On Mon, 2022-06-06 at 10:36 +0200, Vitaly Kuznetsov wrote:
> Handle L2 TLB flush requests by going through all vCPUs and checking
> whether there are vCPUs running the same VM_ID with a VP_ID specified
> in the requests. Perform synthetic exit to L2 upon finish.
>
> Note, while checking VM_ID/VP_ID of running vCPUs seem to be a bit
> racy, we count on the fact that KVM flushes the whole L2 VPID upon
> transition. Also, KVM_REQ_HV_TLB_FLUSH request needs to be done upon
> transition between L1 and L2 to make sure all pending requests are
> always processed.
>
> For the reference, Hyper-V TLFS refers to the feature as "Direct
> Virtual Flush".
>
> Note, nVMX/nSVM code does not handle VMCALL/VMMCALL from L2 yet.
>
> Signed-off-by: Vitaly Kuznetsov <vkuznets@xxxxxxxxxx>
> ---
> arch/x86/kvm/hyperv.c | 84 +++++++++++++++++++++++++++++++++++--------
> arch/x86/kvm/hyperv.h | 14 ++++----
> arch/x86/kvm/trace.h | 21 ++++++-----
> arch/x86/kvm/x86.c | 4 +--
> 4 files changed, 91 insertions(+), 32 deletions(-)
>
> diff --git a/arch/x86/kvm/hyperv.c b/arch/x86/kvm/hyperv.c
> index 3075e9661696..740190917c1c 100644
> --- a/arch/x86/kvm/hyperv.c
> +++ b/arch/x86/kvm/hyperv.c
> @@ -34,6 +34,7 @@
> #include <linux/eventfd.h>
>
> #include <asm/apicdef.h>
> +#include <asm/mshyperv.h>
> #include <trace/events/kvm.h>
>
> #include "trace.h"
> @@ -1835,9 +1836,10 @@ static int kvm_hv_get_tlb_flush_entries(struct kvm *kvm, struct kvm_hv_hcall *hc
> entries, consumed_xmm_halves, offset);
> }
>
> -static void hv_tlb_flush_enqueue(struct kvm_vcpu *vcpu, u64 *entries, int count)
> +static void hv_tlb_flush_enqueue(struct kvm_vcpu *vcpu,
> + struct kvm_vcpu_hv_tlb_flush_fifo *tlb_flush_fifo,
> + u64 *entries, int count)
> {
> - struct kvm_vcpu_hv_tlb_flush_fifo *tlb_flush_fifo;
> struct kvm_vcpu_hv *hv_vcpu = to_hv_vcpu(vcpu);
> u64 entry = KVM_HV_TLB_FLUSHALL_ENTRY;
> unsigned long flags;
> @@ -1845,9 +1847,6 @@ static void hv_tlb_flush_enqueue(struct kvm_vcpu *vcpu, u64 *entries, int count)
> if (!hv_vcpu)
> return;
>
> - /* kvm_hv_flush_tlb() is not ready to handle requests for L2s yet */
> - tlb_flush_fifo = &hv_vcpu->tlb_flush_fifo[HV_L1_TLB_FLUSH_FIFO];
> -
> spin_lock_irqsave(&tlb_flush_fifo->write_lock, flags);
>
> /*
> @@ -1883,7 +1882,7 @@ void kvm_hv_vcpu_flush_tlb(struct kvm_vcpu *vcpu)
> return;
> }
>
> - tlb_flush_fifo = kvm_hv_get_tlb_flush_fifo(vcpu);
> + tlb_flush_fifo = kvm_hv_get_tlb_flush_fifo(vcpu, is_guest_mode(vcpu));
>
> count = kfifo_out(&tlb_flush_fifo->entries, entries, KVM_HV_TLB_FLUSH_FIFO_SIZE);
>
> @@ -1916,6 +1915,7 @@ static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
> struct hv_tlb_flush_ex flush_ex;
> struct hv_tlb_flush flush;
> DECLARE_BITMAP(vcpu_mask, KVM_MAX_VCPUS);
> + struct kvm_vcpu_hv_tlb_flush_fifo *tlb_flush_fifo;
> /*
> * Normally, there can be no more than 'KVM_HV_TLB_FLUSH_FIFO_SIZE'
> * entries on the TLB flush fifo. The last entry, however, needs to be
> @@ -1959,7 +1959,8 @@ static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
> }
>
> trace_kvm_hv_flush_tlb(flush.processor_mask,
> - flush.address_space, flush.flags);
> + flush.address_space, flush.flags,
> + is_guest_mode(vcpu));
>
> valid_bank_mask = BIT_ULL(0);
> sparse_banks[0] = flush.processor_mask;
> @@ -1990,7 +1991,7 @@ static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
> trace_kvm_hv_flush_tlb_ex(flush_ex.hv_vp_set.valid_bank_mask,
> flush_ex.hv_vp_set.format,
> flush_ex.address_space,
> - flush_ex.flags);
> + flush_ex.flags, is_guest_mode(vcpu));
>
> valid_bank_mask = flush_ex.hv_vp_set.valid_bank_mask;
> all_cpus = flush_ex.hv_vp_set.format !=
> @@ -2028,19 +2029,57 @@ static u64 kvm_hv_flush_tlb(struct kvm_vcpu *vcpu, struct kvm_hv_hcall *hc)
> * vcpu->arch.cr3 may not be up-to-date for running vCPUs so we can't
> * analyze it here, flush TLB regardless of the specified address space.
> */
> - if (all_cpus) {
> - kvm_for_each_vcpu(i, v, kvm)
> - hv_tlb_flush_enqueue(v, tlb_flush_entries, hc->rep_cnt);
> + if (all_cpus && !is_guest_mode(vcpu)) {
> + kvm_for_each_vcpu(i, v, kvm) {
> + tlb_flush_fifo = kvm_hv_get_tlb_flush_fifo(v, false);
> + hv_tlb_flush_enqueue(v, tlb_flush_fifo,
> + tlb_flush_entries, hc->rep_cnt);
> + }
>
> kvm_make_all_cpus_request(kvm, KVM_REQ_HV_TLB_FLUSH);
> - } else {
> + } else if (!is_guest_mode(vcpu)) {
> sparse_set_to_vcpu_mask(kvm, sparse_banks, valid_bank_mask, vcpu_mask);
>
> for_each_set_bit(i, vcpu_mask, KVM_MAX_VCPUS) {
> v = kvm_get_vcpu(kvm, i);
> if (!v)
> continue;
> - hv_tlb_flush_enqueue(v, tlb_flush_entries, hc->rep_cnt);
> + tlb_flush_fifo = kvm_hv_get_tlb_flush_fifo(v, false);
> + hv_tlb_flush_enqueue(v, tlb_flush_fifo,
> + tlb_flush_entries, hc->rep_cnt);
> + }
> +
> + kvm_make_vcpus_request_mask(kvm, KVM_REQ_HV_TLB_FLUSH, vcpu_mask);
> + } else {
> + struct kvm_vcpu_hv *hv_v;
> +
> + bitmap_zero(vcpu_mask, KVM_MAX_VCPUS);
> +
> + kvm_for_each_vcpu(i, v, kvm) {
> + hv_v = to_hv_vcpu(v);
> +
> + /*
> + * The following check races with nested vCPUs entering/exiting
> + * and/or migrating between L1's vCPUs, however the only case when
> + * KVM *must* flush the TLB is when the target L2 vCPU keeps
> + * running on the same L1 vCPU from the moment of the request until
> + * kvm_hv_flush_tlb() returns. TLB is fully flushed in all other
> + * cases, e.g. when the target L2 vCPU migrates to a different L1
> + * vCPU or when the corresponding L1 vCPU temporary switches to a
> + * different L2 vCPU while the request is being processed.
Looks great!
