Skip to contentSkip to conversation

QEMU Guest memfd + QEMU TDX support

PATCH Superseded Xiaoyao Li 5 participants 94 messages Jan 25, 2024 – Feb 27, 2024 1 message not in the archive

Subsystems (5)

v4 of 9 Superseded Latest is v9 →

lore
Share & tools
lore

Xiaoyao Li

author
#0 lore raw

v4 cover QEMU Guest memfd + QEMU TDX support

This v4 series still carry both QEMU guest memfd support and QEMU TDX support. Because TDX is going to be the first user of guest memfd in QEMU, put them together can provide us a full picture of how they work.

KVM provides guest memfd, which cannot be mapped, read, or written by userspace. It's designed to serve as private memory for confidential VMs, like Intel TDX and AMD sev-snp.

Patches 1 - 9 add support of guest memfd into QEMU, associating it with RAMBlock. For the VM types that require private memory (see tdx_kvm_init() in patch 14), QEMU will automatically create guest memfd for each RAM backend.

Follwoing patches 10 to 66, enables TDX support to allow creating and booting a TD (TDX VM) with QEMU.

This series needs to work with KVM TDX v18 series:
https://lore.kernel.org/all/[email protected]/T/#u

(TDX host kernel support and guest memfd KVM support lands in Linux v6.8-rc1.)

===
This series is also available in github:
https://github.com/intel/qemu-tdx/tree/tdx-qemu-upstream-v4

It's based on several patches that haven't get merged:
https://lore.kernel.org/qemu-devel/[email protected]/
https://lore.kernel.org/qemu-devel/[email protected]/
https://lore.kernel.org/all/[email protected]/

Luckily, the absence of them doesn't block applying this series nor affecting the functionality.

== 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.

===
Main changes in v4:
guset memfd part:
- Drop the patch that supports KVM_GUEST_MEMFD_ALLOW_HUGEPAGE flag because KVM side leaves it as future work;
- send patch "physmem: replace function name with __func__ in ram_block_discard_range()" separately;
- drop ram_block_convert_range() and open code it instead.
- move the supported attribute bit check into kvm_set_memory_attributes() (per request from Wei)
- Clarify more for the desgin consideration in the commit messages

TDX part:
- Re-implement get-quote support by introduing TdxQuoteGenerator object and make the socket communication non-blocking; (per Daniel)
- Various code and doc refinement;

v3:
https://lore.kernel.org/qemu-devel/[email protected]/

Main changes in v3:
gmem memfd part:
- Since KVM side renamed gmem to guest_memfd in the uapi, this version renames it accordingly;
- Drop the 'private' property of memory backend. (see comment[1]) Now QEMU decides whether need to create guest memfd based on specific vm type (or specific VM implementation, please see patch *X* and *Y*);
- Drop sw_protected_vm implementation;

TDX part:
- improve the error report in various patches by utilizing 'errp';
- drop the vm-type interface;
- rename __tdx_ioctl() to tdx_ioctl_internal();
- refine the description of 'sept-ve-disable' in qom.json;
- use base64 for mrconfigif/mrowner/mrownerconfig instread of hex-string;
- use type SocketAddress for quote-generation-service;

[1] https://lore.kernel.org/qemu-devel/[email protected]/

Chao Peng (3):
kvm: Enable KVM_SET_USER_MEMORY_REGION2 for memslot
kvm: handle KVM_EXIT_MEMORY_FAULT
i386/tdx: register TDVF as private memory

Isaku Yamahata (14):
trace/kvm: Add trace for page convertion between shared and private i386/tdx: Make sept_ve_disable set by default
i386/tdx: Support user configurable mrconfigid/mrowner/mrownerconfig 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 pci-host/q35: Move PAM initialization above SMRAM initialization
q35: Introduce smm_ranges property for q35-pci-host 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 (47):
linux-headers: Update to Linux v6.8-rc1
RAMBlock: Add support of KVM private guest memfd
HostMem: Add mechanism to opt in kvm guest memfd via MachineState trace/kvm: Split address space and slot id in
trace_kvm_set_user_memory()
kvm: Introduce support for memory_attributes
physmem: Introduce ram_block_discard_guest_memfd_range() *** HACK *** linux-headers: Update headers to pull in TDX API changes
i386: Introduce tdx-guest object
target/i386: Implement mc->kvm_type() to get 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: Make Intel-PT unsupported for TD guest
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 kvm/memory: Introduce the infrastructure to set the default
shared/private value
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_guest_memfd() 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 TDX_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                           |  278 +++-
 accel/kvm/trace-events                        |    3 +-
 accel/stubs/kvm-stub.c                        |    5 +
 backends/hostmem-file.c                       |    1 +
 backends/hostmem-memfd.c                      |    1 +
 backends/hostmem-ram.c                        |    1 +
 backends/hostmem.c                            |    1 +
 configs/devices/i386-softmmu/default.mak      |    1 +
 docs/system/confidential-guest-support.rst    |    1 +
 docs/system/i386/tdx.rst                      |  113 ++
 docs/system/target-i386.rst                   |    1 +
 hw/core/machine.c                             |    5 +
 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_q35.c                              |    2 +
 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                                 |   51 +-
 hw/pci-host/q35.c                             |   61 +-
 include/exec/cpu-common.h                     |    2 +
 include/exec/memory.h                         |   33 +-
 include/exec/ram_addr.h                       |    2 +-
 include/exec/ramblock.h                       |    1 +
 include/hw/boards.h                           |    2 +
 include/hw/i386/pc.h                          |    1 +
 include/hw/i386/tdvf.h                        |   58 +
 include/hw/i386/x86.h                         |    2 +
 include/hw/pci-host/q35.h                     |    1 +
 include/standard-headers/drm/drm_fourcc.h     |   10 +-
 include/standard-headers/linux/ethtool.h      |   41 +-
 .../standard-headers/linux/virtio_config.h    |    8 +-
 include/standard-headers/linux/virtio_pci.h   |   68 +
 include/standard-headers/linux/virtio_pmem.h  |    7 +
 include/standard-headers/uefi/uefi.h          |  198 +++
 include/sysemu/hostmem.h                      |    1 +
 include/sysemu/kvm.h                          |    8 +
 include/sysemu/kvm_int.h                      |    2 +
 linux-headers/asm-generic/unistd.h            |   15 +-
 linux-headers/asm-loongarch/kvm.h             |    1 +
 linux-headers/asm-mips/mman.h                 |    2 +-
 linux-headers/asm-mips/unistd_n32.h           |    5 +
 linux-headers/asm-mips/unistd_n64.h           |    5 +
 linux-headers/asm-mips/unistd_o32.h           |    5 +
 linux-headers/asm-powerpc/unistd_32.h         |    5 +
 linux-headers/asm-powerpc/unistd_64.h         |    5 +
 linux-headers/asm-riscv/kvm.h                 |   13 +
 linux-headers/asm-s390/unistd_32.h            |    5 +
 linux-headers/asm-s390/unistd_64.h            |    5 +
 linux-headers/asm-x86/kvm.h                   |   94 ++
 linux-headers/asm-x86/unistd_32.h             |    5 +
 linux-headers/asm-x86/unistd_64.h             |    5 +
 linux-headers/asm-x86/unistd_x32.h            |    5 +
 linux-headers/linux/iommufd.h                 |   79 +
 linux-headers/linux/kvm.h                     |  229 +--
 linux-headers/linux/userfaultfd.h             |   29 +-
 linux-headers/linux/vfio.h                    |    1 +
 qapi/qom.json                                 |   33 +
 qapi/run-state.json                           |   28 +-
 system/memory.c                               |   45 +
 system/physmem.c                              |   47 +-
 system/runstate.c                             |   54 +
 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                         |  611 +++++---
 target/i386/kvm/kvm_i386.h                    |    6 +
 target/i386/kvm/meson.build                   |    2 +
 target/i386/kvm/tdx-quote-generator.c         |  170 ++
 target/i386/kvm/tdx-quote-generator.h         |   95 ++
 target/i386/kvm/tdx-stub.c                    |   23 +
 target/i386/kvm/tdx.c                         | 1373 +++++++++++++++++
 target/i386/kvm/tdx.h                         |   82 +
 target/i386/sev.c                             |    1 -
 target/i386/sev.h                             |    2 +
 80 files changed, 4156 insertions(+), 501 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-quote-generator.c
 create mode 100644 target/i386/kvm/tdx-quote-generator.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
author
#1 lore raw

PATCH v4 1/66 linux-headers: Update to Linux v6.8-rc1

Guest memfd support in QEMU requires corresponding KVM guest memfd APIs, which lands in Linux from v6.8-rc1.

---
 include/standard-headers/drm/drm_fourcc.h     |  10 +-
 include/standard-headers/linux/ethtool.h      |  41 +++--
 .../standard-headers/linux/virtio_config.h    |   8 +-
 include/standard-headers/linux/virtio_pci.h   |  68 +++++++++
 include/standard-headers/linux/virtio_pmem.h  |   7 +
 linux-headers/asm-generic/unistd.h            |  15 +-
 linux-headers/asm-loongarch/kvm.h             |   1 +
 linux-headers/asm-mips/mman.h                 |   2 +-
 linux-headers/asm-mips/unistd_n32.h           |   5 +
 linux-headers/asm-mips/unistd_n64.h           |   5 +
 linux-headers/asm-mips/unistd_o32.h           |   5 +
 linux-headers/asm-powerpc/unistd_32.h         |   5 +
 linux-headers/asm-powerpc/unistd_64.h         |   5 +
 linux-headers/asm-riscv/kvm.h                 |  13 ++
 linux-headers/asm-s390/unistd_32.h            |   5 +
 linux-headers/asm-s390/unistd_64.h            |   5 +
 linux-headers/asm-x86/kvm.h                   |   3 +
 linux-headers/asm-x86/unistd_32.h             |   5 +
 linux-headers/asm-x86/unistd_64.h             |   5 +
 linux-headers/asm-x86/unistd_x32.h            |   5 +
 linux-headers/linux/iommufd.h                 |  79 ++++++++++
 linux-headers/linux/kvm.h                     | 140 +++++++-----------
 linux-headers/linux/userfaultfd.h             |  29 +++-
 linux-headers/linux/vfio.h                    |   1 +
 24 files changed, 355 insertions(+), 112 deletions(-)
include/standard-headers/drm/drm_fourcc.h +5 -5
index 3afb70160f0b..b72917073d8d 100644
@@ -53,7 +53,7 @@ extern "C" {
  * Format modifiers may change any property of the buffer, including the number
  * of planes and/or the required allocation size. Format modifiers are
  * vendor-namespaced, and as such the relationship between a fourcc code and a
- * modifier is specific to the modifer being used. For example, some modifiers
+ * modifier is specific to the modifier being used. For example, some modifiers
  * may preserve meaning - such as number of planes - from the fourcc code,
  * whereas others may not.
  *
@@ -78,7 +78,7 @@ extern "C" {
  *   format.
  * - Higher-level programs interfacing with KMS/GBM/EGL/Vulkan/etc: these users
  *   see modifiers as opaque tokens they can check for equality and intersect.
- *   These users musn't need to know to reason about the modifier value
+ *   These users mustn't need to know to reason about the modifier value
  *   (i.e. they are not expected to extract information out of the modifier).
  *
  * Vendors should document their modifier usage in as much detail as
@@ -539,7 +539,7 @@ extern "C" {
  * This is a tiled layout using 4Kb tiles in row-major layout.
  * Within the tile pixels are laid out in 16 256 byte units / sub-tiles which
  * are arranged in four groups (two wide, two high) with column-major layout.
- * Each group therefore consits out of four 256 byte units, which are also laid
+ * Each group therefore consists out of four 256 byte units, which are also laid
  * out as 2x2 column-major.
  * 256 byte units are made out of four 64 byte blocks of pixels, producing
  * either a square block or a 2:1 unit.
@@ -1102,7 +1102,7 @@ drm_fourcc_canonicalize_nvidia_format_mod(uint64_t modifier)
  */
 
 /*
- * The top 4 bits (out of the 56 bits alloted for specifying vendor specific
+ * The top 4 bits (out of the 56 bits allotted for specifying vendor specific
  * modifiers) denote the category for modifiers. Currently we have three
  * categories of modifiers ie AFBC, MISC and AFRC. We can have a maximum of
  * sixteen different categories.
@@ -1418,7 +1418,7 @@ drm_fourcc_canonicalize_nvidia_format_mod(uint64_t modifier)
  * Amlogic FBC Memory Saving mode
  *
  * Indicates the storage is packed when pixel size is multiple of word
- * boudaries, i.e. 8bit should be stored in this mode to save allocation
+ * boundaries, i.e. 8bit should be stored in this mode to save allocation
  * memory.
  *
  * This mode reduces body layout to 3072 bytes per 64x32 superblock with
include/standard-headers/linux/ethtool.h +28 -13

71 lines not on this page · raw message

include/standard-headers/linux/virtio_config.h +7 -1
index bfd1ca643e7f..45be0fa1bcdb 100644
@@ -52,7 +52,7 @@
  * rest are per-device feature bits.
  */
 #define VIRTIO_TRANSPORT_F_START	28
-#define VIRTIO_TRANSPORT_F_END		41
+#define VIRTIO_TRANSPORT_F_END		42
 
 #ifndef VIRTIO_CONFIG_NO_LEGACY
 /* Do we get callbacks when the ring is completely used, even if we've
@@ -112,4 +112,10 @@
  * This feature indicates that the driver can reset a queue individually.
  */
 #define VIRTIO_F_RING_RESET		40
+
+/*
+ * This feature indicates that the device support administration virtqueues.
+ */
+#define VIRTIO_F_ADMIN_VQ		41
+
 #endif /* _LINUX_VIRTIO_CONFIG_H */
include/standard-headers/linux/virtio_pci.h +68 -0

81 lines not on this page · raw message

include/standard-headers/linux/virtio_pmem.h +7 -0
index fc029de7988e..1a2576d0178e 100644
@@ -14,6 +14,13 @@
 #include "standard-headers/linux/virtio_ids.h"
 #include "standard-headers/linux/virtio_config.h"
 
+/* Feature bits */
+/* guest physical address range will be indicated as shared memory region 0 */
+#define VIRTIO_PMEM_F_SHMEM_REGION 0
+
+/* shmid of the shared memory region corresponding to the pmem */
+#define VIRTIO_PMEM_SHMEM_REGION_ID 0
+
 struct virtio_pmem_config {
 	uint64_t start;
 	uint64_t size;
linux-headers/asm-generic/unistd.h +14 -1
index 756b013fb832..75f00965ab15 100644
@@ -829,8 +829,21 @@ __SYSCALL(__NR_futex_wait, sys_futex_wait)
 #define __NR_futex_requeue 456
 __SYSCALL(__NR_futex_requeue, sys_futex_requeue)
 
+#define __NR_statmount   457
+__SYSCALL(__NR_statmount, sys_statmount)
+
+#define __NR_listmount   458
+__SYSCALL(__NR_listmount, sys_listmount)
+
+#define __NR_lsm_get_self_attr 459
+__SYSCALL(__NR_lsm_get_self_attr, sys_lsm_get_self_attr)
+#define __NR_lsm_set_self_attr 460
+__SYSCALL(__NR_lsm_set_self_attr, sys_lsm_set_self_attr)
+#define __NR_lsm_list_modules 461
+__SYSCALL(__NR_lsm_list_modules, sys_lsm_list_modules)
+
 #undef __NR_syscalls
-#define __NR_syscalls 457
+#define __NR_syscalls 462
 
 /*
  * 32 bit systems traditionally used different
linux-headers/asm-loongarch/kvm.h +1 -0
index c6ad2ee6106c..923d0bd38294 100644
@@ -79,6 +79,7 @@ struct kvm_fpu {
 #define LOONGARCH_REG_64(TYPE, REG)	(TYPE | KVM_REG_SIZE_U64 | (REG << LOONGARCH_REG_SHIFT))
 #define KVM_IOC_CSRID(REG)		LOONGARCH_REG_64(KVM_REG_LOONGARCH_CSR, REG)
 #define KVM_IOC_CPUCFG(REG)		LOONGARCH_REG_64(KVM_REG_LOONGARCH_CPUCFG, REG)
+#define KVM_LOONGARCH_VCPU_CPUCFG	0
 
 struct kvm_debug_exit_arch {
 };
linux-headers/asm-mips/mman.h +1 -1
index c6e1fc77c996..9c48d9a21aa0 100644
@@ -88,7 +88,7 @@
 #define MADV_HUGEPAGE	14		/* Worth backing with hugepages */
 #define MADV_NOHUGEPAGE 15		/* Not worth backing with hugepages */
 
-#define MADV_DONTDUMP	16		/* Explicity exclude from the core dump,
+#define MADV_DONTDUMP	16		/* Explicitly exclude from core dump,
 					   overrides the coredump filter bits */
 #define MADV_DODUMP	17		/* Clear the MADV_NODUMP flag */
 
linux-headers/asm-mips/unistd_n32.h +5 -0
index 994b6f008f54..ce2e050a9ba4 100644
@@ -385,5 +385,10 @@
 #define __NR_futex_wake (__NR_Linux + 454)
 #define __NR_futex_wait (__NR_Linux + 455)
 #define __NR_futex_requeue (__NR_Linux + 456)
+#define __NR_statmount (__NR_Linux + 457)
+#define __NR_listmount (__NR_Linux + 458)
+#define __NR_lsm_get_self_attr (__NR_Linux + 459)
+#define __NR_lsm_set_self_attr (__NR_Linux + 460)
+#define __NR_lsm_list_modules (__NR_Linux + 461)
 
 #endif /* _ASM_UNISTD_N32_H */
linux-headers/asm-mips/unistd_n64.h +5 -0
index 41dcf5877a1b..5bfb3733ffdf 100644
@@ -361,5 +361,10 @@
 #define __NR_futex_wake (__NR_Linux + 454)
 #define __NR_futex_wait (__NR_Linux + 455)
 #define __NR_futex_requeue (__NR_Linux + 456)
+#define __NR_statmount (__NR_Linux + 457)
+#define __NR_listmount (__NR_Linux + 458)
+#define __NR_lsm_get_self_attr (__NR_Linux + 459)
+#define __NR_lsm_set_self_attr (__NR_Linux + 460)
+#define __NR_lsm_list_modules (__NR_Linux + 461)
 
 #endif /* _ASM_UNISTD_N64_H */
linux-headers/asm-mips/unistd_o32.h +5 -0
index ae9d334d96e3..02eaecd020ec 100644
@@ -431,5 +431,10 @@
 #define __NR_futex_wake (__NR_Linux + 454)
 #define __NR_futex_wait (__NR_Linux + 455)
 #define __NR_futex_requeue (__NR_Linux + 456)
+#define __NR_statmount (__NR_Linux + 457)
+#define __NR_listmount (__NR_Linux + 458)
+#define __NR_lsm_get_self_attr (__NR_Linux + 459)
+#define __NR_lsm_set_self_attr (__NR_Linux + 460)
+#define __NR_lsm_list_modules (__NR_Linux + 461)
 
 #endif /* _ASM_UNISTD_O32_H */
linux-headers/asm-powerpc/unistd_32.h +5 -0
index b9b23d66d7d9..bbab08d6ec26 100644
@@ -438,6 +438,11 @@
 #define __NR_futex_wake 454
 #define __NR_futex_wait 455
 #define __NR_futex_requeue 456
+#define __NR_statmount 457
+#define __NR_listmount 458
+#define __NR_lsm_get_self_attr 459
+#define __NR_lsm_set_self_attr 460
+#define __NR_lsm_list_modules 461
 
 
 #endif /* _ASM_UNISTD_32_H */
linux-headers/asm-powerpc/unistd_64.h +5 -0
index cbb4b3e8f7c2..af34cde70f20 100644
@@ -410,6 +410,11 @@
 #define __NR_futex_wake 454
 #define __NR_futex_wait 455
 #define __NR_futex_requeue 456
+#define __NR_statmount 457
+#define __NR_listmount 458
+#define __NR_lsm_get_self_attr 459
+#define __NR_lsm_set_self_attr 460
+#define __NR_lsm_list_modules 461
 
 
 #endif /* _ASM_UNISTD_64_H */
linux-headers/asm-riscv/kvm.h +13 -0
index 60d3b21dead7..d6b7a5b95874 100644
@@ -157,9 +157,16 @@ enum KVM_RISCV_SBI_EXT_ID {
 	KVM_RISCV_SBI_EXT_EXPERIMENTAL,
 	KVM_RISCV_SBI_EXT_VENDOR,
 	KVM_RISCV_SBI_EXT_DBCN,
+	KVM_RISCV_SBI_EXT_STA,
 	KVM_RISCV_SBI_EXT_MAX,
 };
 
+/* SBI STA extension registers for KVM_GET_ONE_REG and KVM_SET_ONE_REG */
+struct kvm_riscv_sbi_sta {
+	unsigned long shmem_lo;
+	unsigned long shmem_hi;
+};
+
 /* Possible states for kvm_riscv_timer */
 #define KVM_RISCV_TIMER_STATE_OFF	0
 #define KVM_RISCV_TIMER_STATE_ON	1
@@ -241,6 +248,12 @@ enum KVM_RISCV_SBI_EXT_ID {
 #define KVM_REG_RISCV_VECTOR_REG(n)	\
 		((n) + sizeof(struct __riscv_v_ext_state) / sizeof(unsigned long))
 
+/* Registers for specific SBI extensions are mapped as type 10 */
+#define KVM_REG_RISCV_SBI_STATE		(0x0a << KVM_REG_RISCV_TYPE_SHIFT)
+#define KVM_REG_RISCV_SBI_STA		(0x0 << KVM_REG_RISCV_SUBTYPE_SHIFT)
+#define KVM_REG_RISCV_SBI_STA_REG(name)		\
+		(offsetof(struct kvm_riscv_sbi_sta, name) / sizeof(unsigned long))
+
 /* Device Control API: RISC-V AIA */
 #define KVM_DEV_RISCV_APLIC_ALIGN		0x1000
 #define KVM_DEV_RISCV_APLIC_SIZE		0x4000
linux-headers/asm-s390/unistd_32.h +5 -0
index c093e6d5f911..a3ece69d8241 100644
@@ -429,5 +429,10 @@
 #define __NR_futex_wake 454
 #define __NR_futex_wait 455
 #define __NR_futex_requeue 456
+#define __NR_statmount 457
+#define __NR_listmount 458
+#define __NR_lsm_get_self_attr 459
+#define __NR_lsm_set_self_attr 460
+#define __NR_lsm_list_modules 461
 
 #endif /* _ASM_S390_UNISTD_32_H */
linux-headers/asm-s390/unistd_64.h +5 -0
index 114c0569a49a..8c5fd93495ce 100644
@@ -377,5 +377,10 @@
 #define __NR_futex_wake 454
 #define __NR_futex_wait 455
 #define __NR_futex_requeue 456
+#define __NR_statmount 457
+#define __NR_listmount 458
+#define __NR_lsm_get_self_attr 459
+#define __NR_lsm_set_self_attr 460
+#define __NR_lsm_list_modules 461
 
 #endif /* _ASM_S390_UNISTD_64_H */
linux-headers/asm-x86/kvm.h +3 -0
index 2b3a8f7bd2c0..003fb745347c 100644
@@ -560,4 +560,7 @@ struct kvm_pmu_event_filter {
 /* x86-specific KVM_EXIT_HYPERCALL flags. */
 #define KVM_EXIT_HYPERCALL_LONG_MODE	BIT(0)
 
+#define KVM_X86_DEFAULT_VM	0
+#define KVM_X86_SW_PROTECTED_VM	1
+
 #endif /* _ASM_X86_KVM_H */
linux-headers/asm-x86/unistd_32.h +5 -0
index 329649c377be..5c9c329e9390 100644
@@ -447,6 +447,11 @@
 #define __NR_futex_wake 454
 #define __NR_futex_wait 455
 #define __NR_futex_requeue 456
+#define __NR_statmount 457
+#define __NR_listmount 458
+#define __NR_lsm_get_self_attr 459
+#define __NR_lsm_set_self_attr 460
+#define __NR_lsm_list_modules 461
 
 
 #endif /* _ASM_UNISTD_32_H */
linux-headers/asm-x86/unistd_64.h +5 -0
index 4583606ce684..d9aab7ae87d8 100644
@@ -369,6 +369,11 @@
 #define __NR_futex_wake 454
 #define __NR_futex_wait 455
 #define __NR_futex_requeue 456
+#define __NR_statmount 457
+#define __NR_listmount 458
+#define __NR_lsm_get_self_attr 459
+#define __NR_lsm_set_self_attr 460
+#define __NR_lsm_list_modules 461
 
 
 #endif /* _ASM_UNISTD_64_H */
linux-headers/asm-x86/unistd_x32.h +5 -0
index 146d74d8e4b0..63cdd1ee43df 100644
@@ -321,6 +321,11 @@
 #define __NR_futex_wake (__X32_SYSCALL_BIT + 454)
 #define __NR_futex_wait (__X32_SYSCALL_BIT + 455)
 #define __NR_futex_requeue (__X32_SYSCALL_BIT + 456)
+#define __NR_statmount (__X32_SYSCALL_BIT + 457)
+#define __NR_listmount (__X32_SYSCALL_BIT + 458)
+#define __NR_lsm_get_self_attr (__X32_SYSCALL_BIT + 459)
+#define __NR_lsm_set_self_attr (__X32_SYSCALL_BIT + 460)
+#define __NR_lsm_list_modules (__X32_SYSCALL_BIT + 461)
 #define __NR_rt_sigaction (__X32_SYSCALL_BIT + 512)
 #define __NR_rt_sigreturn (__X32_SYSCALL_BIT + 513)
 #define __NR_ioctl (__X32_SYSCALL_BIT + 514)
linux-headers/linux/iommufd.h +79 -0

89 lines not on this page · raw message

linux-headers/linux/kvm.h +50 -90

218 lines not on this page · raw message

linux-headers/linux/userfaultfd.h +28 -1
index 953c75fedae9..4283de22d5b6 100644
@@ -41,7 +41,8 @@
 			   UFFD_FEATURE_WP_HUGETLBFS_SHMEM |	\
 			   UFFD_FEATURE_WP_UNPOPULATED |	\
 			   UFFD_FEATURE_POISON |		\
-			   UFFD_FEATURE_WP_ASYNC)
+			   UFFD_FEATURE_WP_ASYNC |		\
+			   UFFD_FEATURE_MOVE)
 #define UFFD_API_IOCTLS				\
 	((__u64)1 << _UFFDIO_REGISTER |		\
 	 (__u64)1 << _UFFDIO_UNREGISTER |	\
@@ -50,6 +51,7 @@
 	((__u64)1 << _UFFDIO_WAKE |		\
 	 (__u64)1 << _UFFDIO_COPY |		\
 	 (__u64)1 << _UFFDIO_ZEROPAGE |		\
+	 (__u64)1 << _UFFDIO_MOVE |		\
 	 (__u64)1 << _UFFDIO_WRITEPROTECT |	\
 	 (__u64)1 << _UFFDIO_CONTINUE |		\
 	 (__u64)1 << _UFFDIO_POISON)
@@ -73,6 +75,7 @@
 #define _UFFDIO_WAKE			(0x02)
 #define _UFFDIO_COPY			(0x03)
 #define _UFFDIO_ZEROPAGE		(0x04)
+#define _UFFDIO_MOVE			(0x05)
 #define _UFFDIO_WRITEPROTECT		(0x06)
 #define _UFFDIO_CONTINUE		(0x07)
 #define _UFFDIO_POISON			(0x08)
@@ -92,6 +95,8 @@
 				      struct uffdio_copy)
 #define UFFDIO_ZEROPAGE		_IOWR(UFFDIO, _UFFDIO_ZEROPAGE,	\
 				      struct uffdio_zeropage)
+#define UFFDIO_MOVE		_IOWR(UFFDIO, _UFFDIO_MOVE,	\
+				      struct uffdio_move)
 #define UFFDIO_WRITEPROTECT	_IOWR(UFFDIO, _UFFDIO_WRITEPROTECT, \
 				      struct uffdio_writeprotect)
 #define UFFDIO_CONTINUE		_IOWR(UFFDIO, _UFFDIO_CONTINUE,	\
@@ -222,6 +227,9 @@ struct uffdio_api {
 	 * asynchronous mode is supported in which the write fault is
 	 * automatically resolved and write-protection is un-set.
 	 * It implies UFFD_FEATURE_WP_UNPOPULATED.
+	 *
+	 * UFFD_FEATURE_MOVE indicates that the kernel supports moving an
+	 * existing page contents from userspace.
 	 */
 #define UFFD_FEATURE_PAGEFAULT_FLAG_WP		(1<<0)
 #define UFFD_FEATURE_EVENT_FORK			(1<<1)
@@ -239,6 +247,7 @@ struct uffdio_api {
 #define UFFD_FEATURE_WP_UNPOPULATED		(1<<13)
 #define UFFD_FEATURE_POISON			(1<<14)
 #define UFFD_FEATURE_WP_ASYNC			(1<<15)
+#define UFFD_FEATURE_MOVE			(1<<16)
 	__u64 features;
 
 	__u64 ioctls;
@@ -347,6 +356,24 @@ struct uffdio_poison {
 	__s64 updated;
 };
 
+struct uffdio_move {
+	__u64 dst;
+	__u64 src;
+	__u64 len;
+	/*
+	 * Especially if used to atomically remove memory from the
+	 * address space the wake on the dst range is not needed.
+	 */
+#define UFFDIO_MOVE_MODE_DONTWAKE		((__u64)1<<0)
+#define UFFDIO_MOVE_MODE_ALLOW_SRC_HOLES	((__u64)1<<1)
+	__u64 mode;
+	/*
+	 * "move" is written by the ioctl and must be at the end: the
+	 * copy_from_user will not read the last 8 bytes.
+	 */
+	__s64 move;
+};
+
 /*
  * Flags for the userfaultfd(2) system call itself.
  */
linux-headers/linux/vfio.h +1 -0
index 8e175ece31ce..b4be37b2255c 100644
@@ -1219,6 +1219,7 @@ enum vfio_device_mig_state {
 	VFIO_DEVICE_STATE_RUNNING_P2P = 5,
 	VFIO_DEVICE_STATE_PRE_COPY = 6,
 	VFIO_DEVICE_STATE_PRE_COPY_P2P = 7,
+	VFIO_DEVICE_STATE_NR,
 };
 
 /**
Signature · 1 line
2.34.1
author
#2 lore raw

PATCH v4 2/66 RAMBlock: Add support of KVM private guest memfd

In reply to Xiaoyao Li

Add KVM guest_memfd support to RAMBlock so both normal hva based memory and kvm guest memfd based private memory can be associated in one RAMBlock.

Introduce new flag RAM_GUEST_MEMFD. When it's set, it calls KVM ioctl to create private guest_memfd during RAMBlock setup.

Allocating a new RAM_GUEST_MEMFD flag to instruct the setup of guest memfd is more flexible and extensible than simply relying on the VM type because in the future we may have the case that not all the memory of a VM need guest memfd. As a benefit, it also avoid getting MachineState in memory subsystem.

Note, RAM_GUEST_MEMFD is supposed to be set for memory backends of confidential guests, such as TDX VM. How and when to set it for memory backends will be implemented in the following patches.

Introduce memory_region_has_guest_memfd() to query if the MemoryRegion has KVM guest_memfd allocated.

---
Changes in v4:
- Add clarification of RAM_GUEST_MEMFD in commit message; (David Hildenbrand)
- refine the return value and error message; (Daniel P. Berrangé)
- remove flags in ram_block_add(); (David Hildenbrand)

Changes in v3:
- rename gmem to guest_memfd;
- close(guest_memfd) when RAMBlock is released; (Daniel P. Berrangé)
- Suqash the patch that introduces memory_region_has_guest_memfd().

---
 accel/kvm/kvm-all.c     | 25 +++++++++++++++++++++++++
 accel/stubs/kvm-stub.c  |  5 +++++
 include/exec/memory.h   | 20 +++++++++++++++++---
 include/exec/ram_addr.h |  2 +-
 include/exec/ramblock.h |  1 +
 include/sysemu/kvm.h    |  2 ++
 system/memory.c         |  5 +++++
 system/physmem.c        | 24 +++++++++++++++++++++---
 8 files changed, 77 insertions(+), 7 deletions(-)
accel/kvm/kvm-all.c +25 -0
index 49e755ec4ad2..f6b0f8d2db1c 100644
@@ -91,6 +91,7 @@ bool kvm_msi_use_devid;
 static bool kvm_has_guest_debug;
 static int kvm_sstep_flags;
 static bool kvm_immediate_exit;
+static bool kvm_guest_memfd_supported;
 static hwaddr kvm_max_slot_size = ~0;
 
 static const KVMCapabilityInfo kvm_required_capabilites[] = {
@@ -2385,6 +2386,8 @@ static int kvm_init(MachineState *ms)
     }
     s->as = g_new0(struct KVMAs, s->nr_as);
 
+    kvm_guest_memfd_supported = kvm_check_extension(s, KVM_CAP_GUEST_MEMFD);
+
     if (object_property_find(OBJECT(current_machine), "kvm-type")) {
         g_autofree char *kvm_type = object_property_get_str(OBJECT(current_machine),
                                                             "kvm-type",
@@ -4084,3 +4087,25 @@ void query_stats_schemas_cb(StatsSchemaList **result, Error **errp)
         query_stats_schema_vcpu(first_cpu, &stats_args);
     }
 }
+
+int kvm_create_guest_memfd(uint64_t size, uint64_t flags, Error **errp)
+{
+    int fd;
+    struct kvm_create_guest_memfd guest_memfd = {
+        .size = size,
+        .flags = flags,
+    };
+
+    if (!kvm_guest_memfd_supported) {
+        error_setg(errp, "KVM doesn't support guest memfd\n");
+        return -1;
+    }
+
+    fd = kvm_vm_ioctl(kvm_state, KVM_CREATE_GUEST_MEMFD, &guest_memfd);
+    if (fd < 0) {
+        error_setg_errno(errp, errno, "Error creating kvm guest memfd");
+        return -1;
+    }
+
+    return fd;
+}
accel/stubs/kvm-stub.c +5 -0
index 1b37d9a302cc..06cf548a0329 100644
@@ -124,3 +124,8 @@ uint32_t kvm_dirty_ring_size(void)
 {
     return 0;
 }
+
+int kvm_create_guest_memfd(uint64_t size, uint64_t flags, Error **errp)
+{
+    return -ENOSYS;
+}
\ No newline at end of file
include/exec/memory.h +17 -3
index 177be23db709..f11036ead15e 100644
@@ -243,6 +243,9 @@ typedef struct IOMMUTLBEvent {
 /* RAM FD is opened read-only */
 #define RAM_READONLY_FD (1 << 11)
 
+/* RAM can be private that has kvm guest memfd backend */
+#define RAM_GUEST_MEMFD   (1 << 12)
+
 static inline void iommu_notifier_init(IOMMUNotifier *n, IOMMUNotify fn,
                                        IOMMUNotifierFlag flags,
                                        hwaddr start, hwaddr end,
@@ -1307,7 +1310,8 @@ bool memory_region_init_ram_nomigrate(MemoryRegion *mr,
  * @name: Region name, becomes part of RAMBlock name used in migration stream
  *        must be unique within any device
  * @size: size of the region.
- * @ram_flags: RamBlock flags. Supported flags: RAM_SHARED, RAM_NORESERVE.
+ * @ram_flags: RamBlock flags. Supported flags: RAM_SHARED, RAM_NORESERVE,
+ *             RAM_GUEST_MEMFD.
  * @errp: pointer to Error*, to store an error if it happens.
  *
  * Note that this function does not do anything to cause the data in the
@@ -1369,7 +1373,7 @@ bool memory_region_init_resizeable_ram(MemoryRegion *mr,
  *         (getpagesize()) will be used.
  * @ram_flags: RamBlock flags. Supported flags: RAM_SHARED, RAM_PMEM,
  *             RAM_NORESERVE, RAM_PROTECTED, RAM_NAMED_FILE, RAM_READONLY,
- *             RAM_READONLY_FD
+ *             RAM_READONLY_FD, RAM_GUEST_MEMFD
  * @path: the path in which to allocate the RAM.
  * @offset: offset within the file referenced by path
  * @errp: pointer to Error*, to store an error if it happens.
@@ -1399,7 +1403,7 @@ bool memory_region_init_ram_from_file(MemoryRegion *mr,
  * @size: size of the region.
  * @ram_flags: RamBlock flags. Supported flags: RAM_SHARED, RAM_PMEM,
  *             RAM_NORESERVE, RAM_PROTECTED, RAM_NAMED_FILE, RAM_READONLY,
- *             RAM_READONLY_FD
+ *             RAM_READONLY_FD, RAM_GUEST_MEMFD
  * @fd: the fd to mmap.
  * @offset: offset within the file referenced by fd
  * @errp: pointer to Error*, to store an error if it happens.
@@ -1722,6 +1726,16 @@ static inline bool memory_region_is_romd(MemoryRegion *mr)
  */
 bool memory_region_is_protected(MemoryRegion *mr);
 
+/**
+ * memory_region_has_guest_memfd: check whether a memory region has guest_memfd
+ *     associated
+ *
+ * Returns %true if a memory region's ram_block has valid guest_memfd assigned.
+ *
+ * @mr: the memory region being queried
+ */
+bool memory_region_has_guest_memfd(MemoryRegion *mr);
+
 /**
  * memory_region_get_iommu: check whether a memory region is an iommu
  *
include/exec/ram_addr.h +1 -1
index 90676093f5d5..4ebd9ded5e86 100644
@@ -109,7 +109,7 @@ long qemu_maxrampagesize(void);
  *  @mr: the memory region where the ram block is
  *  @ram_flags: RamBlock flags. Supported flags: RAM_SHARED, RAM_PMEM,
  *              RAM_NORESERVE, RAM_PROTECTED, RAM_NAMED_FILE, RAM_READONLY,
- *              RAM_READONLY_FD
+ *              RAM_READONLY_FD, RAM_GUEST_MEMFD
  *  @mem_path or @fd: specify the backing file or device
  *  @offset: Offset into target file
  *  @errp: pointer to Error*, to store an error if it happens
include/exec/ramblock.h +1 -0
index 3eb79723c6a8..03b3a3d40a6d 100644
@@ -41,6 +41,7 @@ struct RAMBlock {
     QLIST_HEAD(, RAMBlockNotifier) ramblock_notifiers;
     int fd;
     uint64_t fd_offset;
+    int guest_memfd;
     size_t page_size;
     /* dirty bitmap used during migration */
     unsigned long *bmap;
include/sysemu/kvm.h +2 -0
index d61487816421..fedc28c7d17f 100644
@@ -538,4 +538,6 @@ bool kvm_arch_cpu_check_are_resettable(void);
 bool kvm_dirty_ring_enabled(void);
 
 uint32_t kvm_dirty_ring_size(void);
+
+int kvm_create_guest_memfd(uint64_t size, uint64_t flags, Error **errp);
 #endif
system/memory.c +5 -0
index a229a79988fc..c756950c0c0f 100644
@@ -1850,6 +1850,11 @@ bool memory_region_is_protected(MemoryRegion *mr)
     return mr->ram && (mr->ram_block->flags & RAM_PROTECTED);
 }
 
+bool memory_region_has_guest_memfd(MemoryRegion *mr)
+{
+    return mr->ram_block && mr->ram_block->guest_memfd >= 0;
+}
+
 uint8_t memory_region_get_dirty_log_mask(MemoryRegion *mr)
 {
     uint8_t mask = mr->dirty_log_mask;
system/physmem.c +21 -3
index c1b22bac77c2..4735b0462ed9 100644
@@ -1841,6 +1841,17 @@ static void ram_block_add(RAMBlock *new_block, Error **errp)
         }
     }
 
+    if (kvm_enabled() && (new_block->flags & RAM_GUEST_MEMFD) &&
+        new_block->guest_memfd < 0) {
+        /* TODO: to decide if KVM_GUEST_MEMFD_ALLOW_HUGEPAGE is supported */
+        new_block->guest_memfd = kvm_create_guest_memfd(new_block->max_length,
+                                                        0, errp);
+        if (new_block->guest_memfd < 0) {
+            qemu_mutex_unlock_ramlist();
+            return;
+        }
+    }
+
     new_ram_size = MAX(old_ram_size,
               (new_block->offset + new_block->max_length) >> TARGET_PAGE_BITS);
     if (new_ram_size > old_ram_size) {
@@ -1903,7 +1914,7 @@ RAMBlock *qemu_ram_alloc_from_fd(ram_addr_t size, MemoryRegion *mr,
     /* Just support these ram flags by now. */
     assert((ram_flags & ~(RAM_SHARED | RAM_PMEM | RAM_NORESERVE |
                           RAM_PROTECTED | RAM_NAMED_FILE | RAM_READONLY |
-                          RAM_READONLY_FD)) == 0);
+                          RAM_READONLY_FD | RAM_GUEST_MEMFD)) == 0);
 
     if (xen_enabled()) {
         error_setg(errp, "-mem-path not supported with Xen");
@@ -1938,6 +1949,7 @@ RAMBlock *qemu_ram_alloc_from_fd(ram_addr_t size, MemoryRegion *mr,
     new_block->used_length = size;
     new_block->max_length = size;
     new_block->flags = ram_flags;
+    new_block->guest_memfd = -1;
     new_block->host = file_ram_alloc(new_block, size, fd, !file_size, offset,
                                      errp);
     if (!new_block->host) {
@@ -2016,7 +2028,7 @@ RAMBlock *qemu_ram_alloc_internal(ram_addr_t size, ram_addr_t max_size,
     Error *local_err = NULL;
 
     assert((ram_flags & ~(RAM_SHARED | RAM_RESIZEABLE | RAM_PREALLOC |
-                          RAM_NORESERVE)) == 0);
+                          RAM_NORESERVE| RAM_GUEST_MEMFD)) == 0);
     assert(!host ^ (ram_flags & RAM_PREALLOC));
 
     size = HOST_PAGE_ALIGN(size);
@@ -2028,6 +2040,7 @@ RAMBlock *qemu_ram_alloc_internal(ram_addr_t size, ram_addr_t max_size,
     new_block->max_length = max_size;
     assert(max_size >= size);
     new_block->fd = -1;
+    new_block->guest_memfd = -1;
     new_block->page_size = qemu_real_host_page_size();
     new_block->host = host;
     new_block->flags = ram_flags;
@@ -2050,7 +2063,7 @@ RAMBlock *qemu_ram_alloc_from_ptr(ram_addr_t size, void *host,
 RAMBlock *qemu_ram_alloc(ram_addr_t size, uint32_t ram_flags,
                          MemoryRegion *mr, Error **errp)
 {
-    assert((ram_flags & ~(RAM_SHARED | RAM_NORESERVE)) == 0);
+    assert((ram_flags & ~(RAM_SHARED | RAM_NORESERVE | RAM_GUEST_MEMFD)) == 0);
     return qemu_ram_alloc_internal(size, size, NULL, NULL, ram_flags, mr, errp);
 }
 
@@ -2078,6 +2091,11 @@ static void reclaim_ramblock(RAMBlock *block)
     } else {
         qemu_anon_ram_free(block->host, block->max_length);
     }
+
+    if (block->guest_memfd >= 0) {
+        close(block->guest_memfd);
+    }
+
     g_free(block);
 }
 
Signature · 1 line
2.34.1
Reviewed-by on v4 2/66
#3 lore raw

Re: [PATCH v4 02/66] RAMBlock: Add support of KVM private guest memfd

system/physmem.c static void ram_block_add(RAMBlock *new_block, Error **errp) · line 1845

Quoted 13 lines
  uint8_t memory_region_get_dirty_log_mask(MemoryRegion *mr)
  {
      uint8_t mask = mr->dirty_log_mask;
system/physmem.c +2 -0
index c1b22bac77c2..4735b0462ed9 100644
@@ -1841,6 +1841,17 @@ static void ram_block_add(RAMBlock *new_block, Error **errp)
          }
      }
 
+    if (kvm_enabled() && (new_block->flags & RAM_GUEST_MEMFD) &&
+        new_block->guest_memfd < 0) {

How could we have a guest_memfd already at this point? Smells more like an assert(new_block->guest_memfd < 0);

system/physmem.c static void ram_block_add(RAMBlock *new_block, Error **errp) · line 1846

Quoted 1 line
system/physmem.c +1 -0
+        /* TODO: to decide if KVM_GUEST_MEMFD_ALLOW_HUGEPAGE is supported */

I suggest dropping that completely. As long as it's not upstream, not even the name of that thing is stable.

system/physmem.c static void ram_block_add(RAMBlock *new_block, Error **errp) · line 1854

Quoted 8 lines
system/physmem.c +8 -0
+        new_block->guest_memfd = kvm_create_guest_memfd(new_block->max_length,
+                                                        0, errp);
+        if (new_block->guest_memfd < 0) {
+            qemu_mutex_unlock_ramlist();
+            return;
+        }
+    }
+

In general, LGTM. With the two nits above:

Signature · 3 lines
Cheers,

David / dhildenb
author
#4 lore raw

Re: [PATCH v4 02/66] RAMBlock: Add support of KVM private guest memfd

Quoted 7 lines · David Hildenbrand
Quoted 7 lines

  uint8_t memory_region_get_dirty_log_mask(MemoryRegion *mr) {
      uint8_t mask = mr->dirty_log_mask;

system/physmem.c +0 -0
index c1b22bac77c2..4735b0462ed9 100644
Quoted 8 lines

system/physmem.c static void ram_block_add(RAMBlock *new_block, · line 1845

Quoted 6 lines
system/physmem.c +2 -0
@@ -1841,6 +1841,17 @@ static void ram_block_add(RAMBlock *new_block, 
 Error **errp)
           }
       }
+    if (kvm_enabled() && (new_block->flags & RAM_GUEST_MEMFD) &&
+        new_block->guest_memfd < 0) {

How could we have a guest_memfd already at this point? Smells more like an assert(new_block->guest_memfd < 0);

you are right. I will change it to the assert()

Quoted 4 lines
Quoted 2 lines
system/physmem.c +1 -0
+        /* TODO: to decide if KVM_GUEST_MEMFD_ALLOW_HUGEPAGE is 

supported */

I suggest dropping that completely. As long as it's not upstream, not even the name of that thing is stable.

OK

Quoted 11 lines
Quoted 9 lines

+        new_block->guest_memfd = kvm_create_guest_memfd(new_block->max_length,
+                                                        0, errp); +        if (new_block->guest_memfd < 0) {
+            qemu_mutex_unlock_ramlist();
+            return;
+        }
+    }
+

Quoted 4 lines

In general, LGTM. With the two nits above:

Thanks!

author
#5 lore raw

PATCH v4 3/66 HostMem: Add mechanism to opt in kvm guest memfd via MachineState

In reply to Xiaoyao Li

Add a new member "guest_memfd" to memory backends. When it's set to true, it enables RAM_GUEST_MEMFD in ram_flags, thus private kvm guest_memfd will be allocated during RAMBlock allocation.

Memory backend's @guest_memfd is wired with @require_guest_memfd field of MachineState. It avoid looking up the machine in phymem.c.

MachineState::require_guest_memfd is supposed to be set by any VMs that requires KVM guest memfd as private memory, e.g., TDX VM.

---
Changes in v4:
- rename "require_guest_memfd" to "guest_memfd" in struct
HostMemoryBackend; (David Hildenbrand)

---
 backends/hostmem-file.c  | 1 +
 backends/hostmem-memfd.c | 1 +
 backends/hostmem-ram.c   | 1 +
 backends/hostmem.c       | 1 +
 hw/core/machine.c        | 5 +++++
 include/hw/boards.h      | 2 ++
 include/sysemu/hostmem.h | 1 +
 7 files changed, 12 insertions(+)
backends/hostmem-file.c +1 -0
index ac3e433cbddb..3c69db794607 100644
@@ -85,6 +85,7 @@ file_backend_memory_alloc(HostMemoryBackend *backend, Error **errp)
     ram_flags |= fb->readonly ? RAM_READONLY_FD : 0;
     ram_flags |= fb->rom == ON_OFF_AUTO_ON ? RAM_READONLY : 0;
     ram_flags |= backend->reserve ? 0 : RAM_NORESERVE;
+    ram_flags |= backend->guest_memfd ? RAM_GUEST_MEMFD : 0;
     ram_flags |= fb->is_pmem ? RAM_PMEM : 0;
     ram_flags |= RAM_NAMED_FILE;
     return memory_region_init_ram_from_file(&backend->mr, OBJECT(backend), name,
backends/hostmem-memfd.c +1 -0
index 3923ea9364d5..745ead0034dc 100644
@@ -55,6 +55,7 @@ memfd_backend_memory_alloc(HostMemoryBackend *backend, Error **errp)
     name = host_memory_backend_get_name(backend);
     ram_flags = backend->share ? RAM_SHARED : 0;
     ram_flags |= backend->reserve ? 0 : RAM_NORESERVE;
+    ram_flags |= backend->guest_memfd ? RAM_GUEST_MEMFD : 0;
     return memory_region_init_ram_from_fd(&backend->mr, OBJECT(backend), name,
                                           backend->size, ram_flags, fd, 0, errp);
 }
backends/hostmem-ram.c +1 -0
index d121249f0f45..f7d81af783a5 100644
@@ -30,6 +30,7 @@ ram_backend_memory_alloc(HostMemoryBackend *backend, Error **errp)
     name = host_memory_backend_get_name(backend);
     ram_flags = backend->share ? RAM_SHARED : 0;
     ram_flags |= backend->reserve ? 0 : RAM_NORESERVE;
+    ram_flags |= backend->guest_memfd ? RAM_GUEST_MEMFD : 0;
     return memory_region_init_ram_flags_nomigrate(&backend->mr, OBJECT(backend),
                                                   name, backend->size,
                                                   ram_flags, errp);
backends/hostmem.c +1 -0
index 30f69b2cb5e9..b1c39e8065e2 100644
@@ -276,6 +276,7 @@ static void host_memory_backend_init(Object *obj)
     /* TODO: convert access to globals to compat properties */
     backend->merge = machine_mem_merge(machine);
     backend->dump = machine_dump_guest_core(machine);
+    backend->guest_memfd = machine_require_guest_memfd(machine);
     backend->reserve = true;
     backend->prealloc_threads = machine->smp.cpus;
 }
hw/core/machine.c +5 -0
index fb5afdcae4cc..514a201dc570 100644
@@ -1192,6 +1192,11 @@ bool machine_mem_merge(MachineState *machine)
     return machine->mem_merge;
 }
 
+bool machine_require_guest_memfd(MachineState *machine)
+{
+    return machine->require_guest_memfd;
+}
+
 static char *cpu_slot_to_string(const CPUArchId *cpu)
 {
     GString *s = g_string_new(NULL);
include/hw/boards.h +2 -0
index bcfde8a84d10..2e6156ae96b8 100644
@@ -36,6 +36,7 @@ bool machine_usb(MachineState *machine);
 int machine_phandle_start(MachineState *machine);
 bool machine_dump_guest_core(MachineState *machine);
 bool machine_mem_merge(MachineState *machine);
+bool machine_require_guest_memfd(MachineState *machine);
 HotpluggableCPUList *machine_query_hotpluggable_cpus(MachineState *machine);
 void machine_set_cpu_numa_node(MachineState *machine,
                                const CpuInstanceProperties *props,
@@ -370,6 +371,7 @@ struct MachineState {
     char *dt_compatible;
     bool dump_guest_core;
     bool mem_merge;
+    bool require_guest_memfd;
     bool usb;
     bool usb_disabled;
     char *firmware;
include/sysemu/hostmem.h +1 -0
index 0e411aaa29e1..04b884bf42a6 100644
@@ -74,6 +74,7 @@ struct HostMemoryBackend {
     uint64_t size;
     bool merge, dump, use_canonical_path;
     bool prealloc, is_mapped, share, reserve;
+    bool guest_memfd;
     uint32_t prealloc_threads;
     ThreadContext *prealloc_context;
     DECLARE_BITMAP(host_nodes, MAX_NODES + 1);
Signature · 1 line
2.34.1
Reviewed-by on v4 3/66
#6 lore raw

Re: [PATCH v4 03/66] HostMem: Add mechanism to opt in kvm guest memfd via MachineState

Quoted 4 lines · Xiaoyao Li

Add a new member "guest_memfd" to memory backends. When it's set to true, it enables RAM_GUEST_MEMFD in ram_flags, thus private kvm guest_memfd will be allocated during RAMBlock allocation.

Quoted 8 lines

Memory backend's @guest_memfd is wired with @require_guest_memfd field of MachineState. It avoid looking up the machine in phymem.c.

MachineState::require_guest_memfd is supposed to be set by any VMs that requires KVM guest memfd as private memory, e.g., TDX VM.

---

Signature · 3 lines
Cheers,

David / dhildenb
author
#7 lore raw

PATCH v4 4/66 trace/kvm: Split address space and slot id in trace_kvm_set_user_memory()

In reply to Xiaoyao Li

The upper 16 bits of kvm_userspace_memory_region::slot are address space id. Parse it separately in trace_kvm_set_user_memory().

---
 accel/kvm/kvm-all.c    | 5 +++--
 accel/kvm/trace-events | 2 +-
 2 files changed, 4 insertions(+), 3 deletions(-)
accel/kvm/kvm-all.c +3 -2
index f6b0f8d2db1c..56b41a4ea8dc 100644
@@ -304,8 +304,9 @@ static int kvm_set_user_memory_region(KVMMemoryListener *kml, KVMSlot *slot, boo
     ret = kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
     slot->old_flags = mem.flags;
 err:
-    trace_kvm_set_user_memory(mem.slot, mem.flags, mem.guest_phys_addr,
-                              mem.memory_size, mem.userspace_addr, ret);
+    trace_kvm_set_user_memory(mem.slot >> 16, (uint16_t)mem.slot, mem.flags,
+                              mem.guest_phys_addr, mem.memory_size,
+                              mem.userspace_addr, ret);
     if (ret < 0) {
         error_report("%s: KVM_SET_USER_MEMORY_REGION failed, slot=%d,"
                      " start=0x%" PRIx64 ", size=0x%" PRIx64 ": %s",
accel/kvm/trace-events +1 -1
index a25902597b1b..9f599abc172c 100644
@@ -15,7 +15,7 @@ kvm_irqchip_update_msi_route(int virq) "Updating MSI route virq=%d"
 kvm_irqchip_release_virq(int virq) "virq %d"
 kvm_set_ioeventfd_mmio(int fd, uint64_t addr, uint32_t val, bool assign, uint32_t size, bool datamatch) "fd: %d @0x%" PRIx64 " val=0x%x assign: %d size: %d match: %d"
 kvm_set_ioeventfd_pio(int fd, uint16_t addr, uint32_t val, bool assign, uint32_t size, bool datamatch) "fd: %d @0x%x val=0x%x assign: %d size: %d match: %d"
-kvm_set_user_memory(uint32_t slot, uint32_t flags, uint64_t guest_phys_addr, uint64_t memory_size, uint64_t userspace_addr, int ret) "Slot#%d flags=0x%x gpa=0x%"PRIx64 " size=0x%"PRIx64 " ua=0x%"PRIx64 " ret=%d"
+kvm_set_user_memory(uint16_t as, uint16_t slot, uint32_t flags, uint64_t guest_phys_addr, uint64_t memory_size, uint64_t userspace_addr, int ret) "AddrSpace#%d Slot#%d flags=0x%x gpa=0x%"PRIx64 " size=0x%"PRIx64 " ua=0x%"PRIx64 " ret=%d"
 kvm_clear_dirty_log(uint32_t slot, uint64_t start, uint32_t size) "slot#%"PRId32" start 0x%"PRIx64" size 0x%"PRIx32
 kvm_resample_fd_notify(int gsi) "gsi %d"
 kvm_dirty_ring_full(int id) "vcpu %d"
Signature · 1 line
2.34.1
author
#8 lore raw

PATCH v4 5/66 kvm: Enable KVM_SET_USER_MEMORY_REGION2 for memslot

In reply to Xiaoyao Li

From: Chao Peng <[email protected]>

Switch to KVM_SET_USER_MEMORY_REGION2 when supported by KVM.

With KVM_SET_USER_MEMORY_REGION2, QEMU can set up memory region that backend'ed both by hva-based shared memory and guest memfd based private memory.

---
Changes in v4:
- update KVM_MEM_PRIVATE to KVM_MEM_GUEST_MEMFD; (Isaku)

---
 accel/kvm/kvm-all.c      | 56 ++++++++++++++++++++++++++++++++++------
 accel/kvm/trace-events   |  2 +-
 include/sysemu/kvm_int.h |  2 ++
 3 files changed, 51 insertions(+), 9 deletions(-)
accel/kvm/kvm-all.c +48 -8
index 56b41a4ea8dc..50c10becf772 100644
@@ -283,35 +283,69 @@ int kvm_physical_memory_addr_from_host(KVMState *s, void *ram,
 static int kvm_set_user_memory_region(KVMMemoryListener *kml, KVMSlot *slot, bool new)
 {
     KVMState *s = kvm_state;
-    struct kvm_userspace_memory_region mem;
+    struct kvm_userspace_memory_region2 mem;
+    static int cap_user_memory2 = -1;
     int ret;
 
+    if (cap_user_memory2 == -1) {
+        cap_user_memory2 = kvm_check_extension(s, KVM_CAP_USER_MEMORY2);
+    }
+
+    if (!cap_user_memory2 && slot->guest_memfd >= 0) {
+        error_report("%s, KVM doesn't support KVM_CAP_USER_MEMORY2,"
+                     " which is required by guest memfd!", __func__);
+        exit(1);
+    }
+
     mem.slot = slot->slot | (kml->as_id << 16);
     mem.guest_phys_addr = slot->start_addr;
     mem.userspace_addr = (unsigned long)slot->ram;
     mem.flags = slot->flags;
+    mem.guest_memfd = slot->guest_memfd;
+    mem.guest_memfd_offset = slot->guest_memfd_offset;
 
     if (slot->memory_size && !new && (mem.flags ^ slot->old_flags) & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;
-        ret = kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
+
+        if (cap_user_memory2) {
+            ret = kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION2, &mem);
+        } else {
+            ret = kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
+        }
         if (ret < 0) {
             goto err;
         }
     }
     mem.memory_size = slot->memory_size;
-    ret = kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
+    if (cap_user_memory2) {
+        ret = kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION2, &mem);
+    } else {
+        ret = kvm_vm_ioctl(s, KVM_SET_USER_MEMORY_REGION, &mem);
+    }
     slot->old_flags = mem.flags;
 err:
     trace_kvm_set_user_memory(mem.slot >> 16, (uint16_t)mem.slot, mem.flags,
                               mem.guest_phys_addr, mem.memory_size,
-                              mem.userspace_addr, ret);
+                              mem.userspace_addr, mem.guest_memfd,
+                              mem.guest_memfd_offset, ret);
     if (ret < 0) {
-        error_report("%s: KVM_SET_USER_MEMORY_REGION failed, slot=%d,"
-                     " start=0x%" PRIx64 ", size=0x%" PRIx64 ": %s",
-                     __func__, mem.slot, slot->start_addr,
-                     (uint64_t)mem.memory_size, strerror(errno));
+        if (cap_user_memory2) {
+                error_report("%s: KVM_SET_USER_MEMORY_REGION2 failed, slot=%d,"
+                        " start=0x%" PRIx64 ", size=0x%" PRIx64 ","
+                        " flags=0x%" PRIx32 ", guest_memfd=%" PRId32 ","
+                        " guest_memfd_offset=0x%" PRIx64 ": %s",
+                        __func__, mem.slot, slot->start_addr,
+                        (uint64_t)mem.memory_size, mem.flags,
+                        mem.guest_memfd, (uint64_t)mem.guest_memfd_offset,
+                        strerror(errno));
+        } else {
+                error_report("%s: KVM_SET_USER_MEMORY_REGION failed, slot=%d,"
+                            " start=0x%" PRIx64 ", size=0x%" PRIx64 ": %s",
+                            __func__, mem.slot, slot->start_addr,
+                            (uint64_t)mem.memory_size, strerror(errno));
+        }
     }
     return ret;
 }
@@ -466,6 +500,9 @@ static int kvm_mem_flags(MemoryRegion *mr)
     if (readonly && kvm_readonly_mem_allowed) {
         flags |= KVM_MEM_READONLY;
     }
+    if (memory_region_has_guest_memfd(mr)) {
+        flags |= KVM_MEM_GUEST_MEMFD;
+    }
     return flags;
 }
 
@@ -1353,6 +1390,9 @@ static void kvm_set_phys_mem(KVMMemoryListener *kml,
         mem->ram_start_offset = ram_start_offset;
         mem->ram = ram;
         mem->flags = kvm_mem_flags(mr);
+        mem->guest_memfd = mr->ram_block->guest_memfd;
+        mem->guest_memfd_offset = (uint8_t*)ram - mr->ram_block->host;
+
         kvm_slot_init_dirty_bitmap(mem);
         err = kvm_set_user_memory_region(kml, mem, true);
         if (err) {
accel/kvm/trace-events +1 -1
index 9f599abc172c..e8c52cb9e7a1 100644
@@ -15,7 +15,7 @@ kvm_irqchip_update_msi_route(int virq) "Updating MSI route virq=%d"
 kvm_irqchip_release_virq(int virq) "virq %d"
 kvm_set_ioeventfd_mmio(int fd, uint64_t addr, uint32_t val, bool assign, uint32_t size, bool datamatch) "fd: %d @0x%" PRIx64 " val=0x%x assign: %d size: %d match: %d"
 kvm_set_ioeventfd_pio(int fd, uint16_t addr, uint32_t val, bool assign, uint32_t size, bool datamatch) "fd: %d @0x%x val=0x%x assign: %d size: %d match: %d"
-kvm_set_user_memory(uint16_t as, uint16_t slot, uint32_t flags, uint64_t guest_phys_addr, uint64_t memory_size, uint64_t userspace_addr, int ret) "AddrSpace#%d Slot#%d flags=0x%x gpa=0x%"PRIx64 " size=0x%"PRIx64 " ua=0x%"PRIx64 " ret=%d"
+kvm_set_user_memory(uint16_t as, uint16_t slot, uint32_t flags, uint64_t guest_phys_addr, uint64_t memory_size, uint64_t userspace_addr, uint32_t fd, uint64_t fd_offset, int ret) "AddrSpace#%d Slot#%d flags=0x%x gpa=0x%"PRIx64 " size=0x%"PRIx64 " ua=0x%"PRIx64 " guest_memfd=%d" " guest_memfd_offset=0x%" PRIx64 " ret=%d"
 kvm_clear_dirty_log(uint32_t slot, uint64_t start, uint32_t size) "slot#%"PRId32" start 0x%"PRIx64" size 0x%"PRIx32
 kvm_resample_fd_notify(int gsi) "gsi %d"
 kvm_dirty_ring_full(int id) "vcpu %d"
include/sysemu/kvm_int.h +2 -0
index 882e37e12c5b..cad763e240c1 100644
@@ -30,6 +30,8 @@ typedef struct KVMSlot
     int as_id;
     /* Cache of the offset in ram address space */
     ram_addr_t ram_start_offset;
+    int guest_memfd;
+    hwaddr guest_memfd_offset;
 } KVMSlot;
 
 typedef struct KVMMemoryUpdate {
Signature · 1 line
2.34.1
author
#9 lore raw

PATCH v4 6/66 kvm: Introduce support for memory_attributes

In reply to Xiaoyao Li

Introduce the helper functions to set the attributes of a range of memory to private or shared.

This is necessary to notify KVM the private/shared attribute of each gpa range. KVM needs the information to decide the GPA needs to be mapped at hva-based shared memory or guest_memfd based private memory.

---
Changes in v4:
- move the check of kvm_supported_memory_attributes to the common kvm_set_memory_attributes(); (Wang Wei)
- change warn_report() to error_report() in kvm_set_memory_attributes() and drop the __func__; (Daniel)

---
 accel/kvm/kvm-all.c  | 44 ++++++++++++++++++++++++++++++++++++++++++++
 include/sysemu/kvm.h |  3 +++
 2 files changed, 47 insertions(+)
accel/kvm/kvm-all.c +44 -0
index 50c10becf772..3bc78a16b48d 100644
@@ -92,6 +92,7 @@ static bool kvm_has_guest_debug;
 static int kvm_sstep_flags;
 static bool kvm_immediate_exit;
 static bool kvm_guest_memfd_supported;
+static uint64_t kvm_supported_memory_attributes;
 static hwaddr kvm_max_slot_size = ~0;
 
 static const KVMCapabilityInfo kvm_required_capabilites[] = {
@@ -1294,6 +1295,46 @@ void kvm_set_max_memslot_size(hwaddr max_slot_size)
     kvm_max_slot_size = max_slot_size;
 }
 
+static int kvm_set_memory_attributes(hwaddr start, hwaddr size, uint64_t attr)
+{
+    struct kvm_memory_attributes attrs;
+    int r;
+
+    if (kvm_supported_memory_attributes == 0) {
+        error_report("No memory attribute supported by KVM\n");
+        return -EINVAL;
+    }
+
+    if ((attr & kvm_supported_memory_attributes) != attr) {
+        error_report("memory attribute 0x%lx not supported by KVM,"
+                     " supported bits are 0x%lx\n",
+                     attr, kvm_supported_memory_attributes);
+        return -EINVAL;
+    }
+
+    attrs.attributes = attr;
+    attrs.address = start;
+    attrs.size = size;
+    attrs.flags = 0;
+
+    r = kvm_vm_ioctl(kvm_state, KVM_SET_MEMORY_ATTRIBUTES, &attrs);
+    if (r) {
+        error_report("failed to set memory (0x%lx+%#zx) with attr 0x%lx error '%s'",
+                     start, size, attr, strerror(errno));
+    }
+    return r;
+}
+
+int kvm_set_memory_attributes_private(hwaddr start, hwaddr size)
+{
+    return kvm_set_memory_attributes(start, size, KVM_MEMORY_ATTRIBUTE_PRIVATE);
+}
+
+int kvm_set_memory_attributes_shared(hwaddr start, hwaddr size)
+{
+    return kvm_set_memory_attributes(start, size, 0);
+}
+
 /* Called with KVMMemoryListener.slots_lock held */
 static void kvm_set_phys_mem(KVMMemoryListener *kml,
                              MemoryRegionSection *section, bool add)
@@ -2429,6 +2470,9 @@ static int kvm_init(MachineState *ms)
 
     kvm_guest_memfd_supported = kvm_check_extension(s, KVM_CAP_GUEST_MEMFD);
 
+    ret = kvm_check_extension(s, KVM_CAP_MEMORY_ATTRIBUTES);
+    kvm_supported_memory_attributes = ret > 0 ? ret : 0;
+
     if (object_property_find(OBJECT(current_machine), "kvm-type")) {
         g_autofree char *kvm_type = object_property_get_str(OBJECT(current_machine),
                                                             "kvm-type",
include/sysemu/kvm.h +3 -0
index fedc28c7d17f..0e88958190a4 100644
@@ -540,4 +540,7 @@ bool kvm_dirty_ring_enabled(void);
 uint32_t kvm_dirty_ring_size(void);
 
 int kvm_create_guest_memfd(uint64_t size, uint64_t flags, Error **errp);
+
+int kvm_set_memory_attributes_private(hwaddr start, hwaddr size);
+int kvm_set_memory_attributes_shared(hwaddr start, hwaddr size);
 #endif
Signature · 1 line
2.34.1
author
#10 lore raw

PATCH v4 7/66 physmem: Introduce ram_block_discard_guest_memfd_range()

In reply to Xiaoyao Li

When memory page is converted from private to shared, the original private memory is back'ed by guest_memfd. Introduce ram_block_discard_guest_memfd_range() for discarding memory in guest_memfd.

Originally-from: Isaku Yamahata <[email protected]>

---
Changes in in v4:
- Drop ram_block_convert_range() and open code its implementation in the next Patch.

---
 include/exec/cpu-common.h |  2 ++
 system/physmem.c          | 23 +++++++++++++++++++++++
 2 files changed, 25 insertions(+)
include/exec/cpu-common.h +2 -0
index fef3138d29fc..05610efa8b4f 100644
@@ -175,6 +175,8 @@ typedef int (RAMBlockIterFunc)(RAMBlock *rb, void *opaque);
 
 int qemu_ram_foreach_block(RAMBlockIterFunc func, void *opaque);
 int ram_block_discard_range(RAMBlock *rb, uint64_t start, size_t length);
+int ram_block_discard_guest_memfd_range(RAMBlock *rb, uint64_t start,
+                                        size_t length);
 
 #endif
 
system/physmem.c +23 -0
index 4735b0462ed9..fc59470191ef 100644
@@ -3626,6 +3626,29 @@ err:
     return ret;
 }
 
+int ram_block_discard_guest_memfd_range(RAMBlock *rb, uint64_t start,
+                                        size_t length)
+{
+    int ret = -1;
+
+#ifdef CONFIG_FALLOCATE_PUNCH_HOLE
+    ret = fallocate(rb->guest_memfd, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
+                    start, length);
+
+    if (ret) {
+        ret = -errno;
+        error_report("%s: Failed to fallocate %s:%" PRIx64 " +%zx (%d)",
+                     __func__, rb->idstr, start, length, ret);
+    }
+#else
+    ret = -ENOSYS;
+    error_report("%s: fallocate not available %s:%" PRIx64 " +%zx (%d)",
+                 __func__, rb->idstr, start, length, ret);
+#endif
+
+    return ret;
+}
+
 bool ramblock_is_pmem(RAMBlock *rb)
 {
     return rb->flags & RAM_PMEM;
Signature · 1 line
2.34.1
Reviewed-by on v4 7/66
#11 lore raw

Re: [PATCH v4 07/66] physmem: Introduce ram_block_discard_guest_memfd_range()

Quoted 54 lines · Xiaoyao Li

When memory page is converted from private to shared, the original private memory is back'ed by guest_memfd. Introduce ram_block_discard_guest_memfd_range() for discarding memory in guest_memfd.

Originally-from: Isaku Yamahata <[email protected]>

---
Changes in in v4:
- Drop ram_block_convert_range() and open code its implementation in the next Patch.

---
  include/exec/cpu-common.h |  2 ++
  system/physmem.c          | 23 +++++++++++++++++++++++
  2 files changed, 25 insertions(+)
include/exec/cpu-common.h +2 -0
index fef3138d29fc..05610efa8b4f 100644
@@ -175,6 +175,8 @@ typedef int (RAMBlockIterFunc)(RAMBlock *rb, void *opaque);
 
  int qemu_ram_foreach_block(RAMBlockIterFunc func, void *opaque);
  int ram_block_discard_range(RAMBlock *rb, uint64_t start, size_t length);
+int ram_block_discard_guest_memfd_range(RAMBlock *rb, uint64_t start,
+                                        size_t length);
 
  #endif
 
system/physmem.c +16 -0
index 4735b0462ed9..fc59470191ef 100644
@@ -3626,6 +3626,29 @@ err:
      return ret;
  }
 
+int ram_block_discard_guest_memfd_range(RAMBlock *rb, uint64_t start,
+                                        size_t length)
+{
+    int ret = -1;
+
+#ifdef CONFIG_FALLOCATE_PUNCH_HOLE
+    ret = fallocate(rb->guest_memfd, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
+                    start, length);
+
+    if (ret) {
+        ret = -errno;
+        error_report("%s: Failed to fallocate %s:%" PRIx64 " +%zx (%d)",
+                     __func__, rb->idstr, start, length, ret);
+    }
+#else
+    ret = -ENOSYS;

system/physmem.c err: · line 3654

Quoted 10 lines
system/physmem.c +7 -0
+    error_report("%s: fallocate not available %s:%" PRIx64 " +%zx (%d)",
+                 __func__, rb->idstr, start, length, ret);
+#endif
+
+    return ret;
+}
+
  bool ramblock_is_pmem(RAMBlock *rb)
  {
      return rb->flags & RAM_PMEM;
Signature · 3 lines
Cheers,

David / dhildenb
author
#12 lore raw

PATCH v4 8/66 kvm: handle KVM_EXIT_MEMORY_FAULT

In reply to Xiaoyao Li

From: Chao Peng <[email protected]>

When geeting KVM_EXIT_MEMORY_FAULT exit, it indicates userspace needs to do the memory conversion on the RAMBlock to turn the memory into desired attribute, i.e., private/shared.

Currently only KVM_MEMORY_EXIT_FLAG_PRIVATE in flags is valid when KVM_EXIT_MEMORY_FAULT happens.

Note, KVM_EXIT_MEMORY_FAULT makes sense only when the RAMBlock has guest_memfd memory backend.

Note, KVM_EXIT_MEMORY_FAULT returns with -EFAULT, so special handling is added.

When page is converted from shared to private, the original shared memory can be discarded via ram_block_discard_range(). Note, shared memory can be discarded only when it's not back'ed by hugetlb because hugetlb is supposed to be pre-allocated and no need for discarding.

---
Changes in v4:
- open-coded ram_block_discard logic;
- change warn_report() to error_report(); (Daniel)

---
 accel/kvm/kvm-all.c | 94 ++++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 84 insertions(+), 10 deletions(-)
accel/kvm/kvm-all.c +84 -10
index 3bc78a16b48d..2421b8a5b314 100644
@@ -2893,6 +2893,68 @@ static void kvm_eat_signals(CPUState *cpu)
     } while (sigismember(&chkset, SIG_IPI));
 }
 
+static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private)
+{
+    MemoryRegionSection section;
+    ram_addr_t offset;
+    MemoryRegion *mr;
+    RAMBlock *rb;
+    void *addr;
+    int ret = -1;
+
+    if (!QEMU_PTR_IS_ALIGNED(start, qemu_host_page_size) ||
+        !QEMU_PTR_IS_ALIGNED(size, qemu_host_page_size)) {
+        return -1;
+    }
+
+    if (!size) {
+        return -1;
+    }
+
+    section = memory_region_find(get_system_memory(), start, size);
+    mr = section.mr;
+    if (!mr) {
+        return -1;
+    }
+
+    if (memory_region_has_guest_memfd(mr)) {
+        if (to_private) {
+            ret = kvm_set_memory_attributes_private(start, size);
+        } else {
+            ret = kvm_set_memory_attributes_shared(start, size);
+        }
+
+        if (ret) {
+            memory_region_unref(section.mr);
+            return ret;
+        }
+
+        addr = memory_region_get_ram_ptr(mr) + section.offset_within_region;
+        rb = qemu_ram_block_from_host(addr, false, &offset);
+
+        if (to_private) {
+            if (rb->page_size != qemu_host_page_size) {
+                /*
+                * shared memory is back'ed by  hugetlb, which is supposed to be
+                * pre-allocated and doesn't need to be discarded
+                */
+                return 0;
+            } else {
+                ret = ram_block_discard_range(rb, offset, size);
+            }
+        } else {
+            ret = ram_block_discard_guest_memfd_range(rb, offset, size);
+        }
+    } else {
+        error_report("Convert non guest_memfd backed memory region "
+                    "(0x%"HWADDR_PRIx" ,+ 0x%"HWADDR_PRIx") to %s",
+                    start, size, to_private ? "private" : "shared");
+    }
+
+    memory_region_unref(section.mr);
+    return ret;
+}
+
 int kvm_cpu_exec(CPUState *cpu)
 {
     struct kvm_run *run = cpu->kvm_run;
@@ -2960,18 +3022,20 @@ int kvm_cpu_exec(CPUState *cpu)
                 ret = EXCP_INTERRUPT;
                 break;
             }
-            fprintf(stderr, "error: kvm run failed %s\n",
-                    strerror(-run_ret));
+            if (!(run_ret == -EFAULT && run->exit_reason == KVM_EXIT_MEMORY_FAULT)) {
+                fprintf(stderr, "error: kvm run failed %s\n",
+                        strerror(-run_ret));
 #ifdef TARGET_PPC
-            if (run_ret == -EBUSY) {
-                fprintf(stderr,
-                        "This is probably because your SMT is enabled.\n"
-                        "VCPU can only run on primary threads with all "
-                        "secondary threads offline.\n");
-            }
+                if (run_ret == -EBUSY) {
+                    fprintf(stderr,
+                            "This is probably because your SMT is enabled.\n"
+                            "VCPU can only run on primary threads with all "
+                            "secondary threads offline.\n");
+                }
 #endif
-            ret = -1;
-            break;
+                ret = -1;
+                break;
+            }
         }
 
         trace_kvm_run_exit(cpu->cpu_index, run->exit_reason);
@@ -3054,6 +3118,16 @@ int kvm_cpu_exec(CPUState *cpu)
                 break;
             }
             break;
+        case KVM_EXIT_MEMORY_FAULT:
+            if (run->memory_fault.flags & ~KVM_MEMORY_EXIT_FLAG_PRIVATE) {
+                error_report("KVM_EXIT_MEMORY_FAULT: Unknown flag 0x%" PRIx64,
+                             (uint64_t)run->memory_fault.flags);
+                ret = -1;
+                break;
+            }
+            ret = kvm_convert_memory(run->memory_fault.gpa, run->memory_fault.size,
+                                     run->memory_fault.flags & KVM_MEMORY_EXIT_FLAG_PRIVATE);
+            break;
         default:
             ret = kvm_arch_handle_exit(cpu, run);
             break;
Signature · 1 line
2.34.1
author
#13 lore raw

PATCH v4 9/66 trace/kvm: Add trace for page convertion between shared and private

In reply to Xiaoyao Li

From: Isaku Yamahata <[email protected]>

---
 accel/kvm/kvm-all.c    | 2 ++
 accel/kvm/trace-events | 1 +
 2 files changed, 3 insertions(+)
accel/kvm/kvm-all.c +2 -0
index 2421b8a5b314..def2b46b9de2 100644
@@ -2902,6 +2902,8 @@ static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private)
     void *addr;
     int ret = -1;
 
+    trace_kvm_convert_memory(start, size, to_private ? "shared_to_private" : "private_to_shared");
+
     if (!QEMU_PTR_IS_ALIGNED(start, qemu_host_page_size) ||
         !QEMU_PTR_IS_ALIGNED(size, qemu_host_page_size)) {
         return -1;
accel/kvm/trace-events +1 -0
index e8c52cb9e7a1..31175fed97ce 100644
@@ -31,3 +31,4 @@ kvm_cpu_exec(void) ""
 kvm_interrupt_exit_request(void) ""
 kvm_io_window_exit(void) ""
 kvm_run_exit_system_event(int cpu_index, uint32_t event_type) "cpu_index %d, system_even_type %"PRIu32
+kvm_convert_memory(uint64_t start, uint64_t size, const char *msg) "start 0x%" PRIx64 " size 0x%" PRIx64 " %s"
Signature · 1 line
2.34.1
author
#14 lore raw

PATCH v4 10/66 *** HACK *** linux-headers: Update headers to pull in TDX API changes

In reply to Xiaoyao Li

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 | 91 +++++++++++++++++++++++++++++++++++++
 linux-headers/linux/kvm.h   | 89 ++++++++++++++++++++++++++++++++++++
 2 files changed, 180 insertions(+)
linux-headers/asm-x86/kvm.h +91 -0
index 003fb745347c..cf708ea9472e 100644
@@ -562,5 +562,96 @@ 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_RELEASE_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 +89 -0

107 lines not on this page · raw message

Signature · 1 line
2.34.1
author
#15 lore raw

PATCH v4 11/66 i386: Introduce tdx-guest object

In reply to Xiaoyao Li

Introduce 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-guest,id=tdx0

It has only one member 'attributes' with fixed value 0 and not configurable so far.

---
Changes in v4:
- update the new qapi since filed from 8.2 to 9.0

Changes in v1
- 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
index 598c6646dfc0..9b5ec59d65b0 100644
@@ -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
index a1846be6f761..c0ccf50ac3ef 100644
@@ -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
index 95516ba325e5..5b3c3146947f 100644
@@ -895,6 +895,16 @@
             'reduced-phys-bits': 'uint32',
             '*kernel-hashes': 'bool' } }
 
+##
+# @TdxGuestProperties:
+#
+# Properties for tdx-guest objects.
+#
+# Since: 9.0
+##
+{ 'struct': 'TdxGuestProperties',
+  'data': { }}
+
 ##
 # @ThreadContextProperties:
 #
@@ -974,6 +984,7 @@
     'sev-guest',
     'thread-context',
     's390-pv-guest',
+    'tdx-guest',
     'throttle-group',
     'tls-creds-anon',
     'tls-creds-psk',
@@ -1041,6 +1052,7 @@
       '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
index 84d9143e6029..6ea0ce27b757 100644
@@ -9,6 +9,8 @@ i386_kvm_ss.add(when: 'CONFIG_XEN_EMU', if_true: files('xen-emu.c'))
 
 i386_kvm_ss.add(when: 'CONFIG_SEV', if_false: files('sev-stub.c'))
 
+i386_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_kvm_ss)
target/i386/kvm/tdx.c new file +40 -0
new file mode 100644
index 000000000000..d3792d4a3d56
@@ -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
new file mode 100644
index 000000000000..415aeb5af746
@@ -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
#16 lore raw

Re: [PATCH v4 11/66] i386: Introduce tdx-guest object

Quoted 8 lines · Xiaoyao Li

Introduce 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-guest,id=tdx0

It has only one member 'attributes' with fixed value 0 and not configurable so far.

Really? Can't see it.

Suggest to add something like "Configuration properties will be added later in this series."

Quoted 3 lines

[...]

Quoted 37 lines
qapi/qom.json +12 -0
index 95516ba325e5..5b3c3146947f 100644
@@ -895,6 +895,16 @@
             'reduced-phys-bits': 'uint32',
             '*kernel-hashes': 'bool' } }
 
+##
+# @TdxGuestProperties:
+#
+# Properties for tdx-guest objects.
+#
+# Since: 9.0
+##
+{ 'struct': 'TdxGuestProperties',
+  'data': { }}
+
 ##
 # @ThreadContextProperties:
 #
@@ -974,6 +984,7 @@
     'sev-guest',
     'thread-context',
     's390-pv-guest',
+    'tdx-guest',
     'throttle-group',
     'tls-creds-anon',
     'tls-creds-psk',
@@ -1041,6 +1052,7 @@
       'secret_keyring':             { 'type': 'SecretKeyringProperties',
                                       'if': 'CONFIG_SECRET_KEYRING' },
       'sev-guest':                  'SevGuestProperties',
+      'tdx-guest':                  'TdxGuestProperties',
       'thread-context':             'ThreadContextProperties',
       'throttle-group':             'ThrottleGroupProperties',
       'tls-creds-anon':             'TlsCredsAnonProperties',

[...]

#17 lore raw

Re: [PATCH v4 11/66] i386: Introduce tdx-guest object

Quoted 5 lines · Markus Armbruster
Quoted 3 lines · Xiaoyao Li

Introduce tdx-guest object which implements the interface of CONFIDENTIAL_GUEST_SUPPORT, and will be used to create TDX VMs (TDs) by

Quoted 7 lines
Quoted 5 lines
  qemu -machine ...,confidential-guest-support=tdx0	\
       -object tdx-guest,id=tdx0

It has only one member 'attributes' with fixed value 0 and not configurable so far.

Really? Can't see it.

The 'attributes' referred to is an internal struct field, rather than a QAPI declared member.

Quoted 41 lines

Suggest to add something like "Configuration properties will be added later in this series."

Quoted 3 lines

[...]

Quoted 31 lines
qapi/qom.json +11 -0
index 95516ba325e5..5b3c3146947f 100644
@@ -895,6 +895,16 @@
             'reduced-phys-bits': 'uint32',
             '*kernel-hashes': 'bool' } }
 
+##
+# @TdxGuestProperties:
+#
+# Properties for tdx-guest objects.
+#
+# Since: 9.0
+##
+{ 'struct': 'TdxGuestProperties',
+  'data': { }}
+
 ##
 # @ThreadContextProperties:
 #
@@ -974,6 +984,7 @@
     'sev-guest',
     'thread-context',
     's390-pv-guest',
+    'tdx-guest',
     'throttle-group',
     'tls-creds-anon',
     'tls-creds-psk',
@@ -1041,6 +1052,7 @@
       'secret_keyring':             { 'type': 'SecretKeyringProperties',
Quoted 9 lines
Quoted 6 lines
qapi/qom.json +1 -0
                                       'if': 'CONFIG_SECRET_KEYRING' },
       'sev-guest':                  'SevGuestProperties',
+      'tdx-guest':                  'TdxGuestProperties',
       'thread-context':             'ThreadContextProperties',
       'throttle-group':             'ThrottleGroupProperties',
       'tls-creds-anon':             'TlsCredsAnonProperties',

[...]

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 :|
author
#18 lore raw

PATCH v4 12/66 target/i386: Implement mc->kvm_type() to get VM type

In reply to Xiaoyao Li

TDX VM requires VM type KVM_X86_TDX_VM to be passed to kvm_ioctl(KVM_CREATE_VM). Hence implement mc->kvm_type() for i386 architecture.

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. Otherwise, it's KVM_X86_DEFAULT_VM.

Also store the vm_type in MachineState for other code to query what the VM type is.

---
Changes in v4:
- fix the build error of kvm_get_vm_type() when --disable-kvm;

---
 hw/i386/x86.c              | 12 ++++++++++++
 include/hw/i386/x86.h      |  1 +
 target/i386/kvm/kvm.c      | 30 ++++++++++++++++++++++++++++++
 target/i386/kvm/kvm_i386.h |  1 +
 4 files changed, 44 insertions(+)
hw/i386/x86.c +12 -0
index 2b6291ad8d5f..f66a92f6c9cc 100644
@@ -1386,6 +1386,17 @@ static void machine_set_sgx_epc(Object *obj, Visitor *v, const char *name,
     qapi_free_SgxEPCList(list);
 }
 
+static int x86_kvm_type(MachineState *ms, const char *vm_type)
+{
+    X86MachineState *x86ms = X86_MACHINE(ms);
+    int kvm_type;
+
+    kvm_type = kvm_enabled() ? kvm_get_vm_type(ms, vm_type) : 0;
+    x86ms->vm_type = kvm_type;
+
+    return kvm_type;
+}
+
 static void x86_machine_initfn(Object *obj)
 {
     X86MachineState *x86ms = X86_MACHINE(obj);
@@ -1410,6 +1421,7 @@ static void x86_machine_class_init(ObjectClass *oc, void *data)
     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
+    mc->kvm_type = x86_kvm_type;
     x86mc->save_tsc_khz = true;
     x86mc->fwcfg_dma_enabled = true;
     nc->nmi_monitor_handler = x86_nmi;
include/hw/i386/x86.h +1 -0
index da19ae15463a..ab1d38569019 100644
@@ -41,6 +41,7 @@ struct X86MachineState {
     MachineState parent;
 
     /*< public >*/
+    unsigned int vm_type;
 
     /* Pointers to devices and objects: */
     ISADevice *rtc;
target/i386/kvm/kvm.c +30 -0
index 42970ab046fa..c961846777cc 100644
@@ -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"
@@ -161,6 +162,35 @@ static KVMMSRHandlers msr_handlers[KVM_MSR_FILTER_MAX_RANGES];
 static RateLimit bus_lock_ratelimit_ctrl;
 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_TDX_VM] = "tdx",
+};
+
+int kvm_get_vm_type(MachineState *ms, const char *vm_type)
+{
+    int kvm_type = KVM_X86_DEFAULT_VM;
+
+    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
+     */
+    if (kvm_type == KVM_X86_DEFAULT_VM) {
+        return kvm_type;
+    }
+
+    if (!(kvm_check_extension(KVM_STATE(ms->accelerator), KVM_CAP_VM_TYPES) & BIT(kvm_type))) {
+        error_report("vm-type %s not supported by KVM", vm_type_name[kvm_type]);
+        exit(1);
+    }
+
+    return kvm_type;
+}
+
 bool kvm_has_smm(void)
 {
     return kvm_vm_check_extension(kvm_state, KVM_CAP_X86_SMM);
target/i386/kvm/kvm_i386.h +1 -0
index 30fedcffea3e..55fb25fa8e2e 100644
@@ -37,6 +37,7 @@ bool kvm_hv_vpindex_settable(void);
 bool kvm_enable_sgx_provisioning(KVMState *s);
 bool kvm_hyperv_expand_features(X86CPU *cpu, Error **errp);
 
+int kvm_get_vm_type(MachineState *ms, const char *vm_type);
 void kvm_arch_reset_vcpu(X86CPU *cs);
 void kvm_arch_after_reset_vcpu(X86CPU *cpu);
 void kvm_arch_do_init_vcpu(X86CPU *cs);
Signature · 1 line
2.34.1
author
#19 lore raw

PATCH v4 13/66 target/i386: Introduce kvm_confidential_guest_init()

In reply to Xiaoyao Li

Introduce a separate function kvm_confidential_guest_init(), which dispatches specific confidential guest initialization function by ms->cgs type.

---
 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
index c961846777cc..f9a774925cf6 100644
@@ -2541,6 +2541,15 @@ int kvm_arch_get_default_type(MachineState *ms)
     return 0;
 }
 
+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)
 {
     uint64_t identity_base = 0xfffbc000;
@@ -2561,7 +2570,7 @@ 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
index 173de91afe7d..27d58702d6dc 100644
@@ -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
index e7499c95b1e8..1fe25d096dc4 100644
@@ -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
#20 lore raw

Re: [PATCH v4 13/66] target/i386: Introduce kvm_confidential_guest_init()

Quoted 19 lines · Xiaoyao Li

Introduce a separate function kvm_confidential_guest_init(), which dispatches specific confidential guest initialization function by ms->cgs type.

---
 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 +0 -0
index c961846777cc..f9a774925cf6 100644
@@ -2541,6 +2541,15 @@ int kvm_arch_get_default_type(MachineState *ms)
     return 0;

target/i386/kvm/kvm.c int kvm_arch_get_default_type(MachineState *ms) · line 2551

Quoted 10 lines
target/i386/kvm/kvm.c +8 -0
 }
 
+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;
+}

if/else ladders checking object subclass type and then invoking a subclass specific method are quite an object oriented code anti-pattern. I think this suggests that ConfidentialGuestSupportClass should gain a member

   void (*kvm_init)(ConfidentialGuestSpport *cgs, Error **errp);

and then an impl

   void confidential_guest_kvm_init(ConfidentialGuestSupport *cgs, Error *errp) {
       ConfidentialGuestSupportClass *klass = CONFIDENTIAL_GUEST_SUPPORT_GET_CLASS(cgs);
       klass->kvm_init(cgs, errp)
   }

that way, kvm.c arch code doesn't get directly bound to the APIs for the different CVM technologies, the ConfidentialGuestSupport object will act as a proper isolation layer.

This is likely to apply in other parts of KVM code that need to call into SEV/TDX specific functions too.

Quoted 41 lines
target/i386/kvm/kvm.c +2 -1
+
 int kvm_arch_init(MachineState *ms, KVMState *s)
 {
     uint64_t identity_base = 0xfffbc000;
@@ -2561,7 +2570,7 @@ 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
index 173de91afe7d..27d58702d6dc 100644
@@ -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
index e7499c95b1e8..1fe25d096dc4 100644
@@ -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

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 :|
author
#21 lore raw

PATCH v4 14/66 i386/tdx: Implement tdx_kvm_init() to initialize TDX VM context

In reply to Xiaoyao Li

Introduce tdx_kvm_init() and invoke it in kvm_confidential_guest_init() if it's a TDX VM.

Set ms->require_guest_memfd to require kvm guest memfd allocation for any memory backend. More TDX specific 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       |  9 +++++++++
 target/i386/kvm/tdx.h       |  2 ++
 5 files changed, 26 insertions(+), 10 deletions(-)
 create mode 100644 target/i386/kvm/tdx-stub.c
target/i386/kvm/kvm.c +6 -9
index f9a774925cf6..3f1d2272fb06 100644
@@ -62,6 +62,7 @@
 #include "migration/blocker.h"
 #include "exec/memattrs.h"
 #include "trace.h"
+#include "tdx.h"
 
 #include CONFIG_DEVICES
 
@@ -2545,6 +2546,8 @@ 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;
@@ -2559,16 +2562,10 @@ 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
index 6ea0ce27b757..30a90b4d371d 100644
@@ -9,7 +9,7 @@ i386_kvm_ss.add(when: 'CONFIG_XEN_EMU', if_true: files('xen-emu.c'))
 
 i386_kvm_ss.add(when: 'CONFIG_SEV', if_false: files('sev-stub.c'))
 
-i386_kvm_ss.add(when: 'CONFIG_TDX', if_true: files('tdx.c'))
+i386_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
new file mode 100644
index 000000000000..1d866d5496bf
@@ -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 +9 -0
index d3792d4a3d56..621a05beeb4e 100644
@@ -12,10 +12,19 @@
  */
 
 #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)
+{
+    ms->require_guest_memfd = true;
+
+    return 0;
+}
+
 /* tdx guest */
 OBJECT_DEFINE_TYPE_WITH_INTERFACES(TdxGuest,
                                    tdx_guest,
target/i386/kvm/tdx.h +2 -0
index 415aeb5af746..c8a23d95258d 100644
@@ -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
author
#22 lore raw

PATCH v4 15/66 i386/tdx: Get tdx_capabilities via KVM_TDX_CAPABILITIES

In reply to Xiaoyao Li

KVM 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 in v4:
- use {} to initialize struct kvm_tdx_cmd, to avoid memset();
- remove tdx_platform_ioctl() because no user;

Changes in v3:
- rename __tdx_ioctl() to tdx_ioctl_internal()
- Pass errp in get_tdx_capabilities();

changes in v2:
- Make the error message more clear;

changes in v1:
- 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      | 91 +++++++++++++++++++++++++++++++++++++-
 3 files changed, 92 insertions(+), 3 deletions(-)
target/i386/kvm/kvm.c +0 -2
index 3f1d2272fb06..ddbcd1071a7e 100644
@@ -1686,8 +1686,6 @@ 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
index 55fb25fa8e2e..c3ef46a97a7b 100644
@@ -13,6 +13,8 @@
 
 #include "sysemu/kvm.h"
 
+#define KVM_MAX_CPUID_ENTRIES  100
+
 #ifdef CONFIG_KVM
 
 #define kvm_pit_in_kernel() \
target/i386/kvm/tdx.c +90 -1

107 lines not on this page · raw message

Signature · 1 line
2.34.1
author
#23 lore raw

PATCH v4 16/66 i386/tdx: Introduce is_tdx_vm() helper and cache tdx_guest object

In reply to 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_guest object thus no need to cast from ms->cgs every time.

---
changes in v3:
- replace object_dynamic_cast with TDX_GUEST();

---
 target/i386/kvm/tdx.c | 15 ++++++++++++++-
 target/i386/kvm/tdx.h | 10 ++++++++++
 2 files changed, 24 insertions(+), 1 deletion(-)
target/i386/kvm/tdx.c +14 -1
index ad76abd58373..3c59b3b45419 100644
@@ -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_VM_IOCTL,
     TDX_VCPU_IOCTL,
@@ -103,15 +111,20 @@ static int get_tdx_capabilities(Error **errp)
 
 int tdx_kvm_init(MachineState *ms, Error **errp)
 {
+    TdxGuest *tdx = TDX_GUEST(OBJECT(ms->cgs));
     int r = 0;
 
     ms->require_guest_memfd = true;
 
     if (!tdx_caps) {
         r = get_tdx_capabilities(errp);
+        if (r) {
+            return r;
+        }
     }
 
-    return r;
+    tdx_guest = tdx;
+    return 0;
 }
 
 /* tdx guest */
target/i386/kvm/tdx.h +10 -0
index c8a23d95258d..4036ca2f3f99 100644
@@ -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
author
#24 lore raw

PATCH v4 17/66 i386/tdx: Adjust the supported CPUID based on TDX restrictions

In reply to 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_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;

- depends_on_vmm_cap: The bits that are configurable from the view of TDX module. But they requires emulation of VMM when 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;

(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 | 263 ++++++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.h |   2 +
 4 files changed, 285 insertions(+)
target/i386/cpu.h +16 -0
index 7f0786e8b98f..5b6bcba778ae 100644
@@ -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
index ddbcd1071a7e..64dfa6c36a88 100644
@@ -521,6 +521,10 @@ 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 +263 -0

280 lines not on this page · raw message

target/i386/kvm/tdx.h +2 -0
index 4036ca2f3f99..06599b65b827 100644
@@ -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
author
#25 lore raw

PATCH v4 18/66 i386/tdx: Make Intel-PT unsupported for TD guest

In reply to Xiaoyao Li

Due to the fact that Intel-PT virtualization support has been broken in QEMU since Sapphire Rapids generation[1], below warning is triggered when luanching TD guest:

warning: host doesn't support requested feature: CPUID.07H:EBX.intel-pt [bit 25]

Before Intel-pt is fixed in QEMU, just make Intel-PT unsupported for TD guest, to avoid the confusing warning.

[1] https://lore.kernel.org/qemu-devel/[email protected]/

---
Changes in v4:
- newly added patch;

---
 target/i386/kvm/tdx.c | 5 +++++
 1 file changed, 5 insertions(+)
target/i386/kvm/tdx.c +5 -0
index a962f8a4ea60..2703e97f991d 100644
@@ -292,6 +292,11 @@ void tdx_get_supported_cpuid(uint32_t function, uint32_t index, int reg,
     if (function == 1 && reg == R_ECX && !enable_cpu_pm) {
         *ret &= ~CPUID_EXT_MONITOR;
     }
+
+    /* QEMU Intel-pt support is broken, don't advertise Intel-PT */
+    if (function == 7 && reg == R_EBX) {
+        *ret &= ~CPUID_7_0_EBX_INTEL_PT;
+    }
 }
 
 enum tdx_ioctl_level{
Signature · 1 line
2.34.1
author
#26 lore raw

PATCH v4 19/66 i386/tdx: Update tdx_cpuid_lookup[].tdx_fixed0/1 by tdx_caps.cpuid_config[]

In reply to Xiaoyao Li

tdx_cpuid_lookup[].tdx_fixed0/1 is QEMU maintained data which reflects TDX restrictions regrading what bits are fixed by TDX module.

It's retrieved from TDX spec and static. However, TDX may evolve and 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 | 34 ++++++++++++++++++++++++++++++++++
 1 file changed, 34 insertions(+)
target/i386/kvm/tdx.c +34 -0
index 2703e97f991d..5dfea0378f26 100644
@@ -377,6 +377,38 @@ static int get_tdx_capabilities(Error **errp)
     return 0;
 }
 
+static void update_tdx_cpuid_lookup_by_tdx_caps(void)
+{
+    KvmTdxCpuidLookup *entry;
+    FeatureWordInfo *fi;
+    uint32_t config;
+    FeatureWord w;
+
+    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);
+
+        if (!config) {
+            continue;
+        }
+
+        /*
+         * Remove the configurable bits from tdx_fixed0/1 in case QEMU
+         * maintained fixed0/1 values is outdated to TDX module.
+         */
+        entry->tdx_fixed0 &= ~config;
+        entry->tdx_fixed1 &= ~config;
+    }
+}
+
 int tdx_kvm_init(MachineState *ms, Error **errp)
 {
     TdxGuest *tdx = TDX_GUEST(OBJECT(ms->cgs));
@@ -391,6 +423,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp)
         }
     }
 
+    update_tdx_cpuid_lookup_by_tdx_caps();
+
     tdx_guest = tdx;
     return 0;
 }
Signature · 1 line
2.34.1
author
#27 lore raw

PATCH v4 20/66 i386/tdx: Integrate tdx_caps->xfam_fixed0/1 into tdx_cpuid_lookup

In reply to Xiaoyao Li

KVM 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
index 03822d9ba8ee..160ba8c940a2 100644
@@ -1575,9 +1575,6 @@ 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
index 5b6bcba778ae..23d187d7cc5f 100644
@@ -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
index 5dfea0378f26..4c8455783e36 100644
@@ -407,6 +407,30 @@ 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
author
#28 lore raw

PATCH v4 21/66 i386/tdx: Integrate tdx_caps->attrs_fixed0/1 to tdx_cpuid_lookup

In reply to Xiaoyao Li

Some 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.

---
Changes in v4:
- reverse the meaning of tdx_caps->attr_fixed0, because value 0 of bit means the bit must be fixed 0.

---
 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
index 9baac5c0b450..e980f6e3147f 100644
@@ -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
index 160ba8c940a2..39b07f8e1204 100644
@@ -1442,15 +1442,6 @@ 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
index 23d187d7cc5f..2091451a140e 100644
@@ -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
index 4c8455783e36..6caefb27d90a 100644
@@ -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;
@@ -383,6 +390,8 @@ static void update_tdx_cpuid_lookup_by_tdx_caps(void)
     FeatureWordInfo *fi;
     uint32_t config;
     FeatureWord w;
+    FeatureMask *fm;
+    int i;
 
     for (w = 0; w < FEATURE_WORDS; w++) {
         fi = &feature_word_info[w];
@@ -408,6 +417,18 @@ 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
author
#29 lore raw

PATCH v4 22/66 i386/kvm: Move architectural CPUID leaf generation to separate helper

In reply to Xiaoyao Li

From: 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      | 459 +++++++++++++++++++------------------
 target/i386/kvm/kvm_i386.h |   3 +
 2 files changed, 240 insertions(+), 222 deletions(-)
target/i386/kvm/kvm.c +237 -222
index 64dfa6c36a88..8e3b491fb071 100644
@@ -1732,6 +1732,241 @@ 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;
+
+    if (cpuid_i >  KVM_MAX_CPUID_ENTRIES) {
+        error_report("exceeded cpuid index (%d) for entries[]", cpuid_i);
+        abort();
+    }
+
+    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 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 0x7:
+        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 {
@@ -1748,8 +1983,7 @@ 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;
@@ -1902,8 +2136,6 @@ 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) {
@@ -1914,224 +2146,7 @@ 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 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 0x7:
-        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
index c3ef46a97a7b..cbf52c1c6d17 100644
@@ -24,6 +24,9 @@
 #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
author
#30 lore raw

PATCH v4 23/66 kvm: Introduce kvm_arch_pre_create_vcpu()

In reply to 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 call TDX_INIT_VM before creating any vcpu.

---
Changes in v3:
- pass @errp to kvm_arch_pre_create_vcpu(); (Per Daniel)

---
 accel/kvm/kvm-all.c  | 10 ++++++++++
 include/sysemu/kvm.h |  1 +
 2 files changed, 11 insertions(+)
accel/kvm/kvm-all.c +10 -0
index def2b46b9de2..7b177e3ee7c5 100644
@@ -418,6 +418,11 @@ 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, Error **errp)
+{
+    return 0;
+}
+
 int kvm_init_vcpu(CPUState *cpu, Error **errp)
 {
     KVMState *s = kvm_state;
@@ -426,6 +431,11 @@ 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, errp);
+    if (ret < 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
index 0e88958190a4..2f6592859ac6 100644
@@ -341,6 +341,7 @@ int kvm_arch_get_default_type(MachineState *ms);
 
 int kvm_arch_init(MachineState *ms, KVMState *s);
 
+int kvm_arch_pre_create_vcpu(CPUState *cpu, Error **errp);
 int kvm_arch_init_vcpu(CPUState *cpu);
 int kvm_arch_destroy_vcpu(CPUState *cpu);
 
Signature · 1 line
2.34.1
author
#31 lore raw

PATCH v4 24/66 i386/tdx: Initialize TDX before creating TD vcpus

In reply to Xiaoyao Li

Invoke 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().

---
Changes in v4:
- mark init_vm with g_autofree() and use QEMU_LOCK_GUARD() to eliminate the goto labels; (Daniel)
Changes in v3:
- Pass @errp in tdx_pre_create_vcpu() and pass error info to it. (Daniel)

---
 accel/kvm/kvm-all.c        |  9 ++++++++-
 target/i386/kvm/kvm.c      |  9 +++++++++
 target/i386/kvm/tdx-stub.c |  5 +++++
 target/i386/kvm/tdx.c      | 41 ++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.h      |  4 ++++
 5 files changed, 67 insertions(+), 1 deletion(-)
accel/kvm/kvm-all.c +8 -1
index 7b177e3ee7c5..55f69d0f049a 100644
@@ -431,8 +431,15 @@ 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, errp);
     if (ret < 0) {
+        cpu->kvm_state = NULL;
         goto err;
     }
 
@@ -440,11 +447,11 @@ 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 +9 -0
index 8e3b491fb071..fa07daad607a 100644
@@ -2272,6 +2272,15 @@ int kvm_arch_init_vcpu(CPUState *cs)
     return r;
 }
 
+int kvm_arch_pre_create_vcpu(CPUState *cpu, Error **errp)
+{
+    if (is_tdx_vm()) {
+        return tdx_pre_create_vcpu(cpu, errp);
+    }
+
+    return 0;
+}
+
 int kvm_arch_destroy_vcpu(CPUState *cs)
 {
     X86CPU *cpu = X86_CPU(cs);
target/i386/kvm/tdx-stub.c +5 -0
index 1d866d5496bf..3877d432a397 100644
@@ -6,3 +6,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp)
 {
     return -EINVAL;
 }
+
+int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
+{
+    return -EINVAL;
+}
target/i386/kvm/tdx.c +41 -0

53 lines not on this page · raw message

target/i386/kvm/tdx.h +4 -0
index 06599b65b827..432077723ac5 100644
@@ -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, Error **errp);
 
 #endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
author
#32 lore raw

PATCH v4 25/66 i386/tdx: Add property sept-ve-disable for tdx-guest object

In reply to 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.

---
Changes in v4:
- collect Acked-by from Markus

Changes in v3:
- update the comment of property @sept-ve-disable to make it more descriptive and use new format. (Daniel and Markus)

---
 qapi/qom.json         |  7 ++++++-
 target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++
 2 files changed, 30 insertions(+), 1 deletion(-)
qapi/qom.json +6 -1
index 5b3c3146947f..2177f3101382 100644
@@ -900,10 +900,15 @@
 #
 # Properties for tdx-guest objects.
 #
+# @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.
+#
 # Since: 9.0
 ##
 { 'struct': 'TdxGuestProperties',
-  'data': { }}
+  'data': { '*sept-ve-disable': 'bool' } }
 
 ##
 # @ThreadContextProperties:
target/i386/kvm/tdx.c +24 -0
index 634aca504b32..51db64bec7ce 100644
@@ -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      64
 
 static FeatureMask tdx_attrs_ctrl_fields[TDX_ATTRIBUTES_MAX_BITS] = {
@@ -513,6 +515,24 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
     return 0;
 }
 
+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,
@@ -528,6 +548,10 @@ 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
author
#33 lore raw

PATCH v4 26/66 i386/tdx: Make sept_ve_disable set by default

In reply to Xiaoyao Li

From: 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
index 51db64bec7ce..91cd116fb153 100644
@@ -547,7 +547,7 @@ 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
author
#34 lore raw

PATCH v4 27/66 i386/tdx: Wire CPU features up with attributes of TD guest

In reply to Xiaoyao Li

For 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 | 13 +++++++++++++
 1 file changed, 13 insertions(+)
target/i386/kvm/tdx.c +13 -0
index 91cd116fb153..1cb38b5d6221 100644
@@ -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
 
@@ -476,6 +478,15 @@ 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, Error **errp)
 {
     MachineState *ms = MACHINE(qdev_get_machine());
@@ -498,6 +509,8 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
         return r;
     }
 
+    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;
Signature · 1 line
2.34.1
author
#35 lore raw

PATCH v4 28/66 i386/tdx: Validate TD attributes

In reply to Xiaoyao Li

Validate 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.

---
Changes in v3:
- using error_setg() for error report; (Daniel)

---
 target/i386/kvm/tdx.c | 29 +++++++++++++++++++++++++++--
 1 file changed, 27 insertions(+), 2 deletions(-)
target/i386/kvm/tdx.c +27 -2
index 1cb38b5d6221..8c2bf512397e 100644
@@ -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)
@@ -478,13 +479,34 @@ 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, Error **errp)
+{
+    if (((tdx->attributes & tdx_caps->attrs_fixed0) | tdx_caps->attrs_fixed1) !=
+        tdx->attributes) {
+            error_setg(errp, "Invalid attributes 0x%lx for TDX VM "
+                       "(fixed0 0x%llx, fixed1 0x%llx)",
+                       tdx->attributes, tdx_caps->attrs_fixed0,
+                       tdx_caps->attrs_fixed1);
+            return -1;
+    }
+
+    if (tdx->attributes & TDX_TD_ATTRIBUTES_DEBUG) {
+        error_setg(errp, "Current QEMU doesn't support attributes.debug[bit 0] for TDX VM");
+        return -1;
+    }
+
+    return 0;
+}
+
+static int setup_td_guest_attributes(X86CPU *x86cpu, Error **errp)
 {
     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, errp);
 }
 
 int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
@@ -509,7 +531,10 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
         return r;
     }
 
-    setup_td_guest_attributes(x86cpu);
+    r = setup_td_guest_attributes(x86cpu, errp);
+    if (r) {
+        return r;
+    }
 
     init_vm->cpuid.nent = kvm_x86_arch_cpuid(env, init_vm->cpuid.entries, 0);
 
Signature · 1 line
2.34.1
author
#36 lore raw

PATCH v4 29/66 i386/tdx: Support user configurable mrconfigid/mrowner/mrownerconfig

In reply to Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Three sha384 hash values, mrconfigid, mrowner and mrownerconfig, of a TD can be provided for TDX attestation. Detailed meaning of them can be
found: https://lore.kernel.org/qemu-devel/[email protected]/

Allow user to specify those values via property mrconfigid, mrowner and mrownerconfig. They are all in base64 format.

example
-object tdx-guest, \
mrconfigid=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v,\
mrowner=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v,\
mrownerconfig=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v

---
Changes in v4:
- describe more of there fields in qom.json
- free the old value before set new value to avoid memory leak in _setter(); (Daniel)

Changes in v3:
- use base64 encoding instread of hex-string;

---
 qapi/qom.json         | 14 ++++++-
 target/i386/kvm/tdx.c | 87 +++++++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.h |  3 ++
 3 files changed, 103 insertions(+), 1 deletion(-)
qapi/qom.json +13 -1
index 2177f3101382..15445f9e41fc 100644
@@ -905,10 +905,22 @@
 #     pages.  Some guest OS (e.g., Linux TD guest) may require this to
 #     be set, otherwise they refuse to boot.
 #
+# @mrconfigid: ID for non-owner-defined configuration of the guest TD,
+#     e.g., run-time or OS configuration.  base64 encoded SHA384 digest.
+#
+# @mrowner: ID for the guest TD’s owner.  base64 encoded SHA384 digest.
+#
+# @mrownerconfig: ID for owner-defined configuration of the guest TD,
+#     e.g., specific to the workload rather than the run-time or OS.
+#     base64 encoded SHA384 digest.
+#
 # Since: 9.0
 ##
 { 'struct': 'TdxGuestProperties',
-  'data': { '*sept-ve-disable': 'bool' } }
+  'data': { '*sept-ve-disable': 'bool',
+            '*mrconfigid': 'str',
+            '*mrowner': 'str',
+            '*mrownerconfig': 'str' } }
 
 ##
 # @ThreadContextProperties:
target/i386/kvm/tdx.c +87 -0

117 lines not on this page · raw message

target/i386/kvm/tdx.h +3 -0
index 432077723ac5..6e39ef3bac13 100644
@@ -21,6 +21,9 @@ typedef struct TdxGuest {
 
     bool initialized;
     uint64_t attributes;    /* TD attributes */
+    char *mrconfigid;       /* base64 encoded sha348 digest */
+    char *mrowner;          /* base64 encoded sha348 digest */
+    char *mrownerconfig;    /* base64 encoded sha348 digest */
 } TdxGuest;
 
 #ifdef CONFIG_TDX
Signature · 1 line
2.34.1
#37 lore raw

Re: [PATCH v4 29/66] i386/tdx: Support user configurable mrconfigid/mrowner/mrownerconfig

Quoted 38 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Three sha384 hash values, mrconfigid, mrowner and mrownerconfig, of a TD can be provided for TDX attestation. Detailed meaning of them can be
found: https://lore.kernel.org/qemu-devel/[email protected]/

Allow user to specify those values via property mrconfigid, mrowner and mrownerconfig. They are all in base64 format.

example
-object tdx-guest, \
mrconfigid=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v,\
mrowner=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v,\
mrownerconfig=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v

---
Changes in v4:
- describe more of there fields in qom.json
- free the old value before set new value to avoid memory leak in _setter(); (Daniel)

Changes in v3:
- use base64 encoding instread of hex-string;

---
 qapi/qom.json         | 14 ++++++-
 target/i386/kvm/tdx.c | 87 +++++++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.h |  3 ++
 3 files changed, 103 insertions(+), 1 deletion(-)
qapi/qom.json +0 -0
index 2177f3101382..15445f9e41fc 100644

qapi/qom.json line 909

Quoted 5 lines
qapi/qom.json +2 -0
@@ -905,10 +905,22 @@
 #     pages.  Some guest OS (e.g., Linux TD guest) may require this to
 #     be set, otherwise they refuse to boot.
 #
+# @mrconfigid: ID for non-owner-defined configuration of the guest TD,
+#     e.g., run-time or OS configuration.  base64 encoded SHA384 digest.

"base64 encoded SHA384" is not a sentence.

Double-checking: the data being hashed here is the "non-owner-defined configuration of the guest TD", and the resulting hash is the "ID"?

qapi/qom.json line 911

Quoted 2 lines
qapi/qom.json +2 -0
+#
+# @mrowner: ID for the guest TD’s owner.  base64 encoded SHA384 digest.

Likewise.

qapi/qom.json line 915

Quoted 4 lines
qapi/qom.json +4 -0
+#
+# @mrownerconfig: ID for owner-defined configuration of the guest TD,
+#     e.g., specific to the workload rather than the run-time or OS.
+#     base64 encoded SHA384 digest.

Likewise.

qapi/qom.json line 923

Quoted 9 lines
qapi/qom.json +5 -1
+#
 # Since: 9.0
 ##
 { 'struct': 'TdxGuestProperties',
-  'data': { '*sept-ve-disable': 'bool' } }
+  'data': { '*sept-ve-disable': 'bool',
+            '*mrconfigid': 'str',
+            '*mrowner': 'str',
+            '*mrownerconfig': 'str' } }

The new members are optional, but their description in the doc comment doesn't explain behavior when present vs. behavior when absent.

Quoted 3 lines
qapi/qom.json +0 -0
 ##
 # @ThreadContextProperties:

[...]

author
#38 lore raw

Re: [PATCH v4 29/66] i386/tdx: Support user configurable mrconfigid/mrowner/mrownerconfig

Quoted 48 lines · Markus Armbruster
Quoted 38 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Three sha384 hash values, mrconfigid, mrowner and mrownerconfig, of a TD can be provided for TDX attestation. Detailed meaning of them can be
found: https://lore.kernel.org/qemu-devel/[email protected]/

Allow user to specify those values via property mrconfigid, mrowner and mrownerconfig. They are all in base64 format.

example

-object tdx-guest, \
   mrconfigid=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v,\
   mrowner=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v,\
   mrownerconfig=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v

---
Changes in v4:
- describe more of there fields in qom.json
- free the old value before set new value to avoid memory leak in

    _setter(); (Daniel)

Changes in v3:
- use base64 encoding instread of hex-string;

---
  qapi/qom.json         | 14 ++++++-
  target/i386/kvm/tdx.c | 87 +++++++++++++++++++++++++++++++++++++++++++
  target/i386/kvm/tdx.h |  3 ++
  3 files changed, 103 insertions(+), 1 deletion(-)
qapi/qom.json +0 -0
index 2177f3101382..15445f9e41fc 100644

qapi/qom.json line 909

Quoted 5 lines
qapi/qom.json +2 -0
@@ -905,10 +905,22 @@
  #     pages.  Some guest OS (e.g., Linux TD guest) may require this to
  #     be set, otherwise they refuse to boot.
  #
+# @mrconfigid: ID for non-owner-defined configuration of the guest TD,
+#     e.g., run-time or OS configuration.  base64 encoded SHA384 digest.

"base64 encoded SHA384" is not a sentence.

Quoted 2 lines

Double-checking: the data being hashed here is the "non-owner-defined configuration of the guest TD", and the resulting hash is the "ID"?

yes. The "ID" here means the resulting hash.

The reason to use "ID" here because in the TDX spec, it's description is

Software-defined ID for non-owner-defined configuration of the guest TD - e.g., run-time or OS configuration.

If ID is confusing, how about

SHA384 hash of non-owner-defined configuration of the guest TD, e.g., run-time of OS configuration. It's base64 encoded.

Quoted 32 lines

qapi/qom.json line 911

Quoted 2 lines
qapi/qom.json +2 -0
+#
+# @mrowner: ID for the guest TD’s owner.  base64 encoded SHA384 digest.

Likewise.

qapi/qom.json line 915

Quoted 4 lines
qapi/qom.json +4 -0
+#
+# @mrownerconfig: ID for owner-defined configuration of the guest TD,
+#     e.g., specific to the workload rather than the run-time or OS.
+#     base64 encoded SHA384 digest.

Likewise.

qapi/qom.json line 923

Quoted 9 lines
qapi/qom.json +5 -1
+#
  # Since: 9.0
  ##
  { 'struct': 'TdxGuestProperties',
-  'data': { '*sept-ve-disable': 'bool' } }
+  'data': { '*sept-ve-disable': 'bool',
+            '*mrconfigid': 'str',
+            '*mrowner': 'str',
+            '*mrownerconfig': 'str' } }

The new members are optional, but their description in the doc comment doesn't explain behavior when present vs. behavior when absent.

Quoted 3 lines
qapi/qom.json +0 -0
  ##
  # @ThreadContextProperties:

[...]

#39 lore raw

Re: [PATCH v4 29/66] i386/tdx: Support user configurable mrconfigid/mrowner/mrownerconfig

Quoted 56 lines · Xiaoyao Li
Quoted 48 lines · Markus Armbruster
Quoted 38 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Three sha384 hash values, mrconfigid, mrowner and mrownerconfig, of a TD can be provided for TDX attestation. Detailed meaning of them can be
found: https://lore.kernel.org/qemu-devel/[email protected]/

Allow user to specify those values via property mrconfigid, mrowner and mrownerconfig. They are all in base64 format.

example

-object tdx-guest, \
   mrconfigid=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v,\
   mrowner=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v,\
   mrownerconfig=ASNFZ4mrze8BI0VniavN7wEjRWeJq83vASNFZ4mrze8BI0VniavN7wEjRWeJq83v

---
Changes in v4:
- describe more of there fields in qom.json
- free the old value before set new value to avoid memory leak in

    _setter(); (Daniel)

Changes in v3:
- use base64 encoding instread of hex-string;

---
  qapi/qom.json         | 14 ++++++-
  target/i386/kvm/tdx.c | 87 +++++++++++++++++++++++++++++++++++++++++++
  target/i386/kvm/tdx.h |  3 ++
  3 files changed, 103 insertions(+), 1 deletion(-)
qapi/qom.json +0 -0
index 2177f3101382..15445f9e41fc 100644

qapi/qom.json line 909

Quoted 5 lines
qapi/qom.json +2 -0
@@ -905,10 +905,22 @@
  #     pages.  Some guest OS (e.g., Linux TD guest) may require this to
  #     be set, otherwise they refuse to boot.
  #
+# @mrconfigid: ID for non-owner-defined configuration of the guest TD,
+#     e.g., run-time or OS configuration.  base64 encoded SHA384 digest.

"base64 encoded SHA384" is not a sentence.

Quoted 2 lines

Double-checking: the data being hashed here is the "non-owner-defined configuration of the guest TD", and the resulting hash is the "ID"?

yes. The "ID" here means the resulting hash.

The reason to use "ID" here because in the TDX spec, it's description is

Quoted 7 lines

Software-defined ID for non-owner-defined configuration of the guest TD - e.g., run-time or OS configuration.

If ID is confusing, how about

SHA384 hash of non-owner-defined configuration of the guest TD, e.g., run-time of OS configuration. It's base64 encoded.

I guess staying close to the TDX spec makes sense.

We still need to mention the base64 encoding.

What about something like

ID for non-owner-defined configuration of the guest TD, e.g., run-time or OS configuration (base64 encoded SHA384 digest)

or, if we decide that the fact it's SHA384 digest is irrelevant for QMP

ID for non-owner-defined configuration of the guest TD, e.g., run-time or OS configuration (base64 encoded)

Quoted 32 lines
Quoted 32 lines

qapi/qom.json line 911

Quoted 2 lines
qapi/qom.json +2 -0
+#
+# @mrowner: ID for the guest TD’s owner.  base64 encoded SHA384 digest.

Likewise.

qapi/qom.json line 915

Quoted 4 lines
qapi/qom.json +4 -0
+#
+# @mrownerconfig: ID for owner-defined configuration of the guest TD,
+#     e.g., specific to the workload rather than the run-time or OS.
+#     base64 encoded SHA384 digest.

Likewise.

qapi/qom.json line 923

Quoted 9 lines
qapi/qom.json +5 -1
+#
  # Since: 9.0
  ##
  { 'struct': 'TdxGuestProperties',
-  'data': { '*sept-ve-disable': 'bool' } }
+  'data': { '*sept-ve-disable': 'bool',
+            '*mrconfigid': 'str',
+            '*mrowner': 'str',
+            '*mrownerconfig': 'str' } }

The new members are optional, but their description in the doc comment doesn't explain behavior when present vs. behavior when absent.

Quoted 3 lines
qapi/qom.json +0 -0
  ##
  # @ThreadContextProperties:

[...]

author
#40 lore raw

PATCH v4 30/66 i386/tdx: Implement user specified tsc frequency

In reply to Xiaoyao Li

Reuse "-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 in v3:
- use @errp to report error info; (Daniel)

Changes in v1:
- 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 | 25 +++++++++++++++++++++++++
 2 files changed, 34 insertions(+)
target/i386/kvm/kvm.c +9 -0
index fa07daad607a..348f76a9cb81 100644
@@ -794,6 +794,15 @@ 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 +25 -0
index ead19923f318..a27c7b068a07 100644
@@ -33,6 +33,9 @@
                                      (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)
@@ -565,6 +568,28 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
         return r;
     }
 
+    if (env->tsc_khz && (env->tsc_khz < TDX_MIN_TSC_FREQUENCY_KHZ ||
+                         env->tsc_khz > TDX_MAX_TSC_FREQUENCY_KHZ)) {
+        error_setg(errp, "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);
+       return -EINVAL;
+    }
+
+    if (env->tsc_khz % (25 * 1000)) {
+        error_setg(errp, "Invalid TSC %ld KHz, it must be multiple of 25MHz",
+                   env->tsc_khz);
+        return -EINVAL;
+    }
+
+    /* 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_setg_errno(errp, -r, "Unable to set TSC frequency to %" PRId64 " kHz",
+                         env->tsc_khz);
+        return r;
+    }
+
     r = setup_td_guest_attributes(x86cpu, errp);
     if (r) {
         return r;
Signature · 1 line
2.34.1
author
#41 lore raw

PATCH v4 31/66 i386/tdx: Set kvm_readonly_mem_enabled to false for TDX VM

In reply to Xiaoyao Li

TDX 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
index a27c7b068a07..7b250d80bc1d 100644
@@ -479,6 +479,15 @@ 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
author
#42 lore raw

PATCH v4 32/66 kvm/memory: Introduce the infrastructure to set the default shared/private value

In reply to Xiaoyao Li

Introduce new flag RAM_DEFAULT_PRIVATE for RAMBlock. It's used to indicate the default attribute, private or not.

Set the RAM range to private explicitly when it's default private.

Originated-from: Isaku Yamahata <[email protected]>

---
 accel/kvm/kvm-all.c   | 10 ++++++++++
 include/exec/memory.h |  6 ++++++
 system/memory.c       | 13 +++++++++++++
 3 files changed, 29 insertions(+)
accel/kvm/kvm-all.c +10 -0
index 55f69d0f049a..094ce7695e16 100644
@@ -1458,6 +1458,16 @@ static void kvm_set_phys_mem(KVMMemoryListener *kml,
                     strerror(-err));
             abort();
         }
+
+        if (memory_region_is_default_private(mr)) {
+            err = kvm_set_memory_attributes_private(start_addr, slot_size);
+            if (err) {
+                error_report("%s: failed to set memory attribute private: %s\n",
+                             __func__, strerror(-err));
+                exit(1);
+            }
+        }
+
         start_addr += slot_size;
         ram_start_offset += slot_size;
         ram += slot_size;
include/exec/memory.h +6 -0
index f11036ead15e..7229fcc0415f 100644
@@ -246,6 +246,9 @@ typedef struct IOMMUTLBEvent {
 /* RAM can be private that has kvm guest memfd backend */
 #define RAM_GUEST_MEMFD   (1 << 12)
 
+/* RAM is default private */
+#define RAM_DEFAULT_PRIVATE     (1 << 13)
+
 static inline void iommu_notifier_init(IOMMUNotifier *n, IOMMUNotify fn,
                                        IOMMUNotifierFlag flags,
                                        hwaddr start, hwaddr end,
@@ -1736,6 +1739,9 @@ bool memory_region_is_protected(MemoryRegion *mr);
  */
 bool memory_region_has_guest_memfd(MemoryRegion *mr);
 
+void memory_region_set_default_private(MemoryRegion *mr);
+bool memory_region_is_default_private(MemoryRegion *mr);
+
 /**
  * memory_region_get_iommu: check whether a memory region is an iommu
  *
system/memory.c +13 -0
index c756950c0c0f..74f647f2e56f 100644
@@ -1855,6 +1855,19 @@ bool memory_region_has_guest_memfd(MemoryRegion *mr)
     return mr->ram_block && mr->ram_block->guest_memfd >= 0;
 }
 
+bool memory_region_is_default_private(MemoryRegion *mr)
+{
+    return memory_region_has_guest_memfd(mr) &&
+           (mr->ram_block->flags & RAM_DEFAULT_PRIVATE);
+}
+
+void memory_region_set_default_private(MemoryRegion *mr)
+{
+    if (memory_region_has_guest_memfd(mr)) {
+        mr->ram_block->flags |= RAM_DEFAULT_PRIVATE;
+    }
+}
+
 uint8_t memory_region_get_dirty_log_mask(MemoryRegion *mr)
 {
     uint8_t mask = mr->dirty_log_mask;
Signature · 1 line
2.34.1
author
#43 lore raw

PATCH v4 33/66 i386/tdx: Make memory type private by default

In reply to Xiaoyao Li

By 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]>

---
 include/exec/memory.h |  1 +
 target/i386/kvm/tdx.c | 20 ++++++++++++++++++++
 2 files changed, 21 insertions(+)
include/exec/memory.h +1 -0
index 7229fcc0415f..f25959f6d30f 100644
@@ -850,6 +850,7 @@ struct IOMMUMemoryRegion {
 #define MEMORY_LISTENER_PRIORITY_MIN            0
 #define MEMORY_LISTENER_PRIORITY_ACCEL          10
 #define MEMORY_LISTENER_PRIORITY_DEV_BACKEND    10
+#define MEMORY_LISTENER_PRIORITY_ACCEL_HIGH     20
 
 /**
  * struct MemoryListener: callbacks structure for updates to the physical memory map
target/i386/kvm/tdx.c +20 -0
index 7b250d80bc1d..f892551821ce 100644
@@ -19,6 +19,7 @@
 #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"
@@ -621,6 +622,19 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
     return 0;
 }
 
+static void tdx_guest_region_add(MemoryListener *listener,
+                                 MemoryRegionSection *section)
+{
+    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);
@@ -695,6 +709,12 @@ 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
#44 lore raw

Re: [PATCH v4 33/66] i386/tdx: Make memory type private by default

Quoted 56 lines · Xiaoyao Li

By 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]>

---
  include/exec/memory.h |  1 +
  target/i386/kvm/tdx.c | 20 ++++++++++++++++++++
  2 files changed, 21 insertions(+)
include/exec/memory.h +1 -0
index 7229fcc0415f..f25959f6d30f 100644
@@ -850,6 +850,7 @@ struct IOMMUMemoryRegion {
  #define MEMORY_LISTENER_PRIORITY_MIN            0
  #define MEMORY_LISTENER_PRIORITY_ACCEL          10
  #define MEMORY_LISTENER_PRIORITY_DEV_BACKEND    10
+#define MEMORY_LISTENER_PRIORITY_ACCEL_HIGH     20
 
  /**
   * struct MemoryListener: callbacks structure for updates to the physical memory map
target/i386/kvm/tdx.c +1 -0
index 7b250d80bc1d..f892551821ce 100644
@@ -19,6 +19,7 @@
  #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"

target/i386/kvm/tdx.c int tdx_pre_create_vcpu(CPUState *cpu, Error **errp) · line 629

Quoted 8 lines
target/i386/kvm/tdx.c +5 -0
@@ -621,6 +622,19 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
      return 0;
  }
 
+static void tdx_guest_region_add(MemoryListener *listener,
+                                 MemoryRegionSection *section)
+{
+    memory_region_set_default_private(section->mr);
+}

That looks fishy. Why is TDX to decide what happens to other memory regions it doesn't own?

We should define that behavior when creating these memory region, and TDX could sanity check that they have been setup properly.

Let me ask differently: For which memory region where we have RAM_GUEST_MEMFD set would we *not* want to set private as default right from the start?

Signature · 3 lines
Cheers,

David / dhildenb
author
#45 lore raw

Re: [PATCH v4 33/66] i386/tdx: Make memory type private by default

Quoted 65 lines · David Hildenbrand
Quoted 54 lines · Xiaoyao Li

By 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]>

---
  include/exec/memory.h |  1 +
  target/i386/kvm/tdx.c | 20 ++++++++++++++++++++

  2 files changed, 21 insertions(+)
include/exec/memory.h +1 -0
index 7229fcc0415f..f25959f6d30f 100644
@@ -850,6 +850,7 @@ struct IOMMUMemoryRegion {
   #define MEMORY_LISTENER_PRIORITY_MIN            0
   #define MEMORY_LISTENER_PRIORITY_ACCEL          10
   #define MEMORY_LISTENER_PRIORITY_DEV_BACKEND    10
+#define MEMORY_LISTENER_PRIORITY_ACCEL_HIGH     20
   /**
    * struct MemoryListener: callbacks structure for updates to the physical memory map
target/i386/kvm/tdx.c +1 -0
index 7b250d80bc1d..f892551821ce 100644
@@ -19,6 +19,7 @@
   #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"

target/i386/kvm/tdx.c int tdx_pre_create_vcpu(CPUState *cpu, Error **errp) · line 628

Quoted 7 lines
target/i386/kvm/tdx.c +5 -0
@@ -621,6 +622,19 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
       return 0;
   }
+static void tdx_guest_region_add(MemoryListener *listener,
+                                 MemoryRegionSection *section)
+{
+    memory_region_set_default_private(section->mr);
+}

That looks fishy. Why is TDX to decide what happens to other memory regions it doesn't own?

Quoted 3 lines

We should define that behavior when creating these memory region, and TDX could sanity check that they have been setup properly.

Let me ask differently: For which memory region where we have RAM_GUEST_MEMFD set would we *not* want to set private as default right from the start?

All memory regions have RAM_GUEST_MEMFD set will benefit from being private as default, for TDX guest.

I will update the implementation to set RAM_DEFAULT_PRIVATE flag when guest_memfd is created successfully, like

system/physmem.c +2 -0
index fc59470191ef..60676689c807 100644
@@ -1850,6 +1850,8 @@ static void ram_block_add(RAMBlock *new_block, Error **errp)
             qemu_mutex_unlock_ramlist();
             return;
         }
+
+        new_block->flags |= RAM_DEFAULT_PRIVATE;
     }
 

then this patch can be dropped, and the calling of memory_region_set_default_private(mr) of Patch 45 can be dropped too.

I think this is what you suggested, right?

#46 lore raw

Re: [PATCH v4 33/66] i386/tdx: Make memory type private by default

Quoted 97 lines · Xiaoyao Li
Quoted 67 lines · David Hildenbrand
Quoted 55 lines · Xiaoyao Li

By 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]>

---
   include/exec/memory.h |  1 +
   target/i386/kvm/tdx.c | 20 ++++++++++++++++++++
   2 files changed, 21 insertions(+)
include/exec/memory.h +1 -0
index 7229fcc0415f..f25959f6d30f 100644
@@ -850,6 +850,7 @@ struct IOMMUMemoryRegion {
   #define MEMORY_LISTENER_PRIORITY_MIN            0
   #define MEMORY_LISTENER_PRIORITY_ACCEL          10
   #define MEMORY_LISTENER_PRIORITY_DEV_BACKEND    10
+#define MEMORY_LISTENER_PRIORITY_ACCEL_HIGH     20
   /**
    * struct MemoryListener: callbacks structure for updates to the

physical memory map

target/i386/kvm/tdx.c +1 -0
index 7b250d80bc1d..f892551821ce 100644
@@ -19,6 +19,7 @@
   #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"

target/i386/kvm/tdx.c int tdx_pre_create_vcpu(CPUState *cpu, Error **errp) · line 628

Quoted 7 lines
target/i386/kvm/tdx.c +5 -0
@@ -621,6 +622,19 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
       return 0;
   }
+static void tdx_guest_region_add(MemoryListener *listener,
+                                 MemoryRegionSection *section)
+{
+    memory_region_set_default_private(section->mr);
+}

That looks fishy. Why is TDX to decide what happens to other memory regions it doesn't own?

Quoted 6 lines

We should define that behavior when creating these memory region, and TDX could sanity check that they have been setup properly.

Let me ask differently: For which memory region where we have RAM_GUEST_MEMFD set would we *not* want to set private as default right from the start?

All memory regions have RAM_GUEST_MEMFD set will benefit from being private as default, for TDX guest.

I will update the implementation to set RAM_DEFAULT_PRIVATE flag when guest_memfd is created successfully, like

system/physmem.c +2 -0
index fc59470191ef..60676689c807 100644
@@ -1850,6 +1850,8 @@ static void ram_block_add(RAMBlock *new_block,
 Error **errp)
               qemu_mutex_unlock_ramlist();
               return;
           }
+
+        new_block->flags |= RAM_DEFAULT_PRIVATE;
       }
 

then this patch can be dropped, and the calling of memory_region_set_default_private(mr) of Patch 45 can be dropped too.

Quoted 1 line

I think this is what you suggested, right?

Yes, if that works, great!

Signature · 3 lines
Cheers,

David / dhildenb
author
#47 lore raw

PATCH v4 34/66 kvm/tdx: Don't complain when converting vMMIO region to shared

In reply to 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.

---
 accel/kvm/kvm-all.c | 19 ++++++++++++++++---
 1 file changed, 16 insertions(+), 3 deletions(-)
accel/kvm/kvm-all.c +16 -3
index 094ce7695e16..e26e9121b30d 100644
@@ -2975,9 +2975,22 @@ static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private)
             ret = ram_block_discard_guest_memfd_range(rb, offset, size);
         }
     } else {
-        error_report("Convert non guest_memfd backed memory region "
-                    "(0x%"HWADDR_PRIx" ,+ 0x%"HWADDR_PRIx") to %s",
-                    start, size, to_private ? "private" : "shared");
+        /*
+         * 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)) {
+		    ret = 0;
+        } else {
+            error_report("Convert non guest_memfd backed memory region "
+                        "(0x%"HWADDR_PRIx" ,+ 0x%"HWADDR_PRIx") to %s",
+                        start, size, to_private ? "private" : "shared");
+        }
     }
 
     memory_region_unref(section.mr);
Signature · 1 line
2.34.1
author
#48 lore raw

PATCH v4 35/66 kvm/tdx: Ignore memory conversion to shared of unassigned region

In reply to Xiaoyao Li

From: 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
index e26e9121b30d..eb8b3925dbe1 100644
@@ -2943,6 +2943,18 @@ static int kvm_convert_memory(hwaddr start, hwaddr size, bool to_private)
     section = memory_region_find(get_system_memory(), start, size);
     mr = section.mr;
     if (!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) {
+            return 0;
+        }
         return -1;
     }
 
Signature · 1 line
2.34.1
author
#49 lore raw

PATCH v4 36/66 i386/tdvf: Introduce function to parse TDVF metadata

In reply to Xiaoyao Li

From: 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 in v1:
- rename tdvf_parse_section_entry() to tdvf_parse_and_check_section_entry()
Changes in RFC 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
index c0ccf50ac3ef..4e6c8905f077 100644
@@ -12,6 +12,7 @@ config SGX
 
 config TDX
     bool
+    select X86_FW_OVMF
     depends on KVM
 
 config PC
hw/i386/meson.build +1 -0
index 369c6bf823bb..6808bd4e3032 100644
@@ -27,6 +27,7 @@ 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
author
#50 lore raw

PATCH v4 37/66 i386/tdx: Parse TDVF metadata for TDX VM

In reply to Xiaoyao Li

TDX 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 loading 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
index c8d9e71b889b..cf63434ba89d 100644
@@ -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
index f66a92f6c9cc..343a929e1825 100644
@@ -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"
@@ -1156,7 +1157,7 @@ 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
index 3877d432a397..587dbeeed196 100644
@@ -11,3 +11,8 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
 {
     return -EINVAL;
 }
+
+int tdx_parse_tdvf(void *flash_ptr, int size)
+{
+    return -EINVAL;
+}
target/i386/kvm/tdx.c +5 -0
index f892551821ce..c743340c145c 100644
@@ -622,6 +622,11 @@ int tdx_pre_create_vcpu(CPUState *cpu, Error **errp)
     return 0;
 }
 
+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
index 6e39ef3bac13..a46af433135f 100644
@@ -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 {
     char *mrconfigid;       /* base64 encoded sha348 digest */
     char *mrowner;          /* base64 encoded sha348 digest */
     char *mrownerconfig;    /* base64 encoded 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, Error **errp);
+int tdx_parse_tdvf(void *flash_ptr, int size);
 
 #endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
author
#51 lore raw

PATCH v4 38/66 i386/tdx: Skip BIOS shadowing setup

In reply to Xiaoyao Li

TDX 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 in v1:
- 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
index 343a929e1825..f7352b06c3e6 100644
@@ -1179,17 +1179,20 @@ 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
author
#52 lore raw

PATCH v4 39/66 i386/tdx: Don't initialize pc.rom for TDX VMs

In reply to Xiaoyao Li

For 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
index 5c80b5fe8a0e..f38694902764 100644
@@ -43,6 +43,7 @@
 #include "sysemu/xen.h"
 #include "sysemu/reset.h"
 #include "kvm/kvm_i386.h"
+#include "kvm/tdx.h"
 #include "hw/xen/xen.h"
 #include "qapi/qmp/qlist.h"
 #include "qemu/error-report.h"
@@ -1039,16 +1040,18 @@ 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
author
#53 lore raw

PATCH v4 40/66 i386/tdx: Track mem_ptr for each firmware entry of TDVF

In reply to Xiaoyao Li

For 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
index ff51f40088f0..0a6445705160 100644
@@ -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
index 593341eb2e93..d880af245a73 100644
@@ -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
index c743340c145c..825a432255ff 100644
@@ -14,6 +14,7 @@
 #include "qemu/osdep.h"
 #include "qemu/error-report.h"
 #include "qemu/base64.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/address-spaces.h"
 
 #include "hw/i386/x86.h"
+#include "hw/i386/tdvf.h"
 #include "kvm_i386.h"
 #include "tdx.h"
 #include "../cpu-internal.h"
@@ -464,6 +466,33 @@ 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 = TDX_GUEST(OBJECT(ms->cgs));
@@ -489,6 +518,8 @@ 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
author
#54 lore raw

PATCH v4 41/66 i386/tdx: Track RAM entries for TDX VM

In reply to 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_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 in v3:
- use enum TdxRamType in struct TdxRamEntry; (Isaku)
- Fix the indention; (Daniel)

Changes in v1:
- 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 | 111 ++++++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.h |  14 ++++++
 2 files changed, 125 insertions(+)
target/i386/kvm/tdx.c +111 -0
index 825a432255ff..c8c6ed9f811b 100644
@@ -22,6 +22,7 @@
 #include "sysemu/sysemu.h"
 #include "exec/address-spaces.h"
 
+#include "hw/i386/e820_memory_layout.h"
 #include "hw/i386/x86.h"
 #include "hw/i386/tdvf.h"
 #include "kvm_i386.h"
@@ -466,11 +467,117 @@ 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,
+                              enum TdxRamType 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:
@@ -481,12 +588,16 @@ 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
index a46af433135f..3a35a2bc0900 100644
@@ -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;
+    enum TdxRamType type;
+} TdxRamEntry;
+
 typedef struct TdxGuest {
     ConfidentialGuestSupport parent_obj;
 
@@ -27,6 +38,9 @@ typedef struct TdxGuest {
     char *mrownerconfig;    /* base64 encoded sha348 digest */
 
     TdxFirmware tdvf;
+
+    uint32_t nr_ram_entries;
+    TdxRamEntry *ram_entries;
 } TdxGuest;
 
 #ifdef CONFIG_TDX
Signature · 1 line
2.34.1
author
#55 lore raw

PATCH v4 42/66 headers: Add definitions from UEFI spec for volumes, resources, etc...

In reply to 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 [1], PI spec [2] and TDVF design guide[3].

[1] UEFI Specification v2.1.0 https://uefi.org/sites/default/files/resources/UEFI_Spec_2_10_Aug29.pdf
[2] UEFI PI spec v1.8 https://uefi.org/sites/default/files/resources/UEFI_PI_Spec_1_8_March3.pdf
[3] https://software.intel.com/content/dam/develop/external/us/en/documents/tdx-virtual-firmware-design-guide-rev-1.pdf

---
 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
new file mode 100644
index 000000000000..b15aba796156
@@ -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
author
#56 lore raw

PATCH v4 43/66 i386/tdx: Setup the TD HOB list

In reply to Xiaoyao Li

The 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 in v1:
- 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
index 6808bd4e3032..118fd9dae610 100644
@@ -27,7 +27,7 @@ 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
new file mode 100644
index 000000000000..1b737e946a8d
@@ -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
index c8c6ed9f811b..7f84aa0e64a9 100644
@@ -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"
@@ -467,6 +468,19 @@ 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,
                               enum TdxRamType type)
 {
@@ -598,6 +612,8 @@ 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
author
#57 lore raw

PATCH v4 44/66 i386/tdx: Add TDVF memory via KVM_TDX_INIT_MEM_REGION

In reply to Xiaoyao Li

From: 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 in v1:
- rename variable @metadata to @flags

---
 target/i386/kvm/tdx.c | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)
target/i386/kvm/tdx.c +24 -0
index 7f84aa0e64a9..af4107514fc9 100644
@@ -589,6 +589,7 @@ static void tdx_finalize_vm(Notifier *notifier, void *unused)
 {
     TdxFirmware *tdvf = &tdx_guest->tdvf;
     TdxFirmwareEntry *entry;
+    int r;
 
     tdx_init_ram_entries();
 
@@ -614,6 +615,29 @@ 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
author
#58 lore raw

PATCH v4 45/66 memory: Introduce memory_region_init_ram_guest_memfd()

In reply to Xiaoyao Li

Introduce memory_region_init_ram_guest_memfd() to allocate private guset memfd on the MemoryRegion initialization. It's for the use case of TDVF, which must be private on TDX case.

---
 include/exec/memory.h |  6 ++++++
 system/memory.c       | 27 +++++++++++++++++++++++++++
 2 files changed, 33 insertions(+)
include/exec/memory.h +6 -0
index f25959f6d30f..3a7f41b030e8 100644
@@ -1607,6 +1607,12 @@ bool memory_region_init_ram(MemoryRegion *mr,
                             uint64_t size,
                             Error **errp);
 
+bool memory_region_init_ram_guest_memfd(MemoryRegion *mr,
+                                        Object *owner,
+                                        const char *name,
+                                        uint64_t size,
+                                        Error **errp);
+
 /**
  * memory_region_init_rom: Initialize a ROM memory region.
  *
system/memory.c +27 -0
index 74f647f2e56f..41049d3e4c9a 100644
@@ -3619,6 +3619,33 @@ bool memory_region_init_ram(MemoryRegion *mr,
     return true;
 }
 
+bool memory_region_init_ram_guest_memfd(MemoryRegion *mr,
+                                        Object *owner,
+                                        const char *name,
+                                        uint64_t size,
+                                        Error **errp)
+{
+    DeviceState *owner_dev;
+
+    if (!memory_region_init_ram_flags_nomigrate(mr, owner, name, size,
+                                                RAM_GUEST_MEMFD, errp)) {
+        return false;
+    }
+
+    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);
+
+    return true;
+}
+
 bool memory_region_init_rom(MemoryRegion *mr,
                             Object *owner,
                             const char *name,
Signature · 1 line
2.34.1
#59 lore raw

Re: [PATCH v4 45/66] memory: Introduce memory_region_init_ram_guest_memfd()

Quoted 39 lines · Xiaoyao Li

Introduce memory_region_init_ram_guest_memfd() to allocate private guset memfd on the MemoryRegion initialization. It's for the use case of TDVF, which must be private on TDX case.

---
  include/exec/memory.h |  6 ++++++
  system/memory.c       | 27 +++++++++++++++++++++++++++
  2 files changed, 33 insertions(+)
include/exec/memory.h +6 -0
index f25959f6d30f..3a7f41b030e8 100644
@@ -1607,6 +1607,12 @@ bool memory_region_init_ram(MemoryRegion *mr,
                              uint64_t size,
                              Error **errp);
 
+bool memory_region_init_ram_guest_memfd(MemoryRegion *mr,
+                                        Object *owner,
+                                        const char *name,
+                                        uint64_t size,
+                                        Error **errp);
+
  /**
   * memory_region_init_rom: Initialize a ROM memory region.
   *
system/memory.c +4 -0
index 74f647f2e56f..41049d3e4c9a 100644
@@ -3619,6 +3619,33 @@ bool memory_region_init_ram(MemoryRegion *mr,
      return true;
  }
 
+bool memory_region_init_ram_guest_memfd(MemoryRegion *mr,
+                                        Object *owner,
+                                        const char *name,
+                                        uint64_t size,

system/memory.c bool memory_region_init_ram(MemoryRegion *mr, · line 3635

Quoted 10 lines
system/memory.c +10 -0
+                                        Error **errp)
+{
+    DeviceState *owner_dev;
+
+    if (!memory_region_init_ram_flags_nomigrate(mr, owner, name, size,
+                                                RAM_GUEST_MEMFD, errp)) {
+        return false;
+    }
+
+    memory_region_set_default_private(mr);

It would be much cleaner to just pass the flag when initializing the memory region.

Ideally, we can get rid of memory_region_set_default_private().

What you do in patch #33 feels like a layer violation.

Signature · 3 lines
Cheers,

David / dhildenb
author
#60 lore raw

PATCH v4 46/66 i386/tdx: register TDVF as private memory

In reply to Xiaoyao Li

From: Chao Peng <[email protected]>

Allocate private guest memfd memory for BIOS if it's TD VM.

---
 hw/i386/x86.c         | 10 +++++++++-
 target/i386/kvm/tdx.c | 18 ++++++++++++++++++
 target/i386/kvm/tdx.h |  2 ++
 3 files changed, 29 insertions(+), 1 deletion(-)
hw/i386/x86.c +9 -1
index f7352b06c3e6..f13f49069d40 100644
@@ -1155,8 +1155,16 @@ 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_guest_memfd(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 +18 -0
index af4107514fc9..0f6a6e9bb024 100644
@@ -21,6 +21,7 @@
 #include "sysemu/kvm.h"
 #include "sysemu/sysemu.h"
 #include "exec/address-spaces.h"
+#include "exec/ramblock.h"
 
 #include "hw/i386/e820_memory_layout.h"
 #include "hw/i386/x86.h"
@@ -468,6 +469,12 @@ 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;
@@ -589,6 +596,7 @@ 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();
@@ -623,6 +631,12 @@ 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;
 
@@ -638,6 +652,10 @@ 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
index 3a35a2bc0900..5fb20a5f06bb 100644
@@ -38,6 +38,7 @@ typedef struct TdxGuest {
     char *mrownerconfig;    /* base64 encoded 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, Error **errp);
+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
author
#61 lore raw

PATCH v4 47/66 i386/tdx: Call KVM_TDX_INIT_VCPU to initialize TDX vcpu

In reply to Xiaoyao Li

TDX 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
index 0f6a6e9bb024..341b4e76bf7d 100644
@@ -592,6 +592,22 @@ 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;
@@ -624,6 +640,8 @@ 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
author
#62 lore raw

PATCH v4 48/66 i386/tdx: Finalize TDX VM

In reply to Xiaoyao Li

Invoke 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
index 341b4e76bf7d..69a697608219 100644
@@ -674,6 +674,13 @@ 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
author
#63 lore raw

PATCH v4 49/66 i386/tdx: handle TDG.VP.VMCALL<SetupEventNotifyInterrupt>

In reply to Xiaoyao Li

From: 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      |  8 ++++++
 target/i386/kvm/tdx-stub.c |  5 ++++
 target/i386/kvm/tdx.c      | 53 ++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.h      | 14 ++++++++++
 4 files changed, 80 insertions(+)
target/i386/kvm/kvm.c +8 -0
index 348f76a9cb81..e36ece874246 100644
@@ -5426,6 +5426,14 @@ 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;
+        }
+        ret = tdx_handle_exit(cpu, &run->tdx);
+        break;
     default:
         fprintf(stderr, "KVM: unknown exit reason %d\n", run->exit_reason);
         ret = -1;
target/i386/kvm/tdx-stub.c +5 -0
index 587dbeeed196..ffaf659bc9be 100644
@@ -16,3 +16,8 @@ int tdx_parse_tdvf(void *flash_ptr, int size)
 {
     return -EINVAL;
 }
+
+int tdx_handle_exit(X86CPU *cpu, struct kvm_tdx_exit *tdx_exit)
+{
+    return -EINVAL;
+}
target/i386/kvm/tdx.c +53 -0

65 lines not on this page · raw message

target/i386/kvm/tdx.h +14 -0
index 5fb20a5f06bb..992916e4c905 100644
@@ -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)
@@ -15,6 +16,14 @@ typedef struct TdxGuestClass {
     ConfidentialGuestSupportClass parent_class;
 } TdxGuestClass;
 
+#define TDG_VP_VMCALL_SETUP_EVENT_NOTIFY_INTERRUPT      0x10004ULL
+
+#define TDG_VP_VMCALL_SUCCESS           0x0000000000000000ULL
+#define TDG_VP_VMCALL_RETRY             0x0000000000000001ULL
+#define TDG_VP_VMCALL_INVALID_OPERAND   0x8000000000000000ULL
+#define TDG_VP_VMCALL_GPA_INUSE         0x8000000000000001ULL
+#define TDG_VP_VMCALL_ALIGN_ERROR       0x8000000000000002ULL
+
 enum TdxRamType{
     TDX_RAM_UNACCEPTED,
     TDX_RAM_ADDED,
@@ -42,6 +51,10 @@ typedef struct TdxGuest {
 
     uint32_t nr_ram_entries;
     TdxRamEntry *ram_entries;
+
+    /* runtime state */
+    uint32_t event_notify_vector;
+    uint32_t event_notify_apicid;
 } TdxGuest;
 
 #ifdef CONFIG_TDX
@@ -56,5 +69,6 @@ void tdx_get_supported_cpuid(uint32_t function, uint32_t index, int reg,
 int tdx_pre_create_vcpu(CPUState *cpu, Error **errp);
 void tdx_set_tdvf_region(MemoryRegion *tdvf_region);
 int tdx_parse_tdvf(void *flash_ptr, int size);
+int tdx_handle_exit(X86CPU *cpu, struct kvm_tdx_exit *tdx_exit);
 
 #endif /* QEMU_I386_TDX_H */
Signature · 1 line
2.34.1
author
#64 lore raw

PATCH v4 50/66 i386/tdx: handle TDG.VP.VMCALL<GetQuote>

In reply to Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Add property "quote-generation-socket" to tdx-guest, which is a property of type SocketAddress to specify Quote Generation Service(QGS).

On request of GetQuote, it connects to the QGS socket, read request data from shared guest memory, send the request data to the QGS, and store the response into shared guest memory, at last notify TD guest by interrupt.

command line example:

  qemu-system-x86_64 \
    -object '{"qom-type":"tdx-guest","id":"tdx0","quote-generation-socket":{"type": "vsock", "cid":"1","port":"1234"}}' \
    -machine confidential-guest-support=tdx0

Note, above example uses vsock type socket because the QGS we used implements the vsock socket. It can be other types, like UNIX socket, which depends on the implementation of QGS.

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.

---
Changes in v4:
- merge next patch "i386/tdx: setup a timer for the qio channel";

Changes in v3:
- rename property "quote-generation-service" to "quote-generation-socket";
- change the type of "quote-generation-socket" from str to SocketAddress;
- squash next patch into this one;

---
 qapi/qom.json                         |   6 +-
 target/i386/kvm/meson.build           |   2 +-
 target/i386/kvm/tdx-quote-generator.c | 170 ++++++++++++++++++++
 target/i386/kvm/tdx-quote-generator.h |  95 +++++++++++
 target/i386/kvm/tdx.c                 | 216 ++++++++++++++++++++++++++
 target/i386/kvm/tdx.h                 |   6 +
 6 files changed, 493 insertions(+), 2 deletions(-)
 create mode 100644 target/i386/kvm/tdx-quote-generator.c
 create mode 100644 target/i386/kvm/tdx-quote-generator.h
qapi/qom.json +5 -1
index 15445f9e41fc..c60fb5710961 100644
@@ -914,13 +914,17 @@
 #     e.g., specific to the workload rather than the run-time or OS.
 #     base64 encoded SHA384 digest.
 #
+# @quote-generation-socket: socket address for Quote Generation
+#     Service(QGS)
+#
 # Since: 9.0
 ##
 { 'struct': 'TdxGuestProperties',
   'data': { '*sept-ve-disable': 'bool',
             '*mrconfigid': 'str',
             '*mrowner': 'str',
-            '*mrownerconfig': 'str' } }
+            '*mrownerconfig': 'str',
+            '*quote-generation-socket': 'SocketAddress' } }
 
 ##
 # @ThreadContextProperties:
target/i386/kvm/meson.build +1 -1
index 30a90b4d371d..2fd7c27b76c4 100644
@@ -9,7 +9,7 @@ i386_kvm_ss.add(when: 'CONFIG_XEN_EMU', if_true: files('xen-emu.c'))
 
 i386_kvm_ss.add(when: 'CONFIG_SEV', if_false: files('sev-stub.c'))
 
-i386_kvm_ss.add(when: 'CONFIG_TDX', if_true: files('tdx.c'), if_false: files('tdx-stub.c'))
+i386_kvm_ss.add(when: 'CONFIG_TDX', if_true: files('tdx.c', 'tdx-quote-generator.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-quote-generator.c new file +170 -0

170 lines not on this page · raw message

target/i386/kvm/tdx-quote-generator.h new file +95 -0

95 lines not on this page · raw message

target/i386/kvm/tdx.c +216 -0

258 lines not on this page · raw message

target/i386/kvm/tdx.h +6 -0
index 992916e4c905..c6e4c275262d 100644
@@ -9,6 +9,8 @@
 #include "hw/i386/tdvf.h"
 #include "sysemu/kvm.h"
 
+#include "tdx-quote-generator.h"
+
 #define TYPE_TDX_GUEST "tdx-guest"
 #define TDX_GUEST(obj)  OBJECT_CHECK(TdxGuest, (obj), TYPE_TDX_GUEST)
 
@@ -16,6 +18,7 @@ typedef struct TdxGuestClass {
     ConfidentialGuestSupportClass parent_class;
 } TdxGuestClass;
 
+#define TDG_VP_VMCALL_GET_QUOTE                         0x10002ULL
 #define TDG_VP_VMCALL_SETUP_EVENT_NOTIFY_INTERRUPT      0x10004ULL
 
 #define TDG_VP_VMCALL_SUCCESS           0x0000000000000000ULL
@@ -55,6 +58,9 @@ typedef struct TdxGuest {
     /* runtime state */
     uint32_t event_notify_vector;
     uint32_t event_notify_apicid;
+
+    /* GetQuote */
+    TdxQuoteGenerator *quote_generator;
 } TdxGuest;
 
 #ifdef CONFIG_TDX
Signature · 1 line
2.34.1

A message that is not in the archive (sent off-list or not yet received): [email protected]

#66 lore raw

Re: [PATCH v4 50/66] i386/tdx: handle TDG.VP.VMCALL<GetQuote>

Quoted 63 lines · Markus Armbruster
Quoted 53 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Add property "quote-generation-socket" to tdx-guest, which is a property of type SocketAddress to specify Quote Generation Service(QGS).

On request of GetQuote, it connects to the QGS socket, read request data from shared guest memory, send the request data to the QGS, and store the response into shared guest memory, at last notify TD guest by interrupt.

command line example:

  qemu-system-x86_64 \
    -object '{"qom-type":"tdx-guest","id":"tdx0","quote-generation-socket":{"type": "vsock", "cid":"1","port":"1234"}}' \
    -machine confidential-guest-support=tdx0

Note, above example uses vsock type socket because the QGS we used implements the vsock socket. It can be other types, like UNIX socket, which depends on the implementation of QGS.

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.

---
Changes in v4:
- merge next patch "i386/tdx: setup a timer for the qio channel";

Changes in v3:
- rename property "quote-generation-service" to "quote-generation-socket";
- change the type of "quote-generation-socket" from str to SocketAddress;
- squash next patch into this one;

---
 qapi/qom.json                         |   6 +-
 target/i386/kvm/meson.build           |   2 +-
 target/i386/kvm/tdx-quote-generator.c | 170 ++++++++++++++++++++
 target/i386/kvm/tdx-quote-generator.h |  95 +++++++++++
 target/i386/kvm/tdx.c                 | 216 ++++++++++++++++++++++++++
 target/i386/kvm/tdx.h                 |   6 +
 6 files changed, 493 insertions(+), 2 deletions(-)
 create mode 100644 target/i386/kvm/tdx-quote-generator.c
 create mode 100644 target/i386/kvm/tdx-quote-generator.h
qapi/qom.json +0 -0
index 15445f9e41fc..c60fb5710961 100644

qapi/qom.json line 918

Quoted 5 lines
qapi/qom.json +2 -0
@@ -914,13 +914,17 @@
 #     e.g., specific to the workload rather than the run-time or OS.
 #     base64 encoded SHA384 digest.
 #
+# @quote-generation-socket: socket address for Quote Generation
+#     Service(QGS)

Space between "Service" and "(QGS)", please.

Quoted 2 lines

The description feels too terse. What is the "Quote Generation Service", and why should I care?

The "Quote Generation Service" is a daemon running on the host. The reference implementation is at

https://github.com/intel/SGXDataCenterAttestationPrimitives/tree/master/QuoteGeneration/quote_wrapper/qgs

If you don't provide this, then quests won't bet able to generate quotes needed for attestation. So although this is technically optional, in practice for a sane deployment, an admin should always provide this

Quoted 12 lines
Quoted 12 lines
qapi/qom.json +3 -1
+#
 # Since: 9.0
 ##
 { 'struct': 'TdxGuestProperties',
   'data': { '*sept-ve-disable': 'bool',
             '*mrconfigid': 'str',
             '*mrowner': 'str',
-            '*mrownerconfig': 'str' } }
+            '*mrownerconfig': 'str',
+            '*quote-generation-socket': 'SocketAddress' } }
 
Quoted 5 lines
Quoted 2 lines
qapi/qom.json +0 -0
 ##
 # @ThreadContextProperties:

[...]

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 :|
#67 lore raw

Re: [PATCH v4 50/66] i386/tdx: handle TDG.VP.VMCALL<GetQuote>

Quoted 70 lines · Daniel P. Berrangé
Quoted 63 lines · Markus Armbruster
Quoted 53 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Add property "quote-generation-socket" to tdx-guest, which is a property of type SocketAddress to specify Quote Generation Service(QGS).

On request of GetQuote, it connects to the QGS socket, read request data from shared guest memory, send the request data to the QGS, and store the response into shared guest memory, at last notify TD guest by interrupt.

command line example:

  qemu-system-x86_64 \
    -object '{"qom-type":"tdx-guest","id":"tdx0","quote-generation-socket":{"type": "vsock", "cid":"1","port":"1234"}}' \
    -machine confidential-guest-support=tdx0

Note, above example uses vsock type socket because the QGS we used implements the vsock socket. It can be other types, like UNIX socket, which depends on the implementation of QGS.

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.

---
Changes in v4:
- merge next patch "i386/tdx: setup a timer for the qio channel";

Changes in v3:
- rename property "quote-generation-service" to "quote-generation-socket";
- change the type of "quote-generation-socket" from str to SocketAddress;
- squash next patch into this one;

---
 qapi/qom.json                         |   6 +-
 target/i386/kvm/meson.build           |   2 +-
 target/i386/kvm/tdx-quote-generator.c | 170 ++++++++++++++++++++
 target/i386/kvm/tdx-quote-generator.h |  95 +++++++++++
 target/i386/kvm/tdx.c                 | 216 ++++++++++++++++++++++++++
 target/i386/kvm/tdx.h                 |   6 +
 6 files changed, 493 insertions(+), 2 deletions(-)
 create mode 100644 target/i386/kvm/tdx-quote-generator.c
 create mode 100644 target/i386/kvm/tdx-quote-generator.h
qapi/qom.json +0 -0
index 15445f9e41fc..c60fb5710961 100644

qapi/qom.json line 918

Quoted 5 lines
qapi/qom.json +2 -0
@@ -914,13 +914,17 @@
 #     e.g., specific to the workload rather than the run-time or OS.
 #     base64 encoded SHA384 digest.
 #
+# @quote-generation-socket: socket address for Quote Generation
+#     Service(QGS)

Space between "Service" and "(QGS)", please.

Quoted 2 lines

The description feels too terse. What is the "Quote Generation Service", and why should I care?

The "Quote Generation Service" is a daemon running on the host. The reference implementation is at

Quoted 6 lines

https://github.com/intel/SGXDataCenterAttestationPrimitives/tree/master/QuoteGeneration/quote_wrapper/qgs

If you don't provide this, then quests won't bet able to generate quotes needed for attestation. So although this is technically optional, in practice for a sane deployment, an admin should always provide this

Thanks. Care to work some of this information into the doc comment?

Quoted 19 lines
Quoted 11 lines
Quoted 11 lines
qapi/qom.json +3 -1
+#
 # Since: 9.0
 ##
 { 'struct': 'TdxGuestProperties',
   'data': { '*sept-ve-disable': 'bool',
             '*mrconfigid': 'str',
             '*mrowner': 'str',
-            '*mrownerconfig': 'str' } }
+            '*mrownerconfig': 'str',
+            '*quote-generation-socket': 'SocketAddress' } }
 
Quoted 5 lines
Quoted 2 lines
qapi/qom.json +0 -0
 ##
 # @ThreadContextProperties:

[...]

With regards,
Daniel

author
#68 lore raw

Re: [PATCH v4 50/66] i386/tdx: handle TDG.VP.VMCALL<GetQuote>

Quoted 72 lines · Markus Armbruster
Quoted 70 lines · Daniel P. Berrangé
Quoted 63 lines · Markus Armbruster
Quoted 53 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Add property "quote-generation-socket" to tdx-guest, which is a property of type SocketAddress to specify Quote Generation Service(QGS).

On request of GetQuote, it connects to the QGS socket, read request data from shared guest memory, send the request data to the QGS, and store the response into shared guest memory, at last notify TD guest by interrupt.

command line example:

   qemu-system-x86_64 \
     -object '{"qom-type":"tdx-guest","id":"tdx0","quote-generation-socket":{"type": "vsock", "cid":"1","port":"1234"}}' \
     -machine confidential-guest-support=tdx0

Note, above example uses vsock type socket because the QGS we used implements the vsock socket. It can be other types, like UNIX socket, which depends on the implementation of QGS.

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.

---
Changes in v4:
- merge next patch "i386/tdx: setup a timer for the qio channel";

Changes in v3:
- rename property "quote-generation-service" to "quote-generation-socket";
- change the type of "quote-generation-socket" from str to SocketAddress;
- squash next patch into this one;

---
  qapi/qom.json                         |   6 +-
  target/i386/kvm/meson.build           |   2 +-
  target/i386/kvm/tdx-quote-generator.c | 170 ++++++++++++++++++++
  target/i386/kvm/tdx-quote-generator.h |  95 +++++++++++
  target/i386/kvm/tdx.c                 | 216 ++++++++++++++++++++++++++
  target/i386/kvm/tdx.h                 |   6 +
  6 files changed, 493 insertions(+), 2 deletions(-)

create mode 100644 target/i386/kvm/tdx-quote-generator.c create mode 100644 target/i386/kvm/tdx-quote-generator.h

qapi/qom.json +0 -0
index 15445f9e41fc..c60fb5710961 100644

qapi/qom.json line 918

Quoted 5 lines
qapi/qom.json +2 -0
@@ -914,13 +914,17 @@
  #     e.g., specific to the workload rather than the run-time or OS.
  #     base64 encoded SHA384 digest.
  #
+# @quote-generation-socket: socket address for Quote Generation
+#     Service(QGS)

Space between "Service" and "(QGS)", please.

Quoted 2 lines

The description feels too terse. What is the "Quote Generation Service", and why should I care?

The "Quote Generation Service" is a daemon running on the host. The reference implementation is at

Quoted 8 lines
Quoted 6 lines

https://github.com/intel/SGXDataCenterAttestationPrimitives/tree/master/QuoteGeneration/quote_wrapper/qgs

If you don't provide this, then quests won't bet able to generate quotes needed for attestation. So although this is technically optional, in practice for a sane deployment, an admin should always provide this

Thanks. Care to work some of this information into the doc comment?

Sure. Will add a new section of Attestation in the last patch.

Quoted 20 lines
Quoted 19 lines
Quoted 11 lines
Quoted 11 lines
qapi/qom.json +3 -1
+#
  # Since: 9.0
  ##
  { 'struct': 'TdxGuestProperties',
    'data': { '*sept-ve-disable': 'bool',
              '*mrconfigid': 'str',
              '*mrowner': 'str',
-            '*mrownerconfig': 'str' } }
+            '*mrownerconfig': 'str',
+            '*quote-generation-socket': 'SocketAddress' } }
 
Quoted 5 lines
Quoted 2 lines
qapi/qom.json +0 -0
  ##
  # @ThreadContextProperties:

[...]

With regards,
Daniel

#69 lore raw

Re: [PATCH v4 50/66] i386/tdx: handle TDG.VP.VMCALL<GetQuote>

In reply to Xiaoyao Li
Quoted 10 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Add property "quote-generation-socket" to tdx-guest, which is a property of type SocketAddress to specify Quote Generation Service(QGS).

On request of GetQuote, it connects to the QGS socket, read request data from shared guest memory, send the request data to the QGS, and store the response into shared guest memory, at last notify TD guest by interrupt.

Quoted 8 lines

command line example:

  qemu-system-x86_64 \
    -object '{"qom-type":"tdx-guest","id":"tdx0","quote-generation-socket":{"type": "vsock", "cid":"1","port":"1234"}}' \
    -machine confidential-guest-support=tdx0

Note, above example uses vsock type socket because the QGS we used implements the vsock socket. It can be other types, like UNIX socket, which depends on the implementation of QGS.

Can you confirm again exactly what QGS impl you are testing against ?

I've tried the impl at

https://github.com/intel/SGXDataCenterAttestationPrimitives/tree/master/QuoteGeneration/quote_wrapper/qgs

which supports UNIX sockets and VSOCK. In both cases, however, it appears to be speaking a different protocol than your QEMU impl below uses.

Specifically here:

https://github.com/intel/SGXDataCenterAttestationPrimitives/blob/master/QuoteGeneration/quote_wrapper/qgs/qgs_server.cpp#L143

it is reading 4 bytes of header, which are interpreted as the length of the payload which will then be read off the wire. IIUC the payload it expects is the TDREPORT struct.

Your QEMU patches here meanwhile are just sending the payload from the GetQuote hypercall which is the TDREPORT struct.

IOW, QEMU is not sending the 4 byte length header the QGS expects. and whole thing fails.

Quoted 11 lines

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.

---

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 :|
author
#70 lore raw

Re: [PATCH v4 50/66] i386/tdx: handle TDG.VP.VMCALL<GetQuote>

+ Feng Qiu,

Quoted 37 lines · Daniel P. Berrangé
Quoted 10 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Add property "quote-generation-socket" to tdx-guest, which is a property of type SocketAddress to specify Quote Generation Service(QGS).

On request of GetQuote, it connects to the QGS socket, read request data from shared guest memory, send the request data to the QGS, and store the response into shared guest memory, at last notify TD guest by interrupt.

Quoted 8 lines

command line example:

   qemu-system-x86_64 \
     -object '{"qom-type":"tdx-guest","id":"tdx0","quote-generation-socket":{"type": "vsock", "cid":"1","port":"1234"}}' \
     -machine confidential-guest-support=tdx0

Note, above example uses vsock type socket because the QGS we used implements the vsock socket. It can be other types, like UNIX socket, which depends on the implementation of QGS.

Can you confirm again exactly what QGS impl you are testing against ?

Quoted 1 line

I've tried the impl at

https://github.com/intel/SGXDataCenterAttestationPrimitives/tree/master/QuoteGeneration/quote_wrapper/qgs

which supports UNIX sockets and VSOCK. In both cases, however, it appears to be speaking a different protocol than your QEMU impl below uses.

Specifically here:

https://github.com/intel/SGXDataCenterAttestationPrimitives/blob/master/QuoteGeneration/quote_wrapper/qgs/qgs_server.cpp#L143

it is reading 4 bytes of header, which are interpreted as the length of the payload which will then be read off the wire. IIUC the payload it expects is the TDREPORT struct.

Quoted 5 lines

Your QEMU patches here meanwhile are just sending the payload from the GetQuote hypercall which is the TDREPORT struct.

IOW, QEMU is not sending the 4 byte length header the QGS expects. and whole thing fails.

I'm using the one provided by internal folks, which supports interpreting the payload without the header.

I don't know when will the updated implementation show up in public github. @Feng Liu can help on it.

Quoted 14 lines
Quoted 11 lines

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.

---

With regards,
Daniel

#71 lore raw

Re: [PATCH v4 50/66] i386/tdx: handle TDG.VP.VMCALL<GetQuote>

Actually the 4 byte length header is provided by client library(https://github.com/intel/SGXDataCenterAttestationPrimitives/blob/master/QuoteGeneration/quote_wrapper/tdx_attest/tdx_attest.c#L295), not QEMU. QEMUjust treats the how payload including the header a whole blob.
BTW, in the latest stable kernel, the TDX guest driver changed to TSM based solution(https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/drivers/virt/coco/tdx-guest/tdx-guest.c?h=v6.7.5) and it will only send raw report without 4 byte length header and other stuff. Existing official QGS doesn't compatible with this change and we will deliver compatible QGS in the end of Q1.

Quoted 67 lines · Xiaoyao Li

+ Feng Qiu,

Quoted 39 lines · Daniel P. Berrangé
Quoted 10 lines · Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Add property "quote-generation-socket" to tdx-guest, which is a property of type SocketAddress to specify Quote Generation Service(QGS).

On request of GetQuote, it connects to the QGS socket, read request data from shared guest memory, send the request data to the QGS, and store the response into shared guest memory, at last notify TD guest by interrupt.

Quoted 8 lines

command line example:
   qemu-system-x86_64 \
     -object '{"qom-type":"tdx-guest","id":"tdx0","quote-generation-socket":{"type": "vsock", "cid":"1","port":"1234"}}' \
     -machine confidential-guest-support=tdx0

Note, above example uses vsock type socket because the QGS we used implements the vsock socket. It can be other types, like UNIX socket, which depends on the implementation of QGS.

Can you confirm again exactly what QGS impl you are testing against ?

Quoted 1 line

I've tried the impl at

https://github.com/intel/SGXDataCenterAttestationPrimitives/tree/master/QuoteGeneration/quote_wrapper/qgs

which supports UNIX sockets and VSOCK. In both cases, however, it appears to be speaking a different protocol than your QEMU impl below uses.

Specifically here:

https://github.com/intel/SGXDataCenterAttestationPrimitives/blob/master/QuoteGeneration/quote_wrapper/qgs/qgs_server.cpp#L143

it is reading 4 bytes of header, which are interpreted as the length of the payload which will then be read off the wire. IIUC the payload it expects is the TDREPORT struct.

Quoted 5 lines

Your QEMU patches here meanwhile are just sending the payload from the GetQuote hypercall which is the TDREPORT struct.

IOW, QEMU is not sending the 4 byte length header the QGS expects. and whole thing fails.

I'm using the one provided by internal folks, which supports interpreting the payload without the header.

I don't know when will the updated implementation show up in public github. @Feng Liu can help on it.

Quoted 14 lines
Quoted 11 lines

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.

---

With regards,
Daniel

author
#72 lore raw

PATCH v4 51/66 i386/tdx: handle TDG.VP.VMCALL<MapGPA> hypercall

In reply to Xiaoyao Li

From: 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 | 54 +++++++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.h |  1 +
 4 files changed, 58 insertions(+), 1 deletion(-)
accel/kvm/kvm-all.c +1 -1
index eb8b3925dbe1..04309a8623d9 100644
@@ -2920,7 +2920,7 @@ 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;
     ram_addr_t offset;
include/sysemu/kvm.h +2 -0
index 2f6592859ac6..e0061848b053 100644
@@ -544,4 +544,6 @@ int kvm_create_guest_memfd(uint64_t size, uint64_t flags, Error **errp);
 
 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 +54 -0

66 lines not on this page · raw message

target/i386/kvm/tdx.h +1 -0
index c6e4c275262d..cedc7d7e226f 100644
@@ -18,6 +18,7 @@ typedef struct TdxGuestClass {
     ConfidentialGuestSupportClass parent_class;
 } TdxGuestClass;
 
+#define TDG_VP_VMCALL_MAP_GPA                           0x10001ULL
 #define TDG_VP_VMCALL_GET_QUOTE                         0x10002ULL
 #define TDG_VP_VMCALL_SETUP_EVENT_NOTIFY_INTERRUPT      0x10004ULL
 
Signature · 1 line
2.34.1
author
#73 lore raw

PATCH v4 52/66 i386/tdx: Handle TDG.VP.VMCALL<REPORT_FATAL_ERROR>

In reply to Xiaoyao Li

TD guest can use TDG.VP.VMCALL<REPORT_FATAL_ERROR> to request termination with error message encoded in GPRs.

Parse and print the error message, and terminate the TD guest in the handler.

---
 target/i386/kvm/tdx.c | 39 +++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.h |  1 +
 2 files changed, 40 insertions(+)
target/i386/kvm/tdx.c +39 -0
index f1c60274c448..1c79032ca262 100644
@@ -1078,6 +1078,43 @@ static int tdx_handle_get_quote(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall)
     return 0;
 }
 
+static int 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("TDX: REPORT_FATAL_ERROR: invalid error code: "
+                     "0x%lx\n", error_code);
+        return -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 ? : "");
+    return -1;
+}
+
 static int tdx_handle_setup_event_notify_interrupt(X86CPU *cpu,
                                                    struct kvm_tdx_vmcall *vmcall)
 {
@@ -1112,6 +1149,8 @@ static int tdx_handle_vmcall(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall)
         return tdx_handle_map_gpa(cpu, vmcall);
     case TDG_VP_VMCALL_GET_QUOTE:
         return tdx_handle_get_quote(cpu, vmcall);
+    case TDG_VP_VMCALL_REPORT_FATAL_ERROR:
+        return  tdx_handle_report_fatal_error(cpu, vmcall);
     case TDG_VP_VMCALL_SETUP_EVENT_NOTIFY_INTERRUPT:
         return tdx_handle_setup_event_notify_interrupt(cpu, vmcall);
     default:
target/i386/kvm/tdx.h +1 -0
index cedc7d7e226f..37fdd845f462 100644
@@ -20,6 +20,7 @@ typedef struct TdxGuestClass {
 
 #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
Signature · 1 line
2.34.1
author
#74 lore raw

PATCH v4 53/66 i386/tdx: Wire TDX_REPORT_FATAL_ERROR with GuestPanic facility

In reply to Xiaoyao Li

Integrate TDX's TDX_REPORT_FATAL_ERROR into QEMU GuestPanic facility

Originated-from: Isaku Yamahata <[email protected]>

---
Changes in v4:
- refine the documentation; (Markus)

Changes in v3:
- Add docmentation of new type and struct; (Daniel)
- refine the error message handling; (Daniel)

---
 qapi/run-state.json   | 28 ++++++++++++++++++++--
 system/runstate.c     | 54 +++++++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.c | 24 ++++++++++++++++++-
 3 files changed, 103 insertions(+), 3 deletions(-)
qapi/run-state.json +26 -2
index 08bc99cb8561..5429116679e3 100644
@@ -485,10 +485,12 @@
 #
 # @s390: s390 guest panic information type (Since: 2.12)
 #
+# @tdx: tdx guest panic information type (Since: 8.2)
+#
 # Since: 2.9
 ##
 { 'enum': 'GuestPanicInformationType',
-  'data': [ 'hyper-v', 's390' ] }
+  'data': [ 'hyper-v', 's390', 'tdx' ] }
 
 ##
 # @GuestPanicInformation:
@@ -503,7 +505,8 @@
  'base': {'type': 'GuestPanicInformationType'},
  'discriminator': 'type',
  'data': {'hyper-v': 'GuestPanicInformationHyperV',
-          's390': 'GuestPanicInformationS390'}}
+          's390': 'GuestPanicInformationS390',
+          'tdx' : 'GuestPanicInformationTdx'}}
 
 ##
 # @GuestPanicInformationHyperV:
@@ -566,6 +569,27 @@
           'psw-addr': 'uint64',
           'reason': 'S390CrashReason'}}
 
+##
+# @GuestPanicInformationTdx:
+#
+# TDX Guest panic information specific to TDX GCHI
+# TDG.VP.VMCALL<ReportFatalError>.
+#
+# @error-code: TD-specific error code
+#
+# @gpa: guest-physical address of a page that contains additional
+#     error data, in forms of zero-terminated string.
+#
+# @message: Human-readable error message provided by the guest. Not
+#     to be trusted.
+#
+# Since: 9.0
+##
+{'struct': 'GuestPanicInformationTdx',
+ 'data': {'error-code': 'uint64',
+          'gpa': 'uint64',
+          'message': 'str'}}
+
 ##
 # @MEMORY_FAILURE:
 #
system/runstate.c +54 -0

67 lines not on this page · raw message

target/i386/kvm/tdx.c +23 -1
index 1c79032ca262..4fbb18135951 100644
@@ -20,6 +20,7 @@
 #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"
@@ -1078,11 +1079,26 @@ static int tdx_handle_get_quote(X86CPU *cpu, struct kvm_tdx_vmcall *vmcall)
     return 0;
 }
 
+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 = message;
+
+    qemu_system_guest_panicked(panic_info);
+}
+
 static int 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("TDX: REPORT_FATAL_ERROR: invalid error code: "
@@ -1111,7 +1127,13 @@ static int tdx_handle_report_fatal_error(X86CPU *cpu,
         assert((char *)tmp == message + GUEST_PANIC_INFO_TDX_MESSAGE_MAX);
     }
 
-    error_report("TD guest reports fatal error. %s\n", message ? : "");
+#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);
+
     return -1;
 }
 
Signature · 1 line
2.34.1
#75 lore raw

Re: [PATCH v4 53/66] i386/tdx: Wire TDX_REPORT_FATAL_ERROR with GuestPanic facility

Quoted 46 lines · Xiaoyao Li

Integrate TDX's TDX_REPORT_FATAL_ERROR into QEMU GuestPanic facility

Originated-from: Isaku Yamahata <[email protected]>

---
Changes in v4:
- refine the documentation; (Markus)

Changes in v3:
- Add docmentation of new type and struct; (Daniel)
- refine the error message handling; (Daniel)

---
 qapi/run-state.json   | 28 ++++++++++++++++++++--
 system/runstate.c     | 54 +++++++++++++++++++++++++++++++++++++++++++
 target/i386/kvm/tdx.c | 24 ++++++++++++++++++-
 3 files changed, 103 insertions(+), 3 deletions(-)
qapi/run-state.json +5 -2
index 08bc99cb8561..5429116679e3 100644
@@ -485,10 +485,12 @@
 #
 # @s390: s390 guest panic information type (Since: 2.12)
 #
+# @tdx: tdx guest panic information type (Since: 8.2)
+#
 # Since: 2.9
 ##
 { 'enum': 'GuestPanicInformationType',
-  'data': [ 'hyper-v', 's390' ] }
+  'data': [ 'hyper-v', 's390', 'tdx' ] }
 
 ##
 # @GuestPanicInformation:
@@ -503,7 +505,8 @@
  'base': {'type': 'GuestPanicInformationType'},
  'discriminator': 'type',
  'data': {'hyper-v': 'GuestPanicInformationHyperV',
-          's390': 'GuestPanicInformationS390'}}
+          's390': 'GuestPanicInformationS390',
+          'tdx' : 'GuestPanicInformationTdx'}}
 
 ##
 # @GuestPanicInformationHyperV:

qapi/run-state.json line 578

Quoted 10 lines
qapi/run-state.json +7 -0
@@ -566,6 +569,27 @@
           'psw-addr': 'uint64',
           'reason': 'S390CrashReason'}}
 
+##
+# @GuestPanicInformationTdx:
+#
+# TDX Guest panic information specific to TDX GCHI
+# TDG.VP.VMCALL<ReportFatalError>.
+#
+# @error-code: TD-specific error code

Where could a user find information on these error codes?

qapi/run-state.json line 581

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @gpa: guest-physical address of a page that contains additional
+#     error data, in forms of zero-terminated string.

"in the form of a zero-terminated string"

qapi/run-state.json line 584

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @message: Human-readable error message provided by the guest. Not
+#     to be trusted.

How is this message related to the one pointed to by @gpa?

Quoted 11 lines
qapi/run-state.json +8 -0
+#
+# Since: 9.0
+##
+{'struct': 'GuestPanicInformationTdx',
+ 'data': {'error-code': 'uint64',
+          'gpa': 'uint64',
+          'message': 'str'}}
+
 ##
 # @MEMORY_FAILURE:
 #

[...]

author
#76 lore raw

Re: [PATCH v4 53/66] i386/tdx: Wire TDX_REPORT_FATAL_ERROR with GuestPanic facility

Quoted 51 lines · Markus Armbruster
Quoted 49 lines · Xiaoyao Li

Integrate TDX's TDX_REPORT_FATAL_ERROR into QEMU GuestPanic facility

Originated-from: Isaku Yamahata <[email protected]>

---
Changes in v4:
- refine the documentation; (Markus)

Changes in v3:
- Add docmentation of new type and struct; (Daniel)
- refine the error message handling; (Daniel)

---
  qapi/run-state.json   | 28 ++++++++++++++++++++--
  system/runstate.c     | 54 +++++++++++++++++++++++++++++++++++++++++++
  target/i386/kvm/tdx.c | 24 ++++++++++++++++++-
  3 files changed, 103 insertions(+), 3 deletions(-)
qapi/run-state.json +5 -2
index 08bc99cb8561..5429116679e3 100644
@@ -485,10 +485,12 @@
  #
  # @s390: s390 guest panic information type (Since: 2.12)
  #
+# @tdx: tdx guest panic information type (Since: 8.2)
+#
  # Since: 2.9
  ##
  { 'enum': 'GuestPanicInformationType',
-  'data': [ 'hyper-v', 's390' ] }
+  'data': [ 'hyper-v', 's390', 'tdx' ] }
 
  ##
  # @GuestPanicInformation:
@@ -503,7 +505,8 @@
   'base': {'type': 'GuestPanicInformationType'},
   'discriminator': 'type',
   'data': {'hyper-v': 'GuestPanicInformationHyperV',
-          's390': 'GuestPanicInformationS390'}}
+          's390': 'GuestPanicInformationS390',
+          'tdx' : 'GuestPanicInformationTdx'}}
 
  ##
  # @GuestPanicInformationHyperV:
@@ -566,6 +569,27 @@
            'psw-addr': 'uint64',
            'reason': 'S390CrashReason'}}
 
Quoted 9 lines

qapi/run-state.json line 578

Quoted 7 lines
qapi/run-state.json +7 -0
+##
+# @GuestPanicInformationTdx:
+#
+# TDX Guest panic information specific to TDX GCHI
+# TDG.VP.VMCALL<ReportFatalError>.
+#
+# @error-code: TD-specific error code

Where could a user find information on these error codes?

TDX GHCI (Guset-host-communication-Interface)spec. It defines all the TDVMCALL leaves.

0: panic;
0x1 - 0xffffffff: reserved.

Quoted 5 lines

qapi/run-state.json line 581

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @gpa: guest-physical address of a page that contains additional
+#     error data, in forms of zero-terminated string.

"in the form of a zero-terminated string"

fixed.

Quoted 5 lines

qapi/run-state.json line 584

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @message: Human-readable error message provided by the guest. Not
+#     to be trusted.

How is this message related to the one pointed to by @gpa?

In general, @message contains a brief message of the error. While @gpa (when valid) contains a verbose message.

The reason why we need both is because sometime when TD guest hits a fatal error, its memory may get corrupted so we cannot pass information via @gpa. Information in @message is passed through GPRs.

Quoted 14 lines
Quoted 11 lines
qapi/run-state.json +8 -0
+#
+# Since: 9.0
+##
+{'struct': 'GuestPanicInformationTdx',
+ 'data': {'error-code': 'uint64',
+          'gpa': 'uint64',
+          'message': 'str'}}
+
  ##
  # @MEMORY_FAILURE:
  #

[...]

#77 lore raw

Re: [PATCH v4 53/66] i386/tdx: Wire TDX_REPORT_FATAL_ERROR with GuestPanic facility

Quoted 62 lines · Xiaoyao Li
Quoted 51 lines · Markus Armbruster
Quoted 49 lines · Xiaoyao Li

Integrate TDX's TDX_REPORT_FATAL_ERROR into QEMU GuestPanic facility

Originated-from: Isaku Yamahata <[email protected]>

---
Changes in v4:
- refine the documentation; (Markus)

Changes in v3:
- Add docmentation of new type and struct; (Daniel)
- refine the error message handling; (Daniel)

---
  qapi/run-state.json   | 28 ++++++++++++++++++++--
  system/runstate.c     | 54 +++++++++++++++++++++++++++++++++++++++++++
  target/i386/kvm/tdx.c | 24 ++++++++++++++++++-
  3 files changed, 103 insertions(+), 3 deletions(-)
qapi/run-state.json +5 -2
index 08bc99cb8561..5429116679e3 100644
@@ -485,10 +485,12 @@
 #
 # @s390: s390 guest panic information type (Since: 2.12)
 #
+# @tdx: tdx guest panic information type (Since: 8.2)
+#
 # Since: 2.9
 ##
 { 'enum': 'GuestPanicInformationType',
-  'data': [ 'hyper-v', 's390' ] }
+  'data': [ 'hyper-v', 's390', 'tdx' ] }
 
 ##
 # @GuestPanicInformation:
@@ -503,7 +505,8 @@
   'base': {'type': 'GuestPanicInformationType'},
   'discriminator': 'type',
   'data': {'hyper-v': 'GuestPanicInformationHyperV',
-          's390': 'GuestPanicInformationS390'}}
+          's390': 'GuestPanicInformationS390',
+          'tdx' : 'GuestPanicInformationTdx'}}
 
 ##
 # @GuestPanicInformationHyperV:
@@ -566,6 +569,27 @@
            'psw-addr': 'uint64',
            'reason': 'S390CrashReason'}}
 
Quoted 9 lines

qapi/run-state.json line 578

Quoted 7 lines
qapi/run-state.json +7 -0
+##
+# @GuestPanicInformationTdx:
+#
+# TDX Guest panic information specific to TDX GCHI
+# TDG.VP.VMCALL<ReportFatalError>.
+#
+# @error-code: TD-specific error code

Where could a user find information on these error codes?

Quoted 5 lines

TDX GHCI (Guset-host-communication-Interface)spec. It defines all the TDVMCALL leaves.

0: panic;
0x1 - 0xffffffff: reserved.

Would it make sense to add a reference?

Quoted 14 lines
Quoted 5 lines

qapi/run-state.json line 581

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @gpa: guest-physical address of a page that contains additional
+#     error data, in forms of zero-terminated string.

"in the form of a zero-terminated string"

fixed.

Quoted 5 lines

qapi/run-state.json line 584

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @message: Human-readable error message provided by the guest. Not
+#     to be trusted.

How is this message related to the one pointed to by @gpa?

Quoted 6 lines

In general, @message contains a brief message of the error. While @gpa (when valid) contains a verbose message.

The reason why we need both is because sometime when TD guest hits a fatal error, its memory may get corrupted so we cannot pass information via @gpa. Information in @message is passed through GPRs.

Well, we do pass information via @gpa, always. I guess it page's contents can be corrupted.

Perhaps something like

# @message: Human-readable error message provided by the guest. Not
# to be trusted.
#
# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Note that guest
# memory corruption can corrupt the page's contents.

