[PATCH v2] mtd: nand: write bad block marker even with BBT

[Brian Norris]

On Fri, Dec 23, 2011 at 12:13 AM, Shmulik Ladkani
<shmulik.ladkani at gmail.com> wrote:
> On Wed, 21 Dec 2011 17:15:29 -0800 Brian Norris <computersforpeace at gmail.com> wrote:
>> Another question: I noticed that when writing bad block markers to
>> OOB, we simply overwrite without erasing first. This is a problem on
>> MLC, which aren't designed to be written twice; they usually can write
>> some of the bits to 0, but not all of them. Is it reasonable to change
>> nand_default_block_markbad() to always have it erase the block before
>> writing?
>
> Not an MLC expert, but could it be that for these MLC chips,
> NAND_BBT_USE_FLASH is usually set, hence the complexities invloved in
> re-programming the OOB are simply avoided?

I can't comment on others' use of MLC + NAND_BBT_USE_FLASH, but I use
NAND_BBT_USE_FLASH for all NAND.

NAND_BBT_USE_FLASH only applies to the bad block *table* location -
that it will be stored in flash. This does not exclude the possibility
of still writing/recognizing bad block markers in OOB. Thus, I do not
understand your how "the complexities invloved in re-programming the
OOB are simply avoided". In fact, I feel that (when possible) it's
best to write to *both* a flash-based BBT and the OOB region of the
bad block, so I do not plan to avoid the complexities of reprogramming
the OOB. Please correct me if I'm missing your point.

> If so, it stregthens your NAND_BBT_WRITE_BBM/NAND_NO_WRITE_OOB approach,
> which goes hand-in-hand with Sebastian's scenario where there's no room
> in the OOB.

I will proceed with v2 adding a NAND_NO_WRITE_OOB flag to provide an
option for cases like Sebastian's, where there is no writeable OOB
space whatsoever.

Perhaps the paragraph beginning with "Another question:" should be
relegated to a separate patch/thread to avoid confusion between two
related but distinct issues. I will try to write a patch soon that
implements my suggestion and tag it onto a v2 patch series.

Brian