From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id B3F36C43334 for ; Mon, 27 Jun 2022 16:06:32 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S238368AbiF0QGa (ORCPT ); Mon, 27 Jun 2022 12:06:30 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:39498 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S235656AbiF0QG2 (ORCPT ); Mon, 27 Jun 2022 12:06:28 -0400 Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by lindbergh.monkeyblade.net (Postfix) with ESMTP id 454C5B483 for ; Mon, 27 Jun 2022 09:06:27 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1656345986; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: in-reply-to:in-reply-to:references:references; bh=Et0fBT88jZFdIetfnOLa7eYQfFaPq0dKwFkCGFAcX7M=; b=NrtvoVi5e71lfXGt+dTQLNCIgNJ5lxN41IaYn+fSMqsFDwyajDWmdbilou9iabauOhvypH TMiDDgMulG2F7v6fbXa6HifaIxyrQmXL0+OdstGusDzEkmFClgvTLriQCi1FiZDlvLXuCc xtA9pirFcdk0dr8ZJejfmC+gnRqGdlc= Received: from mail-wm1-f70.google.com (mail-wm1-f70.google.com [209.85.128.70]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id us-mta-6-sSrWIMV4N_m8cv_f2Z7iiw-1; Mon, 27 Jun 2022 12:06:25 -0400 X-MC-Unique: sSrWIMV4N_m8cv_f2Z7iiw-1 Received: by mail-wm1-f70.google.com with SMTP id o28-20020a05600c511c00b003a04f97f27aso260448wms.9 for ; Mon, 27 Jun 2022 09:06:24 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=x-gm-message-state:from:to:cc:subject:in-reply-to:references:date :message-id:mime-version; bh=Et0fBT88jZFdIetfnOLa7eYQfFaPq0dKwFkCGFAcX7M=; b=zPp14lLCg0EpAtLWbZdDfDtyq8x+j9N17CoMs35L7LAT3i/WX4p9Xh4dk49J/y/z6d u2+g82iTLnkBvAsoT6t0RAoC/Oz7fQ2adGS+QE8QPktrEO24Cg1qkab/lZZoCKaT0YAZ ubF0uTafGsDFSyI9TUickVPmuC79pyACLMjBwirEJZCCRk4a8NZRefxELUEb8/OoW7PU ADQF9zVu6p//kcloOQPasWEeQjP4VFH/kfi+71kmHkc9d9aRPyKYT+YupUpuWhF2MxSX 428oR+MYI6GRR/H0JqvGqyu5R2tiFrWJralYjqVOFW+/yvHvxxKEQ/ubLkU2vs1L8H+K dECg== X-Gm-Message-State: AJIora+B2WYUe/6fUgW5lIOEu/MxtFgmxsWB1AeayiEWyFcfkbozND7q 6EY32CjA90gIE7AhU7lHLdCSR/LrRZSbSL/Ei/p+n6yA7D5OqNMxxXyBLu3/IdaU2UnmE1yH+bY zFb1FM1/QRUnzcI4QKXy34OmwI7A59v6XLlJR5t72kq4W1vEzpF8JMVPMRsB7YWDk8sx44PAHxo ZK X-Received: by 2002:a05:600c:4f08:b0:39c:9437:da06 with SMTP id l8-20020a05600c4f0800b0039c9437da06mr15905007wmq.181.1656345983802; Mon, 27 Jun 2022 09:06:23 -0700 (PDT) X-Google-Smtp-Source: AGRyM1v+tzLlkmyLWgs6sbrQRofFBaYY7XIFbbta3ZJQIjl0O65OYMJ11jwxkV5OFeVW9JYuXDa+jw== X-Received: by 2002:a05:600c:4f08:b0:39c:9437:da06 with SMTP id l8-20020a05600c4f0800b0039c9437da06mr15904963wmq.181.1656345983407; Mon, 27 Jun 2022 09:06:23 -0700 (PDT) Received: from fedora (nat-2.ign.cz. [91.219.240.2]) by smtp.gmail.com with ESMTPSA id z22-20020a05600c221600b003a0499df21asm5169981wml.25.2022.06.27.09.06.22 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Mon, 27 Jun 2022 09:06:22 -0700 (PDT) From: Vitaly Kuznetsov To: Sean Christopherson Cc: kvm@vger.kernel.org, Paolo Bonzini , Anirudh Rayabharam , Wanpeng Li , Jim Mattson , Maxim Levitsky , linux-hyperv@vger.kernel.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH RFC v1 00/10] KVM: nVMX: Use vmcs_config for setting up nested VMX MSRs In-Reply-To: References: <20220622164432.194640-1-vkuznets@redhat.com> Date: Mon, 27 Jun 2022 18:06:21 +0200 Message-ID: <87tu86um4i.fsf@redhat.com> MIME-Version: 1.0 Content-Type: text/plain Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Sean Christopherson writes: > On Wed, Jun 22, 2022, Vitaly Kuznetsov wrote: >> vmcs_config is a sanitized version of host VMX MSRs where some controls are >> filtered out (e.g. when Enlightened VMCS is enabled, some know bugs are >> discovered, some inconsistencies in controls are detected,...) but >> nested_vmx_setup_ctls_msrs() uses raw host MSRs instead. This may end up >> in exposing undesired controls to L1. Switch to using vmcs_config instead. >> >> RFC part: vmcs_config's sanitization now is a mix of "what can't be enabled" >> and "what KVM doesn't want" and we need to separate these as for nested VMX >> MSRs only the first category makes sense. This gives vmcs_config a slightly >> different meaning "controls which can be (theoretically) used". An alternative >> approach would be to store sanitized host MSRs values separately, sanitize >> them and and use in nested_vmx_setup_ctls_msrs() but currently I don't see >> any benefits. Comments welcome! > > I like the idea overall, even though it's a decent amount of churn. It seems > easier to maintain than separate paths for nested. The alternative would be to > add common helpers to adjust the baseline configurations, but I don't see any > way to programmatically make that approach more robust. > > An idea to further harden things. Or: an excuse to extend macro shenanigans :-) > > If we throw all of the "opt" and "min" lists into macros, e.g. KVM_REQUIRED_* > and KVM_OPTIONAL_*, and then use those to define a KVM_KNOWN_* field, we can > prevent using the mutators to set/clear unknown bits at runtime via BUILD_BUG_ON(). > The core builders, e.g. vmx_exec_control(), can still set unknown bits, i.e. set > bits that aren't enumerated to L1, but that's easier to audit and this would catch > regressions for the issues fixed in patches. > > It'll required making add_atomic_switch_msr_special() __always_inline (or just > open code it), but that's not a big deal. > > E.g. if we have > > #define KVM_REQUIRED_CPU_BASED_VM_EXEC_CONTROL > #define KVM_OPTIONAL_CPU_BASED_VM_EXEC_CONTROL > > Then the builders for the controls shadows can do: > > diff --git a/arch/x86/kvm/vmx/vmx.h b/arch/x86/kvm/vmx/vmx.h > index 286c88e285ea..5eb75822a09e 100644 > --- a/arch/x86/kvm/vmx/vmx.h > +++ b/arch/x86/kvm/vmx/vmx.h > @@ -468,6 +468,8 @@ static inline u8 vmx_get_rvi(void) > } > > #define BUILD_CONTROLS_SHADOW(lname, uname, bits) \ > +#define KVM_KNOWN_ ## uname \ > + (KVM_REQUIRED_ ## uname | KVM_OPTIONAL_ ## uname) \ I'm certainly not a macro jedi and I failed to make this compile as gcc hates when I put '#define's in macros but I made a simpler version with (presumeably) the same outcome. v1 is out, thanks for the suggestion! > static inline void lname##_controls_set(struct vcpu_vmx *vmx, u##bits val) \ > { \ > if (vmx->loaded_vmcs->controls_shadow.lname != val) { \ > @@ -485,10 +487,12 @@ static inline u##bits lname##_controls_get(struct vcpu_vmx *vmx) \ > } \ > static inline void lname##_controls_setbit(struct vcpu_vmx *vmx, u##bits val) \ > { \ > + BUILD_BUG_ON(!(val & KVM_KNOWN_ ## uname)); \ > lname##_controls_set(vmx, lname##_controls_get(vmx) | val); \ > } \ > static inline void lname##_controls_clearbit(struct vcpu_vmx *vmx, u##bits val) \ > { \ > + BUILD_BUG_ON(!(val & KVM_KNOWN_ ## uname)); \ > lname##_controls_set(vmx, lname##_controls_get(vmx) & ~val); \ > } > BUILD_CONTROLS_SHADOW(vm_entry, VM_ENTRY_CONTROLS, 32) > > > > Handling the controls that are restricted to CONFIG_X86_64=y will be midly annoying, > but adding a base set isn't too bad, e.g. > > #define __KVM_REQUIRED_CPU_BASED_VM_EXEC_CONTROL > #ifdef CONFIG_X86_64 > #define KVM_REQUIRED_CPU_BASED_VM_EXEC_CONTROL (__KVM_REQUIRED_CPU_BASED_VM_EXEC_CONTROL | \ > CPU_BASED_CR8_LOAD_EXITING | \ > CPU_BASED_CR8_STORE_EXITING) > #else > #define KVM_REQUIRED_CPU_BASED_VM_EXEC_CONTROL __KVM_REQUIRED_CPU_BASED_VM_EXEC_CONTROL > #endif > -- Vitaly