Hello! > The need to add num_irq to this prototype reveals the ugliness > of it as an interface.
> I don't think we gain much by making these two functions common, > and we do get a lot of churn in the existing code below. You know, i was also thinking about it. And actually there is a way to resolve this: make some common structure holding num_irqs, num_cpus and dev_fd, and share it between GICv2 and GICv3 code. But wouldn't this be ugly too? So, i decided to make some compromise and leave it this way, at least for now. Regarding the latter two functions, i thought that they are useful for future live migration, aren't they? > It would be much nicer to convert the > GIC to using named GPIO arrays for its incoming interrupt > lines, so that you can handle the different cases of SPIs > and PPIs naturally rather than exposing the awkward mapping > from multiple different sets of interrupts into a single > integer space. No no, not now. I believe this would require lots if changes in all machne models, i'm not ready for this... Kind regards, Pavel Fedin Expert Engineer Samsung Electronics Research center Russia
