| Message ID | 20210112044508.427338-14-david@gibson.dropbear.id.au (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Generalize memory encryption models | expand |
On 12.01.21 05:45, David Gibson wrote: > At least some s390 cpu models support "Protected Virtualization" (PV), > a mechanism to protect guests from eavesdropping by a compromised > hypervisor. > > This is similar in function to other mechanisms like AMD's SEV and > POWER's PEF, which are controlled by the "confidential-guest-support" > machine option. s390 is a slightly special case, because we already > supported PV, simply by using a CPU model with the required feature > (S390_FEAT_UNPACK). > > To integrate this with the option used by other platforms, we > implement the following compromise: > > - When the confidential-guest-support option is set, s390 will > recognize it, verify that the CPU can support PV (failing if not) > and set virtio default options necessary for encrypted or protected > guests, as on other platforms. i.e. if confidential-guest-support > is set, we will either create a guest capable of entering PV mode, > or fail outright. > > - If confidential-guest-support is not set, guests might still be > able to enter PV mode, if the CPU has the right model. This may be > a little surprising, but shouldn't actually be harmful. > > To start a guest supporting Protected Virtualization using the new > option use the command line arguments: > -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 This results in [cborntra@t35lp61 qemu]$ qemu-system-s390x -enable-kvm -nographic -m 2G -kernel ~/full.normal ** ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) Bail out! ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) Aborted (core dumped) > > Signed-off-by: David Gibson <david@gibson.dropbear.id.au> > --- > docs/confidential-guest-support.txt | 3 ++ > docs/system/s390x/protvirt.rst | 19 +++++++--- > hw/s390x/pv.c | 58 +++++++++++++++++++++++++++++ > include/hw/s390x/pv.h | 1 + > target/s390x/kvm.c | 3 ++ > 5 files changed, 78 insertions(+), 6 deletions(-) > > diff --git a/docs/confidential-guest-support.txt b/docs/confidential-guest-support.txt > index d466aa79d5..b4912f66c2 100644 > --- a/docs/confidential-guest-support.txt > +++ b/docs/confidential-guest-support.txt > @@ -42,4 +42,7 @@ AMD Secure Encrypted Virtualization (SEV) > > POWER Protected Execution Facility (PEF) > > +s390x Protected Virtualization (PV) > + docs/system/s390x/protvirt.rst > + > Other mechanisms may be supported in future. > diff --git a/docs/system/s390x/protvirt.rst b/docs/system/s390x/protvirt.rst > index 712974ad87..0f481043d9 100644 > --- a/docs/system/s390x/protvirt.rst > +++ b/docs/system/s390x/protvirt.rst > @@ -22,15 +22,22 @@ If those requirements are met, the capability `KVM_CAP_S390_PROTECTED` > will indicate that KVM can support PVMs on that LPAR. > > > -QEMU Settings > -------------- > +Running a Protected Virtual Machine > +----------------------------------- > > -To indicate to the VM that it can transition into protected mode, the > +To run a PVM you will need to select a CPU model which includes the > `Unpack facility` (stfle bit 161 represented by the feature > -`unpack`/`S390_FEAT_UNPACK`) needs to be part of the cpu model of > -the VM. > +`unpack`/`S390_FEAT_UNPACK`), and add these options to the command line:: > + > + -object s390-pv-guest,id=pv0 \ > + -machine confidential-guest-support=pv0 > + > +Adding these options will: > + > +* Ensure the `unpack` facility is available > +* Enable the IOMMU by default for all I/O devices > +* Initialize the PV mechanism > > -All I/O devices need to use the IOMMU. > Passthrough (vfio) devices are currently not supported. > > Host huge page backings are not supported. However guests can use huge > diff --git a/hw/s390x/pv.c b/hw/s390x/pv.c > index ab3a2482aa..85592e100a 100644 > --- a/hw/s390x/pv.c > +++ b/hw/s390x/pv.c > @@ -14,8 +14,11 @@ > #include <linux/kvm.h> > > #include "cpu.h" > +#include "qapi/error.h" > #include "qemu/error-report.h" > #include "sysemu/kvm.h" > +#include "qom/object_interfaces.h" > +#include "exec/confidential-guest-support.h" > #include "hw/s390x/ipl.h" > #include "hw/s390x/pv.h" > > @@ -111,3 +114,58 @@ void s390_pv_inject_reset_error(CPUState *cs) > /* Report that we are unable to enter protected mode */ > env->regs[r1 + 1] = DIAG_308_RC_INVAL_FOR_PV; > } > + > +#define TYPE_S390_PV_GUEST "s390-pv-guest" > +#define S390_PV_GUEST(obj) \ > + OBJECT_CHECK(S390PVGuestState, (obj), TYPE_S390_PV_GUEST) > + > +typedef struct S390PVGuestState S390PVGuestState; > + > +/** > + * S390PVGuestState: > + * > + * The S390PVGuestState object is basically a dummy used to tell the > + * confidential guest support system to use s390's PV mechanism. > + * > + * # $QEMU \ > + * -object s390-pv-guest,id=pv0 \ > + * -machine ...,confidential-guest-support=pv0 > + */ > +struct S390PVGuestState { > + Object parent_obj; > +}; > + > +int s390_pv_init(ConfidentialGuestSupport *cgs, Error **errp) > +{ > + if (!object_dynamic_cast(OBJECT(cgs), TYPE_S390_PV_GUEST)) { > + return 0; > + } > + > + if (!s390_has_feat(S390_FEAT_UNPACK)) { > + error_setg(errp, > + "CPU model does not support Protected Virtualization"); > + return -1; > + } > + > + cgs->ready = true; > + > + return 0; > +} > + > +static const TypeInfo s390_pv_guest_info = { > + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, > + .name = TYPE_S390_PV_GUEST, > + .instance_size = sizeof(S390PVGuestState), > + .interfaces = (InterfaceInfo[]) { > + { TYPE_USER_CREATABLE }, > + { } > + } > +}; > + > +static void > +s390_pv_register_types(void) > +{ > + type_register_static(&s390_pv_guest_info); > +} > + > +type_init(s390_pv_register_types); > diff --git a/include/hw/s390x/pv.h b/include/hw/s390x/pv.h > index aee758bc2d..9bbf66f356 100644 > --- a/include/hw/s390x/pv.h > +++ b/include/hw/s390x/pv.h > @@ -43,6 +43,7 @@ void s390_pv_prep_reset(void); > int s390_pv_verify(void); > void s390_pv_unshare(void); > void s390_pv_inject_reset_error(CPUState *cs); > +int s390_pv_init(ConfidentialGuestSupport *cgs, Error **errp); > #else /* CONFIG_KVM */ > static inline bool s390_is_pv(void) { return false; } > static inline int s390_pv_vm_enable(void) { return 0; } > diff --git a/target/s390x/kvm.c b/target/s390x/kvm.c > index b8385e6b95..d2435664dc 100644 > --- a/target/s390x/kvm.c > +++ b/target/s390x/kvm.c > @@ -387,6 +387,9 @@ int kvm_arch_init(MachineState *ms, KVMState *s) > } > > kvm_set_max_memslot_size(KVM_SLOT_MAX_BYTES); > + > + s390_pv_init(ms->cgs, &error_fatal); > + > return 0; > } > >
On Tue, Jan 12, 2021 at 03:45:08PM +1100, David Gibson wrote: > At least some s390 cpu models support "Protected Virtualization" (PV), > a mechanism to protect guests from eavesdropping by a compromised > hypervisor. > > This is similar in function to other mechanisms like AMD's SEV and > POWER's PEF, which are controlled by the "confidential-guest-support" > machine option. s390 is a slightly special case, because we already > supported PV, simply by using a CPU model with the required feature > (S390_FEAT_UNPACK). > > To integrate this with the option used by other platforms, we > implement the following compromise: > > - When the confidential-guest-support option is set, s390 will > recognize it, verify that the CPU can support PV (failing if not) > and set virtio default options necessary for encrypted or protected > guests, as on other platforms. i.e. if confidential-guest-support > is set, we will either create a guest capable of entering PV mode, > or fail outright. > > - If confidential-guest-support is not set, guests might still be > able to enter PV mode, if the CPU has the right model. This may be > a little surprising, but shouldn't actually be harmful. > > To start a guest supporting Protected Virtualization using the new > option use the command line arguments: > -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 > > Signed-off-by: David Gibson <david@gibson.dropbear.id.au> > --- > docs/confidential-guest-support.txt | 3 ++ > docs/system/s390x/protvirt.rst | 19 +++++++--- > hw/s390x/pv.c | 58 +++++++++++++++++++++++++++++ > include/hw/s390x/pv.h | 1 + > target/s390x/kvm.c | 3 ++ > 5 files changed, 78 insertions(+), 6 deletions(-) > diff --git a/hw/s390x/pv.c b/hw/s390x/pv.c > index ab3a2482aa..85592e100a 100644 > --- a/hw/s390x/pv.c > +++ b/hw/s390x/pv.c > @@ -14,8 +14,11 @@ > #include <linux/kvm.h> > > #include "cpu.h" > +#include "qapi/error.h" > #include "qemu/error-report.h" > #include "sysemu/kvm.h" > +#include "qom/object_interfaces.h" > +#include "exec/confidential-guest-support.h" > #include "hw/s390x/ipl.h" > #include "hw/s390x/pv.h" > > @@ -111,3 +114,58 @@ void s390_pv_inject_reset_error(CPUState *cs) > /* Report that we are unable to enter protected mode */ > env->regs[r1 + 1] = DIAG_308_RC_INVAL_FOR_PV; > } > + > +#define TYPE_S390_PV_GUEST "s390-pv-guest" > +#define S390_PV_GUEST(obj) \ > + OBJECT_CHECK(S390PVGuestState, (obj), TYPE_S390_PV_GUEST) > + > +typedef struct S390PVGuestState S390PVGuestState; Can use OBJECT_DECLARE_TYPE for this. Also same comment about struct naming vs type naming. > + > +/** > + * S390PVGuestState: > + * > + * The S390PVGuestState object is basically a dummy used to tell the > + * confidential guest support system to use s390's PV mechanism. > + * > + * # $QEMU \ > + * -object s390-pv-guest,id=pv0 \ > + * -machine ...,confidential-guest-support=pv0 > + */ > +struct S390PVGuestState { > + Object parent_obj; > +}; > + > +int s390_pv_init(ConfidentialGuestSupport *cgs, Error **errp) > +{ > + if (!object_dynamic_cast(OBJECT(cgs), TYPE_S390_PV_GUEST)) { > + return 0; > + } > + > + if (!s390_has_feat(S390_FEAT_UNPACK)) { > + error_setg(errp, > + "CPU model does not support Protected Virtualization"); > + return -1; > + } > + > + cgs->ready = true; > + > + return 0; > +} > + > +static const TypeInfo s390_pv_guest_info = { > + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, > + .name = TYPE_S390_PV_GUEST, > + .instance_size = sizeof(S390PVGuestState), > + .interfaces = (InterfaceInfo[]) { > + { TYPE_USER_CREATABLE }, > + { } > + } > +}; > + > +static void > +s390_pv_register_types(void) > +{ > + type_register_static(&s390_pv_guest_info); > +} > + > +type_init(s390_pv_register_types); This can all use OBJECT_DEFINE_TYPE_WITH_INTERFACES Regards, Daniel
On Tue, 12 Jan 2021 09:15:26 +0100 Christian Borntraeger <borntraeger@de.ibm.com> wrote: > On 12.01.21 05:45, David Gibson wrote: > > At least some s390 cpu models support "Protected Virtualization" (PV), > > a mechanism to protect guests from eavesdropping by a compromised > > hypervisor. > > > > This is similar in function to other mechanisms like AMD's SEV and > > POWER's PEF, which are controlled by the "confidential-guest-support" > > machine option. s390 is a slightly special case, because we already > > supported PV, simply by using a CPU model with the required feature > > (S390_FEAT_UNPACK). > > > > To integrate this with the option used by other platforms, we > > implement the following compromise: > > > > - When the confidential-guest-support option is set, s390 will > > recognize it, verify that the CPU can support PV (failing if not) > > and set virtio default options necessary for encrypted or protected > > guests, as on other platforms. i.e. if confidential-guest-support > > is set, we will either create a guest capable of entering PV mode, > > or fail outright. > > > > - If confidential-guest-support is not set, guests might still be > > able to enter PV mode, if the CPU has the right model. This may be > > a little surprising, but shouldn't actually be harmful. > > > > To start a guest supporting Protected Virtualization using the new > > option use the command line arguments: > > -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 > > > This results in > > [cborntra@t35lp61 qemu]$ qemu-system-s390x -enable-kvm -nographic -m 2G -kernel ~/full.normal > ** > ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) > Bail out! ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) > Aborted (core dumped) > > > +static const TypeInfo s390_pv_guest_info = { > > + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, > > + .name = TYPE_S390_PV_GUEST, > > + .instance_size = sizeof(S390PVGuestState), > > + .interfaces = (InterfaceInfo[]) { > > + { TYPE_USER_CREATABLE }, > > + { } > > + } > > +}; I think this needs TYPE_OBJECT in .parent and TYPE_CONFIDENTIAL_GUEST_SUPPORT as an interface to fix the crash.
On 12.01.21 12:36, Cornelia Huck wrote: > On Tue, 12 Jan 2021 09:15:26 +0100 > Christian Borntraeger <borntraeger@de.ibm.com> wrote: > >> On 12.01.21 05:45, David Gibson wrote: >>> At least some s390 cpu models support "Protected Virtualization" (PV), >>> a mechanism to protect guests from eavesdropping by a compromised >>> hypervisor. >>> >>> This is similar in function to other mechanisms like AMD's SEV and >>> POWER's PEF, which are controlled by the "confidential-guest-support" >>> machine option. s390 is a slightly special case, because we already >>> supported PV, simply by using a CPU model with the required feature >>> (S390_FEAT_UNPACK). >>> >>> To integrate this with the option used by other platforms, we >>> implement the following compromise: >>> >>> - When the confidential-guest-support option is set, s390 will >>> recognize it, verify that the CPU can support PV (failing if not) >>> and set virtio default options necessary for encrypted or protected >>> guests, as on other platforms. i.e. if confidential-guest-support >>> is set, we will either create a guest capable of entering PV mode, >>> or fail outright. >>> >>> - If confidential-guest-support is not set, guests might still be >>> able to enter PV mode, if the CPU has the right model. This may be >>> a little surprising, but shouldn't actually be harmful. >>> >>> To start a guest supporting Protected Virtualization using the new >>> option use the command line arguments: >>> -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 >> >> >> This results in >> >> [cborntra@t35lp61 qemu]$ qemu-system-s390x -enable-kvm -nographic -m 2G -kernel ~/full.normal >> ** >> ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) >> Bail out! ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) >> Aborted (core dumped) >> > >>> +static const TypeInfo s390_pv_guest_info = { >>> + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, >>> + .name = TYPE_S390_PV_GUEST, >>> + .instance_size = sizeof(S390PVGuestState), >>> + .interfaces = (InterfaceInfo[]) { >>> + { TYPE_USER_CREATABLE }, >>> + { } >>> + } >>> +}; > > I think this needs TYPE_OBJECT in .parent and > TYPE_CONFIDENTIAL_GUEST_SUPPORT as an interface to fix the crash. It fixes the crash but it does not work: qemu-system-s390x -enable-kvm -nographic -m 2G -kernel ~/image -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 qemu-system-s390x: Invalid parameter type for 'confidential-guest-support', expected: confidential-guest-support
On Tue, Jan 12, 2021 at 12:36:07PM +0100, Cornelia Huck wrote: > On Tue, 12 Jan 2021 09:15:26 +0100 > Christian Borntraeger <borntraeger@de.ibm.com> wrote: > > > On 12.01.21 05:45, David Gibson wrote: > > > At least some s390 cpu models support "Protected Virtualization" (PV), > > > a mechanism to protect guests from eavesdropping by a compromised > > > hypervisor. > > > > > > This is similar in function to other mechanisms like AMD's SEV and > > > POWER's PEF, which are controlled by the "confidential-guest-support" > > > machine option. s390 is a slightly special case, because we already > > > supported PV, simply by using a CPU model with the required feature > > > (S390_FEAT_UNPACK). > > > > > > To integrate this with the option used by other platforms, we > > > implement the following compromise: > > > > > > - When the confidential-guest-support option is set, s390 will > > > recognize it, verify that the CPU can support PV (failing if not) > > > and set virtio default options necessary for encrypted or protected > > > guests, as on other platforms. i.e. if confidential-guest-support > > > is set, we will either create a guest capable of entering PV mode, > > > or fail outright. > > > > > > - If