Re: [PATCH 1/4] serial: 8250 driver replaceable i/o functions.

[David Daney <ddaney@caviumnetworks.com>]

H. Peter Anvin wrote:
> David Daney wrote:
>> /* sane hardware needs no mapping */
>> -static inline int map_8250_in_reg(struct uart_8250_port *up, int offset)
>> +static inline int map_8250_in_reg(struct uart_port *p, int offset)
>> {
>> -    if (up->port.iotype != UPIO_AU)
>> +    if (p->iotype != UPIO_AU)
>>         return offset;
>>     return au_io_in_map[offset];
>> }
> 
> With your changes, these functions cannot be called with p->iotype != 
> UPIO_AU anymore, correct?  So there is no need for this test...
> 

I think you are probably correct.  However, with the patch it is 
possible to move all this target specific code out of the driver.  So if 
the patch is accepted, a better follow up would be to get rid of the 
UPIO_AU things altogether.

I gave an example of how that could be done with UPIO_TSI here:

http://marc.info/?l=linux-serial&m=122333633802691&w=2

David Daney