[PATCH] mtd: fsl_elbc_nand: set NAND_NO_SUBPAGE_WRITE flag if using hardware ECC

[PATCH] mtd: fsl_elbc_nand: set NAND_NO_SUBPAGE_WRITE flag if using hardware ECC

[Nikita Yushchenko]

Per MPC8572E manual, sec 14.4.3.1.3: "Transfers shorter than a full page,
however, require software to prepare the appropriate ECC in the spare
region"

Need to set NAND_NO_SUBPAGE_WRITE flag. If this is not done, then
generic nand_write_subpage_hwecc() is called instead of driver's
write routine, which results into OOPS dereferencing NULL chip->ecc.hwctl
pointer.

Signed-off-by: Nikita Yushchenko <nyushchenko@dev.rtsoft.ru>
---
 drivers/mtd/nand/fsl_elbc_nand.c |    2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/mtd/nand/fsl_elbc_nand.c b/drivers/mtd/nand/fsl_elbc_nand.c
index ec549cd..bb146a0 100644
--- a/drivers/mtd/nand/fsl_elbc_nand.c
+++ b/drivers/mtd/nand/fsl_elbc_nand.c
@@ -766,6 +766,8 @@ static int fsl_elbc_chip_init(struct fsl_elbc_mtd *priv)
 	if ((in_be32(&lbc->bank[priv->bank].br) & BR_DECC) ==
 	    BR_DECC_CHK_GEN) {
 		chip->ecc.mode = NAND_ECC_HW;
+		/* hardware can't calculate ECC for subpages */
+		chip->options |= NAND_NO_SUBPAGE_WRITE;
 		/* put in small page settings and adjust later if needed */
 		chip->ecc.layout = (priv->fmr & FMR_ECCM) ?
 				&fsl_elbc_oob_sp_eccm1 : &fsl_elbc_oob_sp_eccm0;
-- 
1.7.10.4

RE: [PATCH] mtd: fsl_elbc_nand: set NAND_NO_SUBPAGE_WRITE flag if using hardware ECC

[Gupta, Pekon]

Hi Nikita,

>From: Nikita Yushchenko
>
>Per MPC8572E manual, sec 14.4.3.1.3: "Transfers shorter than a full page,
>however, require software to prepare the appropriate ECC in the spare
>region"
>
>Need to set NAND_NO_SUBPAGE_WRITE flag. If this is not done, then
>generic nand_write_subpage_hwecc() is called instead of driver's
>write routine, which results into OOPS dereferencing NULL chip->ecc.hwctl
>pointer.
>
>Signed-off-by: Nikita Yushchenko <nyushchenko@dev.rtsoft.ru>
>---

This one is already solved.

Please refer below discussion and its previous thread on
why NAND_NO_SUBPAGE_WRITE should _not_ be set,
because it breaks compatibility with older UBIFS images.

http://lists.infradead.org/pipermail/linux-mtd/2014-April/053262.html

Also below is the appropriate fix without breaking backward compatibility.
http://lists.infradead.org/pipermail/linux-mtd/2014-May/053745.html


with regards, pekon