Re: [PATCH 1/2] xen/arm: Improve handling of nr_spis

[Bertrand Marquis <Bertrand.Marquis@xxxxxxx>]

Hi Michal,

> On 11 Mar 2025, at 10:04, Michal Orzel <michal.orzel@xxxxxxx> wrote:
> 
> At the moment, we print a warning about max number of IRQs supported by
> GIC bigger than vGIC only for hardware domain. This check is not hwdom
> special, and should be made common. Also, in case of user not specifying
> nr_spis for dom0less domUs, we should take into account max number of
> IRQs supported by vGIC if it's smaller than for GIC.
> 
> Introduce VGIC_MAX_IRQS macro and use it instead of hardcoded 992 value.
> Fix calculation of nr_spis for dom0less domUs and make the GIC/vGIC max
> IRQs comparison common.
> 
> Signed-off-by: Michal Orzel <michal.orzel@xxxxxxx>
> ---
> xen/arch/arm/dom0less-build.c   | 2 +-
> xen/arch/arm/domain_build.c     | 9 ++-------
> xen/arch/arm/gic.c              | 3 +++
> xen/arch/arm/include/asm/vgic.h | 3 +++
> 4 files changed, 9 insertions(+), 8 deletions(-)
> 
> diff --git a/xen/arch/arm/dom0less-build.c b/xen/arch/arm/dom0less-build.c
> index 31f31c38da3f..9a84fee94119 100644
> --- a/xen/arch/arm/dom0less-build.c
> +++ b/xen/arch/arm/dom0less-build.c
> @@ -1018,7 +1018,7 @@ void __init create_domUs(void)
>         {
>             int vpl011_virq = GUEST_VPL011_SPI;
> 
> -            d_cfg.arch.nr_spis = gic_number_lines() - 32;
> +            d_cfg.arch.nr_spis = min(gic_number_lines(), VGIC_MAX_IRQS) - 32;

I would suggest to introduce a static inline gic_nr_spis in a gic header ...

> 
>             /*
>              * The VPL011 virq is GUEST_VPL011_SPI, unless direct-map is
> diff --git a/xen/arch/arm/domain_build.c b/xen/arch/arm/domain_build.c
> index 7cc141ef75e9..b99c4e3a69bf 100644
> --- a/xen/arch/arm/domain_build.c
> +++ b/xen/arch/arm/domain_build.c
> @@ -2371,13 +2371,8 @@ void __init create_dom0(void)
> 
>     /* The vGIC for DOM0 is exactly emulating the hardware GIC */
>     dom0_cfg.arch.gic_version = XEN_DOMCTL_CONFIG_GIC_NATIVE;
> -    /*
> -     * Xen vGIC supports a maximum of 992 interrupt lines.
> -     * 32 are substracted to cover local IRQs.
> -     */
> -    dom0_cfg.arch.nr_spis = min(gic_number_lines(), (unsigned int) 992) - 32;
> -    if ( gic_number_lines() > 992 )
> -        printk(XENLOG_WARNING "Maximum number of vGIC IRQs exceeded.\n");
> +    /* 32 are substracted to cover local IRQs */
> +    dom0_cfg.arch.nr_spis = min(gic_number_lines(), VGIC_MAX_IRQS) - 32;

and reuse it here to make sure the value used is always the same.

>     dom0_cfg.arch.tee_type = tee_get_type();
>     dom0_cfg.max_vcpus = dom0_max_vcpus();
> 
> diff --git a/xen/arch/arm/gic.c b/xen/arch/arm/gic.c
> index acf61a4de373..e80fe0ca2421 100644
> --- a/xen/arch/arm/gic.c
> +++ b/xen/arch/arm/gic.c
> @@ -251,6 +251,9 @@ void __init gic_init(void)
>         panic("Failed to initialize the GIC drivers\n");
>     /* Clear LR mask for cpu0 */
>     clear_cpu_lr_mask();
> +
> +    if ( gic_number_lines() > VGIC_MAX_IRQS )
> +        printk(XENLOG_WARNING "Maximum number of vGIC IRQs exceeded\n");

I am a bit unsure with this one.
If this is the case, it means any gicv2 or gicv3 init detected an impossible 
value and
any usage of gic_number_lines would be using an impossible value.

Shouldn't this somehow result in a panic as the gic detection was wrong ?
Do you think we can continue to work safely if this value is over VGIC_MAX_IRQS.
There are other places using gic_number_lines like in irq.c.

Cheers
Bertrand


> }
> 
> void send_SGI_mask(const cpumask_t *cpumask, enum gic_sgi sgi)
> diff --git a/xen/arch/arm/include/asm/vgic.h b/xen/arch/arm/include/asm/vgic.h
> index e309dca1ad01..c549e5840bfa 100644
> --- a/xen/arch/arm/include/asm/vgic.h
> +++ b/xen/arch/arm/include/asm/vgic.h
> @@ -329,6 +329,9 @@ extern void vgic_check_inflight_irqs_pending(struct vcpu 
> *v,
>  */
> #define vgic_num_irqs(d)        ((d)->arch.vgic.nr_spis + 32)
> 
> +/* Maximum number of IRQs supported by vGIC */
> +#define VGIC_MAX_IRQS 992U
> +
> /*
>  * Allocate a guest VIRQ
>  *  - spi == 0 => allocate a PPI. It will be the same on every vCPU
> -- 
> 2.25.1
>