xref: /illumos-gate/usr/src/cmd/fm/fmd/common/fmd_xprt.c (revision 7e934d3acc051b7ee3ef0d11571fd1225800a607)
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 /*
23  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
24  */
25 
26 /*
27  * FMD Transport Subsystem
28  *
29  * A transport module uses some underlying mechanism to transport events.
30  * This mechanism may use any underlying link-layer protocol and may support
31  * additional link-layer packets unrelated to FMA.  Some appropriate link-
32  * layer mechanism to create the underlying connection is expected to be
33  * called prior to calling fmd_xprt_open() itself.  Alternatively, a transport
34  * may be created in the suspended state by specifying the FMD_XPRT_SUSPENDED
35  * flag as part of the call to fmd_xprt_open(), and then may be resumed later.
36  * The underlying transport mechanism is *required* to provide ordering: that
37  * is, the sequences of bytes written across the transport must be read by
38  * the remote peer in the order that they are written, even across separate
39  * calls to fmdo_send().  As an example, the Internet TCP protocol would be
40  * a valid transport as it guarantees ordering, whereas the Internet UDP
41  * protocol would not because UDP datagrams may be delivered in any order
42  * as a result of delays introduced when datagrams pass through routers.
43  *
44  * Similar to sending events, a transport module receives events that are from
45  * its peer remote endpoint using some transport-specific mechanism that is
46  * unknown to FMD.  As each event is received, the transport module is
47  * responsible for constructing a valid nvlist_t object from the data and then
48  * calling fmd_xprt_post() to post the event to the containing FMD's dispatch
49  * queue, making it available to all local modules that are not transport
50  * modules that have subscribed to the event.
51  *
52  * The following state machine is used for each transport.  The initial state
53  * is either SYN, ACK, or RUN, depending on the flags specified to xprt_create.
54  *
55  *       FMD_XPRT_ACCEPT   !FMD_XPRT_ACCEPT
56  *             |                 |
57  * waiting  +--v--+           +--v--+  waiting
58  * for syn  | SYN |--+     --+| ACK |  for ack
59  * event    +-----+   \   /   +-----+  event
60  *             |       \ /       |
61  * drop all +--v--+     X     +--v--+  send subscriptions,
62  * events   | ERR |<---+ +--->| SUB |  recv subscriptions,
63  *          +-----+           +-----+  wait for run event
64  *             ^                 |
65  *             |     +-----+     |
66  *             +-----| RUN |<----+
67  *                   +--^--+
68  *                      |
69  *               FMD_XPRT_RDONLY
70  *
71  * When fmd_xprt_open() is called without FMD_XPRT_ACCEPT, the Common Transport
72  * Layer enqueues a "syn" event for the module in its event queue and sets the
73  * state to ACK.  In state ACK, we are waiting for the transport to get an
74  * "ack" event and call fmd_xprt_post() on this event.  Other events will be
75  * discarded.  If an "ack" is received, we transition to state SUB.  If a
76  * configurable timeout occurs or if the "ack" is invalid (e.g. invalid version
77  * exchange), we transition to state ERR.  Once in state ERR, no further
78  * operations are valid except fmd_xprt_close() and fmd_xprt_error() will
79  * return a non-zero value to the caller indicating the transport has failed.
80  *
81  * When fmd_xprt_open() is called with FMD_XPRT_ACCEPT, the Common Transport
82  * Layer assumes this transport is being used to accept a virtual connection
83  * from a remote peer that is sending a "syn", and sets the initial state to
84  * SYN.  In this state, the transport waits for a "syn" event, validates it,
85  * and then transitions to state SUB if it is valid or state ERR if it is not.
86  *
87  * Once in state SUB, the transport module is expected to receive a sequence of
88  * zero or more "subscribe" events from the remote peer, followed by a "run"
89  * event.  Once in state RUN, the transport is active and any events can be
90  * sent or received.  The transport module is free to call fmd_xprt_close()
91  * from any state.  The fmd_xprt_error() function will return zero if the
92  * transport is not in the ERR state, or non-zero if it is in the ERR state.
93  *
94  * Once the state machine reaches RUN, other FMA protocol events can be sent
95  * and received across the transport in addition to the various control events.
96  *
97  * Table of Common Transport Layer Control Events
98  * ==============================================
99  *
100  * FMA Class                     Payload
101  * ---------                     -------
102  * resource.fm.xprt.uuclose      string (uuid of case)
103  * resource.fm.xprt.uuresolved   string (uuid of case)
104  * resource.fm.xprt.updated      string (uuid of case)
105  * resource.fm.xprt.subscribe    string (class pattern)
106  * resource.fm.xprt.unsubscribe  string (class pattern)
107  * resource.fm.xprt.unsuback     string (class pattern)
108  * resource.fm.xprt.syn          version information
109  * resource.fm.xprt.ack          version information
110  * resource.fm.xprt.run          version information
111  *
112  * Control events are used to add and delete proxy subscriptions on the remote
113  * transport peer module, and to set up connections.  When a "syn" event is
114  * sent, FMD will include in the payload the highest version of the FMA event
115  * protocol that is supported by the sender.  When a "syn" event is received,
116  * the receiving FMD will use the minimum of this version and its version of
117  * the protocol, and reply with this new minimum version in the "ack" event.
118  * The receiver will then use this new minimum for subsequent event semantics.
119  */
120 
121 #include <sys/fm/protocol.h>
122 #include <strings.h>
123 #include <limits.h>
124 
125 #include <fmd_alloc.h>
126 #include <fmd_error.h>
127 #include <fmd_conf.h>
128 #include <fmd_subr.h>
129 #include <fmd_string.h>
130 #include <fmd_protocol.h>
131 #include <fmd_thread.h>
132 #include <fmd_eventq.h>
133 #include <fmd_dispq.h>
134 #include <fmd_ctl.h>
135 #include <fmd_log.h>
136 #include <fmd_ustat.h>
137 #include <fmd_case.h>
138 #include <fmd_api.h>
139 #include <fmd_fmri.h>
140 #include <fmd_asru.h>
141 #include <fmd_xprt.h>
142 
143 #include <fmd.h>
144 
145 /*
146  * The states shown above in the transport state machine diagram are encoded
147  * using arrays of class patterns and a corresponding action function.  These
148  * arrays are then passed to fmd_xprt_transition() to change transport states.
149  */
150 
151 const fmd_xprt_rule_t _fmd_xprt_state_syn[] = {
152 { "resource.fm.xprt.syn", fmd_xprt_event_syn },
153 { "*", fmd_xprt_event_error },
154 { NULL, NULL }
155 };
156 
157 const fmd_xprt_rule_t _fmd_xprt_state_ack[] = {
158 { "resource.fm.xprt.ack", fmd_xprt_event_ack },
159 { "*", fmd_xprt_event_error },
160 };
161 
162 const fmd_xprt_rule_t _fmd_xprt_state_err[] = {
163 { "*", fmd_xprt_event_drop },
164 { NULL, NULL }
165 };
166 
167 const fmd_xprt_rule_t _fmd_xprt_state_sub[] = {
168 { "resource.fm.xprt.subscribe", fmd_xprt_event_sub },
169 { "resource.fm.xprt.run", fmd_xprt_event_run },
170 { "resource.fm.xprt.*", fmd_xprt_event_error },
171 { "*", fmd_xprt_event_drop },
172 { NULL, NULL }
173 };
174 
175 const fmd_xprt_rule_t _fmd_xprt_state_run[] = {
176 { "resource.fm.xprt.subscribe", fmd_xprt_event_sub },
177 { "resource.fm.xprt.unsubscribe", fmd_xprt_event_unsub },
178 { "resource.fm.xprt.unsuback", fmd_xprt_event_unsuback },
179 { "resource.fm.xprt.uuclose", fmd_xprt_event_uuclose },
180 { "resource.fm.xprt.uuresolved", fmd_xprt_event_uuresolved },
181 { "resource.fm.xprt.updated", fmd_xprt_event_updated },
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_xprt_t *xp = (fmd_xprt_t *)xip;
372 	fmd_event_t *e;
373 	nvlist_t *nvl;
374 	char *s;
375 
376 	TRACE((FMD_DBG_XPRT, "xprt %u -> %s\n", xip->xi_id, tag));
377 
378 	xip->xi_state = state;
379 	s = fmd_strdup(tag, FMD_SLEEP);
380 
381 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
382 	fmd_strfree(xip->xi_stats->xs_state.fmds_value.str);
383 	xip->xi_stats->xs_state.fmds_value.str = s;
384 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
385 
386 	/*
387 	 * If we've reached the SUB state, take out the big hammer and snapshot
388 	 * all of the subscriptions of all of the loaded modules.  Then queue a
389 	 * run event for our remote peer indicating that it can enter RUN.
390 	 */
391 	if (state == _fmd_xprt_state_sub) {
392 		fmd_xprt_subscribe_modhash(xip, fmd.d_mod_hash);
393 
394 		/*
395 		 * For read-write transports, we always want to set up remote
396 		 * subscriptions to the bultin list.* events, regardless of
397 		 * whether any agents have subscribed to them.
