TDX QEMU support
v2 cover TDX QEMU support
This is v2 series of adding TDX suppot in QEMU.
This patch series aims to enable TDX support to allow creating and booting a TD (TDX VM) with QEMU. It needs to work with corresponding KVM v15 patch [1]. TDX related documents can be found in [2].
This series is based on QEMU gmem implemntation, which is posted at [3]. And This series is also available in github:
https://github.com/intel/qemu-tdx/tree/tdx-qemu-upstream-v2
This version aims to update the TDX QEMU side to match with latest TDX KVM side implementation, which expose gmem for private memory. This version is not targeted as the final version because how to support KVM gmem in QEMU is not finalized yet. Though, any review comment is welcomed.
[1] KVM TDX basic feature support v15
https://lore.kernel.org/kvm/[email protected]/
[3] https://lore.kernel.org/all/[email protected]/
== Limitation and future work ==
- Readonly memslot
TDX only support readonly (write protection) memslot for shared memory, but not for private memory. For simplicity, just mark readonly memslot not supported entirely for TDX.
- CPU model
We cannot create a TD with arbitrary CPU model like what for non-TDX VMs, because only a subset of features can be configured for TD.
- It's recommended to use '-cpu host' to create TD;
- '+feature/-feature' might not work as expected;
future work: To introduce specific CPU model for TDs and enhance +/-features
for TDs.
- gdb suppport
gdb support to debug a TD of off-debug mode is future work.
== Change history ==
Changes from v1:
[v1] https://lore.kernel.org/qemu-devel/[email protected]/
- Switch to KVM gmem interface for private memory;
- Add TDVMCALL and its sub leaves support;
- mark LMCE as unsupported for TD VM;
- bing back the support of mrconfigid/mrowner/mrownerconfig;
- update documentation;
Changes from RFC v4:
[RFC v4] https://lore.kernel.org/qemu-devel/[email protected]/
- Add 3 more patches(9, 10, 11) to improve the tdx_get_supported_cpuid();
- make attributes of object tdx-guest not settable by user;
- improve get_tdx_capabilities() by using a known starting value and limiting the loop with a known size;
- clarify why isa.bios needs to be skipped;
- remove the MMIO hob setup since OVMF sets them up itself;
Changes from RFC v3:
[RFC v3] https://lore.kernel.org/qemu-devel/[email protected]/
- Load TDVF with -bios interface;
- Adapt to KVM API changes;
- KVM_TDX_CAPABILITIES changes back to KVM-scope;
- struct kvm_tdx_init_vm changes;
- Define TDX_SUPPORTED_KVM_FEATURES;
- Drop the patch of introducing property sept-ve-disable since it's not public yet;
- some misc cleanups
Changes from RFC v2:
[RFC v2] https://lore.kernel.org/qemu-devel/[email protected]/
- Get vm-type from confidential-guest-support object type;
- Drop machine_init_done_late_notifiers;
- Refactor tdx_ioctl implementation;
- re-use existing pflash interface to load TDVF (i.e., OVMF binaries);
- introduce new date structure to track memory type instead of changing e820 table;
- Force smm to off for TDX VM;
- Drop the patches that suppress level-trigger/SMI/INIT/SIPI since KVM will ingore them;
- Add documentation;
Changes from RFC v1:
[RFC v1] https://lore.kernel.org/qemu-devel/[email protected]/
- suppress level trigger/SMI/INIT/SIPI related to IOAPIC.
- add VM attribute sha384 to TD measurement.
- guest TSC Hz specification
Chao Peng (1):
i386/tdx: register TDVF as private memory
Chenyi Qiang (2):
i386/tdx: register the fd read callback with the main loop to read the
quote data
i386/tdx: setup a timer for the qio channel
Isaku Yamahata (14):
i386/tdx: Make sept_ve_disable set by default
qom: implement property helper for sha384
i386/tdx: Allows mrconfigid/mrowner/mrownerconfig for TDX_INIT_VM i386/tdx: Create kvm gmem for TD
kvm/tdx: Don't complain when converting vMMIO region to shared kvm/tdx: Ignore memory conversion to shared of unassigned region i386/tdvf: Introduce function to parse TDVF metadata i386/tdx: Add TDVF memory via KVM_TDX_INIT_MEM_REGION i386/tdx: handle TDG.VP.VMCALL<SetupEventNotifyInterrupt> i386/tdx: handle TDG.VP.VMCALL<GetQuote>
i386/tdx: handle TDG.VP.VMCALL<MapGPA> hypercall i386/tdx: Limit the range size for MapGPA
hw/i386: add option to forcibly report edge trigger in acpi tables i386/tdx: Don't synchronize guest tsc for TDs
Sean Christopherson (2):
i386/kvm: Move architectural CPUID leaf generation to separate helper i386/tdx: Don't get/put guest state for TDX VMs
Xiaoyao Li (39):
*** HACK *** linux-headers: Update headers to pull in TDX API changes
i386: Introduce tdx-guest object
target/i386: Parse TDX vm type
target/i386: Introduce kvm_confidential_guest_init() i386/tdx: Implement tdx_kvm_init() to initialize TDX VM context i386/tdx: Get tdx_capabilities via KVM_TDX_CAPABILITIES i386/tdx: Introduce is_tdx_vm() helper and cache tdx_guest object i386/tdx: Adjust the supported CPUID based on TDX restrictions i386/tdx: Update tdx_cpuid_lookup[].tdx_fixed0/1 by
tdx_caps.cpuid_config[]
i386/tdx: Integrate tdx_caps->xfam_fixed0/1 into tdx_cpuid_lookup i386/tdx: Integrate tdx_caps->attrs_fixed0/1 to tdx_cpuid_lookup
kvm: Introduce kvm_arch_pre_create_vcpu()
i386/tdx: Initialize TDX before creating TD vcpus i386/tdx: Add property sept-ve-disable for tdx-guest object i386/tdx: Wire CPU features up with attributes of TD guest i386/tdx: Validate TD attributes
i386/tdx: Implement user specified tsc frequency
i386/tdx: Set kvm_readonly_mem_enabled to false for TDX VM i386/tdx: Make memory type private by default
i386/tdx: Parse TDVF metadata for TDX VM
i386/tdx: Skip BIOS shadowing setup
i386/tdx: Don't initialize pc.rom for TDX VMs
i386/tdx: Track mem_ptr for each firmware entry of TDVF i386/tdx: Track RAM entries for TDX VM
headers: Add definitions from UEFI spec for volumes, resources, etc... i386/tdx: Setup the TD HOB list
memory: Introduce memory_region_init_ram_gmem()
i386/tdx: Call KVM_TDX_INIT_VCPU to initialize TDX vcpu i386/tdx: Finalize TDX VM
i386/tdx: Handle TDG.VP.VMCALL<REPORT_FATAL_ERROR> i386/tdx: Wire REPORT_FATAL_ERROR with GuestPanic facility i386/tdx: Disable SMM for TDX VMs
i386/tdx: Disable PIC for TDX VMs
i386/tdx: Don't allow system reset for TDX VMs
i386/tdx: LMCE is not supported for TDX
hw/i386: add eoi_intercept_unsupported member to X86MachineState i386/tdx: Only configure MSR_IA32_UCODE_REV in kvm_init_msrs() for TDs i386/tdx: Skip kvm_put_apicbase() for TDs
docs: Add TDX documentation
accel/kvm/kvm-all.c | 55 +- configs/devices/i386-softmmu/default.mak | 1 + docs/system/confidential-guest-support.rst | 1 + docs/system/i386/tdx.rst | 114 ++ docs/system/target-i386.rst | 1 + hw/i386/Kconfig | 6 + hw/i386/acpi-build.c | 99 +- hw/i386/acpi-common.c | 50 +- hw/i386/meson.build | 1 + hw/i386/pc.c | 21 +- hw/i386/pc_sysfw.c | 7 + hw/i386/tdvf-hob.c | 147 ++ hw/i386/tdvf-hob.h | 24 + hw/i386/tdvf.c | 200 +++ hw/i386/x86.c | 38 +- include/exec/memory.h | 6 + include/hw/i386/tdvf.h | 58 + include/hw/i386/x86.h | 1 + include/qom/object.h | 17 + include/standard-headers/uefi/uefi.h | 198 +++ include/sysemu/kvm.h | 3 + linux-headers/asm-x86/kvm.h | 90 ++ linux-headers/linux/kvm.h | 87 ++ qapi/qom.json | 26 + qapi/run-state.json | 17 +- qom/object.c | 76 + softmmu/memory.c | 52 + softmmu/runstate.c | 49 + target/i386/cpu-internal.h | 9 + target/i386/cpu.c | 12 - target/i386/cpu.h | 21 + target/i386/kvm/kvm-cpu.c | 5 + target/i386/kvm/kvm.c | 586 ++++---- target/i386/kvm/kvm_i386.h | 5 + target/i386/kvm/meson.build | 2 + target/i386/kvm/tdx-stub.c | 22 + target/i386/kvm/tdx.c | 1543 ++++++++++++++++++++ target/i386/kvm/tdx.h | 73 + target/i386/sev.c | 1 - target/i386/sev.h | 2 + 40 files changed, 3382 insertions(+), 344 deletions(-) create mode 100644 docs/system/i386/tdx.rst create mode 100644 hw/i386/tdvf-hob.c create mode 100644 hw/i386/tdvf-hob.h create mode 100644 hw/i386/tdvf.c create mode 100644 include/hw/i386/tdvf.h create mode 100644 include/standard-headers/uefi/uefi.h create mode 100644 target/i386/kvm/tdx-stub.c create mode 100644 target/i386/kvm/tdx.c create mode 100644 target/i386/kvm/tdx.h
Signature · 1 line
2.34.1
PATCH v2 1/58 *** HACK *** linux-headers: Update headers to pull in TDX API changes
Pull in recent TDX updates, which are not backwards compatible.
It's just to make this series runnable. It will be updated by script
scripts/update-linux-headers.sh
once TDX support is upstreamed in linux kernel
--- linux-headers/asm-x86/kvm.h | 90 +++++++++++++++++++++++++++++++++++++ linux-headers/linux/kvm.h | 87 +++++++++++++++++++++++++++++++++++ 2 files changed, 177 insertions(+)
linux-headers/asm-x86/kvm.h +90 -0
@@ -562,5 +562,95 @@ struct kvm_pmu_event_filter {#define KVM_X86_DEFAULT_VM 0#define KVM_X86_SW_PROTECTED_VM 1 +#define KVM_X86_TDX_VM 2 +#define KVM_X86_SNP_VM 3 + +/* Trust Domain eXtension sub-ioctl() commands. */ +enum kvm_tdx_cmd_id { +KVM_TDX_CAPABILITIES =0 , +KVM_TDX_INIT_VM , +KVM_TDX_INIT_VCPU , +KVM_TDX_INIT_MEM_REGION , +KVM_TDX_FINALIZE_VM , + +KVM_TDX_CMD_NR_MAX , +}; + +struct kvm_tdx_cmd { +/* enum kvm_tdx_cmd_id */ +__u32 id; +/* flags for sub-commend. If sub-command doesn't use this, set zero. */ +__u32 flags; +/* +* data for each sub-command. An immediate or a pointer to the actual +* data in process virtual address. If sub-command doesn't use it, +* set zero. +*/ +__u64 data; +/* +* Auxiliary error code. The sub-command may return TDX SEAMCALL +* status code in addition to -Exxx. +* Defined for consistency with struct kvm_sev_cmd. +*/ +__u64 error; +}; + +struct kvm_tdx_cpuid_config { +__u32 leaf; +__u32 sub_leaf; +__u32 eax; +__u32 ebx; +__u32 ecx; +__u32 edx; +}; + +struct kvm_tdx_capabilities { +__u64 attrs_fixed0; +__u64 attrs_fixed1; +__u64 xfam_fixed0; +__u64 xfam_fixed1; +#define TDX_CAP_GPAW_48 (1 << 0) +#define TDX_CAP_GPAW_52 (1 << 1) +__u32 supported_gpaw; +__u32 padding; +__u64 reserved[251 ]; + +__u32 nr_cpuid_configs; +struct kvm_tdx_cpuid_config cpuid_configs[]; +}; + +struct kvm_tdx_init_vm { +__u64 attributes; +__u64 mrconfigid[6 ];/* sha384 digest */ +__u64 mrowner[6 ];/* sha384 digest */ +__u64 mrownerconfig[6 ];/* sha348 digest */ +/* +* For future extensibility to make sizeof(struct kvm_tdx_init_vm) = 8KB. +* This should be enough given sizeof(TD_PARAMS) = 1024. +* 8KB was chosen given because +* sizeof(struct kvm_cpuid_entry2) * KVM_MAX_CPUID_ENTRIES(=256) = 8KB. +*/ +__u64 reserved[1004 ]; + +/* +* Call KVM_TDX_INIT_VM before vcpu creation, thus before +* KVM_SET_CPUID2. +* This configuration supersedes KVM_SET_CPUID2s for VCPUs because the +* TDX module directly virtualizes those CPUIDs without VMM. The user +* space VMM, e.g. qemu, should make KVM_SET_CPUID2 consistent with +* those values. If it doesn't, KVM may have wrong idea of vCPUIDs of +* the guest, and KVM may wrongly emulate CPUIDs or MSRs that the TDX +* module doesn't virtualize. +*/ +struct kvm_cpuid2 cpuid; +}; + +#define KVM_TDX_MEASURE_MEMORY_REGION (1UL << 0) + +struct kvm_tdx_init_mem_region { +__u64 source_addr; +__u64 gpa; +__u64 nr_pages; +};#endif /* _ASM_X86_KVM_H */
linux-headers/linux/kvm.h +87 -0
105 lines not on this page · raw message
Signature · 1 line
2.34.1
PATCH v2 2/58 i386: Introduce tdx-guest object
In reply to Xiaoyao LiIntroduce tdx-guest object which implements the interface of CONFIDENTIAL_GUEST_SUPPORT, and will be used to create TDX VMs (TDs) by
qemu -machine ...,confidential-guest-support=tdx0 \
-object tdx-guset,id=tdx0It has only one property 'attributes' with fixed value 0 and not configurable so far.
---
changes from RFC-V4
- make @attributes not user-settable
--- configs/devices/i386-softmmu/default.mak | 1 + hw/i386/Kconfig | 5 +++ qapi/qom.json | 12 +++++++ target/i386/kvm/meson.build | 2 ++ target/i386/kvm/tdx.c | 40 ++++++++++++++++++++++++ target/i386/kvm/tdx.h | 19 +++++++++++ 6 files changed, 79 insertions(+) create mode 100644 target/i386/kvm/tdx.c create mode 100644 target/i386/kvm/tdx.h
configs/devices/i386-softmmu/default.mak +1 -0
@@ -18,6 +18,7 @@ #CONFIG_QXL=n #CONFIG_SEV=n #CONFIG_SGA=n +#CONFIG_TDX=n #CONFIG_TEST_DEVICES=n #CONFIG_TPM_CRB=n #CONFIG_TPM_TIS_ISA=n
hw/i386/Kconfig +5 -0
@@ -10,6 +10,10 @@ config SGX bool depends on KVM +config TDX + bool + depends on KVM + config PC bool imply APPLESMC
@@ -26,6 +30,7 @@ config PC imply QXL imply SEV imply SGX + imply TDX imply TEST_DEVICES imply TPM_CRB imply TPM_TIS_ISA
qapi/qom.json +12 -0
@@ -866,6 +866,16 @@ 'reduced-phys-bits': 'uint32', '*kernel-hashes': 'bool' } } +## +# @TdxGuestProperties: +# +# Properties for tdx-guest objects. +# +# Since: 8.2 +## +{ 'struct': 'TdxGuestProperties', + 'data': { }} + ## # @ThreadContextProperties: #
@@ -944,6 +954,7 @@ 'sev-guest', 'thread-context', 's390-pv-guest', + 'tdx-guest', 'throttle-group', 'tls-creds-anon', 'tls-creds-psk',
@@ -1010,6 +1021,7 @@ 'secret_keyring': { 'type': 'SecretKeyringProperties', 'if': 'CONFIG_SECRET_KEYRING' }, 'sev-guest': 'SevGuestProperties', + 'tdx-guest': 'TdxGuestProperties', 'thread-context': 'ThreadContextProperties', 'throttle-group': 'ThrottleGroupProperties', 'tls-creds-anon': 'TlsCredsAnonProperties',
target/i386/kvm/meson.build +2 -0
@@ -11,6 +11,8 @@ i386_softmmu_kvm_ss.add(when: 'CONFIG_XEN_EMU', if_true: files('xen-emu.c')) i386_softmmu_kvm_ss.add(when: 'CONFIG_SEV', if_false: files('sev-stub.c')) +i386_softmmu_kvm_ss.add(when: 'CONFIG_TDX', if_true: files('tdx.c')) + i386_system_ss.add(when: 'CONFIG_HYPERV', if_true: files('hyperv.c'), if_false: files('hyperv-stub.c')) i386_system_ss.add_all(when: 'CONFIG_KVM', if_true: i386_softmmu_kvm_ss)
target/i386/kvm/tdx.c new file +40 -0
@@ -0,0 +1,40 @@ +/* +* QEMU TDX support +* +* Copyright Intel +* +* Author: +* Xiaoyao Li <[email protected]> +* +* This work is licensed under the terms of the GNU GPL, version 2 or later. +* See the COPYING file in the top-level directory +* +*/ + +#include "qemu/osdep.h" +#include "qom/object_interfaces.h" + +#include "tdx.h" + +/* tdx guest */ +OBJECT_DEFINE_TYPE_WITH_INTERFACES (TdxGuest, + tdx_guest, +TDX_GUEST , +CONFIDENTIAL_GUEST_SUPPORT , + { TYPE_USER_CREATABLE }, + { NULL }) + +static void tdx_guest_init (Object *obj) +{ +TdxGuest *tdx =TDX_GUEST (obj); + + tdx->attributes =0 ; +} + +static void tdx_guest_finalize (Object *obj) +{ +} + +static void tdx_guest_class_init (ObjectClass *oc,void *data) +{ +}
target/i386/kvm/tdx.h new file +19 -0
@@ -0,0 +1,19 @@ +#ifndef QEMU_I386_TDX_H +#define QEMU_I386_TDX_H + +#include "exec/confidential-guest-support.h" + +#define TYPE_TDX_GUEST "tdx-guest" +#define TDX_GUEST (obj) OBJECT_CHECK(TdxGuest, (obj), TYPE_TDX_GUEST) + +typedef struct TdxGuestClass { +ConfidentialGuestSupportClass parent_class; +}TdxGuestClass ; + +typedef struct TdxGuest { +ConfidentialGuestSupport parent_obj; + +uint64_t attributes;/* TD attributes */ +}TdxGuest ; + +#endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
Re: [PATCH v2 02/58] i386: Introduce tdx-guest object
Quoted 5 lines · Xiaoyao Li
Introduce tdx-guest object which implements the interface of
CONFIDENTIAL_GUEST_SUPPORT, and will be used to create TDX VMs (TDs) byqemu -machine ...,confidential-guest-support=tdx0 \ -object tdx-guset,id=tdx0
Typo: tdx-guest
Quoted 2 lines
It has only one property 'attributes' with fixed value 0 and not configurable so far.
This must refer to TdxGuest member @attributes.
"Property" suggests QOM property, which @attributes isn't, at least not in this patch. Will it become a QOM property later in this series?
Hmm, @attributes appears to remain unused until PATCH 14. Recommend to delay its addition until then.
Quoted 81 lines
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>---
changes from RFC-V4
- make @attributes not user-settable--- configs/devices/i386-softmmu/default.mak | 1 + hw/i386/Kconfig | 5 +++ qapi/qom.json | 12 +++++++ target/i386/kvm/meson.build | 2 ++ target/i386/kvm/tdx.c | 40 ++++++++++++++++++++++++ target/i386/kvm/tdx.h | 19 +++++++++++ 6 files changed, 79 insertions(+) create mode 100644 target/i386/kvm/tdx.c create mode 100644 target/i386/kvm/tdx.hconfigs/devices/i386-softmmu/default.mak +1
-0@@ -18,6 +18,7 @@ #CONFIG_QXL=n #CONFIG_SEV=n #CONFIG_SGA=n +#CONFIG_TDX=n #CONFIG_TEST_DEVICES=n #CONFIG_TPM_CRB=n #CONFIG_TPM_TIS_ISA=nhw/i386/Kconfig +5
-0@@ -10,6 +10,10 @@ config SGX bool depends on KVM +config TDX + bool + depends on KVM + config PC bool imply APPLESMC@@ -26,6 +30,7 @@ config PC imply QXL imply SEV imply SGX + imply TDX imply TEST_DEVICES imply TPM_CRB imply TPM_TIS_ISAqapi/qom.json +11
-0@@ -866,6 +866,16 @@ 'reduced-phys-bits': 'uint32', '*kernel-hashes': 'bool' } } +## +# @TdxGuestProperties: +# +# Properties for tdx-guest objects. +# +# Since: 8.2 +## +{ 'struct': 'TdxGuestProperties', + 'data': { }} + ## # @ThreadContextProperties: #@@ -944,6 +954,7 @@ 'sev-guest', 'thread-context', 's390-pv-guest', + 'tdx-guest', 'throttle-group', 'tls-creds-anon', 'tls-creds-psk',
qapi/qom.json line 1027
Quoted 7 lines
qapi/qom.json +1
-0@@ -1010,6 +1021,7 @@ 'secret_keyring': { 'type': 'SecretKeyringProperties', 'if': 'CONFIG_SECRET_KEYRING' }, 'sev-guest': 'SevGuestProperties', + 'tdx-guest': 'TdxGuestProperties', 'thread-context': 'ThreadContextProperties', 'throttle-group': 'ThrottleGroupProperties', 'tls-creds-anon': 'TlsCredsAnonProperties',
Actually useful only when CONFIG_TDX is on, but can't make it conditional here, as CONFIG_TDX is poisoned.
Quoted 74 lines
target/i386/kvm/meson.build +2
-0@@ -11,6 +11,8 @@ i386_softmmu_kvm_ss.add(when: 'CONFIG_XEN_EMU', if_true: files('xen-emu.c')) i386_softmmu_kvm_ss.add(when: 'CONFIG_SEV', if_false: files('sev-stub.c')) +i386_softmmu_kvm_ss.add(when: 'CONFIG_TDX', if_true: files('tdx.c')) + i386_system_ss.add(when: 'CONFIG_HYPERV', if_true: files('hyperv.c'), if_false: files('hyperv-stub.c')) i386_system_ss.add_all(when: 'CONFIG_KVM', if_true: i386_softmmu_kvm_ss)target/i386/kvm/tdx.c new file +40
-0@@ -0,0 +1,40 @@ +/* +* QEMU TDX support +* +* Copyright Intel +* +* Author: +* Xiaoyao Li <[email protected]> +* +* This work is licensed under the terms of the GNU GPL, version 2 or later. +* See the COPYING file in the top-level directory +* +*/ + +#include "qemu/osdep.h" +#include "qom/object_interfaces.h" + +#include "tdx.h" + +/* tdx guest */ +OBJECT_DEFINE_TYPE_WITH_INTERFACES (TdxGuest, + tdx_guest, +TDX_GUEST , +CONFIDENTIAL_GUEST_SUPPORT , + { TYPE_USER_CREATABLE }, + { NULL }) + +static void tdx_guest_init (Object *obj) +{ +TdxGuest *tdx =TDX_GUEST (obj); + + tdx->attributes =0 ; +} + +static void tdx_guest_finalize (Object *obj) +{ +} + +static void tdx_guest_class_init (ObjectClass *oc,void *data) +{ +}target/i386/kvm/tdx.h new file +9
-0@@ -0,0 +1,19 @@ +#ifndef QEMU_I386_TDX_H +#define QEMU_I386_TDX_H + +#include "exec/confidential-guest-support.h" + +#define TYPE_TDX_GUEST "tdx-guest" +#define TDX_GUEST(obj) OBJECT_CHECK(TdxGuest, (obj), TYPE_TDX_GUEST) + +typedef struct TdxGuestClass {
target/i386/kvm/tdx.h line 19
Quoted 10 lines
target/i386/kvm/tdx.h new file +10
-0+ ConfidentialGuestSupportClass parent_class; +} TdxGuestClass; + +typedef struct TdxGuest { +ConfidentialGuestSupport parent_obj; + +uint64_t attributes;/* TD attributes */ +}TdxGuest ; + +#endif /* QEMU_I386_TDX_H */
QAPI schema
Re: [PATCH v2 02/58] i386: Introduce tdx-guest object
Quoted 9 lines · Markus Armbruster
Quoted 5 lines · Xiaoyao Li
Introduce tdx-guest object which implements the interface of
CONFIDENTIAL_GUEST_SUPPORT, and will be used to create TDX VMs (TDs) byqemu -machine ...,confidential-guest-support=tdx0 \ -object tdx-guset,id=tdx0Typo: tdx-guest
Will fix it.
Quoted 7 lines
Quoted 2 lines
It has only one property 'attributes' with fixed value 0 and not configurable so far.
This must refer to TdxGuest member @attributes.
"Property" suggests QOM property, which @attributes isn't, at least not in this patch. Will it become a QOM property later in this series?
At least not in this series. Maybe in the future there is request to directly configure the whole attributes via QOM property, but none from now.
I will change the description of it to avoid confusion.
Quoted 2 lines
Hmm, @attributes appears to remain unused until PATCH 14. Recommend to delay its addition until then.
IMHO, it's not suitable to introduce it in patch 14. Using a separate patch seems unnecessary. I'll leave it in this patch unless strong objection on it.
Quoted 83 lines
Quoted 83 lines
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>---
changes from RFC-V4
- make @attributes not user-settable--- configs/devices/i386-softmmu/default.mak | 1 + hw/i386/Kconfig | 5 +++ qapi/qom.json | 12 +++++++ target/i386/kvm/meson.build | 2 ++ target/i386/kvm/tdx.c | 40 ++++++++++++++++++++++++ target/i386/kvm/tdx.h | 19 +++++++++++ 6 files changed, 79 insertions(+)create mode 100644
target/i386/kvm/tdx.c
create mode 100644target/i386/kvm/tdx.hconfigs/devices/i386-softmmu/default.mak +1
-0@@ -18,6 +18,7 @@ #CONFIG_QXL=n #CONFIG_SEV=n #CONFIG_SGA=n +#CONFIG_TDX=n #CONFIG_TEST_DEVICES=n #CONFIG_TPM_CRB=n #CONFIG_TPM_TIS_ISA=nhw/i386/Kconfig +5
-0@@ -10,6 +10,10 @@ config SGX bool depends on KVM +config TDX + bool + depends on KVM + config PC bool imply APPLESMC@@ -26,6 +30,7 @@ config PC imply QXL imply SEV imply SGX + imply TDX imply TEST_DEVICES imply TPM_CRB imply TPM_TIS_ISAqapi/qom.json +11
-0@@ -866,6 +866,16 @@ 'reduced-phys-bits': 'uint32', '*kernel-hashes': 'bool' } } +## +# @TdxGuestProperties: +# +# Properties for tdx-guest objects. +# +# Since: 8.2 +## +{ 'struct': 'TdxGuestProperties', + 'data': { }} + ## # @ThreadContextProperties: #@@ -944,6 +954,7 @@ 'sev-guest', 'thread-context', 's390-pv-guest', + 'tdx-guest', 'throttle-group', 'tls-creds-anon', 'tls-creds-psk',@@ -1010,6 +1021,7 @@ 'secret_keyring': { 'type': 'SecretKeyringProperties', 'if': 'CONFIG_SECRET_KEYRING' },
Quoted 8 lines
qapi/qom.json line 1027
Quoted 5 lines
qapi/qom.json +1
-0'sev-guest': 'SevGuestProperties', + 'tdx-guest': 'TdxGuestProperties', 'thread-context': 'ThreadContextProperties', 'throttle-group': 'ThrottleGroupProperties', 'tls-creds-anon': 'TlsCredsAnonProperties',Actually useful only when
CONFIG_TDXis on, but can't make it conditional here, asCONFIG_TDXis poisoned.
In fact, I just followed what SEV did.
To me, it looks OK to make it conditional on CONFIG_TDX. Could you please elaborate "but can't make it conditional here, as CONFIG_TDX is poisoned." ?
Quoted 86 lines
Quoted 74 lines
target/i386/kvm/meson.build +2
-0@@ -11,6 +11,8 @@ i386_softmmu_kvm_ss.add(when: 'CONFIG_XEN_EMU', if_true: files('xen-emu.c')) i386_softmmu_kvm_ss.add(when: 'CONFIG_SEV', if_false: files('sev-stub.c')) +i386_softmmu_kvm_ss.add(when: 'CONFIG_TDX', if_true: files('tdx.c')) + i386_system_ss.add(when: 'CONFIG_HYPERV', if_true: files('hyperv.c'), if_false: files('hyperv-stub.c')) i386_system_ss.add_all(when: 'CONFIG_KVM', if_true: i386_softmmu_kvm_ss)target/i386/kvm/tdx.c new file +40
-0@@ -0,0 +1,40 @@ +/* +* QEMU TDX support +* +* Copyright Intel +* +* Author: +* Xiaoyao Li <[email protected]> +* +* This work is licensed under the terms of the GNU GPL, version 2 or later. +* See the COPYING file in the top-level directory +* +*/ + +#include "qemu/osdep.h" +#include "qom/object_interfaces.h" + +#include "tdx.h" + +/* tdx guest */ +OBJECT_DEFINE_TYPE_WITH_INTERFACES (TdxGuest, + tdx_guest, +TDX_GUEST , +CONFIDENTIAL_GUEST_SUPPORT , + { TYPE_USER_CREATABLE }, + { NULL }) + +static void tdx_guest_init (Object *obj) +{ +TdxGuest *tdx =TDX_GUEST (obj); + + tdx->attributes =0 ; +} + +static void tdx_guest_finalize (Object *obj) +{ +} + +static void tdx_guest_class_init (ObjectClass *oc,void *data) +{ +}target/i386/kvm/tdx.h new file +9
-0@@ -0,0 +1,19 @@ +#ifndef QEMU_I386_TDX_H +#define QEMU_I386_TDX_H + +#include "exec/confidential-guest-support.h" + +#define TYPE_TDX_GUEST "tdx-guest" +#define TDX_GUEST(obj) OBJECT_CHECK(TdxGuest, (obj), TYPE_TDX_GUEST) + +typedef struct TdxGuestClass {target/i386/kvm/tdx.h line 19
Quoted 10 lines
target/i386/kvm/tdx.h new file +10
-0+ ConfidentialGuestSupportClass parent_class; +} TdxGuestClass; + +typedef struct TdxGuest { +ConfidentialGuestSupport parent_obj; + +uint64_t attributes;/* TD attributes */ +}TdxGuest ; + +#endif /* QEMU_I386_TDX_H */ QAPI schema
Quoted 1 line
Thank you!
Re: [PATCH v2 02/58] i386: Introduce tdx-guest object
Quoted 20 lines · Xiaoyao Li
Quoted 9 lines · Markus Armbruster
Quoted 5 lines · Xiaoyao Li
Introduce tdx-guest object which implements the interface of
CONFIDENTIAL_GUEST_SUPPORT, and will be used to create TDX VMs (TDs) byqemu -machine ...,confidential-guest-support=tdx0 \ -object tdx-guset,id=tdx0Typo: tdx-guest
Will fix it.
Quoted 6 lines
Quoted 2 lines
It has only one property 'attributes' with fixed value 0 and not configurable so far.
This must refer to TdxGuest member @attributes.
"Property" suggests QOM property, which @attributes isn't, at least not in this patch. Will it become a QOM property later in this series?
Quoted 8 lines
At least not in this series. Maybe in the future there is request to directly configure the whole attributes via QOM property, but none from now.
I will change the description of it to avoid confusion.
Quoted 2 lines
Hmm, @attributes appears to remain unused until PATCH 14. Recommend to delay its addition until then.
IMHO, it's not suitable to introduce it in patch 14. Using a separate patch seems unnecessary. I'll leave it in this patch unless strong objection on it.
Not worth arguing about.
Quoted 90 lines
Quoted 81 lines
Quoted 81 lines
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>---
changes from RFC-V4
- make @attributes not user-settable--- configs/devices/i386-softmmu/default.mak | 1 + hw/i386/Kconfig | 5 +++ qapi/qom.json | 12 +++++++ target/i386/kvm/meson.build | 2 ++ target/i386/kvm/tdx.c | 40 ++++++++++++++++++++++++ target/i386/kvm/tdx.h | 19 +++++++++++ 6 files changed, 79 insertions(+)create mode 100644
target/i386/kvm/tdx.c
create mode 100644target/i386/kvm/tdx.hconfigs/devices/i386-softmmu/default.mak +1
-0@@ -18,6 +18,7 @@ #CONFIG_QXL=n #CONFIG_SEV=n #CONFIG_SGA=n +#CONFIG_TDX=n #CONFIG_TEST_DEVICES=n #CONFIG_TPM_CRB=n #CONFIG_TPM_TIS_ISA=nhw/i386/Kconfig +4
-0@@ -10,6 +10,10 @@ config SGX bool depends on KVM +config TDX + bool + depends on KVM + config PC bool imply APPLESMC@@ -26,6 +30,7 @@ config PC imply QXL imply SEV imply SGX + imply TDX imply TEST_DEVICES imply TPM_CRB imply TPM_TIS_ISAqapi/qom.json +10
-0@@ -866,6 +866,16 @@ 'reduced-phys-bits': 'uint32', '*kernel-hashes': 'bool' } } +## +# @TdxGuestProperties: +# +# Properties for tdx-guest objects. +# +# Since: 8.2 +## +{ 'struct': 'TdxGuestProperties', + 'data': { }} + ## # @ThreadContextProperties: #@@ -944,6 +954,7 @@ 'sev-guest', 'thread-context', 's390-pv-guest', + 'tdx-guest', 'throttle-group', 'tls-creds-anon', 'tls-creds-psk',@@ -1010,6 +1021,7 @@ 'secret_keyring': { 'type': 'SecretKeyringProperties', 'if': 'CONFIG_SECRET_KEYRING' },Quoted 8 lines
qapi/qom.json line 1027
Quoted 5 lines
qapi/qom.json +1
-0'sev-guest': 'SevGuestProperties', + 'tdx-guest': 'TdxGuestProperties', 'thread-context': 'ThreadContextProperties', 'throttle-group': 'ThrottleGroupProperties', 'tls-creds-anon': 'TlsCredsAnonProperties',Actually useful only when
CONFIG_TDXis on, but can't make it conditional here, asCONFIG_TDXis poisoned.
Quoted 1 line
In fact, I just followed what SEV did.
Yup.
Quoted 1 line
To me, it looks OK to make it conditional on
CONFIG_TDX. Could you please elaborate "but can't make it conditional here, asCONFIG_TDXis poisoned." ?
CONFIG_TDX is one of the macros that can only be used in target-dependent code. Enforced by config-poison.h's
#pragma GCC poison CONFIG_TDX
The code generated from qom.json is target-independent.
To use 'if': 'CONFIG_TDX', we'd have to move the definition to a target-dependent QAPI module, say qom-machine.json. Sadly, that's more trouble than it's worth.
[...]
PATCH v2 3/58 target/i386: Parse TDX vm type
In reply to Xiaoyao LiTDX VM requires VM type KVM_X86_TDX_VM to be passed to kvm_ioctl(KVM_CREATE_VM).
If tdx-guest object is specified to confidential-guest-support, like,
qemu -machine ...,confidential-guest-support=tdx0 \
-object tdx-guest,id=tdx0,...it parses VM type as KVM_X86_TDX_VM.
--- target/i386/kvm/kvm.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)
target/i386/kvm/kvm.c +9 -1
@@ -32,6 +32,7 @@#include "sysemu/runstate.h" #include "kvm_i386.h" #include "sev.h" +#include "tdx.h" #include "xen-emu.h" #include "hyperv.h" #include "hyperv-proto.h"
@@ -158,6 +159,7 @@ static int kvm_get_one_msr(X86CPU *cpu, int index, uint64_t *value);static const char * vm_type_name[] = { [KVM_X86_DEFAULT_VM ] ="default" , [KVM_X86_SW_PROTECTED_VM ] ="sw-protected-vm" , + [KVM_X86_TDX_VM ] ="tdx" , };int kvm_get_vm_type (MachineState *ms,const char *vm_type)
@@ -170,12 +172,18 @@ int kvm_get_vm_type(MachineState *ms, const char *vm_type) kvm_type =KVM_X86_DEFAULT_VM ; }else if (!g_ascii_strcasecmp (vm_type,"sw-protected-vm" )) { kvm_type =KVM_X86_SW_PROTECTED_VM ; - }else { + }else if (!g_ascii_strcasecmp (vm_type,"tdx" )) { + kvm_type =KVM_X86_TDX_VM ; + }else {error_report ("Unknown kvm-type specified '%s'" , vm_type);exit (1 ); } } +if (ms->cgs &&object_dynamic_cast (OBJECT (ms->cgs),TYPE_TDX_GUEST )) { + kvm_type =KVM_X86_TDX_VM ; + } +/* * old KVM doesn't support KVM_CAP_VM_TYPES and KVM_X86_DEFAULT_VM * is always supported
Signature · 1 line
2.34.1
Re: [PATCH v2 03/58] target/i386: Parse TDX vm type
Quoted 37 lines · Xiaoyao Li
TDX VM requires VM type
KVM_X86_TDX_VMto be passed tokvm_ioctl(KVM_CREATE_VM).If tdx-guest object is specified to confidential-guest-support, like,
qemu -machine ...,confidential-guest-support=tdx0 \ -object tdx-guest,id=tdx0,...it parses VM type as
KVM_X86_TDX_VM.Signed-off-by Xiaoyao Li <[email protected]>--- target/i386/kvm/kvm.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)target/i386/kvm/kvm.c +2
-0@@ -32,6 +32,7 @@#include "sysemu/runstate.h" #include "kvm_i386.h" #include "sev.h" +#include "tdx.h" #include "xen-emu.h" #include "hyperv.h" #include "hyperv-proto.h" @@ -158,6 +159,7 @@ static int kvm_get_one_msr(X86CPU *cpu, int index, uint64_t *value);static const char * vm_type_name[] = { [KVM_X86_DEFAULT_VM ] ="default" , [KVM_X86_SW_PROTECTED_VM ] ="sw-protected-vm" , + [KVM_X86_TDX_VM ] ="tdx" , };int kvm_get_vm_type (MachineState *ms,const char *vm_type)@@ -170,12 +172,18 @@ int kvm_get_vm_type(MachineState *ms, const char *vm_type) kvm_type =KVM_X86_DEFAULT_VM ;
target/i386/kvm/kvm.c int kvm_get_vm_type(MachineState *ms, const char *vm_type) · line 181
Quoted 10 lines
target/i386/kvm/kvm.c +3
-1}else if (!g_ascii_strcasecmp (vm_type,"sw-protected-vm" )) { kvm_type =KVM_X86_SW_PROTECTED_VM ; - }else { + }else if (!g_ascii_strcasecmp (vm_type,"tdx" )) { + kvm_type =KVM_X86_TDX_VM ; + }else {error_report ("Unknown kvm-type specified '%s'" , vm_type);exit (1 ); } }
This whole block of code should go away - as this should not exist as a user visible property. It should be sufficient to use the tdx-guest object type to identify use of TDX.
Quoted 11 lines
target/i386/kvm/kvm.c +4
-0+if (ms->cgs &&object_dynamic_cast (OBJECT (ms->cgs),TYPE_TDX_GUEST )) { + kvm_type =KVM_X86_TDX_VM ; + } +/* * old KVM doesn't support KVM_CAP_VM_TYPES and KVM_X86_DEFAULT_VM * is always supported Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 03/58] target/i386: Parse TDX vm type
Quoted 44 lines · Daniel P. Berrangé
Quoted 43 lines · Xiaoyao Li
TDX VM requires VM type
KVM_X86_TDX_VMto be passed tokvm_ioctl(KVM_CREATE_VM).If tdx-guest object is specified to confidential-guest-support, like,
qemu -machine ...,confidential-guest-support=tdx0 \ -object tdx-guest,id=tdx0,...it parses VM type as
KVM_X86_TDX_VM.Signed-off-by Xiaoyao Li <[email protected]>--- target/i386/kvm/kvm.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)target/i386/kvm/kvm.c +5
-1@@ -32,6 +32,7 @@#include "sysemu/runstate.h" #include "kvm_i386.h" #include "sev.h" +#include "tdx.h" #include "xen-emu.h" #include "hyperv.h" #include "hyperv-proto.h" @@ -158,6 +159,7 @@ static int kvm_get_one_msr(X86CPU *cpu, int index, uint64_t *value);static const char * vm_type_name[] = { [KVM_X86_DEFAULT_VM ] ="default" , [KVM_X86_SW_PROTECTED_VM ] ="sw-protected-vm" , + [KVM_X86_TDX_VM ] ="tdx" , };int kvm_get_vm_type (MachineState *ms,const char *vm_type)@@ -170,12 +172,18 @@ int kvm_get_vm_type(MachineState *ms, const char *vm_type) kvm_type =KVM_X86_DEFAULT_VM ; }else if (!g_ascii_strcasecmp (vm_type,"sw-protected-vm" )) { kvm_type =KVM_X86_SW_PROTECTED_VM ; - }else { + }else if (!g_ascii_strcasecmp (vm_type,"tdx" )) { + kvm_type =KVM_X86_TDX_VM ; + }else {
Quoted 9 lines
target/i386/kvm/kvm.c int kvm_get_vm_type(MachineState *ms, const char *vm_type) · line 181
Quoted 4 lines
target/i386/kvm/kvm.c +0
-0error_report ("Unknown kvm-type specified '%s'" , vm_type);exit (1 ); } }This whole block of code should go away - as this should not exist as a user visible property. It should be sufficient to use the tdx-guest object type to identify use of TDX.
yes, agreed.
It's here because this series is based on the gmem series, which introduced property. I'm sorry that I forgot to mention it in the commit message.
Next gmem series will drop the implementation of kvm-type property [1] and above code will be dropped in next version as well.
PATCH v2 4/58 target/i386: Introduce kvm_confidential_guest_init()
In reply to Xiaoyao LiIntroduce a separate function kvm_confidential_guest_init() for SEV (and future TDX).
--- target/i386/kvm/kvm.c | 11 ++++++++++- target/i386/sev.c | 1 - target/i386/sev.h | 2 ++ 3 files changed, 12 insertions(+), 2 deletions(-)
target/i386/kvm/kvm.c +10 -1
@@ -2633,6 +2633,15 @@ static MemoryListener kvm_x86_sw_protected_vm_memory_listener = { .priority = MEMORY_LISTENER_PRIORITY_ACCEL_HIGH, }; +static int kvm_confidential_guest_init (MachineState *ms,Error **errp) +{ +if (object_dynamic_cast (OBJECT (ms->cgs),TYPE_SEV_GUEST )) { +return sev_kvm_init (ms->cgs, errp); + } + +return 0 ; +} +int kvm_arch_init (MachineState *ms,KVMState *s) {X86MachineState *x86ms =X86_MACHINE (ms);
@@ -2654,7 +2663,7 @@ int kvm_arch_init(MachineState *ms, KVMState *s)* mechanisms are supported in future (e.g. TDX), they'll need * their own initialization either here or elsewhere. */ - ret =sev_kvm_init (ms->cgs, &local_err); + ret =kvm_confidential_guest_init (ms, &local_err);if (ret <0 ) {error_report_err (local_err);return ret;
target/i386/sev.c +0 -1
@@ -39,7 +39,6 @@#include "hw/i386/pc.h" #include "exec/address-spaces.h" -#define TYPE_SEV_GUEST "sev-guest"OBJECT_DECLARE_SIMPLE_TYPE (SevGuestState,SEV_GUEST )
target/i386/sev.h +2 -0
@@ -20,6 +20,8 @@#include "exec/confidential-guest-support.h" +#define TYPE_SEV_GUEST "sev-guest" +#define SEV_POLICY_NODBG 0x1#define SEV_POLICY_NOKS 0x2#define SEV_POLICY_ES 0x4
Signature · 1 line
2.34.1
Re: [PATCH v2 04/58] target/i386: Introduce kvm_confidential_guest_init()
Quoted 10 lines · Xiaoyao Li
Introduce a separate function
kvm_confidential_guest_init()for SEV (and future TDX).Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- target/i386/kvm/kvm.c | 11 ++++++++++- target/i386/sev.c | 1 - target/i386/sev.h | 2 ++ 3 files changed, 12 insertions(+), 2 deletions(-)
PATCH v2 5/58 i386/tdx: Implement tdx_kvm_init() to initialize TDX VM context
In reply to Xiaoyao LiIntroduce tdx_kvm_init() and invoke it in kvm_confidential_guest_init() if it's a TDX VM. More initialization will be added later.
--- target/i386/kvm/kvm.c | 15 ++++++--------- target/i386/kvm/meson.build | 2 +- target/i386/kvm/tdx-stub.c | 8 ++++++++ target/i386/kvm/tdx.c | 7 +++++++ target/i386/kvm/tdx.h | 2 ++ 5 files changed, 24 insertions(+), 10 deletions(-) create mode 100644 target/i386/kvm/tdx-stub.c
target/i386/kvm/kvm.c +6 -9
@@ -63,6 +63,7 @@#include "migration/blocker.h" #include "exec/memattrs.h" #include "trace.h" +#include "tdx.h" #include CONFIG_DEVICES
@@ -2637,6 +2638,8 @@ static int kvm_confidential_guest_init(MachineState *ms, Error **errp) {if (object_dynamic_cast (OBJECT (ms->cgs),TYPE_SEV_GUEST )) {return sev_kvm_init (ms->cgs, errp); + }else if (object_dynamic_cast (OBJECT (ms->cgs),TYPE_TDX_GUEST )) { +return tdx_kvm_init (ms, errp); }return 0 ;
@@ -2652,16 +2655,10 @@ int kvm_arch_init(MachineState *ms, KVMState *s)Error *local_err =NULL ;/* -* Initialize SEV context, if required +* Initialize confidential guest (SEV/TDX) context, if required * -* If no memory encryption is requested (ms->cgs == NULL) this is -* a no-op. -* -* It's also a no-op if a non-SEV confidential guest support -* mechanism is selected. SEV is the only mechanism available to -* select on x86 at present, so this doesn't arise, but if new -* mechanisms are supported in future (e.g. TDX), they'll need -* their own initialization either here or elsewhere. +* It's a no-op if a non-SEV/non-tdx confidential guest support +* mechanism is selected, i.e., ms->cgs == NULL */ ret =kvm_confidential_guest_init (ms, &local_err);if (ret <0 ) {
target/i386/kvm/meson.build +1 -1
@@ -11,7 +11,7 @@ i386_softmmu_kvm_ss.add(when: 'CONFIG_XEN_EMU', if_true: files('xen-emu.c')) i386_softmmu_kvm_ss.add(when: 'CONFIG_SEV', if_false: files('sev-stub.c')) -i386_softmmu_kvm_ss.add(when: 'CONFIG_TDX', if_true: files('tdx.c')) +i386_softmmu_kvm_ss.add(when: 'CONFIG_TDX', if_true: files('tdx.c'), if_false: files('tdx-stub.c')) i386_system_ss.add(when: 'CONFIG_HYPERV', if_true: files('hyperv.c'), if_false: files('hyperv-stub.c'))
target/i386/kvm/tdx-stub.c new file +8 -0
@@ -0,0 +1,8 @@ +#include "qemu/osdep.h" + +#include "tdx.h" + +int tdx_kvm_init (MachineState *ms,Error **errp) +{ +return -EINVAL ; +}
target/i386/kvm/tdx.c +7 -0
@@ -12,10 +12,17 @@*/ #include "qemu/osdep.h" +#include "qapi/error.h" #include "qom/object_interfaces.h" +#include "hw/i386/x86.h" #include "tdx.h" +int tdx_kvm_init (MachineState *ms,Error **errp) +{ +return 0 ; +} +/* tdx guest */ OBJECT_DEFINE_TYPE_WITH_INTERFACES(TdxGuest, tdx_guest,
target/i386/kvm/tdx.h +2 -0
@@ -16,4 +16,6 @@ typedef struct TdxGuest { uint64_t attributes; /* TD attributes */ } TdxGuest; +int tdx_kvm_init (MachineState *ms,Error **errp); +#endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
PATCH v2 6/58 i386/tdx: Get tdx_capabilities via KVM_TDX_CAPABILITIES
In reply to Xiaoyao LiKVM provides TDX capabilities via sub command KVM_TDX_CAPABILITIES of IOCTL(KVM_MEMORY_ENCRYPT_OP). Get the capabilities when initializing TDX context. It will be used to validate user's setting later.
Since there is no interface reporting how many cpuid configs contains in KVM_TDX_CAPABILITIES, QEMU chooses to try starting with a known number and abort when it exceeds KVM_MAX_CPUID_ENTRIES.
Besides, introduce the interfaces to invoke TDX "ioctls" at different scope (KVM, VM and VCPU) in preparation.
---
changes from v1:
- Make the error message more clear;
changes from RFC v4:
- start from nr_cpuid_configs = 6 for the loop;
- stop the loop when nr_cpuid_configs exceeds KVM_MAX_CPUID_ENTRIES;
--- target/i386/kvm/kvm.c | 2 - target/i386/kvm/kvm_i386.h | 2 + target/i386/kvm/tdx.c | 93 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 95 insertions(+), 2 deletions(-)
target/i386/kvm/kvm.c +0 -2
@@ -1751,8 +1751,6 @@ static int hyperv_init_vcpu(X86CPU *cpu)static Error *invtsc_mig_blocker; -#define KVM_MAX_CPUID_ENTRIES 100 -static void kvm_init_xsave (CPUX86State *env) {if (has_xsave2) {
target/i386/kvm/kvm_i386.h +2 -0
@@ -13,6 +13,8 @@#include "sysemu/kvm.h" +#define KVM_MAX_CPUID_ENTRIES 100 +#define kvm_apic_in_kernel () (kvm_irqchip_in_kernel())#ifdef CONFIG_KVM
target/i386/kvm/tdx.c +93 -0
107 lines not on this page · raw message
Signature · 1 line
2.34.1
Re: [PATCH v2 06/58] i386/tdx: Get tdx_capabilities via KVM_TDX_CAPABILITIES
Quoted 68 lines · Xiaoyao Li
KVM provides TDX capabilities via sub command
KVM_TDX_CAPABILITIESofIOCTL(KVM_MEMORY_ENCRYPT_OP). Get the capabilities when initializing TDX context. It will be used to validate user's setting later.Since there is no interface reporting how many cpuid configs contains in
KVM_TDX_CAPABILITIES, QEMU chooses to try starting with a known number and abort when it exceedsKVM_MAX_CPUID_ENTRIES.Besides, introduce the interfaces to invoke TDX "ioctls" at different scope (KVM, VM and VCPU) in preparation.
Signed-off-by Xiaoyao Li <[email protected]>---
changes from v1:
- Make the error message more clear;changes from RFC v4:
- start fromnr_cpuid_configs= 6 for the loop;
- stop the loop whennr_cpuid_configsexceedsKVM_MAX_CPUID_ENTRIES;--- target/i386/kvm/kvm.c | 2 - target/i386/kvm/kvm_i386.h | 2 + target/i386/kvm/tdx.c | 93 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 95 insertions(+), 2 deletions(-)target/i386/kvm/kvm.c +0
-2@@ -1751,8 +1751,6 @@ static int hyperv_init_vcpu(X86CPU *cpu)static Error *invtsc_mig_blocker; -#define KVM_MAX_CPUID_ENTRIES 100 -static void kvm_init_xsave (CPUX86State *env) {if (has_xsave2) {target/i386/kvm/kvm_i386.h +2
-0@@ -13,6 +13,8 @@#include "sysemu/kvm.h" +#define KVM_MAX_CPUID_ENTRIES 100 +#define kvm_apic_in_kernel () (kvm_irqchip_in_kernel())#ifdef CONFIG_KVM target/i386/kvm/tdx.c +3
-0@@ -12,14 +12,107 @@*/ #include "qemu/osdep.h" +#include "qemu/error-report.h" #include "qapi/error.h" #include "qom/object_interfaces.h" +#include "sysemu/kvm.h" #include "hw/i386/x86.h" +#include "kvm_i386.h" #include "tdx.h"
target/i386/kvm/tdx.c line 33
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+static struct kvm_tdx_capabilities *tdx_caps; + +enum tdx_ioctl_level { +TDX_PLATFORM_IOCTL , +TDX_VM_IOCTL , +TDX_VCPU_IOCTL , +}; + +static int __tdx_ioctl (void *state,enum tdx_ioctl_level level,int cmd_id, +__u32 flags,void *data)
Names with an initial double underscore are reserved for us by the platform implementation, so shouldn't be used in userspace app code.
Quoted 35 lines
target/i386/kvm/tdx.c +35
-0+{ +struct kvm_tdx_cmd tdx_cmd; +int r; + +memset (&tdx_cmd,0x0 ,sizeof (tdx_cmd)); + + tdx_cmd.id = cmd_id; + tdx_cmd.flags = flags; + tdx_cmd.data = (__u64 )(unsigned long )data; + +switch (level) { +case TDX_PLATFORM_IOCTL : + r =kvm_ioctl (kvm_state,KVM_MEMORY_ENCRYPT_OP , &tdx_cmd); +break ; +case TDX_VM_IOCTL : + r =kvm_vm_ioctl (kvm_state,KVM_MEMORY_ENCRYPT_OP , &tdx_cmd); +break ; +case TDX_VCPU_IOCTL : + r =kvm_vcpu_ioctl (state,KVM_MEMORY_ENCRYPT_OP , &tdx_cmd); +break ; +default : +error_report ("Invalid tdx_ioctl_level %d" , level); +exit (1 ); + } + +return r; +} + +static inline int tdx_platform_ioctl (int cmd_id,__u32 flags,void *data) +{ +return __tdx_ioctl (NULL ,TDX_PLATFORM_IOCTL , cmd_id, flags, data); +} + +static inline int tdx_vm_ioctl (int cmd_id,__u32 flags,void *data) +{
target/i386/kvm/tdx.c line 78
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+return __tdx_ioctl (NULL ,TDX_VM_IOCTL , cmd_id, flags, data); +} + +static inline int tdx_vcpu_ioctl (void *vcpu_fd,int cmd_id,__u32 flags, +void *data) +{ +return __tdx_ioctl (vcpu_fd,TDX_VCPU_IOCTL , cmd_id, flags, data); +} + +static void get_tdx_capabilities (void )
Pass in 'Error **errp'
target/i386/kvm/tdx.c line 83
Quoted 5 lines
target/i386/kvm/tdx.c +5
-0+{ +struct kvm_tdx_capabilities *caps; +/* 1st generation of TDX reports 6 cpuid configs */ +int nr_cpuid_configs =6 ; +int r, size;
It is preferrable to use 'size_t' for memory allocation sizes.
target/i386/kvm/tdx.c line 96
Quoted 13 lines
target/i386/kvm/tdx.c +13
-0+ + do { + size = sizeof(struct kvm_tdx_capabilities) + + nr_cpuid_configs * sizeof(struct kvm_tdx_cpuid_config); + caps = g_malloc0(size); + caps->nr_cpuid_configs = nr_cpuid_configs; + + r = tdx_vm_ioctl(KVM_TDX_CAPABILITIES, 0, caps); + if (r == -E2BIG) { + g_free(caps); + nr_cpuid_configs *= 2; + if (nr_cpuid_configs > KVM_MAX_CPUID_ENTRIES) { + error_report("KVM TDX seems broken that number of CPUID entries in kvm_tdx_capabilities exceeds limit");
Include the limit in the error message, so if we ever need to change the limit, it'll be clear what limit the QEMU version was built with.
Also use error_setg(errp, ...);
target/i386/kvm/tdx.c line 97
Quoted 1 line
target/i386/kvm/tdx.c +1
-0+exit (1 );
Return -1
target/i386/kvm/tdx.c line 101
Quoted 4 lines
target/i386/kvm/tdx.c +4
-0+ } + }else if (r <0 ) { +g_free (caps); +error_report ("KVM_TDX_CAPABILITIES failed: %s" ,strerror (-r));
Use error_setg_errno(errp, ...) instead of calling strerror yourself;
target/i386/kvm/tdx.c line 102
Quoted 1 line
target/i386/kvm/tdx.c +1
-0+exit (1 );
Return -1
target/i386/kvm/tdx.c line 107
Quoted 5 lines
target/i386/kvm/tdx.c +5
-0+ } + } +while (r == -E2BIG ); + + tdx_caps = caps;
Return 0
target/i386/kvm/tdx.c line 113
Quoted 6 lines
target/i386/kvm/tdx.c +4
-0+} +int tdx_kvm_init (MachineState *ms,Error **errp) { +if (!tdx_caps) { +get_tdx_capabilities ();
Pass 'errp' into this method, and check return value for failure
Quoted 8 lines
target/i386/kvm/tdx.c +2
-0+ } +return 0 ; }Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 06/58] i386/tdx: Get tdx_capabilities via KVM_TDX_CAPABILITIES
Quoted 69 lines · Daniel P. Berrangé
Quoted 68 lines · Xiaoyao Li
KVM provides TDX capabilities via sub command
KVM_TDX_CAPABILITIESofIOCTL(KVM_MEMORY_ENCRYPT_OP). Get the capabilities when initializing TDX context. It will be used to validate user's setting later.Since there is no interface reporting how many cpuid configs contains in
KVM_TDX_CAPABILITIES, QEMU chooses to try starting with a known number and abort when it exceedsKVM_MAX_CPUID_ENTRIES.Besides, introduce the interfaces to invoke TDX "ioctls" at different scope (KVM, VM and VCPU) in preparation.
Signed-off-by Xiaoyao Li <[email protected]>---
changes from v1:
- Make the error message more clear;changes from RFC v4:
- start fromnr_cpuid_configs= 6 for the loop;
- stop the loop whennr_cpuid_configsexceedsKVM_MAX_CPUID_ENTRIES;--- target/i386/kvm/kvm.c | 2 - target/i386/kvm/kvm_i386.h | 2 + target/i386/kvm/tdx.c | 93 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 95 insertions(+), 2 deletions(-)target/i386/kvm/kvm.c +0
-2@@ -1751,8 +1751,6 @@ static int hyperv_init_vcpu(X86CPU *cpu)static Error *invtsc_mig_blocker; -#define KVM_MAX_CPUID_ENTRIES 100 -static void kvm_init_xsave (CPUX86State *env) {if (has_xsave2) {target/i386/kvm/kvm_i386.h +2
-0@@ -13,6 +13,8 @@#include "sysemu/kvm.h" +#define KVM_MAX_CPUID_ENTRIES 100 +#define kvm_apic_in_kernel () (kvm_irqchip_in_kernel())#ifdef CONFIG_KVM target/i386/kvm/tdx.c +3
-0@@ -12,14 +12,107 @@*/ #include "qemu/osdep.h" +#include "qemu/error-report.h" #include "qapi/error.h" #include "qom/object_interfaces.h" +#include "sysemu/kvm.h" #include "hw/i386/x86.h" +#include "kvm_i386.h" #include "tdx.h"
Quoted 14 lines
target/i386/kvm/tdx.c line 33
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+static struct kvm_tdx_capabilities *tdx_caps; + +enum tdx_ioctl_level { +TDX_PLATFORM_IOCTL , +TDX_VM_IOCTL , +TDX_VCPU_IOCTL , +}; + +static int __tdx_ioctl (void *state,enum tdx_ioctl_level level,int cmd_id, +__u32 flags,void *data)Names with an initial double underscore are reserved for us by the platform implementation, so shouldn't be used in userspace app code.
How about tdx_ioctl_internal() ?
Quoted 39 lines
Quoted 39 lines
target/i386/kvm/tdx.c +39
-0+{ +struct kvm_tdx_cmd tdx_cmd; +int r; + +memset (&tdx_cmd,0x0 ,sizeof (tdx_cmd)); + + tdx_cmd.id = cmd_id; + tdx_cmd.flags = flags; + tdx_cmd.data = (__u64 )(unsigned long )data; + +switch (level) { +case TDX_PLATFORM_IOCTL : + r =kvm_ioctl (kvm_state,KVM_MEMORY_ENCRYPT_OP , &tdx_cmd); +break ; +case TDX_VM_IOCTL : + r =kvm_vm_ioctl (kvm_state,KVM_MEMORY_ENCRYPT_OP , &tdx_cmd); +break ; +case TDX_VCPU_IOCTL : + r =kvm_vcpu_ioctl (state,KVM_MEMORY_ENCRYPT_OP , &tdx_cmd); +break ; +default : +error_report ("Invalid tdx_ioctl_level %d" , level); +exit (1 ); + } + +return r; +} + +static inline int tdx_platform_ioctl (int cmd_id,__u32 flags,void *data) +{ +return __tdx_ioctl (NULL ,TDX_PLATFORM_IOCTL , cmd_id, flags, data); +} + +static inline int tdx_vm_ioctl (int cmd_id,__u32 flags,void *data) +{ +return __tdx_ioctl (NULL ,TDX_VM_IOCTL , cmd_id, flags, data); +} + +static inline int tdx_vcpu_ioctl(void *vcpu_fd, int cmd_id, __u32 flags,
Quoted 8 lines
target/i386/kvm/tdx.c line 78
Quoted 6 lines
target/i386/kvm/tdx.c +6
-0+ void *data) +{ +return __tdx_ioctl (vcpu_fd,TDX_VCPU_IOCTL , cmd_id, flags, data); +} + +static void get_tdx_capabilities (void )Pass in 'Error **errp'
OK. Will do it and all the following.
Thanks!
Quoted 70 lines
target/i386/kvm/tdx.c line 83
Quoted 5 lines
target/i386/kvm/tdx.c +5
-0+{ +struct kvm_tdx_capabilities *caps; +/* 1st generation of TDX reports 6 cpuid configs */ +int nr_cpuid_configs =6 ; +int r, size;It is preferrable to use '
size_t' for memory allocation sizes.target/i386/kvm/tdx.c line 96
Quoted 13 lines
target/i386/kvm/tdx.c +13
-0+ + do { + size = sizeof(struct kvm_tdx_capabilities) + + nr_cpuid_configs * sizeof(struct kvm_tdx_cpuid_config); + caps = g_malloc0(size); + caps->nr_cpuid_configs = nr_cpuid_configs; + + r = tdx_vm_ioctl(KVM_TDX_CAPABILITIES, 0, caps); + if (r == -E2BIG) { + g_free(caps); + nr_cpuid_configs *= 2; + if (nr_cpuid_configs > KVM_MAX_CPUID_ENTRIES) { + error_report("KVM TDX seems broken that number of CPUID entries in kvm_tdx_capabilities exceeds limit");Include the limit in the error message, so if we ever need to change the limit, it'll be clear what limit the QEMU version was built with.
Also use
error_setg(errp, ...);target/i386/kvm/tdx.c line 97
Quoted 1 line
target/i386/kvm/tdx.c +1
-0+exit (1 );Return -1
target/i386/kvm/tdx.c line 101
Quoted 4 lines
target/i386/kvm/tdx.c +4
-0+ } + }else if (r <0 ) { +g_free (caps); +error_report ("KVM_TDX_CAPABILITIES failed: %s" ,strerror (-r));Use
error_setg_errno(errp, ...)instead of calling strerror yourself;target/i386/kvm/tdx.c line 102
Quoted 1 line
target/i386/kvm/tdx.c +1
-0+exit (1 );Return -1
target/i386/kvm/tdx.c line 107
Quoted 5 lines
target/i386/kvm/tdx.c +5
-0+ } + } +while (r == -E2BIG ); + + tdx_caps = caps;Return 0
target/i386/kvm/tdx.c line 113
Quoted 6 lines
target/i386/kvm/tdx.c +4
-0+} +int tdx_kvm_init (MachineState *ms,Error **errp) { +if (!tdx_caps) { +get_tdx_capabilities ();Pass 'errp' into this method, and check return value for failure
Quoted 8 lines
target/i386/kvm/tdx.c +2
-0+ } +return 0 ; }Signature · 1 line
2.34.1With regards,
Daniel
Re: [PATCH v2 06/58] i386/tdx: Get tdx_capabilities via KVM_TDX_CAPABILITIES
Quoted 77 lines · Xiaoyao Li
Quoted 67 lines · Daniel P. Berrangé
Quoted 66 lines · Xiaoyao Li
KVM provides TDX capabilities via sub command
KVM_TDX_CAPABILITIESofIOCTL(KVM_MEMORY_ENCRYPT_OP). Get the capabilities when initializing TDX context. It will be used to validate user's setting later.Since there is no interface reporting how many cpuid configs contains in
KVM_TDX_CAPABILITIES, QEMU chooses to try starting with a known number and abort when it exceedsKVM_MAX_CPUID_ENTRIES.Besides, introduce the interfaces to invoke TDX "ioctls" at different scope (KVM, VM and VCPU) in preparation.
Signed-off-by Xiaoyao Li <[email protected]>---
changes from v1:
- Make the error message more clear;changes from RFC v4:
- start fromnr_cpuid_configs= 6 for the loop;
- stop the loop whennr_cpuid_configsexceedsKVM_MAX_CPUID_ENTRIES;--- target/i386/kvm/kvm.c | 2 - target/i386/kvm/kvm_i386.h | 2 + target/i386/kvm/tdx.c | 93 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 95 insertions(+), 2 deletions(-)target/i386/kvm/kvm.c +0
-2@@ -1751,8 +1751,6 @@ static int hyperv_init_vcpu(X86CPU *cpu)static Error *invtsc_mig_blocker; -#define KVM_MAX_CPUID_ENTRIES 100 -static void kvm_init_xsave (CPUX86State *env) {if (has_xsave2) {target/i386/kvm/kvm_i386.h +2
-0@@ -13,6 +13,8 @@#include "sysemu/kvm.h" +#define KVM_MAX_CPUID_ENTRIES 100 +#define kvm_apic_in_kernel () (kvm_irqchip_in_kernel())#ifdef CONFIG_KVM target/i386/kvm/tdx.c +9
-0@@ -12,14 +12,107 @@*/ #include "qemu/osdep.h" +#include "qemu/error-report.h" #include "qapi/error.h" #include "qom/object_interfaces.h" +#include "sysemu/kvm.h" #include "hw/i386/x86.h" +#include "kvm_i386.h" #include "tdx.h" +static struct kvm_tdx_capabilities *tdx_caps; + +enum tdx_ioctl_level { +TDX_PLATFORM_IOCTL , +TDX_VM_IOCTL , +TDX_VCPU_IOCTL ,Quoted 8 lines
target/i386/kvm/tdx.c line 33
Quoted 4 lines
target/i386/kvm/tdx.c +4
-0+}; + +static int __tdx_ioctl (void *state,enum tdx_ioctl_level level,int cmd_id, +__u32 flags,void *data)Names with an initial double underscore are reserved for us by the platform implementation, so shouldn't be used in userspace app code.
Quoted 1 line
How about
tdx_ioctl_internal()?
Sure, that's fine.
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
PATCH v2 7/58 i386/tdx: Introduce is_tdx_vm() helper and cache tdx_guest object
In reply to Xiaoyao LiIt will need special handling for TDX VMs all around the QEMU. Introduce is_tdx_vm() helper to query if it's a TDX VM.
Cache tdx_guest object thus no need to cast from ms->cgs every time.
--- target/i386/kvm/tdx.c | 13 +++++++++++++ target/i386/kvm/tdx.h | 10 ++++++++++ 2 files changed, 23 insertions(+)
target/i386/kvm/tdx.c +13 -0
@@ -21,8 +21,16 @@#include "kvm_i386.h" #include "tdx.h" +static TdxGuest *tdx_guest; +static struct kvm_tdx_capabilities *tdx_caps; +/* It's valid after kvm_confidential_guest_init()->kvm_tdx_init() */ +bool is_tdx_vm (void ) +{ +return !!tdx_guest; +} +enum tdx_ioctl_level {TDX_PLATFORM_IOCTL ,TDX_VM_IOCTL ,
@@ -109,10 +117,15 @@ static void get_tdx_capabilities(void)int tdx_kvm_init (MachineState *ms,Error **errp) { +TdxGuest *tdx = (TdxGuest *)object_dynamic_cast (OBJECT (ms->cgs), +TYPE_TDX_GUEST ); +if (!tdx_caps) {get_tdx_capabilities (); } + tdx_guest = tdx; +return 0 ; }
target/i386/kvm/tdx.h +10 -0
@@ -1,6 +1,10 @@#ifndef QEMU_I386_TDX_H #define QEMU_I386_TDX_H +#ifndef CONFIG_USER_ONLY +#include CONFIG_DEVICES /* CONFIG_TDX */ +#endif +#include "exec/confidential-guest-support.h" #define TYPE_TDX_GUEST "tdx-guest"
@@ -16,6 +20,12 @@ typedef struct TdxGuest { uint64_t attributes; /* TD attributes */ } TdxGuest; +#ifdef CONFIG_TDX +bool is_tdx_vm (void ); +#else +#define is_tdx_vm () 0 +#endif /* CONFIG_TDX */ +int tdx_kvm_init (MachineState *ms,Error **errp);#endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
Re: [PATCH v2 07/58] i386/tdx: Introduce is_tdx_vm() helper and cache tdx_guest object
Quoted 34 lines · Xiaoyao Li
It will need special handling for TDX VMs all around the QEMU. Introduce
is_tdx_vm()helper to query if it's a TDX VM.Cache
tdx_guestobject thus no need to cast fromms->cgsevery time.Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- target/i386/kvm/tdx.c | 13 +++++++++++++ target/i386/kvm/tdx.h | 10 ++++++++++ 2 files changed, 23 insertions(+)target/i386/kvm/tdx.c +8
-0@@ -21,8 +21,16 @@#include "kvm_i386.h" #include "tdx.h" +static TdxGuest *tdx_guest; +static struct kvm_tdx_capabilities *tdx_caps; +/* It's valid after kvm_confidential_guest_init()->kvm_tdx_init() */ +bool is_tdx_vm (void ) +{ +return !!tdx_guest; +} +enum tdx_ioctl_level {TDX_PLATFORM_IOCTL ,TDX_VM_IOCTL ,
target/i386/kvm/tdx.c static void get_tdx_capabilities(void) · line 121
Quoted 5 lines
target/i386/kvm/tdx.c +2
-0@@ -109,10 +117,15 @@ static void get_tdx_capabilities(void)int tdx_kvm_init (MachineState *ms,Error **errp) { +TdxGuest *tdx = (TdxGuest *)object_dynamic_cast (OBJECT (ms->cgs), +TYPE_TDX_GUEST );
This method can return NULL. Presumably tdx_kvm_init() should only be called if we already checked ms->cgs == TYPE_TDX_GUEST. If so then use object_dynamic_cast_assert() instead.
Quoted 41 lines
target/i386/kvm/tdx.c +3
-0+if (!tdx_caps) {get_tdx_capabilities (); } + tdx_guest = tdx; +return 0 ; }target/i386/kvm/tdx.h +10
-0@@ -1,6 +1,10 @@#ifndef QEMU_I386_TDX_H #define QEMU_I386_TDX_H +#ifndef CONFIG_USER_ONLY +#include CONFIG_DEVICES /* CONFIG_TDX */ +#endif +#include "exec/confidential-guest-support.h" #define TYPE_TDX_GUEST "tdx-guest"@@ -16,6 +20,12 @@ typedef struct TdxGuest { uint64_t attributes; /* TD attributes */ } TdxGuest; +#ifdef CONFIG_TDX +bool is_tdx_vm (void ); +#else +#define is_tdx_vm () 0 +#endif /* CONFIG_TDX */ +int tdx_kvm_init (MachineState *ms,Error **errp);#endif /* QEMU_I386_TDX_H */ Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 07/58] i386/tdx: Introduce is_tdx_vm() helper and cache tdx_guest object
Quoted 36 lines · Daniel P. Berrangé
Quoted 35 lines · Xiaoyao Li
It will need special handling for TDX VMs all around the QEMU. Introduce
is_tdx_vm()helper to query if it's a TDX VM.Cache
tdx_guestobject thus no need to cast fromms->cgsevery time.Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- target/i386/kvm/tdx.c | 13 +++++++++++++ target/i386/kvm/tdx.h | 10 ++++++++++ 2 files changed, 23 insertions(+)target/i386/kvm/tdx.c +8
-0@@ -21,8 +21,16 @@#include "kvm_i386.h" #include "tdx.h" +static TdxGuest *tdx_guest; +static struct kvm_tdx_capabilities *tdx_caps; +/* It's valid after kvm_confidential_guest_init()->kvm_tdx_init() */ +bool is_tdx_vm (void ) +{ +return !!tdx_guest; +} +enum tdx_ioctl_level {TDX_PLATFORM_IOCTL ,TDX_VM_IOCTL ,@@ -109,10 +117,15 @@ static void get_tdx_capabilities(void)
Quoted 9 lines
target/i386/kvm/tdx.c static void get_tdx_capabilities(void) · line 121
Quoted 4 lines
target/i386/kvm/tdx.c +2
-0int tdx_kvm_init (MachineState *ms,Error **errp) { +TdxGuest *tdx = (TdxGuest *)object_dynamic_cast (OBJECT (ms->cgs), +TYPE_TDX_GUEST );This method can return NULL. Presumably
tdx_kvm_init()should only be called if we already checkedms->cgs==TYPE_TDX_GUEST. If so then useobject_dynamic_cast_assert()instead.
object_dynamic_cast_assert() is for OBJECT_CHECK() and INTERFACE_CHECK().
So I will use TDX_GUEST(OBJECT(ms->cgs)) (introduced in patch 2) instead, which is the wrapper of OBJECT_CHECK().
PATCH v2 8/58 i386/tdx: Adjust the supported CPUID based on TDX restrictions
In reply to Xiaoyao LiAccording to Chapter "CPUID Virtualization" in TDX module spec, CPUID bits of TD can be classified into 6 types:
------------------------------------------------------------------------
1 | As configured | configurable by VMM, independent of native value;
------------------------------------------------------------------------
2 | As configured | configurable by VMM if the bit is supported natively
(if native) | Otherwise it equals as native(0).
------------------------------------------------------------------------
3 | Fixed | fixed to 0/1
------------------------------------------------------------------------
4 | Native | reflect the native value
------------------------------------------------------------------------
5 | Calculated | calculated by TDX module.
------------------------------------------------------------------------
6 | Inducing #VE | get #VE exception
------------------------------------------------------------------------
Note:
1. All the configurable XFAM related features and TD attributes related features fall into type #2. And fixed0/1 bits of XFAM and TD attributes fall into type #3.
2. For CPUID leaves not listed in "CPUID virtualization Overview" table in TDX module spec, TDX module injects #VE to TDs when those are queried. For this case, TDs can request CPUID emulation from VMM via TDVMCALL and the values are fully controlled by VMM.
Due to TDX module has its own virtualization policy on CPUID bits, it leads to what reported via KVM_GET_SUPPORTED_CPUID diverges from the supported CPUID bits for TDs. In order to keep a consistent CPUID configuration between VMM and TDs. Adjust supported CPUID for TDs based on TDX restrictions.
Currently only focus on the CPUID leaves recognized by QEMU's feature_word_info[] that are indexed by a FeatureWord.
Introduce a TDX CPUID lookup table, which maintains 1 entry for each FeatureWord. Each entry has below fields:
- tdx_fixed0/1: The bits that are fixed as 0/1;
- vmm_fixup: The bits that are configurable from the view of TDX module.
But they requires emulation of VMM when they are configured as enabled. For those, they are not supported if VMM doesn't report them as supported. So they need be fixed up by checking if VMM supports them.
- inducing_ve: TD gets #VE when querying this CPUID leaf. The result is totally configurable by VMM.
- supported_on_ve: It's valid only when @inducing_ve is true. It represents the maximum feature set supported that be emulated for TDs.
By applying TDX CPUID lookup table and TDX capabilities reported from TDX module, the supported CPUID for TDs can be obtained from following steps:
- get the base of VMM supported feature set;
- if the leaf is not a FeatureWord just return VMM's value without modification;
- if the leaf is an inducing_ve type, applying supported_on_ve mask and return;
- include all native bits, it covers type #2, #4, and parts of type #1. (it also includes some unsupported bits. The following step will correct it.)
- apply fixed0/1 to it (it covers #3, and rectifies the previous step);
- add configurable bits (it covers the other part of type #1);
- fix the ones in vmm_fixup;
- filter the one has valid .supported field;
(Calculated type is ignored since it's determined at runtime).
--- target/i386/cpu.h | 16 +++ target/i386/kvm/kvm.c | 4 + target/i386/kvm/tdx.c | 254 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 2 + 4 files changed, 276 insertions(+)
target/i386/cpu.h +16 -0
@@ -780,6 +780,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,/* Support RDFSBASE/RDGSBASE/WRFSBASE/WRGSBASE */ #define CPUID_7_0_EBX_FSGSBASE (1U << 0) +/* Support for TSC adjustment MSR 0x3B */ +#define CPUID_7_0_EBX_TSC_ADJUST (1U << 1)/* Support SGX */ #define CPUID_7_0_EBX_SGX (1U << 2)/* 1st Group of Advanced Bit Manipulation Extensions */
@@ -798,8 +800,12 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_EBX_INVPCID (1U << 10)/* Restricted Transactional Memory */ #define CPUID_7_0_EBX_RTM (1U << 11) +/* Cache QoS Monitoring */ +#define CPUID_7_0_EBX_PQM (1U << 12)/* Memory Protection Extension */ #define CPUID_7_0_EBX_MPX (1U << 14) +/* Resource Director Technology Allocation */ +#define CPUID_7_0_EBX_RDT_A (1U << 15)/* AVX-512 Foundation */ #define CPUID_7_0_EBX_AVX512F (1U << 16)/* AVX-512 Doubleword & Quadword Instruction */
@@ -855,10 +861,16 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_ECX_AVX512VNNI (1U << 11)/* Support for VPOPCNT[B,W] and VPSHUFBITQMB */ #define CPUID_7_0_ECX_AVX512BITALG (1U << 12) +/* Intel Total Memory Encryption */ +#define CPUID_7_0_ECX_TME (1U << 13)/* POPCNT for vectors of DW/QW */ #define CPUID_7_0_ECX_AVX512_VPOPCNTDQ (1U << 14) +/* Placeholder for bit 15 */ +#define CPUID_7_0_ECX_FZM (1U << 15)/* 5-level Page Tables */ #define CPUID_7_0_ECX_LA57 (1U << 16) +/* MAWAU for MPX */ +#define CPUID_7_0_ECX_MAWAU (31U << 17)/* Read Processor ID */ #define CPUID_7_0_ECX_RDPID (1U << 22)/* Bus Lock Debug Exception */
@@ -869,6 +881,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_ECX_MOVDIRI (1U << 27)/* Move 64 Bytes as Direct Store Instruction */ #define CPUID_7_0_ECX_MOVDIR64B (1U << 28) +/* ENQCMD and ENQCMDS instructions */ +#define CPUID_7_0_ECX_ENQCMD (1U << 29)/* Support SGX Launch Control */ #define CPUID_7_0_ECX_SGX_LC (1U << 30)/* Protection Keys for Supervisor-mode Pages */
@@ -886,6 +900,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_EDX_SERIALIZE (1U << 14)/* TSX Suspend Load Address Tracking instruction */ #define CPUID_7_0_EDX_TSX_LDTRK (1U << 16) +/* PCONFIG instruction */ +#define CPUID_7_0_EDX_PCONFIG (1U << 18)/* Architectural LBRs */ #define CPUID_7_0_EDX_ARCH_LBR (1U << 19)/* AMX_BF16 instruction */
target/i386/kvm/kvm.c +4 -0
@@ -539,6 +539,10 @@ uint32_t kvm_arch_get_supported_cpuid(KVMState *s, uint32_t function, ret |=1U <<KVM_HINTS_REALTIME ; } +if (is_tdx_vm ()) { +tdx_get_supported_cpuid (function, index, reg, &ret); + } +return ret; }
target/i386/kvm/tdx.c +254 -0
271 lines not on this page · raw message
target/i386/kvm/tdx.h +2 -0
@@ -27,5 +27,7 @@ bool is_tdx_vm(void);#endif /* CONFIG_TDX */ int tdx_kvm_init (MachineState *ms,Error **errp); +void tdx_get_supported_cpuid (uint32_t function,uint32_t index,int reg, +uint32_t *ret);#endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
Re: [PATCH v2 08/58] i386/tdx: Adjust the supported CPUID based on TDX restrictions
Quoted 4 lines · Xiaoyao Li
target/i386/kvm/tdx.c +0
-0
...
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+static inline uint32_t host_cpuid_reg (uint32_t function, +uint32_t index,int reg) +{ +uint32_t eax, ebx, ecx, edx; +uint32_t ret =0 ; + +host_cpuid (function, index, &eax, &ebx, &ecx, &edx); + +switch (reg) { +case R_EAX :
target/i386/kvm/tdx.c bool is_tdx_vm(void) · line 171
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+ ret |= eax; +break ; +case R_EBX : + ret |= ebx; +break ; +case R_ECX : + ret |= ecx; +break ; +case R_EDX : + ret |= edx;
Nitpick: "|" isn't needed as we initialize ret = 0 above. Just '='.
Signature · 1 line
Isaku Yamahata <[email protected]>
Re: [PATCH v2 08/58] i386/tdx: Adjust the supported CPUID based on TDX restrictions
Quoted 22 lines · Isaku Yamahata
Quoted 4 lines · Xiaoyao Li
target/i386/kvm/tdx.c +0
-0...
Quoted 14 lines
target/i386/kvm/tdx.c +14
-0+static inline uint32_t host_cpuid_reg (uint32_t function, +uint32_t index,int reg) +{ +uint32_t eax, ebx, ecx, edx; +uint32_t ret =0 ; + +host_cpuid (function, index, &eax, &ebx, &ecx, &edx); + +switch (reg) { +case R_EAX : + ret |= eax; +break ; +case R_EBX : + ret |= ebx;
Quoted 8 lines
target/i386/kvm/tdx.c bool is_tdx_vm(void) · line 171
Quoted 6 lines
target/i386/kvm/tdx.c +6
-0+break ; +case R_ECX : + ret |= ecx; +break ; +case R_EDX : + ret |= edx;Nitpick: "|" isn't needed as we initialize ret = 0 above. Just '='.
Will fix it in next version.
thanks!
Re: [PATCH v2 08/58] i386/tdx: Adjust the supported CPUID based on TDX restrictions
In reply to Xiaoyao LiHi,
Quoted 221 lines · Xiaoyao Li
According to Chapter "CPUID Virtualization" in TDX module spec, CPUID bits of TD can be classified into 6 types:
------------------------------------------------------------------------
1 | As configured | configurable by VMM, independent of native value;
------------------------------------------------------------------------
2 | As configured | configurable by VMM if the bit is supported natively
(if native) | Otherwise it equals as native(0).
------------------------------------------------------------------------
3 | Fixed | fixed to 0/1
------------------------------------------------------------------------
4 | Native | reflect the native value
------------------------------------------------------------------------
5 | Calculated | calculated by TDX module.
------------------------------------------------------------------------
6 | Inducing #VE | get #VE exception
------------------------------------------------------------------------Note:
1. All the configurable XFAM related features and TD attributes related features fall into type #2. And fixed0/1 bits of XFAM and TD attributes fall into type #3.2. For CPUID leaves not listed in "CPUID virtualization Overview" table in TDX module spec, TDX module injects #VE to TDs when those are queried. For this case, TDs can request CPUID emulation from VMM via TDVMCALL and the values are fully controlled by VMM.
Due to TDX module has its own virtualization policy on CPUID bits, it leads to what reported via
KVM_GET_SUPPORTED_CPUIDdiverges from the supported CPUID bits for TDs. In order to keep a consistent CPUID configuration between VMM and TDs. Adjust supported CPUID for TDs based on TDX restrictions.Currently only focus on the CPUID leaves recognized by QEMU's feature_word_info[] that are indexed by a FeatureWord.
Introduce a TDX CPUID lookup table, which maintains 1 entry for each FeatureWord. Each entry has below fields:
- tdx_fixed0/1: The bits that are fixed as 0/1;
-
vmm_fixup: The bits that are configurable from the view of TDX module.
But they requires emulation of VMM when they are configured
as enabled. For those, they are not supported if VMM doesn't report them as supported. So they need be fixed up by checking if VMM supports them.-
inducing_ve: TD gets #VE when querying this CPUID leaf. The result is totally configurable by VMM.-
supported_on_ve: It's valid only when @inducing_ve is true. It represents the maximum feature set supported that be emulated for TDs.By applying TDX CPUID lookup table and TDX capabilities reported from TDX module, the supported CPUID for TDs can be obtained from following steps:
- get the base of VMM supported feature set;
- if the leaf is not a FeatureWord just return VMM's value without modification;
- if the leaf is an
inducing_vetype, applyingsupported_on_vemask and return;- include all native bits, it covers type #2, #4, and parts of type #1. (it also includes some unsupported bits. The following step will correct it.)
- apply fixed0/1 to it (it covers #3, and rectifies the previous step);
- add configurable bits (it covers the other part of type #1);
- fix the ones in
vmm_fixup;- filter the one has valid .supported field;
(Calculated type is ignored since it's determined at runtime).
Co-developed-by Chenyi Qiang <[email protected]>Signed-off-by Chenyi Qiang <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- target/i386/cpu.h | 16 +++ target/i386/kvm/kvm.c | 4 + target/i386/kvm/tdx.c | 254 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 2 + 4 files changed, 276 insertions(+)target/i386/cpu.h +16
-0@@ -780,6 +780,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,/* Support RDFSBASE/RDGSBASE/WRFSBASE/WRGSBASE */ #define CPUID_7_0_EBX_FSGSBASE (1U << 0) +/* Support for TSC adjustment MSR 0x3B */ +#define CPUID_7_0_EBX_TSC_ADJUST (1U << 1)/* Support SGX */ #define CPUID_7_0_EBX_SGX (1U << 2)/* 1st Group of Advanced Bit Manipulation Extensions */ @@ -798,8 +800,12 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_EBX_INVPCID (1U << 10)/* Restricted Transactional Memory */ #define CPUID_7_0_EBX_RTM (1U << 11) +/* Cache QoS Monitoring */ +#define CPUID_7_0_EBX_PQM (1U << 12)/* Memory Protection Extension */ #define CPUID_7_0_EBX_MPX (1U << 14) +/* Resource Director Technology Allocation */ +#define CPUID_7_0_EBX_RDT_A (1U << 15)/* AVX-512 Foundation */ #define CPUID_7_0_EBX_AVX512F (1U << 16)/* AVX-512 Doubleword & Quadword Instruction */ @@ -855,10 +861,16 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_ECX_AVX512VNNI (1U << 11)/* Support for VPOPCNT[B,W] and VPSHUFBITQMB */ #define CPUID_7_0_ECX_AVX512BITALG (1U << 12) +/* Intel Total Memory Encryption */ +#define CPUID_7_0_ECX_TME (1U << 13)/* POPCNT for vectors of DW/QW */ #define CPUID_7_0_ECX_AVX512_VPOPCNTDQ (1U << 14) +/* Placeholder for bit 15 */ +#define CPUID_7_0_ECX_FZM (1U << 15)/* 5-level Page Tables */ #define CPUID_7_0_ECX_LA57 (1U << 16) +/* MAWAU for MPX */ +#define CPUID_7_0_ECX_MAWAU (31U << 17)/* Read Processor ID */ #define CPUID_7_0_ECX_RDPID (1U << 22)/* Bus Lock Debug Exception */ @@ -869,6 +881,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_ECX_MOVDIRI (1U << 27)/* Move 64 Bytes as Direct Store Instruction */ #define CPUID_7_0_ECX_MOVDIR64B (1U << 28) +/* ENQCMD and ENQCMDS instructions */ +#define CPUID_7_0_ECX_ENQCMD (1U << 29)/* Support SGX Launch Control */ #define CPUID_7_0_ECX_SGX_LC (1U << 30)/* Protection Keys for Supervisor-mode Pages */ @@ -886,6 +900,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_EDX_SERIALIZE (1U << 14)/* TSX Suspend Load Address Tracking instruction */ #define CPUID_7_0_EDX_TSX_LDTRK (1U << 16) +/* PCONFIG instruction */ +#define CPUID_7_0_EDX_PCONFIG (1U << 18)/* Architectural LBRs */ #define CPUID_7_0_EDX_ARCH_LBR (1U << 19)/* AMX_BF16 instruction */ target/i386/kvm/kvm.c +4
-0@@ -539,6 +539,10 @@ uint32_t kvm_arch_get_supported_cpuid(KVMState *s, uint32_t function, ret |=1U <<KVM_HINTS_REALTIME ; } +if (is_tdx_vm ()) { +tdx_get_supported_cpuid (function, index, reg, &ret); + } +return ret; }target/i386/kvm/tdx.c +41
-0@@ -15,11 +15,129 @@#include "qemu/error-report.h" #include "qapi/error.h" #include "qom/object_interfaces.h" +#include "standard-headers/asm-x86/kvm_para.h" #include "sysemu/kvm.h" +#include "sysemu/sysemu.h" #include "hw/i386/x86.h" #include "kvm_i386.h" #include "tdx.h" +#include "../cpu-internal.h" + +#define TDX_SUPPORTED_KVM_FEATURES ((1U << KVM_FEATURE_NOP_IO_DELAY) | \ + (1U << KVM_FEATURE_PV_UNHALT) | \ + (1U << KVM_FEATURE_PV_TLB_FLUSH) | \ + (1U << KVM_FEATURE_PV_SEND_IPI) | \ + (1U << KVM_FEATURE_POLL_CONTROL) | \ + (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ + (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) + +typedef struct KvmTdxCpuidLookup { +uint32_t tdx_fixed0; +uint32_t tdx_fixed1; + +/* +* The CPUID bits that are configurable from the view of TDX module +* but require VMM emulation if configured to enabled by VMM. +* +* For those bits, they cannot be enabled actually if VMM (KVM/QEMU) cannot +* virtualize them. +*/ +uint32_t vmm_fixup; + +bool inducing_ve; +/* +* The maximum supported feature set for given inducing-#VE leaf. +* It's valid only when .inducing_ve is true. +*/ +uint32_t supported_on_ve; +}KvmTdxCpuidLookup ; + +/* +* QEMU maintained TDX CPUID lookup tables, which reflects how CPUIDs are +* virtualized for guest TDs based on "CPUID virtualization" of TDX spec. +* +* Note: +* +* This table will be updated runtime by tdx_caps reported by platform. +*
target/i386/kvm/tdx.c line 73
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+*/ +static KvmTdxCpuidLookup tdx_cpuid_lookup[FEATURE_WORDS ] = { + [FEAT_1_EDX ] = { + .tdx_fixed0 = +BIT (10 )/* Reserved */ |BIT (20 )/* Reserved */ |CPUID_IA64 , + .tdx_fixed1 = +CPUID_MSR |CPUID_PAE |CPUID_MCE |CPUID_APIC | +CPUID_MTRR |CPUID_MCA |CPUID_CLFLUSH |CPUID_DTS , + .vmm_fixup = +CPUID_ACPI |CPUID_PBE ,
CPUID_HT might also be needed here, as it's disabled by QEMU when TD guest only has a single processor core.
