On Tue, Jul 12, 2022 at 01:53:03PM +0200, Marek Behún wrote:
> From: Pali Rohár <pali@xxxxxxxxxx>
>
> If setting new baud rate fails, all other drivers leave the device at
> previous baud rate, only ftdi_sio resets to 9600 Baud.
> Change this behavior to that of other drivers so that /dev/ttyUSB*
> devices behave in the same way.
These statements are not true. Several USB serial driver set 9600 baud
on errors for historical reasons. Yet others clamp. It's inconsistent at
best.
> Signed-off-by: Pali Rohár <pali@xxxxxxxxxx>
> Tested-by: Marek Behún <kabel@xxxxxxxxxx>
> Signed-off-by: Marek Behún <kabel@xxxxxxxxxx>
> ---
> Greg pointed out that this make break some people's workflow, that they
> may depend on this behavior.
>
> This is of course possible, although IMO improbable: it would be weird
> to depend on the fall back to 9600 Baud on error, instead of expecting
> that the baud rate didn't change at all (like in other /dev/ttyUSB*
> drivers).
>
> Nonetheless if someone complains that they workflow got broken, we will
> need to revert this.
> static u32 get_ftdi_divisor(struct tty_struct *tty,
> - struct usb_serial_port *port)
> + struct usb_serial_port *port, speed_t old_baud)
> {
> struct ftdi_private *priv = usb_get_serial_port_data(port);
> struct device *dev = &port->dev;
> @@ -1322,6 +1322,8 @@ static u32 get_ftdi_divisor(struct tty_struct *tty,
> __func__, priv->custom_divisor, baud);
> }
>
> + if (!baud)
> + baud = old_baud;
> if (!baud)
> baud = 9600;
This looks like it breaks B0 handling. Either way, an unrelated change.
> switch (priv->chip_type) {
> @@ -1330,8 +1332,12 @@ static u32 get_ftdi_divisor(struct tty_struct *tty,
> if (div_value == -1) {
> dev_dbg(dev, "%s - Baudrate (%d) requested is not supported\n",
> __func__, baud);
> - baud = 9600;
> + baud = old_baud ? old_baud : 9600;
> div_value = ftdi_sio_baud_to_divisor(baud);
> + if (div_value == -1) {
> + baud = 9600;
> + div_value = ftdi_sio_baud_to_divisor(baud);
> + }
> div_okay = 0;
> }
> break;
> @@ -1340,8 +1346,8 @@ static u32 get_ftdi_divisor(struct tty_struct *tty,
> div_value = ftdi_232am_baud_to_divisor(baud);
> } else {
> dev_dbg(dev, "%s - Baud rate too high!\n", __func__);
> - baud = 9600;
> - div_value = ftdi_232am_baud_to_divisor(9600);
> + baud = (old_baud >= 183 && old_baud <= 3000000) ? old_baud : 9600;
And please avoid using the ternary operator which tend to just hurt
readability.
Looks like this needs to be refactored in some way.
> @@ -1525,7 +1536,7 @@ static int set_serial_info(struct tty_struct *tty, struct serial_struct *ss)
> if (priv->flags & ASYNC_SPD_MASK)
> dev_warn_ratelimited(&port->dev, "use of SPD flags is deprecated\n");
>
> - change_speed(tty, port);
> + change_speed(tty, port, 0);
Zero has a special meaning (B0).
> }
> mutex_unlock(&priv->cfg_lock);
> return 0;
> @@ -2774,9 +2785,12 @@ static void ftdi_set_termios(struct tty_struct *tty,
> /* Drop RTS and DTR */
> clear_mctrl(port, TIOCM_DTR | TIOCM_RTS);
> } else {
> + speed_t old_baud =
> + old_termios ? tty_termios_baud_rate(old_termios) : 0;
Just use 9600 if you don't have an old termios. The termios rate should
be valid in that case.
Johan
