The sam926x docs allegedly don't list an "ECC_PARITY" field, and the 
header files in the upstream kernel don't have it either.

Masking with it was useless anyway, so just remove it.
Signed-off-by: Richard Genoud <richard.genoud@gmail.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

This is a patch to use the hardware ECC controller of
the AT91SAM9260 and AT91SAM9263 for the AT91 nand.
On AT91 NAND, there's now a choice between ECC soft,
ECC hard or no ECC (for debug).

It has been tested on AT91SAM9263 with 8 bits large
and small page NAND.
Signed-off-by: Richard Genoud <richard.genoud@gmail.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

We haven't seen bugs in this for a while now, since the rewrite. No need
to be _quite_ so verbose...
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

When _all_ the blocks were on the erase_pending_list, we could't find a
block to GC from but there was no _actually_ free space, and
jffs2_reserve_space() would get a little unhappy.

Handle this case by returning -EAGAIN from jffs2_garbage_collect_pass().
There are two callers of that function -- jffs2_flush_wbuf_gc(), which
will interpret it as an error and flush the writebuffer by other means,
and jffs2_reserve_space(), which we modify to respond to -EAGAIN with an
immediate call to jffs2_erase_pending_blocks() and another run round the
loop.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Just to keep the debug code happy when it's adding all the blocks up.
Otherwise, they disappear for a while while the locks are dropped to
check them and write the cleanmarker.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

UBI scan takes quite a time on some systems, so it is nice
to print a message that we started attaching an MTD device.
Signed-off-by: Artem Bityutskiy <Artem.Bityutskiy@nokia.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

It looks the error paths in jffs2_block_check_erase() have wrong return 
values. A block that failed to be erased never gets marked as bad.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

drivers/mtd/ar7part.c: In function ‘create_mtd_partitions’:
drivers/mtd/ar7part.c:69: warning: passing argument 4 of ‘master->read’ from incompatible pointer type
drivers/mtd/ar7part.c:91: warning: passing argument 4 of ‘master->read’ from incompatible pointer type
drivers/mtd/ar7part.c:99: warning: passing argument 4 of ‘master->read’ from incompatible pointer type
drivers/mtd/ar7part.c:110: warning: passing argument 4 of ‘master->read’ from incompatible pointer type
drivers/mtd/ar7part.c:111: error: ‘SQUASHFS_MAGIC’ undeclared (first use in this function)
drivers/mtd/ar7part.c:111: error: (Each undeclared identifier is reported only once
drivers/mtd/ar7part.c:111: error: for each function it appears in.)
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

collie seems to contain LH28F640BF flash chips. According to
http://sharp-world.com/products/device/flash/pdf/*FUM00701*@E.pdf
(page 83) if they have 0x51 of Extended Query Table (number of hardware
partitions) set to zero, they have a single fixed partition.
This patch makes those chips work.
Signed-off-by: Thomas Kunze <thommycheck@gmx.de>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Haven't had any complaints about it recently, despite having the test
code enabled to verify that the calculated length is correct.

Kill it off, just by #undef TEST_TOTLEN for now; removing it for real
can come a little later.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Signed-off-by: David Woodhouse <dwmw2@infradead.org>

The problem fixed in commit 014b164e
(space leak with in-band cleanmarkers) would have been caught a lot
quicker if our paranoid debugging mode had included adding up the size
counts from all the eraseblocks and comparing the totals with the counts
in the superblock. Add that.

Make jffs2_mark_erased_block() file the newly-erased block on the
free_list before calling the debug function, to make it happy.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

We were accounting for the cleanmarker by calling jffs2_link_node_ref()
(without locking!), which adjusted both superblock and per-eraseblock
accounting, subtracting the size of the cleanmarker from {jeb,c}->free_size
and adding it to {jeb,c}->used_size.

But only _then_ were we adding the size of the newly-erased block back
to the superblock counts, and we were adding each of jeb->{free,used}_size
to the corresponding superblock counts. Thus, the size of the cleanmarker
was effectively subtracted from the superblock's free_size _twice_.

Fix this, by always adding a full eraseblock size to c->free_size when
we've erased a block. And call jffs2_link_node_ref() under the proper
lock, while we're at it.

Thanks to Alexander Yurchenko and/or Damir Shayhutdinov for (almost)
pinpointing the problem.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

... instead of <linux/semaphore.h> which we don't need any more anyway.
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

With modern systems using bus-hold instead of bus pull-up, it can
often lead to erroneous reporting of NAND devices where there are
none. Do a double probe to ensure that the result we got the first
time is repeatable, and if it is not then return that there is no
chip there.
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Add support to disable ECC checking for a given chip
when passed by the board via the platform data.
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Add support for the ECC layout to be passed via the
platform data specified by the board.
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

If a block's ecc field is all 0xff, then ignore the ECC
correction. This is for systems where some of the blocks,
such as the initial cramfs are written without ECC and
need to be loaded on start.
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

This adds support for using large page NAND devices
with the S3C24XX NAND controller. This also adds the
file Documentation/arm/Samsung-S3C24XX/NAND.txt to
describe the differences.
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Commit 03680b1e incorrectly
was assuming S3C2410_NFCONF was being used to select the
NAND chip. Fix this error by ising the sel_reg.
Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

A power loss while writing can result in a page becoming unreadable.  
When the device is mounted again, reading that page gives controller 
errors. Upper level software like JFFS2 treat -EIO as fatal, refusing to 
mount at all.  That means it is necessary to treat the error as an ECC 
error to allow recovery. Note that typically in this case, the 
eraseblock can still be erased and rewritten i.e. it has not become a 
bad block.
Signed-off-by: Adrian Hunter <ext-adrian.hunter@nokia.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Minor tweaks to omap_nor ... as with most platform drivers, its probe
and remove logic can (and should!) safely vanish in most configs.
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

This is a known erratum confirmed by Spansion.  I have an errata document,
but I can't find a link to it anywhere on their site to include here.

Some of the S29GL064N chips report 64 sectors when they should report 128,
and some of S29GL032N chips report 127 sectors when they should report 63.

Note that when the chip dies are fixed by Spansion, they will still have
the same id.  The fix is done in such a way that it won't affect corrected
chips.

The fixups use the extended id made available by a previous patch.  Without
that, virtually all newer AMD/Spansion chips will have the same ID (0x227e)
and it's not possible to apply the fixup to the correct chips.
Signed-off-by: Trent Piepho <tpiepho@freescale.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

AMD/Spansion use a device id of 0x7e to indicate an extended device is
present at offset 0xe and 0xf in the query data.

I've verified with Spansion that all their chips (mfr == 0x01) with an id
of 0x7e use it to indicate an extended id is present.  What's more, there
are no chips with a NON-extended id that is the same as a different chip's
extended id.  In other words, when the extended ID is present, one can
replace the normal id with the extended id without losing any information.
Which is what I've done.
Signed-off-by: Trent Piepho <tpiepho@freescale.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Using current driver elbc sometimes hangs during nand write. Reading back
last byte helps though (thanks to Scott Wood for the idea).
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Acked-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

This patch sets mtd->name to the platform bus ID in the plat_nand
driver, so that you can specify partitions readily with mtdparts=.

Currently it relies on nand_base filling in the name from the device,
which results in names like "NAND 256MiB 3,3V 8-bit", that you can't
use with cmdlineparts.
Signed-off-by: Hamish Moffatt <hamish@cloud.net.au>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Fix a race condition in fsl_elbc_run_command
Fix incorrect usage of clearbits32 that bashed option register
Remove work around for bashed register
Signed-off-by: Mike Hench <mhench@elutions.com>
Acked-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Currently fsl_elbc_nand doesn't initialize mtd->name, and this causes
nand_get_flash_type() to assign name that is equal to chip type, like
this:

   root@b1:~# cat /proc/mtd
   dev:    size   erasesize  name
   mtd0: 00800000 00010000 "fe000000.flash"
   mtd1: 02000000 00004000 "NAND 32MiB 3,3V 8-bit"

mtd0 is physmap_of flash (normal name), and mtd1 is fsl_elbc_nand.

Despite inconsistency, with mtd name like this specifying paritions
from the kernel command line becomes a torture (though, I didn't tried
and not sure if mtdparts= can handle spaces at all). Plus, this causes
real bugs when multiple fsl_elbc_nand chips registered.

With this patch applied fsl_elbc_nand chip will have proper name:

   root@b1:~# cat /proc/mtd
   dev:    size   erasesize  name
   mtd0: 00800000 00010000 "fe000000.flash"
   mtd1: 02000000 00004000 "e0600000.flash"

p.s. We can't use priv->dev->bus_id as in physmap_of, because
fsl_elbc_nand pretends to be a localbus controller, so its bus_id is
"address.localbus", which is incorrect and thus will also not work
for multiple chips.
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Signed-off-by: Matteo Croce <technoboy85@gmail.com>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Eugene Konev <ejka@imfi.kspu.ru>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

This is very simple driver, NAND is connected through localbus,
and User-Programmable Machine is doing various adjustments to
speak NAND. No special efforts needed to do read and write cycles,
though to control ALE and CLE phases, we ask UPM to generate exact
pre-programmed signals on the localbus lines.
Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Add support for the SST 36VF3203 flash chip. It is used on Emerson 
KSI8560 board.
Signed-off-by: Andrei Dolnikov <adolnikov@ru.mvista.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

This enhances plat-ram to take a map_probes argument in
the platform_data structure which allow plat-ram to support
any direct-mapped device that MTD supports (jedec, cfi, amd ..)

A few items are also fixed:
- Don't panic if probes is 0
- Actually use the partition list that is passed in
Signed-off-by: Florian Fainelli <florian.fainelli@telecomint.eu>
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Untested, but shouldn't break anything... Makes MTD_XIP arch
independent. I guess this is why xip_iprefetch() was made for.
Signed-off-by: Paulius Zaleckas <paulius.zaleckas@teltonika.lt>
Acked-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

This is preliminary since:

1. It supports only _one_ chip select at the moment. As there is no
   existing platforms available using two chip selects of the NAND
   controller, it shall really not include code for supporting the
   2nd chip select for now, as such code cannot be verified.

2. It resorts to the default and simpliest memory based badblock
   table

3. Only limited types of nand flash are currently supported. Most
   PXA3xx processors come with on-chip NAND flash dies, so there
   isn't much flexibility for other types of NAND.

4. The NAND controller should be configured to detect the device's
   ID, thus making it difficult to use nand_scan_ident() to assist
   the detection process (though it's not impossible)

TODO: fix all the above limitations of cuz :-)
Signed-off-by: eric miao <eric.miao@marvell.com>
Cc: Sergey Podstavin <spodstavin@ru.mvista.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Acked-By: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Signed-off-by: David Woodhouse <dwmw2@infradead.org>

Return 0 partitions instead of -EINVAL on no mtdpart= argument in kernel
cmdline or missing partition info for device.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Acked-by: Stefan Roese <sr@denx.de>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>

This patch extends the existing MAPS driver for the Nvidia CK804 chipset
(ck804xrom.c) to also work on the Nvidia MCP55 chipset. As both chipsets
are rather similar, suporting them both with the same driver is easy.
Signed-off-by: Carl-Daniel Hailfinger <c-d.hailfinger.devel.2006@gmx.net>
Signed-off-by: David Woodhouse <dwmw2@infradead.org>