diff options
| -rw-r--r-- | Documentation/virt/kvm/api.rst | 115 | ||||
| -rw-r--r-- | arch/x86/include/asm/kvm_host.h | 9 | ||||
| -rw-r--r-- | arch/x86/kvm/Kconfig | 9 | ||||
| -rw-r--r-- | arch/x86/kvm/Makefile | 3 | ||||
| -rw-r--r-- | arch/x86/kvm/hyperv.c | 2 | ||||
| -rw-r--r-- | arch/x86/kvm/mmu/mmu_internal.h | 16 | ||||
| -rw-r--r-- | arch/x86/kvm/svm/svm.c | 1 | ||||
| -rw-r--r-- | arch/x86/kvm/x86.c | 22 | ||||
| -rw-r--r-- | arch/x86/kvm/xen.c | 290 | ||||
| -rw-r--r-- | arch/x86/kvm/xen.h | 64 | ||||
| -rw-r--r-- | include/uapi/linux/kvm.h | 13 | ||||
| -rw-r--r-- | tools/testing/selftests/kvm/lib/kvm_util.c | 6 | ||||
| -rw-r--r-- | tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c | 159 |
13 files changed, 633 insertions, 76 deletions
diff --git a/Documentation/virt/kvm/api.rst b/Documentation/virt/kvm/api.rst index aed52b0fc16e..1a2b5210cdbf 100644 --- a/Documentation/virt/kvm/api.rst +++ b/Documentation/virt/kvm/api.rst @@ -3856,49 +3856,20 @@ base 2 of the page size in the bottom 6 bits. -EFAULT if struct kvm_reinject_control cannot be read, -EINVAL if the supplied shift or flags are invalid, -ENOMEM if unable to allocate the new HPT, - -ENOSPC if there was a hash collision - -:: - - struct kvm_ppc_rmmu_info { - struct kvm_ppc_radix_geom { - __u8 page_shift; - __u8 level_bits[4]; - __u8 pad[3]; - } geometries[8]; - __u32 ap_encodings[8]; - }; - -The geometries[] field gives up to 8 supported geometries for the -radix page table, in terms of the log base 2 of the smallest page -size, and the number of bits indexed at each level of the tree, from -the PTE level up to the PGD level in that order. Any unused entries -will have 0 in the page_shift field. - -The ap_encodings gives the supported page sizes and their AP field -encodings, encoded with the AP value in the top 3 bits and the log -base 2 of the page size in the bottom 6 bits. - -4.102 KVM_PPC_RESIZE_HPT_PREPARE --------------------------------- - -:Capability: KVM_CAP_SPAPR_RESIZE_HPT -:Architectures: powerpc -:Type: vm ioctl -:Parameters: struct kvm_ppc_resize_hpt (in) -:Returns: 0 on successful completion, - >0 if a new HPT is being prepared, the value is an estimated - number of milliseconds until preparation is complete, - -EFAULT if struct kvm_reinject_control cannot be read, - -EINVAL if the supplied shift or flags are invalid,when moving existing - HPT entries to the new HPT, - -EIO on other error conditions Used to implement the PAPR extension for runtime resizing of a guest's Hashed Page Table (HPT). Specifically this starts, stops or monitors the preparation of a new potential HPT for the guest, essentially implementing the H_RESIZE_HPT_PREPARE hypercall. +:: + + struct kvm_ppc_resize_hpt { + __u64 flags; + __u32 shift; + __u32 pad; + }; + If called with shift > 0 when there is no pending HPT for the guest, this begins preparation of a new pending HPT of size 2^(shift) bytes. It then returns a positive integer with the estimated number of @@ -3926,14 +3897,6 @@ Normally this will be called repeatedly with the same parameters until it returns <= 0. The first call will initiate preparation, subsequent ones will monitor preparation until it completes or fails. -:: - - struct kvm_ppc_resize_hpt { - __u64 flags; - __u32 shift; - __u32 pad; - }; - 4.103 KVM_PPC_RESIZE_HPT_COMMIT ------------------------------- @@ -3956,6 +3919,14 @@ Hashed Page Table (HPT). Specifically this requests that the guest be transferred to working with the new HPT, essentially implementing the H_RESIZE_HPT_COMMIT hypercall. +:: + + struct kvm_ppc_resize_hpt { + __u64 flags; + __u32 shift; + __u32 pad; + }; + This should only be called after KVM_PPC_RESIZE_HPT_PREPARE has returned 0 with the same parameters. In other cases KVM_PPC_RESIZE_HPT_COMMIT will return an error (usually -ENXIO or @@ -3971,14 +3942,6 @@ HPT and the previous HPT will be discarded. On failure, the guest will still be operating on its previous HPT. -:: - - struct kvm_ppc_resize_hpt { - __u64 flags; - __u32 shift; - __u32 pad; - }; - 4.104 KVM_X86_GET_MCE_CAP_SUPPORTED ----------------------------------- @@ -4915,6 +4878,14 @@ see KVM_XEN_HVM_SET_ATTR above. union { __u64 gpa; __u64 pad[4]; + struct { + __u64 state; + __u64 state_entry_time; + __u64 time_running; + __u64 time_runnable; + __u64 time_blocked; + __u64 time_offline; + } runstate; } u; }; @@ -4927,6 +4898,31 @@ KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO Sets the guest physical address of an additional pvclock structure for a given vCPU. This is typically used for guest vsyscall support. +KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR + Sets the guest physical address of the vcpu_runstate_info for a given + vCPU. This is how a Xen guest tracks CPU state such as steal time. + +KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT + Sets the runstate (RUNSTATE_running/_runnable/_blocked/_offline) of + the given vCPU from the .u.runstate.state member of the structure. + KVM automatically accounts running and runnable time but blocked + and offline states are only entered explicitly. + +KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA + Sets all fields of the vCPU runstate data from the .u.runstate member + of the structure, including the current runstate. The state_entry_time + must equal the sum of the other four times. + +KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST + This *adds* the contents of the .u.runstate members of the structure + to the corresponding members of the given vCPU's runstate data, thus + permitting atomic adjustments to the runstate times. The adjustment + to the state_entry_time must equal the sum of the adjustments to the + other four times. The state field must be set to -1, or to a valid + runstate value (RUNSTATE_running, RUNSTATE_runnable, RUNSTATE_blocked + or RUNSTATE_offline) to set the current accounted state as of the + adjusted state_entry_time. + 4.130 KVM_XEN_VCPU_GET_ATTR --------------------------- @@ -4939,6 +4935,9 @@ KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO Allows Xen vCPU attributes to be read. For the structure and types, see KVM_XEN_VCPU_SET_ATTR above. +The KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST type may not be used +with the KVM_XEN_VCPU_GET_ATTR ioctl. + 5. The kvm_run structure ======================== @@ -5000,7 +4999,8 @@ local APIC is not used. __u16 flags; More architecture-specific flags detailing state of the VCPU that may -affect the device's behavior. Current defined flags: +affect the device's behavior. Current defined flags:: + /* x86, set if the VCPU is in system management mode */ #define KVM_RUN_X86_SMM (1 << 0) /* x86, set if bus lock detected in VM */ @@ -6217,7 +6217,7 @@ the bus lock vm exit can be preempted by a higher priority VM exit, the exit notifications to userspace can be KVM_EXIT_BUS_LOCK or other reasons. KVM_RUN_BUS_LOCK flag is used to distinguish between them. -7.22 KVM_CAP_PPC_DAWR1 +7.23 KVM_CAP_PPC_DAWR1 ---------------------- :Architectures: ppc @@ -6702,6 +6702,7 @@ PVHVM guests. Valid flags are:: #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) + #define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 2) The KVM_XEN_HVM_CONFIG_HYPERCALL_MSR flag indicates that the KVM_XEN_HVM_CONFIG ioctl is available, for the guest to set its hypercall page. @@ -6716,3 +6717,7 @@ KVM_XEN_HVM_SET_ATTR, KVM_XEN_HVM_GET_ATTR, KVM_XEN_VCPU_SET_ATTR and KVM_XEN_VCPU_GET_ATTR ioctls, as well as the delivery of exception vectors for event channel upcalls when the evtchn_upcall_pending field of a vcpu's vcpu_info is set. + +The KVM_XEN_HVM_CONFIG_RUNSTATE flag indicates that the runstate-related +features KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR/_CURRENT/_DATA/_ADJUST are +supported by the KVM_XEN_VCPU_SET_ATTR/KVM_XEN_VCPU_GET_ATTR ioctls. diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index 0cf71ff2b2e5..877a4025d8da 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -535,10 +535,16 @@ struct kvm_vcpu_hv { /* Xen HVM per vcpu emulation context */ struct kvm_vcpu_xen { u64 hypercall_rip; + u32 current_runstate; bool vcpu_info_set; bool vcpu_time_info_set; + bool runstate_set; struct gfn_to_hva_cache vcpu_info_cache; struct gfn_to_hva_cache vcpu_time_info_cache; + struct gfn_to_hva_cache runstate_cache; + u64 last_steal; + u64 runstate_entry_time; + u64 runstate_times[4]; }; struct kvm_vcpu_arch { @@ -939,9 +945,6 @@ struct kvm_arch { unsigned int indirect_shadow_pages; u8 mmu_valid_gen; struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES]; - /* - * Hash table of struct kvm_mmu_page. - */ struct list_head active_mmu_pages; struct list_head zapped_obsolete_pages; struct list_head lpage_disallowed_mmu_pages; diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig index 7ac592664c52..a788d5120d4d 100644 --- a/arch/x86/kvm/Kconfig +++ b/arch/x86/kvm/Kconfig @@ -103,6 +103,15 @@ config KVM_AMD_SEV Provides support for launching Encrypted VMs (SEV) and Encrypted VMs with Encrypted State (SEV-ES) on AMD processors. +config KVM_XEN + bool "Support for Xen hypercall interface" + depends on KVM + help + Provides KVM support for the hosting Xen HVM guests and + passing Xen hypercalls to userspace. + + If in doubt, say "N". + config KVM_MMU_AUDIT bool "Audit KVM MMU" depends on KVM && TRACEPOINTS diff --git a/arch/x86/kvm/Makefile b/arch/x86/kvm/Makefile index aeab168c5711..1b4766fe1de2 100644 --- a/arch/x86/kvm/Makefile +++ b/arch/x86/kvm/Makefile @@ -14,11 +14,12 @@ kvm-y += $(KVM)/kvm_main.o $(KVM)/coalesced_mmio.o \ $(KVM)/dirty_ring.o kvm-$(CONFIG_KVM_ASYNC_PF) += $(KVM)/async_pf.o -kvm-y += x86.o emulate.o i8259.o irq.o lapic.o xen.o \ +kvm-y += x86.o emulate.o i8259.o irq.o lapic.o \ i8254.o ioapic.o irq_comm.o cpuid.o pmu.o mtrr.o \ hyperv.o debugfs.o mmu/mmu.o mmu/page_track.o \ mmu/spte.o kvm-$(CONFIG_X86_64) += mmu/tdp_iter.o mmu/tdp_mmu.o +kvm-$(CONFIG_KVM_XEN) += xen.o kvm-intel-y += vmx/vmx.o vmx/vmenter.o vmx/pmu_intel.o vmx/vmcs12.o \ vmx/evmcs.o vmx/nested.o vmx/posted_intr.o diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c index 7d2dae92d638..58fa8c029867 100644 --- a/arch/x86/kvm/hyperv.c +++ b/arch/x86/kvm/hyperv.c @@ -159,7 +159,7 @@ static struct kvm_vcpu_hv_synic *synic_get(struct kvm *kvm, u32 vpidx) struct kvm_vcpu_hv_synic *synic; vcpu = get_vcpu_by_vpidx(kvm, vpidx); - if (!vcpu) + if (!vcpu || !to_hv_vcpu(vcpu)) return NULL; synic = to_hv_synic(vcpu); return (synic->active) ? synic : NULL; diff --git a/arch/x86/kvm/mmu/mmu_internal.h b/arch/x86/kvm/mmu/mmu_internal.h index 72b0928f2b2d..ec4fc28b325a 100644 --- a/arch/x86/kvm/mmu/mmu_internal.h +++ b/arch/x86/kvm/mmu/mmu_internal.h @@ -81,15 +81,15 @@ static inline struct kvm_mmu_page *sptep_to_sp(u64 *sptep) static inline bool kvm_vcpu_ad_need_write_protect(struct kvm_vcpu *vcpu) { /* - * When using the EPT page-modification log, the GPAs in the log - * would come from L2 rather than L1. Therefore, we need to rely - * on write protection to record dirty pages. This also bypasses - * PML, since writes now result in a vmexit. Note, this helper will - * tag SPTEs as needing write-protection even if PML is disabled or - * unsupported, but that's ok because the tag is consumed if and only - * if PML is enabled. Omit the PML check to save a few uops. + * When using the EPT page-modification log, the GPAs in the CPU dirty + * log would come from L2 rather than L1. Therefore, we need to rely + * on write protection to record dirty pages, which bypasses PML, since + * writes now result in a vmexit. Note, the check on CPU dirty logging + * being enabled is mandatory as the bits used to denote WP-only SPTEs + * are reserved for NPT w/ PAE (32-bit KVM). */ - return vcpu->arch.mmu == &vcpu->arch.guest_mmu; + return vcpu->arch.mmu == &vcpu->arch.guest_mmu && + kvm_x86_ops.cpu_dirty_log_size; } bool is_nx_huge_page_enabled(void); diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c index c636021b066b..baee91c1e936 100644 --- a/arch/x86/kvm/svm/svm.c +++ b/arch/x86/kvm/svm/svm.c @@ -1200,6 +1200,7 @@ static void init_vmcb(struct vcpu_svm *svm) init_sys_seg(&save->ldtr, SEG_TYPE_LDT); init_sys_seg(&save->tr, SEG_TYPE_BUSY_TSS16); + svm_set_cr4(&svm->vcpu, 0); svm_set_efer(&svm->vcpu, 0); save->dr6 = 0xffff0ff0; kvm_set_rflags(&svm->vcpu, X86_EFLAGS_FIXED); diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 3712bb5245eb..2a20ce60152e 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -2957,6 +2957,11 @@ static void record_steal_time(struct kvm_vcpu *vcpu) struct kvm_host_map map; struct kvm_steal_time *st; + if (kvm_xen_msr_enabled(vcpu->kvm)) { + kvm_xen_runstate_set_running(vcpu); + return; + } + if (!(vcpu->arch.st.msr_val & KVM_MSR_ENABLED)) return; @@ -3756,11 +3761,15 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext) case KVM_CAP_ENFORCE_PV_FEATURE_CPUID: r = 1; break; +#ifdef CONFIG_KVM_XEN case KVM_CAP_XEN_HVM: r = KVM_XEN_HVM_CONFIG_HYPERCALL_MSR | KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL | KVM_XEN_HVM_CONFIG_SHARED_INFO; + if (sched_info_on()) + r |= KVM_XEN_HVM_CONFIG_RUNSTATE; break; +#endif case KVM_CAP_SYNC_REGS: r = KVM_SYNC_X86_VALID_FIELDS; break; @@ -4038,7 +4047,11 @@ void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu) if (vcpu->preempted && !vcpu->arch.guest_state_protected) vcpu->arch.preempted_in_kernel = !static_call(kvm_x86_get_cpl)(vcpu); - kvm_steal_time_set_preempted(vcpu); + if (kvm_xen_msr_enabled(vcpu->kvm)) + kvm_xen_runstate_set_preempted(vcpu); + else + kvm_steal_time_set_preempted(vcpu); + static_call(kvm_x86_vcpu_put)(vcpu); vcpu->arch.last_host_tsc = rdtsc(); /* @@ -5013,6 +5026,7 @@ long kvm_arch_vcpu_ioctl(struct file *filp, case KVM_GET_SUPPORTED_HV_CPUID: r = kvm_ioctl_get_supported_hv_cpuid(vcpu, argp); break; +#ifdef CONFIG_KVM_XEN case KVM_XEN_VCPU_GET_ATTR: { struct kvm_xen_vcpu_attr xva; @@ -5033,6 +5047,7 @@ long kvm_arch_vcpu_ioctl(struct file *filp, r = kvm_xen_vcpu_set_attr(vcpu, &xva); break; } +#endif default: r = -EINVAL; } @@ -5654,6 +5669,7 @@ set_pit2_out: kvm->arch.bsp_vcpu_id = arg; mutex_unlock(&kvm->lock); break; +#ifdef CONFIG_KVM_XEN case KVM_XEN_HVM_CONFIG: { struct kvm_xen_hvm_config xhc; r = -EFAULT; @@ -5682,6 +5698,7 @@ set_pit2_out: r = kvm_xen_hvm_set_attr(kvm, &xha); break; } +#endif case KVM_SET_CLOCK: { struct kvm_clock_data user_ns; u64 now_ns; @@ -8040,7 +8057,10 @@ void kvm_arch_exit(void) kvm_mmu_module_exit(); free_percpu(user_return_msrs); kmem_cache_destroy(x86_fpu_cache); +#ifdef CONFIG_KVM_XEN + static_key_deferred_flush(&kvm_xen_enabled); WARN_ON(static_branch_unlikely(&kvm_xen_enabled.key)); +#endif } static int __kvm_vcpu_halt(struct kvm_vcpu *vcpu, int state, int reason) diff --git a/arch/x86/kvm/xen.c b/arch/x86/kvm/xen.c index af8f6562fce4..ae17250e1efe 100644 --- a/arch/x86/kvm/xen.c +++ b/arch/x86/kvm/xen.c @@ -11,9 +11,11 @@ #include "hyperv.h" #include <linux/kvm_host.h> +#include <linux/sched/stat.h> #include <trace/events/kvm.h> #include <xen/interface/xen.h> +#include <xen/interface/vcpu.h> #include "trace.h" @@ -61,6 +63,132 @@ out: return ret; } +static void kvm_xen_update_runstate(struct kvm_vcpu *v, int state) +{ + struct kvm_vcpu_xen *vx = &v->arch.xen; + u64 now = get_kvmclock_ns(v->kvm); + u64 delta_ns = now - vx->runstate_entry_time; + u64 run_delay = current->sched_info.run_delay; + + if (unlikely(!vx->runstate_entry_time)) + vx->current_runstate = RUNSTATE_offline; + + /* + * Time waiting for the scheduler isn't "stolen" if the + * vCPU wasn't running anyway. + */ + if (vx->current_runstate == RUNSTATE_running) { + u64 steal_ns = run_delay - vx->last_steal; + + delta_ns -= steal_ns; + + vx->runstate_times[RUNSTATE_runnable] += steal_ns; + } + vx->last_steal = run_delay; + + vx->runstate_times[vx->current_runstate] += delta_ns; + vx->current_runstate = state; + vx->runstate_entry_time = now; +} + +void kvm_xen_update_runstate_guest(struct kvm_vcpu *v, int state) +{ + struct kvm_vcpu_xen *vx = &v->arch.xen; + uint64_t state_entry_time; + unsigned int offset; + + kvm_xen_update_runstate(v, state); + + if (!vx->runstate_set) + return; + + BUILD_BUG_ON(sizeof(struct compat_vcpu_runstate_info) != 0x2c); + + offset = offsetof(struct compat_vcpu_runstate_info, state_entry_time); +#ifdef CONFIG_X86_64 + /* + * The only difference is alignment of uint64_t in 32-bit. + * So the first field 'state' is accessed directly using + * offsetof() (where its offset happens to be zero), while the + * remaining fields which are all uint64_t, start at 'offset' + * which we tweak here by adding 4. + */ + BUILD_BUG_ON(offsetof(struct vcpu_runstate_info, state_entry_time) != + offsetof(struct compat_vcpu_runstate_info, state_entry_time) + 4); + BUILD_BUG_ON(offsetof(struct vcpu_runstate_info, time) != + offsetof(struct compat_vcpu_runstate_info, time) + 4); + + if (v->kvm->arch.xen.long_mode) + offset = offsetof(struct vcpu_runstate_info, state_entry_time); +#endif + /* + * First write the updated state_entry_time at the appropriate + * location determined by 'offset'. + */ + state_entry_time = vx->runstate_entry_time; + state_entry_time |= XEN_RUNSTATE_UPDATE; + + BUILD_BUG_ON(sizeof(((struct vcpu_runstate_info *)0)->state_entry_time) != + sizeof(state_entry_time)); + BUILD_BUG_ON(sizeof(((struct compat_vcpu_runstate_info *)0)->state_entry_time) != + sizeof(state_entry_time)); + + if (kvm_write_guest_offset_cached(v->kvm, &v->arch.xen.runstate_cache, + &state_entry_time, offset, + sizeof(state_entry_time))) + return; + smp_wmb(); + + /* + * Next, write the new runstate. This is in the *same* place + * for 32-bit and 64-bit guests, asserted here for paranoia. + */ + BUILD_BUG_ON(offsetof(struct vcpu_runstate_info, state) != + offsetof(struct compat_vcpu_runstate_info, state)); + BUILD_BUG_ON(sizeof(((struct vcpu_runstate_info *)0)->state) != + sizeof(vx->current_runstate)); + BUILD_BUG_ON(sizeof(((struct compat_vcpu_runstate_info *)0)->state) != + sizeof(vx->current_runstate)); + + if (kvm_write_guest_offset_cached(v->kvm, &v->arch.xen.runstate_cache, + &vx->current_runstate, + offsetof(struct vcpu_runstate_info, state), + sizeof(vx->current_runstate))) + return; + + /* + * Write the actual runstate times immediately after the + * runstate_entry_time. + */ + BUILD_BUG_ON(offsetof(struct vcpu_runstate_info, state_entry_time) != + offsetof(struct