linux-2.6-microblaze.git
3 years agospi: dw: Perform IRQ setup in a dedicated function
Serge Semin [Wed, 7 Oct 2020 23:54:59 +0000 (02:54 +0300)]
spi: dw: Perform IRQ setup in a dedicated function

In order to make the transfer_one() callback method more readable and
for unification with the DMA-based transfer, let's detach the IRQ setup
procedure into a dedicated function. While at it rename the IRQ-based
transfer handler function to be dw_spi-prefixe and looking more like the
DMA-related one.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-11-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Refactor IRQ-based SPI transfer procedure
Serge Semin [Wed, 7 Oct 2020 23:54:58 +0000 (02:54 +0300)]
spi: dw: Refactor IRQ-based SPI transfer procedure

Current IRQ-based SPI transfer execution procedure doesn't work well at
the final stage of the execution. If all the Tx data is sent out (written
to the Tx FIFO) but there is some data left to receive, the Tx FIFO Empty
IRQ will constantly happen until all of the requested inbound data is
received. Though for a short period of time, but it will make the system
less responsive. In order to fix that let's refactor the SPI transfer
execution procedure by taking the Rx FIFO Full IRQ into account. We'll read
and write SPI transfer data each time the IRQ happens as before. If all
the outbound data is sent out, we'll disable the Tx FIFO Empty IRQ. If
there is still some data to receive, we'll adjust the Rx FIFO Threshold
level, so the next IRQ would be raised at the moment of all incoming data
being available in the Rx FIFO.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-10-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Refactor data IO procedure
Serge Semin [Wed, 7 Oct 2020 23:54:57 +0000 (02:54 +0300)]
spi: dw: Refactor data IO procedure

The Tx and Rx data write/read procedure can be significantly simplified by
using Tx/Rx transfer lengths instead of the end pointers. By having the
Tx/Rx data leftover lengths (in the number of transfer words) we can get
rid of all subtraction and division operations utilized here and there in
the tx_max(), rx_max(), dw_writer() and dw_reader() methods. Such
modification will not only give us the more optimized IO procedures, but
will make the data IO methods much more readable than before.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-9-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Add DW SPI controller config structure
Serge Semin [Wed, 7 Oct 2020 23:54:56 +0000 (02:54 +0300)]
spi: dw: Add DW SPI controller config structure

DW APB SSI controller can be used by the two SPI core interfaces:
traditional SPI transfers and SPI memory operations. The controller needs
to be accordingly configured at runtime when the corresponding operations
are executed. In order to do that for the both interfaces from a single
function we introduce a new data wrapper for the transfer mode, data
width, number of data frames (for the automatic data transfer) and the bus
frequency. It will be used by the update_config() method to tune the DW
APB SSI up.

The update_config() method is made exported to be used not only by the DW
SPI core driver, but by the glue layer drivers too. This will be required
in a coming further commit.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-8-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Update Rx sample delay in the config function
Serge Semin [Wed, 7 Oct 2020 23:54:55 +0000 (02:54 +0300)]
spi: dw: Update Rx sample delay in the config function

Rx sample delay can be SPI device specific, and should be synchronously
initialized with the rest of the communication and peripheral device
related controller setups. So let's move the Rx-sample delay setup into
the DW APB SSI configuration update method.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-7-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Simplify the SPI bus speed config procedure
Serge Semin [Wed, 7 Oct 2020 23:54:54 +0000 (02:54 +0300)]
spi: dw: Simplify the SPI bus speed config procedure

The code currently responsible for the SPI communication speed setting up
is a bit messy. Most likely for some historical reason the bus frequency
is saved in the peripheral chip private data. It's pointless now since the
custom communication speed is a SPI-transfer-specific thing and only if
there is no SPI transfer data specified (like during the SPI memory
operations) it can be taken from the SPI device structure. But even in the
later case there is no point in having the clock divider and the SPI bus
frequency saved in the chip data, because the controller can be used for
both SPI-transfer-based and SPI-transfer-less communications. From
software point of view keeping the current clock divider in an SPI-device
specific storage may give a small performance gain (to avoid sometimes a
round-up division), but in comparison to the total SPI transfer time it
just doesn't worth saving a few CPU cycles in comparison to the total SPI
transfer time while having the harder to read code. The only optimization,
which could worth preserving in the code is to avoid unnecessary DW SPI
controller registers update if it's possible. So to speak let's simplify
the SPI communication speed update procedure by removing the clock-related
fields from the peripheral chip data and update the DW SPI clock divider
only if it's really changed. The later change is reached by keeping the
effective SPI bus speed in the internal DW SPI private data.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-6-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Update SPI bus speed in a config function
Serge Semin [Wed, 7 Oct 2020 23:54:53 +0000 (02:54 +0300)]
spi: dw: Update SPI bus speed in a config function

The SPI bus speed update functionality will be useful in another parts of
the driver too (like to implement the SPI memory operations and from the
DW SPI glue layers). Let's move it to the update_cr0() method then and
since the later is now updating not only the CTRLR0 register alter its
prototype to have a generic function name not related to CR0.

Leave the too long line with the chip->clk_div setting as is for now,
since it's going to be changed later anyway.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-5-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Detach SPI device specific CR0 config method
Serge Semin [Wed, 7 Oct 2020 23:54:52 +0000 (02:54 +0300)]
spi: dw: Detach SPI device specific CR0 config method

Indeed there is no point in detecting the SPI peripheral device parameters
and initializing the CR0 register fields each time an SPI transfer is
executed. Instead let's define a dedicated CR0 chip-data member, which
will be initialized in accordance with the SPI device settings at the
moment of setting it up.

By doing so we'll finally make the SPI device chip_data serving as it's
supposed to - to preserve the SPI device specific DW SPI configuration.
See spi-fsl-dspi.c, spi-pl022.c, spi-pxa2xx.c drivers for example of the
way the chip data is utilized.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-4-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Add DWC SSI capability
Serge Semin [Wed, 7 Oct 2020 23:54:51 +0000 (02:54 +0300)]
spi: dw: Add DWC SSI capability

Currently DWC SSI core is supported by means of setting up the
core-specific update_cr0() callback. It isn't suitable for multiple
reasons. First of all having exported several methods doing the same thing
but for different chips makes the code harder to maintain. Secondly the
spi-dw-core driver exports the methods, then the spi-dw-mmio driver sets
the private data callback with one of them so to be called by the core
driver again. That makes the code logic too complicated. Thirdly using
callbacks for just updating the CR0 register is problematic, since in case
if the register needed to be updated from different parts of the code,
we'd have to create another callback (for instance the SPI device-specific
parameters don't need to be calculated each time the SPI transfer is
submitted, so it's better to pre-calculate the CR0 data at the SPI-device
setup stage).

So keeping all the above in mind let's discard the update_cr0() callbacks,
define a generic and static dw_spi_update_cr0() method and create the
DW_SPI_CAP_DWC_SSI capability, which when enabled would activate the
alternative CR0 register layout.

While at it add the comments to the code path of the normal DW APB SSI
controller setup to make the dw_spi_update_cr0() method looking coherent.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-3-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Use an explicit set_cs assignment
Serge Semin [Wed, 7 Oct 2020 23:54:50 +0000 (02:54 +0300)]
spi: dw: Use an explicit set_cs assignment

Simplify the dw_spi_add_host() method a bit by replacing the currently
implemented default set_cs callback setting up and later having it
overwritten by a custom function with direct if-else-based callback
assignment.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20201007235511.4935-2-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-mtk-nor: Add power management support
Ikjoon Jang [Tue, 6 Oct 2020 07:54:05 +0000 (15:54 +0800)]
spi: spi-mtk-nor: Add power management support

This patch adds dev_pm_ops to mtk-nor to support suspend/resume,
auto suspend delay is set to -1 by default.

Accessing registers are only permitted after its clock is enabled
to deal with unknown state of operating clk at probe time.

Signed-off-by: Ikjoon Jang <ikjn@chromium.org>
Link: https://lore.kernel.org/r/20201006155010.v5.4.I68983b582d949a91866163bab588ff3c2a0d0275@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-mtk-nor: support 36bit dma addressing
Ikjoon Jang [Tue, 6 Oct 2020 07:54:04 +0000 (15:54 +0800)]
spi: spi-mtk-nor: support 36bit dma addressing

This patch enables 36bit dma address support to spi-mtk-nor.
Currently this is enabled only for mt8192-nor.

Signed-off-by: Ikjoon Jang <ikjn@chromium.org>
Link: https://lore.kernel.org/r/20201006155010.v5.3.Id1cb208392928afc7ceed4de06924243c7858cd0@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-mtk-nor: use dma_alloc_coherent() for bounce buffer
Ikjoon Jang [Tue, 6 Oct 2020 07:54:03 +0000 (15:54 +0800)]
spi: spi-mtk-nor: use dma_alloc_coherent() for bounce buffer

Use dma_alloc_coherent() for bounce buffer instead of kmalloc() to
make sure the bounce buffer to be allocated within its DMAable range.

Signed-off-by: Ikjoon Jang <ikjn@chromium.org>
Link: https://lore.kernel.org/r/20201006155010.v5.2.I06cb65401ab5ad63ea30c4788d26633928d80f38@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agodt-bindings: spi: add mt8192-nor compatible string
Ikjoon Jang [Tue, 6 Oct 2020 07:54:02 +0000 (15:54 +0800)]
dt-bindings: spi: add mt8192-nor compatible string

Add MT8192 spi-nor controller support.

Signed-off-by: Ikjoon Jang <ikjn@chromium.org>
Acked-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20201006155010.v5.1.I4cd089ef1fe576535c6b6e4f1778eaab1c4441cf@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: renesas,sh-msiof: Add r8a77961 support
Geert Uytterhoeven [Mon, 5 Oct 2020 11:25:47 +0000 (13:25 +0200)]
spi: renesas,sh-msiof: Add r8a77961 support

Document R-Car M3-W+ (R8A77961) SoC bindings.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20201005112549.22222-1-geert+renesas@glider.be
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3c64xx: Turn on interrupts upon resume
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:43 +0000 (14:22 +0200)]
spi: spi-s3c64xx: Turn on interrupts upon resume

s3c64xx_spi_hwinit() disables interrupts. In s3c64xx_spi_probe() after
calling s3c64xx_spi_hwinit() they are enabled with the following call.

Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Link: https://lore.kernel.org/r/20201002122243.26849-10-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3c64xx: Increase transfer timeout
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:42 +0000 (14:22 +0200)]
spi: spi-s3c64xx: Increase transfer timeout

Increase timeout by 30 ms for some wiggle room and set the minimum value
to 100 ms. This ensures a non-zero value for short transfers which
may take less than 1 ms. The timeout value does not affect
performance because it is used with a completion.

Similar formula is used in other drivers e.g. sun4i, sun6i.

Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Link: https://lore.kernel.org/r/20201002122243.26849-9-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3c64xx: Ensure cur_speed holds actual clock value
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:41 +0000 (14:22 +0200)]
spi: spi-s3c64xx: Ensure cur_speed holds actual clock value

Make sure the cur_speed value used in s3c64xx_enable_datapath()
to configure DMA channel and in s3c64xx_wait_for_*() to calculate the
transfer timeout is set to the actual value of (half) the clock speed.

Don't change non-CMU case, because no frequency calculation errors have
been reported.

Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Suggested-by: Tomasz Figa <tomasz.figa@gmail.com>
Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Link: https://lore.kernel.org/r/20201002122243.26849-8-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3c64xx: Fix doc comment for struct s3c64xx_spi_driver_data
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:40 +0000 (14:22 +0200)]
spi: spi-s3c64xx: Fix doc comment for struct s3c64xx_spi_driver_data

Remove descriptions for non-existent fields and fix indentation.

Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Link: https://lore.kernel.org/r/20201002122243.26849-7-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3c64xx: Rename S3C64XX_SPI_SLAVE_* to S3C64XX_SPI_CS_*
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:39 +0000 (14:22 +0200)]
spi: spi-s3c64xx: Rename S3C64XX_SPI_SLAVE_* to S3C64XX_SPI_CS_*

Rename S3C64XX_SPI_SLAVE_* to S3C64XX_SPI_CS_* to match documentation.

Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20201002122243.26849-6-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3c64xx: Report more information when errors occur
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:38 +0000 (14:22 +0200)]
spi: spi-s3c64xx: Report more information when errors occur

Report amount of pending data when a transfer stops due to errors.

Report if DMA was used to transfer data and print the status code.

Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Link: https://lore.kernel.org/r/20201002122243.26849-5-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3c64xx: Check return values
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:37 +0000 (14:22 +0200)]
spi: spi-s3c64xx: Check return values

Check return values in prepare_dma() and s3c64xx_spi_config() and
propagate errors upwards.

Fixes: 788437273fa8 ("spi: s3c64xx: move to generic dmaengine API")
Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Link: https://lore.kernel.org/r/20201002122243.26849-4-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3s64xx: Add S3C64XX_SPI_QUIRK_CS_AUTO for Exynos3250
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:36 +0000 (14:22 +0200)]
spi: spi-s3s64xx: Add S3C64XX_SPI_QUIRK_CS_AUTO for Exynos3250

Fix issues with DMA transfers bigger than 512 bytes on Exynos3250. Without
the patches such transfers fail.

The vendor kernel for ARTIK5 handles CS in a simmilar way.

Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20201002122243.26849-3-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-s3c64xx: swap s3c64xx_spi_set_cs() and s3c64xx_enable_datapath()
Łukasz Stelmach [Fri, 2 Oct 2020 12:22:35 +0000 (14:22 +0200)]
spi: spi-s3c64xx: swap s3c64xx_spi_set_cs() and s3c64xx_enable_datapath()

Fix issues with DMA transfers bigger than 512 bytes on Exynos3250. Without
the patches such transfers fail to complete. This solution to the problem
is found in the vendor kernel for ARTIK5 boards based on Exynos3250.

Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Łukasz Stelmach <l.stelmach@samsung.com>
Link: https://lore.kernel.org/r/20201002122243.26849-2-l.stelmach@samsung.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agoMerge series "spi: spi-mtk-nor: make use of full capability of program mode" from...
Mark Brown [Thu, 1 Oct 2020 22:45:29 +0000 (23:45 +0100)]
Merge series "spi: spi-mtk-nor: make use of full capability of program mode" from Chuanhong Guo <gch981213@gmail.com>:

"program" mode on this controller can trigger up to 56 bits of data
shifting. During the operation, data in PRGDATA[0-5] will be
shifted out from MOSI, and data from MISO will be continuously filling
SHREG[0-9].
Currently this mode is used to implement transfer_one_message for 6-byte
full-duplex transfer, but it can execute a transfer for up-to 7 bytes
as long as the last byte is read only.
transfer_one_message is expected to perform full-duplex transfer,
instead of transfer with specific format. mtk_nor_spi_mem_prg is
added here to use this extra byte.

Newer version of this controller can trigger longer data shifting with
shift bytes more than PRGDATA_MAX + SHREG_MAX. This patch is implemented
with that in mind and it checks against both SHREG_MAX and PRG_CNT_MAX
for future support of new controllers.

Patch 3/3 is a fix for:
commit a59b2c7c56bf7 ("spi: spi-mtk-nor: support standard spi properties")
which breaks supports_op logic. But it can't be separated as it depends
on patch 2/3. Fortuantely the broken commit isn't in stable yet.

Chuanhong Guo (3):
  spi: spi-mtk-nor: make use of full capability of prg mode
  spi: spi-mtk-nor: add helper for checking prg mode ops
  spi: spi-mtk-nor: fix op checks in supports_op

 drivers/spi/spi-mtk-nor.c | 179 +++++++++++++++++++++++++++++++++-----
 1 file changed, 158 insertions(+), 21 deletions(-)

--
2.26.2

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

3 years agospi: spi-stm32: remove redundant irqsave and irqrestore in hardIRQ
Barry Song [Sat, 26 Sep 2020 00:16:16 +0000 (12:16 +1200)]
spi: spi-stm32: remove redundant irqsave and irqrestore in hardIRQ

Running in hardIRQ, disabling IRQ is redundant.

Signed-off-by: Barry Song <song.bao.hua@hisilicon.com>
Link: https://lore.kernel.org/r/20200926001616.21292-2-song.bao.hua@hisilicon.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-tegra20-sflash: remove redundant irqsave and irqrestore in hardIRQ
Barry Song [Sat, 26 Sep 2020 00:16:15 +0000 (12:16 +1200)]
spi: spi-tegra20-sflash: remove redundant irqsave and irqrestore in hardIRQ

Running in hardIRQ, disabling IRQ is redundant.

Signed-off-by: Barry Song <song.bao.hua@hisilicon.com>
Link: https://lore.kernel.org/r/20200926001616.21292-1-song.bao.hua@hisilicon.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: atmel: Exposing effective spi speed
Thomas Kopp [Mon, 21 Sep 2020 07:10:36 +0000 (09:10 +0200)]
spi: atmel: Exposing effective spi speed

This patch implements the reporting of the effectively used speed_hz for
the transfer by setting xfer->effective_speed_hz.

See the following patch, which adds this feature to the SPI core for more
information:
commit 5d7e2b5ed585 ("spi: core: allow reporting the effectivly used speed_hz for a transfer")

Signed-off-by: Thomas Kopp <thomas.kopp@microchip.com>
Reviewed-by: Tudor Ambarus <tudor.ambarus@microchip.com>
Link: https://lore.kernel.org/r/20200921071036.2091-1-thomas.kopp@microchip.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-mtk-nor: fix op checks in supports_op
Chuanhong Guo [Thu, 24 Sep 2020 15:27:30 +0000 (23:27 +0800)]
spi: spi-mtk-nor: fix op checks in supports_op

commit a59b2c7c56bf7 ("spi: spi-mtk-nor: support standard spi properties")
tries to inverse the logic of supports_op when adding
spi_mem_default_supports_op check, but it didn't get it done properly.
There are two regressions introduced by this commit:
1. reading ops supported by program mode is rejected.
2. all ops with special controller routines are incorrectly further
   checked against program mode.

This commits inverses the logic back:
1. check spi_mem_default_supports_op and reject unsupported ops first.
2. return true for ops with special controller routines.
3. check the left ops against controller program mode.

Fixes: a59b2c7c56bf7 ("spi: spi-mtk-nor: support standard spi properties")
Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Link: https://lore.kernel.org/r/20200924152730.733243-4-gch981213@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-mtk-nor: add helper for checking prg mode ops
Chuanhong Guo [Thu, 24 Sep 2020 15:27:29 +0000 (23:27 +0800)]
spi: spi-mtk-nor: add helper for checking prg mode ops

op checking/resizing logic for the newly added mtk_nor_spi_mem_prg is
more complicated. Add two helper functions for them:
mtk_nor_match_prg: check whether an op is supported by prg mode.
mtk_nor_adj_prg_size: adjust data size for mtk_nor_spi_mem_prg.

mtk_nor_match_prg isn't called yet because supports_op is currently
broken. It'll be used in the next fix commit.

Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Link: https://lore.kernel.org/r/20200924152730.733243-3-gch981213@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-mtk-nor: make use of full capability of prg mode
Chuanhong Guo [Thu, 24 Sep 2020 15:27:28 +0000 (23:27 +0800)]
spi: spi-mtk-nor: make use of full capability of prg mode

"program" mode on this controller can trigger up to 56 bits of data
shifting. During the operation, data in PRGDATA[0-5] will be
shifted out from MOSI, and data from MISO will be continuously filling
SHREG[0-9].
Currently this mode is used to implement transfer_one_message for 6-byte
full-duplex transfer, but it can execute a transfer for up-to 7 bytes
as long as the last byte is read only.
transfer_one_message is expected to perform full-duplex transfer,
instead of transfer with specific format. mtk_nor_spi_mem_prg is
added here to use this extra byte.

Newer version of this controller can trigger longer data shifting with
shift bytes more than PRGDATA_MAX + SHREG_MAX. This patch is implemented
with that in mind and it checks against both SHREG_MAX and PRG_CNT_MAX
for future support of new controllers.

Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Link: https://lore.kernel.org/r/20200924152730.733243-2-gch981213@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agoMerge series "spi: dw: Add full Baikal-T1 SPI Controllers support" from Serge Semin...
Mark Brown [Tue, 29 Sep 2020 16:22:29 +0000 (17:22 +0100)]
Merge series "spi: dw: Add full Baikal-T1 SPI Controllers support" from Serge Semin <Sergey.Semin@baikalelectronics.ru>:

Originally I intended to merge a dedicated Baikal-T1 System Boot SPI
Controller driver into the kernel and leave the DW APB SSI driver
untouched. But after a long discussion (see the link at the bottom of the
letter) Mark and Andy persuaded me to integrate what we developed there
into the DW APB SSI core driver to be useful for another controllers,
which may have got the same peculiarities/problems as ours:
- No IRQ.
- No DMA.
- No GPIO CS, so a native CS is utilized.
- small Tx/Rx FIFO depth.
- Automatic CS assertion/de-assertion.
- Slow system bus.
All of them have been fixed in the framework of this patchset in some
extent at least for the SPI memory operations. As I expected it wasn't
that easy and the integration took that many patches as you can see from
the subject. Though some of them are mere cleanups or weakly related with
the subject fixes, but we just couldn't leave the code as is at some
places since we were working with the DW APB SSI driver anyway. Here is
what we did to fix the original DW APB SSI driver, to make it less messy.

First two patches are just cleanups to simplify the DW APB SSI device
initialization a bit. We suggest to discard the IRQ threshold macro as
unused and use a ternary operator to initialize the set_cs callback
instead of assigning-and-updating it.

Then we've discovered that the n_bytes field of the driver private data is
used by the DW APB SSI IRQ handler, which requires it to be initialized
before the SMP memory barrier and to be visible from another CPUs. Speaking
about the SMP memory barrier. Having one right after the shared resources
initialization is enough and there is no point in using the spin-lock to
protect the Tx/Rx buffer pointers. The protection functionality is
redundant there by the driver design. (Though I have a doubt whether the
SMP memory barrier is also required there because the normal IO-methods
like readl/writel implies a full memory barrier. So any memory operations
performed before them are supposed to be seen by devices and another CPUs.
See the patch log for details of my concern.)

Thirdly we've found out that there is some confusion in the IRQs
masking/unmasking/clearing in the SPI-transfer procedure. Multiple interrupts
are unmasked on the SPI-transfer initialization, but just TXEI is only
masked back on completion. Similarly IRQ status isn't cleared on the
controller reset, which actually makes the reset being not full and errors
prone in the controller probe procedure.

Another very important optimization is using the IO-relaxed accessors in
the dw_read_io_reg()/dw_write_io_reg() methods. Since the Tx/Rx FIFO data
registers are the most frequently accessible controller resource, using
relaxed accessors there will significantly improve the data read/write
performance. At least on Baikal-T1 SoC such modification opens up a way to
have the DW APB SSI controller working with higher SPI bus speeds, than
without it.

Fifthly we've made an effort to cleanup the code using the SPI-device
private data - chip_data. We suggest to remove the chip type from there
since it isn't used and isn't implemented right anyway. Then instead of
having a bus speed, clock divider, transfer mode preserved there, and
recalculating the CR0 fields of the SPI-device-specific phase, polarity
and frame format each time the SPI transfer is requested, we can save it
in the chip_data instance. By doing so we'll make that structure finally
used as it was supposed to by design (see the spi-fsl-dspi.c, spi-pl022.c,
spi-pxa2xx.c drivers for examples).

Sixthly instead of having the SPI-transfer specific CR0-update callback,
we suggest to implement the DW APB SSI controller capabilities approach.
By doing so we can now inject the vendor-specific peculiarities in
different parts of the DW APB SSI core driver (which is required to
implement both SPI-transfers and the SPI memory operations). This will
also make the code less confusing like defining a callback in the core
driver, setting it up in the glue layer, then calling it from the core
driver again. Seeing the small capabilities implementation embedded
in-situ is more readable than tracking the callbacks assignments. This
will concern the CS-override, Keembay master setup, DW SSI-specific CR0
registers layout capabilities.

Seventhly since there are going to be two types of the transfers
implemented in the DW APB SSI core driver, we need a common method to set
the controller configuration like, Tx/Rx-mode, bus speed, data frame size
and number of data frames to read in case of the memory operations. So we
just detached the corresponding code from the SPI-transfer-one method and
made it to be a part of the new dw_spi_update_config() function, which is
former update_cr0(). Note that the new method will be also useful for the
glue drivers, which due to the hardware design need to create their own
memory operations (for instance, for the dirmap-operations provided in the
Baikal-T System Boot SPI controller driver).

Eighthly it is the data IO procedure and IRQ-based SPI-transfer
implementation refactoring. The former one will look much simpler if the
buffers initial pointers and the buffers length data utilized instead of
the Tx/Rx buffers start and end pointers. The later one currently lacks of
valid execution at the final stage of the SPI-transfer. So if there is no
data left to send, but there is still data which needs to be received, the
Tx FIFO Empty IRQ will constantly happen until all of the requested
inbound data is received. So we suggest to fix that by taking the Rx FIFO
Empty IRQ into account.

Ninthly it's potentially errors prone to enable the DW APB SSI interrupts
before enabling the chip. It specifically concerns a case if for some
reason the DW APB SSI IRQs handler is executed before the controller is
enabled. That will cause a part of the outbound data loss. So we suggest
to reverse the order.

Tenthly in order to be able to pre-initialize the Tx FIFO with data and
only the start the SPI memory operations we need to have any CS
de-activated. We'll fulfil that requirement by explicitly clearing the CS
on the SPI transfer completion and at the explicit controller reset.

Then seeing all the currently available and potentially being created
types of the SPI transfers need to perform the DW APB SSI controller
status register check and the errors handler procedure, we've created a
common method for all of them.

Eleventhly if before we've mostly had a series of fixups, cleanups and
refactorings, here we've finally come to the new functionality
implementation. It concerns the poll-based transfer (as Baikal-T1 System
Boot SPI controller lacks a dedicated IRQ lane connected) and the SPI
memory operations implementation. If the former feature is pretty much
straightforward (see the patch log for details), the later one is a bit
tricky. It's based on the EEPROM-read (write-then-read) and the Tx-only
modes of the DW APB SSI controller, which as performing the automatic data
read and write let's us to implement the faster IO procedure than using
the Tx-Rx-mode-based approach. Having the memory-operations implemented
that way is the best thing we can currently do to provide the errors-less
SPI transfers to SPI devices with native CS attached.

Note the approach utilized here to develop the SPI memory operations can
be also used to create the "automatic CS toggle problem"-free(ish) SPI
transfers (combine SPI-message transfers into two buffers, disable
interrupts, push-pull the combined data). But we don't provide a solution
in the framework of this patchset. It is a matter of a dedicated one,
which we currently don't intend to spend our time on.

Finally at the closure of the this patchset you'll find patches, which
provide the Baikal-T1-specific DW APB SSI controllers support. The SoC has
got three SPI controllers. Two of them are pretty much normal DW APB SSI
interfaces: with IRQ, DMA, FIFOs of 64 words depth, 4x CSs. But the third
one as being a part of the Baikal-T1 System Boot Controller has got a very
limited resources: no IRQ, no DMA, only a single native chip-select and
Tx/Rx FIFOs with just 8 words depth available. In order to provide a
transparent initial boot code execution the System Boot SPI Controller is
also utilized by an vendor-specific IP-block, which exposes an SPI flash
memory direct mapping interface. Please see the corresponding patch for
details.

Link: https://lore.kernel.org/linux-spi/20200508093621.31619-1-Sergey.Semin@baikalelectronics.ru/
[1] "LINUX KERNEL MEMORY BARRIERS", Documentation/memory-barriers.txt,
    Section "KERNEL I/O BARRIER EFFECTS"

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Cc: Alexey Malahov <Alexey.Malahov@baikalelectronics.ru>
Cc: Ramil Zaripov <Ramil.Zaripov@baikalelectronics.ru>
Cc: Pavel Parkhomenko <Pavel.Parkhomenko@baikalelectronics.ru>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Lars Povlsen <lars.povlsen@microchip.com>
Cc: wuxu.wu <wuxu.wu@huawei.com>
Cc: Feng Tang <feng.tang@intel.com>
Cc: Rob Herring <robh+dt@kernel.org>
Cc: linux-spi@vger.kernel.org
Cc: devicetree@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Serge Semin (30):
  spi: dw: Discard IRQ threshold macro
  spi: dw: Use ternary op to init set_cs callback
  spi: dw: Initialize n_bytes before the memory barrier
  Revert: spi: spi-dw: Add lock protect dw_spi rx/tx to prevent
    concurrent calls
  spi: dw: Clear IRQ status on DW SPI controller reset
  spi: dw: Disable all IRQs when controller is unused
  spi: dw: Use relaxed IO-methods to access FIFOs
  spi: dw: Discard DW SSI chip type storages
  spi: dw: Convert CS-override to DW SPI capabilities
  spi: dw: Add KeemBay Master capability
  spi: dw: Add DWC SSI capability
  spi: dw: Detach SPI device specific CR0 config method
  spi: dw: Update SPI bus speed in a config function
  spi: dw: Simplify the SPI bus speed config procedure
  spi: dw: Update Rx sample delay in the config function
  spi: dw: Add DW SPI controller config structure
  spi: dw: Refactor data IO procedure
  spi: dw: Refactor IRQ-based SPI transfer procedure
  spi: dw: Perform IRQ setup in a dedicated function
  spi: dw: Unmask IRQs after enabling the chip
  spi: dw: Discard chip enabling on DMA setup error
  spi: dw: De-assert chip-select on reset
  spi: dw: Explicitly de-assert CS on SPI transfer completion
  spi: dw: Move num-of retries parameter to the header file
  spi: dw: Add generic DW SSI status-check method
  spi: dw: Add memory operations support
  spi: dw: Introduce max mem-ops SPI bus frequency setting
  spi: dw: Add poll-based SPI transfers support
  dt-bindings: spi: dw: Add Baikal-T1 SPI Controllers
  spi: dw: Add Baikal-T1 SPI Controller glue driver

 .../bindings/spi/snps,dw-apb-ssi.yaml         |  33 +-
 drivers/spi/Kconfig                           |  29 +
 drivers/spi/Makefile                          |   1 +
 drivers/spi/spi-dw-bt1.c                      | 339 +++++++++
 drivers/spi/spi-dw-core.c                     | 642 ++++++++++++++----
 drivers/spi/spi-dw-dma.c                      |  16 +-
 drivers/spi/spi-dw-mmio.c                     |  36 +-
 drivers/spi/spi-dw.h                          |  85 ++-
 8 files changed, 960 insertions(+), 221 deletions(-)
 create mode 100644 drivers/spi/spi-dw-bt1.c

--
2.27.0

3 years agospi: spi-dw: Remove extraneous locking
Serge Semin [Sun, 20 Sep 2020 11:28:48 +0000 (14:28 +0300)]
spi: spi-dw: Remove extraneous locking

There is no point in having the commit 19b61392c5a8 ("spi: spi-dw: Add
lock protect dw_spi rx/tx to prevent concurrent calls") applied. The
commit author made an assumption that the problem with the rx data
mismatch was due to the lack of the data protection. While most likely it
was caused by the lack of the memory barrier. So having the
commit bfda044533b2 ("spi: dw: use "smp_mb()" to avoid sending spi data
error") applied would be enough to fix the problem.

Indeed the spin unlock operation makes sure each memory operation issued
before the release will be completed before it's completed. In other words
it works as an implicit one way memory barrier. So having both smp_mb()
and the spin_unlock_irqrestore() here is just redundant. One of them would
be enough. It's better to leave the smp_mb() since the Tx/Rx buffers
consistency is provided by the data transfer algorithm implementation:
first we initialize the buffers pointers, then make sure the assignments
are visible by the other CPUs by calling the smp_mb(), only after that
enable the interrupt, which handler uses the buffers.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-5-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Add KeemBay Master capability
Serge Semin [Sun, 20 Sep 2020 11:28:54 +0000 (14:28 +0300)]
spi: dw: Add KeemBay Master capability

In a further commit we'll have to get rid of the update_cr0() callback and
define a DW SSI capability instead. Since Keem Bay master/slave
functionality is controller by the CTRL0 register bitfield, we need to
first move the master mode selection into the internal corresponding
update_cr0 method, which would be activated by means of the dedicated
DW_SPI_CAP_KEEMBAY_MST capability setup.

Note this will be also useful if the driver will be ever altered to
support the DW SPI slave interface.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-11-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Convert CS-override to DW SPI capabilities
Serge Semin [Sun, 20 Sep 2020 11:28:53 +0000 (14:28 +0300)]
spi: dw: Convert CS-override to DW SPI capabilities

There are several vendor-specific versions of the DW SPI controllers,
each of which may have some peculiarities with respect to the original
IP-core. Seeing it has already caused adding flags and a callback into the
DW SPI private data, let's introduce a generic capabilities interface to
tune the generic DW SPI controller driver up in accordance with the
particular controller specifics. It's done by converting a simple
Alpine-specific CS-override capability into the DW SPI controller
capability activated by setting the DW_SPI_CAP_CS_OVERRIDE flag.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-10-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Discard DW SSI chip type storages
Serge Semin [Sun, 20 Sep 2020 11:28:52 +0000 (14:28 +0300)]
spi: dw: Discard DW SSI chip type storages

Keeping SPI peripheral devices type is pointless since first it hasn't
been functionally utilized by any of the client drivers/code and second it
won't work for Microwire type at the very least. Moreover there is no
point in setting up the type by means of the chip-data in the modern
kernel. The peripheral devices with specific interface type need to be
detected in order to activate the corresponding frame format. It most
likely will require some peripheral device specific DT property or
whatever to find out the interface protocol. So let's remove the serial
interface type fields from the DW APB SSI controller and the SPI
peripheral device private data.

Note we'll preserve the explicit SSI_MOTO_SPI interface type setting up to
signify the only currently supported interface protocol.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-9-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Use relaxed IO-methods to access FIFOs
Serge Semin [Sun, 20 Sep 2020 11:28:51 +0000 (14:28 +0300)]
spi: dw: Use relaxed IO-methods to access FIFOs

In accordance with [1] the relaxed methods are guaranteed to be ordered
with respect to other accesses from the same CPU thread to the same
peripheral.  This is what we need during the data read/write from/to the
controller FIFOs being executed within a single IRQ handler or a kernel
task.

Such optimization shall significantly speed the data reader and writer up.
For instance, the relaxed IO-accessors utilization on Baikal-T1 lets the
driver to support the SPI memory operations with bus frequency three-fold
faster than if normal IO-accessors would be used.

[1] "LINUX KERNEL MEMORY BARRIERS", Documentation/memory-barriers.txt,
    Section "KERNEL I/O BARRIER EFFECTS"

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-8-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Disable all IRQs when controller is unused
Serge Semin [Sun, 20 Sep 2020 11:28:50 +0000 (14:28 +0300)]
spi: dw: Disable all IRQs when controller is unused

It's a good practice to disable all IRQs if a device is fully unused. In
our case it is supposed to be done before requesting the IRQ and after the
last byte of an SPI transfer is received. In the former case it's required
to prevent the IRQ handler invocation before the driver data is fully
initialized (which may happen if the IRQs status has been left uncleared
before the device is probed). So we just moved the spi_hw_init() method
invocation to the earlier stage before requesting the IRQ. In the later
case there is just no point in having any of the IRQs enabled between SPI
transfers and when there is no SPI message currently being processed.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-7-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Clear IRQ status on DW SPI controller reset
Serge Semin [Sun, 20 Sep 2020 11:28:49 +0000 (14:28 +0300)]
spi: dw: Clear IRQ status on DW SPI controller reset

It turns out the IRQ status isn't cleared after switching the controller
off and getting it back on, which may cause raising false error interrupts
if controller has been unsuccessfully used by, for instance, a bootloader
before the driver is loaded. Let's explicitly clear the interrupts status
in the dedicated controller reset method.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-6-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Initialize n_bytes before the memory barrier
Serge Semin [Sun, 20 Sep 2020 11:28:47 +0000 (14:28 +0300)]
spi: dw: Initialize n_bytes before the memory barrier

Since n_bytes field of the DW SPI private data is also utilized by the
IRQ handler, we need to make sure it' initialization is done before the
memory barrier.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-4-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw: Discard IRQ threshold macro
Serge Semin [Sun, 20 Sep 2020 11:28:45 +0000 (14:28 +0300)]
spi: dw: Discard IRQ threshold macro

The macro has been unused since a half of FIFO length was defined to be a
marker of the IRQ. Let's remove it definition.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112914.26501-2-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Add one-by-one SG list entries transfer
Serge Semin [Sun, 20 Sep 2020 11:23:22 +0000 (14:23 +0300)]
spi: dw-dma: Add one-by-one SG list entries transfer

In case if at least one of the requested DMA engine channels doesn't
support the hardware accelerated SG list entries traverse, the DMA driver
will most likely work that around by performing the IRQ-based SG list
entries resubmission. That might and will cause a problem if the DMA Tx
channel is recharged and re-executed before the Rx DMA channel. Due to
non-deterministic IRQ-handler execution latency the DMA Tx channel will
start pushing data to the SPI bus before the Rx DMA channel is even
reinitialized with the next inbound SG list entry. By doing so the DMA
Tx channel will implicitly start filling the DW APB SSI Rx FIFO up, which
while the DMA Rx channel being recharged and re-executed will eventually
be overflown.

In order to solve the problem we have to feed the DMA engine with SG
list entries one-by-one. It shall keep the DW APB SSI Tx and Rx FIFOs
synchronized and prevent the Rx FIFO overflow. Since in general the SPI
tx_sg and rx_sg lists may have different number of entries of different
lengths (though total length should match) we virtually split the
SG-lists to the set of DMA transfers, which length is a minimum of the
ordered SG-entries lengths.

The solution described above is only executed if a full-duplex SPI
transfer is requested and the DMA engine hasn't provided channels with
hardware accelerated SG list traverse capability to handle both SG
lists at once.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20200920112322.24585-12-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Pass exact data to the DMA submit and wait methods
Serge Semin [Sun, 20 Sep 2020 11:23:21 +0000 (14:23 +0300)]
spi: dw-dma: Pass exact data to the DMA submit and wait methods

In order to use the DMA submission and waiting methods in both generic
DMA-based SPI transfer and one-by-one DMA SG entries transmission
functions, we need to modify the dw_spi_dma_wait() and
dw_spi_dma_submit_tx()/dw_spi_dma_submit_rx() prototypes. So instead of
getting the SPI transfer object as the second argument they must accept
the exact data structure instances they imply to use. Those are the
current transfer length and the SPI bus frequency in case of
dw_spi_dma_wait(), and SG list together with number of list entries in
case of the DMA Tx/Rx submission methods.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-11-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Move DMAC register cleanup to DMA transfer method
Serge Semin [Sun, 20 Sep 2020 11:23:20 +0000 (14:23 +0300)]
spi: dw-dma: Move DMAC register cleanup to DMA transfer method

DW APB SSI DMA driver doesn't use the native SPI core wait API since
commit bdbdf0f06337 ("spi: dw: Locally wait for the DMA transfers
completion"). Due to that the driver can now clear the DMAC register
in a single place synchronously with the DMA transactions completion
or failure. After that all the possible code paths are still covered:
1) DMA completion callbacks are executed in case if the corresponding DMA
transactions are finished. When they are, one of them will eventually wake
the SPI messages pump kernel thread and dw_spi_dma_transfer_all() method
will clean the DMAC register as implied by this patch.
2) dma_stop is called when the SPI core detects an error either returned
from the transfer_one() callback or set in the SPI message status field.
Both types of errors will be noticed by the dw_spi_dma_transfer_all()
method.
3) dma_exit is called when either SPI controller driver or the
corresponding device is removed. In any case the SPI core will first
flush the SPI messages pump kernel thread, so any pending or in-fly
SPI transfers will be finished before that.

