xref: /freebsd/sys/kern/kern_boottrace.c (revision ba3c1f5972d7b90feb6e6da47905ff2757e0fe57)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause
3  *
4  * Copyright (c) 2008-2022 NetApp, Inc.
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice, this list of conditions and the following disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  *
15  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
16  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
17  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
18  * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
19  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
20  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
21  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
22  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
23  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
24  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
25  */
26 
27 #include <sys/param.h>
28 #include <sys/systm.h>
29 #include <sys/boottrace.h>
30 #include <sys/kernel.h>
31 #include <sys/malloc.h>
32 #include <sys/pcpu.h>
33 #include <sys/proc.h>
34 #include <sys/resourcevar.h>
35 #include <sys/sbuf.h>
36 #include <sys/syscallsubr.h>
37 #include <sys/sysctl.h>
38 #include <sys/time.h>
39 
40 #include <machine/stdarg.h>
41 
42 #define	dprintf(fmt, ...)						\
43 	do {								\
44 		if (dotrace_debugging)					\
45 			printf(fmt, ##__VA_ARGS__);			\
46 	} while (0);
47 
48 static MALLOC_DEFINE(M_BOOTTRACE, "boottrace", "memory for boot tracing");
49 
50 #define	BT_TABLE_DEFSIZE	3000
51 #define	BT_TABLE_RUNSIZE	2000
52 #define	BT_TABLE_SHTSIZE	1000
53 #define	BT_TABLE_MINSIZE	500
54 
55 /*
56  * Boot-time & shutdown-time event.
57  */
58 struct bt_event {
59 	uint64_t tsc;			/* Timestamp */
60 	uint64_t tick;			/* Kernel tick */
61 	uint64_t cputime;		/* Microseconds of process CPU time */
62 	uint32_t cpuid;			/* CPU on which the event ran */
63 	uint32_t inblock;		/* # of blocks in */
64 	uint32_t oublock;		/* # of blocks out */
65 	pid_t pid;			/* Current PID */
66 	char name[BT_EVENT_NAMELEN];	/* Event name */
67 	char tdname[BT_EVENT_TDNAMELEN]; /* Thread name */
68 };
69 
70 struct bt_table {
71 	uint32_t size;			/* Trace table size */
72 	uint32_t curr;			/* Trace entry to use */
73 	uint32_t wrap;			/* Wrap-around, instead of dropping */
74 	uint32_t drops_early;		/* Trace entries dropped before init */
75 	uint32_t drops_full;		/* Trace entries dropped after full */
76 	struct bt_event *table;
77 };
78 
79 /* Boot-time tracing */
80 static struct bt_table bt;
81 
82 /* Run-time tracing */
83 static struct bt_table rt;
84 
85 /* Shutdown-time tracing */
86 static struct bt_table st;
87 
88 /* Set when system boot is complete, and we've switched to runtime tracing. */
89 static bool bootdone;
90 
91 /* Set when system shutdown has started. */
92 static bool shutdown;
93 
94 /* Turn on dotrace() debug logging to console. */
95 static bool dotrace_debugging;
96 TUNABLE_BOOL("debug.boottrace.dotrace_debugging", &dotrace_debugging);
97 
98 /* Trace kernel events */
99 static bool dotrace_kernel = true;
100 TUNABLE_BOOL("kern.boottrace.dotrace_kernel", &dotrace_kernel);
101 
102 /* Trace userspace events */
103 static bool dotrace_user = true;
104 TUNABLE_BOOL("kern.boottrace.dotrace_user", &dotrace_user);
105 
106 static int sysctl_log(SYSCTL_HANDLER_ARGS);
107 static int sysctl_boottrace(SYSCTL_HANDLER_ARGS);
108 static int sysctl_runtrace(SYSCTL_HANDLER_ARGS);
109 static int sysctl_shuttrace(SYSCTL_HANDLER_ARGS);
110 static int sysctl_boottrace_reset(SYSCTL_HANDLER_ARGS);
111 
112 SYSCTL_NODE(_kern, OID_AUTO, boottrace, CTLFLAG_RD | CTLFLAG_MPSAFE, 0,
113     "boottrace statistics");
114 
115 SYSCTL_PROC(_kern_boottrace, OID_AUTO, log,
116     CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE | CTLFLAG_SKIP,
117     0, 0, sysctl_log, "A",
118     "Print a log of the boottrace trace data");
119 SYSCTL_PROC(_kern_boottrace, OID_AUTO, boottrace,
120     CTLTYPE_STRING | CTLFLAG_WR | CTLFLAG_MPSAFE,
121     0, 0, sysctl_boottrace, "A",
122     "Capture a boot-time trace event");
123 SYSCTL_PROC(_kern_boottrace, OID_AUTO, runtrace,
124     CTLTYPE_STRING | CTLFLAG_WR | CTLFLAG_MPSAFE,
125     0, 0, sysctl_runtrace, "A",
126     "Capture a run-time trace event");
127 SYSCTL_PROC(_kern_boottrace, OID_AUTO, shuttrace,
128     CTLTYPE_STRING | CTLFLAG_WR | CTLFLAG_MPSAFE,
129     0, 0, sysctl_shuttrace, "A",
130     "Capture a shutdown-time trace event");
131 SYSCTL_PROC(_kern_boottrace, OID_AUTO, reset,
132     CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_MPSAFE,
133     0, 0, sysctl_boottrace_reset, "A",
134     "Reset run-time tracing table");
135 
136 /*
137  * Global enable.
138  */
139 bool __read_mostly boottrace_enabled = false;
140 SYSCTL_BOOL(_kern_boottrace, OID_AUTO, enabled, CTLFLAG_RDTUN,
141     &boottrace_enabled, 0,
142     "Boot-time and shutdown-time tracing enabled");
143 
144 /*
145  * Enable dumping of the shutdown trace entries to console.
146  */
147 bool shutdown_trace = false;
148 SYSCTL_BOOL(_kern_boottrace, OID_AUTO, shutdown_trace, CTLFLAG_RWTUN,
149    &shutdown_trace, 0,
150    "Enable kernel shutdown tracing to the console");
151 
152 /*
153  * Set the delta threshold (ms) below which events are ignored, used in
154  * determining what to dump to the console.
155  */
156 static int shutdown_trace_threshold;
157 SYSCTL_INT(_kern_boottrace, OID_AUTO, shutdown_trace_threshold, CTLFLAG_RWTUN,
158    &shutdown_trace_threshold, 0,
159    "Tracing threshold (ms) below which tracing is ignored");
160 
161 /*
162  * Dump a trace to buffer or if buffer is NULL to console.
163  *
164  * Non-zero delta_threshold selectively prints entries based on delta
165  * with current and previous entry. Otherwise, delta_threshold of 0
166  * prints every trace entry and delta.
167  *
168  * Output something like this:
169  *
170  * CPU      msecs      delta process                  event
171  *  11 1228262715          0 init                     shutdown pre sync begin
172  *   3 1228265622       2907 init                     shutdown pre sync complete
173  *   3 1228265623          0 init                     shutdown turned swap off
174  *  18 1228266466        843 init                     shutdown unmounted all filesystems
175  *
176  * How to interpret:
177  *
178  * delta column represents the time in milliseconds between this event and the previous.
179  * Usually that means you can take the previous event, current event, match them
180  * up in the code, and whatever lies between is the culprit taking time.
181  *
182  * For example, above: Pre-Sync is taking 2907ms, and something between swap and unmount
183  * filesystems is taking 843 milliseconds.
184  *
185  * An event with a delta of 0 are 'landmark' events that simply exist in the output
186  * for the developer to know where the time measurement begins. The 0 is an arbitrary
187  * number that can effectively be ignored.
