#define _GNU_SOURCE
#include <sys/types.h>
#include <sys/wait.h>
#include <stdio.h>
#include <sched.h>
#include <signal.h>
#include <unistd.h>
#include <sys/capability.h>

#define STACK_SIZE (1024 * 1024)

static char child_stack[STACK_SIZE];
char* const child_args[] = {
    "/bin/bash",
    NULL
};

void set_uid_map(pid_t pid, int inside_id, int outside_id, int length){
    char path[256];
    sprintf(path, "/proc/%d/uid_map", getpid());
    FILE* uid_map = fopen(path, "w");
    fprintf(uid_map, "%d %d %d", inside_id, outside_id, length);
    fclose(uid_map);
}

void set_gid_map(pid_t pid, int inside_id, int outside_id, int length){
    char path[256];
    sprintf(path, "/proc/%d/gid_map", getpid());
    FILE* gid_map = fopen(path, "w");
    //printf("*****gid---Pid: %d *****\n", getpid());
    fprintf(gid_map, "%d %d %d", inside_id, outside_id, length);
    fclose(gid_map);
}

int child_main(void* args){
    cap_t caps;
    printf("In the Child process !\n");
    //sethostname("NewNameSpace", 12);
    //cap_t caps;
    printf("*****Pid: %d *****\n", getpid());
    set_uid_map(getpid(), 0, 1000, 1);
    set_gid_map(getpid(), 0, 1000, 1);
    printf("eUID = %ld; eGID = %ld; \n\n", (long)geteuid(), (long)getegid());
    caps = cap_get_proc();
    printf("capabilities: %s\n", cap_to_text(caps, NULL));
    execv(child_args[0], child_args);
    return 1;
}

int main(){
    printf("Start: \n");
    //int child_pid = clone(child_main, child_stack + STACK_SIZE, CLONE_NEWUSER | CLONE_NEWNET | CLONE_NEWNS | CLONE_NEWPID | CLONE_NEWIPC | CLONE_NEWUTS | SIGCHLD, NULL);
    int child_pid = clone(child_main, child_stack + STACK_SIZE, CLONE_NEWUSER | SIGCHLD, NULL);
    waitpid(child_pid, NULL, 0);
    printf("Already Exit \n");
    return 0;
}

代碼為 user namespace 實現隔離，但是我在Ubuntu 16.04 GCC 5.4 上編譯時出錯

gcc userns.c -Wall -lcap -o userns.o && ./userns.o

Start:
In the Child process !
*****Pid: 3582 *****
*****gid---Pid: 3582 *****
eUID = 0; eGID = 65534;

capabilities: = cap_chown,cap_dac_override,cap_dac_read_search,cap_fowner,cap_fsetid,cap_kill,cap_setgid,cap_setuid,cap_setpcap,cap_linux_immutable,cap_net_bind_service,cap_net_broadcast,cap_net_admin,cap_net_raw,cap_ipc_lock,cap_ipc_owner,cap_sys_module,cap_sys_rawio,cap_sys_chroot,cap_sys_ptrace,cap_sys_pacct,cap_sys_admin,cap_sys_boot,cap_sys_nice,cap_sys_resource,cap_sys_time,cap_sys_tty_config,cap_mknod,cap_lease,cap_audit_write,cap_audit_control,cap_setfcap,cap_mac_override,cap_mac_admin,cap_syslog,cap_wake_alarm,cap_block_suspend,37+ep

eGID 本應該為 0 的，這里是 65534 ， 我去 /proc/[pid]/gid_map 里看了一下，什么也沒已有，如圖所示

2017-07-16 16-38-33屏幕截圖.png

不知道哪里出問題了?
參考資料：Docker 背后的內核知識：命名空間資源隔離