Add struct kvm_vcpu_arch_ser in include/linux/kho/abi/kvm_x86.h and export x86 register and MSR access helpers for LUO state save/restore.
Signed-off-by: Pasha Tatashin <[email protected]> --- arch/x86/kvm/cpuid.c | 4 +- arch/x86/kvm/cpuid.h | 1 + arch/x86/kvm/msrs.c | 15 ++++++ arch/x86/kvm/msrs.h | 2 + arch/x86/kvm/regs.c | 8 ++-- arch/x86/kvm/regs.h | 4 ++ arch/x86/kvm/x86.c | 8 ++-- arch/x86/kvm/x86.h | 4 ++ include/linux/kho/abi/kvm_x86.h | 84 +++++++++++++++++++++++++++++++++ 9 files changed, 120 insertions(+), 10 deletions(-) create mode 100644 include/linux/kho/abi/kvm_x86.h diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c index 851f151efb35..5aba16e0610b 100644 --- a/arch/x86/kvm/cpuid.c +++ b/arch/x86/kvm/cpuid.c @@ -507,8 +507,8 @@ u64 kvm_vcpu_reserved_gpa_bits_raw(struct kvm_vcpu *vcpu) return rsvd_bits(cpuid_maxphyaddr(vcpu), 63); } -static int kvm_set_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid_entry2 *e2, - int nent) +int kvm_set_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid_entry2 *e2, + int nent) { u32 vcpu_caps[NR_KVM_CPU_CAPS]; int r; diff --git a/arch/x86/kvm/cpuid.h b/arch/x86/kvm/cpuid.h index 8d863f45585d..e5ec112d39cc 100644 --- a/arch/x86/kvm/cpuid.h +++ b/arch/x86/kvm/cpuid.h @@ -21,6 +21,7 @@ static inline void kvm_finalize_cpu_caps(void) } void kvm_vcpu_after_set_cpuid(struct kvm_vcpu *vcpu); +int kvm_set_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid_entry2 *e2, int nent); struct kvm_cpuid_entry2 *kvm_find_cpuid_entry2(struct kvm_cpuid_entry2 *entries, int nent, u32 function, u64 index); /* diff --git a/arch/x86/kvm/msrs.c b/arch/x86/kvm/msrs.c index dd3bb04878ca..ae849e0b7d14 100644 --- a/arch/x86/kvm/msrs.c +++ b/arch/x86/kvm/msrs.c @@ -339,6 +339,21 @@ static u32 msr_based_features[ARRAY_SIZE(msr_based_features_all_except_vmx) + (KVM_LAST_EMULATED_VMX_MSR - KVM_FIRST_EMULATED_VMX_MSR + 1)]; static unsigned int num_msr_based_features; +unsigned int kvm_num_msrs_to_save(void) +{ + return num_msrs_to_save + num_emulated_msrs; +} + +u32 kvm_get_msr_to_save_index(unsigned int i) +{ + if (i < num_msrs_to_save) + return msrs_to_save[i]; + i -= num_msrs_to_save; + if (i < num_emulated_msrs) + return emulated_msrs[i]; + return 0; +} + int kvm_get_msr_index_list(struct kvm_msr_list __user *user_msr_list) { struct kvm_msr_list msr_list; diff --git a/arch/x86/kvm/msrs.h b/arch/x86/kvm/msrs.h index 7cc182a15b3b..3a5e39041d99 100644 --- a/arch/x86/kvm/msrs.h +++ b/arch/x86/kvm/msrs.h @@ -65,6 +65,8 @@ int __kvm_emulate_msr_read(struct kvm_vcpu *vcpu, u32 index, u64 *data); int __kvm_emulate_msr_write(struct kvm_vcpu *vcpu, u32 index, u64 data); int kvm_msr_read(struct kvm_vcpu *vcpu, u32 index, u64 *data); int kvm_msr_write(struct kvm_vcpu *vcpu, u32 index, u64 data); +unsigned int kvm_num_msrs_to_save(void); +u32 kvm_get_msr_to_save_index(unsigned int i); int kvm_emulate_rdmsr(struct kvm_vcpu *vcpu); int kvm_emulate_rdmsr_imm(struct kvm_vcpu *vcpu, u32 msr, int reg); int kvm_emulate_wrmsr(struct kvm_vcpu *vcpu); diff --git a/arch/x86/kvm/regs.c b/arch/x86/kvm/regs.c index 8f66438989e4..02f041c355cc 100644 --- a/arch/x86/kvm/regs.c +++ b/arch/x86/kvm/regs.c @@ -51,7 +51,7 @@ void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags) } EXPORT_SYMBOL_FOR_KVM_INTERNAL(kvm_set_rflags); -static void __get_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs) +void __get_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs) { if (vcpu->arch.emulate_regs_need_sync_to_vcpu) { /* @@ -99,7 +99,7 @@ int kvm_arch_vcpu_ioctl_get_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs) return 0; } -static void __set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs) +void __set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs) { vcpu->arch.emulate_regs_need_sync_from_vcpu = true; vcpu->arch.emulate_regs_need_sync_to_vcpu = false; @@ -498,7 +498,7 @@ static void __get_sregs_common(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs) sregs->apic_base = vcpu->arch.apic_base; } -static void __get_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs) +void __get_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs) { __get_sregs_common(vcpu, sregs); @@ -635,7 +635,7 @@ static int __set_sregs_common(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs, return 0; } -static int __set_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs) +int __set_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs) { int pending_vec, max_bits; int mmu_reset_needed = 0; diff --git a/arch/x86/kvm/regs.h b/arch/x86/kvm/regs.h index 447f0ec3e63e..40947b1be404 100644 --- a/arch/x86/kvm/regs.h +++ b/arch/x86/kvm/regs.h @@ -517,5 +517,9 @@ int kvm_vcpu_ioctl_x86_get_debugregs(struct kvm_vcpu *vcpu, int kvm_vcpu_ioctl_x86_set_debugregs(struct kvm_vcpu *vcpu, struct kvm_debugregs *dbgregs); +void __get_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs); +void __set_regs(struct kvm_vcpu *vcpu, struct kvm_regs *regs); +void __get_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs); +int __set_sregs(struct kvm_vcpu *vcpu, struct kvm_sregs *sregs); #endif diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 79468ddfe473..69e0d67a309b 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -2959,8 +2959,8 @@ void kvm_handle_exception_payload_quirk(struct kvm_vcpu *vcpu) kvm_deliver_exception_payload(vcpu, ex); } -static void kvm_vcpu_ioctl_x86_get_vcpu_events(struct kvm_vcpu *vcpu, - struct kvm_vcpu_events *events) +void kvm_vcpu_ioctl_x86_get_vcpu_events(struct kvm_vcpu *vcpu, + struct kvm_vcpu_events *events) { struct kvm_queued_exception *ex = kvm_get_exception_to_save(vcpu); @@ -3028,8 +3028,8 @@ static void kvm_vcpu_ioctl_x86_get_vcpu_events(struct kvm_vcpu *vcpu, } } -static int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu, - struct kvm_vcpu_events *events) +int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu, + struct kvm_vcpu_events *events) { if (events->flags & ~(KVM_VCPUEVENT_VALID_NMI_PENDING | KVM_VCPUEVENT_VALID_SIPI_VECTOR diff --git a/arch/x86/kvm/x86.h b/arch/x86/kvm/x86.h index 0f5919b092e4..936ee660ccec 100644 --- a/arch/x86/kvm/x86.h +++ b/arch/x86/kvm/x86.h @@ -492,6 +492,10 @@ int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index, int reason, bool has_error_code, u32 error_code); int __kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr); +void kvm_vcpu_ioctl_x86_get_vcpu_events(struct kvm_vcpu *vcpu, + struct kvm_vcpu_events *events); +int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu, + struct kvm_vcpu_events *events); int kvm_emulate_xsetbv(struct kvm_vcpu *vcpu); int kvm_emulate_rdpmc(struct kvm_vcpu *vcpu); diff --git a/include/linux/kho/abi/kvm_x86.h b/include/linux/kho/abi/kvm_x86.h new file mode 100644 index 000000000000..316250482db5 --- /dev/null +++ b/include/linux/kho/abi/kvm_x86.h @@ -0,0 +1,84 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * Copyright (c) 2026, Google LLC. + * Pasha Tatashin <[email protected]> + */ +#ifndef _LINUX_KHO_ABI_KVM_X86_H +#define _LINUX_KHO_ABI_KVM_X86_H + +#ifdef CONFIG_X86_64 + +#include <linux/build_bug.h> +#include <linux/stddef.h> +#include <linux/types.h> +#include <linux/kho/abi/kvm.h> +#include <uapi/linux/kvm.h> +#include <uapi/asm/kvm.h> + +/** + * DOC: x86 KVM Live Update ABI + * + * x86 KVM uses the ABI defined below for preserving architectural VM and vCPU + * state across a kexec reboot using LUO. + * + * The vCPU-level architectural state contains the compact register sets and + * CPUID table for each vCPU. + * + * All sub-structures (struct kvm_regs, struct kvm_sregs, struct kvm_mp_state, + * struct kvm_cpuid_entry2) are uAPI contracts. + */ + +/** + * struct kvm_vcpu_arch_ser - Preserved x86 architectural vCPU state in RAM. + * @regs: General-purpose registers (uAPI struct kvm_regs). + * @sregs: Segment and control registers (uAPI struct kvm_sregs). + * @mp_state: Multiprocessor state (uAPI struct kvm_mp_state). + * @pad: Padding to maintain 64-bit alignment after mp_state. + * @xcrs: Extended control registers including XCR0 (uAPI struct kvm_xcrs). + * @lapic: In-kernel local APIC register state (uAPI struct kvm_lapic_state). + * @pad_xsave: Padding that aligns @xsave to 64 bytes; see the static_assert + * below. Must be zero. + * @xsave: Extended processor state (uAPI struct kvm_xsave). This is the + * only copy of the guest FPU state; there is no kvm_fpu twin. + * @events: vCPU exception, interrupt, NMI and SMI events (uAPI struct kvm_vcpu_events). + * @debugregs: Hardware debug registers DR0-DR7 (uAPI struct kvm_debugregs). + * @num_msrs: Number of valid MSR entries in msrs. + * @cpuid_nent: Number of valid CPUID entries immediately following msrs[num_msrs]. + * @msrs: Array of preserved architectural and paravirtual MSR entries, + * followed by @cpuid_nent struct kvm_cpuid_entry2 entries. + */ +struct kvm_vcpu_arch_ser { + struct kvm_regs regs; + struct kvm_sregs sregs; + struct kvm_mp_state mp_state; + u32 pad; + struct kvm_xcrs xcrs; + struct kvm_lapic_state lapic; + u8 pad_xsave[40]; + struct kvm_xsave xsave; + struct kvm_vcpu_events events; + struct kvm_debugregs debugregs; + u32 num_msrs; + u32 cpuid_nent; + struct kvm_msr_entry msrs[]; +} __packed; + +static_assert(offsetof(struct kvm_vcpu_arch_ser, msrs) % sizeof(u64) == 0, + "msrs must be 64-bit aligned"); + +/* + * @xsave must be a legal XSAVE destination. The structure is __packed, so + * without @pad_xsave the member lands at offset 1880, which is not even + * 16-byte aligned, and the XSAVE/XRSTOR family faults on anything less than + * 64. The allocation itself is fine: kho_alloc_preserve() returns a folio + * address, so the base is page aligned. + * + * If this assert fires, a member above @xsave changed size; adjust + * @pad_xsave rather than deleting the assert. + */ +static_assert(offsetof(struct kvm_vcpu_arch_ser, xsave) % 64 == 0, + "xsave must be 64-byte aligned to be XSAVE-able in place"); + +#endif /* CONFIG_X86_64 */ + +#endif /* _LINUX_KHO_ABI_KVM_X86_H */ -- 2.55.0.1082.g2b9226bbc0-goog