Due to all of that let's simplify the DW APB SSI DMA driver a bit and
move the DMAC register cleanup to a single place in the
dw_spi_dma_transfer_all() method.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-10-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Detach DMA transfer into a dedicated method
Serge Semin [Sun, 20 Sep 2020 11:23:19 +0000 (14:23 +0300)]
spi: dw-dma: Detach DMA transfer into a dedicated method

In order to add an alternative method of DMA-based SPI transfer first we
need to detach the currently available one from the common code. Here we
move the normal DMA-based SPI transfer execution functionality into a
dedicated method. It will be utilized if either the DMA engine supports
an unlimited number SG entries or Tx-only SPI transfer is requested. But
currently just use it for any SPI transfer.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-9-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Remove DMA Tx-desc passing around
Serge Semin [Sun, 20 Sep 2020 11:23:18 +0000 (14:23 +0300)]
spi: dw-dma: Remove DMA Tx-desc passing around

It's pointless to pass the Rx and Tx transfers DMA Tx-descriptors, since
they are used in the Tx/Rx submit method only. Instead just return the
submission status from these methods. This alteration will make the code
less complex.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-8-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Check DMA Tx-desc submission status
Serge Semin [Sun, 20 Sep 2020 11:23:17 +0000 (14:23 +0300)]
spi: dw-dma: Check DMA Tx-desc submission status

We suggest to add the dmaengine_submit() return value test for errors.  It
has been unnecessary while the driver was expected to be utilized in pair
with DW DMAC. But since now the driver can be used with any DMA engine, it
might be useful to track the errors on DMA submissions so not miss them
and get into an unpredictable driver behaviour.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-7-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Move DMA transfers submission to the channels prep methods
Serge Semin [Sun, 20 Sep 2020 11:23:16 +0000 (14:23 +0300)]
spi: dw-dma: Move DMA transfers submission to the channels prep methods

Indeed we can freely move the dmaengine_submit() method invocation and the
Tx and Rx busy flag setting into the DMA Tx/Rx prepare methods. Since the
Tx/Rx preparation method is now mainly used for the DMA transfers
submission, here we suggest to rename it to have the _submit_{r,t}x suffix
instead.

By having this alteration applied first we implement another code
preparation before adding the one-by-one DMA SG entries transmission,
second we now have the dma_async_tx_descriptor descriptor used locally
only in the new DMA transfers submission methods (this will be cleaned up
a bit later), third we make the generic transfer method more readable,
where now the functionality of submission, execution and wait procedures
is transparently split up instead of having a preparation, intermixed
submission/execution and wait procedures.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-6-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Check rx_buf availability in the xfer method
Serge Semin [Sun, 20 Sep 2020 11:23:15 +0000 (14:23 +0300)]
spi: dw-dma: Check rx_buf availability in the xfer method

Checking rx_buf for being NULL and returning NULL from the Rx-channel
preparation method doesn't let us to distinguish that situation from
errors happening during the Rx SG-list preparation. So it's better to make
sure that the rx_buf not-NULL and full-duplex communication is requested
prior calling the Rx preparation method.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-5-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Configure the DMA channels in dma_setup
Serge Semin [Sun, 20 Sep 2020 11:23:14 +0000 (14:23 +0300)]
spi: dw-dma: Configure the DMA channels in dma_setup

Mainly this is a preparation patch before adding one-by-one DMA SG entries
transmission. But logically the Tx and Rx DMA channels setup should be
performed in the dma_setup() callback anyway. So we'll move the DMA slave
channels src/dst burst lengths, address and address width configuration
from the Tx/Rx channels preparation methods to the dedicated functions and
then make sure it's called at the DMA setup stage.

Note we now make sure the return value of the dmaengine_slave_config()
method doesn't indicate an error. It has been unnecessary in case if Dw
DMAC is utilized as a DMA engine, since its device_config() callback
always returns zero (though it might change in future). But since DW APB
SSI driver now supports any DMA back-end we must make sure the DMA device
configuration has been successful before proceeding with further setups.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-4-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Fail DMA-based transfer if no Tx-buffer specified
Serge Semin [Sun, 20 Sep 2020 11:23:13 +0000 (14:23 +0300)]
spi: dw-dma: Fail DMA-based transfer if no Tx-buffer specified

Since commit 46164fde6b78 ("spi: dw: Fix Rx-only DMA transfers") if DMA
interface is enabled, then Tx-buffer must be available in each SPI
transfer. It's required since in order to activate the incoming data
reception either DMA or CPU must be pushing data out to the SPI bus.
But the DW APB SSI DMA driver code is still left in state as if Tx-buffer
might be optional, which is no longer true. Let's fix it so an error would
be returned if no Tx-buffer detected and DMA Tx would be always
enabled.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-3-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: dw-dma: Set DMA Level registers on init
Serge Semin [Sun, 20 Sep 2020 11:23:12 +0000 (14:23 +0300)]
spi: dw-dma: Set DMA Level registers on init

Indeed the registers content doesn't get cleared when the SPI controller
is disabled and enabled. Max burst lengths aren't changed since the Rx and
Tx DMA channels are requested on init stage and are kept acquired until
the device is removed. Obviously SPI controller FIFO depth can't be
changed. Due to all of that we can safely move the DMA Transmit and
Receive data level registers initialization to the SPI controller DMA init
stage (when the SPI controller is being probed) instead of doing it for
each SPI transfer when dma_setup is called. This shall speed the DMA-based
SPI transfer initialization up a bit, particularly if the APB bus is
relatively slow.

Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20200920112322.24585-2-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: hisi-sfc-v3xx: fix spelling mistake "occured" -> "occurred"
Colin Ian King [Mon, 28 Sep 2020 12:30:42 +0000 (13:30 +0100)]
spi: hisi-sfc-v3xx: fix spelling mistake "occured" -> "occurred"

There is a spelling mistake in a dev_err message. Fix it.

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Link: https://lore.kernel.org/r/20200928123042.125359-1-colin.king@canonical.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agoMerge series "Add IRQ mode support for hisi-sfc-v3xx driver and some cleanups" from...
Mark Brown [Fri, 25 Sep 2020 20:40:21 +0000 (21:40 +0100)]
Merge series "Add IRQ mode support for hisi-sfc-v3xx driver and some cleanups" from Yicong Yang <yangyicong@hisilicon.com>:

This series mainly add the IRQ mode support for hisi-sfc-v3xx driver, and some
cleanups for the preparation of the IRQ mode.
After this patch, the device can work in IRQ mode, or if firmware doesn't
declare irq support it will fall back to Poll mode.

Patch 1-2 refactor the .exec_op() path to make it simpler and clearer.
Patch 3 factor the definition of the interrupt bits.
Patch 4 add the IRQ support of the driver.

Yicong Yang (4):
  spi: hisi-sfc-v3xx: factor out IO modes configuration
  spi: hisi-sfc-v3xx: factor out bus config and transfer functions
  spi: hisi-sfc-v3xx: factor out the bit definition of interrupt
    register
  spi: hisi-sfc-v3xx: add support for IRQ mode

 drivers/spi/spi-hisi-sfc-v3xx.c | 261 +++++++++++++++++++++++++++++-----------
 1 file changed, 190 insertions(+), 71 deletions(-)

--
2.8.1

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/

3 years agospi: spi-mtk-nor: fix timeout calculation overflow
Chuanhong Guo [Tue, 22 Sep 2020 11:49:02 +0000 (19:49 +0800)]
spi: spi-mtk-nor: fix timeout calculation overflow

CLK_TO_US macro is used to calculate potential transfer time for various
timeout handling. However it overflows on transfer bigger than 512 bytes
because it first did (len * 8 * 1000000).
This controller typically operates at 45MHz. This patch did 2 things:
1. calculate clock / 1000000 first
2. add a 4M transfer size cap so that the final timeout in DMA reading
   doesn't overflow

Fixes: 881d1ee9fe81f ("spi: add support for mediatek spi-nor controller")
Cc: <stable@vger.kernel.org>
Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Link: https://lore.kernel.org/r/20200922114905.2942859-1-gch981213@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: hisi-sfc-v3xx: add support for IRQ mode
Yicong Yang [Thu, 24 Sep 2020 12:24:30 +0000 (20:24 +0800)]
spi: hisi-sfc-v3xx: add support for IRQ mode

The controller can work with interrupts, so add support for it.
Then we can work under IRQ mode or Poll mode now, if firmware
doesn't declare the IRQ support, it will fall back to Poll mode.

Acked-by: John Garry <john.garry@huawei.com>
Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Link: https://lore.kernel.org/r/1600950270-52536-5-git-send-email-yangyicong@hisilicon.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: hisi-sfc-v3xx: factor out the bit definition of interrupt register
Yicong Yang [Thu, 24 Sep 2020 12:24:29 +0000 (20:24 +0800)]
spi: hisi-sfc-v3xx: factor out the bit definition of interrupt register

The definition of the register field in the interrupt corresponding
registers are the same. So factor them out to public place.

Acked-by: John Garry <john.garry@huawei.com>
Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Link: https://lore.kernel.org/r/1600950270-52536-4-git-send-email-yangyicong@hisilicon.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: hisi-sfc-v3xx: factor out bus config and transfer functions
Yicong Yang [Thu, 24 Sep 2020 12:24:28 +0000 (20:24 +0800)]
spi: hisi-sfc-v3xx: factor out bus config and transfer functions

In hisi_sfc_v3xx_generic_exec_op(), we will write the data to the buffer,
configure and start the transfer, read the data to the buffer and check
whether occurs an error. Factor out the config and transfer start codes
as individual functions, to make the process a bit clearer.

Acked-by: John Garry <john.garry@huawei.com>
Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Link: https://lore.kernel.org/r/1600950270-52536-3-git-send-email-yangyicong@hisilicon.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: hisi-sfc-v3xx: factor out IO modes configuration
Yicong Yang [Thu, 24 Sep 2020 12:24:27 +0000 (20:24 +0800)]
spi: hisi-sfc-v3xx: factor out IO modes configuration

Factor IO modes configuration out of hisi_sfc_v3xx_generic_exec_op()
using an IO modes lookup table. This will make the process a bit clearer
and reduce the cyclomatic complexity. Simplify the IO mode definition
macros a little bit as well.

Also add the .supports_op() method for the controller mem ops, in order
to avoid OOB access.

Acked-by: John Garry <john.garry@huawei.com>
Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
Link: https://lore.kernel.org/r/1600950270-52536-2-git-send-email-yangyicong@hisilicon.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-zynqmp-gqspi: Fix incorrect indentation
Amit Kumar Mahapatra [Thu, 24 Sep 2020 07:11:19 +0000 (09:11 +0200)]
spi: spi-zynqmp-gqspi: Fix incorrect indentation

Fixed incorrect indentation in ZynqMP qspi controller driver.

