Re: [RFC PATCH 1/8] kvm: x86: MSR for setting up scheduler info shared memory

[Vitaly Kuznetsov]

"Vineeth Pillai (Google)" <vineeth@xxxxxxxxxxxxxxx> writes:

> Implement a kvm MSR that guest uses to provide the GPA of shared memory
> for communicating the scheduling information between host and guest.
>
> wrmsr(0) disables the feature. wrmsr(valid_gpa) enables the feature and
> uses the gpa for further communication.
>
> Also add a new cpuid feature flag for the host to advertise the feature
> to the guest.
>
> Co-developed-by: Joel Fernandes (Google) <joel@xxxxxxxxxxxxxxxxx>
> Signed-off-by: Joel Fernandes (Google) <joel@xxxxxxxxxxxxxxxxx>
> Signed-off-by: Vineeth Pillai (Google) <vineeth@xxxxxxxxxxxxxxx>
> ---
>  arch/x86/include/asm/kvm_host.h      | 25 ++++++++++++
>  arch/x86/include/uapi/asm/kvm_para.h | 24 +++++++++++
>  arch/x86/kvm/Kconfig                 | 12 ++++++
>  arch/x86/kvm/cpuid.c                 |  2 +
>  arch/x86/kvm/x86.c                   | 61 ++++++++++++++++++++++++++++
>  include/linux/kvm_host.h             |  5 +++
>  6 files changed, 129 insertions(+)
>
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index f72b30d2238a..f89ba1f07d88 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -987,6 +987,18 @@ struct kvm_vcpu_arch {
>  	/* Protected Guests */
>  	bool guest_state_protected;
>  
> +#ifdef CONFIG_PARAVIRT_SCHED_KVM
> +	/*
> +	 * MSR to setup a shared memory for scheduling
> +	 * information sharing between host and guest.
> +	 */
> +	struct {
> +		enum kvm_vcpu_boost_state boost_status;
> +		u64 msr_val;
> +		struct gfn_to_hva_cache data;
> +	} pv_sched;
> +#endif
> +
>  	/*
>  	 * Set when PDPTS were loaded directly by the userspace without
>  	 * reading the guest memory
> @@ -2217,4 +2229,17 @@ int memslot_rmap_alloc(struct kvm_memory_slot *slot, unsigned long npages);
>   */
>  #define KVM_EXIT_HYPERCALL_MBZ		GENMASK_ULL(31, 1)
>  
> +#ifdef CONFIG_PARAVIRT_SCHED_KVM
> +static inline bool kvm_arch_vcpu_pv_sched_enabled(struct kvm_vcpu_arch *arch)
> +{
> +	return arch->pv_sched.msr_val;
> +}
> +
> +static inline void kvm_arch_vcpu_set_boost_status(struct kvm_vcpu_arch *arch,
> +		enum kvm_vcpu_boost_state boost_status)
> +{
> +	arch->pv_sched.boost_status = boost_status;
> +}
> +#endif
> +
>  #endif /* _ASM_X86_KVM_HOST_H */
> diff --git a/arch/x86/include/uapi/asm/kvm_para.h b/arch/x86/include/uapi/asm/kvm_para.h
> index 6e64b27b2c1e..6b1dea07a563 100644
> --- a/arch/x86/include/uapi/asm/kvm_para.h
> +++ b/arch/x86/include/uapi/asm/kvm_para.h
> @@ -36,6 +36,7 @@
>  #define KVM_FEATURE_MSI_EXT_DEST_ID	15
>  #define KVM_FEATURE_HC_MAP_GPA_RANGE	16
>  #define KVM_FEATURE_MIGRATION_CONTROL	17
> +#define KVM_FEATURE_PV_SCHED		18
>  
>  #define KVM_HINTS_REALTIME      0
>  
> @@ -58,6 +59,7 @@
>  #define MSR_KVM_ASYNC_PF_INT	0x4b564d06
>  #define MSR_KVM_ASYNC_PF_ACK	0x4b564d07
>  #define MSR_KVM_MIGRATION_CONTROL	0x4b564d08
> +#define MSR_KVM_PV_SCHED	0x4b564da0
>  
>  struct kvm_steal_time {
>  	__u64 steal;
> @@ -150,4 +152,26 @@ struct kvm_vcpu_pv_apf_data {
>  #define KVM_PV_EOI_ENABLED KVM_PV_EOI_MASK
>  #define KVM_PV_EOI_DISABLED 0x0
>  
> +/*
> + * VCPU boost state shared between the host and guest.
> + */
> +enum kvm_vcpu_boost_state {
