Re: [PATCH v6] KVM: x86: Add x2APIC "features" to control EOI broadcast suppression

[Sean Christopherson]

On Wed, Jan 28, 2026, Khushit Shah wrote:
> 
> > On 28 Jan 2026, at 9:27 AM, Khushit Shah <khushit.shah@xxxxxxxxxxx> wrote:
> > 
> > 
> > On 28/01/26, 9:19 AM, "David Woodhouse" <dwmw2@xxxxxxxxxxxxx> wrote:
> > 
> > On Wed, 2026-01-28 at 02:22 +0000, Huang, Kai wrote:
> > >  
> > > > Ah, so userspace which checks all the kernel's capabilities *first*
> > > > will not see KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST advertised,
> > > > because it needs to enable KVM_CAP_SPLIT_IRQCHIP first?
> > > > > > I guess that's tolerable¹ but the documentation could make it clearer,
> > > > perhaps? I can see VMMs silently failing to detect the feature because
> > > > they just don't set split-irqchip before checking for it? > > > > > > ¹ although I still kind of hate it and would have preferred to have the
> > > >    I/O APIC patch; userspace still has to intentionally *enable* that
> > > >    combination. But OK, I've reluctantly conceded that.
> > > > To make it even more robust, perhaps we can grab kvm->lock mutex in
> > > kvm_vm_ioctl_enable_cap() for KVM_CAP_X2APIC_API, so that it won't race with
> > > KVM_CREATE_IRQCHIP (which already grabs kvm->lock) and
> > > KVM_CAP_SPLIT_IRQCHIP?
> > > > Even more, we can add additional check in KVM_CREATE_IRQCHIP to return -
> > > EINVAL when it sees kvm->arch.suppress_eoi_broadcast_mode is
> > > KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST?
> > 
> > If we do that, then the query for KVM_CAP_X2APIC_API could advertise
> > the KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST for a freshly created KVM,
> > even before userspace has enabled *either* KVM_CREATE_IRQCHIP nor
> > KVM_CAP_SPLIT_IRQCHIP?
> > 
> > That would be slightly better than the existing proposed awfulness
> > where the kernel doesn't *admit* to having the _ENABLE_ capability
> > until userspace first enables the KVM_CAP_SPLIT_IRQCHIP.

No.  If userspace wants to see if *KVM* supports the feature, then userspace can
do KVM_CHECK_EXTENSION on /dev/kvm.  If userspace does KVM_CHECK_EXTENSION on a
VM fd, then KVM absolutely must report exactly what that VM supports.

> How about we make an explicit _ENABLE_ bit for split IRQCHIP?
> When/if in-kernel IRQCHIP starts supporting I/O APIC 0x20, we
> can add a separate bit for that in the CAP. 

NAK.  Conditionally enumerating support for a feature based on the configuration
of the VM has been KVM's documented behavior since KVM_CHECK_EXTENSION was added
by commit 92b591a4c46b ("KVM: Allow KVM_CHECK_EXTENSION on the vm fd").

I don't see any reason why KVM_X2APIC_ENABLE_SUPPRESS_EOI_BROADCAST needs to do
something different.