Bug in drivers/serial/of_serial.c?

[Grant Likely]

On Sun, Nov 22, 2009 at 3:42 PM, Grant Likely <grant.likely at secretlab.ca> wrote:
> On Sat, Nov 21, 2009 at 12:45 PM, Arnd Bergmann <arnd at arndb.de> wrote:
>> On Saturday 21 November 2009, Grant Likely wrote:
>>> BTW, while looking at that file... Arnd, does the .type = "serial"
>>> stuff really need to be there?
>>
>> Well, serial is one of the few that actually has some preexisting
>> binding with a well defined device-type, so I guess we should use
>> it.
>
> But device-type describes a OpenFirmware API, not a device register
> interface.  Once we're in the kernel and talking to the hardware
> directly, device-type has no real meaning because we're not using the
> OpenFirmware interface to talk to serial devices in a generic manner.

And looking for it in the driver forces us to put meaningless
"device_type" properties in the flat trees, where they make no sense
whatsoever.

g.

-- 
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.