Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
To:     Claudio Imbrenda <imbrenda@linux.ibm.com>, kvm@vger.kernel.org
Cc:     cohuck@redhat.com, borntraeger@de.ibm.com, thuth@redhat.com,
        pasic@linux.ibm.com, david@redhat.com, linux-s390@vger.kernel.org,
        linux-kernel@vger.kernel.org, Ulrich.Weigand@de.ibm.com
References: <20210818132620.46770-1-imbrenda@linux.ibm.com>
 <20210818132620.46770-3-imbrenda@linux.ibm.com>
From:   Janosch Frank <frankja@linux.ibm.com>
Subject: Re: [PATCH v4 02/14] KVM: s390: pv: avoid double free of sida page
Message-ID: <857732a7-906b-8275-a6fb-d2512f50c280@linux.ibm.com>
Date:   Tue, 31 Aug 2021 15:59:05 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.11.0
MIME-Version: 1.0
In-Reply-To: <20210818132620.46770-3-imbrenda@linux.ibm.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 8bit
Precedence: bulk

On 8/18/21 3:26 PM, Claudio Imbrenda wrote:
> If kvm_s390_pv_destroy_cpu is called more than once, we risk calling
> free_page on a random page, since the sidad field is aliased with the
> gbea, which is not guaranteed to be zero.
> 
> The solution is to simply return successfully immediately if the vCPU
> was already non secure.

I have the feeling this has been completely inconsistent from the start.

If we can't destroy a cpu we also won't take the VM out of PV state
because that's only allowed if we have removed all PV CPUs. That means
KVM thinks the VM is in PV mode but most of the CPUs are not.

Granted if the destroy CPU fails it makes more sense to just kill the VM
but looking at QEMU we don't even check the return value of that call.

> 
> Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
> Fixes: 19e1227768863a1469797c13ef8fea1af7beac2c ("KVM: S390: protvirt: Introduce instruction data area bounce buffer")
> ---
>  arch/s390/kvm/pv.c | 19 +++++++++----------
>  1 file changed, 9 insertions(+), 10 deletions(-)
> 
> diff --git a/arch/s390/kvm/pv.c b/arch/s390/kvm/pv.c
> index c8841f476e91..0a854115100b 100644
> --- a/arch/s390/kvm/pv.c
> +++ b/arch/s390/kvm/pv.c
> @@ -16,18 +16,17 @@
>  
>  int kvm_s390_pv_destroy_cpu(struct kvm_vcpu *vcpu, u16 *rc, u16 *rrc)
>  {
> -	int cc = 0;
> +	int cc;
>  
> -	if (kvm_s390_pv_cpu_get_handle(vcpu)) {
> -		cc = uv_cmd_nodata(kvm_s390_pv_cpu_get_handle(vcpu),
> -				   UVC_CMD_DESTROY_SEC_CPU, rc, rrc);
> +	if (!kvm_s390_pv_cpu_get_handle(vcpu))
> +		return 0;
> +
> +	cc = uv_cmd_nodata(kvm_s390_pv_cpu_get_handle(vcpu), UVC_CMD_DESTROY_SEC_CPU, rc, rrc);
> +
> +	KVM_UV_EVENT(vcpu->kvm, 3, "PROTVIRT DESTROY VCPU %d: rc %x rrc %x",
> +		     vcpu->vcpu_id, *rc, *rrc);
> +	WARN_ONCE(cc, "protvirt destroy cpu failed rc %x rrc %x", *rc, *rrc);
>  
> -		KVM_UV_EVENT(vcpu->kvm, 3,
> -			     "PROTVIRT DESTROY VCPU %d: rc %x rrc %x",
> -			     vcpu->vcpu_id, *rc, *rrc);
> -		WARN_ONCE(cc, "protvirt destroy cpu failed rc %x rrc %x",
> -			  *rc, *rrc);
> -	}
>  	/* Intended memory leak for something that should never happen. */
>  	if (!cc)
>  		free_pages(vcpu->arch.pv.stor_base,
>