xref: /titanic_44/usr/src/cmd/fm/fmd/common/fmd_xprt.c (revision 737d277a27d4872543f597e35c470e7510f61f03)
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, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
11  * See the License for the specific language governing permissions
12  * and limitations under the License.
13  *
14  * When distributing Covered Code, include this CDDL HEADER in each
15  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16  * If applicable, add the following below this CDDL HEADER, with the
17  * fields enclosed by brackets "[]" replaced with your own identifying
18  * information: Portions Copyright [yyyy] [name of copyright owner]
19  *
20  * CDDL HEADER END
21  */
22 
23 /*
24  * Copyright 2005 Sun Microsystems, Inc.  All rights reserved.
25  * Use is subject to license terms.
26  */
27 
28 #pragma ident	"%Z%%M%	%I%	%E% SMI"
29 
30 /*
31  * FMD Transport Subsystem
32  *
33  * A transport module uses some underlying mechanism to transport events.
34  * This mechanism may use any underlying link-layer protocol and may support
35  * additional link-layer packets unrelated to FMA.  Some appropriate link-
36  * layer mechanism to create the underlying connection is expected to be
37  * called prior to calling fmd_xprt_open() itself.  Alternatively, a transport
38  * may be created in the suspended state by specifying the FMD_XPRT_SUSPENDED
39  * flag as part of the call to fmd_xprt_open(), and then may be resumed later.
40  * The underlying transport mechanism is *required* to provide ordering: that
41  * is, the sequences of bytes written across the transport must be read by
42  * the remote peer in the order that they are written, even across separate
43  * calls to fmdo_send().  As an example, the Internet TCP protocol would be
44  * a valid transport as it guarantees ordering, whereas the Internet UDP
45  * protocol would not because UDP datagrams may be delivered in any order
46  * as a result of delays introduced when datagrams pass through routers.
47  *
48  * Similar to sending events, a transport module receives events that are from
49  * its peer remote endpoint using some transport-specific mechanism that is
50  * unknown to FMD.  As each event is received, the transport module is
51  * responsible for constructing a valid nvlist_t object from the data and then
52  * calling fmd_xprt_post() to post the event to the containing FMD's dispatch
53  * queue, making it available to all local modules that are not transport
54  * modules that have subscribed to the event.
55  *
56  * The following state machine is used for each transport.  The initial state
57  * is either SYN, ACK, or RUN, depending on the flags specified to xprt_create.
58  *
59  *       FMD_XPRT_ACCEPT   !FMD_XPRT_ACCEPT
60  *             |                 |
61  * waiting  +--v--+           +--v--+  waiting
62  * for syn  | SYN |--+     --+| ACK |  for ack
63  * event    +-----+   \   /   +-----+  event
64  *             |       \ /       |
65  * drop all +--v--+     X     +--v--+  send subscriptions,
66  * events   | ERR |<---+ +--->| SUB |  recv subscriptions,
67  *          +-----+           +-----+  wait for run event
68  *             ^                 |
69  *             |     +-----+     |
70  *             +-----| RUN |<----+
71  *                   +--^--+
72  *                      |
73  *               FMD_XPRT_RDONLY
74  *
75  * When fmd_xprt_open() is called without FMD_XPRT_ACCEPT, the Common Transport
76  * Layer enqueues a "syn" event for the module in its event queue and sets the
77  * state to ACK.  In state ACK, we are waiting for the transport to get an
78  * "ack" event and call fmd_xprt_post() on this event.  Other events will be
79  * discarded.  If an "ack" is received, we transition to state SUB.  If a
80  * configurable timeout occurs or if the "ack" is invalid (e.g. invalid version
81  * exchange), we transition to state ERR.  Once in state ERR, no further
82  * operations are valid except fmd_xprt_close() and fmd_xprt_error() will
83  * return a non-zero value to the caller indicating the transport has failed.
84  *
85  * When fmd_xprt_open() is called with FMD_XPRT_ACCEPT, the Common Transport
86  * Layer assumes this transport is being used to accept a virtual connection
87  * from a remote peer that is sending a "syn", and sets the initial state to
88  * SYN.  In this state, the transport waits for a "syn" event, validates it,
89  * and then transitions to state SUB if it is valid or state ERR if it is not.
90  *
91  * Once in state SUB, the transport module is expected to receive a sequence of
92  * zero or more "subscribe" events from the remote peer, followed by a "run"
93  * event.  Once in state RUN, the transport is active and any events can be
94  * sent or received.  The transport module is free to call fmd_xprt_close()
95  * from any state.  The fmd_xprt_error() function will return zero if the
96  * transport is not in the ERR state, or non-zero if it is in the ERR state.
97  *
98  * Once the state machine reaches RUN, other FMA protocol events can be sent
99  * and received across the transport in addition to the various control events.
100  *
101  * Table of Common Transport Layer Control Events
102  * ==============================================
103  *
104  * FMA Class                     Payload
105  * ---------                     -------
106  * resource.fm.xprt.uuclose      string (uuid of case)
107  * resource.fm.xprt.subscribe    string (class pattern)
108  * resource.fm.xprt.unsubscribe  string (class pattern)
109  * resource.fm.xprt.unsuback     string (class pattern)
110  * resource.fm.xprt.syn          version information
111  * resource.fm.xprt.ack          version information
112  * resource.fm.xprt.run          version information
113  *
114  * Control events are used to add and delete proxy subscriptions on the remote
115  * transport peer module, and to set up connections.  When a "syn" event is
116  * sent, FMD will include in the payload the highest version of the FMA event
117  * protocol that is supported by the sender.  When a "syn" event is received,
118  * the receiving FMD will use the minimum of this version and its version of
119  * the protocol, and reply with this new minimum version in the "ack" event.
120  * The receiver will then use this new minimum for subsequent event semantics.
121  */
122 
123 #include <sys/fm/protocol.h>
124 #include <strings.h>
125 #include <limits.h>
126 
127 #include <fmd_alloc.h>
128 #include <fmd_error.h>
129 #include <fmd_conf.h>
130 #include <fmd_subr.h>
131 #include <fmd_string.h>
132 #include <fmd_protocol.h>
133 #include <fmd_thread.h>
134 #include <fmd_eventq.h>
135 #include <fmd_dispq.h>
136 #include <fmd_ctl.h>
137 #include <fmd_log.h>
138 #include <fmd_ustat.h>
139 #include <fmd_case.h>
140 #include <fmd_api.h>
141 #include <fmd_fmri.h>
142 #include <fmd_asru.h>
143 #include <fmd_xprt.h>
144 
145 #include <fmd.h>
146 
147 /*
148  * The states shown above in the transport state machine diagram are encoded
149  * using arrays of class patterns and a corresponding action function.  These
150  * arrays are then passed to fmd_xprt_transition() to change transport states.
