On Sun, Mar 16, 2014 at 03:44:00PM +1100, Anton Blanchard wrote: > > I was testing KVM on an upstream kernel (3.14.0-rc6-00145-ga4ecdf8) > with Gregory's emulated MMIO fixes. > > I ran with 16 cores and 4 threads (matches the host): > > qemu-system-ppc64 -enable-kvm -smp cores=16,threads=4 -m 32G -M pseries -cpu > POWER7 -nographic -nodefaults -monitor stdio -serial pty -drive file=XXX > -netdev bridge,br=br0,id=net0,helper=/usr/libexec/qemu-bridge-helper -device > virtio-net-pci,netdev=net0 > > Note that in kernel XICS acceleration wasn't enabled (perhaps it should > be the default, it isn't right now): > > qemu-system-ppc64: KVM and IRQ_XICS capability must be present for in-kernel > XICS > > During boot I see processors stuck in the guest: > > [ 16.145166] Processor 1 is stuck. > > And thousands of: > > kvmppc_wait_for_nap timeout 0 1 > kvmppc_wait_for_nap timeout 0 1 > kvmppc_wait_for_nap timeout 0 1 > kvmppc_wait_for_nap timeout 0 1
You need the couple of patches I posted recently, which Paolo Bonzini is hopefully going to get to Linus before 3.14 is released, and which are in the "next" branch of the kvm tree: KVM: PPC: Book3S HV: Remove bogus duplicate code KVM: PPC: Book3S HV: Fix register usage when loading/saving VRSAVE as well as the series that you have in your "le" branch and Alex has in his "kvm-ppc-queue" branch: PPC: KVM: introduce helper to check RESUME_GUEST and related PPC: KVM: fix VCPU run for HV KVM (v2) PPC: KVM: fix RESUME_GUEST check before ending CEDE in kvmppc_run_core() PPC: KVM: fix RESUME_GUEST check before returning from kvmppc_run_core() Paul. _______________________________________________ Linuxppc-dev mailing list Linuxppc-dev@lists.ozlabs.org https://lists.ozlabs.org/listinfo/linuxppc-dev
