Re: [PATCH v2 2/3] x86/vmscape: Replace IBPB with branch history clear on exit to userspace

[Sean Christopherson <seanjc@google.com>]

On Mon, Oct 20, 2025, Pawan Gupta wrote:
> On Mon, Oct 20, 2025 at 09:10:17AM -0700, Sean Christopherson wrote:
> > On Wed, Oct 15, 2025, Pawan Gupta wrote:
> > > diff --git a/arch/x86/include/asm/nospec-branch.h b/arch/x86/include/asm/nospec-branch.h
> > > index 49707e563bdf71bdd05d3827f10dd2b8ac6bca2c..00730cc22c2e7115f6dbb38a1ed8d10383ada5c0 100644
> > > --- a/arch/x86/include/asm/nospec-branch.h
> > > +++ b/arch/x86/include/asm/nospec-branch.h
> > > @@ -534,7 +534,7 @@ void alternative_msr_write(unsigned int msr, u64 val, unsigned int feature)
> > >  		: "memory");
> > >  }
> > >  
> > > -DECLARE_PER_CPU(bool, x86_ibpb_exit_to_user);
> > > +DECLARE_PER_CPU(bool, x86_pred_flush_pending);
> > 
> > Rather than "flush pending", what about using "need" in the name to indicate that
> > a flush is necessary?  That makes it more obvious that e.g. KVM is marking the
> > CPU as needing a flush by some other code, as opposed to implying that KVM itself
> > has a pending flush.
> > 
> > And maybe spell out "prediction"?  Without the context of features being checked,
> > I don't know that I would be able to guess "prediction".
> > 
> > E.g. x86_need_prediction_flush?
> > 
> > Or x86_prediction_flush_exit_to_user if we would prefer to clarify when the flush
> > needs to occur?
> 
> Ok, ya this is more clear. I would want to make a small change, instead of
> "prediction_flush", "predictor_flush" reads better to me. Changing it to:
> x86_predictor_flush_exit_to_user.

LOL, see, told you I couldn't guest the word. :-D

"predictor" is way better, thanks!