[RFC] AI support (13/14 ssb add AI support)

George Kashperko george at znau.edu.ua
Fri Feb 18 23:07:23 UTC 2011


Hi,

> On Fri, 2011-02-18 at 00:31 +0200, George Kashperko wrote: 
> > +#define SSB_AI_BASE			0x18100000U /* base for AI registers */
> 
> #define MOON_EARTH_BASE 1337
> 
> That probably is my major point against this patchset. It just
> merges things that don't belong to each other.
> Is this the base on moon or earth?
> If it's the earth base, why the hell do we prefix the define with MOON?
> 
> That naming confusion continues throughout the whole patchset.
> 
> SSB is end of life, so we should not try to support the new AI hardware
> by making SSB AI aware. Just fork SSB and let legacy SSB die as the
> existing hardware dies. It deserves it. Mercy killing for software!
> 
> _Currently_ it may look like a good idea to use one software base
> for the two busses, but the cruft will hurt you in the future.
> And the future begins one day after this hack was merged.
> 
Well, if that was year ago I would agree with you. Now I doubt it.
Starting from the clean page means another months of no support for
AI-based hw. Going cp ssb bcmai; sed s/ssb/bcmai way will make whole
messup. Making AI over SSB causing confusion.
Either way here for me looks wrong, but among them I would better choose
on going 1+3 rather just 1st one alone.
Ah, well, there might is another 4th way to go if Broadcom will finally
support their bus with their brand code but honestly (no offence to
Broadcom here, just my personal subjective opinion) I dont think I will
be happy to see this code after all that I've already seen.

Clean page is the best. The best is the enemy for the good.

Have nice day,
George





More information about the devel mailing list