Re: Pooled memory allocation for JFFS2

[David Woodhouse <dwmw2 at infradead dot org>]

On Fri, 2003-11-14 at 16:57 +0100, Andrew Lunn wrote:
> We aim to not make any changes to files which are part of the normal
> jffs2 distribution. That way we can simply import the latest jffs2
> when ever it changes.

Except I probably broke it with the recent NAND changes. Shouldn't be
hard to catch up though.

> Turning off compression required changes to the standard jffs2
> files. Such changes you need David to commit to the jffs2 main tree
> first (or at least agree to them and i think Gary can then commit
> them) and then we can import them into eCos.

I've no real objection to this; I'd like to see us sanely handle the
case where we encounter a node with an unknown compression though. At
the moment I'm not entirely convinced we'll do the right thing.

> Your pool changes only touch one file in the normal jffs2 distribution
> and that change could be put somewhere else. 

We already have 'jffs2_init_slab_caches()' which does precisely the same
task for the Linux code. You could use that, perhaps? 

As an added bonus, that way you'll get it done once only, instead of
once per file system.


-- 
dwmw2