[uCsimm] Flash FS

From: Ǽ (kwonsk@mutech.co.kr)
Date: Fri Apr 07 2000 - 19:24:14 EDT

Hi, all

Erwin Authried wrote:

>Hello Kwonsk!
>What's the status of your Flash FS implementation?
>Have you had a look at the Journaling FS implementation
>from the Axis site? I'd like to ask you if you could send
>me the modified modules for mtd and the flash drivers
>for uClinux.

flash drivers for uClinux was improved a little for cpu utilization.
instead of silly busy wait it uses 'check and reschedule if erase
not done' method. i will send this drivers if someone need it.

A problem about FTL copyright was not solved. you could not
use FTL code on the raw flash chip. it's a bad.
I've followed Axis's JFFS and found that it has wear leveling
effect. but it's a side effect. It means JFFS do not sure wear leveling
effect will occur always. and if a small is resides on sector boundary of
a flash, then these two sectors could not used until this file
is deleted (fragmentation problem).
I know that JFFS would be good for almost people (about 99 percent?)
and it free. If you do not change files frequently, then you might
choose JFFS for your flash file system. not FTL on mtd.
(again, FTL on direct flash was not allowd by M-systems who has
copyright about FTL and producer of DiskOnChip)

Nowadays i'm working with VSB(Variable Small Block) from Intel.
I've ported this for uClinux and got success. I could build linux file
system on VSB layer on direct flash chip now.
VSB has a copyright problem also. They did not allow
VSB code for another flash chip (like amd flash). and did not allow
source code distribution, just allow binary only driver.
it just free for intel flash memory.

Free flash usage for linux,uClinux is unclear yet. we must make
new implementations and new scheme for free use. Need your help.


This message resent by the ucsimm@uclinux.com list server http://www.uClinux.com/

This archive was generated by hypermail 2b30 : Sun Apr 07 2002 - 00:01:35 EST