Re: [PATCH v5 13/17] irqdomain: irq_domain_check_msi_remap

[Marc Zyngier]

Hi Eric,

On 04/01/17 13:32, Eric Auger wrote:
> This new function checks whether all platform and PCI
> MSI domains implement IRQ remapping. This is useful to
> understand whether VFIO passthrough is safe with respect
> to interrupts.
> 
> On ARM typically an MSI controller can sit downstream
> to the IOMMU without preventing VFIO passthrough.
> As such any assigned device can write into the MSI doorbell.
> In case the MSI controller implements IRQ remapping, assigned
> devices will not be able to trigger interrupts towards the
> host. On the contrary, the assignment must be emphasized as
> unsafe with respect to interrupts.
> 
> Signed-off-by: Eric Auger <eric.auger@xxxxxxxxxx>
> 
> ---
> 
> v4 -> v5:
> - Handle DOMAIN_BUS_FSL_MC_MSI domains
> - Check parents
> ---
>  include/linux/irqdomain.h |  1 +
>  kernel/irq/irqdomain.c    | 41 +++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 42 insertions(+)
> 
> diff --git a/include/linux/irqdomain.h b/include/linux/irqdomain.h
> index ab017b2..281a40f 100644
> --- a/include/linux/irqdomain.h
> +++ b/include/linux/irqdomain.h
> @@ -219,6 +219,7 @@ struct irq_domain *irq_domain_add_legacy(struct device_node *of_node,
>  					 void *host_data);
>  extern struct irq_domain *irq_find_matching_fwspec(struct irq_fwspec *fwspec,
>  						   enum irq_domain_bus_token bus_token);
> +extern bool irq_domain_check_msi_remap(void);
>  extern void irq_set_default_host(struct irq_domain *host);
>  extern int irq_domain_alloc_descs(int virq, unsigned int nr_irqs,
>  				  irq_hw_number_t hwirq, int node,
> diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c
> index 8c0a0ae..700caea 100644
> --- a/kernel/irq/irqdomain.c
> +++ b/kernel/irq/irqdomain.c
> @@ -278,6 +278,47 @@ struct irq_domain *irq_find_matching_fwspec(struct irq_fwspec *fwspec,
>  EXPORT_SYMBOL_GPL(irq_find_matching_fwspec);
>  
>  /**
> + * irq_domain_is_msi_remap - Check if @domain or any parent
> + * has MSI remapping support
> + * @domain: domain pointer
> + */
> +static bool irq_domain_is_msi_remap(struct irq_domain *domain)
> +{
> +	struct irq_domain *h = domain;
> +
> +	for (; h; h = h->parent) {
> +		if (h->flags & IRQ_DOMAIN_FLAG_MSI_REMAP)
> +			return true;
> +	}
> +	return false;
> +}
> +
> +/**
> + * irq_domain_check_msi_remap() - Checks whether all MSI
> + * irq domains implement IRQ remapping
> + */
> +bool irq_domain_check_msi_remap(void)
> +{
> +	struct irq_domain *h;
> +	bool ret = true;
> +
> +	mutex_lock(&irq_domain_mutex);
> +	list_for_each_entry(h, &irq_domain_list, link) {
> +		if (((h->bus_token & DOMAIN_BUS_PCI_MSI) ||
> +		     (h->bus_token & DOMAIN_BUS_PLATFORM_MSI) ||
> +		     (h->bus_token & DOMAIN_BUS_FSL_MC_MSI)) &&
> +		     !irq_domain_is_msi_remap(h)) {

(h->bus_token & DOMAIN_BUS_PCI_MSI) and co looks quite wrong. bus_token
is not a bitmap, and DOMAIN_BUS_* not a single bit value (see enum
irq_domain_bus_token). Surely this should read
(h->bus_token == DOMAIN_BUS_PCI_MSI).

Thanks,

	M.
-- 
Jazz is not dead. It just smells funny...