On 7/30/21 1:25 AM, Greg Kroah-Hartman wrote:
On Thu, Jul 29, 2021 at 11:43:08AM -0400, Sean Anderson wrote:
On 7/29/21 11:32 AM, Greg Kroah-Hartman wrote:
> On Thu, Jul 29, 2021 at 11:26:59AM -0400, Sean Anderson wrote:
>>
>>
>> On 7/29/21 11:01 AM, Greg Kroah-Hartman wrote:
>> > On Fri, Jul 23, 2021 at 06:31:51PM -0400, Sean Anderson wrote:
>> > > This device does not support changing baud, parity, data bits, stop
>> > > bits, or detecting breaks. Disable "changing" these settings to prevent
>> > > their termios from diverging from the actual state of the uart. To inform
>> > > users of these limitations, warn if the new termios change these
>> > > parameters. We only do this once to avoid spamming the log. These
>> > > warnings are inspired by those in the sifive driver.
>> > >
>> > > Signed-off-by: Sean Anderson <sean.anderson@xxxxxxxx>
>> > > ---
>> > >
>> > > drivers/tty/serial/uartlite.c | 52 +++++++++++++++++++++++++++++++++--
>> > > 1 file changed, 49 insertions(+), 3 deletions(-)
>> > >
>> > > diff --git a/drivers/tty/serial/uartlite.c b/drivers/tty/serial/uartlite.c
>> > > index 39c17ab206ca..0aed70039f46 100644
>> > > --- a/drivers/tty/serial/uartlite.c
>> > > +++ b/drivers/tty/serial/uartlite.c
>> > > @@ -314,7 +314,54 @@ static void ulite_set_termios(struct uart_port *port, struct ktermios *termios,
>> > > struct ktermios *old)
>> > > {
>> > > unsigned long flags;
>> > > - unsigned int baud;
>> > > + struct uartlite_data *pdata = port->private_data;
>> > > + tcflag_t old_cflag;
>> > > +
>> > > + if (termios->c_iflag & BRKINT)
>> > > + dev_err_once(port->dev, "BREAK detection not supported\n");
>> > > + termios->c_iflag &= ~BRKINT;
>> > > +
>> > > + if (termios->c_cflag & CSTOPB)
>> > > + dev_err_once(port->dev, "only one stop bit supported\n");
>> > > + termios->c_cflag &= ~CSTOPB;
>> > > +
>> > > + old_cflag = termios->c_cflag;
>> > > + termios->c_cflag &= ~(PARENB | PARODD);
>> > > + if (pdata->parity == 'e')
>> > > + termios->c_cflag |= PARENB;
>> > > + else if (pdata->parity == 'o')
>> > > + termios->c_cflag |= PARENB | PARODD;
>> > > +
>> > > + if (termios->c_cflag != old_cflag)
>> > > + dev_err_once(port->dev, "only '%c' parity supported\n",
>> > > + pdata->parity);
>> >
>> > Through all of this, you are warning that nothing is supported, yet you
>> > are continuing on as if all of this worked just fine.
>>
>> We don't. The idea is that we see if (e.g.) CSIZE is something the
>> hardware can't produce, warn about it (once), and then set it to what we
>> can support.
>
> So you are ignoring what the user wanted, and doing whatever you wanted.
>
> As you can only support one setting, why even care? Just set it to what
> you want and ignore userspace's requests.
That is exactly what we are doing. We set it to what we can support and
ignore what userspace requested.
If you can only support one set of options, just set it and always fail
the tcsetattr call which will allow userspace to know it shouldn't have
tried to do that.
We can't. set_termios returns void. All we can do is set the termios to
what they should be.
> Of course that is a pain but
> no one is going to notice kernel log messages either, right?
*shrug* Why does sifive_serial_set_termios do it?
I didn't notice it during the review process. Doesn't mean you should
copy a bad example :)
I didn't know it was a bad example :)
Generally, I presume that any recently-added drivers are using the best
practices for their subsystems. If they weren't, I'd have expected them
to have been rejected during review. Of course, that doesn't always
happen (such as in this case), but in the absence of documentation, code
is the next best source of how things should be.
In any case, I am not particularly attached to these particular
warnings, as long as the termios get set to what we support.
--Sean