Regards,
-Tina
Re: [PATCH v2 08/58] i386/tdx: Adjust the supported CPUID based on TDX restrictions
Quoted 238 lines · Tina Zhang
Hi,
Quoted 235 lines · Xiaoyao Li
According to Chapter "CPUID Virtualization" in TDX module spec, CPUID bits of TD can be classified into 6 types:
------------------------------------------------------------------------
1 | As configured | configurable by VMM, independent of native value;
------------------------------------------------------------------------
2 | As configured | configurable by VMM if the bit is supported natively (if native) | Otherwise it equals as native(0).
------------------------------------------------------------------------
3 | Fixed | fixed to 0/1
------------------------------------------------------------------------
4 | Native | reflect the native value
------------------------------------------------------------------------
5 | Calculated | calculated by TDX module.
------------------------------------------------------------------------
6 | Inducing #VE | get #VE exception
------------------------------------------------------------------------Note:
1. All the configurable XFAM related features and TD attributes related features fall into type #2. And fixed0/1 bits of XFAM and TD attributes fall into type #3.2. For CPUID leaves not listed in "CPUID virtualization Overview" table in TDX module spec, TDX module injects #VE to TDs when those are queried. For this case, TDs can request CPUID emulation from VMM via TDVMCALL and the values are fully controlled by VMM.
Due to TDX module has its own virtualization policy on CPUID bits, it leads to what reported via
KVM_GET_SUPPORTED_CPUIDdiverges from the supported CPUID bits for TDs. In order to keep a consistent CPUID configuration between VMM and TDs. Adjust supported CPUID for TDs based on TDX restrictions.Currently only focus on the CPUID leaves recognized by QEMU's feature_word_info[] that are indexed by a FeatureWord.
Introduce a TDX CPUID lookup table, which maintains 1 entry for each FeatureWord. Each entry has below fields:
- tdx_fixed0/1: The bits that are fixed as 0/1;
- vmm_fixup: The bits that are configurable from the view of TDX module. But they requires emulation of VMM when they are configured as enabled. For those, they are not supported if VMM doesn't report them as supported. So they need be fixed up by checking if VMM supports them.
-
inducing_ve: TD gets #VE when querying this CPUID leaf. The result is totally configurable by VMM.-
supported_on_ve: It's valid only when @inducing_ve is true. It represents the maximum feature set supported that be emulated for TDs.By applying TDX CPUID lookup table and TDX capabilities reported from TDX module, the supported CPUID for TDs can be obtained from following steps:
- get the base of VMM supported feature set;
- if the leaf is not a FeatureWord just return VMM's value without modification;
- if the leaf is an
inducing_vetype, applyingsupported_on_vemask and return;- include all native bits, it covers type #2, #4, and parts of type #1. (it also includes some unsupported bits. The following step will correct it.)
- apply fixed0/1 to it (it covers #3, and rectifies the previous step);
- add configurable bits (it covers the other part of type #1);
- fix the ones in
vmm_fixup;- filter the one has valid .supported field;
(Calculated type is ignored since it's determined at runtime).
Co-developed-by Chenyi Qiang <[email protected]>Signed-off-by Chenyi Qiang <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>---
target/i386/cpu.h | 16 +++target/i386/kvm/kvm.c | 4 + target/i386/kvm/tdx.c | 254 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 2 + 4 files changed, 276 insertions(+)target/i386/cpu.h +6
-0@@ -780,6 +780,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w, /* Support RDFSBASE/RDGSBASE/WRFSBASE/WRGSBASE */ #define CPUID_7_0_EBX_FSGSBASE (1U << 0) +/* Support for TSC adjustment MSR 0x3B */ +#define CPUID_7_0_EBX_TSC_ADJUST (1U << 1) /* Support SGX */ #define CPUID_7_0_EBX_SGX (1U << 2) /* 1st Group of Advanced Bit Manipulation Extensions */@@ -798,8 +800,12 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w, #define CPUID_7_0_EBX_INVPCID (1U << 10) /* Restricted Transactional Memory */ #define CPUID_7_0_EBX_RTM (1U << 11) +/* Cache QoS Monitoring */ +#define CPUID_7_0_EBX_PQM (1U << 12) /* Memory Protection Extension */ #define CPUID_7_0_EBX_MPX (1U << 14) +/* Resource Director Technology Allocation */ +#define CPUID_7_0_EBX_RDT_A (1U << 15) /* AVX-512 Foundation */ #define CPUID_7_0_EBX_AVX512F (1U << 16)/* AVX-512 Doubleword & Quadword Instruction */
target/i386/cpu.h +6
-0@@ -855,10 +861,16 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w, #define CPUID_7_0_ECX_AVX512VNNI (1U << 11) /* Support for VPOPCNT[B,W] and VPSHUFBITQMB */ #define CPUID_7_0_ECX_AVX512BITALG (1U << 12) +/* Intel Total Memory Encryption */ +#define CPUID_7_0_ECX_TME (1U << 13) /* POPCNT for vectors of DW/QW */ #define CPUID_7_0_ECX_AVX512_VPOPCNTDQ (1U << 14) +/* Placeholder for bit 15 */ +#define CPUID_7_0_ECX_FZM (1U << 15) /* 5-level Page Tables */ #define CPUID_7_0_ECX_LA57 (1U << 16) +/* MAWAU for MPX */ +#define CPUID_7_0_ECX_MAWAU (31U << 17) /* Read Processor ID */ #define CPUID_7_0_ECX_RDPID (1U << 22)/* Bus Lock Debug Exception */
target/i386/cpu.h +2
-0@@ -869,6 +881,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w, #define CPUID_7_0_ECX_MOVDIRI (1U << 27) /* Move 64 Bytes as Direct Store Instruction */ #define CPUID_7_0_ECX_MOVDIR64B (1U << 28) +/* ENQCMD and ENQCMDS instructions */ +#define CPUID_7_0_ECX_ENQCMD (1U << 29) /* Support SGX Launch Control */ #define CPUID_7_0_ECX_SGX_LC (1U << 30)/* Protection Keys for Supervisor-mode Pages */
target/i386/cpu.h +2
-0@@ -886,6 +900,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w, #define CPUID_7_0_EDX_SERIALIZE (1U << 14) /* TSX Suspend Load Address Tracking instruction */ #define CPUID_7_0_EDX_TSX_LDTRK (1U << 16) +/* PCONFIG instruction */ +#define CPUID_7_0_EDX_PCONFIG (1U << 18) /* Architectural LBRs */ #define CPUID_7_0_EDX_ARCH_LBR (1U << 19)/*
AMX_BF16instruction */target/i386/kvm/kvm.c +4
-0@@ -539,6 +539,10 @@ uint32_t kvm_arch_get_supported_cpuid(KVMState *s, uint32_t function, ret |= 1U << KVM_HINTS_REALTIME; } + if (is_tdx_vm()) { + tdx_get_supported_cpuid(function, index, reg, &ret); + } + return ret; }target/i386/kvm/tdx.c +11
-0@@ -15,11 +15,129 @@ #include "qemu/error-report.h" #include "qapi/error.h" #include "qom/object_interfaces.h" +#include "standard-headers/asm-x86/kvm_para.h" #include "sysemu/kvm.h" +#include "sysemu/sysemu.h" #include "hw/i386/x86.h" #include "kvm_i386.h" #include "tdx.h" +#include "../cpu-internal.h" + +#define TDX_SUPPORTED_KVM_FEATURES ((1U << KVM_FEATURE_NOP_IO_DELAY) | \ + (1U << KVM_FEATURE_PV_UNHALT) | \ + (1U << KVM_FEATURE_PV_TLB_FLUSH) | \ + (1U << KVM_FEATURE_PV_SEND_IPI) | \ + (1U << KVM_FEATURE_POLL_CONTROL) | \ + (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ + (1U <<
KVM_FEATURE_MSI_EXT_DEST_ID))target/i386/kvm/tdx.c +10
-0+ +typedef struct KvmTdxCpuidLookup { + uint32_t tdx_fixed0; + uint32_t tdx_fixed1; + + /* + * The CPUID bits that are configurable from the view of TDX module + * but require VMM emulation if configured to enabled by VMM. + * + * For those bits, they cannot be enabled actually if VMM(KVM/QEMU) cannot
+ * virtualize them.
+ */
+uint32_tvmm_fixup;target/i386/kvm/tdx.c +11
-0+ + bool inducing_ve; + /* + * The maximum supported feature set for given inducing-#VE leaf. + * It's valid only when .inducing_ve is true. + */ + uint32_t supported_on_ve; +} KvmTdxCpuidLookup; + +/* +* QEMU maintained TDX CPUID lookup tables, which reflects how CPUIDs are
+ * virtualized for guest TDs based on "CPUID virtualization" of TDX spec.
+ *
+ * Note:
+ *
+ * This table will be updated runtime bytdx_capsreported by platform. + *
+ */
+static KvmTdxCpuidLookup tdx_cpuid_lookup[FEATURE_WORDS] = { + [FEAT_1_EDX] = {
+ .tdx_fixed0 =
Quoted 8 lines
Quoted 7 lines
+ BIT(10) /* Reserved */ | BIT(20) /* Reserved */ |
CPUID_IA64,
+ .tdx_fixed1 =
+CPUID_MSR|CPUID_PAE|CPUID_MCE|CPUID_APIC| +CPUID_MTRR|CPUID_MCA|CPUID_CLFLUSH|CPUID_DTS, + .vmm_fixup =
+CPUID_ACPI|CPUID_PBE,
CPUID_HTmight also be needed here, as it's disabled by QEMU when TD guest only has a single processor core.
Add CPUID_HT here seems not correct fix. The root cause is that CPUID_HT is wrongly treated as auto_enabled bit, I will sent a fix separately.
Quoted 3 lines
Regards,
-Tina
PATCH v2 9/58 i386/tdx: Update tdx_cpuid_lookup[].tdx_fixed0/1 by tdx_caps.cpuid_config[]
In reply to Xiaoyao Litdx_cpuid_lookup[].tdx_fixed0/1 is QEMU maintained data which reflects TDX restrictions regrading how some CPUIDs are virtualized by TDX.
It's retrieved from TDX spec. However, TDX may change some fixed fields to configurable in the future. Update tdx_cpuid.lookup[].tdx_fixed0/1 fields by removing the bits that reported from TDX module as configurable. This can adapt with the updated TDX (module) automatically.
--- target/i386/kvm/tdx.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+)
target/i386/kvm/tdx.c +30 -0
@@ -369,6 +369,34 @@ static void get_tdx_capabilities(void) tdx_caps = caps; } +static void update_tdx_cpuid_lookup_by_tdx_caps (void ) +{ +KvmTdxCpuidLookup *entry; +FeatureWordInfo *fi; +uint32_t config; +FeatureWord w; + +/* +* Patch tdx_fixed0/1 by tdx_caps that what TDX module reports as +* configurable is not fixed. +*/ +for (w =0 ; w <FEATURE_WORDS ; w++) { + fi = &feature_word_info[w]; + entry = &tdx_cpuid_lookup[w]; + +if (fi->type !=CPUID_FEATURE_WORD ) { +continue ; + } + + config =tdx_cap_cpuid_config (fi->cpuid.eax, + fi->cpuid.needs_ecx ? fi->cpuid.ecx : ~0u , + fi->cpuid.reg); + + entry->tdx_fixed0 &= ~config; + entry->tdx_fixed1 &= ~config; + } +} +int tdx_kvm_init (MachineState *ms,Error **errp) { TdxGuest *tdx = (TdxGuest *)object_dynamic_cast(OBJECT(ms->cgs),
@@ -378,6 +406,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp)get_tdx_capabilities (); } +update_tdx_cpuid_lookup_by_tdx_caps (); + tdx_guest = tdx;return 0 ;
Signature · 1 line
2.34.1
PATCH v2 10/58 i386/tdx: Integrate tdx_caps->xfam_fixed0/1 into tdx_cpuid_lookup
In reply to Xiaoyao LiKVM requires userspace to pass XFAM configuration via CPUID 0xD leaves.
Convert tdx_caps->xfam_fixed0/1 into corresponding tdx_cpuid_lookup[].tdx_fixed0/1 field of CPUID 0xD leaves. Thus the requirement can be applied naturally.
--- target/i386/cpu.c | 3 --- target/i386/cpu.h | 3 +++ target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++ 3 files changed, 27 insertions(+), 3 deletions(-)
target/i386/cpu.c +0 -3
@@ -1568,9 +1568,6 @@ static const X86RegisterInfo32 x86_reg_info_32[CPU_NB_REGS32] = { };#undef REGISTER -/* CPUID feature bits available in XSS */ -#define CPUID_XSTATE_XSS_MASK (XSTATE_ARCH_LBR_MASK) -ExtSaveArea x86_ext_save_areas[XSAVE_STATE_AREA_COUNT ] = { [XSTATE_FP_BIT ] = {/* x87 FP state component is always enabled if XSAVE is supported */
target/i386/cpu.h +3 -0
@@ -588,6 +588,9 @@ typedef enum X86Seg { XSTATE_Hi16_ZMM_MASK | XSTATE_PKRU_MASK | \ XSTATE_XTILE_CFG_MASK | XSTATE_XTILE_DATA_MASK) +/* CPUID feature bits available in XSS */ +#define CPUID_XSTATE_XSS_MASK (XSTATE_ARCH_LBR_MASK) +/* CPUID feature words */ typedef enum FeatureWord { FEAT_1_EDX, /* CPUID[1].EDX */
target/i386/kvm/tdx.c +24 -0
@@ -395,6 +395,30 @@ static void update_tdx_cpuid_lookup_by_tdx_caps(void) entry->tdx_fixed0 &= ~config; entry->tdx_fixed1 &= ~config; } + +/* +* Because KVM gets XFAM settings via CPUID leaves 0xD, map +* tdx_caps->xfam_fixed{0, 1} into tdx_cpuid_lookup[].tdx_fixed{0, 1}. +* +* Then the enforment applies in tdx_get_configurable_cpuid() naturally. +*/ + tdx_cpuid_lookup[FEAT_XSAVE_XCR0_LO ].tdx_fixed0 = + (uint32_t )~tdx_caps->xfam_fixed0 &CPUID_XSTATE_XCR0_MASK ; + tdx_cpuid_lookup[FEAT_XSAVE_XCR0_LO ].tdx_fixed1 = + (uint32_t )tdx_caps->xfam_fixed1 &CPUID_XSTATE_XCR0_MASK ; + tdx_cpuid_lookup[FEAT_XSAVE_XCR0_HI ].tdx_fixed0 = + (~tdx_caps->xfam_fixed0 &CPUID_XSTATE_XCR0_MASK ) >>32 ; + tdx_cpuid_lookup[FEAT_XSAVE_XCR0_HI ].tdx_fixed1 = + (tdx_caps->xfam_fixed1 &CPUID_XSTATE_XCR0_MASK ) >>32 ; + + tdx_cpuid_lookup[FEAT_XSAVE_XSS_LO ].tdx_fixed0 = + (uint32_t )~tdx_caps->xfam_fixed0 &CPUID_XSTATE_XSS_MASK ; + tdx_cpuid_lookup[FEAT_XSAVE_XSS_LO ].tdx_fixed1 = + (uint32_t )tdx_caps->xfam_fixed1 &CPUID_XSTATE_XSS_MASK ; + tdx_cpuid_lookup[FEAT_XSAVE_XSS_HI ].tdx_fixed0 = + (~tdx_caps->xfam_fixed0 &CPUID_XSTATE_XSS_MASK ) >>32 ; + tdx_cpuid_lookup[FEAT_XSAVE_XSS_HI ].tdx_fixed1 = + (tdx_caps->xfam_fixed1 &CPUID_XSTATE_XSS_MASK ) >>32 ; }int tdx_kvm_init (MachineState *ms,Error **errp)
Signature · 1 line
2.34.1
PATCH v2 11/58 i386/tdx: Integrate tdx_caps->attrs_fixed0/1 to tdx_cpuid_lookup
In reply to Xiaoyao LiSome bits in TD attributes have corresponding CPUID feature bits. Reflect the fixed0/1 restriction on TD attributes to their corresponding CPUID bits in tdx_cpuid_lookup[] as well.
--- target/i386/cpu-internal.h | 9 +++++++++ target/i386/cpu.c | 9 --------- target/i386/cpu.h | 2 ++ target/i386/kvm/tdx.c | 21 +++++++++++++++++++++ 4 files changed, 32 insertions(+), 9 deletions(-)
target/i386/cpu-internal.h +9 -0
@@ -20,6 +20,15 @@ #ifndef I386_CPU_INTERNAL_H #define I386_CPU_INTERNAL_H +typedef struct FeatureMask { + FeatureWord index; + uint64_t mask; +} FeatureMask; + +typedef struct FeatureDep { + FeatureMask from, to; +} FeatureDep; + typedef enum FeatureWordType { CPUID_FEATURE_WORD, MSR_FEATURE_WORD,
target/i386/cpu.c +0 -9
@@ -1439,15 +1439,6 @@ FeatureWordInfo feature_word_info[FEATURE_WORDS] = { }, }; -typedef struct FeatureMask { -FeatureWord index; -uint64_t mask; -}FeatureMask ; - -typedef struct FeatureDep { -FeatureMask from, to; -}FeatureDep ; -static FeatureDep feature_dependencies[] = { { .from = {FEAT_7_0_EDX ,CPUID_7_0_EDX_ARCH_CAPABILITIES },
target/i386/cpu.h +2 -0
@@ -876,6 +876,8 @@ uint64_t x86_cpu_get_supported_feature_word(FeatureWord w,#define CPUID_7_0_ECX_MAWAU (31U << 17)/* Read Processor ID */ #define CPUID_7_0_ECX_RDPID (1U << 22) +/* KeyLocker */ +#define CPUID_7_0_ECX_KeyLocker (1U << 23)/* Bus Lock Debug Exception */ #define CPUID_7_0_ECX_BUS_LOCK_DETECT (1U << 24)/* Cache Line Demote Instruction */
target/i386/kvm/tdx.c +21 -0
@@ -32,6 +32,13 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_ATTRIBUTES_MAX_BITS 64 + +static FeatureMask tdx_attrs_ctrl_fields[TDX_ATTRIBUTES_MAX_BITS ] = { + [30 ] = { .index =FEAT_7_0_ECX , .mask =CPUID_7_0_ECX_PKS }, + [31 ] = { .index =FEAT_7_0_ECX , .mask = CPUID_7_0_ECX_KeyLocker}, +}; + typedef struct KvmTdxCpuidLookup { uint32_t tdx_fixed0; uint32_t tdx_fixed1;
@@ -375,6 +382,8 @@ static void update_tdx_cpuid_lookup_by_tdx_caps(void)FeatureWordInfo *fi;uint32_t config;FeatureWord w; +FeatureMask *fm; +int i;/* * Patch tdx_fixed0/1 by tdx_caps that what TDX module reports as
@@ -396,6 +405,18 @@ static void update_tdx_cpuid_lookup_by_tdx_caps(void) entry->tdx_fixed1 &= ~config; } +for (i =0 ; i <ARRAY_SIZE (tdx_attrs_ctrl_fields); i++) { + fm = &tdx_attrs_ctrl_fields[i]; + +if (tdx_caps->attrs_fixed0 & (1ULL << i)) { + tdx_cpuid_lookup[fm->index].tdx_fixed0 |= fm->mask; + } + +if (tdx_caps->attrs_fixed1 & (1ULL << i)) { + tdx_cpuid_lookup[fm->index].tdx_fixed1 |= fm->mask; + } + } +/* * Because KVM gets XFAM settings via CPUID leaves 0xD, map * tdx_caps->xfam_fixed{0, 1} into tdx_cpuid_lookup[].tdx_fixed{0, 1}.
Signature · 1 line
2.34.1
PATCH v2 12/58 i386/kvm: Move architectural CPUID leaf generation to separate helper
In reply to Xiaoyao LiFrom: Sean Christopherson <[email protected]>
Move the architectural (for lack of a better term) CPUID leaf generation to a separate helper so that the generation code can be reused by TDX, which needs to generate a canonical VM-scoped configuration.
--- target/i386/kvm/kvm.c | 454 +++++++++++++++++++------------------ target/i386/kvm/kvm_i386.h | 3 + 2 files changed, 235 insertions(+), 222 deletions(-)
target/i386/kvm/kvm.c +232 -222
@@ -1799,6 +1799,236 @@ static void kvm_init_nested_state(CPUX86State *env) } } +uint32_t kvm_x86_arch_cpuid (CPUX86State *env,struct kvm_cpuid_entry2 *entries, +uint32_t cpuid_i) +{ +uint32_t limit, i, j; +uint32_t unused; +struct kvm_cpuid_entry2 *c; + +cpu_x86_cpuid (env,0 ,0 , &limit, &unused, &unused, &unused); + +for (i =0 ; i <= limit; i++) { +if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { +fprintf (stderr,"unsupported level value: 0x%x\n" , limit); +abort (); + } + c = &entries[cpuid_i++]; + +switch (i) { +case 2 : { +/* Keep reading function 2 till all the input is received */ +int times; + + c->function = i; + c->flags =KVM_CPUID_FLAG_STATEFUL_FUNC | +KVM_CPUID_FLAG_STATE_READ_NEXT ; +cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); + times = c->eax &0xff ; + +for (j =1 ; j < times; ++j) { +if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { +fprintf (stderr,"cpuid_data is full, no space for " +"cpuid(eax:2):eax & 0xf = 0x%x\n" , times); +abort (); + } + c = &entries[cpuid_i++]; + c->function = i; + c->flags =KVM_CPUID_FLAG_STATEFUL_FUNC ; +cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); + } +break ; + } +case 0x1f : +if (env->nr_dies <2 ) { + cpuid_i--; +break ; + } +/* fallthrough */ +case 4 : +case 0xb : +case 0xd : +for (j =0 ; ; j++) { +if (i ==0xd && j ==64 ) { +break ; + } + + c->function = i; + c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; + c->index = j; +cpu_x86_cpuid (env, i, j, &c->eax, &c->ebx, &c->ecx, &c->edx); + +if (i ==4 && c->eax ==0 ) { +break ; + } +if (i ==0xb && !(c->ecx &0xff00 )) { +break ; + } +if (i ==0x1f && !(c->ecx &0xff00 )) { +break ; + } +if (i ==0xd && c->eax ==0 ) { +continue ; + } +if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { +fprintf (stderr,"cpuid_data is full, no space for " +"cpuid(eax:0x%x,ecx:0x%x)\n" , i, j); +abort (); + } + c = &entries[cpuid_i++]; + } +break ; +case 0x7 : +case 0x12 : +for (j =0 ; ; j++) { + c->function = i; + c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; + c->index = j; +cpu_x86_cpuid (env, i, j, &c->eax, &c->ebx, &c->ecx, &c->edx); + +if (j >1 && (c->eax &0xf ) !=1 ) { +break ; + } + +if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { +fprintf (stderr,"cpuid_data is full, no space for " +"cpuid(eax:0x12,ecx:0x%x)\n" , j); +abort (); + } + c = &entries[cpuid_i++]; + } +break ; +case 0x14 : +case 0x1d : +case 0x1e : { +uint32_t times; + + c->function = i; + c->index =0 ; + c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; +cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); + times = c->eax; + +for (j =1 ; j <= times; ++j) { +if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { +fprintf (stderr,"cpuid_data is full, no space for " +"cpuid(eax:0x%x,ecx:0x%x)\n" , i, j); +abort (); + } + c = &entries[cpuid_i++]; + c->function = i; + c->index = j; + c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; +cpu_x86_cpuid (env, i, j, &c->eax, &c->ebx, &c->ecx, &c->edx); + } +break ; + } +default : + c->function = i; + c->flags =0 ; +cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); +if (!c->eax && !c->ebx && !c->ecx && !c->edx) { +/* +* KVM already returns all zeroes if a CPUID entry is missing, +* so we can omit it and avoid hitting KVM's 80-entry limit. +*/ + cpuid_i--; + } +break ; + } + } + +if (limit >=0x0a ) { +uint32_t eax, edx; + +cpu_x86_cpuid (env,0x0a ,0 , &eax, &unused, &unused, &edx); + + has_architectural_pmu_version = eax &0xff ; +if (has_architectural_pmu_version >0 ) { + num_architectural_pmu_gp_counters = (eax &0xff00 ) >>8 ; + +/* Shouldn't be more than 32, since that's the number of bits +* available in EBX to tell us _which_ counters are available. +* Play it safe. +*/ +if (num_architectural_pmu_gp_counters >MAX_GP_COUNTERS ) { + num_architectural_pmu_gp_counters =MAX_GP_COUNTERS ; + } + +if (has_architectural_pmu_version >1 ) { + num_architectural_pmu_fixed_counters = edx &0x1f ; + +if (num_architectural_pmu_fixed_counters >MAX_FIXED_COUNTERS ) { + num_architectural_pmu_fixed_counters =MAX_FIXED_COUNTERS ; + } + } + } + } + +cpu_x86_cpuid (env,0x80000000 ,0 , &limit, &unused, &unused, &unused); + +for (i =0x80000000 ; i <= limit; i++) { +if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { +fprintf (stderr,"unsupported xlevel value: 0x%x\n" , limit); +abort (); + } + c = &entries[cpuid_i++]; + +switch (i) { +case 0x8000001d : +/* Query for all AMD cache information leaves */ +for (j =0 ; ; j++) { + c->function = i; + c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; + c->index = j; +cpu_x86_cpuid (env, i, j, &c->eax, &c->ebx, &c->ecx, &c->edx); + +if (c->eax ==0 ) { +break ; + } +if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { +fprintf (stderr,"cpuid_data is full, no space for " +"cpuid(eax:0x%x,ecx:0x%x)\n" , i, j); +abort (); + } + c = &entries[cpuid_i++]; + } +break ; +default : + c->function = i; + c->flags =0 ; +cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); +if (!c->eax && !c->ebx && !c->ecx && !c->edx) { +/* +* KVM already returns all zeroes if a CPUID entry is missing, +* so we can omit it and avoid hitting KVM's 80-entry limit. +*/ + cpuid_i--; + } +break ; + } + } + +/* Call Centaur's CPUID instructions they are supported. */ +if (env->cpuid_xlevel2 >0 ) { +cpu_x86_cpuid (env,0xC0000000 ,0 , &limit, &unused, &unused, &unused); + +for (i =0xC0000000 ; i <= limit; i++) { +if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { +fprintf (stderr,"unsupported xlevel2 value: 0x%x\n" , limit); +abort (); + } + c = &entries[cpuid_i++]; + + c->function = i; + c->flags =0 ; +cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); + } + } + +return cpuid_i; +} +int kvm_arch_init_vcpu (CPUState *cs) {struct {
@@ -1815,8 +2045,7 @@ int kvm_arch_init_vcpu(CPUState *cs)X86CPU *cpu =X86_CPU (cs);CPUX86State *env = &cpu->env; -uint32_t limit, i, j, cpuid_i; -uint32_t unused; +uint32_t cpuid_i;struct kvm_cpuid_entry2 *c;uint32_t signature[3 ];int kvm_base =KVM_CPUID_SIGNATURE ;
@@ -1965,8 +2194,6 @@ int kvm_arch_init_vcpu(CPUState *cs) c->edx = env->features[FEAT_KVM_HINTS ]; } -cpu_x86_cpuid (env,0 ,0 , &limit, &unused, &unused, &unused); -if (cpu->kvm_pv_enforce_cpuid) { r =kvm_vcpu_enable_cap (cs,KVM_CAP_ENFORCE_PV_FEATURE_CPUID ,0 ,1 );if (r <0 ) {
@@ -1977,224 +2204,7 @@ int kvm_arch_init_vcpu(CPUState *cs) } } -for (i =0 ; i <= limit; i++) { -if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { -fprintf (stderr,"unsupported level value: 0x%x\n" , limit); -abort (); - } - c = &cpuid_data.entries[cpuid_i++]; - -switch (i) { -case 2 : { -/* Keep reading function 2 till all the input is received */ -int times; - - c->function = i; - c->flags =KVM_CPUID_FLAG_STATEFUL_FUNC | -KVM_CPUID_FLAG_STATE_READ_NEXT ; -cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); - times = c->eax &0xff ; - -for (j =1 ; j < times; ++j) { -if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { -fprintf (stderr,"cpuid_data is full, no space for " -"cpuid(eax:2):eax & 0xf = 0x%x\n" , times); -abort (); - } - c = &cpuid_data.entries[cpuid_i++]; - c->function = i; - c->flags =KVM_CPUID_FLAG_STATEFUL_FUNC ; -cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); - } -break ; - } -case 0x1f : -if (env->nr_dies <2 ) { - cpuid_i--; -break ; - } -/* fallthrough */ -case 4 : -case 0xb : -case 0xd : -for (j =0 ; ; j++) { -if (i ==0xd && j ==64 ) { -break ; - } - - c->function = i; - c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; - c->index = j; -cpu_x86_cpuid (env, i, j, &c->eax, &c->ebx, &c->ecx, &c->edx); - -if (i ==4 && c->eax ==0 ) { -break ; - } -if (i ==0xb && !(c->ecx &0xff00 )) { -break ; - } -if (i ==0x1f && !(c->ecx &0xff00 )) { -break ; - } -if (i ==0xd && c->eax ==0 ) { -continue ; - } -if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { -fprintf (stderr,"cpuid_data is full, no space for " -"cpuid(eax:0x%x,ecx:0x%x)\n" , i, j); -abort (); - } - c = &cpuid_data.entries[cpuid_i++]; - } -break ; -case 0x7 : -case 0x12 : -for (j =0 ; ; j++) { - c->function = i; - c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; - c->index = j; -cpu_x86_cpuid (env, i, j, &c->eax, &c->ebx, &c->ecx, &c->edx); - -if (j >1 && (c->eax &0xf ) !=1 ) { -break ; - } - -if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { -fprintf (stderr,"cpuid_data is full, no space for " -"cpuid(eax:0x12,ecx:0x%x)\n" , j); -abort (); - } - c = &cpuid_data.entries[cpuid_i++]; - } -break ; -case 0x14 : -case 0x1d : -case 0x1e : { -uint32_t times; - - c->function = i; - c->index =0 ; - c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; -cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); - times = c->eax; - -for (j =1 ; j <= times; ++j) { -if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { -fprintf (stderr,"cpuid_data is full, no space for " -"cpuid(eax:0x%x,ecx:0x%x)\n" , i, j); -abort (); - } - c = &cpuid_data.entries[cpuid_i++]; - c->function = i; - c->index = j; - c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; -cpu_x86_cpuid (env, i, j, &c->eax, &c->ebx, &c->ecx, &c->edx); - } -break ; - } -default : - c->function = i; - c->flags =0 ; -cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); -if (!c->eax && !c->ebx && !c->ecx && !c->edx) { -/* -* KVM already returns all zeroes if a CPUID entry is missing, -* so we can omit it and avoid hitting KVM's 80-entry limit. -*/ - cpuid_i--; - } -break ; - } - } - -if (limit >=0x0a ) { -uint32_t eax, edx; - -cpu_x86_cpuid (env,0x0a ,0 , &eax, &unused, &unused, &edx); - - has_architectural_pmu_version = eax &0xff ; -if (has_architectural_pmu_version >0 ) { - num_architectural_pmu_gp_counters = (eax &0xff00 ) >>8 ; - -/* Shouldn't be more than 32, since that's the number of bits -* available in EBX to tell us _which_ counters are available. -* Play it safe. -*/ -if (num_architectural_pmu_gp_counters >MAX_GP_COUNTERS ) { - num_architectural_pmu_gp_counters =MAX_GP_COUNTERS ; - } - -if (has_architectural_pmu_version >1 ) { - num_architectural_pmu_fixed_counters = edx &0x1f ; - -if (num_architectural_pmu_fixed_counters >MAX_FIXED_COUNTERS ) { - num_architectural_pmu_fixed_counters =MAX_FIXED_COUNTERS ; - } - } - } - } - -cpu_x86_cpuid (env,0x80000000 ,0 , &limit, &unused, &unused, &unused); - -for (i =0x80000000 ; i <= limit; i++) { -if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { -fprintf (stderr,"unsupported xlevel value: 0x%x\n" , limit); -abort (); - } - c = &cpuid_data.entries[cpuid_i++]; - -switch (i) { -case 0x8000001d : -/* Query for all AMD cache information leaves */ -for (j =0 ; ; j++) { - c->function = i; - c->flags =KVM_CPUID_FLAG_SIGNIFCANT_INDEX ; - c->index = j; -cpu_x86_cpuid (env, i, j, &c->eax, &c->ebx, &c->ecx, &c->edx); - -if (c->eax ==0 ) { -break ; - } -if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { -fprintf (stderr,"cpuid_data is full, no space for " -"cpuid(eax:0x%x,ecx:0x%x)\n" , i, j); -abort (); - } - c = &cpuid_data.entries[cpuid_i++]; - } -break ; -default : - c->function = i; - c->flags =0 ; -cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); -if (!c->eax && !c->ebx && !c->ecx && !c->edx) { -/* -* KVM already returns all zeroes if a CPUID entry is missing, -* so we can omit it and avoid hitting KVM's 80-entry limit. -*/ - cpuid_i--; - } -break ; - } - } - -/* Call Centaur's CPUID instructions they are supported. */ -if (env->cpuid_xlevel2 >0 ) { -cpu_x86_cpuid (env,0xC0000000 ,0 , &limit, &unused, &unused, &unused); - -for (i =0xC0000000 ; i <= limit; i++) { -if (cpuid_i ==KVM_MAX_CPUID_ENTRIES ) { -fprintf (stderr,"unsupported xlevel2 value: 0x%x\n" , limit); -abort (); - } - c = &cpuid_data.entries[cpuid_i++]; - - c->function = i; - c->flags =0 ; -cpu_x86_cpuid (env, i,0 , &c->eax, &c->ebx, &c->ecx, &c->edx); - } - } - + cpuid_i =kvm_x86_arch_cpuid (env, cpuid_data.entries, cpuid_i); cpuid_data.cpuid.nent = cpuid_i; if (((env->cpuid_version >> 8)&0xF) >= 6
target/i386/kvm/kvm_i386.h +3 -0
@@ -26,6 +26,9 @@#define kvm_ioapic_in_kernel () \ (kvm_irqchip_in_kernel() && !kvm_irqchip_is_split()) +uint32_t kvm_x86_arch_cpuid (CPUX86State *env,struct kvm_cpuid_entry2 *entries, +uint32_t cpuid_i); +#else #define kvm_pit_in_kernel () 0
Signature · 1 line
2.34.1
PATCH v2 13/58 kvm: Introduce kvm_arch_pre_create_vcpu()
In reply to Xiaoyao LiIntroduce kvm_arch_pre_create_vcpu(), to perform arch-dependent work prior to create any vcpu. This is for i386 TDX because it needs call TDX_INIT_VM before creating any vcpu.
--- accel/kvm/kvm-all.c | 12 ++++++++++++ include/sysemu/kvm.h | 1 + 2 files changed, 13 insertions(+)
accel/kvm/kvm-all.c +12 -0
@@ -422,6 +422,11 @@ static int kvm_get_vcpu(KVMState *s, unsigned long vcpu_id)return kvm_vm_ioctl (s,KVM_CREATE_VCPU , (void *)vcpu_id); } +int __attribute__ ((weak))kvm_arch_pre_create_vcpu (CPUState *cpu) +{ +return 0 ; +} +int kvm_init_vcpu (CPUState *cpu,Error **errp) {KVMState *s = kvm_state;
@@ -430,6 +435,13 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)trace_kvm_init_vcpu (cpu->cpu_index,kvm_arch_vcpu_id (cpu)); + ret =kvm_arch_pre_create_vcpu (cpu); +if (ret <0 ) { +error_setg_errno (errp, -ret,"%s: kvm_arch_pre_create_vcpu() failed" , + __func__); + goto err; + } + ret =kvm_get_vcpu (s,kvm_arch_vcpu_id (cpu));if (ret <0 ) { error_setg_errno(errp, -ret, "kvm_init_vcpu: kvm_get_vcpu failed (%lu)",
include/sysemu/kvm.h +1 -0
@@ -371,6 +371,7 @@ int kvm_arch_put_registers(CPUState *cpu, int level);int kvm_arch_init (MachineState *ms,KVMState *s); +int kvm_arch_pre_create_vcpu (CPUState *cpu);int kvm_arch_init_vcpu (CPUState *cpu);int kvm_arch_destroy_vcpu (CPUState *cpu);
Signature · 1 line
2.34.1
Re: [PATCH v2 13/58] kvm: Introduce kvm_arch_pre_create_vcpu()
Quoted 28 lines · Xiaoyao Li
Introduce
kvm_arch_pre_create_vcpu(), to perform arch-dependent work prior to create any vcpu. This is for i386 TDX because it needs callTDX_INIT_VMbefore creating any vcpu.Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- accel/kvm/kvm-all.c | 12 ++++++++++++ include/sysemu/kvm.h | 1 + 2 files changed, 13 insertions(+)accel/kvm/kvm-all.c +5
-0@@ -422,6 +422,11 @@ static int kvm_get_vcpu(KVMState *s, unsigned long vcpu_id)return kvm_vm_ioctl (s,KVM_CREATE_VCPU , (void *)vcpu_id); } +int __attribute__ ((weak))kvm_arch_pre_create_vcpu (CPUState *cpu) +{ +return 0 ; +} +int kvm_init_vcpu (CPUState *cpu,Error **errp) {KVMState *s = kvm_state;
accel/kvm/kvm-all.c int kvm_init_vcpu(CPUState *cpu, Error **errp) · line 441
Quoted 7 lines
accel/kvm/kvm-all.c +4
-0@@ -430,6 +435,13 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)trace_kvm_init_vcpu (cpu->cpu_index,kvm_arch_vcpu_id (cpu)); + ret =kvm_arch_pre_create_vcpu (cpu); +if (ret <0 ) { +error_setg_errno (errp, -ret,"%s: kvm_arch_pre_create_vcpu() failed" , + __func__);
Don't report generic error messages here, when kvm_arch_pre_create_vcpu can provide a better error - pass the 'errp' into the kvm_arch_pre_create_vcpu method.
Quoted 21 lines
accel/kvm/kvm-all.c +3
-0+ goto err; + } + ret =kvm_get_vcpu (s,kvm_arch_vcpu_id (cpu));if (ret <0 ) { error_setg_errno(errp, -ret, "kvm_init_vcpu: kvm_get_vcpu failed (%lu)",include/sysemu/kvm.h +1
-0@@ -371,6 +371,7 @@ int kvm_arch_put_registers(CPUState *cpu, int level);int kvm_arch_init (MachineState *ms,KVMState *s); +int kvm_arch_pre_create_vcpu (CPUState *cpu);int kvm_arch_init_vcpu (CPUState *cpu);int kvm_arch_destroy_vcpu (CPUState *cpu);Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 13/58] kvm: Introduce kvm_arch_pre_create_vcpu()
In reply to Xiaoyao LiQuoted 16 lines · Xiaoyao Li
Introduce
kvm_arch_pre_create_vcpu(), to perform arch-dependent work prior to create any vcpu. This is for i386 TDX because it needs callTDX_INIT_VMbefore creating any vcpu.Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- accel/kvm/kvm-all.c | 12 ++++++++++++ include/sysemu/kvm.h | 1 + 2 files changed, 13 insertions(+)accel/kvm/kvm-all.c +0
-0
accel/kvm/kvm-all.c static int kvm_get_vcpu(KVMState *s, unsigned long vcpu_id) · line 428
Quoted 7 lines
accel/kvm/kvm-all.c +4
-0@@ -422,6 +422,11 @@ static int kvm_get_vcpu(KVMState *s, unsigned long vcpu_id)return kvm_vm_ioctl (s,KVM_CREATE_VCPU , (void *)vcpu_id); } +int __attribute__ ((weak))kvm_arch_pre_create_vcpu (CPUState *cpu) +{ +return 0 ; +}
kvm_arch_init_vcpu() is implemented for each arch. Why not use the same approach here?
Quoted 29 lines
accel/kvm/kvm-all.c +7
-0int kvm_init_vcpu (CPUState *cpu,Error **errp) {KVMState *s = kvm_state;@@ -430,6 +435,13 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)trace_kvm_init_vcpu (cpu->cpu_index,kvm_arch_vcpu_id (cpu)); + ret =kvm_arch_pre_create_vcpu (cpu); +if (ret <0 ) { +error_setg_errno (errp, -ret,"%s: kvm_arch_pre_create_vcpu() failed" , + __func__); + goto err; + } + ret =kvm_get_vcpu (s,kvm_arch_vcpu_id (cpu));if (ret <0 ) { error_setg_errno(errp, -ret, "kvm_init_vcpu: kvm_get_vcpu failed (%lu)",include/sysemu/kvm.h +1
-0@@ -371,6 +371,7 @@ int kvm_arch_put_registers(CPUState *cpu, int level);int kvm_arch_init (MachineState *ms,KVMState *s); +int kvm_arch_pre_create_vcpu (CPUState *cpu);int kvm_arch_init_vcpu (CPUState *cpu);int kvm_arch_destroy_vcpu (CPUState *cpu);
Re: [PATCH v2 13/58] kvm: Introduce kvm_arch_pre_create_vcpu()
Quoted 12 lines · Philippe Mathieu-Daudé
Quoted 11 lines · Xiaoyao Li
Introduce
kvm_arch_pre_create_vcpu(), to perform arch-dependent work prior to create any vcpu. This is for i386 TDX because it needs callTDX_INIT_VMbefore creating any vcpu.Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>---
accel/kvm/kvm-all.c | 12 ++++++++++++
include/sysemu/kvm.h| 1 +2 files changed, 13 insertions(+)
Quoted 15 lines
accel/kvm/kvm-all.c static int kvm_get_vcpu(KVMState *s, unsigned · line 428
Quoted 12 lines
accel/kvm/kvm-all.c +4
-0@@ -422,6 +422,11 @@ static int kvm_get_vcpu(KVMState *s, unsigned long vcpu_id) return kvm_vm_ioctl(s, KVM_CREATE_VCPU, (void *)vcpu_id); } +int __attribute__ ((weak))kvm_arch_pre_create_vcpu (CPUState *cpu) +{ + return 0; +}
kvm_arch_init_vcpu()is implemented for each arch. Why not use the same approach here?
Because only x86 needs it currently, for TDX. Other arches don't require an implementation.
If don't provide the _weak_ function, it needs to implement the empty function (justing return 0) in all the other arches just as the placeholder. If QEMU community prefers this approach, I can change to it in next version.
Quoted 26 lines
Quoted 25 lines
accel/kvm/kvm-all.c +7
-0int kvm_init_vcpu(CPUState *cpu, Error **errp) { KVMState *s = kvm_state;@@ -430,6 +435,13 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp) trace_kvm_init_vcpu(cpu->cpu_index, kvm_arch_vcpu_id(cpu)); + ret = kvm_arch_pre_create_vcpu(cpu); + if (ret < 0) { + error_setg_errno(errp, -ret, "%s: kvm_arch_pre_create_vcpu() failed", + __func__); + goto err; + } + ret = kvm_get_vcpu(s, kvm_arch_vcpu_id(cpu)); if (ret < 0) { error_setg_errno(errp, -ret, "kvm_init_vcpu: kvm_get_vcpu failed (%lu)",include/sysemu/kvm.h +1
-0@@ -371,6 +371,7 @@ int kvm_arch_put_registers(CPUState *cpu, int level); int kvm_arch_init(MachineState *ms, KVMState *s); +int kvm_arch_pre_create_vcpu (CPUState *cpu); int kvm_arch_init_vcpu(CPUState *cpu); int kvm_arch_destroy_vcpu(CPUState *cpu);
Re: [PATCH v2 13/58] kvm: Introduce kvm_arch_pre_create_vcpu()
Quoted 29 lines · Xiaoyao Li
Quoted 12 lines · Philippe Mathieu-Daudé
Quoted 11 lines · Xiaoyao Li
Introduce
kvm_arch_pre_create_vcpu(), to perform arch-dependent work prior to create any vcpu. This is for i386 TDX because it needs callTDX_INIT_VMbefore creating any vcpu.Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>---
accel/kvm/kvm-all.c | 12 ++++++++++++
include/sysemu/kvm.h| 1 +2 files changed, 13 insertions(+)Quoted 15 lines
accel/kvm/kvm-all.c static int kvm_get_vcpu(KVMState *s, unsigned · line 428
Quoted 12 lines
accel/kvm/kvm-all.c +4
-0@@ -422,6 +422,11 @@ static int kvm_get_vcpu(KVMState *s, unsigned long vcpu_id) return kvm_vm_ioctl(s, KVM_CREATE_VCPU, (void *)vcpu_id); } +int __attribute__ ((weak))kvm_arch_pre_create_vcpu (CPUState *cpu) +{ + return 0; +}
kvm_arch_init_vcpu()is implemented for each arch. Why not use the same approach here?
Quoted 6 lines
Because only x86 needs it currently, for TDX. Other arches don't require an implementation.
If don't provide the _weak_ function, it needs to implement the empty function (justing return 0) in all the other arches just as the placeholder. If QEMU community prefers this approach, I can change to it in next version.
Alternative is to move the hook to x86 specific function, not common kvm function. With my quick grepping, x86_cpus_init() or x86_cpu_realizefn().
Signature · 1 line
Isaku Yamahata <[email protected]>
PATCH v2 14/58 i386/tdx: Initialize TDX before creating TD vcpus
In reply to Xiaoyao LiInvoke KVM_TDX_INIT in kvm_arch_pre_create_vcpu() that KVM_TDX_INIT configures global TD configurations, e.g. the canonical CPUID config, and must be executed prior to creating vCPUs.
Use kvm_x86_arch_cpuid() to setup the CPUID settings for TDX VM.
Note, this doesn't address the fact that QEMU may change the CPUID configuration when creating vCPUs, i.e. punts on refactoring QEMU to provide a stable CPUID config prior to kvm_arch_init().
--- accel/kvm/kvm-all.c | 9 +++++++- target/i386/kvm/kvm.c | 8 +++++++ target/i386/kvm/tdx-stub.c | 5 +++++ target/i386/kvm/tdx.c | 45 ++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 4 ++++ 5 files changed, 70 insertions(+), 1 deletion(-)
accel/kvm/kvm-all.c +8 -1
@@ -435,10 +435,17 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)trace_kvm_init_vcpu (cpu->cpu_index,kvm_arch_vcpu_id (cpu)); +/* +* tdx_pre_create_vcpu() may call cpu_x86_cpuid(). It in turn may call +* kvm_vm_ioctl(). Set cpu->kvm_state in advance to avoid NULL pointer +* dereference. +*/ + cpu->kvm_state = s; ret =kvm_arch_pre_create_vcpu (cpu);if (ret <0 ) {error_setg_errno (errp, -ret,"%s: kvm_arch_pre_create_vcpu() failed" , __func__); + cpu->kvm_state =NULL ; goto err; }
@@ -446,11 +453,11 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)if (ret <0 ) {error_setg_errno (errp, -ret,"kvm_init_vcpu: kvm_get_vcpu failed (%lu)" ,kvm_arch_vcpu_id (cpu)); + cpu->kvm_state =NULL ; goto err; } cpu->kvm_fd = ret; - cpu->kvm_state = s; cpu->vcpu_dirty = true; cpu->dirty_pages =0 ; cpu->throttle_us_per_full =0 ;
target/i386/kvm/kvm.c +8 -0
@@ -2331,6 +2331,14 @@ int kvm_arch_init_vcpu(CPUState *cs)return r; } +int kvm_arch_pre_create_vcpu (CPUState *cpu) +{ +if (is_tdx_vm ()) +return tdx_pre_create_vcpu (cpu); + +return 0 ; +} +int kvm_arch_destroy_vcpu (CPUState *cs) {X86CPU *cpu =X86_CPU (cs);
target/i386/kvm/tdx-stub.c +5 -0
@@ -6,3 +6,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp) {return -EINVAL ; } + +int tdx_pre_create_vcpu (CPUState *cpu) +{ +return -EINVAL ; +}
target/i386/kvm/tdx.c +45 -0
57 lines not on this page · raw message
target/i386/kvm/tdx.h +4 -0
@@ -17,6 +17,9 @@ typedef struct TdxGuestClass {typedef struct TdxGuest {ConfidentialGuestSupport parent_obj; +QemuMutex lock; + +bool initialized;uint64_t attributes;/* TD attributes */ }TdxGuest ;
@@ -29,5 +32,6 @@ bool is_tdx_vm(void);int tdx_kvm_init (MachineState *ms,Error **errp);void tdx_get_supported_cpuid (uint32_t function,uint32_t index,int reg,uint32_t *ret); +int tdx_pre_create_vcpu (CPUState *cpu);#endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
Re: [PATCH v2 14/58] i386/tdx: Initialize TDX before creating TD vcpus
Quoted 60 lines · Xiaoyao Li
Invoke
KVM_TDX_INITinkvm_arch_pre_create_vcpu()thatKVM_TDX_INITconfigures global TD configurations, e.g. the canonical CPUID config, and must be executed prior to creating vCPUs.Use
kvm_x86_arch_cpuid()to setup the CPUID settings for TDX VM.Note, this doesn't address the fact that QEMU may change the CPUID configuration when creating vCPUs, i.e. punts on refactoring QEMU to provide a stable CPUID config prior to
kvm_arch_init().Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- accel/kvm/kvm-all.c | 9 +++++++- target/i386/kvm/kvm.c | 8 +++++++ target/i386/kvm/tdx-stub.c | 5 +++++ target/i386/kvm/tdx.c | 45 ++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 4 ++++ 5 files changed, 70 insertions(+), 1 deletion(-)accel/kvm/kvm-all.c +8
-1@@ -435,10 +435,17 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)trace_kvm_init_vcpu (cpu->cpu_index,kvm_arch_vcpu_id (cpu)); +/* +* tdx_pre_create_vcpu() may call cpu_x86_cpuid(). It in turn may call +* kvm_vm_ioctl(). Set cpu->kvm_state in advance to avoid NULL pointer +* dereference. +*/ + cpu->kvm_state = s; ret =kvm_arch_pre_create_vcpu (cpu);if (ret <0 ) {error_setg_errno (errp, -ret,"%s: kvm_arch_pre_create_vcpu() failed" , __func__); + cpu->kvm_state =NULL ; goto err; }@@ -446,11 +453,11 @@ int kvm_init_vcpu(CPUState *cpu, Error **errp)if (ret <0 ) {error_setg_errno (errp, -ret,"kvm_init_vcpu: kvm_get_vcpu failed (%lu)" ,kvm_arch_vcpu_id (cpu)); + cpu->kvm_state =NULL ; goto err; } cpu->kvm_fd = ret; - cpu->kvm_state = s; cpu->vcpu_dirty = true; cpu->dirty_pages =0 ; cpu->throttle_us_per_full =0 ;target/i386/kvm/kvm.c +0
-0
target/i386/kvm/kvm.c int kvm_arch_init_vcpu(CPUState *cs) · line 2337
Quoted 7 lines
target/i386/kvm/kvm.c +4
-0@@ -2331,6 +2331,14 @@ int kvm_arch_init_vcpu(CPUState *cs)return r; } +int kvm_arch_pre_create_vcpu (CPUState *cpu) +{ +if (is_tdx_vm ()) +return tdx_pre_create_vcpu (cpu);
Curly braces needed for coding style - run 'scripts/checkpatch.pl' to validate it.
Quoted 25 lines
target/i386/kvm/kvm.c +4
-0+ +return 0 ; +} +int kvm_arch_destroy_vcpu (CPUState *cs) {X86CPU *cpu =X86_CPU (cs);target/i386/kvm/tdx-stub.c +5
-0@@ -6,3 +6,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp) {return -EINVAL ; } + +int tdx_pre_create_vcpu (CPUState *cpu) +{ +return -EINVAL ; +}target/i386/kvm/tdx.c +0
-0
target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 461
Quoted 4 lines
target/i386/kvm/tdx.c +1
-0@@ -458,6 +458,49 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return 0 ; } +int tdx_pre_create_vcpu (CPUState *cpu)
Add 'Error **errp' to this method
Quoted 8 lines
target/i386/kvm/tdx.c +8
-0+{ +MachineState *ms =MACHINE (qdev_get_machine ()); +X86CPU *x86cpu =X86_CPU (cpu); +CPUX86State *env = &x86cpu->env; +struct kvm_tdx_init_vm *init_vm; +int r =0 ; + +qemu_mutex_lock (&tdx_guest->lock);
target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 479
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+if (tdx_guest->initialized) { + goto out; + } + + init_vm =g_malloc0 (sizeof (struct kvm_tdx_init_vm ) + +sizeof (struct kvm_cpuid_entry2 ) *KVM_MAX_CPUID_ENTRIES ); + + r =kvm_vm_enable_cap (kvm_state,KVM_CAP_MAX_VCPUS ,0 , ms->smp.cpus); +if (r <0 ) { +error_report ("Unable to set MAX VCPUS to %d" , ms->smp.cpus);
Use error_setg / error_setg_errno in this method.
Quoted 60 lines
target/i386/kvm/tdx.c +26
-0+ goto out_free; + } + + init_vm->cpuid.nent =kvm_x86_arch_cpuid (env, init_vm->cpuid.entries,0 ); + + init_vm->attributes = tdx_guest->attributes; + +do { + r =tdx_vm_ioctl (KVM_TDX_INIT_VM ,0 , init_vm); + }while (r == -EAGAIN ); +if (r <0 ) { +error_report ("KVM_TDX_INIT_VM failed %s" ,strerror (-r)); + goto out_free; + } + + tdx_guest->initialized = true; + +out_free: +g_free (init_vm); +out: +qemu_mutex_unlock (&tdx_guest->lock); +return r; +} +/* tdx guest */ OBJECT_DEFINE_TYPE_WITH_INTERFACES(TdxGuest, tdx_guest,@@ -470,6 +513,8 @@ static void tdx_guest_init(Object *obj) {TdxGuest *tdx =TDX_GUEST (obj); +qemu_mutex_init (&tdx->lock); + tdx->attributes =0 ; }target/i386/kvm/tdx.h +4
-0@@ -17,6 +17,9 @@ typedef struct TdxGuestClass {typedef struct TdxGuest {ConfidentialGuestSupport parent_obj; +QemuMutex lock; + +bool initialized;uint64_t attributes;/* TD attributes */ }TdxGuest ;@@ -29,5 +32,6 @@ bool is_tdx_vm(void);int tdx_kvm_init (MachineState *ms,Error **errp);void tdx_get_supported_cpuid (uint32_t function,uint32_t index,int reg,uint32_t *ret); +int tdx_pre_create_vcpu (CPUState *cpu);#endif /* QEMU_I386_TDX_H */ Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
PATCH v2 15/58 i386/tdx: Add property sept-ve-disable for tdx-guest object
In reply to Xiaoyao LiBit 28 of TD attribute, named SEPT_VE_DISABLE. When set to 1, it disables EPT violation conversion to #VE on guest TD access of PENDING pages.
Some guest OS (e.g., Linux TD guest) may require this bit as 1. Otherwise refuse to boot.
Add sept-ve-disable property for tdx-guest object, for user to configure this bit.
--- qapi/qom.json | 4 +++- target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++ 2 files changed, 27 insertions(+), 1 deletion(-)
qapi/qom.json +3 -1
@@ -871,10 +871,12 @@ # # Properties for tdx-guest objects. # +# @sept-ve-disable: bit 28 of TD attributes (default: 0) +# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', - 'data': { }} + 'data': { '*sept-ve-disable': 'bool' } } ## # @ThreadContextProperties:
target/i386/kvm/tdx.c +24 -0
@@ -32,6 +32,8 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28) +#define TDX_ATTRIBUTES_MAX_BITS 64static FeatureMask tdx_attrs_ctrl_fields[TDX_ATTRIBUTES_MAX_BITS ] = {
@@ -501,6 +503,24 @@ out:return r; } +static bool tdx_guest_get_sept_ve_disable (Object *obj,Error **errp) +{ +TdxGuest *tdx =TDX_GUEST (obj); + +return !!(tdx->attributes &TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE ); +} + +static void tdx_guest_set_sept_ve_disable (Object *obj,bool value,Error **errp) +{ +TdxGuest *tdx =TDX_GUEST (obj); + +if (value) { + tdx->attributes |=TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE ; + }else { + tdx->attributes &= ~TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE ; + } +} +/* tdx guest */ OBJECT_DEFINE_TYPE_WITH_INTERFACES(TdxGuest, tdx_guest,
@@ -516,6 +536,10 @@ static void tdx_guest_init(Object *obj)qemu_mutex_init (&tdx->lock); tdx->attributes =0 ; + +object_property_add_bool (obj,"sept-ve-disable" , + tdx_guest_get_sept_ve_disable, + tdx_guest_set_sept_ve_disable); }static void tdx_guest_finalize (Object *obj)
Signature · 1 line
2.34.1
Re: [PATCH v2 15/58] i386/tdx: Add property sept-ve-disable for tdx-guest object
Quoted 21 lines · Xiaoyao Li
Bit 28 of TD attribute, named
SEPT_VE_DISABLE. When set to 1, it disables EPT violation conversion to #VE on guest TD access of PENDING pages.Some guest OS (e.g., Linux TD guest) may require this bit as 1. Otherwise refuse to boot.
Add sept-ve-disable property for tdx-guest object, for user to configure this bit.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- qapi/qom.json | 4 +++- target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++ 2 files changed, 27 insertions(+), 1 deletion(-)qapi/qom.json +0
-0
qapi/qom.json line 874
Quoted 4 lines
qapi/qom.json +1
-0@@ -871,10 +871,12 @@ # # Properties for tdx-guest objects. # +# @sept-ve-disable: bit 28 of TD attributes (default: 0)
This description isn't very useful as it forces the user to go off and read the TDX specification to find out what bit 28 means. You've got a more useful description in the commit message, so please use that in the docs too. eg something like this
@sept-ve-disable: toggle bit 28 of TD attributes to control disabling
of EPT violation conversion to #VE on guest TD access of PENDING pages. Some guest OS (e.g. Linux TD guest) may require this set, otherwise they refuse to boot.
Quoted 61 lines
qapi/qom.json +2
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', - 'data': { }} + 'data': { '*sept-ve-disable': 'bool' } } ## # @ThreadContextProperties:target/i386/kvm/tdx.c +24
-0@@ -32,6 +32,8 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28) +#define TDX_ATTRIBUTES_MAX_BITS 64static FeatureMask tdx_attrs_ctrl_fields[TDX_ATTRIBUTES_MAX_BITS ] = {@@ -501,6 +503,24 @@ out:return r; } +static bool tdx_guest_get_sept_ve_disable (Object *obj,Error **errp) +{ +TdxGuest *tdx =TDX_GUEST (obj); + +return !!(tdx->attributes &TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE ); +} + +static void tdx_guest_set_sept_ve_disable (Object *obj,bool value,Error **errp) +{ +TdxGuest *tdx =TDX_GUEST (obj); + +if (value) { + tdx->attributes |=TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE ; + }else { + tdx->attributes &= ~TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE ; + } +} +/* tdx guest */ OBJECT_DEFINE_TYPE_WITH_INTERFACES(TdxGuest, tdx_guest,@@ -516,6 +536,10 @@ static void tdx_guest_init(Object *obj)qemu_mutex_init (&tdx->lock); tdx->attributes =0 ; + +object_property_add_bool (obj,"sept-ve-disable" , + tdx_guest_get_sept_ve_disable, + tdx_guest_set_sept_ve_disable); }static void tdx_guest_finalize (Object *obj)Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 15/58] i386/tdx: Add property sept-ve-disable for tdx-guest object
Quoted 17 lines · Daniel P. Berrangé
Quoted 16 lines · Xiaoyao Li
Bit 28 of TD attribute, named
SEPT_VE_DISABLE. When set to 1, it disables EPT violation conversion to #VE on guest TD access of PENDING pages.Some guest OS (e.g., Linux TD guest) may require this bit as 1. Otherwise refuse to boot.
Add sept-ve-disable property for tdx-guest object, for user to configure this bit.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- qapi/qom.json | 4 +++- target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++ 2 files changed, 27 insertions(+), 1 deletion(-)
Quoted 12 lines
qapi/qom.json line 874
Quoted 9 lines
qapi/qom.json +1
-0@@ -871,10 +871,12 @@ # # Properties for tdx-guest objects. # +# @sept-ve-disable: bit 28 of TD attributes (default: 0)This description isn't very useful as it forces the user to go off and read the TDX specification to find out what bit 28 means. You've got a
Seconded.
Quoted 8 lines
more useful description in the commit message, so please use that in the docs too. eg something like this
@sept-ve-disable: toggle bit 28 of TD attributes to control disabling
of EPT violation conversion to #VE on guest TD access of PENDING pages. Some guest OS (e.g. Linux TD guest) may require this set, otherwise they refuse to boot.
But please format like
# @sept-ve-disable: toggle bit 28 of TD attributes to control disabling
# of EPT violation conversion to #VE on guest TD access of PENDING # pages. Some guest OS (e.g. Linux TD guest) may require this to # be set, otherwise they refuse to boot.
to blend in with recent commit a937b6aa739 (qapi: Reformat doc comments to conform to current conventions).
Quoted 9 lines
Quoted 9 lines
qapi/qom.json +2
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', - 'data': { }} + 'data': { '*sept-ve-disable': 'bool' } } ## # @ThreadContextProperties:
[...]
Re: [PATCH v2 15/58] i386/tdx: Add property sept-ve-disable for tdx-guest object
Quoted 43 lines · Markus Armbruster
Quoted 17 lines · Daniel P. Berrangé
Quoted 16 lines · Xiaoyao Li
Bit 28 of TD attribute, named
SEPT_VE_DISABLE. When set to 1, it disables EPT violation conversion to #VE on guest TD access of PENDING pages.Some guest OS (e.g., Linux TD guest) may require this bit as 1. Otherwise refuse to boot.
Add sept-ve-disable property for tdx-guest object, for user to configure this bit.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- qapi/qom.json | 4 +++- target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++ 2 files changed, 27 insertions(+), 1 deletion(-)Quoted 12 lines
qapi/qom.json line 874
Quoted 9 lines
qapi/qom.json +1
-0@@ -871,10 +871,12 @@ # # Properties for tdx-guest objects. # +# @sept-ve-disable: bit 28 of TD attributes (default: 0)This description isn't very useful as it forces the user to go off and read the TDX specification to find out what bit 28 means. You've got a
Seconded.
Quoted 8 lines
more useful description in the commit message, so please use that in the docs too. eg something like this
@sept-ve-disable: toggle bit 28 of TD attributes to control disabling
of EPT violation conversion to #VE on guest TD access of PENDING pages. Some guest OS (e.g. Linux TD guest) may require this set, otherwise they refuse to boot.
Quoted 7 lines
But please format like
# @sept-ve-disable: toggle bit 28 of TD attributes to control disabling
# of EPT violation conversion to #VE on guest TD access of PENDING # pages. Some guest OS (e.g. Linux TD guest) may require this to # be set, otherwise they refuse to boot.
Thank you, Daniel and Markus.
Will use above in the next version.
Quoted 15 lines
to blend in with recent commit
a937b6aa739(qapi: Reformat doc comments to conform to current conventions).Quoted 9 lines
Quoted 9 lines
qapi/qom.json +2
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', - 'data': { }} + 'data': { '*sept-ve-disable': 'bool' } } ## # @ThreadContextProperties:[...]
PATCH v2 16/58 i386/tdx: Make sept_ve_disable set by default
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
For TDX KVM use case, Linux guest is the most major one. It requires sept_ve_disable set. Make it default for the main use case. For other use case, it can be enabled/disabled via qemu command line.
--- target/i386/kvm/tdx.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
target/i386/kvm/tdx.c +1 -1
@@ -535,7 +535,7 @@ static void tdx_guest_init(Object *obj)qemu_mutex_init (&tdx->lock); - tdx->attributes =0 ; + tdx->attributes =TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE ; object_property_add_bool(obj, "sept-ve-disable", tdx_guest_get_sept_ve_disable,
Signature · 1 line
2.34.1
PATCH v2 17/58 i386/tdx: Wire CPU features up with attributes of TD guest
In reply to Xiaoyao LiFor QEMU VMs, PKS is configured via CPUID_7_0_ECX_PKS and PMU is configured by x86cpu->enable_pmu. Reuse the existing configuration interface for TDX VMs.
--- target/i386/kvm/tdx.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-)
target/i386/kvm/tdx.c +13 -1
@@ -33,6 +33,8 @@ (1U << KVM_FEATURE_MSI_EXT_DEST_ID))#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28) +#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30) +#define TDX_TD_ATTRIBUTES_PERFMON BIT_ULL(63)#define TDX_ATTRIBUTES_MAX_BITS 64
@@ -460,6 +462,15 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return 0 ; } +static void setup_td_guest_attributes (X86CPU *x86cpu) +{ +CPUX86State *env = &x86cpu->env; + + tdx_guest->attributes |= (env->features[FEAT_7_0_ECX ] &CPUID_7_0_ECX_PKS ) ? +TDX_TD_ATTRIBUTES_PKS :0 ; + tdx_guest->attributes |= x86cpu->enable_pmu ?TDX_TD_ATTRIBUTES_PERFMON :0 ; +} +int tdx_pre_create_vcpu (CPUState *cpu) {MachineState *ms =MACHINE (qdev_get_machine ());
@@ -482,8 +493,9 @@ int tdx_pre_create_vcpu(CPUState *cpu) goto out_free; } +setup_td_guest_attributes (x86cpu); + init_vm->cpuid.nent =kvm_x86_arch_cpuid (env, init_vm->cpuid.entries,0 ); - init_vm->attributes = tdx_guest->attributes; do {
Signature · 1 line
2.34.1
PATCH v2 18/58 i386/tdx: Validate TD attributes
In reply to Xiaoyao LiValidate TD attributes with tdx_caps that fixed-0 bits must be zero and fixed-1 bits must be set.
Besides, sanity check the attribute bits that have not been supported by QEMU yet. e.g., debug bit, it will be allowed in the future when debug TD support lands in QEMU.
--- target/i386/kvm/tdx.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-)
target/i386/kvm/tdx.c +25 -2
@@ -32,6 +32,7 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_TD_ATTRIBUTES_DEBUG BIT_ULL(0)#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28)#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30)#define TDX_TD_ATTRIBUTES_PERFMON BIT_ULL(63)
@@ -462,13 +463,32 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return 0 ; } -static void setup_td_guest_attributes (X86CPU *x86cpu) +static int tdx_validate_attributes (TdxGuest *tdx) +{ +if (((tdx->attributes & tdx_caps->attrs_fixed0) | tdx_caps->attrs_fixed1) != + tdx->attributes) { +error_report ("Invalid attributes 0x%lx for TDX VM (fixed0 0x%llx, fixed1 0x%llx)" , + tdx->attributes, tdx_caps->attrs_fixed0, tdx_caps->attrs_fixed1); +return -EINVAL ; + } + +if (tdx->attributes &TDX_TD_ATTRIBUTES_DEBUG ) { +error_report ("Current QEMU doesn't support attributes.debug[bit 0] for TDX VM" ); +return -EINVAL ; + } + +return 0 ; +} + +static int setup_td_guest_attributes (X86CPU *x86cpu) {CPUX86State *env = &x86cpu->env; tdx_guest->attributes |= (env->features[FEAT_7_0_ECX ] &CPUID_7_0_ECX_PKS ) ?TDX_TD_ATTRIBUTES_PKS :0 ; tdx_guest->attributes |= x86cpu->enable_pmu ?TDX_TD_ATTRIBUTES_PERFMON :0 ; + +return tdx_validate_attributes (tdx_guest); }int tdx_pre_create_vcpu (CPUState *cpu)
@@ -493,7 +513,10 @@ int tdx_pre_create_vcpu(CPUState *cpu) goto out_free; } -setup_td_guest_attributes (x86cpu); + r =setup_td_guest_attributes (x86cpu); +if (r) { + goto out; + } init_vm->cpuid.nent =kvm_x86_arch_cpuid (env, init_vm->cpuid.entries,0 ); init_vm->attributes = tdx_guest->attributes;
Signature · 1 line
2.34.1
Re: [PATCH v2 18/58] i386/tdx: Validate TD attributes
Quoted 33 lines · Xiaoyao Li
Validate TD attributes with
tdx_capsthat fixed-0 bits must be zero and fixed-1 bits must be set.Besides, sanity check the attribute bits that have not been supported by QEMU yet. e.g., debug bit, it will be allowed in the future when debug TD support lands in QEMU.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- target/i386/kvm/tdx.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-)target/i386/kvm/tdx.c +4
-1@@ -32,6 +32,7 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_TD_ATTRIBUTES_DEBUG BIT_ULL(0)#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28)#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30)#define TDX_TD_ATTRIBUTES_PERFMON BIT_ULL(63)@@ -462,13 +463,32 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return 0 ; } -static void setup_td_guest_attributes (X86CPU *x86cpu) +static int tdx_validate_attributes (TdxGuest *tdx) +{ + if (((tdx->attributes & tdx_caps->attrs_fixed0) | tdx_caps->attrs_fixed1) !=
target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 478
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+ tdx->attributes) { +error_report ("Invalid attributes 0x%lx for TDX VM (fixed0 0x%llx, fixed1 0x%llx)" , + tdx->attributes, tdx_caps->attrs_fixed0, tdx_caps->attrs_fixed1); +return -EINVAL ; + } + +if (tdx->attributes &TDX_TD_ATTRIBUTES_DEBUG ) { +error_report ("Current QEMU doesn't support attributes.debug[bit 0] for TDX VM" ); +return -EINVAL ; + }
Use error_setg() in both cases, passing in a 'Error **errp' object, and 'return -1' instead of returning an errno value.
target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 491
Quoted 13 lines
target/i386/kvm/tdx.c +7
-0+ +return 0 ; +} + +static int setup_td_guest_attributes (X86CPU *x86cpu) {CPUX86State *env = &x86cpu->env; tdx_guest->attributes |= (env->features[FEAT_7_0_ECX ] &CPUID_7_0_ECX_PKS ) ?TDX_TD_ATTRIBUTES_PKS :0 ; tdx_guest->attributes |= x86cpu->enable_pmu ?TDX_TD_ATTRIBUTES_PERFMON :0 ; + +return tdx_validate_attributes (tdx_guest);
Pass along "errp" into this
target/i386/kvm/tdx.c int tdx_pre_create_vcpu(CPUState *cpu)
Quoted 4 lines
target/i386/kvm/tdx.c +0
-0}int tdx_pre_create_vcpu (CPUState *cpu)@@ -493,7 +513,10 @@ int tdx_pre_create_vcpu(CPUState *cpu)
In an earlier patch I suggested adding 'Error **errp' to this method...
target/i386/kvm/tdx.c int tdx_pre_create_vcpu(CPUState *cpu) · line 516
Quoted 5 lines
target/i386/kvm/tdx.c +1
-1goto out_free; } -setup_td_guest_attributes (x86cpu); + r =setup_td_guest_attributes (x86cpu);
...it can also be passed into this method
Quoted 9 lines
target/i386/kvm/tdx.c +3
-0+if (r) { + goto out; + } init_vm->cpuid.nent =kvm_x86_arch_cpuid (env, init_vm->cpuid.entries,0 ); init_vm->attributes = tdx_guest->attributes;Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 18/58] i386/tdx: Validate TD attributes
Quoted 39 lines · Daniel P. Berrangé
Quoted 38 lines · Xiaoyao Li
Validate TD attributes with
tdx_capsthat fixed-0 bits must be zero and fixed-1 bits must be set.Besides, sanity check the attribute bits that have not been supported by QEMU yet. e.g., debug bit, it will be allowed in the future when debug TD support lands in QEMU.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- target/i386/kvm/tdx.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-)target/i386/kvm/tdx.c +9
-1@@ -32,6 +32,7 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_TD_ATTRIBUTES_DEBUG BIT_ULL(0)#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28)#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30)#define TDX_TD_ATTRIBUTES_PERFMON BIT_ULL(63)@@ -462,13 +463,32 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return 0 ; } -static void setup_td_guest_attributes (X86CPU *x86cpu) +static int tdx_validate_attributes (TdxGuest *tdx) +{ +if (((tdx->attributes & tdx_caps->attrs_fixed0) | tdx_caps->attrs_fixed1) != + tdx->attributes) { +error_report ("Invalid attributes 0x%lx for TDX VM (fixed0 0x%llx, fixed1 0x%llx)" , + tdx->attributes, tdx_caps->attrs_fixed0, tdx_caps->attrs_fixed1); +return -EINVAL ; + }
Quoted 8 lines
target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 478
Quoted 5 lines
target/i386/kvm/tdx.c +5
-0+ +if (tdx->attributes &TDX_TD_ATTRIBUTES_DEBUG ) { +error_report ("Current QEMU doesn't support attributes.debug[bit 0] for TDX VM" ); +return -EINVAL ; + }Use
error_setg()in both cases, passing in a 'Error **errp' object, and 'return -1' instead of returning an errno value.
Will do it in next version.
thanks!
Quoted 43 lines
target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 491
Quoted 13 lines
target/i386/kvm/tdx.c +7
-0+ +return 0 ; +} + +static int setup_td_guest_attributes (X86CPU *x86cpu) {CPUX86State *env = &x86cpu->env; tdx_guest->attributes |= (env->features[FEAT_7_0_ECX ] &CPUID_7_0_ECX_PKS ) ?TDX_TD_ATTRIBUTES_PKS :0 ; tdx_guest->attributes |= x86cpu->enable_pmu ?TDX_TD_ATTRIBUTES_PERFMON :0 ; + +return tdx_validate_attributes (tdx_guest);Pass along "errp" into this
target/i386/kvm/tdx.c int tdx_pre_create_vcpu(CPUState *cpu)
Quoted 4 lines
target/i386/kvm/tdx.c +0
-0}int tdx_pre_create_vcpu (CPUState *cpu)@@ -493,7 +513,10 @@ int tdx_pre_create_vcpu(CPUState *cpu)In an earlier patch I suggested adding 'Error **errp' to this method...
target/i386/kvm/tdx.c int tdx_pre_create_vcpu(CPUState *cpu) · line 516
Quoted 5 lines
target/i386/kvm/tdx.c +1
-1goto out_free; } -setup_td_guest_attributes (x86cpu); + r =setup_td_guest_attributes (x86cpu);...it can also be passed into this method
Quoted 9 lines
target/i386/kvm/tdx.c +3
-0+if (r) { + goto out; + } init_vm->cpuid.nent =kvm_x86_arch_cpuid (env, init_vm->cpuid.entries,0 ); init_vm->attributes = tdx_guest->attributes;Signature · 1 line
2.34.1With regards,
Daniel
Re: [PATCH v2 18/58] i386/tdx: Validate TD attributes
In reply to Daniel P. BerrangéQuoted 39 lines · Daniel P. Berrangé
Quoted 38 lines · Xiaoyao Li
Validate TD attributes with
tdx_capsthat fixed-0 bits must be zero and fixed-1 bits must be set.Besides, sanity check the attribute bits that have not been supported by QEMU yet. e.g., debug bit, it will be allowed in the future when debug TD support lands in QEMU.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- target/i386/kvm/tdx.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-)target/i386/kvm/tdx.c +9
-1@@ -32,6 +32,7 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_TD_ATTRIBUTES_DEBUG BIT_ULL(0)#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28)#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30)#define TDX_TD_ATTRIBUTES_PERFMON BIT_ULL(63)@@ -462,13 +463,32 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return 0 ; } -static void setup_td_guest_attributes (X86CPU *x86cpu) +static int tdx_validate_attributes (TdxGuest *tdx) +{ +if (((tdx->attributes & tdx_caps->attrs_fixed0) | tdx_caps->attrs_fixed1) != + tdx->attributes) { +error_report ("Invalid attributes 0x%lx for TDX VM (fixed0 0x%llx, fixed1 0x%llx)" , + tdx->attributes, tdx_caps->attrs_fixed0, tdx_caps->attrs_fixed1); +return -EINVAL ; + }
Quoted 9 lines
target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 478
Quoted 5 lines
target/i386/kvm/tdx.c +5
-0+ +if (tdx->attributes &TDX_TD_ATTRIBUTES_DEBUG ) { +error_report ("Current QEMU doesn't support attributes.debug[bit 0] for TDX VM" ); +return -EINVAL ; + }Use
error_setg()in both cases, passing in a 'Error **errp' object, and 'return -1' instead of returning an errno value.
why return -1 instead of -EINVAL?
Re: [PATCH v2 18/58] i386/tdx: Validate TD attributes
Quoted 48 lines · Xiaoyao Li
Quoted 37 lines · Daniel P. Berrangé
Quoted 36 lines · Xiaoyao Li
Validate TD attributes with
tdx_capsthat fixed-0 bits must be zero and fixed-1 bits must be set.Besides, sanity check the attribute bits that have not been supported by QEMU yet. e.g., debug bit, it will be allowed in the future when debug TD support lands in QEMU.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- target/i386/kvm/tdx.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-)target/i386/kvm/tdx.c +9
-1@@ -32,6 +32,7 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_TD_ATTRIBUTES_DEBUG BIT_ULL(0)#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28)#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30)#define TDX_TD_ATTRIBUTES_PERFMON BIT_ULL(63)@@ -462,13 +463,32 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return 0 ; } -static void setup_td_guest_attributes (X86CPU *x86cpu) +static int tdx_validate_attributes (TdxGuest *tdx) +{ +if (((tdx->attributes & tdx_caps->attrs_fixed0) | tdx_caps->attrs_fixed1) != + tdx->attributes) { +error_report ("Invalid attributes 0x%lx for TDX VM (fixed0 0x%llx, fixed1 0x%llx)" , + tdx->attributes, tdx_caps->attrs_fixed0, tdx_caps->attrs_fixed1); +return -EINVAL ; + }Quoted 9 lines
target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 478
Quoted 5 lines
target/i386/kvm/tdx.c +5
-0+ +if (tdx->attributes &TDX_TD_ATTRIBUTES_DEBUG ) { +error_report ("Current QEMU doesn't support attributes.debug[bit 0] for TDX VM" ); +return -EINVAL ; + }Use
error_setg()in both cases, passing in a 'Error **errp' object, and 'return -1' instead of returning an errno value.
Quoted 1 line
why return -1 instead of
-EINVAL?
Returning errno values is useful if the method isn't providing an "Error **errp" parameter, because it lets the caller report a more detailed error message via strerror(). Once you add a Error ** parameter though, there is almost never any reason for the caller to care about the original errno value, and so we use 0 / -1 as success/fail indicators.
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 18/58] i386/tdx: Validate TD attributes
Quoted 49 lines · Daniel P. Berrangé
Quoted 48 lines · Xiaoyao Li
Quoted 46 lines · Daniel P. Berrangé
Quoted 31 lines · Xiaoyao Li
Validate TD attributes with
tdx_capsthat fixed-0 bits must be zero and fixed-1 bits must be set.Besides, sanity check the attribute bits that have not been supported by QEMU yet. e.g., debug bit, it will be allowed in the future when debug TD support lands in QEMU.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>--- target/i386/kvm/tdx.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-)target/i386/kvm/tdx.c +4
-1@@ -32,6 +32,7 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_TD_ATTRIBUTES_DEBUG BIT_ULL(0)#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28)#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30)#define TDX_TD_ATTRIBUTES_PERFMON BIT_ULL(63)@@ -462,13 +463,32 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return 0 ; } -static void setup_td_guest_attributes (X86CPU *x86cpu) +static int tdx_validate_attributes (TdxGuest *tdx) +{ + if (((tdx->attributes & tdx_caps->attrs_fixed0) | tdx_caps->attrs_fixed1) !=target/i386/kvm/tdx.c int tdx_kvm_init(MachineState *ms, Error **errp) · line 477
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+ tdx->attributes) { +error_report ("Invalid attributes 0x%lx for TDX VM (fixed0 0x%llx, fixed1 0x%llx)" , + tdx->attributes, tdx_caps->attrs_fixed0, tdx_caps->attrs_fixed1); +return -EINVAL ; + } + +if (tdx->attributes &TDX_TD_ATTRIBUTES_DEBUG ) { +error_report ("Current QEMU doesn't support attributes.debug[bit 0] for TDX VM" ); +return -EINVAL ; + }Use
error_setg()in both cases, passing in a 'Error **errp' object, and 'return -1' instead of returning an errno value.
Quoted 8 lines
Quoted 1 line
why return -1 instead of
-EINVAL?Returning errno values is useful if the method isn't providing an "Error **errp" parameter, because it lets the caller report a more detailed error message via
strerror(). Once you add a Error ** parameter though, there is almost never any reason for the caller to care about the original errno value, and so we use 0 / -1 as success/fail indicators.
I see.
Thanks,
-Xiaoyao
Quoted 2 lines
With regards,
Daniel
PATCH v2 19/58 qom: implement property helper for sha384
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
Implement property_add_sha384() which converts hex string <-> uint8_t[48] It will be used for TDX which uses sha384 for measurement.
--- include/qom/object.h | 17 ++++++++++ qom/object.c | 76 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 93 insertions(+)
include/qom/object.h +17 -0
@@ -1887,6 +1887,23 @@ ObjectProperty *object_property_add_alias(Object *obj, const char *name,ObjectProperty *object_property_add_const_link (Object *obj,const char *name,Object *target); + +/** +* object_property_add_sha384: +* @obj: the object to add a property to +* @name: the name of the property +* @v: pointer to value +* @flags: bitwise-or'd ObjectPropertyFlags +* +* Add an sha384 property in memory. This function will add a +* property of type 'sha384'. +* +* Returns: The newly added property on success, or %NULL on failure. +*/ +ObjectProperty *object_property_add_sha384 (Object *obj,const char *name, +const uint8_t *v, +ObjectPropertyFlags flags); +/** * object_property_set_description: * @obj: the object owning the property
qom/object.c +76 -0
88 lines not on this page · raw message
Signature · 1 line
2.34.1
Re: [PATCH v2 19/58] qom: implement property helper for sha384
Quoted 4 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
Implement
property_add_sha384()which converts hex string <-> uint8_t[48] It will be used for TDX which uses sha384 for measurement.
I think it is likely a better idea to use base64 for the encoding the binary hash - we use base64 for all the sev-guest properties that were binary data.
At which points the property set/get logic is much simpler as it is just needing a call to g_base64_encode / g_base64_decode and length validation for the decode case.
Quoted 133 lines
Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- include/qom/object.h | 17 ++++++++++ qom/object.c | 76 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 93 insertions(+)include/qom/object.h +17
-0@@ -1887,6 +1887,23 @@ ObjectProperty *object_property_add_alias(Object *obj, const char *name,ObjectProperty *object_property_add_const_link (Object *obj,const char *name,Object *target); + +/** +* object_property_add_sha384: +* @obj: the object to add a property to +* @name: the name of the property +* @v: pointer to value +* @flags: bitwise-or'd ObjectPropertyFlags +* +* Add an sha384 property in memory. This function will add a +* property of type 'sha384'. +* +* Returns: The newly added property on success, or %NULL on failure. +*/ +ObjectProperty *object_property_add_sha384 (Object *obj,const char *name, +const uint8_t *v, +ObjectPropertyFlags flags); +/** * object_property_set_description: * @obj: the object owning the property qom/object.c +76
-0@@ -15,6 +15,7 @@#include "qapi/error.h" #include "qom/object.h" #include "qom/object_interfaces.h" +#include "qemu/ctype.h" #include "qemu/cutils.h" #include "qemu/memalign.h" #include "qapi/visitor.h" @@ -2781,6 +2782,81 @@ object_property_add_alias(Object *obj, const char *name,return op; } +#define SHA384_DIGEST_SIZE 48 +static void property_get_sha384 (Object *obj,Visitor *v,const char *name, +void *opaque,Error **errp) +{ +uint8_t *value = (uint8_t *)opaque; +char str[SHA384_DIGEST_SIZE *2 +1 ]; +char *str_ = (char *)str; +size_t i; + +for (i =0 ; i <SHA384_DIGEST_SIZE ; i++) { +char *buf; + buf = &str[i *2 ]; + +sprintf (buf,"%02hhx" , value[i]); + } + str[SHA384_DIGEST_SIZE *2 ] ='\0' ; + +visit_type_str (v, name, &str_, errp); +} + +static void property_set_sha384 (Object *obj,Visitor *v,const char *name, +void *opaque,Error **errp) +{ +uint8_t *value = (uint8_t *)opaque; +char * str; +size_t len; +size_t i; + +if (!visit_type_str (v, name, &str, errp)) { + goto err; + } + + len =strlen (str); +if (len !=SHA384_DIGEST_SIZE *2 ) { +error_setg (errp,"invalid length for sha348 hex string %s. " +"it must be 48 * 2 hex" , name); + goto err; + } + +for (i =0 ; i <SHA384_DIGEST_SIZE ; i++) { +if (!qemu_isxdigit (str[i *2 ]) || !qemu_isxdigit (str[i *2 +1 ])) { +error_setg (errp,"invalid char for sha318 hex string %s at %c%c" , + name, str[i *2 ], str[i *2 +1 ]); + goto err; + } + +if (sscanf (str + i *2 ,"%02hhx" , &value[i]) !=1 ) { +error_setg (errp,"invalid format for sha318 hex string %s" , name); + goto err; + } + } + +err: +g_free (str); +} + +ObjectProperty * +object_property_add_sha384 (Object *obj,const char *name, +const uint8_t *v,ObjectPropertyFlags flags) +{ +ObjectPropertyAccessor *getter =NULL ; +ObjectPropertyAccessor *setter =NULL ; + +if ((flags &OBJ_PROP_FLAG_READ ) ==OBJ_PROP_FLAG_READ ) { + getter = property_get_sha384; + } + +if ((flags &OBJ_PROP_FLAG_WRITE ) ==OBJ_PROP_FLAG_WRITE ) { + setter = property_set_sha384; + } + +return object_property_add (obj, name,"sha384" , + getter, setter,NULL , (void *)v); +} +void object_property_set_description (Object *obj,const char *name,const char *description) {Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 19/58] qom: implement property helper for sha384
Quoted 6 lines · Daniel P. Berrangé
Quoted 4 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
Implement
property_add_sha384()which converts hex string <-> uint8_t[48] It will be used for TDX which uses sha384 for measurement.
Quoted 7 lines
I think it is likely a better idea to use base64 for the encoding the binary hash - we use base64 for all the sev-guest properties that were binary data.
At which points the property set/get logic is much simpler as it is just needing a call to
g_base64_encode/g_base64_decodeand length validation for the decode case.
Hex string is poplar to show hash value, isn't it? Anyway it's easy for human operator, shell scripts, libvirt or whatever to convert those representations with utility commands like base64 or xxd, or library call. Either way would work.
Signature · 1 line
Isaku Yamahata <[email protected]>
PATCH v2 20/58 i386/tdx: Allows mrconfigid/mrowner/mrownerconfig for TDX_INIT_VM
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
When creating TDX vm, three sha384 hash values can be provided for TDX attestation.
So far they were hard coded as 0. Now allow user to specify those values via property mrconfigid, mrowner and mrownerconfig. Choose hex-encoded string as format since it's friendly for user to input.
example
-object tdx-guest, \
mrconfigid=0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef, \
mrowner=fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210, \
mrownerconfig=0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef
---
TODO:
- community requests to use base64 encoding if no special reason
--- qapi/qom.json | 11 ++++++++++- target/i386/kvm/tdx.c | 13 +++++++++++++ target/i386/kvm/tdx.h | 3 +++ 3 files changed, 26 insertions(+), 1 deletion(-)
qapi/qom.json +10 -1
@@ -873,10 +873,19 @@ # # @sept-ve-disable: bit 28 of TD attributes (default: 0) # +# @mrconfigid: MRCONFIGID SHA384 hex string of 48 * 2 length (default: 0) +# +# @mrowner: MROWNER SHA384 hex string of 48 * 2 length (default: 0) +# +# @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) +# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', - 'data': { '*sept-ve-disable': 'bool' } } + 'data': { '*sept-ve-disable': 'bool', + '*mrconfigid': 'str', + '*mrowner': 'str', + '*mrownerconfig': 'str' } } ## # @ThreadContextProperties:
target/i386/kvm/tdx.c +13 -0
@@ -521,6 +521,13 @@ int tdx_pre_create_vcpu(CPUState *cpu) init_vm->cpuid.nent =kvm_x86_arch_cpuid (env, init_vm->cpuid.entries,0 ); init_vm->attributes = tdx_guest->attributes; +QEMU_BUILD_BUG_ON (sizeof (init_vm->mrconfigid) !=sizeof (tdx_guest->mrconfigid)); +QEMU_BUILD_BUG_ON (sizeof (init_vm->mrowner) !=sizeof (tdx_guest->mrowner)); +QEMU_BUILD_BUG_ON (sizeof (init_vm->mrownerconfig) !=sizeof (tdx_guest->mrownerconfig)); +memcpy (init_vm->mrconfigid, tdx_guest->mrconfigid,sizeof (tdx_guest->mrconfigid)); +memcpy (init_vm->mrowner, tdx_guest->mrowner,sizeof (tdx_guest->mrowner)); +memcpy (init_vm->mrownerconfig, tdx_guest->mrownerconfig,sizeof (tdx_guest->mrownerconfig)); +do { r =tdx_vm_ioctl (KVM_TDX_INIT_VM ,0 , init_vm); }while (r == -EAGAIN );
@@ -575,6 +582,12 @@ static void tdx_guest_init(Object *obj)object_property_add_bool (obj,"sept-ve-disable" , tdx_guest_get_sept_ve_disable, tdx_guest_set_sept_ve_disable); +object_property_add_sha384 (obj,"mrconfigid" , tdx->mrconfigid, +OBJ_PROP_FLAG_READWRITE ); +object_property_add_sha384 (obj,"mrowner" , tdx->mrowner, +OBJ_PROP_FLAG_READWRITE ); +object_property_add_sha384 (obj,"mrownerconfig" , tdx->mrownerconfig, +OBJ_PROP_FLAG_READWRITE ); }static void tdx_guest_finalize (Object *obj)
target/i386/kvm/tdx.h +3 -0
@@ -21,6 +21,9 @@ typedef struct TdxGuest {bool initialized;uint64_t attributes;/* TD attributes */ +uint8_t mrconfigid[48 ];/* sha348 digest */ +uint8_t mrowner[48 ];/* sha348 digest */ +uint8_t mrownerconfig[48 ];/* sha348 digest */ } TdxGuest;#ifdef CONFIG_TDX
Signature · 1 line
2.34.1
Re: [PATCH v2 20/58] i386/tdx: Allows mrconfigid/mrowner/mrownerconfig for TDX_INIT_VM
Quoted 31 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
When creating TDX vm, three sha384 hash values can be provided for TDX attestation.
So far they were hard coded as 0. Now allow user to specify those values via property mrconfigid, mrowner and mrownerconfig. Choose hex-encoded string as format since it's friendly for user to input.
example
-object tdx-guest, \
mrconfigid=0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef, \
mrowner=fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210, \
mrownerconfig=0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdefSigned-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>---
TODO:
- community requests to use base64 encoding if no special reason--- qapi/qom.json | 11 ++++++++++- target/i386/kvm/tdx.c | 13 +++++++++++++ target/i386/kvm/tdx.h | 3 +++ 3 files changed, 26 insertions(+), 1 deletion(-)qapi/qom.json +0
-0
qapi/qom.json line 880
Quoted 8 lines
qapi/qom.json +5
-0@@ -873,10 +873,19 @@ # # @sept-ve-disable: bit 28 of TD attributes (default: 0) # +# @mrconfigid: MRCONFIGID SHA384 hex string of 48 * 2 length (default: 0) +# +# @mrowner: MROWNER SHA384 hex string of 48 * 2 length (default: 0) +# +# @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0)
Per previous patch, I suggest these should all be passed in base64 instead of hex. Also 'default: 0' makes no sense for a string, which would be 'default: nil', and no need to document that as the default is implicit from the fact that its an optional string field. So eg
@mrconfigid: base64 encoded MRCONFIGID SHA384 digest
Quoted 60 lines
qapi/qom.json +5
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', - 'data': { '*sept-ve-disable': 'bool' } } + 'data': { '*sept-ve-disable': 'bool', + '*mrconfigid': 'str', + '*mrowner': 'str', + '*mrownerconfig': 'str' } } ## # @ThreadContextProperties:target/i386/kvm/tdx.c +13
-0@@ -521,6 +521,13 @@ int tdx_pre_create_vcpu(CPUState *cpu) init_vm->cpuid.nent =kvm_x86_arch_cpuid (env, init_vm->cpuid.entries,0 ); init_vm->attributes = tdx_guest->attributes; +QEMU_BUILD_BUG_ON (sizeof (init_vm->mrconfigid) !=sizeof (tdx_guest->mrconfigid)); +QEMU_BUILD_BUG_ON (sizeof (init_vm->mrowner) !=sizeof (tdx_guest->mrowner)); +QEMU_BUILD_BUG_ON (sizeof (init_vm->mrownerconfig) !=sizeof (tdx_guest->mrownerconfig)); +memcpy (init_vm->mrconfigid, tdx_guest->mrconfigid,sizeof (tdx_guest->mrconfigid)); +memcpy (init_vm->mrowner, tdx_guest->mrowner,sizeof (tdx_guest->mrowner)); +memcpy (init_vm->mrownerconfig, tdx_guest->mrownerconfig,sizeof (tdx_guest->mrownerconfig)); +do { r =tdx_vm_ioctl (KVM_TDX_INIT_VM ,0 , init_vm); }while (r == -EAGAIN );@@ -575,6 +582,12 @@ static void tdx_guest_init(Object *obj)object_property_add_bool (obj,"sept-ve-disable" , tdx_guest_get_sept_ve_disable, tdx_guest_set_sept_ve_disable); +object_property_add_sha384 (obj,"mrconfigid" , tdx->mrconfigid, +OBJ_PROP_FLAG_READWRITE ); +object_property_add_sha384 (obj,"mrowner" , tdx->mrowner, +OBJ_PROP_FLAG_READWRITE ); +object_property_add_sha384 (obj,"mrownerconfig" , tdx->mrownerconfig, +OBJ_PROP_FLAG_READWRITE ); }static void tdx_guest_finalize (Object *obj)target/i386/kvm/tdx.h +3
-0@@ -21,6 +21,9 @@ typedef struct TdxGuest {bool initialized;uint64_t attributes;/* TD attributes */ +uint8_t mrconfigid[48 ];/* sha348 digest */ +uint8_t mrowner[48 ];/* sha348 digest */ +uint8_t mrownerconfig[48 ];/* sha348 digest */ } TdxGuest;#ifdef CONFIG_TDX Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 20/58] i386/tdx: Allows mrconfigid/mrowner/mrownerconfig for TDX_INIT_VM
Quoted 27 lines · Daniel P. Berrangé
Quoted 26 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
When creating TDX vm, three sha384 hash values can be provided for TDX attestation.
So far they were hard coded as 0. Now allow user to specify those values via property mrconfigid, mrowner and mrownerconfig. Choose hex-encoded string as format since it's friendly for user to input.
example
-object tdx-guest, \
mrconfigid=0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef, \
mrowner=fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210fedcba9876543210, \
mrownerconfig=0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdefSigned-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>---
TODO:
- community requests to use base64 encoding if no special reason--- qapi/qom.json | 11 ++++++++++- target/i386/kvm/tdx.c | 13 +++++++++++++ target/i386/kvm/tdx.h | 3 +++ 3 files changed, 26 insertions(+), 1 deletion(-)
Quoted 16 lines
qapi/qom.json line 880
Quoted 13 lines
qapi/qom.json +5
-0@@ -873,10 +873,19 @@ # # @sept-ve-disable: bit 28 of TD attributes (default: 0) # +# @mrconfigid: MRCONFIGID SHA384 hex string of 48 * 2 length (default: 0) +# +# @mrowner: MROWNER SHA384 hex string of 48 * 2 length (default: 0) +# +# @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0)Per previous patch, I suggest these should all be passed in base64 instead of hex.
I'm upgrading this suggestion to a demand: we use base64 for encoding binary data everywhere in QAPI/QMP. Consistency matters.
Quoted 6 lines
Also 'default: 0' makes no sense for a string,
which would be 'default: nil', and no need to document that as the default is implicit from the fact that its an optional string field. So eg@mrconfigid: base64 encoded MRCONFIGID SHA384 digest
Agree.
The member names are abbreviations all run together, wheras QAPI/QMP favors words-separated-with-dashes. If you invented them, please change them to QAPI/QMP style. If they are established TDX terminology, keep them as they are, but please point to your evidence.
Quoted 12 lines
Quoted 12 lines
qapi/qom.json +5
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', - 'data': { '*sept-ve-disable': 'bool' } } + 'data': { '*sept-ve-disable': 'bool', + '*mrconfigid': 'str', + '*mrowner': 'str', + '*mrownerconfig': 'str' } } ## # @ThreadContextProperties:
[...]
PATCH v2 21/58 i386/tdx: Implement user specified tsc frequency
In reply to Xiaoyao LiReuse "-cpu,tsc-frequency=" to get user wanted tsc frequency and call VM scope VM_SET_TSC_KHZ to set the tsc frequency of TD before KVM_TDX_INIT_VM.
Besides, sanity check the tsc frequency to be in the legal range and legal granularity (required by TDX module).
---
Changes from RFC v4:
- Use VM scope VM_SET_TSC_KHZ to set the TSC frequency of TD since KVM side drop the @tsc_khz field in struct kvm_tdx_init_vm
--- target/i386/kvm/kvm.c | 9 +++++++++ target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++ 2 files changed, 33 insertions(+)
target/i386/kvm/kvm.c +9 -0
@@ -859,6 +859,15 @@ static int kvm_arch_set_tsc_khz(CPUState *cs)int r, cur_freq;bool set_ioctl = false; +/* +* TSC of TD vcpu is immutable, it cannot be set/changed via vcpu scope +* VM_SET_TSC_KHZ, but only be initialized via VM scope VM_SET_TSC_KHZ +* before ioctl KVM_TDX_INIT_VM in tdx_pre_create_vcpu() +*/ +if (is_tdx_vm ()) { +return 0 ; + } +if (!env->tsc_khz) {return 0 ; }
target/i386/kvm/tdx.c +24 -0
@@ -32,6 +32,9 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_MIN_TSC_FREQUENCY_KHZ (100 * 1000) +#define TDX_MAX_TSC_FREQUENCY_KHZ (10 * 1000 * 1000) +#define TDX_TD_ATTRIBUTES_DEBUG BIT_ULL(0)#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28)#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30)
@@ -513,6 +516,27 @@ int tdx_pre_create_vcpu(CPUState *cpu) goto out_free; } + r = -EINVAL ; +if (env->tsc_khz && (env->tsc_khz <TDX_MIN_TSC_FREQUENCY_KHZ || + env->tsc_khz >TDX_MAX_TSC_FREQUENCY_KHZ )) { +error_report ("Invalid TSC %ld KHz, must specify cpu_frequency between [%d, %d] kHz" , + env->tsc_khz,TDX_MIN_TSC_FREQUENCY_KHZ , +TDX_MAX_TSC_FREQUENCY_KHZ ); + goto out; + } + +if (env->tsc_khz % (25 *1000 )) { +error_report ("Invalid TSC %ld KHz, it must be multiple of 25MHz" , env->tsc_khz); + goto out; + } + +/* it's safe even env->tsc_khz is 0. KVM uses host's tsc_khz in this case */ + r =kvm_vm_ioctl (kvm_state,KVM_SET_TSC_KHZ , env->tsc_khz); +if (r <0 ) { +error_report ("Unable to set TSC frequency to %" PRId64" kHz" , env->tsc_khz); + goto out; + } + r =setup_td_guest_attributes (x86cpu);if (r) { goto out;
Signature · 1 line
2.34.1
Re: [PATCH v2 21/58] i386/tdx: Implement user specified tsc frequency
Quoted 65 lines · Xiaoyao Li
Reuse "-cpu,tsc-frequency=" to get user wanted tsc frequency and call VM scope
VM_SET_TSC_KHZto set the tsc frequency of TD beforeKVM_TDX_INIT_VM.Besides, sanity check the tsc frequency to be in the legal range and legal granularity (required by TDX module).
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>---
Changes from RFC v4:
- Use VM scopeVM_SET_TSC_KHZto set the TSC frequency of TD since KVM side drop the @tsc_khz field in structkvm_tdx_init_vm--- target/i386/kvm/kvm.c | 9 +++++++++ target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++ 2 files changed, 33 insertions(+)target/i386/kvm/kvm.c +9
-0@@ -859,6 +859,15 @@ static int kvm_arch_set_tsc_khz(CPUState *cs)int r, cur_freq;bool set_ioctl = false; +/* +* TSC of TD vcpu is immutable, it cannot be set/changed via vcpu scope +* VM_SET_TSC_KHZ, but only be initialized via VM scope VM_SET_TSC_KHZ +* before ioctl KVM_TDX_INIT_VM in tdx_pre_create_vcpu() +*/ +if (is_tdx_vm ()) { +return 0 ; + } +if (!env->tsc_khz) {return 0 ; }target/i386/kvm/tdx.c +13
-0@@ -32,6 +32,9 @@ (1U << KVM_FEATURE_PV_SCHED_YIELD) | \ (1U << KVM_FEATURE_MSI_EXT_DEST_ID)) +#define TDX_MIN_TSC_FREQUENCY_KHZ (100 * 1000) +#define TDX_MAX_TSC_FREQUENCY_KHZ (10 * 1000 * 1000) +#define TDX_TD_ATTRIBUTES_DEBUG BIT_ULL(0)#define TDX_TD_ATTRIBUTES_SEPT_VE_DISABLE BIT_ULL(28)#define TDX_TD_ATTRIBUTES_PKS BIT_ULL(30)@@ -513,6 +516,27 @@ int tdx_pre_create_vcpu(CPUState *cpu) goto out_free; } + r = -EINVAL ; +if (env->tsc_khz && (env->tsc_khz <TDX_MIN_TSC_FREQUENCY_KHZ || + env->tsc_khz >TDX_MAX_TSC_FREQUENCY_KHZ )) { +error_report ("Invalid TSC %ld KHz, must specify cpu_frequency between [%d, %d] kHz" , + env->tsc_khz,TDX_MIN_TSC_FREQUENCY_KHZ , +TDX_MAX_TSC_FREQUENCY_KHZ ); + goto out; + } + +if (env->tsc_khz % (25 *1000 )) {
target/i386/kvm/tdx.c int tdx_pre_create_vcpu(CPUState *cpu) · line 538
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+error_report ("Invalid TSC %ld KHz, it must be multiple of 25MHz" , env->tsc_khz); + goto out; + } + +/* it's safe even env->tsc_khz is 0. KVM uses host's tsc_khz in this case */ + r =kvm_vm_ioctl (kvm_state,KVM_SET_TSC_KHZ , env->tsc_khz); +if (r <0 ) { +error_report ("Unable to set TSC frequency to %" PRId64" kHz" , env->tsc_khz); + goto out; + }
error_setg(errp, ....) in all of these cases.
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
PATCH v2 22/58 i386/tdx: Set kvm_readonly_mem_enabled to false for TDX VM
In reply to Xiaoyao LiTDX only supports readonly for shared memory but not for private memory.
In the view of QEMU, it has no idea whether a memslot is used as shared memory of private. Thus just mark kvm_readonly_mem_enabled to false to TDX VM for simplicity.
--- target/i386/kvm/tdx.c | 9 +++++++++ 1 file changed, 9 insertions(+)
target/i386/kvm/tdx.c +9 -0
@@ -461,6 +461,15 @@ int tdx_kvm_init(MachineState *ms, Error **errp)update_tdx_cpuid_lookup_by_tdx_caps (); +/* +* Set kvm_readonly_mem_allowed to false, because TDX only supports readonly +* memory for shared memory but not for private memory. Besides, whether a +* memslot is private or shared is not determined by QEMU. +* +* Thus, just mark readonly memory not supported for simplicity. +*/ + kvm_readonly_mem_allowed = false; + tdx_guest = tdx;return 0 ;
Signature · 1 line
2.34.1
PATCH v2 23/58 i386/tdx: Make memory type private by default
In reply to Xiaoyao LiBy default (due to the recent UPM change), restricted memory attribute is shared. Convert the memory region from shared to private at the memory slot creation time.
add kvm region registering function to check the flag
and convert the region, and add memory listener to TDX guest code to set the flag to the possible memory region.
Without this patch
- Secure-EPT violation on private area
- KVM_MEMORY_FAULT EXIT (kvm -> qemu)
- qemu converts the 4K page from shared to private
- Resume VCPU execution
- Secure-EPT violation again
- KVM resolves EPT Violation
This also prevents huge page because page conversion is done at 4K granularity. Although it's possible to merge 4K private mapping into 2M large page, it slows guest boot.
With this patch
- After memory slot creation, convert the region from private to shared
- Secure-EPT violation on private area.
- KVM resolves EPT Violation
Originated-from: Isaku Yamahata <[email protected]>
--- target/i386/kvm/tdx.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+)
target/i386/kvm/tdx.c +22 -0
@@ -18,6 +18,7 @@#include "standard-headers/asm-x86/kvm_para.h" #include "sysemu/kvm.h" #include "sysemu/sysemu.h" +#include "exec/address-spaces.h" #include "hw/i386/x86.h" #include "kvm_i386.h"
@@ -578,6 +579,21 @@ out:return r; } +static void tdx_guest_region_add (MemoryListener *listener, +MemoryRegionSection *section) +{ +if (memory_region_can_be_private (section->mr)) { +memory_region_set_default_private (section->mr); + } +} + +static MemoryListener tdx_memory_listener = { + .name =TYPE_TDX_GUEST , + .region_add = tdx_guest_region_add, +/* Higher than KVM memory listener = 10. */ + .priority =MEMORY_LISTENER_PRIORITY_ACCEL_HIGH , +}; +static bool tdx_guest_get_sept_ve_disable (Object *obj,Error **errp) {TdxGuest *tdx =TDX_GUEST (obj);
@@ -607,6 +623,12 @@ OBJECT_DEFINE_TYPE_WITH_INTERFACES(TdxGuest,static void tdx_guest_init (Object *obj) {TdxGuest *tdx =TDX_GUEST (obj); +static bool memory_listener_registered = false; + +if (!memory_listener_registered) { +memory_listener_register (&tdx_memory_listener, &address_space_memory); + memory_listener_registered = true; + }qemu_mutex_init (&tdx->lock);
Signature · 1 line
2.34.1
PATCH v2 24/58 i386/tdx: Create kvm gmem for TD
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
Allocate private gmem for TD guest, if the MemoryRegion is memory backend and has private property on.
--- target/i386/kvm/tdx.c | 27 +++++++++++++++++++++++++-- 1 file changed, 25 insertions(+), 2 deletions(-)
target/i386/kvm/tdx.c +25 -2
@@ -19,6 +19,7 @@#include "sysemu/kvm.h" #include "sysemu/sysemu.h" #include "exec/address-spaces.h" +#include "exec/ramblock.h" #include "hw/i386/x86.h" #include "kvm_i386.h"
@@ -582,8 +583,30 @@ out:static void tdx_guest_region_add (MemoryListener *listener,MemoryRegionSection *section) { -if (memory_region_can_be_private (section->mr)) { -memory_region_set_default_private (section->mr); +MemoryRegion *mr = section->mr; +Object *owner =memory_region_owner (mr); + +if (owner &&object_dynamic_cast (owner,TYPE_MEMORY_BACKEND ) && +object_property_get_bool (owner,"private" ,NULL ) && + mr->ram_block && mr->ram_block->gmem_fd <0 ) { +struct kvm_create_guest_memfd gmem = { + .size =memory_region_size (mr), +/* TODO: add property to hostmem backend for huge pmd */ + .flags =KVM_GUEST_MEMFD_ALLOW_HUGEPAGE , + }; +int fd; + + fd =kvm_vm_ioctl (kvm_state,KVM_CREATE_GUEST_MEMFD , &gmem); +if (fd <0 ) { +fprintf (stderr,"%s: error creating gmem: %s\n" , __func__, +strerror (-fd)); +abort (); + } +memory_region_set_gmem_fd (mr, fd); + } + +if (memory_region_can_be_private (mr)) { +memory_region_set_default_private (mr); } }
Signature · 1 line
2.34.1
PATCH v2 25/58 kvm/tdx: Don't complain when converting vMMIO region to shared
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
Because vMMIO region needs to be shared region, guest TD may explicitly convert such region from private to shared. Don't complain such conversion.
--- accel/kvm/kvm-all.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-)
accel/kvm/kvm-all.c +18 -2
@@ -3094,8 +3094,24 @@ static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private)*/ ram_block_convert_range (rb, offset, size, to_private); }else { -warn_report ("Convert non guest-memfd backed memory region (0x%" HWADDR_PRIx" ,+ 0x%" HWADDR_PRIx") to %s" , - start, size, to_private ?"private" :"shared" ); +MemoryRegion *mr = section.mr; + +/* +* Because vMMIO region must be shared, guest TD may convert vMMIO +* region to shared explicitly. Don't complain such case. See +* memory_region_type() for checking if the region is MMIO region. +*/ +if (to_private || +memory_region_is_ram (mr) || +memory_region_is_ram_device (mr) || +memory_region_is_rom (mr) || +memory_region_is_romd (mr)) { +warn_report ("Convert non guest-memfd backed memory region (0x%" HWADDR_PRIx" ,+ 0x%" HWADDR_PRIx") of %s to %s" , + start, size, mr->name, to_private ?"private" :"shared" ); + }else { + ret =0 ; + } + }memory_region_unref (section.mr);
Signature · 1 line
2.34.1
Re: [PATCH v2 25/58] kvm/tdx: Don't complain when converting vMMIO region to shared
Quoted 24 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
Because vMMIO region needs to be shared region, guest TD may explicitly convert such region from private to shared. Don't complain such conversion.
Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- accel/kvm/kvm-all.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-)accel/kvm/kvm-all.c +2
-2@@ -3094,8 +3094,24 @@ static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private)*/ ram_block_convert_range (rb, offset, size, to_private); }else { -warn_report ("Convert non guest-memfd backed memory region (0x%" HWADDR_PRIx" ,+ 0x%" HWADDR_PRIx") to %s" , - start, size, to_private ?"private" :"shared" ); +MemoryRegion *mr = section.mr; +
accel/kvm/kvm-all.c static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private) · line 3108
Quoted 10 lines
accel/kvm/kvm-all.c +10
-0+/* +* Because vMMIO region must be shared, guest TD may convert vMMIO +* region to shared explicitly. Don't complain such case. See +* memory_region_type() for checking if the region is MMIO region. +*/ +if (to_private || +memory_region_is_ram (mr) || +memory_region_is_ram_device (mr) || +memory_region_is_rom (mr) || +memory_region_is_romd (mr)) {
Should we also have !is_tdx_vm() as the first clause to check, as IIUC you only need the special logic for TDX VMs.
accel/kvm/kvm-all.c static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private) · line 3113
Quoted 5 lines
accel/kvm/kvm-all.c +5
-0+warn_report ("Convert non guest-memfd backed memory region (0x%" HWADDR_PRIx" ,+ 0x%" HWADDR_PRIx") of %s to %s" , + start, size, mr->name, to_private ?"private" :"shared" ); + }else { + ret =0 ; + }
Inconsistent indentation here due to use of tabs
Quoted 7 lines
accel/kvm/kvm-all.c +1
-0+ }memory_region_unref (section.mr);Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
PATCH v2 26/58 kvm/tdx: Ignore memory conversion to shared of unassigned region
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
TDX requires vMMIO region to be shared. For KVM, MMIO region is the region which kvm memslot isn't assigned to (except in-kernel emulation). qemu has the memory region for vMMIO at each device level.
While OVMF issues MapGPA(to-shared) conservatively on 32bit PCI MMIO region, qemu doesn't find corresponding vMMIO region because it's before PCI device allocation and memory_region_find() finds the device region, not PCI bus region. It's safe to ignore MapGPA(to-shared) because when guest accesses those region they use GPA with shared bit set for vMMIO. Ignore memory conversion request of non-assigned region to shared and return success. Otherwise OVMF is confused and panics there.
--- accel/kvm/kvm-all.c | 12 ++++++++++++ 1 file changed, 12 insertions(+)
accel/kvm/kvm-all.c +12 -0
@@ -3070,6 +3070,18 @@ static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private)trace_kvm_convert_memory (start, size, to_private ?"shared_to_private" :"private_to_shared" ); section =memory_region_find (get_system_memory (), start, size);if (!section.mr) { +/* +* Ignore converting non-assigned region to shared. +* +* TDX requires vMMIO region to be shared to inject #VE to guest. +* OVMF issues conservatively MapGPA(shared) on 32bit PCI MMIO region, +* and vIO-APIC 0xFEC00000 4K page. +* OVMF assigns 32bit PCI MMIO region to +* [top of low memory: typically 2GB=0xC000000, 0xFC00000) +*/ +if (!to_private) { + ret =0 ; + }return ret; }
Signature · 1 line
2.34.1
PATCH v2 27/58 i386/tdvf: Introduce function to parse TDVF metadata
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
TDX VM needs to boot with its specialized firmware, Trusted Domain Virtual Firmware (TDVF). QEMU needs to parse TDVF and map it in TD guest memory prior to running the TDX VM.
A TDVF Metadata in TDVF image describes the structure of firmware. QEMU refers to it to setup memory for TDVF. Introduce function tdvf_parse_metadata() to parse the metadata from TDVF image and store the info of each TDVF section.
TDX metadata is located by a TDX metadata offset block, which is a GUID-ed structure. The data portion of the GUID structure contains only an 4-byte field that is the offset of TDX metadata to the end of firmware file.
Select X86_FW_OVMF when TDX is enable to leverage existing functions to parse and search OVMF's GUID-ed structures.
---
Changes from RFC v4:
- rename tdvf_parse_section_entry() to tdvf_parse_and_check_section_entry()
Changes in v4:
- rename TDX_METADATA_GUID to TDX_METADATA_OFFSET_GUID
--- hw/i386/Kconfig | 1 + hw/i386/meson.build | 1 + hw/i386/tdvf.c | 199 +++++++++++++++++++++++++++++++++++++++++ include/hw/i386/tdvf.h | 51 +++++++++++ 4 files changed, 252 insertions(+) create mode 100644 hw/i386/tdvf.c create mode 100644 include/hw/i386/tdvf.h
hw/i386/Kconfig +1 -0
@@ -12,6 +12,7 @@ config SGX config TDX bool + select X86_FW_OVMF depends on KVM config PC
hw/i386/meson.build +1 -0
@@ -28,6 +28,7 @@ i386_ss.add(when: 'CONFIG_PC', if_true: files( 'port92.c')) i386_ss.add(when: 'CONFIG_X86_FW_OVMF', if_true: files('pc_sysfw_ovmf.c'), if_false: files('pc_sysfw_ovmf-stubs.c')) +i386_ss.add(when: 'CONFIG_TDX', if_true: files('tdvf.c')) subdir('kvm') subdir('xen')
hw/i386/tdvf.c new file +199 -0
199 lines not on this page · raw message
include/hw/i386/tdvf.h new file +51 -0
51 lines not on this page · raw message
Signature · 1 line
2.34.1
PATCH v2 28/58 i386/tdx: Parse TDVF metadata for TDX VM
In reply to Xiaoyao LiTDX cannot support pflash device since it doesn't support read-only memslot and doesn't support emulation. Load TDVF(OVMF) with -bios option for TDs.
When boot a TD, besides load TDVF to the address below 4G, it needs parse TDVF metadata.
--- hw/i386/pc_sysfw.c | 7 +++++++ hw/i386/x86.c | 3 ++- target/i386/kvm/tdx-stub.c | 5 +++++ target/i386/kvm/tdx.c | 5 +++++ target/i386/kvm/tdx.h | 4 ++++ 5 files changed, 23 insertions(+), 1 deletion(-)
hw/i386/pc_sysfw.c +7 -0
@@ -37,6 +37,7 @@#include "hw/block/flash.h" #include "sysemu/kvm.h" #include "sev.h" +#include "kvm/tdx.h" #define FLASH_SECTOR_SIZE 4096
@@ -265,5 +266,11 @@ void x86_firmware_configure(void *ptr, int size) }sev_encrypt_flash (ptr, size, &error_fatal); + }else if (is_tdx_vm ()) { + ret =tdx_parse_tdvf (ptr, size); +if (ret) { +error_report ("failed to parse TDVF for TDX VM" ); +exit (1 ); + } } }
hw/i386/x86.c +2 -1
@@ -47,6 +47,7 @@#include "hw/intc/i8259.h" #include "hw/rtc/mc146818rtc.h" #include "target/i386/sev.h" +#include "kvm/tdx.h" #include "hw/acpi/cpu_hotplug.h" #include "hw/irq.h"
@@ -1152,7 +1153,7 @@ void x86_bios_rom_init(MachineState *ms, const char *default_firmware, } bios =g_malloc (sizeof (*bios));memory_region_init_ram (bios,NULL ,"pc.bios" , bios_size, &error_fatal); -if (sev_enabled ()) { +if (sev_enabled () ||is_tdx_vm ()) {/* * The concept of a "reset" simply doesn't exist for * confidential computing guests, we have to destroy and
target/i386/kvm/tdx-stub.c +5 -0
@@ -11,3 +11,8 @@ int tdx_pre_create_vcpu(CPUState *cpu) {return -EINVAL ; } + +int tdx_parse_tdvf (void *flash_ptr,int size) +{ +return -EINVAL ; +}
target/i386/kvm/tdx.c +5 -0
@@ -580,6 +580,11 @@ out:return r; } +int tdx_parse_tdvf (void *flash_ptr,int size) +{ +return tdvf_parse_metadata (&tdx_guest->tdvf, flash_ptr, size); +} +static void tdx_guest_region_add (MemoryListener *listener,MemoryRegionSection *section) {
target/i386/kvm/tdx.h +4 -0
@@ -6,6 +6,7 @@#endif #include "exec/confidential-guest-support.h" +#include "hw/i386/tdvf.h" #define TYPE_TDX_GUEST "tdx-guest"#define TDX_GUEST (obj) OBJECT_CHECK(TdxGuest, (obj), TYPE_TDX_GUEST)
@@ -24,6 +25,8 @@ typedef struct TdxGuest {uint8_t mrconfigid[48 ];/* sha348 digest */ uint8_t mrowner[48 ];/* sha348 digest */ uint8_t mrownerconfig[48 ];/* sha348 digest */ + +TdxFirmware tdvf; } TdxGuest;#ifdef CONFIG_TDX
@@ -36,5 +39,6 @@ int tdx_kvm_init(MachineState *ms, Error **errp);void tdx_get_supported_cpuid (uint32_t function,uint32_t index,int reg,uint32_t *ret);int tdx_pre_create_vcpu (CPUState *cpu); +int tdx_parse_tdvf (void *flash_ptr,int size);#endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
PATCH v2 29/58 i386/tdx: Skip BIOS shadowing setup
In reply to Xiaoyao LiTDX doesn't support map different GPAs to same private memory. Thus, aliasing top 128KB of BIOS as isa-bios is not supported.
On the other hand, TDX guest cannot go to real mode, it can work fine without isa-bios.
---
Changes from RFC v4:
- update commit message and comment to clarify
--- hw/i386/x86.c | 25 ++++++++++++++----------- 1 file changed, 14 insertions(+), 11 deletions(-)
hw/i386/x86.c +14 -11
@@ -1175,17 +1175,20 @@ void x86_bios_rom_init(MachineState *ms, const char *default_firmware, }g_free (filename); -/* map the last 128KB of the BIOS in ISA space */ - isa_bios_size =MIN (bios_size,128 * KiB); - isa_bios =g_malloc (sizeof (*isa_bios)); -memory_region_init_alias (isa_bios,NULL ,"isa-bios" , bios, - bios_size - isa_bios_size, isa_bios_size); -memory_region_add_subregion_overlap (rom_memory, -0x100000 - isa_bios_size, - isa_bios, -1 ); -if (!isapc_ram_fw) { -memory_region_set_readonly (isa_bios, true); +/* For TDX, alias different GPAs to same private memory is not supported */ +if (!is_tdx_vm ()) { +/* map the last 128KB of the BIOS in ISA space */ + isa_bios_size =MIN (bios_size,128 * KiB); + isa_bios =g_malloc (sizeof (*isa_bios)); +memory_region_init_alias (isa_bios,NULL ,"isa-bios" , bios, + bios_size - isa_bios_size, isa_bios_size); +memory_region_add_subregion_overlap (rom_memory, +0x100000 - isa_bios_size, + isa_bios, +1 ); +if (!isapc_ram_fw) { +memory_region_set_readonly (isa_bios, true); + } }/* map all the bios at the top of memory */
Signature · 1 line
2.34.1
PATCH v2 30/58 i386/tdx: Don't initialize pc.rom for TDX VMs
In reply to Xiaoyao LiFor TDX, the address below 1MB are entirely general RAM. No need to initialize pc.rom memory region for TDs.
---
This is more as a workaround of the issue that for q35 machine type, the real memslot update (which requires memslot deletion )for pc.rom happens after tdx_init_memory_region. It leads to the private memory ADD'ed before get lost. I haven't work out a good solution to resolve the order issue. So just skip the pc.rom setup to avoid memslot deletion.
--- hw/i386/pc.c | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-)
hw/i386/pc.c +12 -9
@@ -62,6 +62,7 @@#include "sysemu/reset.h" #include "sysemu/runstate.h" #include "kvm/kvm_i386.h" +#include "kvm/tdx.h" #include "hw/xen/xen.h" #include "hw/xen/start_info.h" #include "ui/qemu-spice.h"
@@ -1095,16 +1096,18 @@ void pc_memory_init(PCMachineState *pcms,/* Initialize PC system firmware */ pc_system_firmware_init (pcms, rom_memory); - option_rom_mr =g_malloc (sizeof (*option_rom_mr)); -memory_region_init_ram (option_rom_mr,NULL ,"pc.rom" ,PC_ROM_SIZE , - &error_fatal); -if (pcmc->pci_enabled) { -memory_region_set_readonly (option_rom_mr, true); +if (!is_tdx_vm ()) { + option_rom_mr =g_malloc (sizeof (*option_rom_mr)); +memory_region_init_ram (option_rom_mr,NULL ,"pc.rom" ,PC_ROM_SIZE , + &error_fatal); +if (pcmc->pci_enabled) { +memory_region_set_readonly (option_rom_mr, true); + } +memory_region_add_subregion_overlap (rom_memory, +PC_ROM_MIN_VGA , + option_rom_mr, +1 ); } -memory_region_add_subregion_overlap (rom_memory, -PC_ROM_MIN_VGA , - option_rom_mr, -1 ); fw_cfg =fw_cfg_arch_create (machine, x86ms->boot_cpus, x86ms->apic_id_limit);
Signature · 1 line
2.34.1
PATCH v2 31/58 i386/tdx: Track mem_ptr for each firmware entry of TDVF
In reply to Xiaoyao LiFor each TDVF sections, QEMU needs to copy the content to guest private memory via KVM API (KVM_TDX_INIT_MEM_REGION).
Introduce a field @mem_ptr for TdxFirmwareEntry to track the memory pointer of each TDVF sections. So that QEMU can add/copy them to guest private memory later.
TDVF sections can be classified into two groups:
- Firmware itself, e.g., TDVF BFV and CFV, that located separately from guest RAM. Its memory pointer is the bios pointer.
- Sections located at guest RAM, e.g., TEMP_MEM and TD_HOB. mmap a new memory range for them.
Register a machine_init_done callback to do the stuff.
--- hw/i386/tdvf.c | 1 + include/hw/i386/tdvf.h | 7 +++++++ target/i386/kvm/tdx.c | 31 +++++++++++++++++++++++++++++++ 3 files changed, 39 insertions(+)
hw/i386/tdvf.c +1 -0
@@ -189,6 +189,7 @@ int tdvf_parse_metadata(TdxFirmware *fw, void *flash_ptr, int size) }g_free (sections); + fw->mem_ptr = flash_ptr;return 0 ; err:
include/hw/i386/tdvf.h +7 -0
@@ -39,13 +39,20 @@ typedef struct TdxFirmwareEntry {uint64_t size;uint32_t type;uint32_t attributes; + + void *mem_ptr; } TdxFirmwareEntry;typedef struct TdxFirmware { +void *mem_ptr; +uint32_t nr_entries;TdxFirmwareEntry *entries; }TdxFirmware ; +#define for_each_tdx_fw_entry (fw, e) \ + for (e = (fw)->entries; e != (fw)->entries + (fw)->nr_entries; e++) +int tdvf_parse_metadata (TdxFirmware *fw,void *flash_ptr,int size);#endif /* HW_I386_TDVF_H */
target/i386/kvm/tdx.c +31 -0
@@ -13,6 +13,7 @@#include "qemu/osdep.h" #include "qemu/error-report.h" +#include "qemu/mmap-alloc.h" #include "qapi/error.h" #include "qom/object_interfaces.h" #include "standard-headers/asm-x86/kvm_para.h"
@@ -22,6 +23,7 @@#include "exec/ramblock.h" #include "hw/i386/x86.h" +#include "hw/i386/tdvf.h" #include "kvm_i386.h" #include "tdx.h" #include "../cpu-internal.h"
@@ -452,6 +454,33 @@ static void update_tdx_cpuid_lookup_by_tdx_caps(void) (tdx_caps->xfam_fixed1 &CPUID_XSTATE_XSS_MASK ) >>32 ; } +static void tdx_finalize_vm (Notifier *notifier,void *unused) +{ +TdxFirmware *tdvf = &tdx_guest->tdvf; +TdxFirmwareEntry *entry; + +for_each_tdx_fw_entry (tdvf, entry) { +switch (entry->type) { +case TDVF_SECTION_TYPE_BFV : +case TDVF_SECTION_TYPE_CFV : + entry->mem_ptr = tdvf->mem_ptr + entry->data_offset; +break ; +case TDVF_SECTION_TYPE_TD_HOB : +case TDVF_SECTION_TYPE_TEMP_MEM : + entry->mem_ptr =qemu_ram_mmap (-1 , entry->size, +qemu_real_host_page_size (),0 ,0 ); +break ; +default : +error_report ("Unsupported TDVF section %d" , entry->type); +exit (1 ); + } + } +} + +static Notifier tdx_machine_done_notify = { + .notify = tdx_finalize_vm, +}; +int tdx_kvm_init (MachineState *ms,Error **errp) { TdxGuest *tdx = (TdxGuest *)object_dynamic_cast(OBJECT(ms->cgs),
@@ -472,6 +501,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp)*/ kvm_readonly_mem_allowed = false; +qemu_add_machine_init_done_notifier (&tdx_machine_done_notify); + tdx_guest = tdx;return 0 ;
Signature · 1 line
2.34.1
PATCH v2 32/58 i386/tdx: Track RAM entries for TDX VM
In reply to Xiaoyao LiThe RAM of TDX VM can be classified into two types:
- TDX_RAM_UNACCEPTED: default type of TDX memory, which needs to be accepted by TDX guest before it can be used and will be all-zeros after being accepted.
- TDX_RAM_ADDED: the RAM that is ADD'ed to TD guest before running, and can be used directly. E.g., TD HOB and TEMP MEM that needed by TDVF.
Maintain TdxRamEntries[] which grabs the initial RAM info from e820 table and mark each RAM range as default type TDX_RAM_UNACCEPTED.
Then turn the range of TD HOB and TEMP MEM to TDX_RAM_ADDED since these ranges will be ADD'ed before TD runs and no need to be accepted runtime.
The TdxRamEntries[] are later used to setup the memory TD resource HOB that passes memory info from QEMU to TDVF.
---
Changes from RFC v4:
- simplify the algorithm of tdx_accept_ram_range() (Suggested-by: Gerd Hoffman)
(1) Change the existing entry to cover the accepted ram range.
(2) If there is room before the accepted ram range add a TDX_RAM_UNACCEPTED entry for that.
(3) If there is room after the accepted ram range add a TDX_RAM_UNACCEPTED entry for that.
--- target/i386/kvm/tdx.c | 110 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 14 ++++++ 2 files changed, 124 insertions(+)
target/i386/kvm/tdx.c +110 -0
@@ -22,6 +22,7 @@#include "exec/address-spaces.h" #include "exec/ramblock.h" +#include "hw/i386/e820_memory_layout.h" #include "hw/i386/x86.h" #include "hw/i386/tdvf.h" #include "kvm_i386.h"
@@ -454,11 +455,116 @@ static void update_tdx_cpuid_lookup_by_tdx_caps(void) (tdx_caps->xfam_fixed1 &CPUID_XSTATE_XSS_MASK ) >>32 ; } +static void tdx_add_ram_entry (uint64_t address,uint64_t length,uint32_t type) +{ +uint32_t nr_entries = tdx_guest->nr_ram_entries; + tdx_guest->ram_entries =g_renew (TdxRamEntry, tdx_guest->ram_entries, + nr_entries +1 ); + + tdx_guest->ram_entries[nr_entries].address = address; + tdx_guest->ram_entries[nr_entries].length = length; + tdx_guest->ram_entries[nr_entries].type = type; + tdx_guest->nr_ram_entries++; +} + +static int tdx_accept_ram_range (uint64_t address,uint64_t length) +{ +uint64_t head_start, tail_start, head_length, tail_length; +uint64_t tmp_address, tmp_length; +TdxRamEntry *e; +int i; + +for (i =0 ; i < tdx_guest->nr_ram_entries; i++) { + e = &tdx_guest->ram_entries[i]; + +if (address + length <= e->address || + e->address + e->length <= address) { +continue ; + } + +/* +* The to-be-accepted ram range must be fully contained by one +* RAM entry. +*/ +if (e->address > address || + e->address + e->length < address + length) { +return -EINVAL ; + } + +if (e->type ==TDX_RAM_ADDED ) { +return -EINVAL ; + } + +break ; + } + +if (i == tdx_guest->nr_ram_entries) { +return -1 ; + } + + tmp_address = e->address; + tmp_length = e->length; + + e->address = address; + e->length = length; + e->type =TDX_RAM_ADDED ; + + head_length = address - tmp_address; +if (head_length >0 ) { + head_start = tmp_address; +tdx_add_ram_entry (head_start, head_length,TDX_RAM_UNACCEPTED ); + } + + tail_start = address + length; +if (tail_start < tmp_address + tmp_length) { + tail_length = tmp_address + tmp_length - tail_start; +tdx_add_ram_entry (tail_start, tail_length,TDX_RAM_UNACCEPTED ); + } + +return 0 ; +} + +static int tdx_ram_entry_compare (const void *lhs_,const void * rhs_) +{ +const TdxRamEntry *lhs = lhs_; +const TdxRamEntry *rhs = rhs_; + +if (lhs->address == rhs->address) { +return 0 ; + } +if (le64_to_cpu (lhs->address) >le64_to_cpu (rhs->address)) { +return 1 ; + } +return -1 ; +} + +static void tdx_init_ram_entries (void ) +{ +unsigned i, j, nr_e820_entries; + + nr_e820_entries =e820_get_num_entries (); + tdx_guest->ram_entries =g_new (TdxRamEntry, nr_e820_entries); + +for (i =0 , j =0 ; i < nr_e820_entries; i++) { +uint64_t addr, len; + +if (e820_get_entry (i,E820_RAM , &addr, &len)) { + tdx_guest->ram_entries[j].address = addr; + tdx_guest->ram_entries[j].length = len; + tdx_guest->ram_entries[j].type =TDX_RAM_UNACCEPTED ; + j++; + } + } + tdx_guest->nr_ram_entries = j; +} +static void tdx_finalize_vm (Notifier *notifier,void *unused) {TdxFirmware *tdvf = &tdx_guest->tdvf;TdxFirmwareEntry *entry; +tdx_init_ram_entries (); +for_each_tdx_fw_entry (tdvf, entry) {switch (entry->type) {case TDVF_SECTION_TYPE_BFV :
@@ -469,12 +575,16 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused)case TDVF_SECTION_TYPE_TEMP_MEM : entry->mem_ptr =qemu_ram_mmap (-1 , entry->size,qemu_real_host_page_size (),0 ,0 ); +tdx_accept_ram_range (entry->address, entry->size);break ;default :error_report ("Unsupported TDVF section %d" , entry->type);exit (1 ); } } + +qsort (tdx_guest->ram_entries, tdx_guest->nr_ram_entries, +sizeof (TdxRamEntry), &tdx_ram_entry_compare); }static Notifier tdx_machine_done_notify = {
target/i386/kvm/tdx.h +14 -0
@@ -15,6 +15,17 @@ typedef struct TdxGuestClass {ConfidentialGuestSupportClass parent_class; } TdxGuestClass; +enum TdxRamType { +TDX_RAM_UNACCEPTED , +TDX_RAM_ADDED , +}; + +typedef struct TdxRamEntry { +uint64_t address; +uint64_t length; +uint32_t type; +}TdxRamEntry ; + typedef struct TdxGuest { ConfidentialGuestSupport parent_obj;
@@ -27,6 +38,9 @@ typedef struct TdxGuest {uint8_t mrownerconfig[48 ];/* sha348 digest */ TdxFirmware tdvf; + +uint32_t nr_ram_entries; +TdxRamEntry *ram_entries; } TdxGuest;#ifdef CONFIG_TDX
Signature · 1 line
2.34.1
Re: [PATCH v2 32/58] i386/tdx: Track RAM entries for TDX VM
Quoted 41 lines · Xiaoyao Li
The RAM of TDX VM can be classified into two types:
-
TDX_RAM_UNACCEPTED: default type of TDX memory, which needs to be accepted by TDX guest before it can be used and will be all-zeros after being accepted.-
TDX_RAM_ADDED: the RAM that is ADD'ed to TD guest before running, and can be used directly. E.g., TD HOB and TEMP MEM that needed by TDVF.Maintain TdxRamEntries[] which grabs the initial RAM info from e820 table and mark each RAM range as default type
TDX_RAM_UNACCEPTED.Then turn the range of TD HOB and TEMP MEM to
TDX_RAM_ADDEDsince these ranges will be ADD'ed before TD runs and no need to be accepted runtime.The TdxRamEntries[] are later used to setup the memory TD resource HOB that passes memory info from QEMU to TDVF.
Signed-off-by Xiaoyao Li <[email protected]>Acked-by Gerd Hoffmann <[email protected]>---
Changes from RFC v4:
- simplify the algorithm oftdx_accept_ram_range()(Suggested-by: Gerd Hoffman)
(1) Change the existing entry to cover the accepted ram range.
(2) If there is room before the accepted ram range add aTDX_RAM_UNACCEPTEDentry for that.
(3) If there is room after the accepted ram range add aTDX_RAM_UNACCEPTEDentry for that.--- target/i386/kvm/tdx.c | 110 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 14 ++++++ 2 files changed, 124 insertions(+)target/i386/kvm/tdx.c +3
-0+static int tdx_accept_ram_range (uint64_t address,uint64_t length) +{ +uint64_t head_start, tail_start, head_length, tail_length;
target/i386/kvm/tdx.c static void update_tdx_cpuid_lookup_by_tdx_caps(void) · line 482
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+uint64_t tmp_address, tmp_length; +TdxRamEntry *e; +int i; + +for (i =0 ; i < tdx_guest->nr_ram_entries; i++) { + e = &tdx_guest->ram_entries[i]; + +if (address + length <= e->address || + e->address + e->length <= address) { +continue ;
Indented too far
Quoted 33 lines
target/i386/kvm/tdx.c +33
-0+ } + +/* +* The to-be-accepted ram range must be fully contained by one +* RAM entry. +*/ +if (e->address > address || + e->address + e->length < address + length) { +return -EINVAL ; + } + +if (e->type ==TDX_RAM_ADDED ) { +return -EINVAL ; + } + +break ; + } + +if (i == tdx_guest->nr_ram_entries) { +return -1 ; + } + + tmp_address = e->address; + tmp_length = e->length; + + e->address = address; + e->length = length; + e->type =TDX_RAM_ADDED ; + + head_length = address - tmp_address; +if (head_length >0 ) { + head_start = tmp_address; +tdx_add_ram_entry (head_start, head_length,TDX_RAM_UNACCEPTED );
target/i386/kvm/tdx.c static void update_tdx_cpuid_lookup_by_tdx_caps(void) · line 525
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+ } + + tail_start = address + length; +if (tail_start < tmp_address + tmp_length) { + tail_length = tmp_address + tmp_length - tail_start; +tdx_add_ram_entry (tail_start, tail_length,TDX_RAM_UNACCEPTED ); + } + +return 0 ; +}
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 32/58] i386/tdx: Track RAM entries for TDX VM
Quoted 10 lines · Daniel P. Berrangé
Quoted 10 lines
target/i386/kvm/tdx.c +6
-0+static int tdx_accept_ram_range (uint64_t address,uint64_t length) +{ +uint64_t head_start, tail_start, head_length, tail_length; +uint64_t tmp_address, tmp_length; +TdxRamEntry *e; +int i;
Quoted 9 lines
target/i386/kvm/tdx.c static void update_tdx_cpuid_lookup_by_tdx_caps(void) · line 482
Quoted 7 lines
target/i386/kvm/tdx.c +7
-0+ +for (i =0 ; i < tdx_guest->nr_ram_entries; i++) { + e = &tdx_guest->ram_entries[i]; + +if (address + length <= e->address || + e->address + e->length <= address) { +continue ;Indented too far
Fixed.
Thanks!
-Xiaoyao
Re: [PATCH v2 32/58] i386/tdx: Track RAM entries for TDX VM
In reply to Xiaoyao LiQuoted 7 lines · Xiaoyao Li
target/i386/kvm/tdx.h +0
-0@@ -15,6 +15,17 @@ typedef struct TdxGuestClass {ConfidentialGuestSupportClass parent_class; } TdxGuestClass;
target/i386/kvm/tdx.h typedef struct TdxGuestClass { · line 26
Quoted 10 lines
target/i386/kvm/tdx.h +9
-0+enum TdxRamType { +TDX_RAM_UNACCEPTED , +TDX_RAM_ADDED , +}; + +typedef struct TdxRamEntry { + uint64_t address; + uint64_t length; + uint32_t type;
nitpick: enum TdxRamType. and related function arguments.
Signature · 1 line
Isaku Yamahata <[email protected]>
Re: [PATCH v2 32/58] i386/tdx: Track RAM entries for TDX VM
Quoted 11 lines · Isaku Yamahata
Quoted 8 lines · Xiaoyao Li
target/i386/kvm/tdx.h +0
-0@@ -15,6 +15,17 @@ typedef struct TdxGuestClass {ConfidentialGuestSupportClass parent_class; } TdxGuestClass;
Quoted 12 lines
target/i386/kvm/tdx.h typedef struct TdxGuestClass { · line 26
Quoted 9 lines
target/i386/kvm/tdx.h +9
-0+enum TdxRamType { +TDX_RAM_UNACCEPTED , +TDX_RAM_ADDED , +}; + +typedef struct TdxRamEntry { + uint64_t address; + uint64_t length; + uint32_t type;nitpick: enum TdxRamType. and related function arguments.
Will do it.
Thanks!
PATCH v2 33/58 headers: Add definitions from UEFI spec for volumes, resources, etc...
In reply to Xiaoyao LiAdd UEFI definitions for literals, enums, structs, GUIDs, etc... that will be used by TDX to build the UEFI Hand-Off Block (HOB) that is passed to the Trusted Domain Virtual Firmware (TDVF).
All values come from the UEFI specification and TDVF design guide. [1]
Note, EFI_RESOURCE_MEMORY_UNACCEPTED will be added in future UEFI spec.
--- include/standard-headers/uefi/uefi.h | 198 +++++++++++++++++++++++++++ 1 file changed, 198 insertions(+) create mode 100644 include/standard-headers/uefi/uefi.h
include/standard-headers/uefi/uefi.h new file +198 -0
@@ -0,0 +1,198 @@ +/* +* Copyright (C) 2020 Intel Corporation +* +* Author: Isaku Yamahata <isaku.yamahata at gmail.com> +* <isaku.yamahata at intel.com> +* +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation; either version 2 of the License, or +* (at your option) any later version. + +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. + +* You should have received a copy of the GNU General Public License along +* with this program; if not, see <http://www.gnu.org/licenses/>. +* +*/ + +#ifndef HW_I386_UEFI_H +#define HW_I386_UEFI_H + +/***************************************************************************/ +/* +* basic EFI definitions +* supplemented with UEFI Specification Version 2.8 (Errata A) +* released February 2020 +*/ +/* UEFI integer is little endian */ + +typedef struct { +uint32_t Data1; +uint16_t Data2; +uint16_t Data3; +uint8_t Data4[8 ]; +}EFI_GUID ; + +typedef enum { + EfiReservedMemoryType, + EfiLoaderCode, + EfiLoaderData, + EfiBootServicesCode, + EfiBootServicesData, + EfiRuntimeServicesCode, + EfiRuntimeServicesData, + EfiConventionalMemory, + EfiUnusableMemory, + EfiACPIReclaimMemory, + EfiACPIMemoryNVS, + EfiMemoryMappedIO, + EfiMemoryMappedIOPortSpace, + EfiPalCode, + EfiPersistentMemory, + EfiUnacceptedMemoryType, + EfiMaxMemoryType +}EFI_MEMORY_TYPE ; + +#define EFI_HOB_HANDOFF_TABLE_VERSION 0x0009 + +#define EFI_HOB_TYPE_HANDOFF 0x0001 +#define EFI_HOB_TYPE_MEMORY_ALLOCATION 0x0002 +#define EFI_HOB_TYPE_RESOURCE_DESCRIPTOR 0x0003 +#define EFI_HOB_TYPE_GUID_EXTENSION 0x0004 +#define EFI_HOB_TYPE_FV 0x0005 +#define EFI_HOB_TYPE_CPU 0x0006 +#define EFI_HOB_TYPE_MEMORY_POOL 0x0007 +#define EFI_HOB_TYPE_FV2 0x0009 +#define EFI_HOB_TYPE_LOAD_PEIM_UNUSED 0x000A +#define EFI_HOB_TYPE_UEFI_CAPSULE 0x000B +#define EFI_HOB_TYPE_FV3 0x000C +#define EFI_HOB_TYPE_UNUSED 0xFFFE +#define EFI_HOB_TYPE_END_OF_HOB_LIST 0xFFFF + +typedef struct { +uint16_t HobType; +uint16_t HobLength; +uint32_t Reserved; +}EFI_HOB_GENERIC_HEADER ; + +typedef uint64_t EFI_PHYSICAL_ADDRESS ; +typedef uint32_t EFI_BOOT_MODE ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; +uint32_t Version; +EFI_BOOT_MODE BootMode; +EFI_PHYSICAL_ADDRESS EfiMemoryTop; +EFI_PHYSICAL_ADDRESS EfiMemoryBottom; +EFI_PHYSICAL_ADDRESS EfiFreeMemoryTop; +EFI_PHYSICAL_ADDRESS EfiFreeMemoryBottom; +EFI_PHYSICAL_ADDRESS EfiEndOfHobList; +}EFI_HOB_HANDOFF_INFO_TABLE ; + +#define EFI_RESOURCE_SYSTEM_MEMORY 0x00000000 +#define EFI_RESOURCE_MEMORY_MAPPED_IO 0x00000001 +#define EFI_RESOURCE_IO 0x00000002 +#define EFI_RESOURCE_FIRMWARE_DEVICE 0x00000003 +#define EFI_RESOURCE_MEMORY_MAPPED_IO_PORT 0x00000004 +#define EFI_RESOURCE_MEMORY_RESERVED 0x00000005 +#define EFI_RESOURCE_IO_RESERVED 0x00000006 +#define EFI_RESOURCE_MEMORY_UNACCEPTED 0x00000007 +#define EFI_RESOURCE_MAX_MEMORY_TYPE 0x00000008 + +#define EFI_RESOURCE_ATTRIBUTE_PRESENT 0x00000001 +#define EFI_RESOURCE_ATTRIBUTE_INITIALIZED 0x00000002 +#define EFI_RESOURCE_ATTRIBUTE_TESTED 0x00000004 +#define EFI_RESOURCE_ATTRIBUTE_SINGLE_BIT_ECC 0x00000008 +#define EFI_RESOURCE_ATTRIBUTE_MULTIPLE_BIT_ECC 0x00000010 +#define EFI_RESOURCE_ATTRIBUTE_ECC_RESERVED_1 0x00000020 +#define EFI_RESOURCE_ATTRIBUTE_ECC_RESERVED_2 0x00000040 +#define EFI_RESOURCE_ATTRIBUTE_READ_PROTECTED 0x00000080 +#define EFI_RESOURCE_ATTRIBUTE_WRITE_PROTECTED 0x00000100 +#define EFI_RESOURCE_ATTRIBUTE_EXECUTION_PROTECTED 0x00000200 +#define EFI_RESOURCE_ATTRIBUTE_UNCACHEABLE 0x00000400 +#define EFI_RESOURCE_ATTRIBUTE_WRITE_COMBINEABLE 0x00000800 +#define EFI_RESOURCE_ATTRIBUTE_WRITE_THROUGH_CACHEABLE 0x00001000 +#define EFI_RESOURCE_ATTRIBUTE_WRITE_BACK_CACHEABLE 0x00002000 +#define EFI_RESOURCE_ATTRIBUTE_16_BIT_IO 0x00004000 +#define EFI_RESOURCE_ATTRIBUTE_32_BIT_IO 0x00008000 +#define EFI_RESOURCE_ATTRIBUTE_64_BIT_IO 0x00010000 +#define EFI_RESOURCE_ATTRIBUTE_UNCACHED_EXPORTED 0x00020000 +#define EFI_RESOURCE_ATTRIBUTE_READ_ONLY_PROTECTED 0x00040000 +#define EFI_RESOURCE_ATTRIBUTE_READ_ONLY_PROTECTABLE 0x00080000 +#define EFI_RESOURCE_ATTRIBUTE_READ_PROTECTABLE 0x00100000 +#define EFI_RESOURCE_ATTRIBUTE_WRITE_PROTECTABLE 0x00200000 +#define EFI_RESOURCE_ATTRIBUTE_EXECUTION_PROTECTABLE 0x00400000 +#define EFI_RESOURCE_ATTRIBUTE_PERSISTENT 0x00800000 +#define EFI_RESOURCE_ATTRIBUTE_PERSISTABLE 0x01000000 +#define EFI_RESOURCE_ATTRIBUTE_MORE_RELIABLE 0x02000000 + +typedef uint32_t EFI_RESOURCE_TYPE ; +typedef uint32_t EFI_RESOURCE_ATTRIBUTE_TYPE ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; +EFI_GUID Owner; +EFI_RESOURCE_TYPE ResourceType; +EFI_RESOURCE_ATTRIBUTE_TYPE ResourceAttribute; +EFI_PHYSICAL_ADDRESS PhysicalStart; +uint64_t ResourceLength; +}EFI_HOB_RESOURCE_DESCRIPTOR ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; +EFI_GUID Name; + +/* guid specific data follows */ +}EFI_HOB_GUID_TYPE ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; +EFI_PHYSICAL_ADDRESS BaseAddress; +uint64_t Length; +}EFI_HOB_FIRMWARE_VOLUME ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; +EFI_PHYSICAL_ADDRESS BaseAddress; +uint64_t Length; +EFI_GUID FvName; +EFI_GUID FileName; +}EFI_HOB_FIRMWARE_VOLUME2 ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; +EFI_PHYSICAL_ADDRESS BaseAddress; +uint64_t Length; +uint32_t AuthenticationStatus; +bool ExtractedFv; +EFI_GUID FvName; +EFI_GUID FileName; +}EFI_HOB_FIRMWARE_VOLUME3 ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; +uint8_t SizeOfMemorySpace; +uint8_t SizeOfIoSpace; +uint8_t Reserved[6 ]; +}EFI_HOB_CPU ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; +}EFI_HOB_MEMORY_POOL ; + +typedef struct { +EFI_HOB_GENERIC_HEADER Header; + +EFI_PHYSICAL_ADDRESS BaseAddress; +uint64_t Length; +}EFI_HOB_UEFI_CAPSULE ; + +#define EFI_HOB_OWNER_ZERO \ + ((EFI_GUID){ 0x00000000, 0x0000, 0x0000, \ + { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 } }) + +#endif
Signature · 1 line
2.34.1
Re: [PATCH v2 33/58] headers: Add definitions from UEFI spec for volumes, resources, etc...
Quoted 9 lines · Xiaoyao Li
Add UEFI definitions for literals, enums, structs, GUIDs, etc... that will be used by TDX to build the UEFI Hand-Off Block (HOB) that is passed to the Trusted Domain Virtual Firmware (TDVF).
All values come from the UEFI specification and TDVF design guide. [1]
Note,
EFI_RESOURCE_MEMORY_UNACCEPTEDwill be added in future UEFI spec.
Nitpick: The specs [1] [2] include unaccepted memory.
[1] UEFI Specification Version 2.10 (released August 2022)
[2] UEFI Platform Initialization Distribution Packaging Specification Version 1.1)
Signature · 1 line
Isaku Yamahata <[email protected]>
Re: [PATCH v2 33/58] headers: Add definitions from UEFI spec for volumes, resources, etc...
Quoted 7 lines · Isaku Yamahata
Quoted 4 lines · Xiaoyao Li
Add UEFI definitions for literals, enums, structs, GUIDs, etc... that will be used by TDX to build the UEFI Hand-Off Block (HOB) that is passed to the Trusted Domain Virtual Firmware (TDVF).
Quoted 7 lines
Quoted 5 lines
All values come from the UEFI specification and TDVF design guide. [1]
Note,
EFI_RESOURCE_MEMORY_UNACCEPTEDwill be added in future UEFI spec.Nitpick: The specs [1] [2] include unaccepted memory.
EfiUnacceptedMemoryType shows in UEFI spec while EFI_RESOURCE_MEMORY_UNACCEPTED is still missing in PI spec.
https://github.com/tianocore/edk2/commit/00bbb1e584ec05547159f405cca383e8ba5e4ddb
Quoted 2 lines
[1] UEFI Specification Version 2.10 (released August 2022)
[2] UEFI Platform Initialization Distribution Packaging Specification Version 1.1)
Re: [PATCH v2 33/58] headers: Add definitions from UEFI spec for volumes, resources, etc...
Quoted 12 lines · Xiaoyao Li
Quoted 11 lines · Isaku Yamahata
Quoted 8 lines · Xiaoyao Li
Add UEFI definitions for literals, enums, structs, GUIDs, etc... that will be used by TDX to build the UEFI Hand-Off Block (HOB) that is passed to the Trusted Domain Virtual Firmware (TDVF).
All values come from the UEFI specification and TDVF design guide. [1]
Note,
EFI_RESOURCE_MEMORY_UNACCEPTEDwill be added in future UEFI spec.
Quoted 7 lines
Quoted 3 lines
Quoted 1 line
Nitpick: The specs [1] [2] include unaccepted memory.
EfiUnacceptedMemoryType shows in UEFI spec while
EFI_RESOURCE_MEMORY_UNACCEPTEDis still missing in PI spec.https://github.com/tianocore/edk2/commit/00bbb1e584ec05547159f405cca383e8ba5e4ddb
Sorry, I just find it shows in latest PI spec.
https://uefi.org/sites/default/files/resources/UEFI_PI_Spec_1_8_March3.pdf
Quoted 4 lines
Quoted 2 lines
[1] UEFI Specification Version 2.10 (released August 2022)
[2] UEFI Platform Initialization Distribution Packaging Specification Version 1.1)
PATCH v2 34/58 i386/tdx: Setup the TD HOB list
In reply to Xiaoyao LiThe TD HOB list is used to pass the information from VMM to TDVF. The TD HOB must include PHIT HOB and Resource Descriptor HOB. More details can be found in TDVF specification and PI specification.
Build the TD HOB in TDX's machine_init_done callback.
---
Changes from RFC v4:
- drop the code of adding mmio resources since OVMF prepares all the MMIO hob itself.
--- hw/i386/meson.build | 2 +- hw/i386/tdvf-hob.c | 147 ++++++++++++++++++++++++++++++++++++++++++ hw/i386/tdvf-hob.h | 24 +++++++ target/i386/kvm/tdx.c | 16 +++++ 4 files changed, 188 insertions(+), 1 deletion(-) create mode 100644 hw/i386/tdvf-hob.c create mode 100644 hw/i386/tdvf-hob.h
hw/i386/meson.build +1 -1
@@ -28,7 +28,7 @@ i386_ss.add(when: 'CONFIG_PC', if_true: files( 'port92.c')) i386_ss.add(when: 'CONFIG_X86_FW_OVMF', if_true: files('pc_sysfw_ovmf.c'), if_false: files('pc_sysfw_ovmf-stubs.c')) -i386_ss.add(when: 'CONFIG_TDX', if_true: files('tdvf.c')) +i386_ss.add(when: 'CONFIG_TDX', if_true: files('tdvf.c', 'tdvf-hob.c')) subdir('kvm') subdir('xen')
hw/i386/tdvf-hob.c new file +147 -0
147 lines not on this page · raw message
hw/i386/tdvf-hob.h new file +24 -0
@@ -0,0 +1,24 @@ +#ifndef HW_I386_TD_HOB_H +#define HW_I386_TD_HOB_H + +#include "hw/i386/tdvf.h" +#include "target/i386/kvm/tdx.h" + +void tdvf_hob_create (TdxGuest *tdx,TdxFirmwareEntry *td_hob); + +#define EFI_RESOURCE_ATTRIBUTE_TDVF_PRIVATE \ + (EFI_RESOURCE_ATTRIBUTE_PRESENT | \ + EFI_RESOURCE_ATTRIBUTE_INITIALIZED | \ + EFI_RESOURCE_ATTRIBUTE_TESTED) + +#define EFI_RESOURCE_ATTRIBUTE_TDVF_UNACCEPTED \ + (EFI_RESOURCE_ATTRIBUTE_PRESENT | \ + EFI_RESOURCE_ATTRIBUTE_INITIALIZED | \ + EFI_RESOURCE_ATTRIBUTE_TESTED) + +#define EFI_RESOURCE_ATTRIBUTE_TDVF_MMIO \ + (EFI_RESOURCE_ATTRIBUTE_PRESENT | \ + EFI_RESOURCE_ATTRIBUTE_INITIALIZED | \ + EFI_RESOURCE_ATTRIBUTE_UNCACHEABLE) + +#endif
target/i386/kvm/tdx.c +16 -0
@@ -25,6 +25,7 @@#include "hw/i386/e820_memory_layout.h" #include "hw/i386/x86.h" #include "hw/i386/tdvf.h" +#include "hw/i386/tdvf-hob.h" #include "kvm_i386.h" #include "tdx.h" #include "../cpu-internal.h"
@@ -455,6 +456,19 @@ static void update_tdx_cpuid_lookup_by_tdx_caps(void) (tdx_caps->xfam_fixed1 &CPUID_XSTATE_XSS_MASK ) >>32 ; } +static TdxFirmwareEntry *tdx_get_hob_entry (TdxGuest *tdx) +{ +TdxFirmwareEntry *entry; + +for_each_tdx_fw_entry (&tdx->tdvf, entry) { +if (entry->type ==TDVF_SECTION_TYPE_TD_HOB ) { +return entry; + } + } +error_report ("TDVF metadata doesn't specify TD_HOB location." ); +exit (1 ); +} +static void tdx_add_ram_entry (uint64_t address,uint64_t length,uint32_t type) {uint32_t nr_entries = tdx_guest->nr_ram_entries;
@@ -585,6 +599,8 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused)qsort (tdx_guest->ram_entries, tdx_guest->nr_ram_entries,sizeof (TdxRamEntry), &tdx_ram_entry_compare); + +tdvf_hob_create (tdx_guest,tdx_get_hob_entry (tdx_guest)); }static Notifier tdx_machine_done_notify = {
Signature · 1 line
2.34.1
PATCH v2 35/58 i386/tdx: Add TDVF memory via KVM_TDX_INIT_MEM_REGION
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
TDVF firmware (CODE and VARS) needs to be added/copied to TD's private memory via KVM_TDX_INIT_MEM_REGION, as well as TD HOB and TEMP memory.
---
Changes from RFC v4:
- rename variable @metadata to @flags
--- target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+)
target/i386/kvm/tdx.c +24 -0
@@ -576,6 +576,7 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused) {TdxFirmware *tdvf = &tdx_guest->tdvf;TdxFirmwareEntry *entry; +int r;tdx_init_ram_entries ();
@@ -601,6 +602,29 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused) sizeof(TdxRamEntry), &tdx_ram_entry_compare);tdvf_hob_create (tdx_guest,tdx_get_hob_entry (tdx_guest)); + +for_each_tdx_fw_entry (tdvf, entry) { +struct kvm_tdx_init_mem_region mem_region = { + .source_addr = (__u64 )entry->mem_ptr, + .gpa = entry->address, + .nr_pages = entry->size /4096 , + }; + +__u32 flags = entry->attributes &TDVF_SECTION_ATTRIBUTES_MR_EXTEND ? +KVM_TDX_MEASURE_MEMORY_REGION :0 ; + + r =tdx_vm_ioctl (KVM_TDX_INIT_MEM_REGION , flags, &mem_region); +if (r <0 ) { +error_report ("KVM_TDX_INIT_MEM_REGION failed %s" ,strerror (-r)); +exit (1 ); + } + +if (entry->type ==TDVF_SECTION_TYPE_TD_HOB || + entry->type ==TDVF_SECTION_TYPE_TEMP_MEM ) { +qemu_ram_munmap (-1 , entry->mem_ptr, entry->size); + entry->mem_ptr =NULL ; + } + } }static Notifier tdx_machine_done_notify = {
Signature · 1 line
2.34.1
PATCH v2 36/58 memory: Introduce memory_region_init_ram_gmem()
In reply to Xiaoyao LiIntroduce memory_region_init_ram_gmem() to allocate private gmem on the MemoryRegion initialization. It's for the usercase of TDVF, which must be private on TDX case.
--- include/exec/memory.h | 6 +++++ softmmu/memory.c | 52 +++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 58 insertions(+)
include/exec/memory.h +6 -0
@@ -1564,6 +1564,12 @@ void memory_region_init_ram(MemoryRegion *mr,uint64_t size, Error **errp); +void memory_region_init_ram_gmem (MemoryRegion *mr, +Object *owner, +const char *name, +uint64_t size, +Error **errp); +/** * memory_region_init_rom: Initialize a ROM memory region. *
softmmu/memory.c +52 -0
64 lines not on this page · raw message
Signature · 1 line
2.34.1
Re: [PATCH v2 36/58] memory: Introduce memory_region_init_ram_gmem()
Quoted 23 lines · Xiaoyao Li
Introduce
memory_region_init_ram_gmem()to allocate private gmem on the MemoryRegion initialization. It's for the usercase of TDVF, which must be private on TDX case.Signed-off-by Xiaoyao Li <[email protected]>--- include/exec/memory.h | 6 +++++ softmmu/memory.c | 52 +++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 58 insertions(+)softmmu/memory.c +1
-0@@ -25,6 +25,7 @@#include "qom/object.h" #include "trace.h" +#include <linux/kvm.h> #include "exec/memory-internal.h" #include "exec/ram_addr.h" #include "sysemu/kvm.h"
softmmu/memory.c void memory_region_init_ram(MemoryRegion *mr, · line 3611
Quoted 9 lines
softmmu/memory.c +6
-0@@ -3602,6 +3603,57 @@ void memory_region_init_ram(MemoryRegion *mr,vmstate_register_ram (mr, owner_dev); } +#ifdef CONFIG_KVM +void memory_region_init_ram_gmem (MemoryRegion *mr, +Object *owner, +const char *name, +uint64_t size, +Error **errp)
Since you have an 'errp' parameter here....
Quoted 14 lines
softmmu/memory.c +14
-0+{ +DeviceState *owner_dev; +Error *err =NULL ; +int priv_fd; + +memory_region_init_ram_nomigrate (mr, owner, name, size, &err); +if (err) { +error_propagate (errp, err); +return ; + } + +if (object_dynamic_cast (OBJECT (current_accel ()),TYPE_KVM_ACCEL )) { +KVMState *s =KVM_STATE (current_accel ()); + struct kvm_create_guest_memfd gmem = {
softmmu/memory.c void memory_region_init_ram(MemoryRegion *mr, · line 3635
Quoted 10 lines
softmmu/memory.c +10
-0+ .size = size, + /* TODO: add property to hostmem backend for huge pmd */ + .flags = KVM_GUEST_MEMFD_ALLOW_HUGEPAGE, + }; + + priv_fd =kvm_vm_ioctl (s,KVM_CREATE_GUEST_MEMFD , &gmem); +if (priv_fd <0 ) { +fprintf (stderr,"%s: error creating gmem: %s\n" , __func__, +strerror (-priv_fd)); +abort ();
It should be using error_setg_errno() here and return not abort
softmmu/memory.c void memory_region_init_ram(MemoryRegion *mr, · line 3639
Quoted 4 lines
softmmu/memory.c +4
-0+ } + }else { +fprintf (stderr,"%s: gmem unsupported accel: %s\n" , __func__, +current_accel_name ());
and error_setg() here and return.
Quoted 23 lines
softmmu/memory.c +17
-0+ abort(); + } + +memory_region_set_gmem_fd (mr, priv_fd); +memory_region_set_default_private (mr); + +/* This will assert if owner is neither NULL nor a DeviceState. +* We only want the owner here for the purposes of defining a +* unique name for migration. TODO: Ideally we should implement +* a naming scheme for Objects which are not DeviceStates, in +* which case we can relax this restriction. +*/ + owner_dev =DEVICE (owner); +vmstate_register_ram (mr, owner_dev); +} +#endif + void memory_region_init_rom(MemoryRegion *mr, Object *owner, const char *name,Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 36/58] memory: Introduce memory_region_init_ram_gmem()
In reply to Xiaoyao LiQuoted 16 lines · Xiaoyao Li
Introduce
memory_region_init_ram_gmem()to allocate private gmem on the MemoryRegion initialization. It's for the usercase of TDVF, which must be private on TDX case.Signed-off-by Xiaoyao Li <[email protected]>--- include/exec/memory.h | 6 +++++ softmmu/memory.c | 52 +++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 58 insertions(+)softmmu/memory.c +0
-0
softmmu/memory.c line 28
Quoted 4 lines
softmmu/memory.c +1
-0@@ -25,6 +25,7 @@#include "qom/object.h" #include "trace.h" +#include <linux/kvm.h>
Unlikely to build on non-Linux hosts.
Quoted 3 lines
softmmu/memory.c +0
-0#include "exec/memory-internal.h" #include "exec/ram_addr.h" #include "sysemu/kvm.h"
Re: [PATCH v2 36/58] memory: Introduce memory_region_init_ram_gmem()
Quoted 12 lines · Philippe Mathieu-Daudé
Quoted 9 lines · Xiaoyao Li
Introduce
memory_region_init_ram_gmem()to allocate private gmem on the MemoryRegion initialization. It's for the usercase of TDVF, which must be private on TDX case.Signed-off-by Xiaoyao Li <[email protected]>---
include/exec/memory.h| 6 +++++
softmmu/memory.c | 52 +++++++++++++++++++++++++++++++++++++++++++2 files changed, 58 insertions(+)
Quoted 10 lines
softmmu/memory.c line 27
Quoted 8 lines
softmmu/memory.c +1
-0@@ -25,6 +25,7 @@ #include "qom/object.h" #include "trace.h" +#include <linux/kvm.h> Unlikely to build on non-Linux hosts.
Thanks for catching it!
Will warp it with CONFIG_KVM.
Anyway, it's the main open of how to integrating KVM gmem into QEMU's memory system, in QMEU gmem series[*]. I'm still working on it.
[*] https://lore.kernel.org/qemu-devel/[email protected]/
Quoted 4 lines
Quoted 3 lines
softmmu/memory.c +0
-0#include "exec/memory-internal.h" #include "exec/ram_addr.h" #include "sysemu/kvm.h"
PATCH v2 37/58 i386/tdx: register TDVF as private memory
In reply to Xiaoyao LiFrom: Chao Peng <[email protected]>
Allocate private gmem memory for BIOS if it's TD VM.
--- hw/i386/x86.c | 9 ++++++++- target/i386/kvm/tdx.c | 17 +++++++++++++++++ target/i386/kvm/tdx.h | 2 ++ 3 files changed, 27 insertions(+), 1 deletion(-)
hw/i386/x86.c +8 -1
@@ -1151,8 +1151,15 @@ void x86_bios_rom_init(MachineState *ms, const char *default_firmware, (bios_size % 65536) != 0) { goto bios_error; } + bios =g_malloc (sizeof (*bios)); -memory_region_init_ram (bios,NULL ,"pc.bios" , bios_size, &error_fatal); +if (is_tdx_vm ()) { +memory_region_init_ram_gmem (bios,NULL ,"pc.bios" , bios_size, &error_fatal); +tdx_set_tdvf_region (bios); + }else { +memory_region_init_ram (bios,NULL ,"pc.bios" , bios_size, &error_fatal); + } +if (sev_enabled () ||is_tdx_vm ()) {/* * The concept of a "reset" simply doesn't exist for
target/i386/kvm/tdx.c +17 -0
@@ -456,6 +456,12 @@ static void update_tdx_cpuid_lookup_by_tdx_caps(void) (tdx_caps->xfam_fixed1 &CPUID_XSTATE_XSS_MASK ) >>32 ; } +void tdx_set_tdvf_region (MemoryRegion *tdvf_region) +{ +assert (!tdx_guest->tdvf_region); + tdx_guest->tdvf_region = tdvf_region; +} +static TdxFirmwareEntry *tdx_get_hob_entry (TdxGuest *tdx) {TdxFirmwareEntry *entry;
@@ -576,6 +582,7 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused) {TdxFirmware *tdvf = &tdx_guest->tdvf;TdxFirmwareEntry *entry; +RAMBlock *ram_block;int r;tdx_init_ram_entries ();
@@ -610,6 +617,12 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused) .nr_pages = entry->size / 4096, }; + r =kvm_set_memory_attributes_private (entry->address, entry->size); +if (r <0 ) { +error_report ("Reserve initial private memory failed %s" ,strerror (-r)); +exit (1 ); + } +__u32 flags = entry->attributes &TDVF_SECTION_ATTRIBUTES_MR_EXTEND ?KVM_TDX_MEASURE_MEMORY_REGION :0 ;
@@ -625,6 +638,10 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused) entry->mem_ptr =NULL ; } } + +/* Tdvf image was copied into private region above. It becomes unnecessary. */ + ram_block = tdx_guest->tdvf_region->ram_block; +ram_block_discard_range (ram_block,0 , ram_block->max_length); }static Notifier tdx_machine_done_notify = {
target/i386/kvm/tdx.h +2 -0
@@ -38,6 +38,7 @@ typedef struct TdxGuest {uint8_t mrownerconfig[48 ];/* sha348 digest */ TdxFirmware tdvf; +MemoryRegion *tdvf_region;uint32_t nr_ram_entries;TdxRamEntry *ram_entries;
@@ -53,6 +54,7 @@ int tdx_kvm_init(MachineState *ms, Error **errp);void tdx_get_supported_cpuid (uint32_t function,uint32_t index,int reg,uint32_t *ret);int tdx_pre_create_vcpu (CPUState *cpu); +void tdx_set_tdvf_region (MemoryRegion *tdvf_region);int tdx_parse_tdvf (void *flash_ptr,int size);#endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
PATCH v2 38/58 i386/tdx: Call KVM_TDX_INIT_VCPU to initialize TDX vcpu
In reply to Xiaoyao LiTDX vcpu needs to be initialized by SEAMCALL(TDH.VP.INIT) and KVM provides vcpu level IOCTL KVM_TDX_INIT_VCPU for it.
KVM_TDX_INIT_VCPU needs the address of the HOB as input. Invoke it for each vcpu after HOB list is created.
--- target/i386/kvm/tdx.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+)
target/i386/kvm/tdx.c +18 -0
@@ -578,6 +578,22 @@ static void tdx_init_ram_entries(void) tdx_guest->nr_ram_entries = j; } +static void tdx_post_init_vcpus (void ) +{ +TdxFirmwareEntry *hob; +CPUState *cpu; +int r; + + hob =tdx_get_hob_entry (tdx_guest); +CPU_FOREACH (cpu) { + r =tdx_vcpu_ioctl (cpu,KVM_TDX_INIT_VCPU ,0 , (void *)hob->address); +if (r <0 ) { +error_report ("KVM_TDX_INIT_VCPU failed %s" ,strerror (-r)); +exit (1 ); + } + } +} +static void tdx_finalize_vm (Notifier *notifier,void *unused) {TdxFirmware *tdvf = &tdx_guest->tdvf;
@@ -610,6 +626,8 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused)tdvf_hob_create (tdx_guest,tdx_get_hob_entry (tdx_guest)); +tdx_post_init_vcpus (); +for_each_tdx_fw_entry (tdvf, entry) {struct kvm_tdx_init_mem_region mem_region = { .source_addr = (__u64 )entry->mem_ptr,
Signature · 1 line
2.34.1
PATCH v2 39/58 i386/tdx: Finalize TDX VM
In reply to Xiaoyao LiInvoke KVM_TDX_FINALIZE_VM to finalize the TD's measurement and make the TD vCPUs runnable once machine initialization is complete.
--- target/i386/kvm/tdx.c | 7 +++++++ 1 file changed, 7 insertions(+)
target/i386/kvm/tdx.c +7 -0
@@ -660,6 +660,13 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused)/* Tdvf image was copied into private region above. It becomes unnecessary. */ ram_block = tdx_guest->tdvf_region->ram_block;ram_block_discard_range (ram_block,0 , ram_block->max_length); + + r =tdx_vm_ioctl (KVM_TDX_FINALIZE_VM ,0 ,NULL ); +if (r <0 ) { +error_report ("KVM_TDX_FINALIZE_VM failed %s" ,strerror (-r)); +exit (0 ); + } + tdx_guest->parent_obj.ready = true; }static Notifier tdx_machine_done_notify = {
Signature · 1 line
2.34.1
PATCH v2 40/58 i386/tdx: handle TDG.VP.VMCALL<SetupEventNotifyInterrupt>
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
For SetupEventNotifyInterrupt, record interrupt vector and the apic id of the vcpu that received this TDVMCALL.
Later it can inject interrupt with given vector to the specific vcpu that received SetupEventNotifyInterrupt.
--- target/i386/kvm/kvm.c | 9 ++++++ target/i386/kvm/tdx-stub.c | 4 +++ target/i386/kvm/tdx.c | 61 ++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 6 ++++ 4 files changed, 80 insertions(+)
target/i386/kvm/kvm.c +9 -0
@@ -5657,6 +5657,15 @@ int kvm_arch_handle_exit(CPUState *cs, struct kvm_run *run) ret =kvm_xen_handle_exit (cpu, &run->xen);break ;#endif +case KVM_EXIT_TDX : +if (!is_tdx_vm ()) { +error_report ("KVM: get KVM_EXIT_TDX for a non-TDX VM." ); + ret =-1 ; +break ; + } +tdx_handle_exit (cpu, &run->tdx); + ret =0 ; +break ;default :fprintf (stderr,"KVM: unknown exit reason %d\n" , run->exit_reason); ret =-1 ;
target/i386/kvm/tdx-stub.c +4 -0
@@ -16,3 +16,7 @@ int tdx_parse_tdvf(void *flash_ptr, int size) {return -EINVAL ; } + +void tdx_handle_exit (X86CPU *cpu,struct kvm_tdx_exit *tdx_exit) +{ +}
target/i386/kvm/tdx.c +61 -0
70 lines not on this page · raw message
target/i386/kvm/tdx.h +6 -0
@@ -7,6 +7,7 @@#include "exec/confidential-guest-support.h" #include "hw/i386/tdvf.h" +#include "sysemu/kvm.h" #define TYPE_TDX_GUEST "tdx-guest"#define TDX_GUEST (obj) OBJECT_CHECK(TdxGuest, (obj), TYPE_TDX_GUEST)
@@ -42,6 +43,10 @@ typedef struct TdxGuest {uint32_t nr_ram_entries;TdxRamEntry *ram_entries; + +/* runtime state */ +int event_notify_interrupt; +uint32_t event_notify_apic_id; } TdxGuest;#ifdef CONFIG_TDX
@@ -56,5 +61,6 @@ void tdx_get_supported_cpuid(uint32_t function, uint32_t index, int reg,int tdx_pre_create_vcpu (CPUState *cpu);void tdx_set_tdvf_region (MemoryRegion *tdvf_region);int tdx_parse_tdvf (void *flash_ptr,int size); +void tdx_handle_exit (X86CPU *cpu,struct kvm_tdx_exit *tdx_exit);#endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
PATCH v2 41/58 i386/tdx: handle TDG.VP.VMCALL<GetQuote>
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \
-object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \
-machine confidential-guest-support=tdx0--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)
qapi/qom.json +4 -1
@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS) +# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', 'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } } ## # @ThreadContextProperties:
target/i386/kvm/tdx.c +380 -0
416 lines not on this page · raw message
target/i386/kvm/tdx.h +7 -0
@@ -5,8 +5,10 @@#include CONFIG_DEVICES /* CONFIG_TDX */ #endif +#include <linux/kvm.h> #include "exec/confidential-guest-support.h" #include "hw/i386/tdvf.h" +#include "io/channel-socket.h" #include "sysemu/kvm.h" #define TYPE_TDX_GUEST "tdx-guest"
@@ -47,6 +49,11 @@ typedef struct TdxGuest {/* runtime state */ int event_notify_interrupt;uint32_t event_notify_apic_id; + +/* GetQuote */ +int quote_generation_num; +char *quote_generation_str; +SocketAddress *quote_generation; } TdxGuest;#ifdef CONFIG_TDX
Signature · 1 line
2.34.1
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
Quoted 34 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +1
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS)
qapi/qom.json line 891
Quoted 10 lines
qapi/qom.json +3
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', 'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }
Why not type SocketAddress?
Quoted 3 lines
qapi/qom.json +0
-0## # @ThreadContextProperties:
[...]
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
Quoted 40 lines · Markus Armbruster
Quoted 38 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +2
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS) +# # Since: 8.2 ## { 'struct': 'TdxGuestProperties',
Quoted 8 lines
qapi/qom.json line 891
Quoted 6 lines
qapi/qom.json +2
-1'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }Why not type SocketAddress?
Yes, the code uses SocketAddress internally when it eventually calls qio_channel_socket_connect_async(), so we should directly use SocketAddress in the QAPI from the start.
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
Quoted 50 lines · Daniel P. Berrangé
Quoted 40 lines · Markus Armbruster
Quoted 38 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +2
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS) +# # Since: 8.2 ## { 'struct': 'TdxGuestProperties',Quoted 8 lines
qapi/qom.json line 891
Quoted 6 lines
qapi/qom.json +2
-1'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }Why not type SocketAddress?
Quoted 3 lines
Yes, the code uses SocketAddress internally when it eventually calls
qio_channel_socket_connect_async(), so we should directly use SocketAddress in the QAPI from the start.
Any benefit to directly use SocketAddress?
"quote-generation-service" here is optional, it seems not trivial to add and parse the SocketAddress type in QEMU command. After I change 'str' to 'SocketAddress' and specify the command like "-object tdx-guest,type=vsock,cid=2,port=1234...", it will report "invalid parameter cid".
Quoted 3 lines
With regards,
Daniel
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
Quoted 53 lines · Chenyi Qiang
Quoted 50 lines · Daniel P. Berrangé
Quoted 48 lines · Markus Armbruster
Quoted 34 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +1
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS)qapi/qom.json line 891
Quoted 10 lines
qapi/qom.json +3
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', 'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }Why not type SocketAddress?
Quoted 5 lines
Quoted 3 lines
Yes, the code uses SocketAddress internally when it eventually calls
qio_channel_socket_connect_async(), so we should directly use SocketAddress in the QAPI from the start.Any benefit to directly use SocketAddress?
We don't want whatever code consumes the configuration to do a second level of parsing to convert the 'str' value into the 'SocketAddress' object it actually needs.
QEMU has a long history of having a second round of ad-hoc parsing of configuration and we've found it to be a serious maintenence burden. Thus we strive to have everything represented in QAPI using the desired final type, and avoid the second round of parsing.
Quoted 5 lines
"quote-generation-service" here is optional, it seems not trivial to add and parse the SocketAddress type in QEMU command. After I change 'str' to 'SocketAddress' and specify the command like "-object tdx-guest,type=vsock,cid=2,port=1234...", it will report "invalid parameter cid".
The -object parameter supports JSON syntax for this reason
-object '{"qom-type":"tdx-guest","quote-generation-service":{"type": "vsock", "cid":"2","port":"1234"}}'
libvirt will always use the JSON syntax for -object with a new enough QEMU.
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
Quoted 64 lines · Daniel P. Berrangé
Quoted 53 lines · Chenyi Qiang
Quoted 50 lines · Daniel P. Berrangé
Quoted 48 lines · Markus Armbruster
Quoted 34 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +1
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS)qapi/qom.json line 891
Quoted 10 lines
qapi/qom.json +3
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', 'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }Why not type SocketAddress?
Quoted 5 lines
Quoted 3 lines
Yes, the code uses SocketAddress internally when it eventually calls
qio_channel_socket_connect_async(), so we should directly use SocketAddress in the QAPI from the start.Any benefit to directly use SocketAddress?
We don't want whatever code consumes the configuration to do a second level of parsing to convert the 'str' value into the 'SocketAddress' object it actually needs.
Quoted 5 lines
QEMU has a long history of having a second round of ad-hoc parsing of configuration and we've found it to be a serious maintenence burden. Thus we strive to have everything represented in QAPI using the desired final type, and avoid the second round of parsing.
Thanks for your explanation.
Quoted 7 lines
Quoted 6 lines
"quote-generation-service" here is optional, it seems not trivial to add and parse the SocketAddress type in QEMU command. After I change 'str' to 'SocketAddress' and specify the command like "-object tdx-guest,type=vsock,cid=2,port=1234...", it will report "invalid parameter cid".
Quoted 6 lines
The -object parameter supports JSON syntax for this reason
-object '{"qom-type":"tdx-guest","quote-generation-service":{"type": "vsock", "cid":"2","port":"1234"}}'libvirt will always use the JSON syntax for -object with a new enough QEMU.
The JSON syntax works for me. Then, we need to add some doc about using JSON syntax when quote-generation-service is required.
Quoted 3 lines
With regards,
Daniel
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
Quoted 84 lines · Chenyi Qiang
Quoted 64 lines · Daniel P. Berrangé
Quoted 53 lines · Chenyi Qiang
Quoted 50 lines · Daniel P. Berrangé
Quoted 48 lines · Markus Armbruster
Quoted 34 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +1
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS)qapi/qom.json line 891
Quoted 10 lines
qapi/qom.json +3
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', 'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }Why not type SocketAddress?
Quoted 5 lines
Quoted 3 lines
Yes, the code uses SocketAddress internally when it eventually calls
qio_channel_socket_connect_async(), so we should directly use SocketAddress in the QAPI from the start.Any benefit to directly use SocketAddress?
We don't want whatever code consumes the configuration to do a second level of parsing to convert the 'str' value into the 'SocketAddress' object it actually needs.
Quoted 5 lines
QEMU has a long history of having a second round of ad-hoc parsing of configuration and we've found it to be a serious maintenence burden. Thus we strive to have everything represented in QAPI using the desired final type, and avoid the second round of parsing.
Thanks for your explanation.
Quoted 9 lines
Quoted 6 lines
"quote-generation-service" here is optional, it seems not trivial to add and parse the SocketAddress type in QEMU command. After I change 'str' to 'SocketAddress' and specify the command like "-object tdx-guest,type=vsock,cid=2,port=1234...", it will report "invalid parameter cid".
The -object parameter supports JSON syntax for this reason
Quoted 7 lines
Quoted 4 lines
-object '{"qom-type":"tdx-guest","quote-generation-service":{"type": "vsock", "cid":"2","port":"1234"}}'libvirt will always use the JSON syntax for -object with a new enough QEMU.
The JSON syntax works for me. Then, we need to add some doc about using JSON syntax when quote-generation-service is required.
This limitation doesn't look reasonable to me.
@Daniel,
Is it acceptable by QEMU community?
Quoted 3 lines
Quoted 3 lines
With regards,
Daniel
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
Quoted 90 lines · Xiaoyao Li
Quoted 89 lines · Chenyi Qiang
Quoted 64 lines · Daniel P. Berrangé
Quoted 53 lines · Chenyi Qiang
Quoted 50 lines · Daniel P. Berrangé
Quoted 48 lines · Markus Armbruster
Quoted 34 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +1
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS)qapi/qom.json line 891
Quoted 10 lines
qapi/qom.json +3
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', 'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }Why not type SocketAddress?
Quoted 5 lines
Quoted 3 lines
Yes, the code uses SocketAddress internally when it eventually calls
qio_channel_socket_connect_async(), so we should directly use SocketAddress in the QAPI from the start.Any benefit to directly use SocketAddress?
We don't want whatever code consumes the configuration to do a second level of parsing to convert the 'str' value into the 'SocketAddress' object it actually needs.
Quoted 5 lines
QEMU has a long history of having a second round of ad-hoc parsing of configuration and we've found it to be a serious maintenence burden. Thus we strive to have everything represented in QAPI using the desired final type, and avoid the second round of parsing.
Thanks for your explanation.
Quoted 13 lines
Quoted 6 lines
"quote-generation-service" here is optional, it seems not trivial to add and parse the SocketAddress type in QEMU command. After I change 'str' to 'SocketAddress' and specify the command like "-object tdx-guest,type=vsock,cid=2,port=1234...", it will report "invalid parameter cid".
The -object parameter supports JSON syntax for this reason
-object '{"qom-type":"tdx-guest","quote-generation-service":{"type": "vsock", "cid":"2","port":"1234"}}'libvirt will always use the JSON syntax for -object with a new enough QEMU.
Quoted 8 lines
Quoted 2 lines
The JSON syntax works for me. Then, we need to add some doc about using JSON syntax when quote-generation-service is required.
This limitation doesn't look reasonable to me.
@Daniel,
Is it acceptable by QEMU community?
This is the expected approach for object types which have non-scalar properties.
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
Quoted 94 lines · Daniel P. Berrangé
Quoted 93 lines · Xiaoyao Li
Quoted 91 lines · Chenyi Qiang
Quoted 64 lines · Daniel P. Berrangé
Quoted 53 lines · Chenyi Qiang
Quoted 50 lines · Daniel P. Berrangé
Quoted 48 lines · Markus Armbruster
Quoted 34 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +1
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS)qapi/qom.json line 891
Quoted 10 lines
qapi/qom.json +3
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', 'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }Why not type SocketAddress?
