blob: 52182ff7519edcde8b7dee746b72282fe6eb832f [file] [log] [blame]
Andy Groverfcd8b7c2009-02-24 15:30:36 +00001/*
2 * Copyright (c) 2006 Oracle. All rights reserved.
3 *
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
9 *
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
12 * conditions are met:
13 *
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
16 * disclaimer.
17 *
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
22 *
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30 * SOFTWARE.
31 *
32 */
33#include <linux/kernel.h>
34#include <linux/in.h>
35#include <linux/device.h>
36#include <linux/dmapool.h>
37
38#include "rds.h"
39#include "rdma.h"
40#include "iw.h"
41
42static void rds_iw_send_rdma_complete(struct rds_message *rm,
43 int wc_status)
44{
45 int notify_status;
46
47 switch (wc_status) {
48 case IB_WC_WR_FLUSH_ERR:
49 return;
50
51 case IB_WC_SUCCESS:
52 notify_status = RDS_RDMA_SUCCESS;
53 break;
54
55 case IB_WC_REM_ACCESS_ERR:
56 notify_status = RDS_RDMA_REMOTE_ERROR;
57 break;
58
59 default:
60 notify_status = RDS_RDMA_OTHER_ERROR;
61 break;
62 }
63 rds_rdma_send_complete(rm, notify_status);
64}
65
66static void rds_iw_send_unmap_rdma(struct rds_iw_connection *ic,
67 struct rds_rdma_op *op)
68{
69 if (op->r_mapped) {
70 ib_dma_unmap_sg(ic->i_cm_id->device,
71 op->r_sg, op->r_nents,
72 op->r_write ? DMA_TO_DEVICE : DMA_FROM_DEVICE);
73 op->r_mapped = 0;
74 }
75}
76
77static void rds_iw_send_unmap_rm(struct rds_iw_connection *ic,
78 struct rds_iw_send_work *send,
79 int wc_status)
80{
81 struct rds_message *rm = send->s_rm;
82
83 rdsdebug("ic %p send %p rm %p\n", ic, send, rm);
84
85 ib_dma_unmap_sg(ic->i_cm_id->device,
86 rm->m_sg, rm->m_nents,
87 DMA_TO_DEVICE);
88
89 if (rm->m_rdma_op != NULL) {
90 rds_iw_send_unmap_rdma(ic, rm->m_rdma_op);
91
92 /* If the user asked for a completion notification on this
93 * message, we can implement three different semantics:
94 * 1. Notify when we received the ACK on the RDS message
95 * that was queued with the RDMA. This provides reliable
96 * notification of RDMA status at the expense of a one-way
97 * packet delay.
98 * 2. Notify when the IB stack gives us the completion event for
99 * the RDMA operation.
100 * 3. Notify when the IB stack gives us the completion event for
101 * the accompanying RDS messages.
102 * Here, we implement approach #3. To implement approach #2,
103 * call rds_rdma_send_complete from the cq_handler. To implement #1,
104 * don't call rds_rdma_send_complete at all, and fall back to the notify
105 * handling in the ACK processing code.
106 *
107 * Note: There's no need to explicitly sync any RDMA buffers using
108 * ib_dma_sync_sg_for_cpu - the completion for the RDMA
109 * operation itself unmapped the RDMA buffers, which takes care
110 * of synching.
111 */
112 rds_iw_send_rdma_complete(rm, wc_status);
113
114 if (rm->m_rdma_op->r_write)
115 rds_stats_add(s_send_rdma_bytes, rm->m_rdma_op->r_bytes);
116 else
117 rds_stats_add(s_recv_rdma_bytes, rm->m_rdma_op->r_bytes);
118 }
119
120 /* If anyone waited for this message to get flushed out, wake
121 * them up now */
122 rds_message_unmapped(rm);
123
124 rds_message_put(rm);
125 send->s_rm = NULL;
126}
127
128void rds_iw_send_init_ring(struct rds_iw_connection *ic)
129{
130 struct rds_iw_send_work *send;
131 u32 i;
132
133 for (i = 0, send = ic->i_sends; i < ic->i_send_ring.w_nr; i++, send++) {
134 struct ib_sge *sge;
135
136 send->s_rm = NULL;
137 send->s_op = NULL;
138 send->s_mapping = NULL;
139
140 send->s_wr.next = NULL;
141 send->s_wr.wr_id = i;
142 send->s_wr.sg_list = send->s_sge;
143 send->s_wr.num_sge = 1;
144 send->s_wr.opcode = IB_WR_SEND;
145 send->s_wr.send_flags = 0;
146 send->s_wr.ex.imm_data = 0;
147
148 sge = rds_iw_data_sge(ic, send->s_sge);
149 sge->lkey = 0;
150
151 sge = rds_iw_header_sge(ic, send->s_sge);
152 sge->addr = ic->i_send_hdrs_dma + (i * sizeof(struct rds_header));
153 sge->length = sizeof(struct rds_header);
154 sge->lkey = 0;
155
156 send->s_mr = ib_alloc_fast_reg_mr(ic->i_pd, fastreg_message_size);
