1 ======================= 1 =======================
2 Kernel Probes (Kprobes) 2 Kernel Probes (Kprobes)
3 ======================= 3 =======================
4 4
5 :Author: Jim Keniston <jkenisto@us.ibm.com> 5 :Author: Jim Keniston <jkenisto@us.ibm.com>
6 :Author: Prasanna S Panchamukhi <prasanna.panch 6 :Author: Prasanna S Panchamukhi <prasanna.panchamukhi@gmail.com>
7 :Author: Masami Hiramatsu <mhiramat@kernel.org> !! 7 :Author: Masami Hiramatsu <mhiramat@redhat.com>
8 8
9 .. CONTENTS 9 .. CONTENTS
10 10
11 1. Concepts: Kprobes, and Return Probes 11 1. Concepts: Kprobes, and Return Probes
12 2. Architectures Supported 12 2. Architectures Supported
13 3. Configuring Kprobes 13 3. Configuring Kprobes
14 4. API Reference 14 4. API Reference
15 5. Kprobes Features and Limitations 15 5. Kprobes Features and Limitations
16 6. Probe Overhead 16 6. Probe Overhead
17 7. TODO 17 7. TODO
18 8. Kprobes Example 18 8. Kprobes Example
19 9. Kretprobes Example 19 9. Kretprobes Example
20 10. Deprecated Features 20 10. Deprecated Features
21 Appendix A: The kprobes debugfs interface 21 Appendix A: The kprobes debugfs interface
22 Appendix B: The kprobes sysctl interface 22 Appendix B: The kprobes sysctl interface
23 Appendix C: References 23 Appendix C: References
24 24
25 Concepts: Kprobes and Return Probes 25 Concepts: Kprobes and Return Probes
26 ========================================= 26 =========================================
27 27
28 Kprobes enables you to dynamically break into 28 Kprobes enables you to dynamically break into any kernel routine and
29 collect debugging and performance information 29 collect debugging and performance information non-disruptively. You
30 can trap at almost any kernel code address [1] 30 can trap at almost any kernel code address [1]_, specifying a handler
31 routine to be invoked when the breakpoint is h 31 routine to be invoked when the breakpoint is hit.
32 32
33 .. [1] some parts of the kernel code can not b 33 .. [1] some parts of the kernel code can not be trapped, see
34 :ref:`kprobes_blacklist`) 34 :ref:`kprobes_blacklist`)
35 35
36 There are currently two types of probes: kprob 36 There are currently two types of probes: kprobes, and kretprobes
37 (also called return probes). A kprobe can be 37 (also called return probes). A kprobe can be inserted on virtually
38 any instruction in the kernel. A return probe 38 any instruction in the kernel. A return probe fires when a specified
39 function returns. 39 function returns.
40 40
41 In the typical case, Kprobes-based instrumenta 41 In the typical case, Kprobes-based instrumentation is packaged as
42 a kernel module. The module's init function i 42 a kernel module. The module's init function installs ("registers")
43 one or more probes, and the exit function unre 43 one or more probes, and the exit function unregisters them. A
44 registration function such as register_kprobe( 44 registration function such as register_kprobe() specifies where
45 the probe is to be inserted and what handler i 45 the probe is to be inserted and what handler is to be called when
46 the probe is hit. 46 the probe is hit.
47 47
48 There are also ``register_/unregister_*probes( 48 There are also ``register_/unregister_*probes()`` functions for batch
49 registration/unregistration of a group of ``*p 49 registration/unregistration of a group of ``*probes``. These functions
50 can speed up unregistration process when you h 50 can speed up unregistration process when you have to unregister
51 a lot of probes at once. 51 a lot of probes at once.
52 52
53 The next four subsections explain how the diff 53 The next four subsections explain how the different types of
54 probes work and how jump optimization works. 54 probes work and how jump optimization works. They explain certain
55 things that you'll need to know in order to ma 55 things that you'll need to know in order to make the best use of
56 Kprobes -- e.g., the difference between a pre_ 56 Kprobes -- e.g., the difference between a pre_handler and
57 a post_handler, and how to use the maxactive a 57 a post_handler, and how to use the maxactive and nmissed fields of
58 a kretprobe. But if you're in a hurry to star 58 a kretprobe. But if you're in a hurry to start using Kprobes, you
59 can skip ahead to :ref:`kprobes_archs_supporte 59 can skip ahead to :ref:`kprobes_archs_supported`.
60 60
61 How Does a Kprobe Work? 61 How Does a Kprobe Work?
62 ----------------------- 62 -----------------------
63 63
64 When a kprobe is registered, Kprobes makes a c 64 When a kprobe is registered, Kprobes makes a copy of the probed
65 instruction and replaces the first byte(s) of 65 instruction and replaces the first byte(s) of the probed instruction
66 with a breakpoint instruction (e.g., int3 on i 66 with a breakpoint instruction (e.g., int3 on i386 and x86_64).
67 67
68 When a CPU hits the breakpoint instruction, a 68 When a CPU hits the breakpoint instruction, a trap occurs, the CPU's
69 registers are saved, and control passes to Kpr 69 registers are saved, and control passes to Kprobes via the
70 notifier_call_chain mechanism. Kprobes execut 70 notifier_call_chain mechanism. Kprobes executes the "pre_handler"
71 associated with the kprobe, passing the handle 71 associated with the kprobe, passing the handler the addresses of the
72 kprobe struct and the saved registers. 72 kprobe struct and the saved registers.
73 73
74 Next, Kprobes single-steps its copy of the pro 74 Next, Kprobes single-steps its copy of the probed instruction.
75 (It would be simpler to single-step the actual 75 (It would be simpler to single-step the actual instruction in place,
76 but then Kprobes would have to temporarily rem 76 but then Kprobes would have to temporarily remove the breakpoint
77 instruction. This would open a small time win 77 instruction. This would open a small time window when another CPU
78 could sail right past the probepoint.) 78 could sail right past the probepoint.)
79 79
80 After the instruction is single-stepped, Kprob 80 After the instruction is single-stepped, Kprobes executes the
81 "post_handler," if any, that is associated wit 81 "post_handler," if any, that is associated with the kprobe.
82 Execution then continues with the instruction 82 Execution then continues with the instruction following the probepoint.
83 83
84 Changing Execution Path 84 Changing Execution Path
85 ----------------------- 85 -----------------------
86 86
87 Since kprobes can probe into a running kernel 87 Since kprobes can probe into a running kernel code, it can change the
88 register set, including instruction pointer. T 88 register set, including instruction pointer. This operation requires
89 maximum care, such as keeping the stack frame, 89 maximum care, such as keeping the stack frame, recovering the execution
90 path etc. Since it operates on a running kerne 90 path etc. Since it operates on a running kernel and needs deep knowledge
91 of computer architecture and concurrent comput 91 of computer architecture and concurrent computing, you can easily shoot
92 your foot. 92 your foot.
93 93
94 If you change the instruction pointer (and set 94 If you change the instruction pointer (and set up other related
95 registers) in pre_handler, you must return !0 95 registers) in pre_handler, you must return !0 so that kprobes stops
96 single stepping and just returns to the given 96 single stepping and just returns to the given address.
97 This also means post_handler should not be cal 97 This also means post_handler should not be called anymore.
98 98
99 Note that this operation may be harder on some 99 Note that this operation may be harder on some architectures which use
100 TOC (Table of Contents) for function call, sin 100 TOC (Table of Contents) for function call, since you have to setup a new
101 TOC for your function in your module, and reco 101 TOC for your function in your module, and recover the old one after
102 returning from it. 102 returning from it.
103 103
104 Return Probes 104 Return Probes
105 ------------- 105 -------------
106 106
107 How Does a Return Probe Work? 107 How Does a Return Probe Work?
108 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 108 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
109 109
110 When you call register_kretprobe(), Kprobes es 110 When you call register_kretprobe(), Kprobes establishes a kprobe at
111 the entry to the function. When the probed fu 111 the entry to the function. When the probed function is called and this
112 probe is hit, Kprobes saves a copy of the retu 112 probe is hit, Kprobes saves a copy of the return address, and replaces
113 the return address with the address of a "tram 113 the return address with the address of a "trampoline." The trampoline
114 is an arbitrary piece of code -- typically jus 114 is an arbitrary piece of code -- typically just a nop instruction.
115 At boot time, Kprobes registers a kprobe at th 115 At boot time, Kprobes registers a kprobe at the trampoline.
116 116
117 When the probed function executes its return i 117 When the probed function executes its return instruction, control
118 passes to the trampoline and that probe is hit 118 passes to the trampoline and that probe is hit. Kprobes' trampoline
119 handler calls the user-specified return handle 119 handler calls the user-specified return handler associated with the
120 kretprobe, then sets the saved instruction poi 120 kretprobe, then sets the saved instruction pointer to the saved return
121 address, and that's where execution resumes up 121 address, and that's where execution resumes upon return from the trap.
122 122
123 While the probed function is executing, its re 123 While the probed function is executing, its return address is
124 stored in an object of type kretprobe_instance 124 stored in an object of type kretprobe_instance. Before calling
125 register_kretprobe(), the user sets the maxact 125 register_kretprobe(), the user sets the maxactive field of the
126 kretprobe struct to specify how many instances 126 kretprobe struct to specify how many instances of the specified
127 function can be probed simultaneously. regist 127 function can be probed simultaneously. register_kretprobe()
128 pre-allocates the indicated number of kretprob 128 pre-allocates the indicated number of kretprobe_instance objects.
129 129
130 For example, if the function is non-recursive 130 For example, if the function is non-recursive and is called with a
131 spinlock held, maxactive = 1 should be enough. 131 spinlock held, maxactive = 1 should be enough. If the function is
132 non-recursive and can never relinquish the CPU 132 non-recursive and can never relinquish the CPU (e.g., via a semaphore
133 or preemption), NR_CPUS should be enough. If 133 or preemption), NR_CPUS should be enough. If maxactive <= 0, it is
134 set to a default value: max(10, 2*NR_CPUS). !! 134 set to a default value. If CONFIG_PREEMPT is enabled, the default
>> 135 is max(10, 2*NR_CPUS). Otherwise, the default is NR_CPUS.
135 136
136 It's not a disaster if you set maxactive too l 137 It's not a disaster if you set maxactive too low; you'll just miss
137 some probes. In the kretprobe struct, the nmi 138 some probes. In the kretprobe struct, the nmissed field is set to
138 zero when the return probe is registered, and 139 zero when the return probe is registered, and is incremented every
139 time the probed function is entered but there 140 time the probed function is entered but there is no kretprobe_instance
140 object available for establishing the return p 141 object available for establishing the return probe.
141 142
142 Kretprobe entry-handler 143 Kretprobe entry-handler
143 ^^^^^^^^^^^^^^^^^^^^^^^ 144 ^^^^^^^^^^^^^^^^^^^^^^^
144 145
145 Kretprobes also provides an optional user-spec 146 Kretprobes also provides an optional user-specified handler which runs
146 on function entry. This handler is specified b 147 on function entry. This handler is specified by setting the entry_handler
147 field of the kretprobe struct. Whenever the kp 148 field of the kretprobe struct. Whenever the kprobe placed by kretprobe at the
148 function entry is hit, the user-defined entry_ 149 function entry is hit, the user-defined entry_handler, if any, is invoked.
149 If the entry_handler returns 0 (success) then 150 If the entry_handler returns 0 (success) then a corresponding return handler
150 is guaranteed to be called upon function retur 151 is guaranteed to be called upon function return. If the entry_handler
151 returns a non-zero error then Kprobes leaves t 152 returns a non-zero error then Kprobes leaves the return address as is, and
152 the kretprobe has no further effect for that p 153 the kretprobe has no further effect for that particular function instance.
153 154
154 Multiple entry and return handler invocations 155 Multiple entry and return handler invocations are matched using the unique
155 kretprobe_instance object associated with them 156 kretprobe_instance object associated with them. Additionally, a user
156 may also specify per return-instance private d 157 may also specify per return-instance private data to be part of each
157 kretprobe_instance object. This is especially 158 kretprobe_instance object. This is especially useful when sharing private
158 data between corresponding user entry and retu 159 data between corresponding user entry and return handlers. The size of each
159 private data object can be specified at kretpr 160 private data object can be specified at kretprobe registration time by
160 setting the data_size field of the kretprobe s 161 setting the data_size field of the kretprobe struct. This data can be
161 accessed through the data field of each kretpr 162 accessed through the data field of each kretprobe_instance object.
162 163
163 In case probed function is entered but there i 164 In case probed function is entered but there is no kretprobe_instance
164 object available, then in addition to incremen 165 object available, then in addition to incrementing the nmissed count,
165 the user entry_handler invocation is also skip 166 the user entry_handler invocation is also skipped.
166 167
167 .. _kprobes_jump_optimization: 168 .. _kprobes_jump_optimization:
168 169
169 How Does Jump Optimization Work? 170 How Does Jump Optimization Work?
170 -------------------------------- 171 --------------------------------
171 172
172 If your kernel is built with CONFIG_OPTPROBES= 173 If your kernel is built with CONFIG_OPTPROBES=y (currently this flag
173 is automatically set 'y' on x86/x86-64, non-pr 174 is automatically set 'y' on x86/x86-64, non-preemptive kernel) and
174 the "debug.kprobes_optimization" kernel parame 175 the "debug.kprobes_optimization" kernel parameter is set to 1 (see
175 sysctl(8)), Kprobes tries to reduce probe-hit 176 sysctl(8)), Kprobes tries to reduce probe-hit overhead by using a jump
176 instruction instead of a breakpoint instructio 177 instruction instead of a breakpoint instruction at each probepoint.
177 178
178 Init a Kprobe 179 Init a Kprobe
179 ^^^^^^^^^^^^^ 180 ^^^^^^^^^^^^^
180 181
181 When a probe is registered, before attempting 182 When a probe is registered, before attempting this optimization,
182 Kprobes inserts an ordinary, breakpoint-based 183 Kprobes inserts an ordinary, breakpoint-based kprobe at the specified
183 address. So, even if it's not possible to opti 184 address. So, even if it's not possible to optimize this particular
184 probepoint, there'll be a probe there. 185 probepoint, there'll be a probe there.
