[PATCH 5/6] tools: secure guest check for AMD in virt-host-validate

Boris Fiuczynski posted 6 patches 5 years, 9 months ago
There is a newer version of this series
[PATCH 5/6] tools: secure guest check for AMD in virt-host-validate
Posted by Boris Fiuczynski 5 years, 9 months ago
Add checking in virt-host-validate for secure guest support
on x86 for AMD Secure Encrypted Virtualization.

Signed-off-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
Reviewed-by: Paulo de Rezende Pinatti <ppinatti@linux.ibm.com>
Reviewed-by: Bjoern Walk <bwalk@linux.ibm.com>
---
 docs/kbase/launch_security_sev.rst |  7 ++++--
 tools/virt-host-validate-common.c  | 36 ++++++++++++++++++++++++++++--
 tools/virt-host-validate-common.h  |  1 +
 3 files changed, 40 insertions(+), 4 deletions(-)

diff --git a/docs/kbase/launch_security_sev.rst b/docs/kbase/launch_security_sev.rst
index fa602c7432..45166b3886 100644
--- a/docs/kbase/launch_security_sev.rst
+++ b/docs/kbase/launch_security_sev.rst
@@ -30,8 +30,11 @@ Enabling SEV on the host
 ========================
 
 Before VMs can make use of the SEV feature you need to make sure your
-AMD CPU does support SEV. You can check whether SEV is among the CPU
-flags with:
+AMD CPU does support SEV. You can run ``libvirt-host-validate``
+(libvirt >= 6.4.0) to check if your host supports secure guests or you
+can follow the manual checks below.
+
+You can manually check whether SEV is among the CPU flags with:
 
 ::
 
diff --git a/tools/virt-host-validate-common.c b/tools/virt-host-validate-common.c
index dd73bd0dea..67aa420d35 100644
--- a/tools/virt-host-validate-common.c
+++ b/tools/virt-host-validate-common.c
@@ -41,7 +41,8 @@ VIR_ENUM_IMPL(virHostValidateCPUFlag,
               "vmx",
               "svm",
               "sie",
-              "158");
+              "158",
+              "sev");
 
 static bool quiet;
 
@@ -447,15 +448,19 @@ int virHostValidateSecureGuests(const char *hvname,
                                 virHostValidateLevel level)
 {
     virBitmapPtr flags;
-    bool hasFac158 = false;
+    bool hasFac158 = false, hasAMDSev = false;
     virArch arch = virArchFromHost();
     g_autofree char *cmdline = NULL;
     static const char *kIBMValues[] = {"y", "Y", "1"};
+    static const char *kAMDValues[] = {"on"};
+    g_autofree char *mod_value = NULL;
 
     flags = virHostValidateGetCPUFlags();
 
     if (flags && virBitmapIsBitSet(flags, VIR_HOST_VALIDATE_CPU_FLAG_FACILITY_158))
         hasFac158 = true;
+    else if (flags && virBitmapIsBitSet(flags, VIR_HOST_VALIDATE_CPU_FLAG_SEV))
+        hasAMDSev = true;
 
     virBitmapFree(flags);
 
@@ -485,6 +490,33 @@ int virHostValidateSecureGuests(const char *hvname,
             virHostMsgFail(level, "Hardware or firmware does not provide "
                                   "support for IBM Secure Execution");
         }
+    } else if (hasAMDSev) {
+        if (virFileReadValueString(&mod_value, "/sys/module/kvm_amd/parameters/sev") < 0) {
+            virHostMsgFail(level, "AMD Secure Encrypted Virtualization not "
+                                  "supported by the currently used kernel");
+            return 0;
+        }
+        if (mod_value[0] != '1') {
+            virHostMsgFail(level,
+                           "AMD Secure Encrypted Virtualization appears to be "
+                           "disabled in kernel. Add mem_encrypt=on "
+                           "kvm_amd.sev=1 to kernel cmdline arguments");
+            return 0;
+        }
+        if (virFileReadValueString(&cmdline, "/proc/cmdline") < 0)
+            return -1;
+        if (virKernelCmdlineMatchParam(cmdline, "mem_encrypt", kAMDValues,
+                                       G_N_ELEMENTS(kAMDValues),
+                                       VIR_KERNEL_CMDLINE_FLAGS_SEARCH_LAST |
+                                       VIR_KERNEL_CMDLINE_FLAGS_CMP_EQ)) {
+            virHostMsgPass();
+            return 1;
+        } else {
+            virHostMsgFail(level,
+                           "AMD Secure Encrypted Virtualization appears to be "
+                           "disabled in kernel. Add mem_encrypt=on "
+                           "kvm_amd.sev=1 to kernel cmdline arguments");
+        }
     } else {
         virHostMsgFail(level,
                        "Unknown if this platform has Secure Guest support");
diff --git a/tools/virt-host-validate-common.h b/tools/virt-host-validate-common.h
index 44b5544a12..3df5ea0c7e 100644
--- a/tools/virt-host-validate-common.h
+++ b/tools/virt-host-validate-common.h
@@ -38,6 +38,7 @@ typedef enum {
     VIR_HOST_VALIDATE_CPU_FLAG_SVM,
     VIR_HOST_VALIDATE_CPU_FLAG_SIE,
     VIR_HOST_VALIDATE_CPU_FLAG_FACILITY_158,
+    VIR_HOST_VALIDATE_CPU_FLAG_SEV,
 
     VIR_HOST_VALIDATE_CPU_FLAG_LAST,
 } virHostValidateCPUFlag;
-- 
2.25.1

Re: [PATCH 5/6] tools: secure guest check for AMD in virt-host-validate
Posted by Erik Skultety 5 years, 8 months ago
On Mon, May 11, 2020 at 06:42:00PM +0200, Boris Fiuczynski wrote:
> Add checking in virt-host-validate for secure guest support
> on x86 for AMD Secure Encrypted Virtualization.
> 
> Signed-off-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
> Reviewed-by: Paulo de Rezende Pinatti <ppinatti@linux.ibm.com>
> Reviewed-by: Bjoern Walk <bwalk@linux.ibm.com>
> ---
>  docs/kbase/launch_security_sev.rst |  7 ++++--
>  tools/virt-host-validate-common.c  | 36 ++++++++++++++++++++++++++++--
>  tools/virt-host-validate-common.h  |  1 +
>  3 files changed, 40 insertions(+), 4 deletions(-)
> 
> diff --git a/docs/kbase/launch_security_sev.rst b/docs/kbase/launch_security_sev.rst
> index fa602c7432..45166b3886 100644
> --- a/docs/kbase/launch_security_sev.rst
> +++ b/docs/kbase/launch_security_sev.rst
> @@ -30,8 +30,11 @@ Enabling SEV on the host
>  ========================
>  
>  Before VMs can make use of the SEV feature you need to make sure your
> -AMD CPU does support SEV. You can check whether SEV is among the CPU
> -flags with:
> +AMD CPU does support SEV. You can run ``libvirt-host-validate``
> +(libvirt >= 6.4.0) to check if your host supports secure guests or you
> +can follow the manual checks below.
> +
> +You can manually check whether SEV is among the CPU flags with:

^this change should go along the (<6.4.0) in one of the earlier patches into a
standalone patch.

Otherwise looking good.

-- 
Erik Skultety

Re: [PATCH 5/6] tools: secure guest check for AMD in virt-host-validate
Posted by Boris Fiuczynski 5 years, 8 months ago
On 5/18/20 3:01 PM, Erik Skultety wrote:
> On Mon, May 11, 2020 at 06:42:00PM +0200, Boris Fiuczynski wrote:
>> Add checking in virt-host-validate for secure guest support
>> on x86 for AMD Secure Encrypted Virtualization.
>>
>> Signed-off-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
>> Reviewed-by: Paulo de Rezende Pinatti <ppinatti@linux.ibm.com>
>> Reviewed-by: Bjoern Walk <bwalk@linux.ibm.com>
>> ---
>>   docs/kbase/launch_security_sev.rst |  7 ++++--
>>   tools/virt-host-validate-common.c  | 36 ++++++++++++++++++++++++++++--
>>   tools/virt-host-validate-common.h  |  1 +
>>   3 files changed, 40 insertions(+), 4 deletions(-)
>>
>> diff --git a/docs/kbase/launch_security_sev.rst b/docs/kbase/launch_security_sev.rst
>> index fa602c7432..45166b3886 100644
>> --- a/docs/kbase/launch_security_sev.rst
>> +++ b/docs/kbase/launch_security_sev.rst
>> @@ -30,8 +30,11 @@ Enabling SEV on the host
>>   ========================
>>   
>>   Before VMs can make use of the SEV feature you need to make sure your
>> -AMD CPU does support SEV. You can check whether SEV is among the CPU
>> -flags with:
>> +AMD CPU does support SEV. You can run ``libvirt-host-validate``
>> +(libvirt >= 6.4.0) to check if your host supports secure guests or you
>> +can follow the manual checks below.
>> +
>> +You can manually check whether SEV is among the CPU flags with:
> 
> ^this change should go along the (<6.4.0) in one of the earlier patches into a
> standalone patch.

Actually the earlier patches fix the stale cap cache and this update is 
because of a new support in libvirt-host-validate. I am not sure that we 
should tie these to into one patch.
I would prefer to keep the two doc changes separate and with the changes 
that caused the update.

> 
> Otherwise looking good.
> 
Thanks but the changes need also to be adjusted as discussed on patch 3.
I will do so in a followup version.


-- 
Mit freundlichen Grüßen/Kind regards
    Boris Fiuczynski

IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Gregor Pillen
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294


Re: [PATCH 5/6] tools: secure guest check for AMD in virt-host-validate
Posted by Erik Skultety 5 years, 8 months ago
...
> > > diff --git a/docs/kbase/launch_security_sev.rst b/docs/kbase/launch_security_sev.rst
> > > index fa602c7432..45166b3886 100644
> > > --- a/docs/kbase/launch_security_sev.rst
> > > +++ b/docs/kbase/launch_security_sev.rst
> > > @@ -30,8 +30,11 @@ Enabling SEV on the host
> > >   ========================
> > >   Before VMs can make use of the SEV feature you need to make sure your
> > > -AMD CPU does support SEV. You can check whether SEV is among the CPU
> > > -flags with:
> > > +AMD CPU does support SEV. You can run ``libvirt-host-validate``
> > > +(libvirt >= 6.4.0) to check if your host supports secure guests or you
> > > +can follow the manual checks below.
> > > +
> > > +You can manually check whether SEV is among the CPU flags with:
> > 
> > ^this change should go along the (<6.4.0) in one of the earlier patches into a
> > standalone patch.
> 
> Actually the earlier patches fix the stale cap cache and this update is
> because of a new support in libvirt-host-validate. I am not sure that we
> should tie these to into one patch.
> I would prefer to keep the two doc changes separate and with the changes
> that caused the update.

I won't argue against that logic. However, both patch 3 and this one update the
same knowledge article. What IMO matters here the most is that once all of the
changes you're introducing are applied as a unit, the article needs to
reflect both the changes. From that perspective, at least to me it makes total
sense to group the docs changes from both 3/6 and this patch to a single update
to the SEV article accordingly.

-- 
Erik Skultety

Re: [PATCH 5/6] tools: secure guest check for AMD in virt-host-validate
Posted by Boris Fiuczynski 5 years, 8 months ago
On 5/19/20 8:25 AM, Erik Skultety wrote:
> ...
>>>> diff --git a/docs/kbase/launch_security_sev.rst b/docs/kbase/launch_security_sev.rst
>>>> index fa602c7432..45166b3886 100644
>>>> --- a/docs/kbase/launch_security_sev.rst
>>>> +++ b/docs/kbase/launch_security_sev.rst
>>>> @@ -30,8 +30,11 @@ Enabling SEV on the host
>>>>    ========================
>>>>    Before VMs can make use of the SEV feature you need to make sure your
>>>> -AMD CPU does support SEV. You can check whether SEV is among the CPU
>>>> -flags with:
>>>> +AMD CPU does support SEV. You can run ``libvirt-host-validate``
>>>> +(libvirt >= 6.4.0) to check if your host supports secure guests or you
>>>> +can follow the manual checks below.
>>>> +
>>>> +You can manually check whether SEV is among the CPU flags with:
>>>
>>> ^this change should go along the (<6.4.0) in one of the earlier patches into a
>>> standalone patch.
>>
>> Actually the earlier patches fix the stale cap cache and this update is
>> because of a new support in libvirt-host-validate. I am not sure that we
>> should tie these to into one patch.
>> I would prefer to keep the two doc changes separate and with the changes
>> that caused the update.
> 
> I won't argue against that logic. However, both patch 3 and this one update the
> same knowledge article. What IMO matters here the most is that once all of the
> changes you're introducing are applied as a unit, the article needs to
> reflect both the changes. From that perspective, at least to me it makes total
> sense to group the docs changes from both 3/6 and this patch to a single update
> to the SEV article accordingly.
> 
Fine, I will sort out the changes and create a single AMD doc update patch.

-- 
Mit freundlichen Grüßen/Kind regards
    Boris Fiuczynski

IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Gregor Pillen
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294