Commit 06bb6f5a ("mtd: spi-nor: stop (ab)using struct
spi_device_id") converted an array into a pointer, which means that
we should be checking if the pointer goes anywhere, not whether the C
string is empty. To do the latter means we dereference a NULL pointer
when we reach the terminating entry, for which 'name' is now NULL
instead of an array { 0, 0, ... }.

Sample crash:

[    1.101371] Unable to handle kernel NULL pointer dereference at virtual address 00000000
[    1.109457] pgd = c0004000
[    1.112157] [00000000] *pgd=00000000
[    1.115736] Internal error: Oops: 5 [#1] SMP ARM
[    1.120345] Modules linked in:
[    1.123405] CPU: 3 PID: 1 Comm: swapper/0 Not tainted 4.2.0-next-20150902+ #61
[    1.130611] Hardware name: Rockchip (Device Tree)
[    1.135306] task: ee0b8d40 ti: ee0ba000 task.ti: ee0ba000
[    1.140697] PC is at spi_nor_scan+0x90/0x8c4
[    1.144958] LR is at spi_nor_scan+0xa4/0x8c4
...
[    1.504112] [<c03cc2e0>] (spi_nor_scan) from [<c03cb188>] (m25p_probe+0xc8/0x11c)
[    1.511583] [<c03cb188>] (m25p_probe) from [<c03cd9d8>] (spi_drv_probe+0x60/0x7c)
[    1.519055] [<c03cd9d8>] (spi_drv_probe) from [<c037faa0>] (driver_probe_device+0x1a0/0x444)
[    1.527478] [<c037faa0>] (driver_probe_device) from [<c037fec8>] (__device_attach_driver+0x94/0xa0)
[    1.536507] [<c037fec8>] (__device_attach_driver) from [<c037db3c>] (bus_for_each_drv+0x94/0xa4)
[    1.545277] [<c037db3c>] (bus_for_each_drv) from [<c037f7e4>] (__device_attach+0xa4/0x144)
[    1.553526] [<c037f7e4>] (__device_attach) from [<c0380058>] (device_initial_probe+0x1c/0x20)
[    1.562035] [<c0380058>] (device_initial_probe) from [<c037ec88>] (bus_probe_device+0x38/0x94)
[    1.570631] [<c037ec88>] (bus_probe_device) from [<c037ccf4>] (device_add+0x430/0x558)
[    1.578534] [<c037ccf4>] (device_add) from [<c03d0240>] (spi_add_device+0xe4/0x174)
[    1.586178] [<c03d0240>] (spi_add_device) from [<c03d0a24>] (spi_register_master+0x698/0x7d4)
[    1.594688] [<c03d0a24>] (spi_register_master) from [<c03d0ba0>] (devm_spi_register_master+0x40/0x7c)
[    1.603892] [<c03d0ba0>] (devm_spi_register_master) from [<c03d2fb4>] (rockchip_spi_probe+0x360/0x3f4)
[    1.613182] [<c03d2fb4>] (rockchip_spi_probe) from [<c0381e34>] (platform_drv_probe+0x58/0xa8)
[    1.621779] [<c0381e34>] (platform_drv_probe) from [<c037faa0>] (driver_probe_device+0x1a0/0x444)
[    1.630635] [<c037faa0>] (driver_probe_device) from [<c037fdc4>] (__driver_attach+0x80/0xa4)
[    1.639058] [<c037fdc4>] (__driver_attach) from [<c037e850>] (bus_for_each_dev+0x98/0xac)
[    1.647221] [<c037e850>] (bus_for_each_dev) from [<c037f448>] (driver_attach+0x28/0x30)
[    1.655210] [<c037f448>] (driver_attach) from [<c037ef74>] (bus_add_driver+0x128/0x250)
[    1.663200] [<c037ef74>] (bus_add_driver) from [<c0380c40>] (driver_register+0xac/0xf0)
[    1.671191] [<c0380c40>] (driver_register) from [<c0381d50>] (__platform_driver_register+0x58/0x6c)
[    1.680221] [<c0381d50>] (__platform_driver_register) from [<c0a467c8>] (rockchip_spi_driver_init+0x18/0x20)
[    1.690033] [<c0a467c8>] (rockchip_spi_driver_init) from [<c00098a4>] (do_one_initcall+0x124/0x1dc)
[    1.699063] [<c00098a4>] (do_one_initcall) from [<c0a19f84>] (kernel_init_freeable+0x218/0x2ec)
[    1.707748] [<c0a19f84>] (kernel_init_freeable) from [<c0719ed8>] (kernel_init+0x1c/0xf4)
[    1.715912] [<c0719ed8>] (kernel_init) from [<c000fe50>] (ret_from_fork+0x14/0x24)
[    1.723460] Code: e3510000 159f67c0 0a00000c e5961000 (e5d13000)
[    1.729564] ---[ end trace 95baa6b3b861ce25 ]---

Fixes: 06bb6f5a ("mtd: spi-nor: stop (ab)using struct spi_device_id")
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Cc: Rafał Miłecki <zajec5@gmail.com>

This chip can be found on Hitex LPC4350 Evaluation Board.
Signed-off-by: NJoachim Eastwood <manabian@gmail.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

According to the datasheet[1], both S25FL129P0 (256kB sectors) and
S25FL129P1 (64kB sectors) support dual read, quad read, dual i/o read
and quad i/o read.

I have verified dual read to be working for S25FL129P1 on a dual
capable spi controller. Quad as well as S25FL129P0 is untested, lacking
hardware to verify.

[1] http://www.spansion.com/Support/Datasheets/S25FL129P_00.pdfSigned-off-by: NJonas Gorski <jogo@openwrt.org>
Reviewed-by: NMarek Vasut <marex@denx.de>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

As both omap2 onenand and omap2 nand driver modules are
named the same i.e. "omap2.ko", only one of them gets shipped
during MODPOST if both are configured as loadable modules.

To avoid this ambiguity let's ship the omap2 nand
driver as "omap2_nand.ko"

Reported by Pierre Neyron via github
https://github.com/beagleboard/linux/issues/40

Cc: Robert Nelson <robertcnelson@gmail.com>
Signed-off-by: NRoger Quadros <rogerq@ti.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

It is a 256KiB flash with 4 KiB erase sectors
and 64KiB overlay blocks.

This is the one available on Hardkernel's Odroid U3 shield.
Signed-off-by: NAlexis Ballier <aballier@gentoo.org>
[Brian: seems like this does NOT require the usual SST_WRITE hacks]
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Add Micron (n25q064a) 8MB flash to the list of supported chips.
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: NJagan Teki <jteki@openedev.com>
[Brian: fixup context]
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

When vary_offset is set (e.g. test case 3), the offset is not always
zero so memcmpshow() will show the wrong offset in the print message.
To fix this we introduce a new function memcmpshowoffset() which takes
offset as a parameter and displays the right offset and use it in
the case where offset is non zero.

The old memcmpshow() functionality is preserved by converting it into
a macro with offset preset to 0.
Signed-off-by: NRoger Quadros <rogerq@ti.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

With gcc 5.1 I get:
warning: switch condition has boolean value [-Wswitch-bool]
Signed-off-by: NTomer Barletz <barletz@gmail.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Using struct spi_device_id for storing list of flash devices comes from
early SPI NOR framework days. Thanks to the commit 70f3ce05 ("mtd:
spi-nor: make spi_nor_scan() take a chip type name, not spi_device_id")
we can stop using spi_device_id and just switch to our own struct.
Signed-off-by: NRafał Miłecki <zajec5@gmail.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Add the full description of the Toshiba TC58NVG0S3E NAND chip in the
nand_ids table so that we can later use the NAND ECC info and ONFI timing
mode in controller drivers.

Tested with asm9260_nand driver. [Brian: driver still under review]
Signed-off-by: NOleksij Rempel <linux@rempel-privat.de>
Reviewed-by: NBoris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

The SPI core currently reports the MODALIAS uevent as "spi:<modalias>"
even for SPI devices that were registered by OF.

That means the OF module alias exported by MODULE_OF_TABLE(of,...) is
currently not used and user-space has no way to autoload this module.

But it is still a good practice to add the OF module alias information
into the kernel module even when it currently is unused so once the SPI
core is changed to report a correct OF modalias uevent, module
autoloading will be working for this driver.
Signed-off-by: NJavier Martinez Canillas <javier@osg.samsung.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

The read ID count should be made as large as the maximum READ_ID size,
so there's no need to have dynamic size. This commit sets the hardware
maximum read ID count, which should be more than enough on all cases.
Also, we get rid of the read_id_bytes, and use a macro instead.
Signed-off-by: NEzequiel Garcia <ezequiel@vanguardiasur.com.ar>
Acked-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

When 2 commands are submitted in a row, and the second is very quick,
the completion of the second command might never come. This happens
especially if the second command is quick, such as a status read after
an erase.

The issue is that in the interrupt handler, the status bits are cleared
after the new command is issued. There is a small temporal window where
this happens :
 - the previous command has set the command done bit
 - the ready for a command bit is set
 - the handler submits the next command
   - just then, the command completes, and the command done bit is still
     set
 - the handler clears the "previous" command done bit
 - the handler exits

In this flow, the "command done" of the next command will never trigger
a new interrupt to finish the status command, as it was cleared for both
commands.

Fix this by clearing the status bit before submitting a new command.
Signed-off-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Acked-by: NEzequiel Garcia <ezequiel@vanguardiasur.com.ar>
Tested-by: NEzequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

When the nand is first probe, and upon the first command start, the
status bits should be cleared before the interrupts are unmasked.

The bug is tricky : if the bootloader left a status bit set, the
unmasking of interrupts does trigger the interrupt handler before the
first command is issued, blocking the good behavior of the nand.

The same would happen if in pxa3xx_nand code flow a status bit is left,
and then a command is started.
Signed-off-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Acked-by: NEzequiel Garcia <ezequiel@vanguardiasur.com.ar>
Tested-by: NEzequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

When keeping the configuration set by the bootloader (by using
the marvell,nand-keep-config property), the pxa3xx_nand_detect_config()
function is called and set the chunk size to 512 as a default value if
NDCR_PAGE_SZ is not set.

In the other case, when not keeping the bootloader configuration, no
chunk size is set. Fix this by adding a default chunk size of 512.

Fixes: 70ed8523 ("mtd: nand: pxa3xx: Introduce multiple page I/O
support")
Signed-off-by: NAntoine Tenart <antoine.tenart@free-electrons.com>
Acked-by: NRobert Jarzmik <robert.jarzmik@free>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Remove extra space after the "platform:" prefix and make the alias matches
driver name.
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Acked-by: NRoger Quadros <rogerq@ti.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

static analysis by smatch caught the following error:

drivers/mtd/maps/physmap_of.c:135 of_get_probes()
   error: potential null dereference 'res'.  (kzalloc returns null)

Check for failed kzalloc and return -ENOMEM in of_flash_probe if
this occurs.
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

If mtd_device_register() fails in nettel_init(), iomap left mapped.

The patch adds failure handling for mtd_device_register().

Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: NAlexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

It is better to have device name prefixed the actual error message.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

In recent kernels we have a lot of helper functions, including
devres API, to make life of device driver developer easy.

Convert the driver using devm_kzalloc() and pcim_enable_device().
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Let's use module_pci_driver() macro to reduce code base of the driver.

There is no functional change.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

There is no need to user to see the core part of the driver.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Spansion S25FL204K is a 4-Mbit 3.0V Serial Flash Memory
with Uniform 4 kB Sectors.
Signed-off-by: NAntony Pavlov <antonynpavlov@gmail.com>
Acked-by: NRafał Miłecki <zajec5@gmail.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

The current "We only connect the NOR to this controller now." text
is not very clear, so explain it better by saying that generic SPI
is not supported by SPI_FSL_QUADSPI and only SPI NOR is.
Signed-off-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Add SPI-NOR driver for the SPI Flash Interface (SPIFI)
controller that is found on newer NXP MCU devices.

The controller supports serial SPI Flash devices with 1-, 2-
and 4-bit width in either SPI mode 0 or 3. The controller
can operate in either command or memory mode. In memory mode
the Flash is exposed as normal memory and can be directly
accessed by the CPU.
Signed-off-by: NJoachim Eastwood <manabian@gmail.com>
Reviewed-by: NEzequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Add device tree binding documentation for the SPI Flash Interface
(SPIFI) found on NXP LPC18xx and LPC43xx devies.
Signed-off-by: NJoachim Eastwood <manabian@gmail.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

The initial buffer is used for the initial commands used to detect
a flash device (STATUS, READID and PARAM).

ONFI param page is 256 bytes, and there are three redundant copies
to be read. JEDEC param page is 512 bytes, and there are also three
redundant copies to be read. Hence this buffer should be at least
512 x 3. This commits rounds the buffer size to 2048.
Signed-off-by: NEzequiel Garcia <ezequiel@vanguardiasur.com.ar>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

While IS_ENABLED() is perfectly fine for CONFIG_* symbols, it is not
for other symbols such as __BIG_ENDIAN that is provided directly by
the compiler.

Switch to use CONFIG_CPU_BIG_ENDIAN instead of __BIG_ENDIAN.

Fixes: 27c5b17c ("mtd: nand: add NAND driver "library" for Broadcom STB NAND controller")
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Erase function will use cmd 0x20 (SPINOR_OP_BE_4K) if kenrel enable option
CONFIG_MTD_SPI_NOR_USE_4K_SECTORS.

This command is not in fsl-quadspi driver LUT. So driver continue report
fsl-quadspi 21e0000.qspi: Unsupported cmd 0x20.

