Re: [Part2 PATCH v7 35/38] KVM: SVM: Pin guest memory when SEV is active

[Borislav Petkov]

On Wed, Nov 01, 2017 at 04:17:20PM -0500, Brijesh Singh wrote:
> The SEV memory encryption engine uses a tweak such that two identical
> plaintext pages at different location will have different ciphertext.
> So swapping or moving ciphertext of two pages will not result in
> plaintext being swapped. Relocating (or migrating) physical backing
> pages for a SEV guest will require some additional steps. The current SEV
> key management spec does not provide commands to swap or migrate (move)
> ciphertext pages. For now, we pin the guest memory registered through
> KVM_MEMORY_ENCRYPT_REGISTER_REGION ioctl.

...

> +static int svm_register_enc_region(struct kvm *kvm,
> +				   struct kvm_enc_region *range)
> +{
> +	struct kvm_sev_info *sev = &kvm->arch.sev_info;
> +	struct enc_region *region;
> +	int ret = 0;
> +
> +	if (!sev_guest(kvm))
> +		return -ENOTTY;
> +
> +	region = kzalloc(sizeof(*region), GFP_KERNEL);
> +	if (!region)
> +		return -ENOMEM;
> +
> +	region->pages = sev_pin_memory(kvm, range->addr, range->size, &region->npages, 1);
> +	if (!region->pages) {
> +		ret = -ENOMEM;
> +		goto e_free;
> +	}
> +
> +	/*
> +	 * The guest may change the memory encryption attribute from C=0 -> C=1
> +	 * or vice versa for this memory range. Lets make sure caches are
> +	 * flushed to ensure that guest data gets written into memory with
> +	 * correct C-bit.
> +	 */
> +	sev_clflush_pages(region->pages, region->npages);
> +
> +	region->uaddr = range->addr;
> +	region->size = range->size;
> +	list_add_tail(&region->list, &sev->regions_list);
> +	return ret;

Nothing's protecting that list from concurrent modifications of adding
and removal of regions.

-- 
Regards/Gruss,
    Boris.

Good mailing practices for 400: avoid top-posting and trim the reply.