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#include <linux/module.h>
35#include <linux/init.h>
36#include <linux/slab.h>
37#include <linux/errno.h>
38#include <linux/netdevice.h>
39#include <linux/inetdevice.h>
40#include <linux/rtnetlink.h>
41#include <linux/if_vlan.h>
42
43#include <rdma/ib_smi.h>
44#include <rdma/ib_user_verbs.h>
45#include <rdma/ib_addr.h>
46
47#include <linux/mlx4/driver.h>
48#include <linux/mlx4/cmd.h>
49
50#include "mlx4_ib.h"
51#include "user.h"
52
53#define DRV_NAME "mlx4_ib"
54#define DRV_VERSION "1.0"
55#define DRV_RELDATE "April 4, 2008"
56
57MODULE_AUTHOR("Roland Dreier");
58MODULE_DESCRIPTION("Mellanox ConnectX HCA InfiniBand driver");
59MODULE_LICENSE("Dual BSD/GPL");
60MODULE_VERSION(DRV_VERSION);
61
62static const char mlx4_ib_version[] =
63 DRV_NAME ": Mellanox ConnectX InfiniBand driver v"
64 DRV_VERSION " (" DRV_RELDATE ")\n";
65
66struct update_gid_work {
67 struct work_struct work;
68 union ib_gid gids[128];
69 struct mlx4_ib_dev *dev;
70 int port;
71};
72
73static struct workqueue_struct *wq;
74
75static void init_query_mad(struct ib_smp *mad)
76{
77 mad->base_version = 1;
78 mad->mgmt_class = IB_MGMT_CLASS_SUBN_LID_ROUTED;
79 mad->class_version = 1;
80 mad->method = IB_MGMT_METHOD_GET;
81}
82
83static union ib_gid zgid;
84
85static int mlx4_ib_query_device(struct ib_device *ibdev,
86 struct ib_device_attr *props)
87{
88 struct mlx4_ib_dev *dev = to_mdev(ibdev);
89 struct ib_smp *in_mad = NULL;
90 struct ib_smp *out_mad = NULL;
91 int err = -ENOMEM;
92
93 in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL);
94 out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
95 if (!in_mad || !out_mad)
96 goto out;
97
98 init_query_mad(in_mad);
99 in_mad->attr_id = IB_SMP_ATTR_NODE_INFO;
100
101 err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, 1, NULL, NULL, in_mad, out_mad);
102 if (err)
103 goto out;
104
105 memset(props, 0, sizeof *props);
106
107 props->fw_ver = dev->dev->caps.fw_ver;
108 props->device_cap_flags = IB_DEVICE_CHANGE_PHY_PORT |
109 IB_DEVICE_PORT_ACTIVE_EVENT |
110 IB_DEVICE_SYS_IMAGE_GUID |
111 IB_DEVICE_RC_RNR_NAK_GEN |
112 IB_DEVICE_BLOCK_MULTICAST_LOOPBACK;
113 if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BAD_PKEY_CNTR)
114 props->device_cap_flags |= IB_DEVICE_BAD_PKEY_CNTR;
115 if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BAD_QKEY_CNTR)
116 props->device_cap_flags |= IB_DEVICE_BAD_QKEY_CNTR;
117 if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_APM)
118 props->device_cap_flags |= IB_DEVICE_AUTO_PATH_MIG;
119 if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_UD_AV_PORT)
120 props->device_cap_flags |= IB_DEVICE_UD_AV_PORT_ENFORCE;
121 if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_IPOIB_CSUM)
122 props->device_cap_flags |= IB_DEVICE_UD_IP_CSUM;
123 if (dev->dev->caps.max_gso_sz && dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BLH)
124 props->device_cap_flags |= IB_DEVICE_UD_TSO;
125 if (dev->dev->caps.bmme_flags & MLX4_BMME_FLAG_RESERVED_LKEY)
126 props->device_cap_flags |= IB_DEVICE_LOCAL_DMA_LKEY;
127 if ((dev->dev->caps.bmme_flags & MLX4_BMME_FLAG_LOCAL_INV) &&
128 (dev->dev->caps.bmme_flags & MLX4_BMME_FLAG_REMOTE_INV) &&
129 (dev->dev->caps.bmme_flags & MLX4_BMME_FLAG_FAST_REG_WR))
130 props->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS;
131 if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC)
132 props->device_cap_flags |= IB_DEVICE_XRC;
133
134 props->vendor_id = be32_to_cpup((__be32 *) (out_mad->data + 36)) &
135 0xffffff;
136 props->vendor_part_id = be16_to_cpup((__be16 *) (out_mad->data + 30));
137 props->hw_ver = be32_to_cpup((__be32 *) (out_mad->data + 32));
138 memcpy(&props->sys_image_guid, out_mad->data + 4, 8);
139
140 props->max_mr_size = ~0ull;
141 props->page_size_cap = dev->dev->caps.page_size_cap;
142 props->max_qp = dev->dev->caps.num_qps - dev->dev->caps.reserved_qps;
143 props->max_qp_wr = dev->dev->caps.max_wqes;
144 props->max_sge = min(dev->dev->caps.max_sq_sg,
145 dev->dev->caps.max_rq_sg);
146 props->max_cq = dev->dev->caps.num_cqs - dev->dev->caps.reserved_cqs;
147 props->max_cqe = dev->dev->caps.max_cqes;
148 props->max_mr = dev->dev->caps.num_mpts - dev->dev->caps.reserved_mrws;
149 props->max_pd = dev->dev->caps.num_pds - dev->dev->caps.reserved_pds;
150 props->max_qp_rd_atom = dev->dev->caps.max_qp_dest_rdma;
151 props->max_qp_init_rd_atom = dev->dev->caps.max_qp_init_rdma;
152 props->max_res_rd_atom = props->max_qp_rd_atom * props->max_qp;
153 props->max_srq = dev->dev->caps.num_srqs - dev->dev->caps.reserved_srqs;
154 props->max_srq_wr = dev->dev->caps.max_srq_wqes - 1;
155 props->max_srq_sge = dev->dev->caps.max_srq_sge;
156 props->max_fast_reg_page_list_len = MLX4_MAX_FAST_REG_PAGES;
157 props->local_ca_ack_delay = dev->dev->caps.local_ca_ack_delay;
158 props->atomic_cap = dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_ATOMIC ?
159 IB_ATOMIC_HCA : IB_ATOMIC_NONE;
160 props->masked_atomic_cap = IB_ATOMIC_HCA;
161 props->max_pkeys = dev->dev->caps.pkey_table_len[1];
162 props->max_mcast_grp = dev->dev->caps.num_mgms + dev->dev->caps.num_amgms;
163 props->max_mcast_qp_attach = dev->dev->caps.num_qp_per_mgm;
164 props->max_total_mcast_qp_attach = props->max_mcast_qp_attach *
165 props->max_mcast_grp;
166 props->max_map_per_fmr = dev->dev->caps.max_fmr_maps;
167
168out:
169 kfree(in_mad);
170 kfree(out_mad);
171
172 return err;
173}
174
175static enum rdma_link_layer
176mlx4_ib_port_link_layer(struct ib_device *device, u8 port_num)
177{
178 struct mlx4_dev *dev = to_mdev(device)->dev;
179
180 return dev->caps.port_mask[port_num] == MLX4_PORT_TYPE_IB ?
