[meta-ti] [PATCH 1/2] u-boot 2011.09: convert beagleboard to SPL build

[Tom Rini]

On Mon, Dec 12, 2011 at 5:42 PM, Andreas Müller <schnitzeltony at gmx.de> wrote:
> On Thursday, December 08, 2011 03:49:47 PM you wrote:
>> On Wed, Dec 7, 2011 at 6:13 PM, Andreas Müller <schnitzeltony at gmx.de> wrote:
>> > On Sunday, November 27, 2011 04:03:37 PM Koen Kooi wrote:
>> >> Runtime tested on an xM rev C
>> >>
>> >> Signed-off-by: Koen Kooi <koen at dominion.thruhere.net>
>> >> ---
>> >>  ...-probe-DDC-bus-for-expansionboard-EEPROMS.patch |  165 ----------
>> >>  ...p3-mem-Comment-enable_gpmc_cs_config-more.patch |   56 ++++
>> >>  ...e-SDRC-dram_init-to-always-call-make_cs1_.patch |   66 ++++
>> >>  ...-a-helper-function-to-set-timings-in-SDRC.patch |  172 +++++++++++
>> >>  ...e-mem_ok-to-clear-again-after-reading-bac.patch |   30 ++
>> >>  .../0005-OMAP3-Remove-get_mem_type-prototype.patch |   28 ++
>> >>  .../0006-omap3-mem-Add-MCFG-helper-macro.patch     |   97 ++++++
>> >>  ...d-optimal-SDRC-autorefresh-control-values.patch |   59 ++++
>> >>  ...x-all-Micron-memory-timing-parts-with-the.patch |   55 ++++
>> >>  ...ework-memory-initalization-and-devkit8000.patch |  174 +++++++++++
>> >>  ...OMAP3-SPL-Add-identify_nand_chip-function.patch |  142 +++++++++
>> >>  ...0011-OMAP3-Add-SPL-support-to-Beagleboard.patch |  315
>> >
>> > Is anybody working on SPL support for gumstix overo?
>>
>> No, but in mainline U-Boot it should be very easy to do so and I'd be
>> glad to help out with questions on the SPL specific side of things
>> (basically overo needs to provide a function that returns the correct
>> timings for the SDRC for the board rev in question).
> Got overo board revisions 0 & 1 booting fine from SDcard ( revision 2 might need
> further alignment.. ) [1].
>
> To prevent modifying multiple locations I left
>
> CONFIG_SYS_TEXT_BASE           0x80080000

This is part of your problem, actually.  You need to move it up to
0x8100'0000, but you can for everything (well, except for NOR, but
that's not SPL :)).

> To test NAND I did on u-boot console ( based on [2] )
>
> | nand erase.chip
> | mmc rescan 0
>
> | # write MLO
> | fatload mmc 0 $loadaddr MLO
> | nandecc hw
> | nand write $loadaddr 0 ${filesize}
>
> | # write u-boot
> | fatload mmc 0 ${loadaddr} u-boot.bin
> | nand write ${loadaddr} 80000 ${filesize}

Two problems, needs to be in 'nandecc hw' mode too and should be
u-boot.img to avoid the problem you get on "signature not found"

> | Boot SPL 2011.09-01172-gfdbe8b9-dirty (Dec 13 2011 - 00:58:36)
> | Texas Instruments Revision detection unimplemented
> Is it correct that this is only implemented for OMAP4/5 so for OMAP3 this
> message is OK?

So, this is omap_rev_parse, or so (check
arch/arm/cpu/armv7/omap-common/spl.c for the exact name) and it's a
weak function to be provided by boards.  On OMAP4/5/am33xx they can
tell at run-time (am33xx reads an i2c eeprom, dunno about omap4/5) but
on omap3 we have this per-board.

Oh, and FWIW feel free to bring this up on the u-boot ML if you want,
we're getting a little off-topic for meta-ti but that's OK :)

-- 
Tom