uprobetracer.rst 7.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186
  1. =========================================
  2. Uprobe-tracer: Uprobe-based Event Tracing
  3. =========================================
  4. :Author: Srikar Dronamraju
  5. Overview
  6. --------
  7. Uprobe based trace events are similar to kprobe based trace events.
  8. To enable this feature, build your kernel with CONFIG_UPROBE_EVENTS=y.
  9. Similar to the kprobe-event tracer, this doesn't need to be activated via
  10. current_tracer. Instead of that, add probe points via
  11. /sys/kernel/debug/tracing/uprobe_events, and enable it via
  12. /sys/kernel/debug/tracing/events/uprobes/<EVENT>/enable.
  13. However unlike kprobe-event tracer, the uprobe event interface expects the
  14. user to calculate the offset of the probepoint in the object.
  15. You can also use /sys/kernel/debug/tracing/dynamic_events instead of
  16. uprobe_events. That interface will provide unified access to other
  17. dynamic events too.
  18. Synopsis of uprobe_tracer
  19. -------------------------
  20. ::
  21. p[:[GRP/]EVENT] PATH:OFFSET [FETCHARGS] : Set a uprobe
  22. r[:[GRP/]EVENT] PATH:OFFSET [FETCHARGS] : Set a return uprobe (uretprobe)
  23. p[:[GRP/]EVENT] PATH:OFFSET%return [FETCHARGS] : Set a return uprobe (uretprobe)
  24. -:[GRP/]EVENT : Clear uprobe or uretprobe event
  25. GRP : Group name. If omitted, "uprobes" is the default value.
  26. EVENT : Event name. If omitted, the event name is generated based
  27. on PATH+OFFSET.
  28. PATH : Path to an executable or a library.
  29. OFFSET : Offset where the probe is inserted.
  30. OFFSET%return : Offset where the return probe is inserted.
  31. FETCHARGS : Arguments. Each probe can have up to 128 args.
  32. %REG : Fetch register REG
  33. @ADDR : Fetch memory at ADDR (ADDR should be in userspace)
  34. @+OFFSET : Fetch memory at OFFSET (OFFSET from same file as PATH)
  35. $stackN : Fetch Nth entry of stack (N >= 0)
  36. $stack : Fetch stack address.
  37. $retval : Fetch return value.(\*1)
  38. $comm : Fetch current task comm.
  39. +|-[u]OFFS(FETCHARG) : Fetch memory at FETCHARG +|- OFFS address.(\*2)(\*3)
  40. \IMM : Store an immediate value to the argument.
  41. NAME=FETCHARG : Set NAME as the argument name of FETCHARG.
  42. FETCHARG:TYPE : Set TYPE as the type of FETCHARG. Currently, basic types
  43. (u8/u16/u32/u64/s8/s16/s32/s64), hexadecimal types
  44. (x8/x16/x32/x64), "string" and bitfield are supported.
  45. (\*1) only for return probe.
  46. (\*2) this is useful for fetching a field of data structures.
  47. (\*3) Unlike kprobe event, "u" prefix will just be ignored, becuse uprobe
  48. events can access only user-space memory.
  49. Types
  50. -----
  51. Several types are supported for fetch-args. Uprobe tracer will access memory
  52. by given type. Prefix 's' and 'u' means those types are signed and unsigned
  53. respectively. 'x' prefix implies it is unsigned. Traced arguments are shown
  54. in decimal ('s' and 'u') or hexadecimal ('x'). Without type casting, 'x32'
  55. or 'x64' is used depends on the architecture (e.g. x86-32 uses x32, and
  56. x86-64 uses x64).
  57. String type is a special type, which fetches a "null-terminated" string from
  58. user space.
  59. Bitfield is another special type, which takes 3 parameters, bit-width, bit-
  60. offset, and container-size (usually 32). The syntax is::
  61. b<bit-width>@<bit-offset>/<container-size>
  62. For $comm, the default type is "string"; any other type is invalid.
  63. Event Profiling
  64. ---------------
  65. You can check the total number of probe hits per event via
  66. /sys/kernel/debug/tracing/uprobe_profile. The first column is the filename,
  67. the second is the event name, the third is the number of probe hits.
  68. Usage examples
  69. --------------
  70. * Add a probe as a new uprobe event, write a new definition to uprobe_events
  71. as below (sets a uprobe at an offset of 0x4245c0 in the executable /bin/bash)::
  72. echo 'p /bin/bash:0x4245c0' > /sys/kernel/debug/tracing/uprobe_events
  73. * Add a probe as a new uretprobe event::
  74. echo 'r /bin/bash:0x4245c0' > /sys/kernel/debug/tracing/uprobe_events
  75. * Unset registered event::
  76. echo '-:p_bash_0x4245c0' >> /sys/kernel/debug/tracing/uprobe_events
  77. * Print out the events that are registered::
  78. cat /sys/kernel/debug/tracing/uprobe_events
  79. * Clear all events::
  80. echo > /sys/kernel/debug/tracing/uprobe_events
  81. Following example shows how to dump the instruction pointer and %ax register
  82. at the probed text address. Probe zfree function in /bin/zsh::
  83. # cd /sys/kernel/debug/tracing/
  84. # cat /proc/`pgrep zsh`/maps | grep /bin/zsh | grep r-xp
  85. 00400000-0048a000 r-xp 00000000 08:03 130904 /bin/zsh
  86. # objdump -T /bin/zsh | grep -w zfree
  87. 0000000000446420 g DF .text 0000000000000012 Base zfree
  88. 0x46420 is the offset of zfree in object /bin/zsh that is loaded at
  89. 0x00400000. Hence the command to uprobe would be::
  90. # echo 'p:zfree_entry /bin/zsh:0x46420 %ip %ax' > uprobe_events
  91. And the same for the uretprobe would be::
  92. # echo 'r:zfree_exit /bin/zsh:0x46420 %ip %ax' >> uprobe_events
  93. .. note:: User has to explicitly calculate the offset of the probe-point
  94. in the object.
  95. We can see the events that are registered by looking at the uprobe_events file.
  96. ::
  97. # cat uprobe_events
  98. p:uprobes/zfree_entry /bin/zsh:0x00046420 arg1=%ip arg2=%ax
  99. r:uprobes/zfree_exit /bin/zsh:0x00046420 arg1=%ip arg2=%ax
  100. Format of events can be seen by viewing the file events/uprobes/zfree_entry/format.
  101. ::
  102. # cat events/uprobes/zfree_entry/format
  103. name: zfree_entry
  104. ID: 922
  105. format:
  106. field:unsigned short common_type; offset:0; size:2; signed:0;
  107. field:unsigned char common_flags; offset:2; size:1; signed:0;
  108. field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
  109. field:int common_pid; offset:4; size:4; signed:1;
  110. field:int common_padding; offset:8; size:4; signed:1;
  111. field:unsigned long __probe_ip; offset:12; size:4; signed:0;
  112. field:u32 arg1; offset:16; size:4; signed:0;
  113. field:u32 arg2; offset:20; size:4; signed:0;
  114. print fmt: "(%lx) arg1=%lx arg2=%lx", REC->__probe_ip, REC->arg1, REC->arg2
  115. Right after definition, each event is disabled by default. For tracing these
  116. events, you need to enable it by::
  117. # echo 1 > events/uprobes/enable
  118. Lets start tracing, sleep for some time and stop tracing.
  119. ::
  120. # echo 1 > tracing_on
  121. # sleep 20
  122. # echo 0 > tracing_on
  123. Also, you can disable the event by::
  124. # echo 0 > events/uprobes/enable
  125. And you can see the traced information via /sys/kernel/debug/tracing/trace.
  126. ::
  127. # cat trace
  128. # tracer: nop
  129. #
  130. # TASK-PID CPU# TIMESTAMP FUNCTION
  131. # | | | | |
  132. zsh-24842 [006] 258544.995456: zfree_entry: (0x446420) arg1=446420 arg2=79
  133. zsh-24842 [007] 258545.000270: zfree_exit: (0x446540 <- 0x446420) arg1=446540 arg2=0
  134. zsh-24842 [002] 258545.043929: zfree_entry: (0x446420) arg1=446420 arg2=79
  135. zsh-24842 [004] 258547.046129: zfree_exit: (0x446540 <- 0x446420) arg1=446540 arg2=0
  136. Output shows us uprobe was triggered for a pid 24842 with ip being 0x446420
  137. and contents of ax register being 79. And uretprobe was triggered with ip at
  138. 0x446540 with counterpart function entry at 0x446420.