RE: [v3 08/26] iommu, x86: Add intel_irq_remapping_capability() for Intel

[Wu, Feng]

> -----Original Message-----
> From: David Woodhouse [mailto:dwmw2@xxxxxxxxxxxxx]
> Sent: Wednesday, January 28, 2015 11:38 PM
> To: Wu, Feng
> Cc: tglx@xxxxxxxxxxxxx; mingo@xxxxxxxxxx; hpa@xxxxxxxxx; x86@xxxxxxxxxx;
> gleb@xxxxxxxxxx; pbonzini@xxxxxxxxxx; joro@xxxxxxxxxx;
> alex.williamson@xxxxxxxxxx; jiang.liu@xxxxxxxxxxxxxxx; eric.auger@xxxxxxxxxx;
> linux-kernel@xxxxxxxxxxxxxxx; iommu@xxxxxxxxxxxxxxxxxxxxxxxxxx;
> kvm@xxxxxxxxxxxxxxx
> Subject: Re: [v3 08/26] iommu, x86: Add intel_irq_remapping_capability() for
> Intel
> 
> On Fri, 2014-12-12 at 23:14 +0800, Feng Wu wrote:
> > Add the Intel side implementation for capability in
> > struct irq_remap_ops.
> >
> > Signed-off-by: Feng Wu <feng.wu@xxxxxxxxx>
> > Reviewed-by: Jiang Liu <jiang.liu@xxxxxxxxxxxxxxx>
> 
> > +static bool intel_irq_remapping_capability(enum irq_remap_cap cap)
> > +{
> > +	struct dmar_drhd_unit *drhd;
> > +	struct intel_iommu *iommu;
> > +
> > +	switch (cap) {
> > +	case IRQ_POSTING_CAP:
> > +		/*
> > +		 * If 1) posted-interrupts is disabled by user
> > +		 * or 2) irq remapping is disabled, posted-interrupts
> > +		 * is not supported.
> > +		 */
> > +		if (disable_irq_post || !irq_remapping_enabled)
> > +			return 0;
> > +
> > +		for_each_iommu(iommu, drhd)
> > +			if (!cap_pi_support(iommu->cap))
> > +				return 0;
> > +
> 
> If a new IOMMU is hotplugged now which doesn't support posted
> interrupts, what happens?

Good question, Just had a offline discussion with Jiang Liu, actually, there
is the same question for IR. In the current implementation, If IR is in use
and a new IOMMU without IR capability is hotplugged, it will reject this
hotplugging. I think I can simple follow the same policy for PI.

Thanks,
Feng

> 
> --
> David Woodhouse                            Open Source Technology
> Centre
> David.Woodhouse@xxxxxxxxx                              Intel
> Corporation