2018-06-27 09:11-0700, Jim Mattson: > On Wed, Jun 27, 2018 at 9:01 AM, Radim Krčmář <rkrcmar@xxxxxxxxxx> wrote: > > >> For consistency, can we also change the various cpu_has_vmx_* > >> functions that test for physical cpu capabilities to > >> cpu_supports_vmx_*? > > > > Seems reasonable, but I thought there weren't any -- can you name an > > example, thanks. > > cpu_has_vmx_msr_bitmap > cpu_has_vmx_tpr_shadow > cpu_has_vmx_virtualize_apic_accesses > cpu_has_vmx_virtualize_x2apic_mode > cpu_has_vmx_apic_register_virt > cpu_has_vmx_virtual_intr_delivery > cpu_has_vmx_preemption_timer > cpu_has_vmx_posted_intr > cpu_has_vmx_apicv > cpu_has_vmx_flexpriority > cpu_has_vmx_ept_execute_only > cpu_has_vmx_ept_2m_page > cpu_has_vmx_ept_1g_page > cpu_has_vmx_ept_4levels > cpu_has_vmx_ept_mt_wb > cpu_has_vmx_ept_5levels > cpu_has_vmx_ept_ad_bits > cpu_has_vmx_invept_context > cpu_has_vmx_invept_global > cpu_has_vmx_invvpid_individual_addr > cpu_has_vmx_invvpid_single > cpu_has_vmx_invvpid_global > cpu_has_vmx_invvpid > cpu_has_vmx_ept > cpu_has_vmx_unrestricted_guest > cpu_has_vmx_ple > cpu_has_vmx_basic_inout > cpu_has_vmx_vpid > cpu_has_vmx_rdtscp > cpu_has_vmx_invpcid > cpu_has_vmx_wbinvd_exit > cpu_has_vmx_shadow_vmcs > cpu_has_vmx_pml > cpu_has_vmx_tsc_scaling > cpu_has_vmx_vmfunc They all seem to be using vmcs_config, which is a snapshot of active VMCS VM controls. This meaning corresponds to what nested_cpu_has* does. This patch was checking the high bits of VMX MSR and those have a different meaning. I think we're consistent as far as the naming goes.
