RE: oops 2.5.68 ohci1394/ IRQ/acpi

[Grover, Andrew]

Ohhh so we need to not just return nonzero, but return 1 (aka
IRQ_HANDLED?) Well, then this makes sense. Sorry about that.

btw I think the line in handle_IRQ_event that reads

if (retval != 1) {

should be

if (retval != IRQ_HANDLED) {

but that's just a nit.

Regards -- Andy

> -----Original Message-----
> From: Andrew Morton [mailto:[e-mail]] 
> Sent: Monday, May 05, 2003 2:16 PM
> To: Nicolas
> Cc: [e-mail]; Grover, Andrew
> Subject: Re: oops 2.5.68 ohci1394/ IRQ/acpi
> Importance: High
> 
> 
> Nicolas <[e-mail]> wrote:
> >
> > 
> > May  5 13:36:56 hal9003 kernel: irq 9: nobody cared!
> > ...
> > May  5 13:36:56 hal9003 kernel: handlers:
> > May  5 13:36:56 hal9003 kernel: [acpi_irq+0/17] (acpi_irq+0x0/0x11) 
> > May  5 13:36:56 hal9003 kernel: [<c01c1ff0>] (acpi_irq+0x0/0x11)
> 
> Look like the ACPI IRQ handler isn't returning an appropriate value.
> 
> Can you test this patch?
> 
> diff -puN drivers/acpi/osl.c~acpi-irq-ret-fix drivers/acpi/osl.c
> --- 25/drivers/acpi/osl.c~acpi-irq-ret-fix	Mon May  5 14:14:24 2003
> +++ 25-akpm/drivers/acpi/osl.c	Mon May  5 14:14:38 2003
> @@ -237,7 +237,7 @@ acpi_os_table_override (struct acpi_tabl  
> static irqreturn_t  acpi_irq(int irq, void *dev_id, struct 
> pt_regs *regs)  {
> -	return (*acpi_irq_handler)(acpi_irq_context);
> +	return (*acpi_irq_handler)(acpi_irq_context) ? 
> IRQ_HANDLED : IRQ_NONE;
>  }
>  
>  acpi_status
> 
> _
>