Currently, the flash-based BBT implementation writes bad block data only
to its flash-based table and not to the OOB marker area. Then, as new bad
blocks are marked over time, the OOB markers become incomplete and the
flash-based table becomes the only source of current bad block
information. This becomes an obvious problem when, for example:

 * bootloader cannot read the flash-based BBT format
 * BBT is corrupted and the flash must be rescanned for bad
   blocks; we want to remember bad blocks that were marked from Linux

So to keep the bad block markers in sync with the flash-based BBT, this
patch changes the default so that we write bad block markers to the proper
OOB area on each block in addition to flash-based BBT. Comments are
updated, expanded, and/or relocated as necessary.

The new flash-based BBT procedure for marking bad blocks:
 (1) erase the affected block, to allow OOB marker to be written cleanly
 (2) update in-memory BBT
 (3) write bad block marker to OOB area of affected block
 (4) update flash-based BBT
Note that we retain the first error encountered in (3) or (4), finish the
procedures, and dump the error in the end.

This should handle power cuts gracefully enough. (1) and (2) are mostly
harmless (note that (1) will not erase an already-recognized bad block).
The OOB and BBT may be "out of sync" if we experience power loss bewteen
(3) and (4), but we can reasonably expect that on next boot, subsequent
I/O operations will discover that the block should be marked bad again,
thus re-syncing the OOB and BBT.

Note that this is a change from the previous default flash-based BBT
behavior. If your system cannot support writing bad block markers to OOB,
use the new NAND_BBT_NO_OOB_BBM option (in combination with
NAND_BBT_USE_FLASH and NAND_BBT_NO_OOB).
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

This patch renames all MTD functions by adding a "_" prefix:

mtd->erase -> mtd->_erase
mtd->read_oob -> mtd->_read_oob
...

The reason is that we are re-working the MTD API and from now on it is
an error to use MTD function pointers directly - we have a corresponding
API call for every pointer. By adding a leading "_" we achieve the following:

1. Make sure we convert every direct pointer users
2. A leading "_" suggests that this interface is internal and it becomes
   less likely that people will use them directly
3. Make sure all the out-of-tree modules stop compiling and the owners
   spot the big API change and amend them.
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

As nand_default_block_markbad() is becoming more complex, it helps to
have code appear only in its relevant codepath(s). Here, the calculation
of `ofs' based on NAND_BBT_SCANLASTPAGE is only useful on paths where we
write bad block markers to OOB. We move the condition/calculation closer
to the `write' operation and update the comment to more correctly
describe the operation.

The variable `wr_ofs' is also used to help isolate our calculation of
the "write" offset from the usage of `ofs' to represent the eraseblock
offset. This will become useful when we reorder operations in the next
patch.

This patch should make no functional change.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

It seems that we have developed a bad-block-marking "feature" out of
pure laziness:

  "We write two bytes per location, so we dont have to mess with 16 bit
  access."

It's relatively simple to write a 1 byte at a time on x8 devices and 2
bytes at a time on x16 devices, so let's do it.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

nand_block_bad() doesn't check the correct pages when
NAND_BBT_SCAN2NDPAGE is enabled. It should scan both the OOB region of
both the 1st and 2nd page of each block.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Many NAND flash systems (especially those with MLC NAND) cannot be
reliably written twice in a row. For instance, when marking a bad block,
the block may already have data written to it, and so we should attempt
to erase the block before writing a bad block marker to its OOB region.

We can ignore erase failures, since the block may be bad such that it
cannot be erased properly; we still attempt to write zeros to its spare
area.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Funny one :) "Heck" fits somehow, too, but I am sure it was meant to be "Check".
Signed-off-by: NWolfram Sang <w.sang@pengutronix.de>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@linux.intel.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@linux.intel.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Our `ops' information was converted to a local variable recently, and
apparently, old code relied on the fact that the global version was
often left in a valid mode. We can't make this assumption on local
structs, and we shouldn't be relying on a previous state anyway.

Instead, we initialize mode to 0 for don't-care situations (i.e., the
operation does not use OOB anyway) and MTD_OPS_PLACE_OOB when we want to
place OOB data.

This fixes a bug with nand_default_block_markbad(), where we catch on
the BUG() call in nand_fill_oob():

Kernel bug detected[#1]:
...
Call Trace:
[<80307350>] nand_fill_oob.clone.5+0xa4/0x15c
[<803075d8>] nand_do_write_oob+0x1d0/0x260
[<803077c4>] nand_default_block_markbad+0x15c/0x1a8
[<802e8c2c>] part_block_markbad+0x80/0x98
[<802ebc74>] mtd_ioctl+0x6d8/0xbd0
[<802ec1a4>] mtd_unlocked_ioctl+0x38/0x5c
[<800d9c60>] do_vfs_ioctl+0xa4/0x6e4
[<800da2e4>] sys_ioctl+0x44/0xa0
[<8001381c>] stack_done+0x20/0x40
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

In rare cases, we are given an unaligned parameter `from' in
`nand_do_read_ops()'. In such cases, we use the page cache
(chip->buffers->databuf) as an intermediate buffer before dumping to the
client buffer. However, there are also cases where this buffer is not
cleanly reusable. In those cases, we need to make sure that we
explicitly invalidate the cache.

