[PATCH 2/2] mtd/nand: Fix tmio_nand ecc correction

[PATCH 2/2] mtd/nand: Fix tmio_nand ecc correction

[Atsushi Nemoto]

This driver may be reading 512 bytes at a times, but still calculates
256-byte sector ECC.  So the nand_correct_data() is not appropriate
for this driver.  Implement its ecc.correct function calling
__nand_correct_data() twice.

Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Acked-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Acked-by: Vimal Singh <vimalsingh@ti.com>
---
 drivers/mtd/nand/tmio_nand.c |   17 ++++++++++++++++-
 1 files changed, 16 insertions(+), 1 deletions(-)

diff --git a/drivers/mtd/nand/tmio_nand.c b/drivers/mtd/nand/tmio_nand.c
index daa6a4c..92c7334 100644
--- a/drivers/mtd/nand/tmio_nand.c
+++ b/drivers/mtd/nand/tmio_nand.c
@@ -301,6 +301,21 @@ static int tmio_nand_calculate_ecc(struct mtd_info *mtd, const u_char *dat,
 	return 0;
 }
 
+static int tmio_nand_correct_data(struct mtd_info *mtd, unsigned char *buf,
+		unsigned char *read_ecc, unsigned char *calc_ecc)
+{
+	int r0, r1;
+
+	/* assume ecc.size = 512 and ecc.bytes = 6 */
+	r0 = __nand_correct_data(buf, read_ecc, calc_ecc, 256);
+	if (r0 < 0)
+		return r0;
+	r1 = __nand_correct_data(buf + 256, read_ecc + 3, calc_ecc + 3, 256);
+	if (r1 < 0)
+		return r1;
+	return r0 + r1;
+}
+
 static int tmio_hw_init(struct platform_device *dev, struct tmio_nand *tmio)
 {
 	struct mfd_cell *cell = (struct mfd_cell *)dev->dev.platform_data;
@@ -424,7 +439,7 @@ static int tmio_probe(struct platform_device *dev)
 	nand_chip->ecc.bytes = 6;
 	nand_chip->ecc.hwctl = tmio_nand_enable_hwecc;
 	nand_chip->ecc.calculate = tmio_nand_calculate_ecc;
-	nand_chip->ecc.correct = nand_correct_data;
+	nand_chip->ecc.correct = tmio_nand_correct_data;
 
 	if (data)
 		nand_chip->badblock_pattern = data->badblock_pattern;
-- 
1.5.6.5

Re: [PATCH 2/2] mtd/nand: Fix tmio_nand ecc correction

[Ian molton]

Atsushi Nemoto wrote:

This looks fine to me, but as I said, wouldnt it be better to implement 
the 6-byte ecc in the core if theres more than one chip that can use it?

no point in duplicating code...

-Ian

> This driver may be reading 512 bytes at a times, but still calculates
> 256-byte sector ECC.  So the nand_correct_data() is not appropriate
> for this driver.  Implement its ecc.correct function calling
> __nand_correct_data() twice.
> 
> Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
> Acked-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
> Acked-by: Vimal Singh <vimalsingh@ti.com>
> ---
>  drivers/mtd/nand/tmio_nand.c |   17 ++++++++++++++++-
>  1 files changed, 16 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/mtd/nand/tmio_nand.c b/drivers/mtd/nand/tmio_nand.c
> index daa6a4c..92c7334 100644
> --- a/drivers/mtd/nand/tmio_nand.c
> +++ b/drivers/mtd/nand/tmio_nand.c
> @@ -301,6 +301,21 @@ static int tmio_nand_calculate_ecc(struct mtd_info *mtd, const u_char *dat,
>  	return 0;
>  }
>  
> +static int tmio_nand_correct_data(struct mtd_info *mtd, unsigned char *buf,
> +		unsigned char *read_ecc, unsigned char *calc_ecc)
> +{
> +	int r0, r1;
> +
> +	/* assume ecc.size = 512 and ecc.bytes = 6 */
> +	r0 = __nand_correct_data(buf, read_ecc, calc_ecc, 256);
> +	if (r0 < 0)
> +		return r0;
> +	r1 = __nand_correct_data(buf + 256, read_ecc + 3, calc_ecc + 3, 256);
> +	if (r1 < 0)
> +		return r1;
> +	return r0 + r1;
> +}
> +
>  static int tmio_hw_init(struct platform_device *dev, struct tmio_nand *tmio)
>  {
>  	struct mfd_cell *cell = (struct mfd_cell *)dev->dev.platform_data;
> @@ -424,7 +439,7 @@ static int tmio_probe(struct platform_device *dev)
>  	nand_chip->ecc.bytes = 6;
>  	nand_chip->ecc.hwctl = tmio_nand_enable_hwecc;
>  	nand_chip->ecc.calculate = tmio_nand_calculate_ecc;
> -	nand_chip->ecc.correct = nand_correct_data;
> +	nand_chip->ecc.correct = tmio_nand_correct_data;
>  
>  	if (data)
>  		nand_chip->badblock_pattern = data->badblock_pattern;

Re: [PATCH 2/2] mtd/nand: Fix tmio_nand ecc correction

[Atsushi Nemoto]

On Tue, 28 Jul 2009 18:12:40 +0100, Ian molton <spyro@f2s.com> wrote:
> This looks fine to me, but as I said, wouldnt it be better to implement 
> the 6-byte ecc in the core if theres more than one chip that can use it?
> 
> no point in duplicating code...

Yes, but I wonder how popular 6-byte ecc per 512 byte block compared
to other combinations.  For now, keep tmio_nand_correct_data in
tmio_nand driver would be reasonable.

If many (well, two might not be enough :-)) xxx_nand_correct_data do
same thing, we can refactor them easily.

---
Atsushi Nemoto