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#include <linux/delay.h>
37#include <linux/moduleparam.h>
38#include <linux/dma-mapping.h>
39#include <linux/slab.h>
40
41#include <linux/ip.h>
42#include <linux/tcp.h>
43
44#include "ipoib.h"
45
46#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG_DATA
47static int data_debug_level;
48
49module_param(data_debug_level, int, 0644);
50MODULE_PARM_DESC(data_debug_level,
51 "Enable data path debug tracing if > 0");
52#endif
53
54static DEFINE_MUTEX(pkey_mutex);
55
56struct ipoib_ah *ipoib_create_ah(struct net_device *dev,
57 struct ib_pd *pd, struct ib_ah_attr *attr)
58{
59 struct ipoib_ah *ah;
60 struct ib_ah *vah;
61
62 ah = kmalloc(sizeof *ah, GFP_KERNEL);
63 if (!ah)
64 return ERR_PTR(-ENOMEM);
65
66 ah->dev = dev;
67 ah->last_send = 0;
68 kref_init(&ah->ref);
69
70 vah = ib_create_ah(pd, attr);
71 if (IS_ERR(vah)) {
72 kfree(ah);
73 ah = (struct ipoib_ah *)vah;
74 } else {
75 ah->ah = vah;
76 ipoib_dbg(netdev_priv(dev), "Created ah %p\n", ah->ah);
77 }
78
79 return ah;
80}
81
82void ipoib_free_ah(struct kref *kref)
83{
84 struct ipoib_ah *ah = container_of(kref, struct ipoib_ah, ref);
85 struct ipoib_dev_priv *priv = netdev_priv(ah->dev);
86
87 unsigned long flags;
88
89 spin_lock_irqsave(&priv->lock, flags);
90 list_add_tail(&ah->list, &priv->dead_ahs);
91 spin_unlock_irqrestore(&priv->lock, flags);
92}
93
94static void ipoib_ud_dma_unmap_rx(struct ipoib_dev_priv *priv,
95 u64 mapping[IPOIB_UD_RX_SG])
96{
97 if (ipoib_ud_need_sg(priv->max_ib_mtu)) {
98 ib_dma_unmap_single(priv->ca, mapping[0], IPOIB_UD_HEAD_SIZE,
99 DMA_FROM_DEVICE);
100 ib_dma_unmap_page(priv->ca, mapping[1], PAGE_SIZE,
101 DMA_FROM_DEVICE);
102 } else
103 ib_dma_unmap_single(priv->ca, mapping[0],
104 IPOIB_UD_BUF_SIZE(priv->max_ib_mtu),
105 DMA_FROM_DEVICE);
106}
107
108static void ipoib_ud_skb_put_frags(struct ipoib_dev_priv *priv,
109 struct sk_buff *skb,
110 unsigned int length)
111{
112 if (ipoib_ud_need_sg(priv->max_ib_mtu)) {
113 skb_frag_t *frag = &skb_shinfo(skb)->frags[0];
114 unsigned int size;
115
116
117
118
119 skb->tail += IPOIB_UD_HEAD_SIZE;
120 skb->len += length;
121
122 size = length - IPOIB_UD_HEAD_SIZE;
123
124 skb_frag_size_set(frag, size);
125 skb->data_len += size;
126 skb->truesize += PAGE_SIZE;
127 } else
128 skb_put(skb, length);
129
130}
131
132static int ipoib_ib_post_receive(struct net_device *dev, int id)
133{
134 struct ipoib_dev_priv *priv = netdev_priv(dev);
135 struct ib_recv_wr *bad_wr;
136 int ret;
137
138 priv->rx_wr.wr_id = id | IPOIB_OP_RECV;
139 priv->rx_sge[0].addr = priv->rx_ring[id].mapping[0];
140 priv->rx_sge[1].addr = priv->rx_ring[id].mapping[1];
141
142
143 ret = ib_post_recv(priv->qp, &priv->rx_wr, &bad_wr);
144 if (unlikely(ret)) {
145 ipoib_warn(priv, "receive failed for buf %d (%d)\n", id, ret);
146 ipoib_ud_dma_unmap_rx(priv, priv->rx_ring[id].mapping);
147 dev_kfree_skb_any(priv->rx_ring[id].skb);
148 priv->rx_ring[id].skb = NULL;
149 }
150
151 return ret;
152}
153
154static struct sk_buff *ipoib_alloc_rx_skb(struct net_device *dev, int id)
155{
156 struct ipoib_dev_priv *priv = netdev_priv(dev);
157 struct sk_buff *skb;
158 int buf_size;
159 int tailroom;
160 u64 *mapping;
161
162 if (ipoib_ud_need_sg(priv->max_ib_mtu)) {
163 buf_size = IPOIB_UD_HEAD_SIZE;
164 tailroom = 128;
165 } else {
166 buf_size = IPOIB_UD_BUF_SIZE(priv->max_ib_mtu);
167 tailroom = 0;
168 }
169
170 skb = dev_alloc_skb(buf_size + tailroom + 4);
171 if (unlikely(!skb))
172 return NULL;
173
174
175
176
177
178
179 skb_reserve(skb, 4);
180
181 mapping = priv->rx_ring[id].mapping;
182 mapping[0] = ib_dma_map_single(priv->ca, skb->data, buf_size,
183 DMA_FROM_DEVICE);
184 if (unlikely(ib_dma_mapping_error(priv->ca, mapping[0])))
185 goto error;
186
187 if (ipoib_ud_need_sg(priv->max_ib_mtu)) {
188 struct page *page = alloc_page(GFP_ATOMIC);
189 if (!page)
190 goto partial_error;
191 skb_fill_page_desc(skb, 0, page, 0, PAGE_SIZE);
192 mapping[1] =
193 ib_dma_map_page(priv->ca, page,
194 0, PAGE_SIZE, DMA_FROM_DEVICE);
195 if (unlikely(ib_dma_mapping_error(priv->ca, mapping[1])))
196 goto partial_error;
197 }
198
199 priv->rx_ring[id].skb = skb;
200 return skb;
201
202partial_error:
203 ib_dma_unmap_single(priv->ca, mapping[0], buf_size, DMA_FROM_DEVICE);
204error:
205 dev_kfree_skb_any(skb);
206 return NULL;
207}
208
209static int ipoib_ib_post_receives(struct net_device *dev)
210{
211 struct ipoib_dev_priv *priv = netdev_priv(dev);
212 int i;
213
214 for (i = 0; i < ipoib_recvq_size; ++i) {
215 if (!ipoib_alloc_rx_skb(dev, i)) {
216 ipoib_warn(priv, "failed to allocate receive buffer %d\n", i);
217 return -ENOMEM;
218 }
219 if (ipoib_ib_post_receive(dev, i)) {
220 ipoib_warn(priv, "ipoib_ib_post_receive failed for buf %d\n", i);
221 return -EIO;
222 }
223 }
224
225 return 0;
226}
227
228static void ipoib_ib_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
229{
230 struct ipoib_dev_priv *priv = netdev_priv(dev);
231 unsigned int wr_id = wc->wr_id & ~IPOIB_OP_RECV;
232 struct sk_buff *skb;
233 u64 mapping[IPOIB_UD_RX_SG];
234 union ib_gid *dgid;
235
236 ipoib_dbg_data(priv, "recv completion: id %d, status: %d\n",
237 wr_id, wc->status);
238
239 if (unlikely(wr_id >= ipoib_recvq_size)) {
240 ipoib_warn(priv, "recv completion event with wrid %d (> %d)\n",
241 wr_id, ipoib_recvq_size);
242 return;
243 }
244
245 skb = priv->rx_ring[wr_id].skb;
246
247 if (unlikely(wc->status != IB_WC_SUCCESS)) {
248 if (wc->status != IB_WC_WR_FLUSH_ERR)
249 ipoib_warn(priv, "failed recv event "
250 "(status=%d, wrid=%d vend_err %x)\n",
251 wc->status, wr_id, wc->vendor_err);
252 ipoib_ud_dma_unmap_rx(priv, priv->rx_ring[wr_id].mapping);
253 dev_kfree_skb_any(skb);
254 priv->rx_ring[wr_id].skb = NULL;
255 return;
256 }
257
258
259
260
261
262 if (wc->slid == priv->local_lid && wc->src_qp == priv->qp->qp_num)
263 goto repost;
264
265 memcpy(mapping, priv->rx_ring[wr_id].mapping,
266 IPOIB_UD_RX_SG * sizeof *mapping);
267
268
269
270
271
272 if (unlikely(!ipoib_alloc_rx_skb(dev, wr_id))) {
273 ++dev->stats.rx_dropped;
274 goto repost;
275 }
276
277 ipoib_dbg_data(priv, "received %d bytes, SLID 0x%04x\n",
278 wc->byte_len, wc->slid);
279
280 ipoib_ud_dma_unmap_rx(priv, mapping);
281 ipoib_ud_skb_put_frags(priv, skb, wc->byte_len);
282
283
284 dgid = &((struct ib_grh *)skb->data)->dgid;
285
286 if (!(wc->wc_flags & IB_WC_GRH) || dgid->raw[0] != 0xff)
287 skb->pkt_type = PACKET_HOST;
288 else if (memcmp(dgid, dev->broadcast + 4, sizeof(union ib_gid)) == 0)
289 skb->pkt_type = PACKET_BROADCAST;
290 else
291 skb->pkt_type = PACKET_MULTICAST;
292
293 skb_pull(skb, IB_GRH_BYTES);
294
295 skb->protocol = ((struct ipoib_header *) skb->data)->proto;
296 skb_reset_mac_header(skb);
297 skb_pull(skb, IPOIB_ENCAP_LEN);
298
299 ++dev->stats.rx_packets;
300 dev->stats.rx_bytes += skb->len;
301
302 skb->dev = dev;
303 if ((dev->features & NETIF_F_RXCSUM) &&
304 likely(wc->wc_flags & IB_WC_IP_CSUM_OK))
305 skb->ip_summed = CHECKSUM_UNNECESSARY;
306
307 napi_gro_receive(&priv->napi, skb);
308
309repost:
310 if (unlikely(ipoib_ib_post_receive(dev, wr_id)))
311 ipoib_warn(priv, "ipoib_ib_post_receive failed "
312 "for buf %d\n", wr_id);
313}
314
315static int ipoib_dma_map_tx(struct ib_device *ca,
316 struct ipoib_tx_buf *tx_req)
317{
318 struct sk_buff *skb = tx_req->skb;
319 u64 *mapping = tx_req->mapping;
320 int i;
321 int off;
322
323 if (skb_headlen(skb)) {
324 mapping[0] = ib_dma_map_single(ca, skb->data, skb_headlen(skb),
325 DMA_TO_DEVICE);
326 if (unlikely(ib_dma_mapping_error(ca, mapping[0])))
327 return -EIO;
328
329 off = 1;
330 } else
331 off = 0;
332
333 for (i = 0; i < skb_shinfo(skb)->nr_frags; ++i) {
334 const skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
335 mapping[i + off] = ib_dma_map_page(ca,
336 skb_frag_page(frag),
337 frag->page_offset, skb_frag_size(frag),
338 DMA_TO_DEVICE);
339 if (unlikely(ib_dma_mapping_error(ca, mapping[i + off])))
340 goto partial_error;
341 }
342 return 0;
343
344partial_error:
345 for (; i > 0; --i) {
346 const skb_frag_t *frag = &skb_shinfo(skb)->frags[i - 1];
347
348 ib_dma_unmap_page(ca, mapping[i - !off], skb_frag_size(frag), DMA_TO_DEVICE);
349 }
350
351 if (off)
352 ib_dma_unmap_single(ca, mapping[0], skb_headlen(skb), DMA_TO_DEVICE);
353
354 return -EIO;
355}
356
357static void ipoib_dma_unmap_tx(struct ib_device *ca,
358 struct ipoib_tx_buf *tx_req)
359{
360 struct sk_buff *skb = tx_req->skb;
361 u64 *mapping = tx_req->mapping;
362 int i;
363 int off;
364
365 if (skb_headlen(skb)) {
366 ib_dma_unmap_single(ca, mapping[0], skb_headlen(skb), DMA_TO_DEVICE);
367 off = 1;
368 } else
369 off = 0;
370
371 for (i = 0; i < skb_shinfo(skb)->nr_frags; ++i) {
372 const skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
373
374 ib_dma_unmap_page(ca, mapping[i + off], skb_frag_size(frag),
375 DMA_TO_DEVICE);
376 }
377}
378
379static void ipoib_ib_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
380{
381 struct ipoib_dev_priv *priv = netdev_priv(dev);
382 unsigned int wr_id = wc->wr_id;
383 struct ipoib_tx_buf *tx_req;
384
385 ipoib_dbg_data(priv, "send completion: id %d, status: %d\n",
386 wr_id, wc->status);
387
388 if (unlikely(wr_id >= ipoib_sendq_size)) {
389 ipoib_warn(priv, "send completion event with wrid %d (> %d)\n",
390 wr_id, ipoib_sendq_size);
391 return;
392 }
393
394 tx_req = &priv->tx_ring[wr_id];
395
396 ipoib_dma_unmap_tx(priv->ca, tx_req);
397
398 ++dev->stats.tx_packets;
399 dev->stats.tx_bytes += tx_req->skb->len;
400
401 dev_kfree_skb_any(tx_req->skb);
402
403 ++priv->tx_tail;
404 if (unlikely(--priv->tx_outstanding == ipoib_sendq_size >> 1) &&
405 netif_queue_stopped(dev) &&
406 test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags))
407 netif_wake_queue(dev);
408
409 if (wc->status != IB_WC_SUCCESS &&
410 wc->status != IB_WC_WR_FLUSH_ERR)
411 ipoib_warn(priv, "failed send event "
412 "(status=%d, wrid=%d vend_err %x)\n",
413 wc->status, wr_id, wc->vendor_err);
414}
415
416static int poll_tx(struct ipoib_dev_priv *priv)
417{
418 int n, i;
419
420 n = ib_poll_cq(priv->send_cq, MAX_SEND_CQE, priv->send_wc);
421 for (i = 0; i < n; ++i)
422 ipoib_ib_handle_tx_wc(priv->dev, priv->send_wc + i);
423
424 return n == MAX_SEND_CQE;
425}
426
427int ipoib_poll(struct napi_struct *napi, int budget)
428{
429 struct ipoib_dev_priv *priv = container_of(napi, struct ipoib_dev_priv, napi);
430 struct net_device *dev = priv->dev;
431 int done;
432 int t;
433 int n, i;
434
435 done = 0;
436
437poll_more:
438 while (done < budget) {
439 int max = (budget - done);
440
441 t = min(IPOIB_NUM_WC, max);
442 n = ib_poll_cq(priv->recv_cq, t, priv->ibwc);
443
444 for (i = 0; i < n; i++) {
445 struct ib_wc *wc = priv->ibwc + i;
446
447 if (wc->wr_id & IPOIB_OP_RECV) {
448 ++done;
449 if (wc->wr_id & IPOIB_OP_CM)
450 ipoib_cm_handle_rx_wc(dev, wc);
451 else
452 ipoib_ib_handle_rx_wc(dev, wc);
453 } else
454 ipoib_cm_handle_tx_wc(priv->dev, wc);
455 }
456
457 if (n != t)
458 break;
459 }
460
461 if (done < budget) {
462 napi_complete(napi);
463 if (unlikely(ib_req_notify_cq(priv->recv_cq,
464 IB_CQ_NEXT_COMP |
465 IB_CQ_REPORT_MISSED_EVENTS)) &&
466 napi_reschedule(napi))
467 goto poll_more;
468 }
469
470 return done;
471}
472
473void ipoib_ib_completion(struct ib_cq *cq, void *dev_ptr)
474{
475 struct net_device *dev = dev_ptr;
476 struct ipoib_dev_priv *priv = netdev_priv(dev);
477
478 napi_schedule(&priv->napi);
479}
480
481static void drain_tx_cq(struct net_device *dev)
482{
483 struct ipoib_dev_priv *priv = netdev_priv(dev);
484
485 netif_tx_lock(dev);
486 while (poll_tx(priv))
487 ;
488
489 if (netif_queue_stopped(dev))
490 mod_timer(&priv->poll_timer, jiffies + 1);
491
492 netif_tx_unlock(dev);
493}
494
495void ipoib_send_comp_handler(struct ib_cq *cq, void *dev_ptr)
496{
497 struct ipoib_dev_priv *priv = netdev_priv(dev_ptr);
498
499 mod_timer(&priv->poll_timer, jiffies);
500}
501
502static inline int post_send(struct ipoib_dev_priv *priv,
503 unsigned int wr_id,
504 struct ib_ah *address, u32 qpn,
505 struct ipoib_tx_buf *tx_req,
506 void *head, int hlen)
507{
508 struct ib_send_wr *bad_wr;
509 int i, off;
510 struct sk_buff *skb = tx_req->skb;
511 skb_frag_t *frags = skb_shinfo(skb)->frags;
512 int nr_frags = skb_shinfo(skb)->nr_frags;
513 u64 *mapping = tx_req->mapping;
514
515 if (skb_headlen(skb)) {
516 priv->tx_sge[0].addr = mapping[0];
517 priv->tx_sge[0].length = skb_headlen(skb);
518 off = 1;
519 } else
520 off = 0;
521
522 for (i = 0; i < nr_frags; ++i) {
523 priv->tx_sge[i + off].addr = mapping[i + off];
524 priv->tx_sge[i + off].length = skb_frag_size(&frags[i]);
525 }
526 priv->tx_wr.num_sge = nr_frags + off;
527 priv->tx_wr.wr_id = wr_id;
528 priv->tx_wr.wr.ud.remote_qpn = qpn;
529 priv->tx_wr.wr.ud.ah = address;
530
531 if (head) {
532 priv->tx_wr.wr.ud.mss = skb_shinfo(skb)->gso_size;
533 priv->tx_wr.wr.ud.header = head;
534 priv->tx_wr.wr.ud.hlen = hlen;
535 priv->tx_wr.opcode = IB_WR_LSO;
536 } else
537 priv->tx_wr.opcode = IB_WR_SEND;
538
539 return ib_post_send(priv->qp, &priv->tx_wr, &bad_wr);
540}
541
542void ipoib_send(struct net_device *dev, struct sk_buff *skb,
543 struct ipoib_ah *address, u32 qpn)
544{
545 struct ipoib_dev_priv *priv = netdev_priv(dev);
546 struct ipoib_tx_buf *tx_req;
547 int hlen, rc;
548 void *phead;
549
550 if (skb_is_gso(skb)) {
551 hlen = skb_transport_offset(skb) + tcp_hdrlen(skb);
552 phead = skb->data;
553 if (unlikely(!skb_pull(skb, hlen))) {
554 ipoib_warn(priv, "linear data too small\n");
555 ++dev->stats.tx_dropped;
556 ++dev->stats.tx_errors;
557 dev_kfree_skb_any(skb);
558 return;
559 }
560 } else {
561 if (unlikely(skb->len > priv->mcast_mtu + IPOIB_ENCAP_LEN)) {
562 ipoib_warn(priv, "packet len %d (> %d) too long to send, dropping\n",
563 skb->len, priv->mcast_mtu + IPOIB_ENCAP_LEN);
564 ++dev->stats.tx_dropped;
565 ++dev->stats.tx_errors;
566 ipoib_cm_skb_too_long(dev, skb, priv->mcast_mtu);
567 return;
568 }
569 phead = NULL;
570 hlen = 0;
571 }
572
573 ipoib_dbg_data(priv, "sending packet, length=%d address=%p qpn=0x%06x\n",
574 skb->len, address, qpn);
575
576
577
578
579
580
581
582
583 tx_req = &priv->tx_ring[priv->tx_head & (ipoib_sendq_size - 1)];
584 tx_req->skb = skb;
585 if (unlikely(ipoib_dma_map_tx(priv->ca, tx_req))) {
586 ++dev->stats.tx_errors;
587 dev_kfree_skb_any(skb);
588 return;
589 }
590
591 if (skb->ip_summed == CHECKSUM_PARTIAL)
592 priv->tx_wr.send_flags |= IB_SEND_IP_CSUM;
593 else
594 priv->tx_wr.send_flags &= ~IB_SEND_IP_CSUM;
595
596 if (++priv->tx_outstanding == ipoib_sendq_size) {
597 ipoib_dbg(priv, "TX ring full, stopping kernel net queue\n");
598 if (ib_req_notify_cq(priv->send_cq, IB_CQ_NEXT_COMP))
599 ipoib_warn(priv, "request notify on send CQ failed\n");
600 netif_stop_queue(dev);
601 }
602
603 skb_orphan(skb);
604 skb_dst_drop(skb);
605
606 rc = post_send(priv, priv->tx_head & (ipoib_sendq_size - 1),
607 address->ah, qpn, tx_req, phead, hlen);
608 if (unlikely(rc)) {
609 ipoib_warn(priv, "post_send failed, error %d\n", rc);
610 ++dev->stats.tx_errors;
611 --priv->tx_outstanding;
612 ipoib_dma_unmap_tx(priv->ca, tx_req);
613 dev_kfree_skb_any(skb);
614 if (netif_queue_stopped(dev))
615 netif_wake_queue(dev);
616 } else {
617 dev->trans_start = jiffies;
618
619 address->last_send = priv->tx_head;
620 ++priv->tx_head;
621 }
622
623 if (unlikely(priv->tx_outstanding > MAX_SEND_CQE))
624 while (poll_tx(priv))
625 ;
626}
627
628static void __ipoib_reap_ah(struct net_device *dev)
629{
630 struct ipoib_dev_priv *priv = netdev_priv(dev);
631 struct ipoib_ah *ah, *tah;
632 LIST_HEAD(remove_list);
633 unsigned long flags;
634
635 netif_tx_lock_bh(dev);
636 spin_lock_irqsave(&priv->lock, flags);
637
638 list_for_each_entry_safe(ah, tah, &priv->dead_ahs, list)
639 if ((int) priv->tx_tail - (int) ah->last_send >= 0) {
640 list_del(&ah->list);
641 ib_destroy_ah(ah->ah);
642 kfree(ah);
643 }
644
645 spin_unlock_irqrestore(&priv->lock, flags);
646 netif_tx_unlock_bh(dev);
647}
648
649void ipoib_reap_ah(struct work_struct *work)
650{
651 struct ipoib_dev_priv *priv =
652 container_of(work, struct ipoib_dev_priv, ah_reap_task.work);
653 struct net_device *dev = priv->dev;
654
655 __ipoib_reap_ah(dev);
656
657 if (!test_bit(IPOIB_STOP_REAPER, &priv->flags))
658 queue_delayed_work(ipoib_workqueue, &priv->ah_reap_task,
659 round_jiffies_relative(HZ));
660}
661
662static void ipoib_ib_tx_timer_func(unsigned long ctx)
663{
664 drain_tx_cq((struct net_device *)ctx);
665}
666
667int ipoib_ib_dev_open(struct net_device *dev)
668{
669 struct ipoib_dev_priv *priv = netdev_priv(dev);
670 int ret;
671
672 if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &priv->pkey_index)) {
673 ipoib_warn(priv, "P_Key 0x%04x not found\n", priv->pkey);
674 clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
675 return -1;
676 }
677 set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
678
679 ret = ipoib_init_qp(dev);
680 if (ret) {
681 ipoib_warn(priv, "ipoib_init_qp returned %d\n", ret);
682 return -1;
683 }
684
685 ret = ipoib_ib_post_receives(dev);
686 if (ret) {
687 ipoib_warn(priv, "ipoib_ib_post_receives returned %d\n", ret);
688 goto dev_stop;
689 }
690
691 ret = ipoib_cm_dev_open(dev);
692 if (ret) {
693 ipoib_warn(priv, "ipoib_cm_dev_open returned %d\n", ret);
694 goto dev_stop;
695 }
696
697 clear_bit(IPOIB_STOP_REAPER, &priv->flags);
698 queue_delayed_work(ipoib_workqueue, &priv->ah_reap_task,
699 round_jiffies_relative(HZ));
700
701 if (!test_and_set_bit(IPOIB_FLAG_INITIALIZED, &priv->flags))
702 napi_enable(&priv->napi);
703
704 return 0;
705dev_stop:
706 if (!test_and_set_bit(IPOIB_FLAG_INITIALIZED, &priv->flags))
707 napi_enable(&priv->napi);
708 ipoib_ib_dev_stop(dev, 1);
709 return -1;
710}
711
712static void ipoib_pkey_dev_check_presence(struct net_device *dev)
713{
714 struct ipoib_dev_priv *priv = netdev_priv(dev);
715 u16 pkey_index = 0;
716
717 if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &pkey_index))
718 clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
719 else
720 set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
721}
722
723int ipoib_ib_dev_up(struct net_device *dev)
724{
725 struct ipoib_dev_priv *priv = netdev_priv(dev);
726
727 ipoib_pkey_dev_check_presence(dev);
728
729 if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) {
730 ipoib_dbg(priv, "PKEY is not assigned.\n");
731 return 0;
732 }
733
734 set_bit(IPOIB_FLAG_OPER_UP, &priv->flags);
735
736 return ipoib_mcast_start_thread(dev);
737}
738
739int ipoib_ib_dev_down(struct net_device *dev, int flush)
740{
741 struct ipoib_dev_priv *priv = netdev_priv(dev);
742
743 ipoib_dbg(priv, "downing ib_dev\n");
744
745 clear_bit(IPOIB_FLAG_OPER_UP, &priv->flags);
746 netif_carrier_off(dev);
747
748
749 if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) {
750 mutex_lock(&pkey_mutex);
751 set_bit(IPOIB_PKEY_STOP, &priv->flags);
752 cancel_delayed_work_sync(&priv->pkey_poll_task);
753 mutex_unlock(&pkey_mutex);
754 }
755
756 ipoib_mcast_stop_thread(dev, flush);
757 ipoib_mcast_dev_flush(dev);
758
759 ipoib_flush_paths(dev);
760
761 return 0;
762}
763
764static int recvs_pending(struct net_device *dev)
765{
766 struct ipoib_dev_priv *priv = netdev_priv(dev);
767 int pending = 0;
768 int i;
769
770 for (i = 0; i < ipoib_recvq_size; ++i)
771 if (priv->rx_ring[i].skb)
772 ++pending;
773
774 return pending;
775}
776
777void ipoib_drain_cq(struct net_device *dev)
778{
779 struct ipoib_dev_priv *priv = netdev_priv(dev);
780 int i, n;
781
782
783
784
785
786
787 local_bh_disable();
788
789 do {
790 n = ib_poll_cq(priv->recv_cq, IPOIB_NUM_WC, priv->ibwc);
791 for (i = 0; i < n; ++i) {
792
793
794
795
796
797 if (priv->ibwc[i].status == IB_WC_SUCCESS)
798 priv->ibwc[i].status = IB_WC_WR_FLUSH_ERR;
799
800 if (priv->ibwc[i].wr_id & IPOIB_OP_RECV) {
801 if (priv->ibwc[i].wr_id & IPOIB_OP_CM)
802 ipoib_cm_handle_rx_wc(dev, priv->ibwc + i);
803 else
804 ipoib_ib_handle_rx_wc(dev, priv->ibwc + i);
805 } else
806 ipoib_cm_handle_tx_wc(dev, priv->ibwc + i);
807 }
808 } while (n == IPOIB_NUM_WC);
809
810 while (poll_tx(priv))
811 ;
812
813 local_bh_enable();
814}
815
816int ipoib_ib_dev_stop(struct net_device *dev, int flush)
817{
818 struct ipoib_dev_priv *priv = netdev_priv(dev);
819 struct ib_qp_attr qp_attr;
820 unsigned long begin;
821 struct ipoib_tx_buf *tx_req;
822 int i;
823
824 if (test_and_clear_bit(IPOIB_FLAG_INITIALIZED, &priv->flags))
825 napi_disable(&priv->napi);
826
827 ipoib_cm_dev_stop(dev);
828
829
830
831
832
833 qp_attr.qp_state = IB_QPS_ERR;
834 if (ib_modify_qp(priv->qp, &qp_attr, IB_QP_STATE))
835 ipoib_warn(priv, "Failed to modify QP to ERROR state\n");
836
837
838 begin = jiffies;
839
840 while (priv->tx_head != priv->tx_tail || recvs_pending(dev)) {
841 if (time_after(jiffies, begin + 5 * HZ)) {
842 ipoib_warn(priv, "timing out; %d sends %d receives not completed\n",
843 priv->tx_head - priv->tx_tail, recvs_pending(dev));
844
845
846
847
848
849 while ((int) priv->tx_tail - (int) priv->tx_head < 0) {
850 tx_req = &priv->tx_ring[priv->tx_tail &
