Re: irqnr fallout in gpiolib on sparc32

[Ingo Molnar <mingo@elte.hu>]

* David Miller <davem@davemloft.net> wrote:

> From: Ingo Molnar <mingo@elte.hu>
> Date: Tue, 6 Jan 2009 13:57:04 +0100
> 
> > Sparc32 on the other hand had a clean IRQ layer long before x86 found its 
> > desire for a clean genirq layer - so genirq is a nuisance for Sparc32 at 
> > best and it deserves none of the not nice actions. What i am hoping for is 
> > that perhaps the Sparc unification changed that equation.
> 
> Not really, the unificiation didn't change much in this area. [...]

Yeah, i mean - "changed the equation" psychologically, the same way it did 
it on x86.

There's now two files close to each other in the namespace:

 earth4:~/tip> ls -l arch/sparc/kernel/irq*.c
 -rw-rw-r-- 1 mingo mingo 16122 2009-01-07 13:50 arch/sparc/kernel/irq_32.c
 -rw-rw-r-- 1 mingo mingo 26556 2009-01-07 12:34 arch/sparc/kernel/irq_64.c

Each of them crying out loud to be unified. Every time you open irq_64.c 
you'll think "why that ugly _64.c postfix, shouldnt this be irq.c 
instead?" ;-)

That kind of gentle pressure to unify comes straight from the fact that 
there's _32.c and _64.c postfixes around and the postfixes mess up command 
completion when those files are opened, so we notice the non-unified-ness 
again and again.

At least this was what drove many of the x86 unifications. (Mind you, 
irq_32/irq_64.c is still not fully unified on x86 ;)

> [...] That doesn't change the fact that I do intend to genirq'ify 
> sparc32 some time soon. :-)

Cool! :-)

	Ingo