Re: [PATCH] - Increase PNP_MAX_PORT. ACPI devices can have a lot IO resource declarations

[Thomas Renninger <trenn@suse.de>]

On Wed, 2007-07-18 at 15:33 -0600, Bjorn Helgaas wrote:
> On Wednesday 18 July 2007 02:21:14 am Thomas Renninger wrote:
> > On Tue, 2007-07-17 at 09:49 -0600, Bjorn Helgaas wrote:
> > > On Monday 16 July 2007 08:21:07 am Thomas Renninger wrote:
> > > > PNP0C02 devices normally have a lot more IO port declarations than
> > > > currently defined in PNP_MAX_PORT
> > > 
> > > Yes.
> > > 
> > > > I also wonder whether other limits like:
> > > >  #define PNP_MAX_MEM        4
> > > >  #define PNP_MAX_IRQ        2
> > > >  #define PNP_MAX_DMA        2
> > > > could get exceeded with pnpacpi?
> > > 
> > > Definitely.  I think the current limits come from the PNP ISA spec
> > > (sec 4.6).  I don't see similar limits in the PNPBIOS or ACPI specs,
> > > so ideally I think they should be dynamically allocated as you suggest.
> > > 
> > I wanted to implement the dynamic approach and used a dynamically
> > allocated array, filled up from beginning. While this is close to the
> > current implementation I thought this is the easiest sufficient way...
> > (I also only did this for io ports where most mem is wasted).
> > Now I am thinking about hotplug (e.g. if a SSDT with resources gets
> > hot-added, removed)... If a device can vanish, the array must get
> > reordered, not a really well fitting structure, a list (a pnp specific
> > set up, or from include/linux/list.h?) should be better?
> 
> I don't understand the array reordering problem.  Either a device exists
> or it doesn't.  Loading or unloading an SSDT should not change the number
> of resources for devices (except that it might add or remove an entire
> device).
Yes, you are right. I feared there could be cases were single resources
should be able to be removed.
> 
> I think for now, it would be sufficient to increase PNP_MAX_IRQ to 8
> and PNP_MAX_PORT to 32 and be done with it.  I don't think it's worth
> getting more complicated unless we dynamically allocate everything.
Ok.

Thanks,

   Thomas