在实际研发过程中，可能会遇到过这样的问题：测试通过 valgrind 验证当前代码存在变量未初始化的问题，但仅通过 valgrind 测试报告，研发无法确认具体的应用场景。本文将通过 valgrind 检测并定位程序中变量未初始化的问题，并给出用户具体的操作场景。

为保证程序的健壮性和可靠性，程序在完成基本的功能测试后，一般来说还要进行内存使用相关的测试。Valgrind 是一个强大的 C 和 C++ 程序调试和性能分析工具。它可以检测内存泄漏、未初始化的内存使用以及其他各种错误。本文将以 MariaDB 为例，通过 valgrind 检测并定位程序中变量未初始化的问题。

1. 为说明问题，在 MariaDB 的 mysql_execute_command 函数中添加如下代码：

   int unitialized_value;

   if (SQLCOM_SHOW_TABLES != thd_sql_command(thd)) {
   unitialized_value = 0;
   }

   if (unitialized_value) {
   sql_print_information("Unitialized value");
   }

2. 使用 valgrind 运行 mariadbd 程序

   valgrind --tool=memcheck --trace-children=yes --track-origins=yes <br /> --leak-check=full --show-leak-kinds=all <br /> --log-file=./valgrind.log --error-limit=no <br /> /home/dbdev/mariadb/bin/mariadbd --defaults-file=./etc/mariadb.cnf &

3. 登录 MariaDB server

   /home/dbdev/mariadb/bin/mysql -S /home/dbdev/mariadb/data/mysqld.sock

4. 执行如下 SQL 语句

   MariaDB [(none)]>
   MariaDB [(none)]> show databases;
   +--------------------+
   | Database |
   +--------------------+
   | information_schema |
   | mysql |
   | performance_schema |
   | sakila |
   | sys |
   | test |
   | world_x |
   +--------------------+
   7 rows in set (0.486 sec)

   MariaDB [(none)]> use test
   Reading table information for completion of table and column names
   You can turn off this feature to get a quicker startup with -A

   Database changed
   MariaDB [test]> show tables;
   +----------------+
   | Tables_in_test |
   +----------------+
   | v1 |
   +----------------+
   1 row in set (0.096 sec)

5. 查看 valgrind.log 日志

   ==5432== Memcheck, a memory error detector
   ==5432== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al.
   ==5432== Using Valgrind-3.18.1 and LibVEX; rerun with -h for copyright info
   ==5432== Command: /home/dbdev/mariadb/bin/mariadbd --defaults-file=./etc/mariadb.cnf
   ==5432== Parent PID: 1055
   ==5432==
   ==5432== Warning: set address range perms: large range [0x124746000, 0x164746000) (defined)
   ==5432== Warning: set address range perms: large range [0x164746000, 0x1a4746000) (defined)
   ==5432== Warning: set address range perms: large range [0x124746000, 0x144063000) (defined)
   ==5432== Warning: set address range perms: large range [0x124746000, 0x144063000) (defined)
   ==5432== Thread 20:
   ==5432== Conditional jump or move depends on uninitialised value(s)
   ==5432== at 0xAA9991: mysql_execute_command(THD, bool) (sql_parse.cc:3482) ==5432== by 0xAB52D3: mysql_parse(THD, char, unsigned int, Parser_state) (sql_parse.cc:7779)
   ==5432== by 0xAA4E5A: dispatch_command(enum_server_command, THD, char, unsigned int, bool) (sql_parse.cc:1892)
   ==5432== by 0xAA3C74: do_command(THD, bool) (sql_parse.cc:1405) ==5432== by 0xC9381C: do_handle_one_connection(CONNECT, bool) (sql_connect.cc:1416)
   ==5432== by 0xC9359C: handle_one_connection (sql_connect.cc:1318)
   ==5432== by 0x11D4C72: pfs_spawn_thread (pfs.cc:2201)
   ==5432== by 0x5330B42: start_thread (pthread_create.c:442)
   ==5432== by 0x53C1BB3: clone (clone.S:100)
   ==5432== Uninitialised value was created by a stack allocation
   ==5432== at 0xAA977B: mysql_execute_command(THD*, bool) (sql_parse.cc:3444)
   ==5432==

