Repository navigation
Expand file tree
/
Copy pathstatsnoop.py
More file actions
executable file
·174 lines (146 loc) · 4.45 KB
/
Copy pathstatsnoop.py
File metadata and controls
executable file
·174 lines (146 loc) · 4.45 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
#!/usr/bin/python
# @lint-avoid-python-3-compatibility-imports
#
# statsnoop Trace stat() syscalls.
# For Linux, uses BCC, eBPF. Embedded C.
#
# USAGE: statsnoop [-h] [-t] [-x] [-p PID]
#
# Copyright 2016 Netflix, Inc.
# Licensed under the Apache License, Version 2.0 (the "License")
#
# 08-Feb-2016 Brendan Gregg Created this.
# 17-Feb-2016 Allan McAleavy updated for BPF_PERF_OUTPUT
from __future__ import print_function
from bcc import BPF
import argparse
import ctypes as ct
# arguments
examples = """examples:
./statsnoop # trace all stat() syscalls
./statsnoop -t # include timestamps
./statsnoop -x # only show failed stats
./statsnoop -p 181 # only trace PID 181
"""
parser = argparse.ArgumentParser(
description="Trace stat() syscalls",
formatter_class=argparse.RawDescriptionHelpFormatter,
epilog=examples)
parser.add_argument("-t", "--timestamp", action="store_true",
help="include timestamp on output")
parser.add_argument("-x", "--failed", action="store_true",
help="only show failed stats")
parser.add_argument("-p", "--pid",
help="trace this PID only")
args = parser.parse_args()
debug = 0
# define BPF program
bpf_text = """
#include <uapi/linux/ptrace.h>
#include <uapi/linux/limits.h>
#include <linux/sched.h>
struct val_t {
const char *fname;
};
struct data_t {
u32 pid;
u64 ts_ns;
int ret;
char comm[TASK_COMM_LEN];
char fname[NAME_MAX];
};
BPF_HASH(args_filename, u32, const char *);
BPF_HASH(infotmp, u32, struct val_t);
BPF_PERF_OUTPUT(events);
int trace_entry(struct pt_regs *ctx, const char __user *filename)
{
struct val_t val = {};
u32 pid = bpf_get_current_pid_tgid();
FILTER
val.fname = filename;
infotmp.update(&pid, &val);
return 0;
};
int trace_return(struct pt_regs *ctx)
{
u32 pid = bpf_get_current_pid_tgid();
struct val_t *valp;
valp = infotmp.lookup(&pid);
if (valp == 0) {
// missed entry
return 0;
}
struct data_t data = {.pid = pid};
bpf_probe_read(&data.fname, sizeof(data.fname), (void *)valp->fname);
bpf_get_current_comm(&data.comm, sizeof(data.comm));
data.ts_ns = bpf_ktime_get_ns();
data.ret = PT_REGS_RC(ctx);
events.perf_submit(ctx, &data, sizeof(data));
infotmp.delete(&pid);
args_filename.delete(&pid);
return 0;
}
"""
if args.pid:
bpf_text = bpf_text.replace('FILTER',
'if (pid != %s) { return 0; }' % args.pid)
else:
bpf_text = bpf_text.replace('FILTER', '')
if debug:
print(bpf_text)
# initialize BPF
b = BPF(text=bpf_text)
# for POSIX compliance, all architectures implement these
# system calls but the name of the actual entry point may
# be different for which we must check if the entry points
# actually exist before attaching the probes
if BPF.ksymname("sys_stat") != -1:
b.attach_kprobe(event="sys_stat", fn_name="trace_entry")
b.attach_kretprobe(event="sys_stat", fn_name="trace_return")
if BPF.ksymname("sys_statfs") != -1:
b.attach_kprobe(event="sys_statfs", fn_name="trace_entry")
b.attach_kretprobe(event="sys_statfs", fn_name="trace_return")
if BPF.ksymname("sys_newstat") != -1:
b.attach_kprobe(event="sys_newstat", fn_name="trace_entry")
b.attach_kretprobe(event="sys_newstat", fn_name="trace_return")
TASK_COMM_LEN = 16 # linux/sched.h
NAME_MAX = 255 # linux/limits.h
class Data(ct.Structure):
_fields_ = [
("pid", ct.c_ulonglong),
("ts_ns", ct.c_ulonglong),
("ret", ct.c_int),
("comm", ct.c_char * TASK_COMM_LEN),
("fname", ct.c_char * NAME_MAX)
]
start_ts = 0
prev_ts = 0
delta = 0
# header
if args.timestamp:
print("%-14s" % ("TIME(s)"), end="")
print("%-6s %-16s %4s %3s %s" % ("PID", "COMM", "FD", "ERR", "PATH"))
# process event
def print_event(cpu, data, size):
event = ct.cast(data, ct.POINTER(Data)).contents
global start_ts
global prev_ts
global delta
global cont
# split return value into FD and errno columns
if event.ret >= 0:
fd_s = event.ret
err = 0
else:
fd_s = -1
err = - event.ret
if start_ts == 0:
start_ts = event.ts_ns
if args.timestamp:
print("%-14.9f" % (float(event.ts_ns - start_ts) / 1000000000), end="")
print("%-6d %-16s %4d %3d %s" % (event.pid, event.comm.decode(),
fd_s, err, event.fname.decode()))
# loop with callback to print_event
b["events"].open_perf_buffer(print_event, page_cnt=64)
while 1:
b.kprobe_poll()