Corner case reads do not work, and return false data and
ECC. This case is typically seen in a ubifs usage, with a
read of type:
 - docg3 docg3: doc_read_oob(from=14882415, mode=1,
 data=(c30eca40:12), oob=(  (null):0))

This results in the following reads:
 - docg3 docg3: doc_read_data_area(buf=  (null), len=111)
 - docg3 docg3: doc_read_data_area(buf=c30eca40, len=12)
 - docg3 docg3: doc_read_data_area(buf=  (null), len=389)
 - docg3 docg3: doc_read_data_area(buf=  (null), len=0)
 - docg3 docg3: doc_read_data_area(buf=  (null), len=16)

If we suppose that the pages content is :
 - bytes 0 .. 111   : 0x0a
 - bytes 112 .. 255 : 0x0f
Then the returned bytes will be :
 - 111 times 0x0a (correct)
 - 0x0a 2 times and 0x0f 10 times (incorrect, should be
 0x0a,0x0f)
 - 0x0f 389 times (correct)
 - nothing
 - correct OOB

The reason seams that the first 111 bytes read ends between
the 2 docg3 planes, and that the first following read (in
the 12 bytes sequence, read of 16 bit word) returns the byte
of the rightmost plane duplicated in high and lower byte of
the word.

Fix this behaviour by ensuring that if the previous read
ended up in-between the 2 planes, there will be a first 1
byte read to get back to the beginning of leftmost plane.
Signed-off-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>

These should be semicolons, not commas.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Spansion S29NS512P flash uses a 16bit transfer to report number
of sectors instead of two 8bit accesses as CFI specifies.

Artem: remove warning message which said that we are applying the
fixup - no need to scary the user unnecessarily.
Signed-off-by: Javier Martin <javier.martin@vista-silicon.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Unlike file data integrity the xattr data integrity was not checked
before some explicit access to the attribute was made.

This could leave in the system a number of corrupted extended attributes
which will be detected only at access time and possibly at a very late
time compared to the time the corruption actually happened.

This patch adds the ability to check for extended attribute integrity
on first GC scan pass (similar to file data integrity check). This allows
for all present attributes to be completly verified before any use of them.

In order to work correctly this patch also needs the patch allowing
JFFS2 to discriminate between recoverable and non recoverable errors
on extended attributes.
Signed-off-by: Jean-Christophe DUBOIS <jcd@tribudubois.net>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

This patch is basically a revert of commit f326966b.

It allows JFFS2 to make the distinction between a potential transient
error (reading or writing the media) and a non recoverable error like a
bad CRC on the extended attribute data or some insconsitent parameters.

In order to make clear that the error is indeed intended to report a
corrupted attribute, a new local error code (JFFS2_XATTR_IS_CORRUPTED)
is introduced rather than returning a confusing positive EIO, which is
what led to the inappropriate "fix" last time.

This error code is never reported to user space and only checked locally
in this file.
Signed-off-by: Jean-Christophe DUBOIS <jcd@tribudubois.net>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Two modes are supported: 4-bit and 8-bit error correction.
Note that 4-bit mode is only confirmed to work on OMAP3630 ES 1.x,
x >= 1. The OMAP3 GPMC hardware BCH engine computes remainder
polynomials, it does not provide automatic error location and
correction: this step is implemented using the BCH library.

This implementation only protects page data, there is no support
for protecting user-defined spare area bytes (this could be added
with few modifications); therefore, it cannot be used with YAFFS2
or other similar filesystems that depend on oob storage.

Before being stored to nand flash, hardware BCH ecc is adjusted
so that an erased page has a valid ecc; thus allowing correction of
bitflips in blank pages (also common on 4-bit devices).

BCH correction mode is selected at runtime by setting platform data
parameter 'ecc_opt' to value OMAP_ECC_BCH4_CODE_HW or
OMAP_ECC_BCH8_CODE_HW.