851 (ipoib_sendq_size - 1)];
852 ipoib_dma_unmap_tx(priv->ca, tx_req);
853 dev_kfree_skb_any(tx_req->skb);
854 ++priv->tx_tail;
855 --priv->tx_outstanding;
856 }
857
858 for (i = 0; i < ipoib_recvq_size; ++i) {
859 struct ipoib_rx_buf *rx_req;
860
861 rx_req = &priv->rx_ring[i];
862 if (!rx_req->skb)
863 continue;
864 ipoib_ud_dma_unmap_rx(priv,
865 priv->rx_ring[i].mapping);
866 dev_kfree_skb_any(rx_req->skb);
867 rx_req->skb = NULL;
868 }
869
870 goto timeout;
871 }
872
873 ipoib_drain_cq(dev);
874
875 msleep(1);
876 }
877
878 ipoib_dbg(priv, "All sends and receives done.\n");
879
880timeout:
881 del_timer_sync(&priv->poll_timer);
882 qp_attr.qp_state = IB_QPS_RESET;
883 if (ib_modify_qp(priv->qp, &qp_attr, IB_QP_STATE))
884 ipoib_warn(priv, "Failed to modify QP to RESET state\n");
885
886
887 set_bit(IPOIB_STOP_REAPER, &priv->flags);
888 cancel_delayed_work(&priv->ah_reap_task);
889 if (flush)
890 flush_workqueue(ipoib_workqueue);
891
892 begin = jiffies;
893
894 while (!list_empty(&priv->dead_ahs)) {
895 __ipoib_reap_ah(dev);
896
897 if (time_after(jiffies, begin + HZ)) {
898 ipoib_warn(priv, "timing out; will leak address handles\n");
899 break;
900 }
901
902 msleep(1);
903 }
904
905 ib_req_notify_cq(priv->recv_cq, IB_CQ_NEXT_COMP);
906
907 return 0;
908}
909
910int ipoib_ib_dev_init(struct net_device *dev, struct ib_device *ca, int port)
911{
912 struct ipoib_dev_priv *priv = netdev_priv(dev);
913
914 priv->ca = ca;
915 priv->port = port;
916 priv->qp = NULL;
917
918 if (ipoib_transport_dev_init(dev, ca)) {
919 printk(KERN_WARNING "%s: ipoib_transport_dev_init failed\n", ca->name);
920 return -ENODEV;
921 }
922
923 setup_timer(&priv->poll_timer, ipoib_ib_tx_timer_func,
924 (unsigned long) dev);
925
926 if (dev->flags & IFF_UP) {
927 if (ipoib_ib_dev_open(dev)) {
928 ipoib_transport_dev_cleanup(dev);
929 return -ENODEV;
930 }
931 }
932
933 return 0;
934}
935
936
937
938
939
940static inline int update_parent_pkey(struct ipoib_dev_priv *priv)
941{
942 int result;
943 u16 prev_pkey;
944
945 prev_pkey = priv->pkey;
946 result = ib_query_pkey(priv->ca, priv->port, 0, &priv->pkey);
947 if (result) {
948 ipoib_warn(priv, "ib_query_pkey port %d failed (ret = %d)\n",
949 priv->port, result);
950 return result;
951 }
952
953 priv->pkey |= 0x8000;
954
955 if (prev_pkey != priv->pkey) {
956 ipoib_dbg(priv, "pkey changed from 0x%x to 0x%x\n",
957 prev_pkey, priv->pkey);
958
959
960
961
962 priv->dev->broadcast[8] = priv->pkey >> 8;
963 priv->dev->broadcast[9] = priv->pkey & 0xff;
964 return 0;
965 }
966
967 return 1;
968}
969
970static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv,
971 enum ipoib_flush_level level)
972{
973 struct ipoib_dev_priv *cpriv;
974 struct net_device *dev = priv->dev;
975 u16 new_index;
976 int result;
977
978 down_read(&priv->vlan_rwsem);
979
980
981
982
983
984 list_for_each_entry(cpriv, &priv->child_intfs, list)
985 __ipoib_ib_dev_flush(cpriv, level);
986
987 up_read(&priv->vlan_rwsem);
988
989 if (!test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags)) {
990
991 if (level == IPOIB_FLUSH_HEAVY &&
992 !test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags))
993 update_parent_pkey(priv);
994 ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_INITIALIZED not set.\n");
995 return;
996 }
997
998 if (!test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags)) {
999 ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_ADMIN_UP not set.\n");
1000 return;
1001 }
1002
1003 if (level == IPOIB_FLUSH_HEAVY) {
