求助：fgets 读文件死锁！

mailsafe

问题：难道fgets函数也死锁，系统上线后出现过几次这类问题。 暂未解决，只能kill掉读取进程。

环境：HP-UX 安腾 64。
数据文件：放在共享存储上，不在本机硬盘上。
描述：两个并发进程进入文件拆分函数，读取同一待拆分文件，然后就没有下文，僵在里面。
错误跟踪日志如下：

--进程A，GDB堆栈信息
xxxx@/home/db/xxxx> gdb xxxxFileLoad 20166
HP gdb 6.2 for HP Itanium (32 or 64 bit) and target HP-UX 11iv2 and 11iv3.
Copyright 1986 - 2011 Free Software Foundation, Inc.
Hewlett-Packard Wildebeest 6.2 (based on GDB) is covered by the
GNU General Public License. Type "show copying" to see the conditions to
change it and/or distribute copies. Type "show warranty" for warranty/support.
..
Attaching to program: /home/db/xxxx/xxxxbatch/bin/xxxxFileLoad, process 20166

warning: The shared libraries were not privately mapped; setting a
breakpoint in a shared library will not work until you rerun the program;
stepping over longjmp calls will not work as expected.
Please set the kernel variable "shlib_debug_enable" to 1 to enable the shared library debugging

warning: Load module /home/db/oracle/product/10.2.0/lib/libnnz10.so has been stripped.  
Debugging information is not available.

0xc00000000054a490:0 in _read_sys+0x30 () from /usr/lib/hpux64/libc.so.1
(gdb) bt
#0  0xc00000000054a490:0 in _read_sys+0x30 () from /usr/lib/hpux64/libc.so.1
#1  0xc00000000055ed80:0 in read+0xe0 () from /usr/lib/hpux64/libc.so.1
#2  0xc000000000527a20:0 in _filbuf+0x190 () from /usr/lib/hpux64/libc.so.1
#3  0xc00000000053a0d0:0 in __fgets_unlocked+0x1f0 () from /usr/lib/hpux64/libc.so.1
#4  0xc00000000053a3e0:0 in fgets+0x140 () from /usr/lib/hpux64/libc.so.1
#5  0x4000000000030320:0 in g_GetFSegment () at p_file.c:301  文件拆分函数：这一行就是 fgets()，其读取缓存大小为一次读取 400K。
#6  0x4000000000006130:0 in InsxxxxFile () at xxxxFileLoad.pc:1060
#7  0x4000000000005380:0 in main () at xxxxFileLoad.pc:107
(gdb) quit
The program is running.  Quit anyway (and detach it)? (y or n) y
Detaching from program: /home/db/xxxx/xxxxbatch/bin/xxxxFileLoad, process 20166


--进程B，GDB堆栈信息
xxxx@/home/db/xxxx> gdb xxxxFileLoad 20167
HP gdb 6.2 for HP Itanium (32 or 64 bit) and target HP-UX 11iv2 and 11iv3.
Copyright 1986 - 2011 Free Software Foundation, Inc.
Hewlett-Packard Wildebeest 6.2 (based on GDB) is covered by the
GNU General Public License. Type "show copying" to see the conditions to
change it and/or distribute copies. Type "show warranty" for warranty/support.
..
Attaching to program: /home/db/xxxx/xxxxbatch/bin/xxxxFileLoad, process 20167

warning: The shared libraries were not privately mapped; setting a
breakpoint in a shared library will not work until you rerun the program;
stepping over longjmp calls will not work as expected.
Please set the kernel variable "shlib_debug_enable" to 1 to enable the shared library debugging

warning: Load module /home/db/oracle/product/10.2.0/lib/libnnz10.so has been stripped.  
Debugging information is not available.

0xc00000000054a490:0 in _read_sys+0x30 () from /usr/lib/hpux64/libc.so.1
(gdb) bt
#0  0xc00000000054a490:0 in _read_sys+0x30 () from /usr/lib/hpux64/libc.so.1
#1  0xc00000000055ed80:0 in read+0xe0 () from /usr/lib/hpux64/libc.so.1
#2  0xc000000000527a20:0 in _filbuf+0x190 () from /usr/lib/hpux64/libc.so.1
#3  0xc00000000053a0d0:0 in __fgets_unlocked+0x1f0 () from /usr/lib/hpux64/libc.so.1
#4  0xc00000000053a3e0:0 in fgets+0x140 () from /usr/lib/hpux64/libc.so.1
#5  0x4000000000030320:0 in g_GetFSegment () at p_file.c:301 文件拆分函数：这一行就是 fgets()，其读取缓存大小为一次读取 400K。
#6  0x4000000000006130:0 in InsxxxxFile () at xxxxFileLoad.pc:1060
#7  0x4000000000005380:0 in main () at xxxxFileLoad.pc:107
(gdb) quit
The program is running.  Quit anyway (and detach it)? (y or n) y
Detaching from program: /home/db/xxxx/xxxxbatch/bin/xxxxFileLoad, process 20167

请教一下可能是什么原因导致此类问题随机产生。

csumck

man手冊里说明了可以用*_unlocked()來做非阻塞操作。

liangzhishao

因为fgets的读操作（read系统调用）是在网络设备上读取文件导致永远阻塞（就是死锁）。

APUE 14.2中描述

1. 低速系统调用是可能会使进程永远阻塞的一类系统调用，它们包括下列调用：
   
2.   * 如果某些文件类型（例如管道、终端设备和[color=Red]网络设备[/color]）的数据并不存在，则读操作可能会使调用者永远阻塞。
   
3.   * 。。。。

复制代码

你的系统使用的共享存储，它不是本地文件系统，所以它属于低速设备。

另外，APUE 中还说

1. 我们也曾说过，虽然读、写磁盘文件会使调用者在短暂时间内阻塞，但并不能将和磁盘I/O有关的系统调用视为“低速”。

复制代码

这一点我不能给出进一步解释，为何磁盘文件不属于“低速”，它是否会永远阻塞，我没有找到明确的答案。
望得到指点。。。

mailsafe

意思是：网络设备可能在某一时点失效，从而使读取发生死锁，可以采用 fgets_unlocked 来规避？