[PATCH v2 0/9] KVM: MMU: fix and improve validation of mmio page fault

From: Xiao Guangrong <guangrong.xiao@linux.intel.com>
To: pbonzini@redhat.com
Cc: gleb@kernel.org, mtosatti@redhat.com, kvm@vger.kernel.org,
	linux-kernel@vger.kernel.org, ru.pchel@gmail.com,
	Xiao Guangrong <guangrong.xiao@linux.intel.com>
Subject: [PATCH v2 0/9] KVM: MMU: fix and improve validation of mmio page fault
Date: Wed,  5 Aug 2015 12:04:18 +0800	[thread overview]
Message-ID: <1438747467-25425-1-git-send-email-guangrong.xiao@linux.intel.com> (raw)

Changelog in v2:
- rename reset_*_rsvds_bits_mask() to reset_*_zero_bits_mask() and
  is_shadow_rsvd_bits_set() to is_shadow_zero_bits_set() to better
  match what we are checking. Thanks for Paolo's suggestion.

Current code validating mmio #PF is buggy, it was spotted by Pavel
Shirshov, the bug is that qemu complained with "KVM: unknown exit,
hardware reason 31" and KVM shown these info:
[84245.284948] EPT: Misconfiguration.
[84245.285056] EPT: GPA: 0xfeda848
[84245.285154] ept_misconfig_inspect_spte: spte 0x5eaef50107 level 4
[84245.285344] ept_misconfig_inspect_spte: spte 0x5f5fadc107 level 3
[84245.285532] ept_misconfig_inspect_spte: spte 0x5141d18107 level 2
[84245.285723] ept_misconfig_inspect_spte: spte 0x52e40dad77 level 1

This is because we got a mmio #PF and the handler see the mmio spte
becomes normal (points to the ram page)

However, this is valid after introducing fast mmio spte invalidation which
increases the generation-number instead of zapping mmio sptes, a example
is as follows:
1. QEMU drops mmio region by adding a new memslot
2. invalidate all mmio sptes
3.

        VCPU 0                        VCPU 1
    access the invalid mmio spte

                            access the region originally was MMIO before
                            set the spte to the normal ram map

    mmio #PF
    check the spte and see it becomes normal ram mapping !!!

The first patch simply fixes the bug by dropping the validation in mmio
handler which is good for backport

Later patches enable fully check reserved bits for shadow page table
entries, since shadow page table and guest page table have the some
format, this patches reuse the logic which checks reserved bits on
guest pte to check sptes

In order to catching as many bugs as possible, we not only check the
reserved bits on hardware but also check other bits that spte never used

Xiao Guangrong (9):
  KVM: MMU: fix validation of mmio page fault
  KVM: MMU: move FNAME(is_rsvd_bits_set) to mmu.c
  KVM: MMU: introduce rsvd_bits_validate
  KVM: MMU: split reset_rsvds_bits_mask
  KVM: MMU: split reset_rsvds_bits_mask_ept
  KVM: MMU: introduce the framework to check zero bits on sptes
  KVM: MMU: introduce is_shadow_zero_bits_set()
  KVM: MMU: fully check zero bits for sptes
  KVM: VMX: drop ept misconfig check

 arch/x86/include/asm/kvm_host.h |  16 ++-
 arch/x86/kvm/mmu.c              | 283 ++++++++++++++++++++++++----------------
 arch/x86/kvm/mmu.h              |   4 +-
 arch/x86/kvm/paging_tmpl.h      |  13 +-
 arch/x86/kvm/svm.c              |   1 +
 arch/x86/kvm/vmx.c              |  74 +----------
 arch/x86/kvm/x86.c              |   3 +-
 7 files changed, 193 insertions(+), 201 deletions(-)

-- 
2.1.0