Re: [Xen-devel] [PATCH v2] xen/arm: gicv2: Export GICv2m register frames to domain0 by device tree

[Wei Chen <wei.chen@xxxxxxxxxx>]

Hi Julien,

On 27 April 2016 at 17:40, Julien Grall <julien.grall@xxxxxxx> wrote:
>
>
> On 27/04/2016 07:05, Wei Chen wrote:
>>
>> Hi Julien,
>
>
> Hi Wei,
>
>> On 26 April 2016 at 18:49, Julien Grall <julien.grall@xxxxxxx> wrote:
>>>
>>> On 25/04/2016 10:39, Wei Chen wrote:
>>>>
>>>> +                   d->domain_id);
>>>> +            return ret;
>>>> +        }
>>>> +
>>>> +        /*
>>>> +         * Map all SPIs that are allocated to MSIs for the frame to the
>>>> +         * domain.
>>>> +         */
>>>> +        for ( spi = v2m_data->spi_start;
>>>> +              spi < (v2m_data->spi_start + v2m_data->nr_spis); spi++ )
>>>> +        {
>>>> +            /*
>>>> +             * MSIs are always edge-triggered. Configure the associated
>>>> SPIs
>>>> +             * to be edge-rising.
>>>
>>>
>>>
>>> How did you find that SPIs should be configured edge-rising?
>>
>> Before route_irq_to_guest, the SPI must be configured. I found Linux
>> v2m driver set
>> the SPI type to edge-rising, so I set edge-rising as v2m SPI default
>> type here too.
>
>
> Well, Linux did it for a good reason (i.e based on a spec). We are trying in
> Xen to document piece of code which may not be straight-forward to deduce.
>
> In this case, the IRQ is configured edge-rising because this is the only
> edge type supported by SPIs. I would expand the second sentence of the
> comment to explain that.
>

Thanks for your explanation. I have addressed your other comments in
the v3 patch.
So I will add the reason of configuring SPIs to edge-rising to v4 patch.

> Regards,
>
> --
> Julien Grall

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