Re: [PATCH 0/3] An alternative to SPI NAND

[Ezequiel Garcia <ezequiel.garcia@imgtec.com>]

Hi Qi Wang,

On 01/07/2015 11:45 PM, Qi Wang 王起 (qiwang) wrote:
> Hi Brian,
> 
> On Thu, Jan 08, 2015 at 9:03:24AM +0000, Brian Norris wrote:
>>
>> On Thu, Jan 08, 2015 at 12:47:24AM +0000, Peter Pan 潘栋 (peterpandong)
>> wrote:
>>> Documentation/devicetree/bindings/mtd/spi-nand.txt |   22 +
>>> drivers/mtd/Kconfig                                |    2 +
>>> drivers/mtd/Makefile                               |    1 +
>>> drivers/mtd/spi-nand/Kconfig                       |    7 +
>>> drivers/mtd/spi-nand/Makefile                      |    3 +
>>> drivers/mtd/spi-nand/spi-nand-base.c               | 2034
>> ++++++++++++++++++++
>>> drivers/mtd/spi-nand/spi-nand-bbt.c                | 1279 ++++++++++++
>>
>> I can already tell by the diffstat that I don't like this. We probably
>> don't need 3000 new lines of code for this, but we especially don't want
>> to duplicate nand_bbt.c. It won't take a lot of work to augment
>> nand_bbt.c to make it shareable. (I can whip that patch up if needed.)
> 
> Yes, I agree with you, Nand_bbt.c do can be shared by Parallel NAND and
> SPI NAND. Actually, we are working at this now. Will send patches to you
> Once we finished it.
> 

Thanks for the quick submission!

However, Brian is right, this code duplication is a no go.

Perhaps a more valid approach would be to first identify the code that
needs to be shared in nand_bbt.c and nand_base.c, and export those
symbols (or maybe do the required refactor).

Then, separate the SPI NAND upper and lower logic (in a similar to my
proposal, which I still consider turned out to be clean).

These two things would lead to a simpler and smaller patchset. I also
suggest to cut off everything that we don't utterly need on a first
submission, so it's easier to review.
-- 
Ezequiel