Cascaded interrupts on PPC405 (Virtex-II Pro)

[Joshua Lamorie]

Gidday there,

Yes, I see this behaviour. When I first boot and load all the modules 
and run my test it appears to work. But when I unload everything and 
start it back up again, I get this problem. Unfortunately, your 
suggestion to clear the IRQ_DISABLED flag in __rt_umount_rtai doesn't 
improve things. My system doesn't have syslog running but I do see 
/proc/interrupts displaying a BAD interrupt.

Shouldn't I be doing this in __rt_mount_rtai() ?

Joshua


Wolfgang Grandegger wrote:

>Normally it gets disabled, when the Linux interrupt displatcher detects
>an unregistered interrupt:
>
>  $ cat <kernel-path>
>  ..
>  void ppc_irq_dispatch_handler(struct pt_regs *regs, int irq)
>  {
>  ...
>        if (!action || !action->handler) {
>
>           ppc_spurious_interrupts++;
>           printk(KERN_DEBUG "Unhandled interrupt %x,  disabled\n",
>                  irq);
>           /* We can't call disable_irq here, it would deadlock */
>           ++desc->depth;
>           desc->status |= IRQ_DISABLED;
>  ...
>
>If this is the case you should realize the above printk message as well.
>The question then is: why is it enabled. To fix this problem you may
>clear the status when RTAI takes over the interrupt handling e.g. in
>rtai.c or hal.c:
>
>  void __rt_umount_rtai(void)
>  {
>        ...
>	for (i = 0; i < NR_IRQS; i++) {
>		IRQ_DESC[i].status &= ~IRQ_DISABLED;
>		IRQ_DESC[i].handler = linux_irq_desc_handler[i];
>	}
>
>Does it work?
>
>Wolfgang.
>
>  
>