Signed-off-by: Amit Kumar Mahapatra <amit.kumar-mahapatra@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Link: https://lore.kernel.org/r/2b246b6f0925c8a2a767a4240e8738ffeefd62be.1600931476.git.michal.simek@xilinx.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-zynqmp-gqspi: Update driver to use spi-mem framework
Amit Kumar Mahapatra [Thu, 24 Sep 2020 07:11:18 +0000 (09:11 +0200)]
spi: spi-zynqmp-gqspi: Update driver to use spi-mem framework

Updated Zynqmp qspi controller driver to use spi-mem framework.

Signed-off-by: Amit Kumar Mahapatra <amit.kumar-mahapatra@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Link: https://lore.kernel.org/r/490a7642a975f4d3dd9618304e9e45f7e2414661.1600931476.git.michal.simek@xilinx.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-zynqmp-gqspi: Fix kernel-doc warnings
Amit Kumar Mahapatra [Thu, 24 Sep 2020 07:11:17 +0000 (09:11 +0200)]
spi: spi-zynqmp-gqspi: Fix kernel-doc warnings

Fix kernel-doc warnings in ZynqMP qspi driver file.

Signed-off-by: Amit Kumar Mahapatra <amit.kumar-mahapatra@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Link: https://lore.kernel.org/r/ba5920c57eee06fafa6f9d1df9859e69819ac301.1600931476.git.michal.simek@xilinx.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-imx: spi_imx_transfer(): add support for effective_speed_hz
Marc Kleine-Budde [Thu, 17 Sep 2020 20:24:20 +0000 (22:24 +0200)]
spi: spi-imx: spi_imx_transfer(): add support for effective_speed_hz

This patch implementes the reporting of the effectivly used speed_hz for the
transfer by setting tfr->effective_speed_hz.

See the following patch, which adds this feature to the SPI core for more
information:

    5d7e2b5ed585 spi: core: allow reporting the effectivly used speed_hz for a transfer

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://lore.kernel.org/r/20200917202420.1914104-1-mkl@pengutronix.de
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: omap2-mcspi: Improve performance waiting for CHSTAT
Aswath Govindraju [Thu, 10 Sep 2020 12:26:24 +0000 (17:56 +0530)]
spi: omap2-mcspi: Improve performance waiting for CHSTAT

This reverts commit 13d515c796 (spi: omap2-mcspi: Switch to
readl_poll_timeout()).

The amount of time spent polling for the MCSPI_CHSTAT bits to be set on
AM335x-icev2 platform is less than 1us (about 0.6us) in most cases, with
or without using DMA. So, in most cases the function need not sleep.
Also, setting the sleep_usecs to zero would not be optimal here because
ktime_add_us() used in readl_poll_timeout() is slower compared to the
direct addition used after the revert. So, it is sub-optimal to use
readl_poll_timeout in this case.

When DMA is not enabled, this revert results in an increase of about 27%
in throughput and decrease of about 20% in CPU usage. However, the CPU
usage and throughput are almost the same when used with DMA.

Therefore, fix this by reverting the commit which switched to using
readl_poll_timeout().

Fixes: 13d515c796ad ("spi: omap2-mcspi: Switch to readl_poll_timeout()")
Signed-off-by: Aswath Govindraju <a-govindraju@ti.com>
Link: https://lore.kernel.org/r/20200910122624.8769-1-a-govindraju@ti.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: npcm-fiu: simplify the return expression of npcm_fiu_probe()
Qinglang Miao [Mon, 21 Sep 2020 13:11:06 +0000 (21:11 +0800)]
spi: npcm-fiu: simplify the return expression of npcm_fiu_probe()

Simplify the return expression.

Signed-off-by: Qinglang Miao <miaoqinglang@huawei.com>
Link: https://lore.kernel.org/r/20200921131106.93228-1-miaoqinglang@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi/topcliff-pch: drop double zeroing
Julia Lawall [Sun, 20 Sep 2020 11:26:23 +0000 (13:26 +0200)]
spi/topcliff-pch: drop double zeroing

sg_init_table zeroes its first argument, so the allocation of that argument
doesn't have to.

the semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@
expression x,n,flags;
@@

x =
- kcalloc
+ kmalloc_array
  (n,sizeof(*x),flags)
...
sg_init_table(x,n)
// </smpl>

Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
Link: https://lore.kernel.org/r/1600601186-7420-12-git-send-email-Julia.Lawall@inria.fr
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agoMerge series "spi: Fixes for FSI-attached controller" from Eddie James <eajames@linux...
Mark Brown [Thu, 17 Sep 2020 18:56:05 +0000 (19:56 +0100)]
Merge series "spi: Fixes for FSI-attached controller" from Eddie James <eajames@linux.ibm.com>:

This series implements a number of fixes for the FSI-attached SPI
controller driver.

Changes since v1:
 - Switch to a new compatible string for the restricted version of the
   SPI controller, rather than a new boolean parameter.

Brad Bishop (3):
  spi: fsi: Handle 9 to 15 byte transfers lengths
  spi: fsi: Fix clock running too fast
  spi: fsi: Fix use of the bneq+ sequencer instruction

Eddie James (3):
  dt-bindings: fsi: fsi2spi: Add compatible string for restricted
    version
  spi: fsi: Implement restricted size for certain controllers
  spi: fsi: Check mux status before transfers

 .../devicetree/bindings/fsi/ibm,fsi2spi.yaml  |   1 +
 drivers/spi/spi-fsi.c                         | 139 ++++++++++++++----
 2 files changed, 109 insertions(+), 31 deletions(-)

--
2.26.2

3 years agospi: dw-pci: free previously allocated IRQs if desc->setup() fails
Jay Fang [Tue, 15 Sep 2020 01:22:49 +0000 (09:22 +0800)]
spi: dw-pci: free previously allocated IRQs if desc->setup() fails

Free previously allocated IRQs when return an error code of desc->setup()
which is not always successful. And simplify the code by adding a goto
label.

Fixes: 8f5c285f3ef5 ("SPI: designware: pci: Switch over to MSI interrupts")
CC: Felipe Balbi <felipe.balbi@linux.intel.com>
Signed-off-by: Jay Fang <f.fangjian@huawei.com>
Link: https://lore.kernel.org/r/1600132969-53037-1-git-send-email-f.fangjian@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-nxp-fspi: Add ACPI support
kuldip dwivedi [Fri, 11 Sep 2020 13:03:31 +0000 (18:33 +0530)]
spi: spi-nxp-fspi: Add ACPI support

Currently NXP fspi  driver has support of DT only. Adding ACPI
support to the driver so that it can be used by UEFI firmware
booting in ACPI mode. This driver will be probed if any firmware
will expose HID "NXP0009" in DSDT table.

Signed-off-by: kuldip dwivedi <kuldip.dwivedi@puresoftware.com>
Reviewed-by: Ashish Kumar <Ashish.Kumar@nxp.com>
Link: https://lore.kernel.org/r/20200911130331.6313-1-kuldip.dwivedi@puresoftware.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: lantiq: remove redundant irqsave and irqrestore in hardIRQ
Barry Song [Wed, 16 Sep 2020 10:10:42 +0000 (22:10 +1200)]
spi: lantiq: remove redundant irqsave and irqrestore in hardIRQ

Running in hardIRQ, disabling irq is redundant.

Signed-off-by: Barry Song <song.bao.hua@hisilicon.com>
Link: https://lore.kernel.org/r/20200916101042.21860-1-song.bao.hua@hisilicon.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: xilinx: Fix info message during probe
Ricardo Ribalda [Tue, 15 Sep 2020 11:29:36 +0000 (13:29 +0200)]
spi: xilinx: Fix info message during probe

The info message was showing the mapped address of the device. To avoid
security problems, all virtual addresses are converted to __ptrval__, so
the message was useless/ugly:

[    2.304949] xilinx_spi b0010000.spi-flash: at 0xB0010000 mapped to 0x(____ptrval____), irq=37

Use %pR instead:

[   15.021354] xilinx_spi b0010000.spi-flash: at [mem 0xb0010000-0xb001ffff], irq=37

Signed-off-by: Ricardo Ribalda <ribalda@kernel.org>
Acked-by: Michal Simek <michal.simek@xilinx.com>
Link: https://lore.kernel.org/r/20200915112936.320647-1-ribalda@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: fsi: fsi2spi: Add compatible string for restricted version
Eddie James [Wed, 9 Sep 2020 22:28:55 +0000 (17:28 -0500)]
spi: fsi: fsi2spi: Add compatible string for restricted version

Add a compatible string for the restricted version of the SPI controller.
The restricted version cannot process sequence loop operations and
therefore has a smaller transfer size.

Signed-off-by: Eddie James <eajames@linux.ibm.com>
Link: https://lore.kernel.org/r/20200909222857.28653-5-eajames@linux.ibm.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: fsi: Check mux status before transfers
Eddie James [Wed, 9 Sep 2020 22:28:57 +0000 (17:28 -0500)]
spi: fsi: Check mux status before transfers

The SPI controllers are not accessible if the mux isn't set. Therefore,
check the mux status before starting a transfer and fail out if it isn't
set.

Signed-off-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Joel Stanley <joel@jms.id.au>
Link: https://lore.kernel.org/r/20200909222857.28653-7-eajames@linux.ibm.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: fsi: Implement restricted size for certain controllers
Eddie James [Wed, 9 Sep 2020 22:28:56 +0000 (17:28 -0500)]
spi: fsi: Implement restricted size for certain controllers

Some of the FSI-attached SPI controllers cannot use the loop command in
programming the sequencer due to security requirements. Check the
devicetree compatibility that indicates this condition and restrict the
size for these controllers. Also, add more transfers directly in the
sequence up to the length of the sequence register.

Fixes: bbb6b2f9865b ("spi: Add FSI-attached SPI controller driver")
Signed-off-by: Eddie James <eajames@linux.ibm.com>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Joel Stanley <joel@jms.id.au>
Link: https://lore.kernel.org/r/20200909222857.28653-6-eajames@linux.ibm.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: fsi: Fix use of the bneq+ sequencer instruction
Brad Bishop [Wed, 9 Sep 2020 22:28:54 +0000 (17:28 -0500)]
spi: fsi: Fix use of the bneq+ sequencer instruction

All of the switches in N2_count_control in the counter configuration are
required to make the branch if not equal and increment command work.
Set them when using bneq+.

A side effect of this mode requires a dummy write to TDR when both
transmitting and receiving otherwise the controller won't start shifting
receive data.

It is likely not possible to avoid TDR underrun errors in this mode and
they are harmless, so do not check for them.

Fixes: bbb6b2f9865b ("spi: Add FSI-attached SPI controller driver")
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Signed-off-by: Eddie James <eajames@linux.ibm.com>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Joel Stanley <joel@jms.id.au>
Link: https://lore.kernel.org/r/20200909222857.28653-4-eajames@linux.ibm.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: fsi: Fix clock running too fast
Brad Bishop [Wed, 9 Sep 2020 22:28:53 +0000 (17:28 -0500)]
spi: fsi: Fix clock running too fast

Use a clock divider tuned to a 200MHz FSI bus frequency (the maximum). Use
of the previous divider at 200MHz results in corrupt data from endpoint
devices. Ideally the clock divider would be calculated from the FSI clock,
but that would require some significant work on the FSI driver. With FSI
frequencies slower than 200MHz, the SPI clock will simply run slower, but
safely.

Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Signed-off-by: Eddie James <eajames@linux.ibm.com>
Signed-off-by: Joel Stanley <joel@jms.id.au>
Link: https://lore.kernel.org/r/20200909222857.28653-3-eajames@linux.ibm.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: fsi: Handle 9 to 15 byte transfers lengths
Brad Bishop [Wed, 9 Sep 2020 22:28:52 +0000 (17:28 -0500)]
spi: fsi: Handle 9 to 15 byte transfers lengths

The trailing <len> - 8 bytes of transfer data in this size range is no
longer ignored.

Fixes: bbb6b2f9865b ("spi: Add FSI-attached SPI controller driver")
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Signed-off-by: Eddie James <eajames@linux.ibm.com>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Joel Stanley <joel@jms.id.au>
Link: https://lore.kernel.org/r/20200909222857.28653-2-eajames@linux.ibm.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: sprd: Simplify with dev_err_probe()
Krzysztof Kozlowski [Thu, 10 Sep 2020 16:07:06 +0000 (18:07 +0200)]
spi: sprd: Simplify with dev_err_probe()

Common pattern of handling deferred probe can be simplified with
dev_err_probe().  Less code and the error value gets printed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Acked-by: Chunyan Zhang <zhang.lyra@gmail.com>
Link: https://lore.kernel.org/r/20200910160706.5883-1-krzk@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-geni-qcom: Don't wait to start 1st transfer if transmitting
Douglas Anderson [Sat, 12 Sep 2020 18:17:25 +0000 (11:17 -0700)]
spi: spi-geni-qcom: Don't wait to start 1st transfer if transmitting

If we're sending bytes over SPI, we know the FIFO is empty at the
start of the transfer.  There's no reason to wait for the interrupt
telling us to start--we can just start right away.  Then if we
transmit everything in one swell foop we don't even need to bother
listening for TX interrupts.

In a test of "flashrom -p ec -r /tmp/foo.bin" interrupts were reduced
from ~30560 to ~29730, about a 3% savings.

This patch looks bigger than it is because I moved a few functions
rather than adding a forward declaration.  The only actual change to
geni_spi_handle_tx() was to make it return a bool indicating if there
is more to tx.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Akash Asthana <akashast@codeaurora.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20200912111716.1.Ied5e843fad0d6b733a1fb8bcfb364dd2fa889eb3@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: bcm2835: Make polling_limit_us static
Jason Yan [Sat, 12 Sep 2020 07:22:11 +0000 (15:22 +0800)]
spi: bcm2835: Make polling_limit_us static

This eliminates the following sparse warning:

drivers/spi/spi-bcm2835.c:78:14: warning: symbol 'polling_limit_us' was
not declared. Should it be static?

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
Link: https://lore.kernel.org/r/20200912072211.602735-1-yanaijie@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-fsl-dspi: use XSPI mode instead of DMA for DPAA2 SoCs
Vladimir Oltean [Thu, 10 Sep 2020 12:15:32 +0000 (15:15 +0300)]
spi: spi-fsl-dspi: use XSPI mode instead of DMA for DPAA2 SoCs

The arch/arm64/boot/dts/freescale/fsl-ls208xa.dtsi device tree lacks DMA
channels for DSPI, so naturally, the driver fails to probe:

[ 2.945302] fsl-dspi 2100000.spi: rx dma channel not available
[ 2.951134] fsl-dspi 2100000.spi: can't get dma channels

In retrospect, this should have been obvious, because LS2080A, LS2085A
LS2088A and LX2160A don't appear to have an eDMA module at all. Looking
again at their datasheets, the CTARE register (which is specific to XSPI
functionality) seems to be documented, so switch them to XSPI mode
instead.

Fixes: 0feaf8f5afe0 ("spi: spi-fsl-dspi: Convert the instantiations that support it to DMA")
Reported-by: Qiang Zhao <qiang.zhao@nxp.com>
Tested-by: Qiang Zhao <qiang.zhao@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://lore.kernel.org/r/20200910121532.1138596-1-olteanv@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-geni-qcom: Don't program CS_TOGGLE again and again
Douglas Anderson [Sat, 12 Sep 2020 21:08:00 +0000 (14:08 -0700)]
spi: spi-geni-qcom: Don't program CS_TOGGLE again and again

We always toggle the chip select manually in spi-geni-qcom so that we
can properly implement the Linux API.  There's no reason to program
this to the hardware on every transfer.  Program it once at init and
be done with it.

This saves some part of a microsecond of overhead on each transfer.
While not really noticeable on any real world benchmarks, we might as
well save the time.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20200912140730.2.I33e571179986850b4ec17042e813d0b08fb1b9c1@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-geni-qcom: Use the FIFO even more
Douglas Anderson [Sat, 12 Sep 2020 21:07:59 +0000 (14:07 -0700)]
spi: spi-geni-qcom: Use the FIFO even more

In commit 902481a78ee4 ("spi: spi-geni-qcom: Actually use our FIFO") I
explained that the maximum size we could program the FIFO was
"mas->tx_fifo_depth - 3" but that I chose "mas->tx_fifo_depth()"
because I was worried about decreased bandwidth.

Since that time:
* All the interconnect patches have landed, making things run at the
  proper speed.
* I've done more measurements.

This lets me confirm that there's really no downside of using the FIFO
more.  Specifically I did "flashrom -p ec -r /tmp/foo.bin" on a
Chromebook and averaged over several runs.

Before: It took 6.66 seconds and 59669 interrupts fired.
After:  It took 6.66 seconds and 47992 interrupts fired.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20200912140730.1.Ie67fa32009b94702d56232c064f1d89065ee8836@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-qcom-qspi: replace spin_lock_irqsave by spin_lock in hard IRQ
Barry Song [Thu, 10 Sep 2020 10:02:46 +0000 (22:02 +1200)]
spi: spi-qcom-qspi: replace spin_lock_irqsave by spin_lock in hard IRQ

It is redundant to do irqsave and irqrestore in hardIRQ context.

Cc: Andy Gross <agross@kernel.org>
Cc: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Barry Song <song.bao.hua@hisilicon.com>
Link: https://lore.kernel.org/r/20200910100246.32696-1-song.bao.hua@hisilicon.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: Add compatible string for brcmstb SoCs
Ray Jui [Thu, 10 Sep 2020 15:25:36 +0000 (08:25 -0700)]
spi: Add compatible string for brcmstb SoCs

Add compatible string for brcmstb 7445 SoCs.

Signed-off-by: Ray Jui <ray.jui@broadcom.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20200910152539.45584-1-ray.jui@broadcom.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: bcm-qspi: Clean up 7425, 7429, and 7435 settings
Ray Jui [Thu, 10 Sep 2020 15:25:39 +0000 (08:25 -0700)]
spi: bcm-qspi: Clean up 7425, 7429, and 7435 settings

The Broadcom QSPI driver now falls back to no MSPI_DEV support as the
default setting in the generic compatible string, explicit settings for
STB chips 7425, 7429, and 7435 can be removed.

Signed-off-by: Ray Jui <ray.jui@broadcom.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20200910152539.45584-4-ray.jui@broadcom.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: bcm-qspi: Fix probe regression on iProc platforms
Ray Jui [Thu, 10 Sep 2020 15:25:38 +0000 (08:25 -0700)]
spi: bcm-qspi: Fix probe regression on iProc platforms

iProc chips have QSPI controller that does not have the MSPI_REV
offset. Reading from that offset will cause a bus error. Fix it by
having MSPI_REV query disabled in the generic compatible string.

Fixes: 3a01f04d74ef ("spi: bcm-qspi: Handle lack of MSPI_REV offset")
Link: https://lore.kernel.org/linux-arm-kernel/20200909211857.4144718-1-f.fainelli@gmail.com/T/#u
Signed-off-by: Ray Jui <ray.jui@broadcom.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20200910152539.45584-3-ray.jui@broadcom.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: bcm-qspi: Add compatible string for BRCMSTB 7445 SoCs
Ray Jui [Thu, 10 Sep 2020 15:25:37 +0000 (08:25 -0700)]
spi: bcm-qspi: Add compatible string for BRCMSTB 7445 SoCs

Add compatible string for BRCMSTB 7445 SoCs and indicate it has MSPI rev
support.

Signed-off-by: Ray Jui <ray.jui@broadcom.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20200910152539.45584-2-ray.jui@broadcom.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: qup: remove redundant assignment to variable ret
Colin Ian King [Thu, 10 Sep 2020 15:04:10 +0000 (16:04 +0100)]
spi: qup: remove redundant assignment to variable ret

The variable ret is being initialized with a value that is
never read and it is being updated later with a new value. The
initialization is redundant and can be removed.

Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Link: https://lore.kernel.org/r/20200910150410.750959-1-colin.king@canonical.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agoMerge series "opp: Unconditionally call dev_pm_opp_of_remove_table()" from Viresh...
Mark Brown [Wed, 9 Sep 2020 15:27:47 +0000 (16:27 +0100)]
Merge series "opp: Unconditionally call dev_pm_opp_of_remove_table()" from Viresh Kumar <viresh.kumar@linaro.org>:

Hello,

This cleans up some of the user code around calls to
dev_pm_opp_of_remove_table().

All the patches can be picked by respective maintainers directly except
for the last patch, which needs the previous two to get merged first.

These are based for 5.9-rc1.

Rajendra, Since most of these changes are related to qcom stuff, it
would be great if you can give them a try. I wasn't able to test them
due to lack of hardware.

Ulf, I had to revise the sdhci patch, sorry about that. Please pick this
one.

Diff between V1 and V2 is mentioned in each of the patches separately.

Viresh Kumar (8):
  cpufreq: imx6q: Unconditionally call dev_pm_opp_of_remove_table()
  drm/lima: Unconditionally call dev_pm_opp_of_remove_table()
  drm/msm: Unconditionally call dev_pm_opp_of_remove_table()
  mmc: sdhci-msm: Unconditionally call dev_pm_opp_of_remove_table()
  spi: spi-geni-qcom: Unconditionally call dev_pm_opp_of_remove_table()
  spi: spi-qcom-qspi: Unconditionally call dev_pm_opp_of_remove_table()
  tty: serial: qcom_geni_serial: Unconditionally call
    dev_pm_opp_of_remove_table()
  qcom-geni-se: remove has_opp_table

 drivers/cpufreq/imx6q-cpufreq.c         | 10 ++--------
 drivers/gpu/drm/lima/lima_devfreq.c     |  6 +-----
 drivers/gpu/drm/lima/lima_devfreq.h     |  1 -
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c | 14 +++++---------
 drivers/gpu/drm/msm/disp/dpu1/dpu_kms.h |  1 -
 drivers/gpu/drm/msm/dsi/dsi_host.c      |  8 ++------
 drivers/mmc/host/sdhci-msm.c            | 14 +++++---------
 drivers/spi/spi-geni-qcom.c             | 13 +++++--------
 drivers/spi/spi-qcom-qspi.c             | 15 ++++++---------
 drivers/tty/serial/qcom_geni_serial.c   | 13 +++++--------
 include/linux/qcom-geni-se.h            |  2 --
 11 files changed, 31 insertions(+), 66 deletions(-)

base-commit: f4d51dffc6c01a9e94650d95ce0104964f8ae822
--
2.25.0.rc1.19.g042ed3e048af

3 years agospi: spidev: Remove redundant initialization of variable status
Jay Fang [Wed, 9 Sep 2020 06:08:24 +0000 (14:08 +0800)]
spi: spidev: Remove redundant initialization of variable status

In spidev_read() and spidev_write(), the variable status is being
initialized with a value that is never read and it is being updated
later with a new value. The initialization is redundant and can be
removed.

Signed-off-by: Jay Fang <f.fangjian@huawei.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/1599631704-53232-1-git-send-email-f.fangjian@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-qcom-qspi: Unconditionally call dev_pm_opp_of_remove_table()
Viresh Kumar [Fri, 28 Aug 2020 06:07:51 +0000 (11:37 +0530)]
spi: spi-qcom-qspi: Unconditionally call dev_pm_opp_of_remove_table()

dev_pm_opp_of_remove_table() doesn't report any errors when it fails to
find the OPP table with error -ENODEV (i.e. OPP table not present for
the device). And we can call dev_pm_opp_of_remove_table()
unconditionally here.

While at it, create a new label and put clkname on errors.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Link: https://lore.kernel.org/r/b77aa0bbe82a580508e321a34da488b4b27966d0.1598594714.git.viresh.kumar@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: spi-geni-qcom: Unconditionally call dev_pm_opp_of_remove_table()
Viresh Kumar [Fri, 28 Aug 2020 06:07:50 +0000 (11:37 +0530)]
spi: spi-geni-qcom: Unconditionally call dev_pm_opp_of_remove_table()

dev_pm_opp_of_remove_table() doesn't report any errors when it fails to
find the OPP table with error -ENODEV (i.e. OPP table not present for
the device). And we can call dev_pm_opp_of_remove_table()
unconditionally here.

While at it, create a new label and put clkname on errors.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Link: https://lore.kernel.org/r/ea0864d41277e61fa31d304fbd4cf9af6b314269.1598594714.git.viresh.kumar@linaro.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agoMerge series "spi: Adding support for Microchip Sparx5 SoC" from Lars Povlsen <lars...
Mark Brown [Tue, 8 Sep 2020 17:19:33 +0000 (18:19 +0100)]
Merge series "spi: Adding support for Microchip Sparx5 SoC" from Lars Povlsen <lars.povlsen@microchip.com>:

The series add support for the Sparx5 SoC SPI controller in the
spi-dw-mmio.c spi driver.

v5 changes:
- rx-sample-delay-ns documentation changes from Rob Herring:
 - Drop superfluous type $ref
 - Add default value = 0

v4 changes:
- Changed snps,rx-sample-delay-ns to snps,rx-sample-delay-ns
  suggested by Rob Herring (rockchip also has this property).
- Added support for controller-level rx-sample-delay-ns value as
  well as per SPI slave value (rockchip has controller-level property).
- Dropped internal mux in favor of suggested spi-mux to
  control bus inteface selection.

v3 changes:
- Added mux support for controlling SPI bus interface. This is new mux
  driver, bindings and added to sparx5 base DT.
- Removed "microchip,spi-interface2" property in favour of
  "mux-controls" property in SPI controller (sparx5 only).
- Changed dw_spi_sparx5_set_cs() to use the mux control instead of
  directly acessing "mux" register. Associated code/defines moved to mux
  driver.
- Changed dw_spi_sparx5_set_cs() to match other similar functions in
  signature and avoid explicit CS toggling.
- Spun off duplicated NAND device DT chunks into separate DT file.

v2 changes:
- Moved all RX sample delay into spi-dw-core.c, using
  the "snps,rx-sample-delay-ns" device property.
- Integrated Sparx5 support directly in spi-dw-mmio.c
- Changed SPI2 configuration to per-slave "microchip,spi-interface2"
  property.
- Added bindings to existing snps,dw-apb-ssi.yaml file
- Dropped patches for polled mode and SPI memory operations.

Lars Povlsen (6):
  spi: dw: Add support for RX sample delay register
  spi: dw: Add Microchip Sparx5 support
  arm64: dts: sparx5: Add SPI controller and associated mmio-mux
  dt-bindings: snps,dw-apb-ssi: Add sparx5 support, plus
    rx-sample-delay-ns property
  arm64: dts: sparx5: Add spi-nor support
  arm64: dts: sparx5: Add spi-nand devices

 .../bindings/spi/snps,dw-apb-ssi.yaml         | 21 ++++++
 arch/arm64/boot/dts/microchip/sparx5.dtsi     | 47 ++++++++++++-
 .../arm64/boot/dts/microchip/sparx5_nand.dtsi | 31 ++++++++
 .../boot/dts/microchip/sparx5_pcb125.dts      | 30 ++++++++
 .../boot/dts/microchip/sparx5_pcb134.dts      |  1 +
 .../dts/microchip/sparx5_pcb134_board.dtsi    | 16 +++++
 .../boot/dts/microchip/sparx5_pcb135.dts      |  1 +
 .../dts/microchip/sparx5_pcb135_board.dtsi    | 16 +++++
 drivers/spi/spi-dw-core.c                     | 26 +++++++
 drivers/spi/spi-dw-mmio.c                     | 70 ++++++++++++++++++-
 drivers/spi/spi-dw.h                          |  3 +
 11 files changed, 260 insertions(+), 2 deletions(-)
 create mode 100644 arch/arm64/boot/dts/microchip/sparx5_nand.dtsi

--
2.27.0

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

3 years agospi: spi-mtk-nor: support standard spi properties
Ikjoon Jang [Wed, 26 Aug 2020 09:18:52 +0000 (17:18 +0800)]
spi: spi-mtk-nor: support standard spi properties

Use default supports_op() to support spi-[rt]x-bus-width properties.
And check dummy op's byte length instead of its bus width for output.

Signed-off-by: Ikjoon Jang <ikjn@chromium.org>
Link: https://lore.kernel.org/r/20200826091852.519138-1-ikjn@chromium.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: tegra20: Simplify with dev_err_probe()
Krzysztof Kozlowski [Tue, 1 Sep 2020 15:27:13 +0000 (17:27 +0200)]
spi: tegra20: Simplify with dev_err_probe()

Common pattern of handling deferred probe can be simplified with
dev_err_probe().  Less code and the error value gets printed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200901152713.18629-11-krzk@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: tegra114: Simplify with dev_err_probe()
Krzysztof Kozlowski [Tue, 1 Sep 2020 15:27:12 +0000 (17:27 +0200)]
spi: tegra114: Simplify with dev_err_probe()

Common pattern of handling deferred probe can be simplified with
dev_err_probe().  Less code and the error value gets printed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200901152713.18629-10-krzk@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: synquacer: Simplify with dev_err_probe()
Krzysztof Kozlowski [Tue, 1 Sep 2020 15:27:11 +0000 (17:27 +0200)]
spi: synquacer: Simplify with dev_err_probe()

Common pattern of handling deferred probe can be simplified with
dev_err_probe().  Less code and the error value gets printed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200901152713.18629-9-krzk@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: stm32: Simplify with dev_err_probe()
Krzysztof Kozlowski [Tue, 1 Sep 2020 15:27:10 +0000 (17:27 +0200)]
spi: stm32: Simplify with dev_err_probe()

Common pattern of handling deferred probe can be simplified with
dev_err_probe().  Less code and the error value gets printed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200901152713.18629-8-krzk@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
3 years agospi: qcom-qspi: Simplify with dev_err_probe()
Krzysztof Kozlowski [Tue, 1 Sep 2020 15:27:09 +0000 (17:27 +0200)]
spi: qcom-qspi: Simplify with dev_err_probe()

Common pattern of handling deferred probe can be simplified with
dev_err_probe().  Less code and the error value gets printed.

Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Link: https://lore.kernel.org/r/20200901152713.18629-7-krzk@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>