151  */
152 
153 const fmd_xprt_rule_t _fmd_xprt_state_syn[] = {
154 { "resource.fm.xprt.syn", fmd_xprt_event_syn },
155 { "*", fmd_xprt_event_error },
156 { NULL, NULL }
157 };
158 
159 const fmd_xprt_rule_t _fmd_xprt_state_ack[] = {
160 { "resource.fm.xprt.ack", fmd_xprt_event_ack },
161 { "*", fmd_xprt_event_error },
162 };
163 
164 const fmd_xprt_rule_t _fmd_xprt_state_err[] = {
165 { "*", fmd_xprt_event_drop },
166 { NULL, NULL }
167 };
168 
169 const fmd_xprt_rule_t _fmd_xprt_state_sub[] = {
170 { "resource.fm.xprt.subscribe", fmd_xprt_event_sub },
171 { "resource.fm.xprt.run", fmd_xprt_event_run },
172 { "resource.fm.xprt.*", fmd_xprt_event_error },
173 { "*", fmd_xprt_event_drop },
174 { NULL, NULL }
175 };
176 
177 const fmd_xprt_rule_t _fmd_xprt_state_run[] = {
178 { "resource.fm.xprt.subscribe", fmd_xprt_event_sub },
179 { "resource.fm.xprt.unsubscribe", fmd_xprt_event_unsub },
180 { "resource.fm.xprt.unsuback", fmd_xprt_event_unsuback },
181 { "resource.fm.xprt.uuclose", fmd_xprt_event_uuclose },
182 { "resource.fm.xprt.*", fmd_xprt_event_error },
183 { NULL, NULL }
184 };
185 
186 /*
187  * Template for per-transport statistics installed by fmd on behalf of each
188  * transport.  These are used to initialize the per-transport xi_stats.  For
189  * each statistic, the name is prepended with "fmd.xprt.%u", where %u is the
190  * transport ID (xi_id) and then are inserted into the per-module stats hash.
191  * The values in this array must match fmd_xprt_stat_t from <fmd_xprt.h>.
192  */
193 static const fmd_xprt_stat_t _fmd_xprt_stat_tmpl = {
194 {
195 { "dispatched", FMD_TYPE_UINT64, "total events dispatched to transport" },
196 { "dequeued", FMD_TYPE_UINT64, "total events dequeued by transport" },
197 { "prdequeued", FMD_TYPE_UINT64, "protocol events dequeued by transport" },
198 { "dropped", FMD_TYPE_UINT64, "total events dropped on queue overflow" },
199 { "wcnt", FMD_TYPE_UINT32, "count of events waiting on queue" },
200 { "wtime", FMD_TYPE_TIME, "total wait time on queue" },
201 { "wlentime", FMD_TYPE_TIME, "total wait length * time product" },
202 { "wlastupdate", FMD_TYPE_TIME, "hrtime of last wait queue update" },
203 { "dtime", FMD_TYPE_TIME, "total processing time after dequeue" },
204 { "dlastupdate", FMD_TYPE_TIME, "hrtime of last event dequeue completion" },
205 },
206 { "module", FMD_TYPE_STRING, "module that owns this transport" },
207 { "authority", FMD_TYPE_STRING, "authority associated with this transport" },
208 { "state", FMD_TYPE_STRING, "current transport state" },
209 { "received", FMD_TYPE_UINT64, "events received by transport" },
210 { "discarded", FMD_TYPE_UINT64, "bad events discarded by transport" },
211 { "retried", FMD_TYPE_UINT64, "retries requested of transport" },
212 { "replayed", FMD_TYPE_UINT64, "events replayed by transport" },
213 { "lost", FMD_TYPE_UINT64, "events lost by transport" },
214 { "timeouts", FMD_TYPE_UINT64, "events received by transport with ttl=0" },
215 { "subscriptions", FMD_TYPE_UINT64, "subscriptions registered to transport" },
216 };
217 
218 static void
219 fmd_xprt_class_hash_create(fmd_xprt_class_hash_t *xch, fmd_eventq_t *eq)
220 {
221 	uint_t hashlen = fmd.d_str_buckets;
222 
223 	xch->xch_queue = eq;
224 	xch->xch_hashlen = hashlen;
225 	xch->xch_hash = fmd_zalloc(sizeof (void *) * hashlen, FMD_SLEEP);
226 }
227 
228 static void
229 fmd_xprt_class_hash_destroy(fmd_xprt_class_hash_t *xch)
230 {
231 	fmd_eventq_t *eq = xch->xch_queue;
232 	fmd_xprt_class_t *xcp, *ncp;
233 	uint_t i;
234 
235 	for (i = 0; i < xch->xch_hashlen; i++) {
236 		for (xcp = xch->xch_hash[i]; xcp != NULL; xcp = ncp) {
237 			ncp = xcp->xc_next;
238 
239 			if (eq != NULL)
240 				fmd_dispq_delete(fmd.d_disp, eq, xcp->xc_class);
241 
242 			fmd_strfree(xcp->xc_class);
243 			fmd_free(xcp, sizeof (fmd_xprt_class_t));
244 		}
245 	}
246 
247 	fmd_free(xch->xch_hash, sizeof (void *) * xch->xch_hashlen);
248 }
249 
250 /*
251  * Insert the specified class into the specified class hash, and return the
252  * reference count.  A return value of one indicates this is the first insert.
253  * If an eventq is associated with the hash, insert a dispq subscription for it.
254  */
255 static uint_t
256 fmd_xprt_class_hash_insert(fmd_xprt_impl_t *xip,
257     fmd_xprt_class_hash_t *xch, const char *class)
258 {
259 	uint_t h = fmd_strhash(class) % xch->xch_hashlen;
260 	fmd_xprt_class_t *xcp;
261 
262 	ASSERT(MUTEX_HELD(&xip->xi_lock));
263 
264 	for (xcp = xch->xch_hash[h]; xcp != NULL; xcp = xcp->xc_next) {
265 		if (strcmp(class, xcp->xc_class) == 0)
266 			return (++xcp->xc_refs);
267 	}
268 
269 	xcp = fmd_alloc(sizeof (fmd_xprt_class_t), FMD_SLEEP);
270 	xcp->xc_class = fmd_strdup(class, FMD_SLEEP);
271 	xcp->xc_next = xch->xch_hash[h];
272 	xcp->xc_refs = 1;
273 	xch->xch_hash[h] = xcp;
274 
275 	if (xch->xch_queue != NULL)
276 		fmd_dispq_insert(fmd.d_disp, xch->xch_queue, class);
277 
278 	return (xcp->xc_refs);
279 }
280 
281 /*
282  * Delete the specified class from the specified class hash, and return the
283  * reference count.  A return value of zero indicates the class was deleted.
284  * If an eventq is associated with the hash, delete the dispq subscription.
285  */
286 static uint_t
287 fmd_xprt_class_hash_delete(fmd_xprt_impl_t *xip,
288     fmd_xprt_class_hash_t *xch, const char *class)
289 {
290 	uint_t h = fmd_strhash(class) % xch->xch_hashlen;
291 	fmd_xprt_class_t *xcp, **pp;
292 
293 	ASSERT(MUTEX_HELD(&xip->xi_lock));
294 	pp = &xch->xch_hash[h];
295 
296 	for (xcp = *pp; xcp != NULL; xcp = xcp->xc_next) {
297 		if (strcmp(class, xcp->xc_class) == 0)
298 			break;
299 		else
300 			pp = &xcp->xc_next;
301 	}
302 
303 	if (xcp == NULL)
304 		return (-1U); /* explicitly permit an invalid delete */
305 
306 	if (--xcp->xc_refs != 0)
307 		return (xcp->xc_refs);
308 
309 	ASSERT(xcp->xc_refs == 0);
310 	*pp = xcp->xc_next;
311 
312 	fmd_strfree(xcp->xc_class);
313 	fmd_free(xcp, sizeof (fmd_xprt_class_t));
314 
315 	if (xch->xch_queue != NULL)
316 		fmd_dispq_delete(fmd.d_disp, xch->xch_queue, class);
317 
318 	return (0);
319 }
320 
321 /*
322  * Queue subscribe events for the specified transport corresponding to all of
323  * the active module subscriptions.  This is an extremely heavyweight operation
324  * that we expect to take place rarely (i.e. when loading a transport module
325  * or when it establishes a connection).  We lock all of the known modules to
326  * prevent them from adding or deleting subscriptions, then snapshot their
327  * subscriptions, and then unlock all of the modules.  We hold the modhash
328  * lock for the duration of this operation to prevent new modules from loading.
