Re: [RFC PATCH] mtd: add per NAND partition ECC config

[Boris BREZILLON]

Hi Ezequiel,

On 11/02/2014 15:01, Ezequiel Garcia wrote:
> Hi Boris,
>
> On Sat, Feb 08, 2014 at 11:26:46AM +0100, Boris BREZILLON wrote:
> > This patch aims to add per partition ECC config for NAND devices.
> > It defines a new field in the mtd struct to store the mtd ECC config and
> > thus each mtd partition device can store its config instead of using the
> > default NAND chip config.
> >
> > This feature is needed to support the sunxi boot0 paritition case:
> > Allwinner boot code (BROM) requires a specific HW ECC for its boot code
> > that may not fit the HW NAND requirements for the entire NAND chip.
> >
> > Signed-off-by: Boris BREZILLON <b.brezillon.dev@xxxxxxxxx>
> > ---
> > Hello,
> >
> > This patch is just a draft that implement per partition ECC config.
> > It's currently not properly splitted (it should be separated in several
> > patches) and not documented either.
> Ah, ah...
>
> > There's at least one point that bother me in the current implementation:
> > I introduced DT notions in the nand core code by the mean of the get_ecc_ctrl
> > callback, and so far this was kept out of mtd/nand core code (I guess it was
> > on purpose).
> >
> > Please let me know if you see other drawbacks.
> >
> > If you think per partition ECC should not be implemented, could you help me
> > find a way to handle sunxi specific case decribed above ?
> >
> >   drivers/mtd/mtdpart.c          |   23 ++-
> >   drivers/mtd/nand/nand_base.c   |  428 ++++++++++++++++++++++++----------------
> I really appreciate your effort, but 428 changed lines is a too big and
> intrusive change. I must admit I'm not smart enough to review such patches.
>
> I honestly think you'll have better luck getting feedback if you take the time
> to properly split and document this.
>
> Yeah, it's annoying and time-consuming, but it's globally cheaper for you to
> invest time on making it easier for reviewers and maintainers, than for each
> of us to invest the time deciphering this :-)

Fair enough.

Anyway, this proposal does not work.

If nobody objects to this ECC per partition concept, I'll propose 
something else soon.
And this time I'll make a proper documentation and patch separation ;-).

Best Regards,

Boris

> Just my point of view, of course.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/