• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1#!/usr/bin/python
2# @lint-avoid-python-3-compatibility-imports
3#
4# biosnoop  Trace block device I/O and print details including issuing PID.
5#       For Linux, uses BCC, eBPF.
6#
7# This uses in-kernel eBPF maps to cache process details (PID and comm) by I/O
8# request, as well as a starting timestamp for calculating I/O latency.
9#
10# Copyright (c) 2015 Brendan Gregg.
11# Licensed under the Apache License, Version 2.0 (the "License")
12#
13# 16-Sep-2015   Brendan Gregg   Created this.
14
15from __future__ import print_function
16from bcc import BPF
17
18# load BPF program
19b = BPF(text="""
20#include <uapi/linux/ptrace.h>
21#include <linux/blkdev.h>
22
23struct val_t {
24    u32 pid;
25    char name[TASK_COMM_LEN];
26};
27
28BPF_HASH(start, struct request *);
29BPF_HASH(infobyreq, struct request *, struct val_t);
30
31// cache PID and comm by-req
32int trace_pid_start(struct pt_regs *ctx, struct request *req)
33{
34    struct val_t val = {};
35
36    if (bpf_get_current_comm(&val.name, sizeof(val.name)) == 0) {
37        val.pid = bpf_get_current_pid_tgid();
38        infobyreq.update(&req, &val);
39    }
40
41    return 0;
42}
43
44// time block I/O
45int trace_req_start(struct pt_regs *ctx, struct request *req)
46{
47    u64 ts;
48
49    ts = bpf_ktime_get_ns();
50    start.update(&req, &ts);
51
52    return 0;
53}
54
55// output
56int trace_req_completion(struct pt_regs *ctx, struct request *req)
57{
58    u64 *tsp, delta;
59    u32 *pidp = 0;
60    struct val_t *valp;
61
62    // fetch timestamp and calculate delta
63    tsp = start.lookup(&req);
64    if (tsp == 0) {
65        // missed tracing issue
66        return 0;
67    }
68    delta = bpf_ktime_get_ns() - *tsp;
69
70    //
71    // Fetch and output issuing pid and comm.
72    // As bpf_trace_prink() is limited to a maximum of 1 string and 2
73    // integers, we'll use more than one to output the data.
74    //
75    valp = infobyreq.lookup(&req);
76    if (valp == 0) {
77        bpf_trace_printk("0 0 ? %d\\n", req->__data_len);
78    } else {
79        bpf_trace_printk("0 %d %s %d\\n", valp->pid, valp->name,
80            req->__data_len);
81    }
82
83    // output remaining details
84    if (req->cmd_flags & REQ_WRITE) {
85        bpf_trace_printk("1 W %s %d %d ?\\n", req->rq_disk->disk_name,
86            req->__sector, delta / 1000);
87    } else {
88        bpf_trace_printk("1 R %s %d %d ?\\n", req->rq_disk->disk_name,
89            req->__sector, delta / 1000);
90    }
91
92    start.delete(&req);
93    infobyreq.delete(&req);
94
95    return 0;
96}
97""", debug=0)
98b.attach_kprobe(event="blk_account_io_start", fn_name="trace_pid_start")
99b.attach_kprobe(event="blk_start_request", fn_name="trace_req_start")
100b.attach_kprobe(event="blk_mq_start_request", fn_name="trace_req_start")
101b.attach_kprobe(event="blk_account_io_completion",
102    fn_name="trace_req_completion")
103
104# header
105print("%-14s %-14s %-6s %-7s %-2s %-9s %-7s %7s" % ("TIME(s)", "COMM", "PID",
106    "DISK", "T", "SECTOR", "BYTES", "LAT(ms)"))
107
108start_ts = 0
109
110# format output
111while 1:
112    (task, pid, cpu, flags, ts, msg) = b.trace_fields()
113    args = msg.split(" ")
114
115    if start_ts == 0:
116        start_ts = ts
117
118    if args[0] == "0":
119        (real_pid, real_comm, bytes_s) = (args[1], args[2], args[3])
120        continue
121    else:
122        (type_s, disk_s, sector_s, us_s) = (args[1], args[2], args[3],
123            args[4])
124
125    ms = float(int(us_s, 10)) / 1000
126
127    print("%-14.9f %-14.14s %-6s %-7s %-2s %-9s %-7s %7.2f" % (
128        ts - start_ts, real_comm, real_pid, disk_s, type_s, sector_s,
129        bytes_s, ms))
130