aboutsummaryrefslogtreecommitdiffstats
path: root/drivers/infiniband/hw/vmw_pvrdma/pvrdma_verbs.c
blob: 08ea4b9e4c6c577f211eedf089feb559cca2e837 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
/*
 * Copyright (c) 2012-2016 VMware, Inc.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of EITHER the GNU General Public License
 * version 2 as published by the Free Software Foundation or the BSD
 * 2-Clause License. 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 version 2 for more details at
 * http://www.gnu.org/licenses/old-licenses/gpl-2.0.en.html.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program available in the file COPYING in the main
 * directory of this source tree.
 *
 * The BSD 2-Clause 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.
 *
 * 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 HOLDER 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 <asm/page.h>
#include <linux/inet.h>
#include <linux/io.h>
#include <rdma/ib_addr.h>
#include <rdma/ib_smi.h>
#include <rdma/ib_user_verbs.h>
#include <rdma/vmw_pvrdma-abi.h>

#include "pvrdma.h"

/**
 * pvrdma_query_device - query device
 * @ibdev: the device to query
 * @props: the device properties
 * @uhw: user data
 *
 * @return: 0 on success, otherwise negative errno
 */
int pvrdma_query_device(struct ib_device *ibdev,
			struct ib_device_attr *props,
			struct ib_udata *uhw)
{
	struct pvrdma_dev *dev = to_vdev(ibdev);

	if (uhw->inlen || uhw->outlen)
		return -EINVAL;

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

	props->fw_ver = dev->dsr->caps.fw_ver;
	props->sys_image_guid = dev->dsr->caps.sys_image_guid;
	props->max_mr_size = dev->dsr->caps.max_mr_size;
	props->page_size_cap = dev->dsr->caps.page_size_cap;
	props->vendor_id = dev->dsr->caps.vendor_id;
	props->vendor_part_id = dev->pdev->device;
	props->hw_ver = dev->dsr->caps.hw_ver;
	props->max_qp = dev->dsr->caps.max_qp;
	props->max_qp_wr = dev->dsr->caps.max_qp_wr;
	props->device_cap_flags = dev->dsr->caps.device_cap_flags;
	props->max_send_sge = dev->dsr->caps.max_sge;
	props->max_recv_sge = dev->dsr->caps.max_sge;
	props->max_sge_rd = PVRDMA_GET_CAP(dev, dev->dsr->caps.max_sge,
					   dev->dsr->caps.max_sge_rd);
	props->max_srq = dev->dsr->caps.max_srq;
	props->max_srq_wr = dev->dsr->caps.max_srq_wr;
	props->max_srq_sge = dev->dsr->caps.max_srq_sge;
	props->max_cq = dev->dsr->caps.max_cq;
	props->max_cqe = dev->dsr->caps.max_cqe;
	props->max_mr = dev->dsr->caps.max_mr;
	props->max_pd = dev->dsr->caps.max_pd;
	props->max_qp_rd_atom = dev->dsr->caps.max_qp_rd_atom;
	props->max_qp_init_rd_atom = dev->dsr->caps.max_qp_init_rd_atom;
	props->atomic_cap =
		dev->dsr->caps.atomic_ops &
		(PVRDMA_ATOMIC_OP_COMP_SWAP | PVRDMA_ATOMIC_OP_FETCH_ADD) ?
		IB_ATOMIC_HCA : IB_ATOMIC_NONE;
	props->masked_atomic_cap = props->atomic_cap;
	props->max_ah = dev->dsr->caps.max_ah;
	props->max_pkeys = dev->dsr->caps.max_pkeys;
	props->local_ca_ack_delay = dev->dsr->caps.local_ca_ack_delay;
	if ((dev->dsr->caps.bmme_flags & PVRDMA_BMME_FLAG_LOCAL_INV) &&
	    (dev->dsr->caps.bmme_flags & PVRDMA_BMME_FLAG_REMOTE_INV) &&
	    (dev->dsr->caps.bmme_flags & PVRDMA_BMME_FLAG_FAST_REG_WR)) {
		props->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS;
		props->max_fast_reg_page_list_len = PVRDMA_GET_CAP(dev,
				PVRDMA_MAX_FAST_REG_PAGES,
				dev->dsr->caps.max_fast_reg_page_list_len);
	}

	props->device_cap_flags |= IB_DEVICE_PORT_ACTIVE_EVENT |
				   IB_DEVICE_RC_RNR_NAK_GEN;

