DMA submit should just add the dma descriptor to a queue, without firing
it. EIO is misleading and hides what happens in DMA. Propagate the
dma_submit_error() error code, don't overwrite it.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-12-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Check the return code of dmaengine_slave_config().
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-11-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Move common code outside if else to avoid code duplication.
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-10-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Else case is not needed after a return, remove it.
Reviewed-by: Andi Shyti <andi.shyti@kernel.org>
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-9-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

of_device_id::data is an opaque pointer. No explicit cast is needed.
Remove unneeded (void *) casts in of_match_table.
Reviewed-by: Andi Shyti <andi.shyti@kernel.org>
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-8-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Fix typo, s/configuartion/configuration.
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-7-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

The platform id is used as an index into the fifo_lvl_mask array.
Platforms can come with a negative device ID, PLATFORM_DEVID_NONE (-1),
thus we risked a negative array index. Catch such cases and fail to
probe.
Reviewed-by: Sam Protsenko <semen.protsenko@linaro.org>
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-6-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

The driver uses u32 and relies on an implicit inclusion of
<linux/types.h>.

It is good practice to directly include all headers used, it avoids
implicit dependencies and spurious breakage if someone rearranges
headers and causes the implicit include to vanish.

Include the missing header.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-5-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

The driver uses GENMASK() but does not include <linux/bits.h>.

It is good practice to directly include all headers used, it avoids
implicit dependencies and spurious breakage if someone rearranges
headers and causes the implicit include to vanish.

Include the missing header.
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-4-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

The driver uses readl() but does not include <linux/io.h>.

It is good practice to directly include all headers used, it avoids
implicit dependencies and spurious breakage if someone rearranges
headers and causes the implicit include to vanish.

Include the missing header.
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-3-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Sorting headers alphabetically helps locating duplicates,
and makes it easier to figure out where to insert new headers.
Reviewed-by: Andi Shyti <andi.shyti@kernel.org>
Reviewed-by: Peter Griffin <peter.griffin@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20240207120431.2766269-2-tudor.ambarus@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

As a general principal, it is best to do as little as possible in an
interrupt handler. This patch reworks the AXI SPI Engine driver to move
timer_delete_sync() and spi_finalize_current_message() out of the
interrupt handler. Instead, spi_finalize_current_message() is moved to
the transfer_one_message function (similar to nearly all other SPI
controllers). A completion is now used to wait for the sync interrupt
that indicates that the message is complete. The watchdog timer is no
longer needed since we can use the wait_for_completion_timeout()
function to wait for the message to complete with the same effect.

As a bonus, these changes also improve throughput of the SPI controller.
For example, this was tested on a ZynqMP with a 80MHz SCLK reading 4
byte samples from an ADC. The max measured throughput increased from
26k to 28k samples per second.
Reviewed-by: Nuno Sa <nuno.sa@analog.com>
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://lore.kernel.org/r/20240207-axi-spi-engine-round-2-1-v2-2-40c0b4e85352@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

Profiling has shown that ida_alloc_range() accounts for about 10% of the
time spent in spi_sync() when using the AXI SPI Engine controller. This
call is used to create a unique id for each SPI message to match to an
IRQ when the message is complete.

Since the core SPI code serializes messages in a message queue, we can
only have one message in flight at a time, namely host->cur_msg. This
means that we can use a fixed value instead of a unique id for each
message since there can never be more than one message pending at a
time.

This patch removes the use of ida for the sync id and replaces it with a
constant value. This simplifies the driver and improves performance.
Reviewed-by: Nuno Sa <nuno.sa@analog.com>
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://lore.kernel.org/r/20240207-axi-spi-engine-round-2-1-v2-1-40c0b4e85352@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

The __spi_split_transfer_maxsize() function has a gpf argument to allow
callers to specify the type of memory allocation that needs to be used.
However, this function only allocates struct spi_transfer and is not
intended to be used from atomic contexts so this type should always be
GFP_KERNEL, so we can just drop the argument.

Some callers of these functions also passed GFP_DMA, but since only
struct spi_transfer is allocated and not any tx/rx buffers, this is
not actually necessary and is removed in this commit.
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://lore.kernel.org/r/20240206200648.1782234-1-dlechner@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

Merge series from andy.shevchenko@gmail.com:

A couple of error handling improvements here:
- unshadowing error code from dmaengine_slave_config()
- making error messages uniform

Signed-off-by: Luis de Arquer <luis.dearquer@inertim.com>
Link: https://lore.kernel.org/r/4d18808e85b85077761c5655083f20ebfd7d3770.camel@gmail.comSigned-off-by: Mark Brown <broonie@kernel.org>

Since spi-rockchip enables use_gpio_descriptors and the
SPI_CONTROLLER_GPIO_SS flag, the spi subsytem may call set_cs()
for spi devices with indexes above ROCKCHIP_SPI_MAX_CS_NUM

