On Fri, Feb 08, 2019 at 04:06:09PM -0700, Shuah Khan wrote: > tty_set_termios() has the following WARN_ON which can be triggered with a > syscall to invoke TIOCSETD __NR_ioctl. > > WARN_ON(tty->driver->type == TTY_DRIVER_TYPE_PTY && > tty->driver->subtype == PTY_TYPE_MASTER); > Reference: https://syzkaller.appspot.com/bug?id=2410d22f1d8e5984217329dd0884b01d99e3e48d > > "The problem started with commit 7721383f4199 ("Bluetooth: hci_uart: Support > operational speed during setup") which introduced a new way for how > tty_set_termios() could end up being called for a master pty." > > Fix hci_uart_tty_open() to check if tty supports set_termios in addition > to write and error out, if it doesn't. > > Reported-by: syzbot+a950165cbb86bdd023a4@xxxxxxxxxxxxxxxxxxxxxxxxx > Cc: Johan Hovold <johan@xxxxxxxxxx> > Cc: Marcel Holtmann <marcel@xxxxxxxxxxxx> > Cc: Al Viro <viro@xxxxxxxxxxxxxxxxxx> > Signed-off-by: Shuah Khan <shuah@xxxxxxxxxx> > --- > drivers/bluetooth/hci_ldisc.c | 6 +++--- > 1 file changed, 3 insertions(+), 3 deletions(-) > > diff --git a/drivers/bluetooth/hci_ldisc.c b/drivers/bluetooth/hci_ldisc.c > index fbf7b4df23ab..087ec2268744 100644 > --- a/drivers/bluetooth/hci_ldisc.c > +++ b/drivers/bluetooth/hci_ldisc.c > @@ -474,10 +474,10 @@ static int hci_uart_tty_open(struct tty_struct *tty) > > BT_DBG("tty %p", tty); > > - /* Error if the tty has no write op instead of leaving an exploitable > - * hole > + /* Error if the tty has no write/set_termios ops instead of leaving > + * an exploitable hole. Why do you think that a tty driver not implementing set_termios() is exploitable? Note that tty_set_termios() handles a missing set_termios() callback just fine without dereferencing a NULL-pointer. > */ > - if (tty->ops->write == NULL) > + if (!tty->ops->write || !tty->ops->set_termios) > return -EOPNOTSUPP; I know Marcel asked you do implement things like this, but what you're really doing is to try to uphold the invariant that tty_set_termios() should never be called for a pty master. The WARN_ON in that function was put in place to make sure line-disciplines use the correct ioctl helpers (and the commit I pointed you to earlier in the paragraph you quote in the commit message broke that invariant). I'm afraid this is totally obscured by the above change and misleading comment update. Johan
