linux-kernel - Re: [PATCH v4] KVM: SEV: Mark nested locking of vcpu->lock

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 for Android: free password hash cracker in your pocket

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

Message-ID: <93a99e53-abf6-db1b-f610-e94cd8d93a5c@redhat.com>
Date:   Mon, 2 May 2022 19:45:30 +0200
From:   Paolo Bonzini <pbonzini@...hat.com>
To:     Peter Gonda <pgonda@...gle.com>, kvm@...r.kernel.org
Cc:     John Sperbeck <jsperbeck@...gle.com>,
        David Rientjes <rientjes@...gle.com>,
        Sean Christopherson <seanjc@...gle.com>,
        Hillf Danton <hdanton@...a.com>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v4] KVM: SEV: Mark nested locking of vcpu->lock

On 5/2/22 18:58, Peter Gonda wrote:
> svm_vm_migrate_from() uses sev_lock_vcpus_for_migration() to lock all
> source and target vcpu->locks. Unfortunately there is an 8 subclass
> limit, so a new subclass cannot be used for each vCPU. Instead maintain
> ownership of the first vcpu's mutex.dep_map using a role specific
> subclass: source vs target. Release the other vcpu's mutex.dep_maps.
> 
> Fixes: b56639318bb2b ("KVM: SEV: Add support for SEV intra host migration")
> Reported-by: John Sperbeck<jsperbeck@...gle.com>
> Suggested-by: David Rientjes <rientjes@...gle.com>
> Suggested-by: Sean Christopherson <seanjc@...gle.com>
> Suggested-by: Paolo Bonzini <pbonzini@...hat.com>
> Cc: Hillf Danton <hdanton@...a.com>
> Cc: kvm@...r.kernel.org
> Cc: linux-kernel@...r.kernel.org
> Signed-off-by: Peter Gonda <pgonda@...gle.com>

Looks good, thanks!

Paolo

> ---
> 
> V4
>   * Due to 8 subclass limit keep dep_map on only the first vcpu and
>     release the others.
> 
> V3
>   * Updated signature to enum to self-document argument.
>   * Updated comment as Seanjc@ suggested.
> 
> Tested by running sev_migrate_tests with lockdep enabled. Before we see
> a warning from sev_lock_vcpus_for_migration(). After we get no warnings.
> 
> ---
>   arch/x86/kvm/svm/sev.c | 46 ++++++++++++++++++++++++++++++++++++++----
>   1 file changed, 42 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
> index 75fa6dd268f0..0239def64eaa 100644
> --- a/arch/x86/kvm/svm/sev.c
> +++ b/arch/x86/kvm/svm/sev.c
> @@ -1591,24 +1591,55 @@ static void sev_unlock_two_vms(struct kvm *dst_kvm, struct kvm *src_kvm)
>   	atomic_set_release(&src_sev->migration_in_progress, 0);
>   }
>   
> +/*
> + * To suppress lockdep false positives, subclass all vCPU mutex locks by
> + * assigning even numbers to the source vCPUs and odd numbers to destination
> + * vCPUs based on the vCPU's index.
> + */
> +enum sev_migration_role {
> +	SEV_MIGRATION_SOURCE = 0,
> +	SEV_MIGRATION_TARGET,
> +	SEV_NR_MIGRATION_ROLES,
> +};
>   
> -static int sev_lock_vcpus_for_migration(struct kvm *kvm)
> +static int sev_lock_vcpus_for_migration(struct kvm *kvm,
> +					enum sev_migration_role role)
>   {
>   	struct kvm_vcpu *vcpu;
>   	unsigned long i, j;
> +	bool first = true;
>   
>   	kvm_for_each_vcpu(i, vcpu, kvm) {
> -		if (mutex_lock_killable(&vcpu->mutex))
> +		if (mutex_lock_killable_nested(&vcpu->mutex, role))
>   			goto out_unlock;
> +
> +		if (first) {
> +			/*
> +			 * Reset the role to one that avoids colliding with
> +			 * the role used for the first vcpu mutex.
> +			 */
> +			role = SEV_NR_MIGRATION_ROLES;
> +			first = false;
> +		} else {
> +			mutex_release(&vcpu->mutex.dep_map, _THIS_IP_);
> +		}
>   	}
>   
>   	return 0;
>   
>   out_unlock:
> +
> +	first = true;
>   	kvm_for_each_vcpu(j, vcpu, kvm) {
>   		if (i == j)
>   			break;
>   
> +		if (first)
> +			first = false;
> +		else
> +			mutex_acquire(&vcpu->mutex.dep_map, role, 0, _THIS_IP_);
> +
> +
>   		mutex_unlock(&vcpu->mutex);
>   	}
>   	return -EINTR;
> @@ -1618,8 +1649,15 @@ static void sev_unlock_vcpus_for_migration(struct kvm *kvm)
>   {
>   	struct kvm_vcpu *vcpu;
>   	unsigned long i;
> +	bool first = true;
>   
>   	kvm_for_each_vcpu(i, vcpu, kvm) {
> +		if (first)
> +			first = false;
> +		else
> +			mutex_acquire(&vcpu->mutex.dep_map,
> +				      SEV_NR_MIGRATION_ROLES, 0, _THIS_IP_);
> +
>   		mutex_unlock(&vcpu->mutex);
>   	}
>   }
> @@ -1745,10 +1783,10 @@ int sev_vm_move_enc_context_from(struct kvm *kvm, unsigned int source_fd)
>   		charged = true;
>   	}
>   
> -	ret = sev_lock_vcpus_for_migration(kvm);
> +	ret = sev_lock_vcpus_for_migration(kvm, SEV_MIGRATION_SOURCE);
>   	if (ret)
>   		goto out_dst_cgroup;
> -	ret = sev_lock_vcpus_for_migration(source_kvm);
> +	ret = sev_lock_vcpus_for_migration(source_kvm, SEV_MIGRATION_TARGET);
>   	if (ret)
>   		goto out_dst_vcpu;
>