[PATCH v2 22/38] ARM: orion: switch to a per-platform handle_irq() function

[Arnd Bergmann]

On Tuesday 22 April 2014 23:53:23 Thomas Petazzoni wrote:
> 
> On Tue, 22 Apr 2014 23:45:55 +0200, Arnd Bergmann wrote:
> > On Tuesday 22 April 2014 23:26:26 Thomas Petazzoni wrote:
> > > However, the common CONFIG_MULTI_IRQ_HANDLER handler for non-DT
> > > platforms in plat-orion/irq.c doesn't match the needs of
> > > Orion5x. Also, it doesn't make much sense for orion_irq_init() to
> > > register the multi-IRQ handler: orion_irq_init() is called once for
> > > each IRQ cause/mask tuple, while the multi-IRQ handler only needs to
> > > be registered once.
> > 
> > Since you now have the code for doing MULTI_IRQ_HANDLER on Orion5x,
> > why not always enable it for both Orion and Kirkwood, and remove
> > the #ifdef?
> 
> You mean the #ifdef CONFIG_MULTI_IRQ_HANDLER ? This is because this
> piece of code is only needed if you build into the same image the DT
> and non-DT code. For pure DT, this is not needed, as the handler is in
> drivers/irqchip/irq-orion.c. For pure non-DT, this is not needed,
> because we don't use MULTI_IRQ_HANDLER. However, for mixed DT and
> non-DT, since the DT stuff selects MULTI_IRQ_HANDLER, the non-DT stuff
> must have its own.
>
> Also, this patch step is only *one* step in the (hopefully) right
> direction. It's already a 38 patches series, so it would be good to
> understand that no all cleanups can be done in this particular patch
> series. What you're suggesting is not a problem introduced by this
> patch series: the original code already had the #ifdef
> CONFIG_MULTI_IRQ_HANDLER, so I would like to take the task of removing
> it as a follow-up task, if you agree.

Nevermind then, I was trying to simplify the patches because you no
longer need the #ifdef after your change. If you prefer not to do
it right away, we can probably skip the change, since the code is
going to be removed anyway once the mach-mvebu version is complete.

	Arnd