[PATCH/RFC] MTD: Striping layer core

[Artem B. Bityutskiy]

On Fri, 2006-03-31 at 14:57 +0200, Jörn Engel wrote:
> Full agreement.
> 
> Only difference is that I would grind my teeth and accept a
> jffs2-specific flag in mtd, if the overall code is improved that way.
> Adding 10 lines of cruft to mtd and removing 50 lines of cruft from
> jffs2 would be a net win.
> 
> If someone smarter comes along and notices that the 10 lines of cruft
> can be removed from mtd without reintroducing 50 lines to jffs2, even
> better.  Just like you showed me to use "can flip bits" instead of
> "can mark obsolete".

Well, features could  be implemented as a separate unit in MTD. This
unit could do the following:

1. define set of features.
2. test if a particular feature is available in a particular MTD devise.

So we would just have:

#include <mtd/feature.h>

#define jffs2_can_mark_obsolete(c) \
	mtd_feature_available(c->mtd, MTD_FEATURE_CHANGE_BIT)

The "feature" module would make the right decision basing on mtd->type.

-- 
Best Regards,
Artem B. Bityutskiy,
St.-Petersburg, Russia.