> + */
> + if (!hv_v || hv_v->nested.vm_id != hv_vcpu->nested.vm_id)
> + continue;
> +
> + if (!all_cpus &&
> + !hv_is_vp_in_sparse_set(hv_v->nested.vp_id, valid_bank_mask,
> + sparse_banks))
> + continue;
> +
> + __set_bit(i, vcpu_mask);
> + tlb_flush_fifo = kvm_hv_get_tlb_flush_fifo(v, true);
> + hv_tlb_flush_enqueue(v, tlb_flush_fifo,
> + tlb_flush_entries, hc->rep_cnt);
> }
>
> kvm_make_vcpus_request_mask(kvm, KVM_REQ_HV_TLB_FLUSH, vcpu_mask);
> @@ -2228,10 +2267,27 @@ static void kvm_hv_hypercall_set_result(struct kvm_vcpu *vcpu, u64 result)
>
> static int kvm_hv_hypercall_complete(struct kvm_vcpu *vcpu, u64 result)
> {
> + int ret;
> +
> trace_kvm_hv_hypercall_done(result);
> kvm_hv_hypercall_set_result(vcpu, result);
> ++vcpu->stat.hypercalls;
> - return kvm_skip_emulated_instruction(vcpu);
> + ret = kvm_skip_emulated_instruction(vcpu);
> +
> + if (unlikely(hv_result_success(result) && is_guest_mode(vcpu)
> + && kvm_hv_is_tlb_flush_hcall(vcpu))) {
> + struct kvm_vcpu_hv *hv_vcpu = to_hv_vcpu(vcpu);
> + u32 tlb_lock_count;
> +
> + if (unlikely(kvm_read_guest(vcpu->kvm, hv_vcpu->nested.pa_page_gpa,
> + &tlb_lock_count, sizeof(tlb_lock_count))))
> + kvm_inject_gp(vcpu, 0);
> +
> + if (tlb_lock_count)
> + kvm_x86_ops.nested_ops->hv_inject_synthetic_vmexit_post_tlb_flush(vcpu);
> + }
> +
> + return ret;
> }
>
> static int kvm_hv_hypercall_complete_userspace(struct kvm_vcpu *vcpu)
> diff --git a/arch/x86/kvm/hyperv.h b/arch/x86/kvm/hyperv.h
> index dc46c5ed5d18..7778b3a5913c 100644
> --- a/arch/x86/kvm/hyperv.h
> +++ b/arch/x86/kvm/hyperv.h
> @@ -148,26 +148,24 @@ int kvm_vm_ioctl_hv_eventfd(struct kvm *kvm, struct kvm_hyperv_eventfd *args);
> int kvm_get_hv_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid2 *cpuid,
> struct kvm_cpuid_entry2 __user *entries);
>
> -static inline struct kvm_vcpu_hv_tlb_flush_fifo *kvm_hv_get_tlb_flush_fifo(struct kvm_vcpu *vcpu)
> +static inline struct kvm_vcpu_hv_tlb_flush_fifo *kvm_hv_get_tlb_flush_fifo(struct kvm_vcpu *vcpu,
> + bool is_guest_mode)
> {
> struct kvm_vcpu_hv *hv_vcpu = to_hv_vcpu(vcpu);
> - int i = !is_guest_mode(vcpu) ? HV_L1_TLB_FLUSH_FIFO :
> - HV_L2_TLB_FLUSH_FIFO;
> -
> - /* KVM does not handle L2 TLB flush requests yet */
> - WARN_ON_ONCE(i != HV_L1_TLB_FLUSH_FIFO);
> + int i = is_guest_mode ? HV_L2_TLB_FLUSH_FIFO :
> + HV_L1_TLB_FLUSH_FIFO;
>
> return &hv_vcpu->tlb_flush_fifo[i];
> }
>
> -static inline void kvm_hv_vcpu_empty_flush_tlb(struct kvm_vcpu *vcpu)
> +static inline void kvm_hv_vcpu_empty_flush_tlb(struct kvm_vcpu *vcpu, bool is_guest_mode)
> {
> struct kvm_vcpu_hv_tlb_flush_fifo *tlb_flush_fifo;
>
> if (!to_hv_vcpu(vcpu) || !kvm_check_request(KVM_REQ_HV_TLB_FLUSH, vcpu))
> return;
>
> - tlb_flush_fifo = kvm_hv_get_tlb_flush_fifo(vcpu);
> + tlb_flush_fifo = kvm_hv_get_tlb_flush_fifo(vcpu, is_guest_mode);
>
> kfifo_reset_out(&tlb_flush_fifo->entries);
> }
> diff --git a/arch/x86/kvm/trace.h b/arch/x86/kvm/trace.h
> index fd28dd40b813..f5e5b8f0342c 100644
> --- a/arch/x86/kvm/trace.h
> +++ b/arch/x86/kvm/trace.h
> @@ -1510,38 +1510,41 @@ TRACE_EVENT(kvm_hv_timer_state,
> * Tracepoint for kvm_hv_flush_tlb.
> */
> TRACE_EVENT(kvm_hv_flush_tlb,
> - TP_PROTO(u64 processor_mask, u64 address_space, u64 flags),
> - TP_ARGS(processor_mask, address_space, flags),
> + TP_PROTO(u64 processor_mask, u64 address_space, u64 flags, bool guest_mode),
> + TP_ARGS(processor_mask, address_space, flags, guest_mode),
>
> TP_STRUCT__entry(
> __field(u64, processor_mask)
> __field(u64, address_space)
> __field(u64, flags)
> + __field(bool, guest_mode)
> ),
>
> TP_fast_assign(
> __entry->processor_mask = processor_mask;
> __entry->address_space = address_space;
> __entry->flags = flags;
> + __entry->guest_mode = guest_mode;
> ),
>
> - TP_printk("processor_mask 0x%llx address_space 0x%llx flags 0x%llx",
> + TP_printk("processor_mask 0x%llx address_space 0x%llx flags 0x%llx %s",
> __entry->processor_mask, __entry->address_space,
> - __entry->flags)
> + __entry->flags, __entry->guest_mode ? "(L2)" : "")
> );
>
> /*
> * Tracepoint for kvm_hv_flush_tlb_ex.
> */
> TRACE_EVENT(kvm_hv_flush_tlb_ex,
> - TP_PROTO(u64 valid_bank_mask, u64 format, u64 address_space, u64 flags),
> - TP_ARGS(valid_bank_mask, format, address_space, flags),
> + TP_PROTO(u64 valid_bank_mask, u64 format, u64 address_space, u64 flags, bool guest_mode),
> + TP_ARGS(valid_bank_mask, format, address_space, flags, guest_mode),
>
> TP_STRUCT__entry(
> __field(u64, valid_bank_mask)
> __field(u64, format)
> __field(u64, address_space)
> __field(u64, flags)
> + __field(bool, guest_mode)
> ),
>
> TP_fast_assign(
> @@ -1549,12 +1552,14 @@ TRACE_EVENT(kvm_hv_flush_tlb_ex,
> __entry->format = format;
> __entry->address_space = address_space;
> __entry->flags = flags;
> + __entry->guest_mode = guest_mode;
> ),
>
> TP_printk("valid_bank_mask 0x%llx format 0x%llx "
> - "address_space 0x%llx flags 0x%llx",
> + "address_space 0x%llx flags 0x%llx %s",
> __entry->valid_bank_mask, __entry->format,
> - __entry->address_space, __entry->flags)
> + __entry->address_space, __entry->flags,
> + __entry->guest_mode ? "(L2)" : "")
> );
>
> /*
> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> index 805db43c2829..8e945500ef50 100644
> --- a/arch/x86/kvm/x86.c
> +++ b/arch/x86/kvm/x86.c
> @@ -3355,12 +3355,12 @@ void kvm_service_local_tlb_flush_requests(struct kvm_vcpu *vcpu)
> {
> if (kvm_check_request(KVM_REQ_TLB_FLUSH_CURRENT, vcpu)) {
> kvm_vcpu_flush_tlb_current(vcpu);
> - kvm_hv_vcpu_empty_flush_tlb(vcpu);
> + kvm_hv_vcpu_empty_flush_tlb(vcpu, is_guest_mode(vcpu));
> }
>
> if (kvm_check_request(KVM_REQ_TLB_FLUSH_GUEST, vcpu)) {
> kvm_vcpu_flush_tlb_guest(vcpu);
> - kvm_hv_vcpu_empty_flush_tlb(vcpu);
> + kvm_hv_vcpu_empty_flush_tlb(vcpu, is_guest_mode(vcpu));
> } else if (kvm_check_request(KVM_REQ_HV_TLB_FLUSH, vcpu)) {
> kvm_hv_vcpu_flush_tlb(vcpu);
> }
Looks good,
Reviewed-by: Maxim Levitsky <mlevitsk@xxxxxxxxxx>
Best regadrds,
Maxim Levitsky
