1 =================================
2 Using ftrace to hook to functions
3 =================================
5 .. Copyright 2017 VMware Inc.
6 .. Author: Steven Rostedt <srostedt@goodmis.org>
7 .. License: The GNU Free Documentation License, Version 1.2
8 .. (dual licensed under the GPL v2)
15 The ftrace infrastructure was originally created to attach callbacks to the
16 beginning of functions in order to record and trace the flow of the kernel.
17 But callbacks to the start of a function can have other use cases. Either
18 for live kernel patching, or for security monitoring. This document describes
19 how to use ftrace to implement your own function callbacks.
26 The ability to add a callback to almost any function within the
27 kernel comes with risks. A callback can be called from any context
28 (normal, softirq, irq, and NMI). Callbacks can also be called just before
29 going to idle, during CPU bring up and takedown, or going to user space.
30 This requires extra care to what can be done inside a callback. A callback
31 can be called outside the protective scope of RCU.
33 There are helper functions to help against recursion, and making sure
34 RCU is watching. These are explained below.
37 The ftrace_ops structure
38 ========================
40 To register a function callback, a ftrace_ops is required. This structure
41 is used to tell ftrace what function should be called as the callback
42 as well as what protections the callback will perform and not require
45 There is only one field that is needed to be set when registering
46 an ftrace_ops with ftrace:
50 struct ftrace_ops ops = {
51 .func = my_callback_func,
52 .flags = MY_FTRACE_FLAGS
53 .private = any_private_data_structure,
56 Both .flags and .private are optional. Only .func is required.
58 To enable tracing call::
60 register_ftrace_function(&ops);
62 To disable tracing call::
64 unregister_ftrace_function(&ops);
66 The above is defined by including the header::
68 #include <linux/ftrace.h>
70 The registered callback will start being called some time after the
71 register_ftrace_function() is called and before it returns. The exact time
72 that callbacks start being called is dependent upon architecture and scheduling
73 of services. The callback itself will have to handle any synchronization if it
74 must begin at an exact moment.
76 The unregister_ftrace_function() will guarantee that the callback is
77 no longer being called by functions after the unregister_ftrace_function()
78 returns. Note that to perform this guarantee, the unregister_ftrace_function()
79 may take some time to finish.
85 The prototype of the callback function is as follows (as of v4.14):
89 void callback_func(unsigned long ip, unsigned long parent_ip,
90 struct ftrace_ops *op, struct pt_regs *regs);
93 This is the instruction pointer of the function that is being traced.
94 (where the fentry or mcount is within the function)
97 This is the instruction pointer of the function that called the
98 the function being traced (where the call of the function occurred).
101 This is a pointer to ftrace_ops that was used to register the callback.
102 This can be used to pass data to the callback via the private pointer.
105 If the FTRACE_OPS_FL_SAVE_REGS or FTRACE_OPS_FL_SAVE_REGS_IF_SUPPORTED
106 flags are set in the ftrace_ops structure, then this will be pointing
107 to the pt_regs structure like it would be if an breakpoint was placed
108 at the start of the function where ftrace was tracing. Otherwise it
109 either contains garbage, or NULL.
111 Protect your callback
112 =====================
114 As functions can be called from anywhere, and it is possible that a function
115 called by a callback may also be traced, and call that same callback,
116 recursion protection must be used. There are two helper functions that
117 can help in this regard. If you start your code with:
123 bit = ftrace_test_recursion_trylock(ip, parent_ip);
131 ftrace_test_recursion_unlock(bit);
133 The code in between will be safe to use, even if it ends up calling a
134 function that the callback is tracing. Note, on success,
135 ftrace_test_recursion_trylock() will disable preemption, and the
136 ftrace_test_recursion_unlock() will enable it again (if it was previously
137 enabled). The instruction pointer (ip) and its parent (parent_ip) is passed to
138 ftrace_test_recursion_trylock() to record where the recursion happened
139 (if CONFIG_FTRACE_RECORD_RECURSION is set).
141 Alternatively, if the FTRACE_OPS_FL_RECURSION flag is set on the ftrace_ops
142 (as explained below), then a helper trampoline will be used to test
143 for recursion for the callback and no recursion test needs to be done.
144 But this is at the expense of a slightly more overhead from an extra
147 If your callback accesses any data or critical section that requires RCU
148 protection, it is best to make sure that RCU is "watching", otherwise
149 that data or critical section will not be protected as expected. In this
154 if (!rcu_is_watching())
157 Alternatively, if the FTRACE_OPS_FL_RCU flag is set on the ftrace_ops
158 (as explained below), then a helper trampoline will be used to test
159 for rcu_is_watching for the callback and no other test needs to be done.
160 But this is at the expense of a slightly more overhead from an extra
167 The ftrace_ops flags are all defined and documented in include/linux/ftrace.h.
168 Some of the flags are used for internal infrastructure of ftrace, but the
169 ones that users should be aware of are the following:
171 FTRACE_OPS_FL_SAVE_REGS
172 If the callback requires reading or modifying the pt_regs
173 passed to the callback, then it must set this flag. Registering
174 a ftrace_ops with this flag set on an architecture that does not
175 support passing of pt_regs to the callback will fail.