vcpu_runstate_info, time) - sizeof(u64)); + BUILD_BUG_ON(offsetof(struct compat_vcpu_runstate_info, state_entry_time) != + offsetof(struct compat_vcpu_runstate_info, time) - sizeof(u64)); + BUILD_BUG_ON(sizeof(((struct vcpu_runstate_info *)0)->time) != + sizeof(((struct compat_vcpu_runstate_info *)0)->time)); + BUILD_BUG_ON(sizeof(((struct vcpu_runstate_info *)0)->time) != + sizeof(vx->runstate_times)); + + if (kvm_write_guest_offset_cached(v->kvm, &v->arch.xen.runstate_cache, + &vx->runstate_times[0], + offset + sizeof(u64), + sizeof(vx->runstate_times))) + return; + + smp_wmb(); + + /* + * Finally, clear the XEN_RUNSTATE_UPDATE bit in the guest's + * runstate_entry_time field. + */ + + state_entry_time &= ~XEN_RUNSTATE_UPDATE; + if (kvm_write_guest_offset_cached(v->kvm, &v->arch.xen.runstate_cache, + &state_entry_time, offset, + sizeof(state_entry_time))) + return; +} + int __kvm_xen_has_interrupt(struct kvm_vcpu *v) { u8 rc = 0; @@ -187,9 +315,12 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data) /* No compat necessary here. */ BUILD_BUG_ON(sizeof(struct vcpu_info) != sizeof(struct compat_vcpu_info)); + BUILD_BUG_ON(offsetof(struct vcpu_info, time) != + offsetof(struct compat_vcpu_info, time)); if (data->u.gpa == GPA_INVALID) { vcpu->arch.xen.vcpu_info_set = false; + r = 0; break; } @@ -206,6 +337,7 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data) case KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO: if (data->u.gpa == GPA_INVALID) { vcpu->arch.xen.vcpu_time_info_set = false; + r = 0; break; } @@ -219,6 +351,121 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data) } break; + case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR: + if (!sched_info_on()) { + r = -EOPNOTSUPP; + break; + } + if (data->u.gpa == GPA_INVALID) { + vcpu->arch.xen.runstate_set = false; + r = 0; + break; + } + + r = kvm_gfn_to_hva_cache_init(vcpu->kvm, + &vcpu->arch.xen.runstate_cache, + data->u.gpa, + sizeof(struct vcpu_runstate_info)); + if (!r) { + vcpu->arch.xen.runstate_set = true; + } + break; + + case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT: + if (!sched_info_on()) { + r = -EOPNOTSUPP; + break; + } + if (data->u.runstate.state > RUNSTATE_offline) { + r = -EINVAL; + break; + } + + kvm_xen_update_runstate(vcpu, data->u.runstate.state); + r = 0; + break; + + case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA: + if (!sched_info_on()) { + r = -EOPNOTSUPP; + break; + } + if (data->u.runstate.state > RUNSTATE_offline) { + r = -EINVAL; + break; + } + if (data->u.runstate.state_entry_time != + (data->u.runstate.time_running + + data->u.runstate.time_runnable + + data->u.runstate.time_blocked + + data->u.runstate.time_offline)) { + r = -EINVAL; + break; + } + if (get_kvmclock_ns(vcpu->kvm) < + data->u.runstate.state_entry_time) { + r = -EINVAL; + break; + } + + vcpu->arch.xen.current_runstate = data->u.runstate.state; + vcpu->arch.xen.runstate_entry_time = + data->u.runstate.state_entry_time; + vcpu->arch.xen.runstate_times[RUNSTATE_running] = + data->u.runstate.time_running; + vcpu->arch.xen.runstate_times[RUNSTATE_runnable] = + data->u.runstate.time_runnable; + vcpu->arch.xen.runstate_times[RUNSTATE_blocked] = + data->u.runstate.time_blocked; + vcpu->arch.xen.runstate_times[RUNSTATE_offline] = + data->u.runstate.time_offline; + vcpu->arch.xen.last_steal = current->sched_info.run_delay; + r = 0; + break; + + case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST: + if (!sched_info_on()) { + r = -EOPNOTSUPP; + break; + } + if (data->u.runstate.state > RUNSTATE_offline && + data->u.runstate.state != (u64)-1) { + r = -EINVAL; + break; + } + /* The adjustment must add up */ + if (data->u.runstate.state_entry_time != + (data->u.runstate.time_running + + data->u.runstate.time_runnable + + data->u.runstate.time_blocked + + data->u.runstate.time_offline)) { + r = -EINVAL; + break; + } + + if (get_kvmclock_ns(vcpu->kvm) < + (vcpu->arch.xen.runstate_entry_time + + data->u.runstate.state_entry_time)) { + r = -EINVAL; + break; + } + + vcpu->arch.xen.runstate_entry_time += + data->u.runstate.state_entry_time; + vcpu->arch.xen.runstate_times[RUNSTATE_running] += + data->u.runstate.time_running; + vcpu->arch.xen.runstate_times[RUNSTATE_runnable] += + data->u.runstate.time_runnable; + vcpu->arch.xen.runstate_times[RUNSTATE_blocked] += + data->u.runstate.time_blocked; + vcpu->arch.xen.runstate_times[RUNSTATE_offline] += + data->u.runstate.time_offline; + + if (data->u.runstate.state <= RUNSTATE_offline) + kvm_xen_update_runstate(vcpu, data->u.runstate.state); + r = 0; + break; + default: break; } @@ -251,6 +498,49 @@ int kvm_xen_vcpu_get_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data) r = 0; break; + case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR: + if (!sched_info_on()) { + r = -EOPNOTSUPP; + break; + } + if (vcpu->arch.xen.runstate_set) { + data->u.gpa = vcpu->arch.xen.runstate_cache.gpa; + r = 0; + } + break; + + case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT: + if (!sched_info_on()) { + r = -EOPNOTSUPP; + break; + } + data->u.runstate.state = vcpu->arch.xen.current_runstate; + r = 0; + break; + + case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA: + if (!sched_info_on()) { + r = -EOPNOTSUPP; + break; + } + data->u.runstate.state = vcpu->arch.xen.current_runstate; + data->u.runstate.state_entry_time = + vcpu->arch.xen.runstate_entry_time; + data->u.runstate.time_running = + vcpu->arch.xen.runstate_times[RUNSTATE_running]; + data->u.runstate.time_runnable = + vcpu->arch.xen.runstate_times[RUNSTATE_runnable]; + data->u.runstate.time_blocked = + vcpu->arch.xen.runstate_times[RUNSTATE_blocked]; + data->u.runstate.time_offline = + vcpu->arch.xen.runstate_times[RUNSTATE_offline]; + r = 0; + break; + + case KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST: + r = -EINVAL; + break; + default: break; } diff --git a/arch/x86/kvm/xen.h b/arch/x86/kvm/xen.h index b66a921776f4..463a7844a8ca 100644 --- a/arch/x86/kvm/xen.h +++ b/arch/x86/kvm/xen.h @@ -9,6 +9,7 @@ #ifndef __ARCH_X86_KVM_XEN_H__ #define __ARCH_X86_KVM_XEN_H__ +#ifdef CONFIG_KVM_XEN #include <linux/jump_label_ratelimit.h> extern struct static_key_false_deferred kvm_xen_enabled; @@ -18,11 +19,16 @@ int kvm_xen_vcpu_set_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data) int kvm_xen_vcpu_get_attr(struct kvm_vcpu *vcpu, struct kvm_xen_vcpu_attr *data); int kvm_xen_hvm_set_attr(struct kvm *kvm, struct kvm_xen_hvm_attr *data); int kvm_xen_hvm_get_attr(struct kvm *kvm, struct kvm_xen_hvm_attr *data); -int kvm_xen_hypercall(struct kvm_vcpu *vcpu); int kvm_xen_write_hypercall_page(struct kvm_vcpu *vcpu, u64 data); int kvm_xen_hvm_config(struct kvm *kvm, struct kvm_xen_hvm_config *xhc); void kvm_xen_destroy_vm(struct kvm *kvm); +static inline bool kvm_xen_msr_enabled(struct kvm *kvm) +{ + return static_branch_unlikely(&kvm_xen_enabled.key) && + kvm->arch.xen_hvm_config.msr; +} + static inline bool kvm_xen_hypercall_enabled(struct kvm *kvm) { return static_branch_unlikely(&kvm_xen_enabled.key) && @@ -38,11 +44,59 @@ static inline int kvm_xen_has_interrupt(struct kvm_vcpu *vcpu) return 0; } +#else +static inline int kvm_xen_write_hypercall_page(struct kvm_vcpu *vcpu, u64 data) +{ + return 1; +} + +static inline void kvm_xen_destroy_vm(struct kvm *kvm) +{ +} + +static inline bool kvm_xen_msr_enabled(struct kvm *kvm) +{ + return false; +} + +static inline bool kvm_xen_hypercall_enabled(struct kvm *kvm) +{ + return false; +} + +static inline int kvm_xen_has_interrupt(struct kvm_vcpu *vcpu) +{ + return 0; +} +#endif + +int kvm_xen_hypercall(struct kvm_vcpu *vcpu); -/* 32-bit compatibility definitions, also used natively in 32-bit build */ #include <asm/pvclock-abi.h> #include <asm/xen/interface.h> +#include <xen/interface/vcpu.h> + +void kvm_xen_update_runstate_guest(struct kvm_vcpu *vcpu, int state); +static inline void kvm_xen_runstate_set_running(struct kvm_vcpu *vcpu) +{ + kvm_xen_update_runstate_guest(vcpu, RUNSTATE_running); +} + +static inline void kvm_xen_runstate_set_preempted(struct kvm_vcpu *vcpu) +{ + /* + * If the vCPU wasn't preempted but took a normal exit for + * some reason (hypercalls, I/O, etc.), that is accounted as + * still RUNSTATE_running, as the VMM is still operating on + * behalf of the vCPU. Only if the VMM does actually block + * does it need to enter RUNSTATE_blocked. + */ + if (vcpu->preempted) + kvm_xen_update_runstate_guest(vcpu, RUNSTATE_runnable); +} + +/* 32-bit compatibility definitions, also used natively in 32-bit build */ struct compat_arch_vcpu_info { unsigned int cr2; unsigned int pad[5]; @@ -75,4 +129,10 @@ struct compat_shared_info { struct compat_arch_shared_info arch; }; +struct compat_vcpu_runstate_info { + int state; + uint64_t state_entry_time; + uint64_t time[4]; +} __attribute__((packed)); + #endif /* __ARCH_X86_KVM_XEN_H__ */ diff --git a/include/uapi/linux/kvm.h b/include/uapi/linux/kvm.h index 8b281f722e5b..f6afee209620 100644 --- a/include/uapi/linux/kvm.h +++ b/include/uapi/linux/kvm.h @@ -1154,6 +1154,7 @@ struct kvm_x86_mce { #define KVM_XEN_HVM_CONFIG_HYPERCALL_MSR (1 << 0) #define KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL (1 << 1) #define KVM_XEN_HVM_CONFIG_SHARED_INFO (1 << 2) +#define KVM_XEN_HVM_CONFIG_RUNSTATE (1 << 3) struct kvm_xen_hvm_config { __u32 flags; @@ -1621,12 +1622,24 @@ struct kvm_xen_vcpu_attr { union { __u64 gpa; __u64 pad[8]; + struct { + __u64 state; + __u64 state_entry_time; + __u64 time_running; + __u64 time_runnable; + __u64 time_blocked; + __u64 time_offline; + } runstate; } u; }; /* Available with KVM_CAP_XEN_HVM / KVM_XEN_HVM_CONFIG_SHARED_INFO */ #define KVM_XEN_VCPU_ATTR_TYPE_VCPU_INFO 0x0 #define KVM_XEN_VCPU_ATTR_TYPE_VCPU_TIME_INFO 0x1 +#define KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR 0x2 +#define KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT 0x3 +#define KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA 0x4 +#define KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST 0x5 /* Secure Encrypted Virtualization command */ enum sev_cmd_id { diff --git a/tools/testing/selftests/kvm/lib/kvm_util.c b/tools/testing/selftests/kvm/lib/kvm_util.c index d787cb802b4a..e5fbf16f725b 100644 --- a/tools/testing/selftests/kvm/lib/kvm_util.c +++ b/tools/testing/selftests/kvm/lib/kvm_util.c @@ -21,6 +21,8 @@ #define KVM_UTIL_PGS_PER_HUGEPG 512 #define KVM_UTIL_MIN_PFN 2 +static int vcpu_mmap_sz(void); + /* Aligns x up to the next multiple of size. Size must be a power of 2. */ static void *align(void *x, size_t size) { @@ -509,7 +511,7 @@ static void vm_vcpu_rm(struct kvm_vm *vm, struct vcpu *vcpu) vcpu->dirty_gfns = NULL; } - ret = munmap(vcpu->state, sizeof(*vcpu->state)); + ret = munmap(vcpu->state, vcpu_mmap_sz()); TEST_ASSERT(ret == 0, "munmap of VCPU fd failed, rc: %i " "errno: %i", ret, errno); close(vcpu->fd); @@ -978,7 +980,7 @@ void vm_vcpu_add(struct kvm_vm *vm, uint32_t vcpuid) TEST_ASSERT(vcpu_mmap_sz() >= sizeof(*vcpu->state), "vcpu mmap size " "smaller than expected, vcpu_mmap_sz: %i expected_min: %zi", vcpu_mmap_sz(), sizeof(*vcpu->state)); - vcpu->state = (struct kvm_run *) mmap(NULL, sizeof(*vcpu->state), + vcpu->state = (struct kvm_run *) mmap(NULL, vcpu_mmap_sz(), PROT_READ | PROT_WRITE, MAP_SHARED, vcpu->fd, 0); TEST_ASSERT(vcpu->state != MAP_FAILED, "mmap vcpu_state failed, " "vcpu id: %u errno: %i", vcpuid, errno); diff --git a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c index 9246ea310587..804ff5ff022d 100644 --- a/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c +++ b/tools/testing/selftests/kvm/x86_64/xen_shinfo_test.c @@ -13,19 +13,27 @@ #include <stdint.h> #include <time.h> +#include <sched.h> +#include <sys/syscall.h> #define VCPU_ID 5 +#define SHINFO_REGION_GVA 0xc0000000ULL #define SHINFO_REGION_GPA 0xc0000000ULL #define SHINFO_REGION_SLOT 10 #define PAGE_SIZE 4096 #define PVTIME_ADDR (SHINFO_REGION_GPA + PAGE_SIZE) +#define RUNSTATE_ADDR (SHINFO_REGION_GPA + PAGE_SIZE + 0x20) + +#define RUNSTATE_VADDR (SHINFO_REGION_GVA + PAGE_SIZE + 0x20) static struct kvm_vm *vm; #define XEN_HYPERCALL_MSR 0x40000000 +#define MIN_STEAL_TIME 50000 + struct pvclock_vcpu_time_info { u32 version; u32 pad0; @@ -43,11 +51,67 @@ struct pvclock_wall_clock { u32 nsec; } __attribute__((__packed__)); +struct vcpu_runstate_info { + uint32_t state; + uint64_t state_entry_time; + uint64_t time[4]; +}; + +#define RUNSTATE_running 0 +#define RUNSTATE_runnable 1 +#define RUNSTATE_blocked 2 +#define RUNSTATE_offline 3 + static void guest_code(void) { + struct vcpu_runstate_info *rs = (void *)RUNSTATE_VADDR; + + /* Test having the host set runstates manually */ + GUEST_SYNC(RUNSTATE_runnable); + GUEST_ASSERT(rs->time[RUNSTATE_runnable] != 0); + GUEST_ASSERT(rs->state == 0); + + GUEST_SYNC(RUNSTATE_blocked); + GUEST_ASSERT(rs->time[RUNSTATE_blocked] != 0); + GUEST_ASSERT(rs->state == 0); + + GUEST_SYNC(RUNSTATE_offline); + GUEST_ASSERT(rs->time[RUNSTATE_offline] != 0); + GUEST_ASSERT(rs->state == 0); + + /* Test runstate time adjust */ + GUEST_SYNC(4); + GUEST_ASSERT(rs->time[RUNSTATE_blocked] == 0x5a); + GUEST_ASSERT(rs->time[RUNSTATE_offline] == 0x6b6b); + + /* Test runstate time set */ + GUEST_SYNC(5); + GUEST_ASSERT(rs->state_entry_time >= 0x8000); + GUEST_ASSERT(rs->time[RUNSTATE_runnable] == 0); + GUEST_ASSERT(rs->time[RUNSTATE_blocked] == 0x6b6b); + GUEST_ASSERT(rs->time[RUNSTATE_offline] == 0x5a); + + /* sched_yield() should result in some 'runnable' time */ + GUEST_SYNC(6); + GUEST_ASSERT(rs->time[RUNSTATE_runnable] >= MIN_STEAL_TIME); + GUEST_DONE(); } +static long get_run_delay(void) +{ + char path[64]; + long val[2]; + FILE *fp; + + sprintf(path, "/proc/%ld/schedstat", syscall(SYS_gettid)); + fp = fopen(path, "r"); + fscanf(fp, "%ld %ld ", &val[0], &val[1]); + fclose(fp); + + return val[1]; +} + static int cmp_timespec(struct timespec *a, struct timespec *b) { if (a->tv_sec > b->tv_sec) @@ -66,12 +130,14 @@ int main(int argc, char *argv[]) { struct timespec min_ts, max_ts, vm_ts; - if (!