* (C) Copyright IBM Corp. 2007, 2009
*
* Authors:
- * Daniel Lezcano <dlezcano at fr.ibm.com>
+ * Daniel Lezcano <daniel.lezcano at free.fr>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
-#include <unistd.h>
-#include <alloca.h>
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE 1
+#endif
#include <errno.h>
+#include <fcntl.h>
+#include <sched.h>
#include <signal.h>
-#include <syscall.h>
#include <sys/param.h>
-#include <sys/types.h>
#include <sys/stat.h>
-#include <fcntl.h>
+#include <sys/syscall.h>
+#include <sys/types.h>
+#include <unistd.h>
-#include "namespace.h"
+#include "config.h"
#include "log.h"
+#include "memory_utils.h"
+#include "namespace.h"
+#include "utils.h"
-#include "setns.h"
-
-lxc_log_define(lxc_namespace, lxc);
+lxc_log_define(namespace, lxc);
struct clone_arg {
int (*fn)(void *);
void *arg;
};
-int setns(int fd, int nstype)
-{
-#ifndef __NR_setns
- errno = ENOSYS;
- return -1;
-#else
- return syscall(__NR_setns, fd, nstype);
-#endif
-}
-
static int do_clone(void *arg)
{
struct clone_arg *clone_arg = arg;
return clone_arg->fn(clone_arg->arg);
}
-pid_t lxc_clone(int (*fn)(void *), void *arg, int flags)
+#define __LXC_STACK_SIZE 4096
+pid_t lxc_clone(int (*fn)(void *), void *arg, int flags, int *pidfd)
{
+ pid_t ret;
struct clone_arg clone_arg = {
- .fn = fn,
- .arg = arg,
+ .fn = fn,
+ .arg = arg,
};
+ void *stack;
- long stack_size = sysconf(_SC_PAGESIZE);
- void *stack = alloca(stack_size);
- pid_t ret;
+ stack = malloc(__LXC_STACK_SIZE);
+ if (!stack) {
+ SYSERROR("Failed to allocate clone stack");
+ return -ENOMEM;
+ }
#ifdef __ia64__
- ret = __clone2(do_clone, stack,
- stack_size, flags | SIGCHLD, &clone_arg);
+ ret = __clone2(fn, stack, __LXC_STACK_SIZE, flags | SIGCHLD, &clone_arg, pidfd);
#else
- ret = clone(do_clone, stack + stack_size, flags | SIGCHLD, &clone_arg);
+ ret = clone(fn, stack + __LXC_STACK_SIZE, flags | SIGCHLD, &clone_arg, pidfd);
#endif
if (ret < 0)
- ERROR("failed to clone(0x%x): %s", flags, strerror(errno));
+ SYSERROR("Failed to clone (%#x)", flags);
return ret;
}
-int lxc_attach(pid_t pid)
+/* Leave the user namespace at the first position in the array of structs so
+ * that we always attach to it first when iterating over the struct and using
+ * setns() to switch namespaces. This especially affects lxc_attach(): Suppose
+ * you cloned a new user namespace and mount namespace as an unprivileged user
+ * on the host and want to setns() to the mount namespace. This requires you to
+ * attach to the user namespace first otherwise the kernel will fail this check:
+ *
+ * if (!ns_capable(mnt_ns->user_ns, CAP_SYS_ADMIN) ||
+ * !ns_capable(current_user_ns(), CAP_SYS_CHROOT) ||
+ * !ns_capable(current_user_ns(), CAP_SYS_ADMIN))
+ * return -EPERM;
+ *
+ * in
+ *
+ * linux/fs/namespace.c:mntns_install().
