Re: [Ltib] patch to mtd-utils to dump ECC layout for a NAND MTD device

[Peter Barada]

On 07/08/2011 06:44 PM, Fritz Mueller wrote:
> Hey folks,
> > Peter Barada wrote:
> >> Adding in this code made life so much easier trying to hunt down 
> and fix
> >> a but in YAFFS/NAND ECC compatibility between Linux and u-boot on my
> >> target....
> I've run up against a similar problem to the one Peter mentions that 
> motivated this patch (in my case, JFFS2/NAND ECC compatibility issue 
> between Linux and u-boot on a Freescale MPC8308 based target.)
> I've already taken a look at the NAND driver code being used by both 
> u-boot and Linux, and at least by inspection both seem to be using the 
> same ECC setup/strategy.  I was wondering if Peter could comment here 
> a little more on the nature of the problem he experienced and the 
> eventual solution?
The problem I had was integrating support for a Micron NAND OMAP PoP 
device where the NAND has an internal 4-bit ECC engine that dictates the 
layout of the OOB/ECC bytes.

However the bootrom internal to the OMAP dictates teh first block use 
Hamming 1-bit ECCs so it can read the  2nd stage bootloader (x-loader) 
into SRAM.  The kernel I was buillding ran on two differnt OMAP 
assemblies, one with the Micron NAND and the other used a Hynix (for 
industrial temperatures) that used a tradtiional 1-bit Hamming ECC.

Adding the code into the MTD utilities was a quick/easy way to prove at 
userspace that the ECC layout is indeed correct for each partition 
defined on each NAND part.

--
Peter Barada
address@hidden