Re: [Xen-devel] [PATCH v6 10/31] xen/arm: ITS: Introduce gic_is_lpi helper function

[Vijay Kilari <vijay.kilari@xxxxxxxxx>]

On Tue, Sep 1, 2015 at 6:32 PM, Julien Grall <julien.grall@xxxxxxxxxx> wrote:
> On 01/09/15 12:56, Vijay Kilari wrote:
>>> BTW, I suggested to create a field nr_lpis but you decided to store the
>>> number of bits supported. Why?
>>
>> I have nr_lpis field in vgic structure (patch #17). But it just tells
>> how LPIs are supported
>> for a domain.
>
> Why are you speaking about vgic structure? I'm only suggesting to
> replace you nr_id_bits by nr_lpis in the hw GIC. AFAICT, there is
> nothing to prevent having 2 field using the same name on 2 differents
> structure...
>
>> Where as nr_id_bits shows total number of  lpis that hw supports.
>
> No nr_id_bits shows the total number of interrupt not LPIs. The total
> number of LPIs is (1 << nr_id_bits) - 8092. Although (1 << nr_id_bits)
> gives you the last LPI interrupt ID supported.
>
> Anyway, as I said earlier, re-calculating the last LPI interrupt ID
> everytime based on the shift is time consuming. You should optimize for
> the common case rather than using copy the raw value (i.e ID bits) from
> the HW directly.
>
> Maybe the name "max_lpi_id" would make more sense to you for a name?

Instead of storing max_lpi_id which is ( 1 << nr_id_bits) - 8192, I prefer
to rename nr_id_bits as nr_irq_ids which can be initialized to (1 << nr_id_bits)
for gicv3 and gicv2_info.nr_lines for gicv2.

and gic_is_lpi() as below will work both gicv2 and gicv3.

bool_t gic_is_lpi(unsigned int irq)
{
    return (irq >= FIRST_GIC_LPI && irq < gic_hw_ops->info->nr_irq_ids);
}

Regards
Vijay

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel