QT2410 changes to load u-boot through JTAG

Ling.Alex at iac.com.tw Ling.Alex at iac.com.tw
Thu Jun 7 03:20:58 CEST 2007

Hi Wener & Harald,
According to board/neo1973/common/lowlevel_init.S:

	.word	TEXT_BASE

.globl lowlevel_init
	/* memory control configuration */
	/* make r0 relative the current location so that it */
	/* reads SMRDATA out of FLASH rather than memory ! */
	ldr     r0, =3DSMRDATA
	ldr	r1, _TEXT_BASE
	sub	r0, r0, r1

I think lowlevel_init is designed to be guaranteed to use SMRDATA stored =
in stepping-stone when booting from NAND or NOR flash when booting from =
NOR instead of SDRAM. Can we get rid of CONFIG_LL_INIT_NAND_ONLY by =
simplifying the logic to perform S3C24x0 memory controller =
initialization ONLY when lowlevel_init are not called from SDRAM?

Best regards,
Tel 021-64853668 ext 2434
Mail Ling.Alex at iac.com.tw
-----Original Message-----
From: openmoko-uboot-bounces at lists.openmoko.org =
[mailto:openmoko-uboot-bounces at lists.openmoko.org] On Behalf Of Werner =
Sent: Tuesday, June 05, 2007 10:49 AM
To: Harald Welte
Cc: openmoko-uboot at lists.openmoko.org
Subject: Re: QT2410 changes to load u-boot through JTAG

I wrote:
>>> 2) include/configs/qt2410.h needs CONFIG_LL_INIT_NAND_ONLY
>> why is that?
> I haven't tracked it down to where it breaks, but it seems that
> the lowlevel init doesn't want to be done twice. Maybe we can
> change that. I'll have a look.

It was due to lowlevel_init.S calculating the address of SMRDATA
with the assumption that we're running from RAM. I've not changed
this to PC-relative.

> Hmm, thinking of it, CONFIG_LL_INIT_NAND_ONLY is probably okay.

With the change above, we don't need it anymore. In fact, we
could probably remove CONFIG_LL_INIT_NAND_ONLY entirely. Shall we
give this a try now, or wait for more quiet days ?

> Good, so we have a solution ;-) I'll make a patch after I've had
> my morning coffee.

Well, it took a little longer, but it's now in start.S. I noticed
that we have quite a bit of patch noise now (had me chasing ghosts
for a good while, too), so perhaps a global refresh wouldn't be so

- Werner


 / Werner Almesberger, Buenos Aires, Argentina     =
werner at almesberger.net /

More information about the openmoko-uboot mailing list