On Thu, 2018-11-22 at 16:38 +0000, Marc Zyngier wrote: > > To add to Lorenzo's comment, we're trying hard to have a *minimal* fix > that can be easily backported. Changing variable and field names as well > as flipping the semantic of other bits of the driver makes it harder to > review, and certainly doesn't help getting things backported to stable > (see the stable kernel rules). > > I'd suggest this kind of repainting is better kept as a separate patch > and merged separately. If you want a minimal fix to backport, you should just use my original patch for 4.15-4.19 stable series. It's a one line change to fix the real bug. It'll be far easier to backport to < 4.17, where the hierarchical irq domain stuff wasn't used in this part of the driver. If you follow the control flow of an irq, you'll see that it ACKs the irq at the same point as putting the ack in a new irq_ack method does. While the driver shouldn't enable or disable on mask, that's also a new bug from 4.17 that doesn't need to be fixed in earlier versions. It's unlikely to come up, in 4.17+, since the mask methods don't get used in normal irq flow. Still a bug.
