mad.c 124.8 KB
Newer Older
M
Mike Marciniszyn 已提交
1
/*
J
Jubin John 已提交
2
 * Copyright(c) 2015, 2016 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 49 50 51 52 53 54
 *
 * 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>
#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"
55
#include "qp.h"
M
Mike Marciniszyn 已提交
56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80

/* 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

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);
}

81 82 83 84 85 86 87 88 89 90
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);
}

M
Mike Marciniszyn 已提交
91 92 93 94
static void send_trap(struct hfi1_ibport *ibp, void *data, unsigned len)
{
	struct ib_mad_send_buf *send_buf;
	struct ib_mad_agent *agent;
95
	struct opa_smp *smp;
M
Mike Marciniszyn 已提交
96 97 98 99 100 101
	int ret;
	unsigned long flags;
	unsigned long timeout;
	int pkey_idx;
	u32 qpn = ppd_from_ibp(ibp)->sm_trap_qp;

102
	agent = ibp->rvp.send_agent;
M
Mike Marciniszyn 已提交
103 104 105 106 107 108 109 110
	if (!agent)
		return;

	/* o14-3.2.1 */
	if (ppd_from_ibp(ibp)->lstate != IB_PORT_ACTIVE)
		return;

	/* o14-2 */
111 112
	if (ibp->rvp.trap_timeout && time_before(jiffies,
						 ibp->rvp.trap_timeout))
M
Mike Marciniszyn 已提交
113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
		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;
129
	smp->base_version = OPA_MGMT_BASE_VERSION;
M
Mike Marciniszyn 已提交
130
	smp->mgmt_class = IB_MGMT_CLASS_SUBN_LID_ROUTED;
131
	smp->class_version = OPA_SMI_CLASS_VERSION;
M
Mike Marciniszyn 已提交
132
	smp->method = IB_MGMT_METHOD_TRAP;
133 134
	ibp->rvp.tid++;
	smp->tid = cpu_to_be64(ibp->rvp.tid);
M
Mike Marciniszyn 已提交
135 136
	smp->attr_id = IB_SMP_ATTR_NOTICE;
	/* o14-1: smp->mkey = 0; */
137
	memcpy(smp->route.lid.data, data, len);
M
Mike Marciniszyn 已提交
138

139
	spin_lock_irqsave(&ibp->rvp.lock, flags);
140
	if (!ibp->rvp.sm_ah) {
141
		if (ibp->rvp.sm_lid != be16_to_cpu(IB_LID_PERMISSIVE)) {
M
Mike Marciniszyn 已提交
142 143
			struct ib_ah *ah;

144
			ah = hfi1_create_qp0_ah(ibp, ibp->rvp.sm_lid);
145
			if (IS_ERR(ah)) {
M
Mike Marciniszyn 已提交
146
				ret = PTR_ERR(ah);
147
			} else {
M
Mike Marciniszyn 已提交
148
				send_buf->ah = ah;
149
				ibp->rvp.sm_ah = ibah_to_rvtah(ah);
M
Mike Marciniszyn 已提交
150 151
				ret = 0;
			}
152
		} else {
M
Mike Marciniszyn 已提交
153
			ret = -EINVAL;
154
		}
M
Mike Marciniszyn 已提交
155
	} else {
156
		send_buf->ah = &ibp->rvp.sm_ah->ibah;
M
Mike Marciniszyn 已提交
157 158
		ret = 0;
	}
159
	spin_unlock_irqrestore(&ibp->rvp.lock, flags);
M
Mike Marciniszyn 已提交
160 161 162 163 164

	if (!ret)
		ret = ib_post_send_mad(send_buf, NULL);
	if (!ret) {
		/* 4.096 usec. */
165 166
		timeout = (4096 * (1UL << ibp->rvp.subnet_timeout)) / 1000;
		ibp->rvp.trap_timeout = jiffies + usecs_to_jiffies(timeout);
M
Mike Marciniszyn 已提交
167 168
	} else {
		ib_free_send_mad(send_buf);
169
		ibp->rvp.trap_timeout = 0;
M
Mike Marciniszyn 已提交
170 171 172 173 174 175 176
	}
}

/*
 * Send a bad [PQ]_Key trap (ch. 14.3.8).
 */
void hfi1_bad_pqkey(struct hfi1_ibport *ibp, __be16 trap_num, u32 key, u32 sl,
177
		    u32 qp1, u32 qp2, u16 lid1, u16 lid2)
M
Mike Marciniszyn 已提交
178
{
179 180 181 182
	struct opa_mad_notice_attr data;
	u32 lid = ppd_from_ibp(ibp)->lid;
	u32 _lid1 = lid1;
	u32 _lid2 = lid2;
M
Mike Marciniszyn 已提交
183

184 185 186
	memset(&data, 0, sizeof(data));

	if (trap_num == OPA_TRAP_BAD_P_KEY)
187
		ibp->rvp.pkey_violations++;
M
Mike Marciniszyn 已提交
188
	else
189 190
		ibp->rvp.qkey_violations++;
	ibp->rvp.n_pkt_drops++;
M
Mike Marciniszyn 已提交
191 192 193 194 195

	/* Send violation trap */
	data.generic_type = IB_NOTICE_TYPE_SECURITY;
	data.prod_type_lsb = IB_NOTICE_PROD_CA;
	data.trap_num = trap_num;
196 197 198 199 200 201 202
	data.issuer_lid = cpu_to_be32(lid);
	data.ntc_257_258.lid1 = cpu_to_be32(_lid1);
	data.ntc_257_258.lid2 = cpu_to_be32(_lid2);
	data.ntc_257_258.key = cpu_to_be32(key);
	data.ntc_257_258.sl = sl << 3;
	data.ntc_257_258.qp1 = cpu_to_be32(qp1);
	data.ntc_257_258.qp2 = cpu_to_be32(qp2);
M
Mike Marciniszyn 已提交
203 204 205 206 207 208 209 210 211 212

	send_trap(ibp, &data, sizeof(data));
}

/*
 * 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)
{
213 214
	struct opa_mad_notice_attr data;
	u32 lid = ppd_from_ibp(ibp)->lid;
M
Mike Marciniszyn 已提交
215

216
	memset(&data, 0, sizeof(data));
M
Mike Marciniszyn 已提交
217 218 219
	/* Send violation trap */
	data.generic_type = IB_NOTICE_TYPE_SECURITY;
	data.prod_type_lsb = IB_NOTICE_PROD_CA;
220 221 222 223 224 225 226
	data.trap_num = OPA_TRAP_BAD_M_KEY;
	data.issuer_lid = cpu_to_be32(lid);
	data.ntc_256.lid = data.issuer_lid;
	data.ntc_256.method = mad->method;
	data.ntc_256.attr_id = mad->attr_id;
	data.ntc_256.attr_mod = mad->attr_mod;
	data.ntc_256.mkey = mkey;
M
Mike Marciniszyn 已提交
227
	if (mad->mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) {
228 229 230 231
		data.ntc_256.dr_slid = dr_slid;
		data.ntc_256.dr_trunc_hop = IB_NOTICE_TRAP_DR_NOTICE;
		if (hop_cnt > ARRAY_SIZE(data.ntc_256.dr_rtn_path)) {
			data.ntc_256.dr_trunc_hop |=
M
Mike Marciniszyn 已提交
232
				IB_NOTICE_TRAP_DR_TRUNC;
233
			hop_cnt = ARRAY_SIZE(data.ntc_256.dr_rtn_path);
M
Mike Marciniszyn 已提交
234
		}
235 236
		data.ntc_256.dr_trunc_hop |= hop_cnt;
		memcpy(data.ntc_256.dr_rtn_path, return_path,
M
Mike Marciniszyn 已提交
237 238 239 240 241 242 243 244 245
		       hop_cnt);
	}

	send_trap(ibp, &data, sizeof(data));
}

/*
 * Send a Port Capability Mask Changed trap (ch. 14.3.11).
 */
246
void hfi1_cap_mask_chg(struct rvt_dev_info *rdi, u8 port_num)
M
Mike Marciniszyn 已提交
247
{
248
	struct opa_mad_notice_attr data;
249 250 251
	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;
252 253 254
	u32 lid = ppd_from_ibp(ibp)->lid;

	memset(&data, 0, sizeof(data));
M
Mike Marciniszyn 已提交
255 256 257

	data.generic_type = IB_NOTICE_TYPE_INFO;
	data.prod_type_lsb = IB_NOTICE_PROD_CA;
258 259 260
	data.trap_num = OPA_TRAP_CHANGE_CAPABILITY;
	data.issuer_lid = cpu_to_be32(lid);
	data.ntc_144.lid = data.issuer_lid;
261
	data.ntc_144.new_cap_mask = cpu_to_be32(ibp->rvp.port_cap_flags);
M
Mike Marciniszyn 已提交
262 263 264 265 266 267 268 269 270

	send_trap(ibp, &data, sizeof(data));
}

/*
 * Send a System Image GUID Changed trap (ch. 14.3.12).
 */
void hfi1_sys_guid_chg(struct hfi1_ibport *ibp)
{
271 272 273 274
	struct opa_mad_notice_attr data;
	u32 lid = ppd_from_ibp(ibp)->lid;

	memset(&data, 0, sizeof(data));
M
Mike Marciniszyn 已提交
275 276 277

	data.generic_type = IB_NOTICE_TYPE_INFO;
	data.prod_type_lsb = IB_NOTICE_PROD_CA;
278 279 280 281
	data.trap_num = OPA_TRAP_CHANGE_SYSGUID;
	data.issuer_lid = cpu_to_be32(lid);
	data.ntc_145.new_sys_guid = ib_hfi1_sys_image_guid;
	data.ntc_145.lid = data.issuer_lid;
M
Mike Marciniszyn 已提交
282 283 284 285 286 287 288 289 290

	send_trap(ibp, &data, sizeof(data));
}

/*
 * Send a Node Description Changed trap (ch. 14.3.13).
 */
void hfi1_node_desc_chg(struct hfi1_ibport *ibp)
{
291 292 293 294
	struct opa_mad_notice_attr data;
	u32 lid = ppd_from_ibp(ibp)->lid;

	memset(&data, 0, sizeof(data));
M
Mike Marciniszyn 已提交
295 296 297

	data.generic_type = IB_NOTICE_TYPE_INFO;
	data.prod_type_lsb = IB_NOTICE_PROD_CA;
298 299 300 301 302
	data.trap_num = OPA_TRAP_CHANGE_CAPABILITY;
	data.issuer_lid = cpu_to_be32(lid);
	data.ntc_144.lid = data.issuer_lid;
	data.ntc_144.change_flags =
		cpu_to_be16(OPA_NOTICE_TRAP_NODE_DESC_CHG);
M
Mike Marciniszyn 已提交
303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422

	send_trap(ibp, &data, sizeof(data));
}

static int __subn_get_opa_nodedesc(struct opa_smp *smp, u32 am,
				   u8 *data, struct ib_device *ibdev,
				   u8 port, u32 *resp_len)
{
	struct opa_node_description *nd;

	if (am) {
		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,
				   u32 *resp_len)
{
	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 */
	if (am || pidx >= dd->num_pports || dd->pport[pidx].guid == 0) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

	ni->port_guid = cpu_to_be64(dd->pport[pidx].guid);
	ni->base_version = OPA_MGMT_BASE_VERSION;
	ni->class_version = OPA_SMI_CLASS_VERSION;
	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;
	/* Use first-port GUID as node */
	ni->node_guid = cpu_to_be64(dd->pport->guid);
	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 ||
	    dd->pport[pidx].guid == 0)
		smp->status |= IB_SMP_INVALID_FIELD;
	else
		nip->port_guid = cpu_to_be64(dd->pport[pidx].guid);

	nip->base_version = OPA_MGMT_BASE_VERSION;
	nip->class_version = OPA_SMI_CLASS_VERSION;
	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;
	 /* Use first-port GUID as node */
	nip->node_guid = cpu_to_be64(dd->pport->guid);
	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? */
423 424
	if (ibp->rvp.mkey_lease_timeout &&
	    time_after_eq(jiffies, ibp->rvp.mkey_lease_timeout)) {
M
Mike Marciniszyn 已提交
425
		/* Clear timeout and mkey protection field. */
426 427
		ibp->rvp.mkey_lease_timeout = 0;
		ibp->rvp.mkeyprot = 0;
M
Mike Marciniszyn 已提交
428 429
	}

430 431
	if ((mad_flags & IB_MAD_IGNORE_MKEY) ||  ibp->rvp.mkey == 0 ||
	    ibp->rvp.mkey == mkey)
M
Mike Marciniszyn 已提交
432 433 434
		valid_mkey = 1;

	/* Unset lease timeout on any valid Get/Set/TrapRepress */
435
	if (valid_mkey && ibp->rvp.mkey_lease_timeout &&
M
Mike Marciniszyn 已提交
436 437 438
	    (mad->method == IB_MGMT_METHOD_GET ||
	     mad->method == IB_MGMT_METHOD_SET ||
	     mad->method == IB_MGMT_METHOD_TRAP_REPRESS))
439
		ibp->rvp.mkey_lease_timeout = 0;
M
Mike Marciniszyn 已提交
440 441 442 443 444

	if (!valid_mkey) {
		switch (mad->method) {
		case IB_MGMT_METHOD_GET:
			/* Bad mkey not a violation below level 2 */
445
			if (ibp->rvp.mkeyprot < 2)
M
Mike Marciniszyn 已提交
446 447 448
				break;
		case IB_MGMT_METHOD_SET:
		case IB_MGMT_METHOD_TRAP_REPRESS:
449 450 451 452 453 454
			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
455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528
			/* 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,
				   u32 *resp_len)
{
	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;
529
	u8 is_beaconing_active;
M
Mike Marciniszyn 已提交
530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545
	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;

	if (num_ports != 1) {
		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;

546
	if (ppd->vls_supported / 2 > ARRAY_SIZE(pi->neigh_mtu.pvlx_to_mtu) ||
547
	    ppd->vls_supported > ARRAY_SIZE(dd->vld)) {
M
Mike Marciniszyn 已提交
548 549 550 551 552 553 554 555
		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 &&
556 557 558 559 560 561 562 563
	      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
564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588
	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;

589
	pi->port_phys_conf = (ppd->port_type & 0xf);
M
Mike Marciniszyn 已提交
590 591 592 593

	pi->port_states.ledenable_offlinereason = ppd->neighbor_normal << 4;
	pi->port_states.ledenable_offlinereason |=
		ppd->is_sm_config_started << 5;
594
	/*
595 596 597
	 * 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
598
	 */
599
	smp_rmb();
600 601
	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
602
	pi->port_states.ledenable_offlinereason |=
603
		ppd->offline_disabled_reason;
M
Mike Marciniszyn 已提交
604 605 606 607

	pi->port_states.portphysstate_portstate =
		(hfi1_ibphys_portstate(ppd) << 4) | state;

608
	pi->mkeyprotect_lmc = (ibp->rvp.mkeyprot << 6) | ppd->lmc;
M
Mike Marciniszyn 已提交
609 610 611 612 613

	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)
614
			pi->neigh_mtu.pvlx_to_mtu[i / 2] |= (mtu << 4);
M
Mike Marciniszyn 已提交
615
		else
616
			pi->neigh_mtu.pvlx_to_mtu[i / 2] |= mtu;
M
Mike Marciniszyn 已提交
617 618 619
	}
	/* don't forget VL 15 */
	mtu = mtu_to_enum(dd->vld[15].mtu, 2048);
620
	pi->neigh_mtu.pvlx_to_mtu[15 / 2] |= mtu;
621
	pi->smsl = ibp->rvp.sm_sl & OPA_PI_MASK_SMSL;
M
Mike Marciniszyn 已提交
622 623 624 625 626 627 628
	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;
629
	pi->mkey_violations = cpu_to_be16(ibp->rvp.mkey_violations);
M
Mike Marciniszyn 已提交
630
	/* P_KeyViolations are counted by hardware. */
631 632
	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
633 634

	pi->vl.cap = ppd->vls_supported;
635
	pi->vl.high_limit = cpu_to_be16(ibp->rvp.vl_high_limit);
M
Mike Marciniszyn 已提交
636 637 638
	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);

639
	pi->clientrereg_subnettimeout = ibp->rvp.subnet_timeout;
M
Mike Marciniszyn 已提交
640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709

	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 =
		cpu_to_be16(OPA_PORT_PACKET_FORMAT_9B);
	pi->port_packet_format.enabled =
		cpu_to_be16(OPA_PORT_PACKET_FORMAT_9B);

	/* 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);

	pi->opa_cap_mask = cpu_to_be16(OPA_CAP_MASK3_IsSharedSpaceSupported);

	/* 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);

710 711 712 713
	/*
	 * this counter is 16 bits wide, but the replay_depth.wire
	 * variable is only 8 bits
	 */
M
Mike Marciniszyn 已提交
714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759
	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,
				    u32 *resp_len)
{
	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);
	}

760
	n_blocks_avail = (u16)(npkeys / OPA_PARTITION_TABLE_BLK_SIZE) + 1;
M
Mike Marciniszyn 已提交
761 762 763 764 765 766 767 768 769 770 771 772 773

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

	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);
	}

774
	p = (__be16 *)data;
M
Mike Marciniszyn 已提交
775 776 777 778 779 780 781 782
	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;
783
	} else {
M
Mike Marciniszyn 已提交
784
		smp->status |= IB_SMP_INVALID_FIELD;
785
	}
M
Mike Marciniszyn 已提交
786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911
	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)
{
	u32 physical_old = driver_physical_state(ppd);
	u32 logical_old = driver_logical_state(ppd);
	int ret, logical_allowed, physical_allowed;

912 913
	ret = logical_transition_allowed(logical_old, logical_new);
	logical_allowed = ret;
M
Mike Marciniszyn 已提交
914 915 916 917 918 919 920 921 922

	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;
	}

923 924
	ret = physical_transition_allowed(physical_old, physical_new);
	physical_allowed = ret;
M
Mike Marciniszyn 已提交
925 926 927 928 929 930 931 932 933 934 935 936 937

	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;

938 939 940 941 942 943 944 945
	/*
	 * 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
946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990
	/*
	 * 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,
			   u32 logical_state, u32 phys_state,
			   int suppress_idle_sma)
{
	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:
991
		if (phys_state == IB_PORTPHYSSTATE_NOP) {
M
Mike Marciniszyn 已提交
992
			link_state = HLS_DN_DOWNDEF;
993
		} else if (phys_state == IB_PORTPHYSSTATE_POLLING) {
M
Mike Marciniszyn 已提交
994
			link_state = HLS_DN_POLL;
995 996
			set_link_down_reason(ppd, OPA_LINKDOWN_REASON_FM_BOUNCE,
					     0, OPA_LINKDOWN_REASON_FM_BOUNCE);
997
		} else if (phys_state == IB_PORTPHYSSTATE_DISABLED) {
M
Mike Marciniszyn 已提交
998
			link_state = HLS_DN_DISABLE;
999
		} else {
M
Mike Marciniszyn 已提交
1000 1001 1002 1003 1004 1005
			pr_warn("SubnSet(OPA_PortInfo) invalid physical state 0x%x\n",
				phys_state);
			smp->status |= IB_SMP_INVALID_FIELD;
			break;
		}

