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/param.h>
28 #include <sys/kernel.h>
29 #include <sys/linker.h>
30 #include <sys/malloc.h>
31 #include <sys/proc.h>
32 #include <sys/sbuf.h>
33 #include <sys/sysctl.h>
34 #include <sys/systm.h>
35 #include <sys/tslog.h>
36
37 #include <machine/atomic.h>
38 #include <machine/cpu.h>
39
40 #ifndef TSLOGSIZE
41 #define TSLOGSIZE 262144
42 #endif
43
44 static volatile long nrecs = 0;
45 static struct timestamp {
46 void * td;
47 int type;
48 const char * f;
49 const char * s;
50 uint64_t tsc;
51 } timestamps[TSLOGSIZE];
52
53 void
tslog(void * td,int type,const char * f,const char * s)54 tslog(void * td, int type, const char * f, const char * s)
55 {
56 uint64_t tsc = get_cyclecount();
57 long pos;
58
59 /* A NULL thread is thread0 before curthread is set. */
60 if (td == NULL)
61 td = &thread0;
62
63 /* Grab a slot. */
64 pos = atomic_fetchadd_long(&nrecs, 1);
65
66 /* Store record. */
67 if (pos < nitems(timestamps)) {
68 timestamps[pos].td = td;
69 timestamps[pos].type = type;
70 timestamps[pos].f = f;
71 timestamps[pos].s = s;
72 timestamps[pos].tsc = tsc;
73 }
74 }
75
76 static int
sysctl_debug_tslog(SYSCTL_HANDLER_ARGS)77 sysctl_debug_tslog(SYSCTL_HANDLER_ARGS)
78 {
79 int error;
80 struct sbuf *sb;
81 size_t i, limit;
82 caddr_t loader_tslog;
83 void * loader_tslog_buf;
84 size_t loader_tslog_len;
85
86 /*
87 * This code can race against the code in tslog() which stores
88 * records: Theoretically we could end up reading a record after
89 * its slots have been reserved but before it has been written.
90 * Since this code takes orders of magnitude longer to run than
91 * tslog() takes to write a record, it is highly unlikely that
92 * anyone will ever experience this race.
93 */
94 sb = sbuf_new_for_sysctl(NULL, NULL, 1024, req);
95
96 /* Get data from the boot loader, if it provided any. */
97 loader_tslog = preload_search_by_type("TSLOG data");
98 if (loader_tslog != NULL) {
99 loader_tslog_buf = preload_fetch_addr(loader_tslog);
100 loader_tslog_len = preload_fetch_size(loader_tslog);
101 sbuf_bcat(sb, loader_tslog_buf, loader_tslog_len);
102 }
103
104 /* Add data logged within the kernel. */
105 limit = MIN(nrecs, nitems(timestamps));
106 for (i = 0; i < limit; i++) {
107 sbuf_printf(sb, "%p", timestamps[i].td);
108 sbuf_printf(sb, " %llu",
109 (unsigned long long)timestamps[i].tsc);
110 switch (timestamps[i].type) {
111 case TS_ENTER:
112 sbuf_cat(sb, " ENTER");
113 break;
114 case TS_EXIT:
115 sbuf_cat(sb, " EXIT");
116 break;
117 case TS_THREAD:
118 sbuf_cat(sb, " THREAD");
119 break;
120 case TS_EVENT:
121 sbuf_cat(sb, " EVENT");
122 break;
123 }
124 sbuf_printf(sb, " %s", timestamps[i].f ? timestamps[i].f : "(null)");
125 if (timestamps[i].s)
126 sbuf_printf(sb, " %s\n", timestamps[i].s);
127 else
128 sbuf_putc(sb, '\n');
129 }
130 error = sbuf_finish(sb);
131 sbuf_delete(sb);
132 return (error);
133 }
134
135 SYSCTL_PROC(_debug, OID_AUTO, tslog,
136 CTLTYPE_STRING|CTLFLAG_RD|CTLFLAG_MPSAFE|CTLFLAG_SKIP,
137 0, 0, sysctl_debug_tslog, "", "Dump recorded event timestamps");
138
139 MALLOC_DEFINE(M_TSLOGUSER, "tsloguser", "Strings used by userland tslog");
140 static struct procdata {
141 pid_t ppid;
142 uint64_t tsc_forked;
143 uint64_t tsc_exited;
144 char * execname;
145 char * namei;
146 int reused;
147 } procs[PID_MAX + 1];
148
149 void
tslog_user(pid_t pid,pid_t ppid,const char * execname,const char * namei)150 tslog_user(pid_t pid, pid_t ppid, const char * execname, const char * namei)
151 {
152 uint64_t tsc = get_cyclecount();
153
154 /* If we wrapped, do nothing. */
155 if (procs[pid].reused)
156 return;
157
158 /* If we have a ppid, we're recording a fork. */
159 if (ppid != (pid_t)(-1)) {
160 /* If we have a ppid already, we wrapped. */
161 if (procs[pid].ppid) {
162 procs[pid].reused = 1;
163 return;
164 }
165
166 /* Fill in some fields. */
167 procs[pid].ppid = ppid;
168 procs[pid].tsc_forked = tsc;
169 return;
170 }
171
172 /* If we have an execname, record it. */
173 if (execname != NULL) {
174 if (procs[pid].execname != NULL)
175 free(procs[pid].execname, M_TSLOGUSER);
176 procs[pid].execname = strdup(execname, M_TSLOGUSER);
177 return;
178 }
179
180 /* Record the first namei for the process. */
181 if (namei != NULL) {
182 if (procs[pid].namei == NULL)
183 procs[pid].namei = strdup(namei, M_TSLOGUSER);
184 return;
185 }
186
187 /* Otherwise we're recording an exit. */
188 procs[pid].tsc_exited = tsc;
189 }
190
191 static int
sysctl_debug_tslog_user(SYSCTL_HANDLER_ARGS)192 sysctl_debug_tslog_user(SYSCTL_HANDLER_ARGS)
193 {
194 int error;
195 struct sbuf *sb;
196 pid_t pid;
197
198 sb = sbuf_new_for_sysctl(NULL, NULL, 1024, req);
199
200 /* Export the data we logged. */
201 for (pid = 0; pid <= PID_MAX; pid++) {
202 sbuf_printf(sb, "%zu", (size_t)pid);
203 sbuf_printf(sb, " %zu", (size_t)procs[pid].ppid);
204 sbuf_printf(sb, " %llu",
205 (unsigned long long)procs[pid].tsc_forked);
206 sbuf_printf(sb, " %llu",
207 (unsigned long long)procs[pid].tsc_exited);
208 sbuf_printf(sb, " \"%s\"", procs[pid].execname ?
209 procs[pid].execname : "");
210 sbuf_printf(sb, " \"%s\"", procs[pid].namei ?
211 procs[pid].namei : "");
212 sbuf_putc(sb, '\n');
213 }
214 error = sbuf_finish(sb);
215 sbuf_delete(sb);
216 return (error);
217 }
218
219 SYSCTL_PROC(_debug, OID_AUTO, tslog_user,
220 CTLTYPE_STRING|CTLFLAG_RD|CTLFLAG_MPSAFE|CTLFLAG_SKIP,
221 0, 0, sysctl_debug_tslog_user,
222 "", "Dump recorded userland event timestamps");
223