Re: [Xen-devel] [PATCH v4 04/12] xen/arm: support for guest SGI

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Mon, 2013-04-29 at 16:47 +0100, Stefano Stabellini wrote:
> On Mon, 29 Apr 2013, Ian Campbell wrote:
> > On Sun, 2013-04-28 at 15:26 +0100, Stefano Stabellini wrote:
> > > On Fri, 26 Apr 2013, Ian Campbell wrote:
> > > > > +
> > > > > +
> > > > > +    for_each_set_bit( vcpuid, &vcpu_mask, d->max_vcpus )
> > > > > +    {
> > > > > +        if ( vcpuid >= d->max_vcpus || (vt = d->vcpu[vcpuid]) == 
> > > > > NULL )
> > > > 
> > > > Is d->vcpu[vcpuid] == NULL sufficient here? A vcpu which exists but has
> > > > not been PSCI'd up will pass this -- what do we do to them? Hopefully we
> > > > don't wake them up? Do we not want to check something like _VPF_down
> > > > too?
> > > 
> > > I think you are right, we should test for _VPF_down too
> > 
> > Are there any interesting races here? Seems like there should be, unless
> > we hold the domain lock or something?
> 
> 
> Considering that max_vcpus is static, d->vcpu[vcpuid] has been
> allocated at domain creation and that test_bit should be atomic, I don't
> think there are any races here.
> What am I missing?

The race between checking the bit and acting on the result.

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel