[Adeos-main] Why is __adeos_irq_regs only updated when a timer IRQ is caught?

[Adeos-main] Why is __adeos_irq_regs only updated when a timer IRQ is caught?

[张 静]

Hi, all,

I have no idea about why __adeos_irq_regs is only updated when a timer IRQ 
is caught. Does it means its value doesn't matter to other irq handlers? 
I'am working on version 2.4r12 for ARM.

Below is the snippet to update the __adeos_irq_regs in __adeos_handle_irq:

if (likely(irq == ADEOS_TIMER_IRQ))
{
    __adeos_irq_regs.ARM_cpsr = regs->ARM_cpsr;
    __adeos_irq_regs.ARM_pc = regs->ARM_pc;
} 

Thanks!

Re: [Adeos-main] Why is __adeos_irq_regs only updated when a timer IRQ is caught?

[Heikki Lindholm]

张 静 kirjoitti:
> Hi, all,
> 
> I have no idea about why __adeos_irq_regs is only updated when a timer 
> IRQ is caught. Does it means its value doesn't matter to other irq 
> handlers? I'am working on version 2.4r12 for ARM.

If you look at the irq handlers you'll see that most don't use the regs
parameter
and there have been talks about removing it altogether from the kernel.
Also,
if you think about the irq replaying scheme of Adeos and how you would
implement
saving the regs for every (to be) replayed interrupt, maybe you'll see
the point of
doing only the necessary reg saving in Adeos.

-- Heikki Lindholm

Re: [Adeos-main] Why is __adeos_irq_regs only updated when a timerIRQ is caught?

[Zhang Jing]

Yes, you are right. I've got it.

Thank you!

Jing Zhang


>From: Heikki Lindholm <holindho@domain.hid>
>To: 张 静 <zhang_jing_bit@domain.hid>
>CC: adeos-main@gna.org
>Subject: Re: [Adeos-main] Why is __adeos_irq_regs only updated when a 
timerIRQ	is caught?
>Date: Mon, 12 Dec 2005 11:11:04 +0200
>
>张 静 kirjoitti:
> > Hi, all,
> >
> > I have no idea about why __adeos_irq_regs is only updated when a timer
> > IRQ is caught. Does it means its value doesn't matter to other irq
> > handlers? I'am working on version 2.4r12 for ARM.
>
>If you look at the irq handlers you'll see that most don't use the regs
>parameter
>and there have been talks about removing it altogether from the kernel.
>Also,
>if you think about the irq replaying scheme of Adeos and how you would
>implement
>saving the regs for every (to be) replayed interrupt, maybe you'll see
>the point of
>doing only the necessary reg saving in Adeos.
>
>-- Heikki Lindholm
>
>_______________________________________________
>Adeos-main mailing list
>Adeos-main@domain.hid
>https://mail.gna.org/listinfo/adeos-main