Re: [PATCH v2 14/18] KVM: SVM: Check that the current CPU supports SVM in kvm_is_svm_supported()

[Huang, Kai]

On Mon, 2023-03-13 at 10:29 -0700, Sean Christopherson wrote:
> On Mon, Mar 13, 2023, Huang, Kai wrote:
> > On Fri, 2023-03-10 at 13:42 -0800, Sean Christopherson wrote:
> > > Check "this" CPU instead of the boot CPU when querying SVM support so that
> > > the per-CPU checks done during hardware enabling actually function as
> > > intended, i.e. will detect issues where SVM isn't support on all CPUs.
> > > 
> > > Disable migration for the use from svm_init() mostly so that the standard
> > > accessors for the per-CPU data can be used without getting yelled at by
> > > CONFIG_DEBUG_PREEMPT=y sanity checks.  Preventing the "disabled by BIOS"
> > > error message from reporting the wrong CPU is largely a bonus, as ensuring
> > > a stable CPU during module load is a non-goal for KVM.
> > > 
> > > Link: https://lore.kernel.org/all/ZAdxNgv0M6P63odE@xxxxxxxxxx
> > > Cc: Kai Huang <kai.huang@xxxxxxxxx>
> > > Cc: Chao Gao <chao.gao@xxxxxxxxx>
> > > Signed-off-by: Sean Christopherson <seanjc@xxxxxxxxxx>
> > 
> > Should we add:
> > 
> > Fixes: c82a5c5c53c5 ("KVM: x86: Do compatibility checks when onlining CPU")
> > 
> > As that commit introduced using raw_smp_processor_id() to get CPU id in
> > kvm_is_svm_supported() and print the CPU id out in error message?
> 
> My vote is to not to add a Fixes because using raw_smp_processor_id() and not disabling
> migration for module probe case was deliberate and is safe.  I don't want to give the
> impression that the existing code is functionally broken.  The only quirk is that
> the reporting could be misleading.
> 
> That said, I'm not against adding a Fixes tag, because I certainly can't argue
> against the reporting being flawed.

Yeah the only issue is the reporting.

And I will leave this to others.

> 
> > > ---
> > >  arch/x86/kvm/svm/svm.c | 25 +++++++++++++++++++------
> > >  1 file changed, 19 insertions(+), 6 deletions(-)
> > > 
> > > diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> > > index 2934f185960d..f04b61c3d9d8 100644
> > > --- a/arch/x86/kvm/svm/svm.c
> > > +++ b/arch/x86/kvm/svm/svm.c
> > > @@ -520,18 +520,20 @@ static void svm_init_osvw(struct kvm_vcpu *vcpu)
> > >  		vcpu->arch.osvw.status |= 1;
> > >  }
> > >  
> > > -static bool kvm_is_svm_supported(void)
> > > +static bool __kvm_is_svm_supported(void)
> > >  {
> > > -	int cpu = raw_smp_processor_id();
> > > +	int cpu = smp_processor_id();
> > 
> > Since we have made sure __kvm_is_svm_supported() is always performed on a stable
> > cpu, should we keep using raw_smp_processor_id()? �
> > 
> > It is faster than smp_processor_id() when CONFIG_DEBUG_PREEMPT=y, but yes the
> > latter can help to catch bug.
> 
> Most kernels with any amount of CONFIG_DEBUG_* options enabled are comically slow
> anyways, I much prefer having the sanity checks than the performance.

Yeah fine to me.