Re: [PATCH] serial: change remove NR_IRQS in 8250.c

[Eric W. Biederman]

Alan Cox <alan@xxxxxxxxxxxxxxxxxxx> writes:

> On Thu, 31 Jul 2008 04:50:21 -0700
> ebiederm@xxxxxxxxxxxx (Eric W. Biederman) wrote:
>
>> > replace
>> > 	[PATCH] serial: change irq_lists to use dyn_array
>> > use small array with index to handle irq locking for serial port
>> > hope 32 slot is enough
>> 
>> Could you size this array by NR_UARTS (our worst case usage)
>> and place irq_no in struct irq_info?
>
> NR_UARTS is likely to go away in time so don't get attached to it

I'm not attached to it, but NR_UARTS is a closer approximation to what
we are trying to do then YH's hard coded 32.  Do you know if we
actually need the list of uarts per irq or if request_irq having a
shared isa would work?

The practical question is how do we cleanly kill the array
irq_lists[NR_IRQS].  YH's hack where he isn't paying attention to what
the code is doing and just trying to avoid the problem is not
something I am fond of seeing being merged.

It is also true that sorting out 8250.c and by extension the other
serial drivers that have cloned it is the significant non-arch
piece of work needed to kill NR_IRQs

Eric


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/