157 if (IS_ERR(send->s_mr)) {
158 printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_mr failed\n");
159 break;
160 }
161
162 send->s_page_list = ib_alloc_fast_reg_page_list(
163 ic->i_cm_id->device, fastreg_message_size);
164 if (IS_ERR(send->s_page_list)) {
165 printk(KERN_WARNING "RDS/IW: ib_alloc_fast_reg_page_list failed\n");
166 break;
167 }
168 }
169}
170
171void rds_iw_send_clear_ring(struct rds_iw_connection *ic)
172{
173 struct rds_iw_send_work *send;
174 u32 i;
175
176 for (i = 0, send = ic->i_sends; i < ic->i_send_ring.w_nr; i++, send++) {
177 BUG_ON(!send->s_mr);
178 ib_dereg_mr(send->s_mr);
179 BUG_ON(!send->s_page_list);
180 ib_free_fast_reg_page_list(send->s_page_list);
181 if (send->s_wr.opcode == 0xdead)
182 continue;
183 if (send->s_rm)
184 rds_iw_send_unmap_rm(ic, send, IB_WC_WR_FLUSH_ERR);
185 if (send->s_op)
186 rds_iw_send_unmap_rdma(ic, send->s_op);
187 }
188}
189
190/*
191 * The _oldest/_free ring operations here race cleanly with the alloc/unalloc
192 * operations performed in the send path. As the sender allocs and potentially
193 * unallocs the next free entry in the ring it doesn't alter which is
194 * the next to be freed, which is what this is concerned with.
195 */
196void rds_iw_send_cq_comp_handler(struct ib_cq *cq, void *context)
197{
198 struct rds_connection *conn = context;
199 struct rds_iw_connection *ic = conn->c_transport_data;
200 struct ib_wc wc;
201 struct rds_iw_send_work *send;
202 u32 completed;
203 u32 oldest;
204 u32 i;
205 int ret;
206
207 rdsdebug("cq %p conn %p\n", cq, conn);
208 rds_iw_stats_inc(s_iw_tx_cq_call);
209 ret = ib_req_notify_cq(cq, IB_CQ_NEXT_COMP);
210 if (ret)
211 rdsdebug("ib_req_notify_cq send failed: %d\n", ret);
212
213 while (ib_poll_cq(cq, 1, &wc) > 0) {
214 rdsdebug("wc wr_id 0x%llx status %u byte_len %u imm_data %u\n",
215 (unsigned long long)wc.wr_id, wc.status, wc.byte_len,
216 be32_to_cpu(wc.ex.imm_data));
217 rds_iw_stats_inc(s_iw_tx_cq_event);
218
219 if (wc.status != IB_WC_SUCCESS) {
220 printk(KERN_ERR "WC Error: status = %d opcode = %d\n", wc.status, wc.opcode);
221 break;
222 }
223
224 if (wc.opcode == IB_WC_LOCAL_INV && wc.wr_id == RDS_IW_LOCAL_INV_WR_ID) {
225 ic->i_fastreg_posted = 0;
226 continue;
227 }
228
229 if (wc.opcode == IB_WC_FAST_REG_MR && wc.wr_id == RDS_IW_FAST_REG_WR_ID) {
230 ic->i_fastreg_posted = 1;
231 continue;
232 }
233
234 if (wc.wr_id == RDS_IW_ACK_WR_ID) {
235 if (ic->i_ack_queued + HZ/2 < jiffies)
236 rds_iw_stats_inc(s_iw_tx_stalled);
237 rds_iw_ack_send_complete(ic);
238 continue;
239 }
240
241 oldest = rds_iw_ring_oldest(&ic->i_send_ring);
242
243 completed = rds_iw_ring_completed(&ic->i_send_ring, wc.wr_id, oldest);
244
245 for (i = 0; i < completed; i++) {
246 send = &ic->i_sends[oldest];
247
248 /* In the error case, wc.opcode sometimes contains garbage */
249 switch (send->s_wr.opcode) {
250 case IB_WR_SEND:
251 if (send->s_rm)
252 rds_iw_send_unmap_rm(ic, send, wc.status);
253 break;
254 case IB_WR_FAST_REG_MR:
255 case IB_WR_RDMA_WRITE:
256 case IB_WR_RDMA_READ:
257 case IB_WR_RDMA_READ_WITH_INV:
258 /* Nothing to be done - the SG list will be unmapped
259 * when the SEND completes. */
260 break;
261 default:
262 if (printk_ratelimit())
263 printk(KERN_NOTICE
264 "RDS/IW: %s: unexpected opcode 0x%x in WR!\n",
265 __func__, send->s_wr.opcode);
266 break;
267 }
268
269 send->s_wr.opcode = 0xdead;
270 send->s_wr.num_sge = 1;
271 if (send->s_queued + HZ/2 < jiffies)
272 rds_iw_stats_inc(s_iw_tx_stalled);
273
274 /* If a RDMA operation produced an error, signal this right
275 * away. If we don't, the subsequent SEND that goes with this
276 * RDMA will be canceled with ERR_WFLUSH, and the application
277 * never learn that the RDMA failed. */
278 if (unlikely(wc.status == IB_WC_REM_ACCESS_ERR && send->s_op)) {
279 struct rds_message *rm;
280
281 rm = rds_send_get_message(conn, send->s_op);
282 if (rm)
283 rds_iw_send_rdma_complete(rm, wc.status);
284 }
285
286 oldest = (oldest + 1) % ic->i_send_ring.w_nr;
287 }
288
