MTD concat layer

Robert Kaiser rob at sysgo.de
Wed Feb 13 06:34:34 EST 2002


On Wed, 13 Feb 2002, David Woodhouse wrote:

> 
> joakim.tjernlund at lumentis.se said:
> >  I thougth this was already supported? Just create a mtd partition
> > that will span two or more banks and mount a JFFS2 FS on top of that
> > partition. 
> 
> Doesn't work if the banks aren't made to appear contiguous by the map 
> driver. The probe code can't handle it. 

Yes. My initial plan _was_ to hide that stuff in the mapping driver
but you (David) persuaded me to look for a more general solution.

BTW, the concat layer even allows to concatenate different kinds of
chips (as long as they are of the same type, e.g. Flash).

> 
> The case in question has something like:
> 
>  --------------
>  |  CFI chip  |
>  --------------
>  |  BIOS chip |
>  --------------
>  |  CFI chip  |
>  --------------
> 
> You could hack up a map driver that appears to have the two CFI chips 
> contiguously, and then the probe code would find them both and make a 
> single MTD device from them -  but there was a reason why he wasn't doing 
> this. Perhaps the 'BIOS' bit is in fact the beginning of the second chip, 
> not actually a separate chip at all? I don't remember.

No, the BIOS is in the middle of the chip's addressing space. All this
_could_ conceivably be solved with a hacked-up map driver, but then again,
one might as well do the partitioning stuff in a hacked-up map driver. The
idea was to have this functionality available in a generic form, so it
it is accessible as a service from the MTD core layer.

Rob

----------------------------------------------------------------
Robert Kaiser                          email: rkaiser at sysgo.de
SYSGO RTS GmbH
Am Pfaffenstein 14
D-55270 Klein-Winternheim / Germany    fax:   (49) 6136 9948-10






More information about the linux-mtd mailing list