Re: KVM on Via Nano (Isaiah) CPUs? <Virus checked>

[Avi Kivity <avi@redhat.com>]

Andreas Tanz wrote:
> Mar 23 14:27:35 [kernel] [ 5252.775939] vmx->handle_exception 10 : handle_rmode_exception(vcpu, intr_info & INTR_INFO_VECTOR_MASK, error_code) returned 0x1
> Mar 23 14:27:35 [kernel] [ 5252.775955] vmx->vmx_vcpu_run() 00 : vmcs_read32(VM_ENTRY_INTR_INFO_FIELD) returned 0x8000040d
> Mar 23 14:27:35 [kernel] [ 5252.775967] vmx->handle_exception 00 : giving some infos
> Mar 23 14:27:35 [kernel] [ 5252.775972] vmx->handle_exception 01 : vect_info: 0x0
> Mar 23 14:27:35 [kernel] [ 5252.775978] vmx->handle_exception 02 : intr_info: 0x80000b0d, is_page_fault()==0
> Mar 23 14:27:35 [kernel] [ 5252.775984] vmx->handle_exception 03 : irq_chip_in_kernel()==1
> Mar 23 14:27:35 [kernel] [ 5252.775989] vmx->handle_exception 04 : is_external_interrupt()==0
> Mar 23 14:27:35 [kernel] [ 5252.775995] vmx->handle_exception 0a : kvm_rip_read(vcpu) returned 0x3154
> Mar 23 14:27:35 [kernel] [ 5252.776000] vmx->handle_exception 0f : vcpu->arch.rmode.active: 0x1
> Mar 23 14:27:35 [kernel] [ 5252.776015] vmx->handle_exception 10 : handle_rmode_exception(vcpu, intr_info & INTR_INFO_VECTOR_MASK, error_code) returned 0x1
>
>   

Okay, the data is already poisoned at this point.  We need something 
from earlier in time.

Try adding at the beginning of kvm_handle_exit() (after the three lines 
of declarations):

    {
        static int counter;

        if (kvm_rip_read(vcpu) == 0x3154 && counter++ >= 2) {
            kvm_run->exit_reason = KVM_EXIT_UNKNOWN;
            kvm_run->hw.hardware_exit_reason = exit_reason;
            return 0;
        }
    }

This time, qemu should exit instead of looping.  Please post the last 
few screenfulls of dmesg.

-- 
error compiling committee.c: too many arguments to function