Re: [PATCH] [RFC] x86: kvm: remove KVM_SOFT_MAX_VCPUS

[Andrew Jones <drjones@redhat.com>]

On Sun, Sep 15, 2013 at 12:03:22PM +0300, Gleb Natapov wrote:
> On Sat, Sep 14, 2013 at 02:16:51PM +0200, Andrew Jones wrote:
> > This patch removes KVM_SOFT_MAX_VCPUS and uses num_online_cpus() for
> > KVM_CAP_NR_VCPUS instead, as ARM does. While the API doc simply says
> > KVM_CAP_NR_VCPUS should return the recommended maximum number of vcpus,
> > it has been returning KVM_SOFT_MAX_VCPUS, which was defined as the
> > maximum tested number of vcpus. As that concept could be
> > distro-specific, this patch uses the other recommended maximum, the
> > number of physical cpus, as we never recommend configuring a guest that
> > has more vcpus than the host has pcpus. Of course a guest can always
> > still be configured with up to KVM_CAP_MAX_VCPUS though anyway.
> > 
> > I've put RFC on this patch because I'm not sure if there are any gotchas
> > lurking with this change. The change now means hosts no longer return
> > the same number for KVM_CAP_NR_VCPUS, and that number is likely going to
> > generally be quite a bit less than what KVM_SOFT_MAX_VCPUS was (160). I
> > can't think of anything other than generating more warnings[1] from qemu
> > with guests that configure more vcpus than pcpus though.
> > 
> Another gotcha is that on a host with more then 160 cpus recommended
> value will grow which is not a good idea without appropriate testing.

Good point. Of course the objective could be to test a guest with
vcpus > 160 on that host, and then the potential warning messages would
need to be ignored. Probably the best place to set the cap on the number
of vcpus used in a stable environment would be in KVM_MAX_VCPUS. That said,
then at least until KVM_SOFT_MAX_VCPUS catches up to KVM_MAX_VCPUS, I guess
we should keep them both to avoid breaking anything.

> 
> > [1] Actually, until 972fc544b6034a in uq/master is merged there won't be
> >     any warnings either.
> > 
> > Signed-off-by: Andrew Jones <drjones@redhat.com>
> > ---
> >  arch/x86/include/asm/kvm_host.h | 1 -
> >  arch/x86/kvm/x86.c              | 2 +-
> >  2 files changed, 1 insertion(+), 2 deletions(-)
> > 
> > diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> > index c76ff74a98f2e..9236c63315a9b 100644
> > --- a/arch/x86/include/asm/kvm_host.h
> > +++ b/arch/x86/include/asm/kvm_host.h
> > @@ -32,7 +32,6 @@
> >  #include <asm/asm.h>
> >  
> >  #define KVM_MAX_VCPUS 255
> > -#define KVM_SOFT_MAX_VCPUS 160
> >  #define KVM_USER_MEM_SLOTS 125
> >  /* memory slots that are not exposed to userspace */
> >  #define KVM_PRIVATE_MEM_SLOTS 3
> > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> > index e5ca72a5cdb6d..d9d3e2ed68ee9 100644
> > --- a/arch/x86/kvm/x86.c
> > +++ b/arch/x86/kvm/x86.c
> > @@ -2604,7 +2604,7 @@ int kvm_dev_ioctl_check_extension(long ext)
> >  		r = !kvm_x86_ops->cpu_has_accelerated_tpr();
> >  		break;
> >  	case KVM_CAP_NR_VCPUS:
> > -		r = KVM_SOFT_MAX_VCPUS;
> > +		r = min(num_online_cpus(), KVM_MAX_VCPUS);
> s/KVM_MAX_VCPUS/KVM_SOFT_MAX_VCPUS/.  Also what about hotplug cpus?

I'll send a v2 with this change.

I thought a bit about hotplug, and thus using num_possible_cpus()
instead, but then decided it made more sense to stick to what's online now
for the recommended number. It's just a recommendation anyway. So as long
as KVM_MAX_VCPUS is >= num_possible_cpus(), then one can still configure
more vcpus to count for all hotplugable cpus, if they wish.

drew