1=========================================
2Uprobe-tracer: Uprobe-based Event Tracing
3=========================================
4
5:Author: Srikar Dronamraju
6
7
8Overview
9--------
10Uprobe based trace events are similar to kprobe based trace events.
11To enable this feature, build your kernel with CONFIG_UPROBE_EVENTS=y.
12
13Similar to the kprobe-event tracer, this doesn't need to be activated via
14current_tracer. Instead of that, add probe points via
15/sys/kernel/debug/tracing/uprobe_events, and enable it via
16/sys/kernel/debug/tracing/events/uprobes/<EVENT>/enabled.
17
18However unlike kprobe-event tracer, the uprobe event interface expects the
19user to calculate the offset of the probepoint in the object.
20
21Synopsis of uprobe_tracer
22-------------------------
23::
24
25  p[:[GRP/]EVENT] PATH:OFFSET [FETCHARGS] : Set a uprobe
26  r[:[GRP/]EVENT] PATH:OFFSET [FETCHARGS] : Set a return uprobe (uretprobe)
27  -:[GRP/]EVENT                           : Clear uprobe or uretprobe event
28
29  GRP           : Group name. If omitted, "uprobes" is the default value.
30  EVENT         : Event name. If omitted, the event name is generated based
31                  on PATH+OFFSET.
32  PATH          : Path to an executable or a library.
33  OFFSET        : Offset where the probe is inserted.
34
35  FETCHARGS     : Arguments. Each probe can have up to 128 args.
36   %REG         : Fetch register REG
37   @ADDR	: Fetch memory at ADDR (ADDR should be in userspace)
38   @+OFFSET	: Fetch memory at OFFSET (OFFSET from same file as PATH)
39   $stackN	: Fetch Nth entry of stack (N >= 0)
40   $stack	: Fetch stack address.
41   $retval	: Fetch return value.(*)
42   $comm	: Fetch current task comm.
43   +|-offs(FETCHARG) : Fetch memory at FETCHARG +|- offs address.(**)
44   NAME=FETCHARG     : Set NAME as the argument name of FETCHARG.
45   FETCHARG:TYPE     : Set TYPE as the type of FETCHARG. Currently, basic types
46		       (u8/u16/u32/u64/s8/s16/s32/s64), hexadecimal types
47		       (x8/x16/x32/x64), "string" and bitfield are supported.
48
49  (*) only for return probe.
50  (**) this is useful for fetching a field of data structures.
51
52Types
53-----
54Several types are supported for fetch-args. Uprobe tracer will access memory
55by given type. Prefix 's' and 'u' means those types are signed and unsigned
56respectively. 'x' prefix implies it is unsigned. Traced arguments are shown
57in decimal ('s' and 'u') or hexadecimal ('x'). Without type casting, 'x32'
58or 'x64' is used depends on the architecture (e.g. x86-32 uses x32, and
59x86-64 uses x64).
60String type is a special type, which fetches a "null-terminated" string from
61user space.
62Bitfield is another special type, which takes 3 parameters, bit-width, bit-
63offset, and container-size (usually 32). The syntax is::
64
65 b<bit-width>@<bit-offset>/<container-size>
66
67For $comm, the default type is "string"; any other type is invalid.
68
69
70Event Profiling
71---------------
72You can check the total number of probe hits and probe miss-hits via
73/sys/kernel/debug/tracing/uprobe_profile.
74The first column is event name, the second is the number of probe hits,
75the third is the number of probe miss-hits.
76
77Usage examples
78--------------
79 * Add a probe as a new uprobe event, write a new definition to uprobe_events
80   as below (sets a uprobe at an offset of 0x4245c0 in the executable /bin/bash)::
81
82    echo 'p /bin/bash:0x4245c0' > /sys/kernel/debug/tracing/uprobe_events
83
84 * Add a probe as a new uretprobe event::
85
86    echo 'r /bin/bash:0x4245c0' > /sys/kernel/debug/tracing/uprobe_events
87
88 * Unset registered event::
89
90    echo '-:p_bash_0x4245c0' >> /sys/kernel/debug/tracing/uprobe_events
91
92 * Print out the events that are registered::
93
94    cat /sys/kernel/debug/tracing/uprobe_events
95
96 * Clear all events::
97
98    echo > /sys/kernel/debug/tracing/uprobe_events
99
100Following example shows how to dump the instruction pointer and %ax register
101at the probed text address. Probe zfree function in /bin/zsh::
102
103    # cd /sys/kernel/debug/tracing/
104    # cat /proc/`pgrep zsh`/maps | grep /bin/zsh | grep r-xp
105    00400000-0048a000 r-xp 00000000 08:03 130904 /bin/zsh
106    # objdump -T /bin/zsh | grep -w zfree
107    0000000000446420 g    DF .text  0000000000000012  Base        zfree
108
1090x46420 is the offset of zfree in object /bin/zsh that is loaded at
1100x00400000. Hence the command to uprobe would be::
111
112    # echo 'p:zfree_entry /bin/zsh:0x46420 %ip %ax' > uprobe_events
113
114And the same for the uretprobe would be::
115
116    # echo 'r:zfree_exit /bin/zsh:0x46420 %ip %ax' >> uprobe_events
117
118.. note:: User has to explicitly calculate the offset of the probe-point
119	in the object.
120
121We can see the events that are registered by looking at the uprobe_events file.
122::
123
124    # cat uprobe_events
125    p:uprobes/zfree_entry /bin/zsh:0x00046420 arg1=%ip arg2=%ax
126    r:uprobes/zfree_exit /bin/zsh:0x00046420 arg1=%ip arg2=%ax
127
128Format of events can be seen by viewing the file events/uprobes/zfree_entry/format.
129::
130
131    # cat events/uprobes/zfree_entry/format
132    name: zfree_entry
133    ID: 922
134    format:
135         field:unsigned short common_type;         offset:0;  size:2; signed:0;
136         field:unsigned char common_flags;         offset:2;  size:1; signed:0;
137         field:unsigned char common_preempt_count; offset:3;  size:1; signed:0;
138         field:int common_pid;                     offset:4;  size:4; signed:1;
139         field:int common_padding;                 offset:8;  size:4; signed:1;
140
141         field:unsigned long __probe_ip;           offset:12; size:4; signed:0;
142         field:u32 arg1;                           offset:16; size:4; signed:0;
143         field:u32 arg2;                           offset:20; size:4; signed:0;
144
145    print fmt: "(%lx) arg1=%lx arg2=%lx", REC->__probe_ip, REC->arg1, REC->arg2
146
147Right after definition, each event is disabled by default. For tracing these
148events, you need to enable it by::
149
150    # echo 1 > events/uprobes/enable
151
152Lets disable the event after sleeping for some time.
153::
154
155    # sleep 20
156    # echo 0 > events/uprobes/enable
157
158And you can see the traced information via /sys/kernel/debug/tracing/trace.
159::
160
161    # cat trace
162    # tracer: nop
163    #
164    #           TASK-PID    CPU#    TIMESTAMP  FUNCTION
165    #              | |       |          |         |
166                 zsh-24842 [006] 258544.995456: zfree_entry: (0x446420) arg1=446420 arg2=79
167                 zsh-24842 [007] 258545.000270: zfree_exit:  (0x446540 <- 0x446420) arg1=446540 arg2=0
168                 zsh-24842 [002] 258545.043929: zfree_entry: (0x446420) arg1=446420 arg2=79
169                 zsh-24842 [004] 258547.046129: zfree_exit:  (0x446540 <- 0x446420) arg1=446540 arg2=0
170
171Output shows us uprobe was triggered for a pid 24842 with ip being 0x446420
172and contents of ax register being 79. And uretprobe was triggered with ip at
1730x446540 with counterpart function entry at 0x446420.
174