Re: [PATCH v5 6/6] x86/kvm: Add kexec support for SEV Live Migration.

[Steve Rutherford]

On Tue, Jun 29, 2021 at 8:14 AM Ashish Kalra <Ashish.Kalra@xxxxxxx> wrote:
>
> From: Ashish Kalra <ashish.kalra@xxxxxxx>
>
> Reset the host's shared pages list related to kernel
> specific page encryption status settings before we load a
> new kernel by kexec. We cannot reset the complete
> shared pages list here as we need to retain the
> UEFI/OVMF firmware specific settings.
>
> The host's shared pages list is maintained for the
> guest to keep track of all unencrypted guest memory regions,
> therefore we need to explicitly mark all shared pages as
> encrypted again before rebooting into the new guest kernel.
>
> Signed-off-by: Ashish Kalra <ashish.kalra@xxxxxxx>
> ---
>  arch/x86/kernel/kvm.c | 25 +++++++++++++++++++++++++
>  1 file changed, 25 insertions(+)
>
> diff --git a/arch/x86/kernel/kvm.c b/arch/x86/kernel/kvm.c
> index a014c9bb5066..a55712ee58a1 100644
> --- a/arch/x86/kernel/kvm.c
> +++ b/arch/x86/kernel/kvm.c
> @@ -869,10 +869,35 @@ static void __init kvm_init_platform(void)
>         if (sev_active() &&
>             kvm_para_has_feature(KVM_FEATURE_MIGRATION_CONTROL)) {
>                 unsigned long nr_pages;
> +               int i;
>
>                 pv_ops.mmu.notify_page_enc_status_changed =
>                         kvm_sev_hc_page_enc_status;
>
> +               /*
> +                * Reset the host's shared pages list related to kernel
> +                * specific page encryption status settings before we load a
> +                * new kernel by kexec. Reset the page encryption status
> +                * during early boot intead of just before kexec to avoid SMP
> +                * races during kvm_pv_guest_cpu_reboot().
> +                * NOTE: We cannot reset the complete shared pages list
> +                * here as we need to retain the UEFI/OVMF firmware
> +                * specific settings.
> +                */
> +
> +               for (i = 0; i < e820_table->nr_entries; i++) {
> +                       struct e820_entry *entry = &e820_table->entries[i];
> +
> +                       if (entry->type != E820_TYPE_RAM)
> +                               continue;
> +
> +                       nr_pages = DIV_ROUND_UP(entry->size, PAGE_SIZE);
> +
> +                       kvm_hypercall3(KVM_HC_MAP_GPA_RANGE, entry->addr,
> +                                      nr_pages,
> +                                      KVM_MAP_GPA_RANGE_ENCRYPTED | KVM_MAP_GPA_RANGE_PAGE_SZ_4K);
> +               }
> +
>                 /*
>                  * Ensure that _bss_decrypted section is marked as decrypted in the
>                  * shared pages list.
> --
> 2.17.1
>

Re-reading things, I've convinced myself that kexec is fine. I was
previously concerned with KEXEC_PRESERVE_CONTEXT. In particular, since
the guest does not re-call the encryption status hypercalls after it
jumps back, the host will be out of the loop, and continue believing
the guest is in the same state as it was while running the
intermediate kernel. As a result, it would have an inaccurate list of
which pages are shared/private after a kexecing and jumping back.

The bit that I neglected is that the new kernel (just like the
original kernel) will disable live migration before jumping back (if
it enabled live migration at all). And the original kernel will never
re-enable. This is sub-optimal, since a VM that previously supported
migration can reach a state where it will stop supporting live
migration. But that's unavoidable. Kexecing a kernel that does not
support live migration does the same thing.

This looks good to me.

Reviewed-by: Steve Rutherford<srutherford@xxxxxxxxxx>