Commit eff67bfb authored by Brendan Gregg's avatar Brendan Gregg

Merge pull request #404 from mcaleavya/master

Migrated killsnoop to bpf_perf_event
parents ca6ac5ad 6262f5d9
......@@ -11,6 +11,10 @@ is sending signals.
This works by tracing the kernel sys_kill() function using dynamic tracing, and
will need updating to match any changes to this function.
This makes use of a Linux 4.5 feature (bpf_perf_event_output());
for kernels older than 4.5, see the version under tools/old,
which uses an older mechanism.
Since this uses BPF, only the root user can use this tool.
.SH REQUIREMENTS
CONFIG_BPF and bcc.
......
......@@ -10,10 +10,12 @@
# Licensed under the Apache License, Version 2.0 (the "License")
#
# 20-Sep-2015 Brendan Gregg Created this.
# 19-Feb-2016 Allan McAleavy migrated to BPF_PERF_OUTPUT
from __future__ import print_function
from bcc import BPF
import argparse
import ctypes as ct
# arguments
examples = """examples:
......@@ -38,35 +40,69 @@ debug = 0
# define BPF program
bpf_text = """
#include <uapi/linux/ptrace.h>
#include <linux/sched.h>
struct val_t {
u64 pid;
u64 ts;
int sig;
int tpid;
char comm[TASK_COMM_LEN];
};
struct data_t {
u64 pid;
u64 tpid;
int sig;
int ret;
u64 ts;
u64 delta;
char comm[TASK_COMM_LEN];
};
BPF_HASH(args_pid, u32, int);
BPF_HASH(args_sig, u32, int);
BPF_HASH(infotmp, u32, struct val_t);
BPF_PERF_OUTPUT(events);
int kprobe__sys_kill(struct pt_regs *ctx, int tpid, int sig)
{
struct val_t val = {};
u32 pid = bpf_get_current_pid_tgid();
FILTER
args_pid.update(&pid, &tpid);
args_sig.update(&pid, &sig);
if (bpf_get_current_comm(&val.comm, sizeof(val.comm)) == 0) {
val.pid = bpf_get_current_pid_tgid();
val.ts = bpf_ktime_get_ns();
val.tpid = tpid;
val.sig = sig;
infotmp.update(&pid, &val);
}
return 0;
};
int kretprobe__sys_kill(struct pt_regs *ctx)
{
int *tpidp, *sigp, ret = ctx->ax;
struct data_t data = {};
struct val_t *valp;
u32 pid = bpf_get_current_pid_tgid();
u64 tsp = bpf_ktime_get_ns();
tpidp = args_pid.lookup(&pid);
sigp = args_sig.lookup(&pid);
if (tpidp == 0 || sigp == 0) {
return 0; // missed entry
valp = infotmp.lookup(&pid);
if (valp == 0) {
// missed entry
return 0;
}
bpf_trace_printk("%d %d %d\\n", *tpidp, *sigp, ret);
args_pid.delete(&pid);
args_sig.delete(&pid);
bpf_probe_read(&data.comm, sizeof(data.comm), valp->comm);
data.pid = pid;
data.delta = tsp - valp->ts;
data.ts = tsp / 1000;
data.tpid = valp->tpid;
data.ret = ctx->ax;
data.sig = valp->sig;
events.perf_submit(ctx, &data, sizeof(data));
infotmp.delete(&pid);
return 0;
}
......@@ -82,25 +118,57 @@ if debug:
# initialize BPF
b = BPF(text=bpf_text)
TASK_COMM_LEN = 16 # linux/sched.h
class Data(ct.Structure):
_fields_ = [
("pid", ct.c_ulonglong),
("tpid", ct.c_ulonglong),
("sig", ct.c_int),
("ret", ct.c_int),
("ts", ct.c_ulonglong),
("delta", ct.c_ulonglong),
("comm", ct.c_char * TASK_COMM_LEN)
]
start_ts = 0
prev_ts = 0
delta = 0
# header
if args.timestamp:
print("%-14s" % ("TIME(s)"), end="")
print("%-6s %-16s %-4s %-6s %s" % ("PID", "COMM", "SIG", "TPID", "RESULT"))
