runsc: Support job control signals in "exec -it".

Terminal support in runsc relies on host tty file descriptors that are imported
into the sandbox. Application tty ioctls are sent directly to the host fd.

However, those host tty ioctls are associated in the host kernel with a host
process (in this case runsc), and the host kernel intercepts job control
characters like ^C and send signals to the host process. Thus, typing ^C into a
"runsc exec" shell will send a SIGINT to the runsc process.

This change makes "runsc exec" handle all signals, and forward them into the
sandbox via the "ContainerSignal" urpc method. Since the "runsc exec" is
associated with a particular container process in the sandbox, the signal must
be associated with the same container process.

One big difficulty is that the signal should not necessarily be sent to the
sandbox process started by "exec", but instead must be sent to the foreground
process group for the tty. For example, we may exec "bash", and from bash call
"sleep 100". A ^C at this point should SIGINT sleep, not bash.

To handle this, tty files inside the sandbox must keep track of their
foreground process group, which is set/get via ioctls. When an incoming
ContainerSignal urpc comes in, we look up the foreground process group via the
tty file. Unfortunately, this means we have to expose and cache the tty file in
the Loader.

Note that "runsc exec" now handles signals properly, but "runs run" does not.
That will come in a later CL, as this one is complex enough already.

Example:
	root@:/usr/local/apache2# sleep 100
	^C

	root@:/usr/local/apache2# sleep 100
	^Z
	[1]+  Stopped                 sleep 100

	root@:/usr/local/apache2# fg
	sleep 100
	^C

	root@:/usr/local/apache2#

PiperOrigin-RevId: 215334554
Change-Id: I53cdce39653027908510a5ba8d08c49f9cf24f39

1 files changed, 185 insertions, 0 deletions

diff --git a/pkg/sentry/fs/host/tty.go b/pkg/sentry/fs/host/tty.go
new file mode 100644
index 000000000..ad1323610
--- /dev/null
+++ b/pkg/sentry/fs/host/tty.go
@@ -0,0 +1,185 @@
+// Copyright 2018 Google Inc.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package host
+
+import (
+	"sync"
+
+	"gvisor.googlesource.com/gvisor/pkg/abi/linux"
+	"gvisor.googlesource.com/gvisor/pkg/sentry/arch"
+	"gvisor.googlesource.com/gvisor/pkg/sentry/context"
+	"gvisor.googlesource.com/gvisor/pkg/sentry/fs"
+	"gvisor.googlesource.com/gvisor/pkg/sentry/kernel"
+	"gvisor.googlesource.com/gvisor/pkg/sentry/usermem"
+	"gvisor.googlesource.com/gvisor/pkg/syserror"
+)
+
+// TTYFileOperations implements fs.FileOperations for a host file descriptor
+// that wraps a TTY FD.
+//
+// +stateify savable
+type TTYFileOperations struct {
+	fileOperations
+
+	// mu protects the fields below.
+	mu sync.Mutex
+
+	// FGProcessGroup is the foreground process group this TTY.  Will be
+	// nil if not set or if this file has been released.
+	fgProcessGroup *kernel.ProcessGroup
+}
+
+// newTTYFile returns a new fs.File that wraps a TTY FD.
+func newTTYFile(ctx context.Context, dirent *fs.Dirent, flags fs.FileFlags, iops *inodeOperations) *fs.File {
+	return fs.NewFile(ctx, dirent, flags, &TTYFileOperations{
+		fileOperations: fileOperations{iops: iops},
+	})
+}
+
+// ForegroundProcessGroup returns the foreground process for the TTY. This will
+// be nil if the foreground process has not been set or if the file has been
+// released.
+func (t *TTYFileOperations) ForegroundProcessGroup() *kernel.ProcessGroup {
+	t.mu.Lock()
+	defer t.mu.Unlock()
+	return t.fgProcessGroup
+}
+
+// Release implements fs.FileOperations.Release.
+func (t *TTYFileOperations) Release() {
+	t.mu.Lock()
+	t.fgProcessGroup = nil
+	t.mu.Unlock()
+
+	t.fileOperations.Release()
+}
+
+// Ioctl implements fs.FileOperations.Ioctl.
+func (t *TTYFileOperations) Ioctl(ctx context.Context, io usermem.IO, args arch.SyscallArguments) (uintptr, error) {
+	// Ignore arg[0].  This is the real FD:
+	fd := t.fileOperations.iops.fileState.FD()
+	ioctl := args[1].Uint64()
+	switch ioctl {
+	case linux.TCGETS:
+		termios, err := ioctlGetTermios(fd)
+		if err != nil {
+			return 0, err
+		}
+		_, err = usermem.CopyObjectOut(ctx, io, args[2].Pointer(), termios, usermem.IOOpts{
+			AddressSpaceActive: true,
+		})
+		return 0, err
+
+	case linux.TCSETS, linux.TCSETSW, linux.TCSETSF:
+		var termios linux.Termios
+		if _, err := usermem.CopyObjectIn(ctx, io, args[2].Pointer(), &termios, usermem.IOOpts{
+			AddressSpaceActive: true,
+		}); err != nil {
+			return 0, err
+		}
+		err := ioctlSetTermios(fd, ioctl, &termios)
+		return 0, err
+
+	case linux.TIOCGPGRP:
+		// Args: pid_t *argp
+		// When successful, equivalent to *argp = tcgetpgrp(fd).
+		// Get the process group ID of the foreground process group on
+		// this terminal.
+
+		t.mu.Lock()
+		defer t.mu.Unlock()
+
+		if t.fgProcessGroup == nil {
+			// No process group has been set yet. Let's just lie
+			// and tell it the process group from the current task.
+			// The app is probably going to set it to something
+			// else very soon anyways.
+			t.fgProcessGroup = kernel.TaskFromContext(ctx).ThreadGroup().ProcessGroup()
+		}
+
+		// Map the ProcessGroup into a ProcessGroupID in the task's PID
+		// namespace.
+		pgID := kernel.TaskFromContext(ctx).ThreadGroup().PIDNamespace().IDOfProcessGroup(t.fgProcessGroup)
+		_, err := usermem.CopyObjectOut(ctx, io, args[2].Pointer(), &pgID, usermem.IOOpts{
+			AddressSpaceActive: true,
+		})
+		return 0, err
+
+	case linux.TIOCSPGRP:
+		// Args: const pid_t *argp
+		// Equivalent to tcsetpgrp(fd, *argp).
+		// Set the foreground process group ID of this terminal.
+
+		var pgID kernel.ProcessGroupID
+		if _, err := usermem.CopyObjectIn(ctx, io, args[2].Pointer(), &pgID, usermem.IOOpts{
+			AddressSpaceActive: true,
+		}); err != nil {
+			return 0, err
+		}
+
+		// pgID must be non-negative.
+		if pgID < 0 {
+			return 0, syserror.EINVAL
+		}
+
+		// Process group with pgID must exist in this PID namespace.
+		task := kernel.TaskFromContext(ctx)
+		pidns := task.PIDNamespace()
+		pg := pidns.ProcessGroupWithID(pgID)
+		if pg == nil {
+			return 0, syserror.ESRCH
+		}
+
+		// Process group must be in same session as calling task's
+		// process group.
+		curSession := task.ThreadGroup().ProcessGroup().Session()
+		curSessionID := pidns.IDOfSession(curSession)
+		if pidns.IDOfSession(pg.Session()) != curSessionID {
+			return 0, syserror.EPERM
+		}
+
+		t.mu.Lock()
+		t.fgProcessGroup = pg
+		t.mu.Unlock()
+		return 0, nil
+
+	case linux.TIOCGWINSZ:
+		// Args: struct winsize *argp
+		// Get window size.
+		winsize, err := ioctlGetWinsize(fd)
+		if err != nil {
+			return 0, err
+		}
+		_, err = usermem.CopyObjectOut(ctx, io, args[2].Pointer(), winsize, usermem.IOOpts{
+			AddressSpaceActive: true,
+		})
+		return 0, err
+
+	case linux.TIOCSWINSZ:
+		// Args: const struct winsize *argp
+		// Set window size.
+		var winsize linux.Winsize
+		if _, err := usermem.CopyObjectIn(ctx, io, args[2].Pointer(), &winsize, usermem.IOOpts{
+			AddressSpaceActive: true,
+		}); err != nil {
+			return 0, err
+		}
+		err := ioctlSetWinsize(fd, &winsize)
+		return 0, err
+
+	default:
+		return 0, syserror.ENOTTY
+	}
+}