Re: [PATCH v19 110/130] KVM: TDX: Handle TDX PV MMIO hypercall

[Binbin Wu]

On 4/18/2024 5:29 PM, Binbin Wu wrote:
> > +
> > +static int tdx_emulate_mmio(struct kvm_vcpu *vcpu)
> > +{
> > +    struct kvm_memory_slot *slot;
> > +    int size, write, r;
> > +    unsigned long val;
> > +    gpa_t gpa;
> > +
> > +    KVM_BUG_ON(vcpu->mmio_needed, vcpu->kvm);
> > +
> > +    size = tdvmcall_a0_read(vcpu);
> > +    write = tdvmcall_a1_read(vcpu);
> > +    gpa = tdvmcall_a2_read(vcpu);
> > +    val = write ? tdvmcall_a3_read(vcpu) : 0;
> > +
> > +    if (size != 1 && size != 2 && size != 4 && size != 8)
> > +        goto error;
> > +    if (write != 0 && write != 1)
> > +        goto error;
> > +
> > +    /* Strip the shared bit, allow MMIO with and without it set. */
> Based on the discussion 
> https://lore.kernel.org/all/ZcUO5sFEAIH68JIA@xxxxxxxxxx/
> Do we still allow the MMIO without shared bit?
>
> > +    gpa = gpa & ~gfn_to_gpa(kvm_gfn_shared_mask(vcpu->kvm));
> > +
> > +    if (size > 8u || ((gpa + size - 1) ^ gpa) & PAGE_MASK)
> "size > 8u" can be removed, since based on the check of size above, it 
> can't be greater than 8.
>
>
> > +        goto error;
> > +
> > +    slot = kvm_vcpu_gfn_to_memslot(vcpu, gpa_to_gfn(gpa));
> > +    if (slot && !(slot->flags & KVM_MEMSLOT_INVALID))
> > +        goto error;
> > +
> > +    if (!kvm_io_bus_write(vcpu, KVM_FAST_MMIO_BUS, gpa, 0, NULL)) {
> Should this be checked for write first?
>
> I check the handle_ept_misconfig() in VMX, it doesn't check write 
> first neither.
>
> Functionally, it should be OK since guest will not read the address 
> range of fast mmio.
> So the read case will be filtered out by ioeventfd_write().
> But it has take a long way to get to ioeventfd_write().
> Isn't it more efficient to check write first?

I got the reason why in handle_ept_misconfig(), it tries to do fast mmio 
write without checking.
It was intended to make fast mmio faster.
And for ept misconfig case, it's not easy to get the info of read/write.

But in this patch, we have already have read/write info, so maybe we can 
add the check for write before fast mmio?


> > +        trace_kvm_fast_mmio(gpa);
> > +        return 1;
> > +    }
> > +
> > +    if (write)
> > +        r = tdx_mmio_write(vcpu, gpa, size, val);
> > +    else
> > +        r = tdx_mmio_read(vcpu, gpa, size);
> > +    if (!r) {
> > +        /* Kernel completed device emulation. */
> > +        tdvmcall_set_return_code(vcpu, TDVMCALL_SUCCESS);
> > +        return 1;
> > +    }
> > +
> > +    /* Request the device emulation to userspace device model. */
> > +    vcpu->mmio_needed = 1;
> > +    vcpu->mmio_is_write = write;
> > +    vcpu->arch.complete_userspace_io = tdx_complete_mmio;
> > +
> > +    vcpu->run->mmio.phys_addr = gpa;
> > +    vcpu->run->mmio.len = size;
> > +    vcpu->run->mmio.is_write = write;
> > +    vcpu->run->exit_reason = KVM_EXIT_MMIO;
> > +
> > +    if (write) {
> > +        memcpy(vcpu->run->mmio.data, &val, size);
> > +    } else {
> > +        vcpu->mmio_fragments[0].gpa = gpa;
> > +        vcpu->mmio_fragments[0].len = size;
> > +        trace_kvm_mmio(KVM_TRACE_MMIO_READ_UNSATISFIED, size, gpa, 
> > NULL);
> > +    }
> > +    return 0;
> > +
> > +error:
> > +    tdvmcall_set_return_code(vcpu, TDVMCALL_INVALID_OPERAND);
> > +    return 1;
> > +}
> > +
> >   static int handle_tdvmcall(struct kvm_vcpu *vcpu)
> >   {
> >       if (tdvmcall_exit_type(vcpu))
> > @@ -1229,6 +1341,8 @@ static int handle_tdvmcall(struct kvm_vcpu *vcpu)
> >           return tdx_emulate_hlt(vcpu);
> >       case EXIT_REASON_IO_INSTRUCTION:
> >           return tdx_emulate_io(vcpu);
> > +    case EXIT_REASON_EPT_VIOLATION:
> > +        return tdx_emulate_mmio(vcpu);
> >       default:
> >           break;
> >       }
> > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> > index 03950368d8db..d5b18cad9dcd 100644
> > --- a/arch/x86/kvm/x86.c
> > +++ b/arch/x86/kvm/x86.c
> > @@ -13975,6 +13975,7 @@ EXPORT_SYMBOL_GPL(kvm_sev_es_string_io);
> >     EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_entry);
> >   EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_exit);
> > +EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_mmio);
> >   EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_fast_mmio);
> >   EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_inj_virq);
> >   EXPORT_TRACEPOINT_SYMBOL_GPL(kvm_page_fault);
> > diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> > index e27c22449d85..bc14e1f2610c 100644
> > --- a/virt/kvm/kvm_main.c
> > +++ b/virt/kvm/kvm_main.c
> > @@ -2689,6 +2689,7 @@ struct kvm_memory_slot 
> > *kvm_vcpu_gfn_to_memslot(struct kvm_vcpu *vcpu, gfn_t gfn
> >         return NULL;
> >   }
> > +EXPORT_SYMBOL_GPL(kvm_vcpu_gfn_to_memslot);
> >     bool kvm_is_visible_gfn(struct kvm *kvm, gfn_t gfn)
> >   {
> > @@ -5992,6 +5993,7 @@ int kvm_io_bus_read(struct kvm_vcpu *vcpu, enum 
> > kvm_bus bus_idx, gpa_t addr,
> >       r = __kvm_io_bus_read(vcpu, bus, &range, val);
> >       return r < 0 ? r : 0;
> >   }
> > +EXPORT_SYMBOL_GPL(kvm_io_bus_read);
> >     int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus 
> > bus_idx, gpa_t addr,
> >                   int len, struct kvm_io_device *dev)