->lmc != (pi->mkeyprotect_lmc & OPA_PI_MASK_LMC))
			hfi1_set_uevent_bits(ppd, _HFI1_EVENT_LMC_CHANGE_BIT);
		hfi1_set_lid(ppd, lid, pi->mkeyprotect_lmc & OPA_PI_MASK_LMC);
		event.event = IB_EVENT_LID_CHANGE;
		ib_dispatch_event(&event);
1430 1431 1432 1433 1434 1435 1436 1437 1438 1439

		if (HFI1_PORT_GUID_INDEX + 1 < HFI1_GUIDS_PER_PORT) {
			/* Manufacture GID from LID to support extended
			 * addresses
			 */
			ppd->guids[HFI1_PORT_GUID_INDEX + 1] =
				be64_to_cpu(OPA_MAKE_ID(lid));
			event.event = IB_EVENT_GID_CHANGE;
			ib_dispatch_event(&event);
		}
M
Mike Marciniszyn 已提交
1440 1441 1442 1443 1444 1445 1446 1447 1448 1449
	}

	msl = pi->smsl & OPA_PI_MASK_SMSL;
	if (pi->partenforce_filterraw & OPA_PI_MASK_LINKINIT_REASON)
		ppd->linkinit_reason =
			(pi->partenforce_filterraw &
			 OPA_PI_MASK_LINKINIT_REASON);

	/* Must be a valid unicast LID address. */
	if ((smlid == 0 && ls_old > IB_PORT_INIT) ||
1450
	     (hfi1_is_16B_mcast(smlid))) {
M
Mike Marciniszyn 已提交
1451 1452
		smp->status |= IB_SMP_INVALID_FIELD;
		pr_warn("SubnSet(OPA_PortInfo) smlid invalid 0x%x\n", smlid);
1453
	} else if (smlid != ibp->rvp.sm_lid || msl != ibp->rvp.sm_sl) {
M
Mike Marciniszyn 已提交
1454
		pr_warn("SubnSet(OPA_PortInfo) smlid 0x%x\n", smlid);
1455
		spin_lock_irqsave(&ibp->rvp.lock, flags);
1456
		if (ibp->rvp.sm_ah) {
1457
			if (smlid != ibp->rvp.sm_lid)
1458
				hfi1_modify_qp0_ah(ibp, ibp->rvp.sm_ah, smlid);
1459
			if (msl != ibp->rvp.sm_sl)
1460
				rdma_ah_set_sl(&ibp->rvp.sm_ah->attr, msl);
M
Mike Marciniszyn 已提交
1461
		}
1462 1463 1464 1465 1466
		spin_unlock_irqrestore(&ibp->rvp.lock, flags);
		if (smlid != ibp->rvp.sm_lid)
			ibp->rvp.sm_lid = smlid;
		if (msl != ibp->rvp.sm_sl)
			ibp->rvp.sm_sl = msl;
M
Mike Marciniszyn 已提交
1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486
		event.event = IB_EVENT_SM_CHANGE;
		ib_dispatch_event(&event);
	}

	if (pi->link_down_reason == 0) {
		ppd->local_link_down_reason.sma = 0;
		ppd->local_link_down_reason.latest = 0;
	}

	if (pi->neigh_link_down_reason == 0) {
		ppd->neigh_link_down_reason.sma = 0;
		ppd->neigh_link_down_reason.latest = 0;
	}

	ppd->sm_trap_qp = be32_to_cpu(pi->sm_trap_qp);
	ppd->sa_qp = be32_to_cpu(pi->sa_qp);

	ppd->port_error_action = be32_to_cpu(pi->port_error_action);
	lwe = be16_to_cpu(pi->link_width.enabled);
	if (lwe) {
1487 1488
		if (lwe == OPA_LINK_WIDTH_RESET ||
		    lwe == OPA_LINK_WIDTH_RESET_OLD)
M
Mike Marciniszyn 已提交
1489 1490 1491 1492 1493 1494 1495 1496
			set_link_width_enabled(ppd, ppd->link_width_supported);
		else if ((lwe & ~ppd->link_width_supported) == 0)
			set_link_width_enabled(ppd, lwe);
		else
			smp->status |= IB_SMP_INVALID_FIELD;
	}
	lwe = be16_to_cpu(pi->link_width_downgrade.enabled);
	/* LWD.E is always applied - 0 means "disabled" */
1497 1498
	if (lwe == OPA_LINK_WIDTH_RESET ||
	    lwe == OPA_LINK_WIDTH_RESET_OLD) {
M
Mike Marciniszyn 已提交
1499
		set_link_width_downgrade_enabled(ppd,
1500 1501 1502
						 ppd->
						 link_width_downgrade_supported
						 );
M
Mike Marciniszyn 已提交
1503 1504 1505 1506 1507 1508
	} else if ((lwe & ~ppd->link_width_downgrade_supported) == 0) {
		/* only set and apply if something changed */
		if (lwe != ppd->link_width_downgrade_enabled) {
			set_link_width_downgrade_enabled(ppd, lwe);
			call_link_downgrade_policy = 1;
		}
1509
	} else {
M
Mike Marciniszyn 已提交
1510
		smp->status |= IB_SMP_INVALID_FIELD;
1511
	}
M
Mike Marciniszyn 已提交
1512 1513 1514 1515 1516 1517 1518 1519
	lse = be16_to_cpu(pi->link_speed.enabled);
	if (lse) {
		if (lse & be16_to_cpu(pi->link_speed.supported))
			set_link_speed_enabled(ppd, lse);
		else
			smp->status |= IB_SMP_INVALID_FIELD;
	}

1520 1521 1522
	ibp->rvp.mkeyprot =
		(pi->mkeyprotect_lmc & OPA_PI_MASK_MKEY_PROT_BIT) >> 6;
	ibp->rvp.vl_high_limit = be16_to_cpu(pi->vl.high_limit) & 0xFF;
M
Mike Marciniszyn 已提交
1523
	(void)hfi1_set_ib_cfg(ppd, HFI1_IB_CFG_VL_HIGH_LIMIT,
1524
				    ibp->rvp.vl_high_limit);
M
Mike Marciniszyn 已提交
1525

1526
	if (ppd->vls_supported / 2 > ARRAY_SIZE(pi->neigh_mtu.pvlx_to_mtu) ||
1527
	    ppd->vls_supported > ARRAY_SIZE(dd->vld)) {
M
Mike Marciniszyn 已提交
1528 1529 1530 1531 1532
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}
	for (i = 0; i < ppd->vls_supported; i++) {
		if ((i % 2) == 0)
1533 1534
			mtu = enum_to_mtu((pi->neigh_mtu.pvlx_to_mtu[i / 2] >>
					   4) & 0xF);
M
Mike Marciniszyn 已提交
1535
		else
1536 1537
			mtu = enum_to_mtu(pi->neigh_mtu.pvlx_to_mtu[i / 2] &
					  0xF);
M
Mike Marciniszyn 已提交
1538 1539 1540 1541 1542 1543 1544 1545 1546
		if (mtu == 0xffff) {
			pr_warn("SubnSet(OPA_PortInfo) mtu invalid %d (0x%x)\n",
				mtu,
				(pi->neigh_mtu.pvlx_to_mtu[0] >> 4) & 0xF);
			smp->status |= IB_SMP_INVALID_FIELD;
			mtu = hfi1_max_mtu; /* use a valid MTU */
		}
		if (dd->vld[i].mtu != mtu) {
			dd_dev_info(dd,
1547 1548
				    "MTU change on vl %d from %d to %d\n",
				    i, dd->vld[i].mtu, mtu);
M
Mike Marciniszyn 已提交
1549 1550 1551 1552 1553 1554 1555
			dd->vld[i].mtu = mtu;
			call_set_mtu++;
		}
	}
	/* As per OPAV1 spec: VL15 must support and be configured
	 * for operation with a 2048 or larger MTU.
	 */
1556
	mtu = enum_to_mtu(pi->neigh_mtu.pvlx_to_mtu[15 / 2] & 0xF);
M
Mike Marciniszyn 已提交
1557 1558 1559 1560
	if (mtu < 2048 || mtu == 0xffff)
		mtu = 2048;
	if (dd->vld[15].mtu != mtu) {
		dd_dev_info(dd,
1561 1562
			    "MTU change on vl 15 from %d to %d\n",
			    dd->vld[15].mtu, mtu);
M
Mike Marciniszyn 已提交
1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577
		dd->vld[15].mtu = mtu;
		call_set_mtu++;
	}
	if (call_set_mtu)
		set_mtu(ppd);

	/* Set operational VLs */
	vls = pi->operational_vls & OPA_PI_MASK_OPERATIONAL_VL;
	if (vls) {
		if (vls > ppd->vls_supported) {
			pr_warn("SubnSet(OPA_PortInfo) VL's supported invalid %d\n",
				pi->operational_vls);
			smp->status |= IB_SMP_INVALID_FIELD;
		} else {
			if (hfi1_set_ib_cfg(ppd, HFI1_IB_CFG_OP_VLS,
1578
					    vls) == -EINVAL)
M
Mike Marciniszyn 已提交
1579 1580 1581 1582 1583
				smp->status |= IB_SMP_INVALID_FIELD;
		}
	}

	if (pi->mkey_violations == 0)
1584
		ibp->rvp.mkey_violations = 0;
M
Mike Marciniszyn 已提交
1585 1586

	if (pi->pkey_violations == 0)
1587
		ibp->rvp.pkey_violations = 0;
M
Mike Marciniszyn 已提交
1588 1589

	if (pi->qkey_violations == 0)
1590
		ibp->rvp.qkey_violations = 0;
M
Mike Marciniszyn 已提交
1591

1592
	ibp->rvp.subnet_timeout =
M
Mike Marciniszyn 已提交
1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615
		pi->clientrereg_subnettimeout & OPA_PI_MASK_SUBNET_TIMEOUT;

	crc_enabled = be16_to_cpu(pi->port_ltp_crc_mode);
	crc_enabled >>= 4;
	crc_enabled &= 0xf;

	if (crc_enabled != 0)
		ppd->port_crc_mode_enabled = port_ltp_to_cap(crc_enabled);

	ppd->is_active_optimize_enabled =
			!!(be16_to_cpu(pi->port_mode)
					& OPA_PI_MASK_PORT_ACTIVE_OPTOMIZE);

	ls_new = pi->port_states.portphysstate_portstate &
			OPA_PI_MASK_PORT_STATE;
	ps_new = (pi->port_states.portphysstate_portstate &
			OPA_PI_MASK_PORT_PHYSICAL_STATE) >> 4;

	if (ls_old == IB_PORT_INIT) {
		if (start_of_sm_config) {
			if (ls_new == ls_old || (ls_new == IB_PORT_ARMED))
				ppd->is_sm_config_started = 1;
		} else if (ls_new == IB_PORT_ARMED) {
1616
			if (ppd->is_sm_config_started == 0) {
M
Mike Marciniszyn 已提交
1617
				invalid = 1;
1618 1619
				smp->status |= IB_SMP_INVALID_FIELD;
			}
M
Mike Marciniszyn 已提交
1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635
		}
	}

	/* Handle CLIENT_REREGISTER event b/c SM asked us for it */
	if (clientrereg) {
		event.event = IB_EVENT_CLIENT_REREGISTER;
		ib_dispatch_event(&event);
	}

	/*
	 * Do the port state change now that the other link parameters
	 * have been set.
	 * Changing the port physical state only makes sense if the link
	 * is down or is being set to down.
	 */

1636
	if (!invalid) {
1637
		ret = set_port_states(ppd, smp, ls_new, ps_new, local_mad);
1638 1639 1640
		if (ret)
			return ret;
	}
M
Mike Marciniszyn 已提交
1641

1642 1643
	ret = __subn_get_opa_portinfo(smp, am, data, ibdev, port, resp_len,
				      max_len);
M
Mike Marciniszyn 已提交
1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659

	/* restore re-reg bit per o14-12.2.1 */
	pi->clientrereg_subnettimeout |= clientrereg;

	/*
	 * Apply the new link downgrade policy.  This may result in a link
	 * bounce.  Do this after everything else so things are settled.
	 * Possible problem: if setting the port state above fails, then
	 * the policy change is not applied.
	 */
	if (call_link_downgrade_policy)
		apply_link_downgrade_policy(ppd, 0);

	return ret;

get_only:
1660 1661
	return __subn_get_opa_portinfo(smp, am, data, ibdev, port, resp_len,
				       max_len);
M
Mike Marciniszyn 已提交
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 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713
}

/**
 * set_pkeys - set the PKEY table for ctxt 0
 * @dd: the hfi1_ib device
 * @port: the IB port number
 * @pkeys: the PKEY table
 */
static int set_pkeys(struct hfi1_devdata *dd, u8 port, u16 *pkeys)
{
	struct hfi1_pportdata *ppd;
	int i;
	int changed = 0;
	int update_includes_mgmt_partition = 0;

	/*
	 * IB port one/two always maps to context zero/one,
	 * always a kernel context, no locking needed
	 * If we get here with ppd setup, no need to check
	 * that rcd is valid.
	 */
	ppd = dd->pport + (port - 1);
	/*
	 * If the update does not include the management pkey, don't do it.
	 */
	for (i = 0; i < ARRAY_SIZE(ppd->pkeys); i++) {
		if (pkeys[i] == LIM_MGMT_P_KEY) {
			update_includes_mgmt_partition = 1;
			break;
		}
	}

	if (!update_includes_mgmt_partition)
		return 1;

	for (i = 0; i < ARRAY_SIZE(ppd->pkeys); i++) {
		u16 key = pkeys[i];
		u16 okey = ppd->pkeys[i];

		if (key == okey)
			continue;
		/*
		 * The SM gives us the complete PKey table. We have
		 * to ensure that we put the PKeys in the matching
		 * slots.
		 */
		ppd->pkeys[i] = key;
		changed = 1;
	}

	if (changed) {
		(void)hfi1_set_ib_cfg(ppd, HFI1_IB_CFG_PKEYS, 0);
1714
		hfi1_event_pkey_change(dd, port);
M
Mike Marciniszyn 已提交
1715
	}
1716

M
Mike Marciniszyn 已提交
1717 1718 1719 1720 1721
	return 0;
}

static int __subn_set_opa_pkeytable(struct opa_smp *smp, u32 am, u8 *data,
				    struct ib_device *ibdev, u8 port,
1722
				    u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1723 1724 1725 1726
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	u32 n_blocks_sent = OPA_AM_NBLK(am);
	u32 start_block = am & 0x7ff;
1727
	u16 *p = (u16 *)data;
M
Mike Marciniszyn 已提交
1728 1729 1730 1731
	__be16 *q = (__be16 *)data;
	int i;
	u16 n_blocks_avail;
	unsigned npkeys = hfi1_get_npkeys(dd);
1732
	u32 size = 0;
M
Mike Marciniszyn 已提交
1733 1734 1735 1736 1737 1738 1739 1740

	if (n_blocks_sent == 0) {
		pr_warn("OPA Get PKey AM Invalid : P = %d; B = 0x%x; N = 0x%x\n",
			port, start_block, n_blocks_sent);
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

1741
	n_blocks_avail = (u16)(npkeys / OPA_PARTITION_TABLE_BLK_SIZE) + 1;
M
Mike Marciniszyn 已提交
1742

1743 1744 1745 1746 1747 1748 1749
	size = sizeof(u16) * (n_blocks_sent * OPA_PARTITION_TABLE_BLK_SIZE);

	if (smp_length_check(size, max_len)) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

M
Mike Marciniszyn 已提交
1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766
	if (start_block + n_blocks_sent > n_blocks_avail ||
	    n_blocks_sent > OPA_NUM_PKEY_BLOCKS_PER_SMP) {
		pr_warn("OPA Set PKey AM Invalid : s 0x%x; req 0x%x; avail 0x%x; blk/smp 0x%lx\n",
			start_block, n_blocks_sent, n_blocks_avail,
			OPA_NUM_PKEY_BLOCKS_PER_SMP);
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	for (i = 0; i < n_blocks_sent * OPA_PARTITION_TABLE_BLK_SIZE; i++)
		p[i] = be16_to_cpu(q[i]);

	if (start_block == 0 && set_pkeys(dd, port, p) != 0) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

1767 1768
	return __subn_get_opa_pkeytable(smp, am, data, ibdev, port, resp_len,
					max_len);
M
Mike Marciniszyn 已提交
1769 1770 1771 1772 1773 1774 1775 1776
}

#define ILLEGAL_VL 12
/*
 * filter_sc2vlt changes mappings to VL15 to ILLEGAL_VL (except
 * for SC15, which must map to VL15). If we don't remap things this
 * way it is possible for VL15 counters to increment when we try to
 * send on a SC which is mapped to an invalid VL.
1777
 * When getting the table convert ILLEGAL_VL back to VL15.
M
Mike Marciniszyn 已提交
1778
 */
1779
static void filter_sc2vlt(void *data, bool set)
M
Mike Marciniszyn 已提交
1780 1781
{
	int i;
1782
	u8 *pd = data;
M
Mike Marciniszyn 已提交
1783 1784 1785 1786

	for (i = 0; i < OPA_MAX_SCS; i++) {
		if (i == 15)
			continue;
1787 1788 1789 1790 1791 1792 1793 1794

		if (set) {
			if ((pd[i] & 0x1f) == 0xf)
				pd[i] = ILLEGAL_VL;
		} else {
			if ((pd[i] & 0x1f) == ILLEGAL_VL)
				pd[i] = 0xf;
		}
M
Mike Marciniszyn 已提交
1795 1796 1797 1798 1799
	}
}

static int set_sc2vlt_tables(struct hfi1_devdata *dd, void *data)
{
1800
	u64 *val = data;
M
Mike Marciniszyn 已提交
1801

1802
	filter_sc2vlt(data, true);
M
Mike Marciniszyn 已提交
1803 1804 1805 1806 1807 1808

	write_csr(dd, SEND_SC2VLT0, *val++);
	write_csr(dd, SEND_SC2VLT1, *val++);
	write_csr(dd, SEND_SC2VLT2, *val++);
	write_csr(dd, SEND_SC2VLT3, *val++);
	write_seqlock_irq(&dd->sc2vl_lock);
1809
	memcpy(dd->sc2vl, data, sizeof(dd->sc2vl));
M
Mike Marciniszyn 已提交
1810 1811 1812 1813
	write_sequnlock_irq(&dd->sc2vl_lock);
	return 0;
}

1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
static int get_sc2vlt_tables(struct hfi1_devdata *dd, void *data)
{
	u64 *val = (u64 *)data;

	*val++ = read_csr(dd, SEND_SC2VLT0);
	*val++ = read_csr(dd, SEND_SC2VLT1);
	*val++ = read_csr(dd, SEND_SC2VLT2);
	*val++ = read_csr(dd, SEND_SC2VLT3);

	filter_sc2vlt((u64 *)data, false);
	return 0;
}

M
Mike Marciniszyn 已提交
1827 1828
static int __subn_get_opa_sl_to_sc(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
1829
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1830 1831
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
1832
	u8 *p = data;
M
Mike Marciniszyn 已提交
1833 1834 1835
	size_t size = ARRAY_SIZE(ibp->sl_to_sc); /* == 32 */
	unsigned i;

1836
	if (am || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	for (i = 0; i < ARRAY_SIZE(ibp->sl_to_sc); i++)
		*p++ = ibp->sl_to_sc[i];

	if (resp_len)
		*resp_len += size;

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_sl_to_sc(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
1852
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1853 1854
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
1855
	u8 *p = data;
1856
	size_t size = ARRAY_SIZE(ibp->sl_to_sc);
M
Mike Marciniszyn 已提交
1857
	int i;
1858
	u8 sc;
M
Mike Marciniszyn 已提交
1859

1860
	if (am || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
1861 1862 1863 1864
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

1865 1866 1867 1868 1869 1870 1871 1872 1873
	for (i = 0; i <  ARRAY_SIZE(ibp->sl_to_sc); i++) {
		sc = *p++;
		if (ibp->sl_to_sc[i] != sc) {
			ibp->sl_to_sc[i] = sc;

			/* Put all stale qps into error state */
			hfi1_error_port_qps(ibp, i);
		}
	}
M
Mike Marciniszyn 已提交
1874

1875 1876
	return __subn_get_opa_sl_to_sc(smp, am, data, ibdev, port, resp_len,
				       max_len);
M
Mike Marciniszyn 已提交
1877 1878 1879 1880
}

static int __subn_get_opa_sc_to_sl(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
1881
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1882 1883
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
1884
	u8 *p = data;
M
Mike Marciniszyn 已提交
1885 1886 1887
	size_t size = ARRAY_SIZE(ibp->sc_to_sl); /* == 32 */
	unsigned i;

1888
	if (am || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	for (i = 0; i < ARRAY_SIZE(ibp->sc_to_sl); i++)
		*p++ = ibp->sc_to_sl[i];

	if (resp_len)
		*resp_len += size;

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_sc_to_sl(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
1904
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1905 1906
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
1907
	size_t size = ARRAY_SIZE(ibp->sc_to_sl);
1908
	u8 *p = data;
M
Mike Marciniszyn 已提交
1909 1910
	int i;

1911
	if (am || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
1912 1913 1914 1915 1916 1917 1918
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	for (i = 0; i < ARRAY_SIZE(ibp->sc_to_sl); i++)
		ibp->sc_to_sl[i] = *p++;

1919 1920
	return __subn_get_opa_sc_to_sl(smp, am, data, ibdev, port, resp_len,
				       max_len);
M
Mike Marciniszyn 已提交
1921 1922 1923 1924
}

static int __subn_get_opa_sc_to_vlt(struct opa_smp *smp, u32 am, u8 *data,
				    struct ib_device *ibdev, u8 port,
1925
				    u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1926 1927 1928
{
	u32 n_blocks = OPA_AM_NBLK(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
1929
	void *vp = (void *)data;
M
Mike Marciniszyn 已提交
1930 1931
	size_t size = 4 * sizeof(u64);

1932
	if (n_blocks != 1 || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	get_sc2vlt_tables(dd, vp);

	if (resp_len)
		*resp_len += size;

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_sc_to_vlt(struct opa_smp *smp, u32 am, u8 *data,
				    struct ib_device *ibdev, u8 port,
1947
				    u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1948 1949 1950 1951
{
	u32 n_blocks = OPA_AM_NBLK(am);
	int async_update = OPA_AM_ASYNC(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
1952
	void *vp = (void *)data;
M
Mike Marciniszyn 已提交
1953 1954
	struct hfi1_pportdata *ppd;
	int lstate;
1955 1956 1957 1958 1959 1960 1961
	/*
	 * set_sc2vlt_tables writes the information contained in *data
	 * to four 64-bit registers SendSC2VLt[0-3]. We need to make
	 * sure *max_len is not greater than the total size of the four
	 * SendSC2VLt[0-3] registers.
	 */
	size_t size = 4 * sizeof(u64);
M
Mike Marciniszyn 已提交
1962

1963
	if (n_blocks != 1 || async_update || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
1964 1965 1966 1967 1968 1969 1970
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	/* IB numbers ports from 1, hw from 0 */
	ppd = dd->pport + (port - 1);
	lstate = driver_lstate(ppd);
1971 1972 1973 1974
	/*
	 * it's known that async_update is 0 by this point, but include
	 * the explicit check for clarity
	 */
M
Mike Marciniszyn 已提交
1975 1976 1977 1978 1979 1980 1981 1982
	if (!async_update &&
	    (lstate == IB_PORT_ARMED || lstate == IB_PORT_ACTIVE)) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	set_sc2vlt_tables(dd, vp);

1983 1984
	return __subn_get_opa_sc_to_vlt(smp, am, data, ibdev, port, resp_len,
					max_len);
M
Mike Marciniszyn 已提交
1985 1986 1987 1988
}

static int __subn_get_opa_sc_to_vlnt(struct opa_smp *smp, u32 am, u8 *data,
				     struct ib_device *ibdev, u8 port,
1989
				     u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
1990 1991 1992 1993
{
	u32 n_blocks = OPA_AM_NPORT(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_pportdata *ppd;
1994
	void *vp = (void *)data;
1995
	int size = sizeof(struct sc2vlnt);
M
Mike Marciniszyn 已提交
1996

1997
	if (n_blocks != 1 || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
1998 1999 2000 2001 2002 2003
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	ppd = dd->pport + (port - 1);

2004
	fm_get_table(ppd, FM_TBL_SC2VLNT, vp);
M
Mike Marciniszyn 已提交
2005 2006 2007 2008 2009 2010 2011 2012 2013

	if (resp_len)
		*resp_len += size;

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_sc_to_vlnt(struct opa_smp *smp, u32 am, u8 *data,
				     struct ib_device *ibdev, u8 port,
2014
				     u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
2015 2016 2017 2018
{
	u32 n_blocks = OPA_AM_NPORT(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_pportdata *ppd;
2019
	void *vp = (void *)data;
M
Mike Marciniszyn 已提交
2020
	int lstate;
2021
	int size = sizeof(struct sc2vlnt);
M
Mike Marciniszyn 已提交
2022

2023
	if (n_blocks != 1 || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	/* IB numbers ports from 1, hw from 0 */
	ppd = dd->pport + (port - 1);
	lstate = driver_lstate(ppd);
	if (lstate == IB_PORT_ARMED || lstate == IB_PORT_ACTIVE) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	ppd = dd->pport + (port - 1);

	fm_set_table(ppd, FM_TBL_SC2VLNT, vp);

	return __subn_get_opa_sc_to_vlnt(smp, am, data, ibdev, port,
2041
					 resp_len, max_len);
M
Mike Marciniszyn 已提交
2042 2043 2044 2045
}

static int __subn_get_opa_psi(struct opa_smp *smp, u32 am, u8 *data,
			      struct ib_device *ibdev, u8 port,
2046
			      u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
2047 2048 2049 2050 2051 2052
{
	u32 nports = OPA_AM_NPORT(am);
	u32 start_of_sm_config = OPA_AM_START_SM_CFG(am);
	u32 lstate;
	struct hfi1_ibport *ibp;
	struct hfi1_pportdata *ppd;
2053
	struct opa_port_state_info *psi = (struct opa_port_state_info *)data;
M
Mike Marciniszyn 已提交
2054

2055
	if (nports != 1 || smp_length_check(sizeof(*psi), max_len)) {
M
Mike Marciniszyn 已提交
2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	ibp = to_iport(ibdev, port);
	ppd = ppd_from_ibp(ibp);

	lstate = driver_lstate(ppd);

	if (start_of_sm_config && (lstate == IB_PORT_INIT))
		ppd->is_sm_config_started = 1;

	psi->port_states.ledenable_offlinereason = ppd->neighbor_normal << 4;
	psi->port_states.ledenable_offlinereason |=
		ppd->is_sm_config_started << 5;
	psi->port_states.ledenable_offlinereason |=
2072
		ppd->offline_disabled_reason;
M
Mike Marciniszyn 已提交
2073 2074

	psi->port_states.portphysstate_portstate =
2075
		(driver_pstate(ppd) << 4) | (lstate & 0xf);
M
Mike Marciniszyn 已提交
2076
	psi->link_width_downgrade_tx_active =
2077
		cpu_to_be16(ppd->link_width_downgrade_tx_active);
M
Mike Marciniszyn 已提交
2078
	psi->link_width_downgrade_rx_active =
2079
		cpu_to_be16(ppd->link_width_downgrade_rx_active);
M
Mike Marciniszyn 已提交
2080 2081 2082 2083 2084 2085 2086 2087
	if (resp_len)
		*resp_len += sizeof(struct opa_port_state_info);

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_psi(struct opa_smp *smp, u32 am, u8 *data,
			      struct ib_device *ibdev, u8 port,
2088
			      u32 *resp_len, u32 max_len, int local_mad)
M
Mike Marciniszyn 已提交
2089 2090 2091 2092 2093 2094 2095
{
	u32 nports = OPA_AM_NPORT(am);
	u32 start_of_sm_config = OPA_AM_START_SM_CFG(am);
	u32 ls_old;
	u8 ls_new, ps_new;
	struct hfi1_ibport *ibp;
	struct hfi1_pportdata *ppd;
2096
	struct opa_port_state_info *psi = (struct opa_port_state_info *)data;
M
Mike Marciniszyn 已提交
2097 2098
	int ret, invalid = 0;

2099
	if (nports != 1 || smp_length_check(sizeof(*psi), max_len)) {
M
Mike Marciniszyn 已提交
2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	ibp = to_iport(ibdev, port);
	ppd = ppd_from_ibp(ibp);

	ls_old = driver_lstate(ppd);

	ls_new = port_states_to_logical_state(&psi->port_states);
	ps_new = port_states_to_phys_state(&psi->port_states);

	if (ls_old == IB_PORT_INIT) {
		if (start_of_sm_config) {
			if (ls_new == ls_old || (ls_new == IB_PORT_ARMED))
				ppd->is_sm_config_started = 1;
		} else if (ls_new == IB_PORT_ARMED) {
2117
			if (ppd->is_sm_config_started == 0) {
M
Mike Marciniszyn 已提交
2118
				invalid = 1;
2119 2120
				smp->status |= IB_SMP_INVALID_FIELD;
			}
M
Mike Marciniszyn 已提交
2121 2122 2123
		}
	}

2124
	if (!invalid) {
2125
		ret = set_port_states(ppd, smp, ls_new, ps_new, local_mad);
2126 2127 2128
		if (ret)
			return ret;
	}
M
Mike Marciniszyn 已提交
2129

2130 2131
	return __subn_get_opa_psi(smp, am, data, ibdev, port, resp_len,
				  max_len);
M
Mike Marciniszyn 已提交
2132 2133 2134 2135
}

static int __subn_get_opa_cable_info(struct opa_smp *smp, u32 am, u8 *data,
				     struct ib_device *ibdev, u8 port,
2136
				     u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
2137 2138 2139 2140 2141 2142
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	u32 addr = OPA_AM_CI_ADDR(am);
	u32 len = OPA_AM_CI_LEN(am) + 1;
	int ret;

2143 2144
	if (dd->pport->port_type != PORT_TYPE_QSFP ||
	    smp_length_check(len, max_len)) {
2145 2146 2147 2148
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

2149
#define __CI_PAGE_SIZE BIT(7) /* 128 bytes */
M
Mike Marciniszyn 已提交
2150 2151 2152
#define __CI_PAGE_MASK ~(__CI_PAGE_SIZE - 1)
#define __CI_PAGE_NUM(a) ((a) & __CI_PAGE_MASK)

2153 2154 2155 2156
	/*
	 * check that addr is within spec, and
	 * addr and (addr + len - 1) are on the same "page"
	 */
M
Mike Marciniszyn 已提交
2157
	if (addr >= 4096 ||
2158
	    (__CI_PAGE_NUM(addr) != __CI_PAGE_NUM(addr + len - 1))) {
M
Mike Marciniszyn 已提交
2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	ret = get_cable_info(dd, port, addr, len, data);

	if (ret == -ENODEV) {
		smp->status |= IB_SMP_UNSUP_METH_ATTR;
		return reply((struct ib_mad_hdr *)smp);
	}

	/* The address range for the CableInfo SMA query is wider than the
	 * memory available on the QSFP cable. We want to return a valid
	 * response, albeit zeroed out, for address ranges beyond available
	 * memory but that are within the CableInfo query spec
	 */
	if (ret < 0 && ret != -ERANGE) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	if (resp_len)
		*resp_len += len;

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_get_opa_bct(struct opa_smp *smp, u32 am, u8 *data,
2187 2188
			      struct ib_device *ibdev, u8 port, u32 *resp_len,
			      u32 max_len)
M
Mike Marciniszyn 已提交
2189 2190 2191 2192
{
	u32 num_ports = OPA_AM_NPORT(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_pportdata *ppd;
2193
	struct buffer_control *p = (struct buffer_control *)data;
2194
	int size = sizeof(struct buffer_control);
M
Mike Marciniszyn 已提交
2195

2196
	if (num_ports != 1 || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
2197 2198 2199 2200 2201
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	ppd = dd->pport + (port - 1);
2202
	fm_get_table(ppd, FM_TBL_BUFFER_CONTROL, p);
M
Mike Marciniszyn 已提交
2203 2204 2205 2206 2207 2208 2209 2210
	trace_bct_get(dd, p);
	if (resp_len)
		*resp_len += size;

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_bct(struct opa_smp *smp, u32 am, u8 *data,
2211 2212
			      struct ib_device *ibdev, u8 port, u32 *resp_len,
			      u32 max_len)
M
Mike Marciniszyn 已提交
2213 2214 2215 2216
{
	u32 num_ports = OPA_AM_NPORT(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_pportdata *ppd;
2217
	struct buffer_control *p = (struct buffer_control *)data;
M
Mike Marciniszyn 已提交
2218

2219
	if (num_ports != 1 || smp_length_check(sizeof(*p), max_len)) {
M
Mike Marciniszyn 已提交
2220 2221 2222 2223 2224 2225 2226 2227 2228 2229
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}
	ppd = dd->pport + (port - 1);
	trace_bct_set(dd, p);
	if (fm_set_table(ppd, FM_TBL_BUFFER_CONTROL, p) < 0) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

2230 2231
	return __subn_get_opa_bct(smp, am, data, ibdev, port, resp_len,
				  max_len);
M
Mike Marciniszyn 已提交
2232 2233 2234 2235
}

static int __subn_get_opa_vl_arb(struct opa_smp *smp, u32 am, u8 *data,
				 struct ib_device *ibdev, u8 port,
2236
				 u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
2237 2238 2239 2240 2241
{
	struct hfi1_pportdata *ppd = ppd_from_ibp(to_iport(ibdev, port));
	u32 num_ports = OPA_AM_NPORT(am);
	u8 section = (am & 0x00ff0000) >> 16;
	u8 *p = data;
2242
	int size = 256;
M
Mike Marciniszyn 已提交
2243

2244
	if (num_ports != 1 || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
2245 2246 2247 2248 2249 2250
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	switch (section) {
	case OPA_VLARB_LOW_ELEMENTS:
2251
		fm_get_table(ppd, FM_TBL_VL_LOW_ARB, p);
M
Mike Marciniszyn 已提交
2252 2253
		break;
	case OPA_VLARB_HIGH_ELEMENTS:
2254
		fm_get_table(ppd, FM_TBL_VL_HIGH_ARB, p);
M
Mike Marciniszyn 已提交
2255 2256
		break;
	case OPA_VLARB_PREEMPT_ELEMENTS:
2257
		fm_get_table(ppd, FM_TBL_VL_PREEMPT_ELEMS, p);
M
Mike Marciniszyn 已提交
2258 2259
		break;
	case OPA_VLARB_PREEMPT_MATRIX:
2260
		fm_get_table(ppd, FM_TBL_VL_PREEMPT_MATRIX, p);
M
Mike Marciniszyn 已提交
2261 2262 2263 2264 2265
		break;
	default:
		pr_warn("OPA SubnGet(VL Arb) AM Invalid : 0x%x\n",
			be32_to_cpu(smp->attr_mod));
		smp->status |= IB_SMP_INVALID_FIELD;
2266
		size = 0;
M
Mike Marciniszyn 已提交
2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277
		break;
	}

	if (size > 0 && resp_len)
		*resp_len += size;

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_vl_arb(struct opa_smp *smp, u32 am, u8 *data,
				 struct ib_device *ibdev, u8 port,
2278
				 u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
2279 2280 2281 2282 2283
{
	struct hfi1_pportdata *ppd = ppd_from_ibp(to_iport(ibdev, port));
	u32 num_ports = OPA_AM_NPORT(am);
	u8 section = (am & 0x00ff0000) >> 16;
	u8 *p = data;
2284
	int size = 256;
M
Mike Marciniszyn 已提交
2285

2286
	if (num_ports != 1 || smp_length_check(size, max_len)) {
M
Mike Marciniszyn 已提交
2287 2288 2289 2290 2291 2292
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	switch (section) {
	case OPA_VLARB_LOW_ELEMENTS:
2293
		(void)fm_set_table(ppd, FM_TBL_VL_LOW_ARB, p);
M
Mike Marciniszyn 已提交
2294 2295
		break;
	case OPA_VLARB_HIGH_ELEMENTS:
2296
		(void)fm_set_table(ppd, FM_TBL_VL_HIGH_ARB, p);
M
Mike Marciniszyn 已提交
2297
		break;
2298 2299 2300 2301
	/*
	 * neither OPA_VLARB_PREEMPT_ELEMENTS, or OPA_VLARB_PREEMPT_MATRIX
	 * can be changed from the default values
	 */
M
Mike Marciniszyn 已提交
2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313
	case OPA_VLARB_PREEMPT_ELEMENTS:
		/* FALLTHROUGH */
	case OPA_VLARB_PREEMPT_MATRIX:
		smp->status |= IB_SMP_UNSUP_METH_ATTR;
		break;
	default:
		pr_warn("OPA SubnSet(VL Arb) AM Invalid : 0x%x\n",
			be32_to_cpu(smp->attr_mod));
		smp->status |= IB_SMP_INVALID_FIELD;
		break;
	}

2314 2315
	return __subn_get_opa_vl_arb(smp, am, data, ibdev, port, resp_len,
				     max_len);
M
Mike Marciniszyn 已提交
2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328
}

struct opa_pma_mad {
	struct ib_mad_hdr mad_hdr;
	u8 data[2024];
} __packed;

struct opa_port_status_req {
	__u8 port_num;
	__u8 reserved[3];
	__be32 vl_select_mask;
};

2329
#define VL_MASK_ALL		0x00000000000080ffUL
M
Mike Marciniszyn 已提交
2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 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

struct opa_port_status_rsp {
	__u8 port_num;
	__u8 reserved[3];
	__be32  vl_select_mask;

	/* Data counters */
	__be64 port_xmit_data;
	__be64 port_rcv_data;
	__be64 port_xmit_pkts;
	__be64 port_rcv_pkts;
	__be64 port_multicast_xmit_pkts;
	__be64 port_multicast_rcv_pkts;
	__be64 port_xmit_wait;
	__be64 sw_port_congestion;
	__be64 port_rcv_fecn;
	__be64 port_rcv_becn;
	__be64 port_xmit_time_cong;
	__be64 port_xmit_wasted_bw;
	__be64 port_xmit_wait_data;
	__be64 port_rcv_bubble;
	__be64 port_mark_fecn;
	/* Error counters */
	__be64 port_rcv_constraint_errors;
	__be64 port_rcv_switch_relay_errors;
	__be64 port_xmit_discards;
	__be64 port_xmit_constraint_errors;
	__be64 port_rcv_remote_physical_errors;
	__be64 local_link_integrity_errors;
	__be64 port_rcv_errors;
	__be64 excessive_buffer_overruns;
	__be64 fm_config_errors;
	__be32 link_error_recovery;
	__be32 link_downed;
	u8 uncorrectable_errors;

	u8 link_quality_indicator; /* 5res, 3bit */
	u8 res2[6];
	struct _vls_pctrs {
		/* per-VL Data counters */
		__be64 port_vl_xmit_data;
		__be64 port_vl_rcv_data;
		__be64 port_vl_xmit_pkts;
		__be64 port_vl_rcv_pkts;
		__be64 port_vl_xmit_wait;
		__be64 sw_port_vl_congestion;
		__be64 port_vl_rcv_fecn;
		__be64 port_vl_rcv_becn;
		__be64 port_xmit_time_cong;
		__be64 port_vl_xmit_wasted_bw;
		__be64 port_vl_xmit_wait_data;
		__be64 port_vl_rcv_bubble;
		__be64 port_vl_mark_fecn;
		__be64 port_vl_xmit_discards;
	} vls[0]; /* real array size defined by # bits set in vl_select_mask */
};

enum counter_selects {
	CS_PORT_XMIT_DATA			= (1 << 31),
	CS_PORT_RCV_DATA			= (1 << 30),
	CS_PORT_XMIT_PKTS			= (1 << 29),
	CS_PORT_RCV_PKTS			= (1 << 28),
	CS_PORT_MCAST_XMIT_PKTS			= (1 << 27),
	CS_PORT_MCAST_RCV_PKTS			= (1 << 26),
	CS_PORT_XMIT_WAIT			= (1 << 25),
	CS_SW_PORT_CONGESTION			= (1 << 24),
	CS_PORT_RCV_FECN			= (1 << 23),
	CS_PORT_RCV_BECN			= (1 << 22),
	CS_PORT_XMIT_TIME_CONG			= (1 << 21),
	CS_PORT_XMIT_WASTED_BW			= (1 << 20),
	CS_PORT_XMIT_WAIT_DATA			= (1 << 19),
	CS_PORT_RCV_BUBBLE			= (1 << 18),
	CS_PORT_MARK_FECN			= (1 << 17),
	CS_PORT_RCV_CONSTRAINT_ERRORS		= (1 << 16),
	CS_PORT_RCV_SWITCH_RELAY_ERRORS		= (1 << 15),
	CS_PORT_XMIT_DISCARDS			= (1 << 14),
	CS_PORT_XMIT_CONSTRAINT_ERRORS		= (1 << 13),
	CS_PORT_RCV_REMOTE_PHYSICAL_ERRORS	= (1 << 12),
	CS_LOCAL_LINK_INTEGRITY_ERRORS		= (1 << 11),
	CS_PORT_RCV_ERRORS			= (1 << 10),
	CS_EXCESSIVE_BUFFER_OVERRUNS		= (1 << 9),
	CS_FM_CONFIG_ERRORS			= (1 << 8),
	CS_LINK_ERROR_RECOVERY			= (1 << 7),
	CS_LINK_DOWNED				= (1 << 6),
	CS_UNCORRECTABLE_ERRORS			= (1 << 5),
};

struct opa_clear_port_status {
	__be64 port_select_mask[4];
	__be32 counter_select_mask;
};

struct opa_aggregate {
	__be16 attr_id;
	__be16 err_reqlength;	/* 1 bit, 8 res, 7 bit */
	__be32 attr_mod;
	u8 data[0];
};

2429 2430 2431 2432 2433 2434 2435 2436
#define MSK_LLI 0x000000f0
#define MSK_LLI_SFT 4
#define MSK_LER 0x0000000f
#define MSK_LER_SFT 0
#define ADD_LLI 8
#define ADD_LER 2

/* Request contains first three fields, response contains those plus the rest */
M
Mike Marciniszyn 已提交
2437 2438 2439
struct opa_port_data_counters_msg {
	__be64 port_select_mask[4];
	__be32 vl_select_mask;
2440
	__be32 resolution;
M
Mike Marciniszyn 已提交
2441 2442 2443 2444 2445 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 2480 2481 2482 2483 2484 2485 2486 2487 2488

	/* Response fields follow */
	struct _port_dctrs {
		u8 port_number;
		u8 reserved2[3];
		__be32 link_quality_indicator; /* 29res, 3bit */

		/* Data counters */
		__be64 port_xmit_data;
		__be64 port_rcv_data;
		__be64 port_xmit_pkts;
		__be64 port_rcv_pkts;
		__be64 port_multicast_xmit_pkts;
		__be64 port_multicast_rcv_pkts;
		__be64 port_xmit_wait;
		__be64 sw_port_congestion;
		__be64 port_rcv_fecn;
		__be64 port_rcv_becn;
		__be64 port_xmit_time_cong;
		__be64 port_xmit_wasted_bw;
		__be64 port_xmit_wait_data;
		__be64 port_rcv_bubble;
		__be64 port_mark_fecn;

		__be64 port_error_counter_summary;
		/* Sum of error counts/port */

		struct _vls_dctrs {
			/* per-VL Data counters */
			__be64 port_vl_xmit_data;
			__be64 port_vl_rcv_data;
			__be64 port_vl_xmit_pkts;
			__be64 port_vl_rcv_pkts;
			__be64 port_vl_xmit_wait;
			__be64 sw_port_vl_congestion;
			__be64 port_vl_rcv_fecn;
			__be64 port_vl_rcv_becn;
			__be64 port_xmit_time_cong;
			__be64 port_vl_xmit_wasted_bw;
			__be64 port_vl_xmit_wait_data;
			__be64 port_vl_rcv_bubble;
			__be64 port_vl_mark_fecn;
		} vls[0];
		/* array size defined by #bits set in vl_select_mask*/
	} port[1]; /* array size defined by  #ports in attribute modifier */
};

struct opa_port_error_counters64_msg {
2489 2490 2491 2492
	/*
	 * Request contains first two fields, response contains the
	 * whole magilla
	 */
M
Mike Marciniszyn 已提交
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 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 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 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603
	__be64 port_select_mask[4];
	__be32 vl_select_mask;

	/* Response-only fields follow */
	__be32 reserved1;
	struct _port_ectrs {
		u8 port_number;
		u8 reserved2[7];
		__be64 port_rcv_constraint_errors;
		__be64 port_rcv_switch_relay_errors;
		__be64 port_xmit_discards;
		__be64 port_xmit_constraint_errors;
		__be64 port_rcv_remote_physical_errors;
		__be64 local_link_integrity_errors;
		__be64 port_rcv_errors;
		__be64 excessive_buffer_overruns;
		__be64 fm_config_errors;
		__be32 link_error_recovery;
		__be32 link_downed;
		u8 uncorrectable_errors;
		u8 reserved3[7];
		struct _vls_ectrs {
			__be64 port_vl_xmit_discards;
		} vls[0];
		/* array size defined by #bits set in vl_select_mask */
	} port[1]; /* array size defined by #ports in attribute modifier */
};

struct opa_port_error_info_msg {
	__be64 port_select_mask[4];
	__be32 error_info_select_mask;
	__be32 reserved1;
	struct _port_ei {
		u8 port_number;
		u8 reserved2[7];

		/* PortRcvErrorInfo */
		struct {
			u8 status_and_code;
			union {
				u8 raw[17];
				struct {
					/* EI1to12 format */
					u8 packet_flit1[8];
					u8 packet_flit2[8];
					u8 remaining_flit_bits12;
				} ei1to12;
				struct {
					u8 packet_bytes[8];
					u8 remaining_flit_bits;
				} ei13;
			} ei;
			u8 reserved3[6];
		} __packed port_rcv_ei;

		/* ExcessiveBufferOverrunInfo */
		struct {
			u8 status_and_sc;
			u8 reserved4[7];
		} __packed excessive_buffer_overrun_ei;

		/* PortXmitConstraintErrorInfo */
		struct {
			u8 status;
			u8 reserved5;
			__be16 pkey;
			__be32 slid;
		} __packed port_xmit_constraint_ei;

		/* PortRcvConstraintErrorInfo */
		struct {
			u8 status;
			u8 reserved6;
			__be16 pkey;
			__be32 slid;
		} __packed port_rcv_constraint_ei;

		/* PortRcvSwitchRelayErrorInfo */
		struct {
			u8 status_and_code;
			u8 reserved7[3];
			__u32 error_info;
		} __packed port_rcv_switch_relay_ei;

		/* UncorrectableErrorInfo */
		struct {
			u8 status_and_code;
			u8 reserved8;
		} __packed uncorrectable_ei;

		/* FMConfigErrorInfo */
		struct {
			u8 status_and_code;
			u8 error_info;
		} __packed fm_config_ei;
		__u32 reserved9;
	} port[1]; /* actual array size defined by #ports in attr modifier */
};

/* opa_port_error_info_msg error_info_select_mask bit definitions */
enum error_info_selects {
	ES_PORT_RCV_ERROR_INFO			= (1 << 31),
	ES_EXCESSIVE_BUFFER_OVERRUN_INFO	= (1 << 30),
	ES_PORT_XMIT_CONSTRAINT_ERROR_INFO	= (1 << 29),
	ES_PORT_RCV_CONSTRAINT_ERROR_INFO	= (1 << 28),
	ES_PORT_RCV_SWITCH_RELAY_ERROR_INFO	= (1 << 27),
	ES_UNCORRECTABLE_ERROR_INFO		= (1 << 26),
	ES_FM_CONFIG_ERROR_INFO			= (1 << 25)
};

static int pma_get_opa_classportinfo(struct opa_pma_mad *pmp,
2604
				     struct ib_device *ibdev, u32 *resp_len)
M
Mike Marciniszyn 已提交
2605 2606 2607 2608 2609 2610 2611 2612 2613 2614
{
	struct opa_class_port_info *p =
		(struct opa_class_port_info *)pmp->data;

	memset(pmp->data, 0, sizeof(pmp->data));

	if (pmp->mad_hdr.attr_mod != 0)
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;

	p->base_version = OPA_MGMT_BASE_VERSION;
2615
	p->class_version = OPA_SM_CLASS_VERSION;
M
Mike Marciniszyn 已提交
2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627
	/*
	 * Expected response time is 4.096 usec. * 2^18 == 1.073741824 sec.
	 */
	p->cap_mask2_resp_time = cpu_to_be32(18);

	if (resp_len)
		*resp_len += sizeof(*p);

	return reply((struct ib_mad_hdr *)pmp);
}

static void a0_portstatus(struct hfi1_pportdata *ppd,
2628
			  struct opa_port_status_rsp *rsp)
M
Mike Marciniszyn 已提交
2629 2630 2631
{
	if (!is_bx(ppd->dd)) {
		unsigned long vl;
2632
		u64 sum_vl_xmit_wait = 0;
2633
		unsigned long vl_all_mask = VL_MASK_ALL;
M
Mike Marciniszyn 已提交
2634

2635
		for_each_set_bit(vl, &vl_all_mask, BITS_PER_LONG) {
2636 2637 2638 2639 2640 2641 2642 2643 2644
			u64 tmp = sum_vl_xmit_wait +
				  read_port_cntr(ppd, C_TX_WAIT_VL,
						 idx_from_vl(vl));
			if (tmp < sum_vl_xmit_wait) {
				/* we wrapped */
				sum_vl_xmit_wait = (u64)~0;
				break;
			}
			sum_vl_xmit_wait = tmp;
M
Mike Marciniszyn 已提交
2645
		}
2646 2647
		if (be64_to_cpu(rsp->port_xmit_wait) > sum_vl_xmit_wait)
			rsp->port_xmit_wait = cpu_to_be64(sum_vl_xmit_wait);
M
Mike Marciniszyn 已提交
2648 2649 2650
	}
}

2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 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 2717 2718 2719 2720 2721 2722 2723
/**
 * tx_link_width - convert link width bitmask to integer
 * value representing actual link width.
 * @link_width: width of active link
 * @return: return index of the bit set in link_width var
 *
 * The function convert and return the index of bit set
 * that indicate the current link width.
 */
u16 tx_link_width(u16 link_width)
{
	int n = LINK_WIDTH_DEFAULT;
	u16 tx_width = n;

	while (link_width && n) {
		if (link_width & (1 << (n - 1))) {
			tx_width = n;
			break;
		}
		n--;
	}

	return tx_width;
}

/**
 * get_xmit_wait_counters - Convert HFI 's SendWaitCnt/SendWaitVlCnt
 * counter in unit of TXE cycle times to flit times.
 * @ppd: info of physical Hfi port
 * @link_width: width of active link
 * @link_speed: speed of active link
 * @vl: represent VL0-VL7, VL15 for PortVLXmitWait counters request
 * and if vl value is C_VL_COUNT, it represent SendWaitCnt
 * counter request
 * @return: return SendWaitCnt/SendWaitVlCnt counter value per vl.
 *
 * Convert SendWaitCnt/SendWaitVlCnt counter from TXE cycle times to
 * flit times. Call this function to samples these counters. This
 * function will calculate for previous state transition and update
 * current state at end of function using ppd->prev_link_width and
 * ppd->port_vl_xmit_wait_last to port_vl_xmit_wait_curr and link_width.
 */
u64 get_xmit_wait_counters(struct hfi1_pportdata *ppd,
			   u16 link_width, u16 link_speed, int vl)
{
	u64 port_vl_xmit_wait_curr;
	u64 delta_vl_xmit_wait;
	u64 xmit_wait_val;

	if (vl > C_VL_COUNT)
		return  0;
	if (vl < C_VL_COUNT)
		port_vl_xmit_wait_curr =
			read_port_cntr(ppd, C_TX_WAIT_VL, vl);
	else
		port_vl_xmit_wait_curr =
			read_port_cntr(ppd, C_TX_WAIT, CNTR_INVALID_VL);

	xmit_wait_val =
		port_vl_xmit_wait_curr -
		ppd->port_vl_xmit_wait_last[vl];
	delta_vl_xmit_wait =
		convert_xmit_counter(xmit_wait_val,
				     ppd->prev_link_width,
				     link_speed);

	ppd->vl_xmit_flit_cnt[vl] += delta_vl_xmit_wait;
	ppd->port_vl_xmit_wait_last[vl] = port_vl_xmit_wait_curr;
	ppd->prev_link_width = link_width;

	return ppd->vl_xmit_flit_cnt[vl];
}

M
Mike Marciniszyn 已提交
2724
static int pma_get_opa_portstatus(struct opa_pma_mad *pmp,
2725 2726
				  struct ib_device *ibdev,
				  u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
2727 2728 2729 2730 2731
{
	struct opa_port_status_req *req =
		(struct opa_port_status_req *)pmp->data;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct opa_port_status_rsp *rsp;
2732
	unsigned long vl_select_mask = be32_to_cpu(req->vl_select_mask);
M
Mike Marciniszyn 已提交
2733 2734 2735 2736
	unsigned long vl;
	size_t response_data_size;
	u32 nports = be32_to_cpu(pmp->mad_hdr.attr_mod) >> 24;
	u8 port_num = req->port_num;
2737
	u8 num_vls = hweight64(vl_select_mask);
M
Mike Marciniszyn 已提交
2738 2739 2740 2741 2742
	struct _vls_pctrs *vlinfo;
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	int vfi;
	u64 tmp, tmp2;
2743 2744
	u16 link_width;
	u16 link_speed;
M
Mike Marciniszyn 已提交
2745 2746 2747 2748 2749 2750 2751 2752

	response_data_size = sizeof(struct opa_port_status_rsp) +
				num_vls * sizeof(struct _vls_pctrs);
	if (response_data_size > sizeof(pmp->data)) {
		pmp->mad_hdr.status |= OPA_PM_STATUS_REQUEST_TOO_LARGE;
		return reply((struct ib_mad_hdr *)pmp);
	}

2753 2754
	if (nports != 1 || (port_num && port_num != port) ||
	    num_vls > OPA_MAX_VLS || (vl_select_mask & ~VL_MASK_ALL)) {
M
Mike Marciniszyn 已提交
2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	memset(pmp->data, 0, sizeof(pmp->data));

	rsp = (struct opa_port_status_rsp *)pmp->data;
	if (port_num)
		rsp->port_num = port_num;
	else
		rsp->port_num = port;

	rsp->port_rcv_constraint_errors =
		cpu_to_be64(read_port_cntr(ppd, C_SW_RCV_CSTR_ERR,
					   CNTR_INVALID_VL));

	hfi1_read_link_quality(dd, &rsp->link_quality_indicator);

2773
	rsp->vl_select_mask = cpu_to_be32((u32)vl_select_mask);
M
Mike Marciniszyn 已提交
2774 2775 2776 2777 2778 2779 2780 2781 2782 2783
	rsp->port_xmit_data = cpu_to_be64(read_dev_cntr(dd, C_DC_XMIT_FLITS,
					  CNTR_INVALID_VL));
	rsp->port_rcv_data = cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_FLITS,
					 CNTR_INVALID_VL));
	rsp->port_xmit_pkts = cpu_to_be64(read_dev_cntr(dd, C_DC_XMIT_PKTS,
					  CNTR_INVALID_VL));
	rsp->port_rcv_pkts = cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_PKTS,
					 CNTR_INVALID_VL));
	rsp->port_multicast_xmit_pkts =
		cpu_to_be64(read_dev_cntr(dd, C_DC_MC_XMIT_PKTS,
2784
					  CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
2785 2786 2787
	rsp->port_multicast_rcv_pkts =
		cpu_to_be64(read_dev_cntr(dd, C_DC_MC_RCV_PKTS,
					  CNTR_INVALID_VL));
2788 2789 2790 2791 2792 2793 2794
	/*
	 * Convert PortXmitWait counter from TXE cycle times
	 * to flit times.
	 */
	link_width =
		tx_link_width(ppd->link_width_downgrade_tx_active);
	link_speed = get_link_speed(ppd->link_speed_active);
M
Mike Marciniszyn 已提交
2795
	rsp->port_xmit_wait =
2796 2797
		cpu_to_be64(get_xmit_wait_counters(ppd, link_width,
						   link_speed, C_VL_COUNT));
M
Mike Marciniszyn 已提交
2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810
	rsp->port_rcv_fecn =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_FCN, CNTR_INVALID_VL));
	rsp->port_rcv_becn =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_BCN, CNTR_INVALID_VL));
	rsp->port_xmit_discards =
		cpu_to_be64(read_port_cntr(ppd, C_SW_XMIT_DSCD,
					   CNTR_INVALID_VL));
	rsp->port_xmit_constraint_errors =
		cpu_to_be64(read_port_cntr(ppd, C_SW_XMIT_CSTR_ERR,
					   CNTR_INVALID_VL));
	rsp->port_rcv_remote_physical_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RMT_PHY_ERR,
					  CNTR_INVALID_VL));
2811 2812 2813
	rsp->local_link_integrity_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RX_REPLAY,
					  CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
2814 2815
	tmp = read_dev_cntr(dd, C_DC_SEQ_CRC_CNT, CNTR_INVALID_VL);
	tmp2 = tmp + read_dev_cntr(dd, C_DC_REINIT_FROM_PEER_CNT,
2816
				   CNTR_INVALID_VL);
M
Mike Marciniszyn 已提交
2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830
	if (tmp2 > (u32)UINT_MAX || tmp2 < tmp) {
		/* overflow/wrapped */
		rsp->link_error_recovery = cpu_to_be32(~0);
	} else {
		rsp->link_error_recovery = cpu_to_be32(tmp2);
	}
	rsp->port_rcv_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_ERR, CNTR_INVALID_VL));
	rsp->excessive_buffer_overruns =
		cpu_to_be64(read_dev_cntr(dd, C_RCV_OVF, CNTR_INVALID_VL));
	rsp->fm_config_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_FM_CFG_ERR,
					  CNTR_INVALID_VL));
	rsp->link_downed = cpu_to_be32(read_port_cntr(ppd, C_SW_LINK_DOWN,
2831
						      CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
2832 2833 2834 2835 2836

