[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <alpine.DEB.2.21.9999.1904111412210.19990@viisi.sifive.com>
Date: Thu, 11 Apr 2019 14:13:24 -0700 (PDT)
From: Paul Walmsley <paul.walmsley@...ive.com>
To: Andreas Schwab <schwab@...e.de>
cc: Paul Walmsley <paul.walmsley@...ive.com>,
linux-kernel@...r.kernel.org, linux-riscv@...ts.infradead.org,
linux-serial@...r.kernel.org, gregkh@...uxfoundation.org,
Paul Walmsley <paul@...an.com>, Jiri Slaby <jslaby@...e.com>,
Palmer Dabbelt <palmer@...ive.com>,
Wesley Terpstra <wesley@...ive.com>,
Julia Lawall <julia.lawall@...6.fr>
Subject: Re: [PATCH v4 2/2] tty: serial: add driver for the SiFive UART
On Thu, 11 Apr 2019, Andreas Schwab wrote:
> On Apr 11 2019, Paul Walmsley <paul.walmsley@...ive.com> wrote:
>
> > +static void sifive_serial_set_termios(struct uart_port *port,
> > + struct ktermios *termios,
> > + struct ktermios *old)
> > +{
> > + struct sifive_serial_port *ssp = port_to_sifive_serial_port(port);
> > + unsigned long flags;
> > + u32 v, old_v;
> > + int rate;
> > + char nstop;
> > +
> > + if ((termios->c_cflag & CSIZE) != CS8) {
> > + dev_err(ssp->port.dev, "only 8-bit words supported\n");
> > + return;
> > + }
> > +
> > + /* Set number of stop bits */
> > + nstop = (termios->c_cflag & CSTOPB) ? 2 : 1;
> > + __ssp_set_stop_bits(ssp, nstop);
> > +
> > + /* Set line rate */
> > + rate = uart_get_baud_rate(port, termios, old, 0, ssp->clkin_rate / 16);
> > + __ssp_update_baud_rate(ssp, rate);
> > +
> > + spin_lock_irqsave(&ssp->port.lock, flags);
> > +
> > + /*
> > + * Update the per-port timeout.
> > + */
> > + uart_update_timeout(port, termios->c_cflag, rate);
> > +
> > + ssp->port.read_status_mask = 0;
> > + if (termios->c_iflag & INPCK) {
> > + dev_err(ssp->port.dev, "INPCK flag not supported\n");
> > + goto ssst_out;
> > + }
> > + if (termios->c_iflag & (BRKINT | PARMRK)) {
> > + dev_err(ssp->port.dev, "BRKINT/PARMRK flag not supported\n");
> > + goto ssst_out;
> > + }
>
> I don't think it is a good idea to print something while the port is
> locked.
>
> And I think set_termios is supposed to apply all applicable settings,
> even if some of them are not supported.
Agreed. Thank you for your help debugging (some of which was off-list).
Clearly there should have been a build-test with lockdep enabled.
I will send a v5.
- Paul
Powered by blists - more mailing lists