[PATCH 00/12] Marvell NAND controller rework with ->exec_op()

[Boris Brezillon]

On Sun, 7 Jan 2018 22:19:11 +0100
Boris Brezillon <boris.brezillon at free-electrons.com> wrote:

> On Sun, 7 Jan 2018 22:09:11 +0100
> Miquel RAYNAL <miquel.raynal at free-electrons.com> wrote:
> 
> > Hi Robert,
> > 
> > On Sun, 07 Jan 2018 21:55:33 +0100
> > Robert Jarzmik <robert.jarzmik at free.fr> wrote:
> >   
> > > Boris Brezillon <boris.brezillon at free-electrons.com> writes:
> > >     
> > > > On Wed, 3 Jan 2018 21:10:28 +0100
> > > > Boris Brezillon <boris.brezillon at free-electrons.com> wrote:      
> > > >> Hm, that's weird. Can you try with the old driver (pxa3xx)?      
> > > Ah you're right, my NAND was damaged ...
> > >     
> > > > Alternatively, you can type 'nand bad' from uboot to check if it
> > > > detects the same bad blocks.      
> > > Mmmh no, the SPL is barebox in my case. Do you have a command in
> > > linux or barebox to do the same thing ?    
> > 
> > Not sure, but nand -i should do the trick.
> > https://www.barebox.org/doc/latest/commands/hwmanip/nand.html
> > 
> > But I am not sure this is still relevant now we know the NAND was
> > damaged by the previous experiments (sorry about that). Can you put the
> > NAND in a clean state and report us if it is still failing?  
> 
> In order to do that you'll have to scrub the blocks storing the BBT, and
> I'm not sure barebox supports that. Linux does not, for sure, so if you
> want to forcibly erase bad blocks from linux, you'll have to comment
> these lines [1].

Apparently there's a "nand -g <offs>" command to mark a block good in
barebox. After doing that you should be able to erase the blocks storing
the BBT (it should be placed in the last 2 blocks of the NAND).

> 
> [1]http://elixir.free-electrons.com/linux/v4.15-rc6/source/drivers/mtd/nand/nand_base.c#L3056