On Sat, 21 Oct 2017, Christoffer Dall wrote: > On Fri, Oct 20, 2017 at 01:56:16PM +0200, Thomas Gleixner wrote: > > On Fri, 20 Oct 2017, Christoffer Dall wrote: > > > > Please Cc lkml when changing genirq core code. > > > > ok, will do so in the future. > > > > It is currently unclear how to set the VCPU affinity for a percpu_devid > > > interrupt , since the Linux irq_data structure describes the state for > > > multiple interrupts, one for each physical CPU on the system. Since > > > each such interrupt can be associated with different VCPUs or none at > > > all, associating a single VCPU state with such an interrupt does not > > > capture the necessary semantics. > > > > > > The implementers of irq_set_affinity are the Intel and AMD IOMMUs, and > > > the ARM GIC irqchip. The Intel and AMD callers do not appear to use > > > percpu_devid interrupts, and the ARM GIC implementation only checks the > > > pointer against NULL vs. non-NULL. > > > > > > Therefore, simply update the function documentation to explain the > > > expected use in the context of percpu_devid interrupts, allowing future > > > changes or additions to irqchip implementers to do the right thing. > > > > > > This allows us to set the VCPU affinity for the virtual timer interrupt > > > in KVM/ARM, which is a percpu_devid (PPI) interrupt. > > > > I have a hard time to understand how adding the comment allows you to set > > VCPU affinity .... > > > > Obviously everything already works without adding the comment, however > it is not entirely clear how to use the vcpu_into for percpu_devid > interrupts, and this patch aims to make that clear, as it is the first > user of irq_set_vcpu_affinity for a percpu_devid interrupt which > actually uses the vcpu_info pointer. > > Would you rather that we did not update the documentation or what do you > suggest? Documentation is welcome of course. I just want a changelog which is not misleading. i.e. it says: Update the function documentation ..... This allow us to set .... Which reads like: Updating the documentation allows us to set ... See? Thanks, tglx
