2019-02-07 02:03:09 +07:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
|
|
|
#define _GNU_SOURCE
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <errno.h>
|
|
|
|
#include <pwd.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <syscall.h>
|
|
|
|
#include <sys/capability.h>
|
|
|
|
#include <sys/types.h>
|
|
|
|
#include <sys/mount.h>
|
|
|
|
#include <sys/prctl.h>
|
|
|
|
#include <sys/wait.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <unistd.h>
|
|
|
|
#include <fcntl.h>
|
|
|
|
#include <stdbool.h>
|
|
|
|
#include <stdarg.h>
|
|
|
|
|
|
|
|
#ifndef CLONE_NEWUSER
|
|
|
|
# define CLONE_NEWUSER 0x10000000
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#define ROOT_USER 0
|
|
|
|
#define RESTRICTED_PARENT 1
|
|
|
|
#define ALLOWED_CHILD1 2
|
|
|
|
#define ALLOWED_CHILD2 3
|
|
|
|
#define NO_POLICY_USER 4
|
|
|
|
|
|
|
|
char* add_whitelist_policy_file = "/sys/kernel/security/safesetid/add_whitelist_policy";
|
|
|
|
|
|
|
|
static void die(char *fmt, ...)
|
|
|
|
{
|
|
|
|
va_list ap;
|
|
|
|
va_start(ap, fmt);
|
|
|
|
vfprintf(stderr, fmt, ap);
|
|
|
|
va_end(ap);
|
|
|
|
exit(EXIT_FAILURE);
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool vmaybe_write_file(bool enoent_ok, char *filename, char *fmt, va_list ap)
|
|
|
|
{
|
|
|
|
char buf[4096];
|
|
|
|
int fd;
|
|
|
|
ssize_t written;
|
|
|
|
int buf_len;
|
|
|
|
|
|
|
|
buf_len = vsnprintf(buf, sizeof(buf), fmt, ap);
|
|
|
|
if (buf_len < 0) {
|
|
|
|
printf("vsnprintf failed: %s\n",
|
|
|
|
strerror(errno));
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
if (buf_len >= sizeof(buf)) {
|
|
|
|
printf("vsnprintf output truncated\n");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
fd = open(filename, O_WRONLY);
|
|
|
|
if (fd < 0) {
|
|
|
|
if ((errno == ENOENT) && enoent_ok)
|
|
|
|
return true;
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
written = write(fd, buf, buf_len);
|
|
|
|
if (written != buf_len) {
|
|
|
|
if (written >= 0) {
|
|
|
|
printf("short write to %s\n", filename);
|
|
|
|
return false;
|
|
|
|
} else {
|
|
|
|
printf("write to %s failed: %s\n",
|
|
|
|
filename, strerror(errno));
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (close(fd) != 0) {
|
|
|
|
printf("close of %s failed: %s\n",
|
|
|
|
filename, strerror(errno));
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool write_file(char *filename, char *fmt, ...)
|
|
|
|
{
|
|
|
|
va_list ap;
|
|
|
|
bool ret;
|
|
|
|
|
|
|
|
va_start(ap, fmt);
|
|
|
|
ret = vmaybe_write_file(false, filename, fmt, ap);
|
|
|
|
va_end(ap);
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ensure_user_exists(uid_t uid)
|
|
|
|
{
|
|
|
|
struct passwd p;
|
|
|
|
|
|
|
|
FILE *fd;
|
|
|
|
char name_str[10];
|
|
|
|
|
|
|
|
if (getpwuid(uid) == NULL) {
|
|
|
|
memset(&p,0x00,sizeof(p));
|
|
|
|
fd=fopen("/etc/passwd","a");
|
|
|
|
if (fd == NULL)
|
|
|
|
die("couldn't open file\n");
|
|
|
|
if (fseek(fd, 0, SEEK_END))
|
|
|
|
die("couldn't fseek\n");
|
|
|
|
snprintf(name_str, 10, "%d", uid);
|
|
|
|
p.pw_name=name_str;
|
|
|
|
p.pw_uid=uid;
|
|
|
|
p.pw_gecos="Test account";
|
|
|
|
p.pw_dir="/dev/null";
|
|
|
|
p.pw_shell="/bin/false";
|
|
|
|
int value = putpwent(&p,fd);
|
|
|
|
if (value != 0)
|
|
|
|
die("putpwent failed\n");
|
|
|
|
if (fclose(fd))
|
|
|
|
die("fclose failed\n");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ensure_securityfs_mounted(void)
|
|
|
|
{
|
|
|
|
int fd = open(add_whitelist_policy_file, O_WRONLY);
|
|
|
|
if (fd < 0) {
|
|
|
|
if (errno == ENOENT) {
|
|
|
|
// Need to mount securityfs
|
|
|
|
if (mount("securityfs", "/sys/kernel/security",
|
|
|
|
"securityfs", 0, NULL) < 0)
|
|
|
|
die("mounting securityfs failed\n");
|
|
|
|
} else {
|
|
|
|
die("couldn't find securityfs for unknown reason\n");
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (close(fd) != 0) {
|
|
|
|
die("close of %s failed: %s\n",
|
|
|
|
add_whitelist_policy_file, strerror(errno));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void write_policies(void)
|
|
|
|
{
|
LSM: SafeSetID: rewrite userspace API to atomic updates
The current API of the SafeSetID LSM uses one write() per rule, and applies
each written rule instantly. This has several downsides:
- While a policy is being loaded, once a single parent-child pair has been
loaded, the parent is restricted to that specific child, even if
subsequent rules would allow transitions to other child UIDs. This means
that during policy loading, set*uid() can randomly fail.
