Re: [PATCHv3 0/3] OMAP UART: Adding support for omap-serial driver.

[Kevin Hilman <khilman@deeprootsystems.com>]

Tony Lindgren <tony@atomide.com> writes:

> * Olof Johansson <olof@lixom.net> [091124 23:05]:
>> G, Manjunath Kondaiah wrote:
>> >>>You should allow both of them to be enabled at the same time, so the
>> >>>same
>> >>>kernel can for example be booted on a ZOOM2 with debug board
>> >>>attached
>> >>>(8250 on GPMC), or on a beagle/overo board.
>> >>>
>> >>>Making them exclusive would be a step backwards.
>> >>More so, selecting both config option really shouldn't break the
>> >>build. If one expect a build break because only one of the
>> >>options can be enabled then clearly the dependency should be
>> >>used so that user can select only one option at a time.
>> >
>> >Ack. There should be dependency in Kconfig so that user can select
>> >either 8250 console(through Quad-uart for zoom2/zoom3) or OMAP-serial
>> >Console(through OMAP UART's).
>> 
>> No. The drivers MUST handle co-existing. You can have a bunch of
>> drivers configured to be _potential_ consoles, which one will end up
>> being used depends on, among other things, the bootargs. But it
>> should be allowed to configure both 8250 and omap-serial at the same
>> time.
>
> Exactly. Even the same board must be capable of using 8250 and
> omap-serial. That's why it must be called ttyO instead of ttyS.

I completely agree with the need for these drivers to co-exist and for
a given board to use both drivers.  For example, Zoom2 would want to
use omap-serial for the on-chip UARTs, and 8250 for the external UART
used for the console.

In my review I complained about using ttyO instead of ttyS, but now I
see more clearly the reason for that.

Kevin