177 FTRACE_OPS_FL_SAVE_REGS_IF_SUPPORTED
178 Similar to SAVE_REGS but the registering of a
179 ftrace_ops on an architecture that does not support passing of regs
180 will not fail with this flag set. But the callback must check if
181 regs is NULL or not to determine if the architecture supports it.
183 FTRACE_OPS_FL_RECURSION
184 By default, it is expected that the callback can handle recursion.
185 But if the callback is not that worried about overehead, then
186 setting this bit will add the recursion protection around the
187 callback by calling a helper function that will do the recursion
188 protection and only call the callback if it did not recurse.
190 Note, if this flag is not set, and recursion does occur, it could
191 cause the system to crash, and possibly reboot via a triple fault.
193 Not, if this flag is set, then the callback will always be called
194 with preemption disabled. If it is not set, then it is possible
195 (but not guaranteed) that the callback will be called in
198 FTRACE_OPS_FL_IPMODIFY
199 Requires FTRACE_OPS_FL_SAVE_REGS set. If the callback is to "hijack"
200 the traced function (have another function called instead of the
201 traced function), it requires setting this flag. This is what live
202 kernel patches uses. Without this flag the pt_regs->ip can not be
205 Note, only one ftrace_ops with FTRACE_OPS_FL_IPMODIFY set may be
206 registered to any given function at a time.
209 If this is set, then the callback will only be called by functions
210 where RCU is "watching". This is required if the callback function
211 performs any rcu_read_lock() operation.
213 RCU stops watching when the system goes idle, the time when a CPU
214 is taken down and comes back online, and when entering from kernel
215 to user space and back to kernel space. During these transitions,
216 a callback may be executed and RCU synchronization will not protect
219 FTRACE_OPS_FL_PERMANENT
220 If this is set on any ftrace ops, then the tracing cannot disabled by
221 writing 0 to the proc sysctl ftrace_enabled. Equally, a callback with
222 the flag set cannot be registered if ftrace_enabled is 0.
224 Livepatch uses it not to lose the function redirection, so the system
228 Filtering which functions to trace
229 ==================================
231 If a callback is only to be called from specific functions, a filter must be
232 set up. The filters are added by name, or ip if it is known.
236 int ftrace_set_filter(struct ftrace_ops *ops, unsigned char *buf,
240 The ops to set the filter with
243 The string that holds the function filter text.
245 The length of the string.
248 Non-zero to reset all filters before applying this filter.
250 Filters denote which functions should be enabled when tracing is enabled.
251 If @buf is NULL and reset is set, all functions will be enabled for tracing.
253 The @buf can also be a glob expression to enable all functions that
254 match a specific pattern.
256 See Filter Commands in :file:`Documentation/trace/ftrace.rst`.
258 To just trace the schedule function:
262 ret = ftrace_set_filter(&ops, "schedule", strlen("schedule"), 0);
264 To add more functions, call the ftrace_set_filter() more than once with the
265 @reset parameter set to zero. To remove the current filter set and replace it
266 with new functions defined by @buf, have @reset be non-zero.
268 To remove all the filtered functions and trace all functions:
272 ret = ftrace_set_filter(&ops, NULL, 0, 1);
275 Sometimes more than one function has the same name. To trace just a specific
276 function in this case, ftrace_set_filter_ip() can be used.
280 ret = ftrace_set_filter_ip(&ops, ip, 0, 0);
282 Although the ip must be the address where the call to fentry or mcount is
283 located in the function. This function is used by perf and kprobes that
284 gets the ip address from the user (usually using debug info from the kernel).
286 If a glob is used to set the filter, functions can be added to a "notrace"
287 list that will prevent those functions from calling the callback.
288 The "notrace" list takes precedence over the "filter" list. If the
289 two lists are non-empty and contain the same functions, the callback will not
290 be called by any function.
292 An empty "notrace" list means to allow all functions defined by the filter
297 int ftrace_set_notrace(struct ftrace_ops *ops, unsigned char *buf,
300 This takes the same parameters as ftrace_set_filter() but will add the
301 functions it finds to not be traced. This is a separate list from the
302 filter list, and this function does not modify the filter list.
304 A non-zero @reset will clear the "notrace" list before adding functions
305 that match @buf to it.
307 Clearing the "notrace" list is the same as clearing the filter list
311 ret = ftrace_set_notrace(&ops, NULL, 0, 1);
313 The filter and notrace lists may be changed at any time. If only a set of
314 functions should call the callback, it is best to set the filters before
315 registering the callback. But the changes may also happen after the callback
318 If a filter is in place, and the @reset is non-zero, and @buf contains a
319 matching glob to functions, the switch will happen during the time of
320 the ftrace_set_filter() call. At no time will all functions call the callback.
324 ftrace_set_filter(&ops, "schedule", strlen("schedule"), 1);
326 register_ftrace_function(&ops);
330 ftrace_set_filter(&ops, "try_to_wake_up", strlen("try_to_wake_up"), 1);
336 ftrace_set_filter(&ops, "schedule", strlen("schedule"), 1);
338 register_ftrace_function(&ops);
342 ftrace_set_filter(&ops, NULL, 0, 1);
344 ftrace_set_filter(&ops, "try_to_wake_up", strlen("try_to_wake_up"), 0);
346 As the latter will have a short time where all functions will call
347 the callback, between the time of the reset, and the time of the
348 new setting of the filter.