329  */
330 static void
331 fmd_xprt_subscribe_modhash(fmd_xprt_impl_t *xip, fmd_modhash_t *mhp)
332 {
333 	fmd_xprt_t *xp = (fmd_xprt_t *)xip;
334 	const fmd_conf_path_t *pap;
335 	fmd_module_t *mp;
336 	uint_t i, j;
337 
338 	(void) pthread_rwlock_rdlock(&mhp->mh_lock);
339 
340 	for (i = 0; i < mhp->mh_hashlen; i++) {
341 		for (mp = mhp->mh_hash[i]; mp != NULL; mp = mp->mod_next)
342 			fmd_module_lock(mp);
343 	}
344 
345 	(void) pthread_mutex_lock(&xip->xi_lock);
346 	ASSERT(!(xip->xi_flags & FMD_XPRT_SUBSCRIBER));
347 	xip->xi_flags |= FMD_XPRT_SUBSCRIBER;
348 	(void) pthread_mutex_unlock(&xip->xi_lock);
349 
350 	for (i = 0; i < mhp->mh_hashlen; i++) {
351 		for (mp = mhp->mh_hash[i]; mp != NULL; mp = mp->mod_next) {
352 			(void) fmd_conf_getprop(mp->mod_conf,
353 			    FMD_PROP_SUBSCRIPTIONS, &pap);
354 			for (j = 0; j < pap->cpa_argc; j++)
355 				fmd_xprt_subscribe(xp, pap->cpa_argv[j]);
356 		}
357 	}
358 
359 	for (i = 0; i < mhp->mh_hashlen; i++) {
360 		for (mp = mhp->mh_hash[i]; mp != NULL; mp = mp->mod_next)
361 			fmd_module_unlock(mp);
362 	}
363 
364 	(void) pthread_rwlock_unlock(&mhp->mh_lock);
365 }
366 
367 static void
368 fmd_xprt_transition(fmd_xprt_impl_t *xip,
369     const fmd_xprt_rule_t *state, const char *tag)
370 {
371 	fmd_event_t *e;
372 	nvlist_t *nvl;
373 	char *s;
374 
375 	TRACE((FMD_DBG_XPRT, "xprt %u -> %s\n", xip->xi_id, tag));
376 
377 	xip->xi_state = state;
378 	s = fmd_strdup(tag, FMD_SLEEP);
379 
380 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
381 	fmd_strfree(xip->xi_stats->xs_state.fmds_value.str);
382 	xip->xi_stats->xs_state.fmds_value.str = s;
383 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
384 
385 	/*
386 	 * If we've reached the SUB state, take out the big hammer and snapshot
387 	 * all of the subscriptions of all of the loaded modules.  Then queue a
388 	 * run event for our remote peer indicating that it can enter RUN.
389 	 */
390 	if (state == _fmd_xprt_state_sub) {
391 		fmd_xprt_subscribe_modhash(xip, fmd.d_mod_hash);
392 
393 		nvl = fmd_protocol_xprt_ctl(xip->xi_queue->eq_mod,
394 		    "resource.fm.xprt.run", xip->xi_version);
395 
396 		(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
397 		e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
398 		fmd_eventq_insert_at_time(xip->xi_queue, e);
399 	}
400 }
401 
402 static void
403 fmd_xprt_authupdate(fmd_xprt_impl_t *xip)
404 {
405 	char *s = fmd_fmri_auth2str(xip->xi_auth);
406 
407 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
408 	fmd_strfree(xip->xi_stats->xs_authority.fmds_value.str);
409 	xip->xi_stats->xs_authority.fmds_value.str = s;
410 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
411 }
412 
413 static int
414 fmd_xprt_vmismatch(fmd_xprt_impl_t *xip, nvlist_t *nvl, uint_t *rversionp)
415 {
416 	uint8_t rversion;
417 
418 	if (nvlist_lookup_uint8(nvl, FM_VERSION, &rversion) != 0) {
419 		(void) pthread_mutex_lock(&xip->xi_stats_lock);
420 		xip->xi_stats->xs_discarded.fmds_value.ui64++;
421 		(void) pthread_mutex_unlock(&xip->xi_stats_lock);
422 
423 		fmd_xprt_transition(xip, _fmd_xprt_state_err, "ERR");
424 		return (1);
425 	}
426 
427 	if (rversion > xip->xi_version) {
428 		fmd_dprintf(FMD_DBG_XPRT, "xprt %u protocol mismatch: %u>%u\n",
429 		    xip->xi_id, rversion, xip->xi_version);
430 
431 		(void) pthread_mutex_lock(&xip->xi_stats_lock);
432 		xip->xi_stats->xs_discarded.fmds_value.ui64++;
433 		(void) pthread_mutex_unlock(&xip->xi_stats_lock);
434 
435 		fmd_xprt_transition(xip, _fmd_xprt_state_err, "ERR");
436 		return (1);
437 	}
438 
439 	if (rversionp != NULL)
440 		*rversionp = rversion;
441 
442 	return (0);
443 }
444 
445 void
446 fmd_xprt_event_syn(fmd_xprt_impl_t *xip, nvlist_t *nvl)
447 {
448 	fmd_event_t *e;
449 	uint_t vers;
450 	char *class;
451 
452 	if (fmd_xprt_vmismatch(xip, nvl, &vers))
453 		return; /* transitioned to error state */
454 
455 	/*
456 	 * If the transport module didn't specify an authority, extract the
457 	 * one that is passed along with the xprt.syn event and use that.
458 	 */
459 	if (xip->xi_auth == NULL &&
460 	    nvlist_lookup_nvlist(nvl, FM_RSRC_RESOURCE, &nvl) == 0 &&
461 	    nvlist_lookup_nvlist(nvl, FM_FMRI_AUTHORITY, &nvl) == 0) {
462 		(void) nvlist_xdup(nvl, &xip->xi_auth, &fmd.d_nva);
463 		fmd_xprt_authupdate(xip);
464 	}
465 
466 	nvl = fmd_protocol_xprt_ctl(xip->xi_queue->eq_mod,
467 	    "resource.fm.xprt.ack", xip->xi_version);
468 
469 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
470 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, class);
471 	fmd_eventq_insert_at_time(xip->xi_queue, e);
472 
473 	xip->xi_version = MIN(FM_RSRC_XPRT_VERSION, vers);
474 	fmd_xprt_transition(xip, _fmd_xprt_state_sub, "SUB");
475 }
476 
477 void
478 fmd_xprt_event_ack(fmd_xprt_impl_t *xip, nvlist_t *nvl)
479 {
480 	uint_t vers;
481 
482 	if (fmd_xprt_vmismatch(xip, nvl, &vers))
483 		return; /* transitioned to error state */
484 
485 	/*
486 	 * If the transport module didn't specify an authority, extract the
487 	 * one that is passed along with the xprt.syn event and use that.
488 	 */
489 	if (xip->xi_auth == NULL &&
490 	    nvlist_lookup_nvlist(nvl, FM_RSRC_RESOURCE, &nvl) == 0 &&
491 	    nvlist_lookup_nvlist(nvl, FM_FMRI_AUTHORITY, &nvl) == 0) {
492 		(void) nvlist_xdup(nvl, &xip->xi_auth, &fmd.d_nva);
493 		fmd_xprt_authupdate(xip);
494 	}
495 
496 	xip->xi_version = MIN(FM_RSRC_XPRT_VERSION, vers);
497 	fmd_xprt_transition(xip, _fmd_xprt_state_sub, "SUB");
498 }
499 
500 /*
501  * Upon transition to RUN, we take every solved case and resend a list.suspect
502  * event for it to our remote peer.  If a case transitions from solved to a
503  * future state (CLOSE_WAIT, CLOSED, or REPAIRED) while we are iterating over
504  * the case hash, we will get it as part of examining the resource cache, next.
505  */
506 static void
507 fmd_xprt_send_case(fmd_case_t *cp, void *arg)
508 {
509 	fmd_case_impl_t *cip = (fmd_case_impl_t *)cp;
510 	fmd_xprt_impl_t *xip = arg;
511 
512 	fmd_event_t *e;
513 	nvlist_t *nvl;
514 	char *class;
515 
516 	if (cip->ci_state != FMD_CASE_SOLVED)
517 		return; /* unsolved, or we'll get it during the ASRU pass */
518 
519 	nvl = fmd_case_mkevent(cp, FM_LIST_SUSPECT_CLASS);
520 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
521 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, class);
522 
523 	fmd_dprintf(FMD_DBG_XPRT, "re-send %s for %s to transport %u\n",
524 	    FM_LIST_SUSPECT_CLASS, cip->ci_uuid, xip->xi_id);
525 
526 	fmd_dispq_dispatch_gid(fmd.d_disp, e, class, xip->xi_queue->eq_sgid);
527 }
528 
529 /*
530  * Upon transition to RUN, we take every ASRU which is in the degraded state
531  * and resend a fault.* event for it to our remote peer, in case the peer is
532  * running in the fault manager that knows how to disable this resource.  If
533  * any new resources are added to the cache during our iteration, this is no
534  * problem because our subscriptions are already proxied and so any new cases
535  * will result in a list.suspect event being transported if that is needed.
536  */
537 static void
538 fmd_xprt_send_asru(fmd_asru_t *ap, void *arg)
539 {
540 	fmd_xprt_impl_t *xip = arg;
541 	nvlist_t *nvl = NULL;
542 	fmd_event_t *e;
543 	char *class;
544 
545 	(void) pthread_mutex_lock(&ap->asru_lock);
546 
547 	if ((ap->asru_flags & (FMD_ASRU_INTERNAL | FMD_ASRU_STATE)) ==
548 	    FMD_ASRU_FAULTY && fmd_case_orphaned(ap->asru_case))
549 		(void) nvlist_xdup(ap->asru_event, &nvl, &fmd.d_nva);
550 
551 	(void) pthread_mutex_unlock(&ap->asru_lock);
552 
553 	if (nvl == NULL)
554 		return; /* asru is internal, unusable, or not faulty */
555 
556 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
557 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, class);
558 
559 	fmd_dprintf(FMD_DBG_XPRT, "re-send %s for %s to transport %u\n",
560 	    class, ap->asru_name, xip->xi_id);
561 
562 	fmd_dispq_dispatch_gid(fmd.d_disp, e, class, xip->xi_queue->eq_sgid);
563 }
564 
565 void
566 fmd_xprt_event_run(fmd_xprt_impl_t *xip, nvlist_t *nvl)
567 {
568 	if (!fmd_xprt_vmismatch(xip, nvl, NULL)) {
569 		fmd_xprt_transition(xip, _fmd_xprt_state_run, "RUN");
570 		fmd_case_hash_apply(fmd.d_cases, fmd_xprt_send_case, xip);
571 		fmd_asru_hash_apply(fmd.d_asrus, fmd_xprt_send_asru, xip);
572 	}
573 }
574 
575 void
576 fmd_xprt_event_sub(fmd_xprt_impl_t *xip, nvlist_t *nvl)
577 {
578 	char *class;
579 
580 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
581 		return; /* transitioned to error state */
582 
583 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_SUBCLASS, &class) != 0)
584 		return; /* malformed protocol event */
585 
586 	(void) pthread_mutex_lock(&xip->xi_lock);
587 	(void) fmd_xprt_class_hash_insert(xip, &xip->xi_lsub, class);
588 	(void) pthread_mutex_unlock(&xip->xi_lock);
589 
590 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
591 	xip->xi_stats->xs_subscriptions.fmds_value.ui64++;
592 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
593 }
594 
595 void
596 fmd_xprt_event_unsub(fmd_xprt_impl_t *xip, nvlist_t *nvl)
597 {
598 	fmd_event_t *e;
599 	char *class;
600 
601 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
602 		return; /* transitioned to error state */
603 
604 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_SUBCLASS, &class) != 0)
605 		return; /* malformed protocol event */
606 
607 	(void) pthread_mutex_lock(&xip->xi_lock);
608 	(void) fmd_xprt_class_hash_delete(xip, &xip->xi_lsub, class);
609 	(void) pthread_mutex_unlock(&xip->xi_lock);
610 
611 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
612 	xip->xi_stats->xs_subscriptions.fmds_value.ui64--;
613 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
614 
615 	nvl = fmd_protocol_xprt_sub(xip->xi_queue->eq_mod,
616 	    "resource.fm.xprt.unsuback", xip->xi_version, class);
617 
618 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
619 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, class);
620 	fmd_eventq_insert_at_time(xip->xi_queue, e);
621 }
622 
623 void
624 fmd_xprt_event_unsuback(fmd_xprt_impl_t *xip, nvlist_t *nvl)
625 {
626 	char *class;
627 
628 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
629 		return; /* transitioned to error state */
630 
631 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_SUBCLASS, &class) != 0)
632 		return; /* malformed protocol event */
633 
634 	(void) pthread_mutex_lock(&xip->xi_lock);
635 	(void) fmd_xprt_class_hash_delete(xip, &xip->xi_usub, class);
636 	(void) pthread_mutex_unlock(&xip->xi_lock);
637 }
638 
639 void
640 fmd_xprt_event_uuclose(fmd_xprt_impl_t *xip, nvlist_t *nvl)
641 {
642 	fmd_case_t *cp;
643 	char *uuid;
644 
645 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
646 		return; /* transitioned to error state */
647 
648 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_UUID, &uuid) == 0 &&
649 	    (cp = fmd_case_hash_lookup(fmd.d_cases, uuid)) != NULL) {
650 		fmd_case_transition(cp, FMD_CASE_CLOSE_WAIT, FMD_CF_ISOLATED);
651 		fmd_case_rele(cp);
652 	}
653 }
654 
655 void
656 fmd_xprt_event_error(fmd_xprt_impl_t *xip, nvlist_t *nvl)
657 {
658 	char *class = "<unknown>";
659 
660 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
661 	xip->xi_stats->xs_discarded.fmds_value.ui64++;
662 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
663 
664 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
665 	TRACE((FMD_DBG_XPRT, "xprt %u bad event %s\n", xip->xi_id, class));
666 
667 	fmd_xprt_transition(xip, _fmd_xprt_state_err, "ERR");
668 }
669 
670 void
671 fmd_xprt_event_drop(fmd_xprt_impl_t *xip, nvlist_t *nvl)
672 {
673 	char *class = "<unknown>";
674 
675 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
676 	xip->xi_stats->xs_discarded.fmds_value.ui64++;
677 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
678 
679 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
680 	TRACE((FMD_DBG_XPRT, "xprt %u drop event %s\n", xip->xi_id, class));
681 
682 }
683 
684 fmd_xprt_t *
685 fmd_xprt_create(fmd_module_t *mp, uint_t flags, nvlist_t *auth, void *data)
686 {
687 	fmd_xprt_impl_t *xip = fmd_zalloc(sizeof (fmd_xprt_impl_t), FMD_SLEEP);
688 	fmd_stat_t *statv;
689 	uint_t i, statc;
690 
691 	char buf[PATH_MAX];
692 	fmd_event_t *e;
693 	nvlist_t *nvl;
694 	char *s;
695 
696 	(void) pthread_mutex_init(&xip->xi_lock, NULL);
697 	(void) pthread_cond_init(&xip->xi_cv, NULL);
698 	(void) pthread_mutex_init(&xip->xi_stats_lock, NULL);
699 
700 	xip->xi_auth = auth;
701 	xip->xi_data = data;
702 	xip->xi_version = FM_RSRC_XPRT_VERSION;
703 	xip->xi_flags = flags;
704 
705 	/*
706 	 * Grab fmd.d_xprt_lock to block fmd_xprt_suspend_all() and then create
707 	 * a transport ID and make it visible in fmd.d_xprt_ids.  If transports
708 	 * were previously suspended, set the FMD_XPRT_DSUSPENDED flag on us to
709 	 * ensure that this transport will not run until fmd_xprt_resume_all().