188  */
189 #define BTC_DELTA_PRINT(bte, msecs, delta) do {				\
190 	if (sbp) {							\
191 		sbuf_printf(sbp, fmt, (bte)->cpuid, msecs, delta,	\
192 		    (bte)->tdname, (bte)->name, (bte)->pid,		\
193 		    (bte)->cputime / 1000000,				\
194 		    ((bte)->cputime % 1000000) / 10000,			\
195 		    (bte)->inblock, (bte)->oublock);			\
196 	} else {							\
197 		printf(fmt, (bte)->cpuid, msecs, delta,			\
198 		    (bte)->tdname, (bte)->name, (bte)->pid,		\
199 		    (bte)->cputime / 1000000,				\
200 		    ((bte)->cputime % 1000000) / 10000,			\
201 		    (bte)->inblock, (bte)->oublock);			\
202 	}								\
203 } while (0)
204 
205 /*
206  * Print the trace entries to the message buffer, or to an sbuf, if provided.
207  *
208  * Entries with a difference less than dthres will not be printed.
209  */
210 static void
211 boottrace_display(struct sbuf *sbp, struct bt_table *btp, uint64_t dthres)
212 {
213 	struct bt_event *evtp;
214 	struct bt_event *last_evtp;
215 	uint64_t msecs;
216 	uint64_t first_msecs;
217 	uint64_t last_msecs;
218 	uint64_t dmsecs;
219 	uint64_t last_dmsecs;
220 	uint64_t total_dmsecs;
221 	uint32_t i;
222 	uint32_t curr;
223 	const char *fmt     = "%3u %10llu %10llu %-24s %-40s %5d %4d.%02d %5u %5u\n";
224 	const char *hdr_fmt = "\n\n%3s %10s %10s %-24s %-40s %5s %6s %5s %5s\n";
225 	bool printed;
226 	bool last_printed;
227 
228 	/* Print the header */
229 	if (sbp != NULL)
230 		sbuf_printf(sbp, hdr_fmt,
231 		    "CPU", "msecs", "delta", "process",
232 		    "event", "PID", "CPUtime", "IBlks", "OBlks");
233 	else
234 		printf(hdr_fmt,
235 		    "CPU", "msecs", "delta", "process",
236 		    "event", "PID", "CPUtime", "IBlks", "OBlks");
237 
238 	first_msecs = 0;
239 	last_evtp = NULL;
240 	last_msecs = 0;
241 	last_dmsecs = 0;
242 	last_printed = false;
243 	i = curr = btp->curr;
244 
245 	do {
246 		evtp = &btp->table[i];
247 		if (evtp->tsc == 0)
248 			goto next;
249 
250 		msecs = cputick2usec(evtp->tick) / 1000;
251 		dmsecs = (last_msecs != 0 && msecs > last_msecs) ?
252 		    msecs - last_msecs : 0;
253 		printed = false;
254 
255 		/*
256 		 * If a threshold is defined, start filtering events by the
257 		 * delta of msecs.
258 		 */
259 		if (dthres != 0 && (dmsecs > dthres)) {
260 			/*
261 			 * Print the previous entry as a landmark, even if it
262 			 * falls below the threshold.
263 			 */
264 			if (last_evtp != NULL && !last_printed)
265 				BTC_DELTA_PRINT(last_evtp, last_msecs,
266 				    last_dmsecs);
267 			BTC_DELTA_PRINT(evtp, msecs, dmsecs);
268 			printed = true;
269 		} else if (dthres == 0) {
270 			BTC_DELTA_PRINT(evtp, msecs, dmsecs);
271 			printed = true;
272 		}
273 		if (first_msecs == 0 || msecs < first_msecs)
274 			first_msecs = msecs;
275 		last_evtp = evtp;
276 		last_msecs = msecs;
277 		last_dmsecs = dmsecs;
278 		last_printed = printed;
279 		maybe_yield();
280 next:
281 		i = (i + 1) % btp->size;
282 	} while (i != curr);
283 
284 	total_dmsecs = last_msecs > first_msecs ?