1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020
		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);
			tune_serdes(ppd);
			start_link(ppd);
		} else {
			set_link_state(ppd, link_state);
		}
M
Mike Marciniszyn 已提交
1021 1022
		if (link_state == HLS_DN_DISABLE &&
		    (ppd->offline_disabled_reason >
1023
		     HFI1_ODR_MASK(OPA_LINKDOWN_REASON_SMA_DISABLED) ||
M
Mike Marciniszyn 已提交
1024
		     ppd->offline_disabled_reason ==
1025
		     HFI1_ODR_MASK(OPA_LINKDOWN_REASON_NONE)))
M
Mike Marciniszyn 已提交
1026
			ppd->offline_disabled_reason =
1027
			HFI1_ODR_MASK(OPA_LINKDOWN_REASON_SMA_DISABLED);
M
Mike Marciniszyn 已提交
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 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
		/*
		 * Don't send a reply if the response would be sent
		 * through the disabled port.
		 */
		if (link_state == HLS_DN_DISABLE && smp->hop_cnt)
			return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED;
		break;
	case IB_PORT_ARMED:
		ret = set_link_state(ppd, HLS_UP_ARMED);
		if ((ret == 0) && (suppress_idle_sma == 0))
			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,
				   u32 *resp_len)
{
	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;
	u32 smlid, opa_lid; /* tmp vars to hold LID values */
	u16 lid;
	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;

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

	opa_lid = be32_to_cpu(pi->lid);
	if (opa_lid & 0xFFFF0000) {
		pr_warn("OPA_PortInfo lid out of range: %X\n", opa_lid);
		smp->status |= IB_SMP_INVALID_FIELD;
		goto get_only;
	}

	lid = (u16)(opa_lid & 0x0000FFFF);

	smlid = be32_to_cpu(pi->sm_lid);
	if (smlid & 0xFFFF0000) {
		pr_warn("OPA_PortInfo SM lid out of range: %X\n", smlid);
		smp->status |= IB_SMP_INVALID_FIELD;
		goto get_only;
	}
	smlid &= 0x0000FFFF;

	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);

1123 1124 1125
	ibp->rvp.mkey = pi->mkey;
	ibp->rvp.gid_prefix = pi->subnet_prefix;
	ibp->rvp.mkey_lease_period = be16_to_cpu(pi->mkey_lease_period);
M
Mike Marciniszyn 已提交
1126 1127 1128

	/* Must be a valid unicast LID address. */
	if ((lid == 0 && ls_old > IB_PORT_INIT) ||
1129
	    lid >= be16_to_cpu(IB_MULTICAST_LID_BASE)) {
M
Mike Marciniszyn 已提交
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
		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);
	}

	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);
	/* enable/disable SW pkey checking as per FM control */
	if (pi->partenforce_filterraw & OPA_PI_MASK_PARTITION_ENFORCE_IN)
		ppd->part_enforce |= HFI1_PART_ENFORCE_IN;
	else
		ppd->part_enforce &= ~HFI1_PART_ENFORCE_IN;

	if (pi->partenforce_filterraw & OPA_PI_MASK_PARTITION_ENFORCE_OUT)
		ppd->part_enforce |= HFI1_PART_ENFORCE_OUT;
	else
		ppd->part_enforce &= ~HFI1_PART_ENFORCE_OUT;

	/* Must be a valid unicast LID address. */
	if ((smlid == 0 && ls_old > IB_PORT_INIT) ||
1162
	    smlid >= be16_to_cpu(IB_MULTICAST_LID_BASE)) {
M
Mike Marciniszyn 已提交
1163 1164
		smp->status |= IB_SMP_INVALID_FIELD;
		pr_warn("SubnSet(OPA_PortInfo) smlid invalid 0x%x\n", smlid);
1165
	} else if (smlid != ibp->rvp.sm_lid || msl != ibp->rvp.sm_sl) {
M
Mike Marciniszyn 已提交
1166
		pr_warn("SubnSet(OPA_PortInfo) smlid 0x%x\n", smlid);
1167
		spin_lock_irqsave(&ibp->rvp.lock, flags);
1168
		if (ibp->rvp.sm_ah) {
1169
			if (smlid != ibp->rvp.sm_lid)
1170
				ibp->rvp.sm_ah->attr.dlid = smlid;
1171
			if (msl != ibp->rvp.sm_sl)
1172
				ibp->rvp.sm_ah->attr.sl = msl;
M
Mike Marciniszyn 已提交
1173
		}
1174 1175 1176 1177 1178
		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
1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
		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) {
1199 1200
		if (lwe == OPA_LINK_WIDTH_RESET ||
		    lwe == OPA_LINK_WIDTH_RESET_OLD)
M
Mike Marciniszyn 已提交
1201 1202 1203 1204 1205 1206 1207 1208
			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" */
1209 1210
	if (lwe == OPA_LINK_WIDTH_RESET ||
	    lwe == OPA_LINK_WIDTH_RESET_OLD) {
M
Mike Marciniszyn 已提交
1211
		set_link_width_downgrade_enabled(ppd,
1212 1213 1214
						 ppd->
						 link_width_downgrade_supported
						 );
M
Mike Marciniszyn 已提交
1215 1216 1217 1218 1219 1220
	} 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;
		}
1221
	} else {
M
Mike Marciniszyn 已提交
1222
		smp->status |= IB_SMP_INVALID_FIELD;
1223
	}
M
Mike Marciniszyn 已提交
1224 1225 1226 1227 1228 1229 1230 1231
	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;
	}

1232 1233 1234
	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
1235
	(void)hfi1_set_ib_cfg(ppd, HFI1_IB_CFG_VL_HIGH_LIMIT,
1236
				    ibp->rvp.vl_high_limit);
M
Mike Marciniszyn 已提交
1237

1238
	if (ppd->vls_supported / 2 > ARRAY_SIZE(pi->neigh_mtu.pvlx_to_mtu) ||
1239
	    ppd->vls_supported > ARRAY_SIZE(dd->vld)) {
M
Mike Marciniszyn 已提交
1240 1241 1242 1243 1244
		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)
1245 1246
			mtu = enum_to_mtu((pi->neigh_mtu.pvlx_to_mtu[i / 2] >>
					   4) & 0xF);
M
Mike Marciniszyn 已提交
1247
		else
1248 1249
			mtu = enum_to_mtu(pi->neigh_mtu.pvlx_to_mtu[i / 2] &
					  0xF);
M
Mike Marciniszyn 已提交
1250 1251 1252 1253 1254 1255 1256 1257 1258
		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,
1259 1260
				    "MTU change on vl %d from %d to %d\n",
				    i, dd->vld[i].mtu, mtu);
M
Mike Marciniszyn 已提交
1261 1262 1263 1264 1265 1266 1267
			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.
	 */
1268
	mtu = enum_to_mtu(pi->neigh_mtu.pvlx_to_mtu[15 / 2] & 0xF);
M
Mike Marciniszyn 已提交
1269 1270 1271 1272
	if (mtu < 2048 || mtu == 0xffff)
		mtu = 2048;
	if (dd->vld[15].mtu != mtu) {
		dd_dev_info(dd,
1273 1274
			    "MTU change on vl 15 from %d to %d\n",
			    dd->vld[15].mtu, mtu);
M
Mike Marciniszyn 已提交
1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289
		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,
1290
					    vls) == -EINVAL)
M
Mike Marciniszyn 已提交
1291 1292 1293 1294 1295
				smp->status |= IB_SMP_INVALID_FIELD;
		}
	}

	if (pi->mkey_violations == 0)
1296
		ibp->rvp.mkey_violations = 0;
M
Mike Marciniszyn 已提交
1297 1298

	if (pi->pkey_violations == 0)
1299
		ibp->rvp.pkey_violations = 0;
M
Mike Marciniszyn 已提交
1300 1301

	if (pi->qkey_violations == 0)
1302
		ibp->rvp.qkey_violations = 0;
M
Mike Marciniszyn 已提交
1303

1304
	ibp->rvp.subnet_timeout =
M
Mike Marciniszyn 已提交
1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 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 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408
		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) {
			if (ppd->is_sm_config_started == 0)
				invalid = 1;
		}
	}

	/* 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.
	 */

	ret = set_port_states(ppd, smp, ls_new, ps_new, invalid);
	if (ret)
		return ret;

	ret = __subn_get_opa_portinfo(smp, am, data, ibdev, port, resp_len);

	/* 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:
	return __subn_get_opa_portinfo(smp, am, data, ibdev, port, resp_len);
}

/**
 * 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;
1409 1410 1411 1412 1413 1414
		/*
		 * Don't update pkeys[2], if an HFI port without MgmtAllowed
		 * by neighbor is a switch.
		 */
		if (i == 2 && !ppd->mgmt_allowed && ppd->neighbor_type == 1)
			continue;
M
Mike Marciniszyn 已提交
1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425
		/*
		 * 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);
1426
		hfi1_event_pkey_change(dd, port);
M
Mike Marciniszyn 已提交
1427
	}
1428

M
Mike Marciniszyn 已提交
1429 1430 1431 1432 1433 1434 1435 1436 1437 1438
	return 0;
}

static int __subn_set_opa_pkeytable(struct opa_smp *smp, u32 am, u8 *data,
				    struct ib_device *ibdev, u8 port,
				    u32 *resp_len)
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	u32 n_blocks_sent = OPA_AM_NBLK(am);
	u32 start_block = am & 0x7ff;
1439
	u16 *p = (u16 *)data;
M
Mike Marciniszyn 已提交
1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451
	__be16 *q = (__be16 *)data;
	int i;
	u16 n_blocks_avail;
	unsigned npkeys = hfi1_get_npkeys(dd);

	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);
	}

1452
	n_blocks_avail = (u16)(npkeys / OPA_PARTITION_TABLE_BLK_SIZE) + 1;
M
Mike Marciniszyn 已提交
1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475

	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);
	}

	return __subn_get_opa_pkeytable(smp, am, data, ibdev, port, resp_len);
}

static int get_sc2vlt_tables(struct hfi1_devdata *dd, void *data)
{
1476
	u64 *val = data;
M
Mike Marciniszyn 已提交
1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494

	*val++ = read_csr(dd, SEND_SC2VLT0);
	*val++ = read_csr(dd, SEND_SC2VLT1);
	*val++ = read_csr(dd, SEND_SC2VLT2);
	*val++ = read_csr(dd, SEND_SC2VLT3);
	return 0;
}

#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.
 */
static void filter_sc2vlt(void *data)
{
	int i;
1495
	u8 *pd = data;
M
Mike Marciniszyn 已提交
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506

	for (i = 0; i < OPA_MAX_SCS; i++) {
		if (i == 15)
			continue;
		if ((pd[i] & 0x1f) == 0xf)
			pd[i] = ILLEGAL_VL;
	}
}

static int set_sc2vlt_tables(struct hfi1_devdata *dd, void *data)
{
1507
	u64 *val = data;
M
Mike Marciniszyn 已提交
1508 1509 1510 1511 1512 1513 1514 1515

	filter_sc2vlt(data);

	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);
1516
	memcpy(dd->sc2vl, data, sizeof(dd->sc2vl));
M
Mike Marciniszyn 已提交
1517 1518 1519 1520 1521 1522 1523 1524 1525
	write_sequnlock_irq(&dd->sc2vl_lock);
	return 0;
}

static int __subn_get_opa_sl_to_sc(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
				   u32 *resp_len)
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
1526
	u8 *p = data;
M
Mike Marciniszyn 已提交
1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548
	size_t size = ARRAY_SIZE(ibp->sl_to_sc); /* == 32 */
	unsigned i;

	if (am) {
		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,
				   u32 *resp_len)
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
1549
	u8 *p = data;
M
Mike Marciniszyn 已提交
1550
	int i;
1551
	u8 sc;
M
Mike Marciniszyn 已提交
1552 1553 1554 1555 1556 1557

	if (am) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

1558 1559 1560 1561 1562 1563 1564 1565 1566
	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
1567 1568 1569 1570 1571 1572 1573 1574 1575

	return __subn_get_opa_sl_to_sc(smp, am, data, ibdev, port, resp_len);
}

static int __subn_get_opa_sc_to_sl(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
				   u32 *resp_len)
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
1576
	u8 *p = data;
M
Mike Marciniszyn 已提交
1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598
	size_t size = ARRAY_SIZE(ibp->sc_to_sl); /* == 32 */
	unsigned i;

	if (am) {
		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,
				   u32 *resp_len)
{
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
1599
	u8 *p = data;
M
Mike Marciniszyn 已提交
1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618
	int i;

	if (am) {
		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++;

	return __subn_get_opa_sc_to_sl(smp, am, data, ibdev, port, resp_len);
}

static int __subn_get_opa_sc_to_vlt(struct opa_smp *smp, u32 am, u8 *data,
				    struct ib_device *ibdev, u8 port,
				    u32 *resp_len)
{
	u32 n_blocks = OPA_AM_NBLK(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
1619
	void *vp = (void *)data;
M
Mike Marciniszyn 已提交
1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641
	size_t size = 4 * sizeof(u64);

	if (n_blocks != 1) {
		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,
				    u32 *resp_len)
{
	u32 n_blocks = OPA_AM_NBLK(am);
	int async_update = OPA_AM_ASYNC(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
1642
	void *vp = (void *)data;
M
Mike Marciniszyn 已提交
1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653
	struct hfi1_pportdata *ppd;
	int lstate;

	if (n_blocks != 1 || async_update) {
		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);
1654 1655 1656 1657
	/*
	 * it's known that async_update is 0 by this point, but include
	 * the explicit check for clarity
	 */
M
Mike Marciniszyn 已提交
1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675
	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);

	return __subn_get_opa_sc_to_vlt(smp, am, data, ibdev, port, resp_len);
}

static int __subn_get_opa_sc_to_vlnt(struct opa_smp *smp, u32 am, u8 *data,
				     struct ib_device *ibdev, u8 port,
				     u32 *resp_len)
{
	u32 n_blocks = OPA_AM_NPORT(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_pportdata *ppd;
1676
	void *vp = (void *)data;
M
Mike Marciniszyn 已提交
1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700
	int size;

	if (n_blocks != 1) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

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

	size = fm_get_table(ppd, FM_TBL_SC2VLNT, vp);

	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,
				     u32 *resp_len)
{
	u32 n_blocks = OPA_AM_NPORT(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_pportdata *ppd;
1701
	void *vp = (void *)data;
M
Mike Marciniszyn 已提交
1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733
	int lstate;

	if (n_blocks != 1) {
		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,
					 resp_len);
}

static int __subn_get_opa_psi(struct opa_smp *smp, u32 am, u8 *data,
			      struct ib_device *ibdev, u8 port,
			      u32 *resp_len)
{
	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;
1734
	struct opa_port_state_info *psi = (struct opa_port_state_info *)data;
M
Mike Marciniszyn 已提交
1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752

	if (nports != 1) {
		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 |=
1753
		ppd->offline_disabled_reason;
M
Mike Marciniszyn 已提交
1754 1755 1756 1757

	psi->port_states.portphysstate_portstate =
		(hfi1_ibphys_portstate(ppd) << 4) | (lstate & 0xf);
	psi->link_width_downgrade_tx_active =
1758
		cpu_to_be16(ppd->link_width_downgrade_tx_active);
M
Mike Marciniszyn 已提交
1759
	psi->link_width_downgrade_rx_active =
1760
		cpu_to_be16(ppd->link_width_downgrade_rx_active);
M
Mike Marciniszyn 已提交
1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
	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,
			      u32 *resp_len)
{
	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;
1777
	struct opa_port_state_info *psi = (struct opa_port_state_info *)data;
M
Mike Marciniszyn 已提交
1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821
	int ret, invalid = 0;

	if (nports != 1) {
		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) {
			if (ppd->is_sm_config_started == 0)
				invalid = 1;
		}
	}

	ret = set_port_states(ppd, smp, ls_new, ps_new, invalid);
	if (ret)
		return ret;

	if (invalid)
		smp->status |= IB_SMP_INVALID_FIELD;

	return __subn_get_opa_psi(smp, am, data, ibdev, port, resp_len);
}

static int __subn_get_opa_cable_info(struct opa_smp *smp, u32 am, u8 *data,
				     struct ib_device *ibdev, u8 port,
				     u32 *resp_len)
{
	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;

1822 1823 1824 1825 1826
	if (dd->pport->port_type != PORT_TYPE_QSFP) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

1827
#define __CI_PAGE_SIZE BIT(7) /* 128 bytes */
M
Mike Marciniszyn 已提交
1828 1829 1830
#define __CI_PAGE_MASK ~(__CI_PAGE_SIZE - 1)
#define __CI_PAGE_NUM(a) ((a) & __CI_PAGE_MASK)

1831 1832 1833 1834
	/*
	 * check that addr is within spec, and
	 * addr and (addr + len - 1) are on the same "page"
	 */
M
Mike Marciniszyn 已提交
1835
	if (addr >= 4096 ||
1836
	    (__CI_PAGE_NUM(addr) != __CI_PAGE_NUM(addr + len - 1))) {
M
Mike Marciniszyn 已提交
1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869
		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,
			      struct ib_device *ibdev, u8 port, u32 *resp_len)
{
	u32 num_ports = OPA_AM_NPORT(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_pportdata *ppd;
1870
	struct buffer_control *p = (struct buffer_control *)data;
M
Mike Marciniszyn 已提交
1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892
	int size;

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

	ppd = dd->pport + (port - 1);
	size = fm_get_table(ppd, FM_TBL_BUFFER_CONTROL, p);
	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,
			      struct ib_device *ibdev, u8 port, u32 *resp_len)
{
	u32 num_ports = OPA_AM_NPORT(am);
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	struct hfi1_pportdata *ppd;
1893
	struct buffer_control *p = (struct buffer_control *)data;
M
Mike Marciniszyn 已提交
1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965

	if (num_ports != 1) {
		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);
	}

	return __subn_get_opa_bct(smp, am, data, ibdev, port, resp_len);
}

static int __subn_get_opa_vl_arb(struct opa_smp *smp, u32 am, u8 *data,
				 struct ib_device *ibdev, u8 port,
				 u32 *resp_len)
{
	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;
	int size = 0;

