[U-Boot] [PATCH 1/3] nand: Add support for unlock.invert

[Scott Wood <scottwood@freescale.com>]

On 08/22/2012 03:34 PM, Joe Hershberger wrote:
> Hi Scott,
> 
> On Fri, Aug 17, 2012 at 3:53 PM, Scott Wood <scottwood@freescale.com> wrote:
>> On 08/17/2012 03:31 PM, Joe Hershberger wrote:
>>> NAND unlock command allows an invert bit to be set to unlock all but
>>> the selected page range.
>>>
>>> Signed-off-by: Joe Hershberger <joe.hershberger@ni.com>
>>> ---
>>>  common/cmd_nand.c            | 13 ++++++++++---
>>>  drivers/mtd/nand/nand_util.c |  9 ++++++---
>>>  include/nand.h               |  4 ++--
>>>  3 files changed, 18 insertions(+), 8 deletions(-)
>>>
>>> @@ -368,6 +369,8 @@ int nand_unlock(struct mtd_info *mtd, ulong start, ulong length)
>>>
>>>       /* submit ADDRESS of LAST page to unlock */
>>>       page += (int)(length >> chip->page_shift);
>>> +     if (invert)
>>> +             page |= 1;
>>>       chip->cmdfunc(mtd, NAND_CMD_UNLOCK2, -1, page & chip->pagemask);
>>
>> Why |= 1?  Is this some magic that the chip recognizes to implement
>> "invert"?  Do all chips that support lock/unlock support this (none of
>> the NAND chip manuals I have document lock/unlock at all as far as I
>> could find)?  What if you want to unlock a non-inverted range that ends
>> in a page with the low bit set?
> 
> According to the data sheet for the part I'm working with
> (MT29F4G08ABADAH4) the unlock command has 2 commands... 0x23 and 0x24.
>  The invert bit only exists for 0x24 (NAND_CMD_UNLOCK2).  The format
> of the unlock commands specifys that block addresses are used,
> therefore the LSb would never be set.  This bit-0 is defined to be
> "invert area" for 0x24 and always LOW for 0x23.

Please add a code comment to this effect.

-Scott