Re: [PATCH]TTY: Fix tty can't be restarted by TCXONC ioctl request

[Peter Hurley <peter@hurleysoftware.com>]

On 05/07/2013 02:47 PM, Wang YanQing wrote:
> I meet emacs hang in start if I do the operation below:
>    1: echo 3 > /proc/sys/vm/drop_caches
>    2: emacs BigFile
>    3: Press CTRL-S follow 2 immediately
>
> Then emacs hang on, CTRL-Q can't resume, the terminal
> hang on, you can do nothing with this terminal except
> close it.
>
> The reason is before emacs takeover control the tty,
> we use CTRL-S to XOFF it. Then when emacs takeover the
> control, it may don't use the flow-control, so emacs hang.
> But after search the emacs's startup codes, I find they use TCXONC
> to workaround this situation:
>
>    /* This code added to insure that, if flow-control is not to be used,
>       we have an unlocked terminal at the start. */
>    if (!tty_out->flow_control) ioctl (fileno (tty_out->input), TCXONC, 1);
>    ioctl (fileno (tty_out->input), TCXONC, 1);
>
> But this workaround never work due the kernel's current code.
> This patch fix it.
>
> Below is the ChangeLog introduce the tty->flow_stopped flag:
>
> Thu Nov 21 10:05:22 1996  Theodre Ts'o  <tytso@localhost.mit.edu>
>
>         * tty_ioctl.c (tty_wait_until_sent): Always check the driver
>                 wait_until_ready routine, even if there are no characters
>                 in the xmit buffer.  (There may be charactes in the device
>                 FIFO.)
>                 (n_tty_ioctl): Add new flag tty->flow_stopped which
>                 indicates whether the tty is stopped due to a request by
>                 the TCXONC ioctl (used by tcflow).  If so, don't let an
>                 incoming XOFF character restart the tty.  The tty can only
>                 be restarted by another TCXONC request.
>
> So I think this patch make TCXONC can restart tty which stopped by
> STOP_CHAR don't break the original meaning.
>
> This patch will fix some strange tty relation hang problem,
> I believe I meet it with vim in ssh, and also see below bug report:
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=465823
>
> Signed-off-by: Wang YanQing <udknight@gmail.com>
> ---
>   drivers/tty/tty_ioctl.c | 5 +++--
>   1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/tty/tty_ioctl.c b/drivers/tty/tty_ioctl.c
> index e4455e0..42e08e5 100644
> --- a/drivers/tty/tty_ioctl.c
> +++ b/drivers/tty/tty_ioctl.c
> @@ -1129,11 +1129,12 @@ int n_tty_ioctl_helper(struct tty_struct *tty, struct file *file,
>   		case TCOOFF:
>   			if (!tty->flow_stopped) {
>   				tty->flow_stopped = 1;
> -				stop_tty(tty);
> +				if (!tty->stopped)
> +					stop_tty(tty);
>   			}
>   			break;
>   		case TCOON:
> -			if (tty->flow_stopped) {
> +			if (tty->flow_stopped || tty->stopped) {
>   				tty->flow_stopped = 0;
>   				start_tty(tty);
>   			}

This should be fixed in n_tty_set_termios() instead of fixing userspace
workarounds.

The problem occurs when the tty has been stopped with STOP_CHAR(tty) and then
termios is changed so that START_CHAR(tty) is subsequently ignored
(in the reported case, I_IXON(tty) is cleared).

Regards,
Peter Hurley