Hi, Bibo,
On Fri, May 24, 2024 at 3:38 PM Bibo Mao <maobibo@xxxxxxxxxxx> wrote:
>
> Steal time feature is added here in kvm side, VM can search supported
> features provided by KVM hypervisor, feature KVM_FEATURE_STEAL_TIME
> is added here. Like x86, steal time structure is saved in guest memory,
> one hypercall function KVM_HCALL_FUNC_NOTIFY is added to notify KVM to
> enable the feature.
>
> One cpu attr ioctl command KVM_LOONGARCH_VCPU_PVTIME_CTRL is added to
> save and restore base address of steal time structure when VM is migrated.
>
> Signed-off-by: Bibo Mao <maobibo@xxxxxxxxxxx>
> ---
> arch/loongarch/include/asm/kvm_host.h | 7 ++
> arch/loongarch/include/asm/kvm_para.h | 10 ++
> arch/loongarch/include/asm/kvm_vcpu.h | 4 +
> arch/loongarch/include/asm/loongarch.h | 1 +
> arch/loongarch/include/uapi/asm/kvm.h | 4 +
> arch/loongarch/kvm/Kconfig | 1 +
> arch/loongarch/kvm/exit.c | 38 +++++++-
> arch/loongarch/kvm/vcpu.c | 124 +++++++++++++++++++++++++
> 8 files changed, 187 insertions(+), 2 deletions(-)
>
> diff --git a/arch/loongarch/include/asm/kvm_host.h b/arch/loongarch/include/asm/kvm_host.h
> index c87b6ea0ec47..2eb2f7572023 100644
> --- a/arch/loongarch/include/asm/kvm_host.h
> +++ b/arch/loongarch/include/asm/kvm_host.h
> @@ -30,6 +30,7 @@
> #define KVM_PRIVATE_MEM_SLOTS 0
>
> #define KVM_HALT_POLL_NS_DEFAULT 500000
> +#define KVM_REQ_STEAL_UPDATE KVM_ARCH_REQ(1)
>
> #define KVM_GUESTDBG_SW_BP_MASK \
> (KVM_GUESTDBG_ENABLE | KVM_GUESTDBG_USE_SW_BP)
> @@ -201,6 +202,12 @@ struct kvm_vcpu_arch {
> struct kvm_mp_state mp_state;
> /* cpucfg */
> u32 cpucfg[KVM_MAX_CPUCFG_REGS];
> + /* paravirt steal time */
> + struct {
> + u64 guest_addr;
> + u64 last_steal;
> + struct gfn_to_hva_cache cache;
> + } st;
> };
>
> static inline unsigned long readl_sw_gcsr(struct loongarch_csrs *csr, int reg)
> diff --git a/arch/loongarch/include/asm/kvm_para.h b/arch/loongarch/include/asm/kvm_para.h
> index 4ba2312e5f8c..a9ba8185d4af 100644
> --- a/arch/loongarch/include/asm/kvm_para.h
> +++ b/arch/loongarch/include/asm/kvm_para.h
> @@ -14,6 +14,7 @@
>
> #define KVM_HCALL_SERVICE HYPERCALL_ENCODE(HYPERVISOR_KVM, KVM_HCALL_CODE_SERVICE)
> #define KVM_HCALL_FUNC_IPI 1
> +#define KVM_HCALL_FUNC_NOTIFY 2
>
> #define KVM_HCALL_SWDBG HYPERCALL_ENCODE(HYPERVISOR_KVM, KVM_HCALL_CODE_SWDBG)
>
> @@ -24,6 +25,15 @@
> #define KVM_HCALL_INVALID_CODE -1UL
> #define KVM_HCALL_INVALID_PARAMETER -2UL
>
> +#define KVM_STEAL_PHYS_VALID BIT_ULL(0)
> +#define KVM_STEAL_PHYS_MASK GENMASK_ULL(63, 6)
> +struct kvm_steal_time {
> + __u64 steal;
> + __u32 version;
> + __u32 flags;
I found that x86 has a preempted field here, in our internal repo the
LoongArch version also has this field. Moreover,
kvm_steal_time_set_preempted() and kvm_steal_time_clear_preempted()
seems needed.