Quoted 5 lines
Quoted 3 lines
Yes, the code uses SocketAddress internally when it eventually calls
qio_channel_socket_connect_async(), so we should directly use SocketAddress in the QAPI from the start.Any benefit to directly use SocketAddress?
We don't want whatever code consumes the configuration to do a second level of parsing to convert the 'str' value into the 'SocketAddress' object it actually needs.
Quoted 5 lines
QEMU has a long history of having a second round of ad-hoc parsing of configuration and we've found it to be a serious maintenence burden. Thus we strive to have everything represented in QAPI using the desired final type, and avoid the second round of parsing.
Thanks for your explanation.
Quoted 7 lines
Quoted 6 lines
"quote-generation-service" here is optional, it seems not trivial to add and parse the SocketAddress type in QEMU command. After I change 'str' to 'SocketAddress' and specify the command like "-object tdx-guest,type=vsock,cid=2,port=1234...", it will report "invalid parameter cid".
Quoted 6 lines
The -object parameter supports JSON syntax for this reason
-object '{"qom-type":"tdx-guest","quote-generation-service":{"type": "vsock", "cid":"2","port":"1234"}}'libvirt will always use the JSON syntax for -object with a new enough QEMU.
The JSON syntax works for me. Then, we need to add some doc about using JSON syntax when quote-generation-service is required.
Quoted 8 lines
Quoted 5 lines
This limitation doesn't look reasonable to me.
@Daniel,
Is it acceptable by QEMU community?
This is the expected approach for object types which have non-scalar properties.
Learned it.
thanks!
Quoted 2 lines
With regards,
Daniel
Re: [PATCH v2 41/58] i386/tdx: handle TDG.VP.VMCALL<GetQuote>
In reply to Chenyi QiangI sent this reply to your question on the same day, but it got eaten by malfunctioning servers, and I noticed only now after another failure made me dig through my logs. Sorry for the inconvenience!
Quoted 51 lines · Chenyi Qiang
Quoted 50 lines · Daniel P. Berrangé
Quoted 48 lines · Markus Armbruster
Quoted 34 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
For GetQuote, delegate a request to Quote Generation Service. Add property of address of quote generation server and On request, connect to the server, read request buffer from shared guest memory, send the request buffer to the server and store the response into shared guest memory and notify TD guest by interrupt.
"quote-generation-service" is a property to specify Quote Generation Service(QGS) in qemu socket address format. The examples of the supported format are "vsock:2:1234", "unix:/run/qgs", "localhost:1234".
command line example:
qemu-system-x86_64 \ -object 'tdx-guest,id=tdx0,quote-generation-service=localhost:1234' \ -machine confidential-guest-support=tdx0Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- qapi/qom.json | 5 +- target/i386/kvm/tdx.c | 380 ++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.h | 7 + 3 files changed, 391 insertions(+), 1 deletion(-)qapi/qom.json +1
-0@@ -879,13 +879,16 @@ # # @mrownerconfig: MROWNERCONFIG SHA384 hex string of 48 * 2 length (default: 0) # +# @quote-generation-service: socket address for Quote Generation Service(QGS)qapi/qom.json line 891
Quoted 10 lines
qapi/qom.json +3
-1+# # Since: 8.2 ## { 'struct': 'TdxGuestProperties', 'data': { '*sept-ve-disable': 'bool', '*mrconfigid': 'str', '*mrowner': 'str', - '*mrownerconfig': 'str' } } + '*mrownerconfig': 'str', + '*quote-generation-service': 'str' } }Why not type SocketAddress?
Quoted 5 lines
Quoted 3 lines
Yes, the code uses SocketAddress internally when it eventually calls
qio_channel_socket_connect_async(), so we should directly use SocketAddress in the QAPI from the start.Any benefit to directly use SocketAddress?
Design principle: use JSON to encode structured data as text in QAPI/QMP.
Do: "mumble": [1, 2, 3]
Don't: "mumble": "1,2,3"
Do: "server": { "type": "inet", "host": "localhost", "port": "12345" }
Don't: "server": "host=localhost,port=12345"
We violate the principle in a couple of places. Some are arguably mistakes, some are pragmatic exceptions.
The principle implies "the only parser QAPI needs is the JSON parser".
The other benefit is consistency with existing interfaces. They use SocketAddress (a few old ones use SocketAddressLegacy).
Quoted 5 lines
"quote-generation-service" here is optional, it seems not trivial to add and parse the SocketAddress type in QEMU command. After I change 'str' to 'SocketAddress' and specify the command like "-object tdx-guest,type=vsock,cid=2,port=1234...", it will report "invalid parameter cid".
Try "quote-generation-service.port=1234".
PATCH v2 42/58 i386/tdx: register the fd read callback with the main loop to read the quote data
In reply to Xiaoyao LiFrom: Chenyi Qiang <[email protected]>
When TD guest invokes getquote tdvmcall, QEMU will register a async qio task with default context when the qio channel is connected. However, as there is a blocking action (recvmsg()) in qio_channel_read() and it will block main thread and make TD guest have no response until the server returns.
Set the io channel non-blocking and register the socket fd with the main loop. Move the read operation into the callback. When the fd is readable, inovke the callback to handle the quote data.
--- target/i386/kvm/tdx.c | 147 +++++++++++++++++++++++++++--------------- 1 file changed, 96 insertions(+), 51 deletions(-)
target/i386/kvm/tdx.c +96 -51
@@ -997,6 +997,8 @@ struct tdx_get_quote_task {uint32_t apic_id;hwaddr gpa;uint64_t buf_len; +char *out_data; +uint64_t out_len;struct tdx_get_quote_header hdr;int event_notify_interrupt;QIOChannelSocket *ioc;
@@ -1082,78 +1084,53 @@ static void tdx_td_notify(struct tdx_get_quote_task *t) } } -/* -* TODO: If QGS doesn't reply for long time, make it an error and interrupt -* guest. -*/ -static void tdx_handle_get_quote_connected (QIOTask *task,gpointer opaque) +static void tdx_get_quote_read (void *opaque) {struct tdx_get_quote_task *t = opaque; +ssize_t size =0 ;Error *err =NULL ; -char *in_data =NULL ; -char *out_data =NULL ; -size_t out_len; -ssize_t size;MachineState *ms;TdxGuest *tdx; - t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_ERROR ); -if (qio_task_propagate_error (task,NULL )) { - t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_QGS_UNAVAILABLE ); - goto error; - } - - in_data =g_malloc (le32_to_cpu (t->hdr.in_len)); -if (address_space_read (&address_space_memory, t->gpa +sizeof (t->hdr), -MEMTXATTRS_UNSPECIFIED , in_data, -le32_to_cpu (t->hdr.in_len)) !=MEMTX_OK ) { - goto error; - } - -if (qio_channel_write_all (QIO_CHANNEL (t->ioc), in_data, -le32_to_cpu (t->hdr.in_len), &err) || - err) { - t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_QGS_UNAVAILABLE ); - goto error; - } - - out_data =g_malloc (t->buf_len); - out_len =0 ; - size =0 ;while (true) {char *buf;size_t buf_size; -if (out_len < t->buf_len) { - buf = out_data + out_len; - buf_size = t->buf_len - out_len; +if (t->out_len < t->buf_len) { + buf = t->out_data + t->out_len; + buf_size = t->buf_len - t->out_len; }else {/* * The received data is too large to fit in the shared GPA. * Discard the received data and try to know the data size. */ - buf = out_data; + buf = t->out_data; buf_size = t->buf_len; } size =qio_channel_read (QIO_CHANNEL (t->ioc), buf, buf_size, &err); -if (err) { +if (!size) {break ; } -if (size <=0 ) { -break ; + +if (size <0 ) { +if (size ==QIO_CHANNEL_ERR_BLOCK ) { +return ; + }else { +break ; + } } - out_len += size; + t->out_len += size; }/* -* Treat partial read as success and let the QGS client to handle it because -* the client knows better about the QGS. +* If partial read successfully but return error at last, also treat it +* as failure. */ -if (out_len ==0 && (err || size <0 )) { +if (size <0 ) { t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_QGS_UNAVAILABLE ); goto error; } -if (out_len >0 && out_len > t->buf_len) { +if (t->out_len >0 && t->out_len > t->buf_len) {/* * There is no specific error code defined for this case(E2BIG) at the * moment.
@@ -1161,20 +1138,20 @@ static void tdx_handle_get_quote_connected(QIOTask *task, gpointer opaque)* update the error code. */ t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_ERROR ); - t->hdr.out_len =cpu_to_le32 (out_len); + t->hdr.out_len =cpu_to_le32 (t->out_len); goto error_hdr; }if (address_space_write ( &address_space_memory, t->gpa +sizeof (t->hdr), -MEMTXATTRS_UNSPECIFIED , out_data, out_len) !=MEMTX_OK ) { +MEMTXATTRS_UNSPECIFIED , t->out_data, t->out_len) !=MEMTX_OK ) { goto error; }/* * Even if out_len == 0, it's a success. It's up to the QGS-client contract * how to interpret the zero-sized message as return message. */ - t->hdr.out_len =cpu_to_le32 (out_len); + t->hdr.out_len =cpu_to_le32 (t->out_len); t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_SUCCESS ); error:
@@ -1185,14 +1162,15 @@ error_hdr:if (address_space_write ( &address_space_memory, t->gpa,MEMTXATTRS_UNSPECIFIED , &t->hdr,sizeof (t->hdr)) !=MEMTX_OK ) { -error_report ("TDX: failed to updsate GetQuote header.\n" ); +error_report ("TDX: failed to update GetQuote header." ); }tdx_td_notify (t); +qemu_set_fd_handler (t->ioc->fd,NULL ,NULL ,NULL );qio_channel_close (QIO_CHANNEL (t->ioc), &err);object_unref (OBJECT (t->ioc)); -g_free (in_data); -g_free (out_data); +g_free (t->out_data); +g_free (t);/* Maintain the number of in-flight requests. */ ms =MACHINE (qdev_get_machine ());
@@ -1200,7 +1178,71 @@ error_hdr:qemu_mutex_lock (&tdx->lock); tdx->quote_generation_num--;qemu_mutex_unlock (&tdx->lock); +} + +/* +* TODO: If QGS doesn't reply for long time, make it an error and interrupt +* guest. +*/ +static void tdx_handle_get_quote_connected (QIOTask *task,gpointer opaque) +{ +struct tdx_get_quote_task *t = opaque; +Error *err =NULL ; +char *in_data =NULL ; +MachineState *ms; +TdxGuest *tdx; + + t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_ERROR ); +if (qio_task_propagate_error (task,NULL )) { + t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_QGS_UNAVAILABLE ); + goto error; + } + + in_data =g_malloc (le32_to_cpu (t->hdr.in_len)); +if (!in_data) { + goto error; + } + +if (address_space_read (&address_space_memory, t->gpa +sizeof (t->hdr), +MEMTXATTRS_UNSPECIFIED , in_data, +le32_to_cpu (t->hdr.in_len)) !=MEMTX_OK ) { + goto error; + } + +qio_channel_set_blocking (QIO_CHANNEL (t->ioc), false,NULL ); + +if (qio_channel_write_all (QIO_CHANNEL (t->ioc), in_data, +le32_to_cpu (t->hdr.in_len), &err) || + err) { + t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_QGS_UNAVAILABLE ); + goto error; + } + +g_free (in_data); +qemu_set_fd_handler (t->ioc->fd, tdx_get_quote_read,NULL , t); + +return ; +error: + t->hdr.out_len =cpu_to_le32 (0 ); +if (address_space_write ( + &address_space_memory, t->gpa, +MEMTXATTRS_UNSPECIFIED , &t->hdr,sizeof (t->hdr)) !=MEMTX_OK ) { +error_report ("TDX: failed to update GetQuote header.\n" ); + } +tdx_td_notify (t); + +qio_channel_close (QIO_CHANNEL (t->ioc), &err); +object_unref (OBJECT (t->ioc)); +g_free (t); +g_free (in_data); + +/* Maintain the number of in-flight requests. */ + ms =MACHINE (qdev_get_machine ()); + tdx =TDX_GUEST (ms->cgs); +qemu_mutex_lock (&tdx->lock); + tdx->quote_generation_num--; +qemu_mutex_unlock (&tdx->lock);return ; }
@@ -1269,6 +1311,8 @@ static void tdx_handle_get_quote(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall) t->apic_id = tdx->event_notify_apic_id; t->gpa = gpa; t->buf_len = buf_len; + t->out_data =g_malloc (t->buf_len); + t->out_len =0 ; t->hdr = hdr; t->ioc = ioc;
@@ -1279,13 +1323,14 @@ static void tdx_handle_get_quote(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall)qemu_mutex_unlock (&tdx->lock); vmcall->status_code =TDG_VP_VMCALL_RETRY ;object_unref (OBJECT (ioc)); +g_free (t->out_data);g_free (t);return ; } tdx->quote_generation_num++; t->event_notify_interrupt = tdx->event_notify_interrupt;qio_channel_socket_connect_async ( - ioc, tdx->quote_generation, tdx_handle_get_quote_connected, t, g_free, + ioc, tdx->quote_generation, tdx_handle_get_quote_connected, t,NULL ,NULL );qemu_mutex_unlock (&tdx->lock);
Signature · 1 line
2.34.1
Re: [PATCH v2 42/58] i386/tdx: register the fd read callback with the main loop to read the quote data
Quoted 12 lines · Xiaoyao Li
From: Chenyi Qiang <[email protected]>
When TD guest invokes getquote tdvmcall, QEMU will register a async qio task with default context when the qio channel is connected. However, as there is a blocking action (
recvmsg()) inqio_channel_read()and it will block main thread and make TD guest have no response until the server returns.Set the io channel non-blocking and register the socket fd with the main loop. Move the read operation into the callback. When the fd is readable, inovke the callback to handle the quote data.
Quoted 6 lines
Signed-off-by Chenyi Qiang <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- target/i386/kvm/tdx.c | 147 +++++++++++++++++++++++++++--------------- 1 file changed, 96 insertions(+), 51 deletions(-)
How about squashing this patch with the previous one? I think this patch is somewhat a bug fix for it to resolve the thread blocking issue.
PATCH v2 43/58 i386/tdx: setup a timer for the qio channel
In reply to Xiaoyao LiFrom: Chenyi Qiang <[email protected]>
To avoid no response from QGS server, setup a timer for the transaction. If timeout, make it an error and interrupt guest. Define the threshold of time to 30s at present, maybe change to other value if not appropriate.
Extract the common cleanup code to make it more clear.
--- target/i386/kvm/tdx.c | 151 ++++++++++++++++++++++++------------------ 1 file changed, 85 insertions(+), 66 deletions(-)
target/i386/kvm/tdx.c +85 -66
@@ -1002,6 +1002,7 @@ struct tdx_get_quote_task {struct tdx_get_quote_header hdr;int event_notify_interrupt;QIOChannelSocket *ioc; +QEMUTimer timer; };struct x86_msi {
@@ -1084,13 +1085,48 @@ static void tdx_td_notify(struct tdx_get_quote_task *t) } } +static void tdx_getquote_task_cleanup (struct tdx_get_quote_task *t,bool outlen_overflow) +{ +MachineState *ms; +TdxGuest *tdx; + +if (t->hdr.error_code !=cpu_to_le64 (TDX_VP_GET_QUOTE_SUCCESS ) && !outlen_overflow) { + t->hdr.out_len =cpu_to_le32 (0 ); + } + +/* Publish the response contents before marking this request completed. */ +smp_wmb (); +if (address_space_write ( + &address_space_memory, t->gpa, +MEMTXATTRS_UNSPECIFIED , &t->hdr,sizeof (t->hdr)) !=MEMTX_OK ) { +error_report ("TDX: failed to update GetQuote header." ); + } +tdx_td_notify (t); + +if (t->ioc->fd >0 ) { +qemu_set_fd_handler (t->ioc->fd,NULL ,NULL ,NULL ); + } +qio_channel_close (QIO_CHANNEL (t->ioc),NULL ); +object_unref (OBJECT (t->ioc)); +timer_del (&t->timer); +g_free (t->out_data); +g_free (t); + +/* Maintain the number of in-flight requests. */ + ms =MACHINE (qdev_get_machine ()); + tdx =TDX_GUEST (ms->cgs); +qemu_mutex_lock (&tdx->lock); + tdx->quote_generation_num--; +qemu_mutex_unlock (&tdx->lock); +} + +static void tdx_get_quote_read (void *opaque) {struct tdx_get_quote_task *t = opaque;ssize_t size =0 ;Error *err =NULL ; -MachineState *ms; -TdxGuest *tdx; +bool outlen_overflow = false;while (true) {char *buf;
@@ -1135,11 +1171,12 @@ static void tdx_get_quote_read(void *opaque)* There is no specific error code defined for this case(E2BIG) at the * moment. * TODO: Once an error code for this case is defined in GHCI spec , -* update the error code. +* update the error code and the tdx_getquote_task_cleanup() argument. */ t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_ERROR ); t->hdr.out_len =cpu_to_le32 (t->out_len); - goto error_hdr; + outlen_overflow = true; + goto error; } if (address_space_write(
@@ -1155,94 +1192,76 @@ static void tdx_get_quote_read(void *opaque) t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_SUCCESS ); error: -if (t->hdr.error_code !=cpu_to_le64 (TDX_VP_GET_QUOTE_SUCCESS )) { - t->hdr.out_len =cpu_to_le32 (0 ); - } -error_hdr: -if (address_space_write ( - &address_space_memory, t->gpa, -MEMTXATTRS_UNSPECIFIED , &t->hdr,sizeof (t->hdr)) !=MEMTX_OK ) { -error_report ("TDX: failed to update GetQuote header." ); - } -tdx_td_notify (t); +tdx_getquote_task_cleanup (t, outlen_overflow); +} + +#define TRANSACTION_TIMEOUT 30000 + +static void getquote_timer_expired (void *opaque) +{ +struct tdx_get_quote_task *t = opaque; + +tdx_getquote_task_cleanup (t, false); +} -qemu_set_fd_handler (t->ioc->fd,NULL ,NULL ,NULL ); -qio_channel_close (QIO_CHANNEL (t->ioc), &err); -object_unref (OBJECT (t->ioc)); -g_free (t->out_data); -g_free (t); +static void tdx_transaction_start (struct tdx_get_quote_task *t) +{ +int64_t time; -/* Maintain the number of in-flight requests. */ - ms =MACHINE (qdev_get_machine ()); - tdx =TDX_GUEST (ms->cgs); -qemu_mutex_lock (&tdx->lock); - tdx->quote_generation_num--; -qemu_mutex_unlock (&tdx->lock); + time =qemu_clock_get_ms (QEMU_CLOCK_VIRTUAL ); +/* +* Timeout callback and fd callback both run in main loop thread, +* thus no need to worry about race condition. +*/ +qemu_set_fd_handler (t->ioc->fd, tdx_get_quote_read,NULL , t); +timer_init_ms (&t->timer,QEMU_CLOCK_VIRTUAL , getquote_timer_expired, t); +timer_mod (&t->timer, time +TRANSACTION_TIMEOUT ); } -/* -* TODO: If QGS doesn't reply for long time, make it an error and interrupt -* guest. -*/ static void tdx_handle_get_quote_connected (QIOTask *task,gpointer opaque) {struct tdx_get_quote_task *t = opaque;Error *err =NULL ;char *in_data =NULL ; -MachineState *ms; -TdxGuest *tdx; +int ret =0 ; t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_ERROR ); -if (qio_task_propagate_error (task,NULL )) { + ret =qio_task_propagate_error (task,NULL ); +if (ret) { t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_QGS_UNAVAILABLE ); - goto error; + goto out; } in_data =g_malloc (le32_to_cpu (t->hdr.in_len));if (!in_data) { - goto error; + ret =-1 ; + goto out; } -if (address_space_read (&address_space_memory, t->gpa +sizeof (t->hdr), -MEMTXATTRS_UNSPECIFIED , in_data, -le32_to_cpu (t->hdr.in_len)) !=MEMTX_OK ) { - goto error; + ret =address_space_read (&address_space_memory, t->gpa +sizeof (t->hdr), +MEMTXATTRS_UNSPECIFIED , in_data, +le32_to_cpu (t->hdr.in_len)); +if (ret) { +g_free (in_data); + goto out; }qio_channel_set_blocking (QIO_CHANNEL (t->ioc), false,NULL ); -if (qio_channel_write_all (QIO_CHANNEL (t->ioc), in_data, -le32_to_cpu (t->hdr.in_len), &err) || - err) { + ret =qio_channel_write_all (QIO_CHANNEL (t->ioc), in_data, +le32_to_cpu (t->hdr.in_len), &err); +if (ret) { t->hdr.error_code =cpu_to_le64 (TDX_VP_GET_QUOTE_QGS_UNAVAILABLE ); - goto error; +g_free (in_data); + goto out; } -g_free (in_data); -qemu_set_fd_handler (t->ioc->fd, tdx_get_quote_read,NULL , t); - -return ; -error: - t->hdr.out_len =cpu_to_le32 (0 ); - -if (address_space_write ( - &address_space_memory, t->gpa, -MEMTXATTRS_UNSPECIFIED , &t->hdr,sizeof (t->hdr)) !=MEMTX_OK ) { -error_report ("TDX: failed to update GetQuote header.\n" ); +out: +if (ret) { +tdx_getquote_task_cleanup (t, false); + }else { +tdx_transaction_start (t); } -tdx_td_notify (t); - -qio_channel_close (QIO_CHANNEL (t->ioc), &err); -object_unref (OBJECT (t->ioc)); -g_free (t); -g_free (in_data); - -/* Maintain the number of in-flight requests. */ - ms =MACHINE (qdev_get_machine ()); - tdx =TDX_GUEST (ms->cgs); -qemu_mutex_lock (&tdx->lock); - tdx->quote_generation_num--; -qemu_mutex_unlock (&tdx->lock);return ; }
Signature · 1 line
2.34.1
Re: [PATCH v2 43/58] i386/tdx: setup a timer for the qio channel
Quoted 44 lines · Xiaoyao Li
From: Chenyi Qiang <[email protected]>
To avoid no response from QGS server, setup a timer for the transaction. If timeout, make it an error and interrupt guest. Define the threshold of time to 30s at present, maybe change to other value if not appropriate.
Extract the common cleanup code to make it more clear.
Signed-off-by Chenyi Qiang <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- target/i386/kvm/tdx.c | 151 ++++++++++++++++++++++++------------------ 1 file changed, 85 insertions(+), 66 deletions(-)target/i386/kvm/tdx.c +15
-0@@ -1002,6 +1002,7 @@ struct tdx_get_quote_task {struct tdx_get_quote_header hdr;int event_notify_interrupt;QIOChannelSocket *ioc; +QEMUTimer timer; };struct x86_msi {@@ -1084,13 +1085,48 @@ static void tdx_td_notify(struct tdx_get_quote_task *t) } } +static void tdx_getquote_task_cleanup (struct tdx_get_quote_task *t,bool outlen_overflow) +{ +MachineState *ms; +TdxGuest *tdx; + +if (t->hdr.error_code !=cpu_to_le64 (TDX_VP_GET_QUOTE_SUCCESS ) && !outlen_overflow) { + t->hdr.out_len =cpu_to_le32 (0 ); + } + +/* Publish the response contents before marking this request completed. */ +smp_wmb (); +if (address_space_write ( + &address_space_memory, t->gpa, +MEMTXATTRS_UNSPECIFIED , &t->hdr,sizeof (t->hdr)) !=MEMTX_OK ) {
target/i386/kvm/tdx.c static void tdx_td_notify(struct tdx_get_quote_task *t) · line 1111
Quoted 10 lines
target/i386/kvm/tdx.c +10
-0+error_report ("TDX: failed to update GetQuote header." ); + } +tdx_td_notify (t); + +if (t->ioc->fd >0 ) { +qemu_set_fd_handler (t->ioc->fd,NULL ,NULL ,NULL ); + } +qio_channel_close (QIO_CHANNEL (t->ioc),NULL ); +object_unref (OBJECT (t->ioc)); +timer_del (&t->timer);
Xiaoyao, I guess you missed a bug fix patch here as t->timer could be uninitialized and then timer_del() will cause segv.
Quoted 11 lines
target/i386/kvm/tdx.c +11
-0+g_free (t->out_data); +g_free (t); + +/* Maintain the number of in-flight requests. */ + ms =MACHINE (qdev_get_machine ()); + tdx =TDX_GUEST (ms->cgs); +qemu_mutex_lock (&tdx->lock); + tdx->quote_generation_num--; +qemu_mutex_unlock (&tdx->lock); +} +
Re: [PATCH v2 43/58] i386/tdx: setup a timer for the qio channel
Quoted 52 lines · Chenyi Qiang
Quoted 49 lines · Xiaoyao Li
From: Chenyi Qiang <[email protected]>
To avoid no response from QGS server, setup a timer for the transaction. If timeout, make it an error and interrupt guest. Define the threshold of time to 30s at present, maybe change to other value if not appropriate.
Extract the common cleanup code to make it more clear.
Signed-off-by Chenyi Qiang <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- target/i386/kvm/tdx.c | 151 ++++++++++++++++++++++++------------------ 1 file changed, 85 insertions(+), 66 deletions(-)target/i386/kvm/tdx.c +20
-0@@ -1002,6 +1002,7 @@ struct tdx_get_quote_task {struct tdx_get_quote_header hdr;int event_notify_interrupt;QIOChannelSocket *ioc; +QEMUTimer timer; };struct x86_msi {@@ -1084,13 +1085,48 @@ static void tdx_td_notify(struct tdx_get_quote_task *t) } } +static void tdx_getquote_task_cleanup (struct tdx_get_quote_task *t,bool outlen_overflow) +{ +MachineState *ms; +TdxGuest *tdx; + +if (t->hdr.error_code !=cpu_to_le64 (TDX_VP_GET_QUOTE_SUCCESS ) && !outlen_overflow) { + t->hdr.out_len =cpu_to_le32 (0 ); + } + +/* Publish the response contents before marking this request completed. */ +smp_wmb (); +if (address_space_write ( + &address_space_memory, t->gpa, +MEMTXATTRS_UNSPECIFIED , &t->hdr,sizeof (t->hdr)) !=MEMTX_OK ) { +error_report ("TDX: failed to update GetQuote header." ); + } +tdx_td_notify (t); + +if (t->ioc->fd >0 ) {
Quoted 8 lines
target/i386/kvm/tdx.c static void tdx_td_notify(struct tdx_get_quote_task *t) · line 1111
Quoted 5 lines
target/i386/kvm/tdx.c +5
-0+qemu_set_fd_handler (t->ioc->fd,NULL ,NULL ,NULL ); + } +qio_channel_close (QIO_CHANNEL (t->ioc),NULL ); +object_unref (OBJECT (t->ioc)); +timer_del (&t->timer);Xiaoyao, I guess you missed a bug fix patch here as
t->timercould be uninitialized and thentimer_del()will cause segv.
Thanks for the reminding.
I'll update this patch to include the fix.
Thanks,
-Xiaoyao
Quoted 12 lines
Quoted 11 lines
target/i386/kvm/tdx.c +11
-0+g_free (t->out_data); +g_free (t); + +/* Maintain the number of in-flight requests. */ + ms =MACHINE (qdev_get_machine ()); + tdx =TDX_GUEST (ms->cgs); +qemu_mutex_lock (&tdx->lock); + tdx->quote_generation_num--; +qemu_mutex_unlock (&tdx->lock); +} +
PATCH v2 44/58 i386/tdx: handle TDG.VP.VMCALL<MapGPA> hypercall
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
MapGPA is a hypercall to convert GPA from/to private GPA to/from shared GPA. As the conversion function is already implemented as kvm_convert_memory, wire it to TDX hypercall exit.
--- accel/kvm/kvm-all.c | 2 +- include/sysemu/kvm.h | 2 ++ target/i386/kvm/tdx.c | 37 +++++++++++++++++++++++++++++++++++++ 3 files changed, 40 insertions(+), 1 deletion(-)
accel/kvm/kvm-all.c +1 -1
@@ -3059,7 +3059,7 @@ static void kvm_eat_signals(CPUState *cpu) }while (sigismember (&chkset,SIG_IPI )); } -static int kvm_convert_memory (hwaddr start,hwaddr size,bool to_private) +int kvm_convert_memory (hwaddr start,hwaddr size,bool to_private) {MemoryRegionSection section;void *addr;
include/sysemu/kvm.h +2 -0
@@ -584,4 +584,6 @@ uint32_t kvm_dirty_ring_size(void);int kvm_set_memory_attributes_private (hwaddr start,hwaddr size);int kvm_set_memory_attributes_shared (hwaddr start,hwaddr size); + +int kvm_convert_memory (hwaddr start,hwaddr size,bool to_private);#endif
target/i386/kvm/tdx.c +37 -0
@@ -934,6 +934,7 @@ static void tdx_guest_class_init(ObjectClass *oc, void *data) { } +#define TDG_VP_VMCALL_MAP_GPA 0x10001ULL#define TDG_VP_VMCALL_GET_QUOTE 0x10002ULL#define TDG_VP_VMCALL_SETUP_EVENT_NOTIFY_INTERRUPT 0x10004ULL
@@ -993,6 +994,39 @@ static hwaddr tdx_shared_bit(X86CPU *cpu)return (cpu->phys_bits >48 ) ?BIT_ULL (51 ) :BIT_ULL (47 ); } +static void tdx_handle_map_gpa (X86CPU *cpu,struct kvm_tdx_vmcall *vmcall) +{ +hwaddr shared_bit =tdx_shared_bit (cpu); +hwaddr gpa = vmcall->in_r12 & ~shared_bit; +bool private = !(vmcall->in_r12 & shared_bit); +hwaddr size = vmcall->in_r13; +int ret =0 ; + + vmcall->status_code =TDG_VP_VMCALL_INVALID_OPERAND ; + +if (!QEMU_IS_ALIGNED (gpa,4096 ) || !QEMU_IS_ALIGNED (size,4096 )) { + vmcall->status_code =TDG_VP_VMCALL_ALIGN_ERROR ; +return ; + } + +/* Overflow case. */ +if (gpa + size < gpa) { +return ; + } +if (gpa >= (1ULL << cpu->phys_bits) || + gpa + size >= (1ULL << cpu->phys_bits)) { +return ; + } + +if (size >0 ) { + ret =kvm_convert_memory (gpa, size, private); + } + +if (!ret) { + vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; + } +} +struct tdx_get_quote_task {uint32_t apic_id;hwaddr gpa;
@@ -1384,6 +1418,9 @@ static void tdx_handle_vmcall(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall) }switch (vmcall->subfunction) { +case TDG_VP_VMCALL_MAP_GPA : +tdx_handle_map_gpa (cpu, vmcall); +break ;case TDG_VP_VMCALL_GET_QUOTE :tdx_handle_get_quote (cpu, vmcall);break ;
Signature · 1 line
2.34.1
PATCH v2 45/58 i386/tdx: Limit the range size for MapGPA
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
If the range for TDG.VP.VMCALL<MapGPA> is too large, process the limited size and return retry error. It's bad for VMM to take too long time, e.g. second order, with blocking vcpu execution. It results in too many missing timer interrupts.
--- target/i386/kvm/tdx.c | 19 ++++++++++++++++++- 1 file changed, 18 insertions(+), 1 deletion(-)
target/i386/kvm/tdx.c +18 -1
@@ -994,12 +994,16 @@ static hwaddr tdx_shared_bit(X86CPU *cpu)return (cpu->phys_bits >48 ) ?BIT_ULL (51 ) :BIT_ULL (47 ); } +/* 64MB at most in one call. What value is appropriate? */ +#define TDX_MAP_GPA_MAX_LEN (64 * 1024 * 1024) +static void tdx_handle_map_gpa (X86CPU *cpu,struct kvm_tdx_vmcall *vmcall) {hwaddr shared_bit =tdx_shared_bit (cpu);hwaddr gpa = vmcall->in_r12 & ~shared_bit;bool private = !(vmcall->in_r12 & shared_bit);hwaddr size = vmcall->in_r13; +bool retry = false;int ret =0 ; vmcall->status_code =TDG_VP_VMCALL_INVALID_OPERAND ;
@@ -1018,12 +1022,25 @@ static void tdx_handle_map_gpa(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall)return ; } +if (size >TDX_MAP_GPA_MAX_LEN ) { + retry = true; + size =TDX_MAP_GPA_MAX_LEN ; + } +if (size >0 ) { ret =kvm_convert_memory (gpa, size, private); }if (!ret) { - vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; +if (retry) { + vmcall->status_code =TDG_VP_VMCALL_RETRY ; + vmcall->out_r11 = gpa + size; +if (!private) { + vmcall->out_r11 |= shared_bit; + } + }else { + vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; + } } }
Signature · 1 line
2.34.1
Re: [PATCH v2 45/58] i386/tdx: Limit the range size for MapGPA
Quoted 6 lines · Xiaoyao Li
From: Isaku Yamahata <[email protected]>
If the range for TDG.VP.VMCALL<MapGPA> is too large, process the limited size and return retry error. It's bad for VMM to take too long time, e.g. second order, with blocking vcpu execution. It results in too many missing timer interrupts.
This patch requires the guest side patch. [1]
Unless with large guest memory, it's unlikely to hit the limit with KVM/qemu, though.
[1] https://lore.kernel.org/all/[email protected]/
Quoted 59 lines
Signed-off-by Isaku Yamahata <[email protected]>Signed-off-by Xiaoyao Li <[email protected]>--- target/i386/kvm/tdx.c | 19 ++++++++++++++++++- 1 file changed, 18 insertions(+), 1 deletion(-)target/i386/kvm/tdx.c +18
-1@@ -994,12 +994,16 @@ static hwaddr tdx_shared_bit(X86CPU *cpu)return (cpu->phys_bits >48 ) ?BIT_ULL (51 ) :BIT_ULL (47 ); } +/* 64MB at most in one call. What value is appropriate? */ +#define TDX_MAP_GPA_MAX_LEN (64 * 1024 * 1024) +static void tdx_handle_map_gpa (X86CPU *cpu,struct kvm_tdx_vmcall *vmcall) {hwaddr shared_bit =tdx_shared_bit (cpu);hwaddr gpa = vmcall->in_r12 & ~shared_bit;bool private = !(vmcall->in_r12 & shared_bit);hwaddr size = vmcall->in_r13; +bool retry = false;int ret =0 ; vmcall->status_code =TDG_VP_VMCALL_INVALID_OPERAND ;@@ -1018,12 +1022,25 @@ static void tdx_handle_map_gpa(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall)return ; } +if (size >TDX_MAP_GPA_MAX_LEN ) { + retry = true; + size =TDX_MAP_GPA_MAX_LEN ; + } +if (size >0 ) { ret =kvm_convert_memory (gpa, size, private); }if (!ret) { - vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; +if (retry) { + vmcall->status_code =TDG_VP_VMCALL_RETRY ; + vmcall->out_r11 = gpa + size; +if (!private) { + vmcall->out_r11 |= shared_bit; + } + }else { + vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; + } } }Signature · 1 line
2.34.1
Signature · 1 line
Isaku Yamahata <[email protected]>
PATCH v2 46/58 i386/tdx: Handle TDG.VP.VMCALL<REPORT_FATAL_ERROR>
In reply to Xiaoyao Li--- target/i386/kvm/tdx.c | 40 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 40 insertions(+)
target/i386/kvm/tdx.c +40 -0
@@ -936,6 +936,7 @@ static void tdx_guest_class_init(ObjectClass *oc, void *data)#define TDG_VP_VMCALL_MAP_GPA 0x10001ULL#define TDG_VP_VMCALL_GET_QUOTE 0x10002ULL +#define TDG_VP_VMCALL_REPORT_FATAL_ERROR 0x10003ULL#define TDG_VP_VMCALL_SETUP_EVENT_NOTIFY_INTERRUPT 0x10004ULL#define TDG_VP_VMCALL_SUCCESS 0x0000000000000000ULL
@@ -1407,6 +1408,42 @@ static void tdx_handle_get_quote(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall) vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; } +static void tdx_handle_report_fatal_error (X86CPU *cpu, +struct kvm_tdx_vmcall *vmcall) +{ +uint64_t error_code = vmcall->in_r12; +char *message =NULL ; + +if (error_code &0xffff ) { +error_report ("invalid error code of TDG.VP.VMCALL<REPORT_FATAL_ERROR>\n" ); +exit (1 ); + } + +/* it has optional message */ +if (vmcall->in_r14) { +uint64_t * tmp; + +#define GUEST_PANIC_INFO_TDX_MESSAGE_MAX 64 + message =g_malloc0 (GUEST_PANIC_INFO_TDX_MESSAGE_MAX +1 ); + + tmp = (uint64_t *)message; +/* The order is defined in TDX GHCI spec */ + *(tmp++) =cpu_to_le64 (vmcall->in_r14); + *(tmp++) =cpu_to_le64 (vmcall->in_r15); + *(tmp++) =cpu_to_le64 (vmcall->in_rbx); + *(tmp++) =cpu_to_le64 (vmcall->in_rdi); + *(tmp++) =cpu_to_le64 (vmcall->in_rsi); + *(tmp++) =cpu_to_le64 (vmcall->in_r8); + *(tmp++) =cpu_to_le64 (vmcall->in_r9); + *(tmp++) =cpu_to_le64 (vmcall->in_rdx); + message[GUEST_PANIC_INFO_TDX_MESSAGE_MAX ] ='\0' ; +assert ((char *)tmp == message +GUEST_PANIC_INFO_TDX_MESSAGE_MAX ); + } + +error_report ("TD guest reports fatal error. %s\n" , message ? :"" ); +exit (1 ); +} +static void tdx_handle_setup_event_notify_interrupt (X86CPU *cpu,struct kvm_tdx_vmcall *vmcall) {
@@ -1441,6 +1478,9 @@ static void tdx_handle_vmcall(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall)case TDG_VP_VMCALL_GET_QUOTE :tdx_handle_get_quote (cpu, vmcall);break ; +case TDG_VP_VMCALL_REPORT_FATAL_ERROR : +tdx_handle_report_fatal_error (cpu, vmcall); +break ;case TDG_VP_VMCALL_SETUP_EVENT_NOTIFY_INTERRUPT :tdx_handle_setup_event_notify_interrupt (cpu, vmcall);break ;
Signature · 1 line
2.34.1
PATCH v2 47/58 i386/tdx: Wire REPORT_FATAL_ERROR with GuestPanic facility
In reply to Xiaoyao LiOriginated-from: Isaku Yamahata <[email protected]>
--- qapi/run-state.json | 17 +++++++++++++-- softmmu/runstate.c | 49 +++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.c | 24 ++++++++++++++++++++- 3 files changed, 87 insertions(+), 3 deletions(-)
qapi/run-state.json +15 -2
@@ -499,7 +499,7 @@ # Since: 2.9 ## { 'enum': 'GuestPanicInformationType', - 'data': [ 'hyper-v', 's390' ] } + 'data': [ 'hyper-v', 's390', 'tdx' ] } ## # @GuestPanicInformation:
@@ -514,7 +514,8 @@ 'base': {'type': 'GuestPanicInformationType'}, 'discriminator': 'type', 'data': {'hyper-v': 'GuestPanicInformationHyperV', - 's390': 'GuestPanicInformationS390'}} + 's390': 'GuestPanicInformationS390', + 'tdx' : 'GuestPanicInformationTdx'}} ## # @GuestPanicInformationHyperV:
@@ -577,6 +578,18 @@ 'psw-addr': 'uint64', 'reason': 'S390CrashReason'}} +## +# @GuestPanicInformationTdx: +# +# TDX GHCI TDG.VP.VMCALL<ReportFatalError> specific guest panic information +# +# Since: 8.2 +## +{'struct': 'GuestPanicInformationTdx', + 'data': {'error-code': 'uint64', + 'gpa': 'uint64', + 'message': 'str'}} + ## # @MEMORY_FAILURE: #
softmmu/runstate.c +49 -0
56 lines not on this page · raw message
target/i386/kvm/tdx.c +23 -1
@@ -18,6 +18,7 @@#include "qom/object_interfaces.h" #include "standard-headers/asm-x86/kvm_para.h" #include "sysemu/kvm.h" +#include "sysemu/runstate.h" #include "sysemu/sysemu.h" #include "exec/address-spaces.h" #include "exec/ramblock.h"
@@ -1408,11 +1409,26 @@ static void tdx_handle_get_quote(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall) vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; } +static void tdx_panicked_on_fatal_error (X86CPU *cpu,uint64_t error_code, +uint64_t gpa,char *message) +{ +GuestPanicInformation *panic_info; + + panic_info =g_new0 (GuestPanicInformation,1 ); + panic_info->type =GUEST_PANIC_INFORMATION_TYPE_TDX ; + panic_info->u.tdx.error_code = error_code; + panic_info->u.tdx.gpa = gpa; + panic_info->u.tdx.message = (char *)message; + +qemu_system_guest_panicked (panic_info); +} +static void tdx_handle_report_fatal_error (X86CPU *cpu,struct kvm_tdx_vmcall *vmcall) {uint64_t error_code = vmcall->in_r12;char *message =NULL ; +uint64_t gpa =-1ull ;if (error_code &0xffff ) {error_report ("invalid error code of TDG.VP.VMCALL<REPORT_FATAL_ERROR>\n" );
@@ -1441,7 +1457,13 @@ static void tdx_handle_report_fatal_error(X86CPU *cpu, }error_report ("TD guest reports fatal error. %s\n" , message ? :"" ); -exit (1 ); + +#define TDX_REPORT_FATAL_ERROR_GPA_VALID BIT_ULL(63) +if (error_code &TDX_REPORT_FATAL_ERROR_GPA_VALID ) { + gpa = vmcall->in_r13; + } + +tdx_panicked_on_fatal_error (cpu, error_code, gpa, message); } static void tdx_handle_setup_event_notify_interrupt(X86CPU *cpu,
Signature · 1 line
2.34.1
Re: [PATCH v2 47/58] i386/tdx: Wire REPORT_FATAL_ERROR with GuestPanic facility
Quoted 13 lines · Xiaoyao Li
Originated-from: Isaku Yamahata <[email protected]>
Signed-off-by Xiaoyao Li <[email protected]>--- qapi/run-state.json | 17 +++++++++++++-- softmmu/runstate.c | 49 +++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.c | 24 ++++++++++++++++++++- 3 files changed, 87 insertions(+), 3 deletions(-)qapi/run-state.json +0
-0
qapi/run-state.json line 502
Quoted 5 lines
qapi/run-state.json +1
-1@@ -499,7 +499,7 @@ # Since: 2.9 ## { 'enum': 'GuestPanicInformationType', - 'data': [ 'hyper-v', 's390' ] } + 'data': [ 'hyper-v', 's390', 'tdx' ] }
Missing documentation for the 'tdx' value
Quoted 14 lines
qapi/run-state.json +2
-1## # @GuestPanicInformation:@@ -514,7 +514,8 @@ 'base': {'type': 'GuestPanicInformationType'}, 'discriminator': 'type', 'data': {'hyper-v': 'GuestPanicInformationHyperV', - 's390': 'GuestPanicInformationS390'}} + 's390': 'GuestPanicInformationS390', + 'tdx' : 'GuestPanicInformationTdx'}} ## # @GuestPanicInformationHyperV:
qapi/run-state.json line 584
Quoted 7 lines
qapi/run-state.json +4
-0@@ -577,6 +578,18 @@ 'psw-addr': 'uint64', 'reason': 'S390CrashReason'}} +## +# @GuestPanicInformationTdx: +# +# TDX GHCI TDG.VP.VMCALL<ReportFatalError> specific guest panic information
Not documented any of the struct members. Especially please include the warning that 'message' comes from the guest and so must not be trusted, not assumed to be well formed.
Quoted 46 lines
qapi/run-state.json +8
-0+# +# Since: 8.2 +## +{'struct': 'GuestPanicInformationTdx', + 'data': {'error-code': 'uint64', + 'gpa': 'uint64', + 'message': 'str'}} + ## # @MEMORY_FAILURE: #softmmu/runstate.c +27
-0@@ -518,7 +518,56 @@ void qemu_system_guest_panicked(GuestPanicInformation *info)S390CrashReason_str (info->u.s390.reason), info->u.s390.psw_mask, info->u.s390.psw_addr); + }else if (info->type ==GUEST_PANIC_INFORMATION_TYPE_TDX ) { +char *buf =NULL ; +bool printable = false; + +/* +* Although message is defined as a json string, we shouldn't +* unconditionally treat it as is because the guest generated it and +* it's not necessarily trustable. +*/ +if (info->u.tdx.message) { +/* The caller guarantees the NUL-terminated string. */ +int len =strlen (info->u.tdx.message); +int i; + + printable = len >0 ; +for (i =0 ; i < len; i++) { +if (!(0x20 <= info->u.tdx.message[i] && + info->u.tdx.message[i] <=0x7e )) { + printable = false; +break ; + } + } + +/* 3 = length of "%02x " */ + buf =g_malloc (len *3 ); +for (i =0 ; i < len; i++) { +if (info->u.tdx.message[i] =='\0' ) {
softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 557
Quoted 10 lines
softmmu/runstate.c +10
-0+break ; + }else { +sprintf (buf +3 * i,"%02x " , info->u.tdx.message[i]); + } + } +if (i >0 ) +/* replace the last ' '(space) to NUL */ + buf[i *3 -1 ] ='\0' ; +else + buf[0 ] ='\0' ;
You're building this escaped buffer but...
softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 565
Quoted 8 lines
softmmu/runstate.c +8
-0+ } + + qemu_log_mask(LOG_GUEST_ERROR, + //" TDX report fatal error:\"%s\" %s", + " TDX report fatal error:\"%s\"" + "error: 0x%016" PRIx64 " gpa page: 0x%016" PRIx64 "\n", + printable ? info->u.tdx.message : "", +//buf ? buf : "",
...then not actually using it
Either delete the 'buf' code, or use it.