285 	    (last_msecs - first_msecs) : 0;
286 	if (sbp != NULL)
287 		sbuf_printf(sbp, "Total measured time: %ju msecs\n",
288 		    (uintmax_t)total_dmsecs);
289 	else
290 		printf("Total measured time: %ju msecs\n",
291 		    (uintmax_t)total_dmsecs);
292 }
293 
294 /*
295  * Dump trace table entries to the console, given a delta threshold.
296  */
297 void
298 boottrace_dump_console(void)
299 {
300 	if (!boottrace_enabled)
301 		return;
302 
303 	if (shutdown || rebooting || KERNEL_PANICKED()) {
304 		boottrace_display(NULL, &st, shutdown_trace_threshold);
305 	} else {
306 		boottrace_display(NULL, &bt, 0);
307 		boottrace_display(NULL, &rt, 0);
308 	}
309 }
310 
311 /*
312  * Records a new tracing event to the specified table.
313  *
314  * We don't use a lock because we want this to be callable from interrupt
315  * context.
316  */
317 static int
318 dotrace(struct bt_table *btp, const char *eventname, const char *tdname)
319 {
320 	uint32_t idx, nxt;
321 	struct rusage usage;
322 
323 	MPASS(boottrace_enabled);
324 	if (tdname == NULL)
325 		tdname = (curproc->p_flag & P_SYSTEM) ?
326 		    curthread->td_name : curproc->p_comm;
327 
328 	dprintf("dotrace[");
329 	dprintf("cpu=%u, pid=%d, tsc=%ju, tick=%d, td='%s', event='%s'",
330 	    PCPU_GET(cpuid), curthread->td_proc->p_pid,
331 	    (uintmax_t)get_cyclecount(), ticks, tdname, eventname);
332 	if (btp->table == NULL) {
333 		btp->drops_early++;
334 		dprintf(", return=ENOSPC_1]\n");
335 		return (ENOSPC);
336 	}
337 
338 	/* Claim a slot in the table. */
339 	do {
340 		idx = btp->curr;
341 		nxt = (idx + 1) % btp->size;
342 		if (nxt == 0 && btp->wrap == 0) {
343 			btp->drops_full++;
344 			dprintf(", return=ENOSPC_2]\n");
345 			return (ENOSPC);
346 		}
347 	} while (!atomic_cmpset_int(&btp->curr, idx, nxt));
348 
349 	btp->table[idx].cpuid = PCPU_GET(cpuid);
350 	btp->table[idx].tsc = get_cyclecount(),
351 	btp->table[idx].tick = cpu_ticks();
352 	btp->table[idx].pid = curthread->td_proc->p_pid;
353 
354 	/*
355 	 * Get the resource usage for the process. Don't attempt this for the
356 	 * kernel proc0 or for critical section activities.
357 	 */
358 	if ((curthread->td_proc == &proc0) || (curthread->td_critnest != 0)) {
359 		btp->table[idx].cputime = 0;
360 		btp->table[idx].inblock = 0;
361 		btp->table[idx].oublock = 0;
362 	} else {
363 		kern_getrusage(curthread, RUSAGE_CHILDREN, &usage);
364 		btp->table[idx].cputime =
365 		    (uint64_t)(usage.ru_utime.tv_sec * 1000000 +
366 		    usage.ru_utime.tv_usec + usage.ru_stime.tv_sec * 1000000 +
367 		    usage.ru_stime.tv_usec);
368 		btp->table[idx].inblock = (uint32_t)usage.ru_inblock;
369 		btp->table[idx].oublock = (uint32_t)usage.ru_oublock;
370 	}
371 	strlcpy(btp->table[idx].name, eventname, BT_EVENT_NAMELEN);
372 	strlcpy(btp->table[idx].tdname, tdname, BT_EVENT_TDNAMELEN);
373 
374 	dprintf(", return=0]\n");
375 	return (0);
376 }
377 
378 /*
379  * Log various boot-time events, with the trace message encoded using
380  * printf-like arguments.