Quoted 7 lines
Quoted 7 lines

qapi/run-state.json line 591

Quoted 7 lines
qapi/run-state.json +7 -0
+#
+# Since: 9.0
+##
+{'struct': 'GuestPanicInformationTdx',
+ 'data': {'error-code': 'uint64',
+          'gpa': 'uint64',
+          'message': 'str'}}

Note that my proposed doc string has the members in a different order. Recommend to use the same order here.

Quoted 7 lines
Quoted 7 lines
Quoted 4 lines
qapi/run-state.json +1 -0
+
  ##
  # @MEMORY_FAILURE:
  #

[...]

author
#78 lore raw

Re: [PATCH v4 53/66] i386/tdx: Wire TDX_REPORT_FATAL_ERROR with GuestPanic facility

Quoted 64 lines · Markus Armbruster
Quoted 62 lines · Xiaoyao Li
Quoted 60 lines · Markus Armbruster
Quoted 46 lines · Xiaoyao Li

Integrate TDX's TDX_REPORT_FATAL_ERROR into QEMU GuestPanic facility

Originated-from: Isaku Yamahata <[email protected]>

---
Changes in v4:
- refine the documentation; (Markus)

Changes in v3:
- Add docmentation of new type and struct; (Daniel)
- refine the error message handling; (Daniel)

---
   qapi/run-state.json   | 28 ++++++++++++++++++++--
   system/runstate.c     | 54 +++++++++++++++++++++++++++++++++++++++++++
   target/i386/kvm/tdx.c | 24 ++++++++++++++++++-
   3 files changed, 103 insertions(+), 3 deletions(-)
qapi/run-state.json +5 -2
index 08bc99cb8561..5429116679e3 100644
@@ -485,10 +485,12 @@
  #
  # @s390: s390 guest panic information type (Since: 2.12)
  #
+# @tdx: tdx guest panic information type (Since: 8.2)
+#
  # Since: 2.9
  ##
  { 'enum': 'GuestPanicInformationType',
-  'data': [ 'hyper-v', 's390' ] }
+  'data': [ 'hyper-v', 's390', 'tdx' ] }
 
  ##
  # @GuestPanicInformation:
@@ -503,7 +505,8 @@
    'base': {'type': 'GuestPanicInformationType'},
    'discriminator': 'type',
    'data': {'hyper-v': 'GuestPanicInformationHyperV',
-          's390': 'GuestPanicInformationS390'}}
+          's390': 'GuestPanicInformationS390',
+          'tdx' : 'GuestPanicInformationTdx'}}
 
  ##
  # @GuestPanicInformationHyperV:

qapi/run-state.json line 578

Quoted 10 lines
qapi/run-state.json +7 -0
@@ -566,6 +569,27 @@
             'psw-addr': 'uint64',
             'reason': 'S390CrashReason'}}
 
+##
+# @GuestPanicInformationTdx:
+#
+# TDX Guest panic information specific to TDX GCHI
+# TDG.VP.VMCALL<ReportFatalError>.
+#
+# @error-code: TD-specific error code

Where could a user find information on these error codes?

Quoted 7 lines
Quoted 5 lines

TDX GHCI (Guset-host-communication-Interface)spec. It defines all the TDVMCALL leaves.

0: panic;
0x1 - 0xffffffff: reserved.

Would it make sense to add a reference?

https://cdrdv2.intel.com/v1/dl/getContent/726792

Quoted 17 lines
Quoted 17 lines
Quoted 5 lines

qapi/run-state.json line 581

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @gpa: guest-physical address of a page that contains additional
+#     error data, in forms of zero-terminated string.

"in the form of a zero-terminated string"

fixed.

Quoted 5 lines

qapi/run-state.json line 584

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @message: Human-readable error message provided by the guest. Not
+#     to be trusted.

How is this message related to the one pointed to by @gpa?

In general, @message contains a brief message of the error. While @gpa (when valid) contains a verbose message.

Quoted 6 lines
Quoted 3 lines

The reason why we need both is because sometime when TD guest hits a fatal error, its memory may get corrupted so we cannot pass information via @gpa. Information in @message is passed through GPRs.

Well, we do pass information via @gpa, always. I guess it page's contents can be corrupted.

No. It's not always. the bit 63 of the error code is "GPA valid" bit. @gpa is valid only when bit 63 of error code is 1.

And current Linux TD guest implementation doesn't use @gpa at all.
https://github.com/torvalds/linux/blob/45ec2f5f6ed3ec3a79ba1329ad585497cdcbe663/arch/x86/coco/tdx/tdx.c#L131

Quoted 28 lines

Perhaps something like

# @message: Human-readable error message provided by the guest. Not
# to be trusted.
#
# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Note that guest
# memory corruption can corrupt the page's contents.

Quoted 7 lines
Quoted 7 lines

qapi/run-state.json line 591

Quoted 7 lines
qapi/run-state.json +7 -0
+#
+# Since: 9.0
+##
+{'struct': 'GuestPanicInformationTdx',
+ 'data': {'error-code': 'uint64',
+          'gpa': 'uint64',
+          'message': 'str'}}

Note that my proposed doc string has the members in a different order. Recommend to use the same order here.

Quoted 7 lines
Quoted 7 lines
Quoted 4 lines
qapi/run-state.json +1 -0
+
   ##
   # @MEMORY_FAILURE:
   #

[...]

#79 lore raw

Re: [PATCH v4 53/66] i386/tdx: Wire TDX_REPORT_FATAL_ERROR with GuestPanic facility

Quoted 27 lines · Xiaoyao Li
Quoted 26 lines · Markus Armbruster
Quoted 24 lines · Xiaoyao Li
Quoted 23 lines · Markus Armbruster
Quoted 21 lines · Xiaoyao Li

Integrate TDX's TDX_REPORT_FATAL_ERROR into QEMU GuestPanic facility

Originated-from: Isaku Yamahata <[email protected]>

---
Changes in v4:
- refine the documentation; (Markus)

Changes in v3:
- Add docmentation of new type and struct; (Daniel)
- refine the error message handling; (Daniel)

---
   qapi/run-state.json   | 28 ++++++++++++++++++++--
   system/runstate.c     | 54 +++++++++++++++++++++++++++++++++++++++++++
   target/i386/kvm/tdx.c | 24 ++++++++++++++++++-
   3 files changed, 103 insertions(+), 3 deletions(-)
qapi/run-state.json +0 -0
index 08bc99cb8561..5429116679e3 100644

[...]

Quoted 17 lines
Quoted 17 lines
Quoted 17 lines
Quoted 4 lines
Quoted 4 lines
qapi/run-state.json +0 -0
@@ -566,6 +569,27 @@
             'psw-addr': 'uint64',
             'reason': 'S390CrashReason'}}
 
Quoted 9 lines

qapi/run-state.json line 578

Quoted 7 lines
qapi/run-state.json +7 -0
+##
+# @GuestPanicInformationTdx:
+#
+# TDX Guest panic information specific to TDX GCHI
+# TDG.VP.VMCALL<ReportFatalError>.
+#
+# @error-code: TD-specific error code

Where could a user find information on these error codes?

TDX GHCI (Guset-host-communication-Interface)spec. It defines all the TDVMCALL leaves.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

0: panic;
0x1 - 0xffffffff: reserved.

Would it make sense to add a reference?

https://cdrdv2.intel.com/v1/dl/getContent/726792

URLs have this annoying tendency to rot.

What about

# @error-code: Error code as defined in "Guest-Hypervisor Communication
# Interface (GHCI) Specification for Intel TDX 1.5"

Quoted 21 lines
Quoted 21 lines
Quoted 20 lines
Quoted 5 lines

qapi/run-state.json line 581

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @gpa: guest-physical address of a page that contains additional
+#     error data, in forms of zero-terminated string.

"in the form of a zero-terminated string"

fixed.

Quoted 5 lines

qapi/run-state.json line 584

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @message: Human-readable error message provided by the guest. Not
+#     to be trusted.

How is this message related to the one pointed to by @gpa?

In general, @message contains a brief message of the error. While @gpa (when valid) contains a verbose message.

The reason why we need both is because sometime when TD guest hits a fatal error, its memory may get corrupted so we cannot pass information via @gpa. Information in @message is passed through GPRs.

Quoted 8 lines
Quoted 2 lines

Well, we do pass information via @gpa, always. I guess it page's contents can be corrupted.

No. It's not always. the bit 63 of the error code is "GPA valid" bit. @gpa is valid only when bit 63 of error code is 1.

And current Linux TD guest implementation doesn't use @gpa at all.
https://github.com/torvalds/linux/blob/45ec2f5f6ed3ec3a79ba1329ad585497cdcbe663/arch/x86/coco/tdx/tdx.c#L131

Aha!

Why would we want to include @gpa when the "GPA valid" bit is off?

If we do want it, then

# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Valid when the
# "GPA valid" bit is set in @error-code.

If we don't, then make @gpa optional, present when valid, and document it like this:

# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Present when the
# "GPA valid" bit is set in @error-code.

Quoted 28 lines
Quoted 28 lines

Perhaps something like

# @message: Human-readable error message provided by the guest. Not
# to be trusted.
#
# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Note that guest
# memory corruption can corrupt the page's contents.

Quoted 7 lines
Quoted 7 lines

qapi/run-state.json line 591

Quoted 7 lines
qapi/run-state.json +7 -0
+#
+# Since: 9.0
+##
+{'struct': 'GuestPanicInformationTdx',
+ 'data': {'error-code': 'uint64',
+          'gpa': 'uint64',
+          'message': 'str'}}

Note that my proposed doc string has the members in a different order. Recommend to use the same order here.

Quoted 7 lines
Quoted 7 lines
Quoted 4 lines
qapi/run-state.json +1 -0
+
   ##
   # @MEMORY_FAILURE:
   #

[...]

author
#80 lore raw

Re: [PATCH v4 53/66] i386/tdx: Wire TDX_REPORT_FATAL_ERROR with GuestPanic facility

On 2/27/2024 9:09 PM, Markus Armbruster wrote:
[...]

Quoted 22 lines
Quoted 22 lines
Quoted 21 lines
Quoted 14 lines
Quoted 4 lines
Quoted 4 lines
qapi/run-state.json +0 -0
@@ -566,6 +569,27 @@
              'psw-addr': 'uint64',
              'reason': 'S390CrashReason'}}
 
Quoted 9 lines

qapi/run-state.json line 578

Quoted 7 lines
qapi/run-state.json +7 -0
+##
+# @GuestPanicInformationTdx:
+#
+# TDX Guest panic information specific to TDX GCHI
+# TDG.VP.VMCALL<ReportFatalError>.
+#
+# @error-code: TD-specific error code

Where could a user find information on these error codes?

Quoted 5 lines

TDX GHCI (Guset-host-communication-Interface)spec. It defines all the TDVMCALL leaves.

0: panic;
0x1 - 0xffffffff: reserved.

Would it make sense to add a reference?

Quoted 8 lines
Quoted 1 line

https://cdrdv2.intel.com/v1/dl/getContent/726792

URLs have this annoying tendency to rot.

What about

# @error-code: Error code as defined in "Guest-Hypervisor Communication
# Interface (GHCI) Specification for Intel TDX 1.5"

I think it gets mentioned at the beginning of @GuestPanicInformationTdx

TDX Guest panic information specific to TDX GHCI TDG.VP.VMCALL<ReportFatalError>.

Do we still to mention it in every single member?

Quoted 40 lines
Quoted 21 lines
Quoted 21 lines
Quoted 20 lines
Quoted 5 lines

qapi/run-state.json line 581

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @gpa: guest-physical address of a page that contains additional
+#     error data, in forms of zero-terminated string.

"in the form of a zero-terminated string"

fixed.

Quoted 5 lines

qapi/run-state.json line 584

Quoted 3 lines
qapi/run-state.json +3 -0
+#
+# @message: Human-readable error message provided by the guest. Not
+#     to be trusted.

How is this message related to the one pointed to by @gpa?

In general, @message contains a brief message of the error. While @gpa (when valid) contains a verbose message.

The reason why we need both is because sometime when TD guest hits a fatal error, its memory may get corrupted so we cannot pass information via @gpa. Information in @message is passed through GPRs.

Quoted 8 lines
Quoted 2 lines

Well, we do pass information via @gpa, always. I guess it page's contents can be corrupted.

No. It's not always. the bit 63 of the error code is "GPA valid" bit. @gpa is valid only when bit 63 of error code is 1.

And current Linux TD guest implementation doesn't use @gpa at all.
https://github.com/torvalds/linux/blob/45ec2f5f6ed3ec3a79ba1329ad585497cdcbe663/arch/x86/coco/tdx/tdx.c#L131

Aha!

Why would we want to include @gpa when the "GPA valid" bit is off?

If we do want it, then

# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Valid when the
# "GPA valid" bit is set in @error-code.

Quoted 6 lines

If we don't, then make @gpa optional, present when valid, and document it like this:

# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Present when the
# "GPA valid" bit is set in @error-code.

I will go this direction.

thanks!

Quoted 30 lines
Quoted 28 lines
Quoted 28 lines

Perhaps something like

# @message: Human-readable error message provided by the guest. Not
# to be trusted.
#
# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Note that guest
# memory corruption can corrupt the page's contents.

Quoted 7 lines
Quoted 7 lines

qapi/run-state.json line 591

Quoted 7 lines
qapi/run-state.json +7 -0
+#
+# Since: 9.0
+##
+{'struct': 'GuestPanicInformationTdx',
+ 'data': {'error-code': 'uint64',
+          'gpa': 'uint64',
+          'message': 'str'}}

Note that my proposed doc string has the members in a different order. Recommend to use the same order here.

Quoted 7 lines
Quoted 7 lines
Quoted 4 lines
qapi/run-state.json +1 -0
+
    ##
    # @MEMORY_FAILURE:
    #

[...]

#81 lore raw

Re: [PATCH v4 53/66] i386/tdx: Wire TDX_REPORT_FATAL_ERROR with GuestPanic facility

Quoted 33 lines · Xiaoyao Li

On 2/27/2024 9:09 PM, Markus Armbruster wrote:
[...]

Quoted 22 lines
Quoted 22 lines
Quoted 21 lines
Quoted 14 lines
Quoted 4 lines
Quoted 4 lines
(quoted diff) +0 -0
@@ -566,6 +569,27 @@
              'psw-addr': 'uint64',
              'reason': 'S390CrashReason'}}
 
Quoted 9 lines

Review line 578

Quoted 7 lines
(quoted diff) +7 -0
+##
+# @GuestPanicInformationTdx:
+#
+# TDX Guest panic information specific to TDX GCHI
+# TDG.VP.VMCALL<ReportFatalError>.
+#
+# @error-code: TD-specific error code

Where could a user find information on these error codes?

Quoted 5 lines

TDX GHCI (Guset-host-communication-Interface)spec. It defines all the TDVMCALL leaves.

0: panic;
0x1 - 0xffffffff: reserved.

Would it make sense to add a reference?

Quoted 8 lines
Quoted 1 line

https://cdrdv2.intel.com/v1/dl/getContent/726792

URLs have this annoying tendency to rot.

What about

# @error-code: Error code as defined in "Guest-Hypervisor Communication
# Interface (GHCI) Specification for Intel TDX 1.5"

Quoted 6 lines

I think it gets mentioned at the beginning of @GuestPanicInformationTdx

TDX Guest panic information specific to TDX GHCI TDG.VP.VMCALL<ReportFatalError>.

Do we still to mention it in every single member?

No, I didn't recognize the alphabet soup there as a reference :)

Let me try again:

##
# @GuestPanicInformationTdx:
#
# TDX Guest panic information specific to TDX, as specified in the # "Guest-Hypervisor Communication Interface (GHCI) Specification",

# section TDG.VP.VMCALL<ReportFatalError>.
#
# @error-code: TD-specific error code
#
[...]
#
# Since: 9.0
##
Quoted 43 lines
Quoted 43 lines
Quoted 21 lines
Quoted 21 lines
Quoted 20 lines
Quoted 5 lines

Review line 581

Quoted 3 lines
(quoted diff) +3 -0
+#
+# @gpa: guest-physical address of a page that contains additional
+#     error data, in forms of zero-terminated string.

"in the form of a zero-terminated string"

fixed.

Quoted 5 lines

Review line 584

Quoted 3 lines
(quoted diff) +3 -0
+#
+# @message: Human-readable error message provided by the guest. Not
+#     to be trusted.

How is this message related to the one pointed to by @gpa?

In general, @message contains a brief message of the error. While @gpa (when valid) contains a verbose message.

The reason why we need both is because sometime when TD guest hits a fatal error, its memory may get corrupted so we cannot pass information via @gpa. Information in @message is passed through GPRs.

Quoted 8 lines
Quoted 2 lines

Well, we do pass information via @gpa, always. I guess it page's contents can be corrupted.

No. It's not always. the bit 63 of the error code is "GPA valid" bit. @gpa is valid only when bit 63 of error code is 1.

And current Linux TD guest implementation doesn't use @gpa at all.
https://github.com/torvalds/linux/blob/45ec2f5f6ed3ec3a79ba1329ad585497cdcbe663/arch/x86/coco/tdx/tdx.c#L131

Aha!

Why would we want to include @gpa when the "GPA valid" bit is off?

If we do want it, then

# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Valid when the
# "GPA valid" bit is set in @error-code.

If we don't, then make @gpa optional, present when valid, and document it like this:

Quoted 7 lines
Quoted 3 lines

# @gpa: guest-physical address of a page that contains more verbose
# error information, as zero-terminated string. Present when the
# "GPA valid" bit is set in @error-code.

I will go this direction.

thanks!

You're welcome!

[...]

author
#82 lore raw

PATCH v4 54/66 pci-host/q35: Move PAM initialization above SMRAM initialization

In reply to Xiaoyao Li

From: Isaku Yamahata <[email protected]>

In mch_realize(), process PAM initialization before SMRAM initialization so that later patch can skill all the SMRAM related with a single check.

---
 hw/pci-host/q35.c | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)
hw/pci-host/q35.c +10 -9
index 0d7d4e3f0860..98d4a7c253a6 100644
@@ -568,6 +568,16 @@ static void mch_realize(PCIDevice *d, Error **errp)
     /* setup pci memory mapping */
     pc_pci_as_mapping_init(mch->system_memory, mch->pci_address_space);
 
+    /* PAM */
+    init_pam(&mch->pam_regions[0], OBJECT(mch), mch->ram_memory,
+             mch->system_memory, mch->pci_address_space,
+             PAM_BIOS_BASE, PAM_BIOS_SIZE);
+    for (i = 0; i < ARRAY_SIZE(mch->pam_regions) - 1; ++i) {
+        init_pam(&mch->pam_regions[i + 1], OBJECT(mch), mch->ram_memory,
+                 mch->system_memory, mch->pci_address_space,
+                 PAM_EXPAN_BASE + i * PAM_EXPAN_SIZE, PAM_EXPAN_SIZE);
+    }
+
     /* if *disabled* show SMRAM to all CPUs */
     memory_region_init_alias(&mch->smram_region, OBJECT(mch), "smram-region",
                              mch->pci_address_space, MCH_HOST_BRIDGE_SMRAM_C_BASE,
@@ -634,15 +644,6 @@ static void mch_realize(PCIDevice *d, Error **errp)
 
     object_property_add_const_link(qdev_get_machine(), "smram",
                                    OBJECT(&mch->smram));
-
-    init_pam(&mch->pam_regions[0], OBJECT(mch), mch->ram_memory,
-             mch->system_memory, mch->pci_address_space,
-             PAM_BIOS_BASE, PAM_BIOS_SIZE);
-    for (i = 0; i < ARRAY_SIZE(mch->pam_regions) - 1; ++i) {
-        init_pam(&mch->pam_regions[i + 1], OBJECT(mch), mch->ram_memory,
-                 mch->system_memory, mch->pci_address_space,
-                 PAM_EXPAN_BASE + i * PAM_EXPAN_SIZE, PAM_EXPAN_SIZE);
-    }
 }
 
 uint64_t mch_mcfg_base(void)
Signature · 1 line
2.34.1
author
#83 lore raw

PATCH v4 55/66 q35: Introduce smm_ranges property for q35-pci-host

In reply to Xiaoyao Li

From: Isaku Yamahata <[email protected]>

Add a q35 property to check whether or not SMM ranges, e.g. SMRAM, TSEG, etc... exist for the target platform. TDX doesn't support SMM and doesn't play nice with QEMU modifying related guest memory ranges.

---
 hw/i386/pc_q35.c          |  2 ++
 hw/pci-host/q35.c         | 42 +++++++++++++++++++++++++++------------
 include/hw/i386/pc.h      |  1 +
 include/hw/pci-host/q35.h |  1 +
 4 files changed, 33 insertions(+), 13 deletions(-)
hw/i386/pc_q35.c +2 -0
index f43d5142b8e5..4e467fbb2f65 100644
@@ -236,6 +236,8 @@ static void pc_q35_init(MachineState *machine)
                             x86ms->above_4g_mem_size, NULL);
     object_property_set_bool(phb, PCI_HOST_BYPASS_IOMMU,
                              pcms->default_bus_bypass_iommu, NULL);
+    object_property_set_bool(phb, PCI_HOST_PROP_SMM_RANGES,
+                             x86_machine_is_smm_enabled(x86ms), NULL);
     sysbus_realize_and_unref(SYS_BUS_DEVICE(phb), &error_fatal);
 
     /* pci */
hw/pci-host/q35.c +29 -13

82 lines not on this page · raw message

include/hw/i386/pc.h +1 -0
index b7c59818df7d..1aa3ed590596 100644
@@ -165,6 +165,7 @@ void pc_guest_info_init(PCMachineState *pcms);
 #define PCI_HOST_PROP_PCI_HOLE64_SIZE  "pci-hole64-size"
 #define PCI_HOST_BELOW_4G_MEM_SIZE     "below-4g-mem-size"
 #define PCI_HOST_ABOVE_4G_MEM_SIZE     "above-4g-mem-size"
+#define PCI_HOST_PROP_SMM_RANGES       "smm-ranges"
 
 
 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
include/hw/pci-host/q35.h +1 -0
index bafcbe675214..22fadfa3ed76 100644
@@ -50,6 +50,7 @@ struct MCHPCIState {
     MemoryRegion tseg_blackhole, tseg_window;
     MemoryRegion smbase_blackhole, smbase_window;
     bool has_smram_at_smbase;
+    bool has_smm_ranges;
     Range pci_hole;
     uint64_t below_4g_mem_size;
     uint64_t above_4g_mem_size;
Signature · 1 line
2.34.1
author
#84 lore raw

PATCH v4 56/66 i386/tdx: Disable SMM for TDX VMs

In reply to Xiaoyao Li

TDX 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
index 4fbb18135951..7eb3628763ae 100644
@@ -695,11 +695,19 @@ static Notifier tdx_machine_done_notify = {
 
 int tdx_kvm_init(MachineState *ms, Error **errp)
 {
+    X86MachineState *x86ms = X86_MACHINE(ms);
     TdxGuest *tdx = TDX_GUEST(OBJECT(ms->cgs));
     int r = 0;
 
     ms->require_guest_memfd = true;
 
+    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) {
         r = get_tdx_capabilities(errp);
         if (r) {
Signature · 1 line
2.34.1
author
#85 lore raw

PATCH v4 57/66 i386/tdx: Disable PIC for TDX VMs

In reply to Xiaoyao Li

Legacy 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
index 7eb3628763ae..838b7916278c 100644
@@ -708,6 +708,13 @@ 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) {
         r = get_tdx_capabilities(errp);
         if (r) {
Signature · 1 line
2.34.1
author
#86 lore raw

PATCH v4 58/66 i386/tdx: Don't allow system reset for TDX VMs

In reply to Xiaoyao Li

TDX 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
index e36ece874246..0ec0584d22ca 100644
@@ -5686,7 +5686,7 @@ 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
author
#87 lore raw

PATCH v4 59/66 i386/tdx: LMCE is not supported for TDX

In reply to Xiaoyao Li

LMCE 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
index 9c791b7b0520..8c618869533c 100644
@@ -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"
 
@@ -60,6 +61,10 @@ 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
author
#88 lore raw

PATCH v4 60/66 hw/i386: add eoi_intercept_unsupported member to X86MachineState

In reply to Xiaoyao Li

Add 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
index f13f49069d40..504575abfa98 100644
@@ -1422,6 +1422,7 @@ 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
index ab1d38569019..b689feb389b3 100644
@@ -59,6 +59,7 @@ 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
index 838b7916278c..c844d753736c 100644
@@ -715,6 +715,8 @@ int tdx_kvm_init(MachineState *ms, Error **errp)
         return -EINVAL;
     }
 
+    x86ms->eoi_intercept_unsupported = true;
+
     if (!tdx_caps) {
         r = get_tdx_capabilities(errp);
         if (r) {
Signature · 1 line
2.34.1
author
#89 lore raw

PATCH v4 61/66 hw/i386: add option to forcibly report edge trigger in acpi tables

In reply to Xiaoyao Li

From: 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
index edc979379c03..3d1e4457ff30 100644
@@ -975,7 +975,8 @@ 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;
@@ -987,7 +988,10 @@ 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));
 
@@ -1011,7 +1015,8 @@ 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;
@@ -1024,7 +1029,10 @@ 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));
 
@@ -1043,7 +1051,7 @@ 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;
@@ -1051,7 +1059,9 @@ 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));
 
@@ -1085,7 +1095,7 @@ 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;
@@ -1098,12 +1108,16 @@ 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");
     {
@@ -1112,7 +1126,9 @@ 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));
@@ -1198,7 +1214,7 @@ 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");
@@ -1237,25 +1253,41 @@ 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;
@@ -1468,7 +1501,7 @@ 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");
@@ -1512,7 +1545,7 @@ 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
author
#90 lore raw

PATCH v4 62/66 i386/tdx: Don't synchronize guest tsc for TDs

In reply to Xiaoyao Li

From: 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
index 0ec0584d22ca..d8819da8dccc 100644
@@ -278,7 +278,7 @@ 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
author
#91 lore raw

PATCH v4 63/66 i386/tdx: Only configure MSR_IA32_UCODE_REV in kvm_init_msrs() for TDs

In reply to Xiaoyao Li

For 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
index d8819da8dccc..128cce929ed3 100644
@@ -3283,32 +3283,34 @@ 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
author
#92 lore raw

PATCH v4 64/66 i386/tdx: Skip kvm_put_apicbase() for TDs

In reply to Xiaoyao Li

KVM 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
index 128cce929ed3..109f778ba14d 100644
@@ -3065,6 +3065,11 @@ 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
author
#93 lore raw

PATCH v4 65/66 i386/tdx: Don't get/put guest state for TDX VMs

In reply to Xiaoyao Li

From: 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
index 109f778ba14d..bad00da37c2f 100644
@@ -4633,6 +4633,11 @@ 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
@@ -4733,6 +4738,12 @@ 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
author
#94 lore raw

PATCH v4 66/66 docs: Add TDX documentation

In reply to Xiaoyao Li

Add 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                   | 113 +++++++++++++++++++++
 docs/system/target-i386.rst                |   1 +
 3 files changed, 115 insertions(+)
 create mode 100644 docs/system/i386/tdx.rst
docs/system/confidential-guest-support.rst +1 -0
index 0c490dbda2b7..66129fbab64c 100644
@@ -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 +113 -0

113 lines not on this page · raw message

docs/system/target-i386.rst +1 -0
index 1b8a1f248abb..4d58cdbc4e06 100644
@@ -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