Remove array cs_asserted[] which held a shadow copy of the state
of the chip select lines with the only purpose of optimizing out
rewriting a chip select line to the current state (no-op)
This case is already handled by spi.c
Signed-off-by: Luis de Arquer <luis.dearquer@inertim.com>
Link: https://lore.kernel.org/r/d0a0c4b94f933f7f43973c34765214303ee82b77.camel@gmail.comSigned-off-by: Mark Brown <broonie@kernel.org>

If there are two flash chips connected flash regions can refer to the
second chip too. In this case we may see the following warning:

  mtd: partition "BIOS" extends beyond the end of device "0000:00:1f.5" --
  	size truncated to 0x400000

For this reason, check the BIOS partition size against the chip size and
make sure it stays within the that.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Link: https://lore.kernel.org/r/20240201121638.207632-2-mika.westerberg@linux.intel.comSigned-off-by: Mark Brown <broonie@kernel.org>

This should make it easier to identify the second chip and also allows
using "mtdparts=" and the like with this chip too.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Link: https://lore.kernel.org/r/20240201121638.207632-1-mika.westerberg@linux.intel.comSigned-off-by: Mark Brown <broonie@kernel.org>

This avoid duplicating the same macros in multiple drivers by reusing
the common AXI macros for the version register.
Signed-off-by: David Lechner <dlechner@baylibre.com>
Reviewed-by: Nuno Sa <nuno.sa@analog.com>
Link: https://lore.kernel.org/r/20240202213132.3863124-2-dlechner@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

The core SPI code will handle splitting transfers if needed as long
as ctlr->max_transfer_size is implemented. It does this in
__spi_pump_transfer_message() immediately before calling
ctlr->prepare_message. So effectively, this change does not
alter the behavior of the driver.

Also, several peripheral drivers make use of spi_max_transfer_size(),
so this should improve compatibility with those drivers.
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://lore.kernel.org/r/20240126220024.3926403-2-dlechner@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

This moves splitting transfers for CS_WORD software emulation to the
same place where we split transfers for controller-specific reasons.

This fixes a few subtle bugs.

The calculation for maxsize was wrong for bit sizes between 17 and 24.
This is fixed by making use of spi_split_transfers_maxwords() which
already has the correct calculation.

Also, since this indirectly calls spi_res_alloc(), to avoid leaking
resources, spi_finalize_current_message() would need to be called
on all error paths in __spi_validate() and callers of __spi_validate()
would need to do the same. This is fixed by moving the call to
__spi_pump_transfer_message() where it is already splitting transfers
for other reasons and correctly releases resources in the subsequent
error paths.

Fixes: cbaa62e0 ("spi: add software implementation for SPI_CS_WORD")
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://lore.kernel.org/r/20240126212358.3916280-2-dlechner@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

Use dev_err_probe() for all messages in dspi_request_dma() for the sake of
making them uniform. While at it, fix indentation issue reported by Vladimir
Oltean.
Signed-off-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Link: https://lore.kernel.org/r/20240204203127.1186621-3-andy.shevchenko@gmail.comSigned-off-by: Mark Brown <broonie@kernel.org>

dmaengine_slave_config() may return different error codes based on
the circumstances. Preserve it instead of shadowing to -EINVAL.
Suggested-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Link: https://lore.kernel.org/r/20240204203127.1186621-2-andy.shevchenko@gmail.comSigned-off-by: Mark Brown <broonie@kernel.org>

As we get a child node in the OF case, we should also clean up the
reference, add code to do so.
Suggested-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://msgid.link/r/20240202103430.951598-1-ckeepax@opensource.cirrus.comSigned-off-by: Mark Brown <broonie@kernel.org>

The members of `struct spi_message` were reordered in commit
ae2ade4b ("spi: Reorder fields in 'struct spi_message'")
but the documentation comments were not updated to match.

This commit updates the comments to match the new order.
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://msgid.link/r/20240131170732.1665105-1-dlechner@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

Signed-off-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Link: https://msgid.link/r/lq6gstev3sd7i4iw2btiq3gg7lhsraj5w74fkbp6lpbl6nkyff@tarta.nabijaczleweli.xyzSigned-off-by: Mark Brown <broonie@kernel.org>

The __spi_sync() function calls __spi_validate() early in the function.
Later, it can call spi_async_locked() which calls __spi_validate()
again. __spi_validate() is an expensive function, so we can improve
performance measurably by avoiding calling it twice.

Instead of calling spi_async_locked(), we can call __spi_async() with
the spin lock held.

spi_async_locked() is removed since there are no more callers.
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://msgid.link/r/20240125234732.3530278-2-dlechner@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

As it devm_pm_runtime_enable() can fail due to memory allocations, it
is best to handle the error.
Suggested-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Link: https://msgid.link/r/20240125103426.2622549-1-ckeepax@opensource.cirrus.comSigned-off-by: Mark Brown <broonie@kernel.org>

Merge series from Sam Protsenko <semen.protsenko@linaro.org>:

This series enables SPI for Exynos850 SoC, there are also some
dependencies that will be needed at runtime which were sent as part of
the same series but will be separately applied:

  1. Enable PDMA, it's needed for SPI (dts, clk)
  2. Propagate SPI src clock rate change up to DIV clocks, to make it
     possible to change SPI frequency (clk driver)
  4. Add SPI nodes to Exynos850 SoC dtsi

All SPI instances were tested using `spidev_test' tool in all 3 possible
modes:

  - Polling mode: xfer_size <= 32
  - IRQ mode: 64 >= xfer_size >= 32
  - DMA mode: xfer_size > 64

with 200 kHz ... 49.9 MHz SPI frequencies. The next 3 approaches were
used:

  1. Software loopback ('-l' option for `spidev_test' tool)
  2. Hardware loopback (by connecting MISO line to MOSI)
  3. By communicating with ATMega found on Sensors Mezzanine board [1],
     programmed to act as an SPI device

and all the transactions were additionally checked on my Logic Analyzer
to make sure the SCK frequencies were actually correct.

[1] https://www.96boards.org/product/sensors-mezzanine/

Previously, __spi_sync() and __spi_async() set message->spi to the spi
device independently after calling __spi_validate(). __spi_validate()
also would conditionally set this if it needed to split the message
since it wasn't set yet.

Since both __spi_sync() and __spi_async() call __spi_validate(), we can
consolidate this into only setting message->spi once (unconditionally)
in __spi_validate(). This will also save any future callers of
__spi_validate() from also needing to set message->spi.
Signed-off-by: David Lechner <dlechner@baylibre.com>
Link: https://msgid.link/r/20240123214946.2616786-1-dlechner@baylibre.comSigned-off-by: Mark Brown <broonie@kernel.org>

Document samsung,exynos850-spi compatible which will be used on
Exynos850 SoC. Exynos850 doesn't have ioclk, so only two clocks are
needed (bus clock and functional SPI clock).
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Reviewed-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://msgid.link/r/20240120012948.8836-3-semen.protsenko@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Add SPI port configuration for Exynos850 SoC. It has 3 USI blocks which
can be configured in SPI mode:

  * spi_0: BLK_PERI_SPI_0 (0x13940000)
  * spi_1: BLK_ALIVE_USI_CMGP00 (0x11d00000)
  * spi_2: BLK_ALIVE_USI_CMGP01 (0x11d20000)

SPI FIFO depth is 64 bytes for all those SPI blocks, so the
.fifo_lvl_mask value is set to 0x7f. All blocks have DIV_4 as the
default internal clock divider, and an internal loopback mode to run
a loopback test.
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Reviewed-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://msgid.link/r/20240120012948.8836-6-semen.protsenko@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Add i.MX93/95 flexspi compatible strings, which are compatible with
i.MX8MM
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Acked-by: Han Xu <han.xu@nxp.com>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://msgid.link/r/20240122091510.2077498-2-peng.fan@oss.nxp.comSigned-off-by: Mark Brown <broonie@kernel.org>

Add i.MX95 LPSPI compatible string, same as i.MX93 compatible with
i.MX7ULP
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://msgid.link/r/20240122091510.2077498-1-peng.fan@oss.nxp.comSigned-off-by: Mark Brown <broonie@kernel.org>

Simplify the code by extracting all cases of FIFO depth calculation into
a dedicated macro. No functional change.
Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Reviewed-by: Andi Shyti <andi.shyti@kernel.org>
Link: https://msgid.link/r/20240120170001.3356-1-semen.protsenko@linaro.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Ensure the command value and LUT entry values have a fixed width. This
way consecutive output lines can be read much easier.
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Link: https://msgid.link/r/20240118121016.3734770-1-alexander.stein@ew.tq-group.comSigned-off-by: Mark Brown <broonie@kernel.org>

Now that the driver core can properly handle constant struct bus_type,
move the spi_bus_type variable to be a constant structure as well,
placing it into read-only memory which can not be modified at runtime.

Cc: Mark Brown <broonie@kernel.org>
Cc:  <linux-spi@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://msgid.link/r/2024010549-erasure-swoop-1cc6@gregkhSigned-off-by: Mark Brown <broonie@kernel.org>

Supports configuring sleep pin control during system suspend to prevent
potential power leakage and additional power consumption.
Signed-off-by: Ruihai Zhou <zhouruihai@huaqin.corp-partner.google.com>
Link: https://msgid.link/r/20240108120802.7601-1-zhouruihai@huaqin.corp-partner.google.comSigned-off-by: Mark Brown <broonie@kernel.org>