[PATCH 1/3] Adapt ipic driver to new host_ops interface, addset_irq_type to set IRQ sense

[Li Yang-r58472]

> -----Original Message-----
> From: Benjamin Herrenschmidt [mailto:benh at kernel.crashing.org]
> Sent: Thursday, August 24, 2006 1:56 PM
> To: Li Yang-r58472
> Cc: Phillips Kim-R1AAHA; linuxppc-dev at ozlabs.org
> Subject: RE: [PATCH 1/3] Adapt ipic driver to new host_ops
> interface,addset_irq_type to set IRQ sense
> 
> 
> > But for IRQ with a specified type, set_irq_type() will be called two
> > times continuously.  Why not call the set_irq_type() in any case in
> > irq_create_of_mapping()?
> 
> well... it would still not be called by irq_create_mapping() (the
non-OF
> one). You need to make sure your stuff is properly initialized in the
> case there it's not called. That is that you have a proper default irq
> type, handler, and HW setup after host_map() returns.
> 
> It's not terribly harmful to have set_irq_type() called twice. Another
> option is to have host_map re-do what your set_type() does, or move
the
> common code into a low level __set_type() that gets called by both
> set_irq_type() and your host_map ...

Currently, it's difficult to get the irq type in host_map() before it is
set in set_type().  Shall we add flow_type to the host_map() parameters?