kernel 2.6.3 on JS20

[Benjamin Herrenschmidt]

On Sat, 2004-03-06 at 02:05, Jake Moilanen wrote:
> > The interrupt is shared with something else ? If not it should have
> > been disabled on the controller in the first place...
>
> What the IDE sequence is:
>
> 	disable_irq()
> 	probe_for_drive()    <- Generates an interrupt
> 	enable_irq()
> 	<< we get interrupted here >>
> 	request_irq()
>
> It appears that the interrupt is queued and not dropped on the floor.
> It is my understanding that it is hardware dependent if it gets queued
> or dropped when ibm,int-off rtas call is made.

The enable_irq should do nothing if there is no hander attached. I
know it's all a bit hairy, the IDE code is definitely asking for
trouble, but that's the way it should work.

Ben.


** Sent via the linuxppc64-dev mail list. See http://lists.linuxppc.org/