398 		 */
399 		if (xip->xi_flags & FMD_XPRT_RDWR) {
400 			fmd_xprt_subscribe(xp, FM_LIST_SUSPECT_CLASS);
401 			fmd_xprt_subscribe(xp, FM_LIST_ISOLATED_CLASS);
402 			fmd_xprt_subscribe(xp, FM_LIST_UPDATED_CLASS);
403 			fmd_xprt_subscribe(xp, FM_LIST_RESOLVED_CLASS);
404 			fmd_xprt_subscribe(xp, FM_LIST_REPAIRED_CLASS);
405 		}
406 
407 		nvl = fmd_protocol_xprt_ctl(xip->xi_queue->eq_mod,
408 		    "resource.fm.xprt.run", xip->xi_version);
409 
410 		(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
411 		e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
412 		fmd_eventq_insert_at_time(xip->xi_queue, e);
413 	}
414 }
415 
416 static void
417 fmd_xprt_authupdate(fmd_xprt_impl_t *xip)
418 {
419 	char *s = fmd_fmri_auth2str(xip->xi_auth);
420 
421 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
422 	fmd_strfree(xip->xi_stats->xs_authority.fmds_value.str);
423 	xip->xi_stats->xs_authority.fmds_value.str = s;
424 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
425 }
426 
427 static int
428 fmd_xprt_vmismatch(fmd_xprt_impl_t *xip, nvlist_t *nvl, uint_t *rversionp)
429 {
430 	uint8_t rversion;
431 
432 	if (nvlist_lookup_uint8(nvl, FM_VERSION, &rversion) != 0) {
433 		(void) pthread_mutex_lock(&xip->xi_stats_lock);
434 		xip->xi_stats->xs_discarded.fmds_value.ui64++;
435 		(void) pthread_mutex_unlock(&xip->xi_stats_lock);
436 
437 		fmd_xprt_transition(xip, _fmd_xprt_state_err, "ERR");
438 		return (1);
439 	}
440 
441 	if (rversion > xip->xi_version) {
442 		fmd_dprintf(FMD_DBG_XPRT, "xprt %u protocol mismatch: %u>%u\n",
443 		    xip->xi_id, rversion, xip->xi_version);
444 
445 		(void) pthread_mutex_lock(&xip->xi_stats_lock);
446 		xip->xi_stats->xs_discarded.fmds_value.ui64++;
447 		(void) pthread_mutex_unlock(&xip->xi_stats_lock);
448 
449 		fmd_xprt_transition(xip, _fmd_xprt_state_err, "ERR");
450 		return (1);
451 	}
452 
453 	if (rversionp != NULL)
454 		*rversionp = rversion;
455 
456 	return (0);
457 }
458 
459 void
460 fmd_xprt_event_syn(fmd_xprt_impl_t *xip, nvlist_t *nvl)
461 {
462 	fmd_event_t *e;
463 	uint_t vers;
464 	char *class;
465 
466 	if (fmd_xprt_vmismatch(xip, nvl, &vers))
467 		return; /* transitioned to error state */
468 
469 	/*
470 	 * If the transport module didn't specify an authority, extract the
471 	 * one that is passed along with the xprt.syn event and use that.
472 	 */
473 	if (xip->xi_auth == NULL &&
474 	    nvlist_lookup_nvlist(nvl, FM_RSRC_RESOURCE, &nvl) == 0 &&
475 	    nvlist_lookup_nvlist(nvl, FM_FMRI_AUTHORITY, &nvl) == 0) {
476 		(void) nvlist_xdup(nvl, &xip->xi_auth, &fmd.d_nva);
477 		fmd_xprt_authupdate(xip);
478 	}
479 
480 	nvl = fmd_protocol_xprt_ctl(xip->xi_queue->eq_mod,
481 	    "resource.fm.xprt.ack", xip->xi_version);
482 
483 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
484 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, class);
485 	fmd_eventq_insert_at_time(xip->xi_queue, e);
486 
487 	xip->xi_version = MIN(FM_RSRC_XPRT_VERSION, vers);
488 	fmd_xprt_transition(xip, _fmd_xprt_state_sub, "SUB");
489 }
490 
491 void
492 fmd_xprt_event_ack(fmd_xprt_impl_t *xip, nvlist_t *nvl)
493 {
494 	uint_t vers;
495 
496 	if (fmd_xprt_vmismatch(xip, nvl, &vers))
497 		return; /* transitioned to error state */
498 
499 	/*
500 	 * If the transport module didn't specify an authority, extract the
501 	 * one that is passed along with the xprt.syn event and use that.
502 	 */
503 	if (xip->xi_auth == NULL &&
504 	    nvlist_lookup_nvlist(nvl, FM_RSRC_RESOURCE, &nvl) == 0 &&
505 	    nvlist_lookup_nvlist(nvl, FM_FMRI_AUTHORITY, &nvl) == 0) {
506 		(void) nvlist_xdup(nvl, &xip->xi_auth, &fmd.d_nva);
507 		fmd_xprt_authupdate(xip);
508 	}
509 
510 	xip->xi_version = MIN(FM_RSRC_XPRT_VERSION, vers);
511 	fmd_xprt_transition(xip, _fmd_xprt_state_sub, "SUB");
512 }
513 
514 /*
515  * Upon transition to RUN, we take every solved case and resend a list.suspect
516  * event for it to our remote peer.  If a case transitions from solved to a
517  * future state (CLOSE_WAIT, CLOSED, or REPAIRED) while we are iterating over
518  * the case hash, we will get it as part of examining the resource cache, next.
519  */
520 static void
521 fmd_xprt_send_case(fmd_case_t *cp, void *arg)
522 {
523 	fmd_case_impl_t *cip = (fmd_case_impl_t *)cp;
524 	fmd_xprt_impl_t *xip = arg;
525 
526 	fmd_event_t *e;
527 	nvlist_t *nvl;
528 	char *class;
529 
530 	if (cip->ci_state != FMD_CASE_SOLVED)
531 		return;
532 
533 	nvl = fmd_case_mkevent(cp, FM_LIST_SUSPECT_CLASS);
534 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
535 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, class);
536 
537 	fmd_dprintf(FMD_DBG_XPRT, "re-send %s for %s to transport %u\n",
538 	    FM_LIST_SUSPECT_CLASS, cip->ci_uuid, xip->xi_id);
539 
540 	fmd_dispq_dispatch_gid(fmd.d_disp, e, class, xip->xi_queue->eq_sgid);
541 }
542 
543 /*
544  * Similar to the above function, but for use with readonly transport. Puts
545  * the event on the module's queue such that it's fmdo_recv function can pick
546  * it up and send it if appropriate.
547  */
548 static void
549 fmd_xprt_send_case_ro(fmd_case_t *cp, void *arg)
550 {
551 	fmd_case_impl_t *cip = (fmd_case_impl_t *)cp;
552 	fmd_module_t *mp = arg;
553 
554 	fmd_event_t *e;
555 	nvlist_t *nvl;
556 	char *class;
557 
558 	if (cip->ci_state != FMD_CASE_SOLVED)
559 		return;
560 
561 	nvl = fmd_case_mkevent(cp, FM_LIST_SUSPECT_CLASS);
562 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
563 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, class);
564 
565 	fmd_dprintf(FMD_DBG_XPRT, "re-send %s for %s to rdonly transport %s\n",
566 	    FM_LIST_SUSPECT_CLASS, cip->ci_uuid, mp->mod_name);
567 
568 	fmd_dispq_dispatch_gid(fmd.d_disp, e, class, mp->mod_queue->eq_sgid);
569 }
570 
571 void
572 fmd_xprt_event_run(fmd_xprt_impl_t *xip, nvlist_t *nvl)
573 {
574 	if (!fmd_xprt_vmismatch(xip, nvl, NULL)) {
575 		fmd_xprt_transition(xip, _fmd_xprt_state_run, "RUN");
576 		fmd_case_hash_apply(fmd.d_cases, fmd_xprt_send_case, xip);
577 	}
578 }
579 
580 void
581 fmd_xprt_event_sub(fmd_xprt_impl_t *xip, nvlist_t *nvl)
582 {
583 	char *class;
584 
585 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
586 		return; /* transitioned to error state */
587 
588 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_SUBCLASS, &class) != 0)
589 		return; /* malformed protocol event */
590 
591 	(void) pthread_mutex_lock(&xip->xi_lock);
592 	(void) fmd_xprt_class_hash_insert(xip, &xip->xi_lsub, class);
593 	(void) pthread_mutex_unlock(&xip->xi_lock);
594 
595 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
596 	xip->xi_stats->xs_subscriptions.fmds_value.ui64++;
597 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
598 }
599 
600 void
601 fmd_xprt_event_unsub(fmd_xprt_impl_t *xip, nvlist_t *nvl)
602 {
603 	fmd_event_t *e;
604 	char *class;
605 
606 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
607 		return; /* transitioned to error state */
608 
609 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_SUBCLASS, &class) != 0)
610 		return; /* malformed protocol event */
611 
612 	(void) pthread_mutex_lock(&xip->xi_lock);
613 	(void) fmd_xprt_class_hash_delete(xip, &xip->xi_lsub, class);
614 	(void) pthread_mutex_unlock(&xip->xi_lock);
615 
616 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
617 	xip->xi_stats->xs_subscriptions.fmds_value.ui64--;
618 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
619 
620 	nvl = fmd_protocol_xprt_sub(xip->xi_queue->eq_mod,
621 	    "resource.fm.xprt.unsuback", xip->xi_version, class);
622 
623 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
624 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, class);
625 	fmd_eventq_insert_at_time(xip->xi_queue, e);
626 }
627 
628 void
629 fmd_xprt_event_unsuback(fmd_xprt_impl_t *xip, nvlist_t *nvl)
630 {
631 	char *class;
632 
633 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
634 		return; /* transitioned to error state */
635 
636 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_SUBCLASS, &class) != 0)
637 		return; /* malformed protocol event */
638 
639 	(void) pthread_mutex_lock(&xip->xi_lock);
640 	(void) fmd_xprt_class_hash_delete(xip, &xip->xi_usub, class);
641 	(void) pthread_mutex_unlock(&xip->xi_lock);
642 }
643 
644 /*
645  * on diagnosing side, receive a uuclose from the proxy.
646  */
647 void
648 fmd_xprt_event_uuclose(fmd_xprt_impl_t *xip, nvlist_t *nvl)
649 {
650 	fmd_case_t *cp;
651 	char *uuid;
652 
653 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
654 		return; /* transitioned to error state */
655 
656 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_UUID, &uuid) == 0 &&
657 	    (cp = fmd_case_hash_lookup(fmd.d_cases, uuid)) != NULL) {
658 		/*
659 		 * update resource cache status and transition case
660 		 */
661 		fmd_case_close_status(cp);
662 		fmd_case_transition(cp, FMD_CASE_CLOSE_WAIT, FMD_CF_ISOLATED);
663 		fmd_case_rele(cp);
664 	}
665 }
666 
667 /*
668  * on diagnosing side, receive a uuresolved from the proxy.
669  */
670 void
671 fmd_xprt_event_uuresolved(fmd_xprt_impl_t *xip, nvlist_t *nvl)
672 {
673 	fmd_case_t *cp;
674 	char *uuid;
675 
676 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
677 		return; /* transitioned to error state */
678 
679 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_UUID, &uuid) == 0 &&
680 	    (cp = fmd_case_hash_lookup(fmd.d_cases, uuid)) != NULL) {
681 		fmd_case_impl_t *cip = (fmd_case_impl_t *)cp;
682 
683 		fmd_case_transition(cp, (cip->ci_state == FMD_CASE_REPAIRED) ?
684 		    FMD_CASE_RESOLVED : (cip->ci_state == FMD_CASE_CLOSED) ?
685 		    FMD_CASE_REPAIRED : FMD_CASE_CLOSE_WAIT, FMD_CF_RESOLVED);
686 		fmd_case_rele(cp);
687 	}
688 }
689 
690 /*
691  * on diagnosing side, receive a repair/acquit from the proxy.
692  */
693 void
694 fmd_xprt_event_updated(fmd_xprt_impl_t *xip, nvlist_t *nvl)
695 {
696 	fmd_case_t *cp;
697 	char *uuid;
698 
699 	if (fmd_xprt_vmismatch(xip, nvl, NULL))
700 		return; /* transitioned to error state */
701 
702 	if (nvlist_lookup_string(nvl, FM_RSRC_XPRT_UUID, &uuid) == 0 &&
703 	    (cp = fmd_case_hash_lookup(fmd.d_cases, uuid)) != NULL) {
704 		uint8_t *statusp, *proxy_asrup = NULL;
705 		uint_t nelem = 0;
706 
707 		/*
708 		 * Only update status with new repairs if "no remote repair"
709 		 * is not set. Do the case_update anyway though (as this will
710 		 * refresh the status on the proxy side).
711 		 */
712 		if (!(xip->xi_flags & FMD_XPRT_NO_REMOTE_REPAIR)) {
713 			if (nvlist_lookup_uint8_array(nvl,
714 			    FM_RSRC_XPRT_FAULT_STATUS, &statusp, &nelem) == 0 &&
715 			    nelem != 0) {
716 				(void) nvlist_lookup_uint8_array(nvl,
717 				    FM_RSRC_XPRT_FAULT_HAS_ASRU, &proxy_asrup,
718 				    &nelem);
719 				fmd_case_update_status(cp, statusp,
720 				    proxy_asrup, NULL);
721 			}
722 			fmd_case_update_containees(cp);
723 		}
724 		fmd_case_update(cp);
725 		fmd_case_rele(cp);
726 	}
727 }
728 
729 void
730 fmd_xprt_event_error(fmd_xprt_impl_t *xip, nvlist_t *nvl)
731 {
732 	char *class = "<unknown>";
733 
734 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
735 	xip->xi_stats->xs_discarded.fmds_value.ui64++;
736 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
737 
738 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
739 	TRACE((FMD_DBG_XPRT, "xprt %u bad event %s\n", xip->xi_id, class));
740 
741 	fmd_xprt_transition(xip, _fmd_xprt_state_err, "ERR");
742 }
743 
744 void
745 fmd_xprt_event_drop(fmd_xprt_impl_t *xip, nvlist_t *nvl)
746 {
747 	char *class = "<unknown>";
748 
749 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
750 	xip->xi_stats->xs_discarded.fmds_value.ui64++;
751 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
752 
753 	(void) nvlist_lookup_string(nvl, FM_CLASS, &class);
754 	TRACE((FMD_DBG_XPRT, "xprt %u drop event %s\n", xip->xi_id, class));
755 
756 }
757 
758 fmd_xprt_t *
759 fmd_xprt_create(fmd_module_t *mp, uint_t flags, nvlist_t *auth, void *data)
760 {
761 	fmd_xprt_impl_t *xip = fmd_zalloc(sizeof (fmd_xprt_impl_t), FMD_SLEEP);
762 	fmd_stat_t *statv;
763 	uint_t i, statc;
764 
765 	char buf[PATH_MAX];
766 	fmd_event_t *e;
767 	nvlist_t *nvl;
768 	char *s;
769 
770 	(void) pthread_mutex_init(&xip->xi_lock, NULL);
771 	(void) pthread_cond_init(&xip->xi_cv, NULL);
772 	(void) pthread_mutex_init(&xip->xi_stats_lock, NULL);
773 
774 	xip->xi_auth = auth;
775 	xip->xi_data = data;
776 	xip->xi_version = FM_RSRC_XPRT_VERSION;
777 	xip->xi_flags = flags;
778 
779 	/*
780 	 * Grab fmd.d_xprt_lock to block fmd_xprt_suspend_all() and then create
781 	 * a transport ID and make it visible in fmd.d_xprt_ids.  If transports
782 	 * were previously suspended, set the FMD_XPRT_DSUSPENDED flag on us to
783 	 * ensure that this transport will not run until fmd_xprt_resume_all().
784 	 */
785 	(void) pthread_mutex_lock(&fmd.d_xprt_lock);
786 	xip->xi_id = fmd_idspace_alloc(fmd.d_xprt_ids, xip);
787 
788 	if (fmd.d_xprt_suspend != 0)
789 		xip->xi_flags |= FMD_XPRT_DSUSPENDED;
790 
791 	(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
792 
793 	/*
794 	 * If the module has not yet finished _fmd_init(), set the ISUSPENDED
795 	 * bit so that fmdo_send() is not called until _fmd_init() completes.
796 	 */
797 	if (!(mp->mod_flags & FMD_MOD_INIT))
798 		xip->xi_flags |= FMD_XPRT_ISUSPENDED;
799 
800 	/*
801 	 * Initialize the transport statistics that we keep on behalf of fmd.
802 	 * These are set up using a template defined at the top of this file.
803 	 * We rename each statistic with a prefix ensuring its uniqueness.
804 	 */
805 	statc = sizeof (_fmd_xprt_stat_tmpl) / sizeof (fmd_stat_t);
806 	statv = fmd_alloc(sizeof (_fmd_xprt_stat_tmpl), FMD_SLEEP);
807 	bcopy(&_fmd_xprt_stat_tmpl, statv, sizeof (_fmd_xprt_stat_tmpl));
808 
809 	for (i = 0; i < statc; i++) {
810 		(void) snprintf(statv[i].fmds_name,
811 		    sizeof (statv[i].fmds_name), "fmd.xprt.%u.%s", xip->xi_id,
812 		    ((fmd_stat_t *)&_fmd_xprt_stat_tmpl + i)->fmds_name);
813 	}
814 
815 	xip->xi_stats = (fmd_xprt_stat_t *)fmd_ustat_insert(
816 	    mp->mod_ustat, FMD_USTAT_NOALLOC, statc, statv, NULL);
817 
818 	if (xip->xi_stats == NULL)
819 		fmd_panic("failed to create xi_stats (%p)\n", (void *)statv);
820 
821 	xip->xi_stats->xs_module.fmds_value.str =
822 	    fmd_strdup(mp->mod_name, FMD_SLEEP);
823 
824 	if (xip->xi_auth != NULL)
825 		fmd_xprt_authupdate(xip);
826 
827 	/*
828 	 * Create the outbound eventq for this transport and link to its stats.
829 	 * If any suspend bits were set above, suspend the eventq immediately.
830 	 */
831 	xip->xi_queue = fmd_eventq_create(mp, &xip->xi_stats->xs_evqstat,
832 	    &xip->xi_stats_lock, mp->mod_stats->ms_xprtqlimit.fmds_value.ui32);
833 
834 	if (xip->xi_flags & FMD_XPRT_SMASK)
835 		fmd_eventq_suspend(xip->xi_queue);
836 
837 	/*
838 	 * Create our subscription hashes: local subscriptions go to xi_queue,
839 	 * remote subscriptions are tracked only for protocol requests, and
840 	 * pending unsubscriptions are associated with the /dev/null eventq.
841 	 */
842 	fmd_xprt_class_hash_create(&xip->xi_lsub, xip->xi_queue);
843 	fmd_xprt_class_hash_create(&xip->xi_rsub, NULL);
844 	fmd_xprt_class_hash_create(&xip->xi_usub, fmd.d_rmod->mod_queue);
845 
846 	/*
847 	 * Determine our initial state based upon the creation flags.  If we're
848 	 * read-only, go directly to RUN.  If we're accepting a new connection,
849 	 * wait for a SYN.  Otherwise send a SYN and wait for an ACK.
850 	 */
851 	if ((flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY) {
852 		/*
853 		 * Send the list.suspects across here for readonly transports.
854 		 * For read-write transport they will be sent on transition to
855 		 * RUN state in fmd_xprt_event_run().
856 		 */
857 		fmd_case_hash_apply(fmd.d_cases, fmd_xprt_send_case_ro, mp);
858 		fmd_xprt_transition(xip, _fmd_xprt_state_run, "RUN");
859 	} else if (flags & FMD_XPRT_ACCEPT)
860 		fmd_xprt_transition(xip, _fmd_xprt_state_syn, "SYN");
861 	else
862 		fmd_xprt_transition(xip, _fmd_xprt_state_ack, "ACK");
863 
864 	/*
865 	 * If client.xprtlog is set to TRUE, create a debugging log for the
866 	 * events received by the transport in var/fm/fmd/xprt/.
867 	 */
868 	(void) fmd_conf_getprop(fmd.d_conf, "client.xprtlog", &i);
869 	(void) fmd_conf_getprop(fmd.d_conf, "log.xprt", &s);
870 
871 	if (i) {
872 		(void) snprintf(buf, sizeof (buf), "%s/%u.log", s, xip->xi_id);
873 		xip->xi_log = fmd_log_open(fmd.d_rootdir, buf, FMD_LOG_XPRT);
874 	}
875 
876 	ASSERT(fmd_module_locked(mp));
877 	fmd_list_append(&mp->mod_transports, xip);
878 
879 	(void) pthread_mutex_lock(&mp->mod_stats_lock);
880 	mp->mod_stats->ms_xprtopen.fmds_value.ui32++;
881 	(void) pthread_mutex_unlock(&mp->mod_stats_lock);
882 
883 	/*
884 	 * If this is a read-only transport, return without creating a send
885 	 * queue thread and setting up any connection events in our queue.
886 	 */
887 	if ((flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
888 		goto out;
889 
890 	/*
891 	 * Once the transport is fully initialized, create a send queue thread
892 	 * and start any connect events flowing to complete our initialization.
893 	 */
894 	if ((xip->xi_thread = fmd_thread_create(mp,
895 	    (fmd_thread_f *)fmd_xprt_send, xip)) == NULL) {
896 
897 		fmd_error(EFMD_XPRT_THR,
898 		    "failed to create thread for transport %u", xip->xi_id);
899 
900 		fmd_xprt_destroy((fmd_xprt_t *)xip);
901 		(void) fmd_set_errno(EFMD_XPRT_THR);
902 		return (NULL);
903 	}
904 
905 	/*
906 	 * If the transport is not being opened to accept an inbound connect,
907 	 * start an outbound connection by enqueuing a SYN event for our peer.
908 	 */
909 	if (!(flags & FMD_XPRT_ACCEPT)) {
910 		nvl = fmd_protocol_xprt_ctl(mp,
911 		    "resource.fm.xprt.syn", FM_RSRC_XPRT_VERSION);
912 
913 		(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
914 		e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
915 		fmd_eventq_insert_at_time(xip->xi_queue, e);
916 	}
917 out:
918 	fmd_dprintf(FMD_DBG_XPRT, "opened transport %u\n", xip->xi_id);
919 	return ((fmd_xprt_t *)xip);
920 }
921 
922 void
923 fmd_xprt_destroy(fmd_xprt_t *xp)
924 {
925 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
926 	fmd_module_t *mp = xip->xi_queue->eq_mod;
927 	uint_t id = xip->xi_id;
928 
929 	fmd_case_impl_t *cip, *nip;
930 	fmd_stat_t *sp;
931 	uint_t i, n;
932 
933 	ASSERT(fmd_module_locked(mp));
934 	fmd_list_delete(&mp->mod_transports, xip);
935 
936 	(void) pthread_mutex_lock(&mp->mod_stats_lock);
937 	mp->mod_stats->ms_xprtopen.fmds_value.ui32--;
938 	(void) pthread_mutex_unlock(&mp->mod_stats_lock);
939 
940 	(void) pthread_mutex_lock(&xip->xi_lock);
941 
942 	while (xip->xi_busy != 0)
943 		(void) pthread_cond_wait(&xip->xi_cv, &xip->xi_lock);
944 
945 	/*
946 	 * Remove the transport from global visibility, cancel its send-side
947 	 * thread, join with it, and then remove the transport from module
948 	 * visibility.  Once all this is done, destroy and free the transport.
949 	 */
950 	(void) fmd_idspace_free(fmd.d_xprt_ids, xip->xi_id);
951 
952 	if (xip->xi_thread != NULL) {
953 		fmd_eventq_abort(xip->xi_queue);
954 		fmd_module_unlock(mp);
955 		fmd_thread_destroy(xip->xi_thread, FMD_THREAD_JOIN);
956 		fmd_module_lock(mp);
957 	}
958 
959 	if (xip->xi_log != NULL)
960 		fmd_log_rele(xip->xi_log);
961 
962 	/*
963 	 * Release every case handle in the module that was cached by this
964 	 * transport.  This will result in these cases disappearing from the
965 	 * local case hash so that fmd_case_uuclose() and fmd_case_repaired()
966 	 * etc can no longer be used.
967 	 */
968 	for (cip = fmd_list_next(&mp->mod_cases); cip != NULL; cip = nip) {
969 		nip = fmd_list_next(cip);
970 		if (cip->ci_xprt == xp)
971 			fmd_case_discard((fmd_case_t *)cip, B_TRUE);
972 	}
973 
974 	/*
975 	 * Destroy every class in the various subscription hashes and remove
976 	 * any corresponding subscriptions from the event dispatch queue.
977 	 */
978 	fmd_xprt_class_hash_destroy(&xip->xi_lsub);
979 	fmd_xprt_class_hash_destroy(&xip->xi_rsub);
980 	fmd_xprt_class_hash_destroy(&xip->xi_usub);
981 
982 	/*
983 	 * Uniquify the stat names exactly as was done in fmd_xprt_create()
984 	 * before calling fmd_ustat_insert(), otherwise fmd_ustat_delete()
985 	 * won't find the entries in the hash table.
986 	 */
987 	n = sizeof (_fmd_xprt_stat_tmpl) / sizeof (fmd_stat_t);
988 	sp = fmd_alloc(sizeof (_fmd_xprt_stat_tmpl), FMD_SLEEP);
989 	bcopy(&_fmd_xprt_stat_tmpl, sp, sizeof (_fmd_xprt_stat_tmpl));
990 	for (i = 0; i < n; i++) {
991 		(void) snprintf(sp[i].fmds_name,
992 		    sizeof (sp[i].fmds_name), "fmd.xprt.%u.%s", xip->xi_id,
993 		    ((fmd_stat_t *)&_fmd_xprt_stat_tmpl + i)->fmds_name);
994 	}
995 	fmd_ustat_delete(mp->mod_ustat, n, sp);
996 	fmd_free(sp, sizeof (_fmd_xprt_stat_tmpl));
997 
998 	fmd_free(xip->xi_stats, sizeof (fmd_xprt_stat_t));
999 	fmd_eventq_destroy(xip->xi_queue);
1000 	nvlist_free(xip->xi_auth);
1001 	fmd_free(xip, sizeof (fmd_xprt_impl_t));
1002 
1003 	fmd_dprintf(FMD_DBG_XPRT, "closed transport %u\n", id);
1004 }
1005 
1006 void
1007 fmd_xprt_xsuspend(fmd_xprt_t *xp, uint_t flags)
1008 {
1009 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1010 	uint_t oflags;
1011 
1012 	ASSERT((flags & ~FMD_XPRT_SMASK) == 0);
1013 	(void) pthread_mutex_lock(&xip->xi_lock);
1014 
1015 	oflags = xip->xi_flags;
1016 	xip->xi_flags |= flags;
1017 
1018 	if (!(oflags & FMD_XPRT_SMASK) && (xip->xi_flags & FMD_XPRT_SMASK) != 0)
1019 		fmd_eventq_suspend(xip->xi_queue);
1020 
1021 	(void) pthread_cond_broadcast(&xip->xi_cv);
1022 
1023 	while (xip->xi_busy != 0)
1024 		(void) pthread_cond_wait(&xip->xi_cv, &xip->xi_lock);
1025 
1026 	(void) pthread_mutex_unlock(&xip->xi_lock);
1027 }
1028 
1029 void
1030 fmd_xprt_xresume(fmd_xprt_t *xp, uint_t flags)
1031 {
1032 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1033 	uint_t oflags;
1034 
1035 	ASSERT((flags & ~FMD_XPRT_SMASK) == 0);
1036 	(void) pthread_mutex_lock(&xip->xi_lock);
1037 
1038 	oflags = xip->xi_flags;
1039 	xip->xi_flags &= ~flags;
1040 
1041 	if ((oflags & FMD_XPRT_SMASK) != 0 && !(xip->xi_flags & FMD_XPRT_SMASK))
1042 		fmd_eventq_resume(xip->xi_queue);
1043 
1044 	(void) pthread_cond_broadcast(&xip->xi_cv);
1045 	(void) pthread_mutex_unlock(&xip->xi_lock);
1046 }
1047 
1048 void
1049 fmd_xprt_send(fmd_xprt_t *xp)
1050 {
1051 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1052 	fmd_module_t *mp = xip->xi_queue->eq_mod;
1053 	fmd_event_t *ep;
1054 	int err;
1055 
1056 	while ((ep = fmd_eventq_delete(xip->xi_queue)) != NULL) {
1057 		if (FMD_EVENT_TTL(ep) == 0) {
1058 			fmd_event_rele(ep);
1059 			continue;
1060 		}
1061 
1062 		fmd_dprintf(FMD_DBG_XPRT, "xprt %u sending %s\n",
1063 		    xip->xi_id, (char *)FMD_EVENT_DATA(ep));
1064 
1065 		err = mp->mod_ops->mop_transport(mp, xp, ep);
1066 		fmd_eventq_done(xip->xi_queue);
1067 
1068 		if (err == FMD_SEND_RETRY) {
1069 			fmd_eventq_insert_at_time(xip->xi_queue, ep);
1070 			(void) pthread_mutex_lock(&xip->xi_stats_lock);
1071 			xip->xi_stats->xs_retried.fmds_value.ui64++;
1072 			(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1073 		}
1074 
1075 		if (err != FMD_SEND_SUCCESS && err != FMD_SEND_RETRY) {
1076 			(void) pthread_mutex_lock(&xip->xi_stats_lock);
1077 			xip->xi_stats->xs_lost.fmds_value.ui64++;
1078 			(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1079 		}
1080 
1081 		fmd_event_rele(ep);
1082 	}
1083 }
1084 
1085 /*
1086  * This function creates a local suspect list. This is used when a suspect list
1087  * is created directly by an external source like fminject.
1088  */
1089 static void
1090 fmd_xprt_list_suspect_local(fmd_xprt_t *xp, nvlist_t *nvl)
1091 {
1092 	nvlist_t **nvlp;
1093 	nvlist_t *de_fmri, *de_fmri_dup = NULL;
1094 	int64_t *diag_time;
1095 	char *code = NULL;
1096 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1097 	fmd_case_t *cp;
1098 	uint_t nelem = 0, nelem2 = 0, i;
1099 	boolean_t injected;
1100 
1101 	fmd_module_lock(xip->xi_queue->eq_mod);
1102 	cp = fmd_case_create(xip->xi_queue->eq_mod, NULL, NULL);
1103 	if (cp == NULL) {
1104 		fmd_module_unlock(xip->xi_queue->eq_mod);
1105 		return;
1106 	}
1107 
1108 	/*
1109 	 * copy diag_code if present
1110 	 */
1111 	(void) nvlist_lookup_string(nvl, FM_SUSPECT_DIAG_CODE, &code);
1112 	if (code != NULL) {
1113 		fmd_case_impl_t *cip = (fmd_case_impl_t *)cp;
1114 
1115 		cip->ci_precanned = 1;
1116 		fmd_case_setcode(cp, code);
1117 	}
1118 
1119 	/*
1120 	 * copy suspects
1121 	 */
1122 	(void) nvlist_lookup_nvlist_array(nvl, FM_SUSPECT_FAULT_LIST, &nvlp,
1123 	    &nelem);
1124 	for (i = 0; i < nelem; i++) {
1125 		nvlist_t *flt_copy, *asru = NULL, *fru = NULL, *rsrc = NULL;
1126 		topo_hdl_t *thp;
1127 		char *loc = NULL;
1128 		int err;
1129 
1130 		thp = fmd_fmri_topo_hold(TOPO_VERSION);
1131 		(void) nvlist_xdup(nvlp[i], &flt_copy, &fmd.d_nva);
1132 		(void) nvlist_lookup_nvlist(nvlp[i], FM_FAULT_RESOURCE, &rsrc);
1133 
1134 		/*
1135 		 * If no fru specified, get it from topo
1136 		 */
1137 		if (nvlist_lookup_nvlist(nvlp[i], FM_FAULT_FRU, &fru) != 0 &&
1138 		    rsrc && topo_fmri_fru(thp, rsrc, &fru, &err) == 0)
1139 			(void) nvlist_add_nvlist(flt_copy, FM_FAULT_FRU, fru);
1140 		/*
1141 		 * If no asru specified, get it from topo
1142 		 */
1143 		if (nvlist_lookup_nvlist(nvlp[i], FM_FAULT_ASRU, &asru) != 0 &&
1144 		    rsrc && topo_fmri_asru(thp, rsrc, &asru, &err) == 0)
1145 			(void) nvlist_add_nvlist(flt_copy, FM_FAULT_ASRU, asru);
1146 		/*
1147 		 * If no location specified, get it from topo
1148 		 */
1149 		if (nvlist_lookup_string(nvlp[i], FM_FAULT_LOCATION,
1150 		    &loc) != 0) {
1151 			if (fru && topo_fmri_label(thp, fru, &loc, &err) == 0)
1152 				(void) nvlist_add_string(flt_copy,
1153 				    FM_FAULT_LOCATION, loc);
1154 			else if (rsrc && topo_fmri_label(thp, rsrc, &loc,
1155 			    &err) == 0)
1156 				(void) nvlist_add_string(flt_copy,
1157 				    FM_FAULT_LOCATION, loc);
1158 			if (loc)
1159 				topo_hdl_strfree(thp, loc);
1160 		}
1161 		nvlist_free(fru);
1162 		nvlist_free(asru);
1163 		nvlist_free(rsrc);
1164 		fmd_fmri_topo_rele(thp);
1165 		fmd_case_insert_suspect(cp, flt_copy);
1166 	}
1167 
1168 	/*
1169 	 * copy diag_time if present
1170 	 */
1171 	if (nvlist_lookup_int64_array(nvl, FM_SUSPECT_DIAG_TIME, &diag_time,
1172 	    &nelem2) == 0 && nelem2 >= 2)
1173 		fmd_case_settime(cp, diag_time[0], diag_time[1]);
1174 
1175 	/*
1176 	 * copy DE fmri if present
1177 	 */
1178 	if (nvlist_lookup_nvlist(nvl, FM_SUSPECT_DE, &de_fmri) == 0) {
1179 		(void) nvlist_xdup(de_fmri, &de_fmri_dup, &fmd.d_nva);
1180 		fmd_case_set_de_fmri(cp, de_fmri_dup);
1181 	}
1182 
1183 	/*
1184 	 * copy injected if present
1185 	 */
1186 	if (nvlist_lookup_boolean_value(nvl, FM_SUSPECT_INJECTED,
1187 	    &injected) == 0 && injected)
1188 		fmd_case_set_injected(cp);
1189 
1190 	fmd_case_transition(cp, FMD_CASE_SOLVED, FMD_CF_SOLVED);
1191 	fmd_module_unlock(xip->xi_queue->eq_mod);
1192 }
1193 
1194 /*
1195  * This function is called to create a proxy case on receipt of a list.suspect
1196  * from the diagnosing side of the transport.
1197  */
1198 static void
1199 fmd_xprt_list_suspect(fmd_xprt_t *xp, nvlist_t *nvl)
1200 {
1201 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1202 	nvlist_t **nvlp;
1203 	uint_t nelem = 0, nelem2 = 0, i;
1204 	int64_t *diag_time;
1205 	topo_hdl_t *thp;
1206 	char *class;
1207 	nvlist_t *rsrc, *asru, *de_fmri, *de_fmri_dup = NULL;
1208 	nvlist_t *flt_copy;
1209 	int err;
1210 	nvlist_t **asrua;
1211 	uint8_t *proxy_asru = NULL;
1212 	int got_proxy_asru = 0;
1213 	int got_hc_rsrc = 0;
1214 	int got_hc_asru = 0;
1215 	int got_present_rsrc = 0;
1216 	uint8_t *diag_asru = NULL;
1217 	char *scheme;
1218 	uint8_t *statusp;
1219 	char *uuid, *code;
1220 	fmd_case_t *cp;
1221 	fmd_case_impl_t *cip;
1222 	int need_update = 0;
1223 	boolean_t injected;
1224 
1225 	if (nvlist_lookup_string(nvl, FM_SUSPECT_UUID, &uuid) != 0)
1226 		return;
1227 	if (nvlist_lookup_string(nvl, FM_SUSPECT_DIAG_CODE, &code) != 0)
1228 		return;
1229 	(void) nvlist_lookup_nvlist_array(nvl, FM_SUSPECT_FAULT_LIST, &nvlp,
1230 	    &nelem);
1231 
1232 	/*
1233 	 * In order to implement FMD_XPRT_HCONLY and FMD_XPRT_HC_PRESENT_ONLY
1234 	 * etc we first scan the suspects to see if
1235 	 * - there was an asru in the received fault
1236 	 * - there was an hc-scheme resource in the received fault
1237 	 * - any hc-scheme resource in the received fault is present in the
1238 	 *   local topology
1239 	 * - any hc-scheme resource in the received fault has an asru in the
1240 	 *   local topology
1241 	 */
1242 	if (nelem > 0) {
1243 		asrua = fmd_zalloc(sizeof (nvlist_t *) * nelem, FMD_SLEEP);
1244 		proxy_asru = fmd_zalloc(sizeof (uint8_t) * nelem, FMD_SLEEP);
1245 		diag_asru = fmd_zalloc(sizeof (uint8_t) * nelem, FMD_SLEEP);
1246 		thp = fmd_fmri_topo_hold(TOPO_VERSION);
1247 		for (i = 0; i < nelem; i++) {
1248 			if (nvlist_lookup_nvlist(nvlp[i], FM_FAULT_ASRU,
1249 			    &asru) == 0 && asru != NULL)
1250 				diag_asru[i] = 1;
1251 			if (nvlist_lookup_string(nvlp[i], FM_CLASS,
1252 			    &class) != 0 || strncmp(class, "fault", 5) != 0)
1253 				continue;
1254 			/*
1255 			 * If there is an hc-scheme asru, use that to find the
1256 			 * real asru. Otherwise if there is an hc-scheme
1257 			 * resource, work out the old asru from that.
1258 			 * This order is to allow a two stage evaluation
1259 			 * of the asru where a fault in the diagnosing side
1260 			 * is in a component not visible to the proxy side,
1261 			 * but prevents a component that is visible from
1262 			 * working. So the diagnosing side sets the asru to
1263 			 * the latter component (in hc-scheme as the diagnosing
1264 			 * side doesn't know about the proxy side's virtual
1265 			 * schemes), and then the proxy side can convert that
1266 			 * to a suitable virtual scheme asru.
1267 			 */
1268 			if (nvlist_lookup_nvlist(nvlp[i], FM_FAULT_ASRU,
1269 			    &asru) == 0 && asru != NULL &&
1270 			    nvlist_lookup_string(asru, FM_FMRI_SCHEME,
1271 			    &scheme) == 0 &&
1272 			    strcmp(scheme, FM_FMRI_SCHEME_HC) == 0) {
1273 				got_hc_asru = 1;
1274 				if (xip->xi_flags & FMD_XPRT_EXTERNAL)
1275 					continue;
1276 				if (topo_fmri_present(thp, asru, &err) != 0)
1277 					got_present_rsrc = 1;
1278 				if (topo_fmri_asru(thp, asru, &asrua[i],
1279 				    &err) == 0) {
1280 					proxy_asru[i] =
1281 					    FMD_PROXY_ASRU_FROM_ASRU;
1282 					got_proxy_asru = 1;
1283 				}
1284 			} else if (nvlist_lookup_nvlist(nvlp[i],
1285 			    FM_FAULT_RESOURCE, &rsrc) == 0 && rsrc != NULL &&
1286 			    nvlist_lookup_string(rsrc, FM_FMRI_SCHEME,
1287 			    &scheme) == 0 &&
1288 			    strcmp(scheme, FM_FMRI_SCHEME_HC) == 0) {
1289 				got_hc_rsrc = 1;
1290 				if (xip->xi_flags & FMD_XPRT_EXTERNAL)
1291 					continue;
1292 				if (topo_fmri_present(thp, rsrc, &err) != 0)
1293 					got_present_rsrc = 1;
1294 				if (topo_fmri_asru(thp, rsrc, &asrua[i],
1295 				    &err) == 0) {
1296 					proxy_asru[i] =
1297 					    FMD_PROXY_ASRU_FROM_RSRC;
1298 					got_proxy_asru = 1;
1299 				}
1300 			}
1301 		}
1302 		fmd_fmri_topo_rele(thp);
1303 	}
1304 
1305 	/*
1306 	 * If we're set up only to report hc-scheme faults, and
1307 	 * there aren't any, then just drop the event.
1308 	 */
1309 	if (got_hc_rsrc == 0 && got_hc_asru == 0 &&
1310 	    (xip->xi_flags & FMD_XPRT_HCONLY)) {
1311 		if (nelem > 0) {
1312 			fmd_free(proxy_asru, sizeof (uint8_t) * nelem);
1313 			fmd_free(diag_asru, sizeof (uint8_t) * nelem);
1314 			fmd_free(asrua, sizeof (nvlist_t *) * nelem);
1315 		}
1316 		return;
1317 	}
1318 
1319 	/*
1320 	 * If we're set up only to report locally present hc-scheme
1321 	 * faults, and there aren't any, then just drop the event.
1322 	 */
1323 	if (got_present_rsrc == 0 &&
1324 	    (xip->xi_flags & FMD_XPRT_HC_PRESENT_ONLY)) {
1325 		if (nelem > 0) {
1326 			for (i = 0; i < nelem; i++)
1327 				nvlist_free(asrua[i]);
1328 			fmd_free(proxy_asru, sizeof (uint8_t) * nelem);
1329 			fmd_free(diag_asru, sizeof (uint8_t) * nelem);
1330 			fmd_free(asrua, sizeof (nvlist_t *) * nelem);
1331 		}
1332 		return;
1333 	}
1334 
1335 	/*
1336 	 * If fmd_case_recreate() returns NULL, UUID is already known.
