[PATCH 1/3] misc: Add crossbar driver

[Santosh Shilimkar]

On Thursday 15 August 2013 04:51 PM, Linus Walleij wrote:
> On Thu, Aug 15, 2013 at 10:26 PM, Santosh Shilimkar
> <santosh.shilimkar at ti.com> wrote:
>> On Thursday 15 August 2013 04:01 PM, Linus Walleij wrote:
>>> On Tue, Aug 13, 2013 at 11:56 AM, Sricharan R <r.sricharan at ti.com> wrote:
>>>
>>>>  Initially irqchip was discussed, but we also have a DMA crossbar
>>>>  to map the dma-requests. Since both irq/dma crossbars should be handled,
>>>>  pinctrl was suggested as the appropriate place to handle this.
>>>
>>> I think it is better to use irqchip.
>>>
>> Did you happen to read the thread why irqchip is in-appropriate
>> for such an IP.
> 
> Sorry I don't understand what thread that is... can you point me there?
> My previous statement on this issue what this:
> http://marc.info/?l=linux-kernel&m=137442541628641&w=2
> 
It was discussed in couple of threads but the main point was the
need of a link needed for the irqchip.

>> As I said earlier, an IRQ-chip always need a
>> real IRQ link (even for the chained one) to the primary irqchip.
>>
>> This IP is just dummy IP makes the connections for the primary
>> irqchip(read GIC). And its use only limited to make the
>> connection between the peripheral IRQ event to the GIC IRQ line.
>>
>> I don't see how you can make this happen with an irqchip
>> infrastructure.
> 
> I think my post above describes this.
> 
Sorry for being dumb but I don't think cascaded irqchip examples
like GPIO and cross-bars are same. If you take an example of
GPIO irqchip, it always have a physical connection even if it
is 1 IRQ line for (32 logical/sparse IRQs). That goes with
other MFD examples too.

So may be I am still missing something in your proposal.

>>> I don't see any way to really abstract this pretty simple crossbar
>>> for reuse across subsystems.
>>>
>> This exactly the reason, i am against idea of over-engineering the
>> simple IP whose only job is to make the physical wire connection
>> in software where as this is generally done in RTL by default on
>> most of the SOCs.
> 
> Well, it was made accessible by software, and if someone has a
> usecase that requires this do be done dynamically, i.e. not just
> being set up by firmware and never touched, and that use case
> is valid, then I guess we need to do something...
> 
> I think it was mentioned in the thread that there is really such
> a usecase?
> 
Actually there is no practical usecase but one but one can manufacture
it ;-)