289 rds_iw_ring_free(&ic->i_send_ring, completed);
290
Joe Perchesf64f9e72009-11-29 16:55:45 -0800291 if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags) ||
292 test_bit(0, &conn->c_map_queued))
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000293 queue_delayed_work(rds_wq, &conn->c_send_w, 0);
294
295 /* We expect errors as the qp is drained during shutdown */
296 if (wc.status != IB_WC_SUCCESS && rds_conn_up(conn)) {
297 rds_iw_conn_error(conn,
298 "send completion on %pI4 "
299 "had status %u, disconnecting and reconnecting\n",
300 &conn->c_faddr, wc.status);
301 }
302 }
303}
304
305/*
306 * This is the main function for allocating credits when sending
307 * messages.
308 *
309 * Conceptually, we have two counters:
310 * - send credits: this tells us how many WRs we're allowed
311 * to submit without overruning the reciever's queue. For
312 * each SEND WR we post, we decrement this by one.
313 *
314 * - posted credits: this tells us how many WRs we recently
315 * posted to the receive queue. This value is transferred
316 * to the peer as a "credit update" in a RDS header field.
317 * Every time we transmit credits to the peer, we subtract
318 * the amount of transferred credits from this counter.
319 *
320 * It is essential that we avoid situations where both sides have
321 * exhausted their send credits, and are unable to send new credits
322 * to the peer. We achieve this by requiring that we send at least
323 * one credit update to the peer before exhausting our credits.
324 * When new credits arrive, we subtract one credit that is withheld
325 * until we've posted new buffers and are ready to transmit these
326 * credits (see rds_iw_send_add_credits below).
327 *
328 * The RDS send code is essentially single-threaded; rds_send_xmit
329 * grabs c_send_lock to ensure exclusive access to the send ring.
330 * However, the ACK sending code is independent and can race with
331 * message SENDs.
332 *
333 * In the send path, we need to update the counters for send credits
334 * and the counter of posted buffers atomically - when we use the
335 * last available credit, we cannot allow another thread to race us
336 * and grab the posted credits counter. Hence, we have to use a
337 * spinlock to protect the credit counter, or use atomics.
338 *
339 * Spinlocks shared between the send and the receive path are bad,
340 * because they create unnecessary delays. An early implementation
341 * using a spinlock showed a 5% degradation in throughput at some
342 * loads.
343 *
344 * This implementation avoids spinlocks completely, putting both
345 * counters into a single atomic, and updating that atomic using
346 * atomic_add (in the receive path, when receiving fresh credits),
347 * and using atomic_cmpxchg when updating the two counters.
348 */
349int rds_iw_send_grab_credits(struct rds_iw_connection *ic,
Steve Wise7b70d032009-04-09 14:09:39 +0000350 u32 wanted, u32 *adv_credits, int need_posted, int max_posted)
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000351{
352 unsigned int avail, posted, got = 0, advertise;
353 long oldval, newval;
354
355 *adv_credits = 0;
356 if (!ic->i_flowctl)
357 return wanted;
358
359try_again:
360 advertise = 0;
361 oldval = newval = atomic_read(&ic->i_credits);
362 posted = IB_GET_POST_CREDITS(oldval);
363 avail = IB_GET_SEND_CREDITS(oldval);
364
365 rdsdebug("rds_iw_send_grab_credits(%u): credits=%u posted=%u\n",
366 wanted, avail, posted);
367
368 /* The last credit must be used to send a credit update. */
369 if (avail && !posted)
370 avail--;
371
372 if (avail < wanted) {
373 struct rds_connection *conn = ic->i_cm_id->context;
374
375 /* Oops, there aren't that many credits left! */
376 set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
377 got = avail;
378 } else {
379 /* Sometimes you get what you want, lalala. */
380 got = wanted;
381 }
382 newval -= IB_SET_SEND_CREDITS(got);
383
384 /*
385 * If need_posted is non-zero, then the caller wants
386 * the posted regardless of whether any send credits are
387 * available.
388 */
389 if (posted && (got || need_posted)) {
Steve Wise7b70d032009-04-09 14:09:39 +0000390 advertise = min_t(unsigned int, posted, max_posted);
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000391 newval -= IB_SET_POST_CREDITS(advertise);
392 }
393
394 /* Finally bill everything */
395 if (atomic_cmpxchg(&ic->i_credits, oldval, newval) != oldval)
396 goto try_again;
397
398 *adv_credits = advertise;
399 return got;
400}
401
402void rds_iw_send_add_credits(struct rds_connection *conn, unsigned int credits)
403{
404 struct rds_iw_connection *ic = conn->c_transport_data;
405
406 if (credits == 0)
407 return;
408
409 rdsdebug("rds_iw_send_add_credits(%u): current=%u%s\n",
410 credits,
411 IB_GET_SEND_CREDITS(atomic_read(&ic->i_credits)),
412 test_bit(RDS_LL_SEND_FULL, &conn->c_flags) ? ", ll_send_full" : "");
413
414 atomic_add(IB_SET_SEND_CREDITS(credits), &ic->i_credits);
415 if (test_and_clear_bit(RDS_LL_SEND_FULL, &conn->c_flags))
416 queue_delayed_work(rds_wq, &conn->c_send_w, 0);
417
418 WARN_ON(IB_GET_SEND_CREDITS(credits) >= 16384);
419
420 rds_iw_stats_inc(s_iw_rx_credit_updates);
421}
422
423void rds_iw_advertise_credits(struct rds_connection *conn, unsigned int posted)
424{
425 struct rds_iw_connection *ic = conn->c_transport_data;
426
427 if (posted == 0)
428 return;
429
430 atomic_add(IB_SET_POST_CREDITS(posted), &ic->i_credits);
431
432 /* Decide whether to send an update to the peer now.
433 * If we would send a credit update for every single buffer we
434 * post, we would end up with an ACK storm (ACK arrives,
435 * consumes buffer, we refill the ring, send ACK to remote
436 * advertising the newly posted buffer... ad inf)
437 *
438 * Performance pretty much depends on how often we send
439 * credit updates - too frequent updates mean lots of ACKs.
440 * Too infrequent updates, and the peer will run out of
441 * credits and has to throttle.
442 * For the time being, 16 seems to be a good compromise.
443 */
444 if (IB_GET_POST_CREDITS(atomic_read(&ic->i_credits)) >= 16)
445 set_bit(IB_ACK_REQUESTED, &ic->i_ack_flags);
446}
447
448static inline void
449rds_iw_xmit_populate_wr(struct rds_iw_connection *ic,
450 struct rds_iw_send_work *send, unsigned int pos,
451 unsigned long buffer, unsigned int length,
452 int send_flags)
453{
454 struct ib_sge *sge;
455
456 WARN_ON(pos != send - ic->i_sends);
457
458 send->s_wr.send_flags = send_flags;
459 send->s_wr.opcode = IB_WR_SEND;
460 send->s_wr.num_sge = 2;
461 send->s_wr.next = NULL;
462 send->s_queued = jiffies;
463 send->s_op = NULL;
464
465 if (length != 0) {
466 sge = rds_iw_data_sge(ic, send->s_sge);
467 sge->addr = buffer;
468 sge->length = length;
469 sge->lkey = rds_iw_local_dma_lkey(ic);
470
471 sge = rds_iw_header_sge(ic, send->s_sge);
472 } else {
473 /* We're sending a packet with no payload. There is only
474 * one SGE */
475 send->s_wr.num_sge = 1;
476 sge = &send->s_sge[0];
477 }
478
479 sge->addr = ic->i_send_hdrs_dma + (pos * sizeof(struct rds_header));
480 sge->length = sizeof(struct rds_header);
481 sge->lkey = rds_iw_local_dma_lkey(ic);
482}
483
484/*
485 * This can be called multiple times for a given message. The first time
486 * we see a message we map its scatterlist into the IB device so that
487 * we can provide that mapped address to the IB scatter gather entries
488 * in the IB work requests. We translate the scatterlist into a series
489 * of work requests that fragment the message. These work requests complete
490 * in order so we pass ownership of the message to the completion handler
491 * once we send the final fragment.
492 *
493 * The RDS core uses the c_send_lock to only enter this function once
494 * per connection. This makes sure that the tx ring alloc/unalloc pairs
495 * don't get out of sync and confuse the ring.
