2 # Architectures that offer an FTRACE implementation should select HAVE_FTRACE:
12 config HAVE_DYNAMIC_FTRACE
15 config HAVE_FTRACE_MCOUNT_RECORD
18 config TRACER_MAX_TRACE
32 bool "Kernel Function Tracer"
33 depends on HAVE_FTRACE
34 depends on DEBUG_KERNEL
37 select CONTEXT_SWITCH_TRACER
39 Enable the kernel to trace every kernel function. This is done
40 by using a compiler feature to insert a small, 5-byte No-Operation
41 instruction to the beginning of every kernel function, which NOP
42 sequence is then dynamically patched into a tracer call when
43 tracing is enabled by the administrator. If it's runtime disabled
44 (the bootup default), then the overhead of the instructions is very
45 small and not measurable even in micro-benchmarks.
48 bool "Interrupts-off Latency Tracer"
50 depends on TRACE_IRQFLAGS_SUPPORT
51 depends on GENERIC_TIME
52 depends on DEBUG_KERNEL
55 select TRACER_MAX_TRACE
57 This option measures the time spent in irqs-off critical
58 sections, with microsecond accuracy.
60 The default measurement method is a maximum search, which is
61 disabled by default and can be runtime (re-)started
64 echo 0 > /debugfs/tracing/tracing_max_latency
66 (Note that kernel size and overhead increases with this option
67 enabled. This option and the preempt-off timing option can be
68 used together or separately.)
71 bool "Preemption-off Latency Tracer"
73 depends on GENERIC_TIME
75 depends on DEBUG_KERNEL
77 select TRACER_MAX_TRACE
79 This option measures the time spent in preemption off critical
80 sections, with microsecond accuracy.
82 The default measurement method is a maximum search, which is
83 disabled by default and can be runtime (re-)started
86 echo 0 > /debugfs/tracing/tracing_max_latency
88 (Note that kernel size and overhead increases with this option
89 enabled. This option and the irqs-off timing option can be
90 used together or separately.)
97 This tracer provides the trace needed by the 'Sysprof' userspace
101 bool "Scheduling Latency Tracer"
102 depends on DEBUG_KERNEL
104 select CONTEXT_SWITCH_TRACER
105 select TRACER_MAX_TRACE
107 This tracer tracks the latency of the highest priority task
108 to be scheduled in, starting from the point it has woken up.
110 config CONTEXT_SWITCH_TRACER
111 bool "Trace process context switches"
112 depends on DEBUG_KERNEL
116 This tracer gets called from the context switch and records
117 all switching of tasks.
120 bool "Trace boot initcalls"
121 depends on DEBUG_KERNEL
124 This tracer helps developers to optimize boot times: it records
125 the timings of the initcalls and traces key events and the identity
126 of tasks that can cause boot delays, such as context-switches.
128 Its aim is to be parsed by the /scripts/bootgraph.pl tool to
129 produce pretty graphics about boot inefficiencies, giving a visual
130 representation of the delays during initcalls - but the raw
131 /debug/tracing/trace text output is readable too.
133 ( Note that tracing self tests can't be enabled if this tracer is
134 selected, because the self-tests are an initcall as well and that
135 would invalidate the boot trace. )
138 bool "Trace max stack"
139 depends on HAVE_FTRACE
140 depends on DEBUG_KERNEL
144 This special tracer records the maximum stack footprint of the
145 kernel and displays it in debugfs/tracing/stack_trace.
147 This tracer works by hooking into every function call that the
148 kernel executes, and keeping a maximum stack depth value and
149 stack-trace saved. Because this logic has to execute in every
150 kernel function, all the time, this option can slow down the
151 kernel measurably and is generally intended for kernel
156 config DYNAMIC_FTRACE
157 bool "enable/disable ftrace tracepoints dynamically"
159 depends on HAVE_DYNAMIC_FTRACE
160 depends on DEBUG_KERNEL
163 This option will modify all the calls to ftrace dynamically
164 (will patch them out of the binary image and replaces them
165 with a No-Op instruction) as they are called. A table is
166 created to dynamically enable them again.
168 This way a CONFIG_FTRACE kernel is slightly larger, but otherwise
169 has native performance as long as no tracing is active.
171 The changes to the code are done by a kernel thread that
172 wakes up once a second and checks to see if any ftrace calls
173 were made. If so, it runs stop_machine (stops all CPUS)
174 and modifies the code to jump over the call to ftrace.
176 config FTRACE_MCOUNT_RECORD
178 depends on DYNAMIC_FTRACE
179 depends on HAVE_FTRACE_MCOUNT_RECORD
181 config FTRACE_SELFTEST
184 config FTRACE_STARTUP_TEST
185 bool "Perform a startup test on ftrace"
186 depends on TRACING && DEBUG_KERNEL && !BOOT_TRACER
187 select FTRACE_SELFTEST
189 This option performs a series of startup tests on ftrace. On bootup
190 a series of tests are made to verify that the tracer is
191 functioning properly. It will do tests on all the configured