hyperStart source code

The files under src directory is the codes of hyperStart,
use autogen.sh && configure && make to generate hyper-initrd.img,
this image will be used to start hyper instance.

Signed-off-by: Gao feng <feng@hyper.sh>

.gitignore

@@ -0,0 +1,37 @@
+*#*#
+*.#*#
+*.a
+*.o
+*.orig
+*.rej
+.#*
+.git
+.deps
+init
+build/hyper_daemon
+build/hyper-initrd.img
+build/root/*
+build/daemon
+build/.*
+src/.*
+Makefile
+Makefile.in
+cscope.files
+cscope.in.out
+cscope.out
+cscope.po.out
+stamp-h
+stamp-h.in
+stamp-h1
+/config.h
+/config.h.in
+/config.log
+/config.status
+/configure
+/aclocal.m4
+/autoscan.log
+/autom4te.cache
+/compile
+/depcomp
+/install-sh
+/missing

Makefile.am

@@ -0,0 +1 @@
+SUBDIRS=src build

README.md

@@ -0,0 +1,2 @@
+The init task for hyper.
+exec ./autogen.sh to compile hyperinit

autogen.sh

@@ -0,0 +1,64 @@
+#!/bin/sh
+
+srcdir=`dirname $0`
+test -z "$srcdir" && srcidr=.
+
+cd $srcdir
+DIE=0
+
+test -f src/init.c || {
+	echo
+	echo "You must run this script in the top-level hyperint drectory."
+	echo
+	DIE=1
+}
+
+(autoconf --version) < /dev/null > /dev/null 2>&1 || {
+	echo
+	echo "You must have autoconf installed to generate the hyperinit."
+	echo
+	DIE=1
+}
+
+(autoheader --version) < /dev/null > /dev/null 2>&1 || {
+	echo
+	echo "You must have autoheader installed to generate the hypernit."
+	echo
+	DIE=1
+}
+
+(automake --version) < /dev/null > /dev/null 2>&1 || {
+	echo
+	echo "You must have automake installed to generate the hypernit."
+	echo
+	DIE=1
+}
+(autoreconf --version) < /dev/null > /dev/null 2>&1 || {
+	echo
+	echo "You must have autoreconf installed to generate the hypernit."
+	echo
+	DIE=1
+}
+
+if test "$DIE" -eq 1; then
+	exit 1
+fi
+
+echo
+echo "Generating build-system with:"
+echo "  aclocal:  $(aclocal --version | head -1)"
+echo "  autoconf:  $(autoconf --version | head -1)"
+echo "  autoheader:  $(autoheader --version | head -1)"
+echo "  automake:  $(automake --version | head -1)"
+echo
+
+rm -rf autom4te.cache
+
+aclocal
+autoconf
+autoheader
+automake --add-missing
+
+echo
+echo "type '$srcdir/configure' and 'make' to compile hyperinit."
+echo

build/Makefile.am

@@ -0,0 +1,4 @@
+bin_PROGRAMS=hyper_daemon
+hyper_daemon_SOURCES=hyper_daemon.c ../src/net.c
+all-local:
+	bash ./make-initrd.sh

build/hyper_daemon.c

@@ -0,0 +1,168 @@
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <sys/socket.h>
+#include <linux/un.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <string.h>
+
+#include "../src/hyper.h"
+#include "../src/net.h"
+
+int test_sendmsg(int fd, unsigned int type, unsigned int len, char *message)
+{
+	uint8_t buf[4096];
+
+
+	/* send hyper info to guest */
+	hyper_set_be32(buf, type);
+
+	len += 8;
+
+	fprintf(stdout, "type is %u, len is %u\n", type, len);
+	hyper_set_be32(buf + 4, len);
+
+	if (message)
+		memcpy(buf + 8, message, len - 8);
+
+	if (write(fd, buf, len) != len) {
+		fprintf(stderr, "send SETDVM MESSAGE failed\n");
+		return -1;
+	}
+
+	fprintf(stdout, "finish sending\n");
+	return 0;
+}
+
+int test_sendmsg_from_file(int fd, unsigned int type, char *file)
+{
+	int file_fd = open(file, O_RDONLY);
+	uint8_t buf[4096];
+	unsigned int len = 0;
+
+	if (file_fd < 0) {
+		perror("fail to open file");
+		return -1;
+	}
+
+	while (len < 4096) {
+		int size = read(file_fd, buf + len, sizeof(buf) - len);
+
+		fprintf(stdout, "read %d data\n", size);
+		if (size < 0) {
+			perror("fail to read data");
+			return -1;
+		} else if (size == 0) {
+			/* buf[len] = '\0';*/
+			fprintf(stdout, "get buf %s, call sendmsg\n", buf);
+			test_sendmsg(fd, type, len, (char *)buf);
+			break;
+		}
+
+		len += size;
+	}
+
+	close(file_fd);
+
+	if (read(fd, buf, 8) != 8) {
+		fprintf(stderr, "read response failed\n");
+		return -1;
+	}
+
+	type = hyper_get_be32(buf);
+	if (type != ACK) {
+		fprintf(stderr, "incorrect type %d\n", type);
+		return -1;
+	}
+
+	return 0;
+}
+
+int main(int argc, char *argv[])
+{
+	int sock, fd = -1;
+	struct sockaddr_un addr;
+	uint8_t buf[8];
+	unsigned int type;
+
+	unlink("/tmp/hyper.sock");
+
+	sock = socket(AF_UNIX, SOCK_STREAM|SOCK_CLOEXEC, 0);
+	if (sock == -1) {
+		perror("create unix socket failed");
+		return -1;
+	}
+
+	addr.sun_family = AF_UNIX;
+	strncpy(addr.sun_path, "/tmp/hyper.sock", UNIX_PATH_MAX);
+	addr.sun_path[UNIX_PATH_MAX - 1] = '\0';
+
+	if (bind(sock, ((struct sockaddr *) &addr), sizeof(addr)) == -1) {
+		perror("bind failed");
+		goto out;
+	}
+
+	if (listen(sock, 1) == -1) {
+		perror("bind failed");
+		goto out;
+	}
+
+	while (fd == -1)
+		fd = accept4(sock, NULL, NULL, SOCK_CLOEXEC);
+
+	fprintf(stdout, "connected\n");
+	if (read(fd, buf, 8) != 8) {
+		fprintf(stderr, "read failed, buf %s\n", buf);
+		goto out1;
+	}
+
+	type = hyper_get_be32(buf);
+	fprintf(stdout, "get type %d\n", type);
+
+	if (type != READY) {
+		fprintf(stderr, "incorrect type %d\n", type);
+		goto out1;
+	}
+
+	fprintf(stdout, "get length %d\n", hyper_get_be32(buf + 4));
+
+	/* test_sendmsg_from_file(fd, SETDVM, "sethyper.json"); */
+	if (test_sendmsg_from_file(fd, STARTPOD, "startpod.json") < 0) {
+		fprintf(stderr, "send startpod message failed\n");
+		goto out1;
+	}
+
+	if (test_sendmsg_from_file(fd, EXECCMD, "execcmd.json") < 0) {
+		fprintf(stderr, "send execcmd message failed\n");
+		goto out1;
+	}
+
+	if (test_sendmsg(fd, STOPPOD, 0, NULL) < 0) {
+		fprintf(stderr, "send stoppod message failed\n");
+		goto out1;
+	}
+
+	if (read(fd, buf, 8) != 8) {
+		fprintf(stderr, "read response failed\n");
+		return -1;
+	}
+
+	type = hyper_get_be32(buf);
+	if (type != ACK) {
+		fprintf(stderr, "incorrect type %d\n", type);
+		return -1;
+	}
+
+	if (read(fd, buf, 8) != 8) {
+		fprintf(stderr, "read response failed\n");
+		return -1;
+	}
+out1:
+	close(fd);
+out:
+	close(sock);
+
+	return 0;
+}

build/make-initrd.sh

@@ -0,0 +1,21 @@
+#!/bin/bash
+
+rm -rf root
+mkdir root
+
+cp ../src/init ./root
+
+ldd ./root/init | while read line
+do
+	arr=(${line// / })
+
+	for lib in ${arr[@]}
+	do
+		if [ "${lib:0:1}" = "/" ]; then
+			dir=root`dirname $lib`
+			mkdir -p "${dir}"
+			cp -f $lib $dir
+		fi
+	done
+done
+cd ./root && find . | cpio -H newc -o | gzip -9 > ../hyper-initrd.img

configure.ac

@@ -0,0 +1,67 @@
+#                                               -*- Autoconf -*-
+# Process this file with autoconf to produce a configure script.
+
+AC_PREREQ([2.69])
+AC_INIT([dvminit], [0.1], [list@getdvm.org])
+AM_INIT_AUTOMAKE([-Wall -Werror foreign subdir-objects])
+AC_CONFIG_SRCDIR([src/init.c])
+AC_CONFIG_HEADERS([config.h])
+
+# Checks for programs.
+AC_PROG_CC
+
+# Checks for libraries.
+
+# Checks for header files.
+AC_CHECK_HEADERS([arpa/inet.h fcntl.h limits.h stddef.h stdint.h stdlib.h string.h sys/mount.h sys/socket.h unistd.h],
+[headers_found=yes],
+[headers_found=no])
+
+if test "x$headers_found" != "xyes"; then
+    AC_MSG_ERROR(Unable to find necessary headers)
+fi
+
+# Checks for typedefs, structures, and compiler characteristics.
+AC_C_INLINE
+AC_TYPE_PID_T
+AC_TYPE_SIZE_T
+AC_TYPE_UINT32_T
+AC_TYPE_UINT8_T
+
+# Checks for library functions.
+AC_FUNC_FORK
+AC_CHECK_FUNCS([dup2 memmove memset mkdir setenv socket strchr strdup strrchr strtoul], [fail=0], [fail=1])
+
+if test "$fail" = "1" ; then
+    AC_MSG_ERROR(Unable to find necessary functions)
+fi
+
+AC_ARG_WITH([ttys],
+            [AS_HELP_STRING([--with-ttys],
+                            [use pci serial device as container tty])],
+            [],[with_ttys=yes])
+
+if test "x$with_ttys" != "xno" ; then
+    AC_DEFINE_UNQUOTED([WITH_TTYS], 1, [where use pci serial device as container tty])
+fi
+AM_CONDITIONAL([WITH_TTYS], [test "x$with_ttys" != "xno"])
+
+AC_CONFIG_FILES([
+  Makefile
+  src/Makefile
+  build/Makefile
+])
+AC_OUTPUT
+
+AC_MSG_RESULT([
+	${PACKAGE}  ${VERSION}
+	prefix:		${prefix}
+
+	with-ttys:	${with_ttys}
+
+	compiler:          ${CC}
+	cflags:            ${CFLAGS}
+	suid cflags:       ${SUID_CFLAGS}
+	ldflags:           ${LDFLAGS}
+	suid ldflags:      ${SUID_LDFLAGS}
+])

src/Makefile.am

@@ -0,0 +1,3 @@
+AM_CFLAGS = -Wall
+bin_PROGRAMS=init
+init_SOURCES=init.c jsmn.c net.c util.c parse.c container.c exec.c event.c

src/container.c

@@ -0,0 +1,546 @@
+#define _GNU_SOURCE
+#include <stdio.h>
+#include <stdlib.h>
+#include <sched.h>
+#include <string.h>
+#include <sys/mount.h>
+#include <sys/types.h>
+#include <sys/socket.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <mntent.h>
+#include <signal.h>
+#include <errno.h>
+#include <fcntl.h>
+#include <dirent.h>
+
+#include "hyper.h"
+
+static int container_setup_env(struct hyper_container *container)
+{
+	int i;
+	struct env *env;
+
+	for (i = 0; i < container->envs_num; i++) {
+		env = &container->envs[i];
+
+		setenv(env->env, env->value, 1);
+	}
+
+	return 0;
+}
+
+static int container_setup_volume(struct hyper_container *container)
+{
+	int i;
+	char dev[512], path[512];
+	struct volume *vol;
+
+	for (i = 0; i < container->vols_num; i++) {
+		vol = &container->vols[i];
+
+		sprintf(dev, "/.oldroot/dev/%s", vol->device);
+		sprintf(path, "/tmp/%s", vol->mountpoint);
+		fprintf(stdout, "mount %s to %s\n", dev, vol->mountpoint);
+
+		if (hyper_mkdir(path) < 0 || hyper_mkdir(vol->mountpoint) < 0) {
+			fprintf(stdout, "mountpoint %s\n", vol->mountpoint);
+			perror("create volume dir failed");
+			continue;
+		}
+
+		if (mount(dev, path, vol->fstype, 0, NULL) < 0) {
+			perror("mount volume device faled");
+			continue;
+		}
+
+		if (mount(path, vol->mountpoint, NULL, MS_BIND, NULL) < 0) {
+			perror("mount volume device faled");
+			continue;
+		}
+
+		if (vol->readonly &&
+		    mount(path, vol->mountpoint, NULL, MS_BIND | MS_REMOUNT | MS_RDONLY, NULL) < 0)
+			perror("mount fsmap faled");
+
+		umount(path);
+	}
+
+	return 0;
+}
+
+static void container_unmount_oldroot(char *path)
+{
+	FILE *mtab;
+	struct mntent *mnt;
+	char *mntlist[128];
+	int i;
+	int n = 0;
+	char *filesys;
+
+	mtab = setmntent("/proc/mounts", "r");
+	if (mtab == NULL) {
+		fprintf(stderr, "cannot open /proc/mount");
+		return;
+	}
+
+	while (n < 128 && (mnt = getmntent(mtab))) {
+		if (strncmp(mnt->mnt_dir, path, strlen(path)))
+			continue;
+		mntlist[n++] = strdup(mnt->mnt_dir);
+	}
+
+	endmntent(mtab);
+
+	for (i = n - 1; i >= 0; i--) {
+		filesys = mntlist[i];
+		fprintf(stdout, "umount %s\n", filesys);
+		if (umount(mntlist[i]) < 0 && umount2(mntlist[i],
+			   MNT_DETACH) < 0) {
+			fprintf(stdout, "umount %s: %s failed\n",
+				filesys, strerror(errno));
+		}
+	}
+}
+
+static int container_setup_mount(struct hyper_container *container)
+{
+	int i, fd;
+	char src[512];
+	struct fsmap *map;
+
+	if (mount("proc", "/proc", "proc", 0, NULL) < 0 ||
+	    mount("sysfs", "/sys", "sysfs", 0, NULL) < 0 ||
+	    mount("devtmpfs", "/dev", "devtmpfs", 0, NULL) < 0) {
+		perror("mount basic filesystem for container failed");
+		return -1;
+	}
+
+	if (hyper_mkdir("/dev/pts") < 0) {
+		fprintf(stderr, "create /dev/pts failed\n");
+		return -1;
+	}
+
+	if (sprintf(src, "/.oldroot/tmp/hyper/%s/devpts", container->id) < 0) {
+		fprintf(stderr, "get container devpts failed\n");
+		return -1;
+	}
+
+	if (mount(src, "/dev/pts/", NULL, MS_BIND, NULL) < 0) {
+		perror("move pts to /dev/pts failed");
+		return -1;
+	}
+
+	unlink("/dev/ptmx");
+	link("/dev/pts/ptmx", "/dev/ptmx");
+
+	for (i = 0; i < container->maps_num; i++) {
+		struct stat st;
+
+		map = &container->maps[i];
+
+		sprintf(src, "/.oldroot/tmp/hyper/shared/%s", map->source);
+		fprintf(stdout, "mount %s to %s\n", src, map->path);
+
+		stat(src, &st);
+		if (st.st_mode & S_IFDIR) {
+			if (hyper_mkdir(map->path) < 0) {
+				perror("create map dir failed");
+				continue;
+			}
+		} else {
+			fd = open(map->path, O_CREAT|O_WRONLY, 0755);
+			if (fd < 0) {
+				perror("create map file failed");
+				continue;
+			}
+			close(fd);
+		}
+
+		if (mount(src, map->path, NULL, MS_BIND, NULL) < 0) {
+			perror("mount fsmap faled");
+			continue;
+		}
+
+		if (map->readonly == 0)
+			continue;
+
+		if (mount(src, map->path, NULL, MS_BIND | MS_REMOUNT | MS_RDONLY, NULL) < 0)
+			perror("mount fsmap faled");
+	}
+
+	return 0;
+}
+
+static int container_setup_workdir(struct hyper_container *container)
+{
+	if (container->workdir && chdir(container->workdir) < 0) {
+		perror("change work directory failed");
+		return -1;
+	}
+
+	return 0;
+}
+
+static int container_setup_tty(int fd, struct hyper_container *container)
+{
+	return hyper_dup_exec_tty(fd, &container->exec);
+}
+
+static int hyper_list_dir(char *path)
+{
+	struct dirent **list;
+	struct dirent *dir;
+	int i, num;
+
+	fprintf(stdout, "list %s\n", path);
+	num = scandir(path, &list, NULL, NULL);
+	if (num < 0) {
+		perror("scan path failed");
+		return -1;
+	}
+
+	for (i = 0; i < num; i++) {
+		dir = list[i];
+		fprintf(stdout, "%s get %s\n", path, dir->d_name);
+	}
+
+	free(list);
+	return 0;
+}
+
+static int hyper_rescan_scsi(void)
+{
+	struct dirent **list;
+	struct dirent *dir;
+	int fd = -1, i, num;
+	char path[256];
+
+	num = scandir("/sys/class/scsi_host/", &list, NULL, NULL);
+	if (num < 0) {
+		perror("scan /sys/calss/virtio-ports/ failed");
+		return -1;
+	}
+
+	memset(path, 0, sizeof(path));
+
+	for (i = 0; i < num; i++) {
+		dir = list[i];
+		if (dir->d_name[0] == '.')
+			continue;
+
+		if (snprintf(path, sizeof(path), "/sys/class/scsi_host/%s/scan",
+			     dir->d_name) < 0) {
+			fprintf(stderr, "get scsi host device %s path failed\n",
+				dir->d_name);
+			continue;
+		}
+
+		fprintf(stdout, "path %s\n", path);
+		fd = open(path, O_WRONLY);
+		if (fd < 0) {
+			perror("open path failed");
+			continue;
+		}
+
+		if (write(fd, "- - - \n", 7) < 0)
+			perror("write to scan failed");
+
+		close(fd);
+	}
+
+	fprintf(stdout, "finish scan scsi\n");
+	return 0;
+	free(list);
+}
+
+struct hyper_container_arg {
+	struct hyper_container	*c;
+	int			pipe[2];
+};
+
+static int hyper_container_init(void *data)
+{
+	struct hyper_container_arg *arg = data;
+	struct hyper_container *container = arg->c;
+	char root[512], oldroot[512];
+
+	fprintf(stdout, "%s in\n", __func__);
+	if (container->exec.argv == NULL) {
+		fprintf(stdout, "no cmd!\n");
+		goto fail;
+	}
+
+	if (hyper_rescan_scsi() < 0) {
+		fprintf(stdout, "rescan scsi failed\n");
+		goto fail;
+	}
+
+	if (container_setup_env(container) < 0) {
+		fprintf(stdout, "setup env failed\n");
+		goto fail;
+	}
+
+	if (mount("", "/", NULL, MS_SLAVE|MS_REC, NULL) < 0) {
+		perror("mount SLAVE failed");
+		goto fail;
+	}
+
+	if (mount("", "/", NULL, MS_PRIVATE|MS_REC, NULL) < 0) {
+		perror("mount PRIVATE failed");
+		goto fail;
+	}
+
+	sprintf(root, "/tmp/hyper/%s/root/", container->id);
+	if (hyper_mkdir(root) < 0) {
+		perror("make root directroy failed");
+		goto fail;
+	}
+
+	fprintf(stdout, "container root directory %s\n", root);
+
+	if (container->fstype) {
+		char dev[128];
+
+		sprintf(dev, "/dev/%s", container->image);
+		fprintf(stdout, "device %s\n", dev);
+
+		if (mount(dev, root, container->fstype, 0, NULL) < 0) {
+			perror("mount device failed");
+			goto fail;
+		}
+	} else {
+		char path[512];
+
+		sprintf(path, "/tmp/hyper/shared/%s/", container->image);
+		fprintf(stdout, "src directory %s\n", path);
+
+		if (mount(path, root, NULL, MS_BIND, NULL) < 0) {
+			perror("mount src dir failed");
+			goto fail;
+		}
+	}
+
+	fprintf(stdout, "root directory for container is %s/%s, init task %s\n",
+		root, container->rootfs, container->exec.argv[0]);
+
+	hyper_list_dir(root);
+	sprintf(oldroot, "%s/%s/.oldroot", root, container->rootfs);
+	if (hyper_mkdir(oldroot) < 0) {
+		perror("make oldroot directroy failed");
+		goto fail;
+	}
+
+	if (mount("/", oldroot, NULL, MS_BIND|MS_REC, NULL) < 0) {
+		perror("bind oldroot failed");
+		goto fail;
+	}
+	/* reuse oldroot array */
+	sprintf(oldroot, "%s/%s/", root, container->rootfs);
+	/* pivot_root won't work, see
+	 * Documention/filesystem/ramfs-rootfs-initramfs.txt */
+	chroot(oldroot);
+
+	chdir("/");
+
+	if (container_setup_volume(container) < 0) {
+		fprintf(stderr, "container sets up voulme failed\n");
+		goto fail;
+	}
+
+	if (container_setup_mount(container) < 0) {
+		fprintf(stderr, "container sets up mount ns failed\n");
+		goto fail;
+	}
+
+	if (container_setup_workdir(container) < 0) {
+		fprintf(stderr, "container sets up work directory failed\n");
+		goto fail;
+	}
+
+	container_unmount_oldroot("/.oldroot");
+
+	fflush(stdout);
+
+	if (container_setup_tty(arg->pipe[1], container) < 0) {
+		fprintf(stdout, "setup tty failed\n");
+		goto fail;
+	}
+
+	close(arg->pipe[0]);
+	close(arg->pipe[1]);
+
+	execvp(container->exec.argv[0], container->exec.argv);
+
+	_exit(-1);
+
+fail:
+	container->exec.code = -1;
+	hyper_send_type_block(arg->pipe[1], ERROR, 0);
+	_exit(-1);
+}
+
+int hyper_start_container(struct hyper_container *container)
+{
+	int stacksize = getpagesize() * 4;
+	void *stack = malloc(stacksize);
+	struct hyper_container_arg arg = {
+		.c = container,
+	};
+	int flags = CLONE_NEWNS | SIGCHLD;
+	uint32_t type;
+	int pid;
+
+	if (container->image == NULL || container->exec.argv == NULL) {
+		fprintf(stdout, "container root image %s, argv %p\n",
+			container->image, container->exec.argv);
+		goto fail;
+	}
+
+	if (socketpair(PF_UNIX, SOCK_STREAM, 0, arg.pipe) < 0) {
+		perror("create pipe between pod init execcmd failed");
+		goto fail;
+	}
+
+	pid = clone(hyper_container_init, stack + stacksize, flags, &arg);
+	free(stack);
+	if (pid < 0) {
+		perror("create child process failed");
+		goto fail;
+	}
+
+	/* wait for ready message */
+	if (hyper_get_type_block(arg.pipe[0], &type) < 0 || type != READY) {
+		fprintf(stdout, "wait for container started failed\n");
+		goto fail;
+	}
+
+	close(arg.pipe[0]);
+	close(arg.pipe[1]);
+
+	container->exec.pid = pid;
+	fprintf(stdout, "container %s init pid is %d\n", container->id, pid);
+	return 0;
+
+fail:
+	fprintf(stdout, "container %s init exit code %d\n", container->id, -1);
+	container->exec.code = -1;
+	return -1;
+}
+
+int hyper_start_containers(struct hyper_pod *pod)
+{
+	int i;
+
+	/* mount new proc directory */
+	if (umount("/proc") < 0) {
+		perror("umount proc filesystem failed\n");
+		return -1;
+	}
+
+	if (mount("proc", "/proc", "proc", 0, NULL) < 0) {
+		perror("mount proc filesystem failed\n");
+		return -1;
+	}
+
+	if (sethostname(pod->hostname, strlen(pod->hostname)) < 0) {
+		perror("set host name failed");
+		return -1;
+	}
+
+	for (i = 0; i < pod->c_num; i++)
+		hyper_start_container(&pod->c[i]);
+
+	return 0;
+}
+
+int hyper_restart_containers(struct hyper_pod *pod)
+{
+	int i;
+	struct hyper_container *c;
+
+	for (i = 0; i < pod->c_num; i++) {
+		c = &pod->c[i];
+
+		if (hyper_start_container(c) < 0) {
+			fprintf(stderr, "restart container %s failed\n", c->id);
+			hyper_send_type(pod->ctl.fd, ERROR);
+			return -1;
+		}
+	}
+
+	if (hyper_send_type(pod->ctl.fd, ACK) < 0)
+		return -1;
+
+	return 0;
+}
+
+struct hyper_container *hyper_find_container(struct hyper_pod *pod, char *id)
+{
+	int i;
+	struct hyper_container *container;
+
+	for (i = 0; i < pod->c_num; i++) {
+		container = &pod->c[i];
+
+		if (strlen(container->id) != strlen(id))
+			continue;
+
+		if (strncmp(container->id, id, strlen(id)))
+			continue;
+
+		return container;
+	}
+
+	return NULL;
+}
+
+void hyper_cleanup_container(struct hyper_pod *pod)
+{
+	int i, j;
+	struct hyper_container *c;
+	struct volume *vol;
+	struct env *env;
+	struct fsmap *map;
+	char root[512];
+
+	for (i = 0; i < pod->c_num; i++) {
+		c = &pod->c[i];
+
+		sprintf(root, "/tmp/hyper/%s/devpts/", c->id);
+		if (umount(root) < 0 && umount2(root, MNT_DETACH))
+			perror("umount devpts failed");
+
+		free(c->id);
+		free(c->rootfs);
+		free(c->image);
+		free(c->workdir);
+		free(c->fstype);
+
+		for (j = 0; j < c->vols_num; j++) {
+			vol = &(c->vols[j]);
+			free(vol->device);
+			free(vol->mountpoint);
+			free(vol->fstype);
+		}
+		free(c->vols);
+
+		for (j = 0; j < c->envs_num; j++) {
+			env = &(c->envs[j]);
+			free(env->env);
+			free(env->value);
+		}
+		free(c->envs);
+
+		for (j = 0; j < c->maps_num; j++) {
+			map = &(c->maps[j]);
+			free(map->source);
+			free(map->path);
+		}
+		free(c->maps);
+	}
+
+	free(pod->c);
+	pod->c = NULL;
+	pod->c_num = 0;
+}

src/container.h

@@ -0,0 +1,47 @@
+#ifndef _CONTAINER_H_
+#define _CONTAINER_H_
+
+#include "exec.h"
+
+struct env {
+	char	*env;
+	char	*value;
+};
+
+struct volume {
+	char	*device;
+	char	*mountpoint;
+	char	*fstype;
+	int	readonly;
+};
+
+struct fsmap {
+	char	*source;
+	char	*path;
+	int	readonly;
+};
+
+struct hyper_container {
+	char			*id;
+	char			*rootfs;
+	char			*image;
+	char			*workdir;
+	char			*fstype;
+	struct volume		*vols;
+	struct env		*envs;
+	struct fsmap		*maps;
+	int			vols_num;
+	int			envs_num;
+	int			maps_num;
+	uint32_t		code;
+	struct hyper_exec	exec;
+};
+
+struct hyper_pod;
+
+int hyper_start_containers(struct hyper_pod *pod);
+struct hyper_container *hyper_find_container(struct hyper_pod *pod, char *id);
+int hyper_restart_containers(struct hyper_pod *pod);
+void hyper_cleanup_container(struct hyper_pod *pod);
+
+#endif

src/event.c

@@ -0,0 +1,91 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+#include <string.h>
+#include <fcntl.h>
+#include <unistd.h>
+
+#include "util.h"
+#include "event.h"
+
+void hyper_reset_event(struct hyper_event *de)
+{
+	free(de->buf.data);
+	de->buf.data = NULL;
+	memset(de, 0, sizeof(*de));
+}
+
+int hyper_init_event(struct hyper_event *de,
+		   struct hyper_event_ops *ops,
+		   uint32_t size, int to, void *arg)
+{
+	struct hyper_buf *buf = &de->buf;
+
+	memset(buf, 0, sizeof(*buf));
+
+	de->ops		= ops;
+	de->ptr		= arg;
+	de->to		= to;
+	buf->size	= size;
+
+	if (size) {
+		buf->data = malloc(size);
+		if (buf->data == NULL) {
+			fprintf(stderr, "allocate data for event failed\n");
+			return -1;
+		}
+	}
+
+	return 0;
+}
+
+int hyper_add_event(int efd, struct hyper_event *de)
+{
+	struct epoll_event event = {
+		.events		= EPOLLIN,
+		.data.ptr	= de,
+	};
+
+	if (hyper_setfd_nonblock(de->fd) < 0) {
+		perror("set fd nonblock failed");
+		return -1;
+	}
+
+	fprintf(stdout, "%s add event fd %d, %p\n", __func__, de->fd, de->ops);
+
+	if (epoll_ctl(efd, EPOLL_CTL_ADD, de->fd, &event) < 0) {
+		perror("epoll_ctl fd failed");
+		return -1;
+	}
+
+	return 0;
+}
+
+void hyper_event_hup(struct hyper_event *de, int efd)
+{
+	if (epoll_ctl(efd, EPOLL_CTL_DEL, de->fd, NULL) < 0)
+		perror("epoll_ctl del epoll event failed");
+	close(de->fd);
+	hyper_reset_event(de);
+}
+
+int hyper_handle_event(int efd, struct epoll_event *event)
+{
+	struct hyper_event *de = event->data.ptr;
+
+	if (event->events & EPOLLHUP) {
+		fprintf(stdout, "%s event EPOLLHUP, de %p, fd %d, %p\n",
+			__func__, de, de->fd, de->ops);
+		if (de->ops->hup)
+			de->ops->hup(de, efd);
+		return 0;
+	} else if (event->events & EPOLLIN) {
+		return de->ops->read(de);
+	} else if (event->events & EPOLLERR) {
+		fprintf(stderr, "get epoll err of not epool in event\n");
+		return -1;
+	}
+
+	fprintf(stdout, "%s get unknown event %d\n", __func__, event->events);
+	return -1;
+}

src/event.h

@@ -0,0 +1,36 @@
+#ifndef _EVENT_H
+#define _EVENT_H
+
+#include <inttypes.h>
+#include <sys/epoll.h>
+
+struct hyper_event;
+
+struct hyper_event_ops {
+	int		(*read)(struct hyper_event *e);
+	int		(*getlen)(struct hyper_event *e, uint32_t *len);
+	int		(*handle)(struct hyper_event *e, uint32_t len);
+	void		(*hup)(struct hyper_event *e, int efd);
+};
+
+struct hyper_buf {
+	uint32_t		get;
+	uint32_t		size;
+	uint8_t			*data;
+};
+
+struct hyper_event {
+	int			fd;
+	int			to;
+	struct hyper_buf	buf;
+	struct hyper_event_ops	*ops;
+	void			*ptr;
+};
+
+int hyper_add_event(int efd, struct hyper_event *de);
+int hyper_init_event(struct hyper_event *de, struct hyper_event_ops *ops,
+		   uint32_t size, int to, void *arg);
+int hyper_handle_event(int efd, struct epoll_event *event);
+void hyper_reset_event(struct hyper_event *de);
+void hyper_event_hup(struct hyper_event *de, int efd);
+#endif

src/exec.c

@@ -0,0 +1,558 @@
+#define _GNU_SOURCE
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/epoll.h>
+#include <sys/ioctl.h>
+#include <sched.h>
+#include <errno.h>
+#include <string.h>
+#include <fcntl.h>
+#include <inttypes.h>
+
+#include "hyper.h"
+#include "util.h"
+#include "parse.h"
+
+static int pts_loop(struct hyper_event *de)
+{
+	int size = 0;
+	uint8_t buf[512];
+	uint8_t seq[12];
+	struct hyper_exec *exec = container_of(de, struct hyper_exec, e);
+
+	fprintf(stdout, "%s\n", __func__);
+	while (1) {
+		size = read(de->fd, buf, sizeof(buf));
+		if (size <= 0) {
+			if (errno == EINTR)
+				continue;
+			if (errno == EAGAIN || errno == EIO)
+				break;
+
+			perror("fail to read tty fd");
+			return -1;
+		}
+
+		hyper_set_be64(seq, exec->seq);
+		hyper_set_be32(seq + 8, size + 12);
+
+		if (hyper_send_data(de->to, seq, 12) < 0)
+			continue;
+
+		hyper_send_data(de->to, buf, size);
+	}
+
+	return 0;
+}
+
+struct hyper_event_ops pts_ops = {
+	.read		= pts_loop,
+	.hup		= hyper_event_hup,
+};
+
+int hyper_setup_exec_tty(struct hyper_exec *e)
+{
+	int unlock = 0;
+	char ptmx[512], path[512];
+
+	if (e->seq == 0)
+		return 0;
+
+	if (e->id) {
+		if (sprintf(path, "/tmp/hyper/%s/devpts/", e->id) < 0) {
+			fprintf(stderr, "get ptmx path failed\n");
+			return -1;
+		}
+	} else {
+		if (sprintf(path, "/dev/pts/") < 0) {
+			fprintf(stderr, "get ptmx path failed\n");
+			return -1;
+		}
+	}
+
+	if (sprintf(ptmx, "%s/ptmx", path) < 0) {
+		fprintf(stderr, "get ptmx path failed\n");
+		return -1;
+	}
+
+	e->e.fd = open(ptmx, O_RDWR | O_NOCTTY | O_NONBLOCK | O_CLOEXEC);
+	if (e->e.fd < 0) {
+		perror("open ptmx device for execcmd failed");
+		return -1;
+	}
+
+	if (ioctl(e->e.fd, TIOCSPTLCK, &unlock) < 0) {
+		perror("ioctl unlock ptmx device failed");
+		return -1;
+	}
+
+	if (ioctl(e->e.fd, TIOCGPTN, &e->ptyno) < 0) {
+		perror("ioctl get execcmd pty device failed");
+		return -1;
+	}
+
+	if (sprintf(ptmx, "%s/%d", path, e->ptyno) < 0) {
+		fprintf(stderr, "get ptmx path failed\n");
+		return -1;
+	}
+
+	e->pty = strdup(ptmx);
+	fprintf(stdout, "get pty device for exec %s\n", e->pty);
+
+	return 0;
+}
+
+int hyper_dup_exec_tty(int to, struct hyper_exec *e)
+{
+	int fd;
+	char pty[128];
+
+	setsid();
+
+	if (e->seq) {
+		if (sprintf(pty, "/dev/pts/%d", e->ptyno) < 0) {
+			perror("get pts device name failed");
+			return -1;
+		}
+	} else {
+		if (sprintf(pty, "/dev/null") < 0) {
+			perror("get pts device name failed");
+			return -1;
+		}
+	}
+
+	fprintf(stdout, "setup pty device %s for exec\n", pty);
+
+	fd = open(pty, O_RDWR | O_NOCTTY);
+	if (fd < 0) {
+		perror("open pty device for execcmd failed");
+		return -1;
+	}
+
+	if (ioctl(fd, TIOCSCTTY, NULL) < 0) {
+		perror("ioctl pty device for execcmd failed");
+		return -1;
+	}
+
+	if (hyper_send_type_block(to, READY, 0) < 0) {
+		fprintf(stderr, "send ready message to hyper init failed\n");
+		return -1;
+	}
+
+	fflush(stdout);
+
+	if (dup2(fd, STDIN_FILENO) < 0) {
+		perror("dup tty device to stdin failed");
+		close(fd);
+		return -1;
+	}
+
+	if (dup2(fd, STDOUT_FILENO) < 0) {
+		perror("dup tty device to stdout failed");
+		close(fd);
+		return -1;
+	}
+
+	if (dup2(fd, STDERR_FILENO) < 0) {
+		perror("dup tty device to stderr failed");
+		close(fd);
+		return -1;
+	}
+
+	close(fd);
+
+	return 0;
+}
+
+int hyper_exec_in_container(struct hyper_pod *pod,
+			    struct hyper_exec *exec)
+{
+	global_exec = exec;
+
+	if (hyper_send_type_block(ctl.ctl.fd, EXECCMD, 1) < 0) {
+		global_exec = NULL;
+		fprintf(stderr, "tell pod init EXECCMD failed\n");
+		return -1;
+	}
+
+	list_add_tail(&exec->list, &pod->ce_head);
+	if (exec->seq == 0)
+		return 0;
+
+	fprintf(stdout, "init container exec pts event %p, ops %p, fd %d\n",
+		&exec->e, &pts_ops, exec->e.fd);
+
+	if (hyper_init_event(&exec->e, &pts_ops, 0, ctl.tty.fd, NULL) < 0 ||
+	    hyper_add_event(ctl.efd, &exec->e) < 0) {
+		fprintf(stderr, "add pts master event failed\n");
+		return -1;
+	}
+
+	return 0;
+}
+
+int hyper_request_restart_containers(struct hyper_pod *pod)
+{
+	int i;
+	struct hyper_exec *exec;
+
+	pod->code = 0;
+	pod->remains = pod->c_num;
+
+	for (i = 0; i < pod->c_num; i++) {
+		exec = &pod->c[i].exec;
+
+		if (hyper_setup_exec_tty(exec) < 0) {
+			fprintf(stdout, "restart setup container tty failed\n");
+			return -1;
+		}
+	}
+
+	if (hyper_send_type_block(ctl.ctl.fd, RESTARTCONTAINER, 1) < 0) {
+		fprintf(stderr, "tell container init RESTARTCONTAINER failed\n");
+		return -1;
+	}
+
+	for (i = 0; i < pod->c_num; i++) {
+		exec = &pod->c[i].exec;
+
+		list_add_tail(&exec->list, &pod->ce_head);
+		if (exec->seq == 0)
+			continue;
+
+		if (hyper_init_event(&exec->e, &pts_ops, 0, ctl.tty.fd, NULL) < 0 ||
+		    hyper_add_event(ctl.efd, &exec->e) < 0) {
+			fprintf(stderr, "add pts master event failed\n");
+			return -1;
+		}
+	}
+
+	return 0;
+}
+
+int hyper_exec_cmd(char *json, int length)
+{
+	struct hyper_exec *exec;
+	struct hyper_pod *pod = &global_pod;
+	int pid, pipe[2];
+
+	fprintf(stdout, "call hyper_exec_cmd, json %s, len %d\n", json, length);
+
+	exec = hyper_parse_execcmd(json, length);
+	if (exec == NULL) {
+		fprintf(stderr, "parse exec cmd failed\n");
+		return -1;
+	}
+
+	if (exec->argv == NULL) {
+		fprintf(stderr, "cmd is %p, seq %" PRIu64 ", container %s\n",
+			exec->argv, exec->seq, exec->id);
+		return -1;
+	}
+
+	if (hyper_setup_exec_tty(exec) < 0) {
+		fprintf(stderr, "setup exec tty failed\n");
+		return -1;
+	}
+
+	if (exec->id) {
+		if (hyper_exec_in_container(pod, exec) < 0) {
+			fprintf(stderr, "notify container exec failed\n");
+			return -1;
+		}
+
+		return 0;
+	}
+
+	if (socketpair(PF_UNIX, SOCK_STREAM, 0, pipe) < 0) {
+		perror("create pipe between pod init execcmd failed");
+		return -1;
+	}
+
+	pid = fork();
+	if (pid < 0) {
+		fprintf(stderr, "fork failed\n");
+		return -1;
+	} else if (pid > 0) {
+		uint32_t type;
+
+		if (hyper_get_type_block(pipe[0], &type) < 0 || type != READY) {
+			fprintf(stderr, "hyper init doesn't get execcmd ready message\n");
+			return -1;
+		}
+
+		close(pipe[0]);
+		close(pipe[1]);
+		fprintf(stdout, "hyper init get ready message\n");
+		exec->pid = pid;
+		fprintf(stdout, "create exec cmd %s pid %d\n", exec->argv[0], pid);
+
+		list_add_tail(&exec->list, &pod->pe_head);
+		if (exec->seq == 0)
+			return 0;
+
+		fprintf(stdout, "init pod exec pts event %p, ops %p, fd %d\n",
+			&exec->e, &pts_ops, exec->e.fd);
+		if (hyper_init_event(&exec->e, &pts_ops, 0, ctl.tty.fd, NULL) < 0 ||
+		    hyper_add_event(ctl.efd, &exec->e) < 0) {
+			fprintf(stderr, "add pts master event failed\n");
+			return -1;
+		}
+
+		return 0;
+	}
+
+	if (hyper_dup_exec_tty(pipe[1], exec) < 0) {
+		fprintf(stderr, "dup pts to exec stdio failed\n");
+		_exit(-1);
+	}
+
+	close(pipe[0]);
+	close(pipe[1]);
+
+	if (execvp(exec->argv[0], exec->argv) < 0) {
+		perror("exec failed");
+		_exit(-1);
+	}
+
+	_exit(0);
+}
+
+int hyper_container_execcmd(struct hyper_pod *pod)
+{
+	struct hyper_exec *exec = global_exec;
+	struct hyper_container *container = NULL;
+	int fd, pid, sock = pod->ctl.fd;
+	char root[512]; int pipe[2];
+
+	global_exec = NULL;
+
+	container = hyper_find_container(pod, exec->id);
+	if (container == NULL) {
+		fprintf(stderr, "can not find container %s\n", exec->id);
+		return -1;
+	}
+
+	if (sprintf(root, "/tmp/hyper/%s/devpts/", exec->id) < 0) {
+		fprintf(stderr, "get container %s pts path failed\n", exec->id);
+		return -1;
+	}
+
+	if (socketpair(PF_UNIX, SOCK_STREAM, 0, pipe) < 0) {
+		perror("create pipe between pod init execcmd failed");
+		return -1;
+	}
+
+	pid = fork();
+	if (pid < 0) {
+		fprintf(stderr, "container init fork failed\n");
+		return -1;
+	} else if (pid > 0) {
+		uint32_t type;
+
+		if (hyper_get_type_block(pipe[0], &type) < 0 || type != READY) {
+			fprintf(stderr, "pod init get execcmd ready message failed\n");
+			hyper_send_type_block(sock, ERROR, 0);
+			return 0;
+		}
+
+		close(pipe[0]);
+		close(pipe[1]);
+		fprintf(stdout, "pod init get ready message\n");
+
+		exec->pid = pid;
+		fprintf(stdout, "create exec cmd %s pid %d\n", exec->argv[0], pid);
+
+		if (hyper_send_type_block(sock, ACK, 0) < 0)
+			return -1;
+
+		return 0;
+	}
+
+	close(pipe[0]);
+	sprintf(root, "/proc/%d/ns/mnt", container->exec.pid);
+
+	fprintf(stdout, "container %s, init pid %d\n",
+		exec->id, container->exec.pid);
+
+	fd = open(root, O_RDONLY);
+	if (fd < 0) {
+		perror("fail to open container mnt ns\n");
+		goto fail;
+	}
+
+	if (setns(fd, CLONE_NEWNS) < 0) {
+		perror("enter mnt ns failed");
+		goto fail;
+	}
+
+	close(fd);
+
+	sprintf(root, "/tmp/hyper/%s/root/%s/",
+		container->id, container->rootfs);
+
+	fprintf(stdout, "root directory for container is %s, exec %s\n",
+		root, exec->argv[0]);
+
+	/* TODO: wait for container finishing setup root */
+	if (chroot(root) < 0) {
+		perror("chroot for exec command failed");
+		goto fail;
+	}
+
+	chdir("/");
+
+	if (hyper_dup_exec_tty(pipe[1], exec)) {
+		fprintf(stderr, "dup pts to stdio failed\n");
+		goto fail;
+	}
+
+	close(pipe[1]);
+	if (execvp(exec->argv[0], exec->argv) < 0)
+		perror("exec failed");
+
+	_exit(-1);
+fail:
+	hyper_send_type_block(pipe[1], ERROR, 0);
+	_exit(-1);
+}
+
+int hyper_release_exec(struct hyper_exec *exec,
+		       struct hyper_pod *pod)
+{
+	int i;
+
+	close(exec->e.fd);
+	free(exec->pty);
+
+	list_del_init(&exec->list);
+
+	fprintf(stdout, "%s exit code %" PRIu8"\n", __func__, exec->code);
+	if (exec->init) {
+		fprintf(stdout, "%s container init exited, type %d, remains %d, policy %d\n",
+			__func__, pod->type, pod->remains, pod->policy);
+
+		/* stop pod, should not restart container */
+		if (pod->type == STOPPOD)
+			return 0;
+
+		if (exec->code)
+			pod->code = exec->code;
+
+		if (--pod->remains > 0)
+			return 0;
+
+		/* should shutdown? */
+		if (pod->policy == POLICY_NEVER ||
+		   ((pod->policy == POLICY_ONFAILURE) && pod->code == 0)) {
+			hyper_shutdown(pod);
+			return 0;
+		}
+
+		if (hyper_request_restart_containers(pod) < 0) {
+			fprintf(stderr, "restart container failed\n");
+			return -1;
+		}
+
+		return 0;
+	}
+
+	free(exec->id);
+
+	for (i = 0; i < exec->argc; i++) {
+		fprintf(stdout, "argv %d %s\n", i, exec->argv[i]);
+		free(exec->argv[i]);
+	}
+
+	free(exec->argv);
+	free(exec);
+
+	return 0;
+}
+
+struct hyper_exec *hyper_find_exec_by_pid(struct list_head *head, int pid)
+{
+	struct hyper_exec *exec;
+
+	list_for_each_entry(exec, head, list) {
+		fprintf(stdout, "exec pid %d, pid %d\n", exec->pid, pid);
+		if (exec->pid != pid)
+			continue;
+
+		return exec;
+	}
+
+	return NULL;
+}
+
+struct hyper_exec *hyper_find_exec_by_seq(struct hyper_pod *pod, uint64_t seq)
+{
+	struct hyper_exec *exec;
+
+	list_for_each_entry(exec, &pod->ce_head, list) {
+		fprintf(stdout, "container exec seq %" PRIu64 ", seq %" PRIu64 "\n",
+			exec->seq, seq);
+		if (exec->seq != seq)
+			continue;
+
+		return exec;
+	}
+
+	list_for_each_entry(exec, &pod->pe_head, list) {
+		fprintf(stdout, "pod exec seq %" PRIu64 ", seq %" PRIu64 "\n",
+			exec->seq, seq);
+		if (exec->seq != seq)
+			continue;
+
+		return exec;
+	}
+
+	return NULL;
+}
+
+int hyper_send_exec_eof(int to, struct hyper_pod *pod,
+		      struct list_head *head, int pid,
+		      uint8_t code)
+{
+	struct hyper_exec *exec;
+	uint8_t seq[12];
+
+	exec = hyper_find_exec_by_pid(head, pid);
+	if (exec == NULL) {
+		fprintf(stdout, "can not find exec whose pid is %d\n",
+			pid);
+		return 0;
+	}
+
+	fprintf(stdout, "%s exec pid %d, seq %" PRIu64 ", container %s\n",
+		__func__, exec->pid, exec->seq, exec->id ? exec->id : "pod");
+
+	exec->code = code;
+
+	if (exec->seq == 0)
+		goto out;
+
+	hyper_set_be64(seq, exec->seq);
+	hyper_set_be32(seq + 8, 12);
+	if (hyper_send_data(to, seq, 12) < 0) {
+		fprintf(stderr, "pod signal_loop send finishcmd failed\n");
+		return -1;
+	}
+out:
+	hyper_release_exec(exec, pod);
+
+	return 0;
+}
+
+void hyper_cleanup_exec(struct hyper_pod *pod)
+{
+	struct hyper_exec *exec, *next;
+
+	list_for_each_entry_safe(exec, next, &pod->ce_head, list) {
+		fprintf(stdout, "cleanup container exec seq %" PRIu64 "\n", exec->seq);
+		hyper_release_exec(exec, pod);
+	}
+}

src/exec.h

@@ -0,0 +1,36 @@
+#ifndef _EXEC_H
+#define _EXEC_H
+
+#include "list.h"
+#include "event.h"
+
+struct hyper_exec {
+	struct list_head	list;
+	struct hyper_event	e;
+	char			*id;
+	char			*pty;
+	char			**argv;
+	int			argc;
+	uint64_t		seq;
+	int			pid;
+	int			ptyno;
+	int			init;
+	uint8_t			code;
+};
+
+struct hyper_pod;
+
+int hyper_exec_cmd(char *json, int length);
+int hyper_release_exec(struct hyper_exec *, struct hyper_pod *);
+int hyper_container_execcmd(struct hyper_pod *pod);
+int hyper_setup_exec_tty(struct hyper_exec *e);
+int hyper_dup_exec_tty(int fd, struct hyper_exec *e);
+struct hyper_exec *hyper_find_exec_by_pid(struct list_head *head, int pid);
+struct hyper_exec *hyper_find_exec_by_seq(struct hyper_pod *pod, uint64_t seq);
+int hyper_send_exec_eof(int to, struct hyper_pod *pod,
+		      struct list_head *head, int pid,
+		      uint8_t code);
+void hyper_cleanup_exec(struct hyper_pod *pod);
+
+extern struct hyper_event_ops pts_ops;
+#endif

src/hyper.h

@@ -0,0 +1,80 @@
+#ifndef _DVM_H_
+#define _DVM_H_
+
+#include <stdint.h>
+
+#include "net.h"
+#include "list.h"
+#include "exec.h"
+#include "event.h"
+#include "container.h"
+
+enum {
+	SETDVM,
+	STARTPOD,
+	GETPOD,
+	STOPPOD,
+	DESTROYPOD,
+	RESTARTCONTAINER,
+	EXECCMD,
+	FINISHCMD,
+	READY,
+	ACK,
+	ERROR,
+	WINSIZE,
+	PING,
+	FINISH,
+};
+
+enum {
+	POLICY_NEVER,
+	POLICY_ALWAYS,
+	POLICY_ONFAILURE,
+};
+
+struct hyper_pod {
+	struct hyper_container	*c;
+	struct hyper_interface	*iface;
+	struct hyper_route	*rt;
+	struct list_head	pe_head;
+	struct list_head	ce_head;
+	char			*hostname;
+	char			*tag;
+	char			*channel;
+	int			init_pid;
+	uint32_t		c_num;
+	uint32_t		i_num;
+	uint32_t		r_num;
+	uint32_t		e_num;
+	uint32_t		type;
+	uint32_t		code;
+	uint32_t		remains;
+	uint8_t			policy;
+	int			efd;
+	struct hyper_event	sig;
+	struct hyper_event	ctl;
+};
+
+struct hyper_win_size {
+	char		*tty;
+	int		row;
+	int		column;
+	uint64_t	seq;
+};
+
+struct hyper_ctl {
+	int			efd;
+	struct hyper_event	sig;
+	struct hyper_event	tty;
+	struct hyper_event	chan;
+	struct hyper_event	ctl;
+};
+
+int hyper_mkdir(char *hyper_path);
+int hyper_open_serial(char *tty);
+struct hyper_container *hyper_find_container(struct hyper_pod *pod, char *id);
+
+extern struct hyper_pod global_pod;
+extern struct hyper_ctl ctl;
+extern struct hyper_exec *global_exec;
+#endif

src/jsmn.c

@@ -0,0 +1,321 @@
+#include <stdlib.h>
+
+#include "jsmn.h"
+
+/**
+ * Allocates a fresh unused token from the token pull.
+ */
+static jsmntok_t *jsmn_alloc_token(jsmn_parser *parser,
+				   jsmntok_t *tokens,
+				   size_t num_tokens)
+{
+	jsmntok_t *tok;
+
+	if (parser->toknext >= num_tokens)
+		return NULL;
+
+	tok = &tokens[parser->toknext++];
+	tok->start = tok->end = -1;
+	tok->size = 0;
+#ifdef JSMN_PARENT_LINKS
+	tok->parent = -1;
+#endif
+	return tok;
+}
+
+/**
+ * Fills token type and boundaries.
+ */
+static void jsmn_fill_token(jsmntok_t *token, jsmntype_t type,
+			    int start, int end)
+{
+	token->type = type;
+	token->start = start;
+	token->end = end;
+	token->size = 0;
+}
+
+/**
+ * Fills next available token with JSON primitive.
+ */
+static jsmnerr_t jsmn_parse_primitive(jsmn_parser *parser, const char *js,
+				      size_t len, jsmntok_t *tokens,
+				      size_t num_tokens) {
+	jsmntok_t *token;
+	int start;
+
+	start = parser->pos;
+
+	for (; parser->pos < len && js[parser->pos] != '\0'; parser->pos++) {
+		switch (js[parser->pos]) {
+#ifndef JSMN_STRICT
+		/* In strict mode primitive must be followed by "," or "}" or "]" */
+		case ':':
+#endif
+		case '\t': case '\r': case '\n': case ' ':
+		case ',': case ']': case '}':
+			goto found;
+		}
+		if (js[parser->pos] < 32 || js[parser->pos] >= 127) {
+			parser->pos = start;
+			return JSMN_ERROR_INVAL;
+		}
+	}
+#ifdef JSMN_STRICT
+	/* In strict mode primitive must be followed by a comma/object/array */
+	parser->pos = start;
+	return JSMN_ERROR_PART;
+#endif
+
+found:
+	if (tokens == NULL) {
+		parser->pos--;
+		return 0;
+	}
+	token = jsmn_alloc_token(parser, tokens, num_tokens);
+	if (token == NULL) {
+		parser->pos = start;
+		return JSMN_ERROR_NOMEM;
+	}
+	jsmn_fill_token(token, JSMN_PRIMITIVE, start, parser->pos);
+#ifdef JSMN_PARENT_LINKS
+	token->parent = parser->toksuper;
+#endif
+	parser->pos--;
+	return 0;
+}
+
+/**
+ * Filsl next token with JSON string.
+ */
+static jsmnerr_t jsmn_parse_string(jsmn_parser *parser, const char *js,
+				   size_t len, jsmntok_t *tokens,
+				   size_t num_tokens)
+{
+	jsmntok_t *token;
+	int start = parser->pos;
+
+	parser->pos++;
+
+	/* Skip starting quote */
+	for (; parser->pos < len && js[parser->pos] != '\0'; parser->pos++) {
+		char c = js[parser->pos];
+
+		/* Quote: end of string */
+		if (c == '\"') {
+			if (tokens == NULL)
+				return 0;
+
+			token = jsmn_alloc_token(parser, tokens, num_tokens);
+			if (token == NULL) {
+				parser->pos = start;
+				return JSMN_ERROR_NOMEM;
+			}
+			jsmn_fill_token(token, JSMN_STRING, start+1, parser->pos);
+#ifdef JSMN_PARENT_LINKS
+			token->parent = parser->toksuper;
+#endif
+			return 0;
+		}
+
+		/* Backslash: Quoted symbol expected */
+		if (c == '\\' && parser->pos + 1 < len) {
+			int i;
+
+			parser->pos++;
+			switch (js[parser->pos]) {
+			/* Allowed escaped symbols */
+			case '\"': case '/': case '\\': case 'b':
+			case 'f': case 'r': case 'n': case 't':
+				break;
+			/* Allows escaped symbol \uXXXX */
+			case 'u':
+				parser->pos++;
+				for (i = 0; i < 4 && parser->pos < len && js[parser->pos] != '\0'; i++) {
+					/* If it isn't a hex character we have an error */
+					if (!((js[parser->pos] >= 48 && js[parser->pos] <= 57) || /* 0-9 */
+					      (js[parser->pos] >= 65 && js[parser->pos] <= 70) || /* A-F */
+					      (js[parser->pos] >= 97 && js[parser->pos] <= 102))) { /* a-f */
+						parser->pos = start;
+						return JSMN_ERROR_INVAL;
+					}
+					parser->pos++;
+				}
+				parser->pos--;
+				break;
+			/* Unexpected symbol */
+			default:
+				parser->pos = start;
+				return JSMN_ERROR_INVAL;
+			}
+		}
+	}
+	parser->pos = start;
+	return JSMN_ERROR_PART;
+}
+
+/**
+ * Parse JSON string and fill tokens.
+ */
+jsmnerr_t jsmn_parse(jsmn_parser *parser, const char *js, size_t len,
+		     jsmntok_t *tokens, unsigned int num_tokens)
+{
+	jsmnerr_t r;
+	int i;
+	jsmntok_t *token;
+	int count = 0;
+
+	for (; parser->pos < len && js[parser->pos] != '\0'; parser->pos++) {
+		char c;
+		jsmntype_t type;
+
+		c = js[parser->pos];
+		switch (c) {
+		case '{': case '[':
+			count++;
+			if (tokens == NULL)
+				break;
+
+			token = jsmn_alloc_token(parser, tokens, num_tokens);
+			if (token == NULL)
+				return JSMN_ERROR_NOMEM;
+			if (parser->toksuper != -1) {
+				tokens[parser->toksuper].size++;
+#ifdef JSMN_PARENT_LINKS
+				token->parent = parser->toksuper;
+#endif
+			}
+			token->type = (c == '{' ? JSMN_OBJECT : JSMN_ARRAY);
+			token->start = parser->pos;
+			parser->toksuper = parser->toknext - 1;
+			break;
+		case '}': case ']':
+			if (tokens == NULL)
+				break;
+			type = (c == '}' ? JSMN_OBJECT : JSMN_ARRAY);
+#ifdef JSMN_PARENT_LINKS
+			if (parser->toknext < 1)
+				return JSMN_ERROR_INVAL;
+
+			token = &tokens[parser->toknext - 1];
+			for (;;) {
+				if (token->start != -1 && token->end == -1) {
+					if (token->type != type)
+						return JSMN_ERROR_INVAL;
+
+					token->end = parser->pos + 1;
+					parser->toksuper = token->parent;
+					break;
+				}
+				if (token->parent == -1)
+					break;
+
+				token = &tokens[token->parent];
+			}
+#else
+			for (i = parser->toknext - 1; i >= 0; i--) {
+				token = &tokens[i];
+				if (token->start != -1 && token->end == -1) {
+					if (token->type != type)
+						return JSMN_ERROR_INVAL;
+
+					parser->toksuper = -1;
+					token->end = parser->pos + 1;
+						break;
+				}
+			}
+			/* Error if unmatched closing bracket */
+			if (i == -1)
+				return JSMN_ERROR_INVAL;
+			for (; i >= 0; i--) {
+				token = &tokens[i];
+				if (token->start != -1 && token->end == -1) {
+					parser->toksuper = i;
+					break;
+				}
+			}
+#endif
+			break;
+		case '\"':
+			r = jsmn_parse_string(parser, js, len, tokens, num_tokens);
+			if (r < 0)
+				return r;
+			count++;
+			if (parser->toksuper != -1 && tokens != NULL)
+				tokens[parser->toksuper].size++;
+			break;
+		case '\t': case '\r': case '\n': case ' ':
+			break;
+		case ':':
+			parser->toksuper = parser->toknext - 1;
+			break;
+		case ',':
+			if (tokens != NULL &&
+			    tokens[parser->toksuper].type != JSMN_ARRAY &&
+			    tokens[parser->toksuper].type != JSMN_OBJECT) {
+#ifdef JSMN_PARENT_LINKS
+				parser->toksuper = tokens[parser->toksuper].parent;
+#else
+				for (i = parser->toknext - 1; i >= 0; i--) {
+					if (tokens[i].type == JSMN_ARRAY || tokens[i].type == JSMN_OBJECT) {
+						if (tokens[i].start != -1 && tokens[i].end == -1) {
+							parser->toksuper = i;
+							break;
+						}
+					}
+				}
+#endif
+			}
+			break;
+#ifdef JSMN_STRICT
+		/* In strict mode primitives are: numbers and booleans */
+		case '-': case '0': case '1': case '2': case '3': case '4':
+		case '5': case '6': case '7': case '8': case '9':
+		case 't': case 'f': case 'n':
+			/* And they must not be keys of the object */
+			if (tokens != NULL) {
+				jsmntok_t *t = &tokens[parser->toksuper];
+
+				if (t->type == JSMN_OBJECT ||
+				   (t->type == JSMN_STRING && t->size != 0))
+					return JSMN_ERROR_INVAL;
+			}
+#else
+		/* In non-strict mode every unquoted value is a primitive */
+		default:
+#endif
+			r = jsmn_parse_primitive(parser, js, len, tokens, num_tokens);
+			if (r < 0)
+				return r;
+			count++;
+			if (parser->toksuper != -1 && tokens != NULL)
+				tokens[parser->toksuper].size++;
+			break;
+
+#ifdef JSMN_STRICT
+		/* Unexpected char in strict mode */
+		default:
+			return JSMN_ERROR_INVAL;
+#endif
+		}
+	}
+
+	for (i = parser->toknext - 1; i >= 0; i--) {
+		/* Unmatched opened object or array */
+		if (tokens[i].start != -1 && tokens[i].end == -1)
+			return JSMN_ERROR_PART;
+	}
+
+	return count;
+}
+
+/**
+ * Creates a new parser based over a given  buffer with an array of tokens
+ * available.
+ */
+void jsmn_init(jsmn_parser *parser)
+{
+	parser->pos = 0;
+	parser->toknext = 0;
+	parser->toksuper = -1;
+}

src/jsmn.h

@@ -0,0 +1,67 @@
+#ifndef __JSMN_H_
+#define __JSMN_H_
+
+#include <stddef.h>
+
+/**
+ * JSON type identifier. Basic types are:
+ * o Object
+ * o Array
+ * o String
+ * o Other primitive: number, boolean (true/false) or null
+ */
+typedef enum {
+	JSMN_PRIMITIVE = 0,
+	JSMN_OBJECT = 1,
+	JSMN_ARRAY = 2,
+	JSMN_STRING = 3
+} jsmntype_t;
+
+typedef enum {
+	/* Not enough tokens were provided */
+	JSMN_ERROR_NOMEM = -1,
+	/* Invalid character inside JSON string */
+	JSMN_ERROR_INVAL = -2,
+	/* The string is not a full JSON packet, more bytes expected */
+	JSMN_ERROR_PART = -3
+} jsmnerr_t;
+
+/**
+ * JSON token description.
+ * @param		type	type (object, array, string etc.)
+ * @param		start	start position in JSON data string
+ * @param		end		end position in JSON data string
+ */
+typedef struct {
+	jsmntype_t type;
+	int start;
+	int end;
+	int size;
+#ifdef JSMN_PARENT_LINKS
+	int parent;
+#endif
+} jsmntok_t;
+
+/**
+ * JSON parser. Contains an array of token blocks available. Also stores
+ * the string being parsed now and current position in that string
+ */
+typedef struct {
+	unsigned int pos; /* offset in the JSON string */
+	unsigned int toknext; /* next token to allocate */
+	int toksuper; /* superior token node, e.g parent object or array */
+} jsmn_parser;
+
+/**
+ * Create JSON parser over an array of tokens
+ */
+void jsmn_init(jsmn_parser *parser);
+
+/**
+ * Run JSON parser. It parses a JSON data string into and array of tokens, each describing
+ * a single JSON object.
+ */
+jsmnerr_t jsmn_parse(jsmn_parser *parser, const char *js, size_t len,
+		jsmntok_t *tokens, unsigned int num_tokens);
+
+#endif /* __JSMN_H_ */

src/list.h

@@ -0,0 +1,191 @@
+#ifndef _LIST_H_
+#define _LIST_H_
+
+#include <stddef.h>
+
+/*
+ * Simple doubly linked list implementation.
+ *
+ * Some of the internal functions ("__xxx") are useful when
+ * manipulating whole lists rather than single entries, as
+ * sometimes we already know the next/prev entries and we can
+ * generate better code by using them directly rather than
+ * using the generic single-entry routines.
+ */
+
+/**
+ * container_of - cast a member of a structure out to the containing structure
+ * @ptr:	the pointer to the member.
+ * @type:	the type of the container struct this is embedded in.
+ * @member:	the name of the member within the struct.
+ *
+ */
+#define container_of(ptr, type, member) ({			\
+	const typeof(((type *)0)->member) * __mptr = (ptr);	\
+	(type *)((char *)__mptr - offsetof(type, member)); })
+
+struct list_head {
+	struct list_head *next, *prev;
+};
+
+#define LIST_HEAD_INIT(name) { &(name), &(name) }
+
+#define LIST_HEAD(name) \
+	struct list_head name = LIST_HEAD_INIT(name)
+
+static inline void INIT_LIST_HEAD(struct list_head *list)
+{
+	list->next = list;
+	list->prev = list;
+}
+
+static inline void __list_add(struct list_head *new,
+			      struct list_head *prev,
+			      struct list_head *next)
+{
+	next->prev = new;
+	new->next = next;
+	new->prev = prev;
+	prev->next = new;
+}
+
+/**
+ * list_add - add a new entry
+ * @new: new entry to be added
+ * @head: list head to add it after
+ *
+ * Insert a new entry after the specified head.
+ * This is good for implementing stacks.
+ */
+static inline void list_add(struct list_head *new, struct list_head *head)
+{
+	__list_add(new, head, head->next);
+}
+
+/**
+ * list_add_tail - add a new entry
+ * @new: new entry to be added
+ * @head: list head to add it before
+ *
+ * Insert a new entry before the specified head.
+ * This is useful for implementing queues.
+ */
+static inline void list_add_tail(struct list_head *new, struct list_head *head)
+{
+	__list_add(new, head->prev, head);
+}
+
+/*
+ * Delete a list entry by making the prev/next entries
+ * point to each other.
+ *
+ * This is only for internal list manipulation where we know
+ * the prev/next entries already!
+ */
+static inline void __list_del(struct list_head *prev, struct list_head *next)
+{
+	next->prev = prev;
+	prev->next = next;
+}
+
+/**
+ * list_del - deletes entry from list.
+ * @entry: the element to delete from the list.
+ * Note: list_empty() on entry does not return true after this, the entry is
+ * in an undefined state.
+ */
+static inline void list_del(struct list_head *entry)
+{
+	__list_del(entry->prev, entry->next);
+}
+
+/**
+ * list_del_init - deletes entry from list and reinitialize it.
+ * @entry: the element to delete from the list.
+ */
+static inline void list_del_init(struct list_head *entry)
+{
+	list_del(entry);
+	INIT_LIST_HEAD(entry);
+}
+
+/**
+ * list_empty - tests whether a list is empty
+ * @head: the list to test.
+ */
+static inline int list_empty(const struct list_head *head)
+{
+	return head->next == head;
+}
+
+/**
+ * list_entry - get the struct for this entry
+ * @ptr:	the &struct list_head pointer.
+ * @type:	the type of the struct this is embedded in.
+ * @member:	the name of the list_head within the struct.
+ */
+#define list_entry(ptr, type, member) \
+	container_of(ptr, type, member)
+
+/**
+ * list_for_each	-	iterate over a list
+ * @pos:	the &struct list_head to use as a loop cursor.
+ * @head:	the head for your list.
+ */
+#define list_for_each(pos, head) \
+	for (pos = (head)->next; pos != (head); pos = pos->next)
+
+/**
+ * list_first_entry - get the first element from a list
+ * @ptr:	the list head to take the element from.
+ * @type:	the type of the struct this is embedded in.
+ * @member:	the name of the list_head within the struct.
+ *
+ * Note, that list is expected to be not empty.
+ */
+#define list_first_entry(ptr, type, member) \
+	list_entry((ptr)->next, type, member)
+
+/**
+ * list_last_entry - get the last element from a list
+ * @ptr:	the list head to take the element from.
+ * @type:	the type of the struct this is embedded in.
+ * @member:	the name of the list_head within the struct.
+ *
+ * Note, that list is expected to be not empty.
+ */
+#define list_last_entry(ptr, type, member) \
+	list_entry((ptr)->prev, type, member)
+
+/**
+ * list_next_entry - get the next element in list
+ * @pos:	the type * to cursor
+ * @member:	the name of the list_head within the struct.
+ */
+#define list_next_entry(pos, member) \
+	list_entry((pos)->member.next, typeof(*(pos)), member)
+
+/**
+ * list_for_each_entry	-	iterate over list of given type
+ * @pos:	the type * to use as a loop cursor.
+ * @head:	the head for your list.
+ * @member:	the name of the list_head within the struct.
+ */
+#define list_for_each_entry(pos, head, member)				\
+	for (pos = list_first_entry(head, typeof(*pos), member);	\
+	     &pos->member != (head);					\
+	     pos = list_next_entry(pos, member))
+
+/**
+ * list_for_each_entry_safe - iterate over list of given type safe against removal of list entry
+ * @pos:	the type * to use as a loop cursor.
+ * @n:		another type * to use as temporary storage
+ * @head:	the head for your list.
+ * @member:	the name of the list_head within the struct.
+ */
+#define list_for_each_entry_safe(pos, n, head, member)			\
+	for (pos = list_first_entry(head, typeof(*pos), member),	\
+	     n = list_next_entry(pos, member);				\
+	     &pos->member != (head); 					\
+	     pos = n, n = list_next_entry(n, member))
+#endif

src/net.c

@@ -0,0 +1,793 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/socket.h>
+#include <limits.h>
+#include <errno.h>
+#include <unistd.h>
+#include <arpa/inet.h>
+#include <fcntl.h>
+
+#include "hyper.h"
+
+void hyper_set_be32(uint8_t *buf, uint32_t val)
+{
+	buf[0] = val >> 24;
+	buf[1] = val >> 16;
+	buf[2] = val >> 8;
+	buf[3] = val;
+}
+
+uint32_t hyper_get_be32(uint8_t *buf)
+{
+	return buf[0] << 24 | buf[1] << 16 | buf[2] << 8 | buf[3];
+}
+
+void hyper_set_be64(uint8_t *buf, uint64_t val)
+{
+	hyper_set_be32(buf, val >> 32);
+	hyper_set_be32(buf + 4, val);
+}
+
+uint64_t hyper_get_be64(uint8_t *buf)
+{
+	uint64_t v;
+
+	v = (uint64_t) hyper_get_be32(buf) << 32;
+	v |= hyper_get_be32(buf + 4);
+	return v;
+}
+
+int hyper_send_data(int fd, uint8_t *data, uint32_t len)
+{
+	int length = 0, size;
+
+	while (length < len) {
+		size = write(fd, data + length, len - length);
+
+		if (size <= 0) {
+			if (errno == EINTR)
+				continue;
+			/* EAGAIN means unblock and the peer of virtio-ports is disappear */
+			if (errno == EAGAIN)
+				return 0;
+
+			perror("send hyper data failed");
+			return -1;
+		}
+
+		length += size;
+	}
+
+	return 0;
+}
+
+int hyper_send_msg(int fd, uint32_t type, uint32_t len,
+		 uint8_t *message)
+{
+	uint8_t buf[8];
+
+	fprintf(stdout, "hyper send type %d, len %d\n", type, len);
+
+	hyper_set_be32(buf, type);
+	hyper_set_be32(buf + 4, len + 8);
+
+	if (hyper_send_data(fd, buf, 8) < 0)
+		return -1;
+
+	if (message && hyper_send_data(fd, message, len) < 0)
+		return -1;
+
+	return 0;
+}
+
+int hyper_send_type(int fd, uint32_t type)
+{
+	return hyper_send_msg(fd, type, 0, NULL);
+}
+
+int hyper_get_type_block(int fd, uint32_t *type)
+{
+	int len = 0, size;
+	uint8_t buf[8];
+
+	while (len < 8) {
+		size = read(fd, buf + len, 8 - len);
+
+		if (size <= 0) {
+			if (errno == EINTR)
+				continue;
+			perror("wait for ack failed");
+			return -1;
+		}
+		len += size;
+	}
+
+	*type = hyper_get_be32(buf);
+	return 0;
+}
+
+int hyper_send_type_block(int fd, uint32_t type, int need_ack)
+{
+	int ret = 0, flags;
+	uint32_t t;
+
+	flags = fcntl(fd, F_GETFL, 0);
+	if (flags < 0) {
+		fprintf(stderr, "get fd flag failed\n");
+		return -1;
+	}
+
+	if (fcntl(fd, F_SETFL, flags & ~O_NONBLOCK) < 0) {
+		perror("set fd BLOCK failed");
+		return -1;
+	}
+
+	ret = hyper_send_msg(fd, type, 0, NULL);
+	if (ret < 0)
+		goto out;
+
+	if (need_ack == 0)
+		goto out;
+
+	ret = hyper_get_type_block(fd, &t);
+	if (ret < 0) {
+		fprintf(stderr, "can not get type\n");
+		goto out;
+	}
+
+	fprintf(stdout, "get type %" PRIu32"\n", type);
+
+	if (t != ACK)
+		ret = -1;
+out:
+	if (fcntl(fd, F_SETFL, flags | O_NONBLOCK) < 0) {
+		perror("set fd BLOCK failed");
+		return -1;
+	}
+
+	return ret;
+}
+
+static int get_addr_ipv4(uint8_t *ap, const char *cp)
+{
+	int i;
+
+	for (i = 0; i < 4; i++) {
+		unsigned long n;
+		char *endp;
+
+		n = strtoul(cp, &endp, 0);
+		if (n > 255)
+			return -1;      /* bogus network value */
+
+		if (endp == cp) /* no digits */
+			return -1;
+		ap[i] = n;
+
+		if (*endp == '\0')
+			break;
+
+		if (i == 3 || *endp != '.')
+			return -1;      /* extra characters */
+
+		cp = endp + 1;
+	}
+
+	return 1;
+}
+
+static int addattr_l(struct nlmsghdr *n, int maxlen, int type, void *data, int alen)
+{
+	int len = RTA_LENGTH(alen);
+	struct rtattr *rta;
+
+	if (NLMSG_ALIGN(n->nlmsg_len) + len > maxlen)
+		return -1;
+	rta = (struct rtattr *)(((char *)n) + NLMSG_ALIGN(n->nlmsg_len));
+	rta->rta_type = type;
+	rta->rta_len = len;
+	memcpy(RTA_DATA(rta), data, alen);
+	n->nlmsg_len = NLMSG_ALIGN(n->nlmsg_len) + len;
+	return 0;
+}
+
+static int hyper_get_ifindex(char *nic)
+{
+	int fd, ifindex = -1;
+	char path[512], buf[4];
+
+	fprintf(stdout, "net device %s\n", nic);
+	sprintf(path, "/sys/class/net/%s/ifindex", nic);
+	fprintf(stdout, "net device sys path is %s\n", path);
+
+	fd = open(path, O_RDONLY);
+	if (fd < 0) {
+		perror("can not open file");
+		return -1;
+	}
+
+	memset(buf, 0, sizeof(buf));
+	if (read(fd, buf, sizeof(buf)) <= 0) {
+		perror("can read open file");
+		goto out;
+	}
+
+	ifindex = atoi(buf);
+	fprintf(stdout, "get ifindex %d\n", ifindex);
+out:
+	close(fd);
+	return ifindex;
+}
+
+static int netlink_open(struct rtnl_handle *rth)
+{
+	memset(rth, 0, sizeof(*rth));
+
+	rth->fd = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
+	if (rth->fd < 0) {
+		perror("cannot open netlink socket");
+		return -1;
+	}
+
+	rth->local.nl_family = AF_NETLINK;
+	rth->local.nl_groups = 0;
+
+	if (bind(rth->fd, (struct sockaddr *)&rth->local, sizeof(rth->local)) < 0) {
+		perror("cannot bind netlink socket");
+		goto out;
+	}
+
+	rth->seq = 0;
+	return 0;
+out:
+	close(rth->fd);
+	return -1;
+}
+
+static void netlink_close(struct rtnl_handle *rth)
+{
+	if (rth->fd > 0)
+		close(rth->fd);
+	rth->fd = -1;
+}
+
+static int rtnl_talk(struct rtnl_handle *rtnl,
+		     struct nlmsghdr *n, pid_t peer,
+		     unsigned groups, struct nlmsghdr *answer)
+{
+	int status;
+	struct sockaddr_nl nladdr;
+	struct iovec iov = { (void *)n, n->nlmsg_len };
+	struct msghdr msg = { (void *)&nladdr, sizeof(nladdr), &iov, 1, NULL, 0, 0 };
+
+	memset(&nladdr, 0, sizeof(nladdr));
+	nladdr.nl_family = AF_NETLINK;
+	nladdr.nl_pid = peer;
+	nladdr.nl_groups = groups;
+	n->nlmsg_seq = ++rtnl->seq;
+	if (answer == NULL)
+		n->nlmsg_flags |= NLM_F_ACK;
+
+	status = sendmsg(rtnl->fd, &msg, 0);
+	if (status < 0)
+		return -1;
+
+	return 0;
+}
+
+static int hyper_up_nic(struct rtnl_handle *rth, int ifindex)
+{
+	struct {
+		struct nlmsghdr n;
+		struct ifinfomsg i;
+		char buf[1024];
+	} req;
+
+	memset(&req, 0, sizeof(req));
+	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg));
+	req.n.nlmsg_flags = NLM_F_REQUEST;
+	req.n.nlmsg_type = RTM_NEWLINK;
+	req.i.ifi_family = AF_UNSPEC;
+	req.i.ifi_change |= IFF_UP;
+	req.i.ifi_flags |= IFF_UP;
+	req.i.ifi_index = ifindex;
+
+	if (rtnl_talk(rth, &req.n, 0, 0, NULL) < 0)
+		return -1;
+
+	return 0;
+}
+
+static int hyper_remove_nic(char *device)
+{
+	char path[256], real[128];
+	int fd;
+
+	sprintf(path, "/sys/class/net/%s", device);
+
+	if (readlink(path, real, 128) < 0) {
+		perror("fail to read link directory");
+		return -1;
+	}
+
+	sprintf(path, "/sys/%s/../../../remove", real + 5);
+
+	fprintf(stdout, "get net sys path %s\n", path);
+
+	fd = open(path, O_WRONLY);
+	if (fd < 0) {
+		perror("open file failed");
+		return -1;
+	}
+
+	if (write(fd, "1\n", 2) < 0) {
+		perror("write 1 to file failed");
+		close(fd);
+		return 1;
+	}
+
+	close(fd);
+	return 0;
+}
+
+static int hyper_down_nic(struct rtnl_handle *rth, int ifindex)
+{
+	struct {
+		struct nlmsghdr n;
+		struct ifinfomsg i;
+		char buf[1024];
+	} req;
+
+	memset(&req, 0, sizeof(req));
+	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifinfomsg));
+	req.n.nlmsg_flags = NLM_F_REQUEST;
+	req.n.nlmsg_type = RTM_NEWLINK;
+	req.i.ifi_family = AF_UNSPEC;
+	req.i.ifi_change |= IFF_UP;
+	req.i.ifi_flags &= ~IFF_UP;
+	req.i.ifi_index = ifindex;
+
+	if (rtnl_talk(rth, &req.n, 0, 0, NULL) < 0)
+		return -1;
+
+	return 0;
+}
+
+static int mask2bits(uint32_t netmask)
+{
+	unsigned bits = 0;
+	uint32_t mask = ntohl(netmask);
+	uint32_t host = ~mask;
+
+	/* a valid netmask must be 2^n - 1 */
+	if ((host & (host + 1)) != 0)
+		return -1;
+
+	for (; mask; mask <<= 1)
+		++bits;
+
+	return bits;
+}
+
+static int get_netmask(unsigned *val, const char *addr)
+{
+	char *ptr;
+	unsigned long res;
+	uint32_t data;
+	int b;
+
+	res = strtoul(addr, &ptr, 0);
+
+	if (!ptr || ptr == addr || *ptr)
+		goto get_addr;
+
+	if (res == ULONG_MAX && errno == ERANGE)
+		goto get_addr;
+
+	if (res > UINT_MAX)
+		goto get_addr;
+
+	*val = res;
+	return 0;
+
+get_addr:
+	if (get_addr_ipv4((uint8_t *)&data, addr) <= 0)
+		return -1;
+
+	b = mask2bits(data);
+	if (b < 0)
+		return -1;
+
+	*val = b;
+	return 0;
+}
+
+static int hyper_setup_route(struct rtnl_handle *rth,
+			   struct hyper_route *rt)
+{
+	uint32_t data;
+	struct {
+		struct nlmsghdr n;
+		struct rtmsg r;
+		char buf[1024];
+	} req;
+
+	if (!rt->dst) {
+		fprintf(stderr, "route dest is null\n");
+		return -1;
+	}
+
+	memset(&req, 0, sizeof(req));
+	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
+	req.n.nlmsg_flags = NLM_F_CREATE | NLM_F_EXCL | NLM_F_REQUEST;
+	req.n.nlmsg_type = RTM_NEWROUTE;
+
+	req.r.rtm_family = AF_INET;
+	req.r.rtm_table = RT_TABLE_MAIN;
+	req.r.rtm_scope = RT_SCOPE_UNIVERSE;
+	req.r.rtm_type = RTN_UNICAST;
+	req.r.rtm_protocol = RTPROT_BOOT;
+	req.r.rtm_dst_len = 0;
+
+	if (rt->gw) {
+		if (get_addr_ipv4((uint8_t *)&data, rt->gw) <= 0) {
+			fprintf(stderr, "get gw failed\n");
+			return -1;
+		}
+
+		if (addattr_l(&req.n, sizeof(req), RTA_GATEWAY, &data, 4)) {
+			fprintf(stderr, "setup gateway attr failed\n");
+			return -1;
+		}
+	}
+
+	if (rt->device) {
+		rt->ifindex = hyper_get_ifindex(rt->device);
+		if (addattr_l(&req.n, sizeof(req), RTA_OIF, &rt->ifindex, 4)) {
+			fprintf(stderr, "setup oif attr failed\n");
+			return -1;
+		}
+	}
+
+	if (strcmp(rt->dst, "default") && strcmp(rt->dst, "any") && strcmp(rt->dst, "all")) {
+		unsigned mask;
+		char *slash = strchr(rt->dst, '/');
+
+		req.r.rtm_dst_len = 32;
+
+		if (slash)
+			*slash = 0;
+
+		if (get_addr_ipv4((uint8_t *)&data, rt->dst) <= 0) {
+			fprintf(stderr, "get dst failed\n");
+			return -1;
+		}
+
+		if (addattr_l(&req.n, sizeof(req), RTA_DST, &data, 4)) {
+			fprintf(stderr, "setup gateway attr failed\n");
+			return -1;
+		}
+
+		if (slash) {
+			if (get_netmask(&mask, slash + 1) < 0) {
+				fprintf(stderr, "get netmask failed\n");
+				return -1;
+			}
+			req.r.rtm_dst_len = mask;
+			*slash = '/';
+		}
+	}
+
+	if (rtnl_talk(rth, &req.n, 0, 0, NULL) < 0) {
+		fprintf(stderr, "rtnl talk failed\n");
+		return -1;
+	}
+
+	return 0;
+}
+
+static int hyper_cleanup_route(struct rtnl_handle *rth, struct hyper_route *rt)
+{
+	uint32_t data;
+	struct {
+		struct nlmsghdr n;
+		struct rtmsg r;
+		char buf[1024];
+	} req;
+
+	if (!rt->dst) {
+		fprintf(stderr, "route dest is null\n");
+		return -1;
+	}
+
+	memset(&req, 0, sizeof(req));
+	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct rtmsg));
+	req.n.nlmsg_flags = NLM_F_REQUEST;
+	req.n.nlmsg_type = RTM_DELROUTE;
+
+	req.r.rtm_family = AF_INET;
+	req.r.rtm_table = RT_TABLE_MAIN;
+	req.r.rtm_scope = RT_SCOPE_UNIVERSE;
+	req.r.rtm_type = RTN_UNICAST;
+	req.r.rtm_protocol = RTPROT_BOOT;
+	req.r.rtm_dst_len = 0;
+
+	if (rt->gw) {
+		if (get_addr_ipv4((uint8_t *)&data, rt->gw) <= 0) {
+			fprintf(stderr, "get gw failed\n");
+			return -1;
+		}
+
+		if (addattr_l(&req.n, sizeof(req), RTA_GATEWAY, &data, 4)) {
+			fprintf(stderr, "setup gateway attr failed\n");
+			return -1;
+		}
+	}
+
+	if (rt->device) {
+		if (addattr_l(&req.n, sizeof(req), RTA_OIF, &rt->ifindex, 4)) {
+			fprintf(stderr, "setup oif attr failed\n");
+			return -1;
+		}
+	}
+
+	if (strcmp(rt->dst, "default") && strcmp(rt->dst, "any") && strcmp(rt->dst, "all")) {
+		unsigned mask;
+		char *slash = strchr(rt->dst, '/');
+
+		req.r.rtm_dst_len = 32;
+
+		if (slash)
+			*slash = 0;
+
+		if (get_addr_ipv4((uint8_t *)&data, rt->dst) <= 0) {
+			fprintf(stderr, "get dst failed\n");
+			return -1;
+		}
+
+		if (addattr_l(&req.n, sizeof(req), RTA_DST, &data, 4)) {
+			fprintf(stderr, "setup gateway attr failed\n");
+			return -1;
+		}
+
+		if (slash) {
+			if (get_netmask(&mask, slash + 1) < 0) {
+				fprintf(stderr, "get netmask failed\n");
+				return -1;
+			}
+			req.r.rtm_dst_len = mask;
+			*slash = '/';
+		}
+	}
+
+	if (rtnl_talk(rth, &req.n, 0, 0, NULL) < 0) {
+		fprintf(stderr, "rtnl talk failed\n");
+		return -1;
+	}
+
+	return 0;
+}
+
+static int hyper_setup_interface(struct rtnl_handle *rth,
+			       struct hyper_interface *iface)
+{
+	uint8_t data[4];
+	unsigned mask;
+	struct {
+		struct nlmsghdr n;
+		struct ifaddrmsg ifa;
+		char buf[256];
+	} req;
+
+	if (!(iface->device && iface->ipaddr && iface->mask)) {
+		fprintf(stderr, "interface information incorrect\n");
+		return -1;
+	}
+
+	memset(&req, 0, sizeof(req));
+	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifaddrmsg));
+	req.n.nlmsg_flags = NLM_F_REQUEST | NLM_F_CREATE | NLM_F_EXCL;
+	req.n.nlmsg_type = RTM_NEWADDR;
+	req.ifa.ifa_family = AF_INET;
+
+	iface->ifindex = hyper_get_ifindex(iface->device);
+	req.ifa.ifa_index = iface->ifindex;
+	req.ifa.ifa_scope = 0;
+
+	if (get_addr_ipv4((uint8_t *)&data, iface->ipaddr) <= 0) {
+		fprintf(stderr, "get addr failed\n");
+		return -1;
+	}
+
+	if (addattr_l(&req.n, sizeof(req), IFA_LOCAL, &data, 4)) {
+		fprintf(stderr, "setup attr failed\n");
+		return -1;
+	}
+
+	if (get_netmask(&mask, iface->mask) < 0) {
+		fprintf(stderr, "get netamsk failed\n");
+		return -1;
+	}
+
+	req.ifa.ifa_prefixlen = mask;
+	fprintf(stdout, "interface get netamsk %d %s\n", req.ifa.ifa_prefixlen, iface->mask);
+	if (rtnl_talk(rth, &req.n, 0, 0, NULL) < 0) {
+		perror("rtnl_talk failed");
+		return -1;
+	}
+
+	if (hyper_up_nic(rth, iface->ifindex) < 0) {
+		fprintf(stderr, "up device %d failed\n", iface->ifindex);
+		return -1;
+	}
+
+	return 0;
+}
+
+static int hyper_cleanup_interface(struct rtnl_handle *rth,
+				 struct hyper_interface *iface)
+{
+	uint8_t data[4];
+	unsigned mask;
+	struct {
+		struct nlmsghdr n;
+		struct ifaddrmsg ifa;
+		char buf[256];
+	} req;
+
+	if (!(iface->device && iface->ipaddr && iface->mask)) {
+		fprintf(stderr, "interface information incorrect\n");
+		return -1;
+	}
+
+	memset(&req, 0, sizeof(req));
+	req.n.nlmsg_len = NLMSG_LENGTH(sizeof(struct ifaddrmsg));
+	req.n.nlmsg_flags = NLM_F_REQUEST;
+	req.n.nlmsg_type = RTM_DELADDR;
+	req.ifa.ifa_family = AF_INET;
+
+	req.ifa.ifa_index = iface->ifindex;
+	req.ifa.ifa_scope = 0;
+
+	if (get_addr_ipv4((uint8_t *)&data, iface->ipaddr) <= 0) {
+		fprintf(stderr, "get addr failed\n");
+		return -1;
+	}
+
+	if (addattr_l(&req.n, sizeof(req), IFA_LOCAL, &data, 4)) {
+		fprintf(stderr, "setup attr failed\n");
+		return -1;
+	}
+
+	if (get_netmask(&mask, iface->mask) < 0) {
+		fprintf(stderr, "get netamsk failed\n");
+		return -1;
+	}
+
+	req.ifa.ifa_prefixlen = mask;
+	fprintf(stdout, "interface get netamsk %d %s\n", req.ifa.ifa_prefixlen, iface->mask);
+	if (rtnl_talk(rth, &req.n, 0, 0, NULL) < 0) {
+		perror("rtnl_talk failed");
+		return -1;
+	}
+
+	if (hyper_down_nic(rth, iface->ifindex) < 0) {
+		fprintf(stderr, "up device %d failed\n", iface->ifindex);
+		return -1;
+	}
+
+	if (hyper_remove_nic(iface->device) < 0) {
+		fprintf(stderr, "remove device %s failed\n", iface->device);
+		return -1;
+	}
+
+	return 0;
+}
+
+int hyper_rescan(void)
+{
+	int fd = open("/sys/bus/pci/rescan", O_WRONLY);
+
+	if (fd < 0) {
+		perror("can not open rescan file");
+		return -1;
+	}
+
+	if (write(fd, "1\n", 2) < 0) {
+		perror("can not open rescan file");
+		close(fd);
+		return -1;
+	}
+	fprintf(stdout, "finish rescan\n");
+	close(fd);
+	return 0;
+}
+
+int hyper_setup_network(struct hyper_pod *pod)
+{
+	int i, ret = 0;
+	struct hyper_interface *iface;
+	struct hyper_route *rt;
+	struct rtnl_handle rth;
+
+	if (hyper_rescan() < 0)
+		return -1;
+
+	if (netlink_open(&rth) < 0)
+		return -1;
+
+	for (i = 0; i < pod->i_num; i++) {
+		iface = &pod->iface[i];
+
+		ret = hyper_setup_interface(&rth, iface);
+		if (ret < 0) {
+			fprintf(stderr, "link up device %s failed\n", iface->device);
+			goto out;
+		}
+	}
+
+	ret = hyper_up_nic(&rth, 1);
+	if (ret < 0) {
+		fprintf(stderr, "link up lo device failed\n");
+		goto out;
+	}
+
+	for (i = 0; i < pod->r_num; i++) {
+		rt = &pod->rt[i];
+
+		ret = hyper_setup_route(&rth, rt);
+		if (ret < 0) {
+			fprintf(stderr, "setup route failed\n");
+			goto out;
+		}
+	}
+
+out:
+	netlink_close(&rth);
+	return ret;
+}
+
+void hyper_cleanup_network(struct hyper_pod *pod)
+{
+	int i;
+	struct rtnl_handle rth;
+	struct hyper_interface *iface;
+	struct hyper_route *rt;
+
+	if (netlink_open(&rth) < 0) {
+		fprintf(stdout, "open netlink failed\n");
+		return;
+	}
+
+	for (i = 0; i < pod->r_num; i++) {
+		rt = &pod->rt[i];
+
+		if (hyper_cleanup_route(&rth, rt) < 0)
+			fprintf(stderr, "cleanup route failed\n");
+
+		free(rt->dst);
+		free(rt->gw);
+		free(rt->device);
+	}
+
+	free(pod->rt);
+	pod->rt = NULL;
+	pod->r_num = 0;
+
+	for (i = 0; i < pod->i_num; i++) {
+		iface = &pod->iface[i];
+
+		if (hyper_cleanup_interface(&rth, iface) < 0)
+			fprintf(stderr, "link down device %s failed\n", iface->device);
+
+		free(iface->device);
+		free(iface->ipaddr);
+		free(iface->mask);
+	}
+
+	free(pod->iface);
+	pod->iface = NULL;
+	pod->i_num = 0;
+	netlink_close(&rth);
+}

src/net.h

@@ -0,0 +1,57 @@
+#ifndef _NET_H_
+#define _NET_H_
+
+#include <stdio.h>
+#include <ctype.h>
+#include <stdint.h>
+#include <sys/types.h>
+#include <sys/socket.h>
+#include <linux/if.h>
+#include <linux/netlink.h>
+#include <linux/rtnetlink.h>
+
+struct rtnl_handle {
+	int fd;
+	struct sockaddr_nl local;
+	struct sockaddr_nl peer;
+	__u32 seq;
+	__u32 dump;
+};
+
+typedef struct {
+	__u8 family;
+	__u8 bytelen;
+	__s16 bitlen;
+	__u32 flags;
+	__u32 data[8];
+} inet_prefix;
+
+struct hyper_interface {
+	char		*device;
+	int		ifindex;
+	char		*ipaddr;
+	char		*mask;
+};
+
+struct hyper_route {
+	char		*dst;
+	char		*gw;
+	char		*device;
+	int		ifindex;
+};
+
+struct hyper_pod;
+int hyper_rescan(void);
+void hyper_set_be32(uint8_t *buf, uint32_t val);
+uint32_t hyper_get_be32(uint8_t *buf);
+void hyper_set_be64(uint8_t *buf, uint64_t val);
+uint64_t hyper_get_be64(uint8_t *buf);
+int hyper_setup_network(struct hyper_pod *pod);
+void hyper_cleanup_network(struct hyper_pod *pod);
+int hyper_get_type_block(int fd, uint32_t *type);
+int hyper_send_type(int fd, uint32_t type);
+int hyper_send_type_block(int fd, uint32_t type, int need_ack);
+int hyper_send_msg(int fd, uint32_t type, uint32_t len,
+		 uint8_t *message);
+int hyper_send_data(int fd, uint8_t *data, uint32_t len);
+#endif

src/parse.c

@@ -0,0 +1,605 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <inttypes.h>
+
+#include "list.h"
+#include "parse.h"
+
+char *json_token_str(char *js, jsmntok_t *t)
+{
+	js[t->end] = '\0';
+	return js + t->start;
+}
+
+int json_token_int(char *js, jsmntok_t *t)
+{
+	return strtol(json_token_str(js, t), 0, 10);
+}
+
+uint64_t json_token_ll(char *js, jsmntok_t *t)
+{
+	return strtoll(json_token_str(js, t), 0, 10);
+}
+
+int json_token_streq(char *js, jsmntok_t *t, char *s)
+{
+	return (strncmp(js + t->start, s, t->end - t->start) == 0 &&
+		strlen(s) == (size_t)(t->end - t->start));
+}
+
+static int container_parse_cmd(struct hyper_container *c, char *json, jsmntok_t *toks)
+{
+	int i = 1, j;
+
+	if (toks[i].type != JSMN_ARRAY) {
+		fprintf(stdout, "cmd need array");
+		return -1;
+	}
+
+	c->exec.argc = toks[i].size;
+
+	c->exec.argv = calloc(c->exec.argc + 1, sizeof(*c->exec.argv));
+	c->exec.argv[c->exec.argc] = NULL;
+
+	for (j = 0; j < c->exec.argc; j++) {
+		i++;
+		c->exec.argv[j] = strdup(json_token_str(json, &toks[i]));
+		fprintf(stdout, "container init arg %d %s\n", j, c->exec.argv[j]);
+	}
+
+	return i;
+}
+
+static int container_parse_volumes(struct hyper_container *c, char *json, jsmntok_t *toks)
+{
+	int i = 1, j;
+
+	if (toks[i].type != JSMN_ARRAY) {
+		fprintf(stdout, "volume need array\n");
+		return -1;
+	}
+	c->vols_num = toks[i].size;
+	fprintf(stdout, "volumes num %d\n", c->vols_num);
+	c->vols = calloc(c->vols_num, sizeof(*c->vols));
+
+	for (j = 0; j < c->vols_num; j++) {
+		int i_volume, next_volume;
+
+		i++;
+		if (toks[i].type != JSMN_OBJECT) {
+			fprintf(stdout, "volume array need object\n");
+			return -1;
+		}
+		next_volume = toks[i].size;
+		for (i_volume = 0; i_volume < next_volume; i_volume++) {
+			i++;
+			if (json_token_streq(json, &toks[i], "device")) {
+				c->vols[j].device =
+				strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "volume %d device %s\n", j, c->vols[j].device);
+			} else if (json_token_streq(json, &toks[i], "mount")) {
+				c->vols[j].mountpoint =
+				strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "volume %d mp %s\n", j, c->vols[j].mountpoint);
+			} else if (json_token_streq(json, &toks[i], "fstype")) {
+				c->vols[j].fstype =
+				strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "volume %d fstype %s\n", j, c->vols[j].fstype);
+			} else if (json_token_streq(json, &toks[i], "readOnly")) {
+				if (!json_token_streq(json, &toks[++i], "false"))
+					c->vols[j].readonly = 1;
+				fprintf(stdout, "volume %d readonly %d\n", j, c->vols[j].readonly);
+			} else {
+				fprintf(stdout, "in voulmes incorrect %s\n", json_token_str(json, &toks[i]));
+			}
+		}
+	}
+
+	return i;
+}
+
+static int container_parse_fsmap(struct hyper_container *c, char *json, jsmntok_t *toks)
+{
+	int i = 1, j;
+
+	if (toks[i].type != JSMN_ARRAY) {
+		fprintf(stdout, "envs need array\n");
+		return -1;
+	}
+
+	c->maps_num = toks[i].size;
+	fprintf(stdout, "fsmap num %d\n", c->maps_num);
+	c->maps = calloc(c->maps_num, sizeof(*c->maps));
+
+	for (j = 0; j < c->maps_num; j++) {
+		int i_map, next_map;
+
+		i++;
+		if (toks[i].type != JSMN_OBJECT) {
+			fprintf(stdout, "fsmap array need object\n");
+			return -1;
+		}
+		next_map = toks[i].size;
+		for (i_map = 0; i_map < next_map; i_map++) {
+			i++;
+			if (json_token_streq(json, &toks[i], "source")) {
+				c->maps[j].source =
+				strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "maps %d source %s\n", j, c->maps[j].source);
+			} else if (json_token_streq(json, &toks[i], "path")) {
+				c->maps[j].path =
+				strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "maps %d path %s\n", j, c->maps[j].path);
+			} else if (json_token_streq(json, &toks[i], "readOnly")) {
+				if (!json_token_streq(json, &toks[++i], "false"))
+					c->maps[j].readonly = 1;
+				fprintf(stdout, "maps %d readonly %d\n", j, c->maps[j].readonly);
+			} else {
+				fprintf(stdout, "in maps incorrect %s\n",
+					json_token_str(json, &toks[i]));
+			}
+		}
+	}
+
+	return i;
+}
+
+static int container_parse_envs(struct hyper_container *c, char *json, jsmntok_t *toks)
+{
+	int i = 1, j;
+
+	if (toks[i].type != JSMN_ARRAY) {
+		fprintf(stdout, "encs need array\n");
+		return -1;
+	}
+
+	c->envs_num = toks[i].size;
+	fprintf(stdout, "envs num %d\n", c->envs_num);
+	c->envs = calloc(c->envs_num, sizeof(*c->envs));
+
+	for (j = 0; j < c->envs_num; j++) {
+		int i_env, next_env;
+
+		i++;
+		if (toks[i].type != JSMN_OBJECT) {
+			fprintf(stdout, "env array need object\n");
+			return -1;
+		}
+		next_env = toks[i].size;
+		for (i_env = 0; i_env < next_env; i_env++) {
+			i++;
+			if (json_token_streq(json, &toks[i], "env")) {
+				c->envs[j].env =
+				strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "envs %d env %s\n", j, c->envs[j].env);
+			} else if (json_token_streq(json, &toks[i], "value")) {
+				c->envs[j].value =
+				strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "envs %d value %s\n", j, c->envs[j].value);
+			} else {
+				fprintf(stdout, "in envs incorrect %s\n", json_token_str(json, &toks[i]));
+			}
+		}
+	}
+
+	return i;
+}
+
+static int hyper_parse_container(struct hyper_pod *pod, struct hyper_container *c,
+			       char *json, jsmntok_t *toks)
+{
+	int i = 1, j, next, next_container;
+	jsmntok_t *t;
+
+	if (toks[i].type != JSMN_OBJECT) {
+		fprintf(stderr, "format incorrect\n");
+		return -1;
+	}
+
+	c->exec.init = 1;
+	c->exec.code = -1;
+
+	next_container = toks[i].size;
+	fprintf(stdout, "next container %d\n", next_container);
+
+	for (j = 0; j < next_container; j++) {
+		i++;
+		t = &toks[i];
+		fprintf(stdout, "%d name %s\n", i, json_token_str(json, t));
+		if (json_token_streq(json, t, "id") && t->size == 1) {
+			i++;
+			c->id = strdup(json_token_str(json, &toks[i]));
+			c->exec.id = strdup(c->id);
+			fprintf(stdout, "container id %s\n", c->id);
+		} else if (json_token_streq(json, t, "cmd") && t->size == 1) {
+			next = container_parse_cmd(c, json, &toks[i]);
+			if (next < 0)
+				return -1;
+			i += next;
+		} else if (json_token_streq(json, t, "rootfs") && t->size == 1) {
+			i++;
+			c->rootfs = strdup(json_token_str(json, &toks[i]));
+			fprintf(stdout, "container rootfs %s\n", c->rootfs);
+		} else if (json_token_streq(json, t, "tty") && t->size == 1) {
+			i++;
+			c->exec.seq = json_token_ll(json, &toks[i]);
+			fprintf(stdout, "container seq %" PRIu64 "\n", c->exec.seq);
+		} else if (json_token_streq(json, t, "workdir") && t->size == 1) {
+			i++;
+			c->workdir = strdup(json_token_str(json, &toks[i]));
+			fprintf(stdout, "container workdir %s\n", c->workdir);
+		} else if (json_token_streq(json, t, "image") && t->size == 1) {
+			i++;
+			c->image = strdup(json_token_str(json, &toks[i]));
+			fprintf(stdout, "container image %s\n", c->image);
+		} else if (json_token_streq(json, t, "fstype") && t->size == 1) {
+			i++;
+			c->fstype = strdup(json_token_str(json, &toks[i]));
+			fprintf(stdout, "container fstype %s\n", c->fstype);
+		} else if (json_token_streq(json, t, "volumes") && t->size == 1) {
+			next = container_parse_volumes(c, json, &toks[i]);
+			if (next < 0)
+				return -1;
+			i += next;
+		} else if (json_token_streq(json, t, "fsmap") && t->size == 1) {
+			next = container_parse_fsmap(c, json, &toks[i]);
+			if (next < 0)
+				return -1;
+			i += next;
+		} else if (json_token_streq(json, t, "envs") && t->size == 1) {
+			next = container_parse_envs(c, json, &toks[i]);
+			if (next < 0)
+				return -1;
+			i += next;
+		} else if (json_token_streq(json, t, "restartPolicy") && t->size == 1) {
+			i++;
+/*
+			if (json_token_streq(json, &toks[i], "always") && t->size == 1)
+				c->exec.flags = POLICY_ALWAYS;
+			else if (json_token_streq(json, &toks[i], "onFailure") && t->size == 1)
+				c->exec.flags = POLICY_ONFAILURE;
+			else
+				c->exec.flags = POLICY_NEVER;
+*/
+			fprintf(stdout, "restartPolicy %s\n", json_token_str(json, &toks[i]));
+		}
+	}
+
+	return i;
+}
+
+static int hyper_parse_containers(struct hyper_pod *pod, char *json, jsmntok_t *toks)
+{
+	int i = 1, j, next;
+
+	if (toks[i].type != JSMN_ARRAY) {
+		fprintf(stdout, "format incorrect\n");
+		return -1;
+	}
+
+	pod->remains = pod->c_num = toks[i].size;
+	fprintf(stdout, "container count %d\n", pod->c_num);
+	pod->c = calloc(pod->c_num, sizeof(*pod->c));
+
+	if (pod->c == NULL) {
+		fprintf(stdout, "alloc memory for container failed\n");
+		return -1;
+	}
+
+	for (j = 0; j < pod->c_num; j++) {
+		next = hyper_parse_container(pod, &pod->c[j], json, toks + i);
+		if (next < 0)
+			return -1;
+
+		i += next;
+	}
+
+	return i;
+}
+
+static int hyper_parse_interfaces(struct hyper_pod *pod, char *json, jsmntok_t *toks)
+{
+	int i = 1, j, next_if;
+	struct hyper_interface *iface;
+
+	if (toks[i].type != JSMN_ARRAY) {
+		fprintf(stdout, "interfaces need array\n");
+		return -1;
+	}
+
+	pod->i_num = toks[i].size;
+	fprintf(stdout, "network interfaces num %d\n", pod->i_num);
+	pod->iface = calloc(pod->i_num, sizeof(*iface));
+
+	for (j = 0; j < pod->i_num; j++) {
+		int i_if;
+
+		iface = &pod->iface[j];
+		i++;
+		if (toks[i].type != JSMN_OBJECT) {
+			fprintf(stdout, "network array need object\n");
+			return -1;
+		}
+		next_if = toks[i].size;
+
+		for (i_if = 0; i_if < next_if; i_if++) {
+			i++;
+			if (json_token_streq(json, &toks[i], "device")) {
+				iface->device = strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "net device is %s\n", iface->device);
+			} else if (json_token_streq(json, &toks[i], "ipAddress")) {
+				iface->ipaddr = strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "net ipaddress is %s\n", iface->ipaddr);
+			} else if (json_token_streq(json, &toks[i], "netMask")) {
+				iface->mask = strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "net mask is %s\n", iface->mask);
+			}
+		}
+	}
+
+	return i;
+}
+
+static int hyper_parse_routes(struct hyper_pod *pod, char *json, jsmntok_t *toks)
+{
+	int i = 1, j, next_rt;
+	struct hyper_route *rt;
+
+	if (toks[i].type != JSMN_ARRAY) {
+		fprintf(stdout, "routes need array\n");
+		return -1;
+	}
+
+	pod->r_num = toks[i].size;
+	fprintf(stdout, "network routes num %d\n", pod->r_num);
+	pod->rt = calloc(pod->r_num, sizeof(*rt));
+
+	for (j = 0; j < pod->r_num; j++) {
+		int i_rt;
+
+		rt = &pod->rt[j];
+		i++;
+		if (toks[i].type != JSMN_OBJECT) {
+			fprintf(stdout, "routes array need object\n");
+			return -1;
+		}
+		next_rt = toks[i].size;
+
+		for (i_rt = 0; i_rt < next_rt; i_rt++) {
+			i++;
+			if (json_token_streq(json, &toks[i], "dest")) {
+				rt->dst = strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "route %d dest is %s\n", j, rt->dst);
+			} else if (json_token_streq(json, &toks[i], "gateway")) {
+				rt->gw = strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "route %d gateway is %s\n", j, rt->gw);
+			} else if (json_token_streq(json, &toks[i], "device")) {
+				rt->device = strdup(json_token_str(json, &toks[++i]));
+				fprintf(stdout, "route %d device is %s\n", j, rt->device);
+			}
+		}
+	}
+
+	return i;
+}
+
+int hyper_parse_pod(struct hyper_pod *pod, char *json, int length)
+{
+	int i, n, next = -1;
+	jsmn_parser p;
+	int toks_num = 100;
+	jsmntok_t *toks = NULL;
+
+realloc:
+	toks = realloc(toks, toks_num * sizeof(jsmntok_t));
+
+	fprintf(stdout, "call hyper_start_pod, json %s, len %d\n", json, length);
+	jsmn_init(&p);
+	n = jsmn_parse(&p, json, length, toks, toks_num);
+	if (n < 0) {
+		fprintf(stdout, "jsmn parse failed, n is %d\n", n);
+		if (n == JSMN_ERROR_NOMEM) {
+			toks_num *= 2;
+			goto realloc;
+		}
+
+		goto out;
+	}
+
+	pod->policy = POLICY_NEVER;
+
+	fprintf(stdout, "jsmn parse successed, n is %d\n", n);
+	next = 0;
+	for (i = 0; i < n; i++) {
+		jsmntok_t *t = &toks[i];
+
+		fprintf(stdout, "token %d, type is %d, size is %d\n", i, t->type, t->size);
+
+		if (t->type != JSMN_STRING)
+			continue;
+
+		if (json_token_streq(json, t, "containers") && t->size == 1) {
+			next = hyper_parse_containers(pod, json, t);
+			if (next < 0)
+				goto out;
+
+			i += next;
+		} else if (json_token_streq(json, t, "interfaces") && t->size == 1) {
+			next = hyper_parse_interfaces(pod, json, t);
+			if (next < 0)
+				goto out;
+
+			i += next;
+		} else if (json_token_streq(json, t, "routes") && t->size == 1) {
+			next = hyper_parse_routes(pod, json, t);
+			if (next < 0)
+				goto out;
+
+			i += next;
+		} else if (json_token_streq(json, t, "socket") && t->size == 1) {
+			pod->channel = strdup(json_token_str(json, &toks[++i]));
+			fprintf(stdout, "channel is %s\n", pod->channel);
+		} else if (json_token_streq(json, t, "shareDir") && t->size == 1) {
+			pod->tag = strdup(json_token_str(json, &toks[++i]));
+			fprintf(stdout, "9p tag is %s\n", pod->tag);
+		} else if (json_token_streq(json, t, "hostname") && t->size == 1) {
+			pod->hostname = strdup(json_token_str(json, &toks[++i]));
+			fprintf(stdout, "hostname is %s\n", pod->hostname);
+		} else if (json_token_streq(json, t, "restartPolicy") && t->size == 1) {
+			i++;
+			if (json_token_streq(json, &toks[i], "always") && t->size == 1)
+				pod->policy = POLICY_ALWAYS;
+			else if (json_token_streq(json, &toks[i], "onFailure") && t->size == 1)
+				pod->policy = POLICY_ONFAILURE;
+			fprintf(stdout, "restartPolicy is %" PRIu8 "\n", pod->policy);
+		}
+	}
+
+out:
+	free(toks);
+	return next;
+}
+
+int hyper_parse_winsize(struct hyper_win_size *ws, char *json, int length)
+{
+	int i, n, ret = 0;
+	jsmn_parser p;
+	int toks_num = 10;
+	jsmntok_t *toks = NULL;
+
+realloc:
+	toks = realloc(toks, toks_num * sizeof(jsmntok_t));
+
+	jsmn_init(&p);
+
+	n = jsmn_parse(&p, json, length,  toks, toks_num);
+	if (n < 0) {
+		fprintf(stdout, "jsmn parse failed, n is %d\n", n);
+		if (n == JSMN_ERROR_NOMEM) {
+			toks_num *= 2;
+			goto realloc;
+		}
+
+		goto fail;
+	}
+
+	for (i = 0; i < n; i++) {
+		jsmntok_t *t = &toks[i];
+
+		if (t->type != JSMN_STRING)
+			continue;
+
+		if (i++ == n)
+			goto fail;
+		if (json_token_streq(json, t, "tty")) {
+			if (toks[i].type != JSMN_STRING)
+				goto fail;
+			ws->tty = strdup(json_token_str(json, &toks[i]));
+		} else if (json_token_streq(json, t, "seq")) {
+			if (toks[i].type != JSMN_PRIMITIVE)
+				goto fail;
+			ws->seq = json_token_ll(json, &toks[i]);
+		} else if (json_token_streq(json, t, "row")) {
+			if (toks[i].type != JSMN_PRIMITIVE)
+				goto fail;
+			ws->row = json_token_int(json, &toks[i]);
+		} else if (json_token_streq(json, t, "column")) {
+			if (toks[i].type != JSMN_PRIMITIVE)
+				goto fail;
+			ws->column = json_token_int(json, &toks[i]);
+		}
+	}
+out:
+	free(toks);
+	return ret;
+fail:
+	ret = -1;
+	goto out;
+}
+
+struct hyper_exec *hyper_parse_execcmd(char *json, int length)
+{
+	int i, j, n, has_seq = 0;
+	struct hyper_exec *exec = NULL;
+	char **argv = NULL;
+
+	jsmn_parser p;
+	int toks_num = 10;
+	jsmntok_t *toks = NULL;
+
+realloc:
+	toks = realloc(toks, toks_num * sizeof(jsmntok_t));
+
+	jsmn_init(&p);
+	n = jsmn_parse(&p, json, length,  toks, toks_num);
+	if (n < 0) {
+		fprintf(stdout, "jsmn parse failed, n is %d\n", n);
+		if (n == JSMN_ERROR_NOMEM) {
+			toks_num *= 2;
+			goto realloc;
+		}
+		goto out;
+	}
+
+	exec = calloc(1, sizeof(*exec));
+	if (exec == NULL)
+		goto out;
+
+	INIT_LIST_HEAD(&exec->list);
+
+	for (i = 0, j = 0; i < n; i++) {
+		jsmntok_t *t = &toks[i];
+
+		if (t->type != JSMN_STRING)
+			continue;
+
+		if (json_token_streq(json, t, "container")) {
+			if (i++ == n)
+				goto fail;
+
+			exec->id = strdup(json_token_str(json, &toks[i]));
+			fprintf(stdout, "get container %s\n", exec->id);
+		} else if (json_token_streq(json, t, "seq")) {
+			if (i++ == n)
+				goto fail;
+			has_seq = 1;
+			exec->seq = json_token_ll(json, &toks[i]);
+			fprintf(stdout, "get seq %"PRIu64"\n", exec->seq);
+		} else if (json_token_streq(json, t, "cmd")) {
+			if (i++ == n)
+				goto fail;
+
+			if (toks[i].type != JSMN_ARRAY) {
+				fprintf(stdout, "execcmd need array\n");
+				goto fail;
+			}
+
+			exec->argc = toks[i].size;
+			argv = calloc(exec->argc + 1, sizeof(*argv));
+			argv[exec->argc] = NULL;
+		} else if (j < exec->argc) {
+			argv[j++] = strdup(json_token_str(json, &toks[i]));
+			fprintf(stdout, "argv %d, %s\n", j - 1, argv[j - 1]);
+		}
+	}
+
+	if (!has_seq) {
+		fprintf(stderr, "execcmd format error, has no seq\n");
+		goto fail;
+	}
+	exec->argv = argv;
+out:
+	free(toks);
+	return exec;
+fail:
+	free(exec->id);
+	for (i = 0; i < exec->argc; i++)
+		free(argv[i]);
+
+	free(exec->argv);
+	free(exec);
+
+	exec = NULL;
+	goto out;
+}

src/parse.h

@@ -0,0 +1,13 @@
+#ifndef _DVM_JSON_H_
+#define _DVM_JSON_H_
+
+#include "hyper.h"
+#include "jsmn.h"
+
+int hyper_parse_pod(struct hyper_pod *pod, char *json, int length);
+struct hyper_exec *hyper_parse_execcmd(char *json, int length);
+char *json_token_str(char *js, jsmntok_t *t);
+int json_token_streq(char *js, jsmntok_t *t, char *s);
+int hyper_parse_winsize(struct hyper_win_size *ws, char *json, int length);
+
+#endif

src/util.c

@@ -0,0 +1,300 @@
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <dirent.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <errno.h>
+#include <signal.h>
+#include <ctype.h>
+#include <mntent.h>
+#include <sys/mount.h>
+#include <sys/reboot.h>
+#include <linux/reboot.h>
+
+#include "net.h"
+#include "util.h"
+
+char *read_cmdline(void)
+{
+	return NULL;
+}
+
+int hyper_mkdir(char *hyper_path)
+{
+	struct stat st;
+	char *p, *path = strdup(hyper_path);
+
+	if (path == NULL) {
+		errno = ENOMEM;
+		return -1;
+	}
+
+	fprintf(stdout, "create directory %s\n", path);
+	if (stat(path, &st) >= 0) {
+		if (S_ISDIR(st.st_mode))
+			return 0;
+		errno = ENOTDIR;
+		return -1;
+	}
+
+	if (errno != ENOENT)
+		return -1;
+
+	p = strrchr(path, '/');
+	if (p == NULL) {
+		errno = EINVAL;
+		return -1;
+	}
+
+	if (p != path) {
+		*p = '\0';
+
+		if (hyper_mkdir(path) < 0)
+			return -1;
+
+		*p = '/';
+	}
+
+	if (mkdir(path, 0755) < 0 && errno != EEXIST)
+		return -1;
+
+	return 0;
+}
+
+int hyper_open_channel(char *channel, int mode)
+{
+	struct dirent **list;
+	struct dirent *dir;
+	int fd = -1, i, num;
+	char path[256], name[128];
+
+	num = scandir("/sys/class/virtio-ports/", &list, NULL, NULL);
+	if (num < 0) {
+		perror("scan /sys/calss/virtio-ports/ failed");
+		return -1;
+	}
+
+	memset(path, 0, sizeof(path));
+
+	for (i = 0; i < num; i++) {
+		dir = list[i];
+		if (dir->d_name[0] == '.')
+			continue;
+
+		if (snprintf(path, sizeof(path), "/sys/class/virtio-ports/%s/name", dir->d_name) < 0) {
+			fprintf(stderr, "get channel device %s path failed\n", dir->d_name);
+			continue;
+		}
+
+		fd = open(path, O_RDONLY);
+
+		memset(name, 0, sizeof(name));
+		if (fd < 0 || read(fd, name, sizeof(name)) < 0)
+			continue;
+
+		close(fd);
+		fd = -1;
+
+		if (strncmp(name, channel, strlen(channel))) {
+			fprintf(stderr, "channel %s, directory %s\n", channel, name);
+			continue;
+		}
+
+		if (snprintf(path, sizeof(path), "/dev/%s", dir->d_name) < 0) {
+			fprintf(stderr, "get channel device %s path failed\n", dir->d_name);
+			continue;
+		}
+
+		fprintf(stdout, "open hyper channel %s\n", path);
+		fd = open(path, O_RDWR | O_CLOEXEC | mode);
+		if (fd < 0)
+			perror("fail to open channel deice");
+
+		break;
+	}
+
+	free(list);
+	return fd;
+}
+
+int hyper_open_serial_dev(char *tty)
+{
+	int fd = open(tty, O_RDWR | O_CLOEXEC | O_NOCTTY);
+
+	if (fd < 0) {
+		perror("fail to open tty device");
+		return -1;
+	}
+
+	return fd;
+}
+
+int hyper_open_serial(char *tty)
+{
+	char path[256];
+
+	memset(path, 0, sizeof(path));
+
+	if (snprintf(path, sizeof(path), "/dev/%s", tty) < 0) {
+		fprintf(stderr, "get channel device %s path failed\n", tty);
+		return -1;
+	}
+
+	fprintf(stdout, "open hyper tty %s\n", path);
+
+	return hyper_open_serial_dev(path);
+}
+
+int hyper_setfd_cloexec(int fd)
+{
+	int flags = fcntl(fd, F_GETFD);
+
+	if (flags < 0) {
+		perror("fcntl F_GETFD failed");
+		return -1;
+	}
+
+	if (fcntl(fd, F_SETFD, flags | FD_CLOEXEC) < 0) {
+		perror("fcntl F_SETFD failed");
+		return -1;
+	}
+
+	return 0;
+}
+
+int hyper_setfd_block(int fd)
+{
+	int flags = fcntl(fd, F_GETFL);
+
+	if (flags < 0) {
+		perror("fcntl F_GETFL failed");
+		return -1;
+	}
+
+	if (fcntl(fd, F_SETFL, flags & ~O_NONBLOCK) < 0) {
+		perror("fcntl F_SETFD failed");
+		return -1;
+	}
+
+	return 0;
+}
+
+int hyper_setfd_nonblock(int fd)
+{
+	int flags = fcntl(fd, F_GETFL);
+
+	if (flags < 0) {
+		perror("fcntl F_GETFL failed");
+		return -1;
+	}
+
+	if (fcntl(fd, F_SETFL, flags | O_NONBLOCK) < 0) {
+		perror("fcntl F_SETFD failed");
+		return -1;
+	}
+
+	return 0;
+}
+
+void hyper_unmount_all(void)
+{
+	FILE *mtab;
+	struct mntent *mnt;
+	char *mntlist[128];
+	int i, n = 0;
+	char *filesys;
+
+	mtab = setmntent("/proc/mounts", "r");
+	if (mtab == NULL) {
+		fprintf(stderr, "cannot open /proc/mount");
+		return;
+	}
+
+	while (n < 128) {
+		mnt = getmntent(mtab);
+		if (mnt == NULL)
+			break;
+
+		if (strcmp(mnt->mnt_type, "devtmpfs") == 0 ||
+		    strcmp(mnt->mnt_type, "proc") == 0 ||
+		    strcmp(mnt->mnt_type, "sysfs") == 0 ||
+		    strcmp(mnt->mnt_type, "ramfs") == 0 ||
+		    strcmp(mnt->mnt_type, "tmpfs") == 0 ||
+		    strcmp(mnt->mnt_type, "rootfs") == 0 ||
+		    strcmp(mnt->mnt_type, "devpts") == 0)
+			continue;
+
+		mntlist[n++] = strdup(mnt->mnt_dir);
+	}
+
+	endmntent(mtab);
+
+	for (i = n - 1; i >= 0; i--) {
+		filesys = mntlist[i];
+		fprintf(stdout, "umount %s\n", filesys);
+		if ((umount(mntlist[i]) < 0) && (umount2(mntlist[i], MNT_DETACH) < 0)) {
+			fprintf(stdout, ("umount %s: %s failed\n"),
+				filesys, strerror(errno));
+		}
+	}
+
+	sync();
+}
+
+void hyper_kill_all(void)
+{
+	int npids = 0;
+	int index = 0;
+	int pid;
+	DIR *dp;
+	struct dirent *de;
+	pid_t *pids = NULL;
+
+	dp = opendir("/proc");
+	if (dp == NULL)
+		return;
+
+	while ((de = readdir(dp)) && de != NULL) {
+		if (!isdigit(de->d_name[0]))
+			continue;
+		pid = atoi(de->d_name);
+		if (pid == 1)
+			continue;
+		if (index <= npids) {
+			pids = realloc(pids, npids + 16384);
+			if (pids == NULL)
+				return;
+			npids += 16384;
+		}
+
+		pids[index++] = pid;
+	}
+
+	fprintf(stdout, "Sending SIGTERM\n");
+
+	for (--index; index >= 0; --index) {
+		fprintf(stdout, "kill process %d\n", pids[index]);
+		kill(pids[index], SIGTERM);
+	}
+
+	free(pids);
+	closedir(dp);
+}
+
+void hyper_shutdown(struct hyper_pod *pod)
+{
+	int i;
+	uint8_t *data = calloc(pod->c_num, 4);
+
+	for (i = 0; i < pod->c_num; i++)
+		hyper_set_be32(data + (i * 4), pod->c[i].exec.code);
+
+	hyper_send_msg(ctl.chan.fd, FINISH, pod->c_num * 4, data);
+
+	hyper_unmount_all();
+	hyper_kill_all();
+	reboot(LINUX_REBOOT_CMD_POWER_OFF);
+}

src/util.h

@@ -0,0 +1,17 @@
+#ifndef _UTIL_H_
+#define _UTIL_H_
+
+#include "hyper.h"
+
+char *read_cmdline(void);
+int hyper_mkdir(char *path);
+int hyper_open_channel(char *channel, int mode);
+int hyper_open_serial_dev(char *tty);
+int hyper_open_serial(char *tty);
+int hyper_setfd_cloexec(int fd);
+int hyper_setfd_block(int fd);
+int hyper_setfd_nonblock(int fd);
+void hyper_shutdown(struct hyper_pod *pod);
+void hyper_kill_all(void);
+void hyper_unmount_all(void);
+#endif