Re: ioctl cleanups: enable sg_io and serial stuff to be shared

[Pavel Machek <pavel@ucw.cz>]

Hi!

>    From: Pavel Machek <pavel@ucw.cz>
>    Date: Thu, 8 May 2003 22:33:14 +0200
> 
>    What about this one: redefine it to (*ioctl)( ...., unsigned *long*,
>    unsinged long). That means we can add 
>    
>    #define IOCTL_COMPAT 0x1 0000 0000
> 
> Bzzt!  Doesn't work on 32-bit.  COMPAT does not mean 64-bit-->32-bit
> translations, stop thinking about the compat layer in this way.
> 
> It is a generic environment translation system.
> 
> Eventually we can use it for things like IBCS2 and stuff like that.

I... do not think so. 

You'd then need

.compat_linux32_ioctl
.compat_IBCS2_ioctl
...

I do not think that is doable.

> Suggest something sane like defining a macro such as
> "compat_task(tsk)" that can be tested by various bits of
> code.

That makes more sense. Unfortunately, that means that case "okay, it
is compatible" can not be told from "we did not bother to check
compat_task()". :-(. Nor do I see a transition phase.
								Pavel

-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]