181 IB_LINK_LAYER_INFINIBAND : IB_LINK_LAYER_ETHERNET;
182}
183
184static int ib_link_query_port(struct ib_device *ibdev, u8 port,
185 struct ib_port_attr *props)
186{
187 struct ib_smp *in_mad = NULL;
188 struct ib_smp *out_mad = NULL;
189 int ext_active_speed;
190 int err = -ENOMEM;
191
192 in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL);
193 out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
194 if (!in_mad || !out_mad)
195 goto out;
196
197 init_query_mad(in_mad);
198 in_mad->attr_id = IB_SMP_ATTR_PORT_INFO;
199 in_mad->attr_mod = cpu_to_be32(port);
200
201 err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL,
202 in_mad, out_mad);
203 if (err)
204 goto out;
205
206
207 props->lid = be16_to_cpup((__be16 *) (out_mad->data + 16));
208 props->lmc = out_mad->data[34] & 0x7;
209 props->sm_lid = be16_to_cpup((__be16 *) (out_mad->data + 18));
210 props->sm_sl = out_mad->data[36] & 0xf;
211 props->state = out_mad->data[32] & 0xf;
212 props->phys_state = out_mad->data[33] >> 4;
213 props->port_cap_flags = be32_to_cpup((__be32 *) (out_mad->data + 20));
214 props->gid_tbl_len = to_mdev(ibdev)->dev->caps.gid_table_len[port];
215 props->max_msg_sz = to_mdev(ibdev)->dev->caps.max_msg_sz;
216 props->pkey_tbl_len = to_mdev(ibdev)->dev->caps.pkey_table_len[port];
217 props->bad_pkey_cntr = be16_to_cpup((__be16 *) (out_mad->data + 46));
218 props->qkey_viol_cntr = be16_to_cpup((__be16 *) (out_mad->data + 48));
219 props->active_width = out_mad->data[31] & 0xf;
220 props->active_speed = out_mad->data[35] >> 4;
221 props->max_mtu = out_mad->data[41] & 0xf;
222 props->active_mtu = out_mad->data[36] >> 4;
223 props->subnet_timeout = out_mad->data[51] & 0x1f;
224 props->max_vl_num = out_mad->data[37] >> 4;
225 props->init_type_reply = out_mad->data[41] >> 4;
226
227
228 if (props->port_cap_flags & IB_PORT_EXTENDED_SPEEDS_SUP) {
229 ext_active_speed = out_mad->data[62] >> 4;
230
231 switch (ext_active_speed) {
232 case 1:
233 props->active_speed = IB_SPEED_FDR;
234 break;
235 case 2:
236 props->active_speed = IB_SPEED_EDR;
237 break;
238 }
239 }
240
241
242 if (props->active_speed == IB_SPEED_QDR) {
243 init_query_mad(in_mad);
244 in_mad->attr_id = MLX4_ATTR_EXTENDED_PORT_INFO;
245 in_mad->attr_mod = cpu_to_be32(port);
246
247 err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port,
248 NULL, NULL, in_mad, out_mad);
249 if (err)
250 goto out;
251
252
253 if (out_mad->data[15] & 0x1)
254 props->active_speed = IB_SPEED_FDR10;
255 }
256
257
258 if (props->state == IB_PORT_DOWN)
259 props->active_speed = IB_SPEED_SDR;
260
261out:
262 kfree(in_mad);
263 kfree(out_mad);
264 return err;
265}
266
267static u8 state_to_phys_state(enum ib_port_state state)
268{
269 return state == IB_PORT_ACTIVE ? 5 : 3;
270}
271
272static int eth_link_query_port(struct ib_device *ibdev, u8 port,
273 struct ib_port_attr *props)
274{
275
276 struct mlx4_ib_dev *mdev = to_mdev(ibdev);
277 struct mlx4_ib_iboe *iboe = &mdev->iboe;
278 struct net_device *ndev;
279 enum ib_mtu tmp;
280 struct mlx4_cmd_mailbox *mailbox;
281 int err = 0;
282
283 mailbox = mlx4_alloc_cmd_mailbox(mdev->dev);
284 if (IS_ERR(mailbox))
285 return PTR_ERR(mailbox);
286
287 err = mlx4_cmd_box(mdev->dev, 0, mailbox->dma, port, 0,
288 MLX4_CMD_QUERY_PORT, MLX4_CMD_TIME_CLASS_B,
289 MLX4_CMD_WRAPPED);
290 if (err)
291 goto out;
292
293 props->active_width = (((u8 *)mailbox->buf)[5] == 0x40) ?
294 IB_WIDTH_4X : IB_WIDTH_1X;
295 props->active_speed = IB_SPEED_QDR;
296 props->port_cap_flags = IB_PORT_CM_SUP;
297 props->gid_tbl_len = mdev->dev->caps.gid_table_len[port];
298 props->max_msg_sz = mdev->dev->caps.max_msg_sz;
299 props->pkey_tbl_len = 1;
300 props->max_mtu = IB_MTU_4096;
301 props->max_vl_num = 2;
302 props->state = IB_PORT_DOWN;
303 props->phys_state = state_to_phys_state(props->state);
304 props->active_mtu = IB_MTU_256;
305 spin_lock(&iboe->lock);
306 ndev = iboe->netdevs[port - 1];
307 if (!ndev)
308 goto out_unlock;
309
310 tmp = iboe_get_mtu(ndev->mtu);
311 props->active_mtu = tmp ? min(props->max_mtu, tmp) : IB_MTU_256;
312
313 props->state = (netif_running(ndev) && netif_carrier_ok(ndev)) ?
314 IB_PORT_ACTIVE : IB_PORT_DOWN;
315 props->phys_state = state_to_phys_state(props->state);
316out_unlock:
317 spin_unlock(&iboe->lock);
318out:
319 mlx4_free_cmd_mailbox(mdev->dev, mailbox);
320 return err;
321}
322
323static int mlx4_ib_query_port(struct ib_device *ibdev, u8 port,
324 struct ib_port_attr *props)
325{
326 int err;
327
328 memset(props, 0, sizeof *props);
329
330 err = mlx4_ib_port_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND ?
