Commit | Line | Data |
---|---|---|
16444a8a ACM |
1 | # |
2 | # Architectures that offer an FTRACE implementation should select HAVE_FTRACE: | |
3 | # | |
4 | config HAVE_FTRACE | |
5 | bool | |
bc0c38d1 | 6 | |
677aa9f7 SR |
7 | config HAVE_DYNAMIC_FTRACE |
8 | bool | |
9 | ||
352ad25a SR |
10 | config TRACER_MAX_TRACE |
11 | bool | |
12 | ||
bc0c38d1 SR |
13 | config TRACING |
14 | bool | |
15 | select DEBUG_FS | |
86387f7e | 16 | select STACKTRACE |
bc0c38d1 | 17 | |
1b29b018 SR |
18 | config FTRACE |
19 | bool "Kernel Function Tracer" | |
694379e9 | 20 | depends on HAVE_FTRACE |
1b29b018 SR |
21 | select FRAME_POINTER |
22 | select TRACING | |
35e8e302 | 23 | select CONTEXT_SWITCH_TRACER |
1b29b018 SR |
24 | help |
25 | Enable the kernel to trace every kernel function. This is done | |
26 | by using a compiler feature to insert a small, 5-byte No-Operation | |
27 | instruction to the beginning of every kernel function, which NOP | |
28 | sequence is then dynamically patched into a tracer call when | |
29 | tracing is enabled by the administrator. If it's runtime disabled | |
30 | (the bootup default), then the overhead of the instructions is very | |
31 | small and not measurable even in micro-benchmarks. | |
35e8e302 | 32 | |
81d68a96 SR |
33 | config IRQSOFF_TRACER |
34 | bool "Interrupts-off Latency Tracer" | |
35 | default n | |
36 | depends on TRACE_IRQFLAGS_SUPPORT | |
37 | depends on GENERIC_TIME | |
c1d2327b | 38 | depends on HAVE_FTRACE |
81d68a96 SR |
39 | select TRACE_IRQFLAGS |
40 | select TRACING | |
41 | select TRACER_MAX_TRACE | |
42 | help | |
43 | This option measures the time spent in irqs-off critical | |
44 | sections, with microsecond accuracy. | |
45 | ||
46 | The default measurement method is a maximum search, which is | |
47 | disabled by default and can be runtime (re-)started | |
48 | via: | |
49 | ||
50 | echo 0 > /debugfs/tracing/tracing_max_latency | |
51 | ||
6cd8a4bb SR |
52 | (Note that kernel size and overhead increases with this option |
53 | enabled. This option and the preempt-off timing option can be | |
54 | used together or separately.) | |
55 | ||
56 | config PREEMPT_TRACER | |
57 | bool "Preemption-off Latency Tracer" | |
58 | default n | |
59 | depends on GENERIC_TIME | |
60 | depends on PREEMPT | |
c1d2327b | 61 | depends on HAVE_FTRACE |
6cd8a4bb SR |
62 | select TRACING |
63 | select TRACER_MAX_TRACE | |
64 | help | |
65 | This option measures the time spent in preemption off critical | |
66 | sections, with microsecond accuracy. | |
67 | ||
68 | The default measurement method is a maximum search, which is | |
69 | disabled by default and can be runtime (re-)started | |
70 | via: | |
71 | ||
72 | echo 0 > /debugfs/tracing/tracing_max_latency | |
73 | ||
74 | (Note that kernel size and overhead increases with this option | |
75 | enabled. This option and the irqs-off timing option can be | |
76 | used together or separately.) | |
77 | ||
352ad25a SR |
78 | config SCHED_TRACER |
79 | bool "Scheduling Latency Tracer" | |
c1d2327b | 80 | depends on HAVE_FTRACE |
352ad25a SR |
81 | select TRACING |
82 | select CONTEXT_SWITCH_TRACER | |
83 | select TRACER_MAX_TRACE | |
84 | help | |
85 | This tracer tracks the latency of the highest priority task | |
86 | to be scheduled in, starting from the point it has woken up. | |
87 | ||
35e8e302 SR |
88 | config CONTEXT_SWITCH_TRACER |
89 | bool "Trace process context switches" | |
c1d2327b | 90 | depends on HAVE_FTRACE |
35e8e302 SR |
91 | select TRACING |
92 | select MARKERS | |
93 | help | |
94 | This tracer gets called from the context switch and records | |
95 | all switching of tasks. | |
96 | ||
3d083395 SR |
97 | config DYNAMIC_FTRACE |
98 | bool "enable/disable ftrace tracepoints dynamically" | |
99 | depends on FTRACE | |
677aa9f7 | 100 | depends on HAVE_DYNAMIC_FTRACE |
3d083395 SR |
101 | default y |
102 | help | |
103 | This option will modify all the calls to ftrace dynamically | |
104 | (will patch them out of the binary image and replaces them | |
105 | with a No-Op instruction) as they are called. A table is | |
106 | created to dynamically enable them again. | |
107 | ||
108 | This way a CONFIG_FTRACE kernel is slightly larger, but otherwise | |
109 | has native performance as long as no tracing is active. | |
110 | ||
111 | The changes to the code are done by a kernel thread that | |
112 | wakes up once a second and checks to see if any ftrace calls | |
113 | were made. If so, it runs stop_machine (stops all CPUS) | |
114 | and modifies the code to jump over the call to ftrace. | |
60a11774 SR |
115 | |
116 | config FTRACE_SELFTEST | |
117 | bool | |
118 | ||
119 | config FTRACE_STARTUP_TEST | |
120 | bool "Perform a startup test on ftrace" | |
121 | depends on TRACING | |
122 | select FTRACE_SELFTEST | |
123 | help | |
124 | This option performs a series of startup tests on ftrace. On bootup | |
125 | a series of tests are made to verify that the tracer is | |
126 | functioning properly. It will do tests on all the configured | |
127 | tracers of ftrace. |