Re: [PATCH v3 11/16] KVM: Move vcpu_load to arch-specific kvm_arch_vcpu_ioctl_set_guest_debug

[Christoffer Dall <christoffer.dall@linaro.org>]

On Mon, Dec 11, 2017 at 01:39:43PM +0100, Cornelia Huck wrote:
> On Mon,  4 Dec 2017 21:35:33 +0100
> Christoffer Dall <cdall@kernel.org> wrote:
> 
> > From: Christoffer Dall <christoffer.dall@linaro.org>
> > 
> > Move vcpu_load() and vcpu_put() into the architecture specific
> > implementations of kvm_arch_vcpu_ioctl_set_guest_debug().
> > 
> > Reviewed-by: David Hildenbrand <david@redhat.com>
> > Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
> > ---
> >  arch/arm64/kvm/guest.c    | 15 ++++++++++++---
> >  arch/powerpc/kvm/book3s.c |  2 ++
> >  arch/powerpc/kvm/booke.c  | 19 +++++++++++++------
> >  arch/s390/kvm/kvm-s390.c  | 16 ++++++++++++----
> >  arch/x86/kvm/x86.c        |  4 +++-
> >  virt/kvm/kvm_main.c       |  2 --
> >  6 files changed, 42 insertions(+), 16 deletions(-)
> > 
> 
> > diff --git a/arch/powerpc/kvm/booke.c b/arch/powerpc/kvm/booke.c
> > index 1b491b89cd43..7cb0e2677e60 100644
> > --- a/arch/powerpc/kvm/booke.c
> > +++ b/arch/powerpc/kvm/booke.c
> > @@ -2018,12 +2018,15 @@ int kvm_arch_vcpu_ioctl_set_guest_debug(struct 
> > kvm_vcpu *vcpu,
> >  {
> >     struct debug_reg *dbg_reg;
> >     int n, b = 0, w = 0;
> > +   int ret = 0;
> > +
> > +   vcpu_load(vcpu);
> >  
> >     if (!(dbg->control & KVM_GUESTDBG_ENABLE)) {
> >             vcpu->arch.dbg_reg.dbcr0 = 0;
> >             vcpu->guest_debug = 0;
> >             kvm_guest_protect_msr(vcpu, MSR_DE, false);
> > -           return 0;
> > +           goto out;
> >     }
> >  
> >     kvm_guest_protect_msr(vcpu, MSR_DE, true);
> > @@ -2055,8 +2058,9 @@ int kvm_arch_vcpu_ioctl_set_guest_debug(struct 
> > kvm_vcpu *vcpu,
> >  #endif
> >  
> >     if (!(vcpu->guest_debug & KVM_GUESTDBG_USE_HW_BP))
> > -           return 0;
> > +           goto out;
> >  
> > +   ret = -EINVAL;
> >     for (n = 0; n < (KVMPPC_BOOKE_IAC_NUM + KVMPPC_BOOKE_DAC_NUM); n++) {
> >             uint64_t addr = dbg->arch.bp[n].addr;
> >             uint32_t type = dbg->arch.bp[n].type;
> > @@ -2067,21 +2071,24 @@ int kvm_arch_vcpu_ioctl_set_guest_debug(struct 
> > kvm_vcpu *vcpu,
> >             if (type & ~(KVMPPC_DEBUG_WATCH_READ |
> >                          KVMPPC_DEBUG_WATCH_WRITE |
> >                          KVMPPC_DEBUG_BREAKPOINT))
> > -                   return -EINVAL;
> > +                   goto out;
> >  
> >             if (type & KVMPPC_DEBUG_BREAKPOINT) {
> >                     /* Setting H/W breakpoint */
> >                     if (kvmppc_booke_add_breakpoint(dbg_reg, addr, b++))
> > -                           return -EINVAL;
> > +                           goto out;
> >             } else {
> >                     /* Setting H/W watchpoint */
> >                     if (kvmppc_booke_add_watchpoint(dbg_reg, addr,
> >                                                     type, w++))
> > -                           return -EINVAL;
> > +                           goto out;
> >             }
> >     }
> >  
> > -   return 0;
> > +   ret = 0;
> 
> I would probably set the -EINVAL in the individual branches (so it is
> clear that something is wrong, and it is not just a benign exit as in
> the cases above), but your code is correct as well.

I think that's better as well actually.  I got probably got a little
used to that pattern after looking the main dispatcher function for a
while.  I'm happy to change it.

> > +out:
> > +   vcpu_put(vcpu);
> > +   return ret;
> >  }
> >  
> >  void kvmppc_booke_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
> 
> In any case,
> 
> Reviewed-by: Cornelia Huck <cohuck@redhat.com>

Thanks!
-Christoffer