On 06.09.2017 15:17, Borislav Petkov wrote:
> On Wed, Sep 06, 2017 at 02:54:05PM +0200, David Hildenbrand wrote:
>> Maybe go one step further and incorporate everything (+vls) into a
>> single if statement?
>
> Or maybe simplify it even more by not even looking at vls. If the user
> disables it, fine, if she enables it but the hw doesn't support it, it
> will be set to false automatically.
>
> Or am I missing a case?
>
> ---
> From: Borislav Petkov <bp@xxxxxxx>
> Date: Tue, 5 Sep 2017 18:59:55 +0200
> Subject: [PATCH] KVM: SVM: Do not issue virtual VMLOAD/VMSAVE
> supported-message
> MIME-Version: 1.0
> Content-Type: text/plain; charset=UTF-8
> Content-Transfer-Encoding: 8bit
>
> There's no need to issue that everytime during boot - we have the
> /proc/cpuinfo flag for people and software to query.
>
> Also, simplify logic which verifies the vls chicken bit setting.
>
> Signed-off-by: Borislav Petkov <bp@xxxxxxx>
> Cc: David Hildenbrand <david@xxxxxxxxxx>
> Cc: Janakarajan Natarajan <Janakarajan.Natarajan@xxxxxxx>
> Cc: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> Cc: Radim Krčmář <rkrcmar@xxxxxxxxxx>
> ---
> arch/x86/kvm/svm.c | 13 ++++---------
> 1 file changed, 4 insertions(+), 9 deletions(-)
>
> diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
> index 8dbd8dbc83eb..d3c481778d9c 100644
> --- a/arch/x86/kvm/svm.c
> +++ b/arch/x86/kvm/svm.c
> @@ -1098,15 +1098,10 @@ static __init int svm_hardware_setup(void)
> }
> }
>
> - if (vls) {
> - if (!npt_enabled ||
> - !boot_cpu_has(X86_FEATURE_V_VMSAVE_VMLOAD) ||
> - !IS_ENABLED(CONFIG_X86_64)) {
> - vls = false;
> - } else {
> - pr_info("Virtual VMLOAD VMSAVE supported\n");
> - }
> - }
> + if (!npt_enabled ||
> + !boot_cpu_has(X86_FEATURE_V_VMSAVE_VMLOAD) ||
> + !IS_ENABLED(CONFIG_X86_64))
> + vls = false;
>
> return 0;
>
>
had the same idea but was worried about runtime. but this is really
only executed once, so
Reviewed-by: David Hildenbrand <david@xxxxxxxxxx>
--
Thanks,
David
