学习建议：

• 弃用root用户
• 重构老代码
• 书要看, 习题代码要敲
• 资料
  • C: https://www.bilibili.com/video/BV18p4y167Md
  • APUE笔记: https://www.cnblogs.com/0xcafebabe/tag/APUE/
  • man手册: vim下用shift+k可跳转到man手册

标准IO

I/O: 落盘是一切实现的基础, 分为2种

• stdio: 标准IO

• sysIO: 系统IO(文件IO)

优先用标准IO

标准是一组接口:

• 移植性好, linux sysIO/win sysIO 都会实现该接口, 例如fopen是标准IO, linux是open的实现, win是openfile的实现
• 合并系统调用: 加速

stdio: FILE类型贯穿始终

fopen

# include <stdio.h>
# include <stdlib.h>
# include <errno.h>
# include <string.h>int main() {FILE *fp;fp = fopen("tmp", "r+");if (fp == NULL) {//        fprintf(stderr, "open failed! errno: %d\n", errno);perror("fopen()");fprintf(stderr, "fopen(): %s\n", strerror(errno));exit(1);}puts("ok");exit(0);
}

• Fopen() 创建的FILE是在堆上(一般有fopen_fclose, malloc-free这种逆操作的都是分配在堆上的)

FILE *open(const char *path, const char *mode){FILE *tmp = NULL;tm = malloc(sizeof(FILE));tmp->xx=xx;...return tmp;
}不可能放在栈是因为, if return &(函数内局部变量), 函数结束后变量就销毁了.
不可能放在静态区, 静态区只会有1份, 是因为该函数若被调用N次, 第2+次之后会覆盖第上一次打开的FILE;

fclose和文件权限

一个进程最多打开几个文件

# include <stdio.h>
# include <stdlib.h>
# include <errno.h>
# include <string.h>int main() {int count = 0;FILE *fp = NULL;while (1) {fp = fopen("tmp", "r");if (fp == NULL) {perror("fopen()");break;}count++;}printf("count = %d\n", count);exit(0);
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ./p5-fopen
fopen(): Too many open files
count = 253用ulimit -a可看到上限256, 再加上stdin, stdout, stderr这3个, 总共253+3=256;~/go/src/github.com/sonnary/apue/y   master ●✚  ulimit -a
-t: cpu time (seconds)              unlimited
-f: file size (blocks)              unlimited
-d: data seg size (kbytes)          unlimited
-s: stack size (kbytes)             8192
-c: core file size (blocks)         0
-v: address space (kbytes)          unlimited
-l: locked-in-memory size (kbytes)  unlimited
-u: processes                       709
-n: file descriptors                256vim /etc/security/limits.conf 改soft nofile和hard nofile上限值(单进程级, reboot生效, ulimit -n当终端退出后则失效) && vim /etc/sysctl.conf 修改fs.file-max值(系统级, reboot生效)
https://www.mdeditor.tw/pl/pFyl
http://www.3mu.me/linux%E4%BF%AE%E6%94%B9open-files%E6%95%B0%E5%8F%8Aulimit%E5%92%8Cfile-max%E7%9A%84%E5%8C%BA%E5%88%AB/

• umask

  新创建的文件的chmod默认是0666 & ~umask()

  例如: umask为0022, 新创建的文件就是0666-0022=0644

root@k8s-wolf-minion-47:~# umask
0022
root@k8s-wolf-minion-47:~# touch 123
root@k8s-wolf-minion-47:~# ll | grep 123
-rw-r--r--  1 root root   27069 Feb 18 23:23 123

fputc和pgetc

• copy命令

# include <stdio.h>
# include <stdlib.h>int main(int argc, char **argv) {FILE *fps = NULL;FILE *fpd = NULL;int ch;int put_ret_code;if (argc != 3) {fprintf(stderr, "Usage: %s <src_file_name>, <dst_file_name> need 2 arguments", argv[0]);exit(1);}fps = fopen(argv[1], "r");if (fps == NULL) {fclose(fps);perror("fopen");exit(1);}fpd = fopen(argv[2], "w");if (fpd == NULL) {perror("fopen");exit(1);}while (1) {ch = fgetc(fps);if (ch == EOF) {break;}put_ret_code = fputc(ch, fpd);if (put_ret_code == EOF) {break;}}fclose(fps);fclose(fpd);
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ./p8-fgetc /etc/services /tmp/out~/go/src/github.com/sonnary/apue/y   master ●✚  diff /etc/services /tmp/out是相同的, 模拟了linux的cp命令边界条件✘  ~/go/src/github.com/sonnary/apue/y   master ●✚  make p8-fgetc && ./p8-fgetc
cc     p8-fgetc.c   -o p8-fgetc
Usage: ./p8-fgetc, need 2 arguments%

• 计算文件的字节数命令

# include <stdio.h>
# include <stdlib.h>int main(int argc, char **argv) {if (argc != 2) {fprintf(stderr, "Usage: need 1 arg\n");exit(1);}FILE *fp = fopen(argv[1], "r");if (fp == NULL) {perror("fopen");exit(1);}long count = 0;while (fgetc(fp) != EOF) {count++;}fclose(fp);fprintf(stdout, "count: %ld\n", count);exit(0);
}

drwxr-xr-x   9 sunyuchuan  staff   288  2 19 00:20 .
drwxr-xr-x  27 sunyuchuan  staff   864  2 19 00:15 ..
-rwxr-xr-x   1 sunyuchuan  staff  8568  2 18 22:56 p5-fopen
-rw-r--r--   1 sunyuchuan  staff   347  2 18 22:55 p5-fopen.c
-rwxr-xr-x   1 sunyuchuan  staff  8756  2 18 23:53 p8-fgetc
-rw-r--r--   1 sunyuchuan  staff   764  2 18 23:55 p8-fgetc.c
-rwxr-xr-x   1 sunyuchuan  staff  8760  2 19 00:20 p8-fgetccount
-rw-r--r--   1 sunyuchuan  staff   428  2 19 00:20 p8-fgetccount.c
-rw-r--r--   1 sunyuchuan  staff     0  2 18 22:42 tmp
~/go/src/github.com/sonnary/apue/y   master ●✚  ./p8-fgetccount p8-fgetc
Usage: need 1 arg
✘  ~/go/src/github.com/sonnary/apue/y   master ●✚  make p8-fgetccount
cc     p8-fgetccount.c   -o p8-fgetccount
~/go/src/github.com/sonnary/apue/y   master ●✚  ./p8-fgetccount p8-fgetc
count: 8756
~/go/src/github.com/sonnary/apue/y   master ●✚  ./p8-fgetccount p8-fgetccount.c
count: 428

fputs fgets

# include <stdio.h>
# include <stdlib.h>
# define BUFSIZE 100int main(int argc, char **argv) {FILE *fps = NULL;FILE *fpd = NULL;char buf[BUFSIZE];if (argc != 3) {fprintf(stderr, "Usage: %s <src_file_name>, <dst_file_name> need 2 arguments", argv[0]);exit(1);}fps = fopen(argv[1], "r");if (fps == NULL) {fclose(fps);perror("fopen");exit(1);}fpd = fopen(argv[2], "w");if (fpd == NULL) {perror("fopen");exit(1);}while ( fgets(buf, BUFSIZE, fps) != NULL ) {fputs(buf, fpd);}fclose(fps);fclose(fpd);exit(0);
}

fread fwrite

• 要注意 这一对儿函数用的时候要把块大小设置为1
  • 因为其行为为返回实际操作的块个数, 所以无法区分, 一般最佳实践为块大小设置为1
    • eg: 如果1个块1有10Bytes, 但拿到了0Bytes, 返回值为0
    • eg: 如果1个块1有10Bytes, 但拿到了5Bytes, 返回值为0
    • eg: 如果1个块1有10Bytes, 但拿到了10Bytes, 返回值为1

# include <stdio.h>
# include <stdlib.h>
# define BUFSIZE 100int main(int argc, char **argv) {FILE *fps = NULL;FILE *fpd = NULL;char buf[BUFSIZE];if (argc != 3) {fprintf(stderr, "Usage: %s <src_file_name>, <dst_file_name> need 2 arguments", argv[0]);exit(1);}fps = fopen(argv[1], "r");if (fps == NULL) {fclose(fps);perror("fopen");exit(1);}fpd = fopen(argv[2], "w");if (fpd == NULL) {perror("fopen");exit(1);}int n = 0;while ( ( n = fread(buf, 1, BUFSIZE, fps) ) > 0 ) {fwrite(buf, 1, n, fpd);}fclose(fps);fclose(fpd);exit(0);
}

atoi和sprintf

是逆操作

fseeko和ftello

C99(为了移植)支持fseek和ftell, 因为返回的是long类型, 且只能用正数部分, 所以为0.5*2^32=2G大小, 不够文件大小用

方言(不考虑移植)推荐用fsseko和ftello, 其返回为off_t, 一般在编译时用gcc a.c -o a -D_FILE_OFFSET_BITS=64定义

文件位置函数

fseek, ftell, rewind

# include <stdio.h>
# include <stdlib.h>int main(int argc, char **argv) {if (argc != 2) {fprintf(stderr, "Usage: need 1 arg\n");exit(1);}FILE *fp = fopen(argv[1], "r");if (fp == NULL) {perror("fopen");exit(1);}int seek_success = fseek(fp, 0, SEEK_END);if (seek_success != 0) {perror("seek fail");exit(1);}printf("%ld\n", ftell(fp));fclose(fp);exit(0);
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ls -l flen.c
-rw-r--r--  1 sunyuchuan  staff  467  2 19 10:05 flen.c~/go/src/github.com/sonnary/apue/y   master ●✚  ./flen flen.c
467

• fseek可用于SEEK_END, 如迅雷下载, 把文件抻开成很大的大小(eg: 2GB), 然后分段用多线程下载
• rewind就是fseek(fp, 0, SEEK_SET)的封装, seek到文件头

缓冲区fflush

printf默认是行缓冲, 所以一般加fprintf("xxx %d\n", xxx);

缓冲区大多是好事, 用于合并系统调用

• 全缓冲: 满了强制刷 (用于非终端设备)
• 行缓冲: 换行or满了时刷新, (用于终端设备(如stdout等))
• 无缓冲: stderr

getline

在makefile中写#define

CFLAGS+=-D_FILE_OFFSET_BITS=64 -D_GNU_SOURCE

相当于在c文件开头写define, 会更易读

#define _GNU_SOURCE
#include <stdio.h>int main() {}

• getline

# include <stdio.h>
# include <stdlib.h>
# include <string.h>int main(int argc, char **argv) {if (argc < 2) {fprintf(stderr, "Usage...\n");exit(1);}FILE *fp = fopen(argv[1], "r");if (fp == NULL) {perror("fopen");exit(1);}char *linebuf = NULL;size_t linesize = 0;while(1) {if (getline(&linebuf, &linesize, fp) < 0 ) {break;}printf("%lu\n", strlen(linebuf));}fclose(fp);exit(0);
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ./getline makefile
45

系统IO

open, close, read, write, lseek

文件描述符是整数(数组下标), 优先返回可用的最小的fd

每个进程有个文件描述符数组(size为ulimit -n的长度), 数组下标为文件描述符, 值为FILE结构体指针, FILE结构体指向了文件inode

        thread 1+-------------+            +-------------+     +-------+
fd1  +val: &FILE1  +------------>             +-----> inode1||             |            |   FILE1     |     +-------++-------------+            +-------------+     +-------+
fd2  +val: &FILE2  |            |             +---->+ inode2||             |            |   FILE2     |     +-------++-------------+            +-------------+|             ||             |+-------------+|    ...      ||             |+-------------+thread 2+-------------+            +-------------+     +-------+
fd1  +val: &FILE1  +------------>             +-----> inode1||             |            |   FILE1     |     +-------++-------------+            +-------------+     +-------+
fd2  +val: &FILE2  |            |             +---->+ inode2||             |            |   FILE2     |     +-------++-------------+            +-------------+|             ||             |+-------------+|    ...      ||             |+-------------+

• man 2 open, C没有重载, 所以C是用变参函数实现的, (若传入的参数个数不符, C中的fprintf(“%d %d”, a,b,c)不会报错, 而C++重载是固定参数的则会报错)

int open(const char *pathname, int flags);
int open(const char *pathname, int flags, mode_t mode);

read write

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/uio.h>
# include <unistd.h>
# include <fcntl.h>
# define BUFSIZE 1024int main(int argc, char **argv) {int sfd, dfd;if (argc < 3) {fprintf(stderr, "Usage: 3 args...");exit(1);}sfd = open(argv[1], O_RDONLY);if (sfd < 0) {perror("open");exit(1);}dfd = open(argv[2], O_WRONLY|O_CREAT|O_TRUNC,0600);if (dfd < 0) {close(sfd);perror("open");exit(1);}char buf[BUFSIZE];int readed = 0, writted = 0;while(1) {readed = read(sfd, buf, BUFSIZE);if (readed < 0) {perror("read fail");break;}if (readed == 0) {fprintf(stdout, "all readed done");break;}int pos = 0;int towrite = readed;while(towrite > 0) {writted = write(dfd, buf + pos, towrite);if (writted < 0) {perror("write fail");exit(1);}towrite -= writted;pos += writted;}}close(sfd);close(dfd);exit(0);
}

标准IO: 引入缓冲区, 可以合并系统调用, 增大吞吐量

标准IO和文件IO不可混用, 因为有缓冲区的区别, 所以pos是不同的

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>int main() {putchar('a');write(1, "b", 1);putchar('a');write(1, "b", 1);putchar('a');write(1, "b", 1);exit(0);
}

root@k8s-wolf-minion-47:~# make write
cc     write.c   -o write
root@k8s-wolf-minion-47:~# ./write
bbbaaastrace ./write 跟踪系统调用
write(1, "b", 1b)                        = 1
write(1, "b", 1b)                        = 1
write(1, "b", 1b)                        = 1
write(1, "aaa", 3aaa)                      = 3
exit_group(0)                           = ?

原子操作

dup() dup2()

• 输出从stdout重定向到文件

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
# include <fcntl.h>
# define TMP_FILE "/tmp/out"
int main() {close(1);int fd = open(TMP_FILE, O_RDONLY|O_WRONLY|O_CREAT|O_TRUNC, 0600);if (fd < 0) {perror("open");exit(1);}puts("hello!");exit(0);
}

~/go/src/github.com/sonnary/apue/y   master ●✚  ./dup
~/go/src/github.com/sonnary/apue/y   master ●✚  cat /tmp/out
hello!

• dup: 复制一个最小的fd, 使原fd和新fd都指向同一个FILE, 该FILE指向inode(注意FILE内使用fd引用计数器的)

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
# include <fcntl.h>
# define TMP_FILE "/tmp/out"
int main() {int fd = open(TMP_FILE, O_RDONLY|O_WRONLY|O_CREAT|O_TRUNC, 0600);if (fd < 0) {perror("open");exit(1);}close(1);dup(fd); close(fd);puts("hello!");exit(0);
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  rm /tmp/out~/go/src/github.com/sonnary/apue/y   master ●✚  ./dup1~/go/src/github.com/sonnary/apue/y   master ●✚  cat /tmp/out
hello!

• dup2是原子操作

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
# include <fcntl.h>
# define TMP_FILE "/tmp/out"
int main() {int fd = open(TMP_FILE, O_RDONLY|O_WRONLY|O_CREAT|O_TRUNC, 0600);if (fd < 0) {perror("open");exit(1);}dup2(fd, 1);if (fd != 1)close(fd);puts("hello!");exit(0);
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ./dup2~/go/src/github.com/sonnary/apue/y   master ●✚  cat /tmp/out
hello!

fcntl 管理文件描述符

Ioctl 设备相关的内容

/dev/fd/目录: 畜牧路, 显示的是当前进程的目录描述符信息

文件系统

• stat: 通过文件路径获取属性, 面对符号链接文件是获取的是所指向文件的属性
• fstat: 通过文件描述符获取属性
• lstat: 面对符号链接文件是获取的是符号链接文阿基内的属性

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
# include <fcntl.h>off_t flen(char *filename) {struct stat data;if (stat(filename, &data) != 0) {return 0;}return data.st_size;
}int main(int argc, char **argv) {if (argc < 2) {fprintf(stderr, "Usage...\n");exit(1);}printf("%lld\n", flen(argv[1]));
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ll flen2.c
-rw-r--r--  1 sunyuchuan  staff   406B  2 19 17:23 flen2.c~/go/src/github.com/sonnary/apue/y   master ●✚  ./flen2 flen2.c
406~/go/src/github.com/sonnary/apue/y   master ●✚  stat flen2.c
16777220 105969084 -rw-r--r-- 1 sunyuchuan staff 0 406 "Feb 19 17:23:40 2021" "Feb 19 17:23:34 2021" "Feb 19 17:23:34 2021" "Feb 19 17:23:34 2021" 4096 8 0 flen2.c

• 文件真正占用磁盘容量是Blocks * BytesPerBlock(一般为512Bytes), 而不是file的size
• linux的cp命令支持空洞文件拷贝, 如果src文件里read出的一段内容都是\0的话, 只会记录却并不会write\0到dst文件, 最终在fst文件做lseek偏移
• 注意设置为LL才不会溢出int上限

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
# include <fcntl.h>int main(int argc, char **argv) {if (argc < 2) {fprintf(stderr, "Usage: ...\n");exit(1);}int fd = open(argv[1], O_WRONLY|O_CREAT|O_TRUNC, 0600);if (fd < 0) {perror("open");exit(1);}lseek(fd, 5LL*1024LL*1024LL*1024LL-1LL, SEEK_SET);write(fd, "", 1);exit(0);
}

root@k8s-wolf-minion-47:~# vim big.c
root@k8s-wolf-minion-47:~# make big
cc     big.c   -o big
root@k8s-wolf-minion-47:~# ./big bigfileroot@k8s-wolf-minion-47:~# ll bigfile
-rw------- 1 root root 5368709120 Feb 19 18:42 bigfileroot@k8s-wolf-minion-47:~# stat bigfileFile: 'bigfile'Size: 5368709120 Blocks: 8          IO Block: 4096   regular file
Device: fc00h/64512d    Inode: 76546720    Links: 1
Access: (0600/-rw-------)  Uid: (    0/    root)   Gid: (    0/    root)
Access: 2021-02-19 18:42:09.195734809 +0800
Modify: 2021-02-19 18:42:09.195734809 +0800
Change: 2021-02-19 18:42:09.195734809 +0800Birth: -root@k8s-wolf-minion-47:~# rm /tmp/bigfile.bk
root@k8s-wolf-minion-47:~# time bigfile /tmp/bigfile.bk
bigfile: command not foundreal  0m0.160s
user    0m0.120s
sys 0m0.040s
root@k8s-wolf-minion-47:~#

• 5G文件占用的是4KB(因为8块, 每块512Bytes)
• st_mode是16位的位图, 表示文件的类型
• 粘住位, 一般是/tmp的最后一位是t
• FAT文件系统: 静态单链表, 轻量级, 缺陷: 上限有限制, 单链表只能单向
• UFS:

文件

硬链接: 就是目录项, 缺点: 不能给分区建立, 不能给目录建立

软链接: 即符号链接, 其值为指向的文件名, 优点: 可以给分区建立, 可以给目录建立

link unlink remove rename

utime: 最后读/更改的时间

目录

opendir()
closedir()
readdir();
rewinddir();

##argc argv

    ./main abc 123 456 789argv+----------+
argv[0]  |          |         +-------+ "./main"|          +-------> +-------++----------+
argv[1]  |          |         +-------+ "abc"|          +-------> +-------++----------+
argv[2]  |          |         +-------+ "123"|          +-------> +-------++----------+
argv[3]  |          |                   "456"|          |+----------+
argv[4]  |          |                   "789"|          |+----------+

glob

解析模式/通配符

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
# include <fcntl.h>
# include <glob.h>
# define PAT "/etc/a*.conf"int main(int argc, char *argv[]) {if (argc != 1) {fprintf(stderr, "Usage: 1 argvs");exit(1);}glob_t globres;int i, err;err = glob(PAT, 0, NULL, &globres);if (err) {printf("Error code = %d\n", err);exit(1);}for (i = 0; i < globres.gl_pathc; i++) {puts(globres.gl_pathv[i]);}globfree(&globres);exit(0);
}

readdir

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
# include <fcntl.h>
# include <glob.h>
# include <dirent.h># define PAT "/etc"int main(int argc, char *argv[]) {DIR *dp;struct dirent *cur;dp = opendir(PAT);if (dp == NULL) {perror("opendir()");exit(1);}while((cur = readdir(dp)) != NULL)puts(cur->d_name);closedir(dp);exit(0);
}

du

占用的磁盘空间大小, KBytes为单位

root@k8s-wolf-minion-47:~# stat read.cFile: 'read.c'Size: 1068        Blocks: 8          IO Block: 4096   regular file
Device: fc00h/64512d    Inode: 76546716    Links: 1
Access: (0644/-rw-r--r--)  Uid: (    0/    root)   Gid: (    0/    root)
Access: 2021-02-19 13:24:30.344781444 +0800
Modify: 2021-02-19 13:24:28.736767302 +0800
Change: 2021-02-19 13:24:28.736767302 +0800Birth: -root@k8s-wolf-minion-47:~# du read.c
4   read.c8个block * 512Bytes/block = 4KBytes

自己实现du如下

# include <stdio.h>
# include <stdlib.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
# include <fcntl.h>
# include <glob.h>
# include <dirent.h>
# include <string.h>
# define PATHSIZE 1024static int path_noloop(const char *path) {char *pos = strrchr(path, '/');if ( pos == NULL )exit(1);if ((strcmp(pos+1, ".") == 0) || (strcmp(pos+1, "..") == 0))return 0;return 1;
}static int mydu(char *path) {struct stat statres;glob_t globres;char nextpath[PATHSIZE];int sum = 0;if (lstat(path, &statres) < 0) {perror("lstat");exit(1);}if (!S_ISDIR(statres.st_mode))return statres.st_blocks;strncpy(nextpath, path, PATHSIZE);strncat(nextpath, "/*", PATHSIZE);glob(nextpath, 0, NULL, &globres);strncpy(nextpath, path, PATHSIZE);strncat(nextpath, "/.*", PATHSIZE);glob(nextpath, GLOB_APPEND, NULL, &globres);int i = 0;for (i = 0; i < globres.gl_pathc; i++) {if (!path_noloop(globres.gl_pathv[i]))continue;sum += mydu(globres.gl_pathv[i]);}sum += statres.st_blocks;globfree(&globres);return sum;
}int main(int argc, char **argv) {if (argc != 2) {fprintf(stderr, "Usage: 1 argv");exit(1);}printf("%d", mydu(argv[1])/2); // 因为512Bytes为一个块, 所以占用的Bytes数为块数*512/1024=块数/2exit(1);
}

root@k8s-wolf-minion-47:./du /etc/
6860    /etc/
root@k8s-wolf-minion-47: ./mydu /etc/
6860%root@k8s-wolf-minion-47: du glob.c
8   glob.c
root@k8s-wolf-minion-47: ./mydu glob.c
8%

• Fdisk -l可以看到磁盘块大小

系统数据文件和信息

/etc/passwd

getpwuid getpwunam

/etc/passwd

Getgrgid getgrnam

/etc/shadow

单进程环境

• main函数

int main(int argc, char *argc[])

进程的终止

正常终止:

• c共main函数返回

int main() {return 0;
}
return 0是给当前进程的父进程看的./main
echo $?打印上次执行结果 若成功则为0

• exit()

int main() {exit(0);
}可返回-128~127

​ atexit钩子函数(压入栈), 即go的defer, c++的析构函数

# include<stdio.h>
# include<stdlib.h>
static void f1(void) {printf("f1 working\n");
}
static void f2(void) {printf("f2 working\n");
}
static void f3(void) {printf("f3 working\n");
}int main(int argc, char **argv) {printf("begin\n");atexit(f1);atexit(f2);atexit(f3);printf("end\n");exit(0);
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ./atexit
begin
end
f3 working
f2 working
f1 working

• _exit() 或_Exit()

  是系统调用, 不执行钩子函数和进程资源清理

• 最后一个现场从其启动例程返回

• 最后一个进程调用pthread_exit

异常终止:

• abort()
• 接到一个信号并终止
• 最后一个线程对其取消请求作出响应

命令行参数分析

getopt()
getopt_long()

• 可以传参

# include <stdio.h>
# include <stdlib.h>
# include <time.h>
# define TIMESTRSIZE 1024int main() {struct tm *tm;char timestr[TIMESTRSIZE];time_t stamp = time(NULL);tm = localtime(&stamp);strftime(timestr, TIMESTRSIZE, "Now:%Y-%m-%d", tm);puts(timestr);exit(0);
}

# include <stdio.h>
# include <stdlib.h>
# include <time.h>
# include <unistd.h>
# include <string.h>
# define TIMESTRSIZE 1024
# define FMTSTRSIZE 1024/**
* -y: year
* -m: month
* -d: day
* -H: hour
* -M: minute
* -s: second
*/
int main(int argc, char **argv) {struct tm *tm;char timestr[TIMESTRSIZE];int c;char fmtstr[FMTSTRSIZE];fmtstr[0] = '\0';time_t stamp = time(NULL);tm = localtime(&stamp);while(1) {c = getopt(argc, argv, "HMSymd");if (c < 0) {break;}switch(c) {case 'H':strncat(fmtstr, "%H ", FMTSTRSIZE);break;case 'M':strncat(fmtstr, "%M ", FMTSTRSIZE);break;case 'S':strncat(fmtstr, "%S ", FMTSTRSIZE);break;case 'y':strncat(fmtstr, "%y ", FMTSTRSIZE);break;case 'm':strncat(fmtstr, "%m ", FMTSTRSIZE);break;case 'd':strncat(fmtstr, "%d ", FMTSTRSIZE);break;default:break;}}strftime(timestr, TIMESTRSIZE, fmtstr, tm);puts(timestr);exit(0);
}

root@k8s-wolf-minion-47:~# date
Sun Feb 21 22:36:13 CST 2021
root@k8s-wolf-minion-47:~# ./getopt -M -S -m -d
36 13 02 21

• 下面是带参数的选项, 就是加冒号

# include <stdio.h>
# include <stdlib.h>
# include <time.h>
# include <unistd.h>
# include <string.h>
# define TIMESTRSIZE 1024
# define FMTSTRSIZE 1024/**
* -y: year
* -m: month
* -d: day
* -H: hour
* -M: minute
* -s: second
*/
int main(int argc, char **argv) {struct tm *tm;char timestr[TIMESTRSIZE];int c;char fmtstr[FMTSTRSIZE];fmtstr[0] = '\0';time_t stamp = time(NULL);tm = localtime(&stamp);while(1) {c = getopt(argc, argv, "H:MSy:md");if (c < 0) {break;}switch(c) {case 'H':if (strcmp(optarg, "12") == 0) strncat(fmtstr, "%I(%P) ", FMTSTRSIZE);else if (strcmp(optarg, "24") == 0) {strncat(fmtstr, "%H ", FMTSTRSIZE);} else {fprintf(stderr, "Invalid arguiments");}strncat(fmtstr, "%H ", FMTSTRSIZE);break;case 'M':strncat(fmtstr, "%M ", FMTSTRSIZE);break;case 'S':strncat(fmtstr, "%S ", FMTSTRSIZE);break;case 'y':if (strcmp(optarg, "2") == 0) strncat(fmtstr, "%y ", FMTSTRSIZE);else if (strcmp(optarg, "4") == 0)strncat(fmtstr, "%Y ", FMTSTRSIZE);else fprintf(stderr, "Invalid arguments of -y");strncat(fmtstr, "%y ", FMTSTRSIZE);break;case 'm':strncat(fmtstr, "%m ", FMTSTRSIZE);break;case 'd':strncat(fmtstr, "%d ", FMTSTRSIZE);break;default:break;}}strftime(timestr, TIMESTRSIZE, fmtstr, tm);puts(timestr);exit(0);
}

root@k8s-wolf-minion-47:~# date
Sun Feb 21 22:40:31 CST 2021
root@k8s-wolf-minion-47:~# ./getopt -M -S -m -d -y 4 -H 12
40 35 02 21 21 10(pm)22root@k8s-wolf-minion-47:~# date
Sun Feb 21 22:41:19 CST 2021
root@k8s-wolf-minion-47:~# ./getopt -H 12
10(pm)22root@k8s-wolf-minion-47:~# date
Sun Feb 21 22:45:39 CST 2021
root@k8s-wolf-minion-47:~# ./getopt -y 4 -m -d
2021 21 02 21
root@k8s-wolf-minion-47:~# ./getopt -y 2 -m -d
21 21 02 21

• 下面是非选项传参, 如果传参有文件&&文件可打开则写入文件, 否则写入stdout
  • 原理是非选项传参会被getopt()函数识别为1

# include <stdio.h>
# include <stdlib.h>
# include <time.h>
# include <unistd.h>
# include <string.h>
# define TIMESTRSIZE 1024
# define FMTSTRSIZE 1024/**
* -y: year
* -m: month
* -d: day
* -H: hour
* -M: minute
* -s: second
*/
int main(int argc, char **argv) {FILE *fp = stdout;struct tm *tm;char timestr[TIMESTRSIZE];int c;char fmtstr[FMTSTRSIZE];fmtstr[0] = '\0';time_t stamp = time(NULL);tm = localtime(&stamp);while(1) {c = getopt(argc, argv, "-H:MSy:md");if (c < 0) {break;}switch(c) {case 1:if (fp == stdout) {fp = fopen(argv[optind-1], "w");if (fp == NULL) {perror("fopen()");fp = stdout;} }break;case 'H':if (strcmp(optarg, "12") == 0) strncat(fmtstr, "%I(%P) ", FMTSTRSIZE);else if (strcmp(optarg, "24") == 0) {strncat(fmtstr, "%H ", FMTSTRSIZE);} else {fprintf(stderr, "Invalid arguiments");}strncat(fmtstr, "%H ", FMTSTRSIZE);break;case 'M':strncat(fmtstr, "%M ", FMTSTRSIZE);break;case 'S':strncat(fmtstr, "%S ", FMTSTRSIZE);break;case 'y':if (strcmp(optarg, "2") == 0) strncat(fmtstr, "%y ", FMTSTRSIZE);else if (strcmp(optarg, "4") == 0)strncat(fmtstr, "%Y ", FMTSTRSIZE);else fprintf(stderr, "Invalid arguments of -y");strncat(fmtstr, "%y ", FMTSTRSIZE);break;case 'm':strncat(fmtstr, "%m ", FMTSTRSIZE);break;case 'd':strncat(fmtstr, "%d ", FMTSTRSIZE);break;default:break;}}strncat(fmtstr, "\n", FMTSTRSIZE);strftime(timestr, TIMESTRSIZE, fmtstr, tm);fputs(timestr, fp);if (fp != stdout)fclose(fp);exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./getopt -y 4 /tmp/o0 -m /tmp/o1 -d /tmp/o2
ubuntu@k8s-wolf-minion-47:~$ cat /tmp/o0
2021 21 02 21
ubuntu@k8s-wolf-minion-47:~$ cat /tmp/o1
cat: /tmp/o1: No such file or directory
ubuntu@k8s-wolf-minion-47:~$ cat /tmp/o2
cat: /tmp/o2: No such file or directory

环境变量

都存在environ的全局变量中(类似errno全局变量), 是一个字符串数组

# include <stdio.h>
# include <stdlib.h>extern char **environ;
int main() {int i = 0;for (i = 0; environ[i] != NULL; i++)puts(environ[i]);exit(0);
}

• 运行结果和export相同, 会打印所有环境变量

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ./myenv
TERM_SESSION_ID=w0t1p0:AF2FA44C-F867-49A5-9309-1D3BCDDC8926
SSH_AUTH_SOCK=/private/tmp/com.apple.launchd.WpTXLeu8Cx/Listeners
LC_TERMINAL_VERSION=3.4.4~/go/src/github.com/sonnary/apue/y   master ●✚  export
TERM_SESSION_ID=w0t1p0:AF2FA44C-F867-49A5-9309-1D3BCDDC8926
SSH_AUTH_SOCK=/private/tmp/com.apple.launchd.WpTXLeu8Cx/Listeners
LC_TERMINAL_VERSION=3.4.4

• 一些函数

getenv();
setenv(); // 申请在堆上, 可指定是否覆盖
unsetenv(); // 释放老堆的值, 在堆上申请新值
putenv(); // 不好用, 因为参数没有const修饰

# include <stdio.h>
# include <stdlib.h>int main() {puts(getenv("PATH"));exit(0);
}

 ~/go/src/github.com/sonnary/apue/y   master ●✚  ./getenv
/Users/sunyuchuan/bin:/usr/local/bin:/usr/local/bin:/usr/bin:/bin:

• C程序的存储空间布局

32位 4G虚拟内存空间, 3G用户态, 1G内核态. 3G=0x08048000~0xC0000000

   4G   +-----------------+
kernel space              ||                 |3G   +---------------------------+ 0XC000000|                 ||                 ||                 ||                 ||                 ||                 |user space               ||                 ||                 ||                 ||                 ||                 ||                 |+---------------------------+0X08408000|                 ||                 |0     +-----------------+

# include <stdio.h>
# include <stdlib.h>int main() {puts(getenv("PATH"));getchar();exit(0);
}

root@k8s-wolf-minion-47:~# ps aux | grep getenv
root      93166  0.0  0.0   4352   672 pts/0    S+   16:21   0:00 ./getenvroot@k8s-wolf-minion-47:~# pmap 93166
93166:   ./getenv
0000000000400000      4K r-x-- getenv
0000000000600000      4K r---- getenv
0000000000601000      4K rw--- getenv
0000000001145000    132K rw---   [ anon ]
00007f5ce8073000   1792K r-x-- libc-2.23.so
00007f5ce8233000   2048K ----- libc-2.23.so
00007f5ce8433000     16K r---- libc-2.23.so
00007f5ce8437000      8K rw--- libc-2.23.so
00007f5ce8439000     16K rw---   [ anon ]
00007f5ce843d000    152K r-x-- ld-2.23.so
00007f5ce8658000     12K rw---   [ anon ]
00007f5ce8662000      4K r---- ld-2.23.so
00007f5ce8663000      4K rw--- ld-2.23.so
00007f5ce8664000      4K rw---   [ anon ]
00007ffc9817b000    132K rw---   [ stack ]
00007ffc981f2000     12K r----   [ anon ]
00007ffc981f5000      8K r-x--   [ anon ]
ffffffffff600000      4K r-x--   [ anon ]total             4356K

• pmap(1)命令

库

• 动态库

• 静态库

• 手工装载库

• 加载动态库的流程: dlopen, dlsym, dlerror, dlclose

  #include <stdio.h>
  #include <stdlib.h>
  #include <dlfcn.h>
  #include <gnu/lib-names.h>  /* Defines LIBM_SO (which will be astring such as "libm.so.6") */
  intmain(void)
  {void *handle;double (*cosine)(double);char *error;handle = dlopen(LIBM_SO, RTLD_LAZY);if (!handle) {fprintf(stderr, "%s\n", dlerror());exit(EXIT_FAILURE);}dlerror();    /* Clear any existing error */cosine = (double (*)(double)) dlsym(handle, "cos");/* According to the ISO C standard, casting between functionpointers and 'void *', as done above, produces undefined results.POSIX.1-2003 and POSIX.1-2008 accepted this state of affairs andproposed the following workaround:*(void **) (&cosine) = dlsym(handle, "cos");This (clumsy) cast conforms with the ISO C standard and willavoid any compiler warnings.The 2013 Technical Corrigendum to POSIX.1-2008 (a.k.a.POSIX.1-2013) improved matters by requiring that conformingimplementations support casting 'void *' to a function pointer.Nevertheless, some compilers (e.g., gcc with the '-pedantic'option) may complain about the cast used in this program. */error = dlerror();if (error != NULL) {fprintf(stderr, "%s\n", error);exit(EXIT_FAILURE);}printf("%f\n", (*cosine)(2.0));dlclose(handle);exit(EXIT_SUCCESS);
  }
  

  栈 静态区管理

函数跳转

setjmp()
longjmp()

• 比goto功能高级: 可以跨函数

# include <stdio.h>
# include <stdlib.h>
static void d(void) {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): End.\n", __FUNCTION__);
}static void c(void) {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): Call d().\n", __FUNCTION__);d();printf("%s():d() returned.\n", __FUNCTION__);printf("%s(): End.\n", __FUNCTION__);
}static void b(void) {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): Call c().\n", __FUNCTION__);c();printf("%s():c() returned.\n", __FUNCTION__);printf("%s(): End.\n", __FUNCTION__);
}static void a(void) {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): Call b().\n", __FUNCTION__);b();printf("%s():b() returned.\n", __FUNCTION__);printf("%s(): End.\n", __FUNCTION__);
}void main() {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): Call a().\n", __FUNCTION__);a();printf("%s():a() returned.\n", __FUNCTION__);printf("%s(): End.\n", __FUNCTION__);exit(0);
}

Main -> a() -> b() -> c() -> d()

 +------------+|            |  d
+-------------+|            ||            |  c+------------+|            ||            |  b|            |+------------+|            ||            |   a|            |+------------+|            ||            |   main|            |+------------+

root@k8s-wolf-minion-47:~# ./jump
main(): Begin.
main(): Call a().
a(): Begin.
a(): Call b().
b(): Begin.
b(): Call c().
c(): Begin.
c(): Call d().
d(): Begin.
d(): End.
c():d() returned.
c(): End.
b():c() returned.
b(): End.
a():b() returned.
a(): End.
main():a() returned.
main(): End.

• 函数a做setjump, 函数d做longjump跳到a处, 改写为如下
  • 能跨函数跳, 交出函数执行权+执行环境
  • setjmp()执行一次, 返回两次

# include <stdio.h>
# include <stdlib.h>
# include <setjmp.h>static jmp_buf save;static void d(void) {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): jump now\n", __FUNCTION__);longjmp(save, 6);printf("%s(): End.\n", __FUNCTION__);
}static void c(void) {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): Call d().\n", __FUNCTION__);d();printf("%s():d() returned.\n", __FUNCTION__);printf("%s(): End.\n", __FUNCTION__);
}static void b(void) {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): Call c().\n", __FUNCTION__);c();printf("%s():c() returned.\n", __FUNCTION__);printf("%s(): End.\n", __FUNCTION__);
}static void a(void) {printf("%s(): Begin.\n", __FUNCTION__);int ret = setjmp(save);if (ret == 0) {printf("%s(): Call b().\n", __FUNCTION__);b();printf("%s():b() returned.\n", __FUNCTION__);} else {printf("%s(): jumped back here with code %d\n", __FUNCTION__, ret);}printf("%s(): End.\n", __FUNCTION__);
}void main() {printf("%s(): Begin.\n", __FUNCTION__);printf("%s(): Call a().\n", __FUNCTION__);a();printf("%s():a() returned.\n", __FUNCTION__);printf("%s(): End.\n", __FUNCTION__);exit(0);
}

root@k8s-wolf-minion-47:~# ./setjmp
main(): Begin.
main(): Call a().
a(): Begin.
a(): Call b().
b(): Begin.
b(): Call c().
c(): Begin.
c(): Call d().
d(): Begin.
d(): jump now
a(): jumped back here with code 6
a(): End.
main():a() returned.
main(): End.

资源获取与控制(如ulimit)

gettlimit();
setrlimit();

组合就是ulimit命令

进程

进程标识符pid

类型pid_t 有符号的16位整数, 最多3w个进程

ps axf

ps axm

ps ax -L看轻量进程

进程号是顺次向下使用

getpid()

getppid()

进程的产生

fork()

• 在2个进程返回

  • 父进程 返回 子进程pid(失败时父进程 返回 -1 && 设置errno)
  • 子进程 返回 0

• 复制父进程(一模一样, 包括已执行到的位置)

• 除了以下不同:

  • pid不同, ppid也不同, fork()返回值不同
  • 未决信号和文件锁 不继承
  • 资源利用量清0

      parent process          child process+----------+                +----------+|          |                |          ||          |                |          ||          |                |          |
+-------------+已 执 行 处      +--------------+  已 执 行 处|          |                |          ||          |                |          ||          |    fork        |          ||          |  +--------->   |          ||          |                |          ||          |                |          ||          |                |          |+----------+                +----------+

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
int main() {pid_t pid;printf("[%d] Begin!\n", getpid());pid = fork();if (pid < 0) {perror("fork()");exit(1);} else if (pid == 0) {printf("[%d]: child is working\n", getpid());} else if (pid > 0) {printf("[%d]: parent is working\n", getpid());}printf("[%d] End!\n", getpid());exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./fork
[160859] Begin!
[160859]: parent is working
[160859] End!
[160860]: child is working
[160860] End!或 ubuntu@k8s-wolf-minion-47:~$ ./fork
[160859] Begin!
[160859]: child is working
[160859] End!
[160860]: parent is working
[160860] End!

• 调度器的调度策略 决定 哪个进程先运行(可能child或parent被调度的顺序是不确定的)
• 下面的代码加了getchar(), 使程序停下来, 用ps观察

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
int main() {pid_t pid;printf("[%d] Begin!\n", getpid());pid = fork();if (pid < 0) {perror("fork()");exit(1);} else if (pid == 0) {printf("[%d]: child is working\n", getpid());} else if (pid > 0) {printf("[%d]: parent is working\n", getpid());}printf("[%d] End!\n", getpid());getchar();exit(0);
}

ps axf 找进程号160027 ?        Ss     0:00  \_ sshd: ubuntu [priv]
160039 ?        S      0:00  |   \_ sshd: ubuntu@pts/0
160040 pts/0    Ss     0:00  |       \_ -bash
161677 pts/0    S+     0:00  |           \_ ./fork
161678 pts/0    S+     0:00  |               \_ ./fork

一定要先fflush()再fork()

• 去掉getchar, 重定向到文件中
  • Begin被打印2次
  • 解决办法, 在fork前用fflush()刷新所有成功打开的流

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
int main() {pid_t pid;printf("[%d] Begin!\n", getpid());pid = fork();if (pid < 0) {perror("fork()");exit(1);} else if (pid == 0) {printf("[%d]: child is working\n", getpid());} else if (pid > 0) {printf("[%d]: parent is working\n", getpid());}printf("[%d] End!\n", getpid());exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./fork > /tmp/out
ubuntu@k8s-wolf-minion-47:~$ cat /tmp/out
[162452] Begin!
[162452]: parent is working
[162452] End!
[162452] Begin!
[162453]: child is working
[162453] End!

• 增加fflush()改进为如下

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
int main() {pid_t pid;printf("[%d] Begin!\n", getpid());fflush(NULL);  pid = fork();if (pid < 0) {perror("fork()");exit(1);} else if (pid == 0) {printf("[%d]: child is working\n", getpid());} else if (pid > 0) {printf("[%d]: parent is working\n", getpid());}printf("[%d] End!\n", getpid());exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./fork > /tmp/out
ubuntu@k8s-wolf-minion-47:~$ cat /tmp/out
[163515] Begin!
[163515]: parent is working
[163515] End!
[163516]: child is working
[163516] End!

在begin放到缓冲区, 然后fork后, 父子进程的缓冲区各有一句"begin", 所以为了避免就需要fflush()

计算质数

• 单机版

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# define LEFT 30000000
# define RIGHT 30000200
int main() {int i, j, mark;for (i = LEFT; i <= RIGHT; i++) {mark = 1;for (j = 2; j < i/2; j++) {if (i % j == 0) {mark = 0;break;}}if (mark) {printf("%d is a primer\n", i);}}exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ time ./primer0 > /dev/nullreal  0m1.187s
user    0m1.188s
sys 0m0.000s

• 201个子进程
  • 父进程fork201次
  • 子进程负责执行, 但子进程不应继续fork孙进程(子进程需要有退出方法)

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# define LEFT 30000000
# define RIGHT 30000200
int main() {pid_t pid;int i, j, mark;for (i = LEFT; i <= RIGHT; i++) {pid = fork();if (pid < 0) {perror("fork");exit(1);}if (pid == 0) {mark = 1;for (j = 2; j < i/2; j++) {if (i % j == 0) {mark = 0;break;}}if (mark) {printf("%d is a primer\n", i);}   exit(0); // 子进程的退出方法, 防止子进程再fork孙进程}}exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./primer1 | wc -l
18
ubuntu@k8s-wolf-minion-47:~$ time ./primer1 > /dev/nullreal 0m0.020s
user    0m0.000s
sys 0m0.020s
// 性能取决于有几核, 若有x个核, 则y个进程排队等待被x个核调度, 则性能会提升x倍, 速度会变为y/x.

• 下面先让父进程结束

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# define LEFT 30000000
# define RIGHT 30000200
int main() {pid_t pid;int i, j, mark;for (i = LEFT; i <= RIGHT; i++) {pid = fork();if (pid < 0) {perror("fork");exit(1);}if (pid == 0) {mark = 1;for (j = 2; j < i/2; j++) {if (i % j == 0) {mark = 0;break;}}if (mark) {printf("%d is a primer\n", i);}   sleep(1000);exit(0); // 子进程的退出方法, 防止子进程再fork孙进程}}exit(0);
}

 42259 pts/0    S+     0:00 ./primer142260 pts/0    S+     0:00 ./primer142261 pts/0    S+     0:00 ./primer142262 pts/0    S+     0:00 ./primer142263 pts/0    S+     0:00 ./primer142264 pts/0    S+     0:00 ./primer142265 pts/0    S+     0:00 ./primer142266 pts/0    S+     0:00 ./primer142267 pts/0    S+     0:00 ./primer142268 pts/0    S+     0:00 ./primer142269 pts/0    S+     0:00 ./primer142270 pts/0    S+     0:00 ./primer142271 pts/0    S+     0:00 ./primer142272 pts/0    S+     0:00 ./primer142273 pts/0    S+     0:00 ./primer142274 pts/0    S+     0:00 ./primer142275 pts/0    S+     0:00 ./primer142276 pts/0    S+     0:00 ./primer142277 pts/0    S+     0:00 ./primer142278 pts/0    S+     0:00 ./primer142279 pts/0    S+     0:00 ./primer142280 pts/0    S+     0:00 ./primer142281 pts/0    S+     0:00 ./primer142282 pts/0    S+     0:00 ./primer142283 pts/0    S+     0:00 ./primer1

• 下面是父进程sleep(1000)
  • 当父进程sleep1000满后, 父进程exit(0), 子进程变为Z+僵尸进程
  • init收留各孤儿进程, 当子进程exit()后, init会将各子进程收尸, 变为孤儿进程(所以僵尸态是一闪而过的)
  • 僵尸态不占什么资源, 基本就是一个结构体指针(包含pid, 进程退出状态), 但是会占pid号(需要释放pid号)

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# define LEFT 30000000
# define RIGHT 30000200
int main() {pid_t pid;int i, j, mark;for (i = LEFT; i <= RIGHT; i++) {pid = fork();if (pid < 0) {perror("fork");exit(1);}if (pid == 0) {mark = 1;for (j = 2; j < i/2; j++) {if (i % j == 0) {mark = 0;break;}}if (mark) {printf("%d is a primer\n", i);}   exit(0); // 子进程的退出方法, 防止子进程再fork孙进程}}sleep(1000);exit(0);
}

36188 ?        Ss     0:00  \_ sshd: ubuntu [priv]36190 ?        S      0:00  |   \_ sshd: ubuntu@pts/036191 pts/0    Ss     0:00  |       \_ -bash43360 pts/0    S+     0:00  |           \_ ./primer143361 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43362 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43363 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43364 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43365 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43366 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43367 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43368 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43369 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43370 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43371 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43372 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43373 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43374 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43375 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43376 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43377 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43378 pts/0    Z+     0:00  |               \_ [primer1] <defunct>43379 pts/0    Z+     0:00  |               \_ [primer1] <defunct>

init进程

• 是1号进程, 是所有进程的祖先进程

vfork()

写时复制已合并到fork里了

进程的消亡及释放资源

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# define LEFT 30000000
# define RIGHT 30000200
int main() {pid_t pid;int i, j, mark;for (i = LEFT; i <= RIGHT; i++) {pid = fork();if (pid < 0) {perror("fork");exit(1);}if (pid == 0) {mark = 1;for (j = 2; j < i/2; j++) {if (i % j == 0) {mark = 0;break;}}if (mark) {printf("%d is a primer\n", i);}   exit(0); // 子进程的退出方法, 防止子进程再fork孙进程}}for (i = LEFT; i <= RIGHT; i++) {wait(NULL); // 循环201次, 等201个子进程exit后, 收201个子进程的收尸}sleep(1000);exit(0);
}

• 分配方法
  • 分块法
  • 交叉分配法
  • 池

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
#include <sys/types.h>
#include <sys/wait.h>
# define LEFT 30000000
# define RIGHT 30000200
# define N 3
int main() {pid_t pid;int i, j, n, mark;for (n = 0; n < N; n++) {pid = fork();if (pid < 0) {perror("");exit(1);}if (pid == 0) {for (i = LEFT+n; i <= RIGHT; i+=N) {mark = 1;for (j = 2; j < i/2; j++) {if (i % j == 0) {mark = 0;break;}}if (mark) {printf("[第%d个进程] 数字%d is a primer\n", n, i);}}exit(0); // 子进程的退出方法, 防止子进程再fork孙进程}}for (n = 0; n < N; n++) {wait(NULL); // 循环3次, 等3个子进程exit后, 收3个子进程的收尸}exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./primerN
[第1个进程] 数字30000001 is a primer
[第2个进程] 数字30000023 is a primer
[第1个进程] 数字30000037 is a primer
[第2个进程] 数字30000041 is a primer
[第2个进程] 数字30000059 is a primer
[第1个进程] 数字30000049 is a primer
[第2个进程] 数字30000071 is a primer
[第1个进程] 数字30000079 is a primer
[第2个进程] 数字30000083 is a primer
[第1个进程] 数字30000109 is a primer
[第2个进程] 数字30000137 is a primer
[第1个进程] 数字30000133 is a primer
[第2个进程] 数字30000149 is a primer
[第1个进程] 数字30000163 is a primer
[第2个进程] 数字30000167 is a primer
[第1个进程] 数字30000169 is a primer
[第1个进程] 数字30000193 is a primer
[第1个进程] 数字30000199 is a primerubuntu@k8s-wolf-minion-47:~$ ./primerN | wc -l
18ubuntu@k8s-wolf-minion-47:~$ time ./primerN > /dev/null
real    0m0.817s
user    0m1.444s
sys 0m0.000s

exec函数族

• 注意先fflush(), 再exec()

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
#include <sys/types.h>
#include <sys/wait.h>/**
* date +%s
*/
int main() {puts("Begin!");fflush(NULL);execl("/bin/date", "date", "+%s", NULL); // 用date进程代替本进程, 若成功则不会返回, 失败才会返回perror("execl");exit(1);puts("End!");exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./ex > /tmp/out
ubuntu@k8s-wolf-minion-47:~$ cat /tmp/out
1613972219

fork+exec+wait组合

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# include <sys/types.h>
# include <sys/wait.h>int main() {puts("BEGIN");fflush(NULL);pid_t pid = fork();if (pid < 0) {perror("fork()");exit(0);}if (pid == 0) {execl("/bin/date", "date", "+%s", NULL);perror("execl()");exit(1);}wait(NULL);puts("END");exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./few
BEGIN
1613972883
END

          father process                                       child processfile desc table         stdin                       file desc table+----------------+              +----+                +------------------------+0  |                |              |    | <-----------+  |                        ||                | +--------->  +----+                +------------------------++----------------+               stdout               |                        |1  |                |            +------+ <--------------+                        ||                | +--------> |      |                |                        |+----------------+            +------+                +------------------------+
2   |                |              stderr                |                        ||                | +--------> +-------+ <-----------+ |                        |+----------------+            +-------+               +------------------------+|                |                                    |                        ||                |                                    |                        ||                |                                    |                        ||                |           fork()                   |                        ||                |  +----------------------------->   |                        ||                |                                    |                        ||                |                                    |                        ||                |                                    |                        ||                |                                    |                        ||                |                                    |                        ||                |                                    |                        ||                |                                    |                        |+----------------+                                    +------------------------+

sleep(100)

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# include <sys/types.h>
# include <sys/wait.h>int main() {puts("BEGIN");fflush(NULL);pid_t pid = fork();if (pid < 0) {perror("fork()");exit(0);}if (pid == 0) {execl("/bin/sleep", "sleep", "100", NULL);perror("execl()");exit(1);}wait(NULL);puts("END");exit(0);
}

 66706 ?        Ss     0:00  \_ sshd: ubuntu [priv]66718 ?        S      0:00  |   \_ sshd: ubuntu@pts/066719 pts/0    Ss     0:00  |       \_ -bash70741 pts/0    S+     0:00  |           \_ ./sleep70742 pts/0    S+     0:00  |               \_ sleep 100

myshell练习

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# include <sys/types.h>
# include <sys/wait.h>
# include <string.h>
# include <glob.h>
# define DELIMS " \t\n"struct cmd_st {glob_t globres;
};static void prompt(void) {printf("mysh-0.1$ ");
}static void parse(char *line, struct cmd_st *res) {char *tok;int i = 0;while(1) {tok = strsep(&line, DELIMS);if (tok == NULL)break;if (tok[0] == '\0')continue;glob(tok, GLOB_NOCHECK|GLOB_APPEND * i, NULL, &res->globres);i = 1; }
}int main() {char *linebuf = NULL;size_t linebuf_size = 0;struct cmd_st cmd;while(1) {prompt();if( getline(&linebuf, &linebuf_size, stdin) < 0 ) {break;}parse(linebuf, &cmd);if (0) { // 是内部命令} else { // 是外部命令pid_t pid = fork();if (pid < 0) {perror("fork()");exit(1);}if (pid == 0) {// 子进程execvp(cmd.globres.gl_pathv[0], cmd.globres.gl_pathv);perror("execvp()");exit(1);}else {// 父进程wait(NULL);}}}
}

ubuntu@k8s-wolf-minion-47:~$ make mysh
cc     mysh.c   -o myshubuntu@k8s-wolf-minion-47:~$ ./mysh
mysh-0.1$ lsbig.c  ex.c  few.c  fork.c    getopt.c  mydu.c  mysh.c        primer0    primer1    primerN    read.c    sleep    time.c
ex     few   fork   getenv.c  jump.c    mysh    platformData  primer0.c  primer1.c  primerN.c  setjmp.c  sleep.c  write.cmysh-0.1$ pwd
/home/ubuntumysh-0.1$ ll
execvp(): No such file or directorymysh-0.1$ ls -l
total 164
-rw-r--r-- 1 ubuntu ubuntu  443 Feb 19 18:42 big.c
-rwxr-xr-x 1 ubuntu ubuntu 8752 Feb 22 13:36 ex
-rw-r--r-- 1 ubuntu ubuntu  342 Feb 22 13:36 ex.c
-rwxr-xr-x 1 ubuntu ubuntu 8904 Feb 22 13:47 few
-rw-r--r-- 1 ubuntu ubuntu  394 Feb 22 13:47 few.c
-rwxr-xr-x 1 ubuntu ubuntu 8864 Feb 22 01:10 fork
-rw-r--r-- 1 ubuntu ubuntu  420 Feb 22 01:10 fork.c
-rw-r--r-- 1 ubuntu ubuntu  105 Feb 21 16:20 getenv.c
-rw-r--r-- 1 ubuntu ubuntu 1922 Feb 21 23:01 getopt.c
-rw-r--r-- 1 ubuntu ubuntu  981 Feb 21 16:58 jump.c
-rw-r--r-- 1 ubuntu ubuntu 1262 Feb 20 11:54 mydu.c
-rwxr-xr-x 1 ubuntu ubuntu 9176 Feb 22 17:42 mysh
-rw-r--r-- 1 ubuntu ubuntu 1168 Feb 22 17:42 mysh.c
lrwxrwxrwx 1 root   root     13 Sep 14 16:31 platformData -> /platformData
-rwxr-xr-x 1 ubuntu ubuntu 8656 Feb 22 10:03 primer0
-rw-r--r-- 1 ubuntu ubuntu  372 Feb 22 10:03 primer0.c
-rwxr-xr-x 1 ubuntu ubuntu 8808 Feb 22 10:33 primer1
-rw-r--r-- 1 ubuntu ubuntu  563 Feb 22 10:33 primer1.c
-rwxr-xr-x 1 ubuntu ubuntu 8808 Feb 22 12:28 primerN
-rw-r--r-- 1 ubuntu ubuntu  770 Feb 22 12:28 primerN.c
-rw-r--r-- 1 ubuntu ubuntu 1068 Feb 19 13:24 read.c
-rw-r--r-- 1 ubuntu ubuntu 1228 Feb 21 21:54 setjmp.c
-rwxr-xr-x 1 ubuntu ubuntu 8912 Feb 22 13:56 sleep
-rw-r--r-- 1 ubuntu ubuntu  396 Feb 22 13:56 sleep.c
-rw-r--r-- 1 ubuntu ubuntu  294 Feb 21 22:21 time.c
-rw-r--r-- 1 ubuntu ubuntu  206 Feb 19 13:41 write.c
mysh-0.1$

用户权限和组权限

u+s, g+s的实现是通过fork + exec然后切换为对应effective user实现的

观摩课: 解释器文件

root@ubuntu:/home/ubuntu# chgrp ubuntu mysu
root@ubuntu:/home/ubuntu# ll mysu.c
-rw-r--r-- 1 root ubuntu 429 Feb 22 18:31 mysuroot@ubuntu:/home/ubuntu# chmod u+s mysu
root@ubuntu:/home/ubuntu# ll mysu.c
-rwSr--r-- 1 root ubuntu 429 Feb 22 18:31 mysu

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# include <sys/types.h>
# include <sys/stat.h>
# include <sys/wait.h>int main(int argc, char **argv) {if (argc < 3) {fprintf(stderr, "Usage...\n");exit(1);}pid_t pid = fork();if (pid < 0) {perror("fork");exit(1);}if (pid ==0) {setuid(atoi(argv[1]));execvp(argv[2], argv+2);perror("execvp");exit(1);}wait(NULL);exit(0);
}

ubuntu@ubuntu:~$ cat /etc/shadow
cat: /etc/shadow: Permission deniedubuntu@ubuntu:~$ ./mysu 0 cat /etc/shadow
root:*:18519:0:99999:7:::
daemon:*:17743:0:99999:7:::
bin:*:17743:0:99999:7:::
sys:*:17743:0:99999:7:::
sync:*:17743:0:99999:7:::
games:*:17743:0:99999:7:::
man:*:17743:0:99999:7:::
lp:*:17743:0:99999:7:::
mail:*:17743:0:99999:7:::
news:*:17743:0:99999:7:::
uucp:*:17743:0:99999:7:::
proxy:*:17743:0:99999:7:::
www-data:*:17743:0:99999:7:::
backup:*:17743:0:99999:7:::
list:*:17743:0:99999:7:::
irc:*:17743:0:99999:7:::
gnats:*:17743:0:99999:7:::
nobody:*:17743:0:99999:7:::
systemd-timesync:*:17743:0:99999:7:::
systemd-network:*:17743:0:99999:7:::
systemd-resolve:*:17743:0:99999:7:::
systemd-bus-proxy:*:17743:0:99999:7:::
syslog:*:17743:0:99999:7:::
_apt:*:17743:0:99999:7:::
lxd:*:18519:0:99999:7:::
messagebus:*:18519:0:99999:7:::
uuidd:*:18519:0:99999:7:::
dnsmasq:*:18519:0:99999:7:::
sshd:*:18519:0:99999:7:::
ubuntu:$6$Be.6KxDz$5MH6Hm2yqudKmqroQO8uA04WdsdBoeE/JLR395vRGmANkRMlcam8/hZra18hWQjE5d57z7BbLPyjsyTHQYv9q.:18519:0:99999:7:::
ftp:*:18519:0:99999:7:::
postgres:*:18519:0:99999:7:::

system()

# include <stdio.h>
# include <stdlib.h>
int main(int argc, char **argv) {system("date +%s > /tmp/out");exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./sys
ubuntu@k8s-wolf-minion-47:~$ cat /tmp/out
1613993971

• 是fork, exec, wait的封装, system()会调/bin/sh -c, 内部实际上是如下原理

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# include <sys/types.h>
# include <sys/wait.h>int main() {puts("BEGIN");fflush(NULL);pid_t pid = fork();if (pid < 0) {perror("fork()");exit(0);}if (pid == 0) {execl("/bin/sh", "sh", "-c", "date +%s", NULL);perror("execl()");exit(1);}wait(NULL);puts("END");exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./sys1
BEGIN
1613994175
END

进程会计

acct

进程时间

times

守护进程

httpd sshd vsftpd dhcp

守护进程是没有stdin的(会重定向或close), (因为守护进程是后台进程, 一旦stdin输入会把守护进程杀死)

如下3个特点:

• 子进程调用setsid()使sessionid, processid, processgroupid相同
• 并且其父进程退出, 使init为其父进程
• tty为空

getpgrp(), getpgid(), setpgid()

• 下面写一个守护进程

#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <sys/stat.h>
#include <fcntl.h>
#define FNAME "/tmp/out"
static int daemonize() {pid_t pid = fork();if (pid < 0) {perror("fork()");return -1;}if (pid > 0)  // parentexit(0);// ------以下为child------// 重定向stdin, stdout, stderrint fd = open("/dev/null", O_RDWR);if (fd < 0) {perror("open");return -1;}dup2(fd, 0);dup2(fd, 1);dup2(fd, 2);if (fd > 2) close(fd);setsid();chdir("/");umask(0);// 如果不产生文件的话, 可以关掉umaskreturn 0;
}int main() {if (daemonize()) {exit(1);}FILE *fp = fopen(FNAME, "w");if (fp == NULL) {perror("fopen()");exit(1);}int i = 0;for ( i = 0; ; i++) {fprintf(fp, "%d", i);fflush(fp);sleep(1);}exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./mydaemon
ubuntu@k8s-wolf-minion-47:~$ cat /tmp/out
01234ubuntu@k8s-wolf-minion-47:~$ cat /tmp/out
0123456ubuntu@k8s-wolf-minion-47:~$ cat /tmp/out
01234567ubuntu@k8s-wolf-minion-47:~$ cat /tmp/out
012345678ubuntu@k8s-wolf-minion-47:~$ubuntu@k8s-wolf-minion-47:~$ ps axjPPID    PID   PGID    SID TTY       TPGID STAT   UID   TIME COMMAND1 126975 126975 126975 ?            -1 Ss    1000   0:00 ./mydaemonubuntu@k8s-wolf-minion-47:~$ kill 126975

• vsftpd的守护进程

root@k8s-wolf-minion-47:~# service vsftpd status
● vsftpd.service - vsftpd FTP serverLoaded: loaded (/lib/systemd/system/vsftpd.service; enabled; vendor preset: enabled)Active: active (running) since Fri 2020-11-06 11:50:53 CST; 3 months 17 days agoMain PID: 1532 (vsftpd)Tasks: 1Memory: 1.2MCPU: 6msCGroup: /system.slice/vsftpd.service└─1532 /usr/sbin/vsftpd /etc/vsftpd.confWarning: Journal has been rotated since unit was started. Log output is incomplete or unavailable.
root@k8s-wolf-minion-47:~# service vsftpd stop
root@k8s-wolf-minion-47:~# service vsftpd start
root@k8s-wolf-minion-47:~# service vsftpd start
root@k8s-wolf-minion-47:~# ps axj  | grep ftp1 145688 145688 145688 ?            -1 Ss       0   0:00 /usr/sbin/vsftpd /etc/vsftpd.conf

单实例守护进程-锁文件

• /var/run/xxx.pid`

root@k8s-wolf-minion-47:/var/run# ll | grep pid
-rw-r--r--  1 root     root        5 Nov  6 11:50 acpid.pid
srw-rw-rw-  1 root     root        0 Nov  6 11:50 acpid.socket=
-rw-r--r--  1 root     root        5 Nov  6 11:50 atd.pid
-rw-r--r--  1 root     root        5 Nov  6 11:50 crond.pid
-rw-r--r--  1 root     root        4 Nov  6 11:50 docker.pid
-rw-r--r--  1 root     root        5 Nov  6 11:50 irqbalance.pid
-rw-------  1 root     root        5 Nov  6 11:50 iscsid.pid
-rw-r--r--  1 root     root        4 Nov  6 11:50 lvmetad.pid
-rw-------  1 root     root        5 Nov  6 11:50 lxcfs.pid
-rw-r--r--  1 root     root        4 Nov  6 11:50 rsyslogd.pid
-rw-r--r--  1 root     root        6 Nov 23 11:14 sshd.pid
-rw-r--r--  1 root     root        5 Dec 14 10:47 supervisord.pid
root@k8s-wolf-minion-47:/var/run# cat supervisord.pid
7148
root@k8s-wolf-minion-47:/var/run# cat sshd.pid
38648root@k8s-wolf-minion-47:/var/run# ps axj | grep 71487148    910    910   7148 ?            -1 S        0   0:00 /bin/bash /home/ubuntu/platformTG/dbtool/latest/sv_start.sh910    911    910   7148 ?            -1 Sl       0  41:26 ./dbtool1   7148   7148   7148 ?            -1 Ss       0  86:27 /usr/bin/python /usr/bin/supervisord -n -c /etc/supervisor/supervisord.confroot@k8s-wolf-minion-47:/var/run# ps axj | grep 386481  38648  38648  38648 ?            -1 Ss       0   0:00 sshd: /usr/sbin/sshd -D -f /etc/ssh/sshd_config [listener] 0 of 10-100 startups38648 139349 139349 139349 ?            -1 Ss       0   0:00 sshd: ubuntu [priv]root@k8s-wolf-minion-47: ps axf -p 714838648 ?        Ss     0:00 sshd: /usr/sbin/sshd -D -f /etc/ssh/sshd_config [listener] 0 of 10-100 startups
139349 ?        Ss     0:00  \_ sshd: ubuntu [priv]
139392 ?        S      0:00      \_ sshd: ubuntu@pts/0
139403 pts/0    Ss     0:00          \_ -bash
143215 pts/0    S      0:00              \_ sudo -i
143216 pts/0    S      0:00                  \_ -bash
146989 pts/0    R+     0:00                      \_ ps axf -p 71487148 ?        Ss    86:27 /usr/bin/python /usr/bin/supervisord -n -c /etc/supervisor/supervisord.conf910 ?        S      0:00  \_ /bin/bash /home/ubuntu/platformTG/dbtool/latest/sv_start.sh911 ?        Sl    41:26      \_ ./dbtool

守护进程开机启动脚本文件

• /etc/rc..., 一般把启动命令(如启动pg/sshd/ftpd)放在这里, 会使守护进程开机启动, 从ubuntu看是软链接到了/etc/init.d

root@k8s-wolf-minion-47:/etc/rc0.d# ll
total 12
drwxr-xr-x   2 root root 4096 Dec 14 10:47 ./
drwxr-xr-x 101 root root 4096 Feb 22 23:14 ../
lrwxrwxrwx   1 root root   13 Sep 14 12:02 K01atd -> ../init.d/atd*
lrwxrwxrwx   1 root root   16 Sep 14 16:32 K01docker -> ../init.d/docker*
lrwxrwxrwx   1 root root   18 Nov 13 16:32 K01ebtables -> ../init.d/ebtables*
lrwxrwxrwx   1 root root   20 Sep 14 12:01 K01irqbalance -> ../init.d/irqbalance*
lrwxrwxrwx   1 root root   22 Sep 14 11:57 K01lvm2-lvmetad -> ../init.d/lvm2-lvmetad*
lrwxrwxrwx   1 root root   23 Sep 14 11:57 K01lvm2-lvmpolld -> ../init.d/lvm2-lvmpolld*
lrwxrwxrwx   1 root root   15 Sep 14 12:01 K01lxcfs -> ../init.d/lxcfs*
lrwxrwxrwx   1 root root   13 Sep 14 12:02 K01lxd -> ../init.d/lxd*
lrwxrwxrwx   1 root root   15 Sep 14 12:02 K01mdadm -> ../init.d/mdadm*
lrwxrwxrwx   1 root root   27 Nov 13 16:35 K01nfs-kernel-server -> ../init.d/nfs-kernel-server*
lrwxrwxrwx   1 root root   20 Sep 14 12:01 K01open-iscsi -> ../init.d/open-iscsi*
lrwxrwxrwx   1 root root   23 Sep 14 12:02 K01open-vm-tools -> ../init.d/open-vm-tools*
lrwxrwxrwx   1 root root   18 Sep 14 12:01 K01plymouth -> ../init.d/plymouth*
lrwxrwxrwx   1 root root   20 Sep 14 16:46 K01postgresql -> ../init.d/postgresql*
lrwxrwxrwx   1 root root   20 Sep 14 11:55 K01resolvconf -> ../init.d/resolvconf*
lrwxrwxrwx   1 root root   20 Dec 14 10:47 K01supervisor -> ../init.d/supervisor*
lrwxrwxrwx   1 root root   29 Sep 14 12:02 K01unattended-upgrades -> ../init.d/unattended-upgrades*
lrwxrwxrwx   1 root root   17 Sep 14 11:55 K01urandom -> ../init.d/urandom*
lrwxrwxrwx   1 root root   15 Sep 14 12:01 K01uuidd -> ../init.d/uuidd*
lrwxrwxrwx   1 root root   16 Sep 14 16:32 K01vsftpd -> ../init.d/vsftpd*
lrwxrwxrwx   1 root root   24 Sep 14 16:32 K02cgroupfs-mount -> ../init.d/cgroupfs-mount*
lrwxrwxrwx   1 root root   16 Sep 14 12:01 K02iscsid -> ../init.d/iscsid*
lrwxrwxrwx   1 root root   18 Sep 14 12:01 K03sendsigs -> ../init.d/sendsigs*
lrwxrwxrwx   1 root root   17 Sep 14 12:01 K04rsyslog -> ../init.d/rsyslog*
lrwxrwxrwx   1 root root   20 Sep 14 12:01 K05hwclock.sh -> ../init.d/hwclock.sh*
lrwxrwxrwx   1 root root   22 Sep 14 12:01 K05umountnfs.sh -> ../init.d/umountnfs.sh*
lrwxrwxrwx   1 root root   17 Nov 13 16:34 K06rpcbind -> ../init.d/rpcbind*
lrwxrwxrwx   1 root root   20 Nov 13 16:34 K07networking -> ../init.d/networking*
lrwxrwxrwx   1 root root   18 Nov 13 16:34 K08umountfs -> ../init.d/umountfs*
lrwxrwxrwx   1 root root   20 Nov 13 16:34 K09cryptdisks -> ../init.d/cryptdisks*
lrwxrwxrwx   1 root root   26 Nov 13 16:34 K10cryptdisks-early -> ../init.d/cryptdisks-early*
lrwxrwxrwx   1 root root   20 Nov 13 16:34 K11umountroot -> ../init.d/umountroot*
lrwxrwxrwx   1 root root   24 Nov 13 16:34 K12mdadm-waitidle -> ../init.d/mdadm-waitidle*
lrwxrwxrwx   1 root root   14 Nov 13 16:34 K13halt -> ../init.d/halt*
-rw-r--r--   1 root root  353 Jan 20  2016 README
root@k8s-wolf-minion-47:/etc/rc0.d# cd ..
root@k8s-wolf-minion-47:/etc# ll rc1.d/
total 12
drwxr-xr-x   2 root root 4096 Dec 14 10:47 ./
drwxr-xr-x 101 root root 4096 Feb 22 23:14 ../
lrwxrwxrwx   1 root root   13 Sep 14 12:02 K01atd -> ../init.d/atd*
lrwxrwxrwx   1 root root   16 Sep 14 16:32 K01docker -> ../init.d/docker*
lrwxrwxrwx   1 root root   18 Nov 13 16:32 K01ebtables -> ../init.d/ebtables*
lrwxrwxrwx   1 root root   20 Sep 14 12:01 K01irqbalance -> ../init.d/irqbalance*
lrwxrwxrwx   1 root root   22 Sep 14 11:57 K01lvm2-lvmetad -> ../init.d/lvm2-lvmetad*
lrwxrwxrwx   1 root root   23 Sep 14 11:57 K01lvm2-lvmpolld -> ../init.d/lvm2-lvmpolld*
lrwxrwxrwx   1 root root   15 Sep 14 12:01 K01lxcfs -> ../init.d/lxcfs*
lrwxrwxrwx   1 root root   13 Sep 14 12:02 K01lxd -> ../init.d/lxd*
lrwxrwxrwx   1 root root   15 Sep 14 12:02 K01mdadm -> ../init.d/mdadm*
lrwxrwxrwx   1 root root   27 Nov 13 16:35 K01nfs-kernel-server -> ../init.d/nfs-kernel-server*
lrwxrwxrwx   1 root root   20 Sep 14 12:01 K01open-iscsi -> ../init.d/open-iscsi*
lrwxrwxrwx   1 root root   23 Sep 14 12:02 K01open-vm-tools -> ../init.d/open-vm-tools*
lrwxrwxrwx   1 root root   20 Sep 14 16:46 K01postgresql -> ../init.d/postgresql*
lrwxrwxrwx   1 root root   20 Dec 14 10:47 K01supervisor -> ../init.d/supervisor*
lrwxrwxrwx   1 root root   13 Sep 14 12:02 K01ufw -> ../init.d/ufw*
lrwxrwxrwx   1 root root   15 Sep 14 12:01 K01uuidd -> ../init.d/uuidd*
lrwxrwxrwx   1 root root   16 Sep 14 16:32 K01vsftpd -> ../init.d/vsftpd*
lrwxrwxrwx   1 root root   24 Sep 14 16:32 K02cgroupfs-mount -> ../init.d/cgroupfs-mount*
lrwxrwxrwx   1 root root   16 Sep 14 12:01 K02iscsid -> ../init.d/iscsid*
lrwxrwxrwx   1 root root   17 Sep 14 12:01 K04rsyslog -> ../init.d/rsyslog*
lrwxrwxrwx   1 root root   17 Nov 13 16:34 K06rpcbind -> ../init.d/rpcbind*
-rw-r--r--   1 root root  369 Jan 20  2016 README
lrwxrwxrwx   1 root root   19 Sep 14 11:55 S01killprocs -> ../init.d/killprocs*
lrwxrwxrwx   1 root root   16 Sep 14 11:55 S02single -> ../init.d/single*
root@k8s-wolf-minion-47:/etc# ll -R rc* | grep postgres
lrwxrwxrwx   1 root root   20 Sep 14 16:46 K01postgresql -> ../init.d/postgresql*
lrwxrwxrwx   1 root root   20 Sep 14 16:46 K01postgresql -> ../init.d/postgresql*
lrwxrwxrwx   1 root root   20 Sep 14 16:46 S02postgresql -> ../init.d/postgresql*
lrwxrwxrwx   1 root root   20 Sep 14 16:46 S02postgresql -> ../init.d/postgresql*
lrwxrwxrwx   1 root root   20 Sep 14 16:46 S02postgresql -> ../init.d/postgresql*
lrwxrwxrwx   1 root root   20 Sep 14 16:46 S02postgresql -> ../init.d/postgresql*
lrwxrwxrwx   1 root root   20 Sep 14 16:46 K01postgresql -> ../init.d/postgresql*

系统日志

ubuntu@k8s-wolf-minion-47:/var/log$ ll | grep -v gz
total 823284
drwxrwxr-x 13 root   syslog        4096 Feb 21 06:25 ./
drwxr-xr-x 14 root   root          4096 Sep 14 16:46 ../
-rw-r--r--  1 root   root           505 Feb 10 11:03 alternatives.log
-rw-r--r--  1 root   root          1000 Feb  3 17:23 alternatives.log.1
-rw-r--r--  1 root   root        149090 Sep 16 16:13 ansible.log
drwxr-xr-x  2 root   root          4096 Feb  4 06:25 apt/
-rw-r-----  1 syslog adm          49055 Feb 22 20:55 auth.log
-rw-r-----  1 syslog adm         224972 Feb 21 06:25 auth.log.1
-rw-r--r--  1 root   root         57457 Jul 31  2018 bootstrap.log
-rw-rw----  1 root   utmp             0 Feb  1 06:25 btmp
-rw-rw----  1 root   utmp             0 Jan  1 06:25 btmp.1
drwxr-xr-x  2 root   root          4096 Jan  5 15:53 containers/
drwxr-xr-x  2 root   root          4096 Apr  9  2018 dist-upgrade/
-rw-r-----  1 root   adm             31 Jul 31  2018 dmesg
-rw-r--r--  1 root   root         27128 Feb 10 11:03 dpkg.log
-rw-r--r--  1 root   root         23078 Feb  3 17:40 dpkg.log.1
-rw-r--r--  1 root   root         32096 Feb 22 17:47 faillog
drwxr-xr-x  2 root   root          4096 Sep 14 11:55 fsck/
drwxr-xr-x  3 root   root          4096 Sep 14 12:05 installer/
-rw-r-----  1 syslog adm              0 Jan 18 06:25 kern.log
-rw-r-----  1 syslog adm           1522 Jan 12 11:26 kern.log.1
-rw-rw-r--  1 root   utmp        292876 Feb 22 20:40 lastlog
drwxr-xr-x  2 root   root          4096 Dec  8  2017 lxd/
drwxr-xr-x  2 root   root          4096 Jan  5 15:53 pods/
drwxrwxr-t  2 root   postgres      4096 Jan 31 06:25 postgresql/
drwxr-xr-x  2 root   root          4096 Dec 14 10:47 supervisor/
-rw-r-----  1 syslog adm      446618132 Feb 22 20:59 syslog
-rw-r-----  1 syslog adm      277778484 Feb 21 06:25 syslog.1
drwxr-xr-x  2 root   root          4096 Dec 11  2017 sysstat/
drwxr-x---  2 root   adm           4096 Oct 20 16:28 unattended-upgrades/
-rw-rw-r--  1 root   utmp         38016 Feb 22 20:50 wtmp
-rw-rw-r--  1 root   utmp         72576 Jan 29 18:09 wtmp.1

• syslogd守护进程负责

root@k8s-wolf-minion-47:~# ps aux | grep syslogd
syslog     1311  0.0  0.0 256392  4304 ?        Ssl   2020  60:28 /usr/sbin/rsyslogd -n

• Openlog

• closelog

• Syslog

#include <stdio.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <string.h>
#include <errno.h>
#include <syslog.h>
#define FNAME "/tmp/out"
static int daemonize() {pid_t pid = fork();if (pid < 0) {return -1;}if (pid > 0)  // parentexit(0);// ------以下为child------// 重定向stdin, stdout, stderrint fd = open("/dev/null", O_RDWR);if (fd < 0) {return -1;}dup2(fd, 0);dup2(fd, 1);dup2(fd, 2);if (fd > 2) close(fd);setsid();chdir("/");umask(0);// 如果不产生文件的话, 可以关掉umaskreturn 0;
}int main() {openlog("mydaemon", LOG_PID, LOG_DAEMON);if (daemonize()) {syslog(LOG_ERR, "daemonize() failed!"); // 注意这里不要加\n, syslog自己控制格式exit(1);} syslog(LOG_INFO, "daemonize() seccessed!");FILE *fp = fopen(FNAME, "w");if (fp == NULL) {syslog(LOG_ERR, "fopen:%s", strerror(errno));exit(1);}syslog(LOG_INFO, "%s was opened.", FNAME);int i = 0;for ( i = 0; ; i++) {fprintf(fp, "%d", i);fflush(fp);syslog(LOG_DEBUG, "%d is printed.", i);sleep(1);}fclose(fp);closelog();exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./mydaemon
ubuntu@k8s-wolf-minion-47:~$ ps -ef | grep mydaemon
ubuntu   143056      1  0 22:56 ?        00:00:00 ./mydaemonroot@k8s-wolf-minion-47:~# grep 143056 /var/log/syslog
Feb 22 22:56:11 k8s-wolf-minion-47 mydaemon[143056]: daemonize() seccessed!
Feb 22 22:56:11 k8s-wolf-minion-47 mydaemon[143056]: /tmp/out was opened.
Feb 22 22:56:11 k8s-wolf-minion-47 mydaemon[143056]: 0 is printed.
Feb 22 22:56:12 k8s-wolf-minion-47 mydaemon[143056]: 1 is printed.
Feb 22 22:56:13 k8s-wolf-minion-47 mydaemon[143056]: 2 is printed.
Feb 22 22:56:14 k8s-wolf-minion-47 mydaemon[143056]: 3 is printed.
Feb 22 22:56:15 k8s-wolf-minion-47 mydaemon[143056]: 4 is printed.
Feb 22 22:56:16 k8s-wolf-minion-47 mydaemon[143056]: 5 is printed.
Feb 22 22:56:17 k8s-wolf-minion-47 mydaemon[143056]: 6 is printed.
Feb 22 22:56:18 k8s-wolf-minion-47 mydaemon[143056]: 7 is printed.
Feb 22 22:56:19 k8s-wolf-minion-47 mydaemon[143056]: 8 is printed.
Feb 22 22:56:20 k8s-wolf-minion-47 mydaemon[143056]: 9 is printed.root@k8s-wolf-minion-47:~# vim /etc/rsyslog.conf

并发

• 同步

• 异步: 事件到来时间未知, 事件到来的后果未知

  • 主动轮询法: 适合于事件发生频率频繁时, (eg钓鱼, 鱼很多, 适合轮询法: 每1min拿一个网子去捞鱼)
  • 被动通知法: 适合于事件发生频率稀疏时, (eg钓鱼, 鱼很少, 适合通知法: 甩钓竿和浮漂, 等浮漂上下浮动过快时通知自己去捞鱼, 省力气)

• 单核: 各进程按时间片分片, 没有真正意义的并行

• 多核: 才有并行

• 信号和多线程一般不会混用

通过信号实现并发

信号的概念

信号是软件中断, 信号(应用软件层)的响应依赖于中断(硬件层).

signal()

ubuntu@k8s-wolf-minion-47:~$ kill -l1) SIGHUP    2) SIGINT   3) SIGQUIT  4) SIGILL   5) SIGTRAP6) SIGABRT    7) SIGBUS   8) SIGFPE   9) SIGKILL 10) SIGUSR1
11) SIGSEGV 12) SIGUSR2 13) SIGPIPE 14) SIGALRM 15) SIGTERM
16) SIGSTKFLT   17) SIGCHLD 18) SIGCONT 19) SIGSTOP 20) SIGTSTP
21) SIGTTIN 22) SIGTTOU 23) SIGURG  24) SIGXCPU 25) SIGXFSZ
26) SIGVTALRM   27) SIGPROF 28) SIGWINCH    29) SIGIO   30) SIGPWR
31) SIGSYS  34) SIGRTMIN    35) SIGRTMIN+1 36) SIGRTMIN+2 37) SIGRTMIN+3
38) SIGRTMIN+4 39) SIGRTMIN+5 40) SIGRTMIN+6 41) SIGRTMIN+7 42) SIGRTMIN+8
43) SIGRTMIN+9 44) SIGRTMIN+10    45) SIGRTMIN+11    46) SIGRTMIN+12    47) SIGRTMIN+13
48) SIGRTMIN+14    49) SIGRTMIN+15    50) SIGRTMAX-14 51) SIGRTMAX-13 52) SIGRTMAX-12
53) SIGRTMAX-11 54) SIGRTMAX-10 55) SIGRTMAX-9  56) SIGRTMAX-8  57) SIGRTMAX-7
58) SIGRTMAX-6  59) SIGRTMAX-5  60) SIGRTMAX-4  61) SIGRTMAX-3  62) SIGRTMAX-2
63) SIGRTMAX-1  64) SIGRTMAX

• 1~31为标准信号

• 34~64为RT, real time实时信号

• 函数原型如下

void (*signal(int signum, void (*func) (int))) (int);typedef void (*sighandler_t)(int);
sighandler_t signal(int signum, sighandler_t handler);

• 例子如下

# include <stdio.h>
# include <stdlib.h>
# include <signal.h>
# include <unistd.h>int main() {int i;for (i = 0; i < 10; i++) {write(1, "*", 1);sleep(1);}exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./star
***^C

^c会发送SIGINT信号

• 第二个例子

# include <stdio.h>
# include <stdlib.h>
# include <signal.h>
# include <unistd.h>int main() {int i;signal(SIGINT, SIG_IGN);for (i = 0; i < 10; i++) {write(1, "*", 1);sleep(1);}exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./star
***^C*******

现在这样确实^c打断不了了, 因为忽略了SIGINT信号

• 第三个例子

# include <stdio.h>
# include <stdlib.h>
# include <signal.h>
# include <unistd.h>static void int_handler(int s) {write(1, "!", 1);
}int main() {int i;signal(SIGINT, int_handler);for (i = 0; i < 10; i++) {write(1, "*", 1);sleep(1);}exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./star
**^C!**^C!*^C!*^C!*^C!*^C!**ubuntu@k8s-wolf-minion-47:~$

• 信号会打断阻塞的系统调用
  • 如下例, 一直按住ctrl +C, 不到10s, 程序就结束了
  • open阻塞, read阻塞, 都会被signal打断, 需要判断errno的返回是否为EINTR, 若是则忽略(因为是被打断了), 否则才是真的出错了

# include <stdio.h>
# include <stdlib.h>
# include <signal.h>
# include <unistd.h>static void int_handler(int s) {write(1, "!", 1);
}int main() {int i;signal(SIGINT, int_handler);for (i = 0; i < 10; i++) {write(1, "*", 1);sleep(1);}exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./star
*^C!*^C!*^C!*^C!*^C!*^C!*^C!*^C!*^C!*^C!ubuntu@k8s-wolf-minion-47:~$ ^C

生成core dumped

[u16.04开启coredump并设置core文件的产生位置]

(https://blog.csdn.net/u013937038/article/details/106757765/)

如何利用core dump + gdb

编译的时候带上gcc -g, 然后ulimit -c设置core size, 然后/etc/sysctl.conf设置core_pattern和core_uses_pid参数, 如果有段错误就会生成core文件, 然后用gdb调试可以找到时哪一行的问题, 注意一般core文件会很大

#include <stdio.h>
int main(void) {int *a = NULL;*a = 0x1;return 0;
}

ubuntu@k8s-wolf-minion-47:~$ ./getline_code /tmp/out
Segmentation faultubuntu@k8s-wolf-minion-47:~$ ulimit -a | grep 'core file size'
core file size          (blocks, -c) 0ubuntu@k8s-wolf-minion-47:~$ ulimit -c 10240
ubuntu@k8s-wolf-minion-47:~$ ulimit -a | grep 'core file size'
core file size          (blocks, -c) 10240root@k8s-wolf-minion-47:/home/ubuntu# gcc a.c -o a -g
root@k8s-wolf-minion-47:~# ./a
Segmentation fault (core dumped)
root@k8s-wolf-minion-47:~# ls
a  a.c  drop_cluster.sh  platform_0815-release-c1f7aa9-20200814  ssh_20200914root@k8s-wolf-minion-47:/home/ubuntu# ll /devo/null
total 116
drwxr-xr-x 2 root root   4096 Feb 23 00:52 ./
drwxr-xr-x 3 root root   4096 Feb 23 00:34 ../
-rw------- 1 root root 249856 Feb 23 00:52 core_a_1614013093_161183root@k8s-wolf-minion-47:/home/ubuntu# gdb a
(gdb) core-file /devo/null/core_a_1614013093_161183
[New LWP 161183]
Core was generated by `./a'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x00000000004004e6 in main () at a.c:4
4     *a = 0x1;

信号的行为不可靠与可重入函数

因为信号是未知的突发情况, 而不是人为写的逻辑代码, 所以是内核布置的现场, 每次可能不一样, 可能第一次调用还未结束就进行了第二次调用, 所以unix引入了可重入函数解决此问题

所有的系统调用都是可重入的, 部分库函数(如memcpy)是可重入的, 部分库函数(如rand, 因为rand的第n次调用的结果依赖于第n-1次的调用)是不可重入的, 一般函数末尾加_r(如rand_r, localtime_r, 尤其是返回值是指针的函数, 很容易是不可重入的, 那么人为指定指针地址就变成可重入的了)的就是可重入的.

信号的响应过程

• 信号从收到到响应有一个不可避免的延迟:(eg下图)
  • 每个进程有2个位图, 一个是mask bitmap, 一个是pending bitmap
  • 进程收到信号后只是把pending bitmap的某信号位置位, 此时进程并不知道收到信号了
  • 当进程遇到中断后, 进程带着 现场 做压栈(把用户态的栈压入内核态), 等待CPU的分时调度
  • 当等待到CPU分时调度时, 进程才有机会计算信号(即用进程内的mask bitmap &&按位与 pending bitmap), 计算后进程才知道收到了SIGINT信号, 此时进程调用SIGINT信号绑定的int_handler()函数处理’

 mask 32bit          pending 32bit+----------+       +---------+
|          |       |         |
|     1    |       |   0     |
+----------+       +---------|
|    1     |       |  1 seted when recv SIGINT
|          |       |         |
+----------+       +---------+
|          |       |    0    |
|    1     |       |         |
+----------+       +---------+
|          |       |         |
|    1     |       |    0    |
+----------+       +---------+
|          |       |    0    |
|    1     |       |         |
+----------+       +---------+
|    1     |       |         |
|          |       |         |        calc: mask bitmap && pending bitmap
|          |       |    0    |
+----------+       +---------+                       ^                                            main    int_handler
|          |       |         |                       |
|     1    |       |    0    |       user space      |                                               +      +
|          |       |         |                       |                                             * |      |
+----------+       +---------+        +---------------------------------------------+              * |      ||                                             * |      |kernel space    |                                             * |      ||       thread saved stack                    * |      ||    +----------+                               |     !||    |          |                             * |      ||    +----------+                               |      |+    |   addr   +-----------------------------> |      |+----------+                               |      ||          |                               v      v+----------+*****^C!**

• 信号响应后(如下图)
  • 把mask bitmap 对应位置位为0(表示正在执行信号处理函数int_handler, 防止此int_handler被后续其他信号处理函数打扰), pending bitmap 对应位置位为0(表示正在或已执行信号处理函数int_handler)
  • 把压入kernel space的thread saved stack中运行到的addr由main函数某行的指针替换为int_handler函数的第一行的指针, 然后执行int_handler, 此时界面打印!
  • 执行完int_handler()后, 把kernel space的thread saved stack中运行到的addr再替换回main函数, 并把mask bitmap的对应位置位为1(表示已执行完信号处理函数, 可以执行新的后续其他信号处理函数了), 然后接着执行main函数后面的部分, 打印后续的***

                mask 32bit          pending 32bit+----------+       +---------+|          |       |         ||     1    |       |   0     |+----------+       +---------+
unset when processing sig handler |  0 unset when proessing sig handler|    0     |       |         ++----------+       +---------+|          |       |    0    ||    1     |       |         |+----------+       +---------+|          |       |         ||    1     |       |    0    |+----------+       +---------+|          |       |    0    ||    1     |       |         |+----------+       +---------+|    1     |       |         ||          |       |         |        calc: mask bitmap && pending bitmap|          |       |    0    |+----------+       +---------+                       ^                                            main    int_handler|          |       |         |                       ||     1    |       |    0    |       user space      |                                               +      +<----------+|          |       |         |                       |                                             * |      |           |+----------+       +---------+        +---------------------------------------------+              * |      |           ||                                             * |      |           |kernel space    |                                             * |      |           ||       thread saved stack                    * |      |           ||    +----------+                               |     !|           ||    |          |                             * |      |           ||    +----------+                               |      |           |+    |   addr   +----------------+              |      |           |+----------+                |              |      |           ||          |                |              v      v           |+----------+                |                                 ||            *****^C!**           ||                                 ||                                 |+---------------------------------+switch stack pointer to signal handler function

信号是从kernel回user的路上, 被响应的

信号被IGNORE内部的原理就是把mask bitmap中对应的位置0实现的

• 单信号响应过程
  • M P
  • 1 0 还未收到信号
  • 1 1 收到信号后
  • 0 0 开始执行信号
  • 1 0 执行完信号
• 多信号响应过程
  • M P
  • 1 0 还未收到信号
  • 1 1 收到信号后
  • 0 0 开始执行信号
  • 0 1 此时又来了N多个SIGINT信号, 则Pending为又被set了N多次, 故pending为变为1了N多次
  • 1 1 执行完信号变为11, 此时发现又有信号来了则继续执行
  • 0 0 开始执行信号
  • 1 0 执行完信号

如何忽略掉一个信号?

• 标准信号为什么要丢失?
  • 因为是一个pending的位图, 如果来1w次相同的SIGINT信号, 还是只能在SIGINT位置位1次1

常用函数

• kill() 发信号

• Raise() 自己给自己发

• alarm() 闹钟计时发信号

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>int main() {alarm(5);while(1);exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./alarm
Alarm clock
ubuntu@k8s-wolf-minion-47:~$ time ./alarm
Alarm clockreal 0m5.002s
user    0m5.000s
sys 0m0.000s

流量控制实验mycat(漏桶)

# include <stdio.h>
# include <stdlib.h>
# include <errno.h>
# include <sys/types.h>
# include <sys/uio.h>
# include <unistd.h>
# include <fcntl.h>
# include <signal.h># define CPS 10
# define BUFSIZE CPS // 速度, 即缓冲区大小static volatile int loop = 0;
static void alrm_handler (int s) {alarm(1);loop = 1;
}int main(int argc, char **argv) {int sfd, dfd = 1; // dfd 为写向stdoutif (argc < 2) {fprintf(stderr, "Usage: 2 args...");exit(1);}signal(SIGALRM, alrm_handler);alarm(1);do {sfd = open(argv[1], O_RDONLY);if (sfd < 0) {if (errno != EINTR) {perror("open");exit(1);}}} while(sfd < 0);char buf[BUFSIZE];int readed = 0, writted = 0;while(1) {while(!loop)pause();loop = 0;while (readed = read(sfd, buf, BUFSIZE) < 0) {if (errno == EINTR)continue;perror("read fail");break;}if (readed == 0) {fprintf(stdout, "all readed done");break;}int pos = 0;int towrite = readed;while(towrite > 0) {writted = write(dfd, buf + pos, towrite);if (writted < 0) {if (errno == EINTR)continue;perror("write fail");exit(1);}towrite -= writted;pos += writted;}sleep(1);}close(sfd);close(dfd);exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./slowcat /etc/services
# Network services, Internet style
#
# Note that it is presently the p

令牌桶

第一个简化版

# include <stdio.h>
# include <stdlib.h>
# include <errno.h>
# include <sys/types.h>
# include <sys/uio.h>
# include <unistd.h>
# include <fcntl.h>
# include <signal.h># define CPS 10
# define BUFSIZE CPS // 速度, 即缓冲区大小
# define BURST 100static volatile int token = 0;
static void alrm_handler (int s) {alarm(1);token++;if (token > BURST)token = BURST;
}int main(int argc, char **argv) {int sfd, dfd = 1; // dfd 为写向stdoutif (argc < 2) {fprintf(stderr, "Usage: 2 args...");exit(1);}signal(SIGALRM, alrm_handler);alarm(1);do {sfd = open(argv[1], O_RDONLY);if (sfd < 0) {if (errno != EINTR) {perror("open");exit(1);}}} while(sfd < 0);char buf[BUFSIZE];int readed = 0, writted = 0;while(1) {while(token <= 0)pause();tolen --;while (readed = read(sfd, buf, BUFSIZE) < 0) {if (errno == EINTR)continue;perror("read fail");break;}if (readed == 0) {fprintf(stdout, "all readed done");break;}int pos = 0;int towrite = readed;while(towrite > 0) {writted = write(dfd, buf + pos, towrite);if (writted < 0) {if (errno == EINTR)continue;perror("write fail");exit(1);}towrite -= writted;pos += writted;}sleep(1);}close(sfd);close(dfd);exit(0);
}

封装为库

• 把slowcat.c封装为一个库 https://www.cnblogs.com/muzihuan/p/5304757.html

     +----+----+----+----+----+----+ job[1024]|  * | *  |    |    |    |    ||    |    |    |    |    |    |+-+--+-+--+----+----+----+----+^    ^|    ||    +----------------+
+--------+-+                 +----------+
|          |                 |  cps=50  |
| cps=10   |                 |          |
+----------+                 +----------+
|          |                 |          |
|          |                 |          |
| burst=100|                 |  burst=5000
|          |                 |          |
+----------+                 +----------+
|          |                 |          |
| token+=cps                 |  token+=cps
|          |                 |          |
+----------+                 +----------+

• main.c

# include <stdio.h>
# include <stdlib.h>
# include <errno.h>
# include <sys/types.h>
# include <sys/uio.h>
# include <unistd.h>
# include <fcntl.h>
# include <signal.h>
# include <string.h>
# include "mytbf.h"# define CPS 10 // 速度, 即缓冲区大小
# define BUFSIZE 1024
# define BURST 100static volatile sig_atomic_t token = 0; // 对该值的改变(如++, --, +2), 会保证为原子操作
static void alrm_handler (int s) {alarm(1);token++;if (token > BURST)token = BURST;
}int main(int argc, char **argv) {int sfd, dfd = 1; // dfd 为写向stdoutprintf("before sfd:%d", sfd);if (argc < 2) {fprintf(stderr, "Usage: 2 args...");exit(1);}mytbf_t *tbf = mytbf_init(CPS, BURST);if (tbf == NULL) {fprintf(stderr, "mytbf_init failed");exit(1);}do {sfd = open(argv[1], O_RDONLY);if (sfd < 0) {if (errno != EINTR) {perror("open");exit(1);}}} while(sfd < 0);printf("after sfd:%d", sfd);char buf[BUFSIZE];int readed = 0, writted = 0;while(1) {int got_token_size = mytbf_fetchtoken(tbf, BUFSIZE);if (got_token_size < 0) {fprintf(stderr, "mytbf_ftchtoken(): %s\n", strerror(-got_token_size));exit(1);}while ((readed = read(sfd, buf, got_token_size)) < 0) {if (errno == EINTR)continue;perror("read fail");break;}if (readed == 0) {printf("read 0");break;}printf("read some");// 归还用不完的tokenif (got_token_size - readed > 0) {mytbf_returntoken(tbf, got_token_size - readed);}int pos = 0;int towrite = readed;while(towrite > 0) {writted = write(dfd, buf + pos, towrite);if (writted < 0) {if (errno == EINTR)continue;perror("write fail");exit(1);}towrite -= writted;pos += writted;}sleep(1);}close(sfd);mytbf_destroy(tbf);exit(0);
}

• makefile

all:mytbf
mytbf: main.o mytbf.ogcc $^ -o $@clean:rm -rf *.o mytbf

• mytbf.c

# include <stdio.h>
# include <unistd.h>
# include <stdlib.h>
# include <errno.h>
# include <signal.h>
# include "mytbf.h"typedef void (*sighandler_t)(int);static struct mytbf_st* job[MYTBF_MAX];
static int inited = 0;
static sighandler_t alrm_handler_save;struct mytbf_st {int cps;int burst;int token;int pos;
};static void alrm_handler(int s) {int i = 0;alarm(1);for (i = 0; i< MYTBF_MAX; i++) {if (job[i] != NULL) {job[i]->token += job[i]->cps;if (job[i]->token > job[i]->burst)job[i]->token = job[i]->burst;}}
}static void module_unload(void) {signal(SIGALRM, alrm_handler_save);alarm(0);int i = 0;for (i = 0; i < MYTBF_MAX; i++)free(job[i]);
}static void module_load(void) {alrm_handler_save = signal(SIGALRM, alrm_handler);alarm(1);atexit(module_unload); // 卸载模块 只被调用一次
}static int get_free_pos(void) {int i = 0;for ( i = 0; i < MYTBF_MAX; i++) {if (job[i] == NULL)return i;}return -1;
}mytbf_t *mytbf_init(int cps, int burst) {struct mytbf_st *me;if (!inited) {module_load();inited = 1;}int pos = get_free_pos();if (pos < 0)return NULL;me = malloc(sizeof(*me));if (me == NULL) {return NULL;}me->token = 0;me->cps = cps;me->burst = burst;me->pos = pos;job[pos] = me;return me;
}static int min(int a, int b) {if (a < b)return a;return b;
}int mytbf_fetchtoken(mytbf_t *ptr , int size){struct mytbf_st *me = ptr; // 空指针强转: 相当于interface{}的reflectif (size <= 0)return -EINVAL;while(me->token <= 0) // 无token则阻塞, 当有token时会发signal, 则pause()的阻塞会被打破pause();int n = min(me->token, size);me->token -= n;return n;
}int mytbf_returntoken(mytbf_t *ptr, int size) {struct mytbf_st *me = ptr;if (size <= 0)return -EINVAL;me->token += size;if (me->token > me->burst)me->token = me->burst;return size;
}int mytbf_destroy(mytbf_t *ptr) {struct mytbf_st *me = ptr;job[me->pos] = NULL;free(ptr);return 0;
}

• mytbf.h

# ifndef MYTBF_H__
# define MYTBF_H__# define MYTBF_MAX 1024
typedef void mytbf_t;mytbf_t *mytbf_init(int cps, int burst);int mytbf_fetchtoken(mytbf_t *, int);int mytbf_returntoken(mytbf_t *, int);int mytbf_destroy(mytbf_t *);# endif

• make

ubuntu@k8s-wolf-minion-47:~/signal$ make clean
rm -rf *.o mytbfubuntu@k8s-wolf-minion-47:~/signal$ make
cc    -c -o main.o main.c
cc    -c -o mytbf.o mytbf.c
gcc main.o mytbf.o -o mytbfubuntu@k8s-wolf-minion-47:~/signal$ ./mytbf /etc/services
# Network services, Internet style
#
# Note that it is prese^C

使用单一计时器实现计数

• 使用单一计时器实现计数, 5s后调函数f1(“aaa”), 2s后调f2(“bbb”), 7s后调函数f1(“ccc”), 输出"Begin!End!..bbb…aaa…ccc…"

  • main.c

  # include <stdio.h>
  # include <stdlib.h>static f1(void *p) {printf("f1():%s\n", p);
  }
  static f2(void *p) {printf("f2():%s\n", p);
  }
  

  • anytimer.c

  # include <stdio.h>
  # include <stdlib.h>int main() {puts("Begin");int job1 = at_addjob(5, f1, "aaa");if (job1 < 0) {fprintf(stderr, "at_addjob():5s\n", strerror(-job1));exit(1);}int job2 = at_addjob(2, f2, "aaa");if (job2 < 0) {fprintf(stderr, "at_addjob():5s\n", strerror(-job2));exit(1);}int job3 = at_addjob(7, f1, "aaa");if (job3 < 0) {fprintf(stderr, "at_addjob():5s\n", strerror(-job3));exit(1);}puts("End!");while(1) {write(1, "", 1);sleep(1);}exit(0);
  }
  

  • anytimer.h

  # ifndef ANYTIMER_H__
  # define ANYTIMER_H__
  typedef void at_jobfunc_t(void*);#define JOB_MAX 1024
  int at_addjob(int sec, at_jobfunc_t *jobp, void *arg);
  /**
  * return >= 0       成功, 返回任务ID
  *        == -EINVAL 失败, 参数非法
  *            == -ENOSPC 失败, 数组满
  *            == -ENOMEM 失败, 内存空间不足
  */int at_canceljob(int id);
  /**
  * return >= 0               成功, 指定任务已取消
  *        == -EINVAL       失败, 参数非法
  *            == -EBUSY        失败, 指定任务已完成
  *            == -ECANCELED  失败, 指定任务重复取消
  */int at_waitjob(int id);
  /**
  * return == 0                 成功, 指定任务成功释放
  *            == -EINVAL   失败, 参数非法
  */// at_pausejob();
  // at_resumejob();# endif
  

  • makefile

  all:anytimer
  mytbf: main.o mytbf.ogcc $^ -o $@clean:rm -rf *.o mytbf
  

• setitimer(), 可替换链式alarm() https://blog.csdn.net/liuguanghui1988/article/details/6987611

#include <unistd.h>
#include <signal.h>
#include <stdio.h>
#include <errno.h>
#include <sys/time.h>void alarm_handler(int sig)
{   printf("hello!--sig:%d\n",sig);
}int main()
{struct itimerval t;   /* 第一次执行前的时间间隔*/t.it_value.tv_usec = 0;   t.it_value.tv_sec = 3; /* 以后每间再次执行的时间间隔*/t.it_interval.tv_usec = 0;    t.it_interval.tv_sec = 1;    if( setitimer(ITIMER_REAL, &t, NULL) < 0 ){perror("settimer");        return -1;    }if (signal( SIGALRM, alarm_handler) == SIG_ERR) {perror("signal");return -1;}    while(1){        pause();    }return 0;

ubuntu@k8s-wolf-minion-47:~$ ./1
000hello!--sig:14
hello!--sig:14
hello!--sig:14
hello!--sig:14
hello!--sig:14
hello!--sig:14
hello!--sig:14
hello!--sig:14
hello!--sig:14
hello!--sig:14

• Pause()

• Abort(): 结束自己 并 产生core dumped文件

• system()

• sleep(): linux是用nanosleep封装的, 但某些平台使用alarm + pause封装的, 但多个alarm会是程序出错, 故一般不用sleep

  • 可用nanosleep, usleep代替

信号集

• 信号集类型 sigset_t

• sigemptyset()

• sigfillset()

• sigaddset()

• sigdelset()

• sigismember()

信号屏蔽字/pending集的处理

• sigprocmask();

# include <stdio.h>
# include <stdlib.h>
# include <signal.h>
# include <unistd.h>static void int_handler(int s) {write(1, "!", 1);
}int main() {int i, j;sigset_t set, saveset;signal(SIGINT, int_handler);sigemptyset(&set);sigaddset(&set, SIGINT);sigprocmask(SIG_UNBLOCK, &set, &saveset); // 记录进此模块之前的saveset值以便后续恢复, 并按此模块业务逻辑初始化set为取消屏蔽信号for (j = 0; i < 1000; j++) {sigprocmask(SIG_BLOCK, &set, NULL); // 屏蔽信号for (i = 0; i < 5; i++) {write(1, "*", 1);sleep(1);}write(1, "\n", 1);sigprocmask(SIG_UNBLOCK, &set, NULL); // 解除信号的屏蔽}sigprocmask(SIG_SETMASK, &saveset, NULL); // 出模块后, 恢复为进此模块之前的saveset值exit(0);
}

扩展

用sigsetjump代替setjump, 用sitlongjump代替longjump, 可以在设置jump点是保存mask信息, 返回时能恢复掩码信息, 才能使得可以在signal_handler中向外跳转.

sigsuspend()

• 原子操作, 用于做信号驱动程序

# include <stdio.h>
# include <stdlib.h>
# include <signal.h>
# include <unistd.h>static void int_handler(int s) {write(1, "!", 1);
}int main() {int i, j;sigset_t set, saveset, oset;signal(SIGINT, int_handler);sigemptyset(&set);sigaddset(&set, SIGINT);sigprocmask(SIG_UNBLOCK, &set, &saveset); // 记录进此模块之前的saveset值以便后续恢复, 并按此模块业务逻辑初始化set为取消屏蔽信号sigprocmask(SIG_BLOCK, &set, &oset); // 屏蔽信号, 记录到oldsetfor (j = 0; i < 1000; j++) {for (i = 0; i < 5; i++) {write(1, "*", 1);sleep(1);}write(1, "\n", 1);sigsuspend(&oset); // 原子操作: 解除屏蔽信号 + 恢复至oldset信号, 开始响应信号/*sigset_t tmpset;sigprocmask(SIG_UNBLOCK, &oset, &tmpset); // 解除信号的屏蔽pause();sigprocmask(SIG_UNBLOCK, &tmpset, NULL); // 解除信号的屏蔽*/}sigprocmask(SIG_SETMASK, &saveset, NULL); // 出模块后, 恢复为进此模块之前的saveset值exit(0);
}

sigaction()

• 用于替代signal(), 更高级
  • 多个信号都有相同的信号处理函数时, 响应一个信号时, 把其他信号先屏蔽
  • 除了拿到信号, 还能拿到信号的来源(si_code), 如来自内核态or用户态or信号队列来的…, 可以做更精细的信号控制(例如只响应来自内核态的信号, 而忽略用户态的信号)

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <sys/types.h>
#include <unistd.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <errno.h>
#include <syslog.h>
#include <signal.h>
#define FNAME "/tmp/out"static int daemonize() {pid_t pid = fork();if (pid < 0) {return -1;}if (pid > 0)  // parentexit(0);// ------以下为child------// 重定向stdin, stdout, stderrint fd = open("/dev/null", O_RDWR);if (fd < 0) {return -1;}dup2(fd, 0);dup2(fd, 1);dup2(fd, 2);if (fd > 2)close(fd);setsid();chdir("/");umask(0);// 如果不产生文件的话, 可以关掉umaskreturn 0;
}static FILE* fp; // 用于daemon_exit的关闭, 和main的打开
static void daemon_exit(void) {/*if (s == SIGINT)printf("SIGINT");else if (s == SIGTERM)printf("SIGTERM");else if (s == SIGTERM)printf("SIGTERM");*/fclose(fp);closelog();exit(0);
}int main() {struct sigaction sa;sa.sa_handler = daemon_exit;sigemptyset(sa.sa_mask);sigaddset(&sa.sa_mask, SIGINT);sigaddset(&sa.sa_mask, SIGQUIT);sigaddset(&sa.sa_mask, SIGTERM);sa.sa_flags = 0;sigaction(SIGINT, &sa, NULL);sigaction(SIGQUIT, &sa, NULL);sigaction(SIGTERM, &sa, NULL);//  signal(SIGNAL, daemon_exit);
//  signal(SIGQUIT, daemon_exit);
//  signal(SIGTERM, daemon_exit);openlog("mydaemon", LOG_PID, LOG_DAEMON);if (daemonize()) {syslog(LOG_ERR, "daemonize() failed!"); // 注意这里不要加\n, syslog自己控制格式exit(1);}syslog(LOG_INFO, "daemonize() seccessed!");fp = fopen(FNAME, "w");if (fp == NULL) {syslog(LOG_ERR, "fopen:%s", strerror(errno));exit(1);}syslog(LOG_INFO, "%s was opened.", FNAME);int i = 0;for ( i = 0; ; i++) {fprintf(fp, "%d", i);fflush(fp);syslog(LOG_DEBUG, "%d is printed.", i);sleep(1);}exit(0);
}

setitimer()

• 用于替代alarm(), 更高级

• 改版后的mytbf如下

  • mytbf.c

# include <stdio.h>
# include <unistd.h>
# include <stdlib.h>
# include <errno.h>
# include <signal.h>
# include <sys/time.h>
# include "mytbf.h"// typedef void (*sighandler_t)(int);static struct mytbf_st* job[MYTBF_MAX];
static int inited = 0;
// static sighandler_t alrm_handler_save;
static struct sigaction alrm_sa_save;struct mytbf_st {int cps;int burst;int token;int pos;
};static void alrm_action(int s, siginfo_t *infop, void *unuesd) {int i = 0;// alarm(1);if(infop->si_code != SI_KERNEL) {return;}for (i = 0; i< MYTBF_MAX; i++) {if (job[i] != NULL) {job[i]->token += job[i]->cps;if (job[i]->token > job[i]->burst)job[i]->token = job[i]->burst;}}
}static void module_unload(void) {//  signal(SIGALRM, alrm_handler_save);
//  alarm(0);sigaction(SIGALRM, &alrm_sa_save, NULL);struct itimerval itv;itv.it_interval.tv_sec = 0;itv.it_interval.tv_usec = 0;itv.it_value.tv_sec = 0;itv.it_value.tv_usec = 1;setitimer(ITIMER_REAL, &itv, NULL);int i = 0;for (i = 0; i < MYTBF_MAX; i++)free(job[i]);
}static void module_load(void) {//  alrm_handler_save = signal(SIGALRM, alrm_handler);
//  alarm(1);struct sigaction sa;sa.sa_sigaction = alrm_action;sigemptyset(&sa.sa_mask);sa.sa_flags = SA_SIGINFO;sigaction(SIGALRM, &sa, &alrm_sa_save);struct itimerval itv;itv.it_interval.tv_sec = 1;itv.it_interval.tv_usec = 0;itv.it_value.tv_sec = 1;itv.it_value.tv_usec = 1;setitimer(ITIMER_REAL, &itv, NULL);atexit(module_unload); // 卸载模块 只被调用一次
}static int get_free_pos(void) {int i = 0;for ( i = 0; i < MYTBF_MAX; i++) {if (job[i] == NULL)return i;}return -1;
}mytbf_t *mytbf_init(int cps, int burst) {struct mytbf_st *me;if (!inited) {module_load();inited = 1;}int pos = get_free_pos();if (pos < 0)return NULL;me = malloc(sizeof(*me));if (me == NULL) {return NULL;}me->token = 0;me->cps = cps;me->burst = burst;me->pos = pos;job[pos] = me;return me;
}static int min(int a, int b) {if (a < b)return a;return b;
}int mytbf_fetchtoken(mytbf_t *ptr , int size){struct mytbf_st *me = ptr; // 空指针强转: 相当于interface{}的reflectif (size <= 0)return -EINVAL;while(me->token <= 0) // 无token则阻塞, 当有token时会发signal, 则pause()的阻塞会被打破pause();int n = min(me->token, size);me->token -= n;return n;
}int mytbf_returntoken(mytbf_t *ptr, int size) {struct mytbf_st *me = ptr;if (size <= 0)return -EINVAL;me->token += size;if (me->token > me->burst)me->token = me->burst;return size;
}int mytbf_destroy(mytbf_t *ptr) {struct mytbf_st *me = ptr;job[me->pos] = NULL;free(ptr);return 0;
}

ubuntu@k8s-wolf-minion-47:~/signal_sa$ ./mytbf /etc/services
# Network services, Internet style
#
# Note that it is presently the policy of I# 另一个终端, 执行, 因为来自用户态, mytbf不会响应来自用户态的信号, 故不会干扰流量控制(流控)算法
root@k8s-wolf-minion-47:~# while true; do kill -ALRM 140749; done

实时信号

• 为了解决标准信号的问题(如未定义行为)

  • 标准信号就是不排队, 就是会丢失; 而实时信号会排队则不会丢失

    • 排队上显示ulimit -a 的 pending signal值

    root@k8s-wolf-minion-47:~# ulimit -a | grep pending
    pending signals                 (-i) 257077
    

kill -l
ubuntu@k8s-wolf-minion-47:~$ kill -l1) SIGHUP   2) SIGINT   3) SIGQUIT  4) SIGILL   5) SIGTRAP6) SIGABRT    7) SIGBUS   8) SIGFPE   9) SIGKILL 10) SIGUSR1
11) SIGSEGV 12) SIGUSR2 13) SIGPIPE 14) SIGALRM 15) SIGTERM
16) SIGSTKFLT   17) SIGCHLD 18) SIGCONT 19) SIGSTOP 20) SIGTSTP
21) SIGTTIN 22) SIGTTOU 23) SIGURG  24) SIGXCPU 25) SIGXFSZ
26) SIGVTALRM   27) SIGPROF 28) SIGWINCH    29) SIGIO   30) SIGPWR
31) SIGSYS  34) SIGRTMIN    35) SIGRTMIN+1 36) SIGRTMIN+2 37) SIGRTMIN+3
38) SIGRTMIN+4 39) SIGRTMIN+5 40) SIGRTMIN+6 41) SIGRTMIN+7 42) SIGRTMIN+8
43) SIGRTMIN+9 44) SIGRTMIN+10    45) SIGRTMIN+11    46) SIGRTMIN+12    47) SIGRTMIN+13
48) SIGRTMIN+14    49) SIGRTMIN+15    50) SIGRTMAX-14 51) SIGRTMAX-13 52) SIGRTMAX-12
53) SIGRTMAX-11 54) SIGRTMAX-10 55) SIGRTMAX-9  56) SIGRTMAX-8  57) SIGRTMAX-7
58) SIGRTMAX-6  59) SIGRTMAX-5  60) SIGRTMAX-4  61) SIGRTMAX-3  62) SIGRTMAX-2
63) SIGRTMAX-1  64) SIGRTMAXvim /usr/include/bits/.h

# include <stdio.h>
# include <stdlib.h>
# include <signal.h>
# include <unistd.h>
# define MYRTSIG (SIGRTMIN+6) // 挑选的一个实时信号static void mysig_handler(int s) {write(1, "!", 1);
}int main() {int i, j;sigset_t set, saveset, oset;signal(MYRTSIG, mysig_handler);sigemptyset(&set);sigaddset(&set, SIGINT);sigprocmask(SIG_UNBLOCK, &set, &saveset); // 记录进此模块之前的saveset值以便后续恢复, 并按此模块业务逻辑初始化set为取消屏蔽信号sigprocmask(SIG_BLOCK, &set, &oset); // 屏蔽信号, 记录到oldsetfor (j = 0; i < 1000; j++) {for (i = 0; i < 5; i++) {write(1, "*", 1);sleep(1);}write(1, "\n", 1);sigsuspend(&oset); // 原子操作: 解除屏蔽信号 + 恢复至oldset信号, 开始响应信号/*sigset_t tmpset;sigprocmask(SIG_UNBLOCK, &oset, &tmpset); // 解除信号的屏蔽pause();sigprocmask(SIG_UNBLOCK, &tmpset, NULL); // 解除信号的屏蔽*/}sigprocmask(SIG_SETMASK, &saveset, NULL); // 出模块后, 恢复为进此模块之前的saveset值exit(0);
}

ubuntu@k8s-wolf-minion-47:~$ ./susprt
*****root@k8s-wolf-minion-47:~# ps -ef | grep susprt
ubuntu   143479 140575  0 23:12 pts/7    00:00:00 ./susprt# 此时发多个实时信号
root@k8s-wolf-minion-47:~# kill -40 143479
root@k8s-wolf-minion-47:~# kill -40 143479
root@k8s-wolf-minion-47:~# kill -40 143479
root@k8s-wolf-minion-47:~# kill -40 143479
root@k8s-wolf-minion-47:~# kill -40 143479
root@k8s-wolf-minion-47:~# kill -40 143479
root@k8s-wolf-minion-47:~# kill -40 143479# 多个实时信号都没丢失
ubuntu@k8s-wolf-minion-47:~$ ./susprt
*****
!*****
!**!*!**
!*!*!***

• 多线程比信号好用得多, 先有标准再有事实
• 信号处理是用于多进程并发的, 信号处理函数尽量少(避免可重入), 先有事实再有标准

其他

CLion工程中只能有一个main函数 &&怎么同时编写多个main函数的C文件

cmake_minimum_required(VERSION 3.13)
project(address C)set (CMAKE_C_STANDARD 90)add_executable(main01 Chapter-01/1.4.3.c)

Unix环境编程中的apue.h和err_quit、err_sys问题

文件和目录

• ls

#include "../apue.3e/include/apue.h"
# include"error.c"#define BUFFSIZE 4096int main(int argc, char *argv[]) {int n;char buf[BUFFSIZE];while ((n = read(STDIN_FILENO, buf, BUFFSIZE)) > 0)if (write(STDOUT_FILENO, buf, n) != n)err_sys("write error");if (n < 0)err_sys("read error");exit(0);
}

输入和输出

有缓冲IO

• 读输入输出

#include "../apue.3e/include/apue.h"
# include"error.c"#define BUFFSIZE 4096int main(int argc, char *argv[]) {int n;char buf[BUFFSIZE];while ((n = read(STDIN_FILENO, buf, BUFFSIZE)) > 0)if (write(STDOUT_FILENO, buf, n) != n)err_sys("write error");if (n < 0)err_sys("read error");exit(0);
}

• 重定向: 控制台输入 -> 控制台输出

 ~/go/src/github.com/sonnary/apue/Chapter-01   master ●  gcc -o a 1.5.3.c~/go/src/github.com/sonnary/apue/Chapter-01   master ●  ./a
1
1
2
2
^C

• 重定向: 控制台输入 -> 文件输出

 ✘  ~/go/src/github.com/sonnary/apue/Chapter-01   master ●  ./a > data
123
456
^C✘  ~/go/src/github.com/sonnary/apue/Chapter-01   master ●  cat data
123
456

• 重定向: 文件输入 -> 文件输出

 ~/go/src/github.com/sonnary/apue/Chapter-01   master ●  cat > infile
indata
^C✘  ~/go/src/github.com/sonnary/apue/Chapter-01   master ●  cat infile
indata~/go/src/github.com/sonnary/apue/Chapter-01   master ●  ./a < infile > outfile~/go/src/github.com/sonnary/apue/Chapter-01   master ●  cat outfile
indata

无缓冲IO

#include "../apue.3e/include/apue.h"
# include "error.c"int main(int argc, char *argv[]) {int c;while ((c = getc(stdin)) != EOF)if (putc(c, stdout) == EOF)err_sys("output error");if (ferror(stdin))err_sys("input error");exit(0);
}

程序和进程

• 进程ID

#include "../apue.3e/include/apue.h"int main(int argc, char *argv[]) {printf("hello world from process ID %ld\n", (long)getpid());exit(0);
}

出错处理

#include "../apue.3e/include/apue.h"#include <errno.h>int main(int argc, char *argv[]) {fprintf(stderr, "EACCES: %s\n", strerror(EACCES));errno = ENOENT;perror(argv[0]);exit(0);
}EACCES: Permission denied
~/go/src/github.com/sonnary/apue/cmake-build-debug/main17: No such file or directory

用户标识

int main(int argc, char *argv[]) {printf("uid = %d, gid = %d\n", getuid(), getgid());exit(0);
}uid = 501, gid = 20✘  /etc  cat /etc/group | grep staff
staff:*:20:root

信号

#include "../apue.3e/include/apue.h"
#include <sys/wait.h>static void sig_int(int);int main(int argc, char *argv[]) {char buf[MAXLINE];pid_t pid;int status;printf("pid: %d\n", getpid());if (signal(SIGINT, sig_int) == SIG_ERR)err_sys("signal error");printf("%% ");while (fgets(buf, MAXLINE, stdin) != NULL) {if (buf[strlen(buf) - 1] == '\n')buf[strlen(buf) - 1] = 0;if ((pid = fork()) < 0) {err_sys("fork error");} else if (pid == 0) {execlp(buf, buf, (char *) 0);err_ret("couldn't execute: %s", buf);exit(127);}if ((pid = waitpid(pid, &status, 0)) < 0)err_sys("waitpid error");printf("%% ");}exit(0);
}void sig_int(int signo) {printf("interrupt\n%% ");
}

时间值

/etc  time cat 1>1
zsh: permission denied: 1
cat > 1  0.00s user 0.00s system 58% cpu 0.001 total✘  /etc  time rm 1
rm: 1: No such file or directory
rm 1  0.00s user 0.00s system 64% cpu 0.003 total

系统调用和库函数

awk -f 2.5.4.awk
用awk转换为c代码, 然后执行该c代码(2.5.4.c)

多线程实现并发

线程概念

一个正在运行的函数

posix线程时一套标准, 而不是实现; openmp线程时另一套线程标准

线程标识: pthread_t(标准只定义需要有此标识, 具体实现可以异构, 例如可能实现方式为int, struct, string…, linux下一般为int)

pthread_equal();

Pthread_self() 即获取线程号

线程的创建

• makefile: 用于gcc的编译和链接

CFLAGS+=-pthread
LDFLAGS+=-pthread

• pthread_create

# include <stdio.h>
# include <stdlib.h>
# include <pthread.h>static void *func(void *p) {puts("Thread is working!");pthread_exit(NULL);
}int main() {pthread_t tid;int err;puts("Begin");err = pthread_create(&tid, NULL, func, NULL);if (err) {fprintf(stderr, "pthread_create(): %s", strerror(err));exit(1);}puts("End");exit(0);
}

root@k8s-wolf-minion-47:~/thread# ./create1
Begin
End# 这是因为线程还没来得及被调度呢, 进程就结束了

• 线程的调度取决于调度器策略

  • 加上sleep(1)就展示出来了

  Begin
  End
  Thread is working!
  

线程的终止

• 有3种方式
  • 线程从启动例程返回, 返回值就是线程的退出码
  • 线程可以被同一进程中的其他线程取消
  • 线程调用pthread_exit()函数: 会清理
• pthread_join() 相当于 进程process中的wait(), 等待线程运行结束
  • 区别: 可以指定收哪个线程

# include <stdio.h>
# include <stdlib.h>
# include <pthread.h>static void *func(void *p) {puts("Thread is working!");pthread_exit(NULL);
}int main() {pthread_t tid;int err;puts("Begin");err = pthread_create(&tid, NULL, func, NULL);if (err) {fprintf(stderr, "pthread_create(): %s", strerror(err));exit(1);}pthread_join(tid, NULL);  puts("End");exit(0);
}

Begin
Thread is working!
End

线程清理

• pthread_cleanup_push();
• pthread_cleanup_pop(); 参数1代表执行钩子函数, 参数0代表不执行钩子函数, (如果pop()函数放到exit后面就默认当做1了)
• 有几个push()函数, 就要有几个pop()函数, 否则编译不过去
• 是用栈的方式做的

# include <stdio.h>
# include <stdlib.h>
# include <pthread.h>
# include <string.h>static void cleanup_func(void *p) {puts(p);
}static void *func(void *p) {puts("Thread is working!");pthread_cleanup_push(cleanup_func, "cleanup:1");pthread_cleanup_push(cleanup_func, "cleanup:2");pthread_cleanup_push(cleanup_func, "cleanup:3");puts("puts over");pthread_cleanup_pop(1);pthread_cleanup_pop(1);pthread_cleanup_pop(1);pthread_exit(NULL);
}int main() {pthread_t tid;int err;puts("Begin");err = pthread_create(&tid, NULL, func, NULL);if (err) {fprintf(stderr, "pthread_create(): %s", strerror(err));exit(1);}pthread_join(tid, NULL);  puts("End");exit(0);
}

root@k8s-wolf-minion-47:~/thread# make cleanup
cc -pthread  -pthread  cleanup.c   -o cleanup
root@k8s-wolf-minion-47:~/thread# ./cleanup
Begin
Thread is working!
puts over
cleanup:3
cleanup:2
cleanup:1
End

线程的取消选项

• 线程取消: pthread_cancel()
• 取消有2种状态: 允许和不允许
  • 允许取消
    • 异步cancel
    • 推迟cancel(默认的): 推迟至cancel点再响应
  • 不允许取消
• POSIX定义的cancel点都是可能引发阻塞的系统调用(如open(), read()等)
• pthread_setcancelstate(): 设置是否允许取消
• pthread_setcanceltype(): 设置取消方式
• pthread_testcancel(): 本函数什么都不做, 就是一个取消点

线程分离

• pthread_detach(): 已经被detach的线程不能再join回来, 因为已经被放弃了…

线程竞争

example

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# include <pthread.h>
# include <string.h>
# include <stdbool.h>
# define LEFT 30000000
# define RIGHT 30000200
# define THRNUM (RIGHT-LEFT+1)struct thr_arg_st {int n;
};static void *thr_prime(void *p) {int i, j;i = ((struct thr_arg_st *)p)->n;bool mark = 1;for (j = 2; j < i/2; j++) {if (i % j == 0) {mark = 0;break;}}if (mark) {printf("%d is a primer\n", i);}pthread_exit(p);
}int main() {int i;pthread_t tid[THRNUM];for (i = LEFT; i <= RIGHT; i++) {struct thr_arg_st *p = malloc(sizeof(*p));if (p == NULL) {perror("malloc");exit(1);}p->n = i;int err = pthread_create(tid+(i-LEFT), NULL, thr_prime, p);if (err) {fprintf(stderr, "pthread_create():%s\n", strerror(err));exit(1);}}void *ptr; // child thread send to main threadfor ( i = LEFT; i <= RIGHT; i++) {pthread_join(tid[i-LEFT], &ptr);free(ptr); // 尽量把malloc和free放在一个func或module中}exit(0);
}

root@k8s-wolf-minion-47:~/thread# ./primer_thread
30000037 is a primer
30000163 is a primer
30000041 is a primer
30000001 is a primer
30000133 is a primer
30000023 is a primer
30000059 is a primer
30000071 is a primer
30000137 is a primer
30000149 is a primer
30000193 is a primer
30000199 is a primer
30000049 is a primer
30000169 is a primer
30000083 is a primer
30000079 is a primer
30000109 is a primer
30000167 is a primer
root@k8s-wolf-minion-47:~/thread# ./primer_thread  | wc -l
18
root@k8s-wolf-minion-47:~/thread# ./primer_thread  | wc -l
18
root@k8s-wolf-minion-47:~/thread# ./primer_thread  | wc -l
18
root@k8s-wolf-minion-47:~/thread# ./primer_thread  | wc -l
18
root@k8s-wolf-minion-47:~/thread# ./primer_thread  | wc -l
18

• 非原子操作 就可能出现线程竞争, 且竞争后果未知(类比十字路口南北&&东西车流), 内核以线程(而非进程为调度单位), for循环i跑的很快, 而pthread_create有开销会慢, 所以一般会导致把200个RIGHT值赋给了200个thread, 传的值不对, 而且因为200个线程拿到的是同样的地址会导致线程竞争

• 32位系统, 未改系统参数情况下, 一个process最多创建300个thread

  • 这些threads, 共用代码段, 独立拥有栈, 栈默认8MB

  root@k8s-wolf-minion-47:~/thread# ulimit -a | grep 'stack size'
  stack size              (kbytes, -s) 8192
  

  • 32位系统, c进程的虚拟内核空间是4GB, 内核态是3GB~4GB的地址空间, 用户态是从0x8000到3GB的地址空间
  • 能创建的thread数, 就是3GB用户态大小/每个栈大小(默认8MB)
  • 64位系统, 进程虚拟内核空间是128TB, pid号比栈消耗的更快

竞争故障

# include <stdio.h>
# include <stdlib.h>
# include <pthread.h>
# define THRNUM 20
# define FNAME "/tmp/out"
# define LINESIZE 1024static void *thr_add(void *p) {FILE *fp = fopen(FNAME, "r+");if (fp == NULL) {perror("fopen()");exit(1);}char linebuf[LINESIZE];fgets(linebuf, LINESIZE, fp);fseek(fp, 0, SEEK_SET);fprintf(fp, "%d\n", atoi(linebuf)+1);fclose(fp);pthread_exit(NULL);
}int main() {int i, err;pthread_t tid[THRNUM];for (i = 0; i < THRNUM; i++) {err = pthread_create(tid+i, NULL, thr_add, NULL);if (err) {fprintf(stderr, "%d", err);exit(1);}}for( i = 0; i < THRNUM; i++) {pthread_join(tid[i], NULL);}exit(0);
}

ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
89
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
106
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
123
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
140
ubuntu@k8s-wolf-minion-47:~/c/thread$
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
155
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
171
ubuntu@k8s-wolf-minion-47:~

线程同步

• 线程同步: 用于解决线程竞争
• 互斥量: 加锁, 锁的是代码(其实锁住的就是pthread_mutex_t变量), 锁住的代码称为临界区

pthread_mutex_t类型
pthread_mutex_init();
pthread_mutex_lock();
pthread_mutex_lock();
pthread_mutex_trylock();
pthread_mutex_unlock();

# include <stdio.h>
# include <stdlib.h>
# include <pthread.h>
# define THRNUM 20
# define FNAME "/tmp/out"
# define LINESIZE 1024static pthread_mutex_t mut = PTHREAD_MUTEX_INITIALIZER;
static void *thr_add(void *p) {FILE *fp = fopen(FNAME, "r+");if (fp == NULL) {perror("fopen()");exit(1);}char linebuf[LINESIZE];pthread_mutex_lock(&mut);fgets(linebuf, LINESIZE, fp);fseek(fp, 0, SEEK_SET);fprintf(fp, "%d\n", atoi(linebuf)+1);fclose(fp);pthread_mutex_unlock(&mut);pthread_exit(NULL);
}int main() {int i, err;pthread_t tid[THRNUM];for (i = 0; i < THRNUM; i++) {err = pthread_create(tid+i, NULL, thr_add, NULL);if (err) {fprintf(stderr, "%d", err);exit(1);}}for( i = 0; i < THRNUM; i++) {pthread_join(tid[i], NULL);}pthread_mutex_destroy(&mut);exit(0);
}

ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
21
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
41
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
61
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
81
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
101
ubuntu@k8s-wolf-minion-47:~/c/thread$ ./add  && cat /tmp/out
121

• 4个线程, 先各自锁上自己, 然后把第一个a线程解锁
  • 之后while循环
    • a线程lock+a线程执行+b线程解锁
    • b线程lock+b线程执行+c线程解锁
    • c线程lock+c线程执行+d线程解锁
    • d线程lock+d线程执行+a线程解锁

# include <stdio.h>
# include <stdlib.h>
# include <pthread.h>
# include <unistd.h>
# include <string.h>
# define THRNUM 4
# define FNAME "/tmp/out"
# define LINESIZE 1024static pthread_mutex_t mut[THRNUM];static int next(int n) {if (n+1 == THRNUM) return 0;return n+1;
}static void *thr_func(void *p) {int n = (int)p;int c = 'a' + n;while(1) {pthread_mutex_lock(mut + n);write(1, &c, 1);pthread_mutex_unlock(mut + next(n)); }pthread_exit(NULL);
}int main() {int i, err;pthread_t tid[THRNUM];for (i = 0; i < THRNUM; i++) {pthread_mutex_init(mut+i, NULL);pthread_mutex_lock(mut+i);err = pthread_create(tid+i, NULL, thr_func, (void *)i);if (err) {fprintf(stderr, "pthread_create(): %s\n", strerror(err));exit(1);}}pthread_mutex_unlock(mut+0);for( i = 0; i < THRNUM; i++) {pthread_join(tid[i], NULL);}exit(0);
}

abcdabcdabcdabcdabcdabcdabcdabcdabcdabcdab

线程池

# include <stdio.h>
# include <stdlib.h>
# include <unistd.h>
# include <pthread.h>
# include <string.h>
# include <stdbool.h>
# define LEFT 30000000
# define RIGHT 30000200
# define THRNUM 4static int num = 0;
static pthread_mutex_t mut_num = PTHREAD_MUTEX_INITILIZER;static void *thr_prime(void *p) {int i, j;i = ((struct thr_arg_st *)p)->n;bool mark = 1;for (j = 2; j < i/2; j++) {if (i % j == 0) {mark = 0;break;}}if (mark) {printf("%d is a primer\n", i);}pthread_exit(p);
}int main() {int i;pthread_t tid[THRNUM];for (i = 0; i <= THRNUM; i++) {int err = pthread_create(tid+i, NULL, thr_prime, (void *)i);if (err) {fprintf(stderr, "pthread_create():%s\n", strerror(err));exit(1);}}for for ( i = 0; i <= THRNUM; i++)pthread_join(tid[i], &ptr);pthread_mutex_destroy(&mut_num);  exit(0);
}

线程版令牌桶

• mytbf.h

# define MYTBF_MAX 20

• mytbf.c

# include <pthread.h>
# include "mytbf.h"static struct mytbf_st * job[MYTBF_MAX];
static pthread_mutex_t mut_job = PTHREAD_MUTEX_INITIALLIZER;
static pthread_once_t init_once = PTHREAD_ONCE_INIT;struct mytbf_st {int cps;int burst;int token;int pos;pthread_mutex_t mut;
};staitc void *thr_alrm(void *p) {while(1) {pthread_mutex_lock(&mut_job);for(i = 0; i < MYTBF_MAX; i++) {if(job[i] != NULL) {pthread_mutex_lock(&job[i]->mut);job[i]->token += job[i]->cps;if (job[i]->token > job[i]->burst) {job[i]->token = job[i]->burst;}pthread_mutex_unlock(&job[i]->mut);}}pthread_mutex_unlock(&mut_job);sleep(1);}
}static void module_unload(void) {int i;pthread_cancel(tid_alrm);pthread_join(tid_alrm, NULL);for(i = 0; i < MUTBF_MAX; i++) {if(job[i] != NULL) {mytbf_destroy(job[i]);}}pthread_mutex_destroy(&mut_job);
}static void module_load(void) {int err = pthread_create(&tid_alrm, NULL, thr_alrm, NULL);if (err) {fprintf(stderr, "pthread_create():%s\n", strerror(err));exit(1);}atexit(module_unload);
}static int get_free_pos_unlocked(void) {int i;for(i = 0; i < MYTBF_MAX; i++) {if(job[i] == NULL) {return i;}}return -1;
}mytbf_t *mytbf_init(int cps, int burst) {struct mytbf_st *me;int pos;pthread_once(&init_once, module_load);me = malloc(sizeof(*me));if (me == NULL) {return NULL;}me->token = 0;me->cps = cps;me->burst=burst;pthread_mutex_init(&me->mut, NULL);pthread_mutex_lock(&mut_job);pos = get_free_pos_unlocked();if (pos < 0) {pthread_mutex_unlock(&mut_job);free(me);return NULL;}me->pos = pos;job[pos] = me;pthread_mutex_unlock(&mut_job);return me;
}static int min(int a, int b) {if (a < b) {return a;}return b;
}int mytbf_fetchtoken(mytbf_t *ptr, int size) {struct mytbf_st *me = ptr;int n;if(size <= 0) {return -EINVAL;}pthread_mutex_lock(&me->mut);while(me->token <= 0) {pthread_mutex_unlock(&me->mut);sched_yield();pthread_mutex_lock(&me->mut);}n = min(me->token, size);me->token = n;pthread_mutex_unlock(&me->mut);return n;
}int mytbf_returntoken(mytbf_t *ptr, int size) {struct mytbf_st *me = ptr;if(size <= 0) {return -EINVAL;}pthread_mutex_lock(&me->mut);me->token += size;if (me->token > me->burst) {me->token = me=>burst;}pthread_mutex_unlock(&me->mut);return size;
}int mytbf_destroy(mytbf_t *ptr) {struct mytbf_st *me = ptr;pthread_mutex_lock(&mut_job);job[me->pos] = NULL;pthread_mutex_unlock(&mut_job);pthread_mutex_destroy(&me->mut);free(ptr);return 0;
}

• 使用

./mytbf /etc/services

条件变量

pthread_cond_init
pthread_cond_destroy
pthread_cond_broadcast 如果对多个线程都会有影响(惊群)
pthread_cond_signal 任意一个
pthread_cond_wait 阻塞等
pthread_cond_timedwait 有条件的等

• mytbf.c中以下函数有更改

# include <pthread.h>
# include "mytbf.h"static struct mytbf_st * job[MYTBF_MAX];
static pthread_mutex_t mut_job = PTHREAD_MUTEX_INITIALLIZER;
static pthread_once_t init_once = PTHREAD_ONCE_INIT;struct mytbf_st {int cps;int burst;int token;int pos;pthread_mutex_t mut;pthread_cond_t cond;
};staitc void *thr_alrm(void *p) {while(1) {pthread_mutex_lock(&mut_job);for(i = 0; i < MYTBF_MAX; i++) {if(job[i] != NULL) {pthread_mutex_lock(&job[i]->mut);job[i]->token += job[i]->cps;if (job[i]->token > job[i]->burst) {job[i]->token = job[i]->burst;}// 因为是给第i个加的, 所以通知时传的参数是: 第i个的消息// 如果数组共有20个令牌, 同时由20个人在等待, 此时只给第1个令牌`加值了`, 需要给所有的人都做通知, 让他们都来抢pthread_cond_broadcast(&job[i]->cond);pthread_mutex_unlock(&job[i]->mut);}}pthread_mutex_unlock(&mut_job);sleep(1);}
}static void module_unload(void) {int i;pthread_cancel(tid_alrm);pthread_join(tid_alrm, NULL);for(i = 0; i < MUTBF_MAX; i++) {if(job[i] != NULL) {mytbf_destroy(job[i]);}}pthread_mutex_destroy(&mut_job);
}static void module_load(void) {int err = pthread_create(&tid_alrm, NULL, thr_alrm, NULL);if (err) {fprintf(stderr, "pthread_create():%s\n", strerror(err));exit(1);}atexit(module_unload);
}static int get_free_pos_unlocked(void) {int i;for(i = 0; i < MYTBF_MAX; i++) {if(job[i] == NULL) {return i;}}return -1;
}mytbf_t *mytbf_init(int cps, int burst) {struct mytbf_st *me;int pos;pthread_once(&init_once, module_load);me = malloc(sizeof(*me));if (me == NULL) {return NULL;}me->token = 0;me->cps = cps;me->burst=burst;pthread_mutex_init(&me->mut, NULL);pthread_cond_init(&me->cond, NULL);pthread_mutex_lock(&mut_job);pos = get_free_pos_unlocked();if (pos < 0) {pthread_mutex_unlock(&mut_job);free(me);return NULL;}me->pos = pos;job[pos] = me;pthread_mutex_unlock(&mut_job);return me;
}static int min(int a, int b) {if (a < b) {return a;}return b;
}int mytbf_fetchtoken(mytbf_t *ptr, int size) {struct mytbf_st *me = ptr;int n;if(size <= 0) {return -EINVAL;}pthread_mutex_lock(&me->mut);while(me->token <= 0) {// 当me->token <= 0时, 先解锁me->mut, 再等待me->cond的通知(通过cond_broadcast或cond_signal产生通知)// 当收到了me->cond的通知时, 抢锁me->mut, 然后判断当me->token <= 0是否成立, 若当me->token >0则退出while循环, 反之继续在while循环中pthread_cond_wait(&me->cond, &me->mut);/*pthread_mutex_unlock(&me->mut);sched_yield();pthread_mutex_lock(&me->mut);*/}n = min(me->token, size);me->token = n;pthread_mutex_unlock(&me->mut);return n;
}int mytbf_returntoken(mytbf_t *ptr, int size) {struct mytbf_st *me = ptr;if(size <= 0) {return -EINVAL;}pthread_mutex_lock(&me->mut);me->token += size;if (me->token > me->burst) {me->token = me=>burst;}pthread_cond_broadcast(&me->cond);pthread_mutex_unlock(&me->mut);return size;
}int mytbf_destroy(mytbf_t *ptr) {struct mytbf_st *me = ptr;pthread_mutex_lock(&mut_job);job[me->pos] = NULL;pthread_mutex_unlock(&mut_job);pthread_mutex_destroy(&me->mut);pthread_cond_destroy(&me->cond);free(ptr);return 0;
}

条件变量的例子

• 筛选质数的例子
  子线程a叫醒子线程b, b叫c, c叫d

信号量

• 比互斥量, 更有共享的味道
• 哲学家吃面问题(就餐问题)
• 数字类型, 有上限
• 用互斥量+条件变量: 实现信号量
• 参考primer0.c

读写锁

• 读锁: 共享锁
• 写锁: 互斥锁

线程属性

pthread_attr_init();
pthread_attr_destroy();
pthread_attr_setstacksize();

线程同步的属性

• 互斥量属性

pthread_mutexattr_init();
pthread_mutexattr_destroy();
pthread_mutexattr_getpshared();
pthread_mutexattr_setpshared();
clone();
pthread_mutexattr_gettype();
pthread_mutexattr_settype();

• 条件变量属性

pthread_condattr_init();
pthread_condattr_destroy();

• 读写锁属性

线程重入

• 一个库需要支持多线程并发, 若不支持则需要在名字上体现
• 多线程中的IO

getchar()为线程安全
getchar_unlocked()为非线程安全

• 线程和信号

pthread_sigmask();
sigwait();
pthread_kill();

• 线程与fork
  posix原语定义: 若父进程有多个线程, 则fork出的子进程只有1个线程

线程标准

posix现场标准, openmp线程标准(https://www.openmp.org)

非阻塞IO

数据中继

• 目标

                  数据加工1+--------------------------->文件L                                      文件R
+--------+                                +----------+
|        |                                |          |
|        |                                |          |
|        |                                |          |
|        |                                |          |
+--------+                                +----------+<---------------------------+数据加工2

• 方式
  • 方式1: 读L, 写R, 读R, 写L: 会阻塞在读L步骤上
  • 方式2: 任务1: 读L写R; 任务2: 读R写L

有限状态机

• 用有限状态机解决复杂的数据中继问题

-rw-r--r-- 1 root root    0 Mar 24 19:01 main.c
-rw-r--r-- 1 root root   86 Mar 24 19:00 makefile
-rw-r--r-- 1 root root 2762 Mar 24 18:50 relayer.c
-rw-r--r-- 1 root root    0 Mar 24 19:01 relayer.h

【apue】UNIX环境高级编程 超详细介绍相关推荐

1. APUE Unix环境高级编程读书笔记

   .. 转载于:https://www.cnblogs.com/solitrarychen/p/5407536.html

2. (三) 一起学 Unix 环境高级编程 (APUE) 之 文件和目录

   . . . . . 目录 (一) 一起学 Unix 环境高级编程 (APUE) 之 标准IO (二) 一起学 Unix 环境高级编程 (APUE) 之 文件 IO (三) 一起学 Unix 环境高级编 ...

3. apue.h头文件（UNIX环境高级编程）

   相信非常多初学<UNIX环境高级编程>的朋友都会遇到一个问题,执行里面的实例(download: http://www.apuebook.com/ )时就出现故障,提示 "错误: ...

4. Unix——学习《Unix环境高级编程》找不到“apue.h”方法

   在运行<UNIX环境高级编程>中的程序时会遇到apue.h包头找不到的情况,这是作者为了方便程序书写封闭了一些功能函数和错误处理等.在http://www.apuebook.com中可以下 ...

5. UNIX环境高级编程（第三版）关于apue.h的用法

   UNIX环境高级编程(第三版)中的例子用到apue.h这个头文件,但是书里面写的地址已经不能访问. 经过一番查找之后,找到如下解决方案: 1.到www.apuebook.com上下载第2版的源码,也可 ...

6. Unix环境高级编程中的apue.h配置

   本文解释Unix环境高级编程的环境搭建中,apue.h如何使用安装及问题解决. 使用的是<Unix环境高级编程>第三版: 系统为CentOS 7.4,64位(Linux系统都可参考): 文 ...

7. Unix环境高级编程 笔记

   Unix环境高级编程(第二版)学习笔记 这是一次较长时间的整理,然而跳跃了一些章节和很多知识点,仍然是很不完善很不全面的. 前言 操作系统某些问题 严格意义上,可将操作系统定义为一种软件,它控制计算机 ...

8. UNIX环境高级编程笔记之文件I/O

   一.总结 在写之前,先唠几句,<UNIX环境高级编程>,简称APUE,这本书简直是本神书,像我这种小白,基本上每看完一章都是"哇"这种很吃惊的表情.其实大概三年前,那会 ...

9. unix环境高级编程基础知识之第二篇（3）

   看了unix环境高级编程第三章,把代码也都自己敲了一遍,另主要讲解了一些IO函数,read/write/fseek/fcntl:这里主要是c函数,比较容易,看多了就熟悉了.对fcntl函数讲解比较到位 ...

10. 《Unix环境高级编程》学习笔记：从点到面

    以前在课堂上学习过<Unix初级教程(第四版)>,对于Unix有了一点了解.由于以后使用的需要,要对它进行比较深入的学习,为此需要阅读不少的书籍,这本<Unix环境高级编程>便 ...

最新文章

1. java 跳转 oop_Java基础之：OOP——接口
2. 2021年最有用的数据清洗 Python 库
3. eclipse hibernate配置文件（*.hbm.xml）加上自动提示功能
4. mysql存储加速_mysql存储过程加速
5. Web推流，Webrtc网页推流rtmp在教学直播录制中应用
6. docker常用命令，安装常用实例，一步式安装mysql
7. indesign如何画弧线_彩铅画入门教程，如何给独角兽设计一款好发型
8. Error和Exception（异常）
9. 在CMakeLists.txt文件中包含Eigen
10. xshell连接虚拟机（后续）
11. PIL-读取与保存图片
12. paip.提升用户体验---搜索功能设计
13. 【数学建模】CUMCM-2014B 创意平板折叠桌 解题思路整理
14. SpringCloud 分布式日志采集方案
15. 交换机 路由器模拟仿真软件命令
16. UE4：转换成VR项目
17. JAVA实习生/应届生面试笔试题（一）
18. 如何使用pyinstaller将Python的代码打包
19. JAVA原生图片处理：Graphics2D海报升级(二)
20. C++程序设计实践里面石头剪刀布版王者农药实例

热门文章

1. JavaScript高级编程之树节点搜索过滤的设计和实现
2. java反编译工具jad安装
3. excel mysql乱码_excel打开是乱码的解法方法
4. 图书管理系统源代码＜Java＞（带文字讲解）(javase)
5. JAVA基础编程50道练习题
6. JAVA删除文件或者文件夹
7. matlab2014A安装教程及下载
8. BAT54C 二极管是如何工作的？
9. 在线SAS统计分析软件使用简介（2019年最新修订）
10. 帆软注册文件_注册信息- FineReport帮助文档|报表开发|报表使用|学习教程