[PATCH] Fix struct device member name in PCMCIA au1000_generic

Andrew Morton akpm at linux-foundation.org
Mon Mar 26 06:11:42 EDT 2007


On Mon, 26 Mar 2007 10:46:02 +0100 Ralf Baechle <ralf at linux-mips.org> wrote:

> On Mon, Mar 26, 2007 at 12:28:29AM -0800, Andrew Morton wrote:
> 
> > Is there any way in which we can get the various au.... drivers to compile on
> > x86?  That way they'd end up a bit better maintained.
> 
> Sounds like a plan - but not without a few problems.
> 
> The PCMCIA driver doesn't look to hard to get to compile on x86 but on a
> more general scale that's somewhat hard.  The Alchemy chips are SOCs which
> are all only available with Alchemy's MIPS32 core, so the drivers
> implementations are somewhat dirt^H^H^H^HMIPS-specific, same for other
> MIPS SOCs.
> 
> In the past several people (the loudest of them was Alan afair) have
> complained if architecture-specific drivers were selectable during the
> kernel configuration for x86 and generally users didn't miss an
> opportunity to machinegun both their feet when configuring their kernel
> so the Kconfig files now limit the selection of drivers to only those
> configurationss where it actually makes sense.

Yes, there are pros and cons.

> How about something like a new configuration flag DRIVERS_FOR_ANY_ARCH:
> 
> config PCMCIA_AU1X00
>         tristate "Au1x00 pcmcia support"
>         depends on (DRIVERS_FOR_ANY_ARCH || SOC_AU1X00) && PCMCIA
> 
> ?
> 
> But no doubt there are plenty of mines on that path ...

That sounds like a good idea.



More information about the linux-pcmcia mailing list