Re: [PATCH] 2/3 ACPI resource handling

[Bjorn Helgaas]

> > +#define copy_field(out, in, field)	out->field = in->field
> > +#define copy_address(out, in)					\
> > +	copy_field(out, in, resource_type);			\
> > +	copy_field(out, in, producer_consumer);			\
> > +	copy_field(out, in, decode);				\
> > +	copy_field(out, in, min_address_fixed);			\
> > +	copy_field(out, in, max_address_fixed);			\
> > +	copy_field(out, in, attribute);				\
> > +	copy_field(out, in, granularity);			\
> > +	copy_field(out, in, min_address_range);			\
> > +	copy_field(out, in, max_address_range);			\
> > +	copy_field(out, in, address_translation_offset);	\
> > +	copy_field(out, in, address_length);			\
> > +	copy_field(out, in, resource_source);
> 
> If ACPI just used normal (ie short) variable names, then ugly macros
> like this would not be necessary.

Well, the length of variable names is really irrelevant to this patch.
I just didn't want to write

	switch (resource->id) {
	case ACPI_RSTYPE_ADDRESS16:
		address16 = (struct acpi_resource_address16 *) &resource->data;
		out->resource_type = address16->resource_type;
		out->producer_consumer = address16->producer_consumer;
		...
	case ACPI_RSTYPE_ADDRESS32:
		address32 = (struct acpi_resource_address32 *) &resource->data;
		out->resource_type = address32->resource_type;
		out->producer_consumer = address32->producer_consumer;
		...
	case ACPI_RSTYPE_ADDRESS64:
		address64 = (struct acpi_resource_address64 *) &resource->data;
		out->resource_type = address64->resource_type;
		out->producer_consumer = address64->producer_consumer;
		...

where all the cases are identical except for the 16/32/64 (and the
occasional cut-and-paste error).

If the macros are considered too ugly, I'd be glad to rewrite the
function as above.  The main thing is to put the 16/32/64 switch
ONE place, rather than duplicating it in every place that consumes
address resources.

Bjorn