]> git.baikalelectronics.ru Git - kernel.git/commit
KVM: nVMX: Ignore limit checks on VMX instructions using flat segments
authorSean Christopherson <sean.j.christopherson@intel.com>
Wed, 23 Jan 2019 22:39:25 +0000 (14:39 -0800)
committerPaolo Bonzini <pbonzini@redhat.com>
Wed, 20 Feb 2019 21:48:24 +0000 (22:48 +0100)
commitf219fd6ef4686a9347ff69fe750c50cec9189f5c
tree490b05c767b5129db5ae289510e434e7652126a2
parent9c768fdd07e91521968d07a783b7ae092e18cb0e
KVM: nVMX: Ignore limit checks on VMX instructions using flat segments

Regarding segments with a limit==0xffffffff, the SDM officially states:

    When the effective limit is FFFFFFFFH (4 GBytes), these accesses may
    or may not cause the indicated exceptions.  Behavior is
    implementation-specific and may vary from one execution to another.

In practice, all CPUs that support VMX ignore limit checks for "flat
segments", i.e. an expand-up data or code segment with base=0 and
limit=0xffffffff.  This is subtly different than wrapping the effective
address calculation based on the address size, as the flat segment
behavior also applies to accesses that would wrap the 4g boundary, e.g.
a 4-byte access starting at 0xffffffff will access linear addresses
0xffffffff, 0x0, 0x1 and 0x2.

Fixes: 07633f529f81 ("KVM: nVMX: VMX instructions: add checks for #GP/#SS exceptions")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/vmx/nested.c