(kvm_check_cap(KVM_CAP_XEN_HVM) & - KVM_XEN_HVM_CONFIG_SHARED_INFO) ) { + int xen_caps = kvm_check_cap(KVM_CAP_XEN_HVM); + if (!(xen_caps & KVM_XEN_HVM_CONFIG_SHARED_INFO) ) { print_skip("KVM_XEN_HVM_CONFIG_SHARED_INFO not available"); exit(KSFT_SKIP); } + bool do_runstate_tests = !!(xen_caps & KVM_XEN_HVM_CONFIG_RUNSTATE); + clock_gettime(CLOCK_REALTIME, &min_ts); vm = vm_create_default(VCPU_ID, 0, (void *) guest_code); @@ -80,6 +146,7 @@ int main(int argc, char *argv[]) /* Map a region for the shared_info page */ vm_userspace_mem_region_add(vm, VM_MEM_SRC_ANONYMOUS, SHINFO_REGION_GPA, SHINFO_REGION_SLOT, 2, 0); + virt_map(vm, SHINFO_REGION_GVA, SHINFO_REGION_GPA, 2, 0); struct kvm_xen_hvm_config hvmc = { .flags = KVM_XEN_HVM_CONFIG_INTERCEPT_HCALL, @@ -111,6 +178,17 @@ int main(int argc, char *argv[]) }; vcpu_ioctl(vm, VCPU_ID, KVM_XEN_VCPU_SET_ATTR, &pvclock); + if (do_runstate_tests) { + struct kvm_xen_vcpu_attr st = { + .type = KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADDR, + .u.gpa = RUNSTATE_ADDR, + }; + vcpu_ioctl(vm, VCPU_ID, KVM_XEN_VCPU_SET_ATTR, &st); + } + + struct vcpu_runstate_info *rs = addr_gpa2hva(vm, RUNSTATE_ADDR);; + rs->state = 0x5a; + for (;;) { volatile struct kvm_run *run = vcpu_state(vm, VCPU_ID); struct ucall uc; @@ -126,8 +204,56 @@ int main(int argc, char *argv[]) case UCALL_ABORT: TEST_FAIL("%s", (const char *)uc.args[0]); /* NOT REACHED */ - case UCALL_SYNC: + case UCALL_SYNC: { + struct kvm_xen_vcpu_attr rst; + long rundelay; + + /* If no runstate support, bail out early */ + if (!do_runstate_tests) + goto done; + + TEST_ASSERT(rs->state_entry_time == rs->time[0] + + rs->time[1] + rs->time[2] + rs->time[3], + "runstate times don't add up"); + + switch (uc.args[1]) { + case RUNSTATE_running...RUNSTATE_offline: + rst.type = KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_CURRENT; + rst.u.runstate.state = uc.args[1]; + vcpu_ioctl(vm, VCPU_ID, KVM_XEN_VCPU_SET_ATTR, &rst); + break; + case 4: + rst.type = KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_ADJUST; + memset(&rst.u, 0, sizeof(rst.u)); + rst.u.runstate.state = (uint64_t)-1; + rst.u.runstate.time_blocked = + 0x5a - rs->time[RUNSTATE_blocked]; + rst.u.runstate.time_offline = + 0x6b6b - rs->time[RUNSTATE_offline]; + rst.u.runstate.time_runnable = -rst.u.runstate.time_blocked - + rst.u.runstate.time_offline; + vcpu_ioctl(vm, VCPU_ID, KVM_XEN_VCPU_SET_ATTR, &rst); + break; + + case 5: + rst.type = KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA; + memset(&rst.u, 0, sizeof(rst.u)); + rst.u.runstate.state = RUNSTATE_running; + rst.u.runstate.state_entry_time = 0x6b6b + 0x5a; + rst.u.runstate.time_blocked = 0x6b6b; + rst.u.runstate.time_offline = 0x5a; + vcpu_ioctl(vm, VCPU_ID, KVM_XEN_VCPU_SET_ATTR, &rst); + break; + case 6: + /* Yield until scheduler delay exceeds target */ + rundelay = get_run_delay() + MIN_STEAL_TIME; + do { + sched_yield(); + } while (get_run_delay() < rundelay); + break; + } break; + } case UCALL_DONE: goto done; default: @@ -162,6 +288,33 @@ int main(int argc, char *argv[]) TEST_ASSERT(ti2->version && !(ti2->version & 1), "Bad time_info version %x", ti->version); + if (do_runstate_tests) { + /* + * Fetch runstate and check sanity. Strictly speaking in the + * general case we might not expect the numbers to be identical + * but in this case we know we aren't running the vCPU any more. + */ + struct kvm_xen_vcpu_attr rst = { + .type = KVM_XEN_VCPU_ATTR_TYPE_RUNSTATE_DATA, + }; + vcpu_ioctl(vm, VCPU_ID, KVM_XEN_VCPU_GET_ATTR, &rst); + + TEST_ASSERT(rs->state == rst.u.runstate.state, "Runstate mismatch"); + TEST_ASSERT(rs->state_entry_time == rst.u.runstate.state_entry_time, + "State entry time mismatch"); + TEST_ASSERT(rs->time[RUNSTATE_running] == rst.u.runstate.time_running, + "Running time mismatch"); + TEST_ASSERT(rs->time[RUNSTATE_runnable] == rst.u.runstate.time_runnable, + "Runnable time mismatch"); + TEST_ASSERT(rs->time[RUNSTATE_blocked] == rst.u.runstate.time_blocked, + "Blocked time mismatch"); + TEST_ASSERT(rs->time[RUNSTATE_offline] == rst.u.runstate.time_offline, + "Offline time mismatch"); + + TEST_ASSERT(rs->state_entry_time == rs->time[0] + + rs->time[1] + rs->time[2] + rs->time[3], + "runstate times don't add up"); + } kvm_vm_free(vm); return 0; } |