185 186
186 Safety Check 187 Safety Check
187 ^^^^^^^^^^^^ 188 ^^^^^^^^^^^^
188 189
189 Before optimizing a probe, Kprobes performs th 190 Before optimizing a probe, Kprobes performs the following safety checks:
190 191
191 - Kprobes verifies that the region that will b 192 - Kprobes verifies that the region that will be replaced by the jump
192 instruction (the "optimized region") lies en 193 instruction (the "optimized region") lies entirely within one function.
193 (A jump instruction is multiple bytes, and s 194 (A jump instruction is multiple bytes, and so may overlay multiple
194 instructions.) 195 instructions.)
195 196
196 - Kprobes analyzes the entire function and ver 197 - Kprobes analyzes the entire function and verifies that there is no
197 jump into the optimized region. Specificall 198 jump into the optimized region. Specifically:
198 199
199 - the function contains no indirect jump; 200 - the function contains no indirect jump;
200 - the function contains no instruction that 201 - the function contains no instruction that causes an exception (since
201 the fixup code triggered by the exception 202 the fixup code triggered by the exception could jump back into the
202 optimized region -- Kprobes checks the exc 203 optimized region -- Kprobes checks the exception tables to verify this);
203 - there is no near jump to the optimized reg 204 - there is no near jump to the optimized region (other than to the first
204 byte). 205 byte).
205 206
206 - For each instruction in the optimized region 207 - For each instruction in the optimized region, Kprobes verifies that
207 the instruction can be executed out of line. 208 the instruction can be executed out of line.
208 209
209 Preparing Detour Buffer 210 Preparing Detour Buffer
210 ^^^^^^^^^^^^^^^^^^^^^^^ 211 ^^^^^^^^^^^^^^^^^^^^^^^
211 212
212 Next, Kprobes prepares a "detour" buffer, whic 213 Next, Kprobes prepares a "detour" buffer, which contains the following
213 instruction sequence: 214 instruction sequence:
214 215
215 - code to push the CPU's registers (emulating 216 - code to push the CPU's registers (emulating a breakpoint trap)
216 - a call to the trampoline code which calls us 217 - a call to the trampoline code which calls user's probe handlers.
217 - code to restore registers 218 - code to restore registers
218 - the instructions from the optimized region 219 - the instructions from the optimized region
219 - a jump back to the original execution path. 220 - a jump back to the original execution path.
220 221
221 Pre-optimization 222 Pre-optimization
222 ^^^^^^^^^^^^^^^^ 223 ^^^^^^^^^^^^^^^^
223 224
224 After preparing the detour buffer, Kprobes ver 225 After preparing the detour buffer, Kprobes verifies that none of the
225 following situations exist: 226 following situations exist:
226 227
227 - The probe has a post_handler. 228 - The probe has a post_handler.
228 - Other instructions in the optimized region a 229 - Other instructions in the optimized region are probed.
229 - The probe is disabled. 230 - The probe is disabled.
230 231
231 In any of the above cases, Kprobes won't start 232 In any of the above cases, Kprobes won't start optimizing the probe.
232 Since these are temporary situations, Kprobes 233 Since these are temporary situations, Kprobes tries to start
233 optimizing it again if the situation is change 234 optimizing it again if the situation is changed.
234 235
235 If the kprobe can be optimized, Kprobes enqueu 236 If the kprobe can be optimized, Kprobes enqueues the kprobe to an
236 optimizing list, and kicks the kprobe-optimize 237 optimizing list, and kicks the kprobe-optimizer workqueue to optimize
237 it. If the to-be-optimized probepoint is hit 238 it. If the to-be-optimized probepoint is hit before being optimized,
238 Kprobes returns control to the original instru 239 Kprobes returns control to the original instruction path by setting
239 the CPU's instruction pointer to the copied co 240 the CPU's instruction pointer to the copied code in the detour buffer
240 -- thus at least avoiding the single-step. 241 -- thus at least avoiding the single-step.
241 242
242 Optimization 243 Optimization
243 ^^^^^^^^^^^^ 244 ^^^^^^^^^^^^
244 245
245 The Kprobe-optimizer doesn't insert the jump i 246 The Kprobe-optimizer doesn't insert the jump instruction immediately;
246 rather, it calls synchronize_rcu() for safety 247 rather, it calls synchronize_rcu() for safety first, because it's
247 possible for a CPU to be interrupted in the mi 248 possible for a CPU to be interrupted in the middle of executing the
248 optimized region [3]_. As you know, synchroni 249 optimized region [3]_. As you know, synchronize_rcu() can ensure
249 that all interruptions that were active when s 250 that all interruptions that were active when synchronize_rcu()
250 was called are done, but only if CONFIG_PREEMP 251 was called are done, but only if CONFIG_PREEMPT=n. So, this version
251 of kprobe optimization supports only kernels w 252 of kprobe optimization supports only kernels with CONFIG_PREEMPT=n [4]_.
252 253
253 After that, the Kprobe-optimizer calls stop_ma 254 After that, the Kprobe-optimizer calls stop_machine() to replace
254 the optimized region with a jump instruction t 255 the optimized region with a jump instruction to the detour buffer,
255 using text_poke_smp(). 256 using text_poke_smp().
256 257
257 Unoptimization 258 Unoptimization
258 ^^^^^^^^^^^^^^ 259 ^^^^^^^^^^^^^^
259 260
260 When an optimized kprobe is unregistered, disa 261 When an optimized kprobe is unregistered, disabled, or blocked by
261 another kprobe, it will be unoptimized. If th 262 another kprobe, it will be unoptimized. If this happens before
262 the optimization is complete, the kprobe is ju 263 the optimization is complete, the kprobe is just dequeued from the
263 optimized list. If the optimization has been 264 optimized list. If the optimization has been done, the jump is
264 replaced with the original code (except for an 265 replaced with the original code (except for an int3 breakpoint in
265 the first byte) by using text_poke_smp(). 266 the first byte) by using text_poke_smp().
266 267
267 .. [3] Please imagine that the 2nd instruction 268 .. [3] Please imagine that the 2nd instruction is interrupted and then
268 the optimizer replaces the 2nd instruction 269 the optimizer replaces the 2nd instruction with the jump *address*
269 while the interrupt handler is running. Whe 270 while the interrupt handler is running. When the interrupt
270 returns to original address, there is no va 271 returns to original address, there is no valid instruction,
271 and it causes an unexpected result. 272 and it causes an unexpected result.
272 273
273 .. [4] This optimization-safety checking may b 274 .. [4] This optimization-safety checking may be replaced with the
274 stop-machine method that ksplice uses for s 275 stop-machine method that ksplice uses for supporting a CONFIG_PREEMPT=y
275 kernel. 276 kernel.
276 277
277 NOTE for geeks: 278 NOTE for geeks:
278 The jump optimization changes the kprobe's pre 279 The jump optimization changes the kprobe's pre_handler behavior.
279 Without optimization, the pre_handler can chan 280 Without optimization, the pre_handler can change the kernel's execution
280 path by changing regs->ip and returning 1. Ho 281 path by changing regs->ip and returning 1. However, when the probe
281 is optimized, that modification is ignored. T 282 is optimized, that modification is ignored. Thus, if you want to
282 tweak the kernel's execution path, you need to 283 tweak the kernel's execution path, you need to suppress optimization,
283 using one of the following techniques: 284 using one of the following techniques:
284 285
285 - Specify an empty function for the kprobe's p 286 - Specify an empty function for the kprobe's post_handler.
286 287
287 or 288 or
288 289
289 - Execute 'sysctl -w debug.kprobes_optimizatio 290 - Execute 'sysctl -w debug.kprobes_optimization=n'
290 291
291 .. _kprobes_blacklist: 292 .. _kprobes_blacklist:
292 293
293 Blacklist 294 Blacklist
294 --------- 295 ---------
295 296
296 Kprobes can probe most of the kernel except it 297 Kprobes can probe most of the kernel except itself. This means
297 that there are some functions where kprobes ca 298 that there are some functions where kprobes cannot probe. Probing
298 (trapping) such functions can cause a recursiv 299 (trapping) such functions can cause a recursive trap (e.g. double
299 fault) or the nested probe handler may never b 300 fault) or the nested probe handler may never be called.
300 Kprobes manages such functions as a blacklist. 301 Kprobes manages such functions as a blacklist.
301 If you want to add a function into the blackli 302 If you want to add a function into the blacklist, you just need
302 to (1) include linux/kprobes.h and (2) use NOK 303 to (1) include linux/kprobes.h and (2) use NOKPROBE_SYMBOL() macro
303 to specify a blacklisted function. 304 to specify a blacklisted function.
304 Kprobes checks the given probe address against 305 Kprobes checks the given probe address against the blacklist and
305 rejects registering it, if the given address i 306 rejects registering it, if the given address is in the blacklist.
306 307
307 .. _kprobes_archs_supported: 308 .. _kprobes_archs_supported:
308 309
309 Architectures Supported 310 Architectures Supported
310 ======================= 311 =======================
311 312
312 Kprobes and return probes are implemented on t 313 Kprobes and return probes are implemented on the following
313 architectures: 314 architectures:
314 315
315 - i386 (Supports jump optimization) 316 - i386 (Supports jump optimization)
316 - x86_64 (AMD-64, EM64T) (Supports jump optimi 317 - x86_64 (AMD-64, EM64T) (Supports jump optimization)
317 - ppc64 318 - ppc64
>> 319 - ia64 (Does not support probes on instruction slot1.)
318 - sparc64 (Return probes not yet implemented.) 320 - sparc64 (Return probes not yet implemented.)
319 - arm 321 - arm
320 - ppc 322 - ppc
321 - mips 323 - mips
322 - s390 324 - s390
323 - parisc 325 - parisc
324 - loongarch <<
325 - riscv <<
326 326
327 Configuring Kprobes 327 Configuring Kprobes
328 =================== 328 ===================
329 329
330 When configuring the kernel using make menucon 330 When configuring the kernel using make menuconfig/xconfig/oldconfig,
331 ensure that CONFIG_KPROBES is set to "y", look !! 331 ensure that CONFIG_KPROBES is set to "y". Under "General setup", look
332 "General architecture-dependent options". !! 332 for "Kprobes".
333 333
334 So that you can load and unload Kprobes-based 334 So that you can load and unload Kprobes-based instrumentation modules,
335 make sure "Loadable module support" (CONFIG_MO 335 make sure "Loadable module support" (CONFIG_MODULES) and "Module
336 unloading" (CONFIG_MODULE_UNLOAD) are set to " 336 unloading" (CONFIG_MODULE_UNLOAD) are set to "y".
337 337
338 Also make sure that CONFIG_KALLSYMS and perhap 338 Also make sure that CONFIG_KALLSYMS and perhaps even CONFIG_KALLSYMS_ALL
339 are set to "y", since kallsyms_lookup_name() i 339 are set to "y", since kallsyms_lookup_name() is used by the in-kernel
340 kprobe address resolution code. 340 kprobe address resolution code.
341 341
342 If you need to insert a probe in the middle of 342 If you need to insert a probe in the middle of a function, you may find
343 it useful to "Compile the kernel with debug in 343 it useful to "Compile the kernel with debug info" (CONFIG_DEBUG_INFO),
344 so you can use "objdump -d -l vmlinux" to see 344 so you can use "objdump -d -l vmlinux" to see the source-to-object
345 code mapping. 345 code mapping.
346 346
347 API Reference 347 API Reference
348 ============= 348 =============
349 349
350 The Kprobes API includes a "register" function 350 The Kprobes API includes a "register" function and an "unregister"
351 function for each type of probe. The API also 351 function for each type of probe. The API also includes "register_*probes"
352 and "unregister_*probes" functions for (un)reg 352 and "unregister_*probes" functions for (un)registering arrays of probes.
353 Here are terse, mini-man-page specifications f 353 Here are terse, mini-man-page specifications for these functions and
354 the associated probe handlers that you'll writ 354 the associated probe handlers that you'll write. See the files in the
355 samples/kprobes/ sub-directory for examples. 355 samples/kprobes/ sub-directory for examples.
356 356
357 register_kprobe 357 register_kprobe
358 --------------- 358 ---------------
359 359
360 :: 360 ::
361 361
362 #include <linux/kprobes.h> 362 #include <linux/kprobes.h>
363 int register_kprobe(struct kprobe *kp) 363 int register_kprobe(struct kprobe *kp);
364 364
365 Sets a breakpoint at the address kp->addr. Wh !! 365 Sets a breakpoint at the address kp->addr. When the breakpoint is
366 calls kp->pre_handler. After the probed instr !! 366 hit, Kprobes calls kp->pre_handler. After the probed instruction
367 calls kp->post_handler. Any or all handlers c !! 367 is single-stepped, Kprobe calls kp->post_handler. If a fault
368 KPROBE_FLAG_DISABLED, that kp will be register !! 368 occurs during execution of kp->pre_handler or kp->post_handler,
369 aren't hit until calling enable_kprobe(kp). !! 369 or during single-stepping of the probed instruction, Kprobes calls
>> 370 kp->fault_handler. Any or all handlers can be NULL. If kp->flags
>> 371 is set KPROBE_FLAG_DISABLED, that kp will be registered but disabled,
>> 372 so, its handlers aren't hit until calling enable_kprobe(kp).
370 373
371 .. note:: 374 .. note::
372 375
373 1. With the introduction of the "symbol_nam 376 1. With the introduction of the "symbol_name" field to struct kprobe,
374 the probepoint address resolution will n 377 the probepoint address resolution will now be taken care of by the kernel.
375 The following will now work:: 378 The following will now work::
376 379
377 kp.symbol_name = "symbol_name"; 380 kp.symbol_name = "symbol_name";
378 381
379 (64-bit powerpc intricacies such as func 382 (64-bit powerpc intricacies such as function descriptors are handled
380 transparently) 383 transparently)
381 384
382 2. Use the "offset" field of struct kprobe 385 2. Use the "offset" field of struct kprobe if the offset into the symbol
383 to install a probepoint is known. This f 386 to install a probepoint is known. This field is used to calculate the
384 probepoint. 387 probepoint.
385 388
386 3. Specify either the kprobe "symbol_name" 389 3. Specify either the kprobe "symbol_name" OR the "addr". If both are
387 specified, kprobe registration will fail 390 specified, kprobe registration will fail with -EINVAL.
388 391
389 4. With CISC architectures (such as i386 an 392 4. With CISC architectures (such as i386 and x86_64), the kprobes code
390 does not validate if the kprobe.addr is 393 does not validate if the kprobe.addr is at an instruction boundary.
391 Use "offset" with caution. 394 Use "offset" with caution.
392 395
393 register_kprobe() returns 0 on success, or a n 396 register_kprobe() returns 0 on success, or a negative errno otherwise.
394 397
395 User's pre-handler (kp->pre_handler):: 398 User's pre-handler (kp->pre_handler)::
396 399
397 #include <linux/kprobes.h> 400 #include <linux/kprobes.h>
398 #include <linux/ptrace.h> 401 #include <linux/ptrace.h>
399 int pre_handler(struct kprobe *p, stru 402 int pre_handler(struct kprobe *p, struct pt_regs *regs);
400 403
401 Called with p pointing to the kprobe associate 404 Called with p pointing to the kprobe associated with the breakpoint,
402 and regs pointing to the struct containing the 405 and regs pointing to the struct containing the registers saved when
403 the breakpoint was hit. Return 0 here unless 406 the breakpoint was hit. Return 0 here unless you're a Kprobes geek.
404 407
405 User's post-handler (kp->post_handler):: 408 User's post-handler (kp->post_handler)::
406 409
407 #include <linux/kprobes.h> 410 #include <linux/kprobes.h>
408 #include <linux/ptrace.h> 411 #include <linux/ptrace.h>
409 void post_handler(struct kprobe *p, st 412 void post_handler(struct kprobe *p, struct pt_regs *regs,
410 unsigned long flags) 413 unsigned long flags);
411 414
412 p and regs are as described for the pre_handle 415 p and regs are as described for the pre_handler. flags always seems
413 to be zero. 416 to be zero.
414 417
>> 418 User's fault-handler (kp->fault_handler)::
>> 419
>> 420 #include <linux/kprobes.h>
>> 421 #include <linux/ptrace.h>
>> 422 int fault_handler(struct kprobe *p, struct pt_regs *regs, int trapnr);
>> 423
>> 424 p and regs are as described for the pre_handler. trapnr is the
>> 425 architecture-specific trap number associated with the fault (e.g.,
>> 426 on i386, 13 for a general protection fault or 14 for a page fault).
>> 427 Returns 1 if it successfully handled the exception.
>> 428
415 register_kretprobe 429 register_kretprobe
416 ------------------ 430 ------------------
417 431
418 :: 432 ::
419 433
420 #include <linux/kprobes.h> 434 #include <linux/kprobes.h>
421 int register_kretprobe(struct kretprob 435 int register_kretprobe(struct kretprobe *rp);
422 436
423 Establishes a return probe for the function wh 437 Establishes a return probe for the function whose address is
424 rp->kp.addr. When that function returns, Kpro 438 rp->kp.addr. When that function returns, Kprobes calls rp->handler.
425 You must set rp->maxactive appropriately befor 439 You must set rp->maxactive appropriately before you call
426 register_kretprobe(); see "How Does a Return P 440 register_kretprobe(); see "How Does a Return Probe Work?" for details.
427 441
428 register_kretprobe() returns 0 on success, or 442 register_kretprobe() returns 0 on success, or a negative errno
429 otherwise. 443 otherwise.
430 444
431 User's return-probe handler (rp->handler):: 445 User's return-probe handler (rp->handler)::
432 446
433 #include <linux/kprobes.h> 447 #include <linux/kprobes.h>
434 #include <linux/ptrace.h> 448 #include <linux/ptrace.h>
435 int kretprobe_handler(struct kretprobe 449 int kretprobe_handler(struct kretprobe_instance *ri,
436 struct pt_regs * 450 struct pt_regs *regs);
437 451
438 regs is as described for kprobe.pre_handler. 452 regs is as described for kprobe.pre_handler. ri points to the
439 kretprobe_instance object, of which the follow 453 kretprobe_instance object, of which the following fields may be
440 of interest: 454 of interest:
441 455
442 - ret_addr: the return address 456 - ret_addr: the return address
443 - rp: points to the corresponding kretprobe ob 457 - rp: points to the corresponding kretprobe object
444 - task: points to the corresponding task struc 458 - task: points to the corresponding task struct
445 - data: points to per return-instance private 459 - data: points to per return-instance private data; see "Kretprobe
446 entry-handler" for details. 460 entry-handler" for details.
447 461
448 The regs_return_value(regs) macro provides a s 462 The regs_return_value(regs) macro provides a simple abstraction to
449 extract the return value from the appropriate 463 extract the return value from the appropriate register as defined by
450 the architecture's ABI. 464 the architecture's ABI.
451 465
452 The handler's return value is currently ignore 466 The handler's return value is currently ignored.
453 467
454 unregister_*probe 468 unregister_*probe
455 ------------------ 469 ------------------
456 470
457 :: 471 ::
458 472
459 #include <linux/kprobes.h> 473 #include <linux/kprobes.h>
460 void unregister_kprobe(struct kprobe * 474 void unregister_kprobe(struct kprobe *kp);
461 void unregister_kretprobe(struct kretp 475 void unregister_kretprobe(struct kretprobe *rp);
462 476
463 Removes the specified probe. The unregister f 477 Removes the specified probe. The unregister function can be called
464 at any time after the probe has been registere 478 at any time after the probe has been registered.
465 479
466 .. note:: 480 .. note::
467 481
468 If the functions find an incorrect probe (e 482 If the functions find an incorrect probe (ex. an unregistered probe),
469 they clear the addr field of the probe. 483 they clear the addr field of the probe.
470 484
471 register_*probes 485 register_*probes
472 ---------------- 486 ----------------
473 487
474 :: 488 ::
475 489
476 #include <linux/kprobes.h> 490 #include <linux/kprobes.h>
477 int register_kprobes(struct kprobe **k 491 int register_kprobes(struct kprobe **kps, int num);
478 int register_kretprobes(struct kretpro 492 int register_kretprobes(struct kretprobe **rps, int num);
479 493
480 Registers each of the num probes in the specif 494 Registers each of the num probes in the specified array. If any
481 error occurs during registration, all probes i 495 error occurs during registration, all probes in the array, up to
482 the bad probe, are safely unregistered before 496 the bad probe, are safely unregistered before the register_*probes
483 function returns. 497 function returns.
484 498
485 - kps/rps: an array of pointers to ``*probe`` 499 - kps/rps: an array of pointers to ``*probe`` data structures
486 - num: the number of the array entries. 500 - num: the number of the array entries.
487 501
488 .. note:: 502 .. note::
489 503
490 You have to allocate(or define) an array of 504 You have to allocate(or define) an array of pointers and set all
491 of the array entries before using these fun 505 of the array entries before using these functions.
492 506
493 unregister_*probes 507 unregister_*probes
494 ------------------ 508 ------------------
495 509
496 :: 510 ::
497 511
498 #include <linux/kprobes.h> 512 #include <linux/kprobes.h>
499 void unregister_kprobes(struct kprobe 513 void unregister_kprobes(struct kprobe **kps, int num);
500 void unregister_kretprobes(struct kret 514 void unregister_kretprobes(struct kretprobe **rps, int num);
501 515
502 Removes each of the num probes in the specifie 516 Removes each of the num probes in the specified array at once.
503 517
504 .. note:: 518 .. note::
505 519
506 If the functions find some incorrect probes 520 If the functions find some incorrect probes (ex. unregistered
507 probes) in the specified array, they clear 521 probes) in the specified array, they clear the addr field of those
508 incorrect probes. However, other probes in 522 incorrect probes. However, other probes in the array are
509 unregistered correctly. 523 unregistered correctly.
510 524
511 disable_*probe 525 disable_*probe
512 -------------- 526 --------------
513 527
514 :: 528 ::
515 529
516 #include <linux/kprobes.h> 530 #include <linux/kprobes.h>
517 int disable_kprobe(struct kprobe *kp); 531 int disable_kprobe(struct kprobe *kp);
518 int disable_kretprobe(struct kretprobe 532 int disable_kretprobe(struct kretprobe *rp);
519 533
520 Temporarily disables the specified ``*probe``. 534 Temporarily disables the specified ``*probe``. You can enable it again by using
521 enable_*probe(). You must specify the probe wh 535 enable_*probe(). You must specify the probe which has been registered.
522 536
523 enable_*probe 537 enable_*probe
524 ------------- 538 -------------
525 539
526 :: 540 ::
527 541
528 #include <linux/kprobes.h> 542 #include <linux/kprobes.h>
529 int enable_kprobe(struct kprobe *kp); 543 int enable_kprobe(struct kprobe *kp);
530 int enable_kretprobe(struct kretprobe 544 int enable_kretprobe(struct kretprobe *rp);
531 545
532 Enables ``*probe`` which has been disabled by 546 Enables ``*probe`` which has been disabled by disable_*probe(). You must specify
533 the probe which has been registered. 547 the probe which has been registered.
534 548
535 Kprobes Features and Limitations 549 Kprobes Features and Limitations
536 ================================ 550 ================================
537 551
538 Kprobes allows multiple probes at the same add 552 Kprobes allows multiple probes at the same address. Also,
539 a probepoint for which there is a post_handler 553 a probepoint for which there is a post_handler cannot be optimized.
540 So if you install a kprobe with a post_handler 554 So if you install a kprobe with a post_handler, at an optimized
541 probepoint, the probepoint will be unoptimized 555 probepoint, the probepoint will be unoptimized automatically.
542 556
543 In general, you can install a probe anywhere i 557 In general, you can install a probe anywhere in the kernel.
544 In particular, you can probe interrupt handler 558 In particular, you can probe interrupt handlers. Known exceptions
545 are discussed in this section. 559 are discussed in this section.
546 560
547 The register_*probe functions will return -EIN 561 The register_*probe functions will return -EINVAL if you attempt
548 to install a probe in the code that implements 562 to install a probe in the code that implements Kprobes (mostly
549 kernel/kprobes.c and ``arch/*/kernel/kprobes.c 563 kernel/kprobes.c and ``arch/*/kernel/kprobes.c``, but also functions such
550 as do_page_fault and notifier_call_chain). 564 as do_page_fault and notifier_call_chain).
551 565
552 If you install a probe in an inline-able funct 566 If you install a probe in an inline-able function, Kprobes makes
553 no attempt to chase down all inline instances 567 no attempt to chase down all inline instances of the function and
554 install probes there. gcc may inline a functi 568 install probes there. gcc may inline a function without being asked,
555 so keep this in mind if you're not seeing the 569 so keep this in mind if you're not seeing the probe hits you expect.
556 570
557 A probe handler can modify the environment of 571 A probe handler can modify the environment of the probed function
558 -- e.g., by modifying kernel data structures, 572 -- e.g., by modifying kernel data structures, or by modifying the
559 contents of the pt_regs struct (which are rest 573 contents of the pt_regs struct (which are restored to the registers
560 upon return from the breakpoint). So Kprobes 574 upon return from the breakpoint). So Kprobes can be used, for example,
561 to install a bug fix or to inject faults for t 575 to install a bug fix or to inject faults for testing. Kprobes, of
562 course, has no way to distinguish the delibera 576 course, has no way to distinguish the deliberately injected faults
563 from the accidental ones. Don't drink and pro 577 from the accidental ones. Don't drink and probe.
564 578
565 Kprobes makes no attempt to prevent probe hand 579 Kprobes makes no attempt to prevent probe handlers from stepping on
566 each other -- e.g., probing printk() and then 580 each other -- e.g., probing printk() and then calling printk() from a
567 probe handler. If a probe handler hits a prob 581 probe handler. If a probe handler hits a probe, that second probe's
568 handlers won't be run in that instance, and th 582 handlers won't be run in that instance, and the kprobe.nmissed member
569 of the second probe will be incremented. 583 of the second probe will be incremented.
570 584
571 As of Linux v2.6.15-rc1, multiple handlers (or 585 As of Linux v2.6.15-rc1, multiple handlers (or multiple instances of
572 the same handler) may run concurrently on diff 586 the same handler) may run concurrently on different CPUs.
573 587
574 Kprobes does not use mutexes or allocate memor 588 Kprobes does not use mutexes or allocate memory except during
575 registration and unregistration. 589 registration and unregistration.
576 590
577 Probe handlers are run with preemption disable 591 Probe handlers are run with preemption disabled or interrupt disabled,
578 which depends on the architecture and optimiza 592 which depends on the architecture and optimization state. (e.g.,
579 kretprobe handlers and optimized kprobe handle 593 kretprobe handlers and optimized kprobe handlers run without interrupt
580 disabled on x86/x86-64). In any case, your ha 594 disabled on x86/x86-64). In any case, your handler should not yield
581 the CPU (e.g., by attempting to acquire a sema 595 the CPU (e.g., by attempting to acquire a semaphore, or waiting I/O).
582 596
583 Since a return probe is implemented by replaci 597 Since a return probe is implemented by replacing the return
584 address with the trampoline's address, stack b 598 address with the trampoline's address, stack backtraces and calls
585 to __builtin_return_address() will typically y 599 to __builtin_return_address() will typically yield the trampoline's
586 address instead of the real return address for 600 address instead of the real return address for kretprobed functions.
587 (As far as we can tell, __builtin_return_addre 601 (As far as we can tell, __builtin_return_address() is used only
588 for instrumentation and error reporting.) 602 for instrumentation and error reporting.)
589 603
590 If the number of times a function is called do 604 If the number of times a function is called does not match the number
591 of times it returns, registering a return prob 605 of times it returns, registering a return probe on that function may
592 produce undesirable results. In such a case, a 606 produce undesirable results. In such a case, a line:
593 kretprobe BUG!: Processing kretprobe d00000000 607 kretprobe BUG!: Processing kretprobe d000000000041aa8 @ c00000000004f48c
594 gets printed. With this information, one will 608 gets printed. With this information, one will be able to correlate the
595 exact instance of the kretprobe that caused th 609 exact instance of the kretprobe that caused the problem. We have the
596 do_exit() case covered. do_execve() and do_for 610 do_exit() case covered. do_execve() and do_fork() are not an issue.
597 We're unaware of other specific cases where th 611 We're unaware of other specific cases where this could be a problem.
598 612
599 If, upon entry to or exit from a function, the 613 If, upon entry to or exit from a function, the CPU is running on
600 a stack other than that of the current task, r 614 a stack other than that of the current task, registering a return
601 probe on that function may produce undesirable 615 probe on that function may produce undesirable results. For this
602 reason, Kprobes doesn't support return probes 616 reason, Kprobes doesn't support return probes (or kprobes)
603 on the x86_64 version of __switch_to(); the re 617 on the x86_64 version of __switch_to(); the registration functions
604 return -EINVAL. 618 return -EINVAL.
605 619
606 On x86/x86-64, since the Jump Optimization of 620 On x86/x86-64, since the Jump Optimization of Kprobes modifies
607 instructions widely, there are some limitation 621 instructions widely, there are some limitations to optimization. To
608 explain it, we introduce some terminology. Ima 622 explain it, we introduce some terminology. Imagine a 3-instruction
609 sequence consisting of a two 2-byte instructio 623 sequence consisting of a two 2-byte instructions and one 3-byte
610 instruction. 624 instruction.
611 625
612 :: 626 ::
613 627
614 IA 628 IA
615 | 629 |
616 [-2][-1][0][1][2][3][4][5][6][7] 630 [-2][-1][0][1][2][3][4][5][6][7]
617 [ins1][ins2][ ins3 ] 631 [ins1][ins2][ ins3 ]
618 [<- DCR ->] 632 [<- DCR ->]
619 [<- JTPR ->] 633 [<- JTPR ->]
620 634
621 ins1: 1st Instruction 635 ins1: 1st Instruction
622 ins2: 2nd Instruction 636 ins2: 2nd Instruction
623 ins3: 3rd Instruction 637 ins3: 3rd Instruction
624 IA: Insertion Address 638 IA: Insertion Address
625 JTPR: Jump Target Prohibition Region 639 JTPR: Jump Target Prohibition Region
626 DCR: Detoured Code Region 640 DCR: Detoured Code Region
627 641
628 The instructions in DCR are copied to the out- 642 The instructions in DCR are copied to the out-of-line buffer
629 of the kprobe, because the bytes in DCR are re 643 of the kprobe, because the bytes in DCR are replaced by
630 a 5-byte jump instruction. So there are severa 644 a 5-byte jump instruction. So there are several limitations.
631 645
632 a) The instructions in DCR must be relocatable 646 a) The instructions in DCR must be relocatable.
633 b) The instructions in DCR must not include a 647 b) The instructions in DCR must not include a call instruction.
634 c) JTPR must not be targeted by any jump or ca 648 c) JTPR must not be targeted by any jump or call instruction.
635 d) DCR must not straddle the border between fu 649 d) DCR must not straddle the border between functions.
636 650
637 Anyway, these limitations are checked by the i 651 Anyway, these limitations are checked by the in-kernel instruction
638 decoder, so you don't need to worry about that 652 decoder, so you don't need to worry about that.
639 653
640 Probe Overhead 654 Probe Overhead
641 ============== 655 ==============
642 656
643 On a typical CPU in use in 2005, a kprobe hit 657 On a typical CPU in use in 2005, a kprobe hit takes 0.5 to 1.0
644 microseconds to process. Specifically, a benc 658 microseconds to process. Specifically, a benchmark that hits the same
645 probepoint repeatedly, firing a simple handler 659 probepoint repeatedly, firing a simple handler each time, reports 1-2
646 million hits per second, depending on the arch 660 million hits per second, depending on the architecture. A return-probe
647 hit typically takes 50-75% longer than a kprob 661 hit typically takes 50-75% longer than a kprobe hit.
648 When you have a return probe set on a function 662 When you have a return probe set on a function, adding a kprobe at
649 the entry to that function adds essentially no 663 the entry to that function adds essentially no overhead.
650 664
651 Here are sample overhead figures (in usec) for 665 Here are sample overhead figures (in usec) for different architectures::
652 666
653 k = kprobe; r = return probe; kr = kprobe + 667 k = kprobe; r = return probe; kr = kprobe + return probe
654 on same function 668 on same function
655 669
656 i386: Intel Pentium M, 1495 MHz, 2957.31 bog 670 i386: Intel Pentium M, 1495 MHz, 2957.31 bogomips
657 k = 0.57 usec; r = 0.92; kr = 0.99 671 k = 0.57 usec; r = 0.92; kr = 0.99
658 672
659 x86_64: AMD Opteron 246, 1994 MHz, 3971.48 b 673 x86_64: AMD Opteron 246, 1994 MHz, 3971.48 bogomips
660 k = 0.49 usec; r = 0.80; kr = 0.82 674 k = 0.49 usec; r = 0.80; kr = 0.82
661 675
662 ppc64: POWER5 (gr), 1656 MHz (SMT disabled, 676 ppc64: POWER5 (gr), 1656 MHz (SMT disabled, 1 virtual CPU per physical CPU)
663 k = 0.77 usec; r = 1.26; kr = 1.45 677 k = 0.77 usec; r = 1.26; kr = 1.45
664 678
665 Optimized Probe Overhead 679 Optimized Probe Overhead
666 ------------------------ 680 ------------------------
667 681
668 Typically, an optimized kprobe hit takes 0.07 682 Typically, an optimized kprobe hit takes 0.07 to 0.1 microseconds to
669 process. Here are sample overhead figures (in 683 process. Here are sample overhead figures (in usec) for x86 architectures::
670 684
671 k = unoptimized kprobe, b = boosted (single- 685 k = unoptimized kprobe, b = boosted (single-step skipped), o = optimized kprobe,
672 r = unoptimized kretprobe, rb = boosted kret 686 r = unoptimized kretprobe, rb = boosted kretprobe, ro = optimized kretprobe.
673 687
674 i386: Intel(R) Xeon(R) E5410, 2.33GHz, 4656. 688 i386: Intel(R) Xeon(R) E5410, 2.33GHz, 4656.90 bogomips
675 k = 0.80 usec; b = 0.33; o = 0.05; r = 1.10; 689 k = 0.80 usec; b = 0.33; o = 0.05; r = 1.10; rb = 0.61; ro = 0.33
676 690
677 x86-64: Intel(R) Xeon(R) E5410, 2.33GHz, 465 691 x86-64: Intel(R) Xeon(R) E5410, 2.33GHz, 4656.90 bogomips
678 k = 0.99 usec; b = 0.43; o = 0.06; r = 1.24; 692 k = 0.99 usec; b = 0.43; o = 0.06; r = 1.24; rb = 0.68; ro = 0.30
679 693
680 TODO 694 TODO
681 ==== 695 ====
682 696
683 a. SystemTap (http://sourceware.org/systemtap) 697 a. SystemTap (http://sourceware.org/systemtap): Provides a simplified
684 programming interface for probe-based instr 698 programming interface for probe-based instrumentation. Try it out.
685 b. Kernel return probes for sparc64. 699 b. Kernel return probes for sparc64.
686 c. Support for other architectures. 700 c. Support for other architectures.
687 d. User-space probes. 701 d. User-space probes.
688 e. Watchpoint probes (which fire on data refer 702 e. Watchpoint probes (which fire on data references).
