rename semantics

[Alexander Larsson]

On Fri, 19 May 2000, Alexander Viro wrote:
> On Fri, 19 May 2000, Alexander Larsson wrote:
> > The 2.2 version will be merged into the MTD cvs tree. 
> > Sébastien Côté <scote1 at matrox.com> is working on it. The changes are
> > sufficiently small to warrant #ifdef's. 
> 
> Notice that 2.3 and 2.2 versions will seriously diverge in interp.c too -
> current code maintains the heck of redundant stuff that already belongs
> to dcache. And I'm not too sure that cleanups in that area will be portable
> to 2.2

Ahh, I see. I didn't think about that, but it would be much better to use
the dcache for the in-memory representation of the filesystem. That change
will surely make a lot of changes in interp.c, but it will mean less code,
use less memory, and will (i think) mean that we will not need to keep the
entire jffs meta-data in memory (parts of it is cached, the rest read on
demand into the dcache from the journal).

That will be a good change, and with the threaded dcache (and other
2.2->2.3 dcache changes) it might mean that 2.2 compatibility will be
hard.

Anyway, I will try to get a well-working version of the current code
before i do any dcache stuff. That version will probably support 2.2 and
2.3/4, and the interp.c changes are small (mainly bugfixes and MTD
support) so that the old inode.c with some small MTD changes could
probably be used for 2.0 support.

/ Alex




To unsubscribe, send "unsubscribe mtd" to majordomo at infradead.org