	/* rsp->uncorrectable_errors is 8 bits wide, and it pegs at 0xff */
	tmp = read_dev_cntr(dd, C_DC_UNC_ERR, CNTR_INVALID_VL);
	rsp->uncorrectable_errors = tmp < 0x100 ? (tmp & 0xff) : 0xff;

2837
	vlinfo = &rsp->vls[0];
M
Mike Marciniszyn 已提交
2838 2839 2840 2841 2842 2843
	vfi = 0;
	/* The vl_select_mask has been checked above, and we know
	 * that it contains only entries which represent valid VLs.
	 * So in the for_each_set_bit() loop below, we don't need
	 * any additional checks for vl.
	 */
2844
	for_each_set_bit(vl, &vl_select_mask, BITS_PER_LONG) {
M
Mike Marciniszyn 已提交
2845 2846 2847 2848 2849 2850 2851
		memset(vlinfo, 0, sizeof(*vlinfo));

		tmp = read_dev_cntr(dd, C_DC_RX_FLIT_VL, idx_from_vl(vl));
		rsp->vls[vfi].port_vl_rcv_data = cpu_to_be64(tmp);

		rsp->vls[vfi].port_vl_rcv_pkts =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RX_PKT_VL,
2852
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2853 2854 2855

		rsp->vls[vfi].port_vl_xmit_data =
			cpu_to_be64(read_port_cntr(ppd, C_TX_FLIT_VL,
2856
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2857 2858 2859

		rsp->vls[vfi].port_vl_xmit_pkts =
			cpu_to_be64(read_port_cntr(ppd, C_TX_PKT_VL,
2860
						   idx_from_vl(vl)));
2861 2862 2863 2864
		/*
		 * Convert PortVlXmitWait counter from TXE cycle
		 * times to flit times.
		 */
M
Mike Marciniszyn 已提交
2865
		rsp->vls[vfi].port_vl_xmit_wait =
2866 2867 2868
			cpu_to_be64(get_xmit_wait_counters(ppd, link_width,
							   link_speed,
							   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2869 2870 2871

		rsp->vls[vfi].port_vl_rcv_fecn =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_FCN_VL,
2872
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2873 2874 2875

		rsp->vls[vfi].port_vl_rcv_becn =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_BCN_VL,
2876
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2877

2878 2879 2880
		rsp->vls[vfi].port_vl_xmit_discards =
			cpu_to_be64(read_port_cntr(ppd, C_SW_XMIT_DSCD_VL,
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2881 2882 2883 2884
		vlinfo++;
		vfi++;
	}

2885
	a0_portstatus(ppd, rsp);
M
Mike Marciniszyn 已提交
2886 2887 2888 2889 2890 2891 2892

	if (resp_len)
		*resp_len += response_data_size;

	return reply((struct ib_mad_hdr *)pmp);
}

2893 2894
static u64 get_error_counter_summary(struct ib_device *ibdev, u8 port,
				     u8 res_lli, u8 res_ler)
M
Mike Marciniszyn 已提交
2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	u64 error_counter_summary = 0, tmp;

	error_counter_summary += read_port_cntr(ppd, C_SW_RCV_CSTR_ERR,
						CNTR_INVALID_VL);
	/* port_rcv_switch_relay_errors is 0 for HFIs */
	error_counter_summary += read_port_cntr(ppd, C_SW_XMIT_DSCD,
						CNTR_INVALID_VL);
	error_counter_summary += read_port_cntr(ppd, C_SW_XMIT_CSTR_ERR,
						CNTR_INVALID_VL);
	error_counter_summary += read_dev_cntr(dd, C_DC_RMT_PHY_ERR,
2909
					       CNTR_INVALID_VL);
2910
	/* local link integrity must be right-shifted by the lli resolution */
2911 2912
	error_counter_summary += (read_dev_cntr(dd, C_DC_RX_REPLAY,
						CNTR_INVALID_VL) >> res_lli);
2913 2914 2915 2916
	/* link error recovery must b right-shifted by the ler resolution */
	tmp = read_dev_cntr(dd, C_DC_SEQ_CRC_CNT, CNTR_INVALID_VL);
	tmp += read_dev_cntr(dd, C_DC_REINIT_FROM_PEER_CNT, CNTR_INVALID_VL);
	error_counter_summary += (tmp >> res_ler);
M
Mike Marciniszyn 已提交
2917
	error_counter_summary += read_dev_cntr(dd, C_DC_RCV_ERR,
2918
					       CNTR_INVALID_VL);
M
Mike Marciniszyn 已提交
2919 2920
	error_counter_summary += read_dev_cntr(dd, C_RCV_OVF, CNTR_INVALID_VL);
	error_counter_summary += read_dev_cntr(dd, C_DC_FM_CFG_ERR,
2921
					       CNTR_INVALID_VL);
M
Mike Marciniszyn 已提交
2922 2923 2924 2925 2926 2927 2928 2929 2930 2931
	/* ppd->link_downed is a 32-bit value */
	error_counter_summary += read_port_cntr(ppd, C_SW_LINK_DOWN,
						CNTR_INVALID_VL);
	tmp = read_dev_cntr(dd, C_DC_UNC_ERR, CNTR_INVALID_VL);
	/* this is an 8-bit quantity */
	error_counter_summary += tmp < 0x100 ? (tmp & 0xff) : 0xff;

	return error_counter_summary;
}

2932
static void a0_datacounters(struct hfi1_pportdata *ppd, struct _port_dctrs *rsp)
M
Mike Marciniszyn 已提交
2933
{
2934
	if (!is_bx(ppd->dd)) {
M
Mike Marciniszyn 已提交
2935
		unsigned long vl;
2936
		u64 sum_vl_xmit_wait = 0;
2937
		unsigned long vl_all_mask = VL_MASK_ALL;
2938

2939
		for_each_set_bit(vl, &vl_all_mask, BITS_PER_LONG) {
M
Mike Marciniszyn 已提交
2940
			u64 tmp = sum_vl_xmit_wait +
2941 2942
				  read_port_cntr(ppd, C_TX_WAIT_VL,
						 idx_from_vl(vl));
M
Mike Marciniszyn 已提交
2943 2944
			if (tmp < sum_vl_xmit_wait) {
				/* we wrapped */
2945
				sum_vl_xmit_wait = (u64)~0;
M
Mike Marciniszyn 已提交
2946 2947 2948 2949 2950 2951 2952 2953 2954
				break;
			}
			sum_vl_xmit_wait = tmp;
		}
		if (be64_to_cpu(rsp->port_xmit_wait) > sum_vl_xmit_wait)
			rsp->port_xmit_wait = cpu_to_be64(sum_vl_xmit_wait);
	}
}

2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975
static void pma_get_opa_port_dctrs(struct ib_device *ibdev,
				   struct _port_dctrs *rsp)
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);

	rsp->port_xmit_data = cpu_to_be64(read_dev_cntr(dd, C_DC_XMIT_FLITS,
						CNTR_INVALID_VL));
	rsp->port_rcv_data = cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_FLITS,
						CNTR_INVALID_VL));
	rsp->port_xmit_pkts = cpu_to_be64(read_dev_cntr(dd, C_DC_XMIT_PKTS,
						CNTR_INVALID_VL));
	rsp->port_rcv_pkts = cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_PKTS,
						CNTR_INVALID_VL));
	rsp->port_multicast_xmit_pkts =
		cpu_to_be64(read_dev_cntr(dd, C_DC_MC_XMIT_PKTS,
					  CNTR_INVALID_VL));
	rsp->port_multicast_rcv_pkts =
		cpu_to_be64(read_dev_cntr(dd, C_DC_MC_RCV_PKTS,
					  CNTR_INVALID_VL));
}

M
Mike Marciniszyn 已提交
2976
static int pma_get_opa_datacounters(struct opa_pma_mad *pmp,
2977 2978
				    struct ib_device *ibdev,
				    u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989
{
	struct opa_port_data_counters_msg *req =
		(struct opa_port_data_counters_msg *)pmp->data;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	struct _port_dctrs *rsp;
	struct _vls_dctrs *vlinfo;
	size_t response_data_size;
	u32 num_ports;
	u8 lq, num_vls;
2990
	u8 res_lli, res_ler;
M
Mike Marciniszyn 已提交
2991
	u64 port_mask;
2992
	u8 port_num;
M
Mike Marciniszyn 已提交
2993
	unsigned long vl;
2994
	unsigned long vl_select_mask;
M
Mike Marciniszyn 已提交
2995
	int vfi;
2996 2997
	u16 link_width;
	u16 link_speed;
M
Mike Marciniszyn 已提交
2998 2999 3000 3001

	num_ports = be32_to_cpu(pmp->mad_hdr.attr_mod) >> 24;
	num_vls = hweight32(be32_to_cpu(req->vl_select_mask));
	vl_select_mask = be32_to_cpu(req->vl_select_mask);
3002 3003 3004 3005
	res_lli = (u8)(be32_to_cpu(req->resolution) & MSK_LLI) >> MSK_LLI_SFT;
	res_lli = res_lli ? res_lli + ADD_LLI : 0;
	res_ler = (u8)(be32_to_cpu(req->resolution) & MSK_LER) >> MSK_LER_SFT;
	res_ler = res_ler ? res_ler + ADD_LER : 0;
M
Mike Marciniszyn 已提交
3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026

	if (num_ports != 1 || (vl_select_mask & ~VL_MASK_ALL)) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	/* Sanity check */
	response_data_size = sizeof(struct opa_port_data_counters_msg) +
				num_vls * sizeof(struct _vls_dctrs);

	if (response_data_size > sizeof(pmp->data)) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	/*
	 * The bit set in the mask needs to be consistent with the
	 * port the request came in on.
	 */
	port_mask = be64_to_cpu(req->port_select_mask[3]);
	port_num = find_first_bit((unsigned long *)&port_mask,
3027
				  sizeof(port_mask) * 8);
M
Mike Marciniszyn 已提交
3028

3029
	if (port_num != port) {
M
Mike Marciniszyn 已提交
3030 3031 3032 3033
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

3034
	rsp = &req->port[0];
M
Mike Marciniszyn 已提交
3035 3036 3037 3038 3039 3040 3041 3042 3043 3044
	memset(rsp, 0, sizeof(*rsp));

	rsp->port_number = port;
	/*
	 * Note that link_quality_indicator is a 32 bit quantity in
	 * 'datacounters' queries (as opposed to 'portinfo' queries,
	 * where it's a byte).
	 */
	hfi1_read_link_quality(dd, &lq);
	rsp->link_quality_indicator = cpu_to_be32((u32)lq);
3045
	pma_get_opa_port_dctrs(ibdev, rsp);
M
Mike Marciniszyn 已提交
3046

3047 3048 3049 3050 3051 3052 3053
	/*
	 * Convert PortXmitWait counter from TXE
	 * cycle times to flit times.
	 */
	link_width =
		tx_link_width(ppd->link_width_downgrade_tx_active);
	link_speed = get_link_speed(ppd->link_speed_active);
M
Mike Marciniszyn 已提交
3054
	rsp->port_xmit_wait =
3055 3056
		cpu_to_be64(get_xmit_wait_counters(ppd, link_width,
						   link_speed, C_VL_COUNT));
M
Mike Marciniszyn 已提交
3057 3058 3059 3060 3061
	rsp->port_rcv_fecn =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_FCN, CNTR_INVALID_VL));
	rsp->port_rcv_becn =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_BCN, CNTR_INVALID_VL));
	rsp->port_error_counter_summary =
3062 3063
		cpu_to_be64(get_error_counter_summary(ibdev, port,
						      res_lli, res_ler));
M
Mike Marciniszyn 已提交
3064

3065
	vlinfo = &rsp->vls[0];
M
Mike Marciniszyn 已提交
3066 3067 3068 3069 3070 3071
	vfi = 0;
	/* The vl_select_mask has been checked above, and we know
	 * that it contains only entries which represent valid VLs.
	 * So in the for_each_set_bit() loop below, we don't need
	 * any additional checks for vl.
	 */
3072
	for_each_set_bit(vl, &vl_select_mask, BITS_PER_LONG) {
M
Mike Marciniszyn 已提交
3073 3074 3075 3076
		memset(vlinfo, 0, sizeof(*vlinfo));

		rsp->vls[vfi].port_vl_xmit_data =
			cpu_to_be64(read_port_cntr(ppd, C_TX_FLIT_VL,
3077
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
3078 3079 3080

		rsp->vls[vfi].port_vl_rcv_data =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RX_FLIT_VL,
3081
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
3082 3083 3084

		rsp->vls[vfi].port_vl_xmit_pkts =
			cpu_to_be64(read_port_cntr(ppd, C_TX_PKT_VL,
3085
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
3086 3087 3088

		rsp->vls[vfi].port_vl_rcv_pkts =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RX_PKT_VL,
3089
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
3090

3091 3092 3093 3094
		/*
		 * Convert PortVlXmitWait counter from TXE
		 * cycle times to flit times.
		 */
M
Mike Marciniszyn 已提交
3095
		rsp->vls[vfi].port_vl_xmit_wait =
3096 3097 3098
			cpu_to_be64(get_xmit_wait_counters(ppd, link_width,
							   link_speed,
							   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
3099 3100 3101

		rsp->vls[vfi].port_vl_rcv_fecn =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_FCN_VL,
3102
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
3103 3104
		rsp->vls[vfi].port_vl_rcv_becn =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_BCN_VL,
3105
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
3106 3107 3108 3109

		/* rsp->port_vl_xmit_time_cong is 0 for HFIs */
		/* rsp->port_vl_xmit_wasted_bw ??? */
		/* port_vl_xmit_wait_data - TXE (table 13-9 HFI spec) ???
3110 3111
		 * does this differ from rsp->vls[vfi].port_vl_xmit_wait
		 */
M
Mike Marciniszyn 已提交
3112
		/*rsp->vls[vfi].port_vl_mark_fecn =
3113 3114 3115
		 *	cpu_to_be64(read_csr(dd, DCC_PRF_PORT_VL_MARK_FECN_CNT
		 *		+ offset));
		 */
M
Mike Marciniszyn 已提交
3116 3117 3118 3119
		vlinfo++;
		vfi++;
	}

3120
	a0_datacounters(ppd, rsp);
M
Mike Marciniszyn 已提交
3121 3122 3123 3124 3125 3126 3127

	if (resp_len)
		*resp_len += response_data_size;

	return reply((struct ib_mad_hdr *)pmp);
}

3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190
static int pma_get_ib_portcounters_ext(struct ib_pma_mad *pmp,
				       struct ib_device *ibdev, u8 port)
{
	struct ib_pma_portcounters_ext *p = (struct ib_pma_portcounters_ext *)
						pmp->data;
	struct _port_dctrs rsp;

	if (pmp->mad_hdr.attr_mod != 0 || p->port_select != port) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		goto bail;
	}

	memset(&rsp, 0, sizeof(rsp));
	pma_get_opa_port_dctrs(ibdev, &rsp);

	p->port_xmit_data = rsp.port_xmit_data;
	p->port_rcv_data = rsp.port_rcv_data;
	p->port_xmit_packets = rsp.port_xmit_pkts;
	p->port_rcv_packets = rsp.port_rcv_pkts;
	p->port_unicast_xmit_packets = 0;
	p->port_unicast_rcv_packets =  0;
	p->port_multicast_xmit_packets = rsp.port_multicast_xmit_pkts;
	p->port_multicast_rcv_packets = rsp.port_multicast_rcv_pkts;

bail:
	return reply((struct ib_mad_hdr *)pmp);
}

static void pma_get_opa_port_ectrs(struct ib_device *ibdev,
				   struct _port_ectrs *rsp, u8 port)
{
	u64 tmp, tmp2;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);

	tmp = read_dev_cntr(dd, C_DC_SEQ_CRC_CNT, CNTR_INVALID_VL);
	tmp2 = tmp + read_dev_cntr(dd, C_DC_REINIT_FROM_PEER_CNT,
					CNTR_INVALID_VL);
	if (tmp2 > (u32)UINT_MAX || tmp2 < tmp) {
		/* overflow/wrapped */
		rsp->link_error_recovery = cpu_to_be32(~0);
	} else {
		rsp->link_error_recovery = cpu_to_be32(tmp2);
	}

	rsp->link_downed = cpu_to_be32(read_port_cntr(ppd, C_SW_LINK_DOWN,
						CNTR_INVALID_VL));
	rsp->port_rcv_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_ERR, CNTR_INVALID_VL));
	rsp->port_rcv_remote_physical_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RMT_PHY_ERR,
					  CNTR_INVALID_VL));
	rsp->port_rcv_switch_relay_errors = 0;
	rsp->port_xmit_discards =
		cpu_to_be64(read_port_cntr(ppd, C_SW_XMIT_DSCD,
					   CNTR_INVALID_VL));
	rsp->port_xmit_constraint_errors =
		cpu_to_be64(read_port_cntr(ppd, C_SW_XMIT_CSTR_ERR,
					   CNTR_INVALID_VL));
	rsp->port_rcv_constraint_errors =
		cpu_to_be64(read_port_cntr(ppd, C_SW_RCV_CSTR_ERR,
					   CNTR_INVALID_VL));
3191 3192 3193
	rsp->local_link_integrity_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RX_REPLAY,
					  CNTR_INVALID_VL));
3194 3195 3196 3197
	rsp->excessive_buffer_overruns =
		cpu_to_be64(read_dev_cntr(dd, C_RCV_OVF, CNTR_INVALID_VL));
}

M
Mike Marciniszyn 已提交
3198
static int pma_get_opa_porterrors(struct opa_pma_mad *pmp,
3199 3200
				  struct ib_device *ibdev,
				  u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
3201 3202 3203
{
	size_t response_data_size;
	struct _port_ectrs *rsp;
3204
	u8 port_num;
M
Mike Marciniszyn 已提交
3205 3206 3207 3208 3209 3210 3211 3212 3213
	struct opa_port_error_counters64_msg *req;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	u32 num_ports;
	u8 num_pslm;
	u8 num_vls;
	struct hfi1_ibport *ibp;
	struct hfi1_pportdata *ppd;
	struct _vls_ectrs *vlinfo;
	unsigned long vl;
3214
	u64 port_mask, tmp;
3215
	unsigned long vl_select_mask;
M
Mike Marciniszyn 已提交
3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242
	int vfi;

	req = (struct opa_port_error_counters64_msg *)pmp->data;

	num_ports = be32_to_cpu(pmp->mad_hdr.attr_mod) >> 24;

	num_pslm = hweight64(be64_to_cpu(req->port_select_mask[3]));
	num_vls = hweight32(be32_to_cpu(req->vl_select_mask));

	if (num_ports != 1 || num_ports != num_pslm) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	response_data_size = sizeof(struct opa_port_error_counters64_msg) +
				num_vls * sizeof(struct _vls_ectrs);

	if (response_data_size > sizeof(pmp->data)) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}
	/*
	 * The bit set in the mask needs to be consistent with the
	 * port the request came in on.
	 */
	port_mask = be64_to_cpu(req->port_select_mask[3]);
	port_num = find_first_bit((unsigned long *)&port_mask,
3243
				  sizeof(port_mask) * 8);
M
Mike Marciniszyn 已提交
3244

3245
	if (port_num != port) {
M
Mike Marciniszyn 已提交
3246 3247 3248 3249
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

3250
	rsp = &req->port[0];
M
Mike Marciniszyn 已提交
3251 3252 3253 3254 3255

	ibp = to_iport(ibdev, port_num);
	ppd = ppd_from_ibp(ibp);

	memset(rsp, 0, sizeof(*rsp));
3256
	rsp->port_number = port_num;
M
Mike Marciniszyn 已提交
3257

3258
	pma_get_opa_port_ectrs(ibdev, rsp, port_num);
M
Mike Marciniszyn 已提交
3259 3260 3261

	rsp->port_rcv_remote_physical_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RMT_PHY_ERR,
3262
					  CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
3263 3264
	rsp->fm_config_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_FM_CFG_ERR,
3265
					  CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
3266
	tmp = read_dev_cntr(dd, C_DC_UNC_ERR, CNTR_INVALID_VL);
3267

M
Mike Marciniszyn 已提交
3268
	rsp->uncorrectable_errors = tmp < 0x100 ? (tmp & 0xff) : 0xff;
3269 3270
	rsp->port_rcv_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_ERR, CNTR_INVALID_VL));
3271
	vlinfo = &rsp->vls[0];
M
Mike Marciniszyn 已提交
3272 3273
	vfi = 0;
	vl_select_mask = be32_to_cpu(req->vl_select_mask);
3274
	for_each_set_bit(vl, &vl_select_mask, BITS_PER_LONG) {
M
Mike Marciniszyn 已提交
3275
		memset(vlinfo, 0, sizeof(*vlinfo));
3276 3277 3278
		rsp->vls[vfi].port_vl_xmit_discards =
			cpu_to_be64(read_port_cntr(ppd, C_SW_XMIT_DSCD_VL,
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
3279 3280 3281 3282 3283 3284 3285 3286 3287 3288
		vlinfo += 1;
		vfi++;
	}

	if (resp_len)
		*resp_len += response_data_size;

	return reply((struct ib_mad_hdr *)pmp);
}

3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373
static int pma_get_ib_portcounters(struct ib_pma_mad *pmp,
				   struct ib_device *ibdev, u8 port)
{
	struct ib_pma_portcounters *p = (struct ib_pma_portcounters *)
		pmp->data;
	struct _port_ectrs rsp;
	u64 temp_link_overrun_errors;
	u64 temp_64;
	u32 temp_32;

	memset(&rsp, 0, sizeof(rsp));
	pma_get_opa_port_ectrs(ibdev, &rsp, port);

	if (pmp->mad_hdr.attr_mod != 0 || p->port_select != port) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		goto bail;
	}

	p->symbol_error_counter = 0; /* N/A for OPA */

	temp_32 = be32_to_cpu(rsp.link_error_recovery);
	if (temp_32 > 0xFFUL)
		p->link_error_recovery_counter = 0xFF;
	else
		p->link_error_recovery_counter = (u8)temp_32;

	temp_32 = be32_to_cpu(rsp.link_downed);
	if (temp_32 > 0xFFUL)
		p->link_downed_counter = 0xFF;
	else
		p->link_downed_counter = (u8)temp_32;

	temp_64 = be64_to_cpu(rsp.port_rcv_errors);
	if (temp_64 > 0xFFFFUL)
		p->port_rcv_errors = cpu_to_be16(0xFFFF);
	else
		p->port_rcv_errors = cpu_to_be16((u16)temp_64);

	temp_64 = be64_to_cpu(rsp.port_rcv_remote_physical_errors);
	if (temp_64 > 0xFFFFUL)
		p->port_rcv_remphys_errors = cpu_to_be16(0xFFFF);
	else
		p->port_rcv_remphys_errors = cpu_to_be16((u16)temp_64);

	temp_64 = be64_to_cpu(rsp.port_rcv_switch_relay_errors);
	p->port_rcv_switch_relay_errors = cpu_to_be16((u16)temp_64);

	temp_64 = be64_to_cpu(rsp.port_xmit_discards);
	if (temp_64 > 0xFFFFUL)
		p->port_xmit_discards = cpu_to_be16(0xFFFF);
	else
		p->port_xmit_discards = cpu_to_be16((u16)temp_64);

	temp_64 = be64_to_cpu(rsp.port_xmit_constraint_errors);
	if (temp_64 > 0xFFUL)
		p->port_xmit_constraint_errors = 0xFF;
	else
		p->port_xmit_constraint_errors = (u8)temp_64;

	temp_64 = be64_to_cpu(rsp.port_rcv_constraint_errors);
	if (temp_64 > 0xFFUL)
		p->port_rcv_constraint_errors = 0xFFUL;
	else
		p->port_rcv_constraint_errors = (u8)temp_64;

	/* LocalLink: 7:4, BufferOverrun: 3:0 */
	temp_64 = be64_to_cpu(rsp.local_link_integrity_errors);
	if (temp_64 > 0xFUL)
		temp_64 = 0xFUL;

	temp_link_overrun_errors = temp_64 << 4;

	temp_64 = be64_to_cpu(rsp.excessive_buffer_overruns);
	if (temp_64 > 0xFUL)
		temp_64 = 0xFUL;
	temp_link_overrun_errors |= temp_64;

	p->link_overrun_errors = (u8)temp_link_overrun_errors;

	p->vl15_dropped = 0; /* N/A for OPA */

bail:
	return reply((struct ib_mad_hdr *)pmp);
}

M
Mike Marciniszyn 已提交
3374
static int pma_get_opa_errorinfo(struct opa_pma_mad *pmp,
3375 3376
				 struct ib_device *ibdev,
				 u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
3377 3378 3379 3380 3381 3382 3383
{
	size_t response_data_size;
	struct _port_ei *rsp;
	struct opa_port_error_info_msg *req;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	u64 port_mask;
	u32 num_ports;
3384
	u8 port_num;
M
Mike Marciniszyn 已提交
3385 3386 3387 3388
	u8 num_pslm;
	u64 reg;

	req = (struct opa_port_error_info_msg *)pmp->data;
3389
	rsp = &req->port[0];
M
Mike Marciniszyn 已提交
3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414

	num_ports = OPA_AM_NPORT(be32_to_cpu(pmp->mad_hdr.attr_mod));
	num_pslm = hweight64(be64_to_cpu(req->port_select_mask[3]));

	memset(rsp, 0, sizeof(*rsp));

	if (num_ports != 1 || num_ports != num_pslm) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	/* Sanity check */
	response_data_size = sizeof(struct opa_port_error_info_msg);

	if (response_data_size > sizeof(pmp->data)) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	/*
	 * The bit set in the mask needs to be consistent with the port
	 * the request came in on.
	 */
	port_mask = be64_to_cpu(req->port_select_mask[3]);
	port_num = find_first_bit((unsigned long *)&port_mask,
3415
				  sizeof(port_mask) * 8);
M
Mike Marciniszyn 已提交
3416

3417
	if (port_num != port) {
M
Mike Marciniszyn 已提交
3418 3419 3420
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}
3421
	rsp->port_number = port;
M
Mike Marciniszyn 已提交
3422 3423 3424 3425 3426

	/* PortRcvErrorInfo */
	rsp->port_rcv_ei.status_and_code =
		dd->err_info_rcvport.status_and_code;
	memcpy(&rsp->port_rcv_ei.ei.ei1to12.packet_flit1,
3427
	       &dd->err_info_rcvport.packet_flit1, sizeof(u64));
M
Mike Marciniszyn 已提交
3428
	memcpy(&rsp->port_rcv_ei.ei.ei1to12.packet_flit2,
3429
	       &dd->err_info_rcvport.packet_flit2, sizeof(u64));
M
Mike Marciniszyn 已提交
3430 3431 3432 3433

	/* ExcessiverBufferOverrunInfo */
	reg = read_csr(dd, RCV_ERR_INFO);
	if (reg & RCV_ERR_INFO_RCV_EXCESS_BUFFER_OVERRUN_SMASK) {
3434 3435 3436 3437
		/*
		 * if the RcvExcessBufferOverrun bit is set, save SC of
		 * first pkt that encountered an excess buffer overrun
		 */
M
Mike Marciniszyn 已提交
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
		u8 tmp = (u8)reg;

		tmp &=  RCV_ERR_INFO_RCV_EXCESS_BUFFER_OVERRUN_SC_SMASK;
		tmp <<= 2;
		rsp->excessive_buffer_overrun_ei.status_and_sc = tmp;
		/* set the status bit */
		rsp->excessive_buffer_overrun_ei.status_and_sc |= 0x80;
	}

	rsp->port_xmit_constraint_ei.status =
		dd->err_info_xmit_constraint.status;
	rsp->port_xmit_constraint_ei.pkey =
		cpu_to_be16(dd->err_info_xmit_constraint.pkey);
	rsp->port_xmit_constraint_ei.slid =
		cpu_to_be32(dd->err_info_xmit_constraint.slid);

	rsp->port_rcv_constraint_ei.status =
		dd->err_info_rcv_constraint.status;
	rsp->port_rcv_constraint_ei.pkey =
		cpu_to_be16(dd->err_info_rcv_constraint.pkey);
	rsp->port_rcv_constraint_ei.slid =
		cpu_to_be32(dd->err_info_rcv_constraint.slid);

	/* UncorrectableErrorInfo */
	rsp->uncorrectable_ei.status_and_code = dd->err_info_uncorrectable;

	/* FMConfigErrorInfo */
	rsp->fm_config_ei.status_and_code = dd->err_info_fmconfig;

	if (resp_len)
		*resp_len += response_data_size;

	return reply((struct ib_mad_hdr *)pmp);
}

static int pma_set_opa_portstatus(struct opa_pma_mad *pmp,
3474 3475
				  struct ib_device *ibdev,
				  u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
3476 3477 3478 3479 3480 3481 3482 3483 3484
{
	struct opa_clear_port_status *req =
		(struct opa_clear_port_status *)pmp->data;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	u32 nports = be32_to_cpu(pmp->mad_hdr.attr_mod) >> 24;
	u64 portn = be64_to_cpu(req->port_select_mask[3]);
	u32 counter_select = be32_to_cpu(req->counter_select_mask);
3485
	unsigned long vl_select_mask = VL_MASK_ALL; /* clear all per-vl cnts */
M
Mike Marciniszyn 已提交
3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515
	unsigned long vl;

	if ((nports != 1) || (portn != 1 << port)) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}
	/*
	 * only counters returned by pma_get_opa_portstatus() are
	 * handled, so when pma_get_opa_portstatus() gets a fix,
	 * the corresponding change should be made here as well.
	 */

	if (counter_select & CS_PORT_XMIT_DATA)
		write_dev_cntr(dd, C_DC_XMIT_FLITS, CNTR_INVALID_VL, 0);

	if (counter_select & CS_PORT_RCV_DATA)
		write_dev_cntr(dd, C_DC_RCV_FLITS, CNTR_INVALID_VL, 0);

	if (counter_select & CS_PORT_XMIT_PKTS)
		write_dev_cntr(dd, C_DC_XMIT_PKTS, CNTR_INVALID_VL, 0);

	if (counter_select & CS_PORT_RCV_PKTS)
		write_dev_cntr(dd, C_DC_RCV_PKTS, CNTR_INVALID_VL, 0);

	if (counter_select & CS_PORT_MCAST_XMIT_PKTS)
		write_dev_cntr(dd, C_DC_MC_XMIT_PKTS, CNTR_INVALID_VL, 0);

	if (counter_select & CS_PORT_MCAST_RCV_PKTS)
		write_dev_cntr(dd, C_DC_MC_RCV_PKTS, CNTR_INVALID_VL, 0);

3516
	if (counter_select & CS_PORT_XMIT_WAIT) {
M
Mike Marciniszyn 已提交
3517
		write_port_cntr(ppd, C_TX_WAIT, CNTR_INVALID_VL, 0);
3518 3519 3520
		ppd->port_vl_xmit_wait_last[C_VL_COUNT] = 0;
		ppd->vl_xmit_flit_cnt[C_VL_COUNT] = 0;
	}
M
Mike Marciniszyn 已提交
3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535
	/* ignore cs_sw_portCongestion for HFIs */

	if (counter_select & CS_PORT_RCV_FECN)
		write_dev_cntr(dd, C_DC_RCV_FCN, CNTR_INVALID_VL, 0);

	if (counter_select & CS_PORT_RCV_BECN)
		write_dev_cntr(dd, C_DC_RCV_BCN, CNTR_INVALID_VL, 0);

	/* ignore cs_port_xmit_time_cong for HFIs */
	/* ignore cs_port_xmit_wasted_bw for now */
	/* ignore cs_port_xmit_wait_data for now */
	if (counter_select & CS_PORT_RCV_BUBBLE)
		write_dev_cntr(dd, C_DC_RCV_BBL, CNTR_INVALID_VL, 0);

	/* Only applicable for switch */
3536 3537 3538
	/* if (counter_select & CS_PORT_MARK_FECN)
	 *	write_csr(dd, DCC_PRF_PORT_MARK_FECN_CNT, 0);
	 */
M
Mike Marciniszyn 已提交
3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552

	if (counter_select & CS_PORT_RCV_CONSTRAINT_ERRORS)
		write_port_cntr(ppd, C_SW_RCV_CSTR_ERR, CNTR_INVALID_VL, 0);

	/* ignore cs_port_rcv_switch_relay_errors for HFIs */
	if (counter_select & CS_PORT_XMIT_DISCARDS)
		write_port_cntr(ppd, C_SW_XMIT_DSCD, CNTR_INVALID_VL, 0);

	if (counter_select & CS_PORT_XMIT_CONSTRAINT_ERRORS)
		write_port_cntr(ppd, C_SW_XMIT_CSTR_ERR, CNTR_INVALID_VL, 0);

	if (counter_select & CS_PORT_RCV_REMOTE_PHYSICAL_ERRORS)
		write_dev_cntr(dd, C_DC_RMT_PHY_ERR, CNTR_INVALID_VL, 0);

3553
	if (counter_select & CS_LOCAL_LINK_INTEGRITY_ERRORS)
M
Mike Marciniszyn 已提交
3554 3555 3556 3557 3558
		write_dev_cntr(dd, C_DC_RX_REPLAY, CNTR_INVALID_VL, 0);

	if (counter_select & CS_LINK_ERROR_RECOVERY) {
		write_dev_cntr(dd, C_DC_SEQ_CRC_CNT, CNTR_INVALID_VL, 0);
		write_dev_cntr(dd, C_DC_REINIT_FROM_PEER_CNT,
3559
			       CNTR_INVALID_VL, 0);
M
Mike Marciniszyn 已提交
3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578
	}

	if (counter_select & CS_PORT_RCV_ERRORS)
		write_dev_cntr(dd, C_DC_RCV_ERR, CNTR_INVALID_VL, 0);

	if (counter_select & CS_EXCESSIVE_BUFFER_OVERRUNS) {
		write_dev_cntr(dd, C_RCV_OVF, CNTR_INVALID_VL, 0);
		dd->rcv_ovfl_cnt = 0;
	}

	if (counter_select & CS_FM_CONFIG_ERRORS)
		write_dev_cntr(dd, C_DC_FM_CFG_ERR, CNTR_INVALID_VL, 0);

	if (counter_select & CS_LINK_DOWNED)
		write_port_cntr(ppd, C_SW_LINK_DOWN, CNTR_INVALID_VL, 0);

	if (counter_select & CS_UNCORRECTABLE_ERRORS)
		write_dev_cntr(dd, C_DC_UNC_ERR, CNTR_INVALID_VL, 0);

3579
	for_each_set_bit(vl, &vl_select_mask, BITS_PER_LONG) {
M
Mike Marciniszyn 已提交
3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591
		if (counter_select & CS_PORT_XMIT_DATA)
			write_port_cntr(ppd, C_TX_FLIT_VL, idx_from_vl(vl), 0);

		if (counter_select & CS_PORT_RCV_DATA)
			write_dev_cntr(dd, C_DC_RX_FLIT_VL, idx_from_vl(vl), 0);

		if (counter_select & CS_PORT_XMIT_PKTS)
			write_port_cntr(ppd, C_TX_PKT_VL, idx_from_vl(vl), 0);

		if (counter_select & CS_PORT_RCV_PKTS)
			write_dev_cntr(dd, C_DC_RX_PKT_VL, idx_from_vl(vl), 0);

3592
		if (counter_select & CS_PORT_XMIT_WAIT) {
M
Mike Marciniszyn 已提交
3593
			write_port_cntr(ppd, C_TX_WAIT_VL, idx_from_vl(vl), 0);
3594 3595 3596
			ppd->port_vl_xmit_wait_last[idx_from_vl(vl)] = 0;
			ppd->vl_xmit_flit_cnt[idx_from_vl(vl)] = 0;
		}
M
Mike Marciniszyn 已提交
3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610

		/* sw_port_vl_congestion is 0 for HFIs */
		if (counter_select & CS_PORT_RCV_FECN)
			write_dev_cntr(dd, C_DC_RCV_FCN_VL, idx_from_vl(vl), 0);

		if (counter_select & CS_PORT_RCV_BECN)
			write_dev_cntr(dd, C_DC_RCV_BCN_VL, idx_from_vl(vl), 0);

		/* port_vl_xmit_time_cong is 0 for HFIs */
		/* port_vl_xmit_wasted_bw ??? */
		/* port_vl_xmit_wait_data - TXE (table 13-9 HFI spec) ??? */
		if (counter_select & CS_PORT_RCV_BUBBLE)
			write_dev_cntr(dd, C_DC_RCV_BBL_VL, idx_from_vl(vl), 0);

3611 3612 3613
		/* if (counter_select & CS_PORT_MARK_FECN)
		 *     write_csr(dd, DCC_PRF_PORT_VL_MARK_FECN_CNT + offset, 0);
		 */
3614 3615 3616
		if (counter_select & C_SW_XMIT_DSCD_VL)
			write_port_cntr(ppd, C_SW_XMIT_DSCD_VL,
					idx_from_vl(vl), 0);
M
Mike Marciniszyn 已提交
3617 3618 3619 3620 3621 3622 3623 3624 3625
	}

	if (resp_len)
		*resp_len += sizeof(*req);

	return reply((struct ib_mad_hdr *)pmp);
}

static int pma_set_opa_errorinfo(struct opa_pma_mad *pmp,
3626 3627
				 struct ib_device *ibdev,
				 u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
3628 3629 3630 3631 3632 3633
{
	struct _port_ei *rsp;
	struct opa_port_error_info_msg *req;
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	u64 port_mask;
	u32 num_ports;
3634
	u8 port_num;
M
Mike Marciniszyn 已提交
3635 3636 3637 3638
	u8 num_pslm;
	u32 error_info_select;

	req = (struct opa_port_error_info_msg *)pmp->data;
3639
	rsp = &req->port[0];
M
Mike Marciniszyn 已提交
3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656

	num_ports = OPA_AM_NPORT(be32_to_cpu(pmp->mad_hdr.attr_mod));
	num_pslm = hweight64(be64_to_cpu(req->port_select_mask[3]));

	memset(rsp, 0, sizeof(*rsp));

	if (num_ports != 1 || num_ports != num_pslm) {
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	/*
	 * The bit set in the mask needs to be consistent with the port
	 * the request came in on.
	 */
	port_mask = be64_to_cpu(req->port_select_mask[3]);
	port_num = find_first_bit((unsigned long *)&port_mask,
3657
				  sizeof(port_mask) * 8);
M
Mike Marciniszyn 已提交
3658

3659
	if (port_num != port) {
M
Mike Marciniszyn 已提交
3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	error_info_select = be32_to_cpu(req->error_info_select_mask);

	/* PortRcvErrorInfo */
	if (error_info_select & ES_PORT_RCV_ERROR_INFO)
		/* turn off status bit */
		dd->err_info_rcvport.status_and_code &= ~OPA_EI_STATUS_SMASK;

	/* ExcessiverBufferOverrunInfo */
	if (error_info_select & ES_EXCESSIVE_BUFFER_OVERRUN_INFO)
3673 3674 3675 3676
		/*
		 * status bit is essentially kept in the h/w - bit 5 of
		 * RCV_ERR_INFO
		 */
M
Mike Marciniszyn 已提交
3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709
		write_csr(dd, RCV_ERR_INFO,
			  RCV_ERR_INFO_RCV_EXCESS_BUFFER_OVERRUN_SMASK);

	if (error_info_select & ES_PORT_XMIT_CONSTRAINT_ERROR_INFO)
		dd->err_info_xmit_constraint.status &= ~OPA_EI_STATUS_SMASK;

	if (error_info_select & ES_PORT_RCV_CONSTRAINT_ERROR_INFO)
		dd->err_info_rcv_constraint.status &= ~OPA_EI_STATUS_SMASK;

	/* UncorrectableErrorInfo */
	if (error_info_select & ES_UNCORRECTABLE_ERROR_INFO)
		/* turn off status bit */
		dd->err_info_uncorrectable &= ~OPA_EI_STATUS_SMASK;

	/* FMConfigErrorInfo */
	if (error_info_select & ES_FM_CONFIG_ERROR_INFO)
		/* turn off status bit */
		dd->err_info_fmconfig &= ~OPA_EI_STATUS_SMASK;

	if (resp_len)
		*resp_len += sizeof(*req);

	return reply((struct ib_mad_hdr *)pmp);
}

struct opa_congestion_info_attr {
	__be16 congestion_info;
	u8 control_table_cap;	/* Multiple of 64 entry unit CCTs */
	u8 congestion_log_length;
} __packed;

static int __subn_get_opa_cong_info(struct opa_smp *smp, u32 am, u8 *data,
				    struct ib_device *ibdev, u8 port,
3710
				    u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
3711 3712 3713 3714 3715 3716
{
	struct opa_congestion_info_attr *p =
		(struct opa_congestion_info_attr *)data;
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);

3717 3718 3719 3720 3721
	if (smp_length_check(sizeof(*p), max_len)) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

M
Mike Marciniszyn 已提交
3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732
	p->congestion_info = 0;
	p->control_table_cap = ppd->cc_max_table_entries;
	p->congestion_log_length = OPA_CONG_LOG_ELEMS;

	if (resp_len)
		*resp_len += sizeof(*p);

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_get_opa_cong_setting(struct opa_smp *smp, u32 am,
3733
				       u8 *data, struct ib_device *ibdev,
3734
				       u8 port, u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
3735 3736 3737
{
	int i;
	struct opa_congestion_setting_attr *p =
3738
		(struct opa_congestion_setting_attr *)data;
M
Mike Marciniszyn 已提交
3739 3740 3741 3742 3743
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	struct opa_congestion_setting_entry_shadow *entries;
	struct cc_state *cc_state;

3744 3745 3746 3747 3748
	if (smp_length_check(sizeof(*p), max_len)) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

M
Mike Marciniszyn 已提交
3749 3750 3751 3752
	rcu_read_lock();

	cc_state = get_cc_state(ppd);

3753
	if (!cc_state) {
M
Mike Marciniszyn 已提交
3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776
		rcu_read_unlock();
		return reply((struct ib_mad_hdr *)smp);
	}

	entries = cc_state->cong_setting.entries;
	p->port_control = cpu_to_be16(cc_state->cong_setting.port_control);
	p->control_map = cpu_to_be32(cc_state->cong_setting.control_map);
	for (i = 0; i < OPA_MAX_SLS; i++) {
		p->entries[i].ccti_increase = entries[i].ccti_increase;
		p->entries[i].ccti_timer = cpu_to_be16(entries[i].ccti_timer);
		p->entries[i].trigger_threshold =
			entries[i].trigger_threshold;
		p->entries[i].ccti_min = entries[i].ccti_min;
	}

	rcu_read_unlock();

	if (resp_len)
		*resp_len += sizeof(*p);

	return reply((struct ib_mad_hdr *)smp);
}

3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794
/*
 * Apply congestion control information stored in the ppd to the
 * active structure.
 */
static void apply_cc_state(struct hfi1_pportdata *ppd)
{
	struct cc_state *old_cc_state, *new_cc_state;

	new_cc_state = kzalloc(sizeof(*new_cc_state), GFP_KERNEL);
	if (!new_cc_state)
		return;

	/*
	 * Hold the lock for updating *and* to prevent ppd information
	 * from changing during the update.
	 */
	spin_lock(&ppd->cc_state_lock);

3795
	old_cc_state = get_cc_state_protected(ppd);
3796 3797 3798 3799 3800 3801 3802 3803 3804
	if (!old_cc_state) {
		/* never active, or shutting down */
		spin_unlock(&ppd->cc_state_lock);
		kfree(new_cc_state);
		return;
	}

	*new_cc_state = *old_cc_state;

3805 3806 3807 3808 3809
	if (ppd->total_cct_entry)
		new_cc_state->cct.ccti_limit = ppd->total_cct_entry - 1;
	else
		new_cc_state->cct.ccti_limit = 0;

3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821
	memcpy(new_cc_state->cct.entries, ppd->ccti_entries,
	       ppd->total_cct_entry * sizeof(struct ib_cc_table_entry));

	new_cc_state->cong_setting.port_control = IB_CC_CCS_PC_SL_BASED;
	new_cc_state->cong_setting.control_map = ppd->cc_sl_control_map;
	memcpy(new_cc_state->cong_setting.entries, ppd->congestion_entries,
	       OPA_MAX_SLS * sizeof(struct opa_congestion_setting_entry));

	rcu_assign_pointer(ppd->cc_state, new_cc_state);

	spin_unlock(&ppd->cc_state_lock);

3822
	kfree_rcu(old_cc_state, rcu);
3823 3824
}

M
Mike Marciniszyn 已提交
3825 3826
static int __subn_set_opa_cong_setting(struct opa_smp *smp, u32 am, u8 *data,
				       struct ib_device *ibdev, u8 port,
3827
				       u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
3828 3829
{
	struct opa_congestion_setting_attr *p =
3830
		(struct opa_congestion_setting_attr *)data;
M
Mike Marciniszyn 已提交
3831 3832 3833 3834 3835
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	struct opa_congestion_setting_entry_shadow *entries;
	int i;

3836 3837 3838 3839 3840
	if (smp_length_check(sizeof(*p), max_len)) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

3841 3842 3843 3844 3845
	/*
	 * Save details from packet into the ppd.  Hold the cc_state_lock so
	 * our information is consistent with anyone trying to apply the state.
	 */
	spin_lock(&ppd->cc_state_lock);
M
Mike Marciniszyn 已提交
3846 3847 3848 3849 3850 3851 3852 3853 3854 3855
	ppd->cc_sl_control_map = be32_to_cpu(p->control_map);

	entries = ppd->congestion_entries;
	for (i = 0; i < OPA_MAX_SLS; i++) {
		entries[i].ccti_increase = p->entries[i].ccti_increase;
		entries[i].ccti_timer = be16_to_cpu(p->entries[i].ccti_timer);
		entries[i].trigger_threshold =
			p->entries[i].trigger_threshold;
		entries[i].ccti_min = p->entries[i].ccti_min;
	}
3856 3857 3858 3859
	spin_unlock(&ppd->cc_state_lock);

	/* now apply the information */
	apply_cc_state(ppd);
M
Mike Marciniszyn 已提交
3860 3861

	return __subn_get_opa_cong_setting(smp, am, data, ibdev, port,
3862
					   resp_len, max_len);
M
Mike Marciniszyn 已提交
3863 3864 3865 3866
}

static int __subn_get_opa_hfi1_cong_log(struct opa_smp *smp, u32 am,
					u8 *data, struct ib_device *ibdev,
3867
					u8 port, u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
3868 3869 3870 3871
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	struct opa_hfi1_cong_log *cong_log = (struct opa_hfi1_cong_log *)data;
3872
	u64 ts;
M
Mike Marciniszyn 已提交
3873 3874
	int i;

3875
	if (am || smp_length_check(sizeof(*cong_log), max_len)) {
M
Mike Marciniszyn 已提交
3876 3877 3878 3879
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

3880
	spin_lock_irq(&ppd->cc_log_lock);
M
Mike Marciniszyn 已提交
3881 3882 3883 3884 3885 3886 3887 3888 3889

	cong_log->log_type = OPA_CC_LOG_TYPE_HFI;
	cong_log->congestion_flags = 0;
	cong_log->threshold_event_counter =
		cpu_to_be16(ppd->threshold_event_counter);
	memcpy(cong_log->threshold_cong_event_map,
	       ppd->threshold_cong_event_map,
	       sizeof(cong_log->threshold_cong_event_map));
	/* keep timestamp in units of 1.024 usec */
3890
	ts = ktime_get_ns() / 1024;
M
Mike Marciniszyn 已提交
3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901
	cong_log->current_time_stamp = cpu_to_be32(ts);
	for (i = 0; i < OPA_CONG_LOG_ELEMS; i++) {
		struct opa_hfi1_cong_log_event_internal *cce =
			&ppd->cc_events[ppd->cc_mad_idx++];
		if (ppd->cc_mad_idx == OPA_CONG_LOG_ELEMS)
			ppd->cc_mad_idx = 0;
		/*
		 * Entries which are older than twice the time
		 * required to wrap the counter are supposed to
		 * be zeroed (CA10-49 IBTA, release 1.2.1, V1).
		 */
3902
		if ((ts - cce->timestamp) / 2 > U32_MAX)
M
Mike Marciniszyn 已提交
3903 3904 3905
			continue;
		memcpy(cong_log->events[i].local_qp_cn_entry, &cce->lqpn, 3);
		memcpy(cong_log->events[i].remote_qp_number_cn_entry,
3906
		       &cce->rqpn, 3);
M
Mike Marciniszyn 已提交
3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922
		cong_log->events[i].sl_svc_type_cn_entry =
			((cce->sl & 0x1f) << 3) | (cce->svc_type & 0x7);
		cong_log->events[i].remote_lid_cn_entry =
			cpu_to_be32(cce->rlid);
		cong_log->events[i].timestamp_cn_entry =
			cpu_to_be32(cce->timestamp);
	}

	/*
	 * Reset threshold_cong_event_map, and threshold_event_counter
	 * to 0 when log is read.
	 */
	memset(ppd->threshold_cong_event_map, 0x0,
	       sizeof(ppd->threshold_cong_event_map));
	ppd->threshold_event_counter = 0;

3923
	spin_unlock_irq(&ppd->cc_log_lock);
M
Mike Marciniszyn 已提交
3924 3925 3926 3927 3928 3929 3930 3931 3932

	if (resp_len)
		*resp_len += sizeof(struct opa_hfi1_cong_log);

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_get_opa_cc_table(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
3933
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
3934 3935
{
	struct ib_cc_table_attr *cc_table_attr =
3936
		(struct ib_cc_table_attr *)data;
M
Mike Marciniszyn 已提交
3937 3938 3939 3940 3941 3942 3943 3944
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	u32 start_block = OPA_AM_START_BLK(am);
	u32 n_blocks = OPA_AM_NBLK(am);
	struct ib_cc_table_entry_shadow *entries;
	int i, j;
	u32 sentry, eentry;
	struct cc_state *cc_state;
3945
	u32 size = sizeof(u16) * (IB_CCT_ENTRIES * n_blocks + 1);
M
Mike Marciniszyn 已提交
3946 3947

	/* sanity check n_blocks, start_block */
3948
	if (n_blocks == 0 || smp_length_check(size, max_len) ||
M
Mike Marciniszyn 已提交
3949 3950 3951 3952 3953 3954 3955 3956 3957
	    start_block + n_blocks > ppd->cc_max_table_entries) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	rcu_read_lock();

	cc_state = get_cc_state(ppd);

3958
	if (!cc_state) {
M
Mike Marciniszyn 已提交
3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977
		rcu_read_unlock();
		return reply((struct ib_mad_hdr *)smp);
	}

	sentry = start_block * IB_CCT_ENTRIES;
	eentry = sentry + (IB_CCT_ENTRIES * n_blocks);

	cc_table_attr->ccti_limit = cpu_to_be16(cc_state->cct.ccti_limit);

	entries = cc_state->cct.entries;

	/* return n_blocks, though the last block may not be full */
	for (j = 0, i = sentry; i < eentry; j++, i++)
		cc_table_attr->ccti_entries[j].entry =
			cpu_to_be16(entries[i].entry);

	rcu_read_unlock();

	if (resp_len)
3978
		*resp_len += size;
M
Mike Marciniszyn 已提交
3979 3980 3981 3982 3983 3984

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_cc_table(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
3985
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
3986
{
3987
	struct ib_cc_table_attr *p = (struct ib_cc_table_attr *)data;
M
Mike Marciniszyn 已提交
3988 3989 3990 3991 3992 3993 3994 3995
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	u32 start_block = OPA_AM_START_BLK(am);
	u32 n_blocks = OPA_AM_NBLK(am);
	struct ib_cc_table_entry_shadow *entries;
	int i, j;
	u32 sentry, eentry;
	u16 ccti_limit;
3996
	u32 size = sizeof(u16) * (IB_CCT_ENTRIES * n_blocks + 1);
M
Mike Marciniszyn 已提交
3997 3998

	/* sanity check n_blocks, start_block */
3999
	if (n_blocks == 0 || smp_length_check(size, max_len) ||
M
Mike Marciniszyn 已提交
4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015
	    start_block + n_blocks > ppd->cc_max_table_entries) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	sentry = start_block * IB_CCT_ENTRIES;
	eentry = sentry + ((n_blocks - 1) * IB_CCT_ENTRIES) +
		 (be16_to_cpu(p->ccti_limit)) % IB_CCT_ENTRIES + 1;

	/* sanity check ccti_limit */
	ccti_limit = be16_to_cpu(p->ccti_limit);
	if (ccti_limit + 1 > eentry) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

4016 4017 4018 4019
	/*
	 * Save details from packet into the ppd.  Hold the cc_state_lock so
	 * our information is consistent with anyone trying to apply the state.
	 */
M
Mike Marciniszyn 已提交
4020 4021
	spin_lock(&ppd->cc_state_lock);
	ppd->total_cct_entry = ccti_limit + 1;
4022
	entries = ppd->ccti_entries;
M
Mike Marciniszyn 已提交
4023 4024 4025 4026
	for (j = 0, i = sentry; i < eentry; j++, i++)
		entries[i].entry = be16_to_cpu(p->ccti_entries[j].entry);
	spin_unlock(&ppd->cc_state_lock);

4027 4028
	/* now apply the information */
	apply_cc_state(ppd);
M
Mike Marciniszyn 已提交
4029

4030 4031
	return __subn_get_opa_cc_table(smp, am, data, ibdev, port, resp_len,
				       max_len);
M
Mike Marciniszyn 已提交
4032 4033 4034 4035 4036 4037 4038 4039
}

struct opa_led_info {
	__be32 rsvd_led_mask;
	__be32 rsvd;
};

#define OPA_LED_SHIFT	31
4040
#define OPA_LED_MASK	BIT(OPA_LED_SHIFT)
M
Mike Marciniszyn 已提交
4041 4042 4043

static int __subn_get_opa_led_info(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
4044
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
4045 4046
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
4047
	struct hfi1_pportdata *ppd = dd->pport;
4048
	struct opa_led_info *p = (struct opa_led_info *)data;
M
Mike Marciniszyn 已提交
4049
	u32 nport = OPA_AM_NPORT(am);
4050
	u32 is_beaconing_active;
M
Mike Marciniszyn 已提交
4051

4052
	if (nport != 1 || smp_length_check(sizeof(*p), max_len)) {
M
Mike Marciniszyn 已提交
4053 4054 4055 4056
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

4057
	/*
4058 4059 4060
	 * This pairs with the memory barrier in hfi1_start_led_override to
	 * ensure that we read the correct state of LED beaconing represented
	 * by led_override_timer_active
4061
	 */
4062
	smp_rmb();
4063 4064
	is_beaconing_active = !!atomic_read(&ppd->led_override_timer_active);
	p->rsvd_led_mask = cpu_to_be32(is_beaconing_active << OPA_LED_SHIFT);
M
Mike Marciniszyn 已提交
4065 4066 4067 4068 4069 4070 4071 4072 4073

	if (resp_len)
		*resp_len += sizeof(struct opa_led_info);

	return reply((struct ib_mad_hdr *)smp);
}

static int __subn_set_opa_led_info(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
4074
				   u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
4075 4076
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
4077
	struct opa_led_info *p = (struct opa_led_info *)data;
M
Mike Marciniszyn 已提交
4078 4079 4080
	u32 nport = OPA_AM_NPORT(am);
	int on = !!(be32_to_cpu(p->rsvd_led_mask) & OPA_LED_MASK);

4081
	if (nport != 1 || smp_length_check(sizeof(*p), max_len)) {
M
Mike Marciniszyn 已提交
4082 4083 4084 4085
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