This patch fix this issue.
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Acked-by: NHan Xu <Han.xu@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

The uboot may run the QuadSpi controler with command:
    #sf probe

So we should reset the module in the probe.
This patch also clear the pending interrupts which arised by the uboot
code.
Signed-off-by: NHuang Shijie <shijie8@gmail.com>
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

QSPI1 cannot wake up CCM from WAIT mode on SX ARD board, add pmqos to
let PM NOT enter WAIT mode when accessing QSPI1, refer to TKT245618.
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NHan Xu <Han.xu@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

We found there is a low probability(5%) QSPI access timeout issue,
usually it happened on kernel boot stage, the first time kernel tried to
access QSPI chip. The READ_ID command was sent but not executed,
consequently the probe function failed.

The root cause is that the divider is not glitchless in i.MX6SX chip.
If qspi clock enabled then change clock frequency by call clk_set_rate,
there will be glitch at low possiblity rate and pass to qspi controller.
The controler will be hang by this glitch.

Based on the new clock flag(CLK_SET_RATE_GATE) and new framework, we
need to change the approach of seting clock rate.
1. Disable clock.
2. call clk_set_rate.
3. Enable clock again.
Signed-off-by: NHan Xu <han.xu@freescale.com>
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

new compatible string: "fsl,imx6ul-qspi".
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Add i.mx6ul chip support
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Acked-by: NHan xu <han.xu@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

new compatible string: "fsl,imx7d-qspi"
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Support i.mx7d.
quadspi in i.mx7d increase rxfifo.
require fill at least 16byte to trigger data transfer.
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NHan Xu <han.xu@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

add several quirk to distinguish different version of qspi module.
Signed-off-by: NHan Xu <han.xu@freescale.com>
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

QSPI may failed to map enough memory (256MB) for AHB read in
previous implementation, especially in 3G/1G memory layout kernel.
Dynamically map memory to avoid such issue.

This implementation generally map QUADSPI_MAX_IOMAP (default 4MB) memory
for AHB read, it should be enough for common scenarios, and the side
effect (0.6% performance drop) is minor.

Previous implementation

root@imx6qdlsolo:~# dd if=/dev/mtd0 of=/dev/null bs=1K count=32K
32768+0 records in
32768+0 records out
33554432 bytes (34 MB) copied, 2.16006 s, 15.5 MB/s

root@imx6qdlsolo:~# dd if=/dev/mtd0 of=/dev/null bs=32M count=1
1+0 records in
1+0 records out
33554432 bytes (34 MB) copied, 1.43149 s, 23.4 MB/s

After applied the patch

root@imx6qdlsolo:~# dd if=/dev/mtd0 of=/dev/null bs=1K count=32K
32768+0 records in
32768+0 records out
33554432 bytes (34 MB) copied, 2.1743 s, 15.4 MB/s

root@imx6qdlsolo:~# dd if=/dev/mtd0 of=/dev/null bs=32M count=1
1+0 records in
1+0 records out
33554432 bytes (34 MB) copied, 1.43158 s, 23.4 MB/s
Signed-off-by: NHan Xu <han.xu@freescale.com>
Signed-off-by: NFrank Li <Frank.Li@freescale.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

When we added the "jedec,spi-nor" compatible string for use in this
driver, we added it as a modalias option. The modalias can be derived in
different ways for platform devices vs. device tree (of_*) matching. But
for device tree matching (the primary target of this identifier string),
the modalias is determined from the first entry in the 'compatible'
property. IOW, the following properties would bind to this driver:

	// Option (a), modalias = "spi-nor"
	compatible = "jedec,spi-nor";

	// Option (b), modalias = "spi-nor"
	compatible = "idontknowwhatimdoing,spi-nor";

But the following would not:

	// Option (c), modalias = "shinynewdevice"
	compatible = "myvendor,shinynewdevice", "jedec,spi-nor";

So, we'd like to match (a) and (c) (even when we don't have an explicit
entry for "shinynewdevice"), and we'd rather not allow (b).

To do this, we
  (1) always (for devices without specific platform data) pass the
      modalias to the spi-nor library;
  (2) rework the spi-nor library to not reject "bad" names, and
      instead just fall back to autodetection; and
  (3) add the .of_match_table to properly catch all "jedec,spi-nor".

This allows (a) and (c) without warnings, and rejects (b).
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Replaces the hard coded "always use EDO" policy with that prescribed
by the ONFI 3.1 specification that EDO mode should always be used if tRC
is below 30ns.
Signed-off-by: NRoy Spliet <r.spliet@ultimaker.com>
Acked-by: NBoris Brezillon <boris.brezillon@free-electrons.com>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>