496 */
497int rds_iw_xmit(struct rds_connection *conn, struct rds_message *rm,
498 unsigned int hdr_off, unsigned int sg, unsigned int off)
499{
500 struct rds_iw_connection *ic = conn->c_transport_data;
501 struct ib_device *dev = ic->i_cm_id->device;
502 struct rds_iw_send_work *send = NULL;
503 struct rds_iw_send_work *first;
504 struct rds_iw_send_work *prev;
505 struct ib_send_wr *failed_wr;
506 struct scatterlist *scat;
507 u32 pos;
508 u32 i;
509 u32 work_alloc;
510 u32 credit_alloc;
511 u32 posted;
512 u32 adv_credits = 0;
513 int send_flags = 0;
514 int sent;
515 int ret;
516 int flow_controlled = 0;
517
518 BUG_ON(off % RDS_FRAG_SIZE);
519 BUG_ON(hdr_off != 0 && hdr_off != sizeof(struct rds_header));
520
521 /* Fastreg support */
Joe Perchesf64f9e72009-11-29 16:55:45 -0800522 if (rds_rdma_cookie_key(rm->m_rdma_cookie) && !ic->i_fastreg_posted) {
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000523 ret = -EAGAIN;
524 goto out;
525 }
526
527 /* FIXME we may overallocate here */
528 if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0)
529 i = 1;
530 else
531 i = ceil(be32_to_cpu(rm->m_inc.i_hdr.h_len), RDS_FRAG_SIZE);
532
533 work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, i, &pos);
534 if (work_alloc == 0) {
535 set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
536 rds_iw_stats_inc(s_iw_tx_ring_full);
537 ret = -ENOMEM;
538 goto out;
539 }
540
541 credit_alloc = work_alloc;
542 if (ic->i_flowctl) {
Steve Wise7b70d032009-04-09 14:09:39 +0000543 credit_alloc = rds_iw_send_grab_credits(ic, work_alloc, &posted, 0, RDS_MAX_ADV_CREDIT);
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000544 adv_credits += posted;
545 if (credit_alloc < work_alloc) {
546 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - credit_alloc);
547 work_alloc = credit_alloc;
548 flow_controlled++;
549 }
550 if (work_alloc == 0) {
Steve Wised39e0602009-04-09 14:09:38 +0000551 set_bit(RDS_LL_SEND_FULL, &conn->c_flags);
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000552 rds_iw_stats_inc(s_iw_tx_throttle);
553 ret = -ENOMEM;
554 goto out;
555 }
556 }
557
558 /* map the message the first time we see it */
559 if (ic->i_rm == NULL) {
560 /*
561 printk(KERN_NOTICE "rds_iw_xmit prep msg dport=%u flags=0x%x len=%d\n",
562 be16_to_cpu(rm->m_inc.i_hdr.h_dport),
563 rm->m_inc.i_hdr.h_flags,
564 be32_to_cpu(rm->m_inc.i_hdr.h_len));
565 */
566 if (rm->m_nents) {
567 rm->m_count = ib_dma_map_sg(dev,
568 rm->m_sg, rm->m_nents, DMA_TO_DEVICE);
569 rdsdebug("ic %p mapping rm %p: %d\n", ic, rm, rm->m_count);
570 if (rm->m_count == 0) {
571 rds_iw_stats_inc(s_iw_tx_sg_mapping_failure);
572 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
573 ret = -ENOMEM; /* XXX ? */
574 goto out;
575 }
576 } else {
577 rm->m_count = 0;
578 }
579
580 ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
581 ic->i_unsignaled_bytes = rds_iw_sysctl_max_unsig_bytes;
582 rds_message_addref(rm);
583 ic->i_rm = rm;
584
585 /* Finalize the header */
586 if (test_bit(RDS_MSG_ACK_REQUIRED, &rm->m_flags))
587 rm->m_inc.i_hdr.h_flags |= RDS_FLAG_ACK_REQUIRED;
588 if (test_bit(RDS_MSG_RETRANSMITTED, &rm->m_flags))
589 rm->m_inc.i_hdr.h_flags |= RDS_FLAG_RETRANSMITTED;
590
591 /* If it has a RDMA op, tell the peer we did it. This is
592 * used by the peer to release use-once RDMA MRs. */
593 if (rm->m_rdma_op) {
594 struct rds_ext_header_rdma ext_hdr;
595
596 ext_hdr.h_rdma_rkey = cpu_to_be32(rm->m_rdma_op->r_key);
597 rds_message_add_extension(&rm->m_inc.i_hdr,
598 RDS_EXTHDR_RDMA, &ext_hdr, sizeof(ext_hdr));
599 }
600 if (rm->m_rdma_cookie) {
601 rds_message_add_rdma_dest_extension(&rm->m_inc.i_hdr,
602 rds_rdma_cookie_key(rm->m_rdma_cookie),
603 rds_rdma_cookie_offset(rm->m_rdma_cookie));
604 }
605
606 /* Note - rds_iw_piggyb_ack clears the ACK_REQUIRED bit, so
607 * we should not do this unless we have a chance of at least
608 * sticking the header into the send ring. Which is why we
609 * should call rds_iw_ring_alloc first. */
610 rm->m_inc.i_hdr.h_ack = cpu_to_be64(rds_iw_piggyb_ack(ic));
611 rds_message_make_checksum(&rm->m_inc.i_hdr);
612
613 /*
614 * Update adv_credits since we reset the ACK_REQUIRED bit.
615 */
Steve Wise7b70d032009-04-09 14:09:39 +0000616 rds_iw_send_grab_credits(ic, 0, &posted, 1, RDS_MAX_ADV_CREDIT - adv_credits);
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000617 adv_credits += posted;
618 BUG_ON(adv_credits > 255);
Andy Grover735f61e2010-03-11 13:49:55 +0000619 }
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000620
621 send = &ic->i_sends[pos];
622 first = send;
623 prev = NULL;
624 scat = &rm->m_sg[sg];
625 sent = 0;
626 i = 0;
627
628 /* Sometimes you want to put a fence between an RDMA
629 * READ and the following SEND.
630 * We could either do this all the time
631 * or when requested by the user. Right now, we let
632 * the application choose.
633 */
634 if (rm->m_rdma_op && rm->m_rdma_op->r_fence)
635 send_flags = IB_SEND_FENCE;
636
637 /*
638 * We could be copying the header into the unused tail of the page.
639 * That would need to be changed in the future when those pages might
640 * be mapped userspace pages or page cache pages. So instead we always
641 * use a second sge and our long-lived ring of mapped headers. We send
642 * the header after the data so that the data payload can be aligned on
643 * the receiver.
644 */
645
646 /* handle a 0-len message */
647 if (be32_to_cpu(rm->m_inc.i_hdr.h_len) == 0) {
648 rds_iw_xmit_populate_wr(ic, send, pos, 0, 0, send_flags);
649 goto add_header;
650 }
651
652 /* if there's data reference it with a chain of work reqs */
653 for (; i < work_alloc && scat != &rm->m_sg[rm->m_count]; i++) {
654 unsigned int len;
655
656 send = &ic->i_sends[pos];
657
658 len = min(RDS_FRAG_SIZE, ib_sg_dma_len(dev, scat) - off);
659 rds_iw_xmit_populate_wr(ic, send, pos,
660 ib_sg_dma_address(dev, scat) + off, len,
661 send_flags);
662
663 /*
664 * We want to delay signaling completions just enough to get
665 * the batching benefits but not so much that we create dead time
666 * on the wire.
667 */
668 if (ic->i_unsignaled_wrs-- == 0) {
669 ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
670 send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
671 }
672
673 ic->i_unsignaled_bytes -= len;
674 if (ic->i_unsignaled_bytes <= 0) {
675 ic->i_unsignaled_bytes = rds_iw_sysctl_max_unsig_bytes;
676 send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
677 }
678
679 /*
680 * Always signal the last one if we're stopping due to flow control.
681 */
682 if (flow_controlled && i == (work_alloc-1))
683 send->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
684
685 rdsdebug("send %p wr %p num_sge %u next %p\n", send,
686 &send->s_wr, send->s_wr.num_sge, send->s_wr.next);
687
688 sent += len;
689 off += len;
690 if (off == ib_sg_dma_len(dev, scat)) {
691 scat++;
692 off = 0;
693 }
694
695add_header:
696 /* Tack on the header after the data. The header SGE should already
697 * have been set up to point to the right header buffer. */
698 memcpy(&ic->i_send_hdrs[pos], &rm->m_inc.i_hdr, sizeof(struct rds_header));
699
700 if (0) {
701 struct rds_header *hdr = &ic->i_send_hdrs[pos];
702
703 printk(KERN_NOTICE "send WR dport=%u flags=0x%x len=%d\n",
704 be16_to_cpu(hdr->h_dport),
705 hdr->h_flags,
706 be32_to_cpu(hdr->h_len));
707 }
708 if (adv_credits) {
709 struct rds_header *hdr = &ic->i_send_hdrs[pos];
710
711 /* add credit and redo the header checksum */
712 hdr->h_credit = adv_credits;
713 rds_message_make_checksum(hdr);
714 adv_credits = 0;
715 rds_iw_stats_inc(s_iw_tx_credit_updates);
716 }
717
718 if (prev)
719 prev->s_wr.next = &send->s_wr;
720 prev = send;
721
722 pos = (pos + 1) % ic->i_send_ring.w_nr;
723 }
724
725 /* Account the RDS header in the number of bytes we sent, but just once.