> + __u32 pad[12];
> +};
> +
> /*
> * Hypercall interface for KVM hypervisor
> *
> diff --git a/arch/loongarch/include/asm/kvm_vcpu.h b/arch/loongarch/include/asm/kvm_vcpu.h
> index 590a92cb5416..d7e51300a89f 100644
> --- a/arch/loongarch/include/asm/kvm_vcpu.h
> +++ b/arch/loongarch/include/asm/kvm_vcpu.h
> @@ -120,4 +120,8 @@ static inline void kvm_write_reg(struct kvm_vcpu *vcpu, int num, unsigned long v
> vcpu->arch.gprs[num] = val;
> }
>
> +static inline bool kvm_pvtime_supported(void)
> +{
> + return !!sched_info_on();
> +}
> #endif /* __ASM_LOONGARCH_KVM_VCPU_H__ */
> diff --git a/arch/loongarch/include/asm/loongarch.h b/arch/loongarch/include/asm/loongarch.h
> index eb09adda54b7..7a4633ef284b 100644
> --- a/arch/loongarch/include/asm/loongarch.h
> +++ b/arch/loongarch/include/asm/loongarch.h
> @@ -169,6 +169,7 @@
> #define KVM_SIGNATURE "KVM\0"
> #define CPUCFG_KVM_FEATURE (CPUCFG_KVM_BASE + 4)
> #define KVM_FEATURE_IPI BIT(1)
> +#define KVM_FEATURE_STEAL_TIME BIT(2)
>
> #ifndef __ASSEMBLY__
>
> diff --git a/arch/loongarch/include/uapi/asm/kvm.h b/arch/loongarch/include/uapi/asm/kvm.h
> index f9abef382317..ddc5cab0ffd0 100644
> --- a/arch/loongarch/include/uapi/asm/kvm.h
> +++ b/arch/loongarch/include/uapi/asm/kvm.h
> @@ -81,7 +81,11 @@ 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)
> +
> +/* Device Control API on vcpu fd */
> #define KVM_LOONGARCH_VCPU_CPUCFG 0
> +#define KVM_LOONGARCH_VCPU_PVTIME_CTRL 1
> +#define KVM_LOONGARCH_VCPU_PVTIME_GPA 0
>
> struct kvm_debug_exit_arch {
> };
> diff --git a/arch/loongarch/kvm/Kconfig b/arch/loongarch/kvm/Kconfig
> index c4ef2b4d9797..248744b4d086 100644
> --- a/arch/loongarch/kvm/Kconfig
> +++ b/arch/loongarch/kvm/Kconfig
> @@ -29,6 +29,7 @@ config KVM
> select KVM_MMIO
> select HAVE_KVM_READONLY_MEM
> select KVM_XFER_TO_GUEST_WORK
> + select SCHED_INFO
> help
> Support hosting virtualized guest machines using
> hardware virtualization extensions. You will need
> diff --git a/arch/loongarch/kvm/exit.c b/arch/loongarch/kvm/exit.c
> index c86e099af5ca..e2abd97fb13f 100644
> --- a/arch/loongarch/kvm/exit.c
> +++ b/arch/loongarch/kvm/exit.c
> @@ -24,7 +24,7 @@
> static int kvm_emu_cpucfg(struct kvm_vcpu *vcpu, larch_inst inst)
> {
> int rd, rj;
> - unsigned int index;
> + unsigned int index, ret;
>
> if (inst.reg2_format.opcode != cpucfg_op)
> return EMULATE_FAIL;
> @@ -50,7 +50,10 @@ static int kvm_emu_cpucfg(struct kvm_vcpu *vcpu, larch_inst inst)
> vcpu->arch.gprs[rd] = *(unsigned int *)KVM_SIGNATURE;
> break;
> case CPUCFG_KVM_FEATURE:
> - vcpu->arch.gprs[rd] = KVM_FEATURE_IPI;
> + ret = KVM_FEATURE_IPI;
> + if (sched_info_on())
What about replacing it with your helper function kvm_pvtime_supported()?
Huacai
> + ret |= KVM_FEATURE_STEAL_TIME;
> + vcpu->arch.gprs[rd] = ret;
> break;
> default:
> vcpu->arch.gprs[rd] = 0;
> @@ -687,6 +690,34 @@ static int kvm_handle_fpu_disabled(struct kvm_vcpu *vcpu)
> return RESUME_GUEST;
> }
>
> +static long kvm_save_notify(struct kvm_vcpu *vcpu)
> +{
> + unsigned long id, data;
> +
> + id = kvm_read_reg(vcpu, LOONGARCH_GPR_A1);
> + data = kvm_read_reg(vcpu, LOONGARCH_GPR_A2);
> + switch (id) {
> + case KVM_FEATURE_STEAL_TIME:
> + if (!kvm_pvtime_supported())
> + return KVM_HCALL_INVALID_CODE;
> +
> + if (data & ~(KVM_STEAL_PHYS_MASK | KVM_STEAL_PHYS_VALID))
> + return KVM_HCALL_INVALID_PARAMETER;
> +
> + vcpu->arch.st.guest_addr = data;
> + if (!(data & KVM_STEAL_PHYS_VALID))
> + break;
> +
> + vcpu->arch.st.last_steal = current->sched_info.run_delay;
> + kvm_make_request(KVM_REQ_STEAL_UPDATE, vcpu);
> + break;
> + default:
> + break;
> + };
> +
> + return 0;
> +};
> +
> /*
> * kvm_handle_lsx_disabled() - Guest used LSX while disabled in root.
> * @vcpu: Virtual CPU context.
> @@ -758,6 +789,9 @@ static void kvm_handle_service(struct kvm_vcpu *vcpu)
> kvm_send_pv_ipi(vcpu);
> ret = KVM_HCALL_SUCCESS;
> break;
> + case KVM_HCALL_FUNC_NOTIFY:
> + ret = kvm_save_notify(vcpu);
> + break;
> default:
> ret = KVM_HCALL_INVALID_CODE;
> break;
> diff --git a/arch/loongarch/kvm/vcpu.c b/arch/loongarch/kvm/vcpu.c
> index 9e8030d45129..382796f1d3e6 100644
> --- a/arch/loongarch/kvm/vcpu.c
> +++ b/arch/loongarch/kvm/vcpu.c
> @@ -31,6 +31,117 @@ const struct kvm_stats_header kvm_vcpu_stats_header = {
> sizeof(kvm_vcpu_stats_desc),
> };
>
> +static void kvm_update_stolen_time(struct kvm_vcpu *vcpu)
> +{
> + struct kvm_steal_time __user *st;
> + struct gfn_to_hva_cache *ghc;
> + struct kvm_memslots *slots;
> + gpa_t gpa;
> + u64 steal;
> + u32 version;
> +
> + ghc = &vcpu->arch.st.cache;
> + gpa = vcpu->arch.st.guest_addr;
> + if (!(gpa & KVM_STEAL_PHYS_VALID))
> + return;
> +
> + gpa &= KVM_STEAL_PHYS_MASK;
> + slots = kvm_memslots(vcpu->kvm);
> + if (slots->generation != ghc->generation || gpa != ghc->gpa) {
> + if (kvm_gfn_to_hva_cache_init(vcpu->kvm, ghc, gpa,
> + sizeof(*st))) {
> + ghc->gpa = INVALID_GPA;
> + return;
> + }
> + }
> +
> + st = (struct kvm_steal_time __user *)ghc->hva;
> + unsafe_get_user(version, &st->version, out);
> + if (version & 1)
> + version += 1;
> + version += 1;
> + unsafe_put_user(version, &st->version, out);
> + smp_wmb();
> +
> + unsafe_get_user(steal, &st->steal, out);
> + steal += current->sched_info.run_delay -
> + vcpu->arch.st.last_steal;
> + vcpu->arch.st.last_steal = current->sched_info.run_delay;
> + unsafe_put_user(steal, &st->steal, out);
> +
> + smp_wmb();
> + version += 1;
> + unsafe_put_user(version, &st->version, out);
> +out:
> + mark_page_dirty_in_slot(vcpu->kvm, ghc->memslot, gpa_to_gfn(ghc->gpa));
> +}
> +
> +static int kvm_loongarch_pvtime_has_attr(struct kvm_vcpu *vcpu,
> + struct kvm_device_attr *attr)
> +{
> + if (!kvm_pvtime_supported() ||
> + attr->attr != KVM_LOONGARCH_VCPU_PVTIME_GPA)
> + return -ENXIO;
> +
