Skip to contentSkip to conversation

NAND ECC mixup?

U-Boot Gary Thomas 1 participant 2 messages Apr 16, 2010

lore
Share & tools
lore
#0 lore raw

I'm having trouble with NAND ECC on my OMAP/3530 platform. It seems that U-Boot and Linux don't agree on the ECC patterns for a given set of data.

In particular, if I create the U-Boot environment in NAND, then update it via fw_setconfig in Linux, U-Boot is no longer happy with the ECC bits.

I noticed that the file drivers/mtd/nand/nand_ecc.c has changed dramatically in the Linux sources. Could this be the problem?

My code is based on git://git.denx.de/u-boot.git
15fba3279b56333bdb65ead366f82c945ed320d1 ppc/85xx: Disable all async interrupt sources when we boot
My Linux kernel is 2.6.31 or newer

I looked in the latest tree and didn't see anything relevant. Am I missing something?

Thanks

Signature · 4 lines
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------
#1 lore raw
Quoted 11 lines · Gary Thomas

I'm having trouble with NAND ECC on my OMAP/3530 platform. It seems that U-Boot and Linux don't agree on the ECC patterns for a given set of data.

In particular, if I create the U-Boot environment in NAND, then update it via fw_setconfig in Linux, U-Boot is no longer happy with the ECC bits.

I noticed that the file drivers/mtd/nand/nand_ecc.c has changed dramatically in the Linux sources. Could this be the problem?

Quoted 6 lines

My code is based on git://git.denx.de/u-boot.git

    15fba3279b56333bdb65ead366f82c945ed320d1 ppc/85xx: Disable all async interrupt sources when we boot

My Linux kernel is 2.6.31 or newer

I looked in the latest tree and didn't see anything relevant. Am I missing something?

This turns out to a misconfiguration of 'CONFIG_MTD_NAND_ECC_SMC' My Linux configuration had it, U-Boot did not.

That said, CONFIG_MTD_NAND_ECC_SMC is incompatible with the version of X-Loader I'm running, so it's just a matter of disabling this in Linux.

Thanks

Signature · 4 lines
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------