On Thu, Jul 08, 2021 at 09:59:08AM +0530, Bharata B Rao wrote:
> On Thu, Jul 08, 2021 at 01:58:04PM +1000, David Gibson wrote:
> > On Tue, Jul 06, 2021 at 10:56:32AM +0530, Bharata B Rao wrote:
> > > On Mon, Jul 05, 2021 at 02:42:33PM +1000, David Gibson wrote:
> > > > On Mon, Jun 21, 2021 at 02:20:00PM +0530, Bharata B Rao wrote:
> > > > > diff --git a/arch/powerpc/include/asm/mmu_context.h b/arch/powerpc/include/asm/mmu_context.h
> > > > > index 4bc45d3ed8b0..b44f291fc909 100644
> > > > > --- a/arch/powerpc/include/asm/mmu_context.h
> > > > > +++ b/arch/powerpc/include/asm/mmu_context.h
> > > > > @@ -124,8 +124,17 @@ static inline bool need_extra_context(struct mm_struct *mm, unsigned long ea)
> > > > >  
> > > > >  #if defined(CONFIG_KVM_BOOK3S_HV_POSSIBLE) && defined(CONFIG_PPC_RADIX_MMU)
> > > > >  extern void radix_kvm_prefetch_workaround(struct mm_struct *mm);
> > > > > +void do_h_rpt_invalidate_prt(unsigned long pid, unsigned long lpid,
> > > > > +			     unsigned long type, unsigned long pg_sizes,
> > > > > +			     unsigned long start, unsigned long end);
> > > > >  #else
> > > > >  static inline void radix_kvm_prefetch_workaround(struct mm_struct *mm) { }
> > > > > +static inline void do_h_rpt_invalidate_prt(unsigned long pid,
> > > > > +					   unsigned long lpid,
> > > > > +					   unsigned long type,
> > > > > +					   unsigned long pg_sizes,
> > > > > +					   unsigned long start,
> > > > > +					   unsigned long end) { }
> > > > 
> > > > Since the only plausible caller is in KVM HV code, why do you need the
> > > > #else clause.
> > > 
> > > The call to the above routine is prevented for non-radix guests
> > > in KVM HV code at runtime using kvm_is_radix() check and not by
> > > CONFIG_PPC_RADIX_MMU. Hence the #else version would be needed.
> > 
> > kvm_is_radix() should evaluate to false at compile time if
> > !defined(CONFIG_PPC_RADIX_MMU), in which case, no you shouldn't need
> > the else version.
> 
> At least in the latest mainline, I don't see the definition of
> kvm_is_radix() being conditional to CONFIG_PPC_RADIX_MMU.

That's probably worth fixing, then.

> Anyway this is what I see in practice if the #else version is
> removed and CONFIG_PPC_RADIX_MMU is turned off:
> 
> arch/powerpc/kvm/book3s_hv.c: In function ‘kvmppc_h_rpt_invalidate’:
> arch/powerpc/kvm/book3s_hv.c:983:2: error: implicit declaration of function ‘do_h_rpt_invalidate_prt’; did you mean ‘do_h_rpt_invalidate_pat’? [-Werror=implicit-function-declaration]
>   983 |  do_h_rpt_invalidate_prt(id, vcpu->kvm->arch.lpid,
>       |  ^~~~~~~~~~~~~~~~~~~~~~~
>       |  do_h_rpt_invalidate_pat

If you really do need the stub, it should probably be a BUG() rather
than a no-op, since this really never should get called if we don't
have radix.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson