1 .. SPDX-License-Identifier: GPL-2.0 1 .. SPDX-License-Identifier: GPL-2.0
2 2
3 ================= 3 =================
4 Boot-time tracing 4 Boot-time tracing
5 ================= 5 =================
6 6
7 :Author: Masami Hiramatsu <mhiramat@kernel.org> 7 :Author: Masami Hiramatsu <mhiramat@kernel.org>
8 8
9 Overview 9 Overview
10 ======== 10 ========
11 11
12 Boot-time tracing allows users to trace boot-t 12 Boot-time tracing allows users to trace boot-time process including
13 device initialization with full features of ft 13 device initialization with full features of ftrace including per-event
14 filter and actions, histograms, kprobe-events 14 filter and actions, histograms, kprobe-events and synthetic-events,
15 and trace instances. 15 and trace instances.
16 Since kernel command line is not enough to con 16 Since kernel command line is not enough to control these complex features,
17 this uses bootconfig file to describe tracing 17 this uses bootconfig file to describe tracing feature programming.
18 18
19 Options in the Boot Config 19 Options in the Boot Config
20 ========================== 20 ==========================
21 21
22 Here is the list of available options list for 22 Here is the list of available options list for boot time tracing in
23 boot config file [1]_. All options are under " 23 boot config file [1]_. All options are under "ftrace." or "kernel."
24 prefix. See kernel parameters for the options 24 prefix. See kernel parameters for the options which starts
25 with "kernel." prefix [2]_. 25 with "kernel." prefix [2]_.
26 26
27 .. [1] See :ref:`Documentation/admin-guide/boo 27 .. [1] See :ref:`Documentation/admin-guide/bootconfig.rst <bootconfig>`
28 .. [2] See :ref:`Documentation/admin-guide/ker 28 .. [2] See :ref:`Documentation/admin-guide/kernel-parameters.rst <kernelparameters>`
29 29
30 Ftrace Global Options 30 Ftrace Global Options
31 --------------------- 31 ---------------------
32 32
33 Ftrace global options have "kernel." prefix in 33 Ftrace global options have "kernel." prefix in boot config, which means
34 these options are passed as a part of kernel l 34 these options are passed as a part of kernel legacy command line.
35 35
36 kernel.tp_printk 36 kernel.tp_printk
37 Output trace-event data on printk buffer to 37 Output trace-event data on printk buffer too.
38 38
39 kernel.dump_on_oops [= MODE] 39 kernel.dump_on_oops [= MODE]
40 Dump ftrace on Oops. If MODE = 1 or omitted 40 Dump ftrace on Oops. If MODE = 1 or omitted, dump trace buffer
41 on all CPUs. If MODE = 2, dump a buffer on 41 on all CPUs. If MODE = 2, dump a buffer on a CPU which kicks Oops.
42 42
43 kernel.traceoff_on_warning 43 kernel.traceoff_on_warning
44 Stop tracing if WARN_ON() occurs. 44 Stop tracing if WARN_ON() occurs.
45 45
46 kernel.fgraph_max_depth = MAX_DEPTH 46 kernel.fgraph_max_depth = MAX_DEPTH
47 Set MAX_DEPTH to maximum depth of fgraph tr 47 Set MAX_DEPTH to maximum depth of fgraph tracer.
48 48
49 kernel.fgraph_filters = FILTER[, FILTER2...] 49 kernel.fgraph_filters = FILTER[, FILTER2...]
50 Add fgraph tracing function filters. 50 Add fgraph tracing function filters.
51 51
52 kernel.fgraph_notraces = FILTER[, FILTER2...] 52 kernel.fgraph_notraces = FILTER[, FILTER2...]
53 Add fgraph non-tracing function filters. 53 Add fgraph non-tracing function filters.
54 54
55 55
56 Ftrace Per-instance Options 56 Ftrace Per-instance Options
57 --------------------------- 57 ---------------------------
58 58
59 These options can be used for each instance in 59 These options can be used for each instance including global ftrace node.
60 60
61 ftrace.[instance.INSTANCE.]options = OPT1[, OP 61 ftrace.[instance.INSTANCE.]options = OPT1[, OPT2[...]]
62 Enable given ftrace options. 62 Enable given ftrace options.
63 63
64 ftrace.[instance.INSTANCE.]tracing_on = 0|1 64 ftrace.[instance.INSTANCE.]tracing_on = 0|1
65 Enable/Disable tracing on this instance whe 65 Enable/Disable tracing on this instance when starting boot-time tracing.
66 (you can enable it by the "traceon" event t 66 (you can enable it by the "traceon" event trigger action)
67 67
68 ftrace.[instance.INSTANCE.]trace_clock = CLOCK 68 ftrace.[instance.INSTANCE.]trace_clock = CLOCK
69 Set given CLOCK to ftrace's trace_clock. 69 Set given CLOCK to ftrace's trace_clock.
70 70
71 ftrace.[instance.INSTANCE.]buffer_size = SIZE 71 ftrace.[instance.INSTANCE.]buffer_size = SIZE
72 Configure ftrace buffer size to SIZE. You c 72 Configure ftrace buffer size to SIZE. You can use "KB" or "MB"
73 for that SIZE. 73 for that SIZE.
74 74
75 ftrace.[instance.INSTANCE.]alloc_snapshot 75 ftrace.[instance.INSTANCE.]alloc_snapshot
76 Allocate snapshot buffer. 76 Allocate snapshot buffer.
77 77
78 ftrace.[instance.INSTANCE.]cpumask = CPUMASK 78 ftrace.[instance.INSTANCE.]cpumask = CPUMASK
79 Set CPUMASK as trace cpu-mask. 79 Set CPUMASK as trace cpu-mask.
80 80
81 ftrace.[instance.INSTANCE.]events = EVENT[, EV 81 ftrace.[instance.INSTANCE.]events = EVENT[, EVENT2[...]]
82 Enable given events on boot. You can use a 82 Enable given events on boot. You can use a wild card in EVENT.
83 83
84 ftrace.[instance.INSTANCE.]tracer = TRACER 84 ftrace.[instance.INSTANCE.]tracer = TRACER
85 Set TRACER to current tracer on boot. (e.g. 85 Set TRACER to current tracer on boot. (e.g. function)
86 86
87 ftrace.[instance.INSTANCE.]ftrace.filters 87 ftrace.[instance.INSTANCE.]ftrace.filters
88 This will take an array of tracing function 88 This will take an array of tracing function filter rules.
89 89
90 ftrace.[instance.INSTANCE.]ftrace.notraces 90 ftrace.[instance.INSTANCE.]ftrace.notraces
91 This will take an array of NON-tracing func 91 This will take an array of NON-tracing function filter rules.
92 92
93 93
94 Ftrace Per-Event Options 94 Ftrace Per-Event Options
95 ------------------------ 95 ------------------------
96 96
97 These options are setting per-event options. 97 These options are setting per-event options.
98 98
99 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.e 99 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.enable
100 Enable GROUP:EVENT tracing. 100 Enable GROUP:EVENT tracing.
101 101
102 ftrace.[instance.INSTANCE.]event.GROUP.enable 102 ftrace.[instance.INSTANCE.]event.GROUP.enable
103 Enable all event tracing within GROUP. 103 Enable all event tracing within GROUP.
104 104
105 ftrace.[instance.INSTANCE.]event.enable 105 ftrace.[instance.INSTANCE.]event.enable
106 Enable all event tracing. 106 Enable all event tracing.
107 107
108 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.f 108 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.filter = FILTER
109 Set FILTER rule to the GROUP:EVENT. 109 Set FILTER rule to the GROUP:EVENT.
110 110
111 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.a 111 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.actions = ACTION[, ACTION2[...]]
112 Set ACTIONs to the GROUP:EVENT. 112 Set ACTIONs to the GROUP:EVENT.
113 113
114 ftrace.[instance.INSTANCE.]event.kprobes.EVENT 114 ftrace.[instance.INSTANCE.]event.kprobes.EVENT.probes = PROBE[, PROBE2[...]]
115 Defines new kprobe event based on PROBEs. I 115 Defines new kprobe event based on PROBEs. It is able to define
116 multiple probes on one event, but those mus 116 multiple probes on one event, but those must have same type of
117 arguments. This option is available only fo 117 arguments. This option is available only for the event which
118 group name is "kprobes". 118 group name is "kprobes".
119 119
120 ftrace.[instance.INSTANCE.]event.synthetic.EVE 120 ftrace.[instance.INSTANCE.]event.synthetic.EVENT.fields = FIELD[, FIELD2[...]]
121 Defines new synthetic event with FIELDs. Ea 121 Defines new synthetic event with FIELDs. Each field should be
122 "type varname". 122 "type varname".
123 123
124 Note that kprobe and synthetic event definitio 124 Note that kprobe and synthetic event definitions can be written under
125 instance node, but those are also visible from 125 instance node, but those are also visible from other instances. So please
126 take care for event name conflict. 126 take care for event name conflict.
127 127
128 Ftrace Histogram Options <<
129 ------------------------ <<
130 <<
131 Since it is too long to write a histogram acti <<
132 action option, there are tree-style options un <<
133 for the histogram actions. For the detail of t <<
134 please read the event histogram document (Docu <<
135 <<
136 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
137 Set histogram key parameters. (Mandatory) <<
138 The 'N' is a digit string for the multiple h <<
139 if there is one histogram on the event. <<
140 <<
141 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
142 Set histogram value parameters. <<
143 <<
144 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
145 Set histogram sort parameter options. <<
146 <<
147 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
148 Set histogram size (number of entries). <<
149 <<
150 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
151 Set histogram name. <<
152 <<
153 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
154 Define a new VARIABLE by EXPR expression. <<
155 <<
156 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
157 Set histogram control parameter. You can set <<
158 <<
159 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
160 Set histogram 'onmatch' handler matching eve <<
161 The 'M' is a digit string for the multiple ' <<
162 if there is one 'onmatch' handler on this hi <<
163 <<
164 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
165 Set histogram 'trace' action for 'onmatch'. <<
166 EVENT must be a synthetic event name, and AR <<
167 for that event. Mandatory if 'onmatch.event' <<
168 <<
169 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
170 Set histogram 'onmax' handler variable param <<
171 <<
172 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
173 Set histogram 'onchange' handler variable pa <<
174 <<
175 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
176 Set histogram 'save' action parameters for ' <<
177 This option or below 'snapshot' option is ma <<
178 'onchange.var' option is set. <<
179 <<
180 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
181 Set histogram 'snapshot' action for 'onmax' <<
182 This option or above 'save' option is mandat <<
183 'onchange.var' option is set. <<
184 <<
185 ftrace.[instance.INSTANCE.]event.GROUP.EVENT.h <<
186 Set histogram filter expression. You don't n <<
187 <<
188 Note that this 'hist' option can conflict with <<
189 option if the 'actions' option has a histogram <<
190 <<
191 128
192 When to Start 129 When to Start
193 ============= 130 =============
194 131
195 All boot-time tracing options starting with `` 132 All boot-time tracing options starting with ``ftrace`` will be enabled at the
196 end of core_initcall. This means you can trace 133 end of core_initcall. This means you can trace the events from postcore_initcall.
197 Most of the subsystems and architecture depend 134 Most of the subsystems and architecture dependent drivers will be initialized
198 after that (arch_initcall or subsys_initcall). 135 after that (arch_initcall or subsys_initcall). Thus, you can trace those with
199 boot-time tracing. 136 boot-time tracing.
200 If you want to trace events before core_initca 137 If you want to trace events before core_initcall, you can use the options
201 starting with ``kernel``. Some of them will be 138 starting with ``kernel``. Some of them will be enabled eariler than the initcall
202 processing (for example,. ``kernel.ftrace=func 139 processing (for example,. ``kernel.ftrace=function`` and ``kernel.trace_event``
203 will start before the initcall.) 140 will start before the initcall.)
204 141
205 142
206 Examples 143 Examples
207 ======== 144 ========
208 145
209 For example, to add filter and actions for eac 146 For example, to add filter and actions for each event, define kprobe
210 events, and synthetic events with histogram, w 147 events, and synthetic events with histogram, write a boot config like
211 below:: 148 below::
212 149
213 ftrace.event { 150 ftrace.event {
214 task.task_newtask { 151 task.task_newtask {
215 filter = "pid < 128" 152 filter = "pid < 128"
216 enable 153 enable
217 } 154 }
218 kprobes.vfs_read { 155 kprobes.vfs_read {
219 probes = "vfs_read $arg1 $arg2 156 probes = "vfs_read $arg1 $arg2"
220 filter = "common_pid < 200" 157 filter = "common_pid < 200"
221 enable 158 enable
222 } 159 }
223 synthetic.initcall_latency { 160 synthetic.initcall_latency {
224 fields = "unsigned long func", 161 fields = "unsigned long func", "u64 lat"
225 hist { !! 162 actions = "hist:keys=func.sym,lat:vals=lat:sort=lat"
226 keys = func.sym, lat <<
227 values = lat <<
228 sort = lat <<
229 } <<
230 } 163 }
231 initcall.initcall_start.hist { !! 164 initcall.initcall_start {
232 keys = func !! 165 actions = "hist:keys=func:ts0=common_timestamp.usecs"
233 var.ts0 = common_timestamp.use <<
234 } 166 }
235 initcall.initcall_finish.hist { !! 167 initcall.initcall_finish {
236 keys = func !! 168 actions = "hist:keys=func:lat=common_timestamp.usecs-$ts0:onmatch(initcall.initcall_start).initcall_latency(func,$lat)"
237 var.lat = common_timestamp.use <<
238 onmatch { <<
239 event = initcall.initc <<
240 trace = initcall_laten <<
241 } <<
242 } 169 }
243 } 170 }
244 171
245 Also, boot-time tracing supports "instance" no 172 Also, boot-time tracing supports "instance" node, which allows us to run
246 several tracers for different purpose at once. 173 several tracers for different purpose at once. For example, one tracer
247 is for tracing functions starting with "user\_ 174 is for tracing functions starting with "user\_", and others tracing
248 "kernel\_" functions, you can write boot confi 175 "kernel\_" functions, you can write boot config as below::
249 176
250 ftrace.instance { 177 ftrace.instance {
251 foo { 178 foo {
252 tracer = "function" 179 tracer = "function"
253 ftrace.filters = "user_*" 180 ftrace.filters = "user_*"
254 } 181 }
255 bar { 182 bar {
256 tracer = "function" 183 tracer = "function"
257 ftrace.filters = "kernel_*" 184 ftrace.filters = "kernel_*"
258 } 185 }
259 } 186 }
260 187
261 The instance node also accepts event nodes so 188 The instance node also accepts event nodes so that each instance
262 can customize its event tracing. 189 can customize its event tracing.
263 190
264 With the trigger action and kprobes, you can t 191 With the trigger action and kprobes, you can trace function-graph while
265 a function is called. For example, this will t 192 a function is called. For example, this will trace all function calls in
266 the pci_proc_init():: 193 the pci_proc_init()::
267 194
268 ftrace { 195 ftrace {
269 tracing_on = 0 196 tracing_on = 0
270 tracer = function_graph 197 tracer = function_graph
271 event.kprobes { 198 event.kprobes {
272 start_event { 199 start_event {
273 probes = "pci_proc_ini 200 probes = "pci_proc_init"
274 actions = "traceon" 201 actions = "traceon"
275 } 202 }
276 end_event { 203 end_event {
277 probes = "pci_proc_ini 204 probes = "pci_proc_init%return"
278 actions = "traceoff" 205 actions = "traceoff"
279 } 206 }
280 } 207 }
281 } 208 }
282 209
283 210
284 This boot-time tracing also supports ftrace ke 211 This boot-time tracing also supports ftrace kernel parameters via boot
285 config. 212 config.
286 For example, following kernel parameters:: 213 For example, following kernel parameters::
287 214
288 trace_options=sym-addr trace_event=initcall:* 215 trace_options=sym-addr trace_event=initcall:* tp_printk trace_buf_size=1M ftrace=function ftrace_filter="vfs*"
289 216
290 This can be written in boot config like below: 217 This can be written in boot config like below::
291 218
292 kernel { 219 kernel {
293 trace_options = sym-addr 220 trace_options = sym-addr
294 trace_event = "initcall:*" 221 trace_event = "initcall:*"
295 tp_printk 222 tp_printk
296 trace_buf_size = 1M 223 trace_buf_size = 1M
297 ftrace = function 224 ftrace = function
298 ftrace_filter = "vfs*" 225 ftrace_filter = "vfs*"
299 } 226 }
300 227
301 Note that parameters start with "kernel" prefi 228 Note that parameters start with "kernel" prefix instead of "ftrace".
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.