Commit 2bd28cee authored by Michael Stapelberg's avatar Michael Stapelberg Committed by Ian Lance Taylor

syscall: correctly set up uid/gid mappings in user namespaces

Before this CL, uid/gid mapping was always set up from the parent
process, which is a privileged operation.

When using unprivileged user namespaces, a process can modify its
uid/gid mapping after the unshare(2) call (but setting the uid/gid
mapping from another process is NOT possible).

Fixes #29789

Change-Id: I8c96a03f5da23fe80bbb83ef051ad89cf185d750
Reviewed-on: https://go-review.googlesource.com/c/go/+/158298
Run-TryBot: Tobias Klauser <tobias.klauser@gmail.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: default avatarIan Lance Taylor <iant@golang.org>
parent 30cc8a46
......@@ -82,10 +82,13 @@ func forkAndExecInChild(argv0 *byte, argv, envv []*byte, chroot, dir *byte, attr
if sys.UidMappings != nil || sys.GidMappings != nil {
Close(p[0])
err := writeUidGidMappings(pid, sys)
var err2 Errno
if err != nil {
err2 = err.(Errno)
// uid/gid mappings will be written after fork and unshare(2) for user
// namespaces.
if sys.Unshareflags&CLONE_NEWUSER == 0 {
if err := writeUidGidMappings(pid, sys); err != nil {
err2 = err.(Errno)
}
}
RawSyscall(SYS_WRITE, uintptr(p[1]), uintptr(unsafe.Pointer(&err2)), unsafe.Sizeof(err2))
Close(p[1])
......@@ -142,12 +145,32 @@ func forkAndExecInChild1(argv0 *byte, argv, envv []*byte, chroot, dir *byte, att
// Declare all variables at top in case any
// declarations require heap allocation (e.g., err1).
var (
err2 Errno
nextfd int
i int
caps caps
err2 Errno
nextfd int
i int
caps caps
fd1 uintptr
puid, psetgroups, pgid []byte
uidmap, setgroups, gidmap []byte
)
if sys.UidMappings != nil {
puid = []byte("/proc/self/uid_map\000")
uidmap = formatIDMappings(sys.UidMappings)
}
if sys.GidMappings != nil {
psetgroups = []byte("/proc/self/setgroups\000")
pgid = []byte("/proc/self/gid_map\000")
if sys.GidMappingsEnableSetgroups {
setgroups = []byte("allow\000")
} else {
setgroups = []byte("deny\000")
}
gidmap = formatIDMappings(sys.GidMappings)
}
// Record parent PID so child can test if it has died.
ppid, _ := rawSyscallNoError(SYS_GETPID, 0, 0, 0)
......@@ -264,6 +287,46 @@ func forkAndExecInChild1(argv0 *byte, argv, envv []*byte, chroot, dir *byte, att
if err1 != 0 {
goto childerror
}
if sys.Unshareflags&CLONE_NEWUSER != 0 && sys.GidMappings != nil {
dirfd := int(_AT_FDCWD)
if fd1, _, err1 = RawSyscall6(SYS_OPENAT, uintptr(dirfd), uintptr(unsafe.Pointer(&psetgroups[0])), uintptr(O_WRONLY), 0, 0, 0); err1 != 0 {
goto childerror
}
r1, _, err1 = RawSyscall(SYS_WRITE, uintptr(fd1), uintptr(unsafe.Pointer(&setgroups[0])), uintptr(len(setgroups)))
if err1 != 0 {
goto childerror
}
if _, _, err1 = RawSyscall(SYS_CLOSE, uintptr(fd1), 0, 0); err1 != 0 {
goto childerror
}
if fd1, _, err1 = RawSyscall6(SYS_OPENAT, uintptr(dirfd), uintptr(unsafe.Pointer(&pgid[0])), uintptr(O_WRONLY), 0, 0, 0); err1 != 0 {
goto childerror
}
r1, _, err1 = RawSyscall(SYS_WRITE, uintptr(fd1), uintptr(unsafe.Pointer(&gidmap[0])), uintptr(len(gidmap)))
if err1 != 0 {
goto childerror
}
if _, _, err1 = RawSyscall(SYS_CLOSE, uintptr(fd1), 0, 0); err1 != 0 {
goto childerror
}
}
if sys.Unshareflags&CLONE_NEWUSER != 0 && sys.UidMappings != nil {
dirfd := int(_AT_FDCWD)
if fd1, _, err1 = RawSyscall6(SYS_OPENAT, uintptr(dirfd), uintptr(unsafe.Pointer(&puid[0])), uintptr(O_WRONLY), 0, 0, 0); err1 != 0 {
goto childerror
}
r1, _, err1 = RawSyscall(SYS_WRITE, uintptr(fd1), uintptr(unsafe.Pointer(&uidmap[0])), uintptr(len(uidmap)))
if err1 != 0 {
goto childerror
}
if _, _, err1 = RawSyscall(SYS_CLOSE, uintptr(fd1), 0, 0); err1 != 0 {
goto childerror
}
}
// The unshare system call in Linux doesn't unshare mount points
// mounted with --shared. Systemd mounts / with --shared. For a
// long discussion of the pros and cons of this see debian bug 739593.
......@@ -480,25 +543,22 @@ func forkExecPipe(p []int) (err error) {
return
}
// writeIDMappings writes the user namespace User ID or Group ID mappings to the specified path.
func writeIDMappings(path string, idMap []SysProcIDMap) error {
fd, err := Open(path, O_RDWR, 0)
if err != nil {
return err
}
data := ""
func formatIDMappings(idMap []SysProcIDMap) []byte {
var data []byte
for _, im := range idMap {
data = data + itoa(im.ContainerID) + " " + itoa(im.HostID) + " " + itoa(im.Size) + "\n"
data = append(data, []byte(itoa(im.ContainerID)+" "+itoa(im.HostID)+" "+itoa(im.Size)+"\n")...)
}
return data
}
bytes, err := ByteSliceFromString(data)
// writeIDMappings writes the user namespace User ID or Group ID mappings to the specified path.
func writeIDMappings(path string, idMap []SysProcIDMap) error {
fd, err := Open(path, O_RDWR, 0)
if err != nil {
Close(fd)
return err
}
if _, err := Write(fd, bytes); err != nil {
if _, err := Write(fd, formatIDMappings(idMap)); err != nil {
Close(fd)
return err
}
......
......@@ -434,6 +434,49 @@ func TestUnshareMountNameSpaceChroot(t *testing.T) {
}
}
func TestUnshareUidGidMappingHelper(*testing.T) {
if os.Getenv("GO_WANT_HELPER_PROCESS") != "1" {
return
}
defer os.Exit(0)
if err := syscall.Chroot(os.TempDir()); err != nil {
fmt.Fprintln(os.Stderr, err)
os.Exit(2)
}
}
// Test for Issue 29789: unshare fails when uid/gid mapping is specified
func TestUnshareUidGidMapping(t *testing.T) {
if os.Getuid() == 0 {
t.Skip("test exercises unprivileged user namespace, fails with privileges")
}
checkUserNS(t)
cmd := exec.Command(os.Args[0], "-test.run=TestUnshareUidGidMappingHelper")
cmd.Env = append(os.Environ(), "GO_WANT_HELPER_PROCESS=1")
cmd.SysProcAttr = &syscall.SysProcAttr{
Unshareflags: syscall.CLONE_NEWNS | syscall.CLONE_NEWUSER,
GidMappingsEnableSetgroups: false,
UidMappings: []syscall.SysProcIDMap{
{
ContainerID: 0,
HostID: syscall.Getuid(),
Size: 1,
},
},
GidMappings: []syscall.SysProcIDMap{
{
ContainerID: 0,
HostID: syscall.Getgid(),
Size: 1,
},
},
}
out, err := cmd.CombinedOutput()
if err != nil {
t.Fatalf("Cmd failed with err %v, output: %s", err, out)
}
}
type capHeader struct {
version uint32
pid int32
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment