On Wed, 2018-11-07 at 18:41 +0000, Marc Zyngier wrote: > On 06/11/18 19:40, Trent Piepho wrote: > > > > What about stable kernels that don't have the hierarchical API? > > My goal is to fix mainline first. Once we have something that works on > mainline, we can look at propagating the fix to other versions. But > mainline always comes first. This is a regression that went into 4.14. Wouldn't the appropriate action for the stable series be to undo the regression? The hierarchical design might allow unmasking the irq to be placed in a better location. And maybe knowing how the hardware worked would allow for a design that maintained a separate pending and active state in hardware, with distinct ack and eoi controls, i.e. allowed an MSI to be queued while still masked. But that doesn't seem like the kind of enhancement that would be backported to stable.
