On Sat, Jul 24, 2004 at 11:21:36AM +0900, Takeshi Nakayama wrote:
> >>> Eduardo Horvath <eeh@NetBSD.org> wrote
> 
> > On Sat, Jul 24, 2004 at 10:08:56AM +0900, Takeshi Nakayama wrote:
> > 
> > > Our interrupt handler pending list has issue that it has
> > > possibilities to link same handler twice, since we use same value
> > > 0 to indicate the list end and a interrupt handler in-use flag.
> > 
> > I don't think this should be an issue since hardware should
> > only be able to generate an interrupt once and the interrupt
> > must be handled and acknowledged (hence unlinked) before it
> > can be re-issued.  Or am I missing something here?
> 
> The case shared same interrupt level with multiple devices
> (handled by intr_list_handler() in sparc64/intr.c), I thought it
> may occur multiple interrupts. Is it my misunderstand?

You can get multiple interrupts at the same level at the same time.
But only one from a single source.

> However, the case of software interrupts?

I suppose these could be a problem.

Eduardo