1 /*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21 /*
22 * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
23 * Use is subject to license terms.
24 * Copyright 2018 Joyent, Inc.
25 */
26
27 /*
28 * General Soft rings - Simulating Rx rings in S/W.
29 *
30 * Soft ring is a data abstraction containing a queue and a worker
31 * thread and represents a hardware Rx ring in software. Each soft
32 * ring set can have a collection of soft rings for separating
33 * L3/L4 specific traffic (IPv4 from IPv6 or TCP from UDP) or for
34 * allowing a higher degree of parallelism by sending traffic to
35 * one of the soft rings for a SRS (using a hash on src IP or port).
36 * Each soft ring worker thread can be bound to a different CPU
37 * allowing the processing for each soft ring to happen in parallel
38 * and independent from each other.
39 *
40 * Protocol soft rings:
41 *
42 * Each SRS has at an minimum 3 softrings. One each for IPv4 TCP,
43 * IPv4 UDP and rest (OTH - for IPv6 and everything else). The
44 * SRS does dynamic polling and enforces link level bandwidth but
45 * it does so for all traffic (IPv4 and IPv6 and all protocols) on
46 * that link. However, each protocol layer wants a different
47 * behaviour. For instance IPv4 TCP has per CPU squeues which
48 * enforce their own polling and flow control so IPv4 TCP traffic
49 * needs to go to a separate soft ring which can be polled by the
50 * TCP squeue. It also allows TCP squeue to push back flow control
51 * all the way to NIC hardware (if it puts its corresponding soft
52 * ring in the poll mode and soft ring queue builds up, the
53 * shared srs_poll_pkt_cnt goes up and SRS automatically stops
54 * more packets from entering the system).
55 *
56 * Similarly, the UDP benefits from a DLS bypass and packet chaining
57 * so sending it to a separate soft ring is desired. All the rest of
58 * the traffic (including IPv6 is sent to OTH softring). The IPv6
59 * traffic current goes through OTH softring and via DLS because
60 * it need more processing to be done. Irrespective of the sap
61 * (IPv4 or IPv6) or the transport, the dynamic polling, B/W enforcement,
62 * cpu assignment, fanout, etc apply to all traffic since they
63 * are implement by the SRS which is agnostic to sap or transport.
64 *
65 * Fanout soft rings:
66 *
67 * On a multithreaded system, we can assign more CPU and multi thread
68 * the stack by creating a soft ring per CPU and spreading traffic
69 * based on a hash computed on src IP etc. Since we still need to
70 * keep the protocol separation, we create a set of 3 soft ring per
71 * CPU (specified by cpu list or degree of fanout).
72 *
73 * NOTE: See the block level comment on top of mac_sched.c
74 */
75
76 #include <sys/types.h>
77 #include <sys/callb.h>
78 #include <sys/sdt.h>
79 #include <sys/strsubr.h>
80 #include <sys/strsun.h>
81 #include <sys/vlan.h>
82 #include <inet/ipsec_impl.h>
83 #include <inet/ip_impl.h>
84 #include <inet/sadb.h>
85 #include <inet/ipsecesp.h>
86 #include <inet/ipsecah.h>
87
88 #include <sys/mac_impl.h>
89 #include <sys/mac_client_impl.h>
90 #include <sys/mac_soft_ring.h>
91 #include <sys/mac_flow_impl.h>
92 #include <sys/mac_stat.h>
93
94 static void mac_rx_soft_ring_drain(mac_soft_ring_t *);
95 static void mac_soft_ring_fire(void *);
96 static void mac_soft_ring_worker(mac_soft_ring_t *);
97 static void mac_tx_soft_ring_drain(mac_soft_ring_t *);
98
99 uint32_t mac_tx_soft_ring_max_q_cnt = 100000;
100 uint32_t mac_tx_soft_ring_hiwat = 1000;
101
102 extern kmem_cache_t *mac_soft_ring_cache;
103
104 #define ADD_SOFTRING_TO_SET(mac_srs, softring) { \
105 if (mac_srs->srs_soft_ring_head == NULL) { \
106 mac_srs->srs_soft_ring_head = softring; \
107 mac_srs->srs_soft_ring_tail = softring; \
108 } else { \
109 /* ADD to the list */ \
110 softring->s_ring_prev = \
111 mac_srs->srs_soft_ring_tail; \
112 mac_srs->srs_soft_ring_tail->s_ring_next = softring; \
113 mac_srs->srs_soft_ring_tail = softring; \
114 } \
115 mac_srs->srs_soft_ring_count++; \
116 }
117
118 /*
119 * mac_soft_ring_worker_wakeup
120 *
121 * Wake up the soft ring worker thread to process the queue as long
122 * as no one else is processing it and upper layer (client) is still
123 * ready to receive packets.
