Re: [PATCH V2] Fix unsynchronized access to sev members through svm_register_enc_region

[Sean Christopherson <seanjc@xxxxxxxxxx>]

On Wed, Jan 27, 2021, Peter Gonda wrote:
> Grab kvm->lock before pinning memory when registering an encrypted
> region; sev_pin_memory() relies on kvm->lock being held to ensure
> correctness when checking and updating the number of pinned pages.
> 
> Add a lockdep assertion to help prevent future regressions.
> 
> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> Cc: "H. Peter Anvin" <hpa@xxxxxxxxx>
> Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> Cc: Joerg Roedel <joro@xxxxxxxxxx>
> Cc: Tom Lendacky <thomas.lendacky@xxxxxxx>
> Cc: Brijesh Singh <brijesh.singh@xxxxxxx>
> Cc: Sean Christopherson <seanjc@xxxxxxxxxx>
> Cc: x86@xxxxxxxxxx
> Cc: kvm@xxxxxxxxxxxxxxx
> Cc: stable@xxxxxxxxxxxxxxx
> Cc: linux-kernel@xxxxxxxxxxxxxxx
> Fixes: 1e80fdc09d12 ("KVM: SVM: Pin guest memory when SEV is active")
> Signed-off-by: Peter Gonda <pgonda@xxxxxxxxxx>
> 
> V2
>  - Fix up patch description
>  - Correct file paths svm.c -> sev.c
>  - Add unlock of kvm->lock on sev_pin_memory error
> 
> V1
>  - https://lore.kernel.org/kvm/20210126185431.1824530-1-pgonda@xxxxxxxxxx/

Put version info, and anything else that shouldn't be in the final commit, below
the three dashes.  AFAIK that requires manually editing the patch file before
sending it.

> 
> ---

Version info goes here.

>  arch/x86/kvm/svm/sev.c | 17 ++++++++++-------
>  1 file changed, 10 insertions(+), 7 deletions(-)
> 
> diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
> index c8ffdbc81709..b80e9bf0a31b 100644
> --- a/arch/x86/kvm/svm/sev.c
> +++ b/arch/x86/kvm/svm/sev.c
> @@ -342,6 +342,8 @@ static struct page **sev_pin_memory(struct kvm *kvm, unsigned long uaddr,
>  	unsigned long first, last;
>  	int ret;
>  
> +	lockdep_assert_held(&kvm->lock);
> +
>  	if (ulen == 0 || uaddr + ulen < uaddr)
>  		return ERR_PTR(-EINVAL);
>  
> @@ -1119,12 +1121,20 @@ int svm_register_enc_region(struct kvm *kvm,
>  	if (!region)
>  		return -ENOMEM;
>  
> +	mutex_lock(&kvm->lock);
>  	region->pages = sev_pin_memory(kvm, range->addr, range->size, &region->npages, 1);
>  	if (IS_ERR(region->pages)) {
>  		ret = PTR_ERR(region->pages);
> +		mutex_unlock(&kvm->lock);
>  		goto e_free;
>  	}
>  
> +	region->uaddr = range->addr;
> +	region->size = range->size;
> +
> +	list_add_tail(&region->list, &sev->regions_list);
> +	mutex_unlock(&kvm->lock);
> +
>  	/*
>  	 * 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
> @@ -1133,13 +1143,6 @@ int svm_register_enc_region(struct kvm *kvm,
>  	 */
>  	sev_clflush_pages(region->pages, region->npages);

I don't think it actually matters, but it feels like the flush should be done
before adding the region to the list.  That would also make this sequence
consistent with the other flows.

Tom, any thoughts?

>  
> -	region->uaddr = range->addr;
> -	region->size = range->size;
> -
> -	mutex_lock(&kvm->lock);
> -	list_add_tail(&region->list, &sev->regions_list);
> -	mutex_unlock(&kvm->lock);
> -
>  	return ret;
>  
>  e_free:
> -- 
> 2.30.0.280.ga3ce27912f-goog
>