[PATCH 1/2] mtd: nand: add OOB argument to NAND {read, write}_page interfaces

[Bastian Hecht]

Hi Brian,

2012/4/18 Brian Norris <computersforpeace at gmail.com>:
> Hi Bastian,
>
> On 4/18/2012 2:37 AM, Bastian Hecht wrote:
>>
>> 2012/4/18 Bastian Hecht<hechtb at googlemail.com>:
>>
>>> I'm currently working on the hardware ECC part of the SH Mobile flctl
>>> and I like the patchset as it makes things more explicit when OOB data
>>> is requested or not and such things. It's cleaner and better, thanks!
>
>
> Glad it will help! It's good to know other drivers will see a benefit.
>
>
>> To be a bit more of a concrete help, if you can give me a use case to
>> test the oob reads/writes I can modify my driver to work with your
>> patches. Right now I rely on nandwrite, nanddump, nandtest, the kernel
>> test modules and ubi. With none of them I produced an error so far
>> although the current implementation of the hardware ECC completely
>> ignores OOB data other than ECC.
>
>
> Well, for me the big "use case" is that I can support my new DMA controller
> (that doesn't read/write OOB) without the rest of MTD/NAND choking on
> anything or returning junk data. So I mostly make sure that
> nanddump/nandwrite work both with and without OOB (the -o flag) or ECC (the
> -n flag); that the kernel test modules work; and that UBI/UBIFS work on top
> of MTD.

Ah very nice! You made me look at the nandwrite -o thing once again. I
used it before without success, there was no write at all when I
issued it, but now ( - I compiled the git version of mtd-utils and
replaced the debian prehistoric prepackaged one with it - ) it works!
Now I can finish my work on the flctl ECC part. Thanks!

> I don't know if there are any more specific tests you would need to run; are
> you looking for something else?

Nope, thanks.

cheers,

 Bastian Hecht


> Brian