124 */
125 void
126 mac_soft_ring_worker_wakeup(mac_soft_ring_t *ringp)
127 {
128 ASSERT(MUTEX_HELD(&ringp->s_ring_lock));
129 if (!(ringp->s_ring_state & S_RING_PROC) &&
130 !(ringp->s_ring_state & S_RING_BLANK) &&
131 (ringp->s_ring_tid == NULL)) {
132 if (ringp->s_ring_wait != 0) {
133 ringp->s_ring_tid =
134 timeout(mac_soft_ring_fire, ringp,
135 ringp->s_ring_wait);
136 } else {
137 /* Schedule the worker thread. */
138 cv_signal(&ringp->s_ring_async);
139 }
140 }
141 }
142
143 /*
144 * mac_soft_ring_create
145 *
146 * Create a soft ring, do the necessary setup and bind the worker
147 * thread to the assigned CPU.
148 */
149 mac_soft_ring_t *
150 mac_soft_ring_create(int id, clock_t wait, uint16_t type,
151 pri_t pri, mac_client_impl_t *mcip, mac_soft_ring_set_t *mac_srs,
152 processorid_t cpuid, mac_direct_rx_t rx_func, void *x_arg1,
153 mac_resource_handle_t x_arg2)
154 {
155 mac_soft_ring_t *ringp;
156 char name[S_RING_NAMELEN];
157
158 bzero(name, 64);
159 ringp = kmem_cache_alloc(mac_soft_ring_cache, KM_SLEEP);
160
161 if (type & ST_RING_TCP) {
162 (void) snprintf(name, sizeof (name),
163 "mac_tcp_soft_ring_%d_%p", id, (void *)mac_srs);
164 } else if (type & ST_RING_UDP) {
165 (void) snprintf(name, sizeof (name),
166 "mac_udp_soft_ring_%d_%p", id, (void *)mac_srs);
167 } else if (type & ST_RING_OTH) {
168 (void) snprintf(name, sizeof (name),
169 "mac_oth_soft_ring_%d_%p", id, (void *)mac_srs);
170 } else {
171 ASSERT(type & ST_RING_TX);
172 (void) snprintf(name, sizeof (name),
173 "mac_tx_soft_ring_%d_%p", id, (void *)mac_srs);
174 }
175
176 bzero(ringp, sizeof (mac_soft_ring_t));
177 (void) strncpy(ringp->s_ring_name, name, S_RING_NAMELEN + 1);
178 ringp->s_ring_name[S_RING_NAMELEN] = '\0';
179 mutex_init(&ringp->s_ring_lock, NULL, MUTEX_DEFAULT, NULL);
180 ringp->s_ring_notify_cb_info.mcbi_lockp = &ringp->s_ring_lock;
181
182 ringp->s_ring_type = type;
183 ringp->s_ring_wait = MSEC_TO_TICK(wait);
184 ringp->s_ring_mcip = mcip;
185 ringp->s_ring_set = mac_srs;
186
187 /*
188 * Protect against access from DR callbacks (mac_walk_srs_bind/unbind)
189 * which can't grab the mac perimeter
190 */
191 mutex_enter(&mac_srs->srs_lock);
192 ADD_SOFTRING_TO_SET(mac_srs, ringp);
193 mutex_exit(&mac_srs->srs_lock);
194
195 /*
196 * set the bind CPU to -1 to indicate
197 * no thread affinity set
198 */
199 ringp->s_ring_cpuid = ringp->s_ring_cpuid_save = -1;
200 ringp->s_ring_worker = thread_create(NULL, 0,
201 mac_soft_ring_worker, ringp, 0, &p0, TS_RUN, pri);
202 if (type & ST_RING_TX) {
203 ringp->s_ring_drain_func = mac_tx_soft_ring_drain;
204 ringp->s_ring_tx_arg1 = x_arg1;
205 ringp->s_ring_tx_arg2 = x_arg2;
206 ringp->s_ring_tx_max_q_cnt = mac_tx_soft_ring_max_q_cnt;
207 ringp->s_ring_tx_hiwat =
208 (mac_tx_soft_ring_hiwat > mac_tx_soft_ring_max_q_cnt) ?
209 mac_tx_soft_ring_max_q_cnt : mac_tx_soft_ring_hiwat;
210 if (mcip->mci_state_flags & MCIS_IS_AGGR_CLIENT) {
211 mac_srs_tx_t *tx = &mac_srs->srs_tx;
212
213 ASSERT(tx->st_soft_rings[
214 ((mac_ring_t *)x_arg2)->mr_index] == NULL);
215 tx->st_soft_rings[((mac_ring_t *)x_arg2)->mr_index] =
216 ringp;
217 }
218 } else {
219 ringp->s_ring_drain_func = mac_rx_soft_ring_drain;
220 ringp->s_ring_rx_func = rx_func;
221 ringp->s_ring_rx_arg1 = x_arg1;
222 ringp->s_ring_rx_arg2 = x_arg2;
223 if (mac_srs->srs_state & SRS_SOFTRING_QUEUE)
224 ringp->s_ring_type |= ST_RING_WORKER_ONLY;
225 }
226 if (cpuid != -1)
227 (void) mac_soft_ring_bind(ringp, cpuid);
228
229 mac_soft_ring_stat_create(ringp);
230
231 return (ringp);
232 }
233
234 /*
235 * mac_soft_ring_free
236 *
237 * Free the soft ring once we are done with it.
238 */
239 void
240 mac_soft_ring_free(mac_soft_ring_t *softring)
241 {
242 ASSERT((softring->s_ring_state &
243 (S_RING_CONDEMNED | S_RING_CONDEMNED_DONE | S_RING_PROC)) ==
244 (S_RING_CONDEMNED | S_RING_CONDEMNED_DONE));
245 mac_pkt_drop(NULL, NULL, softring->s_ring_first, B_FALSE);
246 softring->s_ring_tx_arg2 = NULL;
247 mac_soft_ring_stat_delete(softring);
248 mac_callback_free(softring->s_ring_notify_cb_list);
249 kmem_cache_free(mac_soft_ring_cache, softring);
250 }
251
252 int mac_soft_ring_thread_bind = 1;
253
254 /*
255 * mac_soft_ring_bind
256 *
257 * Bind a soft ring worker thread to supplied CPU.
258 */
259 cpu_t *
260 mac_soft_ring_bind(mac_soft_ring_t *ringp, processorid_t cpuid)
261 {
262 cpu_t *cp;
263 boolean_t clear = B_FALSE;
264
265 ASSERT(MUTEX_HELD(&cpu_lock));
266
267 if (mac_soft_ring_thread_bind == 0) {
268 DTRACE_PROBE1(mac__soft__ring__no__cpu__bound,
269 mac_soft_ring_t *, ringp);
270 return (NULL);
271 }
272
273 cp = cpu_get(cpuid);
274 if (cp == NULL || !cpu_is_online(cp))
275 return (NULL);
276
277 mutex_enter(&ringp->s_ring_lock);
278 ringp->s_ring_state |= S_RING_BOUND;
279 if (ringp->s_ring_cpuid != -1)
280 clear = B_TRUE;
281 ringp->s_ring_cpuid = cpuid;
282 mutex_exit(&ringp->s_ring_lock);
283
284 if (clear)
285 thread_affinity_clear(ringp->s_ring_worker);
286
287 DTRACE_PROBE2(mac__soft__ring__cpu__bound, mac_soft_ring_t *,
288 ringp, processorid_t, cpuid);
289
290 thread_affinity_set(ringp->s_ring_worker, cpuid);
291
292 return (cp);
293 }
294
295 /*
296 * mac_soft_ring_unbind
297 *
298 * Un Bind a soft ring worker thread.
299 */
300 void
301 mac_soft_ring_unbind(mac_soft_ring_t *ringp)
302 {
303 ASSERT(MUTEX_HELD(&cpu_lock));
304
305 mutex_enter(&ringp->s_ring_lock);
306 if (!(ringp->s_ring_state & S_RING_BOUND)) {
307 ASSERT(ringp->s_ring_cpuid == -1);
308 mutex_exit(&ringp->s_ring_lock);
309 return;
310 }
311
312 ringp->s_ring_cpuid = -1;
313 ringp->s_ring_state &= ~S_RING_BOUND;
314 thread_affinity_clear(ringp->s_ring_worker);
315 mutex_exit(&ringp->s_ring_lock);
316 }
317
318 /*
319 * PRIVATE FUNCTIONS
320 */
321
322 static void
323 mac_soft_ring_fire(void *arg)
324 {
325 mac_soft_ring_t *ringp = arg;
326
327 mutex_enter(&ringp->s_ring_lock);
328 if (ringp->s_ring_tid == NULL) {
329 mutex_exit(&ringp->s_ring_lock);
330 return;
331 }
332
333 ringp->s_ring_tid = NULL;
334
335 if (!(ringp->s_ring_state & S_RING_PROC)) {
336 cv_signal(&ringp->s_ring_async);
337 }
338 mutex_exit(&ringp->s_ring_lock);
339 }
340
341 /*
342 * Drain the soft ring pointed to by ringp.
343 *
344 * o s_ring_first: pointer to the queued packet chain.
345 *
346 * o s_ring_rx_func: pointer to to the client's Rx routine.
347 *
348 * o s_ring_rx_{arg1,arg2}: opaque values specific to the client.
349 */
350 static void
351 mac_rx_soft_ring_drain(mac_soft_ring_t *ringp)
352 {
353 mblk_t *mp;
354 void *arg1;
355 mac_resource_handle_t arg2;
356 timeout_id_t tid;
357 mac_direct_rx_t proc;
358 size_t sz;
359 int cnt;
360 mac_soft_ring_set_t *mac_srs = ringp->s_ring_set;
361
362 ringp->s_ring_run = curthread;
363 ASSERT(mutex_owned(&ringp->s_ring_lock));
364 ASSERT(!(ringp->s_ring_state & S_RING_PROC));
365
366 if ((tid = ringp->s_ring_tid) != NULL)
367 ringp->s_ring_tid = NULL;
368
369 ringp->s_ring_state |= S_RING_PROC;
370
371 proc = ringp->s_ring_rx_func;
372 arg1 = ringp->s_ring_rx_arg1;
373 arg2 = ringp->s_ring_rx_arg2;
374
375 while ((ringp->s_ring_first != NULL) &&
376 !(ringp->s_ring_state & S_RING_PAUSE)) {
377 mp = ringp->s_ring_first;
378 ringp->s_ring_first = NULL;
379 ringp->s_ring_last = NULL;
380 cnt = ringp->s_ring_count;
381 ringp->s_ring_count = 0;
382 sz = ringp->s_ring_size;
383 ringp->s_ring_size = 0;
384 mutex_exit(&ringp->s_ring_lock);
385
386 if (tid != NULL) {
387 (void) untimeout(tid);
388 tid = NULL;
389 }
390
391 (*proc)(arg1, arg2, mp, NULL);
392
393 /*
394 * If we have an SRS performing bandwidth control, then
395 * we need to decrement the size and count so the SRS
396 * has an accurate measure of the data queued between
397 * the SRS and its soft rings. We decrement the
398 * counters only when the packet is processed by both
399 * the SRS and the soft ring.
400 */
401 mutex_enter(&mac_srs->srs_lock);
402 MAC_UPDATE_SRS_COUNT_LOCKED(mac_srs, cnt);
403 MAC_UPDATE_SRS_SIZE_LOCKED(mac_srs, sz);
404 mutex_exit(&mac_srs->srs_lock);
405
406 mutex_enter(&ringp->s_ring_lock);
407 }
408 ringp->s_ring_state &= ~S_RING_PROC;
409 if (ringp->s_ring_state & S_RING_CLIENT_WAIT)
410 cv_signal(&ringp->s_ring_client_cv);
411 ringp->s_ring_run = NULL;
412 }
413
414 /*
415 * The soft ring worker routine to process any queued packets. In
416 * normal case, the worker thread is bound to a CPU. If the soft ring
417 * handles TCP packets then the worker thread is bound to the same CPU
418 * as the TCP squeue.
419 */
420 static void
421 mac_soft_ring_worker(mac_soft_ring_t *ringp)
422 {
423 kmutex_t *lock = &ringp->s_ring_lock;
424 kcondvar_t *async = &ringp->s_ring_async;
425 mac_soft_ring_set_t *srs = ringp->s_ring_set;
426 callb_cpr_t cprinfo;
427
428 CALLB_CPR_INIT(&cprinfo, lock, callb_generic_cpr, "mac_soft_ring");
429 mutex_enter(lock);
430 start:
431 for (;;) {
432 while (((ringp->s_ring_first == NULL ||
433 (ringp->s_ring_state & (S_RING_BLOCK|S_RING_BLANK))) &&
434 !(ringp->s_ring_state & S_RING_PAUSE)) ||
435 (ringp->s_ring_state & S_RING_PROC)) {
436
437 CALLB_CPR_SAFE_BEGIN(&cprinfo);
438 cv_wait(async, lock);
439 CALLB_CPR_SAFE_END(&cprinfo, lock);
440 }
441
442 /*
443 * Either we have work to do, or we have been asked to
444 * shutdown temporarily or permanently
445 */
446 if (ringp->s_ring_state & S_RING_PAUSE)
447 goto done;
448
449 ringp->s_ring_drain_func(ringp);
450 }
451 done:
452 mutex_exit(lock);
453 mutex_enter(&srs->srs_lock);
454 mutex_enter(lock);
455
456 ringp->s_ring_state |= S_RING_QUIESCE_DONE;
457 if (!(ringp->s_ring_state & S_RING_CONDEMNED)) {
458 srs->srs_soft_ring_quiesced_count++;
459 cv_broadcast(&srs->srs_async);
460 mutex_exit(&srs->srs_lock);
461 while (!(ringp->s_ring_state &
462 (S_RING_RESTART | S_RING_CONDEMNED)))
463 cv_wait(&ringp->s_ring_async, &ringp->s_ring_lock);
464 mutex_exit(lock);
465 mutex_enter(&srs->srs_lock);
466 mutex_enter(lock);
467 srs->srs_soft_ring_quiesced_count--;
468 if (ringp->s_ring_state & S_RING_RESTART) {
469 ASSERT(!(ringp->s_ring_state & S_RING_CONDEMNED));
470 ringp->s_ring_state &= ~(S_RING_RESTART |
471 S_RING_QUIESCE | S_RING_QUIESCE_DONE);
472 cv_broadcast(&srs->srs_async);
473 mutex_exit(&srs->srs_lock);
474 goto start;
475 }
476 }
477 ASSERT(ringp->s_ring_state & S_RING_CONDEMNED);
478 ringp->s_ring_state |= S_RING_CONDEMNED_DONE;
479 CALLB_CPR_EXIT(&cprinfo);
480 srs->srs_soft_ring_condemned_count++;
481 cv_broadcast(&srs->srs_async);
482 mutex_exit(&srs->srs_lock);
483 thread_exit();
484 }
485
486 /*
487 * mac_soft_ring_intr_enable and mac_soft_ring_intr_disable
488 *
489 * these functions are called to toggle the sending of packets to the
490 * client. They are called by the client. the client gets the name
491 * of these routine and corresponding cookie (pointing to softring)
492 * during capability negotiation at setup time.
493 *
494 * Enabling is allow the processing thread to send packets to the
495 * client while disabling does the opposite.
496 */
497 int
498 mac_soft_ring_intr_enable(void *arg)
499 {
500 mac_soft_ring_t *ringp = (mac_soft_ring_t *)arg;
501 mutex_enter(&ringp->s_ring_lock);
502 ringp->s_ring_state &= ~S_RING_BLANK;
503 if (ringp->s_ring_first != NULL)
504 mac_soft_ring_worker_wakeup(ringp);
505 mutex_exit(&ringp->s_ring_lock);
506 return (0);
507 }
508
509 boolean_t
510 mac_soft_ring_intr_disable(void *arg)
511 {
512 mac_soft_ring_t *ringp = (mac_soft_ring_t *)arg;
513 boolean_t sring_blanked = B_FALSE;
514 /*
515 * Stop worker thread from sending packets above.
516 * Squeue will poll soft ring when it needs packets.
