Hi, On 02/02/2017 19.48, Ard Biesheuvel wrote:
$ git grep -C5 -ni 0x1DE7EC7EDBADC0DE arch/arm64/kvm/sys_regs.h-105-static inline void reset_unknown(struct kvm_vcpu *vcpu, arch/arm64/kvm/sys_regs.h-106- const struct sys_reg_desc *r) arch/arm64/kvm/sys_regs.h-107-{ arch/arm64/kvm/sys_regs.h-108- BUG_ON(!r->reg); arch/arm64/kvm/sys_regs.h-109- BUG_ON(r->reg >= NR_SYS_REGS); arch/arm64/kvm/sys_regs.h:110: vcpu_sys_reg(vcpu, r->reg) = 0x1de7ec7edbadc0deULL; arch/arm64/kvm/sys_regs.h-111-} arch/arm64/kvm/sys_regs.h-112-In other words (or rather, in words), KVM is triggering this exception in the guest deliberately, which I suspect has something to do with the lack of a GIC? Are you using these patches Peter mentions?
No, I'm not using Peter's patches. It's mainline Fedora 24 and rawhide QEMU. - Pekka