726 * The caller has no concept of fragmentation. */
727 if (hdr_off == 0)
728 sent += sizeof(struct rds_header);
729
730 /* if we finished the message then send completion owns it */
731 if (scat == &rm->m_sg[rm->m_count]) {
732 prev->s_rm = ic->i_rm;
733 prev->s_wr.send_flags |= IB_SEND_SIGNALED | IB_SEND_SOLICITED;
734 ic->i_rm = NULL;
735 }
736
737 if (i < work_alloc) {
738 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - i);
739 work_alloc = i;
740 }
741 if (ic->i_flowctl && i < credit_alloc)
742 rds_iw_send_add_credits(conn, credit_alloc - i);
743
744 /* XXX need to worry about failed_wr and partial sends. */
745 failed_wr = &first->s_wr;
746 ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
747 rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
748 first, &first->s_wr, ret, failed_wr);
749 BUG_ON(failed_wr != &first->s_wr);
750 if (ret) {
751 printk(KERN_WARNING "RDS/IW: ib_post_send to %pI4 "
752 "returned %d\n", &conn->c_faddr, ret);
753 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
754 if (prev->s_rm) {
755 ic->i_rm = prev->s_rm;
756 prev->s_rm = NULL;
757 }
758 goto out;
759 }
760
761 ret = sent;
762out:
763 BUG_ON(adv_credits);
764 return ret;
765}
766
767static void rds_iw_build_send_fastreg(struct rds_iw_device *rds_iwdev, struct rds_iw_connection *ic, struct rds_iw_send_work *send, int nent, int len, u64 sg_addr)
768{
769 BUG_ON(nent > send->s_page_list->max_page_list_len);
770 /*
771 * Perform a WR for the fast_reg_mr. Each individual page
772 * in the sg list is added to the fast reg page list and placed
773 * inside the fast_reg_mr WR.
774 */
775 send->s_wr.opcode = IB_WR_FAST_REG_MR;
776 send->s_wr.wr.fast_reg.length = len;
777 send->s_wr.wr.fast_reg.rkey = send->s_mr->rkey;
778 send->s_wr.wr.fast_reg.page_list = send->s_page_list;
779 send->s_wr.wr.fast_reg.page_list_len = nent;
Andy Grover404bb722009-07-17 13:13:34 +0000780 send->s_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
Andy Groverfcd8b7c2009-02-24 15:30:36 +0000781 send->s_wr.wr.fast_reg.access_flags = IB_ACCESS_REMOTE_WRITE;
782 send->s_wr.wr.fast_reg.iova_start = sg_addr;
783
784 ib_update_fast_reg_key(send->s_mr, send->s_remap_count++);
785}
786
787int rds_iw_xmit_rdma(struct rds_connection *conn, struct rds_rdma_op *op)
788{
789 struct rds_iw_connection *ic = conn->c_transport_data;
790 struct rds_iw_send_work *send = NULL;
791 struct rds_iw_send_work *first;
792 struct rds_iw_send_work *prev;
793 struct ib_send_wr *failed_wr;
794 struct rds_iw_device *rds_iwdev;
795 struct scatterlist *scat;
796 unsigned long len;
797 u64 remote_addr = op->r_remote_addr;
798 u32 pos, fr_pos;
799 u32 work_alloc;
800 u32 i;
801 u32 j;
802 int sent;
803 int ret;
804 int num_sge;
805
806 rds_iwdev = ib_get_client_data(ic->i_cm_id->device, &rds_iw_client);
807
808 /* map the message the first time we see it */
809 if (!op->r_mapped) {
810 op->r_count = ib_dma_map_sg(ic->i_cm_id->device,
811 op->r_sg, op->r_nents, (op->r_write) ?
812 DMA_TO_DEVICE : DMA_FROM_DEVICE);
813 rdsdebug("ic %p mapping op %p: %d\n", ic, op, op->r_count);
814 if (op->r_count == 0) {
815 rds_iw_stats_inc(s_iw_tx_sg_mapping_failure);
816 ret = -ENOMEM; /* XXX ? */
817 goto out;
818 }
819
820 op->r_mapped = 1;
821 }
822
823 if (!op->r_write) {
824 /* Alloc space on the send queue for the fastreg */
825 work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, 1, &fr_pos);
826 if (work_alloc != 1) {
827 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
828 rds_iw_stats_inc(s_iw_tx_ring_full);
829 ret = -ENOMEM;
830 goto out;
831 }
832 }
833
834 /*
835 * Instead of knowing how to return a partial rdma read/write we insist that there
836 * be enough work requests to send the entire message.
837 */
838 i = ceil(op->r_count, rds_iwdev->max_sge);
839
840 work_alloc = rds_iw_ring_alloc(&ic->i_send_ring, i, &pos);
841 if (work_alloc != i) {
842 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
843 rds_iw_stats_inc(s_iw_tx_ring_full);
844 ret = -ENOMEM;
845 goto out;
846 }
847
848 send = &ic->i_sends[pos];
849 if (!op->r_write) {
850 first = prev = &ic->i_sends[fr_pos];
851 } else {
852 first = send;
853 prev = NULL;
854 }
855 scat = &op->r_sg[0];
856 sent = 0;
857 num_sge = op->r_count;
858
859 for (i = 0; i < work_alloc && scat != &op->r_sg[op->r_count]; i++) {
860 send->s_wr.send_flags = 0;
861 send->s_queued = jiffies;
862
863 /*
864 * We want to delay signaling completions just enough to get
865 * the batching benefits but not so much that we create dead time on the wire.
