Hello Tom,
On 08.11.23 17:11, Tom Rini wrote:
On Wed, Nov 08, 2023 at 03:53:16PM +0100, Enrico Leto wrote:
The boards were not maintained upstream since the development phase.
Restore SPL and remove unused settings. We need i2c to read the DDR settings from EEPROM, get u-boot from NAND, the console output and the watchdog.
Restore u-boot up to the user prompt. The complete functionality will be restored in future patches.
Signed-off-by: Enrico Leto enrico.leto@siemens.com
I'm glad to see this series and in general it looks fine and is a good start. As some parts of the future cleanup work I'd love to see a file under doc/board/siemens for the family and moving to plain text environment, both of which iot2050 do so there's some in-house experience I am hopeful you can leverage as well if needed. It would also be good to reduce the include/configs/ files to as little as possible, I believe there's a number of dead-comments there, and other defines that are only needed in a local context and can move to a specific file or other header. Thanks!
Yep, good points!
I tried this patchset on my dxr2 thuban board, and see a problem when booting from NAND.
SPL fails to load u-boot.img (wrong ecc calculation)
reverting commit:
commit 04fcd2587321: ("mtd: rawnand: omap_gpmc: Fix BCH6/16 HW based correction")
from Roger Quadros rogerq@kernel.org (as for this reason added to cc)
and SPL/U-Boot 2023.10 comes up fine with Enricos patchset!
remark: for plain U-Boot 2023.10 even SPL does not boot, so Enricos patchset is a big step in getting back a working board again, thanks!
So first, I wonder, if there is any am335x based board booting from NAND with the above commit in?
If the above commit is in U-Boot, the following patch helps also to get this board back booting: """ diff --git a/drivers/mtd/nand/raw/omap_gpmc.c b/drivers/mtd/nand/raw/omap_gpmc.c index 1a5ed0de31..d1c73b389c 100644 --- a/drivers/mtd/nand/raw/omap_gpmc.c +++ b/drivers/mtd/nand/raw/omap_gpmc.c @@ -644,7 +644,11 @@ static int omap_correct_data_bch(struct mtd_info *mtd, uint8_t *dat,
/* check calculated ecc */ for (i = 0; i < ecc->bytes && !ecc_flag; i++) { +#ifdef CONFIG_SPL_BUILD + if (calc_ecc[i] != read_ecc[i]) +#else if (calc_ecc[i] != 0x00) +#endif goto not_ecc_match; } return 0; """
but of course, this is not the solution ...
Any ideas are welcome, Thanks!
bye, Heiko
(Damn, that I have not longer automated testing up and running for this board)