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

	switch (section) {
	case OPA_VLARB_LOW_ELEMENTS:
		size = fm_get_table(ppd, FM_TBL_VL_LOW_ARB, p);
		break;
	case OPA_VLARB_HIGH_ELEMENTS:
		size = fm_get_table(ppd, FM_TBL_VL_HIGH_ARB, p);
		break;
	case OPA_VLARB_PREEMPT_ELEMENTS:
		size = fm_get_table(ppd, FM_TBL_VL_PREEMPT_ELEMS, p);
		break;
	case OPA_VLARB_PREEMPT_MATRIX:
		size = fm_get_table(ppd, FM_TBL_VL_PREEMPT_MATRIX, p);
		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;
		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,
				 u32 *resp_len)
{
	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;

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

	switch (section) {
	case OPA_VLARB_LOW_ELEMENTS:
1966
		(void)fm_set_table(ppd, FM_TBL_VL_LOW_ARB, p);
M
Mike Marciniszyn 已提交
1967 1968
		break;
	case OPA_VLARB_HIGH_ELEMENTS:
1969
		(void)fm_set_table(ppd, FM_TBL_VL_HIGH_ARB, p);
M
Mike Marciniszyn 已提交
1970
		break;
1971 1972 1973 1974
	/*
	 * neither OPA_VLARB_PREEMPT_ELEMENTS, or OPA_VLARB_PREEMPT_MATRIX
	 * can be changed from the default values
	 */
M
Mike Marciniszyn 已提交
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125
	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;
	}

	return __subn_get_opa_vl_arb(smp, am, data, ibdev, port, resp_len);
}

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

struct opa_class_port_info {
	u8 base_version;
	u8 class_version;
	__be16 cap_mask;
	__be32 cap_mask2_resp_time;

	u8 redirect_gid[16];
	__be32 redirect_tc_fl;
	__be32 redirect_lid;
	__be32 redirect_sl_qp;
	__be32 redirect_qkey;

	u8 trap_gid[16];
	__be32 trap_tc_fl;
	__be32 trap_lid;
	__be32 trap_hl_qp;
	__be32 trap_qkey;

	__be16 trap_pkey;
	__be16 redirect_pkey;

	u8 trap_sl_rsvd;
	u8 reserved[3];
} __packed;

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

#define VL_MASK_ALL		0x000080ff

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];
};

2126 2127 2128 2129 2130 2131 2132 2133
#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
2134 2135 2136
struct opa_port_data_counters_msg {
	__be64 port_select_mask[4];
	__be32 vl_select_mask;
2137
	__be32 resolution;
M
Mike Marciniszyn 已提交
2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185

	/* 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 {
2186 2187 2188 2189
	/*
	 * Request contains first two fields, response contains the
	 * whole magilla
	 */
M
Mike Marciniszyn 已提交
2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300
	__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,
2301
				     struct ib_device *ibdev, u32 *resp_len)
M
Mike Marciniszyn 已提交
2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328
{
	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;
	p->class_version = OPA_SMI_CLASS_VERSION;
	/*
	 * 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,
			  struct opa_port_status_rsp *rsp, u32 vl_select_mask)
{
	if (!is_bx(ppd->dd)) {
		unsigned long vl;
2329
		u64 sum_vl_xmit_wait = 0;
M
Mike Marciniszyn 已提交
2330 2331 2332 2333
		u32 vl_all_mask = VL_MASK_ALL;

		for_each_set_bit(vl, (unsigned long *)&(vl_all_mask),
				 8 * sizeof(vl_all_mask)) {
2334 2335 2336 2337 2338 2339 2340 2341 2342
			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
2343
		}
2344 2345
		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
2346 2347 2348 2349
	}
}

static int pma_get_opa_portstatus(struct opa_pma_mad *pmp,
2350 2351
				  struct ib_device *ibdev,
				  u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375
{
	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;
	u32 vl_select_mask = be32_to_cpu(req->vl_select_mask);
	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;
	u8 num_vls = hweight32(vl_select_mask);
	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;

	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);
	}

2376 2377
	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
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
		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);

	rsp->vl_select_mask = cpu_to_be32(vl_select_mask);
	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,
2407
					  CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425
	rsp->port_multicast_rcv_pkts =
		cpu_to_be64(read_dev_cntr(dd, C_DC_MC_RCV_PKTS,
					  CNTR_INVALID_VL));
	rsp->port_xmit_wait =
		cpu_to_be64(read_port_cntr(ppd, C_TX_WAIT, CNTR_INVALID_VL));
	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));
2426 2427 2428
	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
2429 2430
	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,
2431
				   CNTR_INVALID_VL);
M
Mike Marciniszyn 已提交
2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445
	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,
2446
						      CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
2447 2448 2449 2450 2451

	/* 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;

2452
	vlinfo = &rsp->vls[0];
M
Mike Marciniszyn 已提交
2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467
	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.
	 */
	for_each_set_bit(vl, (unsigned long *)&(vl_select_mask),
			 8 * sizeof(vl_select_mask)) {
		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,
2468
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2469 2470 2471

		rsp->vls[vfi].port_vl_xmit_data =
			cpu_to_be64(read_port_cntr(ppd, C_TX_FLIT_VL,
2472
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2473 2474 2475

		rsp->vls[vfi].port_vl_xmit_pkts =
			cpu_to_be64(read_port_cntr(ppd, C_TX_PKT_VL,
2476
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2477 2478 2479

		rsp->vls[vfi].port_vl_xmit_wait =
			cpu_to_be64(read_port_cntr(ppd, C_TX_WAIT_VL,
2480
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2481 2482 2483

		rsp->vls[vfi].port_vl_rcv_fecn =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_FCN_VL,
2484
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2485 2486 2487

		rsp->vls[vfi].port_vl_rcv_becn =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_BCN_VL,
2488
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2489

2490 2491 2492
		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
2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504
		vlinfo++;
		vfi++;
	}

	a0_portstatus(ppd, rsp, vl_select_mask);

	if (resp_len)
		*resp_len += response_data_size;

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

2505 2506
static u64 get_error_counter_summary(struct ib_device *ibdev, u8 port,
				     u8 res_lli, u8 res_ler)
M
Mike Marciniszyn 已提交
2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520
{
	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,
2521
					       CNTR_INVALID_VL);
2522
	/* local link integrity must be right-shifted by the lli resolution */
2523 2524
	error_counter_summary += (read_dev_cntr(dd, C_DC_RX_REPLAY,
						CNTR_INVALID_VL) >> res_lli);
2525 2526 2527 2528
	/* 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
2529
	error_counter_summary += read_dev_cntr(dd, C_DC_RCV_ERR,
2530
					       CNTR_INVALID_VL);
M
Mike Marciniszyn 已提交
2531 2532
	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,
2533
					       CNTR_INVALID_VL);
M
Mike Marciniszyn 已提交
2534 2535 2536 2537 2538 2539 2540 2541 2542 2543
	/* 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;
}

2544
static void a0_datacounters(struct hfi1_pportdata *ppd, struct _port_dctrs *rsp,
M
Mike Marciniszyn 已提交
2545 2546
			    u32 vl_select_mask)
{
2547
	if (!is_bx(ppd->dd)) {
M
Mike Marciniszyn 已提交
2548
		unsigned long vl;
2549
		u64 sum_vl_xmit_wait = 0;
2550
		u32 vl_all_mask = VL_MASK_ALL;
2551

2552 2553
		for_each_set_bit(vl, (unsigned long *)&(vl_all_mask),
				 8 * sizeof(vl_all_mask)) {
M
Mike Marciniszyn 已提交
2554
			u64 tmp = sum_vl_xmit_wait +
2555 2556
				  read_port_cntr(ppd, C_TX_WAIT_VL,
						 idx_from_vl(vl));
M
Mike Marciniszyn 已提交
2557 2558
			if (tmp < sum_vl_xmit_wait) {
				/* we wrapped */
2559
				sum_vl_xmit_wait = (u64)~0;
M
Mike Marciniszyn 已提交
2560 2561 2562 2563 2564 2565 2566 2567 2568
				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);
	}
}

2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589
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
2590
static int pma_get_opa_datacounters(struct opa_pma_mad *pmp,
2591 2592
				    struct ib_device *ibdev,
				    u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604
{
	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 num_pslm;
	u8 lq, num_vls;
2605
	u8 res_lli, res_ler;
M
Mike Marciniszyn 已提交
2606
	u64 port_mask;
2607
	u8 port_num;
M
Mike Marciniszyn 已提交
2608 2609 2610 2611 2612 2613 2614 2615
	unsigned long vl;
	u32 vl_select_mask;
	int vfi;

	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));
	vl_select_mask = be32_to_cpu(req->vl_select_mask);
2616 2617 2618 2619
	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
2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642

	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,
				  sizeof(port_mask));

2643
	if (port_num != port) {
M
Mike Marciniszyn 已提交
2644 2645 2646 2647
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

2648
	rsp = &req->port[0];
M
Mike Marciniszyn 已提交
2649 2650 2651 2652 2653 2654 2655 2656 2657 2658
	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);
2659
	pma_get_opa_port_dctrs(ibdev, rsp);
M
Mike Marciniszyn 已提交
2660 2661 2662 2663 2664 2665 2666 2667

	rsp->port_xmit_wait =
		cpu_to_be64(read_port_cntr(ppd, C_TX_WAIT, CNTR_INVALID_VL));
	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 =
2668 2669
		cpu_to_be64(get_error_counter_summary(ibdev, port,
						      res_lli, res_ler));
M
Mike Marciniszyn 已提交
2670

