xref: /freebsd/sys/kern/kern_tslog.c (revision bdd1243d)
1 /*-
2  * Copyright (c) 2017 Colin Percival
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  *
14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24  * SUCH DAMAGE.
25  */
26 
27 #include <sys/cdefs.h>
28 __FBSDID("$FreeBSD$");
29 
30 #include <sys/param.h>
31 #include <sys/kernel.h>
32 #include <sys/linker.h>
33 #include <sys/malloc.h>
34 #include <sys/proc.h>
35 #include <sys/sbuf.h>
36 #include <sys/sysctl.h>
37 #include <sys/systm.h>
38 #include <sys/tslog.h>
39 
40 #include <machine/atomic.h>
41 #include <machine/cpu.h>
42 
43 #ifndef TSLOGSIZE
44 #define TSLOGSIZE 262144
45 #endif
46 
47 static volatile long nrecs = 0;
48 static struct timestamp {
49 	void * td;
50 	int type;
51 	const char * f;
52 	const char * s;
53 	uint64_t tsc;
54 } timestamps[TSLOGSIZE];
55 
56 void
57 tslog(void * td, int type, const char * f, const char * s)
58 {
59 	uint64_t tsc = get_cyclecount();
60 	long pos;
61 
62 	/* A NULL thread is thread0 before curthread is set. */
63 	if (td == NULL)
64 		td = &thread0;
65 
66 	/* Grab a slot. */
67 	pos = atomic_fetchadd_long(&nrecs, 1);
68 
69 	/* Store record. */
70 	if (pos < nitems(timestamps)) {
71 		timestamps[pos].td = td;
72 		timestamps[pos].type = type;
73 		timestamps[pos].f = f;
74 		timestamps[pos].s = s;
75 		timestamps[pos].tsc = tsc;
76 	}
77 }
78 
79 static int
80 sysctl_debug_tslog(SYSCTL_HANDLER_ARGS)
81 {
82 	int error;
83 	struct sbuf *sb;
84 	size_t i, limit;
85 	caddr_t loader_tslog;
86 	void * loader_tslog_buf;
87 	size_t loader_tslog_len;
88 
89 	/*
90 	 * This code can race against the code in tslog() which stores
91 	 * records: Theoretically we could end up reading a record after
92 	 * its slots have been reserved but before it has been written.
93 	 * Since this code takes orders of magnitude longer to run than
94 	 * tslog() takes to write a record, it is highly unlikely that
95 	 * anyone will ever experience this race.
96 	 */
97 	sb = sbuf_new_for_sysctl(NULL, NULL, 1024, req);
98 
99 	/* Get data from the boot loader, if it provided any. */
100 	loader_tslog = preload_search_by_type("TSLOG data");
101 	if (loader_tslog != NULL) {
102 		loader_tslog_buf = preload_fetch_addr(loader_tslog);
103 		loader_tslog_len = preload_fetch_size(loader_tslog);
104 		sbuf_bcat(sb, loader_tslog_buf, loader_tslog_len);
105 	}
106 
107 	/* Add data logged within the kernel. */
108 	limit = MIN(nrecs, nitems(timestamps));
109 	for (i = 0; i < limit; i++) {
110 		sbuf_printf(sb, "%p", timestamps[i].td);
111 		sbuf_printf(sb, " %llu",
112 		    (unsigned long long)timestamps[i].tsc);
113 		switch (timestamps[i].type) {
114 		case TS_ENTER:
115 			sbuf_printf(sb, " ENTER");
116 			break;
117 		case TS_EXIT:
118 			sbuf_printf(sb, " EXIT");
119 			break;
120 		case TS_THREAD:
121 			sbuf_printf(sb, " THREAD");
122 			break;
123 		case TS_EVENT:
124 			sbuf_printf(sb, " EVENT");
125 			break;
126 		}
127 		sbuf_printf(sb, " %s", timestamps[i].f ? timestamps[i].f : "(null)");
128 		if (timestamps[i].s)
129 			sbuf_printf(sb, " %s\n", timestamps[i].s);
130 		else
131 			sbuf_printf(sb, "\n");
132 	}
133 	error = sbuf_finish(sb);
134 	sbuf_delete(sb);
135 	return (error);
136 }
137 
138 SYSCTL_PROC(_debug, OID_AUTO, tslog,
139     CTLTYPE_STRING|CTLFLAG_RD|CTLFLAG_MPSAFE|CTLFLAG_SKIP,
140     0, 0, sysctl_debug_tslog, "", "Dump recorded event timestamps");
141 
142 MALLOC_DEFINE(M_TSLOGUSER, "tsloguser", "Strings used by userland tslog");
143 static struct procdata {
144 	pid_t ppid;
145 	uint64_t tsc_forked;
146 	uint64_t tsc_exited;
147 	char * execname;
148 	char * namei;
149 	int reused;
150 } procs[PID_MAX + 1];
151 
152 void
153 tslog_user(pid_t pid, pid_t ppid, const char * execname, const char * namei)
154 {
155 	uint64_t tsc = get_cyclecount();
156 
157 	/* If we wrapped, do nothing. */
158 	if (procs[pid].reused)
159 		return;
160 
161 	/* If we have a ppid, we're recording a fork. */
162 	if (ppid != (pid_t)(-1)) {
163 		/* If we have a ppid already, we wrapped. */
164 		if (procs[pid].ppid) {
165 			procs[pid].reused = 1;
166 			return;
167 		}
168 
169 		/* Fill in some fields. */
170 		procs[pid].ppid = ppid;
171 		procs[pid].tsc_forked = tsc;
172 		return;
173 	}
174 
175 	/* If we have an execname, record it. */
176 	if (execname != NULL) {
177 		if (procs[pid].execname != NULL)
178 			free(procs[pid].execname, M_TSLOGUSER);
179 		procs[pid].execname = strdup(execname, M_TSLOGUSER);
180 		return;
181 	}
182 
183 	/* Record the first namei for the process. */
184 	if (namei != NULL) {
185 		if (procs[pid].namei == NULL)
186 			procs[pid].namei = strdup(namei, M_TSLOGUSER);
187 		return;
188 	}
189 
190 	/* Otherwise we're recording an exit. */
191 	procs[pid].tsc_exited = tsc;
192 }
193 
194 static int
195 sysctl_debug_tslog_user(SYSCTL_HANDLER_ARGS)
196 {
197 	int error;
198 	struct sbuf *sb;
199 	pid_t pid;
200 
201 	sb = sbuf_new_for_sysctl(NULL, NULL, 1024, req);
202 
203 	/* Export the data we logged. */
204 	for (pid = 0; pid <= PID_MAX; pid++) {
205 		sbuf_printf(sb, "%zu", (size_t)pid);
206 		sbuf_printf(sb, " %zu", (size_t)procs[pid].ppid);
207 		sbuf_printf(sb, " %llu",
208 		    (unsigned long long)procs[pid].tsc_forked);
209 		sbuf_printf(sb, " %llu",
210 		    (unsigned long long)procs[pid].tsc_exited);
211 		sbuf_printf(sb, " \"%s\"", procs[pid].execname ?
212 		    procs[pid].execname : "");
213 		sbuf_printf(sb, " \"%s\"", procs[pid].namei ?
214 		    procs[pid].namei : "");
215 		sbuf_printf(sb, "\n");
216 	}
217 	error = sbuf_finish(sb);
218 	sbuf_delete(sb);
219 	return (error);
220 }
221 
222 SYSCTL_PROC(_debug, OID_AUTO, tslog_user,
223     CTLTYPE_STRING|CTLFLAG_RD|CTLFLAG_MPSAFE|CTLFLAG_SKIP,
224     0, 0, sysctl_debug_tslog_user,
225     "", "Dump recorded userland event timestamps");
226