存在的问题

测试生成了 valgrind 测试报告 valgrind.log，测试报告内容如下：

==5432== Memcheck, a memory error detector
==5432== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al.
==5432== Using Valgrind-3.18.1 and LibVEX; rerun with -h for copyright info
==5432== Command: /home/dbdev/mariadb/bin/mariadbd --defaults-file=./etc/mariadb.cnf
==5432== Parent PID: 1055
==5432==
==5432== Warning: set address range perms: large range [0x124746000, 0x164746000) (defined)
==5432== Warning: set address range perms: large range [0x164746000, 0x1a4746000) (defined)
==5432== Warning: set address range perms: large range [0x124746000, 0x144063000) (defined)
==5432== Warning: set address range perms: large range [0x124746000, 0x144063000) (defined)
==5432== Thread 20:
==5432== Conditional jump or move depends on uninitialised value(s)
==5432==    at 0xAA9991: mysql_execute_command(THD*, bool) (sql_parse.cc:3482)
==5432==    by 0xAB52D3: mysql_parse(THD*, char*, unsigned int, Parser_state*) (sql_parse.cc:7779)
==5432==    by 0xAA4E5A: dispatch_command(enum_server_command, THD*, char*, unsigned int, bool) (sql_parse.cc:1892)
==5432==    by 0xAA3C74: do_command(THD*, bool) (sql_parse.cc:1405)
==5432==    by 0xC9381C: do_handle_one_connection(CONNECT*, bool) (sql_connect.cc:1416)
==5432==    by 0xC9359C: handle_one_connection (sql_connect.cc:1318)
==5432==    by 0x11D4C72: pfs_spawn_thread (pfs.cc:2201)
==5432==    by 0x5330B42: start_thread (pthread_create.c:442)
==5432==    by 0x53C1BB3: clone (clone.S:100)
==5432==  Uninitialised value was created by a stack allocation
==5432==    at 0xAA977B: mysql_execute_command(THD*, bool) (sql_parse.cc:3444)
==5432==

研发进能够根据该堆栈信息查询到未初始化值的设置地方及使用地方。但是不清楚具体是哪种操作（具体 SQL 语句）导致的，因为具体应用中未初始化 unitialized_value 的地方可能存在多处。

valgrind 配合 gdb 调试 MariaDB

1. 开启终端 T1，使用 gdb 模式运行 valgrind，即指定 --vgdb=yes --vgdb-error=0 选项运行 valgrind

   valgrind --tool=memcheck --trace-children=yes --track-origins=yes <br /> --vgdb=yes --vgdb-error=0 --leak-check=full <br /> --show-leak-kinds=all --log-file=./valgrind.log <br /> --error-limit=no /home/dbdev/mariadb/bin/mariadbd <br /> --defaults-file=./etc/mariadb.cnf &

2. 开启终端 T2，使用 gdb 调试运行 mariadbd

   gdb /home/dbdev/mariadb/bin/mariadbd
   GNU gdb (Ubuntu 13.1-2ubuntu2) 13.1
   Copyright (C) 2023 Free Software Foundation, Inc.
   License GPLv3+: GNU GPL version 3 or later http://gnu.org/licenses/gpl.html
   This is free software: you are free to change and redistribute it.
   There is NO WARRANTY, to the extent permitted by law.
   Type "show copying" and "show warranty" for details.
   This GDB was configured as "x86_64-linux-gnu".
   Type "show configuration" for configuration details.
   For bug reporting instructions, please see:
   https://www.gnu.org/software/gdb/bugs/.
   Find the GDB manual and other documentation resources online at:
   http://www.gnu.org/software/gdb/documentation/.

   For help, type "help".
   Type "apropos word" to search for commands related to "word"…
   Reading symbols from bin/mariadbd…
   (gdb)
   (gdb)

3. 终端 T1 打开 valgrind.log

   cat valgrind.log
   ==7516== Memcheck, a memory error detector
   ==7516== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al.
   ==7516== Using Valgrind-3.18.1 and LibVEX; rerun with -h for copyright info
   ==7516== Command: /home/dbdev/mariadb/bin/mariadbd --defaults-file=./etc/mariadb.cnf
   ==7516== Parent PID: 1055
   ==7516==
   ==7516== (action at startup) vgdb me …
   ==7516==
   ==7516== TO DEBUG THIS PROCESS USING GDB: start GDB like this
   ==7516== /path/to/gdb /home/dbdev/mariadb/bin/mariadbd
   ==7516== and then give GDB the following command
   ==7516== target remote | /usr/bin/vgdb --pid=7516
   ==7516== --pid is optional if only one valgrind process is running
   ==7516==

4. 将 "target remote | /usr/bin/vgdb --pid=7516" 输入到 T2 执行

   (gdb) target remote | /usr/bin/vgdb --pid=7516
   Remote debugging using | /usr/bin/vgdb --pid=7516
   relaying data between gdb and process 7516
   warning: remote target does not support file transfer, attempting to access files from local filesystem.
   Reading symbols from /lib64/ld-linux-x86-64.so.2…
   Reading symbols from /usr/lib/debug/.build-id/bf/789e981fb305f374f3975a4b248eb4cd207ce6.debug…
   0x00000000040202f0 in _start () from /lib64/ld-linux-x86-64.so.2

5. T2 输入 c 命令后程序自动运行，后续用户连接到 MariaDB 实例执行 SQL 语句时，valgrind 在检测到代码可能存在问题时会自动激活 valgrind 内部设置的断点，此时就可以观察程序内部的变量是否正确

   (gdb) target remote | /usr/bin/vgdb --pid=7516
   Remote debugging using | /usr/bin/vgdb --pid=7516
   relaying data between gdb and process 7516
   warning: remote target does not support file transfer, attempting to access files from local filesystem.
   Reading symbols from /lib64/ld-linux-x86-64.so.2…
   Reading symbols from /usr/lib/debug/.build-id/bf/789e981fb305f374f3975a4b248eb4cd207ce6.debug…
   0x00000000040202f0 in _start () from /lib64/ld-linux-x86-64.so.2
   (gdb) c
   Continuing.

6. T1 登录 MariaDB 实例，执行如下 SQL 语句，该操作会激活 T2 的 gdb 断点

   MariaDB [(none)]> use test

7. T2 查看堆栈信息，输出当前的 SQL 语句信息，通过这种方式，我们可以清楚的知道具体是什么操作导致了未初始化内存的使用

   (gdb) bt
   #0 0x0000000000aa99a0 in mysql_execute_command (thd=0x1aa290f88, is_called_from_prepared_stmt=false) at /home/dbdev/workspace/server/sql/sql_parse.cc:3482
   #1 0x0000000000ab52e4 in mysql_parse (thd=0x1aa290f88, rawbuf=0x1aa30bf20 "show tables", length=11, parser_state=0x1aa51b1f0) at /home/dbdev/workspace/server/sql/sql_parse.cc:7779
   #2 0x0000000000aa4e5b in dispatch_command (command=COM_QUERY, thd=0x1aa290f88, packet=0x1aa29b479 "show tables", packet_length=11, blocking=true) at /home/dbdev/workspace/server/sql/sql_parse.cc:1892
   #3 0x0000000000aa3c75 in do_command (thd=0x1aa290f88, blocking=true) at /home/dbdev/workspace/server/sql/sql_parse.cc:1405
   #4 0x0000000000c9382d in do_handle_one_connection (connect=0x1aa28f0c8, put_in_cache=true) at /home/dbdev/workspace/server/sql/sql_connect.cc:1416
   #5 0x0000000000c935ad in handle_one_connection (arg=0x1aa28f0c8) at /home/dbdev/workspace/server/sql/sql_connect.cc:1318
   #6 0x00000000011d4c83 in pfs_spawn_thread (arg=0x1aa290898) at /home/dbdev/workspace/server/storage/perfschema/pfs.cc:2201
   #7 0x0000000005330b43 in start_thread (arg=) at ./nptl/pthread_create.c:442
   #8 0x00000000053c1bb4 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:100
   (gdb) p thd->query_string
   $1 = {string = {str = 0x1aa30bf20 "show tables", length = 11}, cs = 0x24cbda0 }