On 13 April 2010 11:45, Alan Cox <alan@xxxxxxxxxxxxxxxxxxx> wrote:
> +static int iuu_uart_baud(struct usb_serial_port *port, u32 baud_base,
> u32 *actual, u8 parity)
> {
> int status;
> + u32 baud;
> u8 *dataout;
> u8 DataCount = 0;
> u8 T1Frekvens = 0;
> @@ -865,6 +872,8 @@
>
> if (!dataout)
> return -ENOMEM;
> + /*baud = (((priv->clk / 35) * baud_base) / 100000); */
> + baud = baud_base;
>
>
> I'm a bit confused why you create an extra variable here ?
It was initially for added readability. As you can see, the commented
out line makes an adjustment to the baud rate, so I wanted to make it
clear that the rate applied to the hardware was not necessarily the
same as the function entry value. I failed to get this working
reliably, so currently the two rates are the same.
I would like to leave it in, so that if I do get the equation right,
the change would be a single line patch.
>
>
> + /* Just use the ospeed. ispeed should be the same. */
> + baud = tty->termios->c_ospeed;
>
>
> ispeed and ospeed may be different. Prefering the ospeed if you can't set
> them separately is expected so just fine and they will get written back
> correctly by the encode_baud_rate call
Ok. So I don't need to change anything there?
>
> int result;
> + int baud = 9600; /* Fixed for the moment */
> u32 actual;
> struct iuu_private *priv = usb_get_serial_port_data(port);
>
> The expected behaviour on open is to use the termios baud rate so now
> you've allowed it to be set this should be using tty->termios as well
> (and if need be forcing a default when the port is registered)
I will modify the driver to do this. My understanding of tty is not 100%.
It just so happens, that in the user land application I am using, the
device is opened with a baud of 9600, so I guess my code works just
out of co-incidence!
I will test that everything still works after the change, and submit a
new patch.
>
> +static int iuu_ioctl(struct tty_struct *tty, struct file *file,
> + unsigned int cmd, unsigned long arg)
> +{
> + dbg("%s cmd 0x%04x", __func__, cmd);
> +
> + /* Old interface. Returning -ENOIOCTLCMD for proper termios use.
> */
> + return -ENOIOCTLCMD;
> +}
>
> This should not be needed - its the default behaviour
Strange, without this bit, my iuu_set_termios() is never called, and
none of the baud rate functionality works.
Are you sure that the return value of -ENOIOCTLCMD is the normal behaviour?
I agree that is looks to be, but in practice it did not seem to work.
From: usb-serial.c, is seems that without this function in my driver,
the usb-serial.c should return -ENOIOCTLCMD.
static int serial_ioctl(struct tty_struct *tty, struct file *file,
unsigned int cmd, unsigned long arg)
{
struct usb_serial_port *port = tty->driver_data;
int retval = -ENODEV;
dbg("%s - port %d, cmd 0x%.4x", __func__, port->number, cmd);
WARN_ON(!port->port.count);
/* pass on to the driver specific version of this function
if it is available */
if (port->serial->type->ioctl) {
retval = port->serial->type->ioctl(tty, file, cmd, arg);
} else
retval = -ENOIOCTLCMD;
return retval;
}
Kind Regards
James
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