331 ib_link_query_port(ibdev, port, props) :
332 eth_link_query_port(ibdev, port, props);
333
334 return err;
335}
336
337static int __mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
338 union ib_gid *gid)
339{
340 struct ib_smp *in_mad = NULL;
341 struct ib_smp *out_mad = NULL;
342 int err = -ENOMEM;
343
344 in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL);
345 out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
346 if (!in_mad || !out_mad)
347 goto out;
348
349 init_query_mad(in_mad);
350 in_mad->attr_id = IB_SMP_ATTR_PORT_INFO;
351 in_mad->attr_mod = cpu_to_be32(port);
352
353 err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
354 if (err)
355 goto out;
356
357 memcpy(gid->raw, out_mad->data + 8, 8);
358
359 init_query_mad(in_mad);
360 in_mad->attr_id = IB_SMP_ATTR_GUID_INFO;
361 in_mad->attr_mod = cpu_to_be32(index / 8);
362
363 err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
364 if (err)
365 goto out;
366
367 memcpy(gid->raw + 8, out_mad->data + (index % 8) * 8, 8);
368
369out:
370 kfree(in_mad);
371 kfree(out_mad);
372 return err;
373}
374
375static int iboe_query_gid(struct ib_device *ibdev, u8 port, int index,
376 union ib_gid *gid)
377{
378 struct mlx4_ib_dev *dev = to_mdev(ibdev);
379
380 *gid = dev->iboe.gid_table[port - 1][index];
381
382 return 0;
383}
384
385static int mlx4_ib_query_gid(struct ib_device *ibdev, u8 port, int index,
386 union ib_gid *gid)
387{
388 if (rdma_port_get_link_layer(ibdev, port) == IB_LINK_LAYER_INFINIBAND)
389 return __mlx4_ib_query_gid(ibdev, port, index, gid);
390 else
391 return iboe_query_gid(ibdev, port, index, gid);
392}
393
394static int mlx4_ib_query_pkey(struct ib_device *ibdev, u8 port, u16 index,
395 u16 *pkey)
396{
397 struct ib_smp *in_mad = NULL;
398 struct ib_smp *out_mad = NULL;
399 int err = -ENOMEM;
400
401 in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL);
402 out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
403 if (!in_mad || !out_mad)
404 goto out;
405
406 init_query_mad(in_mad);
407 in_mad->attr_id = IB_SMP_ATTR_PKEY_TABLE;
408 in_mad->attr_mod = cpu_to_be32(index / 32);
409
410 err = mlx4_MAD_IFC(to_mdev(ibdev), 1, 1, port, NULL, NULL, in_mad, out_mad);
411 if (err)
412 goto out;
413
414 *pkey = be16_to_cpu(((__be16 *) out_mad->data)[index % 32]);
415
416out:
417 kfree(in_mad);
418 kfree(out_mad);
419 return err;
420}
421
422static int mlx4_ib_modify_device(struct ib_device *ibdev, int mask,
423 struct ib_device_modify *props)
424{
425 struct mlx4_cmd_mailbox *mailbox;
426
427 if (mask & ~IB_DEVICE_MODIFY_NODE_DESC)
428 return -EOPNOTSUPP;
429
430 if (!(mask & IB_DEVICE_MODIFY_NODE_DESC))
431 return 0;
432
433 spin_lock(&to_mdev(ibdev)->sm_lock);
434 memcpy(ibdev->node_desc, props->node_desc, 64);
435 spin_unlock(&to_mdev(ibdev)->sm_lock);
436
437
438
439
440
441 mailbox = mlx4_alloc_cmd_mailbox(to_mdev(ibdev)->dev);
442 if (IS_ERR(mailbox))
443 return 0;
444
445 memset(mailbox->buf, 0, 256);
446 memcpy(mailbox->buf, props->node_desc, 64);
447 mlx4_cmd(to_mdev(ibdev)->dev, mailbox->dma, 1, 0,
448 MLX4_CMD_SET_NODE, MLX4_CMD_TIME_CLASS_A, MLX4_CMD_WRAPPED);
449
450 mlx4_free_cmd_mailbox(to_mdev(ibdev)->dev, mailbox);
451
452 return 0;
453}
454
455static int mlx4_SET_PORT(struct mlx4_ib_dev *dev, u8 port, int reset_qkey_viols,
456 u32 cap_mask)
457{
458 struct mlx4_cmd_mailbox *mailbox;
459 int err;
460 u8 is_eth = dev->dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH;
461
462 mailbox = mlx4_alloc_cmd_mailbox(dev->dev);
463 if (IS_ERR(mailbox))
464 return PTR_ERR(mailbox);
465
466 memset(mailbox->buf, 0, 256);
467
468 if (dev->dev->flags & MLX4_FLAG_OLD_PORT_CMDS) {
469 *(u8 *) mailbox->buf = !!reset_qkey_viols << 6;
470 ((__be32 *) mailbox->buf)[2] = cpu_to_be32(cap_mask);
471 } else {
472 ((u8 *) mailbox->buf)[3] = !!reset_qkey_viols;
473 ((__be32 *) mailbox->buf)[1] = cpu_to_be32(cap_mask);
474 }
475
476 err = mlx4_cmd(dev->dev, mailbox->dma, port, is_eth, MLX4_CMD_SET_PORT,
477 MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
478
479 mlx4_free_cmd_mailbox(dev->dev, mailbox);
480 return err;
481}
482
483static int mlx4_ib_modify_port(struct ib_device *ibdev, u8 port, int mask,
484 struct ib_port_modify *props)
485{
486 struct ib_port_attr attr;
487 u32 cap_mask;
488 int err;
489
490 mutex_lock(&to_mdev(ibdev)->cap_mask_mutex);
491
492 err = mlx4_ib_query_port(ibdev, port, &attr);
493 if (err)
494 goto out;
495
496 cap_mask = (attr.port_cap_flags | props->set_port_cap_mask) &
497 ~props->clr_port_cap_mask;
498
499 err = mlx4_SET_PORT(to_mdev(ibdev), port,
500 !!(mask & IB_PORT_RESET_QKEY_CNTR),
501 cap_mask);
502
503out:
504 mutex_unlock(&to_mdev(ibdev)->cap_mask_mutex);
505 return err;
506}
507
508static struct ib_ucontext *mlx4_ib_alloc_ucontext(struct ib_device *ibdev,
509 struct ib_udata *udata)
510{
511 struct mlx4_ib_dev *dev = to_mdev(ibdev);
512 struct mlx4_ib_ucontext *context;
513 struct mlx4_ib_alloc_ucontext_resp resp;
514 int err;
515
516 if (!dev->ib_active)
517 return ERR_PTR(-EAGAIN);
518
519 resp.qp_tab_size = dev->dev->caps.num_qps;
520 resp.bf_reg_size = dev->dev->caps.bf_reg_size;
521 resp.bf_regs_per_page = dev->dev->caps.bf_regs_per_page;
522
523 context = kmalloc(sizeof *context, GFP_KERNEL);
524 if (!context)
525 return ERR_PTR(-ENOMEM);
526
527 err = mlx4_uar_alloc(to_mdev(ibdev)->dev, &context->uar);
528 if (err) {
529 kfree(context);
530 return ERR_PTR(err);
531 }
532
533 INIT_LIST_HEAD(&context->db_page_list);
534 mutex_init(&context->db_page_mutex);
535
536 err = ib_copy_to_udata(udata, &resp, sizeof resp);
537 if (err) {
538 mlx4_uar_free(to_mdev(ibdev)->dev, &context->uar);
539 kfree(context);
540 return ERR_PTR(-EFAULT);
541 }
542
543 return &context->ibucontext;
544}
545
546static int mlx4_ib_dealloc_ucontext(struct ib_ucontext *ibcontext)
547{
548 struct mlx4_ib_ucontext *context = to_mucontext(ibcontext);
549
550 mlx4_uar_free(to_mdev(ibcontext->device)->dev, &context->uar);
551 kfree(context);
552
553 return 0;
554}
555
556static int mlx4_ib_mmap(struct ib_ucontext *context, struct vm_area_struct *vma)
557{
558 struct mlx4_ib_dev *dev = to_mdev(context->device);
559
560 if (vma->vm_end - vma->vm_start != PAGE_SIZE)
561 return -EINVAL;
562
563 if (vma->vm_pgoff == 0) {
564 vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
565
566 if (io_remap_pfn_range(vma, vma->vm_start,
567 to_mucontext(context)->uar.pfn,
568 PAGE_SIZE, vma->vm_page_prot))
569 return -EAGAIN;
570 } else if (vma->vm_pgoff == 1 && dev->dev->caps.bf_reg_size != 0) {
571 vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
572
573 if (io_remap_pfn_range(vma, vma->vm_start,
574 to_mucontext(context)->uar.pfn +
575 dev->dev->caps.num_uars,
576 PAGE_SIZE, vma->vm_page_prot))
577 return -EAGAIN;
578 } else
579 return -EINVAL;
580
581 return 0;
582}
583
584static struct ib_pd *mlx4_ib_alloc_pd(struct ib_device *ibdev,
585 struct ib_ucontext *context,
586 struct ib_udata *udata)
587{
588 struct mlx4_ib_pd *pd;
589 int err;
590
591 pd = kmalloc(sizeof *pd, GFP_KERNEL);
592 if (!pd)
593 return ERR_PTR(-ENOMEM);
594
595 err = mlx4_pd_alloc(to_mdev(ibdev)->dev, &pd->pdn);
596 if (err) {
597 kfree(pd);
598 return ERR_PTR(err);
599 }
600
601 if (context)
602 if (ib_copy_to_udata(udata, &pd->pdn, sizeof (__u32))) {
603 mlx4_pd_free(to_mdev(ibdev)->dev, pd->pdn);
604 kfree(pd);
605 return ERR_PTR(-EFAULT);
606 }
607
608 return &pd->ibpd;
609}
610
611static int mlx4_ib_dealloc_pd(struct ib_pd *pd)
612{
613 mlx4_pd_free(to_mdev(pd->device)->dev, to_mpd(pd)->pdn);
614 kfree(pd);
615
616 return 0;
617}
618
619static struct ib_xrcd *mlx4_ib_alloc_xrcd(struct ib_device *ibdev,
620 struct ib_ucontext *context,
621 struct ib_udata *udata)
622{
623 struct mlx4_ib_xrcd *xrcd;
624 int err;
625
626 if (!(to_mdev(ibdev)->dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC))
627 return ERR_PTR(-ENOSYS);
628
629 xrcd = kmalloc(sizeof *xrcd, GFP_KERNEL);
630 if (!xrcd)
631 return ERR_PTR(-ENOMEM);
632
633 err = mlx4_xrcd_alloc(to_mdev(ibdev)->dev, &xrcd->xrcdn);
634 if (err)
635 goto err1;
636
637 xrcd->pd = ib_alloc_pd(ibdev);
638 if (IS_ERR(xrcd->pd)) {
639 err = PTR_ERR(xrcd->pd);
640 goto err2;
641 }
642
643 xrcd->cq = ib_create_cq(ibdev, NULL, NULL, xrcd, 1, 0);
644 if (IS_ERR(xrcd->cq)) {
645 err = PTR_ERR(xrcd->cq);
646 goto err3;
647 }
648
649 return &xrcd->ibxrcd;
650
651err3:
652 ib_dealloc_pd(xrcd->pd);
653err2:
654 mlx4_xrcd_free(to_mdev(ibdev)->dev, xrcd->xrcdn);
655err1:
656 kfree(xrcd);
657 return ERR_PTR(err);
658}
659
660static int mlx4_ib_dealloc_xrcd(struct ib_xrcd *xrcd)
661{
662 ib_destroy_cq(to_mxrcd(xrcd)->cq);
663 ib_dealloc_pd(to_mxrcd(xrcd)->pd);
664 mlx4_xrcd_free(to_mdev(xrcd->device)->dev, to_mxrcd(xrcd)->xrcdn);
665 kfree(xrcd);
666
667 return 0;
668}
669
670static int add_gid_entry(struct ib_qp *ibqp, union ib_gid *gid)
671{
672 struct mlx4_ib_qp *mqp = to_mqp(ibqp);
673 struct mlx4_ib_dev *mdev = to_mdev(ibqp->device);
674 struct mlx4_ib_gid_entry *ge;
675
676 ge = kzalloc(sizeof *ge, GFP_KERNEL);
677 if (!ge)
678 return -ENOMEM;
679
680 ge->gid = *gid;
681 if (mlx4_ib_add_mc(mdev, mqp, gid)) {
682 ge->port = mqp->port;
683 ge->added = 1;
684 }
685
686 mutex_lock(&mqp->mutex);
687 list_add_tail(&ge->list, &mqp->gid_list);
688 mutex_unlock(&mqp->mutex);
689
690 return 0;
691}
692
693int mlx4_ib_add_mc(struct mlx4_ib_dev *mdev, struct mlx4_ib_qp *mqp,
694 union ib_gid *gid)
695{
696 u8 mac[6];
697 struct net_device *ndev;
698 int ret = 0;
699
700 if (!mqp->port)
701 return 0;
702
703 spin_lock(&mdev->iboe.lock);
704 ndev = mdev->iboe.netdevs[mqp->port - 1];
705 if (ndev)
706 dev_hold(ndev);
707 spin_unlock(&mdev->iboe.lock);
708
709 if (ndev) {
710 rdma_get_mcast_mac((struct in6_addr *)gid, mac);
711 rtnl_lock();
712 dev_mc_add(mdev->iboe.netdevs[mqp->port - 1], mac);
713 ret = 1;
714 rtnl_unlock();
715 dev_put(ndev);
716 }
717
718 return ret;
719}
720
721static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
722{
723 int err;
724 struct mlx4_ib_dev *mdev = to_mdev(ibqp->device);
725 struct mlx4_ib_qp *mqp = to_mqp(ibqp);
726
727 err = mlx4_multicast_attach(mdev->dev, &mqp->mqp, gid->raw,
728 !!(mqp->flags & MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK),
729 MLX4_PROT_IB_IPV6);
730 if (err)
731 return err;
732
733 err = add_gid_entry(ibqp, gid);
734 if (err)
735 goto err_add;
736
737 return 0;
738
739err_add:
740 mlx4_multicast_detach(mdev->dev, &mqp->mqp, gid->raw, MLX4_PROT_IB_IPV6);
741 return err;
742}
743
744static struct mlx4_ib_gid_entry *find_gid_entry(struct mlx4_ib_qp *qp, u8 *raw)
745{
746 struct mlx4_ib_gid_entry *ge;
747 struct mlx4_ib_gid_entry *tmp;
748 struct mlx4_ib_gid_entry *ret = NULL;
749
750 list_for_each_entry_safe(ge, tmp, &qp->gid_list, list) {
751 if (!memcmp(raw, ge->gid.raw, 16)) {
752 ret = ge;
753 break;
754 }
755 }
756
757 return ret;
758}
759
760static int mlx4_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
761{
762 int err;
763 struct mlx4_ib_dev *mdev = to_mdev(ibqp->device);
764 struct mlx4_ib_qp *mqp = to_mqp(ibqp);
765 u8 mac[6];
766 struct net_device *ndev;
767 struct mlx4_ib_gid_entry *ge;
768
769 err = mlx4_multicast_detach(mdev->dev,
770 &mqp->mqp, gid->raw, MLX4_PROT_IB_IPV6);
771 if (err)
772 return err;
773
774 mutex_lock(&mqp->mutex);
775 ge = find_gid_entry(mqp, gid->raw);
776 if (ge) {
777 spin_lock(&mdev->iboe.lock);
778 ndev = ge->added ? mdev->iboe.netdevs[ge->port - 1] : NULL;
779 if (ndev)
780 dev_hold(ndev);
781 spin_unlock(&mdev->iboe.lock);
782 rdma_get_mcast_mac((struct in6_addr *)gid, mac);
783 if (ndev) {
784 rtnl_lock();
785 dev_mc_del(mdev->iboe.netdevs[ge->port - 1], mac);
786 rtnl_unlock();
787 dev_put(ndev);
788 }
789 list_del(&ge->list);
790 kfree(ge);
791 } else
792 printk(KERN_WARNING "could not find mgid entry\n");
793
794 mutex_unlock(&mqp->mutex);
795
796 return 0;
797}
798
799static int init_node_data(struct mlx4_ib_dev *dev)
800{
801 struct ib_smp *in_mad = NULL;
802 struct ib_smp *out_mad = NULL;
803 int err = -ENOMEM;
804
805 in_mad = kzalloc(sizeof *in_mad, GFP_KERNEL);
806 out_mad = kmalloc(sizeof *out_mad, GFP_KERNEL);
807 if (!in_mad || !out_mad)
808 goto out;
809
810 init_query_mad(in_mad);
811 in_mad->attr_id = IB_SMP_ATTR_NODE_DESC;
812
813 err = mlx4_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad);
814 if (err)
815 goto out;
816
817 memcpy(dev->ib_dev.node_desc, out_mad->data, 64);
818
819 in_mad->attr_id = IB_SMP_ATTR_NODE_INFO;
820
821 err = mlx4_MAD_IFC(dev, 1, 1, 1, NULL, NULL, in_mad, out_mad);
822 if (err)
823 goto out;
824
825 memcpy(&dev->ib_dev.node_guid, out_mad->data + 12, 8);
826
827out:
828 kfree(in_mad);
829 kfree(out_mad);
830 return err;
831}
832
833static ssize_t show_hca(struct device *device, struct device_attribute *attr,
834 char *buf)
835{
836 struct mlx4_ib_dev *dev =
837 container_of(device, struct mlx4_ib_dev, ib_dev.dev);
838 return sprintf(buf, "MT%d\n", dev->dev->pdev->device);
839}
840
841static ssize_t show_fw_ver(struct device *device, struct device_attribute *attr,
842 char *buf)
843{
844 struct mlx4_ib_dev *dev =
845 container_of(device, struct mlx4_ib_dev, ib_dev.dev);
846 return sprintf(buf, "%d.%d.%d\n", (int) (dev->dev->caps.fw_ver >> 32),
847 (int) (dev->dev->caps.fw_ver >> 16) & 0xffff,
848 (int) dev->dev->caps.fw_ver & 0xffff);
849}
850
851static ssize_t show_rev(struct device *device, struct device_attribute *attr,
852 char *buf)
853{
854 struct mlx4_ib_dev *dev =
855 container_of(device, struct mlx4_ib_dev, ib_dev.dev);
856 return sprintf(buf, "%x\n", dev->dev->rev_id);
857}
858
859static ssize_t show_board(struct device *device, struct device_attribute *attr,
860 char *buf)
861{
862 struct mlx4_ib_dev *dev =
863 container_of(device, struct mlx4_ib_dev, ib_dev.dev);
864 return sprintf(buf, "%.*s\n", MLX4_BOARD_ID_LEN,
865 dev->dev->board_id);
866}
867
868static DEVICE_ATTR(hw_rev, S_IRUGO, show_rev, NULL);
869static DEVICE_ATTR(fw_ver, S_IRUGO, show_fw_ver, NULL);
870static DEVICE_ATTR(hca_type, S_IRUGO, show_hca, NULL);
871static DEVICE_ATTR(board_id, S_IRUGO, show_board, NULL);
872
873static struct device_attribute *mlx4_class_attributes[] = {
874 &dev_attr_hw_rev,
875 &dev_attr_fw_ver,
876 &dev_attr_hca_type,
877 &dev_attr_board_id
878};
879
880static void mlx4_addrconf_ifid_eui48(u8 *eui, u16 vlan_id, struct net_device *dev)
881{
882 memcpy(eui, dev->dev_addr, 3);
883 memcpy(eui + 5, dev->dev_addr + 3, 3);
884 if (vlan_id < 0x1000) {
885 eui[3] = vlan_id >> 8;
886 eui[4] = vlan_id & 0xff;
887 } else {
888 eui[3] = 0xff;
889 eui[4] = 0xfe;
890 }
891 eui[0] ^= 2;
892}
893
894static void update_gids_task(struct work_struct *work)
895{
896 struct update_gid_work *gw = container_of(work, struct update_gid_work, work);
897 struct mlx4_cmd_mailbox *mailbox;
898 union ib_gid *gids;
899 int err;
900 struct mlx4_dev *dev = gw->dev->dev;
901 struct ib_event event;
902
903 mailbox = mlx4_alloc_cmd_mailbox(dev);
904 if (IS_ERR(mailbox)) {
905 printk(KERN_WARNING "update gid table failed %ld\n", PTR_ERR(mailbox));
906 return;
907 }
908
909 gids = mailbox->buf;
910 memcpy(gids, gw->gids, sizeof gw->gids);
911
912 err = mlx4_cmd(dev, mailbox->dma, MLX4_SET_PORT_GID_TABLE << 8 | gw->port,
913 1, MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B,
914 MLX4_CMD_NATIVE);
915 if (err)
916 printk(KERN_WARNING "set port command failed\n");
917 else {
918 memcpy(gw->dev->iboe.gid_table[gw->port - 1], gw->gids, sizeof gw->gids);
919 event.device = &gw->dev->ib_dev;
920 event.element.port_num = gw->port;
921 event.event = IB_EVENT_GID_CHANGE;
922 ib_dispatch_event(&event);
923 }
924
925 mlx4_free_cmd_mailbox(dev, mailbox);
926 kfree(gw);
927}
928
929static int update_ipv6_gids(struct mlx4_ib_dev *dev, int port, int clear)
930{
931 struct net_device *ndev = dev->iboe.netdevs[port - 1];
932 struct update_gid_work *work;
933 struct net_device *tmp;
934 int i;
935 u8 *hits;
936 int ret;
937 union ib_gid gid;
938 int free;
939 int found;
940 int need_update = 0;
941 u16 vid;
942
943 work = kzalloc(sizeof *work, GFP_ATOMIC);
944 if (!work)
945 return -ENOMEM;
946
947 hits = kzalloc(128, GFP_ATOMIC);
948 if (!hits) {
949 ret = -ENOMEM;
950 goto out;
951 }
952
953 rcu_read_lock();
954 for_each_netdev_rcu(&init_net, tmp) {
955 if (ndev && (tmp == ndev || rdma_vlan_dev_real_dev(tmp) == ndev)) {
956 gid.global.subnet_prefix = cpu_to_be64(0xfe80000000000000LL);
957 vid = rdma_vlan_dev_vlan_id(tmp);
958 mlx4_addrconf_ifid_eui48(&gid.raw[8], vid, ndev);
959 found = 0;
960 free = -1;
961 for (i = 0; i < 128; ++i) {
962 if (free < 0 &&
963 !memcmp(&dev->iboe.gid_table[port - 1][i], &zgid, sizeof zgid))
964 free = i;
965 if (!memcmp(&dev->iboe.gid_table[port - 1][i], &gid, sizeof gid)) {
966 hits[i] = 1;
967 found = 1;
968 break;
969 }
970 }
971
972 if (!found) {
973 if (tmp == ndev &&
974 (memcmp(&dev->iboe.gid_table[port - 1][0],
975 &gid, sizeof gid) ||
976 !memcmp(&dev->iboe.gid_table[port - 1][0],
977 &zgid, sizeof gid))) {
978 dev->iboe.gid_table[port - 1][0] = gid;
979 ++need_update;
980 hits[0] = 1;
981 } else if (free >= 0) {
982 dev->iboe.gid_table[port - 1][free] = gid;
983 hits[free] = 1;
984 ++need_update;
985 }
986 }
987 }
988 }
989 rcu_read_unlock();
990
991 for (i = 0; i < 128; ++i)
992 if (!hits[i]) {
993 if (memcmp(&dev->iboe.gid_table[port - 1][i], &zgid, sizeof zgid))
994 ++need_update;
995 dev->iboe.gid_table[port - 1][i] = zgid;
996 }
997
998 if (need_update) {
999 memcpy(work->gids, dev->iboe.gid_table[port - 1], sizeof work->gids);
1000 INIT_WORK(&work->work, update_gids_task);
1001 work->port = port;
1002 work->dev = dev;
1003 queue_work(wq, &work->work);
1004 } else
1005 kfree(work);
1006
1007 kfree(hits);
1008 return 0;
1009
1010out:
1011 kfree(work);
1012 return ret;
1013}
1014
1015static void handle_en_event(struct mlx4_ib_dev *dev, int port, unsigned long event)
1016{
1017 switch (event) {
1018 case NETDEV_UP:
1019 case NETDEV_CHANGEADDR:
1020 update_ipv6_gids(dev, port, 0);
1021 break;
1022
1023 case NETDEV_DOWN:
1024 update_ipv6_gids(dev, port, 1);
1025 dev->iboe.netdevs[port - 1] = NULL;
1026 }
1027}
1028
1029static void netdev_added(struct mlx4_ib_dev *dev, int port)
1030{
1031 update_ipv6_gids(dev, port, 0);
1032}
1033
1034static void netdev_removed(struct mlx4_ib_dev *dev, int port)
1035{
1036 update_ipv6_gids(dev, port, 1);
1037}
1038
1039static int mlx4_ib_netdev_event(struct notifier_block *this, unsigned long event,
1040 void *ptr)
1041{
1042 struct net_device *dev = ptr;
1043 struct mlx4_ib_dev *ibdev;
1044 struct net_device *oldnd;
1045 struct mlx4_ib_iboe *iboe;
1046 int port;
1047
1048 if (!net_eq(dev_net(dev), &init_net))
1049 return NOTIFY_DONE;
1050
1051 ibdev = container_of(this, struct mlx4_ib_dev, iboe.nb);
1052 iboe = &ibdev->iboe;
1053
1054 spin_lock(&iboe->lock);
1055 mlx4_foreach_ib_transport_port(port, ibdev->dev) {
1056 oldnd = iboe->netdevs[port - 1];
1057 iboe->netdevs[port - 1] =
1058 mlx4_get_protocol_dev(ibdev->dev, MLX4_PROT_ETH, port);
1059 if (oldnd != iboe->netdevs[port - 1]) {
1060 if (iboe->netdevs[port - 1])
1061 netdev_added(ibdev, port);
1062 else
1063 netdev_removed(ibdev, port);
1064 }
1065 }
1066
1067 if (dev == iboe->netdevs[0] ||
1068 (iboe->netdevs[0] && rdma_vlan_dev_real_dev(dev) == iboe->netdevs[0]))
1069 handle_en_event(ibdev, 1, event);
1070 else if (dev == iboe->netdevs[1]
1071 || (iboe->netdevs[1] && rdma_vlan_dev_real_dev(dev) == iboe->netdevs[1]))
1072 handle_en_event(ibdev, 2, event);
1073
1074 spin_unlock(&iboe->lock);
1075
1076 return NOTIFY_DONE;
1077}
1078
1079static void *mlx4_ib_add(struct mlx4_dev *dev)
1080{
1081 struct mlx4_ib_dev *ibdev;
1082 int num_ports = 0;
1083 int i;
1084 int err;
1085 struct mlx4_ib_iboe *iboe;
1086
1087 printk_once(KERN_INFO "%s", mlx4_ib_version);
1088
1089 if (mlx4_is_mfunc(dev)) {
1090 printk(KERN_WARNING "IB not yet supported in SRIOV\n");
1091 return NULL;
1092 }
1093
1094 mlx4_foreach_ib_transport_port(i, dev)
1095 num_ports++;
1096
1097
1098 if (num_ports == 0)
1099 return NULL;
1100
1101 ibdev = (struct mlx4_ib_dev *) ib_alloc_device(sizeof *ibdev);
1102 if (!ibdev) {
1103 dev_err(&dev->pdev->dev, "Device struct alloc failed\n");
1104 return NULL;
1105 }
1106
1107 iboe = &ibdev->iboe;
1108
1109 if (mlx4_pd_alloc(dev, &ibdev->priv_pdn))
1110 goto err_dealloc;
1111
1112 if (mlx4_uar_alloc(dev, &ibdev->priv_uar))
1113 goto err_pd;
1114
1115 ibdev->uar_map = ioremap((phys_addr_t) ibdev->priv_uar.pfn << PAGE_SHIFT,
1116 PAGE_SIZE);
1117 if (!ibdev->uar_map)
1118 goto err_uar;
1119 MLX4_INIT_DOORBELL_LOCK(&ibdev->uar_lock);
1120
1121 ibdev->dev = dev;
1122
1123 strlcpy(ibdev->ib_dev.name, "mlx4_%d", IB_DEVICE_NAME_MAX);
1124 ibdev->ib_dev.owner = THIS_MODULE;
1125 ibdev->ib_dev.node_type = RDMA_NODE_IB_CA;
1126 ibdev->ib_dev.local_dma_lkey = dev->caps.reserved_lkey;
1127 ibdev->num_ports = num_ports;
1128 ibdev->ib_dev.phys_port_cnt = ibdev->num_ports;
1129 ibdev->ib_dev.num_comp_vectors = dev->caps.num_comp_vectors;
1130 ibdev->ib_dev.dma_device = &dev->pdev->dev;
1131
1132 ibdev->ib_dev.uverbs_abi_ver = MLX4_IB_UVERBS_ABI_VERSION;
1133 ibdev->ib_dev.uverbs_cmd_mask =
1134 (1ull << IB_USER_VERBS_CMD_GET_CONTEXT) |
1135 (1ull << IB_USER_VERBS_CMD_QUERY_DEVICE) |
1136 (1ull << IB_USER_VERBS_CMD_QUERY_PORT) |
1137 (1ull << IB_USER_VERBS_CMD_ALLOC_PD) |
1138 (1ull << IB_USER_VERBS_CMD_DEALLOC_PD) |
1139 (1ull << IB_USER_VERBS_CMD_REG_MR) |
1140 (1ull << IB_USER_VERBS_CMD_DEREG_MR) |
1141 (1ull << IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL) |
1142 (1ull << IB_USER_VERBS_CMD_CREATE_CQ) |
1143 (1ull << IB_USER_VERBS_CMD_RESIZE_CQ) |
1144 (1ull << IB_USER_VERBS_CMD_DESTROY_CQ) |
1145 (1ull << IB_USER_VERBS_CMD_CREATE_QP) |
1146 (1ull << IB_USER_VERBS_CMD_MODIFY_QP) |
1147 (1ull << IB_USER_VERBS_CMD_QUERY_QP) |
1148 (1ull << IB_USER_VERBS_CMD_DESTROY_QP) |
1149 (1ull << IB_USER_VERBS_CMD_ATTACH_MCAST) |
1150 (1ull << IB_USER_VERBS_CMD_DETACH_MCAST) |
1151 (1ull << IB_USER_VERBS_CMD_CREATE_SRQ) |
1152 (1ull << IB_USER_VERBS_CMD_MODIFY_SRQ) |
1153 (1ull << IB_USER_VERBS_CMD_QUERY_SRQ) |
1154 (1ull << IB_USER_VERBS_CMD_DESTROY_SRQ) |
1155 (1ull << IB_USER_VERBS_CMD_CREATE_XSRQ) |
1156 (1ull << IB_USER_VERBS_CMD_OPEN_QP);
1157
1158 ibdev->ib_dev.query_device = mlx4_ib_query_device;
1159 ibdev->ib_dev.query_port = mlx4_ib_query_port;
1160 ibdev->ib_dev.get_link_layer = mlx4_ib_port_link_layer;
1161 ibdev->ib_dev.query_gid = mlx4_ib_query_gid;
1162 ibdev->ib_dev.query_pkey = mlx4_ib_query_pkey;
1163 ibdev->ib_dev.modify_device = mlx4_ib_modify_device;
1164 ibdev->ib_dev.modify_port = mlx4_ib_modify_port;
1165 ibdev->ib_dev.alloc_ucontext = mlx4_ib_alloc_ucontext;
1166 ibdev->ib_dev.dealloc_ucontext = mlx4_ib_dealloc_ucontext;
1167 ibdev->ib_dev.mmap = mlx4_ib_mmap;
1168 ibdev->ib_dev.alloc_pd = mlx4_ib_alloc_pd;
1169 ibdev->ib_dev.dealloc_pd = mlx4_ib_dealloc_pd;
1170 ibdev->ib_dev.create_ah = mlx4_ib_create_ah;
1171 ibdev->ib_dev.query_ah = mlx4_ib_query_ah;
1172 ibdev->ib_dev.destroy_ah = mlx4_ib_destroy_ah;
1173 ibdev->ib_dev.create_srq = mlx4_ib_create_srq;
1174 ibdev->ib_dev.modify_srq = mlx4_ib_modify_srq;
1175 ibdev->ib_dev.query_srq = mlx4_ib_query_srq;
1176 ibdev->ib_dev.destroy_srq = mlx4_ib_destroy_srq;
1177 ibdev->ib_dev.post_srq_recv = mlx4_ib_post_srq_recv;
1178 ibdev->ib_dev.create_qp = mlx4_ib_create_qp;
1179 ibdev->ib_dev.modify_qp = mlx4_ib_modify_qp;
1180 ibdev->ib_dev.query_qp = mlx4_ib_query_qp;
1181 ibdev->ib_dev.destroy_qp = mlx4_ib_destroy_qp;
1182 ibdev->ib_dev.post_send = mlx4_ib_post_send;
1183 ibdev->ib_dev.post_recv = mlx4_ib_post_recv;
1184 ibdev->ib_dev.create_cq = mlx4_ib_create_cq;
1185 ibdev->ib_dev.modify_cq = mlx4_ib_modify_cq;
1186 ibdev->ib_dev.resize_cq = mlx4_ib_resize_cq;
1187 ibdev->ib_dev.destroy_cq = mlx4_ib_destroy_cq;
1188 ibdev->ib_dev.poll_cq = mlx4_ib_poll_cq;
1189 ibdev->ib_dev.req_notify_cq = mlx4_ib_arm_cq;
1190 ibdev->ib_dev.get_dma_mr = mlx4_ib_get_dma_mr;
1191 ibdev->ib_dev.reg_user_mr = mlx4_ib_reg_user_mr;
1192 ibdev->ib_dev.dereg_mr = mlx4_ib_dereg_mr;
1193 ibdev->ib_dev.alloc_fast_reg_mr = mlx4_ib_alloc_fast_reg_mr;