This code has been tested with mtd test modules, UBI and UBIFS on a
BeagleBoard revC3 (OMAP3530 ES3.0 + Micron NAND 256MiB 1,8V 16-bit).
Signed-off-by: Ivan Djelic <ivan.djelic@parrot.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

This patch adds a simple BCH ecc computation api, similar to the
existing Hamming ecc api. It is intended to be used by the MTD layer.
It implements the following features:

- support 4-bit and 8-bit ecc computation
- do not protect user bytes in spare area, only data area is protected
- ecc for an erased NAND page (0xFFs) is also a sequence of 0xFFs

This last feature is obtained by adding a constant polynomial to
the hardware computed ecc. It allows to correct bitflips in blank pages
and is extremely useful to support filesystems such as UBIFS, which expect
erased pages to contain only 0xFFs.

This api has been tested on an OMAP3630 board.

Artem: The OMAP maintainer Tony Lindgren gave us his blessing for merging
this patch via the MTD tree.
Signed-off-by: Ivan Djelic <ivan.djelic@parrot.com>
Acked-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Apparently, there is an implementor of 'read_oob' which may return an
error inidication (e.g. docg4_read_oob may return -EIO).

Test the return value of 'read_oob/read_oob_raw', and if negative,
propagate the error, so it's returned by the '_read_oob' interface.
Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

As of [mtd: nand: remove autoincrement 'sndcmd' code], the
NAND_CMD_READ0 command is issued unconditionally.

Thus, read_oob/read_oob_raw's 'sndcmd' argument is no longer needed, as
well as their return code.

Remove the 'sndcmd' parameter, and set the return code to 0.
Signed-off-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Winbond W25Q80BW is a 8Mbit serial flash memory device.
Signed-off-by: Thomas Abraham <thomas.abraham@linaro.org>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Currently JFFS2 file-system maps the VFS "superblock" abstraction to the
write-buffer. Namely, it uses VFS services to synchronize the write-buffer
periodically.

The whole "superblock write-out" VFS infrastructure is served by the
'sync_supers()' kernel thread, which wakes up every 5 (by default) seconds and
writes out all dirty superblock using the '->write_super()' call-back. But the
problem with this thread is that it wastes power by waking up the system every
5 seconds no matter what. So we want to kill it completely and thus, we need to
make file-systems to stop using the '->write_super' VFS service, and then
remove it together with the kernel thread.

This patch switches the JFFS2 write-buffer management from
'->write_super()'/'->s_dirt' to a delayed work. Instead of setting the 's_dirt'
flag we just schedule a delayed work for synchronizing the write-buffer.
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

We do not need to call 'jffs2_write_super()' on sync. This function
causes a GC pass to make sure the current contents is pushed out with
the data which we already have on the media.

But this is not needed on unmount and only slows sync down unnecessarily.
It is enough to just sync the write-buffer.

This call was added by one of the generic VFS rework patch-sets,
see d579ed00.
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

We do not need to call 'jffs2_write_super()' on unmount. This function
causes a GC pass to make sure the current contents is pushed out with
the data which we already have on the media.

But this is not needed on unmount and only slows unmount down unnecessarily.
It is enough to just sync the write-buffer.

This call was added by one of the generic VFS rework patch-sets,
see 8c85e125.
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

We do not need 'lock_super()'/'unlock_super()' in JFFS2 - kill them.
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

add gpmi support for mx6q.
Signed-off-by: Huang Shijie <b32955@freescale.com>
Signed-off-by: Huang Shijie <shijie8@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

This patch just adds the DT support to gpmi-nand.
Signed-off-by: Huang Shijie <b32955@freescale.com>
Signed-off-by: Huang Shijie <shijie8@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Don't read/write OOB if the caller doesn't require it.
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Don't read OOB if the caller didn't request it.
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Acked-by: Huang Shijie <b32955@freescale.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Don't read OOB if the caller doesn't require it.
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Reviewed-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Don't read OOB if the caller doesn't require it.
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Reviewed-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

We now have an interface for notifying the nand_ecc_ctrl functions when OOB
data must be returned to the upper layers and when it may be left untouched.
This patch fills in the 'oob_required' parameter properly from
nand_do_{read,write}_ops. When utilized properly in the lower layers, this
parameter can improve performance and/or reduce complexity for NAND HW and SW
that can simply avoid transferring the OOB data.
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Reviewed-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Acked-by: Jiandong Zheng <jdzheng@broadcom.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

New NAND controllers can perform read/write via HW engines which don't expose
OOB data in their DMA mode. To reflect this, we should rework the nand_chip /
nand_ecc_ctrl interfaces that assume that drivers will always read/write OOB
data in the nand_chip.oob_poi buffer. A better interface includes a boolean
argument that explicitly tells the callee when OOB data is requested by the
calling layer (for reading/writing to/from nand_chip.oob_poi).

This patch adds the 'oob_required' parameter to each relevant {read,write}_page
interface; all 'oob_required' parameters are left unused for now. The next
patch will set the parameter properly in the nand_base.c callers, and follow-up
patches will make use of 'oob_required' in some of the callee functions.

Note that currently, there is no harm in ignoring the 'oob_required' parameter
and *always* utilizing nand_chip.oob_poi, but there can be
performance/complexity/design benefits from avoiding filling oob_poi in the
common case. I will try to implement this for some drivers which can be ported
easily.

Note: I couldn't compile-test all of these easily, as some had ARCH
dependencies.

[dwmw2: Merge later 1/0 vs. true/false cleanup]
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Reviewed-by: Shmulik Ladkani <shmulik.ladkani@gmail.com>
Acked-by: Jiandong Zheng <jdzheng@broadcom.com>
Acked-by: Mike Dunn <mikedunn@newsguy.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Lantiq SoCs have a External Bus Unit (EBU) that is used to attach MTD media.
As we need to co-exist with PCI on the same bus, certain swapping settings must
be applied. Similar to the NOR map driver we need to apply a fix to make NAND
work. The easiest way is to use byte reads.
Signed-off-by: John Crispin <blogic@openwrt.org>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

This patch sets the of_match_table field inside plat_nand's platform_driver.
We also add a struct mtd_part_parser_data pointer to make sure of_part parsing
works.

If an arch wants to support plat_nand via DT it needs to setup the
platform_nand_data and hook it into the platform_device.
Signed-off-by: John Crispin <blogic@openwrt.org>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Add the id and sector mappings for mx25l2005a flash chips.
Signed-off-by: John Crispin <blogic@openwrt.org>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

This chip isn't JEDEC-compatible.
Signed-off-by: Marek Vasut <marex@denx.de>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

No drivers use auto-increment NAND, so kill the NO_AUTOINCR option entirely.
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

The NAND_NO_AUTOINCR option is always set, so we will kill the option and make
"no autoincrement" the default behavior for nand_base.c. Thus, we should remove
the code which decides whether or not to send the NAND_CMD_READ0 command.
Instead, we unconditionally send the command.
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

The NAND layer always has NAND_NO_AUTOINCR set, so we will never utilize the
AUTOINCR code in nandsim. We will be removing the NAND_NO_AUTOINCR option soon,
and so kill this code as well.
Signed-off-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

The drivers' _read() method, absent an error, returns a non-negative integer
indicating the maximum number of bit errors that were corrected in any one
region comprising an ecc step.  MTD returns -EUCLEAN if this is >=
bitflip_threshold, 0 otherwise.  If bitflip_threshold is zero, the comparison is
not made since these devices lack ECC and always return zero in the non-error
case (thanks Brian)¹.  Note that this is a subtle change to the driver
interface.

This and the preceding patches in this set were tested with ubi on top of the
nandsim and docg4 devices, running the ubi test io_basic from mtd-utils.

