On 22/02/2018 01:07, Jim Mattson wrote:
On Wed, Feb 21, 2018 at 1:41 PM, Paolo Bonzini pbonzini@redhat.com wrote:
We need to change the default all-1s bitmap if the MSRs are _not_ intercepted. However, the code was disabling the intercept when it was _enabled_ in the VMCS01. This is not causing bigger trouble, because vmx_vcpu_run checks the VMCS02's MSR bitmap and would do the right thing even if fed garbage... but it's obviously a bug and it can cause extra MSR reads and writes when running nested guests.
Fixes: d28b387fb74da95d69d2615732f50cceb38e9a4d Fixes: 15d45071523d89b3fb7372e2135fbd72f6af9506 Cc: x86@kernel.org Cc: Radim Krčmář rkrcmar@redhat.com Cc: KarimAllah Ahmed karahmed@amazon.de Cc: David Woodhouse dwmw@amazon.co.uk Cc: Jim Mattson jmattson@google.com Cc: Thomas Gleixner tglx@linutronix.de Cc: Ingo Molnar mingo@kernel.org Cc: stable@vger.kernel.org Signed-off-by: Paolo Bonzini pbonzini@redhat.com
Wasn't this already fixed by 206587a9fb76 ("X86/nVMX: Properly set spec_ctrl and pred_cmd before merging MSRs")?
Ouch, yes, and my patch would have no conflicts at all so it would reintroduce the bug! Will resend v2 without it.
Paolo