一次系统调用时间过长追踪完整教程案例

Ftrace简介

Ftrace是Linux进行代码级实践分析最有效的工具之一，比如我们进行一个系统调用，出来的时间过长，我们想知道时间花哪里去了，利用Ftrace就可以追踪到一级级的时间分布。

Ftrace案例

写一个proc模块，包含一个proc的读和写的入口。test_proc_show()故意调用了一个kill_time()的函数，而kill_time()的函数，又调用了mdelay(2)和kill_moretime()的函数，该函数体内调用mdelay(2)。

kill_time()的函数和kill_moretime()函数前面都加了noinline以避免被编译器inline优化掉。

#include <linux/module.h>

#include <linux/kernel.h>

#include <linux/init.h>

#include <linux/version.h>

#include <linux/proc_fs.h>

#include <linux/seq_file.h>

#include <linux/delay.h>

#include <linux/uaccess.h>



static unsigned int variable;

static struct proc_dir_entry *test_dir, *test_entry;



static noinline void kill_moretime(void)

{

mdelay(2);

}



static noinline void kill_time(void)

{

mdelay(2);

kill_moretime();

}



static int test_proc_show(struct seq_file *seq, void *v)

{

unsigned int *ptr_var = seq->private;

kill_time();

seq_printf(seq, "%u\n", *ptr_var);

return 0;

}



static ssize_t test_proc_write(struct file *file, const char __user *buffer,

size_t count, loff_t *ppos)

{

struct seq_file *seq = file->private_data;

unsigned int *ptr_var = seq->private;

int err;

char *kbuffer;



        if (!buffer || count > PAGE_SIZE - 1)

                return -EINVAL;



kbuffer = (char *)__get_free_page(GFP_KERNEL);

if (!kbuffer)

return -ENOMEM;



err = -EFAULT;

if (copy_from_user(kbuffer, buffer, count))

goto out;

kbuffer[count] = '\0';



*ptr_var = simple_strtoul(kbuffer, NULL, 10);

return count;



out:

free_page((unsigned long)buffer);

return err;

}



static int test_proc_open(struct inode *inode, struct file *file)

{

return single_open(file, test_proc_show, PDE_DATA(inode));

}



static const struct file_operations test_proc_fops =

{

.owner = THIS_MODULE,

.open = test_proc_open,

.read = seq_read,

.write = test_proc_write,

.llseek = seq_lseek,

.release = single_release,

};



static __init int test_proc_init(void)

{

test_dir = proc_mkdir("test_dir", NULL);

if (test_dir) {

test_entry = proc_create_data("test_rw",0666, test_dir, &test_proc_fops, &variable);

if (test_entry)

return 0;

}



return -ENOMEM;

}

module_init(test_proc_init);



static __exit void test_proc_cleanup(void)

{

remove_proc_entry("test_rw", test_dir);

remove_proc_entry("test_dir", NULL);

}

module_exit(test_proc_cleanup);



MODULE_AUTHOR("Barry Song <baohua@kernel.org>");

MODULE_DESCRIPTION("proc exmaple");

MODULE_LICENSE("GPL v2");

模块对应的Makefile如下：

KVERS = $(shell uname -r)



# Kernel modules

obj-m += proc.o



# Specify flags for the module compilation.

#EXTRA_CFLAGS=-g -O0



build: kernel_modules



kernel_modules:

make -C /lib/modules/$(KVERS)/build M=$(CURDIR) modules



clean:

make -C /lib/modules/$(KVERS)/build M=$(CURDIR) clean

编译并且加载：

$ make

baohua@baohua-perf:~/develop/training/debug/ftrace/proc$ 

$ sudo insmod proc.ko

[sudo] password for baohua:

之后/proc目录下/proc/test_dir/test_rw文件可被读写。

下面我们用Ftrace来跟踪test_proc_show()这个函数。

我们把启动ftrace的所有命令写到一个脚本function.sh里面：

#!/bin/bash



debugfs=/sys/kernel/debug

echo nop > $debugfs/tracing/current_tracer

echo 0 > $debugfs/tracing/tracing_on

echo $$ > $debugfs/tracing/set_ftrace_pid

echo function_graph > $debugfs/tracing/current_tracer

#replace test_proc_show by your function name

echo test_proc_show > $debugfs/tracing/set_graph_function

echo 1 > $debugfs/tracing/tracing_on

exec "$@"

然后用这个脚本去启动cat /proc/test_dir/test_rw，这样ftrace下面test_proc_show()函数就被trace了。

# ./function.sh cat /proc/test_dir/test_rw

0

读取trace的结果：

# cat /sys/kernel/debug/tracing/trace > 1

接着用vim打开这个文件1，发现这个文件有600多行：

Ftrace结果怎么读？

Ftrace结果怎么读？答案非常简单：如果是叶子函数，就直接在这个函数的前面显示它占用的时间，如果是非叶子，要等到 }的时候，再显示时间，如下图：

延迟比较大的部分，会有+、#等特殊标号：

 '$' - greater than 1 second
 '@' - greater than 100 milisecond
 '*' - greater than 10 milisecond
 '#' - greater than 1000 microsecond
 '!' - greater than 100 microsecond
 '+' - greater than 10 microsecond
 ' ' - less than or equal to 10 microsecond.

vim对Ftrace进行折叠

上面那个Ftrace文件太大了，大到看不清。我们可以用vim来折叠之，不过需要一个vim的特别配置，我把它存放在了我的~目录，名字叫.fungraph-vim：

" Enable folding for ftrace function_graph traces.

"

" To use, :source this file while viewing a function_graph trace, or use vim's

" -S option to load from the command-line together with a trace.  You can then

" use the usual vim fold commands, such as "za", to open and close nested

" functions.  While closed, a fold will show the total time taken for a call,

" as would normally appear on the line with the closing brace.  Folded

" functions will not include finish_task_switch(), so folding should remain

" relatively sane even through a context switch.

"

" Note that this will almost certainly only work well with a

" single-CPU trace (e.g. trace-cmd report --cpu 1).



function! FunctionGraphFoldExpr(lnum)

  let line = getline(a:lnum)

  if line[-1:] == '{'

    if line =~ 'finish_task_switch() {$'

      return '>1'

    endif

    return 'a1'

  elseif line[-1:] == '}'

    return 's1'

  else

    return '='

  endif

endfunction



function! FunctionGraphFoldText()

  let s = split(getline(v:foldstart), '|', 1)

  if getline(v:foldend+1) =~ 'finish_task_switch() {$'

    let s[2] = ' task switch  '

  else

    let e = split(getline(v:foldend), '|', 1)

    let s[2] = e[2]

  endif

  return join(s, '|')

endfunction



setlocal foldexpr=FunctionGraphFoldExpr(v:lnum)

setlocal foldtext=FunctionGraphFoldText()

setlocal foldcolumn=12

setlocal foldmethod=expr