- To replace the policy without rebooting, it is necessary to first flush
all old rules. This creates a time window in which no constraints are
placed on the use of CAP_SETUID.
- If we want to perform sanity checks on the final policy, this requires
that the policy isn't constructed in a piecemeal fashion without telling
the kernel when it's done.
Other kernel APIs - including things like the userns code and netfilter -
avoid this problem by performing updates atomically. Luckily, SafeSetID
hasn't landed in a stable (upstream) release yet, so maybe it's not too
late to completely change the API.
The new API for SafeSetID is: If you want to change the policy, open
"safesetid/whitelist_policy" and write the entire policy,
newline-delimited, in there.
Signed-off-by: Jann Horn <jannh@google.com>
Signed-off-by: Micah Morton <mortonm@chromium.org>
2019-04-10 23:56:05 +07:00
|
|
|
static char *policy_str =
|
|
|
|
"1:2\n"
|
2019-04-12 03:12:43 +07:00
|
|
|
"1:3\n"
|
|
|
|
"2:2\n"
|
|
|
|
"3:3\n";
|
2019-02-07 02:03:09 +07:00
|
|
|
ssize_t written;
|
|
|
|
int fd;
|
|
|
|
|
|
|
|
fd = open(add_whitelist_policy_file, O_WRONLY);
|
|
|
|
if (fd < 0)
|
|
|
|
die("cant open add_whitelist_policy file\n");
|
LSM: SafeSetID: rewrite userspace API to atomic updates
The current API of the SafeSetID LSM uses one write() per rule, and applies
each written rule instantly. This has several downsides:
- While a policy is being loaded, once a single parent-child pair has been
loaded, the parent is restricted to that specific child, even if
subsequent rules would allow transitions to other child UIDs. This means
that during policy loading, set*uid() can randomly fail.
- To replace the policy without rebooting, it is necessary to first flush
all old rules. This creates a time window in which no constraints are
placed on the use of CAP_SETUID.
- If we want to perform sanity checks on the final policy, this requires
that the policy isn't constructed in a piecemeal fashion without telling
the kernel when it's done.
Other kernel APIs - including things like the userns code and netfilter -
avoid this problem by performing updates atomically. Luckily, SafeSetID
hasn't landed in a stable (upstream) release yet, so maybe it's not too
late to completely change the API.
The new API for SafeSetID is: If you want to change the policy, open
"safesetid/whitelist_policy" and write the entire policy,
newline-delimited, in there.
Signed-off-by: Jann Horn <jannh@google.com>
Signed-off-by: Micah Morton <mortonm@chromium.org>
2019-04-10 23:56:05 +07:00
|
|
|
written = write(fd, policy_str, strlen(policy_str));
|
|
|
|
if (written != strlen(policy_str)) {
|
2019-02-07 02:03:09 +07:00
|
|
|
if (written >= 0) {
|
|
|
|
die("short write to %s\n", add_whitelist_policy_file);
|
|
|
|
} else {
|
|
|
|
die("write to %s failed: %s\n",
|
|
|
|
add_whitelist_policy_file, strerror(errno));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (close(fd) != 0) {
|
|
|
|
die("close of %s failed: %s\n",
|
|
|
|
add_whitelist_policy_file, strerror(errno));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool test_userns(bool expect_success)
|
|
|
|
{
|
|
|
|
uid_t uid;
|
|
|
|
char map_file_name[32];
|
|
|
|
size_t sz = sizeof(map_file_name);
|
|
|
|
pid_t cpid;
|
|
|
|
bool success;
|
|
|
|
|
|
|
|
uid = getuid();
|
|
|
|
|
|
|
|
int clone_flags = CLONE_NEWUSER;
|
|
|
|
cpid = syscall(SYS_clone, clone_flags, NULL);
|
|
|
|
if (cpid == -1) {
|
|
|
|
printf("clone failed");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (cpid == 0) { /* Code executed by child */
|
|
|
|
// Give parent 1 second to write map file
|
|
|
|
sleep(1);
|
|
|
|
exit(EXIT_SUCCESS);
|
|
|
|
} else { /* Code executed by parent */
|
|
|
|
if(snprintf(map_file_name, sz, "/proc/%d/uid_map", cpid) < 0) {
|
|
|
|
printf("preparing file name string failed");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
success = write_file(map_file_name, "0 0 1", uid);
|
|
|
|
return success == expect_success;
|
|
|
|
}
|
|
|
|
|
|
|
|
printf("should not reach here");
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void test_setuid(uid_t child_uid, bool expect_success)
|
|
|
|
{
|
|
|
|
pid_t cpid, w;
|
|
|
|
int wstatus;
|
|
|
|
|
|
|
|
cpid = fork();
|
|
|
|
if (cpid == -1) {
|
|
|
|
die("fork\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
if (cpid == 0) { /* Code executed by child */
|
|
|
|
setuid(child_uid);
|
|
|
|
if (getuid() == child_uid)
|
|
|
|
exit(EXIT_SUCCESS);
|
|
|
|
else
|
|
|
|
exit(EXIT_FAILURE);
|
|
|
|
} else { /* Code executed by parent */
|
|
|
|
do {
|
|
|
|
w = waitpid(cpid, &wstatus, WUNTRACED | WCONTINUED);
|
|
|
|
if (w == -1) {
|
|
|
|
die("waitpid\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
if (WIFEXITED(wstatus)) {
|
|
|
|
if (WEXITSTATUS(wstatus) == EXIT_SUCCESS) {
|
|
|
|
if (expect_success) {
|
|
|
|
return;
|
|
|
|
} else {
|
|
|
|
die("unexpected success\n");
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (expect_success) {
|
|
|
|
die("unexpected failure\n");
|
|
|
|
} else {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else if (WIFSIGNALED(wstatus)) {
|
|
|
|
if (WTERMSIG(wstatus) == 9) {
|
|
|
|
if (expect_success)
|
|
|
|
die("killed unexpectedly\n");
|
|
|
|
else
|
|
|
|
return;
|
|
|
|
} else {
|
|
|
|
die("unexpected signal: %d\n", wstatus);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
die("unexpected status: %d\n", wstatus);
|
|
|
|
}
|
|
|
|
} while (!WIFEXITED(wstatus) && !WIFSIGNALED(wstatus));
|
|
|
|
}
|
|
|
|
|
|
|
|
die("should not reach here\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ensure_users_exist(void)
|
|
|
|
{
|
|
|
|
ensure_user_exists(ROOT_USER);
|
|
|
|
ensure_user_exists(RESTRICTED_PARENT);
|
|
|
|
ensure_user_exists(ALLOWED_CHILD1);
|
|
|
|
ensure_user_exists(ALLOWED_CHILD2);
|
|
|
|
ensure_user_exists(NO_POLICY_USER);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void drop_caps(bool setid_retained)
|
|
|
|
{
|
|
|
|
cap_value_t cap_values[] = {CAP_SETUID, CAP_SETGID};
|
|
|
|
cap_t caps;
|
|
|
|
|
|
|
|
caps = cap_get_proc();
|
|
|
|
if (setid_retained)
|
|
|
|
cap_set_flag(caps, CAP_EFFECTIVE, 2, cap_values, CAP_SET);
|
|
|
|
else
|
|
|
|
cap_clear(caps);
|
|
|
|
cap_set_proc(caps);
|
|
|
|
cap_free(caps);
|
|
|
|
}
|
|
|
|
|
|
|
|
int main(int argc, char **argv)
|
|
|
|
{
|
|
|
|
ensure_users_exist();
|
|
|
|
ensure_securityfs_mounted();
|
|
|
|
write_policies();
|
|
|
|
|
|
|
|
if (prctl(PR_SET_KEEPCAPS, 1L))
|
|
|
|
die("Error with set keepcaps\n");
|
|
|
|
|
|
|
|
// First test to make sure we can write userns mappings from a user
|
|
|
|
// that doesn't have any restrictions (as long as it has CAP_SETUID);
|
|
|
|
setuid(NO_POLICY_USER);
|
|
|
|
setgid(NO_POLICY_USER);
|
|
|
|
|
|
|
|
// Take away all but setid caps
|
|
|
|
drop_caps(true);
|
|
|
|
|
|
|
|
// Need PR_SET_DUMPABLE flag set so we can write /proc/[pid]/uid_map
|
|
|
|
// from non-root parent process.
|
|
|
|
if (prctl(PR_SET_DUMPABLE, 1, 0, 0, 0))
|
|
|
|
die("Error with set dumpable\n");
|
|
|
|
|
|
|
|
if (!test_userns(true)) {
|
|
|
|
die("test_userns failed when it should work\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
setuid(RESTRICTED_PARENT);
|
|
|
|
setgid(RESTRICTED_PARENT);
|
|
|
|
|
|
|
|
test_setuid(ROOT_USER, false);
|
|
|
|
test_setuid(ALLOWED_CHILD1, true);
|
|
|
|
test_setuid(ALLOWED_CHILD2, true);
|
|
|
|
test_setuid(NO_POLICY_USER, false);
|
|
|
|
|
|
|
|
if (!test_userns(false)) {
|
|
|
|
die("test_userns worked when it should fail\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
// Now take away all caps
|
|
|
|
drop_caps(false);
|
|
|
|
test_setuid(2, false);
|
|
|
|
test_setuid(3, false);
|
|
|
|
test_setuid(4, false);
|
|
|
|
|
|
|
|
// NOTE: this test doesn't clean up users that were created in
|
|
|
|
// /etc/passwd or flush policies that were added to the LSM.
|
|
|
|
return EXIT_SUCCESS;
|
|
|
|
}
|