1194 ibdev->ib_dev.alloc_fast_reg_page_list = mlx4_ib_alloc_fast_reg_page_list;
1195 ibdev->ib_dev.free_fast_reg_page_list = mlx4_ib_free_fast_reg_page_list;
1196 ibdev->ib_dev.attach_mcast = mlx4_ib_mcg_attach;
1197 ibdev->ib_dev.detach_mcast = mlx4_ib_mcg_detach;
1198 ibdev->ib_dev.process_mad = mlx4_ib_process_mad;
1199
1200 ibdev->ib_dev.alloc_fmr = mlx4_ib_fmr_alloc;
1201 ibdev->ib_dev.map_phys_fmr = mlx4_ib_map_phys_fmr;
1202 ibdev->ib_dev.unmap_fmr = mlx4_ib_unmap_fmr;
1203 ibdev->ib_dev.dealloc_fmr = mlx4_ib_fmr_dealloc;
1204
1205 if (dev->caps.flags & MLX4_DEV_CAP_FLAG_XRC) {
1206 ibdev->ib_dev.alloc_xrcd = mlx4_ib_alloc_xrcd;
1207 ibdev->ib_dev.dealloc_xrcd = mlx4_ib_dealloc_xrcd;
1208 ibdev->ib_dev.uverbs_cmd_mask |=
1209 (1ull << IB_USER_VERBS_CMD_OPEN_XRCD) |
1210 (1ull << IB_USER_VERBS_CMD_CLOSE_XRCD);
1211 }
1212
1213 spin_lock_init(&iboe->lock);
1214
1215 if (init_node_data(ibdev))
1216 goto err_map;
1217
1218 for (i = 0; i < ibdev->num_ports; ++i) {
1219 if (mlx4_ib_port_link_layer(&ibdev->ib_dev, i + 1) ==
1220 IB_LINK_LAYER_ETHERNET) {
1221 err = mlx4_counter_alloc(ibdev->dev, &ibdev->counters[i]);
1222 if (err)
1223 ibdev->counters[i] = -1;
1224 } else
1225 ibdev->counters[i] = -1;
1226 }
1227
1228 spin_lock_init(&ibdev->sm_lock);
1229 mutex_init(&ibdev->cap_mask_mutex);
1230
1231 if (ib_register_device(&ibdev->ib_dev, NULL))
1232 goto err_counter;
1233
1234 if (mlx4_ib_mad_init(ibdev))
1235 goto err_reg;
1236
1237 if (dev->caps.flags & MLX4_DEV_CAP_FLAG_IBOE && !iboe->nb.notifier_call) {
1238 iboe->nb.notifier_call = mlx4_ib_netdev_event;
1239 err = register_netdevice_notifier(&iboe->nb);
1240 if (err)
1241 goto err_reg;
1242 }
1243
1244 for (i = 0; i < ARRAY_SIZE(mlx4_class_attributes); ++i) {
1245 if (device_create_file(&ibdev->ib_dev.dev,
1246 mlx4_class_attributes[i]))
1247 goto err_notif;
1248 }
1249
1250 ibdev->ib_active = true;
1251
1252 return ibdev;
1253
1254err_notif:
1255 if (unregister_netdevice_notifier(&ibdev->iboe.nb))
1256 printk(KERN_WARNING "failure unregistering notifier\n");
1257 flush_workqueue(wq);
1258
1259err_reg:
1260 ib_unregister_device(&ibdev->ib_dev);
1261
1262err_counter:
1263 for (; i; --i)
1264 if (ibdev->counters[i - 1] != -1)
1265 mlx4_counter_free(ibdev->dev, ibdev->counters[i - 1]);
1266
1267err_map:
1268 iounmap(ibdev->uar_map);
1269
1270err_uar:
1271 mlx4_uar_free(dev, &ibdev->priv_uar);
1272
1273err_pd:
1274 mlx4_pd_free(dev, ibdev->priv_pdn);
1275
1276err_dealloc:
1277 ib_dealloc_device(&ibdev->ib_dev);
1278
1279 return NULL;
1280}
1281
1282static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
1283{
1284 struct mlx4_ib_dev *ibdev = ibdev_ptr;
1285 int p;
1286
1287 mlx4_ib_mad_cleanup(ibdev);
1288 ib_unregister_device(&ibdev->ib_dev);
1289 if (ibdev->iboe.nb.notifier_call) {
1290 if (unregister_netdevice_notifier(&ibdev->iboe.nb))
1291 printk(KERN_WARNING "failure unregistering notifier\n");
1292 ibdev->iboe.nb.notifier_call = NULL;
1293 }
1294 iounmap(ibdev->uar_map);
1295 for (p = 0; p < ibdev->num_ports; ++p)
1296 if (ibdev->counters[p] != -1)
1297 mlx4_counter_free(ibdev->dev, ibdev->counters[p]);
1298 mlx4_foreach_port(p, dev, MLX4_PORT_TYPE_IB)
1299 mlx4_CLOSE_PORT(dev, p);
1300
1301 mlx4_uar_free(dev, &ibdev->priv_uar);
1302 mlx4_pd_free(dev, ibdev->priv_pdn);
1303 ib_dealloc_device(&ibdev->ib_dev);
1304}
1305
1306static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
1307 enum mlx4_dev_event event, int port)
1308{
1309 struct ib_event ibev;
1310 struct mlx4_ib_dev *ibdev = to_mdev((struct ib_device *) ibdev_ptr);
1311
1312 if (port > ibdev->num_ports)
1313 return;
1314
1315 switch (event) {
1316 case MLX4_DEV_EVENT_PORT_UP:
1317 ibev.event = IB_EVENT_PORT_ACTIVE;
1318 break;
1319
1320 case MLX4_DEV_EVENT_PORT_DOWN:
1321 ibev.event = IB_EVENT_PORT_ERR;
1322 break;
1323
1324 case MLX4_DEV_EVENT_CATASTROPHIC_ERROR:
1325 ibdev->ib_active = false;
1326 ibev.event = IB_EVENT_DEVICE_FATAL;
1327 break;
1328
1329 default:
1330 return;
1331 }
1332
1333 ibev.device = ibdev_ptr;
1334 ibev.element.port_num = port;
1335
1336 ib_dispatch_event(&ibev);
1337}
1338
1339static struct mlx4_interface mlx4_ib_interface = {
1340 .add = mlx4_ib_add,
1341 .remove = mlx4_ib_remove,
1342 .event = mlx4_ib_event,
1343 .protocol = MLX4_PROT_IB_IPV6
1344};
1345
1346static int __init mlx4_ib_init(void)
1347{
1348 int err;
1349
1350 wq = create_singlethread_workqueue("mlx4_ib");
1351 if (!wq)
1352 return -ENOMEM;
1353
1354 err = mlx4_register_interface(&mlx4_ib_interface);
1355 if (err) {
1356 destroy_workqueue(wq);
1357 return err;
1358 }
1359
1360 return 0;
1361}
1362
1363static void __exit mlx4_ib_cleanup(void)
1364{
1365 mlx4_unregister_interface(&mlx4_ib_interface);
1366 destroy_workqueue(wq);
1367}
1368
1369module_init(mlx4_ib_init);
1370module_exit(mlx4_ib_cleanup);
1371
