Because there are so many cases of checking, writing, and re-writing of
the bad block table(s), we might as well wait until the we've settled on
a valid, clean copy of the table. This also prevents us from falsely
incrementing the table version. For example, we may have the following:

  Primary table, with version 0x02
  Mirror table, with version 0x01
  Primary table has uncorrectable ECC errors

If we don't have this fix applied, then we will:

  Choose to read the primary table (higher version)
  Set mirror table version to 0x02
  Read back primary table
  Invalidate table because of ECC errors
  Retry readback operation with mirror table, now version 0x02
  Mirrored table reads cleanly
  Writeback BBT to primary table location (with "version 0x02")

However, the mirrored table shouldn't have a new version number.
Instead, we actually want:

  Choose to read the primary table (higher version)
  Read back primary table
  Invalidate table because of ECC errors
  Retry readback with mirror table (version 0x01)
  Mirrored table reads cleanly
  Set both tables to version 0x01
  Writeback BBT to primary table location (version 0x01)
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Now that `read_bbt()' returns ECC error codes properly, we handle those
codes when checking the integrity of our flash-based BBT.

The modifications can be described by this new policy:

*) On any uncorrected ECC error, we invalidate the corresponding table
   and retry our version-checking integrity logic.
*) On corrected bitflips, we mark both tables for re-writing to flash
   (a.k.a. scrubbing).

Current integrity checks (i.e., comparing version numbers, etc.) should
take care of all the cases that result in rescanning the device for bad
blocks or falling back to the BBT as found in the mirror descriptor.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Instead of just printing a warning when encountering ECC errors, we
should return a proper error status and print a more informative
warning. Later, we will handle these error messages in the upper layers
of the BBT scan.

Note that this patch makes our check for ECC error codes a little bit
more restrictive, leaving all unrecognized errors to the generic "else"
clause. This shouldn't cause problems and could even be a benefit.

This code is based on some findings reported by Matthieu Castet.
Reported-by: NMatthieu CASTET <matthieu.castet@parrot.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

These functions can be used instead of referencing -EUCLEAN and -EBADMSG
all over the place. They should help make code a little bit more
readable.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

This is a second step in restructuring `check_create()'. When we don't
rely on goto statements for our main functionality, the code will become
a little easier to manipulate.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

We will begin restructuring the code for check_create so that we can
make some important changes. For now, we should just begin to get rid of
some goto statements to make things cleaner. This is the first step of a
few, which are separated to make them easier to follow.

This step should just be a code refactor.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Remove some extra spaces
Consistently use '0x' prefix for bitfield-like constants
Spelling: "aplies" -> "applies"
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

`writeops' is unnecessary in the function `nand_update_bbt()'
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

There are a few reasons not to ignore ECC errors here.

First, mtd->read_oob is being called in raw mode, so there should be no
error correction in the first place.

Second, if we change this such that there *is* error correction in this
function, then we will want to pass the error message upward.

In fact, the code I introduced to "ignore ECC errors" would have been
better if it had just placed this test down in `scan_block_full()' in
the first place. We would like to ignore ECC errors when we are simply
checking for bad block markers (e.g., factory marked), but we may not
want to ignore ECC errors when scanning OOB for a flash-based BBT
pattern (in `scan_read_raw()'; note that the return codes from
`scan_read_raw()' are not actually handled yet).
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

A few pieces of code are unnecessarily duplicated. For easier
maintenance, we should fix this.

This should have no functional effect.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

The integrator flash has been deleted, even from the Makefile.
Drop the Kconfig entry as well.
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Acked-by: NMarc Zyngier <marc.zyngier@arm.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

The ecctype and eccsize fields have been obsolete for a while. Since they
don't have any users, we can kill them and leave padding in their place
for now.
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>

We're missing a lot of important documentation in include/mtd/mtd-abi.h:

* add a simple description of each ioctl (feel free to expand!)
* give full explanations of recently added and modified operations
* explain the usage of "RAW" that appear in different modes and types of
  operations
* fix some comment style along the way
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Implement a new ioctl for writing both page data and OOB to flash at the
same time. This ioctl is intended to be a generic interface that can
replace other ioctls (MEMWRITEOOB and MEMWRITEOOB64) and cover the
functionality of several other old ones, e.g., MEMWRITE can:

* write autoplaced OOB instead of using ECCGETLAYOUT (deprecated) and
  working around the reserved areas
* write raw (no ECC) OOB instead of using MTDFILEMODE to set the
  per-file-descriptor MTD_FILE_MODE_RAW
* write raw (no ECC) data instead of using MTDFILEMODE
  (MTD_FILE_MODE_RAW) and using standard character device "write"

This ioctl is especially useful for MLC NAND, which cannot be written
twice (i.e., we cannot successfully write the page data and OOB in two
separate operations). Instead, MEMWRITE can write both in a single
operation.

Note that this ioctl is not affected by the MTD file mode (i.e.,
MTD_FILE_MODE_RAW vs. MTD_FILE_MODE_NORMAL), since it receives its write
mode as an input parameter.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

These modes hold their state only for the life of their file descriptor,
and they overlap functionality with the MTD_OPS_* modes. Particularly,
MTD_MODE_RAW and MTD_OPS_RAW cover the same function: to provide raw
(i.e., without ECC) access to the flash. In fact, although it may not be
clear, MTD_MODE_RAW implied that operations should enable the
MTD_OPS_RAW mode.

Thus, we should be specific on what each mode means. This is a start,
where MTD_FILE_MODE_* actually represents a "file mode," not necessarily
a true global MTD mode.
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>

We will want to use the MTD_OOB_{PLACE,AUTO,RAW} modes in user-space
applications through the introduction of new ioctls, so we should make
this enum a shared type.

This enum is now anonymous.

Artem: tweaked the patch.
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>

Signed-off-by: NBrian Norris <computersforpeace@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>

These files contain the common code for the GPMI-NAND driver.
Signed-off-by: NHuang Shijie <b32955@freescale.com>
Acked-by: NMarek Vasut <marek.vasut@gmail.com>
Tested-by: NKoen Beel <koen.beel@barco.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

add the GPMI-NAND driver in the relevant Kconfig and Makefile in the MTD.
Signed-off-by: NHuang Shijie <b32955@freescale.com>
Acked-by: NMarek Vasut <marek.vasut@gmail.com>
Tested-by: NKoen Beel <koen.beel@barco.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

bch-regs.h : registers file for BCH module
gpmi-regs.h: registers file for GPMI module
gpmi-lib.c: helper functions library.
Signed-off-by: NHuang Shijie <b32955@freescale.com>
Acked-by: NMarek Vasut <marek.vasut@gmail.com>
Tested-by: NKoen Beel <koen.beel@barco.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

The code has the check for parts but it called after kmemdup,
kmemdup(parts, sizeof(*parts) * nr_parts,...)
if (!parts)
	return -ENOMEM

In fact, we need check parts before safely using it.
and we also need check the real_parts to make sure kmemdup
allocation sucessfully.
Signed-off-by: NJason Liu <jason.hui@linaro.org>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

MEMSETOOBSEL is completely unused and useless. Remove the definition.

Note: it's probably best not to use this ioctl number in the future for
MTD, since that may cause conflicts between old kernels and new user
software (or new kernels and old user software).

Artem: leave a comment about MEMSETOOBSEL.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

It's been gone for a while.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Previous generations of MTDs all used OOB sizes that were powers of 2,
(e.g., 64, 128). However, newer generations of flash, especially NAND,
use irregular OOB sizes that are not powers of 2 (e.g., 218, 224, 448).
This means we cannot use masks like "mtd->oobsize - 1" to assume that we
will get a proper bitmask for OOB operations.

These masks are really only intended to hide the "page" portion of the
offset, leaving any OOB offset intact, so a masking with the writesize
(which *is* always a power of 2) is valid and makes more sense.

This has been tested for read/write of NAND devices (nanddump/nandwrite)
using nandsim and actual NAND flash.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

This has already been tested with Samsung NAND: K9LAG08U0M
on MX53EVK board, ubi/ubifs has already been tested OK too.
Signed-off-by: NJason Liu <jason.hui@linaro.org>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Convert error handling code to use gotos.  At the same time, this adds
calls to kfree and iounmap in a few cases where they were overlooked.
Signed-off-by: NJulia Lawall <julia@diku.dk>
Acked-by: NJiandong Zheng <jdzheng@broadcom.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Signed-off-by: NMichael Hench <MichaelHench@gmail.com>
Acked-by: NScott Wood <scottwood@freescale.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

clk_get() return a pointer to the struct clk or an ERR_PTR().
Signed-off-by: NJonghwan Choi <jhbird.choi@samsung.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

use MTD_NAND_OMAP2 also for OMAP4 arch.
testes with omap4430
Signed-off-by: NJan Weitzel <j.weitzel@phytec.de>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Some messages that were tied to CONFIG_MTD_DEBUG_VERBOSE can now simply
be enabled using dynamic debugging features, if necessary. There's no
need for special debugging functions here.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>

Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@intel.com>