2671
	vlinfo = &rsp->vls[0];
M
Mike Marciniszyn 已提交
2672 2673 2674 2675 2676 2677 2678
	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.
	 */
	for_each_set_bit(vl, (unsigned long *)&(vl_select_mask),
2679
			 8 * sizeof(req->vl_select_mask)) {
M
Mike Marciniszyn 已提交
2680 2681 2682 2683
		memset(vlinfo, 0, sizeof(*vlinfo));

		rsp->vls[vfi].port_vl_xmit_data =
			cpu_to_be64(read_port_cntr(ppd, C_TX_FLIT_VL,
2684
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2685 2686 2687

		rsp->vls[vfi].port_vl_rcv_data =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RX_FLIT_VL,
2688
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2689 2690 2691

		rsp->vls[vfi].port_vl_xmit_pkts =
			cpu_to_be64(read_port_cntr(ppd, C_TX_PKT_VL,
2692
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2693 2694 2695

		rsp->vls[vfi].port_vl_rcv_pkts =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RX_PKT_VL,
2696
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2697 2698 2699

		rsp->vls[vfi].port_vl_xmit_wait =
			cpu_to_be64(read_port_cntr(ppd, C_TX_WAIT_VL,
2700
						   idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2701 2702 2703

		rsp->vls[vfi].port_vl_rcv_fecn =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_FCN_VL,
2704
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2705 2706
		rsp->vls[vfi].port_vl_rcv_becn =
			cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_BCN_VL,
2707
						  idx_from_vl(vl)));
M
Mike Marciniszyn 已提交
2708 2709 2710 2711

		/* 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) ???
2712 2713
		 * does this differ from rsp->vls[vfi].port_vl_xmit_wait
		 */
M
Mike Marciniszyn 已提交
2714
		/*rsp->vls[vfi].port_vl_mark_fecn =
2715 2716 2717
		 *	cpu_to_be64(read_csr(dd, DCC_PRF_PORT_VL_MARK_FECN_CNT
		 *		+ offset));
		 */
M
Mike Marciniszyn 已提交
2718 2719 2720 2721
		vlinfo++;
		vfi++;
	}

2722
	a0_datacounters(ppd, rsp, vl_select_mask);
M
Mike Marciniszyn 已提交
2723 2724 2725 2726 2727 2728 2729

	if (resp_len)
		*resp_len += response_data_size;

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

2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792
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));
2793 2794 2795
	rsp->local_link_integrity_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RX_REPLAY,
					  CNTR_INVALID_VL));
2796 2797 2798 2799
	rsp->excessive_buffer_overruns =
		cpu_to_be64(read_dev_cntr(dd, C_RCV_OVF, CNTR_INVALID_VL));
}

M
Mike Marciniszyn 已提交
2800
static int pma_get_opa_porterrors(struct opa_pma_mad *pmp,
2801 2802
				  struct ib_device *ibdev,
				  u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
2803 2804 2805
{
	size_t response_data_size;
	struct _port_ectrs *rsp;
2806
	u8 port_num;
M
Mike Marciniszyn 已提交
2807 2808 2809 2810 2811 2812 2813 2814 2815
	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;
2816
	u64 port_mask, tmp;
M
Mike Marciniszyn 已提交
2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844
	u32 vl_select_mask;
	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,
2845
				  sizeof(port_mask));
M
Mike Marciniszyn 已提交
2846

2847
	if (port_num != port) {
M
Mike Marciniszyn 已提交
2848 2849 2850 2851
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

2852
	rsp = &req->port[0];
M
Mike Marciniszyn 已提交
2853 2854 2855 2856 2857

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

	memset(rsp, 0, sizeof(*rsp));
2858
	rsp->port_number = port_num;
M
Mike Marciniszyn 已提交
2859

2860
	pma_get_opa_port_ectrs(ibdev, rsp, port_num);
M
Mike Marciniszyn 已提交
2861 2862 2863

	rsp->port_rcv_remote_physical_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RMT_PHY_ERR,
2864
					  CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
2865 2866
	rsp->fm_config_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_FM_CFG_ERR,
2867
					  CNTR_INVALID_VL));
M
Mike Marciniszyn 已提交
2868
	tmp = read_dev_cntr(dd, C_DC_UNC_ERR, CNTR_INVALID_VL);
2869

M
Mike Marciniszyn 已提交
2870
	rsp->uncorrectable_errors = tmp < 0x100 ? (tmp & 0xff) : 0xff;
2871 2872
	rsp->port_rcv_errors =
		cpu_to_be64(read_dev_cntr(dd, C_DC_RCV_ERR, CNTR_INVALID_VL));
2873
	vlinfo = &rsp->vls[0];
M
Mike Marciniszyn 已提交
2874 2875 2876 2877 2878
	vfi = 0;
	vl_select_mask = be32_to_cpu(req->vl_select_mask);
	for_each_set_bit(vl, (unsigned long *)&(vl_select_mask),
			 8 * sizeof(req->vl_select_mask)) {
		memset(vlinfo, 0, sizeof(*vlinfo));
2879 2880 2881
		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
2882 2883 2884 2885 2886 2887 2888 2889 2890 2891
		vlinfo += 1;
		vfi++;
	}

	if (resp_len)
		*resp_len += response_data_size;

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

2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976
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
2977
static int pma_get_opa_errorinfo(struct opa_pma_mad *pmp,
2978 2979
				 struct ib_device *ibdev,
				 u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
2980 2981 2982 2983 2984 2985 2986
{
	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;
2987
	u8 port_num;
M
Mike Marciniszyn 已提交
2988 2989 2990 2991
	u8 num_pslm;
	u64 reg;

	req = (struct opa_port_error_info_msg *)pmp->data;
2992
	rsp = &req->port[0];
M
Mike Marciniszyn 已提交
2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019

	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,
				  sizeof(port_mask));

3020
	if (port_num != port) {
M
Mike Marciniszyn 已提交
3021 3022 3023 3024 3025 3026 3027 3028
		pmp->mad_hdr.status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)pmp);
	}

	/* PortRcvErrorInfo */
	rsp->port_rcv_ei.status_and_code =
		dd->err_info_rcvport.status_and_code;
	memcpy(&rsp->port_rcv_ei.ei.ei1to12.packet_flit1,
3029
	       &dd->err_info_rcvport.packet_flit1, sizeof(u64));
M
Mike Marciniszyn 已提交
3030
	memcpy(&rsp->port_rcv_ei.ei.ei1to12.packet_flit2,
3031
	       &dd->err_info_rcvport.packet_flit2, sizeof(u64));
M
Mike Marciniszyn 已提交
3032 3033 3034 3035

	/* ExcessiverBufferOverrunInfo */
	reg = read_csr(dd, RCV_ERR_INFO);
	if (reg & RCV_ERR_INFO_RCV_EXCESS_BUFFER_OVERRUN_SMASK) {
3036 3037 3038 3039
		/*
		 * if the RcvExcessBufferOverrun bit is set, save SC of
		 * first pkt that encountered an excess buffer overrun
		 */
M
Mike Marciniszyn 已提交
3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075
		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,
3076 3077
				  struct ib_device *ibdev,
				  u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135
{
	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);
	u32 vl_select_mask = VL_MASK_ALL; /* clear all per-vl cnts */
	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);

	if (counter_select & CS_PORT_XMIT_WAIT)
		write_port_cntr(ppd, C_TX_WAIT, CNTR_INVALID_VL, 0);

	/* 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 */
3136 3137 3138
	/* if (counter_select & CS_PORT_MARK_FECN)
	 *	write_csr(dd, DCC_PRF_PORT_MARK_FECN_CNT, 0);
	 */
M
Mike Marciniszyn 已提交
3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152

	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);

3153
	if (counter_select & CS_LOCAL_LINK_INTEGRITY_ERRORS)
M
Mike Marciniszyn 已提交
3154 3155 3156 3157 3158
		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,
3159
			       CNTR_INVALID_VL, 0);
M
Mike Marciniszyn 已提交
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 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208
	}

	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);

	for_each_set_bit(vl, (unsigned long *)&(vl_select_mask),
			 8 * sizeof(vl_select_mask)) {
		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);

		if (counter_select & CS_PORT_XMIT_WAIT)
			write_port_cntr(ppd, C_TX_WAIT_VL, idx_from_vl(vl), 0);

		/* 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);

3209 3210 3211
		/* if (counter_select & CS_PORT_MARK_FECN)
		 *     write_csr(dd, DCC_PRF_PORT_VL_MARK_FECN_CNT + offset, 0);
		 */
3212 3213 3214
		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
3215 3216 3217 3218 3219 3220 3221 3222 3223
	}

	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,
3224 3225
				 struct ib_device *ibdev,
				 u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
3226 3227 3228 3229 3230 3231
{
	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;
3232
	u8 port_num;
M
Mike Marciniszyn 已提交
3233 3234 3235 3236
	u8 num_pslm;
	u32 error_info_select;

	req = (struct opa_port_error_info_msg *)pmp->data;
3237
	rsp = &req->port[0];
M
Mike Marciniszyn 已提交
3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256

	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,
				  sizeof(port_mask));

3257
	if (port_num != port) {
M
Mike Marciniszyn 已提交
3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270
		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)
3271 3272 3273 3274
		/*
		 * status bit is essentially kept in the h/w - bit 5 of
		 * RCV_ERR_INFO
		 */
M
Mike Marciniszyn 已提交
3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 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
		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,
				    u32 *resp_len)
{
	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);