710 	 */
711 	(void) pthread_mutex_lock(&fmd.d_xprt_lock);
712 	xip->xi_id = fmd_idspace_alloc(fmd.d_xprt_ids, xip);
713 
714 	if (fmd.d_xprt_suspend != 0)
715 		xip->xi_flags |= FMD_XPRT_DSUSPENDED;
716 
717 	(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
718 
719 	/*
720 	 * If the module has not yet finished _fmd_init(), set the ISUSPENDED
721 	 * bit so that fmdo_send() is not called until _fmd_init() completes.
722 	 */
723 	if (!(mp->mod_flags & FMD_MOD_INIT))
724 		xip->xi_flags |= FMD_XPRT_ISUSPENDED;
725 
726 	/*
727 	 * Initialize the transport statistics that we keep on behalf of fmd.
728 	 * These are set up using a template defined at the top of this file.
729 	 * We rename each statistic with a prefix ensuring its uniqueness.
730 	 */
731 	statc = sizeof (_fmd_xprt_stat_tmpl) / sizeof (fmd_stat_t);
732 	statv = fmd_alloc(sizeof (_fmd_xprt_stat_tmpl), FMD_SLEEP);
733 	bcopy(&_fmd_xprt_stat_tmpl, statv, sizeof (_fmd_xprt_stat_tmpl));
734 
735 	for (i = 0; i < statc; i++) {
736 		(void) snprintf(statv[i].fmds_name,
737 		    sizeof (statv[i].fmds_name), "fmd.xprt.%u.%s", xip->xi_id,
738 		    ((fmd_stat_t *)&_fmd_xprt_stat_tmpl + i)->fmds_name);
739 	}
740 
741 	xip->xi_stats = (fmd_xprt_stat_t *)fmd_ustat_insert(
742 	    mp->mod_ustat, FMD_USTAT_NOALLOC, statc, statv, NULL);
743 
744 	if (xip->xi_stats == NULL)
745 		fmd_panic("failed to create xi_stats (%p)\n", (void *)statv);
746 
747 	xip->xi_stats->xs_module.fmds_value.str =
748 	    fmd_strdup(mp->mod_name, FMD_SLEEP);
749 
750 	if (xip->xi_auth != NULL)
751 		fmd_xprt_authupdate(xip);
752 
753 	/*
754 	 * Create the outbound eventq for this transport and link to its stats.
755 	 * If any suspend bits were set above, suspend the eventq immediately.
756 	 */
757 	xip->xi_queue = fmd_eventq_create(mp, &xip->xi_stats->xs_evqstat,
758 	    &xip->xi_stats_lock, mp->mod_stats->ms_xprtqlimit.fmds_value.ui32);
759 
760 	if (xip->xi_flags & FMD_XPRT_SMASK)
761 		fmd_eventq_suspend(xip->xi_queue);
762 
763 	/*
764 	 * Create our subscription hashes: local subscriptions go to xi_queue,
765 	 * remote subscriptions are tracked only for protocol requests, and
766 	 * pending unsubscriptions are associated with the /dev/null eventq.
767 	 */
768 	fmd_xprt_class_hash_create(&xip->xi_lsub, xip->xi_queue);
769 	fmd_xprt_class_hash_create(&xip->xi_rsub, NULL);
770 	fmd_xprt_class_hash_create(&xip->xi_usub, fmd.d_rmod->mod_queue);
771 
772 	/*
773 	 * Determine our initial state based upon the creation flags.  If we're
774 	 * read-only, go directly to RUN.  If we're accepting a new connection,
775 	 * wait for a SYN.  Otherwise send a SYN and wait for an ACK.
776 	 */
777 	if ((flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
778 		fmd_xprt_transition(xip, _fmd_xprt_state_run, "RUN");
779 	else if (flags & FMD_XPRT_ACCEPT)
780 		fmd_xprt_transition(xip, _fmd_xprt_state_syn, "SYN");
781 	else
782 		fmd_xprt_transition(xip, _fmd_xprt_state_ack, "ACK");
783 
784 	/*
785 	 * If client.xprtlog is set to TRUE, create a debugging log for the
786 	 * events received by the transport in var/fm/fmd/xprt/.
787 	 */
788 	(void) fmd_conf_getprop(fmd.d_conf, "client.xprtlog", &i);
789 	(void) fmd_conf_getprop(fmd.d_conf, "log.xprt", &s);
790 
791 	if (i) {
792 		(void) snprintf(buf, sizeof (buf), "%s/%u.log", s, xip->xi_id);
793 		xip->xi_log = fmd_log_open(fmd.d_rootdir, buf, FMD_LOG_XPRT);
794 	}
795 
796 	ASSERT(fmd_module_locked(mp));
797 	fmd_list_append(&mp->mod_transports, xip);
798 
799 	(void) pthread_mutex_lock(&mp->mod_stats_lock);
800 	mp->mod_stats->ms_xprtopen.fmds_value.ui32++;
801 	(void) pthread_mutex_unlock(&mp->mod_stats_lock);
802 
803 	/*
804 	 * If this is a read-only transport, return without creating a send
805 	 * queue thread and setting up any connection events in our queue.
806 	 */
807 	if ((flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
808 		goto out;
809 
810 	/*
811 	 * Once the transport is fully initialized, create a send queue thread
812 	 * and start any connect events flowing to complete our initialization.
813 	 */
814 	if ((xip->xi_thread = fmd_thread_create(mp,
815 	    (fmd_thread_f *)fmd_xprt_send, xip)) == NULL) {
816 
817 		fmd_error(EFMD_XPRT_THR,
818 		    "failed to create thread for transport %u", xip->xi_id);
819 
820 		fmd_xprt_destroy((fmd_xprt_t *)xip);
821 		(void) fmd_set_errno(EFMD_XPRT_THR);
822 		return (NULL);
823 	}
824 
825 	/*
826 	 * If the transport is not being opened to accept an inbound connect,
827 	 * start an outbound connection by enqueuing a SYN event for our peer.
828 	 */
829 	if (!(flags & FMD_XPRT_ACCEPT)) {
830 		nvl = fmd_protocol_xprt_ctl(mp,
831 		    "resource.fm.xprt.syn", FM_RSRC_XPRT_VERSION);
832 
833 		(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
834 		e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
835 		fmd_eventq_insert_at_time(xip->xi_queue, e);
836 	}
837 out:
838 	fmd_dprintf(FMD_DBG_XPRT, "opened transport %u\n", xip->xi_id);
839 	return ((fmd_xprt_t *)xip);
840 }
841 
842 void
843 fmd_xprt_destroy(fmd_xprt_t *xp)
844 {
845 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
846 	fmd_module_t *mp = xip->xi_queue->eq_mod;
847 	uint_t id = xip->xi_id;
848 
849 	fmd_case_impl_t *cip, *nip;
850 	fmd_stat_t *sp;
851 	uint_t i, n;
852 
853 	ASSERT(fmd_module_locked(mp));
854 	fmd_list_delete(&mp->mod_transports, xip);
855 
856 	(void) pthread_mutex_lock(&mp->mod_stats_lock);
857 	mp->mod_stats->ms_xprtopen.fmds_value.ui32--;
858 	(void) pthread_mutex_unlock(&mp->mod_stats_lock);
859 
860 	(void) pthread_mutex_lock(&xip->xi_lock);
861 
862 	while (xip->xi_busy != 0)
863 		(void) pthread_cond_wait(&xip->xi_cv, &xip->xi_lock);
864 
865 	/*
866 	 * Remove the transport from global visibility, cancel its send-side
867 	 * thread, join with it, and then remove the transport from module
868 	 * visibility.  Once all this is done, destroy and free the transport.
869 	 */
870 	(void) fmd_idspace_free(fmd.d_xprt_ids, xip->xi_id);
871 
872 	if (xip->xi_thread != NULL) {
873 		fmd_eventq_abort(xip->xi_queue);
874 		fmd_thread_destroy(xip->xi_thread, FMD_THREAD_JOIN);
875 	}
876 
877 	if (xip->xi_log != NULL)
878 		fmd_log_rele(xip->xi_log);
879 
880 	/*
881 	 * Release every case handle in the module that was cached by this
882 	 * transport.  This will result in these cases disappearing from the
883 	 * local case hash so that fmd_case_uuclose() can no longer be used.
884 	 */
885 	for (cip = fmd_list_next(&mp->mod_cases); cip != NULL; cip = nip) {
886 		nip = fmd_list_next(cip);
887 		if (cip->ci_xprt == xp)
888 			fmd_case_discard((fmd_case_t *)cip);
889 	}
890 
891 	/*
892 	 * Destroy every class in the various subscription hashes and remove
893 	 * any corresponding subscriptions from the event dispatch queue.
894 	 */
895 	fmd_xprt_class_hash_destroy(&xip->xi_lsub);
896 	fmd_xprt_class_hash_destroy(&xip->xi_rsub);
897 	fmd_xprt_class_hash_destroy(&xip->xi_usub);
898 
899 	/*
900 	 * Since our statistics are created by hand, after deleting them from
901 	 * the ustat hash we must manually free them and any embedded strings.
902 	 */
903 	fmd_ustat_delete(mp->mod_ustat, sizeof (_fmd_xprt_stat_tmpl) /
904 	    sizeof (fmd_stat_t), (fmd_stat_t *)&_fmd_xprt_stat_tmpl);
905 
906 	sp = (fmd_stat_t *)xip->xi_stats;
907 	n = sizeof (fmd_xprt_stat_t) / sizeof (fmd_stat_t);
908 
909 	for (i = 0; i < n; i++, sp++) {
910 		if (sp->fmds_type == FMD_TYPE_STRING)
911 			fmd_strfree(sp->fmds_value.str);
912 	}
913 
914 	fmd_free(xip->xi_stats, sizeof (fmd_xprt_stat_t));
915 	fmd_eventq_destroy(xip->xi_queue);
916 	nvlist_free(xip->xi_auth);
917 	fmd_free(xip, sizeof (fmd_xprt_impl_t));
918 
919 	fmd_dprintf(FMD_DBG_XPRT, "closed transport %u\n", id);
920 }
921 
922 void
923 fmd_xprt_xsuspend(fmd_xprt_t *xp, uint_t flags)
924 {
925 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
926 	uint_t oflags;
927 
928 	ASSERT((flags & ~FMD_XPRT_SMASK) == 0);
929 	(void) pthread_mutex_lock(&xip->xi_lock);
930 
931 	oflags = xip->xi_flags;
932 	xip->xi_flags |= flags;
933 
934 	if (!(oflags & FMD_XPRT_SMASK) && (xip->xi_flags & FMD_XPRT_SMASK) != 0)
935 		fmd_eventq_suspend(xip->xi_queue);
936 
937 	(void) pthread_cond_broadcast(&xip->xi_cv);
938 
939 	while (xip->xi_busy != 0)
940 		(void) pthread_cond_wait(&xip->xi_cv, &xip->xi_lock);
941 
942 	(void) pthread_mutex_unlock(&xip->xi_lock);
943 }
944 
945 void
946 fmd_xprt_xresume(fmd_xprt_t *xp, uint_t flags)
947 {
948 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
949 	uint_t oflags;
950 
951 	ASSERT((flags & ~FMD_XPRT_SMASK) == 0);
952 	(void) pthread_mutex_lock(&xip->xi_lock);
953 
954 	oflags = xip->xi_flags;
955 	xip->xi_flags &= ~flags;
956 
957 	if ((oflags & FMD_XPRT_SMASK) != 0 && !(xip->xi_flags & FMD_XPRT_SMASK))
958 		fmd_eventq_resume(xip->xi_queue);
959 
960 	(void) pthread_cond_broadcast(&xip->xi_cv);
961 	(void) pthread_mutex_unlock(&xip->xi_lock);
962 }
963 
964 void
965 fmd_xprt_send(fmd_xprt_t *xp)
966 {
967 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
968 	fmd_module_t *mp = xip->xi_queue->eq_mod;
969 	fmd_event_t *ep;
970 	int err;
971 
972 	while ((ep = fmd_eventq_delete(xip->xi_queue)) != NULL) {
973 		if (FMD_EVENT_TTL(ep) == 0) {
974 			fmd_event_rele(ep);
975 			continue;
976 		}
977 
978 		fmd_dprintf(FMD_DBG_XPRT, "xprt %u sending %s\n",
979 		    xip->xi_id, (char *)FMD_EVENT_DATA(ep));
980 
981 		err = mp->mod_ops->mop_transport(mp, xp, ep);
982 		fmd_eventq_done(xip->xi_queue);
983 
984 		if (err == FMD_SEND_RETRY) {
985 			fmd_eventq_insert_at_time(xip->xi_queue, ep);
986 			(void) pthread_mutex_lock(&xip->xi_stats_lock);
987 			xip->xi_stats->xs_retried.fmds_value.ui64++;
988 			(void) pthread_mutex_unlock(&xip->xi_stats_lock);
989 		}
990 
991 		if (err != FMD_SEND_SUCCESS && err != FMD_SEND_RETRY) {
992 			(void) pthread_mutex_lock(&xip->xi_stats_lock);
993 			xip->xi_stats->xs_lost.fmds_value.ui64++;
994 			(void) pthread_mutex_unlock(&xip->xi_stats_lock);
995 		}
996 
997 		fmd_event_rele(ep);
998 	}
999 }
1000 
1001 void
1002 fmd_xprt_recv(fmd_xprt_t *xp, nvlist_t *nvl, hrtime_t hrt)
1003 {
1004 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1005 	const fmd_xprt_rule_t *xrp;
1006 	fmd_t *dp = &fmd;
1007 
1008 	fmd_event_t *e;
1009 	char *class, *uuid, *code;
1010 	int isproto;
1011 
1012 	uint64_t *tod;
1013 	uint8_t ttl;
1014 	uint_t n;
1015 
1016 	/*
1017 	 * Grab the transport lock and set the busy flag to indicate we are
1018 	 * busy receiving an event.  If [DI]SUSPEND is pending, wait until fmd
1019 	 * resumes the transport before continuing on with the receive.
1020 	 */
1021 	(void) pthread_mutex_lock(&xip->xi_lock);
1022 
1023 	while (xip->xi_flags & (FMD_XPRT_DSUSPENDED | FMD_XPRT_ISUSPENDED))
1024 		(void) pthread_cond_wait(&xip->xi_cv, &xip->xi_lock);
1025 
1026 	xip->xi_busy++;
1027 	ASSERT(xip->xi_busy != 0);
1028 
1029 	(void) pthread_mutex_unlock(&xip->xi_lock);
1030 
1031 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
1032 	xip->xi_stats->xs_received.fmds_value.ui64++;
1033 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1034 
1035 	if (nvlist_lookup_string(nvl, FM_CLASS, &class) != 0) {
1036 		fmd_error(EFMD_XPRT_PAYLOAD, "discarding nvlist %p: missing "
1037 		    "required \"%s\" payload element", (void *)nvl, FM_CLASS);
1038 
1039 		(void) pthread_mutex_lock(&xip->xi_stats_lock);
1040 		xip->xi_stats->xs_discarded.fmds_value.ui64++;
1041 		(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1042 
1043 		nvlist_free(nvl);
1044 		goto done;
1045 	}
1046 
1047 	fmd_dprintf(FMD_DBG_XPRT, "xprt %u posting %s\n", xip->xi_id, class);
1048 
1049 	/*
1050 	 * If a time-to-live value is present in the event and is zero, drop
1051 	 * the event and bump xs_timeouts.  Otherwise decrement the TTL value.
1052 	 */
1053 	if (nvlist_lookup_uint8(nvl, FMD_EVN_TTL, &ttl) == 0) {
1054 		if (ttl == 0) {
1055 			fmd_dprintf(FMD_DBG_XPRT, "xprt %u nvlist %p (%s) "
1056 			    "timeout: event received with ttl=0\n",
1057 			    xip->xi_id, (void *)nvl, class);
1058 
1059 			(void) pthread_mutex_lock(&xip->xi_stats_lock);
1060 			xip->xi_stats->xs_timeouts.fmds_value.ui64++;
1061 			(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1062 
1063 			nvlist_free(nvl);
1064 			goto done;
1065 		}
1066 		(void) nvlist_remove(nvl, FMD_EVN_TTL, DATA_TYPE_UINT8);
1067 		(void) nvlist_add_uint8(nvl, FMD_EVN_TTL, ttl - 1);
1068 	}
1069 
1070 	/*
1071 	 * If we are using the native system clock, the underlying transport
1072 	 * code can provide a tighter event time bound by telling us when the
1073 	 * event was enqueued.  If we're using simulated clocks, this time
1074 	 * has no meaning to us, so just reset the value to use HRT_NOW.
1075 	 */
1076 	if (dp->d_clockops != &fmd_timeops_native)
1077 		hrt = FMD_HRT_NOW;
1078 
1079 	/*
1080 	 * If an event's class is in the FMD_CTL_CLASS family, then create a
1081 	 * control event.  If a FMD_EVN_TOD member is found, create a protocol
1082 	 * event using this time.  Otherwise create a protocol event using hrt.
1083 	 */
1084 	if ((isproto = strncmp(class, FMD_CTL_CLASS, FMD_CTL_CLASS_LEN)) == 0)
1085 		e = fmd_event_create(FMD_EVT_CTL, hrt, nvl, fmd_ctl_init(nvl));
1086 	else if (nvlist_lookup_uint64_array(nvl, FMD_EVN_TOD, &tod, &n) != 0)
1087 		e = fmd_event_create(FMD_EVT_PROTOCOL, hrt, nvl, class);
1088 	else {
1089 		e = fmd_event_recreate(FMD_EVT_PROTOCOL,
1090 		    NULL, nvl, class, NULL, 0, 0);
1091 	}
1092 
1093 	/*
1094 	 * If the debug log is enabled, create a temporary event, log it to the
1095 	 * debug log, and then reset the underlying state of the event.
1096 	 */
1097 	if (xip->xi_log != NULL) {
1098 		fmd_event_impl_t *ep = (fmd_event_impl_t *)e;
1099 
1100 		fmd_log_append(xip->xi_log, e, NULL);
1101 
1102 		ep->ev_flags |= FMD_EVF_VOLATILE;
1103 		ep->ev_off = 0;
1104 		ep->ev_len = 0;
1105 
1106 		if (ep->ev_log != NULL) {
1107 			fmd_log_rele(ep->ev_log);
1108 			ep->ev_log = NULL;
1109 		}
1110 	}
1111 
1112 	/*
1113 	 * Iterate over the rules for the current state trying to match the
1114 	 * event class to one of our special rules.  If a rule is matched, the
1115 	 * event is consumed and not dispatched to other modules.  If the rule
1116 	 * set ends without matching an event, we fall through to dispatching.
1117 	 */
1118 	for (xrp = xip->xi_state; xrp->xr_class != NULL; xrp++) {
1119 		if (fmd_event_match(e, FMD_EVT_PROTOCOL, xrp->xr_class)) {
1120 			fmd_event_hold(e);
1121 			xrp->xr_func(xip, nvl);
1122 			fmd_event_rele(e);
1123 			goto done;
1124 		}
1125 	}
1126 
1127 	/*
1128 	 * Record the event in the errlog if it is an ereport.  This code will
1129 	 * be replaced later with a per-transport intent log instead.
1130 	 */
1131 	if (fmd_event_match(e, FMD_EVT_PROTOCOL, FM_EREPORT_CLASS ".*")) {
1132 		(void) pthread_rwlock_rdlock(&dp->d_log_lock);
1133 		fmd_log_append(dp->d_errlog, e, NULL);
1134 		(void) pthread_rwlock_unlock(&dp->d_log_lock);
1135 	}
1136 
1137 	/*
1138 	 * If a list.suspect event is received, create a case for the specified
1139 	 * UUID in the case hash, with the transport module as its owner.  If
1140 	 * the UUID is already known, fmd_case_recreate() will return NULL and
1141 	 * we simply proceed to our normal event handling regardless.
1142 	 */
1143 	if (fmd_event_match(e, FMD_EVT_PROTOCOL, FM_LIST_SUSPECT_CLASS) &&
1144 	    nvlist_lookup_string(nvl, FM_SUSPECT_UUID, &uuid) == 0 &&
1145 	    nvlist_lookup_string(nvl, FM_SUSPECT_DIAG_CODE, &code) == 0) {
1146 		fmd_module_lock(xip->xi_queue->eq_mod);
1147 		(void) fmd_case_recreate(xip->xi_queue->eq_mod,
1148 		    xp, FMD_CASE_SOLVED, uuid, code);
1149 		fmd_module_unlock(xip->xi_queue->eq_mod);
1150 	}
1151 
1152 	if (isproto)
1153 		fmd_dispq_dispatch(dp->d_disp, e, class);
1154 	else
1155 		fmd_modhash_dispatch(dp->d_mod_hash, e);
1156 done:
1157 	(void) pthread_mutex_lock(&xip->xi_lock);
1158 
1159 	ASSERT(xip->xi_busy != 0);
1160 	xip->xi_busy--;
1161 
1162 	(void) pthread_cond_broadcast(&xip->xi_cv);
1163 	(void) pthread_mutex_unlock(&xip->xi_lock);
1164 }
1165 
1166 void
1167 fmd_xprt_uuclose(fmd_xprt_t *xp, const char *uuid)
1168 {
1169 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1170 
1171 	fmd_event_t *e;
1172 	nvlist_t *nvl;
1173 	char *s;
1174 
1175 	fmd_dprintf(FMD_DBG_XPRT,
1176 	    "xprt %u closing case %s\n", xip->xi_id, uuid);
1177 
1178 	nvl = fmd_protocol_xprt_uuclose(xip->xi_queue->eq_mod,
1179 	    "resource.fm.xprt.uuclose", xip->xi_version, uuid);
1180 
1181 	(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
1182 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
1183 	fmd_eventq_insert_at_time(xip->xi_queue, e);
1184 }
1185 
1186 /*
1187  * Insert the specified class into our remote subscription hash.  If the class
1188  * is already present, bump the reference count; otherwise add it to the hash
1189  * and then enqueue an event for our remote peer to proxy our subscription.
