Re: [Qemu-ppc] [RFC PATCH v2 17/23] xics_kvm: Don't enable KVM_CAP_IRQ_XICS if already enabled

[David Gibson]

On Wed, Mar 25, 2015 at 02:42:24PM +0530, Bharata B Rao wrote:
> On Wed, Mar 25, 2015 at 04:24:39PM +1100, David Gibson wrote:
> > On Mon, Mar 23, 2015 at 07:05:58PM +0530, Bharata B Rao wrote:
> > > When supporting CPU hot removal by parking the vCPU fd and reusing
> > > it during hotplug again, there can be cases where we try to reenable
> > > KVM_CAP_IRQ_XICS CAP for the vCPU for which it was already enabled.
> > > Introduce a boolean member in ICPState to track this and don't
> > > reenable the CAP if it was already enabled earlier.
> > > 
> > > This change allows CPU hot removal to work for sPAPR.
> > > 
> > > Signed-off-by: Bharata B Rao <address@hidden>
> > 
> > Why does double enabling the capability cause problems?  I would have
> > expected it to be unnecessary, but harmless.
> 
> We are reusing the vCPU here w/o closing its fd.
> 
> As things stand currently, enabling this cap again will result in
> kernel trying to create and associate ICP with this vCPU and that
> fails since there is already an ICP associated with it.
> 
> Ref: arch/powerpc/kvm/book3s_xics.c:kvmppc_xics_connect_vcpu() kernel code.

Ah, right.  Sounds like a kernel bug - I would expect enabling a
capability to be idempotent.  But since the bug's there, we have to
work around it now, so ok.

> So this patch will ensure that we don't renable this cap.
> 
> Regards,
> Bharata.
> 

-- 
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

pgpOV4e4CRKVN.pgp
Description: PGP signature