[parisc-linux] rp2470 hang...getting closer

[Grant Grundler]

Matthew Wilcox wrote:
> hmm.. not sure about disable_irq() -- certainly sti() is gone and __sti()
> is local_irq_disable() in 2.5; but i think enable/disable_irq still act
> globally.  i think this is because it's supposed to go and enable/disable
> delivery of interrupts in the (io)(s)(a)pic, rather than playing with
> the cpu interrupt masks.

Normally, disable_irq() will only result in the IRQ being disabled
at the PIC. ie if a regular PCI driver calls disable_irq().
But if HP device (eg Dino/IOSAPIC) calls disable_irq(), the IRQ
is in the CPU region and we muck with EIEM.

grant