Re: [PATCH V2 1/2] dt-bindings: Add optional gpio property to uartps node to support rs485

[Rob Herring <robh@kernel.org>]

On Thu, Oct 12, 2023 at 08:35:59PM +0200, m.brock@vanmierlo.com wrote:
> Manikanta Guntupalli wrote on 2023-10-11 16:56:
> > Add optional gpio property to uartps node and reference to rs485.yaml
> > 
> > On Xilinx/AMD Kria SOM KD240 board rs485 connects via TI ISOW1432
> > Transceiver device, where one GPIO is used for driving DE/RE signals.
> > With rs485 half duplex configuration, DE and RE shorts to each other,
> 
> s/shorts/are connected
> 
> > and at a time, any node acts as either a driver or a receiver.
> > 
> > Here,
> > DE - Driver enable. If pin is floating, driver is disabled.
> > RE - Receiver enable. If pin is floating, receiver buffer is disabled.
> 
> Please use DE and /RE to indicate DE is active high and /RE is active low.
> 
> > xlnx,phy-ctrl-gpios is optional property, because it is not required
> > for uart console node.
> 
> How about introducing an rs485 generic gpios property instead of xlnx
> private one? See also rs485-term-gpios and rs485-rx-during-tx-gpios.
> 
> Also note that every kernel driver expects to use RTS for this purpose.
> So why not give this driver the option to choose a gpio instead of its
> native RTS? And from there on use the rts route?
> What if someone wants to use normal (non-rs485) RTS on a GPIO instead
> of the native pin?
> 
> @Rob Herring
> I am curious to know how the rs485 maintainers look at this.

Ask them.

We already have 'rts-gpios'. If that's what's always used, then perhaps 
we should use that in the RS485 case too?

Rob