381  */
382 int
383 boottrace(const char *tdname, const char *fmt, ...)
384 {
385 	char eventname[BT_EVENT_NAMELEN];
386 	struct bt_table *btp;
387 	va_list ap;
388 
389 	if (!dotrace_kernel)
390 		 return (ENXIO);
391 
392 	va_start(ap, fmt);
393 	vsnprintf(eventname, sizeof(eventname), fmt, ap);
394 	va_end(ap);
395 
396 	btp = &bt;
397 	if (bootdone)
398 		btp = &rt;
399 	if (shutdown || rebooting || KERNEL_PANICKED())
400 		btp = &st;
401 
402 	return (dotrace(btp, eventname, tdname));
403 }
404 
405 /*
406  * Log a run-time event & switch over to run-time tracing mode.
407  */
408 static int
409 runtrace(const char *eventname, const char *tdname)
410 {
411 
412 	bootdone = true;
413 	return (dotrace(&rt, eventname, tdname));
414 }
415 
416 /*
417  * Parse a boottrace message from userspace.
418  *
419  * The input from may contain a ':' to denote tdname. If not, tdname is
420  * inferred from the process' name.
421  *
422  * e.g. reboot(8):SIGINT to init(8)
423  */
424 static void
425 boottrace_parse_message(char *message, char **eventname, char **tdname)
426 {
427 	char *delim;
428 
429 	delim = strchr(message, ':');
430 	if (delim != NULL) {
431 		*delim = '\0';
432 		*tdname = message;
433 		*eventname = ++delim;
434 	} else {
435 		*tdname = curproc->p_comm;
436 		*eventname = message;
437 	}
438 }
439 
440 static int
441 _boottrace_sysctl(struct bt_table *btp, struct sysctl_oid *oidp,
442     struct sysctl_req *req)
443 {
444 	char message[BT_MSGLEN];
445 	char *eventname, *tdname;
446 	int error;
447 
448 	if (!dotrace_user)
449 	       return (ENXIO);
450 
451 	message[0] = '\0';
452 	error = sysctl_handle_string(oidp, message, sizeof(message), req);
453 	if (error)
454 		return (error);
455 
456 	boottrace_parse_message(message, &eventname, &tdname);
457 	error = dotrace(btp, eventname, tdname);
458 	if (error == ENOSPC) {
459 		/* Ignore table full error. */
460 		error = 0;
461 	}
462 	return (error);
463 }
464 
465 static int
466 sysctl_log(SYSCTL_HANDLER_ARGS)
467 {
468 	struct sbuf *sbuf;
469 	int error;
470 
471 	if (!boottrace_enabled || req->newptr != NULL)
472 		return (0);
473 
474 	sbuf = sbuf_new(NULL, NULL, 0, SBUF_AUTOEXTEND);
475 	boottrace_display(sbuf, &bt, 0);
476 	boottrace_display(sbuf, &rt, 0);
477 	sbuf_finish(sbuf);
478 	error = SYSCTL_OUT(req, sbuf_data(sbuf), sbuf_len(sbuf));
479 	sbuf_delete(sbuf);
480 
481 	return (error);
482 }
483 
484 static int
485 sysctl_boottrace(SYSCTL_HANDLER_ARGS)
486 {
487 
488 	if (!boottrace_enabled)
489 		return (0);
490 
491 	/* No output */
492 	if (req->newptr == NULL)
493 		return (0);
494 
495 	/* Trace to rt if we have reached multi-user. */
496 	return (_boottrace_sysctl(bootdone ? &rt : &bt, oidp, req));
497 }
498 
499 /*
500  * Log a run-time event and switch over to run-time tracing mode.
501  */
502 static int
503 sysctl_runtrace(SYSCTL_HANDLER_ARGS)
504 {
505 
506 	if (!boottrace_enabled)
507 		return (0);
508 
509 	/* No output */
510 	if (req->newptr == NULL)
511 		return (0);
512 
513 	bootdone = true;
514 	return (_boottrace_sysctl(&rt, oidp, req));
515 }
516 
517 /*
518  * Log a shutdown-time event and switch over to shutdown tracing mode.
519  */
520 static int
521 sysctl_shuttrace(SYSCTL_HANDLER_ARGS)
522 {
523 
524 	if (!boottrace_enabled)
525 		return (0);
526 
527 	/* No output */
528 	if (req->newptr == NULL)
529 		return (0);
530 
531 	shutdown = true;
532 	return (_boottrace_sysctl(&st, oidp, req));
533 }
534 
535 /*
536  * Reset the runtime tracing buffer.
537  */
538 void
539 boottrace_reset(const char *actor)
540 {
541 	char tmpbuf[64];
542 
543 	snprintf(tmpbuf, sizeof(tmpbuf), "reset: %s", actor);
544 	runtrace(tmpbuf, NULL);
545 }
546 
547 /*
548  * Note that a resize implies a reset, i.e., the index is reset to 0.
549  * We never shrink the array; we can only increase its size.
550  */
551 int
552 boottrace_resize(u_int newsize)
553 {
554 
555 	if (newsize <= rt.size) {
556 		return (EINVAL);
557 	}
558 	rt.table = realloc(rt.table, newsize * sizeof(struct bt_event),
559 	    M_BOOTTRACE, M_WAITOK | M_ZERO);
560 	if (rt.table == NULL)
561 		return (ENOMEM);
562 
563 	rt.size = newsize;
564 	boottrace_reset("boottrace_resize");
565 	return (0);
566 }
567 
568 static int
569 sysctl_boottrace_reset(SYSCTL_HANDLER_ARGS)
570 {
571 
572 	if (!boottrace_enabled)
573 		return (0);
574 
575 	if (req->newptr != NULL)
576 		boottrace_reset("sysctl_boottrace_reset");
577 
578 	return (0);
579 }
580 
581 static void
582 boottrace_init(void)
583 {
584 
585 	if (!boottrace_enabled)
586 		return;
587 
588 	/* Boottime trace table */
589 	bt.size = BT_TABLE_DEFSIZE;
590 	TUNABLE_INT_FETCH("kern.boottrace.table_size", &bt.size);
591 	bt.size = max(bt.size, BT_TABLE_MINSIZE);
592 	bt.table = malloc(bt.size * sizeof(struct bt_event), M_BOOTTRACE,
593 	    M_WAITOK | M_ZERO);
594 
595 	/* Stick in an initial entry. */
596 	bt.table[0].cpuid = PCPU_GET(cpuid);
597 	strlcpy(bt.table[0].tdname, "boottime", BT_EVENT_TDNAMELEN);
598 	strlcpy(bt.table[0].name, "initial event", BT_EVENT_NAMELEN);
599 	bt.curr = 1;
600 
601 	/* Run-time trace table (may wrap-around). */
602 	rt.wrap = 1;
603 	rt.size = BT_TABLE_RUNSIZE;
604 	rt.table = malloc(rt.size * sizeof(struct bt_event), M_BOOTTRACE,
605 	    M_WAITOK | M_ZERO);
606 
607 	/* Shutdown trace table */
608 	st.size = BT_TABLE_SHTSIZE;
609 	st.table = malloc(st.size * sizeof(struct bt_event), M_BOOTTRACE,
610 	    M_WAITOK | M_ZERO);
611 }
612 SYSINIT(boottrace, SI_SUB_CPU, SI_ORDER_ANY, boottrace_init, 0);
613