[PATCH] mtd: nand: ignore ecc errors during bbt reads

[Shmulik Ladkani]

Hi Mike,

On Thu,  7 Jun 2012 11:32:08 -0700 Mike Dunn <mikedunn at newsguy.com> wrote:
> Ignore ecc errors in the scan_read_raw_oob() function.  Also removed code that
> is now redundant.
> 
> Signed-off-by: Mike Dunn <mikedunn at newsguy.com>
> ---
>  drivers/mtd/nand/nand_bbt.c |    6 +++---
>  1 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/mtd/nand/nand_bbt.c b/drivers/mtd/nand/nand_bbt.c
> index 30d1319..585da44 100644
> --- a/drivers/mtd/nand/nand_bbt.c
> +++ b/drivers/mtd/nand/nand_bbt.c
> @@ -319,7 +319,8 @@ static int scan_read_raw_oob(struct mtd_info *mtd, uint8_t *buf, loff_t offs,
>  
>  		res = mtd_read_oob(mtd, offs, &ops);
>  
> -		if (res)
> +		/* Ignore ECC errors when checking for BBM */
> +		if (res && !mtd_is_bitflip_or_eccerr(res))
>  			return res;

IMO this is not necessary.
Note the 'ops.mode' is initialized to MTD_OPS_RAW;  meaning, no ECC is
performed during read ('ecc.read_page_raw' will be invoked).
Thus, EUCLEAN/EBADMSG should not be reported.
Am I missing something here?

> @@ -406,8 +407,7 @@ static int scan_block_full(struct mtd_info *mtd, struct nand_bbt_descr *bd,
>  	int ret, j;
>  
>  	ret = scan_read_raw_oob(mtd, buf, offs, readlen);
> -	/* Ignore ECC errors when checking for BBM */
> -	if (ret && !mtd_is_bitflip_or_eccerr(ret))
> +	if (ret)
>  		return ret;

Don't understand why 'mtd_is_bitflip_or_eccerr' is originally tested.
As noted above, 'scan_read_raw_oob' won't return EUCLEAN/EBADMSG, as it
uses MTD_OPS_RAW.

And what's even stranger is the code in 'scan_block_fast()', quote:

	ops.datbuf = NULL;
	ops.mode = MTD_OPS_PLACE_OOB;

	for (j = 0; j < len; j++) {
		ret = mtd_read_oob(mtd, offs, &ops);
		/* Ignore ECC errors when checking for BBM */
		if (ret && !mtd_is_bitflip_or_eccerr(ret))
			return ret;

Here, 'MTD_OPS_PLACE_OOB' is explicitly used - so *theoretically*
'mtd_read_oob' might return EUCLEAN/EBADMSG, see 'nand_do_read_oob',
(although NO driver YET reports maxbitflips from its 'chip->ecc.read_oob'
method).

Why not use 'MTD_OPS_RAW' in 'scan_block_fast()' as well (as done in
'scan_block_full')?

Regards,
Shmulik