1337 	 */
1338 	fmd_module_lock(xip->xi_queue->eq_mod);
1339 	if ((cp = fmd_case_recreate(xip->xi_queue->eq_mod, xp,
1340 	    FMD_CASE_UNSOLVED, uuid, code)) == NULL) {
1341 		if (nelem > 0) {
1342 			for (i = 0; i < nelem; i++)
1343 				nvlist_free(asrua[i]);
1344 			fmd_free(proxy_asru, sizeof (uint8_t) * nelem);
1345 			fmd_free(diag_asru, sizeof (uint8_t) * nelem);
1346 			fmd_free(asrua, sizeof (nvlist_t *) * nelem);
1347 		}
1348 		fmd_module_unlock(xip->xi_queue->eq_mod);
1349 		return;
1350 	}
1351 
1352 	cip = (fmd_case_impl_t *)cp;
1353 	cip->ci_diag_asru = diag_asru;
1354 	cip->ci_proxy_asru = proxy_asru;
1355 	for (i = 0; i < nelem; i++) {
1356 		(void) nvlist_xdup(nvlp[i], &flt_copy, &fmd.d_nva);
1357 		if (proxy_asru[i] != FMD_PROXY_ASRU_NOT_NEEDED) {
1358 			/*
1359 			 * Copy suspects, but remove/replace asru first. Also if
1360 			 * the original asru was hc-scheme use that as resource.
1361 			 */
1362 			if (proxy_asru[i] == FMD_PROXY_ASRU_FROM_ASRU) {
1363 				(void) nvlist_remove(flt_copy,
1364 				    FM_FAULT_RESOURCE, DATA_TYPE_NVLIST);
1365 				(void) nvlist_lookup_nvlist(flt_copy,
1366 				    FM_FAULT_ASRU, &asru);
1367 				(void) nvlist_add_nvlist(flt_copy,
1368 				    FM_FAULT_RESOURCE, asru);
1369 			}
1370 			(void) nvlist_remove(flt_copy, FM_FAULT_ASRU,
1371 			    DATA_TYPE_NVLIST);
1372 			(void) nvlist_add_nvlist(flt_copy, FM_FAULT_ASRU,
1373 			    asrua[i]);
1374 			nvlist_free(asrua[i]);
1375 		} else if (got_hc_asru == 0 &&
1376 		    nvlist_lookup_nvlist(flt_copy, FM_FAULT_ASRU,
1377 		    &asru) == 0 && asru != NULL) {
1378 			/*
1379 			 * If we have an asru from diag side, but it's not
1380 			 * in hc scheme, then we can't be sure what it
1381 			 * represents, so mark as no retire.
1382 			 */
1383 			(void) nvlist_add_boolean_value(flt_copy,
1384 			    FM_SUSPECT_RETIRE, B_FALSE);
1385 		}
1386 		fmd_case_insert_suspect(cp, flt_copy);
1387 	}
1388 	/*
1389 	 * copy diag_time
1390 	 */
1391 	if (nvlist_lookup_int64_array(nvl, FM_SUSPECT_DIAG_TIME, &diag_time,
1392 	    &nelem2) == 0 && nelem2 >= 2)
1393 		fmd_case_settime(cp, diag_time[0], diag_time[1]);
1394 	/*
1395 	 * copy DE fmri
1396 	 */
1397 	if (nvlist_lookup_nvlist(nvl, FM_SUSPECT_DE, &de_fmri) == 0) {
1398 		(void) nvlist_xdup(de_fmri, &de_fmri_dup, &fmd.d_nva);
1399 		fmd_case_set_de_fmri(cp, de_fmri_dup);
1400 	}
1401 
1402 	/*
1403 	 * copy injected if present
1404 	 */
1405 	if (nvlist_lookup_boolean_value(nvl, FM_SUSPECT_INJECTED,
1406 	    &injected) == 0 && injected)
1407 		fmd_case_set_injected(cp);
1408 
1409 	/*
1410 	 * Transition to solved. This will log the suspect list and create
1411 	 * the resource cache entries.
1412 	 */
1413 	fmd_case_transition(cp, FMD_CASE_SOLVED, FMD_CF_SOLVED);
1414 
1415 	/*
1416 	 * Update status if it is not simply "all faulty" (can happen if
1417 	 * list.suspects are being re-sent when the transport has reconnected).
1418 	 */
1419 	(void) nvlist_lookup_uint8_array(nvl, FM_SUSPECT_FAULT_STATUS, &statusp,
1420 	    &nelem);
1421 	for (i = 0; i < nelem; i++) {
1422 		if ((statusp[i] & (FM_SUSPECT_FAULTY | FM_SUSPECT_UNUSABLE |
1423 		    FM_SUSPECT_NOT_PRESENT | FM_SUSPECT_DEGRADED)) !=
1424 		    FM_SUSPECT_FAULTY)
1425 			need_update = 1;
1426 	}
1427 	if (need_update) {
1428 		fmd_case_update_status(cp, statusp, cip->ci_proxy_asru,
1429 		    cip->ci_diag_asru);
1430 		fmd_case_update_containees(cp);
1431 		fmd_case_update(cp);
1432 	}
1433 
1434 	/*
1435 	 * if asru on proxy side, send an update back to the diagnosing side to
1436 	 * update UNUSABLE/DEGRADED.
1437 	 */
1438 	if (got_proxy_asru)
1439 		fmd_case_xprt_updated(cp);
1440 
1441 	if (nelem > 0)
1442 		fmd_free(asrua, sizeof (nvlist_t *) * nelem);
1443 	fmd_module_unlock(xip->xi_queue->eq_mod);
1444 }
1445 
1446 void
1447 fmd_xprt_recv(fmd_xprt_t *xp, nvlist_t *nvl, hrtime_t hrt, boolean_t logonly)
1448 {
1449 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1450 	const fmd_xprt_rule_t *xrp;
1451 	fmd_t *dp = &fmd;
1452 
1453 	fmd_event_t *e;
1454 	char *class, *uuid;
1455 	boolean_t isproto, isereport, isireport, ishvireport, issysevent;
1456 
1457 	uint64_t *tod;
1458 	uint8_t ttl;
1459 	uint_t n;
1460 	fmd_case_t *cp;
1461 
1462 	/*
1463 	 * Grab the transport lock and set the busy flag to indicate we are
1464 	 * busy receiving an event.  If [DI]SUSPEND is pending, wait until fmd
1465 	 * resumes the transport before continuing on with the receive.
1466 	 */
1467 	(void) pthread_mutex_lock(&xip->xi_lock);
1468 
1469 	while (xip->xi_flags & (FMD_XPRT_DSUSPENDED | FMD_XPRT_ISUSPENDED)) {
1470 
1471 		if (fmd.d_signal != 0) {
1472 			(void) pthread_mutex_unlock(&xip->xi_lock);
1473 			return; /* fmd_destroy() is in progress */
1474 		}
1475 
1476 		(void) pthread_cond_wait(&xip->xi_cv, &xip->xi_lock);
1477 	}
1478 
1479 	xip->xi_busy++;
1480 	ASSERT(xip->xi_busy != 0);
1481 
1482 	(void) pthread_mutex_unlock(&xip->xi_lock);
1483 
1484 	(void) pthread_mutex_lock(&xip->xi_stats_lock);
1485 	xip->xi_stats->xs_received.fmds_value.ui64++;
1486 	(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1487 
1488 	if (nvlist_lookup_string(nvl, FM_CLASS, &class) != 0) {
1489 		fmd_error(EFMD_XPRT_PAYLOAD, "discarding nvlist %p: missing "
1490 		    "required \"%s\" payload element", (void *)nvl, FM_CLASS);
1491 
1492 		(void) pthread_mutex_lock(&xip->xi_stats_lock);
1493 		xip->xi_stats->xs_discarded.fmds_value.ui64++;
1494 		(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1495 
1496 		nvlist_free(nvl);
1497 		goto done;
1498 	}
1499 
1500 	fmd_dprintf(FMD_DBG_XPRT, "xprt %u %s %s\n", xip->xi_id,
1501 	    ((logonly == FMD_B_TRUE) ? "logging" : "posting"), class);
1502 
1503 	isereport = (strncmp(class, FM_EREPORT_CLASS ".",
1504 	    sizeof (FM_EREPORT_CLASS)) == 0) ? FMD_B_TRUE : FMD_B_FALSE;
1505 
1506 	isireport = (strncmp(class, FM_IREPORT_CLASS ".",
1507 	    sizeof (FM_IREPORT_CLASS)) == 0) ?  FMD_B_TRUE : FMD_B_FALSE;
1508 
1509 	issysevent = (strncmp(class, SYSEVENT_RSRC_CLASS,
1510 	    sizeof (SYSEVENT_RSRC_CLASS) - 1)) == 0 ? FMD_B_TRUE : FMD_B_FALSE;
1511 
1512 	if (isireport) {
1513 		char *pri;
1514 
1515 		if (nvlist_lookup_string(nvl, FM_IREPORT_PRIORITY, &pri) == 0 &&
1516 		    strncmp(pri, "high", 5) == 0) {
1517 			ishvireport = 1;
1518 		} else {
1519 			ishvireport = 0;
1520 		}
1521 	}
1522 
1523 	/*
1524 	 * The logonly flag should only be set for ereports.
1525 	 */
1526 	if (logonly == FMD_B_TRUE && isereport == FMD_B_FALSE) {
1527 		fmd_error(EFMD_XPRT_INVAL, "discarding nvlist %p: "
1528 		    "logonly flag is not valid for class %s",
1529 		    (void *)nvl, class);
1530 
1531 		(void) pthread_mutex_lock(&xip->xi_stats_lock);
1532 		xip->xi_stats->xs_discarded.fmds_value.ui64++;
1533 		(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1534 
1535 		nvlist_free(nvl);
1536 		goto done;
1537 	}
1538 
1539 	/*
1540 	 * If a time-to-live value is present in the event and is zero, drop
1541 	 * the event and bump xs_timeouts.  Otherwise decrement the TTL value.
