Re: [PATCH 15/31] KVM: PPC: Book3S HV: XIVE: Fix mapping of passthrough interrupts

[Marc Zyngier <maz@kernel.org>]

On Fri, 14 May 2021 21:51:51 +0100,
Thomas Gleixner <tglx@linutronix.de> wrote:
> 
> On Fri, Apr 30 2021 at 10:03, Cédric Le Goater wrote:
> 
> CC: +Marc

Thanks Thomas.

> 
> > PCI MSI interrupt numbers are now mapped in a PCI-MSI domain but the
> > underlying calls handling the passthrough of the interrupt in the
> > guest need a number in the XIVE IRQ domain.
> >
> > Use the IRQ data mapped in the XIVE IRQ domain and not the one in the
> > PCI-MSI domain.
> >
> > Exporting irq_get_default_host() might not be the best solution.
> >
> > Cc: Thomas Gleixner <tglx@linutronix.de>
> > Cc: Paul Mackerras <paulus@ozlabs.org>
> > Signed-off-by: Cédric Le Goater <clg@kaod.org>
> > ---
> >  arch/powerpc/kvm/book3s_xive.c | 3 ++-
> >  kernel/irq/irqdomain.c         | 1 +
> >  2 files changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/arch/powerpc/kvm/book3s_xive.c b/arch/powerpc/kvm/book3s_xive.c
> > index 3a7da42bed57..81b9f4fc3978 100644
> > --- a/arch/powerpc/kvm/book3s_xive.c
> > +++ b/arch/powerpc/kvm/book3s_xive.c
> > @@ -861,7 +861,8 @@ int kvmppc_xive_set_mapped(struct kvm *kvm, unsigned long guest_irq,
> >  	struct kvmppc_xive *xive = kvm->arch.xive;
> >  	struct kvmppc_xive_src_block *sb;
> >  	struct kvmppc_xive_irq_state *state;
> > -	struct irq_data *host_data = irq_get_irq_data(host_irq);
> > +	struct irq_data *host_data =
> > +		irq_domain_get_irq_data(irq_get_default_host(), host_irq);
> >  	unsigned int hw_irq = (unsigned int)irqd_to_hwirq(host_data);
> >  	u16 idx;
> >  	u8 prio;
> > diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c
> > index d10ab1d689d5..8a073d1ce611 100644
> > --- a/kernel/irq/irqdomain.c
> > +++ b/kernel/irq/irqdomain.c
> > @@ -481,6 +481,7 @@ struct irq_domain *irq_get_default_host(void)
> >  {
> >  	return irq_default_domain;
> >  }
> > +EXPORT_SYMBOL_GPL(irq_get_default_host);
> >  
> >  static void irq_domain_clear_mapping(struct irq_domain *domain,
> >  				     irq_hw_number_t hwirq)
> 

Is there any reason why we should add more users of the "default host"
fallback? I would really hope that new code would actually track their
irqdomain in a more fine-grained way, specially when using the
hierarchical MSi setup, which seems to be the goal of this series.

Don't you have enough topology information that you can make use of to
correctly assign a domain identifier (of_node or otherwise)?

Thanks,

	M.

-- 
Without deviation from the norm, progress is not possible.