	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,
3326 3327
				       u8 *data, struct ib_device *ibdev,
				       u8 port, u32 *resp_len)
M
Mike Marciniszyn 已提交
3328 3329 3330
{
	int i;
	struct opa_congestion_setting_attr *p =
3331
		(struct opa_congestion_setting_attr *)data;
M
Mike Marciniszyn 已提交
3332 3333 3334 3335 3336 3337 3338 3339 3340
	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;

	rcu_read_lock();

	cc_state = get_cc_state(ppd);

3341
	if (!cc_state) {
M
Mike Marciniszyn 已提交
3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364
		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);
}

3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382
/*
 * 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);

3383
	old_cc_state = get_cc_state_protected(ppd);
3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405
	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;

	new_cc_state->cct.ccti_limit = ppd->total_cct_entry - 1;
	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);

3406
	kfree_rcu(old_cc_state, rcu);
3407 3408
}

M
Mike Marciniszyn 已提交
3409 3410 3411 3412 3413
static int __subn_set_opa_cong_setting(struct opa_smp *smp, u32 am, u8 *data,
				       struct ib_device *ibdev, u8 port,
				       u32 *resp_len)
{
	struct opa_congestion_setting_attr *p =
3414
		(struct opa_congestion_setting_attr *)data;
M
Mike Marciniszyn 已提交
3415 3416 3417 3418 3419
	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;

3420 3421 3422 3423 3424
	/*
	 * 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
3425 3426 3427 3428 3429 3430 3431 3432 3433 3434
	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;
	}
3435 3436 3437 3438
	spin_unlock(&ppd->cc_state_lock);

	/* now apply the information */
	apply_cc_state(ppd);
M
Mike Marciniszyn 已提交
3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458

	return __subn_get_opa_cong_setting(smp, am, data, ibdev, port,
					   resp_len);
}

static int __subn_get_opa_hfi1_cong_log(struct opa_smp *smp, u32 am,
					u8 *data, struct ib_device *ibdev,
					u8 port, u32 *resp_len)
{
	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;
	s64 ts;
	int i;

	if (am != 0) {
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

3459
	spin_lock_irq(&ppd->cc_log_lock);
M
Mike Marciniszyn 已提交
3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484

	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 */
	ts = ktime_to_ns(ktime_get()) / 1024;
	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).
		 */
		if ((u64)(ts - cce->timestamp) > (2 * UINT_MAX))
			continue;
		memcpy(cong_log->events[i].local_qp_cn_entry, &cce->lqpn, 3);
		memcpy(cong_log->events[i].remote_qp_number_cn_entry,
3485
		       &cce->rqpn, 3);
M
Mike Marciniszyn 已提交
3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501
		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;

3502
	spin_unlock_irq(&ppd->cc_log_lock);
M
Mike Marciniszyn 已提交
3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514

	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,
				   u32 *resp_len)
{
	struct ib_cc_table_attr *cc_table_attr =
3515
		(struct ib_cc_table_attr *)data;
M
Mike Marciniszyn 已提交
3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535
	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;

	/* sanity check n_blocks, start_block */
	if (n_blocks == 0 ||
	    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);

3536
	if (!cc_state) {
M
Mike Marciniszyn 已提交
3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555
		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)
3556
		*resp_len += sizeof(u16) * (IB_CCT_ENTRIES * n_blocks + 1);
M
Mike Marciniszyn 已提交
3557 3558 3559 3560 3561 3562 3563 3564

	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,
				   u32 *resp_len)
{
3565
	struct ib_cc_table_attr *p = (struct ib_cc_table_attr *)data;
M
Mike Marciniszyn 已提交
3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592
	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;

	/* sanity check n_blocks, start_block */
	if (n_blocks == 0 ||
	    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);
	}

3593 3594 3595 3596
	/*
	 * 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
3597 3598
	spin_lock(&ppd->cc_state_lock);
	ppd->total_cct_entry = ccti_limit + 1;
3599
	entries = ppd->ccti_entries;
M
Mike Marciniszyn 已提交
3600 3601 3602 3603
	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);

3604 3605
	/* now apply the information */
	apply_cc_state(ppd);
M
Mike Marciniszyn 已提交
3606 3607 3608 3609 3610 3611 3612 3613 3614 3615

	return __subn_get_opa_cc_table(smp, am, data, ibdev, port, resp_len);
}

struct opa_led_info {
	__be32 rsvd_led_mask;
	__be32 rsvd;
};

#define OPA_LED_SHIFT	31
3616
#define OPA_LED_MASK	BIT(OPA_LED_SHIFT)
M
Mike Marciniszyn 已提交
3617 3618 3619 3620 3621 3622

static int __subn_get_opa_led_info(struct opa_smp *smp, u32 am, u8 *data,
				   struct ib_device *ibdev, u8 port,
				   u32 *resp_len)
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
3623
	struct hfi1_pportdata *ppd = dd->pport;
3624
	struct opa_led_info *p = (struct opa_led_info *)data;
M
Mike Marciniszyn 已提交
3625
	u32 nport = OPA_AM_NPORT(am);
3626
	u32 is_beaconing_active;
M
Mike Marciniszyn 已提交
3627

3628
	if (nport != 1) {
M
Mike Marciniszyn 已提交
3629 3630 3631 3632
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

3633
	/*
3634 3635 3636
	 * 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
3637
	 */
3638
	smp_rmb();
3639 3640
	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
3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652

	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,
				   u32 *resp_len)
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
3653
	struct opa_led_info *p = (struct opa_led_info *)data;
M
Mike Marciniszyn 已提交
3654 3655 3656
	u32 nport = OPA_AM_NPORT(am);
	int on = !!(be32_to_cpu(p->rsvd_led_mask) & OPA_LED_MASK);

3657
	if (nport != 1) {
M
Mike Marciniszyn 已提交
3658 3659 3660 3661
		smp->status |= IB_SMP_INVALID_FIELD;
		return reply((struct ib_mad_hdr *)smp);
	}

3662
	if (on)
3663
		hfi1_start_led_override(dd->pport, 2000, 1500);
3664
	else
3665
		shutdown_led_override(dd->pport);
M
Mike Marciniszyn 已提交
3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 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

	return __subn_get_opa_led_info(smp, am, data, ibdev, port, resp_len);
}

static int subn_get_opa_sma(__be16 attr_id, struct opa_smp *smp, u32 am,
			    u8 *data, struct ib_device *ibdev, u8 port,
			    u32 *resp_len)
{
	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,
					      resp_len);
		break;
	case IB_SMP_ATTR_NODE_INFO:
		ret = __subn_get_opa_nodeinfo(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case IB_SMP_ATTR_PORT_INFO:
		ret = __subn_get_opa_portinfo(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case IB_SMP_ATTR_PKEY_TABLE:
		ret = __subn_get_opa_pkeytable(smp, am, data, ibdev, port,
					       resp_len);
		break;
	case OPA_ATTRIB_ID_SL_TO_SC_MAP:
		ret = __subn_get_opa_sl_to_sc(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case OPA_ATTRIB_ID_SC_TO_SL_MAP:
		ret = __subn_get_opa_sc_to_sl(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case OPA_ATTRIB_ID_SC_TO_VLT_MAP:
		ret = __subn_get_opa_sc_to_vlt(smp, am, data, ibdev, port,
					       resp_len);
		break;
	case OPA_ATTRIB_ID_SC_TO_VLNT_MAP:
		ret = __subn_get_opa_sc_to_vlnt(smp, am, data, ibdev, port,
3708
						resp_len);
M
Mike Marciniszyn 已提交
3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746
		break;
	case OPA_ATTRIB_ID_PORT_STATE_INFO:
		ret = __subn_get_opa_psi(smp, am, data, ibdev, port,
					 resp_len);
		break;
	case OPA_ATTRIB_ID_BUFFER_CONTROL_TABLE:
		ret = __subn_get_opa_bct(smp, am, data, ibdev, port,
					 resp_len);
		break;
	case OPA_ATTRIB_ID_CABLE_INFO:
		ret = __subn_get_opa_cable_info(smp, am, data, ibdev, port,
						resp_len);
		break;
	case IB_SMP_ATTR_VL_ARB_TABLE:
		ret = __subn_get_opa_vl_arb(smp, am, data, ibdev, port,
					    resp_len);
		break;
	case OPA_ATTRIB_ID_CONGESTION_INFO:
		ret = __subn_get_opa_cong_info(smp, am, data, ibdev, port,
					       resp_len);
		break;
	case OPA_ATTRIB_ID_HFI_CONGESTION_SETTING:
		ret = __subn_get_opa_cong_setting(smp, am, data, ibdev,
						  port, resp_len);
		break;
	case OPA_ATTRIB_ID_HFI_CONGESTION_LOG:
		ret = __subn_get_opa_hfi1_cong_log(smp, am, data, ibdev,
						   port, resp_len);
		break;
	case OPA_ATTRIB_ID_CONGESTION_CONTROL_TABLE:
		ret = __subn_get_opa_cc_table(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case IB_SMP_ATTR_LED_INFO:
		ret = __subn_get_opa_led_info(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case IB_SMP_ATTR_SM_INFO:
3747
		if (ibp->rvp.port_cap_flags & IB_PORT_SM_DISABLED)
M
Mike Marciniszyn 已提交
3748
			return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED;
3749
		if (ibp->rvp.port_cap_flags & IB_PORT_SM)
M
Mike Marciniszyn 已提交
3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789
			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,
			    u32 *resp_len)
{
	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,
					      resp_len);
		break;
	case IB_SMP_ATTR_PKEY_TABLE:
		ret = __subn_set_opa_pkeytable(smp, am, data, ibdev, port,
					       resp_len);
		break;
	case OPA_ATTRIB_ID_SL_TO_SC_MAP:
		ret = __subn_set_opa_sl_to_sc(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case OPA_ATTRIB_ID_SC_TO_SL_MAP:
		ret = __subn_set_opa_sc_to_sl(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case OPA_ATTRIB_ID_SC_TO_VLT_MAP:
		ret = __subn_set_opa_sc_to_vlt(smp, am, data, ibdev, port,
					       resp_len);
		break;
	case OPA_ATTRIB_ID_SC_TO_VLNT_MAP:
		ret = __subn_set_opa_sc_to_vlnt(smp, am, data, ibdev, port,
3790
						resp_len);
M
Mike Marciniszyn 已提交
3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816
		break;
	case OPA_ATTRIB_ID_PORT_STATE_INFO:
		ret = __subn_set_opa_psi(smp, am, data, ibdev, port,
					 resp_len);
		break;
	case OPA_ATTRIB_ID_BUFFER_CONTROL_TABLE:
		ret = __subn_set_opa_bct(smp, am, data, ibdev, port,
					 resp_len);
		break;
	case IB_SMP_ATTR_VL_ARB_TABLE:
		ret = __subn_set_opa_vl_arb(smp, am, data, ibdev, port,
					    resp_len);
		break;
	case OPA_ATTRIB_ID_HFI_CONGESTION_SETTING:
		ret = __subn_set_opa_cong_setting(smp, am, data, ibdev,
						  port, resp_len);
		break;
	case OPA_ATTRIB_ID_CONGESTION_CONTROL_TABLE:
		ret = __subn_set_opa_cc_table(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case IB_SMP_ATTR_LED_INFO:
		ret = __subn_set_opa_led_info(smp, am, data, ibdev, port,
					      resp_len);
		break;
	case IB_SMP_ATTR_SM_INFO:
3817
		if (ibp->rvp.port_cap_flags & IB_PORT_SM_DISABLED)
M
Mike Marciniszyn 已提交
3818
			return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED;
3819
		if (ibp->rvp.port_cap_flags & IB_PORT_SM)
M
Mike Marciniszyn 已提交
3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868
			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);

3869
		(void)subn_get_opa_sma(agg->attr_id, smp, am, agg->data,
M
Mike Marciniszyn 已提交
3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911
					ibdev, port, NULL);
		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,
				  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);
		}

3912
		(void)subn_set_opa_sma(agg->attr_id, smp, am, agg->data,
M
Mike Marciniszyn 已提交
3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034
					ibdev, port, NULL);
		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;
}

/*
 * 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 slid = in_wc->slid;
	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;
	ingress_pkey_table_fail(ppd, pkey, slid);
	return 1;
}

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,
			    u32 *resp_len)
{
	struct opa_smp *smp = (struct opa_smp *)out_mad;
	struct hfi1_ibport *ibp = to_iport(ibdev, port);
	u8 *data;
	u32 am;
	__be16 attr_id;
	int ret;

	*out_mad = *in_mad;
	data = opa_get_smp_data(smp);

	am = be32_to_cpu(smp->attr_mod);
	attr_id = smp->attr_id;
	if (smp->class_version != OPA_SMI_CLASS_VERSION) {
		smp->status |= IB_SMP_UNSUP_VERSION;
		ret = reply((struct ib_mad_hdr *)smp);
4035
		return ret;
M
Mike Marciniszyn 已提交
4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054
	}
	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)
4055
			(void)check_mkey(to_iport(ibdev, port_num),
M
Mike Marciniszyn 已提交
4056 4057 4058 4059 4060
					  (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;
4061
		return ret;
M
Mike Marciniszyn 已提交
4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072
	}

	*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,
					       ibdev, port, resp_len);
4073
			break;
M
Mike Marciniszyn 已提交
4074 4075 4076
		case OPA_ATTRIB_ID_AGGREGATE:
			ret = subn_get_opa_aggregate(smp, ibdev, port,
						     resp_len);
4077
			break;
M
Mike Marciniszyn 已提交
4078
		}
4079
		break;
M
Mike Marciniszyn 已提交
4080 4081 4082 4083 4084
	case IB_MGMT_METHOD_SET:
		switch (attr_id) {
		default:
			ret = subn_set_opa_sma(attr_id, smp, am, data,
					       ibdev, port, resp_len);
4085
			break;
M
Mike Marciniszyn 已提交
4086 4087 4088
		case OPA_ATTRIB_ID_AGGREGATE:
			ret = subn_set_opa_aggregate(smp, ibdev, port,
						     resp_len);
4089
			break;
M
Mike Marciniszyn 已提交
4090
		}
4091
		break;
M
Mike Marciniszyn 已提交
4092 4093 4094 4095 4096 4097 4098 4099 4100 4101
	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;
4102
		break;
M
Mike Marciniszyn 已提交
4103 4104 4105
	default:
		smp->status |= IB_SMP_UNSUP_METHOD;
		ret = reply((struct ib_mad_hdr *)smp);
4106
		break;
M
Mike Marciniszyn 已提交
4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123
	}

	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);
4124
		return ret;
M
Mike Marciniszyn 已提交
4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144
	}

	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)
4145
			(void)check_mkey(to_iport(ibdev, port_num),
4146 4147 4148 4149
					 (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
4150
		ret = IB_MAD_RESULT_FAILURE;
4151
		return ret;
M
Mike Marciniszyn 已提交
4152 4153 4154 4155 4156 4157 4158
	}

	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);
4159
			break;
M
Mike Marciniszyn 已提交
4160 4161 4162
		default:
			smp->status |= IB_SMP_UNSUP_METH_ATTR;
			ret = reply((struct ib_mad_hdr *)smp);
4163
			break;
M
Mike Marciniszyn 已提交
4164
		}
4165
		break;
M
Mike Marciniszyn 已提交
4166 4167 4168 4169 4170
	}

	return ret;
}

4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227
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
4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253
	}

	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;

	if (pmp->mad_hdr.class_version != OPA_SMI_CLASS_VERSION) {
		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);
4254
			break;
M
Mike Marciniszyn 已提交
4255 4256
		case OPA_PM_ATTRIB_ID_PORT_STATUS:
			ret = pma_get_opa_portstatus(pmp, ibdev, port,
4257
						     resp_len);
4258
			break;
M
Mike Marciniszyn 已提交
4259 4260
		case OPA_PM_ATTRIB_ID_DATA_PORT_COUNTERS:
			ret = pma_get_opa_datacounters(pmp, ibdev, port,
4261
						       resp_len);
4262
			break;
M
Mike Marciniszyn 已提交
4263 4264
		case OPA_PM_ATTRIB_ID_ERROR_PORT_COUNTERS:
			ret = pma_get_opa_porterrors(pmp, ibdev, port,
4265
						     resp_len);
4266
			break;
M
Mike Marciniszyn 已提交
4267 4268
		case OPA_PM_ATTRIB_ID_ERROR_INFO:
			ret = pma_get_opa_errorinfo(pmp, ibdev, port,
4269
						    resp_len);
4270
			break;
M
Mike Marciniszyn 已提交
4271 4272 4273
		default:
			pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR;
			ret = reply((struct ib_mad_hdr *)pmp);
4274
			break;
M
Mike Marciniszyn 已提交
4275
		}
4276
		break;
M
Mike Marciniszyn 已提交
4277 4278 4279 4280 4281

	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,
4282
						     resp_len);
4283
			break;
M
Mike Marciniszyn 已提交
4284 4285
		case OPA_PM_ATTRIB_ID_ERROR_INFO:
			ret = pma_set_opa_errorinfo(pmp, ibdev, port,
4286
						    resp_len);
4287
			break;
M
Mike Marciniszyn 已提交
4288 4289 4290
		default:
			pmp->mad_hdr.status |= IB_SMP_UNSUP_METH_ATTR;
			ret = reply((struct ib_mad_hdr *)pmp);
4291
			break;
M
Mike Marciniszyn 已提交
4292
		}
4293
		break;
M
Mike Marciniszyn 已提交
4294 4295 4296 4297 4298 4299 4300 4301 4302

	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;
4303
		break;
M
Mike Marciniszyn 已提交
4304 4305 4306 4307

	default:
		pmp->mad_hdr.status |= IB_SMP_UNSUP_METHOD;
		ret = reply((struct ib_mad_hdr *)pmp);
4308
		break;
M
Mike Marciniszyn 已提交
4309 4310 4311 4312 4313 4314
	}

	return ret;
}

static int hfi1_process_opa_mad(struct ib_device *ibdev, int mad_flags,
4315 4316 4317 4318 4319
				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
4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 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
{
	int ret;
	int pkey_idx;
	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:
		if (is_local_mad(ibp, in_mad, in_wc)) {
			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,
				       out_mad, &resp_len);
		goto bail;
	case IB_MGMT_CLASS_PERF_MGMT:
		ret = process_perf_opa(ibdev, port, in_mad, out_mad,
				       &resp_len);
		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);
4375
		break;
4376 4377 4378
	case IB_MGMT_CLASS_PERF_MGMT:
		ret = process_perf(ibdev, port, in_mad, out_mad);
		break;
M
Mike Marciniszyn 已提交
4379 4380
	default:
		ret = IB_MAD_RESULT_SUCCESS;
4381
		break;
M
Mike Marciniszyn 已提交
4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 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
	}

	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))) {
			dev_err(ibdev->dma_device, "invalid in_mad_size\n");
			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;
}