1542 	 */
1543 	if (nvlist_lookup_uint8(nvl, FMD_EVN_TTL, &ttl) == 0) {
1544 		if (ttl == 0) {
1545 			fmd_dprintf(FMD_DBG_XPRT, "xprt %u nvlist %p (%s) "
1546 			    "timeout: event received with ttl=0\n",
1547 			    xip->xi_id, (void *)nvl, class);
1548 
1549 			(void) pthread_mutex_lock(&xip->xi_stats_lock);
1550 			xip->xi_stats->xs_timeouts.fmds_value.ui64++;
1551 			(void) pthread_mutex_unlock(&xip->xi_stats_lock);
1552 
1553 			nvlist_free(nvl);
1554 			goto done;
1555 		}
1556 		(void) nvlist_remove(nvl, FMD_EVN_TTL, DATA_TYPE_UINT8);
1557 		(void) nvlist_add_uint8(nvl, FMD_EVN_TTL, ttl - 1);
1558 	}
1559 
1560 	/*
1561 	 * If we are using the native system clock, the underlying transport
1562 	 * code can provide a tighter event time bound by telling us when the
1563 	 * event was enqueued.  If we're using simulated clocks, this time
1564 	 * has no meaning to us, so just reset the value to use HRT_NOW.
1565 	 */
1566 	if (dp->d_clockops != &fmd_timeops_native)
1567 		hrt = FMD_HRT_NOW;
1568 
1569 	/*
1570 	 * If an event's class is in the FMD_CTL_CLASS family, then create a
1571 	 * control event.  If a FMD_EVN_TOD member is found, create a protocol
1572 	 * event using this time.  Otherwise create a protocol event using hrt.
1573 	 */
1574 	isproto = (strncmp(class, FMD_CTL_CLASS, FMD_CTL_CLASS_LEN) == 0) ?
1575 	    FMD_B_FALSE : FMD_B_TRUE;
1576 	if (isproto == FMD_B_FALSE)
1577 		e = fmd_event_create(FMD_EVT_CTL, hrt, nvl, fmd_ctl_init(nvl));
1578 	else if (nvlist_lookup_uint64_array(nvl, FMD_EVN_TOD, &tod, &n) != 0)
1579 		e = fmd_event_create(FMD_EVT_PROTOCOL, hrt, nvl, class);
1580 	else {
1581 		e = fmd_event_recreate(FMD_EVT_PROTOCOL,
1582 		    NULL, nvl, class, NULL, 0, 0);
1583 	}
1584 
1585 	/*
1586 	 * If the debug log is enabled, create a temporary event, log it to the
1587 	 * debug log, and then reset the underlying state of the event.
1588 	 */
1589 	if (xip->xi_log != NULL) {
1590 		fmd_event_impl_t *ep = (fmd_event_impl_t *)e;
1591 
1592 		fmd_log_append(xip->xi_log, e, NULL);
1593 
1594 		ep->ev_flags |= FMD_EVF_VOLATILE;
1595 		ep->ev_off = 0;
1596 		ep->ev_len = 0;
1597 
1598 		if (ep->ev_log != NULL) {
1599 			fmd_log_rele(ep->ev_log);
1600 			ep->ev_log = NULL;
1601 		}
1602 	}
1603 
1604 	/*
1605 	 * Iterate over the rules for the current state trying to match the
1606 	 * event class to one of our special rules.  If a rule is matched, the
1607 	 * event is consumed and not dispatched to other modules.  If the rule
1608 	 * set ends without matching an event, we fall through to dispatching.
1609 	 */
1610 	for (xrp = xip->xi_state; xrp->xr_class != NULL; xrp++) {
1611 		if (fmd_event_match(e, FMD_EVT_PROTOCOL, xrp->xr_class)) {
1612 			fmd_event_hold(e);
1613 			xrp->xr_func(xip, nvl);
1614 			fmd_event_rele(e);
1615 			goto done;
1616 		}
1617 	}
1618 
1619 	/*
1620 	 * Record ereports and ireports in the log.  This code will
1621 	 * be replaced later with a per-transport intent log instead.
1622 	 */
1623 	if (isereport == FMD_B_TRUE || isireport == FMD_B_TRUE ||
1624 	    issysevent == B_TRUE) {
1625 		pthread_rwlock_t *lockp;
1626 		fmd_log_t *lp;
1627 
1628 		if (isereport == FMD_B_TRUE) {
1629 			lp = fmd.d_errlog;
1630 			lockp = &fmd.d_log_lock;
1631 		} else {
1632 			if (ishvireport || issysevent) {
1633 				lp = fmd.d_hvilog;
1634 				lockp = &fmd.d_hvilog_lock;
1635 			} else {
1636 				lp = fmd.d_ilog;
1637 				lockp = &fmd.d_ilog_lock;
1638 			}
1639 		}
1640 
1641 		(void) pthread_rwlock_rdlock(lockp);
1642 		fmd_log_append(lp, e, NULL);
1643 		(void) pthread_rwlock_unlock(lockp);
1644 	}
1645 
1646 	/*
1647 	 * If a list.suspect event is received, create a case for the specified
1648 	 * UUID in the case hash, with the transport module as its owner.
1649 	 */
1650 	if (fmd_event_match(e, FMD_EVT_PROTOCOL, FM_LIST_SUSPECT_CLASS)) {
1651 		if (xip->xi_flags & FMD_XPRT_CACHE_AS_LOCAL)
1652 			fmd_xprt_list_suspect_local(xp, nvl);
1653 		else
1654 			fmd_xprt_list_suspect(xp, nvl);
1655 		fmd_event_hold(e);
1656 		fmd_event_rele(e);
1657 		goto done;
1658 	}
1659 
1660 	/*
1661 	 * If a list.updated or list.repaired event is received, update the
1662 	 * resource cache status and the local case.
1663 	 */
1664 	if (fmd_event_match(e, FMD_EVT_PROTOCOL, FM_LIST_REPAIRED_CLASS) ||
1665 	    fmd_event_match(e, FMD_EVT_PROTOCOL, FM_LIST_UPDATED_CLASS)) {
1666 		uint8_t *statusp;
1667 		uint_t nelem = 0;
1668 
1669 		(void) nvlist_lookup_uint8_array(nvl, FM_SUSPECT_FAULT_STATUS,
1670 		    &statusp, &nelem);
1671 		fmd_module_lock(xip->xi_queue->eq_mod);
1672 		if (nvlist_lookup_string(nvl, FM_SUSPECT_UUID, &uuid) == 0 &&
1673 		    (cp = fmd_case_hash_lookup(fmd.d_cases, uuid)) != NULL) {
1674 			fmd_case_impl_t *cip = (fmd_case_impl_t *)cp;
1675 			if (cip->ci_xprt != NULL) {
1676 				fmd_case_update_status(cp, statusp,
1677 				    cip->ci_proxy_asru, cip->ci_diag_asru);
1678 				fmd_case_update_containees(cp);
1679 				fmd_case_update(cp);
1680 			}
1681 			fmd_case_rele(cp);
1682 		}
1683 		fmd_module_unlock(xip->xi_queue->eq_mod);
1684 		fmd_event_hold(e);
1685 		fmd_event_rele(e);
1686 		goto done;
1687 	}
1688 
1689 	/*
1690 	 * If a list.isolated event is received, update resource cache status
1691 	 */
1692 	if (fmd_event_match(e, FMD_EVT_PROTOCOL, FM_LIST_ISOLATED_CLASS)) {
1693 		uint8_t *statusp;
1694 		uint_t nelem = 0;
1695 
1696 		(void) nvlist_lookup_uint8_array(nvl, FM_SUSPECT_FAULT_STATUS,
1697 		    &statusp, &nelem);
1698 		fmd_module_lock(xip->xi_queue->eq_mod);
1699 		if (nvlist_lookup_string(nvl, FM_SUSPECT_UUID, &uuid) == 0 &&
1700 		    (cp = fmd_case_hash_lookup(fmd.d_cases, uuid)) != NULL) {
1701 			fmd_case_impl_t *cip = (fmd_case_impl_t *)cp;
1702 			if (cip->ci_xprt != NULL)
1703 				fmd_case_update_status(cp, statusp,
1704 				    cip->ci_proxy_asru, cip->ci_diag_asru);
1705 			fmd_case_rele(cp);
1706 		}
1707 		fmd_module_unlock(xip->xi_queue->eq_mod);
1708 		fmd_event_hold(e);
1709 		fmd_event_rele(e);
1710 		goto done;
1711 	}
1712 
1713 	/*
1714 	 * If a list.resolved event is received, resolve the local case.
1715 	 */
1716 	if (fmd_event_match(e, FMD_EVT_PROTOCOL, FM_LIST_RESOLVED_CLASS)) {
1717 		fmd_module_lock(xip->xi_queue->eq_mod);
1718 		if (nvlist_lookup_string(nvl, FM_SUSPECT_UUID, &uuid) == 0 &&
1719 		    (cp = fmd_case_hash_lookup(fmd.d_cases, uuid)) != NULL) {
1720 			fmd_case_impl_t *cip = (fmd_case_impl_t *)cp;
1721 			if (cip->ci_xprt != NULL)
1722 				fmd_case_transition(cp, (cip->ci_state ==
1723 				    FMD_CASE_REPAIRED) ? FMD_CASE_RESOLVED :
1724 				    (cip->ci_state == FMD_CASE_CLOSED) ?
1725 				    FMD_CASE_REPAIRED : FMD_CASE_CLOSE_WAIT,
1726 				    FMD_CF_RESOLVED);
1727 			fmd_case_rele(cp);
1728 		}
1729 		fmd_module_unlock(xip->xi_queue->eq_mod);
1730 		fmd_event_hold(e);
1731 		fmd_event_rele(e);
1732 		goto done;
1733 	}
1734 
1735 	if (logonly == FMD_B_TRUE || (xip->xi_flags & FMD_XPRT_EXTERNAL)) {
1736 		/*
1737 		 * Don't proxy ereports on an EXTERNAL transport - we won't
1738 		 * know how to diagnose them with the wrong topology. Note
1739 		 * that here (and above) we have to hold/release the event in
1740 		 * order for it to be freed.
1741 		 */
1742 		fmd_event_hold(e);
1743 		fmd_event_rele(e);
1744 	} else if (isproto == FMD_B_TRUE)
1745 		fmd_dispq_dispatch(dp->d_disp, e, class);
1746 	else
1747 		fmd_modhash_dispatch(dp->d_mod_hash, e);
1748 done:
1749 	(void) pthread_mutex_lock(&xip->xi_lock);
1750 
1751 	ASSERT(xip->xi_busy != 0);
1752 	xip->xi_busy--;
1753 
1754 	(void) pthread_cond_broadcast(&xip->xi_cv);
1755 	(void) pthread_mutex_unlock(&xip->xi_lock);
1756 }
1757 
1758 void
1759 fmd_xprt_uuclose(fmd_xprt_t *xp, const char *uuid)
1760 {
1761 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1762 
1763 	fmd_event_t *e;
1764 	nvlist_t *nvl;
1765 	char *s;
1766 
1767 	if ((xip->xi_flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
1768 		return; /* read-only transports do not proxy uuclose */
1769 
1770 	TRACE((FMD_DBG_XPRT, "xprt %u closing case %s\n", xip->xi_id, uuid));
1771 
1772 	nvl = fmd_protocol_xprt_uuclose(xip->xi_queue->eq_mod,
1773 	    "resource.fm.xprt.uuclose", xip->xi_version, uuid);
1774 
1775 	(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
1776 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
1777 	fmd_eventq_insert_at_time(xip->xi_queue, e);
1778 }
1779 
1780 /*
1781  * On proxy side, send back uuresolved request to diagnosing side
1782  */
1783 void
1784 fmd_xprt_uuresolved(fmd_xprt_t *xp, const char *uuid)
1785 {
1786 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1787 
1788 	fmd_event_t *e;
1789 	nvlist_t *nvl;
1790 	char *s;
1791 
1792 	if ((xip->xi_flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
1793 		return; /* read-only transports do not proxy uuresolved */
1794 
1795 	TRACE((FMD_DBG_XPRT, "xprt %u resolving case %s\n", xip->xi_id, uuid));
1796 
1797 	nvl = fmd_protocol_xprt_uuresolved(xip->xi_queue->eq_mod,
1798 	    "resource.fm.xprt.uuresolved", xip->xi_version, uuid);
1799 
1800 	(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
1801 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
1802 	fmd_eventq_insert_at_time(xip->xi_queue, e);
1803 }
1804 
1805 /*
1806  * On proxy side, send back repair/acquit/etc request to diagnosing side
1807  */
1808 void
1809 fmd_xprt_updated(fmd_xprt_t *xp, const char *uuid, uint8_t *statusp,
1810 	uint8_t *has_asrup, uint_t nelem)
1811 {
1812 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1813 
1814 	fmd_event_t *e;
1815 	nvlist_t *nvl;
1816 	char *s;
1817 
1818 	if ((xip->xi_flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
1819 		return; /* read-only transports do not support remote repairs */
1820 
1821 	TRACE((FMD_DBG_XPRT, "xprt %u updating case %s\n", xip->xi_id, uuid));
1822 
1823 	nvl = fmd_protocol_xprt_updated(xip->xi_queue->eq_mod,
1824 	    "resource.fm.xprt.updated", xip->xi_version, uuid, statusp,
1825 	    has_asrup, nelem);
1826 
1827 	(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
1828 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
1829 	fmd_eventq_insert_at_time(xip->xi_queue, e);
1830 }
1831 
1832 /*
1833  * Insert the specified class into our remote subscription hash.  If the class
1834  * is already present, bump the reference count; otherwise add it to the hash
1835  * and then enqueue an event for our remote peer to proxy our subscription.
1836  */
1837 void
1838 fmd_xprt_subscribe(fmd_xprt_t *xp, const char *class)
1839 {
1840 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1841 
1842 	uint_t refs;
1843 	nvlist_t *nvl;
1844 	fmd_event_t *e;
1845 	char *s;
1846 
1847 	if ((xip->xi_flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
1848 		return; /* read-only transports do not proxy subscriptions */
1849 
1850 	if (!(xip->xi_flags & FMD_XPRT_SUBSCRIBER))
1851 		return; /* transport is not yet an active subscriber */
1852 
1853 	(void) pthread_mutex_lock(&xip->xi_lock);
1854 	refs = fmd_xprt_class_hash_insert(xip, &xip->xi_rsub, class);
1855 	(void) pthread_mutex_unlock(&xip->xi_lock);
1856 
1857 	if (refs > 1)
1858 		return; /* we've already asked our peer for this subscription */
1859 
1860 	fmd_dprintf(FMD_DBG_XPRT,
1861 	    "xprt %u subscribing to %s\n", xip->xi_id, class);
1862 
1863 	nvl = fmd_protocol_xprt_sub(xip->xi_queue->eq_mod,
1864 	    "resource.fm.xprt.subscribe", xip->xi_version, class);
1865 
1866 	(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
1867 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
1868 	fmd_eventq_insert_at_time(xip->xi_queue, e);
1869 }
1870 
1871 /*
1872  * Delete the specified class from the remote subscription hash.  If the
1873  * reference count drops to zero, ask our remote peer to unsubscribe by proxy.
1874  */
1875 void
1876 fmd_xprt_unsubscribe(fmd_xprt_t *xp, const char *class)
1877 {
1878 	fmd_xprt_impl_t *xip = (fmd_xprt_impl_t *)xp;
1879 
1880 	uint_t refs;
1881 	nvlist_t *nvl;
1882 	fmd_event_t *e;
1883 	char *s;
1884 
1885 	if ((xip->xi_flags & FMD_XPRT_RDWR) == FMD_XPRT_RDONLY)
1886 		return; /* read-only transports do not proxy subscriptions */
1887 
1888 	if (!(xip->xi_flags & FMD_XPRT_SUBSCRIBER))
1889 		return; /* transport is not yet an active subscriber */
1890 
1891 	/*
1892 	 * If the subscription reference count drops to zero in xi_rsub, insert
1893 	 * an entry into the xi_usub hash indicating we await an unsuback event.
1894 	 */
1895 	(void) pthread_mutex_lock(&xip->xi_lock);
1896 
1897 	if ((refs = fmd_xprt_class_hash_delete(xip, &xip->xi_rsub, class)) == 0)
1898 		(void) fmd_xprt_class_hash_insert(xip, &xip->xi_usub, class);
1899 
1900 	(void) pthread_mutex_unlock(&xip->xi_lock);
1901 
1902 	if (refs != 0)
1903 		return; /* other subscriptions for this class still active */
1904 
1905 	fmd_dprintf(FMD_DBG_XPRT,
1906 	    "xprt %u unsubscribing from %s\n", xip->xi_id, class);
1907 
1908 	nvl = fmd_protocol_xprt_sub(xip->xi_queue->eq_mod,
1909 	    "resource.fm.xprt.unsubscribe", xip->xi_version, class);
1910 
1911 	(void) nvlist_lookup_string(nvl, FM_CLASS, &s);
1912 	e = fmd_event_create(FMD_EVT_PROTOCOL, FMD_HRT_NOW, nvl, s);
1913 	fmd_eventq_insert_at_time(xip->xi_queue, e);
1914 }
1915 
1916 static void
1917 fmd_xprt_subscribe_xid(fmd_idspace_t *ids, id_t id, void *class)
1918 {
1919 	fmd_xprt_t *xp;
1920 
1921 	if ((xp = fmd_idspace_hold(ids, id)) != NULL) {
1922 		fmd_xprt_subscribe(xp, class);
1923 		fmd_idspace_rele(ids, id);
1924 	}
1925 }
1926 
1927 void
1928 fmd_xprt_subscribe_all(const char *class)
1929 {
1930 	fmd_idspace_t *ids = fmd.d_xprt_ids;
1931 
1932 	if (ids->ids_count != 0)
1933 		fmd_idspace_apply(ids, fmd_xprt_subscribe_xid, (void *)class);
1934 }
1935 
1936 static void
1937 fmd_xprt_unsubscribe_xid(fmd_idspace_t *ids, id_t id, void *class)
1938 {
1939 	fmd_xprt_t *xp;
1940 
1941 	if ((xp = fmd_idspace_hold(ids, id)) != NULL) {
1942 		fmd_xprt_unsubscribe(xp, class);
1943 		fmd_idspace_rele(ids, id);
1944 	}
1945 }
1946 
1947 void
1948 fmd_xprt_unsubscribe_all(const char *class)
1949 {
1950 	fmd_idspace_t *ids = fmd.d_xprt_ids;
1951 
1952 	if (ids->ids_count != 0)
1953 		fmd_idspace_apply(ids, fmd_xprt_unsubscribe_xid, (void *)class);
1954 }
1955 
1956 /*ARGSUSED*/
1957 static void
1958 fmd_xprt_suspend_xid(fmd_idspace_t *ids, id_t id, void *arg)
1959 {
1960 	fmd_xprt_t *xp;
1961 
1962 	if ((xp = fmd_idspace_hold(ids, id)) != NULL) {
1963 		fmd_xprt_xsuspend(xp, FMD_XPRT_DSUSPENDED);
1964 		fmd_idspace_rele(ids, id);
1965 	}
1966 }
1967 
1968 void
1969 fmd_xprt_suspend_all(void)
1970 {
1971 	fmd_idspace_t *ids = fmd.d_xprt_ids;
1972 
1973 	(void) pthread_mutex_lock(&fmd.d_xprt_lock);
1974 
1975 	if (fmd.d_xprt_suspend++ != 0) {
1976 		(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
1977 		return; /* already suspended */
1978 	}
1979 
1980 	if (ids->ids_count != 0)
1981 		fmd_idspace_apply(ids, fmd_xprt_suspend_xid, NULL);
1982 
1983 	(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
1984 }
1985 
1986 /*ARGSUSED*/
1987 static void
1988 fmd_xprt_resume_xid(fmd_idspace_t *ids, id_t id, void *arg)
1989 {
1990 	fmd_xprt_t *xp;
1991 
1992 	if ((xp = fmd_idspace_hold(ids, id)) != NULL) {
1993 		fmd_xprt_xresume(xp, FMD_XPRT_DSUSPENDED);
1994 		fmd_idspace_rele(ids, id);
1995 	}
1996 }
1997 
1998 void
1999 fmd_xprt_resume_all(void)
2000 {
2001 	fmd_idspace_t *ids = fmd.d_xprt_ids;
2002 
2003 	(void) pthread_mutex_lock(&fmd.d_xprt_lock);
2004 
2005 	if (fmd.d_xprt_suspend == 0)
2006 		fmd_panic("fmd_xprt_suspend/resume_all mismatch\n");
2007 
2008 	if (--fmd.d_xprt_suspend != 0) {
2009 		(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
2010 		return; /* not ready to be resumed */
2011 	}
2012 
2013 	if (ids->ids_count != 0)
2014 		fmd_idspace_apply(ids, fmd_xprt_resume_xid, NULL);
2015 
2016 	(void) pthread_mutex_unlock(&fmd.d_xprt_lock);
2017 }
2018