> +	/* Priority boosting feature disabled in host */
> +	VCPU_BOOST_DISABLED = 0,
> +	/*
> +	 * vcpu is not explicitly boosted by the host.
> +	 * (Default priority when the guest started)
> +	 */
> +	VCPU_BOOST_NORMAL,
> +	/* vcpu is boosted by the host */
> +	VCPU_BOOST_BOOSTED
> +};
> +
> +/*
> + * Structure passed in via MSR_KVM_PV_SCHED
> + */
> +struct pv_sched_data {
> +	__u64 boost_status;
> +};
> +
>  #endif /* _UAPI_ASM_X86_KVM_PARA_H */
> diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
> index 89ca7f4c1464..dbcba73fb508 100644
> --- a/arch/x86/kvm/Kconfig
> +++ b/arch/x86/kvm/Kconfig
> @@ -141,4 +141,16 @@ config KVM_XEN
>  config KVM_EXTERNAL_WRITE_TRACKING
>  	bool
>  
> +config PARAVIRT_SCHED_KVM
> +	bool "Enable paravirt scheduling capability for kvm"
> +	depends on KVM
> +	help
> +	  Paravirtualized scheduling facilitates the exchange of scheduling
> +	  related information between the host and guest through shared memory,
> +	  enhancing the efficiency of vCPU thread scheduling by the hypervisor.
> +	  An illustrative use case involves dynamically boosting the priority of
> +	  a vCPU thread when the guest is executing a latency-sensitive workload
> +	  on that specific vCPU.
> +	  This config enables paravirt scheduling in the kvm hypervisor.
> +
>  endif # VIRTUALIZATION
> diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
> index 7bdc66abfc92..960ef6e869f2 100644
> --- a/arch/x86/kvm/cpuid.c
> +++ b/arch/x86/kvm/cpuid.c
> @@ -1113,6 +1113,8 @@ static inline int __do_cpuid_func(struct kvm_cpuid_array *array, u32 function)
>  			     (1 << KVM_FEATURE_POLL_CONTROL) |
>  			     (1 << KVM_FEATURE_PV_SCHED_YIELD) |
>  			     (1 << KVM_FEATURE_ASYNC_PF_INT);
> +		if (IS_ENABLED(CONFIG_PARAVIRT_SCHED_KVM))
> +			entry->eax |= (1 << KVM_FEATURE_PV_SCHED);
>  
>  		if (sched_info_on())
>  			entry->eax |= (1 << KVM_FEATURE_STEAL_TIME);
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 7bcf1a76a6ab..0f475b50ac83 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -3879,6 +3879,33 @@ int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
>  			return 1;
>  		break;
>  
> +#ifdef CONFIG_PARAVIRT_SCHED_KVM
> +	case MSR_KVM_PV_SCHED:
> +		if (!guest_pv_has(vcpu, KVM_FEATURE_PV_SCHED))
> +			return 1;
> +
> +		if (!(data & KVM_MSR_ENABLED))
> +			break;
> +
> +		if (!(data & ~KVM_MSR_ENABLED)) {
> +			/*
> +			 * Disable the feature
> +			 */
> +			vcpu->arch.pv_sched.msr_val = 0;
> +			kvm_set_vcpu_boosted(vcpu, false);
> +		} if (!kvm_gfn_to_hva_cache_init(vcpu->kvm,
> +				&vcpu->arch.pv_sched.data, data & ~KVM_MSR_ENABLED,
> +				sizeof(struct pv_sched_data))) {
> +			vcpu->arch.pv_sched.msr_val = data;
> +			kvm_set_vcpu_boosted(vcpu, false);
> +		} else {
> +			pr_warn("MSR_KVM_PV_SCHED: kvm:%p, vcpu:%p, "
> +				"msr value: %llx, kvm_gfn_to_hva_cache_init failed!\n",
> +				vcpu->kvm, vcpu, data & ~KVM_MSR_ENABLED);

As this is triggerable by the guest please drop this print (which is not
even ratelimited!). I think it would be better to just 'return 1;' in case
of kvm_gfn_to_hva_cache_init() failure but maybe you also need to
account for 'msr_info->host_initiated' to not fail setting this MSR from
the host upon migration.

> +		}
> +		break;
> +#endif
> +
>  	case MSR_KVM_POLL_CONTROL:
>  		if (!guest_pv_has(vcpu, KVM_FEATURE_POLL_CONTROL))
>  			return 1;
> @@ -4239,6 +4266,11 @@ int kvm_get_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
>  
>  		msr_info->data = vcpu->arch.pv_eoi.msr_val;
>  		break;
> +#ifdef CONFIG_PARAVIRT_SCHED_KVM
> +	case MSR_KVM_PV_SCHED:
> +		msr_info->data = vcpu->arch.pv_sched.msr_val;
> +		break;
> +#endif
>  	case MSR_KVM_POLL_CONTROL:
>  		if (!guest_pv_has(vcpu, KVM_FEATURE_POLL_CONTROL))
>  			return 1;
> @@ -9820,6 +9852,29 @@ static int complete_hypercall_exit(struct kvm_vcpu *vcpu)
>  	return kvm_skip_emulated_instruction(vcpu);
>  }
>  
> +#ifdef CONFIG_PARAVIRT_SCHED_KVM
> +static void record_vcpu_boost_status(struct kvm_vcpu *vcpu)
> +{
> +	u64 val = vcpu->arch.pv_sched.boost_status;
> +
> +	if (!kvm_arch_vcpu_pv_sched_enabled(&vcpu->arch))
> +		return;
> +
> +	pagefault_disable();
> +	kvm_write_guest_offset_cached(vcpu->kvm, &vcpu->arch.pv_sched.data,
> +		&val, offsetof(struct pv_sched_data, boost_status), sizeof(u64));
> +	pagefault_enable();
> +}
> +
> +void kvm_set_vcpu_boosted(struct kvm_vcpu *vcpu, bool boosted)
> +{
> +	kvm_arch_vcpu_set_boost_status(&vcpu->arch,
> +			boosted ? VCPU_BOOST_BOOSTED : VCPU_BOOST_NORMAL);
> +
> +	kvm_make_request(KVM_REQ_VCPU_BOOST_UPDATE, vcpu);
> +}
> +#endif
> +
>  int kvm_emulate_hypercall(struct kvm_vcpu *vcpu)
>  {
>  	unsigned long nr, a0, a1, a2, a3, ret;
> @@ -10593,6 +10648,12 @@ static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
>  		}
>  		if (kvm_check_request(KVM_REQ_STEAL_UPDATE, vcpu))
>  			record_steal_time(vcpu);
> +
> +#ifdef CONFIG_PARAVIRT_SCHED_KVM
> +		if (kvm_check_request(KVM_REQ_VCPU_BOOST_UPDATE, vcpu))
> +			record_vcpu_boost_status(vcpu);
> +#endif
> +
>  #ifdef CONFIG_KVM_SMM
>  		if (kvm_check_request(KVM_REQ_SMI, vcpu))
>  			process_smi(vcpu);
> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
> index 9d3ac7720da9..a74aeea55347 100644
> --- a/include/linux/kvm_host.h
> +++ b/include/linux/kvm_host.h
> @@ -167,6 +167,7 @@ static inline bool is_error_page(struct page *page)
>  #define KVM_REQ_VM_DEAD			(1 | KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
>  #define KVM_REQ_UNBLOCK			2
>  #define KVM_REQ_DIRTY_RING_SOFT_FULL	3
> +#define KVM_REQ_VCPU_BOOST_UPDATE	6
>  #define KVM_REQUEST_ARCH_BASE		8
>  
>  /*
> @@ -2287,4 +2288,8 @@ static inline void kvm_account_pgtable_pages(void *virt, int nr)
>  /* Max number of entries allowed for each kvm dirty ring */
>  #define  KVM_DIRTY_RING_MAX_ENTRIES  65536
>  
> +#ifdef CONFIG_PARAVIRT_SCHED_KVM
> +void kvm_set_vcpu_boosted(struct kvm_vcpu *vcpu, bool boosted);
> +#endif
> +
>  #endif

-- 
Vitaly