On 05/07/2018 02:00 PM, van der Linden, Frank wrote:

Hi Boris,

Thanks for the feedback.

On 5/7/18, 8:13 AM, "Boris Ostrovsky" boris.ostrovsky@oracle.com wrote:

> diff --git a/arch/x86/xen/enlighten_hvm.c b/arch/x86/xen/enlighten_hvm.c
> index 6b424da1ce75..c78b3e8fb2e5 100644
> --- a/arch/x86/xen/enlighten_hvm.c
> +++ b/arch/x86/xen/enlighten_hvm.c
> @@ -71,6 +71,19 @@ static void __init xen_hvm_init_mem_mapping(void)
>  {
>  	early_memunmap(HYPERVISOR_shared_info, PAGE_SIZE);
>  	HYPERVISOR_shared_info = __va(PFN_PHYS(shared_info_pfn));
> +
> +	/*
> +	 * The virtual address of the shared_info page has changed, so
> +	 * the vcpu_info pointer for VCPU 0 is now stale.

Is it "has changed" or "has changed if kaslr is on"?

It's "has changed". See commit 4ca83dcf4e3bc0c98836dbb97553792ca7ea5429 . It's a way to make kaslr work, but it's done regardless of whether it's enabled or not.

I completely forgot about this one.

> + * > + * The prepare_boot_cpu callback will re-initialize it via > + * xen_vcpu_setup, but we can't rely on that to be called for > + * old Xen versions (xen_have_vector_callback == 0). > + * > + * It is, in any case, bad to have a stale vcpu_info pointer > + * so reset it now. > + */ > + xen_vcpu_info_reset(0); Why not xen_vcpu_setup(0)? Basically, I wanted to be minimally invasive. xen_vcpu_setup does a little more work (tries to do the VCPU placement hypercall), and will be called later in any case. So doing just the basic xen_vcpu_info_reset for VCPU 0 seems like the best way to do it; it just re-iterates what is done for VCPU 0 earlier in boot, which is also a vcpu_info_reset.

OK, fair enough. This should go to stable as well I think (4.12+), copying them.

Reviewed-by: Boris Ostrovsky boris.ostrovsky@oracle.com