Re: [Xen-devel] [PATCH] x86/vmx: Improvements to vmentry failure handling

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 23.10.15 at 19:52, <andrew.cooper3@xxxxxxxxxx> wrote:
> --- a/xen/arch/x86/hvm/vmx/vmcs.c
> +++ b/xen/arch/x86/hvm/vmx/vmcs.c
> @@ -1588,21 +1588,14 @@ void vmx_destroy_vmcs(struct vcpu *v)
>      free_xenheap_page(v->arch.hvm_vmx.msr_bitmap);
>  }
>  
> -void vm_launch_fail(void)
> -{
> -    unsigned long error;
> -
> -    __vmread(VM_INSTRUCTION_ERROR, &error);
> -    printk("<vm_launch_fail> error code %lx\n", error);
> -    domain_crash_synchronous();
> -}
> -
> -void vm_resume_fail(void)
> +void vmx_vmentry_failure(void)
>  {
> +    struct vcpu *curr = current;
>      unsigned long error;
>  
>      __vmread(VM_INSTRUCTION_ERROR, &error);
> -    printk("<vm_resume_fail> error code %lx\n", error);
> +    printk(XENLOG_ERR "VM%s error: %#lx\n",
> +           curr->arch.hvm_vmx.launched ? "RESUME" : "LAUNCH", error);
>      domain_crash_synchronous();
>  }

Imo this should be gprintk() both for log level purposes (subject to
a follow-up discussion on XSA-152 which I'll kick off in a subsequent
mail) and inclusion of the offending vCPU's ID in what gets printed.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel