[U-Boot] [PATCH v2 3/7] microblaze: intc: Registering interrupt should return value

[Stephan Linz <linz@li-pro.net>]

Am Mittwoch, den 08.08.2012, 10:27 +0200 schrieb Michal Simek: 
> On 08/07/2012 10:10 PM, Stephan Linz wrote:
> > Am Montag, den 06.08.2012, 09:46 +0200 schrieb Michal Simek:
> >> Return value to find out if un/registration was succesful.
> >>
> >> Signed-off-by: Michal Simek <monstr@monstr.eu>
> >>
> >> ---
> >> v2: Add comment to header file to describe parameters and return codes
> >> ---
> >>   arch/microblaze/cpu/interrupts.c              |   16 +++++++++-------
> >>   arch/microblaze/include/asm/microblaze_intc.h |   11 ++++++++++-
> >>   2 files changed, 19 insertions(+), 8 deletions(-)
> >>
> >> diff --git a/arch/microblaze/cpu/interrupts.c b/arch/microblaze/cpu/interrupts.c
> >> index ee67082..08f6bad 100644
> >> --- a/arch/microblaze/cpu/interrupts.c
> >> +++ b/arch/microblaze/cpu/interrupts.c
> >> @@ -91,14 +91,13 @@ static void disable_one_interrupt(int irq)
> >>   #endif
> >>   }
> >>
> >> -/* adding new handler for interrupt */
> >> -void install_interrupt_handler (int irq, interrupt_handler_t * hdlr, void *arg)
> >> +int install_interrupt_handler(int irq, interrupt_handler_t *hdlr, void *arg)
> >>   {
> >>   	struct irq_action *act;
> >>   	/* irq out of range */
> >>   	if ((irq < 0) || (irq > irq_no)) {
> >>   		puts ("IRQ out of range\n");
> >> -		return;
> >> +		return -1;
> >>   	}
> >>   	act = &vecs[irq];
> >>   	if (hdlr) {		/* enable */
> >> @@ -106,11 +105,14 @@ void install_interrupt_handler (int irq, interrupt_handler_t * hdlr, void *arg)
> >>   		act->arg = arg;
> >>   		act->count = 0;
> >>   		enable_one_interrupt (irq);
> >> -	} else {		/* disable */
> >> -		act->handler = (interrupt_handler_t *) def_hdlr;
> >> -		act->arg = (void *)irq;
> >> -		disable_one_interrupt (irq);
> >> +		return 0;
> >>   	}
> >> +
> >> +	/* Disable */
> >> +	act->handler = (interrupt_handler_t *) def_hdlr;
> >> +	act->arg = (void *)irq;
> >> +	disable_one_interrupt(irq);
> >> +	return 1;
> >>   }
> >>
> >>   /* initialization interrupt controller - hardware */
> >> diff --git a/arch/microblaze/include/asm/microblaze_intc.h b/arch/microblaze/include/asm/microblaze_intc.h
> >> index 6142b9c..e9640f5 100644
> >> --- a/arch/microblaze/include/asm/microblaze_intc.h
> >> +++ b/arch/microblaze/include/asm/microblaze_intc.h
> >> @@ -39,7 +39,16 @@ struct irq_action {
> >>   	int count; /* number of interrupt */
> >>   };
> >>
> >> -void install_interrupt_handler (int irq, interrupt_handler_t * hdlr,
> >> +/**
> >> + * Register and unregister interrupt handler rutines
> >> + *
> >> + * @param irq	IRQ number
> >> + * @param hdlr	Interrupt handler rutine
> >> + * @param arg	Pointer to argument which is passed to int. handler rutine
> >> + * @return	0 if registration pass, 1 if unregistration pass,
> >> + *		or an error code < 0 otherwise
> >> + */
> >> +int install_interrupt_handler(int irq, interrupt_handler_t *hdlr,
> >>   				       void *arg);
> >
> > Hi Michal,
> >
> > why not two different functions here, one for registration, another one
> > for unregistration? To mee it is puzzling to use a 'install' function
> > for unregistration ...
> 
> partially agree with that. Maybe we could introduce one macro for that.
> 
> #define uninstall_interrupt_handler(irq) install_interrupt_handler(irq, NULL, NULL)

yes, that is ok ...

> 
> 
> > ... whatever, you should evaluate the return code in fsl_init2() too.
> 
> Not necessary to do it in this patch.

yes, it's another part ...

br,
Stephan