start_ts = 0
# process event
def print_event(cpu, data, size):
event = ct.cast(data, ct.POINTER(Data)).contents
global start_ts
global prev_ts
global delta
# format output
while 1:
(task, pid, cpu, flags, ts, msg) = b.trace_fields()
(tpid_s, sig_s, ret_s) = msg.split(" ")
if start_ts == 0:
prev_ts = start_ts
if start_ts == 1:
delta = float(delta) + (event.ts - prev_ts)
ret = int(ret_s)
if (args.failed and (ret >= 0)):
continue
if (args.failed and (event.ret >= 0)):
start_ts = 1
prev_ts = event.ts
return
# print columns
if args.timestamp:
if start_ts == 0:
start_ts = ts
print("%-14.9f" % (ts - start_ts), end="")
print("%-6d %-16s %-4s %-6s %s" % (pid, task, sig_s, tpid_s, ret_s))
print("%-14.9f" % (delta / 1000000), end="")
print("%-6d %-16s %-4d %-6d %d" % (event.pid, event.comm, event.sig,
event.tpid, event.ret))
prev_ts = event.ts
start_ts = 1
# loop with callback to print_event
b["events"].open_perf_buffer(print_event)
while 1:
b.kprobe_poll()
#!/usr/bin/python
# @lint-avoid-python-3-compatibility-imports
#
# killsnoop Trace signals issued by the kill() syscall.
# For Linux, uses BCC, eBPF. Embedded C.
#
# USAGE: killsnoop [-h] [-t] [-x] [-p PID]
#
# Copyright (c) 2015 Brendan Gregg.
# Licensed under the Apache License, Version 2.0 (the "License")
#
# 20-Sep-2015 Brendan Gregg Created this.
from __future__ import print_function
from bcc import BPF
import argparse
# arguments
examples = """examples:
./killsnoop # trace all kill() signals
./killsnoop -t # include timestamps
./killsnoop -x # only show failed kills
./killsnoop -p 181 # only trace PID 181
"""
parser = argparse.ArgumentParser(
description="Trace signals issued by the kill() syscall",
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 opens")
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>
BPF_HASH(args_pid, u32, int);
BPF_HASH(args_sig, u32, int);
int kprobe__sys_kill(struct pt_regs *ctx, int tpid, int sig)
{
u32 pid = bpf_get_current_pid_tgid();
FILTER
args_pid.update(&pid, &tpid);
args_sig.update(&pid, &sig);
return 0;
};
int kretprobe__sys_kill(struct pt_regs *ctx)
{
int *tpidp, *sigp, ret = ctx->ax;
u32 pid = bpf_get_current_pid_tgid();
tpidp = args_pid.lookup(&pid);
sigp = args_sig.lookup(&pid);
if (tpidp == 0 || sigp == 0) {
return 0; // missed entry
}
bpf_trace_printk("%d %d %d\\n", *tpidp, *sigp, ret);
args_pid.delete(&pid);
args_sig.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)
# header
if args.timestamp:
print("%-14s" % ("TIME(s)"), end="")
print("%-6s %-16s %-4s %-6s %s" % ("PID", "COMM", "SIG", "TPID", "RESULT"))
start_ts = 0
# format output
while 1:
(task, pid, cpu, flags, ts, msg) = b.trace_fields()
(tpid_s, sig_s, ret_s) = msg.split(" ")
ret = int(ret_s)
if (args.failed and (ret >= 0)):
continue
# print columns
if args.timestamp:
if start_ts == 0:
start_ts = ts
print("%-14.9f" % (ts - start_ts), end="")
print("%-6d %-16s %-4s %-6s %s" % (pid, task, sig_s, tpid_s, ret_s))
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment