Re: [PATCH 1/2] KVM: nVMX: fix EPT permissions as reported in exit qualification

From: Xiao Guangrong
Date: Fri May 12 2017 - 00:00:15 EST




On 05/11/2017 07:23 PM, Paolo Bonzini wrote:
This fixes the new ept_access_test_read_only and ept_access_test_read_write
testcases from vmx.flat.

The problem is that gpte_access moves bits around to switch from EPT
bit order (XWR) to ACC_*_MASK bit order (RWX). This results in an
incorrect exit qualification. To fix this, make pt_access and
pte_access operate on raw PTE values (only with NX flipped to mean
"can execute") and call gpte_access at the end of the walk. This
lets us use pte_access to compute the exit qualification with XWR
bit order.

Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx>
---
arch/x86/kvm/paging_tmpl.h | 35 +++++++++++++++++++++--------------
1 file changed, 21 insertions(+), 14 deletions(-)

diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h
index 56241746abbd..b0454c7e4cff 100644
--- a/arch/x86/kvm/paging_tmpl.h
+++ b/arch/x86/kvm/paging_tmpl.h
@@ -283,11 +283,13 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
pt_element_t pte;
pt_element_t __user *uninitialized_var(ptep_user);
gfn_t table_gfn;
- unsigned index, pt_access, pte_access, accessed_dirty, pte_pkey;
+ u64 pt_access, pte_access;
+ unsigned index, accessed_dirty, pte_pkey;
unsigned nested_access;
gpa_t pte_gpa;
bool have_ad;
int offset;
+ u64 walk_nx_mask = 0;
const int write_fault = access & PFERR_WRITE_MASK;
const int user_fault = access & PFERR_USER_MASK;
const int fetch_fault = access & PFERR_FETCH_MASK;
@@ -302,6 +304,7 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
have_ad = PT_HAVE_ACCESSED_DIRTY(mmu);
#if PTTYPE == 64
+ walk_nx_mask = 1ULL << PT64_NX_SHIFT;

We can always make walk_nx_mask = 1ULL << PT64_NX_SHIFT, as:
- for EPT, this bit is useless
- for 32bit, bit 63 is always ZERO, so that the final result should be ZERO too,

if (walker->level == PT32E_ROOT_LEVEL) {
pte = mmu->get_pdptr(vcpu, (addr >> 30) & 3);
trace_kvm_mmu_paging_element(pte, walker->level);
@@ -313,8 +316,6 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
walker->max_level = walker->level;
ASSERT(!(is_long_mode(vcpu) && !is_pae(vcpu)));
- accessed_dirty = have_ad ? PT_GUEST_ACCESSED_MASK : 0;
-
/*
* FIXME: on Intel processors, loads of the PDPTE registers for PAE paging
* by the MOV to CR instruction are treated as reads and do not cause the
@@ -322,14 +323,14 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
*/
nested_access = (have_ad ? PFERR_WRITE_MASK : 0) | PFERR_USER_MASK;
- pt_access = pte_access = ACC_ALL;
+ pte_access = ~0;
++walker->level;
do {
gfn_t real_gfn;
unsigned long host_addr;
- pt_access &= pte_access;
+ pt_access = pte_access;
--walker->level;
index = PT_INDEX(addr, walker->level);
@@ -371,6 +372,12 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
trace_kvm_mmu_paging_element(pte, walker->level);
+ /*
+ * Inverting the NX it lets us AND it like other
+ * permission bits.
+ */
+ pte_access = pt_access & (pte ^ walk_nx_mask);
+
if (unlikely(!FNAME(is_present_gpte)(pte)))
goto error;
@@ -379,14 +386,16 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
goto error;
}
- accessed_dirty &= pte;
- pte_access = pt_access & FNAME(gpte_access)(vcpu, pte);
-
walker->ptes[walker->level - 1] = pte;
} while (!is_last_gpte(mmu, walker->level, pte));
pte_pkey = FNAME(gpte_pkeys)(vcpu, pte);
- errcode = permission_fault(vcpu, mmu, pte_access, pte_pkey, access);
+ accessed_dirty = have_ad ? pte_access & PT_GUEST_ACCESSED_MASK : 0;
+
+ /* Convert to ACC_*_MASK flags for struct guest_walker. */
+ walker->pt_access = FNAME(gpte_access)(vcpu, pt_access ^ walk_nx_mask);
+ walker->pte_access = FNAME(gpte_access)(vcpu, pte_access ^ walk_nx_mask);
+ errcode = permission_fault(vcpu, mmu, walker->pte_access, pte_pkey, access);
if (unlikely(errcode))
goto error;
@@ -403,7 +412,7 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
walker->gfn = real_gpa >> PAGE_SHIFT;
if (!write_fault)
- FNAME(protect_clean_gpte)(mmu, &pte_access, pte);
+ FNAME(protect_clean_gpte)(mmu, &walker->pte_access, pte);
else
/*
* On a write fault, fold the dirty bit into accessed_dirty.
@@ -421,10 +430,8 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
goto retry_walk;
}
- walker->pt_access = pt_access;
- walker->pte_access = pte_access;
pgprintk("%s: pte %llx pte_access %x pt_access %x\n",
- __func__, (u64)pte, pte_access, pt_access);
+ __func__, (u64)pte, walker->pte_access, walker->pt_access);
return 1;
error:
@@ -452,7 +459,7 @@ static int FNAME(walk_addr_generic)(struct guest_walker *walker,
*/
if (!(errcode & PFERR_RSVD_MASK)) {
vcpu->arch.exit_qualification &= 0x187;
- vcpu->arch.exit_qualification |= ((pt_access & pte) & 0x7) << 3;
^ here, the original code
is buggy as pt_access and pte have different bit order, fortunately, this patch fixes it
too. :)

Otherwise it looks good to me, thanks for your fix.

Reviewed-by: Xiao Guangrong <xiaoguangrong@xxxxxxxxxxx>