4086
	if (on)
4087
		hfi1_start_led_override(dd->pport, 2000, 1500);
4088
	else
4089
		shutdown_led_override(dd->pport);
M
Mike Marciniszyn 已提交
4090

4091 4092
	return __subn_get_opa_led_info(smp, am, data, ibdev, port, resp_len,
				       max_len);
M
Mike Marciniszyn 已提交
4093 4094 4095 4096
}

static int subn_get_opa_sma(__be16 attr_id, struct opa_smp *smp, u32 am,
			    u8 *data, struct ib_device *ibdev, u8 port,
4097
			    u32 *resp_len, u32 max_len)
M
Mike Marciniszyn 已提交
4098 4099 4100 4101 4102 4103 4104
{
	int ret;
	struct hfi1_ibport *ibp = to_iport(ibdev, port);

	switch (attr_id) {
	case IB_SMP_ATTR_NODE_DESC:
		ret = __subn_get_opa_nodedesc(smp, am, data, ibdev, port,
4105
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4106 4107 4108
		break;
	case IB_SMP_ATTR_NODE_INFO:
		ret = __subn_get_opa_nodeinfo(smp, am, data, ibdev, port,
4109
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4110 4111 4112
		break;
	case IB_SMP_ATTR_PORT_INFO:
		ret = __subn_get_opa_portinfo(smp, am, data, ibdev, port,
4113
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4114 4115 4116
		break;
	case IB_SMP_ATTR_PKEY_TABLE:
		ret = __subn_get_opa_pkeytable(smp, am, data, ibdev, port,
4117
					       resp_len, max_len);
M
Mike Marciniszyn 已提交
4118 4119 4120
		break;
	case OPA_ATTRIB_ID_SL_TO_SC_MAP:
		ret = __subn_get_opa_sl_to_sc(smp, am, data, ibdev, port,
4121
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4122 4123 4124
		break;
	case OPA_ATTRIB_ID_SC_TO_SL_MAP:
		ret = __subn_get_opa_sc_to_sl(smp, am, data, ibdev, port,
4125
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4126 4127 4128
		break;
	case OPA_ATTRIB_ID_SC_TO_VLT_MAP:
		ret = __subn_get_opa_sc_to_vlt(smp, am, data, ibdev, port,
4129
					       resp_len, max_len);
M
Mike Marciniszyn 已提交
4130 4131 4132
		break;
	case OPA_ATTRIB_ID_SC_TO_VLNT_MAP:
		ret = __subn_get_opa_sc_to_vlnt(smp, am, data, ibdev, port,
4133
						resp_len, max_len);
M
Mike Marciniszyn 已提交
4134 4135 4136
		break;
	case OPA_ATTRIB_ID_PORT_STATE_INFO:
		ret = __subn_get_opa_psi(smp, am, data, ibdev, port,
4137
					 resp_len, max_len);
M
Mike Marciniszyn 已提交
4138 4139 4140
		break;
	case OPA_ATTRIB_ID_BUFFER_CONTROL_TABLE:
		ret = __subn_get_opa_bct(smp, am, data, ibdev, port,
4141
					 resp_len, max_len);
M
Mike Marciniszyn 已提交
4142 4143 4144
		break;
	case OPA_ATTRIB_ID_CABLE_INFO:
		ret = __subn_get_opa_cable_info(smp, am, data, ibdev, port,
4145
						resp_len, max_len);
M
Mike Marciniszyn 已提交
4146 4147 4148
		break;
	case IB_SMP_ATTR_VL_ARB_TABLE:
		ret = __subn_get_opa_vl_arb(smp, am, data, ibdev, port,
4149
					    resp_len, max_len);
M
Mike Marciniszyn 已提交
4150 4151 4152
		break;
	case OPA_ATTRIB_ID_CONGESTION_INFO:
		ret = __subn_get_opa_cong_info(smp, am, data, ibdev, port,
4153
					       resp_len, max_len);
M
Mike Marciniszyn 已提交
4154 4155 4156
		break;
	case OPA_ATTRIB_ID_HFI_CONGESTION_SETTING:
		ret = __subn_get_opa_cong_setting(smp, am, data, ibdev,
4157
						  port, resp_len, max_len);
M
Mike Marciniszyn 已提交
4158 4159 4160
		break;
	case OPA_ATTRIB_ID_HFI_CONGESTION_LOG:
		ret = __subn_get_opa_hfi1_cong_log(smp, am, data, ibdev,
4161
						   port, resp_len, max_len);
M
Mike Marciniszyn 已提交
4162 4163 4164
		break;
	case OPA_ATTRIB_ID_CONGESTION_CONTROL_TABLE:
		ret = __subn_get_opa_cc_table(smp, am, data, ibdev, port,
4165
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4166 4167 4168
		break;
	case IB_SMP_ATTR_LED_INFO:
		ret = __subn_get_opa_led_info(smp, am, data, ibdev, port,
4169
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4170 4171
		break;
	case IB_SMP_ATTR_SM_INFO:
4172
		if (ibp->rvp.port_cap_flags & IB_PORT_SM_DISABLED)
M
Mike Marciniszyn 已提交
4173
			return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED;
4174
		if (ibp->rvp.port_cap_flags & IB_PORT_SM)
M
Mike Marciniszyn 已提交
4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186
			return IB_MAD_RESULT_SUCCESS;
		/* FALLTHROUGH */
	default:
		smp->status |= IB_SMP_UNSUP_METH_ATTR;
		ret = reply((struct ib_mad_hdr *)smp);
		break;
	}
	return ret;
}

static int subn_set_opa_sma(__be16 attr_id, struct opa_smp *smp, u32 am,
			    u8 *data, struct ib_device *ibdev, u8 port,
4187
			    u32 *resp_len, u32 max_len, int local_mad)
M
Mike Marciniszyn 已提交
4188 4189 4190 4191 4192 4193 4194
{
	int ret;
	struct hfi1_ibport *ibp = to_iport(ibdev, port);

	switch (attr_id) {
	case IB_SMP_ATTR_PORT_INFO:
		ret = __subn_set_opa_portinfo(smp, am, data, ibdev, port,
4195
					      resp_len, max_len, local_mad);
M
Mike Marciniszyn 已提交
4196 4197 4198
		break;
	case IB_SMP_ATTR_PKEY_TABLE:
		ret = __subn_set_opa_pkeytable(smp, am, data, ibdev, port,
4199
					       resp_len, max_len);
M
Mike Marciniszyn 已提交
4200 4201 4202
		break;
	case OPA_ATTRIB_ID_SL_TO_SC_MAP:
		ret = __subn_set_opa_sl_to_sc(smp, am, data, ibdev, port,
4203
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4204 4205 4206
		break;
	case OPA_ATTRIB_ID_SC_TO_SL_MAP:
		ret = __subn_set_opa_sc_to_sl(smp, am, data, ibdev, port,
4207
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4208 4209 4210
		break;
	case OPA_ATTRIB_ID_SC_TO_VLT_MAP:
		ret = __subn_set_opa_sc_to_vlt(smp, am, data, ibdev, port,
4211
					       resp_len, max_len);
M
Mike Marciniszyn 已提交
4212 4213 4214
		break;
	case OPA_ATTRIB_ID_SC_TO_VLNT_MAP:
		ret = __subn_set_opa_sc_to_vlnt(smp, am, data, ibdev, port,
4215
						resp_len, max_len);
M
Mike Marciniszyn 已提交
4216 4217 4218
		break;
	case OPA_ATTRIB_ID_PORT_STATE_INFO:
		ret = __subn_set_opa_psi(smp, am, data, ibdev, port,
4219
					 resp_len, max_len, local_mad);
M
Mike Marciniszyn 已提交
4220 4221 4222
		break;
	case OPA_ATTRIB_ID_BUFFER_CONTROL_TABLE:
		ret = __subn_set_opa_bct(smp, am, data, ibdev, port,
4223
					 resp_len, max_len);
M
Mike Marciniszyn 已提交
4224 4225 4226
		break;
	case IB_SMP_ATTR_VL_ARB_TABLE:
		ret = __subn_set_opa_vl_arb(smp, am, data, ibdev, port,
4227
					    resp_len, max_len);
M
Mike Marciniszyn 已提交
4228 4229 4230
		break;
	case OPA_ATTRIB_ID_HFI_CONGESTION_SETTING:
		ret = __subn_set_opa_cong_setting(smp, am, data, ibdev,
4231
						  port, resp_len, max_len);
M
Mike Marciniszyn 已提交
4232 4233 4234
		break;
	case OPA_ATTRIB_ID_CONGESTION_CONTROL_TABLE:
		ret = __subn_set_opa_cc_table(smp, am, data, ibdev, port,
4235
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4236 4237 4238
		break;
	case IB_SMP_ATTR_LED_INFO:
		ret = __subn_set_opa_led_info(smp, am, data, ibdev, port,
4239
					      resp_len, max_len);
M
Mike Marciniszyn 已提交
4240 4241
		break;
	case IB_SMP_ATTR_SM_INFO:
4242
		if (ibp->rvp.port_cap_flags & IB_PORT_SM_DISABLED)
M
Mike Marciniszyn 已提交
4243
			return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED;
4244
		if (ibp->rvp.port_cap_flags & IB_PORT_SM)
M
Mike Marciniszyn 已提交
4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293
			return IB_MAD_RESULT_SUCCESS;
		/* FALLTHROUGH */
	default:
		smp->status |= IB_SMP_UNSUP_METH_ATTR;
		ret = reply((struct ib_mad_hdr *)smp);
		break;
	}
	return ret;
}

static inline void set_aggr_error(struct opa_aggregate *ag)
{
	ag->err_reqlength |= cpu_to_be16(0x8000);
}

static int subn_get_opa_aggregate(struct opa_smp *smp,
				  struct ib_device *ibdev, u8 port,
				  u32 *resp_len)
{
	int i;
	u32 num_attr = be32_to_cpu(smp->attr_mod) & 0x000000ff;
	u8 *next_smp = opa_get_smp_data(smp);

	if (num_attr < 1 || num_attr > 117) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	for (i = 0; i < num_attr; i++) {
		struct opa_aggregate *agg;
		size_t agg_data_len;
		size_t agg_size;
		u32 am;

		agg = (struct opa_aggregate *)next_smp;
		agg_data_len = (be16_to_cpu(agg->err_reqlength) & 0x007f) * 8;
		agg_size = sizeof(*agg) + agg_data_len;
		am = be32_to_cpu(agg->attr_mod);

		*resp_len += agg_size;

		if (next_smp + agg_size > ((u8 *)smp) + sizeof(*smp)) {
			smp->status |= IB_SMP_INVALID_FIELD;
			return reply((struct ib_mad_hdr *)smp);
		}

		/* zero the payload for this segment */
		memset(next_smp + sizeof(*agg), 0, agg_data_len);

4294
		(void)subn_get_opa_sma(agg->attr_id, smp, am, agg->data,
4295 4296 4297 4298
				       ibdev, port, NULL, (u32)agg_data_len);

		if (smp->status & IB_SMP_INVALID_FIELD)
			break;
M
Mike Marciniszyn 已提交
4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310
		if (smp->status & ~IB_SMP_DIRECTION) {
			set_aggr_error(agg);
			return reply((struct ib_mad_hdr *)smp);
		}
		next_smp += agg_size;
	}

	return reply((struct ib_mad_hdr *)smp);
}

static int subn_set_opa_aggregate(struct opa_smp *smp,
				  struct ib_device *ibdev, u8 port,
4311
				  u32 *resp_len, int local_mad)
M
Mike Marciniszyn 已提交
4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339
{
	int i;
	u32 num_attr = be32_to_cpu(smp->attr_mod) & 0x000000ff;
	u8 *next_smp = opa_get_smp_data(smp);

	if (num_attr < 1 || num_attr > 117) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	for (i = 0; i < num_attr; i++) {
		struct opa_aggregate *agg;
		size_t agg_data_len;
		size_t agg_size;
		u32 am;

		agg = (struct opa_aggregate *)next_smp;
		agg_data_len = (be16_to_cpu(agg->err_reqlength) & 0x007f) * 8;
		agg_size = sizeof(*agg) + agg_data_len;
		am = be32_to_cpu(agg->attr_mod);

		*resp_len += agg_size;

		if (next_smp + agg_size > ((u8 *)smp) + sizeof(*smp)) {
			smp->status |= IB_SMP_INVALID_FIELD;
			return reply((struct ib_mad_hdr *)smp);
		}

4340
		(void)subn_set_opa_sma(agg->attr_id, smp, am, agg->data,
4341 4342 4343
				       ibdev, port, NULL, (u32)agg_data_len,
				       local_mad);

4344 4345
		if (smp->status & IB_SMP_INVALID_FIELD)
			break;
M
Mike Marciniszyn 已提交
4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382
		if (smp->status & ~IB_SMP_DIRECTION) {
			set_aggr_error(agg);
			return reply((struct ib_mad_hdr *)smp);
		}
		next_smp += agg_size;
	}

	return reply((struct ib_mad_hdr *)smp);
}

/*
 * OPAv1 specifies that, on the transition to link up, these counters
 * are cleared:
 *   PortRcvErrors [*]
 *   LinkErrorRecovery
 *   LocalLinkIntegrityErrors
 *   ExcessiveBufferOverruns [*]
 *
 * [*] Error info associated with these counters is retained, but the
 * error info status is reset to 0.
 */
void clear_linkup_counters(struct hfi1_devdata *dd)
{
	/* PortRcvErrors */
	write_dev_cntr(dd, C_DC_RCV_ERR, CNTR_INVALID_VL, 0);
	dd->err_info_rcvport.status_and_code &= ~OPA_EI_STATUS_SMASK;
	/* LinkErrorRecovery */
	write_dev_cntr(dd, C_DC_SEQ_CRC_CNT, CNTR_INVALID_VL, 0);
	write_dev_cntr(dd, C_DC_REINIT_FROM_PEER_CNT, CNTR_INVALID_VL, 0);
	/* LocalLinkIntegrityErrors */
	write_dev_cntr(dd, C_DC_RX_REPLAY, CNTR_INVALID_VL, 0);
	/* ExcessiveBufferOverruns */
	write_dev_cntr(dd, C_RCV_OVF, CNTR_INVALID_VL, 0);
	dd->rcv_ovfl_cnt = 0;
	dd->err_info_xmit_constraint.status &= ~OPA_EI_STATUS_SMASK;
}

4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394
static int is_full_mgmt_pkey_in_table(struct hfi1_ibport *ibp)
{
	unsigned int i;
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);

	for (i = 0; i < ARRAY_SIZE(ppd->pkeys); ++i)
		if (ppd->pkeys[i] == FULL_MGMT_P_KEY)
			return 1;

	return 0;
}

M
Mike Marciniszyn 已提交
4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453
/*
 * is_local_mad() returns 1 if 'mad' is sent from, and destined to the
 * local node, 0 otherwise.
 */
static int is_local_mad(struct hfi1_ibport *ibp, const struct opa_mad *mad,
			const struct ib_wc *in_wc)
{
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	const struct opa_smp *smp = (const struct opa_smp *)mad;

	if (smp->mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) {
		return (smp->hop_cnt == 0 &&
			smp->route.dr.dr_slid == OPA_LID_PERMISSIVE &&
			smp->route.dr.dr_dlid == OPA_LID_PERMISSIVE);
	}

	return (in_wc->slid == ppd->lid);
}

/*
 * opa_local_smp_check() should only be called on MADs for which
 * is_local_mad() returns true. It applies the SMP checks that are
 * specific to SMPs which are sent from, and destined to this node.
 * opa_local_smp_check() returns 0 if the SMP passes its checks, 1
 * otherwise.
 *
 * SMPs which arrive from other nodes are instead checked by
 * opa_smp_check().
 */
static int opa_local_smp_check(struct hfi1_ibport *ibp,
			       const struct ib_wc *in_wc)
{
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	u16 pkey;

	if (in_wc->pkey_index >= ARRAY_SIZE(ppd->pkeys))
		return 1;

	pkey = ppd->pkeys[in_wc->pkey_index];
	/*
	 * We need to do the "node-local" checks specified in OPAv1,
	 * rev 0.90, section 9.10.26, which are:
	 *   - pkey is 0x7fff, or 0xffff
	 *   - Source QPN == 0 || Destination QPN == 0
	 *   - the MAD header's management class is either
	 *     IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE or
	 *     IB_MGMT_CLASS_SUBN_LID_ROUTED
	 *   - SLID != 0
	 *
	 * However, we know (and so don't need to check again) that,
	 * for local SMPs, the MAD stack passes MADs with:
	 *   - Source QPN of 0
	 *   - MAD mgmt_class is IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE
	 *   - SLID is either: OPA_LID_PERMISSIVE (0xFFFFFFFF), or
	 *     our own port's lid
	 *
	 */
	if (pkey == LIM_MGMT_P_KEY || pkey == FULL_MGMT_P_KEY)
		return 0;
4454
	ingress_pkey_table_fail(ppd, pkey, in_wc->slid);
M
Mike Marciniszyn 已提交
4455 4456 4457
	return 1;
}

4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514
/**
 * hfi1_pkey_validation_pma - It validates PKEYs for incoming PMA MAD packets.
 * @ibp: IB port data
 * @in_mad: MAD packet with header and data
 * @in_wc: Work completion data such as source LID, port number, etc.
 *
 * These are all the possible logic rules for validating a pkey:
 *
 * a) If pkey neither FULL_MGMT_P_KEY nor LIM_MGMT_P_KEY,
 *    and NOT self-originated packet:
 *     Drop MAD packet as it should always be part of the
 *     management partition unless it's a self-originated packet.
 *
 * b) If pkey_index -> FULL_MGMT_P_KEY, and LIM_MGMT_P_KEY in pkey table:
 *     The packet is coming from a management node and the receiving node
 *     is also a management node, so it is safe for the packet to go through.
 *
 * c) If pkey_index -> FULL_MGMT_P_KEY, and LIM_MGMT_P_KEY is NOT in pkey table:
 *     Drop the packet as LIM_MGMT_P_KEY should always be in the pkey table.
 *     It could be an FM misconfiguration.
 *
 * d) If pkey_index -> LIM_MGMT_P_KEY and FULL_MGMT_P_KEY is NOT in pkey table:
 *     It is safe for the packet to go through since a non-management node is
 *     talking to another non-management node.
 *
 * e) If pkey_index -> LIM_MGMT_P_KEY and FULL_MGMT_P_KEY in pkey table:
 *     Drop the packet because a non-management node is talking to a
 *     management node, and it could be an attack.
 *
 * For the implementation, these rules can be simplied to only checking
 * for (a) and (e). There's no need to check for rule (b) as
 * the packet doesn't need to be dropped. Rule (c) is not possible in
 * the driver as LIM_MGMT_P_KEY is always in the pkey table.
 *
 * Return:
 * 0 - pkey is okay, -EINVAL it's a bad pkey
 */
static int hfi1_pkey_validation_pma(struct hfi1_ibport *ibp,
				    const struct opa_mad *in_mad,
				    const struct ib_wc *in_wc)
{
	u16 pkey_value = hfi1_lookup_pkey_value(ibp, in_wc->pkey_index);

