Currently the docg4's ecc.read_page() method returns -EBADMSG when
uncorrectable bitflips occur. This is wrong; 0 should be returned in
this case. An error code should only be returned by this method in the
case of a hardware error (probably -EIO).
Signed-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>
As a first step towards migrating davinci platforms to use common clock
framework, replace all instances of clk_enable() with clk_prepare_enable()
and clk_disable() with clk_disable_unprepare(). Until the platform is
switched to use the CONFIG_HAVE_CLK_PREPARE Kconfig variable, this just
adds a might_sleep() call and would work without any issues.
This will make it easy later to switch to common clk based implementation
of clk driver from DaVinci specific driver.
Signed-off-by: Murali Karicheri <m-karicheri2@ti.com>
Reviewed-by: Mike Turquette <mturquette@linaro.org>
Acked-by: Mike Turquette <mturquette@linaro.org>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
This adds the double bit error detection test cases listed below:
* Prepare data block with double bit error and ECC data without
corruption, and verify that the uncorrectable error is detected by
__nand_correct_data().
* Prepare data block with single bit error and ECC data with single bit
error, and verify that the uncorrectable error is detected.
* Prepare data block without corruption and ECC data with double bit
error, and verify that the uncorrectable error is detected.
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
This adds the single bit error correction test case listed below:
Prepare data block without corruption and ECC data with single bit error,
and verify that the data block is preserved by __nand_correct_data().
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
This adds no corruptin test case listed below:
Prepare data block and ECC data with no corruption, and verify that
the data block is preserved by __nand_correct_data()
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
This rewrites the entire test routine in order to make it easy to add more
tests by later changes and minimize duplication of each tests as much as
possible.
Now that each test is described by the members of struct nand_ecc_test:
- name: descriptive testname
- prepare: function to prepare data block and ecc with artifical corruption
- verify: function to verify the result of correcting data block
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Currently inject_single_bit_error() is used to inject single bit error
into randomly selected bit position of the 256 or 512 bytes data block.
Later change will add tests which inject bit errors into the ecc code.
Unfortunately, inject_single_bit_error() doesn't work for the ecc code
which is not a multiple of sizeof(unsigned long).
Because bit fliping at random position is done by __change_bit().
For example, flipping bit position 0 by __change_bit(0, addr) modifies
3rd byte (32bit) or 7th byte (64bit) on big-endian systems.
Using little-endian version of bitops can fix this issue. But
little-endian version of __change_bit is not yet available.
So this defines __change_bit_le() locally in a similar fashion to
asm-generic/bitops/le.h and use it.
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
This tests ECC biterror recovery on a single NAND page. Mostly intended
to test ECC hardware and low-level NAND driver.
There are two test modes:
0 - artificially inserting bit errors until the ECC fails
This is the default method and fairly quick. It should
be independent of the quality of the FLASH.
1 - re-writing the same pattern repeatedly until the ECC fails.
This method relies on the physics of NAND FLASH to eventually
generate '0' bits if '1' has been written sufficient times. Depending
on the NAND, the first bit errors will appear after 1000 or
more writes and then will usually snowball, reaching the limits
of the ECC quickly.
The test stops after 10000 cycles, should your FLASH be exceptionally
good and not generate bit errors before that. Try a different page
offset in that case.
Please note that neither of these tests will significantly 'use up' any FLASH
endurance. Only a maximum of two erase operations will be performed.
Signed-off-by: Iwo Mergler <Iwo.Mergler@netcommwireless.com.au>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Huang Shijie <shijie8@gmail.com> explains:
Assume we have a 1GiB(8Gib) NAND chip, and we set the partitions
in the command line like this:
#gpmi-nand:100m(boot),100m(kernel),1g(rootfs)
In this case, the partition truncating occurs. The current code will
get the following result:
----------------------------------
root@freescale ~$ cat /proc/mtd
dev: size erasesize name
mtd0: 06400000 00040000 "boot"
mtd1: 06400000 00040000 "kernel"
----------------------------------
It is obvious that we lost the truncated partition `rootfs` which should
be 824MiB in this case.
Also, forbid 0-sized partitions.
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>
With the new i.mx clock framework the mxc_nand clock is registered as:
clk_register_clkdev(clk[nfc_gate], NULL, "mxc_nand");0")
So we do not need to pass "nfc" string and can use NULL instead.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Quoting from the datasheet for S25FL064P, rev. 05, Nov 18 2011, § 9.17:
"A 64 kB[sic] sector erase (D8h) command issued on 4 kB or 8 kB erase
sectors will erase all sectors in the specified 64 kB region. However,
please note that a 4 kB sector erase (20h) or 8 kB sector erase (40h)
command will not work on a 64 kB sector."
Referring further to Table 8.1 and Table 8.2, it is clearly seen
that most of the sectors are 64KiB; therefore disable this 4KiB
erase support since it's valid only on first/last sectors.
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>
This registers MTD driver for serial flash platform device. Right now it
supports reading only, writing still has to be implemented.
Artem: minor amendments.
Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
We need the following 2 patches from the 'net-next' tree for the BCMA flash
driver:
371a004 bcma: detect and register NAND flash device
d57ef3a bcma: detect and register serial flash device
and this is why we are merging the net-next tree (presumably persistent)
up to commit '371a004'.
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Currently the data blocks which is used to test single bit error
correction is allocated statically and injecting single bit error is
implemented by using __change_bit() which must operate on the memory
aligned to the size of an "unsigned long". But there is no such
guarantee for statically allocated array.
This fix the issue by allocating the data block dynamically by
kmalloc(). It also allocate the ecc code dynamically instead of
allocating statically on stack.
The reason to allocate the ecc code dynamically is that later change
will add tests which inject bit errors into the ecc code by bitops.
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
This includes the message related changes:
- Use pr_* instead of printk
- Print hexdump of ECC code if test fails
- Change log level for hexdump of data from KERN_DEBUG to KERN_INFO
- Factor out the hexdump code into a separate function
Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Huang Shijie
Kim Phillips
Richard Genoud
Mike Dunn
Christian Daudt
m-karicheri2@ti.com
Akinobu Mita
Iwo Mergler
Shmulik Ladkani
Sachin Kamat
Fabio Estevam
Marek Vasut
Rafał Miłecki
David Woodhouse