Implement F_GETLK fcntl.

Fixes #5113.

PiperOrigin-RevId: 353313374

3 files changed, 62 insertions, 16 deletions

diff --git a/pkg/sentry/syscalls/linux/sys_file.go b/pkg/sentry/syscalls/linux/sys_file.go
index c33571f43..a6253626e 100644
--- a/pkg/sentry/syscalls/linux/sys_file.go
+++ b/pkg/sentry/syscalls/linux/sys_file.go
@@ -1014,12 +1014,12 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 			}
 			if cmd == linux.F_SETLK {
 				// Non-blocking lock, provide a nil lock.Blocker.
-				if !file.Dirent.Inode.LockCtx.Posix.LockRegion(t.FDTable(), lock.ReadLock, rng, nil) {
+				if !file.Dirent.Inode.LockCtx.Posix.LockRegionVFS1(t.FDTable(), lock.ReadLock, rng, nil) {
 					return 0, nil, syserror.EAGAIN
 				}
 			} else {
 				// Blocking lock, pass in the task to satisfy the lock.Blocker interface.
-				if !file.Dirent.Inode.LockCtx.Posix.LockRegion(t.FDTable(), lock.ReadLock, rng, t) {
+				if !file.Dirent.Inode.LockCtx.Posix.LockRegionVFS1(t.FDTable(), lock.ReadLock, rng, t) {
 					return 0, nil, syserror.EINTR
 				}
 			}
@@ -1030,12 +1030,12 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 			}
 			if cmd == linux.F_SETLK {
 				// Non-blocking lock, provide a nil lock.Blocker.
-				if !file.Dirent.Inode.LockCtx.Posix.LockRegion(t.FDTable(), lock.WriteLock, rng, nil) {
+				if !file.Dirent.Inode.LockCtx.Posix.LockRegionVFS1(t.FDTable(), lock.WriteLock, rng, nil) {
 					return 0, nil, syserror.EAGAIN
 				}
 			} else {
 				// Blocking lock, pass in the task to satisfy the lock.Blocker interface.
-				if !file.Dirent.Inode.LockCtx.Posix.LockRegion(t.FDTable(), lock.WriteLock, rng, t) {
+				if !file.Dirent.Inode.LockCtx.Posix.LockRegionVFS1(t.FDTable(), lock.WriteLock, rng, t) {
 					return 0, nil, syserror.EINTR
 				}
 			}
@@ -2167,24 +2167,24 @@ func Flock(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 	case linux.LOCK_EX:
 		if nonblocking {
 			// Since we're nonblocking we pass a nil lock.Blocker implementation.
-			if !file.Dirent.Inode.LockCtx.BSD.LockRegion(file, lock.WriteLock, rng, nil) {
+			if !file.Dirent.Inode.LockCtx.BSD.LockRegionVFS1(file, lock.WriteLock, rng, nil) {
 				return 0, nil, syserror.EWOULDBLOCK
 			}
 		} else {
 			// Because we're blocking we will pass the task to satisfy the lock.Blocker interface.
-			if !file.Dirent.Inode.LockCtx.BSD.LockRegion(file, lock.WriteLock, rng, t) {
+			if !file.Dirent.Inode.LockCtx.BSD.LockRegionVFS1(file, lock.WriteLock, rng, t) {
 				return 0, nil, syserror.EINTR
 			}
 		}
 	case linux.LOCK_SH:
 		if nonblocking {
 			// Since we're nonblocking we pass a nil lock.Blocker implementation.
-			if !file.Dirent.Inode.LockCtx.BSD.LockRegion(file, lock.ReadLock, rng, nil) {
+			if !file.Dirent.Inode.LockCtx.BSD.LockRegionVFS1(file, lock.ReadLock, rng, nil) {
 				return 0, nil, syserror.EWOULDBLOCK
 			}
 		} else {
 			// Because we're blocking we will pass the task to satisfy the lock.Blocker interface.
-			if !file.Dirent.Inode.LockCtx.BSD.LockRegion(file, lock.ReadLock, rng, t) {
+			if !file.Dirent.Inode.LockCtx.BSD.LockRegionVFS1(file, lock.ReadLock, rng, t) {
 				return 0, nil, syserror.EINTR
 			}
 		}
diff --git a/pkg/sentry/syscalls/linux/vfs2/fd.go b/pkg/sentry/syscalls/linux/vfs2/fd.go
index 9281952cb..e39f074f2 100644
--- a/pkg/sentry/syscalls/linux/vfs2/fd.go
+++ b/pkg/sentry/syscalls/linux/vfs2/fd.go
@@ -205,8 +205,12 @@ func Fcntl(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 		}
 		err := tmpfs.AddSeals(file, args[2].Uint())
 		return 0, nil, err
-	case linux.F_SETLK, linux.F_SETLKW:
-		return 0, nil, posixLock(t, args, file, cmd)
+	case linux.F_SETLK:
+		return 0, nil, posixLock(t, args, file, false /* blocking */)
+	case linux.F_SETLKW:
+		return 0, nil, posixLock(t, args, file, true /* blocking */)
+	case linux.F_GETLK:
+		return 0, nil, posixTestLock(t, args, file)
 	case linux.F_GETSIG:
 		a := file.AsyncHandler()
 		if a == nil {
@@ -292,7 +296,49 @@ func setAsyncOwner(t *kernel.Task, fd int, file *vfs.FileDescription, ownerType,
 	}
 }
 
-func posixLock(t *kernel.Task, args arch.SyscallArguments, file *vfs.FileDescription, cmd int32) error {
+func posixTestLock(t *kernel.Task, args arch.SyscallArguments, file *vfs.FileDescription) error {
+	// Copy in the lock request.
+	flockAddr := args[2].Pointer()
+	var flock linux.Flock
+	if _, err := flock.CopyIn(t, flockAddr); err != nil {
+		return err
+	}
+	var typ lock.LockType
+	switch flock.Type {
+	case linux.F_RDLCK:
+		typ = lock.ReadLock
+	case linux.F_WRLCK:
+		typ = lock.WriteLock
+	default:
+		return syserror.EINVAL
+	}
+	r, err := file.ComputeLockRange(t, uint64(flock.Start), uint64(flock.Len), flock.Whence)
+	if err != nil {
+		return err
+	}
+
+	newFlock, err := file.TestPOSIX(t, t.FDTable(), typ, r)
+	if err != nil {
+		return err
+	}
+	newFlock.PID = translatePID(t.PIDNamespace().Root(), t.PIDNamespace(), newFlock.PID)
+	if _, err = newFlock.CopyOut(t, flockAddr); err != nil {
+		return err
+	}
+	return nil
+}
+
+// translatePID translates a pid from one namespace to another. Note that this
+// may race with task termination/creation, in which case the original task
+// corresponding to pid may no longer exist. This is used to implement the
+// F_GETLK fcntl, which has the same potential race in Linux as well (i.e.,
+// there is no synchronization between retrieving the lock PID and translating
+// it). See fs/locks.c:posix_lock_to_flock.
+func translatePID(old, new *kernel.PIDNamespace, pid int32) int32 {
+	return int32(new.IDOfTask(old.TaskWithID(kernel.ThreadID(pid))))
+}
+
+func posixLock(t *kernel.Task, args arch.SyscallArguments, file *vfs.FileDescription, blocking bool) error {
 	// Copy in the lock request.
 	flockAddr := args[2].Pointer()
 	var flock linux.Flock
@@ -301,7 +347,7 @@ func posixLock(t *kernel.Task, args arch.SyscallArguments, file *vfs.FileDescrip
 	}
 
 	var blocker lock.Blocker
-	if cmd == linux.F_SETLKW {
+	if blocking {
 		blocker = t
 	}
 
@@ -315,13 +361,13 @@ func posixLock(t *kernel.Task, args arch.SyscallArguments, file *vfs.FileDescrip
 		if !file.IsReadable() {
 			return syserror.EBADF
 		}
-		return file.LockPOSIX(t, t.FDTable(), lock.ReadLock, r, blocker)
+		return file.LockPOSIX(t, t.FDTable(), int32(t.TGIDInRoot()), lock.ReadLock, r, blocker)
 
 	case linux.F_WRLCK:
 		if !file.IsWritable() {
 			return syserror.EBADF
 		}
-		return file.LockPOSIX(t, t.FDTable(), lock.WriteLock, r, blocker)
+		return file.LockPOSIX(t, t.FDTable(), int32(t.TGIDInRoot()), lock.WriteLock, r, blocker)
 
 	case linux.F_UNLCK:
 		return file.UnlockPOSIX(t, t.FDTable(), r)
diff --git a/pkg/sentry/syscalls/linux/vfs2/lock.go b/pkg/sentry/syscalls/linux/vfs2/lock.go
index b910b5a74..d1452a04d 100644
--- a/pkg/sentry/syscalls/linux/vfs2/lock.go
+++ b/pkg/sentry/syscalls/linux/vfs2/lock.go
@@ -44,11 +44,11 @@ func Flock(t *kernel.Task, args arch.SyscallArguments) (uintptr, *kernel.Syscall
 
 	switch operation {
 	case linux.LOCK_EX:
-		if err := file.LockBSD(t, lock.WriteLock, blocker); err != nil {
+		if err := file.LockBSD(t, int32(t.TGIDInRoot()), lock.WriteLock, blocker); err != nil {
 			return 0, nil, err
 		}
 	case linux.LOCK_SH:
-		if err := file.LockBSD(t, lock.ReadLock, blocker); err != nil {
+		if err := file.LockBSD(t, int32(t.TGIDInRoot()), lock.ReadLock, blocker); err != nil {
 			return 0, nil, err
 		}
 	case linux.LOCK_UN: