On Mon, Jan 11, 2016 at 04:13:12PM +0000, Peter Maydell wrote:

On 11 January 2016 at 16:09, Andrew Jones [off-list ref] wrote:

quoted

I think the save/restore case is where I always flip to seeing it as a
bunch of separate per cpu devices. It would feel better to me to
save/restore the cpu-gic registers the same way we do all other cpu
registers.

From QEMU's point of view this would be extremely awkward, which
is one reason I'm happy with the GICv3 API we came up with
and reviewed last year...

OK, I'll shut up about this. As I didn't speak then, I shall now forever
hold my peace :)

drew