| 1 | Kprobe-based Event Tracing |
| 2 | ========================== |
| 3 | |
| 4 | Documentation is written by Masami Hiramatsu |
| 5 | |
| 6 | |
| 7 | Overview |
| 8 | -------- |
| 9 | These events are similar to tracepoint based events. Instead of Tracepoint, |
| 10 | this is based on kprobes (kprobe and kretprobe). So it can probe wherever |
| 11 | kprobes can probe (this means, all functions body except for __kprobes |
| 12 | functions). Unlike the Tracepoint based event, this can be added and removed |
| 13 | dynamically, on the fly. |
| 14 | |
| 15 | To enable this feature, build your kernel with CONFIG_KPROBE_EVENT=y. |
| 16 | |
| 17 | Similar to the events tracer, this doesn't need to be activated via |
| 18 | current_tracer. Instead of that, add probe points via |
| 19 | /sys/kernel/debug/tracing/kprobe_events, and enable it via |
| 20 | /sys/kernel/debug/tracing/events/kprobes/<EVENT>/enabled. |
| 21 | |
| 22 | |
| 23 | Synopsis of kprobe_events |
| 24 | ------------------------- |
| 25 | p[:[GRP/]EVENT] [MOD:]SYM[+offs]|MEMADDR [FETCHARGS] : Set a probe |
| 26 | r[:[GRP/]EVENT] [MOD:]SYM[+0] [FETCHARGS] : Set a return probe |
| 27 | -:[GRP/]EVENT : Clear a probe |
| 28 | |
| 29 | GRP : Group name. If omitted, use "kprobes" for it. |
| 30 | EVENT : Event name. If omitted, the event name is generated |
| 31 | based on SYM+offs or MEMADDR. |
| 32 | MOD : Module name which has given SYM. |
| 33 | SYM[+offs] : Symbol+offset where the probe is inserted. |
| 34 | MEMADDR : Address where the probe is inserted. |
| 35 | |
| 36 | FETCHARGS : Arguments. Each probe can have up to 128 args. |
| 37 | %REG : Fetch register REG |
| 38 | @ADDR : Fetch memory at ADDR (ADDR should be in kernel) |
| 39 | @SYM[+|-offs] : Fetch memory at SYM +|- offs (SYM should be a data symbol) |
| 40 | $stackN : Fetch Nth entry of stack (N >= 0) |
| 41 | $stack : Fetch stack address. |
| 42 | $retval : Fetch return value.(*) |
| 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), "string" and bitfield |
| 47 | are supported. |
| 48 | |
| 49 | (*) only for return probe. |
| 50 | (**) this is useful for fetching a field of data structures. |
| 51 | |
| 52 | Types |
| 53 | ----- |
| 54 | Several types are supported for fetch-args. Kprobe tracer will access memory |
| 55 | by given type. Prefix 's' and 'u' means those types are signed and unsigned |
| 56 | respectively. Traced arguments are shown in decimal (signed) or hex (unsigned). |
| 57 | String type is a special type, which fetches a "null-terminated" string from |
| 58 | kernel space. This means it will fail and store NULL if the string container |
| 59 | has been paged out. |
| 60 | Bitfield is another special type, which takes 3 parameters, bit-width, bit- |
| 61 | offset, and container-size (usually 32). The syntax is; |
| 62 | |
| 63 | b<bit-width>@<bit-offset>/<container-size> |
| 64 | |
| 65 | |
| 66 | Per-Probe Event Filtering |
| 67 | ------------------------- |
| 68 | Per-probe event filtering feature allows you to set different filter on each |
| 69 | probe and gives you what arguments will be shown in trace buffer. If an event |
| 70 | name is specified right after 'p:' or 'r:' in kprobe_events, it adds an event |
| 71 | under tracing/events/kprobes/<EVENT>, at the directory you can see 'id', |
| 72 | 'enabled', 'format' and 'filter'. |
| 73 | |
| 74 | enabled: |
| 75 | You can enable/disable the probe by writing 1 or 0 on it. |
| 76 | |
| 77 | format: |
| 78 | This shows the format of this probe event. |
| 79 | |
| 80 | filter: |
| 81 | You can write filtering rules of this event. |
| 82 | |
| 83 | id: |
| 84 | This shows the id of this probe event. |
| 85 | |
| 86 | |
| 87 | Event Profiling |
| 88 | --------------- |
| 89 | You can check the total number of probe hits and probe miss-hits via |
| 90 | /sys/kernel/debug/tracing/kprobe_profile. |
| 91 | The first column is event name, the second is the number of probe hits, |
| 92 | the third is the number of probe miss-hits. |
| 93 | |
| 94 | |
| 95 | Usage examples |
| 96 | -------------- |
| 97 | To add a probe as a new event, write a new definition to kprobe_events |
| 98 | as below. |
| 99 | |
| 100 | echo 'p:myprobe do_sys_open dfd=%ax filename=%dx flags=%cx mode=+4($stack)' > /sys/kernel/debug/tracing/kprobe_events |
| 101 | |
| 102 | This sets a kprobe on the top of do_sys_open() function with recording |
| 103 | 1st to 4th arguments as "myprobe" event. Note, which register/stack entry is |
| 104 | assigned to each function argument depends on arch-specific ABI. If you unsure |
| 105 | the ABI, please try to use probe subcommand of perf-tools (you can find it |
| 106 | under tools/perf/). |
| 107 | As this example shows, users can choose more familiar names for each arguments. |
| 108 | |
| 109 | echo 'r:myretprobe do_sys_open $retval' >> /sys/kernel/debug/tracing/kprobe_events |
| 110 | |
| 111 | This sets a kretprobe on the return point of do_sys_open() function with |
| 112 | recording return value as "myretprobe" event. |
| 113 | You can see the format of these events via |
| 114 | /sys/kernel/debug/tracing/events/kprobes/<EVENT>/format. |
| 115 | |
| 116 | cat /sys/kernel/debug/tracing/events/kprobes/myprobe/format |
| 117 | name: myprobe |
| 118 | ID: 780 |
| 119 | format: |
| 120 | field:unsigned short common_type; offset:0; size:2; signed:0; |
| 121 | field:unsigned char common_flags; offset:2; size:1; signed:0; |
| 122 | field:unsigned char common_preempt_count; offset:3; size:1;signed:0; |
| 123 | field:int common_pid; offset:4; size:4; signed:1; |
| 124 | |
| 125 | field:unsigned long __probe_ip; offset:12; size:4; signed:0; |
| 126 | field:int __probe_nargs; offset:16; size:4; signed:1; |
| 127 | field:unsigned long dfd; offset:20; size:4; signed:0; |
| 128 | field:unsigned long filename; offset:24; size:4; signed:0; |
| 129 | field:unsigned long flags; offset:28; size:4; signed:0; |
| 130 | field:unsigned long mode; offset:32; size:4; signed:0; |
| 131 | |
| 132 | |
| 133 | print fmt: "(%lx) dfd=%lx filename=%lx flags=%lx mode=%lx", REC->__probe_ip, |
| 134 | REC->dfd, REC->filename, REC->flags, REC->mode |
| 135 | |
| 136 | You can see that the event has 4 arguments as in the expressions you specified. |
| 137 | |
| 138 | echo > /sys/kernel/debug/tracing/kprobe_events |
| 139 | |
| 140 | This clears all probe points. |
| 141 | |
| 142 | Or, |
| 143 | |
| 144 | echo -:myprobe >> kprobe_events |
| 145 | |
| 146 | This clears probe points selectively. |
| 147 | |
| 148 | Right after definition, each event is disabled by default. For tracing these |
| 149 | events, you need to enable it. |
| 150 | |
| 151 | echo 1 > /sys/kernel/debug/tracing/events/kprobes/myprobe/enable |
| 152 | echo 1 > /sys/kernel/debug/tracing/events/kprobes/myretprobe/enable |
| 153 | |
| 154 | And you can see the traced information via /sys/kernel/debug/tracing/trace. |
| 155 | |
| 156 | cat /sys/kernel/debug/tracing/trace |
| 157 | # tracer: nop |
| 158 | # |
| 159 | # TASK-PID CPU# TIMESTAMP FUNCTION |
| 160 | # | | | | | |
| 161 | <...>-1447 [001] 1038282.286875: myprobe: (do_sys_open+0x0/0xd6) dfd=3 filename=7fffd1ec4440 flags=8000 mode=0 |
| 162 | <...>-1447 [001] 1038282.286878: myretprobe: (sys_openat+0xc/0xe <- do_sys_open) $retval=fffffffffffffffe |
| 163 | <...>-1447 [001] 1038282.286885: myprobe: (do_sys_open+0x0/0xd6) dfd=ffffff9c filename=40413c flags=8000 mode=1b6 |
| 164 | <...>-1447 [001] 1038282.286915: myretprobe: (sys_open+0x1b/0x1d <- do_sys_open) $retval=3 |
| 165 | <...>-1447 [001] 1038282.286969: myprobe: (do_sys_open+0x0/0xd6) dfd=ffffff9c filename=4041c6 flags=98800 mode=10 |
| 166 | <...>-1447 [001] 1038282.286976: myretprobe: (sys_open+0x1b/0x1d <- do_sys_open) $retval=3 |
| 167 | |
| 168 | |
| 169 | Each line shows when the kernel hits an event, and <- SYMBOL means kernel |
| 170 | returns from SYMBOL(e.g. "sys_open+0x1b/0x1d <- do_sys_open" means kernel |
| 171 | returns from do_sys_open to sys_open+0x1b). |
| 172 | |