	/* Rule (a) from above */
	if (!is_local_mad(ibp, in_mad, in_wc) &&
	    pkey_value != LIM_MGMT_P_KEY &&
	    pkey_value != FULL_MGMT_P_KEY)
		return -EINVAL;

	/* Rule (e) from above */
	if (pkey_value == LIM_MGMT_P_KEY &&
	    is_full_mgmt_pkey_in_table(ibp))
		return -EINVAL;

	return 0;
}

M
Mike Marciniszyn 已提交
4515 4516 4517
static int process_subn_opa(struct ib_device *ibdev, int mad_flags,
			    u8 port, const struct opa_mad *in_mad,
			    struct opa_mad *out_mad,
4518
			    u32 *resp_len, int local_mad)
M
Mike Marciniszyn 已提交
4519 4520 4521 4522
{
	struct opa_smp *smp = (struct opa_smp *)out_mad;
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	u8 *data;
4523
	u32 am, data_size;
M
Mike Marciniszyn 已提交
4524 4525 4526 4527 4528
	__be16 attr_id;
	int ret;

	*out_mad = *in_mad;
	data = opa_get_smp_data(smp);
4529
	data_size = (u32)opa_get_smp_data_size(smp);
M
Mike Marciniszyn 已提交
4530 4531 4532

	am = be32_to_cpu(smp->attr_mod);
	attr_id = smp->attr_id;
4533
	if (smp->class_version != OPA_SM_CLASS_VERSION) {
M
Mike Marciniszyn 已提交
4534 4535
		smp->status |= IB_SMP_UNSUP_VERSION;
		ret = reply((struct ib_mad_hdr *)smp);
4536
		return ret;
M
Mike Marciniszyn 已提交
4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555
	}
	ret = check_mkey(ibp, (struct ib_mad_hdr *)smp, mad_flags, smp->mkey,
			 smp->route.dr.dr_slid, smp->route.dr.return_path,
			 smp->hop_cnt);
	if (ret) {
		u32 port_num = be32_to_cpu(smp->attr_mod);

		/*
		 * If this is a get/set portinfo, we already check the
		 * M_Key if the MAD is for another port and the M_Key
		 * is OK on the receiving port. This check is needed
		 * to increment the error counters when the M_Key
		 * fails to match on *both* ports.
		 */
		if (attr_id == IB_SMP_ATTR_PORT_INFO &&
		    (smp->method == IB_MGMT_METHOD_GET ||
		     smp->method == IB_MGMT_METHOD_SET) &&
		    port_num && port_num <= ibdev->phys_port_cnt &&
		    port != port_num)
4556
			(void)check_mkey(to_iport(ibdev, port_num),
M
Mike Marciniszyn 已提交
4557 4558 4559 4560 4561
					  (struct ib_mad_hdr *)smp, 0,
					  smp->mkey, smp->route.dr.dr_slid,
					  smp->route.dr.return_path,
					  smp->hop_cnt);
		ret = IB_MAD_RESULT_FAILURE;
4562
		return ret;
M
Mike Marciniszyn 已提交
4563 4564 4565 4566 4567 4568 4569 4570 4571 4572
	}

	*resp_len = opa_get_smp_header_size(smp);

	switch (smp->method) {
	case IB_MGMT_METHOD_GET:
		switch (attr_id) {
		default:
			clear_opa_smp_data(smp);
			ret = subn_get_opa_sma(attr_id, smp, am, data,
4573 4574
					       ibdev, port, resp_len,
					       data_size);
4575
			break;
M
Mike Marciniszyn 已提交
4576 4577 4578
		case OPA_ATTRIB_ID_AGGREGATE:
			ret = subn_get_opa_aggregate(smp, ibdev, port,
						     resp_len);
4579
			break;
M
Mike Marciniszyn 已提交
4580
		}
4581
		break;
M
Mike Marciniszyn 已提交
4582 4583 4584 4585
	case IB_MGMT_METHOD_SET:
		switch (attr_id) {
		default:
			ret = subn_set_opa_sma(attr_id, smp, am, data,
4586
					       ibdev, port, resp_len,
4587
					       data_size, local_mad);
4588
			break;
M
Mike Marciniszyn 已提交
4589 4590
		case OPA_ATTRIB_ID_AGGREGATE:
			ret = subn_set_opa_aggregate(smp, ibdev, port,
4591
						     resp_len, local_mad);
4592
			break;
M
Mike Marciniszyn 已提交
4593
		}
4594
		break;
M
Mike Marciniszyn 已提交
4595 4596 4597 4598 4599 4600 4601 4602 4603 4604
	case IB_MGMT_METHOD_TRAP:
	case IB_MGMT_METHOD_REPORT:
	case IB_MGMT_METHOD_REPORT_RESP:
	case IB_MGMT_METHOD_GET_RESP:
		/*
		 * The ib_mad module will call us to process responses
		 * before checking for other consumers.
		 * Just tell the caller to process it normally.
		 */
		ret = IB_MAD_RESULT_SUCCESS;
4605
		break;
4606 4607 4608 4609 4610
	case IB_MGMT_METHOD_TRAP_REPRESS:
		subn_handle_opa_trap_repress(ibp, smp);
		/* Always successful */
		ret = IB_MAD_RESULT_SUCCESS;
		break;
M
Mike Marciniszyn 已提交
4611 4612 4613
	default:
		smp->status |= IB_SMP_UNSUP_METHOD;
		ret = reply((struct ib_mad_hdr *)smp);
4614
		break;
M
Mike Marciniszyn 已提交
4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631
	}

	return ret;
}

static int process_subn(struct ib_device *ibdev, int mad_flags,
			u8 port, const struct ib_mad *in_mad,
			struct ib_mad *out_mad)
{
	struct ib_smp *smp = (struct ib_smp *)out_mad;
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	int ret;

	*out_mad = *in_mad;
	if (smp->class_version != 1) {
		smp->status |= IB_SMP_UNSUP_VERSION;
		ret = reply((struct ib_mad_hdr *)smp);
4632
		return ret;
M
Mike Marciniszyn 已提交
4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652
	}

	ret = check_mkey(ibp, (struct ib_mad_hdr *)smp, mad_flags,
			 smp->mkey, (__force __be32)smp->dr_slid,
			 smp->return_path, smp->hop_cnt);
	if (ret) {
		u32 port_num = be32_to_cpu(smp->attr_mod);

		/*
		 * If this is a get/set portinfo, we already check the
		 * M_Key if the MAD is for another port and the M_Key
		 * is OK on the receiving port. This check is needed
		 * to increment the error counters when the M_Key
		 * fails to match on *both* ports.
		 */
		if (in_mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO &&
		    (smp->method == IB_MGMT_METHOD_GET ||
		     smp->method == IB_MGMT_METHOD_SET) &&
		    port_num && port_num <= ibdev->phys_port_cnt &&
		    port != port_num)
4653
			(void)check_mkey(to_iport(ibdev, port_num),
4654 4655 4656 4657
					 (struct ib_mad_hdr *)smp, 0,
					 smp->mkey,
					 (__force __be32)smp->dr_slid,
					 smp->return_path, smp->hop_cnt);
M
Mike Marciniszyn 已提交
4658
		ret = IB_MAD_RESULT_FAILURE;
4659
		return ret;
M
Mike Marciniszyn 已提交
4660 4661 4662 4663 4664 4665 4666
	}

	switch (smp->method) {
	case IB_MGMT_METHOD_GET:
		switch (smp->attr_id) {
		case IB_SMP_ATTR_NODE_INFO:
			ret = subn_get_nodeinfo(smp, ibdev, port);
4667
			break;
M
Mike Marciniszyn 已提交
4668 4669 4670
		default:
			smp->status |= IB_SMP_UNSUP_METH_ATTR;
			ret = reply((struct ib_mad_hdr *)smp);
4671
			break;
M
Mike Marciniszyn 已提交
4672
		}
4673
		break;
M
Mike Marciniszyn 已提交
4674 4675 4676 4677 4678
	}

	return ret;
}

4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735
static int process_perf(struct ib_device *ibdev, u8 port,
			const struct ib_mad *in_mad,
			struct ib_mad *out_mad)
{
	struct ib_pma_mad *pmp = (struct ib_pma_mad *)out_mad;
	struct ib_class_port_info *cpi = (struct ib_class_port_info *)
						&pmp->data;
	int ret = IB_MAD_RESULT_FAILURE;

	*out_mad = *in_mad;
	if (pmp->mad_hdr.class_version != 1) {
		pmp->mad_hdr.status |= IB_SMP_UNSUP_VERSION;
		ret = reply((struct ib_mad_hdr *)pmp);
		return ret;
	}

	switch (pmp->mad_hdr.method) {
	case IB_MGMT_METHOD_GET:
		switch (pmp->mad_hdr.attr_id) {
		case IB_PMA_PORT_COUNTERS:
			ret = pma_get_ib_portcounters(pmp, ibdev, port);
			break;
		case IB_PMA_PORT_COUNTERS_EXT:
			ret = pma_get_ib_portcounters_ext(pmp, ibdev, port);
			break;
		case IB_PMA_CLASS_PORT_INFO:
			cpi->capability_mask = IB_PMA_CLASS_CAP_EXT_WIDTH;
			ret = reply((struct ib_mad_hdr *)pmp);
			break;
		default:
			pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR;
			ret = reply((struct ib_mad_hdr *)pmp);
			break;
		}
		break;

	case IB_MGMT_METHOD_SET:
		if (pmp->mad_hdr.attr_id) {
			pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR;
			ret = reply((struct ib_mad_hdr *)pmp);
		}
		break;

	case IB_MGMT_METHOD_TRAP:
	case IB_MGMT_METHOD_GET_RESP:
		/*
		 * The ib_mad module will call us to process responses
		 * before checking for other consumers.
		 * Just tell the caller to process it normally.
		 */
		ret = IB_MAD_RESULT_SUCCESS;
		break;

	default:
		pmp->mad_hdr.status |= IB_SMP_UNSUP_METHOD;
		ret = reply((struct ib_mad_hdr *)pmp);
		break;
M
Mike Marciniszyn 已提交
4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749
	}

	return ret;
}

static int process_perf_opa(struct ib_device *ibdev, u8 port,
			    const struct opa_mad *in_mad,
			    struct opa_mad *out_mad, u32 *resp_len)
{
	struct opa_pma_mad *pmp = (struct opa_pma_mad *)out_mad;
	int ret;

	*out_mad = *in_mad;

4750
	if (pmp->mad_hdr.class_version != OPA_SM_CLASS_VERSION) {
M
Mike Marciniszyn 已提交
4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761
		pmp->mad_hdr.status |= IB_SMP_UNSUP_VERSION;
		return reply((struct ib_mad_hdr *)pmp);
	}

	*resp_len = sizeof(pmp->mad_hdr);

	switch (pmp->mad_hdr.method) {
	case IB_MGMT_METHOD_GET:
		switch (pmp->mad_hdr.attr_id) {
		case IB_PMA_CLASS_PORT_INFO:
			ret = pma_get_opa_classportinfo(pmp, ibdev, resp_len);
4762
			break;
M
Mike Marciniszyn 已提交
4763 4764
		case OPA_PM_ATTRIB_ID_PORT_STATUS:
			ret = pma_get_opa_portstatus(pmp, ibdev, port,
4765
						     resp_len);
4766
			break;
M
Mike Marciniszyn 已提交
4767 4768
		case OPA_PM_ATTRIB_ID_DATA_PORT_COUNTERS:
			ret = pma_get_opa_datacounters(pmp, ibdev, port,
4769
						       resp_len);
4770
			break;
M
Mike Marciniszyn 已提交
4771 4772
		case OPA_PM_ATTRIB_ID_ERROR_PORT_COUNTERS:
			ret = pma_get_opa_porterrors(pmp, ibdev, port,
4773
						     resp_len);
4774
			break;
M
Mike Marciniszyn 已提交
4775 4776
		case OPA_PM_ATTRIB_ID_ERROR_INFO:
			ret = pma_get_opa_errorinfo(pmp, ibdev, port,
4777
						    resp_len);
4778
			break;
M
Mike Marciniszyn 已提交
4779 4780 4781
		default:
			pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR;
			ret = reply((struct ib_mad_hdr *)pmp);
4782
			break;
M
Mike Marciniszyn 已提交
4783
		}
4784
		break;
M
Mike Marciniszyn 已提交
4785 4786 4787 4788 4789

	case IB_MGMT_METHOD_SET:
		switch (pmp->mad_hdr.attr_id) {
		case OPA_PM_ATTRIB_ID_CLEAR_PORT_STATUS:
			ret = pma_set_opa_portstatus(pmp, ibdev, port,
4790
						     resp_len);
4791
			break;
M
Mike Marciniszyn 已提交
4792 4793
		case OPA_PM_ATTRIB_ID_ERROR_INFO:
			ret = pma_set_opa_errorinfo(pmp, ibdev, port,
4794
						    resp_len);
4795
			break;
M
Mike Marciniszyn 已提交
4796 4797 4798
		default:
			pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR;
			ret = reply((struct ib_mad_hdr *)pmp);
4799
			break;
M
Mike Marciniszyn 已提交
4800
		}
4801
		break;
M
Mike Marciniszyn 已提交
4802 4803 4804 4805 4806 4807 4808 4809 4810

	case IB_MGMT_METHOD_TRAP:
	case IB_MGMT_METHOD_GET_RESP:
		/*
		 * The ib_mad module will call us to process responses
		 * before checking for other consumers.
		 * Just tell the caller to process it normally.
		 */
		ret = IB_MAD_RESULT_SUCCESS;
4811
		break;
M
Mike Marciniszyn 已提交
4812 4813 4814 4815

	default:
		pmp->mad_hdr.status |= IB_SMP_UNSUP_METHOD;
		ret = reply((struct ib_mad_hdr *)pmp);
4816
		break;
M
Mike Marciniszyn 已提交
4817 4818 4819 4820 4821 4822
	}

	return ret;
}

static int hfi1_process_opa_mad(struct ib_device *ibdev, int mad_flags,
4823 4824 4825 4826 4827
				u8 port, const struct ib_wc *in_wc,
				const struct ib_grh *in_grh,
				const struct opa_mad *in_mad,
				struct opa_mad *out_mad, size_t *out_mad_size,
				u16 *out_mad_pkey_index)
M
Mike Marciniszyn 已提交
4828 4829 4830
{
	int ret;
	int pkey_idx;
4831
	int local_mad = 0;
M
Mike Marciniszyn 已提交
4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845
	u32 resp_len = 0;
	struct hfi1_ibport *ibp = to_iport(ibdev, port);

	pkey_idx = hfi1_lookup_pkey_idx(ibp, LIM_MGMT_P_KEY);
	if (pkey_idx < 0) {
		pr_warn("failed to find limited mgmt pkey, defaulting 0x%x\n",
			hfi1_get_pkey(ibp, 1));
		pkey_idx = 1;
	}
	*out_mad_pkey_index = (u16)pkey_idx;

	switch (in_mad->mad_hdr.mgmt_class) {
	case IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE:
	case IB_MGMT_CLASS_SUBN_LID_ROUTED:
4846 4847
		local_mad = is_local_mad(ibp, in_mad, in_wc);
		if (local_mad) {
M
Mike Marciniszyn 已提交
4848 4849 4850 4851 4852
			ret = opa_local_smp_check(ibp, in_wc);
			if (ret)
				return IB_MAD_RESULT_FAILURE;
		}
		ret = process_subn_opa(ibdev, mad_flags, port, in_mad,
4853
				       out_mad, &resp_len, local_mad);
M
Mike Marciniszyn 已提交
4854 4855
		goto bail;
	case IB_MGMT_CLASS_PERF_MGMT:
4856 4857 4858 4859 4860
		ret = hfi1_pkey_validation_pma(ibp, in_mad, in_wc);
		if (ret)
			return IB_MAD_RESULT_FAILURE;

		ret = process_perf_opa(ibdev, port, in_mad, out_mad, &resp_len);
M
Mike Marciniszyn 已提交
4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887
		goto bail;

	default:
		ret = IB_MAD_RESULT_SUCCESS;
	}

bail:
	if (ret & IB_MAD_RESULT_REPLY)
		*out_mad_size = round_up(resp_len, 8);
	else if (ret & IB_MAD_RESULT_SUCCESS)
		*out_mad_size = in_wc->byte_len - sizeof(struct ib_grh);

	return ret;
}

static int hfi1_process_ib_mad(struct ib_device *ibdev, int mad_flags, u8 port,
			       const struct ib_wc *in_wc,
			       const struct ib_grh *in_grh,
			       const struct ib_mad *in_mad,
			       struct ib_mad *out_mad)
{
	int ret;

	switch (in_mad->mad_hdr.mgmt_class) {
	case IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE:
	case IB_MGMT_CLASS_SUBN_LID_ROUTED:
		ret = process_subn(ibdev, mad_flags, port, in_mad, out_mad);
4888
		break;
4889 4890 4891
	case IB_MGMT_CLASS_PERF_MGMT:
		ret = process_perf(ibdev, port, in_mad, out_mad);
		break;
M
Mike Marciniszyn 已提交
4892 4893
	default:
		ret = IB_MAD_RESULT_SUCCESS;
4894
		break;
M
Mike Marciniszyn 已提交
4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927
	}

	return ret;
}

/**
 * hfi1_process_mad - process an incoming MAD packet
 * @ibdev: the infiniband device this packet came in on
 * @mad_flags: MAD flags
 * @port: the port number this packet came in on
 * @in_wc: the work completion entry for this packet
 * @in_grh: the global route header for this packet
 * @in_mad: the incoming MAD
 * @out_mad: any outgoing MAD reply
 *
 * Returns IB_MAD_RESULT_SUCCESS if this is a MAD that we are not
 * interested in processing.
 *
 * Note that the verbs framework has already done the MAD sanity checks,
 * and hop count/pointer updating for IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE
 * MADs.
 *
 * This is called by the ib_mad module.
 */
int hfi1_process_mad(struct ib_device *ibdev, int mad_flags, u8 port,
		     const struct ib_wc *in_wc, const struct ib_grh *in_grh,
		     const struct ib_mad_hdr *in_mad, size_t in_mad_size,
		     struct ib_mad_hdr *out_mad, size_t *out_mad_size,
		     u16 *out_mad_pkey_index)
{
	switch (in_mad->base_version) {
	case OPA_MGMT_BASE_VERSION:
		if (unlikely(in_mad_size != sizeof(struct opa_mad))) {
4928
			dev_err(ibdev->dev.parent, "invalid in_mad_size\n");
M
Mike Marciniszyn 已提交
4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947
			return IB_MAD_RESULT_FAILURE;
		}
		return hfi1_process_opa_mad(ibdev, mad_flags, port,
					    in_wc, in_grh,
					    (struct opa_mad *)in_mad,
					    (struct opa_mad *)out_mad,
					    out_mad_size,
					    out_mad_pkey_index);
	case IB_MGMT_BASE_VERSION:
		return hfi1_process_ib_mad(ibdev, mad_flags, port,
					  in_wc, in_grh,
					  (const struct ib_mad *)in_mad,
					  (struct ib_mad *)out_mad);
	default:
		break;
	}

	return IB_MAD_RESULT_FAILURE;
}