517 */
518 mutex_enter(&ringp->s_ring_lock);
519 if (!(ringp->s_ring_state & S_RING_PROC)) {
520 ringp->s_ring_state |= S_RING_BLANK;
521 sring_blanked = B_TRUE;
522 }
523 mutex_exit(&ringp->s_ring_lock);
524 return (sring_blanked);
525 }
526
527 /*
528 * mac_soft_ring_poll
529 *
530 * This routine is called by the client to poll for packets from
531 * the soft ring. The function name and cookie corresponding to
532 * the soft ring is exchanged during capability negotiation during
533 * setup.
534 */
535 mblk_t *
536 mac_soft_ring_poll(mac_soft_ring_t *ringp, size_t bytes_to_pickup)
537 {
538 mblk_t *head, *tail;
539 mblk_t *mp;
540 size_t sz = 0;
541 int cnt = 0;
542 mac_soft_ring_set_t *mac_srs = ringp->s_ring_set;
543
544 ASSERT(mac_srs != NULL);
545
546 mutex_enter(&ringp->s_ring_lock);
547 head = tail = mp = ringp->s_ring_first;
548 if (head == NULL) {
549 mutex_exit(&ringp->s_ring_lock);
550 return (NULL);
551 }
552
553 if (ringp->s_ring_size <= bytes_to_pickup) {
554 head = ringp->s_ring_first;
555 ringp->s_ring_first = NULL;
556 ringp->s_ring_last = NULL;
557 cnt = ringp->s_ring_count;
558 ringp->s_ring_count = 0;
559 sz = ringp->s_ring_size;
560 ringp->s_ring_size = 0;
561 } else {
562 while (mp && sz <= bytes_to_pickup) {
563 sz += msgdsize(mp);
564 cnt++;
565 tail = mp;
566 mp = mp->b_next;
567 }
568 ringp->s_ring_count -= cnt;
569 ringp->s_ring_size -= sz;
570 tail->b_next = NULL;
571 if (mp == NULL) {
572 ringp->s_ring_first = NULL;
573 ringp->s_ring_last = NULL;
574 ASSERT(ringp->s_ring_count == 0);
575 } else {
576 ringp->s_ring_first = mp;
577 }
578 }
579
580 mutex_exit(&ringp->s_ring_lock);
581 /*
582 * Update the shared count and size counters so
583 * that SRS has a accurate idea of queued packets.
584 */
585 mutex_enter(&mac_srs->srs_lock);
586 MAC_UPDATE_SRS_COUNT_LOCKED(mac_srs, cnt);
587 MAC_UPDATE_SRS_SIZE_LOCKED(mac_srs, sz);
588 mutex_exit(&mac_srs->srs_lock);
589 return (head);
590 }
591
592 /*
593 * mac_soft_ring_dls_bypass
594 *
595 * Enable direct client (IP) callback function from the softrings.
596 * Callers need to make sure they don't need any DLS layer processing
597 */
598 void
599 mac_soft_ring_dls_bypass(void *arg, mac_direct_rx_t rx_func, void *rx_arg1)
600 {
601 mac_soft_ring_t *softring = arg;
602 mac_soft_ring_set_t *srs;
603
604 VERIFY3P(rx_func, !=, NULL);
605
606 mutex_enter(&softring->s_ring_lock);
607 softring->s_ring_rx_func = rx_func;
608 softring->s_ring_rx_arg1 = rx_arg1;
609 mutex_exit(&softring->s_ring_lock);
610
611 srs = softring->s_ring_set;
612 mutex_enter(&srs->srs_lock);
613 srs->srs_type |= SRST_DLS_BYPASS;
614 mutex_exit(&srs->srs_lock);
615 }
616
617 /*
618 * mac_soft_ring_signal
619 *
620 * Typically used to set the soft ring state to QUIESCE, CONDEMNED, or
621 * RESTART.
622 *
623 * In the Rx side, the quiescing is done bottom up. After the Rx upcalls
624 * from the driver are done, then the Rx SRS is quiesced and only then can
625 * we signal the soft rings. Thus this function can't be called arbitrarily
626 * without satisfying the prerequisites. On the Tx side, the threads from
627 * top need to quiesced, then the Tx SRS and only then can we signal the
628 * Tx soft rings.
629 */
630 void
631 mac_soft_ring_signal(mac_soft_ring_t *softring, uint_t sr_flag)
632 {
633 mutex_enter(&softring->s_ring_lock);
634 softring->s_ring_state |= sr_flag;
635 cv_signal(&softring->s_ring_async);
636 mutex_exit(&softring->s_ring_lock);
637 }
638
639 /*
640 * mac_tx_soft_ring_drain
641 *
642 * The transmit side drain routine in case the soft ring was being
643 * used to transmit packets.
644 */
645 static void
646 mac_tx_soft_ring_drain(mac_soft_ring_t *ringp)
647 {
648 mblk_t *mp;
649 void *arg1;
650 void *arg2;
651 mblk_t *tail;
652 uint_t saved_pkt_count, saved_size;
653 mac_tx_stats_t stats;
654 mac_soft_ring_set_t *mac_srs = ringp->s_ring_set;
655
656 saved_pkt_count = saved_size = 0;
657 ringp->s_ring_run = curthread;
658 ASSERT(mutex_owned(&ringp->s_ring_lock));
659 ASSERT(!(ringp->s_ring_state & S_RING_PROC));
660
661 ringp->s_ring_state |= S_RING_PROC;
662 arg1 = ringp->s_ring_tx_arg1;
663 arg2 = ringp->s_ring_tx_arg2;
664
665 while (ringp->s_ring_first != NULL) {
666 mp = ringp->s_ring_first;
667 tail = ringp->s_ring_last;
668 saved_pkt_count = ringp->s_ring_count;
669 saved_size = ringp->s_ring_size;
670 ringp->s_ring_first = NULL;
671 ringp->s_ring_last = NULL;
672 ringp->s_ring_count = 0;
673 ringp->s_ring_size = 0;
674 mutex_exit(&ringp->s_ring_lock);
675
676 mp = mac_tx_send(arg1, arg2, mp, &stats);
677
678 mutex_enter(&ringp->s_ring_lock);
679 if (mp != NULL) {
680 /* Device out of tx desc, set block */
681 tail->b_next = ringp->s_ring_first;
682 ringp->s_ring_first = mp;
683 ringp->s_ring_count +=
684 (saved_pkt_count - stats.mts_opackets);
685 ringp->s_ring_size += (saved_size - stats.mts_obytes);
686 if (ringp->s_ring_last == NULL)
687 ringp->s_ring_last = tail;
688
689 if (ringp->s_ring_tx_woken_up) {
690 ringp->s_ring_tx_woken_up = B_FALSE;
691 } else {
692 ringp->s_ring_state |= S_RING_BLOCK;
693 ringp->s_st_stat.mts_blockcnt++;
694 }
695
696 ringp->s_ring_state &= ~S_RING_PROC;
697 ringp->s_ring_run = NULL;
698 return;
699 } else {
700 ringp->s_ring_tx_woken_up = B_FALSE;
701 SRS_TX_STATS_UPDATE(mac_srs, &stats);
702 SOFTRING_TX_STATS_UPDATE(ringp, &stats);
703 }
704 }
705
706 if (ringp->s_ring_count == 0 && ringp->s_ring_state &
707 (S_RING_TX_HIWAT | S_RING_WAKEUP_CLIENT | S_RING_ENQUEUED)) {
708 mac_client_impl_t *mcip = ringp->s_ring_mcip;
709 boolean_t wakeup_required = B_FALSE;
710
711 if (ringp->s_ring_state &
712 (S_RING_TX_HIWAT|S_RING_WAKEUP_CLIENT)) {
713 wakeup_required = B_TRUE;
714 }
715 ringp->s_ring_state &=
716 ~(S_RING_TX_HIWAT | S_RING_WAKEUP_CLIENT | S_RING_ENQUEUED);
717 mutex_exit(&ringp->s_ring_lock);
718 if (wakeup_required) {
719 mac_tx_invoke_callbacks(mcip, (mac_tx_cookie_t)ringp);
720 /*
721 * If the client is not the primary MAC client, then we
722 * need to send the notification to the clients upper
723 * MAC, i.e. mci_upper_mip.
724 */
725 mac_tx_notify(mcip->mci_upper_mip != NULL ?
726 mcip->mci_upper_mip : mcip->mci_mip);
727 }
728 mutex_enter(&ringp->s_ring_lock);
729 }
730 ringp->s_ring_state &= ~S_RING_PROC;
731 ringp->s_ring_run = NULL;
732 }