confidential-guest-support is not set, guests might still be > > > able to enter PV mode, if the CPU has the right model. This may be > > > a little surprising, but shouldn't actually be harmful. > > > > > > To start a guest supporting Protected Virtualization using the new > > > option use the command line arguments: > > > -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 > > > > > > This results in > > > > [cborntra@t35lp61 qemu]$ qemu-system-s390x -enable-kvm -nographic -m 2G -kernel ~/full.normal > > ** > > ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) > > Bail out! ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) > > Aborted (core dumped) > > > > > > +static const TypeInfo s390_pv_guest_info = { > > > + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, > > > + .name = TYPE_S390_PV_GUEST, > > > + .instance_size = sizeof(S390PVGuestState), > > > + .interfaces = (InterfaceInfo[]) { > > > + { TYPE_USER_CREATABLE }, > > > + { } > > > + } > > > +}; > > I think this needs TYPE_OBJECT in .parent and > TYPE_CONFIDENTIAL_GUEST_SUPPORT as an interface to fix the crash. Except TYPE_CONFIDENTIAL_GUEST_SUPPORT is declared as an object, not an interface. Clearly something in this series is wrong though, as the ppc impl uses it as if it were an interface. Regards, Daniel
On Tue, Jan 12, 2021 at 12:36:07PM +0100, Cornelia Huck wrote: 65;6201;1c> On Tue, 12 Jan 2021 09:15:26 +0100 > Christian Borntraeger <borntraeger@de.ibm.com> wrote: > > > On 12.01.21 05:45, David Gibson wrote: > > > At least some s390 cpu models support "Protected Virtualization" (PV), > > > a mechanism to protect guests from eavesdropping by a compromised > > > hypervisor. > > > > > > This is similar in function to other mechanisms like AMD's SEV and > > > POWER's PEF, which are controlled by the "confidential-guest-support" > > > machine option. s390 is a slightly special case, because we already > > > supported PV, simply by using a CPU model with the required feature > > > (S390_FEAT_UNPACK). > > > > > > To integrate this with the option used by other platforms, we > > > implement the following compromise: > > > > > > - When the confidential-guest-support option is set, s390 will > > > recognize it, verify that the CPU can support PV (failing if not) > > > and set virtio default options necessary for encrypted or protected > > > guests, as on other platforms. i.e. if confidential-guest-support > > > is set, we will either create a guest capable of entering PV mode, > > > or fail outright. > > > > > > - If confidential-guest-support is not set, guests might still be > > > able to enter PV mode, if the CPU has the right model. This may be > > > a little surprising, but shouldn't actually be harmful. > > > > > > To start a guest supporting Protected Virtualization using the new > > > option use the command line arguments: > > > -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 > > > > > > This results in > > > > [cborntra@t35lp61 qemu]$ qemu-system-s390x -enable-kvm -nographic -m 2G -kernel ~/full.normal > > ** > > ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) > > Bail out! ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) > > Aborted (core dumped) > > > > > > +static const TypeInfo s390_pv_guest_info = { > > > + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, > > > + .name = TYPE_S390_PV_GUEST, > > > + .instance_size = sizeof(S390PVGuestState), > > > + .interfaces = (InterfaceInfo[]) { > > > + { TYPE_USER_CREATABLE }, > > > + { } > > > + } > > > +}; > > I think this needs TYPE_OBJECT in .parent and > TYPE_CONFIDENTIAL_GUEST_SUPPORT as an interface to fix the crash. No, that was true of an earlier revision, but parent is correct in the current version.
