Re: [PATCH 0/8] Implement a new DTS Source Language

[David Gibson <david-xT8FGy+AXnRB3Ne2BGzF6laj5H9X9Tb+@public.gmane.org>]

On Wed, Sep 24, 2008 at 01:48:55PM -0500, Jon Loeliger wrote:
> On Wed, 2008-09-24 at 11:51 -0500, Jon Loeliger wrote:
> > On Tue, 2008-09-23 at 21:34 -0500, Kumar Gala wrote:
> > 
> > > Any examples/tests of all the cool new features?
> > > 
> > > - k
> > 
> > Here is a QAD example based on the MPC8641 DTS file
> > as found in the arch/powerpc/boot/dts directory today.
> > It is maybe not the best example, but it shows the
> > flavor of what can be done.
> > 
> > Enjoy,
> > jdl
> 
> 
> Also, here is the same refactoring done on the
> mpc8572ds.dts file too.
> 
> And, perhaps more important than doing this work
> on this one file, it shows that the first section
> of both this 8572 DTS file and the 8641 DTS file
> are almost identical throughout the /define/ section.
> Minor refactoring needed in the 8641 variant.
> The only "real" difference is 8641 says "TSEC" and
> the 8572 says "eTSEC", and one weird interrupt thing
> in the serial node.  Dunno.

[snip]
> /define/ make_i2c(\id)
> {
> 	\addr := 0x3000 + \id * 0x100;

Hrm.  Again, not very C-like.  Plus it bothers me that there are 3
somewhat different definition constructs: /define/, /const/ and :=.
At least /define/ and /const/ should be mergeable, with /const/
becoming a zero-parameter /define/.  Maybe local defines need
something different, but the distinction seems dangerously ad-hoc to
me at the moment.

[snip]
> /define/ make_dma_channel(\part, \id, \intr_base)
> {
> 	"dma-channel@" % hexstr(\id * 0x80) {
> 		compatible = < ("fsl,mpc" % \part % "-dma-channel")
> 				("fsl,eloplus-dma-channel") >;

WTF!?  What is a < > construct doing containing strings rather than
cells.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson