Re: [PATCH v4 1/2] target/arm: kvm: Handle DABT with no valid ISS

[Beata Michalska]

On Fri, 17 Apr 2020 at 14:10, Andrew Jones <address@hidden> wrote:
>
> On Fri, Apr 17, 2020 at 11:39:25AM +0100, Peter Maydell wrote:
> > On Mon, 23 Mar 2020 at 11:32, Beata Michalska
> > <address@hidden> wrote:
> > >
> > > On ARMv7 & ARMv8 some load/store instructions might trigger a data abort
> > > exception with no valid ISS info to be decoded. The lack of decode info
> > > makes it at least tricky to emulate those instruction which is one of the
> > > (many) reasons why KVM will not even try to do so.
> > >
> > > Add support for handling those by requesting KVM to inject external
> > > dabt into the quest.
> > >
> > > Signed-off-by: Beata Michalska <address@hidden>
> > > ---
> > >  target/arm/cpu.h     |  2 ++
> > >  target/arm/kvm.c     | 54 
> > > ++++++++++++++++++++++++++++++++++++++++++++++++++++
> > >  target/arm/kvm_arm.h | 11 +++++++++++
> > >  3 files changed, 67 insertions(+)
> > >
> > > diff --git a/target/arm/cpu.h b/target/arm/cpu.h
> > > index 4ffd991..4f834c1 100644
> > > --- a/target/arm/cpu.h
> > > +++ b/target/arm/cpu.h
> > > @@ -560,6 +560,8 @@ typedef struct CPUARMState {
> > >          uint64_t esr;
> > >      } serror;
> > >
> > > +    uint8_t ext_dabt_pending; /* Request for injecting ext DABT */
> >
> > I was trying to work out whether we need to migrate this state,
> > and I'm not sure. Andrew, do you know? I think this comes down
> > to "at what points in QEMU's kvm run loop can migration kick in",
> > and specifically if we get a KVM_EXIT_ARM_NISV do we definitely
> > go round the loop and KVM_RUN again without ever checking
> > to see if we should do a migration ?
> >
>
> I'd prefer a migration expert confirm this, so I've CC'ed David and Juan,
> but afaict there's no way to break out of the KVM_RUN loop after a
> successful (ret=0) call to kvm_arch_handle_exit() until after the next
> KVM_RUN ioctl. This is because even if migration kicks the vcpus between
> kvm_arch_handle_exit() and the next run, the signal won't do anything
> other than prepare the vcpu for an immediate exit.
>
I am definitely not an expert on that one, but if I got things right,
by the time the 'exit_request' gets verified , the external abort
should already be set up , the pending status cleared (through
KVM_SET_VCPU_EVENTS)
and the reg content verified (kvm_arch_pre_run), as all of it is being
 triggered
prior to checking the exit request. So this should not need a
dedicated migration state.

I will hold on with sending the new version though to get the confirmation
whether that is the case.

Thanks,

BR
Beata
>
> Thanks,
> drew
>