Re: [Qemu-devel] [RFC PATCH v1 01/10] KVM: SVM: Add KVM_SEV SEND_START command

["Singh, Brijesh" <brijesh.singh@amd.com>]

On 4/26/19 9:10 AM, Borislav Petkov wrote:
> On Wed, Apr 24, 2019 at 04:09:59PM +0000, Singh, Brijesh wrote:
>> The command is used to create an outgoing SEV guest encryption context.
>>
>> Cc: Thomas Gleixner <tglx@linutronix.de>
>> Cc: Ingo Molnar <mingo@redhat.com>
>> Cc: "H. Peter Anvin" <hpa@zytor.com>
>> Cc: Paolo Bonzini <pbonzini@redhat.com>
>> Cc: "Radim Krčmář" <rkrcmar@redhat.com>
>> Cc: Joerg Roedel <joro@8bytes.org>
>> Cc: Borislav Petkov <bp@suse.de>
>> Cc: Tom Lendacky <thomas.lendacky@amd.com>
>> Cc: x86@kernel.org
>> Cc: kvm@vger.kernel.org
>> Cc: linux-kernel@vger.kernel.org
>> Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
>> ---
>>   .../virtual/kvm/amd-memory-encryption.rst     |  24 +++++
>>   arch/x86/kvm/svm.c                            | 101 ++++++++++++++++++
>>   include/uapi/linux/kvm.h                      |  12 +++
>>   3 files changed, 137 insertions(+)
>>
>> diff --git a/Documentation/virtual/kvm/amd-memory-encryption.rst b/Documentation/virtual/kvm/amd-memory-encryption.rst
>> index 659bbc093b52..340ac4f87321 100644
>> --- a/Documentation/virtual/kvm/amd-memory-encryption.rst
>> +++ b/Documentation/virtual/kvm/amd-memory-encryption.rst
>> @@ -238,6 +238,30 @@ Returns: 0 on success, -negative on error
>>                   __u32 trans_len;
>>           };
>>   
>> +10. KVM_SEV_SEND_START
>> +----------------------
>> +
>> +The KVM_SEV_SEND_START command can be used by the hypervisor to create an
>> +outgoing guest encryption context.
>> +
>> +Parameters (in): struct kvm_sev_send_start
>> +
>> +Returns: 0 on success, -negative on error
>> +
>> +::
>> +        struct kvm_sev_send_start {
>> +                __u32 policy;                 /* guest policy */
>> +
>> +                __u64 pdh_cert_uaddr;         /* platform Diffie-Hellman certificate */
>> +                __u32 pdh_cert_len;
>> +
>> +                __u64 plat_cert_uaddr;        /* platform certificate chain */
>> +                __u32 plat_cert_len;
>> +
>> +                __u64 amd_cert_uaddr;         /* AMD certificate */
>> +                __u32 amd_cert_len;
> 
>          __u64 session_uaddr;
>          __u32 session_len;
> 
> too, right?


Ah good catch, I will fix in next rev. thanks


> 
>> +        };
>> +
>>   References
>>   ==========
>>   
>> diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
>> index 406b558abfef..4c2a225ba546 100644
>> --- a/arch/x86/kvm/svm.c
>> +++ b/arch/x86/kvm/svm.c
>> @@ -6955,6 +6955,104 @@ static int sev_launch_secret(struct kvm *kvm, struct kvm_sev_cmd *argp)
>>   	return ret;
>>   }
>>   
>> +static int sev_send_start(struct kvm *kvm, struct kvm_sev_cmd *argp)
>> +{
>> +	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
>> +	void *amd_cert = NULL, *session_data = NULL;
>> +	void *pdh_cert = NULL, *plat_cert = NULL;
>> +	struct sev_data_send_start *data = NULL;
>> +	struct kvm_sev_send_start params;
>> +	int ret;
>> +
>> +	if (!sev_guest(kvm))
>> +		return -ENOTTY;
>> +
>> +	if (copy_from_user(&params, (void __user *)(uintptr_t)argp->data,
>> +				sizeof(struct kvm_sev_send_start)))
>> +		return -EFAULT;
>> +
>> +	data = kzalloc(sizeof(*data), GFP_KERNEL);
>> +	if (!data)
>> +		return -ENOMEM;
>> +
>> +	/* userspace wants to query the session length */
>> +	if (!params.session_len)
>> +		goto cmd;
>> +
>> +	if (!params.pdh_cert_uaddr || !params.pdh_cert_len ||
>> +	    !params.session_uaddr)
>> +		return -EINVAL;
>> +
>> +	/* copy the certificate blobs from userspace */
>> +	pdh_cert = psp_copy_user_blob(params.pdh_cert_uaddr, params.pdh_cert_len);
>> +	if (IS_ERR(pdh_cert)) {
>> +		ret = PTR_ERR(pdh_cert);
>> +		goto e_free;
>> +	}
>> +
>> +	data->pdh_cert_address = __psp_pa(pdh_cert);
>> +	data->pdh_cert_len = params.pdh_cert_len;
>> +
>> +	plat_cert = psp_copy_user_blob(params.plat_cert_uaddr, params.plat_cert_len);
>> +	if (IS_ERR(plat_cert)) {
>> +		ret = PTR_ERR(plat_cert);
>> +		goto e_free_pdh;
>> +	}
>> +
>> +	data->plat_cert_address = __psp_pa(plat_cert);
>> +	data->plat_cert_len = params.plat_cert_len;
>> +
>> +	amd_cert = psp_copy_user_blob(params.amd_cert_uaddr, params.amd_cert_len);
>> +	if (IS_ERR(amd_cert)) {
>> +		ret = PTR_ERR(amd_cert);
>> +		goto e_free_plat_cert;
>> +	}
>> +
>> +	data->amd_cert_address = __psp_pa(amd_cert);
>> +	data->amd_cert_len = params.amd_cert_len;
>> +
>> +	ret = -ENOMEM;
>> +	session_data = kmalloc(params.session_len, GFP_KERNEL);
> 
> If the user is supposed to query the session length first, you could
> save it in a global variable perhaps and use that value instead of
> trusting the user to give you the correct one in params.session_len for
> the allocation...
> 

Yes that's doable but I am afraid that caching the value may lead us to
wrong path and also divergence from the SEV API spec. The spec says the
returned length is a minimum length but its possible that caller can
give a bigger buffer and FW will still work with it.


>> +	if (!session_data)
>> +		goto e_free_amd_cert;
>> +
>> +	data->session_address = __psp_pa(session_data);
>> +	data->session_len = params.session_len;
>> +cmd:
>> +	data->handle = sev->handle;
>> +	ret = sev_issue_cmd(kvm, SEV_CMD_SEND_START, data, &argp->error);
>> +
>> +	/* if we queried the session length, FW responded with expected data */
> 
> <--- ... here you have the session length from the fw.
>