How read_oob() should work for HW_SYNDROME NAND controller?

[Masahiro Yamada]

Hi.
(CCing Intel engineers because this is related to Denali driver
and comments from them are very appreciated.)

I am tacking on Denali NAND driver rework.

My question is:
Which data should chip->ecc.read_oob() get
in case of a controller with syndrome-like ECC engine.

For example, say we have a NAND chip
with 2048 byte page + 64 byte oob.
(the picture in the right side.)

And let's say the controller's ecc.size = 1024 and ecc.bytes == 14.
I am omitting BBM to make the situation simpler.

The Denali NAND controller interleaves
payload and ECC like the picture in the left side.

  |-----------|    |-----------|
  |           |    |           |
  | Payload0  |    |           |
  |           |    |           |
  | (ecc.size |    |           |
  |  1024B)   |    | Main Page |
  |           |    |   area    |
  |-----------|    |           |
  |  ECC0     |    |   2048B   |
  | (ecc.bytes|    |           |
  |   14B)    |    |           |
  |-----------|    |           |
  |           |    |           |
  | Payload1  |    |           |
  |           |    |           |
  | (ecc.size |    |           |
  |  1024B)   |    |-----------|
  |           |    |           |
  |-----------|    |           |
  |  ECC1     |    | OOB area  |
  | (ecc.bytes|    |           |
  |   14B)    |    |   64B     |
  |-----------|    |           |
  | OOB free  |    |           |
  |    36B    |    |           |
  |-----------|    |-----------|

The controller can transfer
Main page area, OOB area, or both of them,
like the physical structure of the device
in the right-side picture.

This is different from how
Denali controller uses the page logically
(in the left-side picture).


The current read_oob() implementation in the Denali driver
simply get the data in the physical OOB area.
It corresponds the tail of Payload1 + ECC1 + OOB free.


I am afraid the behavior is different from
the reference implementation of
nand_read_page_raw_syndrome()
nand_read_oob_syndrome()


I think we should collect ECC sections
into oob_poi to get along with the ooblayout APIs.


The Denali IP also supports lowlevel
command-base interface to issue NAND_CMD_RNDOUT
and cherry-pick ECC sections.

But, more simply, I can transfer the whole page + oob
into a temporary buffer, then only copy
ECC sections into oob_poi.



Comments are very appreciated.


-- 
Best Regards
Masahiro Yamada