1190  */
1191 void
1192 fmd_xprt_subscribe(fmd_xprt_t *xp, const char *class)
1193 {
1194 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1195 
1196 	uint_t refs;
1197 	nvlist_t *nvl;
1198 	fmd_event_t *e;
1199 	char *s;
1200 
1201 	if ((xip->xi_flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
1202 		return; /* read-only transports do not proxy subscriptions */
1203 
1204 	if (!(xip->xi_flags & FMD_XPRT_SUBSCRIBER))
1205 		return; /* transport is not yet an active subscriber */
1206 
1207 	(void) pthread_mutex_lock(&xip->xi_lock);
1208 	refs = fmd_xprt_class_hash_insert(xip, &xip->xi_rsub, class);
1209 	(void) pthread_mutex_unlock(&xip->xi_lock);
1210 
1211 	if (refs > 1)
1212 		return; /* we've already asked our peer for this subscription */
1213 
1214 	fmd_dprintf(FMD_DBG_XPRT,
1215 	    "xprt %u subscribing to %s\n", xip->xi_id, class);
1216 
1217 	nvl = fmd_protocol_xprt_sub(xip->xi_queue->eq_mod,
1218 	    "resource.fm.xprt.subscribe", xip->xi_version, class);
1219 
1220 	(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
1221 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
1222 	fmd_eventq_insert_at_time(xip->xi_queue, e);
1223 }
1224 
1225 /*
1226  * Delete the specified class from the remote subscription hash.  If the
1227  * reference count drops to zero, ask our remote peer to unsubscribe by proxy.
1228  */
1229 void
1230 fmd_xprt_unsubscribe(fmd_xprt_t *xp, const char *class)
1231 {
1232 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1233 
1234 	uint_t refs;
1235 	nvlist_t *nvl;
1236 	fmd_event_t *e;
1237 	char *s;
1238 
1239 	if ((xip->xi_flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
1240 		return; /* read-only transports do not proxy subscriptions */
1241 
1242 	if (!(xip->xi_flags & FMD_XPRT_SUBSCRIBER))
1243 		return; /* transport is not yet an active subscriber */
1244 
1245 	/*
1246 	 * If the subscription reference count drops to zero in xi_rsub, insert
1247 	 * an entry into the xi_usub hash indicating we await an unsuback event.
1248 	 */
1249 	(void) pthread_mutex_lock(&xip->xi_lock);
1250 
1251 	if ((refs = fmd_xprt_class_hash_delete(xip, &xip->xi_rsub, class)) == 0)
1252 		(void) fmd_xprt_class_hash_insert(xip, &xip->xi_usub, class);
1253 
1254 	(void) pthread_mutex_unlock(&xip->xi_lock);
1255 
1256 	if (refs != 0)
1257 		return; /* other subscriptions for this class still active */
1258 
1259 	fmd_dprintf(FMD_DBG_XPRT,
1260 	    "xprt %u unsubscribing from %s\n", xip->xi_id, class);
1261 
1262 	nvl = fmd_protocol_xprt_sub(xip->xi_queue->eq_mod,
1263 	    "resource.fm.xprt.unsubscribe", xip->xi_version, class);
1264 
1265 	(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
1266 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
1267 	fmd_eventq_insert_at_time(xip->xi_queue, e);
1268 }
1269 
1270 static void
1271 fmd_xprt_subscribe_xid(fmd_idspace_t *ids, id_t id, void *class)
1272 {
1273 	fmd_xprt_t *xp;
1274 
1275 	if ((xp = fmd_idspace_hold(ids, id)) != NULL) {
1276 		fmd_xprt_subscribe(xp, class);
1277 		fmd_idspace_rele(ids, id);
1278 	}
1279 }
1280 
1281 void
1282 fmd_xprt_subscribe_all(const char *class)
1283 {
1284 	fmd_idspace_t *ids = fmd.d_xprt_ids;
1285 
1286 	if (ids->ids_count != 0)
1287 		fmd_idspace_apply(ids, fmd_xprt_subscribe_xid, (void *)class);
1288 }
1289 
1290 static void
1291 fmd_xprt_unsubscribe_xid(fmd_idspace_t *ids, id_t id, void *class)
1292 {
1293 	fmd_xprt_t *xp;
1294 
1295 	if ((xp = fmd_idspace_hold(ids, id)) != NULL) {
1296 		fmd_xprt_unsubscribe(xp, class);
1297 		fmd_idspace_rele(ids, id);
1298 	}
1299 }
1300 
1301 void
1302 fmd_xprt_unsubscribe_all(const char *class)
1303 {
1304 	fmd_idspace_t *ids = fmd.d_xprt_ids;
1305 
1306 	if (ids->ids_count != 0)
1307 		fmd_idspace_apply(ids, fmd_xprt_unsubscribe_xid, (void *)class);
1308 }
1309 
1310 /*ARGSUSED*/
1311 static void
1312 fmd_xprt_suspend_xid(fmd_idspace_t *ids, id_t id, void *arg)
1313 {
1314 	fmd_xprt_t *xp;
1315 
1316 	if ((xp = fmd_idspace_hold(ids, id)) != NULL) {
1317 		fmd_xprt_xsuspend(xp, FMD_XPRT_DSUSPENDED);
1318 		fmd_idspace_rele(ids, id);
1319 	}
1320 }
1321 
1322 void
1323 fmd_xprt_suspend_all(void)
1324 {
1325 	fmd_idspace_t *ids = fmd.d_xprt_ids;
1326 
1327 	(void) pthread_mutex_lock(&fmd.d_xprt_lock);
1328 
1329 	if (fmd.d_xprt_suspend++ != 0) {
1330 		(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
1331 		return; /* already suspended */
1332 	}
1333 
1334 	if (ids->ids_count != 0)
1335 		fmd_idspace_apply(ids, fmd_xprt_suspend_xid, NULL);
1336 
1337 	(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
1338 }
1339 
1340 /*ARGSUSED*/
1341 static void
1342 fmd_xprt_resume_xid(fmd_idspace_t *ids, id_t id, void *arg)
1343 {
1344 	fmd_xprt_t *xp;
1345 
1346 	if ((xp = fmd_idspace_hold(ids, id)) != NULL) {
1347 		fmd_xprt_xresume(xp, FMD_XPRT_DSUSPENDED);
1348 		fmd_idspace_rele(ids, id);
1349 	}
1350 }
1351 
1352 void
1353 fmd_xprt_resume_all(void)
1354 {
1355 	fmd_idspace_t *ids = fmd.d_xprt_ids;
1356 
1357 	(void) pthread_mutex_lock(&fmd.d_xprt_lock);
1358 
1359 	if (fmd.d_xprt_suspend == 0)
1360 		fmd_panic("fmd_xprt_suspend/resume_all mismatch\n");
1361 
1362 	if (--fmd.d_xprt_suspend != 0) {
1363 		(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
1364 		return; /* not ready to be resumed */
1365 	}
1366 
1367 	if (ids->ids_count != 0)
1368 		fmd_idspace_apply(ids, fmd_xprt_resume_xid, NULL);
1369 
1370 	(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
1371 }
1372