KVM: VMX: Fix commit which broke PML
authorKai Huang <kai.huang@linux.intel.com>
Wed, 4 Nov 2015 05:46:05 +0000 (13:46 +0800)
committerPaolo Bonzini <pbonzini@redhat.com>
Thu, 5 Nov 2015 10:34:11 +0000 (11:34 +0100)
commita3eaa8649e4c6a6afdafaa04b9114fb230617bb1
tree878f0e28bd9734f3c7c822771d9f501f7880b707
parent879ae1880449c88db11c1ebdaedc2da79b2fe73f
KVM: VMX: Fix commit which broke PML

I found PML was broken since below commit:

commit feda805fe7c4ed9cf78158e73b1218752e3b4314
Author: Xiao Guangrong <guangrong.xiao@linux.intel.com>
Date:   Wed Sep 9 14:05:55 2015 +0800

KVM: VMX: unify SECONDARY_VM_EXEC_CONTROL update

Unify the update in vmx_cpuid_update()

Signed-off-by: Xiao Guangrong <guangrong.xiao@linux.intel.com>
[Rewrite to use vmcs_set_secondary_exec_control. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The reason is in above commit vmx_cpuid_update calls vmx_secondary_exec_control,
in which currently SECONDARY_EXEC_ENABLE_PML bit is cleared unconditionally (as
PML is enabled in creating vcpu). Therefore if vcpu_cpuid_update is called after
vcpu is created, PML will be disabled unexpectedly while log-dirty code still
thinks PML is used.

Fix this by clearing SECONDARY_EXEC_ENABLE_PML in vmx_secondary_exec_control
only when PML is not supported or not enabled (!enable_pml). This is more
reasonable as PML is currently either always enabled or disabled. With this
explicit updating SECONDARY_EXEC_ENABLE_PML in vmx_enable{disable}_pml is not
needed so also rename vmx_enable{disable}_pml to vmx_create{destroy}_pml_buffer.

Fixes: feda805fe7c4ed9cf78158e73b1218752e3b4314
Signed-off-by: Kai Huang <kai.huang@linux.intel.com>
[While at it, change a wrong ASSERT to an "if".  The condition can happen
 if creating the VCPU fails with ENOMEM. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/vmx.c