1004
1005
1006
1007 if (test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags)) {
1008 if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &new_index)) {
1009 clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
1010 ipoib_ib_dev_down(dev, 0);
1011 ipoib_ib_dev_stop(dev, 0);
1012 if (ipoib_pkey_dev_delay_open(dev))
1013 return;
1014 }
1015
1016 if (test_and_set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags) &&
1017 new_index == priv->pkey_index) {
1018 ipoib_dbg(priv, "Not flushing - P_Key index not changed.\n");
1019 return;
1020 }
1021 priv->pkey_index = new_index;
1022 } else {
1023 result = update_parent_pkey(priv);
1024
1025 if (result) {
1026 ipoib_dbg(priv, "Not flushing - P_Key value not changed.\n");
1027 return;
1028 }
1029 }
1030 }
1031
1032 if (level == IPOIB_FLUSH_LIGHT) {
1033 ipoib_mark_paths_invalid(dev);
1034 ipoib_mcast_dev_flush(dev);
1035 }
1036
1037 if (level >= IPOIB_FLUSH_NORMAL)
1038 ipoib_ib_dev_down(dev, 0);
1039
1040 if (level == IPOIB_FLUSH_HEAVY) {
1041 ipoib_ib_dev_stop(dev, 0);
1042 ipoib_ib_dev_open(dev);
1043 }
1044
1045
1046
1047
1048
1049 if (test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags)) {
1050 if (level >= IPOIB_FLUSH_NORMAL)
1051 ipoib_ib_dev_up(dev);
1052 ipoib_mcast_restart_task(&priv->restart_task);
1053 }
1054}
1055
1056void ipoib_ib_dev_flush_light(struct work_struct *work)
1057{
1058 struct ipoib_dev_priv *priv =
1059 container_of(work, struct ipoib_dev_priv, flush_light);
1060
1061 __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_LIGHT);
1062}
1063
1064void ipoib_ib_dev_flush_normal(struct work_struct *work)
1065{
1066 struct ipoib_dev_priv *priv =
1067 container_of(work, struct ipoib_dev_priv, flush_normal);
1068
1069 __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_NORMAL);
1070}
1071
1072void ipoib_ib_dev_flush_heavy(struct work_struct *work)
1073{
1074 struct ipoib_dev_priv *priv =
1075 container_of(work, struct ipoib_dev_priv, flush_heavy);
1076
1077 __ipoib_ib_dev_flush(priv, IPOIB_FLUSH_HEAVY);
1078}
1079
1080void ipoib_ib_dev_cleanup(struct net_device *dev)
1081{
1082 struct ipoib_dev_priv *priv = netdev_priv(dev);
1083
1084 ipoib_dbg(priv, "cleaning up ib_dev\n");
1085
1086
1087
1088
1089 ipoib_flush_paths(dev);
1090
1091 ipoib_mcast_stop_thread(dev, 1);
1092 ipoib_mcast_dev_flush(dev);
1093
1094 ipoib_transport_dev_cleanup(dev);
1095}
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107void ipoib_pkey_poll(struct work_struct *work)
1108{
1109 struct ipoib_dev_priv *priv =
1110 container_of(work, struct ipoib_dev_priv, pkey_poll_task.work);
1111 struct net_device *dev = priv->dev;
1112
1113 ipoib_pkey_dev_check_presence(dev);
1114
1115 if (test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags))
1116 ipoib_open(dev);
1117 else {
1118 mutex_lock(&pkey_mutex);
1119 if (!test_bit(IPOIB_PKEY_STOP, &priv->flags))
1120 queue_delayed_work(ipoib_workqueue,
1121 &priv->pkey_poll_task,
1122 HZ);
1123 mutex_unlock(&pkey_mutex);
1124 }
1125}
1126
1127int ipoib_pkey_dev_delay_open(struct net_device *dev)
1128{
1129 struct ipoib_dev_priv *priv = netdev_priv(dev);
1130
1131
1132
1133 ipoib_pkey_dev_check_presence(dev);
1134
1135
1136 if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) {
1137 mutex_lock(&pkey_mutex);
1138 clear_bit(IPOIB_PKEY_STOP, &priv->flags);
1139 queue_delayed_work(ipoib_workqueue,
1140 &priv->pkey_poll_task,
1141 HZ);
1142 mutex_unlock(&pkey_mutex);
1143 return 1;
1144 }
1145
1146 return 0;
1147}
1148