On 13.01.21 01:57, David Gibson wrote: > On Tue, Jan 12, 2021 at 12:36:07PM +0100, Cornelia Huck wrote: > 65;6201;1c> On Tue, 12 Jan 2021 09:15:26 +0100 >> Christian Borntraeger <borntraeger@de.ibm.com> wrote: >> >>> On 12.01.21 05:45, David Gibson wrote: >>>> At least some s390 cpu models support "Protected Virtualization" (PV), >>>> a mechanism to protect guests from eavesdropping by a compromised >>>> hypervisor. >>>> >>>> This is similar in function to other mechanisms like AMD's SEV and >>>> POWER's PEF, which are controlled by the "confidential-guest-support" >>>> machine option. s390 is a slightly special case, because we already >>>> supported PV, simply by using a CPU model with the required feature >>>> (S390_FEAT_UNPACK). >>>> >>>> To integrate this with the option used by other platforms, we >>>> implement the following compromise: >>>> >>>> - When the confidential-guest-support option is set, s390 will >>>> recognize it, verify that the CPU can support PV (failing if not) >>>> and set virtio default options necessary for encrypted or protected >>>> guests, as on other platforms. i.e. if confidential-guest-support >>>> is set, we will either create a guest capable of entering PV mode, >>>> or fail outright. >>>> >>>> - If confidential-guest-support is not set, guests might still be >>>> able to enter PV mode, if the CPU has the right model. This may be >>>> a little surprising, but shouldn't actually be harmful. >>>> >>>> To start a guest supporting Protected Virtualization using the new >>>> option use the command line arguments: >>>> -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 >>> >>> >>> This results in >>> >>> [cborntra@t35lp61 qemu]$ qemu-system-s390x -enable-kvm -nographic -m 2G -kernel ~/full.normal >>> ** >>> ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) >>> Bail out! ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) >>> Aborted (core dumped) >>> >> >>>> +static const TypeInfo s390_pv_guest_info = { >>>> + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, >>>> + .name = TYPE_S390_PV_GUEST, >>>> + .instance_size = sizeof(S390PVGuestState), >>>> + .interfaces = (InterfaceInfo[]) { >>>> + { TYPE_USER_CREATABLE }, >>>> + { } >>>> + } >>>> +}; >> >> I think this needs TYPE_OBJECT in .parent and >> TYPE_CONFIDENTIAL_GUEST_SUPPORT as an interface to fix the crash. > > No, that was true of an earlier revision, but parent is correct in the > current version. right now parent is obviously wrong as it triggers the above warning (and all other variants in the previous patches also use TYPE_OBJECT). It is probably the right thing when you fix +struct S390PVGuestState { + Object parent_obj; +}; + and change Object to the proper type I guess.
On Wed, Jan 13, 2021 at 07:57:41AM +0100, Christian Borntraeger wrote: > > > On 13.01.21 01:57, David Gibson wrote: > > On Tue, Jan 12, 2021 at 12:36:07PM +0100, Cornelia Huck wrote: > > 65;6201;1c> On Tue, 12 Jan 2021 09:15:26 +0100 > >> Christian Borntraeger <borntraeger@de.ibm.com> wrote: > >> > >>> On 12.01.21 05:45, David Gibson wrote: > >>>> At least some s390 cpu models support "Protected Virtualization" (PV), > >>>> a mechanism to protect guests from eavesdropping by a compromised > >>>> hypervisor. > >>>> > >>>> This is similar in function to other mechanisms like AMD's SEV and > >>>> POWER's PEF, which are controlled by the "confidential-guest-support" > >>>> machine option. s390 is a slightly special case, because we already > >>>> supported PV, simply by using a CPU model with the required feature > >>>> (S390_FEAT_UNPACK). > >>>> > >>>> To integrate this with the option used by other platforms, we > >>>> implement the following compromise: > >>>> > >>>> - When the confidential-guest-support option is set, s390 will > >>>> recognize it, verify that the CPU can support PV (failing if not) > >>>> and set virtio default options necessary for encrypted or protected > >>>> guests, as on other platforms. i.e. if confidential-guest-support > >>>> is set, we will either create a guest capable of entering PV mode, > >>>> or fail outright. > >>>> > >>>> - If confidential-guest-support is not set, guests might still be > >>>> able to enter PV mode, if the CPU has the right model. This may be > >>>> a little surprising, but shouldn't actually be harmful. > >>>> > >>>> To start a guest supporting Protected Virtualization using the new > >>>> option use the command line arguments: > >>>> -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 > >>> > >>> > >>> This results in > >>> > >>> [cborntra@t35lp61 qemu]$ qemu-system-s390x -enable-kvm -nographic -m 2G -kernel ~/full.normal > >>> ** > >>> ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) > >>> Bail out! ERROR:../qom/object.c:317:type_initialize: assertion failed: (parent->instance_size <= ti->instance_size) > >>> Aborted (core dumped) > >>> > >> > >>>> +static const TypeInfo s390_pv_guest_info = { > >>>> + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, > >>>> + .name = TYPE_S390_PV_GUEST, > >>>> + .instance_size = sizeof(S390PVGuestState), > >>>> + .interfaces = (InterfaceInfo[]) { > >>>> + { TYPE_USER_CREATABLE }, > >>>> + { } > >>>> + } > >>>> +}; > >> > >> I think this needs TYPE_OBJECT in .parent and > >> TYPE_CONFIDENTIAL_GUEST_SUPPORT as an interface to fix the crash. > > > > No, that was true of an earlier revision, but parent is correct in the > > current version. > > right now parent is obviously wrong as it triggers the above warning (and all other > variants in the previous patches also use TYPE_OBJECT). It is probably the right > thing when you fix > > +struct S390PVGuestState { > + Object parent_obj; > +}; > + > > and change Object to the proper type I guess. Yes, I think so. In the next spin I've fixed the parent_obj field (as well as moving to OBJECT_DEFINE_TYPE()) and it's passing the gitlab CI, at least.
diff --git a/docs/confidential-guest-support.txt b/docs/confidential-guest-support.txt index d466aa79d5..b4912f66c2 100644 --- a/docs/confidential-guest-support.txt +++ b/docs/confidential-guest-support.txt @@ -42,4 +42,7 @@ AMD Secure Encrypted Virtualization (SEV) POWER Protected Execution Facility (PEF) +s390x Protected Virtualization (PV) + docs/system/s390x/protvirt.rst + Other mechanisms may be supported in future. diff --git a/docs/system/s390x/protvirt.rst b/docs/system/s390x/protvirt.rst index 712974ad87..0f481043d9 100644 --- a/docs/system/s390x/protvirt.rst +++ b/docs/system/s390x/protvirt.rst @@ -22,15 +22,22 @@ If those requirements are met, the capability `KVM_CAP_S390_PROTECTED` will indicate that KVM can support PVMs on that LPAR. -QEMU Settings -------------- +Running a Protected Virtual Machine +----------------------------------- -To indicate to the VM that it can transition into protected mode, the +To run a PVM you will need to select a CPU model which includes the `Unpack facility` (stfle bit 161 represented by the feature -`unpack`/`S390_FEAT_UNPACK`) needs to be part of the cpu model of -the VM. +`unpack`/`S390_FEAT_UNPACK`), and add these options to the command line:: + + -object s390-pv-guest,id=pv0 \ + -machine confidential-guest-support=pv0 + +Adding these options will: + +* Ensure the `unpack` facility is available +* Enable the IOMMU by default for all I/O devices +* Initialize the PV mechanism -All I/O devices need to use the IOMMU. Passthrough (vfio) devices are currently not supported. Host huge page backings are not supported. However guests can use huge diff --git a/hw/s390x/pv.c b/hw/s390x/pv.c index ab3a2482aa..85592e100a 100644 --- a/hw/s390x/pv.c +++ b/hw/s390x/pv.c @@ -14,8 +14,11 @@ #include <linux/kvm.h> #include "cpu.h" +#include "qapi/error.h" #include "qemu/error-report.h" #include "sysemu/kvm.h" +#include "qom/object_interfaces.h" +#include "exec/confidential-guest-support.h" #include "hw/s390x/ipl.h" #include "hw/s390x/pv.h" @@ -111,3 +114,58 @@ void s390_pv_inject_reset_error(CPUState *cs) /* Report that we are unable to enter protected mode */ env->regs[r1 + 1] = DIAG_308_RC_INVAL_FOR_PV; } + +#define TYPE_S390_PV_GUEST "s390-pv-guest" +#define S390_PV_GUEST(obj) \ + OBJECT_CHECK(S390PVGuestState, (obj), TYPE_S390_PV_GUEST) + +typedef struct S390PVGuestState S390PVGuestState; + +/** + * S390PVGuestState: + * + * The S390PVGuestState object is basically a dummy used to tell the + * confidential guest support system to use s390's PV mechanism. + * + * # $QEMU \ + * -object s390-pv-guest,id=pv0 \ + * -machine ...,confidential-guest-support=pv0 + */ +struct S390PVGuestState { + Object parent_obj; +}; + +int s390_pv_init(ConfidentialGuestSupport *cgs, Error **errp) +{ + if (!object_dynamic_cast(OBJECT(cgs), TYPE_S390_PV_GUEST)) { + return 0; + } + + if (!s390_has_feat(S390_FEAT_UNPACK)) { + error_setg(errp, + "CPU model does not support Protected Virtualization"); + return -1; + } + + cgs->ready = true; + + return 0; +} + +static const TypeInfo s390_pv_guest_info = { + .parent = TYPE_CONFIDENTIAL_GUEST_SUPPORT, + .name = TYPE_S390_PV_GUEST, + .instance_size = sizeof(S390PVGuestState), + .interfaces = (InterfaceInfo[]) { + { TYPE_USER_CREATABLE }, + { } + } +}; + +static void +s390_pv_register_types(void) +{ + type_register_static(&s390_pv_guest_info); +} + +type_init(s390_pv_register_types); diff --git a/include/hw/s390x/pv.h b/include/hw/s390x/pv.h index aee758bc2d..9bbf66f356 100644 --- a/include/hw/s390x/pv.h +++ b/include/hw/s390x/pv.h @@ -43,6 +43,7 @@ void s390_pv_prep_reset(void); int s390_pv_verify(void); void s390_pv_unshare(void); void s390_pv_inject_reset_error(CPUState *cs); +int s390_pv_init(ConfidentialGuestSupport *cgs, Error **errp); #else /* CONFIG_KVM */ static inline bool s390_is_pv(void) { return false; } static inline int s390_pv_vm_enable(void) { return 0; } diff --git a/target/s390x/kvm.c b/target/s390x/kvm.c index b8385e6b95..d2435664dc 100644 --- a/target/s390x/kvm.c +++ b/target/s390x/kvm.c @@ -387,6 +387,9 @@ int kvm_arch_init(MachineState *ms, KVMState *s) } kvm_set_max_memslot_size(KVM_SLOT_MAX_BYTES); + + s390_pv_init(ms->cgs, &error_fatal); + return 0; }
At least some s390 cpu models support "Protected Virtualization" (PV), a mechanism to protect guests from eavesdropping by a compromised hypervisor. This is similar in function to other mechanisms like AMD's SEV and POWER's PEF, which are controlled by the "confidential-guest-support" machine option. s390 is a slightly special case, because we already supported PV, simply by using a CPU model with the required feature (S390_FEAT_UNPACK). To integrate this with the option used by other platforms, we implement the following compromise: - When the confidential-guest-support option is set, s390 will recognize it, verify that the CPU can support PV (failing if not) and set virtio default options necessary for encrypted or protected guests, as on other platforms. i.e. if confidential-guest-support is set, we will either create a guest capable of entering PV mode, or fail outright. - If confidential-guest-support is not set, guests might still be able to enter PV mode, if the CPU has the right model. This may be a little surprising, but shouldn't actually be harmful. To start a guest supporting Protected Virtualization using the new option use the command line arguments: -object s390-pv-guest,id=pv0 -machine confidential-guest-support=pv0 Signed-off-by: David Gibson <david@gibson.dropbear.id.au> --- docs/confidential-guest-support.txt | 3 ++ docs/system/s390x/protvirt.rst | 19 +++++++--- hw/s390x/pv.c | 58 +++++++++++++++++++++++++++++ include/hw/s390x/pv.h | 1 + target/s390x/kvm.c | 3 ++ 5 files changed, 78 insertions(+), 6 deletions(-)