This patch prevents accidental reusage of the page cache, and for me,
this solves some problems I come across when reading a corrupted BBT
from flash (NAND_BBT_USE_FLASH and NAND_BBT_NO_OOB).

Note: the rare "unaligned" case is a result of the extra BBT pattern +
version located in the data area instead of OOB.

Also, this patch disables caching on raw reads, since we are reading
without error correction. This is, obviously, prone to errors and should
not be cached.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

The nand_chip.ops field is a struct that is passed around globally with
no particular reason. Every time it is used, it could just as easily be
replaced with a local struct that is updated on each operation. So make
it local.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

These modes are not necessarily for OOB only. Particularly, MTD_OOB_RAW
affected operations on in-band page data as well. To clarify these
options and to emphasize that their effect is applied per-operation, we
change the primary prefix to MTD_OPS_.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

This fixes issues with `nanddump -n' and the MEMREADOOB[64] ioctls on
hardware that performs error correction when reading only OOB data. A
driver for such hardware needs to know when we're doing a RAW vs. a
normal write, but mtd_do_read_oob does not pass such information to the
lower layers (e.g., NAND). We should pass MTD_OOB_RAW or MTD_OOB_PLACE
based on the MTD file mode.

For now, most drivers can get away with just setting:

  chip->ecc.read_oob_raw = chip->ecc.read_oob

This is done by default; but for systems that behave as described above,
you must supply your own replacement function.

This was tested with nandsim as well as on actual SLC NAND.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Cc: Jim Quinlan <jim2101024@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

This fixes issues with `nandwrite -n -o' and the MEMWRITEOOB[64] ioctls
on hardware that writes ECC when writing OOB. The problem arises as
follows: `nandwrite -n' can write page data to flash without applying
ECC, but when used with the `-o' option, ECC is applied (incorrectly),
contrary to the `--noecc' option.

I found that this is the case because my hardware computes and writes
ECC data to flash upon either OOB write or page write. Thus, to support
a proper "no ECC" write, my driver must know when we're performing a raw
OOB write vs. a normal ECC OOB write. However, MTD does not pass any raw
mode information to the write_oob functions.  This patch addresses the
problems by:

1) Passing MTD_OOB_RAW down to lower layers, instead of just defaulting
   to MTD_OOB_PLACE
2) Handling MTD_OOB_RAW within the NAND layer's `nand_do_write_oob'
3) Adding a new (replaceable) function pointer in struct ecc_ctrl; this
   function should support writing OOB without ECC data. Current
   hardware often can use the same OOB write function when writing
   either with or without ECC

This was tested with nandsim as well as on actual SLC NAND.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Cc: Jim Quinlan <jim2101024@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Start moving away from the MTD_DEBUG_LEVEL messages. The dynamic
debugging feature is a generic kernel feature that provides more
flexibility.

(See Documentation/dynamic-debug-howto.txt)

Also fix some punctuation, indentation, and capitalization that went
along with the affected lines.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

This is a cleanup of some punctuation, indentation, and capitalization
on the lines affected affected by the last patch.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Instead of directly calling printk, it's simpler to use the built-in
pr_* functions. This shortens code and allows easy customization through
the definition of a pr_fmt() macro (not used currently). Ideally, we
could implement much of this with dev_* functions, but the MTD subsystem
does not necessarily register all its master `mtd_info.dev` device, so
we cannot use dev_* consistently. See:
http://lists.infradead.org/pipermail/linux-mtd/2011-July/036950.htmlSigned-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

there is a bug in nand_flash_detect_onfi, busw need to be passed
by pointer to return it.
Signed-off-by: NMatthieu CASTET <matthieu.castet@parrot.com>
Acked-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <dedekind1@gmail.com>

While the standard NAND OOB functions do not do ECC on the spare area,
it is possible for a driver to supply its own OOB ECC functions (e.g., HW
ECC). nand_do_read_oob should act like nand_do_read_ops in checking the
ECC stats and returning -EBADMSG or -EUCLEAN on uncorrectable errors or
correctable bitflips, respectively. These error codes could be used in
flash-based BBT code or in YAFFS, for example.

