Re: [PATCH v1 1/1] KVM: s390: pv: fix asynchronous teardown for small VMs

[Christian Borntraeger <borntraeger@de.ibm.com>]

Am 20.04.23 um 18:01 schrieb Claudio Imbrenda:
> On machines without the Destroy Secure Configuration Fast UVC, the
> topmost level of page tables is set aside and freed asynchronously
> as last step of the asynchronous teardown.
> 
> Each gmap has a host_to_guest radix tree mapping host (userspace)
> addresses (with 1M granularity) to gmap segment table entries (pmds).
> 
> If a guest is smaller than 2GB, the topmost level of page tables is the
> segment table (i.e. there are only 2 levels). Replacing it means that
> the pointers in the host_to_guest mapping would become stale and cause
> all kinds of nasty issues.
> 
> This patch fixes the issue by synchronously destroying all guests with
> only 2 levels of page tables in kvm_s390_pv_set_aside. This will
> speed up the process and avoid the issue altogether.
> 
> Update s390_replace_asce so it refuses to replace segment type ASCEs.
> 
> Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
> Fixes: fb491d5500a7 ("KVM: s390: pv: asynchronous destroy for reboot")
> ---
>   arch/s390/kvm/pv.c  | 35 ++++++++++++++++++++---------------
>   arch/s390/mm/gmap.c |  7 +++++++
>   2 files changed, 27 insertions(+), 15 deletions(-)
> 
> diff --git a/arch/s390/kvm/pv.c b/arch/s390/kvm/pv.c
> index e032ebbf51b9..ceb8cb628d62 100644
> --- a/arch/s390/kvm/pv.c
> +++ b/arch/s390/kvm/pv.c
> @@ -39,6 +39,7 @@ struct pv_vm_to_be_destroyed {
>   	u64 handle;
>   	void *stor_var;
>   	unsigned long stor_base;
> +	bool small;
>   };
>   
>   static void kvm_s390_clear_pv_state(struct kvm *kvm)
> @@ -318,7 +319,11 @@ int kvm_s390_pv_set_aside(struct kvm *kvm, u16 *rc, u16 *rrc)
>   	if (!priv)
>   		return -ENOMEM;
>   
> -	if (is_destroy_fast_available()) {
> +	if ((kvm->arch.gmap->asce & _ASCE_TYPE_MASK) == _ASCE_TYPE_SEGMENT) {
> +		/* No need to do things asynchronously for VMs under 2GB */
> +		res = kvm_s390_pv_deinit_vm(kvm, rc, rrc);
> +		priv->small = true;
> +	} else if (is_destroy_fast_available()) {
>   		res = kvm_s390_pv_deinit_vm_fast(kvm, rc, rrc);
>   	} else {
>   		priv->stor_var = kvm->arch.pv.stor_var;
> @@ -335,7 +340,8 @@ int kvm_s390_pv_set_aside(struct kvm *kvm, u16 *rc, u16 *rrc)
>   		return res;
>   	}
>   
> -	kvm_s390_destroy_lower_2g(kvm);
> +	if (!priv->small)
> +		kvm_s390_destroy_lower_2g(kvm);
>   	kvm_s390_clear_pv_state(kvm);
>   	kvm->arch.pv.set_aside = priv;
>   
> @@ -418,7 +424,10 @@ int kvm_s390_pv_deinit_cleanup_all(struct kvm *kvm, u16 *rc, u16 *rrc)
>   
>   	/* If a previous protected VM was set aside, put it in the need_cleanup list */
>   	if (kvm->arch.pv.set_aside) {
> -		list_add(kvm->arch.pv.set_aside, &kvm->arch.pv.need_cleanup);
> +		if (((struct pv_vm_to_be_destroyed *)kvm->arch.pv.set_aside)->small)
why do we need a cast here?

> +			kfree(kvm->arch.pv.set_aside);
> +		else
> +			list_add(kvm->arch.pv.set_aside, &kvm->arch.pv.need_cleanup);
>   		kvm->arch.pv.set_aside = NULL;
>   	}
>   

With the comment added that Marc asked for

Acked-by: Christian Borntraeger <borntraeger@linux.ibm.com>