flash bbt broken due to unitialized bitflip_threshold?

[Mike Dunn]

On 06/22/2012 01:39 PM, Brian Norris wrote:
> On Sun, Jun 10, 2012 at 12:08 AM, Shmulik Ladkani
> <shmulik.ladkani at gmail.com> wrote:
>> To summarize, return value of the 'mtd_read_oob()' API is
>> inconsistent (sometimes max_bitflips, other times EUCLEAN), and does not
>> adhere to return code policy of 'mtd_read()' - return EUCLEAN to the users,
>> they're not interested with internals such as max_bitflips.
>>
>> This might affect users of 'mtd_read_oob()' which might falsely think
>> the OOB read has failed.
>>
>> Mike, care to take a look and amend if necessary?
>> I think this needs to be fixed pre v3.5 as well.
> 
> Mike, are you planning on handling this?
> 
> I think it would be safe to basically C&P the max_bitflips code from
> mtd_read() into mtd_read_oob(). This would simply pass through the
> EUCLEAN that may be returned for OOB-only case (only for my driver?),
> and it would translate the 'ret_code > 0' case properly for data+OOB
> reads. I'll resend officially if this looks OK. Or I'll defer to Mike
> if he's working on this.


I'm sorry Brian, I missed Shmulik's post and thought this issue was put to bed
with the patch that initialized bitflip_threshold before the nand_scan_tail()
call.

I will take a look at your patch set that addresses this.  Sorry for dropping
the ball.

Mike