virt: rework container detection logic

Instead of accessing /proc/1/environ directly, trying to read the
$container variable from it, let's make PID 1 save the contents of that
variable to /run/systemd/container. This allows us to detect containers
without the need for CAP_SYS_PTRACE, which allows us to drop it from a
number of daemons and from the file capabilities of systemd-detect-virt.

Also, don't consider chroot a container technology anymore. After all,
we don't consider file system namespaces container technology anymore,
and hence chroot() should be considered a container even less.
This commit is contained in:
Lennart Poettering
2014-05-28 18:37:11 +08:00
parent d2edfae0f9
commit fdd2531170
4 changed files with 42 additions and 23 deletions

View File

@@ -1798,9 +1798,6 @@ systemd_detect_virt_SOURCES = \
systemd_detect_virt_LDADD = \
libsystemd-shared.la
systemd-detect-virt-install-hook:
-$(SETCAP) cap_dac_override,cap_sys_ptrace=ep $(DESTDIR)$(bindir)/systemd-detect-virt
INSTALL_EXEC_HOOKS += \
systemd-detect-virt-install-hook

View File

@@ -68,8 +68,6 @@ AC_PATH_PROG([XSLTPROC], [xsltproc])
AC_PATH_PROG([QUOTAON], [quotaon], [/usr/sbin/quotaon], [$PATH:/usr/sbin:/sbin])
AC_PATH_PROG([QUOTACHECK], [quotacheck], [/usr/sbin/quotacheck], [$PATH:/usr/sbin:/sbin])
AC_PATH_PROG([SETCAP], [setcap], [/usr/sbin/setcap], [$PATH:/usr/sbin:/sbin])
AC_PATH_PROG([KILL], [kill], [/usr/bin/kill], [$PATH:/usr/sbin:/sbin])
AC_PATH_PROG([KMOD], [kmod], [/usr/bin/kmod], [$PATH:/usr/sbin:/sbin])

View File

@@ -1261,6 +1261,16 @@ static int status_welcome(void) {
isempty(pretty_name) ? "Linux" : pretty_name);
}
static int write_container_id(void) {
const char *c;
c = getenv("container");
if (isempty(c))
return 0;
return write_string_file("/run/systemd/container", c);
}
int main(int argc, char *argv[]) {
Manager *m = NULL;
int r, retval = EXIT_FAILURE;
@@ -1544,6 +1554,8 @@ int main(int argc, char *argv[]) {
if (virtualization)
log_info("Detected virtualization '%s'.", virtualization);
write_container_id();
log_info("Detected architecture '%s'.", architecture_to_string(uname_architecture()));
if (in_initrd())

View File

@@ -217,8 +217,8 @@ int detect_container(const char **id) {
static thread_local int cached_found = -1;
static thread_local const char *cached_id = NULL;
_cleanup_free_ char *e = NULL;
const char *_id = NULL;
_cleanup_free_ char *m = NULL;
const char *_id = NULL, *e = NULL;
int r;
if (_likely_(cached_found >= 0)) {
@@ -229,17 +229,6 @@ int detect_container(const char **id) {
return cached_found;
}
/* Unfortunately many of these operations require root access
* in one way or another */
r = running_in_chroot();
if (r < 0)
return r;
if (r > 0) {
_id = "chroot";
goto finish;
}
/* /proc/vz exists in container and outside of the container,
* /proc/bc only outside of the container. */
if (access("/proc/vz", F_OK) >= 0 &&
@@ -249,11 +238,32 @@ int detect_container(const char **id) {
goto finish;
}
r = getenv_for_pid(1, "container", &e);
if (r < 0)
return r;
if (r == 0)
goto finish;
if (getpid() == 1) {
/* If we are PID 1 we can just check our own
* environment variable */
e = getenv("container");
if (isempty(e)) {
r = 0;
goto finish;
}
} else {
/* Otherwise, PID 1 dropped this information into a
* file in /run. This is better than accessing
* /proc/1/environ, since we don't need CAP_SYS_PTRACE
* for that. */
r = read_one_line_file("/run/systemd/container", &m);
if (r == -ENOENT) {
r = 0;
goto finish;
}
if (r < 0)
return r;
e = m;
}
/* We only recognize a selected few here, since we want to
* enforce a redacted namespace */
@@ -266,6 +276,8 @@ int detect_container(const char **id) {
else
_id = "other";
r = 1;
finish:
cached_found = r;