On Fri, Oct 05, 2018 at 05:36:56PM +0300, Matwey V. Kornilov wrote:
> пт, 5 окт. 2018 г. в 17:33, Uwe Kleine-König <u.kleine-koenig@xxxxxxxxxxxxxx>:
> >
> > On Fri, Oct 05, 2018 at 01:27:34PM +0300, Matwey V. Kornilov wrote:
> > > чт, 4 окт. 2018 г. в 11:55, Andy Shevchenko <andriy.shevchenko@xxxxxxxxxxxxxxx>:
> > > >
> > > > On Thu, Oct 04, 2018 at 09:23:16AM +0200, Uwe Kleine-König wrote:
> > > > > serial8250_tx_chars() is called from several places and the callers don't
> > > > > consider that sending might not be allowed yet because the
> > > > > rts_before_send delay isn't over yet.
> > > > >
> > > > > So teach serial8250_tx_chars() about rts_before_send and delay the
> > > > > request accordingly if required.
> > > > >
> > > > > Signed-off-by: Uwe Kleine-König <u.kleine-koenig@xxxxxxxxxxxxxx>
> > > > > ---
> > > > > Cc: Tomas Melin <tomas.melin@xxxxxxxxxxx>
> > > > > Cc: Daniel Golle <daniel@xxxxxxxxxxxxxx>
> > > > > Cc: Ricardo Ribalda Delgado <ricardo.ribalda@xxxxxxxxx>
> > > > > Cc: Lukas Wunner <lukas@xxxxxxxxx>
> > > > > Cc: Aaron Sierra <asierra@xxxxxxxxxxx>
> > > > > Cc: Rafael Gago <rafael.gago@xxxxxxxxx>
> > > > > Cc: Daniel Jedrychowski <avistel@xxxxxxxxx>
> > > > >
> > > > > Hello,
> > > > >
> > > > > on a powerpc (MPC8313) this patch prevents sending characters too early
> > > > > after an edge on CTS is detected. As the 8250 driver is quite "advanced"
> > > > > I'd welcome testing feedback for other chip variants, so I added a few
> > > > > people to Cc who did rs485 stuff in the past.
> > > >
> > > > You pretty much missed the author of contribution of RS485 emulation
> > > > to 8250 driver.
> > > >
> > > > Cc'ed to Matwey.
> > >
> > > I still don't understand the following. The only one user for this
> > > code is 8250_omap which is armv7l. Where did MPC8313 came from?
> >
> > It's a device tree based system. On my machine it looks similar to
> > arch/powerpc/boot/dts/mpc8313erdb.dts:
> >
> > serial0: serial@4500 {
> > cell-index = <0>;
> > device_type = "serial";
> > compatible = "fsl,ns16550", "ns16550";
> > reg = <0x4500 0x100>;
> > clock-frequency = <132000000>;
> > interrupts = <9 0x8>;
> > interrupt-parent = <&ipic>;
> > };
> >
> > The corresponding device is instantiated from
> > drivers/tty/serial/8250/8250_of.c.
> >
> > There is:
> >
> > port->handle_irq = fsl8250_handle_irq;
> >
> > and in this function there is a call to serial8250_tx_chars().
> >
> > But there is also serial8250_handle_irq() in
> > drivers/tty/serial/8250/8250_port.c which calls serial8250_tx_chars().
>
> Right. But serial8250_em485_init() is never called for your port
> device. So, em485 is always NULL for you, isn't it?
Ok, you're right. (Testing of this patch wasn't in my hand. I just got
an error description, found the above mentioned issue and provided the
patch. The tester updated from 4.9 to 4.16 + my patch and reported the
issue gone. So the problem was obviously already fixed by the kernel
update.)
Still the issue is real for omap, isn't it?
Best regards
Uwe
--
Pengutronix e.K. | Uwe Kleine-König |
Industrial Linux Solutions | http://www.pengutronix.de/ |