¹ http://lists.infradead.org/pipermail/linux-mtd/2012-March/040468.htmlSigned-off-by: Mike Dunn <mikedunn@newsguy.com>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Acked-by: Brian Norris <computersforpeace@gmail.com>
Ivan Djelic <ivan.djelic@parrot.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

This patch adds sanity checks that ensure that drivers for controllers with
hardware ECC set the 'strength' element in struct nand_ecc_ctrl.  Also stylistic
changes to the line that calculates strength for software ECC.

This v2 simplifies the check.  Thanks Brian!¹

¹ http://lists.infradead.org/pipermail/linux-mtd/2012-April/040890.htmlSigned-off-by: Mike Dunn <mikedunn@newsguy.com>
Acked-by: Brian Norris <computersforpeace@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

The ecc.read_page() method for nand drivers is changed to return the maximum
number of bitflips that were corrected on any one region covering an ecc step,
This patch doesn't change what the nand code returns to mtd.

This v2 includes the change to the fsl_ifc_nand driver requested by Scott¹.

¹ http://lists.infradead.org/pipermail/linux-mtd/2012-April/040883.htmlSigned-off-by: Mike Dunn <mikedunn@newsguy.com>
Acked-by (freescale changes): Scott Wood <scottwood@freescale.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

An element 'bitflip_threshold' is added to struct mtd_info, and also exposed as
a read/write variable in sysfs.  This will be used to determine whether or not
mtd_read() returns -EUCLEAN or 0 (absent a hard error).  If the driver leaves it
as zero, mtd will set it to a default value of ecc_strength.

This v2 adds the line that propagates bitflip_threshold from the master to the
partitions - thanks Ivan¹.

¹ http://lists.infradead.org/pipermail/linux-mtd/2012-April/040900.htmlSigned-off-by: Mike Dunn <mikedunn@newsguy.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

ecc_strength element of struct mtd_info is exposed as a read-only variable in
sysfs.
Signed-off-by: Mike Dunn <mikedunn@newsguy.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

This fixes a couple of ecc strength values for which I earlier made conservative
guesses, but whose correct values were later determined¹ (thanks Ivan).  Also
sets strength for fsl_ifc_nand, which was merged to mainline after the original
patch that set the strength for all drivers.

¹ http://lists.infradead.org/pipermail/linux-mtd/2012-March/040325.htmlSigned-off-by: Mike Dunn <mikedunn@newsguy.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

ecc_strength element of mtd_info will be the strength of one ecc step, not of
the entire writesize, as was previously planned.  This is the appropriate way
because, as was pointed out¹, bit errors in excess of the strength of one
step can cause a hard error if they all occur within the same ecc region.

¹ http://lists.infradead.org/pipermail/linux-mtd/2012-March/040313.htmlSigned-off-by: Mike Dunn <mikedunn@newsguy.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

To make sure the NAND chip is properly programmed we need a status
command before each page write. When CONFIG_MTD_NAND_VERIFY_WRITE=y this
assumption is broken when writing multiple pages consecutively. This
patch fixes this.
Signed-off-by: Bastian Hecht <hechtb@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Replace the verbose `je32_to_cpu(latest_node->csize)' with a shorter
`csize'.
Signed-off-by: Xi Wang <xi.wang@gmail.com>
Cc: Artem Bityutskiy <dedekind1@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

`csize' is read from disk and thus needs validation.  Otherwise a bogus
value 0xffffffff would turn the subsequent kmalloc(csize + 1, ...) into
kmalloc(0, ...), leading to out-of-bounds write.

This patch limits `csize' to JFFS2_MAX_NAME_LEN, which is also used
in jffs2_symlink().

Artem: we actually validate csize by checking CRC, so this 0xFFs cannot
come from empty flash region. But I guess an attacker could feed JFFS2
an image with random csize value, including 0xFFs.
Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Signed-off-by: Peter Meerwald <p.meerwald@bct-electronic.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>