Re: [Patch] Add support for Poseidon board.

Re: [Patch] Add support for Poseidon board.

[Heiko Schocher]

Hello Paul,

Sorry for the long delay, but in Christmas time the idle() function is
never called ;-)

Paul Mundt wrote:
> On Wed, Dec 12, 2007 at 03:08:32PM +0100, Manuel Lauss wrote:
>   
>> On Wed, Dec 12, 2007 at 08:57:57AM +0100, Heiko Schocher wrote:
>>     
>>> --- /dev/null
>>> +++ b/arch/sh/boards/emblux/poseidon/io.c
>>> @@ -0,0 +1,171 @@
>>> +/*
>>> + * linux/arch/sh/boards/emblux/poseidon/io.c
>>> + *
>>> + * Copyright (C) 2007 Heiko Schocher <heiko.schocher@invitel.hu>
>>> + *
>>> + * I/O routine for Poseidon.
>>> + *
>>> + */
>>>       
>
> Please kill this io.c completely. If the generic routines don't work for
> you, we can look at conditionalizing timeouts there if it's absolutely
> needed. Other architectures have similar hooks for broken platforms.
>   

I played with the Bus-timing registers, but without any success :-(
When I disable the cache it works fine with the generic io routines!

Now I dont know, if we have to modify the generic routines, because
I think its a really special problem for me, and that shouldnt go into
generic routines, or?

bye,
Heiko

Re: [Patch] Add support for Poseidon board.

[Paul Mundt]

On Thu, Dec 27, 2007 at 09:34:05AM +0100, Heiko Schocher wrote:
> Paul Mundt wrote:
> > Please kill this io.c completely. If the generic routines don't work for
> > you, we can look at conditionalizing timeouts there if it's absolutely
> > needed. Other architectures have similar hooks for broken platforms.
> 
> I played with the Bus-timing registers, but without any success :-(
> When I disable the cache it works fine with the generic io routines!
> 
> Now I dont know, if we have to modify the generic routines, because
> I think its a really special problem for me, and that shouldnt go into
> generic routines, or?

SLOW_DOWN_IO can be done generically. Look at mips or x86 for an example.

There's no reason to define it for anyone else, and it will be
pre-processed away.