[U-Boot] fsl_elbc_nand.c and nand_boot_fsl_elbc.c for large page devices

[U-Boot] fsl_elbc_nand.c and nand_boot_fsl_elbc.c for large page devices

[Ron Madrid]

So I have a few quick questions/observations.

1.  I don't see anywhere that the AL (address length, or number of address cycles) can be set to anything other than 2.  It appears to be hard coded in both files.

2.  nand_boot_fsl_elbc.c uses ECCM = 1 for large page devices, but I don't see a place where fsl_elbc_nand.c sets ECCM to 1 for similar devices.  The result, I believe, is that if I try to write a new u-boot into the large page flash device it will be written with ECCM = 0.  When I try to reset the board it will be read with ECCM = 1 (inside of nand_boot_fsl_elbc.c) resulting in an ecc error in the LTESR.

I would be more than happy to help make these changes (if necessary) but I don't know if there is already a fix in the works, or if perhaps I am wrong in these two assumptions.

Ron

[U-Boot] fsl_elbc_nand.c and nand_boot_fsl_elbc.c for large page devices

[Scott Wood]

On Tue, Oct 14, 2008 at 01:24:09PM -0700, Ron Madrid wrote:
> 1.  I don't see anywhere that the AL (address length, or number of
> address cycles) can be set to anything other than 2.  It appears to be
> hard coded in both files.

Correct.  AFAICT, setting AL to anything less than 2 is just an
optimization for small flashes.  Linux will make that optimization, but
the way NAND devices are currently probed in u-boot makes that awkward
(no opportunity to run code after size is known, but before
nand_scan_tail).

> 2.  nand_boot_fsl_elbc.c uses ECCM = 1 for large page devices, but I
> don't see a place where fsl_elbc_nand.c sets ECCM to 1 for similar
> devices.

That looks like a bug.

-Scott

[U-Boot] fsl_elbc_nand.c and nand_boot_fsl_elbc.c for large page devices

[Ron Madrid]

So is there a particular convention for which bytes should be used within the OOB area for an ECC?  I've looked for information on this but haven't been able to come up with anything.


--- On Thu, 10/16/08, Scott Wood <scottwood@freescale.com> wrote:

> From: Scott Wood <scottwood@freescale.com>
> Subject: Re: [U-Boot] fsl_elbc_nand.c and nand_boot_fsl_elbc.c for large page devices
> To: "Ron Madrid" <ron_madrid@sbcglobal.net>
> Cc: u-boot at lists.denx.de
> Date: Thursday, October 16, 2008, 12:17 PM
> On Tue, Oct 14, 2008 at 01:24:09PM -0700, Ron Madrid wrote:
> > 1.  I don't see anywhere that the AL (address
> length, or number of
> > address cycles) can be set to anything other than 2. 
> It appears to be
> > hard coded in both files.
> 
> Correct.  AFAICT, setting AL to anything less than 2 is
> just an
> optimization for small flashes.  Linux will make that
> optimization, but
> the way NAND devices are currently probed in u-boot makes
> that awkward
> (no opportunity to run code after size is known, but before
> nand_scan_tail).
> 
> > 2.  nand_boot_fsl_elbc.c uses ECCM = 1 for large page
> devices, but I
> > don't see a place where fsl_elbc_nand.c sets ECCM
> to 1 for similar
> > devices.
> 
> That looks like a bug.
> 
> -Scott

[U-Boot] fsl_elbc_nand.c and nand_boot_fsl_elbc.c for large page devices

[Scott Wood]

Ron Madrid wrote:
> So is there a particular convention for which bytes should be used
> within the OOB area for an ECC?  I've looked for information on this
> but haven't been able to come up with anything.

The 8313 manual says that when booting from NAND, it expects ECCM 0 on 
small page and ECCM 1 on large page.

BTW, can you get your e-mail client to wrap long lines?

-Scott

[U-Boot] fsl_elbc_nand.c and nand_boot_fsl_elbc.c for large page devices

[Ron Madrid]

> BTW, can you get your e-mail client to wrap long lines?

I've tried.  I think there's a problem with yahoo's web mail.

Ron

[U-Boot] fsl_elbc_nand.c and nand_boot_fsl_elbc.c for large page devices

[Wolfgang Denk]

Dear Ron Madrid,

In message <577375.34066.qm@web83501.mail.sp1.yahoo.com> you wrote:
> > BTW, can you get your e-mail client to wrap long lines?
> 
> I've tried.  I think there's a problem with yahoo's web mail.

So how about hitting "ENTER" every now and than?

Best regards,

Wolfgang Denk

-- 
DENX Software Engineering GmbH,     MD: Wolfgang Denk & Detlev Zundel
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd at denx.de
We have found all life forms in the galaxy are  capable  of  superior
development.
	-- Kirk, "The Gamesters of Triskelion", stardate 3211.7