使用C语言getline()函数时出现段错误的原因排查
问题分析:getline判断条件从
!= -1改为>=0触发段错误 核心原因:无符号类型的隐式转换
linelen的类型是size_t,这是一个无符号整数类型,而getline的返回值是ssize_t(有符号整数类型)。当getline读取到文件末尾返回-1时,这个有符号的-1会被隐式转换为无符号的size_t类型,结果就是无符号整数的最大值18446744073709551615(也就是你看到的极大值)。
对于无符号整数来说,所有的值都是非负的,所以判断条件linelen >= 0永远为真,循环不会终止。此时用这个极大的长度调用fwrite,会超出buf的实际分配范围,读取/写入不属于程序的内存区域,最终触发段错误。
而用!= -1判断时,-1同样会被转换为无符号的最大值,此时判断linelen != 这个最大值,当getline返回EOF时,条件不成立,循环正常终止。
代码对比验证
正常运行的代码
#include<stdlib.h> int main(int argc , char *argv[]) { FILE *fp = fopen(argv[argc-1],"r"); if(fp == NULL) { printf("error : can't open file"); exit(1); } char *buf = NULL; size_t linelen = 0; size_t linecapp = 0; while((linelen = getline(&buf,&linecapp,fp)) !=-1) { fwrite(buf,linelen,1,stdout); } fclose(fp); return 0; }
触发段错误的代码
#include<stdio.h> #include<stdlib.h> int main(int argc , char *argv[]) { FILE *fp = fopen(argv[argc-1],"r"); if(fp == NULL) { printf("error : can't open file"); exit(1); } char *buf = NULL; size_t linelen = 0; size_t linecapp = 0; while((linelen = getline(&buf,&linecapp,fp)) >= 0) { fwrite(buf,linelen,1,stdout); } fclose(fp); return 0; }
测试输出佐证
当程序运行到文件末尾时,linelen被赋值为转换后的无符号最大值,后续fwrite尝试写入这个长度的内容,导致输出乱码并最终崩溃:
116 hello world this is a test file to bes used for the purpose of learning basics of unix sys calls and c programming 77 this is just an extra line to elongate the file followed by more extra lines 55 for example this is a line used in elongating the file 22 this is another line 141 this line contain special characters like œ∑´®†¥¨ˆøπåß∂ƒ©˙∆˚¬¬¬…≈ç√∫˜µ≤≥÷¡™£¢∞§¶•ªº– 52 this line contain some numbers 1 2 3 5 7 4 6 8 9 10 18446744073709551615 his line contain some numbers 1 2 3 5 7 4 6 8 9 10 ?¥¨ˆøπåß∂ƒ©˙∆˚¬¬¬…≈ç√∫˜µ≤≥÷¡™£¢∞§¶•ªº– Xjm??ͫ?????@|m? ?5?r?@?@:?#??Zm@?` `@?`?`?Zm???#? ??#????#?????#?@?`X???\/var/folders/l5/h45_tftn3w18ws8vshrpxlsr0000gn/T//private/tmp/com.apple.launchd.pIOsn5DWl2/Listeners0?``SSH_AUTH_SOCK??`??????????????????`??`P ?b??,?,?8Zm@??Xm? ??`(?`?`HOME?Xm?D<?B`38?`P`PATH?Xm?81??` ?`?`TMPDIR?Xm?83?`??!???c?8???W&???ǀ?G???w+???C???????`??`@``?```?Z??`ԥ`???`???`???`?@`?a??\_?x?`?`?`?Z????^??\_?x?`????`?pm?? ?? m ??um?? ?x ?h m? ?@ m?? ?? m?` ?? zsh: segmentation fault ./getlinefunc ../../sampletextfile.txt
解决方案
- 保持原判断条件
!= -1:这是符合getline文档规范的写法,利用无符号转换后的最大值来判断EOF。 - 使用
ssize_t存储返回值:将linelen的类型改为ssize_t,这样可以正确处理有符号的返回值,此时用>=0判断也能正常终止循环:
ssize_t linelen = 0; while((linelen = getline(&buf,&linecapp,fp)) >= 0) { fwrite(buf, (size_t)linelen, 1, stdout); }
注意此时fwrite的第二个参数需要强制转换为size_t,因为fwrite要求该参数是无符号类型。
内容的提问来源于stack exchange,提问作者gunslinger
相关产品推荐
相关产品推荐

