src/syscall/exec_plan9.go

// Copyright 2009 The Go Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.

// Fork, exec, wait, etc.

package syscall

import (
	"runtime"
	"sync"
	"unsafe"
)

// Lock synchronizing creation of new file descriptors with fork.
//
// We want the child in a fork/exec sequence to inherit only the
// file descriptors we intend.  To do that, we mark all file
// descriptors close-on-exec and then, in the child, explicitly
// unmark the ones we want the exec'ed program to keep.
// Unix doesn't make this easy: there is, in general, no way to
// allocate a new file descriptor close-on-exec.  Instead you
// have to allocate the descriptor and then mark it close-on-exec.
// If a fork happens between those two events, the child's exec
// will inherit an unwanted file descriptor.
//
// This lock solves that race: the create new fd/mark close-on-exec
// operation is done holding ForkLock for reading, and the fork itself
// is done holding ForkLock for writing.  At least, that's the idea.
// There are some complications.
//
// Some system calls that create new file descriptors can block
// for arbitrarily long times: open on a hung NFS server or named
// pipe, accept on a socket, and so on.  We can't reasonably grab
// the lock across those operations.
//
// It is worse to inherit some file descriptors than others.
// If a non-malicious child accidentally inherits an open ordinary file,
// that's not a big deal.  On the other hand, if a long-lived child
// accidentally inherits the write end of a pipe, then the reader
// of that pipe will not see EOF until that child exits, potentially
// causing the parent program to hang.  This is a common problem
// in threaded C programs that use popen.
//
// Luckily, the file descriptors that are most important not to
// inherit are not the ones that can take an arbitrarily long time
// to create: pipe returns instantly, and the net package uses
// non-blocking I/O to accept on a listening socket.
// The rules for which file descriptor-creating operations use the
// ForkLock are as follows:
//
// 1) Pipe.    Does not block.  Use the ForkLock.
// 2) Socket.  Does not block.  Use the ForkLock.
// 3) Accept.  If using non-blocking mode, use the ForkLock.
//             Otherwise, live with the race.
// 4) Open.    Can block.  Use O_CLOEXEC if available (Linux).
//             Otherwise, live with the race.
// 5) Dup.     Does not block.  Use the ForkLock.
//             On Linux, could use fcntl F_DUPFD_CLOEXEC
//             instead of the ForkLock, but only for dup(fd, -1).

var ForkLock sync.RWMutex

// StringSlicePtr converts a slice of strings to a slice of pointers
// to NUL-terminated byte arrays. If any string contains a NUL byte
// this function panics instead of returning an error.
//
// Deprecated: Use SlicePtrFromStrings instead.
func StringSlicePtr(ss []string) []*byte {
	bb := make([]*byte, len(ss)+1)
	for i := 0; i < len(ss); i++ {
		bb[i] = StringBytePtr(ss[i])
	}
	bb[len(ss)] = nil
	return bb
}

// SlicePtrFromStrings converts a slice of strings to a slice of
// pointers to NUL-terminated byte arrays. If any string contains
// a NUL byte, it returns (nil, EINVAL).
func SlicePtrFromStrings(ss []string) ([]*byte, error) {
	var err error
	bb := make([]*byte, len(ss)+1)
	for i := 0; i < len(ss); i++ {
		bb[i], err = BytePtrFromString(ss[i])
		if err != nil {
			return nil, err
		}
	}
	bb[len(ss)] = nil
	return bb, nil
}

// readdirnames returns the names of files inside the directory represented by dirfd.
func readdirnames(dirfd int) (names []string, err error) {
	names = make([]string, 0, 100)
	var buf [STATMAX]byte

	for {
		n, e := Read(dirfd, buf[:])
		if e != nil {
			return nil, e
		}
		if n == 0 {
			break
		}
		for i := 0; i < n; {
			m, _ := gbit16(buf[i:])
			m += 2

			if m < STATFIXLEN {
				return nil, ErrBadStat
			}

			s, _, ok := gstring(buf[i+41:])
			if !ok {
				return nil, ErrBadStat
			}
			names = append(names, s)
			i += int(m)
		}
	}
	return
}

// readdupdevice returns a list of currently opened fds (excluding stdin, stdout, stderr) from the dup device #d.
// ForkLock should be write locked before calling, so that no new fds would be created while the fd list is being read.
func readdupdevice() (fds []int, err error) {
	dupdevfd, err := Open("#d", O_RDONLY)
	if err != nil {
		return
	}
	defer Close(dupdevfd)

	names, err := readdirnames(dupdevfd)
	if err != nil {
		return
	}

	fds = make([]int, 0, len(names)/2)
	for _, name := range names {
		if n := len(name); n > 3 && name[n-3:n] == "ctl" {
			continue
		}
		fd := int(atoi([]byte(name)))
		switch fd {
		case 0, 1, 2, dupdevfd:
			continue
		}
		fds = append(fds, fd)
	}
	return
}

var startupFds []int

// Plan 9 does not allow clearing the OCEXEC flag
// from the underlying channel backing an open file descriptor,
// therefore we store a list of already opened file descriptors
// inside startupFds and skip them when manually closing descriptors
// not meant to be passed to a child exec.
func init() {
	startupFds, _ = readdupdevice()
}

// forkAndExecInChild forks the process, calling dup onto 0..len(fd)
// and finally invoking exec(argv0, argvv, envv) in the child.
// If a dup or exec fails, it writes the error string to pipe.
// (The pipe write end is close-on-exec so if exec succeeds, it will be closed.)
//
// In the child, this function must not acquire any locks, because
// they might have been locked at the time of the fork.  This means
// no rescheduling, no malloc calls, and no new stack segments.
// The calls to RawSyscall are okay because they are assembly
// functions that do not grow the stack.
func forkAndExecInChild(argv0 *byte, argv []*byte, envv []envItem, dir *byte, attr *ProcAttr, fdsToClose []int, pipe int, rflag int) (pid int, err error) {
	// Declare all variables at top in case any
	// declarations require heap allocation (e.g., errbuf).
	var (
		r1       uintptr
		nextfd   int
		i        int
		clearenv int
		envfd    int
		errbuf   [ERRMAX]byte
	)

	// Guard against side effects of shuffling fds below.
	// Make sure that nextfd is beyond any currently open files so
	// that we can't run the risk of overwriting any of them.
	fd := make([]int, len(attr.Files))
	nextfd = len(attr.Files)
	for i, ufd := range attr.Files {
		if nextfd < int(ufd) {
			nextfd = int(ufd)
		}
		fd[i] = int(ufd)
	}
	nextfd++

	if envv != nil {
		clearenv = RFCENVG
	}

	// About to call fork.
	// No more allocation or calls of non-assembly functions.
	r1, _, _ = RawSyscall(SYS_RFORK, uintptr(RFPROC|RFFDG|RFREND|clearenv|rflag), 0, 0)

	if r1 != 0 {
		if int32(r1) == -1 {
			return 0, NewError(errstr())
		}
		// parent; return PID
		return int(r1), nil
	}

	// Fork succeeded, now in child.

	// Close fds we don't need.
	for i = 0; i < len(fdsToClose); i++ {
		r1, _, _ = RawSyscall(SYS_CLOSE, uintptr(fdsToClose[i]), 0, 0)
		if int32(r1) == -1 {
			goto childerror
		}
	}

	if envv != nil {
		// Write new environment variables.
		for i = 0; i < len(envv); i++ {
			r1, _, _ = RawSyscall(SYS_CREATE, uintptr(unsafe.Pointer(envv[i].name)), uintptr(O_WRONLY), uintptr(0666))

			if int32(r1) == -1 {
				goto childerror
			}

			envfd = int(r1)

			r1, _, _ = RawSyscall6(SYS_PWRITE, uintptr(envfd), uintptr(unsafe.Pointer(envv[i].value)), uintptr(envv[i].nvalue),
				^uintptr(0), ^uintptr(0), 0)

			if int32(r1) == -1 || int(r1) != envv[i].nvalue {
				goto childerror
			}

			r1, _, _ = RawSyscall(SYS_CLOSE, uintptr(envfd), 0, 0)

			if int32(r1) == -1 {
				goto childerror
			}
		}
	}

	// Chdir
	if dir != nil {
		r1, _, _ = RawSyscall(SYS_CHDIR, uintptr(unsafe.Pointer(dir)), 0, 0)
		if int32(r1) == -1 {
			goto childerror
		}
	}

	// Pass 1: look for fd[i] < i and move those up above len(fd)
	// so that pass 2 won't stomp on an fd it needs later.
	if pipe < nextfd {
		r1, _, _ = RawSyscall(SYS_DUP, uintptr(pipe), uintptr(nextfd), 0)
		if int32(r1) == -1 {
			goto childerror
		}
		pipe = nextfd
		nextfd++
	}
	for i = 0; i < len(fd); i++ {
		if fd[i] >= 0 && fd[i] < int(i) {
			r1, _, _ = RawSyscall(SYS_DUP, uintptr(fd[i]), uintptr(nextfd), 0)
			if int32(r1) == -1 {
				goto childerror
			}

			fd[i] = nextfd
			nextfd++
			if nextfd == pipe { // don't stomp on pipe
				nextfd++
			}
		}
	}

	// Pass 2: dup fd[i] down onto i.
	for i = 0; i < len(fd); i++ {
		if fd[i] == -1 {
			RawSyscall(SYS_CLOSE, uintptr(i), 0, 0)
			continue
		}
		if fd[i] == int(i) {
			continue
		}
		r1, _, _ = RawSyscall(SYS_DUP, uintptr(fd[i]), uintptr(i), 0)
		if int32(r1) == -1 {
			goto childerror
		}
	}

	// Pass 3: close fd[i] if it was moved in the previous pass.
	for i = 0; i < len(fd); i++ {
		if fd[i] >= 0 && fd[i] != int(i) {
			RawSyscall(SYS_CLOSE, uintptr(fd[i]), 0, 0)
		}
	}

	// Time to exec.
	r1, _, _ = RawSyscall(SYS_EXEC,
		uintptr(unsafe.Pointer(argv0)),
		uintptr(unsafe.Pointer(&argv[0])), 0)

childerror:
	// send error string on pipe
	RawSyscall(SYS_ERRSTR, uintptr(unsafe.Pointer(&errbuf[0])), uintptr(len(errbuf)), 0)
	errbuf[len(errbuf)-1] = 0
	i = 0
	for i < len(errbuf) && errbuf[i] != 0 {
		i++
	}

	RawSyscall6(SYS_PWRITE, uintptr(pipe), uintptr(unsafe.Pointer(&errbuf[0])), uintptr(i),
		^uintptr(0), ^uintptr(0), 0)

	for {
		RawSyscall(SYS_EXITS, 0, 0, 0)
	}

	// Calling panic is not actually safe,
	// but the for loop above won't break
	// and this shuts up the compiler.
	panic("unreached")
}

func cexecPipe(p []int) error {
	e := Pipe(p)
	if e != nil {
		return e
	}

	fd, e := Open("#d/"+itoa(p[1]), O_CLOEXEC)
	if e != nil {
		Close(p[0])
		Close(p[1])
		return e
	}

	Close(fd)
	return nil
}

type envItem struct {
	name   *byte
	value  *byte
	nvalue int
}

type ProcAttr struct {
	Dir   string    // Current working directory.
	Env   []string  // Environment.
	Files []uintptr // File descriptors.
	Sys   *SysProcAttr
}

type SysProcAttr struct {
	Rfork int // additional flags to pass to rfork
}

var zeroProcAttr ProcAttr
var zeroSysProcAttr SysProcAttr

func forkExec(argv0 string, argv []string, attr *ProcAttr) (pid int, err error) {
	var (
		p      [2]int
		n      int
		errbuf [ERRMAX]byte
		wmsg   Waitmsg
	)

	if attr == nil {
		attr = &zeroProcAttr
	}
	sys := attr.Sys
	if sys == nil {
		sys = &zeroSysProcAttr
	}

	p[0] = -1
	p[1] = -1

	// Convert args to C form.
	argv0p, err := BytePtrFromString(argv0)
	if err != nil {
		return 0, err
	}
	argvp, err := SlicePtrFromStrings(argv)
	if err != nil {
		return 0, err
	}

	destDir := attr.Dir
	if destDir == "" {
		wdmu.Lock()
		destDir = wdStr
		wdmu.Unlock()
	}
	var dir *byte
	if destDir != "" {
		dir, err = BytePtrFromString(destDir)
		if err != nil {
			return 0, err
		}
	}
	var envvParsed []envItem
	if attr.Env != nil {
		envvParsed = make([]envItem, 0, len(attr.Env))
		for _, v := range attr.Env {
			i := 0
			for i < len(v) && v[i] != '=' {
				i++
			}

			envname, err := BytePtrFromString("/env/" + v[:i])
			if err != nil {
				return 0, err
			}
			envvalue := make([]byte, len(v)-i)
			copy(envvalue, v[i+1:])
			envvParsed = append(envvParsed, envItem{envname, &envvalue[0], len(v) - i})
		}
	}

	// Acquire the fork lock to prevent other threads from creating new fds before we fork.
	ForkLock.Lock()

	// get a list of open fds, excluding stdin,stdout and stderr that need to be closed in the child.
	// no new fds can be created while we hold the ForkLock for writing.
	openFds, e := readdupdevice()
	if e != nil {
		ForkLock.Unlock()
		return 0, e
	}

	fdsToClose := make([]int, 0, len(openFds))
	for _, fd := range openFds {
		doClose := true

		// exclude files opened at startup.
		for _, sfd := range startupFds {
			if fd == sfd {
				doClose = false
				break
			}
		}

		// exclude files explicitly requested by the caller.
		for _, rfd := range attr.Files {
			if fd == int(rfd) {
				doClose = false
				break
			}
		}

		if doClose {
			fdsToClose = append(fdsToClose, fd)
		}
	}

	// Allocate child status pipe close on exec.
	e = cexecPipe(p[:])

	if e != nil {
		return 0, e
	}
	fdsToClose = append(fdsToClose, p[0])

	// Kick off child.
	pid, err = forkAndExecInChild(argv0p, argvp, envvParsed, dir, attr, fdsToClose, p[1], sys.Rfork)

	if err != nil {
		if p[0] >= 0 {
			Close(p[0])
			Close(p[1])
		}
		ForkLock.Unlock()
		return 0, err
	}
	ForkLock.Unlock()

	// Read child error status from pipe.
	Close(p[1])
	n, err = Read(p[0], errbuf[:])
	Close(p[0])

	if err != nil || n != 0 {
		if n != 0 {
			err = NewError(string(errbuf[:n]))
		}

		// Child failed; wait for it to exit, to make sure
		// the zombies don't accumulate.
		for wmsg.Pid != pid {
			Await(&wmsg)
		}
		return 0, err
	}

	// Read got EOF, so pipe closed on exec, so exec succeeded.
	return pid, nil
}

type waitErr struct {
	Waitmsg
	err error
}

var procs struct {
	sync.Mutex
	waits map[int]chan *waitErr
}

// startProcess starts a new goroutine, tied to the OS
// thread, which runs the process and subsequently waits
// for it to finish, communicating the process stats back
// to any goroutines that may have been waiting on it.
//
// Such a dedicated goroutine is needed because on
// Plan 9, only the parent thread can wait for a child,
// whereas goroutines tend to jump OS threads (e.g.,
// between starting a process and running Wait(), the
// goroutine may have been rescheduled).
func startProcess(argv0 string, argv []string, attr *ProcAttr) (pid int, err error) {
	type forkRet struct {
		pid int
		err error
	}

	forkc := make(chan forkRet, 1)
	go func() {
		runtime.LockOSThread()
		var ret forkRet

		ret.pid, ret.err = forkExec(argv0, argv, attr)
		// If fork fails there is nothing to wait for.
		if ret.err != nil || ret.pid == 0 {
			forkc <- ret
			return
		}

		waitc := make(chan *waitErr, 1)

		// Mark that the process is running.
		procs.Lock()
		if procs.waits == nil {
			procs.waits = make(map[int]chan *waitErr)
		}
		procs.waits[ret.pid] = waitc
		procs.Unlock()

		forkc <- ret

		var w waitErr
		for w.err == nil && w.Pid != ret.pid {
			w.err = Await(&w.Waitmsg)
		}
		waitc <- &w
		close(waitc)
	}()
	ret := <-forkc
	return ret.pid, ret.err
}

// Combination of fork and exec, careful to be thread safe.
func ForkExec(argv0 string, argv []string, attr *ProcAttr) (pid int, err error) {
	return startProcess(argv0, argv, attr)
}

// StartProcess wraps ForkExec for package os.
func StartProcess(argv0 string, argv []string, attr *ProcAttr) (pid int, handle uintptr, err error) {
	pid, err = startProcess(argv0, argv, attr)
	return pid, 0, err
}

// Ordinary exec.
func Exec(argv0 string, argv []string, envv []string) (err error) {
	if envv != nil {
		r1, _, _ := RawSyscall(SYS_RFORK, RFCENVG, 0, 0)
		if int32(r1) == -1 {
			return NewError(errstr())
		}

		for _, v := range envv {
			i := 0
			for i < len(v) && v[i] != '=' {
				i++
			}

			fd, e := Create("/env/"+v[:i], O_WRONLY, 0666)
			if e != nil {
				return e
			}

			_, e = Write(fd, []byte(v[i+1:]))
			if e != nil {
				Close(fd)
				return e
			}
			Close(fd)
		}
	}

	argv0p, err := BytePtrFromString(argv0)
	if err != nil {
		return err
	}
	argvp, err := SlicePtrFromStrings(argv)
	if err != nil {
		return err
	}
	_, _, e1 := Syscall(SYS_EXEC,
		uintptr(unsafe.Pointer(argv0p)),
		uintptr(unsafe.Pointer(&argvp[0])),
		0)

	return e1
}

// WaitProcess waits until the pid of a
// running process is found in the queue of
// wait messages. It is used in conjunction
// with ForkExec/StartProcess to wait for a
// running process to exit.
func WaitProcess(pid int, w *Waitmsg) (err error) {
	procs.Lock()
	ch := procs.waits[pid]
	procs.Unlock()

	var wmsg *waitErr
	if ch != nil {
		wmsg = <-ch
		procs.Lock()
		if procs.waits[pid] == ch {
			delete(procs.waits, pid)
		}
		procs.Unlock()
	}
	if wmsg == nil {
		// ch was missing or ch is closed
		return NewError("process not found")
	}
	if wmsg.err != nil {
		return wmsg.err
	}
	if w != nil {
		*w = wmsg.Waitmsg
	}
	return nil
}