bin/bpftrace-tools/dist: Import bpftrace tools v0.12.1

Skipped bashreadline as we link to external library, whereas script
instruments the internal implementation.
Skipped xfsdist as we ship without XFS support in kernel or userspace.
Skipped CMake file as it's not needed.

--HG--
branch : bpftrace-tools

Author: sevan

bin/bpftrace-tools/dist/biolatency.bt

@@ -0,0 +1,34 @@
+#!/usr/bin/env bpftrace
+/*
+ * biolatency.bt	Block I/O latency as a histogram.
+ *			For Linux, uses bpftrace, eBPF.
+ *
+ * This is a bpftrace version of the bcc tool of the same name.
+ *
+ * Copyright 2018 Netflix, Inc.
+ * Licensed under the Apache License, Version 2.0 (the "License")
+ *
+ * 13-Sep-2018	Brendan Gregg	Created this.
+ */
+
+BEGIN
+{
+	printf("Tracing block device I/O... Hit Ctrl-C to end.\n");
+}
+
+kprobe:blk_account_io_start
+{
+	@start[arg0] = nsecs;
+}
+
+kprobe:blk_account_io_done
+/@start[arg0]/
+{
+	@usecs = hist((nsecs - @start[arg0]) / 1000);
+	delete(@start[arg0]);
+}
+
+END
+{
+	clear(@start);
+}

bin/bpftrace-tools/dist/biolatency_example.txt

@@ -0,0 +1,32 @@
+Demonstrations of biolatency, the Linux BPF/bpftrace version.
+
+
+This traces block I/O, and shows latency as a power-of-2 histogram. For example:
+
+# ./biolatency.bt
+Attaching 3 probes...
+Tracing block device I/O... Hit Ctrl-C to end.
+^C
+
+@usecs:
+[256, 512)             2 |                                                    |
+[512, 1K)             10 |@                                                   |
+[1K, 2K)             426 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+[2K, 4K)             230 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@                        |
+[4K, 8K)               9 |@                                                   |
+[8K, 16K)            128 |@@@@@@@@@@@@@@@                                     |
+[16K, 32K)            68 |@@@@@@@@                                            |
+[32K, 64K)             0 |                                                    |
+[64K, 128K)            0 |                                                    |
+[128K, 256K)          10 |@                                                   |
+
+While tracing, this shows that 426 block I/O had a latency of between 1K and 2K
+usecs (1024 and 2048 microseconds), which is between 1 and 2 milliseconds.
+There are also two modes visible, one between 1 and 2 milliseconds, and another
+between 8 and 16 milliseconds: this sounds like cache hits and cache misses.
+There were also 10 I/O with latency 128 to 256 ms: outliers. Other tools and
+instrumentation, like biosnoop.bt, can shed more light on those outliers.
+
+
+There is another version of this tool in bcc: https://github.com/iovisor/bcc
+The bcc version provides options to customize the output.

bin/bpftrace-tools/dist/biosnoop.bt

@@ -0,0 +1,48 @@
+#!/usr/bin/env bpftrace
+#include <linux/blkdev.h>
+/*
+ * biosnoop.bt   Block I/O tracing tool, showing per I/O latency.
+ *               For Linux, uses bpftrace, eBPF.
+ *
+ * TODO: switch to block tracepoints. Add offset and size columns.
+ *
+ * This is a bpftrace version of the bcc tool of the same name.
+ *
+ * 15-Nov-2017	Brendan Gregg	Created this.
+ */
+
+BEGIN
+{
+	printf("%-12s %-7s %-16s %-6s %7s\n", "TIME(ms)", "DISK", "COMM", "PID", "LAT(ms)");
+}
+
+kprobe:blk_account_io_start
+{
+	@start[arg0] = nsecs;
+	@iopid[arg0] = pid;
+	@iocomm[arg0] = comm;
+	@disk[arg0] = ((struct request *)arg0)->rq_disk->disk_name;
+}
+
+kprobe:blk_account_io_done
+/@start[arg0] != 0 && @iopid[arg0] != 0 && @iocomm[arg0] != ""/
+
+{
+	$now = nsecs;
+	printf("%-12u %-7s %-16s %-6d %7d\n",
+	    elapsed / 1e6, @disk[arg0], @iocomm[arg0], @iopid[arg0],
+	    ($now - @start[arg0]) / 1e6);
+
+	delete(@start[arg0]);
+	delete(@iopid[arg0]);
+	delete(@iocomm[arg0]);
+	delete(@disk[arg0]);
+}
+
+END
+{
+	clear(@start);
+	clear(@iopid);
+	clear(@iocomm);
+	clear(@disk);
+}

bin/bpftrace-tools/dist/biosnoop_example.txt

@@ -0,0 +1,47 @@
+Demonstrations of biosnoop, the Linux BPF/bpftrace version.
+
+
+This traces block I/O, and shows the issuing process (at least, the process
+that was on-CPU at the time of queue insert) and the latency of the I/O:
+
+# ./biosnoop.bt
+Attaching 4 probes...
+TIME(ms)     DISK             COMM             PID    LAT(ms)
+611          nvme0n1          bash             4179        10
+611          nvme0n1          cksum            4179         0
+627          nvme0n1          cksum            4179        15
+641          nvme0n1          cksum            4179        13
+644          nvme0n1          cksum            4179         3
+658          nvme0n1          cksum            4179        13
+673          nvme0n1          cksum            4179        14
+686          nvme0n1          cksum            4179        13
+701          nvme0n1          cksum            4179        14
+710          nvme0n1          cksum            4179         8
+717          nvme0n1          cksum            4179         6
+728          nvme0n1          cksum            4179        10
+735          nvme0n1          cksum            4179         6
+751          nvme0n1          cksum            4179        10
+758          nvme0n1          cksum            4179        17
+783          nvme0n1          cksum            4179        12
+796          nvme0n1          cksum            4179        25
+802          nvme0n1          cksum            4179        32
+[...]
+
+This output shows the cksum process was issuing block I/O, which were
+completing with around 12 milliseconds of latency. Each block I/O event is
+printed out, with a completion time as the first column, measured from
+program start.
+
+
+An example of some background flushing:
+
+# ./biosnoop.bt
+Attaching 4 probes...
+TIME(ms)     DISK             COMM             PID    LAT(ms)
+2966         nvme0n1          jbd2/nvme0n1-8   615          0
+2967         nvme0n1          jbd2/nvme0n1-8   615          0
+[...]
+
+
+There is another version of this tool in bcc: https://github.com/iovisor/bcc
+The bcc version provides more fields.

bin/bpftrace-tools/dist/biostacks.bt

@@ -0,0 +1,39 @@
+#!/usr/bin/env bpftrace
+/*
+ * biostacks - Show disk I/O latency with initialization stacks.
+ *
+ * See BPF Performance Tools, Chapter 9, for an explanation of this tool.
+ *
+ * Copyright (c) 2019 Brendan Gregg.
+ * Licensed under the Apache License, Version 2.0 (the "License").
+ * This was originally created for the BPF Performance Tools book
+ * published by Addison Wesley. ISBN-13: 9780136554820
+ * When copying or porting, include this comment.
+ *
+ * 19-Mar-2019  Brendan Gregg   Created this.
+ */
+
+BEGIN
+{
+	printf("Tracing block I/O with init stacks. Hit Ctrl-C to end.\n");
+}
+
+kprobe:blk_account_io_start
+{
+	@reqstack[arg0] = kstack;
+	@reqts[arg0] = nsecs;
+}
+
+kprobe:blk_start_request,
+kprobe:blk_mq_start_request
+/@reqts[arg0]/
+{
+	@usecs[@reqstack[arg0]] = hist(nsecs - @reqts[arg0]);
+	delete(@reqstack[arg0]);
+	delete(@reqts[arg0]);
+}
+
+END
+{
+	clear(@reqstack); clear(@reqts);
+}

bin/bpftrace-tools/dist/biostacks_example.txt

@@ -0,0 +1,63 @@
+Demonstrations of biostacks, the Linux BCC/eBPF version.
+
+
+This tool shows block I/O latency as a histogram, with the kernel stack trace
+that initiated the I/O. This can help explain disk I/O that is not directly
+requested by applications (eg, metadata reads on writes, resilvering, etc).
+For example:
+
+# ./biostacks.bt 
+Attaching 5 probes...
+Tracing block I/O with init stacks. Hit Ctrl-C to end.
+^C
+
+@usecs[
+    blk_account_io_start+1
+    blk_mq_make_request+1102
+    generic_make_request+292
+    submit_bio+115
+    _xfs_buf_ioapply+798
+    xfs_buf_submit+101
+    xlog_bdstrat+43
+    xlog_sync+705
+    xlog_state_release_iclog+108
+    _xfs_log_force+542
+    xfs_log_force+44
+    xfsaild+428
+    kthread+289
+    ret_from_fork+53
+]: 
+[64K, 128K)            1 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+
+[...]
+
+@usecs[
+    blk_account_io_start+1
+    blk_mq_make_request+707
+    generic_make_request+292
+    submit_bio+115
+    xfs_add_to_ioend+455
+    xfs_do_writepage+758
+    write_cache_pages+524
+    xfs_vm_writepages+190
+    do_writepages+75
+    __writeback_single_inode+69
+    writeback_sb_inodes+481
+    __writeback_inodes_wb+103
+    wb_writeback+625
+    wb_workfn+384
+    process_one_work+478
+    worker_thread+50
+    kthread+289
+    ret_from_fork+53
+]: 
+[8K, 16K)            560 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+[16K, 32K)           218 |@@@@@@@@@@@@@@@@@@@@                                |
+[32K, 64K)            26 |@@                                                  |
+[64K, 128K)            2 |                                                    |
+[128K, 256K)          53 |@@@@                                                |
+[256K, 512K)          60 |@@@@@                                               |
+
+This output shows the most frequent stack was XFS writeback, with latencies
+between 8 and 512 microseconds. The other stack included here shows an XFS
+log sync.

bin/bpftrace-tools/dist/bitesize.bt

@@ -0,0 +1,29 @@
+#!/usr/bin/env bpftrace
+/*
+ * bitesize	Show disk I/O size as a histogram.
+ *		For Linux, uses bpftrace and eBPF.
+ *
+ * USAGE: bitesize.bt
+ *
+ * This is a bpftrace version of the bcc tool of the same name.
+ *
+ * Copyright 2018 Netflix, Inc.
+ * Licensed under the Apache License, Version 2.0 (the "License")
+ *
+ * 07-Sep-2018	Brendan Gregg	Created this.
+ */
+
+BEGIN
+{
+	printf("Tracing block device I/O... Hit Ctrl-C to end.\n");
+}
+
+tracepoint:block:block_rq_issue
+{
+	@[args->comm] = hist(args->bytes);
+}
+
+END
+{
+	printf("\nI/O size (bytes) histograms by process name:");
+}

bin/bpftrace-tools/dist/bitesize_example.txt

@@ -0,0 +1,59 @@
+Demonstrations of bitesize, the Linux bpftrace/eBPF version.
+
+
+This traces disk I/O via the block I/O interface, and prints a summary of I/O
+sizes as histograms for each process name. For example:
+
+# ./bitesize.bt
+Attaching 3 probes...
+Tracing block device I/O... Hit Ctrl-C to end.
+^C
+I/O size (bytes) histograms by process name:
+
+@[cleanup]:
+[4K, 8K)               2 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+
+@[postdrop]:
+[4K, 8K)               2 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+
+@[jps]:
+[4K, 8K)               1 |@@@@@@@@@@@@@@@@@@@@@@@@@@                          |
+[8K, 16K)              0 |                                                    |
+[16K, 32K)             0 |                                                    |
+[32K, 64K)             2 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+
+@[kworker/2:1H]:
+[0]                    3 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+[1]                    0 |                                                    |
+[2, 4)                 0 |                                                    |
+[4, 8)                 0 |                                                    |
+[8, 16)                0 |                                                    |
+[16, 32)               0 |                                                    |
+[32, 64)               0 |                                                    |
+[64, 128)              0 |                                                    |
+[128, 256)             0 |                                                    |
+[256, 512)             0 |                                                    |
+[512, 1K)              0 |                                                    |
+[1K, 2K)               0 |                                                    |
+[2K, 4K)               0 |                                                    |
+[4K, 8K)               0 |                                                    |
+[8K, 16K)              0 |                                                    |
+[16K, 32K)             0 |                                                    |
+[32K, 64K)             0 |                                                    |
+[64K, 128K)            1 |@@@@@@@@@@@@@@@@@                                   |
+
+@[jbd2/nvme0n1-8]:
+[4K, 8K)               3 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+[8K, 16K)              0 |                                                    |
+[16K, 32K)             0 |                                                    |
+[32K, 64K)             2 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@                  |
+[64K, 128K)            1 |@@@@@@@@@@@@@@@@@                                   |
+
+@[dd]:
+[16K, 32K)           921 |@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@|
+
+The most active process while tracing was "dd", which issues 921 I/O between
+16 Kbytes and 32 Kbytes in size.
+
+
+There is another version of this tool in bcc: https://github.com/iovisor/bcc