#define _GNU_SOURCE
#include <errno.h>
#include <fcntl.h>
+#include <termios.h>
#include <grp.h>
#include <pwd.h>
#include <signal.h>
#include "lsm/lsm.h"
#include "lxclock.h"
#include "lxcseccomp.h"
+#include "mainloop.h"
#include "namespace.h"
+#include "terminal.h"
#include "utils.h"
#if HAVE_SYS_PERSONALITY_H
#define MS_SLAVE (1 << 19)
#endif
-lxc_log_define(lxc_attach, lxc);
-
-/* /proc/pid-to-str/current\0 = (5 + 21 + 7 + 1) */
-#define __LSMATTRLEN (5 + (LXC_NUMSTRLEN64) + 7 + 1)
-static int lsm_openat(int procfd, pid_t pid, int on_exec)
-{
- int ret = -1;
- int labelfd = -1;
- const char *name;
- char path[__LSMATTRLEN];
-
- name = lsm_name();
-
- if (strcmp(name, "nop") == 0)
- return 0;
-
- if (strcmp(name, "none") == 0)
- return 0;
-
- /* We don't support on-exec with AppArmor */
- if (strcmp(name, "AppArmor") == 0)
- on_exec = 0;
-
- if (on_exec)
- ret = snprintf(path, __LSMATTRLEN, "%d/attr/exec", pid);
- else
- ret = snprintf(path, __LSMATTRLEN, "%d/attr/current", pid);
- if (ret < 0 || ret >= __LSMATTRLEN)
- return -1;
-
- labelfd = openat(procfd, path, O_RDWR);
- if (labelfd < 0) {
- SYSERROR("Unable to open file descriptor to set LSM label.");
- return -1;
- }
-
- return labelfd;
-}
-
-static int lsm_set_label_at(int lsm_labelfd, int on_exec, char *lsm_label)
-{
- int fret = -1;
- const char *name;
- char *command = NULL;
-
- name = lsm_name();
-
- if (strcmp(name, "nop") == 0)
- return 0;
-
- if (strcmp(name, "none") == 0)
- return 0;
-
- /* We don't support on-exec with AppArmor */
- if (strcmp(name, "AppArmor") == 0)
- on_exec = 0;
-
- if (strcmp(name, "AppArmor") == 0) {
- int size;
-
- command =
- malloc(strlen(lsm_label) + strlen("changeprofile ") + 1);
- if (!command) {
- SYSERROR("Failed to write apparmor profile.");
- goto out;
- }
-
- size = sprintf(command, "changeprofile %s", lsm_label);
- if (size < 0) {
- SYSERROR("Failed to write apparmor profile.");
- goto out;
- }
-
- if (write(lsm_labelfd, command, size + 1) < 0) {
- SYSERROR("Unable to set LSM label: %s.", command);
- goto out;
- }
- INFO("Set LSM label to: %s.", command);
- } else if (strcmp(name, "SELinux") == 0) {
- if (write(lsm_labelfd, lsm_label, strlen(lsm_label) + 1) < 0) {
- SYSERROR("Unable to set LSM label: %s.", lsm_label);
- goto out;
- }
- INFO("Set LSM label to: %s.", lsm_label);
- } else {
- ERROR("Unable to restore label for unknown LSM: %s.", name);
- goto out;
- }
- fret = 0;
-
-out:
- free(command);
-
- if (lsm_labelfd != -1)
- close(lsm_labelfd);
-
- return fret;
-}
+lxc_log_define(attach, lxc);
/* /proc/pid-to-str/status\0 = (5 + 21 + 7 + 1) */
#define __PROC_STATUS_LEN (5 + (LXC_NUMSTRLEN64) + 7 + 1)
static void lxc_proc_put_context_info(struct lxc_proc_context_info *ctx)
{
free(ctx->lsm_label);
- if (ctx->container)
+ ctx->lsm_label = NULL;
+
+ if (ctx->container) {
lxc_container_put(ctx->container);
+ ctx->container = NULL;
+ }
+
lxc_proc_close_ns_fd(ctx);
free(ctx);
}
continue;
if (prctl(PR_CAPBSET_DROP, cap, 0, 0, 0)) {
- SYSERROR("Failed to remove capability id %d.", cap);
+ SYSERROR("Failed to drop capability %d", cap);
return -1;
}
+ TRACE("Dropped capability %d", cap);
}
return 0;
}
-static int lxc_attach_set_environment(enum lxc_attach_env_policy_t policy,
+static int lxc_attach_set_environment(struct lxc_proc_context_info *init_ctx,
+ enum lxc_attach_env_policy_t policy,
char **extra_env, char **extra_keep)
{
+ int ret;
+ struct lxc_list *iterator;
+
if (policy == LXC_ATTACH_CLEAR_ENV) {
int path_kept = 0;
char **extra_keep_store = NULL;
if (extra_keep) {
size_t count, i;
- for (count = 0; extra_keep[count]; count++);
+ for (count = 0; extra_keep[count]; count++)
+ ;
extra_keep_store = calloc(count, sizeof(char *));
- if (!extra_keep_store) {
- SYSERROR("Failed to allocate memory for storing current "
- "environment variable values that will be kept.");
+ if (!extra_keep_store)
return -1;
- }
+
for (i = 0; i < count; i++) {
char *v = getenv(extra_keep[i]);
if (v) {
extra_keep_store[i] = strdup(v);
if (!extra_keep_store[i]) {
- SYSERROR("Failed to allocate memory for storing current "
- "environment variable values that will be kept.");
while (i > 0)
free(extra_keep_store[--i]);
free(extra_keep_store);
return -1;
}
+
if (strcmp(extra_keep[i], "PATH") == 0)
path_kept = 1;
}
- /* Calloc sets entire array to zero, so we don't
- * need an else.
- */
}
}
if (clearenv()) {
- char **p;
-
- SYSERROR("Failed to clear environment.");
if (extra_keep_store) {
+ char **p;
+
for (p = extra_keep_store; *p; p++)
free(*p);
+
free(extra_keep_store);
}
+
+ SYSERROR("Failed to clear environment");
return -1;
}
for (i = 0; extra_keep[i]; i++) {
if (extra_keep_store[i]) {
- if (setenv(extra_keep[i], extra_keep_store[i], 1) < 0)
- SYSERROR("Unable to set environment variable.");
+ ret = setenv(extra_keep[i], extra_keep_store[i], 1);
+ if (ret < 0)
+ SYSWARN("Failed to set environment variable");
}
free(extra_keep_store[i]);
}
* out there that just assume that getenv("PATH") is never NULL
* and then die a painful segfault death.
*/
- if (!path_kept)
- setenv("PATH", "/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin", 1);
+ if (!path_kept) {
+ ret = setenv("PATH", "/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin", 1);
+ if (ret < 0)
+ SYSWARN("Failed to set environment variable");
+ }
}
- if (putenv("container=lxc")) {
- SYSERROR("Failed to set environment variable.");
+ ret = putenv("container=lxc");
+ if (ret < 0) {
+ SYSWARN("Failed to set environment variable");
return -1;
}
+ /* Set container environment variables.*/
+ if (init_ctx && init_ctx->container && init_ctx->container->lxc_conf) {
+ lxc_list_for_each(iterator, &init_ctx->container->lxc_conf->environment) {
+ char *env_tmp;
+
+ env_tmp = strdup((char *)iterator->elem);
+ if (!env_tmp)
+ return -1;
+
+ ret = putenv(env_tmp);
+ if (ret < 0) {
+ SYSERROR("Failed to set environment variable: %s", (char *)iterator->elem);
+ return -1;
+ }
+ }
+ }
+
/* Set extra environment variables. */
if (extra_env) {
for (; *extra_env; extra_env++) {
- /* Duplicate the string, just to be on the safe side,
- * because putenv does not do it for us.
- */
- char *p = strdup(*extra_env);
+ char *p;
/* We just assume the user knows what they are doing, so
* we don't do any checks.
*/
- if (!p) {
- SYSERROR("Failed to allocate memory for additional environment "
- "variables.");
+ p = strdup(*extra_env);
+ if (!p)
return -1;
- }
- putenv(p);
+
+ ret = putenv(p);
+ if (ret < 0)
+ SYSWARN("Failed to set environment variable");
}
}
int fd, ret;
pid_t pid;
int pipes[2];
- char *result = NULL;
+ FILE *pipe_f;
+ bool found = false;
+ size_t line_bufsz = 0;
+ char *line = NULL, *result = NULL;
/* We need to fork off a process that runs the getent program, and we
* need to capture its output, so we use a pipe for that purpose.
*/
- ret = pipe(pipes);
+ ret = pipe2(pipes, O_CLOEXEC);
if (ret < 0)
return NULL;
return NULL;
}
- if (pid) {
- int status;
- FILE *pipe_f;
- int found = 0;
- size_t line_bufsz = 0;
- char *line = NULL;
-
- close(pipes[1]);
-
- pipe_f = fdopen(pipes[0], "r");
- while (getline(&line, &line_bufsz, pipe_f) != -1) {
- int i;
- long value;
- char *token;
- char *endptr = NULL, *saveptr = NULL;
-
- /* If we already found something, just continue to read
- * until the pipe doesn't deliver any more data, but
- * don't modify the existing data structure.
- */
- if (found)
- continue;
-
- /* Trim line on the right hand side. */
- for (i = strlen(line); i > 0 && (line[i - 1] == '\n' || line[i - 1] == '\r'); --i)
- line[i - 1] = '\0';
-
- /* Split into tokens: first: user name. */
- token = strtok_r(line, ":", &saveptr);
- if (!token)
- continue;
- /* next: dummy password field */
- token = strtok_r(NULL, ":", &saveptr);
- if (!token)
- continue;
- /* next: user id */
- token = strtok_r(NULL, ":", &saveptr);
- value = token ? strtol(token, &endptr, 10) : 0;
- if (!token || !endptr || *endptr || value == LONG_MIN || value == LONG_MAX)
- continue;
- /* dummy sanity check: user id matches */
- if ((uid_t) value != uid)
- continue;
- /* skip fields: gid, gecos, dir, go to next field 'shell' */
- for (i = 0; i < 4; i++) {
- token = strtok_r(NULL, ":", &saveptr);
- if (!token)
- break;
- }
- if (!token)
- continue;
- free(result);
- result = strdup(token);
-
- /* Sanity check that there are no fields after that. */
- token = strtok_r(NULL, ":", &saveptr);
- if (token)
- continue;
-
- found = 1;
- }
-
- free(line);
- fclose(pipe_f);
- again:
- if (waitpid(pid, &status, 0) < 0) {
- if (errno == EINTR)
- goto again;
- free(result);
- return NULL;
- }
-
- /* Some sanity checks. If anything even hinted at going wrong,
- * we can't be sure we have a valid result, so we assume we
- * don't.
- */
-
- if (!WIFEXITED(status)) {
- free(result);
- return NULL;
- }
-
- if (WEXITSTATUS(status) != 0) {
- free(result);
- return NULL;
- }
-
- if (!found) {
- free(result);
- return NULL;
- }
-
- return result;
- } else {
+ if (!pid) {
char uid_buf[32];
char *arguments[] = {
"getent",
close(pipes[0]);
/* We want to capture stdout. */
- dup2(pipes[1], 1);
+ ret = dup2(pipes[1], STDOUT_FILENO);
close(pipes[1]);
+ if (ret < 0)
+ exit(EXIT_FAILURE);
/* Get rid of stdin/stderr, so we try to associate it with
* /dev/null.
*/
- fd = open("/dev/null", O_RDWR);
+ fd = open_devnull();
if (fd < 0) {
- close(0);
- close(2);
+ close(STDIN_FILENO);
+ close(STDERR_FILENO);
} else {
- dup2(fd, 0);
- dup2(fd, 2);
+ (void)dup3(fd, STDIN_FILENO, O_CLOEXEC);
+ (void)dup3(fd, STDOUT_FILENO, O_CLOEXEC);
close(fd);
}
/* Finish argument list. */
- ret = snprintf(uid_buf, sizeof(uid_buf), "%ld", (long) uid);
- if (ret <= 0)
- exit(-1);
+ ret = snprintf(uid_buf, sizeof(uid_buf), "%ld", (long)uid);
+ if (ret <= 0 || ret >= sizeof(uid_buf))
+ exit(EXIT_FAILURE);
/* Try to run getent program. */
- (void) execvp("getent", arguments);
- exit(-1);
+ (void)execvp("getent", arguments);
+ exit(EXIT_FAILURE);
+ }
+
+ close(pipes[1]);
+
+ pipe_f = fdopen(pipes[0], "r");
+ while (getline(&line, &line_bufsz, pipe_f) != -1) {
+ int i;
+ long value;
+ char *token;
+ char *endptr = NULL, *saveptr = NULL;
+
+ /* If we already found something, just continue to read
+ * until the pipe doesn't deliver any more data, but
+ * don't modify the existing data structure.
+ */
+ if (found)
+ continue;
+
+ /* Trim line on the right hand side. */
+ for (i = strlen(line); i > 0 && (line[i - 1] == '\n' || line[i - 1] == '\r'); --i)
+ line[i - 1] = '\0';
+
+ /* Split into tokens: first: user name. */
+ token = strtok_r(line, ":", &saveptr);
+ if (!token)
+ continue;
+
+ /* next: dummy password field */
+ token = strtok_r(NULL, ":", &saveptr);
+ if (!token)
+ continue;
+
+ /* next: user id */
+ token = strtok_r(NULL, ":", &saveptr);
+ value = token ? strtol(token, &endptr, 10) : 0;
+ if (!token || !endptr || *endptr || value == LONG_MIN ||
+ value == LONG_MAX)
+ continue;
+
+ /* dummy sanity check: user id matches */
+ if ((uid_t)value != uid)
+ continue;
+
+ /* skip fields: gid, gecos, dir, go to next field 'shell' */
+ for (i = 0; i < 4; i++) {
+ token = strtok_r(NULL, ":", &saveptr);
+ if (!token)
+ continue;
+ }
+ if (!token)
+ continue;
+ free(result);
+ result = strdup(token);
+
+ /* Sanity check that there are no fields after that. */
+ token = strtok_r(NULL, ":", &saveptr);
+ if (token)
+ continue;
+
+ found = true;
+ }
+ free(line);
+ fclose(pipe_f);
+
+ ret = wait_for_pid(pid);
+ if (ret < 0) {
+ free(result);
+ return NULL;
+ }
+
+ if (!found) {
+ free(result);
+ return NULL;
}
+
+ return result;
}
static void lxc_attach_get_init_uidgid(uid_t *init_uid, gid_t *init_gid)
uid_t uid = (uid_t)-1;
gid_t gid = (gid_t)-1;
- /* Read capabilities. */
- snprintf(proc_fn, __PROC_STATUS_LEN, "/proc/%d/status", 1);
+ ret = snprintf(proc_fn, __PROC_STATUS_LEN, "/proc/%d/status", 1);
+ if (ret < 0 || ret >= __PROC_STATUS_LEN)
+ return;
proc_file = fopen(proc_fn, "r");
if (!proc_file)
*/
}
-struct attach_clone_payload {
- int ipc_socket;
- lxc_attach_options_t *options;
- struct lxc_proc_context_info *init_ctx;
- lxc_attach_exec_t exec_function;
- void *exec_payload;
-};
-
-static int attach_child_main(void* data);
-
/* Help the optimizer along if it doesn't know that exit always exits. */
#define rexit(c) \
do { \
/* Define default options if no options are supplied by the user. */
static lxc_attach_options_t attach_static_default_options = LXC_ATTACH_OPTIONS_DEFAULT;
-static bool fetch_seccomp(struct lxc_container *c,
- lxc_attach_options_t *options)
+static bool fetch_seccomp(struct lxc_container *c, lxc_attach_options_t *options)
{
+ int ret;
+ bool bret;
char *path;
if (!(options->namespaces & CLONE_NEWNS) ||
}
/* Remove current setting. */
- if (!c->set_config_item(c, "lxc.seccomp", "") &&
- !c->set_config_item(c, "lxc.seccomp.profile", "")) {
+ if (!c->set_config_item(c, "lxc.seccomp.profile", "") &&
+ !c->set_config_item(c, "lxc.seccomp", "")) {
return false;
}
/* Fetch the current profile path over the cmd interface. */
path = c->get_running_config_item(c, "lxc.seccomp.profile");
if (!path) {
- INFO("Failed to get running config item for lxc.seccomp.profile");
+ INFO("Failed to retrieve lxc.seccomp.profile");
path = c->get_running_config_item(c, "lxc.seccomp");
- }
- if (!path) {
- INFO("Failed to get running config item for lxc.seccomp");
- return true;
+ if (!path) {
+ INFO("Failed to retrieve lxc.seccomp");
+ return true;
+ }
}
/* Copy the value into the new lxc_conf. */
- if (!c->set_config_item(c, "lxc.seccomp.profile", path)) {
- free(path);
- return false;
- }
+ bret = c->set_config_item(c, "lxc.seccomp.profile", path);
free(path);
+ if (!bret)
+ return false;
/* Attempt to parse the resulting config. */
- if (lxc_read_seccomp_config(c->lxc_conf) < 0) {
- ERROR("Error reading seccomp policy.");
+ ret = lxc_read_seccomp_config(c->lxc_conf);
+ if (ret < 0) {
+ ERROR("Failed to retrieve seccomp policy");
return false;
}
- INFO("Retrieved seccomp policy.");
+ INFO("Retrieved seccomp policy");
return true;
}
static bool no_new_privs(struct lxc_container *c, lxc_attach_options_t *options)
{
+ bool bret;
char *val;
/* Remove current setting. */
- if (!c->set_config_item(c, "lxc.no_new_privs", ""))
+ if (!c->set_config_item(c, "lxc.no_new_privs", "")) {
+ INFO("Failed to unset lxc.no_new_privs");
return false;
+ }
/* Retrieve currently active setting. */
val = c->get_running_config_item(c, "lxc.no_new_privs");
if (!val) {
- INFO("Failed to get running config item for lxc.no_new_privs.");
+ INFO("Failed to retrieve lxc.no_new_privs");
return false;
}
/* Set currently active setting. */
- if (!c->set_config_item(c, "lxc.no_new_privs", val)) {
- free(val);
- return false;
- }
+ bret = c->set_config_item(c, "lxc.no_new_privs", val);
free(val);
-
- return true;
+ return bret;
}
static signed long get_personality(const char *name, const char *lxcpath)
return ret;
}
-int lxc_attach(const char *name, const char *lxcpath,
- lxc_attach_exec_t exec_function, void *exec_payload,
- lxc_attach_options_t *options, pid_t *attached_process)
-{
- int i, ret, status;
- int ipc_sockets[2];
- char *cwd, *new_cwd;
- signed long personality;
- pid_t attached_pid, expected, init_pid, pid;
+struct attach_clone_payload {
+ int ipc_socket;
+ int terminal_slave_fd;
+ lxc_attach_options_t *options;
struct lxc_proc_context_info *init_ctx;
+ lxc_attach_exec_t exec_function;
+ void *exec_payload;
+};
- ret = access("/proc/self/ns", X_OK);
- if (ret) {
- ERROR("Does this kernel version support namespaces?");
- return -1;
+static void lxc_put_attach_clone_payload(struct attach_clone_payload *p)
+{
+ if (p->ipc_socket >= 0) {
+ shutdown(p->ipc_socket, SHUT_RDWR);
+ close(p->ipc_socket);
+ p->ipc_socket = -EBADF;
}
- if (!options)
- options = &attach_static_default_options;
-
- init_pid = lxc_cmd_get_init_pid(name, lxcpath);
- if (init_pid < 0) {
- ERROR("Failed to get init pid.");
- return -1;
+ if (p->terminal_slave_fd >= 0) {
+ close(p->terminal_slave_fd);
+ p->terminal_slave_fd = -EBADF;
}
- init_ctx = lxc_proc_get_context_info(init_pid);
- if (!init_ctx) {
- ERROR("Failed to get context of init process: %ld", (long)init_pid);
- return -1;
+ if (p->init_ctx) {
+ lxc_proc_put_context_info(p->init_ctx);
+ p->init_ctx = NULL;
}
+}
- personality = get_personality(name, lxcpath);
- if (init_ctx->personality < 0) {
- ERROR("Failed to get personality of the container");
- lxc_proc_put_context_info(init_ctx);
- return -1;
+static int attach_child_main(struct attach_clone_payload *payload)
+{
+ int fd, lsm_fd, ret;
+ uid_t new_uid;
+ gid_t new_gid;
+ lxc_attach_options_t* options = payload->options;
+ struct lxc_proc_context_info* init_ctx = payload->init_ctx;
+ bool needs_lsm = (options->namespaces & CLONE_NEWNS) &&
+ (options->attach_flags & LXC_ATTACH_LSM) &&
+ init_ctx->lsm_label;
+
+ /* A description of the purpose of this functionality is provided in the
+ * lxc-attach(1) manual page. We have to remount here and not in the
+ * parent process, otherwise /proc may not properly reflect the new pid
+ * namespace.
+ */
+ if (!(options->namespaces & CLONE_NEWNS) &&
+ (options->attach_flags & LXC_ATTACH_REMOUNT_PROC_SYS)) {
+ ret = lxc_attach_remount_sys_proc();
+ if (ret < 0)
+ goto on_error;
+ TRACE("Remounted \"/proc\" and \"/sys\"");
}
- init_ctx->personality = personality;
- init_ctx->container = lxc_container_new(name, lxcpath);
- if (!init_ctx->container) {
- lxc_proc_put_context_info(init_ctx);
- return -1;
+/* Now perform additional attachments. */
+#if HAVE_SYS_PERSONALITY_H
+ if (options->attach_flags & LXC_ATTACH_SET_PERSONALITY) {
+ long new_personality;
+
+ if (options->personality < 0)
+ new_personality = init_ctx->personality;
+ else
+ new_personality = options->personality;
+ ret = personality(new_personality);
+ if (ret < 0)
+ goto on_error;
+ TRACE("Set new personality");
}
+#endif
- if (!init_ctx->container->lxc_conf) {
- init_ctx->container->lxc_conf = lxc_conf_init();
- if (!init_ctx->container->lxc_conf) {
- lxc_proc_put_context_info(init_ctx);
- return -ENOMEM;
- }
+ if (options->attach_flags & LXC_ATTACH_DROP_CAPABILITIES) {
+ ret = lxc_attach_drop_privs(init_ctx);
+ if (ret < 0)
+ goto on_error;
+ TRACE("Dropped capabilities");
}
- if (!fetch_seccomp(init_ctx->container, options))
- WARN("Failed to get seccomp policy.");
+ /* Always set the environment (specify (LXC_ATTACH_KEEP_ENV, NULL, NULL)
+ * if you want this to be a no-op).
+ */
+ ret = lxc_attach_set_environment(init_ctx,
+ options->env_policy,
+ options->extra_env_vars,
+ options->extra_keep_env);
+ if (ret < 0)
+ goto on_error;
+ TRACE("Set up environment");
+
+ /* This remark only affects fully unprivileged containers:
+ * Receive fd for LSM security module before we set{g,u}id(). The reason
+ * is that on set{g,u}id() the kernel will a) make us undumpable and b)
+ * we will change our effective uid. This means our effective uid will
+ * be different from the effective uid of the process that created us
+ * which means that this processs no longer has capabilities in our
+ * namespace including CAP_SYS_PTRACE. This means we will not be able to
+ * read and /proc/<pid> files for the process anymore when /proc is
+ * mounted with hidepid={1,2}. So let's get the lsm label fd before the
+ * set{g,u}id().
+ */
+ if (needs_lsm) {
+ ret = lxc_abstract_unix_recv_fds(payload->ipc_socket, &lsm_fd, 1, NULL, 0);
+ if (ret <= 0) {
+ if (ret < 0)
+ SYSERROR("Failed to receive lsm label fd");
- if (!no_new_privs(init_ctx->container, options))
- WARN("Could not determine whether PR_SET_NO_NEW_PRIVS is set.");
+ goto on_error;
+ }
- cwd = getcwd(NULL, 0);
+ TRACE("Received LSM label file descriptor %d from parent", lsm_fd);
+ }
- /* Determine which namespaces the container was created with
- * by asking lxc-start, if necessary.
- */
- if (options->namespaces == -1) {
- options->namespaces = lxc_cmd_get_clone_flags(name, lxcpath);
- /* call failed */
- if (options->namespaces == -1) {
+ if (options->stdin_fd > 0 && isatty(options->stdin_fd)) {
+ ret = lxc_make_controlling_terminal(options->stdin_fd);
+ if (ret < 0)
+ goto on_error;
+ }
+
+ /* Set {u,g}id. */
+ new_uid = 0;
+ new_gid = 0;
+ /* Ignore errors, we will fall back to root in that case (/proc was not
+ * mounted etc.).
+ */
+ if (options->namespaces & CLONE_NEWUSER)
+ lxc_attach_get_init_uidgid(&new_uid, &new_gid);
+
+ if (options->uid != (uid_t)-1)
+ new_uid = options->uid;
+ if (options->gid != (gid_t)-1)
+ new_gid = options->gid;
+
+ /* Try to set the {u,g}id combination. */
+ if (new_uid != 0 || new_gid != 0 || options->namespaces & CLONE_NEWUSER) {
+ ret = lxc_switch_uid_gid(new_uid, new_gid);
+ if (ret < 0)
+ goto on_error;
+ }
+
+ ret = lxc_setgroups(0, NULL);
+ if (ret < 0 && errno != EPERM)
+ goto on_error;
+
+ if ((init_ctx->container && init_ctx->container->lxc_conf &&
+ init_ctx->container->lxc_conf->no_new_privs) ||
+ (options->attach_flags & LXC_ATTACH_NO_NEW_PRIVS)) {
+ ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
+ if (ret < 0)
+ goto on_error;
+ TRACE("Set PR_SET_NO_NEW_PRIVS");
+ }
+
+ if (needs_lsm) {
+ bool on_exec;
+
+ /* Change into our new LSM profile. */
+ on_exec = options->attach_flags & LXC_ATTACH_LSM_EXEC ? true : false;
+ ret = lsm_process_label_set_at(lsm_fd, init_ctx->lsm_label, on_exec);
+ close(lsm_fd);
+ if (ret < 0)
+ goto on_error;
+ TRACE("Set %s LSM label to \"%s\"", lsm_name(), init_ctx->lsm_label);
+ }
+
+ if (init_ctx->container && init_ctx->container->lxc_conf &&
+ init_ctx->container->lxc_conf->seccomp) {
+ ret = lxc_seccomp_load(init_ctx->container->lxc_conf);
+ if (ret < 0)
+ goto on_error;
+ TRACE("Loaded seccomp profile");
+ }
+ shutdown(payload->ipc_socket, SHUT_RDWR);
+ close(payload->ipc_socket);
+ payload->ipc_socket = -EBADF;
+ lxc_proc_put_context_info(init_ctx);
+ payload->init_ctx = NULL;
+
+ /* The following is done after the communication socket is shut down.
+ * That way, all errors that might (though unlikely) occur up until this
+ * point will have their messages printed to the original stderr (if
+ * logging is so configured) and not the fd the user supplied, if any.
+ */
+
+ /* Fd handling for stdin, stdout and stderr; ignore errors here, user
+ * may want to make sure the fds are closed, for example.
+ */
+ if (options->stdin_fd >= 0 && options->stdin_fd != STDIN_FILENO)
+ (void)dup2(options->stdin_fd, STDIN_FILENO);
+
+ if (options->stdout_fd >= 0 && options->stdout_fd != STDOUT_FILENO)
+ (void)dup2(options->stdout_fd, STDOUT_FILENO);
+
+ if (options->stderr_fd >= 0 && options->stderr_fd != STDERR_FILENO)
+ (void)dup2(options->stderr_fd, STDERR_FILENO);
+
+ /* close the old fds */
+ if (options->stdin_fd > STDERR_FILENO)
+ close(options->stdin_fd);
+
+ if (options->stdout_fd > STDERR_FILENO)
+ close(options->stdout_fd);
+
+ if (options->stderr_fd > STDERR_FILENO)
+ close(options->stderr_fd);
+
+ /* Try to remove FD_CLOEXEC flag from stdin/stdout/stderr, but also
+ * here, ignore errors.
+ */
+ for (fd = STDIN_FILENO; fd <= STDERR_FILENO; fd++) {
+ ret = fd_cloexec(fd, false);
+ if (ret < 0) {
+ SYSERROR("Failed to clear FD_CLOEXEC from file descriptor %d", fd);
+ goto on_error;
+ }
+ }
+
+ if (options->attach_flags & LXC_ATTACH_TERMINAL) {
+ ret = lxc_terminal_prepare_login(payload->terminal_slave_fd);
+ if (ret < 0) {
+ SYSERROR("Failed to prepare terminal file descriptor %d", payload->terminal_slave_fd);
+ goto on_error;
+ }
+ TRACE("Prepared terminal file descriptor %d", payload->terminal_slave_fd);
+ }
+
+ /* We're done, so we can now do whatever the user intended us to do. */
+ rexit(payload->exec_function(payload->exec_payload));
+
+on_error:
+ lxc_put_attach_clone_payload(payload);
+ rexit(EXIT_FAILURE);
+}
+
+static int lxc_attach_terminal(struct lxc_conf *conf,
+ struct lxc_terminal *terminal)
+{
+ int ret;
+
+ lxc_terminal_init(terminal);
+
+ ret = lxc_terminal_create(terminal);
+ if (ret < 0) {
+ SYSERROR("Failed to create terminal");
+ return -1;
+ }
+
+ /* Shift ttys to container. */
+ ret = lxc_terminal_map_ids(conf, terminal);
+ if (ret < 0) {
+ ERROR("Failed to chown terminal");
+ goto on_error;
+ }
+
+ return 0;
+
+on_error:
+ lxc_terminal_delete(terminal);
+ lxc_terminal_conf_free(terminal);
+ return -1;
+}
+
+static int lxc_attach_terminal_mainloop_init(struct lxc_terminal *terminal,
+ struct lxc_epoll_descr *descr)
+{
+ int ret;
+
+ ret = lxc_mainloop_open(descr);
+ if (ret < 0) {
+ ERROR("Failed to create mainloop");
+ return -1;
+ }
+
+ ret = lxc_terminal_mainloop_add(descr, terminal);
+ if (ret < 0) {
+ ERROR("Failed to add handlers to mainloop");
+ lxc_mainloop_close(descr);
+ return -1;
+ }
+
+ return 0;
+}
+
+static inline void lxc_attach_terminal_close_master(struct lxc_terminal *terminal)
+{
+ if (terminal->master < 0)
+ return;
+
+ close(terminal->master);
+ terminal->master = -EBADF;
+}
+
+static inline void lxc_attach_terminal_close_slave(struct lxc_terminal *terminal)
+{
+ if (terminal->slave < 0)
+ return;
+
+ close(terminal->slave);
+ terminal->slave = -EBADF;
+}
+
+static inline void lxc_attach_terminal_close_peer(struct lxc_terminal *terminal)
+{
+ if (terminal->peer < 0)
+ return;
+
+ close(terminal->peer);
+ terminal->peer = -EBADF;
+}
+
+static inline void lxc_attach_terminal_close_log(struct lxc_terminal *terminal)
+{
+ if (terminal->log_fd < 0)
+ return;
+
+ close(terminal->log_fd);
+ terminal->log_fd = -EBADF;
+}
+
+int lxc_attach(const char *name, const char *lxcpath,
+ lxc_attach_exec_t exec_function, void *exec_payload,
+ lxc_attach_options_t *options, pid_t *attached_process)
+{
+ int i, ret, status;
+ int ipc_sockets[2];
+ char *cwd, *new_cwd;
+ signed long personality;
+ pid_t attached_pid, init_pid, pid;
+ struct lxc_proc_context_info *init_ctx;
+ struct lxc_terminal terminal;
+ struct lxc_conf *conf;
+ struct attach_clone_payload payload = {0};
+
+ ret = access("/proc/self/ns", X_OK);
+ if (ret) {
+ ERROR("Does this kernel version support namespaces?");
+ return -1;
+ }
+
+ if (!options)
+ options = &attach_static_default_options;
+
+ init_pid = lxc_cmd_get_init_pid(name, lxcpath);
+ if (init_pid < 0) {
+ ERROR("Failed to get init pid");
+ return -1;
+ }
+
+ init_ctx = lxc_proc_get_context_info(init_pid);
+ if (!init_ctx) {
+ ERROR("Failed to get context of init process: %ld", (long)init_pid);
+ return -1;
+ }
+
+ personality = get_personality(name, lxcpath);
+ if (init_ctx->personality < 0) {
+ ERROR("Failed to get personality of the container");
+ lxc_proc_put_context_info(init_ctx);
+ return -1;
+ }
+ init_ctx->personality = personality;
+
+ init_ctx->container = lxc_container_new(name, lxcpath);
+ if (!init_ctx->container) {
+ lxc_proc_put_context_info(init_ctx);
+ return -1;
+ }
+
+ if (!init_ctx->container->lxc_conf) {
+ init_ctx->container->lxc_conf = lxc_conf_init();
+ if (!init_ctx->container->lxc_conf) {
+ lxc_proc_put_context_info(init_ctx);
+ return -ENOMEM;
+ }
+ }
+ conf = init_ctx->container->lxc_conf;
+
+ if (!fetch_seccomp(init_ctx->container, options))
+ WARN("Failed to get seccomp policy");
+
+ if (!no_new_privs(init_ctx->container, options))
+ WARN("Could not determine whether PR_SET_NO_NEW_PRIVS is set");
+
+ cwd = getcwd(NULL, 0);
+
+ /* Determine which namespaces the container was created with
+ * by asking lxc-start, if necessary.
+ */
+ if (options->namespaces == -1) {
+ options->namespaces = lxc_cmd_get_clone_flags(name, lxcpath);
+ /* call failed */
+ if (options->namespaces == -1) {
ERROR("Failed to automatically determine the "
"namespaces which the container uses");
free(cwd);
}
}
- pid = getpid();
+ pid = lxc_raw_getpid();
for (i = 0; i < LXC_NS_MAX; i++) {
int j, saved_errno;
return -1;
}
+ if (options->attach_flags & LXC_ATTACH_TERMINAL) {
+ ret = lxc_attach_terminal(conf, &terminal);
+ if (ret < 0) {
+ ERROR("Failed to setup new terminal");
+ free(cwd);
+ lxc_proc_put_context_info(init_ctx);
+ return -1;
+ }
+
+ terminal.log_fd = options->log_fd;
+ } else {
+ lxc_terminal_init(&terminal);
+ }
+
/* Create a socket pair for IPC communication; set SOCK_CLOEXEC in order
* to make sure we don't irritate other threads that want to fork+exec
* away
*/
ret = socketpair(PF_LOCAL, SOCK_STREAM | SOCK_CLOEXEC, 0, ipc_sockets);
if (ret < 0) {
- SYSERROR("Could not set up required IPC mechanism for attaching.");
+ SYSERROR("Could not set up required IPC mechanism for attaching");
free(cwd);
lxc_proc_put_context_info(init_ctx);
return -1;
}
- /* Create intermediate subprocess, three reasons:
- * 1. Runs all pthread_atfork handlers and the child will no
- * longer be threaded (we can't properly setns() in a threaded
- * process).
- * 2. We can't setns() in the child itself, since we want to make
+ /* Create intermediate subprocess, two reasons:
+ * 1. We can't setns() in the child itself, since we want to make
* sure we are properly attached to the pidns.
- * 3. Also, the initial thread has to put the attached process
+ * 2. Also, the initial thread has to put the attached process
* into the cgroup, which we can only do if we didn't already
* setns() (otherwise, user namespaces will hate us).
*/
pid = fork();
if (pid < 0) {
- SYSERROR("Failed to create first subprocess.");
+ SYSERROR("Failed to create first subprocess");
free(cwd);
lxc_proc_put_context_info(init_ctx);
return -1;
}
if (pid) {
- int procfd = -1;
+ int ret_parent = -1;
pid_t to_cleanup_pid = pid;
+ struct lxc_epoll_descr descr = {0};
- /* close file namespace descriptors */
- lxc_proc_close_ns_fd(init_ctx);
-
- /* Initial thread, we close the socket that is for the
- * subprocesses.
- */
+ /* close unneeded file descriptors */
close(ipc_sockets[1]);
free(cwd);
+ lxc_proc_close_ns_fd(init_ctx);
+ if (options->attach_flags & LXC_ATTACH_TERMINAL)
+ lxc_attach_terminal_close_slave(&terminal);
/* Attach to cgroup, if requested. */
if (options->attach_flags & LXC_ATTACH_MOVE_TO_CGROUP) {
- if (!cgroup_attach(name, lxcpath, pid))
+ struct cgroup_ops *cgroup_ops;
+
+ cgroup_ops = cgroup_init(NULL);
+ if (!cgroup_ops)
+ goto on_error;
+
+ if (!cgroup_ops->attach(cgroup_ops, name, lxcpath, pid))
+ goto on_error;
+
+ cgroup_exit(cgroup_ops);
+ TRACE("Moved intermediate process %d into container's cgroups", pid);
+ }
+
+ /* Setup /proc limits */
+ if (!lxc_list_empty(&conf->procs)) {
+ ret = setup_proc_filesystem(&conf->procs, pid);
+ if (ret < 0)
goto on_error;
}
/* Setup resource limits */
- if (!lxc_list_empty(&init_ctx->container->lxc_conf->limits))
- if (setup_resource_limits(&init_ctx->container->lxc_conf->limits, pid) < 0)
+ if (!lxc_list_empty(&conf->limits)) {
+ ret = setup_resource_limits(&conf->limits, pid);
+ if (ret < 0)
goto on_error;
+ }
- /* Open /proc before setns() to the containers namespace so we
- * don't rely on any information from inside the container.
- */
- procfd = open("/proc", O_DIRECTORY | O_RDONLY | O_CLOEXEC);
- if (procfd < 0) {
- SYSERROR("Unable to open /proc.");
- goto on_error;
+ if (options->attach_flags & LXC_ATTACH_TERMINAL) {
+ ret = lxc_attach_terminal_mainloop_init(&terminal, &descr);
+ if (ret < 0)
+ goto on_error;
+ TRACE("Initialized terminal mainloop");
}
/* Let the child process know to go ahead. */
status = 0;
ret = lxc_write_nointr(ipc_sockets[0], &status, sizeof(status));
- if (ret <= 0) {
- ERROR("Intended to send sequence number 0: %s.",
- strerror(errno));
- goto on_error;
- }
+ if (ret != sizeof(status))
+ goto close_mainloop;
+ TRACE("Told intermediate process to start initializing");
/* Get pid of attached process from intermediate process. */
- ret = lxc_read_nointr_expect(ipc_sockets[0], &attached_pid,
- sizeof(attached_pid), NULL);
- if (ret <= 0) {
- if (ret != 0)
- ERROR("Expected to receive pid: %s.", strerror(errno));
- goto on_error;
- }
+ ret = lxc_read_nointr(ipc_sockets[0], &attached_pid, sizeof(attached_pid));
+ if (ret != sizeof(attached_pid))
+ goto close_mainloop;
+ TRACE("Received pid %d of attached process in parent pid namespace", attached_pid);
/* Ignore SIGKILL (CTRL-C) and SIGQUIT (CTRL-\) - issue #313. */
if (options->stdin_fd == 0) {
/* Reap intermediate process. */
ret = wait_for_pid(pid);
if (ret < 0)
- goto on_error;
+ goto close_mainloop;
+ TRACE("Intermediate process %d exited", pid);
/* We will always have to reap the attached process now. */
to_cleanup_pid = attached_pid;
- /* Tell attached process it may start initializing. */
- status = 0;
- ret = lxc_write_nointr(ipc_sockets[0], &status, sizeof(status));
- if (ret <= 0) {
- ERROR("Intended to send sequence number 0: %s.", strerror(errno));
- goto on_error;
- }
-
- /* Wait for the attached process to finish initializing. */
- expected = 1;
- ret = lxc_read_nointr_expect(ipc_sockets[0], &status,
- sizeof(status), &expected);
- if (ret <= 0) {
- if (ret != 0)
- ERROR("Expected to receive sequence number 1: %s.", strerror(errno));
- goto on_error;
- }
-
- /* Tell attached process we're done. */
- status = 2;
- ret = lxc_write_nointr(ipc_sockets[0], &status, sizeof(status));
- if (ret <= 0) {
- ERROR("Intended to send sequence number 2: %s.", strerror(errno));
- goto on_error;
- }
-
- /* Wait for the (grand)child to tell us that it's ready to set
- * up its LSM labels.
- */
- expected = 3;
- ret = lxc_read_nointr_expect(ipc_sockets[0], &status,
- sizeof(status), &expected);
- if (ret <= 0) {
- ERROR("Expected to receive sequence number 3: %s.",
- strerror(errno));
- goto on_error;
- }
-
/* Open LSM fd and send it to child. */
if ((options->namespaces & CLONE_NEWNS) &&
(options->attach_flags & LXC_ATTACH_LSM) &&
init_ctx->lsm_label) {
- int on_exec, saved_errno;
- int labelfd = -1;
+ int ret = -1;
+ int labelfd;
+ bool on_exec;
- on_exec = options->attach_flags & LXC_ATTACH_LSM_EXEC ? 1 : 0;
- /* Open fd for the LSM security module. */
- labelfd = lsm_openat(procfd, attached_pid, on_exec);
+ on_exec = options->attach_flags & LXC_ATTACH_LSM_EXEC ? true : false;
+ labelfd = lsm_process_label_fd_get(attached_pid, on_exec);
if (labelfd < 0)
- goto on_error;
+ goto close_mainloop;
+ TRACE("Opened LSM label file descriptor %d", labelfd);
/* Send child fd of the LSM security module to write to. */
ret = lxc_abstract_unix_send_fds(ipc_sockets[0], &labelfd, 1, NULL, 0);
- saved_errno = errno;
- close(labelfd);
if (ret <= 0) {
- ERROR("Intended to send file descriptor %d: %s.", labelfd, strerror(saved_errno));
- goto on_error;
+ if (ret < 0)
+ SYSERROR("Failed to send lsm label fd");
+
+ close(labelfd);
+ goto close_mainloop;
}
- }
- if (procfd >= 0)
- close(procfd);
- /* Now shut down communication with child, we're done. */
- shutdown(ipc_sockets[0], SHUT_RDWR);
- close(ipc_sockets[0]);
- lxc_proc_put_context_info(init_ctx);
+ close(labelfd);
+ TRACE("Sent LSM label file descriptor %d to child", labelfd);
+ }
/* We're done, the child process should now execute whatever it
* is that the user requested. The parent can now track it with
*/
*attached_process = attached_pid;
- return 0;
- on_error:
- /* First shut down the socket, then wait for the pid, otherwise
- * the pid we're waiting for may never exit.
- */
- if (procfd >= 0)
- close(procfd);
+ /* Now shut down communication with child, we're done. */
shutdown(ipc_sockets[0], SHUT_RDWR);
close(ipc_sockets[0]);
- if (to_cleanup_pid)
+ ipc_sockets[0] = -1;
+
+ ret_parent = 0;
+ to_cleanup_pid = -1;
+ if (options->attach_flags & LXC_ATTACH_TERMINAL) {
+ ret = lxc_mainloop(&descr, -1);
+ if (ret < 0) {
+ ret_parent = -1;
+ to_cleanup_pid = attached_pid;
+ }
+ }
+
+ close_mainloop:
+ if (options->attach_flags & LXC_ATTACH_TERMINAL)
+ lxc_mainloop_close(&descr);
+
+ on_error:
+ if (ipc_sockets[0] >= 0) {
+ shutdown(ipc_sockets[0], SHUT_RDWR);
+ close(ipc_sockets[0]);
+ }
+
+ if (to_cleanup_pid > 0)
(void)wait_for_pid(to_cleanup_pid);
+
+ if (options->attach_flags & LXC_ATTACH_TERMINAL) {
+ lxc_terminal_delete(&terminal);
+ lxc_terminal_conf_free(&terminal);
+ }
lxc_proc_put_context_info(init_ctx);
- return -1;
+ return ret_parent;
}
- /* First subprocess begins here, we close the socket that is for the
- * initial thread.
- */
+ /* close unneeded file descriptors */
close(ipc_sockets[0]);
+ ipc_sockets[0] = -EBADF;
+ if (options->attach_flags & LXC_ATTACH_TERMINAL) {
+ lxc_attach_terminal_close_master(&terminal);
+ lxc_attach_terminal_close_peer(&terminal);
+ lxc_attach_terminal_close_log(&terminal);
+ }
/* Wait for the parent to have setup cgroups. */
- expected = 0;
- status = -1;
- ret = lxc_read_nointr_expect(ipc_sockets[1], &status, sizeof(status),
- &expected);
- if (ret <= 0) {
- ERROR("Expected to receive sequence number 0: %s.", strerror(errno));
+ ret = lxc_read_nointr(ipc_sockets[1], &status, sizeof(status));
+ if (ret != sizeof(status)) {
shutdown(ipc_sockets[1], SHUT_RDWR);
lxc_proc_put_context_info(init_ctx);
rexit(-1);
}
+ TRACE("Intermediate process starting to initialize");
/* Attach now, create another subprocess later, since pid namespaces
* only really affect the children of the current process.
*/
ret = lxc_attach_to_ns(init_pid, init_ctx);
if (ret < 0) {
- ERROR("Failed to enter namespaces.");
+ ERROR("Failed to enter namespaces");
shutdown(ipc_sockets[1], SHUT_RDWR);
lxc_proc_put_context_info(init_ctx);
rexit(-1);
}
free(cwd);
- /* Now create the real child process. */
- {
- struct attach_clone_payload payload = {
- .ipc_socket = ipc_sockets[1],
- .options = options,
- .init_ctx = init_ctx,
- .exec_function = exec_function,
- .exec_payload = exec_payload,
- };
- /* We use clone_parent here to make this subprocess a direct
- * child of the initial process. Then this intermediate process
- * can exit and the parent can directly track the attached
- * process.
- */
- pid = lxc_clone(attach_child_main, &payload, CLONE_PARENT);
- }
+ /* Create attached process. */
+ payload.ipc_socket = ipc_sockets[1];
+ payload.options = options;
+ payload.init_ctx = init_ctx;
+ payload.terminal_slave_fd = terminal.slave;
+ payload.exec_function = exec_function;
+ payload.exec_payload = exec_payload;
- /* Shouldn't happen, clone() should always return positive pid. */
- if (pid <= 0) {
- SYSERROR("Failed to create subprocess.");
+ pid = lxc_raw_clone(CLONE_PARENT);
+ if (pid < 0) {
+ SYSERROR("Failed to clone attached process");
shutdown(ipc_sockets[1], SHUT_RDWR);
lxc_proc_put_context_info(init_ctx);
rexit(-1);
}
+ if (pid == 0) {
+ ret = attach_child_main(&payload);
+ if (ret < 0)
+ ERROR("Failed to exec");
+ _exit(EXIT_FAILURE);
+ }
+ if (options->attach_flags & LXC_ATTACH_TERMINAL)
+ lxc_attach_terminal_close_slave(&terminal);
+
/* Tell grandparent the pid of the pid of the newly created child. */
ret = lxc_write_nointr(ipc_sockets[1], &pid, sizeof(pid));
if (ret != sizeof(pid)) {
* CLONE_PARENT) so the parent won't be able to reap it and the
* attached process will remain a zombie.
*/
- ERROR("Intended to send pid %d: %s.", pid, strerror(errno));
shutdown(ipc_sockets[1], SHUT_RDWR);
lxc_proc_put_context_info(init_ctx);
rexit(-1);
}
+ TRACE("Sending pid %d of attached process", pid);
/* The rest is in the hands of the initial and the attached process. */
lxc_proc_put_context_info(init_ctx);
rexit(0);
}
-static int attach_child_main(void* data)
-{
- int expected, fd, lsm_labelfd, ret, status;
- long flags;
-#if HAVE_SYS_PERSONALITY_H
- long new_personality;
-#endif
- uid_t new_uid;
- gid_t new_gid;
- struct attach_clone_payload* payload = (struct attach_clone_payload*)data;
- int ipc_socket = payload->ipc_socket;
- lxc_attach_options_t* options = payload->options;
- struct lxc_proc_context_info* init_ctx = payload->init_ctx;
-
- /* Wait for the initial thread to signal us that it's ready for us to
- * start initializing.
- */
- expected = 0;
- status = -1;
- ret = lxc_read_nointr_expect(ipc_socket, &status, sizeof(status), &expected);
- if (ret <= 0) {
- ERROR("Expected to receive sequence number 0: %s.", strerror(errno));
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- /* A description of the purpose of this functionality is provided in the
- * lxc-attach(1) manual page. We have to remount here and not in the
- * parent process, otherwise /proc may not properly reflect the new pid
- * namespace.
- */
- if (!(options->namespaces & CLONE_NEWNS) &&
- (options->attach_flags & LXC_ATTACH_REMOUNT_PROC_SYS)) {
- ret = lxc_attach_remount_sys_proc();
- if (ret < 0) {
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
- }
-
- /* Now perform additional attachments. */
-#if HAVE_SYS_PERSONALITY_H
- if (options->personality < 0)
- new_personality = init_ctx->personality;
- else
- new_personality = options->personality;
-
- if (options->attach_flags & LXC_ATTACH_SET_PERSONALITY) {
- ret = personality(new_personality);
- if (ret < 0) {
- SYSERROR("Could not ensure correct architecture.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
- }
-#endif
-
- if (options->attach_flags & LXC_ATTACH_DROP_CAPABILITIES) {
- ret = lxc_attach_drop_privs(init_ctx);
- if (ret < 0) {
- ERROR("Could not drop privileges.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
- }
-
- /* Always set the environment (specify (LXC_ATTACH_KEEP_ENV, NULL, NULL)
- * if you want this to be a no-op).
- */
- ret = lxc_attach_set_environment(options->env_policy,
- options->extra_env_vars,
- options->extra_keep_env);
- if (ret < 0) {
- ERROR("Could not set initial environment for attached process.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- /* Set {u,g}id. */
- new_uid = 0;
- new_gid = 0;
- /* Ignore errors, we will fall back to root in that case (/proc was not
- * mounted etc.).
- */
- if (options->namespaces & CLONE_NEWUSER)
- lxc_attach_get_init_uidgid(&new_uid, &new_gid);
-
- if (options->uid != (uid_t)-1)
- new_uid = options->uid;
- if (options->gid != (gid_t)-1)
- new_gid = options->gid;
-
- /* Setup the controlling tty. */
- if (options->stdin_fd && isatty(options->stdin_fd)) {
- if (setsid() < 0) {
- SYSERROR("Unable to setsid.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- if (ioctl(options->stdin_fd, TIOCSCTTY, (char *)NULL) < 0) {
- SYSERROR("Unable to set TIOCSTTY.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
- }
-
- /* Try to set the {u,g}id combination. */
- if ((new_gid != 0 || options->namespaces & CLONE_NEWUSER)) {
- if (setgid(new_gid) || setgroups(0, NULL)) {
- SYSERROR("Switching to container gid.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
- }
- if ((new_uid != 0 || options->namespaces & CLONE_NEWUSER) &&
- setuid(new_uid)) {
- SYSERROR("Switching to container uid.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- /* Tell initial process it may now put us into cgroups. */
- status = 1;
- ret = lxc_write_nointr(ipc_socket, &status, sizeof(status));
- if (ret != sizeof(status)) {
- ERROR("Intended to send sequence number 1: %s.", strerror(errno));
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- /* Wait for the initial thread to signal us that it has done everything
- * for us when it comes to cgroups etc.
- */
- expected = 2;
- status = -1;
- ret = lxc_read_nointr_expect(ipc_socket, &status, sizeof(status), &expected);
- if (ret <= 0) {
- ERROR("Expected to receive sequence number 2: %s", strerror(errno));
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- if ((init_ctx->container && init_ctx->container->lxc_conf &&
- init_ctx->container->lxc_conf->no_new_privs) ||
- (options->attach_flags & LXC_ATTACH_NO_NEW_PRIVS)) {
- if (prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0) < 0) {
- SYSERROR("PR_SET_NO_NEW_PRIVS could not be set. "
- "Process can use execve() gainable "
- "privileges.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
- INFO("PR_SET_NO_NEW_PRIVS is set. Process cannot use execve() "
- "gainable privileges.");
- }
-
- /* Tell the (grand)parent to send us LSM label fd. */
- status = 3;
- ret = lxc_write_nointr(ipc_socket, &status, sizeof(status));
- if (ret <= 0) {
- ERROR("Intended to send sequence number 3: %s.", strerror(errno));
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- if ((options->namespaces & CLONE_NEWNS) &&
- (options->attach_flags & LXC_ATTACH_LSM) && init_ctx->lsm_label) {
- int on_exec;
- /* Receive fd for LSM security module. */
- ret = lxc_abstract_unix_recv_fds(ipc_socket, &lsm_labelfd, 1, NULL, 0);
- if (ret <= 0) {
- ERROR("Expected to receive file descriptor: %s.", strerror(errno));
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- /* Change into our new LSM profile. */
- on_exec = options->attach_flags & LXC_ATTACH_LSM_EXEC ? 1 : 0;
- if (lsm_set_label_at(lsm_labelfd, on_exec, init_ctx->lsm_label) < 0) {
- SYSERROR("Failed to set LSM label.");
- shutdown(ipc_socket, SHUT_RDWR);
- close(lsm_labelfd);
- rexit(-1);
- }
- close(lsm_labelfd);
- }
-
- if (init_ctx->container && init_ctx->container->lxc_conf &&
- init_ctx->container->lxc_conf->seccomp &&
- (lxc_seccomp_load(init_ctx->container->lxc_conf) != 0)) {
- ERROR("Failed to load seccomp policy.");
- shutdown(ipc_socket, SHUT_RDWR);
- rexit(-1);
- }
-
- shutdown(ipc_socket, SHUT_RDWR);
- close(ipc_socket);
- lxc_proc_put_context_info(init_ctx);
-
- /* The following is done after the communication socket is shut down.
- * That way, all errors that might (though unlikely) occur up until this
- * point will have their messages printed to the original stderr (if
- * logging is so configured) and not the fd the user supplied, if any.
- */
-
- /* Fd handling for stdin, stdout and stderr; ignore errors here, user
- * may want to make sure the fds are closed, for example.
- */
- if (options->stdin_fd >= 0 && options->stdin_fd != 0)
- dup2(options->stdin_fd, 0);
- if (options->stdout_fd >= 0 && options->stdout_fd != 1)
- dup2(options->stdout_fd, 1);
- if (options->stderr_fd >= 0 && options->stderr_fd != 2)
- dup2(options->stderr_fd, 2);
-
- /* close the old fds */
- if (options->stdin_fd > 2)
- close(options->stdin_fd);
- if (options->stdout_fd > 2)
- close(options->stdout_fd);
- if (options->stderr_fd > 2)
- close(options->stderr_fd);
-
- /* Try to remove FD_CLOEXEC flag from stdin/stdout/stderr, but also
- * here, ignore errors.
- */
- for (fd = 0; fd <= 2; fd++) {
- flags = fcntl(fd, F_GETFL);
- if (flags < 0)
- continue;
- if (flags & FD_CLOEXEC)
- if (fcntl(fd, F_SETFL, flags & ~FD_CLOEXEC) < 0)
- SYSERROR("Unable to clear FD_CLOEXEC from file descriptor.");
- }
-
- /* We're done, so we can now do whatever the user intended us to do. */
- rexit(payload->exec_function(payload->exec_payload));
-}
-
int lxc_attach_run_command(void* payload)
{
lxc_attach_command_t* cmd = (lxc_attach_command_t*)payload;
int lxc_attach_run_shell(void* payload)
{
uid_t uid;
- struct passwd *passwd;
+ struct passwd pwent;
+ struct passwd *pwentp = NULL;
char *user_shell;
+ char *buf;
+ size_t bufsize;
+ int ret;
/* Ignore payload parameter. */
(void)payload;
uid = getuid();
- passwd = getpwuid(uid);
+
+ bufsize = sysconf(_SC_GETPW_R_SIZE_MAX);
+ if (bufsize == -1)
+ bufsize = 1024;
+
+ buf = malloc(bufsize);
+ if (buf) {
+ ret = getpwuid_r(uid, &pwent, buf, bufsize, &pwentp);
+ if (!pwentp) {
+ if (ret == 0)
+ WARN("Could not find matched password record.");
+
+ WARN("Failed to get password record - %u", uid);
+ }
+ }
/* This probably happens because of incompatible nss implementations in
* host and container (remember, this code is still using the host's
* the information by spawning a [getent passwd uid] process and parsing
* the result.
*/
- if (!passwd)
+ if (!pwentp)
user_shell = lxc_attach_getpwshell(uid);
else
- user_shell = passwd->pw_shell;
+ user_shell = pwent.pw_shell;
if (user_shell)
execlp(user_shell, user_shell, (char *)NULL);
*/
execlp("/bin/sh", "/bin/sh", (char *)NULL);
SYSERROR("Failed to execute shell");
- if (!passwd)
+ if (!pwentp)
free(user_shell);
+ free(buf);
return -1;
}