1==========================
2Kprobe-based Event Tracing
3==========================
4
5:Author: Masami Hiramatsu
6
7Overview
8--------
9These events are similar to tracepoint based events. Instead of Tracepoint,
10this is based on kprobes (kprobe and kretprobe). So it can probe wherever
11kprobes can probe (this means, all functions except those with
12__kprobes/nokprobe_inline annotation and those marked NOKPROBE_SYMBOL).
13Unlike the Tracepoint based event, this can be added and removed
14dynamically, on the fly.
15
16To enable this feature, build your kernel with CONFIG_KPROBE_EVENTS=y.
17
18Similar to the events tracer, this doesn't need to be activated via
19current_tracer. Instead of that, add probe points via
20/sys/kernel/debug/tracing/kprobe_events, and enable it via
21/sys/kernel/debug/tracing/events/kprobes/<EVENT>/enable.
22
23You can also use /sys/kernel/debug/tracing/dynamic_events instead of
24kprobe_events. That interface will provide unified access to other
25dynamic events too.
26
27Synopsis of kprobe_events
28-------------------------
29::
30
31  p[:[GRP/]EVENT] [MOD:]SYM[+offs]|MEMADDR [FETCHARGS]	: Set a probe
32  r[MAXACTIVE][:[GRP/]EVENT] [MOD:]SYM[+0] [FETCHARGS]	: Set a return probe
33  -:[GRP/]EVENT						: Clear a probe
34
35 GRP		: Group name. If omitted, use "kprobes" for it.
36 EVENT		: Event name. If omitted, the event name is generated
37		  based on SYM+offs or MEMADDR.
38 MOD		: Module name which has given SYM.
39 SYM[+offs]	: Symbol+offset where the probe is inserted.
40 MEMADDR	: Address where the probe is inserted.
41 MAXACTIVE	: Maximum number of instances of the specified function that
42		  can be probed simultaneously, or 0 for the default value
43		  as defined in Documentation/kprobes.txt section 1.3.1.
44
45 FETCHARGS	: Arguments. Each probe can have up to 128 args.
46  %REG		: Fetch register REG
47  @ADDR		: Fetch memory at ADDR (ADDR should be in kernel)
48  @SYM[+|-offs]	: Fetch memory at SYM +|- offs (SYM should be a data symbol)
49  $stackN	: Fetch Nth entry of stack (N >= 0)
50  $stack	: Fetch stack address.
51  $argN		: Fetch the Nth function argument. (N >= 1) (\*1)
52  $retval	: Fetch return value.(\*2)
53  $comm		: Fetch current task comm.
54  +|-[u]OFFS(FETCHARG) : Fetch memory at FETCHARG +|- OFFS address.(\*3)(\*4)
55  NAME=FETCHARG : Set NAME as the argument name of FETCHARG.
56  FETCHARG:TYPE : Set TYPE as the type of FETCHARG. Currently, basic types
57		  (u8/u16/u32/u64/s8/s16/s32/s64), hexadecimal types
58		  (x8/x16/x32/x64), "string", "ustring" and bitfield
59		  are supported.
60
61  (\*1) only for the probe on function entry (offs == 0).
62  (\*2) only for return probe.
63  (\*3) this is useful for fetching a field of data structures.
64  (\*4) "u" means user-space dereference. See :ref:`user_mem_access`.
65
66Types
67-----
68Several types are supported for fetch-args. Kprobe tracer will access memory
69by given type. Prefix 's' and 'u' means those types are signed and unsigned
70respectively. 'x' prefix implies it is unsigned. Traced arguments are shown
71in decimal ('s' and 'u') or hexadecimal ('x'). Without type casting, 'x32'
72or 'x64' is used depends on the architecture (e.g. x86-32 uses x32, and
73x86-64 uses x64).
74These value types can be an array. To record array data, you can add '[N]'
75(where N is a fixed number, less than 64) to the base type.
76E.g. 'x16[4]' means an array of x16 (2bytes hex) with 4 elements.
77Note that the array can be applied to memory type fetchargs, you can not
78apply it to registers/stack-entries etc. (for example, '$stack1:x8[8]' is
79wrong, but '+8($stack):x8[8]' is OK.)
80String type is a special type, which fetches a "null-terminated" string from
81kernel space. This means it will fail and store NULL if the string container
82has been paged out. "ustring" type is an alternative of string for user-space.
83See :ref:`user_mem_access` for more info..
84The string array type is a bit different from other types. For other base
85types, <base-type>[1] is equal to <base-type> (e.g. +0(%di):x32[1] is same
86as +0(%di):x32.) But string[1] is not equal to string. The string type itself
87represents "char array", but string array type represents "char * array".
88So, for example, +0(%di):string[1] is equal to +0(+0(%di)):string.
89Bitfield is another special type, which takes 3 parameters, bit-width, bit-
90offset, and container-size (usually 32). The syntax is::
91
92 b<bit-width>@<bit-offset>/<container-size>
93
94Symbol type('symbol') is an alias of u32 or u64 type (depends on BITS_PER_LONG)
95which shows given pointer in "symbol+offset" style.
96For $comm, the default type is "string"; any other type is invalid.
97
98.. _user_mem_access:
99User Memory Access
100------------------
101Kprobe events supports user-space memory access. For that purpose, you can use
102either user-space dereference syntax or 'ustring' type.
103
104The user-space dereference syntax allows you to access a field of a data
105structure in user-space. This is done by adding the "u" prefix to the
106dereference syntax. For example, +u4(%si) means it will read memory from the
107address in the register %si offset by 4, and the memory is expected to be in
108user-space. You can use this for strings too, e.g. +u0(%si):string will read
109a string from the address in the register %si that is expected to be in user-
110space. 'ustring' is a shortcut way of performing the same task. That is,
111+0(%si):ustring is equivalent to +u0(%si):string.
112
113Note that kprobe-event provides the user-memory access syntax but it doesn't
114use it transparently. This means if you use normal dereference or string type
115for user memory, it might fail, and may always fail on some archs. The user
116has to carefully check if the target data is in kernel or user space.
117
118Per-Probe Event Filtering
119-------------------------
120Per-probe event filtering feature allows you to set different filter on each
121probe and gives you what arguments will be shown in trace buffer. If an event
122name is specified right after 'p:' or 'r:' in kprobe_events, it adds an event
123under tracing/events/kprobes/<EVENT>, at the directory you can see 'id',
124'enable', 'format', 'filter' and 'trigger'.
125
126enable:
127  You can enable/disable the probe by writing 1 or 0 on it.
128
129format:
130  This shows the format of this probe event.
131
132filter:
133  You can write filtering rules of this event.
134
135id:
136  This shows the id of this probe event.
137
138trigger:
139  This allows to install trigger commands which are executed when the event is
140  hit (for details, see Documentation/trace/events.rst, section 6).
141
142Event Profiling
143---------------
144You can check the total number of probe hits and probe miss-hits via
145/sys/kernel/debug/tracing/kprobe_profile.
146The first column is event name, the second is the number of probe hits,
147the third is the number of probe miss-hits.
148
149Kernel Boot Parameter
150---------------------
151You can add and enable new kprobe events when booting up the kernel by
152"kprobe_event=" parameter. The parameter accepts a semicolon-delimited
153kprobe events, which format is similar to the kprobe_events.
154The difference is that the probe definition parameters are comma-delimited
155instead of space. For example, adding myprobe event on do_sys_open like below
156
157  p:myprobe do_sys_open dfd=%ax filename=%dx flags=%cx mode=+4($stack)
158
159should be below for kernel boot parameter (just replace spaces with comma)
160
161  p:myprobe,do_sys_open,dfd=%ax,filename=%dx,flags=%cx,mode=+4($stack)
162
163
164Usage examples
165--------------
166To add a probe as a new event, write a new definition to kprobe_events
167as below::
168
169  echo 'p:myprobe do_sys_open dfd=%ax filename=%dx flags=%cx mode=+4($stack)' > /sys/kernel/debug/tracing/kprobe_events
170
171This sets a kprobe on the top of do_sys_open() function with recording
1721st to 4th arguments as "myprobe" event. Note, which register/stack entry is
173assigned to each function argument depends on arch-specific ABI. If you unsure
174the ABI, please try to use probe subcommand of perf-tools (you can find it
175under tools/perf/).
176As this example shows, users can choose more familiar names for each arguments.
177::
178
179  echo 'r:myretprobe do_sys_open $retval' >> /sys/kernel/debug/tracing/kprobe_events
180
181This sets a kretprobe on the return point of do_sys_open() function with
182recording return value as "myretprobe" event.
183You can see the format of these events via
184/sys/kernel/debug/tracing/events/kprobes/<EVENT>/format.
185::
186
187  cat /sys/kernel/debug/tracing/events/kprobes/myprobe/format
188  name: myprobe
189  ID: 780
190  format:
191          field:unsigned short common_type;       offset:0;       size:2; signed:0;
192          field:unsigned char common_flags;       offset:2;       size:1; signed:0;
193          field:unsigned char common_preempt_count;       offset:3; size:1;signed:0;
194          field:int common_pid;   offset:4;       size:4; signed:1;
195
196          field:unsigned long __probe_ip; offset:12;      size:4; signed:0;
197          field:int __probe_nargs;        offset:16;      size:4; signed:1;
198          field:unsigned long dfd;        offset:20;      size:4; signed:0;
199          field:unsigned long filename;   offset:24;      size:4; signed:0;
200          field:unsigned long flags;      offset:28;      size:4; signed:0;
201          field:unsigned long mode;       offset:32;      size:4; signed:0;
202
203
204  print fmt: "(%lx) dfd=%lx filename=%lx flags=%lx mode=%lx", REC->__probe_ip,
205  REC->dfd, REC->filename, REC->flags, REC->mode
206
207You can see that the event has 4 arguments as in the expressions you specified.
208::
209
210  echo > /sys/kernel/debug/tracing/kprobe_events
211
212This clears all probe points.
213
214Or,
215::
216
217  echo -:myprobe >> kprobe_events
218
219This clears probe points selectively.
220
221Right after definition, each event is disabled by default. For tracing these
222events, you need to enable it.
223::
224
225  echo 1 > /sys/kernel/debug/tracing/events/kprobes/myprobe/enable
226  echo 1 > /sys/kernel/debug/tracing/events/kprobes/myretprobe/enable
227
228Use the following command to start tracing in an interval.
229::
230
231    # echo 1 > tracing_on
232    Open something...
233    # echo 0 > tracing_on
234
235And you can see the traced information via /sys/kernel/debug/tracing/trace.
236::
237
238  cat /sys/kernel/debug/tracing/trace
239  # tracer: nop
240  #
241  #           TASK-PID    CPU#    TIMESTAMP  FUNCTION
242  #              | |       |          |         |
243             <...>-1447  [001] 1038282.286875: myprobe: (do_sys_open+0x0/0xd6) dfd=3 filename=7fffd1ec4440 flags=8000 mode=0
244             <...>-1447  [001] 1038282.286878: myretprobe: (sys_openat+0xc/0xe <- do_sys_open) $retval=fffffffffffffffe
245             <...>-1447  [001] 1038282.286885: myprobe: (do_sys_open+0x0/0xd6) dfd=ffffff9c filename=40413c flags=8000 mode=1b6
246             <...>-1447  [001] 1038282.286915: myretprobe: (sys_open+0x1b/0x1d <- do_sys_open) $retval=3
247             <...>-1447  [001] 1038282.286969: myprobe: (do_sys_open+0x0/0xd6) dfd=ffffff9c filename=4041c6 flags=98800 mode=10
248             <...>-1447  [001] 1038282.286976: myretprobe: (sys_open+0x1b/0x1d <- do_sys_open) $retval=3
249
250
251Each line shows when the kernel hits an event, and <- SYMBOL means kernel
252returns from SYMBOL(e.g. "sys_open+0x1b/0x1d <- do_sys_open" means kernel
253returns from do_sys_open to sys_open+0x1b).
254
255