> + return 0;
> +}
> +
> +static int kvm_loongarch_pvtime_get_attr(struct kvm_vcpu *vcpu,
> + struct kvm_device_attr *attr)
> +{
> + u64 __user *user = (u64 __user *)attr->addr;
> + u64 gpa;
> +
> + if (!kvm_pvtime_supported() ||
> + attr->attr != KVM_LOONGARCH_VCPU_PVTIME_GPA)
> + return -ENXIO;
> +
> + gpa = vcpu->arch.st.guest_addr;
> + if (put_user(gpa, user))
> + return -EFAULT;
> +
> + return 0;
> +}
> +
> +static int kvm_loongarch_pvtime_set_attr(struct kvm_vcpu *vcpu,
> + struct kvm_device_attr *attr)
> +{
> + u64 __user *user = (u64 __user *)attr->addr;
> + struct kvm *kvm = vcpu->kvm;
> + u64 gpa;
> + int ret = 0;
> + int idx;
> +
> + if (!kvm_pvtime_supported() ||
> + attr->attr != KVM_LOONGARCH_VCPU_PVTIME_GPA)
> + return -ENXIO;
> +
> + if (get_user(gpa, user))
> + return -EFAULT;
> +
> + if (gpa & ~(KVM_STEAL_PHYS_MASK | KVM_STEAL_PHYS_VALID))
> + return -EINVAL;
> +
> + if (!(gpa & KVM_STEAL_PHYS_VALID)) {
> + vcpu->arch.st.guest_addr = gpa;
> + return 0;
> + }
> +
> + /* Check the address is in a valid memslot */
> + idx = srcu_read_lock(&kvm->srcu);
> + if (kvm_is_error_hva(gfn_to_hva(kvm, gpa >> PAGE_SHIFT)))
> + ret = -EINVAL;
> + srcu_read_unlock(&kvm->srcu, idx);
> +
> + if (!ret) {
> + vcpu->arch.st.guest_addr = gpa;
> + vcpu->arch.st.last_steal = current->sched_info.run_delay;
> + kvm_make_request(KVM_REQ_STEAL_UPDATE, vcpu);
> + }
> +
> + return ret;
> +}
> +
> /*
> * kvm_check_requests - check and handle pending vCPU requests
> *
> @@ -48,6 +159,9 @@ static int kvm_check_requests(struct kvm_vcpu *vcpu)
> if (kvm_dirty_ring_check_request(vcpu))
> return RESUME_HOST;
>
> + if (kvm_check_request(KVM_REQ_STEAL_UPDATE, vcpu))
> + kvm_update_stolen_time(vcpu);
> +
> return RESUME_GUEST;
> }
>
> @@ -671,6 +785,9 @@ static int kvm_loongarch_vcpu_has_attr(struct kvm_vcpu *vcpu,
> case KVM_LOONGARCH_VCPU_CPUCFG:
> ret = kvm_loongarch_cpucfg_has_attr(vcpu, attr);
> break;
> + case KVM_LOONGARCH_VCPU_PVTIME_CTRL:
> + ret = kvm_loongarch_pvtime_has_attr(vcpu, attr);
> + break;
> default:
> break;
> }
> @@ -703,6 +820,9 @@ static int kvm_loongarch_vcpu_get_attr(struct kvm_vcpu *vcpu,
> case KVM_LOONGARCH_VCPU_CPUCFG:
> ret = kvm_loongarch_get_cpucfg_attr(vcpu, attr);
> break;
> + case KVM_LOONGARCH_VCPU_PVTIME_CTRL:
> + ret = kvm_loongarch_pvtime_get_attr(vcpu, attr);
> + break;
> default:
> break;
> }
> @@ -725,6 +845,9 @@ static int kvm_loongarch_vcpu_set_attr(struct kvm_vcpu *vcpu,
> case KVM_LOONGARCH_VCPU_CPUCFG:
> ret = kvm_loongarch_cpucfg_set_attr(vcpu, attr);
> break;
> + case KVM_LOONGARCH_VCPU_PVTIME_CTRL:
> + ret = kvm_loongarch_pvtime_set_attr(vcpu, attr);
> + break;
> default:
> break;
> }
> @@ -1084,6 +1207,7 @@ static int _kvm_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
>
> /* Control guest page CCA attribute */
> change_csr_gcfg(CSR_GCFG_MATC_MASK, CSR_GCFG_MATC_ROOT);
> + kvm_make_request(KVM_REQ_STEAL_UPDATE, vcpu);
>
> /* Don't bother restoring registers multiple times unless necessary */
> if (vcpu->arch.aux_inuse & KVM_LARCH_HWCSR_USABLE)
> --
> 2.39.3
>
