linux-kernel - Re: [PATCH 1/2] KVM: Block memslot updates across range_start() and range

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <YO3OomTEhGFo2yee@google.com>
Date:   Tue, 13 Jul 2021 17:34:26 +0000
From:   Sean Christopherson <seanjc@...gle.com>
To:     Paolo Bonzini <pbonzini@...hat.com>
Cc:     linux-kernel@...r.kernel.org, kvm@...r.kernel.org,
        bgardon@...gle.com
Subject: Re: [PATCH 1/2] KVM: Block memslot updates across range_start() and
 range_end()

On Thu, Jun 10, 2021, Paolo Bonzini wrote:
>  static inline struct kvm_memslots *kvm_memslots(struct kvm *kvm)
> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> index fa7e7ebefc79..0dc0726c8d18 100644
> --- a/virt/kvm/kvm_main.c
> +++ b/virt/kvm/kvm_main.c
> @@ -605,10 +605,13 @@ static void kvm_mmu_notifier_change_pte(struct mmu_notifier *mn,
>  
>  	/*
>  	 * .change_pte() must be surrounded by .invalidate_range_{start,end}(),
> -	 * and so always runs with an elevated notifier count.  This obviates
> -	 * the need to bump the sequence count.
> +	 * If mmu_notifier_count is zero, then start() didn't find a relevant
> +	 * memslot and wasn't forced down the slow path; rechecking here is
> +	 * unnecessary.
>  	 */
> -	WARN_ON_ONCE(!kvm->mmu_notifier_count);
> +	WARN_ON_ONCE(!READ_ONCE(kvm->mn_active_invalidate_count));

The sanity check on mn_active_invalidate_count can be added in this patch, but
the optimization to return on !mmu_notifier_count should go in the next patch,
i.e. mmu_notifier_count must be non-zero since __kvm_handle_hva_range() always
takes mmu_lock at the time of this patch.

> +	if (!kvm->mmu_notifier_count)
> +		return;
>  
>  	kvm_handle_hva_range(mn, address, address + 1, pte, kvm_set_spte_gfn);
>  }

...

> @@ -1281,7 +1322,21 @@ static struct kvm_memslots *install_new_memslots(struct kvm *kvm,
>  	WARN_ON(gen & KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS);
>  	slots->generation = gen | KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS;
>  
> +	/*
> +	 * Do not store the new memslots while there are invalidations in
> +	 * progress (preparatory change for the next commit).
> +	 */
> +	spin_lock(&kvm->mn_invalidate_lock);
> +	prepare_to_rcuwait(&kvm->mn_memslots_update_rcuwait);
> +	while (kvm->mn_active_invalidate_count) {

Does this need a READ_ONCE()?  Or are the spin locks guaranteed to prevent the
compiler from caching mn_active_invalidate_count?

> +		set_current_state(TASK_UNINTERRUPTIBLE);
> +		spin_unlock(&kvm->mn_invalidate_lock);
> +		schedule();
> +		spin_lock(&kvm->mn_invalidate_lock);
> +	}
> +	finish_rcuwait(&kvm->mn_memslots_update_rcuwait);
>  	rcu_assign_pointer(kvm->memslots[as_id], slots);
> +	spin_unlock(&kvm->mn_invalidate_lock);
>  
>  	/*
>  	 * Acquired in kvm_set_memslot. Must be released before synchronize
> -- 
> 2.27.0
> 
>