866 */
867 if (ic->i_unsignaled_wrs-- == 0) {
868 ic->i_unsignaled_wrs = rds_iw_sysctl_max_unsig_wrs;
869 send->s_wr.send_flags = IB_SEND_SIGNALED;
870 }
871
872 /* To avoid the need to have the plumbing to invalidate the fastreg_mr used
873 * for local access after RDS is finished with it, using
874 * IB_WR_RDMA_READ_WITH_INV will invalidate it after the read has completed.
875 */
876 if (op->r_write)
877 send->s_wr.opcode = IB_WR_RDMA_WRITE;
878 else
879 send->s_wr.opcode = IB_WR_RDMA_READ_WITH_INV;
880
881 send->s_wr.wr.rdma.remote_addr = remote_addr;
882 send->s_wr.wr.rdma.rkey = op->r_key;
883 send->s_op = op;
884
885 if (num_sge > rds_iwdev->max_sge) {
886 send->s_wr.num_sge = rds_iwdev->max_sge;
887 num_sge -= rds_iwdev->max_sge;
888 } else
889 send->s_wr.num_sge = num_sge;
890
891 send->s_wr.next = NULL;
892
893 if (prev)
894 prev->s_wr.next = &send->s_wr;
895
896 for (j = 0; j < send->s_wr.num_sge && scat != &op->r_sg[op->r_count]; j++) {
897 len = ib_sg_dma_len(ic->i_cm_id->device, scat);
898
899 if (send->s_wr.opcode == IB_WR_RDMA_READ_WITH_INV)
900 send->s_page_list->page_list[j] = ib_sg_dma_address(ic->i_cm_id->device, scat);
901 else {
902 send->s_sge[j].addr = ib_sg_dma_address(ic->i_cm_id->device, scat);
903 send->s_sge[j].length = len;
904 send->s_sge[j].lkey = rds_iw_local_dma_lkey(ic);
905 }
906
907 sent += len;
908 rdsdebug("ic %p sent %d remote_addr %llu\n", ic, sent, remote_addr);
909 remote_addr += len;
910
911 scat++;
912 }
913
914 if (send->s_wr.opcode == IB_WR_RDMA_READ_WITH_INV) {
915 send->s_wr.num_sge = 1;
916 send->s_sge[0].addr = conn->c_xmit_rm->m_rs->rs_user_addr;
917 send->s_sge[0].length = conn->c_xmit_rm->m_rs->rs_user_bytes;
918 send->s_sge[0].lkey = ic->i_sends[fr_pos].s_mr->lkey;
919 }
920
921 rdsdebug("send %p wr %p num_sge %u next %p\n", send,
922 &send->s_wr, send->s_wr.num_sge, send->s_wr.next);
923
924 prev = send;
925 if (++send == &ic->i_sends[ic->i_send_ring.w_nr])
926 send = ic->i_sends;
927 }
928
929 /* if we finished the message then send completion owns it */
930 if (scat == &op->r_sg[op->r_count])
931 first->s_wr.send_flags = IB_SEND_SIGNALED;
932
933 if (i < work_alloc) {
934 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc - i);
935 work_alloc = i;
936 }
937
938 /* On iWARP, local memory access by a remote system (ie, RDMA Read) is not
939 * recommended. Putting the lkey on the wire is a security hole, as it can
940 * allow for memory access to all of memory on the remote system. Some
941 * adapters do not allow using the lkey for this at all. To bypass this use a
942 * fastreg_mr (or possibly a dma_mr)
943 */
944 if (!op->r_write) {
945 rds_iw_build_send_fastreg(rds_iwdev, ic, &ic->i_sends[fr_pos],
946 op->r_count, sent, conn->c_xmit_rm->m_rs->rs_user_addr);
947 work_alloc++;
948 }
949
950 failed_wr = &first->s_wr;
951 ret = ib_post_send(ic->i_cm_id->qp, &first->s_wr, &failed_wr);
952 rdsdebug("ic %p first %p (wr %p) ret %d wr %p\n", ic,
953 first, &first->s_wr, ret, failed_wr);
954 BUG_ON(failed_wr != &first->s_wr);
955 if (ret) {
956 printk(KERN_WARNING "RDS/IW: rdma ib_post_send to %pI4 "
957 "returned %d\n", &conn->c_faddr, ret);
958 rds_iw_ring_unalloc(&ic->i_send_ring, work_alloc);
959 goto out;
960 }
961
962out:
963 return ret;
964}
965
966void rds_iw_xmit_complete(struct rds_connection *conn)
967{
968 struct rds_iw_connection *ic = conn->c_transport_data;
969
970 /* We may have a pending ACK or window update we were unable
971 * to send previously (due to flow control). Try again. */
972 rds_iw_attempt_ack(ic);
973}