Doing this, however, messes with the behavior of mtd_do_readoob. Now,
mtd_do_readoob should check whether we had -EUCLEAN or -EBADMSG errors
and discard those as "non-fatal" so that the ioctls can still succeed
with (possibly uncorrected) data.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

Therefor -> Therefore
[Intern], [Internal] -> [INTERN]
[REPLACABLE] -> [REPLACEABLE]
syndrom, syndom -> syndrome
ecc -> ECC
buswith -> buswidth
endianess -> endianness
dont -> don't
occures -> occurs
independend -> independent
wihin -> within
erease -> erase
blockes -> blocks
...
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

In nand_do_write_ops() code it is possible for a caller to provide
ops.oobbuf populated and ops.mode == MTD_OOB_AUTO, which currently
means that the chip->oob_poi buffer isn't initialised to all 0xFF.
The nand_fill_oob() method then carries out the task of copying
the provided OOB data to oob_poi, but with MTD_OOB_AUTO it skips
areas marked as unavailable by the layout struct, including the
bad block marker bytes.

An example of this causing issues is when the last OOB data read
was from the start of a bad block where the markers are not 0xFF,
and the caller wishes to write new OOB data at the beginning of
another block. In this scenario the caller would provide OOB data,
but nand_fill_oob() would skip the bad block marker bytes in
oob_poi before copying the OOB data provided by the caller.
This means that when the OOB data is written back to NAND,
the block is inadvertently marked as bad without the caller knowing.
This has been witnessed when using YAFFS2 where tags are stored
in the OOB.

To avoid this oob_poi is always initialised to 0xFF to make sure
no left over data is inadvertently written back to the OOB area.

Credits to Brian Norris <computersforpeace@gmail.com> for fixing this
patch.
Signed-off-by: NAdam Thomson <adam.thomson@alcatel-lucent.com>
Signed-off-by: NArtem Bityutskiy <dedekind1@gmail.com>
Cc: stable@kernel.org [2.6.20+]

Artem: while on it, do other commentaries clean-ups:
1. Start one-line comments with capital letter and no dot at the end
2. Turn sparse multi-line comments into one-line comments
3. Change "phrase ?" to "phrase?" and the same with "!".
4. Remove tabs from the kerneldoc parameters comments - they are mixed
   with tabs often, and inconsistent.
5. Put dot at the end of descriptions in kerneldoc comments.
6. Some other small commentaries clean-ups
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

I believe this TODO was unnecessary back when it was introduced:
	commit d1e1f4e4
	mtd: nand: add support for reading ONFI parameters...
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

This delay is meaningless. If delay is needed it is device specific
and must be reimplemented by specific driver, otherwise no delay is
needed.
Signed-off-by: NJiri Pinkava <jiri.pinkava@vscht.cz>
Acked-by: NVimal Singh <vimal.newwork@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

Recall the recently added prefix requirements:
 * "NAND_" for flags in nand.h, used in nand_chip.options
 * "NAND_BBT_" for flags in bbm.h, used in nand_chip.bbt_options
        or in nand_bbt_descr.options

Thus, I am changing NAND_USE_FLASH_BBT to NAND_BBT_USE_FLASH.

Again, this flag is found in bbm.h and so should NOT be used in the
"nand_chip.options" field.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

This patch works with the following three flags from two headers (nand.h
and bbm.h):
  (1) NAND_USE_FLASH_BBT (nand.h)
  (2) NAND_USE_FLASH_BBT_NO_OOB (nand.h)
  (3) NAND_BBT_NO_OOB (bbm.h)

These flags are all related and interdependent, yet they were in
different headers. Flag (2) is simply the combination of (1) and (3) and
can be eliminated.

This patch accomplishes the following:
  * eliminate NAND_USE_FLASH_BBT_NO_OOB (i.e., flag (2))
  * move NAND_USE_FLASH_BBT (i.e., flag (1)) to bbm.h

It's important to note that because (1) and (3) are now both found in
bbm.h, they should NOT be used in the "nand_chip.options" field.

I removed a small section from the mtdnand DocBook because it referes to
NAND_USE_FLASH_BBT in nand.h, which has been moved to bbm.h.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

This patch handles the problems we've been having with using conflicting
flags from nand.h and bbm.h in the same nand_chip.options field. We
should try to separate these two spaces a little more clearly, and so I
have added a bbt_options field to nand_chip.

Important notes about nand_chip fields:
* bbt_options field should contain ONLY flags from bbm.h. They should be
  able to pass safely to a nand_bbt_descr data structure.
    - BBT option flags start with the "NAND_BBT_" prefix.
* options field should contian ONLY flags from nand.h. Ideally, they
  should not be involved in any BBT related options.
    - NAND chip option flags start with the "NAND_" prefix.
* Every flag should have a nice comment explaining what the flag is. While
  this is not yet the case on all existing flags, please be sure to write
  one for new flags. Even better, you can help document the code better
  yourself!