689 703
690 Kprobes Example 704 Kprobes Example
691 =============== 705 ===============
692 706
693 See samples/kprobes/kprobe_example.c 707 See samples/kprobes/kprobe_example.c
694 708
695 Kretprobes Example 709 Kretprobes Example
696 ================== 710 ==================
697 711
698 See samples/kprobes/kretprobe_example.c 712 See samples/kprobes/kretprobe_example.c
699 713
700 Deprecated Features 714 Deprecated Features
701 =================== 715 ===================
702 716
703 Jprobes is now a deprecated feature. People wh 717 Jprobes is now a deprecated feature. People who are depending on it should
704 migrate to other tracing features or use older 718 migrate to other tracing features or use older kernels. Please consider to
705 migrate your tool to one of the following opti 719 migrate your tool to one of the following options:
706 720
707 - Use trace-event to trace target function wit 721 - Use trace-event to trace target function with arguments.
708 722
709 trace-event is a low-overhead (and almost no 723 trace-event is a low-overhead (and almost no visible overhead if it
710 is off) statically defined event interface. 724 is off) statically defined event interface. You can define new events
711 and trace it via ftrace or any other tracing 725 and trace it via ftrace or any other tracing tools.
712 726
713 See the following urls: 727 See the following urls:
714 728
715 - https://lwn.net/Articles/379903/ 729 - https://lwn.net/Articles/379903/
716 - https://lwn.net/Articles/381064/ 730 - https://lwn.net/Articles/381064/
717 - https://lwn.net/Articles/383362/ 731 - https://lwn.net/Articles/383362/
718 732
719 - Use ftrace dynamic events (kprobe event) wit 733 - Use ftrace dynamic events (kprobe event) with perf-probe.
720 734
721 If you build your kernel with debug info (CO 735 If you build your kernel with debug info (CONFIG_DEBUG_INFO=y), you can
722 find which register/stack is assigned to whi 736 find which register/stack is assigned to which local variable or arguments
723 by using perf-probe and set up new event to 737 by using perf-probe and set up new event to trace it.
724 738
725 See following documents: 739 See following documents:
726 740
727 - Documentation/trace/kprobetrace.rst 741 - Documentation/trace/kprobetrace.rst
728 - Documentation/trace/events.rst 742 - Documentation/trace/events.rst
729 - tools/perf/Documentation/perf-probe.txt 743 - tools/perf/Documentation/perf-probe.txt
730 744
731 745
732 The kprobes debugfs interface 746 The kprobes debugfs interface
733 ============================= 747 =============================
734 748
735 749
736 With recent kernels (> 2.6.20) the list of reg 750 With recent kernels (> 2.6.20) the list of registered kprobes is visible
737 under the /sys/kernel/debug/kprobes/ directory 751 under the /sys/kernel/debug/kprobes/ directory (assuming debugfs is mounted at //sys/kernel/debug).
738 752
739 /sys/kernel/debug/kprobes/list: Lists all regi 753 /sys/kernel/debug/kprobes/list: Lists all registered probes on the system::
740 754
741 c015d71a k vfs_read+0x0 755 c015d71a k vfs_read+0x0
742 c03dedc5 r tcp_v4_rcv+0x0 756 c03dedc5 r tcp_v4_rcv+0x0
743 757
744 The first column provides the kernel address w 758 The first column provides the kernel address where the probe is inserted.
745 The second column identifies the type of probe 759 The second column identifies the type of probe (k - kprobe and r - kretprobe)
746 while the third column specifies the symbol+of 760 while the third column specifies the symbol+offset of the probe.
747 If the probed function belongs to a module, th 761 If the probed function belongs to a module, the module name is also
748 specified. Following columns show probe status 762 specified. Following columns show probe status. If the probe is on
749 a virtual address that is no longer valid (mod 763 a virtual address that is no longer valid (module init sections, module
750 virtual addresses that correspond to modules t 764 virtual addresses that correspond to modules that've been unloaded),
751 such probes are marked with [GONE]. If the pro 765 such probes are marked with [GONE]. If the probe is temporarily disabled,
752 such probes are marked with [DISABLED]. If the 766 such probes are marked with [DISABLED]. If the probe is optimized, it is
753 marked with [OPTIMIZED]. If the probe is ftrac 767 marked with [OPTIMIZED]. If the probe is ftrace-based, it is marked with
754 [FTRACE]. 768 [FTRACE].
755 769
756 /sys/kernel/debug/kprobes/enabled: Turn kprobe 770 /sys/kernel/debug/kprobes/enabled: Turn kprobes ON/OFF forcibly.
757 771
758 Provides a knob to globally and forcibly turn 772 Provides a knob to globally and forcibly turn registered kprobes ON or OFF.
759 By default, all kprobes are enabled. By echoin 773 By default, all kprobes are enabled. By echoing "0" to this file, all
760 registered probes will be disarmed, till such 774 registered probes will be disarmed, till such time a "1" is echoed to this
761 file. Note that this knob just disarms and arm 775 file. Note that this knob just disarms and arms all kprobes and doesn't
762 change each probe's disabling state. This mean 776 change each probe's disabling state. This means that disabled kprobes (marked
763 [DISABLED]) will be not enabled if you turn ON 777 [DISABLED]) will be not enabled if you turn ON all kprobes by this knob.
764 778
765 779
766 The kprobes sysctl interface 780 The kprobes sysctl interface
767 ============================ 781 ============================
768 782
769 /proc/sys/debug/kprobes-optimization: Turn kpr 783 /proc/sys/debug/kprobes-optimization: Turn kprobes optimization ON/OFF.
770 784
771 When CONFIG_OPTPROBES=y, this sysctl interface 785 When CONFIG_OPTPROBES=y, this sysctl interface appears and it provides
772 a knob to globally and forcibly turn jump opti 786 a knob to globally and forcibly turn jump optimization (see section
773 :ref:`kprobes_jump_optimization`) ON or OFF. B 787 :ref:`kprobes_jump_optimization`) ON or OFF. By default, jump optimization
774 is allowed (ON). If you echo "0" to this file 788 is allowed (ON). If you echo "0" to this file or set
775 "debug.kprobes_optimization" to 0 via sysctl, 789 "debug.kprobes_optimization" to 0 via sysctl, all optimized probes will be
776 unoptimized, and any new probes registered aft 790 unoptimized, and any new probes registered after that will not be optimized.
777 791
778 Note that this knob *changes* the optimized st 792 Note that this knob *changes* the optimized state. This means that optimized
779 probes (marked [OPTIMIZED]) will be unoptimize 793 probes (marked [OPTIMIZED]) will be unoptimized ([OPTIMIZED] tag will be
780 removed). If the knob is turned on, they will 794 removed). If the knob is turned on, they will be optimized again.
781 795
782 References 796 References
783 ========== 797 ==========
784 798
785 For additional information on Kprobes, refer t 799 For additional information on Kprobes, refer to the following URLs:
786 800
787 - https://lwn.net/Articles/132196/ !! 801 - https://www.ibm.com/developerworks/library/l-kprobes/index.html
788 - https://www.kernel.org/doc/ols/2006/ols2006v 802 - https://www.kernel.org/doc/ols/2006/ols2006v2-pages-109-124.pdf
789 803
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.