kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Sean Christopherson <seanjc@google.com>
To: Reiji Watanabe <reijiw@google.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
	Vitaly Kuznetsov <vkuznets@redhat.com>,
	Wanpeng Li <wanpengli@tencent.com>,
	Jim Mattson <jmattson@google.com>, Joerg Roedel <joro@8bytes.org>,
	kvm@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH 06/43] KVM: x86: Properly reset MMU context at vCPU RESET/INIT
Date: Tue, 18 May 2021 20:23:42 +0000	[thread overview]
Message-ID: <YKQiTlDG1sZ4Zd2E@google.com> (raw)
In-Reply-To: <CAAeT=FzS0bP_7_wz6G6cL8-7pudTD7fhavLCVsOE0KnPXf99dQ@mail.gmail.com>

On Mon, May 17, 2021, Reiji Watanabe wrote:
> >  void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event)
> >  {
> > +       unsigned long old_cr0 = kvm_read_cr0(vcpu);
> > +       unsigned long old_cr4 = kvm_read_cr4(vcpu);
> > +
> >         kvm_lapic_reset(vcpu, init_event);
> >
> >         vcpu->arch.hflags = 0;
> > @@ -10483,6 +10485,10 @@ void kvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event)
> >         vcpu->arch.ia32_xss = 0;
> >
> >         static_call(kvm_x86_vcpu_reset)(vcpu, init_event);
> > +
> > +       if (kvm_cr0_mmu_role_changed(old_cr0, kvm_read_cr0(vcpu)) ||
> > +           kvm_cr4_mmu_role_changed(old_cr4, kvm_read_cr4(vcpu)))
> > +               kvm_mmu_reset_context(vcpu);
> >  }
> 
> I'm wondering if kvm_vcpu_reset() should call kvm_mmu_reset_context()
> for a change in EFER.NX as well.

Oooh.  So there _should_ be no need.   Paging has to be enabled for EFER.NX to
be relevant, and INIT toggles CR0.PG 1=>0 if paging was enabled and so is
guaranteed to trigger a context reset.  And we do want to skip the context reset,
e.g. INIT-SIPI-SIPI when the vCPU has paging disabled should continue using the
same MMU.

But, kvm_calc_mmu_role_common() neglects to ignore NX if CR0.PG=0, and so the
MMU role will be stale if INIT clears EFER.NX without forcing a context reset.
However, that's benign from a functionality perspective because the context
itself correctly incorporates CR0.PG, it's only the role that's borked.  I.e.
KVM will fail to reuse a page/context due to the spurious role.nxe, but the
permission checks are always be correct.

I'll add a comment here and send a patch to fix the role calculation.

  reply	other threads:[~2021-05-18 20:23 UTC|newest]

Thread overview: 88+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-04-24  0:46 [PATCH 00/43] KVM: x86: vCPU RESET/INIT fixes and consolidation Sean Christopherson
2021-04-24  0:46 ` [PATCH 01/43] KVM: nVMX: Set LDTR to its architecturally defined value on nested VM-Exit Sean Christopherson
2021-05-19  5:30   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 02/43] KVM: VMX: Set EDX at INIT with CPUID.0x1, Family-Model-Stepping Sean Christopherson
2021-05-19  5:59   ` Reiji Watanabe
2021-05-19 18:47     ` Sean Christopherson
2021-05-21  7:07       ` Reiji Watanabe
2021-05-21 15:28         ` Sean Christopherson
2021-05-24  4:29           ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 03/43] KVM: SVM: Require exact CPUID.0x1 match when stuffing EDX at INIT Sean Christopherson
2021-05-19  5:30   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 04/43] KVM: SVM: Fall back to KVM's hardcoded value for EDX at RESET/INIT Sean Christopherson
2021-05-19  5:41   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 05/43] KVM: x86: Split out CR0/CR4 MMU role change detectors to separate helpers Sean Christopherson
2021-05-19  5:31   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 06/43] KVM: x86: Properly reset MMU context at vCPU RESET/INIT Sean Christopherson
2021-05-17 16:57   ` Reiji Watanabe
2021-05-18 20:23     ` Sean Christopherson [this message]
2021-05-18 22:42       ` Reiji Watanabe
2021-05-19 17:16         ` Sean Christopherson
2021-05-24  4:57           ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 07/43] KVM: VMX: Remove explicit MMU reset in enter_rmode() Sean Christopherson
2021-04-24  0:46 ` [PATCH 08/43] KVM: SVM: Drop explicit MMU reset at RESET/INIT Sean Christopherson
2021-05-19  5:32   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 09/43] KVM: SVM: Drop a redundant init_vmcb() from svm_create_vcpu() Sean Christopherson
2021-05-19  5:32   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 10/43] KVM: VMX: Move init_vmcs() invocation to vmx_vcpu_reset() Sean Christopherson
2021-05-19  5:33   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 11/43] KVM: x86: WARN if the APIC map is dirty without an in-kernel local APIC Sean Christopherson
2021-04-24  0:46 ` [PATCH 12/43] KVM: x86: Remove defunct BSP "update" in local APIC reset Sean Christopherson
2021-05-26  6:54   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 13/43] KVM: x86: Migrate the PIT only if vcpu0 is migrated, not any BSP Sean Christopherson
2021-04-24  0:46 ` [PATCH 14/43] KVM: x86: Don't force set BSP bit when local APIC is managed by userspace Sean Christopherson
2021-05-26  6:55   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 15/43] KVM: x86: Set BSP bit in reset BSP vCPU's APIC base by default Sean Christopherson
2021-05-26  6:55   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 16/43] KVM: VMX: Stuff vcpu->arch.apic_base directly at vCPU RESET Sean Christopherson
2021-05-26  6:55   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 17/43] KVM: x86: Open code necessary bits of kvm_lapic_set_base() " Sean Christopherson
2021-05-26  7:04   ` Reiji Watanabe
2021-05-26 15:15     ` Sean Christopherson
2021-04-24  0:46 ` [PATCH 18/43] KVM: x86: Consolidate APIC base RESET initialization code Sean Christopherson
2021-05-26  7:04   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 19/43] KVM: x86: Move EDX initialization at vCPU RESET to common code Sean Christopherson
2021-05-19  5:45   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 20/43] KVM: SVM: Don't bother writing vmcb->save.rip at vCPU RESET/INIT Sean Christopherson
2021-05-19  5:34   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 21/43] KVM: VMX: Invert handling of CR0.WP for EPT without unrestricted guest Sean Christopherson
2021-04-24  0:46 ` [PATCH 22/43] KVM: VMX: Remove direct write to vcpu->arch.cr0 during vCPU RESET/INIT Sean Christopherson
2021-05-19  5:34   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 23/43] KVM: VMX: Fold ept_update_paging_mode_cr0() back into vmx_set_cr0() Sean Christopherson
2021-04-24  0:46 ` [PATCH 24/43] KVM: nVMX: Do not clear CR3 load/store exiting bits if L1 wants 'em Sean Christopherson
2021-04-24  0:46 ` [PATCH 25/43] KVM: VMX: Pull GUEST_CR3 from the VMCS iff CR3 load exiting is disabled Sean Christopherson
2021-04-24  0:46 ` [PATCH 26/43] KVM: VMX: Process CR0.PG side effects after setting CR0 assets Sean Christopherson
2021-04-24  0:46 ` [PATCH 27/43] KVM: VMX: Skip emulation required checks during pmode/rmode transitions Sean Christopherson
2021-04-24  0:46 ` [PATCH 28/43] KVM: nVMX: Don't evaluate "emulation required" on VM-Exit Sean Christopherson
2021-04-24  0:46 ` [PATCH 29/43] KVM: SVM: Tweak order of cr0/cr4/efer writes at RESET/INIT Sean Christopherson
2021-05-19 18:16   ` Reiji Watanabe
2021-05-19 19:58     ` Sean Christopherson
2021-05-23 23:04       ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 30/43] KVM: SVM: Drop redundant writes to vmcb->save.cr4 " Sean Christopherson
2021-05-19  5:35   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 31/43] KVM: SVM: Stuff save->dr6 at during VMSA sync, not " Sean Christopherson
2021-04-24  0:46 ` [PATCH 32/43] KVM: VMX: Skip pointless MSR bitmap update when setting EFER Sean Christopherson
2021-04-24  0:46 ` [PATCH 33/43] KVM: VMX: Refresh list of user return MSRs after setting guest CPUID Sean Christopherson
2021-05-19  5:35   ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 34/43] KVM: VMX: Don't _explicitly_ reconfigure user return MSRs on vCPU INIT Sean Christopherson
2021-04-24  0:46 ` [PATCH 35/43] KVM: x86: Move setting of sregs during vCPU RESET/INIT to common x86 Sean Christopherson
2021-05-17 23:50   ` Reiji Watanabe
2021-05-18 21:45     ` Sean Christopherson
2021-05-21  5:19       ` Reiji Watanabe
2021-04-24  0:46 ` [PATCH 36/43] KVM: VMX: Remove obsolete MSR bitmap refresh at vCPU RESET/INIT Sean Christopherson
2021-04-24  0:46 ` [PATCH 37/43] KVM: nVMX: Remove obsolete MSR bitmap refresh at nested transitions Sean Christopherson
2021-04-24  0:46 ` [PATCH 38/43] KVM: VMX: Don't redo x2APIC MSR bitmaps when userspace filter is changed Sean Christopherson
2021-04-24  0:46 ` [PATCH 39/43] KVM: VMX: Remove unnecessary initialization of msr_bitmap_mode Sean Christopherson
2021-04-24  0:46 ` [PATCH 40/43] KVM: VMX: Smush x2APIC MSR bitmap adjustments into single function Sean Christopherson
2021-04-24  0:46 ` [PATCH 41/43] KVM: VMX: Remove redundant write to set vCPU as active at RESET/INIT Sean Christopherson
2021-04-24  0:46 ` [PATCH 42/43] KVM: VMX: Drop VMWRITEs to zero fields at vCPU RESET Sean Christopherson
2021-05-24 21:15   ` Paolo Bonzini
2021-05-24 22:21     ` Jim Mattson
2021-05-24 22:28     ` Sean Christopherson
2021-05-24 22:48       ` Jim Mattson
2021-05-25  1:02         ` Sean Christopherson
2021-04-24  0:46 ` [PATCH 43/43] KVM: x86: Drop pointless @reset_roots from kvm_init_mmu() Sean Christopherson
2021-05-27 19:11   ` Sean Christopherson
2021-05-27 19:25     ` Sean Christopherson
2021-06-10 16:54 ` [PATCH 00/43] KVM: x86: vCPU RESET/INIT fixes and consolidation Paolo Bonzini
2021-06-10 19:22   ` Sean Christopherson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=YKQiTlDG1sZ4Zd2E@google.com \
    --to=seanjc@google.com \
    --cc=jmattson@google.com \
    --cc=joro@8bytes.org \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=pbonzini@redhat.com \
    --cc=reijiw@google.com \
    --cc=vkuznets@redhat.com \
    --cc=wanpengli@tencent.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).