On Thu, 1 Oct 2015 13:47:13 +0100 Jean-Philippe Brucker <jean-philippe.brucker@xxxxxxx> wrote: > Hi, > > This series adds 32bit support to the GICv3 driver. Only SGIs, PPIs and > SPIs support is present, no ITS. > Currently, the main targets for those patches are 32bit guests running under > KVM. They can now use affinity routing and theoretically have a lot more than > 8 CPUs. During my tests, I was able to use 16 virtual CPUs on an AEM model. > > We don't support 32bit hosts with GICv3 at the moment. I added a patch to allow > building a host when CONFIG_ARM_GIC_V3 enabled, as pointed out by Vladimir in > the last series. When attempting to run such host with a device-tree that > declares GICv3, KVM will refuse to start, and throw a -ENODEV error. > > Some time has passed since last series, but I am now available full-time to act > on reviews and comments. > > Changes since v2 [2]: > - rebased on 4.3-rc3, > - add a sixth patch to handle build with CONFIG_KVM + CONFIG_ARM_GIC_V3. > > Changes since v1 [1]: > - fix the mistakes pointed out on the list, > - add a patch to select ARM_GIC_V3 under ARCH_VIRT. > > Cheers, > Jean Hi Jean-Philippe, Looks good to me: Reviewed-by: Marc Zyngier <marc.zyngier@xxxxxxx> There is some conflicts with stuff I have queued already (the 64bit SRE stuff), but I can fix that up pretty easily, assuming Christoffer doesn't have any objection to the last patch. Thanks, M. -- Jazz is not dead. It just smells funny. _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