+ */
+const struct ns_info ns_info[LXC_NS_MAX] = {
+ [LXC_NS_USER] = { "user", CLONE_NEWUSER, "CLONE_NEWUSER", "LXC_USER_NS" },
+ [LXC_NS_MNT] = { "mnt", CLONE_NEWNS, "CLONE_NEWNS", "LXC_MNT_NS" },
+ [LXC_NS_PID] = { "pid", CLONE_NEWPID, "CLONE_NEWPID", "LXC_PID_NS" },
+ [LXC_NS_UTS] = { "uts", CLONE_NEWUTS, "CLONE_NEWUTS", "LXC_UTS_NS" },
+ [LXC_NS_IPC] = { "ipc", CLONE_NEWIPC, "CLONE_NEWIPC", "LXC_IPC_NS" },
+ [LXC_NS_NET] = { "net", CLONE_NEWNET, "CLONE_NEWNET", "LXC_NET_NS" },
+ [LXC_NS_CGROUP] = { "cgroup", CLONE_NEWCGROUP, "CLONE_NEWCGROUP", "LXC_CGROUP_NS" }
+};
+
+int lxc_namespace_2_cloneflag(const char *namespace)
+{
+ int i;
+
+ for (i = 0; i < LXC_NS_MAX; i++)
+ if (!strcasecmp(ns_info[i].proc_name, namespace))
+ return ns_info[i].clone_flag;
+
+ ERROR("Invalid namespace name \"%s\"", namespace);
+ return -EINVAL;
+}
+
+int lxc_namespace_2_ns_idx(const char *namespace)
{
- char path[MAXPATHLEN];
- char *ns[] = { "pid", "mnt", "net", "ipc", "uts" };
- const int size = sizeof(ns) / sizeof(char *);
- int fd[size];
int i;
- sprintf(path, "/proc/%d/ns", pid);
- if (access(path, X_OK)) {
- ERROR("Does this kernel version support 'attach' ?");
+ for (i = 0; i < LXC_NS_MAX; i++)
+ if (!strcmp(ns_info[i].proc_name, namespace))
+ return i;
+
+ ERROR("Invalid namespace name \"%s\"", namespace);
+ return -EINVAL;
+}
+
+extern int lxc_namespace_2_std_identifiers(char *namespaces)
+{
+ char **it;
+ char *del;
+
+ /* The identifiers for namespaces used with lxc-attach and lxc-unshare
+ * as given on the manpage do not align with the standard identifiers.
+ * This affects network, mount, and uts namespaces. The standard identifiers
+ * are: "mnt", "uts", and "net" whereas lxc-attach and lxc-unshare uses
+ * "MOUNT", "UTSNAME", and "NETWORK". So let's use some cheap memmove()s
+ * to replace them by their standard identifiers.
+ * Let's illustrate this with an example:
+ * Assume the string:
+ *
+ * "IPC|MOUNT|PID"
+ *
+ * then we memmove()
+ *
+ * dest: del + 1 == OUNT|PID
+ * src: del + 3 == NT|PID
+ */
+ if (!namespaces)
return -1;
- }
- for (i = 0; i < size; i++) {
- sprintf(path, "/proc/%d/ns/%s", pid, ns[i]);
- fd[i] = open(path, O_RDONLY);
- if (fd[i] < 0) {
- SYSERROR("failed to open '%s'", path);
- return -1;
- }
+ while ((del = strstr(namespaces, "MOUNT")))
+ memmove(del + 1, del + 3, strlen(del) - 2);
+
+ for (it = (char *[]){"NETWORK", "UTSNAME", NULL}; it && *it; it++)
+ while ((del = strstr(namespaces, *it)))
+ memmove(del + 3, del + 7, strlen(del) - 6);
+
+ return 0;
+}
+
+int lxc_fill_namespace_flags(char *flaglist, int *flags)
+{
+ char *token;
+ int aflag;
+
+ if (!flaglist) {
+ ERROR("At least one namespace is needed.");
+ return -1;
}
- for (i = 0; i < size; i++) {
- if (setns(fd[i], 0)) {
- SYSERROR("failed to set namespace '%s'", ns[i]);
+ lxc_iterate_parts(token, flaglist, "|") {
+ aflag = lxc_namespace_2_cloneflag(token);
+ if (aflag < 0)
return -1;
- }
- close(fd[i]);
+ *flags |= aflag;
}
return 0;