Quoted 48 lines
softmmu/runstate.c +4
-0+ info->u.tdx.error_code, + info->u.tdx.gpa); +g_free (buf); } +qapi_free_GuestPanicInformation (info); } }target/i386/kvm/tdx.c +16
-0@@ -18,6 +18,7 @@#include "qom/object_interfaces.h" #include "standard-headers/asm-x86/kvm_para.h" #include "sysemu/kvm.h" +#include "sysemu/runstate.h" #include "sysemu/sysemu.h" #include "exec/address-spaces.h" #include "exec/ramblock.h" @@ -1408,11 +1409,26 @@ static void tdx_handle_get_quote(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall) vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; } +static void tdx_panicked_on_fatal_error (X86CPU *cpu,uint64_t error_code, +uint64_t gpa,char *message) +{ +GuestPanicInformation *panic_info; + + panic_info =g_new0 (GuestPanicInformation,1 ); + panic_info->type =GUEST_PANIC_INFORMATION_TYPE_TDX ; + panic_info->u.tdx.error_code = error_code; + panic_info->u.tdx.gpa = gpa; + panic_info->u.tdx.message = (char *)message; + +qemu_system_guest_panicked (panic_info); +} +static void tdx_handle_report_fatal_error (X86CPU *cpu,struct kvm_tdx_vmcall *vmcall) {uint64_t error_code = vmcall->in_r12;char *message =NULL ; +uint64_t gpa =-1ull ;if (error_code &0xffff ) {error_report ("invalid error code of TDG.VP.VMCALL<REPORT_FATAL_ERROR>\n" );
target/i386/kvm/tdx.c static void tdx_handle_report_fatal_error(X86CPU *cpu, · line 1459
Quoted 3 lines
target/i386/kvm/tdx.c +0
-0@@ -1441,7 +1457,13 @@ static void tdx_handle_report_fatal_error(X86CPU *cpu, }error_report ("TD guest reports fatal error. %s\n" , message ? :"" );
In tdx_panicked_on_fatal_error you're avoiding printing 'message' if it contains non-printable characters, but here you're printing it regardless.
Do we still need this error_report call at all ?
target/i386/kvm/tdx.c static void tdx_handle_report_fatal_error(X86CPU *cpu, · line 1463
Quoted 5 lines
target/i386/kvm/tdx.c +4
-1-exit (1 ); + +#define TDX_REPORT_FATAL_ERROR_GPA_VALID BIT_ULL(63) +if (error_code &TDX_REPORT_FATAL_ERROR_GPA_VALID ) { + gpa = vmcall->in_r13;
Bad indent
Quoted 9 lines
target/i386/kvm/tdx.c +3
-0+ } + +tdx_panicked_on_fatal_error (cpu, error_code, gpa, message); } static void tdx_handle_setup_event_notify_interrupt(X86CPU *cpu,Signature · 1 line
2.34.1
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 47/58] i386/tdx: Wire REPORT_FATAL_ERROR with GuestPanic facility
Quoted 39 lines · Daniel P. Berrangé
Quoted 13 lines · Xiaoyao Li
Originated-from: Isaku Yamahata <[email protected]>
Signed-off-by Xiaoyao Li <[email protected]>--- qapi/run-state.json | 17 +++++++++++++-- softmmu/runstate.c | 49 +++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.c | 24 ++++++++++++++++++++- 3 files changed, 87 insertions(+), 3 deletions(-)qapi/run-state.json +0
-0qapi/run-state.json line 502
Quoted 5 lines
qapi/run-state.json +1
-1@@ -499,7 +499,7 @@ # Since: 2.9 ## { 'enum': 'GuestPanicInformationType', - 'data': [ 'hyper-v', 's390' ] } + 'data': [ 'hyper-v', 's390', 'tdx' ] }Missing documentation for the 'tdx' value
Quoted 17 lines
qapi/run-state.json +2
-1## # @GuestPanicInformation:@@ -514,7 +514,8 @@ 'base': {'type': 'GuestPanicInformationType'}, 'discriminator': 'type', 'data': {'hyper-v': 'GuestPanicInformationHyperV', - 's390': 'GuestPanicInformationS390'}} + 's390': 'GuestPanicInformationS390', + 'tdx' : 'GuestPanicInformationTdx'}} ## # @GuestPanicInformationHyperV:@@ -577,6 +578,18 @@ 'psw-addr': 'uint64', 'reason': 'S390CrashReason'}}
Quoted 8 lines
qapi/run-state.json line 584
Quoted 4 lines
qapi/run-state.json +4
-0+## +# @GuestPanicInformationTdx: +# +# TDX GHCI TDG.VP.VMCALL<ReportFatalError> specific guest panic informationNot documented any of the struct members. Especially please include the warning that 'message' comes from the guest and so must not be trusted, not assumed to be well formed.
Will do it in next version.
thanks!
Quoted 68 lines
Quoted 46 lines
qapi/run-state.json +8
-0+# +# Since: 8.2 +## +{'struct': 'GuestPanicInformationTdx', + 'data': {'error-code': 'uint64', + 'gpa': 'uint64', + 'message': 'str'}} + ## # @MEMORY_FAILURE: #softmmu/runstate.c +27
-0@@ -518,7 +518,56 @@ void qemu_system_guest_panicked(GuestPanicInformation *info)S390CrashReason_str (info->u.s390.reason), info->u.s390.psw_mask, info->u.s390.psw_addr); + }else if (info->type ==GUEST_PANIC_INFORMATION_TYPE_TDX ) { +char *buf =NULL ; +bool printable = false; + +/* +* Although message is defined as a json string, we shouldn't +* unconditionally treat it as is because the guest generated it and +* it's not necessarily trustable. +*/ +if (info->u.tdx.message) { +/* The caller guarantees the NUL-terminated string. */ +int len =strlen (info->u.tdx.message); +int i; + + printable = len >0 ; +for (i =0 ; i < len; i++) { +if (!(0x20 <= info->u.tdx.message[i] && + info->u.tdx.message[i] <=0x7e )) { + printable = false; +break ; + } + } + +/* 3 = length of "%02x " */ + buf =g_malloc (len *3 ); +for (i =0 ; i < len; i++) { +if (info->u.tdx.message[i] =='\0' ) {softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 557
Quoted 10 lines
softmmu/runstate.c +10
-0+break ; + }else { +sprintf (buf +3 * i,"%02x " , info->u.tdx.message[i]); + } + } +if (i >0 ) +/* replace the last ' '(space) to NUL */ + buf[i *3 -1 ] ='\0' ; +else + buf[0 ] ='\0' ;You're building this escaped buffer but...
softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 565
Quoted 8 lines
softmmu/runstate.c +8
-0+ } + + qemu_log_mask(LOG_GUEST_ERROR, + //" TDX report fatal error:\"%s\" %s", + " TDX report fatal error:\"%s\"" + "error: 0x%016" PRIx64 " gpa page: 0x%016" PRIx64 "\n", + printable ? info->u.tdx.message : "", +//buf ? buf : "",
Quoted 3 lines
...then not actually using it
Either delete the 'buf' code, or use it.
Sorry for posting some internal testing version.
Does below look good to you?
@@ -518,7 +518,56 @@ void qemu_system_guest_panicked(GuestPanicInformation *info)
S390CrashReason_str(info->u.s390.reason),
info->u.s390.psw_mask,
info->u.s390.psw_addr);
+ } else if (info->type == GUEST_PANIC_INFORMATION_TYPE_TDX) {
+ bool printable = false;
+ char *buf = NULL;
+ int len = 0, i;
+
+ /*
+ * Although message is defined as a json string, we shouldn't
+ * unconditionally treat it as is because the guest generated it and
+ * it's not necessarily trustable.
+ */+ if (info->u.tdx.message) {
+ /* The caller guarantees the NUL-terminated string. */
+ len = strlen(info->u.tdx.message);
+
+ printable = len > 0;
+ for (i = 0; i < len; i++) {
+ if (!(0x20 <= info->u.tdx.message[i] &&
+ info->u.tdx.message[i] <= 0x7e)) {
+ printable = false;
+ break;
+ }
+ }
+ }
+
+ if (!printable && len) {
+ /* 3 = length of "%02x " */
+ buf = g_malloc(len * 3);
+ for (i = 0; i < len; i++) {+ if (info->u.tdx.message[i] == '\0') {
+ break;
+ } else {
+ sprintf(buf + 3 * i, "%02x ", info->u.tdx.message[i]);
+ }
+ }
+ if (i > 0)
+ /* replace the last ' '(space) to NUL */
+ buf[i * 3 - 1] = '\0';
+ else
+ buf[0] = '\0';
+ }
+
+ qemu_log_mask(LOG_GUEST_ERROR,
+ " TDX guest reports fatal error:\"%s\""
+ " error code: 0x%016" PRIx64 " gpa page: 0x%016" PRIx64 "\n",
+ printable ? info->u.tdx.message : buf,
+ info->u.tdx.error_code,
+ info->u.tdx.gpa);
+ g_free(buf);
}Quoted 50 lines
Quoted 50 lines
softmmu/runstate.c +4
-0+ info->u.tdx.error_code, + info->u.tdx.gpa); +g_free (buf); } +qapi_free_GuestPanicInformation (info); } }target/i386/kvm/tdx.c +16
-0@@ -18,6 +18,7 @@#include "qom/object_interfaces.h" #include "standard-headers/asm-x86/kvm_para.h" #include "sysemu/kvm.h" +#include "sysemu/runstate.h" #include "sysemu/sysemu.h" #include "exec/address-spaces.h" #include "exec/ramblock.h" @@ -1408,11 +1409,26 @@ static void tdx_handle_get_quote(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall) vmcall->status_code =TDG_VP_VMCALL_SUCCESS ; } +static void tdx_panicked_on_fatal_error (X86CPU *cpu,uint64_t error_code, +uint64_t gpa,char *message) +{ +GuestPanicInformation *panic_info; + + panic_info =g_new0 (GuestPanicInformation,1 ); + panic_info->type =GUEST_PANIC_INFORMATION_TYPE_TDX ; + panic_info->u.tdx.error_code = error_code; + panic_info->u.tdx.gpa = gpa; + panic_info->u.tdx.message = (char *)message; + +qemu_system_guest_panicked (panic_info); +} +static void tdx_handle_report_fatal_error (X86CPU *cpu,struct kvm_tdx_vmcall *vmcall) {uint64_t error_code = vmcall->in_r12;char *message =NULL ; +uint64_t gpa =-1ull ;if (error_code &0xffff ) {error_report ("invalid error code of TDG.VP.VMCALL<REPORT_FATAL_ERROR>\n" );@@ -1441,7 +1457,13 @@ static void tdx_handle_report_fatal_error(X86CPU *cpu, }
Quoted 4 lines
target/i386/kvm/tdx.c static void tdx_handle_report_fatal_error(X86CPU *cpu, · line 1459
Quoted 1 line
target/i386/kvm/tdx.c +0
-0error_report ("TD guest reports fatal error. %s\n" , message ? :"" );In
tdx_panicked_on_fatal_erroryou're avoiding printing 'message' if it contains non-printable characters, but here you're printing it regardless.
I guess you meant qemu_system_guest_panicked().
Quoted 1 line
Do we still need this
error_reportcall at all ?
yes. It can and should be dropped before I sent to maillist. I keep it internally for testing purpose because qemu_log_mask() doesn't get printed by default.
Quoted 7 lines
target/i386/kvm/tdx.c static void tdx_handle_report_fatal_error(X86CPU *cpu, · line 1463
Quoted 5 lines
target/i386/kvm/tdx.c +4
-1-exit (1 ); + +#define TDX_REPORT_FATAL_ERROR_GPA_VALID BIT_ULL(63) +if (error_code &TDX_REPORT_FATAL_ERROR_GPA_VALID ) { + gpa = vmcall->in_r13;Bad indent
Fixed.
Quoted 12 lines
Quoted 9 lines
target/i386/kvm/tdx.c +3
-0+ } + +tdx_panicked_on_fatal_error (cpu, error_code, gpa, message); } static void tdx_handle_setup_event_notify_interrupt(X86CPU *cpu,Signature · 1 line
2.34.1With regards,
Daniel
Re: [PATCH v2 47/58] i386/tdx: Wire REPORT_FATAL_ERROR with GuestPanic facility
Quoted 149 lines · Xiaoyao Li
Quoted 89 lines · Daniel P. Berrangé
Quoted 66 lines · Xiaoyao Li
Originated-from: Isaku Yamahata <[email protected]>
Signed-off-by Xiaoyao Li <[email protected]>--- qapi/run-state.json | 17 +++++++++++++-- softmmu/runstate.c | 49 +++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.c | 24 ++++++++++++++++++++- 3 files changed, 87 insertions(+), 3 deletions(-)qapi/run-state.json +9
-1@@ -499,7 +499,7 @@ # Since: 2.9 ## { 'enum': 'GuestPanicInformationType', - 'data': [ 'hyper-v', 's390' ] } + 'data': [ 'hyper-v', 's390', 'tdx' ] } +# +# Since: 8.2 +## +{'struct': 'GuestPanicInformationTdx', + 'data': {'error-code': 'uint64', + 'gpa': 'uint64', + 'message': 'str'}} + ## # @MEMORY_FAILURE: #softmmu/runstate.c +27
-0@@ -518,7 +518,56 @@ void qemu_system_guest_panicked(GuestPanicInformation *info)S390CrashReason_str (info->u.s390.reason), info->u.s390.psw_mask, info->u.s390.psw_addr); + }else if (info->type ==GUEST_PANIC_INFORMATION_TYPE_TDX ) { +char *buf =NULL ; +bool printable = false; + +/* +* Although message is defined as a json string, we shouldn't +* unconditionally treat it as is because the guest generated it and +* it's not necessarily trustable. +*/ +if (info->u.tdx.message) { +/* The caller guarantees the NUL-terminated string. */ +int len =strlen (info->u.tdx.message); +int i; + + printable = len >0 ; +for (i =0 ; i < len; i++) { +if (!(0x20 <= info->u.tdx.message[i] && + info->u.tdx.message[i] <=0x7e )) { + printable = false; +break ; + } + } + +/* 3 = length of "%02x " */ + buf =g_malloc (len *3 ); +for (i =0 ; i < len; i++) { +if (info->u.tdx.message[i] =='\0' ) {softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 557
Quoted 10 lines
softmmu/runstate.c +10
-0+break ; + }else { +sprintf (buf +3 * i,"%02x " , info->u.tdx.message[i]); + } + } +if (i >0 ) +/* replace the last ' '(space) to NUL */ + buf[i *3 -1 ] ='\0' ; +else + buf[0 ] ='\0' ;You're building this escaped buffer but...
softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 565
Quoted 8 lines
softmmu/runstate.c +8
-0+ } + + qemu_log_mask(LOG_GUEST_ERROR, + //" TDX report fatal error:\"%s\" %s", + " TDX report fatal error:\"%s\"" + "error: 0x%016" PRIx64 " gpa page: 0x%016" PRIx64 "\n", + printable ? info->u.tdx.message : "", +//buf ? buf : "", Quoted 3 lines
...then not actually using it
Either delete the 'buf' code, or use it.
Sorry for posting some internal testing version.
Does below look good to you?softmmu/runstate.c +8
-0@@ -518,7 +518,56 @@ void qemu_system_guest_panicked(GuestPanicInformation *info)S390CrashReason_str (info->u.s390.reason), info->u.s390.psw_mask, info->u.s390.psw_addr); + }else if (info->type ==GUEST_PANIC_INFORMATION_TYPE_TDX ) { +bool printable = false; +char *buf =NULL ; +int len =0 , i; + +/* +* Although message is defined as a json string, we shouldn't +* unconditionally treat it as is because the guest generated it and
softmmu/runstate.c +24
-0+* it's not necessarily trustable. +*/ +if (info->u.tdx.message) { +/* The caller guarantees the NUL-terminated string. */ + len =strlen (info->u.tdx.message); + + printable = len >0 ; +for (i =0 ; i < len; i++) { +if (!(0x20 <= info->u.tdx.message[i] && + info->u.tdx.message[i] <=0x7e )) { + printable = false; +break ; + } + } + } + +if (!printable && len) { +/* 3 = length of "%02x " */ + buf =g_malloc (len *3 ); +for (i =0 ; i < len; i++) { +if (info->u.tdx.message[i] =='\0' ) { +break ; + }else { + sprintf(buf + 3 * i, "%02x ",info->u.tdx.message[i]);
softmmu/runstate.c +12
-0+ } + } +if (i >0 ) +/* replace the last ' '(space) to NUL */ + buf[i *3 -1 ] ='\0' ; +else + buf[0 ] ='\0' ; + } + +qemu_log_mask (LOG_GUEST_ERROR , +" TDX guest reports fatal error:\"%s\"" + " error code: 0x%016" PRIx64 " gpa page: 0x%016"PRIx64 "\n",
softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 569
Quoted 5 lines
softmmu/runstate.c +4
-0+ printable ? info->u.tdx.message : buf, + info->u.tdx.error_code, + info->u.tdx.gpa); +g_free (buf); }
Ok that makes more sense now. BTW, probably a nice idea to create a separate helper method that santizes the guest provided JSON into the safe 'buf' string.
With regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
Re: [PATCH v2 47/58] i386/tdx: Wire REPORT_FATAL_ERROR with GuestPanic facility
Quoted 152 lines · Daniel P. Berrangé
Quoted 151 lines · Xiaoyao Li
Quoted 89 lines · Daniel P. Berrangé
Quoted 66 lines · Xiaoyao Li
Originated-from: Isaku Yamahata <[email protected]>
Signed-off-by Xiaoyao Li <[email protected]>--- qapi/run-state.json | 17 +++++++++++++-- softmmu/runstate.c | 49 +++++++++++++++++++++++++++++++++++++++++++ target/i386/kvm/tdx.c | 24 ++++++++++++++++++++- 3 files changed, 87 insertions(+), 3 deletions(-)qapi/run-state.json +9
-1@@ -499,7 +499,7 @@ # Since: 2.9 ## { 'enum': 'GuestPanicInformationType', - 'data': [ 'hyper-v', 's390' ] } + 'data': [ 'hyper-v', 's390', 'tdx' ] } +# +# Since: 8.2 +## +{'struct': 'GuestPanicInformationTdx', + 'data': {'error-code': 'uint64', + 'gpa': 'uint64', + 'message': 'str'}} + ## # @MEMORY_FAILURE: #softmmu/runstate.c +27
-0@@ -518,7 +518,56 @@ void qemu_system_guest_panicked(GuestPanicInformation *info)S390CrashReason_str (info->u.s390.reason), info->u.s390.psw_mask, info->u.s390.psw_addr); + }else if (info->type ==GUEST_PANIC_INFORMATION_TYPE_TDX ) { +char *buf =NULL ; +bool printable = false; + +/* +* Although message is defined as a json string, we shouldn't +* unconditionally treat it as is because the guest generated it and +* it's not necessarily trustable. +*/ +if (info->u.tdx.message) { +/* The caller guarantees the NUL-terminated string. */ +int len =strlen (info->u.tdx.message); +int i; + + printable = len >0 ; +for (i =0 ; i < len; i++) { +if (!(0x20 <= info->u.tdx.message[i] && + info->u.tdx.message[i] <=0x7e )) { + printable = false; +break ; + } + } + +/* 3 = length of "%02x " */ + buf =g_malloc (len *3 ); +for (i =0 ; i < len; i++) { +if (info->u.tdx.message[i] =='\0' ) {softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 557
Quoted 10 lines
softmmu/runstate.c +10
-0+break ; + }else { +sprintf (buf +3 * i,"%02x " , info->u.tdx.message[i]); + } + } +if (i >0 ) +/* replace the last ' '(space) to NUL */ + buf[i *3 -1 ] ='\0' ; +else + buf[0 ] ='\0' ;You're building this escaped buffer but...
softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 565
Quoted 8 lines
softmmu/runstate.c +8
-0+ } + + qemu_log_mask(LOG_GUEST_ERROR, + //" TDX report fatal error:\"%s\" %s", + " TDX report fatal error:\"%s\"" + "error: 0x%016" PRIx64 " gpa page: 0x%016" PRIx64 "\n", + printable ? info->u.tdx.message : "", +//buf ? buf : "", Quoted 3 lines
...then not actually using it
Either delete the 'buf' code, or use it.
Sorry for posting some internal testing version.
Does below look good to you?softmmu/runstate.c +8
-0@@ -518,7 +518,56 @@ void qemu_system_guest_panicked(GuestPanicInformation *info)S390CrashReason_str (info->u.s390.reason), info->u.s390.psw_mask, info->u.s390.psw_addr); + }else if (info->type ==GUEST_PANIC_INFORMATION_TYPE_TDX ) { +bool printable = false; +char *buf =NULL ; +int len =0 , i; + +/* +* Although message is defined as a json string, we shouldn't +* unconditionally treat it as is because the guest generated it and
softmmu/runstate.c +24
-0+* it's not necessarily trustable. +*/ +if (info->u.tdx.message) { +/* The caller guarantees the NUL-terminated string. */ + len =strlen (info->u.tdx.message); + + printable = len >0 ; +for (i =0 ; i < len; i++) { +if (!(0x20 <= info->u.tdx.message[i] && + info->u.tdx.message[i] <=0x7e )) { + printable = false; +break ; + } + } + } + +if (!printable && len) { +/* 3 = length of "%02x " */ + buf =g_malloc (len *3 ); +for (i =0 ; i < len; i++) { +if (info->u.tdx.message[i] =='\0' ) { +break ; + }else { + sprintf(buf + 3 * i, "%02x ",info->u.tdx.message[i]);
softmmu/runstate.c +12
-0+ } + } +if (i >0 ) +/* replace the last ' '(space) to NUL */ + buf[i *3 -1 ] ='\0' ; +else + buf[0 ] ='\0' ; + } + +qemu_log_mask (LOG_GUEST_ERROR , +" TDX guest reports fatal error:\"%s\"" + " error code: 0x%016" PRIx64 " gpa page: 0x%016"PRIx64 "\n",
(quoted diff) +2
-0+ printable ? info->u.tdx.message : buf, + info->u.tdx.error_code,
Quoted 9 lines
softmmu/runstate.c void qemu_system_guest_panicked(GuestPanicInformation *info) · line 569
Quoted 3 lines
softmmu/runstate.c +2
-0+ info->u.tdx.gpa); +g_free (buf); }Ok that makes more sense now. BTW, probably a nice idea to create a separate helper method that santizes the guest provided JSON into the safe 'buf' string.
OK. Thanks for the suggestion.
Will do it in next version.
Quoted 2 lines
With regards,
Daniel
PATCH v2 48/58 i386/tdx: Disable SMM for TDX VMs
In reply to Xiaoyao LiTDX doesn't support SMM and VMM cannot emulate SMM for TDX VMs because VMM cannot manipulate TDX VM's memory.
Disable SMM for TDX VMs and error out if user requests to enable SMM.
--- target/i386/kvm/tdx.c | 8 ++++++++ 1 file changed, 8 insertions(+)
target/i386/kvm/tdx.c +8 -0
@@ -678,9 +678,17 @@ static Notifier tdx_machine_done_notify = {int tdx_kvm_init (MachineState *ms,Error **errp) { +X86MachineState *x86ms =X86_MACHINE (ms);TdxGuest *tdx = (TdxGuest *)object_dynamic_cast (OBJECT (ms->cgs),TYPE_TDX_GUEST ); +if (x86ms->smm ==ON_OFF_AUTO_AUTO ) { + x86ms->smm =ON_OFF_AUTO_OFF ; + }else if (x86ms->smm ==ON_OFF_AUTO_ON ) { +error_setg (errp,"TDX VM doesn't support SMM" ); +return -EINVAL ; + } +if (!tdx_caps) {get_tdx_capabilities (); }
Signature · 1 line
2.34.1
PATCH v2 49/58 i386/tdx: Disable PIC for TDX VMs
In reply to Xiaoyao LiLegacy PIC (8259) cannot be supported for TDX VMs since TDX module doesn't allow directly interrupt injection. Using posted interrupts for the PIC is not a viable option as the guest BIOS/kernel will not do EOI for PIC IRQs, i.e. will leave the vIRR bit set.
Hence disable PIC for TDX VMs and error out if user wants PIC.
--- target/i386/kvm/tdx.c | 7 +++++++ 1 file changed, 7 insertions(+)
target/i386/kvm/tdx.c +7 -0
@@ -689,6 +689,13 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return -EINVAL ; } +if (x86ms->pic ==ON_OFF_AUTO_AUTO ) { + x86ms->pic =ON_OFF_AUTO_OFF ; + }else if (x86ms->pic ==ON_OFF_AUTO_ON ) { +error_setg (errp,"TDX VM doesn't support PIC" ); +return -EINVAL ; + } +if (!tdx_caps) {get_tdx_capabilities (); }
Signature · 1 line
2.34.1
PATCH v2 50/58 i386/tdx: Don't allow system reset for TDX VMs
In reply to Xiaoyao LiTDX CPU state is protected and thus vcpu state cann't be reset by VMM.
--- target/i386/kvm/kvm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
target/i386/kvm/kvm.c +1 -1
@@ -5918,7 +5918,7 @@ bool kvm_has_waitpkg(void)bool kvm_arch_cpu_check_are_resettable (void ) { -return !sev_es_enabled (); +return !sev_es_enabled () && !is_tdx_vm (); }#define ARCH_REQ_XCOMP_GUEST_PERM 0x1025
Signature · 1 line
2.34.1
PATCH v2 51/58 i386/tdx: LMCE is not supported for TDX
In reply to Xiaoyao LiLMCE is not supported TDX since KVM doesn't provide emulation for MSR_IA32_FEAT_CTL.
--- target/i386/kvm/kvm-cpu.c | 5 +++++ 1 file changed, 5 insertions(+)
target/i386/kvm/kvm-cpu.c +5 -0
@@ -15,6 +15,7 @@#include "sysemu/sysemu.h" #include "hw/boards.h" +#include "tdx.h" #include "kvm_i386.h" #include "hw/core/accel-cpu.h"
@@ -59,6 +60,10 @@ static bool lmce_supported(void)if (kvm_ioctl (kvm_state,KVM_X86_GET_MCE_CAP_SUPPORTED , &mce_cap) <0 ) {return false; } + +if (is_tdx_vm ()) +return false; +return !!(mce_cap &MCG_LMCE_P ); }
Signature · 1 line
2.34.1
PATCH v2 52/58 hw/i386: add eoi_intercept_unsupported member to X86MachineState
In reply to Xiaoyao LiAdd a new bool member, eoi_intercept_unsupported, to X86MachineState with default value false. Set true for TDX VM.
Inability to intercept eoi causes impossibility to emulate level triggered interrupt to be re-injected when level is still kept active. which affects interrupt controller emulation.
--- hw/i386/x86.c | 1 + include/hw/i386/x86.h | 1 + target/i386/kvm/tdx.c | 2 ++ 3 files changed, 4 insertions(+)
hw/i386/x86.c +1 -0
@@ -1426,6 +1426,7 @@ static void x86_machine_initfn(Object *obj) x86ms->oem_table_id =g_strndup (ACPI_BUILD_APPNAME8 ,8 ); x86ms->bus_lock_ratelimit =0 ; x86ms->above_4g_mem_start =4 * GiB; + x86ms->eoi_intercept_unsupported = false; }static void x86_machine_class_init (ObjectClass *oc,void *data)
include/hw/i386/x86.h +1 -0
@@ -61,6 +61,7 @@ struct X86MachineState {/* CPU and apic information: */ bool apic_xrupt_override; +bool eoi_intercept_unsupported;unsigned pci_irq_mask;unsigned apic_id_limit;uint16_t boot_cpus;
target/i386/kvm/tdx.c +2 -0
@@ -696,6 +696,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp)return -EINVAL ; } + x86ms->eoi_intercept_unsupported = true; +if (!tdx_caps) {get_tdx_capabilities (); }
Signature · 1 line
2.34.1
PATCH v2 53/58 hw/i386: add option to forcibly report edge trigger in acpi tables
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
When level trigger isn't supported on x86 platform, forcibly report edge trigger in acpi tables.
--- hw/i386/acpi-build.c | 99 ++++++++++++++++++++++++++++--------------- hw/i386/acpi-common.c | 50 ++++++++++++++++------ 2 files changed, 104 insertions(+), 45 deletions(-)
hw/i386/acpi-build.c +66 -33
@@ -976,7 +976,8 @@ static void build_dbg_aml(Aml *table)aml_append (table, scope); } -static Aml *build_link_dev (const char *name,uint8_t uid,Aml *reg) +static Aml *build_link_dev (const char *name,uint8_t uid,Aml *reg, +bool level_trigger_unsupported) {Aml *dev;Aml *crs;
@@ -988,7 +989,10 @@ static Aml *build_link_dev(const char *name, uint8_t uid, Aml *reg)aml_append (dev,aml_name_decl ("_UID" ,aml_int (uid))); crs =aml_resource_template (); -aml_append (crs,aml_interrupt (AML_CONSUMER ,AML_LEVEL ,AML_ACTIVE_HIGH , +aml_append (crs,aml_interrupt (AML_CONSUMER , + level_trigger_unsupported ? +AML_EDGE :AML_LEVEL , +AML_ACTIVE_HIGH ,AML_SHARED , irqs,ARRAY_SIZE (irqs)));aml_append (dev,aml_name_decl ("_PRS" , crs));
@@ -1012,7 +1016,8 @@ static Aml *build_link_dev(const char *name, uint8_t uid, Aml *reg)return dev; } -static Aml *build_gsi_link_dev (const char *name,uint8_t uid,uint8_t gsi) +static Aml *build_gsi_link_dev (const char *name,uint8_t uid, +uint8_t gsi,bool level_trigger_unsupported) {Aml *dev;Aml *crs;
@@ -1025,7 +1030,10 @@ static Aml *build_gsi_link_dev(const char *name, uint8_t uid, uint8_t gsi) crs =aml_resource_template (); irqs = gsi; -aml_append (crs,aml_interrupt (AML_CONSUMER ,AML_LEVEL ,AML_ACTIVE_HIGH , +aml_append (crs,aml_interrupt (AML_CONSUMER , + level_trigger_unsupported ? +AML_EDGE :AML_LEVEL , +AML_ACTIVE_HIGH ,AML_SHARED , &irqs,1 ));aml_append (dev,aml_name_decl ("_PRS" , crs));
@@ -1044,7 +1052,7 @@ static Aml *build_gsi_link_dev(const char *name, uint8_t uid, uint8_t gsi) }/* _CRS method - get current settings */ -static Aml *build_iqcr_method (bool is_piix4) +static Aml *build_iqcr_method (bool is_piix4,bool level_trigger_unsupported) {Aml *if_ctx;uint32_t irqs;
@@ -1052,7 +1060,9 @@ static Aml *build_iqcr_method(bool is_piix4)Aml *crs =aml_resource_template (); irqs =0 ; -aml_append (crs,aml_interrupt (AML_CONSUMER ,AML_LEVEL , +aml_append (crs,aml_interrupt (AML_CONSUMER , + level_trigger_unsupported ? +AML_EDGE :AML_LEVEL ,AML_ACTIVE_HIGH ,AML_SHARED , &irqs,1 ));aml_append (method,aml_name_decl ("PRR0" , crs));
@@ -1086,7 +1096,7 @@ static Aml *build_irq_status_method(void)return method; } -static void build_piix4_pci0_int (Aml *table) +static void build_piix4_pci0_int (Aml *table,bool level_trigger_unsupported) {Aml *dev;Aml *crs;
@@ -1099,12 +1109,16 @@ static void build_piix4_pci0_int(Aml *table)aml_append (sb_scope, pci0_scope);aml_append (sb_scope,build_irq_status_method ()); -aml_append (sb_scope,build_iqcr_method (true)); +aml_append (sb_scope,build_iqcr_method (true, level_trigger_unsupported)); -aml_append (sb_scope,build_link_dev ("LNKA" ,0 ,aml_name ("PRQ0" ))); -aml_append (sb_scope,build_link_dev ("LNKB" ,1 ,aml_name ("PRQ1" ))); -aml_append (sb_scope,build_link_dev ("LNKC" ,2 ,aml_name ("PRQ2" ))); -aml_append (sb_scope,build_link_dev ("LNKD" ,3 ,aml_name ("PRQ3" ))); +aml_append (sb_scope,build_link_dev ("LNKA" ,0 ,aml_name ("PRQ0" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKB" ,1 ,aml_name ("PRQ1" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKC" ,2 ,aml_name ("PRQ2" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKD" ,3 ,aml_name ("PRQ3" ), + level_trigger_unsupported)); dev =aml_device ("LNKS" ); {
@@ -1113,7 +1127,9 @@ static void build_piix4_pci0_int(Aml *table) crs =aml_resource_template (); irqs =9 ; -aml_append (crs,aml_interrupt (AML_CONSUMER ,AML_LEVEL , +aml_append (crs,aml_interrupt (AML_CONSUMER , + level_trigger_unsupported ? +AML_EDGE :AML_LEVEL ,AML_ACTIVE_HIGH ,AML_SHARED , &irqs,1 ));aml_append (dev,aml_name_decl ("_PRS" , crs));
@@ -1199,7 +1215,7 @@ static Aml *build_q35_routing_table(const char *str)return pkg; } -static void build_q35_pci0_int (Aml *table) +static void build_q35_pci0_int (Aml *table,bool level_trigger_unsupported) {Aml *method;Aml *sb_scope =aml_scope ("_SB" );
@@ -1238,25 +1254,41 @@ static void build_q35_pci0_int(Aml *table)aml_append (sb_scope, pci0_scope);aml_append (sb_scope,build_irq_status_method ()); -aml_append (sb_scope,build_iqcr_method (false)); +aml_append (sb_scope,build_iqcr_method (false, level_trigger_unsupported)); -aml_append (sb_scope,build_link_dev ("LNKA" ,0 ,aml_name ("PRQA" ))); -aml_append (sb_scope,build_link_dev ("LNKB" ,1 ,aml_name ("PRQB" ))); -aml_append (sb_scope,build_link_dev ("LNKC" ,2 ,aml_name ("PRQC" ))); -aml_append (sb_scope,build_link_dev ("LNKD" ,3 ,aml_name ("PRQD" ))); -aml_append (sb_scope,build_link_dev ("LNKE" ,4 ,aml_name ("PRQE" ))); -aml_append (sb_scope,build_link_dev ("LNKF" ,5 ,aml_name ("PRQF" ))); -aml_append (sb_scope,build_link_dev ("LNKG" ,6 ,aml_name ("PRQG" ))); -aml_append (sb_scope,build_link_dev ("LNKH" ,7 ,aml_name ("PRQH" ))); +aml_append (sb_scope,build_link_dev ("LNKA" ,0 ,aml_name ("PRQA" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKB" ,1 ,aml_name ("PRQB" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKC" ,2 ,aml_name ("PRQC" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKD" ,3 ,aml_name ("PRQD" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKE" ,4 ,aml_name ("PRQE" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKF" ,5 ,aml_name ("PRQF" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKG" ,6 ,aml_name ("PRQG" ), + level_trigger_unsupported)); +aml_append (sb_scope,build_link_dev ("LNKH" ,7 ,aml_name ("PRQH" ), + level_trigger_unsupported)); -aml_append (sb_scope,build_gsi_link_dev ("GSIA" ,0x10 ,0x10 )); -aml_append (sb_scope,build_gsi_link_dev ("GSIB" ,0x11 ,0x11 )); -aml_append (sb_scope,build_gsi_link_dev ("GSIC" ,0x12 ,0x12 )); -aml_append (sb_scope,build_gsi_link_dev ("GSID" ,0x13 ,0x13 )); -aml_append (sb_scope,build_gsi_link_dev ("GSIE" ,0x14 ,0x14 )); -aml_append (sb_scope,build_gsi_link_dev ("GSIF" ,0x15 ,0x15 )); -aml_append (sb_scope,build_gsi_link_dev ("GSIG" ,0x16 ,0x16 )); -aml_append (sb_scope,build_gsi_link_dev ("GSIH" ,0x17 ,0x17 )); +aml_append (sb_scope,build_gsi_link_dev ("GSIA" ,0x10 ,0x10 , + level_trigger_unsupported)); +aml_append (sb_scope,build_gsi_link_dev ("GSIB" ,0x11 ,0x11 , + level_trigger_unsupported)); +aml_append (sb_scope,build_gsi_link_dev ("GSIC" ,0x12 ,0x12 , + level_trigger_unsupported)); +aml_append (sb_scope,build_gsi_link_dev ("GSID" ,0x13 ,0x13 , + level_trigger_unsupported)); +aml_append (sb_scope,build_gsi_link_dev ("GSIE" ,0x14 ,0x14 , + level_trigger_unsupported)); +aml_append (sb_scope,build_gsi_link_dev ("GSIF" ,0x15 ,0x15 , + level_trigger_unsupported)); +aml_append (sb_scope,build_gsi_link_dev ("GSIG" ,0x16 ,0x16 , + level_trigger_unsupported)); +aml_append (sb_scope,build_gsi_link_dev ("GSIH" ,0x17 ,0x17 , + level_trigger_unsupported));aml_append (table, sb_scope); }
@@ -1436,6 +1468,7 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,PCMachineState *pcms =PC_MACHINE (machine);PCMachineClass *pcmc =PC_MACHINE_GET_CLASS (machine);X86MachineState *x86ms =X86_MACHINE (machine); +bool level_trigger_unsupported = x86ms->eoi_intercept_unsupported;AcpiMcfgInfo mcfg;bool mcfg_valid = !!acpi_get_mcfg (&mcfg);uint32_t nr_mem = machine->ram_slots;
@@ -1469,7 +1502,7 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,if (pm->pcihp_bridge_en || pm->pcihp_root_en) {build_x86_acpi_pci_hotplug (dsdt, pm->pcihp_io_base); } -build_piix4_pci0_int (dsdt); +build_piix4_pci0_int (dsdt, level_trigger_unsupported); }else if (q35) { sb_scope =aml_scope ("_SB" ); dev =aml_device ("PCI0" );
@@ -1514,7 +1547,7 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,if (pm->pcihp_bridge_en) {build_x86_acpi_pci_hotplug (dsdt, pm->pcihp_io_base); } -build_q35_pci0_int (dsdt); +build_q35_pci0_int (dsdt, level_trigger_unsupported); }if (misc->has_hpet) {
hw/i386/acpi-common.c +38 -12
62 lines not on this page · raw message
Signature · 1 line
2.34.1
PATCH v2 54/58 i386/tdx: Don't synchronize guest tsc for TDs
In reply to Xiaoyao LiFrom: Isaku Yamahata <[email protected]>
TSC of TDs is not accessible and KVM doesn't allow access of MSR_IA32_TSC for TDs. To avoid the assert() in kvm_get_tsc, make kvm_synchronize_all_tsc() noop for TDs,
--- target/i386/kvm/kvm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
target/i386/kvm/kvm.c +1 -1
@@ -298,7 +298,7 @@ void kvm_synchronize_all_tsc(void) {CPUState *cpu; -if (kvm_enabled ()) { +if (kvm_enabled () && !is_tdx_vm ()) {CPU_FOREACH (cpu) {run_on_cpu (cpu, do_kvm_synchronize_tsc,RUN_ON_CPU_NULL ); }
Signature · 1 line
2.34.1
PATCH v2 55/58 i386/tdx: Only configure MSR_IA32_UCODE_REV in kvm_init_msrs() for TDs
In reply to Xiaoyao LiFor TDs, only MSR_IA32_UCODE_REV in kvm_init_msrs() can be configured by VMM, while the features enumerated/controlled by other MSRs except MSR_IA32_UCODE_REV in kvm_init_msrs() are not under control of VMM.
Only configure MSR_IA32_UCODE_REV for TDs.
--- target/i386/kvm/kvm.c | 44 ++++++++++++++++++++++--------------------- 1 file changed, 23 insertions(+), 21 deletions(-)
target/i386/kvm/kvm.c +23 -21
@@ -3426,32 +3426,34 @@ static void kvm_init_msrs(X86CPU *cpu)CPUX86State *env = &cpu->env;kvm_msr_buf_reset (cpu); -if (has_msr_arch_capabs) { -kvm_msr_entry_add (cpu,MSR_IA32_ARCH_CAPABILITIES , - env->features[FEAT_ARCH_CAPABILITIES ]); - } - -if (has_msr_core_capabs) { -kvm_msr_entry_add (cpu,MSR_IA32_CORE_CAPABILITY , - env->features[FEAT_CORE_CAPABILITY ]); - } - -if (has_msr_perf_capabs && cpu->enable_pmu) { -kvm_msr_entry_add_perf (cpu, env->features); + +if (!is_tdx_vm ()) { +if (has_msr_arch_capabs) { +kvm_msr_entry_add (cpu,MSR_IA32_ARCH_CAPABILITIES , + env->features[FEAT_ARCH_CAPABILITIES ]); + } + +if (has_msr_core_capabs) { +kvm_msr_entry_add (cpu,MSR_IA32_CORE_CAPABILITY , + env->features[FEAT_CORE_CAPABILITY ]); + } + +if (has_msr_perf_capabs && cpu->enable_pmu) { +kvm_msr_entry_add_perf (cpu, env->features); + } + +/* +* Older kernels do not include VMX MSRs in KVM_GET_MSR_INDEX_LIST, but +* all kernels with MSR features should have them. +*/ +if (kvm_feature_msrs &&cpu_has_vmx (env)) { +kvm_msr_entry_add_vmx (cpu, env->features); + } }if (has_msr_ucode_rev) {kvm_msr_entry_add (cpu,MSR_IA32_UCODE_REV , cpu->ucode_rev); } - -/* -* Older kernels do not include VMX MSRs in KVM_GET_MSR_INDEX_LIST, but -* all kernels with MSR features should have them. -*/ -if (kvm_feature_msrs &&cpu_has_vmx (env)) { -kvm_msr_entry_add_vmx (cpu, env->features); - } -assert (kvm_buf_set_msrs (cpu) ==0 ); }
Signature · 1 line
2.34.1
PATCH v2 56/58 i386/tdx: Skip kvm_put_apicbase() for TDs
In reply to Xiaoyao LiKVM doesn't allow wirting to MSR_IA32_APICBASE for TDs.
--- target/i386/kvm/kvm.c | 5 +++++ 1 file changed, 5 insertions(+)
target/i386/kvm/kvm.c +5 -0
@@ -3208,6 +3208,11 @@ void kvm_put_apicbase(X86CPU *cpu, uint64_t value) {int ret; +/* TODO: Allow accessing guest state for debug TDs. */ +if (is_tdx_vm ()) { +return ; + } + ret =kvm_put_one_msr (cpu,MSR_IA32_APICBASE , value);assert (ret ==1 ); }
Signature · 1 line
2.34.1
PATCH v2 57/58 i386/tdx: Don't get/put guest state for TDX VMs
In reply to Xiaoyao LiFrom: Sean Christopherson <[email protected]>
Don't get/put state of TDX VMs since accessing/mutating guest state of production TDs is not supported.
Note, it will be allowed for a debug TD. Corresponding support will be introduced when debug TD support is implemented in the future.
--- target/i386/kvm/kvm.c | 11 +++++++++++ 1 file changed, 11 insertions(+)
target/i386/kvm/kvm.c +11 -0
@@ -4852,6 +4852,11 @@ int kvm_arch_put_registers(CPUState *cpu, int level)assert (cpu_is_stopped (cpu) ||qemu_cpu_is_self (cpu)); +/* TODO: Allow accessing guest state for debug TDs. */ +if (is_tdx_vm ()) { +return 0 ; + } +/* * Put MSR_IA32_FEATURE_CONTROL first, this ensures the VM gets out of VMX * root operation upon vCPU reset. kvm_put_msr_feature_control() should also
@@ -4962,6 +4967,12 @@ int kvm_arch_get_registers(CPUState *cs)if (ret <0 ) { goto out; } + +/* TODO: Allow accessing guest state for debug TDs. */ +if (is_tdx_vm ()) { +return 0 ; + } + ret =kvm_getput_regs (cpu,0 );if (ret <0 ) { goto out;
Signature · 1 line
2.34.1
PATCH v2 58/58 docs: Add TDX documentation
In reply to Xiaoyao LiAdd docs/system/i386/tdx.rst for TDX support, and add tdx in confidential-guest-support.rst
---
Changes since v1:
- Add prerequisite of private gmem;
- update example command to launch TD;
Changes since RFC v4:
- add the restriction that kernel-irqchip must be split
--- docs/system/confidential-guest-support.rst | 1 + docs/system/i386/tdx.rst | 114 +++++++++++++++++++++ docs/system/target-i386.rst | 1 + 3 files changed, 116 insertions(+) create mode 100644 docs/system/i386/tdx.rst
docs/system/confidential-guest-support.rst +1 -0
@@ -38,6 +38,7 @@ Supported mechanisms Currently supported confidential guest mechanisms are: * AMD Secure Encrypted Virtualization (SEV) (see :doc:`i386/amd-memory-encryption`) +* Intel Trust Domain Extension (TDX) (see :doc:`i386/tdx`) * POWER Protected Execution Facility (PEF) (see :ref:`power-papr-protected-execution-facility-pef`) * s390x Protected Virtualization (PV) (see :doc:`s390x/protvirt`)
docs/system/i386/tdx.rst new file +114 -0
114 lines not on this page · raw message
docs/system/target-i386.rst +1 -0
@@ -29,6 +29,7 @@ Architectural features i386/kvm-pv i386/sgx i386/amd-memory-encryption + i386/tdx OS requirements ~~~~~~~~~~~~~~~
Signature · 1 line
2.34.1