Even if the chip select line is not controlled by the SPFI
hardware, the device select bits need to be set to specify
the chip select line in use for the hardware to know what
parameters to use for the current transfer.
Signed-off-by: Ionela Voinescu <ionela.voinescu@imgtec.com>
Signed-off-by: Andreas Färber <afaerber@suse.de>
Signed-off-by: Mark Brown <broonie@kernel.org>

Variable count, l, mcspi and spi_cntrl are being assigned but are
never used hence they are redundant and can be removed.

Cleans up clang warnings:
warning: variable 'count' set but not used [-Wunused-but-set-variable]
warning: variable 'l' set but not used [-Wunused-but-set-variable]
warning: variable 'mcspi' set but not used [-Wunused-but-set-variable]
warning: variable 'spi_cntrl' set but not used [-Wunused-but-set-variable]
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Because the SPI controller deasserts the chip select when the TX fifo is
empty (which may happen in the middle of a transfer), the CS should be
handled by linux. Unfortunately, some or all of the first four chip
selects are not muxable as GPIOs, depending on the SoC.

There is a way to bitbang those pins by using the SPI boot controller so
use it to set the chip selects.

At init time, it is also necessary to give control of the SPI interface to
the Designware IP.
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Export dw_spi_set_cs so it can be used from the various IP integration
modules.
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Log RX and TX fifo counters when a transfer is done and these are not
zero.
Signed-off-by: Tiago Brusamarello <tiago.brusamarello@datacom.ind.br>
Signed-off-by: Mark Brown <broonie@kernel.org>

Dynamic burst mode allows to group together multiple words and send them
in one continuous burst. When the number of bytes to be sent is not a
strict multiple of the FIFO entry size (32 bits), the controller expects
the non aligned bits to be sent first.

This commit adds support for this particular constraint, avoiding the
need to send the non-aligned bytes one by one at the end of the
transfer, speeding-up transfer speed in that case.

With this method, a transfer is divided into multiple bursts, limited in
size by the maximum amount of data that the controller can transfer in
one continuous burst (which is 512 bytes).

The non-512 byte part of the transfer is sent first. The remaining bytes
to be transferred in the current burst is stored in the 'remainder'
field.

With this method, the read_u32 field is no longer necessary, and is
removed.

This was tested on imx6 solo and imx6 quad.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

The spi_imx_can_dma function computes the watermark level so that
the transfer will fit in exactly N bursts (without a remainder).

The smallest watermark level possible being one FIFO entry per burst, we
can't never have a case where the transfer size isn't divsiible by 1.

Remove the extra check for the wml being different than 0.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

The SPI core enforces that we always use the next power-of-two number of
bytes to store words. As a result, a 24 bits word will be stored in 4
bytes.

This commit fixes the spi_imx_bytes_per_word function to return the
correct number of bytes.

This also allows to get rid of unnecessary checks in the can_dma
function, since the SPI core validates that we always have a transfer
length that is a multiple of the number of bytes per word.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

The dynamic bursts mode allows to group together multiple words into a
single burst. To do so, it's necessary that words can be packed into the
32-bits FIFO entries, so we can't allow using this mode with bit_per_words
different to 8, 16 or 32.

This prevents shitfing out extra clock ticks for transfers with
bit_per_word values not aligned on 8 bits.

With that , we are sure that only the correct number of bits is
shifted out at each transfer, so we don't need to mask out the remaining
parts of the words.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Some fields in struct spi_imx_data are assigned a different value twice
in a row, in spi_imx_setupxfer.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Adopt the SPDX license identifier headers to ease license compliance
management.
Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

This changes the data type of the flags field in struct spi_bitbang from
u8 to u16. This matches the size of the mode field of struct spi_device
where these flags are also used.

This is done in preparation of adding a new SPI mode flag that will be
used with this field that would otherwise not fit in 8 bits.
Signed-off-by: David Lechner <david@lechnology.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Merge tag 'spi-dw-set-cs' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi into spi-4.19

spi: dw: Allow custom set_cs_callback

Allow platform specific drivers to provide their own set_cs callback when
the IP integration requires it.

Allow platform specific drivers to provide their own set_cs callback when
the IP integration requires it.
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Upper layer users of SPI device drivers may rely on 'actual_length',
so it is important that information is correctly reported. One such
example is spi_mem_exec_op() function that will fail if
'actual_length' of the data transferred is not what was requested. Add
necessary code to populate 'actual_length.

Cc: Mark Brown <broonie@kernel.org>
Cc: Sanchayan Maity <maitysanchayan@gmail.com>
Cc: Stefan Agner <stefan@agner.ch>
Cc: cphealy@gmail.com
Cc: linux-spi@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Andrey Smirnov <andrew.smirnov@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Add "rockchip,px30-spi", "rockchip,rk3066-spi" for spi on px30 platform.
Signed-off-by: Liang Chen <cl@rock-chips.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Registers of DSPI should not be accessed before enabling its clock.  On
Toradex Colibri VF50 on Iris carrier board this could be seen during
bootup as imprecise abort:

    Unhandled fault: imprecise external abort (0x1c06) at 0x00000000
    Internal error: : 1c06 [#1] ARM
    Modules linked in:
    CPU: 0 PID: 1 Comm: swapper Not tainted 4.14.39-dirty #97
    Hardware name: Freescale Vybrid VF5xx/VF6xx (Device Tree)
    Backtrace:
    [<804166a8>] (regmap_write) from [<80466b5c>] (dspi_probe+0x1f0/0x8dc)
    [<8046696c>] (dspi_probe) from [<8040107c>] (platform_drv_probe+0x54/0xb8)
    [<80401028>] (platform_drv_probe) from [<803ff53c>] (driver_probe_device+0x280/0x2f8)
    [<803ff2bc>] (driver_probe_device) from [<803ff674>] (__driver_attach+0xc0/0xc4)
    [<803ff5b4>] (__driver_attach) from [<803fd818>] (bus_for_each_dev+0x70/0xa4)
    [<803fd7a8>] (bus_for_each_dev) from [<803fee74>] (driver_attach+0x24/0x28)
    [<803fee50>] (driver_attach) from [<803fe980>] (bus_add_driver+0x1a0/0x218)
    [<803fe7e0>] (bus_add_driver) from [<803fffe8>] (driver_register+0x80/0x100)
    [<803fff68>] (driver_register) from [<80400fdc>] (__platform_driver_register+0x48/0x50)
    [<80400f94>] (__platform_driver_register) from [<8091cf7c>] (fsl_dspi_driver_init+0x1c/0x20)
    [<8091cf60>] (fsl_dspi_driver_init) from [<8010195c>] (do_one_initcall+0x4c/0x174)
    [<80101910>] (do_one_initcall) from [<80900e8c>] (kernel_init_freeable+0x144/0x1d8)
    [<80900d48>] (kernel_init_freeable) from [<805ff6a8>] (kernel_init+0x10/0x114)
    [<805ff698>] (kernel_init) from [<80107be8>] (ret_from_fork+0x14/0x2c)

Cc: <stable@vger.kernel.org>
Fixes: 5ee67b58 ("spi: dspi: clear SPI_SR before enable interrupt")
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Mark Brown <broonie@kernel.org>

Intel Ice Lake SPI host controller follows the Intel Cannon Lake but the
PCI IDs are different. Add the new PCI IDs to the driver supported
devices list.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org

It seems that the proper structure field to use in this particular
case is *regmap_pushr* instead of regmap.

Addresses-Coverity-ID: 1470126 ("Copy-paste error")
Fixes: 58ba07ec ("spi: spi-fsl-dspi: Add support for XSPI mode registers")
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Acked-by: Esben Haabendal <eha@deif.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Set the XSPI bit for devices configured for XSPI mode (currently LS1021A),
and thereby switch to extended SPI mode, allowing for SPI transfers using
from 4 to 32 bits per word instead of 4 to 16 bits per word.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

This implements handling of split CMD and TX FIFO queues for XSPI when
running in TCFQ mode.

It should be simple to add it to EOQ mode also.  Currently, EOQ mode is
only used with coldfire.  So if coldfire DSPI supports XSPI, XSPI FIFO
handling should be added to EOQ mode also.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

This prepares for adding support for extended SPI mode (XSPI), by extending
the regmap with the extra SREX and CTAREx registers.

An additional register map is made for allowing 16 bit access to CMD and TX
FIFO of the PUSHR register separately, which is also needed for XSPI mode
support.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Mark volatile registers to avoid caching bugs.

Note: SPI_MCR is marked volatile because of CLR_TXF and CLR_RXF bits.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

The MCR register is not changed, so initialize it in dspi_init().

The exception is the CLR_TXF and CLR_RXF bits, which should be written to
before each transfer to make sure we start with empty FIFOs.  With MCR
register now configured as volatile, the regmap_update_bits will do a real
read-modify-write cycle.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

This extends the driver with support for all SPI framesizes from 4 to 16
bits, and adds support for per transfer specific bits_per_word, while at
the same time reducing code size and complexity.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Simplify driver by avoiding counter wrapping by clearing transfer counter
on first SPI transfer per interrupt instead of tracking what it was before.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

As of 92dc20d8, transfer->cs_change has
been supported for non-last transfers, but not for last transfer.

This change brings handling of cs_change in line with the specification in
spi.h, implementing handling of transfer->cs_change for all transfers.

The value for CMD FIFO is precalculated with transfer->cs_change field
taken into account, allowing for CS de-activate between transfers and
keeping CS activated after last transfer.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Checking directly against pointer value should be at least as fast as doing
bitmasking and compare, so let's keep it simple.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

The if statement just above this if/else statement triggers on the same
condition, and then invalidates it.
Signed-off-by: Esben Haabendal <eha@deif.com>
Acked-by: Martin Hundebøll <martin@geanix.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

The code did not de-assert any CS GPIOs before probing slaves. This
means that several CS signals could be active at once, garbling the
communication. Whether this was actually a problem depended on the type
of the SPI device attached (so my "spidev" for userspace access worked
correctly because its probe was effectively a no-op), and on the state
of the GPIO pins at SoC's boot.

The code was already iterating through all DT children of the SPI
controller, so this change re-uses that loop for CS GPIO setup as well.
This means that this might change the number of the HW CS signal which
is picked for all GPIO CS devices. Previously, the lowest one was used,
but we now use the first one from the DT.

With this move of the code, we can also finally initialize each GPIO CS
lane before registering the SPI controller (which in turn probes for
slaves).

I tried to fix this in 54424862 already, but that only did it half
way by registering the GPIOs properly. That patch failed to set their
logic signals early enough, though.
Signed-off-by: Jan Kundrát <jan.kundrat@cesnet.cz>
Signed-off-by: Mark Brown <broonie@kernel.org>

The path "spi: cadence: Add usleep_range() for
cdns_spi_fill_tx_fifo()" added a usleep_range() function call,
which cannot be used in atomic context.
However the cdns_spi_fill_tx_fifo() function can be called during
an interrupt which may result in a kernel panic:

BUG: scheduling while atomic: grep/561/0x00010002
Modules linked in:
Preemption disabled at:
[<ffffff800858ea28>] wait_for_common+0x48/0x178
CPU: 0 PID: 561 Comm: grep Not tainted 4.17.0 #1
Hardware name: Cadence CSP (DT)
Call trace:
 dump_backtrace+0x0/0x198
 show_stack+0x14/0x20
 dump_stack+0x8c/0xac
 __schedule_bug+0x6c/0xb8
 __schedule+0x570/0x5d8
 schedule+0x34/0x98
 schedule_hrtimeout_range_clock+0x98/0x110
 schedule_hrtimeout_range+0x10/0x18
 usleep_range+0x64/0x98
 cdns_spi_fill_tx_fifo+0x70/0xb0
 cdns_spi_irq+0xd0/0xe0
 __handle_irq_event_percpu+0x9c/0x128
 handle_irq_event_percpu+0x34/0x88
 handle_irq_event+0x48/0x78
 handle_fasteoi_irq+0xbc/0x1b0
 generic_handle_irq+0x24/0x38
 __handle_domain_irq+0x84/0xf8
 gic_handle_irq+0xc4/0x180

This patch replaces the function call with udelay() which can be
used in an atomic context, like an interrupt.
Signed-off-by: Jan Kotas <jank@cadence.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org

In case of a bi-directional transfer, receive DMA may complete in the
rcar-dmac driver before transmit DMA, due to scheduling latencies.
As the MSIOF driver waits for completion of the receive DMA only, it may
submit the next transmit DMA request before the previous one has
completed.

Make the driver more robust by waiting for the completion of both
receive and transmit DMA, when applicable.

Based on a patch in the BSP by Ryo Kataoka.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Mark Brown <broonie@kernel.org>

include/linux/spi/adi_spi3.h is unused since commit 47838669 ("spi: remove blackfin related host drivers")
Finish the cleaning by removing it.
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>

Pull block fixes from Jens Axboe:
 "A collection of fixes that should go into -rc1. This contains:

   - bsg_open vs bsg_unregister race fix (Anatoliy)

   - NVMe pull request from Christoph, with fixes for regressions in
     this window, FC connect/reconnect path code unification, and a
     trace point addition.

   - timeout fix (Christoph)

   - remove a few unused functions (Christoph)

   - blk-mq tag_set reinit fix (Roman)"

* tag 'for-linus-20180616' of git://git.kernel.dk/linux-block:
  bsg: fix race of bsg_open and bsg_unregister
  block: remov blk_queue_invalidate_tags
  nvme-fabrics: fix and refine state checks in __nvmf_check_ready
  nvme-fabrics: handle the admin-only case properly in nvmf_check_ready
  nvme-fabrics: refactor queue ready check
  blk-mq: remove blk_mq_tagset_iter
  nvme: remove nvme_reinit_tagset
  nvme-fc: fix nulling of queue data on reconnect
  nvme-fc: remove reinit_request routine
  blk-mq: don't time out requests again that are in the timeout handler
  nvme-fc: change controllers first connect to use reconnect path
  nvme: don't rely on the changed namespace list log
  nvmet: free smart-log buffer after use
  nvme-rdma: fix error flow during mapping request data
  nvme: add bio remapping tracepoint
  nvme: fix NULL pointer dereference in nvme_init_subsystem
  blk-mq: reinit q->tag_set_list entry only after grace period

Pull documentation fixes from Mauro Carvalho Chehab:
 "This solves a series of broken links for files under Documentation,
  and improves a script meant to detect such broken links (see
  scripts/documentation-file-ref-check).

  The changes on this series are:

   - can.rst: fix a footnote reference;

   - crypto_engine.rst: Fix two parsing warnings;

   - Fix a lot of broken references to Documentation/*;

   - improve the scripts/documentation-file-ref-check script, in order
     to help detecting/fixing broken references, preventing
     false-positives.

  After this patch series, only 33 broken references to doc files are
  detected by scripts/documentation-file-ref-check"

* tag 'docs-broken-links' of git://linuxtv.org/mchehab/experimental: (26 commits)
  fix a series of Documentation/ broken file name references
  Documentation: rstFlatTable.py: fix a broken reference
  ABI: sysfs-devices-system-cpu: remove a broken reference
  devicetree: fix a series of wrong file references
  devicetree: fix name of pinctrl-bindings.txt
  devicetree: fix some bindings file names
  MAINTAINERS: fix location of DT npcm files
  MAINTAINERS: fix location of some display DT bindings
  kernel-parameters.txt: fix pointers to sound parameters
  bindings: nvmem/zii: Fix location of nvmem.txt
  docs: Fix more broken references
  scripts/documentation-file-ref-check: check tools/*/Documentation
  scripts/documentation-file-ref-check: get rid of false-positives
  scripts/documentation-file-ref-check: hint: dash or underline
  scripts/documentation-file-ref-check: add a fix logic for DT
  scripts/documentation-file-ref-check: accept more wildcards at filenames
  scripts/documentation-file-ref-check: fix help message
  media: max2175: fix location of driver's companion documentation
  media: v4l: fix broken video4linux docs locations
  media: dvb: point to the location of the old README.dvb-usb file
  ...

Pull fsnotify updates from Jan Kara:
 "fsnotify cleanups unifying handling of different watch types.

  This is the shortened fsnotify series from Amir with the last five
  patches pulled out. Amir has modified those patches to not change
  struct inode but obviously it's too late for those to go into this
  merge window"

* tag 'fsnotify_for_v4.18-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  fsnotify: add fsnotify_add_inode_mark() wrappers
  fanotify: generalize fanotify_should_send_event()
  fsnotify: generalize send_to_group()
  fsnotify: generalize iteration of marks by object type
  fsnotify: introduce marks iteration helpers
  fsnotify: remove redundant arguments to handle_event()
  fsnotify: use type id to identify connector object type