Please try to follow these conventions to make everyone's lives easier.

Among the flags that are being moved to the new bbt_options field
throughout various drivers, etc. are:
 * NAND_BBT_SCANLASTPAGE
 * NAND_BBT_SCAN2NDPAGE
and there will be more to come.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

This patch reverts most of:
    commit 58373ff0
    mtd: nand: more BB Detection refactoring and dynamic scan options

According to the discussion at:
    http://lists.infradead.org/pipermail/linux-mtd/2011-May/035696.html
the NAND_BBT_SCANBYTE1AND6 flag, although technically valid, can break
some existing ECC layouts that use the 6th byte in the OOB for ECC data.
Furthermore, we apparently do not need to scan both bytes 1 and 6 in
the OOB region of the devices under consideration; instead, we only need
to scan one or the other.

Thus, the NAND_BBT_SCANBYTE1AND6 flag is at best unnecessary and at
worst a regression.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>

Convert to mtd_device_register() and remove the CONFIG_MTD_PARTITIONS
preprocessor conditionals as partitioning is always available.
Signed-off-by: NJamie Iles <jamie@jamieiles.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

In commit c7b28e25 the initialization of
the backblockbits was accidentally removed. This patch returns it back,
because otherwise some NAND drivers are broken.

This problem was reported by "Saxena, Parth <parth.saxena@ti.com>" here:
http://lists.infradead.org/pipermail/linux-mtd/2011-April/035221.htmlReported-by: NSaxena, Parth <parth.saxena@ti.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Tested-by: NSaxena, Parth <parth.saxena@ti.com>
Acked-by: NSaxena, Parth <parth.saxena@ti.com>
Acked-by: NBrian Norris <computersforpeace@gmail.com>
Cc: stable@kernel.org [2.6.36+]
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Chip deselection is already done in nand_release_device. So only
duplicate code.
Signed-off-by: NJiri Pinkava <jiri.pinkava@vscht.cz>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Fixes generated by 'codespell' and manually reviewed.
Signed-off-by: NLucas De Marchi <lucas.demarchi@profusion.mobi>

This patch adds software BCH ECC support to mtd, in order to handle recent
NAND device ecc requirements (4 bits or more).

It does so by adding a new ecc mode (NAND_ECC_SOFT_BCH) for use by board
drivers, and a new Kconfig option to enable BCH support. It relies on the
generic BCH library introduced in a previous patch.

When a board driver uses mode NAND_ECC_SOFT_BCH, it should also set fields
chip->ecc.size and chip->ecc.bytes to select BCH ecc data size and required
error correction capability. See nand_bch_init() documentation for details.

It has been tested on the following platforms using mtd-utils, UBI and
UBIFS: x86 (with nandsim), arm926ejs.
Signed-off-by: NIvan Djelic <ivan.djelic@parrot.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Signed-off-by: NJason Liu <r64343@freescale.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Initialize mtd->writebufsize to be equal to mtd->writesize.
Signed-off-by: NAnatolij Gustschin <agust@denx.de>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

In checking for the ONFI revision, the first conditional (for checking
"unsupported" ONFI) seems unnecessary.  All ONFI revisions should be
backwards-compatible; even if this is not the case on some newer ONFI
revision, it should simply fail the second version-checking if-else block
(i.e., the bit-fields for 1.0, 2.0, etc. would not be set to 1). Thus, we
move our "unsupported" condition after having checked each bit field.

Also, it's simple enough to add a condition for ONFI revision 2.3. Note
that this does *NOT* mean we handle all new features of ONFI versions
above 1.0.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Acked-by: NFlorian Fainelli <ffainelli@freebox.fr>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

We have the order of the conditional wrong for choosing the ONFI chip name
vs. the ID table name. Without this fix, we will almost *always* choose a
NULL string to print out instead of the correct one.

This has already been suggested by Matthieu Castet.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Acked-by: NFlorian Fainelli <ffainelli@freebox.fr>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

This reverts commit e14feafb.

The commit limits the maximum amount of bytes which can be read
at one go to the OOB size, which is incorrect, because mtd->read_oob()
allows reading multiple pages at a time, see comment near
"struct mtd_oob_ops" at include/linux/mtd/mtd.h. So this patch
breaks ABI and hence, has to be reverted.
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>

When run mtd_oobtest case, there will be one error for step(4),
which turned out it need add one check for out of page read in
nand_do_read_oob just like mtd_do_write_oob did it already.
This commit also fix one typo error for comments in mtd_do_write_oob
Signed-off-by: NJason Liu <r64343@freescale.com>
Signed-off-by: NArtem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>