	return 0;
}

/**
 * pvrdma_query_port - query device port attributes
 * @ibdev: the device to query
 * @port: the port number
 * @props: the device properties
 *
 * @return: 0 on success, otherwise negative errno
 */
int pvrdma_query_port(struct ib_device *ibdev, u8 port,
		      struct ib_port_attr *props)
{
	struct pvrdma_dev *dev = to_vdev(ibdev);
	union pvrdma_cmd_req req;
	union pvrdma_cmd_resp rsp;
	struct pvrdma_cmd_query_port *cmd = &req.query_port;
	struct pvrdma_cmd_query_port_resp *resp = &rsp.query_port_resp;
	int err;

	memset(cmd, 0, sizeof(*cmd));
	cmd->hdr.cmd = PVRDMA_CMD_QUERY_PORT;
	cmd->port_num = port;

	err = pvrdma_cmd_post(dev, &req, &rsp, PVRDMA_CMD_QUERY_PORT_RESP);
	if (err < 0) {
		dev_warn(&dev->pdev->dev,
			 "could not query port, error: %d\n", err);
		return err;
	}

	/* props being zeroed by the caller, avoid zeroing it here */

	props->state = pvrdma_port_state_to_ib(resp->attrs.state);
	props->max_mtu = pvrdma_mtu_to_ib(resp->attrs.max_mtu);
	props->active_mtu = pvrdma_mtu_to_ib(resp->attrs.active_mtu);
	props->gid_tbl_len = resp->attrs.gid_tbl_len;
	props->port_cap_flags =
		pvrdma_port_cap_flags_to_ib(resp->attrs.port_cap_flags);
	props->port_cap_flags |= IB_PORT_CM_SUP;
	props->ip_gids = true;
	props->max_msg_sz = resp->attrs.max_msg_sz;
	props->bad_pkey_cntr = resp->attrs.bad_pkey_cntr;
	props->qkey_viol_cntr = resp->attrs.qkey_viol_cntr;
	props->pkey_tbl_len = resp->attrs.pkey_tbl_len;
	props->lid = resp->attrs.lid;
	props->sm_lid = resp->attrs.sm_lid;
	props->lmc = resp->attrs.lmc;
	props->max_vl_num = resp->attrs.max_vl_num;
	props->sm_sl = resp->attrs.sm_sl;
	props->subnet_timeout = resp->attrs.subnet_timeout;
	props->init_type_reply = resp->attrs.init_type_reply;
	props->active_width = pvrdma_port_width_to_ib(resp->attrs.active_width);
	props->active_speed = pvrdma_port_speed_to_ib(resp->attrs.active_speed);
	props->phys_state = resp->attrs.phys_state;

	return 0;
}

/**
 * pvrdma_query_gid - query device gid
 * @ibdev: the device to query
 * @port: the port number
 * @index: the index
 * @gid: the device gid value
 *
 * @return: 0 on success, otherwise negative errno
 */
int pvrdma_query_gid(struct ib_device *ibdev, u8 port, int index,
		     union ib_gid *gid)
{
	struct pvrdma_dev *dev = to_vdev(ibdev);

	if (index >= dev->dsr->caps.gid_tbl_len)
		return -EINVAL;

	memcpy(gid, &dev->sgid_tbl[index], sizeof(union ib_gid));

	return 0;
}

/**
 * pvrdma_query_pkey - query device port's P_Key table
 * @ibdev: the device to query
 * @port: the port number
 * @index: the index
 * @pkey: the device P_Key value
 *
 * @return: 0 on success, otherwise negative errno
 */
int pvrdma_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
		      u16 *pkey)
{
	int err = 0;
	union pvrdma_cmd_req req;
	union pvrdma_cmd_resp rsp;
	struct pvrdma_cmd_query_pkey *cmd = &req.query_pkey;

	memset(cmd, 0, sizeof(*cmd));
	cmd->hdr.cmd = PVRDMA_CMD_QUERY_PKEY;
	cmd->port_num = port;
	cmd->index = index;

	err = pvrdma_cmd_post(to_vdev(ibdev), &req, &rsp,
			      PVRDMA_CMD_QUERY_PKEY_RESP);
	if (err < 0) {
		dev_warn(&to_vdev(ibdev)->pdev->dev,
			 "could not query pkey, error: %d\n", err);
		return err;
	}

	*pkey = rsp.query_pkey_resp.pkey;

	return 0;
}

enum rdma_link_layer pvrdma_port_link_layer(struct ib_device *ibdev,
					    u8 port)
{
	return IB_LINK_LAYER_ETHERNET;
}

int pvrdma_modify_device(struct ib_device *ibdev, int mask,
			 struct ib_device_modify *props)
{
	unsigned long flags;

	if (mask & ~(IB_DEVICE_MODIFY_SYS_IMAGE_GUID |
		     IB_DEVICE_MODIFY_NODE_DESC)) {
		dev_warn(&to_vdev(ibdev)->pdev->dev,
			 "unsupported device modify mask %#x\n", mask);
		return -EOPNOTSUPP;
	}

	if (mask & IB_DEVICE_MODIFY_NODE_DESC) {
		spin_lock_irqsave(&to_vdev(ibdev)->desc_lock, flags);
		memcpy(ibdev->node_desc, props->node_desc, 64);
		spin_unlock_irqrestore(&to_vdev(ibdev)->desc_lock, flags);
	}

	if (mask & IB_DEVICE_MODIFY_SYS_IMAGE_GUID) {
		mutex_lock(&to_vdev(ibdev)->port_mutex);
		to_vdev(ibdev)->sys_image_guid =
			cpu_to_be64(props->sys_image_guid);
		mutex_unlock(&to_vdev(ibdev)->port_mutex);
	}

	return 0;
}

/**
 * pvrdma_modify_port - modify device port attributes
 * @ibdev: the device to modify
 * @port: the port number
 * @mask: attributes to modify
 * @props: the device properties
 *
 * @return: 0 on success, otherwise negative errno
 */
int pvrdma_modify_port(struct ib_device *ibdev, u8 port, int mask,
		       struct ib_port_modify *props)
{
	struct ib_port_attr attr;
	struct pvrdma_dev *vdev = to_vdev(ibdev);
	int ret;

	if (mask & ~IB_PORT_SHUTDOWN) {
		dev_warn(&vdev->pdev->dev,
			 "unsupported port modify mask %#x\n", mask);
		return -EOPNOTSUPP;
	}

	mutex_lock(&vdev->port_mutex);
	ret = ib_query_port(ibdev, port, &attr);
	if (ret)
		goto out;

	vdev->port_cap_mask |= props->set_port_cap_mask;
	vdev->port_cap_mask &= ~props->clr_port_cap_mask;

	if (mask & IB_PORT_SHUTDOWN)
		vdev->ib_active = false;

out:
	mutex_unlock(&vdev->port_mutex);
	return ret;
}

/**
 * pvrdma_alloc_ucontext - allocate ucontext
 * @ibdev: the IB device
 * @udata: user data
 *
 * @return: the ib_ucontext pointer on success, otherwise errno.
 */
struct ib_ucontext *pvrdma_alloc_ucontext(struct ib_device *ibdev,
					  struct ib_udata *udata)
{
	struct pvrdma_dev *vdev = to_vdev(ibdev);
	struct pvrdma_ucontext *context;
	union pvrdma_cmd_req req;
	union pvrdma_cmd_resp rsp;
	struct pvrdma_cmd_create_uc *cmd = &req.create_uc;
	struct pvrdma_cmd_create_uc_resp *resp = &rsp.create_uc_resp;
	struct pvrdma_alloc_ucontext_resp uresp = {0};
	int ret;
	void *ptr;

	if (!vdev->ib_active)
		return ERR_PTR(-EAGAIN);

	context = kmalloc(sizeof(*context), GFP_KERNEL);
	if (!context)
		return ERR_PTR(-ENOMEM);

	context->dev = vdev;
	ret = pvrdma_uar_alloc(vdev, &context->uar);
	if (ret) {
		kfree(context);
		return ERR_PTR(-ENOMEM);
	}

	/* get ctx_handle from host */
	memset(cmd, 0, sizeof(*cmd));
	cmd->pfn = context->uar.pfn;
	cmd->hdr.cmd = PVRDMA_CMD_CREATE_UC;
	ret = pvrdma_cmd_post(vdev, &req, &rsp, PVRDMA_CMD_CREATE_UC_RESP);
	if (ret < 0) {
		dev_warn(&vdev->pdev->dev,
			 "could not create ucontext, error: %d\n", ret);
		ptr = ERR_PTR(ret);
		goto err;
	}

	context->ctx_handle = resp->ctx_handle;

	/* copy back to user */
	uresp.qp_tab_size = vdev->dsr->caps.max_qp;
	ret = ib_copy_to_udata(udata, &uresp, sizeof(uresp));
	if (ret) {
		pvrdma_uar_free(vdev, &context->uar);
		context->ibucontext.device = ibdev;
		pvrdma_dealloc_ucontext(&context->ibucontext);
		return ERR_PTR(-EFAULT);
	}

	return &context->ibucontext;

err:
	pvrdma_uar_free(vdev, &context->uar);
	kfree(context);
	return ptr;
}

/**
 * pvrdma_dealloc_ucontext - deallocate ucontext
 * @ibcontext: the ucontext
 *
 * @return: 0 on success, otherwise errno.
 */
int pvrdma_dealloc_ucontext(struct ib_ucontext *ibcontext)
{
	struct pvrdma_ucontext *context = to_vucontext(ibcontext);
	union pvrdma_cmd_req req;
	struct pvrdma_cmd_destroy_uc *cmd = &req.destroy_uc;
	int ret;

	memset(cmd, 0, sizeof(*cmd));
	cmd->hdr.cmd = PVRDMA_CMD_DESTROY_UC;
	cmd->ctx_handle = context->ctx_handle;

	ret = pvrdma_cmd_post(context->dev, &req, NULL, 0);
	if (ret < 0)
		dev_warn(&context->dev->pdev->dev,
			 "destroy ucontext failed, error: %d\n", ret);

	/* Free the UAR even if the device command failed */
	pvrdma_uar_free(to_vdev(ibcontext->device), &context->uar);
	kfree(context);

	return ret;
}

/**
 * pvrdma_mmap - create mmap region
 * @ibcontext: the user context
 * @vma: the VMA
 *
 * @return: 0 on success, otherwise errno.
 */
int pvrdma_mmap(struct ib_ucontext *ibcontext, struct vm_area_struct *vma)
{
	struct pvrdma_ucontext *context = to_vucontext(ibcontext);
	unsigned long start = vma->vm_start;
	unsigned long size = vma->vm_end - vma->vm_start;
	unsigned long offset = vma->vm_pgoff << PAGE_SHIFT;

	dev_dbg(&context->dev->pdev->dev, "create mmap region\n");

	if ((size != PAGE_SIZE) || (offset & ~PAGE_MASK)) {
		dev_warn(&context->dev->pdev->dev,
			 "invalid params for mmap region\n");
		return -EINVAL;
	}

	/* Map UAR to kernel space, VM_LOCKED? */
	vma->vm_flags |= VM_DONTCOPY | VM_DONTEXPAND;
	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
	if (io_remap_pfn_range(vma, start, context->uar.pfn, size,
			       vma->vm_page_prot))
		return -EAGAIN;

	return 0;
}

/**
 * pvrdma_alloc_pd - allocate protection domain
 * @ibdev: the IB device
 * @context: user context
 * @udata: user data
 *
 * @return: the ib_pd protection domain pointer on success, otherwise errno.
 */
struct ib_pd *pvrdma_alloc_pd(struct ib_device *ibdev,
			      struct ib_ucontext *context,
			      struct ib_udata *udata)
{
	struct pvrdma_pd *pd;
	struct pvrdma_dev *dev = to_vdev(ibdev);
	union pvrdma_cmd_req req;
	union pvrdma_cmd_resp rsp;
	struct pvrdma_cmd_create_pd *cmd = &req.create_pd;
	struct pvrdma_cmd_create_pd_resp *resp = &rsp.create_pd_resp;
	struct pvrdma_alloc_pd_resp pd_resp = {0};
	int ret;
	void *ptr;

	/* Check allowed max pds */
	if (!atomic_add_unless(&dev->num_pds, 1, dev->dsr->caps.max_pd))
		return ERR_PTR(-ENOMEM);

	pd = kmalloc(sizeof(*pd), GFP_KERNEL);
	if (!pd) {
		ptr = ERR_PTR(-ENOMEM);
		goto err;
	}

	memset(cmd, 0, sizeof(*cmd));
	cmd->hdr.cmd = PVRDMA_CMD_CREATE_PD;
	cmd->ctx_handle = (context) ? to_vucontext(context)->ctx_handle : 0;
	ret = pvrdma_cmd_post(dev, &req, &rsp, PVRDMA_CMD_CREATE_PD_RESP);
	if (ret < 0) {
		dev_warn(&dev->pdev->dev,
			 "failed to allocate protection domain, error: %d\n",
			 ret);
		ptr = ERR_PTR(ret);
		goto freepd;
	}

	pd->privileged = !context;
	pd->pd_handle = resp->pd_handle;
	pd->pdn = resp->pd_handle;
	pd_resp.pdn = resp->pd_handle;

	if (context) {
		if (ib_copy_to_udata(udata, &pd_resp, sizeof(pd_resp))) {
			dev_warn(&dev->pdev->dev,
				 "failed to copy back protection domain\n");
			pvrdma_dealloc_pd(&pd->ibpd);
			return ERR_PTR(-EFAULT);
		}
	}

	/* u32 pd handle */
	return &pd->ibpd;

freepd:
	kfree(pd);
err:
	atomic_dec(&dev->num_pds);
	return ptr;
}

/**
 * pvrdma_dealloc_pd - deallocate protection domain
 * @pd: the protection domain to be released
 *
 * @return: 0 on success, otherwise errno.
 */
int pvrdma_dealloc_pd(struct ib_pd *pd)
{
	struct pvrdma_dev *dev = to_vdev(pd->device);
	union pvrdma_cmd_req req;
	struct pvrdma_cmd_destroy_pd *cmd = &req.destroy_pd;
	int ret;

	memset(cmd, 0, sizeof(*cmd));
	cmd->hdr.cmd = PVRDMA_CMD_DESTROY_PD;
	cmd->pd_handle = to_vpd(pd)->pd_handle;

	ret = pvrdma_cmd_post(dev, &req, NULL, 0);
	if (ret)
		dev_warn(&dev->pdev->dev,
			 "could not dealloc protection domain, error: %d\n",
			 ret);

	kfree(to_vpd(pd));
	atomic_dec(&dev->num_pds);

	return 0;
}

/**
 * pvrdma_create_ah - create an address handle
 * @pd: the protection domain
 * @ah_attr: the attributes of the AH
 * @udata: user data blob
 * @flags: create address handle flags (see enum rdma_create_ah_flags)
 *
 * @return: the ib_ah pointer on success, otherwise errno.
 */
struct ib_ah *pvrdma_create_ah(struct ib_pd *pd, struct rdma_ah_attr *ah_attr,
			       u32 flags, struct ib_udata *udata)
{
	struct pvrdma_dev *dev = to_vdev(pd->device);
	struct pvrdma_ah *ah;
	const struct ib_global_route *grh;
	u8 port_num = rdma_ah_get_port_num(ah_attr);

	if (!(rdma_ah_get_ah_flags(ah_attr) & IB_AH_GRH))
		return ERR_PTR(-EINVAL);

	grh = rdma_ah_read_grh(ah_attr);
	if ((ah_attr->type != RDMA_AH_ATTR_TYPE_ROCE)  ||
	    rdma_is_multicast_addr((struct in6_addr *)grh->dgid.raw))
		return ERR_PTR(-EINVAL);

	if (!atomic_add_unless(&dev->num_ahs, 1, dev->dsr->caps.max_ah))
		return ERR_PTR(-ENOMEM);

	ah = kzalloc(sizeof(*ah), GFP_ATOMIC);
	if (!ah) {
		atomic_dec(&dev->num_ahs);
		return ERR_PTR(-ENOMEM);
	}

	ah->av.port_pd = to_vpd(pd)->pd_handle | (port_num << 24);
	ah->av.src_path_bits = rdma_ah_get_path_bits(ah_attr);
	ah->av.src_path_bits |= 0x80;
	ah->av.gid_index = grh->sgid_index;
	ah->av.hop_limit = grh->hop_limit;
	ah->av.sl_tclass_flowlabel = (grh->traffic_class << 20) |
				      grh->flow_label;
	memcpy(ah->av.dgid, grh->dgid.raw, 16);
	memcpy(ah->av.dmac, ah_attr->roce.dmac, ETH_ALEN);

	ah->ibah.device = pd->device;
	ah->ibah.pd = pd;
	ah->ibah.uobject = NULL;

	return &ah->ibah;
}

/**
 * pvrdma_destroy_ah - destroy an address handle
 * @ah: the address handle to destroyed
 *
 * @return: 0 on success.
 */
int pvrdma_destroy_ah(struct ib_ah *ah)
{
	struct pvrdma_dev *dev = to_vdev(ah->device);

	kfree(to_vah(ah));
	atomic_dec(&dev->num_ahs);

	return 0;
}