[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ceeb0b22-9c50-0cfd-5b3e-f1fb34caa104@redhat.com>
Date: Thu, 4 May 2017 09:03:49 +0200
From: Paolo Bonzini <pbonzini@...hat.com>
To: Bandan Das <bsd@...hat.com>, kvm@...r.kernel.org
Cc: linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3/3] nVMX: Advertise PML to L1 hypervisor
On 04/05/2017 00:14, Bandan Das wrote:
> Advertise the PML bit in vmcs12 but clear it out
> before running L2 since we don't depend on hardware support
> for PML emulation.
>
> Signed-off-by: Bandan Das <bsd@...hat.com>
> ---
> arch/x86/kvm/vmx.c | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> index 5e5abb7..df71116 100644
> --- a/arch/x86/kvm/vmx.c
> +++ b/arch/x86/kvm/vmx.c
> @@ -2763,8 +2763,11 @@ static void nested_vmx_setup_ctls_msrs(struct vcpu_vmx *vmx)
> vmx->nested.nested_vmx_ept_caps |= VMX_EPT_EXTENT_GLOBAL_BIT |
> VMX_EPT_EXTENT_CONTEXT_BIT | VMX_EPT_2MB_PAGE_BIT |
> VMX_EPT_1GB_PAGE_BIT;
> - if (enable_ept_ad_bits)
> + if (enable_ept_ad_bits) {
> + vmx->nested.nested_vmx_secondary_ctls_high |=
> + SECONDARY_EXEC_ENABLE_PML;
> vmx->nested.nested_vmx_ept_caps |= VMX_EPT_AD_BIT;
> + }
> } else
> vmx->nested.nested_vmx_ept_caps = 0;
>
> @@ -10080,6 +10083,7 @@ static int prepare_vmcs02(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12,
> if (exec_control & SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES)
> vmcs_write64(APIC_ACCESS_ADDR, -1ull);
>
> + exec_control &= ~SECONDARY_EXEC_ENABLE_PML;
> vmcs_write32(SECONDARY_VM_EXEC_CONTROL, exec_control);
L0 is still using its own page modification log when running L2, so you
have to clear the bit here instead:
exec_control |= vmcs12->secondary_vm_exec_control;
and set up PML_ADDRESS and GUEST_PML_INDEX. Though, the lack of
PML_ADDRESS and GUEST_PML_INDEX initialization is a pre-existing bug.
Paolo
> }
>
>
Powered by blists - more mailing lists